[RFC v2 00/12] NFS (just) async support for COPY

[RFC v2 00/12] NFS (just) async support for COPY

[Olga Kornievskaia]

In version 2:
This series separates support for async COPY (no inter support). 

Simple timings to demonstrate the usefulness of the async copy
implementation.
This is async "intra" SSC vs normal copy vs upstream sync SSC
time ./nfscopy /mnt/1024MB.txt 0 /mnt/1024 0 10240000000
real    1m47.036s
user    0m0.001s
sys     0m0.002s

time cp /mnt/1024MB.txt /mnt/1024
real    3m10.121s
user    0m0.060s
sys     0m9.365s

(existing "intra" client against "intra" server)
time ./nfscopy /mnt/1024MB.txt 0 /mnt/1024 0 10240000000
real    2m45.975s
user    0m0.022s
sys     0m0.452s

>From version 1 cover letter:
Client sends an asynchronous COPY to the server (always). If server 
errs with ERR_OFFLOAD_NOREQS the copy will be re-sent as a synchronous
COPY. If application cancels an in-flight COPY, OFFLOAD_CANCEL is sent
to the source server.

If server replies to the COPY with the copy stateid, client will go
wait on the CB_OFFLOAD. To fend off the race between CB_OFFLOAD and
COPY reply, we check the list of pending callbacks before going to
wait. Client adds the copy to the global list of copy stateids for the
callback to look thru and signal the waiting copy.

If application cancels async COPY after reply is received, wait will be
interrupted and client will send OFFLOAD_CANCEL to the server (sending 
it as an async RPC in the context of the nfsiod_workqueue).

When the client receives reply from the CB_OFFLOAD with some bytes and
committed how is UNSTABLE, then COMMIT is sent to the server. The results
arep propagated to the VFS and application. Assuming that application
will deal with a partial result and continue from the new offset if needed.

Handling reboot of the destination server when client is waiting on the
CB_OFFLOAD happens when SEQUENCE discovers that destination server rebooted.
The open state initially is marked to be NFS_CLNT_DST_SSC_COPY_STATE
during the COPY. Then during the recovery if state is marked as such,
then look thru the list of copies for the server and see if any are
associated with this recovering open, if so mark the copy rebooted and
wake up the waiting copy. Upon wake up the waiting copy, will restart the
copy from scratch.

If CB_OFFLOAD returned an error and non negative value of partial copy
and error is not ENOSPC, then ignore the error and send the commit
and return partial result to the client to start the next copy.

Olga Kornievskaia (12):
  NFS CB_OFFLOAD xdr
  NFS OFFLOAD_STATUS xdr
  NFS OFFLOAD_STATUS op
  NFS OFFLOAD_CANCEL xdr
  NFS COPY xdr handle async reply
  NFS add support for asynchronous COPY
  NFS handle COPY reply CB_OFFLOAD call race
  NFS send OFFLOAD_CANCEL when COPY killed
  NFS make COPY synchronous xdr configurable
  NFS handle COPY ERR_OFFLOAD_NO_REQS
  NFS if we got partial copy ignore errors
  NFS recover from destination server reboot for copies

 fs/nfs/callback.h         |  13 +++
 fs/nfs/callback_proc.c    |  52 +++++++++++
 fs/nfs/callback_xdr.c     |  80 ++++++++++++++++-
 fs/nfs/client.c           |   1 +
 fs/nfs/nfs42.h            |   3 +-
 fs/nfs/nfs42proc.c        | 222 ++++++++++++++++++++++++++++++++++++++++++++--
 fs/nfs/nfs42xdr.c         | 177 +++++++++++++++++++++++++++++++++---
 fs/nfs/nfs4_fs.h          |   4 +
 fs/nfs/nfs4client.c       |  15 ++++
 fs/nfs/nfs4file.c         |   9 +-
 fs/nfs/nfs4proc.c         |   2 +
 fs/nfs/nfs4state.c        |  15 ++++
 fs/nfs/nfs4xdr.c          |   2 +
 include/linux/nfs4.h      |   2 +
 include/linux/nfs_fs.h    |  11 +++
 include/linux/nfs_fs_sb.h |   4 +
 include/linux/nfs_xdr.h   |  14 +++
 17 files changed, 605 insertions(+), 21 deletions(-)

-- 
1.8.3.1

[RFC v2 01/12] NFS CB_OFFLOAD xdr

[Olga Kornievskaia]

Signed-off-by: Olga Kornievskaia <kolga@netapp.com>
---
 fs/nfs/callback.h      | 13 ++++++++
 fs/nfs/callback_proc.c |  7 +++++
 fs/nfs/callback_xdr.c  | 80 +++++++++++++++++++++++++++++++++++++++++++++++++-
 3 files changed, 99 insertions(+), 1 deletion(-)

diff --git a/fs/nfs/callback.h b/fs/nfs/callback.h
index c701c30..e4ab65d 100644
--- a/fs/nfs/callback.h
+++ b/fs/nfs/callback.h
@@ -189,6 +189,19 @@ extern __be32 nfs4_callback_notify_lock(struct cb_notify_lock_args *args,
 					 void *dummy,
 					 struct cb_process_state *cps);
 #endif /* CONFIG_NFS_V4_1 */
+#ifdef CONFIG_NFS_V4_2
+struct cb_offloadargs {
+	struct nfs_fh		coa_fh;
+	nfs4_stateid		coa_stateid;
+	uint32_t		error;
+	uint64_t		wr_count;
+	struct nfs_writeverf	wr_writeverf;
+};
+
+extern __be32 nfs4_callback_offload(
+	struct cb_offloadargs *args,
+	void *dummy, struct cb_process_state *cps);
+#endif /* CONFIG_NFS_V4_2 */
 extern int check_gss_callback_principal(struct nfs_client *, struct svc_rqst *);
 extern __be32 nfs4_callback_getattr(struct cb_getattrargs *args,
 				    struct cb_getattrres *res,
diff --git a/fs/nfs/callback_proc.c b/fs/nfs/callback_proc.c
index f073a6d2c..b68b803 100644
--- a/fs/nfs/callback_proc.c
+++ b/fs/nfs/callback_proc.c
@@ -680,3 +680,10 @@ __be32 nfs4_callback_notify_lock(struct cb_notify_lock_args *args, void *dummy,
 	return htonl(NFS4_OK);
 }
 #endif /* CONFIG_NFS_V4_1 */
+#ifdef CONFIG_NFS_V4_2
+__be32 nfs4_callback_offload(struct cb_offloadargs *args, void *dummy,
+				struct cb_process_state *cps)
+{
+	return 0;
+}
+#endif /* CONFIG_NFS_V4_2 */
diff --git a/fs/nfs/callback_xdr.c b/fs/nfs/callback_xdr.c
index d051fc3..97f9d78 100644
--- a/fs/nfs/callback_xdr.c
+++ b/fs/nfs/callback_xdr.c
@@ -37,6 +37,9 @@
 #define CB_OP_RECALLSLOT_RES_MAXSZ	(CB_OP_HDR_RES_MAXSZ)
 #define CB_OP_NOTIFY_LOCK_RES_MAXSZ	(CB_OP_HDR_RES_MAXSZ)
 #endif /* CONFIG_NFS_V4_1 */
+#ifdef CONFIG_NFS_V4_2
+#define CB_OP_OFFLOAD_RES_MAXSZ		(CB_OP_HDR_RES_MAXSZ)
+#endif /* CONFIG_NFS_V4_2 */
 
 #define NFSDBG_FACILITY NFSDBG_CALLBACK
 
@@ -565,7 +568,72 @@ static __be32 decode_notify_lock_args(struct svc_rqst *rqstp, struct xdr_stream
 }
 
 #endif /* CONFIG_NFS_V4_1 */
+#ifdef CONFIG_NFS_V4_2
+static __be32 decode_write_response(struct xdr_stream *xdr,
+					struct cb_offloadargs *args)
+{
+	__be32 *p;
+	__be32 dummy;
+
+	/* skip the always zero field */
+	p = read_buf(xdr, 4);
+	if (unlikely(!p))
+		goto out;
+	dummy = ntohl(*p++);
+
+	/* decode count, stable_how, verifier */
+	p = xdr_inline_decode(xdr, 8 + 4);
+	if (unlikely(!p))
+		goto out;
+	p = xdr_decode_hyper(p, &args->wr_count);
+	args->wr_writeverf.committed = be32_to_cpup(p);
+	p = xdr_inline_decode(xdr, NFS4_VERIFIER_SIZE);
+	if (likely(p)) {
+		memcpy(&args->wr_writeverf.verifier.data[0], p,
+			NFS4_VERIFIER_SIZE);
+		return 0;
+	}
+out:
+	return htonl(NFS4ERR_RESOURCE);
+}
 
