From 4522ad764d7e99a79de817fcbc682ebcc75d01fe Mon Sep 17 00:00:00 2001 From: Stanislaw Gruszka Date: Thu, 23 Mar 2023 13:54:57 +0100 Subject: [PATCH 01/11] accel/ivpu: Do not access HW registers after unbind We should not access hardware after we unbind from the bus. Use drm_dev_enter() / drm_dev_exit() to mark code sections where hardware is accessed (and not already protected by other locks) and drm_dev_unplug() to mark device is gone. Fixes: 35b137630f08 ("accel/ivpu: Introduce a new DRM driver for Intel VPU") Signed-off-by: Stanislaw Gruszka Reviewed-by: Jeffrey Hugo Signed-off-by: Jacek Lawrynowicz Link: https://patchwork.freedesktop.org/patch/msgid/20230323125504.2586442-2-stanislaw.gruszka@linux.intel.com --- drivers/accel/ivpu/ivpu_drv.c | 8 ++++++-- drivers/accel/ivpu/ivpu_drv.h | 1 + drivers/accel/ivpu/ivpu_job.c | 11 +++++++++-- 3 files changed, 16 insertions(+), 4 deletions(-) diff --git a/drivers/accel/ivpu/ivpu_drv.c b/drivers/accel/ivpu/ivpu_drv.c index 231f29bb5025..ac06bbfca920 100644 --- a/drivers/accel/ivpu/ivpu_drv.c +++ b/drivers/accel/ivpu/ivpu_drv.c @@ -8,7 +8,6 @@ #include #include -#include #include #include #include @@ -118,6 +117,10 @@ static int ivpu_get_param_ioctl(struct drm_device *dev, void *data, struct drm_f struct pci_dev *pdev = to_pci_dev(vdev->drm.dev); struct drm_ivpu_param *args = data; int ret = 0; + int idx; + + if (!drm_dev_enter(dev, &idx)) + return -ENODEV; switch (args->param) { case DRM_IVPU_PARAM_DEVICE_ID: @@ -171,6 +174,7 @@ static int ivpu_get_param_ioctl(struct drm_device *dev, void *data, struct drm_f break; } + drm_dev_exit(idx); return ret; } @@ -622,7 +626,7 @@ static void ivpu_remove(struct pci_dev *pdev) { struct ivpu_device *vdev = pci_get_drvdata(pdev); - drm_dev_unregister(&vdev->drm); + drm_dev_unplug(&vdev->drm); ivpu_dev_fini(vdev); } diff --git a/drivers/accel/ivpu/ivpu_drv.h b/drivers/accel/ivpu/ivpu_drv.h index f47b4965db2e..1b2aa05840ad 100644 --- a/drivers/accel/ivpu/ivpu_drv.h +++ b/drivers/accel/ivpu/ivpu_drv.h @@ -7,6 +7,7 @@ #define __IVPU_DRV_H__ #include +#include #include #include #include diff --git a/drivers/accel/ivpu/ivpu_job.c b/drivers/accel/ivpu/ivpu_job.c index 94068aedf97c..910301fae435 100644 --- a/drivers/accel/ivpu/ivpu_job.c +++ b/drivers/accel/ivpu/ivpu_job.c @@ -489,12 +489,12 @@ unlock_reservations: int ivpu_submit_ioctl(struct drm_device *dev, void *data, struct drm_file *file) { - int ret = 0; struct ivpu_file_priv *file_priv = file->driver_priv; struct ivpu_device *vdev = file_priv->vdev; struct drm_ivpu_submit *params = data; struct ivpu_job *job; u32 *buf_handles; + int idx, ret; if (params->engine > DRM_IVPU_ENGINE_COPY) return -EINVAL; @@ -523,6 +523,11 @@ int ivpu_submit_ioctl(struct drm_device *dev, void *data, struct drm_file *file) goto free_handles; } + if (!drm_dev_enter(&vdev->drm, &idx)) { + ret = -ENODEV; + goto free_handles; + } + ivpu_dbg(vdev, JOB, "Submit ioctl: ctx %u buf_count %u\n", file_priv->ctx.id, params->buffer_count); @@ -530,7 +535,7 @@ int ivpu_submit_ioctl(struct drm_device *dev, void *data, struct drm_file *file) if (!job) { ivpu_err(vdev, "Failed to create job\n"); ret = -ENOMEM; - goto free_handles; + goto dev_exit; } ret = ivpu_job_prepare_bos_for_submit(file, job, buf_handles, params->buffer_count, @@ -548,6 +553,8 @@ int ivpu_submit_ioctl(struct drm_device *dev, void *data, struct drm_file *file) job_put: job_put(job); +dev_exit: + drm_dev_exit(idx); free_handles: kfree(buf_handles); From 6013aa84ee45fd0faa23fc6312a1af60b06d408b Mon Sep 17 00:00:00 2001 From: Stanislaw Gruszka Date: Thu, 23 Mar 2023 13:54:58 +0100 Subject: [PATCH 02/11] accel/ivpu: Cancel recovery work Prevent running recovery_work after device is removed. Fixes: 852be13f3bd3 ("accel/ivpu: Add PM support") Signed-off-by: Stanislaw Gruszka Reviewed-by: Jeffrey Hugo Signed-off-by: Jacek Lawrynowicz Link: https://patchwork.freedesktop.org/patch/msgid/20230323125504.2586442-3-stanislaw.gruszka@linux.intel.com --- drivers/accel/ivpu/ivpu_drv.c | 2 ++ drivers/accel/ivpu/ivpu_pm.c | 17 ++++++++++++++--- drivers/accel/ivpu/ivpu_pm.h | 1 + 3 files changed, 17 insertions(+), 3 deletions(-) diff --git a/drivers/accel/ivpu/ivpu_drv.c b/drivers/accel/ivpu/ivpu_drv.c index ac06bbfca920..d9e311b40348 100644 --- a/drivers/accel/ivpu/ivpu_drv.c +++ b/drivers/accel/ivpu/ivpu_drv.c @@ -580,6 +580,8 @@ static void ivpu_dev_fini(struct ivpu_device *vdev) ivpu_pm_disable(vdev); ivpu_shutdown(vdev); ivpu_job_done_thread_fini(vdev); + ivpu_pm_cancel_recovery(vdev); + ivpu_ipc_fini(vdev); ivpu_fw_fini(vdev); ivpu_mmu_global_context_fini(vdev); diff --git a/drivers/accel/ivpu/ivpu_pm.c b/drivers/accel/ivpu/ivpu_pm.c index 553bcbd787b3..7df72fa8100f 100644 --- a/drivers/accel/ivpu/ivpu_pm.c +++ b/drivers/accel/ivpu/ivpu_pm.c @@ -98,12 +98,18 @@ retry: static void ivpu_pm_recovery_work(struct work_struct *work) { struct ivpu_pm_info *pm = container_of(work, struct ivpu_pm_info, recovery_work); - struct ivpu_device *vdev = pm->vdev; + struct ivpu_device *vdev = pm->vdev; char *evt[2] = {"IVPU_PM_EVENT=IVPU_RECOVER", NULL}; int ret; - ret = pci_reset_function(to_pci_dev(vdev->drm.dev)); - if (ret) +retry: + ret = pci_try_reset_function(to_pci_dev(vdev->drm.dev)); + if (ret == -EAGAIN && !drm_dev_is_unplugged(&vdev->drm)) { + cond_resched(); + goto retry; + } + + if (ret && ret != -EAGAIN) ivpu_err(vdev, "Failed to reset VPU: %d\n", ret); kobject_uevent_env(&vdev->drm.dev->kobj, KOBJ_CHANGE, evt); @@ -306,6 +312,11 @@ int ivpu_pm_init(struct ivpu_device *vdev) return 0; } +void ivpu_pm_cancel_recovery(struct ivpu_device *vdev) +{ + cancel_work_sync(&vdev->pm->recovery_work); +} + void ivpu_pm_enable(struct ivpu_device *vdev) { struct device *dev = vdev->drm.dev; diff --git a/drivers/accel/ivpu/ivpu_pm.h b/drivers/accel/ivpu/ivpu_pm.h index dc1b3758e13f..baca98187255 100644 --- a/drivers/accel/ivpu/ivpu_pm.h +++ b/drivers/accel/ivpu/ivpu_pm.h @@ -21,6 +21,7 @@ struct ivpu_pm_info { int ivpu_pm_init(struct ivpu_device *vdev); void ivpu_pm_enable(struct ivpu_device *vdev); void ivpu_pm_disable(struct ivpu_device *vdev); +void ivpu_pm_cancel_recovery(struct ivpu_device *vdev); int ivpu_pm_suspend_cb(struct device *dev); int ivpu_pm_resume_cb(struct device *dev); From 3ff6edbc17dbab87f705b1da93854ffff3912673 Mon Sep 17 00:00:00 2001 From: Stanislaw Gruszka Date: Thu, 23 Mar 2023 13:54:59 +0100 Subject: [PATCH 03/11] accel/ivpu: Do not use SSID 1 The SSID=1 is used by the firmware as default value in case SSID mapping is not initialized. This allows detecting use of miss-configured memory contexts. The future FW versions may not allow using SSID=1. SSID=65 is valid value, number of contexts are limited by number of available command queues, but SSID can be any u16 