Jim Meyering wrote: > Pádraig Brady wrote: >> FYI https://bugzilla.redhat.com/show_bug.cgi?id=655096 > > Ouch! Thanks for forwarding that. > > Here's one way to test for the bug: > > seq 100000 > in > mkfifo fifo > (for i in $(seq 12); do read line; echo $i; sleep .1; done > cat > /dev/null) < fifo & > (ulimit -t 1; ./sort in > fifo \ > || echo killed via $(env kill -l $(expr $? - 128))) > > It fails on multi-core systems: > > 1 > 2 > 3 > 4 > 5 > 6 > 7 > 8 > 9 > killed via XCPU > 10 > 11 > 12 > > But succeeds with the sort from coreutils-8.5.
Here's the beginning of a test suite addition for that bug. I've added it to the XFAIL list, for now. As I'm posting this I realized it will fail (i.e, pass) on uniprocessor systems, or if you set OMP_NUM_THREADS=1 in your environment, which led to another patch, one that ensures an OMP_NUM_THREADS setting does not perturb tests. That's below. I'll probably push the OMP_NUM_THREADS-adding test shortly. >From fbb47bf42457f8126cc128ff2c4590ad34d5e730 Mon Sep 17 00:00:00 2001 From: Jim Meyering <meyer...@redhat.com> Date: Sat, 27 Nov 2010 15:50:43 +0100 Subject: [PATCH 1/2] tests: test for parallel sort spinlock abuse * tests/misc/sort-spinlock-abuse: New file. * tests/Makefile.am (TESTS): Add it. (XFAIL_TESTS): Mark this as an expected-to-fail (for now) test. --- tests/Makefile.am | 4 ++++ tests/misc/sort-spinlock-abuse | 37 +++++++++++++++++++++++++++++++++++++ 2 files changed, 41 insertions(+), 0 deletions(-) create mode 100755 tests/misc/sort-spinlock-abuse diff --git a/tests/Makefile.am b/tests/Makefile.am index 2e33063..b3be4df 100644 --- a/tests/Makefile.am +++ b/tests/Makefile.am @@ -237,6 +237,7 @@ TESTS = \ misc/sort-merge-fdlimit \ misc/sort-month \ misc/sort-rand \ + misc/sort-spinlock-abuse \ misc/sort-unique \ misc/sort-version \ misc/split-a \ @@ -654,4 +655,7 @@ pr_data = \ pr/ttb3-FF \ pr/w72l24f-ll +XFAIL_TESTS = \ + misc/sort-spinlock-abuse + include $(srcdir)/check.mk diff --git a/tests/misc/sort-spinlock-abuse b/tests/misc/sort-spinlock-abuse new file mode 100755 index 0000000..974e3d5 --- /dev/null +++ b/tests/misc/sort-spinlock-abuse @@ -0,0 +1,37 @@ +#!/bin/sh +# trigger a bug that would make parallel sort use 100% of one or more +# CPU while blocked on output. + +# Copyright (C) 2010 Free Software Foundation, Inc. + +# This program is free software: you can redistribute it and/or modify +# it under the terms of the GNU General Public License as published by +# the Free Software Foundation, either version 3 of the License, or +# (at your option) any later version. + +# This program is distributed in the hope that it will be useful, +# but WITHOUT ANY WARRANTY; without even the implied warranty of +# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the +# GNU General Public License for more details. + +# You should have received a copy of the GNU General Public License +# along with this program. If not, see <http://www.gnu.org/licenses/>. + +. "${srcdir=.}/init.sh"; path_prepend_ ../src +print_ver_ sort + +test "$(nproc)" = 1 && skip_ "requires a multi-core system" + +seq 100000 > in || framework_failure_ +mkfifo fifo || framework_failure_ + +# Arrange for sort to require 1.2+ seconds of wall-clock time, +# while actually using far less than 1 second of CPU time. +(for i in $(seq 12); do read line; echo $i; sleep .1; done + cat > /dev/null) < fifo & + +# Limit CPU time to 1 second +ulimit -t 1 +sort --parallel=2 in > fifo || fail=1 + +Exit $fail -- 1.7.3.2.846.gf4b062 >From 43d1112d01c0251076b5ec61605e45b101ab3e12 Mon Sep 17 00:00:00 2001 From: Jim Meyering <meyer...@redhat.com> Date: Sat, 27 Nov 2010 15:58:51 +0100 Subject: [PATCH 2/2] tests: don't let the OMP_NUM_THREADS envvar affect sort tests * tests/envvar-check (vars): Add OMP_NUM_THREADS. --- tests/envvar-check | 1 + 1 files changed, 1 insertions(+), 0 deletions(-) diff --git a/tests/envvar-check b/tests/envvar-check index 1dee9af..ac53ee4 100644 --- a/tests/envvar-check +++ b/tests/envvar-check @@ -34,6 +34,7 @@ vars=' LANGUAGE LS_BLOCK_SIZE LS_COLORS + OMP_NUM_THREADS POSIXLY_CORRECT QUOTING_STYLE SIMPLE_BACKUP_SUFFIX -- 1.7.3.2.846.gf4b062