nbd.c 19.4 KB
Newer Older
L
Linus Torvalds 已提交
1 2 3 4 5 6 7 8 9
/*
 * 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>
 *
10
 * This file is released under GPLv2 or later.
L
Linus Torvalds 已提交
11
 *
12
 * (part of code stolen from loop.c)
L
Linus Torvalds 已提交
13 14 15 16 17 18 19 20 21 22 23 24 25 26
 */

#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>
27 28 29
#include <linux/compiler.h>
#include <linux/err.h>
#include <linux/kernel.h>
L
Linus Torvalds 已提交
30
#include <net/sock.h>
31
#include <linux/net.h>
32
#include <linux/kthread.h>
L
Linus Torvalds 已提交
33 34

#include <asm/uaccess.h>
35
#include <asm/system.h>
L
Linus Torvalds 已提交
36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56
#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 */

57
static unsigned int nbds_max = 16;
58
static struct nbd_device *nbd_dev;
L
Linus Torvalds 已提交
59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103

/*
 * 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)
{
104
	int error = req->errors ? -EIO : 0;
105
	struct request_queue *q = req->q;
L
Linus Torvalds 已提交
106 107 108
	unsigned long flags;

	dprintk(DBG_BLKDEV, "%s: request %p: %s\n", req->rq_disk->disk_name,
109
			req, error ? "failed" : "done");
L
Linus Torvalds 已提交
110 111

	spin_lock_irqsave(q->queue_lock, flags);
112
	__blk_end_request(req, error, req->nr_sectors << 9);
L
Linus Torvalds 已提交
113 114 115
	spin_unlock_irqrestore(q->queue_lock, flags);
}

116 117 118 119 120 121 122 123 124 125 126 127 128
static void sock_shutdown(struct nbd_device *lo, int lock)
{
	/* 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 */
	if (lock)
		mutex_lock(&lo->tx_lock);
	if (lo->sock) {
		printk(KERN_WARNING "%s: shutting down socket\n",
			lo->disk->disk_name);
129
		kernel_sock_shutdown(lo->sock, SHUT_RDWR);
130 131 132 133 134 135 136 137 138 139 140 141 142 143 144
		lo->sock = NULL;
	}
	if (lock)
		mutex_unlock(&lo->tx_lock);
}

static void nbd_xmit_timeout(unsigned long arg)
{
	struct task_struct *task = (struct task_struct *)arg;

	printk(KERN_WARNING "nbd: killing hung xmit (%s, pid: %d)\n",
		task->comm, task->pid);
	force_sig(SIGKILL, task);
}

L
Linus Torvalds 已提交
145 146 147
/*
 *  Send or receive packet.
 */
148
static int sock_xmit(struct nbd_device *lo, int send, void *buf, int size,
L
Linus Torvalds 已提交
149 150
		int msg_flags)
{
151
	struct socket *sock = lo->sock;
L
Linus Torvalds 已提交
152 153 154
	int result;
	struct msghdr msg;
	struct kvec iov;
155
	sigset_t blocked, oldset;
L
Linus Torvalds 已提交
156

157 158 159 160 161 162
	if (unlikely(!sock)) {
		printk(KERN_ERR "%s: Attempted %s on closed socket in sock_xmit\n",
		       lo->disk->disk_name, (send ? "send" : "recv"));
		return -EINVAL;
	}

L
Linus Torvalds 已提交
163 164
	/* Allow interception of SIGKILL only
	 * Don't allow other signals to interrupt the transmission */
165 166
	siginitsetinv(&blocked, sigmask(SIGKILL));
	sigprocmask(SIG_SETMASK, &blocked, &oldset);
L
Linus Torvalds 已提交
167 168 169 170 171 172 173 174 175 176 177

	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;

178 179 180 181 182 183 184 185 186 187
		if (send) {
			struct timer_list ti;

			if (lo->xmit_timeout) {
				init_timer(&ti);
				ti.function = nbd_xmit_timeout;
				ti.data = (unsigned long)current;
				ti.expires = jiffies + lo->xmit_timeout;
				add_timer(&ti);
			}
L
Linus Torvalds 已提交
188
			result = kernel_sendmsg(sock, &msg, &iov, 1, size);
189 190 191
			if (lo->xmit_timeout)
				del_timer_sync(&ti);
		} else
L
Linus Torvalds 已提交
192 193 194 195 196
			result = kernel_recvmsg(sock, &msg, &iov, 1, size, 0);

		if (signal_pending(current)) {
			siginfo_t info;
			printk(KERN_WARNING "nbd (pid %d: %s) got signal %d\n",
197
				task_pid_nr(current), current->comm,
198
				dequeue_signal_lock(current, &current->blocked, &info));
L
Linus Torvalds 已提交
199
			result = -EINTR;
200
			sock_shutdown(lo, !send);
L
Linus Torvalds 已提交
201 202 203 204 205 206 207 208 209 210 211 212
			break;
		}

