NFSv4.1: layoutcommit

The filelayout driver sends LAYOUTCOMMIT only when COMMIT goes to
the data server (as opposed to the MDS) and the data server WRITE
is not NFS_FILE_SYNC.

Only whole file layout support means that there is only one IOMODE_RW layout
segment.

Signed-off-by: Andy Adamson <[email protected]>
Signed-off-by: Alexandros Batsakis <[email protected]>
Signed-off-by: Boaz Harrosh <[email protected]>
Signed-off-by: Dean Hildebrand <[email protected]>
Signed-off-by: Fred Isaman <[email protected]>
Signed-off-by: Mingyang Guo <[email protected]>
Signed-off-by: Tao Guo <[email protected]>
Signed-off-by: Zhang Jingwang <[email protected]>
Tested-by: Boaz Harrosh <[email protected]>
Signed-off-by: Benny Halevy <[email protected]>
Signed-off-by: Fred Isaman <[email protected]>
Signed-off-by: Trond Myklebust <[email protected]>

fs/nfs/file.c

@@ -326,6 +326,9 @@ nfs_file_fsync(struct file *file, int datasync)
 		ret = xchg(&ctx->error, 0);
 	if (!ret && status < 0)
 		ret = status;
+	if (!ret && !datasync)
+		/* application has asked for meta-data sync */
+		ret = pnfs_layoutcommit_inode(inode, 1);
 	return ret;
 }
 

fs/nfs/nfs4_fs.h

@@ -262,6 +262,8 @@ extern int nfs4_proc_destroy_session(struct nfs4_session *);
 extern int nfs4_init_session(struct nfs_server *server);
 extern int nfs4_proc_get_lease_time(struct nfs_client *clp,
 		struct nfs_fsinfo *fsinfo);
+extern int nfs4_proc_layoutcommit(struct nfs4_layoutcommit_data *data,
+				  int sync);
 
 static inline bool
 is_ds_only_client(struct nfs_client *clp)

fs/nfs/nfs4filelayout.c

@@ -153,6 +153,23 @@ static int filelayout_read_done_cb(struct rpc_task *task,
 	return 0;
 }
 
+/*
+ * We reference the rpc_cred of the first WRITE that triggers the need for
+ * a LAYOUTCOMMIT, and use it to send the layoutcommit compound.
+ * rfc5661 is not clear about which credential should be used.
+ */
+static void
+filelayout_set_layoutcommit(struct nfs_write_data *wdata)
+{
+	if (FILELAYOUT_LSEG(wdata->lseg)->commit_through_mds ||
+	    wdata->res.verf->committed == NFS_FILE_SYNC)
+		return;
+
+	pnfs_set_layoutcommit(wdata);
+	dprintk("%s ionde %lu pls_end_pos %lu\n", __func__, wdata->inode->i_ino,
+		(unsigned long) wdata->lseg->pls_end_pos);
+}
+
 /*
  * Call ops for the async read/write cases
  * In the case of dense layouts, the offset needs to be reset to its
@@ -210,6 +227,7 @@ static int filelayout_write_done_cb(struct rpc_task *task,
 		return -EAGAIN;
 	}
 
+	filelayout_set_layoutcommit(data);
 	return 0;
 }
 

fs/nfs/nfs4proc.c

@@ -5616,6 +5616,100 @@ int nfs4_proc_getdeviceinfo(struct nfs_server *server, struct pnfs_device *pdev)
 }
 EXPORT_SYMBOL_GPL(nfs4_proc_getdeviceinfo);
 
