korg/linux
0
0
Fork 0
mirror of https://mirrors.bfsu.edu.cn/git/linux.git synced 2024-12-28 21:45:01 +08:00
Code Issues Packages Projects Releases Wiki Activity
7b4cf994e4
linux/drivers/gpu/drm/radeon/r600.c

4326 lines
128 KiB
C
Raw Normal View History

drm/radeon: introduce kernel modesetting for radeon hardware Add kernel modesetting support to radeon driver, use the ttm memory manager to manage memory and DRM/GEM to provide userspace API. In order to avoid backward compatibility issue and to allow clean design and code the radeon kernel modesetting use different code path than old radeon/drm driver. When kernel modesetting is enabled the IOCTL of radeon/drm driver are considered as invalid and an error message is printed in the log and they return failure. KMS enabled userspace will use new API to talk with the radeon/drm driver. The new API provide functions to create/destroy/share/mmap buffer object which are then managed by the kernel memory manager (here TTM). In order to submit command to the GPU the userspace provide a buffer holding the command stream, along this buffer userspace have to provide a list of buffer object used by the command stream. The kernel radeon driver will then place buffer in GPU accessible memory and will update command stream to reflect the position of the different buffers. The kernel will also perform security check on command stream provided by the user, we want to catch and forbid any illegal use of the GPU such as DMA into random system memory or into memory not owned by the process supplying the command stream. This part of the code is still incomplete and this why we propose that patch as a staging driver addition, future security might forbid current experimental userspace to run. This code support the following hardware : R1XX,R2XX,R3XX,R4XX,R5XX (radeon up to X1950). Works is underway to provide support for R6XX, R7XX and newer hardware (radeon from HD2XXX to HD4XXX). Authors: Jerome Glisse <jglisse@redhat.com> Dave Airlie <airlied@redhat.com> Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com> Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-06-05 20:42:42 +08:00
/*
* Copyright 2008 Advanced Micro Devices, Inc.
* Copyright 2008 Red Hat Inc.
* Copyright 2009 Jerome Glisse.
*
* Permission is hereby granted, free of charge, to any person obtaining a
* copy of this software and associated documentation files (the "Software"),
* to deal in the Software without restriction, including without limitation
* the rights to use, copy, modify, merge, publish, distribute, sublicense,
* and/or sell copies of the Software, and to permit persons to whom the
* Software is furnished to do so, subject to the following conditions:
*
* The above copyright notice and this permission notice shall be included in
* all copies or substantial portions of the Software.
*
* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
* IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
* FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
* THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
* OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
* ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
* OTHER DEALINGS IN THE SOFTWARE.
*
* Authors: Dave Airlie
* Alex Deucher
* Jerome Glisse
*/
include cleanup: Update gfp.h and slab.h includes to prepare for breaking implicit slab.h inclusion from percpu.h percpu.h is included by sched.h and module.h and thus ends up being included when building most .c files. percpu.h includes slab.h which in turn includes gfp.h making everything defined by the two files universally available and complicating inclusion dependencies. percpu.h -> slab.h dependency is about to be removed. Prepare for this change by updating users of gfp and slab facilities include those headers directly instead of assuming availability. As this conversion needs to touch large number of source files, the following script is used as the basis of conversion. http://userweb.kernel.org/~tj/misc/slabh-sweep.py The script does the followings. * Scan files for gfp and slab usages and update includes such that only the necessary includes are there. ie. if only gfp is used, gfp.h, if slab is used, slab.h. * When the script inserts a new include, it looks at the include blocks and try to put the new include such that its order conforms to its surrounding. It's put in the include block which contains core kernel includes, in the same order that the rest are ordered - alphabetical, Christmas tree, rev-Xmas-tree or at the end if there doesn't seem to be any matching order. * If the script can't find a place to put a new include (mostly because the file doesn't have fitting include block), it prints out an error message indicating which .h file needs to be added to the file. The conversion was done in the following steps. 1. The initial automatic conversion of all .c files updated slightly over 4000 files, deleting around 700 includes and adding ~480 gfp.h and ~3000 slab.h inclusions. The script emitted errors for ~400 files. 2. Each error was manually checked. Some didn't need the inclusion, some needed manual addition while adding it to implementation .h or embedding .c file was more appropriate for others. This step added inclusions to around 150 files. 3. The script was run again and the output was compared to the edits from #2 to make sure no file was left behind. 4. Several build tests were done and a couple of problems were fixed. e.g. lib/decompress_*.c used malloc/free() wrappers around slab APIs requiring slab.h to be added manually. 5. The script was run on all .h files but without automatically editing them as sprinkling gfp.h and slab.h inclusions around .h files could easily lead to inclusion dependency hell. Most gfp.h inclusion directives were ignored as stuff from gfp.h was usually wildly available and often used in preprocessor macros. Each slab.h inclusion directive was examined and added manually as necessary. 6. percpu.h was updated not to include slab.h. 7. Build test were done on the following configurations and failures were fixed. CONFIG_GCOV_KERNEL was turned off for all tests (as my distributed build env didn't work with gcov compiles) and a few more options had to be turned off depending on archs to make things build (like ipr on powerpc/64 which failed due to missing writeq). * x86 and x86_64 UP and SMP allmodconfig and a custom test config. * powerpc and powerpc64 SMP allmodconfig * sparc and sparc64 SMP allmodconfig * ia64 SMP allmodconfig * s390 SMP allmodconfig * alpha SMP allmodconfig * um on x86_64 SMP allmodconfig 8. percpu.h modifications were reverted so that it could be applied as a separate patch and serve as bisection point. Given the fact that I had only a couple of failures from tests on step 6, I'm fairly confident about the coverage of this conversion patch. If there is a breakage, it's likely to be something in one of the arch headers which should be easily discoverable easily on most builds of the specific arch. Signed-off-by: Tejun Heo <tj@kernel.org> Guess-its-ok-by: Christoph Lameter <cl@linux-foundation.org> Cc: Ingo Molnar <mingo@redhat.com> Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
2010-03-24 16:04:11 +08:00
#include <linux/slab.h>
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
#include <linux/seq_file.h>
#include <linux/firmware.h>
#include <linux/platform_device.h>
gpu: add module.h to drivers/gpu files as required. So that we don't get build failures once the implicit module.h presence is removed. Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>
2011-08-30 23:04:30 +08:00
#include <linux/module.h>
UAPI: (Scripted) Convert #include "..." to #include <path/...> in drivers/gpu/ Convert #include "..." to #include <path/...> in drivers/gpu/. Signed-off-by: David Howells <dhowells@redhat.com> Acked-by: Dave Airlie <airlied@redhat.com> Acked-by: Arnd Bergmann <arnd@arndb.de> Acked-by: Thomas Gleixner <tglx@linutronix.de> Acked-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Acked-by: Dave Jones <davej@redhat.com>
2012-10-03 01:01:07 +08:00
#include <drm/drmP.h>
#include <drm/radeon_drm.h>
drm/radeon: introduce kernel modesetting for radeon hardware Add kernel modesetting support to radeon driver, use the ttm memory manager to manage memory and DRM/GEM to provide userspace API. In order to avoid backward compatibility issue and to allow clean design and code the radeon kernel modesetting use different code path than old radeon/drm driver. When kernel modesetting is enabled the IOCTL of radeon/drm driver are considered as invalid and an error message is printed in the log and they return failure. KMS enabled userspace will use new API to talk with the radeon/drm driver. The new API provide functions to create/destroy/share/mmap buffer object which are then managed by the kernel memory manager (here TTM). In order to submit command to the GPU the userspace provide a buffer holding the command stream, along this buffer userspace have to provide a list of buffer object used by the command stream. The kernel radeon driver will then place buffer in GPU accessible memory and will update command stream to reflect the position of the different buffers. The kernel will also perform security check on command stream provided by the user, we want to catch and forbid any illegal use of the GPU such as DMA into random system memory or into memory not owned by the process supplying the command stream. This part of the code is still incomplete and this why we propose that patch as a staging driver addition, future security might forbid current experimental userspace to run. This code support the following hardware : R1XX,R2XX,R3XX,R4XX,R5XX (radeon up to X1950). Works is underway to provide support for R6XX, R7XX and newer hardware (radeon from HD2XXX to HD4XXX). Authors: Jerome Glisse <jglisse@redhat.com> Dave Airlie <airlied@redhat.com> Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com> Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-06-05 20:42:42 +08:00
#include "radeon.h"
drm/radeon: include radeon_asic.h in the asic specific files In essence this creates a home for all asic specific declarations in radeon_asic.h Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-03-12 05:19:17 +08:00
#include "radeon_asic.h"
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
#include "radeon_mode.h"
#include "r600d.h"
#include "atom.h"
drm/radeon/kms: Convert RV515 to new init path and associated cleanup Convert the rv515 asic support to new init path also add an explanation in radeon.h about the new init path. There is also few cleanups associated with this change (others asic calling rv515 helper functions). Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@linux.ie>
2009-09-29 00:34:43 +08:00
#include "avivod.h"
drm/radeon: introduce kernel modesetting for radeon hardware Add kernel modesetting support to radeon driver, use the ttm memory manager to manage memory and DRM/GEM to provide userspace API. In order to avoid backward compatibility issue and to allow clean design and code the radeon kernel modesetting use different code path than old radeon/drm driver. When kernel modesetting is enabled the IOCTL of radeon/drm driver are considered as invalid and an error message is printed in the log and they return failure. KMS enabled userspace will use new API to talk with the radeon/drm driver. The new API provide functions to create/destroy/share/mmap buffer object which are then managed by the kernel memory manager (here TTM). In order to submit command to the GPU the userspace provide a buffer holding the command stream, along this buffer userspace have to provide a list of buffer object used by the command stream. The kernel radeon driver will then place buffer in GPU accessible memory and will update command stream to reflect the position of the different buffers. The kernel will also perform security check on command stream provided by the user, we want to catch and forbid any illegal use of the GPU such as DMA into random system memory or into memory not owned by the process supplying the command stream. This part of the code is still incomplete and this why we propose that patch as a staging driver addition, future security might forbid current experimental userspace to run. This code support the following hardware : R1XX,R2XX,R3XX,R4XX,R5XX (radeon up to X1950). Works is underway to provide support for R6XX, R7XX and newer hardware (radeon from HD2XXX to HD4XXX). Authors: Jerome Glisse <jglisse@redhat.com> Dave Airlie <airlied@redhat.com> Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com> Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-06-05 20:42:42 +08:00
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
#define PFP_UCODE_SIZE 576
#define PM4_UCODE_SIZE 1792
drm/radeon/kms: Add support for interrupts on r6xx/r7xx chips (v3) This enables the use of interrupts on r6xx/r7xx hardware. Interrupts are implemented via a ring buffer. The GPU adds interrupts vectors to the ring and the host reads them off in the interrupt handler. The interrupt controller requires firmware like the CP. This firmware must be installed and accessble to the firmware loader for interrupts to function. MSIs don't seem to work on my RS780. They work fine on all my discrete cards. I'm not sure about other RS780s or RS880s. I've disabled MSIs on RS780 and RS880, but it would probably be worth checking on some other systems. v2 - fix some checkpatch.pl problems; re-read the disp int status reg if we restart the ih; v3 - remove the irq handler if r600_irq_init() fails; remove spinlock in r600_ih_ring_fini(); move ih rb overflow check to r600_get_ih_wptr(); move irq ack to separate function; Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-02 02:43:46 +08:00
#define RLC_UCODE_SIZE 768
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
#define R700_PFP_UCODE_SIZE 848
#define R700_PM4_UCODE_SIZE 1360
drm/radeon/kms: Add support for interrupts on r6xx/r7xx chips (v3) This enables the use of interrupts on r6xx/r7xx hardware. Interrupts are implemented via a ring buffer. The GPU adds interrupts vectors to the ring and the host reads them off in the interrupt handler. The interrupt controller requires firmware like the CP. This firmware must be installed and accessble to the firmware loader for interrupts to function. MSIs don't seem to work on my RS780. They work fine on all my discrete cards. I'm not sure about other RS780s or RS880s. I've disabled MSIs on RS780 and RS880, but it would probably be worth checking on some other systems. v2 - fix some checkpatch.pl problems; re-read the disp int status reg if we restart the ih; v3 - remove the irq handler if r600_irq_init() fails; remove spinlock in r600_ih_ring_fini(); move ih rb overflow check to r600_get_ih_wptr(); move irq ack to separate function; Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-02 02:43:46 +08:00
#define R700_RLC_UCODE_SIZE 1024
drm/radeon/kms/evergreen: setup and enable the CP The command processor (CP) fetches command buffers and feeds the GPU. This patch requires the evergreen family me and pfp ucode files. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-03-25 01:36:43 +08:00
#define EVERGREEN_PFP_UCODE_SIZE 1120
#define EVERGREEN_PM4_UCODE_SIZE 1376
drm/radeon/kms/evergreen: implement irq support Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-03-25 01:55:51 +08:00
#define EVERGREEN_RLC_UCODE_SIZE 768
drm/radeon/kms: add support for cayman irqs Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-03-03 09:07:32 +08:00
#define CAYMAN_RLC_UCODE_SIZE 1024
drm/radeon/kms: add support for ucode loading on trinity (v2) v2: fix check for MC ucode from Tom. Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2012-03-21 05:18:39 +08:00
#define ARUBA_RLC_UCODE_SIZE 1536
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
/* Firmware Names */
MODULE_FIRMWARE("radeon/R600_pfp.bin");
MODULE_FIRMWARE("radeon/R600_me.bin");
MODULE_FIRMWARE("radeon/RV610_pfp.bin");
MODULE_FIRMWARE("radeon/RV610_me.bin");
MODULE_FIRMWARE("radeon/RV630_pfp.bin");
MODULE_FIRMWARE("radeon/RV630_me.bin");
MODULE_FIRMWARE("radeon/RV620_pfp.bin");
MODULE_FIRMWARE("radeon/RV620_me.bin");
MODULE_FIRMWARE("radeon/RV635_pfp.bin");
MODULE_FIRMWARE("radeon/RV635_me.bin");
MODULE_FIRMWARE("radeon/RV670_pfp.bin");
MODULE_FIRMWARE("radeon/RV670_me.bin");
MODULE_FIRMWARE("radeon/RS780_pfp.bin");
MODULE_FIRMWARE("radeon/RS780_me.bin");
MODULE_FIRMWARE("radeon/RV770_pfp.bin");
MODULE_FIRMWARE("radeon/RV770_me.bin");
MODULE_FIRMWARE("radeon/RV730_pfp.bin");
MODULE_FIRMWARE("radeon/RV730_me.bin");
MODULE_FIRMWARE("radeon/RV710_pfp.bin");
MODULE_FIRMWARE("radeon/RV710_me.bin");
drm/radeon/kms: Add support for interrupts on r6xx/r7xx chips (v3) This enables the use of interrupts on r6xx/r7xx hardware. Interrupts are implemented via a ring buffer. The GPU adds interrupts vectors to the ring and the host reads them off in the interrupt handler. The interrupt controller requires firmware like the CP. This firmware must be installed and accessble to the firmware loader for interrupts to function. MSIs don't seem to work on my RS780. They work fine on all my discrete cards. I'm not sure about other RS780s or RS880s. I've disabled MSIs on RS780 and RS880, but it would probably be worth checking on some other systems. v2 - fix some checkpatch.pl problems; re-read the disp int status reg if we restart the ih; v3 - remove the irq handler if r600_irq_init() fails; remove spinlock in r600_ih_ring_fini(); move ih rb overflow check to r600_get_ih_wptr(); move irq ack to separate function; Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-02 02:43:46 +08:00
MODULE_FIRMWARE("radeon/R600_rlc.bin");
MODULE_FIRMWARE("radeon/R700_rlc.bin");
drm/radeon/kms/evergreen: setup and enable the CP The command processor (CP) fetches command buffers and feeds the GPU. This patch requires the evergreen family me and pfp ucode files. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-03-25 01:36:43 +08:00
MODULE_FIRMWARE("radeon/CEDAR_pfp.bin");
MODULE_FIRMWARE("radeon/CEDAR_me.bin");
drm/radeon/kms/evergreen: implement irq support Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-03-25 01:55:51 +08:00
MODULE_FIRMWARE("radeon/CEDAR_rlc.bin");
drm/radeon/kms/evergreen: setup and enable the CP The command processor (CP) fetches command buffers and feeds the GPU. This patch requires the evergreen family me and pfp ucode files. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-03-25 01:36:43 +08:00
MODULE_FIRMWARE("radeon/REDWOOD_pfp.bin");
MODULE_FIRMWARE("radeon/REDWOOD_me.bin");
drm/radeon/kms/evergreen: implement irq support Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-03-25 01:55:51 +08:00
MODULE_FIRMWARE("radeon/REDWOOD_rlc.bin");
drm/radeon/kms/evergreen: setup and enable the CP The command processor (CP) fetches command buffers and feeds the GPU. This patch requires the evergreen family me and pfp ucode files. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-03-25 01:36:43 +08:00
MODULE_FIRMWARE("radeon/JUNIPER_pfp.bin");
MODULE_FIRMWARE("radeon/JUNIPER_me.bin");
drm/radeon/kms/evergreen: implement irq support Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-03-25 01:55:51 +08:00
MODULE_FIRMWARE("radeon/JUNIPER_rlc.bin");
drm/radeon: fix cypress firmware typo. Pointed out by Dave Witbrodt. Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-04-09 13:31:09 +08:00
MODULE_FIRMWARE("radeon/CYPRESS_pfp.bin");
drm/radeon/kms/evergreen: setup and enable the CP The command processor (CP) fetches command buffers and feeds the GPU. This patch requires the evergreen family me and pfp ucode files. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-03-25 01:36:43 +08:00
MODULE_FIRMWARE("radeon/CYPRESS_me.bin");
drm/radeon/kms/evergreen: implement irq support Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-03-25 01:55:51 +08:00
MODULE_FIRMWARE("radeon/CYPRESS_rlc.bin");
drm/radeon/kms: add Ontario APU ucode loading support Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-11-23 06:56:31 +08:00
MODULE_FIRMWARE("radeon/PALM_pfp.bin");
MODULE_FIRMWARE("radeon/PALM_me.bin");
MODULE_FIRMWARE("radeon/SUMO_rlc.bin");
drm/radeon/kms: add support for Llano Fusion APUs - add gpu init support - add blit support - add ucode loader Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-06-01 03:42:48 +08:00
MODULE_FIRMWARE("radeon/SUMO_pfp.bin");
MODULE_FIRMWARE("radeon/SUMO_me.bin");
MODULE_FIRMWARE("radeon/SUMO2_pfp.bin");
MODULE_FIRMWARE("radeon/SUMO2_me.bin");
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
int r600_debugfs_mc_info_init(struct radeon_device *rdev);
drm/radeon: introduce kernel modesetting for radeon hardware Add kernel modesetting support to radeon driver, use the ttm memory manager to manage memory and DRM/GEM to provide userspace API. In order to avoid backward compatibility issue and to allow clean design and code the radeon kernel modesetting use different code path than old radeon/drm driver. When kernel modesetting is enabled the IOCTL of radeon/drm driver are considered as invalid and an error message is printed in the log and they return failure. KMS enabled userspace will use new API to talk with the radeon/drm driver. The new API provide functions to create/destroy/share/mmap buffer object which are then managed by the kernel memory manager (here TTM). In order to submit command to the GPU the userspace provide a buffer holding the command stream, along this buffer userspace have to provide a list of buffer object used by the command stream. The kernel radeon driver will then place buffer in GPU accessible memory and will update command stream to reflect the position of the different buffers. The kernel will also perform security check on command stream provided by the user, we want to catch and forbid any illegal use of the GPU such as DMA into random system memory or into memory not owned by the process supplying the command stream. This part of the code is still incomplete and this why we propose that patch as a staging driver addition, future security might forbid current experimental userspace to run. This code support the following hardware : R1XX,R2XX,R3XX,R4XX,R5XX (radeon up to X1950). Works is underway to provide support for R6XX, R7XX and newer hardware (radeon from HD2XXX to HD4XXX). Authors: Jerome Glisse <jglisse@redhat.com> Dave Airlie <airlied@redhat.com> Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com> Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-06-05 20:42:42 +08:00
drm/radeon/kms: Fix AGP support for R600/RV770 family (v2) For AGP to work unmapped access must cover VRAM & AGP as AGP is treated like VRAM by the GPU (ie physical address). This patch properly setup the virtual memory system aperture to cover AGP if AGP is enabled. It seems that there is memory corruption after resume when using AGP (RV770 seems unaffected thought). Version 2 just fix merge issue with updated AGP fallback patch. Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-10-07 01:04:30 +08:00
/* r600,rv610,rv630,rv620,rv635,rv670 */
drm/radeon: introduce kernel modesetting for radeon hardware Add kernel modesetting support to radeon driver, use the ttm memory manager to manage memory and DRM/GEM to provide userspace API. In order to avoid backward compatibility issue and to allow clean design and code the radeon kernel modesetting use different code path than old radeon/drm driver. When kernel modesetting is enabled the IOCTL of radeon/drm driver are considered as invalid and an error message is printed in the log and they return failure. KMS enabled userspace will use new API to talk with the radeon/drm driver. The new API provide functions to create/destroy/share/mmap buffer object which are then managed by the kernel memory manager (here TTM). In order to submit command to the GPU the userspace provide a buffer holding the command stream, along this buffer userspace have to provide a list of buffer object used by the command stream. The kernel radeon driver will then place buffer in GPU accessible memory and will update command stream to reflect the position of the different buffers. The kernel will also perform security check on command stream provided by the user, we want to catch and forbid any illegal use of the GPU such as DMA into random system memory or into memory not owned by the process supplying the command stream. This part of the code is still incomplete and this why we propose that patch as a staging driver addition, future security might forbid current experimental userspace to run. This code support the following hardware : R1XX,R2XX,R3XX,R4XX,R5XX (radeon up to X1950). Works is underway to provide support for R6XX, R7XX and newer hardware (radeon from HD2XXX to HD4XXX). Authors: Jerome Glisse <jglisse@redhat.com> Dave Airlie <airlied@redhat.com> Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com> Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-06-05 20:42:42 +08:00
int r600_mc_wait_for_idle(struct radeon_device *rdev);
drm/radeon: Mark all possible functions / structs as static Let's allow GCC to optimize better. This exposed some five unused functions, but this patch doesn't remove them. Signed-off-by: Lauri Kasanen <cand@gmx.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2012-09-01 01:43:50 +08:00
static void r600_gpu_init(struct radeon_device *rdev);
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
void r600_fini(struct radeon_device *rdev);
drm/radeon/kms/evergreen: implement irq support Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-03-25 01:55:51 +08:00
void r600_irq_disable(struct radeon_device *rdev);
drm/radeon/kms: add support for gen2 pcie link speeds Supported on rv6xx/r7xx/evergreen. Cards come up in gen1 mode. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-01-07 07:49:35 +08:00
static void r600_pcie_gen2_enable(struct radeon_device *rdev);
drm/radeon: introduce kernel modesetting for radeon hardware Add kernel modesetting support to radeon driver, use the ttm memory manager to manage memory and DRM/GEM to provide userspace API. In order to avoid backward compatibility issue and to allow clean design and code the radeon kernel modesetting use different code path than old radeon/drm driver. When kernel modesetting is enabled the IOCTL of radeon/drm driver are considered as invalid and an error message is printed in the log and they return failure. KMS enabled userspace will use new API to talk with the radeon/drm driver. The new API provide functions to create/destroy/share/mmap buffer object which are then managed by the kernel memory manager (here TTM). In order to submit command to the GPU the userspace provide a buffer holding the command stream, along this buffer userspace have to provide a list of buffer object used by the command stream. The kernel radeon driver will then place buffer in GPU accessible memory and will update command stream to reflect the position of the different buffers. The kernel will also perform security check on command stream provided by the user, we want to catch and forbid any illegal use of the GPU such as DMA into random system memory or into memory not owned by the process supplying the command stream. This part of the code is still incomplete and this why we propose that patch as a staging driver addition, future security might forbid current experimental userspace to run. This code support the following hardware : R1XX,R2XX,R3XX,R4XX,R5XX (radeon up to X1950). Works is underway to provide support for R6XX, R7XX and newer hardware (radeon from HD2XXX to HD4XXX). Authors: Jerome Glisse <jglisse@redhat.com> Dave Airlie <airlied@redhat.com> Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com> Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-06-05 20:42:42 +08:00
drm/radeon/kms: add support for internal thermal sensors (v3) rv6xx/rv7xx/evergreen families supported; older asics did not have an internal thermal sensor. Note, not all oems use the internal thermal sensor, so it's only exposed in cases where it is used. Note also, that most laptops use an oem specific ACPI solution for GPU thermal information rather than using the internal thermal sensor directly. v2: export millidegrees celsius, use hwmon device properly. v3: fix Kconfig Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-07-03 00:58:16 +08:00
/* get temperature in millidegrees */
drm/radeon/kms: rv6xx+ thermal sensor fixes Some fixes to the thermal sensor code: - handle negative numbers - properly handle temp calculation on different asics Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-02-02 05:12:34 +08:00
int rv6xx_get_temp(struct radeon_device *rdev)
drm/radeon/kms: add support for internal thermal sensors (v3) rv6xx/rv7xx/evergreen families supported; older asics did not have an internal thermal sensor. Note, not all oems use the internal thermal sensor, so it's only exposed in cases where it is used. Note also, that most laptops use an oem specific ACPI solution for GPU thermal information rather than using the internal thermal sensor directly. v2: export millidegrees celsius, use hwmon device properly. v3: fix Kconfig Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-07-03 00:58:16 +08:00
{
u32 temp = (RREG32(CG_THERMAL_STATUS) & ASIC_T_MASK) >>
ASIC_T_SHIFT;
drm/radeon/kms: rv6xx+ thermal sensor fixes Some fixes to the thermal sensor code: - handle negative numbers - properly handle temp calculation on different asics Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-02-02 05:12:34 +08:00
int actual_temp = temp & 0xff;
drm/radeon/kms: add support for internal thermal sensors (v3) rv6xx/rv7xx/evergreen families supported; older asics did not have an internal thermal sensor. Note, not all oems use the internal thermal sensor, so it's only exposed in cases where it is used. Note also, that most laptops use an oem specific ACPI solution for GPU thermal information rather than using the internal thermal sensor directly. v2: export millidegrees celsius, use hwmon device properly. v3: fix Kconfig Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-07-03 00:58:16 +08:00
drm/radeon/kms: rv6xx+ thermal sensor fixes Some fixes to the thermal sensor code: - handle negative numbers - properly handle temp calculation on different asics Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-02-02 05:12:34 +08:00
if (temp & 0x100)
actual_temp -= 256;
return actual_temp * 1000;
drm/radeon/kms: add support for internal thermal sensors (v3) rv6xx/rv7xx/evergreen families supported; older asics did not have an internal thermal sensor. Note, not all oems use the internal thermal sensor, so it's only exposed in cases where it is used. Note also, that most laptops use an oem specific ACPI solution for GPU thermal information rather than using the internal thermal sensor directly. v2: export millidegrees celsius, use hwmon device properly. v3: fix Kconfig Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-07-03 00:58:16 +08:00
}
drm/radeon/kms/pm: rework power management - Separate dynpm and profile based power management methods. You can select the pm method by echoing the selected method ("dynpm" or "profile") to power_method in sysfs. - Expose basic 4 profile in profile method "default" - default clocks "auto" - select between low and high based on ac/dc state "low" - DC, low power mode "high" - AC, performance mode The current base profile is "default", but it should switched to "auto" once we've tested on more systems. Switching the state is a matter of echoing the requested profile to power_profile in sysfs. The lowest power states are selected automatically when dpms turns the monitors off in all states but default. - Remove dynamic fence-based reclocking for the moment. We can revisit this later once we have basic pm in. - Move pm init/fini to modesetting path. pm is tightly coupled with display state. Make sure display side is initialized before pm. - Add pm suspend/resume functions to make sure pm state is properly reinitialized on resume. - Remove dynpm module option. It's now selectable via sysfs. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-05-08 03:10:16 +08:00
void r600_pm_get_dynpm_state(struct radeon_device *rdev)
drm/radeon/kms/pm: add asic specific callbacks for getting power state (v2) This also simplifies the code and enables reclocking with multiple heads active by tracking whether the power states are single or multi-head capable. Eventually, we will want to select a power state based on external factors (AC/DC state, user selection, etc.). (v2) Update for evergreen Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-04-23 02:03:55 +08:00
{
int i;
drm/radeon/kms/pm: rework power management - Separate dynpm and profile based power management methods. You can select the pm method by echoing the selected method ("dynpm" or "profile") to power_method in sysfs. - Expose basic 4 profile in profile method "default" - default clocks "auto" - select between low and high based on ac/dc state "low" - DC, low power mode "high" - AC, performance mode The current base profile is "default", but it should switched to "auto" once we've tested on more systems. Switching the state is a matter of echoing the requested profile to power_profile in sysfs. The lowest power states are selected automatically when dpms turns the monitors off in all states but default. - Remove dynamic fence-based reclocking for the moment. We can revisit this later once we have basic pm in. - Move pm init/fini to modesetting path. pm is tightly coupled with display state. Make sure display side is initialized before pm. - Add pm suspend/resume functions to make sure pm state is properly reinitialized on resume. - Remove dynpm module option. It's now selectable via sysfs. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-05-08 03:10:16 +08:00
rdev->pm.dynpm_can_upclock = true;
rdev->pm.dynpm_can_downclock = true;
drm/radeon/kms/pm: add asic specific callbacks for getting power state (v2) This also simplifies the code and enables reclocking with multiple heads active by tracking whether the power states are single or multi-head capable. Eventually, we will want to select a power state based on external factors (AC/DC state, user selection, etc.). (v2) Update for evergreen Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-04-23 02:03:55 +08:00
/* power state array is low to high, default is first */
if ((rdev->flags & RADEON_IS_IGP) || (rdev->family == CHIP_R600)) {
int min_power_state_index = 0;
if (rdev->pm.num_power_states > 2)
min_power_state_index = 1;
drm/radeon/kms/pm: rework power management - Separate dynpm and profile based power management methods. You can select the pm method by echoing the selected method ("dynpm" or "profile") to power_method in sysfs. - Expose basic 4 profile in profile method "default" - default clocks "auto" - select between low and high based on ac/dc state "low" - DC, low power mode "high" - AC, performance mode The current base profile is "default", but it should switched to "auto" once we've tested on more systems. Switching the state is a matter of echoing the requested profile to power_profile in sysfs. The lowest power states are selected automatically when dpms turns the monitors off in all states but default. - Remove dynamic fence-based reclocking for the moment. We can revisit this later once we have basic pm in. - Move pm init/fini to modesetting path. pm is tightly coupled with display state. Make sure display side is initialized before pm. - Add pm suspend/resume functions to make sure pm state is properly reinitialized on resume. - Remove dynpm module option. It's now selectable via sysfs. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-05-08 03:10:16 +08:00
switch (rdev->pm.dynpm_planned_action) {
case DYNPM_ACTION_MINIMUM:
drm/radeon/kms/pm: add asic specific callbacks for getting power state (v2) This also simplifies the code and enables reclocking with multiple heads active by tracking whether the power states are single or multi-head capable. Eventually, we will want to select a power state based on external factors (AC/DC state, user selection, etc.). (v2) Update for evergreen Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-04-23 02:03:55 +08:00
rdev->pm.requested_power_state_index = min_power_state_index;
rdev->pm.requested_clock_mode_index = 0;
drm/radeon/kms/pm: rework power management - Separate dynpm and profile based power management methods. You can select the pm method by echoing the selected method ("dynpm" or "profile") to power_method in sysfs. - Expose basic 4 profile in profile method "default" - default clocks "auto" - select between low and high based on ac/dc state "low" - DC, low power mode "high" - AC, performance mode The current base profile is "default", but it should switched to "auto" once we've tested on more systems. Switching the state is a matter of echoing the requested profile to power_profile in sysfs. The lowest power states are selected automatically when dpms turns the monitors off in all states but default. - Remove dynamic fence-based reclocking for the moment. We can revisit this later once we have basic pm in. - Move pm init/fini to modesetting path. pm is tightly coupled with display state. Make sure display side is initialized before pm. - Add pm suspend/resume functions to make sure pm state is properly reinitialized on resume. - Remove dynpm module option. It's now selectable via sysfs. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-05-08 03:10:16 +08:00
rdev->pm.dynpm_can_downclock = false;
drm/radeon/kms/pm: add asic specific callbacks for getting power state (v2) This also simplifies the code and enables reclocking with multiple heads active by tracking whether the power states are single or multi-head capable. Eventually, we will want to select a power state based on external factors (AC/DC state, user selection, etc.). (v2) Update for evergreen Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-04-23 02:03:55 +08:00
break;
drm/radeon/kms/pm: rework power management - Separate dynpm and profile based power management methods. You can select the pm method by echoing the selected method ("dynpm" or "profile") to power_method in sysfs. - Expose basic 4 profile in profile method "default" - default clocks "auto" - select between low and high based on ac/dc state "low" - DC, low power mode "high" - AC, performance mode The current base profile is "default", but it should switched to "auto" once we've tested on more systems. Switching the state is a matter of echoing the requested profile to power_profile in sysfs. The lowest power states are selected automatically when dpms turns the monitors off in all states but default. - Remove dynamic fence-based reclocking for the moment. We can revisit this later once we have basic pm in. - Move pm init/fini to modesetting path. pm is tightly coupled with display state. Make sure display side is initialized before pm. - Add pm suspend/resume functions to make sure pm state is properly reinitialized on resume. - Remove dynpm module option. It's now selectable via sysfs. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-05-08 03:10:16 +08:00
case DYNPM_ACTION_DOWNCLOCK:
drm/radeon/kms/pm: add asic specific callbacks for getting power state (v2) This also simplifies the code and enables reclocking with multiple heads active by tracking whether the power states are single or multi-head capable. Eventually, we will want to select a power state based on external factors (AC/DC state, user selection, etc.). (v2) Update for evergreen Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-04-23 02:03:55 +08:00
if (rdev->pm.current_power_state_index == min_power_state_index) {
rdev->pm.requested_power_state_index = rdev->pm.current_power_state_index;
drm/radeon/kms/pm: rework power management - Separate dynpm and profile based power management methods. You can select the pm method by echoing the selected method ("dynpm" or "profile") to power_method in sysfs. - Expose basic 4 profile in profile method "default" - default clocks "auto" - select between low and high based on ac/dc state "low" - DC, low power mode "high" - AC, performance mode The current base profile is "default", but it should switched to "auto" once we've tested on more systems. Switching the state is a matter of echoing the requested profile to power_profile in sysfs. The lowest power states are selected automatically when dpms turns the monitors off in all states but default. - Remove dynamic fence-based reclocking for the moment. We can revisit this later once we have basic pm in. - Move pm init/fini to modesetting path. pm is tightly coupled with display state. Make sure display side is initialized before pm. - Add pm suspend/resume functions to make sure pm state is properly reinitialized on resume. - Remove dynpm module option. It's now selectable via sysfs. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-05-08 03:10:16 +08:00
rdev->pm.dynpm_can_downclock = false;
drm/radeon/kms/pm: add asic specific callbacks for getting power state (v2) This also simplifies the code and enables reclocking with multiple heads active by tracking whether the power states are single or multi-head capable. Eventually, we will want to select a power state based on external factors (AC/DC state, user selection, etc.). (v2) Update for evergreen Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-04-23 02:03:55 +08:00
} else {
if (rdev->pm.active_crtc_count > 1) {
for (i = 0; i < rdev->pm.num_power_states; i++) {
drm/radeon/kms/pm: add support for no display power states The lowest power states often cause display problems, so only enable them when all displays are off. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-05-03 13:13:14 +08:00
if (rdev->pm.power_state[i].flags & RADEON_PM_STATE_SINGLE_DISPLAY_ONLY)
drm/radeon/kms/pm: add asic specific callbacks for getting power state (v2) This also simplifies the code and enables reclocking with multiple heads active by tracking whether the power states are single or multi-head capable. Eventually, we will want to select a power state based on external factors (AC/DC state, user selection, etc.). (v2) Update for evergreen Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-04-23 02:03:55 +08:00
continue;
else if (i >= rdev->pm.current_power_state_index) {
rdev->pm.requested_power_state_index =
rdev->pm.current_power_state_index;
break;
} else {
rdev->pm.requested_power_state_index = i;
break;
}
}
drm/radeon/kms/pm: fix power state indexing on igp chips in dynpm mode Fixes: https://bugs.freedesktop.org/show_bug.cgi?id=28745 Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Tested-by: Rafał Miłecki <zajec5@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-06-26 04:21:27 +08:00
} else {
if (rdev->pm.current_power_state_index == 0)
rdev->pm.requested_power_state_index =
rdev->pm.num_power_states - 1;
else
rdev->pm.requested_power_state_index =
rdev->pm.current_power_state_index - 1;
}
drm/radeon/kms/pm: add asic specific callbacks for getting power state (v2) This also simplifies the code and enables reclocking with multiple heads active by tracking whether the power states are single or multi-head capable. Eventually, we will want to select a power state based on external factors (AC/DC state, user selection, etc.). (v2) Update for evergreen Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-04-23 02:03:55 +08:00
}
rdev->pm.requested_clock_mode_index = 0;
drm/radeon/kms/pm: add support for no display power states The lowest power states often cause display problems, so only enable them when all displays are off. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-05-03 13:13:14 +08:00
/* don't use the power state if crtcs are active and no display flag is set */
if ((rdev->pm.active_crtc_count > 0) &&
(rdev->pm.power_state[rdev->pm.requested_power_state_index].
clock_info[rdev->pm.requested_clock_mode_index].flags &
RADEON_PM_MODE_NO_DISPLAY)) {
rdev->pm.requested_power_state_index++;
}
drm/radeon/kms/pm: add asic specific callbacks for getting power state (v2) This also simplifies the code and enables reclocking with multiple heads active by tracking whether the power states are single or multi-head capable. Eventually, we will want to select a power state based on external factors (AC/DC state, user selection, etc.). (v2) Update for evergreen Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-04-23 02:03:55 +08:00
break;
drm/radeon/kms/pm: rework power management - Separate dynpm and profile based power management methods. You can select the pm method by echoing the selected method ("dynpm" or "profile") to power_method in sysfs. - Expose basic 4 profile in profile method "default" - default clocks "auto" - select between low and high based on ac/dc state "low" - DC, low power mode "high" - AC, performance mode The current base profile is "default", but it should switched to "auto" once we've tested on more systems. Switching the state is a matter of echoing the requested profile to power_profile in sysfs. The lowest power states are selected automatically when dpms turns the monitors off in all states but default. - Remove dynamic fence-based reclocking for the moment. We can revisit this later once we have basic pm in. - Move pm init/fini to modesetting path. pm is tightly coupled with display state. Make sure display side is initialized before pm. - Add pm suspend/resume functions to make sure pm state is properly reinitialized on resume. - Remove dynpm module option. It's now selectable via sysfs. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-05-08 03:10:16 +08:00
case DYNPM_ACTION_UPCLOCK:
drm/radeon/kms/pm: add asic specific callbacks for getting power state (v2) This also simplifies the code and enables reclocking with multiple heads active by tracking whether the power states are single or multi-head capable. Eventually, we will want to select a power state based on external factors (AC/DC state, user selection, etc.). (v2) Update for evergreen Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-04-23 02:03:55 +08:00
if (rdev->pm.current_power_state_index == (rdev->pm.num_power_states - 1)) {
rdev->pm.requested_power_state_index = rdev->pm.current_power_state_index;
drm/radeon/kms/pm: rework power management - Separate dynpm and profile based power management methods. You can select the pm method by echoing the selected method ("dynpm" or "profile") to power_method in sysfs. - Expose basic 4 profile in profile method "default" - default clocks "auto" - select between low and high based on ac/dc state "low" - DC, low power mode "high" - AC, performance mode The current base profile is "default", but it should switched to "auto" once we've tested on more systems. Switching the state is a matter of echoing the requested profile to power_profile in sysfs. The lowest power states are selected automatically when dpms turns the monitors off in all states but default. - Remove dynamic fence-based reclocking for the moment. We can revisit this later once we have basic pm in. - Move pm init/fini to modesetting path. pm is tightly coupled with display state. Make sure display side is initialized before pm. - Add pm suspend/resume functions to make sure pm state is properly reinitialized on resume. - Remove dynpm module option. It's now selectable via sysfs. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-05-08 03:10:16 +08:00
rdev->pm.dynpm_can_upclock = false;
drm/radeon/kms/pm: add asic specific callbacks for getting power state (v2) This also simplifies the code and enables reclocking with multiple heads active by tracking whether the power states are single or multi-head capable. Eventually, we will want to select a power state based on external factors (AC/DC state, user selection, etc.). (v2) Update for evergreen Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-04-23 02:03:55 +08:00
} else {
if (rdev->pm.active_crtc_count > 1) {
for (i = (rdev->pm.num_power_states - 1); i >= 0; i--) {
drm/radeon/kms/pm: add support for no display power states The lowest power states often cause display problems, so only enable them when all displays are off. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-05-03 13:13:14 +08:00
if (rdev->pm.power_state[i].flags & RADEON_PM_STATE_SINGLE_DISPLAY_ONLY)
drm/radeon/kms/pm: add asic specific callbacks for getting power state (v2) This also simplifies the code and enables reclocking with multiple heads active by tracking whether the power states are single or multi-head capable. Eventually, we will want to select a power state based on external factors (AC/DC state, user selection, etc.). (v2) Update for evergreen Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-04-23 02:03:55 +08:00
continue;
else if (i <= rdev->pm.current_power_state_index) {
rdev->pm.requested_power_state_index =
rdev->pm.current_power_state_index;
break;
} else {
rdev->pm.requested_power_state_index = i;
break;
}
}
} else
rdev->pm.requested_power_state_index =
rdev->pm.current_power_state_index + 1;
}
rdev->pm.requested_clock_mode_index = 0;
break;
drm/radeon/kms/pm: rework power management - Separate dynpm and profile based power management methods. You can select the pm method by echoing the selected method ("dynpm" or "profile") to power_method in sysfs. - Expose basic 4 profile in profile method "default" - default clocks "auto" - select between low and high based on ac/dc state "low" - DC, low power mode "high" - AC, performance mode The current base profile is "default", but it should switched to "auto" once we've tested on more systems. Switching the state is a matter of echoing the requested profile to power_profile in sysfs. The lowest power states are selected automatically when dpms turns the monitors off in all states but default. - Remove dynamic fence-based reclocking for the moment. We can revisit this later once we have basic pm in. - Move pm init/fini to modesetting path. pm is tightly coupled with display state. Make sure display side is initialized before pm. - Add pm suspend/resume functions to make sure pm state is properly reinitialized on resume. - Remove dynpm module option. It's now selectable via sysfs. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-05-08 03:10:16 +08:00
case DYNPM_ACTION_DEFAULT:
drm/radeon/kms/pm: restore default power state on exit Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-03-23 01:31:08 +08:00
rdev->pm.requested_power_state_index = rdev->pm.default_power_state_index;
rdev->pm.requested_clock_mode_index = 0;
drm/radeon/kms/pm: rework power management - Separate dynpm and profile based power management methods. You can select the pm method by echoing the selected method ("dynpm" or "profile") to power_method in sysfs. - Expose basic 4 profile in profile method "default" - default clocks "auto" - select between low and high based on ac/dc state "low" - DC, low power mode "high" - AC, performance mode The current base profile is "default", but it should switched to "auto" once we've tested on more systems. Switching the state is a matter of echoing the requested profile to power_profile in sysfs. The lowest power states are selected automatically when dpms turns the monitors off in all states but default. - Remove dynamic fence-based reclocking for the moment. We can revisit this later once we have basic pm in. - Move pm init/fini to modesetting path. pm is tightly coupled with display state. Make sure display side is initialized before pm. - Add pm suspend/resume functions to make sure pm state is properly reinitialized on resume. - Remove dynpm module option. It's now selectable via sysfs. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-05-08 03:10:16 +08:00
rdev->pm.dynpm_can_upclock = false;
drm/radeon/kms/pm: restore default power state on exit Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-03-23 01:31:08 +08:00
break;
drm/radeon/kms/pm: rework power management - Separate dynpm and profile based power management methods. You can select the pm method by echoing the selected method ("dynpm" or "profile") to power_method in sysfs. - Expose basic 4 profile in profile method "default" - default clocks "auto" - select between low and high based on ac/dc state "low" - DC, low power mode "high" - AC, performance mode The current base profile is "default", but it should switched to "auto" once we've tested on more systems. Switching the state is a matter of echoing the requested profile to power_profile in sysfs. The lowest power states are selected automatically when dpms turns the monitors off in all states but default. - Remove dynamic fence-based reclocking for the moment. We can revisit this later once we have basic pm in. - Move pm init/fini to modesetting path. pm is tightly coupled with display state. Make sure display side is initialized before pm. - Add pm suspend/resume functions to make sure pm state is properly reinitialized on resume. - Remove dynpm module option. It's now selectable via sysfs. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-05-08 03:10:16 +08:00
case DYNPM_ACTION_NONE:
drm/radeon/kms/pm: add asic specific callbacks for getting power state (v2) This also simplifies the code and enables reclocking with multiple heads active by tracking whether the power states are single or multi-head capable. Eventually, we will want to select a power state based on external factors (AC/DC state, user selection, etc.). (v2) Update for evergreen Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-04-23 02:03:55 +08:00
default:
DRM_ERROR("Requested mode for not defined action\n");
return;
}
} else {
/* XXX select a power state based on AC/DC, single/dualhead, etc. */
/* for now just select the first power state and switch between clock modes */
/* power state array is low to high, default is first (0) */
if (rdev->pm.active_crtc_count > 1) {
rdev->pm.requested_power_state_index = -1;
/* start at 1 as we don't want the default mode */
for (i = 1; i < rdev->pm.num_power_states; i++) {
drm/radeon/kms/pm: add support for no display power states The lowest power states often cause display problems, so only enable them when all displays are off. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-05-03 13:13:14 +08:00
if (rdev->pm.power_state[i].flags & RADEON_PM_STATE_SINGLE_DISPLAY_ONLY)
drm/radeon/kms/pm: add asic specific callbacks for getting power state (v2) This also simplifies the code and enables reclocking with multiple heads active by tracking whether the power states are single or multi-head capable. Eventually, we will want to select a power state based on external factors (AC/DC state, user selection, etc.). (v2) Update for evergreen Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-04-23 02:03:55 +08:00
continue;
else if ((rdev->pm.power_state[i].type == POWER_STATE_TYPE_PERFORMANCE) ||
(rdev->pm.power_state[i].type == POWER_STATE_TYPE_BATTERY)) {
rdev->pm.requested_power_state_index = i;
break;
}
}
/* if nothing selected, grab the default state. */
if (rdev->pm.requested_power_state_index == -1)
rdev->pm.requested_power_state_index = 0;
} else
rdev->pm.requested_power_state_index = 1;
drm/radeon/kms/pm: rework power management - Separate dynpm and profile based power management methods. You can select the pm method by echoing the selected method ("dynpm" or "profile") to power_method in sysfs. - Expose basic 4 profile in profile method "default" - default clocks "auto" - select between low and high based on ac/dc state "low" - DC, low power mode "high" - AC, performance mode The current base profile is "default", but it should switched to "auto" once we've tested on more systems. Switching the state is a matter of echoing the requested profile to power_profile in sysfs. The lowest power states are selected automatically when dpms turns the monitors off in all states but default. - Remove dynamic fence-based reclocking for the moment. We can revisit this later once we have basic pm in. - Move pm init/fini to modesetting path. pm is tightly coupled with display state. Make sure display side is initialized before pm. - Add pm suspend/resume functions to make sure pm state is properly reinitialized on resume. - Remove dynpm module option. It's now selectable via sysfs. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-05-08 03:10:16 +08:00
switch (rdev->pm.dynpm_planned_action) {
case DYNPM_ACTION_MINIMUM:
drm/radeon/kms/pm: add asic specific callbacks for getting power state (v2) This also simplifies the code and enables reclocking with multiple heads active by tracking whether the power states are single or multi-head capable. Eventually, we will want to select a power state based on external factors (AC/DC state, user selection, etc.). (v2) Update for evergreen Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-04-23 02:03:55 +08:00
rdev->pm.requested_clock_mode_index = 0;
drm/radeon/kms/pm: rework power management - Separate dynpm and profile based power management methods. You can select the pm method by echoing the selected method ("dynpm" or "profile") to power_method in sysfs. - Expose basic 4 profile in profile method "default" - default clocks "auto" - select between low and high based on ac/dc state "low" - DC, low power mode "high" - AC, performance mode The current base profile is "default", but it should switched to "auto" once we've tested on more systems. Switching the state is a matter of echoing the requested profile to power_profile in sysfs. The lowest power states are selected automatically when dpms turns the monitors off in all states but default. - Remove dynamic fence-based reclocking for the moment. We can revisit this later once we have basic pm in. - Move pm init/fini to modesetting path. pm is tightly coupled with display state. Make sure display side is initialized before pm. - Add pm suspend/resume functions to make sure pm state is properly reinitialized on resume. - Remove dynpm module option. It's now selectable via sysfs. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-05-08 03:10:16 +08:00
rdev->pm.dynpm_can_downclock = false;
drm/radeon/kms/pm: add asic specific callbacks for getting power state (v2) This also simplifies the code and enables reclocking with multiple heads active by tracking whether the power states are single or multi-head capable. Eventually, we will want to select a power state based on external factors (AC/DC state, user selection, etc.). (v2) Update for evergreen Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-04-23 02:03:55 +08:00
break;
drm/radeon/kms/pm: rework power management - Separate dynpm and profile based power management methods. You can select the pm method by echoing the selected method ("dynpm" or "profile") to power_method in sysfs. - Expose basic 4 profile in profile method "default" - default clocks "auto" - select between low and high based on ac/dc state "low" - DC, low power mode "high" - AC, performance mode The current base profile is "default", but it should switched to "auto" once we've tested on more systems. Switching the state is a matter of echoing the requested profile to power_profile in sysfs. The lowest power states are selected automatically when dpms turns the monitors off in all states but default. - Remove dynamic fence-based reclocking for the moment. We can revisit this later once we have basic pm in. - Move pm init/fini to modesetting path. pm is tightly coupled with display state. Make sure display side is initialized before pm. - Add pm suspend/resume functions to make sure pm state is properly reinitialized on resume. - Remove dynpm module option. It's now selectable via sysfs. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-05-08 03:10:16 +08:00
case DYNPM_ACTION_DOWNCLOCK:
drm/radeon/kms/pm: add asic specific callbacks for getting power state (v2) This also simplifies the code and enables reclocking with multiple heads active by tracking whether the power states are single or multi-head capable. Eventually, we will want to select a power state based on external factors (AC/DC state, user selection, etc.). (v2) Update for evergreen Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-04-23 02:03:55 +08:00
if (rdev->pm.requested_power_state_index == rdev->pm.current_power_state_index) {
if (rdev->pm.current_clock_mode_index == 0) {
rdev->pm.requested_clock_mode_index = 0;
drm/radeon/kms/pm: rework power management - Separate dynpm and profile based power management methods. You can select the pm method by echoing the selected method ("dynpm" or "profile") to power_method in sysfs. - Expose basic 4 profile in profile method "default" - default clocks "auto" - select between low and high based on ac/dc state "low" - DC, low power mode "high" - AC, performance mode The current base profile is "default", but it should switched to "auto" once we've tested on more systems. Switching the state is a matter of echoing the requested profile to power_profile in sysfs. The lowest power states are selected automatically when dpms turns the monitors off in all states but default. - Remove dynamic fence-based reclocking for the moment. We can revisit this later once we have basic pm in. - Move pm init/fini to modesetting path. pm is tightly coupled with display state. Make sure display side is initialized before pm. - Add pm suspend/resume functions to make sure pm state is properly reinitialized on resume. - Remove dynpm module option. It's now selectable via sysfs. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-05-08 03:10:16 +08:00
rdev->pm.dynpm_can_downclock = false;
drm/radeon/kms/pm: add asic specific callbacks for getting power state (v2) This also simplifies the code and enables reclocking with multiple heads active by tracking whether the power states are single or multi-head capable. Eventually, we will want to select a power state based on external factors (AC/DC state, user selection, etc.). (v2) Update for evergreen Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-04-23 02:03:55 +08:00
} else
rdev->pm.requested_clock_mode_index =
rdev->pm.current_clock_mode_index - 1;
} else {
rdev->pm.requested_clock_mode_index = 0;
drm/radeon/kms/pm: rework power management - Separate dynpm and profile based power management methods. You can select the pm method by echoing the selected method ("dynpm" or "profile") to power_method in sysfs. - Expose basic 4 profile in profile method "default" - default clocks "auto" - select between low and high based on ac/dc state "low" - DC, low power mode "high" - AC, performance mode The current base profile is "default", but it should switched to "auto" once we've tested on more systems. Switching the state is a matter of echoing the requested profile to power_profile in sysfs. The lowest power states are selected automatically when dpms turns the monitors off in all states but default. - Remove dynamic fence-based reclocking for the moment. We can revisit this later once we have basic pm in. - Move pm init/fini to modesetting path. pm is tightly coupled with display state. Make sure display side is initialized before pm. - Add pm suspend/resume functions to make sure pm state is properly reinitialized on resume. - Remove dynpm module option. It's now selectable via sysfs. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-05-08 03:10:16 +08:00
rdev->pm.dynpm_can_downclock = false;
drm/radeon/kms/pm: add asic specific callbacks for getting power state (v2) This also simplifies the code and enables reclocking with multiple heads active by tracking whether the power states are single or multi-head capable. Eventually, we will want to select a power state based on external factors (AC/DC state, user selection, etc.). (v2) Update for evergreen Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-04-23 02:03:55 +08:00
}
drm/radeon/kms/pm: add support for no display power states The lowest power states often cause display problems, so only enable them when all displays are off. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-05-03 13:13:14 +08:00
/* don't use the power state if crtcs are active and no display flag is set */
if ((rdev->pm.active_crtc_count > 0) &&
(rdev->pm.power_state[rdev->pm.requested_power_state_index].
clock_info[rdev->pm.requested_clock_mode_index].flags &
RADEON_PM_MODE_NO_DISPLAY)) {
rdev->pm.requested_clock_mode_index++;
}
drm/radeon/kms/pm: add asic specific callbacks for getting power state (v2) This also simplifies the code and enables reclocking with multiple heads active by tracking whether the power states are single or multi-head capable. Eventually, we will want to select a power state based on external factors (AC/DC state, user selection, etc.). (v2) Update for evergreen Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-04-23 02:03:55 +08:00
break;
drm/radeon/kms/pm: rework power management - Separate dynpm and profile based power management methods. You can select the pm method by echoing the selected method ("dynpm" or "profile") to power_method in sysfs. - Expose basic 4 profile in profile method "default" - default clocks "auto" - select between low and high based on ac/dc state "low" - DC, low power mode "high" - AC, performance mode The current base profile is "default", but it should switched to "auto" once we've tested on more systems. Switching the state is a matter of echoing the requested profile to power_profile in sysfs. The lowest power states are selected automatically when dpms turns the monitors off in all states but default. - Remove dynamic fence-based reclocking for the moment. We can revisit this later once we have basic pm in. - Move pm init/fini to modesetting path. pm is tightly coupled with display state. Make sure display side is initialized before pm. - Add pm suspend/resume functions to make sure pm state is properly reinitialized on resume. - Remove dynpm module option. It's now selectable via sysfs. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-05-08 03:10:16 +08:00
case DYNPM_ACTION_UPCLOCK:
drm/radeon/kms/pm: add asic specific callbacks for getting power state (v2) This also simplifies the code and enables reclocking with multiple heads active by tracking whether the power states are single or multi-head capable. Eventually, we will want to select a power state based on external factors (AC/DC state, user selection, etc.). (v2) Update for evergreen Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-04-23 02:03:55 +08:00
if (rdev->pm.requested_power_state_index == rdev->pm.current_power_state_index) {
if (rdev->pm.current_clock_mode_index ==
(rdev->pm.power_state[rdev->pm.requested_power_state_index].num_clock_modes - 1)) {
rdev->pm.requested_clock_mode_index = rdev->pm.current_clock_mode_index;
drm/radeon/kms/pm: rework power management - Separate dynpm and profile based power management methods. You can select the pm method by echoing the selected method ("dynpm" or "profile") to power_method in sysfs. - Expose basic 4 profile in profile method "default" - default clocks "auto" - select between low and high based on ac/dc state "low" - DC, low power mode "high" - AC, performance mode The current base profile is "default", but it should switched to "auto" once we've tested on more systems. Switching the state is a matter of echoing the requested profile to power_profile in sysfs. The lowest power states are selected automatically when dpms turns the monitors off in all states but default. - Remove dynamic fence-based reclocking for the moment. We can revisit this later once we have basic pm in. - Move pm init/fini to modesetting path. pm is tightly coupled with display state. Make sure display side is initialized before pm. - Add pm suspend/resume functions to make sure pm state is properly reinitialized on resume. - Remove dynpm module option. It's now selectable via sysfs. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-05-08 03:10:16 +08:00
rdev->pm.dynpm_can_upclock = false;
drm/radeon/kms/pm: add asic specific callbacks for getting power state (v2) This also simplifies the code and enables reclocking with multiple heads active by tracking whether the power states are single or multi-head capable. Eventually, we will want to select a power state based on external factors (AC/DC state, user selection, etc.). (v2) Update for evergreen Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-04-23 02:03:55 +08:00
} else
rdev->pm.requested_clock_mode_index =
rdev->pm.current_clock_mode_index + 1;
} else {
rdev->pm.requested_clock_mode_index =
rdev->pm.power_state[rdev->pm.requested_power_state_index].num_clock_modes - 1;
drm/radeon/kms/pm: rework power management - Separate dynpm and profile based power management methods. You can select the pm method by echoing the selected method ("dynpm" or "profile") to power_method in sysfs. - Expose basic 4 profile in profile method "default" - default clocks "auto" - select between low and high based on ac/dc state "low" - DC, low power mode "high" - AC, performance mode The current base profile is "default", but it should switched to "auto" once we've tested on more systems. Switching the state is a matter of echoing the requested profile to power_profile in sysfs. The lowest power states are selected automatically when dpms turns the monitors off in all states but default. - Remove dynamic fence-based reclocking for the moment. We can revisit this later once we have basic pm in. - Move pm init/fini to modesetting path. pm is tightly coupled with display state. Make sure display side is initialized before pm. - Add pm suspend/resume functions to make sure pm state is properly reinitialized on resume. - Remove dynpm module option. It's now selectable via sysfs. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-05-08 03:10:16 +08:00
rdev->pm.dynpm_can_upclock = false;
drm/radeon/kms/pm: add asic specific callbacks for getting power state (v2) This also simplifies the code and enables reclocking with multiple heads active by tracking whether the power states are single or multi-head capable. Eventually, we will want to select a power state based on external factors (AC/DC state, user selection, etc.). (v2) Update for evergreen Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-04-23 02:03:55 +08:00
}
break;
drm/radeon/kms/pm: rework power management - Separate dynpm and profile based power management methods. You can select the pm method by echoing the selected method ("dynpm" or "profile") to power_method in sysfs. - Expose basic 4 profile in profile method "default" - default clocks "auto" - select between low and high based on ac/dc state "low" - DC, low power mode "high" - AC, performance mode The current base profile is "default", but it should switched to "auto" once we've tested on more systems. Switching the state is a matter of echoing the requested profile to power_profile in sysfs. The lowest power states are selected automatically when dpms turns the monitors off in all states but default. - Remove dynamic fence-based reclocking for the moment. We can revisit this later once we have basic pm in. - Move pm init/fini to modesetting path. pm is tightly coupled with display state. Make sure display side is initialized before pm. - Add pm suspend/resume functions to make sure pm state is properly reinitialized on resume. - Remove dynpm module option. It's now selectable via sysfs. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-05-08 03:10:16 +08:00
case DYNPM_ACTION_DEFAULT:
drm/radeon/kms/pm: restore default power state on exit Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-03-23 01:31:08 +08:00
rdev->pm.requested_power_state_index = rdev->pm.default_power_state_index;
rdev->pm.requested_clock_mode_index = 0;
drm/radeon/kms/pm: rework power management - Separate dynpm and profile based power management methods. You can select the pm method by echoing the selected method ("dynpm" or "profile") to power_method in sysfs. - Expose basic 4 profile in profile method "default" - default clocks "auto" - select between low and high based on ac/dc state "low" - DC, low power mode "high" - AC, performance mode The current base profile is "default", but it should switched to "auto" once we've tested on more systems. Switching the state is a matter of echoing the requested profile to power_profile in sysfs. The lowest power states are selected automatically when dpms turns the monitors off in all states but default. - Remove dynamic fence-based reclocking for the moment. We can revisit this later once we have basic pm in. - Move pm init/fini to modesetting path. pm is tightly coupled with display state. Make sure display side is initialized before pm. - Add pm suspend/resume functions to make sure pm state is properly reinitialized on resume. - Remove dynpm module option. It's now selectable via sysfs. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-05-08 03:10:16 +08:00
rdev->pm.dynpm_can_upclock = false;
drm/radeon/kms/pm: restore default power state on exit Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-03-23 01:31:08 +08:00
break;
drm/radeon/kms/pm: rework power management - Separate dynpm and profile based power management methods. You can select the pm method by echoing the selected method ("dynpm" or "profile") to power_method in sysfs. - Expose basic 4 profile in profile method "default" - default clocks "auto" - select between low and high based on ac/dc state "low" - DC, low power mode "high" - AC, performance mode The current base profile is "default", but it should switched to "auto" once we've tested on more systems. Switching the state is a matter of echoing the requested profile to power_profile in sysfs. The lowest power states are selected automatically when dpms turns the monitors off in all states but default. - Remove dynamic fence-based reclocking for the moment. We can revisit this later once we have basic pm in. - Move pm init/fini to modesetting path. pm is tightly coupled with display state. Make sure display side is initialized before pm. - Add pm suspend/resume functions to make sure pm state is properly reinitialized on resume. - Remove dynpm module option. It's now selectable via sysfs. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-05-08 03:10:16 +08:00
case DYNPM_ACTION_NONE:
drm/radeon/kms/pm: add asic specific callbacks for getting power state (v2) This also simplifies the code and enables reclocking with multiple heads active by tracking whether the power states are single or multi-head capable. Eventually, we will want to select a power state based on external factors (AC/DC state, user selection, etc.). (v2) Update for evergreen Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-04-23 02:03:55 +08:00
default:
DRM_ERROR("Requested mode for not defined action\n");
return;
}
}
drm/radeon/kms: move a bunch of modesetting debug to correct debug usage. This migrates a bunch of DRM_DEBUG->DRM_DEBUG_KMS so we can get more modesetting related info without all the other ioctl handling easily. Also the PM code moves to DRM_DEBUG_DRIVER mostly. Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-08-02 08:42:55 +08:00
DRM_DEBUG_DRIVER("Requested: e: %d m: %d p: %d\n",
drm/radeon/kms/pm: make pm spam debug only Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-05-08 04:58:27 +08:00
rdev->pm.power_state[rdev->pm.requested_power_state_index].
clock_info[rdev->pm.requested_clock_mode_index].sclk,
rdev->pm.power_state[rdev->pm.requested_power_state_index].
clock_info[rdev->pm.requested_clock_mode_index].mclk,
rdev->pm.power_state[rdev->pm.requested_power_state_index].
