c/mirror_zfs
1
0
Fork 0
mirror of https://git.proxmox.com/git/mirror_zfs.git synced 2026-05-22 10:37:35 +03:00
Code Issues Packages Projects Releases Wiki Activity

Fix snapshot automount expiry cancellation deadlock

Browse Source 
A deadlock occurs when snapshot expiry tasks are cancelled while holding
locks. The snapshot expiry task (snapentry_expire) spawns an umount
process and waits for it to complete. Concurrently, ARC memory pressure
triggers arc_prune which calls zfs_exit_fs(), attempting to cancel the
expiry task while holding locks. The umount process spawned by the
expiry task blocks trying to acquire locks held by arc_prune, which is
blocked waiting for the expiry task to complete. This creates a circular
dependency: expiry task waits for umount, umount waits for arc_prune,
arc_prune waits for expiry task.

Fix by adding non-blocking cancellation support to taskq_cancel_id().
The zfs_exit_fs() path calls zfsctl_snapshot_unmount_delay() to
reschedule the unmount, which needs to cancel any existing expiry task.
It now uses non-blocking cancellation to avoid waiting while holding
locks, breaking the deadlock by returning immediately when the task is
already running.

The per-entry se_taskqid_lock has been removed, with all taskqid
operations now protected by the global zfs_snapshot_lock held as
WRITER. Additionally, an se_in_umount flag prevents recursive waits when
zfsctl_destroy() is called during unmount. The taskqid is now only
cleared by the caller on successful cancellation; running tasks clear
their own taskqid upon completion.

Reviewed-by: Alexander Motin <alexander.motin@TrueNAS.com>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Ameer Hamza <ahamza@ixsystems.com>
Closes #17941
This commit is contained in:
Ameer Hamza
2025-12-02 03:43:42 +05:00
committed by Brian Behlendorf
parent 663dc86de2
commit 74bbdda1ef
12 changed files with 69 additions and 48 deletions
Download Patch File Download Diff File Expand all files Collapse all files
module/os/linux/spl/spl-kmem-cache.c
+1 -1
View File
@@ -840,7 +840,7 @@ spl_kmem_cache_destroy(spl_kmem_cache_t *skc)
id = skc->skc_taskqid;
spin_unlock(&skc->skc_lock);
taskq_cancel_id(spl_kmem_cache_taskq, id);
taskq_cancel_id(spl_kmem_cache_taskq, id, B_TRUE);
/*
* Wait until all current callers complete, this is mainly
module/os/linux/spl/spl-taskq.c
+20 -7
View File
@@ -600,13 +600,22 @@ taskq_of_curthread(void)
EXPORT_SYMBOL(taskq_of_curthread);
/*
* Cancel an already dispatched task given the task id. Still pending tasks
* will be immediately canceled, and if the task is active the function will
* block until it completes. Preallocated tasks which are canceled must be
* freed by the caller.
* Cancel a dispatched task. Pending tasks are cancelled immediately.
* If the task is running, behavior depends on wait parameter:
* - wait=B_TRUE: Block until task completes
* - wait=B_FALSE: Return EBUSY immediately
*
* Return values:
* 0 - Cancelled before execution. Caller must release resources.
* EBUSY - Task running (wait=B_FALSE only). Will self-cleanup.
* ENOENT - Not found, or completed after waiting. Already cleaned up.
*
* Note: wait=B_TRUE returns ENOENT (not EBUSY) after waiting because
* the task no longer exists. This distinguishes "cancelled before run"
* from "completed naturally" for proper resource management.
*/
int
taskq_cancel_id(taskq_t *tq, taskqid_t id)
taskq_cancel_id(taskq_t *tq, taskqid_t id, boolean_t wait)
{
taskq_ent_t *t;
int rc = ENOENT;
@@ -669,8 +678,12 @@ taskq_cancel_id(taskq_t *tq, taskqid_t id)
spin_unlock_irqrestore(&tq->tq_lock, flags);
if (t == ERR_PTR(-EBUSY)) {
taskq_wait_id(tq, id);
rc = EBUSY;
if (wait) {
taskq_wait_id(tq, id);
rc = ENOENT; /* Completed, no longer exists */
} else {
rc = EBUSY; /* Still running */
}
}
return (rc);