value. Fixes: 35b137630f08 ("accel/ivpu: Introduce a new DRM driver for Intel VPU") Co-developed-by: Andrzej Kacprowski Signed-off-by: Andrzej Kacprowski Signed-off-by: Stanislaw Gruszka Reviewed-by: Jeffrey Hugo Signed-off-by: Jacek Lawrynowicz Link: https://patchwork.freedesktop.org/patch/msgid/20230323125504.2586442-4-stanislaw.gruszka@linux.intel.com --- drivers/accel/ivpu/ivpu_drv.c | 4 ++-- drivers/accel/ivpu/ivpu_drv.h | 5 ++++- 2 files changed, 6 insertions(+), 3 deletions(-) diff --git a/drivers/accel/ivpu/ivpu_drv.c b/drivers/accel/ivpu/ivpu_drv.c index d9e311b40348..70245cf84593 100644 --- a/drivers/accel/ivpu/ivpu_drv.c +++ b/drivers/accel/ivpu/ivpu_drv.c @@ -474,8 +474,8 @@ static int ivpu_dev_init(struct ivpu_device *vdev) vdev->hw->ops = &ivpu_hw_mtl_ops; vdev->platform = IVPU_PLATFORM_INVALID; - vdev->context_xa_limit.min = IVPU_GLOBAL_CONTEXT_MMU_SSID + 1; - vdev->context_xa_limit.max = IVPU_CONTEXT_LIMIT; + vdev->context_xa_limit.min = IVPU_USER_CONTEXT_MIN_SSID; + vdev->context_xa_limit.max = IVPU_USER_CONTEXT_MAX_SSID; atomic64_set(&vdev->unique_id_counter, 0); xa_init_flags(&vdev->context_xa, XA_FLAGS_ALLOC); xa_init_flags(&vdev->submitted_jobs_xa, XA_FLAGS_ALLOC1); diff --git a/drivers/accel/ivpu/ivpu_drv.h b/drivers/accel/ivpu/ivpu_drv.h index 1b2aa05840ad..ef12a38e06e1 100644 --- a/drivers/accel/ivpu/ivpu_drv.h +++ b/drivers/accel/ivpu/ivpu_drv.h @@ -25,7 +25,10 @@ #define PCI_DEVICE_ID_MTL 0x7d1d #define IVPU_GLOBAL_CONTEXT_MMU_SSID 0 -#define IVPU_CONTEXT_LIMIT 64 +/* SSID 1 is used by the VPU to represent invalid context */ +#define IVPU_USER_CONTEXT_MIN_SSID 2 +#define IVPU_USER_CONTEXT_MAX_SSID (IVPU_USER_CONTEXT_MIN_SSID + 63) + #define IVPU_NUM_ENGINES 2 #define IVPU_PLATFORM_SILICON 0 From a8fed6d1e0b9bbea2cbe602831f84a2cff4653e3 Mon Sep 17 00:00:00 2001 From: Stanislaw Gruszka Date: Thu, 23 Mar 2023 13:55:00 +0100 Subject: [PATCH 04/11] accel/ivpu: Fix power down sequence Remove FPGA workaround on power_down to skip checking for noc quiescent state. Put VPU in reset before powering it down and skip manipulating registers that are reset by the VPU reset. This fixes power down errors where VPU is powered down just after VPU is booted. Fixes: 35b137630f08 ("accel/ivpu: Introduce a new DRM driver for Intel VPU") Signed-off-by: Stanislaw Gruszka Reviewed-by: Jeffrey Hugo Signed-off-by: Jacek Lawrynowicz Link: https://patchwork.freedesktop.org/patch/msgid/20230323125504.2586442-5-stanislaw.gruszka@linux.intel.com --- drivers/accel/ivpu/ivpu_hw_mtl.c | 37 ++------------------------------ 1 file changed, 2 insertions(+), 35 deletions(-) diff --git a/drivers/accel/ivpu/ivpu_hw_mtl.c b/drivers/accel/ivpu/ivpu_hw_mtl.c index 62bfaa9081c4..70ca6de78060 100644 --- a/drivers/accel/ivpu/ivpu_hw_mtl.c +++ b/drivers/accel/ivpu/ivpu_hw_mtl.c @@ -403,11 +403,6 @@ static int ivpu_boot_host_ss_axi_enable(struct ivpu_device *vdev) return ivpu_boot_host_ss_axi_drive(vdev, true); } -static int ivpu_boot_host_ss_axi_disable(struct ivpu_device *vdev) -{ - return ivpu_boot_host_ss_axi_drive(vdev, false); -} - static int ivpu_boot_host_ss_top_noc_drive(struct ivpu_device *vdev, bool enable) { int ret; @@ -441,11 +436,6 @@ static int ivpu_boot_host_ss_top_noc_enable(struct ivpu_device *vdev) return ivpu_boot_host_ss_top_noc_drive(vdev, true); } -static int ivpu_boot_host_ss_top_noc_disable(struct ivpu_device *vdev) -{ - return ivpu_boot_host_ss_top_noc_drive(vdev, false); -} - static void ivpu_boot_pwr_island_trickle_drive(struct ivpu_device *vdev, bool enable) { u32 val = REGV_RD32(MTL_VPU_HOST_SS_AON_PWR_ISLAND_TRICKLE_EN0); @@ -504,16 +494,6 @@ static void ivpu_boot_dpu_active_drive(struct ivpu_device *vdev, bool enable) REGV_WR32(MTL_VPU_HOST_SS_AON_DPU_ACTIVE, val); } -static int ivpu_boot_pwr_domain_disable(struct ivpu_device *vdev) -{ - ivpu_boot_dpu_active_drive(vdev, false); - ivpu_boot_pwr_island_isolation_drive(vdev, true); - ivpu_boot_pwr_island_trickle_drive(vdev, false); - ivpu_boot_pwr_island_drive(vdev, false); - - return ivpu_boot_wait_for_pwr_island_status(vdev, 0x0); -} - static int ivpu_boot_pwr_domain_enable(struct ivpu_device *vdev) { int ret; @@ -797,21 +777,8 @@ static int ivpu_hw_mtl_power_down(struct ivpu_device *vdev) { int ret = 0; - /* FPGA requires manual clearing of IP_Reset bit by enabling quiescent state */ - if (ivpu_is_fpga(vdev)) { - if (ivpu_boot_host_ss_top_noc_disable(vdev)) { - ivpu_err(vdev, "Failed to disable TOP NOC\n"); - ret = -EIO; - } - - if (ivpu_boot_host_ss_axi_disable(vdev)) { - ivpu_err(vdev, "Failed to disable AXI\n"); - ret = -EIO; - } - } - - if (ivpu_boot_pwr_domain_disable(vdev)) { - ivpu_err(vdev, "Failed to disable power domain\n"); + if (ivpu_hw_mtl_reset(vdev)) { + ivpu_err(vdev, "Failed to reset the VPU\n"); ret = -EIO; } From 3d8b2727d8de4720c6865917dd05249b1f449f34 Mon Sep 17 00:00:00 2001 From: Stanislaw Gruszka Date: Thu, 23 Mar 2023 13:55:01 +0100 Subject: [PATCH 05/11] accel/ivpu: Disable buttress on device removal Use pci_set_power_state() to disable buttress when device is removed. This is workaround of hardware bug that hangs the system. Additionally not disabling buttress prevents CPU enter deeper Pkg-C states when the driver is unloaded or fail to probe. Fixes: 35b137630f08 ("accel/ivpu: Introduce a new DRM driver for Intel VPU") Signed-off-by: Stanislaw Gruszka Reviewed-by: Jeffrey Hugo Signed-off-by: Jacek Lawrynowicz Link: https://patchwork.freedesktop.org/patch/msgid/20230323125504.2586442-6-stanislaw.gruszka@linux.intel.com --- drivers/accel/ivpu/ivpu_drv.c | 4 ++++ drivers/accel/ivpu/ivpu_drv.h | 1 + drivers/accel/ivpu/ivpu_hw_mtl.c | 1 + 3 files changed, 6 insertions(+) diff --git a/drivers/accel/ivpu/ivpu_drv.c b/drivers/accel/ivpu/ivpu_drv.c index 70245cf84593..6a320a73e3cc 100644 --- a/drivers/accel/ivpu/ivpu_drv.c +++ b/drivers/accel/ivpu/ivpu_drv.c @@ -569,6 +569,8 @@ err_mmu_gctx_fini: ivpu_mmu_global_context_fini(vdev); err_power_down: ivpu_hw_power_down(vdev); + if (IVPU_WA(d3hot_after_power_off)) + pci_set_power_state(to_pci_dev(vdev->drm.dev), PCI_D3hot); err_xa_destroy: xa_destroy(&vdev->submitted_jobs_xa); xa_destroy(&vdev->context_xa); @@ -579,6 +581,8 @@ static void ivpu_dev_fini(struct ivpu_device *vdev) { ivpu_pm_disable(vdev); ivpu_shutdown(vdev); + if (IVPU_WA(d3hot_after_power_off)) + pci_set_power_state(to_pci_dev(vdev->drm.dev), PCI_D3hot); ivpu_job_done_thread_fini(vdev); ivpu_pm_cancel_recovery(vdev); diff --git a/drivers/accel/ivpu/ivpu_drv.h b/drivers/accel/ivpu/ivpu_drv.h index ef12a38e06e1..d3013fbd13b3 100644 --- a/drivers/accel/ivpu/ivpu_drv.h +++ b/drivers/accel/ivpu/ivpu_drv.h @@ -74,6 +74,7 @@ struct ivpu_wa_table { bool punit_disabled; bool clear_runtime_mem; + bool d3hot_after_power_off; }; struct ivpu_hw_info; diff --git