Import xenbus sources, from linux sparse tree in the 20060107 xen-3.0

snapshot.

sys/arch/xen/xenbus/Makefile

@@ -0,0 +1,8 @@
+obj-y	+= xenbus.o
+
+xenbus-objs =
+xenbus-objs += xenbus_client.o 
+xenbus-objs += xenbus_comms.o
+xenbus-objs += xenbus_xs.o
+xenbus-objs += xenbus_probe.o 
+xenbus-objs += xenbus_dev.o 

sys/arch/xen/xenbus/xenbus_client.c

@@ -0,0 +1,256 @@
+/******************************************************************************
+ * Client-facing interface for the Xenbus driver.  In other words, the
+ * interface between the Xenbus and the device-specific code, be it the
+ * frontend or the backend of that driver.
+ *
+ * Copyright (C) 2005 XenSource Ltd
+ * 
+ * This file may be distributed separately from the Linux kernel, or
+ * incorporated into other software packages, subject to the following license:
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this source file (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy, modify,
+ * merge, publish, distribute, sublicense, and/or sell copies of the Software,
+ * and to permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ * 
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
+ * IN THE SOFTWARE.
+ */
+
+
+#if 0
+#define DPRINTK(fmt, args...) \
+    printk("xenbus_client (%s:%d) " fmt ".\n", __FUNCTION__, __LINE__, ##args)
+#else
+#define DPRINTK(fmt, args...) ((void)0)
+#endif
+
+
+#include <asm-xen/evtchn.h>
+#include <asm-xen/gnttab.h>
+#include <asm-xen/xenbus.h>
+
+
+int xenbus_watch_path(struct xenbus_device *dev, const char *path,
+		      struct xenbus_watch *watch, 
+		      void (*callback)(struct xenbus_watch *,
+				       const char **, unsigned int))
+{
+	int err;
+
+	watch->node = path;
+	watch->callback = callback;
+
+	err = register_xenbus_watch(watch);
+
+	if (err) {
+		watch->node = NULL;
+		watch->callback = NULL;
+		xenbus_dev_fatal(dev, err, "adding watch on %s", path);
+	}
+
+	return err;
+}
+EXPORT_SYMBOL(xenbus_watch_path);
+
+
+int xenbus_watch_path2(struct xenbus_device *dev, const char *path,
+		       const char *path2, struct xenbus_watch *watch, 
+		       void (*callback)(struct xenbus_watch *,
+					const char **, unsigned int))
+{
+	int err;
+	char *state =
+		kmalloc(strlen(path) + 1 + strlen(path2) + 1, GFP_KERNEL);
+	if (!state) {
+		xenbus_dev_fatal(dev, -ENOMEM, "allocating path for watch");
+		return -ENOMEM;
+	}
+	strcpy(state, path);
+	strcat(state, "/");
+	strcat(state, path2);
+
+	err = xenbus_watch_path(dev, state, watch, callback);
+
+	if (err) {
+		kfree(state);
+	}
+	return err;
+}
+EXPORT_SYMBOL(xenbus_watch_path2);
+
+
+int xenbus_switch_state(struct xenbus_device *dev,
+			struct xenbus_transaction *xbt,
+			XenbusState state)
+{
+	/* We check whether the state is currently set to the given value, and
+	   if not, then the state is set.  We don't want to unconditionally
+	   write the given state, because we don't want to fire watches
+	   unnecessarily.  Furthermore, if the node has gone, we don't write
+	   to it, as the device will be tearing down, and we don't want to
+	   resurrect that directory.
+	 */
+
+	int current_state;
+
+	int err = xenbus_scanf(xbt, dev->nodename, "state", "%d",
+			       &current_state);
+	if ((err == 1 && (XenbusState)current_state == state) ||
+	    err == -ENOENT)
+		return 0;
+
+	err = xenbus_printf(xbt, dev->nodename, "state", "%d", state);
+	if (err) {
+		xenbus_dev_fatal(dev, err, "writing new state");
+		return err;
+	}
+	return 0;
+}
+EXPORT_SYMBOL(xenbus_switch_state);
+
+
+/**
+ * Return the path to the error node for the given device, or NULL on failure.
+ * If the value returned is non-NULL, then it is the caller's to kfree.
+ */
+static char *error_path(struct xenbus_device *dev)
+{
+	char *path_buffer = kmalloc(strlen("error/") + strlen(dev->nodename) +
+				    1, GFP_KERNEL);
+	if (path_buffer == NULL) {
+		return NULL;
+	}
+
+	strcpy(path_buffer, "error/");
+	strcpy(path_buffer + strlen("error/"), dev->nodename);
+
+	return path_buffer;
+}
+
+
+void _dev_error(struct xenbus_device *dev, int err, const char *fmt,
+		va_list ap)
+{
+	int ret;
+	unsigned int len;
+	char *printf_buffer = NULL, *path_buffer = NULL;
+
+#define PRINTF_BUFFER_SIZE 4096
+	printf_buffer = kmalloc(PRINTF_BUFFER_SIZE, GFP_KERNEL);
+	if (printf_buffer == NULL)
+		goto fail;
+
+	len = sprintf(printf_buffer, "%i ", -err);
+	ret = vsnprintf(printf_buffer+len, PRINTF_BUFFER_SIZE-len, fmt, ap);
+
+	BUG_ON(len + ret > PRINTF_BUFFER_SIZE-1);
+	dev->has_error = 1;
+
+	path_buffer = error_path(dev);
+
+	if (path_buffer == NULL) {
+		printk("xenbus: failed to write error node for %s (%s)\n",
+		       dev->nodename, printf_buffer);
+		goto fail;
+	}
+
+	if (xenbus_write(NULL, path_buffer, "error", printf_buffer) != 0) {
+		printk("xenbus: failed to write error node for %s (%s)\n",
+		       dev->nodename, printf_buffer);
+		goto fail;
+	}
+
+fail:
+	if (printf_buffer)
+		kfree(printf_buffer);
+	if (path_buffer)
+		kfree(path_buffer);
+}
+
+
+void xenbus_dev_error(struct xenbus_device *dev, int err, const char *fmt,
+		      ...)
