1 files changed, 520 insertions, 231 deletions

diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_events.c b/drivers/gpu/drm/amd/amdkfd/kfd_events.c
index 3eea4edee355..5a190dd6be4e 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_events.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_events.c
@@ -1,5 +1,6 @@
+// SPDX-License-Identifier: GPL-2.0 OR MIT
 /*
- * Copyright 2014 Advanced Micro Devices, Inc.
+ * Copyright 2014-2022 Advanced Micro Devices, Inc.
  *
  * Permission is hereby granted, free of charge, to any person obtaining a
  * copy of this software and associated documentation files (the "Software"),
@@ -30,7 +31,7 @@
 #include <linux/memory.h>
 #include "kfd_priv.h"
 #include "kfd_events.h"
-#include "kfd_iommu.h"
+#include "kfd_device_queue_manager.h"
 #include <linux/device.h>
 
 /*
@@ -40,6 +41,7 @@ struct kfd_event_waiter {
 	wait_queue_entry_t wait;
 	struct kfd_event *event; /* Event to wait for */
 	bool activated;		 /* Becomes true when event is signaled */
+	bool event_age_enabled;  /* set to true when last_event_age is non-zero */
 };
 
 /*
@@ -55,7 +57,6 @@ struct kfd_signal_page {
 	bool need_to_free_pages;
 };
 
-
 static uint64_t *page_slots(struct kfd_signal_page *page)
 {
 	return page->kernel_address;
@@ -92,7 +93,8 @@ fail_alloc_signal_store:
 }
 
 static int allocate_event_notification_slot(struct kfd_process *p,
-					    struct kfd_event *ev)
+					    struct kfd_event *ev,
+					    const int *restore_id)
 {
 	int id;
 
@@ -104,14 +106,19 @@ static int allocate_event_notification_slot(struct kfd_process *p,
 		p->signal_mapped_size = 256*8;
 	}
 
-	/*
-	 * Compatibility with old user mode: Only use signal slots
-	 * user mode has mapped, may be less than
-	 * KFD_SIGNAL_EVENT_LIMIT. This also allows future increase
-	 * of the event limit without breaking user mode.
-	 */
-	id = idr_alloc(&p->event_idr, ev, 0, p->signal_mapped_size / 8,
-		       GFP_KERNEL);
+	if (restore_id) {
+		id = idr_alloc(&p->event_idr, ev, *restore_id, *restore_id + 1,
+				GFP_KERNEL);
+	} else {
+		/*
+		 * Compatibility with old user mode: Only use signal slots
+		 * user mode has mapped, may be less than
+		 * KFD_SIGNAL_EVENT_LIMIT. This also allows future increase
+		 * of the event limit without breaking user mode.
+		 */
+		id = idr_alloc(&p->event_idr, ev, 0, p->signal_mapped_size / 8,
+				GFP_KERNEL);
+	}
 	if (id < 0)
 		return id;
 
@@ -122,8 +129,8 @@ static int allocate_event_notification_slot(struct kfd_process *p,
 }
 
 /*
- * Assumes that p->event_mutex is held and of course that p is not going
- * away (current or locked).
+ * Assumes that p->event_mutex or rcu_readlock is held and of course that p is
+ * not going away.
  */
 static struct kfd_event *lookup_event_by_id(struct kfd_process *p, uint32_t id)
 {
@@ -178,9 +185,8 @@ static struct kfd_event *lookup_signaled_event_by_partial_id(
 	return ev;
 }
 
-static int create_signal_event(struct file *devkfd,
-				struct kfd_process *p,
-				struct kfd_event *ev)
+static int create_signal_event(struct file *devkfd, struct kfd_process *p,
+				struct kfd_event *ev, const int *restore_id)
 {
 	int ret;
 
@@ -193,7 +199,7 @@ static int create_signal_event(struct file *devkfd,
 		return -ENOSPC;
 	}
 
-	ret = allocate_event_notification_slot(p, ev);
+	ret = allocate_event_notification_slot(p, ev, restore_id);
 	if (ret) {
 		pr_warn("Signal event wasn't created because out of kernel memory\n");
 		return ret;
@@ -209,16 +215,22 @@ static int create_signal_event(struct file *devkfd,
 	return 0;
 }
 
-static int create_other_event(struct kfd_process *p, struct kfd_event *ev)
+static int create_other_event(struct kfd_process *p, struct kfd_event *ev, const int *restore_id)
 {
-	/* Cast KFD_LAST_NONSIGNAL_EVENT to uint32_t. This allows an
-	 * intentional integer overflow to -1 without a compiler
-	 * warning. idr_alloc treats a negative value as "maximum
-	 * signed integer".
-	 */
-	int id = idr_alloc(&p->event_idr, ev, KFD_FIRST_NONSIGNAL_EVENT_ID,
-			   (uint32_t)KFD_LAST_NONSIGNAL_EVENT_ID + 1,
-			   GFP_KERNEL);
+	int id;
+
+	if (restore_id)
+		id = idr_alloc(&p->event_idr, ev, *restore_id, *restore_id + 1,
+			GFP_KERNEL);
+	else
+		/* Cast KFD_LAST_NONSIGNAL_EVENT to uint32_t. This allows an
+		 * intentional integer overflow to -1 without a compiler
+		 * warning. idr_alloc treats a negative value as "maximum
+		 * signed integer".
+		 */
+		id = idr_alloc(&p->event_idr, ev, KFD_FIRST_NONSIGNAL_EVENT_ID,
+				(uint32_t)KFD_LAST_NONSIGNAL_EVENT_ID + 1,
+				GFP_KERNEL);
 
 	if (id < 0)
 		return id;
@@ -227,12 +239,24 @@ static int create_other_event(struct kfd_process *p, struct kfd_event *ev)
 	return 0;
 }
 
