Fixed MTP to work with TWRP

net/9p/Kconfig

@@ -0,0 +1,36 @@
+#
+# 9P protocol configuration
+#
+
+menuconfig NET_9P
+	depends on NET
+	tristate "Plan 9 Resource Sharing Support (9P2000)"
+	help
+	  If you say Y here, you will get experimental support for
+	  Plan 9 resource sharing via the 9P2000 protocol.
+
+	  See <http://v9fs.sf.net> for more information.
+
+	  If unsure, say N.
+
+if NET_9P
+
+config NET_9P_VIRTIO
+	depends on VIRTIO
+	tristate "9P Virtio Transport"
+	help
+	  This builds support for a transports between
+	  guest partitions and a host partition.
+
+config NET_9P_RDMA
+	depends on INET && INFINIBAND && INFINIBAND_ADDR_TRANS
+	tristate "9P RDMA Transport (Experimental)"
+	help
+	  This builds support for an RDMA transport.
+
+config NET_9P_DEBUG
+	bool "Debug information"
+	help
+	  Say Y if you want the 9P subsystem to log debug information.
+
+endif

net/9p/Makefile

@@ -0,0 +1,18 @@
+obj-$(CONFIG_NET_9P) := 9pnet.o
+obj-$(CONFIG_NET_9P_VIRTIO) += 9pnet_virtio.o
+obj-$(CONFIG_NET_9P_RDMA) += 9pnet_rdma.o
+
+9pnet-objs := \
+	mod.o \
+	client.o \
+	error.o \
+	util.o \
+	protocol.o \
+	trans_fd.o \
+	trans_common.o \
+
+9pnet_virtio-objs := \
+	trans_virtio.o \
+
+9pnet_rdma-objs := \
+	trans_rdma.o \

net/9p/error.c

@@ -0,0 +1,247 @@
+/*
+ * linux/fs/9p/error.c
+ *
+ * Error string handling
+ *
+ * Plan 9 uses error strings, Unix uses error numbers.  These functions
+ * try to help manage that and provide for dynamically adding error
+ * mappings.
+ *
+ *  Copyright (C) 2004 by Eric Van Hensbergen <ericvh@gmail.com>
+ *  Copyright (C) 2002 by Ron Minnich <rminnich@lanl.gov>
+ *
+ *  This program is free software; you can redistribute it and/or modify
+ *  it under the terms of the GNU General Public License version 2
+ *  as published by the Free Software Foundation.
+ *
+ *  This program is distributed in the hope that it will be useful,
+ *  but WITHOUT ANY WARRANTY; without even the implied warranty of
+ *  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ *  GNU General Public License for more details.
+ *
+ *  You should have received a copy of the GNU General Public License
+ *  along with this program; if not, write to:
+ *  Free Software Foundation
+ *  51 Franklin Street, Fifth Floor
+ *  Boston, MA  02111-1301  USA
+ *
+ */
+
+#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
+
+#include <linux/module.h>
+#include <linux/list.h>
+#include <linux/jhash.h>
+#include <linux/errno.h>
+#include <net/9p/9p.h>
+
+/**
+ * struct errormap - map string errors from Plan 9 to Linux numeric ids
+ * @name: string sent over 9P
+ * @val: numeric id most closely representing @name
+ * @namelen: length of string
+ * @list: hash-table list for string lookup
+ */
+struct errormap {
+	char *name;
+	int val;
+
+	int namelen;
+	struct hlist_node list;
+};
+
+#define ERRHASHSZ		32
+static struct hlist_head hash_errmap[ERRHASHSZ];
+
+/* FixMe - reduce to a reasonable size */
+static struct errormap errmap[] = {
+	{"Operation not permitted", EPERM},
+	{"wstat prohibited", EPERM},
+	{"No such file or directory", ENOENT},
+	{"directory entry not found", ENOENT},
+	{"file not found", ENOENT},
+	{"Interrupted system call", EINTR},
+	{"Input/output error", EIO},
+	{"No such device or address", ENXIO},
+	{"Argument list too long", E2BIG},
+	{"Bad file descriptor", EBADF},
+	{"Resource temporarily unavailable", EAGAIN},
+	{"Cannot allocate memory", ENOMEM},
+	{"Permission denied", EACCES},
+	{"Bad address", EFAULT},
+	{"Block device required", ENOTBLK},
+	{"Device or resource busy", EBUSY},
+	{"File exists", EEXIST},
+	{"Invalid cross-device link", EXDEV},
+	{"No such device", ENODEV},
+	{"Not a directory", ENOTDIR},
+	{"Is a directory", EISDIR},
+	{"Invalid argument", EINVAL},
+	{"Too many open files in system", ENFILE},
+	{"Too many open files", EMFILE},
+	{"Text file busy", ETXTBSY},
+	{"File too large", EFBIG},
+	{"No space left on device", ENOSPC},
+	{"Illegal seek", ESPIPE},
+	{"Read-only file system", EROFS},
+	{"Too many links", EMLINK},
+	{"Broken pipe", EPIPE},
+	{"Numerical argument out of domain", EDOM},
+	{"Numerical result out of range", ERANGE},
+	{"Resource deadlock avoided", EDEADLK},
+	{"File name too long", ENAMETOOLONG},
+	{"No locks available", ENOLCK},
+	{"Function not implemented", ENOSYS},
+	{"Directory not empty", ENOTEMPTY},
+	{"Too many levels of symbolic links", ELOOP},
+	{"No message of desired type", ENOMSG},
+	{"Identifier removed", EIDRM},
+	{"No data available", ENODATA},
+	{"Machine is not on the network", ENONET},
+	{"Package not installed", ENOPKG},
+	{"Object is remote", EREMOTE},
+	{"Link has been severed", ENOLINK},
+	{"Communication error on send", ECOMM},
+	{"Protocol error", EPROTO},
+	{"Bad message", EBADMSG},
+	{"File descriptor in bad state", EBADFD},
+	{"Streams pipe error", ESTRPIPE},
+	{"Too many users", EUSERS},
+	{"Socket operation on non-socket", ENOTSOCK},
+	{"Message too long", EMSGSIZE},
+	{"Protocol not available", ENOPROTOOPT},
+	{"Protocol not supported", EPROTONOSUPPORT},
+	{"Socket type not supported", ESOCKTNOSUPPORT},
+	{"Operation not supported", EOPNOTSUPP},
+	{"Protocol family not supported", EPFNOSUPPORT},
+	{"Network is down", ENETDOWN},
+	{"Network is unreachable", ENETUNREACH},
+	{"Network dropped connection on reset", ENETRESET},
+	{"Software caused connection abort", ECONNABORTED},
+	{"Connection reset by peer", ECONNRESET},
+	{"No buffer space available", ENOBUFS},
+	{"Transport endpoint is already connected", EISCONN},
+	{"Transport endpoint is not connected", ENOTCONN},
+	{"Cannot send after transport endpoint shutdown", ESHUTDOWN},
+	{"Connection timed out", ETIMEDOUT},
+	{"Connection refused", ECONNREFUSED},
+	{"Host is down", EHOSTDOWN},
+	{"No route to host", EHOSTUNREACH},
+	{"Operation already in progress", EALREADY},
+	{"Operation now in progress", EINPROGRESS},
+	{"Is a named type file", EISNAM},
+	{"Remote I/O error", EREMOTEIO},
+	{"Disk quota exceeded", EDQUOT},
+/* errors from fossil, vacfs, and u9fs */
+	{"fid unknown or out of range", EBADF},
+	{"permission denied", EACCES},
+	{"file does not exist", ENOENT},
+	{"authentication failed", ECONNREFUSED},
+	{"bad offset in directory read", ESPIPE},
+	{"bad use of fid", EBADF},
+	{"wstat can't convert between files and directories", EPERM},
+	{"directory is not empty", ENOTEMPTY},
+	{"file exists", EEXIST},
+	{"file already exists", EEXIST},
+	{"file or directory already exists", EEXIST},
+	{"fid already in use", EBADF},
+	{"file in use", ETXTBSY},
+	{"i/o error", EIO},
+	{"file already open for I/O", ETXTBSY},
+	{"illegal mode", EINVAL},
+	{"illegal name", ENAMETOOLONG},
+	{"not a directory", ENOTDIR},
+	{"not a member of proposed group", EPERM},
+	{"not owner", EACCES},
+	{"only owner can change group in wstat", EACCES},
+	{"read only file system", EROFS},
+	{"no access to special file", EPERM},
+	{"i/o count too large", EIO},
+	{"unknown group", EINVAL},
+	{"unknown user", EINVAL},
+	{"bogus wstat buffer", EPROTO},
+	{"exclusive use file already open", EAGAIN},
+	{"corrupted directory entry", EIO},
+	{"corrupted file entry", EIO},
+	{"corrupted block label", EIO},
+	{"corrupted meta data", EIO},
+	{"illegal offset", EINVAL},
+	{"illegal path element", ENOENT},
+	{"root of file system is corrupted", EIO},
+	{"corrupted super block", EIO},
+	{"protocol botch", EPROTO},
+	{"file system is full", ENOSPC},
+	{"file is in use", EAGAIN},
+	{"directory entry is not allocated", ENOENT},
+	{"file is read only", EROFS},
+	{"file has been removed", EIDRM},
+	{"only support truncation to zero length", EPERM},
+	{"cannot remove root", EPERM},
+	{"file too big", EFBIG},
+	{"venti i/o error", EIO},
+	/* these are not errors */
+	{"u9fs rhostsauth: no authentication required", 0},
+	{"u9fs authnone: no authentication required", 0},
+	{NULL, -1}
+};
+
+/**
+ * p9_error_init - preload mappings into hash list
+ *
+ */
+
+int p9_error_init(void)
+{
+	struct errormap *c;
+	int bucket;
+
+	/* initialize hash table */
+	for (bucket = 0; bucket < ERRHASHSZ; bucket++)
+		INIT_HLIST_HEAD(&hash_errmap[bucket]);
+
+	/* load initial error map into hash table */
+	for (c = errmap; c->name != NULL; c++) {
+		c->namelen = strlen(c->name);
+		bucket = jhash(c->name, c->namelen, 0) % ERRHASHSZ;
+		INIT_HLIST_NODE(&c->list);
+		hlist_add_head(&c->list, &hash_errmap[bucket]);
+	}
+
+	return 1;
+}
+EXPORT_SYMBOL(p9_error_init);
+
+/**
+ * errstr2errno - convert error string to error number
+ * @errstr: error string
+ * @len: length of error string
+ *
+ */
+
+int p9_errstr2errno(char *errstr, int len)
+{
+	int errno;
+	struct errormap *c;
+	int bucket;
+
+	errno = 0;
+	c = NULL;
+	bucket = jhash(errstr, len, 0) % ERRHASHSZ;
+	hlist_for_each_entry(c, &hash_errmap[bucket], list) {
+		if (c->namelen == len && !memcmp(c->name, errstr, len)) {
+			errno = c->val;
+			break;
+		}
+	}
+
+	if (errno == 0) {
+		/* TODO: if error isn't found, add it dynamically */
+		errstr[len] = 0;
+		pr_err("%s: server reported unknown error %s\n",
+		       __func__, errstr);
+		errno = ESERVERFAULT;
+	}
+
+	return -errno;
+}
+EXPORT_SYMBOL(p9_errstr2errno);