+{
+	va_list ap;
+
+	va_start(ap, fmt);
+	_dev_error(dev, err, fmt, ap);
+	va_end(ap);
+}
+EXPORT_SYMBOL(xenbus_dev_error);
+
+
+void xenbus_dev_fatal(struct xenbus_device *dev, int err, const char *fmt,
+		      ...)
+{
+	va_list ap;
+
+	va_start(ap, fmt);
+	_dev_error(dev, err, fmt, ap);
+	va_end(ap);
+	
+	xenbus_switch_state(dev, NULL, XenbusStateClosing);
+}
+EXPORT_SYMBOL(xenbus_dev_fatal);
+
+
+int xenbus_grant_ring(struct xenbus_device *dev, unsigned long ring_mfn)
+{
+	int err = gnttab_grant_foreign_access(dev->otherend_id, ring_mfn, 0);
+	if (err < 0)
+		xenbus_dev_fatal(dev, err, "granting access to ring page");
+	return err;
+}
+EXPORT_SYMBOL(xenbus_grant_ring);
+
+
+int xenbus_alloc_evtchn(struct xenbus_device *dev, int *port)
+{
+	evtchn_op_t op = {
+		.cmd = EVTCHNOP_alloc_unbound,
+		.u.alloc_unbound.dom = DOMID_SELF,
+		.u.alloc_unbound.remote_dom = dev->otherend_id };
+
+	int err = HYPERVISOR_event_channel_op(&op);
+	if (err)
+		xenbus_dev_fatal(dev, err, "allocating event channel");
+	else
+		*port = op.u.alloc_unbound.port;
+	return err;
+}
+EXPORT_SYMBOL(xenbus_alloc_evtchn);
+
+
+XenbusState xenbus_read_driver_state(const char *path)
+{
+	XenbusState result;
+
+	int err = xenbus_gather(NULL, path, "state", "%d", &result, NULL);
+	if (err)
+		result = XenbusStateClosed;
+
+	return result;
+}
+EXPORT_SYMBOL(xenbus_read_driver_state);
+
+
+/*
+ * Local variables:
+ *  c-file-style: "linux"
+ *  indent-tabs-mode: t
+ *  c-indent-level: 8
+ *  c-basic-offset: 8
+ *  tab-width: 8
+ * End:
+ */

sys/arch/xen/xenbus/xenbus_comms.c

@@ -0,0 +1,203 @@
+/******************************************************************************
+ * xenbus_comms.c
+ *
+ * Low level code to talks to Xen Store: ringbuffer and event channel.
+ *
+ * Copyright (C) 2005 Rusty Russell, IBM Corporation
+ * 
+ * This file may be distributed separately from the Linux kernel, or
+ * incorporated into other software packages, subject to the following license:
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this source file (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy, modify,
+ * merge, publish, distribute, sublicense, and/or sell copies of the Software,
+ * and to permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ * 
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
+ * IN THE SOFTWARE.
+ */
+
+#include <asm/hypervisor.h>
+#include <asm-xen/evtchn.h>
+#include <linux/wait.h>
+#include <linux/interrupt.h>
+#include <linux/sched.h>
+#include <linux/err.h>
+#include <asm-xen/xenbus.h>
+#include "xenbus_comms.h"
+
+static int xenbus_irq;
+
+extern void xenbus_probe(void *); 
+extern int xenstored_ready; 
+static DECLARE_WORK(probe_work, xenbus_probe, NULL);
+
+DECLARE_WAIT_QUEUE_HEAD(xb_waitq);
+
+static inline struct xenstore_domain_interface *xenstore_domain_interface(void)
+{
+	return mfn_to_virt(xen_start_info->store_mfn);
+}
+
+static irqreturn_t wake_waiting(int irq, void *unused, struct pt_regs *regs)
+{
+	if (unlikely(xenstored_ready == 0)) {
+		xenstored_ready = 1; 
+		schedule_work(&probe_work); 
+	} 
+
+	wake_up(&xb_waitq);
+	return IRQ_HANDLED;
+}
+
+static int check_indexes(XENSTORE_RING_IDX cons, XENSTORE_RING_IDX prod)
+{
+	return ((prod - cons) <= XENSTORE_RING_SIZE);
+}
+
+static void *get_output_chunk(XENSTORE_RING_IDX cons,
+			      XENSTORE_RING_IDX prod,
+			      char *buf, uint32_t *len)
+{
+	*len = XENSTORE_RING_SIZE - MASK_XENSTORE_IDX(prod);
+	if ((XENSTORE_RING_SIZE - (prod - cons)) < *len)
+		*len = XENSTORE_RING_SIZE - (prod - cons);
+	return buf + MASK_XENSTORE_IDX(prod);
+}
+
+static const void *get_input_chunk(XENSTORE_RING_IDX cons,
+				   XENSTORE_RING_IDX prod,
+				   const char *buf, uint32_t *len)
+{
+	*len = XENSTORE_RING_SIZE - MASK_XENSTORE_IDX(cons);
+	if ((prod - cons) < *len)
+		*len = prod - cons;
+	return buf + MASK_XENSTORE_IDX(cons);
+}
+
+int xb_write(const void *data, unsigned len)
+{
+	struct xenstore_domain_interface *intf = xenstore_domain_interface();
+	XENSTORE_RING_IDX cons, prod;
+
+	while (len != 0) {
+		void *dst;
+		unsigned int avail;
+
+		wait_event_interruptible(xb_waitq,
+					 (intf->req_prod - intf->req_cons) !=
