bug: shut up format attribute warning for clang as well

PATCH Superseded Arnd Bergmann 6 participants 7 messages Mar 20 – Mar 24 Library code

lore

Share & tools

lore

Arnd Bergmann

author

2026-03-20 21:15 UTC #0 lore raw

PATCH 1/1 bug: shut up format attribute warning for clang as well

From: Arnd Bergmann <[email protected]>

Like gcc, clang-22 now also warns about a function that it incorrectly identifies as a printf-style format:

lib/bug.c:190:22: error: diagnostic behavior may be improved by adding the 'format(printf, 1, 0)' attribute to the declaration of '__warn_printf' [-Werror,-Wmissing-format-attribute]

  179 | static void __warn_printf(const char *fmt, struct pt_regs *regs)
      | __attribute__((format(printf, 1, 0)))
  180 | {
  181 |         if (!fmt)
  182 |                 return;
  183 |
  184 | #ifdef HAVE_ARCH_BUG_FORMAT_ARGS
  185 |         if (regs) {
  186 |                 struct arch_va_list _args;
  187 |                 va_list *args = __warn_args(&_args, regs);
  188 |
  189 |                 if (args) {
  190 |                         vprintk(fmt, *args);
      |                                           ^

Turn off this warning for all compilers and versions.

Fixes d36067d6ea00 ("bug: Hush suggest-attribute=format for __warn_printf()")

Signed-off-by Arnd Bergmann <[email protected]>

---
 lib/bug.c | 3 +--
 1 file changed, 1 insertion(+), 2 deletions(-)

lib/bug.c +1 -2

index bbc301097749..374feb144f0b 100644
@@ -174,8 +174,7 @@ struct bug_entry *find_bug(unsigned long bugaddr)
 }
 
 __diag_push();
-__diag_ignore(GCC, all, "-Wsuggest-attribute=format",
-	      "Not a valid __printf() conversion candidate.");
+__diag_ignore_all("-Wmissing-format-attribute", "Not a valid __printf() conversion candidate.");
 static void __warn_printf(const char *fmt, struct pt_regs *regs)
 {
 	if (!fmt)

Signature · 1 line

2.39.5

Nathan Chancellor

Reviewed-by on v1 1/1

2026-03-21 00:08 UTC #1 lore raw

Quoted 21 lines · Arnd Bergmann

From: Arnd Bergmann <[email protected]>
Like gcc, clang-22 now also warns about a function that it incorrectly identifies as a printf-style format:
lib/bug.c:190:22: error: diagnostic behavior may be improved by adding the 'format(printf, 1, 0)' attribute to the declaration of '__warn_printf' [-Werror,-Wmissing-format-attribute]
  179 | static void __warn_printf(const char *fmt, struct pt_regs *regs)
      | __attribute__((format(printf, 1, 0)))
  180 | {
  181 |         if (!fmt)
  182 |                 return;
  183 |
  184 | #ifdef HAVE_ARCH_BUG_FORMAT_ARGS
  185 |         if (regs) {
  186 |                 struct arch_va_list _args;
  187 |                 va_list *args = __warn_args(&_args, regs);
  188 |
  189 |                 if (args) {
  190 |                         vprintk(fmt, *args);
      |                                           ^

Quoted 4 lines

Turn off this warning for all compilers and versions.
Fixes d36067d6ea00 ("bug: Hush suggest-attribute=format for __warn_printf()")
Signed-off-by Arnd Bergmann <[email protected]>

Yeah, it looks like this flag has been recognized by clang for a long time for GCC compatibility but it only recently started matching GCC's behavior.

Reviewed-by Nathan Chancellor <[email protected]>

Quoted 21 lines

---
 lib/bug.c | 3 +--
 1 file changed, 1 insertion(+), 2 deletions(-)

lib/bug.c +1 -2

index bbc301097749..374feb144f0b 100644
@@ -174,8 +174,7 @@ struct bug_entry *find_bug(unsigned long bugaddr)
 }
 
 __diag_push();
-__diag_ignore(GCC, all, "-Wsuggest-attribute=format",
-	      "Not a valid __printf() conversion candidate.");
+__diag_ignore_all("-Wmissing-format-attribute", "Not a valid __printf() conversion candidate.");
 static void __warn_printf(const char *fmt, struct pt_regs *regs)
 {
 	if (!fmt)

Signature · 1 line

2.39.5

Brendan Jackman

2026-03-23 10:54 UTC #2 lore raw

In reply to Arnd Bergmann

Quoted 34 lines · Arnd Bergmann

From: Arnd Bergmann <[email protected]>
Like gcc, clang-22 now also warns about a function that it incorrectly identifies as a printf-style format:
lib/bug.c:190:22: error: diagnostic behavior may be improved by adding the 'format(printf, 1, 0)' attribute to the declaration of '__warn_printf' [-Werror,-Wmissing-format-attribute]
  179 | static void __warn_printf(const char *fmt, struct pt_regs *regs)
      | __attribute__((format(printf, 1, 0)))
  180 | {
  181 |         if (!fmt)
  182 |                 return;
  183 |
  184 | #ifdef HAVE_ARCH_BUG_FORMAT_ARGS
  185 |         if (regs) {
  186 |                 struct arch_va_list _args;
  187 |                 va_list *args = __warn_args(&_args, regs);
  188 |
  189 |                 if (args) {
  190 |                         vprintk(fmt, *args);
      |                                           ^
Turn off this warning for all compilers and versions.
Fixes d36067d6ea00 ("bug: Hush suggest-attribute=format for __warn_printf()")
Signed-off-by Arnd Bergmann <[email protected]>
---
 lib/bug.c | 3 +--
 1 file changed, 1 insertion(+), 2 deletions(-)
lib/bug.c +0 -0
index bbc301097749..374feb144f0b 100644