+static void nfs4_layoutcommit_prepare(struct rpc_task *task, void *calldata)
+{
+	struct nfs4_layoutcommit_data *data = calldata;
+	struct nfs_server *server = NFS_SERVER(data->args.inode);
+
+	if (nfs4_setup_sequence(server, &data->args.seq_args,
+				&data->res.seq_res, 1, task))
+		return;
+	rpc_call_start(task);
+}
+
+static void
+nfs4_layoutcommit_done(struct rpc_task *task, void *calldata)
+{
+	struct nfs4_layoutcommit_data *data = calldata;
+	struct nfs_server *server = NFS_SERVER(data->args.inode);
+
+	if (!nfs4_sequence_done(task, &data->res.seq_res))
+		return;
+
+	switch (task->tk_status) { /* Just ignore these failures */
+	case NFS4ERR_DELEG_REVOKED: /* layout was recalled */
+	case NFS4ERR_BADIOMODE:     /* no IOMODE_RW layout for range */
+	case NFS4ERR_BADLAYOUT:     /* no layout */
+	case NFS4ERR_GRACE:	    /* loca_recalim always false */
+		task->tk_status = 0;
+	}
+
+	if (nfs4_async_handle_error(task, server, NULL) == -EAGAIN) {
+		nfs_restart_rpc(task, server->nfs_client);
+		return;
+	}
+
+	if (task->tk_status == 0)
+		nfs_post_op_update_inode_force_wcc(data->args.inode,
+						   data->res.fattr);
+}
+
+static void nfs4_layoutcommit_release(void *calldata)
+{
+	struct nfs4_layoutcommit_data *data = calldata;
+
+	/* Matched by references in pnfs_set_layoutcommit */
+	put_lseg(data->lseg);
+	put_rpccred(data->cred);
+	kfree(data);
+}
+
+static const struct rpc_call_ops nfs4_layoutcommit_ops = {
+	.rpc_call_prepare = nfs4_layoutcommit_prepare,
+	.rpc_call_done = nfs4_layoutcommit_done,
+	.rpc_release = nfs4_layoutcommit_release,
+};
+
+int
+nfs4_proc_layoutcommit(struct nfs4_layoutcommit_data *data, int sync)
+{
+	struct rpc_message msg = {
+		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_LAYOUTCOMMIT],
+		.rpc_argp = &data->args,
+		.rpc_resp = &data->res,
+		.rpc_cred = data->cred,
+	};
+	struct rpc_task_setup task_setup_data = {
+		.task = &data->task,
+		.rpc_client = NFS_CLIENT(data->args.inode),
+		.rpc_message = &msg,
+		.callback_ops = &nfs4_layoutcommit_ops,
+		.callback_data = data,
+		.flags = RPC_TASK_ASYNC,
+	};
+	struct rpc_task *task;
+	int status = 0;
+
+	dprintk("NFS: %4d initiating layoutcommit call. sync %d "
+		"lbw: %llu inode %lu\n",
+		data->task.tk_pid, sync,
+		data->args.lastbytewritten,
+		data->args.inode->i_ino);
+
+	task = rpc_run_task(&task_setup_data);
+	if (IS_ERR(task))
+		return PTR_ERR(task);
+	if (!sync)
+		goto out;
+	status = nfs4_wait_for_completion_rpc_task(task);
+	if (status != 0)
+		goto out;
+	status = task->tk_status;
+out:
+	dprintk("%s: status %d\n", __func__, status);
+	rpc_put_task(task);
+	return status;
+}
 #endif /* CONFIG_NFS_V4_1 */
 
 struct nfs4_state_recovery_ops nfs40_reboot_recovery_ops = {

fs/nfs/nfs4xdr.c

@@ -324,6 +324,18 @@ static int nfs4_stat_to_errno(int);
 #define decode_layoutget_maxsz	(op_decode_hdr_maxsz + 8 + \
 				decode_stateid_maxsz + \
 				XDR_QUADLEN(PNFS_LAYOUT_MAXSIZE))
+#define encode_layoutcommit_maxsz (op_encode_hdr_maxsz +          \
+				2 /* offset */ + \
+				2 /* length */ + \
+				1 /* reclaim */ + \
+				encode_stateid_maxsz + \
+				1 /* new offset (true) */ + \
+				2 /* last byte written */ + \
+				1 /* nt_timechanged (false) */ + \
+				1 /* layoutupdate4 layout type */ + \
+				1 /* NULL filelayout layoutupdate4 payload */)
+#define decode_layoutcommit_maxsz (op_decode_hdr_maxsz + 3)
+
 #else /* CONFIG_NFS_V4_1 */
 #define encode_sequence_maxsz	0
 #define decode_sequence_maxsz	0
@@ -727,6 +739,17 @@ static int nfs4_stat_to_errno(int);
 				decode_sequence_maxsz + \
 				decode_putfh_maxsz +        \
 				decode_layoutget_maxsz)
+#define NFS4_enc_layoutcommit_sz (compound_encode_hdr_maxsz + \
+				encode_sequence_maxsz +\
+				encode_putfh_maxsz + \
+				encode_layoutcommit_maxsz + \
+				encode_getattr_maxsz)
+#define NFS4_dec_layoutcommit_sz (compound_decode_hdr_maxsz + \
+				decode_sequence_maxsz + \
+				decode_putfh_maxsz + \
+				decode_layoutcommit_maxsz + \
+				decode_getattr_maxsz)
+
 
 const u32 nfs41_maxwrite_overhead = ((RPC_MAX_HEADER_WITH_AUTH +
 				      compound_encode_hdr_maxsz +
@@ -1816,6 +1839,34 @@ encode_layoutget(struct xdr_stream *xdr,
 	hdr->nops++;
 	hdr->replen += decode_layoutget_maxsz;
 }
+
+static int
+encode_layoutcommit(struct xdr_stream *xdr,
+		    const struct nfs4_layoutcommit_args *args,
+		    struct compound_hdr *hdr)
+{
+	__be32 *p;
+
+	dprintk("%s: lbw: %llu type: %d\n", __func__, args->lastbytewritten,
+		NFS_SERVER(args->inode)->pnfs_curr_ld->id);
+
+	p = reserve_space(xdr, 48 + NFS4_STATEID_SIZE);
+	*p++ = cpu_to_be32(OP_LAYOUTCOMMIT);
+	/* Only whole file layouts */
+	p = xdr_encode_hyper(p, 0); /* offset */
+	p = xdr_encode_hyper(p, NFS4_MAX_UINT64); /* length */
+	*p++ = cpu_to_be32(0); /* reclaim */
+	p = xdr_encode_opaque_fixed(p, args->stateid.data, NFS4_STATEID_SIZE);
+	*p++ = cpu_to_be32(1); /* newoffset = TRUE */
+	p = xdr_encode_hyper(p, args->lastbytewritten);
+	*p++ = cpu_to_be32(0); /* Never send time_modify_changed */
+	*p++ = cpu_to_be32(NFS_SERVER(args->inode)->pnfs_curr_ld->id);/* type */
+	*p++ = cpu_to_be32(0); /* no file layout payload */
+
+	hdr->nops++;
+	hdr->replen += decode_layoutcommit_maxsz;
+	return 0;
+}
 #endif /* CONFIG_NFS_V4_1 */
 
