[mirror_ubuntu-hirsute-kernel.git] / drivers / block / nbd.c

/*
 * Network block device - make block devices work over TCP
 *
 * Note that you can not swap over this thing, yet. Seems to work but
 * deadlocks sometimes - you can not swap over TCP in general.
 * 
 * Copyright 1997-2000 Pavel Machek <pavel@ucw.cz>
 * Parts copyright 2001 Steven Whitehouse <steve@chygwyn.com>
 *
 * This file is released under GPLv2 or later.
 *
 * (part of code stolen from loop.c)
 */

#include <linux/major.h>

#include <linux/blkdev.h>
#include <linux/module.h>
#include <linux/init.h>
#include <linux/sched.h>
#include <linux/fs.h>
#include <linux/bio.h>
#include <linux/stat.h>
#include <linux/errno.h>
#include <linux/file.h>
#include <linux/ioctl.h>
#include <linux/compiler.h>
#include <linux/err.h>
#include <linux/kernel.h>
#include <net/sock.h>

#include <asm/uaccess.h>
#include <asm/system.h>
#include <asm/types.h>

#include <linux/nbd.h>

#define LO_MAGIC 0x68797548

#ifdef NDEBUG
#define dprintk(flags, fmt...)
#else /* NDEBUG */
#define dprintk(flags, fmt...) do { \
	if (debugflags & (flags)) printk(KERN_DEBUG fmt); \
} while (0)
#define DBG_IOCTL       0x0004
#define DBG_INIT        0x0010
#define DBG_EXIT        0x0020
#define DBG_BLKDEV      0x0100
#define DBG_RX          0x0200
#define DBG_TX          0x0400
static unsigned int debugflags;
#endif /* NDEBUG */

static unsigned int nbds_max = 16;
static struct nbd_device nbd_dev[MAX_NBD];

/*
 * Use just one lock (or at most 1 per NIC). Two arguments for this:
 * 1. Each NIC is essentially a synchronization point for all servers
 *    accessed through that NIC so there's no need to have more locks
 *    than NICs anyway.
 * 2. More locks lead to more "Dirty cache line bouncing" which will slow
 *    down each lock to the point where they're actually slower than just
 *    a single lock.
 * Thanks go to Jens Axboe and Al Viro for their LKML emails explaining this!
 */
static DEFINE_SPINLOCK(nbd_lock);

#ifndef NDEBUG
static const char *ioctl_cmd_to_ascii(int cmd)
{
	switch (cmd) {
	case NBD_SET_SOCK: return "set-sock";
	case NBD_SET_BLKSIZE: return "set-blksize";
	case NBD_SET_SIZE: return "set-size";
	case NBD_DO_IT: return "do-it";
	case NBD_CLEAR_SOCK: return "clear-sock";
	case NBD_CLEAR_QUE: return "clear-que";
	case NBD_PRINT_DEBUG: return "print-debug";
	case NBD_SET_SIZE_BLOCKS: return "set-size-blocks";
	case NBD_DISCONNECT: return "disconnect";
	case BLKROSET: return "set-read-only";
	case BLKFLSBUF: return "flush-buffer-cache";
	}
	return "unknown";
}

static const char *nbdcmd_to_ascii(int cmd)
{
	switch (cmd) {
	case  NBD_CMD_READ: return "read";
	case NBD_CMD_WRITE: return "write";
	case  NBD_CMD_DISC: return "disconnect";
	}
	return "invalid";
}
#endif /* NDEBUG */

static void nbd_end_request(struct request *req)
{
	int uptodate = (req->errors == 0) ? 1 : 0;
	request_queue_t *q = req->q;
	unsigned long flags;

	dprintk(DBG_BLKDEV, "%s: request %p: %s\n", req->rq_disk->disk_name,
			req, uptodate? "done": "failed");

	spin_lock_irqsave(q->queue_lock, flags);
	if (!end_that_request_first(req, uptodate, req->nr_sectors)) {
		end_that_request_last(req, uptodate);
	}
	spin_unlock_irqrestore(q->queue_lock, flags);
}

/*
 *  Send or receive packet.
 */
static int sock_xmit(struct socket *sock, int send, void *buf, int size,
		int msg_flags)
{
	int result;
	struct msghdr msg;
	struct kvec iov;
	unsigned long flags;
	sigset_t oldset;

	/* Allow interception of SIGKILL only
	 * Don't allow other signals to interrupt the transmission */
	spin_lock_irqsave(&current->sighand->siglock, flags);
	oldset = current->blocked;
	sigfillset(&current->blocked);
	sigdelsetmask(&current->blocked, sigmask(SIGKILL));
	recalc_sigpending();
	spin_unlock_irqrestore(&current->sighand->siglock, flags);

	do {
		sock->sk->sk_allocation = GFP_NOIO;
		iov.iov_base = buf;
		iov.iov_len = size;
		msg.msg_name = NULL;
		msg.msg_namelen = 0;
		msg.msg_control = NULL;
		msg.msg_controllen = 0;
		msg.msg_flags = msg_flags | MSG_NOSIGNAL;

		if (send)
			result = kernel_sendmsg(sock, &msg, &iov, 1, size);
		else
			result = kernel_recvmsg(sock, &msg, &iov, 1, size, 0);

		if (signal_pending(current)) {
			siginfo_t info;
			spin_lock_irqsave(&current->sighand->siglock, flags);
			printk(KERN_WARNING "nbd (pid %d: %s) got signal %d\n",
				current->pid, current->comm, 
				dequeue_signal(current, &current->blocked, &info));
			spin_unlock_irqrestore(&current->sighand->siglock, flags);
			result = -EINTR;
			break;
		}

		if (result <= 0) {
			if (result == 0)
				result = -EPIPE; /* short read */
			break;
		}
		size -= result;
		buf += result;
	} while (size > 0);

	spin_lock_irqsave(&current->sighand->siglock, flags);
	current->blocked = oldset;
	recalc_sigpending();
	spin_unlock_irqrestore(&current->sighand->siglock, flags);

	return result;
}

static inline int sock_send_bvec(struct socket *sock, struct bio_vec *bvec,
		int flags)
{
	int result;
	void *kaddr = kmap(bvec->bv_page);
	result = sock_xmit(sock, 1, kaddr + bvec->bv_offset, bvec->bv_len,
			flags);
	kunmap(bvec->bv_page);
	return result;
}

static int nbd_send_req(struct nbd_device *lo, struct request *req)
{
	int result, i, flags;
	struct nbd_request request;
	unsigned long size = req->nr_sectors << 9;
	struct socket *sock = lo->sock;

	request.magic = htonl(NBD_REQUEST_MAGIC);
	request.type = htonl(nbd_cmd(req));
	request.from = cpu_to_be64((u64) req->sector << 9);
	request.len = htonl(size);
	memcpy(request.handle, &req, sizeof(req));

