[Bug c/55896] New: Annoying compiler warning

[Bug c/55896] New: Annoying compiler warning

[sworddragon2 at aol dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55896

             Bug #: 55896
           Summary: Annoying compiler warning
    Classification: Unclassified
           Product: gcc
           Version: 4.7.2
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: c
        AssignedTo: unassigned@gcc.gnu.org
        ReportedBy: sworddragon2@aol.com


Created attachment 29097
  --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=29097
Example code

After changing a global struct to a static variable I got a compiler warning. I
have analyzed this and tried to produce a minimum testcase. In the attachments
is an example code and compiling it with "gcc -O3 -Wall -Wextra -o test
-pedantic test.c" will result in the following warning:

In file included from /usr/include/string.h:638:0,
                 from test.c:3:
In function 'strncat',
    inlined from 'load_app_settings' at test.c:39:10:
/usr/include/x86_64-linux-gnu/bits/string3.h:150:3: warning: call to
__builtin___strncat_chk might overflow destination buffer [enabled by default]

Removing static from "static struct status STATUS;" solves the problem. Any
random change in load_app_settings() can solve the problem too even if the
change has nothing to do with the problem. For example if "strncat(content_2,
content_1, w_strlen(content_1) - 1);" is changed to "strncat(content_2,
content_1, w_strlen(content_1));" or " && content_1[0] == '['" is removed in
the condition.

[Bug c/55896] Annoying compiler warning

[mpolacek at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55896

Marek Polacek <mpolacek at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |mpolacek at gcc dot gnu.org

--- Comment #1 from Marek Polacek <mpolacek at gcc dot gnu.org> ---
I think there's a bug in the program.  We inline w_strlen function.  Now, if
buffer is NULL, we don't compute w_strlen, as the result is 0.  But since the
call to strncat is strncat(content_2, content_1, w_strlen(content_1) - 1);, we
pass -1 as n to strncat, which is casted to size_t, that means we end up with
__builtin___strncat_chk (&content_2, &content_1, 18446744073709551615, 1024);.

[Bug c/55896] Annoying compiler warning

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55896

Jakub Jelinek <jakub at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |jakub at gcc dot gnu.org

--- Comment #2 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
But there is  && content_1[0] == '[' guard around it, but likely just no
optimization figures out that the inlined w_strlen (why aren't you using
strlen???) will always return at least 1 and thus the jump threaded variant
with zero length is dead.  I suppose instead of using w_strlen (contents_1) - 1
you could just use w_strlen (contents_1 + 1) and the warning could go away.

[Bug c/55896] Annoying compiler warning

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55896

--- Comment #4 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
(In reply to sworddragon2 from comment #3)
> > why aren't you using strlen???
> 
> Because of this bug (if it should be still valid):
> https://bugs.launchpad.net/ubuntu/+source/gcc-4.7/+bug/1035321

That is a valgrind issue, not a GCC bug though.

[Bug c/55896] Annoying compiler warning

[mpolacek at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55896

--- Comment #5 from Marek Polacek <mpolacek at gcc dot gnu.org> ---
(In reply to Jakub Jelinek from comment #2)
> you could just use w_strlen (contents_1 + 1) and the warning could go away.

This causes that two warnings are emitted :(.  If strlen can't be used (ugh),
then e.g. if (w_strlen(content_1) != 0) before strncat makes the warning go
away.