[Python-bugs-list] [ python-Bugs-553000 ] "./configure" crashes
noreply@sourceforge.net
noreply@sourceforge.net
Mon, 13 May 2002 14:44:13 -0700
Bugs item #553000, was opened at 2002-05-06 22:27
You can respond by visiting:
http://sourceforge.net/tracker/?func=detail&atid=105470&aid=553000&group_id=5470
Category: Build
Group: Platform-specific
Status: Closed
Resolution: Wont Fix
Priority: 5
Submitted By: Jarek Sacha (jsacha)
Assigned to: Nobody/Anonymous (nobody)
>Summary: "./configure" crashes
Initial Comment:
I am trying to built Python 2.2.1. I am getting a core
dump while running "./configure" on IRIX 6.5.
Last messages before the crash:
checking LDSHARED... ld -shared -all
checking CCSHARED... -shared
checking LINKFORSHARED...
checking CFLAGSFORSHARED...
checking for dlopen in -ldl... yes
checking for shl_load in -ldld... no
checking for --with-libs... no
checking for --with-signal-module... yes
checking for --with-dec-threads... no
checking for --with-threads... yes
checking for _POSIX_THREADS in unistd.h... yes
checking for mach/cthreads.h... no
checking for --with-pth... no
checking for pthread_create in -lpthread... yes
checking if PTHREAD_SCOPE_SYSTEM is supported... no
checking for pthread_sigmask... yes
checking for usconfig in -lmpc... no
checking for thr_create in -lthread... no
checking if --enable-ipv6 is specified... Segmentation
fault (core dumped)
Please fing attached "config.log"
Regards,
Jarek Sacha
----------------------------------------------------------------------
>Comment By: Jack Jansen (jackjansen)
Date: 2002-05-13 23:44
Message:
Logged In: YES
user_id=45365
I think you're misreading Guido's closing remark here, and it should be read as "there's nothing more the Python community can do for you". There's a workaround, and if you want the bug to be fixed you should pass it on to either the autoconf developers or SGI.
----------------------------------------------------------------------
Comment By: Jarek Sacha (jsacha)
Date: 2002-05-13 22:32
Message:
Logged In: YES
user_id=32232
there's nothing that Python can do for me...
Oh well, back to Java.
----------------------------------------------------------------------
Comment By: Guido van Rossum (gvanrossum)
Date: 2002-05-13 19:32
Message:
Logged In: YES
user_id=6380
Closing the bug report, as there's nothing that Python can
do for you.
----------------------------------------------------------------------
Comment By: Jarek Sacha (jsacha)
Date: 2002-05-13 19:19
Message:
Logged In: YES
user_id=32232
Looks that there is a workaround. I installed bash then
edited first line of 'configure' to use bash instead of
/bin/sh. Now configuration works. The only problem is that
bash 'configure' crashes too. So, this is only a solution if
you can get bash binaries.
----------------------------------------------------------------------
Comment By: Jarek Sacha (jsacha)
Date: 2002-05-13 18:40
Message:
Logged In: YES
user_id=32232
"file core" outputs:
core: IRIX N32 core dump of 'configure'
"versions eoe.sw.base" outputs
I eoe 03/05/2002 IRIX Execution
Environment, 6.5.15f
I eoe.sw 03/05/2002 IRIX Execution
Environment Software
I eoe.sw.base 03/05/2002 IRIX Base Execution
Environment
I am using /bin/sh that comes with the system (I am not the
system administrator). Looks to me that this is a problem
with /bin/sh. I tried to remove from configure whole block
that tests ipv6 but this just crushes configure same way on
the next test.Tried to put "echo" commands in side and
outside the in line 4360 (where the script crushes). Last
echo prints just before if, never inside, or after. As is
/bin/sh crashes while parsing the script(?).
----------------------------------------------------------------------
Comment By: Sjoerd Mullender (sjoerd)
Date: 2002-05-13 17:30
Message:
Logged In: YES
user_id=43607
The error message indicates that it is the shell that runs
the configure script that crashes. Otherwise the script
would continue (and possibly give other error messages).
Try "file core", this should tell you which program produced
the core (I expect configure). If this is indeed the case,
the bug is not in Python but in the shell.
