diff options
| author | Dave Airlie <airlied@redhat.com> | 2021-04-08 14:02:14 +1000 |
|---|---|---|
| committer | Dave Airlie <airlied@redhat.com> | 2021-04-08 14:02:21 +1000 |
| commit | 9c0fed84d5750e1eea6c664e073ffa2534a17743 (patch) | |
| tree | 19fd185bfceddc79ca7c9eaff670a24e35da7216 /drivers/gpu/drm/i915/gt/intel_ggtt_fencing.c | |
| parent | 41d1d0c51f5ffd5c2c35e82e4a675b185cccea13 (diff) | |
| parent | 81f1f8f1e1489c0bf051d5241ec10da07869b911 (diff) | |
Merge tag 'drm-intel-next-2021-04-01' of git://anongit.freedesktop.org/drm/drm-intel into drm-next
Features:
- Add support for FBs requiring a power-of-two stride padding (Imre)
Refactoring:
- Disassociate display version from gen (Matt)
- Refactor legacy DP and HDMI code to separate files (Ville)
- Refactor FB plane code to a separate file (Imre)
- Refactor VBT child device info parsing and usage (Jani)
- Refactor KBL/TGL/ADL-S display and gt stepping schemes (Jani)
Fixes:
- DP Link-Training Tunable PHY Repeaters (LTTPR) fixes (Imre)
- HDCP fixes (Anshuman)
- DP 2.0 HDMI 2.1 PCON Fixed Rate Link (FRL) fixes (Ankit)
- Set HDA link parameters in driver (Kai)
- Fix enabled_planes bitmask (Ville)
- Fix transposed arguments to skl_plane_wm_level() (Ville)
- Stop adding planes to the commit needlessly (Ville)
Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Jani Nikula <jani.nikula@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/87v996ml17.fsf@intel.com
Diffstat (limited to 'drivers/gpu/drm/i915/gt/intel_ggtt_fencing.c')
| -rw-r--r-- | drivers/gpu/drm/i915/gt/intel_ggtt_fencing.c | 13 |
1 files changed, 12 insertions, 1 deletions
diff --git a/drivers/gpu/drm/i915/gt/intel_ggtt_fencing.c b/drivers/gpu/drm/i915/gt/intel_ggtt_fencing.c index e891552611d5..e72b7a0dc316 100644 --- a/drivers/gpu/drm/i915/gt/intel_ggtt_fencing.c +++ b/drivers/gpu/drm/i915/gt/intel_ggtt_fencing.c @@ -298,7 +298,18 @@ void i915_vma_revoke_fence(struct i915_vma *vma) WRITE_ONCE(fence->vma, NULL); vma->fence = NULL; - with_intel_runtime_pm_if_in_use(fence_to_uncore(fence)->rpm, wakeref) + /* + * Skip the write to HW if and only if the device is currently + * suspended. + * + * If the driver does not currently hold a wakeref (if_in_use == 0), + * the device may currently be runtime suspended, or it may be woken + * up before the suspend takes place. If the device is not suspended + * (powered down) and we skip clearing the fence register, the HW is + * left in an undefined state where we may end up with multiple + * registers overlapping. + */ + with_intel_runtime_pm_if_active(fence_to_uncore(fence)->rpm, wakeref) fence_write(fence); } |