-void kfd_event_init_process(struct kfd_process *p)
+int kfd_event_init_process(struct kfd_process *p)
 {
+	int id;
+
 	mutex_init(&p->event_mutex);
 	idr_init(&p->event_idr);
 	p->signal_page = NULL;
-	p->signal_event_count = 0;
+	p->signal_event_count = 1;
+	/* Allocate event ID 0. It is used for a fast path to ignore bogus events
+	 * that are sent by the CP without a context ID
+	 */
+	id = idr_alloc(&p->event_idr, NULL, 0, 1, GFP_KERNEL);
+	if (id < 0) {
+		idr_destroy(&p->event_idr);
+		mutex_destroy(&p->event_mutex);
+		return id;
+	}
+	return 0;
 }
 
 static void destroy_event(struct kfd_process *p, struct kfd_event *ev)
@@ -240,16 +264,18 @@ static void destroy_event(struct kfd_process *p, struct kfd_event *ev)
 	struct kfd_event_waiter *waiter;
 
 	/* Wake up pending waiters. They will return failure */
+	spin_lock(&ev->lock);
 	list_for_each_entry(waiter, &ev->wq.head, wait.entry)
-		waiter->event = NULL;
+		WRITE_ONCE(waiter->event, NULL);
 	wake_up_all(&ev->wq);
+	spin_unlock(&ev->lock);
 
 	if (ev->type == KFD_EVENT_TYPE_SIGNAL ||
 	    ev->type == KFD_EVENT_TYPE_DEBUG)
 		p->signal_event_count--;
 
 	idr_remove(&p->event_idr, ev->event_id);
-	kfree(ev);
+	kfree_rcu(ev, rcu);
 }
 
 static void destroy_events(struct kfd_process *p)
@@ -258,8 +284,10 @@ static void destroy_events(struct kfd_process *p)
 	uint32_t id;
 
 	idr_for_each_entry(&p->event_idr, ev, id)
-		destroy_event(p, ev);
+		if (ev)
+			destroy_event(p, ev);
 	idr_destroy(&p->event_idr);
+	mutex_destroy(&p->event_mutex);
 }
 
 /*
@@ -295,8 +323,8 @@ static bool event_can_be_cpu_signaled(const struct kfd_event *ev)
 	return ev->type == KFD_EVENT_TYPE_SIGNAL;
 }
 
-int kfd_event_page_set(struct kfd_process *p, void *kernel_address,
-		       uint64_t size)
+static int kfd_event_page_set(struct kfd_process *p, void *kernel_address,
+		       uint64_t size, uint64_t user_handle)
 {
 	struct kfd_signal_page *page;
 
@@ -315,10 +343,56 @@ int kfd_event_page_set(struct kfd_process *p, void *kernel_address,
 
 	p->signal_page = page;
 	p->signal_mapped_size = size;
-
+	p->signal_handle = user_handle;
 	return 0;
 }
 
+int kfd_kmap_event_page(struct kfd_process *p, uint64_t event_page_offset)
+{
+	struct kfd_node *kfd;
+	struct kfd_process_device *pdd;
+	void *mem, *kern_addr;
+	uint64_t size;
+	int err = 0;
+
+	if (p->signal_page) {
+		pr_err("Event page is already set\n");
+		return -EINVAL;
+	}
+
+	pdd = kfd_process_device_data_by_id(p, GET_GPU_ID(event_page_offset));
+	if (!pdd) {
+		pr_err("Getting device by id failed in %s\n", __func__);
+		return -EINVAL;
+	}
+	kfd = pdd->dev;
+
+	pdd = kfd_bind_process_to_device(kfd, p);
+	if (IS_ERR(pdd))
+		return PTR_ERR(pdd);
+
+	mem = kfd_process_device_translate_handle(pdd,
+			GET_IDR_HANDLE(event_page_offset));
+	if (!mem) {
+		pr_err("Can't find BO, offset is 0x%llx\n", event_page_offset);
+		return -EINVAL;
+	}
+
+	err = amdgpu_amdkfd_gpuvm_map_gtt_bo_to_kernel(mem, &kern_addr, &size);
+	if (err) {
+		pr_err("Failed to map event page to kernel\n");
+		return err;
+	}
+
+	err = kfd_event_page_set(p, kern_addr, size, event_page_offset);
+	if (err) {
+		pr_err("Failed to set event page\n");
+		amdgpu_amdkfd_gpuvm_unmap_gtt_bo_from_kernel(mem);
+		return err;
+	}
+	return err;
+}
+
 int kfd_event_create(struct file *devkfd, struct kfd_process *p,
 		     uint32_t event_type, bool auto_reset, uint32_t node_id,
 		     uint32_t *event_id, uint32_t *event_trigger_data,
@@ -334,6 +408,7 @@ int kfd_event_create(struct file *devkfd, struct kfd_process *p,
 	ev->auto_reset = auto_reset;
 	ev->signaled = false;
 
+	spin_lock_init(&ev->lock);
 	init_waitqueue_head(&ev->wq);
 
