Skip Menu |

This queue is for tickets about the Math-Pari CPAN distribution.

Report information
The Basics
Id: 1201
Status: new
Priority: 0/
Queue: Math-Pari

People
Owner: Nobody in particular
Requestors: selsky [...] columbia.edu
Cc:
AdminCc:

Bug Information
Severity: Normal
Broken in: 2.010305
Fixed in: (no value)



Subject: hiremainder symbol not found
I'm running perl5.6.1 on solaris8. I tried building the latest version of Math::Pari like this: perl Makefile.PL [fetch libPARI] make I don't get any errors. When I do the 'make test' I get this: PERL_DL_NONLAZY=1 /opt/local/bin/perl -Iblib/arch -Iblib/lib -I/opt/perl5.6.1/lib/5.6.1/sun4-solaris -I/opt/perl5.6.1/lib/5.6.1 -e 'use Test::Harness qw(&runtests $verbose); $verbose=0; runtests @ARGV;' t/*.t t/Pari........Can't load 'blib/arch/auto/Math/Pari/Pari.so' for module Math::Pari: ld.so.1: /opt/local/bin/perl: fatal: relocation error: file blib/arch/auto/Math/Pari/Pari.so: symbol hiremainder: referenced symbol not found at /opt/perl5.6.1/lib/5.6.1/sun4-solaris/DynaLoader.pm line 206. at t/Pari.t line 15 Compilation failed in require at t/Pari.t line 15. BEGIN failed--compilation aborted at t/Pari.t line 15. t/Pari........dubious Test returned status 255 (wstat 65280, 0xff00) This continues throughout the tests with the hiremainder symbol not being found. When I do an ldd on Pari.so, everything resolves. When I do nm on the so I get this: $ nm Pari.so | grep remain U hiremainder Any idea what the problem is? Please let me know if I can provide you with more information.
From: selsky [...] columbia.edu
I think there is a problem with the sparcv8micro assembler. When I specify to use C code instead of assembler, the hiremainder symbol is found. $ perl Makefile.PL machine=none I hope that helps.
Date: Sat, 27 Jul 2002 06:41:52 -0400
From: Ilya Zakharevich <ilya [...] math.ohio-state.edu>
To: Guest via RT <bug-Math-Pari [...] rt.cpan.org>
Cc: "'AdminCc of cpan Ticket #1201'": ;
Subject: Re: [cpan #1201] hiremainder symbol not found
On Thu, Jun 27, 2002 at 06:55:59PM -0400, Guest via RT wrote: Show quoted text
> PERL_DL_NONLAZY=1 /opt/local/bin/perl -Iblib/arch -Iblib/lib -I/opt/perl5.6.1/lib/5.6.1/sun4-solaris -I/opt/perl5.6.1/lib/5.6.1 -e 'use Test::Harness qw(&runtests $verbose); $verbose=0; runtests @ARGV;' t/*.t > t/Pari........Can't load 'blib/arch/auto/Math/Pari/Pari.so' for module Math::Pari: ld.so.1: /opt/local/bin/perl: fatal: relocation error: file blib/arch/auto/Math/Pari/Pari.so: symbol hiremainder: referenced symbol not found at /opt/perl5.6.1/lib/5.6.1/sun4-solaris/DynaLoader.pm line 206. > at t/Pari.t line 15
Looks like a cousin to the 'overflow' mentioned in README. What happens if you do with `Configure' on Makeperl.PL line (per documentation)? This looks like a "missing underscore" or a similar problem; apparently you have some weird combination of compilers and linkers... [Now I got a handful of reports of such a failure, but so far everybody fails to communicate long enough for me to remote-debug this.] Ilya
From: selsky [...] columbia.edu
Show quoted text
> Looks like a cousin to the 'overflow' mentioned in README. What > happens if you do with `Configure' on Makeperl.PL line (per > documentation)? > This looks like a "missing underscore" or a similar problem; > apparently you have some weird combination of compilers and linkers... > > [Now I got a handful of reports of such a failure, but so far > everybody fails to communicate long enough for me to remote-debug > this.]
I am using gcc-2.95.2 with Sun ld in /usr/ccs/bin which is not in my path. Normally gcc invokes ld so it knows where to find Sun ld in /usr/ccs/bin Configure looks for ld directly and finds the wrong one. I have gnu binutils installs since many of the other programs in it actually work on solaris. Is configure sending gnu ld options to solaris ld?
