[PATCH v3 0/6] tracing/user_events: Add auto cleanup and a flag to persist events

[PATCH v3 0/6] tracing/user_events: Add auto cleanup and a flag to persist events

[Beau Belgrave]

As part of the discussions for user_events aligning to be used with eBPF
it became clear [1] we needed a way to delete events without having to rely
upon the delete IOCTL. Steven suggested that we simply have an owner
for the event, however, the event can be held by more than just the
first register FD, such as perf/ftrace or additional registers. In order
to handle all those cases, we must only delete after all references are
gone from both user and kernel space.

This series adds a internal register flag, USER_EVENT_REG_PERSIST, which
causes the event to not delete itself upon the last put reference. We
cannot fully drop the delete IOCTL, since we still want to enable events to
be registered early via dynamic_events and persist. Events that do not use
this new flag are auto-cleaned up upon no longer being used. This new flag
is not exposed to user processes yet while we work out the details, however,
the code is updated to accomodate it once it is ready.

1: https://lore.kernel.org/linux-trace-kernel/20230518093600.3f119d68@rorschach.local.home/

Change history

v3:
  Rebased to trace/urgent as requested by Steven

  Removed user-facing references to USER_EVENT_REG_PERSIST while we work out
  the details

  Added check for any flags, even from dynamic_events, will return -EINVAL

  Added clarifying comment about event_mutex protection of refcounts

  Updated selftests to work given the absence of persist events

  Updated documentation to not include /sys/kernel/tracing/dynamic_events since
  they cannot be used until persist flags exist and a note about auto-deletions

v2:
  Renamed series to "Add auto cleanup and a flag to persist events"

  Changed auto-delete to be default behavior, with new flag to persist events

Beau Belgrave (6):
  tracing/user_events: Store register flags on events
  tracing/user_events: Track refcount consistently via put/get
  tracing/user_events: Add auto cleanup and future persist flag
  selftests/user_events: Ensure auto cleanup works as expected
  selftests/user_events: Adapt dyn_test to non-persist events
  tracing/user_events: Document auto-cleanup and remove dyn_event refs

 Documentation/trace/user_events.rst           |   7 +-
 kernel/trace/trace_events_user.c              | 208 ++++++++++++++----
 .../testing/selftests/user_events/dyn_test.c  | 177 +++++++++++----
 .../selftests/user_events/ftrace_test.c       |  49 ++++-
 4 files changed, 346 insertions(+), 95 deletions(-)


base-commit: a8c5c6a5c87ea10189ec43483f44800a4ec97785
-- 
2.25.1

[PATCH v3 1/6] tracing/user_events: Store register flags on events

[Beau Belgrave]

Currently we don't have any available flags for user processes to use to
indicate options for user_events. We will soon have a flag to indicate
the event should or should not auto-delete once it's not being used by
anyone.

Add a reg_flags field to user_events and parameters to existing
functions to allow for this in future patches.

Signed-off-by: Beau Belgrave <beaub@linux.microsoft.com>
---
 kernel/trace/trace_events_user.c | 16 ++++++++++------
 1 file changed, 10 insertions(+), 6 deletions(-)

diff --git a/kernel/trace/trace_events_user.c b/kernel/trace/trace_events_user.c
index cf6d4c02c363..629823e21447 100644
--- a/kernel/trace/trace_events_user.c
+++ b/kernel/trace/trace_events_user.c
@@ -87,6 +87,7 @@ struct user_event {
 	struct list_head		validators;
 	refcount_t			refcnt;
 	int				min_size;
+	int				reg_flags;
 	char				status;
 };
 
@@ -165,7 +166,7 @@ typedef void (*user_event_func_t) (struct user_event *user, struct iov_iter *i,
 
 static int user_event_parse(struct user_event_group *group, char *name,
 			    char *args, char *flags,
-			    struct user_event **newuser);
+			    struct user_event **newuser, int reg_flags);
 
 static struct user_event_mm *user_event_mm_get(struct user_event_mm *mm);
 static struct user_event_mm *user_event_mm_get_all(struct user_event *user);
@@ -810,7 +811,8 @@ static struct list_head *user_event_get_fields(struct trace_event_call *call)
  * Upon success user_event has its ref count increased by 1.
  */
 static int user_event_parse_cmd(struct user_event_group *group,
-				char *raw_command, struct user_event **newuser)
+				char *raw_command, struct user_event **newuser,
+				int reg_flags)
 {
 	char *name = raw_command;
 	char *args = strpbrk(name, " ");
@@ -824,7 +826,7 @@ static int user_event_parse_cmd(struct user_event_group *group,
 	if (flags)
 		*flags++ = '\0';
 
