[PATCH 04/14] drm/amdkfd: Update SMI events for GFX9.4.3

Alex Deucher alexander.deucher at amd.com
Wed Mar 29 20:14:16 UTC 2023


From: Mukul Joshi <mukul.joshi at amd.com>

On GFX 9.4.3, there can be multiple KFD nodes. As a result,
SMI events for SVM, queue evict/restore should be raised for
each node independently.

Signed-off-by: Mukul Joshi <mukul.joshi at amd.com>
Reviewed-by: Felix Kuehling <Felix.Kuehling at amd.com>
Signed-off-by: Alex Deucher <alexander.deucher at amd.com>
---
 drivers/gpu/drm/amd/amdkfd/kfd_migrate.c    | 16 ++++-----
 drivers/gpu/drm/amd/amdkfd/kfd_process.c    |  6 ++--
 drivers/gpu/drm/amd/amdkfd/kfd_smi_events.c | 38 ++++++++++-----------
 drivers/gpu/drm/amd/amdkfd/kfd_smi_events.h | 14 ++++----
 drivers/gpu/drm/amd/amdkfd/kfd_svm.c        |  6 ++--
 5 files changed, 40 insertions(+), 40 deletions(-)

diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_migrate.c b/drivers/gpu/drm/amd/amdkfd/kfd_migrate.c
index e7e5abc32c84..42e599912e52 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_migrate.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_migrate.c
@@ -423,9 +423,9 @@ svm_migrate_vma_to_vram(struct kfd_node *node, struct svm_range *prange,
 	migrate.dst = migrate.src + npages;
 	scratch = (dma_addr_t *)(migrate.dst + npages);
 
-	kfd_smi_event_migration_start(adev->kfd.dev, p->lead_thread->pid,
+	kfd_smi_event_migration_start(node, p->lead_thread->pid,
 				      start >> PAGE_SHIFT, end >> PAGE_SHIFT,
-				      0, adev->kfd.dev->nodes[0]->id, prange->prefetch_loc,
+				      0, node->id, prange->prefetch_loc,
 				      prange->preferred_loc, trigger);
 
 	r = migrate_vma_setup(&migrate);
@@ -456,9 +456,9 @@ svm_migrate_vma_to_vram(struct kfd_node *node, struct svm_range *prange,
 	svm_migrate_copy_done(adev, mfence);
 	migrate_vma_finalize(&migrate);
 
-	kfd_smi_event_migration_end(adev->kfd.dev, p->lead_thread->pid,
+	kfd_smi_event_migration_end(node, p->lead_thread->pid,
 				    start >> PAGE_SHIFT, end >> PAGE_SHIFT,
-				    0, adev->kfd.dev->nodes[0]->id, trigger);
+				    0, node->id, trigger);
 
 	svm_range_dma_unmap(adev->dev, scratch, 0, npages);
 	svm_range_free_dma_mappings(prange);
@@ -702,9 +702,9 @@ svm_migrate_vma_to_ram(struct kfd_node *node, struct svm_range *prange,
 	migrate.fault_page = fault_page;
 	scratch = (dma_addr_t *)(migrate.dst + npages);
 
-	kfd_smi_event_migration_start(adev->kfd.dev, p->lead_thread->pid,
+	kfd_smi_event_migration_start(node, p->lead_thread->pid,
 				      start >> PAGE_SHIFT, end >> PAGE_SHIFT,
-				      adev->kfd.dev->nodes[0]->id, 0, prange->prefetch_loc,
+				      node->id, 0, prange->prefetch_loc,
 				      prange->preferred_loc, trigger);
 
 	r = migrate_vma_setup(&migrate);
@@ -738,9 +738,9 @@ svm_migrate_vma_to_ram(struct kfd_node *node, struct svm_range *prange,
 	svm_migrate_copy_done(adev, mfence);
 	migrate_vma_finalize(&migrate);
 
