NFSD 6.8 Release Notes

The bulk of the patches for this release are clean-ups and minor bug
 fixes.
 
 There is one significant revert to mention: support for RDMA Read
 operations in the server's RPC-over-RDMA transport implementation
 has been fixed so it waits for Read completion in a way that avoids
 tying up an nfsd thread. This prevents a possible DoS vector if an
 RPC-over-RDMA client should become unresponsive during RDMA Read
 operations.
 
 As always I am grateful to NFSD contributors, reviewers, and
 testers.
 -----BEGIN PGP SIGNATURE-----
 
 iQIzBAABCAAdFiEEKLLlsBKG3yQ88j7+M2qzM29mf5cFAmWdW34ACgkQM2qzM29m
 f5fKmw/+PcjoNDWR55kTmOo8j0h4HF8rhunvP2C50svnnsX63y1WKkLaxyAFN/Hl
 UFucJDQBjJvwi+PEbGOXcjkizuG5mhRBFvFIYDJYGWsE1s7B/v3E/Servvt1wSek
 UjoTjknYrqH6R3YfA8zBaWRJUXwvVQW3Bzo4mShrQK7He9/7nBHdUe0aWbAA9oW3
 QgzKH/FzqCS03MvuxQv74KgBcl3diIrDaj041A3CtSnXzSKqwc3LaUAd5B4BL+oq
 GnxpV1rtZla50M4Ntddi+vSjUvHWZySQ1GEJj7rKLTwpGXkxM2NuMkGx676WR4Iv
 sYDX0fsica2elKbqJem8pk68qi6XEdZVAdoOHdgNJRClmYHby8xkrL/TYKiQZf42
 IN9FogoVSZ+vSdI158Weim9+0Jqf+ffIh57ZtOyQQQAGZkdhB6GhcbdHJhQ9eOgB
 LAiAL7bsoWvDmBh5m9KnBmQYGpZoDUa6AT0bIvGD2O4/MdpHBkyT8Xwt+210nPOK
 mpBtxe5O8cUcg7A5/TwnVRg5jKp4CF8VWh2R8sGDhcYV8UfRthB38h4rHNhv4vxt
 l6ZUgmtTxrs1rCeh6aoiWTKXeQmI8meWlcet7cxw/axAsaTXkYPi5mslxF9f4O8u
 nQ8q7LuZQy2CKZO/t98STwx7s9OJcDOwcy51rnKK85TlCwnxFWg=
 =mIKg
 -----END PGP SIGNATURE-----

Merge tag 'nfsd-6.8' of git://git.kernel.org/pub/scm/linux/kernel/git/cel/linux

Pull nfsd updates from Chuck Lever:
 "The bulk of the patches for this release are clean-ups and minor bug
  fixes.

  There is one significant revert to mention: support for RDMA Read
  operations in the server's RPC-over-RDMA transport implementation has
  been fixed so it waits for Read completion in a way that avoids tying
  up an nfsd thread. This prevents a possible DoS vector if an
  RPC-over-RDMA client should become unresponsive during RDMA Read
  operations.

  As always I am grateful to NFSD contributors, reviewers, and testers"

* tag 'nfsd-6.8' of git://git.kernel.org/pub/scm/linux/kernel/git/cel/linux: (56 commits)
  nfsd: rename nfsd_last_thread() to nfsd_destroy_serv()
  SUNRPC: discard sv_refcnt, and svc_get/svc_put
  svc: don't hold reference for poolstats, only mutex.
  SUNRPC: remove printk when back channel request not found
  svcrdma: Implement multi-stage Read completion again
  svcrdma: Copy construction of svc_rqst::rq_arg to rdma_read_complete()
  svcrdma: Add back svcxprt_rdma::sc_read_complete_q
  svcrdma: Add back svc_rdma_recv_ctxt::rc_pages
  svcrdma: Clean up comment in svc_rdma_accept()
  svcrdma: Remove queue-shortening warnings
  svcrdma: Remove pointer addresses shown in dprintk()
  svcrdma: Optimize svc_rdma_cc_init()
  svcrdma: De-duplicate completion ID initialization helpers
  svcrdma: Move the svc_rdma_cc_init() call
  svcrdma: Remove struct svc_rdma_read_info
  svcrdma: Update the synopsis of svc_rdma_read_special()
  svcrdma: Update the synopsis of svc_rdma_read_call_chunk()
  svcrdma: Update synopsis of svc_rdma_read_multiple_chunks()
  svcrdma: Update synopsis of svc_rdma_copy_inline_range()
  svcrdma: Update the synopsis of svc_rdma_read_data_item()
  ...

fs/lockd/svc.c

@@ -345,10 +345,10 @@ static int lockd_get(void)
 
 	serv->sv_maxconn = nlm_max_connections;
 	error = svc_set_num_threads(serv, NULL, 1);
-	/* The thread now holds the only reference */
+	if (error < 0) {
-	svc_put(serv);
+		svc_destroy(&serv);
-	if (error < 0)
 		return error;
+	}
 
 	nlmsvc_serv = serv;
 	register_inetaddr_notifier(&lockd_inetaddr_notifier);
@@ -372,11 +372,9 @@ static void lockd_put(void)
 	unregister_inet6addr_notifier(&lockd_inet6addr_notifier);
 #endif
 
-	svc_get(nlmsvc_serv);
 	svc_set_num_threads(nlmsvc_serv, NULL, 0);
-	svc_put(nlmsvc_serv);
 	timer_delete_sync(&nlmsvc_retry);
-	nlmsvc_serv = NULL;
+	svc_destroy(&nlmsvc_serv);
 	dprintk("lockd_down: service destroyed\n");
 }
 

fs/nfs/callback.c

@@ -187,7 +187,7 @@ static struct svc_serv *nfs_callback_create_svc(int minorversion)
 	 * Check whether we're already up and running.
 	 */
 	if (cb_info->serv)
-		return svc_get(cb_info->serv);
+		return cb_info->serv;
 
 	/*
 	 * Sanity check: if there's no task,
@@ -245,9 +245,10 @@ int nfs_callback_up(u32 minorversion, struct rpc_xprt *xprt)
 
 	cb_info->users++;
 err_net:
-	if (!cb_info->users)
+	if (!cb_info->users) {
-		cb_info->serv = NULL;
+		svc_set_num_threads(cb_info->serv, NULL, 0);
-	svc_put(serv);
+		svc_destroy(&cb_info->serv);
+	}
 err_create:
 	mutex_unlock(&nfs_callback_mutex);
 	return ret;
@@ -271,11 +272,9 @@ void nfs_callback_down(int minorversion, struct net *net)
 	nfs_callback_down_net(minorversion, serv, net);
 	cb_info->users--;
 	if (cb_info->users == 0) {
-		svc_get(serv);
 		svc_set_num_threads(serv, NULL, 0);
-		svc_put(serv);
 		dprintk("nfs_callback_down: service destroyed\n");
-		cb_info->serv = NULL;
+		svc_destroy(&cb_info->serv);
 	}
 	mutex_unlock(&nfs_callback_mutex);
 }

fs/nfs/callback.h

@@ -19,25 +19,6 @@ enum nfs4_callback_procnum {
 	CB_COMPOUND = 1,
 };
 
-enum nfs4_callback_opnum {
-	OP_CB_GETATTR = 3,
-	OP_CB_RECALL  = 4,
-/* Callback operations new to NFSv4.1 */
-	OP_CB_LAYOUTRECALL  = 5,
-	OP_CB_NOTIFY        = 6,
-	OP_CB_PUSH_DELEG    = 7,
-	OP_CB_RECALL_ANY    = 8,
-	OP_CB_RECALLABLE_OBJ_AVAIL = 9,
-	OP_CB_RECALL_SLOT   = 10,
-	OP_CB_SEQUENCE      = 11,
-	OP_CB_WANTS_CANCELLED = 12,
-	OP_CB_NOTIFY_LOCK   = 13,
-	OP_CB_NOTIFY_DEVICEID = 14,
-/* Callback operations new to NFSv4.2 */
-	OP_CB_OFFLOAD = 15,
-	OP_CB_ILLEGAL = 10044,
-};
-
 struct nfs4_slot;
 struct cb_process_state {
 	__be32			drc_status;

fs/nfsd/Kconfig

@@ -158,3 +158,19 @@ config NFSD_V4_SECURITY_LABEL
 
 	If you do not wish to enable fine-grained security labels SELinux or
 	Smack policies on NFSv4 files, say N.
+
+config NFSD_LEGACY_CLIENT_TRACKING
+	bool "Support legacy NFSv4 client tracking methods (DEPRECATED)"
+	depends on NFSD_V4
+	default n
+	help
+	  The NFSv4 server needs to store a small amount of information on
+	  stable storage in order to handle state recovery after reboot. Most
+	  modern deployments upcall to a userland daemon for this (nfsdcld),
+	  but older NFS servers may store information directly in a
+	  recoverydir, or spawn a process directly using a usermodehelper
+	  upcall.
+
+	  These legacy client tracking methods have proven to be probelmatic
+	  and will be removed in the future. Say Y here if you need support
+	  for them in the interim.

fs/nfsd/filecache.c

@@ -717,7 +717,7 @@ nfsd_file_cache_init(void)
 		return ret;
 
 	ret = -ENOMEM;
-	nfsd_filecache_wq = alloc_workqueue("nfsd_filecache", 0, 0);
+	nfsd_filecache_wq = alloc_workqueue("nfsd_filecache", WQ_UNBOUND, 0);
 	if (!nfsd_filecache_wq)
 		goto out;
 

fs/nfsd/netns.h

@@ -123,14 +123,9 @@ struct nfsd_net {
 	u32 clientid_counter;
 	u32 clverifier_counter;
 
-	struct svc_serv *nfsd_serv;
+	struct svc_info nfsd_info;
-	/* When a listening socket is added to nfsd, keep_active is set
+#define nfsd_serv nfsd_info.serv
-	 * and this justifies a reference on nfsd_serv.  This stops
+
-	 * nfsd_serv from being freed.  When the number of threads is
-	 * set, keep_active is cleared and the reference is dropped.  So
-	 * when the last thread exits, the service will be destroyed.
-	 */
-	int keep_active;
 
 	/*
 	 * clientid and stateid data for construction of net unique COPY

fs/nfsd/nfs4callback.c

@@ -31,6 +31,7 @@
  *  SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
  */
 
+#include <linux/nfs4.h>
 #include <linux/sunrpc/clnt.h>
 #include <linux/sunrpc/xprt.h>
 #include <linux/sunrpc/svc_xprt.h>
@@ -87,31 +88,6 @@ static void encode_bitmap4(struct xdr_stream *xdr, const __u32 *bitmap,
 	WARN_ON_ONCE(xdr_stream_encode_uint32_array(xdr, bitmap, len) < 0);
 }
 
-/*
- *	nfs_cb_opnum4
- *
- *	enum nfs_cb_opnum4 {
- *		OP_CB_GETATTR		= 3,
- *		  ...
- *	};
- */
-enum nfs_cb_opnum4 {
-	OP_CB_GETATTR			= 3,
-	OP_CB_RECALL			= 4,
-	OP_CB_LAYOUTRECALL		= 5,
-	OP_CB_NOTIFY			= 6,
-	OP_CB_PUSH_DELEG		= 7,
-	OP_CB_RECALL_ANY		= 8,
-	OP_CB_RECALLABLE_OBJ_AVAIL	= 9,
-	OP_CB_RECALL_SLOT		= 10,
-	OP_CB_SEQUENCE			= 11,
-	OP_CB_WANTS_CANCELLED		= 12,
-	OP_CB_NOTIFY_LOCK		= 13,
-	OP_CB_NOTIFY_DEVICEID		= 14,
-	OP_CB_OFFLOAD			= 15,
-	OP_CB_ILLEGAL			= 10044
-};
-
 static void encode_nfs_cb_opnum4(struct xdr_stream *xdr, enum nfs_cb_opnum4 op)
 {
 	__be32 *p;

fs/nfsd/nfs4proc.c

@@ -970,8 +970,11 @@ nfsd4_read(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
 	 * To ensure proper ordering, we therefore turn off zero copy if
 	 * the client wants us to do more in this compound:
 	 */
-	if (!nfsd4_last_compound_op(rqstp))
+	if (!nfsd4_last_compound_op(rqstp)) {
-		clear_bit(RQ_SPLICE_OK, &rqstp->rq_flags);
+		struct nfsd4_compoundargs *argp = rqstp->rq_argp;
+
+		argp->splice_ok = false;
+	}
 
 	/* check stateid */
 	status = nfs4_preprocess_stateid_op(rqstp, cstate, &cstate->current_fh,

fs/nfsd/nfs4recover.c

@@ -66,6 +66,7 @@ struct nfsd4_client_tracking_ops {
 static const struct nfsd4_client_tracking_ops nfsd4_cld_tracking_ops;
 static const struct nfsd4_client_tracking_ops nfsd4_cld_tracking_ops_v2;
 
+#ifdef CONFIG_NFSD_LEGACY_CLIENT_TRACKING
 /* Globals */
 static char user_recovery_dirname[PATH_MAX] = "/var/lib/nfs/v4recovery";
 
@@ -720,6 +721,7 @@ static const struct nfsd4_client_tracking_ops nfsd4_legacy_tracking_ops = {
 	.version	= 1,
 	.msglen		= 0,
 };
+#endif /* CONFIG_NFSD_LEGACY_CLIENT_TRACKING */
 
 /* Globals */
 #define NFSD_PIPE_DIR		"nfsd"
@@ -731,8 +733,10 @@ struct cld_net {
 	spinlock_t		 cn_lock;
 	struct list_head	 cn_list;
 	unsigned int		 cn_xid;
-	bool			 cn_has_legacy;
 	struct crypto_shash	*cn_tfm;
+#ifdef CONFIG_NFSD_LEGACY_CLIENT_TRACKING
+	bool			 cn_has_legacy;
+#endif
 };
 
 struct cld_upcall {
@@ -793,7 +797,6 @@ __cld_pipe_inprogress_downcall(const struct cld_msg_v2 __user *cmsg,
 	uint8_t cmd, princhashlen;
 	struct xdr_netobj name, princhash = { .len = 0, .data = NULL };
 	uint16_t namelen;
-	struct cld_net *cn = nn->cld_net;
 
 	if (get_user(cmd, &cmsg->cm_cmd)) {
 		dprintk("%s: error when copying cmd from userspace", __func__);
@@ -833,11 +836,15 @@ __cld_pipe_inprogress_downcall(const struct cld_msg_v2 __user *cmsg,
 				return PTR_ERR(name.data);
 			name.len = namelen;
 		}
+#ifdef CONFIG_NFSD_LEGACY_CLIENT_TRACKING
 		if (name.len > 5 && memcmp(name.data, "hash:", 5) == 0) {
+			struct cld_net *cn = nn->cld_net;
+
 			name.len = name.len - 5;
 			memmove(name.data, name.data + 5, name.len);
 			cn->cn_has_legacy = true;
 		}
+#endif
 		if (!nfs4_client_to_reclaim(name, princhash, nn)) {
 			kfree(name.data);
 			kfree(princhash.data);
@@ -1010,7 +1017,9 @@ __nfsd4_init_cld_pipe(struct net *net)
 	}
 
 	cn->cn_pipe->dentry = dentry;
+#ifdef CONFIG_NFSD_LEGACY_CLIENT_TRACKING
 	cn->cn_has_legacy = false;
+#endif
 	nn->cld_net = cn;
 	return 0;
 
@@ -1282,10 +1291,6 @@ nfsd4_cld_check(struct nfs4_client *clp)
 {
 	struct nfs4_client_reclaim *crp;
 	struct nfsd_net *nn = net_generic(clp->net, nfsd_net_id);
-	struct cld_net *cn = nn->cld_net;
-	int status;
-	char dname[HEXDIR_LEN];
-	struct xdr_netobj name;
 
 	/* did we already find that this client is stable? */
 	if (test_bit(NFSD4_CLIENT_STABLE, &clp->cl_flags))
@@ -1296,7 +1301,12 @@ nfsd4_cld_check(struct nfs4_client *clp)
 	if (crp)
 		goto found;
 
-	if (cn->cn_has_legacy) {
+#ifdef CONFIG_NFSD_LEGACY_CLIENT_TRACKING
+	if (nn->cld_net->cn_has_legacy) {
+		int status;
+		char dname[HEXDIR_LEN];
+		struct xdr_netobj name;
+
 		status = nfs4_make_rec_clidname(dname, &clp->cl_name);
 		if (status)
 			return -ENOENT;
@@ -1314,6 +1324,7 @@ nfsd4_cld_check(struct nfs4_client *clp)
 			goto found;
 
 	}
+#endif
 	return -ENOENT;
 found:
 	crp->cr_clp = clp;
@@ -1327,8 +1338,6 @@ nfsd4_cld_check_v2(struct nfs4_client *clp)
 	struct nfsd_net *nn = net_generic(clp->net, nfsd_net_id);
 	struct cld_net *cn = nn->cld_net;
 	int status;
-	char dname[HEXDIR_LEN];
-	struct xdr_netobj name;
 	struct crypto_shash *tfm = cn->cn_tfm;
 	struct xdr_netobj cksum;
 	char *principal = NULL;
@@ -1342,7 +1351,11 @@ nfsd4_cld_check_v2(struct nfs4_client *clp)
 	if (crp)
 		goto found;
 
+#ifdef CONFIG_NFSD_LEGACY_CLIENT_TRACKING
 	if (cn->cn_has_legacy) {
+		struct xdr_netobj name;
+		char dname[HEXDIR_LEN];
+
 		status = nfs4_make_rec_clidname(dname, &clp->cl_name);
 		if (status)
 			return -ENOENT;
@@ -1360,6 +1373,7 @@ nfsd4_cld_check_v2(struct nfs4_client *clp)
 			goto found;
 
 	}
+#endif
 	return -ENOENT;
 found:
 	if (crp->cr_princhash.len) {
@@ -1663,6 +1677,7 @@ static const struct nfsd4_client_tracking_ops nfsd4_cld_tracking_ops_v2 = {
 	.msglen		= sizeof(struct cld_msg_v2),
 };
 
+#ifdef CONFIG_NFSD_LEGACY_CLIENT_TRACKING
 /* upcall via usermodehelper */
 static char cltrack_prog[PATH_MAX] = "/sbin/nfsdcltrack";
 module_param_string(cltrack_prog, cltrack_prog, sizeof(cltrack_prog),
@@ -2007,12 +2022,46 @@ static const struct nfsd4_client_tracking_ops nfsd4_umh_tracking_ops = {
 	.msglen		= 0,
 };
 
+static inline int check_for_legacy_methods(int status, struct net *net)
+{
+	struct nfsd_net *nn = net_generic(net, nfsd_net_id);
+	struct path path;
+
+	/*
+	 * Next, try the UMH upcall.
+	 */
+	nn->client_tracking_ops = &nfsd4_umh_tracking_ops;
+	status = nn->client_tracking_ops->init(net);
+	if (!status)
+		return status;
+
+	/*
+	 * Finally, See if the recoverydir exists and is a directory.
+	 * If it is, then use the legacy ops.
+	 */
+	nn->client_tracking_ops = &nfsd4_legacy_tracking_ops;
+	status = kern_path(nfs4_recoverydir(), LOOKUP_FOLLOW, &path);
+	if (!status) {
+		status = !d_is_dir(path.dentry);
+		path_put(&path);
+		if (status)
+			return -ENOTDIR;
+		status = nn->client_tracking_ops->init(net);
+	}
+	return status;
+}
+#else
+static inline int check_for_legacy_methods(int status, struct net *net)
+{
+	return status;
+}
+#endif /* CONFIG_LEGACY_NFSD_CLIENT_TRACKING */
+
 int
 nfsd4_client_tracking_init(struct net *net)
 {
-	int status;
-	struct path path;
 	struct nfsd_net *nn = net_generic(net, nfsd_net_id);
+	int status;
 