-	return user_event_parse(group, name, args, flags, newuser);
+	return user_event_parse(group, name, args, flags, newuser, reg_flags);
 }
 
 static int user_field_array_size(const char *type)
@@ -1588,7 +1590,7 @@ static int user_event_create(const char *raw_command)
 
 	mutex_lock(&group->reg_mutex);
 
-	ret = user_event_parse_cmd(group, name, &user);
+	ret = user_event_parse_cmd(group, name, &user, 0);
 
 	if (!ret)
 		refcount_dec(&user->refcnt);
@@ -1751,7 +1753,7 @@ static int user_event_trace_register(struct user_event *user)
  */
 static int user_event_parse(struct user_event_group *group, char *name,
 			    char *args, char *flags,
-			    struct user_event **newuser)
+			    struct user_event **newuser, int reg_flags)
 {
 	int ret;
 	u32 key;
@@ -1846,6 +1848,8 @@ static int user_event_parse(struct user_event_group *group, char *name,
 	if (ret)
 		goto put_user_lock;
 
+	user->reg_flags = reg_flags;
+
 	/* Ensure we track self ref and caller ref (2) */
 	refcount_set(&user->refcnt, 2);
 
@@ -2144,7 +2148,7 @@ static long user_events_ioctl_reg(struct user_event_file_info *info,
 		return ret;
 	}
 
-	ret = user_event_parse_cmd(info->group, name, &user);
+	ret = user_event_parse_cmd(info->group, name, &user, reg.flags);
 
 	if (ret) {
 		kfree(name);
-- 
2.25.1

[PATCH v3 2/6] tracing/user_events: Track refcount consistently via put/get

[Beau Belgrave]

Various parts of the code today track user_event's refcnt field directly
via a refcount_add/dec. This makes it hard to modify the behavior of the
last reference decrement in all code paths consistently. For example, in
the future we will auto-delete events upon the last reference going
away. This last reference could happen in many places, but we want it to
be consistently handled.

Add user_event_get() and user_event_put() for the add/dec. Update all
places where direct refcounts are being used to utilize these new
functions. In each location pass if event_mutex is locked or not. This
allows us to drop events automatically in future patches clearly. Ensure
when caller states the lock is held, it really is (or is not) held.

Signed-off-by: Beau Belgrave <beaub@linux.microsoft.com>
---
 kernel/trace/trace_events_user.c | 69 +++++++++++++++++++-------------
 1 file changed, 41 insertions(+), 28 deletions(-)

diff --git a/kernel/trace/trace_events_user.c b/kernel/trace/trace_events_user.c
index 629823e21447..c064458eea5c 100644
--- a/kernel/trace/trace_events_user.c
+++ b/kernel/trace/trace_events_user.c
@@ -177,6 +177,28 @@ static u32 user_event_key(char *name)
 	return jhash(name, strlen(name), 0);
 }
 
+static struct user_event *user_event_get(struct user_event *user)
+{
+	refcount_inc(&user->refcnt);
+
+	return user;
+}
+
+static void user_event_put(struct user_event *user, bool locked)
+{
+#ifdef CONFIG_LOCKDEP
+	if (locked)
+		lockdep_assert_held(&event_mutex);
+	else
+		lockdep_assert_not_held(&event_mutex);
+#endif
+
+	if (unlikely(!user))
+		return;
+
+	refcount_dec(&user->refcnt);
+}
+
 static void user_event_group_destroy(struct user_event_group *group)
 {
 	kfree(group->system_name);
@@ -228,12 +250,13 @@ static struct user_event_group *user_event_group_create(void)
 	return NULL;
 };
 
-static void user_event_enabler_destroy(struct user_event_enabler *enabler)
+static void user_event_enabler_destroy(struct user_event_enabler *enabler,
+				       bool locked)
 {
 	list_del_rcu(&enabler->mm_enablers_link);
 
 	/* No longer tracking the event via the enabler */
-	refcount_dec(&enabler->event->refcnt);
+	user_event_put(enabler->event, locked);
 
 	kfree(enabler);
 }
@@ -295,7 +318,7 @@ static void user_event_enabler_fault_fixup(struct work_struct *work)
 
 	/* User asked for enabler to be removed during fault */
 	if (test_bit(ENABLE_VAL_FREEING_BIT, ENABLE_BITOPS(enabler))) {
-		user_event_enabler_destroy(enabler);
+		user_event_enabler_destroy(enabler, true);
 		goto out;
 	}
 
@@ -470,14 +493,12 @@ static bool user_event_enabler_dup(struct user_event_enabler *orig,
 	if (!enabler)
 		return false;
 
