New bootstrap failure on ARM systems
Jan Hubicka
jh@suse.cz
Wed Jan 7 11:30:00 GMT 2004
> >
> > > > stage1/xgcc -Bstage1/ -B/work/rearnsha/gnu/testinstall/arm-netbsdelf/bin/
> > > > -c -O2 -g -mcpu=strongarm -DIN_GCC -W -Wall -Wwrite-strings
> > > > -Wstrict-prototypes -Wmissing-prototypes -pedantic -Wno-long-long
> > > > -Wold-style-definition -Werror -DHAVE_CONFIG_H -DGENERATOR_FILE -I.
> > > > -I. -I/work/rearnsha/gnusrc/gcc/gcc -I/work/rearnsha/gnusrc/gcc/gcc/.
> > > > -I/work/rearnsha/gnusrc/gcc/gcc/../include -I../intl print-rtl1.c -o
> > > > print-rtl1.o
> > > > print-rtl1.c: In function `print_rtx':
> > > > print-rtl1.c:616: internal compiler error: in verify_local_live_at_start,
> > > > at flow.c:546
> > >
> > > Hmm, that sounds like problem with conditional execution and
> > > bbreordering. This may be the reason why things were ordered the other
> > > way around than it would be natural. Do you have some kind of testcase?
> > >
>
> Is this going to be fixed soon, or would you rather that I reverted the
> patch? It's been impossible to bootstrap the trunk on ARM for nearly a
> week now, which at this stage the development process really isn't an
> acceptable situation.
Can I have a testcase for the problem? I can surely reorder the passes
for targets with conditional execution, but I would rather try to fix it
more correctly.
Honza
>
> R.
>
More information about the Gcc-patches
mailing list