+					 XENSTORE_RING_SIZE);
+
+		/* Read indexes, then verify. */
+		cons = intf->req_cons;
+		prod = intf->req_prod;
+		mb();
+		if (!check_indexes(cons, prod))
+			return -EIO;
+
+		dst = get_output_chunk(cons, prod, intf->req, &avail);
+		if (avail == 0)
+			continue;
+		if (avail > len)
+			avail = len;
+
+		memcpy(dst, data, avail);
+		data += avail;
+		len -= avail;
+
+		/* Other side must not see new header until data is there. */
+		wmb();
+		intf->req_prod += avail;
+
+		/* This implies mb() before other side sees interrupt. */
+		notify_remote_via_evtchn(xen_start_info->store_evtchn);
+	}
+
+	return 0;
+}
+
+int xb_read(void *data, unsigned len)
+{
+	struct xenstore_domain_interface *intf = xenstore_domain_interface();
+	XENSTORE_RING_IDX cons, prod;
+
+	while (len != 0) {
+		unsigned int avail;
+		const char *src;
+
+		wait_event_interruptible(xb_waitq,
+					 intf->rsp_cons != intf->rsp_prod);
+
+		/* Read indexes, then verify. */
+		cons = intf->rsp_cons;
+		prod = intf->rsp_prod;
+		mb();
+		if (!check_indexes(cons, prod))
+			return -EIO;
+
+		src = get_input_chunk(cons, prod, intf->rsp, &avail);
+		if (avail == 0)
+			continue;
+		if (avail > len)
+			avail = len;
+
+		/* We must read header before we read data. */
+		rmb();
+
+		memcpy(data, src, avail);
+		data += avail;
+		len -= avail;
+
+		/* Other side must not see free space until we've copied out */
+		mb();
+		intf->rsp_cons += avail;
+
+		pr_debug("Finished read of %i bytes (%i to go)\n", avail, len);
+
+		/* Implies mb(): they will see new header. */
+		notify_remote_via_evtchn(xen_start_info->store_evtchn);
+	}
+
+	return 0;
+}
+
+/* Set up interrupt handler off store event channel. */
+int xb_init_comms(void)
+{
+	int err;
+
+	if (xenbus_irq)
+		unbind_from_irqhandler(xenbus_irq, &xb_waitq);
+
+	err = bind_evtchn_to_irqhandler(
+		xen_start_info->store_evtchn, wake_waiting,
+		0, "xenbus", &xb_waitq);
+	if (err <= 0) {
+		printk(KERN_ERR "XENBUS request irq failed %i\n", err);
+		return err;
+	}
+
+	xenbus_irq = err;
+
+	return 0;
+}
+
+/*
+ * Local variables:
+ *  c-file-style: "linux"
+ *  indent-tabs-mode: t
+ *  c-indent-level: 8
+ *  c-basic-offset: 8
+ *  tab-width: 8
+ * End:
+ */

sys/arch/xen/xenbus/xenbus_comms.h

@@ -0,0 +1,50 @@
+/*
+ * Private include for xenbus communications.
+ * 
+ * Copyright (C) 2005 Rusty Russell, IBM Corporation
+ *
+ * This file may be distributed separately from the Linux kernel, or
+ * incorporated into other software packages, subject to the following license:
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this source file (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy, modify,
+ * merge, publish, distribute, sublicense, and/or sell copies of the Software,
+ * and to permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ * 
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
+ * IN THE SOFTWARE.
+ */
+
+#ifndef _XENBUS_COMMS_H
+#define _XENBUS_COMMS_H
+
+int xs_init(void);
+int xb_init_comms(void);
+
+/* Low level routines. */
+int xb_write(const void *data, unsigned len);
+int xb_read(void *data, unsigned len);
+int xs_input_avail(void);
+extern wait_queue_head_t xb_waitq;
+
+#endif /* _XENBUS_COMMS_H */
+
+/*
+ * Local variables:
+ *  c-file-style: "linux"
+ *  indent-tabs-mode: t
+ *  c-indent-level: 8
+ *  c-basic-offset: 8
+ *  tab-width: 8
+ * End:
+ */

sys/arch/xen/xenbus/xenbus_dev.c

@@ -0,0 +1,241 @@
+/*
+ * xenbus_dev.c
+ * 
+ * Driver giving user-space access to the kernel's xenbus connection
+ * to xenstore.
+ * 
+ * Copyright (c) 2005, Christian Limpach
+ * Copyright (c) 2005, Rusty Russell, IBM Corporation
+ * 
+ * This file may be distributed separately from the Linux kernel, or
+ * incorporated into other software packages, subject to the following license:
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this source file (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy, modify,
+ * merge, publish, distribute, sublicense, and/or sell copies of the Software,
+ * and to permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ * 
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
+ * IN THE SOFTWARE.