+static __be32 decode_offload_args(struct svc_rqst *rqstp,
+					struct xdr_stream *xdr,
+					struct cb_offloadargs *args)
+{
+	__be32 *p;
+	__be32 status;
+
+	/* decode fh */
+	status = decode_fh(xdr, &args->coa_fh);
+	if (unlikely(status != 0))
+		return status;
+
+	/* decode stateid */
+	status = decode_stateid(xdr, &args->coa_stateid);
+	if (unlikely(status != 0))
+		return status;
+
+	/* decode status */
+	p = read_buf(xdr, 4);
+	if (unlikely(!p))
+		goto out;
+	args->error = ntohl(*p++);
+	if (!args->error) {
+		status = decode_write_response(xdr, args);
+		if (unlikely(status != 0))
+			return status;
+	} else {
+		p = xdr_inline_decode(xdr, 8);
+		if (unlikely(!p))
+			goto out;
+		p = xdr_decode_hyper(p, &args->wr_count);
+	}
+	return 0;
+out:
+	return htonl(NFS4ERR_RESOURCE);
+}
+#endif /* CONFIG_NFS_V4_2 */
 static __be32 encode_string(struct xdr_stream *xdr, unsigned int len, const char *str)
 {
 	if (unlikely(xdr_stream_encode_opaque(xdr, str, len) < 0))
@@ -833,7 +901,10 @@ static void nfs4_cb_free_slot(struct cb_process_state *cps)
 	if (status != htonl(NFS4ERR_OP_ILLEGAL))
 		return status;
 
-	if (op_nr == OP_CB_OFFLOAD)
+	if (op_nr == OP_CB_OFFLOAD) {
+		*op = &callback_ops[op_nr];
+		return htonl(NFS_OK);
+	} else
 		return htonl(NFS4ERR_NOTSUPP);
 	return htonl(NFS4ERR_OP_ILLEGAL);
 }
@@ -1038,6 +1109,13 @@ static __be32 nfs4_callback_compound(struct svc_rqst *rqstp, void *argp, void *r
 		.res_maxsize = CB_OP_NOTIFY_LOCK_RES_MAXSZ,
 	},
 #endif /* CONFIG_NFS_V4_1 */
+#ifdef CONFIG_NFS_V4_2
+	[OP_CB_OFFLOAD] = {
+		.process_op = (callback_process_op_t)nfs4_callback_offload,
+		.decode_args = (callback_decode_arg_t)decode_offload_args,
+		.res_maxsize = CB_OP_OFFLOAD_RES_MAXSZ,
+	},
+#endif /* CONFIG_NFS_V4_2 */
 };
 
 /*
-- 
1.8.3.1

[RFC v2 02/12] NFS OFFLOAD_STATUS xdr

[Olga Kornievskaia]

Signed-off-by: Olga Kornievskaia <kolga@netapp.com>
---
 fs/nfs/nfs42xdr.c         | 87 +++++++++++++++++++++++++++++++++++++++++++++++
 fs/nfs/nfs4proc.c         |  1 +
 fs/nfs/nfs4xdr.c          |  1 +
 include/linux/nfs4.h      |  1 +
 include/linux/nfs_fs_sb.h |  1 +
 include/linux/nfs_xdr.h   | 12 +++++++
 6 files changed, 103 insertions(+)

diff --git a/fs/nfs/nfs42xdr.c b/fs/nfs/nfs42xdr.c
index 6c72964..66f6193 100644
--- a/fs/nfs/nfs42xdr.c
+++ b/fs/nfs/nfs42xdr.c
@@ -25,6 +25,11 @@
 					 NFS42_WRITE_RES_SIZE + \
 					 1 /* cr_consecutive */ + \
 					 1 /* cr_synchronous */)
+#define encode_offload_status_maxsz	(op_encode_hdr_maxsz + \
+					 XDR_QUADLEN(NFS4_STATEID_SIZE))
+#define decode_offload_status_maxsz	(op_decode_hdr_maxsz + \
+					 2 /* osr_count */ + \
+					 1 + 1 /* osr_complete<1> */)
 #define encode_deallocate_maxsz		(op_encode_hdr_maxsz + \
 					 encode_fallocate_maxsz)
 #define decode_deallocate_maxsz		(op_decode_hdr_maxsz)
@@ -72,6 +77,12 @@
 					 decode_savefh_maxsz + \
 					 decode_putfh_maxsz + \
 					 decode_copy_maxsz)
+#define NFS4_enc_offload_status_sz	(compound_encode_hdr_maxsz + \
+					 encode_putfh_maxsz + \
+					 encode_offload_status_maxsz)
+#define NFS4_dec_offload_status_sz	(compound_decode_hdr_maxsz + \
+					 decode_putfh_maxsz + \
+					 decode_offload_status_maxsz)
 #define NFS4_enc_deallocate_sz		(compound_encode_hdr_maxsz + \
 					 encode_putfh_maxsz + \
 					 encode_deallocate_maxsz + \
@@ -142,6 +153,14 @@ static void encode_copy(struct xdr_stream *xdr,
 	encode_uint32(xdr, 0); /* src server list */
 }
 
+static void encode_offload_status(struct xdr_stream *xdr,
+				  struct nfs42_offload_status_args *args,
+				  struct compound_hdr *hdr)
+{
+	encode_op_hdr(xdr, OP_OFFLOAD_STATUS, decode_offload_status_maxsz, hdr);
+	encode_nfs4_stateid(xdr, &args->osa_stateid);
+}
+
 static void encode_deallocate(struct xdr_stream *xdr,
 			      struct nfs42_falloc_args *args,
 			      struct compound_hdr *hdr)
@@ -243,6 +262,24 @@ static void nfs4_xdr_enc_copy(struct rpc_rqst *req,
 }
 
 /*
+ * Encode OFFLOAD_STATUS request
+ */
+static void nfs4_xdr_enc_offload_status(struct rpc_rqst *req,
+					struct xdr_stream *xdr,
+					struct nfs42_offload_status_args *args)
+{
+	struct compound_hdr hdr = {
+		.minorversion = nfs4_xdr_minorversion(&args->osa_seq_args),
+	};
+
+	encode_compound_hdr(xdr, req, &hdr);
+	encode_sequence(xdr, &args->osa_seq_args, &hdr);
+	encode_putfh(xdr, args->osa_src_fh, &hdr);
+	encode_offload_status(xdr, args, &hdr);
+	encode_nops(&hdr);
+}
+
+/*
  * Encode DEALLOCATE request
  */
 static void nfs4_xdr_enc_deallocate(struct rpc_rqst *req,
@@ -391,6 +428,31 @@ static int decode_copy(struct xdr_stream *xdr, struct nfs42_copy_res *res)
 	return decode_copy_requirements(xdr, res);
 }
 
+static int decode_offload_status(struct xdr_stream *xdr,
+				 struct nfs42_offload_status_res *res)
+{
+	__be32 *p;
+	uint32_t count;
+	int status;
+
+	status = decode_op_hdr(xdr, OP_OFFLOAD_STATUS);
+	if (status)
+		return status;
+	p = xdr_inline_decode(xdr, 8 + 4);
+	if (unlikely(!p))
+		goto out_overflow;
+	p = xdr_decode_hyper(p, &res->osr_count);
+	count = be32_to_cpup(p++);
+	if (count) {
+		p = xdr_inline_decode(xdr, 4);
+		res->osr_status = be32_to_cpup(p);
+	}
+	return 0;
+out_overflow:
+	print_overflow_msg(__func__, xdr);
+	return -EIO;
+}
+
 static int decode_deallocate(struct xdr_stream *xdr, struct nfs42_falloc_res *res)
 {
 	return decode_op_hdr(xdr, OP_DEALLOCATE);
@@ -486,6 +548,31 @@ static int nfs4_xdr_dec_copy(struct rpc_rqst *rqstp,
 }
 