pcie_lanes);
drm/radeon/kms/pm: add asic specific callbacks for getting power state (v2) This also simplifies the code and enables reclocking with multiple heads active by tracking whether the power states are single or multi-head capable. Eventually, we will want to select a power state based on external factors (AC/DC state, user selection, etc.). (v2) Update for evergreen Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-04-23 02:03:55 +08:00
}
drm/radeon/kms/pm: rework power management - Separate dynpm and profile based power management methods. You can select the pm method by echoing the selected method ("dynpm" or "profile") to power_method in sysfs. - Expose basic 4 profile in profile method "default" - default clocks "auto" - select between low and high based on ac/dc state "low" - DC, low power mode "high" - AC, performance mode The current base profile is "default", but it should switched to "auto" once we've tested on more systems. Switching the state is a matter of echoing the requested profile to power_profile in sysfs. The lowest power states are selected automatically when dpms turns the monitors off in all states but default. - Remove dynamic fence-based reclocking for the moment. We can revisit this later once we have basic pm in. - Move pm init/fini to modesetting path. pm is tightly coupled with display state. Make sure display side is initialized before pm. - Add pm suspend/resume functions to make sure pm state is properly reinitialized on resume. - Remove dynpm module option. It's now selectable via sysfs. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-05-08 03:10:16 +08:00
void rs780_pm_init_profile(struct radeon_device *rdev)
{
if (rdev->pm.num_power_states == 2) {
/* default */
rdev->pm.profiles[PM_PROFILE_DEFAULT_IDX].dpms_off_ps_idx = rdev->pm.default_power_state_index;
rdev->pm.profiles[PM_PROFILE_DEFAULT_IDX].dpms_on_ps_idx = rdev->pm.default_power_state_index;
rdev->pm.profiles[PM_PROFILE_DEFAULT_IDX].dpms_off_cm_idx = 0;
rdev->pm.profiles[PM_PROFILE_DEFAULT_IDX].dpms_on_cm_idx = 0;
/* low sh */
rdev->pm.profiles[PM_PROFILE_LOW_SH_IDX].dpms_off_ps_idx = 0;
rdev->pm.profiles[PM_PROFILE_LOW_SH_IDX].dpms_on_ps_idx = 0;
rdev->pm.profiles[PM_PROFILE_LOW_SH_IDX].dpms_off_cm_idx = 0;
rdev->pm.profiles[PM_PROFILE_LOW_SH_IDX].dpms_on_cm_idx = 0;
drm/radeon/kms/pm: add mid profile This adds an additional profile, mid, to the pm profile code which takes the place of the old low profile. The default behavior remains the same, e.g., auto profile now selects between mid and high profiles based on power source, however, you can now manually force the low profile which was previously only available as a dpms off state. Enabling the low profile when the displays are on has been known to cause display corruption in some cases. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-06-03 05:56:01 +08:00
/* mid sh */
rdev->pm.profiles[PM_PROFILE_MID_SH_IDX].dpms_off_ps_idx = 0;
rdev->pm.profiles[PM_PROFILE_MID_SH_IDX].dpms_on_ps_idx = 0;
rdev->pm.profiles[PM_PROFILE_MID_SH_IDX].dpms_off_cm_idx = 0;
rdev->pm.profiles[PM_PROFILE_MID_SH_IDX].dpms_on_cm_idx = 0;
drm/radeon/kms/pm: rework power management - Separate dynpm and profile based power management methods. You can select the pm method by echoing the selected method ("dynpm" or "profile") to power_method in sysfs. - Expose basic 4 profile in profile method "default" - default clocks "auto" - select between low and high based on ac/dc state "low" - DC, low power mode "high" - AC, performance mode The current base profile is "default", but it should switched to "auto" once we've tested on more systems. Switching the state is a matter of echoing the requested profile to power_profile in sysfs. The lowest power states are selected automatically when dpms turns the monitors off in all states but default. - Remove dynamic fence-based reclocking for the moment. We can revisit this later once we have basic pm in. - Move pm init/fini to modesetting path. pm is tightly coupled with display state. Make sure display side is initialized before pm. - Add pm suspend/resume functions to make sure pm state is properly reinitialized on resume. - Remove dynpm module option. It's now selectable via sysfs. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-05-08 03:10:16 +08:00
/* high sh */
rdev->pm.profiles[PM_PROFILE_HIGH_SH_IDX].dpms_off_ps_idx = 0;
rdev->pm.profiles[PM_PROFILE_HIGH_SH_IDX].dpms_on_ps_idx = 1;
rdev->pm.profiles[PM_PROFILE_HIGH_SH_IDX].dpms_off_cm_idx = 0;
rdev->pm.profiles[PM_PROFILE_HIGH_SH_IDX].dpms_on_cm_idx = 0;
/* low mh */
rdev->pm.profiles[PM_PROFILE_LOW_MH_IDX].dpms_off_ps_idx = 0;
rdev->pm.profiles[PM_PROFILE_LOW_MH_IDX].dpms_on_ps_idx = 0;
rdev->pm.profiles[PM_PROFILE_LOW_MH_IDX].dpms_off_cm_idx = 0;
rdev->pm.profiles[PM_PROFILE_LOW_MH_IDX].dpms_on_cm_idx = 0;
drm/radeon/kms/pm: add mid profile This adds an additional profile, mid, to the pm profile code which takes the place of the old low profile. The default behavior remains the same, e.g., auto profile now selects between mid and high profiles based on power source, however, you can now manually force the low profile which was previously only available as a dpms off state. Enabling the low profile when the displays are on has been known to cause display corruption in some cases. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-06-03 05:56:01 +08:00
/* mid mh */
rdev->pm.profiles[PM_PROFILE_MID_MH_IDX].dpms_off_ps_idx = 0;
rdev->pm.profiles[PM_PROFILE_MID_MH_IDX].dpms_on_ps_idx = 0;
rdev->pm.profiles[PM_PROFILE_MID_MH_IDX].dpms_off_cm_idx = 0;
rdev->pm.profiles[PM_PROFILE_MID_MH_IDX].dpms_on_cm_idx = 0;
drm/radeon/kms/pm: rework power management - Separate dynpm and profile based power management methods. You can select the pm method by echoing the selected method ("dynpm" or "profile") to power_method in sysfs. - Expose basic 4 profile in profile method "default" - default clocks "auto" - select between low and high based on ac/dc state "low" - DC, low power mode "high" - AC, performance mode The current base profile is "default", but it should switched to "auto" once we've tested on more systems. Switching the state is a matter of echoing the requested profile to power_profile in sysfs. The lowest power states are selected automatically when dpms turns the monitors off in all states but default. - Remove dynamic fence-based reclocking for the moment. We can revisit this later once we have basic pm in. - Move pm init/fini to modesetting path. pm is tightly coupled with display state. Make sure display side is initialized before pm. - Add pm suspend/resume functions to make sure pm state is properly reinitialized on resume. - Remove dynpm module option. It's now selectable via sysfs. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-05-08 03:10:16 +08:00
/* high mh */
rdev->pm.profiles[PM_PROFILE_HIGH_MH_IDX].dpms_off_ps_idx = 0;
rdev->pm.profiles[PM_PROFILE_HIGH_MH_IDX].dpms_on_ps_idx = 1;
rdev->pm.profiles[PM_PROFILE_HIGH_MH_IDX].dpms_off_cm_idx = 0;
rdev->pm.profiles[PM_PROFILE_HIGH_MH_IDX].dpms_on_cm_idx = 0;
} else if (rdev->pm.num_power_states == 3) {
/* default */
rdev->pm.profiles[PM_PROFILE_DEFAULT_IDX].dpms_off_ps_idx = rdev->pm.default_power_state_index;
rdev->pm.profiles[PM_PROFILE_DEFAULT_IDX].dpms_on_ps_idx = rdev->pm.default_power_state_index;
rdev->pm.profiles[PM_PROFILE_DEFAULT_IDX].dpms_off_cm_idx = 0;
rdev->pm.profiles[PM_PROFILE_DEFAULT_IDX].dpms_on_cm_idx = 0;
/* low sh */
rdev->pm.profiles[PM_PROFILE_LOW_SH_IDX].dpms_off_ps_idx = 1;
rdev->pm.profiles[PM_PROFILE_LOW_SH_IDX].dpms_on_ps_idx = 1;
rdev->pm.profiles[PM_PROFILE_LOW_SH_IDX].dpms_off_cm_idx = 0;
rdev->pm.profiles[PM_PROFILE_LOW_SH_IDX].dpms_on_cm_idx = 0;
drm/radeon/kms/pm: add mid profile This adds an additional profile, mid, to the pm profile code which takes the place of the old low profile. The default behavior remains the same, e.g., auto profile now selects between mid and high profiles based on power source, however, you can now manually force the low profile which was previously only available as a dpms off state. Enabling the low profile when the displays are on has been known to cause display corruption in some cases. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-06-03 05:56:01 +08:00
/* mid sh */
rdev->pm.profiles[PM_PROFILE_MID_SH_IDX].dpms_off_ps_idx = 1;
rdev->pm.profiles[PM_PROFILE_MID_SH_IDX].dpms_on_ps_idx = 1;
rdev->pm.profiles[PM_PROFILE_MID_SH_IDX].dpms_off_cm_idx = 0;
rdev->pm.profiles[PM_PROFILE_MID_SH_IDX].dpms_on_cm_idx = 0;
drm/radeon/kms/pm: rework power management - Separate dynpm and profile based power management methods. You can select the pm method by echoing the selected method ("dynpm" or "profile") to power_method in sysfs. - Expose basic 4 profile in profile method "default" - default clocks "auto" - select between low and high based on ac/dc state "low" - DC, low power mode "high" - AC, performance mode The current base profile is "default", but it should switched to "auto" once we've tested on more systems. Switching the state is a matter of echoing the requested profile to power_profile in sysfs. The lowest power states are selected automatically when dpms turns the monitors off in all states but default. - Remove dynamic fence-based reclocking for the moment. We can revisit this later once we have basic pm in. - Move pm init/fini to modesetting path. pm is tightly coupled with display state. Make sure display side is initialized before pm. - Add pm suspend/resume functions to make sure pm state is properly reinitialized on resume. - Remove dynpm module option. It's now selectable via sysfs. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-05-08 03:10:16 +08:00
/* high sh */
rdev->pm.profiles[PM_PROFILE_HIGH_SH_IDX].dpms_off_ps_idx = 1;
rdev->pm.profiles[PM_PROFILE_HIGH_SH_IDX].dpms_on_ps_idx = 2;
rdev->pm.profiles[PM_PROFILE_HIGH_SH_IDX].dpms_off_cm_idx = 0;
rdev->pm.profiles[PM_PROFILE_HIGH_SH_IDX].dpms_on_cm_idx = 0;
/* low mh */
rdev->pm.profiles[PM_PROFILE_LOW_MH_IDX].dpms_off_ps_idx = 1;
rdev->pm.profiles[PM_PROFILE_LOW_MH_IDX].dpms_on_ps_idx = 1;
rdev->pm.profiles[PM_PROFILE_LOW_MH_IDX].dpms_off_cm_idx = 0;
rdev->pm.profiles[PM_PROFILE_LOW_MH_IDX].dpms_on_cm_idx = 0;
drm/radeon/kms/pm: add mid profile This adds an additional profile, mid, to the pm profile code which takes the place of the old low profile. The default behavior remains the same, e.g., auto profile now selects between mid and high profiles based on power source, however, you can now manually force the low profile which was previously only available as a dpms off state. Enabling the low profile when the displays are on has been known to cause display corruption in some cases. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-06-03 05:56:01 +08:00
/* mid mh */
rdev->pm.profiles[PM_PROFILE_MID_MH_IDX].dpms_off_ps_idx = 1;
rdev->pm.profiles[PM_PROFILE_MID_MH_IDX].dpms_on_ps_idx = 1;
rdev->pm.profiles[PM_PROFILE_MID_MH_IDX].dpms_off_cm_idx = 0;
rdev->pm.profiles[PM_PROFILE_MID_MH_IDX].dpms_on_cm_idx = 0;
drm/radeon/kms/pm: rework power management - Separate dynpm and profile based power management methods. You can select the pm method by echoing the selected method ("dynpm" or "profile") to power_method in sysfs. - Expose basic 4 profile in profile method "default" - default clocks "auto" - select between low and high based on ac/dc state "low" - DC, low power mode "high" - AC, performance mode The current base profile is "default", but it should switched to "auto" once we've tested on more systems. Switching the state is a matter of echoing the requested profile to power_profile in sysfs. The lowest power states are selected automatically when dpms turns the monitors off in all states but default. - Remove dynamic fence-based reclocking for the moment. We can revisit this later once we have basic pm in. - Move pm init/fini to modesetting path. pm is tightly coupled with display state. Make sure display side is initialized before pm. - Add pm suspend/resume functions to make sure pm state is properly reinitialized on resume. - Remove dynpm module option. It's now selectable via sysfs. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-05-08 03:10:16 +08:00
/* high mh */
rdev->pm.profiles[PM_PROFILE_HIGH_MH_IDX].dpms_off_ps_idx = 1;
rdev->pm.profiles[PM_PROFILE_HIGH_MH_IDX].dpms_on_ps_idx = 2;
rdev->pm.profiles[PM_PROFILE_HIGH_MH_IDX].dpms_off_cm_idx = 0;
rdev->pm.profiles[PM_PROFILE_HIGH_MH_IDX].dpms_on_cm_idx = 0;
} else {
/* default */
rdev->pm.profiles[PM_PROFILE_DEFAULT_IDX].dpms_off_ps_idx = rdev->pm.default_power_state_index;
rdev->pm.profiles[PM_PROFILE_DEFAULT_IDX].dpms_on_ps_idx = rdev->pm.default_power_state_index;
rdev->pm.profiles[PM_PROFILE_DEFAULT_IDX].dpms_off_cm_idx = 0;
rdev->pm.profiles[PM_PROFILE_DEFAULT_IDX].dpms_on_cm_idx = 0;
/* low sh */
rdev->pm.profiles[PM_PROFILE_LOW_SH_IDX].dpms_off_ps_idx = 2;
rdev->pm.profiles[PM_PROFILE_LOW_SH_IDX].dpms_on_ps_idx = 2;
rdev->pm.profiles[PM_PROFILE_LOW_SH_IDX].dpms_off_cm_idx = 0;
rdev->pm.profiles[PM_PROFILE_LOW_SH_IDX].dpms_on_cm_idx = 0;
drm/radeon/kms/pm: add mid profile This adds an additional profile, mid, to the pm profile code which takes the place of the old low profile. The default behavior remains the same, e.g., auto profile now selects between mid and high profiles based on power source, however, you can now manually force the low profile which was previously only available as a dpms off state. Enabling the low profile when the displays are on has been known to cause display corruption in some cases. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-06-03 05:56:01 +08:00
/* mid sh */
rdev->pm.profiles[PM_PROFILE_MID_SH_IDX].dpms_off_ps_idx = 2;
rdev->pm.profiles[PM_PROFILE_MID_SH_IDX].dpms_on_ps_idx = 2;
rdev->pm.profiles[PM_PROFILE_MID_SH_IDX].dpms_off_cm_idx = 0;
rdev->pm.profiles[PM_PROFILE_MID_SH_IDX].dpms_on_cm_idx = 0;
drm/radeon/kms/pm: rework power management - Separate dynpm and profile based power management methods. You can select the pm method by echoing the selected method ("dynpm" or "profile") to power_method in sysfs. - Expose basic 4 profile in profile method "default" - default clocks "auto" - select between low and high based on ac/dc state "low" - DC, low power mode "high" - AC, performance mode The current base profile is "default", but it should switched to "auto" once we've tested on more systems. Switching the state is a matter of echoing the requested profile to power_profile in sysfs. The lowest power states are selected automatically when dpms turns the monitors off in all states but default. - Remove dynamic fence-based reclocking for the moment. We can revisit this later once we have basic pm in. - Move pm init/fini to modesetting path. pm is tightly coupled with display state. Make sure display side is initialized before pm. - Add pm suspend/resume functions to make sure pm state is properly reinitialized on resume. - Remove dynpm module option. It's now selectable via sysfs. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-05-08 03:10:16 +08:00
/* high sh */
rdev->pm.profiles[PM_PROFILE_HIGH_SH_IDX].dpms_off_ps_idx = 2;
rdev->pm.profiles[PM_PROFILE_HIGH_SH_IDX].dpms_on_ps_idx = 3;
rdev->pm.profiles[PM_PROFILE_HIGH_SH_IDX].dpms_off_cm_idx = 0;
rdev->pm.profiles[PM_PROFILE_HIGH_SH_IDX].dpms_on_cm_idx = 0;
/* low mh */
rdev->pm.profiles[PM_PROFILE_LOW_MH_IDX].dpms_off_ps_idx = 2;
rdev->pm.profiles[PM_PROFILE_LOW_MH_IDX].dpms_on_ps_idx = 0;
rdev->pm.profiles[PM_PROFILE_LOW_MH_IDX].dpms_off_cm_idx = 0;
rdev->pm.profiles[PM_PROFILE_LOW_MH_IDX].dpms_on_cm_idx = 0;
drm/radeon/kms/pm: add mid profile This adds an additional profile, mid, to the pm profile code which takes the place of the old low profile. The default behavior remains the same, e.g., auto profile now selects between mid and high profiles based on power source, however, you can now manually force the low profile which was previously only available as a dpms off state. Enabling the low profile when the displays are on has been known to cause display corruption in some cases. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-06-03 05:56:01 +08:00
/* mid mh */
rdev->pm.profiles[PM_PROFILE_MID_MH_IDX].dpms_off_ps_idx = 2;
rdev->pm.profiles[PM_PROFILE_MID_MH_IDX].dpms_on_ps_idx = 0;
rdev->pm.profiles[PM_PROFILE_MID_MH_IDX].dpms_off_cm_idx = 0;
rdev->pm.profiles[PM_PROFILE_MID_MH_IDX].dpms_on_cm_idx = 0;
drm/radeon/kms/pm: rework power management - Separate dynpm and profile based power management methods. You can select the pm method by echoing the selected method ("dynpm" or "profile") to power_method in sysfs. - Expose basic 4 profile in profile method "default" - default clocks "auto" - select between low and high based on ac/dc state "low" - DC, low power mode "high" - AC, performance mode The current base profile is "default", but it should switched to "auto" once we've tested on more systems. Switching the state is a matter of echoing the requested profile to power_profile in sysfs. The lowest power states are selected automatically when dpms turns the monitors off in all states but default. - Remove dynamic fence-based reclocking for the moment. We can revisit this later once we have basic pm in. - Move pm init/fini to modesetting path. pm is tightly coupled with display state. Make sure display side is initialized before pm. - Add pm suspend/resume functions to make sure pm state is properly reinitialized on resume. - Remove dynpm module option. It's now selectable via sysfs. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-05-08 03:10:16 +08:00
/* high mh */
rdev->pm.profiles[PM_PROFILE_HIGH_MH_IDX].dpms_off_ps_idx = 2;
rdev->pm.profiles[PM_PROFILE_HIGH_MH_IDX].dpms_on_ps_idx = 3;
rdev->pm.profiles[PM_PROFILE_HIGH_MH_IDX].dpms_off_cm_idx = 0;
rdev->pm.profiles[PM_PROFILE_HIGH_MH_IDX].dpms_on_cm_idx = 0;
}
}
drm/radeon/kms/pm: add asic specific callbacks for setting power state (v2) (v2) Add evergreen vbl checks Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-04-23 01:38:05 +08:00
drm/radeon/kms/pm: rework power management - Separate dynpm and profile based power management methods. You can select the pm method by echoing the selected method ("dynpm" or "profile") to power_method in sysfs. - Expose basic 4 profile in profile method "default" - default clocks "auto" - select between low and high based on ac/dc state "low" - DC, low power mode "high" - AC, performance mode The current base profile is "default", but it should switched to "auto" once we've tested on more systems. Switching the state is a matter of echoing the requested profile to power_profile in sysfs. The lowest power states are selected automatically when dpms turns the monitors off in all states but default. - Remove dynamic fence-based reclocking for the moment. We can revisit this later once we have basic pm in. - Move pm init/fini to modesetting path. pm is tightly coupled with display state. Make sure display side is initialized before pm. - Add pm suspend/resume functions to make sure pm state is properly reinitialized on resume. - Remove dynpm module option. It's now selectable via sysfs. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-05-08 03:10:16 +08:00
void r600_pm_init_profile(struct radeon_device *rdev)
{
drm/radeon/kms: optimize r600_pm_profile_init Avoid a lot of extra loops through the pm state array. Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-11-04 22:09:42 +08:00
int idx;
drm/radeon/kms/pm: rework power management - Separate dynpm and profile based power management methods. You can select the pm method by echoing the selected method ("dynpm" or "profile") to power_method in sysfs. - Expose basic 4 profile in profile method "default" - default clocks "auto" - select between low and high based on ac/dc state "low" - DC, low power mode "high" - AC, performance mode The current base profile is "default", but it should switched to "auto" once we've tested on more systems. Switching the state is a matter of echoing the requested profile to power_profile in sysfs. The lowest power states are selected automatically when dpms turns the monitors off in all states but default. - Remove dynamic fence-based reclocking for the moment. We can revisit this later once we have basic pm in. - Move pm init/fini to modesetting path. pm is tightly coupled with display state. Make sure display side is initialized before pm. - Add pm suspend/resume functions to make sure pm state is properly reinitialized on resume. - Remove dynpm module option. It's now selectable via sysfs. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-05-08 03:10:16 +08:00
if (rdev->family == CHIP_R600) {
/* XXX */
/* default */
rdev->pm.profiles[PM_PROFILE_DEFAULT_IDX].dpms_off_ps_idx = rdev->pm.default_power_state_index;
rdev->pm.profiles[PM_PROFILE_DEFAULT_IDX].dpms_on_ps_idx = rdev->pm.default_power_state_index;
rdev->pm.profiles[PM_PROFILE_DEFAULT_IDX].dpms_off_cm_idx = 0;
drm/radeon/kms/pm: fix r6xx+ profile setup This patch is a combination of the previous two profile patches, but without the index bugs. It cleans up and fixes some issues with pm profile setup on r6xx chips. Some tables have different orderings for the power states, also, r600 only has 1 clock mode per power state. On desktop cards there are no battery modes, so the low and high power states are the same. For the low profile case, choose the lower clock mode. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-05-18 07:41:26 +08:00
rdev->pm.profiles[PM_PROFILE_DEFAULT_IDX].dpms_on_cm_idx = 0;
drm/radeon/kms/pm: rework power management - Separate dynpm and profile based power management methods. You can select the pm method by echoing the selected method ("dynpm" or "profile") to power_method in sysfs. - Expose basic 4 profile in profile method "default" - default clocks "auto" - select between low and high based on ac/dc state "low" - DC, low power mode "high" - AC, performance mode The current base profile is "default", but it should switched to "auto" once we've tested on more systems. Switching the state is a matter of echoing the requested profile to power_profile in sysfs. The lowest power states are selected automatically when dpms turns the monitors off in all states but default. - Remove dynamic fence-based reclocking for the moment. We can revisit this later once we have basic pm in. - Move pm init/fini to modesetting path. pm is tightly coupled with display state. Make sure display side is initialized before pm. - Add pm suspend/resume functions to make sure pm state is properly reinitialized on resume. - Remove dynpm module option. It's now selectable via sysfs. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-05-08 03:10:16 +08:00
/* low sh */
rdev->pm.profiles[PM_PROFILE_LOW_SH_IDX].dpms_off_ps_idx = rdev->pm.default_power_state_index;
rdev->pm.profiles[PM_PROFILE_LOW_SH_IDX].dpms_on_ps_idx = rdev->pm.default_power_state_index;
rdev->pm.profiles[PM_PROFILE_LOW_SH_IDX].dpms_off_cm_idx = 0;
drm/radeon/kms/pm: fix r6xx+ profile setup This patch is a combination of the previous two profile patches, but without the index bugs. It cleans up and fixes some issues with pm profile setup on r6xx chips. Some tables have different orderings for the power states, also, r600 only has 1 clock mode per power state. On desktop cards there are no battery modes, so the low and high power states are the same. For the low profile case, choose the lower clock mode. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-05-18 07:41:26 +08:00
rdev->pm.profiles[PM_PROFILE_LOW_SH_IDX].dpms_on_cm_idx = 0;
drm/radeon/kms/pm: add mid profile This adds an additional profile, mid, to the pm profile code which takes the place of the old low profile. The default behavior remains the same, e.g., auto profile now selects between mid and high profiles based on power source, however, you can now manually force the low profile which was previously only available as a dpms off state. Enabling the low profile when the displays are on has been known to cause display corruption in some cases. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-06-03 05:56:01 +08:00
/* mid sh */
rdev->pm.profiles[PM_PROFILE_MID_SH_IDX].dpms_off_ps_idx = rdev->pm.default_power_state_index;
rdev->pm.profiles[PM_PROFILE_MID_SH_IDX].dpms_on_ps_idx = rdev->pm.default_power_state_index;
rdev->pm.profiles[PM_PROFILE_MID_SH_IDX].dpms_off_cm_idx = 0;
rdev->pm.profiles[PM_PROFILE_MID_SH_IDX].dpms_on_cm_idx = 0;
drm/radeon/kms/pm: rework power management - Separate dynpm and profile based power management methods. You can select the pm method by echoing the selected method ("dynpm" or "profile") to power_method in sysfs. - Expose basic 4 profile in profile method "default" - default clocks "auto" - select between low and high based on ac/dc state "low" - DC, low power mode "high" - AC, performance mode The current base profile is "default", but it should switched to "auto" once we've tested on more systems. Switching the state is a matter of echoing the requested profile to power_profile in sysfs. The lowest power states are selected automatically when dpms turns the monitors off in all states but default. - Remove dynamic fence-based reclocking for the moment. We can revisit this later once we have basic pm in. - Move pm init/fini to modesetting path. pm is tightly coupled with display state. Make sure display side is initialized before pm. - Add pm suspend/resume functions to make sure pm state is properly reinitialized on resume. - Remove dynpm module option. It's now selectable via sysfs. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-05-08 03:10:16 +08:00
/* high sh */
rdev->pm.profiles[PM_PROFILE_HIGH_SH_IDX].dpms_off_ps_idx = rdev->pm.default_power_state_index;
rdev->pm.profiles[PM_PROFILE_HIGH_SH_IDX].dpms_on_ps_idx = rdev->pm.default_power_state_index;
rdev->pm.profiles[PM_PROFILE_HIGH_SH_IDX].dpms_off_cm_idx = 0;
drm/radeon/kms/pm: fix r6xx+ profile setup This patch is a combination of the previous two profile patches, but without the index bugs. It cleans up and fixes some issues with pm profile setup on r6xx chips. Some tables have different orderings for the power states, also, r600 only has 1 clock mode per power state. On desktop cards there are no battery modes, so the low and high power states are the same. For the low profile case, choose the lower clock mode. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-05-18 07:41:26 +08:00
rdev->pm.profiles[PM_PROFILE_HIGH_SH_IDX].dpms_on_cm_idx = 0;
drm/radeon/kms/pm: rework power management - Separate dynpm and profile based power management methods. You can select the pm method by echoing the selected method ("dynpm" or "profile") to power_method in sysfs. - Expose basic 4 profile in profile method "default" - default clocks "auto" - select between low and high based on ac/dc state "low" - DC, low power mode "high" - AC, performance mode The current base profile is "default", but it should switched to "auto" once we've tested on more systems. Switching the state is a matter of echoing the requested profile to power_profile in sysfs. The lowest power states are selected automatically when dpms turns the monitors off in all states but default. - Remove dynamic fence-based reclocking for the moment. We can revisit this later once we have basic pm in. - Move pm init/fini to modesetting path. pm is tightly coupled with display state. Make sure display side is initialized before pm. - Add pm suspend/resume functions to make sure pm state is properly reinitialized on resume. - Remove dynpm module option. It's now selectable via sysfs. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-05-08 03:10:16 +08:00
/* low mh */
rdev->pm.profiles[PM_PROFILE_LOW_MH_IDX].dpms_off_ps_idx = rdev->pm.default_power_state_index;
rdev->pm.profiles[PM_PROFILE_LOW_MH_IDX].dpms_on_ps_idx = rdev->pm.default_power_state_index;
rdev->pm.profiles[PM_PROFILE_LOW_MH_IDX].dpms_off_cm_idx = 0;
drm/radeon/kms/pm: fix r6xx+ profile setup This patch is a combination of the previous two profile patches, but without the index bugs. It cleans up and fixes some issues with pm profile setup on r6xx chips. Some tables have different orderings for the power states, also, r600 only has 1 clock mode per power state. On desktop cards there are no battery modes, so the low and high power states are the same. For the low profile case, choose the lower clock mode. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-05-18 07:41:26 +08:00
rdev->pm.profiles[PM_PROFILE_LOW_MH_IDX].dpms_on_cm_idx = 0;
drm/radeon/kms/pm: add mid profile This adds an additional profile, mid, to the pm profile code which takes the place of the old low profile. The default behavior remains the same, e.g., auto profile now selects between mid and high profiles based on power source, however, you can now manually force the low profile which was previously only available as a dpms off state. Enabling the low profile when the displays are on has been known to cause display corruption in some cases. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-06-03 05:56:01 +08:00
/* mid mh */
rdev->pm.profiles[PM_PROFILE_MID_MH_IDX].dpms_off_ps_idx = rdev->pm.default_power_state_index;
rdev->pm.profiles[PM_PROFILE_MID_MH_IDX].dpms_on_ps_idx = rdev->pm.default_power_state_index;
rdev->pm.profiles[PM_PROFILE_MID_MH_IDX].dpms_off_cm_idx = 0;
rdev->pm.profiles[PM_PROFILE_MID_MH_IDX].dpms_on_cm_idx = 0;
drm/radeon/kms/pm: rework power management - Separate dynpm and profile based power management methods. You can select the pm method by echoing the selected method ("dynpm" or "profile") to power_method in sysfs. - Expose basic 4 profile in profile method "default" - default clocks "auto" - select between low and high based on ac/dc state "low" - DC, low power mode "high" - AC, performance mode The current base profile is "default", but it should switched to "auto" once we've tested on more systems. Switching the state is a matter of echoing the requested profile to power_profile in sysfs. The lowest power states are selected automatically when dpms turns the monitors off in all states but default. - Remove dynamic fence-based reclocking for the moment. We can revisit this later once we have basic pm in. - Move pm init/fini to modesetting path. pm is tightly coupled with display state. Make sure display side is initialized before pm. - Add pm suspend/resume functions to make sure pm state is properly reinitialized on resume. - Remove dynpm module option. It's now selectable via sysfs. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-05-08 03:10:16 +08:00
/* high mh */
rdev->pm.profiles[PM_PROFILE_HIGH_MH_IDX].dpms_off_ps_idx = rdev->pm.default_power_state_index;
rdev->pm.profiles[PM_PROFILE_HIGH_MH_IDX].dpms_on_ps_idx = rdev->pm.default_power_state_index;
rdev->pm.profiles[PM_PROFILE_HIGH_MH_IDX].dpms_off_cm_idx = 0;
drm/radeon/kms/pm: fix r6xx+ profile setup This patch is a combination of the previous two profile patches, but without the index bugs. It cleans up and fixes some issues with pm profile setup on r6xx chips. Some tables have different orderings for the power states, also, r600 only has 1 clock mode per power state. On desktop cards there are no battery modes, so the low and high power states are the same. For the low profile case, choose the lower clock mode. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-05-18 07:41:26 +08:00
rdev->pm.profiles[PM_PROFILE_HIGH_MH_IDX].dpms_on_cm_idx = 0;
drm/radeon/kms/pm: rework power management - Separate dynpm and profile based power management methods. You can select the pm method by echoing the selected method ("dynpm" or "profile") to power_method in sysfs. - Expose basic 4 profile in profile method "default" - default clocks "auto" - select between low and high based on ac/dc state "low" - DC, low power mode "high" - AC, performance mode The current base profile is "default", but it should switched to "auto" once we've tested on more systems. Switching the state is a matter of echoing the requested profile to power_profile in sysfs. The lowest power states are selected automatically when dpms turns the monitors off in all states but default. - Remove dynamic fence-based reclocking for the moment. We can revisit this later once we have basic pm in. - Move pm init/fini to modesetting path. pm is tightly coupled with display state. Make sure display side is initialized before pm. - Add pm suspend/resume functions to make sure pm state is properly reinitialized on resume. - Remove dynpm module option. It's now selectable via sysfs. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-05-08 03:10:16 +08:00
} else {
if (rdev->pm.num_power_states < 4) {
/* default */
rdev->pm.profiles[PM_PROFILE_DEFAULT_IDX].dpms_off_ps_idx = rdev->pm.default_power_state_index;
rdev->pm.profiles[PM_PROFILE_DEFAULT_IDX].dpms_on_ps_idx = rdev->pm.default_power_state_index;
rdev->pm.profiles[PM_PROFILE_DEFAULT_IDX].dpms_off_cm_idx = 0;
rdev->pm.profiles[PM_PROFILE_DEFAULT_IDX].dpms_on_cm_idx = 2;
/* low sh */
drm/radeon/kms/pm: fix r6xx+ profile setup This patch is a combination of the previous two profile patches, but without the index bugs. It cleans up and fixes some issues with pm profile setup on r6xx chips. Some tables have different orderings for the power states, also, r600 only has 1 clock mode per power state. On desktop cards there are no battery modes, so the low and high power states are the same. For the low profile case, choose the lower clock mode. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-05-18 07:41:26 +08:00
rdev->pm.profiles[PM_PROFILE_LOW_SH_IDX].dpms_off_ps_idx = 1;
rdev->pm.profiles[PM_PROFILE_LOW_SH_IDX].dpms_on_ps_idx = 1;
drm/radeon/kms/pm: rework power management - Separate dynpm and profile based power management methods. You can select the pm method by echoing the selected method ("dynpm" or "profile") to power_method in sysfs. - Expose basic 4 profile in profile method "default" - default clocks "auto" - select between low and high based on ac/dc state "low" - DC, low power mode "high" - AC, performance mode The current base profile is "default", but it should switched to "auto" once we've tested on more systems. Switching the state is a matter of echoing the requested profile to power_profile in sysfs. The lowest power states are selected automatically when dpms turns the monitors off in all states but default. - Remove dynamic fence-based reclocking for the moment. We can revisit this later once we have basic pm in. - Move pm init/fini to modesetting path. pm is tightly coupled with display state. Make sure display side is initialized before pm. - Add pm suspend/resume functions to make sure pm state is properly reinitialized on resume. - Remove dynpm module option. It's now selectable via sysfs. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-05-08 03:10:16 +08:00
rdev->pm.profiles[PM_PROFILE_LOW_SH_IDX].dpms_off_cm_idx = 0;
drm/radeon/kms/pm: add mid profile This adds an additional profile, mid, to the pm profile code which takes the place of the old low profile. The default behavior remains the same, e.g., auto profile now selects between mid and high profiles based on power source, however, you can now manually force the low profile which was previously only available as a dpms off state. Enabling the low profile when the displays are on has been known to cause display corruption in some cases. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-06-03 05:56:01 +08:00
rdev->pm.profiles[PM_PROFILE_LOW_SH_IDX].dpms_on_cm_idx = 0;
/* mid sh */
rdev->pm.profiles[PM_PROFILE_MID_SH_IDX].dpms_off_ps_idx = 1;
rdev->pm.profiles[PM_PROFILE_MID_SH_IDX].dpms_on_ps_idx = 1;
rdev->pm.profiles[PM_PROFILE_MID_SH_IDX].dpms_off_cm_idx = 0;
rdev->pm.profiles[PM_PROFILE_MID_SH_IDX].dpms_on_cm_idx = 1;
drm/radeon/kms/pm: rework power management - Separate dynpm and profile based power management methods. You can select the pm method by echoing the selected method ("dynpm" or "profile") to power_method in sysfs. - Expose basic 4 profile in profile method "default" - default clocks "auto" - select between low and high based on ac/dc state "low" - DC, low power mode "high" - AC, performance mode The current base profile is "default", but it should switched to "auto" once we've tested on more systems. Switching the state is a matter of echoing the requested profile to power_profile in sysfs. The lowest power states are selected automatically when dpms turns the monitors off in all states but default. - Remove dynamic fence-based reclocking for the moment. We can revisit this later once we have basic pm in. - Move pm init/fini to modesetting path. pm is tightly coupled with display state. Make sure display side is initialized before pm. - Add pm suspend/resume functions to make sure pm state is properly reinitialized on resume. - Remove dynpm module option. It's now selectable via sysfs. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-05-08 03:10:16 +08:00
/* high sh */
drm/radeon/kms/pm: fix r6xx+ profile setup This patch is a combination of the previous two profile patches, but without the index bugs. It cleans up and fixes some issues with pm profile setup on r6xx chips. Some tables have different orderings for the power states, also, r600 only has 1 clock mode per power state. On desktop cards there are no battery modes, so the low and high power states are the same. For the low profile case, choose the lower clock mode. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-05-18 07:41:26 +08:00
rdev->pm.profiles[PM_PROFILE_HIGH_SH_IDX].dpms_off_ps_idx = 1;
rdev->pm.profiles[PM_PROFILE_HIGH_SH_IDX].dpms_on_ps_idx = 1;
drm/radeon/kms/pm: rework power management - Separate dynpm and profile based power management methods. You can select the pm method by echoing the selected method ("dynpm" or "profile") to power_method in sysfs. - Expose basic 4 profile in profile method "default" - default clocks "auto" - select between low and high based on ac/dc state "low" - DC, low power mode "high" - AC, performance mode The current base profile is "default", but it should switched to "auto" once we've tested on more systems. Switching the state is a matter of echoing the requested profile to power_profile in sysfs. The lowest power states are selected automatically when dpms turns the monitors off in all states but default. - Remove dynamic fence-based reclocking for the moment. We can revisit this later once we have basic pm in. - Move pm init/fini to modesetting path. pm is tightly coupled with display state. Make sure display side is initialized before pm. - Add pm suspend/resume functions to make sure pm state is properly reinitialized on resume. - Remove dynpm module option. It's now selectable via sysfs. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-05-08 03:10:16 +08:00
rdev->pm.profiles[PM_PROFILE_HIGH_SH_IDX].dpms_off_cm_idx = 0;
rdev->pm.profiles[PM_PROFILE_HIGH_SH_IDX].dpms_on_cm_idx = 2;
/* low mh */
drm/radeon/kms/pm: fix r6xx+ profile setup This patch is a combination of the previous two profile patches, but without the index bugs. It cleans up and fixes some issues with pm profile setup on r6xx chips. Some tables have different orderings for the power states, also, r600 only has 1 clock mode per power state. On desktop cards there are no battery modes, so the low and high power states are the same. For the low profile case, choose the lower clock mode. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-05-18 07:41:26 +08:00
rdev->pm.profiles[PM_PROFILE_LOW_MH_IDX].dpms_off_ps_idx = 2;
rdev->pm.profiles[PM_PROFILE_LOW_MH_IDX].dpms_on_ps_idx = 2;
drm/radeon/kms/pm: rework power management - Separate dynpm and profile based power management methods. You can select the pm method by echoing the selected method ("dynpm" or "profile") to power_method in sysfs. - Expose basic 4 profile in profile method "default" - default clocks "auto" - select between low and high based on ac/dc state "low" - DC, low power mode "high" - AC, performance mode The current base profile is "default", but it should switched to "auto" once we've tested on more systems. Switching the state is a matter of echoing the requested profile to power_profile in sysfs. The lowest power states are selected automatically when dpms turns the monitors off in all states but default. - Remove dynamic fence-based reclocking for the moment. We can revisit this later once we have basic pm in. - Move pm init/fini to modesetting path. pm is tightly coupled with display state. Make sure display side is initialized before pm. - Add pm suspend/resume functions to make sure pm state is properly reinitialized on resume. - Remove dynpm module option. It's now selectable via sysfs. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-05-08 03:10:16 +08:00
rdev->pm.profiles[PM_PROFILE_LOW_MH_IDX].dpms_off_cm_idx = 0;
drm/radeon/kms/pm: add mid profile This adds an additional profile, mid, to the pm profile code which takes the place of the old low profile. The default behavior remains the same, e.g., auto profile now selects between mid and high profiles based on power source, however, you can now manually force the low profile which was previously only available as a dpms off state. Enabling the low profile when the displays are on has been known to cause display corruption in some cases. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-06-03 05:56:01 +08:00
rdev->pm.profiles[PM_PROFILE_LOW_MH_IDX].dpms_on_cm_idx = 0;
/* low mh */
rdev->pm.profiles[PM_PROFILE_MID_MH_IDX].dpms_off_ps_idx = 2;
rdev->pm.profiles[PM_PROFILE_MID_MH_IDX].dpms_on_ps_idx = 2;
rdev->pm.profiles[PM_PROFILE_MID_MH_IDX].dpms_off_cm_idx = 0;
rdev->pm.profiles[PM_PROFILE_MID_MH_IDX].dpms_on_cm_idx = 1;
drm/radeon/kms/pm: rework power management - Separate dynpm and profile based power management methods. You can select the pm method by echoing the selected method ("dynpm" or "profile") to power_method in sysfs. - Expose basic 4 profile in profile method "default" - default clocks "auto" - select between low and high based on ac/dc state "low" - DC, low power mode "high" - AC, performance mode The current base profile is "default", but it should switched to "auto" once we've tested on more systems. Switching the state is a matter of echoing the requested profile to power_profile in sysfs. The lowest power states are selected automatically when dpms turns the monitors off in all states but default. - Remove dynamic fence-based reclocking for the moment. We can revisit this later once we have basic pm in. - Move pm init/fini to modesetting path. pm is tightly coupled with display state. Make sure display side is initialized before pm. - Add pm suspend/resume functions to make sure pm state is properly reinitialized on resume. - Remove dynpm module option. It's now selectable via sysfs. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-05-08 03:10:16 +08:00
/* high mh */
drm/radeon/kms/pm: fix r6xx+ profile setup This patch is a combination of the previous two profile patches, but without the index bugs. It cleans up and fixes some issues with pm profile setup on r6xx chips. Some tables have different orderings for the power states, also, r600 only has 1 clock mode per power state. On desktop cards there are no battery modes, so the low and high power states are the same. For the low profile case, choose the lower clock mode. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-05-18 07:41:26 +08:00
rdev->pm.profiles[PM_PROFILE_HIGH_MH_IDX].dpms_off_ps_idx = 2;
rdev->pm.profiles[PM_PROFILE_HIGH_MH_IDX].dpms_on_ps_idx = 2;
drm/radeon/kms/pm: rework power management - Separate dynpm and profile based power management methods. You can select the pm method by echoing the selected method ("dynpm" or "profile") to power_method in sysfs. - Expose basic 4 profile in profile method "default" - default clocks "auto" - select between low and high based on ac/dc state "low" - DC, low power mode "high" - AC, performance mode The current base profile is "default", but it should switched to "auto" once we've tested on more systems. Switching the state is a matter of echoing the requested profile to power_profile in sysfs. The lowest power states are selected automatically when dpms turns the monitors off in all states but default. - Remove dynamic fence-based reclocking for the moment. We can revisit this later once we have basic pm in. - Move pm init/fini to modesetting path. pm is tightly coupled with display state. Make sure display side is initialized before pm. - Add pm suspend/resume functions to make sure pm state is properly reinitialized on resume. - Remove dynpm module option. It's now selectable via sysfs. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-05-08 03:10:16 +08:00
rdev->pm.profiles[PM_PROFILE_HIGH_MH_IDX].dpms_off_cm_idx = 0;
rdev->pm.profiles[PM_PROFILE_HIGH_MH_IDX].dpms_on_cm_idx = 2;
} else {
/* default */
rdev->pm.profiles[PM_PROFILE_DEFAULT_IDX].dpms_off_ps_idx = rdev->pm.default_power_state_index;
rdev->pm.profiles[PM_PROFILE_DEFAULT_IDX].dpms_on_ps_idx = rdev->pm.default_power_state_index;
rdev->pm.profiles[PM_PROFILE_DEFAULT_IDX].dpms_off_cm_idx = 0;
rdev->pm.profiles[PM_PROFILE_DEFAULT_IDX].dpms_on_cm_idx = 2;
/* low sh */
drm/radeon/kms: optimize r600_pm_profile_init Avoid a lot of extra loops through the pm state array. Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-11-04 22:09:42 +08:00
if (rdev->flags & RADEON_IS_MOBILITY)
idx = radeon_pm_get_type_index(rdev, POWER_STATE_TYPE_BATTERY, 0);
else
idx = radeon_pm_get_type_index(rdev, POWER_STATE_TYPE_PERFORMANCE, 0);
rdev->pm.profiles[PM_PROFILE_LOW_SH_IDX].dpms_off_ps_idx = idx;
rdev->pm.profiles[PM_PROFILE_LOW_SH_IDX].dpms_on_ps_idx = idx;
rdev->pm.profiles[PM_PROFILE_LOW_SH_IDX].dpms_off_cm_idx = 0;
rdev->pm.profiles[PM_PROFILE_LOW_SH_IDX].dpms_on_cm_idx = 0;
drm/radeon/kms/pm: add mid profile This adds an additional profile, mid, to the pm profile code which takes the place of the old low profile. The default behavior remains the same, e.g., auto profile now selects between mid and high profiles based on power source, however, you can now manually force the low profile which was previously only available as a dpms off state. Enabling the low profile when the displays are on has been known to cause display corruption in some cases. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-06-03 05:56:01 +08:00
/* mid sh */
drm/radeon/kms: optimize r600_pm_profile_init Avoid a lot of extra loops through the pm state array. Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-11-04 22:09:42 +08:00
rdev->pm.profiles[PM_PROFILE_MID_SH_IDX].dpms_off_ps_idx = idx;
rdev->pm.profiles[PM_PROFILE_MID_SH_IDX].dpms_on_ps_idx = idx;
rdev->pm.profiles[PM_PROFILE_MID_SH_IDX].dpms_off_cm_idx = 0;
rdev->pm.profiles[PM_PROFILE_MID_SH_IDX].dpms_on_cm_idx = 1;
drm/radeon/kms/pm: rework power management - Separate dynpm and profile based power management methods. You can select the pm method by echoing the selected method ("dynpm" or "profile") to power_method in sysfs. - Expose basic 4 profile in profile method "default" - default clocks "auto" - select between low and high based on ac/dc state "low" - DC, low power mode "high" - AC, performance mode The current base profile is "default", but it should switched to "auto" once we've tested on more systems. Switching the state is a matter of echoing the requested profile to power_profile in sysfs. The lowest power states are selected automatically when dpms turns the monitors off in all states but default. - Remove dynamic fence-based reclocking for the moment. We can revisit this later once we have basic pm in. - Move pm init/fini to modesetting path. pm is tightly coupled with display state. Make sure display side is initialized before pm. - Add pm suspend/resume functions to make sure pm state is properly reinitialized on resume. - Remove dynpm module option. It's now selectable via sysfs. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-05-08 03:10:16 +08:00
/* high sh */
drm/radeon/kms: optimize r600_pm_profile_init Avoid a lot of extra loops through the pm state array. Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-11-04 22:09:42 +08:00
idx = radeon_pm_get_type_index(rdev, POWER_STATE_TYPE_PERFORMANCE, 0);
rdev->pm.profiles[PM_PROFILE_HIGH_SH_IDX].dpms_off_ps_idx = idx;
rdev->pm.profiles[PM_PROFILE_HIGH_SH_IDX].dpms_on_ps_idx = idx;
drm/radeon/kms/pm: rework power management - Separate dynpm and profile based power management methods. You can select the pm method by echoing the selected method ("dynpm" or "profile") to power_method in sysfs. - Expose basic 4 profile in profile method "default" - default clocks "auto" - select between low and high based on ac/dc state "low" - DC, low power mode "high" - AC, performance mode The current base profile is "default", but it should switched to "auto" once we've tested on more systems. Switching the state is a matter of echoing the requested profile to power_profile in sysfs. The lowest power states are selected automatically when dpms turns the monitors off in all states but default. - Remove dynamic fence-based reclocking for the moment. We can revisit this later once we have basic pm in. - Move pm init/fini to modesetting path. pm is tightly coupled with display state. Make sure display side is initialized before pm. - Add pm suspend/resume functions to make sure pm state is properly reinitialized on resume. - Remove dynpm module option. It's now selectable via sysfs. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-05-08 03:10:16 +08:00
rdev->pm.profiles[PM_PROFILE_HIGH_SH_IDX].dpms_off_cm_idx = 0;
rdev->pm.profiles[PM_PROFILE_HIGH_SH_IDX].dpms_on_cm_idx = 2;
/* low mh */
drm/radeon/kms: optimize r600_pm_profile_init Avoid a lot of extra loops through the pm state array. Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-11-04 22:09:42 +08:00
if (rdev->flags & RADEON_IS_MOBILITY)
idx = radeon_pm_get_type_index(rdev, POWER_STATE_TYPE_BATTERY, 1);
else
idx = radeon_pm_get_type_index(rdev, POWER_STATE_TYPE_PERFORMANCE, 1);
rdev->pm.profiles[PM_PROFILE_LOW_MH_IDX].dpms_off_ps_idx = idx;
rdev->pm.profiles[PM_PROFILE_LOW_MH_IDX].dpms_on_ps_idx = idx;
rdev->pm.profiles[PM_PROFILE_LOW_MH_IDX].dpms_off_cm_idx = 0;
rdev->pm.profiles[PM_PROFILE_LOW_MH_IDX].dpms_on_cm_idx = 0;
drm/radeon/kms/pm: add mid profile This adds an additional profile, mid, to the pm profile code which takes the place of the old low profile. The default behavior remains the same, e.g., auto profile now selects between mid and high profiles based on power source, however, you can now manually force the low profile which was previously only available as a dpms off state. Enabling the low profile when the displays are on has been known to cause display corruption in some cases. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-06-03 05:56:01 +08:00
/* mid mh */
drm/radeon/kms: optimize r600_pm_profile_init Avoid a lot of extra loops through the pm state array. Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-11-04 22:09:42 +08:00
rdev->pm.profiles[PM_PROFILE_MID_MH_IDX].dpms_off_ps_idx = idx;
rdev->pm.profiles[PM_PROFILE_MID_MH_IDX].dpms_on_ps_idx = idx;
rdev->pm.profiles[PM_PROFILE_MID_MH_IDX].dpms_off_cm_idx = 0;
rdev->pm.profiles[PM_PROFILE_MID_MH_IDX].dpms_on_cm_idx = 1;
drm/radeon/kms/pm: rework power management - Separate dynpm and profile based power management methods. You can select the pm method by echoing the selected method ("dynpm" or "profile") to power_method in sysfs. - Expose basic 4 profile in profile method "default" - default clocks "auto" - select between low and high based on ac/dc state "low" - DC, low power mode "high" - AC, performance mode The current base profile is "default", but it should switched to "auto" once we've tested on more systems. Switching the state is a matter of echoing the requested profile to power_profile in sysfs. The lowest power states are selected automatically when dpms turns the monitors off in all states but default. - Remove dynamic fence-based reclocking for the moment. We can revisit this later once we have basic pm in. - Move pm init/fini to modesetting path. pm is tightly coupled with display state. Make sure display side is initialized before pm. - Add pm suspend/resume functions to make sure pm state is properly reinitialized on resume. - Remove dynpm module option. It's now selectable via sysfs. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-05-08 03:10:16 +08:00
/* high mh */
drm/radeon/kms: optimize r600_pm_profile_init Avoid a lot of extra loops through the pm state array. Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-11-04 22:09:42 +08:00
idx = radeon_pm_get_type_index(rdev, POWER_STATE_TYPE_PERFORMANCE, 1);
rdev->pm.profiles[PM_PROFILE_HIGH_MH_IDX].dpms_off_ps_idx = idx;
rdev->pm.profiles[PM_PROFILE_HIGH_MH_IDX].dpms_on_ps_idx = idx;
drm/radeon/kms/pm: rework power management - Separate dynpm and profile based power management methods. You can select the pm method by echoing the selected method ("dynpm" or "profile") to power_method in sysfs. - Expose basic 4 profile in profile method "default" - default clocks "auto" - select between low and high based on ac/dc state "low" - DC, low power mode "high" - AC, performance mode The current base profile is "default", but it should switched to "auto" once we've tested on more systems. Switching the state is a matter of echoing the requested profile to power_profile in sysfs. The lowest power states are selected automatically when dpms turns the monitors off in all states but default. - Remove dynamic fence-based reclocking for the moment. We can revisit this later once we have basic pm in. - Move pm init/fini to modesetting path. pm is tightly coupled with display state. Make sure display side is initialized before pm. - Add pm suspend/resume functions to make sure pm state is properly reinitialized on resume. - Remove dynpm module option. It's now selectable via sysfs. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-05-08 03:10:16 +08:00
rdev->pm.profiles[PM_PROFILE_HIGH_MH_IDX].dpms_off_cm_idx = 0;
rdev->pm.profiles[PM_PROFILE_HIGH_MH_IDX].dpms_on_cm_idx = 2;
}
}
drm/radeon/kms/pm: add asic specific callbacks for setting power state (v2) (v2) Add evergreen vbl checks Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-04-23 01:38:05 +08:00
}
drm/radeon/kms/pm: add additional asic callbacks - pm_misc() - handles voltage, pcie lanes, and other non clock related power mode settings. Currently disabled. Needs further debugging - pm_prepare() - disables crtc mem requests right now. All memory clients need to be disabled when changing memory clocks. This function can be expanded to include disabling fb access as well. - pm_finish() - enable active memory clients. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-04-24 05:57:27 +08:00
void r600_pm_misc(struct radeon_device *rdev)
{
drm/radeon/kms/r600+: use voltage from requested clock mode (v3) This fixes FDO bug #28375, it's kind of regression, so quite important to have it for .35. V2: Fix on RV770+ as well. All other chipsets have only one clock mode per state. V3: I'm out of luck today. Grepped for voltage in r*.c and missed evergreen. agd5f: rebased Signed-off-by: Rafał Miłecki <zajec5@gmail.com> Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-06-08 06:20:25 +08:00
int req_ps_idx = rdev->pm.requested_power_state_index;
int req_cm_idx = rdev->pm.requested_clock_mode_index;
struct radeon_power_state *ps = &rdev->pm.power_state[req_ps_idx];
struct radeon_voltage *voltage = &ps->clock_info[req_cm_idx].voltage;
drm/radeon/kms/pm: add support for SetVoltage cmd table (V2) - This enables voltage adjustment on r6xx+ and certain r5xx asics. - Voltage drop support is already available for most r1xx-r5xx asics. V2: endian fix for voltage table. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-05-28 07:25:54 +08:00
drm/radeon/kms/pm: track current voltage (v2) track the current voltage level and avoid setting it if the requested voltage is already set. v2: check voltage type before checking current voltage Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-06-08 06:15:18 +08:00
if ((voltage->type == VOLTAGE_SW) && voltage->voltage) {
drm/radeon/kms/r6xx+: voltage fixes 0xff01 is not an actual voltage value, but a flag for the driver. If the power state as that value, skip setting the voltage. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-06-21 01:00:31 +08:00
/* 0xff01 is a flag rather then an actual voltage */
if (voltage->voltage == 0xff01)
return;
drm/radeon/kms/pm: track current voltage (v2) track the current voltage level and avoid setting it if the requested voltage is already set. v2: check voltage type before checking current voltage Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-06-08 06:15:18 +08:00
if (voltage->voltage != rdev->pm.current_vddc) {
drm/radeon/kms: add voltage type to atom set voltage function This is needed for setting voltages other than vddc. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-04-13 02:49:23 +08:00
radeon_atom_set_voltage(rdev, voltage->voltage, SET_VOLTAGE_TYPE_ASIC_VDDC);
drm/radeon/kms/pm: track current voltage (v2) track the current voltage level and avoid setting it if the requested voltage is already set. v2: check voltage type before checking current voltage Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-06-08 06:15:18 +08:00
rdev->pm.current_vddc = voltage->voltage;
drm/radeon/kms: move a bunch of modesetting debug to correct debug usage. This migrates a bunch of DRM_DEBUG->DRM_DEBUG_KMS so we can get more modesetting related info without all the other ioctl handling easily. Also the PM code moves to DRM_DEBUG_DRIVER mostly. Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-08-02 08:42:55 +08:00
DRM_DEBUG_DRIVER("Setting: v: %d\n", voltage->voltage);
drm/radeon/kms/pm: track current voltage (v2) track the current voltage level and avoid setting it if the requested voltage is already set. v2: check voltage type before checking current voltage Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-06-08 06:15:18 +08:00
}
}
drm/radeon/kms/pm: add additional asic callbacks - pm_misc() - handles voltage, pcie lanes, and other non clock related power mode settings. Currently disabled. Needs further debugging - pm_prepare() - disables crtc mem requests right now. All memory clients need to be disabled when changing memory clocks. This function can be expanded to include disabling fb access as well. - pm_finish() - enable active memory clients. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-04-24 05:57:27 +08:00
}
drm/radeon/kms: add gui_idle callback Check to see if the GUI engine and related blocks (2D, 3D, CP, etc) are idle or not. There are a number of cases when we need to know if the drawing engine is busy. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-04-23 00:39:58 +08:00
bool r600_gui_idle(struct radeon_device *rdev)
{
if (RREG32(GRBM_STATUS) & GUI_ACTIVE)
return false;
else
return true;
}
drm/radeon/kms: add hpd support for r6xx/r7xx/rs780/rs880 asics This just adds the functionality, it's not hooked up yet. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-05 04:12:21 +08:00
/* hpd for digital panel detect/disconnect */
bool r600_hpd_sense(struct radeon_device *rdev, enum radeon_hpd_id hpd)
{
bool connected = false;
if (ASIC_IS_DCE3(rdev)) {
switch (hpd) {
case RADEON_HPD_1:
if (RREG32(DC_HPD1_INT_STATUS) & DC_HPDx_SENSE)
connected = true;
break;
case RADEON_HPD_2:
if (RREG32(DC_HPD2_INT_STATUS) & DC_HPDx_SENSE)
connected = true;
break;
case RADEON_HPD_3:
if (RREG32(DC_HPD3_INT_STATUS) & DC_HPDx_SENSE)
connected = true;
break;
case RADEON_HPD_4:
if (RREG32(DC_HPD4_INT_STATUS) & DC_HPDx_SENSE)
connected = true;
break;
/* DCE 3.2 */
case RADEON_HPD_5:
if (RREG32(DC_HPD5_INT_STATUS) & DC_HPDx_SENSE)
connected = true;
break;
case RADEON_HPD_6:
if (RREG32(DC_HPD6_INT_STATUS) & DC_HPDx_SENSE)
connected = true;
break;
default:
break;
}
} else {
switch (hpd) {
case RADEON_HPD_1:
if (RREG32(DC_HOT_PLUG_DETECT1_INT_STATUS) & DC_HOT_PLUG_DETECTx_SENSE)
connected = true;
break;
case RADEON_HPD_2:
if (RREG32(DC_HOT_PLUG_DETECT2_INT_STATUS) & DC_HOT_PLUG_DETECTx_SENSE)
connected = true;
break;
case RADEON_HPD_3:
if (RREG32(DC_HOT_PLUG_DETECT3_INT_STATUS) & DC_HOT_PLUG_DETECTx_SENSE)
connected = true;
break;
default:
break;
}
}
return connected;
}
void r600_hpd_set_polarity(struct radeon_device *rdev,
drm/radeon/kms: add asic callbacks for hpd Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-05 04:26:55 +08:00
enum radeon_hpd_id hpd)
drm/radeon/kms: add hpd support for r6xx/r7xx/rs780/rs880 asics This just adds the functionality, it's not hooked up yet. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-05 04:12:21 +08:00
{
u32 tmp;
bool connected = r600_hpd_sense(rdev, hpd);
if (ASIC_IS_DCE3(rdev)) {
switch (hpd) {
case RADEON_HPD_1:
tmp = RREG32(DC_HPD1_INT_CONTROL);
if (connected)
tmp &= ~DC_HPDx_INT_POLARITY;
else
tmp |= DC_HPDx_INT_POLARITY;
WREG32(DC_HPD1_INT_CONTROL, tmp);
break;
case RADEON_HPD_2:
tmp = RREG32(DC_HPD2_INT_CONTROL);
if (connected)
tmp &= ~DC_HPDx_INT_POLARITY;
else
tmp |= DC_HPDx_INT_POLARITY;
WREG32(DC_HPD2_INT_CONTROL, tmp);
break;
case RADEON_HPD_3:
tmp = RREG32(DC_HPD3_INT_CONTROL);
if (connected)
tmp &= ~DC_HPDx_INT_POLARITY;
else
tmp |= DC_HPDx_INT_POLARITY;
WREG32(DC_HPD3_INT_CONTROL, tmp);
break;
case RADEON_HPD_4:
tmp = RREG32(DC_HPD4_INT_CONTROL);
if (connected)
tmp &= ~DC_HPDx_INT_POLARITY;
else
tmp |= DC_HPDx_INT_POLARITY;
WREG32(DC_HPD4_INT_CONTROL, tmp);
break;
case RADEON_HPD_5:
tmp = RREG32(DC_HPD5_INT_CONTROL);
if (connected)
tmp &= ~DC_HPDx_INT_POLARITY;
else
tmp |= DC_HPDx_INT_POLARITY;
WREG32(DC_HPD5_INT_CONTROL, tmp);
break;
/* DCE 3.2 */
case RADEON_HPD_6:
tmp = RREG32(DC_HPD6_INT_CONTROL);
if (connected)
tmp &= ~DC_HPDx_INT_POLARITY;
else
tmp |= DC_HPDx_INT_POLARITY;
WREG32(DC_HPD6_INT_CONTROL, tmp);
break;
default:
break;
}
} else {
switch (hpd) {
case RADEON_HPD_1:
tmp = RREG32(DC_HOT_PLUG_DETECT1_INT_CONTROL);
if (connected)
tmp &= ~DC_HOT_PLUG_DETECTx_INT_POLARITY;
else
tmp |= DC_HOT_PLUG_DETECTx_INT_POLARITY;
WREG32(DC_HOT_PLUG_DETECT1_INT_CONTROL, tmp);
break;
case RADEON_HPD_2:
tmp = RREG32(DC_HOT_PLUG_DETECT2_INT_CONTROL);
if (connected)
tmp &= ~DC_HOT_PLUG_DETECTx_INT_POLARITY;
else
tmp |= DC_HOT_PLUG_DETECTx_INT_POLARITY;
WREG32(DC_HOT_PLUG_DETECT2_INT_CONTROL, tmp);
break;
case RADEON_HPD_3:
tmp = RREG32(DC_HOT_PLUG_DETECT3_INT_CONTROL);
if (connected)
tmp &= ~DC_HOT_PLUG_DETECTx_INT_POLARITY;
else
tmp |= DC_HOT_PLUG_DETECTx_INT_POLARITY;
WREG32(DC_HOT_PLUG_DETECT3_INT_CONTROL, tmp);
break;
default:
break;
}
}
}
void r600_hpd_init(struct radeon_device *rdev)
{
struct drm_device *dev = rdev->ddev;
struct drm_connector *connector;
drm/radeon: apply Murphy's law to the kms irq code v3 1. It is really dangerous to have more than one spinlock protecting the same information. 2. radeon_irq_set sometimes wasn't called with lock protection, so it can happen that more than one CPU would tamper with the irq regs at the same time. 3. The pm.gui_idle variable was assuming that the 3D engine wasn't becoming idle between testing the register and setting the variable. So just remove it and test the register directly. v2: Also handle the hpd irq code the same way. v3: Rename hpd parameter for clarification. Signed-off-by: Christian Koenig <christian.koenig@amd.com> Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
2012-05-17 07:33:30 +08:00
unsigned enable = 0;
drm/radeon/kms: add hpd support for r6xx/r7xx/rs780/rs880 asics This just adds the functionality, it's not hooked up yet. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-05 04:12:21 +08:00
drm/radeon/kms: set HPD polarity in hpd_init() Polarity needs to be set accordingly to connector status (connected or disconnected). Set it up in hpd_init() so first hotplug works reliably no matter what is the initial set of connector. hpd_init() also covers resume so HPD will work correctly after resume as well. Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Cc: Jerome Glisse <j.glisse@gmail.com> Cc: stable@kernel.org Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-11-03 23:21:39 +08:00
list_for_each_entry(connector, &dev->mode_config.connector_list, head) {
struct radeon_connector *radeon_connector = to_radeon_connector(connector);
drm/radeon: don't mess with hot plug detect for eDP or LVDS connector v2 It seems imac pannel doesn't like whe we change the hot plug setup and then refuse to work. This help but doesn't fully fix: https://bugzilla.redhat.com/show_bug.cgi?id=726143 v2: fix typo and improve commit message Signed-off-by: Matthew Garrett <mjg@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Reviewed-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2012-05-04 23:06:22 +08:00
if (connector->connector_type == DRM_MODE_CONNECTOR_eDP ||
connector->connector_type == DRM_MODE_CONNECTOR_LVDS) {
/* don't try to enable hpd on eDP or LVDS avoid breaking the
* aux dp channel on imac and help (but not completely fix)
* https://bugzilla.redhat.com/show_bug.cgi?id=726143
*/
continue;
}
drm/radeon/kms: set HPD polarity in hpd_init() Polarity needs to be set accordingly to connector status (connected or disconnected). Set it up in hpd_init() so first hotplug works reliably no matter what is the initial set of connector. hpd_init() also covers resume so HPD will work correctly after resume as well. Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Cc: Jerome Glisse <j.glisse@gmail.com> Cc: stable@kernel.org Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-11-03 23:21:39 +08:00
if (ASIC_IS_DCE3(rdev)) {
u32 tmp = DC_HPDx_CONNECTION_TIMER(0x9c4) | DC_HPDx_RX_INT_TIMER(0xfa);
if (ASIC_IS_DCE32(rdev))
tmp |= DC_HPDx_EN;
drm/radeon/kms: add hpd support for r6xx/r7xx/rs780/rs880 asics This just adds the functionality, it's not hooked up yet. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-05 04:12:21 +08:00
switch (radeon_connector->hpd.hpd) {
case RADEON_HPD_1:
WREG32(DC_HPD1_CONTROL, tmp);
break;
case RADEON_HPD_2:
WREG32(DC_HPD2_CONTROL, tmp);
break;
case RADEON_HPD_3:
WREG32(DC_HPD3_CONTROL, tmp);
break;
case RADEON_HPD_4:
WREG32(DC_HPD4_CONTROL, tmp);
break;
/* DCE 3.2 */
case RADEON_HPD_5:
WREG32(DC_HPD5_CONTROL, tmp);
break;
case RADEON_HPD_6:
WREG32(DC_HPD6_CONTROL, tmp);
break;
default:
break;
}
drm/radeon/kms: set HPD polarity in hpd_init() Polarity needs to be set accordingly to connector status (connected or disconnected). Set it up in hpd_init() so first hotplug works reliably no matter what is the initial set of connector. hpd_init() also covers resume so HPD will work correctly after resume as well. Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Cc: Jerome Glisse <j.glisse@gmail.com> Cc: stable@kernel.org Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-11-03 23:21:39 +08:00
} else {
drm/radeon/kms: add hpd support for r6xx/r7xx/rs780/rs880 asics This just adds the functionality, it's not hooked up yet. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-05 04:12:21 +08:00
switch (radeon_connector->hpd.hpd) {
case RADEON_HPD_1:
WREG32(DC_HOT_PLUG_DETECT1_CONTROL, DC_HOT_PLUG_DETECTx_EN);
break;
case RADEON_HPD_2:
WREG32(DC_HOT_PLUG_DETECT2_CONTROL, DC_HOT_PLUG_DETECTx_EN);
break;
case RADEON_HPD_3:
WREG32(DC_HOT_PLUG_DETECT3_CONTROL, DC_HOT_PLUG_DETECTx_EN);
break;
default:
break;
}
}
drm/radeon: apply Murphy's law to the kms irq code v3 1. It is really dangerous to have more than one spinlock protecting the same information. 2. radeon_irq_set sometimes wasn't called with lock protection, so it can happen that more than one CPU would tamper with the irq regs at the same time. 3. The pm.gui_idle variable was assuming that the 3D engine wasn't becoming idle between testing the register and setting the variable. So just remove it and test the register directly. v2: Also handle the hpd irq code the same way. v3: Rename hpd parameter for clarification. Signed-off-by: Christian Koenig <christian.koenig@amd.com> Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
2012-05-17 07:33:30 +08:00
enable |= 1 << radeon_connector->hpd.hpd;
drm/radeon/kms: set HPD polarity in hpd_init() Polarity needs to be set accordingly to connector status (connected or disconnected). Set it up in hpd_init() so first hotplug works reliably no matter what is the initial set of connector. hpd_init() also covers resume so HPD will work correctly after resume as well. Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Cc: Jerome Glisse <j.glisse@gmail.com> Cc: stable@kernel.org Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-11-03 23:21:39 +08:00
radeon_hpd_set_polarity(rdev, radeon_connector->hpd.hpd);
drm/radeon/kms: add hpd support for r6xx/r7xx/rs780/rs880 asics This just adds the functionality, it's not hooked up yet. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-05 04:12:21 +08:00
}
drm/radeon: apply Murphy's law to the kms irq code v3 1. It is really dangerous to have more than one spinlock protecting the same information. 2. radeon_irq_set sometimes wasn't called with lock protection, so it can happen that more than one CPU would tamper with the irq regs at the same time. 3. The pm.gui_idle variable was assuming that the 3D engine wasn't becoming idle between testing the register and setting the variable. So just remove it and test the register directly. v2: Also handle the hpd irq code the same way. v3: Rename hpd parameter for clarification. Signed-off-by: Christian Koenig <christian.koenig@amd.com> Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
2012-05-17 07:33:30 +08:00
radeon_irq_kms_enable_hpd(rdev, enable);
drm/radeon/kms: add hpd support for r6xx/r7xx/rs780/rs880 asics This just adds the functionality, it's not hooked up yet. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-05 04:12:21 +08:00
}
void r600_hpd_fini(struct radeon_device *rdev)
{
struct drm_device *dev = rdev->ddev;
struct drm_connector *connector;
drm/radeon: apply Murphy's law to the kms irq code v3 1. It is really dangerous to have more than one spinlock protecting the same information. 2. radeon_irq_set sometimes wasn't called with lock protection, so it can happen that more than one CPU would tamper with the irq regs at the same time. 3. The pm.gui_idle variable was assuming that the 3D engine wasn't becoming idle between testing the register and setting the variable. So just remove it and test the register directly. v2: Also handle the hpd irq code the same way. v3: Rename hpd parameter for clarification. Signed-off-by: Christian Koenig <christian.koenig@amd.com> Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
2012-05-17 07:33:30 +08:00
unsigned disable = 0;
drm/radeon/kms: add hpd support for r6xx/r7xx/rs780/rs880 asics This just adds the functionality, it's not hooked up yet. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-05 04:12:21 +08:00
drm/radeon: apply Murphy's law to the kms irq code v3 1. It is really dangerous to have more than one spinlock protecting the same information. 2. radeon_irq_set sometimes wasn't called with lock protection, so it can happen that more than one CPU would tamper with the irq regs at the same time. 3. The pm.gui_idle variable was assuming that the 3D engine wasn't becoming idle between testing the register and setting the variable. So just remove it and test the register directly. v2: Also handle the hpd irq code the same way. v3: Rename hpd parameter for clarification. Signed-off-by: Christian Koenig <christian.koenig@amd.com> Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
2012-05-17 07:33:30 +08:00
list_for_each_entry(connector, &dev->mode_config.connector_list, head) {
struct radeon_connector *radeon_connector = to_radeon_connector(connector);
if (ASIC_IS_DCE3(rdev)) {
drm/radeon/kms: add hpd support for r6xx/r7xx/rs780/rs880 asics This just adds the functionality, it's not hooked up yet. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-05 04:12:21 +08:00
switch (radeon_connector->hpd.hpd) {
case RADEON_HPD_1:
WREG32(DC_HPD1_CONTROL, 0);
break;
case RADEON_HPD_2:
WREG32(DC_HPD2_CONTROL, 0);
break;
case RADEON_HPD_3:
WREG32(DC_HPD3_CONTROL, 0);
break;
case RADEON_HPD_4:
WREG32(DC_HPD4_CONTROL, 0);
break;
/* DCE 3.2 */
case RADEON_HPD_5:
WREG32(DC_HPD5_CONTROL, 0);
break;
case RADEON_HPD_6:
WREG32(DC_HPD6_CONTROL, 0);
break;
default:
break;
}
drm/radeon: apply Murphy's law to the kms irq code v3 1. It is really dangerous to have more than one spinlock protecting the same information. 2. radeon_irq_set sometimes wasn't called with lock protection, so it can happen that more than one CPU would tamper with the irq regs at the same time. 3. The pm.gui_idle variable was assuming that the 3D engine wasn't becoming idle between testing the register and setting the variable. So just remove it and test the register directly. v2: Also handle the hpd irq code the same way. v3: Rename hpd parameter for clarification. Signed-off-by: Christian Koenig <christian.koenig@amd.com> Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
2012-05-17 07:33:30 +08:00
} else {
drm/radeon/kms: add hpd support for r6xx/r7xx/rs780/rs880 asics This just adds the functionality, it's not hooked up yet. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-05 04:12:21 +08:00
switch (radeon_connector->hpd.hpd) {
case RADEON_HPD_1:
WREG32(DC_HOT_PLUG_DETECT1_CONTROL, 0);
break;
case RADEON_HPD_2:
WREG32(DC_HOT_PLUG_DETECT2_CONTROL, 0);
break;
case RADEON_HPD_3:
WREG32(DC_HOT_PLUG_DETECT3_CONTROL, 0);
break;
default:
break;
}
}
drm/radeon: apply Murphy's law to the kms irq code v3 1. It is really dangerous to have more than one spinlock protecting the same information. 2. radeon_irq_set sometimes wasn't called with lock protection, so it can happen that more than one CPU would tamper with the irq regs at the same time. 3. The pm.gui_idle variable was assuming that the 3D engine wasn't becoming idle between testing the register and setting the variable. So just remove it and test the register directly. v2: Also handle the hpd irq code the same way. v3: Rename hpd parameter for clarification. Signed-off-by: Christian Koenig <christian.koenig@amd.com> Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
2012-05-17 07:33:30 +08:00
disable |= 1 << radeon_connector->hpd.hpd;
drm/radeon/kms: add hpd support for r6xx/r7xx/rs780/rs880 asics This just adds the functionality, it's not hooked up yet. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-05 04:12:21 +08:00
}
drm/radeon: apply Murphy's law to the kms irq code v3 1. It is really dangerous to have more than one spinlock protecting the same information. 2. radeon_irq_set sometimes wasn't called with lock protection, so it can happen that more than one CPU would tamper with the irq regs at the same time. 3. The pm.gui_idle variable was assuming that the 3D engine wasn't becoming idle between testing the register and setting the variable. So just remove it and test the register directly. v2: Also handle the hpd irq code the same way. v3: Rename hpd parameter for clarification. Signed-off-by: Christian Koenig <christian.koenig@amd.com> Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
2012-05-17 07:33:30 +08:00
radeon_irq_kms_disable_hpd(rdev, disable);
drm/radeon/kms: add hpd support for r6xx/r7xx/rs780/rs880 asics This just adds the functionality, it's not hooked up yet. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-05 04:12:21 +08:00
}
drm/radeon: introduce kernel modesetting for radeon hardware Add kernel modesetting support to radeon driver, use the ttm memory manager to manage memory and DRM/GEM to provide userspace API. In order to avoid backward compatibility issue and to allow clean design and code the radeon kernel modesetting use different code path than old radeon/drm driver. When kernel modesetting is enabled the IOCTL of radeon/drm driver are considered as invalid and an error message is printed in the log and they return failure. KMS enabled userspace will use new API to talk with the radeon/drm driver. The new API provide functions to create/destroy/share/mmap buffer object which are then managed by the kernel memory manager (here TTM). In order to submit command to the GPU the userspace provide a buffer holding the command stream, along this buffer userspace have to provide a list of buffer object used by the command stream. The kernel radeon driver will then place buffer in GPU accessible memory and will update command stream to reflect the position of the different buffers. The kernel will also perform security check on command stream provided by the user, we want to catch and forbid any illegal use of the GPU such as DMA into random system memory or into memory not owned by the process supplying the command stream. This part of the code is still incomplete and this why we propose that patch as a staging driver addition, future security might forbid current experimental userspace to run. This code support the following hardware : R1XX,R2XX,R3XX,R4XX,R5XX (radeon up to X1950). Works is underway to provide support for R6XX, R7XX and newer hardware (radeon from HD2XXX to HD4XXX). Authors: Jerome Glisse <jglisse@redhat.com> Dave Airlie <airlied@redhat.com> Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com> Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-06-05 20:42:42 +08:00
/*
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
* R600 PCIE GART
drm/radeon: introduce kernel modesetting for radeon hardware Add kernel modesetting support to radeon driver, use the ttm memory manager to manage memory and DRM/GEM to provide userspace API. In order to avoid backward compatibility issue and to allow clean design and code the radeon kernel modesetting use different code path than old radeon/drm driver. When kernel modesetting is enabled the IOCTL of radeon/drm driver are considered as invalid and an error message is printed in the log and they return failure. KMS enabled userspace will use new API to talk with the radeon/drm driver. The new API provide functions to create/destroy/share/mmap buffer object which are then managed by the kernel memory manager (here TTM). In order to submit command to the GPU the userspace provide a buffer holding the command stream, along this buffer userspace have to provide a list of buffer object used by the command stream. The kernel radeon driver will then place buffer in GPU accessible memory and will update command stream to reflect the position of the different buffers. The kernel will also perform security check on command stream provided by the user, we want to catch and forbid any illegal use of the GPU such as DMA into random system memory or into memory not owned by the process supplying the command stream. This part of the code is still incomplete and this why we propose that patch as a staging driver addition, future security might forbid current experimental userspace to run. This code support the following hardware : R1XX,R2XX,R3XX,R4XX,R5XX (radeon up to X1950). Works is underway to provide support for R6XX, R7XX and newer hardware (radeon from HD2XXX to HD4XXX). Authors: Jerome Glisse <jglisse@redhat.com> Dave Airlie <airlied@redhat.com> Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com> Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-06-05 20:42:42 +08:00
*/
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
void r600_pcie_gart_tlb_flush(struct radeon_device *rdev)
{
unsigned i;
u32 tmp;
drm/radeon/kms: flush HDP cache on GART table updates. Suggested by Alex Deucher @ AMD Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-02-15 13:54:45 +08:00
/* flush hdp cache so updates hit vram */
drm/radeon/kms: don't apply 7xx HDP flush workaround on AGP It should be required for all 7xx asics, but seems to cause problems on some AGP 7xx chips. Fixes: https://bugzilla.kernel.org/show_bug.cgi?id=19002 Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Reported-and-Tested-by: Duncan <1i5t5.duncan@cox.net> Cc: stable@kernel.org Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-12-08 23:05:34 +08:00
if ((rdev->family >= CHIP_RV770) && (rdev->family <= CHIP_RV740) &&
!(rdev->flags & RADEON_IS_AGP)) {
drm/radeon/kms: consolidate GART code, fix segfault after GPU lockup V2 After GPU lockup VRAM gart table is unpinned and thus its pointer becomes unvalid. This patch move the unpin code to a common helper function and set pointer to NULL so that page update code can check if it should update GPU page table or not. That way bo still bound to GART can be unbound (pci_unmap_page for all there page) properly while there is no need to update the GPU page table. V2 move the test for null gart out of the loop, small optimization Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-11-03 23:16:49 +08:00
void __iomem *ptr = (void *)rdev->gart.ptr;
drm/radeon/kms/r7xx: add workaround for hw issue with HDP flush Use of HDP_*_COHERENCY_FLUSH_CNTL can cause a hang in certain situations. Add workaround. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Cc: stable@kernel.org Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-07-27 06:51:53 +08:00
u32 tmp;
/* r7xx hw bug. write to HDP_DEBUG1 followed by fb read
* rather than write to HDP_REG_COHERENCY_FLUSH_CNTL
drm/radeon/kms: don't apply 7xx HDP flush workaround on AGP It should be required for all 7xx asics, but seems to cause problems on some AGP 7xx chips. Fixes: https://bugzilla.kernel.org/show_bug.cgi?id=19002 Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Reported-and-Tested-by: Duncan <1i5t5.duncan@cox.net> Cc: stable@kernel.org Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-12-08 23:05:34 +08:00
* This seems to cause problems on some AGP cards. Just use the old
* method for them.
