Skip to contentSkip to conversation

seq_file: Replace strncpy()+nul by strscpy()

PATCH Applied Andy Shevchenko 3 participants 8 messages Jul 17, 2023 – Jul 19, 2023

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 seq_file: Replace strncpy()+nul by strscpy()

Privided seq_show_option_n() macro breaks build with -Werror and W=1, e.g.:

In function ‘strncpy’,
inlined from ‘ocfs2_show_options’ at fs/ocfs2/super.c:1520:3:
include/linux/fortify-string.h:68:33: error: ‘__builtin_strncpy’ output may be truncated copying 4 bytes from a string of length 4 [-Werror=stringop-truncation]

   68 | #define __underlying_strncpy    __builtin_strncpy
      |                                 ^
include/linux/fortify-string.h:151:16: note: in expansion of macro ‘__underlying_strncpy’
  151 |         return __underlying_strncpy(p, q, size);
      |                ^~~~~~~~~~~~~~~~~~~~

cc1: all warnings being treated as errors

While -Werror wasn't enabled by default at the time of the original code landed into mainline, strscpy() was already there and preferred over strncpy(). Due to above mentioned issues, use the latter in seq_show_option_n().

Fixes a068acf2ee77 ("fs: create and use seq_show_option for escaping")
---
 include/linux/seq_file.h | 3 +--
 1 file changed, 1 insertion(+), 2 deletions(-)
include/linux/seq_file.h +1 -2
index bd023dd38ae6..e87d635ca24f 100644
@@ -260,8 +260,7 @@ static inline void seq_show_option(struct seq_file *m, const char *name,
  */
 #define seq_show_option_n(m, name, value, length) {	\
 	char val_buf[length + 1];			\
-	strncpy(val_buf, value, length);		\
-	val_buf[length] = '\0';				\
+	strscpy(val_buf, value, sizeof(val_buf));	\
 	seq_show_option(m, name, val_buf);		\
 }
 
Signature · 1 line
2.40.0.1.gaa8946217a0b
Reviewed-by on v1 1/1
#1 lore raw
Quoted 8 lines · Andy Shevchenko

Privided seq_show_option_n() macro breaks build with -Werror and W=1, e.g.:

In function ‘strncpy’,
inlined from ‘ocfs2_show_options’ at fs/ocfs2/super.c:1520:3:
include/linux/fortify-string.h:68:33: error: ‘__builtin_strncpy’ output may be truncated copying 4 bytes from a string of length 4 [-Werror=stringop-truncation]

   68 | #define __underlying_strncpy    __builtin_strncpy
      |                                 ^

While I totally agree with the removal of strncpy(), I'm confused about how this warning is being produced:

                seq_show_option_n(s, "cluster_stack", osb->osb_cluster_stack,
                                  OCFS2_STACK_LABEL_LEN);

fs/ocfs2/ocfs2.h:389: char osb_cluster_stack[OCFS2_STACK_LABEL_LEN + 1];

fs/ocfs2/ocfs2_fs.h:#define OCFS2_STACK_LABEL_LEN 4

#define seq_show_option_n(m, name, value, length) {  \
     char val_buf[length + 1];                       \
     strncpy(val_buf, value, length);                \

...

the source buffer is OCFS2_STACK_LABEL_LEN + 1 long, and the dest buffer is OCFS2_STACK_LABEL_LEN + 1 long. ??

I think this doesn't need to use seq_show_option_n() at all.

Quoted 20 lines

include/linux/fortify-string.h:151:16: note: in expansion of macro ‘__underlying_strncpy’

  151 |         return __underlying_strncpy(p, q, size);
      |                ^~~~~~~~~~~~~~~~~~~~

cc1: all warnings being treated as errors

While -Werror wasn't enabled by default at the time of the original code landed into mainline, strscpy() was already there and preferred over strncpy(). Due to above mentioned issues, use the latter in seq_show_option_n().