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

213
	sigprocmask(SIG_SETMASK, &oldset, NULL);
L
Linus Torvalds 已提交
214 215 216 217

	return result;
}

218
static inline int sock_send_bvec(struct nbd_device *lo, struct bio_vec *bvec,
L
Linus Torvalds 已提交
219 220 221 222
		int flags)
{
	int result;
	void *kaddr = kmap(bvec->bv_page);
223
	result = sock_xmit(lo, 1, kaddr + bvec->bv_offset, bvec->bv_len, flags);
L
Linus Torvalds 已提交
224 225 226 227
	kunmap(bvec->bv_page);
	return result;
}

228
/* always call with the tx_lock held */
L
Linus Torvalds 已提交
229 230
static int nbd_send_req(struct nbd_device *lo, struct request *req)
{
231
	int result, flags;
L
Linus Torvalds 已提交
232 233 234 235 236 237 238 239 240 241 242 243 244 245
	struct nbd_request request;
	unsigned long size = req->nr_sectors << 9;

	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);
246 247
	result = sock_xmit(lo, 1, &request, sizeof(request),
			(nbd_cmd(req) == NBD_CMD_WRITE) ? MSG_MORE : 0);
L
Linus Torvalds 已提交
248 249 250 251 252 253 254
	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) {
255 256
		struct req_iterator iter;
		struct bio_vec *bvec;
L
Linus Torvalds 已提交
257 258 259 260
		/*
		 * we are really probing at internals to determine
		 * whether to set MSG_MORE or not...
		 */
261
		rq_for_each_segment(bvec, req, iter) {
262 263 264 265 266
			flags = 0;
			if (!rq_iter_last(req, iter))
				flags = MSG_MORE;
			dprintk(DBG_TX, "%s: request %p: sending %d bytes data\n",
					lo->disk->disk_name, req, bvec->bv_len);
267
			result = sock_send_bvec(lo, bvec, flags);
268 269 270 271 272
			if (result <= 0) {
				printk(KERN_ERR "%s: Send data failed (result %d)\n",
						lo->disk->disk_name, result);
				goto error_out;
			}
L
Linus Torvalds 已提交
273 274 275 276 277 278 279 280
		}
	}
	return 0;

error_out:
	return 1;
}

281 282
static struct request *nbd_find_request(struct nbd_device *lo,
					struct request *xreq)
L
Linus Torvalds 已提交
283
{
284
	struct request *req, *tmp;
285
	int err;
L
Linus Torvalds 已提交
286

287 288 289 290
	err = wait_event_interruptible(lo->active_wq, lo->active_req != xreq);
	if (unlikely(err))
		goto out;

L
Linus Torvalds 已提交
291
	spin_lock(&lo->queue_lock);
292
	list_for_each_entry_safe(req, tmp, &lo->queue_head, queuelist) {
L
Linus Torvalds 已提交
293 294 295 296 297 298 299
		if (req != xreq)
			continue;
		list_del_init(&req->queuelist);
		spin_unlock(&lo->queue_lock);
		return req;
	}
	spin_unlock(&lo->queue_lock);
300 301 302 303 304

	err = -ENOENT;

out:
	return ERR_PTR(err);
L
Linus Torvalds 已提交
305 306
}

307
static inline int sock_recv_bvec(struct nbd_device *lo, struct bio_vec *bvec)
L
Linus Torvalds 已提交
308 309 310
{
	int result;
	void *kaddr = kmap(bvec->bv_page);
311
	result = sock_xmit(lo, 0, kaddr + bvec->bv_offset, bvec->bv_len,
L
Linus Torvalds 已提交
312 313 314 315 316 317 318 319 320 321 322 323 324
			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;

	reply.magic = 0;
325
	result = sock_xmit(lo, 0, &reply, sizeof(reply), MSG_WAITALL);
L
Linus Torvalds 已提交
326 327 328 329 330
	if (result <= 0) {
		printk(KERN_ERR "%s: Receive control failed (result %d)\n",
				lo->disk->disk_name, result);
		goto harderror;
	}
331 332 333 334 335 336 337 338 339

	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;
	}

340
	req = nbd_find_request(lo, *(struct request **)reply.handle);
341 342 343 344 345
	if (unlikely(IS_ERR(req))) {
		result = PTR_ERR(req);
		if (result != -ENOENT)
			goto harderror;

L
Linus Torvalds 已提交
346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361
		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) {
362 363 364 365
		struct req_iterator iter;
		struct bio_vec *bvec;

		rq_for_each_segment(bvec, req, iter) {
366
			result = sock_recv_bvec(lo, bvec);
367 368 369 370 371 372 373 374
			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);
L
Linus Torvalds 已提交
375 376 377 378 379 380 381 382
		}
	}
	return req;
harderror:
	lo->harderror = result;
	return NULL;
}

