From: J. Bruce Fields bfields@redhat.com
[ Upstream commit ae3c57b5ca47665dc3416447a5534a9796096d86 ]
The nfsd open code has always kept separate read-only, read-write, and write-only opens as necessary to ensure that when a client closes or downgrades, we don't retain more access than necessary.
Also, I didn't realize the cache behaved this way when I wrote 94415b06eb8a "nfsd4: a client's own opens needn't prevent delegations". There I assumed fi_fds[O_WRONLY] and fi_fds[O_RDWR] would always be distinct. The violation of that assumption is triggering a WARN_ON_ONCE() and could also cause the server to give out a delegation when it shouldn't.
Fixes: 94415b06eb8a ("nfsd4: a client's own opens needn't prevent delegations") Tested-by: Chuck Lever chuck.lever@oracle.com Signed-off-by: J. Bruce Fields bfields@redhat.com Signed-off-by: Sasha Levin sashal@kernel.org --- fs/nfsd/filecache.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/fs/nfsd/filecache.c b/fs/nfsd/filecache.c index c8b9d2667ee6f..3c6c2f7d1688b 100644 --- a/fs/nfsd/filecache.c +++ b/fs/nfsd/filecache.c @@ -889,7 +889,7 @@ nfsd_file_find_locked(struct inode *inode, unsigned int may_flags,
hlist_for_each_entry_rcu(nf, &nfsd_file_hashtbl[hashval].nfb_head, nf_node, lockdep_is_held(&nfsd_file_hashtbl[hashval].nfb_lock)) { - if ((need & nf->nf_may) != need) + if (nf->nf_may != need) continue; if (nf->nf_inode != inode) continue;