[PATCH v3 0/3] drm/xe: Fix survivability

[Lucas De Marchi <lucas.demarchi@intel.com>]

It turns out commit d40f275d96e8 ("drm/xe: Move survivability entirely
to xe_pci") did a bad job moving things to xe_pci. The fix provided by
Riana in 20250306055407.511405-1-riana.tauro@intel.com fixes it
partially, but injecting a failure in xe_pcode_probe_early still causes
the kernel to give warnings/errors.

Correct the course and better split what is done in xe_pci vs xe_device.
This time, also add a patch to test we can handle errors in
xe_pcode_probe_early().

Entering survivability mode was tested with an additional one line to
change the return of xe_survivability_mode_capable(). If we want to
inject error, we'd need to change it's return type, but there's also
another patch series to force it via configs, so this doesn't seem very
important right now.

Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
---
Changes in v3:
- Add another fix for heci
- Rename function according to review feedback
- Link to v2: https://lore.kernel.org/r/20250311-fix-survivability-v2-0-729ce081155e@intel.com

Changes in v2:
- Cover more error injections in the second patch
- Link to v1: https://lore.kernel.org/r/20250310-fix-survivability-v1-0-7af31432bbd0@intel.com

---
Lucas De Marchi (3):
      drm/xe: Move survivability back to xe
      drm/xe: Set survivability mode before heci init
      drm/xe: Allow to inject error in early probe

 drivers/gpu/drm/xe/xe_device.c             | 16 +++++++++++++++-
 drivers/gpu/drm/xe/xe_mmio.c               |  1 +
 drivers/gpu/drm/xe/xe_pci.c                | 16 +++++++---------
 drivers/gpu/drm/xe/xe_pcode.c              |  2 ++
 drivers/gpu/drm/xe/xe_survivability_mode.c | 29 +++++++++++++++++++++--------
 drivers/gpu/drm/xe/xe_survivability_mode.h |  1 -
 6 files changed, 46 insertions(+), 19 deletions(-)
---
base-commit: aba848f9b752cf51474c0c3b1abcf0f572f774dc
change-id: 20250310-fix-survivability-703246c0c480

Best regards,
-- 
Lucas De Marchi <lucas.demarchi@intel.com>