[PATCH v3] drm/amd/amdgpu implement tdr advanced mode
Andrey Grodzovsky
andrey.grodzovsky at amd.com
Mon Mar 8 22:03:06 UTC 2021
On 2021-03-08 7:33 a.m., Jack Zhang wrote:
> [Why]
> Previous tdr design treats the first job in job_timeout as the bad job.
> But sometimes a later bad compute job can block a good gfx job and
> cause an unexpected gfx job timeout because gfx and compute ring share
> internal GC HW mutually.
>
> [How]
> This patch implements an advanced tdr mode.
> 1.It synchronized resubmit wait step to find the real bad job. If the
> job's hw fence get timeout, we decrease the old job's karma, treat
> the new found one as a guilty one,do hw reset to recover hw. After
> that, we conitue the resubmit step to resubmit the left jobs.
>
> 2. For whole gpu reset(vram lost), do resubmit as the old style.
>
> Signed-off-by: Jack Zhang <Jack.Zhang1 at amd.com>
> ---
> drivers/gpu/drm/amd/amdgpu/amdgpu_device.c | 57 ++++++++++++++++++++--
> drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c | 2 +-
> drivers/gpu/drm/amd/amdgpu/amdgpu_job.c | 33 +++++++++++++
> 3 files changed, 88 insertions(+), 4 deletions(-)
>
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
> index e247c3a2ec08..fa53c6c00ee9 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
> @@ -4639,7 +4639,7 @@ int amdgpu_device_gpu_recover(struct amdgpu_device *adev,
> int i, r = 0;
> bool need_emergency_restart = false;
> bool audio_suspended = false;
> -
> + int tmp_vram_lost_counter;
> /*
> * Special case: RAS triggered and full reset isn't supported
> */
> @@ -4788,6 +4788,7 @@ int amdgpu_device_gpu_recover(struct amdgpu_device *adev,
> }
> }
>
> + tmp_vram_lost_counter = atomic_read(&((adev)->vram_lost_counter));
> /* Actual ASIC resets if needed.*/
> /* TODO Implement XGMI hive reset logic for SRIOV */
> if (amdgpu_sriov_vf(adev)) {
> @@ -4807,17 +4808,67 @@ int amdgpu_device_gpu_recover(struct amdgpu_device *adev,
>
> for (i = 0; i < AMDGPU_MAX_RINGS; ++i) {
> struct amdgpu_ring *ring = tmp_adev->rings[i];
> + int ret = 0;
> + struct drm_sched_job *s_job = NULL;
>
> if (!ring || !ring->sched.thread)
> continue;
>
> /* No point to resubmit jobs if we didn't HW reset*/
> - if (!tmp_adev->asic_reset_res && !job_signaled)
> + if (!tmp_adev->asic_reset_res && !job_signaled) {
> drm_sched_resubmit_jobs(&ring->sched);
>
> - drm_sched_start(&ring->sched, !tmp_adev->asic_reset_res);
> + if (amdgpu_gpu_recovery == 2 &&
> + !list_empty(&ring->sched->ring_mirror_list)
> + && !(tmp_vram_lost_counter < atomic_read(&adev->vram_lost_counter)
> + ) {
> +
> + s_job = list_first_entry_or_null(&ring->sched->ring_mirror_list, struct drm_sched_job, node);
Seems better to check for NULL here and skip checking for list_empty
above
> + ret = dma_fence_wait_timeout(s_job->s_fence->parent, false, ring->sched->timeout);
> + if (ret == 0) { /* timeout */
> + /*reset karma to the right job*/
> + if (job && job != s_job)
> + amdgpu_sched_decrease_karma(&job->base);
> + drm_sched_increase_karma(s_job);
> +
> + /* do hw reset */
> + if (amdgpu_sriov_vf(adev)) {
> + amdgpu_virt_fini_data_exchange(adev);
> + r = amdgpu_device_reset_sriov(adev, false);
> + if (r)
> + adev->asic_reset_res = r;
> + } else {
> + r = amdgpu_do_asic_reset(hive, device_list_handle, &need_full_reset, false);
