5.15-stable review patch. If anyone has any objections, please let me know.
------------------
From: Amir Goldstein amir73il@gmail.com
[ Upstream commit aabb45fdcb31f00f1e7cae2bce83e83474a87c03 ]
Before commit 9542e6a643fc6 ("nfsd: Containerise filecache laundrette") nfsd would close open files in direct reclaim context. There is no guarantee that others memory shrinkers don't do the same and no guarantee that future shrinkers won't do that.
For example, if overlayfs implements inode cache of fscache would keep open files to cached objects, inode shrinkers could end up closing open files to underlying fs.
Direct reclaim from dnotify mark allocation context may try to close open files that have dnotify marks of the same group and hit a deadlock on mark_mutex.
Set the FSNOTIFY_GROUP_NOFS flag to prevent going into direct reclaim from allocations under dnotify group lock and use the safe group lock helpers.
Link: https://lore.kernel.org/r/20220422120327.3459282-11-amir73il@gmail.com Suggested-by: Jan Kara jack@suse.cz Link: https://lore.kernel.org/r/20220321112310.vpr7oxro2xkz5llh@quack3.lan/ Signed-off-by: Amir Goldstein amir73il@gmail.com Signed-off-by: Jan Kara jack@suse.cz Signed-off-by: Chuck Lever chuck.lever@oracle.com --- fs/notify/dnotify/dnotify.c | 13 +++++++------ 1 file changed, 7 insertions(+), 6 deletions(-)
diff --git a/fs/notify/dnotify/dnotify.c b/fs/notify/dnotify/dnotify.c index 6c586802c50e6..fa81c59a2ad41 100644 --- a/fs/notify/dnotify/dnotify.c +++ b/fs/notify/dnotify/dnotify.c @@ -150,7 +150,7 @@ void dnotify_flush(struct file *filp, fl_owner_t id) return; dn_mark = container_of(fsn_mark, struct dnotify_mark, fsn_mark);
- mutex_lock(&dnotify_group->mark_mutex); + fsnotify_group_lock(dnotify_group);
spin_lock(&fsn_mark->lock); prev = &dn_mark->dn; @@ -173,7 +173,7 @@ void dnotify_flush(struct file *filp, fl_owner_t id) free = true; }
- mutex_unlock(&dnotify_group->mark_mutex); + fsnotify_group_unlock(dnotify_group);
if (free) fsnotify_free_mark(fsn_mark); @@ -306,7 +306,7 @@ int fcntl_dirnotify(int fd, struct file *filp, unsigned long arg) new_dn_mark->dn = NULL;
/* this is needed to prevent the fcntl/close race described below */ - mutex_lock(&dnotify_group->mark_mutex); + fsnotify_group_lock(dnotify_group);
/* add the new_fsn_mark or find an old one. */ fsn_mark = fsnotify_find_mark(&inode->i_fsnotify_marks, dnotify_group); @@ -316,7 +316,7 @@ int fcntl_dirnotify(int fd, struct file *filp, unsigned long arg) } else { error = fsnotify_add_inode_mark_locked(new_fsn_mark, inode, 0); if (error) { - mutex_unlock(&dnotify_group->mark_mutex); + fsnotify_group_unlock(dnotify_group); goto out_err; } spin_lock(&new_fsn_mark->lock); @@ -365,7 +365,7 @@ int fcntl_dirnotify(int fd, struct file *filp, unsigned long arg)
if (destroy) fsnotify_detach_mark(fsn_mark); - mutex_unlock(&dnotify_group->mark_mutex); + fsnotify_group_unlock(dnotify_group); if (destroy) fsnotify_free_mark(fsn_mark); fsnotify_put_mark(fsn_mark); @@ -383,7 +383,8 @@ static int __init dnotify_init(void) SLAB_PANIC|SLAB_ACCOUNT); dnotify_mark_cache = KMEM_CACHE(dnotify_mark, SLAB_PANIC|SLAB_ACCOUNT);
- dnotify_group = fsnotify_alloc_group(&dnotify_fsnotify_ops, 0); + dnotify_group = fsnotify_alloc_group(&dnotify_fsnotify_ops, + FSNOTIFY_GROUP_NOFS); if (IS_ERR(dnotify_group)) panic("unable to allocate fsnotify group for dnotify\n"); return 0;