Since commit 4bdadb9785 ("drm/i915:
Selectively enable self-reclaim"), we've been passing GFP_MOVABLE to the
i915 page allocator where we weren't before due to some over-eager
removal of the page mapping gfp_flags games the code used to play.
This caused hibernate on Intel hardware to result in a lot of memory
corruptions on resume. See for example
http://bugzilla.kernel.org/show_bug.cgi?id=13811
Reported-by: Evengi Golov (in bugzilla)
Signed-off-by: Dave Airlie <airlied@redhat.com>
Tested-by: M. Vefa Bicakci <bicave@superonline.com>
Cc: stable@kernel.org
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Hugh Dickins <hugh.dickins@tiscali.co.uk>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Some RV100 cards with 2 VGA ports show up with DVI+VGA, however
some boards with DVI+VGA have the same subsystem ids. Better
to have a VGA port show up as DVI than having a non-useable
DVI port.
reported by DHR in irc.
Signed-off-by: Alex Deucher <alexdeucher@gmail.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
There is a regression from 2.6.34 related to the recent radeon power
management changes, caused by attempting to cancel a delayed work
item that's never been scheduled. However, the code as is has some
other issues potentially leading to visible problems.
First, the mutex around cancel_delayed_work() in radeon_pm_suspend()
doesn't really serve any purpose, because cancel_delayed_work() only
tries to delete the work's timer. Moreover, it doesn't prevent the
work handler from running, so the handler can do some wrong things if
it wins the race and in that case it will rearm itself to do some
more wrong things going forward. So, I think it's better to wait for
the handler to return in case it's already been queued up for
execution. Also, it should be prevented from rearming itself in that
case.
Second, in radeon_set_pm_method() the cancel_delayed_work() is not
sufficient to prevent the work handler from running and queing up
itself for the next run (the failure scenario is that
cancel_delayed_work() returns 0, so the handler is run, it waits on
the mutex and then rearms itself after the mutex has been released),
so again the work handler should be prevented from rearming itself in
that case..
Finally, there's a potential deadlock in radeon_pm_fini(), because
cancel_delayed_work_sync() is called under rdev->pm.mutex, but the
work handler tries to acquire the same mutex (if it wins the race).
Fix the issues described above.
Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
Reviewed-by: Alex Deucher <alexdeucher@gmail.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
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>
For some reason on resume, executing the BIOS scripts locks up the whole chipset, by avoiding the dynclk table the machine resumes properly and seems to function okay.
Signed-off-by: Dave Airlie <airlied@redhat.com>
They don't have an atombios so don't attempt to use it for
eng/mem clocks.
Reported by spoonb on #radeon
fixes:
https://bugs.freedesktop.org/show_bug.cgi?id=28671
Signed-off-by: Alex Deucher <alexdeucher@gmail.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
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>
Since the VGA switcheroo, we'd attempt to read the BIOS from VRAM on startup
but on some unposted cards this can cause hangs/crashes.
Fixes: https://bugs.freedesktop.org/show_bug.cgi?id=28592
(further problem pointed out by agd5f on IGP systems)
Reported-by: Reilithion on #radeon
Signed-off-by: Dave Airlie <airlied@redhat.com>
We were returning 0 in both the success and failure paths. Noticed while
investigating FDO bug 26403.
Signed-off-by: Matt Turner <mattst88@gmail.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Non pooled page allocation should have GFP_USER set so allocation
can wait and reclaim page from other process (ie non atomic).
Signed-off-by: Jerome Glisse <jglisse@redhat.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
This needs similar handling to other compressed textures.
Fixes:
https://bugs.freedesktop.org/show_bug.cgi?id=26428
Signed-off-by: sroland@vmware.com
Signed-off-by: Alex Deucher <alexdeucher@gmail.com>
Cc: stable <stable@kernel.org>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Code did not handle projected 2d and depth coordinates, meaning potentially
set 3d or cube special handling might stick.
