Yiqing raised a problem of negative fence refcount for resubmitted jobs in amdgpu and suggested a workaround in [1]. I took a look myself and discovered some deeper problems both in amdgpu and scheduler code. Yiqing helped with testing the new code and also drew a detailed refcount and flow tracing diagram for parent (HW) fence life cycle and refcount under various cases for the proposed patchset at [2]. v2: Update race preventionby fixing by swithing from amdgpu_irq_get/put to enable/desable_irq (Christian) Drop refcount fix for amdgpu_job->external_hw_fence as it was causing underflow in direct submissions TODO - Follow up cleanup to totally get rid of amdgpu_job->external_hw_fence [1] - https://lore.kernel.org/all/731b7ff1-3cc9-e314-df2a-7c51b76d4db0@xxxxxxx/t/#r00c728fcc069b1276642c325bfa9d82bf8fa21a3 [2] - https://drive.google.com/file/d/1yEoeW6OQC9WnwmzFW6NBLhFP_jD0xcHm/view?usp=sharing Andrey Grodzovsky (4): drm/amdgpu: Add put fence in amdgpu_fence_driver_clear_job_fences drm/amdgpu: Prevent race between late signaled fences and GPU reset. drm/sched: Partial revert of 'drm/sched: Keep s_fence->parent pointer' drm/amdgpu: Follow up change to previous drm scheduler change. drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c | 2 ++ drivers/gpu/drm/amd/amdgpu/amdgpu_device.c | 31 +++++++++++++++++----- drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c | 29 ++++++++++++++++++-- drivers/gpu/drm/amd/amdgpu/amdgpu_job.c | 4 --- drivers/gpu/drm/amd/amdgpu/amdgpu_ring.h | 1 + drivers/gpu/drm/scheduler/sched_main.c | 13 ++++++--- 6 files changed, 65 insertions(+), 15 deletions(-) -- 2.25.1
