NFS: "open code" the NFS direct write rescheduler

An NFSv3/v4 client must reschedule on-the-wire writes if the writes are
UNSTABLE, and the server reboots before the client can complete a
subsequent COMMIT request.

To support direct asynchronous scatter-gather writes, the write
rescheduler in fs/nfs/direct.c must not depend on the I/O parameters
in the controlling nfs_direct_req structure.  iovecs can be somewhat
arbitrarily complex, so there could be an unbounded amount of information
to save for a rarely encountered requirement.

Refactor the direct write rescheduler so it uses information from each
nfs_write_data structure to reschedule writes, instead of caching that
information in the controlling nfs_direct_req structure.

Signed-off-by: Chuck Lever <cel@netapp.com>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>

1 files changed, 43 insertions, 8 deletions

diff --git a/fs/nfs/direct.c b/fs/nfs/direct.c
index d78c61a41ec3..7101405713e1 100644
--- a/fs/nfs/direct.c
+++ b/fs/nfs/direct.c
@@ -94,8 +94,8 @@ struct nfs_direct_req {
         struct nfs_writeverf    verf;           /* unstable write verifier */
 };
 
-static void nfs_direct_write_schedule(struct nfs_direct_req *dreq, int sync);
 static void nfs_direct_write_complete(struct nfs_direct_req *dreq, struct inode *inode);
+static const struct rpc_call_ops nfs_write_direct_ops;
 
 static inline void get_dreq(struct nfs_direct_req *dreq)
 {
@@ -435,14 +435,51 @@ static void nfs_direct_free_writedata(struct nfs_direct_req *dreq)
 #if defined(CONFIG_NFS_V3) || defined(CONFIG_NFS_V4)
 static void nfs_direct_write_reschedule(struct nfs_direct_req *dreq)
 {
-        struct list_head *pos;
+        struct inode *inode = dreq->inode;
+        struct list_head *p;
+        struct nfs_write_data *data;
 
-        list_splice_init(&dreq->rewrite_list, &dreq->list);
-        list_for_each(pos, &dreq->list)
-                get_dreq(dreq);
         dreq->count = 0;
+        get_dreq(dreq);
+
+        list_for_each(p, &dreq->rewrite_list) {
+                data = list_entry(p, struct nfs_write_data, pages);
+
+                get_dreq(dreq);
+
+                /*
+                 * Reset data->res.
+                 */
+                nfs_fattr_init(&data->fattr);
+                data->res.count = data->args.count;
+                memset(&data->verf, 0, sizeof(data->verf));
+
+                /*
+                 * Reuse data->task; data->args should not have changed
+                 * since the original request was sent.
+                 */
+                rpc_init_task(&data->task, NFS_CLIENT(inode), RPC_TASK_ASYNC,
+                                &nfs_write_direct_ops, data);
+                NFS_PROTO(inode)->write_setup(data, FLUSH_STABLE);
+
+                data->task.tk_priority = RPC_PRIORITY_NORMAL;
+                data->task.tk_cookie = (unsigned long) inode;
+
+                /*
+                 * We're called via an RPC callback, so BKL is already held.
+                 */
+                rpc_execute(&data->task);
+
+                dprintk("NFS: %5u rescheduled direct write call (req %s/%Ld, %u bytes @ offset %Lu)\n",
+                                data->task.tk_pid,
+                                inode->i_sb->s_id,
+                                (long long)NFS_FILEID(inode),
+                                data->args.count,
+                                (unsigned long long)data->args.offset);
+        }
 
-        nfs_direct_write_schedule(dreq, FLUSH_STABLE);
+        if (put_dreq(dreq))
+                nfs_direct_write_complete(dreq, inode);
 }
 
 static void nfs_direct_commit_result(struct rpc_task *task, void *calldata)
@@ -612,8 +649,6 @@ static void nfs_direct_write_result(struct rpc_task *task, void *calldata)
                                 }
                 }
         }
-        /* In case we have to resend */
-        data->args.stable = NFS_FILE_SYNC;
 
         spin_unlock(&dreq->lock);
 }