kernel/linux.git/drivers/gpu/drm/imagination/pvr_queue.c, branch v6.18.21

drm/sched: Rename DRM_GPU_SCHED_STAT_NOMINAL to DRM_GPU_SCHED_STAT_RESET

2025-07-15T11:27:00+00:00

Among the scheduler's statuses, the only one that indicates an error is DRM_GPU_SCHED_STAT_ENODEV. Any status other than DRM_GPU_SCHED_STAT_ENODEV signifies that the operation succeeded and the GPU is in a nominal state. However, to provide more information about the GPU's status, it is needed to convey more information than just "OK". Therefore, rename DRM_GPU_SCHED_STAT_NOMINAL to DRM_GPU_SCHED_STAT_RESET, which better communicates the meaning of this status. The status DRM_GPU_SCHED_STAT_RESET indicates that the GPU has hung, but it has been successfully reset and is now in a nominal state again. Reviewed-by: Philipp Stanner Link: https://lore.kernel.org/r/20250714-sched-skip-reset-v6-1-5c5ba4f55039@igalia.com Signed-off-by: Maíra Canal

drm/sched: Store the drm client_id in drm_sched_fence

2025-05-28T14:15:58+00:00

This will be used in a later commit to trace the drm client_id in some of the gpu_scheduler trace events. This requires changing all the users of drm_sched_job_init to add an extra parameter. The newly added drm_client_id field in the drm_sched_fence is a bit of a duplicate of the owner one. One suggestion I received was to merge those 2 fields - this can't be done right now as amdgpu uses some special values (AMDGPU_FENCE_OWNER_*) that can't really be translated into a client id. Christian is working on getting rid of those; when it's done we should be able to squash owner/drm_client_id together. Reviewed-by: Christian König Signed-off-by: Pierre-Eric Pelloux-Prayer Signed-off-by: Philipp Stanner Link: https://lore.kernel.org/r/20250526125505.2360-3-pierre-eric.pelloux-prayer@amd.com

Merge drm/drm-fixes into drm-misc-fixes

2025-04-08T08:15:47+00:00

Backmerging to get updates from v6.15-rc1. Signed-off-by: Thomas Zimmermann

drm/imagination: take paired job reference

2025-03-25T17:17:37+00:00

For paired jobs, have the fragment job take a reference on the geometry job, so that the geometry job cannot be freed until the fragment job has finished with it. The geometry job structure is accessed when the fragment job is being prepared by the GPU scheduler. Taking the reference prevents the geometry job being freed until the fragment job no longer requires it. Fixes a use after free bug detected by KASAN: [ 124.256386] BUG: KASAN: slab-use-after-free in pvr_queue_prepare_job+0x108/0x868 [powervr] [ 124.264893] Read of size 1 at addr ffff0000084cb960 by task kworker/u16:4/63 Cc: stable@vger.kernel.org Fixes: eaf01ee5ba28 ("drm/imagination: Implement job submission and scheduling") Signed-off-by: Brendan King Reviewed-by: Matt Coster Link: https://lore.kernel.org/r/20250318-ddkopsrc-1337-use-after-free-in-pvr_queue_prepare_job-v1-1-80fb30d044a6@imgtec.com Signed-off-by: Matt Coster

Backmerge tag 'v6.14-rc6' into drm-next

2025-03-11T23:43:12+00:00

This is a backmerge from Linux 6.14-rc6, needed for the nova PR. Signed-off-by: Dave Airlie

drm/imagination: only init job done fences once

2025-03-03T12:10:13+00:00

Ensure job done fences are only initialised once. This fixes a memory manager not clean warning from drm_mm_takedown on module unload. Cc: stable@vger.kernel.org Fixes: eaf01ee5ba28 ("drm/imagination: Implement job submission and scheduling") Signed-off-by: Brendan King Reviewed-by: Matt Coster Link: https://patchwork.freedesktop.org/patch/msgid/20250226-init-done-fences-once-v2-1-c1b2f556b329@imgtec.com Signed-off-by: Matt Coster