	dprintk(DBG_TX, "%s: request %p: sending control (%s@%llu,%luB)\n",
			lo->disk->disk_name, req,
			nbdcmd_to_ascii(nbd_cmd(req)),
			(unsigned long long)req->sector << 9,
			req->nr_sectors << 9);
	result = sock_xmit(sock, 1, &request, sizeof(request),
			(nbd_cmd(req) == NBD_CMD_WRITE)? MSG_MORE: 0);
	if (result <= 0) {
		printk(KERN_ERR "%s: Send control failed (result %d)\n",
				lo->disk->disk_name, result);
		goto error_out;
	}

	if (nbd_cmd(req) == NBD_CMD_WRITE) {
		struct bio *bio;
		/*
		 * we are really probing at internals to determine
		 * whether to set MSG_MORE or not...
		 */
		rq_for_each_bio(bio, req) {
			struct bio_vec *bvec;
			bio_for_each_segment(bvec, bio, i) {
				flags = 0;
				if ((i < (bio->bi_vcnt - 1)) || bio->bi_next)
					flags = MSG_MORE;
				dprintk(DBG_TX, "%s: request %p: sending %d bytes data\n",
						lo->disk->disk_name, req,
						bvec->bv_len);
				result = sock_send_bvec(sock, bvec, flags);
				if (result <= 0) {
					printk(KERN_ERR "%s: Send data failed (result %d)\n",
							lo->disk->disk_name,
							result);
					goto error_out;
				}
			}
		}
	}
	return 0;

error_out:
	return 1;
}

static struct request *nbd_find_request(struct nbd_device *lo, char *handle)
{
	struct request *req;
	struct list_head *tmp;
	struct request *xreq;
	int err;

	memcpy(&xreq, handle, sizeof(xreq));

	err = wait_event_interruptible(lo->active_wq, lo->active_req != xreq);
	if (unlikely(err))
		goto out;

	spin_lock(&lo->queue_lock);
	list_for_each(tmp, &lo->queue_head) {
		req = list_entry(tmp, struct request, queuelist);
		if (req != xreq)
			continue;
		list_del_init(&req->queuelist);
		spin_unlock(&lo->queue_lock);
		return req;
	}
	spin_unlock(&lo->queue_lock);

	err = -ENOENT;

out:
	return ERR_PTR(err);
}

static inline int sock_recv_bvec(struct socket *sock, struct bio_vec *bvec)
{
	int result;
	void *kaddr = kmap(bvec->bv_page);
	result = sock_xmit(sock, 0, kaddr + bvec->bv_offset, bvec->bv_len,
			MSG_WAITALL);
	kunmap(bvec->bv_page);
	return result;
}

/* NULL returned = something went wrong, inform userspace */
static struct request *nbd_read_stat(struct nbd_device *lo)
{
	int result;
	struct nbd_reply reply;
	struct request *req;
	struct socket *sock = lo->sock;

	reply.magic = 0;
	result = sock_xmit(sock, 0, &reply, sizeof(reply), MSG_WAITALL);
	if (result <= 0) {
		printk(KERN_ERR "%s: Receive control failed (result %d)\n",
				lo->disk->disk_name, result);
		goto harderror;
	}

	if (ntohl(reply.magic) != NBD_REPLY_MAGIC) {
		printk(KERN_ERR "%s: Wrong magic (0x%lx)\n",
				lo->disk->disk_name,
				(unsigned long)ntohl(reply.magic));
		result = -EPROTO;
		goto harderror;
	}

	req = nbd_find_request(lo, reply.handle);
	if (unlikely(IS_ERR(req))) {
		result = PTR_ERR(req);
		if (result != -ENOENT)
			goto harderror;

		printk(KERN_ERR "%s: Unexpected reply (%p)\n",
				lo->disk->disk_name, reply.handle);
		result = -EBADR;
		goto harderror;
	}

	if (ntohl(reply.error)) {
		printk(KERN_ERR "%s: Other side returned error (%d)\n",
				lo->disk->disk_name, ntohl(reply.error));
		req->errors++;
		return req;
	}

	dprintk(DBG_RX, "%s: request %p: got reply\n",
			lo->disk->disk_name, req);
	if (nbd_cmd(req) == NBD_CMD_READ) {
		int i;
		struct bio *bio;
		rq_for_each_bio(bio, req) {
			struct bio_vec *bvec;
			bio_for_each_segment(bvec, bio, i) {
				result = sock_recv_bvec(sock, bvec);
				if (result <= 0) {
					printk(KERN_ERR "%s: Receive data failed (result %d)\n",
							lo->disk->disk_name,
							result);
					req->errors++;
					return req;
				}
				dprintk(DBG_RX, "%s: request %p: got %d bytes data\n",
					lo->disk->disk_name, req, bvec->bv_len);
			}
		}
	}
	return req;
harderror:
	lo->harderror = result;
	return NULL;
}

static ssize_t pid_show(struct gendisk *disk, char *page)
{
	return sprintf(page, "%ld\n",
		(long) ((struct nbd_device *)disk->private_data)->pid);
}

static struct disk_attribute pid_attr = {
	.attr = { .name = "pid", .mode = S_IRUGO },
	.show = pid_show,
};

static int nbd_do_it(struct nbd_device *lo)
{
	struct request *req;
	int ret;

	BUG_ON(lo->magic != LO_MAGIC);

	lo->pid = current->pid;
	ret = sysfs_create_file(&lo->disk->kobj, &pid_attr.attr);
	if (ret) {
		printk(KERN_ERR "nbd: sysfs_create_file failed!");
		return ret;
	}

	while ((req = nbd_read_stat(lo)) != NULL)
		nbd_end_request(req);

	sysfs_remove_file(&lo->disk->kobj, &pid_attr.attr);
	return 0;
}

static void nbd_clear_que(struct nbd_device *lo)
{
	struct request *req;

	BUG_ON(lo->magic != LO_MAGIC);

	/*
	 * Because we have set lo->sock to NULL under the tx_lock, all
	 * modifications to the list must have completed by now.  For
	 * the same reason, the active_req must be NULL.
	 *
	 * As a consequence, we don't need to take the spin lock while
	 * purging the list here.
	 */
	BUG_ON(lo->sock);
	BUG_ON(lo->active_req);

	while (!list_empty(&lo->queue_head)) {
		req = list_entry(lo->queue_head.next, struct request,
				 queuelist);
		list_del_init(&req->queuelist);
		req->errors++;
		nbd_end_request(req);
	}
}

