Re: git coreutils 'make check' hangs

[Pádraig Brady]

Ralf Wildenhues wrote:
> Hello,
> 
> for some time now, 'make check' in the git coreutils tree hangs for me:
> 
> ralf 26931  0.0  0.0   9516  1444 pts/4    S    Nov29   0:00 /bin/sh -c 
> p='tail-2/inotify-race'; case $- in *e*) set +e;; esac; srcdirstrip=`echo "." 
> | sed 's|.|.|g'`; case $p in ./*) f=`echo "$p" | sed "s|^$srcdirstrip/||"`;; 
> *) f=$p;; esac; srcdir=.; export srcdir; rm -f tail-2/inotify-race.log-t; 
> trap 'st=$?; rm -f '\''/tmp/coreutils/tests/tail-2/inotify-race.log-t'\''; 
> (exit $st); exit $st' 1 2 13 15; am__odir=`echo "./tail-2/inotify-race.log" | 
> sed 's|/[^/]*$||'`; test "x$am__odir" = x. || /bin/mkdir -p "$am__odir" || 
> exit $?; if test -f "./$f"; then dir=./; elif test -f "$f"; then dir=; else 
> dir="./"; fi; tst=$dir$f; log='tail-2/inotify-race.log'; __SAVED_TERM=$TERM; 
> . ./lang-default; tmp__=$TMPDIR; test -d "$tmp__" || tmp__=.; . 
> ./envvar-check; TMPDIR=$tmp__; export TMPDIR; exec 9>&2; shell_or_perl_() { 
> if grep '^#!/usr/bin/perl' "$1" > /dev/null; then if perl -e 'use warnings' > 
> /dev/null 2>&1; then grep '^#!/usr/bin/perl -T' "$1" > /dev/null && T_=T || 
> T_=; perl -w
$T_ -I. -MCoreutils -M"CuTmpdir qw($f)" -- "$1"; else echo 1>&2 "$tst: 
configure did not find a usable version of Perl," "so skipping this test"; 
(exit 77); fi; else /bin/sh "$1"; fi; }; export LOCALE_FR='none' 
LOCALE_FR_UTF8='none' abs_top_builddir='/tmp/coreutils' 
abs_top_srcdir='/tmp/coreutils' abs_srcdir='/tmp/coreutils/tests' 
built_programs="`(cd ../src && MAKEFLAGS= make -s built_programs.list)`" 
host_os=linux-gnu host_triplet='x86_64-unknown-linux-gnu' srcdir='.' 
top_srcdir='..' CONFIG_HEADER='/tmp/coreutils/lib/config.h' 
CU_TEST_NAME=`basename '/tmp/coreutils/tests'`,$tst CC='gcc -std=gnu99' 
AWK='gawk' EGREP='/bin/grep -E' EXEEXT='' MAKE=make 
PACKAGE_BUGREPORT='address@hidden' PACKAGE_VERSION=8.1.9-9cf9f-dirty 
PERL='perl' PREFERABLY_POSIX_SHELL='/bin/sh' REPLACE_GETCWD=1 
PATH='/tmp/coreutils/src:'"$PATH" ; shell_or_perl_    "$tst" 
>tail-2/inotify-race.log-t 2>&1; estatus=$?; if test -n '' && test $estatus -eq 
99; then estatus=1; fi; TERM=$__SAVED_TERM; export T
ERM; red=; grn=; lgn=; blu=; std=; test "X" != Xno && test "X$TERM" != Xdumb && 
{ test "X" = Xalways || test -t 1 2>/dev/null; } && { red='?[0;31m'; 
grn='?[0;32m'; lgn='?[1;32m'; blu='?[1;34m'; std='?[m'; }; xfailed=PASS; case " 
 " in *[\ \?]$f[\ \?]* | *[\ \?]$dir$f[\ \?]*) xfailed=XFAIL;; esac; case 
$estatus:$xfailed in 0:XFAIL) col=$red; res=XPASS;; 0:*)     col=$grn; res=PASS 
;; 77:*)    col=$blu; res=SKIP ;; 99:*)    col=$red; res=FAIL ;; *:XFAIL) 
col=$lgn; res=XFAIL;; *:*)     col=$red; res=FAIL ;; esac; echo 
"${col}$res${std}: $f"; echo "$res: $f (exit: $estatus)" | sed 'p;s/./=/g;p;g' 
>tail-2/inotify-race.log; cat tail-2/inotify-race.log-t 
>>tail-2/inotify-race.log; rm -f tail-2/inotify-race.log-t
> 
> This is a bit ugly since for one, hitting CTRL-c fails to kill the
> testsuite, '<CTRL>-Z bg; kill %1' will not clean up the testsuite
> processes.  I just removed a number of zombies from week-old runs.
> 
> I'm typically using parallel make check, but it seems that doesn't
> matter for this issue.  What should I be looking for?

I think it may be a gdb issue:
http://lists.gnu.org/archive/html/bug-coreutils/2009-12/msg00025.htm
For the moment we've marked that test as "very expensive"
so that it will not be run by default.

cheers,
Pádraig.