+ */
+
+#include <linux/config.h>
+#include <linux/kernel.h>
+#include <linux/errno.h>
+#include <linux/uio.h>
+#include <linux/notifier.h>
+#include <linux/wait.h>
+#include <linux/fs.h>
+
+#include "xenbus_comms.h"
+
+#include <asm/uaccess.h>
+#include <asm/hypervisor.h>
+#include <asm-xen/xenbus.h>
+#include <asm-xen/xen_proc.h>
+#include <asm/hypervisor.h>
+
+struct xenbus_dev_transaction {
+	struct list_head list;
+	struct xenbus_transaction *handle;
+};
+
+struct xenbus_dev_data {
+	/* In-progress transaction. */
+	struct list_head transactions;
+
+	/* Partial request. */
+	unsigned int len;
+	union {
+		struct xsd_sockmsg msg;
+		char buffer[PAGE_SIZE];
+	} u;
+
+	/* Response queue. */
+#define MASK_READ_IDX(idx) ((idx)&(PAGE_SIZE-1))
+	char read_buffer[PAGE_SIZE];
+	unsigned int read_cons, read_prod;
+	wait_queue_head_t read_waitq;
+};
+
+static struct proc_dir_entry *xenbus_dev_intf;
+
+static ssize_t xenbus_dev_read(struct file *filp,
+			       char __user *ubuf,
+			       size_t len, loff_t *ppos)
+{
+	struct xenbus_dev_data *u = filp->private_data;
+	int i;
+
+	if (wait_event_interruptible(u->read_waitq,
+				     u->read_prod != u->read_cons))
+		return -EINTR;
+
+	for (i = 0; i < len; i++) {
+		if (u->read_cons == u->read_prod)
+			break;
+		put_user(u->read_buffer[MASK_READ_IDX(u->read_cons)], ubuf+i);
+		u->read_cons++;
+	}
+
+	return i;
+}
+
+static void queue_reply(struct xenbus_dev_data *u,
+			char *data, unsigned int len)
+{
+	int i;
+
+	for (i = 0; i < len; i++, u->read_prod++)
+		u->read_buffer[MASK_READ_IDX(u->read_prod)] = data[i];
+
+	BUG_ON((u->read_prod - u->read_cons) > sizeof(u->read_buffer));
+
+	wake_up(&u->read_waitq);
+}
+
+static ssize_t xenbus_dev_write(struct file *filp,
+				const char __user *ubuf,
+				size_t len, loff_t *ppos)
+{
+	struct xenbus_dev_data *u = filp->private_data;
+	struct xenbus_dev_transaction *trans;
+	void *reply;
+	int err = 0;
+
+	if ((len + u->len) > sizeof(u->u.buffer))
+		return -EINVAL;
+
+	if (copy_from_user(u->u.buffer + u->len, ubuf, len) != 0)
+		return -EFAULT;
+
+	u->len += len;
+	if (u->len < (sizeof(u->u.msg) + u->u.msg.len))
+		return len;
+
+	switch (u->u.msg.type) {
+	case XS_TRANSACTION_START:
+	case XS_TRANSACTION_END:
+	case XS_DIRECTORY:
+	case XS_READ:
+	case XS_GET_PERMS:
+	case XS_RELEASE:
+	case XS_GET_DOMAIN_PATH:
+	case XS_WRITE:
+	case XS_MKDIR:
+	case XS_RM:
+	case XS_SET_PERMS:
+		reply = xenbus_dev_request_and_reply(&u->u.msg);
+		if (IS_ERR(reply)) {
+			err = PTR_ERR(reply);
+		} else {
+			if (u->u.msg.type == XS_TRANSACTION_START) {
+				trans = kmalloc(sizeof(*trans), GFP_KERNEL);
+				trans->handle = (struct xenbus_transaction *)
+					simple_strtoul(reply, NULL, 0);
+				list_add(&trans->list, &u->transactions);
+			} else if (u->u.msg.type == XS_TRANSACTION_END) {
+				list_for_each_entry(trans, &u->transactions,
+						    list)
+					if ((unsigned long)trans->handle ==
+					    (unsigned long)u->u.msg.tx_id)
+						break;
+				BUG_ON(&trans->list == &u->transactions);
+				list_del(&trans->list);
+				kfree(trans);
+			}
+			queue_reply(u, (char *)&u->u.msg, sizeof(u->u.msg));
+			queue_reply(u, (char *)reply, u->u.msg.len);
+			kfree(reply);
+		}
+		break;
+
+	default:
+		err = -EINVAL;
+		break;
+	}
+
+	if (err == 0) {
+		u->len = 0;
+		err = len;
+	}
+
+	return err;
+}
+
+static int xenbus_dev_open(struct inode *inode, struct file *filp)
+{
+	struct xenbus_dev_data *u;
+
+	if (xen_start_info->store_evtchn == 0)
+		return -ENOENT;
+
+	nonseekable_open(inode, filp);
+
+	u = kmalloc(sizeof(*u), GFP_KERNEL);
+	if (u == NULL)
+		return -ENOMEM;
+
+	memset(u, 0, sizeof(*u));
+	INIT_LIST_HEAD(&u->transactions);
+	init_waitqueue_head(&u->read_waitq);
+
+	filp->private_data = u;
+
+	return 0;
+}
+
+static int xenbus_dev_release(struct inode *inode, struct file *filp)
+{
+	struct xenbus_dev_data *u = filp->private_data;
+	struct xenbus_dev_transaction *trans, *tmp;
+
+	list_for_each_entry_safe(trans, tmp, &u->transactions, list) {
+		xenbus_transaction_end(trans->handle, 1);
+		list_del(&trans->list);
+		kfree(trans);
+	}
+
+	kfree(u);
+
+	return 0;
+}
+
+static struct file_operations xenbus_dev_file_ops = {
+	.read = xenbus_dev_read,
+	.write = xenbus_dev_write,
+	.open = xenbus_dev_open,
+	.release = xenbus_dev_release,
+};
+
+static int __init
+xenbus_dev_init(void)
+{
+	xenbus_dev_intf = create_xen_proc_entry("xenbus", 0400);
+	if (xenbus_dev_intf)
+		xenbus_dev_intf->proc_fops = &xenbus_dev_file_ops;
+
+	return 0;
+}
+
+__initcall(xenbus_dev_init);
+
+/*
+ * Local variables:
+ *  c-file-style: "linux"
+ *  indent-tabs-mode: t
+ *  c-indent-level: 8
+ *  c-basic-offset: 8
+ *  tab-width: 8
+ * End:
+ */

sys/arch/xen/xenbus/xenbus_xs.c

@@ -0,0 +1,827 @@
+/******************************************************************************
+ * xenbus_xs.c
+ *
+ * This is the kernel equivalent of the "xs" library.  We don't need everything
+ * and we use xenbus_comms for communication.
+ *
+ * Copyright (C) 2005 Rusty Russell, IBM Corporation
+ * 
+ * This file may be distributed separately from the Linux kernel, or
+ * incorporated into other software packages, subject to the following license:
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this source file (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy, modify,
+ * merge, publish, distribute, sublicense, and/or sell copies of the Software,
+ * and to permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ * 
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
+ * IN THE SOFTWARE.