drm/radeon/kms/r7xx: add workaround for hw issue with HDP flush Use of HDP_*_COHERENCY_FLUSH_CNTL can cause a hang in certain situations. Add workaround. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Cc: stable@kernel.org Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-07-27 06:51:53 +08:00
*/
WREG32(HDP_DEBUG1, 0);
tmp = readl((void __iomem *)ptr);
} else
WREG32(R_005480_HDP_MEM_COHERENCY_FLUSH_CNTL, 0x1);
drm/radeon/kms: flush HDP cache on GART table updates. Suggested by Alex Deucher @ AMD Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-02-15 13:54:45 +08:00
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
WREG32(VM_CONTEXT0_INVALIDATION_LOW_ADDR, rdev->mc.gtt_start >> 12);
WREG32(VM_CONTEXT0_INVALIDATION_HIGH_ADDR, (rdev->mc.gtt_end - 1) >> 12);
WREG32(VM_CONTEXT0_REQUEST_RESPONSE, REQUEST_TYPE(1));
for (i = 0; i < rdev->usec_timeout; i++) {
/* read MC_STATUS */
tmp = RREG32(VM_CONTEXT0_REQUEST_RESPONSE);
tmp = (tmp & RESPONSE_TYPE_MASK) >> RESPONSE_TYPE_SHIFT;
if (tmp == 2) {
printk(KERN_WARNING "[drm] r600 flush TLB failed\n");
return;
}
if (tmp) {
return;
}
udelay(1);
}
}
drm/radeon/kms: clear confusion in GART init/deinit path GART static one time initialization was mixed up with GART enabling/disabling which could happen several time for instance during suspend/resume cycles. This patch splits all GART handling into 4 differents function. gart_init is for one time initialization, gart_deinit is called upon module unload to free resources allocated by gart_init, gart_enable enable the GART and is intented to be call after first initialization and at each resume cycle or reset cycle. Finaly gart_disable stop the GART and is intended to be call at suspend time or when unloading the module. Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-15 00:29:49 +08:00
int r600_pcie_gart_init(struct radeon_device *rdev)
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
{
drm/radeon/kms: clear confusion in GART init/deinit path GART static one time initialization was mixed up with GART enabling/disabling which could happen several time for instance during suspend/resume cycles. This patch splits all GART handling into 4 differents function. gart_init is for one time initialization, gart_deinit is called upon module unload to free resources allocated by gart_init, gart_enable enable the GART and is intented to be call after first initialization and at each resume cycle or reset cycle. Finaly gart_disable stop the GART and is intended to be call at suspend time or when unloading the module. Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-15 00:29:49 +08:00
int r;
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
drm/radeon/kms: consolidate GART code, fix segfault after GPU lockup V2 After GPU lockup VRAM gart table is unpinned and thus its pointer becomes unvalid. This patch move the unpin code to a common helper function and set pointer to NULL so that page update code can check if it should update GPU page table or not. That way bo still bound to GART can be unbound (pci_unmap_page for all there page) properly while there is no need to update the GPU page table. V2 move the test for null gart out of the loop, small optimization Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-11-03 23:16:49 +08:00
if (rdev->gart.robj) {
drivers/gpu/drm: Update WARN uses Coalesce long formats. Align arguments. Add missing newlines. Signed-off-by: Joe Perches <joe@perches.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-10-31 05:08:30 +08:00
WARN(1, "R600 PCIE GART already initialized\n");
drm/radeon/kms: clear confusion in GART init/deinit path GART static one time initialization was mixed up with GART enabling/disabling which could happen several time for instance during suspend/resume cycles. This patch splits all GART handling into 4 differents function. gart_init is for one time initialization, gart_deinit is called upon module unload to free resources allocated by gart_init, gart_enable enable the GART and is intented to be call after first initialization and at each resume cycle or reset cycle. Finaly gart_disable stop the GART and is intended to be call at suspend time or when unloading the module. Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-15 00:29:49 +08:00
return 0;
}
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
/* Initialize common gart structure */
r = radeon_gart_init(rdev);
drm/radeon/kms: clear confusion in GART init/deinit path GART static one time initialization was mixed up with GART enabling/disabling which could happen several time for instance during suspend/resume cycles. This patch splits all GART handling into 4 differents function. gart_init is for one time initialization, gart_deinit is called upon module unload to free resources allocated by gart_init, gart_enable enable the GART and is intented to be call after first initialization and at each resume cycle or reset cycle. Finaly gart_disable stop the GART and is intended to be call at suspend time or when unloading the module. Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-15 00:29:49 +08:00
if (r)
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
return r;
rdev->gart.table_size = rdev->gart.num_gpu_pages * 8;
drm/radeon/kms: clear confusion in GART init/deinit path GART static one time initialization was mixed up with GART enabling/disabling which could happen several time for instance during suspend/resume cycles. This patch splits all GART handling into 4 differents function. gart_init is for one time initialization, gart_deinit is called upon module unload to free resources allocated by gart_init, gart_enable enable the GART and is intented to be call after first initialization and at each resume cycle or reset cycle. Finaly gart_disable stop the GART and is intended to be call at suspend time or when unloading the module. Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-15 00:29:49 +08:00
return radeon_gart_table_vram_alloc(rdev);
}
drm/radeon: Mark all possible functions / structs as static Let's allow GCC to optimize better. This exposed some five unused functions, but this patch doesn't remove them. Signed-off-by: Lauri Kasanen <cand@gmx.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2012-09-01 01:43:50 +08:00
static int r600_pcie_gart_enable(struct radeon_device *rdev)
drm/radeon/kms: clear confusion in GART init/deinit path GART static one time initialization was mixed up with GART enabling/disabling which could happen several time for instance during suspend/resume cycles. This patch splits all GART handling into 4 differents function. gart_init is for one time initialization, gart_deinit is called upon module unload to free resources allocated by gart_init, gart_enable enable the GART and is intented to be call after first initialization and at each resume cycle or reset cycle. Finaly gart_disable stop the GART and is intended to be call at suspend time or when unloading the module. Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-15 00:29:49 +08:00
{
u32 tmp;
int r, i;
drm/radeon/kms: consolidate GART code, fix segfault after GPU lockup V2 After GPU lockup VRAM gart table is unpinned and thus its pointer becomes unvalid. This patch move the unpin code to a common helper function and set pointer to NULL so that page update code can check if it should update GPU page table or not. That way bo still bound to GART can be unbound (pci_unmap_page for all there page) properly while there is no need to update the GPU page table. V2 move the test for null gart out of the loop, small optimization Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-11-03 23:16:49 +08:00
if (rdev->gart.robj == NULL) {
drm/radeon/kms: clear confusion in GART init/deinit path GART static one time initialization was mixed up with GART enabling/disabling which could happen several time for instance during suspend/resume cycles. This patch splits all GART handling into 4 differents function. gart_init is for one time initialization, gart_deinit is called upon module unload to free resources allocated by gart_init, gart_enable enable the GART and is intented to be call after first initialization and at each resume cycle or reset cycle. Finaly gart_disable stop the GART and is intended to be call at suspend time or when unloading the module. Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-15 00:29:49 +08:00
dev_err(rdev->dev, "No VRAM object for PCIE GART.\n");
return -EINVAL;
drm/radeon: introduce kernel modesetting for radeon hardware Add kernel modesetting support to radeon driver, use the ttm memory manager to manage memory and DRM/GEM to provide userspace API. In order to avoid backward compatibility issue and to allow clean design and code the radeon kernel modesetting use different code path than old radeon/drm driver. When kernel modesetting is enabled the IOCTL of radeon/drm driver are considered as invalid and an error message is printed in the log and they return failure. KMS enabled userspace will use new API to talk with the radeon/drm driver. The new API provide functions to create/destroy/share/mmap buffer object which are then managed by the kernel memory manager (here TTM). In order to submit command to the GPU the userspace provide a buffer holding the command stream, along this buffer userspace have to provide a list of buffer object used by the command stream. The kernel radeon driver will then place buffer in GPU accessible memory and will update command stream to reflect the position of the different buffers. The kernel will also perform security check on command stream provided by the user, we want to catch and forbid any illegal use of the GPU such as DMA into random system memory or into memory not owned by the process supplying the command stream. This part of the code is still incomplete and this why we propose that patch as a staging driver addition, future security might forbid current experimental userspace to run. This code support the following hardware : R1XX,R2XX,R3XX,R4XX,R5XX (radeon up to X1950). Works is underway to provide support for R6XX, R7XX and newer hardware (radeon from HD2XXX to HD4XXX). Authors: Jerome Glisse <jglisse@redhat.com> Dave Airlie <airlied@redhat.com> Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com> Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-06-05 20:42:42 +08:00
}
drm/radeon/kms: clear confusion in GART init/deinit path GART static one time initialization was mixed up with GART enabling/disabling which could happen several time for instance during suspend/resume cycles. This patch splits all GART handling into 4 differents function. gart_init is for one time initialization, gart_deinit is called upon module unload to free resources allocated by gart_init, gart_enable enable the GART and is intented to be call after first initialization and at each resume cycle or reset cycle. Finaly gart_disable stop the GART and is intended to be call at suspend time or when unloading the module. Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-15 00:29:49 +08:00
r = radeon_gart_table_vram_pin(rdev);
if (r)
return r;
drm/radeon/kms: set gart pages to invalid on unbind and point to dummy page this uses a new entrypoint to invalidate gart entries instead of using 0. Changed to rather than pointing to 0 address point empty entry to dummy page. This might help to avoid hard lockup if for some wrong reasons GPU try to access unmapped GART entry. I'm not 100% sure this is going to work, we probably need to allocate a dummy page and point all the GTT entries at it similiar to what AGP does. but we can test this first I suppose. Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-02-05 14:00:07 +08:00
radeon_gart_restore(rdev);
drm/radeon/r600: fix some issues with suspend/resume. a) don't zero gart table on gart enable b) move pinning shader object into resume path c) unpin shader object on suspend d) set cp ready to false after cp shutdown on suspend. Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-15 09:07:52 +08:00
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
/* Setup L2 cache */
WREG32(VM_L2_CNTL, ENABLE_L2_CACHE | ENABLE_L2_FRAGMENT_PROCESSING |
ENABLE_L2_PTE_CACHE_LRU_UPDATE_BY_WRITE |
EFFECTIVE_L2_QUEUE_SIZE(7));
WREG32(VM_L2_CNTL2, 0);
WREG32(VM_L2_CNTL3, BANK_SELECT_0(0) | BANK_SELECT_1(1));
/* Setup TLB control */
tmp = ENABLE_L1_TLB | ENABLE_L1_FRAGMENT_PROCESSING |
SYSTEM_ACCESS_MODE_NOT_IN_SYS |
EFFECTIVE_L1_TLB_SIZE(5) | EFFECTIVE_L1_QUEUE_SIZE(5) |
ENABLE_WAIT_L2_QUERY;
WREG32(MC_VM_L1_TLB_MCB_RD_SYS_CNTL, tmp);
WREG32(MC_VM_L1_TLB_MCB_WR_SYS_CNTL, tmp);
WREG32(MC_VM_L1_TLB_MCB_RD_HDP_CNTL, tmp | ENABLE_L1_STRICT_ORDERING);
WREG32(MC_VM_L1_TLB_MCB_WR_HDP_CNTL, tmp);
WREG32(MC_VM_L1_TLB_MCD_RD_A_CNTL, tmp);
WREG32(MC_VM_L1_TLB_MCD_WR_A_CNTL, tmp);
WREG32(MC_VM_L1_TLB_MCD_RD_B_CNTL, tmp);
WREG32(MC_VM_L1_TLB_MCD_WR_B_CNTL, tmp);
WREG32(MC_VM_L1_TLB_MCB_RD_GFX_CNTL, tmp);
WREG32(MC_VM_L1_TLB_MCB_WR_GFX_CNTL, tmp);
WREG32(MC_VM_L1_TLB_MCB_RD_PDMA_CNTL, tmp);
WREG32(MC_VM_L1_TLB_MCB_WR_PDMA_CNTL, tmp);
WREG32(MC_VM_L1_TLB_MCB_RD_SEM_CNTL, tmp | ENABLE_SEMAPHORE_MODE);
WREG32(MC_VM_L1_TLB_MCB_WR_SEM_CNTL, tmp | ENABLE_SEMAPHORE_MODE);
WREG32(VM_CONTEXT0_PAGE_TABLE_START_ADDR, rdev->mc.gtt_start >> 12);
drm/radeon/kms: Fix AGP support for R600/RV770 family (v2) For AGP to work unmapped access must cover VRAM & AGP as AGP is treated like VRAM by the GPU (ie physical address). This patch properly setup the virtual memory system aperture to cover AGP if AGP is enabled. It seems that there is memory corruption after resume when using AGP (RV770 seems unaffected thought). Version 2 just fix merge issue with updated AGP fallback patch. Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-10-07 01:04:30 +08:00
WREG32(VM_CONTEXT0_PAGE_TABLE_END_ADDR, rdev->mc.gtt_end >> 12);
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
WREG32(VM_CONTEXT0_PAGE_TABLE_BASE_ADDR, rdev->gart.table_addr >> 12);
WREG32(VM_CONTEXT0_CNTL, ENABLE_CONTEXT | PAGE_TABLE_DEPTH(0) |
RANGE_PROTECTION_FAULT_ENABLE_DEFAULT);
WREG32(VM_CONTEXT0_PROTECTION_FAULT_DEFAULT_ADDR,
(u32)(rdev->dummy_page.addr >> 12));
for (i = 1; i < 7; i++)
WREG32(VM_CONTEXT0_CNTL + (i * 4), 0);
drm/radeon: introduce kernel modesetting for radeon hardware Add kernel modesetting support to radeon driver, use the ttm memory manager to manage memory and DRM/GEM to provide userspace API. In order to avoid backward compatibility issue and to allow clean design and code the radeon kernel modesetting use different code path than old radeon/drm driver. When kernel modesetting is enabled the IOCTL of radeon/drm driver are considered as invalid and an error message is printed in the log and they return failure. KMS enabled userspace will use new API to talk with the radeon/drm driver. The new API provide functions to create/destroy/share/mmap buffer object which are then managed by the kernel memory manager (here TTM). In order to submit command to the GPU the userspace provide a buffer holding the command stream, along this buffer userspace have to provide a list of buffer object used by the command stream. The kernel radeon driver will then place buffer in GPU accessible memory and will update command stream to reflect the position of the different buffers. The kernel will also perform security check on command stream provided by the user, we want to catch and forbid any illegal use of the GPU such as DMA into random system memory or into memory not owned by the process supplying the command stream. This part of the code is still incomplete and this why we propose that patch as a staging driver addition, future security might forbid current experimental userspace to run. This code support the following hardware : R1XX,R2XX,R3XX,R4XX,R5XX (radeon up to X1950). Works is underway to provide support for R6XX, R7XX and newer hardware (radeon from HD2XXX to HD4XXX). Authors: Jerome Glisse <jglisse@redhat.com> Dave Airlie <airlied@redhat.com> Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com> Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-06-05 20:42:42 +08:00
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
r600_pcie_gart_tlb_flush(rdev);
drm/radeon: Print gart initialization details on all chipsets This was previously done for r300 only. Use %016llX instead of %08X for printing the table address. Also fix typos in gart warning messages. Signed-off-by: Tormod Volden <debian.tormod@gmail.com> Reviewed-by: Michel Dänzer <michel@daenzer.net> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-09-01 05:54:07 +08:00
DRM_INFO("PCIE GART of %uM enabled (table at 0x%016llX).\n",
(unsigned)(rdev->mc.gtt_size >> 20),
(unsigned long long)rdev->gart.table_addr);
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
rdev->gart.ready = true;
drm/radeon: introduce kernel modesetting for radeon hardware Add kernel modesetting support to radeon driver, use the ttm memory manager to manage memory and DRM/GEM to provide userspace API. In order to avoid backward compatibility issue and to allow clean design and code the radeon kernel modesetting use different code path than old radeon/drm driver. When kernel modesetting is enabled the IOCTL of radeon/drm driver are considered as invalid and an error message is printed in the log and they return failure. KMS enabled userspace will use new API to talk with the radeon/drm driver. The new API provide functions to create/destroy/share/mmap buffer object which are then managed by the kernel memory manager (here TTM). In order to submit command to the GPU the userspace provide a buffer holding the command stream, along this buffer userspace have to provide a list of buffer object used by the command stream. The kernel radeon driver will then place buffer in GPU accessible memory and will update command stream to reflect the position of the different buffers. The kernel will also perform security check on command stream provided by the user, we want to catch and forbid any illegal use of the GPU such as DMA into random system memory or into memory not owned by the process supplying the command stream. This part of the code is still incomplete and this why we propose that patch as a staging driver addition, future security might forbid current experimental userspace to run. This code support the following hardware : R1XX,R2XX,R3XX,R4XX,R5XX (radeon up to X1950). Works is underway to provide support for R6XX, R7XX and newer hardware (radeon from HD2XXX to HD4XXX). Authors: Jerome Glisse <jglisse@redhat.com> Dave Airlie <airlied@redhat.com> Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com> Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-06-05 20:42:42 +08:00
return 0;
}
drm/radeon: Mark all possible functions / structs as static Let's allow GCC to optimize better. This exposed some five unused functions, but this patch doesn't remove them. Signed-off-by: Lauri Kasanen <cand@gmx.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2012-09-01 01:43:50 +08:00
static void r600_pcie_gart_disable(struct radeon_device *rdev)
drm/radeon: introduce kernel modesetting for radeon hardware Add kernel modesetting support to radeon driver, use the ttm memory manager to manage memory and DRM/GEM to provide userspace API. In order to avoid backward compatibility issue and to allow clean design and code the radeon kernel modesetting use different code path than old radeon/drm driver. When kernel modesetting is enabled the IOCTL of radeon/drm driver are considered as invalid and an error message is printed in the log and they return failure. KMS enabled userspace will use new API to talk with the radeon/drm driver. The new API provide functions to create/destroy/share/mmap buffer object which are then managed by the kernel memory manager (here TTM). In order to submit command to the GPU the userspace provide a buffer holding the command stream, along this buffer userspace have to provide a list of buffer object used by the command stream. The kernel radeon driver will then place buffer in GPU accessible memory and will update command stream to reflect the position of the different buffers. The kernel will also perform security check on command stream provided by the user, we want to catch and forbid any illegal use of the GPU such as DMA into random system memory or into memory not owned by the process supplying the command stream. This part of the code is still incomplete and this why we propose that patch as a staging driver addition, future security might forbid current experimental userspace to run. This code support the following hardware : R1XX,R2XX,R3XX,R4XX,R5XX (radeon up to X1950). Works is underway to provide support for R6XX, R7XX and newer hardware (radeon from HD2XXX to HD4XXX). Authors: Jerome Glisse <jglisse@redhat.com> Dave Airlie <airlied@redhat.com> Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com> Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-06-05 20:42:42 +08:00
{
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
u32 tmp;
drm/radeon/kms: consolidate GART code, fix segfault after GPU lockup V2 After GPU lockup VRAM gart table is unpinned and thus its pointer becomes unvalid. This patch move the unpin code to a common helper function and set pointer to NULL so that page update code can check if it should update GPU page table or not. That way bo still bound to GART can be unbound (pci_unmap_page for all there page) properly while there is no need to update the GPU page table. V2 move the test for null gart out of the loop, small optimization Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-11-03 23:16:49 +08:00
int i;
drm/radeon: introduce kernel modesetting for radeon hardware Add kernel modesetting support to radeon driver, use the ttm memory manager to manage memory and DRM/GEM to provide userspace API. In order to avoid backward compatibility issue and to allow clean design and code the radeon kernel modesetting use different code path than old radeon/drm driver. When kernel modesetting is enabled the IOCTL of radeon/drm driver are considered as invalid and an error message is printed in the log and they return failure. KMS enabled userspace will use new API to talk with the radeon/drm driver. The new API provide functions to create/destroy/share/mmap buffer object which are then managed by the kernel memory manager (here TTM). In order to submit command to the GPU the userspace provide a buffer holding the command stream, along this buffer userspace have to provide a list of buffer object used by the command stream. The kernel radeon driver will then place buffer in GPU accessible memory and will update command stream to reflect the position of the different buffers. The kernel will also perform security check on command stream provided by the user, we want to catch and forbid any illegal use of the GPU such as DMA into random system memory or into memory not owned by the process supplying the command stream. This part of the code is still incomplete and this why we propose that patch as a staging driver addition, future security might forbid current experimental userspace to run. This code support the following hardware : R1XX,R2XX,R3XX,R4XX,R5XX (radeon up to X1950). Works is underway to provide support for R6XX, R7XX and newer hardware (radeon from HD2XXX to HD4XXX). Authors: Jerome Glisse <jglisse@redhat.com> Dave Airlie <airlied@redhat.com> Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com> Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-06-05 20:42:42 +08:00
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
/* Disable all tables */
for (i = 0; i < 7; i++)
WREG32(VM_CONTEXT0_CNTL + (i * 4), 0);
drm/radeon: introduce kernel modesetting for radeon hardware Add kernel modesetting support to radeon driver, use the ttm memory manager to manage memory and DRM/GEM to provide userspace API. In order to avoid backward compatibility issue and to allow clean design and code the radeon kernel modesetting use different code path than old radeon/drm driver. When kernel modesetting is enabled the IOCTL of radeon/drm driver are considered as invalid and an error message is printed in the log and they return failure. KMS enabled userspace will use new API to talk with the radeon/drm driver. The new API provide functions to create/destroy/share/mmap buffer object which are then managed by the kernel memory manager (here TTM). In order to submit command to the GPU the userspace provide a buffer holding the command stream, along this buffer userspace have to provide a list of buffer object used by the command stream. The kernel radeon driver will then place buffer in GPU accessible memory and will update command stream to reflect the position of the different buffers. The kernel will also perform security check on command stream provided by the user, we want to catch and forbid any illegal use of the GPU such as DMA into random system memory or into memory not owned by the process supplying the command stream. This part of the code is still incomplete and this why we propose that patch as a staging driver addition, future security might forbid current experimental userspace to run. This code support the following hardware : R1XX,R2XX,R3XX,R4XX,R5XX (radeon up to X1950). Works is underway to provide support for R6XX, R7XX and newer hardware (radeon from HD2XXX to HD4XXX). Authors: Jerome Glisse <jglisse@redhat.com> Dave Airlie <airlied@redhat.com> Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com> Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-06-05 20:42:42 +08:00
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
/* Disable L2 cache */
WREG32(VM_L2_CNTL, ENABLE_L2_FRAGMENT_PROCESSING |
EFFECTIVE_L2_QUEUE_SIZE(7));
WREG32(VM_L2_CNTL3, BANK_SELECT_0(0) | BANK_SELECT_1(1));
/* Setup L1 TLB control */
tmp = EFFECTIVE_L1_TLB_SIZE(5) | EFFECTIVE_L1_QUEUE_SIZE(5) |
ENABLE_WAIT_L2_QUERY;
WREG32(MC_VM_L1_TLB_MCD_RD_A_CNTL, tmp);
WREG32(MC_VM_L1_TLB_MCD_WR_A_CNTL, tmp);
WREG32(MC_VM_L1_TLB_MCD_RD_B_CNTL, tmp);
WREG32(MC_VM_L1_TLB_MCD_WR_B_CNTL, tmp);
WREG32(MC_VM_L1_TLB_MCB_RD_GFX_CNTL, tmp);
WREG32(MC_VM_L1_TLB_MCB_WR_GFX_CNTL, tmp);
WREG32(MC_VM_L1_TLB_MCB_RD_PDMA_CNTL, tmp);
WREG32(MC_VM_L1_TLB_MCB_WR_PDMA_CNTL, tmp);
WREG32(MC_VM_L1_TLB_MCB_RD_SEM_CNTL, tmp);
WREG32(MC_VM_L1_TLB_MCB_WR_SEM_CNTL, tmp);
WREG32(MC_VM_L1_TLB_MCB_RD_SYS_CNTL, tmp);
WREG32(MC_VM_L1_TLB_MCB_WR_SYS_CNTL, tmp);
WREG32(MC_VM_L1_TLB_MCB_RD_HDP_CNTL, tmp);
WREG32(MC_VM_L1_TLB_MCB_WR_HDP_CNTL, tmp);
drm/radeon/kms: consolidate GART code, fix segfault after GPU lockup V2 After GPU lockup VRAM gart table is unpinned and thus its pointer becomes unvalid. This patch move the unpin code to a common helper function and set pointer to NULL so that page update code can check if it should update GPU page table or not. That way bo still bound to GART can be unbound (pci_unmap_page for all there page) properly while there is no need to update the GPU page table. V2 move the test for null gart out of the loop, small optimization Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-11-03 23:16:49 +08:00
radeon_gart_table_vram_unpin(rdev);
drm/radeon/kms: clear confusion in GART init/deinit path GART static one time initialization was mixed up with GART enabling/disabling which could happen several time for instance during suspend/resume cycles. This patch splits all GART handling into 4 differents function. gart_init is for one time initialization, gart_deinit is called upon module unload to free resources allocated by gart_init, gart_enable enable the GART and is intented to be call after first initialization and at each resume cycle or reset cycle. Finaly gart_disable stop the GART and is intended to be call at suspend time or when unloading the module. Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-15 00:29:49 +08:00
}
drm/radeon: Mark all possible functions / structs as static Let's allow GCC to optimize better. This exposed some five unused functions, but this patch doesn't remove them. Signed-off-by: Lauri Kasanen <cand@gmx.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2012-09-01 01:43:50 +08:00
static void r600_pcie_gart_fini(struct radeon_device *rdev)
drm/radeon/kms: clear confusion in GART init/deinit path GART static one time initialization was mixed up with GART enabling/disabling which could happen several time for instance during suspend/resume cycles. This patch splits all GART handling into 4 differents function. gart_init is for one time initialization, gart_deinit is called upon module unload to free resources allocated by gart_init, gart_enable enable the GART and is intented to be call after first initialization and at each resume cycle or reset cycle. Finaly gart_disable stop the GART and is intended to be call at suspend time or when unloading the module. Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-15 00:29:49 +08:00
{
drm/radeon/kms: avoid possible oops (call gart_fini before gart_disable) radeon_gart_fini might call GART unbind callback function which might try to access GART table but if gart_disable is call first the GART table will be unmapped so any access to it will oops. Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-03-17 22:44:29 +08:00
radeon_gart_fini(rdev);
drm/radeon/kms: clear confusion in GART init/deinit path GART static one time initialization was mixed up with GART enabling/disabling which could happen several time for instance during suspend/resume cycles. This patch splits all GART handling into 4 differents function. gart_init is for one time initialization, gart_deinit is called upon module unload to free resources allocated by gart_init, gart_enable enable the GART and is intented to be call after first initialization and at each resume cycle or reset cycle. Finaly gart_disable stop the GART and is intended to be call at suspend time or when unloading the module. Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-15 00:29:49 +08:00
r600_pcie_gart_disable(rdev);
radeon_gart_table_vram_free(rdev);
drm/radeon: introduce kernel modesetting for radeon hardware Add kernel modesetting support to radeon driver, use the ttm memory manager to manage memory and DRM/GEM to provide userspace API. In order to avoid backward compatibility issue and to allow clean design and code the radeon kernel modesetting use different code path than old radeon/drm driver. When kernel modesetting is enabled the IOCTL of radeon/drm driver are considered as invalid and an error message is printed in the log and they return failure. KMS enabled userspace will use new API to talk with the radeon/drm driver. The new API provide functions to create/destroy/share/mmap buffer object which are then managed by the kernel memory manager (here TTM). In order to submit command to the GPU the userspace provide a buffer holding the command stream, along this buffer userspace have to provide a list of buffer object used by the command stream. The kernel radeon driver will then place buffer in GPU accessible memory and will update command stream to reflect the position of the different buffers. The kernel will also perform security check on command stream provided by the user, we want to catch and forbid any illegal use of the GPU such as DMA into random system memory or into memory not owned by the process supplying the command stream. This part of the code is still incomplete and this why we propose that patch as a staging driver addition, future security might forbid current experimental userspace to run. This code support the following hardware : R1XX,R2XX,R3XX,R4XX,R5XX (radeon up to X1950). Works is underway to provide support for R6XX, R7XX and newer hardware (radeon from HD2XXX to HD4XXX). Authors: Jerome Glisse <jglisse@redhat.com> Dave Airlie <airlied@redhat.com> Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com> Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-06-05 20:42:42 +08:00
}
drm/radeon: Mark all possible functions / structs as static Let's allow GCC to optimize better. This exposed some five unused functions, but this patch doesn't remove them. Signed-off-by: Lauri Kasanen <cand@gmx.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2012-09-01 01:43:50 +08:00
static void r600_agp_enable(struct radeon_device *rdev)
drm/radeon/kms: Fix AGP support for R600/RV770 family (v2) For AGP to work unmapped access must cover VRAM & AGP as AGP is treated like VRAM by the GPU (ie physical address). This patch properly setup the virtual memory system aperture to cover AGP if AGP is enabled. It seems that there is memory corruption after resume when using AGP (RV770 seems unaffected thought). Version 2 just fix merge issue with updated AGP fallback patch. Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-10-07 01:04:30 +08:00
{
u32 tmp;
int i;
/* Setup L2 cache */
WREG32(VM_L2_CNTL, ENABLE_L2_CACHE | ENABLE_L2_FRAGMENT_PROCESSING |
ENABLE_L2_PTE_CACHE_LRU_UPDATE_BY_WRITE |
EFFECTIVE_L2_QUEUE_SIZE(7));
WREG32(VM_L2_CNTL2, 0);
WREG32(VM_L2_CNTL3, BANK_SELECT_0(0) | BANK_SELECT_1(1));
/* Setup TLB control */
tmp = ENABLE_L1_TLB | ENABLE_L1_FRAGMENT_PROCESSING |
SYSTEM_ACCESS_MODE_NOT_IN_SYS |
EFFECTIVE_L1_TLB_SIZE(5) | EFFECTIVE_L1_QUEUE_SIZE(5) |
ENABLE_WAIT_L2_QUERY;
WREG32(MC_VM_L1_TLB_MCB_RD_SYS_CNTL, tmp);
WREG32(MC_VM_L1_TLB_MCB_WR_SYS_CNTL, tmp);
WREG32(MC_VM_L1_TLB_MCB_RD_HDP_CNTL, tmp | ENABLE_L1_STRICT_ORDERING);
WREG32(MC_VM_L1_TLB_MCB_WR_HDP_CNTL, tmp);
WREG32(MC_VM_L1_TLB_MCD_RD_A_CNTL, tmp);
WREG32(MC_VM_L1_TLB_MCD_WR_A_CNTL, tmp);
WREG32(MC_VM_L1_TLB_MCD_RD_B_CNTL, tmp);
WREG32(MC_VM_L1_TLB_MCD_WR_B_CNTL, tmp);
WREG32(MC_VM_L1_TLB_MCB_RD_GFX_CNTL, tmp);
WREG32(MC_VM_L1_TLB_MCB_WR_GFX_CNTL, tmp);
WREG32(MC_VM_L1_TLB_MCB_RD_PDMA_CNTL, tmp);
WREG32(MC_VM_L1_TLB_MCB_WR_PDMA_CNTL, tmp);
WREG32(MC_VM_L1_TLB_MCB_RD_SEM_CNTL, tmp | ENABLE_SEMAPHORE_MODE);
WREG32(MC_VM_L1_TLB_MCB_WR_SEM_CNTL, tmp | ENABLE_SEMAPHORE_MODE);
for (i = 0; i < 7; i++)
WREG32(VM_CONTEXT0_CNTL + (i * 4), 0);
}
drm/radeon: introduce kernel modesetting for radeon hardware Add kernel modesetting support to radeon driver, use the ttm memory manager to manage memory and DRM/GEM to provide userspace API. In order to avoid backward compatibility issue and to allow clean design and code the radeon kernel modesetting use different code path than old radeon/drm driver. When kernel modesetting is enabled the IOCTL of radeon/drm driver are considered as invalid and an error message is printed in the log and they return failure. KMS enabled userspace will use new API to talk with the radeon/drm driver. The new API provide functions to create/destroy/share/mmap buffer object which are then managed by the kernel memory manager (here TTM). In order to submit command to the GPU the userspace provide a buffer holding the command stream, along this buffer userspace have to provide a list of buffer object used by the command stream. The kernel radeon driver will then place buffer in GPU accessible memory and will update command stream to reflect the position of the different buffers. The kernel will also perform security check on command stream provided by the user, we want to catch and forbid any illegal use of the GPU such as DMA into random system memory or into memory not owned by the process supplying the command stream. This part of the code is still incomplete and this why we propose that patch as a staging driver addition, future security might forbid current experimental userspace to run. This code support the following hardware : R1XX,R2XX,R3XX,R4XX,R5XX (radeon up to X1950). Works is underway to provide support for R6XX, R7XX and newer hardware (radeon from HD2XXX to HD4XXX). Authors: Jerome Glisse <jglisse@redhat.com> Dave Airlie <airlied@redhat.com> Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com> Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-06-05 20:42:42 +08:00
int r600_mc_wait_for_idle(struct radeon_device *rdev)
{
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
unsigned i;
u32 tmp;
for (i = 0; i < rdev->usec_timeout; i++) {
/* read MC_STATUS */
tmp = RREG32(R_000E50_SRBM_STATUS) & 0x3F00;
if (!tmp)
return 0;
udelay(1);
}
return -1;
drm/radeon: introduce kernel modesetting for radeon hardware Add kernel modesetting support to radeon driver, use the ttm memory manager to manage memory and DRM/GEM to provide userspace API. In order to avoid backward compatibility issue and to allow clean design and code the radeon kernel modesetting use different code path than old radeon/drm driver. When kernel modesetting is enabled the IOCTL of radeon/drm driver are considered as invalid and an error message is printed in the log and they return failure. KMS enabled userspace will use new API to talk with the radeon/drm driver. The new API provide functions to create/destroy/share/mmap buffer object which are then managed by the kernel memory manager (here TTM). In order to submit command to the GPU the userspace provide a buffer holding the command stream, along this buffer userspace have to provide a list of buffer object used by the command stream. The kernel radeon driver will then place buffer in GPU accessible memory and will update command stream to reflect the position of the different buffers. The kernel will also perform security check on command stream provided by the user, we want to catch and forbid any illegal use of the GPU such as DMA into random system memory or into memory not owned by the process supplying the command stream. This part of the code is still incomplete and this why we propose that patch as a staging driver addition, future security might forbid current experimental userspace to run. This code support the following hardware : R1XX,R2XX,R3XX,R4XX,R5XX (radeon up to X1950). Works is underway to provide support for R6XX, R7XX and newer hardware (radeon from HD2XXX to HD4XXX). Authors: Jerome Glisse <jglisse@redhat.com> Dave Airlie <airlied@redhat.com> Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com> Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-06-05 20:42:42 +08:00
}
drm/radeon/kms: Fix R600/RV770 startup path & reset We were calling reset unconditionaly in the startup path this is bad we need to call GPU reset for a good reason as after reset the GPU is in unknown states. To avoid any more bad things to happen we now also unconditionaly reinitialize the GPU after reset. This patch fix few issues reported by different people regarding KMS & R6XX/RV7XX hw. Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-10-02 00:02:13 +08:00
static void r600_mc_program(struct radeon_device *rdev)
drm/radeon: introduce kernel modesetting for radeon hardware Add kernel modesetting support to radeon driver, use the ttm memory manager to manage memory and DRM/GEM to provide userspace API. In order to avoid backward compatibility issue and to allow clean design and code the radeon kernel modesetting use different code path than old radeon/drm driver. When kernel modesetting is enabled the IOCTL of radeon/drm driver are considered as invalid and an error message is printed in the log and they return failure. KMS enabled userspace will use new API to talk with the radeon/drm driver. The new API provide functions to create/destroy/share/mmap buffer object which are then managed by the kernel memory manager (here TTM). In order to submit command to the GPU the userspace provide a buffer holding the command stream, along this buffer userspace have to provide a list of buffer object used by the command stream. The kernel radeon driver will then place buffer in GPU accessible memory and will update command stream to reflect the position of the different buffers. The kernel will also perform security check on command stream provided by the user, we want to catch and forbid any illegal use of the GPU such as DMA into random system memory or into memory not owned by the process supplying the command stream. This part of the code is still incomplete and this why we propose that patch as a staging driver addition, future security might forbid current experimental userspace to run. This code support the following hardware : R1XX,R2XX,R3XX,R4XX,R5XX (radeon up to X1950). Works is underway to provide support for R6XX, R7XX and newer hardware (radeon from HD2XXX to HD4XXX). Authors: Jerome Glisse <jglisse@redhat.com> Dave Airlie <airlied@redhat.com> Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com> Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-06-05 20:42:42 +08:00
{
drm/radeon/kms: Fix R600/RV770 startup path & reset We were calling reset unconditionaly in the startup path this is bad we need to call GPU reset for a good reason as after reset the GPU is in unknown states. To avoid any more bad things to happen we now also unconditionaly reinitialize the GPU after reset. This patch fix few issues reported by different people regarding KMS & R6XX/RV7XX hw. Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-10-02 00:02:13 +08:00
struct rv515_mc_save save;
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
u32 tmp;
int i, j;
drm/radeon: introduce kernel modesetting for radeon hardware Add kernel modesetting support to radeon driver, use the ttm memory manager to manage memory and DRM/GEM to provide userspace API. In order to avoid backward compatibility issue and to allow clean design and code the radeon kernel modesetting use different code path than old radeon/drm driver. When kernel modesetting is enabled the IOCTL of radeon/drm driver are considered as invalid and an error message is printed in the log and they return failure. KMS enabled userspace will use new API to talk with the radeon/drm driver. The new API provide functions to create/destroy/share/mmap buffer object which are then managed by the kernel memory manager (here TTM). In order to submit command to the GPU the userspace provide a buffer holding the command stream, along this buffer userspace have to provide a list of buffer object used by the command stream. The kernel radeon driver will then place buffer in GPU accessible memory and will update command stream to reflect the position of the different buffers. The kernel will also perform security check on command stream provided by the user, we want to catch and forbid any illegal use of the GPU such as DMA into random system memory or into memory not owned by the process supplying the command stream. This part of the code is still incomplete and this why we propose that patch as a staging driver addition, future security might forbid current experimental userspace to run. This code support the following hardware : R1XX,R2XX,R3XX,R4XX,R5XX (radeon up to X1950). Works is underway to provide support for R6XX, R7XX and newer hardware (radeon from HD2XXX to HD4XXX). Authors: Jerome Glisse <jglisse@redhat.com> Dave Airlie <airlied@redhat.com> Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com> Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-06-05 20:42:42 +08:00
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
/* Initialize HDP */
for (i = 0, j = 0; i < 32; i++, j += 0x18) {
WREG32((0x2c14 + j), 0x00000000);
WREG32((0x2c18 + j), 0x00000000);
WREG32((0x2c1c + j), 0x00000000);
WREG32((0x2c20 + j), 0x00000000);
WREG32((0x2c24 + j), 0x00000000);
}
WREG32(HDP_REG_COHERENCY_FLUSH_CNTL, 0);
drm/radeon: introduce kernel modesetting for radeon hardware Add kernel modesetting support to radeon driver, use the ttm memory manager to manage memory and DRM/GEM to provide userspace API. In order to avoid backward compatibility issue and to allow clean design and code the radeon kernel modesetting use different code path than old radeon/drm driver. When kernel modesetting is enabled the IOCTL of radeon/drm driver are considered as invalid and an error message is printed in the log and they return failure. KMS enabled userspace will use new API to talk with the radeon/drm driver. The new API provide functions to create/destroy/share/mmap buffer object which are then managed by the kernel memory manager (here TTM). In order to submit command to the GPU the userspace provide a buffer holding the command stream, along this buffer userspace have to provide a list of buffer object used by the command stream. The kernel radeon driver will then place buffer in GPU accessible memory and will update command stream to reflect the position of the different buffers. The kernel will also perform security check on command stream provided by the user, we want to catch and forbid any illegal use of the GPU such as DMA into random system memory or into memory not owned by the process supplying the command stream. This part of the code is still incomplete and this why we propose that patch as a staging driver addition, future security might forbid current experimental userspace to run. This code support the following hardware : R1XX,R2XX,R3XX,R4XX,R5XX (radeon up to X1950). Works is underway to provide support for R6XX, R7XX and newer hardware (radeon from HD2XXX to HD4XXX). Authors: Jerome Glisse <jglisse@redhat.com> Dave Airlie <airlied@redhat.com> Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com> Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-06-05 20:42:42 +08:00
drm/radeon/kms: Fix R600/RV770 startup path & reset We were calling reset unconditionaly in the startup path this is bad we need to call GPU reset for a good reason as after reset the GPU is in unknown states. To avoid any more bad things to happen we now also unconditionaly reinitialize the GPU after reset. This patch fix few issues reported by different people regarding KMS & R6XX/RV7XX hw. Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-10-02 00:02:13 +08:00
rv515_mc_stop(rdev, &save);
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
if (r600_mc_wait_for_idle(rdev)) {
drm/radeon/kms: Fix R600/RV770 startup path & reset We were calling reset unconditionaly in the startup path this is bad we need to call GPU reset for a good reason as after reset the GPU is in unknown states. To avoid any more bad things to happen we now also unconditionaly reinitialize the GPU after reset. This patch fix few issues reported by different people regarding KMS & R6XX/RV7XX hw. Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-10-02 00:02:13 +08:00
dev_warn(rdev->dev, "Wait for MC idle timedout !\n");
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
}
drm/radeon/kms: Fix R600/RV770 startup path & reset We were calling reset unconditionaly in the startup path this is bad we need to call GPU reset for a good reason as after reset the GPU is in unknown states. To avoid any more bad things to happen we now also unconditionaly reinitialize the GPU after reset. This patch fix few issues reported by different people regarding KMS & R6XX/RV7XX hw. Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-10-02 00:02:13 +08:00
/* Lockout access through VGA aperture (doesn't exist before R600) */
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
WREG32(VGA_HDP_CONTROL, VGA_MEMORY_DISABLE);
/* Update configuration */
drm/radeon/kms: Fix AGP support for R600/RV770 family (v2) For AGP to work unmapped access must cover VRAM & AGP as AGP is treated like VRAM by the GPU (ie physical address). This patch properly setup the virtual memory system aperture to cover AGP if AGP is enabled. It seems that there is memory corruption after resume when using AGP (RV770 seems unaffected thought). Version 2 just fix merge issue with updated AGP fallback patch. Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-10-07 01:04:30 +08:00
if (rdev->flags & RADEON_IS_AGP) {
if (rdev->mc.vram_start < rdev->mc.gtt_start) {
/* VRAM before AGP */
WREG32(MC_VM_SYSTEM_APERTURE_LOW_ADDR,
rdev->mc.vram_start >> 12);
WREG32(MC_VM_SYSTEM_APERTURE_HIGH_ADDR,
rdev->mc.gtt_end >> 12);
} else {
/* VRAM after AGP */
WREG32(MC_VM_SYSTEM_APERTURE_LOW_ADDR,
rdev->mc.gtt_start >> 12);
WREG32(MC_VM_SYSTEM_APERTURE_HIGH_ADDR,
rdev->mc.vram_end >> 12);
}
} else {
WREG32(MC_VM_SYSTEM_APERTURE_LOW_ADDR, rdev->mc.vram_start >> 12);
WREG32(MC_VM_SYSTEM_APERTURE_HIGH_ADDR, rdev->mc.vram_end >> 12);
}
drm/radeon/kms: allocate vram scratch page on 6xx+ The vram scratch was originally only used on some 7xx asics to work around a hw bug. Allocate the scratch page on all 6xx+ radeons and set the MC_VM_SYSTEM_APERTURE_DEFAULT_ADDR to point to it. We shouldn't ever hit it since we limit the system aperture to vram or vram and AGP, but better safe than sorry. Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-10-28 22:30:02 +08:00
WREG32(MC_VM_SYSTEM_APERTURE_DEFAULT_ADDR, rdev->vram_scratch.gpu_addr >> 12);
drm/radeon/kms: Fix AGP support for R600/RV770 family (v2) For AGP to work unmapped access must cover VRAM & AGP as AGP is treated like VRAM by the GPU (ie physical address). This patch properly setup the virtual memory system aperture to cover AGP if AGP is enabled. It seems that there is memory corruption after resume when using AGP (RV770 seems unaffected thought). Version 2 just fix merge issue with updated AGP fallback patch. Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-10-07 01:04:30 +08:00
tmp = ((rdev->mc.vram_end >> 24) & 0xFFFF) << 16;
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
tmp |= ((rdev->mc.vram_start >> 24) & 0xFFFF);
WREG32(MC_VM_FB_LOCATION, tmp);
WREG32(HDP_NONSURFACE_BASE, (rdev->mc.vram_start >> 8));
WREG32(HDP_NONSURFACE_INFO, (2 << 7));
drm/radeon/kms: Force HDP_NONSURF to maximum size HDP non surface should cover the whole VRAM but we were misscomputing the size and we endup in some case not covering the VRAM at all (if VRAM size were > 1G). Covering more than the VRAM size shouldn't be an issue. Fix : https://bugs.freedesktop.org/show_bug.cgi?id=28016 [airlied: add evergreen fix] Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-06-04 01:34:48 +08:00
WREG32(HDP_NONSURFACE_SIZE, 0x3FFFFFFF);
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
if (rdev->flags & RADEON_IS_AGP) {
drm/radeon/kms: Fix AGP support for R600/RV770 family (v2) For AGP to work unmapped access must cover VRAM & AGP as AGP is treated like VRAM by the GPU (ie physical address). This patch properly setup the virtual memory system aperture to cover AGP if AGP is enabled. It seems that there is memory corruption after resume when using AGP (RV770 seems unaffected thought). Version 2 just fix merge issue with updated AGP fallback patch. Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-10-07 01:04:30 +08:00
WREG32(MC_VM_AGP_TOP, rdev->mc.gtt_end >> 22);
WREG32(MC_VM_AGP_BOT, rdev->mc.gtt_start >> 22);
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
WREG32(MC_VM_AGP_BASE, rdev->mc.agp_base >> 22);
} else {
WREG32(MC_VM_AGP_BASE, 0);
WREG32(MC_VM_AGP_TOP, 0x0FFFFFFF);
WREG32(MC_VM_AGP_BOT, 0x0FFFFFFF);
}
if (r600_mc_wait_for_idle(rdev)) {
drm/radeon/kms: Fix R600/RV770 startup path & reset We were calling reset unconditionaly in the startup path this is bad we need to call GPU reset for a good reason as after reset the GPU is in unknown states. To avoid any more bad things to happen we now also unconditionaly reinitialize the GPU after reset. This patch fix few issues reported by different people regarding KMS & R6XX/RV7XX hw. Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-10-02 00:02:13 +08:00
dev_warn(rdev->dev, "Wait for MC idle timedout !\n");
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
}
drm/radeon/kms: Fix R600/RV770 startup path & reset We were calling reset unconditionaly in the startup path this is bad we need to call GPU reset for a good reason as after reset the GPU is in unknown states. To avoid any more bad things to happen we now also unconditionaly reinitialize the GPU after reset. This patch fix few issues reported by different people regarding KMS & R6XX/RV7XX hw. Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-10-02 00:02:13 +08:00
rv515_mc_resume(rdev, &save);
drm/radeon/kms: disable VGA rendering engine before taking over VRAM Before we use any of VRAM, we need to disable the VGA rendering engine, this render text mode into a graphical framebuffer for scanout, however it does this on vblank, and can end up overwriting the GART table and r600 shader objects. Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-18 12:16:38 +08:00
/* we need to own VRAM, so turn off the VGA renderer here
* to stop it overwriting our objects */
drm/radeon/kms: Convert RV515 to new init path and associated cleanup Convert the rv515 asic support to new init path also add an explanation in radeon.h about the new init path. There is also few cleanups associated with this change (others asic calling rv515 helper functions). Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@linux.ie>
2009-09-29 00:34:43 +08:00
rv515_vga_render_disable(rdev);
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
}
drm/radeon/kms: simplify memory controller setup V2 Get rid of _location and use _start/_end also simplify the computation of vram_start|end & gtt_start|end. For R1XX-R2XX we place VRAM at the same address of PCI aperture, those GPU shouldn't have much memory and seems to behave better when setup that way. For R3XX and newer we place VRAM at 0. For R6XX-R7XX AGP we place VRAM before or after AGP aperture this might limit to limit the VRAM size but it's very unlikely. For IGP we don't change the VRAM placement. Tested on (compiz,quake3,suspend/resume): PCI/PCIE:RV280,R420,RV515,RV570,RV610,RV710 AGP:RV100,RV280,R420,RV350,RV620(RPB*),RV730 IGP:RS480(RPB*),RS690,RS780(RPB*),RS880 RPB: resume previously broken V2 correct commit message to reflect more accurately the bug and move VRAM placement to 0 for most of the GPU to avoid limiting VRAM. Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-02-18 05:54:29 +08:00
/**
* r600_vram_gtt_location - try to find VRAM & GTT location
* @rdev: radeon device structure holding all necessary informations
* @mc: memory controller structure holding memory informations
*
* Function will place try to place VRAM at same place as in CPU (PCI)
* address space as some GPU seems to have issue when we reprogram at
* different address space.
*
* If there is not enough space to fit the unvisible VRAM after the
* aperture then we limit the VRAM size to the aperture.
*
* If we are using AGP then place VRAM adjacent to AGP aperture are we need
* them to be in one from GPU point of view so that we can program GPU to
* catch access outside them (weird GPU policy see ??).
*
* This function will never fails, worst case are limiting VRAM or GTT.
*
* Note: GTT start, end, size should be initialized before calling this
* function on AGP platform.
*/
drm/radeon/kms: move r7xx/evergreen to its own vram_gtt setup function MC_VM_FB_LOCATION is at a different offset between r6xx and r7xx/evergreen. The location is needed for vram setup on fusion chips. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-11-23 06:56:26 +08:00
static void r600_vram_gtt_location(struct radeon_device *rdev, struct radeon_mc *mc)
drm/radeon/kms: simplify memory controller setup V2 Get rid of _location and use _start/_end also simplify the computation of vram_start|end & gtt_start|end. For R1XX-R2XX we place VRAM at the same address of PCI aperture, those GPU shouldn't have much memory and seems to behave better when setup that way. For R3XX and newer we place VRAM at 0. For R6XX-R7XX AGP we place VRAM before or after AGP aperture this might limit to limit the VRAM size but it's very unlikely. For IGP we don't change the VRAM placement. Tested on (compiz,quake3,suspend/resume): PCI/PCIE:RV280,R420,RV515,RV570,RV610,RV710 AGP:RV100,RV280,R420,RV350,RV620(RPB*),RV730 IGP:RS480(RPB*),RS690,RS780(RPB*),RS880 RPB: resume previously broken V2 correct commit message to reflect more accurately the bug and move VRAM placement to 0 for most of the GPU to avoid limiting VRAM. Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-02-18 05:54:29 +08:00
{
u64 size_bf, size_af;
if (mc->mc_vram_size > 0xE0000000) {
/* leave room for at least 512M GTT */
dev_warn(rdev->dev, "limiting VRAM\n");
mc->real_vram_size = 0xE0000000;
mc->mc_vram_size = 0xE0000000;
}
if (rdev->flags & RADEON_IS_AGP) {
size_bf = mc->gtt_start;
radeon: fix r600/agp when vram is after AGP (v3) If AGP is placed in the middle, the size_af is off-by-one, it results in VRAM being placed at 0x7fffffff instead of 0x8000000. v2: fix the vram_start setup. v3: also fix r7xx & newer ASIC Reported-by: russiane39 on #radeon Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2012-04-18 04:51:38 +08:00
size_af = 0xFFFFFFFF - mc->gtt_end;
drm/radeon/kms: simplify memory controller setup V2 Get rid of _location and use _start/_end also simplify the computation of vram_start|end & gtt_start|end. For R1XX-R2XX we place VRAM at the same address of PCI aperture, those GPU shouldn't have much memory and seems to behave better when setup that way. For R3XX and newer we place VRAM at 0. For R6XX-R7XX AGP we place VRAM before or after AGP aperture this might limit to limit the VRAM size but it's very unlikely. For IGP we don't change the VRAM placement. Tested on (compiz,quake3,suspend/resume): PCI/PCIE:RV280,R420,RV515,RV570,RV610,RV710 AGP:RV100,RV280,R420,RV350,RV620(RPB*),RV730 IGP:RS480(RPB*),RS690,RS780(RPB*),RS880 RPB: resume previously broken V2 correct commit message to reflect more accurately the bug and move VRAM placement to 0 for most of the GPU to avoid limiting VRAM. Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-02-18 05:54:29 +08:00
if (size_bf > size_af) {
if (mc->mc_vram_size > size_bf) {
dev_warn(rdev->dev, "limiting VRAM\n");
mc->real_vram_size = size_bf;
mc->mc_vram_size = size_bf;
}
mc->vram_start = mc->gtt_start - mc->mc_vram_size;
} else {
if (mc->mc_vram_size > size_af) {
dev_warn(rdev->dev, "limiting VRAM\n");
mc->real_vram_size = size_af;
mc->mc_vram_size = size_af;
}
radeon: fix r600/agp when vram is after AGP (v3) If AGP is placed in the middle, the size_af is off-by-one, it results in VRAM being placed at 0x7fffffff instead of 0x8000000. v2: fix the vram_start setup. v3: also fix r7xx & newer ASIC Reported-by: russiane39 on #radeon Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2012-04-18 04:51:38 +08:00
mc->vram_start = mc->gtt_end + 1;
drm/radeon/kms: simplify memory controller setup V2 Get rid of _location and use _start/_end also simplify the computation of vram_start|end & gtt_start|end. For R1XX-R2XX we place VRAM at the same address of PCI aperture, those GPU shouldn't have much memory and seems to behave better when setup that way. For R3XX and newer we place VRAM at 0. For R6XX-R7XX AGP we place VRAM before or after AGP aperture this might limit to limit the VRAM size but it's very unlikely. For IGP we don't change the VRAM placement. Tested on (compiz,quake3,suspend/resume): PCI/PCIE:RV280,R420,RV515,RV570,RV610,RV710 AGP:RV100,RV280,R420,RV350,RV620(RPB*),RV730 IGP:RS480(RPB*),RS690,RS780(RPB*),RS880 RPB: resume previously broken V2 correct commit message to reflect more accurately the bug and move VRAM placement to 0 for most of the GPU to avoid limiting VRAM. Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-02-18 05:54:29 +08:00
}
mc->vram_end = mc->vram_start + mc->mc_vram_size - 1;
dev_info(rdev->dev, "VRAM: %lluM 0x%08llX - 0x%08llX (%lluM used)\n",
mc->mc_vram_size >> 20, mc->vram_start,
mc->vram_end, mc->real_vram_size >> 20);
} else {
u64 base = 0;
drm/radeon/kms: fix vram base calculation on rs780/rs880 Avoid overflowing a 32 bit value. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Cc: stable@kernel.org Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-12-04 03:37:22 +08:00
if (rdev->flags & RADEON_IS_IGP) {
base = RREG32(MC_VM_FB_LOCATION) & 0xFFFF;
base <<= 24;
}
drm/radeon/kms: simplify memory controller setup V2 Get rid of _location and use _start/_end also simplify the computation of vram_start|end & gtt_start|end. For R1XX-R2XX we place VRAM at the same address of PCI aperture, those GPU shouldn't have much memory and seems to behave better when setup that way. For R3XX and newer we place VRAM at 0. For R6XX-R7XX AGP we place VRAM before or after AGP aperture this might limit to limit the VRAM size but it's very unlikely. For IGP we don't change the VRAM placement. Tested on (compiz,quake3,suspend/resume): PCI/PCIE:RV280,R420,RV515,RV570,RV610,RV710 AGP:RV100,RV280,R420,RV350,RV620(RPB*),RV730 IGP:RS480(RPB*),RS690,RS780(RPB*),RS880 RPB: resume previously broken V2 correct commit message to reflect more accurately the bug and move VRAM placement to 0 for most of the GPU to avoid limiting VRAM. Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-02-18 05:54:29 +08:00
radeon_vram_location(rdev, &rdev->mc, base);
drm/radeon/kms: fix gtt MC base alignment on rs4xx/rs690/rs740 asics The asics in question have the following requirements with regard to their gart setups: 1. The GART aperture size has to be in the form of 2^X bytes, where X is from 25 to 31 2. The GART aperture MC base has to be aligned to a boundary equal to the size of the aperture. 3. The GART page table has to be aligned to the boundary equal to the size of the table. 4. The GART page table size is: table_entry_size * (aperture_size / page_size) 5. The GART page table has to be allocated in non-paged, non-cached, contiguous system memory. This patch takes care 2. The rest should already be handled properly. This fixes a regression noticed by: Torsten Kaiser <just.for.lkml@googlemail.com> Tested-by: Torsten Kaiser <just.for.lkml@googlemail.com> Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-07-15 22:51:10 +08:00
rdev->mc.gtt_base_align = 0;
drm/radeon/kms: simplify memory controller setup V2 Get rid of _location and use _start/_end also simplify the computation of vram_start|end & gtt_start|end. For R1XX-R2XX we place VRAM at the same address of PCI aperture, those GPU shouldn't have much memory and seems to behave better when setup that way. For R3XX and newer we place VRAM at 0. For R6XX-R7XX AGP we place VRAM before or after AGP aperture this might limit to limit the VRAM size but it's very unlikely. For IGP we don't change the VRAM placement. Tested on (compiz,quake3,suspend/resume): PCI/PCIE:RV280,R420,RV515,RV570,RV610,RV710 AGP:RV100,RV280,R420,RV350,RV620(RPB*),RV730 IGP:RS480(RPB*),RS690,RS780(RPB*),RS880 RPB: resume previously broken V2 correct commit message to reflect more accurately the bug and move VRAM placement to 0 for most of the GPU to avoid limiting VRAM. Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-02-18 05:54:29 +08:00
radeon_gtt_location(rdev, mc);
}
}
drm/radeon: Mark all possible functions / structs as static Let's allow GCC to optimize better. This exposed some five unused functions, but this patch doesn't remove them. Signed-off-by: Lauri Kasanen <cand@gmx.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2012-09-01 01:43:50 +08:00
static int r600_mc_init(struct radeon_device *rdev)
drm/radeon: introduce kernel modesetting for radeon hardware Add kernel modesetting support to radeon driver, use the ttm memory manager to manage memory and DRM/GEM to provide userspace API. In order to avoid backward compatibility issue and to allow clean design and code the radeon kernel modesetting use different code path than old radeon/drm driver. When kernel modesetting is enabled the IOCTL of radeon/drm driver are considered as invalid and an error message is printed in the log and they return failure. KMS enabled userspace will use new API to talk with the radeon/drm driver. The new API provide functions to create/destroy/share/mmap buffer object which are then managed by the kernel memory manager (here TTM). In order to submit command to the GPU the userspace provide a buffer holding the command stream, along this buffer userspace have to provide a list of buffer object used by the command stream. The kernel radeon driver will then place buffer in GPU accessible memory and will update command stream to reflect the position of the different buffers. The kernel will also perform security check on command stream provided by the user, we want to catch and forbid any illegal use of the GPU such as DMA into random system memory or into memory not owned by the process supplying the command stream. This part of the code is still incomplete and this why we propose that patch as a staging driver addition, future security might forbid current experimental userspace to run. This code support the following hardware : R1XX,R2XX,R3XX,R4XX,R5XX (radeon up to X1950). Works is underway to provide support for R6XX, R7XX and newer hardware (radeon from HD2XXX to HD4XXX). Authors: Jerome Glisse <jglisse@redhat.com> Dave Airlie <airlied@redhat.com> Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com> Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-06-05 20:42:42 +08:00
{
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
u32 tmp;
drm/radeon/kms: fix vram_width calculation on r6xx/r7xx Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-10-20 05:23:33 +08:00
int chansize, numchan;
drm/radeon: introduce kernel modesetting for radeon hardware Add kernel modesetting support to radeon driver, use the ttm memory manager to manage memory and DRM/GEM to provide userspace API. In order to avoid backward compatibility issue and to allow clean design and code the radeon kernel modesetting use different code path than old radeon/drm driver. When kernel modesetting is enabled the IOCTL of radeon/drm driver are considered as invalid and an error message is printed in the log and they return failure. KMS enabled userspace will use new API to talk with the radeon/drm driver. The new API provide functions to create/destroy/share/mmap buffer object which are then managed by the kernel memory manager (here TTM). In order to submit command to the GPU the userspace provide a buffer holding the command stream, along this buffer userspace have to provide a list of buffer object used by the command stream. The kernel radeon driver will then place buffer in GPU accessible memory and will update command stream to reflect the position of the different buffers. The kernel will also perform security check on command stream provided by the user, we want to catch and forbid any illegal use of the GPU such as DMA into random system memory or into memory not owned by the process supplying the command stream. This part of the code is still incomplete and this why we propose that patch as a staging driver addition, future security might forbid current experimental userspace to run. This code support the following hardware : R1XX,R2XX,R3XX,R4XX,R5XX (radeon up to X1950). Works is underway to provide support for R6XX, R7XX and newer hardware (radeon from HD2XXX to HD4XXX). Authors: Jerome Glisse <jglisse@redhat.com> Dave Airlie <airlied@redhat.com> Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com> Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-06-05 20:42:42 +08:00
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
/* Get VRAM informations */
drm/radeon: introduce kernel modesetting for radeon hardware Add kernel modesetting support to radeon driver, use the ttm memory manager to manage memory and DRM/GEM to provide userspace API. In order to avoid backward compatibility issue and to allow clean design and code the radeon kernel modesetting use different code path than old radeon/drm driver. When kernel modesetting is enabled the IOCTL of radeon/drm driver are considered as invalid and an error message is printed in the log and they return failure. KMS enabled userspace will use new API to talk with the radeon/drm driver. The new API provide functions to create/destroy/share/mmap buffer object which are then managed by the kernel memory manager (here TTM). In order to submit command to the GPU the userspace provide a buffer holding the command stream, along this buffer userspace have to provide a list of buffer object used by the command stream. The kernel radeon driver will then place buffer in GPU accessible memory and will update command stream to reflect the position of the different buffers. The kernel will also perform security check on command stream provided by the user, we want to catch and forbid any illegal use of the GPU such as DMA into random system memory or into memory not owned by the process supplying the command stream. This part of the code is still incomplete and this why we propose that patch as a staging driver addition, future security might forbid current experimental userspace to run. This code support the following hardware : R1XX,R2XX,R3XX,R4XX,R5XX (radeon up to X1950). Works is underway to provide support for R6XX, R7XX and newer hardware (radeon from HD2XXX to HD4XXX). Authors: Jerome Glisse <jglisse@redhat.com> Dave Airlie <airlied@redhat.com> Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com> Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-06-05 20:42:42 +08:00
rdev->mc.vram_is_ddr = true;
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
tmp = RREG32(RAMCFG);
if (tmp & CHANSIZE_OVERRIDE) {
drm/radeon: introduce kernel modesetting for radeon hardware Add kernel modesetting support to radeon driver, use the ttm memory manager to manage memory and DRM/GEM to provide userspace API. In order to avoid backward compatibility issue and to allow clean design and code the radeon kernel modesetting use different code path than old radeon/drm driver. When kernel modesetting is enabled the IOCTL of radeon/drm driver are considered as invalid and an error message is printed in the log and they return failure. KMS enabled userspace will use new API to talk with the radeon/drm driver. The new API provide functions to create/destroy/share/mmap buffer object which are then managed by the kernel memory manager (here TTM). In order to submit command to the GPU the userspace provide a buffer holding the command stream, along this buffer userspace have to provide a list of buffer object used by the command stream. The kernel radeon driver will then place buffer in GPU accessible memory and will update command stream to reflect the position of the different buffers. The kernel will also perform security check on command stream provided by the user, we want to catch and forbid any illegal use of the GPU such as DMA into random system memory or into memory not owned by the process supplying the command stream. This part of the code is still incomplete and this why we propose that patch as a staging driver addition, future security might forbid current experimental userspace to run. This code support the following hardware : R1XX,R2XX,R3XX,R4XX,R5XX (radeon up to X1950). Works is underway to provide support for R6XX, R7XX and newer hardware (radeon from HD2XXX to HD4XXX). Authors: Jerome Glisse <jglisse@redhat.com> Dave Airlie <airlied@redhat.com> Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com> Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-06-05 20:42:42 +08:00
chansize = 16;
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
} else if (tmp & CHANSIZE_MASK) {
drm/radeon: introduce kernel modesetting for radeon hardware Add kernel modesetting support to radeon driver, use the ttm memory manager to manage memory and DRM/GEM to provide userspace API. In order to avoid backward compatibility issue and to allow clean design and code the radeon kernel modesetting use different code path than old radeon/drm driver. When kernel modesetting is enabled the IOCTL of radeon/drm driver are considered as invalid and an error message is printed in the log and they return failure. KMS enabled userspace will use new API to talk with the radeon/drm driver. The new API provide functions to create/destroy/share/mmap buffer object which are then managed by the kernel memory manager (here TTM). In order to submit command to the GPU the userspace provide a buffer holding the command stream, along this buffer userspace have to provide a list of buffer object used by the command stream. The kernel radeon driver will then place buffer in GPU accessible memory and will update command stream to reflect the position of the different buffers. The kernel will also perform security check on command stream provided by the user, we want to catch and forbid any illegal use of the GPU such as DMA into random system memory or into memory not owned by the process supplying the command stream. This part of the code is still incomplete and this why we propose that patch as a staging driver addition, future security might forbid current experimental userspace to run. This code support the following hardware : R1XX,R2XX,R3XX,R4XX,R5XX (radeon up to X1950). Works is underway to provide support for R6XX, R7XX and newer hardware (radeon from HD2XXX to HD4XXX). Authors: Jerome Glisse <jglisse@redhat.com> Dave Airlie <airlied@redhat.com> Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com> Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-06-05 20:42:42 +08:00
chansize = 64;
} else {
chansize = 32;
}
drm/radeon/kms: fix vram_width calculation on r6xx/r7xx Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-10-20 05:23:33 +08:00
tmp = RREG32(CHMAP);
switch ((tmp & NOOFCHAN_MASK) >> NOOFCHAN_SHIFT) {
case 0:
default:
numchan = 1;
break;
case 1:
numchan = 2;
break;
case 2:
numchan = 4;
break;
case 3:
numchan = 8;
break;
drm/radeon: introduce kernel modesetting for radeon hardware Add kernel modesetting support to radeon driver, use the ttm memory manager to manage memory and DRM/GEM to provide userspace API. In order to avoid backward compatibility issue and to allow clean design and code the radeon kernel modesetting use different code path than old radeon/drm driver. When kernel modesetting is enabled the IOCTL of radeon/drm driver are considered as invalid and an error message is printed in the log and they return failure. KMS enabled userspace will use new API to talk with the radeon/drm driver. The new API provide functions to create/destroy/share/mmap buffer object which are then managed by the kernel memory manager (here TTM). In order to submit command to the GPU the userspace provide a buffer holding the command stream, along this buffer userspace have to provide a list of buffer object used by the command stream. The kernel radeon driver will then place buffer in GPU accessible memory and will update command stream to reflect the position of the different buffers. The kernel will also perform security check on command stream provided by the user, we want to catch and forbid any illegal use of the GPU such as DMA into random system memory or into memory not owned by the process supplying the command stream. This part of the code is still incomplete and this why we propose that patch as a staging driver addition, future security might forbid current experimental userspace to run. This code support the following hardware : R1XX,R2XX,R3XX,R4XX,R5XX (radeon up to X1950). Works is underway to provide support for R6XX, R7XX and newer hardware (radeon from HD2XXX to HD4XXX). Authors: Jerome Glisse <jglisse@redhat.com> Dave Airlie <airlied@redhat.com> Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com> Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-06-05 20:42:42 +08:00
}
drm/radeon/kms: fix vram_width calculation on r6xx/r7xx Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-10-20 05:23:33 +08:00
rdev->mc.vram_width = numchan * chansize;
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
/* Could aper size report 0 ? */
drm: Remove drm_resource wrappers Remove the drm_resource wrappers and directly use the actual PCI and/or platform functions in their place. [airlied: fixup nouveau properly to build] Signed-off-by: Jordan Crouse <jcrouse@codeaurora.org> Reviewed-by: Matt Turner <mattst88@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-05-28 03:40:24 +08:00
rdev->mc.aper_base = pci_resource_start(rdev->pdev, 0);
rdev->mc.aper_size = pci_resource_len(rdev->pdev, 0);
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
/* Setup GPU memory space */
rdev->mc.mc_vram_size = RREG32(CONFIG_MEMSIZE);
rdev->mc.real_vram_size = RREG32(CONFIG_MEMSIZE);
drm/radeon/kms: force pinning buffer into visible VRAM This patch properly set visible VRAM and enforce any pinned buffer to be into visible VRAM. We might later add a flag to release this constraint for some newer hw more clever than previous. Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-02-19 22:33:54 +08:00
rdev->mc.visible_vram_size = rdev->mc.aper_size;
drm/radeon/kms: simplify memory controller setup V2 Get rid of _location and use _start/_end also simplify the computation of vram_start|end & gtt_start|end. For R1XX-R2XX we place VRAM at the same address of PCI aperture, those GPU shouldn't have much memory and seems to behave better when setup that way. For R3XX and newer we place VRAM at 0. For R6XX-R7XX AGP we place VRAM before or after AGP aperture this might limit to limit the VRAM size but it's very unlikely. For IGP we don't change the VRAM placement. Tested on (compiz,quake3,suspend/resume): PCI/PCIE:RV280,R420,RV515,RV570,RV610,RV710 AGP:RV100,RV280,R420,RV350,RV620(RPB*),RV730 IGP:RS480(RPB*),RS690,RS780(RPB*),RS880 RPB: resume previously broken V2 correct commit message to reflect more accurately the bug and move VRAM placement to 0 for most of the GPU to avoid limiting VRAM. Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-02-18 05:54:29 +08:00
r600_vram_gtt_location(rdev, &rdev->mc);
drm/radeon/kms: display watermark fixes - rs780/880 were using the wrong bandwidth functions - convert r1xx-r4xx to use the same pm sclk/mclk structs as r5xx+ - move bandwidth setup to a common function Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-03-17 08:54:38 +08:00
drm/radeon/kms/igp: fix possible divide by 0 in bandwidth code (v2) Some IGP systems specify the system memory clock in the Firmware table rather than the IGP info table. Check both and make sure we have a value system memory clock value. v2: make sure rs690_pm_info is called on rs780/rs880 as well. fixes a regression since 07d4190327b02ab3aaad25a2d168f79d92e8f8c2. Reported-by: Markus Trippelsdorf <markus@trippelsdorf.de> Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Tested-by: Markus Trippelsdorf <markus@trippelsdorf.de> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-07-01 00:02:03 +08:00
if (rdev->flags & RADEON_IS_IGP) {
rs690_pm_info(rdev);
drm/radeon/kms: detect sideport memory on IGP chips This detects if the sideport memory is enabled and if it is VRAM is evicted on suspend/resume. This should fix s/r issues on some IGPs. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-01-06 00:27:29 +08:00
rdev->mc.igp_sideport_enabled = radeon_atombios_sideport_present(rdev);
drm/radeon/kms/igp: fix possible divide by 0 in bandwidth code (v2) Some IGP systems specify the system memory clock in the Firmware table rather than the IGP info table. Check both and make sure we have a value system memory clock value. v2: make sure rs690_pm_info is called on rs780/rs880 as well. fixes a regression since 07d4190327b02ab3aaad25a2d168f79d92e8f8c2. Reported-by: Markus Trippelsdorf <markus@trippelsdorf.de> Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Tested-by: Markus Trippelsdorf <markus@trippelsdorf.de> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-07-01 00:02:03 +08:00
}
drm/radeon/kms: display watermark fixes - rs780/880 were using the wrong bandwidth functions - convert r1xx-r4xx to use the same pm sclk/mclk structs as r5xx+ - move bandwidth setup to a common function Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-03-17 08:54:38 +08:00
radeon_update_bandwidth_info(rdev);
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
return 0;
drm/radeon: introduce kernel modesetting for radeon hardware Add kernel modesetting support to radeon driver, use the ttm memory manager to manage memory and DRM/GEM to provide userspace API. In order to avoid backward compatibility issue and to allow clean design and code the radeon kernel modesetting use different code path than old radeon/drm driver. When kernel modesetting is enabled the IOCTL of radeon/drm driver are considered as invalid and an error message is printed in the log and they return failure. KMS enabled userspace will use new API to talk with the radeon/drm driver. The new API provide functions to create/destroy/share/mmap buffer object which are then managed by the kernel memory manager (here TTM). In order to submit command to the GPU the userspace provide a buffer holding the command stream, along this buffer userspace have to provide a list of buffer object used by the command stream. The kernel radeon driver will then place buffer in GPU accessible memory and will update command stream to reflect the position of the different buffers. The kernel will also perform security check on command stream provided by the user, we want to catch and forbid any illegal use of the GPU such as DMA into random system memory or into memory not owned by the process supplying the command stream. This part of the code is still incomplete and this why we propose that patch as a staging driver addition, future security might forbid current experimental userspace to run. This code support the following hardware : R1XX,R2XX,R3XX,R4XX,R5XX (radeon up to X1950). Works is underway to provide support for R6XX, R7XX and newer hardware (radeon from HD2XXX to HD4XXX). Authors: Jerome Glisse <jglisse@redhat.com> Dave Airlie <airlied@redhat.com> Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com> Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-06-05 20:42:42 +08:00
}
drm/radeon/kms: allocate vram scratch page on 6xx+ The vram scratch was originally only used on some 7xx asics to work around a hw bug. Allocate the scratch page on all 6xx+ radeons and set the MC_VM_SYSTEM_APERTURE_DEFAULT_ADDR to point to it. We shouldn't ever hit it since we limit the system aperture to vram or vram and AGP, but better safe than sorry. Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-10-28 22:30:02 +08:00
int r600_vram_scratch_init(struct radeon_device *rdev)
{
int r;
if (rdev->vram_scratch.robj == NULL) {
r = radeon_bo_create(rdev, RADEON_GPU_PAGE_SIZE,
PAGE_SIZE, true, RADEON_GEM_DOMAIN_VRAM,
drm/radeon: add PRIME support (v2) This adds prime->fd and fd->prime support to radeon. It passes the sg object to ttm and then populates the gart entries using it. Compile tested only. v2: stub kmap + use new helpers + add reimporting Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Reviewed-by: Jerome Glisse <jglisse@redhat.com> Reviewed-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2012-05-11 06:33:13 +08:00
NULL, &rdev->vram_scratch.robj);
drm/radeon/kms: allocate vram scratch page on 6xx+ The vram scratch was originally only used on some 7xx asics to work around a hw bug. Allocate the scratch page on all 6xx+ radeons and set the MC_VM_SYSTEM_APERTURE_DEFAULT_ADDR to point to it. We shouldn't ever hit it since we limit the system aperture to vram or vram and AGP, but better safe than sorry. Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-10-28 22:30:02 +08:00
if (r) {
return r;
}
}
r = radeon_bo_reserve(rdev->vram_scratch.robj, false);
if (unlikely(r != 0))
return r;
r = radeon_bo_pin(rdev->vram_scratch.robj,
RADEON_GEM_DOMAIN_VRAM, &rdev->vram_scratch.gpu_addr);
if (r) {
radeon_bo_unreserve(rdev->vram_scratch.robj);
return r;
}
r = radeon_bo_kmap(rdev->vram_scratch.robj,
(void **)&rdev->vram_scratch.ptr);
if (r)
radeon_bo_unpin(rdev->vram_scratch.robj);
radeon_bo_unreserve(rdev->vram_scratch.robj);
return r;
}
void r600_vram_scratch_fini(struct radeon_device *rdev)
{
int r;
if (rdev->vram_scratch.robj == NULL) {
return;
}
r = radeon_bo_reserve(rdev->vram_scratch.robj, false);
if (likely(r == 0)) {
radeon_bo_kunmap(rdev->vram_scratch.robj);
radeon_bo_unpin(rdev->vram_scratch.robj);
radeon_bo_unreserve(rdev->vram_scratch.robj);
}
radeon_bo_unref(&rdev->vram_scratch.robj);
}
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
/* We doesn't check that the GPU really needs a reset we simply do the
* reset, it's up to the caller to determine if the GPU needs one. We
* might add an helper function to check that.
