All of lore.kernel.org
 help / color / mirror / Atom feed
From: <vitaly.prosyak@amd.com>
To: <igt-dev@lists.freedesktop.org>
Cc: Vitaly Prosyak <vitaly.prosyak@amd.com>
Subject: [PATCH v1 0/2] tests/amdgpu: Add shared lockdep checks for reset tests
Date: Wed, 16 Sep 2026 18:58:45 -0400	[thread overview]
Message-ID: <cover.1789597588.git.vitaly.prosyak@amd.com> (raw)

From: Vitaly Prosyak <vitaly.prosyak@amd.com>

This series extends the existing amdgpu lockdep validation coverage by
moving the common /dev/kmsg and kernel taint checking logic out of
amd_lockdep and into shared amdgpu helpers, then using those helpers in
additional reset-oriented tests.

Patch 1 is a refactor only and makes no functional change to
amd_lockdep. The test still requires CONFIG_LOCKDEP through
/proc/lockdep_stats, opens /dev/kmsg at the start of each
lockdep-sensitive subtest, records the starting taint state, and fails
if a lockdep splat or new TAINT_WARN appears. The helper move keeps that
logic in one place so other amdgpu tests can use the same behavior
without duplicating the plumbing.

Patch 2 enables the shared optional lockdep checks in amd_dispatch,
amd_deadlock, and amd_queue_reset. These tests mostly exercise queue
reset and GPU reset paths, including user queue coverage, so they are a
useful place to catch lockdep issues while reset and recovery paths are
being stressed.

Unlike amd_lockdep, these tests are not dedicated lockdep tests and must
continue to run on kernels without CONFIG_LOCKDEP. The new begin/end
helper checks /proc/lockdep_stats first; when lockdep is unavailable the
check is a no-op, and when lockdep is available each subtest fails if new
lockdep traces or TAINT_WARN are observed.

Vitaly Prosyak (2):
  lib/amdgpu: Add shared lockdep violation helpers
  tests/amdgpu: Check lockdep violations in reset tests

 lib/amdgpu/amd_utils.c         | 100 ++++++++++++++++++++++++++++++
 lib/amdgpu/amd_utils.h         |  14 +++++
 tests/amdgpu/amd_deadlock.c    |  45 ++++++++++++++
 tests/amdgpu/amd_dispatch.c    |  25 ++++++++
 tests/amdgpu/amd_lockdep.c     | 110 ++++-----------------------------
 tests/amdgpu/amd_queue_reset.c |   3 +
 6 files changed, 199 insertions(+), 98 deletions(-)

-- 
2.55.0


             reply	other threads:[~2026-09-16 22:59 UTC|newest]

Thread overview: 8+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-09-16 22:58 vitaly.prosyak [this message]
2026-09-16 22:58 ` [PATCH v1 1/2] lib/amdgpu: Add shared lockdep violation helpers vitaly.prosyak
2026-09-16 22:58 ` [PATCH v1 2/2] tests/amdgpu: Check lockdep violations in reset tests vitaly.prosyak
2026-09-17  1:06   ` Zhang, Jesse(Jie)
2026-09-16 23:50 ` ✓ Xe.CI.BAT: success for tests/amdgpu: Add shared lockdep checks for " Patchwork
2026-09-17  0:26 ` ✓ i915.CI.BAT: " Patchwork
2026-09-17  4:37 ` ✓ Xe.CI.FULL: " Patchwork
2026-09-17  8:11 ` ✓ i915.CI.Full: " Patchwork

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=cover.1789597588.git.vitaly.prosyak@amd.com \
    --to=vitaly.prosyak@amd.com \
    --cc=igt-dev@lists.freedesktop.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.