On Jan 13, 2014, at 2:55, shaobingqing <shaobingqing@xxxxxxxxxxxxx> wrote:
> When a file is in NFS_INO_LAYOUTCOMMITING status, its isize perhaps has not been
> transferred to the metadate server. So the isize getting from the metadata server
> perhaps is out of date and cannot be used to update the isize of the client.
>
> Signed-off-by: shaobingqing <shaobingqing@xxxxxxxxxxxxx>
> ---
> fs/nfs/inode.c | 3 ++-
> 1 files changed, 2 insertions(+), 1 deletions(-)
>
> diff --git a/fs/nfs/inode.c b/fs/nfs/inode.c
> index ebeb94c..caf50a1 100644
> --- a/fs/nfs/inode.c
> +++ b/fs/nfs/inode.c
> @@ -1402,7 +1402,8 @@ static int nfs_update_inode(struct inode *inode, struct nfs_fattr *fattr)
> if (new_isize != cur_isize) {
> /* Do we perhaps have any outstanding writes, or has
> * the file grown beyond our last write? */
> - if ((nfsi->npages == 0 && !test_bit(NFS_INO_LAYOUTCOMMIT, &nfsi->flags)) ||
> + if ((nfsi->npages == 0 && !test_bit(NFS_INO_LAYOUTCOMMIT, &nfsi->flags) &&
> + !test_bit(NFS_INO_LAYOUTCOMMITTING, &nfsi->flags)) ||
> new_isize > cur_isize) {
> i_size_write(inode, new_isize);
> invalid |= NFS_INO_INVALID_ATTR|NFS_INO_INVALID_DATA;
Ugh. Is there any reason why we can’t just keep NFS_INO_LAYOUTCOMMIT set until the layout commit operation is finished instead of multiplying the tests for it in generic NFS code paths such as nfs_update_inode?
--
Trond Myklebust
Linux NFS client maintainer
--
To unsubscribe from this list: send the line "unsubscribe linux-nfs" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