 /*
+ * Decode OFFLOAD_STATUS response
+ */
+static int nfs4_xdr_dec_offload_status(struct rpc_rqst *rqstp,
+				       struct xdr_stream *xdr,
+				       struct nfs42_offload_status_res *res)
+{
+	struct compound_hdr hdr;
+	int status;
+
+	status = decode_compound_hdr(xdr, &hdr);
+	if (status)
+		goto out;
+	status = decode_sequence(xdr, &res->osr_seq_res, rqstp);
+	if (status)
+		goto out;
+	status = decode_putfh(xdr);
+	if (status)
+		goto out;
+	status = decode_offload_status(xdr, res);
+
+out:
+	return status;
+}
+
+/*
  * Decode DEALLOCATE request
  */
 static int nfs4_xdr_dec_deallocate(struct rpc_rqst *rqstp,
diff --git a/fs/nfs/nfs4proc.c b/fs/nfs/nfs4proc.c
index 1b18368..3caed60 100644
--- a/fs/nfs/nfs4proc.c
+++ b/fs/nfs/nfs4proc.c
@@ -9245,6 +9245,7 @@ static bool nfs4_match_stateid(const nfs4_stateid *s1,
 		| NFS_CAP_ATOMIC_OPEN_V1
 		| NFS_CAP_ALLOCATE
 		| NFS_CAP_COPY
+		| NFS_CAP_OFFLOAD_STATUS
 		| NFS_CAP_DEALLOCATE
 		| NFS_CAP_SEEK
 		| NFS_CAP_LAYOUTSTATS
diff --git a/fs/nfs/nfs4xdr.c b/fs/nfs/nfs4xdr.c
index f0369e3..f7f695a 100644
--- a/fs/nfs/nfs4xdr.c
+++ b/fs/nfs/nfs4xdr.c
@@ -7573,6 +7573,7 @@ struct rpc_procinfo	nfs4_procedures[] = {
 	PROC(LAYOUTSTATS,	enc_layoutstats,	dec_layoutstats),
 	PROC(CLONE,		enc_clone,		dec_clone),
 	PROC(COPY,		enc_copy,		dec_copy),
+	PROC(OFFLOAD_STATUS,	enc_offload_status,	dec_offload_status),
 #endif /* CONFIG_NFS_V4_2 */
 };
 
diff --git a/include/linux/nfs4.h b/include/linux/nfs4.h
index 1b1ca04..d67c350 100644
--- a/include/linux/nfs4.h
+++ b/include/linux/nfs4.h
@@ -523,6 +523,7 @@ enum {
 	NFSPROC4_CLNT_LAYOUTSTATS,
 	NFSPROC4_CLNT_CLONE,
 	NFSPROC4_CLNT_COPY,
+	NFSPROC4_CLNT_OFFLOAD_STATUS,
 };
 
 /* nfs41 types */
diff --git a/include/linux/nfs_fs_sb.h b/include/linux/nfs_fs_sb.h
index b34097c..a53f9ca 100644
--- a/include/linux/nfs_fs_sb.h
+++ b/include/linux/nfs_fs_sb.h
@@ -250,5 +250,6 @@ struct nfs_server {
 #define NFS_CAP_LAYOUTSTATS	(1U << 22)
 #define NFS_CAP_CLONE		(1U << 23)
 #define NFS_CAP_COPY		(1U << 24)
+#define NFS_CAP_OFFLOAD_STATUS	(1U << 25)
 
 #endif
diff --git a/include/linux/nfs_xdr.h b/include/linux/nfs_xdr.h
index 348f7c1..aafafde 100644
--- a/include/linux/nfs_xdr.h
+++ b/include/linux/nfs_xdr.h
@@ -1385,6 +1385,18 @@ struct nfs42_copy_res {
 	bool				synchronous;
 };
 
+struct nfs42_offload_status_args {
+	struct nfs4_sequence_args	osa_seq_args;
+	struct nfs_fh			*osa_src_fh;
+	nfs4_stateid			osa_stateid;
+};
+
+struct nfs42_offload_status_res {
+	struct nfs4_sequence_res	osr_seq_res;
+	uint64_t			osr_count;
+	int				osr_status;
+};
+
 struct nfs42_seek_args {
 	struct nfs4_sequence_args	seq_args;
 
-- 
1.8.3.1

[RFC v2 03/12] NFS OFFLOAD_STATUS op

[Olga Kornievskaia]

Signed-off-by: Olga Kornievskaia <kolga@netapp.com>
---
 fs/nfs/nfs42.h     |  3 ++-
 fs/nfs/nfs42proc.c | 42 ++++++++++++++++++++++++++++++++++++++++++
 2 files changed, 44 insertions(+), 1 deletion(-)

diff --git a/fs/nfs/nfs42.h b/fs/nfs/nfs42.h
index b6cd153..90ef9b4 100644
--- a/fs/nfs/nfs42.h
+++ b/fs/nfs/nfs42.h
@@ -19,5 +19,6 @@
 int nfs42_proc_layoutstats_generic(struct nfs_server *,
 				   struct nfs42_layoutstat_data *);
 int nfs42_proc_clone(struct file *, struct file *, loff_t, loff_t, loff_t);
-
+int nfs42_proc_offload_status(struct file *, nfs4_stateid *,
+			      struct nfs42_offload_status_res *);
 #endif /* __LINUX_FS_NFS_NFS4_2_H */
diff --git a/fs/nfs/nfs42proc.c b/fs/nfs/nfs42proc.c
index 1e486c7..d4e7f60 100644
--- a/fs/nfs/nfs42proc.c
+++ b/fs/nfs/nfs42proc.c
@@ -254,6 +254,48 @@ ssize_t nfs42_proc_copy(struct file *src, loff_t pos_src,
 	return err;
 }
 
+int _nfs42_proc_offload_status(struct file *dst, nfs4_stateid *stateid,
+				struct nfs42_offload_status_res *res)
+{
+	struct nfs42_offload_status_args args;
+	struct nfs_server *dst_server = NFS_SERVER(file_inode(dst));
+	struct rpc_message msg = {
+		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_OFFLOAD_STATUS],
+		.rpc_resp = res,
+	};
+	int status;
+
+	args.osa_src_fh = NFS_FH(file_inode(dst));
+	memcpy(&args.osa_stateid, stateid, sizeof(args.osa_stateid));
+	msg.rpc_argp = &args;
+	status = nfs4_call_sync(dst_server->client, dst_server, &msg,
+				&args.osa_seq_args, &res->osr_seq_res, 0);
+	if (status == -ENOTSUPP)
+		dst_server->caps &= ~NFS_CAP_OFFLOAD_STATUS;
+
+	return status;
+}
+
+int nfs42_proc_offload_status(struct file *dst, nfs4_stateid *stateid,
+				struct nfs42_offload_status_res *res)
+{
+	struct nfs_server *dst_server = NFS_SERVER(file_inode(dst));
+	struct nfs4_exception exception = { };
+	int status;
+
+	if (!(dst_server->caps & NFS_CAP_OFFLOAD_STATUS))
+		return -EOPNOTSUPP;
+
+	do {
+		status = _nfs42_proc_offload_status(dst, stateid, res);
+		if (status == -ENOTSUPP)
+			return -EOPNOTSUPP;
+		status = nfs4_handle_exception(dst_server, status, &exception);
+	} while (exception.retry);
+
+	return status;
+}
+
 static loff_t _nfs42_proc_llseek(struct file *filep,
 		struct nfs_lock_context *lock, loff_t offset, int whence)
 {
-- 
1.8.3.1

[RFC v2 04/12] NFS OFFLOAD_CANCEL xdr

[Olga Kornievskaia]