-	enabler->event = orig->event;
+	enabler->event = user_event_get(orig->event);
 	enabler->addr = orig->addr;
 
 	/* Only dup part of value (ignore future flags, etc) */
 	enabler->values = orig->values & ENABLE_VAL_DUP_MASK;
 
-	refcount_inc(&enabler->event->refcnt);
-
 	/* Enablers not exposed yet, RCU not required */
 	list_add(&enabler->mm_enablers_link, &mm->enablers);
 
@@ -594,7 +615,7 @@ static void user_event_mm_destroy(struct user_event_mm *mm)
 	struct user_event_enabler *enabler, *next;
 
 	list_for_each_entry_safe(enabler, next, &mm->enablers, mm_enablers_link)
-		user_event_enabler_destroy(enabler);
+		user_event_enabler_destroy(enabler, false);
 
 	mmdrop(mm->mm);
 	kfree(mm);
@@ -749,7 +770,7 @@ static struct user_event_enabler
 	 * exit or run exec(), which includes forks and clones.
 	 */
 	if (!*write_result) {
-		refcount_inc(&enabler->event->refcnt);
+		user_event_get(user);
 		list_add_rcu(&enabler->mm_enablers_link, &user_mm->enablers);
 	}
 
@@ -1337,10 +1358,8 @@ static struct user_event *find_user_event(struct user_event_group *group,
 	*outkey = key;
 
 	hash_for_each_possible(group->register_table, user, node, key)
-		if (!strcmp(EVENT_NAME(user), name)) {
-			refcount_inc(&user->refcnt);
-			return user;
-		}
+		if (!strcmp(EVENT_NAME(user), name))
+			return user_event_get(user);
 
 	return NULL;
 }
@@ -1554,12 +1573,12 @@ static int user_event_reg(struct trace_event_call *call,
 
 	return ret;
 inc:
-	refcount_inc(&user->refcnt);
+	user_event_get(user);
 	update_enable_bit_for(user);
 	return 0;
 dec:
 	update_enable_bit_for(user);
-	refcount_dec(&user->refcnt);
+	user_event_put(user, true);
 	return 0;
 }
 
@@ -1593,7 +1612,7 @@ static int user_event_create(const char *raw_command)
 	ret = user_event_parse_cmd(group, name, &user, 0);
 
 	if (!ret)
-		refcount_dec(&user->refcnt);
+		user_event_put(user, false);
 
 	mutex_unlock(&group->reg_mutex);
 
@@ -1794,7 +1813,7 @@ static int user_event_parse(struct user_event_group *group, char *name,
 
 		return 0;
 error:
-		refcount_dec(&user->refcnt);
+		user_event_put(user, false);
 		return ret;
 	}
 
@@ -1883,7 +1902,7 @@ static int delete_user_event(struct user_event_group *group, char *name)
 	if (!user)
 		return -ENOENT;
 
-	refcount_dec(&user->refcnt);
+	user_event_put(user, true);
 
 	if (!user_event_last_ref(user))
 		return -EBUSY;
@@ -2042,9 +2061,7 @@ static int user_events_ref_add(struct user_event_file_info *info,
 	for (i = 0; i < count; ++i)
 		new_refs->events[i] = refs->events[i];
 
-	new_refs->events[i] = user;
-
-	refcount_inc(&user->refcnt);
+	new_refs->events[i] = user_event_get(user);
 
 	rcu_assign_pointer(info->refs, new_refs);
 
@@ -2158,7 +2175,7 @@ static long user_events_ioctl_reg(struct user_event_file_info *info,
 	ret = user_events_ref_add(info, user);
 
 	/* No longer need parse ref, ref_add either worked or not */
-	refcount_dec(&user->refcnt);
+	user_event_put(user, false);
 
 	/* Positive number is index and valid */
 	if (ret < 0)
@@ -2307,7 +2324,7 @@ static long user_events_ioctl_unreg(unsigned long uarg)
 			set_bit(ENABLE_VAL_FREEING_BIT, ENABLE_BITOPS(enabler));
 
 			if (!test_bit(ENABLE_VAL_FAULTING_BIT, ENABLE_BITOPS(enabler)))
-				user_event_enabler_destroy(enabler);
+				user_event_enabler_destroy(enabler, true);
 
 			/* Removed at least one */
 			ret = 0;
@@ -2365,7 +2382,6 @@ static int user_events_release(struct inode *node, struct file *file)
 	struct user_event_file_info *info = file->private_data;
 	struct user_event_group *group;
 	struct user_event_refs *refs;
-	struct user_event *user;
 	int i;
 
 	if (!info)
@@ -2389,12 +2405,9 @@ static int user_events_release(struct inode *node, struct file *file)
 	 * The underlying user_events are ref counted, and cannot be freed.
 	 * After this decrement, the user_events may be freed elsewhere.
 	 */
-	for (i = 0; i < refs->count; ++i) {
-		user = refs->events[i];
+	for (i = 0; i < refs->count; ++i)
+		user_event_put(refs->events[i], false);
 
