1 files changed, 64 insertions, 17 deletions

diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_events.c b/drivers/gpu/drm/amd/amdkfd/kfd_events.c
index 9b33d9d2c9ad..5a190dd6be4e 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_events.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_events.c
@@ -31,6 +31,7 @@
 #include <linux/memory.h>
 #include "kfd_priv.h"
 #include "kfd_events.h"
+#include "kfd_device_queue_manager.h"
 #include <linux/device.h>
 
 /*
@@ -726,7 +727,7 @@ void kfd_signal_event_interrupt(u32 pasid, uint32_t partial_id,
 	 * to process context, kfd_process could attempt to exit while we are
 	 * running so the lookup function increments the process ref count.
 	 */
-	struct kfd_process *p = kfd_lookup_process_by_pasid(pasid);
+	struct kfd_process *p = kfd_lookup_process_by_pasid(pasid, NULL);
 
 	if (!p)
 		return; /* Presumably process exited. */
@@ -1127,8 +1128,8 @@ static void lookup_events_by_type_and_signal(struct kfd_process *p,
 
 	if (type == KFD_EVENT_TYPE_MEMORY) {
 		dev_warn(kfd_device,
-			"Sending SIGSEGV to process %d (pasid 0x%x)",
-				p->lead_thread->pid, p->pasid);
+			"Sending SIGSEGV to process pid %d",
+				p->lead_thread->pid);
 		send_sig(SIGSEGV, p->lead_thread, 0);
 	}
 
@@ -1136,13 +1137,13 @@ static void lookup_events_by_type_and_signal(struct kfd_process *p,
 	if (send_signal) {
 		if (send_sigterm) {
 			dev_warn(kfd_device,
-				"Sending SIGTERM to process %d (pasid 0x%x)",
-					p->lead_thread->pid, p->pasid);
+				"Sending SIGTERM to process pid %d",
+					p->lead_thread->pid);
 			send_sig(SIGTERM, p->lead_thread, 0);
 		} else {
 			dev_err(kfd_device,
-				"Process %d (pasid 0x%x) got unhandled exception",
-				p->lead_thread->pid, p->pasid);
+				"Process pid %d got unhandled exception",
+				p->lead_thread->pid);
 		}
 	}
 
@@ -1156,7 +1157,7 @@ void kfd_signal_hw_exception_event(u32 pasid)
 	 * to process context, kfd_process could attempt to exit while we are
 	 * running so the lookup function increments the process ref count.
 	 */
-	struct kfd_process *p = kfd_lookup_process_by_pasid(pasid);
+	struct kfd_process *p = kfd_lookup_process_by_pasid(pasid, NULL);
 
 	if (!p)
 		return; /* Presumably process exited. */
@@ -1165,22 +1166,39 @@ void kfd_signal_hw_exception_event(u32 pasid)
 	kfd_unref_process(p);
 }
 
-void kfd_signal_vm_fault_event(struct kfd_node *dev, u32 pasid,
+void kfd_signal_vm_fault_event_with_userptr(struct kfd_process *p, uint64_t gpu_va)
+{
+	struct kfd_process_device *pdd;
+	struct kfd_hsa_memory_exception_data exception_data;
+	int i;
+
+	memset(&exception_data, 0, sizeof(exception_data));
+	exception_data.va = gpu_va;
+	exception_data.failure.NotPresent = 1;
+
+	// Send VM seg fault to all kfd process device
+	for (i = 0; i < p->n_pdds; i++) {
+		pdd = p->pdds[i];
+		exception_data.gpu_id = pdd->user_gpu_id;
+		kfd_evict_process_device(pdd);
+		kfd_signal_vm_fault_event(pdd, NULL, &exception_data);
+	}
+}
+
+void kfd_signal_vm_fault_event(struct kfd_process_device *pdd,
 				struct kfd_vm_fault_info *info,
 				struct kfd_hsa_memory_exception_data *data)
 {
 	struct kfd_event *ev;
 	uint32_t id;
-	struct kfd_process *p = kfd_lookup_process_by_pasid(pasid);
+	struct kfd_process *p = pdd->process;
 	struct kfd_hsa_memory_exception_data memory_exception_data;
 	int user_gpu_id;
 