Date: Tue, 30 Jul 2002 18:17:16 -0400
From: Ilya Zakharevich <ilya [...] math.ohio-state.edu>
To: Guest via RT <bug-Math-Pari [...] rt.cpan.org>
Cc: "'AdminCc of cpan Ticket #1201'": ;
Subject: Re: [cpan #1201] hiremainder symbol not found
On Tue, Jul 30, 2002 at 03:12:27AM -0400, Guest via RT wrote: Show quoted text
> > [Now I got a handful of reports of such a failure, but so far > > everybody fails to communicate long enough for me to remote-debug > > this.]
> > I am using gcc-2.95.2 with Sun ld in /usr/ccs/bin which is not in my > path. Normally gcc invokes ld so it knows where to find Sun ld in > /usr/ccs/bin Configure looks for ld directly and finds the wrong one. > I have gnu binutils installs since many of the other programs in it > actually work on solaris. Is configure sending gnu ld options to > solaris ld?
Sorry, I have no idea what you are talking about. Which Configure, one that built perl? Can you send the result of perl -V and discuss your findings in terms of the names in this report? Ilya
From: dneedles [...] comcast.net
[ilya@math.ohio-state.edu - Tue Jul 30 18:17:19 2002]: Show quoted text
> On Tue, Jul 30, 2002 at 03:12:27AM -0400, Guest via RT wrote:
> > > [Now I got a handful of reports of such a failure, but so far > > > everybody fails to communicate long enough for me to remote-debug > > > this.]
> > > > I am using gcc-2.95.2 with Sun ld in /usr/ccs/bin which is not in my > > path. Normally gcc invokes ld so it knows where to find Sun ld in > > /usr/ccs/bin Configure looks for ld directly and finds the wrong
one. Show quoted text
> > I have gnu binutils installs since many of the other programs in it > > actually work on solaris. Is configure sending gnu ld options to > > solaris ld?
> > Sorry, I have no idea what you are talking about. Which Configure, > one that built perl? Can you send the result of perl -V and discuss > your findings in terms of the names in this report? > > Ilya >
There is a comment in the PERI MACHINES file #27 that GCC 2.95.2 had problems with the O3 optimizing command. This was fixed in GCC 2.95.3.
From: selsky [...] columbia.edu
Show quoted text
> There is a comment in the PERI MACHINES file #27 that GCC 2.95.2 had > problems with the O3 optimizing command. This was fixed in GCC 2.95.3.
I retested this with gcc 2.95.3 and Math::Pari 2.010603. Same problem. By the way, I'm using only the O optimizing command.
Date: Tue, 11 Oct 2005 02:48:53 -0700
From: Ilya Zakharevich <nospam-abuse [...] ilyaz.org>
To: Guest via RT <bug-Math-Pari [...] rt.cpan.org>
CC: undisclosed-recipients: ;
Subject: Re: [cpan #1201] hiremainder symbol not found
RT-Send-Cc:
On Tue, Oct 11, 2005 at 05:11:01AM -0400, Guest via RT wrote: Show quoted text
> > There is a comment in the PERI MACHINES file #27 that GCC 2.95.2 had > > problems with the O3 optimizing command. This was fixed in GCC 2.95.3.
Show quoted text
> I retested this with gcc 2.95.3 and Math::Pari 2.010603. Same problem. > By the way, I'm using only the O optimizing command.
I'm somehow confused: is it a continuation of some conversation? One remark: quite often, hiremainder and overflow problems are signs of Perl being not locally compiled (thus misinstalled). Also see the end of ./INSTALL. Hope this helps, Ilya
From: selsky [...] columbia.edu
Show quoted text
> One remark: quite often, hiremainder and overflow problems are signs > of Perl being not locally compiled (thus misinstalled). > > Also see the end of ./INSTALL.
Thanks for the speedy response. Perl is locally compiled. gcc is not using the "-fno-common" flag. The following CFLAGS are set: CCFLAGS = -fno-strict-aliasing -I/opt/local/include -D_LARGEFILE_SOURCE -D_FILE_OFFSET_BITS=64
Date: Tue, 11 Oct 2005 09:20:08 -0700