(Not sure what depth coord actually does, but I guess handling it
like a normal coordinate is the right thing to do.)
Might be related to https://bugs.freedesktop.org/show_bug.cgi?id=26428
Signed-off-by: sroland@vmware.com
Signed-off-by: Alex Deucher <alexdeucher@gmail.com>
Cc: stable <stable@kernel.org>
Signed-off-by: Dave Airlie <airlied@redhat.com>
fixes:
https://bugs.freedesktop.org/show_bug.cgi?id=28459
agd5f: apply to r1xx/r2xx as well.
Signed-off-by: Roland Scheidegger <sroland@vmware.com>
Cc: stable@kernel.org
Signed-off-by: Alex Deucher <alexdeucher@gmail.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
This works well enough on a js21, but it would be nice if IBM could supply
more tables for the later Power6/7 machines.
Signed-off-by: Dave Airlie <airlied@redhat.com>
Reduced blanking is valid only when doing CVT modes. Also, generate GTF
modes unless CVT was requested; CVT devices are required to support GTF,
but the reverse is not true.
[airlied: fix typo]
Signed-off-by: Adam Jackson <ajax@redhat.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
The transmitter needs to be enabled before the link is trained.
Reported-By: Lars Doelle <lars.doelle@on-line.de>
Signed-off-by: Alex Deucher <alexdeucher@gmail.com>
Cc: stable <stable@kernel.org>
Signed-off-by: Dave Airlie <airlied@redhat.com>
When suspending, we turn the display hw off, at resume the screen will stay black.
This patch turn it on. Fixes:
https://bugzilla.kernel.org/show_bug.cgi?id=16180
Signed-off-by: Cedric Godin <cedric.godin@skynet.be>
Signed-off-by: Alex Deucher <alexdeucher@gmail.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Installing 2.6.34 on a Power5/rn50 combo machine, X showed buggy sw rendering,
enabling tiling in the DDX fixed it. Investigation showed that a further /16
was needed in the untiled case on this chipset. Need further investigations
on what other chips this could affect, possibly rv100->rv280.
Signed-off-by: Dave Airlie <airlied@redhat.com>
Plugged in FireMV with the rv250 on it, and the second crtc/dac didn't work,
we were reading/writing different registers than we were modifying in the code.
Signed-off-by: Dave Airlie <airlied@redhat.com>
This reverts commit cfecde435d, since it
seems to cause some systems to not come up with any video output at all
(or video that only comes on when X starts up).
Fixes bugzilla:
http://bugzilla.kernel.org/show_bug.cgi?id=16163
Reported-and-tested-by: David John <davidjon@xenontk.org>
Tested-by: Nick Bowler <nbowler@elliptictech.com>
Acked-by: Carl Worth <cworth@cworth.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
So when we added output polling, we'd suddenly use this code more often, and the fact that it always takes over crtc2 and messes with it during probing isn't what we really want to be happening. A more complete fix would to change it to use whatever crtc was free at the time, but for now lets stay simple and just don't poll if crtc2 is already in use.
Although a more correct fix was found I suspect we should do this as well, until we get a chance to readdres the tv out polling issues.
Reported-by: Torsten Kaiser <just.for.lkml@googlemail.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
We really don't want to be polling tv-out but since we weren't forcing the
i2c lines to invalid (tv-out has no DDC), we were adding tv connectors to the
polling setup and this was causing blinking on secondary displays.
This fixes the regression Torsten reported.
Reported-by: Torsten Kaiser <just.for.lkml@googlemail.com>
Tested-by: Torsten Kaiser <just.for.lkml@googlemail.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
dcb->i2c[] has DCB_MAX_NUM_I2C_ENTRIES entries.
Signed-off-by: Dan Carpenter <error27@gmail.com>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
If "gpio->line" is 32 then "nv50_gpio_reg[gpio->line >> 3]" reads past the
end of the array.
