[PATCH v2 0/3] tracing: More clean ups of triggers

[PATCH v2 0/3] tracing: More clean ups of triggers

[Steven Rostedt]

This is based on top of:

   https://lore.kernel.org/linux-trace-kernel/20251125200857.975524485@kernel.org


Changes since v1: https://lore.kernel.org/linux-trace-kernel/20251120205600.570673392@kernel.org/

- Moved include of llist.h to trace.h as it is used there (Masami Hiramatsu)


Steven Rostedt (3):
      tracing: Remove unneeded event_mutex lock in event_trigger_regex_release()
      tracing: Add bulk garbage collection of freeing event_trigger_data
      tracing: Use strim() in trigger_process_regex() instead of skip_spaces()

----
 kernel/trace/trace.h                |  2 ++
 kernel/trace/trace_events_trigger.c | 64 +++++++++++++++++++++++++++++++------
 2 files changed, 56 insertions(+), 10 deletions(-)

[PATCH v2 1/3] tracing: Remove unneeded event_mutex lock in event_trigger_regex_release()

[Steven Rostedt]

From: Steven Rostedt <rostedt@goodmis.org>

In event_trigger_regex_release(), the only code is:

	mutex_lock(&event_mutex);
	if (file->f_mode & FMODE_READ)
		seq_release(inode, file);
	mutex_unlock(&event_mutex);

	return 0;

There's nothing special about the file->f_mode or the seq_release() that
requires any locking. Remove the unnecessary locks.

Acked-by: Masami Hiramatsu (Google) <mhiramat@kernel.org>
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
---
 kernel/trace/trace_events_trigger.c | 4 ----
 1 file changed, 4 deletions(-)

diff --git a/kernel/trace/trace_events_trigger.c b/kernel/trace/trace_events_trigger.c
index 7795af600466..e5dcfcbb2cd5 100644
--- a/kernel/trace/trace_events_trigger.c
+++ b/kernel/trace/trace_events_trigger.c
@@ -314,13 +314,9 @@ static ssize_t event_trigger_regex_write(struct file *file,
 
 static int event_trigger_regex_release(struct inode *inode, struct file *file)
 {
-	mutex_lock(&event_mutex);
-
 	if (file->f_mode & FMODE_READ)
 		seq_release(inode, file);
 
-	mutex_unlock(&event_mutex);
-
 	return 0;
 }
 
-- 
2.51.0

Re: [PATCH v2 1/3] tracing: Remove unneeded event_mutex lock in event_trigger_regex_release()

[Tom Zanussi]

On Tue, 2025-11-25 at 16:40 -0500, Steven Rostedt wrote:
> From: Steven Rostedt <rostedt@goodmis.org>
> 
> In event_trigger_regex_release(), the only code is:
> 
> 	mutex_lock(&event_mutex);
> 	if (file->f_mode & FMODE_READ)
> 		seq_release(inode, file);
> 	mutex_unlock(&event_mutex);
> 
> 	return 0;
> 
> There's nothing special about the file->f_mode or the seq_release()
> that
> requires any locking. Remove the unnecessary locks.
> 
> Acked-by: Masami Hiramatsu (Google) <mhiramat@kernel.org>
> Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>


Looks good to me.

Reviewed-by: Tom Zanussi <zanussi@kernel.org>

> ---
>  kernel/trace/trace_events_trigger.c | 4 ----
>  1 file changed, 4 deletions(-)
> 
> diff --git a/kernel/trace/trace_events_trigger.c
> b/kernel/trace/trace_events_trigger.c
> index 7795af600466..e5dcfcbb2cd5 100644
> --- a/kernel/trace/trace_events_trigger.c
> +++ b/kernel/trace/trace_events_trigger.c
> @@ -314,13 +314,9 @@ static ssize_t event_trigger_regex_write(struct
> file *file,
>  
>  static int event_trigger_regex_release(struct inode *inode, struct
> file *file)
>  {
> -	mutex_lock(&event_mutex);
> -
>  	if (file->f_mode & FMODE_READ)
>  		seq_release(inode, file);
>  
> -	mutex_unlock(&event_mutex);
> -
>  	return 0;
>  }
>  

[PATCH v2 2/3] tracing: Add bulk garbage collection of freeing event_trigger_data

[Steven Rostedt]

From: Steven Rostedt <rostedt@goodmis.org>

The event trigger data requires a full tracepoint_synchronize_unregister()
call before freeing. That call can take 100s of milliseconds to complete.
In order to allow for bulk freeing of the trigger data, it can not call
the tracepoint_synchronize_unregister() for every individual trigger data
being free.