From: Ilya Zakharevich <nospam-abuse [...] ilyaz.org>
To: Guest via RT <bug-Math-Pari [...] rt.cpan.org>
Subject: Re: [cpan #1201] hiremainder symbol not found
RT-Send-Cc:
On Tue, Oct 11, 2005 at 05:57:32AM -0400, Guest via RT wrote: Show quoted text
> > Also see the end of ./INSTALL.
> > Thanks for the speedy response. Perl is locally compiled. gcc is not > using the "-fno-common" flag. The following CFLAGS are set: > > CCFLAGS = -fno-strict-aliasing -I/opt/local/include -D_LARGEFILE_SOURCE > -D_FILE_OFFSET_BITS=64
Please follow the full procedure for bug reporting. See ./README. Thanks, Ilya
Date: Fri, 28 Oct 2005 13:19:31 -0700
From: Ilya Zakharevich <nospam-abuse [...] ilyaz.org>
To: Guest via RT <bug-Math-Pari [...] rt.cpan.org>
CC: undisclosed-recipients: ;
Subject: Re: [cpan #1201] hiremainder symbol not found
RT-Send-Cc:
On Tue, Oct 11, 2005 at 05:57:32AM -0400, Guest via RT wrote: Show quoted text
> > This message about Math-Pari was sent to you by guest <> via rt.cpan.org > > Full context and any attached attachments can be found at: > <URL: https://rt.cpan.org/Ticket/Display.html?id=1201 > >
> > One remark: quite often, hiremainder and overflow problems are signs > > of Perl being not locally compiled (thus misinstalled). > > > > Also see the end of ./INSTALL.
> > Thanks for the speedy response. Perl is locally compiled. gcc is not > using the "-fno-common" flag. The following CFLAGS are set: > > CCFLAGS = -fno-strict-aliasing -I/opt/local/include -D_LARGEFILE_SOURCE > -D_FILE_OFFSET_BITS=64
Please recheck with The uploaded file Math-Pari-2.010700.tar.gz has entered CPAN as file: $CPAN/authors/id/I/IL/ILYAZ/modules/Math-Pari-2.010700.tar.gz size: 104354 bytes md5: 3acee76566692ddf8dc1391297b5202b (Should propagate to a server near you soon...) Thanks for your patience, Ilya
From: selsky [...] columbia.edu
Show quoted text
> Please recheck with > Math-Pari-2.010700.tar.gz
Same problem. The INSTALL file says to build libpari separately and include the output of nm on the kernel.o file: $ nm ./Osolaris-sparcv9/kernel.o 00000000 T bfffo 0000005c T divll U pari_err 00000000 d tabshi kernel.o is built like this: /opt/local/bin/as -o kernel.o kernel1.s This is the nm output from the kernel.o that is built automatically by Math-Pari: $ nm libPARI/kernel.o 00000000 T bfffo 0000005c T divll U hiremainder U pari_err 00000000 d tabshi kernel.o is built like this: as -P -T -I. -o kernel.o kernel1.s Is this helpful?
Date: Fri, 28 Oct 2005 16:57:57 -0700
From: Ilya Zakharevich <nospam-abuse [...] ilyaz.org>
To: Guest via RT <bug-Math-Pari [...] rt.cpan.org>
Subject: Re: [cpan #1201] hiremainder symbol not found
RT-Send-Cc:
On Fri, Oct 28, 2005 at 06:17:44PM -0400, Guest via RT wrote: Show quoted text
> > Please recheck with > > Math-Pari-2.010700.tar.gz
Show quoted text
> Same problem.
Show quoted text
> $ nm ./Osolaris-sparcv9/kernel.o > 00000000 T bfffo > 0000005c T divll > U pari_err > 00000000 d tabshi > > kernel.o is built like this: > > /opt/local/bin/as -o kernel.o kernel1.s
This is not enough. How is kernel1.s built? Show quoted text
> This is the nm output from the kernel.o that is built automatically by Math-Pari: > > $ nm libPARI/kernel.o > 00000000 T bfffo > 0000005c T divll > U hiremainder > U pari_err > 00000000 d tabshi > > kernel.o is built like this: > > as -P -T -I. -o kernel.o kernel1.s
Show quoted text
> Is this helpful?
I have no idea what to do. You could check whether just copying kernel.o from GP/PARI helps. If this helps, you could check the contents of kernel1.s files. You can check whether removing -P -T -I. helps. You can check whether copying kernel1.s from GP/PARI helps. Anyway, gcc 2.95 is a very bad release. And I do not recall ever seeing any bug report from you (see ./INSTALL ./README). Thanks, Ilya