383 384
static ssize_t pid_show(struct device *dev,
			struct device_attribute *attr, char *buf)
385
{
386 387 388
	struct gendisk *disk = dev_to_disk(dev);

	return sprintf(buf, "%ld\n",
389 390 391
		(long) ((struct nbd_device *)disk->private_data)->pid);
}

392 393
static struct device_attribute pid_attr = {
	.attr = { .name = "pid", .mode = S_IRUGO, .owner = THIS_MODULE },
394 395 396
	.show = pid_show,
};

397
static int nbd_do_it(struct nbd_device *lo)
L
Linus Torvalds 已提交
398 399
{
	struct request *req;
400
	int ret;
L
Linus Torvalds 已提交
401 402 403

	BUG_ON(lo->magic != LO_MAGIC);

404
	lo->pid = current->pid;
405
	ret = sysfs_create_file(&lo->disk->dev.kobj, &pid_attr.attr);
406 407 408 409
	if (ret) {
		printk(KERN_ERR "nbd: sysfs_create_file failed!");
		return ret;
	}
410

L
Linus Torvalds 已提交
411 412
	while ((req = nbd_read_stat(lo)) != NULL)
		nbd_end_request(req);
413

414
	sysfs_remove_file(&lo->disk->dev.kobj, &pid_attr.attr);
415
	return 0;
L
Linus Torvalds 已提交
416 417 418 419 420 421 422 423
}

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

	BUG_ON(lo->magic != LO_MAGIC);

424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441
	/*
	 * 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);
	}
L
Linus Torvalds 已提交
442 443
}

444

445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523
static void nbd_handle_req(struct nbd_device *lo, struct request *req)
{
	if (!blk_fs_request(req))
		goto error_out;

	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;

	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);
		return;
	}

	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);

	return;

error_out:
	req->errors++;
	nbd_end_request(req);
}

static int nbd_thread(void *data)
{
	struct nbd_device *lo = data;
	struct request *req;

	set_user_nice(current, -20);
	while (!kthread_should_stop() || !list_empty(&lo->waiting_queue)) {
		/* wait for something to do */
		wait_event_interruptible(lo->waiting_wq,
					 kthread_should_stop() ||
					 !list_empty(&lo->waiting_queue));

		/* extract request */
		if (list_empty(&lo->waiting_queue))
			continue;

		spin_lock_irq(&lo->queue_lock);
		req = list_entry(lo->waiting_queue.next, struct request,
				 queuelist);
		list_del_init(&req->queuelist);
		spin_unlock_irq(&lo->queue_lock);

		/* handle request */
		nbd_handle_req(lo, req);
	}
	return 0;
}

L
Linus Torvalds 已提交
524 525 526
/*
 * We always wait for result of write, for now. It would be nice to make it optional
 * in future
527
 * if ((rq_data_dir(req) == WRITE) && (lo->flags & NBD_WRITE_NOCHK))
L
Linus Torvalds 已提交
528 529 530
 *   { printk( "Warning: Ignoring result!\n"); nbd_end_request( req ); }
 */

531
static void do_nbd_request(struct request_queue * q)
L
Linus Torvalds 已提交
532 533 534 535 536 537 538
{
	struct request *req;
	
	while ((req = elv_next_request(q)) != NULL) {
		struct nbd_device *lo;

		blkdev_dequeue_request(req);
539 540 541

		spin_unlock_irq(q->queue_lock);

542 543
		dprintk(DBG_BLKDEV, "%s: request %p: dequeued (flags=%x)\n",
				req->rq_disk->disk_name, req, req->cmd_type);
L
Linus Torvalds 已提交
544 545 546 547 548

		lo = req->rq_disk->private_data;

		BUG_ON(lo->magic != LO_MAGIC);

549 550 551
		spin_lock_irq(&lo->queue_lock);
		list_add_tail(&req->queuelist, &lo->waiting_queue);
		spin_unlock_irq(&lo->queue_lock);
L
Linus Torvalds 已提交
552

553
		wake_up(&lo->waiting_wq);
554

L
Linus Torvalds 已提交
555 556 557 558 559 560 561 562 563 564
		spin_lock_irq(q->queue_lock);
	}
}

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 ;
565
	struct task_struct *thread;
L
Linus Torvalds 已提交
566 567 568 569 570 571 572 573 574 575 576 577 578