 	*event_page_offset = 0;
@@ -343,20 +418,21 @@ int kfd_event_create(struct file *devkfd, struct kfd_process *p,
 	switch (event_type) {
 	case KFD_EVENT_TYPE_SIGNAL:
 	case KFD_EVENT_TYPE_DEBUG:
-		ret = create_signal_event(devkfd, p, ev);
+		ret = create_signal_event(devkfd, p, ev, NULL);
 		if (!ret) {
 			*event_page_offset = KFD_MMAP_TYPE_EVENTS;
 			*event_slot_index = ev->event_id;
 		}
 		break;
 	default:
-		ret = create_other_event(p, ev);
+		ret = create_other_event(p, ev, NULL);
 		break;
 	}
 
 	if (!ret) {
 		*event_id = ev->event_id;
 		*event_trigger_data = ev->event_id;
+		ev->event_age = 1;
 	} else {
 		kfree(ev);
 	}
@@ -366,6 +442,166 @@ int kfd_event_create(struct file *devkfd, struct kfd_process *p,
 	return ret;
 }
 
+int kfd_criu_restore_event(struct file *devkfd,
+			   struct kfd_process *p,
+			   uint8_t __user *user_priv_ptr,
+			   uint64_t *priv_data_offset,
+			   uint64_t max_priv_data_size)
+{
+	struct kfd_criu_event_priv_data *ev_priv;
+	struct kfd_event *ev = NULL;
+	int ret = 0;
+
+	ev_priv = kmalloc(sizeof(*ev_priv), GFP_KERNEL);
+	if (!ev_priv)
+		return -ENOMEM;
+
+	ev = kzalloc(sizeof(*ev), GFP_KERNEL);
+	if (!ev) {
+		ret = -ENOMEM;
+		goto exit;
+	}
+
+	if (*priv_data_offset + sizeof(*ev_priv) > max_priv_data_size) {
+		ret = -EINVAL;
+		goto exit;
+	}
+
+	ret = copy_from_user(ev_priv, user_priv_ptr + *priv_data_offset, sizeof(*ev_priv));
+	if (ret) {
+		ret = -EFAULT;
+		goto exit;
+	}
+	*priv_data_offset += sizeof(*ev_priv);
+
+	if (ev_priv->user_handle) {
+		ret = kfd_kmap_event_page(p, ev_priv->user_handle);
+		if (ret)
+			goto exit;
+	}
+
+	ev->type = ev_priv->type;
+	ev->auto_reset = ev_priv->auto_reset;
+	ev->signaled = ev_priv->signaled;
+
+	spin_lock_init(&ev->lock);
+	init_waitqueue_head(&ev->wq);
+
+	mutex_lock(&p->event_mutex);
+	switch (ev->type) {
+	case KFD_EVENT_TYPE_SIGNAL:
+	case KFD_EVENT_TYPE_DEBUG:
+		ret = create_signal_event(devkfd, p, ev, &ev_priv->event_id);
+		break;
+	case KFD_EVENT_TYPE_MEMORY:
+		memcpy(&ev->memory_exception_data,
+			&ev_priv->memory_exception_data,
+			sizeof(struct kfd_hsa_memory_exception_data));
+
+		ret = create_other_event(p, ev, &ev_priv->event_id);
+		break;
+	case KFD_EVENT_TYPE_HW_EXCEPTION:
+		memcpy(&ev->hw_exception_data,
+			&ev_priv->hw_exception_data,
+			sizeof(struct kfd_hsa_hw_exception_data));
+
+		ret = create_other_event(p, ev, &ev_priv->event_id);
+		break;
+	}
+	mutex_unlock(&p->event_mutex);
+
+exit:
+	if (ret)
+		kfree(ev);
+
+	kfree(ev_priv);
+
+	return ret;
+}
+
+int kfd_criu_checkpoint_events(struct kfd_process *p,
+			 uint8_t __user *user_priv_data,
+			 uint64_t *priv_data_offset)
+{
+	struct kfd_criu_event_priv_data *ev_privs;
+	int i = 0;
+	int ret =  0;
+	struct kfd_event *ev;
+	uint32_t ev_id;
+
+	uint32_t num_events = kfd_get_num_events(p);
+
+	if (!num_events)
+		return 0;
+
+	ev_privs = kvzalloc(num_events * sizeof(*ev_privs), GFP_KERNEL);
+	if (!ev_privs)
+		return -ENOMEM;
+
+
+	idr_for_each_entry(&p->event_idr, ev, ev_id) {
+		struct kfd_criu_event_priv_data *ev_priv;
+
+		/*
+		 * Currently, all events have same size of private_data, but the current ioctl's
+		 * and CRIU plugin supports private_data of variable sizes
+		 */
+		ev_priv = &ev_privs[i];
+
+		ev_priv->object_type = KFD_CRIU_OBJECT_TYPE_EVENT;
+
+		/* We store the user_handle with the first event */
+		if (i == 0 && p->signal_page)
+			ev_priv->user_handle = p->signal_handle;
+
+		ev_priv->event_id = ev->event_id;
+		ev_priv->auto_reset = ev->auto_reset;
+		ev_priv->type = ev->type;
+		ev_priv->signaled = ev->signaled;
+
+		if (ev_priv->type == KFD_EVENT_TYPE_MEMORY)
+			memcpy(&ev_priv->memory_exception_data,
+				&ev->memory_exception_data,
+				sizeof(struct kfd_hsa_memory_exception_data));
+		else if (ev_priv->type == KFD_EVENT_TYPE_HW_EXCEPTION)
+			memcpy(&ev_priv->hw_exception_data,
+				&ev->hw_exception_data,
+				sizeof(struct kfd_hsa_hw_exception_data));
+
+		pr_debug("Checkpointed event[%d] id = 0x%08x auto_reset = %x type = %x signaled = %x\n",
+			  i,
+			  ev_priv->event_id,
+			  ev_priv->auto_reset,
+			  ev_priv->type,
+			  ev_priv->signaled);
+		i++;
+	}
+
+	ret = copy_to_user(user_priv_data + *priv_data_offset,
+			   ev_privs, num_events * sizeof(*ev_privs));
+	if (ret) {
+		pr_err("Failed to copy events priv to user\n");
+		ret = -EFAULT;
+	}
+
+	*priv_data_offset += num_events * sizeof(*ev_privs);
+
+	kvfree(ev_privs);
+	return ret;
+}
+
+int kfd_get_num_events(struct kfd_process *p)
+{
+	struct kfd_event *ev;
+	uint32_t id;
+	u32 num_events = 0;
+
+	idr_for_each_entry(&p->event_idr, ev, id)
+		num_events++;
+
+	return num_events;
+}
+
 /* Assumes that p is current. */
 int kfd_event_destroy(struct kfd_process *p, uint32_t event_id)
 {
@@ -391,13 +627,18 @@ static void set_event(struct kfd_event *ev)
 
