Bug 649023
| Summary: | python-devel on ppc64 requires "-m64" on gcc command line | ||
|---|---|---|---|
| Product: | Red Hat Enterprise Linux 6 | Reporter: | Dave Malcolm <dmalcolm> |
| Component: | python | Assignee: | Dave Malcolm <dmalcolm> |
| Status: | CLOSED WONTFIX | QA Contact: | BaseOS QE - Apps <qe-baseos-apps> |
| Severity: | medium | Docs Contact: | |
| Priority: | low | ||
| Version: | 6.0 | CC: | jgalipea, lockhart, sgallagh |
| Target Milestone: | rc | ||
| Target Release: | --- | ||
| Hardware: | ppc64 | ||
| OS: | Unspecified | ||
| Whiteboard: | |||
| Fixed In Version: | Doc Type: | Bug Fix | |
| Doc Text: | Story Points: | --- | |
| Clone Of: | Environment: | ||
| Last Closed: | 2010-11-02 21:24:58 UTC | Type: | --- |
| Regression: | --- | Mount Type: | --- |
| Documentation: | --- | CRM: | |
| Verified Versions: | Category: | --- | |
| oVirt Team: | --- | RHEL 7.3 requirements from Atomic Host: | |
| Cloudforms Team: | --- | Target Upstream Version: | |
| Embargoed: | |||
$ rpm -qf /usr/lib/python2.6/config/Makefile
python-2.6.5-3.el6.ppc
/usr/lib/python2.6/config/Makefile has:
OPT= -DNDEBUG -O2 -g -pipe -Wall -Wp,-D_FORTIFY_SOURCE=2 -fexceptions -fstack-protector --param=ssp-buffer-size=4 -m32 -D_GNU_SOURCE -fPIC -fwrapv
BASECFLAGS= -fno-strict-aliasing
CFLAGS= $(BASECFLAGS) $(OPT) $(EXTRA_CFLAGS)
$ rpm -qf /usr/lib64/python2.6/config/Makefile
python-2.6.5-3.el6.ppc64
whereas /usr/lib64/python2.6/config/Makefile has:
OPT= -DNDEBUG -O2 -g -pipe -Wall -Wp,-D_FORTIFY_SOURCE=2 -fexceptions -fstack-protector --param=ssp-buffer-size=4 -m64 -mminimal-toc -D_GNU_SOURCE -fPIC -fwrapv
BASECFLAGS= -fno-strict-aliasing
CFLAGS= $(BASECFLAGS) $(OPT) $(EXTRA_CFLAGS)
So a module using:
distutils.sysconfig.get_config_vars("CFLAGS")
will work, but one using "BASECFLAGS" will fail on ppc64, unless "-m64" also is used (e.g. by setting CFLAGS).
It looks like rpmbuild on pp64 sets CFLAGS to a value that includes "-m64", so in theory this shouldn't be a problem.
Looks similar to bug 647413 (though that's python26.x86_64 on EPEL5) We split pyconfig.h into pyconfig-32.h and pyconfig-64.h in order to allow both to be installed on 64-bit boxes (see bug 139911 and bug 192747). We are relying on #include <bits/wordsize.h> setting __WORDSIZE to 32 or 64 as appropriate. On ppc64 it appears that gcc requires -m64 to be passed on the command line, but I don't see a workaround for that without breaking the fix for bug 139911). If there's a specific build scenario in which this fails, please open a bug. Development Management has reviewed and declined this request. You may appeal this decision by reopening this request. |
Description of problem: $ rpm -q python python-libs python-devel gcc python-2.6.5-3.el6.ppc64 python-libs-2.6.5-3.el6.ppc64 python-libs-2.6.5-3.el6.ppc python-devel-2.6.5-3.el6.ppc64 gcc-4.4.4-13.el6.ppc64 $ cat test.c #include <Python.h> int main() { return 0; } Compiles OK on ppc64 with "-m64": $ gcc -m64 -I/usr/include/python2.6 test.c $ echo $? 0 Fails without "-m64": $ gcc -I/usr/include/python2.6 test.c In file included from /usr/include/python2.6/Python.h:8, from test.c:1: /usr/include/python2.6/pyconfig.h:4:25: error: pyconfig-32.h: No such file or directory Version-Release number of selected component (if applicable): python-2.6.5-3.el6.ppc64 python-libs-2.6.5-3.el6.ppc64 python-libs-2.6.5-3.el6.ppc python-devel-2.6.5-3.el6.ppc64 gcc-4.4.4-13.el6.ppc64 glibc-headers-2.12-1.7.el6_0.3.ppc64 How reproducible: 100% Steps to Reproduce: As above Additional info: pyconfig.h is a downstream-modified copy in our RPMs: #include <bits/wordsize.h> #if __WORDSIZE == 32 #include "pyconfig-32.h" #elif __WORDSIZE == 64 #include "pyconfig-64.h" #else #error "Unknown word size" #endif glibc-headers-2.12-1.7.el6_0.3.ppc64's /usr/include/bits/wordsize.h begins: /* Determine the wordsize from the preprocessor defines. */ #if defined __powerpc64__ # define __WORDSIZE 64 # define __WORDSIZE_COMPAT32 1 #else # define __WORDSIZE 32 #endif and this is coming out as "32" on ppc64 without -m64