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 nofs group lock
helpers.
Suggested-by: Jan Kara <jack@xxxxxxx>
Link: https://lore.kernel.org/r/20220321112310.vpr7oxro2xkz5llh@xxxxxxxxxx/
Signed-off-by: Amir Goldstein <amir73il@xxxxxxxxx>
---
fs/notify/dnotify/dnotify.c | 13 ++++++++-----
1 file changed, 8 insertions(+), 5 deletions(-)
diff --git a/fs/notify/dnotify/dnotify.c b/fs/notify/dnotify/dnotify.c
index 829dd4a61b66..b291141104ea 100644
--- a/fs/notify/dnotify/dnotify.c
+++ b/fs/notify/dnotify/dnotify.c
@@ -139,6 +139,7 @@ static void dnotify_free_mark(struct fsnotify_mark *fsn_mark)
}
static const struct fsnotify_ops dnotify_fsnotify_ops = {
+ .group_flags = FSNOTIFY_GROUP_NOFS,
.handle_inode_event = dnotify_handle_event,
.free_mark = dnotify_free_mark,
};
@@ -158,6 +159,7 @@ void dnotify_flush(struct file *filp, fl_owner_t id)
struct dnotify_struct **prev;
struct inode *inode;
bool free = false;
+ unsigned int nofs;
inode = file_inode(filp);
if (!S_ISDIR(inode->i_mode))
@@ -168,7 +170,7 @@ 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);
+ nofs = fsnotify_group_nofs_lock(dnotify_group);
spin_lock(&fsn_mark->lock);
prev = &dn_mark->dn;
@@ -191,7 +193,7 @@ void dnotify_flush(struct file *filp, fl_owner_t id)
free = true;
}
- mutex_unlock(&dnotify_group->mark_mutex);
+ fsnotify_group_nofs_unlock(dnotify_group, nofs);
if (free)
fsnotify_free_mark(fsn_mark);
@@ -267,6 +269,7 @@ int fcntl_dirnotify(int fd, struct file *filp, unsigned long arg)
fl_owner_t id = current->files;
struct file *f;
int destroy = 0, error = 0;
+ unsigned int nofs;
__u32 mask;
/* we use these to tell if we need to kfree */
@@ -324,7 +327,7 @@ 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);
+ nofs = fsnotify_group_nofs_lock(dnotify_group);
/* add the new_fsn_mark or find an old one. */
fsn_mark = fsnotify_find_mark(&inode->i_fsnotify_marks, dnotify_group);
@@ -334,7 +337,7 @@ 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_nofs_unlock(dnotify_group, nofs);
goto out_err;
}
spin_lock(&new_fsn_mark->lock);
@@ -383,7 +386,7 @@ 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_nofs_unlock(dnotify_group, nofs);
if (destroy)
fsnotify_free_mark(fsn_mark);
fsnotify_put_mark(fsn_mark);
--
2.25.1