+ */
+
+#include <linux/unistd.h>
+#include <linux/errno.h>
+#include <linux/types.h>
+#include <linux/uio.h>
+#include <linux/kernel.h>
+#include <linux/string.h>
+#include <linux/err.h>
+#include <linux/slab.h>
+#include <linux/fcntl.h>
+#include <linux/kthread.h>
+#include <asm-xen/xenbus.h>
+#include "xenbus_comms.h"
+
+#define streq(a, b) (strcmp((a), (b)) == 0)
+
+struct xs_stored_msg {
+	struct list_head list;
+
+	struct xsd_sockmsg hdr;
+
+	union {
+		/* Queued replies. */
+		struct {
+			char *body;
+		} reply;
+
+		/* Queued watch events. */
+		struct {
+			struct xenbus_watch *handle;
+			char **vec;
+			unsigned int vec_size;
+		} watch;
+	} u;
+};
+
+struct xs_handle {
+	/* A list of replies. Currently only one will ever be outstanding. */
+	struct list_head reply_list;
+	spinlock_t reply_lock;
+	wait_queue_head_t reply_waitq;
+
+	/* One request at a time. */
+	struct semaphore request_mutex;
+
+	/* Protect transactions against save/restore. */
+	struct rw_semaphore suspend_mutex;
+};
+
+static struct xs_handle xs_state;
+
+/* List of registered watches, and a lock to protect it. */
+static LIST_HEAD(watches);
+static DEFINE_SPINLOCK(watches_lock);
+
+/* List of pending watch callback events, and a lock to protect it. */
+static LIST_HEAD(watch_events);
+static DEFINE_SPINLOCK(watch_events_lock);
+
+/*
+ * Details of the xenwatch callback kernel thread. The thread waits on the
+ * watch_events_waitq for work to do (queued on watch_events list). When it
+ * wakes up it acquires the xenwatch_mutex before reading the list and
+ * carrying out work.
+ */
+static pid_t xenwatch_pid;
+/* static */ DECLARE_MUTEX(xenwatch_mutex);
+static DECLARE_WAIT_QUEUE_HEAD(watch_events_waitq);
+
+static int get_error(const char *errorstring)
+{
+	unsigned int i;
+
+	for (i = 0; !streq(errorstring, xsd_errors[i].errstring); i++) {
+		if (i == ARRAY_SIZE(xsd_errors) - 1) {
+			printk(KERN_WARNING
+			       "XENBUS xen store gave: unknown error %s",
+			       errorstring);
+			return EINVAL;
+		}
+	}
+	return xsd_errors[i].errnum;
+}
+
+static void *read_reply(enum xsd_sockmsg_type *type, unsigned int *len)
+{
+	struct xs_stored_msg *msg;
+	char *body;
+
+	spin_lock(&xs_state.reply_lock);
+
+	while (list_empty(&xs_state.reply_list)) {
+		spin_unlock(&xs_state.reply_lock);
+		wait_event_interruptible(xs_state.reply_waitq,
+					 !list_empty(&xs_state.reply_list));
+		spin_lock(&xs_state.reply_lock);
+	}
+
+	msg = list_entry(xs_state.reply_list.next,
+			 struct xs_stored_msg, list);
+	list_del(&msg->list);
+
+	spin_unlock(&xs_state.reply_lock);
+
+	*type = msg->hdr.type;
+	if (len)
+		*len = msg->hdr.len;
+	body = msg->u.reply.body;
+
+	kfree(msg);
+
+	return body;
+}
+
+/* Emergency write. */
+void xenbus_debug_write(const char *str, unsigned int count)
+{
+	struct xsd_sockmsg msg = { 0 };
+
+	msg.type = XS_DEBUG;
+	msg.len = sizeof("print") + count + 1;
+
+	down(&xs_state.request_mutex);
+	xb_write(&msg, sizeof(msg));
+	xb_write("print", sizeof("print"));
+	xb_write(str, count);
+	xb_write("", 1);
+	up(&xs_state.request_mutex);
+}
+
+void *xenbus_dev_request_and_reply(struct xsd_sockmsg *msg)
+{
+	void *ret;
+	struct xsd_sockmsg req_msg = *msg;
+	int err;
+
+	if (req_msg.type == XS_TRANSACTION_START)
+		down_read(&xs_state.suspend_mutex);
+
+	down(&xs_state.request_mutex);
+
+	err = xb_write(msg, sizeof(*msg) + msg->len);
+	if (err) {
+		msg->type = XS_ERROR;
+		ret = ERR_PTR(err);
+	} else {
+		ret = read_reply(&msg->type, &msg->len);
+	}
+
+	up(&xs_state.request_mutex);
+
+	if ((msg->type == XS_TRANSACTION_END) ||
+	    ((req_msg.type == XS_TRANSACTION_START) &&
+	     (msg->type == XS_ERROR)))
+		up_read(&xs_state.suspend_mutex);
+
+	return ret;
+}
+
+/* Send message to xs, get kmalloc'ed reply.  ERR_PTR() on error. */
+static void *xs_talkv(struct xenbus_transaction *t,
+		      enum xsd_sockmsg_type type,
+		      const struct kvec *iovec,
+		      unsigned int num_vecs,
+		      unsigned int *len)
+{
+	struct xsd_sockmsg msg;
+	void *ret = NULL;
+	unsigned int i;
+	int err;
+
+	msg.tx_id = (u32)(unsigned long)t;
+	msg.req_id = 0;
+	msg.type = type;
+	msg.len = 0;
+	for (i = 0; i < num_vecs; i++)
+		msg.len += iovec[i].iov_len;
+
+	down(&xs_state.request_mutex);
+
+	err = xb_write(&msg, sizeof(msg));
+	if (err) {
+		up(&xs_state.request_mutex);
+		return ERR_PTR(err);
+	}
+
+	for (i = 0; i < num_vecs; i++) {
+		err = xb_write(iovec[i].iov_base, iovec[i].iov_len);;
+		if (err) {
+			up(&xs_state.request_mutex);
+			return ERR_PTR(err);
+		}
+	}
+
+	ret = read_reply(&msg.type, len);
+
+	up(&xs_state.request_mutex);
+
+	if (IS_ERR(ret))
+		return ret;
+
+	if (msg.type == XS_ERROR) {
+		err = get_error(ret);
+		kfree(ret);
+		return ERR_PTR(-err);
+	}
+
+	BUG_ON(msg.type != type);
+	return ret;
+}
+
