[PATCH v3 07/12] drm/sched: Prevent any job recoveries after device is unplugged.

Christian König ckoenig.leichtzumerken at gmail.com
Mon Nov 23 08:06:05 UTC 2020


Am 23.11.20 um 06:37 schrieb Andrey Grodzovsky:
>
> On 11/22/20 6:57 AM, Christian König wrote:
>> Am 21.11.20 um 06:21 schrieb Andrey Grodzovsky:
>>> No point to try recovery if device is gone, it's meaningless.
>>
>> I think that this should go into the device specific recovery 
>> function and not in the scheduler.
>
>
> The timeout timer is rearmed here, so this prevents any new recovery 
> work to restart from here
> after drm_dev_unplug was executed from amdgpu_pci_remove.It will not 
> cover other places like
> job cleanup or starting new job but those should stop once the 
> scheduler thread is stopped later.

Yeah, but this is rather unclean. We should probably return an error 
code instead if the timer should be rearmed or not.

Christian.

>
> Andrey
>
>
>>
>> Christian.
>>
>>>
>>> Signed-off-by: Andrey Grodzovsky <andrey.grodzovsky at amd.com>
>>> ---
>>>   drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c |  2 +-
>>>   drivers/gpu/drm/etnaviv/etnaviv_sched.c   |  3 ++-
>>>   drivers/gpu/drm/lima/lima_sched.c         |  3 ++-
>>>   drivers/gpu/drm/panfrost/panfrost_job.c   |  2 +-
>>>   drivers/gpu/drm/scheduler/sched_main.c    | 15 ++++++++++++++-
>>>   drivers/gpu/drm/v3d/v3d_sched.c           | 15 ++++++++++-----
>>>   include/drm/gpu_scheduler.h               |  6 +++++-
>>>   7 files changed, 35 insertions(+), 11 deletions(-)
>>>
>>> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c 
>>> b/drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c
>>> index d56f402..d0b0021 100644
>>> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c
>>> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c
>>> @@ -487,7 +487,7 @@ int amdgpu_fence_driver_init_ring(struct 
>>> amdgpu_ring *ring,
>>>             r = drm_sched_init(&ring->sched, &amdgpu_sched_ops,
>>>                      num_hw_submission, amdgpu_job_hang_limit,
>>> -                   timeout, ring->name);
>>> +                   timeout, ring->name, &adev->ddev);
>>>           if (r) {
>>>               DRM_ERROR("Failed to create scheduler on ring %s.\n",
>>>                     ring->name);
>>> diff --git a/drivers/gpu/drm/etnaviv/etnaviv_sched.c 
>>> b/drivers/gpu/drm/etnaviv/etnaviv_sched.c
>>> index cd46c88..7678287 100644
>>> --- a/drivers/gpu/drm/etnaviv/etnaviv_sched.c
>>> +++ b/drivers/gpu/drm/etnaviv/etnaviv_sched.c
>>> @@ -185,7 +185,8 @@ int etnaviv_sched_init(struct etnaviv_gpu *gpu)
>>>         ret = drm_sched_init(&gpu->sched, &etnaviv_sched_ops,
>>>                    etnaviv_hw_jobs_limit, etnaviv_job_hang_limit,
>>> -                 msecs_to_jiffies(500), dev_name(gpu->dev));
>>> +                 msecs_to_jiffies(500), dev_name(gpu->dev),
>>> +                 gpu->drm);
>>>       if (ret)
>>>           return ret;
>>>   diff --git a/drivers/gpu/drm/lima/lima_sched.c 
>>> b/drivers/gpu/drm/lima/lima_sched.c
>>> index dc6df9e..8a7e5d7ca 100644
>>> --- a/drivers/gpu/drm/lima/lima_sched.c
>>> +++ b/drivers/gpu/drm/lima/lima_sched.c
>>> @@ -505,7 +505,8 @@ int lima_sched_pipe_init(struct lima_sched_pipe 
>>> *pipe, const char *name)
>>>         return drm_sched_init(&pipe->base, &lima_sched_ops, 1,
>>>                     lima_job_hang_limit, msecs_to_jiffies(timeout),
>>> -                  name);
>>> +                  name,
>>> +                  pipe->ldev->ddev);
>>>   }
>>>     void lima_sched_pipe_fini(struct lima_sched_pipe *pipe)
