[PATCH] tests: avoid sort-spinlock-abuse false positive under heavy load

[Jim Meyering]

I have just seen this unwarranted failure again:

    sort (GNU coreutils) 8.12.86-cd27-dirty
    Copyright (C) 2011 Free Software Foundation, Inc.
    License GPLv3+: GNU GPL version 3 or later 
<http://gnu.org/licenses/gpl.html>.
    This is free software: you are free to change and redistribute it.
    There is NO WARRANTY, to the extent permitted by law.

    Written by Mike Haertel and Paul Eggert.
    + grep '^#define HAVE_PTHREAD_T 1' 
/h/j/w/co/cu/tests/torture/coreutils/test/coreutils-8.12.86-cd27-dirty/lib/config.h
    + seq 100000
    + mkfifo_or_skip_ fifo
    + test 1 = 1
    + mkfifo fifo
    + ulimit -t 7
    + sort --parallel=2 in
    ++ seq 80
    + for i in '$(seq 80)'
    + read line
    + echo 1
    1
    + sleep .1
    ./misc/sort-spinlock-abuse: line 37: 8828 Killed sort --parallel=2 in > fifo
    + fail=1
    + Exit 1
    + set +e
    + exit 1
    + exit 1
    + remove_tmp_
    + __st=1
    + cleanup_

That was one time too many, so with this change,
the offending test is relegated to being run at least
once per release, when I run the very expensive tests
as outlined in README-release.

>From 640023b63fe4609e6e98b1b988b292ef2707ea60 Mon Sep 17 00:00:00 2001
From: Jim Meyering <address@hidden>
Date: Fri, 17 Jun 2011 12:01:10 +0200
Subject: [PATCH] tests: avoid sort-spinlock-abuse false positive under heavy
 load

* tests/misc/sort-spinlock-abuse: Classify as "very expensive" to
avoid unwarranted failure once and for all.
---
 tests/misc/sort-spinlock-abuse |    6 ++++++
 1 files changed, 6 insertions(+), 0 deletions(-)

diff --git a/tests/misc/sort-spinlock-abuse b/tests/misc/sort-spinlock-abuse
index 4cf8866..0e1c86b 100755
--- a/tests/misc/sort-spinlock-abuse
+++ b/tests/misc/sort-spinlock-abuse
@@ -20,6 +20,12 @@
 . "${srcdir=.}/init.sh"; path_prepend_ ../src
 print_ver_ sort

+# This isn't terribly expensive, but it must not be run under heavy load.
+# Since the "very expensive" are already run only with -j1, adding this
+# test to the list ensures it still gets _some_ (albeit minimal) coverage
+# while not causing false-positive failures in day to day runs.
+very_expensive_
+
 grep '^#define HAVE_PTHREAD_T 1' "$CONFIG_HEADER" > /dev/null ||
   skip_ 'requires pthreads'

--
1.7.6.rc0.293.g40857