*/
drm/radeon: switch to a finer grained reset for r6xx/7xx No change in functionality as we currently set all the reset flags. Reviewed-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2013-01-04 01:20:35 +08:00
static void r600_gpu_soft_reset_gfx(struct radeon_device *rdev)
drm/radeon: introduce kernel modesetting for radeon hardware Add kernel modesetting support to radeon driver, use the ttm memory manager to manage memory and DRM/GEM to provide userspace API. In order to avoid backward compatibility issue and to allow clean design and code the radeon kernel modesetting use different code path than old radeon/drm driver. When kernel modesetting is enabled the IOCTL of radeon/drm driver are considered as invalid and an error message is printed in the log and they return failure. KMS enabled userspace will use new API to talk with the radeon/drm driver. The new API provide functions to create/destroy/share/mmap buffer object which are then managed by the kernel memory manager (here TTM). In order to submit command to the GPU the userspace provide a buffer holding the command stream, along this buffer userspace have to provide a list of buffer object used by the command stream. The kernel radeon driver will then place buffer in GPU accessible memory and will update command stream to reflect the position of the different buffers. The kernel will also perform security check on command stream provided by the user, we want to catch and forbid any illegal use of the GPU such as DMA into random system memory or into memory not owned by the process supplying the command stream. This part of the code is still incomplete and this why we propose that patch as a staging driver addition, future security might forbid current experimental userspace to run. This code support the following hardware : R1XX,R2XX,R3XX,R4XX,R5XX (radeon up to X1950). Works is underway to provide support for R6XX, R7XX and newer hardware (radeon from HD2XXX to HD4XXX). Authors: Jerome Glisse <jglisse@redhat.com> Dave Airlie <airlied@redhat.com> Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com> Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-06-05 20:42:42 +08:00
{
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
u32 grbm_busy_mask = S_008010_VC_BUSY(1) | S_008010_VGT_BUSY_NO_DMA(1) |
S_008010_VGT_BUSY(1) | S_008010_TA03_BUSY(1) |
S_008010_TC_BUSY(1) | S_008010_SX_BUSY(1) |
S_008010_SH_BUSY(1) | S_008010_SPI03_BUSY(1) |
S_008010_SMX_BUSY(1) | S_008010_SC_BUSY(1) |
S_008010_PA_BUSY(1) | S_008010_DB03_BUSY(1) |
S_008010_CR_BUSY(1) | S_008010_CB03_BUSY(1) |
S_008010_GUI_ACTIVE(1);
u32 grbm2_busy_mask = S_008014_SPI0_BUSY(1) | S_008014_SPI1_BUSY(1) |
S_008014_SPI2_BUSY(1) | S_008014_SPI3_BUSY(1) |
S_008014_TA0_BUSY(1) | S_008014_TA1_BUSY(1) |
S_008014_TA2_BUSY(1) | S_008014_TA3_BUSY(1) |
S_008014_DB0_BUSY(1) | S_008014_DB1_BUSY(1) |
S_008014_DB2_BUSY(1) | S_008014_DB3_BUSY(1) |
S_008014_CB0_BUSY(1) | S_008014_CB1_BUSY(1) |
S_008014_CB2_BUSY(1) | S_008014_CB3_BUSY(1);
drm/radeon/kms: Fix R600/RV770 startup path & reset We were calling reset unconditionaly in the startup path this is bad we need to call GPU reset for a good reason as after reset the GPU is in unknown states. To avoid any more bad things to happen we now also unconditionaly reinitialize the GPU after reset. This patch fix few issues reported by different people regarding KMS & R6XX/RV7XX hw. Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-10-02 00:02:13 +08:00
u32 tmp;
drm/radeon: introduce kernel modesetting for radeon hardware Add kernel modesetting support to radeon driver, use the ttm memory manager to manage memory and DRM/GEM to provide userspace API. In order to avoid backward compatibility issue and to allow clean design and code the radeon kernel modesetting use different code path than old radeon/drm driver. When kernel modesetting is enabled the IOCTL of radeon/drm driver are considered as invalid and an error message is printed in the log and they return failure. KMS enabled userspace will use new API to talk with the radeon/drm driver. The new API provide functions to create/destroy/share/mmap buffer object which are then managed by the kernel memory manager (here TTM). In order to submit command to the GPU the userspace provide a buffer holding the command stream, along this buffer userspace have to provide a list of buffer object used by the command stream. The kernel radeon driver will then place buffer in GPU accessible memory and will update command stream to reflect the position of the different buffers. The kernel will also perform security check on command stream provided by the user, we want to catch and forbid any illegal use of the GPU such as DMA into random system memory or into memory not owned by the process supplying the command stream. This part of the code is still incomplete and this why we propose that patch as a staging driver addition, future security might forbid current experimental userspace to run. This code support the following hardware : R1XX,R2XX,R3XX,R4XX,R5XX (radeon up to X1950). Works is underway to provide support for R6XX, R7XX and newer hardware (radeon from HD2XXX to HD4XXX). Authors: Jerome Glisse <jglisse@redhat.com> Dave Airlie <airlied@redhat.com> Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com> Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-06-05 20:42:42 +08:00
drm/radeon/kms: match r6xx/r7xx/evergreen asic_reset with previous asics Don't reset if the engine isn't busy. This matches the behavior of previous asics. Reseting a non-hung block can lead to a hang. Fixes: https://bugs.freedesktop.org/show_bug.cgi?id=33272 Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@gmail.com>
2011-01-21 23:38:22 +08:00
if (!(RREG32(GRBM_STATUS) & GUI_ACTIVE))
drm/radeon: switch to a finer grained reset for r6xx/7xx No change in functionality as we currently set all the reset flags. Reviewed-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2013-01-04 01:20:35 +08:00
return;
drm/radeon/kms: match r6xx/r7xx/evergreen asic_reset with previous asics Don't reset if the engine isn't busy. This matches the behavior of previous asics. Reseting a non-hung block can lead to a hang. Fixes: https://bugs.freedesktop.org/show_bug.cgi?id=33272 Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@gmail.com>
2011-01-21 23:38:22 +08:00
drm/radeon: reset dma engine on gpu reset (v2) This try to reset the dma engine when performing gpu reset. Hopefully bringing back the gpu dma engine in sane state. v2: agd5f: fix dma reset on cayman/TN, add support for SI Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2013-01-03 06:30:35 +08:00
dev_info(rdev->dev, " R_008010_GRBM_STATUS = 0x%08X\n",
drm/radeon/kms: Fix AGP support for R600/RV770 family (v2) For AGP to work unmapped access must cover VRAM & AGP as AGP is treated like VRAM by the GPU (ie physical address). This patch properly setup the virtual memory system aperture to cover AGP if AGP is enabled. It seems that there is memory corruption after resume when using AGP (RV770 seems unaffected thought). Version 2 just fix merge issue with updated AGP fallback patch. Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-10-07 01:04:30 +08:00
RREG32(R_008010_GRBM_STATUS));
drm/radeon: reset dma engine on gpu reset (v2) This try to reset the dma engine when performing gpu reset. Hopefully bringing back the gpu dma engine in sane state. v2: agd5f: fix dma reset on cayman/TN, add support for SI Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2013-01-03 06:30:35 +08:00
dev_info(rdev->dev, " R_008014_GRBM_STATUS2 = 0x%08X\n",
drm/radeon/kms: Fix R600/RV770 startup path & reset We were calling reset unconditionaly in the startup path this is bad we need to call GPU reset for a good reason as after reset the GPU is in unknown states. To avoid any more bad things to happen we now also unconditionaly reinitialize the GPU after reset. This patch fix few issues reported by different people regarding KMS & R6XX/RV7XX hw. Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-10-02 00:02:13 +08:00
RREG32(R_008014_GRBM_STATUS2));
drm/radeon: reset dma engine on gpu reset (v2) This try to reset the dma engine when performing gpu reset. Hopefully bringing back the gpu dma engine in sane state. v2: agd5f: fix dma reset on cayman/TN, add support for SI Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2013-01-03 06:30:35 +08:00
dev_info(rdev->dev, " R_000E50_SRBM_STATUS = 0x%08X\n",
drm/radeon/kms: Fix AGP support for R600/RV770 family (v2) For AGP to work unmapped access must cover VRAM & AGP as AGP is treated like VRAM by the GPU (ie physical address). This patch properly setup the virtual memory system aperture to cover AGP if AGP is enabled. It seems that there is memory corruption after resume when using AGP (RV770 seems unaffected thought). Version 2 just fix merge issue with updated AGP fallback patch. Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-10-07 01:04:30 +08:00
RREG32(R_000E50_SRBM_STATUS));
drm/radeon: improve GPU lockup debugging info on r6xx/r7xx/r8xx/r9xx Print various CP register that have valuable informations regarding GPU lockup. Signed-off-by: Jerome Glisse <jglisse@redhat.com> Reviewed-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2012-06-28 00:25:01 +08:00
dev_info(rdev->dev, " R_008674_CP_STALLED_STAT1 = 0x%08X\n",
RREG32(CP_STALLED_STAT1));
dev_info(rdev->dev, " R_008678_CP_STALLED_STAT2 = 0x%08X\n",
RREG32(CP_STALLED_STAT2));
dev_info(rdev->dev, " R_00867C_CP_BUSY_STAT = 0x%08X\n",
RREG32(CP_BUSY_STAT));
dev_info(rdev->dev, " R_008680_CP_STAT = 0x%08X\n",
RREG32(CP_STAT));
drm/radeon: reset dma engine on gpu reset (v2) This try to reset the dma engine when performing gpu reset. Hopefully bringing back the gpu dma engine in sane state. v2: agd5f: fix dma reset on cayman/TN, add support for SI Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2013-01-03 06:30:35 +08:00
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
/* Disable CP parsing/prefetching */
drm/radeon/kms: simplify & improve GPU reset V2 This simplify and improve GPU reset for R1XX-R6XX hw, it's not 100% reliable here are result: - R1XX/R2XX works bunch of time in a row, sometimes it seems it can work indifinitly - R3XX/R3XX the most unreliable one, sometimes you will be able to reset few times, sometimes not even once - R5XX more reliable than previous hw, seems to work most of the times but once in a while it fails for no obvious reasons (same status than previous reset just no same happy ending) - R6XX/R7XX are lot more reliable with this patch, still it seems that it can fail after a bunch (reset every 2sec for 3hour bring down the GPU & computer) This have been tested on various hw, for some odd reasons i wasn't able to lockup RS480/RS690 (while they use to love locking up). Note that on R1XX-R5XX the cursor will disapear after lockup haven't checked why, switch to console and back to X will restore cursor. Next step is to record the bogus command that leaded to the lockup. V2 Fix r6xx resume path to avoid reinitializing blit module, use the gpu_lockup boolean to avoid entering inifinite waiting loop on fence while reiniting the GPU Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-03-09 22:45:12 +08:00
WREG32(R_0086D8_CP_ME_CNTL, S_0086D8_CP_ME_HALT(1));
drm/radeon: reset dma engine on gpu reset (v2) This try to reset the dma engine when performing gpu reset. Hopefully bringing back the gpu dma engine in sane state. v2: agd5f: fix dma reset on cayman/TN, add support for SI Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2013-01-03 06:30:35 +08:00
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
/* Check if any of the rendering block is busy and reset it */
if ((RREG32(R_008010_GRBM_STATUS) & grbm_busy_mask) ||
(RREG32(R_008014_GRBM_STATUS2) & grbm2_busy_mask)) {
drm/radeon/kms: Fix R600/RV770 startup path & reset We were calling reset unconditionaly in the startup path this is bad we need to call GPU reset for a good reason as after reset the GPU is in unknown states. To avoid any more bad things to happen we now also unconditionaly reinitialize the GPU after reset. This patch fix few issues reported by different people regarding KMS & R6XX/RV7XX hw. Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-10-02 00:02:13 +08:00
tmp = S_008020_SOFT_RESET_CR(1) |
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
S_008020_SOFT_RESET_DB(1) |
S_008020_SOFT_RESET_CB(1) |
S_008020_SOFT_RESET_PA(1) |
S_008020_SOFT_RESET_SC(1) |
S_008020_SOFT_RESET_SMX(1) |
S_008020_SOFT_RESET_SPI(1) |
S_008020_SOFT_RESET_SX(1) |
S_008020_SOFT_RESET_SH(1) |
S_008020_SOFT_RESET_TC(1) |
S_008020_SOFT_RESET_TA(1) |
S_008020_SOFT_RESET_VC(1) |
drm/radeon/kms: Fix R600/RV770 startup path & reset We were calling reset unconditionaly in the startup path this is bad we need to call GPU reset for a good reason as after reset the GPU is in unknown states. To avoid any more bad things to happen we now also unconditionaly reinitialize the GPU after reset. This patch fix few issues reported by different people regarding KMS & R6XX/RV7XX hw. Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-10-02 00:02:13 +08:00
S_008020_SOFT_RESET_VGT(1);
drm/radeon/kms: Fix AGP support for R600/RV770 family (v2) For AGP to work unmapped access must cover VRAM & AGP as AGP is treated like VRAM by the GPU (ie physical address). This patch properly setup the virtual memory system aperture to cover AGP if AGP is enabled. It seems that there is memory corruption after resume when using AGP (RV770 seems unaffected thought). Version 2 just fix merge issue with updated AGP fallback patch. Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-10-07 01:04:30 +08:00
dev_info(rdev->dev, " R_008020_GRBM_SOFT_RESET=0x%08X\n", tmp);
drm/radeon/kms: Fix R600/RV770 startup path & reset We were calling reset unconditionaly in the startup path this is bad we need to call GPU reset for a good reason as after reset the GPU is in unknown states. To avoid any more bad things to happen we now also unconditionaly reinitialize the GPU after reset. This patch fix few issues reported by different people regarding KMS & R6XX/RV7XX hw. Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-10-02 00:02:13 +08:00
WREG32(R_008020_GRBM_SOFT_RESET, tmp);
drm/radeon/kms: simplify & improve GPU reset V2 This simplify and improve GPU reset for R1XX-R6XX hw, it's not 100% reliable here are result: - R1XX/R2XX works bunch of time in a row, sometimes it seems it can work indifinitly - R3XX/R3XX the most unreliable one, sometimes you will be able to reset few times, sometimes not even once - R5XX more reliable than previous hw, seems to work most of the times but once in a while it fails for no obvious reasons (same status than previous reset just no same happy ending) - R6XX/R7XX are lot more reliable with this patch, still it seems that it can fail after a bunch (reset every 2sec for 3hour bring down the GPU & computer) This have been tested on various hw, for some odd reasons i wasn't able to lockup RS480/RS690 (while they use to love locking up). Note that on R1XX-R5XX the cursor will disapear after lockup haven't checked why, switch to console and back to X will restore cursor. Next step is to record the bogus command that leaded to the lockup. V2 Fix r6xx resume path to avoid reinitializing blit module, use the gpu_lockup boolean to avoid entering inifinite waiting loop on fence while reiniting the GPU Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-03-09 22:45:12 +08:00
RREG32(R_008020_GRBM_SOFT_RESET);
mdelay(15);
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
WREG32(R_008020_GRBM_SOFT_RESET, 0);
}
/* Reset CP (we always reset CP) */
drm/radeon/kms: Fix R600/RV770 startup path & reset We were calling reset unconditionaly in the startup path this is bad we need to call GPU reset for a good reason as after reset the GPU is in unknown states. To avoid any more bad things to happen we now also unconditionaly reinitialize the GPU after reset. This patch fix few issues reported by different people regarding KMS & R6XX/RV7XX hw. Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-10-02 00:02:13 +08:00
tmp = S_008020_SOFT_RESET_CP(1);
dev_info(rdev->dev, "R_008020_GRBM_SOFT_RESET=0x%08X\n", tmp);
WREG32(R_008020_GRBM_SOFT_RESET, tmp);
drm/radeon/kms: simplify & improve GPU reset V2 This simplify and improve GPU reset for R1XX-R6XX hw, it's not 100% reliable here are result: - R1XX/R2XX works bunch of time in a row, sometimes it seems it can work indifinitly - R3XX/R3XX the most unreliable one, sometimes you will be able to reset few times, sometimes not even once - R5XX more reliable than previous hw, seems to work most of the times but once in a while it fails for no obvious reasons (same status than previous reset just no same happy ending) - R6XX/R7XX are lot more reliable with this patch, still it seems that it can fail after a bunch (reset every 2sec for 3hour bring down the GPU & computer) This have been tested on various hw, for some odd reasons i wasn't able to lockup RS480/RS690 (while they use to love locking up). Note that on R1XX-R5XX the cursor will disapear after lockup haven't checked why, switch to console and back to X will restore cursor. Next step is to record the bogus command that leaded to the lockup. V2 Fix r6xx resume path to avoid reinitializing blit module, use the gpu_lockup boolean to avoid entering inifinite waiting loop on fence while reiniting the GPU Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-03-09 22:45:12 +08:00
RREG32(R_008020_GRBM_SOFT_RESET);
mdelay(15);
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
WREG32(R_008020_GRBM_SOFT_RESET, 0);
drm/radeon: switch to a finer grained reset for r6xx/7xx No change in functionality as we currently set all the reset flags. Reviewed-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2013-01-04 01:20:35 +08:00
drm/radeon: reset dma engine on gpu reset (v2) This try to reset the dma engine when performing gpu reset. Hopefully bringing back the gpu dma engine in sane state. v2: agd5f: fix dma reset on cayman/TN, add support for SI Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2013-01-03 06:30:35 +08:00
dev_info(rdev->dev, " R_008010_GRBM_STATUS = 0x%08X\n",
drm/radeon/kms: Fix AGP support for R600/RV770 family (v2) For AGP to work unmapped access must cover VRAM & AGP as AGP is treated like VRAM by the GPU (ie physical address). This patch properly setup the virtual memory system aperture to cover AGP if AGP is enabled. It seems that there is memory corruption after resume when using AGP (RV770 seems unaffected thought). Version 2 just fix merge issue with updated AGP fallback patch. Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-10-07 01:04:30 +08:00
RREG32(R_008010_GRBM_STATUS));
drm/radeon: reset dma engine on gpu reset (v2) This try to reset the dma engine when performing gpu reset. Hopefully bringing back the gpu dma engine in sane state. v2: agd5f: fix dma reset on cayman/TN, add support for SI Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2013-01-03 06:30:35 +08:00
dev_info(rdev->dev, " R_008014_GRBM_STATUS2 = 0x%08X\n",
drm/radeon/kms: Fix AGP support for R600/RV770 family (v2) For AGP to work unmapped access must cover VRAM & AGP as AGP is treated like VRAM by the GPU (ie physical address). This patch properly setup the virtual memory system aperture to cover AGP if AGP is enabled. It seems that there is memory corruption after resume when using AGP (RV770 seems unaffected thought). Version 2 just fix merge issue with updated AGP fallback patch. Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-10-07 01:04:30 +08:00
RREG32(R_008014_GRBM_STATUS2));
drm/radeon: reset dma engine on gpu reset (v2) This try to reset the dma engine when performing gpu reset. Hopefully bringing back the gpu dma engine in sane state. v2: agd5f: fix dma reset on cayman/TN, add support for SI Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2013-01-03 06:30:35 +08:00
dev_info(rdev->dev, " R_000E50_SRBM_STATUS = 0x%08X\n",
drm/radeon/kms: Fix AGP support for R600/RV770 family (v2) For AGP to work unmapped access must cover VRAM & AGP as AGP is treated like VRAM by the GPU (ie physical address). This patch properly setup the virtual memory system aperture to cover AGP if AGP is enabled. It seems that there is memory corruption after resume when using AGP (RV770 seems unaffected thought). Version 2 just fix merge issue with updated AGP fallback patch. Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-10-07 01:04:30 +08:00
RREG32(R_000E50_SRBM_STATUS));
drm/radeon: improve GPU lockup debugging info on r6xx/r7xx/r8xx/r9xx Print various CP register that have valuable informations regarding GPU lockup. Signed-off-by: Jerome Glisse <jglisse@redhat.com> Reviewed-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2012-06-28 00:25:01 +08:00
dev_info(rdev->dev, " R_008674_CP_STALLED_STAT1 = 0x%08X\n",
RREG32(CP_STALLED_STAT1));
dev_info(rdev->dev, " R_008678_CP_STALLED_STAT2 = 0x%08X\n",
RREG32(CP_STALLED_STAT2));
dev_info(rdev->dev, " R_00867C_CP_BUSY_STAT = 0x%08X\n",
RREG32(CP_BUSY_STAT));
dev_info(rdev->dev, " R_008680_CP_STAT = 0x%08X\n",
RREG32(CP_STAT));
drm/radeon: switch to a finer grained reset for r6xx/7xx No change in functionality as we currently set all the reset flags. Reviewed-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2013-01-04 01:20:35 +08:00
}
static void r600_gpu_soft_reset_dma(struct radeon_device *rdev)
{
u32 tmp;
if (RREG32(DMA_STATUS_REG) & DMA_IDLE)
return;
drm/radeon: print dma status reg on lockup (v2) To help debug dma related lockup. v2: agd5f: update SI as well Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2013-01-03 04:12:15 +08:00
dev_info(rdev->dev, " R_00D034_DMA_STATUS_REG = 0x%08X\n",
RREG32(DMA_STATUS_REG));
drm/radeon: switch to a finer grained reset for r6xx/7xx No change in functionality as we currently set all the reset flags. Reviewed-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2013-01-04 01:20:35 +08:00
/* Disable DMA */
tmp = RREG32(DMA_RB_CNTL);
tmp &= ~DMA_RB_ENABLE;
WREG32(DMA_RB_CNTL, tmp);
/* Reset dma */
if (rdev->family >= CHIP_RV770)
WREG32(SRBM_SOFT_RESET, RV770_SOFT_RESET_DMA);
else
WREG32(SRBM_SOFT_RESET, SOFT_RESET_DMA);
RREG32(SRBM_SOFT_RESET);
udelay(50);
WREG32(SRBM_SOFT_RESET, 0);
dev_info(rdev->dev, " R_00D034_DMA_STATUS_REG = 0x%08X\n",
RREG32(DMA_STATUS_REG));
}
static int r600_gpu_soft_reset(struct radeon_device *rdev, u32 reset_mask)
{
struct rv515_mc_save save;
if (reset_mask == 0)
return 0;
dev_info(rdev->dev, "GPU softreset: 0x%08X\n", reset_mask);
rv515_mc_stop(rdev, &save);
if (r600_mc_wait_for_idle(rdev)) {
dev_warn(rdev->dev, "Wait for MC idle timedout !\n");
}
if (reset_mask & (RADEON_RESET_GFX | RADEON_RESET_COMPUTE))
r600_gpu_soft_reset_gfx(rdev);
if (reset_mask & RADEON_RESET_DMA)
r600_gpu_soft_reset_dma(rdev);
/* Wait a little for things to settle down */
mdelay(1);
drm/radeon/kms: Fix R600/RV770 startup path & reset We were calling reset unconditionaly in the startup path this is bad we need to call GPU reset for a good reason as after reset the GPU is in unknown states. To avoid any more bad things to happen we now also unconditionaly reinitialize the GPU after reset. This patch fix few issues reported by different people regarding KMS & R6XX/RV7XX hw. Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-10-02 00:02:13 +08:00
rv515_mc_resume(rdev, &save);
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
return 0;
}
drm/radeon: rename struct radeon_cp to radeon_ring That naming seems to make more sense, since we not only want to run PM4 rings with it. Signed-off-by: Christian König <deathsimple@vodafone.de> Reviewed-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-10-23 18:56:27 +08:00
bool r600_gpu_is_lockup(struct radeon_device *rdev, struct radeon_ring *ring)
drm/radeon/kms: fence cleanup + more reliable GPU lockup detection V4 This patch cleanup the fence code, it drops the timeout field of fence as the time to complete each IB is unpredictable and shouldn't be bound. The fence cleanup lead to GPU lockup detection improvement, this patch introduce a callback, allowing to do asic specific test for lockup detection. In this patch the CP is use as a first indicator of GPU lockup. If CP doesn't make progress during 1second we assume we are facing a GPU lockup. To avoid overhead of testing GPU lockup frequently due to fence taking time to be signaled we query the lockup callback every 500msec. There is plenty code comment explaining the design & choise inside the code. This have been tested mostly on R3XX/R5XX hw, in normal running destkop (compiz firefox, quake3 running) the lockup callback wasn't call once (1 hour session). Also tested with forcing GPU lockup and lockup was reported after the 1s CP activity timeout. V2 switch to 500ms timeout so GPU lockup get call at least 2 times in less than 2sec. V3 store last jiffies in fence struct so on ERESTART, EBUSY we keep track of how long we already wait for a given fence V4 make sure we got up to date cp read pointer so we don't have false positive Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-03-09 22:45:10 +08:00
{
u32 srbm_status;
u32 grbm_status;
u32 grbm_status2;
srbm_status = RREG32(R_000E50_SRBM_STATUS);
grbm_status = RREG32(R_008010_GRBM_STATUS);
grbm_status2 = RREG32(R_008014_GRBM_STATUS2);
if (!G_008010_GUI_ACTIVE(grbm_status)) {
drm/radeon: move lockup detection code into radeon_ring.c It isn't chipset specific, so it makes no sense to have that inside r100.c. Signed-off-by: Christian König <deathsimple@vodafone.de> Reviewed-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2012-05-02 21:11:20 +08:00
radeon_ring_lockup_update(ring);
drm/radeon/kms: fence cleanup + more reliable GPU lockup detection V4 This patch cleanup the fence code, it drops the timeout field of fence as the time to complete each IB is unpredictable and shouldn't be bound. The fence cleanup lead to GPU lockup detection improvement, this patch introduce a callback, allowing to do asic specific test for lockup detection. In this patch the CP is use as a first indicator of GPU lockup. If CP doesn't make progress during 1second we assume we are facing a GPU lockup. To avoid overhead of testing GPU lockup frequently due to fence taking time to be signaled we query the lockup callback every 500msec. There is plenty code comment explaining the design & choise inside the code. This have been tested mostly on R3XX/R5XX hw, in normal running destkop (compiz firefox, quake3 running) the lockup callback wasn't call once (1 hour session). Also tested with forcing GPU lockup and lockup was reported after the 1s CP activity timeout. V2 switch to 500ms timeout so GPU lockup get call at least 2 times in less than 2sec. V3 store last jiffies in fence struct so on ERESTART, EBUSY we keep track of how long we already wait for a given fence V4 make sure we got up to date cp read pointer so we don't have false positive Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-03-09 22:45:10 +08:00
return false;
}
/* force CP activities */
drm/radeon: make forcing ring activity a common function Nothing chipset or ring specific with it, so also move it to radon_ring. Signed-off-by: Christian König <deathsimple@vodafone.de> Reviewed-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2012-05-02 21:11:23 +08:00
radeon_ring_force_activity(rdev, ring);
drm/radeon: move lockup detection code into radeon_ring.c It isn't chipset specific, so it makes no sense to have that inside r100.c. Signed-off-by: Christian König <deathsimple@vodafone.de> Reviewed-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2012-05-02 21:11:20 +08:00
return radeon_ring_test_lockup(rdev, ring);
drm/radeon/kms: fence cleanup + more reliable GPU lockup detection V4 This patch cleanup the fence code, it drops the timeout field of fence as the time to complete each IB is unpredictable and shouldn't be bound. The fence cleanup lead to GPU lockup detection improvement, this patch introduce a callback, allowing to do asic specific test for lockup detection. In this patch the CP is use as a first indicator of GPU lockup. If CP doesn't make progress during 1second we assume we are facing a GPU lockup. To avoid overhead of testing GPU lockup frequently due to fence taking time to be signaled we query the lockup callback every 500msec. There is plenty code comment explaining the design & choise inside the code. This have been tested mostly on R3XX/R5XX hw, in normal running destkop (compiz firefox, quake3 running) the lockup callback wasn't call once (1 hour session). Also tested with forcing GPU lockup and lockup was reported after the 1s CP activity timeout. V2 switch to 500ms timeout so GPU lockup get call at least 2 times in less than 2sec. V3 store last jiffies in fence struct so on ERESTART, EBUSY we keep track of how long we already wait for a given fence V4 make sure we got up to date cp read pointer so we don't have false positive Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-03-09 22:45:10 +08:00
}
drm/radeon/kms: Add initial support for async DMA on r6xx/r7xx Uses the new multi-ring infrastucture. 6xx/7xx has a single async DMA ring. Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2012-09-28 03:08:35 +08:00
/**
* r600_dma_is_lockup - Check if the DMA engine is locked up
*
* @rdev: radeon_device pointer
* @ring: radeon_ring structure holding ring information
*
* Check if the async DMA engine is locked up (r6xx-evergreen).
* Returns true if the engine appears to be locked up, false if not.
*/
bool r600_dma_is_lockup(struct radeon_device *rdev, struct radeon_ring *ring)
{
u32 dma_status_reg;
dma_status_reg = RREG32(DMA_STATUS_REG);
if (dma_status_reg & DMA_IDLE) {
radeon_ring_lockup_update(ring);
return false;
}
/* force ring activities */
radeon_ring_force_activity(rdev, ring);
return radeon_ring_test_lockup(rdev, ring);
}
drm/radeon/kms: rename gpu_reset to asic_reset Patch rename gpu_reset to asic_reset in prevision of having gpu_reset doing more stuff than just basic asic reset. Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-03-09 22:45:11 +08:00
int r600_asic_reset(struct radeon_device *rdev)
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
{
drm/radeon: switch to a finer grained reset for r6xx/7xx No change in functionality as we currently set all the reset flags. Reviewed-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2013-01-04 01:20:35 +08:00
return r600_gpu_soft_reset(rdev, (RADEON_RESET_GFX |
RADEON_RESET_COMPUTE |
RADEON_RESET_DMA));
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
}
drm/radeon: fixup tiling group size and backendmap on r6xx-r9xx (v4) Tiling group size is always 256bits on r6xx/r7xx/r8xx/9xx. Also fix and simplify render backend map. This now properly sets up the backend map on r6xx-9xx which should improve 3D performance. Vadim benchmarked also: Some benchmarks on juniper (5750), fullscreen 1920x1080, first result - kernel 3.4.0+ (fb21affa), second - with these patches: Lightsmark: 91 fps => 123 fps +35% Doom3: 74 fps => 101 fps +36% Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2012-06-01 07:00:25 +08:00
u32 r6xx_remap_render_backend(struct radeon_device *rdev,
u32 tiling_pipe_num,
u32 max_rb_num,
u32 total_max_rb_num,
u32 disabled_rb_mask)
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
{
drm/radeon: fixup tiling group size and backendmap on r6xx-r9xx (v4) Tiling group size is always 256bits on r6xx/r7xx/r8xx/9xx. Also fix and simplify render backend map. This now properly sets up the backend map on r6xx-9xx which should improve 3D performance. Vadim benchmarked also: Some benchmarks on juniper (5750), fullscreen 1920x1080, first result - kernel 3.4.0+ (fb21affa), second - with these patches: Lightsmark: 91 fps => 123 fps +35% Doom3: 74 fps => 101 fps +36% Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2012-06-01 07:00:25 +08:00
u32 rendering_pipe_num, rb_num_width, req_rb_num;
u32 pipe_rb_ratio, pipe_rb_remain;
u32 data = 0, mask = 1 << (max_rb_num - 1);
unsigned i, j;
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
drm/radeon: fixup tiling group size and backendmap on r6xx-r9xx (v4) Tiling group size is always 256bits on r6xx/r7xx/r8xx/9xx. Also fix and simplify render backend map. This now properly sets up the backend map on r6xx-9xx which should improve 3D performance. Vadim benchmarked also: Some benchmarks on juniper (5750), fullscreen 1920x1080, first result - kernel 3.4.0+ (fb21affa), second - with these patches: Lightsmark: 91 fps => 123 fps +35% Doom3: 74 fps => 101 fps +36% Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2012-06-01 07:00:25 +08:00
/* mask out the RBs that don't exist on that asic */
disabled_rb_mask |= (0xff << max_rb_num) & 0xff;
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
drm/radeon: fixup tiling group size and backendmap on r6xx-r9xx (v4) Tiling group size is always 256bits on r6xx/r7xx/r8xx/9xx. Also fix and simplify render backend map. This now properly sets up the backend map on r6xx-9xx which should improve 3D performance. Vadim benchmarked also: Some benchmarks on juniper (5750), fullscreen 1920x1080, first result - kernel 3.4.0+ (fb21affa), second - with these patches: Lightsmark: 91 fps => 123 fps +35% Doom3: 74 fps => 101 fps +36% Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2012-06-01 07:00:25 +08:00
rendering_pipe_num = 1 << tiling_pipe_num;
req_rb_num = total_max_rb_num - r600_count_pipe_bits(disabled_rb_mask);
BUG_ON(rendering_pipe_num < req_rb_num);
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
drm/radeon: fixup tiling group size and backendmap on r6xx-r9xx (v4) Tiling group size is always 256bits on r6xx/r7xx/r8xx/9xx. Also fix and simplify render backend map. This now properly sets up the backend map on r6xx-9xx which should improve 3D performance. Vadim benchmarked also: Some benchmarks on juniper (5750), fullscreen 1920x1080, first result - kernel 3.4.0+ (fb21affa), second - with these patches: Lightsmark: 91 fps => 123 fps +35% Doom3: 74 fps => 101 fps +36% Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2012-06-01 07:00:25 +08:00
pipe_rb_ratio = rendering_pipe_num / req_rb_num;
pipe_rb_remain = rendering_pipe_num - pipe_rb_ratio * req_rb_num;
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
drm/radeon: fixup tiling group size and backendmap on r6xx-r9xx (v4) Tiling group size is always 256bits on r6xx/r7xx/r8xx/9xx. Also fix and simplify render backend map. This now properly sets up the backend map on r6xx-9xx which should improve 3D performance. Vadim benchmarked also: Some benchmarks on juniper (5750), fullscreen 1920x1080, first result - kernel 3.4.0+ (fb21affa), second - with these patches: Lightsmark: 91 fps => 123 fps +35% Doom3: 74 fps => 101 fps +36% Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2012-06-01 07:00:25 +08:00
if (rdev->family <= CHIP_RV740) {
/* r6xx/r7xx */
rb_num_width = 2;
} else {
/* eg+ */
rb_num_width = 4;
}
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
drm/radeon: fixup tiling group size and backendmap on r6xx-r9xx (v4) Tiling group size is always 256bits on r6xx/r7xx/r8xx/9xx. Also fix and simplify render backend map. This now properly sets up the backend map on r6xx-9xx which should improve 3D performance. Vadim benchmarked also: Some benchmarks on juniper (5750), fullscreen 1920x1080, first result - kernel 3.4.0+ (fb21affa), second - with these patches: Lightsmark: 91 fps => 123 fps +35% Doom3: 74 fps => 101 fps +36% Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2012-06-01 07:00:25 +08:00
for (i = 0; i < max_rb_num; i++) {
if (!(mask & disabled_rb_mask)) {
for (j = 0; j < pipe_rb_ratio; j++) {
data <<= rb_num_width;
data |= max_rb_num - i - 1;
}
if (pipe_rb_remain) {
data <<= rb_num_width;
data |= max_rb_num - i - 1;
pipe_rb_remain--;
}
}
mask >>= 1;
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
}
drm/radeon: fixup tiling group size and backendmap on r6xx-r9xx (v4) Tiling group size is always 256bits on r6xx/r7xx/r8xx/9xx. Also fix and simplify render backend map. This now properly sets up the backend map on r6xx-9xx which should improve 3D performance. Vadim benchmarked also: Some benchmarks on juniper (5750), fullscreen 1920x1080, first result - kernel 3.4.0+ (fb21affa), second - with these patches: Lightsmark: 91 fps => 123 fps +35% Doom3: 74 fps => 101 fps +36% Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2012-06-01 07:00:25 +08:00
return data;
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
}
int r600_count_pipe_bits(uint32_t val)
{
drm/radeon: Use hweight32 Use hweight32 instead of counting for each bit Signed-off-by: Akinobu Mita <akinobu.mita@gmail.com> Cc: David Airlie <airlied@linux.ie> Cc: dri-devel@lists.freedesktop.org Signed-off-by: Dave Airlie <airlied@redhat.com>
2012-11-09 20:10:41 +08:00
return hweight32(val);
drm/radeon: introduce kernel modesetting for radeon hardware Add kernel modesetting support to radeon driver, use the ttm memory manager to manage memory and DRM/GEM to provide userspace API. In order to avoid backward compatibility issue and to allow clean design and code the radeon kernel modesetting use different code path than old radeon/drm driver. When kernel modesetting is enabled the IOCTL of radeon/drm driver are considered as invalid and an error message is printed in the log and they return failure. KMS enabled userspace will use new API to talk with the radeon/drm driver. The new API provide functions to create/destroy/share/mmap buffer object which are then managed by the kernel memory manager (here TTM). In order to submit command to the GPU the userspace provide a buffer holding the command stream, along this buffer userspace have to provide a list of buffer object used by the command stream. The kernel radeon driver will then place buffer in GPU accessible memory and will update command stream to reflect the position of the different buffers. The kernel will also perform security check on command stream provided by the user, we want to catch and forbid any illegal use of the GPU such as DMA into random system memory or into memory not owned by the process supplying the command stream. This part of the code is still incomplete and this why we propose that patch as a staging driver addition, future security might forbid current experimental userspace to run. This code support the following hardware : R1XX,R2XX,R3XX,R4XX,R5XX (radeon up to X1950). Works is underway to provide support for R6XX, R7XX and newer hardware (radeon from HD2XXX to HD4XXX). Authors: Jerome Glisse <jglisse@redhat.com> Dave Airlie <airlied@redhat.com> Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com> Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-06-05 20:42:42 +08:00
}
drm/radeon: Mark all possible functions / structs as static Let's allow GCC to optimize better. This exposed some five unused functions, but this patch doesn't remove them. Signed-off-by: Lauri Kasanen <cand@gmx.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2012-09-01 01:43:50 +08:00
static void r600_gpu_init(struct radeon_device *rdev)
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
{
u32 tiling_config;
u32 ramcfg;
drm/radeon: fixes for r6xx/r7xx gfx init - updated swizzle modes for backend map setup - fix programming of a few gfx regs - properly handle pipe/backend setup on LE cards Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-02-20 05:22:31 +08:00
u32 cc_rb_backend_disable;
u32 cc_gc_shader_pipe_config;
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
u32 tmp;
int i, j;
u32 sq_config;
u32 sq_gpr_resource_mgmt_1 = 0;
u32 sq_gpr_resource_mgmt_2 = 0;
u32 sq_thread_resource_mgmt = 0;
u32 sq_stack_resource_mgmt_1 = 0;
u32 sq_stack_resource_mgmt_2 = 0;
drm/radeon: fixup tiling group size and backendmap on r6xx-r9xx (v4) Tiling group size is always 256bits on r6xx/r7xx/r8xx/9xx. Also fix and simplify render backend map. This now properly sets up the backend map on r6xx-9xx which should improve 3D performance. Vadim benchmarked also: Some benchmarks on juniper (5750), fullscreen 1920x1080, first result - kernel 3.4.0+ (fb21affa), second - with these patches: Lightsmark: 91 fps => 123 fps +35% Doom3: 74 fps => 101 fps +36% Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2012-06-01 07:00:25 +08:00
u32 disabled_rb_mask;
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
drm/radeon: fixup tiling group size and backendmap on r6xx-r9xx (v4) Tiling group size is always 256bits on r6xx/r7xx/r8xx/9xx. Also fix and simplify render backend map. This now properly sets up the backend map on r6xx-9xx which should improve 3D performance. Vadim benchmarked also: Some benchmarks on juniper (5750), fullscreen 1920x1080, first result - kernel 3.4.0+ (fb21affa), second - with these patches: Lightsmark: 91 fps => 123 fps +35% Doom3: 74 fps => 101 fps +36% Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2012-06-01 07:00:25 +08:00
rdev->config.r600.tiling_group_size = 256;
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
switch (rdev->family) {
case CHIP_R600:
rdev->config.r600.max_pipes = 4;
rdev->config.r600.max_tile_pipes = 8;
rdev->config.r600.max_simds = 4;
rdev->config.r600.max_backends = 4;
rdev->config.r600.max_gprs = 256;
rdev->config.r600.max_threads = 192;
rdev->config.r600.max_stack_entries = 256;
rdev->config.r600.max_hw_contexts = 8;
rdev->config.r600.max_gs_threads = 16;
rdev->config.r600.sx_max_export_size = 128;
rdev->config.r600.sx_max_export_pos_size = 16;
rdev->config.r600.sx_max_export_smx_size = 128;
rdev->config.r600.sq_num_cf_insts = 2;
break;
case CHIP_RV630:
case CHIP_RV635:
rdev->config.r600.max_pipes = 2;
rdev->config.r600.max_tile_pipes = 2;
rdev->config.r600.max_simds = 3;
rdev->config.r600.max_backends = 1;
rdev->config.r600.max_gprs = 128;
rdev->config.r600.max_threads = 192;
rdev->config.r600.max_stack_entries = 128;
rdev->config.r600.max_hw_contexts = 8;
rdev->config.r600.max_gs_threads = 4;
rdev->config.r600.sx_max_export_size = 128;
rdev->config.r600.sx_max_export_pos_size = 16;
rdev->config.r600.sx_max_export_smx_size = 128;
rdev->config.r600.sq_num_cf_insts = 2;
break;
case CHIP_RV610:
case CHIP_RV620:
case CHIP_RS780:
case CHIP_RS880:
rdev->config.r600.max_pipes = 1;
rdev->config.r600.max_tile_pipes = 1;
rdev->config.r600.max_simds = 2;
rdev->config.r600.max_backends = 1;
rdev->config.r600.max_gprs = 128;
rdev->config.r600.max_threads = 192;
rdev->config.r600.max_stack_entries = 128;
rdev->config.r600.max_hw_contexts = 4;
rdev->config.r600.max_gs_threads = 4;
rdev->config.r600.sx_max_export_size = 128;
rdev->config.r600.sx_max_export_pos_size = 16;
rdev->config.r600.sx_max_export_smx_size = 128;
rdev->config.r600.sq_num_cf_insts = 1;
break;
case CHIP_RV670:
rdev->config.r600.max_pipes = 4;
rdev->config.r600.max_tile_pipes = 4;
rdev->config.r600.max_simds = 4;
rdev->config.r600.max_backends = 4;
rdev->config.r600.max_gprs = 192;
rdev->config.r600.max_threads = 192;
rdev->config.r600.max_stack_entries = 256;
rdev->config.r600.max_hw_contexts = 8;
rdev->config.r600.max_gs_threads = 16;
rdev->config.r600.sx_max_export_size = 128;
rdev->config.r600.sx_max_export_pos_size = 16;
rdev->config.r600.sx_max_export_smx_size = 128;
rdev->config.r600.sq_num_cf_insts = 2;
break;
default:
break;
}
/* Initialize HDP */
for (i = 0, j = 0; i < 32; i++, j += 0x18) {
WREG32((0x2c14 + j), 0x00000000);
WREG32((0x2c18 + j), 0x00000000);
WREG32((0x2c1c + j), 0x00000000);
WREG32((0x2c20 + j), 0x00000000);
WREG32((0x2c24 + j), 0x00000000);
}
WREG32(GRBM_CNTL, GRBM_READ_TIMEOUT(0xff));
/* Setup tiling */
tiling_config = 0;
ramcfg = RREG32(RAMCFG);
switch (rdev->config.r600.max_tile_pipes) {
case 1:
tiling_config |= PIPE_TILING(0);
break;
case 2:
tiling_config |= PIPE_TILING(1);
break;
case 4:
tiling_config |= PIPE_TILING(2);
break;
case 8:
tiling_config |= PIPE_TILING(3);
break;
default:
break;
}
drm/radeon: fixes for r6xx/r7xx gfx init - updated swizzle modes for backend map setup - fix programming of a few gfx regs - properly handle pipe/backend setup on LE cards Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-02-20 05:22:31 +08:00
rdev->config.r600.tiling_npipes = rdev->config.r600.max_tile_pipes;
drm/radeon/kms: r600/r700 command stream checker This patch add cs checker to r600/r700 hw. Command stream checking will rewrite some of the cs value in order to restrict GPU access to BO size. This doesn't break old userspace but just enforce safe value. It should break any things that was using the r600/r700 cs ioctl to do forbidden things (malicious software), though we are not aware of such things. Here is the list of thing we check : - enforcing resource size - enforcing color buffer slice tile max, will restrict cb access - enforcing db buffer slice tile max, will restrict db access We don't check for shader bigger than the BO in which they are supposed to be, such use would lead to GPU lockup and is harmless from security POV, as far as we can tell (note that even checking for this wouldn't prevent someone to write bogus shader that lead to lockup). This patch has received as much testing as humanly possible with old userspace to check that it didn't break such configuration. However not all the applications out there were tested, thus it might broke some odd, rare applications. [airlied: fix rules for cs checker for parallel builds] Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-02-11 06:30:05 +08:00
rdev->config.r600.tiling_nbanks = 4 << ((ramcfg & NOOFBANK_MASK) >> NOOFBANK_SHIFT);
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
tiling_config |= BANK_TILING((ramcfg & NOOFBANK_MASK) >> NOOFBANK_SHIFT);
drm/radeon/kms: properly compute group_size on 6xx/7xx Needed for tiled surfaces. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Cc: stable@kernel.org Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-10-19 11:54:56 +08:00
tiling_config |= GROUP_SIZE((ramcfg & BURSTLENGTH_MASK) >> BURSTLENGTH_SHIFT);
drm/radeon: fixup tiling group size and backendmap on r6xx-r9xx (v4) Tiling group size is always 256bits on r6xx/r7xx/r8xx/9xx. Also fix and simplify render backend map. This now properly sets up the backend map on r6xx-9xx which should improve 3D performance. Vadim benchmarked also: Some benchmarks on juniper (5750), fullscreen 1920x1080, first result - kernel 3.4.0+ (fb21affa), second - with these patches: Lightsmark: 91 fps => 123 fps +35% Doom3: 74 fps => 101 fps +36% Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2012-06-01 07:00:25 +08:00
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
tmp = (ramcfg & NOOFROWS_MASK) >> NOOFROWS_SHIFT;
if (tmp > 3) {
tiling_config |= ROW_TILING(3);
tiling_config |= SAMPLE_SPLIT(3);
} else {
tiling_config |= ROW_TILING(tmp);
tiling_config |= SAMPLE_SPLIT(tmp);
}
tiling_config |= BANK_SWAPS(1);
drm/radeon: fixes for r6xx/r7xx gfx init - updated swizzle modes for backend map setup - fix programming of a few gfx regs - properly handle pipe/backend setup on LE cards Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-02-20 05:22:31 +08:00
cc_rb_backend_disable = RREG32(CC_RB_BACKEND_DISABLE) & 0x00ff0000;
drm/radeon: fixup tiling group size and backendmap on r6xx-r9xx (v4) Tiling group size is always 256bits on r6xx/r7xx/r8xx/9xx. Also fix and simplify render backend map. This now properly sets up the backend map on r6xx-9xx which should improve 3D performance. Vadim benchmarked also: Some benchmarks on juniper (5750), fullscreen 1920x1080, first result - kernel 3.4.0+ (fb21affa), second - with these patches: Lightsmark: 91 fps => 123 fps +35% Doom3: 74 fps => 101 fps +36% Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2012-06-01 07:00:25 +08:00
tmp = R6XX_MAX_BACKENDS -
r600_count_pipe_bits((cc_rb_backend_disable >> 16) & R6XX_MAX_BACKENDS_MASK);
if (tmp < rdev->config.r600.max_backends) {
rdev->config.r600.max_backends = tmp;
}
cc_gc_shader_pipe_config = RREG32(CC_GC_SHADER_PIPE_CONFIG) & 0x00ffff00;
tmp = R6XX_MAX_PIPES -
r600_count_pipe_bits((cc_gc_shader_pipe_config >> 8) & R6XX_MAX_PIPES_MASK);
if (tmp < rdev->config.r600.max_pipes) {
rdev->config.r600.max_pipes = tmp;
}
tmp = R6XX_MAX_SIMDS -
r600_count_pipe_bits((cc_gc_shader_pipe_config >> 16) & R6XX_MAX_SIMDS_MASK);
if (tmp < rdev->config.r600.max_simds) {
rdev->config.r600.max_simds = tmp;
}
disabled_rb_mask = (RREG32(CC_RB_BACKEND_DISABLE) >> 16) & R6XX_MAX_BACKENDS_MASK;
tmp = (tiling_config & PIPE_TILING__MASK) >> PIPE_TILING__SHIFT;
tmp = r6xx_remap_render_backend(rdev, tmp, rdev->config.r600.max_backends,
R6XX_MAX_BACKENDS, disabled_rb_mask);
tiling_config |= tmp << 16;
rdev->config.r600.backend_map = tmp;
drm/radeon/kms/r6xx+: add query for tile config (v2) Userspace needs this information to access tiled buffers via the CPU. v2: rebased on evergreen accel changes Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-06-05 01:10:12 +08:00
rdev->config.r600.tile_config = tiling_config;
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
WREG32(GB_TILING_CONFIG, tiling_config);
WREG32(DCP_TILING_CONFIG, tiling_config & 0xffff);
WREG32(HDP_TILING_CONFIG, tiling_config & 0xffff);
drm/radeon/kms: Add initial support for async DMA on r6xx/r7xx Uses the new multi-ring infrastucture. 6xx/7xx has a single async DMA ring. Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2012-09-28 03:08:35 +08:00
WREG32(DMA_TILING_CONFIG, tiling_config & 0xffff);
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
drm/radeon: fixes for r6xx/r7xx gfx init - updated swizzle modes for backend map setup - fix programming of a few gfx regs - properly handle pipe/backend setup on LE cards Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-02-20 05:22:31 +08:00
tmp = R6XX_MAX_PIPES - r600_count_pipe_bits((cc_gc_shader_pipe_config & INACTIVE_QD_PIPES_MASK) >> 8);
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
WREG32(VGT_OUT_DEALLOC_CNTL, (tmp * 4) & DEALLOC_DIST_MASK);
WREG32(VGT_VERTEX_REUSE_BLOCK_CNTL, ((tmp * 4) - 2) & VTX_REUSE_DEPTH_MASK);
/* Setup some CP states */
WREG32(CP_QUEUE_THRESHOLDS, (ROQ_IB1_START(0x16) | ROQ_IB2_START(0x2b)));
WREG32(CP_MEQ_THRESHOLDS, (MEQ_END(0x40) | ROQ_END(0x40)));
WREG32(TA_CNTL_AUX, (DISABLE_CUBE_ANISO | SYNC_GRADIENT |
SYNC_WALKER | SYNC_ALIGNER));
/* Setup various GPU states */
if (rdev->family == CHIP_RV670)
WREG32(ARB_GDEC_RD_CNTL, 0x00000021);
tmp = RREG32(SX_DEBUG_1);
tmp |= SMX_EVENT_RELEASE;
if ((rdev->family > CHIP_R600))
tmp |= ENABLE_NEW_SMX_ADDRESS;
WREG32(SX_DEBUG_1, tmp);
if (((rdev->family) == CHIP_R600) ||
((rdev->family) == CHIP_RV630) ||
((rdev->family) == CHIP_RV610) ||
((rdev->family) == CHIP_RV620) ||
drm/radeon/kms/r600: fix rs880 support v2 Lots of cases were wrong or missing. v2: rebased against drm-next Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-11-06 02:11:46 +08:00
((rdev->family) == CHIP_RS780) ||
((rdev->family) == CHIP_RS880)) {
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
WREG32(DB_DEBUG, PREZ_MUST_WAIT_FOR_POSTZ_DONE);
} else {
WREG32(DB_DEBUG, 0);
}
WREG32(DB_WATERMARKS, (DEPTH_FREE(4) | DEPTH_CACHELINE_FREE(16) |
DEPTH_FLUSH(16) | DEPTH_PENDING_FREE(4)));
WREG32(PA_SC_MULTI_CHIP_CNTL, 0);
WREG32(VGT_NUM_INSTANCES, 0);
WREG32(SPI_CONFIG_CNTL, GPR_WRITE_PRIORITY(0));
WREG32(SPI_CONFIG_CNTL_1, VTX_DONE_DELAY(0));
tmp = RREG32(SQ_MS_FIFO_SIZES);
if (((rdev->family) == CHIP_RV610) ||
((rdev->family) == CHIP_RV620) ||
drm/radeon/kms/r600: fix rs880 support v2 Lots of cases were wrong or missing. v2: rebased against drm-next Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-11-06 02:11:46 +08:00
((rdev->family) == CHIP_RS780) ||
((rdev->family) == CHIP_RS880)) {
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
tmp = (CACHE_FIFO_SIZE(0xa) |
FETCH_FIFO_HIWATER(0xa) |
DONE_FIFO_HIWATER(0xe0) |
ALU_UPDATE_FIFO_HIWATER(0x8));
} else if (((rdev->family) == CHIP_R600) ||
((rdev->family) == CHIP_RV630)) {
tmp &= ~DONE_FIFO_HIWATER(0xff);
tmp |= DONE_FIFO_HIWATER(0x4);
}
WREG32(SQ_MS_FIFO_SIZES, tmp);
/* SQ_CONFIG, SQ_GPR_RESOURCE_MGMT, SQ_THREAD_RESOURCE_MGMT, SQ_STACK_RESOURCE_MGMT
* should be adjusted as needed by the 2D/3D drivers. This just sets default values
*/
sq_config = RREG32(SQ_CONFIG);
sq_config &= ~(PS_PRIO(3) |
VS_PRIO(3) |
GS_PRIO(3) |
ES_PRIO(3));
sq_config |= (DX9_CONSTS |
VC_ENABLE |
PS_PRIO(0) |
VS_PRIO(1) |
GS_PRIO(2) |
ES_PRIO(3));
if ((rdev->family) == CHIP_R600) {
sq_gpr_resource_mgmt_1 = (NUM_PS_GPRS(124) |
NUM_VS_GPRS(124) |
NUM_CLAUSE_TEMP_GPRS(4));
sq_gpr_resource_mgmt_2 = (NUM_GS_GPRS(0) |
NUM_ES_GPRS(0));
sq_thread_resource_mgmt = (NUM_PS_THREADS(136) |
NUM_VS_THREADS(48) |
NUM_GS_THREADS(4) |
NUM_ES_THREADS(4));
sq_stack_resource_mgmt_1 = (NUM_PS_STACK_ENTRIES(128) |
NUM_VS_STACK_ENTRIES(128));
sq_stack_resource_mgmt_2 = (NUM_GS_STACK_ENTRIES(0) |
NUM_ES_STACK_ENTRIES(0));
} else if (((rdev->family) == CHIP_RV610) ||
((rdev->family) == CHIP_RV620) ||
drm/radeon/kms/r600: fix rs880 support v2 Lots of cases were wrong or missing. v2: rebased against drm-next Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-11-06 02:11:46 +08:00
((rdev->family) == CHIP_RS780) ||
((rdev->family) == CHIP_RS880)) {
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
/* no vertex cache */
sq_config &= ~VC_ENABLE;
sq_gpr_resource_mgmt_1 = (NUM_PS_GPRS(44) |
NUM_VS_GPRS(44) |
NUM_CLAUSE_TEMP_GPRS(2));
sq_gpr_resource_mgmt_2 = (NUM_GS_GPRS(17) |
NUM_ES_GPRS(17));
sq_thread_resource_mgmt = (NUM_PS_THREADS(79) |
NUM_VS_THREADS(78) |
NUM_GS_THREADS(4) |
NUM_ES_THREADS(31));
sq_stack_resource_mgmt_1 = (NUM_PS_STACK_ENTRIES(40) |
NUM_VS_STACK_ENTRIES(40));
sq_stack_resource_mgmt_2 = (NUM_GS_STACK_ENTRIES(32) |
NUM_ES_STACK_ENTRIES(16));
} else if (((rdev->family) == CHIP_RV630) ||
((rdev->family) == CHIP_RV635)) {
sq_gpr_resource_mgmt_1 = (NUM_PS_GPRS(44) |
NUM_VS_GPRS(44) |
NUM_CLAUSE_TEMP_GPRS(2));
sq_gpr_resource_mgmt_2 = (NUM_GS_GPRS(18) |
NUM_ES_GPRS(18));
sq_thread_resource_mgmt = (NUM_PS_THREADS(79) |
NUM_VS_THREADS(78) |
NUM_GS_THREADS(4) |
NUM_ES_THREADS(31));
sq_stack_resource_mgmt_1 = (NUM_PS_STACK_ENTRIES(40) |
NUM_VS_STACK_ENTRIES(40));
sq_stack_resource_mgmt_2 = (NUM_GS_STACK_ENTRIES(32) |
NUM_ES_STACK_ENTRIES(16));
} else if ((rdev->family) == CHIP_RV670) {
sq_gpr_resource_mgmt_1 = (NUM_PS_GPRS(44) |
NUM_VS_GPRS(44) |
NUM_CLAUSE_TEMP_GPRS(2));
sq_gpr_resource_mgmt_2 = (NUM_GS_GPRS(17) |
NUM_ES_GPRS(17));
sq_thread_resource_mgmt = (NUM_PS_THREADS(79) |
NUM_VS_THREADS(78) |
NUM_GS_THREADS(4) |
NUM_ES_THREADS(31));
sq_stack_resource_mgmt_1 = (NUM_PS_STACK_ENTRIES(64) |
NUM_VS_STACK_ENTRIES(64));
sq_stack_resource_mgmt_2 = (NUM_GS_STACK_ENTRIES(64) |
NUM_ES_STACK_ENTRIES(64));
}
WREG32(SQ_CONFIG, sq_config);
WREG32(SQ_GPR_RESOURCE_MGMT_1, sq_gpr_resource_mgmt_1);
WREG32(SQ_GPR_RESOURCE_MGMT_2, sq_gpr_resource_mgmt_2);
WREG32(SQ_THREAD_RESOURCE_MGMT, sq_thread_resource_mgmt);
WREG32(SQ_STACK_RESOURCE_MGMT_1, sq_stack_resource_mgmt_1);
WREG32(SQ_STACK_RESOURCE_MGMT_2, sq_stack_resource_mgmt_2);
if (((rdev->family) == CHIP_RV610) ||
((rdev->family) == CHIP_RV620) ||
drm/radeon/kms/r600: fix rs880 support v2 Lots of cases were wrong or missing. v2: rebased against drm-next Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-11-06 02:11:46 +08:00
((rdev->family) == CHIP_RS780) ||
((rdev->family) == CHIP_RS880)) {
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
WREG32(VGT_CACHE_INVALIDATION, CACHE_INVALIDATION(TC_ONLY));
} else {
WREG32(VGT_CACHE_INVALIDATION, CACHE_INVALIDATION(VC_AND_TC));
}
/* More default values. 2D/3D driver should adjust as needed */
WREG32(PA_SC_AA_SAMPLE_LOCS_2S, (S0_X(0xc) | S0_Y(0x4) |
S1_X(0x4) | S1_Y(0xc)));
WREG32(PA_SC_AA_SAMPLE_LOCS_4S, (S0_X(0xe) | S0_Y(0xe) |
S1_X(0x2) | S1_Y(0x2) |
S2_X(0xa) | S2_Y(0x6) |
S3_X(0x6) | S3_Y(0xa)));
WREG32(PA_SC_AA_SAMPLE_LOCS_8S_WD0, (S0_X(0xe) | S0_Y(0xb) |
S1_X(0x4) | S1_Y(0xc) |
S2_X(0x1) | S2_Y(0x6) |
S3_X(0xa) | S3_Y(0xe)));
WREG32(PA_SC_AA_SAMPLE_LOCS_8S_WD1, (S4_X(0x6) | S4_Y(0x1) |
S5_X(0x0) | S5_Y(0x0) |
S6_X(0xb) | S6_Y(0x4) |
S7_X(0x7) | S7_Y(0x8)));
WREG32(VGT_STRMOUT_EN, 0);
tmp = rdev->config.r600.max_pipes * 16;
switch (rdev->family) {
case CHIP_RV610:
case CHIP_RV620:
drm/radeon/kms/r600: fix rs880 support v2 Lots of cases were wrong or missing. v2: rebased against drm-next Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-11-06 02:11:46 +08:00
case CHIP_RS780:
case CHIP_RS880:
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
tmp += 32;
break;
case CHIP_RV670:
tmp += 128;
break;
default:
break;
}
if (tmp > 256) {
tmp = 256;
}
WREG32(VGT_ES_PER_GS, 128);
WREG32(VGT_GS_PER_ES, tmp);
WREG32(VGT_GS_PER_VS, 2);
WREG32(VGT_GS_VERTEX_REUSE, 16);
/* more default values. 2D/3D driver should adjust as needed */
WREG32(PA_SC_LINE_STIPPLE_STATE, 0);
WREG32(VGT_STRMOUT_EN, 0);
WREG32(SX_MISC, 0);
WREG32(PA_SC_MODE_CNTL, 0);
WREG32(PA_SC_AA_CONFIG, 0);
WREG32(PA_SC_LINE_STIPPLE, 0);
WREG32(SPI_INPUT_Z, 0);
WREG32(SPI_PS_IN_CONTROL_0, NUM_INTERP(2));
WREG32(CB_COLOR7_FRAG, 0);
/* Clear render buffer base addresses */
WREG32(CB_COLOR0_BASE, 0);
WREG32(CB_COLOR1_BASE, 0);
WREG32(CB_COLOR2_BASE, 0);
WREG32(CB_COLOR3_BASE, 0);
WREG32(CB_COLOR4_BASE, 0);
WREG32(CB_COLOR5_BASE, 0);
WREG32(CB_COLOR6_BASE, 0);
WREG32(CB_COLOR7_BASE, 0);
WREG32(CB_COLOR7_FRAG, 0);
switch (rdev->family) {
case CHIP_RV610:
case CHIP_RV620:
drm/radeon/kms/r600: fix rs880 support v2 Lots of cases were wrong or missing. v2: rebased against drm-next Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-11-06 02:11:46 +08:00
case CHIP_RS780:
case CHIP_RS880:
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
tmp = TC_L2_SIZE(8);
break;
case CHIP_RV630:
case CHIP_RV635:
tmp = TC_L2_SIZE(4);
break;
case CHIP_R600:
tmp = TC_L2_SIZE(0) | L2_DISABLE_LATE_HIT;
break;
default:
tmp = TC_L2_SIZE(0);
break;
}
WREG32(TC_CNTL, tmp);
tmp = RREG32(HDP_HOST_PATH_CNTL);
WREG32(HDP_HOST_PATH_CNTL, tmp);
tmp = RREG32(ARB_POP);
tmp |= ENABLE_TC128;
WREG32(ARB_POP, tmp);
WREG32(PA_SC_MULTI_CHIP_CNTL, 0);
WREG32(PA_CL_ENHANCE, (CLIP_VTX_REORDER_ENA |
NUM_CLIP_SEQ(3)));
WREG32(PA_SC_ENHANCE, FORCE_EOV_MAX_CLK_CNT(4095));
drm/radeon: add some additional 6xx/7xx/EG register init - SMX_SAR_CTL0 needs to be programmed correctly to prevent problems with memory exports in certain cases. - VC_ENHANCE needs to be initialized on 6xx/7xx. Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Cc: stable@vger.kernel.org Signed-off-by: Dave Airlie <airlied@redhat.com>
2012-06-15 04:06:36 +08:00
WREG32(VC_ENHANCE, 0);
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
}
drm/radeon: introduce kernel modesetting for radeon hardware Add kernel modesetting support to radeon driver, use the ttm memory manager to manage memory and DRM/GEM to provide userspace API. In order to avoid backward compatibility issue and to allow clean design and code the radeon kernel modesetting use different code path than old radeon/drm driver. When kernel modesetting is enabled the IOCTL of radeon/drm driver are considered as invalid and an error message is printed in the log and they return failure. KMS enabled userspace will use new API to talk with the radeon/drm driver. The new API provide functions to create/destroy/share/mmap buffer object which are then managed by the kernel memory manager (here TTM). In order to submit command to the GPU the userspace provide a buffer holding the command stream, along this buffer userspace have to provide a list of buffer object used by the command stream. The kernel radeon driver will then place buffer in GPU accessible memory and will update command stream to reflect the position of the different buffers. The kernel will also perform security check on command stream provided by the user, we want to catch and forbid any illegal use of the GPU such as DMA into random system memory or into memory not owned by the process supplying the command stream. This part of the code is still incomplete and this why we propose that patch as a staging driver addition, future security might forbid current experimental userspace to run. This code support the following hardware : R1XX,R2XX,R3XX,R4XX,R5XX (radeon up to X1950). Works is underway to provide support for R6XX, R7XX and newer hardware (radeon from HD2XXX to HD4XXX). Authors: Jerome Glisse <jglisse@redhat.com> Dave Airlie <airlied@redhat.com> Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com> Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-06-05 20:42:42 +08:00
/*
* Indirect registers accessor
*/
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
u32 r600_pciep_rreg(struct radeon_device *rdev, u32 reg)
{
u32 r;
WREG32(PCIE_PORT_INDEX, ((reg) & 0xff));
(void)RREG32(PCIE_PORT_INDEX);
r = RREG32(PCIE_PORT_DATA);
return r;
}
void r600_pciep_wreg(struct radeon_device *rdev, u32 reg, u32 v)
{
WREG32(PCIE_PORT_INDEX, ((reg) & 0xff));
(void)RREG32(PCIE_PORT_INDEX);
WREG32(PCIE_PORT_DATA, (v));
(void)RREG32(PCIE_PORT_DATA);
}
/*
* CP & Ring
*/
void r600_cp_stop(struct radeon_device *rdev)
{
drm/radeon: fix problem with changing active VRAM size. (v2) So we used to use lpfn directly to restrict VRAM when we couldn't access the unmappable area, however this was removed in 93225b0d7bc030f4a93165347a65893685822d70 as it also restricted the gtt placements. However it was only later noticed that this broke on some hw. This removes the active_vram_size, and just explicitly sets it when it changes, TTM/drm_mm will always use the real_vram_size, and the active vram size will change the TTM size used for lpfn setting. We should re-work the fpfn/lpfn to per-placement at some point I suspect, but that is too late for this kernel. Hopefully this addresses: https://bugs.freedesktop.org/show_bug.cgi?id=35254 v2: fix reported useful VRAM size to userspace to be correct. Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-03-14 07:47:24 +08:00
radeon_ttm_set_active_vram_size(rdev, rdev->mc.visible_vram_size);
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
WREG32(R_0086D8_CP_ME_CNTL, S_0086D8_CP_ME_HALT(1));
drm/radeon/kms: enable writeback (v2) When writeback is enabled, the GPU shadows writes to certain registers into a buffer in memory. The driver can then read the values from the shadow rather than reading back from the register across the bus. Writeback can be disabled by setting the no_wb module param to 1. On r6xx/r7xx/evergreen, the following registers are shadowed: - CP scratch registers - CP read pointer - IH write pointer On r1xx-rr5xx, the following registers are shadowed: - CP scratch registers - CP read pointer v2: - Combine wb patches for r6xx-evergreen and r1xx-r5xx - Writeback is disabled on AGP boards since it tends to be unreliable on AGP using the gart. - Check radeon_wb_init return values properly. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-08-28 06:25:25 +08:00
WREG32(SCRATCH_UMSK, 0);
drm/radeon/kms: Add initial support for async DMA on r6xx/r7xx Uses the new multi-ring infrastucture. 6xx/7xx has a single async DMA ring. Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2012-09-28 03:08:35 +08:00
rdev->ring[RADEON_RING_TYPE_GFX_INDEX].ready = false;
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
}
drm/radeon/kms: Add support for interrupts on r6xx/r7xx chips (v3) This enables the use of interrupts on r6xx/r7xx hardware. Interrupts are implemented via a ring buffer. The GPU adds interrupts vectors to the ring and the host reads them off in the interrupt handler. The interrupt controller requires firmware like the CP. This firmware must be installed and accessble to the firmware loader for interrupts to function. MSIs don't seem to work on my RS780. They work fine on all my discrete cards. I'm not sure about other RS780s or RS880s. I've disabled MSIs on RS780 and RS880, but it would probably be worth checking on some other systems. v2 - fix some checkpatch.pl problems; re-read the disp int status reg if we restart the ih; v3 - remove the irq handler if r600_irq_init() fails; remove spinlock in r600_ih_ring_fini(); move ih rb overflow check to r600_get_ih_wptr(); move irq ack to separate function; Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-02 02:43:46 +08:00
int r600_init_microcode(struct radeon_device *rdev)
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
{
struct platform_device *pdev;
const char *chip_name;
drm/radeon/kms: Add support for interrupts on r6xx/r7xx chips (v3) This enables the use of interrupts on r6xx/r7xx hardware. Interrupts are implemented via a ring buffer. The GPU adds interrupts vectors to the ring and the host reads them off in the interrupt handler. The interrupt controller requires firmware like the CP. This firmware must be installed and accessble to the firmware loader for interrupts to function. MSIs don't seem to work on my RS780. They work fine on all my discrete cards. I'm not sure about other RS780s or RS880s. I've disabled MSIs on RS780 and RS880, but it would probably be worth checking on some other systems. v2 - fix some checkpatch.pl problems; re-read the disp int status reg if we restart the ih; v3 - remove the irq handler if r600_irq_init() fails; remove spinlock in r600_ih_ring_fini(); move ih rb overflow check to r600_get_ih_wptr(); move irq ack to separate function; Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-02 02:43:46 +08:00
const char *rlc_chip_name;
size_t pfp_req_size, me_req_size, rlc_req_size;
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
char fw_name[30];
int err;
DRM_DEBUG("\n");
pdev = platform_device_register_simple("radeon_cp", 0, NULL, 0);
err = IS_ERR(pdev);
if (err) {
printk(KERN_ERR "radeon_cp: Failed to register firmware\n");
return -EINVAL;
}
switch (rdev->family) {
drm/radeon/kms: Add support for interrupts on r6xx/r7xx chips (v3) This enables the use of interrupts on r6xx/r7xx hardware. Interrupts are implemented via a ring buffer. The GPU adds interrupts vectors to the ring and the host reads them off in the interrupt handler. The interrupt controller requires firmware like the CP. This firmware must be installed and accessble to the firmware loader for interrupts to function. MSIs don't seem to work on my RS780. They work fine on all my discrete cards. I'm not sure about other RS780s or RS880s. I've disabled MSIs on RS780 and RS880, but it would probably be worth checking on some other systems. v2 - fix some checkpatch.pl problems; re-read the disp int status reg if we restart the ih; v3 - remove the irq handler if r600_irq_init() fails; remove spinlock in r600_ih_ring_fini(); move ih rb overflow check to r600_get_ih_wptr(); move irq ack to separate function; Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-02 02:43:46 +08:00
case CHIP_R600:
chip_name = "R600";
rlc_chip_name = "R600";
break;
case CHIP_RV610:
chip_name = "RV610";
rlc_chip_name = "R600";
break;
case CHIP_RV630:
chip_name = "RV630";
rlc_chip_name = "R600";
break;
case CHIP_RV620:
chip_name = "RV620";
rlc_chip_name = "R600";
break;
case CHIP_RV635:
chip_name = "RV635";
rlc_chip_name = "R600";
break;
case CHIP_RV670:
chip_name = "RV670";
rlc_chip_name = "R600";
break;
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
case CHIP_RS780:
drm/radeon/kms: Add support for interrupts on r6xx/r7xx chips (v3) This enables the use of interrupts on r6xx/r7xx hardware. Interrupts are implemented via a ring buffer. The GPU adds interrupts vectors to the ring and the host reads them off in the interrupt handler. The interrupt controller requires firmware like the CP. This firmware must be installed and accessble to the firmware loader for interrupts to function. MSIs don't seem to work on my RS780. They work fine on all my discrete cards. I'm not sure about other RS780s or RS880s. I've disabled MSIs on RS780 and RS880, but it would probably be worth checking on some other systems. v2 - fix some checkpatch.pl problems; re-read the disp int status reg if we restart the ih; v3 - remove the irq handler if r600_irq_init() fails; remove spinlock in r600_ih_ring_fini(); move ih rb overflow check to r600_get_ih_wptr(); move irq ack to separate function; Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-02 02:43:46 +08:00
case CHIP_RS880:
chip_name = "RS780";
rlc_chip_name = "R600";
break;
case CHIP_RV770:
chip_name = "RV770";
rlc_chip_name = "R700";
break;
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
case CHIP_RV730:
drm/radeon/kms: Add support for interrupts on r6xx/r7xx chips (v3) This enables the use of interrupts on r6xx/r7xx hardware. Interrupts are implemented via a ring buffer. The GPU adds interrupts vectors to the ring and the host reads them off in the interrupt handler. The interrupt controller requires firmware like the CP. This firmware must be installed and accessble to the firmware loader for interrupts to function. MSIs don't seem to work on my RS780. They work fine on all my discrete cards. I'm not sure about other RS780s or RS880s. I've disabled MSIs on RS780 and RS880, but it would probably be worth checking on some other systems. v2 - fix some checkpatch.pl problems; re-read the disp int status reg if we restart the ih; v3 - remove the irq handler if r600_irq_init() fails; remove spinlock in r600_ih_ring_fini(); move ih rb overflow check to r600_get_ih_wptr(); move irq ack to separate function; Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-02 02:43:46 +08:00
case CHIP_RV740:
chip_name = "RV730";
rlc_chip_name = "R700";
break;
case CHIP_RV710:
chip_name = "RV710";
rlc_chip_name = "R700";
break;
drm/radeon/kms/evergreen: setup and enable the CP The command processor (CP) fetches command buffers and feeds the GPU. This patch requires the evergreen family me and pfp ucode files. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-03-25 01:36:43 +08:00
case CHIP_CEDAR:
chip_name = "CEDAR";
drm/radeon/kms/evergreen: implement irq support Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-03-25 01:55:51 +08:00
rlc_chip_name = "CEDAR";
drm/radeon/kms/evergreen: setup and enable the CP The command processor (CP) fetches command buffers and feeds the GPU. This patch requires the evergreen family me and pfp ucode files. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-03-25 01:36:43 +08:00
break;
case CHIP_REDWOOD:
chip_name = "REDWOOD";
drm/radeon/kms/evergreen: implement irq support Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-03-25 01:55:51 +08:00
rlc_chip_name = "REDWOOD";
drm/radeon/kms/evergreen: setup and enable the CP The command processor (CP) fetches command buffers and feeds the GPU. This patch requires the evergreen family me and pfp ucode files. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-03-25 01:36:43 +08:00
break;
case CHIP_JUNIPER:
chip_name = "JUNIPER";
drm/radeon/kms/evergreen: implement irq support Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-03-25 01:55:51 +08:00
rlc_chip_name = "JUNIPER";
drm/radeon/kms/evergreen: setup and enable the CP The command processor (CP) fetches command buffers and feeds the GPU. This patch requires the evergreen family me and pfp ucode files. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-03-25 01:36:43 +08:00
break;
case CHIP_CYPRESS:
case CHIP_HEMLOCK:
chip_name = "CYPRESS";
drm/radeon/kms/evergreen: implement irq support Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-03-25 01:55:51 +08:00
rlc_chip_name = "CYPRESS";
drm/radeon/kms/evergreen: setup and enable the CP The command processor (CP) fetches command buffers and feeds the GPU. This patch requires the evergreen family me and pfp ucode files. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-03-25 01:36:43 +08:00
break;
drm/radeon/kms: add Ontario APU ucode loading support Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-11-23 06:56:31 +08:00
case CHIP_PALM:
chip_name = "PALM";
rlc_chip_name = "SUMO";
break;
drm/radeon/kms: add support for Llano Fusion APUs - add gpu init support - add blit support - add ucode loader Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-06-01 03:42:48 +08:00
case CHIP_SUMO:
chip_name = "SUMO";
rlc_chip_name = "SUMO";
break;
case CHIP_SUMO2:
chip_name = "SUMO2";
rlc_chip_name = "SUMO";
break;
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
default: BUG();
}
drm/radeon/kms/evergreen: setup and enable the CP The command processor (CP) fetches command buffers and feeds the GPU. This patch requires the evergreen family me and pfp ucode files. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-03-25 01:36:43 +08:00
if (rdev->family >= CHIP_CEDAR) {
pfp_req_size = EVERGREEN_PFP_UCODE_SIZE * 4;
me_req_size = EVERGREEN_PM4_UCODE_SIZE * 4;
drm/radeon/kms/evergreen: implement irq support Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-03-25 01:55:51 +08:00
rlc_req_size = EVERGREEN_RLC_UCODE_SIZE * 4;
drm/radeon/kms/evergreen: setup and enable the CP The command processor (CP) fetches command buffers and feeds the GPU. This patch requires the evergreen family me and pfp ucode files. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-03-25 01:36:43 +08:00
} else if (rdev->family >= CHIP_RV770) {
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
pfp_req_size = R700_PFP_UCODE_SIZE * 4;
me_req_size = R700_PM4_UCODE_SIZE * 4;
drm/radeon/kms: Add support for interrupts on r6xx/r7xx chips (v3) This enables the use of interrupts on r6xx/r7xx hardware. Interrupts are implemented via a ring buffer. The GPU adds interrupts vectors to the ring and the host reads them off in the interrupt handler. The interrupt controller requires firmware like the CP. This firmware must be installed and accessble to the firmware loader for interrupts to function. MSIs don't seem to work on my RS780. They work fine on all my discrete cards. I'm not sure about other RS780s or RS880s. I've disabled MSIs on RS780 and RS880, but it would probably be worth checking on some other systems. v2 - fix some checkpatch.pl problems; re-read the disp int status reg if we restart the ih; v3 - remove the irq handler if r600_irq_init() fails; remove spinlock in r600_ih_ring_fini(); move ih rb overflow check to r600_get_ih_wptr(); move irq ack to separate function; Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-02 02:43:46 +08:00
rlc_req_size = R700_RLC_UCODE_SIZE * 4;
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
} else {
pfp_req_size = PFP_UCODE_SIZE * 4;
me_req_size = PM4_UCODE_SIZE * 12;
drm/radeon/kms: Add support for interrupts on r6xx/r7xx chips (v3) This enables the use of interrupts on r6xx/r7xx hardware. Interrupts are implemented via a ring buffer. The GPU adds interrupts vectors to the ring and the host reads them off in the interrupt handler. The interrupt controller requires firmware like the CP. This firmware must be installed and accessble to the firmware loader for interrupts to function. MSIs don't seem to work on my RS780. They work fine on all my discrete cards. I'm not sure about other RS780s or RS880s. I've disabled MSIs on RS780 and RS880, but it would probably be worth checking on some other systems. v2 - fix some checkpatch.pl problems; re-read the disp int status reg if we restart the ih; v3 - remove the irq handler if r600_irq_init() fails; remove spinlock in r600_ih_ring_fini(); move ih rb overflow check to r600_get_ih_wptr(); move irq ack to separate function; Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-02 02:43:46 +08:00
rlc_req_size = RLC_UCODE_SIZE * 4;
