[PULL] drm-xe-fixes

Pull-request

https://gitlab.freedesktop.org/drm/xe/kernel.git tags/drm-xe-fixes-2025-03-13

Message

Rodrigo Vivi March 13, 2025, 4:35 p.m. UTC

Hi Dave and Sima,

Here goes xe fixes for this week.

It is worth mention that we are disabling D3Cold on BMG because
we found some bugs where depending on the combination of BMG
card and the host, the PCI upstream port link port re-training
might fail on D3Cold -> D0 blowing things up. But while we
work with that and propagating this protection to the end
users, we are not disabling in our drm-tip to ensure that our
developers and CI can continue working with D3Cold enabled.
Everything we disable behind flags for developers will likely
be forgotten and regressions will pile up and be harder later
to re-enable it.

So, topic/xe-for-CI re-enables D3Cold. If developers ends up
having a bad combination they can workaround it by boot
parameter pcie_port_pm=off or with
echo 0 > /sys/bus/pci/devices/<bdf>/vram_d3cold_threshold

Also, the topic branch solution helps us to avoid adding the
controversial module parameters. But if there's any concern
or issues with this approach, please let me know.

Thanks,
Rodrigo.

drm-xe-fixes-2025-03-13:
- Release guc ids before cancelling work (Tejas)
- Fix new warnings around userptr (Thomas)
- Temporaritly disable D3Cold on BMG (Rodrigo)
- Retry and wait longer for GuC PC to start (Rodrigo)
- Remove redundant check in xe_vm_create_ioctl (Xin)
The following changes since commit 80e54e84911a923c40d7bee33a34c1b4be148d7a:

  Linux 6.14-rc6 (2025-03-09 13:45:25 -1000)

are available in the Git repository at:

  https://gitlab.freedesktop.org/drm/xe/kernel.git tags/drm-xe-fixes-2025-03-13

for you to fetch changes up to f5d4e81774c42d9c2ea3980e570f3330ff2ed5d2:

  drm/xe: remove redundant check in xe_vm_create_ioctl() (2025-03-10 14:01:43 -0400)

----------------------------------------------------------------
- Release guc ids before cancelling work (Tejas)
- Fix new warnings around userptr (Thomas)
- Temporaritly disable D3Cold on BMG (Rodrigo)
- Retry and wait longer for GuC PC to start (Rodrigo)
- Remove redundant check in xe_vm_create_ioctl (Xin)

----------------------------------------------------------------
Rodrigo Vivi (2):
      drm/xe/pm: Temporarily disable D3Cold on BMG
      drm/xe/guc_pc: Retry and wait longer for GuC PC start

Tejas Upadhyay (1):
      drm/xe: Release guc ids before cancelling work

Thomas Hellström (1):
      drm/xe/userptr: Fix an incorrect assert

Xin Wang (1):
      drm/xe: remove redundant check in xe_vm_create_ioctl()

 drivers/gpu/drm/xe/xe_guc_pc.c     | 53 ++++++++++++++++++++++++++++----------
 drivers/gpu/drm/xe/xe_guc_submit.c |  2 +-
 drivers/gpu/drm/xe/xe_hmm.c        |  6 ++++-
 drivers/gpu/drm/xe/xe_pm.c         | 13 +++++++++-
 drivers/gpu/drm/xe/xe_vm.c         |  3 ---
 5 files changed, 58 insertions(+), 19 deletions(-)