net/9p/mod.c

@@ -0,0 +1,201 @@
+/*
+ *  net/9p/9p.c
+ *
+ *  9P entry point
+ *
+ *  Copyright (C) 2007 by Latchesar Ionkov <lucho@ionkov.net>
+ *  Copyright (C) 2004 by Eric Van Hensbergen <ericvh@gmail.com>
+ *  Copyright (C) 2002 by Ron Minnich <rminnich@lanl.gov>
+ *
+ *  This program is free software; you can redistribute it and/or modify
+ *  it under the terms of the GNU General Public License version 2
+ *  as published by the Free Software Foundation.
+ *
+ *  This program is distributed in the hope that it will be useful,
+ *  but WITHOUT ANY WARRANTY; without even the implied warranty of
+ *  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ *  GNU General Public License for more details.
+ *
+ *  You should have received a copy of the GNU General Public License
+ *  along with this program; if not, write to:
+ *  Free Software Foundation
+ *  51 Franklin Street, Fifth Floor
+ *  Boston, MA  02111-1301  USA
+ *
+ */
+
+#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
+
+#include <linux/module.h>
+#include <linux/errno.h>
+#include <linux/sched.h>
+#include <linux/moduleparam.h>
+#include <net/9p/9p.h>
+#include <linux/fs.h>
+#include <linux/parser.h>
+#include <net/9p/client.h>
+#include <net/9p/transport.h>
+#include <linux/list.h>
+#include <linux/spinlock.h>
+
+#ifdef CONFIG_NET_9P_DEBUG
+unsigned int p9_debug_level = 0;	/* feature-rific global debug level  */
+EXPORT_SYMBOL(p9_debug_level);
+module_param_named(debug, p9_debug_level, uint, 0);
+MODULE_PARM_DESC(debug, "9P debugging level");
+
+void _p9_debug(enum p9_debug_flags level, const char *func,
+		const char *fmt, ...)
+{
+	struct va_format vaf;
+	va_list args;
+
+	if ((p9_debug_level & level) != level)
+		return;
+
+	va_start(args, fmt);
+
+	vaf.fmt = fmt;
+	vaf.va = &args;
+
+	if (level == P9_DEBUG_9P)
+		pr_notice("(%8.8d) %pV", task_pid_nr(current), &vaf);
+	else
+		pr_notice("-- %s (%d): %pV", func, task_pid_nr(current), &vaf);
+
+	va_end(args);
+}
+EXPORT_SYMBOL(_p9_debug);
+#endif
+
+/*
+ * Dynamic Transport Registration Routines
+ *
+ */
+
+static DEFINE_SPINLOCK(v9fs_trans_lock);
+static LIST_HEAD(v9fs_trans_list);
+
+/**
+ * v9fs_register_trans - register a new transport with 9p
+ * @m: structure describing the transport module and entry points
+ *
+ */
+void v9fs_register_trans(struct p9_trans_module *m)
+{
+	spin_lock(&v9fs_trans_lock);
+	list_add_tail(&m->list, &v9fs_trans_list);
+	spin_unlock(&v9fs_trans_lock);
+}
+EXPORT_SYMBOL(v9fs_register_trans);
+
+/**
+ * v9fs_unregister_trans - unregister a 9p transport
+ * @m: the transport to remove
+ *
+ */
+void v9fs_unregister_trans(struct p9_trans_module *m)
+{
+	spin_lock(&v9fs_trans_lock);
+	list_del_init(&m->list);
+	spin_unlock(&v9fs_trans_lock);
+}
+EXPORT_SYMBOL(v9fs_unregister_trans);
+
+/**
+ * v9fs_get_trans_by_name - get transport with the matching name
+ * @name: string identifying transport
+ *
+ */
+struct p9_trans_module *v9fs_get_trans_by_name(char *s)
+{
+	struct p9_trans_module *t, *found = NULL;
+
+	spin_lock(&v9fs_trans_lock);
+
+	list_for_each_entry(t, &v9fs_trans_list, list)
+		if (strcmp(t->name, s) == 0 &&
+		    try_module_get(t->owner)) {
+			found = t;
+			break;
+		}
+
+	spin_unlock(&v9fs_trans_lock);
+	return found;
+}
+EXPORT_SYMBOL(v9fs_get_trans_by_name);
+
+/**
+ * v9fs_get_default_trans - get the default transport
+ *
+ */
+
+struct p9_trans_module *v9fs_get_default_trans(void)
+{
+	struct p9_trans_module *t, *found = NULL;
+
+	spin_lock(&v9fs_trans_lock);
+
+	list_for_each_entry(t, &v9fs_trans_list, list)
+		if (t->def && try_module_get(t->owner)) {
+			found = t;
+			break;
+		}
+
+	if (!found)
+		list_for_each_entry(t, &v9fs_trans_list, list)
+			if (try_module_get(t->owner)) {
+				found = t;
+				break;
+			}
+
+	spin_unlock(&v9fs_trans_lock);
+	return found;
+}
+EXPORT_SYMBOL(v9fs_get_default_trans);
+
+/**
+ * v9fs_put_trans - put trans
+ * @m: transport to put
+ *
+ */
+void v9fs_put_trans(struct p9_trans_module *m)
+{
+	if (m)
+		module_put(m->owner);
+}
+
+/**
+ * init_p9 - Initialize module
+ *
+ */
+static int __init init_p9(void)
+{
+	int ret = 0;
+
+	p9_error_init();
+	pr_info("Installing 9P2000 support\n");
+	p9_trans_fd_init();
+
+	return ret;
+}
+
+/**
+ * exit_p9 - shutdown module
+ *
+ */
+
+static void __exit exit_p9(void)
+{
+	pr_info("Unloading 9P2000 support\n");
+
+	p9_trans_fd_exit();
+}
+
+module_init(init_p9)
+module_exit(exit_p9)
+
+MODULE_AUTHOR("Latchesar Ionkov <lucho@ionkov.net>");
+MODULE_AUTHOR("Eric Van Hensbergen <ericvh@gmail.com>");
+MODULE_AUTHOR("Ron Minnich <rminnich@lanl.gov>");
+MODULE_LICENSE("GPL");