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
}
drm/radeon/kms: Add support for interrupts on r6xx/r7xx chips (v3) This enables the use of interrupts on r6xx/r7xx hardware. Interrupts are implemented via a ring buffer. The GPU adds interrupts vectors to the ring and the host reads them off in the interrupt handler. The interrupt controller requires firmware like the CP. This firmware must be installed and accessble to the firmware loader for interrupts to function. MSIs don't seem to work on my RS780. They work fine on all my discrete cards. I'm not sure about other RS780s or RS880s. I've disabled MSIs on RS780 and RS880, but it would probably be worth checking on some other systems. v2 - fix some checkpatch.pl problems; re-read the disp int status reg if we restart the ih; v3 - remove the irq handler if r600_irq_init() fails; remove spinlock in r600_ih_ring_fini(); move ih rb overflow check to r600_get_ih_wptr(); move irq ack to separate function; Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-02 02:43:46 +08:00
DRM_INFO("Loading %s Microcode\n", chip_name);
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
snprintf(fw_name, sizeof(fw_name), "radeon/%s_pfp.bin", chip_name);
err = request_firmware(&rdev->pfp_fw, fw_name, &pdev->dev);
if (err)
goto out;
if (rdev->pfp_fw->size != pfp_req_size) {
printk(KERN_ERR
"r600_cp: Bogus length %zu in firmware \"%s\"\n",
rdev->pfp_fw->size, fw_name);
err = -EINVAL;
goto out;
}
snprintf(fw_name, sizeof(fw_name), "radeon/%s_me.bin", chip_name);
err = request_firmware(&rdev->me_fw, fw_name, &pdev->dev);
if (err)
goto out;
if (rdev->me_fw->size != me_req_size) {
printk(KERN_ERR
"r600_cp: Bogus length %zu in firmware \"%s\"\n",
rdev->me_fw->size, fw_name);
err = -EINVAL;
}
drm/radeon/kms: Add support for interrupts on r6xx/r7xx chips (v3) This enables the use of interrupts on r6xx/r7xx hardware. Interrupts are implemented via a ring buffer. The GPU adds interrupts vectors to the ring and the host reads them off in the interrupt handler. The interrupt controller requires firmware like the CP. This firmware must be installed and accessble to the firmware loader for interrupts to function. MSIs don't seem to work on my RS780. They work fine on all my discrete cards. I'm not sure about other RS780s or RS880s. I've disabled MSIs on RS780 and RS880, but it would probably be worth checking on some other systems. v2 - fix some checkpatch.pl problems; re-read the disp int status reg if we restart the ih; v3 - remove the irq handler if r600_irq_init() fails; remove spinlock in r600_ih_ring_fini(); move ih rb overflow check to r600_get_ih_wptr(); move irq ack to separate function; Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-02 02:43:46 +08:00
snprintf(fw_name, sizeof(fw_name), "radeon/%s_rlc.bin", rlc_chip_name);
err = request_firmware(&rdev->rlc_fw, fw_name, &pdev->dev);
if (err)
goto out;
if (rdev->rlc_fw->size != rlc_req_size) {
printk(KERN_ERR
"r600_rlc: Bogus length %zu in firmware \"%s\"\n",
rdev->rlc_fw->size, fw_name);
err = -EINVAL;
}
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
out:
platform_device_unregister(pdev);
if (err) {
if (err != -EINVAL)
printk(KERN_ERR
"r600_cp: Failed to load firmware \"%s\"\n",
fw_name);
release_firmware(rdev->pfp_fw);
rdev->pfp_fw = NULL;
release_firmware(rdev->me_fw);
rdev->me_fw = NULL;
drm/radeon/kms: Add support for interrupts on r6xx/r7xx chips (v3) This enables the use of interrupts on r6xx/r7xx hardware. Interrupts are implemented via a ring buffer. The GPU adds interrupts vectors to the ring and the host reads them off in the interrupt handler. The interrupt controller requires firmware like the CP. This firmware must be installed and accessble to the firmware loader for interrupts to function. MSIs don't seem to work on my RS780. They work fine on all my discrete cards. I'm not sure about other RS780s or RS880s. I've disabled MSIs on RS780 and RS880, but it would probably be worth checking on some other systems. v2 - fix some checkpatch.pl problems; re-read the disp int status reg if we restart the ih; v3 - remove the irq handler if r600_irq_init() fails; remove spinlock in r600_ih_ring_fini(); move ih rb overflow check to r600_get_ih_wptr(); move irq ack to separate function; Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-02 02:43:46 +08:00
release_firmware(rdev->rlc_fw);
rdev->rlc_fw = NULL;
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
}
return err;
}
static int r600_cp_load_microcode(struct radeon_device *rdev)
{
const __be32 *fw_data;
int i;
if (!rdev->me_fw || !rdev->pfp_fw)
return -EINVAL;
r600_cp_stop(rdev);
drm/radeon/kms: 6xx/7xx big endian fixes agd5f: minor cleanups Signed-off-by: Cédric Cano <ccano@interfaceconcept.com> Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-02-12 08:45:38 +08:00
WREG32(CP_RB_CNTL,
#ifdef __BIG_ENDIAN
BUF_SWAP_32BIT |
#endif
RB_NO_UPDATE | RB_BLKSZ(15) | RB_BUFSZ(3));
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
/* Reset cp */
WREG32(GRBM_SOFT_RESET, SOFT_RESET_CP);
RREG32(GRBM_SOFT_RESET);
mdelay(15);
WREG32(GRBM_SOFT_RESET, 0);
WREG32(CP_ME_RAM_WADDR, 0);
fw_data = (const __be32 *)rdev->me_fw->data;
WREG32(CP_ME_RAM_WADDR, 0);
for (i = 0; i < PM4_UCODE_SIZE * 3; i++)
WREG32(CP_ME_RAM_DATA,
be32_to_cpup(fw_data++));
fw_data = (const __be32 *)rdev->pfp_fw->data;
WREG32(CP_PFP_UCODE_ADDR, 0);
for (i = 0; i < PFP_UCODE_SIZE; i++)
WREG32(CP_PFP_UCODE_DATA,
be32_to_cpup(fw_data++));
WREG32(CP_PFP_UCODE_ADDR, 0);
WREG32(CP_ME_RAM_WADDR, 0);
WREG32(CP_ME_RAM_RADDR, 0);
return 0;
}
int r600_cp_start(struct radeon_device *rdev)
{
drm/radeon: rename struct radeon_cp to radeon_ring That naming seems to make more sense, since we not only want to run PM4 rings with it. Signed-off-by: Christian König <deathsimple@vodafone.de> Reviewed-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-10-23 18:56:27 +08:00
struct radeon_ring *ring = &rdev->ring[RADEON_RING_TYPE_GFX_INDEX];
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
int r;
uint32_t cp_me;
drm/radeon: rename struct radeon_cp to radeon_ring That naming seems to make more sense, since we not only want to run PM4 rings with it. Signed-off-by: Christian König <deathsimple@vodafone.de> Reviewed-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-10-23 18:56:27 +08:00
r = radeon_ring_lock(rdev, ring, 7);
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
if (r) {
DRM_ERROR("radeon: cp failed to lock ring (%d).\n", r);
return r;
}
drm/radeon: rename struct radeon_cp to radeon_ring That naming seems to make more sense, since we not only want to run PM4 rings with it. Signed-off-by: Christian König <deathsimple@vodafone.de> Reviewed-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-10-23 18:56:27 +08:00
radeon_ring_write(ring, PACKET3(PACKET3_ME_INITIALIZE, 5));
radeon_ring_write(ring, 0x1);
drm/radeon/kms/evergreen: fix gpu hangs in userspace accel code These VGT regs need to be programmed via the ring rather than MMIO as on previous asics (r6xx/r7xx). Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Cc: stable@kernel.org Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-09-03 09:32:32 +08:00
if (rdev->family >= CHIP_RV770) {
drm/radeon: rename struct radeon_cp to radeon_ring That naming seems to make more sense, since we not only want to run PM4 rings with it. Signed-off-by: Christian König <deathsimple@vodafone.de> Reviewed-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-10-23 18:56:27 +08:00
radeon_ring_write(ring, 0x0);
radeon_ring_write(ring, rdev->config.rv770.max_hw_contexts - 1);
drm/radeon/kms/evergreen: setup and enable the CP The command processor (CP) fetches command buffers and feeds the GPU. This patch requires the evergreen family me and pfp ucode files. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-03-25 01:36:43 +08:00
} else {
drm/radeon: rename struct radeon_cp to radeon_ring That naming seems to make more sense, since we not only want to run PM4 rings with it. Signed-off-by: Christian König <deathsimple@vodafone.de> Reviewed-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-10-23 18:56:27 +08:00
radeon_ring_write(ring, 0x3);
radeon_ring_write(ring, rdev->config.r600.max_hw_contexts - 1);
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
}
drm/radeon: rename struct radeon_cp to radeon_ring That naming seems to make more sense, since we not only want to run PM4 rings with it. Signed-off-by: Christian König <deathsimple@vodafone.de> Reviewed-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-10-23 18:56:27 +08:00
radeon_ring_write(ring, PACKET3_ME_INITIALIZE_DEVICE_ID(1));
radeon_ring_write(ring, 0);
radeon_ring_write(ring, 0);
radeon_ring_unlock_commit(rdev, ring);
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
cp_me = 0xff;
WREG32(R_0086D8_CP_ME_CNTL, cp_me);
return 0;
}
int r600_cp_resume(struct radeon_device *rdev)
{
drm/radeon: rename struct radeon_cp to radeon_ring That naming seems to make more sense, since we not only want to run PM4 rings with it. Signed-off-by: Christian König <deathsimple@vodafone.de> Reviewed-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-10-23 18:56:27 +08:00
struct radeon_ring *ring = &rdev->ring[RADEON_RING_TYPE_GFX_INDEX];
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
u32 tmp;
u32 rb_bufsz;
int r;
/* Reset cp */
WREG32(GRBM_SOFT_RESET, SOFT_RESET_CP);
RREG32(GRBM_SOFT_RESET);
mdelay(15);
WREG32(GRBM_SOFT_RESET, 0);
/* Set ring buffer size */
drm/radeon: rename struct radeon_cp to radeon_ring That naming seems to make more sense, since we not only want to run PM4 rings with it. Signed-off-by: Christian König <deathsimple@vodafone.de> Reviewed-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-10-23 18:56:27 +08:00
rb_bufsz = drm_order(ring->ring_size / 8);
drm/radeon/kms: enable writeback (v2) When writeback is enabled, the GPU shadows writes to certain registers into a buffer in memory. The driver can then read the values from the shadow rather than reading back from the register across the bus. Writeback can be disabled by setting the no_wb module param to 1. On r6xx/r7xx/evergreen, the following registers are shadowed: - CP scratch registers - CP read pointer - IH write pointer On r1xx-rr5xx, the following registers are shadowed: - CP scratch registers - CP read pointer v2: - Combine wb patches for r6xx-evergreen and r1xx-r5xx - Writeback is disabled on AGP boards since it tends to be unreliable on AGP using the gart. - Check radeon_wb_init return values properly. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-08-28 06:25:25 +08:00
tmp = (drm_order(RADEON_GPU_PAGE_SIZE/8) << 8) | rb_bufsz;
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
#ifdef __BIG_ENDIAN
drm/radeon/kms: Don't RMW CP_RB_CNTL Immediate readback seems faulty on some chips. I suspect it takes a while to get through the fifo to the actual register backbone. There's no need to read it back, so, just write the driver's copy of the register's value directly. Should fix bug 24535 and possibly 24218 Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-11-03 05:01:27 +08:00
tmp |= BUF_SWAP_32BIT;
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
#endif
drm/radeon/kms: Don't RMW CP_RB_CNTL Immediate readback seems faulty on some chips. I suspect it takes a while to get through the fifo to the actual register backbone. There's no need to read it back, so, just write the driver's copy of the register's value directly. Should fix bug 24535 and possibly 24218 Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-11-03 05:01:27 +08:00
WREG32(CP_RB_CNTL, tmp);
drm/radeon/kms: add support for semaphores v3 They are used to sync between rings, while fences sync between a ring and the cpu. v2 Fix radeon_semaphore_driver_fini when no semaphore were allocated. v3 Initialize list early on to avoid issue in case or early error Signed-off-by: Christian König <deathsimple@vodafone.de> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-09-16 01:02:22 +08:00
WREG32(CP_SEM_WAIT_TIMER, 0x0);
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
/* Set the write pointer delay */
WREG32(CP_RB_WPTR_DELAY, 0);
/* Initialize the ring buffer's read and write pointers */
WREG32(CP_RB_CNTL, tmp | RB_RPTR_WR_ENA);
WREG32(CP_RB_RPTR_WR, 0);
drm/radeon: rename struct radeon_cp to radeon_ring That naming seems to make more sense, since we not only want to run PM4 rings with it. Signed-off-by: Christian König <deathsimple@vodafone.de> Reviewed-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-10-23 18:56:27 +08:00
ring->wptr = 0;
WREG32(CP_RB_WPTR, ring->wptr);
drm/radeon/kms: enable writeback (v2) When writeback is enabled, the GPU shadows writes to certain registers into a buffer in memory. The driver can then read the values from the shadow rather than reading back from the register across the bus. Writeback can be disabled by setting the no_wb module param to 1. On r6xx/r7xx/evergreen, the following registers are shadowed: - CP scratch registers - CP read pointer - IH write pointer On r1xx-rr5xx, the following registers are shadowed: - CP scratch registers - CP read pointer v2: - Combine wb patches for r6xx-evergreen and r1xx-r5xx - Writeback is disabled on AGP boards since it tends to be unreliable on AGP using the gart. - Check radeon_wb_init return values properly. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-08-28 06:25:25 +08:00
/* set the wb address whether it's enabled or not */
drm/radeon/kms: 6xx/7xx big endian fixes agd5f: minor cleanups Signed-off-by: Cédric Cano <ccano@interfaceconcept.com> Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-02-12 08:45:38 +08:00
WREG32(CP_RB_RPTR_ADDR,
((rdev->wb.gpu_addr + RADEON_WB_CP_RPTR_OFFSET) & 0xFFFFFFFC));
drm/radeon/kms: enable writeback (v2) When writeback is enabled, the GPU shadows writes to certain registers into a buffer in memory. The driver can then read the values from the shadow rather than reading back from the register across the bus. Writeback can be disabled by setting the no_wb module param to 1. On r6xx/r7xx/evergreen, the following registers are shadowed: - CP scratch registers - CP read pointer - IH write pointer On r1xx-rr5xx, the following registers are shadowed: - CP scratch registers - CP read pointer v2: - Combine wb patches for r6xx-evergreen and r1xx-r5xx - Writeback is disabled on AGP boards since it tends to be unreliable on AGP using the gart. - Check radeon_wb_init return values properly. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-08-28 06:25:25 +08:00
WREG32(CP_RB_RPTR_ADDR_HI, upper_32_bits(rdev->wb.gpu_addr + RADEON_WB_CP_RPTR_OFFSET) & 0xFF);
WREG32(SCRATCH_ADDR, ((rdev->wb.gpu_addr + RADEON_WB_SCRATCH_OFFSET) >> 8) & 0xFFFFFFFF);
if (rdev->wb.enabled)
WREG32(SCRATCH_UMSK, 0xff);
else {
tmp |= RB_NO_UPDATE;
WREG32(SCRATCH_UMSK, 0);
}
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
mdelay(1);
WREG32(CP_RB_CNTL, tmp);
drm/radeon: rename struct radeon_cp to radeon_ring That naming seems to make more sense, since we not only want to run PM4 rings with it. Signed-off-by: Christian König <deathsimple@vodafone.de> Reviewed-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-10-23 18:56:27 +08:00
WREG32(CP_RB_BASE, ring->gpu_addr >> 8);
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
WREG32(CP_DEBUG, (1 << 27) | (1 << 28));
drm/radeon: rename struct radeon_cp to radeon_ring That naming seems to make more sense, since we not only want to run PM4 rings with it. Signed-off-by: Christian König <deathsimple@vodafone.de> Reviewed-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-10-23 18:56:27 +08:00
ring->rptr = RREG32(CP_RB_RPTR);
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
r600_cp_start(rdev);
drm/radeon: rename struct radeon_cp to radeon_ring That naming seems to make more sense, since we not only want to run PM4 rings with it. Signed-off-by: Christian König <deathsimple@vodafone.de> Reviewed-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-10-23 18:56:27 +08:00
ring->ready = true;
drm/radeon/kms: make ring_start, ring_test, and ib_test per ring Each ring type may need a different variant. Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Reviewed-by: Christian König<christian.koenig@amd.com> Reviewed-by: Michel Dänzer <michel.daenzer@amd.com> Reviewed-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2012-02-24 06:53:45 +08:00
r = radeon_ring_test(rdev, RADEON_RING_TYPE_GFX_INDEX, ring);
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
if (r) {
drm/radeon: rename struct radeon_cp to radeon_ring That naming seems to make more sense, since we not only want to run PM4 rings with it. Signed-off-by: Christian König <deathsimple@vodafone.de> Reviewed-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-10-23 18:56:27 +08:00
ring->ready = false;
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
return r;
}
return 0;
}
drm/radeon: rename struct radeon_cp to radeon_ring That naming seems to make more sense, since we not only want to run PM4 rings with it. Signed-off-by: Christian König <deathsimple@vodafone.de> Reviewed-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-10-23 18:56:27 +08:00
void r600_ring_init(struct radeon_device *rdev, struct radeon_ring *ring, unsigned ring_size)
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
{
u32 rb_bufsz;
drm/radeon: record what is next valid wptr for each ring v4 Before emitting any indirect buffer, emit the offset of the next valid ring content if any. This allow code that want to resume ring to resume ring right after ib that caused GPU lockup. v2: use scratch registers instead of storing it into memory v3: skip over the surface sync for ni and si as well v4: use SET_CONFIG_REG instead of PACKET0 Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Christian König <deathsimple@vodafone.de> Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
2012-07-06 22:22:55 +08:00
int r;
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
/* Align ring size */
rb_bufsz = drm_order(ring_size / 8);
ring_size = (1 << (rb_bufsz + 1)) * 4;
drm/radeon: rename struct radeon_cp to radeon_ring That naming seems to make more sense, since we not only want to run PM4 rings with it. Signed-off-by: Christian König <deathsimple@vodafone.de> Reviewed-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-10-23 18:56:27 +08:00
ring->ring_size = ring_size;
ring->align_mask = 16 - 1;
drm/radeon: record what is next valid wptr for each ring v4 Before emitting any indirect buffer, emit the offset of the next valid ring content if any. This allow code that want to resume ring to resume ring right after ib that caused GPU lockup. v2: use scratch registers instead of storing it into memory v3: skip over the surface sync for ni and si as well v4: use SET_CONFIG_REG instead of PACKET0 Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Christian König <deathsimple@vodafone.de> Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
2012-07-06 22:22:55 +08:00
drm/radeon: update rptr saving logic for memory buffers Add support for using memory buffers rather than scratch registers. Some rings may not be able to write to scratch registers. Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Reviewed-by: Christian König <christian.koenig@amd.com>
2012-07-18 02:02:31 +08:00
if (radeon_ring_supports_scratch_reg(rdev, ring)) {
r = radeon_scratch_get(rdev, &ring->rptr_save_reg);
if (r) {
DRM_ERROR("failed to get scratch reg for rptr save (%d).\n", r);
ring->rptr_save_reg = 0;
}
drm/radeon: record what is next valid wptr for each ring v4 Before emitting any indirect buffer, emit the offset of the next valid ring content if any. This allow code that want to resume ring to resume ring right after ib that caused GPU lockup. v2: use scratch registers instead of storing it into memory v3: skip over the surface sync for ni and si as well v4: use SET_CONFIG_REG instead of PACKET0 Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Christian König <deathsimple@vodafone.de> Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
2012-07-06 22:22:55 +08:00
}
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
}
drm/radeon/kms: don't call suspend path before cleaning up GPU In suspend path we unmap the GART table while in cleaning up path we will unbind buffer and thus try to write to unmapped GART leading to oops. In order to avoid this we don't call the suspend path in cleanup path. Cleanup path is clever enough to desactive GPU like the suspend path is doing, thus this was redondant. Tested on: RV370, R420, RV515, RV570, RV610, RV770 (all PCIE) Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-02-02 18:51:45 +08:00
void r600_cp_fini(struct radeon_device *rdev)
{
drm/radeon: record what is next valid wptr for each ring v4 Before emitting any indirect buffer, emit the offset of the next valid ring content if any. This allow code that want to resume ring to resume ring right after ib that caused GPU lockup. v2: use scratch registers instead of storing it into memory v3: skip over the surface sync for ni and si as well v4: use SET_CONFIG_REG instead of PACKET0 Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Christian König <deathsimple@vodafone.de> Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
2012-07-06 22:22:55 +08:00
struct radeon_ring *ring = &rdev->ring[RADEON_RING_TYPE_GFX_INDEX];
drm/radeon/kms: don't call suspend path before cleaning up GPU In suspend path we unmap the GART table while in cleaning up path we will unbind buffer and thus try to write to unmapped GART leading to oops. In order to avoid this we don't call the suspend path in cleanup path. Cleanup path is clever enough to desactive GPU like the suspend path is doing, thus this was redondant. Tested on: RV370, R420, RV515, RV570, RV610, RV770 (all PCIE) Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-02-02 18:51:45 +08:00
r600_cp_stop(rdev);
drm/radeon: record what is next valid wptr for each ring v4 Before emitting any indirect buffer, emit the offset of the next valid ring content if any. This allow code that want to resume ring to resume ring right after ib that caused GPU lockup. v2: use scratch registers instead of storing it into memory v3: skip over the surface sync for ni and si as well v4: use SET_CONFIG_REG instead of PACKET0 Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Christian König <deathsimple@vodafone.de> Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
2012-07-06 22:22:55 +08:00
radeon_ring_fini(rdev, ring);
radeon_scratch_free(rdev, ring->rptr_save_reg);
drm/radeon/kms: don't call suspend path before cleaning up GPU In suspend path we unmap the GART table while in cleaning up path we will unbind buffer and thus try to write to unmapped GART leading to oops. In order to avoid this we don't call the suspend path in cleanup path. Cleanup path is clever enough to desactive GPU like the suspend path is doing, thus this was redondant. Tested on: RV370, R420, RV515, RV570, RV610, RV770 (all PCIE) Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-02-02 18:51:45 +08:00
}
drm/radeon/kms: Add initial support for async DMA on r6xx/r7xx Uses the new multi-ring infrastucture. 6xx/7xx has a single async DMA ring. Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2012-09-28 03:08:35 +08:00
/*
* DMA
* Starting with R600, the GPU has an asynchronous
* DMA engine. The programming model is very similar
* to the 3D engine (ring buffer, IBs, etc.), but the
* DMA controller has it's own packet format that is
* different form the PM4 format used by the 3D engine.
* It supports copying data, writing embedded data,
* solid fills, and a number of other things. It also
* has support for tiling/detiling of buffers.
*/
/**
* r600_dma_stop - stop the async dma engine
*
* @rdev: radeon_device pointer
*
* Stop the async dma engine (r6xx-evergreen).
*/
void r600_dma_stop(struct radeon_device *rdev)
{
u32 rb_cntl = RREG32(DMA_RB_CNTL);
radeon_ttm_set_active_vram_size(rdev, rdev->mc.visible_vram_size);
rb_cntl &= ~DMA_RB_ENABLE;
WREG32(DMA_RB_CNTL, rb_cntl);
rdev->ring[R600_RING_TYPE_DMA_INDEX].ready = false;
}
/**
* r600_dma_resume - setup and start the async dma engine
*
* @rdev: radeon_device pointer
*
* Set up the DMA ring buffer and enable it. (r6xx-evergreen).
* Returns 0 for success, error for failure.
*/
int r600_dma_resume(struct radeon_device *rdev)
{
struct radeon_ring *ring = &rdev->ring[R600_RING_TYPE_DMA_INDEX];
u32 rb_cntl, dma_cntl;
u32 rb_bufsz;
int r;
/* Reset dma */
if (rdev->family >= CHIP_RV770)
WREG32(SRBM_SOFT_RESET, RV770_SOFT_RESET_DMA);
else
WREG32(SRBM_SOFT_RESET, SOFT_RESET_DMA);
RREG32(SRBM_SOFT_RESET);
udelay(50);
WREG32(SRBM_SOFT_RESET, 0);
WREG32(DMA_SEM_INCOMPLETE_TIMER_CNTL, 0);
WREG32(DMA_SEM_WAIT_FAIL_TIMER_CNTL, 0);
/* Set ring buffer size in dwords */
rb_bufsz = drm_order(ring->ring_size / 4);
rb_cntl = rb_bufsz << 1;
#ifdef __BIG_ENDIAN
rb_cntl |= DMA_RB_SWAP_ENABLE | DMA_RPTR_WRITEBACK_SWAP_ENABLE;
#endif
WREG32(DMA_RB_CNTL, rb_cntl);
/* Initialize the ring buffer's read and write pointers */
WREG32(DMA_RB_RPTR, 0);
WREG32(DMA_RB_WPTR, 0);
/* set the wb address whether it's enabled or not */
WREG32(DMA_RB_RPTR_ADDR_HI,
upper_32_bits(rdev->wb.gpu_addr + R600_WB_DMA_RPTR_OFFSET) & 0xFF);
WREG32(DMA_RB_RPTR_ADDR_LO,
((rdev->wb.gpu_addr + R600_WB_DMA_RPTR_OFFSET) & 0xFFFFFFFC));
if (rdev->wb.enabled)
rb_cntl |= DMA_RPTR_WRITEBACK_ENABLE;
WREG32(DMA_RB_BASE, ring->gpu_addr >> 8);
/* enable DMA IBs */
WREG32(DMA_IB_CNTL, DMA_IB_ENABLE);
dma_cntl = RREG32(DMA_CNTL);
dma_cntl &= ~CTXEMPTY_INT_ENABLE;
WREG32(DMA_CNTL, dma_cntl);
if (rdev->family >= CHIP_RV770)
WREG32(DMA_MODE, 1);
ring->wptr = 0;
WREG32(DMA_RB_WPTR, ring->wptr << 2);
ring->rptr = RREG32(DMA_RB_RPTR) >> 2;
WREG32(DMA_RB_CNTL, rb_cntl | DMA_RB_ENABLE);
ring->ready = true;
r = radeon_ring_test(rdev, R600_RING_TYPE_DMA_INDEX, ring);
if (r) {
ring->ready = false;
return r;
}
radeon_ttm_set_active_vram_size(rdev, rdev->mc.real_vram_size);
return 0;
}
/**
* r600_dma_fini - tear down the async dma engine
*
* @rdev: radeon_device pointer
*
* Stop the async dma engine and free the ring (r6xx-evergreen).
*/
void r600_dma_fini(struct radeon_device *rdev)
{
r600_dma_stop(rdev);
radeon_ring_fini(rdev, &rdev->ring[R600_RING_TYPE_DMA_INDEX]);
}
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
/*
* GPU scratch registers helpers function.
*/
void r600_scratch_init(struct radeon_device *rdev)
{
int i;
rdev->scratch.num_reg = 7;
drm/radeon/kms: enable writeback (v2) When writeback is enabled, the GPU shadows writes to certain registers into a buffer in memory. The driver can then read the values from the shadow rather than reading back from the register across the bus. Writeback can be disabled by setting the no_wb module param to 1. On r6xx/r7xx/evergreen, the following registers are shadowed: - CP scratch registers - CP read pointer - IH write pointer On r1xx-rr5xx, the following registers are shadowed: - CP scratch registers - CP read pointer v2: - Combine wb patches for r6xx-evergreen and r1xx-r5xx - Writeback is disabled on AGP boards since it tends to be unreliable on AGP using the gart. - Check radeon_wb_init return values properly. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-08-28 06:25:25 +08:00
rdev->scratch.reg_base = SCRATCH_REG0;
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
for (i = 0; i < rdev->scratch.num_reg; i++) {
rdev->scratch.free[i] = true;
drm/radeon/kms: enable writeback (v2) When writeback is enabled, the GPU shadows writes to certain registers into a buffer in memory. The driver can then read the values from the shadow rather than reading back from the register across the bus. Writeback can be disabled by setting the no_wb module param to 1. On r6xx/r7xx/evergreen, the following registers are shadowed: - CP scratch registers - CP read pointer - IH write pointer On r1xx-rr5xx, the following registers are shadowed: - CP scratch registers - CP read pointer v2: - Combine wb patches for r6xx-evergreen and r1xx-r5xx - Writeback is disabled on AGP boards since it tends to be unreliable on AGP using the gart. - Check radeon_wb_init return values properly. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-08-28 06:25:25 +08:00
rdev->scratch.reg[i] = rdev->scratch.reg_base + (i * 4);
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
}
}
drm/radeon: rename struct radeon_cp to radeon_ring That naming seems to make more sense, since we not only want to run PM4 rings with it. Signed-off-by: Christian König <deathsimple@vodafone.de> Reviewed-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-10-23 18:56:27 +08:00
int r600_ring_test(struct radeon_device *rdev, struct radeon_ring *ring)
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
{
uint32_t scratch;
uint32_t tmp = 0;
drm/radeon: remove radeon_ring_index() Just store the index in the ring structure. Idea taken from one of Jerome's wip rptr patches. Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Reviewed-by: Christian König <christian.koenig@amd.com>
2012-07-18 02:02:30 +08:00
unsigned i;
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
int r;
r = radeon_scratch_get(rdev, &scratch);
if (r) {
DRM_ERROR("radeon: cp failed to get scratch reg (%d).\n", r);
return r;
}
WREG32(scratch, 0xCAFEDEAD);
drm/radeon: rename struct radeon_cp to radeon_ring That naming seems to make more sense, since we not only want to run PM4 rings with it. Signed-off-by: Christian König <deathsimple@vodafone.de> Reviewed-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-10-23 18:56:27 +08:00
r = radeon_ring_lock(rdev, ring, 3);
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
if (r) {
drm/radeon: remove radeon_ring_index() Just store the index in the ring structure. Idea taken from one of Jerome's wip rptr patches. Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Reviewed-by: Christian König <christian.koenig@amd.com>
2012-07-18 02:02:30 +08:00
DRM_ERROR("radeon: cp failed to lock ring %d (%d).\n", ring->idx, r);
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
radeon_scratch_free(rdev, scratch);
return r;
}
drm/radeon: rename struct radeon_cp to radeon_ring That naming seems to make more sense, since we not only want to run PM4 rings with it. Signed-off-by: Christian König <deathsimple@vodafone.de> Reviewed-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-10-23 18:56:27 +08:00
radeon_ring_write(ring, PACKET3(PACKET3_SET_CONFIG_REG, 1));
radeon_ring_write(ring, ((scratch - PACKET3_SET_CONFIG_REG_OFFSET) >> 2));
radeon_ring_write(ring, 0xDEADBEEF);
radeon_ring_unlock_commit(rdev, ring);
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
for (i = 0; i < rdev->usec_timeout; i++) {
tmp = RREG32(scratch);
if (tmp == 0xDEADBEEF)
break;
DRM_UDELAY(1);
}
if (i < rdev->usec_timeout) {
drm/radeon: remove radeon_ring_index() Just store the index in the ring structure. Idea taken from one of Jerome's wip rptr patches. Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Reviewed-by: Christian König <christian.koenig@amd.com>
2012-07-18 02:02:30 +08:00
DRM_INFO("ring test on %d succeeded in %d usecs\n", ring->idx, i);
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
} else {
drm/radeon: make cp variable an array Replace cp, cp1 and cp2 members with just an array of radeon_cp structs. Signed-off-by: Christian König <deathsimple@vodafone.de> Reviewed-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-10-13 19:19:22 +08:00
DRM_ERROR("radeon: ring %d test failed (scratch(0x%04X)=0x%08X)\n",
drm/radeon: remove radeon_ring_index() Just store the index in the ring structure. Idea taken from one of Jerome's wip rptr patches. Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Reviewed-by: Christian König <christian.koenig@amd.com>
2012-07-18 02:02:30 +08:00
ring->idx, scratch, tmp);
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
r = -EINVAL;
}
radeon_scratch_free(rdev, scratch);
return r;
}
drm/radeon/kms: Add initial support for async DMA on r6xx/r7xx Uses the new multi-ring infrastucture. 6xx/7xx has a single async DMA ring. Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2012-09-28 03:08:35 +08:00
/**
* r600_dma_ring_test - simple async dma engine test
*
* @rdev: radeon_device pointer
* @ring: radeon_ring structure holding ring information
*
* Test the DMA engine by writing using it to write an
* value to memory. (r6xx-SI).
* Returns 0 for success, error for failure.
*/
int r600_dma_ring_test(struct radeon_device *rdev,
struct radeon_ring *ring)
{
unsigned i;
int r;
void __iomem *ptr = (void *)rdev->vram_scratch.ptr;
u32 tmp;
if (!ptr) {
DRM_ERROR("invalid vram scratch pointer\n");
return -EINVAL;
}
tmp = 0xCAFEDEAD;
writel(tmp, ptr);
r = radeon_ring_lock(rdev, ring, 4);
if (r) {
DRM_ERROR("radeon: dma failed to lock ring %d (%d).\n", ring->idx, r);
return r;
}
radeon_ring_write(ring, DMA_PACKET(DMA_PACKET_WRITE, 0, 0, 1));
radeon_ring_write(ring, rdev->vram_scratch.gpu_addr & 0xfffffffc);
radeon_ring_write(ring, upper_32_bits(rdev->vram_scratch.gpu_addr) & 0xff);
radeon_ring_write(ring, 0xDEADBEEF);
radeon_ring_unlock_commit(rdev, ring);
for (i = 0; i < rdev->usec_timeout; i++) {
tmp = readl(ptr);
if (tmp == 0xDEADBEEF)
break;
DRM_UDELAY(1);
}
if (i < rdev->usec_timeout) {
DRM_INFO("ring test on %d succeeded in %d usecs\n", ring->idx, i);
} else {
DRM_ERROR("radeon: ring %d test failed (0x%08X)\n",
ring->idx, tmp);
r = -EINVAL;
}
return r;
}
/*
* CP fences/semaphores
*/
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
void r600_fence_ring_emit(struct radeon_device *rdev,
struct radeon_fence *fence)
{
drm/radeon: rename struct radeon_cp to radeon_ring That naming seems to make more sense, since we not only want to run PM4 rings with it. Signed-off-by: Christian König <deathsimple@vodafone.de> Reviewed-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-10-23 18:56:27 +08:00
struct radeon_ring *ring = &rdev->ring[fence->ring];
drm/radeon: make all functions work with multiple rings. Give all asic and radeon_ring_* functions a radeon_cp parameter, so they know the ring to work with. Signed-off-by: Christian König <deathsimple@vodafone.de> Reviewed-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-09-23 21:11:23 +08:00
drm/radeon/kms/r6xx+: use new style fencing (v3) On r6xx+ a newer fence mechanism was implemented to replace the old wait_until plus scratch regs setup. A single EOP event will flush the destination caches, write a fence value, and generate an interrupt. This is the recommended fence mechanism on r6xx+ asics. This requires my previous writeback patch. v2: fix typo that enabled event fence checking on all asics rather than just r6xx+. v3: properly enable EOP interrupts Should fix: https://bugs.freedesktop.org/show_bug.cgi?id=29972 Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-09-04 17:04:34 +08:00
if (rdev->wb.use_event) {
drm/radeon: precompute fence cpu/gpu addr once v3 Add a start fence driver helper function which will be call once for each ring and will compute cpu/gpu addr for fence depending on wether to use wb buffer or scratch reg. This patch replace initialize fence driver separately which was broken in regard of GPU lockup. The fence list for created, emited, signaled must be initialize once and only from the asic init callback not from the startup call back which is call from the gpu reset. v2: With this in place we no longer need to know the number of rings in fence_driver_init, also writing to the scratch reg before knowing its offset is a bad idea. v3: rebase on top of change to previous patch in the serie Signed-off-by: Christian König <deathsimple@vodafone.de> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-11-21 04:45:34 +08:00
u64 addr = rdev->fence_drv[fence->ring].gpu_addr;
drm/radeon: flush read cache for gtt with fence on r6xx and newer GPU V3 Cayman seems to be particularly sensitive to read cache returning old data after bind/unbind to GTT. Flush read cache for GTT range with each fences for all new hw. Should fix several rendering glitches. Like V2 flush whole address space V3 also flush shader read cache https://bugs.freedesktop.org/show_bug.cgi?id=40221 https://bugs.freedesktop.org/show_bug.cgi?id=38022 https://bugzilla.redhat.com/show_bug.cgi?id=738790 Signed-off-by: Jerome Glisse <jglisse@redhat.com> Reviewed-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-10-26 23:41:22 +08:00
/* flush read cache over gart */
drm/radeon: rename struct radeon_cp to radeon_ring That naming seems to make more sense, since we not only want to run PM4 rings with it. Signed-off-by: Christian König <deathsimple@vodafone.de> Reviewed-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-10-23 18:56:27 +08:00
radeon_ring_write(ring, PACKET3(PACKET3_SURFACE_SYNC, 3));
radeon_ring_write(ring, PACKET3_TC_ACTION_ENA |
PACKET3_VC_ACTION_ENA |
PACKET3_SH_ACTION_ENA);
radeon_ring_write(ring, 0xFFFFFFFF);
radeon_ring_write(ring, 0);
radeon_ring_write(ring, 10); /* poll interval */
drm/radeon/kms/r6xx+: use new style fencing (v3) On r6xx+ a newer fence mechanism was implemented to replace the old wait_until plus scratch regs setup. A single EOP event will flush the destination caches, write a fence value, and generate an interrupt. This is the recommended fence mechanism on r6xx+ asics. This requires my previous writeback patch. v2: fix typo that enabled event fence checking on all asics rather than just r6xx+. v3: properly enable EOP interrupts Should fix: https://bugs.freedesktop.org/show_bug.cgi?id=29972 Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-09-04 17:04:34 +08:00
/* EVENT_WRITE_EOP - flush caches, send int */
drm/radeon: rename struct radeon_cp to radeon_ring That naming seems to make more sense, since we not only want to run PM4 rings with it. Signed-off-by: Christian König <deathsimple@vodafone.de> Reviewed-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-10-23 18:56:27 +08:00
radeon_ring_write(ring, PACKET3(PACKET3_EVENT_WRITE_EOP, 4));
radeon_ring_write(ring, EVENT_TYPE(CACHE_FLUSH_AND_INV_EVENT_TS) | EVENT_INDEX(5));
radeon_ring_write(ring, addr & 0xffffffff);
radeon_ring_write(ring, (upper_32_bits(addr) & 0xff) | DATA_SEL(1) | INT_SEL(2));
radeon_ring_write(ring, fence->seq);
radeon_ring_write(ring, 0);
drm/radeon/kms/r6xx+: use new style fencing (v3) On r6xx+ a newer fence mechanism was implemented to replace the old wait_until plus scratch regs setup. A single EOP event will flush the destination caches, write a fence value, and generate an interrupt. This is the recommended fence mechanism on r6xx+ asics. This requires my previous writeback patch. v2: fix typo that enabled event fence checking on all asics rather than just r6xx+. v3: properly enable EOP interrupts Should fix: https://bugs.freedesktop.org/show_bug.cgi?id=29972 Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-09-04 17:04:34 +08:00
} else {
drm/radeon: flush read cache for gtt with fence on r6xx and newer GPU V3 Cayman seems to be particularly sensitive to read cache returning old data after bind/unbind to GTT. Flush read cache for GTT range with each fences for all new hw. Should fix several rendering glitches. Like V2 flush whole address space V3 also flush shader read cache https://bugs.freedesktop.org/show_bug.cgi?id=40221 https://bugs.freedesktop.org/show_bug.cgi?id=38022 https://bugzilla.redhat.com/show_bug.cgi?id=738790 Signed-off-by: Jerome Glisse <jglisse@redhat.com> Reviewed-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-10-26 23:41:22 +08:00
/* flush read cache over gart */
drm/radeon: rename struct radeon_cp to radeon_ring That naming seems to make more sense, since we not only want to run PM4 rings with it. Signed-off-by: Christian König <deathsimple@vodafone.de> Reviewed-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-10-23 18:56:27 +08:00
radeon_ring_write(ring, PACKET3(PACKET3_SURFACE_SYNC, 3));
radeon_ring_write(ring, PACKET3_TC_ACTION_ENA |
PACKET3_VC_ACTION_ENA |
PACKET3_SH_ACTION_ENA);
radeon_ring_write(ring, 0xFFFFFFFF);
radeon_ring_write(ring, 0);
radeon_ring_write(ring, 10); /* poll interval */
radeon_ring_write(ring, PACKET3(PACKET3_EVENT_WRITE, 0));
radeon_ring_write(ring, EVENT_TYPE(CACHE_FLUSH_AND_INV_EVENT) | EVENT_INDEX(0));
drm/radeon/kms/r6xx+: use new style fencing (v3) On r6xx+ a newer fence mechanism was implemented to replace the old wait_until plus scratch regs setup. A single EOP event will flush the destination caches, write a fence value, and generate an interrupt. This is the recommended fence mechanism on r6xx+ asics. This requires my previous writeback patch. v2: fix typo that enabled event fence checking on all asics rather than just r6xx+. v3: properly enable EOP interrupts Should fix: https://bugs.freedesktop.org/show_bug.cgi?id=29972 Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-09-04 17:04:34 +08:00
/* wait for 3D idle clean */
drm/radeon: rename struct radeon_cp to radeon_ring That naming seems to make more sense, since we not only want to run PM4 rings with it. Signed-off-by: Christian König <deathsimple@vodafone.de> Reviewed-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-10-23 18:56:27 +08:00
radeon_ring_write(ring, PACKET3(PACKET3_SET_CONFIG_REG, 1));
radeon_ring_write(ring, (WAIT_UNTIL - PACKET3_SET_CONFIG_REG_OFFSET) >> 2);
radeon_ring_write(ring, WAIT_3D_IDLE_bit | WAIT_3D_IDLECLEAN_bit);
drm/radeon/kms/r6xx+: use new style fencing (v3) On r6xx+ a newer fence mechanism was implemented to replace the old wait_until plus scratch regs setup. A single EOP event will flush the destination caches, write a fence value, and generate an interrupt. This is the recommended fence mechanism on r6xx+ asics. This requires my previous writeback patch. v2: fix typo that enabled event fence checking on all asics rather than just r6xx+. v3: properly enable EOP interrupts Should fix: https://bugs.freedesktop.org/show_bug.cgi?id=29972 Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-09-04 17:04:34 +08:00
/* Emit fence sequence & fire IRQ */
drm/radeon: rename struct radeon_cp to radeon_ring That naming seems to make more sense, since we not only want to run PM4 rings with it. Signed-off-by: Christian König <deathsimple@vodafone.de> Reviewed-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-10-23 18:56:27 +08:00
radeon_ring_write(ring, PACKET3(PACKET3_SET_CONFIG_REG, 1));
radeon_ring_write(ring, ((rdev->fence_drv[fence->ring].scratch_reg - PACKET3_SET_CONFIG_REG_OFFSET) >> 2));
radeon_ring_write(ring, fence->seq);
drm/radeon/kms/r6xx+: use new style fencing (v3) On r6xx+ a newer fence mechanism was implemented to replace the old wait_until plus scratch regs setup. A single EOP event will flush the destination caches, write a fence value, and generate an interrupt. This is the recommended fence mechanism on r6xx+ asics. This requires my previous writeback patch. v2: fix typo that enabled event fence checking on all asics rather than just r6xx+. v3: properly enable EOP interrupts Should fix: https://bugs.freedesktop.org/show_bug.cgi?id=29972 Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-09-04 17:04:34 +08:00
/* CP_INTERRUPT packet 3 no longer exists, use packet 0 */
drm/radeon: rename struct radeon_cp to radeon_ring That naming seems to make more sense, since we not only want to run PM4 rings with it. Signed-off-by: Christian König <deathsimple@vodafone.de> Reviewed-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-10-23 18:56:27 +08:00
radeon_ring_write(ring, PACKET0(CP_INT_STATUS, 0));
radeon_ring_write(ring, RB_INT_STAT);
drm/radeon/kms/r6xx+: use new style fencing (v3) On r6xx+ a newer fence mechanism was implemented to replace the old wait_until plus scratch regs setup. A single EOP event will flush the destination caches, write a fence value, and generate an interrupt. This is the recommended fence mechanism on r6xx+ asics. This requires my previous writeback patch. v2: fix typo that enabled event fence checking on all asics rather than just r6xx+. v3: properly enable EOP interrupts Should fix: https://bugs.freedesktop.org/show_bug.cgi?id=29972 Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-09-04 17:04:34 +08:00
}
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
}
drm/radeon/kms: add support for semaphores v3 They are used to sync between rings, while fences sync between a ring and the cpu. v2 Fix radeon_semaphore_driver_fini when no semaphore were allocated. v3 Initialize list early on to avoid issue in case or early error Signed-off-by: Christian König <deathsimple@vodafone.de> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-09-16 01:02:22 +08:00
void r600_semaphore_ring_emit(struct radeon_device *rdev,
drm/radeon: rename struct radeon_cp to radeon_ring That naming seems to make more sense, since we not only want to run PM4 rings with it. Signed-off-by: Christian König <deathsimple@vodafone.de> Reviewed-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-10-23 18:56:27 +08:00
struct radeon_ring *ring,
drm/radeon/kms: add support for semaphores v3 They are used to sync between rings, while fences sync between a ring and the cpu. v2 Fix radeon_semaphore_driver_fini when no semaphore were allocated. v3 Initialize list early on to avoid issue in case or early error Signed-off-by: Christian König <deathsimple@vodafone.de> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-09-16 01:02:22 +08:00
struct radeon_semaphore *semaphore,
drm/radeon: make all functions work with multiple rings. Give all asic and radeon_ring_* functions a radeon_cp parameter, so they know the ring to work with. Signed-off-by: Christian König <deathsimple@vodafone.de> Reviewed-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-09-23 21:11:23 +08:00
bool emit_wait)
drm/radeon/kms: add support for semaphores v3 They are used to sync between rings, while fences sync between a ring and the cpu. v2 Fix radeon_semaphore_driver_fini when no semaphore were allocated. v3 Initialize list early on to avoid issue in case or early error Signed-off-by: Christian König <deathsimple@vodafone.de> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-09-16 01:02:22 +08:00
{
uint64_t addr = semaphore->gpu_addr;
unsigned sel = emit_wait ? PACKET3_SEM_SEL_WAIT : PACKET3_SEM_SEL_SIGNAL;
drm/radeon: fix a semaphore deadlock on pre cayman asics The out of order execution of semaphore commands on pre cayman asics doesn't work correctly and can cause deadlocks, so turn it off for now. Signed-off-by: Christian König <deathsimple@vodafone.de> Reviewed-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2012-03-07 18:28:57 +08:00
if (rdev->family < CHIP_CAYMAN)
sel |= PACKET3_SEM_WAIT_ON_SIGNAL;
drm/radeon: rename struct radeon_cp to radeon_ring That naming seems to make more sense, since we not only want to run PM4 rings with it. Signed-off-by: Christian König <deathsimple@vodafone.de> Reviewed-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-10-23 18:56:27 +08:00
radeon_ring_write(ring, PACKET3(PACKET3_MEM_SEMAPHORE, 1));
radeon_ring_write(ring, addr & 0xffffffff);
radeon_ring_write(ring, (upper_32_bits(addr) & 0xff) | sel);
drm/radeon/kms: add support for semaphores v3 They are used to sync between rings, while fences sync between a ring and the cpu. v2 Fix radeon_semaphore_driver_fini when no semaphore were allocated. v3 Initialize list early on to avoid issue in case or early error Signed-off-by: Christian König <deathsimple@vodafone.de> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-09-16 01:02:22 +08:00
}
drm/radeon/kms: Add initial support for async DMA on r6xx/r7xx Uses the new multi-ring infrastucture. 6xx/7xx has a single async DMA ring. Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2012-09-28 03:08:35 +08:00
/*
* DMA fences/semaphores
*/
/**
* r600_dma_fence_ring_emit - emit a fence on the DMA ring
*
* @rdev: radeon_device pointer
* @fence: radeon fence object
*
* Add a DMA fence packet to the ring to write
* the fence seq number and DMA trap packet to generate
* an interrupt if needed (r6xx-r7xx).
*/
void r600_dma_fence_ring_emit(struct radeon_device *rdev,
struct radeon_fence *fence)
{
struct radeon_ring *ring = &rdev->ring[fence->ring];
u64 addr = rdev->fence_drv[fence->ring].gpu_addr;
drm/radeon: fix fence driver for dma ring when wb is disabled The dma ring can't write to register thus have to write to memory its fence value. This ensure that it doesn't try to use scratch register for dma ring fence driver. Should fix: https://bugs.freedesktop.org/show_bug.cgi?id=58166 Signed-off-by: Jerome Glisse <jglisse@redhat.com> Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
2012-12-13 05:43:15 +08:00
drm/radeon/kms: Add initial support for async DMA on r6xx/r7xx Uses the new multi-ring infrastucture. 6xx/7xx has a single async DMA ring. Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2012-09-28 03:08:35 +08:00
/* write the fence */
radeon_ring_write(ring, DMA_PACKET(DMA_PACKET_FENCE, 0, 0, 0));
radeon_ring_write(ring, addr & 0xfffffffc);
radeon_ring_write(ring, (upper_32_bits(addr) & 0xff));
drm/radeon: fix fence driver for dma ring when wb is disabled The dma ring can't write to register thus have to write to memory its fence value. This ensure that it doesn't try to use scratch register for dma ring fence driver. Should fix: https://bugs.freedesktop.org/show_bug.cgi?id=58166 Signed-off-by: Jerome Glisse <jglisse@redhat.com> Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
2012-12-13 05:43:15 +08:00
radeon_ring_write(ring, lower_32_bits(fence->seq));
drm/radeon/kms: Add initial support for async DMA on r6xx/r7xx Uses the new multi-ring infrastucture. 6xx/7xx has a single async DMA ring. Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2012-09-28 03:08:35 +08:00
/* generate an interrupt */
radeon_ring_write(ring, DMA_PACKET(DMA_PACKET_TRAP, 0, 0, 0));
}
/**
* r600_dma_semaphore_ring_emit - emit a semaphore on the dma ring
*
* @rdev: radeon_device pointer
* @ring: radeon_ring structure holding ring information
* @semaphore: radeon semaphore object
* @emit_wait: wait or signal semaphore
*
* Add a DMA semaphore packet to the ring wait on or signal
* other rings (r6xx-SI).
*/
void r600_dma_semaphore_ring_emit(struct radeon_device *rdev,
struct radeon_ring *ring,
struct radeon_semaphore *semaphore,
bool emit_wait)
{
u64 addr = semaphore->gpu_addr;
u32 s = emit_wait ? 0 : 1;
radeon_ring_write(ring, DMA_PACKET(DMA_PACKET_SEMAPHORE, 0, s, 0));
radeon_ring_write(ring, addr & 0xfffffffc);
radeon_ring_write(ring, upper_32_bits(addr) & 0xff);
}
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
int r600_copy_blit(struct radeon_device *rdev,
drm/radeon/kms: Make GPU/CPU page size handling consistent in blit code (v2) The BO blit code inconsistenly handled the page size. This wasn't an issue on system with 4k pages since the GPU's page size is 4k as well. Switch the driver blit callbacks to take num pages in GPU page units. Fixes lemote mipsel systems using AMD rs780/rs880 chipsets. v2: incorporate suggestions from Michel. Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Reviewed-by: Michel Dänzer <michel.daenzer@amd.com> Cc: stable@kernel.org Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-09-17 00:04:08 +08:00
uint64_t src_offset,
uint64_t dst_offset,
unsigned num_gpu_pages,
drm/radeon: remove radeon_fence_create It is completely unnecessary to create fences before they are emitted, so remove it and a bunch of checks if fences are emitted or not. Signed-off-by: Christian König <deathsimple@vodafone.de> Reviewed-by: Jerome Glisse <jglisse@redhat.com>
2012-05-08 20:24:01 +08:00
struct radeon_fence **fence)
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
{
drm/radeon: rework ring syncing code Move inter ring syncing with semaphores into the existing ring allocations, with that we need to lock the ring mutex only once. Signed-off-by: Christian König <deathsimple@vodafone.de> Reviewed-by: Jerome Glisse <jglisse@redhat.com>
2012-05-10 22:46:43 +08:00
struct radeon_semaphore *sem = NULL;
drm/radeon: remove r600 blit mutex v2 If we don't store local data into global variables it isn't necessary to lock anything. v2: rebased on new SA interface Signed-off-by: Christian König <deathsimple@vodafone.de> Signed-off-by: Dave Airlie <airlied@redhat.com>
2012-05-09 21:35:01 +08:00
struct radeon_sa_bo *vb = NULL;
drm/radeon/kms: Bailout of blit if error happen & protect with mutex V3 If an error happen in r600_blit_prepare_copy report it rather than WARNING and keeping execution. For instance if ib allocation failed we did just warn about but then latter tried to access NULL ib ptr causing oops. This patch also protect r600_copy_blit with a mutex as otherwise one process might overwrite blit temporary data with new one possibly leading to GPU lockup. Should partialy or totaly fix: https://bugzilla.redhat.com/show_bug.cgi?id=553279 V2 failing blit initialization is not fatal, fallback to memcpy when this happen V3 init blit before startup as we pin in startup, remove duplicate code (this one was actualy tested unlike V2) Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-01-22 22:19:00 +08:00
int r;
drm/radeon: rework ring syncing code Move inter ring syncing with semaphores into the existing ring allocations, with that we need to lock the ring mutex only once. Signed-off-by: Christian König <deathsimple@vodafone.de> Reviewed-by: Jerome Glisse <jglisse@redhat.com>
2012-05-10 22:46:43 +08:00
r = r600_blit_prepare_copy(rdev, num_gpu_pages, fence, &vb, &sem);
drm/radeon/kms: Bailout of blit if error happen & protect with mutex V3 If an error happen in r600_blit_prepare_copy report it rather than WARNING and keeping execution. For instance if ib allocation failed we did just warn about but then latter tried to access NULL ib ptr causing oops. This patch also protect r600_copy_blit with a mutex as otherwise one process might overwrite blit temporary data with new one possibly leading to GPU lockup. Should partialy or totaly fix: https://bugzilla.redhat.com/show_bug.cgi?id=553279 V2 failing blit initialization is not fatal, fallback to memcpy when this happen V3 init blit before startup as we pin in startup, remove duplicate code (this one was actualy tested unlike V2) Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-01-22 22:19:00 +08:00
if (r) {
return r;
}
drm/radeon: remove r600 blit mutex v2 If we don't store local data into global variables it isn't necessary to lock anything. v2: rebased on new SA interface Signed-off-by: Christian König <deathsimple@vodafone.de> Signed-off-by: Dave Airlie <airlied@redhat.com>
2012-05-09 21:35:01 +08:00
r600_kms_blit_copy(rdev, src_offset, dst_offset, num_gpu_pages, vb);
drm/radeon: rework ring syncing code Move inter ring syncing with semaphores into the existing ring allocations, with that we need to lock the ring mutex only once. Signed-off-by: Christian König <deathsimple@vodafone.de> Reviewed-by: Jerome Glisse <jglisse@redhat.com>
2012-05-10 22:46:43 +08:00
r600_blit_done_copy(rdev, fence, vb, sem);
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
return 0;
}
drm/radeon/kms: Add initial support for async DMA on r6xx/r7xx Uses the new multi-ring infrastucture. 6xx/7xx has a single async DMA ring. Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2012-09-28 03:08:35 +08:00
/**
* r600_copy_dma - copy pages using the DMA engine
*
* @rdev: radeon_device pointer
* @src_offset: src GPU address
* @dst_offset: dst GPU address
* @num_gpu_pages: number of GPU pages to xfer
* @fence: radeon fence object
*
drm/radeon: split r6xx and r7xx copy_dma functions - r6xx actually uses a slightly different packet format, although both formats seem to work ok. - r7xx doesn't have the count multiple of 2 limitation. Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2013-01-04 22:24:18 +08:00
* Copy GPU paging using the DMA engine (r6xx).
drm/radeon/kms: Add initial support for async DMA on r6xx/r7xx Uses the new multi-ring infrastucture. 6xx/7xx has a single async DMA ring. Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2012-09-28 03:08:35 +08:00
* Used by the radeon ttm implementation to move pages if
* registered as the asic copy callback.
*/
int r600_copy_dma(struct radeon_device *rdev,
uint64_t src_offset, uint64_t dst_offset,
unsigned num_gpu_pages,
struct radeon_fence **fence)
{
struct radeon_semaphore *sem = NULL;
int ring_index = rdev->asic->copy.dma_ring_index;
struct radeon_ring *ring = &rdev->ring[ring_index];
u32 size_in_dw, cur_size_in_dw;
int i, num_loops;
int r = 0;
r = radeon_semaphore_create(rdev, &sem);
if (r) {
DRM_ERROR("radeon: moving bo (%d).\n", r);
return r;
}
size_in_dw = (num_gpu_pages << RADEON_GPU_PAGE_SHIFT) / 4;
drm/radeon: split r6xx and r7xx copy_dma functions - r6xx actually uses a slightly different packet format, although both formats seem to work ok. - r7xx doesn't have the count multiple of 2 limitation. Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2013-01-04 22:24:18 +08:00
num_loops = DIV_ROUND_UP(size_in_dw, 0xFFFE);
r = radeon_ring_lock(rdev, ring, num_loops * 4 + 8);
drm/radeon/kms: Add initial support for async DMA on r6xx/r7xx Uses the new multi-ring infrastucture. 6xx/7xx has a single async DMA ring. Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2012-09-28 03:08:35 +08:00
if (r) {
DRM_ERROR("radeon: moving bo (%d).\n", r);
radeon_semaphore_free(rdev, &sem, NULL);
return r;
}
if (radeon_fence_need_sync(*fence, ring->idx)) {
radeon_semaphore_sync_rings(rdev, sem, (*fence)->ring,
ring->idx);
radeon_fence_note_sync(*fence, ring->idx);
} else {
radeon_semaphore_free(rdev, &sem, NULL);
}
for (i = 0; i < num_loops; i++) {
cur_size_in_dw = size_in_dw;
drm/radeon/r6xx: fix DMA engine for ttm bo transfers count must be a multiple of 2. Fixes crashes on R6xx chips reported by a number of people. Cc: Borislav Petkov <bp@alien8.de> Cc: Markus Trippelsdorf <markus@trippelsdorf.de> Reviewed-by: Jerome Glisse <jglisse@redhat.com> Tested-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2013-01-03 07:30:21 +08:00
if (cur_size_in_dw > 0xFFFE)
cur_size_in_dw = 0xFFFE;
drm/radeon/kms: Add initial support for async DMA on r6xx/r7xx Uses the new multi-ring infrastucture. 6xx/7xx has a single async DMA ring. Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2012-09-28 03:08:35 +08:00
size_in_dw -= cur_size_in_dw;
radeon_ring_write(ring, DMA_PACKET(DMA_PACKET_COPY, 0, 0, cur_size_in_dw));
radeon_ring_write(ring, dst_offset & 0xfffffffc);
radeon_ring_write(ring, src_offset & 0xfffffffc);
drm/radeon: split r6xx and r7xx copy_dma functions - r6xx actually uses a slightly different packet format, although both formats seem to work ok. - r7xx doesn't have the count multiple of 2 limitation. Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2013-01-04 22:24:18 +08:00
radeon_ring_write(ring, (((upper_32_bits(dst_offset) & 0xff) << 16) |
(upper_32_bits(src_offset) & 0xff)));
drm/radeon/kms: Add initial support for async DMA on r6xx/r7xx Uses the new multi-ring infrastucture. 6xx/7xx has a single async DMA ring. Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2012-09-28 03:08:35 +08:00
src_offset += cur_size_in_dw * 4;
dst_offset += cur_size_in_dw * 4;
}
r = radeon_fence_emit(rdev, fence, ring->idx);
if (r) {
radeon_ring_unlock_undo(rdev, ring);
return r;
}
radeon_ring_unlock_commit(rdev, ring);
radeon_semaphore_free(rdev, &sem, *fence);
return r;
}
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
int r600_set_surface_reg(struct radeon_device *rdev, int reg,
uint32_t tiling_flags, uint32_t pitch,
uint32_t offset, uint32_t obj_size)
{
/* FIXME: implement */
return 0;
}
void r600_clear_surface_reg(struct radeon_device *rdev, int reg)
{
/* FIXME: implement */
}
drm/radeon: Mark all possible functions / structs as static Let's allow GCC to optimize better. This exposed some five unused functions, but this patch doesn't remove them. Signed-off-by: Lauri Kasanen <cand@gmx.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2012-09-01 01:43:50 +08:00
static int r600_startup(struct radeon_device *rdev)
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
{
drm/radeon/kms: Add initial support for async DMA on r6xx/r7xx Uses the new multi-ring infrastucture. 6xx/7xx has a single async DMA ring. Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2012-09-28 03:08:35 +08:00
struct radeon_ring *ring;
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
int r;
drm/radeon/kms: add support for gen2 pcie link speeds Supported on rv6xx/r7xx/evergreen. Cards come up in gen1 mode. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-01-07 07:49:35 +08:00
/* enable pcie gen2 link */
r600_pcie_gen2_enable(rdev);
drm/radeon/kms/r600/r700: fallback gracefully on ucode failure Sent the wrong patch earlier. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-10 08:31:44 +08:00
if (!rdev->me_fw || !rdev->pfp_fw || !rdev->rlc_fw) {
r = r600_init_microcode(rdev);
if (r) {
DRM_ERROR("Failed to load firmware!\n");
return r;
}
}
drm/radeon/kms: allocate vram scratch page on 6xx+ The vram scratch was originally only used on some 7xx asics to work around a hw bug. Allocate the scratch page on all 6xx+ radeons and set the MC_VM_SYSTEM_APERTURE_DEFAULT_ADDR to point to it. We shouldn't ever hit it since we limit the system aperture to vram or vram and AGP, but better safe than sorry. Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-10-28 22:30:02 +08:00
r = r600_vram_scratch_init(rdev);
if (r)
return r;
drm/radeon/kms: Fix R600/RV770 startup path & reset We were calling reset unconditionaly in the startup path this is bad we need to call GPU reset for a good reason as after reset the GPU is in unknown states. To avoid any more bad things to happen we now also unconditionaly reinitialize the GPU after reset. This patch fix few issues reported by different people regarding KMS & R6XX/RV7XX hw. Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-10-02 00:02:13 +08:00
r600_mc_program(rdev);
drm/radeon/kms: Fix AGP support for R600/RV770 family (v2) For AGP to work unmapped access must cover VRAM & AGP as AGP is treated like VRAM by the GPU (ie physical address). This patch properly setup the virtual memory system aperture to cover AGP if AGP is enabled. It seems that there is memory corruption after resume when using AGP (RV770 seems unaffected thought). Version 2 just fix merge issue with updated AGP fallback patch. Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-10-07 01:04:30 +08:00
if (rdev->flags & RADEON_IS_AGP) {
r600_agp_enable(rdev);
} else {
r = r600_pcie_gart_enable(rdev);
if (r)
return r;
}
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
r600_gpu_init(rdev);
drm/radeon/kms: move blit initialization after we disabled VGA VGA might be overwritting VRAM and corrupt our blit shader leading to corruption, it likely won't happen if you load fbcon right after radeon. Thanks to Shawn Starr and Andre Maasikas for tracking down this issue. Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-02-05 00:27:27 +08:00
r = r600_blit_init(rdev);
if (r) {
r600_blit_fini(rdev);
drm/radeon/kms: reorganize copy callbacks tidy up the radeon_asic struct, handle multiple rings better. Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Reviewed-by: Christian König<christian.koenig@amd.com> Reviewed-by: Michel Dänzer <michel.daenzer@amd.com> Reviewed-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2012-02-24 06:53:42 +08:00
rdev->asic->copy.copy = NULL;
drm/radeon/kms: move blit initialization after we disabled VGA VGA might be overwritting VRAM and corrupt our blit shader leading to corruption, it likely won't happen if you load fbcon right after radeon. Thanks to Shawn Starr and Andre Maasikas for tracking down this issue. Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-02-05 00:27:27 +08:00
dev_warn(rdev->dev, "failed blitter (%d) falling back to memcpy\n", r);
}
drm/radeon/kms: clean up r6xx/r7xx blit init (v2) Move common code to init function. v2: make sure the bo is pinned after init as well. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-08-07 09:36:58 +08:00
drm/radeon/kms: enable writeback (v2) When writeback is enabled, the GPU shadows writes to certain registers into a buffer in memory. The driver can then read the values from the shadow rather than reading back from the register across the bus. Writeback can be disabled by setting the no_wb module param to 1. On r6xx/r7xx/evergreen, the following registers are shadowed: - CP scratch registers - CP read pointer - IH write pointer On r1xx-rr5xx, the following registers are shadowed: - CP scratch registers - CP read pointer v2: - Combine wb patches for r6xx-evergreen and r1xx-r5xx - Writeback is disabled on AGP boards since it tends to be unreliable on AGP using the gart. - Check radeon_wb_init return values properly. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-08-28 06:25:25 +08:00
/* allocate wb buffer */
r = radeon_wb_init(rdev);
if (r)
return r;
drm/radeon: precompute fence cpu/gpu addr once v3 Add a start fence driver helper function which will be call once for each ring and will compute cpu/gpu addr for fence depending on wether to use wb buffer or scratch reg. This patch replace initialize fence driver separately which was broken in regard of GPU lockup. The fence list for created, emited, signaled must be initialize once and only from the asic init callback not from the startup call back which is call from the gpu reset. v2: With this in place we no longer need to know the number of rings in fence_driver_init, also writing to the scratch reg before knowing its offset is a bad idea. v3: rebase on top of change to previous patch in the serie Signed-off-by: Christian König <deathsimple@vodafone.de> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-11-21 04:45:34 +08:00
r = radeon_fence_driver_start_ring(rdev, RADEON_RING_TYPE_GFX_INDEX);
if (r) {
dev_err(rdev->dev, "failed initializing CP fences (%d).\n", r);
return r;
}
drm/radeon/kms: Add initial support for async DMA on r6xx/r7xx Uses the new multi-ring infrastucture. 6xx/7xx has a single async DMA ring. Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2012-09-28 03:08:35 +08:00
r = radeon_fence_driver_start_ring(rdev, R600_RING_TYPE_DMA_INDEX);
if (r) {
dev_err(rdev->dev, "failed initializing DMA fences (%d).\n", r);
return r;
}
drm/radeon/kms: Add support for interrupts on r6xx/r7xx chips (v3) This enables the use of interrupts on r6xx/r7xx hardware. Interrupts are implemented via a ring buffer. The GPU adds interrupts vectors to the ring and the host reads them off in the interrupt handler. The interrupt controller requires firmware like the CP. This firmware must be installed and accessble to the firmware loader for interrupts to function. MSIs don't seem to work on my RS780. They work fine on all my discrete cards. I'm not sure about other RS780s or RS880s. I've disabled MSIs on RS780 and RS880, but it would probably be worth checking on some other systems. v2 - fix some checkpatch.pl problems; re-read the disp int status reg if we restart the ih; v3 - remove the irq handler if r600_irq_init() fails; remove spinlock in r600_ih_ring_fini(); move ih rb overflow check to r600_get_ih_wptr(); move irq ack to separate function; Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-02 02:43:46 +08:00
/* Enable IRQ */
r = r600_irq_init(rdev);
if (r) {
DRM_ERROR("radeon: IH init failed (%d).\n", r);
radeon_irq_kms_fini(rdev);
return r;
}
r600_irq_set(rdev);
drm/radeon/kms: Add initial support for async DMA on r6xx/r7xx Uses the new multi-ring infrastucture. 6xx/7xx has a single async DMA ring. Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2012-09-28 03:08:35 +08:00
ring = &rdev->ring[RADEON_RING_TYPE_GFX_INDEX];
drm/radeon: rename struct radeon_cp to radeon_ring That naming seems to make more sense, since we not only want to run PM4 rings with it. Signed-off-by: Christian König <deathsimple@vodafone.de> Reviewed-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-10-23 18:56:27 +08:00
r = radeon_ring_init(rdev, ring, ring->ring_size, RADEON_WB_CP_RPTR_OFFSET,
drm/radeon/kms: add some new ring params to better handle other ring types Some rptr/wptrs fields have different offsets and not all rings are pm4 so add a new nop field. Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Christian König <deathsimple@vodafone.de> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-11-18 03:25:56 +08:00
R600_CP_RB_RPTR, R600_CP_RB_WPTR,
0, 0xfffff, RADEON_CP_PACKET2);
drm/radeon/kms: Add initial support for async DMA on r6xx/r7xx Uses the new multi-ring infrastucture. 6xx/7xx has a single async DMA ring. Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2012-09-28 03:08:35 +08:00
if (r)
return r;
drm/radeon: make ring rptr and wptr register offsets variable Every ring seems to have the concept of read and write pointers. Make the register offset variable so we can use the functions for different types of rings. Signed-off-by: Christian König <deathsimple@vodafone.de> Reviewed-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-10-13 18:48:45 +08:00
drm/radeon/kms: Add initial support for async DMA on r6xx/r7xx Uses the new multi-ring infrastucture. 6xx/7xx has a single async DMA ring. Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2012-09-28 03:08:35 +08:00
ring = &rdev->ring[R600_RING_TYPE_DMA_INDEX];
r = radeon_ring_init(rdev, ring, ring->ring_size, R600_WB_DMA_RPTR_OFFSET,
DMA_RB_RPTR, DMA_RB_WPTR,
2, 0x3fffc, DMA_PACKET(DMA_PACKET_NOP, 0, 0, 0));
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
if (r)
return r;
drm/radeon/kms: Add initial support for async DMA on r6xx/r7xx Uses the new multi-ring infrastucture. 6xx/7xx has a single async DMA ring. Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2012-09-28 03:08:35 +08:00
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
r = r600_cp_load_microcode(rdev);
if (r)
return r;
r = r600_cp_resume(rdev);
if (r)
return r;
drm/radeon/kms: enable writeback (v2) When writeback is enabled, the GPU shadows writes to certain registers into a buffer in memory. The driver can then read the values from the shadow rather than reading back from the register across the bus. Writeback can be disabled by setting the no_wb module param to 1. On r6xx/r7xx/evergreen, the following registers are shadowed: - CP scratch registers - CP read pointer - IH write pointer On r1xx-rr5xx, the following registers are shadowed: - CP scratch registers - CP read pointer v2: - Combine wb patches for r6xx-evergreen and r1xx-r5xx - Writeback is disabled on AGP boards since it tends to be unreliable on AGP using the gart. - Check radeon_wb_init return values properly. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-08-28 06:25:25 +08:00
drm/radeon/kms: Add initial support for async DMA on r6xx/r7xx Uses the new multi-ring infrastucture. 6xx/7xx has a single async DMA ring. Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2012-09-28 03:08:35 +08:00
r = r600_dma_resume(rdev);
if (r)
return r;
drm/radeon: remove ip_pool start/suspend The IB pool is in gart memory, so it is completely superfluous to unpin / repin it on suspend / resume. Signed-off-by: Christian König <deathsimple@vodafone.de> Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
2012-07-05 17:55:34 +08:00
r = radeon_ib_pool_init(rdev);
if (r) {
dev_err(rdev->dev, "IB initialization failed (%d).\n", r);
drm/radeon: introduce a sub allocator and convert ib pool to it v4 Somewhat specializaed sub-allocator designed to perform sub-allocation for command buffer not only for current cs ioctl but for future command submission ioctl as well. Patch also convert current ib pool to use the sub allocator. Idea is that ib poll buffer can be share with other command buffer submission not having 64K granularity. v2 Harmonize pool handling and add suspend/resume callback to pin/unpin sa bo (tested on rv280, rv370, r420, rv515, rv610, rv710, redwood, cayman, rs480, rs690, rs880) v3 Simplify allocator v4 Fix radeon_ib_get error path to properly free fence Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-11-16 00:48:34 +08:00
return r;
drm/radeon: remove ip_pool start/suspend The IB pool is in gart memory, so it is completely superfluous to unpin / repin it on suspend / resume. Signed-off-by: Christian König <deathsimple@vodafone.de> Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
2012-07-05 17:55:34 +08:00
}
drm/radeon: introduce a sub allocator and convert ib pool to it v4 Somewhat specializaed sub-allocator designed to perform sub-allocation for command buffer not only for current cs ioctl but for future command submission ioctl as well. Patch also convert current ib pool to use the sub allocator. Idea is that ib poll buffer can be share with other command buffer submission not having 64K granularity. v2 Harmonize pool handling and add suspend/resume callback to pin/unpin sa bo (tested on rv280, rv370, r420, rv515, rv610, rv710, redwood, cayman, rs480, rs690, rs880) v3 Simplify allocator v4 Fix radeon_ib_get error path to properly free fence Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-11-16 00:48:34 +08:00
drm/radeon: make audio_init consistent across asics Call it in the asic startup callback on all asics. Previously r600 and rv770 called it in the startup and resume callbacks while all the other asics called it in the startup callback. Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Reviewed-by: Rafał Miłecki <zajec5@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2012-06-05 05:18:51 +08:00
r = r600_audio_init(rdev);
if (r) {
DRM_ERROR("radeon: audio init failed\n");
return r;
}
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
return 0;
}
drm/vgaarb: add VGA arbitration support to the drm and kms. VGA arb requires DRM support for non-kms drivers, to turn on/off irqs when disabling the mem/io regions. VGA arb requires KMS support for GPUs where we can turn off VGA decoding. Currently we know how to do this for intel and radeon kms drivers, which allows them to be removed from the arbiter. This patch comes from Fedora rawhide kernel. Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-21 12:33:58 +08:00
void r600_vga_set_state(struct radeon_device *rdev, bool state)
{
uint32_t temp;
temp = RREG32(CONFIG_CNTL);
if (state == false) {
temp &= ~(1<<0);
temp |= (1<<1);
} else {
temp &= ~(1<<1);
}
WREG32(CONFIG_CNTL, temp);
}
drm/radeon/kms: move around new init path code to avoid posting at init We really don't want to post the card at init, it takes a relatively long time and isn't required, so split the resume path into a startup path called by both init/resume and separate resume entry point to do posting. Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-18 13:19:37 +08:00
int r600_resume(struct radeon_device *rdev)
{
int r;
drm/radeon/kms: Fix AGP support for R600/RV770 family (v2) For AGP to work unmapped access must cover VRAM & AGP as AGP is treated like VRAM by the GPU (ie physical address). This patch properly setup the virtual memory system aperture to cover AGP if AGP is enabled. It seems that there is memory corruption after resume when using AGP (RV770 seems unaffected thought). Version 2 just fix merge issue with updated AGP fallback patch. Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-10-07 01:04:30 +08:00
/* Do not reset GPU before posting, on r600 hw unlike on r500 hw,
* posting will perform necessary task to bring back GPU into good
* shape.