 	/* Auto reset if the list is non-empty and we're waking
 	 * someone. waitqueue_active is safe here because we're
-	 * protected by the p->event_mutex, which is also held when
+	 * protected by the ev->lock, which is also held when
 	 * updating the wait queues in kfd_wait_on_events.
 	 */
 	ev->signaled = !ev->auto_reset || !waitqueue_active(&ev->wq);
+	if (!(++ev->event_age)) {
+		/* Never wrap back to reserved/default event age 0/1 */
+		ev->event_age = 2;
+		WARN_ONCE(1, "event_age wrap back!");
+	}
 
 	list_for_each_entry(waiter, &ev->wq.head, wait.entry)
-		waiter->activated = true;
+		WRITE_ONCE(waiter->activated, true);
 
 	wake_up_all(&ev->wq);
 }
@@ -408,16 +649,23 @@ int kfd_set_event(struct kfd_process *p, uint32_t event_id)
 	int ret = 0;
 	struct kfd_event *ev;
 
-	mutex_lock(&p->event_mutex);
+	rcu_read_lock();
 
 	ev = lookup_event_by_id(p, event_id);
+	if (!ev) {
+		ret = -EINVAL;
+		goto unlock_rcu;
+	}
+	spin_lock(&ev->lock);
 
-	if (ev && event_can_be_cpu_signaled(ev))
+	if (event_can_be_cpu_signaled(ev))
 		set_event(ev);
 	else
 		ret = -EINVAL;
 
-	mutex_unlock(&p->event_mutex);
+	spin_unlock(&ev->lock);
+unlock_rcu:
+	rcu_read_unlock();
 	return ret;
 }
 
@@ -432,23 +680,30 @@ int kfd_reset_event(struct kfd_process *p, uint32_t event_id)
 	int ret = 0;
 	struct kfd_event *ev;
 
-	mutex_lock(&p->event_mutex);
+	rcu_read_lock();
 
 	ev = lookup_event_by_id(p, event_id);
+	if (!ev) {
+		ret = -EINVAL;
+		goto unlock_rcu;
+	}
+	spin_lock(&ev->lock);
 
-	if (ev && event_can_be_cpu_signaled(ev))
+	if (event_can_be_cpu_signaled(ev))
 		reset_event(ev);
 	else
 		ret = -EINVAL;
 
-	mutex_unlock(&p->event_mutex);
+	spin_unlock(&ev->lock);
+unlock_rcu:
+	rcu_read_unlock();
 	return ret;
 
 }
 
 static void acknowledge_signal(struct kfd_process *p, struct kfd_event *ev)
 {
-	page_slots(p->signal_page)[ev->event_id] = UNSIGNALED_EVENT_SLOT;
+	WRITE_ONCE(page_slots(p->signal_page)[ev->event_id], UNSIGNALED_EVENT_SLOT);
 }
 
 static void set_event_from_interrupt(struct kfd_process *p,
@@ -456,7 +711,9 @@ static void set_event_from_interrupt(struct kfd_process *p,
 {
 	if (ev && event_can_be_gpu_signaled(ev)) {
 		acknowledge_signal(p, ev);
+		spin_lock(&ev->lock);
 		set_event(ev);
+		spin_unlock(&ev->lock);
 	}
 }
 
@@ -470,12 +727,12 @@ 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. */
 
-	mutex_lock(&p->event_mutex);
+	rcu_read_lock();
 
 	if (valid_id_bits)
 		ev = lookup_signaled_event_by_partial_id(p, partial_id,
@@ -503,7 +760,7 @@ void kfd_signal_event_interrupt(u32 pasid, uint32_t partial_id,
 				if (id >= KFD_SIGNAL_EVENT_LIMIT)
 					break;
 
-				if (slots[id] != UNSIGNALED_EVENT_SLOT)
+				if (READ_ONCE(slots[id]) != UNSIGNALED_EVENT_SLOT)
 					set_event_from_interrupt(p, ev);
 			}
 		} else {
@@ -511,15 +768,15 @@ void kfd_signal_event_interrupt(u32 pasid, uint32_t partial_id,
 			 * iterate over the signal slots and lookup
 			 * only signaled events from the IDR.
 			 */
-			for (id = 0; id < KFD_SIGNAL_EVENT_LIMIT; id++)
-				if (slots[id] != UNSIGNALED_EVENT_SLOT) {
+			for (id = 1; id < KFD_SIGNAL_EVENT_LIMIT; id++)
+				if (READ_ONCE(slots[id]) != UNSIGNALED_EVENT_SLOT) {
 					ev = lookup_event_by_id(p, id);
 					set_event_from_interrupt(p, ev);
 				}
 		}
 	}
 