a/drivers/accel/ivpu/ivpu_hw_mtl.c b/drivers/accel/ivpu/ivpu_hw_mtl.c index 70ca6de78060..133ba33d2866 100644 --- a/drivers/accel/ivpu/ivpu_hw_mtl.c +++ b/drivers/accel/ivpu/ivpu_hw_mtl.c @@ -101,6 +101,7 @@ static void ivpu_hw_wa_init(struct ivpu_device *vdev) { vdev->wa.punit_disabled = ivpu_is_fpga(vdev); vdev->wa.clear_runtime_mem = false; + vdev->wa.d3hot_after_power_off = true; } static void ivpu_hw_timeouts_init(struct ivpu_device *vdev) From 392b35bf92605deb620a9b0c94c14cad1c69fd34 Mon Sep 17 00:00:00 2001 From: Stanislaw Gruszka Date: Thu, 23 Mar 2023 13:55:02 +0100 Subject: [PATCH 06/11] accel/ivpu: Remove support for 1 tile SKUs The support for single tile SKUs was dropped from MTL. Note that we can still boot the VPU with 1-tile work point config - this is independent from number of tiles present in the VPU. Co-developed-by: Andrzej Kacprowski Signed-off-by: Andrzej Kacprowski Signed-off-by: Stanislaw Gruszka Reviewed-by: Jeffrey Hugo Signed-off-by: Jacek Lawrynowicz Link: https://patchwork.freedesktop.org/patch/msgid/20230323125504.2586442-7-stanislaw.gruszka@linux.intel.com --- drivers/accel/ivpu/ivpu_hw_mtl.c | 57 ++++++++++---------------------- 1 file changed, 17 insertions(+), 40 deletions(-) diff --git a/drivers/accel/ivpu/ivpu_hw_mtl.c b/drivers/accel/ivpu/ivpu_hw_mtl.c index 133ba33d2866..98c8a4aa25f0 100644 --- a/drivers/accel/ivpu/ivpu_hw_mtl.c +++ b/drivers/accel/ivpu/ivpu_hw_mtl.c @@ -12,20 +12,20 @@ #include "ivpu_mmu.h" #include "ivpu_pm.h" -#define TILE_FUSE_ENABLE_BOTH 0x0 -#define TILE_FUSE_ENABLE_UPPER 0x1 -#define TILE_FUSE_ENABLE_LOWER 0x2 - -#define TILE_SKU_BOTH_MTL 0x3630 -#define TILE_SKU_LOWER_MTL 0x3631 -#define TILE_SKU_UPPER_MTL 0x3632 +#define TILE_FUSE_ENABLE_BOTH 0x0 +#define TILE_SKU_BOTH_MTL 0x3630 /* Work point configuration values */ -#define WP_CONFIG_1_TILE_5_3_RATIO 0x0101 -#define WP_CONFIG_1_TILE_4_3_RATIO 0x0102 -#define WP_CONFIG_2_TILE_5_3_RATIO 0x0201 -#define WP_CONFIG_2_TILE_4_3_RATIO 0x0202 -#define WP_CONFIG_0_TILE_PLL_OFF 0x0000 +#define CONFIG_1_TILE 0x01 +#define CONFIG_2_TILE 0x02 +#define PLL_RATIO_5_3 0x01 +#define PLL_RATIO_4_3 0x02 +#define WP_CONFIG(tile, ratio) (((tile) << 8) | (ratio)) +#define WP_CONFIG_1_TILE_5_3_RATIO WP_CONFIG(CONFIG_1_TILE, PLL_RATIO_5_3) +#define WP_CONFIG_1_TILE_4_3_RATIO WP_CONFIG(CONFIG_1_TILE, PLL_RATIO_4_3) +#define WP_CONFIG_2_TILE_5_3_RATIO WP_CONFIG(CONFIG_2_TILE, PLL_RATIO_5_3) +#define WP_CONFIG_2_TILE_4_3_RATIO WP_CONFIG(CONFIG_2_TILE, PLL_RATIO_4_3) +#define WP_CONFIG_0_TILE_PLL_OFF WP_CONFIG(0, 0) #define PLL_REF_CLK_FREQ (50 * 1000000) #define PLL_SIMULATION_FREQ (10 * 1000000) @@ -219,7 +219,8 @@ static int ivpu_pll_drive(struct ivpu_device *vdev, bool enable) config = 0; } - ivpu_dbg(vdev, PM, "PLL workpoint request: %d Hz\n", PLL_RATIO_TO_FREQ(target_ratio)); + ivpu_dbg(vdev, PM, "PLL workpoint request: config 0x%04x pll ratio 0x%x\n", + config, target_ratio); ret = ivpu_pll_cmd_send(vdev, hw->pll.min_ratio, hw->pll.max_ratio, target_ratio, config); if (ret) { @@ -610,34 +611,10 @@ static int ivpu_boot_d0i3_drive(struct ivpu_device *vdev, bool enable) static int ivpu_hw_mtl_info_init(struct ivpu_device *vdev) { struct ivpu_hw_info *hw = vdev->hw; - u32 tile_fuse; - tile_fuse = REGB_RD32(MTL_BUTTRESS_TILE_FUSE); - if (!REG_TEST_FLD(MTL_BUTTRESS_TILE_FUSE, VALID, tile_fuse)) - ivpu_warn(vdev, "Tile Fuse: Invalid (0x%x)\n", tile_fuse); - - hw->tile_fuse = REG_GET_FLD(MTL_BUTTRESS_TILE_FUSE, SKU, tile_fuse); - switch (hw->tile_fuse) { - case TILE_FUSE_ENABLE_LOWER: - hw->sku = TILE_SKU_LOWER_MTL; - hw->config = WP_CONFIG_1_TILE_5_3_RATIO; - ivpu_dbg(vdev, MISC, "Tile Fuse: Enable Lower\n"); - break; - case TILE_FUSE_ENABLE_UPPER: - hw->sku = TILE_SKU_UPPER_MTL; - hw->config = WP_CONFIG_1_TILE_4_3_RATIO; - ivpu_dbg(vdev, MISC, "Tile Fuse: Enable Upper\n"); - break; - case TILE_FUSE_ENABLE_BOTH: - hw->sku = TILE_SKU_BOTH_MTL; - hw->config = WP_CONFIG_2_TILE_5_3_RATIO; - ivpu_dbg(vdev, MISC, "Tile Fuse: Enable Both\n"); - break; - default: - hw->config = WP_CONFIG_0_TILE_PLL_OFF; - ivpu_dbg(vdev, MISC, "Tile Fuse: Disable\n"); - break; - } + hw->tile_fuse = TILE_FUSE_ENABLE_BOTH; + hw->sku = TILE_SKU_BOTH_MTL; + hw->config = WP_CONFIG_2_TILE_4_3_RATIO; ivpu_pll_init_frequency_ratios(vdev); From 252776b2973ea8c7e2e6db09993e6f33c81b776e Mon Sep 17 00:00:00 2001 From: Stanislaw Gruszka Date: Thu, 23 Mar 2023 13:55:03 +0100 Subject: [PATCH 07/11] accel/ivpu: Fix VPU clock calculation The driver calculates the wrong frequency because it ignores the workpoint config and this cause undesired power/performance characteristics. Fix this by using the workpoint config in the freq calculations. Fixes: 35b137630f08 ("accel/ivpu: Introduce a new DRM driver for Intel VPU") Co-developed-by: Andrzej Kacprowski Signed-off-by: Andrzej Kacprowski Signed-off-by: Stanislaw Gruszka Reviewed-by: Jeffrey Hugo Signed-off-by: Jacek Lawrynowicz Link: https://patchwork.freedesktop.org/patch/msgid/20230323125504.2586442-8-stanislaw.gruszka@linux.intel.com --- drivers/accel/ivpu/ivpu_hw_mtl.c | 16 ++++++++++++++-- 1 file changed, 14 insertions(+), 2 deletions(-) diff --git a/drivers/accel/ivpu/ivpu_hw_mtl.c b/drivers/accel/ivpu/ivpu_hw_mtl.c index 98c8a4aa25f0..382ec127be8e 100644 --- a/drivers/accel/ivpu/ivpu_hw_mtl.c +++ b/drivers/accel/ivpu/ivpu_hw_mtl.c @@ -29,7 +29,6 @@ #define PLL_REF_CLK_FREQ (50 * 1000000) #define PLL_SIMULATION_FREQ (10 * 1000000) -#define PLL_RATIO_TO_FREQ(x) ((x) * PLL_REF_CLK_FREQ) #define PLL_DEFAULT_EPP_VALUE 0x80 #define TIM_SAFE_ENABLE 0xf1d0dead @@ -789,6 +788,19 @@ static void ivpu_hw_mtl_wdt_disable(struct ivpu_device *vdev) REGV_WR32(MTL_VPU_CPU_SS_TIM_GEN_CONFIG, val); } +static u32 ivpu_hw_mtl_pll_to_freq(u32 ratio, u32 config) +{ + u32 pll_clock = PLL_REF_CLK_FREQ * ratio; + u32 cpu_clock; + + if ((config & 0xff) == PLL_RATIO_4_3) + cpu_clock = pll_clock * 2 / 4; + else + cpu_clock = pll_clock * 2 / 5; + + return cpu_clock; +} + /* Register indirect accesses */ static u32 ivpu_hw_mtl_reg_pll_freq_get(struct ivpu_device *vdev) { @@ -800,7 +812,7 @@ static u32 ivpu_hw_mtl_reg_pll_freq_get(struct ivpu_device *vdev) if (!ivpu_is_silicon(vdev)) return PLL_SIMULATION_FREQ; - return PLL_RATIO_TO_FREQ(pll_curr_ratio); + return ivpu_hw_mtl_pll_to_freq(pll_curr_ratio, vdev->hw->config); } static u32 ivpu_hw_mtl_reg_telemetry_offset_get(struct ivpu_device *vdev) From 34224133458ca48caf2b57df335e870f1abb7af5 Mon Sep 17 00:00:00 2001 From: Andrzej Kacprowski Date: Thu, 23 Mar 2023 13:55:04 +0100 Subject: [PATCH 08/11] accel/ivpu: Fix IPC buffer header status field value IPC messages transmitted to the device must be marked as allocated - status field must be set to 1. The VPU driver has IVPU_IPC_HDR_ALLOCATED incorrectly defined. Future VPU firmware versions will reject all IPC messages with invalid status and will not work with a VPU driver that is missing this fix. Fixes: 5d7422cfb498 ("accel/ivpu: Add IPC driver and JSM messages") Signed-off-by: Andrzej Kacprowski Signed-off-by: Stanislaw Gruszka Reviewed-by: Jeffrey Hugo Signed-off-by: Jacek Lawrynowicz Link: https://patchwork.freedesktop.org/patch/msgid/20230323125504.2586442-9-stanislaw.gruszka@linux.intel.com --- drivers/accel/ivpu/ivpu_ipc.