This is the mail archive of the gcc-bugs@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]

Re: today's flow.c change breaks build.


Robert Lipe wrote:
> 
> Jean-Pierre Radley wrote:
> 
> > I updated the source tree, again, and now, bootstrapping a plain (i.e.,
> > not a udk-enabled) gcc, yet another failure:
> > 
> >         ./xgcc -B/usr/local/i586--sco3.2v5/bin/ -B./ -isystem /usr/local/i586--sco3.2v5/include -O2   -DIN_GCC    -g -isystem ./include   -g1 -DIN_LIBGCC2 -D__GCC_FLOAT_NOT_NEEDED  -I. -I/s/tools/egcs/gcc -I/s/tools/egcs/gcc/config -I/s/tools/egcs/gcc
/../include  -DL_muldi3 -c /s/tools/egcs/gcc/libgcc2.c -o libgcc/./_
> > 	muldi3.o
> > 	/s/tools/egcs/gcc/libgcc2.c: In function `__muldi3':
> > 	/s/tools/egcs/gcc/libgcc2.c:204: Internal compiler error in `?', at emit-rtl.c:2514
> > 	Please submit a full bug report.
> 
> The culprit seems to be this change in flow.c.  I don't know why it
> fails, but with this lone file reverted one version, the build works
> again.
> 
> revision 1.281
> date: 2000/05/09 18:41:25;  author: denisc;  state: Exp;  lines: +2 -1
>         * flow.c (flow_delete_insn): Use INSN_DELETED_P for marking insn
>         as deleted.

I don't think this is the problem.

I think the problem is caused by the test which seems a little fishy?
I don't see why it's illegal to add an insn after a delete insn.
Actually, the rtx which is triggering the abort isn't even a deleted insn;
it's actually a basic block note which has /v set, and somehow it's
recognized as a delete insn.

Toshi

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]