Re: [PATCH] nfs: check a crash in nfs_lookup_revalidate

[Peng Huang]

On Wed, May 11, 2011 at 5:08 PM, Trond Myklebust
<Trond.Myklebust@xxxxxxxxxx> wrote:
> On Wed, 2011-05-11 at 17:03 -0400, Peng Huang wrote:
>> lookup_one_len() may call nfs_loopup_revalidate() with nd == NULL
>> indirectly, that causes the kernel crash.
>>
>> RIP: 0010:[<ffffffffa0ba3b41>] Â[<ffffffffa0ba3b41>]
>> nfs_lookup_revalidate+0x21/0x4a0 [nfs]
>> RSP: 0018:ffff88018f00fae8 ÂEFLAGS: 00010286
>>
>> Call Trace:
>> Â[<ffffffff81164a17>] do_revalidate+0x17/0x60
>> Â[<ffffffff81164e9b>] __lookup_hash+0xcb/0x140
>> Â[<ffffffff811653c4>] lookup_one_len+0x94/0xe0
>> Â[<ffffffff81241ef1>] ecryptfs_lookup+0x91/0x1d0
>> Â[<ffffffff81164d85>] d_alloc_and_lookup+0x45/0x90
>> Â[<ffffffff8116f7b5>] ? d_lookup+0x35/0x60
>> Â[<ffffffff811669b2>] do_lookup+0x192/0x2d0
>> Â[<ffffffff811763be>] ? vfsmount_lock_local_unlock+0x1e/0x30
>> Â[<ffffffff8126d09c>] ? security_inode_permission+0x1c/0x30
>> Â[<ffffffff81167d67>] link_path_walk+0x597/0xae0
>> Â[<ffffffff8117638e>] ? vfsmount_lock_local_lock+0x1e/0x30
>> Â[<ffffffff81165905>] ? path_init_rcu+0xa5/0x210
>> Â[<ffffffff8116858b>] do_path_lookup+0x5b/0x140
>> Â[<ffffffff811692f7>] user_path_at+0x57/0xa0
>> Â[<ffffffff8159fd08>] ? do_page_fault+0x1e8/0x4e0
>> Â[<ffffffff8115eb86>] vfs_fstatat+0x46/0x80
>> Â[<ffffffff8116b990>] ? filldir+0x0/0xe0
>> Â[<ffffffff8115ec2e>] vfs_lstat+0x1e/0x20
>> Â[<ffffffff8115ec54>] sys_newlstat+0x24/0x50
>> Â[<ffffffff8159c995>] ? page_fault+0x25/0x30
>> Â[<ffffffff8100bfc2>] system_call_fastpath+0x16/0x1b
>>
>> Signed-off-by: Peng Huang <shawn.p.huang@xxxxxxxxx>
>> ---
>> Âfs/nfs/dir.c | Â Â2 +-
>> Â1 files changed, 1 insertions(+), 1 deletions(-)
>>
>> diff --git a/fs/nfs/dir.c b/fs/nfs/dir.c
>> index 2c3eb33..9452aa5 100644
>> --- a/fs/nfs/dir.c
>> +++ b/fs/nfs/dir.c
>> @@ -1028,7 +1028,7 @@ static int nfs_lookup_revalidate(struct dentry *dentry, struct nameidata *nd)
>> Â Â Â struct nfs_fattr *fattr = NULL;
>> Â Â Â int error;
>>
>> - Â Â if (nd->flags & LOOKUP_RCU)
>> + Â Â if (nd != NULL && nd->flags & LOOKUP_RCU)
>> Â Â Â Â Â Â Â return -ECHILD;
>>
>> Â Â Â parent = dget_parent(dentry);
>
> That's exactly what Tyler Hicks proposed last week and which was NACKed.
> We simply won't support layered filesystems that don't do intents.
>
> IOW: Feel free to change the above to.
>
> if (nd == NULL)
> Â Â return -EIO;
>

I tested returning -EIO when nd is NULL. Kernel does not crash, but
ecryptfs can not work on nfs anymore.

Peng Huang

>
>
>
> --
> Trond Myklebust
> Linux NFS client maintainer
>
> NetApp
> Trond.Myklebust@xxxxxxxxxx
> www.netapp.com
>
>
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/