All of lore.kernel.org
 help / color / mirror / Atom feed
From: Al Viro <viro@zeniv.linux.org.uk>
To: linux-fsdevel@vger.kernel.org
Cc: Christian Brauner <brauner@kernel.org>,
	Christoph Hellwig <hch@lst.de>,
	Linus Torvalds <torvalds@linux-foundation.org>,
	Namjae Jeon <linkinjeon@kernel.org>,
	David Sterba <dsterba@suse.com>,
	David Howells <dhowells@redhat.com>,
	Miklos Szeredi <miklos@szeredi.hu>,
	Amir Goldstein <amir73il@gmail.com>,
	Trond Myklebust <trond.myklebust@hammerspace.com>,
	Bob Peterson <rpeterso@redhat.com>,
	Steve French <sfrench@samba.org>,
	Luis Chamberlain <mcgrof@kernel.org>
Subject: [PATCH 09/15] nfs: make nfs_set_verifier() safe for use in RCU pathwalk
Date: Mon, 2 Oct 2023 03:34:16 +0100	[thread overview]
Message-ID: <20231002023416.GJ3389589@ZenIV> (raw)
In-Reply-To: <20231002022846.GA3389589@ZenIV>

nfs_set_verifier() relies upon dentry being pinned; if that's
the case, grabbing ->d_lock stabilizes ->d_parent and guarantees
that ->d_parent points to a positive dentry.  For something
we'd run into in RCU mode that is *not* true - dentry might've
been through dentry_kill() just as we grabbed ->d_lock, with
its parent going through the same just as we get to into
nfs_set_verifier_locked().  It might get to detaching inode
(and zeroing ->d_inode) before nfs_set_verifier_locked() gets
to fetching that; we get an oops as the result.

That can happen in nfs{,4} ->d_revalidate(); we check that
parent is positive, but that's done before we get to
nfs_set_verifier() and it's possible for memory pressure
to pick our dentry as eviction candidate by that time.
If that happens, back-to-back attempts to kill dentry and
its parent are quite normal.

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
---
 fs/nfs/dir.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/fs/nfs/dir.c b/fs/nfs/dir.c
index e6a51fd94fea..8ffc1f78ba51 100644
--- a/fs/nfs/dir.c
+++ b/fs/nfs/dir.c
@@ -1431,9 +1431,9 @@ static bool nfs_verifier_is_delegated(struct dentry *dentry)
 static void nfs_set_verifier_locked(struct dentry *dentry, unsigned long verf)
 {
 	struct inode *inode = d_inode(dentry);
-	struct inode *dir = d_inode(dentry->d_parent);
+	struct inode *dir = d_inode_rcu(dentry->d_parent);
 
-	if (!nfs_verify_change_attribute(dir, verf))
+	if (!dir || !nfs_verify_change_attribute(dir, verf))
 		return;
 	if (inode && NFS_PROTO(inode)->have_delegation(inode, FMODE_READ))
 		nfs_set_verifier_delegated(&verf);
-- 
2.39.2


  parent reply	other threads:[~2023-10-02  2:34 UTC|newest]

Thread overview: 38+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2023-10-02  2:28 [RFC][PATCHES] fixes in methods exposed to rcu pathwalk Al Viro
2023-10-02  2:28 ` Al Viro
2023-10-02  2:29   ` [PATCH 01/15] rcu pathwalk: prevent bogus hard errors from may_lookup() Al Viro
2023-10-02  2:30   ` [PATCH 02/15] exfat: move freeing sbi, upcase table and dropping nls into rcu-delayed helper Al Viro
2023-10-02 16:10     ` Linus Torvalds
2023-10-02 18:04       ` Al Viro
2023-10-02  2:30   ` [PATCH 03/15] affs: free affs_sb_info with kfree_rcu() Al Viro
2023-10-02  2:31   ` [PATCH 04/15] hfsplus: switch to rcu-delayed unloading of nls and freeing ->s_fs_info Al Viro
2023-10-02  6:49     ` Christoph Hellwig
2023-10-02  7:14       ` Al Viro
2023-10-02  7:21         ` Al Viro
2023-10-02 18:09           ` Al Viro
2023-10-04 19:04             ` Linus Torvalds
2023-10-04 19:06               ` Linus Torvalds
2023-10-02  2:31   ` [PATCH 05/15] cifs_get_link(): bail out in unsafe case Al Viro
2023-10-02  2:32   ` [PATCH 06/15] procfs: move dropping pde and pid from ->evict_inode() to ->free_inode() Al Viro
2023-10-02  2:33   ` [PATCH 07/15] procfs: make freeing proc_fs_info rcu-delayed Al Viro
2023-10-02  2:33   ` [PATCH 08/15] gfs2: fix an oops in gfs2_permission() Al Viro
2023-10-02 11:46     ` Bob Peterson
2023-10-02 12:59       ` Al Viro
2023-10-02 14:16         ` Al Viro
2023-10-03 14:46           ` Andreas Grünbacher
2023-10-02  2:34   ` Al Viro [this message]
2023-10-02  2:34   ` [PATCH 10/15] nfs: fix UAF on pathwalk running into umount Al Viro
2023-10-02  2:35   ` [PATCH 11/15] fuse: fix UAF in rcu pathwalks Al Viro
2023-10-02  2:35   ` [PATCH 12/15] afs: fix __afs_break_callback() / afs_drop_open_mmap() race Al Viro
2023-10-02  2:36   ` [PATCH 13/15] overlayfs: move freeing ovl_entry past rcu delay Al Viro
2023-10-02  2:36     ` [PATCH 14/15] ovl_dentry_revalidate_common(): fetch inode once Al Viro
2023-10-02  2:37       ` [PATCH 15/15] overlayfs: make use of ->layers safe in rcu pathwalk Al Viro
2023-10-02  6:40         ` Amir Goldstein
2023-10-02  7:23           ` Al Viro
2023-10-02  8:53             ` Amir Goldstein
2023-10-03 20:47               ` Al Viro
2023-10-02  5:47       ` [PATCH 14/15] ovl_dentry_revalidate_common(): fetch inode once Amir Goldstein
2023-10-02  5:56         ` Amir Goldstein
2023-10-02 14:47           ` Amir Goldstein
2023-10-02  5:51     ` [PATCH 13/15] overlayfs: move freeing ovl_entry past rcu delay Amir Goldstein
2023-10-02  2:52   ` [RFC][PATCHES] fixes in methods exposed to rcu pathwalk Al Viro

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20231002023416.GJ3389589@ZenIV \
    --to=viro@zeniv.linux.org.uk \
    --cc=amir73il@gmail.com \
    --cc=brauner@kernel.org \
    --cc=dhowells@redhat.com \
    --cc=dsterba@suse.com \
    --cc=hch@lst.de \
    --cc=linkinjeon@kernel.org \
    --cc=linux-fsdevel@vger.kernel.org \
    --cc=mcgrof@kernel.org \
    --cc=miklos@szeredi.hu \
    --cc=rpeterso@redhat.com \
    --cc=sfrench@samba.org \
    --cc=torvalds@linux-foundation.org \
    --cc=trond.myklebust@hammerspace.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.