*/
drm/radeon/kms: move around new init path code to avoid posting at init We really don't want to post the card at init, it takes a relatively long time and isn't required, so split the resume path into a startup path called by both init/resume and separate resume entry point to do posting. Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-18 13:19:37 +08:00
/* post card */
drm/radeon/kms: R600/RV770 remove dead code and print message for wrong BIOS R600 & RV770 family are all using atombios so remove dead code and print an error message if we fail to find a valid atombios. Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-10-02 00:02:15 +08:00
atom_asic_init(rdev->mode_info.atom_context);
drm/radeon/kms: move around new init path code to avoid posting at init We really don't want to post the card at init, it takes a relatively long time and isn't required, so split the resume path into a startup path called by both init/resume and separate resume entry point to do posting. Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-18 13:19:37 +08:00
drm/radeon: introduce a sub allocator and convert ib pool to it v4 Somewhat specializaed sub-allocator designed to perform sub-allocation for command buffer not only for current cs ioctl but for future command submission ioctl as well. Patch also convert current ib pool to use the sub allocator. Idea is that ib poll buffer can be share with other command buffer submission not having 64K granularity. v2 Harmonize pool handling and add suspend/resume callback to pin/unpin sa bo (tested on rv280, rv370, r420, rv515, rv610, rv710, redwood, cayman, rs480, rs690, rs880) v3 Simplify allocator v4 Fix radeon_ib_get error path to properly free fence Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-11-16 00:48:34 +08:00
rdev->accel_working = true;
drm/radeon/kms: move around new init path code to avoid posting at init We really don't want to post the card at init, it takes a relatively long time and isn't required, so split the resume path into a startup path called by both init/resume and separate resume entry point to do posting. Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-18 13:19:37 +08:00
r = r600_startup(rdev);
if (r) {
DRM_ERROR("r600 startup failed on resume\n");
drm/radeon/kms: properly set accel working flag and bailout when false If accel is not working many subsystem such as the ib pool might not be initialized properly that can lead to segfault inside kernel when cs ioctl is call with non working acceleration. To avoid this make sure the accel working flag is false when an error in GPU startup happen and return EBUSY from cs ioctl if accel is not working. Signed-off-by: Jerome Glisse <jglisse@redhat.com> Reviewed-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2012-02-21 06:57:20 +08:00
rdev->accel_working = false;
drm/radeon/kms: move around new init path code to avoid posting at init We really don't want to post the card at init, it takes a relatively long time and isn't required, so split the resume path into a startup path called by both init/resume and separate resume entry point to do posting. Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-18 13:19:37 +08:00
return r;
}
return r;
}
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
int r600_suspend(struct radeon_device *rdev)
{
drm/radeon/kms: suspend and resume audio stuff Fixes FDO bug #26214 Signed-off-by: Rafał Miłecki <zajec5@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-01-29 01:16:30 +08:00
r600_audio_fini(rdev);
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
r600_cp_stop(rdev);
drm/radeon/kms: Add initial support for async DMA on r6xx/r7xx Uses the new multi-ring infrastucture. 6xx/7xx has a single async DMA ring. Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2012-09-28 03:08:35 +08:00
r600_dma_stop(rdev);
drm/radeon/kms: r600/r700 disable irq at suspend To avoid hw doing anythings after we disabled PCIE GART, fully disable IRQ at suspend. Also cleanup a bit the ih structure and process function. Signed-off-by: Jerome Glisse <jglisse@redhat.com> Reviewed-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@linux.ie>
2010-01-15 21:44:37 +08:00
r600_irq_suspend(rdev);
drm/radeon/kms: enable writeback (v2) When writeback is enabled, the GPU shadows writes to certain registers into a buffer in memory. The driver can then read the values from the shadow rather than reading back from the register across the bus. Writeback can be disabled by setting the no_wb module param to 1. On r6xx/r7xx/evergreen, the following registers are shadowed: - CP scratch registers - CP read pointer - IH write pointer On r1xx-rr5xx, the following registers are shadowed: - CP scratch registers - CP read pointer v2: - Combine wb patches for r6xx-evergreen and r1xx-r5xx - Writeback is disabled on AGP boards since it tends to be unreliable on AGP using the gart. - Check radeon_wb_init return values properly. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-08-28 06:25:25 +08:00
radeon_wb_disable(rdev);
drm/radeon/kms: clear confusion in GART init/deinit path GART static one time initialization was mixed up with GART enabling/disabling which could happen several time for instance during suspend/resume cycles. This patch splits all GART handling into 4 differents function. gart_init is for one time initialization, gart_deinit is called upon module unload to free resources allocated by gart_init, gart_enable enable the GART and is intented to be call after first initialization and at each resume cycle or reset cycle. Finaly gart_disable stop the GART and is intended to be call at suspend time or when unloading the module. Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-15 00:29:49 +08:00
r600_pcie_gart_disable(rdev);
drm/radeon/kms: make r600-NI blit suspend code common r600-NI shared the same blit suspend code. Clean it up and make it a shared function. Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Reviewed-by: Ilija Hadzic <ihadzic@research.bell-labs.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-10-14 22:51:22 +08:00
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
return 0;
}
/* Plan is to move initialization in that function and use
* helper function so that radeon_device_init pretty much
* do nothing more than calling asic specific function. This
* should also allow to remove a bunch of callback function
* like vram_info.
*/
int r600_init(struct radeon_device *rdev)
drm/radeon: introduce kernel modesetting for radeon hardware Add kernel modesetting support to radeon driver, use the ttm memory manager to manage memory and DRM/GEM to provide userspace API. In order to avoid backward compatibility issue and to allow clean design and code the radeon kernel modesetting use different code path than old radeon/drm driver. When kernel modesetting is enabled the IOCTL of radeon/drm driver are considered as invalid and an error message is printed in the log and they return failure. KMS enabled userspace will use new API to talk with the radeon/drm driver. The new API provide functions to create/destroy/share/mmap buffer object which are then managed by the kernel memory manager (here TTM). In order to submit command to the GPU the userspace provide a buffer holding the command stream, along this buffer userspace have to provide a list of buffer object used by the command stream. The kernel radeon driver will then place buffer in GPU accessible memory and will update command stream to reflect the position of the different buffers. The kernel will also perform security check on command stream provided by the user, we want to catch and forbid any illegal use of the GPU such as DMA into random system memory or into memory not owned by the process supplying the command stream. This part of the code is still incomplete and this why we propose that patch as a staging driver addition, future security might forbid current experimental userspace to run. This code support the following hardware : R1XX,R2XX,R3XX,R4XX,R5XX (radeon up to X1950). Works is underway to provide support for R6XX, R7XX and newer hardware (radeon from HD2XXX to HD4XXX). Authors: Jerome Glisse <jglisse@redhat.com> Dave Airlie <airlied@redhat.com> Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com> Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-06-05 20:42:42 +08:00
{
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
int r;
drm/radeon: introduce kernel modesetting for radeon hardware Add kernel modesetting support to radeon driver, use the ttm memory manager to manage memory and DRM/GEM to provide userspace API. In order to avoid backward compatibility issue and to allow clean design and code the radeon kernel modesetting use different code path than old radeon/drm driver. When kernel modesetting is enabled the IOCTL of radeon/drm driver are considered as invalid and an error message is printed in the log and they return failure. KMS enabled userspace will use new API to talk with the radeon/drm driver. The new API provide functions to create/destroy/share/mmap buffer object which are then managed by the kernel memory manager (here TTM). In order to submit command to the GPU the userspace provide a buffer holding the command stream, along this buffer userspace have to provide a list of buffer object used by the command stream. The kernel radeon driver will then place buffer in GPU accessible memory and will update command stream to reflect the position of the different buffers. The kernel will also perform security check on command stream provided by the user, we want to catch and forbid any illegal use of the GPU such as DMA into random system memory or into memory not owned by the process supplying the command stream. This part of the code is still incomplete and this why we propose that patch as a staging driver addition, future security might forbid current experimental userspace to run. This code support the following hardware : R1XX,R2XX,R3XX,R4XX,R5XX (radeon up to X1950). Works is underway to provide support for R6XX, R7XX and newer hardware (radeon from HD2XXX to HD4XXX). Authors: Jerome Glisse <jglisse@redhat.com> Dave Airlie <airlied@redhat.com> Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com> Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-06-05 20:42:42 +08:00
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
if (r600_debugfs_mc_info_init(rdev)) {
DRM_ERROR("Failed to register debugfs file for mc !\n");
}
/* Read BIOS */
if (!radeon_get_bios(rdev)) {
if (ASIC_IS_AVIVO(rdev))
return -EINVAL;
}
/* Must be an ATOMBIOS */
drm/radeon/kms: R600/RV770 remove dead code and print message for wrong BIOS R600 & RV770 family are all using atombios so remove dead code and print an error message if we fail to find a valid atombios. Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-10-02 00:02:15 +08:00
if (!rdev->is_atom_bios) {
dev_err(rdev->dev, "Expecting atombios for R600 GPU\n");
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
return -EINVAL;
drm/radeon/kms: R600/RV770 remove dead code and print message for wrong BIOS R600 & RV770 family are all using atombios so remove dead code and print an error message if we fail to find a valid atombios. Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-10-02 00:02:15 +08:00
}
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
r = radeon_atombios_init(rdev);
if (r)
return r;
/* Post card if necessary */
drm/radeon/kms: remove duplicate card_posted() functions Use the common one for all asics. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-01-12 07:08:59 +08:00
if (!radeon_card_posted(rdev)) {
drm/radeon/kms: ignore unposted GPUs with no BIOS. If we find a GPU but we can't find its BIOS and it isn't posted, then ignore it. Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-01 12:06:31 +08:00
if (!rdev->bios) {
dev_err(rdev->dev, "Card not posted and no BIOS - ignoring\n");
return -EINVAL;
}
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
DRM_INFO("GPU not posted. posting now...\n");
atom_asic_init(rdev->mode_info.atom_context);
}
/* Initialize scratch registers */
r600_scratch_init(rdev);
/* Initialize surface registers */
radeon_surface_init(rdev);
drm/radeon/kms: add debugfs for power management for AtomBIOS devices Signed-off-by: Rafał Miłecki <zajec5@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-11-03 07:53:02 +08:00
/* Initialize clocks */
drm/radeon/kms: Move radeon_get_clock_info() call out of radeon_clocks_init(). Someone on IRC reported problems after commit 95a8f1bf4f48b434c9f839ab5a0773f66b39d7c6 ('drm/radeon/kms: Move radeon_clocks_init() call back after getting VRAM info.'). And indeed, at least some ASIC vram_info hooks use the clock info obtained by radeon_get_clock_info(). So, move that call out of radeon_clocks_init(), ahead of the radeon_vram_info() call. [airlied - fixup missing r600/rv770 calls] Signed-off-by: Michel Dänzer <daenzer@vmware.com> Signed-off-by: Dave Airlie <airlied@redhat.com> drm/radeon/kms: fix get clock info calls for r600/rv770 init path. These were missed when it got split out. Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-17 15:42:28 +08:00
radeon_get_clock_info(rdev->ddev);
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
/* Fence driver */
drm/radeon: precompute fence cpu/gpu addr once v3 Add a start fence driver helper function which will be call once for each ring and will compute cpu/gpu addr for fence depending on wether to use wb buffer or scratch reg. This patch replace initialize fence driver separately which was broken in regard of GPU lockup. The fence list for created, emited, signaled must be initialize once and only from the asic init callback not from the startup call back which is call from the gpu reset. v2: With this in place we no longer need to know the number of rings in fence_driver_init, also writing to the scratch reg before knowing its offset is a bad idea. v3: rebase on top of change to previous patch in the serie Signed-off-by: Christian König <deathsimple@vodafone.de> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-11-21 04:45:34 +08:00
r = radeon_fence_driver_init(rdev);
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
if (r)
return r;
drm/radeon/kms: Use radeon_agp_disable when disabling AGP Use same common function to disable agp so we replace the GART callback by the proper one when we do so. This fix oops if radeon_agp_init report failure. This patch also move radeon_agp_init out of *_mc_init for r600 & rv770 so that we can have a similar behavior than for previous hw, ie if agp_init fails it will fallback to GPU GART and disable AGP. Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-01-13 22:16:38 +08:00
if (rdev->flags & RADEON_IS_AGP) {
r = radeon_agp_init(rdev);
if (r)
radeon_agp_disable(rdev);
}
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
r = r600_mc_init(rdev);
drm/radeon/kms: Fallback to non AGP when acceleration fails to initialize (v2) When GPU acceleration is not working with AGP try to fallback to non AGP GART (either PCI or PCIE GART). This should make KMS failure on AGP less painfull. We still need to find out what is wrong when AGP fails but at least user have a lot of more chances to get a working configuration with acceleration. This patch also cleanup R600/RV770 fallback path so they use same code as others asics. Version 2 factorize agp disabling logic to avoid code duplication and bugs. Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-10-07 01:04:29 +08:00
if (r)
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
return r;
/* Memory manager */
drm/radeon/kms: Rework radeon object handling The locking & protection of radeon object was somewhat messy. This patch completely rework it to now use ttm reserve as a protection for the radeon object structure member. It also shrink down the various radeon object structure by removing field which were redondant with the ttm information. Last it converts few simple functions to inline which should with performances. airlied: rebase on top of r600 and other changes. Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-11-20 21:29:23 +08:00
r = radeon_bo_init(rdev);
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
if (r)
return r;
drm/radeon/kms: Add support for interrupts on r6xx/r7xx chips (v3) This enables the use of interrupts on r6xx/r7xx hardware. Interrupts are implemented via a ring buffer. The GPU adds interrupts vectors to the ring and the host reads them off in the interrupt handler. The interrupt controller requires firmware like the CP. This firmware must be installed and accessble to the firmware loader for interrupts to function. MSIs don't seem to work on my RS780. They work fine on all my discrete cards. I'm not sure about other RS780s or RS880s. I've disabled MSIs on RS780 and RS880, but it would probably be worth checking on some other systems. v2 - fix some checkpatch.pl problems; re-read the disp int status reg if we restart the ih; v3 - remove the irq handler if r600_irq_init() fails; remove spinlock in r600_ih_ring_fini(); move ih rb overflow check to r600_get_ih_wptr(); move irq ack to separate function; Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-02 02:43:46 +08:00
r = radeon_irq_kms_init(rdev);
if (r)
return r;
drm/radeon: rename struct radeon_cp to radeon_ring That naming seems to make more sense, since we not only want to run PM4 rings with it. Signed-off-by: Christian König <deathsimple@vodafone.de> Reviewed-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-10-23 18:56:27 +08:00
rdev->ring[RADEON_RING_TYPE_GFX_INDEX].ring_obj = NULL;
r600_ring_init(rdev, &rdev->ring[RADEON_RING_TYPE_GFX_INDEX], 1024 * 1024);
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
drm/radeon/kms: Add initial support for async DMA on r6xx/r7xx Uses the new multi-ring infrastucture. 6xx/7xx has a single async DMA ring. Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2012-09-28 03:08:35 +08:00
rdev->ring[R600_RING_TYPE_DMA_INDEX].ring_obj = NULL;
r600_ring_init(rdev, &rdev->ring[R600_RING_TYPE_DMA_INDEX], 64 * 1024);
drm/radeon/kms: Add support for interrupts on r6xx/r7xx chips (v3) This enables the use of interrupts on r6xx/r7xx hardware. Interrupts are implemented via a ring buffer. The GPU adds interrupts vectors to the ring and the host reads them off in the interrupt handler. The interrupt controller requires firmware like the CP. This firmware must be installed and accessble to the firmware loader for interrupts to function. MSIs don't seem to work on my RS780. They work fine on all my discrete cards. I'm not sure about other RS780s or RS880s. I've disabled MSIs on RS780 and RS880, but it would probably be worth checking on some other systems. v2 - fix some checkpatch.pl problems; re-read the disp int status reg if we restart the ih; v3 - remove the irq handler if r600_irq_init() fails; remove spinlock in r600_ih_ring_fini(); move ih rb overflow check to r600_get_ih_wptr(); move irq ack to separate function; Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-02 02:43:46 +08:00
rdev->ih.ring_obj = NULL;
r600_ih_ring_init(rdev, 64 * 1024);
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
drm/radeon/kms: clear confusion in GART init/deinit path GART static one time initialization was mixed up with GART enabling/disabling which could happen several time for instance during suspend/resume cycles. This patch splits all GART handling into 4 differents function. gart_init is for one time initialization, gart_deinit is called upon module unload to free resources allocated by gart_init, gart_enable enable the GART and is intented to be call after first initialization and at each resume cycle or reset cycle. Finaly gart_disable stop the GART and is intended to be call at suspend time or when unloading the module. Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-15 00:29:49 +08:00
r = r600_pcie_gart_init(rdev);
if (r)
return r;
drm/radeon/kms/r600/r700: fallback gracefully on ucode failure Sent the wrong patch earlier. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-10 08:31:44 +08:00
rdev->accel_working = true;
drm/radeon/kms: move around new init path code to avoid posting at init We really don't want to post the card at init, it takes a relatively long time and isn't required, so split the resume path into a startup path called by both init/resume and separate resume entry point to do posting. Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-18 13:19:37 +08:00
r = r600_startup(rdev);
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
if (r) {
drm/radeon/kms: don't call suspend path before cleaning up GPU In suspend path we unmap the GART table while in cleaning up path we will unbind buffer and thus try to write to unmapped GART leading to oops. In order to avoid this we don't call the suspend path in cleanup path. Cleanup path is clever enough to desactive GPU like the suspend path is doing, thus this was redondant. Tested on: RV370, R420, RV515, RV570, RV610, RV770 (all PCIE) Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-02-02 18:51:45 +08:00
dev_err(rdev->dev, "disabling GPU acceleration\n");
r600_cp_fini(rdev);
drm/radeon/kms: Add initial support for async DMA on r6xx/r7xx Uses the new multi-ring infrastucture. 6xx/7xx has a single async DMA ring. Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2012-09-28 03:08:35 +08:00
r600_dma_fini(rdev);
drm/radeon/kms: don't call suspend path before cleaning up GPU In suspend path we unmap the GART table while in cleaning up path we will unbind buffer and thus try to write to unmapped GART leading to oops. In order to avoid this we don't call the suspend path in cleanup path. Cleanup path is clever enough to desactive GPU like the suspend path is doing, thus this was redondant. Tested on: RV370, R420, RV515, RV570, RV610, RV770 (all PCIE) Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-02-02 18:51:45 +08:00
r600_irq_fini(rdev);
drm/radeon/kms: enable writeback (v2) When writeback is enabled, the GPU shadows writes to certain registers into a buffer in memory. The driver can then read the values from the shadow rather than reading back from the register across the bus. Writeback can be disabled by setting the no_wb module param to 1. On r6xx/r7xx/evergreen, the following registers are shadowed: - CP scratch registers - CP read pointer - IH write pointer On r1xx-rr5xx, the following registers are shadowed: - CP scratch registers - CP read pointer v2: - Combine wb patches for r6xx-evergreen and r1xx-r5xx - Writeback is disabled on AGP boards since it tends to be unreliable on AGP using the gart. - Check radeon_wb_init return values properly. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-08-28 06:25:25 +08:00
radeon_wb_fini(rdev);
drm/radeon: remove ip_pool start/suspend The IB pool is in gart memory, so it is completely superfluous to unpin / repin it on suspend / resume. Signed-off-by: Christian König <deathsimple@vodafone.de> Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
2012-07-05 17:55:34 +08:00
radeon_ib_pool_fini(rdev);
drm/radeon/kms: don't call suspend path before cleaning up GPU In suspend path we unmap the GART table while in cleaning up path we will unbind buffer and thus try to write to unmapped GART leading to oops. In order to avoid this we don't call the suspend path in cleanup path. Cleanup path is clever enough to desactive GPU like the suspend path is doing, thus this was redondant. Tested on: RV370, R420, RV515, RV570, RV610, RV770 (all PCIE) Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-02-02 18:51:45 +08:00
radeon_irq_kms_fini(rdev);
drm/radeon/kms: Fix R600/RV770 disable acceleration path When acceleration doesn't work we should free associated memory and stop GPU block responsible for hardware acceleration so we don't waste resource or let think one component of the driver that a GPU feature is working/running while it doesn't actualy work. Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-10-02 00:02:14 +08:00
r600_pcie_gart_fini(rdev);
drm/radeon/kms: don't fail if we fail to init GPU acceleration Userspace can query if acceleration is working or not true get info ioctl and could fallback to software if for some reason kernel failed to initialize KMS. This should allow to give a working KMS setup in all case (even with non functionning accel). Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-16 21:24:21 +08:00
rdev->accel_working = false;
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
}
drm/radeon/kms: HDMI support for R600 KMS Adding basic HDMI support for R600 KMS, ported from radeonhd ddx. [airlied:- checkpatch cleanups] Signed-off-by: Christian Koenig <deathsimple@vodafone.de> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-10-12 05:49:13 +08:00
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
return 0;
}
void r600_fini(struct radeon_device *rdev)
{
drm/radeon/kms: HDMI support for R600 KMS Adding basic HDMI support for R600 KMS, ported from radeonhd ddx. [airlied:- checkpatch cleanups] Signed-off-by: Christian Koenig <deathsimple@vodafone.de> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-10-12 05:49:13 +08:00
r600_audio_fini(rdev);
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
r600_blit_fini(rdev);
drm/radeon/kms: don't call suspend path before cleaning up GPU In suspend path we unmap the GART table while in cleaning up path we will unbind buffer and thus try to write to unmapped GART leading to oops. In order to avoid this we don't call the suspend path in cleanup path. Cleanup path is clever enough to desactive GPU like the suspend path is doing, thus this was redondant. Tested on: RV370, R420, RV515, RV570, RV610, RV770 (all PCIE) Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-02-02 18:51:45 +08:00
r600_cp_fini(rdev);
drm/radeon/kms: Add initial support for async DMA on r6xx/r7xx Uses the new multi-ring infrastucture. 6xx/7xx has a single async DMA ring. Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2012-09-28 03:08:35 +08:00
r600_dma_fini(rdev);
drm/radeon/kms: Add support for interrupts on r6xx/r7xx chips (v3) This enables the use of interrupts on r6xx/r7xx hardware. Interrupts are implemented via a ring buffer. The GPU adds interrupts vectors to the ring and the host reads them off in the interrupt handler. The interrupt controller requires firmware like the CP. This firmware must be installed and accessble to the firmware loader for interrupts to function. MSIs don't seem to work on my RS780. They work fine on all my discrete cards. I'm not sure about other RS780s or RS880s. I've disabled MSIs on RS780 and RS880, but it would probably be worth checking on some other systems. v2 - fix some checkpatch.pl problems; re-read the disp int status reg if we restart the ih; v3 - remove the irq handler if r600_irq_init() fails; remove spinlock in r600_ih_ring_fini(); move ih rb overflow check to r600_get_ih_wptr(); move irq ack to separate function; Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-02 02:43:46 +08:00
r600_irq_fini(rdev);
drm/radeon/kms: enable writeback (v2) When writeback is enabled, the GPU shadows writes to certain registers into a buffer in memory. The driver can then read the values from the shadow rather than reading back from the register across the bus. Writeback can be disabled by setting the no_wb module param to 1. On r6xx/r7xx/evergreen, the following registers are shadowed: - CP scratch registers - CP read pointer - IH write pointer On r1xx-rr5xx, the following registers are shadowed: - CP scratch registers - CP read pointer v2: - Combine wb patches for r6xx-evergreen and r1xx-r5xx - Writeback is disabled on AGP boards since it tends to be unreliable on AGP using the gart. - Check radeon_wb_init return values properly. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-08-28 06:25:25 +08:00
radeon_wb_fini(rdev);
drm/radeon: remove ip_pool start/suspend The IB pool is in gart memory, so it is completely superfluous to unpin / repin it on suspend / resume. Signed-off-by: Christian König <deathsimple@vodafone.de> Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
2012-07-05 17:55:34 +08:00
radeon_ib_pool_fini(rdev);
drm/radeon/kms: Add support for interrupts on r6xx/r7xx chips (v3) This enables the use of interrupts on r6xx/r7xx hardware. Interrupts are implemented via a ring buffer. The GPU adds interrupts vectors to the ring and the host reads them off in the interrupt handler. The interrupt controller requires firmware like the CP. This firmware must be installed and accessble to the firmware loader for interrupts to function. MSIs don't seem to work on my RS780. They work fine on all my discrete cards. I'm not sure about other RS780s or RS880s. I've disabled MSIs on RS780 and RS880, but it would probably be worth checking on some other systems. v2 - fix some checkpatch.pl problems; re-read the disp int status reg if we restart the ih; v3 - remove the irq handler if r600_irq_init() fails; remove spinlock in r600_ih_ring_fini(); move ih rb overflow check to r600_get_ih_wptr(); move irq ack to separate function; Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-02 02:43:46 +08:00
radeon_irq_kms_fini(rdev);
drm/radeon/kms: clear confusion in GART init/deinit path GART static one time initialization was mixed up with GART enabling/disabling which could happen several time for instance during suspend/resume cycles. This patch splits all GART handling into 4 differents function. gart_init is for one time initialization, gart_deinit is called upon module unload to free resources allocated by gart_init, gart_enable enable the GART and is intented to be call after first initialization and at each resume cycle or reset cycle. Finaly gart_disable stop the GART and is intended to be call at suspend time or when unloading the module. Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-15 00:29:49 +08:00
r600_pcie_gart_fini(rdev);
drm/radeon/kms: allocate vram scratch page on 6xx+ The vram scratch was originally only used on some 7xx asics to work around a hw bug. Allocate the scratch page on all 6xx+ radeons and set the MC_VM_SYSTEM_APERTURE_DEFAULT_ADDR to point to it. We shouldn't ever hit it since we limit the system aperture to vram or vram and AGP, but better safe than sorry. Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-10-28 22:30:02 +08:00
r600_vram_scratch_fini(rdev);
drm/radeon/kms: don't call suspend path before cleaning up GPU In suspend path we unmap the GART table while in cleaning up path we will unbind buffer and thus try to write to unmapped GART leading to oops. In order to avoid this we don't call the suspend path in cleanup path. Cleanup path is clever enough to desactive GPU like the suspend path is doing, thus this was redondant. Tested on: RV370, R420, RV515, RV570, RV610, RV770 (all PCIE) Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-02-02 18:51:45 +08:00
radeon_agp_fini(rdev);
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
radeon_gem_fini(rdev);
radeon_fence_driver_fini(rdev);
drm/radeon/kms: Rework radeon object handling The locking & protection of radeon object was somewhat messy. This patch completely rework it to now use ttm reserve as a protection for the radeon object structure member. It also shrink down the various radeon object structure by removing field which were redondant with the ttm information. Last it converts few simple functions to inline which should with performances. airlied: rebase on top of r600 and other changes. Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-11-20 21:29:23 +08:00
radeon_bo_fini(rdev);
drm/radeon/kms: R600/RV770 remove dead code and print message for wrong BIOS R600 & RV770 family are all using atombios so remove dead code and print an error message if we fail to find a valid atombios. Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-10-02 00:02:15 +08:00
radeon_atombios_fini(rdev);
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
kfree(rdev->bios);
rdev->bios = NULL;
}
/*
* CS stuff
*/
void r600_ring_ib_execute(struct radeon_device *rdev, struct radeon_ib *ib)
{
drm/radeon: remove radeon_fence_create It is completely unnecessary to create fences before they are emitted, so remove it and a bunch of checks if fences are emitted or not. Signed-off-by: Christian König <deathsimple@vodafone.de> Reviewed-by: Jerome Glisse <jglisse@redhat.com>
2012-05-08 20:24:01 +08:00
struct radeon_ring *ring = &rdev->ring[ib->ring];
drm/radeon: update rptr saving logic for memory buffers Add support for using memory buffers rather than scratch registers. Some rings may not be able to write to scratch registers. Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Reviewed-by: Christian König <christian.koenig@amd.com>
2012-07-18 02:02:31 +08:00
u32 next_rptr;
drm/radeon: make all functions work with multiple rings. Give all asic and radeon_ring_* functions a radeon_cp parameter, so they know the ring to work with. Signed-off-by: Christian König <deathsimple@vodafone.de> Reviewed-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-09-23 21:11:23 +08:00
drm/radeon: record what is next valid wptr for each ring v4 Before emitting any indirect buffer, emit the offset of the next valid ring content if any. This allow code that want to resume ring to resume ring right after ib that caused GPU lockup. v2: use scratch registers instead of storing it into memory v3: skip over the surface sync for ni and si as well v4: use SET_CONFIG_REG instead of PACKET0 Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Christian König <deathsimple@vodafone.de> Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
2012-07-06 22:22:55 +08:00
if (ring->rptr_save_reg) {
drm/radeon: update rptr saving logic for memory buffers Add support for using memory buffers rather than scratch registers. Some rings may not be able to write to scratch registers. Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Reviewed-by: Christian König <christian.koenig@amd.com>
2012-07-18 02:02:31 +08:00
next_rptr = ring->wptr + 3 + 4;
drm/radeon: record what is next valid wptr for each ring v4 Before emitting any indirect buffer, emit the offset of the next valid ring content if any. This allow code that want to resume ring to resume ring right after ib that caused GPU lockup. v2: use scratch registers instead of storing it into memory v3: skip over the surface sync for ni and si as well v4: use SET_CONFIG_REG instead of PACKET0 Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Christian König <deathsimple@vodafone.de> Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
2012-07-06 22:22:55 +08:00
radeon_ring_write(ring, PACKET3(PACKET3_SET_CONFIG_REG, 1));
radeon_ring_write(ring, ((ring->rptr_save_reg -
PACKET3_SET_CONFIG_REG_OFFSET) >> 2));
radeon_ring_write(ring, next_rptr);
drm/radeon: update rptr saving logic for memory buffers Add support for using memory buffers rather than scratch registers. Some rings may not be able to write to scratch registers. Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Reviewed-by: Christian König <christian.koenig@amd.com>
2012-07-18 02:02:31 +08:00
} else if (rdev->wb.enabled) {
next_rptr = ring->wptr + 5 + 4;
radeon_ring_write(ring, PACKET3(PACKET3_MEM_WRITE, 3));
radeon_ring_write(ring, ring->next_rptr_gpu_addr & 0xfffffffc);
radeon_ring_write(ring, (upper_32_bits(ring->next_rptr_gpu_addr) & 0xff) | (1 << 18));
radeon_ring_write(ring, next_rptr);
radeon_ring_write(ring, 0);
drm/radeon: record what is next valid wptr for each ring v4 Before emitting any indirect buffer, emit the offset of the next valid ring content if any. This allow code that want to resume ring to resume ring right after ib that caused GPU lockup. v2: use scratch registers instead of storing it into memory v3: skip over the surface sync for ni and si as well v4: use SET_CONFIG_REG instead of PACKET0 Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Christian König <deathsimple@vodafone.de> Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
2012-07-06 22:22:55 +08:00
}
drm/radeon: rename struct radeon_cp to radeon_ring That naming seems to make more sense, since we not only want to run PM4 rings with it. Signed-off-by: Christian König <deathsimple@vodafone.de> Reviewed-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-10-23 18:56:27 +08:00
radeon_ring_write(ring, PACKET3(PACKET3_INDIRECT_BUFFER, 2));
radeon_ring_write(ring,
drm/radeon/kms: 6xx/7xx big endian fixes agd5f: minor cleanups Signed-off-by: Cédric Cano <ccano@interfaceconcept.com> Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-02-12 08:45:38 +08:00
#ifdef __BIG_ENDIAN
(2 << 0) |
#endif
(ib->gpu_addr & 0xFFFFFFFC));
drm/radeon: rename struct radeon_cp to radeon_ring That naming seems to make more sense, since we not only want to run PM4 rings with it. Signed-off-by: Christian König <deathsimple@vodafone.de> Reviewed-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-10-23 18:56:27 +08:00
radeon_ring_write(ring, upper_32_bits(ib->gpu_addr) & 0xFF);
radeon_ring_write(ring, ib->length_dw);
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
}
drm/radeon/kms: make ring_start, ring_test, and ib_test per ring Each ring type may need a different variant. Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Reviewed-by: Christian König<christian.koenig@amd.com> Reviewed-by: Michel Dänzer <michel.daenzer@amd.com> Reviewed-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2012-02-24 06:53:45 +08:00
int r600_ib_test(struct radeon_device *rdev, struct radeon_ring *ring)
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
{
drm/radeon: make the ib an inline object No need to malloc it any more. Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Christian König <deathsimple@vodafone.de> Signed-off-by: Dave Airlie <airlied@redhat.com>
2012-05-09 21:35:02 +08:00
struct radeon_ib ib;
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
uint32_t scratch;
uint32_t tmp = 0;
unsigned i;
int r;
r = radeon_scratch_get(rdev, &scratch);
if (r) {
DRM_ERROR("radeon: failed to get scratch reg (%d).\n", r);
return r;
}
WREG32(scratch, 0xCAFEDEAD);
drm/radeon: cleanup VM id handling a bit Store a reference to the VM into the IB structure, that makes calculating the IBs address a bit less complicated. Signed-off-by: Christian König <deathsimple@vodafone.de> Reviewed-by: Jerome Glisse <jglisse@redhat.com>
2012-08-07 00:57:44 +08:00
r = radeon_ib_get(rdev, ring->idx, &ib, NULL, 256);
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
if (r) {
DRM_ERROR("radeon: failed to get ib (%d).\n", r);
drm/radeon: Fix scratch register leak in IB test. Restructure the code to jump out via labels instead of directly returning early. Also make error reporting consistent across all hardware generations. Signed-off-by: Michel Dänzer <michel.daenzer@amd.com> Reviewed-by: Simon Kitching <skitching@vonos.net> Reviewed-by: Christian König <christian.koenig@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2012-09-20 16:31:10 +08:00
goto free_scratch;
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
}
drm/radeon: make the ib an inline object No need to malloc it any more. Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Christian König <deathsimple@vodafone.de> Signed-off-by: Dave Airlie <airlied@redhat.com>
2012-05-09 21:35:02 +08:00
ib.ptr[0] = PACKET3(PACKET3_SET_CONFIG_REG, 1);
ib.ptr[1] = ((scratch - PACKET3_SET_CONFIG_REG_OFFSET) >> 2);
ib.ptr[2] = 0xDEADBEEF;
ib.length_dw = 3;
drm/radeon: fix const IB handling v2 Const IBs are executed on the CE not the CP, so we can't fence them in the normal way. So submit them directly before the IB instead, just as the documentation says. v2: keep the extra documentation Signed-off-by: Christian König <deathsimple@vodafone.de> Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
2012-07-13 19:06:00 +08:00
r = radeon_ib_schedule(rdev, &ib, NULL);
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
if (r) {
DRM_ERROR("radeon: failed to schedule ib (%d).\n", r);
drm/radeon: Fix scratch register leak in IB test. Restructure the code to jump out via labels instead of directly returning early. Also make error reporting consistent across all hardware generations. Signed-off-by: Michel Dänzer <michel.daenzer@amd.com> Reviewed-by: Simon Kitching <skitching@vonos.net> Reviewed-by: Christian König <christian.koenig@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2012-09-20 16:31:10 +08:00
goto free_ib;
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
}
drm/radeon: make the ib an inline object No need to malloc it any more. Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Christian König <deathsimple@vodafone.de> Signed-off-by: Dave Airlie <airlied@redhat.com>
2012-05-09 21:35:02 +08:00
r = radeon_fence_wait(ib.fence, false);
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
if (r) {
DRM_ERROR("radeon: fence wait failed (%d).\n", r);
drm/radeon: Fix scratch register leak in IB test. Restructure the code to jump out via labels instead of directly returning early. Also make error reporting consistent across all hardware generations. Signed-off-by: Michel Dänzer <michel.daenzer@amd.com> Reviewed-by: Simon Kitching <skitching@vonos.net> Reviewed-by: Christian König <christian.koenig@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2012-09-20 16:31:10 +08:00
goto free_ib;
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
}
for (i = 0; i < rdev->usec_timeout; i++) {
tmp = RREG32(scratch);
if (tmp == 0xDEADBEEF)
break;
DRM_UDELAY(1);
}
if (i < rdev->usec_timeout) {
drm/radeon: make the ib an inline object No need to malloc it any more. Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Christian König <deathsimple@vodafone.de> Signed-off-by: Dave Airlie <airlied@redhat.com>
2012-05-09 21:35:02 +08:00
DRM_INFO("ib test on ring %d succeeded in %u usecs\n", ib.fence->ring, i);
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
} else {
drm: radeon cleanup fixes... Fix string interpreted as trigraph and typo. Signed-off-by: Daniel J Blueman <daniel.blueman@gmail.com> Reviewed-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-09-23 00:57:19 +08:00
DRM_ERROR("radeon: ib test failed (scratch(0x%04X)=0x%08X)\n",
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
scratch, tmp);
r = -EINVAL;
}
drm/radeon: Fix scratch register leak in IB test. Restructure the code to jump out via labels instead of directly returning early. Also make error reporting consistent across all hardware generations. Signed-off-by: Michel Dänzer <michel.daenzer@amd.com> Reviewed-by: Simon Kitching <skitching@vonos.net> Reviewed-by: Christian König <christian.koenig@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2012-09-20 16:31:10 +08:00
free_ib:
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
radeon_ib_free(rdev, &ib);
drm/radeon: Fix scratch register leak in IB test. Restructure the code to jump out via labels instead of directly returning early. Also make error reporting consistent across all hardware generations. Signed-off-by: Michel Dänzer <michel.daenzer@amd.com> Reviewed-by: Simon Kitching <skitching@vonos.net> Reviewed-by: Christian König <christian.koenig@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2012-09-20 16:31:10 +08:00
free_scratch:
radeon_scratch_free(rdev, scratch);
drm/radeon: introduce kernel modesetting for radeon hardware Add kernel modesetting support to radeon driver, use the ttm memory manager to manage memory and DRM/GEM to provide userspace API. In order to avoid backward compatibility issue and to allow clean design and code the radeon kernel modesetting use different code path than old radeon/drm driver. When kernel modesetting is enabled the IOCTL of radeon/drm driver are considered as invalid and an error message is printed in the log and they return failure. KMS enabled userspace will use new API to talk with the radeon/drm driver. The new API provide functions to create/destroy/share/mmap buffer object which are then managed by the kernel memory manager (here TTM). In order to submit command to the GPU the userspace provide a buffer holding the command stream, along this buffer userspace have to provide a list of buffer object used by the command stream. The kernel radeon driver will then place buffer in GPU accessible memory and will update command stream to reflect the position of the different buffers. The kernel will also perform security check on command stream provided by the user, we want to catch and forbid any illegal use of the GPU such as DMA into random system memory or into memory not owned by the process supplying the command stream. This part of the code is still incomplete and this why we propose that patch as a staging driver addition, future security might forbid current experimental userspace to run. This code support the following hardware : R1XX,R2XX,R3XX,R4XX,R5XX (radeon up to X1950). Works is underway to provide support for R6XX, R7XX and newer hardware (radeon from HD2XXX to HD4XXX). Authors: Jerome Glisse <jglisse@redhat.com> Dave Airlie <airlied@redhat.com> Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com> Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-06-05 20:42:42 +08:00
return r;
}
drm/radeon/kms: Add initial support for async DMA on r6xx/r7xx Uses the new multi-ring infrastucture. 6xx/7xx has a single async DMA ring. Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2012-09-28 03:08:35 +08:00
/**
* r600_dma_ib_test - test an IB on the DMA engine
*
* @rdev: radeon_device pointer
* @ring: radeon_ring structure holding ring information
*
* Test a simple IB in the DMA ring (r6xx-SI).
* Returns 0 on success, error on failure.
*/
int r600_dma_ib_test(struct radeon_device *rdev, struct radeon_ring *ring)
{
struct radeon_ib ib;
unsigned i;
int r;
void __iomem *ptr = (void *)rdev->vram_scratch.ptr;
u32 tmp = 0;
if (!ptr) {
DRM_ERROR("invalid vram scratch pointer\n");
return -EINVAL;
}
tmp = 0xCAFEDEAD;
writel(tmp, ptr);
r = radeon_ib_get(rdev, ring->idx, &ib, NULL, 256);
if (r) {
DRM_ERROR("radeon: failed to get ib (%d).\n", r);
return r;
}
ib.ptr[0] = DMA_PACKET(DMA_PACKET_WRITE, 0, 0, 1);
ib.ptr[1] = rdev->vram_scratch.gpu_addr & 0xfffffffc;
ib.ptr[2] = upper_32_bits(rdev->vram_scratch.gpu_addr) & 0xff;
ib.ptr[3] = 0xDEADBEEF;
ib.length_dw = 4;
r = radeon_ib_schedule(rdev, &ib, NULL);
if (r) {
radeon_ib_free(rdev, &ib);
DRM_ERROR("radeon: failed to schedule ib (%d).\n", r);
return r;
}
r = radeon_fence_wait(ib.fence, false);
if (r) {
DRM_ERROR("radeon: fence wait failed (%d).\n", r);
return r;
}
for (i = 0; i < rdev->usec_timeout; i++) {
tmp = readl(ptr);
if (tmp == 0xDEADBEEF)
break;
DRM_UDELAY(1);
}
if (i < rdev->usec_timeout) {
DRM_INFO("ib test on ring %d succeeded in %u usecs\n", ib.fence->ring, i);
} else {
DRM_ERROR("radeon: ib test failed (0x%08X)\n", tmp);
r = -EINVAL;
}
radeon_ib_free(rdev, &ib);
return r;
}
/**
* r600_dma_ring_ib_execute - Schedule an IB on the DMA engine
*
* @rdev: radeon_device pointer
* @ib: IB object to schedule
*
* Schedule an IB in the DMA ring (r6xx-r7xx).
*/
void r600_dma_ring_ib_execute(struct radeon_device *rdev, struct radeon_ib *ib)
{
struct radeon_ring *ring = &rdev->ring[ib->ring];
if (rdev->wb.enabled) {
u32 next_rptr = ring->wptr + 4;
while ((next_rptr & 7) != 5)
next_rptr++;
next_rptr += 3;
radeon_ring_write(ring, DMA_PACKET(DMA_PACKET_WRITE, 0, 0, 1));
radeon_ring_write(ring, ring->next_rptr_gpu_addr & 0xfffffffc);
radeon_ring_write(ring, upper_32_bits(ring->next_rptr_gpu_addr) & 0xff);
radeon_ring_write(ring, next_rptr);
}
/* The indirect buffer packet must end on an 8 DW boundary in the DMA ring.
* Pad as necessary with NOPs.
*/
while ((ring->wptr & 7) != 5)
radeon_ring_write(ring, DMA_PACKET(DMA_PACKET_NOP, 0, 0, 0));
radeon_ring_write(ring, DMA_PACKET(DMA_PACKET_INDIRECT_BUFFER, 0, 0, 0));
radeon_ring_write(ring, (ib->gpu_addr & 0xFFFFFFE0));
radeon_ring_write(ring, (ib->length_dw << 16) | (upper_32_bits(ib->gpu_addr) & 0xFF));
}
drm/radeon/kms: Add support for interrupts on r6xx/r7xx chips (v3) This enables the use of interrupts on r6xx/r7xx hardware. Interrupts are implemented via a ring buffer. The GPU adds interrupts vectors to the ring and the host reads them off in the interrupt handler. The interrupt controller requires firmware like the CP. This firmware must be installed and accessble to the firmware loader for interrupts to function. MSIs don't seem to work on my RS780. They work fine on all my discrete cards. I'm not sure about other RS780s or RS880s. I've disabled MSIs on RS780 and RS880, but it would probably be worth checking on some other systems. v2 - fix some checkpatch.pl problems; re-read the disp int status reg if we restart the ih; v3 - remove the irq handler if r600_irq_init() fails; remove spinlock in r600_ih_ring_fini(); move ih rb overflow check to r600_get_ih_wptr(); move irq ack to separate function; Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-02 02:43:46 +08:00
/*
* Interrupts
*
* Interrupts use a ring buffer on r6xx/r7xx hardware. It works pretty
* the same as the CP ring buffer, but in reverse. Rather than the CPU
* writing to the ring and the GPU consuming, the GPU writes to the ring
* and host consumes. As the host irq handler processes interrupts, it
* increments the rptr. When the rptr catches up with the wptr, all the
* current interrupts have been processed.
*/
void r600_ih_ring_init(struct radeon_device *rdev, unsigned ring_size)
{
u32 rb_bufsz;
/* Align ring size */
rb_bufsz = drm_order(ring_size / 4);
ring_size = (1 << rb_bufsz) * 4;
rdev->ih.ring_size = ring_size;
drm/radeon/kms: r600/r700 disable irq at suspend To avoid hw doing anythings after we disabled PCIE GART, fully disable IRQ at suspend. Also cleanup a bit the ih structure and process function. Signed-off-by: Jerome Glisse <jglisse@redhat.com> Reviewed-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@linux.ie>
2010-01-15 21:44:37 +08:00
rdev->ih.ptr_mask = rdev->ih.ring_size - 1;
rdev->ih.rptr = 0;
drm/radeon/kms: Add support for interrupts on r6xx/r7xx chips (v3) This enables the use of interrupts on r6xx/r7xx hardware. Interrupts are implemented via a ring buffer. The GPU adds interrupts vectors to the ring and the host reads them off in the interrupt handler. The interrupt controller requires firmware like the CP. This firmware must be installed and accessble to the firmware loader for interrupts to function. MSIs don't seem to work on my RS780. They work fine on all my discrete cards. I'm not sure about other RS780s or RS880s. I've disabled MSIs on RS780 and RS880, but it would probably be worth checking on some other systems. v2 - fix some checkpatch.pl problems; re-read the disp int status reg if we restart the ih; v3 - remove the irq handler if r600_irq_init() fails; remove spinlock in r600_ih_ring_fini(); move ih rb overflow check to r600_get_ih_wptr(); move irq ack to separate function; Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-02 02:43:46 +08:00
}
drm/radeon/kms: add support for interrupts on SI This is mostly identical to evergreen/ni, however there are some additional fields in the IV vector for RINGID and VMID. Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2012-03-21 05:18:22 +08:00
int r600_ih_ring_alloc(struct radeon_device *rdev)
drm/radeon/kms: Add support for interrupts on r6xx/r7xx chips (v3) This enables the use of interrupts on r6xx/r7xx hardware. Interrupts are implemented via a ring buffer. The GPU adds interrupts vectors to the ring and the host reads them off in the interrupt handler. The interrupt controller requires firmware like the CP. This firmware must be installed and accessble to the firmware loader for interrupts to function. MSIs don't seem to work on my RS780. They work fine on all my discrete cards. I'm not sure about other RS780s or RS880s. I've disabled MSIs on RS780 and RS880, but it would probably be worth checking on some other systems. v2 - fix some checkpatch.pl problems; re-read the disp int status reg if we restart the ih; v3 - remove the irq handler if r600_irq_init() fails; remove spinlock in r600_ih_ring_fini(); move ih rb overflow check to r600_get_ih_wptr(); move irq ack to separate function; Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-02 02:43:46 +08:00
{
int r;
/* Allocate ring buffer */
if (rdev->ih.ring_obj == NULL) {
drm/radeon: embed struct drm_gem_object Unconditionally initialize the drm gem object - it's not worth the trouble not to for the few kernel objects. This patch only changes the place of the drm gem object, access is still done via pointers. v2: Uncoditionally align the size in radeon_bo_create. At least the r600/evergreen blit code didn't to this, angering the paranoid gem code. Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-02-19 00:59:16 +08:00
r = radeon_bo_create(rdev, rdev->ih.ring_size,
drm/radeon/kms: fix alignment when allocating buffers We were previously dropping alignment requests on the floor when allocating buffers so we always ended up page aligned. Certain tiling modes on 6xx+ require larger alignment which wasn't happening before. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Cc: Jerome Glisse <j.glisse@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-11-18 08:00:26 +08:00
PAGE_SIZE, true,
drm/radeon/kms: Rework radeon object handling The locking & protection of radeon object was somewhat messy. This patch completely rework it to now use ttm reserve as a protection for the radeon object structure member. It also shrink down the various radeon object structure by removing field which were redondant with the ttm information. Last it converts few simple functions to inline which should with performances. airlied: rebase on top of r600 and other changes. Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-11-20 21:29:23 +08:00
RADEON_GEM_DOMAIN_GTT,
drm/radeon: add PRIME support (v2) This adds prime->fd and fd->prime support to radeon. It passes the sg object to ttm and then populates the gart entries using it. Compile tested only. v2: stub kmap + use new helpers + add reimporting Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Reviewed-by: Jerome Glisse <jglisse@redhat.com> Reviewed-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2012-05-11 06:33:13 +08:00
NULL, &rdev->ih.ring_obj);
drm/radeon/kms: Add support for interrupts on r6xx/r7xx chips (v3) This enables the use of interrupts on r6xx/r7xx hardware. Interrupts are implemented via a ring buffer. The GPU adds interrupts vectors to the ring and the host reads them off in the interrupt handler. The interrupt controller requires firmware like the CP. This firmware must be installed and accessble to the firmware loader for interrupts to function. MSIs don't seem to work on my RS780. They work fine on all my discrete cards. I'm not sure about other RS780s or RS880s. I've disabled MSIs on RS780 and RS880, but it would probably be worth checking on some other systems. v2 - fix some checkpatch.pl problems; re-read the disp int status reg if we restart the ih; v3 - remove the irq handler if r600_irq_init() fails; remove spinlock in r600_ih_ring_fini(); move ih rb overflow check to r600_get_ih_wptr(); move irq ack to separate function; Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-02 02:43:46 +08:00
if (r) {
DRM_ERROR("radeon: failed to create ih ring buffer (%d).\n", r);
return r;
}
drm/radeon/kms: Rework radeon object handling The locking & protection of radeon object was somewhat messy. This patch completely rework it to now use ttm reserve as a protection for the radeon object structure member. It also shrink down the various radeon object structure by removing field which were redondant with the ttm information. Last it converts few simple functions to inline which should with performances. airlied: rebase on top of r600 and other changes. Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-11-20 21:29:23 +08:00
r = radeon_bo_reserve(rdev->ih.ring_obj, false);
if (unlikely(r != 0))
return r;
r = radeon_bo_pin(rdev->ih.ring_obj,
RADEON_GEM_DOMAIN_GTT,
&rdev->ih.gpu_addr);
drm/radeon/kms: Add support for interrupts on r6xx/r7xx chips (v3) This enables the use of interrupts on r6xx/r7xx hardware. Interrupts are implemented via a ring buffer. The GPU adds interrupts vectors to the ring and the host reads them off in the interrupt handler. The interrupt controller requires firmware like the CP. This firmware must be installed and accessble to the firmware loader for interrupts to function. MSIs don't seem to work on my RS780. They work fine on all my discrete cards. I'm not sure about other RS780s or RS880s. I've disabled MSIs on RS780 and RS880, but it would probably be worth checking on some other systems. v2 - fix some checkpatch.pl problems; re-read the disp int status reg if we restart the ih; v3 - remove the irq handler if r600_irq_init() fails; remove spinlock in r600_ih_ring_fini(); move ih rb overflow check to r600_get_ih_wptr(); move irq ack to separate function; Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-02 02:43:46 +08:00
if (r) {
drm/radeon/kms: Rework radeon object handling The locking & protection of radeon object was somewhat messy. This patch completely rework it to now use ttm reserve as a protection for the radeon object structure member. It also shrink down the various radeon object structure by removing field which were redondant with the ttm information. Last it converts few simple functions to inline which should with performances. airlied: rebase on top of r600 and other changes. Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-11-20 21:29:23 +08:00
radeon_bo_unreserve(rdev->ih.ring_obj);
drm/radeon/kms: Add support for interrupts on r6xx/r7xx chips (v3) This enables the use of interrupts on r6xx/r7xx hardware. Interrupts are implemented via a ring buffer. The GPU adds interrupts vectors to the ring and the host reads them off in the interrupt handler. The interrupt controller requires firmware like the CP. This firmware must be installed and accessble to the firmware loader for interrupts to function. MSIs don't seem to work on my RS780. They work fine on all my discrete cards. I'm not sure about other RS780s or RS880s. I've disabled MSIs on RS780 and RS880, but it would probably be worth checking on some other systems. v2 - fix some checkpatch.pl problems; re-read the disp int status reg if we restart the ih; v3 - remove the irq handler if r600_irq_init() fails; remove spinlock in r600_ih_ring_fini(); move ih rb overflow check to r600_get_ih_wptr(); move irq ack to separate function; Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-02 02:43:46 +08:00
DRM_ERROR("radeon: failed to pin ih ring buffer (%d).\n", r);
return r;
}
drm/radeon/kms: Rework radeon object handling The locking & protection of radeon object was somewhat messy. This patch completely rework it to now use ttm reserve as a protection for the radeon object structure member. It also shrink down the various radeon object structure by removing field which were redondant with the ttm information. Last it converts few simple functions to inline which should with performances. airlied: rebase on top of r600 and other changes. Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-11-20 21:29:23 +08:00
r = radeon_bo_kmap(rdev->ih.ring_obj,
(void **)&rdev->ih.ring);
radeon_bo_unreserve(rdev->ih.ring_obj);
drm/radeon/kms: Add support for interrupts on r6xx/r7xx chips (v3) This enables the use of interrupts on r6xx/r7xx hardware. Interrupts are implemented via a ring buffer. The GPU adds interrupts vectors to the ring and the host reads them off in the interrupt handler. The interrupt controller requires firmware like the CP. This firmware must be installed and accessble to the firmware loader for interrupts to function. MSIs don't seem to work on my RS780. They work fine on all my discrete cards. I'm not sure about other RS780s or RS880s. I've disabled MSIs on RS780 and RS880, but it would probably be worth checking on some other systems. v2 - fix some checkpatch.pl problems; re-read the disp int status reg if we restart the ih; v3 - remove the irq handler if r600_irq_init() fails; remove spinlock in r600_ih_ring_fini(); move ih rb overflow check to r600_get_ih_wptr(); move irq ack to separate function; Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-02 02:43:46 +08:00
if (r) {
DRM_ERROR("radeon: failed to map ih ring buffer (%d).\n", r);
return r;
}
}
return 0;
}
drm/radeon/kms: add support for interrupts on SI This is mostly identical to evergreen/ni, however there are some additional fields in the IV vector for RINGID and VMID. Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2012-03-21 05:18:22 +08:00
void r600_ih_ring_fini(struct radeon_device *rdev)
drm/radeon/kms: Add support for interrupts on r6xx/r7xx chips (v3) This enables the use of interrupts on r6xx/r7xx hardware. Interrupts are implemented via a ring buffer. The GPU adds interrupts vectors to the ring and the host reads them off in the interrupt handler. The interrupt controller requires firmware like the CP. This firmware must be installed and accessble to the firmware loader for interrupts to function. MSIs don't seem to work on my RS780. They work fine on all my discrete cards. I'm not sure about other RS780s or RS880s. I've disabled MSIs on RS780 and RS880, but it would probably be worth checking on some other systems. v2 - fix some checkpatch.pl problems; re-read the disp int status reg if we restart the ih; v3 - remove the irq handler if r600_irq_init() fails; remove spinlock in r600_ih_ring_fini(); move ih rb overflow check to r600_get_ih_wptr(); move irq ack to separate function; Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-02 02:43:46 +08:00
{
drm/radeon/kms: Rework radeon object handling The locking & protection of radeon object was somewhat messy. This patch completely rework it to now use ttm reserve as a protection for the radeon object structure member. It also shrink down the various radeon object structure by removing field which were redondant with the ttm information. Last it converts few simple functions to inline which should with performances. airlied: rebase on top of r600 and other changes. Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-11-20 21:29:23 +08:00
int r;
drm/radeon/kms: Add support for interrupts on r6xx/r7xx chips (v3) This enables the use of interrupts on r6xx/r7xx hardware. Interrupts are implemented via a ring buffer. The GPU adds interrupts vectors to the ring and the host reads them off in the interrupt handler. The interrupt controller requires firmware like the CP. This firmware must be installed and accessble to the firmware loader for interrupts to function. MSIs don't seem to work on my RS780. They work fine on all my discrete cards. I'm not sure about other RS780s or RS880s. I've disabled MSIs on RS780 and RS880, but it would probably be worth checking on some other systems. v2 - fix some checkpatch.pl problems; re-read the disp int status reg if we restart the ih; v3 - remove the irq handler if r600_irq_init() fails; remove spinlock in r600_ih_ring_fini(); move ih rb overflow check to r600_get_ih_wptr(); move irq ack to separate function; Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-02 02:43:46 +08:00
if (rdev->ih.ring_obj) {
drm/radeon/kms: Rework radeon object handling The locking & protection of radeon object was somewhat messy. This patch completely rework it to now use ttm reserve as a protection for the radeon object structure member. It also shrink down the various radeon object structure by removing field which were redondant with the ttm information. Last it converts few simple functions to inline which should with performances. airlied: rebase on top of r600 and other changes. Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-11-20 21:29:23 +08:00
r = radeon_bo_reserve(rdev->ih.ring_obj, false);
if (likely(r == 0)) {
radeon_bo_kunmap(rdev->ih.ring_obj);
radeon_bo_unpin(rdev->ih.ring_obj);
radeon_bo_unreserve(rdev->ih.ring_obj);
}
radeon_bo_unref(&rdev->ih.ring_obj);
drm/radeon/kms: Add support for interrupts on r6xx/r7xx chips (v3) This enables the use of interrupts on r6xx/r7xx hardware. Interrupts are implemented via a ring buffer. The GPU adds interrupts vectors to the ring and the host reads them off in the interrupt handler. The interrupt controller requires firmware like the CP. This firmware must be installed and accessble to the firmware loader for interrupts to function. MSIs don't seem to work on my RS780. They work fine on all my discrete cards. I'm not sure about other RS780s or RS880s. I've disabled MSIs on RS780 and RS880, but it would probably be worth checking on some other systems. v2 - fix some checkpatch.pl problems; re-read the disp int status reg if we restart the ih; v3 - remove the irq handler if r600_irq_init() fails; remove spinlock in r600_ih_ring_fini(); move ih rb overflow check to r600_get_ih_wptr(); move irq ack to separate function; Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-02 02:43:46 +08:00
rdev->ih.ring = NULL;
rdev->ih.ring_obj = NULL;
}
}
drm/radeon/kms/evergreen: implement irq support Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-03-25 01:55:51 +08:00
void r600_rlc_stop(struct radeon_device *rdev)
drm/radeon/kms: Add support for interrupts on r6xx/r7xx chips (v3) This enables the use of interrupts on r6xx/r7xx hardware. Interrupts are implemented via a ring buffer. The GPU adds interrupts vectors to the ring and the host reads them off in the interrupt handler. The interrupt controller requires firmware like the CP. This firmware must be installed and accessble to the firmware loader for interrupts to function. MSIs don't seem to work on my RS780. They work fine on all my discrete cards. I'm not sure about other RS780s or RS880s. I've disabled MSIs on RS780 and RS880, but it would probably be worth checking on some other systems. v2 - fix some checkpatch.pl problems; re-read the disp int status reg if we restart the ih; v3 - remove the irq handler if r600_irq_init() fails; remove spinlock in r600_ih_ring_fini(); move ih rb overflow check to r600_get_ih_wptr(); move irq ack to separate function; Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-02 02:43:46 +08:00
{
drm/radeon/kms/evergreen: implement irq support Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-03-25 01:55:51 +08:00
if ((rdev->family >= CHIP_RV770) &&
(rdev->family <= CHIP_RV740)) {
drm/radeon/kms: Add support for interrupts on r6xx/r7xx chips (v3) This enables the use of interrupts on r6xx/r7xx hardware. Interrupts are implemented via a ring buffer. The GPU adds interrupts vectors to the ring and the host reads them off in the interrupt handler. The interrupt controller requires firmware like the CP. This firmware must be installed and accessble to the firmware loader for interrupts to function. MSIs don't seem to work on my RS780. They work fine on all my discrete cards. I'm not sure about other RS780s or RS880s. I've disabled MSIs on RS780 and RS880, but it would probably be worth checking on some other systems. v2 - fix some checkpatch.pl problems; re-read the disp int status reg if we restart the ih; v3 - remove the irq handler if r600_irq_init() fails; remove spinlock in r600_ih_ring_fini(); move ih rb overflow check to r600_get_ih_wptr(); move irq ack to separate function; Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-02 02:43:46 +08:00
/* r7xx asics need to soft reset RLC before halting */
WREG32(SRBM_SOFT_RESET, SOFT_RESET_RLC);
RREG32(SRBM_SOFT_RESET);
drm/radeon: replace udelay with mdelay for long timeouts Some architectures require that delays longer than a few miliseconds are called through mdelay. This was triggered on ARM randconfig builds. Signed-off-by: Arnd Bergmann <arnd@arndb.de> Signed-off-by: Mathieu Poirier <mathieu.poirier@linaro.org> Signed-off-by: Dave Airlie <airlied@redhat.com>
2012-04-06 02:58:22 +08:00
mdelay(15);
drm/radeon/kms: Add support for interrupts on r6xx/r7xx chips (v3) This enables the use of interrupts on r6xx/r7xx hardware. Interrupts are implemented via a ring buffer. The GPU adds interrupts vectors to the ring and the host reads them off in the interrupt handler. The interrupt controller requires firmware like the CP. This firmware must be installed and accessble to the firmware loader for interrupts to function. MSIs don't seem to work on my RS780. They work fine on all my discrete cards. I'm not sure about other RS780s or RS880s. I've disabled MSIs on RS780 and RS880, but it would probably be worth checking on some other systems. v2 - fix some checkpatch.pl problems; re-read the disp int status reg if we restart the ih; v3 - remove the irq handler if r600_irq_init() fails; remove spinlock in r600_ih_ring_fini(); move ih rb overflow check to r600_get_ih_wptr(); move irq ack to separate function; Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-02 02:43:46 +08:00
WREG32(SRBM_SOFT_RESET, 0);
RREG32(SRBM_SOFT_RESET);
}
WREG32(RLC_CNTL, 0);
}
static void r600_rlc_start(struct radeon_device *rdev)
{
WREG32(RLC_CNTL, RLC_ENABLE);
}
static int r600_rlc_init(struct radeon_device *rdev)
{
u32 i;
const __be32 *fw_data;
if (!rdev->rlc_fw)
return -EINVAL;
r600_rlc_stop(rdev);
WREG32(RLC_HB_CNTL, 0);
drm/radeon/kms: add support for ucode loading on trinity (v2) v2: fix check for MC ucode from Tom. Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2012-03-21 05:18:39 +08:00
if (rdev->family == CHIP_ARUBA) {
WREG32(TN_RLC_SAVE_AND_RESTORE_BASE, rdev->rlc.save_restore_gpu_addr >> 8);
WREG32(TN_RLC_CLEAR_STATE_RESTORE_BASE, rdev->rlc.clear_state_gpu_addr >> 8);
}
if (rdev->family <= CHIP_CAYMAN) {
WREG32(RLC_HB_BASE, 0);
WREG32(RLC_HB_RPTR, 0);
WREG32(RLC_HB_WPTR, 0);
}
drm/radeon/kms: add support for cayman irqs Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-03-03 09:07:32 +08:00
if (rdev->family <= CHIP_CAICOS) {
WREG32(RLC_HB_WPTR_LSB_ADDR, 0);
WREG32(RLC_HB_WPTR_MSB_ADDR, 0);
}
drm/radeon/kms: Add support for interrupts on r6xx/r7xx chips (v3) This enables the use of interrupts on r6xx/r7xx hardware. Interrupts are implemented via a ring buffer. The GPU adds interrupts vectors to the ring and the host reads them off in the interrupt handler. The interrupt controller requires firmware like the CP. This firmware must be installed and accessble to the firmware loader for interrupts to function. MSIs don't seem to work on my RS780. They work fine on all my discrete cards. I'm not sure about other RS780s or RS880s. I've disabled MSIs on RS780 and RS880, but it would probably be worth checking on some other systems. v2 - fix some checkpatch.pl problems; re-read the disp int status reg if we restart the ih; v3 - remove the irq handler if r600_irq_init() fails; remove spinlock in r600_ih_ring_fini(); move ih rb overflow check to r600_get_ih_wptr(); move irq ack to separate function; Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-02 02:43:46 +08:00
WREG32(RLC_MC_CNTL, 0);
WREG32(RLC_UCODE_CNTL, 0);
fw_data = (const __be32 *)rdev->rlc_fw->data;
drm/radeon/kms: add support for ucode loading on trinity (v2) v2: fix check for MC ucode from Tom. Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2012-03-21 05:18:39 +08:00
if (rdev->family >= CHIP_ARUBA) {
for (i = 0; i < ARUBA_RLC_UCODE_SIZE; i++) {
WREG32(RLC_UCODE_ADDR, i);
WREG32(RLC_UCODE_DATA, be32_to_cpup(fw_data++));
}
} else if (rdev->family >= CHIP_CAYMAN) {
drm/radeon/kms: add support for cayman irqs Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-03-03 09:07:32 +08:00
for (i = 0; i < CAYMAN_RLC_UCODE_SIZE; i++) {
WREG32(RLC_UCODE_ADDR, i);
WREG32(RLC_UCODE_DATA, be32_to_cpup(fw_data++));
}
} else if (rdev->family >= CHIP_CEDAR) {
drm/radeon/kms/evergreen: implement irq support Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-03-25 01:55:51 +08:00
for (i = 0; i < EVERGREEN_RLC_UCODE_SIZE; i++) {
WREG32(RLC_UCODE_ADDR, i);
WREG32(RLC_UCODE_DATA, be32_to_cpup(fw_data++));
}
} else if (rdev->family >= CHIP_RV770) {
drm/radeon/kms: Add support for interrupts on r6xx/r7xx chips (v3) This enables the use of interrupts on r6xx/r7xx hardware. Interrupts are implemented via a ring buffer. The GPU adds interrupts vectors to the ring and the host reads them off in the interrupt handler. The interrupt controller requires firmware like the CP. This firmware must be installed and accessble to the firmware loader for interrupts to function. MSIs don't seem to work on my RS780. They work fine on all my discrete cards. I'm not sure about other RS780s or RS880s. I've disabled MSIs on RS780 and RS880, but it would probably be worth checking on some other systems. v2 - fix some checkpatch.pl problems; re-read the disp int status reg if we restart the ih; v3 - remove the irq handler if r600_irq_init() fails; remove spinlock in r600_ih_ring_fini(); move ih rb overflow check to r600_get_ih_wptr(); move irq ack to separate function; Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-02 02:43:46 +08:00
for (i = 0; i < R700_RLC_UCODE_SIZE; i++) {
WREG32(RLC_UCODE_ADDR, i);
WREG32(RLC_UCODE_DATA, be32_to_cpup(fw_data++));
}
} else {
for (i = 0; i < RLC_UCODE_SIZE; i++) {
WREG32(RLC_UCODE_ADDR, i);
WREG32(RLC_UCODE_DATA, be32_to_cpup(fw_data++));
}
}
WREG32(RLC_UCODE_ADDR, 0);
r600_rlc_start(rdev);
return 0;
}
static void r600_enable_interrupts(struct radeon_device *rdev)
{
u32 ih_cntl = RREG32(IH_CNTL);
u32 ih_rb_cntl = RREG32(IH_RB_CNTL);
ih_cntl |= ENABLE_INTR;
ih_rb_cntl |= IH_RB_ENABLE;