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/accel/ivpu/ivpu_ipc.h b/drivers/accel/ivpu/ivpu_ipc.h index 9838202ecfad..68f5b6668e00 100644 --- a/drivers/accel/ivpu/ivpu_ipc.h +++ b/drivers/accel/ivpu/ivpu_ipc.h @@ -21,7 +21,7 @@ struct ivpu_bo; #define IVPU_IPC_ALIGNMENT 64 #define IVPU_IPC_HDR_FREE 0 -#define IVPU_IPC_HDR_ALLOCATED 0 +#define IVPU_IPC_HDR_ALLOCATED 1 /** * struct ivpu_ipc_hdr - The IPC message header structure, exchanged From 30fb97ba4a8e082ba0a5432479d6995472edbd7b Mon Sep 17 00:00:00 2001 From: Hans de Goede Date: Sun, 26 Mar 2023 22:54:33 +0200 Subject: [PATCH 09/11] drm/nouveau/kms: Fix backlight registration The nouveau code used to call drm_fb_helper_initial_config() from nouveau_fbcon_init() before calling drm_dev_register(). This would probe all connectors so that drm_connector->status could be used during backlight registration which runs from nouveau_connector_late_register(). After commit 4a16dd9d18a0 ("drm/nouveau/kms: switch to drm fbdev helpers") the fbdev emulation code, which now is a drm-client, can only run after drm_dev_register(). So during backlight registration the connectors are not probed yet and the drm_connector->status == connected check in nv50_backlight_init() would now always fail. Replace the drm_connector->status == connected check with a drm_helper_probe_detect() == connected check to fix nv_backlight no longer getting registered because of this. Fixes: 4a16dd9d18a0 ("drm/nouveau/kms: switch to drm fbdev helpers") Link: https://gitlab.freedesktop.org/drm/nouveau/-/issues/202 Link: https://bugzilla.redhat.com/show_bug.cgi?id=2181941 Signed-off-by: Hans de Goede Reviewed-by: Lyude Paul Link: https://patchwork.freedesktop.org/patch/msgid/20230326205433.36485-1-hdegoede@redhat.com --- drivers/gpu/drm/nouveau/nouveau_backlight.c | 7 ++++++- 1 file changed, 6 insertions(+), 1 deletion(-) diff --git a/drivers/gpu/drm/nouveau/nouveau_backlight.c b/drivers/gpu/drm/nouveau/nouveau_backlight.c index 40409a29f5b6..91b5ecc57538 100644 --- a/drivers/gpu/drm/nouveau/nouveau_backlight.c +++ b/drivers/gpu/drm/nouveau/nouveau_backlight.c @@ -33,6 +33,7 @@ #include #include #include +#include #include "nouveau_drv.h" #include "nouveau_reg.h" @@ -299,8 +300,12 @@ nv50_backlight_init(struct nouveau_backlight *bl, struct nouveau_drm *drm = nouveau_drm(nv_encoder->base.base.dev); struct nvif_object *device = &drm->client.device.object; + /* + * Note when this runs the connectors have not been probed yet, + * so nv_conn->base.status is not set yet. + */ if (!nvif_rd32(device, NV50_PDISP_SOR_PWM_CTL(ffs(nv_encoder->dcb->or) - 1)) || - nv_conn->base.status != connector_status_connected) + drm_helper_probe_detect(&nv_conn->base, NULL, false) != connector_status_connected) return -ENODEV; if (nv_conn->type == DCB_CONNECTOR_eDP) { From 4453545b5b4c3eff941f69a5530f916d899db025 Mon Sep 17 00:00:00 2001 From: David Gow