glibc + gcc-3.2 problems

[Andre C. van Veen]

http://bugs.gnu.org/cgi-bin/gnatsweb.pl?cmd=view%20audit-trail&pr=4489&database=default

and also

http://bugs.gnu.org/cgi-bin/gnatsweb.pl?cmd=view%20audit-trail&pr=4671&database=default

Hi to all out there

I can easily reproduce the problem described in problem report #4489.
For the moment you will get a working glibc when you compile the same
source with gcc-2.95.3. However this is an inermediate solution as you
will be forced to use gcc-3.2 from glibc-2.3 on !!!!

>From my bitter experience with a similar system (i686, linux 2.4.19), I
can not recommend any installation attempts of the glibc-2.3.x (x=0,1)
series compiled with '-fomit-frame-pointer'. I have been already down
for 1 day with this nice (REALLY risky) one. If you like to have a lot
of fun just try to get glibc-2.3.X to run (but download and burn an
iso-image of an emergency system, e.g. suse liveeval, first ;) ).

In view of problem report #4671, I agree on the report that the pthreads
are broken (none of the tests works correct) when '-fomit-frame-pointer'
is used, both for glibc-2.3 and also in the new release 2.3.1. This
might easily explain why the system is not even booting up anymore (end
of story, when it attempts to execute something like the scripts after
starting init), but I also get an error at the test in string. However,
this last one might have other reasons:

snip....

make[2]: *** [<<PATH>>/compile/string/tester.out] Error 1
make[2]: Leaving directory `<<PATH>>/glibc-2.3.1/string'
make[1]: *** [string/tests] Error 2
make[1]: Leaving directory `<<PATH>>/glibc-2.3.1'
make: *** [check] Error 2

and tester.out...

stpcpy flunked test 1
1 errors.

...snap

I can not decide if this a gcc or a glibc problem, but in any case I
think it is a bug.

regards,
Andre