lib/bug.c struct bug_entry *find_bug(unsigned long bugaddr) · line 180

Quoted 9 lines

lib/bug.c +1 -2

@@ -174,8 +174,7 @@ struct bug_entry *find_bug(unsigned long bugaddr)
 }
 
 __diag_push();
-__diag_ignore(GCC, all, "-Wsuggest-attribute=format",
-	      "Not a valid __printf() conversion candidate.");
+__diag_ignore_all("-Wmissing-format-attribute", "Not a valid __printf() conversion candidate.");
 static void __warn_printf(const char *fmt, struct pt_regs *regs)
 {
 	if (!fmt)

There is actually a superior fix for this warning that Andy Schevchenko posted here (mine just happened to get merged first):

https://lore.kernel.org/all/[email protected]/T/#u

Should we just switch over to that? Looks like it should work for both compilers.

Arnd Bergmann

2026-03-23 15:39 UTC #3 lore raw

Quoted 9 lines · Brendan Jackman

On Fri Mar 20, 2026 at 9:14 PM UTC, Arnd Bergmann wrote:
There is actually a superior fix for this warning that Andy Schevchenko posted here (mine just happened to get merged first):
https://lore.kernel.org/all/[email protected]/T/#u
Should we just switch over to that? Looks like it should work for both compilers.

Right, I tried that again and it works. I had confused it with the similar logic to disable the warning on va_format(), which is not that easy to avoid.

Arnd

Andrew Morton

maintainer

2026-03-23 20:45 UTC #4 lore raw

Quoted 6 lines · On Mon, 23 Mar 2026 16:38:14 +0100 "Arnd Bergmann" <[email protected]> wrote:

Quoted 5 lines · Brendan Jackman
On Fri Mar 20, 2026 at 9:14 PM UTC, Arnd Bergmann wrote:
There is actually a superior fix for this warning that Andy Schevchenko posted here (mine just happened to get merged first):

Quoted 9 lines

Quoted 4 lines
https://lore.kernel.org/all/[email protected]/T/#u
Should we just switch over to that? Looks like it should work for both compilers.
Right, I tried that again and it works. I had confused it with the similar logic to disable the warning on va_format(), which is not that easy to avoid.

What I think I'm hearing is that we should revert Brendan's d36067d6ea00 ("bug: Hush suggest-attribute=format for __warn_printf()") then apply Andy's
https://lore.kernel.org/all/[email protected]/T/#u,
yes?

Could someone please prepare a suitably changelogged and tested patch for that?

Arnd Bergmann

2026-03-23 20:57 UTC #5 lore raw

Quoted 18 lines · Andrew Morton

Quoted 6 lines · On Mon, 23 Mar 2026 16:38:14 +0100 "Arnd Bergmann" <[email protected]> wrote:
Quoted 5 lines · Brendan Jackman
On Fri Mar 20, 2026 at 9:14 PM UTC, Arnd Bergmann wrote:
There is actually a superior fix for this warning that Andy Schevchenko posted here (mine just happened to get merged first):
Quoted 9 lines
Quoted 4 lines
https://lore.kernel.org/all/[email protected]/T/#u
Should we just switch over to that? Looks like it should work for both compilers.
Right, I tried that again and it works. I had confused it with the similar logic to disable the warning on va_format(), which is not that easy to avoid.

Quoted 8 lines

What I think I'm hearing is that we should revert Brendan's d36067d6ea00 ("bug: Hush suggest-attribute=format for __warn_printf()") then apply Andy's
https://lore.kernel.org/all/[email protected]/T/#u,
yes?
Could someone please prepare a suitably changelogged and tested patch for that?

I now sent the version that I had in my tree already as a v2 of my patch. I tested this with gcc-16 and clang-22 in random configurations.

Arnd

Andy Shevchenko

2026-03-24 11:26 UTC #6 lore raw

Quoted 23 lines · Arnd Bergmann

Quoted 22 lines · Andrew Morton
Quoted 6 lines · On Mon, 23 Mar 2026 16:38:14 +0100 "Arnd Bergmann" <[email protected]> wrote:
Quoted 5 lines · Brendan Jackman
On Fri Mar 20, 2026 at 9:14 PM UTC, Arnd Bergmann wrote:
There is actually a superior fix for this warning that Andy Schevchenko posted here (mine just happened to get merged first):
Quoted 8 lines
Quoted 4 lines
https://lore.kernel.org/all/[email protected]/T/#u
Should we just switch over to that? Looks like it should work for both compilers.
Right, I tried that again and it works. I had confused it with the similar logic to disable the warning on va_format(), which is not that easy to avoid.
What I think I'm hearing is that we should revert Brendan's d36067d6ea00 ("bug: Hush suggest-attribute=format for __warn_printf()") then apply Andy's
https://lore.kernel.org/all/[email protected]/T/#u,
yes?

Quoted 6 lines

Quoted 2 lines
Could someone please prepare a suitably changelogged and tested patch for that?
I now sent the version that I had in my tree already as a v2 of my patch. I tested this with gcc-16 and clang-22 in random configurations.

Thank you!

Signature · 2 lines

With Best Regards,
Andy Shevchenko