Package: coreutils;
Reported by: "Nelson H. F. Beebe" <beebe <at> math.utah.edu>
Date: Tue, 6 Nov 2012 13:44:02 UTC
Severity: normal
Done: Assaf Gordon <assafgordon <at> gmail.com>
Bug is archived. No further changes may be made.
View this message in rfc822 format
From: Pádraig Brady <P <at> draigBrady.com> To: "Nelson H. F. Beebe" <beebe <at> math.utah.edu> Cc: 12815 <at> debbugs.gnu.org Subject: bug#12815: [bug-coreutils] build glitches in coreutils-8.20 Date: Tue, 06 Nov 2012 14:17:12 +0000
On 11/06/2012 01:40 PM, Nelson H. F. Beebe wrote:
> I've now successfully built and installed coreutils-8.20 on about 20
> flavors of Unix on several different CPU families. The same single
> build repair was needed on several systems, and was easily fixed by
> manually tweaking a compiler command:
>
> Linking of src/factor requires addition of -liconv and -lgmp,
> because they are missing in the Makefile
Thanks a lot for all this Nelson!
The -liconv issue should be handled by:
http://git.sv.gnu.org/gitweb/?p=coreutils.git;a=commitdiff;h=88a62019
I'm surprised that -lgmp is needed, so it seems gmp.m4 is inadequate:
http://git.sv.gnu.org/gitweb/?p=coreutils.git;a=blob;f=m4/gmp.m4;hb=HEAD
I notice that LIB_GMP is set within {}. Does that depend on the shell
not running that within a subshell?
> Here is a summary of the test failures reported by "make check" for
> those systems where I had to apply the above tweak:
>
> System types:
> gen-ppc64 Gentoo 2.1 PowerPC-64
> gen-sparc Gentoo 2.1 SPARC
> macosx-ppc Mac OS X 10.5.8 PowerPC
> macosx-x64 Mac OS X 10.5.8 x86_64
> mirbsd-x86 MirBSD 10 x86
> osuse-x64 openSUSE 11.4 x86_64
> rh5-x64 Red Hat 5.8 x86_64
> rh6-x64 Red Hat 6.3 x86_64
> sol10-sparc Solaris 10 SPARC
> sol11-x64 Solaris 11 x86_64
> ubuntu-x86 Ubuntu 11.10 x86
>
Do you have tests/test-suite.log files and compile flags for the following?
> ============================================================
> failing test name (sort key) system type
> ============================================================
> expr rh6-x64
> factor rh5-x64
> factor macosx-ppc
> factor rh6-x64
> test-getlogin gen-sparc
> test-getlogin osuse-x64
> test-getlogin ubuntu-x86
> test-getlogin (exit: 134) gen-sparc
> test-getlogin (exit: 134) osuse-x64
> test-getlogin (exit: 134) ubuntu-x86
There are many reports of getlogin gnulib tests failing.
I'll send a patch to skip on some exit codes and
also provide more diagnostics
> test-mbsalign mirbsd-x86
> test-mbsalign (exit: 134) mirbsd-x86
> test-parse-datetime mirbsd-x86
> test-parse-datetime (exit: 134) mirbsd-x86
> test-posixtm mirbsd-x86
> test-posixtm (exit: 1) mirbsd-x86
> test-select mirbsd-x86
> test-select (exit: 1) mirbsd-x86
> test-verify.sh mirbsd-x86
> test-verify.sh (exit: 1) mirbsd-x86
> tests/cp/fiemap-FMR.sh gen-ppc64
> tests/cp/fiemap-FMR.sh (exit: 1) gen-ppc64
> tests/cp/nfs-removal-race.sh macosx-ppc
> tests/cp/nfs-removal-race.sh sol10-sparc
> tests/cp/nfs-removal-race.sh (exit: 99) macosx-ppc
> tests/cp/same-file.sh sol11-x64
> tests/cp/sparse.sh rh6-x64
> tests/cp/sparse.sh (exit: 99) rh6-x64
> tests/dd/reblock.sh sol11-x64
> tests/du/basic.sh rh6-x64
> tests/du/basic.sh (exit: 1) rh6-x64
