On Mon, Aug 12, 2024 at 08:02:36PM +0100, Matthew Wilcox wrote: > On Mon, Aug 12, 2024 at 04:39:56PM +0000, Pankaj Raghav (Samsung) wrote: > > On Mon, Aug 12, 2024 at 08:31:01AM +0200, Christoph Hellwig wrote: > > > diff --git a/fs/xfs/libxfs/xfs_ag.c b/fs/xfs/libxfs/xfs_ag.c > > > index 7e80732cb54708..5efb1e8b4107a9 100644 > > > --- a/fs/xfs/libxfs/xfs_ag.c > > > +++ b/fs/xfs/libxfs/xfs_ag.c > > > @@ -46,7 +46,7 @@ xfs_perag_get( > > > struct xfs_perag *pag; > > > > > > rcu_read_lock(); > > xa_load() already calls rcu_read_lock(). So we can get rid of this I > > guess? > > Almost certainly not; I assume pag is RCU-freed, so you'd be introducing > a UAF if the RCU lock is dropped before getting a refcount on the pag. Ah, yes. rcu lock is needed until we get a refcount on pag.