Create a kthread that gets created the first time a trigger data is freed,
and have it use the lockless llist to get the list of data to free, run
the tracepoint_synchronize_unregister() then free everything in the list.

By freeing hundreds of event_trigger_data elements together, it only
requires two runs of the synchronization function, and not hundreds of
runs. This speeds up the operation by orders of magnitude (milliseconds
instead of several seconds).

Acked-by: Masami Hiramatsu (Google) <mhiramat@kernel.org>
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
---
Changes since v1: https://patch.msgid.link/20251120205710.151041470@kernel.org

- Moved include of llist.h to trace.h as it is used there (Masami Hiramatsu)

 kernel/trace/trace.h                |  2 ++
 kernel/trace/trace_events_trigger.c | 55 +++++++++++++++++++++++++++--
 2 files changed, 54 insertions(+), 3 deletions(-)

diff --git a/kernel/trace/trace.h b/kernel/trace/trace.h
index 5863800b1ab3..911fc75dc6c4 100644
--- a/kernel/trace/trace.h
+++ b/kernel/trace/trace.h
@@ -22,6 +22,7 @@
 #include <linux/ctype.h>
 #include <linux/once_lite.h>
 #include <linux/ftrace_regs.h>
+#include <linux/llist.h>
 
 #include "pid_list.h"
 
@@ -1808,6 +1809,7 @@ struct event_trigger_data {
 	char				*name;
 	struct list_head		named_list;
 	struct event_trigger_data	*named_data;
+	struct llist_node		llist;
 };
 
 /* Avoid typos */
diff --git a/kernel/trace/trace_events_trigger.c b/kernel/trace/trace_events_trigger.c
index e5dcfcbb2cd5..3b97c242b795 100644
--- a/kernel/trace/trace_events_trigger.c
+++ b/kernel/trace/trace_events_trigger.c
@@ -6,6 +6,7 @@
  */
 
 #include <linux/security.h>
+#include <linux/kthread.h>
 #include <linux/module.h>
 #include <linux/ctype.h>
 #include <linux/mutex.h>
@@ -17,15 +18,63 @@
 static LIST_HEAD(trigger_commands);
 static DEFINE_MUTEX(trigger_cmd_mutex);
 
+static struct task_struct *trigger_kthread;
+static struct llist_head trigger_data_free_list;
+static DEFINE_MUTEX(trigger_data_kthread_mutex);
+
+/* Bulk garbage collection of event_trigger_data elements */
+static int trigger_kthread_fn(void *ignore)
+{
+	struct event_trigger_data *data, *tmp;
+	struct llist_node *llnodes;
+
+	/* Once this task starts, it lives forever */
+	for (;;) {
+		set_current_state(TASK_INTERRUPTIBLE);
+		if (llist_empty(&trigger_data_free_list))
+			schedule();
+
+		__set_current_state(TASK_RUNNING);
+
+		llnodes = llist_del_all(&trigger_data_free_list);
+
+		/* make sure current triggers exit before free */
+		tracepoint_synchronize_unregister();
+
+		llist_for_each_entry_safe(data, tmp, llnodes, llist)
+			kfree(data);
+	}
+
+	return 0;
+}
+
 void trigger_data_free(struct event_trigger_data *data)
 {
 	if (data->cmd_ops->set_filter)
 		data->cmd_ops->set_filter(NULL, data, NULL);
 