net/9p/protocol.c

@@ -0,0 +1,636 @@
+/*
+ * net/9p/protocol.c
+ *
+ * 9P Protocol Support Code
+ *
+ *  Copyright (C) 2008 by Eric Van Hensbergen <ericvh@gmail.com>
+ *
+ *  Base on code from Anthony Liguori <aliguori@us.ibm.com>
+ *  Copyright (C) 2008 by IBM, Corp.
+ *
+ *  This program is free software; you can redistribute it and/or modify
+ *  it under the terms of the GNU General Public License version 2
+ *  as published by the Free Software Foundation.
+ *
+ *  This program is distributed in the hope that it will be useful,
+ *  but WITHOUT ANY WARRANTY; without even the implied warranty of
+ *  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ *  GNU General Public License for more details.
+ *
+ *  You should have received a copy of the GNU General Public License
+ *  along with this program; if not, write to:
+ *  Free Software Foundation
+ *  51 Franklin Street, Fifth Floor
+ *  Boston, MA  02111-1301  USA
+ *
+ */
+
+#include <linux/module.h>
+#include <linux/errno.h>
+#include <linux/kernel.h>
+#include <linux/uaccess.h>
+#include <linux/slab.h>
+#include <linux/sched.h>
+#include <linux/stddef.h>
+#include <linux/types.h>
+#include <net/9p/9p.h>
+#include <net/9p/client.h>
+#include "protocol.h"
+
+#include <trace/events/9p.h>
+
+static int
+p9pdu_writef(struct p9_fcall *pdu, int proto_version, const char *fmt, ...);
+
+void p9stat_free(struct p9_wstat *stbuf)
+{
+	kfree(stbuf->name);
+	kfree(stbuf->uid);
+	kfree(stbuf->gid);
+	kfree(stbuf->muid);
+	kfree(stbuf->extension);
+}
+EXPORT_SYMBOL(p9stat_free);
+
+size_t pdu_read(struct p9_fcall *pdu, void *data, size_t size)
+{
+	size_t len = min(pdu->size - pdu->offset, size);
+	memcpy(data, &pdu->sdata[pdu->offset], len);
+	pdu->offset += len;
+	return size - len;
+}
+
+static size_t pdu_write(struct p9_fcall *pdu, const void *data, size_t size)
+{
+	size_t len = min(pdu->capacity - pdu->size, size);
+	memcpy(&pdu->sdata[pdu->size], data, len);
+	pdu->size += len;
+	return size - len;
+}
+
+static size_t
+pdu_write_u(struct p9_fcall *pdu, const char __user *udata, size_t size)
+{
+	size_t len = min(pdu->capacity - pdu->size, size);
+	if (copy_from_user(&pdu->sdata[pdu->size], udata, len))
+		len = 0;
+
+	pdu->size += len;
+	return size - len;
+}
+
+/*
+	b - int8_t
+	w - int16_t
+	d - int32_t
+	q - int64_t
+	s - string
+	u - numeric uid
+	g - numeric gid
+	S - stat
+	Q - qid
+	D - data blob (int32_t size followed by void *, results are not freed)
+	T - array of strings (int16_t count, followed by strings)
+	R - array of qids (int16_t count, followed by qids)
+	A - stat for 9p2000.L (p9_stat_dotl)
+	? - if optional = 1, continue parsing
+*/
+
+static int
+p9pdu_vreadf(struct p9_fcall *pdu, int proto_version, const char *fmt,
+	va_list ap)
+{
+	const char *ptr;
+	int errcode = 0;
+
+	for (ptr = fmt; *ptr; ptr++) {
+		switch (*ptr) {
+		case 'b':{
+				int8_t *val = va_arg(ap, int8_t *);
+				if (pdu_read(pdu, val, sizeof(*val))) {
+					errcode = -EFAULT;
+					break;
+				}
+			}
+			break;
+		case 'w':{
+				int16_t *val = va_arg(ap, int16_t *);
+				__le16 le_val;
+				if (pdu_read(pdu, &le_val, sizeof(le_val))) {
+					errcode = -EFAULT;
+					break;
+				}
+				*val = le16_to_cpu(le_val);
+			}
+			break;
+		case 'd':{
+				int32_t *val = va_arg(ap, int32_t *);
+				__le32 le_val;
+				if (pdu_read(pdu, &le_val, sizeof(le_val))) {
+					errcode = -EFAULT;
+					break;
+				}
+				*val = le32_to_cpu(le_val);
+			}
+			break;
+		case 'q':{
+				int64_t *val = va_arg(ap, int64_t *);
+				__le64 le_val;
+				if (pdu_read(pdu, &le_val, sizeof(le_val))) {
+					errcode = -EFAULT;
+					break;
+				}
+				*val = le64_to_cpu(le_val);
+			}
+			break;
+		case 's':{
+				char **sptr = va_arg(ap, char **);
+				uint16_t len;
+
+				errcode = p9pdu_readf(pdu, proto_version,
+								"w", &len);
+				if (errcode)
+					break;
+
+				*sptr = kmalloc(len + 1, GFP_NOFS);
+				if (*sptr == NULL) {
+					errcode = -EFAULT;
+					break;
+				}
+				if (pdu_read(pdu, *sptr, len)) {
+					errcode = -EFAULT;
+					kfree(*sptr);
+					*sptr = NULL;
+				} else
+					(*sptr)[len] = 0;
+			}
+			break;
+		case 'u': {
+				kuid_t *uid = va_arg(ap, kuid_t *);
+				__le32 le_val;
+				if (pdu_read(pdu, &le_val, sizeof(le_val))) {
+					errcode = -EFAULT;
+					break;
+				}
+				*uid = make_kuid(&init_user_ns,
+						 le32_to_cpu(le_val));
+			} break;
+		case 'g': {
+				kgid_t *gid = va_arg(ap, kgid_t *);
+				__le32 le_val;
+				if (pdu_read(pdu, &le_val, sizeof(le_val))) {
+					errcode = -EFAULT;
+					break;
+				}
+				*gid = make_kgid(&init_user_ns,
+						 le32_to_cpu(le_val));
+			} break;
+		case 'Q':{
+				struct p9_qid *qid =
+				    va_arg(ap, struct p9_qid *);
+
+				errcode = p9pdu_readf(pdu, proto_version, "bdq",
+						      &qid->type, &qid->version,
+						      &qid->path);
+			}
+			break;
+		case 'S':{
+				struct p9_wstat *stbuf =
+				    va_arg(ap, struct p9_wstat *);
+
+				memset(stbuf, 0, sizeof(struct p9_wstat));
+				stbuf->n_uid = stbuf->n_muid = INVALID_UID;
+				stbuf->n_gid = INVALID_GID;
+
+				errcode =
+				    p9pdu_readf(pdu, proto_version,
+						"wwdQdddqssss?sugu",
+						&stbuf->size, &stbuf->type,
+						&stbuf->dev, &stbuf->qid,
+						&stbuf->mode, &stbuf->atime,
+						&stbuf->mtime, &stbuf->length,
+						&stbuf->name, &stbuf->uid,
+						&stbuf->gid, &stbuf->muid,
+						&stbuf->extension,
+						&stbuf->n_uid, &stbuf->n_gid,
+						&stbuf->n_muid);
+				if (errcode)
+					p9stat_free(stbuf);
+			}
+			break;
+		case 'D':{
+				uint32_t *count = va_arg(ap, uint32_t *);
+				void **data = va_arg(ap, void **);
+
+				errcode =
+				    p9pdu_readf(pdu, proto_version, "d", count);
+				if (!errcode) {
+					*count =
+					    min_t(uint32_t, *count,
+						  pdu->size - pdu->offset);
+					*data = &pdu->sdata[pdu->offset];
+				}
+			}
+			break;
+		case 'T':{
+				uint16_t *nwname = va_arg(ap, uint16_t *);
+				char ***wnames = va_arg(ap, char ***);
+
+				errcode = p9pdu_readf(pdu, proto_version,
+								"w", nwname);
+				if (!errcode) {
+					*wnames =
+					    kmalloc(sizeof(char *) * *nwname,
+						    GFP_NOFS);
+					if (!*wnames)
+						errcode = -ENOMEM;
+				}
+
+				if (!errcode) {
+					int i;
+
+					for (i = 0; i < *nwname; i++) {
+						errcode =
+						    p9pdu_readf(pdu,
+								proto_version,
+								"s",
+								&(*wnames)[i]);
+						if (errcode)
+							break;
+					}
+				}
+
+				if (errcode) {
+					if (*wnames) {
+						int i;
+
+						for (i = 0; i < *nwname; i++)
+							kfree((*wnames)[i]);
+					}
+					kfree(*wnames);
+					*wnames = NULL;
+				}
+			}
+			break;
+		case 'R':{
+				int16_t *nwqid = va_arg(ap, int16_t *);
+				struct p9_qid **wqids =
+				    va_arg(ap, struct p9_qid **);
+
+				*wqids = NULL;
+
+				errcode =
+				    p9pdu_readf(pdu, proto_version, "w", nwqid);
+				if (!errcode) {
+					*wqids =
+					    kmalloc(*nwqid *
+						    sizeof(struct p9_qid),
+						    GFP_NOFS);
+					if (*wqids == NULL)
+						errcode = -ENOMEM;
+				}
+
+				if (!errcode) {
+					int i;
+
+					for (i = 0; i < *nwqid; i++) {
+						errcode =
+						    p9pdu_readf(pdu,
+								proto_version,
+								"Q",
+								&(*wqids)[i]);
+						if (errcode)
+							break;
+					}
+				}
+
+				if (errcode) {
+					kfree(*wqids);
+					*wqids = NULL;
+				}
+			}
+			break;
+		case 'A': {
+				struct p9_stat_dotl *stbuf =
+				    va_arg(ap, struct p9_stat_dotl *);
+
+				memset(stbuf, 0, sizeof(struct p9_stat_dotl));
+				errcode =
+				    p9pdu_readf(pdu, proto_version,
+					"qQdugqqqqqqqqqqqqqqq",
+					&stbuf->st_result_mask,
+					&stbuf->qid,
+					&stbuf->st_mode,
+					&stbuf->st_uid, &stbuf->st_gid,
+					&stbuf->st_nlink,
+					&stbuf->st_rdev, &stbuf->st_size,
+					&stbuf->st_blksize, &stbuf->st_blocks,
+					&stbuf->st_atime_sec,
+					&stbuf->st_atime_nsec,
+					&stbuf->st_mtime_sec,
+					&stbuf->st_mtime_nsec,
+					&stbuf->st_ctime_sec,
+					&stbuf->st_ctime_nsec,
+					&stbuf->st_btime_sec,
+					&stbuf->st_btime_nsec,
+					&stbuf->st_gen,
+					&stbuf->st_data_version);
+			}
+			break;
+		case '?':
+			if ((proto_version != p9_proto_2000u) &&
+				(proto_version != p9_proto_2000L))
+				return 0;
+			break;
+		default:
+			BUG();
+			break;
+		}
+
+		if (errcode)
+			break;
+	}
+
+	return errcode;
+}
+
+int
+p9pdu_vwritef(struct p9_fcall *pdu, int proto_version, const char *fmt,
+	va_list ap)
+{
+	const char *ptr;
+	int errcode = 0;
+
+	for (ptr = fmt; *ptr; ptr++) {
+		switch (*ptr) {
+		case 'b':{
+				int8_t val = va_arg(ap, int);
+				if (pdu_write(pdu, &val, sizeof(val)))
+					errcode = -EFAULT;
+			}
+			break;
+		case 'w':{
+				__le16 val = cpu_to_le16(va_arg(ap, int));
+				if (pdu_write(pdu, &val, sizeof(val)))
+					errcode = -EFAULT;
+			}
+			break;
+		case 'd':{
+				__le32 val = cpu_to_le32(va_arg(ap, int32_t));
+				if (pdu_write(pdu, &val, sizeof(val)))
+					errcode = -EFAULT;
+			}
+			break;
+		case 'q':{
+				__le64 val = cpu_to_le64(va_arg(ap, int64_t));
+				if (pdu_write(pdu, &val, sizeof(val)))
+					errcode = -EFAULT;
+			}
+			break;
+		case 's':{
+				const char *sptr = va_arg(ap, const char *);
+				uint16_t len = 0;
+				if (sptr)
+					len = min_t(size_t, strlen(sptr),
+								USHRT_MAX);
+
+				errcode = p9pdu_writef(pdu, proto_version,
+								"w", len);
+				if (!errcode && pdu_write(pdu, sptr, len))
+					errcode = -EFAULT;
+			}
+			break;
+		case 'u': {
+				kuid_t uid = va_arg(ap, kuid_t);
+				__le32 val = cpu_to_le32(
+						from_kuid(&init_user_ns, uid));
+				if (pdu_write(pdu, &val, sizeof(val)))
+					errcode = -EFAULT;
+			} break;
+		case 'g': {
+				kgid_t gid = va_arg(ap, kgid_t);
+				__le32 val = cpu_to_le32(
+						from_kgid(&init_user_ns, gid));
+				if (pdu_write(pdu, &val, sizeof(val)))
+					errcode = -EFAULT;
+			} break;
+		case 'Q':{
+				const struct p9_qid *qid =
+				    va_arg(ap, const struct p9_qid *);
+				errcode =
+				    p9pdu_writef(pdu, proto_version, "bdq",
+						 qid->type, qid->version,
+						 qid->path);
+			} break;
+		case 'S':{
+				const struct p9_wstat *stbuf =
+				    va_arg(ap, const struct p9_wstat *);
+				errcode =
+				    p9pdu_writef(pdu, proto_version,
+						 "wwdQdddqssss?sugu",
+						 stbuf->size, stbuf->type,
+						 stbuf->dev, &stbuf->qid,
+						 stbuf->mode, stbuf->atime,
+						 stbuf->mtime, stbuf->length,
+						 stbuf->name, stbuf->uid,
+						 stbuf->gid, stbuf->muid,
+						 stbuf->extension, stbuf->n_uid,
+						 stbuf->n_gid, stbuf->n_muid);
+			} break;
+		case 'D':{
+				uint32_t count = va_arg(ap, uint32_t);
+				const void *data = va_arg(ap, const void *);
+
+				errcode = p9pdu_writef(pdu, proto_version, "d",
+									count);
+				if (!errcode && pdu_write(pdu, data, count))
+					errcode = -EFAULT;
+			}
+			break;
+		case 'U':{
+				int32_t count = va_arg(ap, int32_t);
+				const char __user *udata =
+						va_arg(ap, const void __user *);
+				errcode = p9pdu_writef(pdu, proto_version, "d",
+									count);
+				if (!errcode && pdu_write_u(pdu, udata, count))
+					errcode = -EFAULT;
+			}
+			break;
+		case 'T':{
+				uint16_t nwname = va_arg(ap, int);
+				const char **wnames = va_arg(ap, const char **);
+
+				errcode = p9pdu_writef(pdu, proto_version, "w",
+									nwname);
+				if (!errcode) {
+					int i;
+
+					for (i = 0; i < nwname; i++) {
+						errcode =
+						    p9pdu_writef(pdu,
+								proto_version,
+								 "s",
+								 wnames[i]);
+						if (errcode)
+							break;
+					}
+				}
+			}
+			break;
+		case 'R':{
+				int16_t nwqid = va_arg(ap, int);
+				struct p9_qid *wqids =
+				    va_arg(ap, struct p9_qid *);
+
+				errcode = p9pdu_writef(pdu, proto_version, "w",
+									nwqid);
+				if (!errcode) {
+					int i;
+
+					for (i = 0; i < nwqid; i++) {
+						errcode =
+						    p9pdu_writef(pdu,
+								proto_version,
+								 "Q",
+								 &wqids[i]);
+						if (errcode)
+							break;
+					}
+				}
+			}
+			break;
+		case 'I':{
+				struct p9_iattr_dotl *p9attr = va_arg(ap,
+							struct p9_iattr_dotl *);
+
+				errcode = p9pdu_writef(pdu, proto_version,
+							"ddugqqqqq",
+							p9attr->valid,
+							p9attr->mode,
+							p9attr->uid,
+							p9attr->gid,
+							p9attr->size,
+							p9attr->atime_sec,
+							p9attr->atime_nsec,
+							p9attr->mtime_sec,
+							p9attr->mtime_nsec);
+			}
+			break;
+		case '?':
+			if ((proto_version != p9_proto_2000u) &&
+				(proto_version != p9_proto_2000L))
+				return 0;
+			break;
+		default:
+			BUG();
+			break;
+		}
+
+		if (errcode)
+			break;
+	}
+
+	return errcode;
+}
+
+int p9pdu_readf(struct p9_fcall *pdu, int proto_version, const char *fmt, ...)
+{
+	va_list ap;
+	int ret;
+
+	va_start(ap, fmt);
+	ret = p9pdu_vreadf(pdu, proto_version, fmt, ap);
+	va_end(ap);
+
+	return ret;
+}
+
+static int
+p9pdu_writef(struct p9_fcall *pdu, int proto_version, const char *fmt, ...)
+{
+	va_list ap;
+	int ret;
+
+	va_start(ap, fmt);
+	ret = p9pdu_vwritef(pdu, proto_version, fmt, ap);
+	va_end(ap);
+
+	return ret;
+}
+
+int p9stat_read(struct p9_client *clnt, char *buf, int len, struct p9_wstat *st)
+{
+	struct p9_fcall fake_pdu;
+	int ret;
+
+	fake_pdu.size = len;
+	fake_pdu.capacity = len;
+	fake_pdu.sdata = buf;
+	fake_pdu.offset = 0;
+
+	ret = p9pdu_readf(&fake_pdu, clnt->proto_version, "S", st);
+	if (ret) {
+		p9_debug(P9_DEBUG_9P, "<<< p9stat_read failed: %d\n", ret);
+		trace_9p_protocol_dump(clnt, &fake_pdu);
+	}
+
+	return ret;
+}
+EXPORT_SYMBOL(p9stat_read);
+
+int p9pdu_prepare(struct p9_fcall *pdu, int16_t tag, int8_t type)
+{
+	pdu->id = type;
+	return p9pdu_writef(pdu, 0, "dbw", 0, type, tag);
+}
+
+int p9pdu_finalize(struct p9_client *clnt, struct p9_fcall *pdu)
+{
+	int size = pdu->size;
+	int err;
+
+	pdu->size = 0;
+	err = p9pdu_writef(pdu, 0, "d", size);
+	pdu->size = size;
+
+	trace_9p_protocol_dump(clnt, pdu);
+	p9_debug(P9_DEBUG_9P, ">>> size=%d type: %d tag: %d\n",
+		 pdu->size, pdu->id, pdu->tag);
+
+	return err;
+}
+
+void p9pdu_reset(struct p9_fcall *pdu)
+{
+	pdu->offset = 0;
+	pdu->size = 0;
+}
+
+int p9dirent_read(struct p9_client *clnt, char *buf, int len,
+		  struct p9_dirent *dirent)
+{
+	struct p9_fcall fake_pdu;
+	int ret;
+	char *nameptr;
+
+	fake_pdu.size = len;
+	fake_pdu.capacity = len;
+	fake_pdu.sdata = buf;
+	fake_pdu.offset = 0;
+
+	ret = p9pdu_readf(&fake_pdu, clnt->proto_version, "Qqbs", &dirent->qid,
+			  &dirent->d_off, &dirent->d_type, &nameptr);
+	if (ret) {
+		p9_debug(P9_DEBUG_9P, "<<< p9dirent_read failed: %d\n", ret);
+		trace_9p_protocol_dump(clnt, &fake_pdu);
+		goto out;
+	}
+
+	strcpy(dirent->d_name, nameptr);
+	kfree(nameptr);
+
+out:
+	return fake_pdu.offset;
+}
+EXPORT_SYMBOL(p9dirent_read);

net/9p/protocol.h

@@ -0,0 +1,34 @@
+/*
+ * net/9p/protocol.h
+ *
+ * 9P Protocol Support Code
+ *
+ *  Copyright (C) 2008 by Eric Van Hensbergen <ericvh@gmail.com>
+ *
+ *  Base on code from Anthony Liguori <aliguori@us.ibm.com>
+ *  Copyright (C) 2008 by IBM, Corp.
+ *
+ *  This program is free software; you can redistribute it and/or modify
+ *  it under the terms of the GNU General Public License version 2
+ *  as published by the Free Software Foundation.
+ *
+ *  This program is distributed in the hope that it will be useful,
+ *  but WITHOUT ANY WARRANTY; without even the implied warranty of
+ *  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ *  GNU General Public License for more details.
+ *
+ *  You should have received a copy of the GNU General Public License
+ *  along with this program; if not, write to:
+ *  Free Software Foundation
+ *  51 Franklin Street, Fifth Floor
+ *  Boston, MA  02111-1301  USA
+ *
+ */
+
+int p9pdu_vwritef(struct p9_fcall *pdu, int proto_version, const char *fmt,
+								va_list ap);
+int p9pdu_readf(struct p9_fcall *pdu, int proto_version, const char *fmt, ...);
+int p9pdu_prepare(struct p9_fcall *pdu, int16_t tag, int8_t type);
+int p9pdu_finalize(struct p9_client *clnt, struct p9_fcall *pdu);
+void p9pdu_reset(struct p9_fcall *pdu);
+size_t pdu_read(struct p9_fcall *pdu, void *data, size_t size);

net/9p/trans_common.c

@@ -0,0 +1,69 @@
+/*
+ * Copyright IBM Corporation, 2010
+ * Author Venkateswararao Jujjuri <jvrao@linux.vnet.ibm.com>
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms of version 2.1 of the GNU Lesser General Public License
+ * as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful, but
+ * WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
+ *
+ */
+
+#include <linux/slab.h>
+#include <linux/module.h>
+#include <net/9p/9p.h>
+#include <net/9p/client.h>
+#include <linux/scatterlist.h>
+#include "trans_common.h"
+
+/**
+ *  p9_release_req_pages - Release pages after the transaction.
+ */
+void p9_release_pages(struct page **pages, int nr_pages)
+{
+	int i;
+
+	for (i = 0; i < nr_pages; i++)
+		if (pages[i])
+			put_page(pages[i]);
+}
+EXPORT_SYMBOL(p9_release_pages);
+
+/**
+ * p9_nr_pages - Return number of pages needed to accommodate the payload.
+ */
+int p9_nr_pages(char *data, int len)
+{
+	unsigned long start_page, end_page;
+	start_page =  (unsigned long)data >> PAGE_SHIFT;
+	end_page = ((unsigned long)data + len + PAGE_SIZE - 1) >> PAGE_SHIFT;
+	return end_page - start_page;
+}
+EXPORT_SYMBOL(p9_nr_pages);
+
+/**
+ * payload_gup - Translates user buffer into kernel pages and
+ * pins them either for read/write through get_user_pages_fast().
+ * @req: Request to be sent to server.
+ * @pdata_off: data offset into the first page after translation (gup).
+ * @pdata_len: Total length of the IO. gup may not return requested # of pages.
+ * @nr_pages: number of pages to accommodate the payload
+ * @rw: Indicates if the pages are for read or write.
+ */
+
+int p9_payload_gup(char *data, int *nr_pages, struct page **pages, int write)
+{
+	int nr_mapped_pages;
+
+	nr_mapped_pages = get_user_pages_fast((unsigned long)data,
+					      *nr_pages, write, pages);
+	if (nr_mapped_pages <= 0)
+		return nr_mapped_pages;
+
+	*nr_pages = nr_mapped_pages;
+	return 0;
+}
+EXPORT_SYMBOL(p9_payload_gup);

net/9p/trans_common.h

@@ -0,0 +1,17 @@
+/*
+ * Copyright IBM Corporation, 2010
+ * Author Venkateswararao Jujjuri <jvrao@linux.vnet.ibm.com>
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms of version 2.1 of the GNU Lesser General Public License
+ * as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful, but
+ * WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
+ *
+ */
+
+void p9_release_pages(struct page **, int);
+int p9_payload_gup(char *, int *, struct page **, int);
+int p9_nr_pages(char *, int);

net/9p/trans_rdma.c

@@ -0,0 +1,765 @@
+/*
+ * linux/fs/9p/trans_rdma.c
+ *
+ * RDMA transport layer based on the trans_fd.c implementation.
+ *
+ *  Copyright (C) 2008 by Tom Tucker <tom@opengridcomputing.com>
+ *  Copyright (C) 2006 by Russ Cox <rsc@swtch.com>
+ *  Copyright (C) 2004-2005 by Latchesar Ionkov <lucho@ionkov.net>
+ *  Copyright (C) 2004-2008 by Eric Van Hensbergen <ericvh@gmail.com>
+ *  Copyright (C) 1997-2002 by Ron Minnich <rminnich@sarnoff.com>
+ *
+ *  This program is free software; you can redistribute it and/or modify
+ *  it under the terms of the GNU General Public License version 2
+ *  as published by the Free Software Foundation.
+ *
+ *  This program is distributed in the hope that it will be useful,
+ *  but WITHOUT ANY WARRANTY; without even the implied warranty of
+ *  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ *  GNU General Public License for more details.
+ *
+ *  You should have received a copy of the GNU General Public License
+ *  along with this program; if not, write to:
+ *  Free Software Foundation
+ *  51 Franklin Street, Fifth Floor
+ *  Boston, MA  02111-1301  USA
+ *
+ */
+
+#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
+
+#include <linux/in.h>
+#include <linux/module.h>
+#include <linux/net.h>
+#include <linux/ipv6.h>
+#include <linux/kthread.h>
+#include <linux/errno.h>
+#include <linux/kernel.h>
+#include <linux/un.h>
+#include <linux/uaccess.h>
+#include <linux/inet.h>
+#include <linux/idr.h>
+#include <linux/file.h>
+#include <linux/parser.h>
+#include <linux/semaphore.h>
+#include <linux/slab.h>
+#include <net/9p/9p.h>
+#include <net/9p/client.h>
+#include <net/9p/transport.h>
+#include <rdma/ib_verbs.h>
+#include <rdma/rdma_cm.h>
+
+#define P9_PORT			5640
+#define P9_RDMA_SQ_DEPTH	32
+#define P9_RDMA_RQ_DEPTH	32
+#define P9_RDMA_SEND_SGE	4
+#define P9_RDMA_RECV_SGE	4
+#define P9_RDMA_IRD		0
+#define P9_RDMA_ORD		0
+#define P9_RDMA_TIMEOUT		30000		/* 30 seconds */
+#define P9_RDMA_MAXSIZE		(1024*1024)	/* 1MB */
+
+/**
+ * struct p9_trans_rdma - RDMA transport instance
+ *
+ * @state: tracks the transport state machine for connection setup and tear down
+ * @cm_id: The RDMA CM ID
+ * @pd: Protection Domain pointer
+ * @qp: Queue Pair pointer
+ * @cq: Completion Queue pointer
+ * @dm_mr: DMA Memory Region pointer
+ * @lkey: The local access only memory region key
+ * @timeout: Number of uSecs to wait for connection management events
+ * @sq_depth: The depth of the Send Queue
+ * @sq_sem: Semaphore for the SQ
+ * @rq_depth: The depth of the Receive Queue.
+ * @rq_sem: Semaphore for the RQ
+ * @excess_rc : Amount of posted Receive Contexts without a pending request.
+ *		See rdma_request()
+ * @addr: The remote peer's address
+ * @req_lock: Protects the active request list
+ * @cm_done: Completion event for connection management tracking
+ */
+struct p9_trans_rdma {
+	enum {
+		P9_RDMA_INIT,
+		P9_RDMA_ADDR_RESOLVED,
+		P9_RDMA_ROUTE_RESOLVED,
+		P9_RDMA_CONNECTED,
+		P9_RDMA_FLUSHING,
+		P9_RDMA_CLOSING,
+		P9_RDMA_CLOSED,
+	} state;
+	struct rdma_cm_id *cm_id;
+	struct ib_pd *pd;
+	struct ib_qp *qp;
+	struct ib_cq *cq;
+	struct ib_mr *dma_mr;
+	u32 lkey;
+	long timeout;
+	int sq_depth;
+	struct semaphore sq_sem;
+	int rq_depth;
+	struct semaphore rq_sem;
+	atomic_t excess_rc;
+	struct sockaddr_in addr;
+	spinlock_t req_lock;
+
+	struct completion cm_done;
+};
+
+/**
+ * p9_rdma_context - Keeps track of in-process WR
+ *
+ * @wc_op: The original WR op for when the CQE completes in error.
+ * @busa: Bus address to unmap when the WR completes
+ * @req: Keeps track of requests (send)
+ * @rc: Keepts track of replies (receive)
+ */
+struct p9_rdma_req;
+struct p9_rdma_context {
+	enum ib_wc_opcode wc_op;
+	dma_addr_t busa;
+	union {
+		struct p9_req_t *req;
+		struct p9_fcall *rc;
+	};
+};
+
+/**
+ * p9_rdma_opts - Collection of mount options
+ * @port: port of connection
+ * @sq_depth: The requested depth of the SQ. This really doesn't need
+ * to be any deeper than the number of threads used in the client
+ * @rq_depth: The depth of the RQ. Should be greater than or equal to SQ depth
+ * @timeout: Time to wait in msecs for CM events
+ */
+struct p9_rdma_opts {
+	short port;
+	int sq_depth;
+	int rq_depth;
+	long timeout;
+};
+
+/*
+ * Option Parsing (code inspired by NFS code)
+ */
+enum {
+	/* Options that take integer arguments */
+	Opt_port, Opt_rq_depth, Opt_sq_depth, Opt_timeout, Opt_err,
+};
+
+static match_table_t tokens = {
+	{Opt_port, "port=%u"},
+	{Opt_sq_depth, "sq=%u"},
+	{Opt_rq_depth, "rq=%u"},
+	{Opt_timeout, "timeout=%u"},
+	{Opt_err, NULL},
+};
+
+/**
+ * parse_opts - parse mount options into rdma options structure
+ * @params: options string passed from mount
+ * @opts: rdma transport-specific structure to parse options into
+ *
+ * Returns 0 upon success, -ERRNO upon failure
+ */
+static int parse_opts(char *params, struct p9_rdma_opts *opts)
+{
+	char *p;
+	substring_t args[MAX_OPT_ARGS];
+	int option;
+	char *options, *tmp_options;
+
+	opts->port = P9_PORT;
+	opts->sq_depth = P9_RDMA_SQ_DEPTH;
+	opts->rq_depth = P9_RDMA_RQ_DEPTH;
+	opts->timeout = P9_RDMA_TIMEOUT;
+
+	if (!params)
+		return 0;
+
+	tmp_options = kstrdup(params, GFP_KERNEL);
+	if (!tmp_options) {
+		p9_debug(P9_DEBUG_ERROR,
+			 "failed to allocate copy of option string\n");
+		return -ENOMEM;
+	}
+	options = tmp_options;
+
+	while ((p = strsep(&options, ",")) != NULL) {
+		int token;
+		int r;
+		if (!*p)
+			continue;
+		token = match_token(p, tokens, args);
+		if (token == Opt_err)
+			continue;
+		r = match_int(&args[0], &option);
+		if (r < 0) {
+			p9_debug(P9_DEBUG_ERROR,
+				 "integer field, but no integer?\n");
+			continue;
+		}
+		switch (token) {
+		case Opt_port:
+			opts->port = option;
+			break;
+		case Opt_sq_depth:
+			opts->sq_depth = option;
+			break;
+		case Opt_rq_depth:
+			opts->rq_depth = option;
+			break;
+		case Opt_timeout:
+			opts->timeout = option;
+			break;
+		default:
+			continue;
+		}
+	}
+	/* RQ must be at least as large as the SQ */
+	opts->rq_depth = max(opts->rq_depth, opts->sq_depth);
+	kfree(tmp_options);
+	return 0;
+}
+
+static int
+p9_cm_event_handler(struct rdma_cm_id *id, struct rdma_cm_event *event)
+{
+	struct p9_client *c = id->context;
+	struct p9_trans_rdma *rdma = c->trans;
+	switch (event->event) {
+	case RDMA_CM_EVENT_ADDR_RESOLVED:
+		BUG_ON(rdma->state != P9_RDMA_INIT);
+		rdma->state = P9_RDMA_ADDR_RESOLVED;
+		break;
+
+	case RDMA_CM_EVENT_ROUTE_RESOLVED:
+		BUG_ON(rdma->state != P9_RDMA_ADDR_RESOLVED);
+		rdma->state = P9_RDMA_ROUTE_RESOLVED;
+		break;
+
+	case RDMA_CM_EVENT_ESTABLISHED:
+		BUG_ON(rdma->state != P9_RDMA_ROUTE_RESOLVED);
+		rdma->state = P9_RDMA_CONNECTED;
+		break;
+
+	case RDMA_CM_EVENT_DISCONNECTED:
+		if (rdma)
+			rdma->state = P9_RDMA_CLOSED;
+		if (c)
+			c->status = Disconnected;
+		break;
+
+	case RDMA_CM_EVENT_TIMEWAIT_EXIT:
+		break;
+
+	case RDMA_CM_EVENT_ADDR_CHANGE:
+	case RDMA_CM_EVENT_ROUTE_ERROR:
+	case RDMA_CM_EVENT_DEVICE_REMOVAL:
+	case RDMA_CM_EVENT_MULTICAST_JOIN:
+	case RDMA_CM_EVENT_MULTICAST_ERROR:
+	case RDMA_CM_EVENT_REJECTED:
+	case RDMA_CM_EVENT_CONNECT_REQUEST:
+	case RDMA_CM_EVENT_CONNECT_RESPONSE:
+	case RDMA_CM_EVENT_CONNECT_ERROR:
+	case RDMA_CM_EVENT_ADDR_ERROR:
+	case RDMA_CM_EVENT_UNREACHABLE:
+		c->status = Disconnected;
+		rdma_disconnect(rdma->cm_id);
+		break;
+	default:
+		BUG();
+	}
+	complete(&rdma->cm_done);
+	return 0;
+}
+
+static void
+handle_recv(struct p9_client *client, struct p9_trans_rdma *rdma,
+	    struct p9_rdma_context *c, enum ib_wc_status status, u32 byte_len)
+{
+	struct p9_req_t *req;
+	int err = 0;
+	int16_t tag;
+
+	req = NULL;
+	ib_dma_unmap_single(rdma->cm_id->device, c->busa, client->msize,
+							 DMA_FROM_DEVICE);
+
+	if (status != IB_WC_SUCCESS)
+		goto err_out;
+
+	err = p9_parse_header(c->rc, NULL, NULL, &tag, 1);
+	if (err)
+		goto err_out;
+
+	req = p9_tag_lookup(client, tag);
+	if (!req)
+		goto err_out;
+
+	/* Check that we have not yet received a reply for this request.
+	 */
+	if (unlikely(req->rc)) {
+		pr_err("Duplicate reply for request %d", tag);
+		goto err_out;
+	}
+
+	req->rc = c->rc;
+	p9_client_cb(client, req, REQ_STATUS_RCVD);
+
+	return;
+
+ err_out:
+	p9_debug(P9_DEBUG_ERROR, "req %p err %d status %d\n", req, err, status);
+	rdma->state = P9_RDMA_FLUSHING;
+	client->status = Disconnected;
+}
+
+static void
+handle_send(struct p9_client *client, struct p9_trans_rdma *rdma,
+	    struct p9_rdma_context *c, enum ib_wc_status status, u32 byte_len)
+{
+	ib_dma_unmap_single(rdma->cm_id->device,
+			    c->busa, c->req->tc->size,
+			    DMA_TO_DEVICE);
+}
+
+static void qp_event_handler(struct ib_event *event, void *context)
+{
+	p9_debug(P9_DEBUG_ERROR, "QP event %d context %p\n",
+		 event->event, context);
+}
+
+static void cq_comp_handler(struct ib_cq *cq, void *cq_context)
+{
+	struct p9_client *client = cq_context;
+	struct p9_trans_rdma *rdma = client->trans;
+	int ret;
+	struct ib_wc wc;
+
+	ib_req_notify_cq(rdma->cq, IB_CQ_NEXT_COMP);
+	while ((ret = ib_poll_cq(cq, 1, &wc)) > 0) {
+		struct p9_rdma_context *c = (void *) (unsigned long) wc.wr_id;
+
+		switch (c->wc_op) {
+		case IB_WC_RECV:
+			handle_recv(client, rdma, c, wc.status, wc.byte_len);
+			up(&rdma->rq_sem);
+			break;
+
+		case IB_WC_SEND:
+			handle_send(client, rdma, c, wc.status, wc.byte_len);
+			up(&rdma->sq_sem);
+			break;
+
+		default:
+			pr_err("unexpected completion type, c->wc_op=%d, wc.opcode=%d, status=%d\n",
+			       c->wc_op, wc.opcode, wc.status);
+			break;
+		}
+		kfree(c);
+	}
+}
+
+static void cq_event_handler(struct ib_event *e, void *v)
+{
+	p9_debug(P9_DEBUG_ERROR, "CQ event %d context %p\n", e->event, v);
+}
+
+static void rdma_destroy_trans(struct p9_trans_rdma *rdma)
+{
+	if (!rdma)
+		return;
+
+	if (rdma->dma_mr && !IS_ERR(rdma->dma_mr))
+		ib_dereg_mr(rdma->dma_mr);
+
+	if (rdma->qp && !IS_ERR(rdma->qp))
+		ib_destroy_qp(rdma->qp);
+
+	if (rdma->pd && !IS_ERR(rdma->pd))
+		ib_dealloc_pd(rdma->pd);
+
+	if (rdma->cq && !IS_ERR(rdma->cq))
+		ib_destroy_cq(rdma->cq);
+
+	if (rdma->cm_id && !IS_ERR(rdma->cm_id))
+		rdma_destroy_id(rdma->cm_id);
+
+	kfree(rdma);
+}
+
+static int
+post_recv(struct p9_client *client, struct p9_rdma_context *c)
+{
+	struct p9_trans_rdma *rdma = client->trans;
+	struct ib_recv_wr wr, *bad_wr;
+	struct ib_sge sge;
+
+	c->busa = ib_dma_map_single(rdma->cm_id->device,
+				    c->rc->sdata, client->msize,
+				    DMA_FROM_DEVICE);
+	if (ib_dma_mapping_error(rdma->cm_id->device, c->busa))
+		goto error;
+
+	sge.addr = c->busa;
+	sge.length = client->msize;
+	sge.lkey = rdma->lkey;
+
+	wr.next = NULL;
+	c->wc_op = IB_WC_RECV;
+	wr.wr_id = (unsigned long) c;
+	wr.sg_list = &sge;
+	wr.num_sge = 1;
+	return ib_post_recv(rdma->qp, &wr, &bad_wr);
+
+ error:
+	p9_debug(P9_DEBUG_ERROR, "EIO\n");
+	return -EIO;
+}
+
+static int rdma_request(struct p9_client *client, struct p9_req_t *req)
+{
+	struct p9_trans_rdma *rdma = client->trans;
+	struct ib_send_wr wr, *bad_wr;
+	struct ib_sge sge;
+	int err = 0;
+	unsigned long flags;
+	struct p9_rdma_context *c = NULL;
+	struct p9_rdma_context *rpl_context = NULL;
+
+	/* When an error occurs between posting the recv and the send,
+	 * there will be a receive context posted without a pending request.
+	 * Since there is no way to "un-post" it, we remember it and skip
+	 * post_recv() for the next request.
+	 * So here,
+	 * see if we are this `next request' and need to absorb an excess rc.
+	 * If yes, then drop and free our own, and do not recv_post().
+	 **/
+	if (unlikely(atomic_read(&rdma->excess_rc) > 0)) {
+		if ((atomic_sub_return(1, &rdma->excess_rc) >= 0)) {
+			/* Got one ! */
+			kfree(req->rc);
+			req->rc = NULL;
+			goto dont_need_post_recv;
+		} else {
+			/* We raced and lost. */
+			atomic_inc(&rdma->excess_rc);
+		}
+	}
+
+	/* Allocate an fcall for the reply */
+	rpl_context = kmalloc(sizeof *rpl_context, GFP_NOFS);
+	if (!rpl_context) {
+		err = -ENOMEM;
+		goto recv_error;
+	}
+	rpl_context->rc = req->rc;
+
+	/*
+	 * Post a receive buffer for this request. We need to ensure
+	 * there is a reply buffer available for every outstanding
+	 * request. A flushed request can result in no reply for an
+	 * outstanding request, so we must keep a count to avoid
+	 * overflowing the RQ.
+	 */
+	if (down_interruptible(&rdma->rq_sem)) {
+		err = -EINTR;
+		goto recv_error;
+	}
+
+	err = post_recv(client, rpl_context);
+	if (err) {
+		p9_debug(P9_DEBUG_FCALL, "POST RECV failed\n");
+		goto recv_error;
+	}
+	/* remove posted receive buffer from request structure */
+	req->rc = NULL;
+
+dont_need_post_recv:
+	/* Post the request */
+	c = kmalloc(sizeof *c, GFP_NOFS);
+	if (!c) {
+		err = -ENOMEM;
+		goto send_error;
+	}
+	c->req = req;
+
+	c->busa = ib_dma_map_single(rdma->cm_id->device,
+				    c->req->tc->sdata, c->req->tc->size,
+				    DMA_TO_DEVICE);
+	if (ib_dma_mapping_error(rdma->cm_id->device, c->busa)) {
+		err = -EIO;
+		goto send_error;
+	}
+
+	sge.addr = c->busa;
+	sge.length = c->req->tc->size;
+	sge.lkey = rdma->lkey;
+
+	wr.next = NULL;
+	c->wc_op = IB_WC_SEND;
+	wr.wr_id = (unsigned long) c;
+	wr.opcode = IB_WR_SEND;
+	wr.send_flags = IB_SEND_SIGNALED;
+	wr.sg_list = &sge;
+	wr.num_sge = 1;
+
+	if (down_interruptible(&rdma->sq_sem)) {
+		err = -EINTR;
+		goto send_error;
+	}
+
+	/* Mark request as `sent' *before* we actually send it,
+	 * because doing if after could erase the REQ_STATUS_RCVD
+	 * status in case of a very fast reply.
+	 */
+	req->status = REQ_STATUS_SENT;
+	err = ib_post_send(rdma->qp, &wr, &bad_wr);
+	if (err)
+		goto send_error;
+
+	/* Success */
+	return 0;
+
+ /* Handle errors that happened during or while preparing the send: */
+ send_error:
+	req->status = REQ_STATUS_ERROR;
+	kfree(c);
+	p9_debug(P9_DEBUG_ERROR, "Error %d in rdma_request()\n", err);
+
+	/* Ach.
+	 *  We did recv_post(), but not send. We have one recv_post in excess.
+	 */
+	atomic_inc(&rdma->excess_rc);
+	return err;
+
+ /* Handle errors that happened during or while preparing post_recv(): */
+ recv_error:
+	kfree(rpl_context);
+	spin_lock_irqsave(&rdma->req_lock, flags);
+	if (rdma->state < P9_RDMA_CLOSING) {
+		rdma->state = P9_RDMA_CLOSING;
+		spin_unlock_irqrestore(&rdma->req_lock, flags);
+		rdma_disconnect(rdma->cm_id);
+	} else
+		spin_unlock_irqrestore(&rdma->req_lock, flags);
+	return err;
+}
+
+static void rdma_close(struct p9_client *client)
+{
+	struct p9_trans_rdma *rdma;
+
+	if (!client)
+		return;
+
+	rdma = client->trans;
+	if (!rdma)
+		return;
+
+	client->status = Disconnected;
+	rdma_disconnect(rdma->cm_id);
+	rdma_destroy_trans(rdma);
+}
+
+/**
+ * alloc_rdma - Allocate and initialize the rdma transport structure
+ * @opts: Mount options structure
+ */
+static struct p9_trans_rdma *alloc_rdma(struct p9_rdma_opts *opts)
+{
+	struct p9_trans_rdma *rdma;
+
+	rdma = kzalloc(sizeof(struct p9_trans_rdma), GFP_KERNEL);
+	if (!rdma)
+		return NULL;
+
+	rdma->sq_depth = opts->sq_depth;
+	rdma->rq_depth = opts->rq_depth;
+	rdma->timeout = opts->timeout;
+	spin_lock_init(&rdma->req_lock);
+	init_completion(&rdma->cm_done);
+	sema_init(&rdma->sq_sem, rdma->sq_depth);
+	sema_init(&rdma->rq_sem, rdma->rq_depth);
+	atomic_set(&rdma->excess_rc, 0);
+
+	return rdma;
+}
+
+static int rdma_cancel(struct p9_client *client, struct p9_req_t *req)
+{
+	/* Nothing to do here.
+	 * We will take care of it (if we have to) in rdma_cancelled()
+	 */
+	return 1;
+}
+
+/* A request has been fully flushed without a reply.
+ * That means we have posted one buffer in excess.
+ */
+static int rdma_cancelled(struct p9_client *client, struct p9_req_t *req)
+{
+	struct p9_trans_rdma *rdma = client->trans;
+	atomic_inc(&rdma->excess_rc);
+	return 0;
+}
+
+/**
+ * trans_create_rdma - Transport method for creating atransport instance
+ * @client: client instance
+ * @addr: IP address string
+ * @args: Mount options string
+ */
+static int
+rdma_create_trans(struct p9_client *client, const char *addr, char *args)
+{
+	int err;
+	struct p9_rdma_opts opts;
+	struct p9_trans_rdma *rdma;
+	struct rdma_conn_param conn_param;
+	struct ib_qp_init_attr qp_attr;
+	struct ib_device_attr devattr;
+
+	/* Parse the transport specific mount options */
+	err = parse_opts(args, &opts);
+	if (err < 0)
+		return err;
+
+	/* Create and initialize the RDMA transport structure */
+	rdma = alloc_rdma(&opts);
+	if (!rdma)
+		return -ENOMEM;
+
+	/* Create the RDMA CM ID */
+	rdma->cm_id = rdma_create_id(p9_cm_event_handler, client, RDMA_PS_TCP,
+				     IB_QPT_RC);
+	if (IS_ERR(rdma->cm_id))
+		goto error;
+
+	/* Associate the client with the transport */
+	client->trans = rdma;
+
+	/* Resolve the server's address */
+	rdma->addr.sin_family = AF_INET;
+	rdma->addr.sin_addr.s_addr = in_aton(addr);
+	rdma->addr.sin_port = htons(opts.port);
+	err = rdma_resolve_addr(rdma->cm_id, NULL,
+				(struct sockaddr *)&rdma->addr,
+				rdma->timeout);
+	if (err)
+		goto error;
+	err = wait_for_completion_interruptible(&rdma->cm_done);
+	if (err || (rdma->state != P9_RDMA_ADDR_RESOLVED))
+		goto error;
+
+	/* Resolve the route to the server */
+	err = rdma_resolve_route(rdma->cm_id, rdma->timeout);
+	if (err)
+		goto error;
+	err = wait_for_completion_interruptible(&rdma->cm_done);
+	if (err || (rdma->state != P9_RDMA_ROUTE_RESOLVED))
+		goto error;
+
+	/* Query the device attributes */
+	err = ib_query_device(rdma->cm_id->device, &devattr);
+	if (err)
+		goto error;
+
+	/* Create the Completion Queue */
+	rdma->cq = ib_create_cq(rdma->cm_id->device, cq_comp_handler,
+				cq_event_handler, client,
+				opts.sq_depth + opts.rq_depth + 1, 0);
+	if (IS_ERR(rdma->cq))
+		goto error;
+	ib_req_notify_cq(rdma->cq, IB_CQ_NEXT_COMP);
+
+	/* Create the Protection Domain */
+	rdma->pd = ib_alloc_pd(rdma->cm_id->device);
+	if (IS_ERR(rdma->pd))
+		goto error;
+
+	/* Cache the DMA lkey in the transport */
+	rdma->dma_mr = NULL;
+	if (devattr.device_cap_flags & IB_DEVICE_LOCAL_DMA_LKEY)
+		rdma->lkey = rdma->cm_id->device->local_dma_lkey;
+	else {
+		rdma->dma_mr = ib_get_dma_mr(rdma->pd, IB_ACCESS_LOCAL_WRITE);
+		if (IS_ERR(rdma->dma_mr))
+			goto error;
+		rdma->lkey = rdma->dma_mr->lkey;
+	}
+
+	/* Create the Queue Pair */
+	memset(&qp_attr, 0, sizeof qp_attr);
+	qp_attr.event_handler = qp_event_handler;
+	qp_attr.qp_context = client;
+	qp_attr.cap.max_send_wr = opts.sq_depth;
+	qp_attr.cap.max_recv_wr = opts.rq_depth;
+	qp_attr.cap.max_send_sge = P9_RDMA_SEND_SGE;
+	qp_attr.cap.max_recv_sge = P9_RDMA_RECV_SGE;
+	qp_attr.sq_sig_type = IB_SIGNAL_REQ_WR;
+	qp_attr.qp_type = IB_QPT_RC;
+	qp_attr.send_cq = rdma->cq;
+	qp_attr.recv_cq = rdma->cq;
+	err = rdma_create_qp(rdma->cm_id, rdma->pd, &qp_attr);
+	if (err)
+		goto error;
+	rdma->qp = rdma->cm_id->qp;
+
+	/* Request a connection */
+	memset(&conn_param, 0, sizeof(conn_param));
+	conn_param.private_data = NULL;
+	conn_param.private_data_len = 0;
+	conn_param.responder_resources = P9_RDMA_IRD;
+	conn_param.initiator_depth = P9_RDMA_ORD;
+	err = rdma_connect(rdma->cm_id, &conn_param);
+	if (err)
+		goto error;
+	err = wait_for_completion_interruptible(&rdma->cm_done);
+	if (err || (rdma->state != P9_RDMA_CONNECTED))
+		goto error;
+
+	client->status = Connected;
+
+	return 0;
+
+error:
+	rdma_destroy_trans(rdma);
+	return -ENOTCONN;
+}
+
+static struct p9_trans_module p9_rdma_trans = {
+	.name = "rdma",
+	.maxsize = P9_RDMA_MAXSIZE,
+	.def = 0,
+	.owner = THIS_MODULE,
+	.create = rdma_create_trans,
+	.close = rdma_close,
+	.request = rdma_request,
+	.cancel = rdma_cancel,
+	.cancelled = rdma_cancelled,
+};
+
+/**
+ * p9_trans_rdma_init - Register the 9P RDMA transport driver
+ */
+static int __init p9_trans_rdma_init(void)
+{
+	v9fs_register_trans(&p9_rdma_trans);
+	return 0;
+}
+
+static void __exit p9_trans_rdma_exit(void)
+{
+	v9fs_unregister_trans(&p9_rdma_trans);
+}
+
+module_init(p9_trans_rdma_init);
+module_exit(p9_trans_rdma_exit);
+
+MODULE_AUTHOR("Tom Tucker <tom@opengridcomputing.com>");
+MODULE_DESCRIPTION("RDMA Transport for 9P");
+MODULE_LICENSE("Dual BSD/GPL");

net/9p/trans_virtio.c

@@ -0,0 +1,730 @@
+/*
+ * The Virtio 9p transport driver
+ *
+ * This is a block based transport driver based on the lguest block driver
+ * code.
+ *
+ *  Copyright (C) 2007, 2008 Eric Van Hensbergen, IBM Corporation
+ *
+ *  Based on virtio console driver
+ *  Copyright (C) 2006, 2007 Rusty Russell, IBM Corporation
+ *
+ *  This program is free software; you can redistribute it and/or modify
+ *  it under the terms of the GNU General Public License version 2
+ *  as published by the Free Software Foundation.
+ *
+ *  This program is distributed in the hope that it will be useful,
+ *  but WITHOUT ANY WARRANTY; without even the implied warranty of
+ *  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ *  GNU General Public License for more details.
+ *
+ *  You should have received a copy of the GNU General Public License
+ *  along with this program; if not, write to:
+ *  Free Software Foundation
+ *  51 Franklin Street, Fifth Floor
+ *  Boston, MA  02111-1301  USA
+ *
+ */
+
+#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
+
+#include <linux/in.h>
+#include <linux/module.h>
+#include <linux/net.h>
+#include <linux/ipv6.h>
+#include <linux/errno.h>
+#include <linux/kernel.h>
+#include <linux/un.h>
+#include <linux/uaccess.h>
+#include <linux/inet.h>
+#include <linux/idr.h>
+#include <linux/file.h>
+#include <linux/highmem.h>
+#include <linux/slab.h>
+#include <net/9p/9p.h>
+#include <linux/parser.h>
+#include <net/9p/client.h>
+#include <net/9p/transport.h>
+#include <linux/scatterlist.h>
+#include <linux/swap.h>
+#include <linux/virtio.h>
+#include <linux/virtio_9p.h>
+#include "trans_common.h"
+
+#define VIRTQUEUE_NUM	128
+
+/* a single mutex to manage channel initialization and attachment */
+static DEFINE_MUTEX(virtio_9p_lock);
+static DECLARE_WAIT_QUEUE_HEAD(vp_wq);
+static atomic_t vp_pinned = ATOMIC_INIT(0);
+
+/**
+ * struct virtio_chan - per-instance transport information
+ * @initialized: whether the channel is initialized
+ * @inuse: whether the channel is in use
+ * @lock: protects multiple elements within this structure
+ * @client: client instance
+ * @vdev: virtio dev associated with this channel
+ * @vq: virtio queue associated with this channel
+ * @sg: scatter gather list which is used to pack a request (protected?)
+ *
+ * We keep all per-channel information in a structure.
+ * This structure is allocated within the devices dev->mem space.
+ * A pointer to the structure will get put in the transport private.
+ *
+ */
+
+struct virtio_chan {
+	bool inuse;
+
+	spinlock_t lock;
+
+	struct p9_client *client;
+	struct virtio_device *vdev;
+	struct virtqueue *vq;
+	int ring_bufs_avail;
+	wait_queue_head_t *vc_wq;
+	/* This is global limit. Since we don't have a global structure,
+	 * will be placing it in each channel.
+	 */
+	unsigned long p9_max_pages;
+	/* Scatterlist: can be too big for stack. */
+	struct scatterlist sg[VIRTQUEUE_NUM];
+
+	int tag_len;
+	/*
+	 * tag name to identify a mount Non-null terminated
+	 */
+	char *tag;
+
+	struct list_head chan_list;
+};
+
+static struct list_head virtio_chan_list;
+
+/* How many bytes left in this page. */
+static unsigned int rest_of_page(void *data)
+{
+	return PAGE_SIZE - ((unsigned long)data % PAGE_SIZE);
+}
+
+/**
+ * p9_virtio_close - reclaim resources of a channel
+ * @client: client instance
+ *
+ * This reclaims a channel by freeing its resources and
+ * reseting its inuse flag.
+ *
+ */
+
+static void p9_virtio_close(struct p9_client *client)
+{
+	struct virtio_chan *chan = client->trans;
+
+	mutex_lock(&virtio_9p_lock);
+	if (chan)
+		chan->inuse = false;
+	mutex_unlock(&virtio_9p_lock);
+}
+
+/**
+ * req_done - callback which signals activity from the server
+ * @vq: virtio queue activity was received on
+ *
+ * This notifies us that the server has triggered some activity
+ * on the virtio channel - most likely a response to request we
+ * sent.  Figure out which requests now have responses and wake up
+ * those threads.
+ *
+ * Bugs: could do with some additional sanity checking, but appears to work.
+ *
+ */
+
+static void req_done(struct virtqueue *vq)
+{
+	struct virtio_chan *chan = vq->vdev->priv;
+	struct p9_fcall *rc;
+	unsigned int len;
+	struct p9_req_t *req;
+	unsigned long flags;
+
+	p9_debug(P9_DEBUG_TRANS, ": request done\n");
+
+	while (1) {
+		spin_lock_irqsave(&chan->lock, flags);
+		rc = virtqueue_get_buf(chan->vq, &len);
+		if (rc == NULL) {
+			spin_unlock_irqrestore(&chan->lock, flags);
+			break;
+		}
+		chan->ring_bufs_avail = 1;
+		spin_unlock_irqrestore(&chan->lock, flags);
+		/* Wakeup if anyone waiting for VirtIO ring space. */
+		wake_up(chan->vc_wq);
+		p9_debug(P9_DEBUG_TRANS, ": rc %p\n", rc);
+		p9_debug(P9_DEBUG_TRANS, ": lookup tag %d\n", rc->tag);
+		req = p9_tag_lookup(chan->client, rc->tag);
+		p9_client_cb(chan->client, req, REQ_STATUS_RCVD);
+	}
+}
+
+/**
+ * pack_sg_list - pack a scatter gather list from a linear buffer
+ * @sg: scatter/gather list to pack into
+ * @start: which segment of the sg_list to start at
+ * @limit: maximum segment to pack data to
+ * @data: data to pack into scatter/gather list
+ * @count: amount of data to pack into the scatter/gather list
+ *
+ * sg_lists have multiple segments of various sizes.  This will pack
+ * arbitrary data into an existing scatter gather list, segmenting the
+ * data as necessary within constraints.
+ *
+ */
+
+static int pack_sg_list(struct scatterlist *sg, int start,
+			int limit, char *data, int count)
+{
+	int s;
+	int index = start;
+
+	while (count) {
+		s = rest_of_page(data);
+		if (s > count)
+			s = count;
+		BUG_ON(index > limit);
+		/* Make sure we don't terminate early. */
+		sg_unmark_end(&sg[index]);
+		sg_set_buf(&sg[index++], data, s);
+		count -= s;
+		data += s;
+	}
+	if (index-start)
+		sg_mark_end(&sg[index - 1]);
+	return index-start;
+}
+
+/* We don't currently allow canceling of virtio requests */
+static int p9_virtio_cancel(struct p9_client *client, struct p9_req_t *req)
+{
+	return 1;
+}
+
+/**
+ * pack_sg_list_p - Just like pack_sg_list. Instead of taking a buffer,
+ * this takes a list of pages.
+ * @sg: scatter/gather list to pack into
+ * @start: which segment of the sg_list to start at
+ * @pdata: a list of pages to add into sg.
+ * @nr_pages: number of pages to pack into the scatter/gather list
+ * @data: data to pack into scatter/gather list
+ * @count: amount of data to pack into the scatter/gather list
+ */
+static int
+pack_sg_list_p(struct scatterlist *sg, int start, int limit,
+	       struct page **pdata, int nr_pages, char *data, int count)
+{
+	int i = 0, s;
+	int data_off;
+	int index = start;
+
+	BUG_ON(nr_pages > (limit - start));
+	/*
+	 * if the first page doesn't start at
+	 * page boundary find the offset
+	 */
+	data_off = offset_in_page(data);
+	while (nr_pages) {
+		s = rest_of_page(data);
+		if (s > count)
+			s = count;
+		/* Make sure we don't terminate early. */
+		sg_unmark_end(&sg[index]);
+		sg_set_page(&sg[index++], pdata[i++], s, data_off);
+		data_off = 0;
+		data += s;
+		count -= s;
+		nr_pages--;
+	}
+
+	if (index-start)
+		sg_mark_end(&sg[index - 1]);
+	return index - start;
+}
+
+/**
+ * p9_virtio_request - issue a request
+ * @client: client instance issuing the request
+ * @req: request to be issued
+ *
+ */
+
+static int
+p9_virtio_request(struct p9_client *client, struct p9_req_t *req)
+{
+	int err;
+	int in, out, out_sgs, in_sgs;
+	unsigned long flags;
+	struct virtio_chan *chan = client->trans;
+	struct scatterlist *sgs[2];
+
+	p9_debug(P9_DEBUG_TRANS, "9p debug: virtio request\n");
+
+	req->status = REQ_STATUS_SENT;
+req_retry:
+	spin_lock_irqsave(&chan->lock, flags);
+
+	out_sgs = in_sgs = 0;
+	/* Handle out VirtIO ring buffers */
+	out = pack_sg_list(chan->sg, 0,
+			   VIRTQUEUE_NUM, req->tc->sdata, req->tc->size);
+	if (out)
+		sgs[out_sgs++] = chan->sg;
+
+	in = pack_sg_list(chan->sg, out,
+			  VIRTQUEUE_NUM, req->rc->sdata, req->rc->capacity);
+	if (in)
+		sgs[out_sgs + in_sgs++] = chan->sg + out;
+
+	err = virtqueue_add_sgs(chan->vq, sgs, out_sgs, in_sgs, req->tc,
+				GFP_ATOMIC);
+	if (err < 0) {
+		if (err == -ENOSPC) {
+			chan->ring_bufs_avail = 0;
+			spin_unlock_irqrestore(&chan->lock, flags);
+			err = wait_event_interruptible(*chan->vc_wq,
+							chan->ring_bufs_avail);
+			if (err  == -ERESTARTSYS)
+				return err;
+
+			p9_debug(P9_DEBUG_TRANS, "Retry virtio request\n");
+			goto req_retry;
+		} else {
+			spin_unlock_irqrestore(&chan->lock, flags);
+			p9_debug(P9_DEBUG_TRANS,
+				 "virtio rpc add_sgs returned failure\n");
+			return -EIO;
+		}
+	}
+	virtqueue_kick(chan->vq);
+	spin_unlock_irqrestore(&chan->lock, flags);
+
+	p9_debug(P9_DEBUG_TRANS, "virtio request kicked\n");
+	return 0;
+}
+
+static int p9_get_mapped_pages(struct virtio_chan *chan,
+			       struct page **pages, char *data,
+			       int nr_pages, int write, int kern_buf)
+{
+	int err;
+	if (!kern_buf) {
+		/*
+		 * We allow only p9_max_pages pinned. We wait for the
+		 * Other zc request to finish here
+		 */
+		if (atomic_read(&vp_pinned) >= chan->p9_max_pages) {
+			err = wait_event_interruptible(vp_wq,
+			      (atomic_read(&vp_pinned) < chan->p9_max_pages));
+			if (err == -ERESTARTSYS)
+				return err;
+		}
+		err = p9_payload_gup(data, &nr_pages, pages, write);
+		if (err < 0)
+			return err;
+		atomic_add(nr_pages, &vp_pinned);
+	} else {
+		/* kernel buffer, no need to pin pages */
+		int s, index = 0;
+		int count = nr_pages;
+		while (nr_pages) {
+			s = rest_of_page(data);
+			if (is_vmalloc_addr(data))
+				pages[index++] = vmalloc_to_page(data);
+			else
+				pages[index++] = kmap_to_page(data);
+			data += s;
+			nr_pages--;
+		}
+		nr_pages = count;
+	}
+	return nr_pages;
+}
+
+/**
+ * p9_virtio_zc_request - issue a zero copy request
+ * @client: client instance issuing the request
+ * @req: request to be issued
+ * @uidata: user bffer that should be ued for zero copy read
+ * @uodata: user buffer that shoud be user for zero copy write
+ * @inlen: read buffer size
+ * @olen: write buffer size
+ * @hdrlen: reader header size, This is the size of response protocol data
+ *
+ */
+static int
+p9_virtio_zc_request(struct p9_client *client, struct p9_req_t *req,
+		     char *uidata, char *uodata, int inlen,
+		     int outlen, int in_hdr_len, int kern_buf)
+{
+	int in, out, err, out_sgs, in_sgs;
+	unsigned long flags;
+	int in_nr_pages = 0, out_nr_pages = 0;
+	struct page **in_pages = NULL, **out_pages = NULL;
+	struct virtio_chan *chan = client->trans;
+	struct scatterlist *sgs[4];
+
+	p9_debug(P9_DEBUG_TRANS, "virtio request\n");
+
+	if (uodata) {
+		out_nr_pages = p9_nr_pages(uodata, outlen);
+		out_pages = kmalloc(sizeof(struct page *) * out_nr_pages,
+				    GFP_NOFS);
+		if (!out_pages) {
+			err = -ENOMEM;
+			goto err_out;
+		}
+		out_nr_pages = p9_get_mapped_pages(chan, out_pages, uodata,
+						   out_nr_pages, 0, kern_buf);
+		if (out_nr_pages < 0) {
+			err = out_nr_pages;
+			kfree(out_pages);
+			out_pages = NULL;
+			goto err_out;
+		}
+	}
+	if (uidata) {
+		in_nr_pages = p9_nr_pages(uidata, inlen);
+		in_pages = kmalloc(sizeof(struct page *) * in_nr_pages,
+				   GFP_NOFS);
+		if (!in_pages) {
+			err = -ENOMEM;
+			goto err_out;
+		}
+		in_nr_pages = p9_get_mapped_pages(chan, in_pages, uidata,
+						  in_nr_pages, 1, kern_buf);
+		if (in_nr_pages < 0) {
+			err = in_nr_pages;
+			kfree(in_pages);
+			in_pages = NULL;
+			goto err_out;
+		}
+	}
+	req->status = REQ_STATUS_SENT;
+req_retry_pinned:
+	spin_lock_irqsave(&chan->lock, flags);
+
+	out_sgs = in_sgs = 0;
+
+	/* out data */
+	out = pack_sg_list(chan->sg, 0,
+			   VIRTQUEUE_NUM, req->tc->sdata, req->tc->size);
+
+	if (out)
+		sgs[out_sgs++] = chan->sg;
+
+	if (out_pages) {
+		sgs[out_sgs++] = chan->sg + out;
+		out += pack_sg_list_p(chan->sg, out, VIRTQUEUE_NUM,
+				      out_pages, out_nr_pages, uodata, outlen);
+	}
+		
+	/*
+	 * Take care of in data
+	 * For example TREAD have 11.
+	 * 11 is the read/write header = PDU Header(7) + IO Size (4).
+	 * Arrange in such a way that server places header in the
+	 * alloced memory and payload onto the user buffer.
+	 */
+	in = pack_sg_list(chan->sg, out,
+			  VIRTQUEUE_NUM, req->rc->sdata, in_hdr_len);
+	if (in)
+		sgs[out_sgs + in_sgs++] = chan->sg + out;
+
+	if (in_pages) {
+		sgs[out_sgs + in_sgs++] = chan->sg + out + in;
+		in += pack_sg_list_p(chan->sg, out + in, VIRTQUEUE_NUM,
+				     in_pages, in_nr_pages, uidata, inlen);
+	}
+
+	BUG_ON(out_sgs + in_sgs > ARRAY_SIZE(sgs));
+	err = virtqueue_add_sgs(chan->vq, sgs, out_sgs, in_sgs, req->tc,
+				GFP_ATOMIC);
+	if (err < 0) {
+		if (err == -ENOSPC) {
+			chan->ring_bufs_avail = 0;
+			spin_unlock_irqrestore(&chan->lock, flags);
+			err = wait_event_interruptible(*chan->vc_wq,
+						       chan->ring_bufs_avail);
+			if (err  == -ERESTARTSYS)
+				goto err_out;
+
+			p9_debug(P9_DEBUG_TRANS, "Retry virtio request\n");
+			goto req_retry_pinned;
+		} else {
+			spin_unlock_irqrestore(&chan->lock, flags);
+			p9_debug(P9_DEBUG_TRANS,
+				 "virtio rpc add_sgs returned failure\n");
+			err = -EIO;
+			goto err_out;
+		}
+	}
+	virtqueue_kick(chan->vq);
+	spin_unlock_irqrestore(&chan->lock, flags);
+	p9_debug(P9_DEBUG_TRANS, "virtio request kicked\n");
+	err = wait_event_interruptible(*req->wq,
+				       req->status >= REQ_STATUS_RCVD);
+	/*
+	 * Non kernel buffers are pinned, unpin them
+	 */
+err_out:
+	if (!kern_buf) {
+		if (in_pages) {
+			p9_release_pages(in_pages, in_nr_pages);
+			atomic_sub(in_nr_pages, &vp_pinned);
+		}
+		if (out_pages) {
+			p9_release_pages(out_pages, out_nr_pages);
+			atomic_sub(out_nr_pages, &vp_pinned);
+		}
+		/* wakeup anybody waiting for slots to pin pages */
+		wake_up(&vp_wq);
+	}
+	kfree(in_pages);
+	kfree(out_pages);
+	return err;
+}
+
+static ssize_t p9_mount_tag_show(struct device *dev,
+				struct device_attribute *attr, char *buf)
+{
+	struct virtio_chan *chan;
+	struct virtio_device *vdev;
+
+	vdev = dev_to_virtio(dev);
+	chan = vdev->priv;
+
+	return snprintf(buf, chan->tag_len + 1, "%s", chan->tag);
+}
+
+static DEVICE_ATTR(mount_tag, 0444, p9_mount_tag_show, NULL);
+
+/**
+ * p9_virtio_probe - probe for existence of 9P virtio channels
+ * @vdev: virtio device to probe
+ *
+ * This probes for existing virtio channels.
+ *
+ */
+
+static int p9_virtio_probe(struct virtio_device *vdev)
+{
+	__u16 tag_len;
+	char *tag;
+	int err;
+	struct virtio_chan *chan;
+
+	chan = kmalloc(sizeof(struct virtio_chan), GFP_KERNEL);
+	if (!chan) {
+		pr_err("Failed to allocate virtio 9P channel\n");
+		err = -ENOMEM;
+		goto fail;
+	}
+
+	chan->vdev = vdev;
+
+	/* We expect one virtqueue, for requests. */
+	chan->vq = virtio_find_single_vq(vdev, req_done, "requests");
+	if (IS_ERR(chan->vq)) {
+		err = PTR_ERR(chan->vq);
+		goto out_free_vq;
+	}
+	chan->vq->vdev->priv = chan;
+	spin_lock_init(&chan->lock);
+
+	sg_init_table(chan->sg, VIRTQUEUE_NUM);
+
+	chan->inuse = false;
+	if (virtio_has_feature(vdev, VIRTIO_9P_MOUNT_TAG)) {
+		virtio_cread(vdev, struct virtio_9p_config, tag_len, &tag_len);
+	} else {
+		err = -EINVAL;
+		goto out_free_vq;
+	}
+	tag = kmalloc(tag_len, GFP_KERNEL);
+	if (!tag) {
+		err = -ENOMEM;
+		goto out_free_vq;
+	}
+
+	virtio_cread_bytes(vdev, offsetof(struct virtio_9p_config, tag),
+			   tag, tag_len);
+	chan->tag = tag;
+	chan->tag_len = tag_len;
+	err = sysfs_create_file(&(vdev->dev.kobj), &dev_attr_mount_tag.attr);
+	if (err) {
+		goto out_free_tag;
+	}
+	chan->vc_wq = kmalloc(sizeof(wait_queue_head_t), GFP_KERNEL);
+	if (!chan->vc_wq) {
+		err = -ENOMEM;
+		goto out_free_tag;
+	}
+	init_waitqueue_head(chan->vc_wq);
+	chan->ring_bufs_avail = 1;
+	/* Ceiling limit to avoid denial of service attacks */
+	chan->p9_max_pages = nr_free_buffer_pages()/4;
+
+	virtio_device_ready(vdev);
+
+	mutex_lock(&virtio_9p_lock);
+	list_add_tail(&chan->chan_list, &virtio_chan_list);
+	mutex_unlock(&virtio_9p_lock);
+
+	/* Let udev rules use the new mount_tag attribute. */
+	kobject_uevent(&(vdev->dev.kobj), KOBJ_CHANGE);
+
+	return 0;
+
+out_free_tag:
+	kfree(tag);
+out_free_vq:
+	vdev->config->del_vqs(vdev);
+	kfree(chan);
+fail:
+	return err;
+}
+
+
+/**
+ * p9_virtio_create - allocate a new virtio channel
+ * @client: client instance invoking this transport
+ * @devname: string identifying the channel to connect to (unused)
+ * @args: args passed from sys_mount() for per-transport options (unused)
+ *
+ * This sets up a transport channel for 9p communication.  Right now
+ * we only match the first available channel, but eventually we couldlook up
+ * alternate channels by matching devname versus a virtio_config entry.
+ * We use a simple reference count mechanism to ensure that only a single
+ * mount has a channel open at a time.
+ *
+ */
+
+static int
+p9_virtio_create(struct p9_client *client, const char *devname, char *args)
+{
+	struct virtio_chan *chan;
+	int ret = -ENOENT;
+	int found = 0;
+
+	mutex_lock(&virtio_9p_lock);
+	list_for_each_entry(chan, &virtio_chan_list, chan_list) {
+		if (!strncmp(devname, chan->tag, chan->tag_len) &&
+		    strlen(devname) == chan->tag_len) {
+			if (!chan->inuse) {
+				chan->inuse = true;
+				found = 1;
+				break;
+			}
+			ret = -EBUSY;
+		}
+	}
+	mutex_unlock(&virtio_9p_lock);
+
+	if (!found) {
+		pr_err("no channels available\n");
+		return ret;
+	}
+
+	client->trans = (void *)chan;
+	client->status = Connected;
+	chan->client = client;
+
+	return 0;
+}
+
+/**
+ * p9_virtio_remove - clean up resources associated with a virtio device
+ * @vdev: virtio device to remove
+ *
+ */
+
+static void p9_virtio_remove(struct virtio_device *vdev)
+{
+	struct virtio_chan *chan = vdev->priv;
+
+	if (chan->inuse)
+		p9_virtio_close(chan->client);
+	vdev->config->del_vqs(vdev);
+
+	mutex_lock(&virtio_9p_lock);
+	list_del(&chan->chan_list);
+	mutex_unlock(&virtio_9p_lock);
+	sysfs_remove_file(&(vdev->dev.kobj), &dev_attr_mount_tag.attr);
+	kobject_uevent(&(vdev->dev.kobj), KOBJ_CHANGE);
+	kfree(chan->tag);
+	kfree(chan->vc_wq);
+	kfree(chan);
+
+}
+
+static struct virtio_device_id id_table[] = {
+	{ VIRTIO_ID_9P, VIRTIO_DEV_ANY_ID },
+	{ 0 },
+};
+
+static unsigned int features[] = {
+	VIRTIO_9P_MOUNT_TAG,
+};
+
+/* The standard "struct lguest_driver": */
+static struct virtio_driver p9_virtio_drv = {
+	.feature_table  = features,
+	.feature_table_size = ARRAY_SIZE(features),
+	.driver.name    = KBUILD_MODNAME,
+	.driver.owner	= THIS_MODULE,
+	.id_table	= id_table,
+	.probe		= p9_virtio_probe,
+	.remove		= p9_virtio_remove,
+};
+
+static struct p9_trans_module p9_virtio_trans = {
+	.name = "virtio",
+	.create = p9_virtio_create,
+	.close = p9_virtio_close,
+	.request = p9_virtio_request,
+	.zc_request = p9_virtio_zc_request,
+	.cancel = p9_virtio_cancel,
+	/*
+	 * We leave one entry for input and one entry for response
+	 * headers. We also skip one more entry to accomodate, address
+	 * that are not at page boundary, that can result in an extra
+	 * page in zero copy.
+	 */
+	.maxsize = PAGE_SIZE * (VIRTQUEUE_NUM - 3),
+	.def = 1,
+	.owner = THIS_MODULE,
+};
+
+/* The standard init function */
+static int __init p9_virtio_init(void)
+{
+	INIT_LIST_HEAD(&virtio_chan_list);
+
+	v9fs_register_trans(&p9_virtio_trans);
+	return register_virtio_driver(&p9_virtio_drv);
+}
+
+static void __exit p9_virtio_cleanup(void)
+{
+	unregister_virtio_driver(&p9_virtio_drv);
+	v9fs_unregister_trans(&p9_virtio_trans);
+}
+
+module_init(p9_virtio_init);
+module_exit(p9_virtio_cleanup);
+
+MODULE_DEVICE_TABLE(virtio, id_table);
+MODULE_AUTHOR("Eric Van Hensbergen <ericvh@gmail.com>");
+MODULE_DESCRIPTION("Virtio 9p Transport");
+MODULE_LICENSE("GPL");

net/9p/util.c

@@ -0,0 +1,141 @@
+/*
+ *  net/9p/util.c
+ *
+ *  This file contains some helper functions
+ *
+ *  Copyright (C) 2007 by Latchesar Ionkov <lucho@ionkov.net>
+ *  Copyright (C) 2004 by Eric Van Hensbergen <ericvh@gmail.com>
+ *  Copyright (C) 2002 by Ron Minnich <rminnich@lanl.gov>
+ *
+ *  This program is free software; you can redistribute it and/or modify
+ *  it under the terms of the GNU General Public License version 2
+ *  as published by the Free Software Foundation.
+ *
+ *  This program is distributed in the hope that it will be useful,
+ *  but WITHOUT ANY WARRANTY; without even the implied warranty of
+ *  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ *  GNU General Public License for more details.
+ *
+ *  You should have received a copy of the GNU General Public License
+ *  along with this program; if not, write to:
+ *  Free Software Foundation
+ *  51 Franklin Street, Fifth Floor
+ *  Boston, MA  02111-1301  USA
+ *
+ */
+
+#include <linux/module.h>
+#include <linux/errno.h>
+#include <linux/fs.h>
+#include <linux/sched.h>
+#include <linux/parser.h>
+#include <linux/idr.h>
+#include <linux/slab.h>
+#include <net/9p/9p.h>
+
+/**
+ * struct p9_idpool - per-connection accounting for tag idpool
+ * @lock: protects the pool
+ * @pool: idr to allocate tag id from
+ *
+ */
+
+struct p9_idpool {
+	spinlock_t lock;
+	struct idr pool;
+};
+
+/**
+ * p9_idpool_create - create a new per-connection id pool
+ *
+ */
+
+struct p9_idpool *p9_idpool_create(void)
+{
+	struct p9_idpool *p;
+
+	p = kmalloc(sizeof(struct p9_idpool), GFP_KERNEL);
+	if (!p)
+		return ERR_PTR(-ENOMEM);
+
+	spin_lock_init(&p->lock);
+	idr_init(&p->pool);
+
+	return p;
+}
+EXPORT_SYMBOL(p9_idpool_create);
+
+/**
+ * p9_idpool_destroy - create a new per-connection id pool
+ * @p: idpool to destroy
+ */
+
+void p9_idpool_destroy(struct p9_idpool *p)
+{
+	idr_destroy(&p->pool);
+	kfree(p);
+}
+EXPORT_SYMBOL(p9_idpool_destroy);
+
+/**
+ * p9_idpool_get - allocate numeric id from pool
+ * @p: pool to allocate from
+ *
+ * Bugs: This seems to be an awful generic function, should it be in idr.c with
+ *            the lock included in struct idr?
+ */
+
+int p9_idpool_get(struct p9_idpool *p)
+{
+	int i;
+	unsigned long flags;
+
+	idr_preload(GFP_NOFS);
+	spin_lock_irqsave(&p->lock, flags);
+
+	/* no need to store exactly p, we just need something non-null */
+	i = idr_alloc(&p->pool, p, 0, 0, GFP_NOWAIT);
+
+	spin_unlock_irqrestore(&p->lock, flags);
+	idr_preload_end();
+	if (i < 0)
+		return -1;
+
+	p9_debug(P9_DEBUG_MUX, " id %d pool %p\n", i, p);
+	return i;
+}
+EXPORT_SYMBOL(p9_idpool_get);
+
+/**
+ * p9_idpool_put - release numeric id from pool
+ * @id: numeric id which is being released
+ * @p: pool to release id into
+ *
+ * Bugs: This seems to be an awful generic function, should it be in idr.c with
+ *            the lock included in struct idr?
+ */
+
+void p9_idpool_put(int id, struct p9_idpool *p)
+{
+	unsigned long flags;
+
+	p9_debug(P9_DEBUG_MUX, " id %d pool %p\n", id, p);
+
+	spin_lock_irqsave(&p->lock, flags);
+	idr_remove(&p->pool, id);
+	spin_unlock_irqrestore(&p->lock, flags);
+}
+EXPORT_SYMBOL(p9_idpool_put);
+
+/**
+ * p9_idpool_check - check if the specified id is available
+ * @id: id to check
+ * @p: pool to check
+ */
+
+int p9_idpool_check(int id, struct p9_idpool *p)
+{
+	return idr_find(&p->pool, id) != NULL;
+}
+EXPORT_SYMBOL(p9_idpool_check);
+