	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);
579
		sreq.cmd_type = REQ_TYPE_SPECIAL;
L
Linus Torvalds 已提交
580 581 582 583 584 585 586 587 588 589
		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;
590
		mutex_lock(&lo->tx_lock);
L
Linus Torvalds 已提交
591
                nbd_send_req(lo, &sreq);
592
		mutex_unlock(&lo->tx_lock);
L
Linus Torvalds 已提交
593 594 595 596
                return 0;
 
	case NBD_CLEAR_SOCK:
		error = 0;
597
		mutex_lock(&lo->tx_lock);
L
Linus Torvalds 已提交
598
		lo->sock = NULL;
599
		mutex_unlock(&lo->tx_lock);
L
Linus Torvalds 已提交
600 601 602
		file = lo->file;
		lo->file = NULL;
		nbd_clear_que(lo);
603
		BUG_ON(!list_empty(&lo->queue_head));
L
Linus Torvalds 已提交
604 605 606 607 608 609 610 611 612
		if (file)
			fput(file);
		return error;
	case NBD_SET_SOCK:
		if (lo->file)
			return -EBUSY;
		error = -EINVAL;
		file = fget(arg);
		if (file) {
J
Josef Sipek 已提交
613
			inode = file->f_path.dentry->d_inode;
L
Linus Torvalds 已提交
614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635
			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;
636 637 638
	case NBD_SET_TIMEOUT:
		lo->xmit_timeout = arg * HZ;
		return 0;
L
Linus Torvalds 已提交
639 640 641 642 643 644 645 646 647
	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;
648 649 650 651
		thread = kthread_create(nbd_thread, lo, lo->disk->disk_name);
		if (IS_ERR(thread))
			return PTR_ERR(thread);
		wake_up_process(thread);
652
		error = nbd_do_it(lo);
653
		kthread_stop(thread);
654 655
		if (error)
			return error;
656
		sock_shutdown(lo, 1);
L
Linus Torvalds 已提交
657 658 659 660 661 662
		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);
663 664 665
		lo->bytesize = 0;
		inode->i_bdev->bd_inode->i_size = 0;
		set_capacity(lo->disk, 0);
L
Linus Torvalds 已提交
666 667
		return lo->harderror;
	case NBD_CLEAR_QUE:
668 669 670 671 672
		/*
		 * 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));
L
Linus Torvalds 已提交
673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699
		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;

700
	BUILD_BUG_ON(sizeof(struct nbd_request) != 28);
L
Linus Torvalds 已提交
701

702 703 704
	nbd_dev = kcalloc(nbds_max, sizeof(*nbd_dev), GFP_KERNEL);
	if (!nbd_dev)
		return -ENOMEM;
705 706

	for (i = 0; i < nbds_max; i++) {
L
Linus Torvalds 已提交
707
		struct gendisk *disk = alloc_disk(1);
708
		elevator_t *old_e;
L
Linus Torvalds 已提交
709 710 711 712 713 714 715 716 717 718 719 720 721
		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;
		}
722 723 724 725 726
		old_e = disk->queue->elevator;
		if (elevator_init(disk->queue, "deadline") == 0 ||
			elevator_init(disk->queue, "noop") == 0) {
				elevator_exit(old_e);
		}
L
Linus Torvalds 已提交
727 728 729 730 731 732 733 734 735 736
	}

	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);

737
	for (i = 0; i < nbds_max; i++) {
L
Linus Torvalds 已提交
738 739 740 741
		struct gendisk *disk = nbd_dev[i].disk;
		nbd_dev[i].file = NULL;
		nbd_dev[i].magic = LO_MAGIC;
		nbd_dev[i].flags = 0;
742
		INIT_LIST_HEAD(&nbd_dev[i].waiting_queue);
L
Linus Torvalds 已提交
743 744
		spin_lock_init(&nbd_dev[i].queue_lock);
		INIT_LIST_HEAD(&nbd_dev[i].queue_head);
745
		mutex_init(&nbd_dev[i].tx_lock);
746
		init_waitqueue_head(&nbd_dev[i].active_wq);
747
		init_waitqueue_head(&nbd_dev[i].waiting_wq);
L
Linus Torvalds 已提交
748
		nbd_dev[i].blksize = 1024;
749
		nbd_dev[i].bytesize = 0;
L
Linus Torvalds 已提交
750 751 752 753 754 755
		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);
756
		set_capacity(disk, 0);
L
Linus Torvalds 已提交
757 758 759 760 761 762 763 764 765 766 767 768 769 770 771
		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;
772
	for (i = 0; i < nbds_max; i++) {
L
Linus Torvalds 已提交
773
		struct gendisk *disk = nbd_dev[i].disk;
774
		nbd_dev[i].magic = 0;
L
Linus Torvalds 已提交
775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790
		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");

791 792
module_param(nbds_max, int, 0444);
MODULE_PARM_DESC(nbds_max, "How many network block devices to initialize.");
L
Linus Torvalds 已提交
793 794 795 796
#ifndef NDEBUG
module_param(debugflags, int, 0644);
MODULE_PARM_DESC(debugflags, "flags for controlling debug output");
#endif