today's work

Andreas Jaeger aj@suse.de
Thu Aug 9 07:43:00 GMT 2001


Ulrich Drepper <drepper@redhat.com> writes:

> You might have seen that I checked in a substantial amount of changes
> today.  I get results from a test suite which mostly are correct in
> identifying problems.  It's unfortunate that it's so late but better
> than never.  Tomorrow I have another full day of debugging ahead of me
> and once this is done I fear we have enough changes to invalidate
> previous results.
>
> But since almost all the changes affect wide char streams and other
> wide char functions I doubt that another test release will be of much
> help.  Instead I think it's much better to simply have the
> distribution makers run their tests (and builds) and see whether the
> changes cause problems.

I just compiled and run make check on the following architectures:
- i686-linux
- sparc-linux (32-bit) (still running, I report if I notice problems)
- sparc-linux (64-bit) (still running, I report if I notice problems)
- alpha-linux
- ppc-linux
- s390-linux (31-bit)
- ia64-linux

The problems I noticed so far are:
On i686:
Err: Cannot run on the en_US.UTF-8 locale
make[2]: *** [/builds/glibc/main-gcc-2.95/libio/tst-ungetwc2.out] Error 1

I've committed the patch to generated the en_USE.UTF-8 locale.

On ia64:
In file included from test-tgmath.c:139:
test-tgmath.c: In function `compile_test':
test-tgmath.c:221: warning: implicit declaration of function `__tgmath_real_type'
test-tgmath.c:221: parse error before `__tgmres'

This looks like a compiler bug.

On Alpha:
pthread_create failed
make[2]: *** [/usr/src/packages/BUILD/glibc-2.2/cc/linuxthreads/ex7.out] Error 134
/usr/src/packages/BUILD/glibc-2.2/cc/linuxthreads/ex9: cannot create thread
make[2]: *** [/usr/src/packages/BUILD/glibc-2.2/cc/linuxthreads/ex9.out] Error 1
/usr/src/packages/BUILD/glibc-2.2/cc/linuxthreads/ex10: cannot create thread
make[2]: *** [/usr/src/packages/BUILD/glibc-2.2/cc/linuxthreads/ex10.out] Error 1
make[2]: *** [/usr/src/packages/BUILD/glibc-2.2/cc/linuxthreads/ex11.out] Error 1
make[2]: *** [/usr/src/packages/BUILD/glibc-2.2/cc/linuxthreads/ex14.out] Error 1

Seeems that threads are broken - but I have no time to investigate
today/tomorrow where the problem is (kernel, glibc, binutils or GCC).

On s390:
pthread_create failed
make[2]: *** [/usr/src/packages/BUILD/glibc-2.2/cc/linuxthreads/ex7.out] Error 134
make[2]: *** [/usr/src/packages/BUILD/glibc-2.2/cc/linuxthreads/ex11.out] Error 1
make[2]: *** [/usr/src/packages/BUILD/glibc-2.2/cc/linuxthreads/ex14.out] Error 1

Seems that threads are broken - but I have no time to investigate
today/tomorrow where the problem is.


I'd like to ask the alpha and s390 maintainers to double check that
the current glibc runs on their systems and passes the testsuite,

Andreas
-- 
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.0.6 (GNU/Linux)
Comment: For info see http://www.gnupg.org

iD8DBQE7cqGCOJpWPMJyoSYRAi/iAJ0WR4w2ZHN8lRTgRiNWzVucLpFTbACfRC4W
aJipRgxxqLSX0FffQNjb5ec=
=brHp
-----END PGP SIGNATURE-----From jakub@redhat.com Thu Aug 09 09:38:00 2001
From: Jakub Jelinek <jakub@redhat.com>
To: Andreas Jaeger <aj@suse.de>
Cc: Ulrich Drepper <drepper@cygnus.com>, GNU libc hacker <libc-hacker@sourceware.cygnus.com>, laroche@redhat.com
Subject: Re: today's work
Date: Thu, 09 Aug 2001 09:38:00 -0000
Message-id: <20010809184022.B532@sunsite.ms.mff.cuni.cz>
References: <m3elql62em.fsf@otr.mynet> <ho3d71xpj2.fsf@gee.suse.de>
X-SW-Source: 2001-08/msg00035.html
Content-length: 2028

On Thu, Aug 09, 2001 at 04:43:13PM +0200, Andreas Jaeger wrote:
> The problems I noticed so far are:
> On i686:
> Err: Cannot run on the en_US.UTF-8 locale
> make[2]: *** [/builds/glibc/main-gcc-2.95/libio/tst-ungetwc2.out] Error 1

Worked for me, dunno, guess because it found en_US.UTF-8 in /usr/lib/locale.
Both --enable-kernel=2.2.5 and --enable-kernel=2.4.1 i686 builds passed make
check completely.

> On ia64:
> In file included from test-tgmath.c:139:
> test-tgmath.c: In function `compile_test':
> test-tgmath.c:221: warning: implicit declaration of function `__tgmath_real_type'
> test-tgmath.c:221: parse error before `__tgmres'
> 
> This looks like a compiler bug.

Yep, looks like a compiler bug.
gcc-2.96-RH (particularly 2.96-95) passed full make check.
Are you using Jim Wilson's snapshot? Guess it is a preprocessor bug which is
fixed in 2.96-RH (the 2nd generation cpp lexer did not work well on this,
while current 3rd generation cpp lexer does (the one in 3.0 or 2.96-RH).

> On Alpha:
> pthread_create failed
> make[2]: *** [/usr/src/packages/BUILD/glibc-2.2/cc/linuxthreads/ex7.out] Error 134
> /usr/src/packages/BUILD/glibc-2.2/cc/linuxthreads/ex9: cannot create thread
> make[2]: *** [/usr/src/packages/BUILD/glibc-2.2/cc/linuxthreads/ex9.out] Error 1
> /usr/src/packages/BUILD/glibc-2.2/cc/linuxthreads/ex10: cannot create thread
> make[2]: *** [/usr/src/packages/BUILD/glibc-2.2/cc/linuxthreads/ex10.out] Error 1
> make[2]: *** [/usr/src/packages/BUILD/glibc-2.2/cc/linuxthreads/ex11.out] Error 1
> make[2]: *** [/usr/src/packages/BUILD/glibc-2.2/cc/linuxthreads/ex14.out] Error 1
> 
> Seeems that threads are broken - but I have no time to investigate
> today/tomorrow where the problem is (kernel, glibc, binutils or GCC).

Worked for me too. Alpha passed with just one problem - two tilde tests in
globtest.out failed, apparently because $HOME pointed to directory which was
not mounted in that particular buildroot.

Cannot check s390, Florian, can you check that (glibc-2.2.4-3.src.rpm)?

	Jakub



More information about the Libc-hacker mailing list