Re: [PATCH 01/34] namespace: take lock_mount_hash() directly when changing flags

On Sun, Nov 01, 2020 at 02:41:08PM +0000, Christoph Hellwig wrote:

quoted

index cebaa3e81794..20ee291a7af4 100644

--- a/fs/namespace.c
+++ b/fs/namespace.c

@@ -463,7 +463,6 @@ static int mnt_make_readonly(struct mount *mnt)
 {
 	int ret = 0;
 
-	lock_mount_hash();

What about adding a lockdep_assert_lock_held in all the functions
that used to take the lock to document the assumptions?

Good idea and will do. I wanted to do this but then didn't because I
haven't seen widespread use of lockdep assert in fs/namespace.c.

quoted

 static int __mnt_unmake_readonly(struct mount *mnt)
 {
-	lock_mount_hash();
 	mnt->mnt.mnt_flags &= ~MNT_READONLY;
-	unlock_mount_hash();
 	return 0;

This helper is rather pointless now.

Ok, will remove.

quoted

 static void set_mount_attributes(struct mount *mnt, unsigned int mnt_flags)
 {
-	lock_mount_hash();
 	mnt_flags |= mnt->mnt.mnt_flags & ~MNT_USER_SETTABLE_MASK;
 	mnt->mnt.mnt_flags = mnt_flags;
 	touch_mnt_namespace(mnt->mnt_ns);
-	unlock_mount_hash();

In linux-next there is an additional notify_mount after the unlock here.

Thanks! I can try rebasing on -next.

Also while you touch this lock_mount_hash/unlock_mount_hash could be
moved to namespace.c and maked static now.

Ok, will try to do that.