WREG32(IH_CNTL, ih_cntl);
WREG32(IH_RB_CNTL, ih_rb_cntl);
rdev->ih.enabled = true;
}
drm/radeon/kms/evergreen: implement irq support Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-03-25 01:55:51 +08:00
void r600_disable_interrupts(struct radeon_device *rdev)
drm/radeon/kms: Add support for interrupts on r6xx/r7xx chips (v3) This enables the use of interrupts on r6xx/r7xx hardware. Interrupts are implemented via a ring buffer. The GPU adds interrupts vectors to the ring and the host reads them off in the interrupt handler. The interrupt controller requires firmware like the CP. This firmware must be installed and accessble to the firmware loader for interrupts to function. MSIs don't seem to work on my RS780. They work fine on all my discrete cards. I'm not sure about other RS780s or RS880s. I've disabled MSIs on RS780 and RS880, but it would probably be worth checking on some other systems. v2 - fix some checkpatch.pl problems; re-read the disp int status reg if we restart the ih; v3 - remove the irq handler if r600_irq_init() fails; remove spinlock in r600_ih_ring_fini(); move ih rb overflow check to r600_get_ih_wptr(); move irq ack to separate function; Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-02 02:43:46 +08:00
{
u32 ih_rb_cntl = RREG32(IH_RB_CNTL);
u32 ih_cntl = RREG32(IH_CNTL);
ih_rb_cntl &= ~IH_RB_ENABLE;
ih_cntl &= ~ENABLE_INTR;
WREG32(IH_RB_CNTL, ih_rb_cntl);
WREG32(IH_CNTL, ih_cntl);
/* set rptr, wptr to 0 */
WREG32(IH_RB_RPTR, 0);
WREG32(IH_RB_WPTR, 0);
rdev->ih.enabled = false;
rdev->ih.rptr = 0;
}
drm/radeon/kms: add hpd support for r6xx/r7xx/rs780/rs880 asics This just adds the functionality, it's not hooked up yet. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-05 04:12:21 +08:00
static void r600_disable_interrupt_state(struct radeon_device *rdev)
{
u32 tmp;
drm/radeon/kms: leave certain CP int bits enabled These bits are used for internal communication and should be left enabled. This may fix s/r issues on some systems. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-10-09 00:09:12 +08:00
WREG32(CP_INT_CNTL, CNTX_BUSY_INT_ENABLE | CNTX_EMPTY_INT_ENABLE);
drm/radeon/kms: Add initial support for async DMA on r6xx/r7xx Uses the new multi-ring infrastucture. 6xx/7xx has a single async DMA ring. Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2012-09-28 03:08:35 +08:00
tmp = RREG32(DMA_CNTL) & ~TRAP_ENABLE;
WREG32(DMA_CNTL, tmp);
drm/radeon/kms: add hpd support for r6xx/r7xx/rs780/rs880 asics This just adds the functionality, it's not hooked up yet. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-05 04:12:21 +08:00
WREG32(GRBM_INT_CNTL, 0);
WREG32(DxMODE_INT_MASK, 0);
drm/radeon/kms: add pageflip ioctl support (v3) This adds support for dri2 pageflipping. v2: precision updates from Mario Kleiner. v3: Multihead fixes from Mario Kleiner; missing crtc offset add note about update pending bit on pre-avivo chips Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Mario Kleiner <mario.kleiner@tuebingen.mpg.de> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-11-21 23:59:01 +08:00
WREG32(D1GRPH_INTERRUPT_CONTROL, 0);
WREG32(D2GRPH_INTERRUPT_CONTROL, 0);
drm/radeon/kms: add hpd support for r6xx/r7xx/rs780/rs880 asics This just adds the functionality, it's not hooked up yet. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-05 04:12:21 +08:00
if (ASIC_IS_DCE3(rdev)) {
WREG32(DCE3_DACA_AUTODETECT_INT_CONTROL, 0);
WREG32(DCE3_DACB_AUTODETECT_INT_CONTROL, 0);
tmp = RREG32(DC_HPD1_INT_CONTROL) & DC_HPDx_INT_POLARITY;
WREG32(DC_HPD1_INT_CONTROL, tmp);
tmp = RREG32(DC_HPD2_INT_CONTROL) & DC_HPDx_INT_POLARITY;
WREG32(DC_HPD2_INT_CONTROL, tmp);
tmp = RREG32(DC_HPD3_INT_CONTROL) & DC_HPDx_INT_POLARITY;
WREG32(DC_HPD3_INT_CONTROL, tmp);
tmp = RREG32(DC_HPD4_INT_CONTROL) & DC_HPDx_INT_POLARITY;
WREG32(DC_HPD4_INT_CONTROL, tmp);
if (ASIC_IS_DCE32(rdev)) {
tmp = RREG32(DC_HPD5_INT_CONTROL) & DC_HPDx_INT_POLARITY;
drm/radeon/kms: fix some typos in r6xx/r7xx hpd setup Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-03-25 01:57:29 +08:00
WREG32(DC_HPD5_INT_CONTROL, tmp);
drm/radeon/kms: add hpd support for r6xx/r7xx/rs780/rs880 asics This just adds the functionality, it's not hooked up yet. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-05 04:12:21 +08:00
tmp = RREG32(DC_HPD6_INT_CONTROL) & DC_HPDx_INT_POLARITY;
drm/radeon/kms: fix some typos in r6xx/r7xx hpd setup Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-03-25 01:57:29 +08:00
WREG32(DC_HPD6_INT_CONTROL, tmp);
drm/radeon/kms/hdmi: use relative offsets, official regs Signed-off-by: Rafał Miłecki <zajec5@gmail.com> Tested-by: Christian König <deathsimple@vodafone.de> Reviewed-by: Christian König <deathsimple@vodafone.de> Signed-off-by: Dave Airlie <airlied@redhat.com>
2012-04-29 05:35:24 +08:00
tmp = RREG32(AFMT_AUDIO_PACKET_CONTROL + DCE3_HDMI_OFFSET0) & ~HDMI0_AZ_FORMAT_WTRIG_MASK;
WREG32(AFMT_AUDIO_PACKET_CONTROL + DCE3_HDMI_OFFSET0, tmp);
tmp = RREG32(AFMT_AUDIO_PACKET_CONTROL + DCE3_HDMI_OFFSET1) & ~HDMI0_AZ_FORMAT_WTRIG_MASK;
WREG32(AFMT_AUDIO_PACKET_CONTROL + DCE3_HDMI_OFFSET1, tmp);
drm/radeon/kms: fix up audio interrupt handling - add support for rs6xx - add support for DCE4/5 - fixup 6xx/7xx Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2012-03-30 20:59:57 +08:00
} else {
tmp = RREG32(HDMI0_AUDIO_PACKET_CONTROL) & ~HDMI0_AZ_FORMAT_WTRIG_MASK;
WREG32(HDMI0_AUDIO_PACKET_CONTROL, tmp);
tmp = RREG32(DCE3_HDMI1_AUDIO_PACKET_CONTROL) & ~HDMI0_AZ_FORMAT_WTRIG_MASK;
WREG32(DCE3_HDMI1_AUDIO_PACKET_CONTROL, tmp);
drm/radeon/kms: add hpd support for r6xx/r7xx/rs780/rs880 asics This just adds the functionality, it's not hooked up yet. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-05 04:12:21 +08:00
}
} else {
WREG32(DACA_AUTODETECT_INT_CONTROL, 0);
WREG32(DACB_AUTODETECT_INT_CONTROL, 0);
tmp = RREG32(DC_HOT_PLUG_DETECT1_INT_CONTROL) & DC_HOT_PLUG_DETECTx_INT_POLARITY;
drm/radeon/kms: fix some typos in r6xx/r7xx hpd setup Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-03-25 01:57:29 +08:00
WREG32(DC_HOT_PLUG_DETECT1_INT_CONTROL, tmp);
drm/radeon/kms: add hpd support for r6xx/r7xx/rs780/rs880 asics This just adds the functionality, it's not hooked up yet. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-05 04:12:21 +08:00
tmp = RREG32(DC_HOT_PLUG_DETECT2_INT_CONTROL) & DC_HOT_PLUG_DETECTx_INT_POLARITY;
drm/radeon/kms: fix some typos in r6xx/r7xx hpd setup Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-03-25 01:57:29 +08:00
WREG32(DC_HOT_PLUG_DETECT2_INT_CONTROL, tmp);
drm/radeon/kms: add hpd support for r6xx/r7xx/rs780/rs880 asics This just adds the functionality, it's not hooked up yet. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-05 04:12:21 +08:00
tmp = RREG32(DC_HOT_PLUG_DETECT3_INT_CONTROL) & DC_HOT_PLUG_DETECTx_INT_POLARITY;
drm/radeon/kms: fix some typos in r6xx/r7xx hpd setup Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-03-25 01:57:29 +08:00
WREG32(DC_HOT_PLUG_DETECT3_INT_CONTROL, tmp);
drm/radeon/kms: fix up audio interrupt handling - add support for rs6xx - add support for DCE4/5 - fixup 6xx/7xx Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2012-03-30 20:59:57 +08:00
tmp = RREG32(HDMI0_AUDIO_PACKET_CONTROL) & ~HDMI0_AZ_FORMAT_WTRIG_MASK;
WREG32(HDMI0_AUDIO_PACKET_CONTROL, tmp);
tmp = RREG32(HDMI1_AUDIO_PACKET_CONTROL) & ~HDMI0_AZ_FORMAT_WTRIG_MASK;
WREG32(HDMI1_AUDIO_PACKET_CONTROL, tmp);
drm/radeon/kms: add hpd support for r6xx/r7xx/rs780/rs880 asics This just adds the functionality, it's not hooked up yet. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-05 04:12:21 +08:00
}
}
drm/radeon/kms: Add support for interrupts on r6xx/r7xx chips (v3) This enables the use of interrupts on r6xx/r7xx hardware. Interrupts are implemented via a ring buffer. The GPU adds interrupts vectors to the ring and the host reads them off in the interrupt handler. The interrupt controller requires firmware like the CP. This firmware must be installed and accessble to the firmware loader for interrupts to function. MSIs don't seem to work on my RS780. They work fine on all my discrete cards. I'm not sure about other RS780s or RS880s. I've disabled MSIs on RS780 and RS880, but it would probably be worth checking on some other systems. v2 - fix some checkpatch.pl problems; re-read the disp int status reg if we restart the ih; v3 - remove the irq handler if r600_irq_init() fails; remove spinlock in r600_ih_ring_fini(); move ih rb overflow check to r600_get_ih_wptr(); move irq ack to separate function; Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-02 02:43:46 +08:00
int r600_irq_init(struct radeon_device *rdev)
{
int ret = 0;
int rb_bufsz;
u32 interrupt_cntl, ih_cntl, ih_rb_cntl;
/* allocate ring */
drm/radeon/kms: r600/r700 disable irq at suspend To avoid hw doing anythings after we disabled PCIE GART, fully disable IRQ at suspend. Also cleanup a bit the ih structure and process function. Signed-off-by: Jerome Glisse <jglisse@redhat.com> Reviewed-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@linux.ie>
2010-01-15 21:44:37 +08:00
ret = r600_ih_ring_alloc(rdev);
drm/radeon/kms: Add support for interrupts on r6xx/r7xx chips (v3) This enables the use of interrupts on r6xx/r7xx hardware. Interrupts are implemented via a ring buffer. The GPU adds interrupts vectors to the ring and the host reads them off in the interrupt handler. The interrupt controller requires firmware like the CP. This firmware must be installed and accessble to the firmware loader for interrupts to function. MSIs don't seem to work on my RS780. They work fine on all my discrete cards. I'm not sure about other RS780s or RS880s. I've disabled MSIs on RS780 and RS880, but it would probably be worth checking on some other systems. v2 - fix some checkpatch.pl problems; re-read the disp int status reg if we restart the ih; v3 - remove the irq handler if r600_irq_init() fails; remove spinlock in r600_ih_ring_fini(); move ih rb overflow check to r600_get_ih_wptr(); move irq ack to separate function; Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-02 02:43:46 +08:00
if (ret)
return ret;
/* disable irqs */
r600_disable_interrupts(rdev);
/* init rlc */
ret = r600_rlc_init(rdev);
if (ret) {
r600_ih_ring_fini(rdev);
return ret;
}
/* setup interrupt control */
/* set dummy read address to ring address */
WREG32(INTERRUPT_CNTL2, rdev->ih.gpu_addr >> 8);
interrupt_cntl = RREG32(INTERRUPT_CNTL);
/* IH_DUMMY_RD_OVERRIDE=0 - dummy read disabled with msi, enabled without msi
* IH_DUMMY_RD_OVERRIDE=1 - dummy read controlled by IH_DUMMY_RD_EN
*/
interrupt_cntl &= ~IH_DUMMY_RD_OVERRIDE;
/* IH_REQ_NONSNOOP_EN=1 if ring is in non-cacheable memory, e.g., vram */
interrupt_cntl &= ~IH_REQ_NONSNOOP_EN;
WREG32(INTERRUPT_CNTL, interrupt_cntl);
WREG32(IH_RB_BASE, rdev->ih.gpu_addr >> 8);
rb_bufsz = drm_order(rdev->ih.ring_size / 4);
ih_rb_cntl = (IH_WPTR_OVERFLOW_ENABLE |
IH_WPTR_OVERFLOW_CLEAR |
(rb_bufsz << 1));
drm/radeon/kms: enable writeback (v2) When writeback is enabled, the GPU shadows writes to certain registers into a buffer in memory. The driver can then read the values from the shadow rather than reading back from the register across the bus. Writeback can be disabled by setting the no_wb module param to 1. On r6xx/r7xx/evergreen, the following registers are shadowed: - CP scratch registers - CP read pointer - IH write pointer On r1xx-rr5xx, the following registers are shadowed: - CP scratch registers - CP read pointer v2: - Combine wb patches for r6xx-evergreen and r1xx-r5xx - Writeback is disabled on AGP boards since it tends to be unreliable on AGP using the gart. - Check radeon_wb_init return values properly. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-08-28 06:25:25 +08:00
if (rdev->wb.enabled)
ih_rb_cntl |= IH_WPTR_WRITEBACK_ENABLE;
/* set the writeback address whether it's enabled or not */
WREG32(IH_RB_WPTR_ADDR_LO, (rdev->wb.gpu_addr + R600_WB_IH_WPTR_OFFSET) & 0xFFFFFFFC);
WREG32(IH_RB_WPTR_ADDR_HI, upper_32_bits(rdev->wb.gpu_addr + R600_WB_IH_WPTR_OFFSET) & 0xFF);
drm/radeon/kms: Add support for interrupts on r6xx/r7xx chips (v3) This enables the use of interrupts on r6xx/r7xx hardware. Interrupts are implemented via a ring buffer. The GPU adds interrupts vectors to the ring and the host reads them off in the interrupt handler. The interrupt controller requires firmware like the CP. This firmware must be installed and accessble to the firmware loader for interrupts to function. MSIs don't seem to work on my RS780. They work fine on all my discrete cards. I'm not sure about other RS780s or RS880s. I've disabled MSIs on RS780 and RS880, but it would probably be worth checking on some other systems. v2 - fix some checkpatch.pl problems; re-read the disp int status reg if we restart the ih; v3 - remove the irq handler if r600_irq_init() fails; remove spinlock in r600_ih_ring_fini(); move ih rb overflow check to r600_get_ih_wptr(); move irq ack to separate function; Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-02 02:43:46 +08:00
WREG32(IH_RB_CNTL, ih_rb_cntl);
/* set rptr, wptr to 0 */
WREG32(IH_RB_RPTR, 0);
WREG32(IH_RB_WPTR, 0);
/* Default settings for IH_CNTL (disabled at first) */
ih_cntl = MC_WRREQ_CREDIT(0x10) | MC_WR_CLEAN_CNT(0x10);
/* RPTR_REARM only works if msi's are enabled */
if (rdev->msi_enabled)
ih_cntl |= RPTR_REARM;
WREG32(IH_CNTL, ih_cntl);
/* force the active interrupt state to all disabled */
drm/radeon/kms/evergreen: implement irq support Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-03-25 01:55:51 +08:00
if (rdev->family >= CHIP_CEDAR)
evergreen_disable_interrupt_state(rdev);
else
r600_disable_interrupt_state(rdev);
drm/radeon/kms: Add support for interrupts on r6xx/r7xx chips (v3) This enables the use of interrupts on r6xx/r7xx hardware. Interrupts are implemented via a ring buffer. The GPU adds interrupts vectors to the ring and the host reads them off in the interrupt handler. The interrupt controller requires firmware like the CP. This firmware must be installed and accessble to the firmware loader for interrupts to function. MSIs don't seem to work on my RS780. They work fine on all my discrete cards. I'm not sure about other RS780s or RS880s. I've disabled MSIs on RS780 and RS880, but it would probably be worth checking on some other systems. v2 - fix some checkpatch.pl problems; re-read the disp int status reg if we restart the ih; v3 - remove the irq handler if r600_irq_init() fails; remove spinlock in r600_ih_ring_fini(); move ih rb overflow check to r600_get_ih_wptr(); move irq ack to separate function; Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-02 02:43:46 +08:00
drm/radeon: enable pci bus mastering after card is initialised (v2) This closes a race seen with kexec where we enable PCI bus mastering but the card has been reinitialised fully yet. This was previously fixed by a patch from Jerome, but this should close the race completely. v2: add SI support as suggested by Alex. Reported-and-tested-by: Markus Trippelsdorf <markus@trippelsdorf.de> Signed-off-by: Dave Airlie <airlied@redhat.com>
2012-04-03 18:53:05 +08:00
/* at this point everything should be setup correctly to enable master */
pci_set_master(rdev->pdev);
drm/radeon/kms: Add support for interrupts on r6xx/r7xx chips (v3) This enables the use of interrupts on r6xx/r7xx hardware. Interrupts are implemented via a ring buffer. The GPU adds interrupts vectors to the ring and the host reads them off in the interrupt handler. The interrupt controller requires firmware like the CP. This firmware must be installed and accessble to the firmware loader for interrupts to function. MSIs don't seem to work on my RS780. They work fine on all my discrete cards. I'm not sure about other RS780s or RS880s. I've disabled MSIs on RS780 and RS880, but it would probably be worth checking on some other systems. v2 - fix some checkpatch.pl problems; re-read the disp int status reg if we restart the ih; v3 - remove the irq handler if r600_irq_init() fails; remove spinlock in r600_ih_ring_fini(); move ih rb overflow check to r600_get_ih_wptr(); move irq ack to separate function; Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-02 02:43:46 +08:00
/* enable irqs */
r600_enable_interrupts(rdev);
return ret;
}
drm/radeon/kms: r600/r700 disable irq at suspend To avoid hw doing anythings after we disabled PCIE GART, fully disable IRQ at suspend. Also cleanup a bit the ih structure and process function. Signed-off-by: Jerome Glisse <jglisse@redhat.com> Reviewed-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@linux.ie>
2010-01-15 21:44:37 +08:00
void r600_irq_suspend(struct radeon_device *rdev)
drm/radeon/kms: Add support for interrupts on r6xx/r7xx chips (v3) This enables the use of interrupts on r6xx/r7xx hardware. Interrupts are implemented via a ring buffer. The GPU adds interrupts vectors to the ring and the host reads them off in the interrupt handler. The interrupt controller requires firmware like the CP. This firmware must be installed and accessble to the firmware loader for interrupts to function. MSIs don't seem to work on my RS780. They work fine on all my discrete cards. I'm not sure about other RS780s or RS880s. I've disabled MSIs on RS780 and RS880, but it would probably be worth checking on some other systems. v2 - fix some checkpatch.pl problems; re-read the disp int status reg if we restart the ih; v3 - remove the irq handler if r600_irq_init() fails; remove spinlock in r600_ih_ring_fini(); move ih rb overflow check to r600_get_ih_wptr(); move irq ack to separate function; Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-02 02:43:46 +08:00
{
drm/radeon/kms/evergreen: implement irq support Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-03-25 01:55:51 +08:00
r600_irq_disable(rdev);
drm/radeon/kms: Add support for interrupts on r6xx/r7xx chips (v3) This enables the use of interrupts on r6xx/r7xx hardware. Interrupts are implemented via a ring buffer. The GPU adds interrupts vectors to the ring and the host reads them off in the interrupt handler. The interrupt controller requires firmware like the CP. This firmware must be installed and accessble to the firmware loader for interrupts to function. MSIs don't seem to work on my RS780. They work fine on all my discrete cards. I'm not sure about other RS780s or RS880s. I've disabled MSIs on RS780 and RS880, but it would probably be worth checking on some other systems. v2 - fix some checkpatch.pl problems; re-read the disp int status reg if we restart the ih; v3 - remove the irq handler if r600_irq_init() fails; remove spinlock in r600_ih_ring_fini(); move ih rb overflow check to r600_get_ih_wptr(); move irq ack to separate function; Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-02 02:43:46 +08:00
r600_rlc_stop(rdev);
drm/radeon/kms: r600/r700 disable irq at suspend To avoid hw doing anythings after we disabled PCIE GART, fully disable IRQ at suspend. Also cleanup a bit the ih structure and process function. Signed-off-by: Jerome Glisse <jglisse@redhat.com> Reviewed-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@linux.ie>
2010-01-15 21:44:37 +08:00
}
void r600_irq_fini(struct radeon_device *rdev)
{
r600_irq_suspend(rdev);
drm/radeon/kms: Add support for interrupts on r6xx/r7xx chips (v3) This enables the use of interrupts on r6xx/r7xx hardware. Interrupts are implemented via a ring buffer. The GPU adds interrupts vectors to the ring and the host reads them off in the interrupt handler. The interrupt controller requires firmware like the CP. This firmware must be installed and accessble to the firmware loader for interrupts to function. MSIs don't seem to work on my RS780. They work fine on all my discrete cards. I'm not sure about other RS780s or RS880s. I've disabled MSIs on RS780 and RS880, but it would probably be worth checking on some other systems. v2 - fix some checkpatch.pl problems; re-read the disp int status reg if we restart the ih; v3 - remove the irq handler if r600_irq_init() fails; remove spinlock in r600_ih_ring_fini(); move ih rb overflow check to r600_get_ih_wptr(); move irq ack to separate function; Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-02 02:43:46 +08:00
r600_ih_ring_fini(rdev);
}
int r600_irq_set(struct radeon_device *rdev)
{
drm/radeon/kms: add hpd support for r6xx/r7xx/rs780/rs880 asics This just adds the functionality, it's not hooked up yet. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-05 04:12:21 +08:00
u32 cp_int_cntl = CNTX_BUSY_INT_ENABLE | CNTX_EMPTY_INT_ENABLE;
u32 mode_int = 0;
u32 hpd1, hpd2, hpd3, hpd4 = 0, hpd5 = 0, hpd6 = 0;
drm/radeon/kms: add support for gui idle interrupts (v4) Useful for certain power management operations. You need to wait for the GUI engine (2D, 3D, CP, etc.) to be idle before changing clocks or adjusting engine parameters. (v2) Fix gui idle enable on pre-r6xx asics (v3) The gui idle interrrupt status bit is permanently asserted on pre-r6xx chips, but the interrrupt is still generated. workaround it in the driver. (v4) Add support for evergreen Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-04-23 00:52:11 +08:00
u32 grbm_int_cntl = 0;
drm/radeon/kms: fix up audio interrupt handling - add support for rs6xx - add support for DCE4/5 - fixup 6xx/7xx Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2012-03-30 20:59:57 +08:00
u32 hdmi0, hdmi1;
drm/radeon/kms: add pageflip ioctl support (v3) This adds support for dri2 pageflipping. v2: precision updates from Mario Kleiner. v3: Multihead fixes from Mario Kleiner; missing crtc offset add note about update pending bit on pre-avivo chips Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Mario Kleiner <mario.kleiner@tuebingen.mpg.de> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-11-21 23:59:01 +08:00
u32 d1grph = 0, d2grph = 0;
drm/radeon/kms: Add initial support for async DMA on r6xx/r7xx Uses the new multi-ring infrastucture. 6xx/7xx has a single async DMA ring. Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2012-09-28 03:08:35 +08:00
u32 dma_cntl;
drm/radeon/kms: Add support for interrupts on r6xx/r7xx chips (v3) This enables the use of interrupts on r6xx/r7xx hardware. Interrupts are implemented via a ring buffer. The GPU adds interrupts vectors to the ring and the host reads them off in the interrupt handler. The interrupt controller requires firmware like the CP. This firmware must be installed and accessble to the firmware loader for interrupts to function. MSIs don't seem to work on my RS780. They work fine on all my discrete cards. I'm not sure about other RS780s or RS880s. I've disabled MSIs on RS780 and RS880, but it would probably be worth checking on some other systems. v2 - fix some checkpatch.pl problems; re-read the disp int status reg if we restart the ih; v3 - remove the irq handler if r600_irq_init() fails; remove spinlock in r600_ih_ring_fini(); move ih rb overflow check to r600_get_ih_wptr(); move irq ack to separate function; Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-02 02:43:46 +08:00
drm/radeon/kms: Don't try to enable IRQ if we have no handler installed If for any reason we haven't installed handler we shouldn't try to enable IRQ/MSI on the hw so we don't get unhandled IRQ/MSI which makes the kernel sad. Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-01-07 22:39:14 +08:00
if (!rdev->irq.installed) {
drivers/gpu/drm: Update WARN uses Coalesce long formats. Align arguments. Add missing newlines. Signed-off-by: Joe Perches <joe@perches.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-10-31 05:08:30 +08:00
WARN(1, "Can't enable IRQ/MSI because no handler is installed\n");
drm/radeon/kms: Don't try to enable IRQ if we have no handler installed If for any reason we haven't installed handler we shouldn't try to enable IRQ/MSI on the hw so we don't get unhandled IRQ/MSI which makes the kernel sad. Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-01-07 22:39:14 +08:00
return -EINVAL;
}
drm/radeon/kms: Add support for interrupts on r6xx/r7xx chips (v3) This enables the use of interrupts on r6xx/r7xx hardware. Interrupts are implemented via a ring buffer. The GPU adds interrupts vectors to the ring and the host reads them off in the interrupt handler. The interrupt controller requires firmware like the CP. This firmware must be installed and accessble to the firmware loader for interrupts to function. MSIs don't seem to work on my RS780. They work fine on all my discrete cards. I'm not sure about other RS780s or RS880s. I've disabled MSIs on RS780 and RS880, but it would probably be worth checking on some other systems. v2 - fix some checkpatch.pl problems; re-read the disp int status reg if we restart the ih; v3 - remove the irq handler if r600_irq_init() fails; remove spinlock in r600_ih_ring_fini(); move ih rb overflow check to r600_get_ih_wptr(); move irq ack to separate function; Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-02 02:43:46 +08:00
/* don't enable anything if the ih is disabled */
drm/radeon/kms: r600/r700 don't process IRQ if not initialized In some rare case the wptr returned from the hw wasn't 0 and leaded to trick r600_process_irq that their were irq to process. Add a check to bail out if irq hasn't been initialized this will avoid oops provoqued by the rare wptr != 0 on initialization. Signed-off-by: Jerome Glisse <jglisse@redhat.com> Reviewed-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@linux.ie>
2010-01-15 21:44:38 +08:00
if (!rdev->ih.enabled) {
r600_disable_interrupts(rdev);
/* force the active interrupt state to all disabled */
r600_disable_interrupt_state(rdev);
drm/radeon/kms: Add support for interrupts on r6xx/r7xx chips (v3) This enables the use of interrupts on r6xx/r7xx hardware. Interrupts are implemented via a ring buffer. The GPU adds interrupts vectors to the ring and the host reads them off in the interrupt handler. The interrupt controller requires firmware like the CP. This firmware must be installed and accessble to the firmware loader for interrupts to function. MSIs don't seem to work on my RS780. They work fine on all my discrete cards. I'm not sure about other RS780s or RS880s. I've disabled MSIs on RS780 and RS880, but it would probably be worth checking on some other systems. v2 - fix some checkpatch.pl problems; re-read the disp int status reg if we restart the ih; v3 - remove the irq handler if r600_irq_init() fails; remove spinlock in r600_ih_ring_fini(); move ih rb overflow check to r600_get_ih_wptr(); move irq ack to separate function; Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-02 02:43:46 +08:00
return 0;
drm/radeon/kms: r600/r700 don't process IRQ if not initialized In some rare case the wptr returned from the hw wasn't 0 and leaded to trick r600_process_irq that their were irq to process. Add a check to bail out if irq hasn't been initialized this will avoid oops provoqued by the rare wptr != 0 on initialization. Signed-off-by: Jerome Glisse <jglisse@redhat.com> Reviewed-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@linux.ie>
2010-01-15 21:44:38 +08:00
}
drm/radeon/kms: Add support for interrupts on r6xx/r7xx chips (v3) This enables the use of interrupts on r6xx/r7xx hardware. Interrupts are implemented via a ring buffer. The GPU adds interrupts vectors to the ring and the host reads them off in the interrupt handler. The interrupt controller requires firmware like the CP. This firmware must be installed and accessble to the firmware loader for interrupts to function. MSIs don't seem to work on my RS780. They work fine on all my discrete cards. I'm not sure about other RS780s or RS880s. I've disabled MSIs on RS780 and RS880, but it would probably be worth checking on some other systems. v2 - fix some checkpatch.pl problems; re-read the disp int status reg if we restart the ih; v3 - remove the irq handler if r600_irq_init() fails; remove spinlock in r600_ih_ring_fini(); move ih rb overflow check to r600_get_ih_wptr(); move irq ack to separate function; Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-02 02:43:46 +08:00
drm/radeon/kms: add hpd support for r6xx/r7xx/rs780/rs880 asics This just adds the functionality, it's not hooked up yet. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-05 04:12:21 +08:00
if (ASIC_IS_DCE3(rdev)) {
hpd1 = RREG32(DC_HPD1_INT_CONTROL) & ~DC_HPDx_INT_EN;
hpd2 = RREG32(DC_HPD2_INT_CONTROL) & ~DC_HPDx_INT_EN;
hpd3 = RREG32(DC_HPD3_INT_CONTROL) & ~DC_HPDx_INT_EN;
hpd4 = RREG32(DC_HPD4_INT_CONTROL) & ~DC_HPDx_INT_EN;
if (ASIC_IS_DCE32(rdev)) {
hpd5 = RREG32(DC_HPD5_INT_CONTROL) & ~DC_HPDx_INT_EN;
hpd6 = RREG32(DC_HPD6_INT_CONTROL) & ~DC_HPDx_INT_EN;
drm/radeon/kms/hdmi: use relative offsets, official regs Signed-off-by: Rafał Miłecki <zajec5@gmail.com> Tested-by: Christian König <deathsimple@vodafone.de> Reviewed-by: Christian König <deathsimple@vodafone.de> Signed-off-by: Dave Airlie <airlied@redhat.com>
2012-04-29 05:35:24 +08:00
hdmi0 = RREG32(AFMT_AUDIO_PACKET_CONTROL + DCE3_HDMI_OFFSET0) & ~AFMT_AZ_FORMAT_WTRIG_MASK;
hdmi1 = RREG32(AFMT_AUDIO_PACKET_CONTROL + DCE3_HDMI_OFFSET1) & ~AFMT_AZ_FORMAT_WTRIG_MASK;
drm/radeon/kms: fix up audio interrupt handling - add support for rs6xx - add support for DCE4/5 - fixup 6xx/7xx Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2012-03-30 20:59:57 +08:00
} else {
hdmi0 = RREG32(HDMI0_AUDIO_PACKET_CONTROL) & ~HDMI0_AZ_FORMAT_WTRIG_MASK;
hdmi1 = RREG32(DCE3_HDMI1_AUDIO_PACKET_CONTROL) & ~HDMI0_AZ_FORMAT_WTRIG_MASK;
drm/radeon/kms: add hpd support for r6xx/r7xx/rs780/rs880 asics This just adds the functionality, it's not hooked up yet. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-05 04:12:21 +08:00
}
} else {
hpd1 = RREG32(DC_HOT_PLUG_DETECT1_INT_CONTROL) & ~DC_HPDx_INT_EN;
hpd2 = RREG32(DC_HOT_PLUG_DETECT2_INT_CONTROL) & ~DC_HPDx_INT_EN;
hpd3 = RREG32(DC_HOT_PLUG_DETECT3_INT_CONTROL) & ~DC_HPDx_INT_EN;
drm/radeon/kms: fix up audio interrupt handling - add support for rs6xx - add support for DCE4/5 - fixup 6xx/7xx Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2012-03-30 20:59:57 +08:00
hdmi0 = RREG32(HDMI0_AUDIO_PACKET_CONTROL) & ~HDMI0_AZ_FORMAT_WTRIG_MASK;
hdmi1 = RREG32(HDMI1_AUDIO_PACKET_CONTROL) & ~HDMI0_AZ_FORMAT_WTRIG_MASK;
drm/radeon/kms: add hpd support for r6xx/r7xx/rs780/rs880 asics This just adds the functionality, it's not hooked up yet. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-05 04:12:21 +08:00
}
drm/radeon/kms: Add initial support for async DMA on r6xx/r7xx Uses the new multi-ring infrastucture. 6xx/7xx has a single async DMA ring. Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2012-09-28 03:08:35 +08:00
dma_cntl = RREG32(DMA_CNTL) & ~TRAP_ENABLE;
drm/radeon/kms: add hpd support for r6xx/r7xx/rs780/rs880 asics This just adds the functionality, it's not hooked up yet. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-05 04:12:21 +08:00
drm/radeon: replace pflip and sw_int counters with atomics So we can skip the locking. Also renames sw_int to ring_int, cause that better matches its purpose. Signed-off-by: Christian Koenig <christian.koenig@amd.com>
2012-05-18 01:52:00 +08:00
if (atomic_read(&rdev->irq.ring_int[RADEON_RING_TYPE_GFX_INDEX])) {
drm/radeon/kms: Add support for interrupts on r6xx/r7xx chips (v3) This enables the use of interrupts on r6xx/r7xx hardware. Interrupts are implemented via a ring buffer. The GPU adds interrupts vectors to the ring and the host reads them off in the interrupt handler. The interrupt controller requires firmware like the CP. This firmware must be installed and accessble to the firmware loader for interrupts to function. MSIs don't seem to work on my RS780. They work fine on all my discrete cards. I'm not sure about other RS780s or RS880s. I've disabled MSIs on RS780 and RS880, but it would probably be worth checking on some other systems. v2 - fix some checkpatch.pl problems; re-read the disp int status reg if we restart the ih; v3 - remove the irq handler if r600_irq_init() fails; remove spinlock in r600_ih_ring_fini(); move ih rb overflow check to r600_get_ih_wptr(); move irq ack to separate function; Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-02 02:43:46 +08:00
DRM_DEBUG("r600_irq_set: sw int\n");
cp_int_cntl |= RB_INT_ENABLE;
drm/radeon/kms/r6xx+: use new style fencing (v3) On r6xx+ a newer fence mechanism was implemented to replace the old wait_until plus scratch regs setup. A single EOP event will flush the destination caches, write a fence value, and generate an interrupt. This is the recommended fence mechanism on r6xx+ asics. This requires my previous writeback patch. v2: fix typo that enabled event fence checking on all asics rather than just r6xx+. v3: properly enable EOP interrupts Should fix: https://bugs.freedesktop.org/show_bug.cgi?id=29972 Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-09-04 17:04:34 +08:00
cp_int_cntl |= TIME_STAMP_INT_ENABLE;
drm/radeon/kms: Add support for interrupts on r6xx/r7xx chips (v3) This enables the use of interrupts on r6xx/r7xx hardware. Interrupts are implemented via a ring buffer. The GPU adds interrupts vectors to the ring and the host reads them off in the interrupt handler. The interrupt controller requires firmware like the CP. This firmware must be installed and accessble to the firmware loader for interrupts to function. MSIs don't seem to work on my RS780. They work fine on all my discrete cards. I'm not sure about other RS780s or RS880s. I've disabled MSIs on RS780 and RS880, but it would probably be worth checking on some other systems. v2 - fix some checkpatch.pl problems; re-read the disp int status reg if we restart the ih; v3 - remove the irq handler if r600_irq_init() fails; remove spinlock in r600_ih_ring_fini(); move ih rb overflow check to r600_get_ih_wptr(); move irq ack to separate function; Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-02 02:43:46 +08:00
}
drm/radeon/kms: Add initial support for async DMA on r6xx/r7xx Uses the new multi-ring infrastucture. 6xx/7xx has a single async DMA ring. Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2012-09-28 03:08:35 +08:00
if (atomic_read(&rdev->irq.ring_int[R600_RING_TYPE_DMA_INDEX])) {
DRM_DEBUG("r600_irq_set: sw int dma\n");
dma_cntl |= TRAP_ENABLE;
}
drm/radeon/kms: add pageflip ioctl support (v3) This adds support for dri2 pageflipping. v2: precision updates from Mario Kleiner. v3: Multihead fixes from Mario Kleiner; missing crtc offset add note about update pending bit on pre-avivo chips Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Mario Kleiner <mario.kleiner@tuebingen.mpg.de> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-11-21 23:59:01 +08:00
if (rdev->irq.crtc_vblank_int[0] ||
drm/radeon: replace pflip and sw_int counters with atomics So we can skip the locking. Also renames sw_int to ring_int, cause that better matches its purpose. Signed-off-by: Christian Koenig <christian.koenig@amd.com>
2012-05-18 01:52:00 +08:00
atomic_read(&rdev->irq.pflip[0])) {
drm/radeon/kms: Add support for interrupts on r6xx/r7xx chips (v3) This enables the use of interrupts on r6xx/r7xx hardware. Interrupts are implemented via a ring buffer. The GPU adds interrupts vectors to the ring and the host reads them off in the interrupt handler. The interrupt controller requires firmware like the CP. This firmware must be installed and accessble to the firmware loader for interrupts to function. MSIs don't seem to work on my RS780. They work fine on all my discrete cards. I'm not sure about other RS780s or RS880s. I've disabled MSIs on RS780 and RS880, but it would probably be worth checking on some other systems. v2 - fix some checkpatch.pl problems; re-read the disp int status reg if we restart the ih; v3 - remove the irq handler if r600_irq_init() fails; remove spinlock in r600_ih_ring_fini(); move ih rb overflow check to r600_get_ih_wptr(); move irq ack to separate function; Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-02 02:43:46 +08:00
DRM_DEBUG("r600_irq_set: vblank 0\n");
mode_int |= D1MODE_VBLANK_INT_MASK;
}
drm/radeon/kms: add pageflip ioctl support (v3) This adds support for dri2 pageflipping. v2: precision updates from Mario Kleiner. v3: Multihead fixes from Mario Kleiner; missing crtc offset add note about update pending bit on pre-avivo chips Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Mario Kleiner <mario.kleiner@tuebingen.mpg.de> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-11-21 23:59:01 +08:00
if (rdev->irq.crtc_vblank_int[1] ||
drm/radeon: replace pflip and sw_int counters with atomics So we can skip the locking. Also renames sw_int to ring_int, cause that better matches its purpose. Signed-off-by: Christian Koenig <christian.koenig@amd.com>
2012-05-18 01:52:00 +08:00
atomic_read(&rdev->irq.pflip[1])) {
drm/radeon/kms: Add support for interrupts on r6xx/r7xx chips (v3) This enables the use of interrupts on r6xx/r7xx hardware. Interrupts are implemented via a ring buffer. The GPU adds interrupts vectors to the ring and the host reads them off in the interrupt handler. The interrupt controller requires firmware like the CP. This firmware must be installed and accessble to the firmware loader for interrupts to function. MSIs don't seem to work on my RS780. They work fine on all my discrete cards. I'm not sure about other RS780s or RS880s. I've disabled MSIs on RS780 and RS880, but it would probably be worth checking on some other systems. v2 - fix some checkpatch.pl problems; re-read the disp int status reg if we restart the ih; v3 - remove the irq handler if r600_irq_init() fails; remove spinlock in r600_ih_ring_fini(); move ih rb overflow check to r600_get_ih_wptr(); move irq ack to separate function; Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-02 02:43:46 +08:00
DRM_DEBUG("r600_irq_set: vblank 1\n");
mode_int |= D2MODE_VBLANK_INT_MASK;
}
drm/radeon/kms: add hpd support for r6xx/r7xx/rs780/rs880 asics This just adds the functionality, it's not hooked up yet. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-05 04:12:21 +08:00
if (rdev->irq.hpd[0]) {
DRM_DEBUG("r600_irq_set: hpd 1\n");
hpd1 |= DC_HPDx_INT_EN;
}
if (rdev->irq.hpd[1]) {
DRM_DEBUG("r600_irq_set: hpd 2\n");
hpd2 |= DC_HPDx_INT_EN;
}
if (rdev->irq.hpd[2]) {
DRM_DEBUG("r600_irq_set: hpd 3\n");
hpd3 |= DC_HPDx_INT_EN;
}
if (rdev->irq.hpd[3]) {
DRM_DEBUG("r600_irq_set: hpd 4\n");
hpd4 |= DC_HPDx_INT_EN;
}
if (rdev->irq.hpd[4]) {
DRM_DEBUG("r600_irq_set: hpd 5\n");
hpd5 |= DC_HPDx_INT_EN;
}
if (rdev->irq.hpd[5]) {
DRM_DEBUG("r600_irq_set: hpd 6\n");
hpd6 |= DC_HPDx_INT_EN;
}
drm/radeon/kms: fix up audio interrupt handling - add support for rs6xx - add support for DCE4/5 - fixup 6xx/7xx Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2012-03-30 20:59:57 +08:00
if (rdev->irq.afmt[0]) {
DRM_DEBUG("r600_irq_set: hdmi 0\n");
hdmi0 |= HDMI0_AZ_FORMAT_WTRIG_MASK;
drm/radeon/kms: HDMI irq support Implements irq support for HDMI audio output. Now the polling timer is only enabled if irq support isn't available. Signed-off-by: Christian König <deathsimple@vodafone.de> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-04-10 09:13:16 +08:00
}
drm/radeon/kms: fix up audio interrupt handling - add support for rs6xx - add support for DCE4/5 - fixup 6xx/7xx Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2012-03-30 20:59:57 +08:00
if (rdev->irq.afmt[1]) {
DRM_DEBUG("r600_irq_set: hdmi 0\n");
hdmi1 |= HDMI0_AZ_FORMAT_WTRIG_MASK;
drm/radeon/kms: HDMI irq support Implements irq support for HDMI audio output. Now the polling timer is only enabled if irq support isn't available. Signed-off-by: Christian König <deathsimple@vodafone.de> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-04-10 09:13:16 +08:00
}
drm/radeon/kms: Add support for interrupts on r6xx/r7xx chips (v3) This enables the use of interrupts on r6xx/r7xx hardware. Interrupts are implemented via a ring buffer. The GPU adds interrupts vectors to the ring and the host reads them off in the interrupt handler. The interrupt controller requires firmware like the CP. This firmware must be installed and accessble to the firmware loader for interrupts to function. MSIs don't seem to work on my RS780. They work fine on all my discrete cards. I'm not sure about other RS780s or RS880s. I've disabled MSIs on RS780 and RS880, but it would probably be worth checking on some other systems. v2 - fix some checkpatch.pl problems; re-read the disp int status reg if we restart the ih; v3 - remove the irq handler if r600_irq_init() fails; remove spinlock in r600_ih_ring_fini(); move ih rb overflow check to r600_get_ih_wptr(); move irq ack to separate function; Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-02 02:43:46 +08:00
WREG32(CP_INT_CNTL, cp_int_cntl);
drm/radeon/kms: Add initial support for async DMA on r6xx/r7xx Uses the new multi-ring infrastucture. 6xx/7xx has a single async DMA ring. Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2012-09-28 03:08:35 +08:00
WREG32(DMA_CNTL, dma_cntl);
drm/radeon/kms: Add support for interrupts on r6xx/r7xx chips (v3) This enables the use of interrupts on r6xx/r7xx hardware. Interrupts are implemented via a ring buffer. The GPU adds interrupts vectors to the ring and the host reads them off in the interrupt handler. The interrupt controller requires firmware like the CP. This firmware must be installed and accessble to the firmware loader for interrupts to function. MSIs don't seem to work on my RS780. They work fine on all my discrete cards. I'm not sure about other RS780s or RS880s. I've disabled MSIs on RS780 and RS880, but it would probably be worth checking on some other systems. v2 - fix some checkpatch.pl problems; re-read the disp int status reg if we restart the ih; v3 - remove the irq handler if r600_irq_init() fails; remove spinlock in r600_ih_ring_fini(); move ih rb overflow check to r600_get_ih_wptr(); move irq ack to separate function; Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-02 02:43:46 +08:00
WREG32(DxMODE_INT_MASK, mode_int);
drm/radeon/kms: add pageflip ioctl support (v3) This adds support for dri2 pageflipping. v2: precision updates from Mario Kleiner. v3: Multihead fixes from Mario Kleiner; missing crtc offset add note about update pending bit on pre-avivo chips Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Mario Kleiner <mario.kleiner@tuebingen.mpg.de> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-11-21 23:59:01 +08:00
WREG32(D1GRPH_INTERRUPT_CONTROL, d1grph);
WREG32(D2GRPH_INTERRUPT_CONTROL, d2grph);
drm/radeon/kms: add support for gui idle interrupts (v4) Useful for certain power management operations. You need to wait for the GUI engine (2D, 3D, CP, etc.) to be idle before changing clocks or adjusting engine parameters. (v2) Fix gui idle enable on pre-r6xx asics (v3) The gui idle interrrupt status bit is permanently asserted on pre-r6xx chips, but the interrrupt is still generated. workaround it in the driver. (v4) Add support for evergreen Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-04-23 00:52:11 +08:00
WREG32(GRBM_INT_CNTL, grbm_int_cntl);
drm/radeon/kms: add hpd support for r6xx/r7xx/rs780/rs880 asics This just adds the functionality, it's not hooked up yet. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-05 04:12:21 +08:00
if (ASIC_IS_DCE3(rdev)) {
WREG32(DC_HPD1_INT_CONTROL, hpd1);
WREG32(DC_HPD2_INT_CONTROL, hpd2);
WREG32(DC_HPD3_INT_CONTROL, hpd3);
WREG32(DC_HPD4_INT_CONTROL, hpd4);
if (ASIC_IS_DCE32(rdev)) {
WREG32(DC_HPD5_INT_CONTROL, hpd5);
WREG32(DC_HPD6_INT_CONTROL, hpd6);
drm/radeon/kms/hdmi: use relative offsets, official regs Signed-off-by: Rafał Miłecki <zajec5@gmail.com> Tested-by: Christian König <deathsimple@vodafone.de> Reviewed-by: Christian König <deathsimple@vodafone.de> Signed-off-by: Dave Airlie <airlied@redhat.com>
2012-04-29 05:35:24 +08:00
WREG32(AFMT_AUDIO_PACKET_CONTROL + DCE3_HDMI_OFFSET0, hdmi0);
WREG32(AFMT_AUDIO_PACKET_CONTROL + DCE3_HDMI_OFFSET1, hdmi1);
drm/radeon/kms: fix up audio interrupt handling - add support for rs6xx - add support for DCE4/5 - fixup 6xx/7xx Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2012-03-30 20:59:57 +08:00
} else {
WREG32(HDMI0_AUDIO_PACKET_CONTROL, hdmi0);
WREG32(DCE3_HDMI1_AUDIO_PACKET_CONTROL, hdmi1);
drm/radeon/kms: add hpd support for r6xx/r7xx/rs780/rs880 asics This just adds the functionality, it's not hooked up yet. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-05 04:12:21 +08:00
}
} else {
WREG32(DC_HOT_PLUG_DETECT1_INT_CONTROL, hpd1);
WREG32(DC_HOT_PLUG_DETECT2_INT_CONTROL, hpd2);
WREG32(DC_HOT_PLUG_DETECT3_INT_CONTROL, hpd3);
drm/radeon/kms: fix up audio interrupt handling - add support for rs6xx - add support for DCE4/5 - fixup 6xx/7xx Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2012-03-30 20:59:57 +08:00
WREG32(HDMI0_AUDIO_PACKET_CONTROL, hdmi0);
WREG32(HDMI1_AUDIO_PACKET_CONTROL, hdmi1);
drm/radeon/kms: add hpd support for r6xx/r7xx/rs780/rs880 asics This just adds the functionality, it's not hooked up yet. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-05 04:12:21 +08:00
}
drm/radeon/kms: Add support for interrupts on r6xx/r7xx chips (v3) This enables the use of interrupts on r6xx/r7xx hardware. Interrupts are implemented via a ring buffer. The GPU adds interrupts vectors to the ring and the host reads them off in the interrupt handler. The interrupt controller requires firmware like the CP. This firmware must be installed and accessble to the firmware loader for interrupts to function. MSIs don't seem to work on my RS780. They work fine on all my discrete cards. I'm not sure about other RS780s or RS880s. I've disabled MSIs on RS780 and RS880, but it would probably be worth checking on some other systems. v2 - fix some checkpatch.pl problems; re-read the disp int status reg if we restart the ih; v3 - remove the irq handler if r600_irq_init() fails; remove spinlock in r600_ih_ring_fini(); move ih rb overflow check to r600_get_ih_wptr(); move irq ack to separate function; Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-02 02:43:46 +08:00
return 0;
}
drm/radeon: Move more code out of line With this patch I'm only about 50k larger with DRM debugging enables (why is that enabled by default?!?), and slightly smaller without. [airlied: moved r100.c additions to radeon_ring.c] Signed-off-by: Andi Kleen <ak@linux.intel.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-10-14 07:08:47 +08:00
static void r600_irq_ack(struct radeon_device *rdev)
drm/radeon/kms: Add support for interrupts on r6xx/r7xx chips (v3) This enables the use of interrupts on r6xx/r7xx hardware. Interrupts are implemented via a ring buffer. The GPU adds interrupts vectors to the ring and the host reads them off in the interrupt handler. The interrupt controller requires firmware like the CP. This firmware must be installed and accessble to the firmware loader for interrupts to function. MSIs don't seem to work on my RS780. They work fine on all my discrete cards. I'm not sure about other RS780s or RS880s. I've disabled MSIs on RS780 and RS880, but it would probably be worth checking on some other systems. v2 - fix some checkpatch.pl problems; re-read the disp int status reg if we restart the ih; v3 - remove the irq handler if r600_irq_init() fails; remove spinlock in r600_ih_ring_fini(); move ih rb overflow check to r600_get_ih_wptr(); move irq ack to separate function; Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-02 02:43:46 +08:00
{
drm/radeon/kms: add hpd support for r6xx/r7xx/rs780/rs880 asics This just adds the functionality, it's not hooked up yet. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-05 04:12:21 +08:00
u32 tmp;
if (ASIC_IS_DCE3(rdev)) {
drm/radeon/kms: add pageflip ioctl support (v3) This adds support for dri2 pageflipping. v2: precision updates from Mario Kleiner. v3: Multihead fixes from Mario Kleiner; missing crtc offset add note about update pending bit on pre-avivo chips Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Mario Kleiner <mario.kleiner@tuebingen.mpg.de> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-11-21 23:59:01 +08:00
rdev->irq.stat_regs.r600.disp_int = RREG32(DCE3_DISP_INTERRUPT_STATUS);
rdev->irq.stat_regs.r600.disp_int_cont = RREG32(DCE3_DISP_INTERRUPT_STATUS_CONTINUE);
rdev->irq.stat_regs.r600.disp_int_cont2 = RREG32(DCE3_DISP_INTERRUPT_STATUS_CONTINUE2);
drm/radeon/kms: fix up audio interrupt handling - add support for rs6xx - add support for DCE4/5 - fixup 6xx/7xx Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2012-03-30 20:59:57 +08:00
if (ASIC_IS_DCE32(rdev)) {
drm/radeon/kms/hdmi: use relative offsets, official regs Signed-off-by: Rafał Miłecki <zajec5@gmail.com> Tested-by: Christian König <deathsimple@vodafone.de> Reviewed-by: Christian König <deathsimple@vodafone.de> Signed-off-by: Dave Airlie <airlied@redhat.com>
2012-04-29 05:35:24 +08:00
rdev->irq.stat_regs.r600.hdmi0_status = RREG32(AFMT_STATUS + DCE3_HDMI_OFFSET0);
rdev->irq.stat_regs.r600.hdmi1_status = RREG32(AFMT_STATUS + DCE3_HDMI_OFFSET1);
drm/radeon/kms: fix up audio interrupt handling - add support for rs6xx - add support for DCE4/5 - fixup 6xx/7xx Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2012-03-30 20:59:57 +08:00
} else {
rdev->irq.stat_regs.r600.hdmi0_status = RREG32(HDMI0_STATUS);
rdev->irq.stat_regs.r600.hdmi1_status = RREG32(DCE3_HDMI1_STATUS);
}
drm/radeon/kms: add hpd support for r6xx/r7xx/rs780/rs880 asics This just adds the functionality, it's not hooked up yet. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-05 04:12:21 +08:00
} else {
drm/radeon/kms: add pageflip ioctl support (v3) This adds support for dri2 pageflipping. v2: precision updates from Mario Kleiner. v3: Multihead fixes from Mario Kleiner; missing crtc offset add note about update pending bit on pre-avivo chips Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Mario Kleiner <mario.kleiner@tuebingen.mpg.de> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-11-21 23:59:01 +08:00
rdev->irq.stat_regs.r600.disp_int = RREG32(DISP_INTERRUPT_STATUS);
rdev->irq.stat_regs.r600.disp_int_cont = RREG32(DISP_INTERRUPT_STATUS_CONTINUE);
rdev->irq.stat_regs.r600.disp_int_cont2 = 0;
drm/radeon/kms: fix up audio interrupt handling - add support for rs6xx - add support for DCE4/5 - fixup 6xx/7xx Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2012-03-30 20:59:57 +08:00
rdev->irq.stat_regs.r600.hdmi0_status = RREG32(HDMI0_STATUS);
rdev->irq.stat_regs.r600.hdmi1_status = RREG32(HDMI1_STATUS);
drm/radeon/kms: add pageflip ioctl support (v3) This adds support for dri2 pageflipping. v2: precision updates from Mario Kleiner. v3: Multihead fixes from Mario Kleiner; missing crtc offset add note about update pending bit on pre-avivo chips Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Mario Kleiner <mario.kleiner@tuebingen.mpg.de> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-11-21 23:59:01 +08:00
}
rdev->irq.stat_regs.r600.d1grph_int = RREG32(D1GRPH_INTERRUPT_STATUS);
rdev->irq.stat_regs.r600.d2grph_int = RREG32(D2GRPH_INTERRUPT_STATUS);
if (rdev->irq.stat_regs.r600.d1grph_int & DxGRPH_PFLIP_INT_OCCURRED)
WREG32(D1GRPH_INTERRUPT_STATUS, DxGRPH_PFLIP_INT_CLEAR);
if (rdev->irq.stat_regs.r600.d2grph_int & DxGRPH_PFLIP_INT_OCCURRED)
WREG32(D2GRPH_INTERRUPT_STATUS, DxGRPH_PFLIP_INT_CLEAR);
if (rdev->irq.stat_regs.r600.disp_int & LB_D1_VBLANK_INTERRUPT)
drm/radeon/kms: Add support for interrupts on r6xx/r7xx chips (v3) This enables the use of interrupts on r6xx/r7xx hardware. Interrupts are implemented via a ring buffer. The GPU adds interrupts vectors to the ring and the host reads them off in the interrupt handler. The interrupt controller requires firmware like the CP. This firmware must be installed and accessble to the firmware loader for interrupts to function. MSIs don't seem to work on my RS780. They work fine on all my discrete cards. I'm not sure about other RS780s or RS880s. I've disabled MSIs on RS780 and RS880, but it would probably be worth checking on some other systems. v2 - fix some checkpatch.pl problems; re-read the disp int status reg if we restart the ih; v3 - remove the irq handler if r600_irq_init() fails; remove spinlock in r600_ih_ring_fini(); move ih rb overflow check to r600_get_ih_wptr(); move irq ack to separate function; Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-02 02:43:46 +08:00
WREG32(D1MODE_VBLANK_STATUS, DxMODE_VBLANK_ACK);
drm/radeon/kms: add pageflip ioctl support (v3) This adds support for dri2 pageflipping. v2: precision updates from Mario Kleiner. v3: Multihead fixes from Mario Kleiner; missing crtc offset add note about update pending bit on pre-avivo chips Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Mario Kleiner <mario.kleiner@tuebingen.mpg.de> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-11-21 23:59:01 +08:00
if (rdev->irq.stat_regs.r600.disp_int & LB_D1_VLINE_INTERRUPT)
drm/radeon/kms: Add support for interrupts on r6xx/r7xx chips (v3) This enables the use of interrupts on r6xx/r7xx hardware. Interrupts are implemented via a ring buffer. The GPU adds interrupts vectors to the ring and the host reads them off in the interrupt handler. The interrupt controller requires firmware like the CP. This firmware must be installed and accessble to the firmware loader for interrupts to function. MSIs don't seem to work on my RS780. They work fine on all my discrete cards. I'm not sure about other RS780s or RS880s. I've disabled MSIs on RS780 and RS880, but it would probably be worth checking on some other systems. v2 - fix some checkpatch.pl problems; re-read the disp int status reg if we restart the ih; v3 - remove the irq handler if r600_irq_init() fails; remove spinlock in r600_ih_ring_fini(); move ih rb overflow check to r600_get_ih_wptr(); move irq ack to separate function; Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-02 02:43:46 +08:00
WREG32(D1MODE_VLINE_STATUS, DxMODE_VLINE_ACK);
drm/radeon/kms: add pageflip ioctl support (v3) This adds support for dri2 pageflipping. v2: precision updates from Mario Kleiner. v3: Multihead fixes from Mario Kleiner; missing crtc offset add note about update pending bit on pre-avivo chips Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Mario Kleiner <mario.kleiner@tuebingen.mpg.de> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-11-21 23:59:01 +08:00
if (rdev->irq.stat_regs.r600.disp_int & LB_D2_VBLANK_INTERRUPT)
drm/radeon/kms: Add support for interrupts on r6xx/r7xx chips (v3) This enables the use of interrupts on r6xx/r7xx hardware. Interrupts are implemented via a ring buffer. The GPU adds interrupts vectors to the ring and the host reads them off in the interrupt handler. The interrupt controller requires firmware like the CP. This firmware must be installed and accessble to the firmware loader for interrupts to function. MSIs don't seem to work on my RS780. They work fine on all my discrete cards. I'm not sure about other RS780s or RS880s. I've disabled MSIs on RS780 and RS880, but it would probably be worth checking on some other systems. v2 - fix some checkpatch.pl problems; re-read the disp int status reg if we restart the ih; v3 - remove the irq handler if r600_irq_init() fails; remove spinlock in r600_ih_ring_fini(); move ih rb overflow check to r600_get_ih_wptr(); move irq ack to separate function; Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-02 02:43:46 +08:00
WREG32(D2MODE_VBLANK_STATUS, DxMODE_VBLANK_ACK);
drm/radeon/kms: add pageflip ioctl support (v3) This adds support for dri2 pageflipping. v2: precision updates from Mario Kleiner. v3: Multihead fixes from Mario Kleiner; missing crtc offset add note about update pending bit on pre-avivo chips Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Mario Kleiner <mario.kleiner@tuebingen.mpg.de> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-11-21 23:59:01 +08:00
if (rdev->irq.stat_regs.r600.disp_int & LB_D2_VLINE_INTERRUPT)
drm/radeon/kms: Add support for interrupts on r6xx/r7xx chips (v3) This enables the use of interrupts on r6xx/r7xx hardware. Interrupts are implemented via a ring buffer. The GPU adds interrupts vectors to the ring and the host reads them off in the interrupt handler. The interrupt controller requires firmware like the CP. This firmware must be installed and accessble to the firmware loader for interrupts to function. MSIs don't seem to work on my RS780. They work fine on all my discrete cards. I'm not sure about other RS780s or RS880s. I've disabled MSIs on RS780 and RS880, but it would probably be worth checking on some other systems. v2 - fix some checkpatch.pl problems; re-read the disp int status reg if we restart the ih; v3 - remove the irq handler if r600_irq_init() fails; remove spinlock in r600_ih_ring_fini(); move ih rb overflow check to r600_get_ih_wptr(); move irq ack to separate function; Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-02 02:43:46 +08:00
WREG32(D2MODE_VLINE_STATUS, DxMODE_VLINE_ACK);
drm/radeon/kms: add pageflip ioctl support (v3) This adds support for dri2 pageflipping. v2: precision updates from Mario Kleiner. v3: Multihead fixes from Mario Kleiner; missing crtc offset add note about update pending bit on pre-avivo chips Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Mario Kleiner <mario.kleiner@tuebingen.mpg.de> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-11-21 23:59:01 +08:00
if (rdev->irq.stat_regs.r600.disp_int & DC_HPD1_INTERRUPT) {
drm/radeon/kms: add hpd support for r6xx/r7xx/rs780/rs880 asics This just adds the functionality, it's not hooked up yet. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-05 04:12:21 +08:00
if (ASIC_IS_DCE3(rdev)) {
tmp = RREG32(DC_HPD1_INT_CONTROL);
tmp |= DC_HPDx_INT_ACK;
WREG32(DC_HPD1_INT_CONTROL, tmp);
} else {
tmp = RREG32(DC_HOT_PLUG_DETECT1_INT_CONTROL);
tmp |= DC_HPDx_INT_ACK;
WREG32(DC_HOT_PLUG_DETECT1_INT_CONTROL, tmp);
}
}
drm/radeon/kms: add pageflip ioctl support (v3) This adds support for dri2 pageflipping. v2: precision updates from Mario Kleiner. v3: Multihead fixes from Mario Kleiner; missing crtc offset add note about update pending bit on pre-avivo chips Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Mario Kleiner <mario.kleiner@tuebingen.mpg.de> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-11-21 23:59:01 +08:00
if (rdev->irq.stat_regs.r600.disp_int & DC_HPD2_INTERRUPT) {
drm/radeon/kms: add hpd support for r6xx/r7xx/rs780/rs880 asics This just adds the functionality, it's not hooked up yet. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-05 04:12:21 +08:00
if (ASIC_IS_DCE3(rdev)) {
tmp = RREG32(DC_HPD2_INT_CONTROL);
tmp |= DC_HPDx_INT_ACK;
WREG32(DC_HPD2_INT_CONTROL, tmp);
} else {
tmp = RREG32(DC_HOT_PLUG_DETECT2_INT_CONTROL);
tmp |= DC_HPDx_INT_ACK;
WREG32(DC_HOT_PLUG_DETECT2_INT_CONTROL, tmp);
}
}
drm/radeon/kms: add pageflip ioctl support (v3) This adds support for dri2 pageflipping. v2: precision updates from Mario Kleiner. v3: Multihead fixes from Mario Kleiner; missing crtc offset add note about update pending bit on pre-avivo chips Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Mario Kleiner <mario.kleiner@tuebingen.mpg.de> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-11-21 23:59:01 +08:00
if (rdev->irq.stat_regs.r600.disp_int_cont & DC_HPD3_INTERRUPT) {
drm/radeon/kms: add hpd support for r6xx/r7xx/rs780/rs880 asics This just adds the functionality, it's not hooked up yet. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-05 04:12:21 +08:00
if (ASIC_IS_DCE3(rdev)) {
tmp = RREG32(DC_HPD3_INT_CONTROL);
tmp |= DC_HPDx_INT_ACK;
WREG32(DC_HPD3_INT_CONTROL, tmp);
} else {
tmp = RREG32(DC_HOT_PLUG_DETECT3_INT_CONTROL);
tmp |= DC_HPDx_INT_ACK;
WREG32(DC_HOT_PLUG_DETECT3_INT_CONTROL, tmp);
}
}
drm/radeon/kms: add pageflip ioctl support (v3) This adds support for dri2 pageflipping. v2: precision updates from Mario Kleiner. v3: Multihead fixes from Mario Kleiner; missing crtc offset add note about update pending bit on pre-avivo chips Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Mario Kleiner <mario.kleiner@tuebingen.mpg.de> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-11-21 23:59:01 +08:00
if (rdev->irq.stat_regs.r600.disp_int_cont & DC_HPD4_INTERRUPT) {
drm/radeon/kms: add hpd support for r6xx/r7xx/rs780/rs880 asics This just adds the functionality, it's not hooked up yet. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-05 04:12:21 +08:00
tmp = RREG32(DC_HPD4_INT_CONTROL);
tmp |= DC_HPDx_INT_ACK;
WREG32(DC_HPD4_INT_CONTROL, tmp);
}
if (ASIC_IS_DCE32(rdev)) {
drm/radeon/kms: add pageflip ioctl support (v3) This adds support for dri2 pageflipping. v2: precision updates from Mario Kleiner. v3: Multihead fixes from Mario Kleiner; missing crtc offset add note about update pending bit on pre-avivo chips Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Mario Kleiner <mario.kleiner@tuebingen.mpg.de> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-11-21 23:59:01 +08:00
if (rdev->irq.stat_regs.r600.disp_int_cont2 & DC_HPD5_INTERRUPT) {
drm/radeon/kms: add hpd support for r6xx/r7xx/rs780/rs880 asics This just adds the functionality, it's not hooked up yet. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-05 04:12:21 +08:00
tmp = RREG32(DC_HPD5_INT_CONTROL);
tmp |= DC_HPDx_INT_ACK;
WREG32(DC_HPD5_INT_CONTROL, tmp);
}
drm/radeon/kms: add pageflip ioctl support (v3) This adds support for dri2 pageflipping. v2: precision updates from Mario Kleiner. v3: Multihead fixes from Mario Kleiner; missing crtc offset add note about update pending bit on pre-avivo chips Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Mario Kleiner <mario.kleiner@tuebingen.mpg.de> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-11-21 23:59:01 +08:00
if (rdev->irq.stat_regs.r600.disp_int_cont2 & DC_HPD6_INTERRUPT) {
drm/radeon/kms: add hpd support for r6xx/r7xx/rs780/rs880 asics This just adds the functionality, it's not hooked up yet. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-05 04:12:21 +08:00
tmp = RREG32(DC_HPD5_INT_CONTROL);
tmp |= DC_HPDx_INT_ACK;
WREG32(DC_HPD6_INT_CONTROL, tmp);
}
drm/radeon/kms: fix up audio interrupt handling - add support for rs6xx - add support for DCE4/5 - fixup 6xx/7xx Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2012-03-30 20:59:57 +08:00
if (rdev->irq.stat_regs.r600.hdmi0_status & AFMT_AZ_FORMAT_WTRIG) {
drm/radeon/kms/hdmi: use relative offsets, official regs Signed-off-by: Rafał Miłecki <zajec5@gmail.com> Tested-by: Christian König <deathsimple@vodafone.de> Reviewed-by: Christian König <deathsimple@vodafone.de> Signed-off-by: Dave Airlie <airlied@redhat.com>
2012-04-29 05:35:24 +08:00
tmp = RREG32(AFMT_AUDIO_PACKET_CONTROL + DCE3_HDMI_OFFSET0);
drm/radeon/kms: fix up audio interrupt handling - add support for rs6xx - add support for DCE4/5 - fixup 6xx/7xx Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2012-03-30 20:59:57 +08:00
tmp |= AFMT_AZ_FORMAT_WTRIG_ACK;
drm/radeon/kms/hdmi: use relative offsets, official regs Signed-off-by: Rafał Miłecki <zajec5@gmail.com> Tested-by: Christian König <deathsimple@vodafone.de> Reviewed-by: Christian König <deathsimple@vodafone.de> Signed-off-by: Dave Airlie <airlied@redhat.com>
2012-04-29 05:35:24 +08:00
WREG32(AFMT_AUDIO_PACKET_CONTROL + DCE3_HDMI_OFFSET0, tmp);
drm/radeon/kms: fix up audio interrupt handling - add support for rs6xx - add support for DCE4/5 - fixup 6xx/7xx Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2012-03-30 20:59:57 +08:00
}
if (rdev->irq.stat_regs.r600.hdmi1_status & AFMT_AZ_FORMAT_WTRIG) {
drm/radeon/kms/hdmi: use relative offsets, official regs Signed-off-by: Rafał Miłecki <zajec5@gmail.com> Tested-by: Christian König <deathsimple@vodafone.de> Reviewed-by: Christian König <deathsimple@vodafone.de> Signed-off-by: Dave Airlie <airlied@redhat.com>
2012-04-29 05:35:24 +08:00
tmp = RREG32(AFMT_AUDIO_PACKET_CONTROL + DCE3_HDMI_OFFSET1);
drm/radeon/kms: fix up audio interrupt handling - add support for rs6xx - add support for DCE4/5 - fixup 6xx/7xx Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2012-03-30 20:59:57 +08:00
tmp |= AFMT_AZ_FORMAT_WTRIG_ACK;
drm/radeon/kms/hdmi: use relative offsets, official regs Signed-off-by: Rafał Miłecki <zajec5@gmail.com> Tested-by: Christian König <deathsimple@vodafone.de> Reviewed-by: Christian König <deathsimple@vodafone.de> Signed-off-by: Dave Airlie <airlied@redhat.com>
2012-04-29 05:35:24 +08:00
WREG32(AFMT_AUDIO_PACKET_CONTROL + DCE3_HDMI_OFFSET1, tmp);
drm/radeon/kms: HDMI irq support Implements irq support for HDMI audio output. Now the polling timer is only enabled if irq support isn't available. Signed-off-by: Christian König <deathsimple@vodafone.de> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-04-10 09:13:16 +08:00
}
} else {
drm/radeon/kms: fix up audio interrupt handling - add support for rs6xx - add support for DCE4/5 - fixup 6xx/7xx Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2012-03-30 20:59:57 +08:00
if (rdev->irq.stat_regs.r600.hdmi0_status & HDMI0_AZ_FORMAT_WTRIG) {
tmp = RREG32(HDMI0_AUDIO_PACKET_CONTROL);
tmp |= HDMI0_AZ_FORMAT_WTRIG_ACK;
WREG32(HDMI0_AUDIO_PACKET_CONTROL, tmp);
}
if (rdev->irq.stat_regs.r600.hdmi1_status & HDMI0_AZ_FORMAT_WTRIG) {
if (ASIC_IS_DCE3(rdev)) {
tmp = RREG32(DCE3_HDMI1_AUDIO_PACKET_CONTROL);
tmp |= HDMI0_AZ_FORMAT_WTRIG_ACK;
WREG32(DCE3_HDMI1_AUDIO_PACKET_CONTROL, tmp);
} else {
tmp = RREG32(HDMI1_AUDIO_PACKET_CONTROL);
tmp |= HDMI0_AZ_FORMAT_WTRIG_ACK;
WREG32(HDMI1_AUDIO_PACKET_CONTROL, tmp);
}
drm/radeon/kms: HDMI irq support Implements irq support for HDMI audio output. Now the polling timer is only enabled if irq support isn't available. Signed-off-by: Christian König <deathsimple@vodafone.de> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-04-10 09:13:16 +08:00
}
}
drm/radeon/kms: Add support for interrupts on r6xx/r7xx chips (v3) This enables the use of interrupts on r6xx/r7xx hardware. Interrupts are implemented via a ring buffer. The GPU adds interrupts vectors to the ring and the host reads them off in the interrupt handler. The interrupt controller requires firmware like the CP. This firmware must be installed and accessble to the firmware loader for interrupts to function. MSIs don't seem to work on my RS780. They work fine on all my discrete cards. I'm not sure about other RS780s or RS880s. I've disabled MSIs on RS780 and RS880, but it would probably be worth checking on some other systems. v2 - fix some checkpatch.pl problems; re-read the disp int status reg if we restart the ih; v3 - remove the irq handler if r600_irq_init() fails; remove spinlock in r600_ih_ring_fini(); move ih rb overflow check to r600_get_ih_wptr(); move irq ack to separate function; Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-02 02:43:46 +08:00
}
void r600_irq_disable(struct radeon_device *rdev)
{
r600_disable_interrupts(rdev);
/* Wait and acknowledge irq */
mdelay(1);
drm/radeon/kms: add pageflip ioctl support (v3) This adds support for dri2 pageflipping. v2: precision updates from Mario Kleiner. v3: Multihead fixes from Mario Kleiner; missing crtc offset add note about update pending bit on pre-avivo chips Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Mario Kleiner <mario.kleiner@tuebingen.mpg.de> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-11-21 23:59:01 +08:00
r600_irq_ack(rdev);
drm/radeon/kms: add hpd support for r6xx/r7xx/rs780/rs880 asics This just adds the functionality, it's not hooked up yet. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-05 04:12:21 +08:00
r600_disable_interrupt_state(rdev);
drm/radeon/kms: Add support for interrupts on r6xx/r7xx chips (v3) This enables the use of interrupts on r6xx/r7xx hardware. Interrupts are implemented via a ring buffer. The GPU adds interrupts vectors to the ring and the host reads them off in the interrupt handler. The interrupt controller requires firmware like the CP. This firmware must be installed and accessble to the firmware loader for interrupts to function. MSIs don't seem to work on my RS780. They work fine on all my discrete cards. I'm not sure about other RS780s or RS880s. I've disabled MSIs on RS780 and RS880, but it would probably be worth checking on some other systems. v2 - fix some checkpatch.pl problems; re-read the disp int status reg if we restart the ih; v3 - remove the irq handler if r600_irq_init() fails; remove spinlock in r600_ih_ring_fini(); move ih rb overflow check to r600_get_ih_wptr(); move irq ack to separate function; Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-02 02:43:46 +08:00
}
drm/radeon: Move more code out of line With this patch I'm only about 50k larger with DRM debugging enables (why is that enabled by default?!?), and slightly smaller without. [airlied: moved r100.c additions to radeon_ring.c] Signed-off-by: Andi Kleen <ak@linux.intel.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-10-14 07:08:47 +08:00
static u32 r600_get_ih_wptr(struct radeon_device *rdev)
drm/radeon/kms: Add support for interrupts on r6xx/r7xx chips (v3) This enables the use of interrupts on r6xx/r7xx hardware. Interrupts are implemented via a ring buffer. The GPU adds interrupts vectors to the ring and the host reads them off in the interrupt handler. The interrupt controller requires firmware like the CP. This firmware must be installed and accessble to the firmware loader for interrupts to function. MSIs don't seem to work on my RS780. They work fine on all my discrete cards. I'm not sure about other RS780s or RS880s. I've disabled MSIs on RS780 and RS880, but it would probably be worth checking on some other systems. v2 - fix some checkpatch.pl problems; re-read the disp int status reg if we restart the ih; v3 - remove the irq handler if r600_irq_init() fails; remove spinlock in r600_ih_ring_fini(); move ih rb overflow check to r600_get_ih_wptr(); move irq ack to separate function; Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-02 02:43:46 +08:00
{
u32 wptr, tmp;
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
drm/radeon/kms: enable writeback (v2) When writeback is enabled, the GPU shadows writes to certain registers into a buffer in memory. The driver can then read the values from the shadow rather than reading back from the register across the bus. Writeback can be disabled by setting the no_wb module param to 1. On r6xx/r7xx/evergreen, the following registers are shadowed: - CP scratch registers - CP read pointer - IH write pointer On r1xx-rr5xx, the following registers are shadowed: - CP scratch registers - CP read pointer v2: - Combine wb patches for r6xx-evergreen and r1xx-r5xx - Writeback is disabled on AGP boards since it tends to be unreliable on AGP using the gart. - Check radeon_wb_init return values properly. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-08-28 06:25:25 +08:00
if (rdev->wb.enabled)
drm/radeon/kms: fix IH writeback on r6xx+ on big endian machines agd5f: fix commit message. Signed-off-by: Cedric Cano <ccano@interfaceconcept.com> Reviewed-by: Michel Dänzer <michel@daenzer.net> Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-04-19 23:07:13 +08:00
wptr = le32_to_cpu(rdev->wb.wb[R600_WB_IH_WPTR_OFFSET/4]);
drm/radeon/kms: enable writeback (v2) When writeback is enabled, the GPU shadows writes to certain registers into a buffer in memory. The driver can then read the values from the shadow rather than reading back from the register across the bus. Writeback can be disabled by setting the no_wb module param to 1. On r6xx/r7xx/evergreen, the following registers are shadowed: - CP scratch registers - CP read pointer - IH write pointer On r1xx-rr5xx, the following registers are shadowed: - CP scratch registers - CP read pointer v2: - Combine wb patches for r6xx-evergreen and r1xx-r5xx - Writeback is disabled on AGP boards since it tends to be unreliable on AGP using the gart. - Check radeon_wb_init return values properly. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-08-28 06:25:25 +08:00
else
wptr = RREG32(IH_RB_WPTR);
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
drm/radeon/kms: Add support for interrupts on r6xx/r7xx chips (v3) This enables the use of interrupts on r6xx/r7xx hardware. Interrupts are implemented via a ring buffer. The GPU adds interrupts vectors to the ring and the host reads them off in the interrupt handler. The interrupt controller requires firmware like the CP. This firmware must be installed and accessble to the firmware loader for interrupts to function. MSIs don't seem to work on my RS780. They work fine on all my discrete cards. I'm not sure about other RS780s or RS880s. I've disabled MSIs on RS780 and RS880, but it would probably be worth checking on some other systems. v2 - fix some checkpatch.pl problems; re-read the disp int status reg if we restart the ih; v3 - remove the irq handler if r600_irq_init() fails; remove spinlock in r600_ih_ring_fini(); move ih rb overflow check to r600_get_ih_wptr(); move irq ack to separate function; Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-02 02:43:46 +08:00
if (wptr & RB_OVERFLOW) {
drm/radeon/kms: r600 handle irq vector ring overflow In some rare case i faced an irq overflow quickly followed by a GPU lockup (hard hang) this patch try to deal with irq vector ring overflow, so far haven't been able to reproduce it with the patch. Signed-off-by: Jerome Glisse <jglisse@redhat.com> Reviewed-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@linux.ie>
2010-01-15 21:44:39 +08:00
/* When a ring buffer overflow happen start parsing interrupt
* from the last not overwritten vector (wptr + 16). Hopefully
* this should allow us to catchup.
