[PATCH 0/2] fgraph: fixes for thresh_return return handler

[PATCH 0/2] fgraph: fixes for thresh_return return handler

[hu.shengming]

[-- Attachment #1.1.1: Type: text/plain, Size: 729 bytes --]

From: Shengming Hu <hu.shengming@zte.com.cn>

Hi,

This series fixes two issues in trace_graph_thresh_return(), the function
graph return handler used when tracing_thresh is enabled:

PATCH1: Clear the per-task TRACE_GRAPH_NOTRACE state like
trace_graph_return() does.
PATCH2: Avoid double no-sleep-time adjustment by emitting
the return event directly when the threshold is met.

Patch details are in the individual commit messages.

Shengming Hu (2):
  [PATCH 1/2] fgraph: fix thresh_return clear per-task notrace
  [PATCH 2/2] fgraph: fix thresh_return nosleeptime double-adjust

 kernel/trace/trace_functions_graph.c | 19 +++++++++++++------
 1 file changed, 13 insertions(+), 6 deletions(-)

-- 
2.25.1

[-- Attachment #1.1.2: Type: text/html , Size: 1619 bytes --]

[PATCH 1/2] fgraph: fix thresh_return clear per-task notrace

[hu.shengming]

From: Shengming Hu <hu.shengming@zte.com.cn>

When tracing_thresh is enabled, function graph tracing uses
trace_graph_thresh_return() as the return handler. Unlike
trace_graph_return(), it did not clear the per-task
TRACE_GRAPH_NOTRACE flag set by the entry handler for
set_graph_notrace addresses. This could leave the task
permanently in "notrace" state and effectively disable
function graph tracing for that task.

Mirror trace_graph_return()'s per-task notrace handling by
clearing TRACE_GRAPH_NOTRACE and returning early when set.

Fixes: b84214890a9bc ("function_graph: Move graph notrace bit to
shadow stack global var")
Signed-off-by: Shengming Hu <hu.shengming@zte.com.cn>
---
 kernel/trace/trace_functions_graph.c | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)

diff --git a/kernel/trace/trace_functions_graph.c b/kernel/trace/trace_functions_graph.c
index 1de6f1573..cbe43680c 100644
--- a/kernel/trace/trace_functions_graph.c
+++ b/kernel/trace/trace_functions_graph.c
@@ -400,14 +400,15 @@ static void trace_graph_thresh_return(struct ftrace_graph_ret *trace,
 				      struct fgraph_ops *gops,
 				      struct ftrace_regs *fregs)
 {
+	unsigned long *task_var = fgraph_get_task_var(gops);
 	struct fgraph_times *ftimes;
 	struct trace_array *tr;
 	int size;

 	ftrace_graph_addr_finish(gops, trace);

-	if (trace_recursion_test(TRACE_GRAPH_NOTRACE_BIT)) {
-		trace_recursion_clear(TRACE_GRAPH_NOTRACE_BIT);
+	if (*task_var & TRACE_GRAPH_NOTRACE) {
+		*task_var &= ~TRACE_GRAPH_NOTRACE;
 		return;
 	}

-- 
2.25.1

[PATCH 2/2] fgraph: fix thresh_return nosleeptime double-adjust

[hu.shengming]

From: Shengming Hu <hu.shengming@zte.com.cn>

trace_graph_thresh_return() called handle_nosleeptime() and then
delegated to trace_graph_return(), which calls
handle_nosleeptime() again. When sleep-time accounting is
disabled this double-adjusts calltime and can produce bogus
durations (including underflow).

Fix this by computing rettime once, applying
handle_nosleeptime() only once, using the adjusted calltime
for threshold comparison, and writing the return event
directly via __trace_graph_return() when the threshold is met.

Fixes: 3c9880f3ab52b ("ftrace: Use a running sleeptime instead of
saving on shadow stack")
Signed-off-by: Shengming Hu <hu.shengming@zte.com.cn>
---
 kernel/trace/trace_functions_graph.c | 14 ++++++++++----
 1 file changed, 10 insertions(+), 4 deletions(-)

diff --git a/kernel/trace/trace_functions_graph.c b/kernel/trace/trace_functions_graph.c
index cbe43680c..b9c81fbd9 100644
--- a/kernel/trace/trace_functions_graph.c
+++ b/kernel/trace/trace_functions_graph.c
@@ -403,8 +403,12 @@ static void trace_graph_thresh_return(struct ftrace_graph_ret *trace,
 	unsigned long *task_var = fgraph_get_task_var(gops);
 	struct fgraph_times *ftimes;
 	struct trace_array *tr;
+	unsigned int trace_ctx;
+	u64 calltime, rettime;
 	int size;

+	rettime = trace_clock_local();
+
 	ftrace_graph_addr_finish(gops, trace);

 	if (*task_var & TRACE_GRAPH_NOTRACE) {
@@ -419,11 +423,13 @@ static void trace_graph_thresh_return(struct ftrace_graph_ret *trace,
 	tr = gops->private;
 	handle_nosleeptime(tr, trace, ftimes, size);

-	if (tracing_thresh &&
-	    (trace_clock_local() - ftimes->calltime < tracing_thresh))
+	calltime = ftimes->calltime;
+
+	if (tracing_thresh && (rettime - calltime < tracing_thresh))
 		return;
-	else
-		trace_graph_return(trace, gops, fregs);
+
+	trace_ctx = tracing_gen_ctx();
+	__trace_graph_return(tr, trace, trace_ctx, calltime, rettime);
 }

 static struct fgraph_ops funcgraph_ops = {
-- 
2.25.1

Re: [PATCH 1/2] fgraph: fix thresh_return clear per-task notrace

[Masami Hiramatsu]