Fixes a068acf2ee77 ("fs: create and use seq_show_option for escaping")
---
 include/linux/seq_file.h | 3 +--
 1 file changed, 1 insertion(+), 2 deletions(-)
include/linux/seq_file.h +0 -0
index bd023dd38ae6..e87d635ca24f 100644

include/linux/seq_file.h static inline void seq_show_option(struct seq_file *m, const char *name, · line 265

Quoted 8 lines
include/linux/seq_file.h +1 -2
@@ -260,8 +260,7 @@ static inline void seq_show_option(struct seq_file *m, const char *name,
  */
 #define seq_show_option_n(m, name, value, length) {	\
 	char val_buf[length + 1];			\
-	strncpy(val_buf, value, length);		\
-	val_buf[length] = '\0';				\
+	strscpy(val_buf, value, sizeof(val_buf));	\
 	seq_show_option(m, name, val_buf);		\
 }

-Kees

Signature · 1 line
Kees Cook
#2 lore raw
Quoted 1 line · Kees Cook

On Mon, Jul 17, 2023 at 12:33:32PM +0300, Andy Shevchenko wrote:

...

Quoted 1 line

I think this doesn't need to use seq_show_option_n() at all.

Quite likely. Nevertheless, it's one of the dozens (?) warnings like this.

...

Quoted 1 line

Thank you for the review!

I think it's you who may take it as seq_file.h seems everybody's playground.

Signature · 2 lines
With Best Regards,
Andy Shevchenko
#3 lore raw
Quoted 9 lines · Andy Shevchenko
Quoted 1 line · Kees Cook

On Mon, Jul 17, 2023 at 12:33:32PM +0300, Andy Shevchenko wrote:

...

Quoted 1 line

I think this doesn't need to use seq_show_option_n() at all.

Quite likely. Nevertheless, it's one of the dozens (?) warnings like this.

Quoted 7 lines

...

Quoted 1 line

Thank you for the review!

I think it's you who may take it as seq_file.h seems everybody's playground.

Ah, good point. :P

Signature · 1 line
Kees Cook
Applied
#4 lore raw
In reply to Andy Shevchenko
Quoted 6 lines · Andy Shevchenko

Privided seq_show_option_n() macro breaks build with -Werror and W=1, e.g.:

In function ‘strncpy’,
inlined from ‘ocfs2_show_options’ at fs/ocfs2/super.c:1520:3:
include/linux/fortify-string.h:68:33: error: ‘__builtin_strncpy’ output may be truncated copying 4 bytes from a string of length 4 [-Werror=stringop-truncation]

Quoted 8 lines
   68 | #define __underlying_strncpy    __builtin_strncpy
      |                                 ^
include/linux/fortify-string.h:151:16: note: in expansion of macro ‘__underlying_strncpy’
  151 |         return __underlying_strncpy(p, q, size);
      |                ^~~~~~~~~~~~~~~~~~~~

cc1: all warnings being treated as errors

[...]

Applied, thanks!

[1/1] seq_file: Replace strncpy()+nul by strscpy()

      https://git.kernel.org/kees/c/c30417b20f49

Best regards,

Signature · 1 line
Kees Cook
#5 lore raw
Quoted 15 lines · Kees Cook
Quoted 13 lines · Andy Shevchenko

Privided seq_show_option_n() macro breaks build with -Werror and W=1, e.g.:

In function ‘strncpy’,
inlined from ‘ocfs2_show_options’ at fs/ocfs2/super.c:1520:3:
include/linux/fortify-string.h:68:33: error: ‘__builtin_strncpy’ output may be truncated copying 4 bytes from a string of length 4 [-Werror=stringop-truncation]

   68 | #define __underlying_strncpy    __builtin_strncpy
      |                                 ^
include/linux/fortify-string.h:151:16: note: in expansion of macro ‘__underlying_strncpy’
  151 |         return __underlying_strncpy(p, q, size);
      |                ^~~~~~~~~~~~~~~~~~~~

cc1: all warnings being treated as errors

Quoted 6 lines
Quoted 1 line

[...]