-	kfd_smi_event_migration_end(adev->kfd.dev, p->lead_thread->pid,
+	kfd_smi_event_migration_end(node, p->lead_thread->pid,
 				    start >> PAGE_SHIFT, end >> PAGE_SHIFT,
-				    adev->kfd.dev->nodes[0]->id, 0, trigger);
+				    node->id, 0, trigger);
 
 	svm_range_dma_unmap(adev->dev, scratch, 0, npages);
 
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_process.c b/drivers/gpu/drm/amd/amdkfd/kfd_process.c
index 273002466dd9..0a9103326157 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_process.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_process.c
@@ -1817,7 +1817,7 @@ int kfd_process_evict_queues(struct kfd_process *p, uint32_t trigger)
 	for (i = 0; i < p->n_pdds; i++) {
 		struct kfd_process_device *pdd = p->pdds[i];
 
-		kfd_smi_event_queue_eviction(pdd->dev->kfd, p->lead_thread->pid,
+		kfd_smi_event_queue_eviction(pdd->dev, p->lead_thread->pid,
 					     trigger);
 
 		r = pdd->dev->dqm->ops.evict_process_queues(pdd->dev->dqm,
@@ -1845,7 +1845,7 @@ int kfd_process_evict_queues(struct kfd_process *p, uint32_t trigger)
 		if (n_evicted == 0)
 			break;
 
-		kfd_smi_event_queue_restore(pdd->dev->kfd, p->lead_thread->pid);
+		kfd_smi_event_queue_restore(pdd->dev, p->lead_thread->pid);
 
 		if (pdd->dev->dqm->ops.restore_process_queues(pdd->dev->dqm,
 							      &pdd->qpd))
@@ -1866,7 +1866,7 @@ int kfd_process_restore_queues(struct kfd_process *p)
 	for (i = 0; i < p->n_pdds; i++) {
 		struct kfd_process_device *pdd = p->pdds[i];
 
-		kfd_smi_event_queue_restore(pdd->dev->kfd, p->lead_thread->pid);
+		kfd_smi_event_queue_restore(pdd->dev, p->lead_thread->pid);
 
 		r = pdd->dev->dqm->ops.restore_process_queues(pdd->dev->dqm,
 							      &pdd->qpd);
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_smi_events.c b/drivers/gpu/drm/amd/amdkfd/kfd_smi_events.c
index b703da59e067..d9953c2b2661 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_smi_events.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_smi_events.c
@@ -250,58 +250,58 @@ void kfd_smi_event_update_vmfault(struct kfd_node *dev, uint16_t pasid)
 			  task_info.pid, task_info.task_name);
 }
 
-void kfd_smi_event_page_fault_start(struct kfd_dev *dev, pid_t pid,
+void kfd_smi_event_page_fault_start(struct kfd_node *node, pid_t pid,
 				    unsigned long address, bool write_fault,
 				    ktime_t ts)
 {
-	kfd_smi_event_add(pid, dev->nodes[0], KFD_SMI_EVENT_PAGE_FAULT_START,
+	kfd_smi_event_add(pid, node, KFD_SMI_EVENT_PAGE_FAULT_START,
 			  "%lld -%d @%lx(%x) %c\n", ktime_to_ns(ts), pid,
-			  address, dev->nodes[0]->id, write_fault ? 'W' : 'R');
+			  address, node->id, write_fault ? 'W' : 'R');
 }
 
-void kfd_smi_event_page_fault_end(struct kfd_dev *dev, pid_t pid,
+void kfd_smi_event_page_fault_end(struct kfd_node *node, pid_t pid,
 				  unsigned long address, bool migration)
 {
-	kfd_smi_event_add(pid, dev->nodes[0], KFD_SMI_EVENT_PAGE_FAULT_END,
+	kfd_smi_event_add(pid, node, KFD_SMI_EVENT_PAGE_FAULT_END,
 			  "%lld -%d @%lx(%x) %c\n", ktime_get_boottime_ns(),
-			  pid, address, dev->nodes[0]->id, migration ? 'M' : 'U');
+			  pid, address, node->id, migration ? 'M' : 'U');
 }
 
