[PATCH 0/3] Rework work queue usage

[Lucas De Marchi]

+Tejun

On Thu, Mar 28, 2024 at 11:21:44AM -0700, Matthew Brost wrote:
>Avoid sleeping or grabbing locks in work queues shared with the system.
>Recent changes to work queues [1] have exposed deadlocks [2] in Xe.
>
>I believe only patch #1 is required to unblock CI in 6.9rc1 [2] but
>likely all patches should be included in stable kernels.
>
>Matt
>
>[1] https://lore.kernel.org/lkml/Ze7Jz9cQcUMtCqfk@slm.duckdns.org/
>[2] https://intel-gfx-ci.01.org/tree/intel-xe/xe-984-093d82c02921344d8d400989af156aaf79b9d961/bat-adlp-7/igt_runner0.txt

I think we need some of this information in the commit message in patch
1. Because patch 1 simply says it's moving to a device private wq to
avoid hogging the system one, but the issue is much more serious.

Also, is the "Fixes:"  really correct? It seems more like a regression
from the wq changes and there could be other drivers showing similar
issues now. But it could alos be my lack of understanding of the real
issue.

Lucas De Marchi

>
>Matthew Brost (3):
>  drm/xe: Use ordered wq for preempt fence waiting
>  drm/xe: Use device, gt ordered work queues for resource cleanup
>  drm/xe: Use ordered WQ for TLB invalidation fences
>
> drivers/gpu/drm/xe/xe_device.c        | 7 ++++++-
> drivers/gpu/drm/xe/xe_device_types.h  | 8 +++++++-
> drivers/gpu/drm/xe/xe_execlist.c      | 2 +-
> drivers/gpu/drm/xe/xe_gt_types.h      | 5 ++++-
> drivers/gpu/drm/xe/xe_guc_submit.c    | 2 +-
> drivers/gpu/drm/xe/xe_preempt_fence.c | 2 +-
> drivers/gpu/drm/xe/xe_pt.c            | 2 +-
> drivers/gpu/drm/xe/xe_vm.c            | 4 ++--
> 8 files changed, 23 insertions(+), 9 deletions(-)
>
>-- 
>2.34.1
>