Signed-off-by: Dan Carpenter <error27@gmail.com>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
On cards where there's a specific BAR for PRAMIN, we used to try and fall
back to the "legacy" aperture within the mmio BAR.
This is doomed to cause problems, so lets just fail completely as there's
obviously something else very wrong anyway.
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
It appears version 0x21 'U' and 'd' tables require us to take the SOR link
into account when selecting the appropriate table for a particular output.
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
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>
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>
The vddc value in the power tables is not an actual voltage
like on discrete r6xx/r7xx/evergreen systems, but instead has
a symbolic meaning (e.g., NONE, LOW, HIGH, etc.). See atombios.h
Most RS780/RS880 vbioses don't have a SetVoltage table anyway,
so it shouldn't be doing anything to the hardware at the moment.
I need to figure out how voltage is supposed to work on the newer
IGPs; until then, disable it.
Signed-off-by: Alex Deucher <alexdeucher@gmail.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
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>
- don't rest the power state in pm_init()
We already boot up to the default power state. Note this
patch relies on:
drm/radeon/kms/pm: patch default power state with default clocks/voltages on r6xx+
To make sure the default power state matches the boot up state.
- In the pm resume path asic init will have set the power state
back to the default so reset the tracking state values.
Signed-off-by: Alex Deucher <alexdeucher@gmail.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
This is needed to enable accel in the ddx. However,
due to a bug in older versions of the ddx, it relies
on accel being disabled in order to load properly on
evergreen chips. To maintain compatility, we add a new
get accel param and call that from the ddx. The old one
always returns false for evergreen cards.
[this fixes a regression with older userspaces on newer kernels].
Signed-off-by: Alex Deucher <alexdeucher@gmail.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
copy_to/from_user() returns the number of bytes remaining to be copied
but we want to return a negative error code here. This gets returned to
userspace.
Signed-off-by: Dan Carpenter <error27@gmail.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
copy_from_user() returns the number of bytes left to be copied but we
want to return a negative error code here. This is in the ioctl handler
so the error code get returned to userspace.
Signed-off-by: Dan Carpenter <error27@gmail.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
using DRM_ERROR, results in people blaming the drm code for the oops, and
not looking at the oops.
(sadly yes I've gotten reports).
Signed-off-by: Dave Airlie <airlied@redhat.com>
PM attemps to unmap objects that aren't actually mapped into userspace ever,
so just don't bother unmapping them at this point, since all you are doing
is nothing. We should be making sure all access to these objects are locked in
kernel space instead. In theory the VRAM gart table is already done, and both
the shaders and stolen vga memory blocks are never accessed at runtime.
fixes: https://bugzilla.kernel.org/show_bug.cgi?id=16127
Reported-by: Jure Repnic <jlp.bugs@gmail.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
The previous commit fixes the problem, these commits make sure we actually
fail properly if it happens again.
I've squashed the commits from Chris since they are all fixing one issue.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Dave Airlie <airlied@redhat.com>
(regression fix since fbdev/kms rework).
My fb rework didn't remember about the 84/65s.
Reported-by: Ondrej Zary <linux@rainbow-software.org>
Tested-by: Ondrej Zary <linux@rainbow-software.org>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Cursors need to be in the GTT domain when being accessed by the GPU.
Previously this was a fortuitous byproduct of userspace using pwrite()
to upload the image data into the cursor. The redundant clflush was
removed in commit 9b8c4a and so the image was no longer being flushed
out of the caches into main memory. One could also devise a scenario
where the cursor was rendered by the GPU, prior to being attached as the
cursor, resulting in similar corruption due to the missing MI_FLUSH.
Fixes:
Bug 28335 - Cursor corruption caused by commit 9b8c4a0b21https://bugs.freedesktop.org/show_bug.cgi?id=28335
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Reported-and-tested-by: Jeff Chua <jeff.chua.linux@gmail.com>
Tested-by: Linus Torvalds <torvalds@linux-foundation.org>
Reported-by: Andy Isaacson <adi@hexapodia.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
