Message ID | 20241127061117.25622-1-ville.syrjala@linux.intel.com (mailing list archive) |
---|---|
Headers | show |
Series | drm/i915/dpt: Try to make DPT shrinkable again | expand |
Hello Ville, Thank you very much for the series. With this, we don’t see older suspend/resume issue with DPT enabled. Tested-by: Vidya Srinivas <vidya.srinivas@intel.com> > -----Original Message----- > From: Ville Syrjala <ville.syrjala@linux.intel.com> > Sent: 27 November 2024 11:41 > To: intel-gfx@lists.freedesktop.org > Cc: Brian Geffon <bgeffon@google.com>; Srinivas, Vidya > <vidya.srinivas@intel.com> > Subject: [PATCH 0/4] drm/i915/dpt: Try to make DPT shrinkable again > > From: Ville Syrjälä <ville.syrjala@linux.intel.com> > > Try to make DPT objects shrinakble once again. To overcome the earlier > suspend/resume issues we'll just make sure all DPT VMAs are evicted during > suspend, and thus resume won't care whether the DPT objects got kicked out > or not. > > The proper solution would perhaps be to do the DPT CPU mapping on demand > in dpt insert_{page,range}() but that would require a bunch of extra thought > (especially for DPTs allocated from stolen). > > Cc: Brian Geffon <bgeffon@google.com> > Cc: Vidya Srinivas <vidya.srinivas@intel.com> > > Ville Syrjälä (4): > drm/i915: Don't reuse commit_work for the cleanup > drm/i915: Intruduce display.wq.cleanup > drm/i915/dpt: Evict all DPT VMAs on suspend > Revert "drm/i915/dpt: Make DPT object unshrinkable" > > drivers/gpu/drm/i915/display/intel_display.c | 6 +++--- > .../gpu/drm/i915/display/intel_display_core.h | 3 +++ > .../drm/i915/display/intel_display_driver.c | 6 ++++++ > .../drm/i915/display/intel_display_types.h | 2 ++ > drivers/gpu/drm/i915/display/intel_dpt.c | 4 ++-- > drivers/gpu/drm/i915/gem/i915_gem_object.h | 4 +--- > drivers/gpu/drm/i915/gt/intel_ggtt.c | 19 ++++++++++++++----- > drivers/gpu/drm/i915/gt/intel_gtt.h | 4 ++-- > 8 files changed, 33 insertions(+), 15 deletions(-) > > -- > 2.45.2
From: Ville Syrjälä <ville.syrjala@linux.intel.com> Try to make DPT objects shrinakble once again. To overcome the earlier suspend/resume issues we'll just make sure all DPT VMAs are evicted during suspend, and thus resume won't care whether the DPT objects got kicked out or not. The proper solution would perhaps be to do the DPT CPU mapping on demand in dpt insert_{page,range}() but that would require a bunch of extra thought (especially for DPTs allocated from stolen). Cc: Brian Geffon <bgeffon@google.com> Cc: Vidya Srinivas <vidya.srinivas@intel.com> Ville Syrjälä (4): drm/i915: Don't reuse commit_work for the cleanup drm/i915: Intruduce display.wq.cleanup drm/i915/dpt: Evict all DPT VMAs on suspend Revert "drm/i915/dpt: Make DPT object unshrinkable" drivers/gpu/drm/i915/display/intel_display.c | 6 +++--- .../gpu/drm/i915/display/intel_display_core.h | 3 +++ .../drm/i915/display/intel_display_driver.c | 6 ++++++ .../drm/i915/display/intel_display_types.h | 2 ++ drivers/gpu/drm/i915/display/intel_dpt.c | 4 ++-- drivers/gpu/drm/i915/gem/i915_gem_object.h | 4 +--- drivers/gpu/drm/i915/gt/intel_ggtt.c | 19 ++++++++++++++----- drivers/gpu/drm/i915/gt/intel_gtt.h | 4 ++-- 8 files changed, 33 insertions(+), 15 deletions(-)