This is a note to let you know that I've just added the patch titled NFS: Don't revalidate the directory permissions on a lookup failure to the 5.11-stable tree which can be found at: http://www.kernel.org/git/?p=linux/kernel/git/stable/stable-queue.git;a=summary The filename of the patch is: nfs-don-t-revalidate-the-directory-permissions-on-a-.patch and it can be found in the queue-5.11 subdirectory. If you, or anyone else, feels it should not be added to the stable tree, please let <stable@xxxxxxxxxxxxxxx> know about it. commit 9d9f6ad45143661b3a7fe02f77a914aebc3b01c0 Author: Trond Myklebust <trond.myklebust@xxxxxxxxxxxxxxx> Date: Mon Mar 8 14:42:51 2021 -0500 NFS: Don't revalidate the directory permissions on a lookup failure [ Upstream commit 82e7ca1334ab16e2e04fafded1cab9dfcdc11b40 ] There should be no reason to expect the directory permissions to change just because the directory contents changed or a negative lookup timed out. So let's avoid doing a full call to nfs_mark_for_revalidate() in that case. Furthermore, if this is a negative dentry, and we haven't actually done a new lookup, then we have no reason yet to believe the directory has changed at all. So let's remove the gratuitous directory inode invalidation altogether when called from nfs_lookup_revalidate_negative(). Reported-by: Geert Jansen <gerardu@xxxxxxxxxx> Fixes: 5ceb9d7fdaaf ("NFS: Refactor nfs_lookup_revalidate()") Signed-off-by: Trond Myklebust <trond.myklebust@xxxxxxxxxxxxxxx> Signed-off-by: Anna Schumaker <Anna.Schumaker@xxxxxxxxxx> Signed-off-by: Sasha Levin <sashal@xxxxxxxxxx> diff --git a/fs/nfs/dir.c b/fs/nfs/dir.c index ef827ae193d2..7bcc6fcf1096 100644 --- a/fs/nfs/dir.c +++ b/fs/nfs/dir.c @@ -1401,6 +1401,15 @@ int nfs_lookup_verify_inode(struct inode *inode, unsigned int flags) goto out; } +static void nfs_mark_dir_for_revalidate(struct inode *inode) +{ + struct nfs_inode *nfsi = NFS_I(inode); + + spin_lock(&inode->i_lock); + nfsi->cache_validity |= NFS_INO_REVAL_PAGECACHE; + spin_unlock(&inode->i_lock); +} + /* * We judge how long we want to trust negative * dentries by looking at the parent inode mtime. @@ -1435,7 +1444,6 @@ nfs_lookup_revalidate_done(struct inode *dir, struct dentry *dentry, __func__, dentry); return 1; case 0: - nfs_mark_for_revalidate(dir); if (inode && S_ISDIR(inode->i_mode)) { /* Purge readdir caches. */ nfs_zap_caches(inode); @@ -1525,6 +1533,13 @@ nfs_lookup_revalidate_dentry(struct inode *dir, struct dentry *dentry, nfs_free_fattr(fattr); nfs_free_fhandle(fhandle); nfs4_label_free(label); + + /* + * If the lookup failed despite the dentry change attribute being + * a match, then we should revalidate the directory cache. + */ + if (!ret && nfs_verify_change_attribute(dir, dentry->d_time)) + nfs_mark_dir_for_revalidate(dir); return nfs_lookup_revalidate_done(dir, dentry, inode, ret); } @@ -1567,7 +1582,7 @@ nfs_do_lookup_revalidate(struct inode *dir, struct dentry *dentry, error = nfs_lookup_verify_inode(inode, flags); if (error) { if (error == -ESTALE) - nfs_zap_caches(dir); + nfs_mark_dir_for_revalidate(dir); goto out_bad; } nfs_advise_use_readdirplus(dir); @@ -2064,7 +2079,6 @@ nfs_add_or_obtain(struct dentry *dentry, struct nfs_fh *fhandle, dput(parent); return d; out_error: - nfs_mark_for_revalidate(dir); d = ERR_PTR(error); goto out; }