 	/* just run the init if it the method is already decided */
 	if (nn->client_tracking_ops)
@@ -2030,29 +2079,9 @@ nfsd4_client_tracking_init(struct net *net)
 			return status;
 	}
 
-	/*
+	status = check_for_legacy_methods(status, net);
-	 * Next, try the UMH upcall.
+	if (status)
-	 */
+		goto out;
-	nn->client_tracking_ops = &nfsd4_umh_tracking_ops;
-	status = nn->client_tracking_ops->init(net);
-	if (!status)
-		return status;
-
-	/*
-	 * Finally, See if the recoverydir exists and is a directory.
-	 * If it is, then use the legacy ops.
-	 */
-	nn->client_tracking_ops = &nfsd4_legacy_tracking_ops;
-	status = kern_path(nfs4_recoverydir(), LOOKUP_FOLLOW, &path);
-	if (!status) {
-		status = d_is_dir(path.dentry);
-		path_put(&path);
-		if (!status) {
-			status = -EINVAL;
-			goto out;
-		}
-	}
-
 do_init:
 	status = nn->client_tracking_ops->init(net);
 out:

fs/nfsd/nfs4state.c

@@ -6575,7 +6575,7 @@ unlock:
 	spin_unlock(&nn->s2s_cp_lock);
 	if (!state)
 		return nfserr_bad_stateid;
-	if (!clp && state)
+	if (!clp)
 		*cps = state;
 	return 0;
 }

fs/nfsd/nfs4xdr.c

@@ -2524,8 +2524,9 @@ nfsd4_decode_compound(struct nfsd4_compoundargs *argp)
 	svc_reserve(argp->rqstp, max_reply + readbytes);
 	argp->rqstp->rq_cachetype = cachethis ? RC_REPLBUFF : RC_NOCACHE;
 
+	argp->splice_ok = nfsd_read_splice_ok(argp->rqstp);
 	if (readcount > 1 || max_reply > PAGE_SIZE - auth_slack)
-		clear_bit(RQ_SPLICE_OK, &argp->rqstp->rq_flags);
+		argp->splice_ok = false;
 
 	return true;
 }
@@ -4375,12 +4376,13 @@ static __be32
 nfsd4_encode_read(struct nfsd4_compoundres *resp, __be32 nfserr,
 		  union nfsd4_op_u *u)
 {
+	struct nfsd4_compoundargs *argp = resp->rqstp->rq_argp;
 	struct nfsd4_read *read = &u->read;
-	bool splice_ok = test_bit(RQ_SPLICE_OK, &resp->rqstp->rq_flags);
-	unsigned long maxcount;
 	struct xdr_stream *xdr = resp->xdr;
-	struct file *file;
 	int starting_len = xdr->buf->len;
+	bool splice_ok = argp->splice_ok;
+	unsigned long maxcount;
+	struct file *file;
 	__be32 *p;
 
 	if (nfserr)
@@ -5201,9 +5203,10 @@ static __be32
 nfsd4_encode_read_plus_data(struct nfsd4_compoundres *resp,
 			    struct nfsd4_read *read)
 {
-	bool splice_ok = test_bit(RQ_SPLICE_OK, &resp->rqstp->rq_flags);
+	struct nfsd4_compoundargs *argp = resp->rqstp->rq_argp;
 	struct file *file = read->rd_nf->nf_file;
 	struct xdr_stream *xdr = resp->xdr;
+	bool splice_ok = argp->splice_ok;
 	unsigned long maxcount;
 	__be32 nfserr, *p;
 

fs/nfsd/nfscache.c

@@ -364,8 +364,6 @@ nfsd_reply_cache_scan(struct shrinker *shrink, struct shrink_control *sc)
 		if (freed > sc->nr_to_scan)
 			break;
 	}
-
-	trace_nfsd_drc_gc(nn, freed);
 	return freed;
 }
 
@@ -508,7 +506,6 @@ int nfsd_cache_lookup(struct svc_rqst *rqstp, unsigned int start,
 	__wsum			csum;
 	struct nfsd_drc_bucket	*b;
 	int type = rqstp->rq_cachetype;
-	unsigned long freed;
 	LIST_HEAD(dispose);
 	int rtn = RC_DOIT;
 
@@ -538,8 +535,7 @@ int nfsd_cache_lookup(struct svc_rqst *rqstp, unsigned int start,
 	nfsd_prune_bucket_locked(nn, b, 3, &dispose);
 	spin_unlock(&b->cache_lock);
 
-	freed = nfsd_cacherep_dispose(&dispose);
+	nfsd_cacherep_dispose(&dispose);
-	trace_nfsd_drc_gc(nn, freed);
 
 	nfsd_stats_rc_misses_inc();
 	atomic_inc(&nn->num_drc_entries);

fs/nfsd/nfsctl.c

@@ -76,7 +76,9 @@ static ssize_t write_maxconn(struct file *file, char *buf, size_t size);
 #ifdef CONFIG_NFSD_V4
 static ssize_t write_leasetime(struct file *file, char *buf, size_t size);
 static ssize_t write_gracetime(struct file *file, char *buf, size_t size);
+#ifdef CONFIG_NFSD_LEGACY_CLIENT_TRACKING
 static ssize_t write_recoverydir(struct file *file, char *buf, size_t size);
+#endif
 static ssize_t write_v4_end_grace(struct file *file, char *buf, size_t size);
 #endif
 
@@ -93,7 +95,9 @@ static ssize_t (*const write_op[])(struct file *, char *, size_t) = {
 #ifdef CONFIG_NFSD_V4
 	[NFSD_Leasetime] = write_leasetime,
 	[NFSD_Gracetime] = write_gracetime,
+#ifdef CONFIG_NFSD_LEGACY_CLIENT_TRACKING
 	[NFSD_RecoveryDir] = write_recoverydir,
+#endif
 	[NFSD_V4EndGrace] = write_v4_end_grace,
 #endif
 };
@@ -179,7 +183,7 @@ static const struct file_operations pool_stats_operations = {
 	.open		= nfsd_pool_stats_open,
 	.read		= seq_read,
 	.llseek		= seq_lseek,
-	.release	= nfsd_pool_stats_release,
+	.release	= seq_release,
 };
 
 DEFINE_SHOW_ATTRIBUTE(nfsd_reply_cache_stats);
@@ -707,12 +711,9 @@ static ssize_t __write_ports_addfd(char *buf, struct net *net, const struct cred
 	serv = nn->nfsd_serv;
 	err = svc_addsock(serv, net, fd, buf, SIMPLE_TRANSACTION_LIMIT, cred);
 
-	if (err < 0 && !serv->sv_nrthreads && !nn->keep_active)
+	if (!serv->sv_nrthreads && list_empty(&nn->nfsd_serv->sv_permsocks))
-		nfsd_last_thread(net);
+		nfsd_destroy_serv(net);
-	else if (err >= 0 && !serv->sv_nrthreads && !xchg(&nn->keep_active, 1))
-		svc_get(serv);
 
-	svc_put(serv);
 	return err;
 }
 
@@ -750,10 +751,6 @@ static ssize_t __write_ports_addxprt(char *buf, struct net *net, const struct cr
 	if (err < 0 && err != -EAFNOSUPPORT)
 		goto out_close;
 
-	if (!serv->sv_nrthreads && !xchg(&nn->keep_active, 1))
-		svc_get(serv);
-
-	svc_put(serv);
 	return 0;
 out_close:
 	xprt = svc_find_xprt(serv, transport, net, PF_INET, port);
@@ -762,10 +759,9 @@ out_close:
 		svc_xprt_put(xprt);
 	}
 out_err:
-	if (!serv->sv_nrthreads && !nn->keep_active)
+	if (!serv->sv_nrthreads && list_empty(&nn->nfsd_serv->sv_permsocks))
-		nfsd_last_thread(net);
+		nfsd_destroy_serv(net);
 
-	svc_put(serv);
 	return err;
 }
 
@@ -1021,6 +1017,7 @@ static ssize_t write_gracetime(struct file *file, char *buf, size_t size)
 	return nfsd4_write_time(file, buf, size, &nn->nfsd4_grace, nn);
 }
 
+#ifdef CONFIG_NFSD_LEGACY_CLIENT_TRACKING
 static ssize_t __write_recoverydir(struct file *file, char *buf, size_t size,
 				   struct nfsd_net *nn)
 {
@@ -1081,6 +1078,7 @@ static ssize_t write_recoverydir(struct file *file, char *buf, size_t size)
 	mutex_unlock(&nfsd_mutex);
 	return rv;
 }
+#endif
 
 /*
  * write_v4_end_grace - release grace period for nfsd's v4.x lock manager

fs/nfsd/nfsd.h

@@ -148,7 +148,7 @@ int nfsd_vers(struct nfsd_net *nn, int vers, enum vers_op change);
 int nfsd_minorversion(struct nfsd_net *nn, u32 minorversion, enum vers_op change);
 void nfsd_reset_versions(struct nfsd_net *nn);
 int nfsd_create_serv(struct net *net);
-void nfsd_last_thread(struct net *net);
+void nfsd_destroy_serv(struct net *net);
 
 extern int nfsd_max_blksize;
 

fs/nfsd/nfssvc.c

@@ -59,15 +59,6 @@ static __be32			nfsd_init_request(struct svc_rqst *,
  * nfsd_mutex protects nn->nfsd_serv -- both the pointer itself and some members
  * of the svc_serv struct such as ->sv_temp_socks and ->sv_permsocks.
  *
- * If (out side the lock) nn->nfsd_serv is non-NULL, then it must point to a
- * properly initialised 'struct svc_serv' with ->sv_nrthreads > 0 (unless
- * nn->keep_active is set).  That number of nfsd threads must
- * exist and each must be listed in ->sp_all_threads in some entry of
- * ->sv_pools[].
- *
- * Each active thread holds a counted reference on nn->nfsd_serv, as does
- * the nn->keep_active flag and various transient calls to svc_get().
- *
  * Finally, the nfsd_mutex also protects some of the global variables that are
  * accessed when nfsd starts and that are settable via the write_* routines in
  * nfsctl.c. In particular:
@@ -359,13 +350,12 @@ static bool nfsd_needs_lockd(struct nfsd_net *nn)
  */
 void nfsd_copy_write_verifier(__be32 verf[2], struct nfsd_net *nn)
 {
-	int seq = 0;
+	unsigned int seq;
 
 	do {
-		read_seqbegin_or_lock(&nn->writeverf_lock, &seq);
+		seq = read_seqbegin(&nn->writeverf_lock);
 		memcpy(verf, nn->writeverf, sizeof(nn->writeverf));
-	} while (need_seqretry(&nn->writeverf_lock, seq));
+	} while (read_seqretry(&nn->writeverf_lock, seq));
-	done_seqretry(&nn->writeverf_lock, seq);
 }
 
 static void nfsd_reset_write_verifier_locked(struct nfsd_net *nn)
@@ -542,7 +532,11 @@ static struct notifier_block nfsd_inet6addr_notifier = {
 /* Only used under nfsd_mutex, so this atomic may be overkill: */
 static atomic_t nfsd_notifier_refcount = ATOMIC_INIT(0);
 
-void nfsd_last_thread(struct net *net)
+/**
+ * nfsd_destroy_serv - tear down NFSD's svc_serv for a namespace
+ * @net: network namespace the NFS service is associated with
+ */
+void nfsd_destroy_serv(struct net *net)
 {
 	struct nfsd_net *nn = net_generic(net, nfsd_net_id);
 	struct svc_serv *serv = nn->nfsd_serv;
@@ -564,7 +558,7 @@ void nfsd_last_thread(struct net *net)
 	/*
 	 * write_ports can create the server without actually starting
 	 * any threads--if we get shut down before any threads are
-	 * started, then nfsd_last_thread will be run before any of this
+	 * started, then nfsd_destroy_serv will be run before any of this
 	 * other initialization has been done except the rpcb information.
 	 */
 	svc_rpcb_cleanup(serv, net);
@@ -573,6 +567,7 @@ void nfsd_last_thread(struct net *net)
 
 	nfsd_shutdown_net(net);
 	nfsd_export_flush(net);
+	svc_destroy(&serv);
 }
 
 void nfsd_reset_versions(struct nfsd_net *nn)
@@ -647,11 +642,9 @@ void nfsd_shutdown_threads(struct net *net)
 		return;
 	}
 
-	svc_get(serv);
 	/* Kill outstanding nfsd threads */
 	svc_set_num_threads(serv, NULL, 0);
-	nfsd_last_thread(net);
+	nfsd_destroy_serv(net);
-	svc_put(serv);
 	mutex_unlock(&nfsd_mutex);
 }
 
@@ -667,10 +660,9 @@ int nfsd_create_serv(struct net *net)
 	struct svc_serv *serv;
 
 	WARN_ON(!mutex_is_locked(&nfsd_mutex));
-	if (nn->nfsd_serv) {
+	if (nn->nfsd_serv)
-		svc_get(nn->nfsd_serv);
 		return 0;
-	}
+
 	if (nfsd_max_blksize == 0)
 		nfsd_max_blksize = nfsd_get_default_max_blksize();
 	nfsd_reset_versions(nn);
@@ -681,10 +673,11 @@ int nfsd_create_serv(struct net *net)
 	serv->sv_maxconn = nn->max_connections;
 	error = svc_bind(serv, net);
 	if (error < 0) {
-		svc_put(serv);
+		svc_destroy(&serv);
 		return error;
 	}
 	spin_lock(&nfsd_notifier_lock);
+	nn->nfsd_info.mutex = &nfsd_mutex;
 	nn->nfsd_serv = serv;
 	spin_unlock(&nfsd_notifier_lock);
 
@@ -764,7 +757,6 @@ int nfsd_set_nrthreads(int n, int *nthreads, struct net *net)
 		nthreads[0] = 1;
 
 	/* apply the new numbers */
-	svc_get(nn->nfsd_serv);
 	for (i = 0; i < n; i++) {
 		err = svc_set_num_threads(nn->nfsd_serv,
 					  &nn->nfsd_serv->sv_pools[i],
@@ -772,7 +764,6 @@ int nfsd_set_nrthreads(int n, int *nthreads, struct net *net)
 		if (err)
 			break;
 	}
-	svc_put(nn->nfsd_serv);
 	return err;
 }
 
@@ -814,13 +805,8 @@ nfsd_svc(int nrservs, struct net *net, const struct cred *cred)
 		goto out_put;
 	error = serv->sv_nrthreads;
 out_put:
-	/* Threads now hold service active */
-	if (xchg(&nn->keep_active, 0))
-		svc_put(serv);
-
 	if (serv->sv_nrthreads == 0)
-		nfsd_last_thread(net);
+		nfsd_destroy_serv(net);
-	svc_put(serv);
 out:
 	mutex_unlock(&nfsd_mutex);
 	return error;
@@ -1083,28 +1069,7 @@ bool nfssvc_encode_voidres(struct svc_rqst *rqstp, struct xdr_stream *xdr)
 
 int nfsd_pool_stats_open(struct inode *inode, struct file *file)
 {
-	int ret;
 	struct nfsd_net *nn = net_generic(inode->i_sb->s_fs_info, nfsd_net_id);
 
-	mutex_lock(&nfsd_mutex);
+	return svc_pool_stats_open(&nn->nfsd_info, file);
-	if (nn->nfsd_serv == NULL) {
-		mutex_unlock(&nfsd_mutex);
-		return -ENODEV;
-	}
-	svc_get(nn->nfsd_serv);
-	ret = svc_pool_stats_open(nn->nfsd_serv, file);
-	mutex_unlock(&nfsd_mutex);
-	return ret;
-}
-
-int nfsd_pool_stats_release(struct inode *inode, struct file *file)
-{
-	struct seq_file *seq = file->private_data;
-	struct svc_serv *serv = seq->private;
-	int ret = seq_release(inode, file);
-
-	mutex_lock(&nfsd_mutex);
-	svc_put(serv);
-	mutex_unlock(&nfsd_mutex);
-	return ret;
 }

fs/nfsd/trace.h

@@ -1262,28 +1262,6 @@ TRACE_EVENT(nfsd_drc_mismatch,
 		__entry->ingress)
 );
 
-TRACE_EVENT_CONDITION(nfsd_drc_gc,
-	TP_PROTO(
-		const struct nfsd_net *nn,
-		unsigned long freed
-	),
-	TP_ARGS(nn, freed),
-	TP_CONDITION(freed > 0),
-	TP_STRUCT__entry(
-		__field(unsigned long long, boot_time)
-		__field(unsigned long, freed)
-		__field(int, total)
-	),
-	TP_fast_assign(
-		__entry->boot_time = nn->boot_time;
-		__entry->freed = freed;
-		__entry->total = atomic_read(&nn->num_drc_entries);
-	),
-	TP_printk("boot_time=%16llx total=%d freed=%lu",
-		__entry->boot_time, __entry->total, __entry->freed
-	)
-);
-
 TRACE_EVENT(nfsd_cb_args,
 	TP_PROTO(
 		const struct nfs4_client *clp,

fs/nfsd/vfs.c

@@ -1210,6 +1210,30 @@ out_nfserr:
 	return nfserr;
 }
 
+/**
+ * nfsd_read_splice_ok - check if spliced reading is supported
+ * @rqstp: RPC transaction context
+ *
+ * Return values:
+ *   %true: nfsd_splice_read() may be used
+ *   %false: nfsd_splice_read() must not be used
+ *
+ * NFS READ normally uses splice to send data in-place. However the
+ * data in cache can change after the reply's MIC is computed but
+ * before the RPC reply is sent. To prevent the client from
+ * rejecting the server-computed MIC in this somewhat rare case, do
+ * not use splice with the GSS integrity and privacy services.
+ */
+bool nfsd_read_splice_ok(struct svc_rqst *rqstp)
+{
+	switch (svc_auth_flavor(rqstp)) {
+	case RPC_AUTH_GSS_KRB5I:
+	case RPC_AUTH_GSS_KRB5P:
+		return false;
+	}
+	return true;
+}
+
 /**
  * nfsd_read - Read data from a file
  * @rqstp: RPC transaction context
@@ -1239,7 +1263,7 @@ __be32 nfsd_read(struct svc_rqst *rqstp, struct svc_fh *fhp,
 		return err;
 
 	file = nf->nf_file;
-	if (file->f_op->splice_read && test_bit(RQ_SPLICE_OK, &rqstp->rq_flags))
+	if (file->f_op->splice_read && nfsd_read_splice_ok(rqstp))
 		err = nfsd_splice_read(rqstp, fhp, file, offset, count, eof);
 	else
 		err = nfsd_iter_read(rqstp, fhp, file, offset, count, 0, eof);
@@ -2103,9 +2127,23 @@ static __be32 nfsd_buffered_readdir(struct file *file, struct svc_fh *fhp,
 	return cdp->err;
 }
 