Applied, thanks!

[1/1] seq_file: Replace strncpy()+nul by strscpy()

      https://git.kernel.org/kees/c/c30417b20f49

Gah, I dropped this from my tree since it was actually wrong[1]. This is an ugly corner case with strscpy vs strncpy: the cast be32 from hfs/hfsplus[2] looks unterminated to strscpy, so it would return -E2BIG, but really FORTIFY noticed the over-read (strscpy is correctly checking the 5th byte for NUL).

So... I think we need to fix seq_show_option_n() using memcpy+NUL, drop the ocfs2 usage, and clarify that the seq_show_option_n() docs mean "n means _exactly_ n bytes"...

-Kees

[1] https://lore.kernel.org/lkml/[email protected]/
[2] https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/fs/hfsplus/options.c?h=v6.4#n221

Signature · 1 line
Kees Cook
#6 lore raw
Quoted 28 lines · Kees Cook
Quoted 14 lines · Kees Cook
Quoted 13 lines · Andy Shevchenko

Privided seq_show_option_n() macro breaks build with -Werror and W=1, e.g.:

In function ‘strncpy’,
inlined from ‘ocfs2_show_options’ at fs/ocfs2/super.c:1520:3:
include/linux/fortify-string.h:68:33: error: ‘__builtin_strncpy’ output may be truncated copying 4 bytes from a string of length 4 [-Werror=stringop-truncation]

   68 | #define __underlying_strncpy    __builtin_strncpy
      |                                 ^
include/linux/fortify-string.h:151:16: note: in expansion of macro ‘__underlying_strncpy’
  151 |         return __underlying_strncpy(p, q, size);
      |                ^~~~~~~~~~~~~~~~~~~~

cc1: all warnings being treated as errors

Quoted 6 lines
Quoted 1 line

[...]

Applied, thanks!

[1/1] seq_file: Replace strncpy()+nul by strscpy()

      https://git.kernel.org/kees/c/c30417b20f49

Gah, I dropped this from my tree since it was actually wrong[1]. This is an ugly corner case with strscpy vs strncpy: the cast be32 from hfs/hfsplus[2] looks unterminated to strscpy, so it would return -E2BIG, but really FORTIFY noticed the over-read (strscpy is correctly checking the 5th byte for NUL).

Quoted 3 lines

So... I think we need to fix seq_show_option_n() using memcpy+NUL, drop the ocfs2 usage, and clarify that the seq_show_option_n() docs mean "n means _exactly_ n bytes"...

Sounds like a plan!

Please go ahead with that. My intention is to make eventually build kernel with make W=1 when CONFIG_WERROR=y.

Quoted 2 lines

[1] https://lore.kernel.org/lkml/[email protected]/
[2] https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/fs/hfsplus/options.c?h=v6.4#n221

Signature · 2 lines
With Best Regards,
Andy Shevchenko
#7 lore raw
In reply to Andy Shevchenko

From: Andy Shevchenko

Quoted 1 line

Sent: 17 July 2023 10:34

...

include/linux/seq_file.h static inline void seq_show_option(struct seq_file *m, const char *name, · line 265

Quoted 7 lines

#define seq_show_option_n(m, name, value, length) { \

include/linux/seq_file.h +1 -2
 	char val_buf[length + 1];			\
-	strncpy(val_buf, value, length);		\
-	val_buf[length] = '\0';				\
+	strscpy(val_buf, value, sizeof(val_buf));	\
 	seq_show_option(m, name, val_buf);		\
 }

Why the extra double-copy with (potentially) a VLA?

seq_show_option() calls seq_escape_str(), seq_escape_str calls seq_escape_mem(... strlen(src) ...).

Implement seq_show_option() as seq_show_option_n(... strlen(value)) and use seq_escape_mem() for the value.

David

-
Registered Address Lakeside, Bramley Road, Mount Farm, Milton Keynes, MK1 1PT, UK Registration No: 1397386 (Wales)