Signed-off-by: Olga Kornievskaia <kolga@netapp.com>
---
 fs/nfs/nfs42xdr.c         | 66 +++++++++++++++++++++++++++++++++++++++++++++++
 fs/nfs/nfs4proc.c         |  1 +
 fs/nfs/nfs4xdr.c          |  1 +
 include/linux/nfs4.h      |  1 +
 include/linux/nfs_fs_sb.h |  1 +
 5 files changed, 70 insertions(+)

diff --git a/fs/nfs/nfs42xdr.c b/fs/nfs/nfs42xdr.c
index 66f6193..7634a77 100644
--- a/fs/nfs/nfs42xdr.c
+++ b/fs/nfs/nfs42xdr.c
@@ -30,6 +30,9 @@
 #define decode_offload_status_maxsz	(op_decode_hdr_maxsz + \
 					 2 /* osr_count */ + \
 					 1 + 1 /* osr_complete<1> */)
+#define encode_offload_cancel_maxsz	(op_encode_hdr_maxsz + \
+					 XDR_QUADLEN(NFS4_STATEID_SIZE))
+#define decode_offload_cancel_maxsz	(op_decode_hdr_maxsz)
 #define encode_deallocate_maxsz		(op_encode_hdr_maxsz + \
 					 encode_fallocate_maxsz)
 #define decode_deallocate_maxsz		(op_decode_hdr_maxsz)
@@ -83,6 +86,12 @@
 #define NFS4_dec_offload_status_sz	(compound_decode_hdr_maxsz + \
 					 decode_putfh_maxsz + \
 					 decode_offload_status_maxsz)
+#define NFS4_enc_offload_cancel_sz	(compound_encode_hdr_maxsz + \
+					 encode_putfh_maxsz + \
+					 encode_offload_cancel_maxsz)
+#define NFS4_dec_offload_cancel_sz	(compound_decode_hdr_maxsz + \
+					 decode_putfh_maxsz + \
+					 decode_offload_cancel_maxsz)
 #define NFS4_enc_deallocate_sz		(compound_encode_hdr_maxsz + \
 					 encode_putfh_maxsz + \
 					 encode_deallocate_maxsz + \
@@ -161,6 +170,14 @@ static void encode_offload_status(struct xdr_stream *xdr,
 	encode_nfs4_stateid(xdr, &args->osa_stateid);
 }
 
+static void encode_offload_cancel(struct xdr_stream *xdr,
+				  struct nfs42_offload_status_args *args,
+				  struct compound_hdr *hdr)
+{
+	encode_op_hdr(xdr, OP_OFFLOAD_CANCEL, decode_offload_cancel_maxsz, hdr);
+	encode_nfs4_stateid(xdr, &args->osa_stateid);
+}
+
 static void encode_deallocate(struct xdr_stream *xdr,
 			      struct nfs42_falloc_args *args,
 			      struct compound_hdr *hdr)
@@ -280,6 +297,24 @@ static void nfs4_xdr_enc_offload_status(struct rpc_rqst *req,
 }
 
 /*
+ * Encode OFFLOAD_CANEL request
+ */
+static void nfs4_xdr_enc_offload_cancel(struct rpc_rqst *req,
+					struct xdr_stream *xdr,
+					struct nfs42_offload_status_args *args)
+{
+	struct compound_hdr hdr = {
+		.minorversion = nfs4_xdr_minorversion(&args->osa_seq_args),
+	};
+
+	encode_compound_hdr(xdr, req, &hdr);
+	encode_sequence(xdr, &args->osa_seq_args, &hdr);
+	encode_putfh(xdr, args->osa_src_fh, &hdr);
+	encode_offload_cancel(xdr, args, &hdr);
+	encode_nops(&hdr);
+}
+
+/*
  * Encode DEALLOCATE request
  */
 static void nfs4_xdr_enc_deallocate(struct rpc_rqst *req,
@@ -453,6 +488,12 @@ static int decode_offload_status(struct xdr_stream *xdr,
 	return -EIO;
 }
 
+static int decode_offload_cancel(struct xdr_stream *xdr,
+				 struct nfs42_offload_status_res *res)
+{
+	return decode_op_hdr(xdr, OP_OFFLOAD_CANCEL);
+}
+
 static int decode_deallocate(struct xdr_stream *xdr, struct nfs42_falloc_res *res)
 {
 	return decode_op_hdr(xdr, OP_DEALLOCATE);
@@ -573,6 +614,31 @@ static int nfs4_xdr_dec_offload_status(struct rpc_rqst *rqstp,
 }
 
 /*
+ * Decode OFFLOAD_CANCEL response
+ */
+static int nfs4_xdr_dec_offload_cancel(struct rpc_rqst *rqstp,
+				       struct xdr_stream *xdr,
+				       struct nfs42_offload_status_res *res)
+{
+	struct compound_hdr hdr;
+	int status;
+
+	status = decode_compound_hdr(xdr, &hdr);
+	if (status)
+		goto out;
+	status = decode_sequence(xdr, &res->osr_seq_res, rqstp);
+	if (status)
+		goto out;
+	status = decode_putfh(xdr);
+	if (status)
+		goto out;
+	status = decode_offload_cancel(xdr, res);
+
+out:
+	return status;
+}
+
+/*
  * Decode DEALLOCATE request
  */
 static int nfs4_xdr_dec_deallocate(struct rpc_rqst *rqstp,
diff --git a/fs/nfs/nfs4proc.c b/fs/nfs/nfs4proc.c
index 3caed60..5450926 100644
--- a/fs/nfs/nfs4proc.c
+++ b/fs/nfs/nfs4proc.c
@@ -9246,6 +9246,7 @@ static bool nfs4_match_stateid(const nfs4_stateid *s1,
 		| NFS_CAP_ALLOCATE
 		| NFS_CAP_COPY
 		| NFS_CAP_OFFLOAD_STATUS
+		| NFS_CAP_OFFLOAD_CANCEL
 		| NFS_CAP_DEALLOCATE
 		| NFS_CAP_SEEK
 		| NFS_CAP_LAYOUTSTATS
diff --git a/fs/nfs/nfs4xdr.c b/fs/nfs/nfs4xdr.c
index f7f695a..cf07efc 100644
--- a/fs/nfs/nfs4xdr.c
+++ b/fs/nfs/nfs4xdr.c
@@ -7574,6 +7574,7 @@ struct rpc_procinfo	nfs4_procedures[] = {
 	PROC(CLONE,		enc_clone,		dec_clone),
 	PROC(COPY,		enc_copy,		dec_copy),
 	PROC(OFFLOAD_STATUS,	enc_offload_status,	dec_offload_status),
+	PROC(OFFLOAD_CANCEL,    enc_offload_cancel,     dec_offload_cancel),
 #endif /* CONFIG_NFS_V4_2 */
 };
 
diff --git a/include/linux/nfs4.h b/include/linux/nfs4.h
index d67c350..7262908 100644
--- a/include/linux/nfs4.h
+++ b/include/linux/nfs4.h
@@ -524,6 +524,7 @@ enum {
 	NFSPROC4_CLNT_CLONE,
 	NFSPROC4_CLNT_COPY,
 	NFSPROC4_CLNT_OFFLOAD_STATUS,
+	NFSPROC4_CLNT_OFFLOAD_CANCEL,
 };
 
 /* nfs41 types */
diff --git a/include/linux/nfs_fs_sb.h b/include/linux/nfs_fs_sb.h
index a53f9ca..298ce5a 100644
--- a/include/linux/nfs_fs_sb.h
+++ b/include/linux/nfs_fs_sb.h
@@ -251,5 +251,6 @@ struct nfs_server {
 #define NFS_CAP_CLONE		(1U << 23)
 #define NFS_CAP_COPY		(1U << 24)
 #define NFS_CAP_OFFLOAD_STATUS	(1U << 25)
+#define NFS_CAP_OFFLOAD_CANCEL	(1U << 26)
 
 #endif
-- 
1.8.3.1

[RFC v2 05/12] NFS COPY xdr handle async reply

[Olga Kornievskaia]

If server returns async reply, it must include a callback stateid,
wr_callback_id in the write_response4.

Signed-off-by: Olga Kornievskaia <kolga@netapp.com>
---
 fs/nfs/nfs42xdr.c       | 22 ++++++++++++----------
 include/linux/nfs_xdr.h |  1 +
 2 files changed, 13 insertions(+), 10 deletions(-)

diff --git a/fs/nfs/nfs42xdr.c b/fs/nfs/nfs42xdr.c
index 7634a77..83dce11 100644
--- a/fs/nfs/nfs42xdr.c
+++ b/fs/nfs/nfs42xdr.c
@@ -403,21 +403,23 @@ static int decode_write_response(struct xdr_stream *xdr,
 				 struct nfs42_write_res *res)
 {
 	__be32 *p;
+	int status, count;
 
-	p = xdr_inline_decode(xdr, 4 + 8 + 4);
+	p = xdr_inline_decode(xdr, 4);
 	if (unlikely(!p))
 		goto out_overflow;
-
-	/*
-	 * We never use asynchronous mode, so warn if a server returns
-	 * a stateid.
-	 */
-	if (unlikely(*p != 0)) {
-		pr_err_once("%s: server has set unrequested "
-				"asynchronous mode\n", __func__);
+	count = be32_to_cpup(p);
+	if (count > 1)
 		return -EREMOTEIO;
+	else if (count == 1) {
+		status = decode_opaque_fixed(xdr, &res->stateid,
+				NFS4_STATEID_SIZE);
+		if (unlikely(status))
+			goto out_overflow;
 	}
-	p++;
+	p = xdr_inline_decode(xdr, 8 + 4);
+	if (unlikely(!p))
+		goto out_overflow;
 	p = xdr_decode_hyper(p, &res->count);
 	res->verifier.committed = be32_to_cpup(p);
 	return decode_verifier(xdr, &res->verifier.verifier);
diff --git a/include/linux/nfs_xdr.h b/include/linux/nfs_xdr.h
index aafafde..6c6130f 100644
--- a/include/linux/nfs_xdr.h
+++ b/include/linux/nfs_xdr.h
@@ -1374,6 +1374,7 @@ struct nfs42_copy_args {
 };
 
 struct nfs42_write_res {
+	nfs4_stateid		stateid;
 	u64			count;
 	struct nfs_writeverf	verifier;
 };
-- 
1.8.3.1

[RFC v2 06/12] NFS add support for asynchronous COPY

[Olga Kornievskaia]

Change xdr to always send COPY asynchronously.