-	mutex_unlock(&p->event_mutex);
+	rcu_read_unlock();
 	kfd_unref_process(p);
 }
 
@@ -528,43 +785,44 @@ static struct kfd_event_waiter *alloc_event_waiters(uint32_t num_events)
 	struct kfd_event_waiter *event_waiters;
 	uint32_t i;
 
-	event_waiters = kmalloc_array(num_events,
-					sizeof(struct kfd_event_waiter),
-					GFP_KERNEL);
+	event_waiters = kcalloc(num_events, sizeof(struct kfd_event_waiter),
+				GFP_KERNEL);
+	if (!event_waiters)
+		return NULL;
 
-	for (i = 0; (event_waiters) && (i < num_events) ; i++) {
+	for (i = 0; i < num_events; i++)
 		init_wait(&event_waiters[i].wait);
-		event_waiters[i].activated = false;
-	}
 
 	return event_waiters;
 }
 
-static int init_event_waiter_get_status(struct kfd_process *p,
+static int init_event_waiter(struct kfd_process *p,
 		struct kfd_event_waiter *waiter,
-		uint32_t event_id)
+		struct kfd_event_data *event_data)
 {
-	struct kfd_event *ev = lookup_event_by_id(p, event_id);
+	struct kfd_event *ev = lookup_event_by_id(p, event_data->event_id);
 
 	if (!ev)
 		return -EINVAL;
 
+	spin_lock(&ev->lock);
 	waiter->event = ev;
 	waiter->activated = ev->signaled;
 	ev->signaled = ev->signaled && !ev->auto_reset;
 
-	return 0;
-}
-
-static void init_event_waiter_add_to_waitlist(struct kfd_event_waiter *waiter)
-{
-	struct kfd_event *ev = waiter->event;
+	/* last_event_age = 0 reserved for backward compatible */
+	if (waiter->event->type == KFD_EVENT_TYPE_SIGNAL &&
+		event_data->signal_event_data.last_event_age) {
+		waiter->event_age_enabled = true;
+		if (ev->event_age != event_data->signal_event_data.last_event_age)
+			waiter->activated = true;
+	}
 
-	/* Only add to the wait list if we actually need to
-	 * wait on this event.
-	 */
 	if (!waiter->activated)
 		add_wait_queue(&ev->wq, &waiter->wait);
+	spin_unlock(&ev->lock);
+
+	return 0;
 }
 
 /* test_event_condition - Test condition of events being waited for
@@ -584,10 +842,10 @@ static uint32_t test_event_condition(bool all, uint32_t num_events,
 	uint32_t activated_count = 0;
 
 	for (i = 0; i < num_events; i++) {
-		if (!event_waiters[i].event)
+		if (!READ_ONCE(event_waiters[i].event))
 			return KFD_IOC_WAIT_RESULT_FAIL;
 
-		if (event_waiters[i].activated) {
+		if (READ_ONCE(event_waiters[i].activated)) {
 			if (!all)
 				return KFD_IOC_WAIT_RESULT_COMPLETE;
 
@@ -607,30 +865,40 @@ static int copy_signaled_event_data(uint32_t num_events,
 		struct kfd_event_waiter *event_waiters,
 		struct kfd_event_data __user *data)
 {
-	struct kfd_hsa_memory_exception_data *src;
-	struct kfd_hsa_memory_exception_data __user *dst;
+	void *src;
+	void __user *dst;
 	struct kfd_event_waiter *waiter;
 	struct kfd_event *event;
-	uint32_t i;
+	uint32_t i, size = 0;
 
 	for (i = 0; i < num_events; i++) {
 		waiter = &event_waiters[i];
 		event = waiter->event;
-		if (waiter->activated && event->type == KFD_EVENT_TYPE_MEMORY) {
-			dst = &data[i].memory_exception_data;
-			src = &event->memory_exception_data;
-			if (copy_to_user(dst, src,
-				sizeof(struct kfd_hsa_memory_exception_data)))
+		if (!event)
+			return -EINVAL; /* event was destroyed */
+		if (waiter->activated) {
+			if (event->type == KFD_EVENT_TYPE_MEMORY) {
+				dst = &data[i].memory_exception_data;
+				src = &event->memory_exception_data;
+				size = sizeof(struct kfd_hsa_memory_exception_data);
+			} else if (event->type == KFD_EVENT_TYPE_HW_EXCEPTION) {
+				dst = &data[i].memory_exception_data;
+				src = &event->hw_exception_data;
+				size = sizeof(struct kfd_hsa_hw_exception_data);
+			} else if (event->type == KFD_EVENT_TYPE_SIGNAL &&
+				waiter->event_age_enabled) {
+				dst = &data[i].signal_event_data.last_event_age;
+				src = &event->event_age;
+				size = sizeof(u64);
+			}
+			if (size && copy_to_user(dst, src, size))
 				return -EFAULT;
 		}
 	}
 
 	return 0;
-
 }
 
-
-
 static long user_timeout_to_jiffies(uint32_t user_timeout_ms)
 {
 	if (user_timeout_ms == KFD_EVENT_TIMEOUT_IMMEDIATE)
@@ -649,21 +917,28 @@ static long user_timeout_to_jiffies(uint32_t user_timeout_ms)
 	return msecs_to_jiffies(user_timeout_ms) + 1;
 }
 
