The following reply was made to PR gdb/2188; it has been noted by GNATS.
From: Greg Watson <g.watson@xxxxxxxxxxxx>
To: Daniel Jacobowitz <drow@xxxxxxxxx>
Cc: Clement Chu <clement.chu@xxxxxxxxxxxxxxxxxxxxxx>,
gdb-gnats@xxxxxxxxxxxxxxxxxx
Subject: Re: gdb/2188: gdb 6.5 -- internal error with "-var-update *" command
Date: Sat, 13 Jan 2007 22:36:42 -0700
Daniel,
CVS doesn't build on FC6. I get a whole lot of warnings like "p-
valprint.c: warning: 'tmp_objstack.alloc_failed' may be used
uninitialized" which cause the build to fail. However I did manage to
build one of the 6.6 snapshots. Using this version I no longer see
the assertion for the argc issue, so it appears to have been fixed in
the later release.
BTW, I've confirmed that the bug is present when compiling with gcc
4.1.0, but not gcc 4.0.3, so it appears to be a 4.1 bug.
Regards,
Greg
On Jan 10, 2007, at 7:00 PM, Daniel Jacobowitz wrote:
> On Wed, Nov 22, 2006 at 01:59:02AM +1100, Clement Chu wrote:
>> Hi Daniel,
>>
>> The following is the message from gdb. You can see my step and
>> the
>> error message. Please find the attached file.
>
> This is caused by a bug in GCC. The debug information claims the
> variables are at 0(%ecx) and 4(%ecx) due to the stack realignment code
> added to the beginning of main (not added to any other functions, just
> main).
>
> As for the failure, I do not get an internal error on either gdb
> 6.5 or
> gdb from CVS. Please try building CVS. I know that the code where
> you
> are encountering an assertion was recently improved.
>
> --
> Daniel Jacobowitz
> CodeSourcery
|