>>> diff --git a/drivers/gpu/drm/panfrost/panfrost_job.c 
>>> b/drivers/gpu/drm/panfrost/panfrost_job.c
>>> index 30e7b71..37b03b01 100644
>>> --- a/drivers/gpu/drm/panfrost/panfrost_job.c
>>> +++ b/drivers/gpu/drm/panfrost/panfrost_job.c
>>> @@ -520,7 +520,7 @@ int panfrost_job_init(struct panfrost_device 
>>> *pfdev)
>>>           ret = drm_sched_init(&js->queue[j].sched,
>>>                        &panfrost_sched_ops,
>>>                        1, 0, msecs_to_jiffies(500),
>>> -                     "pan_js");
>>> +                     "pan_js", pfdev->ddev);
>>>           if (ret) {
>>>               dev_err(pfdev->dev, "Failed to create scheduler: %d.", 
>>> ret);
>>>               goto err_sched;
>>> diff --git a/drivers/gpu/drm/scheduler/sched_main.c 
>>> b/drivers/gpu/drm/scheduler/sched_main.c
>>> index c3f0bd0..95db8c6 100644
>>> --- a/drivers/gpu/drm/scheduler/sched_main.c
>>> +++ b/drivers/gpu/drm/scheduler/sched_main.c
>>> @@ -53,6 +53,7 @@
>>>   #include <drm/drm_print.h>
>>>   #include <drm/gpu_scheduler.h>
>>>   #include <drm/spsc_queue.h>
>>> +#include <drm/drm_drv.h>
>>>     #define CREATE_TRACE_POINTS
>>>   #include "gpu_scheduler_trace.h"
>>> @@ -283,8 +284,16 @@ static void drm_sched_job_timedout(struct 
>>> work_struct *work)
>>>       struct drm_gpu_scheduler *sched;
>>>       struct drm_sched_job *job;
>>>   +    int idx;
>>> +
>>>       sched = container_of(work, struct drm_gpu_scheduler, 
>>> work_tdr.work);
>>>   +    if (!drm_dev_enter(sched->ddev, &idx)) {
>>> +        DRM_INFO("%s - device unplugged skipping recovery on 
>>> scheduler:%s",
>>> +             __func__, sched->name);
>>> +        return;
>>> +    }
>>> +
>>>       /* Protects against concurrent deletion in 
>>> drm_sched_get_cleanup_job */
>>>       spin_lock(&sched->job_list_lock);
>>>       job = list_first_entry_or_null(&sched->ring_mirror_list,
>>> @@ -316,6 +325,8 @@ static void drm_sched_job_timedout(struct 
>>> work_struct *work)
>>>       spin_lock(&sched->job_list_lock);
>>>       drm_sched_start_timeout(sched);
>>>       spin_unlock(&sched->job_list_lock);
>>> +
>>> +    drm_dev_exit(idx);
>>>   }
>>>      /**
>>> @@ -845,7 +856,8 @@ int drm_sched_init(struct drm_gpu_scheduler *sched,
>>>              unsigned hw_submission,
>>>              unsigned hang_limit,
>>>              long timeout,
>>> -           const char *name)
>>> +           const char *name,
>>> +           struct drm_device *ddev)
>>>   {
>>>       int i, ret;
>>>       sched->ops = ops;
>>> @@ -853,6 +865,7 @@ int drm_sched_init(struct drm_gpu_scheduler *sched,
>>>       sched->name = name;
>>>       sched->timeout = timeout;
>>>       sched->hang_limit = hang_limit;
>>> +    sched->ddev = ddev;
>>>       for (i = DRM_SCHED_PRIORITY_MIN; i < DRM_SCHED_PRIORITY_COUNT; 
>>> i++)
>>>           drm_sched_rq_init(sched, &sched->sched_rq[i]);
>>>   diff --git a/drivers/gpu/drm/v3d/v3d_sched.c 
>>> b/drivers/gpu/drm/v3d/v3d_sched.c
>>> index 0747614..f5076e5 100644
>>> --- a/drivers/gpu/drm/v3d/v3d_sched.c
>>> +++ b/drivers/gpu/drm/v3d/v3d_sched.c
>>> @@ -401,7 +401,8 @@ v3d_sched_init(struct v3d_dev *v3d)
>>>                    &v3d_bin_sched_ops,
>>>                    hw_jobs_limit, job_hang_limit,
>>>                    msecs_to_jiffies(hang_limit_ms),
>>> -                 "v3d_bin");
>>> +                 "v3d_bin",
>>> +                 &v3d->drm);
>>>       if (ret) {
>>>           dev_err(v3d->drm.dev, "Failed to create bin scheduler: 
>>> %d.", ret);
>>>           return ret;