-static void free_waiters(uint32_t num_events, struct kfd_event_waiter *waiters)
+static void free_waiters(uint32_t num_events, struct kfd_event_waiter *waiters,
+			 bool undo_auto_reset)
 {
 	uint32_t i;
 
 	for (i = 0; i < num_events; i++)
-		if (waiters[i].event)
+		if (waiters[i].event) {
+			spin_lock(&waiters[i].event->lock);
 			remove_wait_queue(&waiters[i].event->wq,
 					  &waiters[i].wait);
+			if (undo_auto_reset && waiters[i].activated &&
+			    waiters[i].event && waiters[i].event->auto_reset)
+				set_event(waiters[i].event);
+			spin_unlock(&waiters[i].event->lock);
+		}
 
 	kfree(waiters);
 }
 
 int kfd_wait_on_events(struct kfd_process *p,
 		       uint32_t num_events, void __user *data,
-		       bool all, uint32_t user_timeout_ms,
+		       bool all, uint32_t *user_timeout_ms,
 		       uint32_t *wait_result)
 {
 	struct kfd_event_data __user *events =
@@ -672,7 +947,7 @@ int kfd_wait_on_events(struct kfd_process *p,
 	int ret = 0;
 
 	struct kfd_event_waiter *event_waiters = NULL;
-	long timeout = user_timeout_to_jiffies(user_timeout_ms);
+	long timeout = user_timeout_to_jiffies(*user_timeout_ms);
 
 	event_waiters = alloc_event_waiters(num_events);
 	if (!event_waiters) {
@@ -680,6 +955,9 @@ int kfd_wait_on_events(struct kfd_process *p,
 		goto out;
 	}
 
+	/* Use p->event_mutex here to protect against concurrent creation and
+	 * destruction of events while we initialize event_waiters.
+	 */
 	mutex_lock(&p->event_mutex);
 
 	for (i = 0; i < num_events; i++) {
@@ -691,8 +969,7 @@ int kfd_wait_on_events(struct kfd_process *p,
 			goto out_unlock;
 		}
 
-		ret = init_event_waiter_get_status(p, &event_waiters[i],
-				event_data.event_id);
+		ret = init_event_waiter(p, &event_waiters[i], &event_data);
 		if (ret)
 			goto out_unlock;
 	}
@@ -710,10 +987,6 @@ int kfd_wait_on_events(struct kfd_process *p,
 		goto out_unlock;
 	}
 
-	/* Add to wait lists if we need to wait. */
-	for (i = 0; i < num_events; i++)
-		init_event_waiter_add_to_waitlist(&event_waiters[i]);
-
 	mutex_unlock(&p->event_mutex);
 
 	while (true) {
@@ -723,15 +996,11 @@ int kfd_wait_on_events(struct kfd_process *p,
 		}
 
 		if (signal_pending(current)) {
-			/*
-			 * This is wrong when a nonzero, non-infinite timeout
-			 * is specified. We need to use
-			 * ERESTARTSYS_RESTARTBLOCK, but struct restart_block
-			 * contains a union with data for each user and it's
-			 * in generic kernel code that I don't want to
-			 * touch yet.
-			 */
 			ret = -ERESTARTSYS;
+			if (*user_timeout_ms != KFD_EVENT_TIMEOUT_IMMEDIATE &&
+			    *user_timeout_ms != KFD_EVENT_TIMEOUT_INFINITE)
+				*user_timeout_ms = jiffies_to_msecs(
+					max(0l, timeout-1));
 			break;
 		}
 
@@ -758,16 +1027,21 @@ int kfd_wait_on_events(struct kfd_process *p,
 	}
 	__set_current_state(TASK_RUNNING);
 
+	mutex_lock(&p->event_mutex);
 	/* copy_signaled_event_data may sleep. So this has to happen
 	 * after the task state is set back to RUNNING.
+	 *
+	 * The event may also have been destroyed after signaling. So
+	 * copy_signaled_event_data also must confirm that the event
+	 * still exists. Therefore this must be under the p->event_mutex
+	 * which is also held when events are destroyed.
 	 */
 	if (!ret && *wait_result == KFD_IOC_WAIT_RESULT_COMPLETE)
 		ret = copy_signaled_event_data(num_events,
 					       event_waiters, events);
 
-	mutex_lock(&p->event_mutex);
 out_unlock:
-	free_waiters(num_events, event_waiters);
+	free_waiters(num_events, event_waiters, ret == -ERESTARTSYS);
 	mutex_unlock(&p->event_mutex);
 out:
 	if (ret)
@@ -801,8 +1075,8 @@ int kfd_event_mmap(struct kfd_process *p, struct vm_area_struct *vma)
 	pfn = __pa(page->kernel_address);
 	pfn >>= PAGE_SHIFT;
 
-	vma->vm_flags |= VM_IO | VM_DONTCOPY | VM_DONTEXPAND | VM_NORESERVE
-		       | VM_DONTDUMP | VM_PFNMAP;
+	vm_flags_set(vma, VM_IO | VM_DONTCOPY | VM_DONTEXPAND | VM_NORESERVE
+		       | VM_DONTDUMP | VM_PFNMAP);
 
 	pr_debug("Mapping signal page\n");
 	pr_debug("     start user address  == 0x%08lx\n", vma->vm_start);
@@ -824,8 +1098,7 @@ int kfd_event_mmap(struct kfd_process *p, struct vm_area_struct *vma)
 }
 
 /*
- * Assumes that p->event_mutex is held and of course
- * that p is not going away (current or locked).
+ * Assumes that p is not going away.
  */
 static void lookup_events_by_type_and_signal(struct kfd_process *p,
 		int type, void *event_data)
@@ -837,6 +1110,8 @@ static void lookup_events_by_type_and_signal(struct kfd_process *p,
 
 	ev_data = (struct kfd_hsa_memory_exception_data *) event_data;
 