-/*
+/**
- * Read entries from a directory.
+ * nfsd_readdir - Read entries from a directory
- * The  NFSv3/4 verifier we ignore for now.
+ * @rqstp: RPC transaction context
+ * @fhp: NFS file handle of directory to be read
+ * @offsetp: OUT: seek offset of final entry that was read
+ * @cdp: OUT: an eof error value
+ * @func: entry filler actor
+ *
+ * This implementation ignores the NFSv3/4 verifier cookie.
+ *
+ * NB: normal system calls hold file->f_pos_lock when calling
+ * ->iterate_shared and ->llseek, but nfsd_readdir() does not.
+ * Because the struct file acquired here is not visible to other
+ * threads, it's internal state does not need mutex protection.
+ *
+ * Returns nfs_ok on success, otherwise an nfsstat code is
+ * returned.
  */
 __be32
 nfsd_readdir(struct svc_rqst *rqstp, struct svc_fh *fhp, loff_t *offsetp, 

fs/nfsd/vfs.h

@@ -114,6 +114,7 @@ __be32		nfsd_iter_read(struct svc_rqst *rqstp, struct svc_fh *fhp,
 				struct file *file, loff_t offset,
 				unsigned long *count, unsigned int base,
 				u32 *eof);
+bool		nfsd_read_splice_ok(struct svc_rqst *rqstp);
 __be32		nfsd_read(struct svc_rqst *rqstp, struct svc_fh *fhp,
 				loff_t offset, unsigned long *count,
 				u32 *eof);

fs/nfsd/xdr4.h

@@ -840,6 +840,7 @@ struct nfsd4_compoundargs {
 	u32				minorversion;
 	u32				client_opcnt;
 	u32				opcnt;
+	bool				splice_ok;
 	struct nfsd4_op			*ops;
 	struct nfsd4_op			iops[8];
 };

include/linux/nfs4.h

@@ -869,4 +869,26 @@ enum {
 	RCA4_TYPE_MASK_OTHER_LAYOUT_MAX	= 15,
 };
 
+enum nfs_cb_opnum4 {
+	OP_CB_GETATTR = 3,
+	OP_CB_RECALL  = 4,
+
+	/* Callback operations new to NFSv4.1 */
+	OP_CB_LAYOUTRECALL  = 5,
+	OP_CB_NOTIFY        = 6,
+	OP_CB_PUSH_DELEG    = 7,
+	OP_CB_RECALL_ANY    = 8,
+	OP_CB_RECALLABLE_OBJ_AVAIL = 9,
+	OP_CB_RECALL_SLOT   = 10,
+	OP_CB_SEQUENCE      = 11,
+	OP_CB_WANTS_CANCELLED = 12,
+	OP_CB_NOTIFY_LOCK   = 13,
+	OP_CB_NOTIFY_DEVICEID = 14,
+
+	/* Callback operations new to NFSv4.2 */
+	OP_CB_OFFLOAD = 15,
+
+	OP_CB_ILLEGAL = 10044,
+};
+
 #endif

include/linux/sunrpc/svc.h

@@ -69,7 +69,6 @@ struct svc_serv {
 	struct svc_program *	sv_program;	/* RPC program */
 	struct svc_stat *	sv_stats;	/* RPC statistics */
 	spinlock_t		sv_lock;
-	struct kref		sv_refcnt;
 	unsigned int		sv_nrthreads;	/* # of server threads */
 	unsigned int		sv_maxconn;	/* max connections allowed or
 						 * '0' causing max to be based
@@ -97,31 +96,13 @@ struct svc_serv {
 #endif /* CONFIG_SUNRPC_BACKCHANNEL */
 };
 
-/**
+/* This is used by pool_stats to find and lock an svc */
- * svc_get() - increment reference count on a SUNRPC serv
+struct svc_info {
- * @serv:  the svc_serv to have count incremented
+	struct svc_serv		*serv;
- *
+	struct mutex		*mutex;
- * Returns: the svc_serv that was passed in.
+};
- */
-static inline struct svc_serv *svc_get(struct svc_serv *serv)
-{
-	kref_get(&serv->sv_refcnt);
-	return serv;
-}
 
-void svc_destroy(struct kref *);
+void svc_destroy(struct svc_serv **svcp);
-
-/**
- * svc_put - decrement reference count on a SUNRPC serv
- * @serv:  the svc_serv to have count decremented
- *
- * When the reference count reaches zero, svc_destroy()
- * is called to clean up and free the serv.
- */
-static inline void svc_put(struct svc_serv *serv)
-{
-	kref_put(&serv->sv_refcnt, svc_destroy);
-}
 
 /*
  * Maximum payload size supported by a kernel RPC server.
@@ -260,8 +241,6 @@ enum {
 	RQ_LOCAL,		/* local request */
 	RQ_USEDEFERRAL,		/* use deferral */
 	RQ_DROPME,		/* drop current reply */
-	RQ_SPLICE_OK,		/* turned off in gss privacy to prevent
-				 * encrypting page cache pages */
 	RQ_VICTIM,		/* Have agreed to shut down */
 	RQ_DATA,		/* request has data */
 };
@@ -433,7 +412,7 @@ void		   svc_exit_thread(struct svc_rqst *);
 struct svc_serv *  svc_create_pooled(struct svc_program *, unsigned int,
 				     int (*threadfn)(void *data));
 int		   svc_set_num_threads(struct svc_serv *, struct svc_pool *, int);
-int		   svc_pool_stats_open(struct svc_serv *serv, struct file *file);
+int		   svc_pool_stats_open(struct svc_info *si, struct file *file);
 void		   svc_process(struct svc_rqst *rqstp);
 void		   svc_process_bc(struct rpc_rqst *req, struct svc_rqst *rqstp);
 int		   svc_register(const struct svc_serv *, struct net *, const int,

include/linux/sunrpc/svc_rdma.h

@@ -65,6 +65,7 @@ extern unsigned int svcrdma_ord;
 extern unsigned int svcrdma_max_requests;
 extern unsigned int svcrdma_max_bc_requests;
 extern unsigned int svcrdma_max_req_size;
+extern struct workqueue_struct *svcrdma_wq;
 
 extern struct percpu_counter svcrdma_stat_read;
 extern struct percpu_counter svcrdma_stat_recv;
@@ -97,6 +98,7 @@ struct svcxprt_rdma {
 	u32		     sc_pending_recvs;
 	u32		     sc_recv_batch;
 	struct list_head     sc_rq_dto_q;
+	struct list_head     sc_read_complete_q;
 	spinlock_t	     sc_rq_dto_lock;
 	struct ib_qp         *sc_qp;
 	struct ib_cq         *sc_rq_cq;
@@ -115,6 +117,13 @@ struct svcxprt_rdma {
 /* sc_flags */
 #define RDMAXPRT_CONN_PENDING	3
 
+static inline struct svcxprt_rdma *svc_rdma_rqst_rdma(struct svc_rqst *rqstp)
+{
+	struct svc_xprt *xprt = rqstp->rq_xprt;
+
+	return container_of(xprt, struct svcxprt_rdma, sc_xprt);
+}
+
 /*
  * Default connection parameters
  */
@@ -126,6 +135,43 @@ enum {
 
 #define RPCSVC_MAXPAYLOAD_RDMA	RPCSVC_MAXPAYLOAD
 
+/**
+ * svc_rdma_send_cid_init - Initialize a Receive Queue completion ID
+ * @rdma: controlling transport
+ * @cid: completion ID to initialize
+ */
+static inline void svc_rdma_recv_cid_init(struct svcxprt_rdma *rdma,
+					  struct rpc_rdma_cid *cid)
+{
+	cid->ci_queue_id = rdma->sc_rq_cq->res.id;
+	cid->ci_completion_id = atomic_inc_return(&rdma->sc_completion_ids);
+}
+
+/**
+ * svc_rdma_send_cid_init - Initialize a Send Queue completion ID
+ * @rdma: controlling transport
+ * @cid: completion ID to initialize
+ */
+static inline void svc_rdma_send_cid_init(struct svcxprt_rdma *rdma,
+					  struct rpc_rdma_cid *cid)
+{
+	cid->ci_queue_id = rdma->sc_sq_cq->res.id;
+	cid->ci_completion_id = atomic_inc_return(&rdma->sc_completion_ids);
+}
+
+/*
+ * A chunk context tracks all I/O for moving one Read or Write
+ * chunk. This is a set of rdma_rw's that handle data movement
+ * for all segments of one chunk.
+ */
+struct svc_rdma_chunk_ctxt {
+	struct rpc_rdma_cid	cc_cid;
+	struct ib_cqe		cc_cqe;
+	struct list_head	cc_rwctxts;
+	ktime_t			cc_posttime;
+	int			cc_sqecount;
+};
+
 struct svc_rdma_recv_ctxt {
 	struct llist_node	rc_node;
 	struct list_head	rc_list;
@@ -136,22 +182,33 @@ struct svc_rdma_recv_ctxt {
 	void			*rc_recv_buf;
 	struct xdr_stream	rc_stream;
 	u32			rc_byte_len;
-	unsigned int		rc_page_count;
 	u32			rc_inv_rkey;
 	__be32			rc_msgtype;
 
+	/* State for pulling a Read chunk */
+	unsigned int		rc_pageoff;
+	unsigned int		rc_curpage;
+	unsigned int		rc_readbytes;
+	struct xdr_buf		rc_saved_arg;
+	struct svc_rdma_chunk_ctxt	rc_cc;
+
 	struct svc_rdma_pcl	rc_call_pcl;
 
 	struct svc_rdma_pcl	rc_read_pcl;
 	struct svc_rdma_chunk	*rc_cur_result_payload;
 	struct svc_rdma_pcl	rc_write_pcl;
 	struct svc_rdma_pcl	rc_reply_pcl;
+
+	unsigned int		rc_page_count;
+	struct page		*rc_pages[RPCSVC_MAXPAGES];
 };
 
 struct svc_rdma_send_ctxt {
 	struct llist_node	sc_node;
 	struct rpc_rdma_cid	sc_cid;
+	struct work_struct	sc_work;
 
+	struct svcxprt_rdma	*sc_rdma;
 	struct ib_send_wr	sc_send_wr;
 	struct ib_cqe		sc_cqe;
 	struct xdr_buf		sc_hdrbuf;
@@ -180,6 +237,11 @@ extern int svc_rdma_recvfrom(struct svc_rqst *);
 
 /* svc_rdma_rw.c */
 extern void svc_rdma_destroy_rw_ctxts(struct svcxprt_rdma *rdma);
+extern void svc_rdma_cc_init(struct svcxprt_rdma *rdma,
+			     struct svc_rdma_chunk_ctxt *cc);
+extern void svc_rdma_cc_release(struct svcxprt_rdma *rdma,
+				struct svc_rdma_chunk_ctxt *cc,
+				enum dma_data_direction dir);
 extern int svc_rdma_send_write_chunk(struct svcxprt_rdma *rdma,
 				     const struct svc_rdma_chunk *chunk,
 				     const struct xdr_buf *xdr);
@@ -200,7 +262,8 @@ extern int svc_rdma_send(struct svcxprt_rdma *rdma,
 			 struct svc_rdma_send_ctxt *ctxt);
 extern int svc_rdma_map_reply_msg(struct svcxprt_rdma *rdma,
 				  struct svc_rdma_send_ctxt *sctxt,
-				  const struct svc_rdma_recv_ctxt *rctxt,
+				  const struct svc_rdma_pcl *write_pcl,
+				  const struct svc_rdma_pcl *reply_pcl,
 				  const struct xdr_buf *xdr);
 extern void svc_rdma_send_error_msg(struct svcxprt_rdma *rdma,
 				    struct svc_rdma_send_ctxt *sctxt,

include/linux/sunrpc/svcauth.h

@@ -131,8 +131,11 @@ enum svc_auth_status {
  *   This call releases a domain.
  *
  * set_client()
- *   Givens a pending request (struct svc_rqst), finds and assigns
+ *   Given a pending request (struct svc_rqst), finds and assigns
  *   an appropriate 'auth_domain' as the client.
+ *
+ * pseudoflavor()
+ *   Returns RPC_AUTH pseudoflavor in use by @rqstp.
  */
 struct auth_ops {
 	char *	name;
@@ -143,11 +146,13 @@ struct auth_ops {
 	int			(*release)(struct svc_rqst *rqstp);
 	void			(*domain_release)(struct auth_domain *dom);
 	enum svc_auth_status	(*set_client)(struct svc_rqst *rqstp);
+	rpc_authflavor_t	(*pseudoflavor)(struct svc_rqst *rqstp);
 };
 
 struct svc_xprt;
 
 extern enum svc_auth_status svc_authenticate(struct svc_rqst *rqstp);
+extern rpc_authflavor_t svc_auth_flavor(struct svc_rqst *rqstp);
 extern int	svc_authorise(struct svc_rqst *rqstp);
 extern enum svc_auth_status svc_set_client(struct svc_rqst *rqstp);
 extern int	svc_auth_register(rpc_authflavor_t flavor, struct auth_ops *aops);

include/trace/events/rpcrdma.h

@@ -22,6 +22,36 @@
  ** Event classes
  **/
 
+DECLARE_EVENT_CLASS(rpcrdma_simple_cid_class,
+	TP_PROTO(
+		const struct rpc_rdma_cid *cid
+	),
+
+	TP_ARGS(cid),
+
+	TP_STRUCT__entry(
+		__field(u32, cq_id)
+		__field(int, completion_id)
+	),
+
+	TP_fast_assign(
+		__entry->cq_id = cid->ci_queue_id;
+		__entry->completion_id = cid->ci_completion_id;
+	),
+
+	TP_printk("cq.id=%d cid=%d",
+		__entry->cq_id, __entry->completion_id
+	)
+);
+
+#define DEFINE_SIMPLE_CID_EVENT(name)					\
+		DEFINE_EVENT(rpcrdma_simple_cid_class, name,		\
+				TP_PROTO(				\
+					const struct rpc_rdma_cid *cid	\
+				),					\
+				TP_ARGS(cid)				\
+		)
+
 DECLARE_EVENT_CLASS(rpcrdma_completion_class,
 	TP_PROTO(
 		const struct ib_wc *wc,
@@ -62,37 +92,6 @@ DECLARE_EVENT_CLASS(rpcrdma_completion_class,
 				),					\
 				TP_ARGS(wc, cid))
 
-DECLARE_EVENT_CLASS(rpcrdma_send_completion_class,
-	TP_PROTO(
-		const struct ib_wc *wc,
-		const struct rpc_rdma_cid *cid
-	),
-
-	TP_ARGS(wc, cid),
-
-	TP_STRUCT__entry(
-		__field(u32, cq_id)
-		__field(int, completion_id)
-	),
-
-	TP_fast_assign(
-		__entry->cq_id = cid->ci_queue_id;
-		__entry->completion_id = cid->ci_completion_id;
-	),
-
-	TP_printk("cq.id=%u cid=%d",
-		__entry->cq_id, __entry->completion_id
-	)
-);
-
-#define DEFINE_SEND_COMPLETION_EVENT(name)				\
-		DEFINE_EVENT(rpcrdma_send_completion_class, name,	\
-				TP_PROTO(				\
-					const struct ib_wc *wc,		\
-					const struct rpc_rdma_cid *cid	\
-				),					\
-				TP_ARGS(wc, cid))
-
 DECLARE_EVENT_CLASS(rpcrdma_send_flush_class,
 	TP_PROTO(
 		const struct ib_wc *wc,
@@ -978,27 +977,7 @@ TRACE_EVENT(xprtrdma_post_send_err,
 	)
 );
 
-TRACE_EVENT(xprtrdma_post_recv,
+DEFINE_SIMPLE_CID_EVENT(xprtrdma_post_recv);
-	TP_PROTO(
-		const struct rpcrdma_rep *rep
-	),
-
-	TP_ARGS(rep),
-
-	TP_STRUCT__entry(
-		__field(u32, cq_id)
-		__field(int, completion_id)
-	),
-
-	TP_fast_assign(
-		__entry->cq_id = rep->rr_cid.ci_queue_id;
-		__entry->completion_id = rep->rr_cid.ci_completion_id;
-	),
-
-	TP_printk("cq.id=%d cid=%d",
-		__entry->cq_id, __entry->completion_id
-	)
-);
 
 TRACE_EVENT(xprtrdma_post_recvs,
 	TP_PROTO(
@@ -1783,29 +1762,29 @@ DEFINE_ERROR_EVENT(chunk);
 
 DECLARE_EVENT_CLASS(svcrdma_dma_map_class,
 	TP_PROTO(
-		const struct svcxprt_rdma *rdma,
+		const struct rpc_rdma_cid *cid,
 		u64 dma_addr,
 		u32 length
 	),
 
-	TP_ARGS(rdma, dma_addr, length),
+	TP_ARGS(cid, dma_addr, length),
 
 	TP_STRUCT__entry(
+		__field(u32, cq_id)
+		__field(int, completion_id)
 		__field(u64, dma_addr)
 		__field(u32, length)
-		__string(device, rdma->sc_cm_id->device->name)
-		__string(addr, rdma->sc_xprt.xpt_remotebuf)
 	),
 
 	TP_fast_assign(
+		__entry->cq_id = cid->ci_queue_id;
+		__entry->completion_id = cid->ci_completion_id;
 		__entry->dma_addr = dma_addr;
 		__entry->length = length;
-		__assign_str(device, rdma->sc_cm_id->device->name);
-		__assign_str(addr, rdma->sc_xprt.xpt_remotebuf);
 	),
 
-	TP_printk("addr=%s device=%s dma_addr=%llu length=%u",
+	TP_printk("cq.id=%u cid=%d dma_addr=%llu length=%u",
-		__get_str(addr), __get_str(device),
+		__entry->cq_id, __entry->completion_id,
 		__entry->dma_addr, __entry->length
 	)
 );
@@ -1813,11 +1792,12 @@ DECLARE_EVENT_CLASS(svcrdma_dma_map_class,
 #define DEFINE_SVC_DMA_EVENT(name)					\
 		DEFINE_EVENT(svcrdma_dma_map_class, svcrdma_##name,	\
 				TP_PROTO(				\
-					const struct svcxprt_rdma *rdma,\
+					const struct rpc_rdma_cid *cid, \
 					u64 dma_addr,			\
 					u32 length			\
 				),					\
-				TP_ARGS(rdma, dma_addr, length))
+				TP_ARGS(cid, dma_addr, length)		\
+		)
 
 DEFINE_SVC_DMA_EVENT(dma_map_page);
 DEFINE_SVC_DMA_EVENT(dma_map_err);
@@ -1826,33 +1806,37 @@ DEFINE_SVC_DMA_EVENT(dma_unmap_page);
 TRACE_EVENT(svcrdma_dma_map_rw_err,
 	TP_PROTO(
 		const struct svcxprt_rdma *rdma,
+		u64 offset,
+		u32 handle,
 		unsigned int nents,
 		int status
 	),
 
-	TP_ARGS(rdma, nents, status),
+	TP_ARGS(rdma, offset, handle, nents, status),
 
 	TP_STRUCT__entry(
-		__field(int, status)
+		__field(u32, cq_id)
+		__field(u32, handle)
+		__field(u64, offset)
 		__field(unsigned int, nents)
-		__string(device, rdma->sc_cm_id->device->name)
+		__field(int, status)
-		__string(addr, rdma->sc_xprt.xpt_remotebuf)
 	),
 
 	TP_fast_assign(
-		__entry->status = status;
+		__entry->cq_id = rdma->sc_sq_cq->res.id;
+		__entry->handle = handle;
+		__entry->offset = offset;
 		__entry->nents = nents;
-		__assign_str(device, rdma->sc_cm_id->device->name);
+		__entry->status = status;
-		__assign_str(addr, rdma->sc_xprt.xpt_remotebuf);
 	),
 
-	TP_printk("addr=%s device=%s nents=%u status=%d",
+	TP_printk("cq.id=%u 0x%016llx:0x%08x nents=%u status=%d",
-		__get_str(addr), __get_str(device), __entry->nents,
+		__entry->cq_id, (unsigned long long)__entry->offset,
-		__entry->status
+		__entry->handle, __entry->nents, __entry->status
 	)
 );
 
-TRACE_EVENT(svcrdma_no_rwctx_err,
+TRACE_EVENT(svcrdma_rwctx_empty,
 	TP_PROTO(
 		const struct svcxprt_rdma *rdma,
 		unsigned int num_sges
@@ -1861,79 +1845,75 @@ TRACE_EVENT(svcrdma_no_rwctx_err,
 	TP_ARGS(rdma, num_sges),
 
