]> www.infradead.org Git - users/jedix/linux-maple.git/commit
drm/i915/guc: Flush ct receive tasklet during reset preparation
authorZhanjun Dong <zhanjun.dong@intel.com>
Mon, 4 Nov 2024 21:41:03 +0000 (13:41 -0800)
committerJohn Harrison <John.C.Harrison@Intel.com>
Wed, 6 Nov 2024 20:29:30 +0000 (12:29 -0800)
commitb939a08bc378a7c716ad7a9486b48794b95d22f5
tree5af79e872b43f104370f140ace8635437947d270
parent79367b7a58c82d0b1c0a7b0ef748f7aafa91d048
drm/i915/guc: Flush ct receive tasklet during reset preparation

GuC to host communication is interrupt driven, the handling has 3
parts: interrupt context, tasklet and request queue worker.
During GuC reset prepare, interrupt is disabled before destroy
contexts steps start. The IRQ and worker are flushed to finish
any outstanding in-progress message handling. But, the tasklet
flush is missing, it might causes 2 race conditions:
1. Tasklet runs after IRQ flushed, add request to queue after worker
flush started, causes unexpected G2H message request processing,
meanwhile, reset prepare code already get the context destroyed.
This will causes error reported about bad context state.
(https://gitlab.freedesktop.org/drm/i915/kernel/-/issues/11349 and
https://gitlab.freedesktop.org/drm/i915/kernel/-/issues/12303)
2. Tasklet runs after intel_guc_submission_reset_prepare,
ct_try_receive_message start to run, while intel_uc_reset_prepare
already finished guc sanitize and set ct->enable to false. This will
causes warning on incorrect ct->enable state.
(https://gitlab.freedesktop.org/drm/i915/kernel/-/issues/12439)

Add the missing tasklet flush to flush all 3 parts.

Signed-off-by: Zhanjun Dong <zhanjun.dong@intel.com>
Reviewed-by: Alan Previn <alan.previn.teres.alexis@intel.com>
Signed-off-by: John Harrison <John.C.Harrison@Intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20241104214103.214702-1-zhanjun.dong@intel.com
drivers/gpu/drm/i915/gt/uc/intel_guc_submission.c