drm/xe/pf: Use migration-friendly GGTT auto-provisioning

Browse source

Instead of trying very hard to find the largest fair GGTT size that
could be allocated for VFs on the current tile, pick some smaller
rounded down to power-of-two value that is more likely to be
provisioned in the same manner by the other PF instance:

  num VFs | GGTT space (MiB)
  --------+-----------------
   63..57 | 56
   56..29 | 64
   28..15 | 128
   14..8  | 256
    7..4  | 512
    3..2  | 1024
       1  | 2048 (regular PF)
       1  | 3584 (admin only PF)

Note that due to FW/HW limitations we can't share all 4GiB GGTT
address space with VFs, so for the larger (>7) number of the VFs
the change in the outcome is happening at different points than
we have in case of GuC contexts/doorbells IDs.

Signed-off-by: Michal Wajdeczko <michal.wajdeczko@intel.com>
Reviewed-by: Piotr Piórkowski <piotr.piorkowski@intel.com>
Link: https://patch.msgid.link/20251112124408.8094-1-michal.wajdeczko@intel.com

...

This commit is contained in:

Michal Wajdeczko 2025-11-12 13:44:08 +01:00

parent aaecfadc22

commit e448372e8a

2 changed files with 69 additions and 0 deletions

Show all changes Ignore whitespace when comparing lines Ignore changes in amount of whitespace Ignore changes in whitespace at EOL

Download patch file Download diff file Expand all files Collapse all files

46

drivers/gpu/drm/xe/tests/xe_gt_sriov_pf_config_kunit.c

View file

23

drivers/gpu/drm/xe/xe_gt_sriov_pf_config.c

View file