+/* Simplified version of xs_talkv: single message. */
+static void *xs_single(struct xenbus_transaction *t,
+		       enum xsd_sockmsg_type type,
+		       const char *string,
+		       unsigned int *len)
+{
+	struct kvec iovec;
+
+	iovec.iov_base = (void *)string;
+	iovec.iov_len = strlen(string) + 1;
+	return xs_talkv(t, type, &iovec, 1, len);
+}
+
+/* Many commands only need an ack, don't care what it says. */
+static int xs_error(char *reply)
+{
+	if (IS_ERR(reply))
+		return PTR_ERR(reply);
+	kfree(reply);
+	return 0;
+}
+
+static unsigned int count_strings(const char *strings, unsigned int len)
+{
+	unsigned int num;
+	const char *p;
+
+	for (p = strings, num = 0; p < strings + len; p += strlen(p) + 1)
+		num++;
+
+	return num;
+}
+
+/* Return the path to dir with /name appended. Buffer must be kfree()'ed. */ 
+static char *join(const char *dir, const char *name)
+{
+	char *buffer;
+
+	buffer = kmalloc(strlen(dir) + strlen("/") + strlen(name) + 1,
+			 GFP_KERNEL);
+	if (buffer == NULL)
+		return ERR_PTR(-ENOMEM);
+
+	strcpy(buffer, dir);
+	if (!streq(name, "")) {
+		strcat(buffer, "/");
+		strcat(buffer, name);
+	}
+
+	return buffer;
+}
+
+static char **split(char *strings, unsigned int len, unsigned int *num)
+{
+	char *p, **ret;
+
+	/* Count the strings. */
+	*num = count_strings(strings, len);
+
+	/* Transfer to one big alloc for easy freeing. */
+	ret = kmalloc(*num * sizeof(char *) + len, GFP_KERNEL);
+	if (!ret) {
+		kfree(strings);
+		return ERR_PTR(-ENOMEM);
+	}
+	memcpy(&ret[*num], strings, len);
+	kfree(strings);
+
+	strings = (char *)&ret[*num];
+	for (p = strings, *num = 0; p < strings + len; p += strlen(p) + 1)
+		ret[(*num)++] = p;
+
+	return ret;
+}
+
+char **xenbus_directory(struct xenbus_transaction *t,
+			const char *dir, const char *node, unsigned int *num)
+{
+	char *strings, *path;
+	unsigned int len;
+
+	path = join(dir, node);
+	if (IS_ERR(path))
+		return (char **)path;
+
+	strings = xs_single(t, XS_DIRECTORY, path, &len);
+	kfree(path);
+	if (IS_ERR(strings))
+		return (char **)strings;
+
+	return split(strings, len, num);
+}
+EXPORT_SYMBOL(xenbus_directory);
+
+/* Check if a path exists. Return 1 if it does. */
+int xenbus_exists(struct xenbus_transaction *t,
+		  const char *dir, const char *node)
+{
+	char **d;
+	int dir_n;
+
+	d = xenbus_directory(t, dir, node, &dir_n);
+	if (IS_ERR(d))
+		return 0;
+	kfree(d);
+	return 1;
+}
+EXPORT_SYMBOL(xenbus_exists);
+
+/* Get the value of a single file.
+ * Returns a kmalloced value: call free() on it after use.
+ * len indicates length in bytes.
+ */
+void *xenbus_read(struct xenbus_transaction *t,
+		  const char *dir, const char *node, unsigned int *len)
+{
+	char *path;
+	void *ret;
+
+	path = join(dir, node);
+	if (IS_ERR(path))
+		return (void *)path;
+
+	ret = xs_single(t, XS_READ, path, len);
+	kfree(path);
+	return ret;
+}
+EXPORT_SYMBOL(xenbus_read);
+
+/* Write the value of a single file.
+ * Returns -err on failure.
+ */
+int xenbus_write(struct xenbus_transaction *t,
+		 const char *dir, const char *node, const char *string)
+{
+	const char *path;
+	struct kvec iovec[2];
+	int ret;
+
+	path = join(dir, node);
+	if (IS_ERR(path))
+		return PTR_ERR(path);
+
+	iovec[0].iov_base = (void *)path;
+	iovec[0].iov_len = strlen(path) + 1;
+	iovec[1].iov_base = (void *)string;
+	iovec[1].iov_len = strlen(string);
+
+	ret = xs_error(xs_talkv(t, XS_WRITE, iovec, ARRAY_SIZE(iovec), NULL));
+	kfree(path);
+	return ret;
+}
+EXPORT_SYMBOL(xenbus_write);
+
+/* Create a new directory. */
+int xenbus_mkdir(struct xenbus_transaction *t,
+		 const char *dir, const char *node)
+{
+	char *path;
+	int ret;
+
+	path = join(dir, node);
+	if (IS_ERR(path))
+		return PTR_ERR(path);
+
+	ret = xs_error(xs_single(t, XS_MKDIR, path, NULL));
+	kfree(path);
+	return ret;
+}
+EXPORT_SYMBOL(xenbus_mkdir);
+
+/* Destroy a file or directory (directories must be empty). */
+int xenbus_rm(struct xenbus_transaction *t, const char *dir, const char *node)
+{
+	char *path;
+	int ret;
+
+	path = join(dir, node);
+	if (IS_ERR(path))
+		return PTR_ERR(path);
+
+	ret = xs_error(xs_single(t, XS_RM, path, NULL));
+	kfree(path);
+	return ret;
+}
+EXPORT_SYMBOL(xenbus_rm);
+
+/* Start a transaction: changes by others will not be seen during this
+ * transaction, and changes will not be visible to others until end.
+ */
+struct xenbus_transaction *xenbus_transaction_start(void)
+{
+	char *id_str;
+	unsigned long id;
+
+	down_read(&xs_state.suspend_mutex);
+
+	id_str = xs_single(NULL, XS_TRANSACTION_START, "", NULL);
+	if (IS_ERR(id_str)) {
+		up_read(&xs_state.suspend_mutex);
+		return (struct xenbus_transaction *)id_str;
+	}
+
+	id = simple_strtoul(id_str, NULL, 0);
+	kfree(id_str);
+
+	return (struct xenbus_transaction *)id;
+}
+EXPORT_SYMBOL(xenbus_transaction_start);
+
+/* End a transaction.
