Fix race in parallel mount's thread dispatching algorithm

Strategy of parallel mount is as follows.

1) Initial thread dispatching is to select sets of mount points that
 don't have dependencies on other sets, hence threads can/should run
 lock-less and shouldn't race with other threads for other sets. Each
 thread dispatched corresponds to top level directory which may or may
 not have datasets to be mounted on sub directories.

2) Subsequent recursive thread dispatching for each thread from 1)
 is to mount datasets for each set of mount points. The mount points
 within each set have dependencies (i.e. child directories), so child
 directories are processed only after parent directory completes.

The problem is that the initial thread dispatching in
zfs_foreach_mountpoint() can be multi-threaded when it needs to be
single-threaded, and this puts threads under race condition. This race
appeared as mount/unmount issues on ZoL for ZoL having different
timing regarding mount(2) execution due to fork(2)/exec(2) of mount(8).
`zfs unmount -a` which expects proper mount order can't unmount if the
mounts were reordered by the race condition.

There are currently two known patterns of input list `handles` in
`zfs_foreach_mountpoint(..,handles,..)` which cause the race condition.

1) #8833 case where input is `/a /a /a/b` after sorting.
 The problem is that libzfs_path_contains() can't correctly handle an
 input list with two same top level directories.
 There is a race between two POSIX threads A and B,
  * ThreadA for "/a" for test1 and "/a/b"
  * ThreadB for "/a" for test0/a
 and in case of #8833, ThreadA won the race. Two threads were created
 because "/a" wasn't considered as `"/a" contains "/a"`.

2) #8450 case where input is `/ /var/data /var/data/test` after sorting.
 The problem is that libzfs_path_contains() can't correctly handle an
 input list containing "/".
 There is a race between two POSIX threads A and B,
  * ThreadA for "/" and "/var/data/test"
  * ThreadB for "/var/data"
 and in case of #8450, ThreadA won the race. Two threads were created
 because "/var/data" wasn't considered as `"/" contains "/var/data"`.
 In other words, if there is (at least one) "/" in the input list,
 the initial thread dispatching must be single-threaded since every
 directory is a child of "/", meaning they all directly or indirectly
 depend on "/".

In both cases, the first non_descendant_idx() call fails to correctly
determine "path1-contains-path2", and as a result the initial thread
dispatching creates another thread when it needs to be single-threaded.
Fix a conditional in libzfs_path_contains() to consider above two.

Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed by: Sebastien Roy <sebastien.roy@delphix.com>
Signed-off-by: Tomohiro Kusumi <kusumi.tomohiro@gmail.com>
Closes #8450
Closes #8833
Closes #8878

diff --git a/lib/libzfs/libzfs_mount.c b/lib/libzfs/libzfs_mount.c
index 7497a72233ad5ceaf3f8a2cd4a68f1955ba74203..5bd3c67bec5e082f09631343eae915156ae79329 100644 (file)
--- a/lib/libzfs/libzfs_mount.c
+++ b/lib/libzfs/libzfs_mount.c
@@ -1336,12 +1336,14 @@ mountpoint_cmp(const void *arga, const void *argb)
 }
 
 /*
- * Return true if path2 is a child of path1.
+ * Return true if path2 is a child of path1 or path2 equals path1 or
+ * path1 is "/" (path2 is always a child of "/").
  */
 static boolean_t
 libzfs_path_contains(const char *path1, const char *path2)
 {
-       return (strstr(path2, path1) == path2 && path2[strlen(path1)] == '/');
+       return (strcmp(path1, path2) == 0 || strcmp(path1, "/") == 0 ||
+           (strstr(path2, path1) == path2 && path2[strlen(path1)] == '/'));
 }
 
 /*

diff --git a/tests/runfiles/linux.run b/tests/runfiles/linux.run
index d8b2a0b42b313d80587fb9bf2e7d863dd1ee5990..43c7a748fcf662a350f9c64869f3bb4fa978fae0 100644 (file)
--- a/tests/runfiles/linux.run
+++ b/tests/runfiles/linux.run
@@ -182,7 +182,8 @@ tests = ['zfs_mount_001_pos', 'zfs_mount_002_pos', 'zfs_mount_003_pos',
     'zfs_mount_007_pos', 'zfs_mount_008_pos', 'zfs_mount_009_neg',
     'zfs_mount_010_neg', 'zfs_mount_011_neg', 'zfs_mount_012_neg',
     'zfs_mount_all_001_pos', 'zfs_mount_encrypted', 'zfs_mount_remount',
-    'zfs_multi_mount', 'zfs_mount_all_fail', 'zfs_mount_all_mountpoints']
+    'zfs_multi_mount', 'zfs_mount_all_fail', 'zfs_mount_all_mountpoints',
+    'zfs_mount_test_race']
 tags = ['functional', 'cli_root', 'zfs_mount']
 
