[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [coreutils] over aggressive threads in sort
From: |
Jim Meyering |
Subject: |
Re: [coreutils] over aggressive threads in sort |
Date: |
Sat, 27 Nov 2010 16:33:03 +0100 |
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 <address@hidden>
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 <address@hidden>
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
- [coreutils] over aggressive threads in sort, Pádraig Brady, 2010/11/22
- Re: [coreutils] over aggressive threads in sort, DJ Lucas, 2010/11/27
- Re: bug#7489: [coreutils] over aggressive threads in sort, Pádraig Brady, 2010/11/26
- Re: bug#7489: [coreutils] over aggressive threads in sort, Paul Eggert, 2010/11/27
- Re: bug#7489: [coreutils] over aggressive threads in sort, Paul Eggert, 2010/11/27
- Re: bug#7489: [coreutils] over aggressive threads in sort, Paul Eggert, 2010/11/27
- Re: bug#7489: [coreutils] over aggressive threads in sort, Pádraig Brady, 2010/11/27
- Re: bug#7489: [coreutils] over aggressive threads in sort, DJ Lucas, 2010/11/27