Skip to contentSkip to conversation

fsnotify: avoid gcc-10 zero-length-bounds warning

PATCH In review Arnd Bergmann 4 participants 7 messages May 5, 2020 – May 7, 2020 VFS

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 fsnotify: avoid gcc-10 zero-length-bounds warning

gcc-10 warns about accesses into the f_handle[] zero-length array.

fs/notify/fdinfo.c: In function 'show_mark_fhandle':
fs/notify/fdinfo.c:66:47: error: array subscript 'i' is outside the bounds of an interior zero-length array 'unsigned char[0]' [-Werror=zero-length-bounds]

   66 |   seq_printf(m, "%02x", (int)f.handle.f_handle[i]);
      |                              ~~~~~~~~~~~~~~~~~^~~
In file included from fs/notify/fdinfo.c:3:
include/linux/fs.h:988:16: note: while referencing 'f_handle'
  988 |  unsigned char f_handle[0];
      |                ^~~~~~~~

This is solved by using a flexible array instead.

Cc Gustavo A. R. Silva <[email protected]>

---
Gustavo has done the same thing as part of a treewide change, but keeping this separate lets us backport it to stable kernels more easily later.

---
 include/linux/fs.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
include/linux/fs.h +1 -1
index 8690dc56e883..b229c55a8232 100644
@@ -986,7 +986,7 @@ struct file_handle {
 	__u32 handle_bytes;
 	int handle_type;
 	/* file identifier */
-	unsigned char f_handle[0];
+	unsigned char f_handle[];
 };
 
 static inline struct file *get_file(struct file *f)
Signature · 1 line
2.26.0
#1 lore raw
Quoted 11 lines · Arnd Bergmann

gcc-10 warns about accesses into the f_handle[] zero-length array.

fs/notify/fdinfo.c: In function 'show_mark_fhandle':
fs/notify/fdinfo.c:66:47: error: array subscript 'i' is outside the bounds of an interior zero-length array 'unsigned char[0]' [-Werror=zero-length-bounds]

   66 |   seq_printf(m, "%02x", (int)f.handle.f_handle[i]);
      |                              ~~~~~~~~~~~~~~~~~^~~
In file included from fs/notify/fdinfo.c:3:
include/linux/fs.h:988:16: note: while referencing 'f_handle'
  988 |  unsigned char f_handle[0];
      |                ^~~~~~~~
Quoted 7 lines

This is solved by using a flexible array instead.

Cc Gustavo A. R. Silva <[email protected]>

---
Gustavo has done the same thing as part of a treewide change, but keeping this separate lets us backport it to stable kernels more easily later.

Arnd,

I wonder why would we need to backport these changes to -stable... merely because of the use of a new version of GCC?

Thanks
--
Gustavo

#2 lore raw
Quoted 20 lines · Gustavo A. R. Silva
Quoted 11 lines · Arnd Bergmann

gcc-10 warns about accesses into the f_handle[] zero-length array.

fs/notify/fdinfo.c: In function 'show_mark_fhandle':
fs/notify/fdinfo.c:66:47: error: array subscript 'i' is outside the bounds of an interior zero-length array 'unsigned char[0]' [-Werror=zero-length-bounds]

   66 |   seq_printf(m, "%02x", (int)f.handle.f_handle[i]);
      |                              ~~~~~~~~~~~~~~~~~^~~
In file included from fs/notify/fdinfo.c:3:
include/linux/fs.h:988:16: note: while referencing 'f_handle'
  988 |  unsigned char f_handle[0];
      |                ^~~~~~~~
Quoted 7 lines

This is solved by using a flexible array instead.

Cc Gustavo A. R. Silva <[email protected]>

---
Gustavo has done the same thing as part of a treewide change, but keeping this separate lets us backport it to stable kernels more easily later.

Quoted 4 lines

Arnd,

I wonder why would we need to backport these changes to -stable... merely because of the use of a new version of GCC?

Yes, we usually backport trivial warning fixes to stable kernels to allow building those with any modern compiler version.

Arnd

#3 lore raw
Quoted 22 lines · Arnd Bergmann
Quoted 20 lines · Gustavo A. R. Silva
Quoted 18 lines · Arnd Bergmann

gcc-10 warns about accesses into the f_handle[] zero-length array.

fs/notify/fdinfo.c: In function 'show_mark_fhandle':
fs/notify/fdinfo.c:66:47: error: array subscript 'i' is outside the bounds of an interior zero-length array 'unsigned char[0]' [-Werror=zero-length-bounds]

   66 |   seq_printf(m, "%02x", (int)f.handle.f_handle[i]);
      |                              ~~~~~~~~~~~~~~~~~^~~
In file included from fs/notify/fdinfo.c:3:
include/linux/fs.h:988:16: note: while referencing 'f_handle'
  988 |  unsigned char f_handle[0];
      |                ^~~~~~~~

This is solved by using a flexible array instead.

Cc Gustavo A. R. Silva <[email protected]>

---
Gustavo has done the same thing as part of a treewide change, but keeping this separate lets us backport it to stable kernels more easily later.

Quoted 8 lines
Quoted 4 lines

Arnd,

I wonder why would we need to backport these changes to -stable... merely because of the use of a new version of GCC?

Yes, we usually backport trivial warning fixes to stable kernels to allow building those with any modern compiler version.

OK. So, if you anticipate that this is going to happen, I can split up my treewide patch into separate per-subsystem patches. I can replace the treewide patch in my tree today, so the changes are reflected in tomorrow's linux-next.

--
Gustavo

#4 lore raw
Quoted 7 lines · Gustavo A. R. Silva
Quoted 6 lines · Arnd Bergmann
Quoted 3 lines · Gustavo A. R. Silva

On 5/5/20 09:30, Arnd Bergmann wrote:
I wonder why would we need to backport these changes to -stable... merely because of the use of a new version of GCC?

Quoted 8 lines
Quoted 3 lines

Yes, we usually backport trivial warning fixes to stable kernels to allow building those with any modern compiler version.

OK. So, if you anticipate that this is going to happen, I can split up my treewide patch into separate per-subsystem patches. I can replace the treewide patch in my tree today, so the changes are reflected in tomorrow's linux-next.

I only needed a few patches to address all the warnings, so you don't need to split up the patch for this purpose, though it may be easier to get it merged anyway.

I see now that Linus has already applied the same fix as part of
https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=9d82973e032e2
It's just not yet in today's linux-next, but my patch is now obsolete.

Linus, let me know if you would like me to Cc you on the other gcc-10 warning fixes I have and possibly apply some directly. I have patches for all gcc-10 and clang-10 warnings now, and am in the process of getting them out to the subsystem maintainers.

Arnd

#5 lore raw
Quoted 3 lines · Arnd Bergmann

Linus, let me know if you would like me to Cc you on the other gcc-10 warning fixes I have and possibly apply some directly.

Sure. If you have any of the "trivially correct, and doesn't make code look worse", push them my way.

I only did the ones that looked trivial and fairly core - didn't want to step on any driver toes etc.

Linus

#6 lore raw
In reply to Arnd Bergmann

From: Arnd Bergmann

Quoted 1 line

Sent: 05 May 2020 16:00

...

Quoted 2 lines

Yes, we usually backport trivial warning fixes to stable kernels to allow building those with any modern compiler version.

In this case wouldn't it be better to backport a change that disables the specific compiler warning?

David

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