/*
 * We always wait for result of write, for now. It would be nice to make it optional
 * in future
 * if ((rq_data_dir(req) == WRITE) && (lo->flags & NBD_WRITE_NOCHK))
 *   { printk( "Warning: Ignoring result!\n"); nbd_end_request( req ); }
 */

static void do_nbd_request(request_queue_t * q)
{
	struct request *req;
	
	while ((req = elv_next_request(q)) != NULL) {
		struct nbd_device *lo;

		blkdev_dequeue_request(req);
		dprintk(DBG_BLKDEV, "%s: request %p: dequeued (flags=%x)\n",
				req->rq_disk->disk_name, req, req->cmd_type);

		if (!blk_fs_request(req))
			goto error_out;

		lo = req->rq_disk->private_data;

		BUG_ON(lo->magic != LO_MAGIC);

		nbd_cmd(req) = NBD_CMD_READ;
		if (rq_data_dir(req) == WRITE) {
			nbd_cmd(req) = NBD_CMD_WRITE;
			if (lo->flags & NBD_READ_ONLY) {
				printk(KERN_ERR "%s: Write on read-only\n",
						lo->disk->disk_name);
				goto error_out;
			}
		}

		req->errors = 0;
		spin_unlock_irq(q->queue_lock);

		mutex_lock(&lo->tx_lock);
		if (unlikely(!lo->sock)) {
			mutex_unlock(&lo->tx_lock);
			printk(KERN_ERR "%s: Attempted send on closed socket\n",
			       lo->disk->disk_name);
			req->errors++;
			nbd_end_request(req);
			spin_lock_irq(q->queue_lock);
			continue;
		}

		lo->active_req = req;

		if (nbd_send_req(lo, req) != 0) {
			printk(KERN_ERR "%s: Request send failed\n",
					lo->disk->disk_name);
			req->errors++;
			nbd_end_request(req);
		} else {
			spin_lock(&lo->queue_lock);
			list_add(&req->queuelist, &lo->queue_head);
			spin_unlock(&lo->queue_lock);
		}

		lo->active_req = NULL;
		mutex_unlock(&lo->tx_lock);
		wake_up_all(&lo->active_wq);

		spin_lock_irq(q->queue_lock);
		continue;

error_out:
		req->errors++;
		spin_unlock(q->queue_lock);
		nbd_end_request(req);
		spin_lock(q->queue_lock);
	}
	return;
}

static int nbd_ioctl(struct inode *inode, struct file *file,
		     unsigned int cmd, unsigned long arg)
{
	struct nbd_device *lo = inode->i_bdev->bd_disk->private_data;
	int error;
	struct request sreq ;

	if (!capable(CAP_SYS_ADMIN))
		return -EPERM;

	BUG_ON(lo->magic != LO_MAGIC);

	/* Anyone capable of this syscall can do *real bad* things */
	dprintk(DBG_IOCTL, "%s: nbd_ioctl cmd=%s(0x%x) arg=%lu\n",
			lo->disk->disk_name, ioctl_cmd_to_ascii(cmd), cmd, arg);

	switch (cmd) {
	case NBD_DISCONNECT:
	        printk(KERN_INFO "%s: NBD_DISCONNECT\n", lo->disk->disk_name);
		sreq.cmd_type = REQ_TYPE_SPECIAL;
		nbd_cmd(&sreq) = NBD_CMD_DISC;
		/*
		 * Set these to sane values in case server implementation
		 * fails to check the request type first and also to keep
		 * debugging output cleaner.
		 */
		sreq.sector = 0;
		sreq.nr_sectors = 0;
                if (!lo->sock)
			return -EINVAL;
                nbd_send_req(lo, &sreq);
                return 0;
 
	case NBD_CLEAR_SOCK:
		error = 0;
		mutex_lock(&lo->tx_lock);
		lo->sock = NULL;
		mutex_unlock(&lo->tx_lock);
		file = lo->file;
		lo->file = NULL;
		nbd_clear_que(lo);
		BUG_ON(!list_empty(&lo->queue_head));
		if (file)
			fput(file);
		return error;
	case NBD_SET_SOCK:
		if (lo->file)
			return -EBUSY;
		error = -EINVAL;
		file = fget(arg);
		if (file) {
			inode = file->f_path.dentry->d_inode;
			if (S_ISSOCK(inode->i_mode)) {
				lo->file = file;
				lo->sock = SOCKET_I(inode);
				error = 0;
			} else {
				fput(file);
			}
		}
		return error;
	case NBD_SET_BLKSIZE:
		lo->blksize = arg;
		lo->bytesize &= ~(lo->blksize-1);
		inode->i_bdev->bd_inode->i_size = lo->bytesize;
		set_blocksize(inode->i_bdev, lo->blksize);
		set_capacity(lo->disk, lo->bytesize >> 9);
		return 0;
	case NBD_SET_SIZE:
		lo->bytesize = arg & ~(lo->blksize-1);
		inode->i_bdev->bd_inode->i_size = lo->bytesize;
		set_blocksize(inode->i_bdev, lo->blksize);
		set_capacity(lo->disk, lo->bytesize >> 9);
		return 0;
	case NBD_SET_SIZE_BLOCKS:
		lo->bytesize = ((u64) arg) * lo->blksize;
		inode->i_bdev->bd_inode->i_size = lo->bytesize;
		set_blocksize(inode->i_bdev, lo->blksize);
		set_capacity(lo->disk, lo->bytesize >> 9);
		return 0;
	case NBD_DO_IT:
		if (!lo->file)
			return -EINVAL;
		error = nbd_do_it(lo);
		if (error)
			return error;
		/* on return tidy up in case we have a signal */
		/* Forcibly shutdown the socket causing all listeners
		 * to error
		 *
		 * FIXME: This code is duplicated from sys_shutdown, but
		 * there should be a more generic interface rather than
		 * calling socket ops directly here */
		mutex_lock(&lo->tx_lock);
		if (lo->sock) {
			printk(KERN_WARNING "%s: shutting down socket\n",
				lo->disk->disk_name);
			lo->sock->ops->shutdown(lo->sock,
				SEND_SHUTDOWN|RCV_SHUTDOWN);
			lo->sock = NULL;
		}
		mutex_unlock(&lo->tx_lock);
		file = lo->file;
		lo->file = NULL;
		nbd_clear_que(lo);
		printk(KERN_WARNING "%s: queue cleared\n", lo->disk->disk_name);
		if (file)
			fput(file);
		return lo->harderror;
	case NBD_CLEAR_QUE:
		/*
		 * This is for compatibility only.  The queue is always cleared
		 * by NBD_DO_IT or NBD_CLEAR_SOCK.
		 */
		BUG_ON(!lo->sock && !list_empty(&lo->queue_head));
		return 0;
	case NBD_PRINT_DEBUG:
		printk(KERN_INFO "%s: next = %p, prev = %p, head = %p\n",
			inode->i_bdev->bd_disk->disk_name,
			lo->queue_head.next, lo->queue_head.prev,
			&lo->queue_head);
		return 0;
	}
	return -EINVAL;
}

static struct block_device_operations nbd_fops =
{
	.owner =	THIS_MODULE,
	.ioctl =	nbd_ioctl,
};

/*
 * And here should be modules and kernel interface 
 *  (Just smiley confuses emacs :-)
 */

static int __init nbd_init(void)
{
	int err = -ENOMEM;
	int i;

	BUILD_BUG_ON(sizeof(struct nbd_request) != 28);

	if (nbds_max > MAX_NBD) {
		printk(KERN_CRIT "nbd: cannot allocate more than %u nbds; %u requested.\n", MAX_NBD,
				nbds_max);
		return -EINVAL;
	}

	for (i = 0; i < nbds_max; i++) {
		struct gendisk *disk = alloc_disk(1);
		if (!disk)
			goto out;
		nbd_dev[i].disk = disk;
		/*
		 * The new linux 2.5 block layer implementation requires
		 * every gendisk to have its very own request_queue struct.
		 * These structs are big so we dynamically allocate them.
		 */
		disk->queue = blk_init_queue(do_nbd_request, &nbd_lock);
		if (!disk->queue) {
			put_disk(disk);
			goto out;
		}
	}

	if (register_blkdev(NBD_MAJOR, "nbd")) {
		err = -EIO;
		goto out;
	}

	printk(KERN_INFO "nbd: registered device at major %d\n", NBD_MAJOR);
	dprintk(DBG_INIT, "nbd: debugflags=0x%x\n", debugflags);

	for (i = 0; i < nbds_max; i++) {
		struct gendisk *disk = nbd_dev[i].disk;
		nbd_dev[i].file = NULL;
		nbd_dev[i].magic = LO_MAGIC;
		nbd_dev[i].flags = 0;
		spin_lock_init(&nbd_dev[i].queue_lock);
		INIT_LIST_HEAD(&nbd_dev[i].queue_head);
		mutex_init(&nbd_dev[i].tx_lock);
		init_waitqueue_head(&nbd_dev[i].active_wq);
		nbd_dev[i].blksize = 1024;
		nbd_dev[i].bytesize = 0x7ffffc00ULL << 10; /* 2TB */
		disk->major = NBD_MAJOR;
		disk->first_minor = i;
		disk->fops = &nbd_fops;
		disk->private_data = &nbd_dev[i];
		disk->flags |= GENHD_FL_SUPPRESS_PARTITION_INFO;
		sprintf(disk->disk_name, "nbd%d", i);
		set_capacity(disk, 0x7ffffc00ULL << 1); /* 2 TB */
		add_disk(disk);
	}

	return 0;
out:
	while (i--) {
		blk_cleanup_queue(nbd_dev[i].disk->queue);
		put_disk(nbd_dev[i].disk);
	}
	return err;
}

static void __exit nbd_cleanup(void)
{
	int i;
	for (i = 0; i < nbds_max; i++) {
		struct gendisk *disk = nbd_dev[i].disk;
		nbd_dev[i].magic = 0;
		if (disk) {
			del_gendisk(disk);
			blk_cleanup_queue(disk->queue);
			put_disk(disk);
		}
	}
	unregister_blkdev(NBD_MAJOR, "nbd");
	printk(KERN_INFO "nbd: unregistered device at major %d\n", NBD_MAJOR);
}

module_init(nbd_init);
module_exit(nbd_cleanup);

MODULE_DESCRIPTION("Network Block Device");
MODULE_LICENSE("GPL");

