linux.git/drivers/gpu/drm/amd/amdgpu/amdgpu_userq.c, branch v7.1-rc7

drm/amdgpu/userq: move wptr_obj cleanup in mqd_destroy

2026-06-03T18:55:48+00:00

In case when queue_create fails and mqd has already been
allocated and hence wptr_obj is not cleaned up.

So moving that cleanup part to mqd_destroy so it takes
care of all the cases of clean up and during tear down of
the queue.

Signed-off-by: Sunil Khatri 
Reviewed-by: Christian König 
Signed-off-by: Alex Deucher 
(cherry picked from commit 43355f62cd2ef5386c2693df537c232ea0f2ce6c)

drm/amdgpu/userq: remove the vital queue unmap logging

2026-06-03T18:55:23+00:00

Mesa userqueues free does not wait for the free to complete and go ahead
in unmapping the vital bos while kernel is still in queue free and
corresponding cleanup.

So ideally we don't need the logging for that and hence remove the warn
message as this is expected behaviour and functionally, we are making
sure to wait for the required fences before unmap.

Signed-off-by: Sunil Khatri 
Reviewed-by: Christian König 
Signed-off-by: Alex Deucher 
(cherry picked from commit 758a868043dcb07eca923bc451c16da3e73dc47c)

drm/amdgpu: unmap userq for evicting user queue

2026-06-03T18:48:49+00:00

If the driver only preempts queues, there can still be inflight waves,
pending dispatch state, or resume/redispatch possibility tied to the
same queue. Then the VM/TTM side may proceed to move/unmap queue related
BOs during evicting userq objects while shader TCP clients still need to
access them.

So for eviction, unmap is safer because it makes the queue nonrunnable
before memory backing is invalidated. Meanwhile, for a idle queue it's
more sutiable for unmapping it rather preempt and unmapping also can
save more processing time than preempt.

Signed-off-by: Prike Liang 
Reviewed-by: Christian König 
Reviewed-by: Alex Deucher 
Signed-off-by: Alex Deucher 
(cherry picked from commit d87c9d86727a0bcc95c3009a213a1b27a11b691e)

drm/amdgpu/userq: use array instead of list for userq_vas

2026-05-27T16:01:42+00:00

Use arrays instead of list for userq_vas since we have fixed no
of bos. Also, we dont have to worry to free that memory later
since this array would be free along with queue only.

Signed-off-by: Sunil Khatri 
Reviewed-by: Christian König 
Signed-off-by: Alex Deucher 
(cherry picked from commit ef7dc711a664b0c548ecfdf13a00436b7446b8e7)

drm/amdgpu/userq: move mqd_destroy to later stage to keep core obj valid

2026-05-27T16:01:35+00:00

mqd_destroy cleans up queue core objects like mqd and fw_object
which are needed for any pending fence to signal properly.

Signed-off-by: Sunil Khatri 
Reviewed-by: Christian König 
Signed-off-by: Alex Deucher 
(cherry picked from commit 4ad65d610096498c8e265615aba42b3c47441bb5)

drm/amdgpu/userq: remove amdgpu_userq_create/destroy_object wrapper

2026-05-27T16:01:00+00:00

Remove the amdgpu_userq_create/destroy_object wrappers and
use directly the kernel bo allocation function which does all the
things which are done in wrapper.

Signed-off-by: Sunil Khatri 
Suggested-by: Christian König 
Reviewed-by: Christian König 
Signed-off-by: Alex Deucher 
(cherry picked from commit deb02080ca5d3f015cf71e56067a39ef2f141998)

drm/amdgpu/userq: make sure queue is valid in the hang_detect_work

2026-05-27T15:58:31+00:00

Thread 1: Running amdgpu_userq_destroy which eventually remove
the queue from door bell and set userq_mgr = NULL.

Thread2: An interrupt might have scheduled the hang_detect_work
which still need userq_mgr to be valid but could get an NULL
ptrs.

To fix that make sure we cancel the hang_detect_work again before
setting userq_mgr to NULL.

Along with that we also need all the queue va to remain valid till
we could be running anything on the queue and hence moving the
userq_va post hang_detect handler is cancelled.

Signed-off-by: Sunil Khatri 
Reviewed-by: Christian König 
Signed-off-by: Alex Deucher 
(cherry picked from commit 1a66ceb98b137d18d303b9889f0e7d8c4db73943)

drm/amdgpu/userq: reserve root bo without interruption

2026-05-27T15:58:25+00:00

Fix the code to make it an uninterruptible reservation
for root bo.

Signed-off-by: Sunil Khatri 
Reviewed-by: Christian König 
Signed-off-by: Alex Deucher 
(cherry picked from commit d409ab4e387d94b2e593d558b54b7bfd315e0e75)

drm/amdgpu: simplify return value in amdgpu_userq_get_doorbell_index

2026-05-27T15:58:10+00:00

amdgpu_userq_get_doorbell_index returns a uint64 type index
as well as a int type failure values. Simplifying this and
using a int type return value and getting the index in input pointer
of type uint64 type.

Also since it's used at once place making it static would be better.

Signed-off-by: Sunil Khatri 
Reviewed-by: Christian König 
Signed-off-by: Alex Deucher 
(cherry picked from commit e947ec9d0529d5f93dbdb33cd197347f6a7b2922)

drm/amdgpu/userq: Fix the mutex_init cleanup for fence_drv_lock

2026-05-27T15:55:38+00:00

mutex fence_drv_lock is destroyed in amdgpu_userq_fence_driver_free
also in one of the jump condition mutex_destroy is also called leading
to double mutex_destroy.

So rearranging the code so amdgpu_userq_fence_driver_free takes care
of the clean up along with mutex_destroy.

Signed-off-by: Sunil Khatri 
Reviewed-by: Christian König 
Signed-off-by: Alex Deucher 
(cherry picked from commit 384dbef269d101e5b671fc7b942c56734cd1d186)