[PATCH] NFS: Fix cache consistency races

If the data cache has been marked as potentially invalid by nfs_refresh_inode,
we should invalidate it rather than assume that changes are due to our own
activity.

Also ensure that we always start with a valid cache before declaring it
to be protected by a delegation.

Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

1 files changed, 2 insertions, 5 deletions

diff --git a/fs/nfs/inode.c b/fs/nfs/inode.c
index 6922469d6fc5..6be46d21c01e 100644
--- a/fs/nfs/inode.c
+++ b/fs/nfs/inode.c
@@ -1226,10 +1226,6 @@ int nfs_refresh_inode(struct inode *inode, struct nfs_fattr *fattr)
         loff_t cur_size, new_isize;
         int data_unstable;
 
-        /* Do we hold a delegation? */
-        if (nfs_have_delegation(inode, FMODE_READ))
-                return 0;
-
         spin_lock(&inode->i_lock);
 
         /* Are we in the process of updating data on the server? */
@@ -1350,7 +1346,8 @@ static int nfs_update_inode(struct inode *inode, struct nfs_fattr *fattr, unsign
         nfsi->read_cache_jiffies = fattr->timestamp;
 
         /* Are we racing with known updates of the metadata on the server? */
-        data_unstable = ! nfs_verify_change_attribute(inode, verifier);
+        data_unstable = ! (nfs_verify_change_attribute(inode, verifier) ||
+                (nfsi->cache_validity & NFS_INO_REVAL_PAGECACHE));
 
         /* Check if our cached file size is stale */
         new_isize = nfs_size_to_loff_t(fattr->size);