drm/scheduler: remove full_recover from drm_sched_start

This was basically just another one of amdgpus hacks. The parameter
allowed to restart the scheduler without turning fence signaling on
again.

That this is absolutely not a good idea should be obvious by now since
the fences will then just sit there and never signal.

While at it cleanup the code a bit.
Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240722083816.99685-1-christian.koenig@amd.com

drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_arcturus.c

@@ -300,7 +300,7 @@ static int suspend_resume_compute_scheduler(struct amdgpu_device *adev, bool sus
 			if (r)
 				goto out;
 		} else {
-			drm_sched_start(&ring->sched, false);
+			drm_sched_start(&ring->sched);
 		}
 	}
 

drivers/gpu/drm/amd/amdgpu/amdgpu_device.c

@@ -5862,7 +5862,7 @@ int amdgpu_device_gpu_recover(struct amdgpu_device *adev,
 			if (!amdgpu_ring_sched_ready(ring))
 				continue;
 
-			drm_sched_start(&ring->sched, true);
+			drm_sched_start(&ring->sched);
 		}
 
 		if (!drm_drv_uses_atomic_modeset(adev_to_drm(tmp_adev)) && !job_signaled)
@@ -6360,7 +6360,7 @@ void amdgpu_pci_resume(struct pci_dev *pdev)
 		if (!amdgpu_ring_sched_ready(ring))
 			continue;
 
-		drm_sched_start(&ring->sched, true);
+		drm_sched_start(&ring->sched);
 	}
 
 	amdgpu_device_unset_mp1_state(adev);

drivers/gpu/drm/etnaviv/etnaviv_sched.c

@@ -72,12 +72,12 @@ static enum drm_gpu_sched_stat etnaviv_sched_timedout_job(struct drm_sched_job
 
 	drm_sched_resubmit_jobs(&gpu->sched);
 
-	drm_sched_start(&gpu->sched, true);
+	drm_sched_start(&gpu->sched);
 	return DRM_GPU_SCHED_STAT_NOMINAL;
 
 out_no_timeout:
 	/* restart scheduler after GPU is usable again */
-	drm_sched_start(&gpu->sched, true);
+	drm_sched_start(&gpu->sched);
 	return DRM_GPU_SCHED_STAT_NOMINAL;
 }
 

drivers/gpu/drm/imagination/pvr_queue.c

@@ -782,7 +782,7 @@ static void pvr_queue_start(struct pvr_queue *queue)
 		}
 	}
 
-	drm_sched_start(&queue->scheduler, true);
+	drm_sched_start(&queue->scheduler);
 }
 
 /**
@@ -842,7 +842,7 @@ pvr_queue_timedout_job(struct drm_sched_job *s_job)
 	}
 	mutex_unlock(&pvr_dev->queues.lock);
 
-	drm_sched_start(sched, true);
+	drm_sched_start(sched);
 
 	return DRM_GPU_SCHED_STAT_NOMINAL;
 }

drivers/gpu/drm/lima/lima_sched.c

@@ -463,7 +463,7 @@ static enum drm_gpu_sched_stat lima_sched_timedout_job(struct drm_sched_job *job
 	lima_pm_idle(ldev);
 
 	drm_sched_resubmit_jobs(&pipe->base);
-	drm_sched_start(&pipe->base, true);
+	drm_sched_start(&pipe->base);
 
 	return DRM_GPU_SCHED_STAT_NOMINAL;
 }

drivers/gpu/drm/nouveau/nouveau_sched.c

@@ -379,7 +379,7 @@ nouveau_sched_timedout_job(struct drm_sched_job *sched_job)
 	else
 		NV_PRINTK(warn, job->cli, "Generic job timeout.\n");
 
-	drm_sched_start(sched, true);
+	drm_sched_start(sched);
 
 	return stat;
 }

drivers/gpu/drm/panfrost/panfrost_job.c

@@ -727,7 +727,7 @@ panfrost_reset(struct panfrost_device *pfdev,
 
 	/* Restart the schedulers */
 	for (i = 0; i < NUM_JOB_SLOTS; i++)
-		drm_sched_start(&pfdev->js->queue[i].sched, true);
+		drm_sched_start(&pfdev->js->queue[i].sched);
 