+ * If abandon is true, transaction is discarded instead of committed.
+ */
+int xenbus_transaction_end(struct xenbus_transaction *t, int abort)
+{
+	char abortstr[2];
+	int err;
+
+	if (abort)
+		strcpy(abortstr, "F");
+	else
+		strcpy(abortstr, "T");
+
+	err = xs_error(xs_single(t, XS_TRANSACTION_END, abortstr, NULL));
+
+	up_read(&xs_state.suspend_mutex);
+
+	return err;
+}
+EXPORT_SYMBOL(xenbus_transaction_end);
+
+/* Single read and scanf: returns -errno or num scanned. */
+int xenbus_scanf(struct xenbus_transaction *t,
+		 const char *dir, const char *node, const char *fmt, ...)
+{
+	va_list ap;
+	int ret;
+	char *val;
+
+	val = xenbus_read(t, dir, node, NULL);
+	if (IS_ERR(val))
+		return PTR_ERR(val);
+
+	va_start(ap, fmt);
+	ret = vsscanf(val, fmt, ap);
+	va_end(ap);
+	kfree(val);
+	/* Distinctive errno. */
+	if (ret == 0)
+		return -ERANGE;
+	return ret;
+}
+EXPORT_SYMBOL(xenbus_scanf);
+
+/* Single printf and write: returns -errno or 0. */
+int xenbus_printf(struct xenbus_transaction *t,
+		  const char *dir, const char *node, const char *fmt, ...)
+{
+	va_list ap;
+	int ret;
+#define PRINTF_BUFFER_SIZE 4096
+	char *printf_buffer;
+
+	printf_buffer = kmalloc(PRINTF_BUFFER_SIZE, GFP_KERNEL);
+	if (printf_buffer == NULL)
+		return -ENOMEM;
+
+	va_start(ap, fmt);
+	ret = vsnprintf(printf_buffer, PRINTF_BUFFER_SIZE, fmt, ap);
+	va_end(ap);
+
+	BUG_ON(ret > PRINTF_BUFFER_SIZE-1);
+	ret = xenbus_write(t, dir, node, printf_buffer);
+
+	kfree(printf_buffer);
+
+	return ret;
+}
+EXPORT_SYMBOL(xenbus_printf);
+
+/* Takes tuples of names, scanf-style args, and void **, NULL terminated. */
+int xenbus_gather(struct xenbus_transaction *t, const char *dir, ...)
+{
+	va_list ap;
+	const char *name;
+	int ret = 0;
+
+	va_start(ap, dir);
+	while (ret == 0 && (name = va_arg(ap, char *)) != NULL) {
+		const char *fmt = va_arg(ap, char *);
+		void *result = va_arg(ap, void *);
+		char *p;
+
+		p = xenbus_read(t, dir, name, NULL);
+		if (IS_ERR(p)) {
+			ret = PTR_ERR(p);
+			break;
+		}
+		if (fmt) {
+			if (sscanf(p, fmt, result) == 0)
+				ret = -EINVAL;
+			kfree(p);
+		} else
+			*(char **)result = p;
+	}
+	va_end(ap);
+	return ret;
+}
+EXPORT_SYMBOL(xenbus_gather);
+
+static int xs_watch(const char *path, const char *token)
+{
+	struct kvec iov[2];
+
+	iov[0].iov_base = (void *)path;
+	iov[0].iov_len = strlen(path) + 1;
+	iov[1].iov_base = (void *)token;
+	iov[1].iov_len = strlen(token) + 1;
+
+	return xs_error(xs_talkv(NULL, XS_WATCH, iov,
+				 ARRAY_SIZE(iov), NULL));
+}
+
+static int xs_unwatch(const char *path, const char *token)
+{
+	struct kvec iov[2];
+
+	iov[0].iov_base = (char *)path;
+	iov[0].iov_len = strlen(path) + 1;
+	iov[1].iov_base = (char *)token;
+	iov[1].iov_len = strlen(token) + 1;
+
+	return xs_error(xs_talkv(NULL, XS_UNWATCH, iov,
+				 ARRAY_SIZE(iov), NULL));
+}
+
+static struct xenbus_watch *find_watch(const char *token)
+{
+	struct xenbus_watch *i, *cmp;
+
+	cmp = (void *)simple_strtoul(token, NULL, 16);
+
+	list_for_each_entry(i, &watches, list)
+		if (i == cmp)
+			return i;
+
+	return NULL;
+}
+
+/* Register callback to watch this node. */
+int register_xenbus_watch(struct xenbus_watch *watch)
+{
+	/* Pointer in ascii is the token. */
+	char token[sizeof(watch) * 2 + 1];
+	int err;
+
+	sprintf(token, "%lX", (long)watch);
+
+	down_read(&xs_state.suspend_mutex);
+
+	spin_lock(&watches_lock);
+	BUG_ON(find_watch(token));
+	list_add(&watch->list, &watches);
+	spin_unlock(&watches_lock);
+
+	err = xs_watch(watch->node, token);
+
+	/* Ignore errors due to multiple registration. */
+	if ((err != 0) && (err != -EEXIST)) {
+		spin_lock(&watches_lock);
+		list_del(&watch->list);
+		spin_unlock(&watches_lock);
+	}
+
+	up_read(&xs_state.suspend_mutex);
+
+	return err;
+}
+EXPORT_SYMBOL(register_xenbus_watch);
+
+void unregister_xenbus_watch(struct xenbus_watch *watch)
+{
+	struct xs_stored_msg *msg, *tmp;
+	char token[sizeof(watch) * 2 + 1];
+	int err;
+
+	sprintf(token, "%lX", (long)watch);
+
+	down_read(&xs_state.suspend_mutex);
+
+	spin_lock(&watches_lock);
+	BUG_ON(!find_watch(token));
+	list_del(&watch->list);
+	spin_unlock(&watches_lock);
+