drm/imagination: avoid deadlock on fence release

2025-03-03T12:07:56+00:00

Do scheduler queue fence release processing on a workqueue, rather than in the release function itself. Fixes deadlock issues such as the following: [ 607.400437] ============================================ [ 607.405755] WARNING: possible recursive locking detected [ 607.415500] -------------------------------------------- [ 607.420817] weston:zfq0/24149 is trying to acquire lock: [ 607.426131] ffff000017d041a0 (reservation_ww_class_mutex){+.+.}-{3:3}, at: pvr_gem_object_vunmap+0x40/0xc0 [powervr] [ 607.436728] but task is already holding lock: [ 607.442554] ffff000017d105a0 (reservation_ww_class_mutex){+.+.}-{3:3}, at: dma_buf_ioctl+0x250/0x554 [ 607.451727] other info that might help us debug this: [ 607.458245] Possible unsafe locking scenario: [ 607.464155] CPU0 [ 607.466601] ---- [ 607.469044] lock(reservation_ww_class_mutex); [ 607.473584] lock(reservation_ww_class_mutex); [ 607.478114] *** DEADLOCK *** Cc: stable@vger.kernel.org Fixes: eaf01ee5ba28 ("drm/imagination: Implement job submission and scheduling") Signed-off-by: Brendan King Reviewed-by: Matt Coster Link: https://patchwork.freedesktop.org/patch/msgid/20250226-fence-release-deadlock-v2-1-6fed2fc1fe88@imgtec.com Signed-off-by: Matt Coster

drm/sched: Use struct for drm_sched_init() params

2025-02-12T10:59:52+00:00

drm_sched_init() has a great many parameters and upcoming new functionality for the scheduler might add even more. Generally, the great number of parameters reduces readability and has already caused one missnaming, addressed in: commit 6f1cacf4eba7 ("drm/nouveau: Improve variable name in nouveau_sched_init()"). Introduce a new struct for the scheduler init parameters and port all users. Reviewed-by: Liviu Dudau Acked-by: Matthew Brost # for Xe Reviewed-by: Boris Brezillon # for Panfrost and Panthor Reviewed-by: Christian Gmeiner # for Etnaviv Reviewed-by: Frank Binns # for Imagination Reviewed-by: Tvrtko Ursulin # for Sched Reviewed-by: Maíra Canal # for v3d Reviewed-by: Danilo Krummrich Reviewed-by: Lizhi Hou # for amdxdna Signed-off-by: Philipp Stanner Link: https://patchwork.freedesktop.org/patch/msgid/20250211111422.21235-2-phasta@kernel.org

drm/sched: add optional errno to drm_sched_start()

2024-09-06T16:05:52+00:00

The current implementation of drm_sched_start uses a hardcoded -ECANCELED to dispose of a job when the parent/hw fence is NULL. This results in drm_sched_job_done being called with -ECANCELED for each job with a NULL parent in the pending list, making it difficult to distinguish between recovery methods, whether a queue reset or a full GPU reset was used. To improve this, we first try a soft recovery for timeout jobs and use the error code -ENODATA. If soft recovery fails, we proceed with a queue reset, where the error code remains -ENODATA for the job. Finally, for a full GPU reset, we use error codes -ECANCELED or -ETIME. This patch adds an error code parameter to drm_sched_start, allowing us to differentiate between queue reset and GPU reset failures. This enables user mode and test applications to validate the expected correctness of the requested operation. After a successful queue reset, the only way to continue normal operation is to call drm_sched_job_done with the specific error code -ENODATA. v1: Initial implementation by Jesse utilized amdgpu_device_lock_reset_domain and amdgpu_device_unlock_reset_domain to allow user mode to track the queue reset status and distinguish between queue reset and GPU reset. v2: Christian suggested using the error codes -ENODATA for queue reset and -ECANCELED or -ETIME for GPU reset, returned to amdgpu_cs_wait_ioctl. v3: To meet the requirements, we introduce a new function drm_sched_start_ex with an additional parameter to set dma_fence_set_error, allowing us to handle the specific error codes appropriately and dispose of bad jobs with the selected error code depending on whether it was a queue reset or GPU reset. v4: Alex suggested using a new name, drm_sched_start_with_recovery_error, which more accurately describes the function's purpose. Additionally, it was recommended to add documentation details about the new method. v5: Fixed declaration of new function drm_sched_start_with_recovery_error.(Alex) v6 (chk): rebase on upstream changes, cleanup the commit message, drop the new function again and update all callers, apply the errno also to scheduler fences with hw fences v7 (chk): rebased Signed-off-by: Jesse Zhang Signed-off-by: Vitaly Prosyak Signed-off-by: Christian König Acked-by: Daniel Vetter Reviewed-by: Alex Deucher Link: https://patchwork.freedesktop.org/patch/msgid/20240826122541.85663-1-christian.koenig@amd.com

drm/scheduler: remove full_recover from drm_sched_start

2024-07-25T12:05:12+00:00

This was basically just another one of amdgpus hacks. The parameter allowed to restart the scheduler without turning fence signaling on again. That this is absolutely not a good idea should be obvious by now since the fences will then just sit there and never signal. While at it cleanup the code a bit. Signed-off-by: Christian König Reviewed-by: Matthew Brost Link: https://patchwork.freedesktop.org/patch/msgid/20240722083816.99685-1-christian.koenig@amd.com