-	/* make sure current triggers exit before free */
-	tracepoint_synchronize_unregister();
+	if (unlikely(!trigger_kthread)) {
+		guard(mutex)(&trigger_data_kthread_mutex);
+		/* Check again after taking mutex */
+		if (!trigger_kthread) {
+			struct task_struct *kthread;
+
+			kthread = kthread_create(trigger_kthread_fn, NULL,
+						 "trigger_data_free");
+			if (!IS_ERR(kthread))
+				WRITE_ONCE(trigger_kthread, kthread);
+		}
+	}
+
+	if (!trigger_kthread) {
+		/* Do it the slow way */
+		tracepoint_synchronize_unregister();
+		kfree(data);
+		return;
+	}
 
-	kfree(data);
+	llist_add(&data->llist, &trigger_data_free_list);
+	wake_up_process(trigger_kthread);
 }
 
 static inline void data_ops_trigger(struct event_trigger_data *data,
-- 
2.51.0

Re: [PATCH v2 2/3] tracing: Add bulk garbage collection of freeing event_trigger_data

[Tom Zanussi]

On Tue, 2025-11-25 at 16:40 -0500, Steven Rostedt wrote:
> From: Steven Rostedt <rostedt@goodmis.org>
> 
> The event trigger data requires a full tracepoint_synchronize_unregister()
> call before freeing. That call can take 100s of milliseconds to complete.
> In order to allow for bulk freeing of the trigger data, it can not call
> the tracepoint_synchronize_unregister() for every individual trigger data
> being free.
> 
> Create a kthread that gets created the first time a trigger data is freed,
> and have it use the lockless llist to get the list of data to free, run
> the tracepoint_synchronize_unregister() then free everything in the list.
> 
> By freeing hundreds of event_trigger_data elements together, it only
> requires two runs of the synchronization function, and not hundreds of
> runs. This speeds up the operation by orders of magnitude (milliseconds
> instead of several seconds).
> 
> Acked-by: Masami Hiramatsu (Google) <mhiramat@kernel.org>
> Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>

Very nice!

