| Message ID | 20190819164448.22132-3-chris@chris-wilson.co.uk (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | [01/10] drm/i915: Use 0 for the unordered context | expand |
On Mon, 19 Aug 2019 at 17:45, Chris Wilson <chris@chris-wilson.co.uk> wrote: > > Inside gtt_restore_mappings() we currently take the obj->resv->lock, but > in the future we need to avoid taking this fs-reclaim tainted lock as we > need to extend the coverage of the vm->mutex. Take advantage of the > single-threaded nature of the early resume phase, and do a single > wbinvd() to flush all the GTT objects en masse. > > Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk> Reviewed-by: Matthew Auld <matthew.auld@intel.com>
diff --git a/drivers/gpu/drm/i915/i915_gem_gtt.c b/drivers/gpu/drm/i915/i915_gem_gtt.c index 5413c2ff51a2..e48df11a19fb 100644 --- a/drivers/gpu/drm/i915/i915_gem_gtt.c +++ b/drivers/gpu/drm/i915/i915_gem_gtt.c @@ -3274,6 +3274,7 @@ void i915_ggtt_disable_guc(struct i915_ggtt *ggtt) static void ggtt_restore_mappings(struct i915_ggtt *ggtt) { struct i915_vma *vma, *vn; + bool flush = false; intel_gt_check_and_clear_faults(ggtt->vm.gt); @@ -3298,10 +3299,9 @@ static void ggtt_restore_mappings(struct i915_ggtt *ggtt) WARN_ON(i915_vma_bind(vma, obj ? obj->cache_level : 0, PIN_UPDATE)); - if (obj) { - i915_gem_object_lock(obj); - WARN_ON(i915_gem_object_set_to_gtt_domain(obj, false)); - i915_gem_object_unlock(obj); + if (obj) { /* only used during resume => exclusive access */ + flush |= fetch_and_zero(&obj->write_domain); + obj->read_domains |= I915_GEM_DOMAIN_GTT; } lock: @@ -3312,6 +3312,9 @@ static void ggtt_restore_mappings(struct i915_ggtt *ggtt) ggtt->invalidate(ggtt); mutex_unlock(&ggtt->vm.mutex); + + if (flush) + wbinvd_on_all_cpus(); } void i915_gem_restore_gtt_mappings(struct drm_i915_private *i915)
Inside gtt_restore_mappings() we currently take the obj->resv->lock, but in the future we need to avoid taking this fs-reclaim tainted lock as we need to extend the coverage of the vm->mutex. Take advantage of the single-threaded nature of the early resume phase, and do a single wbinvd() to flush all the GTT objects en masse. Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk> --- drivers/gpu/drm/i915/i915_gem_gtt.c | 11 +++++++---- 1 file changed, 7 insertions(+), 4 deletions(-)