module_param(nbds_max, int, 0444);
MODULE_PARM_DESC(nbds_max, "How many network block devices to initialize.");
#ifndef NDEBUG
module_param(debugflags, int, 0644);
MODULE_PARM_DESC(debugflags, "flags for controlling debug output");
#endif
Commit	Line	Data
1da177e4 LT	1	/*
	2	* Network block device - make block devices work over TCP
	3	*
	4	* Note that you can not swap over this thing, yet. Seems to work but
	5	* deadlocks sometimes - you can not swap over TCP in general.
	6	*
	7	* Copyright 1997-2000 Pavel Machek <pavel@ucw.cz>
	8	* Parts copyright 2001 Steven Whitehouse <steve@chygwyn.com>
	9	*
dbf492d6	10	* This file is released under GPLv2 or later.
1da177e4	11	*
dbf492d6	12	* (part of code stolen from loop.c)
1da177e4 LT	13	*/
	14
	15	#include <linux/major.h>
	16
	17	#include <linux/blkdev.h>
	18	#include <linux/module.h>
	19	#include <linux/init.h>
	20	#include <linux/sched.h>
	21	#include <linux/fs.h>
	22	#include <linux/bio.h>
	23	#include <linux/stat.h>
	24	#include <linux/errno.h>
	25	#include <linux/file.h>
	26	#include <linux/ioctl.h>
4b2f0260 HX	27	#include <linux/compiler.h>
	28	#include <linux/err.h>
	29	#include <linux/kernel.h>
1da177e4 LT	30	#include <net/sock.h>
1da177e4 LT	31
1da177e4	32	#include <asm/uaccess.h>
4b2f0260	33	#include <asm/system.h>
1da177e4 LT	34	#include <asm/types.h>
	35
	36	#include <linux/nbd.h>
	37
	38	#define LO_MAGIC 0x68797548
	39
	40	#ifdef NDEBUG
	41	#define dprintk(flags, fmt...)
	42	#else /* NDEBUG */
	43	#define dprintk(flags, fmt...) do { \
	44	if (debugflags & (flags)) printk(KERN_DEBUG fmt); \
	45	} while (0)
	46	#define DBG_IOCTL 0x0004
	47	#define DBG_INIT 0x0010
	48	#define DBG_EXIT 0x0020
	49	#define DBG_BLKDEV 0x0100
	50	#define DBG_RX 0x0200
	51	#define DBG_TX 0x0400
	52	static unsigned int debugflags;
	53	#endif /* NDEBUG */
	54
9c7a4169	55	static unsigned int nbds_max = 16;
1da177e4 LT	56	static struct nbd_device nbd_dev[MAX_NBD];
	57
	58	/*
	59	* Use just one lock (or at most 1 per NIC). Two arguments for this:
	60	* 1. Each NIC is essentially a synchronization point for all servers
	61	* accessed through that NIC so there's no need to have more locks
	62	* than NICs anyway.
	63	* 2. More locks lead to more "Dirty cache line bouncing" which will slow
	64	* down each lock to the point where they're actually slower than just
	65	* a single lock.
	66	* Thanks go to Jens Axboe and Al Viro for their LKML emails explaining this!
	67	*/
	68	static DEFINE_SPINLOCK(nbd_lock);
	69
	70	#ifndef NDEBUG
	71	static const char *ioctl_cmd_to_ascii(int cmd)
	72	{
	73	switch (cmd) {
	74	case NBD_SET_SOCK: return "set-sock";
	75	case NBD_SET_BLKSIZE: return "set-blksize";
	76	case NBD_SET_SIZE: return "set-size";
	77	case NBD_DO_IT: return "do-it";
	78	case NBD_CLEAR_SOCK: return "clear-sock";
	79	case NBD_CLEAR_QUE: return "clear-que";
	80	case NBD_PRINT_DEBUG: return "print-debug";
	81	case NBD_SET_SIZE_BLOCKS: return "set-size-blocks";
	82	case NBD_DISCONNECT: return "disconnect";
	83	case BLKROSET: return "set-read-only";
	84	case BLKFLSBUF: return "flush-buffer-cache";
	85	}
	86	return "unknown";
	87	}
	88
	89	static const char *nbdcmd_to_ascii(int cmd)
	90	{
	91	switch (cmd) {
	92	case NBD_CMD_READ: return "read";
	93	case NBD_CMD_WRITE: return "write";
	94	case NBD_CMD_DISC: return "disconnect";
	95	}
	96	return "invalid";
	97	}
	98	#endif /* NDEBUG */
	99
	100	static void nbd_end_request(struct request *req)
	101	{
	102	int uptodate = (req->errors == 0) ? 1 : 0;
	103	request_queue_t *q = req->q;
	104	unsigned long flags;
	105
	106	dprintk(DBG_BLKDEV, "%s: request %p: %s\n", req->rq_disk->disk_name,
	107	req, uptodate? "done": "failed");
	108
	109	spin_lock_irqsave(q->queue_lock, flags);
	110	if (!end_that_request_first(req, uptodate, req->nr_sectors)) {
8ffdc655	111	end_that_request_last(req, uptodate);
1da177e4 LT	112	}
	113	spin_unlock_irqrestore(q->queue_lock, flags);
	114	}
	115
	116	/*
	117	* Send or receive packet.
	118	*/
	119	static int sock_xmit(struct socket sock, int send, void buf, int size,
	120	int msg_flags)
	121	{
	122	int result;
	123	struct msghdr msg;
	124	struct kvec iov;
	125	unsigned long flags;
	126	sigset_t oldset;
	127
	128	/* Allow interception of SIGKILL only
	129	* Don't allow other signals to interrupt the transmission */
	130	spin_lock_irqsave(&current->sighand->siglock, flags);
	131	oldset = current->blocked;
	132	sigfillset(&current->blocked);
	133	sigdelsetmask(&current->blocked, sigmask(SIGKILL));
	134	recalc_sigpending();
	135	spin_unlock_irqrestore(&current->sighand->siglock, flags);
	136
	137	do {
	138	sock->sk->sk_allocation = GFP_NOIO;
	139	iov.iov_base = buf;
	140	iov.iov_len = size;
	141	msg.msg_name = NULL;
	142	msg.msg_namelen = 0;
	143	msg.msg_control = NULL;
	144	msg.msg_controllen = 0;
1da177e4 LT	145	msg.msg_flags = msg_flags \| MSG_NOSIGNAL;
	146
	147	if (send)
	148	result = kernel_sendmsg(sock, &msg, &iov, 1, size);
	149	else
	150	result = kernel_recvmsg(sock, &msg, &iov, 1, size, 0);
	151
	152	if (signal_pending(current)) {
	153	siginfo_t info;
	154	spin_lock_irqsave(&current->sighand->siglock, flags);
	155	printk(KERN_WARNING "nbd (pid %d: %s) got signal %d\n",
	156	current->pid, current->comm,
	157	dequeue_signal(current, &current->blocked, &info));
	158	spin_unlock_irqrestore(&current->sighand->siglock, flags);
	159	result = -EINTR;
	160	break;
	161	}
	162
	163	if (result <= 0) {
	164	if (result == 0)
	165	result = -EPIPE; /* short read */
	166	break;
	167	}
	168	size -= result;
	169	buf += result;