Reviewed-by: Tom Zanussi <zanussi@kernel.org>

> ---
> Changes since v1: https://patch.msgid.link/20251120205710.151041470@kernel.org
> 
> - Moved include of llist.h to trace.h as it is used there (Masami Hiramatsu)
> 
>  kernel/trace/trace.h                |  2 ++
>  kernel/trace/trace_events_trigger.c | 55 +++++++++++++++++++++++++++--
>  2 files changed, 54 insertions(+), 3 deletions(-)
> 
> diff --git a/kernel/trace/trace.h b/kernel/trace/trace.h
> index 5863800b1ab3..911fc75dc6c4 100644
> --- a/kernel/trace/trace.h
> +++ b/kernel/trace/trace.h
> @@ -22,6 +22,7 @@
>  #include <linux/ctype.h>
>  #include <linux/once_lite.h>
>  #include <linux/ftrace_regs.h>
> +#include <linux/llist.h>
>  
>  #include "pid_list.h"
>  
> @@ -1808,6 +1809,7 @@ struct event_trigger_data {
>  	char				*name;
>  	struct list_head		named_list;
>  	struct event_trigger_data	*named_data;
> +	struct llist_node		llist;
>  };
>  
>  /* Avoid typos */
> diff --git a/kernel/trace/trace_events_trigger.c b/kernel/trace/trace_events_trigger.c
> index e5dcfcbb2cd5..3b97c242b795 100644
> --- a/kernel/trace/trace_events_trigger.c
> +++ b/kernel/trace/trace_events_trigger.c
> @@ -6,6 +6,7 @@
>   */
>  
>  #include <linux/security.h>
> +#include <linux/kthread.h>
>  #include <linux/module.h>
>  #include <linux/ctype.h>
>  #include <linux/mutex.h>
> @@ -17,15 +18,63 @@
>  static LIST_HEAD(trigger_commands);
>  static DEFINE_MUTEX(trigger_cmd_mutex);
>  
> +static struct task_struct *trigger_kthread;
> +static struct llist_head trigger_data_free_list;
> +static DEFINE_MUTEX(trigger_data_kthread_mutex);
> +
> +/* Bulk garbage collection of event_trigger_data elements */
> +static int trigger_kthread_fn(void *ignore)
> +{
> +	struct event_trigger_data *data, *tmp;
> +	struct llist_node *llnodes;
> +
> +	/* Once this task starts, it lives forever */
> +	for (;;) {
> +		set_current_state(TASK_INTERRUPTIBLE);
> +		if (llist_empty(&trigger_data_free_list))
> +			schedule();
> +
> +		__set_current_state(TASK_RUNNING);
> +
> +		llnodes = llist_del_all(&trigger_data_free_list);
> +
> +		/* make sure current triggers exit before free */
> +		tracepoint_synchronize_unregister();
> +
> +		llist_for_each_entry_safe(data, tmp, llnodes, llist)
> +			kfree(data);
> +	}
> +
> +	return 0;
> +}
> +
>  void trigger_data_free(struct event_trigger_data *data)
>  {
>  	if (data->cmd_ops->set_filter)
>  		data->cmd_ops->set_filter(NULL, data, NULL);
>  
> -	/* make sure current triggers exit before free */
> -	tracepoint_synchronize_unregister();
> +	if (unlikely(!trigger_kthread)) {
> +		guard(mutex)(&trigger_data_kthread_mutex);
> +		/* Check again after taking mutex */
> +		if (!trigger_kthread) {
> +			struct task_struct *kthread;
> +
> +			kthread = kthread_create(trigger_kthread_fn, NULL,
> +						 "trigger_data_free");
> +			if (!IS_ERR(kthread))
> +				WRITE_ONCE(trigger_kthread, kthread);
> +		}
> +	}
> +
> +	if (!trigger_kthread) {
> +		/* Do it the slow way */
> +		tracepoint_synchronize_unregister();
> +		kfree(data);
> +		return;
> +	}
>  
> -	kfree(data);
> +	llist_add(&data->llist, &trigger_data_free_list);
> +	wake_up_process(trigger_kthread);
>  }
>  
>  static inline void data_ops_trigger(struct event_trigger_data *data,

[PATCH v2 3/3] tracing: Use strim() in trigger_process_regex() instead of skip_spaces()

[Steven Rostedt]

From: Steven Rostedt <rostedt@goodmis.org>

The function trigger_process_regex() is called by a few functions, where
only one calls strim() on the buffer passed to it. That leaves the other
functions not trimming the end of the buffer passed in and making it a
little inconsistent.

Remove the strim() from event_trigger_regex_write() and have
trigger_process_regex() use strim() instead of skip_spaces(). The buff
variable is not passed in as const, so it can be modified.

Acked-by: Masami Hiramatsu (Google) <mhiramat@kernel.org>
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
---
 kernel/trace/trace_events_trigger.c | 5 ++---
 1 file changed, 2 insertions(+), 3 deletions(-)

diff --git a/kernel/trace/trace_events_trigger.c b/kernel/trace/trace_events_trigger.c
index 3b97c242b795..96aad82b1628 100644
--- a/kernel/trace/trace_events_trigger.c
+++ b/kernel/trace/trace_events_trigger.c
@@ -308,7 +308,8 @@ int trigger_process_regex(struct trace_event_file *file, char *buff)
 	char *command, *next;
 	struct event_command *p;
 
-	next = buff = skip_spaces(buff);
+	next = buff = strim(buff);
+
 	command = strsep(&next, ": \t");
 	if (next) {
 		next = skip_spaces(next);
@@ -345,8 +346,6 @@ static ssize_t event_trigger_regex_write(struct file *file,
 	if (IS_ERR(buf))
 		return PTR_ERR(buf);
 
-	strim(buf);
-
 	guard(mutex)(&event_mutex);
 
 	event_file = event_file_file(file);
-- 
2.51.0

Re: [PATCH v2 3/3] tracing: Use strim() in trigger_process_regex() instead of skip_spaces()

[Tom Zanussi]

On Tue, 2025-11-25 at 16:40 -0500, Steven Rostedt wrote:
> From: Steven Rostedt <rostedt@goodmis.org>
> 
> The function trigger_process_regex() is called by a few functions, where
> only one calls strim() on the buffer passed to it. That leaves the other
> functions not trimming the end of the buffer passed in and making it a
> little inconsistent.
> 
> Remove the strim() from event_trigger_regex_write() and have
> trigger_process_regex() use strim() instead of skip_spaces(). The buff
> variable is not passed in as const, so it can be modified.
> 
> Acked-by: Masami Hiramatsu (Google) <mhiramat@kernel.org>
> Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>

Looks good to me.