+	rcu_read_lock();
+
 	id = KFD_FIRST_NONSIGNAL_EVENT_ID;
 	idr_for_each_entry_continue(&p->event_idr, ev, id)
 		if (ev->type == type) {
@@ -844,15 +1119,17 @@ static void lookup_events_by_type_and_signal(struct kfd_process *p,
 			dev_dbg(kfd_device,
 					"Event found: id %X type %d",
 					ev->event_id, ev->type);
+			spin_lock(&ev->lock);
 			set_event(ev);
 			if (ev->type == KFD_EVENT_TYPE_MEMORY && ev_data)
 				ev->memory_exception_data = *ev_data;
+			spin_unlock(&ev->lock);
 		}
 
 	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);
 	}
 
@@ -860,95 +1137,18 @@ 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);
 		}
 	}
-}
-
-#ifdef KFD_SUPPORT_IOMMU_V2
-void kfd_signal_iommu_event(struct kfd_dev *dev, u32 pasid,
-		unsigned long address, bool is_write_requested,
-		bool is_execute_requested)
-{
-	struct kfd_hsa_memory_exception_data memory_exception_data;
-	struct vm_area_struct *vma;
-
-	/*
-	 * Because we are called from arbitrary context (workqueue) as opposed
-	 * 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 mm_struct *mm;
-
-	if (!p)
-		return; /* Presumably process exited. */
-
-	/* Take a safe reference to the mm_struct, which may otherwise
-	 * disappear even while the kfd_process is still referenced.
-	 */
-	mm = get_task_mm(p->lead_thread);
-	if (!mm) {
-		kfd_unref_process(p);
-		return; /* Process is exiting */
-	}
-
-	memset(&memory_exception_data, 0, sizeof(memory_exception_data));
-
-	mmap_read_lock(mm);
-	vma = find_vma(mm, address);
-
-	memory_exception_data.gpu_id = dev->id;
-	memory_exception_data.va = address;
-	/* Set failure reason */
-	memory_exception_data.failure.NotPresent = 1;
-	memory_exception_data.failure.NoExecute = 0;
-	memory_exception_data.failure.ReadOnly = 0;
-	if (vma && address >= vma->vm_start) {
-		memory_exception_data.failure.NotPresent = 0;
-
-		if (is_write_requested && !(vma->vm_flags & VM_WRITE))
-			memory_exception_data.failure.ReadOnly = 1;
-		else
-			memory_exception_data.failure.ReadOnly = 0;
-
-		if (is_execute_requested && !(vma->vm_flags & VM_EXEC))
-			memory_exception_data.failure.NoExecute = 1;
-		else
-			memory_exception_data.failure.NoExecute = 0;
-	}
-
-	mmap_read_unlock(mm);
-	mmput(mm);
 
-	pr_debug("notpresent %d, noexecute %d, readonly %d\n",
-			memory_exception_data.failure.NotPresent,
-			memory_exception_data.failure.NoExecute,
-			memory_exception_data.failure.ReadOnly);
-
-	/* Workaround on Raven to not kill the process when memory is freed
-	 * before IOMMU is able to finish processing all the excessive PPRs
-	 */
-	if (dev->device_info->asic_family != CHIP_RAVEN &&
-	    dev->device_info->asic_family != CHIP_RENOIR) {
-		mutex_lock(&p->event_mutex);
-
-		/* Lookup events by type and signal them */
-		lookup_events_by_type_and_signal(p, KFD_EVENT_TYPE_MEMORY,
-				&memory_exception_data);
-
-		mutex_unlock(&p->event_mutex);
-	}
-
-	kfd_unref_process(p);
+	rcu_read_unlock();
 }
-#endif /* KFD_SUPPORT_IOMMU_V2 */
 
 void kfd_signal_hw_exception_event(u32 pasid)
 {
@@ -957,58 +1157,87 @@ 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. */
 
-	mutex_lock(&p->event_mutex);
-
-	/* Lookup events by type and signal them */
 	lookup_events_by_type_and_signal(p, KFD_EVENT_TYPE_HW_EXCEPTION, NULL);
-
-	mutex_unlock(&p->event_mutex);
 	kfd_unref_process(p);
 }
 
-void kfd_signal_vm_fault_event(struct kfd_dev *dev, u32 pasid,
-				struct kfd_vm_fault_info *info)
+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. */
-	memset(&memory_exception_data, 0, sizeof(memory_exception_data));
-	memory_exception_data.gpu_id = dev->id;
-	memory_exception_data.failure.imprecise = true;
-	/* Set failure reason */
-	if (info) {
-		memory_exception_data.va = (info->page_addr) << PAGE_SHIFT;
-		memory_exception_data.failure.NotPresent =
-			info->prot_valid ? 1 : 0;
-		memory_exception_data.failure.NoExecute =
-			info->prot_exec ? 1 : 0;
-		memory_exception_data.failure.ReadOnly =
-			info->prot_write ? 1 : 0;
-		memory_exception_data.failure.imprecise = 0;
+	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",
+			  pdd->dev->id);
+		return;
 	}
-	mutex_lock(&p->event_mutex);
+
+	/* SoC15 chips and onwards will pass in data from now on. */
+	if (!data) {
+		memset(&memory_exception_data, 0, sizeof(memory_exception_data));
+		memory_exception_data.gpu_id = user_gpu_id;
+		memory_exception_data.failure.imprecise = true;
+
+		/* Set failure reason */
+		if (info) {
+			memory_exception_data.va = (info->page_addr) <<
+								PAGE_SHIFT;
+			memory_exception_data.failure.NotPresent =
+				info->prot_valid ? 1 : 0;
+			memory_exception_data.failure.NoExecute =
+				info->prot_exec ? 1 : 0;
+			memory_exception_data.failure.ReadOnly =
+				info->prot_write ? 1 : 0;
+			memory_exception_data.failure.imprecise = 0;
+		}
+	}
+
+	rcu_read_lock();
 