	170	} while (size > 0);
	171
	172	spin_lock_irqsave(&current->sighand->siglock, flags);
	173	current->blocked = oldset;
	174	recalc_sigpending();
	175	spin_unlock_irqrestore(&current->sighand->siglock, flags);
	176
	177	return result;
	178	}
	179
	180	static inline int sock_send_bvec(struct socket sock, struct bio_vec bvec,
	181	int flags)
	182	{
	183	int result;
	184	void *kaddr = kmap(bvec->bv_page);
	185	result = sock_xmit(sock, 1, kaddr + bvec->bv_offset, bvec->bv_len,
	186	flags);
	187	kunmap(bvec->bv_page);
	188	return result;
	189	}
	190
	191	static int nbd_send_req(struct nbd_device lo, struct request req)
	192	{
	193	int result, i, flags;
	194	struct nbd_request request;
	195	unsigned long size = req->nr_sectors << 9;
	196	struct socket *sock = lo->sock;
	197
	198	request.magic = htonl(NBD_REQUEST_MAGIC);
	199	request.type = htonl(nbd_cmd(req));
	200	request.from = cpu_to_be64((u64) req->sector << 9);
	201	request.len = htonl(size);
	202	memcpy(request.handle, &req, sizeof(req));
	203
1da177e4 LT	204	dprintk(DBG_TX, "%s: request %p: sending control (%s@%llu,%luB)\n",
	205	lo->disk->disk_name, req,
	206	nbdcmd_to_ascii(nbd_cmd(req)),
	207	(unsigned long long)req->sector << 9,
	208	req->nr_sectors << 9);
	209	result = sock_xmit(sock, 1, &request, sizeof(request),
	210	(nbd_cmd(req) == NBD_CMD_WRITE)? MSG_MORE: 0);
	211	if (result <= 0) {
	212	printk(KERN_ERR "%s: Send control failed (result %d)\n",
	213	lo->disk->disk_name, result);
	214	goto error_out;
	215	}
	216
	217	if (nbd_cmd(req) == NBD_CMD_WRITE) {
	218	struct bio *bio;
	219	/*
	220	* we are really probing at internals to determine
	221	* whether to set MSG_MORE or not...
	222	*/
	223	rq_for_each_bio(bio, req) {
	224	struct bio_vec *bvec;
	225	bio_for_each_segment(bvec, bio, i) {
	226	flags = 0;
	227	if ((i < (bio->bi_vcnt - 1)) \|\| bio->bi_next)
	228	flags = MSG_MORE;
	229	dprintk(DBG_TX, "%s: request %p: sending %d bytes data\n",
	230	lo->disk->disk_name, req,
	231	bvec->bv_len);
	232	result = sock_send_bvec(sock, bvec, flags);
	233	if (result <= 0) {
	234	printk(KERN_ERR "%s: Send data failed (result %d)\n",
	235	lo->disk->disk_name,
	236	result);
	237	goto error_out;
	238	}
	239	}
	240	}
	241	}
1da177e4 LT	242	return 0;
	243
	244	error_out:
1da177e4 LT	245	return 1;
	246	}
	247
	248	static struct request nbd_find_request(struct nbd_device lo, char *handle)
	249	{
	250	struct request *req;
	251	struct list_head *tmp;
	252	struct request *xreq;
4b2f0260	253	int err;
1da177e4 LT	254
	255	memcpy(&xreq, handle, sizeof(xreq));
	256
4b2f0260 HX	257	err = wait_event_interruptible(lo->active_wq, lo->active_req != xreq);
	258	if (unlikely(err))
	259	goto out;
	260
1da177e4 LT	261	spin_lock(&lo->queue_lock);
	262	list_for_each(tmp, &lo->queue_head) {
	263	req = list_entry(tmp, struct request, queuelist);
	264	if (req != xreq)
	265	continue;
	266	list_del_init(&req->queuelist);
	267	spin_unlock(&lo->queue_lock);
	268	return req;
	269	}
	270	spin_unlock(&lo->queue_lock);
4b2f0260 HX	271
	272	err = -ENOENT;
	273
	274	out:
	275	return ERR_PTR(err);
1da177e4 LT	276	}
	277
	278	static inline int sock_recv_bvec(struct socket sock, struct bio_vec bvec)
	279	{
	280	int result;
	281	void *kaddr = kmap(bvec->bv_page);
	282	result = sock_xmit(sock, 0, kaddr + bvec->bv_offset, bvec->bv_len,
	283	MSG_WAITALL);
	284	kunmap(bvec->bv_page);
	285	return result;
	286	}
	287
	288	/* NULL returned = something went wrong, inform userspace */
	289	static struct request nbd_read_stat(struct nbd_device lo)
	290	{
	291	int result;
	292	struct nbd_reply reply;
	293	struct request *req;
	294	struct socket *sock = lo->sock;
	295
	296	reply.magic = 0;
	297	result = sock_xmit(sock, 0, &reply, sizeof(reply), MSG_WAITALL);
	298	if (result <= 0) {
	299	printk(KERN_ERR "%s: Receive control failed (result %d)\n",
	300	lo->disk->disk_name, result);
	301	goto harderror;
	302	}
e4b57e08 MF	303
	304	if (ntohl(reply.magic) != NBD_REPLY_MAGIC) {
	305	printk(KERN_ERR "%s: Wrong magic (0x%lx)\n",
	306	lo->disk->disk_name,
	307	(unsigned long)ntohl(reply.magic));
	308	result = -EPROTO;
	309	goto harderror;
	310	}
	311
1da177e4	312	req = nbd_find_request(lo, reply.handle);
4b2f0260 HX	313	if (unlikely(IS_ERR(req))) {
	314	result = PTR_ERR(req);
	315	if (result != -ENOENT)
	316	goto harderror;
	317
1da177e4 LT	318	printk(KERN_ERR "%s: Unexpected reply (%p)\n",
	319	lo->disk->disk_name, reply.handle);
	320	result = -EBADR;
	321	goto harderror;
	322	}
	323
1da177e4 LT	324	if (ntohl(reply.error)) {
	325	printk(KERN_ERR "%s: Other side returned error (%d)\n",
	326	lo->disk->disk_name, ntohl(reply.error));
	327	req->errors++;
	328	return req;
	329	}
	330
	331	dprintk(DBG_RX, "%s: request %p: got reply\n",
	332	lo->disk->disk_name, req);
	333	if (nbd_cmd(req) == NBD_CMD_READ) {
	334	int i;
	335	struct bio *bio;
	336	rq_for_each_bio(bio, req) {
	337	struct bio_vec *bvec;
	338	bio_for_each_segment(bvec, bio, i) {
	339	result = sock_recv_bvec(sock, bvec);
	340	if (result <= 0) {
	341	printk(KERN_ERR "%s: Receive data failed (result %d)\n",
	342	lo->disk->disk_name,
	343	result);
f0df33bc MF	344	req->errors++;
f0df33bc MF	345	return req;
1da177e4 LT	346	}
	347	dprintk(DBG_RX, "%s: request %p: got %d bytes data\n",
	348	lo->disk->disk_name, req, bvec->bv_len);
	349	}
	350	}
	351	}
	352	return req;
	353	harderror:
	354	lo->harderror = result;
	355	return NULL;
	356	}
	357
6b39bb65 PC	358	static ssize_t pid_show(struct gendisk disk, char page)
	359	{
	360	return sprintf(page, "%ld\n",