Keep the list copies send in a list under a server structure.
Once copy is sent, it waits on a completion structure that will
be signalled by the callback thread that receives CB_OFFLOAD.

Signed-off-by: Olga Kornievskaia <kolga@netapp.com>
---
 fs/nfs/callback_proc.c    | 34 ++++++++++++++++++++++++++++++++++
 fs/nfs/client.c           |  1 +
 fs/nfs/nfs42proc.c        | 46 ++++++++++++++++++++++++++++++++++++++++++----
 fs/nfs/nfs42xdr.c         |  2 +-
 include/linux/nfs_fs.h    |  9 +++++++++
 include/linux/nfs_fs_sb.h |  1 +
 6 files changed, 88 insertions(+), 5 deletions(-)

diff --git a/fs/nfs/callback_proc.c b/fs/nfs/callback_proc.c
index b68b803..c1b081c 100644
--- a/fs/nfs/callback_proc.c
+++ b/fs/nfs/callback_proc.c
@@ -681,9 +681,43 @@ __be32 nfs4_callback_notify_lock(struct cb_notify_lock_args *args, void *dummy,
 }
 #endif /* CONFIG_NFS_V4_1 */
 #ifdef CONFIG_NFS_V4_2
+static void nfs4_copy_cb_args(struct nfs4_copy_state *cp_state,
+				struct cb_offloadargs *args)
+{
+	cp_state->count = args->wr_count;
+	cp_state->error = args->error;
+	if (!args->error) {
+		cp_state->verf.committed = args->wr_writeverf.committed;
+		memcpy(&cp_state->verf.verifier.data[0],
+			&args->wr_writeverf.verifier.data[0],
+			NFS4_VERIFIER_SIZE);
+	}
+}
+
 __be32 nfs4_callback_offload(struct cb_offloadargs *args, void *dummy,
 				struct cb_process_state *cps)
 {
+	struct nfs_server *server;
+	struct nfs4_copy_state *copy;
+
+	rcu_read_lock();
+	list_for_each_entry_rcu(server, &cps->clp->cl_superblocks,
+				client_link) {
+		spin_lock(&server->nfs_client->cl_lock);
+		list_for_each_entry(copy, &server->ss_copies, copies) {
+			if (memcmp(args->coa_stateid.other,
+					copy->stateid.other,
+					sizeof(args->coa_stateid.other)))
+				continue;
+			nfs4_copy_cb_args(copy, args);
+			complete(&copy->completion);
+			spin_unlock(&server->nfs_client->cl_lock);
+			goto out;
+		}
+		spin_unlock(&server->nfs_client->cl_lock);
+	}
+out:
+	rcu_read_unlock();
 	return 0;
 }
 #endif /* CONFIG_NFS_V4_2 */
diff --git a/fs/nfs/client.c b/fs/nfs/client.c
index 91a8d61..a72cd1d 100644
--- a/fs/nfs/client.c
+++ b/fs/nfs/client.c
@@ -885,6 +885,7 @@ struct nfs_server *nfs_alloc_server(void)
 	INIT_LIST_HEAD(&server->delegations);
 	INIT_LIST_HEAD(&server->layouts);
 	INIT_LIST_HEAD(&server->state_owners_lru);
+	INIT_LIST_HEAD(&server->ss_copies);
 
 	atomic_set(&server->active, 0);
 
diff --git a/fs/nfs/nfs42proc.c b/fs/nfs/nfs42proc.c
index d4e7f60..27a6eb8 100644
--- a/fs/nfs/nfs42proc.c
+++ b/fs/nfs/nfs42proc.c
@@ -129,6 +129,38 @@ int nfs42_proc_deallocate(struct file *filep, loff_t offset, loff_t len)
 	return err;
 }
 
+static int handle_async_copy(struct nfs42_copy_res *res,
+			     struct nfs_server *server,
+			     struct file *src,
+			     struct file *dst,
+			     nfs4_stateid *src_stateid,
+			     uint64_t *ret_count)
+{
+	struct nfs4_copy_state *copy;
+	int status;
+
+	copy = kzalloc(sizeof(struct nfs4_copy_state), GFP_NOFS);
+	if (!copy)
+		return -ENOMEM;
+	memcpy(&copy->stateid, &res->write_res.stateid, NFS4_STATEID_SIZE);
+	init_completion(&copy->completion);
+
+	spin_lock(&server->nfs_client->cl_lock);
+	list_add_tail(&copy->copies, &server->ss_copies);
+	spin_unlock(&server->nfs_client->cl_lock);
+
+	wait_for_completion_interruptible(&copy->completion);
+	spin_lock(&server->nfs_client->cl_lock);
+	list_del_init(&copy->copies);
+	spin_unlock(&server->nfs_client->cl_lock);
+	*ret_count = copy->count;
+	status = -copy->error;
+	if (copy->count && copy->verf.committed != NFS_FILE_SYNC)
+		status = nfs_commit_file(dst, &copy->verf.verifier);
+	kfree(copy);
+	return status;
+}
+
 static ssize_t _nfs42_proc_copy(struct file *src,
 				struct nfs_lock_context *src_lock,
 				struct file *dst,
@@ -147,6 +179,7 @@ static ssize_t _nfs42_proc_copy(struct file *src,
 	loff_t pos_dst = args->dst_pos;
 	size_t count = args->count;
 	int status;
+	uint64_t ret_count;
 
 	status = nfs4_set_rw_stateid(&args->src_stateid, src_lock->open_context,
 				     src_lock, FMODE_READ);
@@ -174,16 +207,21 @@ static ssize_t _nfs42_proc_copy(struct file *src,
 	if (status)
 		return status;
 
-	if (res->write_res.verifier.committed != NFS_FILE_SYNC) {
+	ret_count = res->write_res.count;
+	if (!res->synchronous) {
+		status = handle_async_copy(res, server, src, dst,
+				&args->src_stateid, &ret_count);
+		if (status)
+			return status;
+	} else if (res->write_res.verifier.committed != NFS_FILE_SYNC) {
 		status = nfs_commit_file(dst, &res->write_res.verifier.verifier);
 		if (status)
 			return status;
 	}
 
-	truncate_pagecache_range(dst_inode, pos_dst,
-				 pos_dst + res->write_res.count);
+	truncate_pagecache_range(dst_inode, pos_dst, pos_dst + ret_count);
 
-	return res->write_res.count;
+	return ret_count;
 }
 
 ssize_t nfs42_proc_copy(struct file *src, loff_t pos_src,
diff --git a/fs/nfs/nfs42xdr.c b/fs/nfs/nfs42xdr.c
index 83dce11..059d074 100644
--- a/fs/nfs/nfs42xdr.c
+++ b/fs/nfs/nfs42xdr.c
@@ -158,7 +158,7 @@ static void encode_copy(struct xdr_stream *xdr,
 	encode_uint64(xdr, args->count);
 
 	encode_uint32(xdr, 1); /* consecutive = true */
-	encode_uint32(xdr, 1); /* synchronous = true */
+	encode_uint32(xdr, 0); /* synchronous = false */
 	encode_uint32(xdr, 0); /* src server list */
 }
 
diff --git a/include/linux/nfs_fs.h b/include/linux/nfs_fs.h
index 287f341..a862b7f 100644
--- a/include/linux/nfs_fs.h
+++ b/include/linux/nfs_fs.h
@@ -182,6 +182,15 @@ struct nfs_inode {
 	struct inode		vfs_inode;
 };
 
+struct nfs4_copy_state {
+	struct list_head	copies;
+	nfs4_stateid		stateid;
+	struct completion	completion;
+	uint64_t		count;
+	struct nfs_writeverf	verf;
+	int			error;
+};
+
 /*
  * Cache validity bit flags
  */
diff --git a/include/linux/nfs_fs_sb.h b/include/linux/nfs_fs_sb.h
index 298ce5a..9b252bb 100644
--- a/include/linux/nfs_fs_sb.h
+++ b/include/linux/nfs_fs_sb.h
@@ -206,6 +206,7 @@ struct nfs_server {
 	struct list_head	state_owners_lru;
 	struct list_head	layouts;
 	struct list_head	delegations;
+	struct list_head	ss_copies;
 
 	unsigned long		mig_gen;
 	unsigned long		mig_status;
-- 
1.8.3.1

[RFC v2 07/12] NFS handle COPY reply CB_OFFLOAD call race

[Olga Kornievskaia]

It's possible that server replies back with CB_OFFLOAD call and
COPY reply at the same time such that client will process
CB_OFFLOAD before reply to COPY. For that keep a list of pending
callback stateids received and them before waiting on completion
check the pending list.

Cleanup any pending copies on the client shutdown.

Signed-off-by: Olga Kornievskaia <kolga@netapp.com>
---
 fs/nfs/callback_proc.c    | 17 ++++++++++++++---
 fs/nfs/nfs42proc.c        | 24 +++++++++++++++++++++---
 fs/nfs/nfs4client.c       | 15 +++++++++++++++
 include/linux/nfs_fs_sb.h |  1 +
 4 files changed, 51 insertions(+), 6 deletions(-)

diff --git a/fs/nfs/callback_proc.c b/fs/nfs/callback_proc.c
index c1b081c..01bcb4b 100644
--- a/fs/nfs/callback_proc.c
+++ b/fs/nfs/callback_proc.c
@@ -699,11 +699,12 @@ __be32 nfs4_callback_offload(struct cb_offloadargs *args, void *dummy,
 {
 	struct nfs_server *server;
 	struct nfs4_copy_state *copy;
+	bool found = false;
 
+	spin_lock(&cps->clp->cl_lock);
 	rcu_read_lock();
 	list_for_each_entry_rcu(server, &cps->clp->cl_superblocks,
 				client_link) {
-		spin_lock(&server->nfs_client->cl_lock);
 		list_for_each_entry(copy, &server->ss_copies, copies) {
 			if (memcmp(args->coa_stateid.other,
 					copy->stateid.other,
@@ -711,13 +712,23 @@ __be32 nfs4_callback_offload(struct cb_offloadargs *args, void *dummy,
 				continue;
 			nfs4_copy_cb_args(copy, args);
 			complete(&copy->completion);
-			spin_unlock(&server->nfs_client->cl_lock);
+			found = true;
 			goto out;
 		}
-		spin_unlock(&server->nfs_client->cl_lock);
 	}
 out:
 	rcu_read_unlock();
+	if (!found) {
+		copy = kzalloc(sizeof(struct nfs4_copy_state), GFP_NOFS);
+		if (!copy) {
+			spin_unlock(&cps->clp->cl_lock);
+			return -ENOMEM;
+		}
+		memcpy(&copy->stateid, &args->coa_stateid, NFS4_STATEID_SIZE);
+		nfs4_copy_cb_args(copy, args);
+		list_add_tail(&copy->copies, &cps->clp->pending_cb_stateids);
+	}
+	spin_unlock(&cps->clp->cl_lock);
 	return 0;
 }
 #endif /* CONFIG_NFS_V4_2 */
diff --git a/fs/nfs/nfs42proc.c b/fs/nfs/nfs42proc.c
index 27a6eb8..8ee9fb8 100644
--- a/fs/nfs/nfs42proc.c
+++ b/fs/nfs/nfs42proc.c
@@ -137,15 +137,32 @@ static int handle_async_copy(struct nfs42_copy_res *res,
 			     uint64_t *ret_count)
 {
 	struct nfs4_copy_state *copy;
-	int status;
+	int status = NFS4_OK;
+	bool found_pending = false;
+
+	spin_lock(&server->nfs_client->cl_lock);
+	list_for_each_entry(copy, &server->nfs_client->pending_cb_stateids,
+				copies) {
+		if (memcmp(&res->write_res.stateid, &copy->stateid,
+				NFS4_STATEID_SIZE))
+			continue;
+		found_pending = true;
+		list_del(&copy->copies);
+		break;
+	}
+	if (found_pending) {
+		spin_unlock(&server->nfs_client->cl_lock);
+		goto out;
+	}
 
 	copy = kzalloc(sizeof(struct nfs4_copy_state), GFP_NOFS);
-	if (!copy)
+	if (!copy) {
+		spin_unlock(&server->nfs_client->cl_lock);
 		return -ENOMEM;
+	}
 	memcpy(&copy->stateid, &res->write_res.stateid, NFS4_STATEID_SIZE);
 	init_completion(&copy->completion);
 
-	spin_lock(&server->nfs_client->cl_lock);
 	list_add_tail(&copy->copies, &server->ss_copies);
 	spin_unlock(&server->nfs_client->cl_lock);
 
@@ -153,6 +170,7 @@ static int handle_async_copy(struct nfs42_copy_res *res,
 	spin_lock(&server->nfs_client->cl_lock);
 	list_del_init(&copy->copies);
 	spin_unlock(&server->nfs_client->cl_lock);
+out:
 	*ret_count = copy->count;
 	status = -copy->error;
 	if (copy->count && copy->verf.committed != NFS_FILE_SYNC)
diff --git a/fs/nfs/nfs4client.c b/fs/nfs/nfs4client.c
index 5ae9d64..bebb97e1 100644
--- a/fs/nfs/nfs4client.c
+++ b/fs/nfs/nfs4client.c
@@ -156,9 +156,23 @@ struct rpc_clnt *
 	}
 }
 
+static void
+nfs4_cleanup_callback(struct nfs_client *clp)
+{
+	struct nfs4_copy_state *cp_state;
+
+	while (!list_empty(&clp->pending_cb_stateids)) {
+		cp_state = list_entry(clp->pending_cb_stateids.next,
+					struct nfs4_copy_state, copies);
+		list_del(&cp_state->copies);
+		kfree(cp_state);
+	}
+}
+
 void nfs41_shutdown_client(struct nfs_client *clp)
 {
 	if (nfs4_has_session(clp)) {
+		nfs4_cleanup_callback(clp);
 		nfs4_shutdown_ds_clients(clp);
 		nfs4_destroy_session(clp->cl_session);
 		nfs4_destroy_clientid(clp);
@@ -202,6 +216,7 @@ struct nfs_client *nfs4_alloc_client(const struct nfs_client_initdata *cl_init)
 #if IS_ENABLED(CONFIG_NFS_V4_1)
 	init_waitqueue_head(&clp->cl_lock_waitq);
 #endif
+	INIT_LIST_HEAD(&clp->pending_cb_stateids);
 	return clp;
 
 error:
diff --git a/include/linux/nfs_fs_sb.h b/include/linux/nfs_fs_sb.h
index 9b252bb..04f6cb4 100644
--- a/include/linux/nfs_fs_sb.h
+++ b/include/linux/nfs_fs_sb.h
@@ -118,6 +118,7 @@ struct nfs_client {
 #endif
 
 	struct net		*cl_net;
+	struct list_head	pending_cb_stateids;
 };
 
 /*
-- 
1.8.3.1

[RFC v2 08/12] NFS send OFFLOAD_CANCEL when COPY killed

[Olga Kornievskaia]

When COPY is killed send OFFLOAD_CANCEL to both destination and
source servers.

Signed-off-by: Olga Kornievskaia <kolga@netapp.com>
---
 fs/nfs/nfs42proc.c | 93 ++++++++++++++++++++++++++++++++++++++++++++++++++++--
 1 file changed, 91 insertions(+), 2 deletions(-)

diff --git a/fs/nfs/nfs42proc.c b/fs/nfs/nfs42proc.c
index 8ee9fb8..2ffd96d 100644
--- a/fs/nfs/nfs42proc.c
+++ b/fs/nfs/nfs42proc.c
@@ -16,6 +16,7 @@
 #include "internal.h"
 
 #define NFSDBG_FACILITY NFSDBG_PROC
+static int nfs42_do_offload_cancel_async(struct file *dst, nfs4_stateid *std);
 
 static int _nfs42_proc_fallocate(struct rpc_message *msg, struct file *filep,
 		struct nfs_lock_context *lock, loff_t offset, loff_t len)
@@ -166,10 +167,15 @@ static int handle_async_copy(struct nfs42_copy_res *res,
 	list_add_tail(&copy->copies, &server->ss_copies);
 	spin_unlock(&server->nfs_client->cl_lock);
 
-	wait_for_completion_interruptible(&copy->completion);
+	status = wait_for_completion_interruptible(&copy->completion);
 	spin_lock(&server->nfs_client->cl_lock);
 	list_del_init(&copy->copies);
 	spin_unlock(&server->nfs_client->cl_lock);
+	if (status == -ERESTARTSYS) {
+		nfs42_do_offload_cancel_async(dst, &copy->stateid);
+		kfree(copy);
+		return status;
+	}
 out:
 	*ret_count = copy->count;
 	status = -copy->error;
@@ -220,8 +226,13 @@ static ssize_t _nfs42_proc_copy(struct file *src,
 
 	status = nfs4_call_sync(server->client, server, &msg,
 				&args->seq_args, &res->seq_res, 0);
-	if (status == -ENOTSUPP)
+	switch (status) {
+	case -ENOTSUPP:
 		server->caps &= ~NFS_CAP_COPY;
+		break;
+	case -ERESTARTSYS:
+		nfs42_do_offload_cancel_async(src, &args->src_stateid);
+	}
 	if (status)
 		return status;
 
@@ -310,6 +321,84 @@ ssize_t nfs42_proc_copy(struct file *src, loff_t pos_src,
 	return err;
 }
 
+struct nfs42_offloadcancel_data {
+	struct nfs_server *seq_server;
+	struct nfs42_offload_status_args args;
+	struct nfs42_offload_status_res res;
+};
+
+static void nfs42_offload_cancel_prepare(struct rpc_task *task, void *calldata)
+{
+	struct nfs42_offloadcancel_data *data = calldata;
+
+	nfs4_setup_sequence(data->seq_server->nfs_client,
+				&data->args.osa_seq_args,
+				&data->res.osr_seq_res, task);
+}
+
+static void nfs42_offload_cancel_done(struct rpc_task *task, void *calldata)
+{
+	struct nfs42_offloadcancel_data *data = calldata;
+
+	nfs41_sequence_done(task, &data->res.osr_seq_res);
+	kfree(data);
+}
+
+static void nfs42_free_offloadcancel_data(void *data)
+{
+	kfree(data);
+}
+
+static const struct rpc_call_ops nfs42_offload_cancel_ops = {
+	.rpc_call_prepare = nfs42_offload_cancel_prepare,
+	.rpc_call_done = nfs42_offload_cancel_done,
+	.rpc_release = nfs42_free_offloadcancel_data,
+};
+
+static int nfs42_do_offload_cancel_async(struct file *dst,
+				nfs4_stateid *stateid)
+{
+	struct nfs_server *dst_server = NFS_SERVER(file_inode(dst));
+	struct nfs42_offloadcancel_data *data = NULL;
+	struct nfs_open_context *ctx = nfs_file_open_context(dst);
+	struct rpc_task *task;
+	struct rpc_message msg = {
+		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_OFFLOAD_CANCEL],
+		.rpc_cred = ctx->cred,
+	};
+	struct rpc_task_setup task_setup_data = {
+		.rpc_client = dst_server->client,
+		.rpc_message = &msg,
+		.callback_ops = &nfs42_offload_cancel_ops,
+		.workqueue = nfsiod_workqueue,
+		.flags = RPC_TASK_ASYNC,
+	};
+	int status;
+
+	if (!(dst_server->caps & NFS_CAP_OFFLOAD_CANCEL))
+		return -EOPNOTSUPP;
+
+	data = kzalloc(sizeof(struct nfs42_offloadcancel_data), GFP_NOFS);
+	if (data == NULL)
+		return -ENOMEM;
+	data->seq_server = dst_server;
+	data->args.osa_src_fh = NFS_FH(file_inode(dst));
+	memcpy(&data->args.osa_stateid, stateid,
+		sizeof(data->args.osa_stateid));
+	msg.rpc_argp = &data->args;
+	msg.rpc_resp = &data->res;
+	task_setup_data.callback_data = data;
+	nfs4_init_sequence(&data->args.osa_seq_args, &data->res.osr_seq_res, 1);
+	task = rpc_run_task(&task_setup_data);
+	if (IS_ERR(task))
+		return PTR_ERR(task);
+	status = rpc_wait_for_completion_task(task);
+	if (status == -ENOTSUPP)
+		dst_server->caps &= ~NFS_CAP_OFFLOAD_CANCEL;
+	rpc_put_task(task);
+	return status;
+}
+
 int _nfs42_proc_offload_status(struct file *dst, nfs4_stateid *stateid,
 				struct nfs42_offload_status_res *res)
 {
-- 
1.8.3.1

[RFC v2 09/12] NFS make COPY synchronous xdr configurable

[Olga Kornievskaia]

Instead of hardcoding always sending async copy, make it
configurable. In case, we get ERR_OFFLOAD_NO_REQS and we need to
resend the operation as synchronous.

Signed-off-by: Olga Kornievskaia <kolga@netapp.com>
---
 fs/nfs/nfs42proc.c      | 2 ++
 fs/nfs/nfs42xdr.c       | 2 +-
 include/linux/nfs_xdr.h | 1 +
 3 files changed, 4 insertions(+), 1 deletion(-)

diff --git a/fs/nfs/nfs42proc.c b/fs/nfs/nfs42proc.c
index 2ffd96d..32b08ea 100644
--- a/fs/nfs/nfs42proc.c
+++ b/fs/nfs/nfs42proc.c
@@ -204,6 +204,7 @@ static ssize_t _nfs42_proc_copy(struct file *src,
 	size_t count = args->count;
 	int status;
 	uint64_t ret_count;
+	bool sync = false;
 
 	status = nfs4_set_rw_stateid(&args->src_stateid, src_lock->open_context,
 				     src_lock, FMODE_READ);
@@ -224,6 +225,7 @@ static ssize_t _nfs42_proc_copy(struct file *src,
 	if (status)
 		return status;
 
+	args->sync = sync;
 	status = nfs4_call_sync(server->client, server, &msg,
 				&args->seq_args, &res->seq_res, 0);
 	switch (status) {
diff --git a/fs/nfs/nfs42xdr.c b/fs/nfs/nfs42xdr.c
index 059d074..1f12574 100644
--- a/fs/nfs/nfs42xdr.c
+++ b/fs/nfs/nfs42xdr.c
@@ -158,7 +158,7 @@ static void encode_copy(struct xdr_stream *xdr,
 	encode_uint64(xdr, args->count);
 
 	encode_uint32(xdr, 1); /* consecutive = true */
-	encode_uint32(xdr, 0); /* synchronous = false */
+	encode_uint32(xdr, args->sync);
 	encode_uint32(xdr, 0); /* src server list */
 }
 
diff --git a/include/linux/nfs_xdr.h b/include/linux/nfs_xdr.h
index 6c6130f..a04862c 100644
--- a/include/linux/nfs_xdr.h
+++ b/include/linux/nfs_xdr.h
@@ -1371,6 +1371,7 @@ struct nfs42_copy_args {
 	u64				dst_pos;
 
 	u64				count;
+	bool				sync;
 };
 
 struct nfs42_write_res {
-- 
1.8.3.1

[RFC v2 10/12] NFS handle COPY ERR_OFFLOAD_NO_REQS

[Olga Kornievskaia]

If client sent async COPY and server replied with
ERR_OFFLOAD_NO_REQS, client should retry with a synchronous copy.

Signed-off-by: Olga Kornievskaia <kolga@netapp.com>
---
 fs/nfs/nfs42proc.c | 7 +++++++
 1 file changed, 7 insertions(+)

diff --git a/fs/nfs/nfs42proc.c b/fs/nfs/nfs42proc.c
index 32b08ea..850fa10 100644
--- a/fs/nfs/nfs42proc.c
+++ b/fs/nfs/nfs42proc.c
@@ -225,10 +225,17 @@ static ssize_t _nfs42_proc_copy(struct file *src,
 	if (status)
 		return status;
 
+retry:
 	args->sync = sync;
 	status = nfs4_call_sync(server->client, server, &msg,
 				&args->seq_args, &res->seq_res, 0);
 	switch (status) {
+	case -NFS4ERR_OFFLOAD_NO_REQS:
+		if (!sync) {
+			sync = true;
+			goto retry;
+		}
+		break;
 	case -ENOTSUPP:
 		server->caps &= ~NFS_CAP_COPY;
 		break;
-- 
1.8.3.1

[RFC v2 11/12] NFS if we got partial copy ignore errors

[Olga Kornievskaia]

Don't ignore ENOSPC. Otherwise, try next copy chunk

Signed-off-by: Olga Kornievskaia <kolga@netapp.com>
---
 fs/nfs/nfs42proc.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/fs/nfs/nfs42proc.c b/fs/nfs/nfs42proc.c
index 850fa10..acd884d 100644
--- a/fs/nfs/nfs42proc.c
+++ b/fs/nfs/nfs42proc.c
@@ -178,7 +178,8 @@ static int handle_async_copy(struct nfs42_copy_res *res,
 	}
 out:
 	*ret_count = copy->count;
-	status = -copy->error;
+	if (copy->count < 0 || copy->error == ENOSPC)
+		status = -copy->error;
 	if (copy->count && copy->verf.committed != NFS_FILE_SYNC)
 		status = nfs_commit_file(dst, &copy->verf.verifier);
 	kfree(copy);
-- 
1.8.3.1

[RFC v2 12/12] NFS recover from destination server reboot for copies

[Olga Kornievskaia]

Mark the destination state to indicate a server-side copy is
happening. On detecting a reboot and recovering open state check
if any state is engaged in a server-side copy, if so, find the
copy and mark it and then signal the waiting thread. Upon wakeup,
if copy was marked then propage EAGAIN to the nfsd_copy_file_range
and restart the copy from scratch (no partial state is being
queried at this point).

Signed-off-by: Olga Kornievskaia <kolga@netapp.com>
---
 fs/nfs/nfs42proc.c     | 20 ++++++++++++++++----
 fs/nfs/nfs4_fs.h       |  4 ++++
 fs/nfs/nfs4file.c      |  9 +++++++--
 fs/nfs/nfs4state.c     | 15 +++++++++++++++
 include/linux/nfs_fs.h |  2 ++
 5 files changed, 44 insertions(+), 6 deletions(-)

diff --git a/fs/nfs/nfs42proc.c b/fs/nfs/nfs42proc.c
index acd884d..bab21ba5 100644
--- a/fs/nfs/nfs42proc.c
+++ b/fs/nfs/nfs42proc.c
@@ -140,6 +140,7 @@ static int handle_async_copy(struct nfs42_copy_res *res,
 	struct nfs4_copy_state *copy;
 	int status = NFS4_OK;
 	bool found_pending = false;
+	struct nfs_open_context *ctx = nfs_file_open_context(dst);
 
 	spin_lock(&server->nfs_client->cl_lock);
 	list_for_each_entry(copy, &server->nfs_client->pending_cb_stateids,
@@ -163,6 +164,7 @@ static int handle_async_copy(struct nfs42_copy_res *res,
 	}
 	memcpy(&copy->stateid, &res->write_res.stateid, NFS4_STATEID_SIZE);
 	init_completion(&copy->completion);
+	copy->parent_state = ctx->state;
 
 	list_add_tail(&copy->copies, &server->ss_copies);
 	spin_unlock(&server->nfs_client->cl_lock);
@@ -172,9 +174,10 @@ static int handle_async_copy(struct nfs42_copy_res *res,
 	list_del_init(&copy->copies);
 	spin_unlock(&server->nfs_client->cl_lock);
 	if (status == -ERESTARTSYS) {
-		nfs42_do_offload_cancel_async(dst, &copy->stateid);
-		kfree(copy);
-		return status;
+		goto out_cancel;
+	} else if (copy->flags) {
+		status = -EAGAIN;
+		goto out_cancel;
 	}
 out:
 	*ret_count = copy->count;
@@ -184,6 +187,10 @@ static int handle_async_copy(struct nfs42_copy_res *res,
 		status = nfs_commit_file(dst, &copy->verf.verifier);
 	kfree(copy);
 	return status;
+out_cancel:
+	nfs42_do_offload_cancel_async(dst, &copy->stateid);
+	kfree(copy);
+	return status;
 }
 
 static ssize_t _nfs42_proc_copy(struct file *src,
@@ -226,6 +233,8 @@ static ssize_t _nfs42_proc_copy(struct file *src,
 	if (status)
 		return status;
 
+	set_bit(NFS_CLNT_DST_SSC_COPY_STATE,
+		&dst_lock->open_context->state->flags);
 retry:
 	args->sync = sync;
 	status = nfs4_call_sync(server->client, server, &msg,
@@ -314,9 +323,12 @@ ssize_t nfs42_proc_copy(struct file *src, loff_t pos_src,
 
 		if (err >= 0)
 			break;
-		if (err == -ENOTSUPP) {
+		switch (err) {
+		case -ENOTSUPP:
 			err = -EOPNOTSUPP;
 			break;
+		case -EAGAIN:
+			break;
 		}
 
 		err2 = nfs4_handle_exception(server, err, &src_exception);
diff --git a/fs/nfs/nfs4_fs.h b/fs/nfs/nfs4_fs.h
index af285cc..681c389 100644
--- a/fs/nfs/nfs4_fs.h
+++ b/fs/nfs/nfs4_fs.h
@@ -161,6 +161,10 @@ enum {
 	NFS_STATE_POSIX_LOCKS,		/* Posix locks are supported */
 	NFS_STATE_RECOVERY_FAILED,	/* OPEN stateid state recovery failed */
 	NFS_STATE_MAY_NOTIFY_LOCK,	/* server may CB_NOTIFY_LOCK */
+#ifdef CONFIG_NFS_V4_2
+	NFS_CLNT_DST_SSC_COPY_STATE,	/* dst server open state on client*/
+#endif /* CONFIG_NFS_V4_2 */
+
 };
 
 struct nfs4_state {
diff --git a/fs/nfs/nfs4file.c b/fs/nfs/nfs4file.c
index 0efba77..881b819 100644
--- a/fs/nfs/nfs4file.c
+++ b/fs/nfs/nfs4file.c
@@ -133,10 +133,15 @@ static ssize_t nfs4_copy_file_range(struct file *file_in, loff_t pos_in,
 				    struct file *file_out, loff_t pos_out,
 				    size_t count, unsigned int flags)
 {
+	ssize_t ret;
+
 	if (file_inode(file_in) == file_inode(file_out))
 		return -EINVAL;
-
-	return nfs42_proc_copy(file_in, pos_in, file_out, pos_out, count);
+retry:
+	ret = nfs42_proc_copy(file_in, pos_in, file_out, pos_out, count);
+	if (ret == -EAGAIN)
+		goto retry;
+	return ret;
 }
 
 static loff_t nfs4_file_llseek(struct file *filep, loff_t offset, int whence)
diff --git a/fs/nfs/nfs4state.c b/fs/nfs/nfs4state.c
index 8156bad..12dbaf0 100644
--- a/fs/nfs/nfs4state.c
+++ b/fs/nfs/nfs4state.c
@@ -1540,6 +1540,21 @@ static int nfs4_reclaim_open_state(struct nfs4_state_owner *sp, const struct nfs
 					&state->flags);
 				nfs4_put_open_state(state);
 				spin_lock(&sp->so_lock);
+#ifdef CONFIG_NFS_V4_2
+				if (test_bit(NFS_CLNT_DST_SSC_COPY_STATE, &state->flags)) {
+					struct nfs4_copy_state *copy;
+
+					spin_lock(&sp->so_server->nfs_client->cl_lock);
+					list_for_each_entry(copy, &sp->so_server->ss_copies, copies) {
+						if (memcmp(&state->stateid.other, &copy->parent_state->stateid.other, NFS4_STATEID_SIZE))
+							continue;
+						copy->flags = 1;
+						complete(&copy->completion);
+						break;
+					}
+					spin_unlock(&sp->so_server->nfs_client->cl_lock);
+				}
+#endif /* CONFIG_NFS_V4_2 */
 				goto restart;
 			}
 		}
diff --git a/include/linux/nfs_fs.h b/include/linux/nfs_fs.h
index a862b7f..d8164b6 100644
--- a/include/linux/nfs_fs.h
+++ b/include/linux/nfs_fs.h
@@ -189,6 +189,8 @@ struct nfs4_copy_state {
 	uint64_t		count;
 	struct nfs_writeverf	verf;
 	int			error;
+	int			flags;
+	struct nfs4_state	*parent_state;
 };
 
 /*
-- 
1.8.3.1