 	id = KFD_FIRST_NONSIGNAL_EVENT_ID;
 	idr_for_each_entry_continue(&p->event_idr, ev, id)
 		if (ev->type == KFD_EVENT_TYPE_MEMORY) {
-			ev->memory_exception_data = memory_exception_data;
+			spin_lock(&ev->lock);
+			ev->memory_exception_data = data ? *data :
+							memory_exception_data;
 			set_event(ev);
+			spin_unlock(&ev->lock);
 		}
 
-	mutex_unlock(&p->event_mutex);
-	kfd_unref_process(p);
+	rcu_read_unlock();
 }
 
-void kfd_signal_reset_event(struct kfd_dev *dev)
+void kfd_signal_reset_event(struct kfd_node *dev)
 {
 	struct kfd_hsa_hw_exception_data hw_exception_data;
 	struct kfd_hsa_memory_exception_data memory_exception_data;
@@ -1022,69 +1251,129 @@ void kfd_signal_reset_event(struct kfd_dev *dev)
 
 	/* Whole gpu reset caused by GPU hang and memory is lost */
 	memset(&hw_exception_data, 0, sizeof(hw_exception_data));
-	hw_exception_data.gpu_id = dev->id;
 	hw_exception_data.memory_lost = 1;
 	hw_exception_data.reset_cause = reset_cause;
 
 	memset(&memory_exception_data, 0, sizeof(memory_exception_data));
 	memory_exception_data.ErrorType = KFD_MEM_ERR_SRAM_ECC;
-	memory_exception_data.gpu_id = dev->id;
 	memory_exception_data.failure.imprecise = true;
 
 	idx = srcu_read_lock(&kfd_processes_srcu);
 	hash_for_each_rcu(kfd_processes_table, temp, p, kfd_processes) {
-		mutex_lock(&p->event_mutex);
+		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;
 		idr_for_each_entry_continue(&p->event_idr, ev, id) {
 			if (ev->type == KFD_EVENT_TYPE_HW_EXCEPTION) {
+				spin_lock(&ev->lock);
 				ev->hw_exception_data = hw_exception_data;
+				ev->hw_exception_data.gpu_id = user_gpu_id;
 				set_event(ev);
+				spin_unlock(&ev->lock);
 			}
 			if (ev->type == KFD_EVENT_TYPE_MEMORY &&
 			    reset_cause == KFD_HW_EXCEPTION_ECC) {
+				spin_lock(&ev->lock);
 				ev->memory_exception_data = memory_exception_data;
+				ev->memory_exception_data.gpu_id = user_gpu_id;
 				set_event(ev);
+				spin_unlock(&ev->lock);
 			}
 		}
-		mutex_unlock(&p->event_mutex);
+
+		rcu_read_unlock();
 	}
 	srcu_read_unlock(&kfd_processes_srcu, idx);
 }
 
-void kfd_signal_poison_consumed_event(struct kfd_dev *dev, u32 pasid)
+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;
 	uint32_t id = KFD_FIRST_NONSIGNAL_EVENT_ID;
+	int user_gpu_id;
 
-	if (!p)
+	if (!p) {
+		dev_warn(dev->adev->dev, "Not find process with pasid:%d\n", pasid);
 		return; /* Presumably process exited. */
+	}
+
+	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;
+	}
 
 	memset(&hw_exception_data, 0, sizeof(hw_exception_data));
-	hw_exception_data.gpu_id = dev->id;
+	hw_exception_data.gpu_id = user_gpu_id;
 	hw_exception_data.memory_lost = 1;
 	hw_exception_data.reset_cause = KFD_HW_EXCEPTION_ECC;
 
 	memset(&memory_exception_data, 0, sizeof(memory_exception_data));
 	memory_exception_data.ErrorType = KFD_MEM_ERR_POISON_CONSUMED;
-	memory_exception_data.gpu_id = dev->id;
+	memory_exception_data.gpu_id = user_gpu_id;
 	memory_exception_data.failure.imprecise = true;
 
-	mutex_lock(&p->event_mutex);
+	rcu_read_lock();
+
 	idr_for_each_entry_continue(&p->event_idr, ev, id) {
 		if (ev->type == KFD_EVENT_TYPE_HW_EXCEPTION) {
+			spin_lock(&ev->lock);
 			ev->hw_exception_data = hw_exception_data;
 			set_event(ev);
+			spin_unlock(&ev->lock);
 		}
 
 		if (ev->type == KFD_EVENT_TYPE_MEMORY) {
+			spin_lock(&ev->lock);
 			ev->memory_exception_data = memory_exception_data;
 			set_event(ev);
+			spin_unlock(&ev->lock);
 		}
 	}
-	mutex_unlock(&p->event_mutex);
+
+	dev_warn(dev->adev->dev, "Send SIGBUS to process %s(pasid:%d)\n",
+		p->lead_thread->comm, pasid);
+	rcu_read_unlock();
 
 	/* user application will handle SIGBUS signal */
 	send_sig(SIGBUS, p->lead_thread, 0);