]> git.ipfire.org Git - thirdparty/kernel/linux.git/commit
accel/amdxdna: Check for device hang on job timeout
authorLizhi Hou <lizhi.hou@amd.com>
Thu, 9 Apr 2026 17:58:26 +0000 (10:58 -0700)
committerLizhi Hou <lizhi.hou@amd.com>
Mon, 13 Apr 2026 16:40:11 +0000 (09:40 -0700)
commit9022f010977fe7b8f3246647fa9eeaa8a84336fe
tree080f4c46f5cd9d46f7177764b72220a2e8747d04
parente96a086ada9112475dc6c8a3a4735291927b5f3c
accel/amdxdna: Check for device hang on job timeout

A job timeout does not necessarily indicate that the device is hung, as
it may still be processing other jobs.

Track whether any jobs have been successfully submitted or completed,
and use this information to determine if the device is making forward
progress. If so, return DRM_GPU_SCHED_STAT_NO_HANG instead of treating
the timeout as a device hang.

In the meanwhile the timeout interval is changed to 2 seconds which meets
the userspace requirement.

Reviewed-by: Mario Limonciello (AMD) <superm1@kernel.org>
Signed-off-by: Lizhi Hou <lizhi.hou@amd.com>
Link: https://patch.msgid.link/20260409175826.195665-1-lizhi.hou@amd.com
drivers/accel/amdxdna/aie2_ctx.c
drivers/accel/amdxdna/aie2_pci.h