On 21/09/2025 18:30, Bruno Haible wrote:
Pádraig Brady wrote:
and was able to pass or skip always with the attached.
Thanks. With this patch, I confirm, the fold/fold-zero-width failure is fixed.
But on CentOS 7, I see a new test failure:
FAIL: tests/du/move-dir-while-traversing
Here's the relevant test-suite.log part.
That test is inherently racy.
I presume the failure was intermittent (or a crazy fast machine).
I'll look at fixing for the next release.
In the meantime we can minimize the race
without impacting the run time,
as done in the attached.
cheers,
Padraig
From 7d522ed4011fd7ef4709dc47182e3bb72833163f Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?P=C3=A1draig=20Brady?= <[email protected]>
Date: Sun, 21 Sep 2025 20:14:10 +0100
Subject: [PATCH] tests: du: avoid false failure in racy test
* tests/du/move-dir-while-traversing.sh: Expand the work to avoid
a false failure where du completes before the directory is moved.
Also expand the timeout to our more standard 10s to avoid the
"directory mover" being killed before du processes the directory.
This doesn't perceptibly impact the run time of the test.
Reported by Bruno Haible on a CentOS 7 system.
---
tests/du/move-dir-while-traversing.sh | 7 +++----
1 file changed, 3 insertions(+), 4 deletions(-)
diff --git a/tests/du/move-dir-while-traversing.sh b/tests/du/move-dir-while-traversing.sh
index 1d0a359bf..adf482b8e 100755
--- a/tests/du/move-dir-while-traversing.sh
+++ b/tests/du/move-dir-while-traversing.sh
@@ -66,9 +66,8 @@ long=d/e/f/g/h/i/j/k/l/m/n/o/p/q/r/s/t/u/v/w/x/y/z
# du sometimes completes its traversal before the above rename.
# Five iterations was not enough in 2 of 7 "make -j20 check" runs on a
# 6/12-core system. However, using "10", I saw no failure in 20 trials.
-# Using 10 iterations was not enough, either.
-# Using 50, I saw no failure in 200 trials.
-for i in $(seq 50); do
+# 2011 this was set at 50, 2025 this was set at 99
+for i in $(seq 99); do
mkdir -p $t/3/a/b/c/$i/$long || framework_failure_
done
@@ -78,7 +77,7 @@ cleanup_() { kill $pid 2>/dev/null && wait $pid; }
# Prohibit suspension, which could otherwise cause a timeout-induced FP failure.
trap '' TSTP
-timeout 6 ./inotify-watch-for-dir-access.py $t/3/a/b > start-msg & pid=$!
+timeout 10 ./inotify-watch-for-dir-access.py $t/3/a/b > start-msg & pid=$!
# Wait for the watcher to start...
nonempty() { sleep $1; test -s start-msg; }
--
2.50.1
