This is the mail archive of the
gcc-bugs@gcc.gnu.org
mailing list for the GCC project.
Re: egcs1.1.1 bogus warning: "might be used uninitialized"
- To: "Brad M. Garcia" <bgarcia at fore dot com>
- Subject: Re: egcs1.1.1 bogus warning: "might be used uninitialized"
- From: Jeffrey A Law <law at hurl dot cygnus dot com>
- Date: Wed, 10 Feb 1999 16:58:06 -0700
- cc: d-dooling at nwu dot edu, egcs-bugs at egcs dot cygnus dot com
- Reply-To: law at cygnus dot com
In message <Pine.LNX.4.04.9902101547550.31679-100000@spud.eng.fore.com>you wr
ite:
> On Wed, 10 Feb 1999, Jeffrey A Law wrote:
>
> > Now, changing code in one function should not be changing the warnings gc
> c
> > emits for an unrelated function *unless* the function changed is actually
> > a template or is inlined. If neither of those cases are true, then there
> 's
> > a serious bug that needs to be addressed.
>
> In my case, this is plain C code, being compiled with gcc, not g++.
> They are not templates, and are not inlined.
>
> On the linux machine, both warnings were in the same function.
> On sunos, the 3rd additional warning was in an unrelated function.
>
> When I deleted some other, unrelated, functions from the file, the
> warnings went away! If instead, I initialized *one* of those
> variables, then *all* the warnings would again go away.
>
> So, yes, it looks like a serious bug. But luckily the only consequence
> appears to be bogus warning messages. That's still a pain though,
> since we always like to compile with -Wall -Werror.
Can you get me a testcase? This is *real* serious since it is a
non-deterministic problem -- which typically means memory corruption or a
dangling pointer.
It is critical that we nail those down as soon as they appear since they tend
to disappear without notice.
jeff