On Sat, 21 Feb 2026 11:30:07 +0800 (CST)
<hu.shengming@zte.com.cn> wrote:

> From: Shengming Hu <hu.shengming@zte.com.cn>
> 
> When tracing_thresh is enabled, function graph tracing uses
> trace_graph_thresh_return() as the return handler. Unlike
> trace_graph_return(), it did not clear the per-task
> TRACE_GRAPH_NOTRACE flag set by the entry handler for
> set_graph_notrace addresses. This could leave the task
> permanently in "notrace" state and effectively disable
> function graph tracing for that task.
> 
> Mirror trace_graph_return()'s per-task notrace handling by
> clearing TRACE_GRAPH_NOTRACE and returning early when set.
> 

Good catch!

Acked-by: Masami Hiramatsu (Google) <mhiramat@kernel.org>

Thanks!


> Fixes: b84214890a9bc ("function_graph: Move graph notrace bit to
> shadow stack global var")
> Signed-off-by: Shengming Hu <hu.shengming@zte.com.cn>
> ---
>  kernel/trace/trace_functions_graph.c | 5 +++--
>  1 file changed, 3 insertions(+), 2 deletions(-)
> 
> diff --git a/kernel/trace/trace_functions_graph.c b/kernel/trace/trace_functions_graph.c
> index 1de6f1573..cbe43680c 100644
> --- a/kernel/trace/trace_functions_graph.c
> +++ b/kernel/trace/trace_functions_graph.c
> @@ -400,14 +400,15 @@ static void trace_graph_thresh_return(struct ftrace_graph_ret *trace,
>  				      struct fgraph_ops *gops,
>  				      struct ftrace_regs *fregs)
>  {
> +	unsigned long *task_var = fgraph_get_task_var(gops);
>  	struct fgraph_times *ftimes;
>  	struct trace_array *tr;
>  	int size;
> 
>  	ftrace_graph_addr_finish(gops, trace);
> 
> -	if (trace_recursion_test(TRACE_GRAPH_NOTRACE_BIT)) {
> -		trace_recursion_clear(TRACE_GRAPH_NOTRACE_BIT);
> +	if (*task_var & TRACE_GRAPH_NOTRACE) {
> +		*task_var &= ~TRACE_GRAPH_NOTRACE;
>  		return;
>  	}
> 
> -- 
> 2.25.1


-- 
Masami Hiramatsu (Google) <mhiramat@kernel.org>

Re: [PATCH 2/2] fgraph: fix thresh_return nosleeptime double-adjust

[Masami Hiramatsu]

On Sat, 21 Feb 2026 11:33:14 +0800 (CST)
<hu.shengming@zte.com.cn> wrote:

> From: Shengming Hu <hu.shengming@zte.com.cn>
> 
> trace_graph_thresh_return() called handle_nosleeptime() and then
> delegated to trace_graph_return(), which calls
> handle_nosleeptime() again. When sleep-time accounting is
> disabled this double-adjusts calltime and can produce bogus
> durations (including underflow).
> 
> Fix this by computing rettime once, applying
> handle_nosleeptime() only once, using the adjusted calltime
> for threshold comparison, and writing the return event
> directly via __trace_graph_return() when the threshold is met.
> 

Looks good to me.

Acked-by: Masami Hiramatsu (Google) <mhiramat@kernel.org>

Thanks!

> Fixes: 3c9880f3ab52b ("ftrace: Use a running sleeptime instead of
> saving on shadow stack")
> Signed-off-by: Shengming Hu <hu.shengming@zte.com.cn>
> ---
>  kernel/trace/trace_functions_graph.c | 14 ++++++++++----
>  1 file changed, 10 insertions(+), 4 deletions(-)
> 
> diff --git a/kernel/trace/trace_functions_graph.c b/kernel/trace/trace_functions_graph.c
> index cbe43680c..b9c81fbd9 100644
> --- a/kernel/trace/trace_functions_graph.c
> +++ b/kernel/trace/trace_functions_graph.c
> @@ -403,8 +403,12 @@ static void trace_graph_thresh_return(struct ftrace_graph_ret *trace,
>  	unsigned long *task_var = fgraph_get_task_var(gops);
>  	struct fgraph_times *ftimes;
>  	struct trace_array *tr;
> +	unsigned int trace_ctx;
> +	u64 calltime, rettime;
>  	int size;
> 
> +	rettime = trace_clock_local();
> +
>  	ftrace_graph_addr_finish(gops, trace);
> 
>  	if (*task_var & TRACE_GRAPH_NOTRACE) {
> @@ -419,11 +423,13 @@ static void trace_graph_thresh_return(struct ftrace_graph_ret *trace,
>  	tr = gops->private;
>  	handle_nosleeptime(tr, trace, ftimes, size);
> 
> -	if (tracing_thresh &&
> -	    (trace_clock_local() - ftimes->calltime < tracing_thresh))
> +	calltime = ftimes->calltime;
> +
> +	if (tracing_thresh && (rettime - calltime < tracing_thresh))
>  		return;
> -	else
> -		trace_graph_return(trace, gops, fregs);
> +
> +	trace_ctx = tracing_gen_ctx();
> +	__trace_graph_return(tr, trace, trace_ctx, calltime, rettime);
>  }
> 
>  static struct fgraph_ops funcgraph_ops = {
> -- 
> 2.25.1
> 


-- 
Masami Hiramatsu (Google) <mhiramat@kernel.org>