	361	(long) ((struct nbd_device *)disk->private_data)->pid);
	362	}
	363
	364	static struct disk_attribute pid_attr = {
	365	.attr = { .name = "pid", .mode = S_IRUGO },
	366	.show = pid_show,
	367	};
	368
84963048	369	static int nbd_do_it(struct nbd_device *lo)
1da177e4 LT	370	{
1da177e4 LT	371	struct request *req;
84963048	372	int ret;
1da177e4 LT	373
	374	BUG_ON(lo->magic != LO_MAGIC);
	375
6b39bb65	376	lo->pid = current->pid;
84963048 WC	377	ret = sysfs_create_file(&lo->disk->kobj, &pid_attr.attr);
	378	if (ret) {
	379	printk(KERN_ERR "nbd: sysfs_create_file failed!");
	380	return ret;
	381	}
6b39bb65	382
1da177e4 LT	383	while ((req = nbd_read_stat(lo)) != NULL)
1da177e4 LT	384	nbd_end_request(req);
6b39bb65 PC	385
6b39bb65 PC	386	sysfs_remove_file(&lo->disk->kobj, &pid_attr.attr);
84963048	387	return 0;
1da177e4 LT	388	}
	389
	390	static void nbd_clear_que(struct nbd_device *lo)
	391	{
	392	struct request *req;
	393
	394	BUG_ON(lo->magic != LO_MAGIC);
	395
4b2f0260 HX	396	/*
	397	* Because we have set lo->sock to NULL under the tx_lock, all
	398	* modifications to the list must have completed by now. For
	399	* the same reason, the active_req must be NULL.
	400	*
	401	* As a consequence, we don't need to take the spin lock while
	402	* purging the list here.
	403	*/
	404	BUG_ON(lo->sock);
	405	BUG_ON(lo->active_req);
	406
	407	while (!list_empty(&lo->queue_head)) {
	408	req = list_entry(lo->queue_head.next, struct request,
	409	queuelist);
	410	list_del_init(&req->queuelist);
	411	req->errors++;
	412	nbd_end_request(req);
	413	}
1da177e4 LT	414	}
	415
	416	/*
	417	* We always wait for result of write, for now. It would be nice to make it optional
	418	* in future
e654bc43	419	* if ((rq_data_dir(req) == WRITE) && (lo->flags & NBD_WRITE_NOCHK))
1da177e4 LT	420	* { printk( "Warning: Ignoring result!\n"); nbd_end_request( req ); }
	421	*/
	422
	423	static void do_nbd_request(request_queue_t * q)
	424	{
	425	struct request *req;
	426
	427	while ((req = elv_next_request(q)) != NULL) {
	428	struct nbd_device *lo;
	429
	430	blkdev_dequeue_request(req);
4aff5e23 JA	431	dprintk(DBG_BLKDEV, "%s: request %p: dequeued (flags=%x)\n",
4aff5e23 JA	432	req->rq_disk->disk_name, req, req->cmd_type);
1da177e4	433
4aff5e23	434	if (!blk_fs_request(req))
1da177e4 LT	435	goto error_out;
	436
	437	lo = req->rq_disk->private_data;
	438
	439	BUG_ON(lo->magic != LO_MAGIC);
	440
1da177e4 LT	441	nbd_cmd(req) = NBD_CMD_READ;
	442	if (rq_data_dir(req) == WRITE) {
	443	nbd_cmd(req) = NBD_CMD_WRITE;
	444	if (lo->flags & NBD_READ_ONLY) {
	445	printk(KERN_ERR "%s: Write on read-only\n",
	446	lo->disk->disk_name);
	447	goto error_out;
	448	}
	449	}
	450
	451	req->errors = 0;
	452	spin_unlock_irq(q->queue_lock);
	453
82d4dc5a	454	mutex_lock(&lo->tx_lock);
4b2f0260	455	if (unlikely(!lo->sock)) {
82d4dc5a	456	mutex_unlock(&lo->tx_lock);
4b2f0260 HX	457	printk(KERN_ERR "%s: Attempted send on closed socket\n",
4b2f0260 HX	458	lo->disk->disk_name);
1da177e4 LT	459	req->errors++;
	460	nbd_end_request(req);
	461	spin_lock_irq(q->queue_lock);
	462	continue;
	463	}
	464
4b2f0260	465	lo->active_req = req;
1da177e4 LT	466
	467	if (nbd_send_req(lo, req) != 0) {
	468	printk(KERN_ERR "%s: Request send failed\n",
	469	lo->disk->disk_name);
4b2f0260 HX	470	req->errors++;
	471	nbd_end_request(req);
	472	} else {
	473	spin_lock(&lo->queue_lock);
	474	list_add(&req->queuelist, &lo->queue_head);
	475	spin_unlock(&lo->queue_lock);
1da177e4 LT	476	}
1da177e4 LT	477
4b2f0260	478	lo->active_req = NULL;
82d4dc5a	479	mutex_unlock(&lo->tx_lock);
4b2f0260 HX	480	wake_up_all(&lo->active_wq);
4b2f0260 HX	481
1da177e4 LT	482	spin_lock_irq(q->queue_lock);
	483	continue;
	484
	485	error_out:
	486	req->errors++;
	487	spin_unlock(q->queue_lock);
	488	nbd_end_request(req);
	489	spin_lock(q->queue_lock);
	490	}
	491	return;
	492	}
	493
	494	static int nbd_ioctl(struct inode inode, struct file file,
	495	unsigned int cmd, unsigned long arg)
	496	{
	497	struct nbd_device *lo = inode->i_bdev->bd_disk->private_data;
	498	int error;
	499	struct request sreq ;
	500
	501	if (!capable(CAP_SYS_ADMIN))
	502	return -EPERM;
	503
	504	BUG_ON(lo->magic != LO_MAGIC);
	505
	506	/* Anyone capable of this syscall can do real bad things */
	507	dprintk(DBG_IOCTL, "%s: nbd_ioctl cmd=%s(0x%x) arg=%lu\n",
	508	lo->disk->disk_name, ioctl_cmd_to_ascii(cmd), cmd, arg);
	509
	510	switch (cmd) {
	511	case NBD_DISCONNECT:
	512	printk(KERN_INFO "%s: NBD_DISCONNECT\n", lo->disk->disk_name);
4aff5e23	513	sreq.cmd_type = REQ_TYPE_SPECIAL;
1da177e4 LT	514	nbd_cmd(&sreq) = NBD_CMD_DISC;
	515	/*
	516	* Set these to sane values in case server implementation
	517	* fails to check the request type first and also to keep
	518	* debugging output cleaner.
	519	*/
	520	sreq.sector = 0;
	521	sreq.nr_sectors = 0;
	522	if (!lo->sock)
	523	return -EINVAL;
	524	nbd_send_req(lo, &sreq);
	525	return 0;
	526
	527	case NBD_CLEAR_SOCK:
	528	error = 0;
82d4dc5a	529	mutex_lock(&lo->tx_lock);
1da177e4	530	lo->sock = NULL;
82d4dc5a	531	mutex_unlock(&lo->tx_lock);
1da177e4 LT	532	file = lo->file;
1da177e4 LT	533	lo->file = NULL;
1da177e4	534	nbd_clear_que(lo);
4b2f0260	535	BUG_ON(!list_empty(&lo->queue_head));
1da177e4 LT	536	if (file)
	537	fput(file);
	538	return error;
	539	case NBD_SET_SOCK:
	540	if (lo->file)
	541	return -EBUSY;
	542	error = -EINVAL;
	543	file = fget(arg);
	544	if (file) {
17506041	545	inode = file->f_path.dentry->d_inode;
1da177e4 LT	546	if (S_ISSOCK(inode->i_mode)) {
	547	lo->file = file;