> tests/install/install-C.sh macosx-ppc
> tests/install/install-C.sh (exit: 1) macosx-ppc
> tests/ls/getxattr-speedup.sh macosx-ppc
> tests/ls/getxattr-speedup.sh (exit: 99) macosx-ppc
> tests/ls/stat-free-color.sh rh6-x64
> tests/ls/stat-free-color.sh (exit: 1) rh6-x64
> tests/misc/cat-buf.sh sol11-x64
> tests/misc/expr.pl rh6-x64
> tests/misc/expr.pl (exit: 1) rh6-x64
> tests/misc/factor.pl rh5-x64
> tests/misc/factor.pl macosx-ppc
> tests/misc/factor.pl rh6-x64
> tests/misc/factor.pl (exit: 1) rh5-x64
> tests/misc/factor.pl (exit: 1) macosx-ppc
> tests/misc/factor.pl (exit: 1) rh6-x64
> tests/misc/help-version.sh rh5-x64
> tests/misc/help-version.sh macosx-ppc
> tests/misc/help-version.sh rh6-x64
> tests/misc/help-version.sh (exit: 1) rh5-x64
> tests/misc/help-version.sh (exit: 1) macosx-ppc
> tests/misc/help-version.sh (exit: 1) rh6-x64
> tests/misc/invalid-opt.pl rh5-x64
> tests/misc/invalid-opt.pl rh6-x64
> tests/misc/invalid-opt.pl (exit: 1) rh5-x64
> tests/misc/invalid-opt.pl (exit: 1) rh6-x64
> tests/misc/od-multiple-t.sh rh6-x64
> tests/misc/od-multiple-t.sh (exit: 1) rh6-x64
> tests/misc/sort-discrim.sh rh6-x64
> tests/misc/sort-discrim.sh (exit: 1) rh6-x64
> tests/misc/sort-u-FMR.sh gen-ppc64
> tests/misc/sort-u-FMR.sh (exit: 1) gen-ppc64
> tests/misc/stat-birthtime.sh rh6-x64
> tests/misc/stat-birthtime.sh sol11-x64
> tests/misc/stat-birthtime.sh (exit: 1) rh6-x64
> tests/misc/stdbuf.sh sol11-x64
> tests/misc/timeout-parameters.sh sol10-sparc
> tests/misc/timeout-parameters.sh rh6-x64
> tests/misc/timeout-parameters.sh (exit: 1) rh6-x64
> tests/mv/i-3.sh sol11-x64
> tests/mv/part-symlink.sh sol11-x64
> tests/mv/part-symlink.sh (exit: 1) sol11-x64
> tests/mv/perm-1.sh sol11-x64
> tests/mv/perm-1.sh (exit: 1) sol11-x64
> tests/rm/cycle.sh sol11-x64
> tests/rm/cycle.sh (exit: 1) sol11-x64
> tests/rm/dangling-symlink.sh sol11-x64
> tests/rm/rm1.sh sol11-x64
> tests/rm/rm1.sh (exit: 1) sol11-x64
> tests/rm/unread3.sh macosx-ppc
> tests/rm/unread3.sh (exit: 1) macosx-ppc
> tests/split/suffix-length.sh rh6-x64
> tests/split/suffix-length.sh (exit: 1) rh6-x64
> tests/tail-2/F-vs-missing.sh rh6-x64
> tests/tail-2/F-vs-missing.sh sol11-x64
> tests/tail-2/F-vs-missing.sh (exit: 1) rh6-x64
> tests/tail-2/F-vs-rename.sh sol11-x64
> tests/tail-2/flush-initial.sh sol11-x64
> tests/tail-2/inotify-hash-abuse.sh rh6-x64
> tests/tail-2/inotify-hash-abuse.sh sol11-x64
> tests/tail-2/inotify-hash-abuse.sh (exit: 1) rh6-x64
> tests/tail-2/pipe-f2.sh rh6-x64
> tests/tail-2/pipe-f2.sh sol11-x64
> tests/tail-2/pipe-f2.sh (exit: 1) rh6-x64
> tests/tail-2/wait.sh sol11-x64
> ============================================================
>
> One or more builds on these systems reported "All nnn tests passed",
> with NO test failures at all:
>
> Fedora 17 IA-32 (Pentium)
> Red Hat 5.6 IA-32 (Pentium)
> Red Hat 5.6 IA-64 (Itanium)
> gen-sparc
> osuse-x64
> sol10-sparc
> ubuntu-x86
>
> There is overlap between those two lists because builds were done with
> different compilers and compilation options on most systems.
thanks again,
Pádraig.
GNU bug tracking system
Copyright (C) 1999 Darren O. Benham,
1997,2003 nCipher Corporation Ltd,
1994-97 Ian Jackson.