-void kfd_smi_event_migration_start(struct kfd_dev *dev, pid_t pid,
+void kfd_smi_event_migration_start(struct kfd_node *node, pid_t pid,
 				   unsigned long start, unsigned long end,
 				   uint32_t from, uint32_t to,
 				   uint32_t prefetch_loc, uint32_t preferred_loc,
 				   uint32_t trigger)
 {
-	kfd_smi_event_add(pid, dev->nodes[0], KFD_SMI_EVENT_MIGRATE_START,
+	kfd_smi_event_add(pid, node, KFD_SMI_EVENT_MIGRATE_START,
 			  "%lld -%d @%lx(%lx) %x->%x %x:%x %d\n",
 			  ktime_get_boottime_ns(), pid, start, end - start,
 			  from, to, prefetch_loc, preferred_loc, trigger);
 }
 
-void kfd_smi_event_migration_end(struct kfd_dev *dev, pid_t pid,
+void kfd_smi_event_migration_end(struct kfd_node *node, pid_t pid,
 				 unsigned long start, unsigned long end,
 				 uint32_t from, uint32_t to, uint32_t trigger)
 {
-	kfd_smi_event_add(pid, dev->nodes[0], KFD_SMI_EVENT_MIGRATE_END,
+	kfd_smi_event_add(pid, node, KFD_SMI_EVENT_MIGRATE_END,
 			  "%lld -%d @%lx(%lx) %x->%x %d\n",
 			  ktime_get_boottime_ns(), pid, start, end - start,
 			  from, to, trigger);
 }
 
-void kfd_smi_event_queue_eviction(struct kfd_dev *dev, pid_t pid,
+void kfd_smi_event_queue_eviction(struct kfd_node *node, pid_t pid,
 				  uint32_t trigger)
 {
-	kfd_smi_event_add(pid, dev->nodes[0], KFD_SMI_EVENT_QUEUE_EVICTION,
+	kfd_smi_event_add(pid, node, KFD_SMI_EVENT_QUEUE_EVICTION,
 			  "%lld -%d %x %d\n", ktime_get_boottime_ns(), pid,
-			  dev->nodes[0]->id, trigger);
+			  node->id, trigger);
 }
 
-void kfd_smi_event_queue_restore(struct kfd_dev *dev, pid_t pid)
+void kfd_smi_event_queue_restore(struct kfd_node *node, pid_t pid)
 {
-	kfd_smi_event_add(pid, dev->nodes[0], KFD_SMI_EVENT_QUEUE_RESTORE,
+	kfd_smi_event_add(pid, node, KFD_SMI_EVENT_QUEUE_RESTORE,
 			  "%lld -%d %x\n", ktime_get_boottime_ns(), pid,
-			  dev->nodes[0]->id);
+			  node->id);
 }
 
 void kfd_smi_event_queue_restore_rescheduled(struct mm_struct *mm)
@@ -324,13 +324,13 @@ void kfd_smi_event_queue_restore_rescheduled(struct mm_struct *mm)
 	kfd_unref_process(p);
 }
 
-void kfd_smi_event_unmap_from_gpu(struct kfd_dev *dev, pid_t pid,
+void kfd_smi_event_unmap_from_gpu(struct kfd_node *node, pid_t pid,
 				  unsigned long address, unsigned long last,
 				  uint32_t trigger)
 {
-	kfd_smi_event_add(pid, dev->nodes[0], KFD_SMI_EVENT_UNMAP_FROM_GPU,
+	kfd_smi_event_add(pid, node, KFD_SMI_EVENT_UNMAP_FROM_GPU,
 			  "%lld -%d @%lx(%lx) %x %d\n", ktime_get_boottime_ns(),
-			  pid, address, last - address + 1, dev->nodes[0]->id, trigger);
+			  pid, address, last - address + 1, node->id, trigger);
 }
 