 /*
@@ -2607,6 +2658,26 @@ static void nfs4_xdr_enc_layoutget(struct rpc_rqst *req,
 	encode_layoutget(xdr, args, &hdr);
 	encode_nops(&hdr);
 }
+
+/*
+ *  Encode LAYOUTCOMMIT request
+ */
+static int nfs4_xdr_enc_layoutcommit(struct rpc_rqst *req,
+				     struct xdr_stream *xdr,
+				     struct nfs4_layoutcommit_args *args)
+{
+	struct compound_hdr hdr = {
+		.minorversion = nfs4_xdr_minorversion(&args->seq_args),
+	};
+
+	encode_compound_hdr(xdr, req, &hdr);
+	encode_sequence(xdr, &args->seq_args, &hdr);
+	encode_putfh(xdr, NFS_FH(args->inode), &hdr);
+	encode_layoutcommit(xdr, args, &hdr);
+	encode_getfattr(xdr, args->bitmask, &hdr);
+	encode_nops(&hdr);
+	return 0;
+}
 #endif /* CONFIG_NFS_V4_1 */
 
 static void print_overflow_msg(const char *func, const struct xdr_stream *xdr)
@@ -5007,6 +5078,35 @@ static int decode_layoutget(struct xdr_stream *xdr, struct rpc_rqst *req,
 	print_overflow_msg(__func__, xdr);
 	return -EIO;
 }
+
+static int decode_layoutcommit(struct xdr_stream *xdr,
+			       struct rpc_rqst *req,
+			       struct nfs4_layoutcommit_res *res)
+{
+	__be32 *p;
+	__u32 sizechanged;
+	int status;
+
+	status = decode_op_hdr(xdr, OP_LAYOUTCOMMIT);
+	if (status)
+		return status;
+
+	p = xdr_inline_decode(xdr, 4);
+	if (unlikely(!p))
+		goto out_overflow;
+	sizechanged = be32_to_cpup(p);
+
+	if (sizechanged) {
+		/* throw away new size */
+		p = xdr_inline_decode(xdr, 8);
+		if (unlikely(!p))
+			goto out_overflow;
+	}
+	return 0;
+out_overflow:
+	print_overflow_msg(__func__, xdr);
+	return -EIO;
+}
 #endif /* CONFIG_NFS_V4_1 */
 
 /*
@@ -6068,6 +6168,34 @@ static int nfs4_xdr_dec_layoutget(struct rpc_rqst *rqstp,
 out:
 	return status;
 }
+
+/*
+ * Decode LAYOUTCOMMIT response
+ */
+static int nfs4_xdr_dec_layoutcommit(struct rpc_rqst *rqstp,
+				     struct xdr_stream *xdr,
+				     struct nfs4_layoutcommit_res *res)
+{
+	struct compound_hdr hdr;
+	int status;
+
+	status = decode_compound_hdr(xdr, &hdr);
+	if (status)
+		goto out;
+	status = decode_sequence(xdr, &res->seq_res, rqstp);
+	if (status)
+		goto out;
+	status = decode_putfh(xdr);
+	if (status)
+		goto out;
+	status = decode_layoutcommit(xdr, rqstp, res);
+	if (status)
+		goto out;
+	decode_getfattr(xdr, res->fattr, res->server,
+			!RPC_IS_ASYNC(rqstp->rq_task));
+out:
+	return status;
+}
 #endif /* CONFIG_NFS_V4_1 */
 
 /**
@@ -6269,6 +6397,7 @@ struct rpc_procinfo	nfs4_procedures[] = {
 	PROC(RECLAIM_COMPLETE,	enc_reclaim_complete,	dec_reclaim_complete),
 	PROC(GETDEVICEINFO,	enc_getdeviceinfo,	dec_getdeviceinfo),
 	PROC(LAYOUTGET,		enc_layoutget,		dec_layoutget),
+	PROC(LAYOUTCOMMIT,	enc_layoutcommit,	dec_layoutcommit),
 #endif /* CONFIG_NFS_V4_1 */
 };