 	TP_STRUCT__entry(
+		__field(u32, cq_id)
 		__field(unsigned int, num_sges)
-		__string(device, rdma->sc_cm_id->device->name)
-		__string(addr, rdma->sc_xprt.xpt_remotebuf)
 	),
 
 	TP_fast_assign(
+		__entry->cq_id = rdma->sc_sq_cq->res.id;
 		__entry->num_sges = num_sges;
-		__assign_str(device, rdma->sc_cm_id->device->name);
-		__assign_str(addr, rdma->sc_xprt.xpt_remotebuf);
 	),
 
-	TP_printk("addr=%s device=%s num_sges=%d",
+	TP_printk("cq.id=%u num_sges=%d",
-		__get_str(addr), __get_str(device), __entry->num_sges
+		__entry->cq_id, __entry->num_sges
 	)
 );
 
 TRACE_EVENT(svcrdma_page_overrun_err,
 	TP_PROTO(
-		const struct svcxprt_rdma *rdma,
+		const struct rpc_rdma_cid *cid,
-		const struct svc_rqst *rqst,
 		unsigned int pageno
 	),
 
-	TP_ARGS(rdma, rqst, pageno),
+	TP_ARGS(cid, pageno),
 
 	TP_STRUCT__entry(
+		__field(u32, cq_id)
+		__field(int, completion_id)
 		__field(unsigned int, pageno)
-		__field(u32, xid)
-		__string(device, rdma->sc_cm_id->device->name)
-		__string(addr, rdma->sc_xprt.xpt_remotebuf)
 	),
 
 	TP_fast_assign(
+		__entry->cq_id = cid->ci_queue_id;
+		__entry->completion_id = cid->ci_completion_id;
 		__entry->pageno = pageno;
-		__entry->xid = __be32_to_cpu(rqst->rq_xid);
-		__assign_str(device, rdma->sc_cm_id->device->name);
-		__assign_str(addr, rdma->sc_xprt.xpt_remotebuf);
 	),
 
-	TP_printk("addr=%s device=%s xid=0x%08x pageno=%u", __get_str(addr),
+	TP_printk("cq.id=%u cid=%d pageno=%u",
-		__get_str(device), __entry->xid, __entry->pageno
+		__entry->cq_id, __entry->completion_id,
+		__entry->pageno
 	)
 );
 
 TRACE_EVENT(svcrdma_small_wrch_err,
 	TP_PROTO(
-		const struct svcxprt_rdma *rdma,
+		const struct rpc_rdma_cid *cid,
 		unsigned int remaining,
 		unsigned int seg_no,
 		unsigned int num_segs
 	),
 
-	TP_ARGS(rdma, remaining, seg_no, num_segs),
+	TP_ARGS(cid, remaining, seg_no, num_segs),
 
 	TP_STRUCT__entry(
+		__field(u32, cq_id)
+		__field(int, completion_id)
 		__field(unsigned int, remaining)
 		__field(unsigned int, seg_no)
 		__field(unsigned int, num_segs)
-		__string(device, rdma->sc_cm_id->device->name)
-		__string(addr, rdma->sc_xprt.xpt_remotebuf)
 	),
 
 	TP_fast_assign(
+		__entry->cq_id = cid->ci_queue_id;
+		__entry->completion_id = cid->ci_completion_id;
 		__entry->remaining = remaining;
 		__entry->seg_no = seg_no;
 		__entry->num_segs = num_segs;
-		__assign_str(device, rdma->sc_cm_id->device->name);
-		__assign_str(addr, rdma->sc_xprt.xpt_remotebuf);
 	),
 
-	TP_printk("addr=%s device=%s remaining=%u seg_no=%u num_segs=%u",
+	TP_printk("cq.id=%u cid=%d remaining=%u seg_no=%u num_segs=%u",
-		__get_str(addr), __get_str(device), __entry->remaining,
+		__entry->cq_id, __entry->completion_id,
-		__entry->seg_no, __entry->num_segs
+		__entry->remaining, __entry->seg_no, __entry->num_segs
 	)
 );
 
@@ -2020,31 +2000,11 @@ TRACE_EVENT(svcrdma_post_send,
 	)
 );
 
-DEFINE_SEND_COMPLETION_EVENT(svcrdma_wc_send);
+DEFINE_SIMPLE_CID_EVENT(svcrdma_wc_send);
 DEFINE_SEND_FLUSH_EVENT(svcrdma_wc_send_flush);
 DEFINE_SEND_FLUSH_EVENT(svcrdma_wc_send_err);
 
-TRACE_EVENT(svcrdma_post_recv,
+DEFINE_SIMPLE_CID_EVENT(svcrdma_post_recv);
-	TP_PROTO(
-		const struct svc_rdma_recv_ctxt *ctxt
-	),
-
-	TP_ARGS(ctxt),
-
-	TP_STRUCT__entry(
-		__field(u32, cq_id)
-		__field(int, completion_id)
-	),
-
-	TP_fast_assign(
-		__entry->cq_id = ctxt->rc_cid.ci_queue_id;
-		__entry->completion_id = ctxt->rc_cid.ci_completion_id;
-	),
-
-	TP_printk("cq.id=%d cid=%d",
-		__entry->cq_id, __entry->completion_id
-	)
-);
 
 DEFINE_RECEIVE_SUCCESS_EVENT(svcrdma_wc_recv);
 DEFINE_RECEIVE_FLUSH_EVENT(svcrdma_wc_recv_flush);
@@ -2152,8 +2112,9 @@ TRACE_EVENT(svcrdma_wc_read,
 
 DEFINE_SEND_FLUSH_EVENT(svcrdma_wc_read_flush);
 DEFINE_SEND_FLUSH_EVENT(svcrdma_wc_read_err);
+DEFINE_SIMPLE_CID_EVENT(svcrdma_read_finished);
 
-DEFINE_SEND_COMPLETION_EVENT(svcrdma_wc_write);
+DEFINE_SIMPLE_CID_EVENT(svcrdma_wc_write);
 DEFINE_SEND_FLUSH_EVENT(svcrdma_wc_write_flush);
 DEFINE_SEND_FLUSH_EVENT(svcrdma_wc_write_err);
 
@@ -2184,65 +2145,74 @@ TRACE_EVENT(svcrdma_qp_error,
 	)
 );
 
-DECLARE_EVENT_CLASS(svcrdma_sendqueue_event,
+DECLARE_EVENT_CLASS(svcrdma_sendqueue_class,
 	TP_PROTO(
-		const struct svcxprt_rdma *rdma
+		const struct svcxprt_rdma *rdma,
+		const struct rpc_rdma_cid *cid
 	),
 
-	TP_ARGS(rdma),
+	TP_ARGS(rdma, cid),
 
 	TP_STRUCT__entry(
+		__field(u32, cq_id)
+		__field(int, completion_id)
 		__field(int, avail)
 		__field(int, depth)
-		__string(addr, rdma->sc_xprt.xpt_remotebuf)
 	),
 
 	TP_fast_assign(
+		__entry->cq_id = cid->ci_queue_id;
+		__entry->completion_id = cid->ci_completion_id;
 		__entry->avail = atomic_read(&rdma->sc_sq_avail);
 		__entry->depth = rdma->sc_sq_depth;
-		__assign_str(addr, rdma->sc_xprt.xpt_remotebuf);
 	),
 
-	TP_printk("addr=%s sc_sq_avail=%d/%d",
+	TP_printk("cq.id=%u cid=%d sc_sq_avail=%d/%d",
-		__get_str(addr), __entry->avail, __entry->depth
+		__entry->cq_id, __entry->completion_id,
+		__entry->avail, __entry->depth
 	)
 );
 
 #define DEFINE_SQ_EVENT(name)						\
-		DEFINE_EVENT(svcrdma_sendqueue_event, svcrdma_sq_##name,\
+		DEFINE_EVENT(svcrdma_sendqueue_class, name,		\
-				TP_PROTO(				\
+			TP_PROTO(					\
-					const struct svcxprt_rdma *rdma \
+				const struct svcxprt_rdma *rdma,	\
-				),					\
+				const struct rpc_rdma_cid *cid		\
-				TP_ARGS(rdma))
+			),						\
+			TP_ARGS(rdma, cid)				\
+		)
 
-DEFINE_SQ_EVENT(full);
+DEFINE_SQ_EVENT(svcrdma_sq_full);
-DEFINE_SQ_EVENT(retry);
+DEFINE_SQ_EVENT(svcrdma_sq_retry);
 
 TRACE_EVENT(svcrdma_sq_post_err,
 	TP_PROTO(
 		const struct svcxprt_rdma *rdma,
+		const struct rpc_rdma_cid *cid,
 		int status
 	),
 
-	TP_ARGS(rdma, status),
+	TP_ARGS(rdma, cid, status),
 
 	TP_STRUCT__entry(
+		__field(u32, cq_id)
+		__field(int, completion_id)
 		__field(int, avail)
 		__field(int, depth)
 		__field(int, status)
-		__string(addr, rdma->sc_xprt.xpt_remotebuf)
 	),
 
 	TP_fast_assign(
+		__entry->cq_id = cid->ci_queue_id;
+		__entry->completion_id = cid->ci_completion_id;
 		__entry->avail = atomic_read(&rdma->sc_sq_avail);
 		__entry->depth = rdma->sc_sq_depth;
 		__entry->status = status;
-		__assign_str(addr, rdma->sc_xprt.xpt_remotebuf);
 	),
 
-	TP_printk("addr=%s sc_sq_avail=%d/%d status=%d",
+	TP_printk("cq.id=%u cid=%d sc_sq_avail=%d/%d status=%d",
-		__get_str(addr), __entry->avail, __entry->depth,
+		__entry->cq_id, __entry->completion_id,
-		__entry->status
+		__entry->avail, __entry->depth, __entry->status
 	)
 );
 

include/trace/events/sunrpc.h

@@ -1675,7 +1675,6 @@ DEFINE_SVCXDRBUF_EVENT(sendto);
 	svc_rqst_flag(LOCAL)						\
 	svc_rqst_flag(USEDEFERRAL)					\
 	svc_rqst_flag(DROPME)						\
-	svc_rqst_flag(SPLICE_OK)					\
 	svc_rqst_flag(VICTIM)						\
 	svc_rqst_flag_end(DATA)
 

net/sunrpc/auth_gss/svcauth_gss.c

@@ -866,14 +866,6 @@ svcauth_gss_unwrap_integ(struct svc_rqst *rqstp, u32 seq, struct gss_ctx *ctx)
 	struct xdr_buf databody_integ;
 	struct xdr_netobj checksum;
 
-	/* NFS READ normally uses splice to send data in-place. However
-	 * the data in cache can change after the reply's MIC is computed
-	 * but before the RPC reply is sent. To prevent the client from
-	 * rejecting the server-computed MIC in this somewhat rare case,
-	 * do not use splice with the GSS integrity service.
-	 */
-	clear_bit(RQ_SPLICE_OK, &rqstp->rq_flags);
-
 	/* Did we already verify the signature on the original pass through? */
 	if (rqstp->rq_deferred)
 		return 0;
@@ -948,8 +940,6 @@ svcauth_gss_unwrap_priv(struct svc_rqst *rqstp, u32 seq, struct gss_ctx *ctx)
 	struct xdr_buf *buf = xdr->buf;
 	unsigned int saved_len;
 
-	clear_bit(RQ_SPLICE_OK, &rqstp->rq_flags);
-
 	if (xdr_stream_decode_u32(xdr, &len) < 0)
 		goto unwrap_failed;
 	if (rqstp->rq_deferred) {
@@ -2014,6 +2004,11 @@ svcauth_gss_domain_release(struct auth_domain *dom)
 	call_rcu(&dom->rcu_head, svcauth_gss_domain_release_rcu);
 }
 
+static rpc_authflavor_t svcauth_gss_pseudoflavor(struct svc_rqst *rqstp)
+{
+	return svcauth_gss_flavor(rqstp->rq_gssclient);
+}
+
 static struct auth_ops svcauthops_gss = {
 	.name		= "rpcsec_gss",
 	.owner		= THIS_MODULE,
@@ -2022,6 +2017,7 @@ static struct auth_ops svcauthops_gss = {
 	.release	= svcauth_gss_release,
 	.domain_release = svcauth_gss_domain_release,
 	.set_client	= svcauth_gss_set_client,
+	.pseudoflavor	= svcauth_gss_pseudoflavor,
 };
 
 static int rsi_cache_create_net(struct net *net)

net/sunrpc/svc.c

@@ -463,7 +463,6 @@ __svc_create(struct svc_program *prog, unsigned int bufsize, int npools,
 		return NULL;
 	serv->sv_name      = prog->pg_name;
 	serv->sv_program   = prog;
-	kref_init(&serv->sv_refcnt);
 	serv->sv_stats     = prog->pg_stats;
 	if (bufsize > RPCSVC_MAXPAYLOAD)
 		bufsize = RPCSVC_MAXPAYLOAD;
@@ -564,11 +563,13 @@ EXPORT_SYMBOL_GPL(svc_create_pooled);
  * protect sv_permsocks and sv_tempsocks.
  */
 void
-svc_destroy(struct kref *ref)
+svc_destroy(struct svc_serv **servp)
 {
-	struct svc_serv *serv = container_of(ref, struct svc_serv, sv_refcnt);
+	struct svc_serv *serv = *servp;
 	unsigned int i;
 
+	*servp = NULL;
+
 	dprintk("svc: svc_destroy(%s)\n", serv->sv_program->pg_name);
 	timer_shutdown_sync(&serv->sv_temptimer);
 
@@ -675,7 +676,6 @@ svc_prepare_thread(struct svc_serv *serv, struct svc_pool *pool, int node)
 	if (!rqstp)
 		return ERR_PTR(-ENOMEM);
 
-	svc_get(serv);
 	spin_lock_bh(&serv->sv_lock);
 	serv->sv_nrthreads += 1;
 	spin_unlock_bh(&serv->sv_lock);
@@ -935,11 +935,6 @@ svc_exit_thread(struct svc_rqst *rqstp)
 
 	svc_rqst_free(rqstp);
 
-	svc_put(serv);
-	/* That svc_put() cannot be the last, because the thread
-	 * waiting for SP_VICTIM_REMAINS to clear must hold
-	 * a reference. So it is still safe to access pool.
-	 */
 	clear_and_wake_up_bit(SP_VICTIM_REMAINS, &pool->sp_flags);
 }
 EXPORT_SYMBOL_GPL(svc_exit_thread);
@@ -1305,8 +1300,6 @@ svc_process_common(struct svc_rqst *rqstp)
 	int			rc;
 	__be32			*p;
 
-	/* Will be turned off by GSS integrity and privacy services */
-	set_bit(RQ_SPLICE_OK, &rqstp->rq_flags);
 	/* Will be turned off only when NFSv4 Sessions are used */
 	set_bit(RQ_USEDEFERRAL, &rqstp->rq_flags);
 	clear_bit(RQ_DROPME, &rqstp->rq_flags);

net/sunrpc/svc_xprt.c

@@ -1362,29 +1362,36 @@ int svc_xprt_names(struct svc_serv *serv, char *buf, const int buflen)
 }
 EXPORT_SYMBOL_GPL(svc_xprt_names);
 
-
 /*----------------------------------------------------------------------------*/
 
 static void *svc_pool_stats_start(struct seq_file *m, loff_t *pos)
 {
 	unsigned int pidx = (unsigned int)*pos;
-	struct svc_serv *serv = m->private;
+	struct svc_info *si = m->private;
 
 	dprintk("svc_pool_stats_start, *pidx=%u\n", pidx);
 
+	mutex_lock(si->mutex);
+
 	if (!pidx)
 		return SEQ_START_TOKEN;
-	return (pidx > serv->sv_nrpools ? NULL : &serv->sv_pools[pidx-1]);
+	if (!si->serv)
+		return NULL;
+	return pidx > si->serv->sv_nrpools ? NULL
+		: &si->serv->sv_pools[pidx - 1];
 }
 
 static void *svc_pool_stats_next(struct seq_file *m, void *p, loff_t *pos)
 {
 	struct svc_pool *pool = p;
-	struct svc_serv *serv = m->private;
+	struct svc_info *si = m->private;
+	struct svc_serv *serv = si->serv;
 
 	dprintk("svc_pool_stats_next, *pos=%llu\n", *pos);
 
-	if (p == SEQ_START_TOKEN) {
+	if (!serv) {
+		pool = NULL;
+	} else if (p == SEQ_START_TOKEN) {
 		pool = &serv->sv_pools[0];
 	} else {
 		unsigned int pidx = (pool - &serv->sv_pools[0]);
@@ -1399,6 +1406,9 @@ static void *svc_pool_stats_next(struct seq_file *m, void *p, loff_t *pos)
 
 static void svc_pool_stats_stop(struct seq_file *m, void *p)
 {
+	struct svc_info *si = m->private;
+
+	mutex_unlock(si->mutex);
 }
 
 static int svc_pool_stats_show(struct seq_file *m, void *p)
@@ -1426,14 +1436,18 @@ static const struct seq_operations svc_pool_stats_seq_ops = {
 	.show	= svc_pool_stats_show,
 };
 
-int svc_pool_stats_open(struct svc_serv *serv, struct file *file)
+int svc_pool_stats_open(struct svc_info *info, struct file *file)
 {
+	struct seq_file *seq;
 	int err;
 
 	err = seq_open(file, &svc_pool_stats_seq_ops);
-	if (!err)
+	if (err)
-		((struct seq_file *) file->private_data)->private = serv;
+		return err;
-	return err;
+	seq = file->private_data;
+	seq->private = info;
+
+	return 0;
 }
 EXPORT_SYMBOL(svc_pool_stats_open);
 

net/sunrpc/svcauth.c

@@ -160,6 +160,22 @@ svc_auth_unregister(rpc_authflavor_t flavor)
 }
 EXPORT_SYMBOL_GPL(svc_auth_unregister);
 
+/**
+ * svc_auth_flavor - return RPC transaction's RPC_AUTH flavor
+ * @rqstp: RPC transaction context
+ *
+ * Returns an RPC flavor or GSS pseudoflavor.
+ */
+rpc_authflavor_t svc_auth_flavor(struct svc_rqst *rqstp)
+{
+	struct auth_ops *aops = rqstp->rq_authop;
+
+	if (!aops->pseudoflavor)
+		return aops->flavour;
+	return aops->pseudoflavor(rqstp);
+}
+EXPORT_SYMBOL_GPL(svc_auth_flavor);
+
 /**************************************************
  * 'auth_domains' are stored in a hash table indexed by name.
  * When the last reference to an 'auth_domain' is dropped,

net/sunrpc/svcsock.c

@@ -1049,18 +1049,14 @@ static int receive_cb_reply(struct svc_sock *svsk, struct svc_rqst *rqstp)
 	struct rpc_rqst *req = NULL;
 	struct kvec *src, *dst;
 	__be32 *p = (__be32 *)rqstp->rq_arg.head[0].iov_base;
-	__be32 xid;
+	__be32 xid = *p;
-	__be32 calldir;
-
-	xid = *p++;
-	calldir = *p;
 
 	if (!bc_xprt)
 		return -EAGAIN;
 	spin_lock(&bc_xprt->queue_lock);
 	req = xprt_lookup_rqst(bc_xprt, xid);
 	if (!req)
-		goto unlock_notfound;
+		goto unlock_eagain;
 
 	memcpy(&req->rq_private_buf, &req->rq_rcv_buf, sizeof(struct xdr_buf));
 	/*
@@ -1077,12 +1073,6 @@ static int receive_cb_reply(struct svc_sock *svsk, struct svc_rqst *rqstp)
 	rqstp->rq_arg.len = 0;
 	spin_unlock(&bc_xprt->queue_lock);
 	return 0;
-unlock_notfound:
-	printk(KERN_NOTICE
-		"%s: Got unrecognized reply: "
-		"calldir 0x%x xpt_bc_xprt %p xid %08x\n",
-		__func__, ntohl(calldir),
-		bc_xprt, ntohl(xid));
 unlock_eagain:
 	spin_unlock(&bc_xprt->queue_lock);
 	return -EAGAIN;

net/sunrpc/xprtrdma/svc_rdma.c

@@ -256,28 +256,44 @@ out_err:
 	return rc;
 }
 
+struct workqueue_struct *svcrdma_wq;
+
 void svc_rdma_cleanup(void)
 {
-	dprintk("SVCRDMA Module Removed, deregister RPC RDMA transport\n");
 	svc_unreg_xprt_class(&svc_rdma_class);
 	svc_rdma_proc_cleanup();
+	if (svcrdma_wq) {
+		struct workqueue_struct *wq = svcrdma_wq;
+
+		svcrdma_wq = NULL;
+		destroy_workqueue(wq);
+	}
+
+	dprintk("SVCRDMA Module Removed, deregister RPC RDMA transport\n");
 }
 
 int svc_rdma_init(void)
 {
+	struct workqueue_struct *wq;
 	int rc;
 