Reviewed-by: Tom Zanussi <zanussi@kernel.org>

> ---
>  kernel/trace/trace_events_trigger.c | 5 ++---
>  1 file changed, 2 insertions(+), 3 deletions(-)
> 
> diff --git a/kernel/trace/trace_events_trigger.c b/kernel/trace/trace_events_trigger.c
> index 3b97c242b795..96aad82b1628 100644
> --- a/kernel/trace/trace_events_trigger.c
> +++ b/kernel/trace/trace_events_trigger.c
> @@ -308,7 +308,8 @@ int trigger_process_regex(struct trace_event_file *file, char *buff)
>  	char *command, *next;
>  	struct event_command *p;
>  
> -	next = buff = skip_spaces(buff);
> +	next = buff = strim(buff);
> +
>  	command = strsep(&next, ": \t");
>  	if (next) {
>  		next = skip_spaces(next);
> @@ -345,8 +346,6 @@ static ssize_t event_trigger_regex_write(struct file *file,
>  	if (IS_ERR(buf))
>  		return PTR_ERR(buf);
>  
> -	strim(buf);
> -
>  	guard(mutex)(&event_mutex);
>  
>  	event_file = event_file_file(file);

Re: [PATCH v2 3/3] tracing: Use strim() in trigger_process_regex() instead of skip_spaces()

[Steven Rostedt]

On Wed, 03 Dec 2025 16:40:04 -0600
Tom Zanussi <zanussi@kernel.org> wrote:

> > Remove the strim() from event_trigger_regex_write() and have
> > trigger_process_regex() use strim() instead of skip_spaces(). The buff
> > variable is not passed in as const, so it can be modified.
> > 
> > Acked-by: Masami Hiramatsu (Google) <mhiramat@kernel.org>
> > Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>  
> 
> Looks good to me.
> 
> Reviewed-by: Tom Zanussi <zanussi@kernel.org>

Thanks for all the reviews, but I already did the pull request ;-)

Although it hasn't been accepted yet and there's always a chance Linus may
have an issue with one of the commits (I had a lot of commits this time,
affecting various aspects of the code, but couldn't use topic branches due
to too many conflicts). If it is rejected, I'll add your reviewed-by tags.

-- Steve

Re: [PATCH v2 3/3] tracing: Use strim() in trigger_process_regex() instead of skip_spaces()

[Tom Zanussi]

On Wed, 2025-12-03 at 17:46 -0500, Steven Rostedt wrote:
> On Wed, 03 Dec 2025 16:40:04 -0600
> Tom Zanussi <zanussi@kernel.org> wrote:
> 
> > > Remove the strim() from event_trigger_regex_write() and have
> > > trigger_process_regex() use strim() instead of skip_spaces(). The buff
> > > variable is not passed in as const, so it can be modified.
> > > 
> > > Acked-by: Masami Hiramatsu (Google) <mhiramat@kernel.org>
> > > Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>  
> > 
> > Looks good to me.
> > 
> > Reviewed-by: Tom Zanussi <zanussi@kernel.org>
> 
> Thanks for all the reviews, but I already did the pull request ;-)
> 
> Although it hasn't been accepted yet and there's always a chance Linus may
> have an issue with one of the commits (I had a lot of commits this time,
> affecting various aspects of the code, but couldn't use topic branches due
> to too many conflicts). If it is rejected, I'll add your reviewed-by tags.
> 
> -- Steve

No problem, my fault for being so slow to get to these. :-/

Tom