-		if (user)
-			refcount_dec(&user->refcnt);
-	}
 out:
 	file->private_data = NULL;
 
-- 
2.25.1

[PATCH v3 3/6] tracing/user_events: Add auto cleanup and future persist flag

[Beau Belgrave]

Currently user events need to be manually deleted via the delete IOCTL
call or via the dynamic_events file. Most operators and processes wish
to have these events auto cleanup when they are no longer used by
anything to prevent them piling without manual maintenance. However,
some operators may not want this, such as pre-registering events via the
dynamic_events tracefs file.

Update user_event_put() to attempt an auto delete of the event if it's
the last reference. The auto delete must run in a work queue to ensure
proper behavior of class->reg() invocations that don't expect the call
to go away from underneath them during the unregister. Add work_struct
to user_event struct to ensure we can do this reliably.

Add a persist flag, that is not yet exposed, to ensure we can toggle
between auto-cleanup and leaving the events existing in the future. When
a non-zero flag is seen during register, return -EINVAL to ensure ABI
is clear for the user processes while we work out the best approach for
persistent events.

Link: https://lore.kernel.org/linux-trace-kernel/20230518093600.3f119d68@rorschach.local.home/

Suggested-by: Steven Rostedt <rostedt@goodmis.org>
Signed-off-by: Beau Belgrave <beaub@linux.microsoft.com>
---
 kernel/trace/trace_events_user.c | 139 ++++++++++++++++++++++++++++---
 1 file changed, 126 insertions(+), 13 deletions(-)

diff --git a/kernel/trace/trace_events_user.c b/kernel/trace/trace_events_user.c
index c064458eea5c..8df0550415e7 100644
--- a/kernel/trace/trace_events_user.c
+++ b/kernel/trace/trace_events_user.c
@@ -49,6 +49,18 @@
 #define EVENT_STATUS_PERF BIT(1)
 #define EVENT_STATUS_OTHER BIT(7)
 
+/*
+ * User register flags are not allowed yet, keep them here until we are
+ * ready to expose them out to the user ABI.
+ */
+enum user_reg_flag {
+	/* Event will not delete upon last reference closing */
+	USER_EVENT_REG_PERSIST		= 1U << 0,
+
+	/* This value or above is currently non-ABI */
+	USER_EVENT_REG_MAX		= 1U << 1,
+};
+
 /*
  * Stores the system name, tables, and locks for a group of events. This
  * allows isolation for events by various means.
@@ -85,6 +97,7 @@ struct user_event {
 	struct hlist_node		node;
 	struct list_head		fields;
 	struct list_head		validators;
+	struct work_struct		put_work;
 	refcount_t			refcnt;
 	int				min_size;
 	int				reg_flags;
@@ -171,6 +184,7 @@ static int user_event_parse(struct user_event_group *group, char *name,
 static struct user_event_mm *user_event_mm_get(struct user_event_mm *mm);
 static struct user_event_mm *user_event_mm_get_all(struct user_event *user);
 static void user_event_mm_put(struct user_event_mm *mm);
+static int destroy_user_event(struct user_event *user);
 
 static u32 user_event_key(char *name)
 {
@@ -184,19 +198,103 @@ static struct user_event *user_event_get(struct user_event *user)
 	return user;
 }
 
+static void delayed_destroy_user_event(struct work_struct *work)
+{
+	struct user_event *user = container_of(
+		work, struct user_event, put_work);
+
+	mutex_lock(&event_mutex);
+
+	if (!refcount_dec_and_test(&user->refcnt))
+		goto out;
+
+	if (destroy_user_event(user)) {
+		/*
+		 * The only reason this would fail here is if we cannot
+		 * update the visibility of the event. In this case the
+		 * event stays in the hashtable, waiting for someone to
+		 * attempt to delete it later.
+		 */
+		pr_warn("user_events: Unable to delete event\n");
+		refcount_set(&user->refcnt, 1);
+	}
+out:
+	mutex_unlock(&event_mutex);
+}
+
 static void user_event_put(struct user_event *user, bool locked)
 {
-#ifdef CONFIG_LOCKDEP
-	if (locked)
-		lockdep_assert_held(&event_mutex);
-	else
-		lockdep_assert_not_held(&event_mutex);
-#endif
+	bool delete;
 
 	if (unlikely(!user))
 		return;
 
