gcl-devel
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[Gcl-devel] Re: Update: Re: [ 101720 ] gcl 2.4.4 compile problems solar


From: Camm Maguire
Subject: [Gcl-devel] Re: Update: Re: [ 101720 ] gcl 2.4.4 compile problems solaris 8
Date: 15 Jan 2003 11:19:39 -0500

Greetings!

"Greg T. Harber" <address@hidden> writes:

> It's working somewhat.  I had to edit the gcl script:
> 
> #!/bin/sh  caused the "not an identifier" problem
> 
> $ gcl
> /usr/local/bin/gcl: C_INCLUDE_PATH=/usr/local/lib/gcl-2.5.0/h: is not an 
> identifier
> 

Are you using GCC as the C compiler?  The goal of that line is to
enable the underlying C compiler to find the cmpinclude.h file
required for compiling all lisp code if the :system-p flag is set.
Without this flag, the cmpinclude.h will be dumped into each c file.  

Can one not set environment variables in this way with your /bin/sh?
I thought that was POSIX?  

> Then it core dumped:
> 
> address@hidden root]# gcl
> GCL (GNU Common Lisp)  Version(2.5.0) Tue Jan 14 10:01:04 CST 2003
> Licensed under GNU Library General Public License
> Contains Enhancements by W. Schelter
> Bus Error (core dumped)
> address@hidden root]# gdb -c core
> GNU gdb 5.0
> Copyright 2000 Free Software Foundation, Inc.
> GDB is free software, covered by the GNU General Public License, and you are
> welcome to change it and/or distribute copies of it under certain conditions.
> Type "show copying" to see the conditions.
> There is absolutely no warranty for GDB.  Type "show warranty" for details.
> This GDB was configured as "sparc-sun-solaris2.8".
> Core was generated by `/usr/local/lib/gcl-2.5.0/unixport/saved_gcl -dir 
> /usr/local/lib/gcl-2.5.0/unixp'.
> Program terminated with signal 10, Bus Error.
> #0  0x8538b8 in ?? ()
> (gdb) quit
> 

I have seen bus errors on Debian/sparc a long time ago, and traced it
to the fact that even on 32bit sparc, memory needs to be 8 byte
aligned in many cases.  This is accomplished in GCL with the PRT_ALIGN
macro, which is set to 8 in h/solaris.h and h/sparc-linux.h.  Please
verify that you are indeed using the former configuration. (Should be
evident from the configure output, machine=...)

Beyond this, it would be helpful if you could recompile with
--enable-debug passed to configure, get the same bus error in gdb, and
send me a back trace ("bt").

Please also download the prebuilt solaris package at
ftp://ftp.gnu.org/pub/gnu/gcl/cvs and see if the same error appears on
your machine.  It doesn't on my solaris build machine:

Last login: Tue Jan 14 09:03:58 2003 from h-66-134-96-17.
Sun Microsystems Inc.   SunOS 5.8       Generic February 2000
$ cd gcl*
$ pwd
/u/camm/gcl
$ unixport/saved_gcl -dir /usr/local/lib/gcl-2.5.0/unixport
GCL (GNU Common Lisp)  Version(2.5.0) Fri Dec  6 01:25:26 CST 2002
Licensed under GNU Library General Public License
Contains Enhancements by W. Schelter

>*system-directory*)

"/usr/local/lib/gcl-2.5.0/unixport/"

>(by)

If none of these options sheds any light, please let me know, and I
will send you my full configure and build output on solaris to compare
with your own.

