ICE in optimized build of 20000103

[Martin v. Loewis]

> http://gcc.gnu.org/ml/gcc-bugs/1999-12/msg00242.html

Some questions about that. How come 

-march=pentium -funroll-loops -fomit-frame-pointer -fstrict-aliasing

is passed to the compiler? I can't find that in the sources, and it
does not happen for me. Did you pass some options to configure that
you did not report? Still, cc1 does not crash for me even with these
options...

Also, what is the output of 'gcc -v' when compiling libgcc2.c with all
these options?

> Also, while I don't follow the nightly cvs updates, I was reporting
> on sources that where just updated with cvs update, is there another
> tree that I should be following?

No. In one of your reports, you mentioned you use snapshots, as you
had to erase your CVS trees. In the other reports, I could not find an
indication what source code you were using.

> P.S., maybe I'm not seeing it, but why should the binutils /ld
> version be an issue at this stage?

It normally shouldn't matter. There might be some weird bug in
binutils that corrupts the cc1 executable. Also, there might be a
hardware error somewhere. Did you try to reproduce the problem on a
different machine?

Anyway, as I cannot reproduce the problem, I guess I cannot do much
about it. If you really need to solve it, I recommend to get hold of
the cc1 command line, and analyse the crash in a debugger. Otherwise,
you'd just have to stick with the released compilers.

Regards,
Martin