-	refcount_dec(&user->refcnt);
+	/*
+	 * When the event is not enabled for auto-delete there will always
+	 * be at least 1 reference to the event. During the event creation
+	 * we initially set the refcnt to 2 to achieve this. In those cases
+	 * the caller must acquire event_mutex and after decrement check if
+	 * the refcnt is 1, meaning this is the last reference. When auto
+	 * delete is enabled, there will only be 1 ref, IE: refcnt will be
+	 * only set to 1 during creation to allow the below checks to go
+	 * through upon the last put. The last put must always be done with
+	 * the event mutex held.
+	 */
+	if (!locked) {
+		lockdep_assert_not_held(&event_mutex);
+		delete = refcount_dec_and_mutex_lock(&user->refcnt, &event_mutex);
+	} else {
+		lockdep_assert_held(&event_mutex);
+		delete = refcount_dec_and_test(&user->refcnt);
+	}
+
+	if (!delete)
+		return;
+
+	/*
+	 * We now have the event_mutex in all cases, which ensures that
+	 * no new references will be taken until event_mutex is released.
+	 * New references come through find_user_event(), which requires
+	 * the event_mutex to be held.
+	 */
+
+	if (user->reg_flags & USER_EVENT_REG_PERSIST) {
+		/* We should not get here when persist flag is set */
+		pr_alert("BUG: Auto-delete engaged on persistent event\n");
+		goto out;
+	}
+
+	/*
+	 * Unfortunately we have to attempt the actual destroy in a work
+	 * queue. This is because not all cases handle a trace_event_call
+	 * being removed within the class->reg() operation for unregister.
+	 */
+	INIT_WORK(&user->put_work, delayed_destroy_user_event);
+
+	/*
+	 * Since the event is still in the hashtable, we have to re-inc
+	 * the ref count to 1. This count will be decremented and checked
+	 * in the work queue to ensure it's still the last ref. This is
+	 * needed because a user-process could register the same event in
+	 * between the time of event_mutex release and the work queue
+	 * running the delayed destroy. If we removed the item now from
+	 * the hashtable, this would result in a timing window where a
+	 * user process would fail a register because the trace_event_call
+	 * register would fail in the tracing layers.
+	 */
+	refcount_set(&user->refcnt, 1);
+
+	if (WARN_ON_ONCE(!schedule_work(&user->put_work))) {
+		/*
+		 * If we fail we must wait for an admin to attempt delete or
+		 * another register/close of the event, whichever is first.
+		 */
+		pr_warn("user_events: Unable to queue delayed destroy\n");
+	}
+out:
+	/* Ensure if we didn't have event_mutex before we unlock it */
+	if (!locked)
+		mutex_unlock(&event_mutex);
 }
 
 static void user_event_group_destroy(struct user_event_group *group)
@@ -793,7 +891,12 @@ static struct user_event_enabler
 static __always_inline __must_check
 bool user_event_last_ref(struct user_event *user)
 {
-	return refcount_read(&user->refcnt) == 1;
+	int last = 0;
+
+	if (user->reg_flags & USER_EVENT_REG_PERSIST)
+		last = 1;
+
+	return refcount_read(&user->refcnt) == last;
 }
 
 static __always_inline __must_check
@@ -1609,7 +1712,8 @@ static int user_event_create(const char *raw_command)
 
 	mutex_lock(&group->reg_mutex);
 
-	ret = user_event_parse_cmd(group, name, &user, 0);
+	/* Dyn events persist, otherwise they would cleanup immediately */
+	ret = user_event_parse_cmd(group, name, &user, USER_EVENT_REG_PERSIST);
 
 	if (!ret)
 		user_event_put(user, false);
@@ -1780,6 +1884,10 @@ static int user_event_parse(struct user_event_group *group, char *name,
 	int argc = 0;
 	char **argv;
 
+	/* User register flags are not ready yet */
+	if (reg_flags != 0 || flags != NULL)
+		return -EINVAL;
+
 	/* Prevent dyn_event from racing */
 	mutex_lock(&event_mutex);
 	user = find_user_event(group, name, &key);
@@ -1869,8 +1977,13 @@ static int user_event_parse(struct user_event_group *group, char *name,
 
 	user->reg_flags = reg_flags;
 
-	/* Ensure we track self ref and caller ref (2) */
-	refcount_set(&user->refcnt, 2);
+	if (user->reg_flags & USER_EVENT_REG_PERSIST) {
+		/* Ensure we track self ref and caller ref (2) */
+		refcount_set(&user->refcnt, 2);
+	} else {
+		/* Ensure we track only caller ref (1) */
+		refcount_set(&user->refcnt, 1);
+	}
 
 	dyn_event_init(&user->devent, &user_event_dops);
 	dyn_event_add(&user->devent, &user->call);
@@ -2092,8 +2205,8 @@ static long user_reg_get(struct user_reg __user *ureg, struct user_reg *kreg)
 	if (ret)
 		return ret;
 