*/
dev_warn(rdev->dev, "IH ring buffer overflow (0x%08X, %d, %d)\n",
wptr, rdev->ih.rptr, (wptr + 16) + rdev->ih.ptr_mask);
rdev->ih.rptr = (wptr + 16) & rdev->ih.ptr_mask;
drm/radeon/kms: Add support for interrupts on r6xx/r7xx chips (v3) This enables the use of interrupts on r6xx/r7xx hardware. Interrupts are implemented via a ring buffer. The GPU adds interrupts vectors to the ring and the host reads them off in the interrupt handler. The interrupt controller requires firmware like the CP. This firmware must be installed and accessble to the firmware loader for interrupts to function. MSIs don't seem to work on my RS780. They work fine on all my discrete cards. I'm not sure about other RS780s or RS880s. I've disabled MSIs on RS780 and RS880, but it would probably be worth checking on some other systems. v2 - fix some checkpatch.pl problems; re-read the disp int status reg if we restart the ih; v3 - remove the irq handler if r600_irq_init() fails; remove spinlock in r600_ih_ring_fini(); move ih rb overflow check to r600_get_ih_wptr(); move irq ack to separate function; Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-02 02:43:46 +08:00
tmp = RREG32(IH_RB_CNTL);
tmp |= IH_WPTR_OVERFLOW_CLEAR;
WREG32(IH_RB_CNTL, tmp);
}
drm/radeon/kms: r600/r700 disable irq at suspend To avoid hw doing anythings after we disabled PCIE GART, fully disable IRQ at suspend. Also cleanup a bit the ih structure and process function. Signed-off-by: Jerome Glisse <jglisse@redhat.com> Reviewed-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@linux.ie>
2010-01-15 21:44:37 +08:00
return (wptr & rdev->ih.ptr_mask);
drm/radeon/kms: Add support for interrupts on r6xx/r7xx chips (v3) This enables the use of interrupts on r6xx/r7xx hardware. Interrupts are implemented via a ring buffer. The GPU adds interrupts vectors to the ring and the host reads them off in the interrupt handler. The interrupt controller requires firmware like the CP. This firmware must be installed and accessble to the firmware loader for interrupts to function. MSIs don't seem to work on my RS780. They work fine on all my discrete cards. I'm not sure about other RS780s or RS880s. I've disabled MSIs on RS780 and RS880, but it would probably be worth checking on some other systems. v2 - fix some checkpatch.pl problems; re-read the disp int status reg if we restart the ih; v3 - remove the irq handler if r600_irq_init() fails; remove spinlock in r600_ih_ring_fini(); move ih rb overflow check to r600_get_ih_wptr(); move irq ack to separate function; Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-02 02:43:46 +08:00
}
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
drm/radeon/kms: Add support for interrupts on r6xx/r7xx chips (v3) This enables the use of interrupts on r6xx/r7xx hardware. Interrupts are implemented via a ring buffer. The GPU adds interrupts vectors to the ring and the host reads them off in the interrupt handler. The interrupt controller requires firmware like the CP. This firmware must be installed and accessble to the firmware loader for interrupts to function. MSIs don't seem to work on my RS780. They work fine on all my discrete cards. I'm not sure about other RS780s or RS880s. I've disabled MSIs on RS780 and RS880, but it would probably be worth checking on some other systems. v2 - fix some checkpatch.pl problems; re-read the disp int status reg if we restart the ih; v3 - remove the irq handler if r600_irq_init() fails; remove spinlock in r600_ih_ring_fini(); move ih rb overflow check to r600_get_ih_wptr(); move irq ack to separate function; Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-02 02:43:46 +08:00
/* r600 IV Ring
* Each IV ring entry is 128 bits:
* [7:0] - interrupt source id
* [31:8] - reserved
* [59:32] - interrupt source data
* [127:60] - reserved
*
* The basic interrupt vector entries
* are decoded as follows:
* src_id src_data description
* 1 0 D1 Vblank
* 1 1 D1 Vline
* 5 0 D2 Vblank
* 5 1 D2 Vline
* 19 0 FP Hot plug detection A
* 19 1 FP Hot plug detection B
* 19 2 DAC A auto-detection
* 19 3 DAC B auto-detection
drm/radeon/kms: HDMI irq support Implements irq support for HDMI audio output. Now the polling timer is only enabled if irq support isn't available. Signed-off-by: Christian König <deathsimple@vodafone.de> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-04-10 09:13:16 +08:00
* 21 4 HDMI block A
* 21 5 HDMI block B
drm/radeon/kms: Add support for interrupts on r6xx/r7xx chips (v3) This enables the use of interrupts on r6xx/r7xx hardware. Interrupts are implemented via a ring buffer. The GPU adds interrupts vectors to the ring and the host reads them off in the interrupt handler. The interrupt controller requires firmware like the CP. This firmware must be installed and accessble to the firmware loader for interrupts to function. MSIs don't seem to work on my RS780. They work fine on all my discrete cards. I'm not sure about other RS780s or RS880s. I've disabled MSIs on RS780 and RS880, but it would probably be worth checking on some other systems. v2 - fix some checkpatch.pl problems; re-read the disp int status reg if we restart the ih; v3 - remove the irq handler if r600_irq_init() fails; remove spinlock in r600_ih_ring_fini(); move ih rb overflow check to r600_get_ih_wptr(); move irq ack to separate function; Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-02 02:43:46 +08:00
* 176 - CP_INT RB
* 177 - CP_INT IB1
* 178 - CP_INT IB2
* 181 - EOP Interrupt
* 233 - GUI Idle
*
* Note, these are based on r600 and may need to be
* adjusted or added to on newer asics
*/
int r600_irq_process(struct radeon_device *rdev)
{
drm/radeon: avoid warnings from r600/eg irq handlers on powered off card. Since we were calling the wptr function before checking if the IH was even enabled, or the GPU wasn't shutdown, we'd get spam in the logs when the GPU readback 0xffffffff. This reorders things so we return early in the no IH and GPU shutdown cases. Reported-and-tested-by: ManDay on #radeon Reviewed-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-06-18 11:59:51 +08:00
u32 wptr;
u32 rptr;
drm/radeon/kms: Add support for interrupts on r6xx/r7xx chips (v3) This enables the use of interrupts on r6xx/r7xx hardware. Interrupts are implemented via a ring buffer. The GPU adds interrupts vectors to the ring and the host reads them off in the interrupt handler. The interrupt controller requires firmware like the CP. This firmware must be installed and accessble to the firmware loader for interrupts to function. MSIs don't seem to work on my RS780. They work fine on all my discrete cards. I'm not sure about other RS780s or RS880s. I've disabled MSIs on RS780 and RS880, but it would probably be worth checking on some other systems. v2 - fix some checkpatch.pl problems; re-read the disp int status reg if we restart the ih; v3 - remove the irq handler if r600_irq_init() fails; remove spinlock in r600_ih_ring_fini(); move ih rb overflow check to r600_get_ih_wptr(); move irq ack to separate function; Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-02 02:43:46 +08:00
u32 src_id, src_data;
drm/radeon/kms: add pageflip ioctl support (v3) This adds support for dri2 pageflipping. v2: precision updates from Mario Kleiner. v3: Multihead fixes from Mario Kleiner; missing crtc offset add note about update pending bit on pre-avivo chips Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Mario Kleiner <mario.kleiner@tuebingen.mpg.de> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-11-21 23:59:01 +08:00
u32 ring_index;
drm/radeon/kms: enable hpd support This enabled interrupt driven hpd support for all radeon chips. Assuming the hpd pin is wired up correctly, the driver will generate uevents on digital monitor connect and disconnect and retrain DP monitors automatically. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-05 05:56:37 +08:00
bool queue_hotplug = false;
drm/radeon/kms: fix up audio interrupt handling - add support for rs6xx - add support for DCE4/5 - fixup 6xx/7xx Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2012-03-30 20:59:57 +08:00
bool queue_hdmi = false;
drm/radeon/kms: Add support for interrupts on r6xx/r7xx chips (v3) This enables the use of interrupts on r6xx/r7xx hardware. Interrupts are implemented via a ring buffer. The GPU adds interrupts vectors to the ring and the host reads them off in the interrupt handler. The interrupt controller requires firmware like the CP. This firmware must be installed and accessble to the firmware loader for interrupts to function. MSIs don't seem to work on my RS780. They work fine on all my discrete cards. I'm not sure about other RS780s or RS880s. I've disabled MSIs on RS780 and RS880, but it would probably be worth checking on some other systems. v2 - fix some checkpatch.pl problems; re-read the disp int status reg if we restart the ih; v3 - remove the irq handler if r600_irq_init() fails; remove spinlock in r600_ih_ring_fini(); move ih rb overflow check to r600_get_ih_wptr(); move irq ack to separate function; Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-02 02:43:46 +08:00
drm/radeon: avoid warnings from r600/eg irq handlers on powered off card. Since we were calling the wptr function before checking if the IH was even enabled, or the GPU wasn't shutdown, we'd get spam in the logs when the GPU readback 0xffffffff. This reorders things so we return early in the no IH and GPU shutdown cases. Reported-and-tested-by: ManDay on #radeon Reviewed-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-06-18 11:59:51 +08:00
if (!rdev->ih.enabled || rdev->shutdown)
drm/radeon/kms: r600/r700 don't process IRQ if not initialized In some rare case the wptr returned from the hw wasn't 0 and leaded to trick r600_process_irq that their were irq to process. Add a check to bail out if irq hasn't been initialized this will avoid oops provoqued by the rare wptr != 0 on initialization. Signed-off-by: Jerome Glisse <jglisse@redhat.com> Reviewed-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@linux.ie>
2010-01-15 21:44:38 +08:00
return IRQ_NONE;
drm/radeon/kms: Add support for interrupts on r6xx/r7xx chips (v3) This enables the use of interrupts on r6xx/r7xx hardware. Interrupts are implemented via a ring buffer. The GPU adds interrupts vectors to the ring and the host reads them off in the interrupt handler. The interrupt controller requires firmware like the CP. This firmware must be installed and accessble to the firmware loader for interrupts to function. MSIs don't seem to work on my RS780. They work fine on all my discrete cards. I'm not sure about other RS780s or RS880s. I've disabled MSIs on RS780 and RS880, but it would probably be worth checking on some other systems. v2 - fix some checkpatch.pl problems; re-read the disp int status reg if we restart the ih; v3 - remove the irq handler if r600_irq_init() fails; remove spinlock in r600_ih_ring_fini(); move ih rb overflow check to r600_get_ih_wptr(); move irq ack to separate function; Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-02 02:43:46 +08:00
drm/radeon: Do an MMIO read on interrupts when not uisng MSIs When not using MSIs, there is no guarantee that DMA from the device has been fully flushed to point where it's visible to the CPU when taking an interrupt. To get this guarantee, we need to perform an MMIO read from the device, which will flush all outstanding DMAs from bridges between the device and the system. Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-07-13 14:28:22 +08:00
/* No MSIs, need a dummy read to flush PCI DMAs */
if (!rdev->msi_enabled)
RREG32(IH_RB_WPTR);
drm/radeon: avoid warnings from r600/eg irq handlers on powered off card. Since we were calling the wptr function before checking if the IH was even enabled, or the GPU wasn't shutdown, we'd get spam in the logs when the GPU readback 0xffffffff. This reorders things so we return early in the no IH and GPU shutdown cases. Reported-and-tested-by: ManDay on #radeon Reviewed-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-06-18 11:59:51 +08:00
wptr = r600_get_ih_wptr(rdev);
drm/radeon/kms: Add support for interrupts on r6xx/r7xx chips (v3) This enables the use of interrupts on r6xx/r7xx hardware. Interrupts are implemented via a ring buffer. The GPU adds interrupts vectors to the ring and the host reads them off in the interrupt handler. The interrupt controller requires firmware like the CP. This firmware must be installed and accessble to the firmware loader for interrupts to function. MSIs don't seem to work on my RS780. They work fine on all my discrete cards. I'm not sure about other RS780s or RS880s. I've disabled MSIs on RS780 and RS880, but it would probably be worth checking on some other systems. v2 - fix some checkpatch.pl problems; re-read the disp int status reg if we restart the ih; v3 - remove the irq handler if r600_irq_init() fails; remove spinlock in r600_ih_ring_fini(); move ih rb overflow check to r600_get_ih_wptr(); move irq ack to separate function; Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-02 02:43:46 +08:00
drm/radeon: fix & improve ih ring handling v3 The spinlock was actually there to protect the rptr, but rptr was read outside of the locked area. Also we don't really need a spinlock here, an atomic should to quite fine since we only need to prevent it from being reentrant. v2: Keep the spinlock.... v3: Back to an atomic again after finding & fixing the real bug. Signed-off-by: Christian Koenig <christian.koenig@amd.com>
2012-05-17 03:45:24 +08:00
restart_ih:
/* is somebody else already processing irqs? */
if (atomic_xchg(&rdev->ih.lock, 1))
drm/radeon/kms: Add support for interrupts on r6xx/r7xx chips (v3) This enables the use of interrupts on r6xx/r7xx hardware. Interrupts are implemented via a ring buffer. The GPU adds interrupts vectors to the ring and the host reads them off in the interrupt handler. The interrupt controller requires firmware like the CP. This firmware must be installed and accessble to the firmware loader for interrupts to function. MSIs don't seem to work on my RS780. They work fine on all my discrete cards. I'm not sure about other RS780s or RS880s. I've disabled MSIs on RS780 and RS880, but it would probably be worth checking on some other systems. v2 - fix some checkpatch.pl problems; re-read the disp int status reg if we restart the ih; v3 - remove the irq handler if r600_irq_init() fails; remove spinlock in r600_ih_ring_fini(); move ih rb overflow check to r600_get_ih_wptr(); move irq ack to separate function; Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-02 02:43:46 +08:00
return IRQ_NONE;
drm/radeon: fix & improve ih ring handling v3 The spinlock was actually there to protect the rptr, but rptr was read outside of the locked area. Also we don't really need a spinlock here, an atomic should to quite fine since we only need to prevent it from being reentrant. v2: Keep the spinlock.... v3: Back to an atomic again after finding & fixing the real bug. Signed-off-by: Christian Koenig <christian.koenig@amd.com>
2012-05-17 03:45:24 +08:00
rptr = rdev->ih.rptr;
DRM_DEBUG("r600_irq_process start: rptr %d, wptr %d\n", rptr, wptr);
drm/radeon: Add a rmb() in IH processing We should have a read memory barrier between reading the WPTR from memory and reading ring entries based on that value (ie, we need to ensure both loads are done in order by the CPU). It could be argued that the MMIO reads in r600_ack_irq() might be enough to get that barrier but I prefer keeping an explicit one just in case. [airlied: fix evergreen + r/w mixup] Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org> Reviewed-by: Alex Deucher <alexander.deucher@amd.com> Reviewed-by: Matt Turner <mattst88@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-07-13 14:28:19 +08:00
/* Order reading of wptr vs. reading of IH ring data */
rmb();
drm/radeon/kms: Add support for interrupts on r6xx/r7xx chips (v3) This enables the use of interrupts on r6xx/r7xx hardware. Interrupts are implemented via a ring buffer. The GPU adds interrupts vectors to the ring and the host reads them off in the interrupt handler. The interrupt controller requires firmware like the CP. This firmware must be installed and accessble to the firmware loader for interrupts to function. MSIs don't seem to work on my RS780. They work fine on all my discrete cards. I'm not sure about other RS780s or RS880s. I've disabled MSIs on RS780 and RS880, but it would probably be worth checking on some other systems. v2 - fix some checkpatch.pl problems; re-read the disp int status reg if we restart the ih; v3 - remove the irq handler if r600_irq_init() fails; remove spinlock in r600_ih_ring_fini(); move ih rb overflow check to r600_get_ih_wptr(); move irq ack to separate function; Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-02 02:43:46 +08:00
/* display interrupts */
drm/radeon/kms: add pageflip ioctl support (v3) This adds support for dri2 pageflipping. v2: precision updates from Mario Kleiner. v3: Multihead fixes from Mario Kleiner; missing crtc offset add note about update pending bit on pre-avivo chips Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Mario Kleiner <mario.kleiner@tuebingen.mpg.de> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-11-21 23:59:01 +08:00
r600_irq_ack(rdev);
drm/radeon/kms: Add support for interrupts on r6xx/r7xx chips (v3) This enables the use of interrupts on r6xx/r7xx hardware. Interrupts are implemented via a ring buffer. The GPU adds interrupts vectors to the ring and the host reads them off in the interrupt handler. The interrupt controller requires firmware like the CP. This firmware must be installed and accessble to the firmware loader for interrupts to function. MSIs don't seem to work on my RS780. They work fine on all my discrete cards. I'm not sure about other RS780s or RS880s. I've disabled MSIs on RS780 and RS880, but it would probably be worth checking on some other systems. v2 - fix some checkpatch.pl problems; re-read the disp int status reg if we restart the ih; v3 - remove the irq handler if r600_irq_init() fails; remove spinlock in r600_ih_ring_fini(); move ih rb overflow check to r600_get_ih_wptr(); move irq ack to separate function; Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-02 02:43:46 +08:00
while (rptr != wptr) {
/* wptr/rptr are in bytes! */
ring_index = rptr / 4;
drm/radeon/kms: 6xx/7xx big endian fixes agd5f: minor cleanups Signed-off-by: Cédric Cano <ccano@interfaceconcept.com> Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-02-12 08:45:38 +08:00
src_id = le32_to_cpu(rdev->ih.ring[ring_index]) & 0xff;
src_data = le32_to_cpu(rdev->ih.ring[ring_index + 1]) & 0xfffffff;
drm/radeon/kms: Add support for interrupts on r6xx/r7xx chips (v3) This enables the use of interrupts on r6xx/r7xx hardware. Interrupts are implemented via a ring buffer. The GPU adds interrupts vectors to the ring and the host reads them off in the interrupt handler. The interrupt controller requires firmware like the CP. This firmware must be installed and accessble to the firmware loader for interrupts to function. MSIs don't seem to work on my RS780. They work fine on all my discrete cards. I'm not sure about other RS780s or RS880s. I've disabled MSIs on RS780 and RS880, but it would probably be worth checking on some other systems. v2 - fix some checkpatch.pl problems; re-read the disp int status reg if we restart the ih; v3 - remove the irq handler if r600_irq_init() fails; remove spinlock in r600_ih_ring_fini(); move ih rb overflow check to r600_get_ih_wptr(); move irq ack to separate function; Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-02 02:43:46 +08:00
switch (src_id) {
case 1: /* D1 vblank/vline */
switch (src_data) {
case 0: /* D1 vblank */
drm/radeon/kms: add pageflip ioctl support (v3) This adds support for dri2 pageflipping. v2: precision updates from Mario Kleiner. v3: Multihead fixes from Mario Kleiner; missing crtc offset add note about update pending bit on pre-avivo chips Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Mario Kleiner <mario.kleiner@tuebingen.mpg.de> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-11-21 23:59:01 +08:00
if (rdev->irq.stat_regs.r600.disp_int & LB_D1_VBLANK_INTERRUPT) {
if (rdev->irq.crtc_vblank_int[0]) {
drm_handle_vblank(rdev->ddev, 0);
rdev->pm.vblank_sync = true;
wake_up(&rdev->irq.vblank_queue);
}
drm/radeon: replace pflip and sw_int counters with atomics So we can skip the locking. Also renames sw_int to ring_int, cause that better matches its purpose. Signed-off-by: Christian Koenig <christian.koenig@amd.com>
2012-05-18 01:52:00 +08:00
if (atomic_read(&rdev->irq.pflip[0]))
drm/kms/radeon: Reorder vblank and pageflip interrupt handling. In the vblank irq handler, calls to actual vblank handling, or at least drm_handle_vblank(), need to happen before calls to radeon_crtc_handle_flip(). Reason: The high precision pageflip timestamping and some other pageflip optimizations will need the updated vblank count and timestamps for the current vblank interval. These are calculated in drm_handle_vblank(), therefore it must go first. Signed-off-by: Mario Kleiner <mario.kleiner@tuebingen.mpg.de> Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-11-21 23:59:02 +08:00
radeon_crtc_handle_flip(rdev, 0);
drm/radeon/kms: add pageflip ioctl support (v3) This adds support for dri2 pageflipping. v2: precision updates from Mario Kleiner. v3: Multihead fixes from Mario Kleiner; missing crtc offset add note about update pending bit on pre-avivo chips Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Mario Kleiner <mario.kleiner@tuebingen.mpg.de> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-11-21 23:59:01 +08:00
rdev->irq.stat_regs.r600.disp_int &= ~LB_D1_VBLANK_INTERRUPT;
drm/radeon/kms: Add support for interrupts on r6xx/r7xx chips (v3) This enables the use of interrupts on r6xx/r7xx hardware. Interrupts are implemented via a ring buffer. The GPU adds interrupts vectors to the ring and the host reads them off in the interrupt handler. The interrupt controller requires firmware like the CP. This firmware must be installed and accessble to the firmware loader for interrupts to function. MSIs don't seem to work on my RS780. They work fine on all my discrete cards. I'm not sure about other RS780s or RS880s. I've disabled MSIs on RS780 and RS880, but it would probably be worth checking on some other systems. v2 - fix some checkpatch.pl problems; re-read the disp int status reg if we restart the ih; v3 - remove the irq handler if r600_irq_init() fails; remove spinlock in r600_ih_ring_fini(); move ih rb overflow check to r600_get_ih_wptr(); move irq ack to separate function; Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-02 02:43:46 +08:00
DRM_DEBUG("IH: D1 vblank\n");
}
break;
case 1: /* D1 vline */
drm/radeon/kms: add pageflip ioctl support (v3) This adds support for dri2 pageflipping. v2: precision updates from Mario Kleiner. v3: Multihead fixes from Mario Kleiner; missing crtc offset add note about update pending bit on pre-avivo chips Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Mario Kleiner <mario.kleiner@tuebingen.mpg.de> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-11-21 23:59:01 +08:00
if (rdev->irq.stat_regs.r600.disp_int & LB_D1_VLINE_INTERRUPT) {
rdev->irq.stat_regs.r600.disp_int &= ~LB_D1_VLINE_INTERRUPT;
drm/radeon/kms: Add support for interrupts on r6xx/r7xx chips (v3) This enables the use of interrupts on r6xx/r7xx hardware. Interrupts are implemented via a ring buffer. The GPU adds interrupts vectors to the ring and the host reads them off in the interrupt handler. The interrupt controller requires firmware like the CP. This firmware must be installed and accessble to the firmware loader for interrupts to function. MSIs don't seem to work on my RS780. They work fine on all my discrete cards. I'm not sure about other RS780s or RS880s. I've disabled MSIs on RS780 and RS880, but it would probably be worth checking on some other systems. v2 - fix some checkpatch.pl problems; re-read the disp int status reg if we restart the ih; v3 - remove the irq handler if r600_irq_init() fails; remove spinlock in r600_ih_ring_fini(); move ih rb overflow check to r600_get_ih_wptr(); move irq ack to separate function; Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-02 02:43:46 +08:00
DRM_DEBUG("IH: D1 vline\n");
}
break;
default:
drm/radeon/kms/r6xx+: make irq handler less verbose Unhandled vectors can be safely ignored, no need to spam the kernel log by default. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-01-12 08:47:38 +08:00
DRM_DEBUG("Unhandled interrupt: %d %d\n", src_id, src_data);
drm/radeon/kms: Add support for interrupts on r6xx/r7xx chips (v3) This enables the use of interrupts on r6xx/r7xx hardware. Interrupts are implemented via a ring buffer. The GPU adds interrupts vectors to the ring and the host reads them off in the interrupt handler. The interrupt controller requires firmware like the CP. This firmware must be installed and accessble to the firmware loader for interrupts to function. MSIs don't seem to work on my RS780. They work fine on all my discrete cards. I'm not sure about other RS780s or RS880s. I've disabled MSIs on RS780 and RS880, but it would probably be worth checking on some other systems. v2 - fix some checkpatch.pl problems; re-read the disp int status reg if we restart the ih; v3 - remove the irq handler if r600_irq_init() fails; remove spinlock in r600_ih_ring_fini(); move ih rb overflow check to r600_get_ih_wptr(); move irq ack to separate function; Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-02 02:43:46 +08:00
break;
}
break;
case 5: /* D2 vblank/vline */
switch (src_data) {
case 0: /* D2 vblank */
drm/radeon/kms: add pageflip ioctl support (v3) This adds support for dri2 pageflipping. v2: precision updates from Mario Kleiner. v3: Multihead fixes from Mario Kleiner; missing crtc offset add note about update pending bit on pre-avivo chips Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Mario Kleiner <mario.kleiner@tuebingen.mpg.de> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-11-21 23:59:01 +08:00
if (rdev->irq.stat_regs.r600.disp_int & LB_D2_VBLANK_INTERRUPT) {
if (rdev->irq.crtc_vblank_int[1]) {
drm_handle_vblank(rdev->ddev, 1);
rdev->pm.vblank_sync = true;
wake_up(&rdev->irq.vblank_queue);
}
drm/radeon: replace pflip and sw_int counters with atomics So we can skip the locking. Also renames sw_int to ring_int, cause that better matches its purpose. Signed-off-by: Christian Koenig <christian.koenig@amd.com>
2012-05-18 01:52:00 +08:00
if (atomic_read(&rdev->irq.pflip[1]))
drm/kms/radeon: Reorder vblank and pageflip interrupt handling. In the vblank irq handler, calls to actual vblank handling, or at least drm_handle_vblank(), need to happen before calls to radeon_crtc_handle_flip(). Reason: The high precision pageflip timestamping and some other pageflip optimizations will need the updated vblank count and timestamps for the current vblank interval. These are calculated in drm_handle_vblank(), therefore it must go first. Signed-off-by: Mario Kleiner <mario.kleiner@tuebingen.mpg.de> Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-11-21 23:59:02 +08:00
radeon_crtc_handle_flip(rdev, 1);
drm/radeon/kms: add pageflip ioctl support (v3) This adds support for dri2 pageflipping. v2: precision updates from Mario Kleiner. v3: Multihead fixes from Mario Kleiner; missing crtc offset add note about update pending bit on pre-avivo chips Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Mario Kleiner <mario.kleiner@tuebingen.mpg.de> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-11-21 23:59:01 +08:00
rdev->irq.stat_regs.r600.disp_int &= ~LB_D2_VBLANK_INTERRUPT;
drm/radeon/kms: Add support for interrupts on r6xx/r7xx chips (v3) This enables the use of interrupts on r6xx/r7xx hardware. Interrupts are implemented via a ring buffer. The GPU adds interrupts vectors to the ring and the host reads them off in the interrupt handler. The interrupt controller requires firmware like the CP. This firmware must be installed and accessble to the firmware loader for interrupts to function. MSIs don't seem to work on my RS780. They work fine on all my discrete cards. I'm not sure about other RS780s or RS880s. I've disabled MSIs on RS780 and RS880, but it would probably be worth checking on some other systems. v2 - fix some checkpatch.pl problems; re-read the disp int status reg if we restart the ih; v3 - remove the irq handler if r600_irq_init() fails; remove spinlock in r600_ih_ring_fini(); move ih rb overflow check to r600_get_ih_wptr(); move irq ack to separate function; Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-02 02:43:46 +08:00
DRM_DEBUG("IH: D2 vblank\n");
}
break;
case 1: /* D1 vline */
drm/radeon/kms: add pageflip ioctl support (v3) This adds support for dri2 pageflipping. v2: precision updates from Mario Kleiner. v3: Multihead fixes from Mario Kleiner; missing crtc offset add note about update pending bit on pre-avivo chips Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Mario Kleiner <mario.kleiner@tuebingen.mpg.de> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-11-21 23:59:01 +08:00
if (rdev->irq.stat_regs.r600.disp_int & LB_D2_VLINE_INTERRUPT) {
rdev->irq.stat_regs.r600.disp_int &= ~LB_D2_VLINE_INTERRUPT;
drm/radeon/kms: Add support for interrupts on r6xx/r7xx chips (v3) This enables the use of interrupts on r6xx/r7xx hardware. Interrupts are implemented via a ring buffer. The GPU adds interrupts vectors to the ring and the host reads them off in the interrupt handler. The interrupt controller requires firmware like the CP. This firmware must be installed and accessble to the firmware loader for interrupts to function. MSIs don't seem to work on my RS780. They work fine on all my discrete cards. I'm not sure about other RS780s or RS880s. I've disabled MSIs on RS780 and RS880, but it would probably be worth checking on some other systems. v2 - fix some checkpatch.pl problems; re-read the disp int status reg if we restart the ih; v3 - remove the irq handler if r600_irq_init() fails; remove spinlock in r600_ih_ring_fini(); move ih rb overflow check to r600_get_ih_wptr(); move irq ack to separate function; Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-02 02:43:46 +08:00
DRM_DEBUG("IH: D2 vline\n");
}
break;
default:
drm/radeon/kms/r6xx+: make irq handler less verbose Unhandled vectors can be safely ignored, no need to spam the kernel log by default. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-01-12 08:47:38 +08:00
DRM_DEBUG("Unhandled interrupt: %d %d\n", src_id, src_data);
drm/radeon/kms: Add support for interrupts on r6xx/r7xx chips (v3) This enables the use of interrupts on r6xx/r7xx hardware. Interrupts are implemented via a ring buffer. The GPU adds interrupts vectors to the ring and the host reads them off in the interrupt handler. The interrupt controller requires firmware like the CP. This firmware must be installed and accessble to the firmware loader for interrupts to function. MSIs don't seem to work on my RS780. They work fine on all my discrete cards. I'm not sure about other RS780s or RS880s. I've disabled MSIs on RS780 and RS880, but it would probably be worth checking on some other systems. v2 - fix some checkpatch.pl problems; re-read the disp int status reg if we restart the ih; v3 - remove the irq handler if r600_irq_init() fails; remove spinlock in r600_ih_ring_fini(); move ih rb overflow check to r600_get_ih_wptr(); move irq ack to separate function; Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-02 02:43:46 +08:00
break;
}
break;
drm/radeon/kms: add hpd support for r6xx/r7xx/rs780/rs880 asics This just adds the functionality, it's not hooked up yet. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-05 04:12:21 +08:00
case 19: /* HPD/DAC hotplug */
switch (src_data) {
case 0:
drm/radeon/kms: add pageflip ioctl support (v3) This adds support for dri2 pageflipping. v2: precision updates from Mario Kleiner. v3: Multihead fixes from Mario Kleiner; missing crtc offset add note about update pending bit on pre-avivo chips Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Mario Kleiner <mario.kleiner@tuebingen.mpg.de> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-11-21 23:59:01 +08:00
if (rdev->irq.stat_regs.r600.disp_int & DC_HPD1_INTERRUPT) {
rdev->irq.stat_regs.r600.disp_int &= ~DC_HPD1_INTERRUPT;
drm/radeon/kms: enable hpd support This enabled interrupt driven hpd support for all radeon chips. Assuming the hpd pin is wired up correctly, the driver will generate uevents on digital monitor connect and disconnect and retrain DP monitors automatically. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-05 05:56:37 +08:00
queue_hotplug = true;
DRM_DEBUG("IH: HPD1\n");
drm/radeon/kms: add hpd support for r6xx/r7xx/rs780/rs880 asics This just adds the functionality, it's not hooked up yet. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-05 04:12:21 +08:00
}
break;
case 1:
drm/radeon/kms: add pageflip ioctl support (v3) This adds support for dri2 pageflipping. v2: precision updates from Mario Kleiner. v3: Multihead fixes from Mario Kleiner; missing crtc offset add note about update pending bit on pre-avivo chips Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Mario Kleiner <mario.kleiner@tuebingen.mpg.de> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-11-21 23:59:01 +08:00
if (rdev->irq.stat_regs.r600.disp_int & DC_HPD2_INTERRUPT) {
rdev->irq.stat_regs.r600.disp_int &= ~DC_HPD2_INTERRUPT;
drm/radeon/kms: enable hpd support This enabled interrupt driven hpd support for all radeon chips. Assuming the hpd pin is wired up correctly, the driver will generate uevents on digital monitor connect and disconnect and retrain DP monitors automatically. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-05 05:56:37 +08:00
queue_hotplug = true;
DRM_DEBUG("IH: HPD2\n");
drm/radeon/kms: add hpd support for r6xx/r7xx/rs780/rs880 asics This just adds the functionality, it's not hooked up yet. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-05 04:12:21 +08:00
}
break;
case 4:
drm/radeon/kms: add pageflip ioctl support (v3) This adds support for dri2 pageflipping. v2: precision updates from Mario Kleiner. v3: Multihead fixes from Mario Kleiner; missing crtc offset add note about update pending bit on pre-avivo chips Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Mario Kleiner <mario.kleiner@tuebingen.mpg.de> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-11-21 23:59:01 +08:00
if (rdev->irq.stat_regs.r600.disp_int_cont & DC_HPD3_INTERRUPT) {
rdev->irq.stat_regs.r600.disp_int_cont &= ~DC_HPD3_INTERRUPT;
drm/radeon/kms: enable hpd support This enabled interrupt driven hpd support for all radeon chips. Assuming the hpd pin is wired up correctly, the driver will generate uevents on digital monitor connect and disconnect and retrain DP monitors automatically. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-05 05:56:37 +08:00
queue_hotplug = true;
DRM_DEBUG("IH: HPD3\n");
drm/radeon/kms: add hpd support for r6xx/r7xx/rs780/rs880 asics This just adds the functionality, it's not hooked up yet. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-05 04:12:21 +08:00
}
break;
case 5:
drm/radeon/kms: add pageflip ioctl support (v3) This adds support for dri2 pageflipping. v2: precision updates from Mario Kleiner. v3: Multihead fixes from Mario Kleiner; missing crtc offset add note about update pending bit on pre-avivo chips Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Mario Kleiner <mario.kleiner@tuebingen.mpg.de> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-11-21 23:59:01 +08:00
if (rdev->irq.stat_regs.r600.disp_int_cont & DC_HPD4_INTERRUPT) {
rdev->irq.stat_regs.r600.disp_int_cont &= ~DC_HPD4_INTERRUPT;
drm/radeon/kms: enable hpd support This enabled interrupt driven hpd support for all radeon chips. Assuming the hpd pin is wired up correctly, the driver will generate uevents on digital monitor connect and disconnect and retrain DP monitors automatically. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-05 05:56:37 +08:00
queue_hotplug = true;
DRM_DEBUG("IH: HPD4\n");
drm/radeon/kms: add hpd support for r6xx/r7xx/rs780/rs880 asics This just adds the functionality, it's not hooked up yet. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-05 04:12:21 +08:00
}
break;
case 10:
drm/radeon/kms: add pageflip ioctl support (v3) This adds support for dri2 pageflipping. v2: precision updates from Mario Kleiner. v3: Multihead fixes from Mario Kleiner; missing crtc offset add note about update pending bit on pre-avivo chips Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Mario Kleiner <mario.kleiner@tuebingen.mpg.de> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-11-21 23:59:01 +08:00
if (rdev->irq.stat_regs.r600.disp_int_cont2 & DC_HPD5_INTERRUPT) {
rdev->irq.stat_regs.r600.disp_int_cont2 &= ~DC_HPD5_INTERRUPT;
drm/radeon/kms: enable hpd support This enabled interrupt driven hpd support for all radeon chips. Assuming the hpd pin is wired up correctly, the driver will generate uevents on digital monitor connect and disconnect and retrain DP monitors automatically. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-05 05:56:37 +08:00
queue_hotplug = true;
DRM_DEBUG("IH: HPD5\n");
drm/radeon/kms: add hpd support for r6xx/r7xx/rs780/rs880 asics This just adds the functionality, it's not hooked up yet. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-05 04:12:21 +08:00
}
break;
case 12:
drm/radeon/kms: add pageflip ioctl support (v3) This adds support for dri2 pageflipping. v2: precision updates from Mario Kleiner. v3: Multihead fixes from Mario Kleiner; missing crtc offset add note about update pending bit on pre-avivo chips Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Mario Kleiner <mario.kleiner@tuebingen.mpg.de> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-11-21 23:59:01 +08:00
if (rdev->irq.stat_regs.r600.disp_int_cont2 & DC_HPD6_INTERRUPT) {
rdev->irq.stat_regs.r600.disp_int_cont2 &= ~DC_HPD6_INTERRUPT;
drm/radeon/kms: enable hpd support This enabled interrupt driven hpd support for all radeon chips. Assuming the hpd pin is wired up correctly, the driver will generate uevents on digital monitor connect and disconnect and retrain DP monitors automatically. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-05 05:56:37 +08:00
queue_hotplug = true;
DRM_DEBUG("IH: HPD6\n");
drm/radeon/kms: add hpd support for r6xx/r7xx/rs780/rs880 asics This just adds the functionality, it's not hooked up yet. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-05 04:12:21 +08:00
}
break;
default:
drm/radeon/kms/r6xx+: make irq handler less verbose Unhandled vectors can be safely ignored, no need to spam the kernel log by default. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-01-12 08:47:38 +08:00
DRM_DEBUG("Unhandled interrupt: %d %d\n", src_id, src_data);
drm/radeon/kms: add hpd support for r6xx/r7xx/rs780/rs880 asics This just adds the functionality, it's not hooked up yet. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-05 04:12:21 +08:00
break;
}
break;
drm/radeon/kms: fix up audio interrupt handling - add support for rs6xx - add support for DCE4/5 - fixup 6xx/7xx Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2012-03-30 20:59:57 +08:00
case 21: /* hdmi */
switch (src_data) {
case 4:
if (rdev->irq.stat_regs.r600.hdmi0_status & HDMI0_AZ_FORMAT_WTRIG) {
rdev->irq.stat_regs.r600.hdmi0_status &= ~HDMI0_AZ_FORMAT_WTRIG;
queue_hdmi = true;
DRM_DEBUG("IH: HDMI0\n");
}
break;
case 5:
if (rdev->irq.stat_regs.r600.hdmi1_status & HDMI0_AZ_FORMAT_WTRIG) {
rdev->irq.stat_regs.r600.hdmi1_status &= ~HDMI0_AZ_FORMAT_WTRIG;
queue_hdmi = true;
DRM_DEBUG("IH: HDMI1\n");
}
break;
default:
DRM_ERROR("Unhandled interrupt: %d %d\n", src_id, src_data);
break;
}
drm/radeon/kms: HDMI irq support Implements irq support for HDMI audio output. Now the polling timer is only enabled if irq support isn't available. Signed-off-by: Christian König <deathsimple@vodafone.de> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-04-10 09:13:16 +08:00
break;
drm/radeon/kms: Add support for interrupts on r6xx/r7xx chips (v3) This enables the use of interrupts on r6xx/r7xx hardware. Interrupts are implemented via a ring buffer. The GPU adds interrupts vectors to the ring and the host reads them off in the interrupt handler. The interrupt controller requires firmware like the CP. This firmware must be installed and accessble to the firmware loader for interrupts to function. MSIs don't seem to work on my RS780. They work fine on all my discrete cards. I'm not sure about other RS780s or RS880s. I've disabled MSIs on RS780 and RS880, but it would probably be worth checking on some other systems. v2 - fix some checkpatch.pl problems; re-read the disp int status reg if we restart the ih; v3 - remove the irq handler if r600_irq_init() fails; remove spinlock in r600_ih_ring_fini(); move ih rb overflow check to r600_get_ih_wptr(); move irq ack to separate function; Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-02 02:43:46 +08:00
case 176: /* CP_INT in ring buffer */
case 177: /* CP_INT in IB1 */
case 178: /* CP_INT in IB2 */
DRM_DEBUG("IH: CP int: 0x%08x\n", src_data);
drm/radeon/kms: add support for multiple fence queues v2 For supporting multiple CP ring buffers, async DMA engines and UVD. We still need a way to synchronize between engines. v2 initialize unused fence driver ring to avoid issue in suspend/unload Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Christian König <deathsimple@vodafone.de> Reviewed-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-08-26 01:39:48 +08:00
radeon_fence_process(rdev, RADEON_RING_TYPE_GFX_INDEX);
drm/radeon/kms: Add support for interrupts on r6xx/r7xx chips (v3) This enables the use of interrupts on r6xx/r7xx hardware. Interrupts are implemented via a ring buffer. The GPU adds interrupts vectors to the ring and the host reads them off in the interrupt handler. The interrupt controller requires firmware like the CP. This firmware must be installed and accessble to the firmware loader for interrupts to function. MSIs don't seem to work on my RS780. They work fine on all my discrete cards. I'm not sure about other RS780s or RS880s. I've disabled MSIs on RS780 and RS880, but it would probably be worth checking on some other systems. v2 - fix some checkpatch.pl problems; re-read the disp int status reg if we restart the ih; v3 - remove the irq handler if r600_irq_init() fails; remove spinlock in r600_ih_ring_fini(); move ih rb overflow check to r600_get_ih_wptr(); move irq ack to separate function; Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-02 02:43:46 +08:00
break;
case 181: /* CP EOP event */
DRM_DEBUG("IH: CP EOP\n");
drm/radeon/kms: add support for multiple fence queues v2 For supporting multiple CP ring buffers, async DMA engines and UVD. We still need a way to synchronize between engines. v2 initialize unused fence driver ring to avoid issue in suspend/unload Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Christian König <deathsimple@vodafone.de> Reviewed-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-08-26 01:39:48 +08:00
radeon_fence_process(rdev, RADEON_RING_TYPE_GFX_INDEX);
drm/radeon/kms: Add support for interrupts on r6xx/r7xx chips (v3) This enables the use of interrupts on r6xx/r7xx hardware. Interrupts are implemented via a ring buffer. The GPU adds interrupts vectors to the ring and the host reads them off in the interrupt handler. The interrupt controller requires firmware like the CP. This firmware must be installed and accessble to the firmware loader for interrupts to function. MSIs don't seem to work on my RS780. They work fine on all my discrete cards. I'm not sure about other RS780s or RS880s. I've disabled MSIs on RS780 and RS880, but it would probably be worth checking on some other systems. v2 - fix some checkpatch.pl problems; re-read the disp int status reg if we restart the ih; v3 - remove the irq handler if r600_irq_init() fails; remove spinlock in r600_ih_ring_fini(); move ih rb overflow check to r600_get_ih_wptr(); move irq ack to separate function; Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-02 02:43:46 +08:00
break;
drm/radeon/kms: Add initial support for async DMA on r6xx/r7xx Uses the new multi-ring infrastucture. 6xx/7xx has a single async DMA ring. Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2012-09-28 03:08:35 +08:00
case 224: /* DMA trap event */
DRM_DEBUG("IH: DMA trap\n");
radeon_fence_process(rdev, R600_RING_TYPE_DMA_INDEX);
break;
drm/radeon/kms: add support for gui idle interrupts (v4) Useful for certain power management operations. You need to wait for the GUI engine (2D, 3D, CP, etc.) to be idle before changing clocks or adjusting engine parameters. (v2) Fix gui idle enable on pre-r6xx asics (v3) The gui idle interrrupt status bit is permanently asserted on pre-r6xx chips, but the interrrupt is still generated. workaround it in the driver. (v4) Add support for evergreen Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-04-23 00:52:11 +08:00
case 233: /* GUI IDLE */
drm/radeon: fix GUI idle IH debug statements debug statement for GUI idle interrupt is wrong and incorrectly reports CP EOP interrupt; trivial issue, but confusing for someone trying to distinguish interrupt sources while debugging ... fixed Signed-off-by: Ilija Hadzic <ihadzic@research.bell-labs.com> Reviewed-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-06-08 02:54:48 +08:00
DRM_DEBUG("IH: GUI idle\n");
drm/radeon/kms: add support for gui idle interrupts (v4) Useful for certain power management operations. You need to wait for the GUI engine (2D, 3D, CP, etc.) to be idle before changing clocks or adjusting engine parameters. (v2) Fix gui idle enable on pre-r6xx asics (v3) The gui idle interrrupt status bit is permanently asserted on pre-r6xx chips, but the interrrupt is still generated. workaround it in the driver. (v4) Add support for evergreen Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-04-23 00:52:11 +08:00
break;
drm/radeon/kms: Add support for interrupts on r6xx/r7xx chips (v3) This enables the use of interrupts on r6xx/r7xx hardware. Interrupts are implemented via a ring buffer. The GPU adds interrupts vectors to the ring and the host reads them off in the interrupt handler. The interrupt controller requires firmware like the CP. This firmware must be installed and accessble to the firmware loader for interrupts to function. MSIs don't seem to work on my RS780. They work fine on all my discrete cards. I'm not sure about other RS780s or RS880s. I've disabled MSIs on RS780 and RS880, but it would probably be worth checking on some other systems. v2 - fix some checkpatch.pl problems; re-read the disp int status reg if we restart the ih; v3 - remove the irq handler if r600_irq_init() fails; remove spinlock in r600_ih_ring_fini(); move ih rb overflow check to r600_get_ih_wptr(); move irq ack to separate function; Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-02 02:43:46 +08:00
default:
drm/radeon/kms/r6xx+: make irq handler less verbose Unhandled vectors can be safely ignored, no need to spam the kernel log by default. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-01-12 08:47:38 +08:00
DRM_DEBUG("Unhandled interrupt: %d %d\n", src_id, src_data);
drm/radeon/kms: Add support for interrupts on r6xx/r7xx chips (v3) This enables the use of interrupts on r6xx/r7xx hardware. Interrupts are implemented via a ring buffer. The GPU adds interrupts vectors to the ring and the host reads them off in the interrupt handler. The interrupt controller requires firmware like the CP. This firmware must be installed and accessble to the firmware loader for interrupts to function. MSIs don't seem to work on my RS780. They work fine on all my discrete cards. I'm not sure about other RS780s or RS880s. I've disabled MSIs on RS780 and RS880, but it would probably be worth checking on some other systems. v2 - fix some checkpatch.pl problems; re-read the disp int status reg if we restart the ih; v3 - remove the irq handler if r600_irq_init() fails; remove spinlock in r600_ih_ring_fini(); move ih rb overflow check to r600_get_ih_wptr(); move irq ack to separate function; Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-02 02:43:46 +08:00
break;
}
/* wptr/rptr are in bytes! */
drm/radeon/kms: r600/r700 disable irq at suspend To avoid hw doing anythings after we disabled PCIE GART, fully disable IRQ at suspend. Also cleanup a bit the ih structure and process function. Signed-off-by: Jerome Glisse <jglisse@redhat.com> Reviewed-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@linux.ie>
2010-01-15 21:44:37 +08:00
rptr += 16;
rptr &= rdev->ih.ptr_mask;
drm/radeon/kms: Add support for interrupts on r6xx/r7xx chips (v3) This enables the use of interrupts on r6xx/r7xx hardware. Interrupts are implemented via a ring buffer. The GPU adds interrupts vectors to the ring and the host reads them off in the interrupt handler. The interrupt controller requires firmware like the CP. This firmware must be installed and accessble to the firmware loader for interrupts to function. MSIs don't seem to work on my RS780. They work fine on all my discrete cards. I'm not sure about other RS780s or RS880s. I've disabled MSIs on RS780 and RS880, but it would probably be worth checking on some other systems. v2 - fix some checkpatch.pl problems; re-read the disp int status reg if we restart the ih; v3 - remove the irq handler if r600_irq_init() fails; remove spinlock in r600_ih_ring_fini(); move ih rb overflow check to r600_get_ih_wptr(); move irq ack to separate function; Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-02 02:43:46 +08:00
}
drm/radeon/kms: enable hpd support This enabled interrupt driven hpd support for all radeon chips. Assuming the hpd pin is wired up correctly, the driver will generate uevents on digital monitor connect and disconnect and retrain DP monitors automatically. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-05 05:56:37 +08:00
if (queue_hotplug)
drm/radeon: use system_wq instead of dev_priv->wq With cmwq, there's no reason for radeon to use a dedicated workqueue. Drop dev_priv->wq and use system_wq instead. Because radeon_driver_irq_uninstall_kms() may be called from unsleepable context, the work items can't be flushed from there. Instead, init and flush from radeon_irq_kms_init/fini(). While at it, simplify canceling/flushing of rdev->pm.dynpm_idle_work. Always initialize and sync cancel instead of being unnecessarily smart about it. Signed-off-by: Tejun Heo <tj@kernel.org> Acked-by: Alex Deucher <alexdeucher@gmail.com> Cc: dri-devel@lists.freedesktop.org Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-01-03 21:49:32 +08:00
schedule_work(&rdev->hotplug_work);
drm/radeon/kms: fix up audio interrupt handling - add support for rs6xx - add support for DCE4/5 - fixup 6xx/7xx Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2012-03-30 20:59:57 +08:00
if (queue_hdmi)
schedule_work(&rdev->audio_work);
drm/radeon/kms: Add support for interrupts on r6xx/r7xx chips (v3) This enables the use of interrupts on r6xx/r7xx hardware. Interrupts are implemented via a ring buffer. The GPU adds interrupts vectors to the ring and the host reads them off in the interrupt handler. The interrupt controller requires firmware like the CP. This firmware must be installed and accessble to the firmware loader for interrupts to function. MSIs don't seem to work on my RS780. They work fine on all my discrete cards. I'm not sure about other RS780s or RS880s. I've disabled MSIs on RS780 and RS880, but it would probably be worth checking on some other systems. v2 - fix some checkpatch.pl problems; re-read the disp int status reg if we restart the ih; v3 - remove the irq handler if r600_irq_init() fails; remove spinlock in r600_ih_ring_fini(); move ih rb overflow check to r600_get_ih_wptr(); move irq ack to separate function; Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-02 02:43:46 +08:00
rdev->ih.rptr = rptr;
WREG32(IH_RB_RPTR, rdev->ih.rptr);
drm/radeon: fix & improve ih ring handling v3 The spinlock was actually there to protect the rptr, but rptr was read outside of the locked area. Also we don't really need a spinlock here, an atomic should to quite fine since we only need to prevent it from being reentrant. v2: Keep the spinlock.... v3: Back to an atomic again after finding & fixing the real bug. Signed-off-by: Christian Koenig <christian.koenig@amd.com>
2012-05-17 03:45:24 +08:00
atomic_set(&rdev->ih.lock, 0);
/* make sure wptr hasn't changed while processing */
wptr = r600_get_ih_wptr(rdev);
if (wptr != rptr)
goto restart_ih;
drm/radeon/kms: Add support for interrupts on r6xx/r7xx chips (v3) This enables the use of interrupts on r6xx/r7xx hardware. Interrupts are implemented via a ring buffer. The GPU adds interrupts vectors to the ring and the host reads them off in the interrupt handler. The interrupt controller requires firmware like the CP. This firmware must be installed and accessble to the firmware loader for interrupts to function. MSIs don't seem to work on my RS780. They work fine on all my discrete cards. I'm not sure about other RS780s or RS880s. I've disabled MSIs on RS780 and RS880, but it would probably be worth checking on some other systems. v2 - fix some checkpatch.pl problems; re-read the disp int status reg if we restart the ih; v3 - remove the irq handler if r600_irq_init() fails; remove spinlock in r600_ih_ring_fini(); move ih rb overflow check to r600_get_ih_wptr(); move irq ack to separate function; Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-12-02 02:43:46 +08:00
return IRQ_HANDLED;
}
drm/radeon/kms: add r600 KMS support This adds the r600 KMS + CS support to the Linux kernel. The r600 TTM support is quite basic and still needs more work esp around using interrupts, but the polled fencing should work okay for now. Also currently TTM is using memcpy to do VRAM moves, the code is here to use a 3D blit to do this, but isn't fully debugged yet. Authors: Alex Deucher <alexdeucher@gmail.com> Dave Airlie <airlied@redhat.com> Jerome Glisse <jglisse@redhat.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-09-08 08:10:24 +08:00
/*
* Debugfs info
*/
#if defined(CONFIG_DEBUG_FS)
static int r600_debugfs_mc_info(struct seq_file *m, void *data)
{
struct drm_info_node *node = (struct drm_info_node *) m->private;
struct drm_device *dev = node->minor->dev;
struct radeon_device *rdev = dev->dev_private;
DREG32_SYS(m, rdev, R_000E50_SRBM_STATUS);
DREG32_SYS(m, rdev, VM_L2_STATUS);
return 0;
}
static struct drm_info_list r600_mc_info_list[] = {
{"r600_mc_info", r600_debugfs_mc_info, 0, NULL},
};
#endif
int r600_debugfs_mc_info_init(struct radeon_device *rdev)
{
#if defined(CONFIG_DEBUG_FS)
return radeon_debugfs_add_files(rdev, r600_mc_info_list, ARRAY_SIZE(r600_mc_info_list));
#else
return 0;
#endif
drm/radeon: introduce kernel modesetting for radeon hardware Add kernel modesetting support to radeon driver, use the ttm memory manager to manage memory and DRM/GEM to provide userspace API. In order to avoid backward compatibility issue and to allow clean design and code the radeon kernel modesetting use different code path than old radeon/drm driver. When kernel modesetting is enabled the IOCTL of radeon/drm driver are considered as invalid and an error message is printed in the log and they return failure. KMS enabled userspace will use new API to talk with the radeon/drm driver. The new API provide functions to create/destroy/share/mmap buffer object which are then managed by the kernel memory manager (here TTM). In order to submit command to the GPU the userspace provide a buffer holding the command stream, along this buffer userspace have to provide a list of buffer object used by the command stream. The kernel radeon driver will then place buffer in GPU accessible memory and will update command stream to reflect the position of the different buffers. The kernel will also perform security check on command stream provided by the user, we want to catch and forbid any illegal use of the GPU such as DMA into random system memory or into memory not owned by the process supplying the command stream. This part of the code is still incomplete and this why we propose that patch as a staging driver addition, future security might forbid current experimental userspace to run. This code support the following hardware : R1XX,R2XX,R3XX,R4XX,R5XX (radeon up to X1950). Works is underway to provide support for R6XX, R7XX and newer hardware (radeon from HD2XXX to HD4XXX). Authors: Jerome Glisse <jglisse@redhat.com> Dave Airlie <airlied@redhat.com> Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com> Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-06-05 20:42:42 +08:00
}
drm/radeon/kms: fix regression rendering issue on R6XX/R7XX It seems that some R6XX/R7XX silently ignore HDP flush when programmed through ring, this patch addback an ioctl callback to allow R6XX/R7XX hw to perform such flush through MMIO in order to fix a regression. For more details see: http://bugzilla.kernel.org/show_bug.cgi?id=15186 Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-02-05 03:36:39 +08:00
/**
* r600_ioctl_wait_idle - flush host path cache on wait idle ioctl
* rdev: radeon device structure
* bo: buffer object struct which userspace is waiting for idle
*
* Some R6XX/R7XX doesn't seems to take into account HDP flush performed
* through ring buffer, this leads to corruption in rendering, see
* http://bugzilla.kernel.org/show_bug.cgi?id=15186 to avoid this we
* directly perform HDP flush by writing register through MMIO.
*/
void r600_ioctl_wait_idle(struct radeon_device *rdev, struct radeon_bo *bo)
{
drm/radeon/kms/r7xx: add workaround for hw issue with HDP flush Use of HDP_*_COHERENCY_FLUSH_CNTL can cause a hang in certain situations. Add workaround. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Cc: stable@kernel.org Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-07-27 06:51:53 +08:00
/* r7xx hw bug. write to HDP_DEBUG1 followed by fb read
drm/radeon/kms: don't apply 7xx HDP flush workaround on AGP It should be required for all 7xx asics, but seems to cause problems on some AGP 7xx chips. Fixes: https://bugzilla.kernel.org/show_bug.cgi?id=19002 Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Reported-and-Tested-by: Duncan <1i5t5.duncan@cox.net> Cc: stable@kernel.org Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-12-08 23:05:34 +08:00
* rather than write to HDP_REG_COHERENCY_FLUSH_CNTL.
* This seems to cause problems on some AGP cards. Just use the old
* method for them.
drm/radeon/kms/r7xx: add workaround for hw issue with HDP flush Use of HDP_*_COHERENCY_FLUSH_CNTL can cause a hang in certain situations. Add workaround. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Cc: stable@kernel.org Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-07-27 06:51:53 +08:00
*/
drm/radeon/kms: fix potential segfault in r600_ioctl_wait_idle radeon_gem_wait_idle_ioctl can apparently get called prior to the vram page being set up or even if accel if false, so make sure it's valid before using it. Should fix: http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=597636 https://bugs.freedesktop.org/show_bug.cgi?id=29834 Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Cc: stable@kernel.org Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-09-27 22:57:10 +08:00
if ((rdev->family >= CHIP_RV770) && (rdev->family <= CHIP_RV740) &&
drm/radeon/kms: don't apply 7xx HDP flush workaround on AGP It should be required for all 7xx asics, but seems to cause problems on some AGP 7xx chips. Fixes: https://bugzilla.kernel.org/show_bug.cgi?id=19002 Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Reported-and-Tested-by: Duncan <1i5t5.duncan@cox.net> Cc: stable@kernel.org Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-12-08 23:05:34 +08:00
rdev->vram_scratch.ptr && !(rdev->flags & RADEON_IS_AGP)) {
drm/radeon/kms: fix a regression on r7xx AGP due to the HDP flush fix commit: 812d046915f48236657f02c06d7dc47140e9ceda drm/radeon/kms/r7xx: add workaround for hw issue with HDP flush breaks on AGP boards since there is no VRAM gart table. This patch fixes the issue by creating a VRAM scratch page so that can be used on both AGP and PCIE. Fixes: https://bugs.freedesktop.org/show_bug.cgi?id=29834 Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Cc: stable@kernel.org Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-08-28 01:59:54 +08:00
void __iomem *ptr = (void *)rdev->vram_scratch.ptr;
drm/radeon/kms/r7xx: add workaround for hw issue with HDP flush Use of HDP_*_COHERENCY_FLUSH_CNTL can cause a hang in certain situations. Add workaround. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Cc: stable@kernel.org Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-07-27 06:51:53 +08:00
u32 tmp;
WREG32(HDP_DEBUG1, 0);
tmp = readl((void __iomem *)ptr);
} else
WREG32(R_005480_HDP_MEM_COHERENCY_FLUSH_CNTL, 0x1);
drm/radeon/kms: fix regression rendering issue on R6XX/R7XX It seems that some R6XX/R7XX silently ignore HDP flush when programmed through ring, this patch addback an ioctl callback to allow R6XX/R7XX hw to perform such flush through MMIO in order to fix a regression. For more details see: http://bugzilla.kernel.org/show_bug.cgi?id=15186 Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-02-05 03:36:39 +08:00
}
drm/radeon/kms: add pcie get/set lane support for r6xx/r7xx/evergreen Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-01-07 07:49:34 +08:00
void r600_set_pcie_lanes(struct radeon_device *rdev, int lanes)
{
u32 link_width_cntl, mask, target_reg;
if (rdev->flags & RADEON_IS_IGP)
return;
if (!(rdev->flags & RADEON_IS_PCIE))
return;
/* x2 cards have a special sequence */
if (ASIC_IS_X2(rdev))
return;
/* FIXME wait for idle */
switch (lanes) {
case 0:
mask = RADEON_PCIE_LC_LINK_WIDTH_X0;
break;
case 1:
mask = RADEON_PCIE_LC_LINK_WIDTH_X1;
break;
case 2:
mask = RADEON_PCIE_LC_LINK_WIDTH_X2;
break;
case 4:
mask = RADEON_PCIE_LC_LINK_WIDTH_X4;
break;
case 8:
mask = RADEON_PCIE_LC_LINK_WIDTH_X8;
break;
case 12:
mask = RADEON_PCIE_LC_LINK_WIDTH_X12;
break;
case 16:
default:
mask = RADEON_PCIE_LC_LINK_WIDTH_X16;
break;
}
link_width_cntl = RREG32_PCIE_P(RADEON_PCIE_LC_LINK_WIDTH_CNTL);
if ((link_width_cntl & RADEON_PCIE_LC_LINK_WIDTH_RD_MASK) ==
(mask << RADEON_PCIE_LC_LINK_WIDTH_RD_SHIFT))
return;
if (link_width_cntl & R600_PCIE_LC_UPCONFIGURE_DIS)
return;
link_width_cntl &= ~(RADEON_PCIE_LC_LINK_WIDTH_MASK |
RADEON_PCIE_LC_RECONFIG_NOW |
R600_PCIE_LC_RENEGOTIATE_EN |
R600_PCIE_LC_RECONFIG_ARC_MISSING_ESCAPE);
link_width_cntl |= mask;
WREG32_PCIE_P(RADEON_PCIE_LC_LINK_WIDTH_CNTL, link_width_cntl);
/* some northbridges can renegotiate the link rather than requiring
* a complete re-config.
* e.g., AMD 780/790 northbridges (pci ids: 0x5956, 0x5957, 0x5958, etc.)
*/
if (link_width_cntl & R600_PCIE_LC_RENEGOTIATION_SUPPORT)
link_width_cntl |= R600_PCIE_LC_RENEGOTIATE_EN | R600_PCIE_LC_UPCONFIGURE_SUPPORT;
else
link_width_cntl |= R600_PCIE_LC_RECONFIG_ARC_MISSING_ESCAPE;
WREG32_PCIE_P(RADEON_PCIE_LC_LINK_WIDTH_CNTL, (link_width_cntl |
RADEON_PCIE_LC_RECONFIG_NOW));
if (rdev->family >= CHIP_RV770)
target_reg = R700_TARGET_AND_CURRENT_PROFILE_INDEX;
else
target_reg = R600_TARGET_AND_CURRENT_PROFILE_INDEX;
/* wait for lane set to complete */
link_width_cntl = RREG32(target_reg);
while (link_width_cntl == 0xffffffff)
link_width_cntl = RREG32(target_reg);
}
int r600_get_pcie_lanes(struct radeon_device *rdev)
{
u32 link_width_cntl;
if (rdev->flags & RADEON_IS_IGP)
return 0;
if (!(rdev->flags & RADEON_IS_PCIE))
return 0;
/* x2 cards have a special sequence */
if (ASIC_IS_X2(rdev))
return 0;
/* FIXME wait for idle */
link_width_cntl = RREG32_PCIE_P(RADEON_PCIE_LC_LINK_WIDTH_CNTL);
switch ((link_width_cntl & RADEON_PCIE_LC_LINK_WIDTH_RD_MASK) >> RADEON_PCIE_LC_LINK_WIDTH_RD_SHIFT) {
case RADEON_PCIE_LC_LINK_WIDTH_X0:
return 0;
case RADEON_PCIE_LC_LINK_WIDTH_X1:
return 1;
case RADEON_PCIE_LC_LINK_WIDTH_X2:
return 2;
case RADEON_PCIE_LC_LINK_WIDTH_X4:
return 4;
case RADEON_PCIE_LC_LINK_WIDTH_X8:
return 8;
case RADEON_PCIE_LC_LINK_WIDTH_X16:
default:
return 16;
}
}
drm/radeon/kms: add support for gen2 pcie link speeds Supported on rv6xx/r7xx/evergreen. Cards come up in gen1 mode. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-01-07 07:49:35 +08:00
static void r600_pcie_gen2_enable(struct radeon_device *rdev)
{
u32 link_width_cntl, lanes, speed_cntl, training_cntl, tmp;
u16 link_cntl2;
drm/radeon/kms: auto detect pcie link speed from root port This check the root ports supported link speeds and enables GEN2 mode if the 5.0 GT link speed is available. The first 3.0 cards are SI so they will probably need more investigation. Signed-off-by: Dave Airlie <airlied@redhat.com>
2012-06-27 15:35:54 +08:00
u32 mask;
int ret;
drm/radeon/kms: add support for gen2 pcie link speeds Supported on rv6xx/r7xx/evergreen. Cards come up in gen1 mode. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-01-07 07:49:35 +08:00
drm/radeon/kms: add module option for pcie gen2 Switching to pcie gen2 causes problems on some boards. Add a module option to turn it on/off. There are gen2 compatability issues with some motherboards it seems. Fixes: https://bugs.freedesktop.org/show_bug.cgi?id=33027 Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-01-13 09:05:11 +08:00
if (radeon_pcie_gen2 == 0)
return;
drm/radeon/kms: add support for gen2 pcie link speeds Supported on rv6xx/r7xx/evergreen. Cards come up in gen1 mode. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-01-07 07:49:35 +08:00
if (rdev->flags & RADEON_IS_IGP)
return;
if (!(rdev->flags & RADEON_IS_PCIE))
return;
/* x2 cards have a special sequence */
if (ASIC_IS_X2(rdev))
return;
/* only RV6xx+ chips are supported */
if (rdev->family <= CHIP_R600)
return;
drm/radeon/kms: auto detect pcie link speed from root port This check the root ports supported link speeds and enables GEN2 mode if the 5.0 GT link speed is available. The first 3.0 cards are SI so they will probably need more investigation. Signed-off-by: Dave Airlie <airlied@redhat.com>
2012-06-27 15:35:54 +08:00
ret = drm_pcie_get_speed_cap_mask(rdev->ddev, &mask);
if (ret != 0)
return;
if (!(mask & DRM_PCIE_SPEED_50))
return;
drm/radeon: check if pcie gen 2 is already enabled (v2) If so, skip enabling it to save time. v2: coding style fixes Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2012-10-09 05:46:27 +08:00
speed_cntl = RREG32_PCIE_P(PCIE_LC_SPEED_CNTL);
if (speed_cntl & LC_CURRENT_DATA_RATE) {
DRM_INFO("PCIE gen 2 link speeds already enabled\n");
return;
}
drm/radeon/kms: auto detect pcie link speed from root port This check the root ports supported link speeds and enables GEN2 mode if the 5.0 GT link speed is available. The first 3.0 cards are SI so they will probably need more investigation. Signed-off-by: Dave Airlie <airlied@redhat.com>
2012-06-27 15:35:54 +08:00
DRM_INFO("enabling PCIE gen 2 link speeds, disable with radeon.pcie_gen2=0\n");
drm/radeon/kms: add support for gen2 pcie link speeds Supported on rv6xx/r7xx/evergreen. Cards come up in gen1 mode. Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-01-07 07:49:35 +08:00
/* 55 nm r6xx asics */
if ((rdev->family == CHIP_RV670) ||
(rdev->family == CHIP_RV620) ||
(rdev->family == CHIP_RV635)) {
/* advertise upconfig capability */
link_width_cntl = RREG32_PCIE_P(PCIE_LC_LINK_WIDTH_CNTL);
link_width_cntl &= ~LC_UPCONFIGURE_DIS;
WREG32_PCIE_P(PCIE_LC_LINK_WIDTH_CNTL, link_width_cntl);
link_width_cntl = RREG32_PCIE_P(PCIE_LC_LINK_WIDTH_CNTL);
if (link_width_cntl & LC_RENEGOTIATION_SUPPORT) {
lanes = (link_width_cntl & LC_LINK_WIDTH_RD_MASK) >> LC_LINK_WIDTH_RD_SHIFT;
link_width_cntl &= ~(LC_LINK_WIDTH_MASK |
LC_RECONFIG_ARC_MISSING_ESCAPE);
link_width_cntl |= lanes | LC_RECONFIG_NOW | LC_RENEGOTIATE_EN;
WREG32_PCIE_P(PCIE_LC_LINK_WIDTH_CNTL, link_width_cntl);
} else {
link_width_cntl |= LC_UPCONFIGURE_DIS;
WREG32_PCIE_P(PCIE_LC_LINK_WIDTH_CNTL, link_width_cntl);
}
}
speed_cntl = RREG32_PCIE_P(PCIE_LC_SPEED_CNTL);
if ((speed_cntl & LC_OTHER_SIDE_EVER_SENT_GEN2) &&
(speed_cntl & LC_OTHER_SIDE_SUPPORTS_GEN2)) {
/* 55 nm r6xx asics */
if ((rdev->family == CHIP_RV670) ||
(rdev->family == CHIP_RV620) ||
(rdev->family == CHIP_RV635)) {
WREG32(MM_CFGREGS_CNTL, 0x8);
link_cntl2 = RREG32(0x4088);
WREG32(MM_CFGREGS_CNTL, 0);
/* not supported yet */
if (link_cntl2 & SELECTABLE_DEEMPHASIS)
return;
}
speed_cntl &= ~LC_SPEED_CHANGE_ATTEMPTS_ALLOWED_MASK;
speed_cntl |= (0x3 << LC_SPEED_CHANGE_ATTEMPTS_ALLOWED_SHIFT);
speed_cntl &= ~LC_VOLTAGE_TIMER_SEL_MASK;
speed_cntl &= ~LC_FORCE_DIS_HW_SPEED_CHANGE;
speed_cntl |= LC_FORCE_EN_HW_SPEED_CHANGE;
WREG32_PCIE_P(PCIE_LC_SPEED_CNTL, speed_cntl);
tmp = RREG32(0x541c);
WREG32(0x541c, tmp | 0x8);
WREG32(MM_CFGREGS_CNTL, MM_WR_TO_CFG_EN);
link_cntl2 = RREG16(0x4088);
link_cntl2 &= ~TARGET_LINK_SPEED_MASK;
link_cntl2 |= 0x2;
WREG16(0x4088, link_cntl2);
WREG32(MM_CFGREGS_CNTL, 0);
if ((rdev->family == CHIP_RV670) ||
(rdev->family == CHIP_RV620) ||
(rdev->family == CHIP_RV635)) {
training_cntl = RREG32_PCIE_P(PCIE_LC_TRAINING_CNTL);
training_cntl &= ~LC_POINT_7_PLUS_EN;
WREG32_PCIE_P(PCIE_LC_TRAINING_CNTL, training_cntl);
} else {
speed_cntl = RREG32_PCIE_P(PCIE_LC_SPEED_CNTL);
speed_cntl &= ~LC_TARGET_LINK_SPEED_OVERRIDE_EN;
WREG32_PCIE_P(PCIE_LC_SPEED_CNTL, speed_cntl);
}
speed_cntl = RREG32_PCIE_P(PCIE_LC_SPEED_CNTL);
speed_cntl |= LC_GEN2_EN_STRAP;
WREG32_PCIE_P(PCIE_LC_SPEED_CNTL, speed_cntl);
} else {
link_width_cntl = RREG32_PCIE_P(PCIE_LC_LINK_WIDTH_CNTL);
/* XXX: only disable it if gen1 bridge vendor == 0x111d or 0x1106 */
if (1)
link_width_cntl |= LC_UPCONFIGURE_DIS;
else
link_width_cntl &= ~LC_UPCONFIGURE_DIS;
WREG32_PCIE_P(PCIE_LC_LINK_WIDTH_CNTL, link_width_cntl);
}
}
drm/radeon/kms: implement timestamp userspace query (v2) Returns a snapshot of the GPU clock counter. Needed for certain OpenGL extensions. v2: agd5f - address Jerome's comments - add function documentation Signed-off-by: Marek Olšák <maraeo@gmail.com> Reviewed-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2012-08-09 22:34:17 +08:00
/**
* r600_get_gpu_clock - return GPU clock counter snapshot
*
* @rdev: radeon_device pointer
*
* Fetches a GPU clock counter snapshot (R6xx-cayman).
* Returns the 64 bit clock counter snapshot.
*/
uint64_t r600_get_gpu_clock(struct radeon_device *rdev)
{
uint64_t clock;
mutex_lock(&rdev->gpu_clock_mutex);
WREG32(RLC_CAPTURE_GPU_CLOCK_COUNT, 1);
clock = (uint64_t)RREG32(RLC_GPU_CLOCK_COUNT_LSB) |
((uint64_t)RREG32(RLC_GPU_CLOCK_COUNT_MSB) << 32ULL);
mutex_unlock(&rdev->gpu_clock_mutex);
return clock;
}
Reference in New Issue Copy Permalink