Date: Wed, 29 Mar 2023 14:55:32 +0800 Subject: [PATCH 10/11] drm: buddy_allocator: Fix buddy allocator init on 32-bit systems MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit The drm buddy allocator tests were broken on 32-bit systems, as rounddown_pow_of_two() takes a long, and the buddy allocator handles 64-bit sizes even on 32-bit systems. This can be reproduced with the drm_buddy_allocator KUnit tests on i386: ./tools/testing/kunit/kunit.py run --arch i386 \ --kunitconfig ./drivers/gpu/drm/tests drm_buddy (It results in kernel BUG_ON() when too many blocks are created, due to the block size being too small.) This was independently uncovered (and fixed) by Luís Mendes, whose patch added a new u64 variant of rounddown_pow_of_two(). This version instead recalculates the size based on the order. Reported-by: Luís Mendes Link: https://lore.kernel.org/lkml/CAEzXK1oghXAB_KpKpm=-CviDQbNaH0qfgYTSSjZgvvyj4U78AA@mail.gmail.com/T/ Signed-off-by: David Gow Acked-by: Christian König Reviewed-by: Arunpravin Paneer Selvam Link: https://patchwork.freedesktop.org/patch/msgid/20230329065532.2122295-1-davidgow@google.com Signed-off-by: Christian König --- drivers/gpu/drm/drm_buddy.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/drivers/gpu/drm/drm_buddy.c b/drivers/gpu/drm/drm_buddy.c index 3d1f50f481cf..7098f125b54a 100644 --- a/drivers/gpu/drm/drm_buddy.c +++ b/drivers/gpu/drm/drm_buddy.c @@ -146,8 +146,8 @@ int drm_buddy_init(struct drm_buddy *mm, u64 size, u64 chunk_size) unsigned int order; u64 root_size; - root_size = rounddown_pow_of_two(size); - order = ilog2(root_size) - ilog2(chunk_size); + order = ilog2(size) - ilog2(chunk_size); + root_size = chunk_size << order; root = drm_block_alloc(mm, NULL, order, offset); if (!root) From 25bbe844ef5c4fb4d7d8dcaa0080f922b7cd3a16 Mon Sep 17 00:00:00 2001 From: David Gow Date: Wed, 29 Mar 2023 14:55:34 +0800 Subject: [PATCH 11/11] drm: test: Fix 32-bit issue in drm_buddy_test MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit The drm_buddy_test KUnit tests verify that returned blocks have sizes which are powers of two using is_power_of_2(). However, is_power_of_2() operations on a 'long', but the block size is a u64. So on systems where long is 32-bit, this can sometimes fail even on correctly sized blocks. This only reproduces randomly, as the parameters passed to the buddy allocator in this test are random. The seed 0xb2e06022 reproduced it fine here. For now, just hardcode an is_power_of_2() implementation using x & (x - 1). Signed-off-by: David Gow Acked-by: Christian König Reviewed-by: Maíra Canal Reviewed-by: Arunpravin Paneer Selvam Link: https://patchwork.freedesktop.org/patch/msgid/20230329065532.2122295-2-davidgow@google.com Signed-off-by: Christian König --- drivers/gpu/drm/tests/drm_buddy_test.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/drivers/gpu/drm/tests/drm_buddy_test.c b/drivers/gpu/drm/tests/drm_buddy_test.c index f8ee714df396..09ee6f6af896 100644 --- a/drivers/gpu/drm/tests/drm_buddy_test.c +++ b/drivers/gpu/drm/tests/drm_buddy_test.c @@ -89,7 +89,8 @@ static int check_block(struct kunit *test, struct drm_buddy *mm, err = -EINVAL; } - if (!is_power_of_2(block_size)) { + /* We can't use is_power_of_2() for a u64 on 32-bit systems. */ + if (block_size & (block_size - 1)) { kunit_err(test, "block size not power of two\n"); err = -EINVAL; }