Re: m4-1.4.10 segfault with x86_64 during ./configure when checking whether printf supports infinite 'long double' arguments
Gilles Espinasse wrote:
It compile fine. I didn't look the result of tests.
The problem stay un-noticed for some times until I was looking at the
console screen during compilation.
So the only problem is that the configure test crashes? This is easy to
fix; I'm applying this patch:
2007-12-17 Bruno Haible [EMAIL PROTECTED]
* m4/printf.m4 (gl_PRINTF_INFINITE_LONG_DOUBLE): Use GL_NOCRASH to
avoid a segmentation fault of the configure test on x86_64 systems.
*** m4/printf.m4.orig 2007-12-17 10:18:35.0 +0100
--- m4/printf.m42007-12-17 10:17:58.0 +0100
***
*** 1,4
! # printf.m4 serial 20
dnl Copyright (C) 2003, 2007 Free Software Foundation, Inc.
dnl This file is free software; the Free Software Foundation
dnl gives unlimited permission to copy and/or distribute it,
--- 1,4
! # printf.m4 serial 21
dnl Copyright (C) 2003, 2007 Free Software Foundation, Inc.
dnl This file is free software; the Free Software Foundation
dnl gives unlimited permission to copy and/or distribute it,
***
*** 240,245
--- 240,246
[gl_cv_func_printf_infinite_long_double],
[
AC_TRY_RUN([
+ ]GL_NOCRASH[
#include float.h
#include stdio.h
#include string.h
***
*** 265,270
--- 266,272
static long double zeroL = 0.0L;
int main ()
{
+ nocrash_init();
if (sprintf (buf, %Lf, 1.0L / 0.0L) 0
|| (strcmp (buf, inf) != 0 strcmp (buf, infinity) != 0))
return 1;
___
Bug-m4 mailing list
Bug-m4@gnu.org
http://lists.gnu.org/mailman/listinfo/bug-m4
Re: m4-1.4.10 segfault with x86_64 during ./configure when checking whether printf supports infinite 'long double' arguments
On Thursday 13 December 2007, Eric Blake wrote: According to Gilles Espinasse on 12/12/2007 3:29 PM: Hi Gilles, -- System Information: Debian Release: 4.0 APT prefers stable APT policy: (500, 'stable') Architecture: amd64 (x86_64) Shell: /bin/sh linked to /bin/bash Kernel: Linux 2.6.18-5-amd64 Locale: LANG=fr_FR.UTF-8, LC_CTYPE=fr_FR.UTF-8 (charmap=UTF-8) m4-1.4.10 ./configure segfault on amd64 conftest[number vary]: segfault at 502000 rip 2b3b28099a7 rsp number vary error 6 I try this patch with no change http://lists.gnu.org/archive/html/bug-m4/2007-09/msg7.html On the same machine but with Etch 32b version, it just run with trouble (no test with the patch) Thanks for the report; I'm adding gnulib, since that would be where the fix (if any) should be made. I need to make an m4 snapshot with a more recent gnulib snapshot, to see if we've addressed this in the meantime. Meanwhile, was configure still coming up with the correct answer, and just printing bogus intermediate text, or was it failing to compile because of the segv? Perhaps attaching the complete config.log would be helpful. on Gentoo, the configure test fails, but otherwise things seem to build/run fine. the little test for infinity handling segfaults. so if the correct answer is obtained by interpreting the segfault the same way ... -mike signature.asc Description: This is a digitally signed message part. ___ Bug-m4 mailing list Bug-m4@gnu.org http://lists.gnu.org/mailman/listinfo/bug-m4
Re: m4-1.4.10 segfault with x86_64 during ./configure when checking whether printf supports infinite 'long double' arguments
-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 According to Gilles Espinasse on 12/12/2007 3:29 PM: Hi Gilles, -- System Information: Debian Release: 4.0 APT prefers stable APT policy: (500, 'stable') Architecture: amd64 (x86_64) Shell: /bin/sh linked to /bin/bash Kernel: Linux 2.6.18-5-amd64 Locale: LANG=fr_FR.UTF-8, LC_CTYPE=fr_FR.UTF-8 (charmap=UTF-8) m4-1.4.10 ./configure segfault on amd64 conftest[number vary]: segfault at 502000 rip 2b3b28099a7 rsp number vary error 6 I try this patch with no change http://lists.gnu.org/archive/html/bug-m4/2007-09/msg7.html On the same machine but with Etch 32b version, it just run with trouble (no test with the patch) Thanks for the report; I'm adding gnulib, since that would be where the fix (if any) should be made. I need to make an m4 snapshot with a more recent gnulib snapshot, to see if we've addressed this in the meantime. Meanwhile, was configure still coming up with the correct answer, and just printing bogus intermediate text, or was it failing to compile because of the segv? Perhaps attaching the complete config.log would be helpful. Gilles (I am not subscribed, please CC me) - -- Don't work too hard, make some time for fun as well! Eric Blake [EMAIL PROTECTED] -BEGIN PGP SIGNATURE- Version: GnuPG v1.4.5 (Cygwin) Comment: Public key at home.comcast.net/~ericblake/eblake.gpg Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org iD8DBQFHYe2784KuGfSFAYARAvSWAKCsaLuTfsfRhTvygQX8CHtGJSc02ACghCg0 4Z8l09P/EMDC+vBY8fGtgmY= =AVzD -END PGP SIGNATURE- ___ Bug-m4 mailing list Bug-m4@gnu.org http://lists.gnu.org/mailman/listinfo/bug-m4
Re: m4-1.4.10 segfault with x86_64 during ./configure when checking whether printf supports infinite 'long double' arguments
- Original Message - From: Eric Blake [EMAIL PROTECTED] To: Gilles Espinasse [EMAIL PROTECTED] Cc: bug-m4@gnu.org; bug-gnulib [EMAIL PROTECTED] Sent: Friday, December 14, 2007 3:43 AM Subject: Re: m4-1.4.10 segfault with x86_64 during ./configure when checking whether printf supports infinite 'long double' arguments -BEGIN PGP SIGNED MESSAGE- Hash: SHA1 According to Gilles Espinasse on 12/12/2007 3:29 PM: Hi Gilles, -- System Information: Debian Release: 4.0 APT prefers stable APT policy: (500, 'stable') Architecture: amd64 (x86_64) Shell: /bin/sh linked to /bin/bash Kernel: Linux 2.6.18-5-amd64 Locale: LANG=fr_FR.UTF-8, LC_CTYPE=fr_FR.UTF-8 (charmap=UTF-8) m4-1.4.10 ./configure segfault on amd64 conftest[number vary]: segfault at 502000 rip 2b3b28099a7 rsp number vary error 6 I try this patch with no change http://lists.gnu.org/archive/html/bug-m4/2007-09/msg7.html On the same machine but with Etch 32b version, it just run with trouble (no test with the patch) Thanks for the report; I'm adding gnulib, since that would be where the fix (if any) should be made. I need to make an m4 snapshot with a more recent gnulib snapshot, to see if we've addressed this in the meantime. Meanwhile, was configure still coming up with the correct answer, and just printing bogus intermediate text, or was it failing to compile because of the segv? Perhaps attaching the complete config.log would be helpful. It compile fine. I didn't look the result of tests. The problem stay un-noticed for some times until I was looking at the console screen during compilation. Nothing special appear in compilation log (m4 is only one of the numerous packages to compile a full distrib à la LSF). I have attached config.log. Gilles config.log.bz2 Description: Binary data ___ Bug-m4 mailing list Bug-m4@gnu.org http://lists.gnu.org/mailman/listinfo/bug-m4