+	wq = alloc_workqueue("svcrdma", WQ_UNBOUND, 0);
+	if (!wq)
+		return -ENOMEM;
+
+	rc = svc_rdma_proc_init();
+	if (rc) {
+		destroy_workqueue(wq);
+		return rc;
+	}
+
+	svcrdma_wq = wq;
+	svc_reg_xprt_class(&svc_rdma_class);
+
 	dprintk("SVCRDMA Module Init, register RPC RDMA transport\n");
 	dprintk("\tsvcrdma_ord      : %d\n", svcrdma_ord);
 	dprintk("\tmax_requests     : %u\n", svcrdma_max_requests);
 	dprintk("\tmax_bc_requests  : %u\n", svcrdma_max_bc_requests);
 	dprintk("\tmax_inline       : %d\n", svcrdma_max_req_size);
-
-	rc = svc_rdma_proc_init();
-	if (rc)
-		return rc;
-
-	/* Register RDMA with the SVC transport switch */
-	svc_reg_xprt_class(&svc_rdma_class);
 	return 0;
 }

net/sunrpc/xprtrdma/svc_rdma_backchannel.c

@@ -76,15 +76,12 @@ static int svc_rdma_bc_sendto(struct svcxprt_rdma *rdma,
 			      struct rpc_rqst *rqst,
 			      struct svc_rdma_send_ctxt *sctxt)
 {
-	struct svc_rdma_recv_ctxt *rctxt;
+	struct svc_rdma_pcl empty_pcl;
 	int ret;
 
-	rctxt = svc_rdma_recv_ctxt_get(rdma);
+	pcl_init(&empty_pcl);
-	if (!rctxt)
+	ret = svc_rdma_map_reply_msg(rdma, sctxt, &empty_pcl, &empty_pcl,
-		return -EIO;
+				     &rqst->rq_snd_buf);
-
-	ret = svc_rdma_map_reply_msg(rdma, sctxt, rctxt, &rqst->rq_snd_buf);
-	svc_rdma_recv_ctxt_put(rdma, rctxt);
 	if (ret < 0)
 		return -EIO;
 

net/sunrpc/xprtrdma/svc_rdma_recvfrom.c

@@ -115,13 +115,6 @@ svc_rdma_next_recv_ctxt(struct list_head *list)
 					rc_list);
 }
 
-static void svc_rdma_recv_cid_init(struct svcxprt_rdma *rdma,
-				   struct rpc_rdma_cid *cid)
-{
-	cid->ci_queue_id = rdma->sc_rq_cq->res.id;
-	cid->ci_completion_id = atomic_inc_return(&rdma->sc_completion_ids);
-}
-
 static struct svc_rdma_recv_ctxt *
 svc_rdma_recv_ctxt_alloc(struct svcxprt_rdma *rdma)
 {
@@ -130,7 +123,7 @@ svc_rdma_recv_ctxt_alloc(struct svcxprt_rdma *rdma)
 	dma_addr_t addr;
 	void *buffer;
 
-	ctxt = kmalloc_node(sizeof(*ctxt), GFP_KERNEL, node);
+	ctxt = kzalloc_node(sizeof(*ctxt), GFP_KERNEL, node);
 	if (!ctxt)
 		goto fail0;
 	buffer = kmalloc_node(rdma->sc_max_req_size, GFP_KERNEL, node);
@@ -156,6 +149,7 @@ svc_rdma_recv_ctxt_alloc(struct svcxprt_rdma *rdma)
 	ctxt->rc_recv_sge.length = rdma->sc_max_req_size;
 	ctxt->rc_recv_sge.lkey = rdma->sc_pd->local_dma_lkey;
 	ctxt->rc_recv_buf = buffer;
+	svc_rdma_cc_init(rdma, &ctxt->rc_cc);
 	return ctxt;
 
 fail2:
@@ -204,18 +198,11 @@ struct svc_rdma_recv_ctxt *svc_rdma_recv_ctxt_get(struct svcxprt_rdma *rdma)
 
 	node = llist_del_first(&rdma->sc_recv_ctxts);
 	if (!node)
-		goto out_empty;
+		return NULL;
-	ctxt = llist_entry(node, struct svc_rdma_recv_ctxt, rc_node);
 
-out:
+	ctxt = llist_entry(node, struct svc_rdma_recv_ctxt, rc_node);
 	ctxt->rc_page_count = 0;
 	return ctxt;
-
-out_empty:
-	ctxt = svc_rdma_recv_ctxt_alloc(rdma);
-	if (!ctxt)
-		return NULL;
-	goto out;
 }
 
 /**
@@ -227,6 +214,13 @@ out_empty:
 void svc_rdma_recv_ctxt_put(struct svcxprt_rdma *rdma,
 			    struct svc_rdma_recv_ctxt *ctxt)
 {
+	svc_rdma_cc_release(rdma, &ctxt->rc_cc, DMA_FROM_DEVICE);
+
+	/* @rc_page_count is normally zero here, but error flows
+	 * can leave pages in @rc_pages.
+	 */
+	release_pages(ctxt->rc_pages, ctxt->rc_page_count);
+
 	pcl_free(&ctxt->rc_call_pcl);
 	pcl_free(&ctxt->rc_read_pcl);
 	pcl_free(&ctxt->rc_write_pcl);
@@ -271,13 +265,13 @@ static bool svc_rdma_refresh_recvs(struct svcxprt_rdma *rdma,
 		if (!ctxt)
 			break;
 
-		trace_svcrdma_post_recv(ctxt);
+		trace_svcrdma_post_recv(&ctxt->rc_cid);
 		ctxt->rc_recv_wr.next = recv_chain;
 		recv_chain = &ctxt->rc_recv_wr;
 		rdma->sc_pending_recvs++;
 	}
 	if (!recv_chain)
-		return false;
+		return true;
 
 	ret = ib_post_recv(rdma->sc_qp, recv_chain, &bad_wr);
 	if (ret)
@@ -301,10 +295,27 @@ err_free:
  * svc_rdma_post_recvs - Post initial set of Recv WRs
  * @rdma: fresh svcxprt_rdma
  *
- * Returns true if successful, otherwise false.
+ * Return values:
+ *   %true: Receive Queue initialization successful
+ *   %false: memory allocation or DMA error
  */
 bool svc_rdma_post_recvs(struct svcxprt_rdma *rdma)
 {
+	unsigned int total;
+
+	/* For each credit, allocate enough recv_ctxts for one
+	 * posted Receive and one RPC in process.
+	 */
+	total = (rdma->sc_max_requests * 2) + rdma->sc_recv_batch;
+	while (total--) {
+		struct svc_rdma_recv_ctxt *ctxt;
+
+		ctxt = svc_rdma_recv_ctxt_alloc(rdma);
+		if (!ctxt)
+			return false;
+		llist_add(&ctxt->rc_node, &rdma->sc_recv_ctxts);
+	}
+
 	return svc_rdma_refresh_recvs(rdma, rdma->sc_max_requests);
 }
 
@@ -373,6 +384,10 @@ void svc_rdma_flush_recv_queues(struct svcxprt_rdma *rdma)
 {
 	struct svc_rdma_recv_ctxt *ctxt;
 
+	while ((ctxt = svc_rdma_next_recv_ctxt(&rdma->sc_read_complete_q))) {
+		list_del(&ctxt->rc_list);
+		svc_rdma_recv_ctxt_put(rdma, ctxt);
+	}
 	while ((ctxt = svc_rdma_next_recv_ctxt(&rdma->sc_rq_dto_q))) {
 		list_del(&ctxt->rc_list);
 		svc_rdma_recv_ctxt_put(rdma, ctxt);
@@ -754,6 +769,122 @@ static bool svc_rdma_is_reverse_direction_reply(struct svc_xprt *xprt,
 	return true;
 }
 
+/* Finish constructing the RPC Call message in rqstp::rq_arg.
+ *
+ * The incoming RPC/RDMA message is an RDMA_MSG type message
+ * with a single Read chunk (only the upper layer data payload
+ * was conveyed via RDMA Read).
+ */
+static void svc_rdma_read_complete_one(struct svc_rqst *rqstp,
+				       struct svc_rdma_recv_ctxt *ctxt)
+{
+	struct svc_rdma_chunk *chunk = pcl_first_chunk(&ctxt->rc_read_pcl);
+	struct xdr_buf *buf = &rqstp->rq_arg;
+	unsigned int length;
+
+	/* Split the Receive buffer between the head and tail
+	 * buffers at Read chunk's position. XDR roundup of the
+	 * chunk is not included in either the pagelist or in
+	 * the tail.
+	 */
+	buf->tail[0].iov_base = buf->head[0].iov_base + chunk->ch_position;
+	buf->tail[0].iov_len = buf->head[0].iov_len - chunk->ch_position;
+	buf->head[0].iov_len = chunk->ch_position;
+
+	/* Read chunk may need XDR roundup (see RFC 8166, s. 3.4.5.2).
+	 *
+	 * If the client already rounded up the chunk length, the
+	 * length does not change. Otherwise, the length of the page
+	 * list is increased to include XDR round-up.
+	 *
+	 * Currently these chunks always start at page offset 0,
+	 * thus the rounded-up length never crosses a page boundary.
+	 */
+	buf->pages = &rqstp->rq_pages[0];
+	length = xdr_align_size(chunk->ch_length);
+	buf->page_len = length;
+	buf->len += length;
+	buf->buflen += length;
+}
+
+/* Finish constructing the RPC Call message in rqstp::rq_arg.
+ *
+ * The incoming RPC/RDMA message is an RDMA_MSG type message
+ * with payload in multiple Read chunks and no PZRC.
+ */
+static void svc_rdma_read_complete_multiple(struct svc_rqst *rqstp,
+					    struct svc_rdma_recv_ctxt *ctxt)
+{
+	struct xdr_buf *buf = &rqstp->rq_arg;
+
+	buf->len += ctxt->rc_readbytes;
+	buf->buflen += ctxt->rc_readbytes;
+
+	buf->head[0].iov_base = page_address(rqstp->rq_pages[0]);
+	buf->head[0].iov_len = min_t(size_t, PAGE_SIZE, ctxt->rc_readbytes);
+	buf->pages = &rqstp->rq_pages[1];
+	buf->page_len = ctxt->rc_readbytes - buf->head[0].iov_len;
+}
+
+/* Finish constructing the RPC Call message in rqstp::rq_arg.
+ *
+ * The incoming RPC/RDMA message is an RDMA_NOMSG type message
+ * (the RPC message body was conveyed via RDMA Read).
+ */
+static void svc_rdma_read_complete_pzrc(struct svc_rqst *rqstp,
+					struct svc_rdma_recv_ctxt *ctxt)
+{
+	struct xdr_buf *buf = &rqstp->rq_arg;
+
+	buf->len += ctxt->rc_readbytes;
+	buf->buflen += ctxt->rc_readbytes;
+
+	buf->head[0].iov_base = page_address(rqstp->rq_pages[0]);
+	buf->head[0].iov_len = min_t(size_t, PAGE_SIZE, ctxt->rc_readbytes);
+	buf->pages = &rqstp->rq_pages[1];
+	buf->page_len = ctxt->rc_readbytes - buf->head[0].iov_len;
+}
+
+static noinline void svc_rdma_read_complete(struct svc_rqst *rqstp,
+					    struct svc_rdma_recv_ctxt *ctxt)
+{
+	unsigned int i;
+
+	/* Transfer the Read chunk pages into @rqstp.rq_pages, replacing
+	 * the rq_pages that were already allocated for this rqstp.
+	 */
+	release_pages(rqstp->rq_respages, ctxt->rc_page_count);
+	for (i = 0; i < ctxt->rc_page_count; i++)
+		rqstp->rq_pages[i] = ctxt->rc_pages[i];
+
+	/* Update @rqstp's result send buffer to start after the
+	 * last page in the RDMA Read payload.
+	 */
+	rqstp->rq_respages = &rqstp->rq_pages[ctxt->rc_page_count];
+	rqstp->rq_next_page = rqstp->rq_respages + 1;
+
+	/* Prevent svc_rdma_recv_ctxt_put() from releasing the
+	 * pages in ctxt::rc_pages a second time.
+	 */
+	ctxt->rc_page_count = 0;
+
+	/* Finish constructing the RPC Call message. The exact
+	 * procedure for that depends on what kind of RPC/RDMA
+	 * chunks were provided by the client.
+	 */
+	rqstp->rq_arg = ctxt->rc_saved_arg;
+	if (pcl_is_empty(&ctxt->rc_call_pcl)) {
+		if (ctxt->rc_read_pcl.cl_count == 1)
+			svc_rdma_read_complete_one(rqstp, ctxt);
+		else
+			svc_rdma_read_complete_multiple(rqstp, ctxt);
+	} else {
+		svc_rdma_read_complete_pzrc(rqstp, ctxt);
+	}
+
+	trace_svcrdma_read_finished(&ctxt->rc_cid);
+}
+
 /**
  * svc_rdma_recvfrom - Receive an RPC call
  * @rqstp: request structure into which to receive an RPC Call
@@ -798,8 +929,15 @@ int svc_rdma_recvfrom(struct svc_rqst *rqstp)
 
 	rqstp->rq_xprt_ctxt = NULL;
 
-	ctxt = NULL;
 	spin_lock(&rdma_xprt->sc_rq_dto_lock);
+	ctxt = svc_rdma_next_recv_ctxt(&rdma_xprt->sc_read_complete_q);
+	if (ctxt) {
+		list_del(&ctxt->rc_list);
+		spin_unlock(&rdma_xprt->sc_rq_dto_lock);
+		svc_xprt_received(xprt);
+		svc_rdma_read_complete(rqstp, ctxt);
+		goto complete;
+	}
 	ctxt = svc_rdma_next_recv_ctxt(&rdma_xprt->sc_rq_dto_q);
 	if (ctxt)
 		list_del(&ctxt->rc_list);
@@ -831,12 +969,10 @@ int svc_rdma_recvfrom(struct svc_rqst *rqstp)
 	svc_rdma_get_inv_rkey(rdma_xprt, ctxt);
 
 	if (!pcl_is_empty(&ctxt->rc_read_pcl) ||
-	    !pcl_is_empty(&ctxt->rc_call_pcl)) {
+	    !pcl_is_empty(&ctxt->rc_call_pcl))
-		ret = svc_rdma_process_read_list(rdma_xprt, rqstp, ctxt);
+		goto out_readlist;
-		if (ret < 0)
-			goto out_readfail;
-	}
 
+complete:
 	rqstp->rq_xprt_ctxt = ctxt;
 	rqstp->rq_prot = IPPROTO_MAX;
 	svc_xprt_copy_addrs(rqstp, xprt);
@@ -848,12 +984,23 @@ out_err:
 	svc_rdma_recv_ctxt_put(rdma_xprt, ctxt);
 	return 0;
 
-out_readfail:
+out_readlist:
-	if (ret == -EINVAL)
+	/* This @rqstp is about to be recycled. Save the work
-		svc_rdma_send_error(rdma_xprt, ctxt, ret);
+	 * already done constructing the Call message in rq_arg
-	svc_rdma_recv_ctxt_put(rdma_xprt, ctxt);
+	 * so it can be restored when the RDMA Reads have
-	svc_xprt_deferred_close(xprt);
+	 * completed.
-	return -ENOTCONN;
+	 */
+	ctxt->rc_saved_arg = rqstp->rq_arg;
+
+	ret = svc_rdma_process_read_list(rdma_xprt, rqstp, ctxt);
+	if (ret < 0) {
+		if (ret == -EINVAL)
+			svc_rdma_send_error(rdma_xprt, ctxt, ret);
+		svc_rdma_recv_ctxt_put(rdma_xprt, ctxt);
+		svc_xprt_deferred_close(xprt);
+		return ret;
+	}
+	return 0;
 
 out_backchannel:
 	svc_rdma_handle_bc_reply(rqstp, ctxt);

net/sunrpc/xprtrdma/svc_rdma_rw.c

@@ -39,6 +39,7 @@ struct svc_rdma_rw_ctxt {
 	struct list_head	rw_list;
 	struct rdma_rw_ctx	rw_ctx;
 	unsigned int		rw_nents;
+	unsigned int		rw_first_sgl_nents;
 	struct sg_table		rw_sg_table;
 	struct scatterlist	rw_first_sgl[];
 };
@@ -53,6 +54,8 @@ svc_rdma_next_ctxt(struct list_head *list)
 static struct svc_rdma_rw_ctxt *
 svc_rdma_get_rw_ctxt(struct svcxprt_rdma *rdma, unsigned int sges)
 {
+	struct ib_device *dev = rdma->sc_cm_id->device;
+	unsigned int first_sgl_nents = dev->attrs.max_send_sge;
 	struct svc_rdma_rw_ctxt *ctxt;
 	struct llist_node *node;
 
@@ -62,32 +65,33 @@ svc_rdma_get_rw_ctxt(struct svcxprt_rdma *rdma, unsigned int sges)
 	if (node) {
 		ctxt = llist_entry(node, struct svc_rdma_rw_ctxt, rw_node);
 	} else {
-		ctxt = kmalloc_node(struct_size(ctxt, rw_first_sgl, SG_CHUNK_SIZE),
+		ctxt = kmalloc_node(struct_size(ctxt, rw_first_sgl, first_sgl_nents),
-				    GFP_KERNEL, ibdev_to_node(rdma->sc_cm_id->device));
+				    GFP_KERNEL, ibdev_to_node(dev));
 		if (!ctxt)
 			goto out_noctx;
 
 		INIT_LIST_HEAD(&ctxt->rw_list);
+		ctxt->rw_first_sgl_nents = first_sgl_nents;
 	}
 
 	ctxt->rw_sg_table.sgl = ctxt->rw_first_sgl;
 	if (sg_alloc_table_chained(&ctxt->rw_sg_table, sges,
 				   ctxt->rw_sg_table.sgl,
-				   SG_CHUNK_SIZE))
+				   first_sgl_nents))
 		goto out_free;
 	return ctxt;
 
 out_free:
 	kfree(ctxt);
 out_noctx:
-	trace_svcrdma_no_rwctx_err(rdma, sges);
+	trace_svcrdma_rwctx_empty(rdma, sges);
 	return NULL;
 }
 
 static void __svc_rdma_put_rw_ctxt(struct svc_rdma_rw_ctxt *ctxt,
 				   struct llist_head *list)
 {
-	sg_free_table_chained(&ctxt->rw_sg_table, SG_CHUNK_SIZE);
+	sg_free_table_chained(&ctxt->rw_sg_table, ctxt->rw_first_sgl_nents);
 	llist_add(&ctxt->rw_node, list);
 }
 
