Intel-XE Archive on lore.kernel.org
 help / color / mirror / Atom feed
From: Brian Nguyen <brian3.nguyen@intel.com>
To: intel-xe@lists.freedesktop.org
Cc: matthew.brost@intel.com
Subject: [PATCH 0/4] Page-reclaim fixes and PRL stats addition
Date: Wed,  7 Jan 2026 09:04:48 +0800	[thread overview]
Message-ID: <20260107010447.4125005-6-brian3.nguyen@intel.com> (raw)

This series provides fixes to page reclamation and additional stats
for the page reclamation feature.

Patch 1 is just a small fixup to remove a debug comment.

Patch 2 makes the distinction between PRL invalidation done as
part of a typical workflow and invalidation done to abort and
fallback to normal TLB invalidation.

Patch 3 fixes the incorrect handling of larger page sizes for PRL entry
generation. 64KB are just contigious 4KB PTE so only need to add one.
2MB also needs to be handled separately because page directory are
handled differently. Also additional handling of racy pte reads and
page walk completions before PTE.

Patch 4 adds PRL related stats to GT stats to assist in debugging and
validation of the feature.

Modifications from the fixes will slightly conflict with the addition of
stats so it seemed better to push patches as a series together.

Thanks,
Brian

v2:
 - Remove comment erroneously added in initial commit.
 - Add explicit abort func to clearly define invalidation done
   to indicate end of scope and invalidation done to abort PRL.
 - Added PRL_ABORTED_STATS. (Matthew B)

v3:
 - Convert abort function to macro. (Matthew B)
 - Reworked logic for one PRL abort handling. (Matthew B)

Brian Nguyen (4):
  drm/xe: Remove debug comment in page reclaim
  drm/xe: Add explicit abort page reclaim list
  drm/xe: Fix page reclaim entry handling for large pages
  drm/xe: Add page reclamation related stats

 drivers/gpu/drm/xe/xe_gt_stats.c       |  5 ++
 drivers/gpu/drm/xe/xe_gt_stats_types.h |  5 ++
 drivers/gpu/drm/xe/xe_guc_tlb_inval.c  |  3 +
 drivers/gpu/drm/xe/xe_page_reclaim.c   |  2 +-
 drivers/gpu/drm/xe/xe_page_reclaim.h   | 20 ++++++
 drivers/gpu/drm/xe/xe_pt.c             | 84 ++++++++++++++++++--------
 6 files changed, 92 insertions(+), 27 deletions(-)

-- 
2.52.0


             reply	other threads:[~2026-01-07  1:04 UTC|newest]

Thread overview: 12+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-01-07  1:04 Brian Nguyen [this message]
2026-01-07  1:04 ` [PATCH 1/4] drm/xe: Remove debug comment in page reclaim Brian Nguyen
2026-01-07  1:04 ` [PATCH 2/4] drm/xe: Add explicit abort page reclaim list Brian Nguyen
2026-01-07 19:57   ` Matthew Brost
2026-01-07  1:04 ` [PATCH 3/4] drm/xe: Fix page reclaim entry handling for large pages Brian Nguyen
2026-01-08 16:22   ` Matthew Brost
2026-01-07  1:04 ` [PATCH 4/4] drm/xe: Add page reclamation related stats Brian Nguyen
2026-01-08 16:24   ` Matthew Brost
2026-01-07  1:19 ` ✓ CI.KUnit: success for Page-reclaim fixes and PRL stats addition (rev3) Patchwork
2026-01-07  2:02 ` ✓ Xe.CI.BAT: " Patchwork
2026-01-07  4:45 ` ✗ Xe.CI.Full: failure " Patchwork
  -- strict thread matches above, loose matches on Subject: below --
2026-01-05 23:33 [PATCH 0/4] Page-reclaim fixes and PRL stats addition Brian Nguyen

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=20260107010447.4125005-6-brian3.nguyen@intel.com \
    --to=brian3.nguyen@intel.com \
    --cc=intel-xe@lists.freedesktop.org \
    --cc=matthew.brost@intel.com \
    /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 a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox