fourteen smb3 client fixes, most smbdirect related

-----BEGIN PGP SIGNATURE-----
 
 iQGzBAABCgAdFiEE6fsu8pdIjtWE/DpLiiy9cAdyT1EFAmhEsJkACgkQiiy9cAdy
 T1Gjlgv+P1I4uD8iDbkJvyrh7mVcnIDnjttbgN1pTZZnpcX7JIyMPR9YPUn+QTKf
 3GStGgLHqKtnXYYCdrkpPipsH/fjt7kjVLUC/YMhUGEr5dSFKzygOVaRTo4LDgNO
 sSlOxTPyTaPgxdvIXriU5CeJGtm95GqowLDG29We9ubX1P5GLjIjGlMbAisAk4/Q
 ek478D0qBSO5u6FkMYUFuSbjGwOYWAlabMxeOVHpjgEH/36tVtDKOc/bORnVZtHB
 6bRjhymn9CZQOE2P+gDSqF6pZwHWhU7uYJYW+kbDbmDIgWDk9xl5/kbbqZ6RR3t5
 joPWzZkO1I58avmJ31aauhvRpYIOXfOhx2dHB6VM/7bSHTvE6CW60RIlpXo1xbO6
 d30pTy7ICqP/RWNZlt5ZgoPTDB1hPmVDdMXoELp7uZzChhrAz4qc/2BSD0Y5fCjE
 e/U0bPfsyjscB7NtpcrQYVMbOsrweQ4qKUgolEk1klLolsKxDGE176wbPmLNAzo4
 C9SzyxU3
 =RvlW
 -----END PGP SIGNATURE-----

Merge tag '6.16-rc-part2-smb3-client-fixes' of git://git.samba.org/sfrench/cifs-2.6

Pull more smb client updates from Steve French:

 - multichannel/reconnect fixes

 - move smbdirect (smb over RDMA) defines to fs/smb/common so they will
   be able to be used in the future more broadly, and a documentation
   update explaining setting up smbdirect mounts

 - update email address for Paulo

* tag '6.16-rc-part2-smb3-client-fixes' of git://git.samba.org/sfrench/cifs-2.6:
  cifs: update internal version number
  MAINTAINERS, mailmap: Update Paulo Alcantara's email address
  cifs: add documentation for smbdirect setup
  cifs: do not disable interface polling on failure
  cifs: serialize other channels when query server interfaces is pending
  cifs: deal with the channel loading lag while picking channels
  smb: client: make use of common smbdirect_socket_parameters
  smb: smbdirect: introduce smbdirect_socket_parameters
  smb: client: make use of common smbdirect_socket
  smb: smbdirect: add smbdirect_socket.h
  smb: client: make use of common smbdirect.h
  smb: smbdirect: add smbdirect.h with public structures
  smb: client: make use of common smbdirect_pdu.h
  smb: smbdirect: add smbdirect_pdu.h with protocol definitions

.mailmap

@@ -602,6 +602,12 @@ Paul Mackerras <paulus@ozlabs.org> <paulus@samba.org>
 Paul Mackerras <paulus@ozlabs.org> <paulus@au1.ibm.com>
 Paul Moore <paul@paul-moore.com> <paul.moore@hp.com>
 Paul Moore <paul@paul-moore.com> <pmoore@redhat.com>
+Paulo Alcantara <pc@manguebit.org> <pcacjr@zytor.com>
+Paulo Alcantara <pc@manguebit.org> <paulo@paulo.ac>
+Paulo Alcantara <pc@manguebit.org> <pc@cjr.nz>
+Paulo Alcantara <pc@manguebit.org> <palcantara@suse.de>
+Paulo Alcantara <pc@manguebit.org> <palcantara@suse.com>
+Paulo Alcantara <pc@manguebit.org> <pc@manguebit.com>
 Pavankumar Kondeti <quic_pkondeti@quicinc.com> <pkondeti@codeaurora.org>
 Peter A Jonsson <pj@ludd.ltu.se>
 Peter Oruba <peter.oruba@amd.com>

Documentation/filesystems/smb/index.rst

@@ -8,3 +8,4 @@ CIFS
 
    ksmbd
    cifsroot
+   smbdirect

Documentation/filesystems/smb/smbdirect.rst

@@ -0,0 +1,103 @@
+.. SPDX-License-Identifier: GPL-2.0
+
+===========================
+SMB Direct - SMB3 over RDMA
+===========================
+
+This document describes how to set up the Linux SMB client and server to
+use RDMA.
+
+Overview
+========
+The Linux SMB kernel client supports SMB Direct, which is a transport
+scheme for SMB3 that uses RDMA (Remote Direct Memory Access) to provide
+high throughput and low latencies by bypassing the traditional TCP/IP
+stack.
+SMB Direct on the Linux SMB client can be tested against KSMBD - a
+kernel-space SMB server.
+
+Installation
+=============
+- Install an RDMA device. As long as the RDMA device driver is supported
+  by the kernel, it should work. This includes both software emulators (soft
+  RoCE, soft iWARP) and hardware devices (InfiniBand, RoCE, iWARP).
+
+- Install a kernel with SMB Direct support. The first kernel release to
+  support SMB Direct on both the client and server side is 5.15. Therefore,
+  a distribution compatible with kernel 5.15 or later is required.
+
+- Install cifs-utils, which provides the `mount.cifs` command to mount SMB
+  shares.
+
+- Configure the RDMA stack
+
+  Make sure that your kernel configuration has RDMA support enabled. Under
+  Device Drivers -> Infiniband support, update the kernel configuration to
+  enable Infiniband support.
+
+  Enable the appropriate IB HCA support or iWARP adapter support,
+  depending on your hardware.
+
+  If you are using InfiniBand, enable IP-over-InfiniBand support.
+
+  For soft RDMA, enable either the soft iWARP (`RDMA _SIW`) or soft RoCE
+  (`RDMA_RXE`) module. Install the `iproute2` package and use the
+  `rdma link add` command to load the module and create an
+  RDMA interface.
+
+  e.g. if your local ethernet interface is `eth0`, you can use:
+
+    .. code-block:: bash
+
+        sudo rdma link add siw0 type siw netdev eth0
+
+- Enable SMB Direct support for both the server and the client in the kernel
+  configuration.
+
+    Server Setup
+
+    .. code-block:: text
+
+        Network File Systems  --->
+            <M> SMB3 server support
+                [*] Support for SMB Direct protocol
+
+    Client Setup
+
+    .. code-block:: text
+
+        Network File Systems  --->
+            <M> SMB3 and CIFS support (advanced network filesystem)
+                [*] SMB Direct support
+
+- Build and install the kernel. SMB Direct support will be enabled in the
+  cifs.ko and ksmbd.ko modules.
+
+Setup and Usage
+================
+
+- Set up and start a KSMBD server as described in the `KSMBD documentation
+  <https://www.kernel.org/doc/Documentation/filesystems/smb/ksmbd.rst>`_.
+  Also add the "server multi channel support = yes" parameter to ksmbd.conf.
+
+- On the client, mount the share with `rdma` mount option to use SMB Direct
+  (specify a SMB version 3.0 or higher using `vers`).
+
+  For example:
+
+    .. code-block:: bash
+
+        mount -t cifs //server/share /mnt/point -o vers=3.1.1,rdma
+
+- To verify that the mount is using SMB Direct, you can check dmesg for the
+  following log line after mounting:
+
+    .. code-block:: text
+
+        CIFS: VFS: RDMA transport established
+
+  Or, verify `rdma` mount option for the share in `/proc/mounts`:
+
+    .. code-block:: bash
+
+        cat /proc/mounts | grep cifs

MAINTAINERS

@@ -5986,7 +5986,7 @@ X:	drivers/clk/clkdev.c
 COMMON INTERNET FILE SYSTEM CLIENT (CIFS and SMB3)
 M:	Steve French <sfrench@samba.org>
 M:	Steve French <smfrench@gmail.com>
-R:	Paulo Alcantara <pc@manguebit.com> (DFS, global name space)
+R:	Paulo Alcantara <pc@manguebit.org> (DFS, global name space)
 R:	Ronnie Sahlberg <ronniesahlberg@gmail.com> (directory leases, sparse files)
 R:	Shyam Prasad N <sprasad@microsoft.com> (multichannel)
 R:	Tom Talpey <tom@talpey.com> (RDMA, smbdirect)
@@ -9280,7 +9280,7 @@ F:	include/linux/iomap.h
 
 FILESYSTEMS [NETFS LIBRARY]
 M:	David Howells <dhowells@redhat.com>
-M:	Paulo Alcantara <pc@manguebit.com>
+M:	Paulo Alcantara <pc@manguebit.org>
 L:	netfs@lists.linux.dev
 L:	linux-fsdevel@vger.kernel.org
 S:	Supported

fs/smb/client/cifs_debug.c

@@ -362,6 +362,10 @@ static int cifs_debug_data_proc_show(struct seq_file *m, void *v)
 	c = 0;
 	spin_lock(&cifs_tcp_ses_lock);
 	list_for_each_entry(server, &cifs_tcp_ses_list, tcp_ses_list) {
+#ifdef CONFIG_CIFS_SMB_DIRECT
+		struct smbdirect_socket_parameters *sp;
+#endif
+
 		/* channel info will be printed as a part of sessions below */
 		if (SERVER_IS_CHAN(server))
 			continue;
@@ -383,25 +387,26 @@ static int cifs_debug_data_proc_show(struct seq_file *m, void *v)
 			seq_printf(m, "\nSMBDirect transport not available");
 			goto skip_rdma;
 		}
+		sp = &server->smbd_conn->socket.parameters;
 
 		seq_printf(m, "\nSMBDirect (in hex) protocol version: %x "
 			"transport status: %x",
 			server->smbd_conn->protocol,
-			server->smbd_conn->transport_status);
+			server->smbd_conn->socket.status);
 		seq_printf(m, "\nConn receive_credit_max: %x "
 			"send_credit_target: %x max_send_size: %x",
-			server->smbd_conn->receive_credit_max,
+			sp->recv_credit_max,
-			server->smbd_conn->send_credit_target,
+			sp->send_credit_target,
-			server->smbd_conn->max_send_size);
+			sp->max_send_size);
 		seq_printf(m, "\nConn max_fragmented_recv_size: %x "
 			"max_fragmented_send_size: %x max_receive_size:%x",
-			server->smbd_conn->max_fragmented_recv_size,
+			sp->max_fragmented_recv_size,
-			server->smbd_conn->max_fragmented_send_size,
+			sp->max_fragmented_send_size,
-			server->smbd_conn->max_receive_size);
+			sp->max_recv_size);
 		seq_printf(m, "\nConn keep_alive_interval: %x "
 			"max_readwrite_size: %x rdma_readwrite_threshold: %x",
-			server->smbd_conn->keep_alive_interval,
+			sp->keepalive_interval_msec * 1000,
-			server->smbd_conn->max_readwrite_size,
+			sp->max_read_write_size,
 			server->smbd_conn->rdma_readwrite_threshold);
 		seq_printf(m, "\nDebug count_get_receive_buffer: %x "
 			"count_put_receive_buffer: %x count_send_empty: %x",

fs/smb/client/cifsfs.h

@@ -145,6 +145,6 @@ extern const struct export_operations cifs_export_ops;
 #endif /* CONFIG_CIFS_NFSD_EXPORT */
 
 /* when changing internal version - update following two lines at same time */
-#define SMB3_PRODUCT_BUILD 54
+#define SMB3_PRODUCT_BUILD 55
-#define CIFS_VERSION   "2.54"
+#define CIFS_VERSION   "2.55"
 #endif				/* _CIFSFS_H */

fs/smb/client/cifsglob.h

@@ -1085,6 +1085,7 @@ struct cifs_chan {
 };
 
 #define CIFS_SES_FLAG_SCALE_CHANNELS (0x1)
+#define CIFS_SES_FLAGS_PENDING_QUERY_INTERFACES (0x2)
 
 /*
  * Session structure.  One of these for each uid session with a particular host

fs/smb/client/connect.c

@@ -116,13 +116,9 @@ static void smb2_query_server_interfaces(struct work_struct *work)
 	rc = server->ops->query_server_interfaces(xid, tcon, false);
 	free_xid(xid);
 
-	if (rc) {
+	if (rc)
-		if (rc == -EOPNOTSUPP)
-			return;
-
 		cifs_dbg(FYI, "%s: failed to query server interfaces: %d\n",
 				__func__, rc);
-	}
 
 	queue_delayed_work(cifsiod_wq, &tcon->query_interfaces,
 			   (SMB_INTERFACE_POLL_INTERVAL * HZ));

fs/smb/client/smb2ops.c

@@ -504,6 +504,9 @@ smb3_negotiate_wsize(struct cifs_tcon *tcon, struct smb3_fs_context *ctx)
 	wsize = min_t(unsigned int, wsize, server->max_write);
 #ifdef CONFIG_CIFS_SMB_DIRECT
 	if (server->rdma) {
+		struct smbdirect_socket_parameters *sp =
+			&server->smbd_conn->socket.parameters;
+
 		if (server->sign)
 			/*
 			 * Account for SMB2 data transfer packet header and
@@ -511,12 +514,12 @@ smb3_negotiate_wsize(struct cifs_tcon *tcon, struct smb3_fs_context *ctx)
 			 */
 			wsize = min_t(unsigned int,
 				wsize,
-				server->smbd_conn->max_fragmented_send_size -
+				sp->max_fragmented_send_size -
 					SMB2_READWRITE_PDU_HEADER_SIZE -
 					sizeof(struct smb2_transform_hdr));
 		else
 			wsize = min_t(unsigned int,
-				wsize, server->smbd_conn->max_readwrite_size);
+				wsize, sp->max_read_write_size);
 	}
 #endif
 	if (!(server->capabilities & SMB2_GLOBAL_CAP_LARGE_MTU))
@@ -552,6 +555,9 @@ smb3_negotiate_rsize(struct cifs_tcon *tcon, struct smb3_fs_context *ctx)
 	rsize = min_t(unsigned int, rsize, server->max_read);
 #ifdef CONFIG_CIFS_SMB_DIRECT
 	if (server->rdma) {
+		struct smbdirect_socket_parameters *sp =
+			&server->smbd_conn->socket.parameters;
+
 		if (server->sign)
 			/*
 			 * Account for SMB2 data transfer packet header and
@@ -559,12 +565,12 @@ smb3_negotiate_rsize(struct cifs_tcon *tcon, struct smb3_fs_context *ctx)
 			 */
 			rsize = min_t(unsigned int,
 				rsize,
-				server->smbd_conn->max_fragmented_recv_size -
+				sp->max_fragmented_recv_size -
 					SMB2_READWRITE_PDU_HEADER_SIZE -
 					sizeof(struct smb2_transform_hdr));
 		else
 			rsize = min_t(unsigned int,
-				rsize, server->smbd_conn->max_readwrite_size);
+				rsize, sp->max_read_write_size);
 	}
 #endif
 

fs/smb/client/smb2pdu.c

@@ -36,6 +36,7 @@
 #include "smb2glob.h"
 #include "cifspdu.h"
 #include "cifs_spnego.h"
+#include "../common/smbdirect/smbdirect.h"
 #include "smbdirect.h"
 #include "trace.h"
 #ifdef CONFIG_CIFS_DFS_UPCALL
@@ -411,14 +412,23 @@ smb2_reconnect(__le16 smb2_command, struct cifs_tcon *tcon,
 	if (!rc &&
 	    (server->capabilities & SMB2_GLOBAL_CAP_MULTI_CHANNEL) &&
 	    server->ops->query_server_interfaces) {
-		mutex_unlock(&ses->session_mutex);
-
 		/*
-		 * query server network interfaces, in case they change
+		 * query server network interfaces, in case they change.
+		 * Also mark the session as pending this update while the query
+		 * is in progress. This will be used to avoid calling
+		 * smb2_reconnect recursively.
 		 */
+		ses->flags |= CIFS_SES_FLAGS_PENDING_QUERY_INTERFACES;
 		xid = get_xid();
 		rc = server->ops->query_server_interfaces(xid, tcon, false);
 		free_xid(xid);
+		ses->flags &= ~CIFS_SES_FLAGS_PENDING_QUERY_INTERFACES;
+
+		/* regardless of rc value, setup polling */
+		queue_delayed_work(cifsiod_wq, &tcon->query_interfaces,
+				   (SMB_INTERFACE_POLL_INTERVAL * HZ));
+
+		mutex_unlock(&ses->session_mutex);
 
 		if (rc == -EOPNOTSUPP && ses->chan_count > 1) {
 			/*
@@ -438,11 +448,8 @@ smb2_reconnect(__le16 smb2_command, struct cifs_tcon *tcon,
 		if (ses->chan_max > ses->chan_count &&
 		    ses->iface_count &&
 		    !SERVER_IS_CHAN(server)) {
-			if (ses->chan_count == 1) {
+			if (ses->chan_count == 1)
 				cifs_server_dbg(VFS, "supports multichannel now\n");
-				queue_delayed_work(cifsiod_wq, &tcon->query_interfaces,
-						 (SMB_INTERFACE_POLL_INTERVAL * HZ));
-			}
 
 			cifs_try_adding_channels(ses);
 		}
@@ -560,11 +567,18 @@ static int smb2_ioctl_req_init(u32 opcode, struct cifs_tcon *tcon,
 			       struct TCP_Server_Info *server,
 			       void **request_buf, unsigned int *total_len)
 {
-	/* Skip reconnect only for FSCTL_VALIDATE_NEGOTIATE_INFO IOCTLs */
+	/*
-	if (opcode == FSCTL_VALIDATE_NEGOTIATE_INFO) {
+	 * Skip reconnect in one of the following cases:
+	 * 1. For FSCTL_VALIDATE_NEGOTIATE_INFO IOCTLs
+	 * 2. For FSCTL_QUERY_NETWORK_INTERFACE_INFO IOCTL when called from
+	 * smb2_reconnect (indicated by CIFS_SES_FLAG_SCALE_CHANNELS ses flag)
+	 */
+	if (opcode == FSCTL_VALIDATE_NEGOTIATE_INFO ||
+	    (opcode == FSCTL_QUERY_NETWORK_INTERFACE_INFO &&
+	     (tcon->ses->flags & CIFS_SES_FLAGS_PENDING_QUERY_INTERFACES)))
 		return __smb2_plain_req_init(SMB2_IOCTL, tcon, server,
 					     request_buf, total_len);
-	}
+
 	return smb2_plain_req_init(SMB2_IOCTL, tcon, server,
 				   request_buf, total_len);
 }
@@ -4449,10 +4463,10 @@ smb2_new_read_req(void **buf, unsigned int *total_len,
 #ifdef CONFIG_CIFS_SMB_DIRECT
 	/*
 	 * If we want to do a RDMA write, fill in and append
-	 * smbd_buffer_descriptor_v1 to the end of read request
+	 * smbdirect_buffer_descriptor_v1 to the end of read request
 	 */
 	if (rdata && smb3_use_rdma_offload(io_parms)) {
-		struct smbd_buffer_descriptor_v1 *v1;
+		struct smbdirect_buffer_descriptor_v1 *v1;
 		bool need_invalidate = server->dialect == SMB30_PROT_ID;
 
 		rdata->mr = smbd_register_mr(server->smbd_conn, &rdata->subreq.io_iter,
@@ -4466,8 +4480,8 @@ smb2_new_read_req(void **buf, unsigned int *total_len,
 		req->ReadChannelInfoOffset =
 			cpu_to_le16(offsetof(struct smb2_read_req, Buffer));
 		req->ReadChannelInfoLength =
-			cpu_to_le16(sizeof(struct smbd_buffer_descriptor_v1));
+			cpu_to_le16(sizeof(struct smbdirect_buffer_descriptor_v1));
-		v1 = (struct smbd_buffer_descriptor_v1 *) &req->Buffer[0];
+		v1 = (struct smbdirect_buffer_descriptor_v1 *) &req->Buffer[0];
 		v1->offset = cpu_to_le64(rdata->mr->mr->iova);
 		v1->token = cpu_to_le32(rdata->mr->mr->rkey);
 		v1->length = cpu_to_le32(rdata->mr->mr->length);
@@ -4975,10 +4989,10 @@ smb2_async_writev(struct cifs_io_subrequest *wdata)
 #ifdef CONFIG_CIFS_SMB_DIRECT
 	/*
 	 * If we want to do a server RDMA read, fill in and append
-	 * smbd_buffer_descriptor_v1 to the end of write request
+	 * smbdirect_buffer_descriptor_v1 to the end of write request
 	 */
 	if (smb3_use_rdma_offload(io_parms)) {
-		struct smbd_buffer_descriptor_v1 *v1;
+		struct smbdirect_buffer_descriptor_v1 *v1;
 		bool need_invalidate = server->dialect == SMB30_PROT_ID;
 
 		wdata->mr = smbd_register_mr(server->smbd_conn, &wdata->subreq.io_iter,
@@ -4997,8 +5011,8 @@ smb2_async_writev(struct cifs_io_subrequest *wdata)
 		req->WriteChannelInfoOffset =
 			cpu_to_le16(offsetof(struct smb2_write_req, Buffer));
 		req->WriteChannelInfoLength =
-			cpu_to_le16(sizeof(struct smbd_buffer_descriptor_v1));
+			cpu_to_le16(sizeof(struct smbdirect_buffer_descriptor_v1));
-		v1 = (struct smbd_buffer_descriptor_v1 *) &req->Buffer[0];
+		v1 = (struct smbdirect_buffer_descriptor_v1 *) &req->Buffer[0];
 		v1->offset = cpu_to_le64(wdata->mr->mr->iova);
 		v1->token = cpu_to_le32(wdata->mr->mr->rkey);
 		v1->length = cpu_to_le32(wdata->mr->mr->length);

fs/smb/client/smbdirect.h

@@ -15,6 +15,9 @@
 #include <rdma/rdma_cm.h>
 #include <linux/mempool.h>
 
+#include "../common/smbdirect/smbdirect.h"
+#include "../common/smbdirect/smbdirect_socket.h"
+
 extern int rdma_readwrite_threshold;
 extern int smbd_max_frmr_depth;
 extern int smbd_keep_alive_interval;
@@ -50,14 +53,8 @@ enum smbd_connection_status {
  * 5. mempools for allocating packets
  */
 struct smbd_connection {
-	enum smbd_connection_status transport_status;
+	struct smbdirect_socket socket;
 
-	/* RDMA related */
-	struct rdma_cm_id *id;
-	struct ib_qp_init_attr qp_attr;
-	struct ib_pd *pd;
-	struct ib_cq *send_cq, *recv_cq;
-	struct ib_device_attr dev_attr;
 	int ri_rc;
 	struct completion ri_done;
 	wait_queue_head_t conn_wait;
@@ -72,15 +69,7 @@ struct smbd_connection {
 	spinlock_t lock_new_credits_offered;
 	int new_credits_offered;
 
-	/* Connection parameters defined in [MS-SMBD] 3.1.1.1 */
+	/* dynamic connection parameters defined in [MS-SMBD] 3.1.1.1 */
-	int receive_credit_max;
-	int send_credit_target;
-	int max_send_size;
-	int max_fragmented_recv_size;
-	int max_fragmented_send_size;
-	int max_receive_size;
-	int keep_alive_interval;
-	int max_readwrite_size;
 	enum keep_alive_status keep_alive_requested;
 	int protocol;
 	atomic_t send_credits;
@@ -177,54 +166,6 @@ enum smbd_message_type {
 	SMBD_TRANSFER_DATA,
 };
 
-#define SMB_DIRECT_RESPONSE_REQUESTED 0x0001
-
-/* SMBD negotiation request packet [MS-SMBD] 2.2.1 */
-struct smbd_negotiate_req {
-	__le16 min_version;
-	__le16 max_version;
-	__le16 reserved;
-	__le16 credits_requested;
-	__le32 preferred_send_size;
-	__le32 max_receive_size;
-	__le32 max_fragmented_size;
-} __packed;
-
-/* SMBD negotiation response packet [MS-SMBD] 2.2.2 */
-struct smbd_negotiate_resp {
-	__le16 min_version;
-	__le16 max_version;
-	__le16 negotiated_version;
-	__le16 reserved;
-	__le16 credits_requested;
-	__le16 credits_granted;
-	__le32 status;
-	__le32 max_readwrite_size;
-	__le32 preferred_send_size;
-	__le32 max_receive_size;
-	__le32 max_fragmented_size;
-} __packed;
-
-/* SMBD data transfer packet with payload [MS-SMBD] 2.2.3 */
-struct smbd_data_transfer {
-	__le16 credits_requested;
-	__le16 credits_granted;
-	__le16 flags;
-	__le16 reserved;
-	__le32 remaining_data_length;
-	__le32 data_offset;
-	__le32 data_length;
-	__le32 padding;
-	__u8 buffer[];
-} __packed;
-
-/* The packet fields for a registered RDMA buffer */
-struct smbd_buffer_descriptor_v1 {
-	__le64 offset;
-	__le32 token;
-	__le32 length;
-} __packed;
-
 /* Maximum number of SGEs used by smbdirect.c in any send work request */
 #define SMBDIRECT_MAX_SEND_SGE	6
 

fs/smb/client/transport.c

@@ -1018,14 +1018,16 @@ struct TCP_Server_Info *cifs_pick_channel(struct cifs_ses *ses)
 	uint index = 0;
 	unsigned int min_in_flight = UINT_MAX, max_in_flight = 0;
 	struct TCP_Server_Info *server = NULL;
-	int i;
+	int i, start, cur;
 
 	if (!ses)
 		return NULL;
 
 	spin_lock(&ses->chan_lock);
+	start = atomic_inc_return(&ses->chan_seq);
 	for (i = 0; i < ses->chan_count; i++) {
-		server = ses->chans[i].server;
+		cur = (start + i) % ses->chan_count;
+		server = ses->chans[cur].server;
 		if (!server || server->terminate)
 			continue;
 
@@ -1042,17 +1044,15 @@ struct TCP_Server_Info *cifs_pick_channel(struct cifs_ses *ses)
 		 */
 		if (server->in_flight < min_in_flight) {
 			min_in_flight = server->in_flight;
-			index = i;
+			index = cur;
 		}
 		if (server->in_flight > max_in_flight)
 			max_in_flight = server->in_flight;
 	}
 
 	/* if all channels are equally loaded, fall back to round-robin */
-	if (min_in_flight == max_in_flight) {
+	if (min_in_flight == max_in_flight)
-		index = (uint)atomic_inc_return(&ses->chan_seq);
+		index = (uint)start % ses->chan_count;
-		index %= ses->chan_count;
-	}
 
 	server = ses->chans[index].server;
 	spin_unlock(&ses->chan_lock);

fs/smb/common/smbdirect/smbdirect.h

@@ -0,0 +1,37 @@
+/* SPDX-License-Identifier: GPL-2.0-or-later */
+/*
+ *   Copyright (C) 2017, Microsoft Corporation.
+ *   Copyright (C) 2018, LG Electronics.
+ */
+
+#ifndef __FS_SMB_COMMON_SMBDIRECT_SMBDIRECT_H__
+#define __FS_SMB_COMMON_SMBDIRECT_SMBDIRECT_H__
+
+/* SMB-DIRECT buffer descriptor V1 structure [MS-SMBD] 2.2.3.1 */
+struct smbdirect_buffer_descriptor_v1 {
+	__le64 offset;
+	__le32 token;
+	__le32 length;
+} __packed;
+
+/*
+ * Connection parameters mostly from [MS-SMBD] 3.1.1.1
+ *
+ * These are setup and negotiated at the beginning of a
+ * connection and remain constant unless explicitly changed.
+ *
+ * Some values are important for the upper layer.
+ */
+struct smbdirect_socket_parameters {
+	__u16 recv_credit_max;
+	__u16 send_credit_target;
+	__u32 max_send_size;
+	__u32 max_fragmented_send_size;
+	__u32 max_recv_size;
+	__u32 max_fragmented_recv_size;
+	__u32 max_read_write_size;
+	__u32 keepalive_interval_msec;
+	__u32 keepalive_timeout_msec;
+} __packed;
+
+#endif /* __FS_SMB_COMMON_SMBDIRECT_SMBDIRECT_H__ */