@@ -135,57 +139,40 @@ static int svc_rdma_rw_ctx_init(struct svcxprt_rdma *rdma,
 			       ctxt->rw_sg_table.sgl, ctxt->rw_nents,
 			       0, offset, handle, direction);
 	if (unlikely(ret < 0)) {
+		trace_svcrdma_dma_map_rw_err(rdma, offset, handle,
+					     ctxt->rw_nents, ret);
 		svc_rdma_put_rw_ctxt(rdma, ctxt);
-		trace_svcrdma_dma_map_rw_err(rdma, ctxt->rw_nents, ret);
 	}
 	return ret;
 }
 
-/* A chunk context tracks all I/O for moving one Read or Write
+/**
- * chunk. This is a set of rdma_rw's that handle data movement
+ * svc_rdma_cc_init - Initialize an svc_rdma_chunk_ctxt
- * for all segments of one chunk.
+ * @rdma: controlling transport instance
- *
+ * @cc: svc_rdma_chunk_ctxt to be initialized
- * These are small, acquired with a single allocator call, and
- * no more than one is needed per chunk. They are allocated on
- * demand, and not cached.
  */
-struct svc_rdma_chunk_ctxt {
+void svc_rdma_cc_init(struct svcxprt_rdma *rdma,
-	struct rpc_rdma_cid	cc_cid;
+		      struct svc_rdma_chunk_ctxt *cc)
-	struct ib_cqe		cc_cqe;
-	struct svcxprt_rdma	*cc_rdma;
-	struct list_head	cc_rwctxts;
-	ktime_t			cc_posttime;
-	int			cc_sqecount;
-	enum ib_wc_status	cc_status;
-	struct completion	cc_done;
-};
-
-static void svc_rdma_cc_cid_init(struct svcxprt_rdma *rdma,
-				 struct rpc_rdma_cid *cid)
 {
-	cid->ci_queue_id = rdma->sc_sq_cq->res.id;
+	struct rpc_rdma_cid *cid = &cc->cc_cid;
-	cid->ci_completion_id = atomic_inc_return(&rdma->sc_completion_ids);
-}
 
-static void svc_rdma_cc_init(struct svcxprt_rdma *rdma,
+	if (unlikely(!cid->ci_completion_id))
-			     struct svc_rdma_chunk_ctxt *cc)
+		svc_rdma_send_cid_init(rdma, cid);
-{
-	svc_rdma_cc_cid_init(rdma, &cc->cc_cid);
-	cc->cc_rdma = rdma;
 
 	INIT_LIST_HEAD(&cc->cc_rwctxts);
 	cc->cc_sqecount = 0;
 }
 
-/*
+/**
- * The consumed rw_ctx's are cleaned and placed on a local llist so
+ * svc_rdma_cc_release - Release resources held by a svc_rdma_chunk_ctxt
- * that only one atomic llist operation is needed to put them all
+ * @rdma: controlling transport instance
- * back on the free list.
+ * @cc: svc_rdma_chunk_ctxt to be released
+ * @dir: DMA direction
  */
-static void svc_rdma_cc_release(struct svc_rdma_chunk_ctxt *cc,
+void svc_rdma_cc_release(struct svcxprt_rdma *rdma,
-				enum dma_data_direction dir)
+			 struct svc_rdma_chunk_ctxt *cc,
+			 enum dma_data_direction dir)
 {
-	struct svcxprt_rdma *rdma = cc->cc_rdma;
 	struct llist_node *first, *last;
 	struct svc_rdma_rw_ctxt *ctxt;
 	LLIST_HEAD(free);
@@ -215,6 +202,8 @@ static void svc_rdma_cc_release(struct svc_rdma_chunk_ctxt *cc,
  *  - Stores arguments for the SGL constructor functions
  */
 struct svc_rdma_write_info {
+	struct svcxprt_rdma	*wi_rdma;
+
 	const struct svc_rdma_chunk	*wi_chunk;
 
 	/* write state of this chunk */
@@ -227,6 +216,7 @@ struct svc_rdma_write_info {
 	unsigned int		wi_next_off;
 
 	struct svc_rdma_chunk_ctxt	wi_cc;
+	struct work_struct	wi_work;
 };
 
 static struct svc_rdma_write_info *
@@ -235,23 +225,31 @@ svc_rdma_write_info_alloc(struct svcxprt_rdma *rdma,
 {
 	struct svc_rdma_write_info *info;
 
-	info = kmalloc_node(sizeof(*info), GFP_KERNEL,
+	info = kzalloc_node(sizeof(*info), GFP_KERNEL,
 			    ibdev_to_node(rdma->sc_cm_id->device));
 	if (!info)
 		return info;
 
+	info->wi_rdma = rdma;
 	info->wi_chunk = chunk;
-	info->wi_seg_off = 0;
-	info->wi_seg_no = 0;
 	svc_rdma_cc_init(rdma, &info->wi_cc);
 	info->wi_cc.cc_cqe.done = svc_rdma_write_done;
 	return info;
 }
 
+static void svc_rdma_write_info_free_async(struct work_struct *work)
+{
+	struct svc_rdma_write_info *info;
+
+	info = container_of(work, struct svc_rdma_write_info, wi_work);
+	svc_rdma_cc_release(info->wi_rdma, &info->wi_cc, DMA_TO_DEVICE);
+	kfree(info);
+}
+
 static void svc_rdma_write_info_free(struct svc_rdma_write_info *info)
 {
-	svc_rdma_cc_release(&info->wi_cc, DMA_TO_DEVICE);
+	INIT_WORK(&info->wi_work, svc_rdma_write_info_free_async);
-	kfree(info);
+	queue_work(svcrdma_wq, &info->wi_work);
 }
 
 /**
@@ -263,16 +261,16 @@ static void svc_rdma_write_info_free(struct svc_rdma_write_info *info)
  */
 static void svc_rdma_write_done(struct ib_cq *cq, struct ib_wc *wc)
 {
+	struct svcxprt_rdma *rdma = cq->cq_context;
 	struct ib_cqe *cqe = wc->wr_cqe;
 	struct svc_rdma_chunk_ctxt *cc =
 			container_of(cqe, struct svc_rdma_chunk_ctxt, cc_cqe);
-	struct svcxprt_rdma *rdma = cc->cc_rdma;
 	struct svc_rdma_write_info *info =
 			container_of(cc, struct svc_rdma_write_info, wi_cc);
 
 	switch (wc->status) {
 	case IB_WC_SUCCESS:
-		trace_svcrdma_wc_write(wc, &cc->cc_cid);
+		trace_svcrdma_wc_write(&cc->cc_cid);
 		break;
 	case IB_WC_WR_FLUSH_ERR:
 		trace_svcrdma_wc_write_flush(wc, &cc->cc_cid);
@@ -289,39 +287,6 @@ static void svc_rdma_write_done(struct ib_cq *cq, struct ib_wc *wc)
 	svc_rdma_write_info_free(info);
 }
 
-/* State for pulling a Read chunk.
- */
-struct svc_rdma_read_info {
-	struct svc_rqst			*ri_rqst;
-	struct svc_rdma_recv_ctxt	*ri_readctxt;
-	unsigned int			ri_pageno;
-	unsigned int			ri_pageoff;
-	unsigned int			ri_totalbytes;
-
-	struct svc_rdma_chunk_ctxt	ri_cc;
-};
-
-static struct svc_rdma_read_info *
-svc_rdma_read_info_alloc(struct svcxprt_rdma *rdma)
-{
-	struct svc_rdma_read_info *info;
-
-	info = kmalloc_node(sizeof(*info), GFP_KERNEL,
-			    ibdev_to_node(rdma->sc_cm_id->device));
-	if (!info)
-		return info;
-
-	svc_rdma_cc_init(rdma, &info->ri_cc);
-	info->ri_cc.cc_cqe.done = svc_rdma_wc_read_done;
-	return info;
-}
-
-static void svc_rdma_read_info_free(struct svc_rdma_read_info *info)
-{
-	svc_rdma_cc_release(&info->ri_cc, DMA_FROM_DEVICE);
-	kfree(info);
-}
-
 /**
  * svc_rdma_wc_read_done - Handle completion of an RDMA Read ctx
  * @cq: controlling Completion Queue
@@ -330,17 +295,27 @@ static void svc_rdma_read_info_free(struct svc_rdma_read_info *info)
  */
 static void svc_rdma_wc_read_done(struct ib_cq *cq, struct ib_wc *wc)
 {
+	struct svcxprt_rdma *rdma = cq->cq_context;
 	struct ib_cqe *cqe = wc->wr_cqe;
 	struct svc_rdma_chunk_ctxt *cc =
 			container_of(cqe, struct svc_rdma_chunk_ctxt, cc_cqe);
-	struct svc_rdma_read_info *info;
+	struct svc_rdma_recv_ctxt *ctxt;
 
+	svc_rdma_wake_send_waiters(rdma, cc->cc_sqecount);
+
+	ctxt = container_of(cc, struct svc_rdma_recv_ctxt, rc_cc);
 	switch (wc->status) {
 	case IB_WC_SUCCESS:
-		info = container_of(cc, struct svc_rdma_read_info, ri_cc);
+		trace_svcrdma_wc_read(wc, &cc->cc_cid, ctxt->rc_readbytes,
-		trace_svcrdma_wc_read(wc, &cc->cc_cid, info->ri_totalbytes,
 				      cc->cc_posttime);
-		break;
+
+		spin_lock(&rdma->sc_rq_dto_lock);
+		list_add_tail(&ctxt->rc_list, &rdma->sc_read_complete_q);
+		/* the unlock pairs with the smp_rmb in svc_xprt_ready */
+		set_bit(XPT_DATA, &rdma->sc_xprt.xpt_flags);
+		spin_unlock(&rdma->sc_rq_dto_lock);
+		svc_xprt_enqueue(&rdma->sc_xprt);
+		return;
 	case IB_WC_WR_FLUSH_ERR:
 		trace_svcrdma_wc_read_flush(wc, &cc->cc_cid);
 		break;
@@ -348,10 +323,13 @@ static void svc_rdma_wc_read_done(struct ib_cq *cq, struct ib_wc *wc)
 		trace_svcrdma_wc_read_err(wc, &cc->cc_cid);
 	}
 
-	svc_rdma_wake_send_waiters(cc->cc_rdma, cc->cc_sqecount);
+	/* The RDMA Read has flushed, so the incoming RPC message
-	cc->cc_status = wc->status;
+	 * cannot be constructed and must be dropped. Signal the
-	complete(&cc->cc_done);
+	 * loss to the client by closing the connection.
-	return;
+	 */
+	svc_rdma_cc_release(rdma, cc, DMA_FROM_DEVICE);
+	svc_rdma_recv_ctxt_put(rdma, ctxt);
+	svc_xprt_deferred_close(&rdma->sc_xprt);
 }
 
 /*
@@ -360,9 +338,9 @@ static void svc_rdma_wc_read_done(struct ib_cq *cq, struct ib_wc *wc)
  *   even if one or more WRs are flushed. This is true when posting
  *   an rdma_rw_ctx or when posting a single signaled WR.
  */
-static int svc_rdma_post_chunk_ctxt(struct svc_rdma_chunk_ctxt *cc)
+static int svc_rdma_post_chunk_ctxt(struct svcxprt_rdma *rdma,
+				    struct svc_rdma_chunk_ctxt *cc)
 {
-	struct svcxprt_rdma *rdma = cc->cc_rdma;
 	struct ib_send_wr *first_wr;
 	const struct ib_send_wr *bad_wr;
 	struct list_head *tmp;
@@ -396,14 +374,14 @@ static int svc_rdma_post_chunk_ctxt(struct svc_rdma_chunk_ctxt *cc)
 		}
 
 		percpu_counter_inc(&svcrdma_stat_sq_starve);
-		trace_svcrdma_sq_full(rdma);
+		trace_svcrdma_sq_full(rdma, &cc->cc_cid);
 		atomic_add(cc->cc_sqecount, &rdma->sc_sq_avail);
 		wait_event(rdma->sc_send_wait,
 			   atomic_read(&rdma->sc_sq_avail) > cc->cc_sqecount);
-		trace_svcrdma_sq_retry(rdma);
+		trace_svcrdma_sq_retry(rdma, &cc->cc_cid);
 	} while (1);
 
-	trace_svcrdma_sq_post_err(rdma, ret);
+	trace_svcrdma_sq_post_err(rdma, &cc->cc_cid, ret);
 	svc_xprt_deferred_close(&rdma->sc_xprt);
 
 	/* If even one was posted, there will be a completion. */
@@ -473,7 +451,7 @@ svc_rdma_build_writes(struct svc_rdma_write_info *info,
 		      unsigned int remaining)
 {
 	struct svc_rdma_chunk_ctxt *cc = &info->wi_cc;
-	struct svcxprt_rdma *rdma = cc->cc_rdma;
+	struct svcxprt_rdma *rdma = info->wi_rdma;
 	const struct svc_rdma_segment *seg;
 	struct svc_rdma_rw_ctxt *ctxt;
 	int ret;
@@ -516,7 +494,7 @@ svc_rdma_build_writes(struct svc_rdma_write_info *info,
 	return 0;
 
 out_overflow:
-	trace_svcrdma_small_wrch_err(rdma, remaining, info->wi_seg_no,
+	trace_svcrdma_small_wrch_err(&cc->cc_cid, remaining, info->wi_seg_no,
 				     info->wi_chunk->ch_segcount);
 	return -E2BIG;
 }
@@ -633,7 +611,7 @@ int svc_rdma_send_write_chunk(struct svcxprt_rdma *rdma,
 		goto out_err;
 
 	trace_svcrdma_post_write_chunk(&cc->cc_cid, cc->cc_sqecount);
-	ret = svc_rdma_post_chunk_ctxt(cc);
+	ret = svc_rdma_post_chunk_ctxt(rdma, cc);
 	if (ret < 0)
 		goto out_err;
 	return xdr->len;
@@ -680,7 +658,7 @@ int svc_rdma_send_reply_chunk(struct svcxprt_rdma *rdma,
 		goto out_err;
 
 	trace_svcrdma_post_reply_chunk(&cc->cc_cid, cc->cc_sqecount);
-	ret = svc_rdma_post_chunk_ctxt(cc);
+	ret = svc_rdma_post_chunk_ctxt(rdma, cc);
 	if (ret < 0)
 		goto out_err;
 
@@ -693,7 +671,8 @@ out_err:
 
 /**
  * svc_rdma_build_read_segment - Build RDMA Read WQEs to pull one RDMA segment
- * @info: context for ongoing I/O
+ * @rqstp: RPC transaction context
+ * @head: context for ongoing I/O
  * @segment: co-ordinates of remote memory to be read
  *
  * Returns:
@@ -702,20 +681,20 @@ out_err:
  *   %-ENOMEM: allocating a local resources failed
  *   %-EIO: a DMA mapping error occurred
  */
-static int svc_rdma_build_read_segment(struct svc_rdma_read_info *info,
+static int svc_rdma_build_read_segment(struct svc_rqst *rqstp,
+				       struct svc_rdma_recv_ctxt *head,
 				       const struct svc_rdma_segment *segment)
 {
-	struct svc_rdma_recv_ctxt *head = info->ri_readctxt;
+	struct svcxprt_rdma *rdma = svc_rdma_rqst_rdma(rqstp);
-	struct svc_rdma_chunk_ctxt *cc = &info->ri_cc;
+	struct svc_rdma_chunk_ctxt *cc = &head->rc_cc;
-	struct svc_rqst *rqstp = info->ri_rqst;
 	unsigned int sge_no, seg_len, len;
 	struct svc_rdma_rw_ctxt *ctxt;
 	struct scatterlist *sg;
 	int ret;
 
 	len = segment->rs_length;
-	sge_no = PAGE_ALIGN(info->ri_pageoff + len) >> PAGE_SHIFT;
+	sge_no = PAGE_ALIGN(head->rc_pageoff + len) >> PAGE_SHIFT;
-	ctxt = svc_rdma_get_rw_ctxt(cc->cc_rdma, sge_no);
+	ctxt = svc_rdma_get_rw_ctxt(rdma, sge_no);
 	if (!ctxt)
 		return -ENOMEM;
 	ctxt->rw_nents = sge_no;
@@ -723,29 +702,27 @@ static int svc_rdma_build_read_segment(struct svc_rdma_read_info *info,
 	sg = ctxt->rw_sg_table.sgl;
 	for (sge_no = 0; sge_no < ctxt->rw_nents; sge_no++) {
 		seg_len = min_t(unsigned int, len,
-				PAGE_SIZE - info->ri_pageoff);
+				PAGE_SIZE - head->rc_pageoff);
 
-		if (!info->ri_pageoff)
+		if (!head->rc_pageoff)
 			head->rc_page_count++;
 
-		sg_set_page(sg, rqstp->rq_pages[info->ri_pageno],
+		sg_set_page(sg, rqstp->rq_pages[head->rc_curpage],
-			    seg_len, info->ri_pageoff);
+			    seg_len, head->rc_pageoff);
 		sg = sg_next(sg);
 
-		info->ri_pageoff += seg_len;
+		head->rc_pageoff += seg_len;
-		if (info->ri_pageoff == PAGE_SIZE) {
+		if (head->rc_pageoff == PAGE_SIZE) {
-			info->ri_pageno++;
+			head->rc_curpage++;
-			info->ri_pageoff = 0;
+			head->rc_pageoff = 0;
 		}
 		len -= seg_len;
 
-		/* Safety check */
+		if (len && ((head->rc_curpage + 1) > ARRAY_SIZE(rqstp->rq_pages)))
-		if (len &&
-		    &rqstp->rq_pages[info->ri_pageno + 1] > rqstp->rq_page_end)
 			goto out_overrun;
 	}
 
-	ret = svc_rdma_rw_ctx_init(cc->cc_rdma, ctxt, segment->rs_offset,
+	ret = svc_rdma_rw_ctx_init(rdma, ctxt, segment->rs_offset,
 				   segment->rs_handle, DMA_FROM_DEVICE);
 	if (ret < 0)
 		return -EIO;
@@ -756,13 +733,14 @@ static int svc_rdma_build_read_segment(struct svc_rdma_read_info *info,
 	return 0;
 
 out_overrun:
-	trace_svcrdma_page_overrun_err(cc->cc_rdma, rqstp, info->ri_pageno);
+	trace_svcrdma_page_overrun_err(&cc->cc_cid, head->rc_curpage);
 	return -EINVAL;
 }
 
 /**
  * svc_rdma_build_read_chunk - Build RDMA Read WQEs to pull one RDMA chunk
- * @info: context for ongoing I/O
+ * @rqstp: RPC transaction context
+ * @head: context for ongoing I/O
  * @chunk: Read chunk to pull
  *
  * Return values:
@@ -771,7 +749,8 @@ out_overrun:
  *   %-ENOMEM: allocating a local resources failed
  *   %-EIO: a DMA mapping error occurred
  */
-static int svc_rdma_build_read_chunk(struct svc_rdma_read_info *info,
+static int svc_rdma_build_read_chunk(struct svc_rqst *rqstp,
+				     struct svc_rdma_recv_ctxt *head,
 				     const struct svc_rdma_chunk *chunk)
 {
 	const struct svc_rdma_segment *segment;
@@ -779,56 +758,56 @@ static int svc_rdma_build_read_chunk(struct svc_rdma_read_info *info,
 
 	ret = -EINVAL;
 	pcl_for_each_segment(segment, chunk) {
-		ret = svc_rdma_build_read_segment(info, segment);
+		ret = svc_rdma_build_read_segment(rqstp, head, segment);
 		if (ret < 0)
 			break;
-		info->ri_totalbytes += segment->rs_length;
+		head->rc_readbytes += segment->rs_length;
 	}
 	return ret;
 }
 