-	/* Ensure no flags, since we don't support any yet */
-	if (kreg->flags != 0)
+	/* Ensure only valid flags */
+	if (kreg->flags & ~(USER_EVENT_REG_MAX-1))
 		return -EINVAL;
 
 	/* Ensure supported size */
-- 
2.25.1

[PATCH v3 4/6] selftests/user_events: Ensure auto cleanup works as expected

[Beau Belgrave]

User events now auto cleanup upon the last reference put. Update
ftrace_test to ensure this works as expected. Ensure EBUSY delays
while event is being deleted do not cause transient failures by
waiting and re-attempting.

Signed-off-by: Beau Belgrave <beaub@linux.microsoft.com>
---
 .../selftests/user_events/ftrace_test.c       | 49 ++++++++++++++-----
 1 file changed, 38 insertions(+), 11 deletions(-)

diff --git a/tools/testing/selftests/user_events/ftrace_test.c b/tools/testing/selftests/user_events/ftrace_test.c
index abfb49558a26..eb6904d89f14 100644
--- a/tools/testing/selftests/user_events/ftrace_test.c
+++ b/tools/testing/selftests/user_events/ftrace_test.c
@@ -102,30 +102,56 @@ static int get_print_fmt(char *buffer, int len)
 	return -1;
 }
 
+static bool wait_for_delete(void)
+{
+	int i;
+
+	for (i = 0; i < 1000; ++i) {
+		int fd = open(enable_file, O_RDONLY);
+
+		if (fd == -1)
+			return true;
+
+		close(fd);
+		usleep(1000);
+	}
+
+	return false;
+}
+
 static int clear(int *check)
 {
 	struct user_unreg unreg = {0};
+	int fd;
 
 	unreg.size = sizeof(unreg);
 	unreg.disable_bit = 31;
 	unreg.disable_addr = (__u64)check;
 
-	int fd = open(data_file, O_RDWR);
+	fd = open(data_file, O_RDWR);
 
 	if (fd == -1)
 		return -1;
 
 	if (ioctl(fd, DIAG_IOCSUNREG, &unreg) == -1)
 		if (errno != ENOENT)
-			return -1;
-
-	if (ioctl(fd, DIAG_IOCSDEL, "__test_event") == -1)
-		if (errno != ENOENT)
-			return -1;
+			goto fail;
+
+	if (ioctl(fd, DIAG_IOCSDEL, "__test_event") == -1) {
+		if (errno == EBUSY) {
+			if (!wait_for_delete())
+				goto fail;
+		} else if (errno != ENOENT)
+			goto fail;
+	}
 
 	close(fd);
 
 	return 0;
+fail:
+	close(fd);
+
+	return -1;
 }
 
 static int check_print_fmt(const char *event, const char *expected, int *check)
@@ -155,9 +181,8 @@ static int check_print_fmt(const char *event, const char *expected, int *check)
 	/* Register should work */
 	ret = ioctl(fd, DIAG_IOCSREG, &reg);
 
-	close(fd);
-
 	if (ret != 0) {
+		close(fd);
 		printf("Reg failed in fmt\n");
 		return ret;
 	}
@@ -165,6 +190,8 @@ static int check_print_fmt(const char *event, const char *expected, int *check)
 	/* Ensure correct print_fmt */
 	ret = get_print_fmt(print_fmt, sizeof(print_fmt));
 
+	close(fd);
+
 	if (ret != 0)
 		return ret;
 
@@ -256,10 +283,10 @@ TEST_F(user, register_events) {
 	unreg.disable_bit = 30;
 	ASSERT_EQ(0, ioctl(self->data_fd, DIAG_IOCSUNREG, &unreg));
 
-	/* Delete should work only after close and unregister */
+	/* Delete should have been auto-done after close and unregister */
 	close(self->data_fd);
-	self->data_fd = open(data_file, O_RDWR);
-	ASSERT_EQ(0, ioctl(self->data_fd, DIAG_IOCSDEL, "__test_event"));
+
+	ASSERT_EQ(true, wait_for_delete());
 }
 
 TEST_F(user, write_events) {
-- 
2.25.1

[PATCH v3 5/6] selftests/user_events: Adapt dyn_test to non-persist events

[Beau Belgrave]

Now that user_events does not honor persist events the dynamic_events
file cannot be easily used to test parsing and matching cases.