-	if (!p)
-		return; /* Presumably process exited. */
-
-	user_gpu_id = kfd_process_get_user_gpu_id(p, dev->id);
+	user_gpu_id = kfd_process_get_user_gpu_id(p, pdd->dev->id);
 	if (unlikely(user_gpu_id == -EINVAL)) {
-		WARN_ONCE(1, "Could not get user_gpu_id from dev->id:%x\n", dev->id);
+		WARN_ONCE(1, "Could not get user_gpu_id from dev->id:%x\n",
+			  pdd->dev->id);
 		return;
 	}
 
@@ -1217,7 +1235,6 @@ void kfd_signal_vm_fault_event(struct kfd_node *dev, u32 pasid,
 		}
 
 	rcu_read_unlock();
-	kfd_unref_process(p);
 }
 
 void kfd_signal_reset_event(struct kfd_node *dev)
@@ -1244,12 +1261,41 @@ void kfd_signal_reset_event(struct kfd_node *dev)
 	idx = srcu_read_lock(&kfd_processes_srcu);
 	hash_for_each_rcu(kfd_processes_table, temp, p, kfd_processes) {
 		int user_gpu_id = kfd_process_get_user_gpu_id(p, dev->id);
+		struct kfd_process_device *pdd = kfd_get_process_device_data(dev, p);
 
 		if (unlikely(user_gpu_id == -EINVAL)) {
 			WARN_ONCE(1, "Could not get user_gpu_id from dev->id:%x\n", dev->id);
 			continue;
 		}
 
+		if (unlikely(!pdd)) {
+			WARN_ONCE(1, "Could not get device data from process pid:%d\n",
+				  p->lead_thread->pid);
+			continue;
+		}
+
+		if (dev->dqm->detect_hang_count && !pdd->has_reset_queue)
+			continue;
+
+		if (dev->dqm->detect_hang_count) {
+			struct amdgpu_task_info *ti;
+			struct amdgpu_fpriv *drv_priv;
+
+			if (unlikely(amdgpu_file_to_fpriv(pdd->drm_file, &drv_priv))) {
+				WARN_ONCE(1, "Could not get vm for device %x from pid:%d\n",
+					  dev->id, p->lead_thread->pid);
+				continue;
+			}
+
+			ti = amdgpu_vm_get_task_info_vm(&drv_priv->vm);
+			if (ti) {
+				dev_err(dev->adev->dev,
+					"Queues reset on process %s tid %d thread %s pid %d\n",
+					ti->process_name, ti->tgid, ti->task.comm, ti->task.pid);
+				amdgpu_vm_put_task_info(ti);
+			}
+		}
+
 		rcu_read_lock();
 
 		id = KFD_FIRST_NONSIGNAL_EVENT_ID;
@@ -1278,7 +1324,7 @@ void kfd_signal_reset_event(struct kfd_node *dev)
 
 void kfd_signal_poison_consumed_event(struct kfd_node *dev, u32 pasid)
 {
-	struct kfd_process *p = kfd_lookup_process_by_pasid(pasid);
+	struct kfd_process *p = kfd_lookup_process_by_pasid(pasid, NULL);
 	struct kfd_hsa_memory_exception_data memory_exception_data;
 	struct kfd_hsa_hw_exception_data hw_exception_data;
 	struct kfd_event *ev;
@@ -1293,6 +1339,7 @@ void kfd_signal_poison_consumed_event(struct kfd_node *dev, u32 pasid)
 	user_gpu_id = kfd_process_get_user_gpu_id(p, dev->id);
 	if (unlikely(user_gpu_id == -EINVAL)) {
 		WARN_ONCE(1, "Could not get user_gpu_id from dev->id:%x\n", dev->id);
+		kfd_unref_process(p);
 		return;
 	}