>>> @@ -411,7 +412,8 @@ v3d_sched_init(struct v3d_dev *v3d)
>>>                    &v3d_render_sched_ops,
>>>                    hw_jobs_limit, job_hang_limit,
>>>                    msecs_to_jiffies(hang_limit_ms),
>>> -                 "v3d_render");
>>> +                 "v3d_render",
>>> +                 &v3d->drm);
>>>       if (ret) {
>>>           dev_err(v3d->drm.dev, "Failed to create render scheduler: 
>>> %d.",
>>>               ret);
>>> @@ -423,7 +425,8 @@ v3d_sched_init(struct v3d_dev *v3d)
>>>                    &v3d_tfu_sched_ops,
>>>                    hw_jobs_limit, job_hang_limit,
>>>                    msecs_to_jiffies(hang_limit_ms),
>>> -                 "v3d_tfu");
>>> +                 "v3d_tfu",
>>> +                 &v3d->drm);
>>>       if (ret) {
>>>           dev_err(v3d->drm.dev, "Failed to create TFU scheduler: %d.",
>>>               ret);
>>> @@ -436,7 +439,8 @@ v3d_sched_init(struct v3d_dev *v3d)
>>>                        &v3d_csd_sched_ops,
>>>                        hw_jobs_limit, job_hang_limit,
>>>                        msecs_to_jiffies(hang_limit_ms),
>>> -                     "v3d_csd");
>>> +                     "v3d_csd",
>>> +                     &v3d->drm);
>>>           if (ret) {
>>>               dev_err(v3d->drm.dev, "Failed to create CSD scheduler: 
>>> %d.",
>>>                   ret);
>>> @@ -448,7 +452,8 @@ v3d_sched_init(struct v3d_dev *v3d)
>>>                        &v3d_cache_clean_sched_ops,
>>>                        hw_jobs_limit, job_hang_limit,
>>>                        msecs_to_jiffies(hang_limit_ms),
>>> -                     "v3d_cache_clean");
>>> +                     "v3d_cache_clean",
>>> +                     &v3d->drm);
>>>           if (ret) {
>>>               dev_err(v3d->drm.dev, "Failed to create CACHE_CLEAN 
>>> scheduler: %d.",
>>>                   ret);
>>> diff --git a/include/drm/gpu_scheduler.h b/include/drm/gpu_scheduler.h
>>> index 9243655..a980709 100644
>>> --- a/include/drm/gpu_scheduler.h
>>> +++ b/include/drm/gpu_scheduler.h
>>> @@ -32,6 +32,7 @@
>>>     struct drm_gpu_scheduler;
>>>   struct drm_sched_rq;
>>> +struct drm_device;
>>>     /* These are often used as an (initial) index
>>>    * to an array, and as such should start at 0.
>>> @@ -267,6 +268,7 @@ struct drm_sched_backend_ops {
>>>    * @score: score to help loadbalancer pick a idle sched
>>>    * @ready: marks if the underlying HW is ready to work
>>>    * @free_guilty: A hit to time out handler to free the guilty job.
>>> + * @ddev: Pointer to drm device of this scheduler.
>>>    *
>>>    * One scheduler is implemented for each hardware ring.
>>>    */
>>> @@ -288,12 +290,14 @@ struct drm_gpu_scheduler {
>>>       atomic_t                        score;
>>>       bool                ready;
>>>       bool                free_guilty;
>>> +    struct drm_device        *ddev;
>>>   };
>>>     int drm_sched_init(struct drm_gpu_scheduler *sched,
>>>              const struct drm_sched_backend_ops *ops,
>>>              uint32_t hw_submission, unsigned hang_limit, long timeout,
>>> -           const char *name);
>>> +           const char *name,
>>> +           struct drm_device *ddev);
>>>     void drm_sched_fini(struct drm_gpu_scheduler *sched);
>>>   int drm_sched_job_init(struct drm_sched_job *job,
>>
> _______________________________________________
> amd-gfx mailing list
> amd-gfx at lists.freedesktop.org
> https://lists.freedesktop.org/mailman/listinfo/amd-gfx



More information about the amd-gfx mailing list