+	err = xs_unwatch(watch->node, token);
+	if (err)
+		printk(KERN_WARNING
+		       "XENBUS Failed to release watch %s: %i\n",
+		       watch->node, err);
+
+	up_read(&xs_state.suspend_mutex);
+
+	/* Cancel pending watch events. */
+	spin_lock(&watch_events_lock);
+	list_for_each_entry_safe(msg, tmp, &watch_events, list) {
+		if (msg->u.watch.handle != watch)
+			continue;
+		list_del(&msg->list);
+		kfree(msg->u.watch.vec);
+		kfree(msg);
+	}
+	spin_unlock(&watch_events_lock);
+
+	/* Flush any currently-executing callback, unless we are it. :-) */
+	if (current->pid != xenwatch_pid) {
+		down(&xenwatch_mutex);
+		up(&xenwatch_mutex);
+	}
+}
+EXPORT_SYMBOL(unregister_xenbus_watch);
+
+void xs_suspend(void)
+{
+	down_write(&xs_state.suspend_mutex);
+	down(&xs_state.request_mutex);
+}
+
+void xs_resume(void)
+{
+	struct xenbus_watch *watch;
+	char token[sizeof(watch) * 2 + 1];
+
+	up(&xs_state.request_mutex);
+
+	/* No need for watches_lock: the suspend_mutex is sufficient. */
+	list_for_each_entry(watch, &watches, list) {
+		sprintf(token, "%lX", (long)watch);
+		xs_watch(watch->node, token);
+	}
+
+	up_write(&xs_state.suspend_mutex);
+}
+
+static int xenwatch_thread(void *unused)
+{
+	struct list_head *ent;
+	struct xs_stored_msg *msg;
+
+	for (;;) {
+		wait_event_interruptible(watch_events_waitq,
+					 !list_empty(&watch_events));
+
+		down(&xenwatch_mutex);
+
+		spin_lock(&watch_events_lock);
+		ent = watch_events.next;
+		if (ent != &watch_events)
+			list_del(ent);
+		spin_unlock(&watch_events_lock);
+
+		if (ent != &watch_events) {
+			msg = list_entry(ent, struct xs_stored_msg, list);
+			msg->u.watch.handle->callback(
+				msg->u.watch.handle,
+				(const char **)msg->u.watch.vec,
+				msg->u.watch.vec_size);
+			kfree(msg->u.watch.vec);
+			kfree(msg);
+		}
+
+		up(&xenwatch_mutex);
+	}
+}
+
+static int process_msg(void)
+{
+	struct xs_stored_msg *msg;
+	char *body;
+	int err;
+
+	msg = kmalloc(sizeof(*msg), GFP_KERNEL);
+	if (msg == NULL)
+		return -ENOMEM;
+
+	err = xb_read(&msg->hdr, sizeof(msg->hdr));
+	if (err) {
+		kfree(msg);
+		return err;
+	}
+
+	body = kmalloc(msg->hdr.len + 1, GFP_KERNEL);
+	if (body == NULL) {
+		kfree(msg);
+		return -ENOMEM;
+	}
+
+	err = xb_read(body, msg->hdr.len);
+	if (err) {
+		kfree(body);
+		kfree(msg);
+		return err;
+	}
+	body[msg->hdr.len] = '\0';
+
+	if (msg->hdr.type == XS_WATCH_EVENT) {
+		msg->u.watch.vec = split(body, msg->hdr.len,
+					 &msg->u.watch.vec_size);
+		if (IS_ERR(msg->u.watch.vec)) {
+			kfree(msg);
+			return PTR_ERR(msg->u.watch.vec);
+		}
+
+		spin_lock(&watches_lock);
+		msg->u.watch.handle = find_watch(
+			msg->u.watch.vec[XS_WATCH_TOKEN]);
+		if (msg->u.watch.handle != NULL) {
+			spin_lock(&watch_events_lock);
+			list_add_tail(&msg->list, &watch_events);
+			wake_up(&watch_events_waitq);
+			spin_unlock(&watch_events_lock);
+		} else {
+			kfree(msg->u.watch.vec);
+			kfree(msg);
+		}
+		spin_unlock(&watches_lock);
+	} else {
+		msg->u.reply.body = body;
+		spin_lock(&xs_state.reply_lock);
+		list_add_tail(&msg->list, &xs_state.reply_list);
+		spin_unlock(&xs_state.reply_lock);
+		wake_up(&xs_state.reply_waitq);
+	}
+
+	return 0;
+}
+
+static int xenbus_thread(void *unused)
+{
+	int err;
+
+	for (;;) {
+		err = process_msg();
+		if (err)
+			printk(KERN_WARNING "XENBUS error %d while reading "
+			       "message\n", err);
+	}
+}
+
+int xs_init(void)
+{
+	int err;
+	struct task_struct *task;
+
+	INIT_LIST_HEAD(&xs_state.reply_list);
+	spin_lock_init(&xs_state.reply_lock);
+	init_waitqueue_head(&xs_state.reply_waitq);
+
+	init_MUTEX(&xs_state.request_mutex);
+	init_rwsem(&xs_state.suspend_mutex);
+
+	/* Initialize the shared memory rings to talk to xenstored */
+	err = xb_init_comms();
+	if (err)
+		return err;
+
+	task = kthread_run(xenwatch_thread, NULL, "xenwatch");
+	if (IS_ERR(task))
+		return PTR_ERR(task);
+	xenwatch_pid = task->pid;
+
+	task = kthread_run(xenbus_thread, NULL, "xenbus");
+	if (IS_ERR(task))
+		return PTR_ERR(task);
+
+	return 0;
+}
+
+/*
+ * Local variables:
+ *  c-file-style: "linux"
+ *  indent-tabs-mode: t
+ *  c-indent-level: 8
+ *  c-basic-offset: 8
+ *  tab-width: 8
+ * End:
+ */