If the test is executed on NFS, it may print:
rm: cannot remove `/mnt/ltp-xeJYpESVKz/acltest01.fCogoo4gn0': Directory not
empty
It seems the following happen:
1. After 'unmount -d' 'acltest01.fCogoo4gn0/blkext3' is still in use by udev.
See kernel commit a1ecac3b0656a68259927c234e505804d33a7b83
("loop: Make explicit loop device destruction lazy")
2. Due to the "NFS silly rename" feature, command 'rm -f
acltest01.fCogoo4gn0/blkext3'
renames 'blkext3' to '.nfs*'
3. The removal of 'acltest01.fCogoo4gn0' directory fails, because it
still contains this '.nfs*' file.
Intorucing a wait cycle should fix this problem.
Signed-off-by: Stanislav Kholmanskikh <stanislav.kholmansk...@oracle.com>
---
It's a follow up to:
http://sourceforge.net/p/ltp/mailman/message/32987268/
Indeed, 'sync' was wrong there.
testcases/kernel/fs/acls/acl_test01 | 16 ++++++++++++++++
1 files changed, 16 insertions(+), 0 deletions(-)
diff --git a/testcases/kernel/fs/acls/acl_test01
b/testcases/kernel/fs/acls/acl_test01
index 4dc15ef..b4c2c70 100755
--- a/testcases/kernel/fs/acls/acl_test01
+++ b/testcases/kernel/fs/acls/acl_test01
@@ -99,6 +99,22 @@ do_cleanup() {
MOUNT_POINT=$(readlink -f "mount-ext3")
mount | grep -q $MOUNT_POINT && umount -d $MOUNT_POINT
+ # Cope with "NFS silly rename"
+ if [ $(tst_fs_type .) = "NFS" ]; then
+ rm -f blkext3
+
+ TIMEOUT=5
+ while $(ls -la | grep -q .nfs); do
+ sleep 1
+
+ TIMEOUT=$(( $TIMEOUT - 1 ))
+ if [ $TIMEOUT -eq 0 ]; then
+ tst_resm TBROK "The loop device is still in use"
+ break
+ fi
+ done
+ fi
+
tst_rmdir
# We set it back to Enforcing.
--
1.7.1
------------------------------------------------------------------------------
_______________________________________________
Ltp-list mailing list
Ltp-list@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/ltp-list
