Re: bug in bison 1.875

[Martin MOKREJŠ]

On Sat, 28 Feb 2003, Paul Eggert wrote:

> Martin MOKREJŠ <address@hidden> writes:
>
> > cc: Error: configure, line 4919: Bug found in compiler: 
> > generate_address_constant : Unexpected opcode. (compilerbug)
> >           bool e = &s;
> > -------------------^
> > Internal compiler error detected at line 1512 in file 
> > /usr/proj/decc2/ccdtk-v4-2/src/me_gemc/initgen.c.
> > cc: Fatal: A memory access violation (bus error or segmentation fault)
> > has occurred.  Please submit a problem report.
>
> That should be OK.  "configure" will notice that the compiler doesn't
> support "bool", and will cause Bison to not use "bool".
>
> > So I'd go for the test-code above or just check version output of the
> > compiler.
>
> I'd prefer a solution that actually reflects the problem, not some
> arbitrary thing like a version number, or a compiler that happens to
> fail with some seemingly unrelated property like "bool".
> Unfortunately we don't know what the real problem is yet.

But the bug really is in the compiler. Non -O2 optimized code jsut works
fine, the one with -O2 works sometimes bad.

>
> > Actually, I think I forgot to run make check. ;( A lot of them failed now.
>
> With as many test failures and compiler errors as you're reporting,
> I suspect that your compiler installation is broken somehow.

Actually, I've upgraded compiler recently to have better C99 support, and
so far the upgrade helped.

>
> > $ cc -V
> > Compaq C V6.5-207 (dtk) on Compaq Tru64 UNIX V5.1A (Rev. 1885)
> > Compiler Driver V6.5-207 (dtk) (dtk) cc Driver
>
> I just searched groups.google.com, and the only other mention I saw of
> this compiler indicated that it had serious internal errors.  So
> perhaps we should just chalk this up to a seriously bad compiler, and
> leave it at that.  See:
>
> http://groups.google.com/groups?selm=fa.j9bd2tb.1j1cm8l%40ifi.uio.no

I don't use 'spike' utility. It's a nice tools to speedup binaries, but it
has nothing to do with my compiler optimization bug.

Regarding those many bugs, which appear even when -O is used, I'm not sure
what to do. I'll try to downgrade and re-test.
-- 
Martin Mokrejs <address@hidden>, <address@hidden>
PGP5.0i key is at http://www.natur.cuni.cz/~mmokrejs
MIPS / Institute for Bioinformatics <http://mips.gsf.de>
GSF - National Research Center for Environment and Health
Ingolstaedter Landstrasse 1, D-85764 Neuherberg, Germany
tel.: +49-89-3187 3683 , fax: +49-89-3187 3585