	548	lo->sock = SOCKET_I(inode);
	549	error = 0;
	550	} else {
	551	fput(file);
	552	}
	553	}
	554	return error;
	555	case NBD_SET_BLKSIZE:
	556	lo->blksize = arg;
	557	lo->bytesize &= ~(lo->blksize-1);
	558	inode->i_bdev->bd_inode->i_size = lo->bytesize;
	559	set_blocksize(inode->i_bdev, lo->blksize);
	560	set_capacity(lo->disk, lo->bytesize >> 9);
	561	return 0;
	562	case NBD_SET_SIZE:
	563	lo->bytesize = arg & ~(lo->blksize-1);
	564	inode->i_bdev->bd_inode->i_size = lo->bytesize;
	565	set_blocksize(inode->i_bdev, lo->blksize);
	566	set_capacity(lo->disk, lo->bytesize >> 9);
	567	return 0;
	568	case NBD_SET_SIZE_BLOCKS:
	569	lo->bytesize = ((u64) arg) * lo->blksize;
	570	inode->i_bdev->bd_inode->i_size = lo->bytesize;
	571	set_blocksize(inode->i_bdev, lo->blksize);
	572	set_capacity(lo->disk, lo->bytesize >> 9);
	573	return 0;
	574	case NBD_DO_IT:
	575	if (!lo->file)
	576	return -EINVAL;
84963048 WC	577	error = nbd_do_it(lo);
	578	if (error)
	579	return error;
1da177e4 LT	580	/* on return tidy up in case we have a signal */
	581	/* Forcibly shutdown the socket causing all listeners
	582	* to error
	583	*
	584	* FIXME: This code is duplicated from sys_shutdown, but
	585	* there should be a more generic interface rather than
	586	* calling socket ops directly here */
82d4dc5a	587	mutex_lock(&lo->tx_lock);
1da177e4 LT	588	if (lo->sock) {
	589	printk(KERN_WARNING "%s: shutting down socket\n",
	590	lo->disk->disk_name);
	591	lo->sock->ops->shutdown(lo->sock,
	592	SEND_SHUTDOWN\|RCV_SHUTDOWN);
	593	lo->sock = NULL;
	594	}
82d4dc5a	595	mutex_unlock(&lo->tx_lock);
1da177e4 LT	596	file = lo->file;
1da177e4 LT	597	lo->file = NULL;
1da177e4 LT	598	nbd_clear_que(lo);
	599	printk(KERN_WARNING "%s: queue cleared\n", lo->disk->disk_name);
	600	if (file)
	601	fput(file);
	602	return lo->harderror;
	603	case NBD_CLEAR_QUE:
4b2f0260 HX	604	/*
	605	* This is for compatibility only. The queue is always cleared
	606	* by NBD_DO_IT or NBD_CLEAR_SOCK.
	607	*/
	608	BUG_ON(!lo->sock && !list_empty(&lo->queue_head));
1da177e4 LT	609	return 0;
	610	case NBD_PRINT_DEBUG:
	611	printk(KERN_INFO "%s: next = %p, prev = %p, head = %p\n",
	612	inode->i_bdev->bd_disk->disk_name,
	613	lo->queue_head.next, lo->queue_head.prev,
	614	&lo->queue_head);
	615	return 0;
	616	}
	617	return -EINVAL;
	618	}
	619
	620	static struct block_device_operations nbd_fops =
	621	{
	622	.owner = THIS_MODULE,
	623	.ioctl = nbd_ioctl,
	624	};
	625
	626	/*
	627	* And here should be modules and kernel interface
	628	* (Just smiley confuses emacs :-)
	629	*/
	630
	631	static int __init nbd_init(void)
	632	{
	633	int err = -ENOMEM;
	634	int i;
	635
5b7b18cc	636	BUILD_BUG_ON(sizeof(struct nbd_request) != 28);
1da177e4	637
40be0c28 LMB	638	if (nbds_max > MAX_NBD) {
	639	printk(KERN_CRIT "nbd: cannot allocate more than %u nbds; %u requested.\n", MAX_NBD,
	640	nbds_max);
	641	return -EINVAL;
	642	}
	643
	644	for (i = 0; i < nbds_max; i++) {
1da177e4 LT	645	struct gendisk *disk = alloc_disk(1);
	646	if (!disk)
	647	goto out;
	648	nbd_dev[i].disk = disk;
	649	/*
	650	* The new linux 2.5 block layer implementation requires
	651	* every gendisk to have its very own request_queue struct.
	652	* These structs are big so we dynamically allocate them.
	653	*/
	654	disk->queue = blk_init_queue(do_nbd_request, &nbd_lock);
	655	if (!disk->queue) {
	656	put_disk(disk);
	657	goto out;
	658	}
	659	}
	660
	661	if (register_blkdev(NBD_MAJOR, "nbd")) {
	662	err = -EIO;
	663	goto out;
	664	}
	665
	666	printk(KERN_INFO "nbd: registered device at major %d\n", NBD_MAJOR);
	667	dprintk(DBG_INIT, "nbd: debugflags=0x%x\n", debugflags);
	668
40be0c28	669	for (i = 0; i < nbds_max; i++) {
1da177e4 LT	670	struct gendisk *disk = nbd_dev[i].disk;
	671	nbd_dev[i].file = NULL;
	672	nbd_dev[i].magic = LO_MAGIC;
	673	nbd_dev[i].flags = 0;
	674	spin_lock_init(&nbd_dev[i].queue_lock);
	675	INIT_LIST_HEAD(&nbd_dev[i].queue_head);
82d4dc5a	676	mutex_init(&nbd_dev[i].tx_lock);
4b2f0260	677	init_waitqueue_head(&nbd_dev[i].active_wq);
1da177e4 LT	678	nbd_dev[i].blksize = 1024;
	679	nbd_dev[i].bytesize = 0x7ffffc00ULL << 10; /* 2TB */
	680	disk->major = NBD_MAJOR;
	681	disk->first_minor = i;
	682	disk->fops = &nbd_fops;
	683	disk->private_data = &nbd_dev[i];
	684	disk->flags \|= GENHD_FL_SUPPRESS_PARTITION_INFO;
	685	sprintf(disk->disk_name, "nbd%d", i);
1da177e4 LT	686	set_capacity(disk, 0x7ffffc00ULL << 1); /* 2 TB */
	687	add_disk(disk);
	688	}
	689
	690	return 0;
	691	out:
	692	while (i--) {
	693	blk_cleanup_queue(nbd_dev[i].disk->queue);
	694	put_disk(nbd_dev[i].disk);
	695	}
	696	return err;
	697	}
	698
	699	static void __exit nbd_cleanup(void)
	700	{
	701	int i;
40be0c28	702	for (i = 0; i < nbds_max; i++) {
1da177e4	703	struct gendisk *disk = nbd_dev[i].disk;
40be0c28	704	nbd_dev[i].magic = 0;
1da177e4 LT	705	if (disk) {
	706	del_gendisk(disk);
	707	blk_cleanup_queue(disk->queue);
	708	put_disk(disk);
	709	}
	710	}
1da177e4 LT	711	unregister_blkdev(NBD_MAJOR, "nbd");
	712	printk(KERN_INFO "nbd: unregistered device at major %d\n", NBD_MAJOR);
	713	}
	714
	715	module_init(nbd_init);
	716	module_exit(nbd_cleanup);
	717
	718	MODULE_DESCRIPTION("Network Block Device");
	719	MODULE_LICENSE("GPL");
	720
40be0c28 LMB	721	module_param(nbds_max, int, 0444);
40be0c28 LMB	722	MODULE_PARM_DESC(nbds_max, "How many network block devices to initialize.");
1da177e4 LT	723	#ifndef NDEBUG
	724	module_param(debugflags, int, 0644);
	725	MODULE_PARM_DESC(debugflags, "flags for controlling debug output");
	726	#endif