This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: Mainline bootstrap failure (Re: libstdc++ execute tests)
At 10:40 12.04.2001, Gabriel Dos Reis wrote:
>Neil Booth <neil@daikokuya.demon.co.uk> writes:
>
>| Neil Booth wrote:-
>|
>| > Given you think it was broken sometime on the 28th, I'm going to try
>| > 2001-03-29 00:00 and expect it to be broken. Could you try 2001-03-28
>| > 18:00?
>|
>| I can confirm that 2001-03-29 00:00 fails:-
>|
>| Bootstrap comparison failure!
>| fold-const.o differs
>| real.o differs
>| tree.o differs
>|
>| (--enable-languages=c). So we have it down to a 12-hour interval.
>
>The breakage seems to happen between 16:00 and 18:00. Currently I'm
>suspecting this patch:
>
>2001-03-28 DJ Delorie <dj@redhat.com>
>
> * Makefile.in (stage1_build): Revert CFLAGS patch.
If you have problems with this reversal, try the attached patch from my 3.0
queue which makes STAGE1_CFLAGS overrideable from the toplevel. Then a
make bootstrap STAGE1_CFLAGS="-O2 -g"
should give you the same setup as before the patch reversal. You can also
play with with different STAGE1_CFLAGS a bit to see when it breaks, maybe
we are chasing a -O0 miscompilation caused by some uninitialized
memory/flag/etc. in the -O0 case.
Franz.
ChangeLog:
* Makefile.in (STAGE1_CFLAGS): Pass down.
gcc/ChangeLog:
* Makefile.in (stage1_build): Pass down STAGE1_CFLAGS and MAKEINFO.
stage1_cflags.patch