 	/* Re-enable job interrupts now that everything has been restarted. */
 	job_write(pfdev, JOB_INT_MASK,

drivers/gpu/drm/panthor/panthor_mmu.c

@@ -827,7 +827,7 @@ static void panthor_vm_stop(struct panthor_vm *vm)
 
 static void panthor_vm_start(struct panthor_vm *vm)
 {
-	drm_sched_start(&vm->sched, true);
+	drm_sched_start(&vm->sched);
 }
 
 /**

drivers/gpu/drm/panthor/panthor_sched.c

@@ -2525,7 +2525,7 @@ static void queue_start(struct panthor_queue *queue)
 	list_for_each_entry(job, &queue->scheduler.pending_list, base.list)
 		job->base.s_fence->parent = dma_fence_get(job->done_fence);
 
-	drm_sched_start(&queue->scheduler, true);
+	drm_sched_start(&queue->scheduler);
 }
 
 static void panthor_group_stop(struct panthor_group *group)

drivers/gpu/drm/scheduler/sched_main.c

@@ -674,13 +674,11 @@ EXPORT_SYMBOL(drm_sched_stop);
  * drm_sched_start - recover jobs after a reset
  *
  * @sched: scheduler instance
- * @full_recovery: proceed with complete sched restart
  *
  */
-void drm_sched_start(struct drm_gpu_scheduler *sched, bool full_recovery)
+void drm_sched_start(struct drm_gpu_scheduler *sched)
 {
 	struct drm_sched_job *s_job, *tmp;
-	int r;
 
 	/*
 	 * Locking the list is not required here as the sched thread is parked
@@ -692,24 +690,17 @@ void drm_sched_start(struct drm_gpu_scheduler *sched, bool full_recovery)
 
 		atomic_add(s_job->credits, &sched->credit_count);
 
-		if (!full_recovery)
+		if (!fence) {
+			drm_sched_job_done(s_job, -ECANCELED);
 			continue;
+		}
 
-		if (fence) {
-			r = dma_fence_add_callback(fence, &s_job->cb,
-						   drm_sched_job_done_cb);
-			if (r == -ENOENT)
+		if (dma_fence_add_callback(fence, &s_job->cb,
+					   drm_sched_job_done_cb))
 			drm_sched_job_done(s_job, fence->error);
-			else if (r)
-				DRM_DEV_ERROR(sched->dev, "fence add callback failed (%d)\n",
-					  r);
-		} else
-			drm_sched_job_done(s_job, -ECANCELED);
 	}
 
-	if (full_recovery)
 	drm_sched_start_timeout_unlocked(sched);
-
 	drm_sched_wqueue_start(sched);
 }
 EXPORT_SYMBOL(drm_sched_start);

drivers/gpu/drm/v3d/v3d_sched.c

@@ -653,7 +653,7 @@ v3d_gpu_reset_for_timeout(struct v3d_dev *v3d, struct drm_sched_job *sched_job)
 
 	/* Unblock schedulers and restart their jobs. */
 	for (q = 0; q < V3D_MAX_QUEUES; q++) {
-		drm_sched_start(&v3d->queue[q].sched, true);
+		drm_sched_start(&v3d->queue[q].sched);
 	}
 
 	mutex_unlock(&v3d->reset_lock);

include/drm/gpu_scheduler.h

@@ -579,7 +579,7 @@ bool drm_sched_wqueue_ready(struct drm_gpu_scheduler *sched);
 void drm_sched_wqueue_stop(struct drm_gpu_scheduler *sched);
 void drm_sched_wqueue_start(struct drm_gpu_scheduler *sched);
 void drm_sched_stop(struct drm_gpu_scheduler *sched, struct drm_sched_job *bad);
-void drm_sched_start(struct drm_gpu_scheduler *sched, bool full_recovery);
+void drm_sched_start(struct drm_gpu_scheduler *sched);
 void drm_sched_resubmit_jobs(struct drm_gpu_scheduler *sched);
 void drm_sched_increase_karma(struct drm_sched_job *bad);
 void drm_sched_reset_karma(struct drm_sched_job *bad);