On Wed, 28 Feb 2024 at 17:34, Bernd Schubert bernd.schubert@fastmail.fm wrote:
On 2/28/24 17:02, Miklos Szeredi wrote:
The root inode is assumed to be always hashed. Do not unhash the root inode even if it is marked BAD.
Fixes: 5d069dbe8aaf ("fuse: fix bad inode") Cc: stable@vger.kernel.org # v5.11 Signed-off-by: Miklos Szeredi mszeredi@redhat.com
fs/fuse/fuse_i.h | 1 - fs/fuse/inode.c | 7 +++++-- 2 files changed, 5 insertions(+), 3 deletions(-)
diff --git a/fs/fuse/fuse_i.h b/fs/fuse/fuse_i.h index 7bd3552b1e80..4ef6087f0e5c 100644 --- a/fs/fuse/fuse_i.h +++ b/fs/fuse/fuse_i.h @@ -994,7 +994,6 @@ static inline bool fuse_stale_inode(const struct inode *inode, int generation,
static inline void fuse_make_bad(struct inode *inode) {
remove_inode_hash(inode); set_bit(FUSE_I_BAD, &get_fuse_inode(inode)->state);}
Hmm, what about callers like fuse_direntplus_link? It now never removes the inode hash for these? Depend on lookup/revalidate?
Good questions.
In that case the dentry will be unhashed, and after retrying it will go through fuse_iget(), which will unhash the inode.
So AFAICS the only place the inode needs to be unhashed is in fuse_iget(), which is the real fix in 775c5033a0d1 ("fuse: fix live lock in fuse_iget()").
Thanks, Miklos