Update dyn_test to use the direct ABI file instead of dynamic_events so
that we still have testing coverage until persist events and
dynamic_events file integration has been decided.

Signed-off-by: Beau Belgrave <beaub@linux.microsoft.com>
---
 .../testing/selftests/user_events/dyn_test.c  | 177 ++++++++++++++----
 1 file changed, 136 insertions(+), 41 deletions(-)

diff --git a/tools/testing/selftests/user_events/dyn_test.c b/tools/testing/selftests/user_events/dyn_test.c
index 8879a7b04c6a..d6979a48478f 100644
--- a/tools/testing/selftests/user_events/dyn_test.c
+++ b/tools/testing/selftests/user_events/dyn_test.c
@@ -16,42 +16,140 @@
 
 #include "../kselftest_harness.h"
 
-const char *dyn_file = "/sys/kernel/tracing/dynamic_events";
-const char *clear = "!u:__test_event";
+const char *abi_file = "/sys/kernel/tracing/user_events_data";
+const char *enable_file = "/sys/kernel/tracing/events/user_events/__test_event/enable";
 
-static int Append(const char *value)
+static bool wait_for_delete(void)
 {
-	int fd = open(dyn_file, O_RDWR | O_APPEND);
-	int ret = write(fd, value, strlen(value));
+	int i;
+
+	for (i = 0; i < 1000; ++i) {
+		int fd = open(enable_file, O_RDONLY);
+
+		if (fd == -1)
+			return true;
+
+		close(fd);
+		usleep(1000);
+	}
+
+	return false;
+}
+
+static int reg_event(int fd, int *check, int bit, const char *value)
+{
+	struct user_reg reg = {0};
+
+	reg.size = sizeof(reg);
+	reg.name_args = (__u64)value;
+	reg.enable_bit = bit;
+	reg.enable_addr = (__u64)check;
+	reg.enable_size = sizeof(*check);
+
+	if (ioctl(fd, DIAG_IOCSREG, &reg) == -1)
+		return -1;
+
+	return 0;
+}
+
+static int unreg_event(int fd, int *check, int bit)
+{
+	struct user_unreg unreg = {0};
+
+	unreg.size = sizeof(unreg);
+	unreg.disable_bit = bit;
+	unreg.disable_addr = (__u64)check;
+
+	return ioctl(fd, DIAG_IOCSUNREG, &unreg);
+}
+
+static int parse(int *check, const char *value)
+{
+	int fd = open(abi_file, O_RDWR);
+	int ret;
+
+	if (fd == -1)
+		return -1;
+
+	/* Until we have persist flags via dynamic events, use the base name */
+	if (value[0] != 'u' || value[1] != ':') {
+		close(fd);
+		return -1;
+	}
+
+	ret = reg_event(fd, check, 31, value + 2);
+
+	if (ret != -1) {
+		if (unreg_event(fd, check, 31) == -1)
+			printf("WARN: Couldn't unreg event\n");
+	}
 
 	close(fd);
+
 	return ret;
 }
 
-#define CLEAR() \
+static int check_match(int *check, const char *first, const char *second, bool *match)
+{
+	int fd = open(abi_file, O_RDWR);
+	int ret = -1;
+
+	if (fd == -1)
+		return -1;
+
+	if (reg_event(fd, check, 31, first) == -1)
+		goto cleanup;
+
+	if (reg_event(fd, check, 30, second) == -1) {
+		if (errno == EADDRINUSE) {
+			/* Name is in use, with different fields */
+			*match = false;
+			ret = 0;
+		}
+
+		goto cleanup;
+	}
+
+	*match = true;
+	ret = 0;
+cleanup:
+	unreg_event(fd, check, 31);
+	unreg_event(fd, check, 30);
+
+	close(fd);
+
+	wait_for_delete();
+
+	return ret;
+}
+
+#define TEST_MATCH(x, y) \
 do { \
-	int ret = Append(clear); \
-	if (ret == -1) \
-		ASSERT_EQ(ENOENT, errno); \
+	bool match; \
+	ASSERT_NE(-1, check_match(&self->check, x, y, &match)); \
+	ASSERT_EQ(true, match); \
 } while (0)
 
-#define TEST_PARSE(x) \
+#define TEST_NMATCH(x, y) \
 do { \
-	ASSERT_NE(-1, Append(x)); \
-	CLEAR(); \
+	bool match; \
+	ASSERT_NE(-1, check_match(&self->check, x, y, &match)); \
+	ASSERT_EQ(false, match); \
 } while (0)
 
-#define TEST_NPARSE(x) ASSERT_EQ(-1, Append(x))
+#define TEST_PARSE(x) ASSERT_NE(-1, parse(&self->check, x))
+
+#define TEST_NPARSE(x) ASSERT_EQ(-1, parse(&self->check, x))
 