> + if (r && r == -EAGAIN)
> + goto retry;
> +
> + /* add reset counter so that the following resubmitted job could flush vmid */
> + atomic_inc(&tmp_adev->gpu_reset_counter);
> +
> + //resubmit again the left jobs
> + drm_sched_resubmit_jobs(&ring->sched);
> + }
> + }
> + }
> + }
> + if (amdgpu_gpu_recovery != 2)
> + drm_sched_start(&ring->sched, !tmp_adev->asic_reset_res);
> + }
> +
> + if (amdgpu_gpu_recovery == 2) {
> + /* delay sched start until all jobs are submitted for advanced tdr mode */
> + for (i = 0; i < AMDGPU_MAX_RINGS; ++i) {
> + struct amdgpu_ring *ring = tmp_adev->rings[i];
> + int ret = 0;
> +
> + if (!ring || !ring->sched.thread)
> + continue;
> +
> + drm_sched_start(&ring->sched, !tmp_adev->asic_reset_res);
> + }
> }
>
> +
> if (!amdgpu_device_has_dc_support(tmp_adev) && !job_signaled) {
> drm_helper_resume_force_mode(adev_to_drm(tmp_adev));
> }
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c
> index 865f924772b0..9c3f4edb7532 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c
> @@ -509,7 +509,7 @@ module_param_named(compute_multipipe, amdgpu_compute_multipipe, int, 0444);
> * DOC: gpu_recovery (int)
> * Set to enable GPU recovery mechanism (1 = enable, 0 = disable). The default is -1 (auto, disabled except SRIOV).
> */
> -MODULE_PARM_DESC(gpu_recovery, "Enable GPU recovery mechanism, (1 = enable, 0 = disable, -1 = auto)");
> +MODULE_PARM_DESC(gpu_recovery, "Enable GPU recovery mechanism, (2 = advanced tdr mode, 1 = enable, 0 = disable, -1 = auto)");
> module_param_named(gpu_recovery, amdgpu_gpu_recovery, int, 0444);
>
> /**
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_job.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_job.c
> index 759b34799221..3457792b5d20 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_job.c
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_job.c
> @@ -281,6 +281,39 @@ void amdgpu_job_stop_all_jobs_on_sched(struct drm_gpu_scheduler *sched)
> }
> }
>
> +void amdgpu_sched_decrease_karma(struct drm_sched_job *bad)
> +{
> + int i;
> + struct drm_sched_entity *tmp;
> + struct drm_sched_entity *entity;
> + struct drm_gpu_scheduler *sched = bad->sched;
> +
> + /* don't decrease @bad's karma if it's from KERNEL RQ,
> + * because sometimes GPU hang would cause kernel jobs (like VM updating jobs)
> + * corrupt but keep in mind that kernel jobs always considered good.
> + */
> + if (bad->s_priority != DRM_SCHED_PRIORITY_KERNEL) {
> + atomic_dec(&bad->karma);
> + for (i = DRM_SCHED_PRIORITY_MIN; i < DRM_SCHED_PRIORITY_KERNEL;
> + i++) {
> + struct drm_sched_rq *rq = &sched->sched_rq[i];
> +
> + spin_lock(&rq->lock);
> + list_for_each_entry_safe(entity, tmp, &rq->entities, list) {
> + if (bad->s_fence->scheduled.context ==
> + entity->fence_context) {
> + if (entity->guilty)
> + atomic_set(entity->guilty, 0);
> + break;
> + }
> + }
> + spin_unlock(&rq->lock);
> + if (&entity->list != &rq->entities)
> + break;
> + }
> + }
> +}
> +
This function is practically identical to amdgpu_sched_increase_karma,
probably better to make one function with some flag for the increase/
decrease cases.
Andrey
> const struct drm_sched_backend_ops amdgpu_sched_ops = {
> .dependency = amdgpu_job_dependency,
> .run_job = amdgpu_job_run,
>
More information about the amd-gfx
mailing list