In preparation for clean driver release, attempts to drain work queues and release freed objects are taken at driver remove time. However, GT buffer pools are now not flushed before the driver release phase. Since unused objects may stay there for up to one second, some may survive until driver release is attempted. That can potentially explain sporadic then hardly reproducible issues observed at driver release time, like non-zero shrink counter or outstanding address space areas.
Flush buffer pools on GT remove as a fix. On driver release, don't flush the pools again, just assert that the flush was called and nothing added more in between.
Signed-off-by: Janusz Krzysztofik janusz.krzysztofik@linux.intel.com Cc: Chris Wilson chris@chris-wilson.co.uk --- Resending with Cc: dri-devel@lists.freedesktop.org as requested, and a typo in commit description fixed.
Thanks, Janusz
drivers/gpu/drm/i915/gt/intel_gt.c | 2 ++ drivers/gpu/drm/i915/gt/intel_gt_buffer_pool.c | 2 -- 2 files changed, 2 insertions(+), 2 deletions(-)
diff --git a/drivers/gpu/drm/i915/gt/intel_gt.c b/drivers/gpu/drm/i915/gt/intel_gt.c index 62d40c986642..8f322a4ecd87 100644 --- a/drivers/gpu/drm/i915/gt/intel_gt.c +++ b/drivers/gpu/drm/i915/gt/intel_gt.c @@ -737,6 +737,8 @@ void intel_gt_driver_remove(struct intel_gt *gt) intel_uc_driver_remove(>->uc);
intel_engines_release(gt); + + intel_gt_flush_buffer_pool(gt); }
void intel_gt_driver_unregister(struct intel_gt *gt) diff --git a/drivers/gpu/drm/i915/gt/intel_gt_buffer_pool.c b/drivers/gpu/drm/i915/gt/intel_gt_buffer_pool.c index aa0a59c5b614..acc49c56a9f3 100644 --- a/drivers/gpu/drm/i915/gt/intel_gt_buffer_pool.c +++ b/drivers/gpu/drm/i915/gt/intel_gt_buffer_pool.c @@ -245,8 +245,6 @@ void intel_gt_fini_buffer_pool(struct intel_gt *gt) struct intel_gt_buffer_pool *pool = >->buffer_pool; int n;
- intel_gt_flush_buffer_pool(gt); - for (n = 0; n < ARRAY_SIZE(pool->cache_list); n++) GEM_BUG_ON(!list_empty(&pool->cache_list[n])); }