 /**
  * svc_rdma_copy_inline_range - Copy part of the inline content into pages
- * @info: context for RDMA Reads
+ * @rqstp: RPC transaction context
+ * @head: context for ongoing I/O
  * @offset: offset into the Receive buffer of region to copy
  * @remaining: length of region to copy
  *
  * Take a page at a time from rqstp->rq_pages and copy the inline
  * content from the Receive buffer into that page. Update
- * info->ri_pageno and info->ri_pageoff so that the next RDMA Read
+ * head->rc_curpage and head->rc_pageoff so that the next RDMA Read
  * result will land contiguously with the copied content.
  *
  * Return values:
  *   %0: Inline content was successfully copied
  *   %-EINVAL: offset or length was incorrect
  */
-static int svc_rdma_copy_inline_range(struct svc_rdma_read_info *info,
+static int svc_rdma_copy_inline_range(struct svc_rqst *rqstp,
+				      struct svc_rdma_recv_ctxt *head,
 				      unsigned int offset,
 				      unsigned int remaining)
 {
-	struct svc_rdma_recv_ctxt *head = info->ri_readctxt;
 	unsigned char *dst, *src = head->rc_recv_buf;
-	struct svc_rqst *rqstp = info->ri_rqst;
 	unsigned int page_no, numpages;
 
-	numpages = PAGE_ALIGN(info->ri_pageoff + remaining) >> PAGE_SHIFT;
+	numpages = PAGE_ALIGN(head->rc_pageoff + remaining) >> PAGE_SHIFT;
 	for (page_no = 0; page_no < numpages; page_no++) {
 		unsigned int page_len;
 
 		page_len = min_t(unsigned int, remaining,
-				 PAGE_SIZE - info->ri_pageoff);
+				 PAGE_SIZE - head->rc_pageoff);
 
-		if (!info->ri_pageoff)
+		if (!head->rc_pageoff)
 			head->rc_page_count++;
 
-		dst = page_address(rqstp->rq_pages[info->ri_pageno]);
+		dst = page_address(rqstp->rq_pages[head->rc_curpage]);
-		memcpy(dst + info->ri_pageno, src + offset, page_len);
+		memcpy(dst + head->rc_curpage, src + offset, page_len);
 
-		info->ri_totalbytes += page_len;
+		head->rc_readbytes += page_len;
-		info->ri_pageoff += page_len;
+		head->rc_pageoff += page_len;
-		if (info->ri_pageoff == PAGE_SIZE) {
+		if (head->rc_pageoff == PAGE_SIZE) {
-			info->ri_pageno++;
+			head->rc_curpage++;
-			info->ri_pageoff = 0;
+			head->rc_pageoff = 0;
 		}
 		remaining -= page_len;
 		offset += page_len;
@@ -839,7 +818,8 @@ static int svc_rdma_copy_inline_range(struct svc_rdma_read_info *info,
 
 /**
  * svc_rdma_read_multiple_chunks - Construct RDMA Reads to pull data item Read chunks
- * @info: context for RDMA Reads
+ * @rqstp: RPC transaction context
+ * @head: context for ongoing I/O
  *
  * The chunk data lands in rqstp->rq_arg as a series of contiguous pages,
  * like an incoming TCP call.
@@ -851,11 +831,11 @@ static int svc_rdma_copy_inline_range(struct svc_rdma_read_info *info,
  *   %-ENOTCONN: posting failed (connection is lost),
  *   %-EIO: rdma_rw initialization failed (DMA mapping, etc).
  */
-static noinline int svc_rdma_read_multiple_chunks(struct svc_rdma_read_info *info)
+static noinline int
+svc_rdma_read_multiple_chunks(struct svc_rqst *rqstp,
+			      struct svc_rdma_recv_ctxt *head)
 {
-	struct svc_rdma_recv_ctxt *head = info->ri_readctxt;
 	const struct svc_rdma_pcl *pcl = &head->rc_read_pcl;
-	struct xdr_buf *buf = &info->ri_rqst->rq_arg;
 	struct svc_rdma_chunk *chunk, *next;
 	unsigned int start, length;
 	int ret;
@@ -863,12 +843,12 @@ static noinline int svc_rdma_read_multiple_chunks(struct svc_rdma_read_info *inf
 	start = 0;
 	chunk = pcl_first_chunk(pcl);
 	length = chunk->ch_position;
-	ret = svc_rdma_copy_inline_range(info, start, length);
+	ret = svc_rdma_copy_inline_range(rqstp, head, start, length);
 	if (ret < 0)
 		return ret;
 
 	pcl_for_each_chunk(chunk, pcl) {
-		ret = svc_rdma_build_read_chunk(info, chunk);
+		ret = svc_rdma_build_read_chunk(rqstp, head, chunk);
 		if (ret < 0)
 			return ret;
 
@@ -877,31 +857,21 @@ static noinline int svc_rdma_read_multiple_chunks(struct svc_rdma_read_info *inf
 			break;
 
 		start += length;
-		length = next->ch_position - info->ri_totalbytes;
+		length = next->ch_position - head->rc_readbytes;
-		ret = svc_rdma_copy_inline_range(info, start, length);
+		ret = svc_rdma_copy_inline_range(rqstp, head, start, length);
 		if (ret < 0)
 			return ret;
 	}
 
 	start += length;
 	length = head->rc_byte_len - start;
-	ret = svc_rdma_copy_inline_range(info, start, length);
+	return svc_rdma_copy_inline_range(rqstp, head, start, length);
-	if (ret < 0)
-		return ret;
-
-	buf->len += info->ri_totalbytes;
-	buf->buflen += info->ri_totalbytes;
-
-	buf->head[0].iov_base = page_address(info->ri_rqst->rq_pages[0]);
-	buf->head[0].iov_len = min_t(size_t, PAGE_SIZE, info->ri_totalbytes);
-	buf->pages = &info->ri_rqst->rq_pages[1];
-	buf->page_len = info->ri_totalbytes - buf->head[0].iov_len;
-	return 0;
 }
 
 /**
  * svc_rdma_read_data_item - Construct RDMA Reads to pull data item Read chunks
- * @info: context for RDMA Reads
+ * @rqstp: RPC transaction context
+ * @head: context for ongoing I/O
  *
  * The chunk data lands in the page list of rqstp->rq_arg.pages.
  *
@@ -916,50 +886,17 @@ static noinline int svc_rdma_read_multiple_chunks(struct svc_rdma_read_info *inf
  *   %-ENOTCONN: posting failed (connection is lost),
  *   %-EIO: rdma_rw initialization failed (DMA mapping, etc).
  */
-static int svc_rdma_read_data_item(struct svc_rdma_read_info *info)
+static int svc_rdma_read_data_item(struct svc_rqst *rqstp,
+				   struct svc_rdma_recv_ctxt *head)
 {
-	struct svc_rdma_recv_ctxt *head = info->ri_readctxt;
+	return svc_rdma_build_read_chunk(rqstp, head,
-	struct xdr_buf *buf = &info->ri_rqst->rq_arg;
+					 pcl_first_chunk(&head->rc_read_pcl));
-	struct svc_rdma_chunk *chunk;
-	unsigned int length;
-	int ret;
-
-	chunk = pcl_first_chunk(&head->rc_read_pcl);
-	ret = svc_rdma_build_read_chunk(info, chunk);
-	if (ret < 0)
-		goto out;
-
-	/* Split the Receive buffer between the head and tail
-	 * buffers at Read chunk's position. XDR roundup of the
-	 * chunk is not included in either the pagelist or in
-	 * the tail.
-	 */
-	buf->tail[0].iov_base = buf->head[0].iov_base + chunk->ch_position;
-	buf->tail[0].iov_len = buf->head[0].iov_len - chunk->ch_position;
-	buf->head[0].iov_len = chunk->ch_position;
-
-	/* Read chunk may need XDR roundup (see RFC 8166, s. 3.4.5.2).
-	 *
-	 * If the client already rounded up the chunk length, the
-	 * length does not change. Otherwise, the length of the page
-	 * list is increased to include XDR round-up.
-	 *
-	 * Currently these chunks always start at page offset 0,
-	 * thus the rounded-up length never crosses a page boundary.
-	 */
-	buf->pages = &info->ri_rqst->rq_pages[0];
-	length = xdr_align_size(chunk->ch_length);
-	buf->page_len = length;
-	buf->len += length;
-	buf->buflen += length;
-
-out:
-	return ret;
 }
 
 /**
- * svc_rdma_read_chunk_range - Build RDMA Read WQEs for portion of a chunk
+ * svc_rdma_read_chunk_range - Build RDMA Read WRs for portion of a chunk
- * @info: context for RDMA Reads
+ * @rqstp: RPC transaction context
+ * @head: context for ongoing I/O
  * @chunk: parsed Call chunk to pull
  * @offset: offset of region to pull
  * @length: length of region to pull
@@ -971,7 +908,8 @@ out:
  *   %-ENOTCONN: posting failed (connection is lost),
  *   %-EIO: rdma_rw initialization failed (DMA mapping, etc).
  */
-static int svc_rdma_read_chunk_range(struct svc_rdma_read_info *info,
+static int svc_rdma_read_chunk_range(struct svc_rqst *rqstp,
+				     struct svc_rdma_recv_ctxt *head,
 				     const struct svc_rdma_chunk *chunk,
 				     unsigned int offset, unsigned int length)
 {
@@ -991,11 +929,11 @@ static int svc_rdma_read_chunk_range(struct svc_rdma_read_info *info,
 		dummy.rs_length = min_t(u32, length, segment->rs_length) - offset;
 		dummy.rs_offset = segment->rs_offset + offset;
 
-		ret = svc_rdma_build_read_segment(info, &dummy);
+		ret = svc_rdma_build_read_segment(rqstp, head, &dummy);
 		if (ret < 0)
 			break;
 
-		info->ri_totalbytes += dummy.rs_length;
+		head->rc_readbytes += dummy.rs_length;
 		length -= dummy.rs_length;
 		offset = 0;
 	}
@@ -1004,7 +942,8 @@ static int svc_rdma_read_chunk_range(struct svc_rdma_read_info *info,
 
 /**
  * svc_rdma_read_call_chunk - Build RDMA Read WQEs to pull a Long Message
- * @info: context for RDMA Reads
+ * @rqstp: RPC transaction context
+ * @head: context for ongoing I/O
  *
  * Return values:
  *   %0: RDMA Read WQEs were successfully built
@@ -1013,9 +952,9 @@ static int svc_rdma_read_chunk_range(struct svc_rdma_read_info *info,
  *   %-ENOTCONN: posting failed (connection is lost),
  *   %-EIO: rdma_rw initialization failed (DMA mapping, etc).
  */
-static int svc_rdma_read_call_chunk(struct svc_rdma_read_info *info)
+static int svc_rdma_read_call_chunk(struct svc_rqst *rqstp,
+				    struct svc_rdma_recv_ctxt *head)
 {
-	struct svc_rdma_recv_ctxt *head = info->ri_readctxt;
 	const struct svc_rdma_chunk *call_chunk =
 			pcl_first_chunk(&head->rc_call_pcl);
 	const struct svc_rdma_pcl *pcl = &head->rc_read_pcl;
@@ -1024,17 +963,18 @@ static int svc_rdma_read_call_chunk(struct svc_rdma_read_info *info)
 	int ret;
 
 	if (pcl_is_empty(pcl))
-		return svc_rdma_build_read_chunk(info, call_chunk);
+		return svc_rdma_build_read_chunk(rqstp, head, call_chunk);
 
 	start = 0;
 	chunk = pcl_first_chunk(pcl);
 	length = chunk->ch_position;
-	ret = svc_rdma_read_chunk_range(info, call_chunk, start, length);
+	ret = svc_rdma_read_chunk_range(rqstp, head, call_chunk,
+					start, length);
 	if (ret < 0)
 		return ret;
 
 	pcl_for_each_chunk(chunk, pcl) {
-		ret = svc_rdma_build_read_chunk(info, chunk);
+		ret = svc_rdma_build_read_chunk(rqstp, head, chunk);
 		if (ret < 0)
 			return ret;
 
@@ -1043,8 +983,8 @@ static int svc_rdma_read_call_chunk(struct svc_rdma_read_info *info)
 			break;
 
 		start += length;
-		length = next->ch_position - info->ri_totalbytes;
+		length = next->ch_position - head->rc_readbytes;
-		ret = svc_rdma_read_chunk_range(info, call_chunk,
+		ret = svc_rdma_read_chunk_range(rqstp, head, call_chunk,
 						start, length);
 		if (ret < 0)
 			return ret;
@@ -1052,12 +992,14 @@ static int svc_rdma_read_call_chunk(struct svc_rdma_read_info *info)
 
 	start += length;
 	length = call_chunk->ch_length - start;
-	return svc_rdma_read_chunk_range(info, call_chunk, start, length);
+	return svc_rdma_read_chunk_range(rqstp, head, call_chunk,
+					 start, length);
 }
 
 /**
  * svc_rdma_read_special - Build RDMA Read WQEs to pull a Long Message
- * @info: context for RDMA Reads
+ * @rqstp: RPC transaction context
+ * @head: context for ongoing I/O
  *
  * The start of the data lands in the first page just after the
  * Transport header, and the rest lands in rqstp->rq_arg.pages.
@@ -1073,25 +1015,31 @@ static int svc_rdma_read_call_chunk(struct svc_rdma_read_info *info)
  *   %-ENOTCONN: posting failed (connection is lost),
  *   %-EIO: rdma_rw initialization failed (DMA mapping, etc).
  */
-static noinline int svc_rdma_read_special(struct svc_rdma_read_info *info)
+static noinline int svc_rdma_read_special(struct svc_rqst *rqstp,
+					  struct svc_rdma_recv_ctxt *head)
 {
-	struct xdr_buf *buf = &info->ri_rqst->rq_arg;
+	return svc_rdma_read_call_chunk(rqstp, head);
-	int ret;
+}
 
-	ret = svc_rdma_read_call_chunk(info);
+/* Pages under I/O have been copied to head->rc_pages. Ensure that
-	if (ret < 0)
+ * svc_xprt_release() does not put them when svc_rdma_recvfrom()
-		goto out;
+ * returns. This has to be done after all Read WRs are constructed
+ * to properly handle a page that happens to be part of I/O on behalf
+ * of two different RDMA segments.
+ *
+ * Note: if the subsequent post_send fails, these pages have already
+ * been moved to head->rc_pages and thus will be cleaned up by
+ * svc_rdma_recv_ctxt_put().
+ */
+static void svc_rdma_clear_rqst_pages(struct svc_rqst *rqstp,
+				      struct svc_rdma_recv_ctxt *head)
+{
+	unsigned int i;
 
-	buf->len += info->ri_totalbytes;
+	for (i = 0; i < head->rc_page_count; i++) {
-	buf->buflen += info->ri_totalbytes;
+		head->rc_pages[i] = rqstp->rq_pages[i];
-
+		rqstp->rq_pages[i] = NULL;
-	buf->head[0].iov_base = page_address(info->ri_rqst->rq_pages[0]);
+	}
-	buf->head[0].iov_len = min_t(size_t, PAGE_SIZE, info->ri_totalbytes);
-	buf->pages = &info->ri_rqst->rq_pages[1];
-	buf->page_len = info->ri_totalbytes - buf->head[0].iov_len;
-
-out:
-	return ret;
 }
 
 /**
@@ -1121,49 +1069,27 @@ int svc_rdma_process_read_list(struct svcxprt_rdma *rdma,
 			       struct svc_rqst *rqstp,
 			       struct svc_rdma_recv_ctxt *head)
 {
-	struct svc_rdma_read_info *info;
+	struct svc_rdma_chunk_ctxt *cc = &head->rc_cc;
-	struct svc_rdma_chunk_ctxt *cc;
 	int ret;
 
-	info = svc_rdma_read_info_alloc(rdma);
+	cc->cc_cqe.done = svc_rdma_wc_read_done;
-	if (!info)
+	cc->cc_sqecount = 0;
-		return -ENOMEM;
+	head->rc_pageoff = 0;
-	cc = &info->ri_cc;
+	head->rc_curpage = 0;
-	info->ri_rqst = rqstp;
+	head->rc_readbytes = 0;
-	info->ri_readctxt = head;
-	info->ri_pageno = 0;
-	info->ri_pageoff = 0;
-	info->ri_totalbytes = 0;
 
 	if (pcl_is_empty(&head->rc_call_pcl)) {
 		if (head->rc_read_pcl.cl_count == 1)
-			ret = svc_rdma_read_data_item(info);
+			ret = svc_rdma_read_data_item(rqstp, head);
 		else
-			ret = svc_rdma_read_multiple_chunks(info);
+			ret = svc_rdma_read_multiple_chunks(rqstp, head);
 	} else
-		ret = svc_rdma_read_special(info);
+		ret = svc_rdma_read_special(rqstp, head);
+	svc_rdma_clear_rqst_pages(rqstp, head);
 	if (ret < 0)
-		goto out_err;
+		return ret;
 
 	trace_svcrdma_post_read_chunk(&cc->cc_cid, cc->cc_sqecount);
-	init_completion(&cc->cc_done);
+	ret = svc_rdma_post_chunk_ctxt(rdma, cc);
-	ret = svc_rdma_post_chunk_ctxt(cc);
+	return ret < 0 ? ret : 1;
-	if (ret < 0)
-		goto out_err;
-
-	ret = 1;
-	wait_for_completion(&cc->cc_done);
-	if (cc->cc_status != IB_WC_SUCCESS)
-		ret = -EIO;
-
-	/* rq_respages starts after the last arg page */
-	rqstp->rq_respages = &rqstp->rq_pages[head->rc_page_count];
-	rqstp->rq_next_page = rqstp->rq_respages + 1;
-
-	/* Ensure svc_rdma_recv_ctxt_put() does not try to release pages */
-	head->rc_page_count = 0;
-
-out_err:
-	svc_rdma_read_info_free(info);
-	return ret;
 }

net/sunrpc/xprtrdma/svc_rdma_sendto.c

@@ -113,13 +113,6 @@
 
 static void svc_rdma_wc_send(struct ib_cq *cq, struct ib_wc *wc);
 
-static void svc_rdma_send_cid_init(struct svcxprt_rdma *rdma,
-				   struct rpc_rdma_cid *cid)
-{
-	cid->ci_queue_id = rdma->sc_sq_cq->res.id;
-	cid->ci_completion_id = atomic_inc_return(&rdma->sc_completion_ids);
-}
-
 static struct svc_rdma_send_ctxt *
 svc_rdma_send_ctxt_alloc(struct svcxprt_rdma *rdma)
 {
@@ -129,7 +122,7 @@ svc_rdma_send_ctxt_alloc(struct svcxprt_rdma *rdma)
 	void *buffer;
 	int i;
 
-	ctxt = kmalloc_node(struct_size(ctxt, sc_sges, rdma->sc_max_send_sges),
+	ctxt = kzalloc_node(struct_size(ctxt, sc_sges, rdma->sc_max_send_sges),
 			    GFP_KERNEL, node);
 	if (!ctxt)
 		goto fail0;
@@ -143,6 +136,7 @@ svc_rdma_send_ctxt_alloc(struct svcxprt_rdma *rdma)
 
 	svc_rdma_send_cid_init(rdma, &ctxt->sc_cid);
 