 int kfd_smi_event_open(struct kfd_node *dev, uint32_t *fd)
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_smi_events.h b/drivers/gpu/drm/amd/amdkfd/kfd_smi_events.h
index 59cd089f80d1..fa95c2dfd587 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_smi_events.h
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_smi_events.h
@@ -29,24 +29,24 @@ void kfd_smi_event_update_vmfault(struct kfd_node *dev, uint16_t pasid);
 void kfd_smi_event_update_thermal_throttling(struct kfd_node *dev,
 					     uint64_t throttle_bitmask);
 void kfd_smi_event_update_gpu_reset(struct kfd_node *dev, bool post_reset);
-void kfd_smi_event_page_fault_start(struct kfd_dev *dev, pid_t pid,
+void kfd_smi_event_page_fault_start(struct kfd_node *node, pid_t pid,
 				    unsigned long address, bool write_fault,
 				    ktime_t ts);
-void kfd_smi_event_page_fault_end(struct kfd_dev *dev, pid_t pid,
+void kfd_smi_event_page_fault_end(struct kfd_node *node, pid_t pid,
 				  unsigned long address, bool migration);
-void kfd_smi_event_migration_start(struct kfd_dev *dev, pid_t pid,
+void kfd_smi_event_migration_start(struct kfd_node *node, pid_t pid,
 			     unsigned long start, unsigned long end,
 			     uint32_t from, uint32_t to,
 			     uint32_t prefetch_loc, uint32_t preferred_loc,
 			     uint32_t trigger);
-void kfd_smi_event_migration_end(struct kfd_dev *dev, pid_t pid,
+void kfd_smi_event_migration_end(struct kfd_node *node, pid_t pid,
 			     unsigned long start, unsigned long end,
 			     uint32_t from, uint32_t to, uint32_t trigger);
-void kfd_smi_event_queue_eviction(struct kfd_dev *dev, pid_t pid,
+void kfd_smi_event_queue_eviction(struct kfd_node *node, pid_t pid,
 				  uint32_t trigger);
-void kfd_smi_event_queue_restore(struct kfd_dev *dev, pid_t pid);
+void kfd_smi_event_queue_restore(struct kfd_node *node, pid_t pid);
 void kfd_smi_event_queue_restore_rescheduled(struct mm_struct *mm);
-void kfd_smi_event_unmap_from_gpu(struct kfd_dev *dev, pid_t pid,
+void kfd_smi_event_unmap_from_gpu(struct kfd_node *node, pid_t pid,
 				  unsigned long address, unsigned long last,
 				  uint32_t trigger);
 #endif
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_svm.c b/drivers/gpu/drm/amd/amdkfd/kfd_svm.c
index b3c10d8fbe0b..0e2b21ec468c 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_svm.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_svm.c
@@ -1271,7 +1271,7 @@ svm_range_unmap_from_gpus(struct svm_range *prange, unsigned long start,
 			return -EINVAL;
 		}
 
-		kfd_smi_event_unmap_from_gpu(pdd->dev->kfd, p->lead_thread->pid,
+		kfd_smi_event_unmap_from_gpu(pdd->dev, p->lead_thread->pid,
 					     start, last, trigger);
 
 		r = svm_range_unmap_from_gpu(pdd->dev->adev,
@@ -2923,7 +2923,7 @@ svm_range_restore_pages(struct amdgpu_device *adev, unsigned int pasid,
 		 svms, prange->start, prange->last, best_loc,
 		 prange->actual_loc);
 
-	kfd_smi_event_page_fault_start(adev->kfd.dev, p->lead_thread->pid, addr,
+	kfd_smi_event_page_fault_start(node, p->lead_thread->pid, addr,
 				       write_fault, timestamp);
 
 	if (prange->actual_loc != best_loc) {
@@ -2961,7 +2961,7 @@ svm_range_restore_pages(struct amdgpu_device *adev, unsigned int pasid,
 		pr_debug("failed %d to map svms 0x%p [0x%lx 0x%lx] to gpus\n",
 			 r, svms, prange->start, prange->last);
 
-	kfd_smi_event_page_fault_end(adev->kfd.dev, p->lead_thread->pid, addr,
+	kfd_smi_event_page_fault_end(node, p->lead_thread->pid, addr,
 				     migration);
 
 out_unlock_range:
-- 
2.39.2



More information about the amd-gfx mailing list