fs/smb/common/smbdirect/smbdirect_pdu.h

@@ -0,0 +1,55 @@
+/* SPDX-License-Identifier: GPL-2.0-or-later */
+/*
+ *   Copyright (c) 2017 Stefan Metzmacher
+ */
+
+#ifndef __FS_SMB_COMMON_SMBDIRECT_SMBDIRECT_PDU_H__
+#define __FS_SMB_COMMON_SMBDIRECT_SMBDIRECT_PDU_H__
+
+#define SMBDIRECT_V1 0x0100
+
+/* SMBD negotiation request packet [MS-SMBD] 2.2.1 */
+struct smbdirect_negotiate_req {
+	__le16 min_version;
+	__le16 max_version;
+	__le16 reserved;
+	__le16 credits_requested;
+	__le32 preferred_send_size;
+	__le32 max_receive_size;
+	__le32 max_fragmented_size;
+} __packed;
+
+/* SMBD negotiation response packet [MS-SMBD] 2.2.2 */
+struct smbdirect_negotiate_resp {
+	__le16 min_version;
+	__le16 max_version;
+	__le16 negotiated_version;
+	__le16 reserved;
+	__le16 credits_requested;
+	__le16 credits_granted;
+	__le32 status;
+	__le32 max_readwrite_size;
+	__le32 preferred_send_size;
+	__le32 max_receive_size;
+	__le32 max_fragmented_size;
+} __packed;
+
+#define SMBDIRECT_DATA_MIN_HDR_SIZE 0x14
+#define SMBDIRECT_DATA_OFFSET       0x18
+
+#define SMBDIRECT_FLAG_RESPONSE_REQUESTED 0x0001
+
+/* SMBD data transfer packet with payload [MS-SMBD] 2.2.3 */
+struct smbdirect_data_transfer {
+	__le16 credits_requested;
+	__le16 credits_granted;
+	__le16 flags;
+	__le16 reserved;
+	__le32 remaining_data_length;
+	__le32 data_offset;
+	__le32 data_length;
+	__le32 padding;
+	__u8 buffer[];
+} __packed;
+
+#endif /* __FS_SMB_COMMON_SMBDIRECT_SMBDIRECT_PDU_H__ */

fs/smb/common/smbdirect/smbdirect_socket.h

@@ -0,0 +1,43 @@
+/* SPDX-License-Identifier: GPL-2.0-or-later */
+/*
+ *   Copyright (c) 2025 Stefan Metzmacher
+ */
+
+#ifndef __FS_SMB_COMMON_SMBDIRECT_SMBDIRECT_SOCKET_H__
+#define __FS_SMB_COMMON_SMBDIRECT_SMBDIRECT_SOCKET_H__
+
+enum smbdirect_socket_status {
+	SMBDIRECT_SOCKET_CREATED,
+	SMBDIRECT_SOCKET_CONNECTING,
+	SMBDIRECT_SOCKET_CONNECTED,
+	SMBDIRECT_SOCKET_NEGOTIATE_FAILED,
+	SMBDIRECT_SOCKET_DISCONNECTING,
+	SMBDIRECT_SOCKET_DISCONNECTED,
+	SMBDIRECT_SOCKET_DESTROYED
+};
+
+struct smbdirect_socket {
+	enum smbdirect_socket_status status;
+
+	/* RDMA related */
+	struct {
+		struct rdma_cm_id *cm_id;
+	} rdma;
+
+	/* IB verbs related */
+	struct {
+		struct ib_pd *pd;
+		struct ib_cq *send_cq;
+		struct ib_cq *recv_cq;
+
+		/*
+		 * shortcuts for rdma.cm_id->{qp,device};
+		 */
+		struct ib_qp *qp;
+		struct ib_device *dev;
+	} ib;
+
+	struct smbdirect_socket_parameters parameters;
+};
+
+#endif /* __FS_SMB_COMMON_SMBDIRECT_SMBDIRECT_SOCKET_H__ */