+	ctxt->sc_rdma = rdma;
 	ctxt->sc_send_wr.next = NULL;
 	ctxt->sc_send_wr.wr_cqe = &ctxt->sc_cqe;
 	ctxt->sc_send_wr.sg_list = ctxt->sc_sges;
@@ -200,10 +194,11 @@ struct svc_rdma_send_ctxt *svc_rdma_send_ctxt_get(struct svcxprt_rdma *rdma)
 
 	spin_lock(&rdma->sc_send_lock);
 	node = llist_del_first(&rdma->sc_send_ctxts);
+	spin_unlock(&rdma->sc_send_lock);
 	if (!node)
 		goto out_empty;
+
 	ctxt = llist_entry(node, struct svc_rdma_send_ctxt, sc_node);
-	spin_unlock(&rdma->sc_send_lock);
 
 out:
 	rpcrdma_set_xdrlen(&ctxt->sc_hdrbuf, 0);
@@ -216,13 +211,45 @@ out:
 	return ctxt;
 
 out_empty:
-	spin_unlock(&rdma->sc_send_lock);
 	ctxt = svc_rdma_send_ctxt_alloc(rdma);
 	if (!ctxt)
 		return NULL;
 	goto out;
 }
 
+static void svc_rdma_send_ctxt_release(struct svcxprt_rdma *rdma,
+				       struct svc_rdma_send_ctxt *ctxt)
+{
+	struct ib_device *device = rdma->sc_cm_id->device;
+	unsigned int i;
+
+	if (ctxt->sc_page_count)
+		release_pages(ctxt->sc_pages, ctxt->sc_page_count);
+
+	/* The first SGE contains the transport header, which
+	 * remains mapped until @ctxt is destroyed.
+	 */
+	for (i = 1; i < ctxt->sc_send_wr.num_sge; i++) {
+		trace_svcrdma_dma_unmap_page(&ctxt->sc_cid,
+					     ctxt->sc_sges[i].addr,
+					     ctxt->sc_sges[i].length);
+		ib_dma_unmap_page(device,
+				  ctxt->sc_sges[i].addr,
+				  ctxt->sc_sges[i].length,
+				  DMA_TO_DEVICE);
+	}
+
+	llist_add(&ctxt->sc_node, &rdma->sc_send_ctxts);
+}
+
+static void svc_rdma_send_ctxt_put_async(struct work_struct *work)
+{
+	struct svc_rdma_send_ctxt *ctxt;
+
+	ctxt = container_of(work, struct svc_rdma_send_ctxt, sc_work);
+	svc_rdma_send_ctxt_release(ctxt->sc_rdma, ctxt);
+}
+
 /**
  * svc_rdma_send_ctxt_put - Return send_ctxt to free list
  * @rdma: controlling svcxprt_rdma
@@ -233,26 +260,8 @@ out_empty:
 void svc_rdma_send_ctxt_put(struct svcxprt_rdma *rdma,
 			    struct svc_rdma_send_ctxt *ctxt)
 {
-	struct ib_device *device = rdma->sc_cm_id->device;
+	INIT_WORK(&ctxt->sc_work, svc_rdma_send_ctxt_put_async);
-	unsigned int i;
+	queue_work(svcrdma_wq, &ctxt->sc_work);
-
-	if (ctxt->sc_page_count)
-		release_pages(ctxt->sc_pages, ctxt->sc_page_count);
-
-	/* The first SGE contains the transport header, which
-	 * remains mapped until @ctxt is destroyed.
-	 */
-	for (i = 1; i < ctxt->sc_send_wr.num_sge; i++) {
-		ib_dma_unmap_page(device,
-				  ctxt->sc_sges[i].addr,
-				  ctxt->sc_sges[i].length,
-				  DMA_TO_DEVICE);
-		trace_svcrdma_dma_unmap_page(rdma,
-					     ctxt->sc_sges[i].addr,
-					     ctxt->sc_sges[i].length);
-	}
-
-	llist_add(&ctxt->sc_node, &rdma->sc_send_ctxts);
 }
 
 /**
@@ -289,7 +298,7 @@ static void svc_rdma_wc_send(struct ib_cq *cq, struct ib_wc *wc)
 	if (unlikely(wc->status != IB_WC_SUCCESS))
 		goto flushed;
 
-	trace_svcrdma_wc_send(wc, &ctxt->sc_cid);
+	trace_svcrdma_wc_send(&ctxt->sc_cid);
 	svc_rdma_send_ctxt_put(rdma, ctxt);
 	return;
 
@@ -327,13 +336,13 @@ int svc_rdma_send(struct svcxprt_rdma *rdma, struct svc_rdma_send_ctxt *ctxt)
 	while (1) {
 		if ((atomic_dec_return(&rdma->sc_sq_avail) < 0)) {
 			percpu_counter_inc(&svcrdma_stat_sq_starve);
-			trace_svcrdma_sq_full(rdma);
+			trace_svcrdma_sq_full(rdma, &ctxt->sc_cid);
 			atomic_inc(&rdma->sc_sq_avail);
 			wait_event(rdma->sc_send_wait,
 				   atomic_read(&rdma->sc_sq_avail) > 1);
 			if (test_bit(XPT_CLOSE, &rdma->sc_xprt.xpt_flags))
 				return -ENOTCONN;
-			trace_svcrdma_sq_retry(rdma);
+			trace_svcrdma_sq_retry(rdma, &ctxt->sc_cid);
 			continue;
 		}
 
@@ -344,7 +353,7 @@ int svc_rdma_send(struct svcxprt_rdma *rdma, struct svc_rdma_send_ctxt *ctxt)
 		return 0;
 	}
 
-	trace_svcrdma_sq_post_err(rdma, ret);
+	trace_svcrdma_sq_post_err(rdma, &ctxt->sc_cid, ret);
 	svc_xprt_deferred_close(&rdma->sc_xprt);
 	wake_up(&rdma->sc_send_wait);
 	return ret;
@@ -534,14 +543,14 @@ static int svc_rdma_page_dma_map(void *data, struct page *page,
 	if (ib_dma_mapping_error(dev, dma_addr))
 		goto out_maperr;
 
-	trace_svcrdma_dma_map_page(rdma, dma_addr, len);
+	trace_svcrdma_dma_map_page(&ctxt->sc_cid, dma_addr, len);
 	ctxt->sc_sges[ctxt->sc_cur_sge_no].addr = dma_addr;
 	ctxt->sc_sges[ctxt->sc_cur_sge_no].length = len;
 	ctxt->sc_send_wr.num_sge++;
 	return 0;
 
 out_maperr:
-	trace_svcrdma_dma_map_err(rdma, dma_addr, len);
+	trace_svcrdma_dma_map_err(&ctxt->sc_cid, dma_addr, len);
 	return -EIO;
 }
 
@@ -653,7 +662,7 @@ static int svc_rdma_xb_count_sges(const struct xdr_buf *xdr,
  * svc_rdma_pull_up_needed - Determine whether to use pull-up
  * @rdma: controlling transport
  * @sctxt: send_ctxt for the Send WR
- * @rctxt: Write and Reply chunks provided by client
+ * @write_pcl: Write chunk list provided by client
  * @xdr: xdr_buf containing RPC message to transmit
  *
  * Returns:
@@ -662,7 +671,7 @@ static int svc_rdma_xb_count_sges(const struct xdr_buf *xdr,
  */
 static bool svc_rdma_pull_up_needed(const struct svcxprt_rdma *rdma,
 				    const struct svc_rdma_send_ctxt *sctxt,
-				    const struct svc_rdma_recv_ctxt *rctxt,
+				    const struct svc_rdma_pcl *write_pcl,
 				    const struct xdr_buf *xdr)
 {
 	/* Resources needed for the transport header */
@@ -672,7 +681,7 @@ static bool svc_rdma_pull_up_needed(const struct svcxprt_rdma *rdma,
 	};
 	int ret;
 
-	ret = pcl_process_nonpayloads(&rctxt->rc_write_pcl, xdr,
+	ret = pcl_process_nonpayloads(write_pcl, xdr,
 				      svc_rdma_xb_count_sges, &args);
 	if (ret < 0)
 		return false;
@@ -728,7 +737,7 @@ static int svc_rdma_xb_linearize(const struct xdr_buf *xdr,
  * svc_rdma_pull_up_reply_msg - Copy Reply into a single buffer
  * @rdma: controlling transport
  * @sctxt: send_ctxt for the Send WR; xprt hdr is already prepared
- * @rctxt: Write and Reply chunks provided by client
+ * @write_pcl: Write chunk list provided by client
  * @xdr: prepared xdr_buf containing RPC message
  *
  * The device is not capable of sending the reply directly.
@@ -743,7 +752,7 @@ static int svc_rdma_xb_linearize(const struct xdr_buf *xdr,
  */
 static int svc_rdma_pull_up_reply_msg(const struct svcxprt_rdma *rdma,
 				      struct svc_rdma_send_ctxt *sctxt,
-				      const struct svc_rdma_recv_ctxt *rctxt,
+				      const struct svc_rdma_pcl *write_pcl,
 				      const struct xdr_buf *xdr)
 {
 	struct svc_rdma_pullup_data args = {
@@ -751,7 +760,7 @@ static int svc_rdma_pull_up_reply_msg(const struct svcxprt_rdma *rdma,
 	};
 	int ret;
 
-	ret = pcl_process_nonpayloads(&rctxt->rc_write_pcl, xdr,
+	ret = pcl_process_nonpayloads(write_pcl, xdr,
 				      svc_rdma_xb_linearize, &args);
 	if (ret < 0)
 		return ret;
@@ -764,7 +773,8 @@ static int svc_rdma_pull_up_reply_msg(const struct svcxprt_rdma *rdma,
 /* svc_rdma_map_reply_msg - DMA map the buffer holding RPC message
  * @rdma: controlling transport
  * @sctxt: send_ctxt for the Send WR
- * @rctxt: Write and Reply chunks provided by client
+ * @write_pcl: Write chunk list provided by client
+ * @reply_pcl: Reply chunk provided by client
  * @xdr: prepared xdr_buf containing RPC message
  *
  * Returns:
@@ -776,7 +786,8 @@ static int svc_rdma_pull_up_reply_msg(const struct svcxprt_rdma *rdma,
  */
 int svc_rdma_map_reply_msg(struct svcxprt_rdma *rdma,
 			   struct svc_rdma_send_ctxt *sctxt,
-			   const struct svc_rdma_recv_ctxt *rctxt,
+			   const struct svc_rdma_pcl *write_pcl,
+			   const struct svc_rdma_pcl *reply_pcl,
 			   const struct xdr_buf *xdr)
 {
 	struct svc_rdma_map_data args = {
@@ -789,18 +800,18 @@ int svc_rdma_map_reply_msg(struct svcxprt_rdma *rdma,
 	sctxt->sc_sges[0].length = sctxt->sc_hdrbuf.len;
 
 	/* If there is a Reply chunk, nothing follows the transport
-	 * header, and we're done here.
+	 * header, so there is nothing to map.
 	 */
-	if (!pcl_is_empty(&rctxt->rc_reply_pcl))
+	if (!pcl_is_empty(reply_pcl))
 		return 0;
 
 	/* For pull-up, svc_rdma_send() will sync the transport header.
 	 * No additional DMA mapping is necessary.
 	 */
-	if (svc_rdma_pull_up_needed(rdma, sctxt, rctxt, xdr))
+	if (svc_rdma_pull_up_needed(rdma, sctxt, write_pcl, xdr))
-		return svc_rdma_pull_up_reply_msg(rdma, sctxt, rctxt, xdr);
+		return svc_rdma_pull_up_reply_msg(rdma, sctxt, write_pcl, xdr);
 
-	return pcl_process_nonpayloads(&rctxt->rc_write_pcl, xdr,
+	return pcl_process_nonpayloads(write_pcl, xdr,
 				       svc_rdma_xb_dma_map, &args);
 }
 
@@ -848,7 +859,8 @@ static int svc_rdma_send_reply_msg(struct svcxprt_rdma *rdma,
 {
 	int ret;
 
-	ret = svc_rdma_map_reply_msg(rdma, sctxt, rctxt, &rqstp->rq_res);
+	ret = svc_rdma_map_reply_msg(rdma, sctxt, &rctxt->rc_write_pcl,
+				     &rctxt->rc_reply_pcl, &rqstp->rq_res);
 	if (ret < 0)
 		return ret;
 

net/sunrpc/xprtrdma/svc_rdma_transport.c

@@ -125,6 +125,9 @@ static void qp_event_handler(struct ib_event *event, void *context)
 static struct svcxprt_rdma *svc_rdma_create_xprt(struct svc_serv *serv,
 						 struct net *net, int node)
 {
+	static struct lock_class_key svcrdma_rwctx_lock;
+	static struct lock_class_key svcrdma_sctx_lock;
+	static struct lock_class_key svcrdma_dto_lock;
 	struct svcxprt_rdma *cma_xprt;
 
 	cma_xprt = kzalloc_node(sizeof(*cma_xprt), GFP_KERNEL, node);
@@ -134,6 +137,7 @@ static struct svcxprt_rdma *svc_rdma_create_xprt(struct svc_serv *serv,
 	svc_xprt_init(net, &svc_rdma_class, &cma_xprt->sc_xprt, serv);
 	INIT_LIST_HEAD(&cma_xprt->sc_accept_q);
 	INIT_LIST_HEAD(&cma_xprt->sc_rq_dto_q);
+	INIT_LIST_HEAD(&cma_xprt->sc_read_complete_q);
 	init_llist_head(&cma_xprt->sc_send_ctxts);
 	init_llist_head(&cma_xprt->sc_recv_ctxts);
 	init_llist_head(&cma_xprt->sc_rw_ctxts);
@@ -141,8 +145,11 @@ static struct svcxprt_rdma *svc_rdma_create_xprt(struct svc_serv *serv,
 
 	spin_lock_init(&cma_xprt->sc_lock);
 	spin_lock_init(&cma_xprt->sc_rq_dto_lock);
+	lockdep_set_class(&cma_xprt->sc_rq_dto_lock, &svcrdma_dto_lock);
 	spin_lock_init(&cma_xprt->sc_send_lock);
+	lockdep_set_class(&cma_xprt->sc_send_lock, &svcrdma_sctx_lock);
 	spin_lock_init(&cma_xprt->sc_rw_ctxt_lock);
+	lockdep_set_class(&cma_xprt->sc_rw_ctxt_lock, &svcrdma_rwctx_lock);
 
 	/*
 	 * Note that this implies that the underlying transport support
@@ -391,37 +398,35 @@ static struct svc_xprt *svc_rdma_accept(struct svc_xprt *xprt)
 	dev = newxprt->sc_cm_id->device;
 	newxprt->sc_port_num = newxprt->sc_cm_id->port_num;
 
-	/* Qualify the transport resource defaults with the
+	newxprt->sc_max_req_size = svcrdma_max_req_size;
-	 * capabilities of this particular device */
+	newxprt->sc_max_requests = svcrdma_max_requests;
+	newxprt->sc_max_bc_requests = svcrdma_max_bc_requests;
+	newxprt->sc_recv_batch = RPCRDMA_MAX_RECV_BATCH;
+	newxprt->sc_fc_credits = cpu_to_be32(newxprt->sc_max_requests);
+
+	/* Qualify the transport's resource defaults with the
+	 * capabilities of this particular device.
+	 */
+
 	/* Transport header, head iovec, tail iovec */
 	newxprt->sc_max_send_sges = 3;
 	/* Add one SGE per page list entry */
 	newxprt->sc_max_send_sges += (svcrdma_max_req_size / PAGE_SIZE) + 1;
 	if (newxprt->sc_max_send_sges > dev->attrs.max_send_sge)
 		newxprt->sc_max_send_sges = dev->attrs.max_send_sge;
-	newxprt->sc_max_req_size = svcrdma_max_req_size;
-	newxprt->sc_max_requests = svcrdma_max_requests;
-	newxprt->sc_max_bc_requests = svcrdma_max_bc_requests;
-	newxprt->sc_recv_batch = RPCRDMA_MAX_RECV_BATCH;
 	rq_depth = newxprt->sc_max_requests + newxprt->sc_max_bc_requests +
 		   newxprt->sc_recv_batch;
 	if (rq_depth > dev->attrs.max_qp_wr) {
-		pr_warn("svcrdma: reducing receive depth to %d\n",
-			dev->attrs.max_qp_wr);
 		rq_depth = dev->attrs.max_qp_wr;
 		newxprt->sc_recv_batch = 1;
 		newxprt->sc_max_requests = rq_depth - 2;
 		newxprt->sc_max_bc_requests = 2;
 	}
-	newxprt->sc_fc_credits = cpu_to_be32(newxprt->sc_max_requests);
 	ctxts = rdma_rw_mr_factor(dev, newxprt->sc_port_num, RPCSVC_MAXPAGES);
 	ctxts *= newxprt->sc_max_requests;
 	newxprt->sc_sq_depth = rq_depth + ctxts;
-	if (newxprt->sc_sq_depth > dev->attrs.max_qp_wr) {
+	if (newxprt->sc_sq_depth > dev->attrs.max_qp_wr)
-		pr_warn("svcrdma: reducing send depth to %d\n",
-			dev->attrs.max_qp_wr);
 		newxprt->sc_sq_depth = dev->attrs.max_qp_wr;
-	}
 	atomic_set(&newxprt->sc_sq_avail, newxprt->sc_sq_depth);
 
 	newxprt->sc_pd = ib_alloc_pd(dev, 0);
@@ -451,8 +456,6 @@ static struct svc_xprt *svc_rdma_accept(struct svc_xprt *xprt)
 	qp_attr.qp_type = IB_QPT_RC;
 	qp_attr.send_cq = newxprt->sc_sq_cq;
 	qp_attr.recv_cq = newxprt->sc_rq_cq;
-	dprintk("svcrdma: newxprt->sc_cm_id=%p, newxprt->sc_pd=%p\n",
-		newxprt->sc_cm_id, newxprt->sc_pd);
 	dprintk("    cap.max_send_wr = %d, cap.max_recv_wr = %d\n",
 		qp_attr.cap.max_send_wr, qp_attr.cap.max_recv_wr);
 	dprintk("    cap.max_send_sge = %d, cap.max_recv_sge = %d\n",
@@ -506,7 +509,7 @@ static struct svc_xprt *svc_rdma_accept(struct svc_xprt *xprt)
 	}
 
 #if IS_ENABLED(CONFIG_SUNRPC_DEBUG)
-	dprintk("svcrdma: new connection %p accepted:\n", newxprt);
+	dprintk("svcrdma: new connection accepted on device %s:\n", dev->name);
 	sap = (struct sockaddr *)&newxprt->sc_cm_id->route.addr.src_addr;
 	dprintk("    local address   : %pIS:%u\n", sap, rpc_get_port(sap));
 	sap = (struct sockaddr *)&newxprt->sc_cm_id->route.addr.dst_addr;
@@ -547,6 +550,7 @@ static void __svc_rdma_free(struct work_struct *work)
 	/* This blocks until the Completion Queues are empty */
 	if (rdma->sc_qp && !IS_ERR(rdma->sc_qp))
 		ib_drain_qp(rdma->sc_qp);
+	flush_workqueue(svcrdma_wq);
 
 	svc_rdma_flush_recv_queues(rdma);
 

net/sunrpc/xprtrdma/verbs.c

@@ -1364,7 +1364,7 @@ void rpcrdma_post_recvs(struct rpcrdma_xprt *r_xprt, int needed, bool temp)
 		}
 
 		rep->rr_cid.ci_queue_id = ep->re_attr.recv_cq->res.id;
-		trace_xprtrdma_post_recv(rep);
+		trace_xprtrdma_post_recv(&rep->rr_cid);
 		rep->rr_recv_wr.next = wr;
 		wr = &rep->rr_recv_wr;
 		--needed;