Take care,


> So I commented out the options on saved_gcl
> 
> #!/bin/ksh
> export C_INCLUDE_PATH=/usr/local/lib/gcl-2.5.0/h
> exec /usr/local/lib/gcl-2.5.0/unixport/saved_gcl \
> #   -dir /usr/local/lib/gcl-2.5.0/unixport \
>     -libdir /usr/local/lib/gcl-2.5.0/ \
>     -eval '(setq si::*allow-gzipped-file* t)' \
>     -eval '(setq si::*tk-library* "/usr/local/lib/tk8.4")' \
>     -eval '(si::init-readline)' \
>       "$@"
> # other options: -load /tmp/foo.o -load jo.lsp -eval "(joe 3)"
> 
> I'm not too comfortable with my solution but I will get me lisp guys to try it
> out.
> 
> gth
> 
> Camm Maguire wrote:
> > Greetings!
> > 
> > Here is the script that I successfully use to autobuild GCL cvs to
> > synchronize with Debian GCL package releases.  The resulting binary
> > package with source can be found at ftp://ftp.gnu.org/pub/gnu/gcl/cvs.
> > 
> > =============================================================================
> > #!/bin/bash
> > 
> > set -e
> > 
> > export PATH=$HOME/bin:/usr/ccs/bin/:$PATH
> > 
> > HAVE=$(ls -1rt gcl*tar.gz 2>/dev/null| tail -1l)
> > if [ "$HAVE" != "" ] ; then
> > mv $HAVE foo
> > rm -f gcl*tar.gz
> > mv foo $HAVE
> > fi
> > 
> > RHAVE=$(
> > (
> >  ftp ftp.debian.org <<EOF
> >  cd debian/pool/main/g/gcl
> >  ls -1rt 
> >  quit
> > EOF
> > ) | grep tar.gz | tail -1l)
> > 
> > if [ "$HAVE" != "$RHAVE" ] ; then
> > 
> > ftp ftp.debian.org <<EOF1
> > cd debian/pool/main/g/gcl
> > bin
> > get $RHAVE
> > quit
> > EOF1
> > 
> > rm -f $HAVE
> > HAVE=$RHAVE
> > 
> > fi
> > 
> > export NAME="gcl_$(echo $HAVE | cut -f2 -d_ | sed 's,.tar.gz,,1').solaris"
> > 
> > if [ -e $NAME.tgz ] ; then
> > exit 0
> > fi
> >   
> > rm -f gcl*solaris.tgz
> > 
> > gzcat $HAVE | tar xf -
> > DIR=$(echo $HAVE | cut -f1 -d\- | tr '_' '-')
> > rm -rf gcl
> > mv $DIR gcl
> > 
> > cd gcl
> > 
> > ./configure --disable-statsysbfd --enable-locbfd
> > cd gmp3
> > ./configure --build=sparc-sun-solaris2.8
> > cd ..
> > make
> > rm -rf tmp
> > mkdir tmp
> > make install DESTDIR=$(pwd)/tmp
> > cd tmp
> > tar cf $HOME/$NAME.tar .
> > cd $HOME
> > gzip $NAME.tar
> > mv $NAME.tar.gz $NAME.tgz
> > 
> > exit 0
> > =============================================================================
> > 
> > Notice that 
> > 
> > 1) I don't use the system bfd, but compile the local bfd supplied with
> >    the GCL source. ( --disable-statsysbfd --enable-locbfd)
> > 
> > 2) I reconfigure the gmp3 subbuild to use only vanilla sparc
> >    instructions:  ( --build=sparc-sun-solaris2.8)
> > 
> > This build has been successfully compiling maxima and acl2 for some
> > time now.
> > 
> > 2.4.4 should work similarly as far as I can tell, but I would
> >     recommend going with a CVS snapshot, as many bug fixes are already
> >     incorporated.  2.4.4 is known to correctly build maxima, however.
> >     You can get current CVS at
> >     :pserver:address@hidden:/cvsroot/gcl, or retrieve a
> >     snapshot tarball at ftp://ftp.gnu.org/pub/gnu/cvs/gcl....tar.gz
> > 
> > Please let me know how it goes.
> > 
> > Take care,
> > 
> > address@hidden writes:
> > 
> > 
> >>Support Request #101720, was updated on 2003-Jan-08 17:09
> >>You can respond by visiting: 
> >>http://savannah.gnu.org/support/?func=detailsupport&support_id=101720&group_id=879
> >>
> >>Category: None
> >>Status: Open
> >>Priority: 5
> >>Summary: gcl 2.4.4 compile problems solaris 8
> >>
> >>By: steve
> >>Date: 2003-Jan-13 09:30
> >>Logged In: NO 
> >>Browser: Mozilla/5.0 (X11; U; SunOS sun4u; en-US; rv:0.9.4.1) 
> >>Gecko/20020518 Netscape6/6.2.3
> >>
> >>We are unable to build gcl on solaris. Here;s the note from
> >>the system administrator (address@hidden).
> >>
> >>Have you read any horror stories of people compiling this
> >>thing??? I've spent the last three hours on it, and I can't
> >>get it to behave at all.  When I tell it to build 32-bit
> >>binaries ONLY, it attempts to build 64-bit binaries.  When I
> >>tell it nothing, it builds 64-bit binaries but complains the
> >>the linker is expecting 32-bit binaries.I'm more confused
> >>than I was the last time I tried to build it!  If you have
> >>some magic potion for this.... PLEASE SHARE!!!
> >>
> >>----------------------------------------------------------------------
> >>
> >>By: gth
> >>Date: 2003-Jan-08 17:09
> >>Logged In: NO 
> >>Browser: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.0.1) Gecko/20021003
> >>
> >>normal ./configure make sequence creates a 32 bit main.o
> >>and a 64 bit gmp  which leads to the dreaded:
> >>
> >>ld: fatal: file ../o/main.o: wrong ELF class: ELFCLASS32
> >>
> >>(if I add  -m64 into makedefs)
> >>
> >># for main link of raw_gcl
> >>LIBS= -lm -m64 -lsocket -lnsl -lreadline -lncurses
> >>
> >>or  if I leave makedefs alone
> >>
> >>ld: warning: file ../gmp/libgmp.a(mp_set_fns.o): wrong
> >>ELF class: ELFCLASS64
> >>
> >>
> >>How can I work around this?
> >>
> >>----------------------------------------------------------------------
> >>You can respond by visiting: 
> >>http://savannah.gnu.org/support/?func=detailsupport&support_id=101720&group_id=879
> >>
> >>
> > 
> > 
> 
> 
> 
> -- 
> Greg T. Harber                        address@hidden
> Director, McGee Computing Laboratory  936 468-2508
> Department of Computer Science        P.O. Box 13063
> Stephen F. Austin State University    Nacogdoches, TX 75962 USA
> 
> 
> 

-- 
Camm Maguire                                            address@hidden
==========================================================================
"The earth is but one country, and mankind its citizens."  --  Baha'u'llah




reply via email to

[Prev in Thread] Current Thread [Next in Thread]