 [tests/functional/cli_root/zfs_program]

diff --git a/tests/zfs-tests/tests/functional/cli_root/zfs_mount/Makefile.am b/tests/zfs-tests/tests/functional/cli_root/zfs_mount/Makefile.am
index b2de98934b7496395bae97c4f0b897e06f866683..c208a1c378dc1fa6859ad886eb2340a8ade90f42 100644 (file)
--- a/tests/zfs-tests/tests/functional/cli_root/zfs_mount/Makefile.am
+++ b/tests/zfs-tests/tests/functional/cli_root/zfs_mount/Makefile.am
@@ -19,6 +19,7 @@ dist_pkgdata_SCRIPTS = \
        zfs_mount_all_mountpoints.ksh \
        zfs_mount_encrypted.ksh \
        zfs_mount_remount.ksh \
+       zfs_mount_test_race.sh \
        zfs_multi_mount.ksh
 
 dist_pkgdata_DATA = \

diff --git a/tests/zfs-tests/tests/functional/cli_root/zfs_mount/zfs_mount_test_race.sh b/tests/zfs-tests/tests/functional/cli_root/zfs_mount/zfs_mount_test_race.sh
new file mode 100755 (executable)
index 0000000..404770b
--- /dev/null
+++ b/tests/zfs-tests/tests/functional/cli_root/zfs_mount/zfs_mount_test_race.sh
@@ -0,0 +1,116 @@
+#!/bin/ksh
+
+#
+# This file and its contents are supplied under the terms of the
+# Common Development and Distribution License ("CDDL"), version 1.0.
+# You may only use this file in accordance with the terms of version
+# 1.0 of the CDDL.
+#
+# A full copy of the text of the CDDL should have accompanied this
+# source.  A copy of the CDDL is also available via the Internet at
+# http://www.illumos.org/license/CDDL.
+#
+
+#
+# Copyright (c) 2019 by Tomohiro Kusumi. All rights reserved.
+#
+
+. $STF_SUITE/include/libtest.shlib
+. $STF_SUITE/tests/functional/cli_root/zfs_mount/zfs_mount.cfg
+
+#
+# DESCRIPTION:
+# Verify parallel mount ordering is consistent.
+#
+# There was a bug in initial thread dispatching algorithm which put threads
+# under race condition which resulted in undefined mount order.  The purpose
+# of this test is to verify `zfs unmount -a` succeeds (not `zfs mount -a`
+# succeeds, it always does) after `zfs mount -a`, which could fail if threads
+# race.  See github.com/zfsonlinux/zfs/issues/{8450,8833,8878} for details.
+#
+# STRATEGY:
+# 1. Create pools and filesystems.
+# 2. Set same mount point for >1 datasets.
+# 3. Unmount all datasets.
+# 4. Mount all datasets.
+# 5. Unmount all datasets (verify this succeeds).
+#
+
+verify_runnable "both"
+
+TMPDIR=${TMPDIR:-$TEST_BASE_DIR}
+MNTPT=$TMPDIR/zfs_mount_test_race_mntpt
+DISK1="$TMPDIR/zfs_mount_test_race_disk1"
+DISK2="$TMPDIR/zfs_mount_test_race_disk2"
+
+TESTPOOL1=zfs_mount_test_race_tp1
+TESTPOOL2=zfs_mount_test_race_tp2
+
+export __ZFS_POOL_RESTRICT="$TESTPOOL1 $TESTPOOL2"
+log_must zfs $unmountall
+unset __ZFS_POOL_RESTRICT
+
+function cleanup
+{
+       zpool destroy $TESTPOOL1
+       zpool destroy $TESTPOOL2
+       rm -rf $MNTPT
+       rm -rf /$TESTPOOL1
+       rm -rf /$TESTPOOL2
+       rm -f $DISK1
+       rm -f $DISK2
+       export __ZFS_POOL_RESTRICT="$TESTPOOL1 $TESTPOOL2"
+       log_must zfs $mountall
+       unset __ZFS_POOL_RESTRICT
+}
+log_onexit cleanup
+
+log_note "Verify parallel mount ordering is consistent"
+
+log_must truncate -s $MINVDEVSIZE $DISK1
+log_must truncate -s $MINVDEVSIZE $DISK2
+
+log_must zpool create -f $TESTPOOL1 $DISK1
+log_must zpool create -f $TESTPOOL2 $DISK2
+
+log_must zfs create $TESTPOOL1/$TESTFS1
+log_must zfs create $TESTPOOL2/$TESTFS2
+
+log_must zfs set mountpoint=none $TESTPOOL1
+log_must zfs set mountpoint=$MNTPT $TESTPOOL1/$TESTFS1
+
+# Note that unmount can fail (due to race condition on `zfs mount -a`) with or
+# without `canmount=off`.  The race has nothing to do with canmount property,
+# but turn it off for convenience of mount layout used in this test case.
+log_must zfs set canmount=off $TESTPOOL2
+log_must zfs set mountpoint=$MNTPT $TESTPOOL2
+
+# At this point, layout of datasets in two pools will look like below.
+# Previously, on next `zfs mount -a`, pthreads assigned to TESTFS1 and TESTFS2
+# could race, and TESTFS2 usually (actually always) won in ZoL.  Note that the
+# problem is how two or more threads could initially be assigned to the same
+# top level directory, not this specific layout.  This layout is just an example
+# that can reproduce race, and is also the layout reported in #8833.
+#
+# NAME                  MOUNTED  MOUNTPOINT
+# ----------------------------------------------
+# /$TESTPOOL1           no       none
+# /$TESTPOOL1/$TESTFS1  yes      $MNTPT
+# /$TESTPOOL2           no       $MNTPT
+# /$TESTPOOL2/$TESTFS2  yes      $MNTPT/$TESTFS2
+
+# Apparently two datasets must be mounted.
+log_must ismounted $TESTPOOL1/$TESTFS1
+log_must ismounted $TESTPOOL2/$TESTFS2
+# This unmount always succeeds, because potential race hasn't happened yet.
+log_must zfs unmount -a
+# This mount always succeeds, whether threads are under race condition or not.
+log_must zfs mount -a
+
+# Verify datasets are mounted (TESTFS2 fails if the race broke mount order).
+log_must ismounted $TESTPOOL1/$TESTFS1
+log_must ismounted $TESTPOOL2/$TESTFS2
+# Verify unmount succeeds (fails if the race broke mount order).
+log_must zfs unmount -a
+
+log_pass "Verify parallel mount ordering is consistent passed"