Do you have a non-standard /bin/sh?
Which version of IRIX 6.5 are you running (try "versions
eoe.sw.base")?
----------------------------------------------------------------------
Comment By: Guido van Rossum (gvanrossum)
Date: 2002-05-13 16:47
Message:
Logged In: YES
user_id=6380
I see no reason why this should be priority 8, since all
evidence points towards the platform as the culprit. (IRIX
is notorious, though this is a new one.)
----------------------------------------------------------------------
Comment By: Michael Hudson (mwh)
Date: 2002-05-10 10:41
Message:
Logged In: YES
user_id=6656
Is there some way to tell which executable dumped the core
file? It'd be "size core" on linux, dunno what on IRIX.
If it's /bin/sh, I'm not really sure what we can do...
----------------------------------------------------------------------
Comment By: Jarek Sacha (jsacha)
Date: 2002-05-09 20:18
Message:
Logged In: YES
user_id=32232
I tried running with --disable-ipv6 and got exactly the same
crash. I edited 'configure' and added extra instructions to
have some idea what is happening. I modified fragment at
line 4361 from:
============================================
if test "${enable_ipv6+set}" = set; then
enableval="$enable_ipv6"
case "$enableval" in
no)
echo "$ac_t""no" 1>&6
ipv6=no
;;
*) echo "$ac_t""yes" 1>&6
cat >> confdefs.h <<\EOF
#define ENABLE_IPV6 1
EOF
ipv6=yes
;;
esac
else
============================================
to
============================================
if test "${enable_ipv6+set}" = set; then
echo "l. 4362"
else
echo "l. 4364"
fi
if test "${enable_ipv6+set}" = set; then
echo "l. 4367"
enableval="$enable_ipv6"
case "$enableval" in
no)
echo "$ac_t""no" 1>&6
ipv6=no
;;
*) echo "$ac_t""yes" 1>&6
cat >> confdefs.h <<\EOF
#define ENABLE_IPV6 1
EOF
ipv6=yes
;;
esac
else
echo "l. 4383"
===========================================
The interesting part is that "l.4362" or "l.4364" are
printed depending on option "--disable-ipv6" being used, as
expected. However, the program crashes befere either
"l.4367" or "l.4383" is printed. For instance if I run
"./configure --disable-ipv6" the last lines of output are:
checking for usconfig in -lmpc... no
checking for thr_create in -lthread... no
checking if --enable-ipv6 is specified... l. 4362
Segmentation fault (core dumped)
The modified config is attached. Maybe there is some problem
with interpreting the 'configure' itself?
----------------------------------------------------------------------
Comment By: Jack Jansen (jackjansen)
Date: 2002-05-08 10:45
Message:
Logged In: YES
user_id=45365
The test for IPv6 support on your machine actually tries to run a program that uses IPv6. So, could it be that you have IPv6 support partially installed? For instance, you have the headers (does AF_INET6 occur anywhere in <sys/socket.h> or any flie it includes?) but not the corresponding C library.
This guess is easy to check: try a "./configure --disable-ipv6". If that works fine then the oabove is your problem, probably, and you also have a workaround:-)
----------------------------------------------------------------------
Comment By: Jarek Sacha (jsacha)
Date: 2002-05-08 05:42
Message:
Logged In: YES
user_id=32232
I tried "--without-gcc" and getting crash the same point.
Tried to use --verbose option, but did not get extra
information.
Do you know if there is a way to figure out what action is
made by 'configure' just before the crash
----------------------------------------------------------------------
Comment By: Jack Jansen (jackjansen)
Date: 2002-05-07 22:31
Message:
Logged In: YES
user_id=45365
I can't repeat this, but I don't have gcc installed (and from your config.log it appears that you do).
Try building with configure --without-gcc.
There have always been problems with gcc on Irix, which is why I've stayed away from it, but I'm not sure whether building Python with gcc on Irix is supported (I found no notes to the contrary). So, I'm leaving this report open and someone else can pick it up, if they like.
----------------------------------------------------------------------
You can respond by visiting:
http://sourceforge.net/tracker/?func=detail&atid=105470&aid=553000&group_id=5470