 FIXTURE(user) {
+	int check;
 };
 
 FIXTURE_SETUP(user) {
-	CLEAR();
 }
 
 FIXTURE_TEARDOWN(user) {
-	CLEAR();
+	wait_for_delete();
 }
 
 TEST_F(user, basic_types) {
@@ -95,33 +193,30 @@ TEST_F(user, size_types) {
 	TEST_NPARSE("u:__test_event char a 20");
 }
 
-TEST_F(user, flags) {
-	/* Should work */
-	TEST_PARSE("u:__test_event:BPF_ITER u32 a");
-	/* Forward compat */
-	TEST_PARSE("u:__test_event:BPF_ITER,FLAG_FUTURE u32 a");
-}
-
 TEST_F(user, matching) {
-	/* Register */
-	ASSERT_NE(-1, Append("u:__test_event struct custom a 20"));
-	/* Should not match */
-	TEST_NPARSE("!u:__test_event struct custom b");
-	/* Should match */
-	TEST_PARSE("!u:__test_event struct custom a");
-	/* Multi field reg */
-	ASSERT_NE(-1, Append("u:__test_event u32 a; u32 b"));
-	/* Non matching cases */
-	TEST_NPARSE("!u:__test_event u32 a");
-	TEST_NPARSE("!u:__test_event u32 b");
-	TEST_NPARSE("!u:__test_event u32 a; u32 ");
-	TEST_NPARSE("!u:__test_event u32 a; u32 a");
-	/* Matching case */
-	TEST_PARSE("!u:__test_event u32 a; u32 b");
-	/* Register */
-	ASSERT_NE(-1, Append("u:__test_event u32 a; u32 b"));
-	/* Ensure trailing semi-colon case */
-	TEST_PARSE("!u:__test_event u32 a; u32 b;");
+	/* Single name matches */
+	TEST_MATCH("__test_event u32 a",
+		   "__test_event u32 a");
+
+	/* Multiple names match */
+	TEST_MATCH("__test_event u32 a; u32 b",
+		   "__test_event u32 a; u32 b");
+
+	/* Multiple names match with dangling ; */
+	TEST_MATCH("__test_event u32 a; u32 b",
+		   "__test_event u32 a; u32 b;");
+
+	/* Single name doesn't match */
+	TEST_NMATCH("__test_event u32 a",
+		    "__test_event u32 b");
+
+	/* Multiple names don't match */
+	TEST_NMATCH("__test_event u32 a; u32 b",
+		    "__test_event u32 b; u32 a");
+
+	/* Types don't match */
+	TEST_NMATCH("__test_event u64 a; u64 b",
+		    "__test_event u32 a; u32 b");
 }
 
 int main(int argc, char **argv)
-- 
2.25.1

[PATCH v3 6/6] tracing/user_events: Document auto-cleanup and remove dyn_event refs

[Beau Belgrave]

Now user_events auto-cleanup upon the last reference by default. This
makes it not possible to use the dynamics event file via tracefs.

Document that auto-cleanup is enabled by default and remove the refernce
to /sys/kernel/tracing/dynamic_events file to make this clear.

Signed-off-by: Beau Belgrave <beaub@linux.microsoft.com>
---
 Documentation/trace/user_events.rst | 7 +++----
 1 file changed, 3 insertions(+), 4 deletions(-)

diff --git a/Documentation/trace/user_events.rst b/Documentation/trace/user_events.rst
index f79987e16cf4..e7b07313550a 100644
--- a/Documentation/trace/user_events.rst
+++ b/Documentation/trace/user_events.rst
@@ -14,10 +14,6 @@ Programs can view status of the events via
 /sys/kernel/tracing/user_events_status and can both register and write
 data out via /sys/kernel/tracing/user_events_data.
 
-Programs can also use /sys/kernel/tracing/dynamic_events to register and
-delete user based events via the u: prefix. The format of the command to
-dynamic_events is the same as the ioctl with the u: prefix applied.
-
 Typically programs will register a set of events that they wish to expose to
 tools that can read trace_events (such as ftrace and perf). The registration
 process tells the kernel which address and bit to reflect if any tool has
@@ -144,6 +140,9 @@ its name. Delete will only succeed if there are no references left to the
 event (in both user and kernel space). User programs should use a separate file
 to request deletes than the one used for registration due to this.
 
+**NOTE:** By default events will auto-delete when there are no references left
+to the event. Flags in the future may change this logic.
+
 Unregistering
 -------------
 If after registering an event it is no longer wanted to be updated then it can
-- 
2.25.1