On Tue, 2012-03-20 at 12:25 -0400, Fred Isaman wrote:
> The decrement is handled by each call to nfs_request_remove_commit_list,
> no need to do it again in nfs_scan_commit.
>
> Signed-off-by: Fred Isaman <iisaman@xxxxxxxxxx>
> ---
> fs/nfs/write.c | 5 +----
> 1 files changed, 1 insertions(+), 4 deletions(-)
>
> diff --git a/fs/nfs/write.c b/fs/nfs/write.c
> index 0de19f4..e39ddfd 100644
> --- a/fs/nfs/write.c
> +++ b/fs/nfs/write.c
> @@ -595,12 +595,9 @@ nfs_scan_commit(struct inode *inode, struct list_head *dst)
> spin_lock(&inode->i_lock);
> if (nfsi->ncommit > 0) {
> const int max = INT_MAX;
> - int pnfs_ret;
>
> ret = nfs_scan_commit_list(&nfsi->commit_list, dst, max);
> - pnfs_ret = pnfs_scan_commit_lists(inode, max - ret);
> - ret += pnfs_ret;
No! If we've scanned pNFS pages to commit, then we need to report this
to nfs_commit_inode. Otherwise it will skip sending the RPC calls.
> - nfsi->ncommit -= ret;
^^^^^^^^^^^^^^^^^^^^^^
I agree with this... I forgot to remove it.
> + pnfs_scan_commit_lists(inode, max - ret);
> }
> spin_unlock(&inode->i_lock);
> return ret;
--
Trond Myklebust
Linux NFS client maintainer
NetApp
Trond.Myklebust@xxxxxxxxxx
www.netapp.com
��.n��������+%������w��{.n�����{��w��� jg��� �����ݢj����G���� ���j:+v���w�m������w�������h�����٥
