Linux Trace Kernel
 help / color / mirror / Atom feed
* [PATCH] tracing: Fix enabling of tracing on file release
@ 2025-12-02 21:17 Steven Rostedt
  0 siblings, 0 replies; only message in thread
From: Steven Rostedt @ 2025-12-02 21:17 UTC (permalink / raw)
  To: LKML, Linux Trace Kernel
  Cc: Masami Hiramatsu, Mathieu Desnoyers, syzkaller-bugs

From: Steven Rostedt <rostedt@goodmis.org>

The trace file will pause tracing if the tracing instance has the
"pause-on-trace" option is set. This happens when the file is opened, and
it is unpaused when the file is closed. When this was first added, there
was only one user that paused tracing. On open, the check to pause was:

   if (!iter->snapshot && (tr->trace_flags & TRACE_ITER(PAUSE_ON_TRACE)))

Where if it is not the snapshot tracer and the "pause-on-trace" option is
set, then it increments a "stop_count" of the trace instance.

On close, the check is:

   if (!iter->snapshot && tr->stop_count)

That is, if it is not the snapshot buffer and it was stopped, it will
re-enable tracing.

Now there's more places that stop tracing. This means, if something else
stops tracing the tr->stop_count will be non-zero, and that means if the
trace file is closed, it will decrement the stop_count even though it
never incremented it. This causes a warning because when the user that
stopped tracing enables it again, the stop_count goes below zero.

Instead of relying on the stop_count being set to know if the close of
the trace file should enable tracing again, add a new flag to the trace
iterator. The trace iterator is unique per open of the trace file, and if
the open stops tracing set the trace iterator PAUSE flag. On close, if the
PAUSE flag is set, then re-enable it again.

Fixes: 06e0a548bad0f ("tracing: Do not disable tracing when reading the trace file")
Reported-by: syzbot+ccdec3bfe0beec58a38d@syzkaller.appspotmail.com
Closes: https://lore.kernel.org/all/692f44a5.a70a0220.2ea503.00c8.GAE@google.com/
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
---
 include/linux/trace_events.h | 1 +
 kernel/trace/trace.c         | 6 ++++--
 2 files changed, 5 insertions(+), 2 deletions(-)

diff --git a/include/linux/trace_events.h b/include/linux/trace_events.h
index 04307a19cde3..3690221ba3d8 100644
--- a/include/linux/trace_events.h
+++ b/include/linux/trace_events.h
@@ -138,6 +138,7 @@ enum trace_iter_flags {
 	TRACE_FILE_LAT_FMT	= 1,
 	TRACE_FILE_ANNOTATE	= 2,
 	TRACE_FILE_TIME_IN_NS	= 4,
+	TRACE_FILE_PAUSE	= 8,
 };
 
 
diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c
index 3d433a426e5f..7b9d93147d01 100644
--- a/kernel/trace/trace.c
+++ b/kernel/trace/trace.c
@@ -4709,8 +4709,10 @@ __tracing_open(struct inode *inode, struct file *file, bool snapshot)
 	 * If pause-on-trace is enabled, then stop the trace while
 	 * dumping, unless this is the "snapshot" file
 	 */
-	if (!iter->snapshot && (tr->trace_flags & TRACE_ITER(PAUSE_ON_TRACE)))
+	if (!iter->snapshot && (tr->trace_flags & TRACE_ITER(PAUSE_ON_TRACE))) {
+		iter->iter_flags |= TRACE_FILE_PAUSE;
 		tracing_stop_tr(tr);
+	}
 
 	if (iter->cpu_file == RING_BUFFER_ALL_CPUS) {
 		for_each_tracing_cpu(cpu) {
@@ -4842,7 +4844,7 @@ static int tracing_release(struct inode *inode, struct file *file)
 	if (iter->trace && iter->trace->close)
 		iter->trace->close(iter);
 
-	if (!iter->snapshot && tr->stop_count)
+	if (iter->iter_flags & TRACE_FILE_PAUSE)
 		/* reenable tracing if it was previously enabled */
 		tracing_start_tr(tr);
 
-- 
2.51.0


^ permalink raw reply related	[flat|nested] only message in thread

only message in thread, other threads:[~2025-12-02 21:16 UTC | newest]

Thread overview: (only message) (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2025-12-02 21:17 [PATCH] tracing: Fix enabling of tracing on file release Steven Rostedt

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox