drm/amdgpu: add gang submit backend v2

Allows submitting jobs as gang which needs to run on multiple
engines at the same time.

Basic idea is that we have a global gang submit fence representing when the
gang leader is finally pushed to run on the hardware last.

Jobs submitted as gang are never re-submitted in case of a GPU reset since this
won't work and will just deadlock the hardware immediately again.

v2: fix logic inversion, improve documentation, fix rcu

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drivers/gpu/drm/amd/amdgpu/amdgpu.h

@@ -885,6 +885,7 @@ struct amdgpu_device {
 	u64				fence_context;
 	unsigned			num_rings;
 	struct amdgpu_ring		*rings[AMDGPU_MAX_RINGS];
+	struct dma_fence __rcu		*gang_submit;
 	bool				ib_pool_ready;
 	struct amdgpu_sa_manager	ib_pools[AMDGPU_IB_POOL_MAX];
 	struct amdgpu_sched		gpu_sched[AMDGPU_HW_IP_NUM][AMDGPU_RING_PRIO_MAX];
@@ -1294,6 +1295,8 @@ u32 amdgpu_device_pcie_port_rreg(struct amdgpu_device *adev,
 				u32 reg);
 void amdgpu_device_pcie_port_wreg(struct amdgpu_device *adev,
 				u32 reg, u32 v);
+struct dma_fence *amdgpu_device_switch_gang(struct amdgpu_device *adev,
+					    struct dma_fence *gang);
 
 /* atpx handler */
 #if defined(CONFIG_VGA_SWITCHEROO)

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

@@ -3511,6 +3511,7 @@ int amdgpu_device_init(struct amdgpu_device *adev,
 	adev->gmc.gart_size = 512 * 1024 * 1024;
 	adev->accel_working = false;
 	adev->num_rings = 0;
+	RCU_INIT_POINTER(adev->gang_submit, dma_fence_get_stub());
 	adev->mman.buffer_funcs = NULL;
 	adev->mman.buffer_funcs_ring = NULL;
 	adev->vm_manager.vm_pte_funcs = NULL;
@@ -3992,6 +3993,7 @@ void amdgpu_device_fini_sw(struct amdgpu_device *adev)
 	release_firmware(adev->firmware.gpu_info_fw);
 	adev->firmware.gpu_info_fw = NULL;
 	adev->accel_working = false;
+	dma_fence_put(rcu_dereference_protected(adev->gang_submit, true));
 
 	amdgpu_reset_fini(adev);
 
@@ -5960,3 +5962,36 @@ void amdgpu_device_pcie_port_wreg(struct amdgpu_device *adev,
 	(void)RREG32(data);
 	spin_unlock_irqrestore(&adev->pcie_idx_lock, flags);
 }
+
+/**
+ * amdgpu_device_switch_gang - switch to a new gang
+ * @adev: amdgpu_device pointer
+ * @gang: the gang to switch to
+ *
+ * Try to switch to a new gang.
+ * Returns: NULL if we switched to the new gang or a reference to the current
+ * gang leader.
+ */
+struct dma_fence *amdgpu_device_switch_gang(struct amdgpu_device *adev,
+					    struct dma_fence *gang)
+{
+	struct dma_fence *old = NULL;
+
+	do {
+		dma_fence_put(old);
+		rcu_read_lock();
+		old = dma_fence_get_rcu_safe(&adev->gang_submit);
+		rcu_read_unlock();
+
+		if (old == gang)
+			break;
+
+		if (!dma_fence_is_signaled(old))
+			return old;
+
+	} while (cmpxchg((struct dma_fence __force **)&adev->gang_submit,
+			 old, gang) != old);
+
+	dma_fence_put(old);
+	return NULL;
+}

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

@@ -173,11 +173,29 @@ static void amdgpu_job_free_cb(struct drm_sched_job *s_job)
 	dma_fence_put(&job->hw_fence);
 }
 
+void amdgpu_job_set_gang_leader(struct amdgpu_job *job,
+				struct amdgpu_job *leader)
+{
+	struct dma_fence *fence = &leader->base.s_fence->scheduled;
+
+	WARN_ON(job->gang_submit);
+
+	/*
+	 * Don't add a reference when we are the gang leader to avoid circle
+	 * dependency.
+	 */
+	if (job != leader)
+		dma_fence_get(fence);
+	job->gang_submit = fence;
+}
+
 void amdgpu_job_free(struct amdgpu_job *job)
 {
 	amdgpu_job_free_resources(job);
 	amdgpu_sync_free(&job->sync);
 	amdgpu_sync_free(&job->sched_sync);
+	if (job->gang_submit != &job->base.s_fence->scheduled)
+		dma_fence_put(job->gang_submit);
 
 	if (!job->hw_fence.ops)
 		kfree(job);
@@ -247,12 +265,16 @@ static struct dma_fence *amdgpu_job_dependency(struct drm_sched_job *sched_job,
 		fence = amdgpu_sync_get_fence(&job->sync);
 	}
 
+	if (!fence && job->gang_submit)
+		fence = amdgpu_device_switch_gang(ring->adev, job->gang_submit);
+
 	return fence;
 }
 
 static struct dma_fence *amdgpu_job_run(struct drm_sched_job *sched_job)
 {
 	struct amdgpu_ring *ring = to_amdgpu_ring(sched_job->sched);
+	struct amdgpu_device *adev = ring->adev;
 	struct dma_fence *fence = NULL, *finished;
 	struct amdgpu_job *job;
 	int r = 0;
@@ -264,8 +286,10 @@ static struct dma_fence *amdgpu_job_run(struct drm_sched_job *sched_job)
 
 	trace_amdgpu_sched_run_job(job);
 
-	if (job->vram_lost_counter != atomic_read(&ring->adev->vram_lost_counter))
-		dma_fence_set_error(finished, -ECANCELED);/* skip IB as well if VRAM lost */
+	/* Skip job if VRAM is lost and never resubmit gangs */
+	if (job->vram_lost_counter != atomic_read(&adev->vram_lost_counter) ||
+	    (job->job_run_counter && job->gang_submit))
+		dma_fence_set_error(finished, -ECANCELED);
 
 	if (finished->error < 0) {
 		DRM_INFO("Skip scheduling IBs!\n");

drivers/gpu/drm/amd/amdgpu/amdgpu_job.h

@@ -50,6 +50,7 @@ struct amdgpu_job {
 	struct amdgpu_sync	sync;
 	struct amdgpu_sync	sched_sync;
 	struct dma_fence	hw_fence;
+	struct dma_fence	*gang_submit;
 	uint32_t		preamble_status;
 	uint32_t                preemption_status;
 	bool                    vm_needs_flush;
@@ -84,6 +85,8 @@ int amdgpu_job_alloc_with_ib(struct amdgpu_device *adev, unsigned size,
 void amdgpu_job_set_resources(struct amdgpu_job *job, struct amdgpu_bo *gds,
 			      struct amdgpu_bo *gws, struct amdgpu_bo *oa);
 void amdgpu_job_free_resources(struct amdgpu_job *job);
+void amdgpu_job_set_gang_leader(struct amdgpu_job *job,
+				struct amdgpu_job *leader);
 void amdgpu_job_free(struct amdgpu_job *job);
 int amdgpu_job_submit(struct amdgpu_job *job, struct drm_sched_entity *entity,
 		      void *owner, struct dma_fence **f);