mirror of https://lore.kernel.org/lkml/
 help / color / mirror / Atom feed
* [PATCH] tracing/synthetic: avoid comparing string against NULL
@ 2026-10-13 20:53 Arnd Bergmann
  2026-11-17 15:08 ` Steven Rostedt
  0 siblings, 1 reply; 2+ messages in thread
From: Arnd Bergmann @ 2026-10-13 20:53 UTC (permalink / raw)
  To: Steven Rostedt, Masami Hiramatsu
  Cc: Arnd Bergmann, Mathieu Desnoyers, Yu Peng, Thomas Gleixner,
	Thomas Weißschuh (Schneider Electric),
	Namhyung Kim, linux-kernel, linux-trace-kernel

From: Arnd Bergmann <arnd@arndb.de>

gcc-10 warns about passing a NULL argument into __synth_event_add_val(),
which uses it in an unchecked strcmp():

In function '__synth_event_add_val',
    inlined from 'synth_event_add_next_val' at kernel/trace/trace_events_synth.c:2162:9:
kernel/trace/trace_events_synth.c:2076:8: error: argument 2 null where non-null expected [-Werror=nonnull]
 2076 |    if (strcmp(field->name, field_name) == 0)
      |        ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
In file included from include/linux/string.h:65,
                 from kernel/trace/trace_events_synth.c:8:
kernel/trace/trace_events_synth.c: In function 'synth_event_add_next_val':
arch/s390/include/asm/string.h:39:5: note: in a call to function 'strcmp' declared here
   39 | int strcmp(const char *s1, const char *s2);
      |     ^~~~~~

The code is actually correct here, as the trace_state->add_name
does check for it, but this is hard for the compiler to prove.

Avoid the problem by repeating the check where the compiler
sees it.

Signed-off-by: Arnd Bergmann <arnd@arndb.de>
---
 kernel/trace/trace_events_synth.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/kernel/trace/trace_events_synth.c b/kernel/trace/trace_events_synth.c
index dc15658a887c..c1287ca95aff 100644
--- a/kernel/trace/trace_events_synth.c
+++ b/kernel/trace/trace_events_synth.c
@@ -2073,7 +2073,8 @@ static int __synth_event_add_val(const char *field_name, u64 val,
 	if (trace_state->add_name) {
 		for (i = 0; i < event->n_fields; i++) {
 			field = event->fields[i];
-			if (strcmp(field->name, field_name) == 0)
+			if (!field_name ||
+			    strcmp(field->name, field_name) == 0)
 				break;
 		}
 		if (!field) {
-- 
2.53.0


^ permalink raw reply	[flat|nested] 2+ messages in thread

* Re: [PATCH] tracing/synthetic: avoid comparing string against NULL
  2026-10-13 20:53 [PATCH] tracing/synthetic: avoid comparing string against NULL Arnd Bergmann
@ 2026-11-17 15:08 ` Steven Rostedt
  0 siblings, 0 replies; 2+ messages in thread
From: Steven Rostedt @ 2026-11-17 15:08 UTC (permalink / raw)
  To: Arnd Bergmann
  Cc: Masami Hiramatsu, Arnd Bergmann, Mathieu Desnoyers, Yu Peng,
	Thomas Gleixner, Thomas Weißschuh (Schneider Electric),
	Namhyung Kim, linux-kernel, linux-trace-kernel

On Tue, 13 Oct 2026 22:53:52 +0200
Arnd Bergmann <arnd@kernel.org> wrote:

> From: Arnd Bergmann <arnd@arndb.de>
> 
> gcc-10 warns about passing a NULL argument into __synth_event_add_val(),
> which uses it in an unchecked strcmp():
> 
> In function '__synth_event_add_val',
>     inlined from 'synth_event_add_next_val' at kernel/trace/trace_events_synth.c:2162:9:
> kernel/trace/trace_events_synth.c:2076:8: error: argument 2 null where non-null expected [-Werror=nonnull]
>  2076 |    if (strcmp(field->name, field_name) == 0)
>       |        ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
> In file included from include/linux/string.h:65,
>                  from kernel/trace/trace_events_synth.c:8:
> kernel/trace/trace_events_synth.c: In function 'synth_event_add_next_val':
> arch/s390/include/asm/string.h:39:5: note: in a call to function 'strcmp' declared here
>    39 | int strcmp(const char *s1, const char *s2);
>       |     ^~~~~~
> 
> The code is actually correct here, as the trace_state->add_name
> does check for it, but this is hard for the compiler to prove.
> 
> Avoid the problem by repeating the check where the compiler
> sees it.

Yeah, but the compiler is being stupid here. It's not that difficult for it
to see it.


	if (field_name) {                      <<-- checks field_name
		if (trace_state->add_next) {
			ret = -EINVAL;
			goto out;
		}
		trace_state->add_name = true;  <<-- set add_name
	} else {
		if (trace_state->add_name) {  <<-- If field_name is NULL and add name is set, error out here
			ret = -EINVAL;
			goto out;
		}
		trace_state->add_next = true;
	}

	if (trace_state->disabled)
		goto out;

	event = trace_state->event;
	if (trace_state->add_name) {  <<-- Only get here if add_name is set (thus field_name is not NULL)
		for (i = 0; i < event->n_fields; i++) {
			field = event->fields[i];
			if (strcmp(field->name, field_name) == 0)   <<-- test field_name
				break;
		}

> 
> Signed-off-by: Arnd Bergmann <arnd@arndb.de>
> ---
>  kernel/trace/trace_events_synth.c | 3 ++-
>  1 file changed, 2 insertions(+), 1 deletion(-)
> 
> diff --git a/kernel/trace/trace_events_synth.c b/kernel/trace/trace_events_synth.c
> index dc15658a887c..c1287ca95aff 100644
> --- a/kernel/trace/trace_events_synth.c
> +++ b/kernel/trace/trace_events_synth.c
> @@ -2073,7 +2073,8 @@ static int __synth_event_add_val(const char *field_name, u64 val,
>  	if (trace_state->add_name) {
>  		for (i = 0; i < event->n_fields; i++) {
>  			field = event->fields[i];
> -			if (strcmp(field->name, field_name) == 0)
> +			if (!field_name ||
> +			    strcmp(field->name, field_name) == 0)

I don't like making the code look worse to please something stupid the
compiler is doing.

-- Steve


>  				break;
>  		}
>  		if (!field) {


^ permalink raw reply	[flat|nested] 2+ messages in thread

end of thread, other threads:[~2026-11-17 15:09 UTC | newest]

Thread overview: 2+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2026-10-13 20:53 [PATCH] tracing/synthetic: avoid comparing string against NULL Arnd Bergmann
2026-11-17 15:08 ` Steven Rostedt

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

all inboxes | Powered by JetHome®