trans_fd.c 23.9 KB
Newer Older
1 2 3 4 5 6 7
/*
 * linux/fs/9p/trans_fd.c
 *
 * Fd transport layer.  Includes deprecated socket layer.
 *
 *  Copyright (C) 2006 by Russ Cox <rsc@swtch.com>
 *  Copyright (C) 2004-2005 by Latchesar Ionkov <lucho@ionkov.net>
8
 *  Copyright (C) 2004-2008 by Eric Van Hensbergen <ericvh@gmail.com>
9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27
 *  Copyright (C) 1997-2002 by Ron Minnich <rminnich@sarnoff.com>
 *
 *  This program is free software; you can redistribute it and/or modify
 *  it under the terms of the GNU General Public License version 2
 *  as published by the Free Software Foundation.
 *
 *  This program is distributed in the hope that it will be useful,
 *  but WITHOUT ANY WARRANTY; without even the implied warranty of
 *  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 *  GNU General Public License for more details.
 *
 *  You should have received a copy of the GNU General Public License
 *  along with this program; if not, write to:
 *  Free Software Foundation
 *  51 Franklin Street, Fifth Floor
 *  Boston, MA  02111-1301  USA
 *
 */

28 29
#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt

30 31 32 33
#include <linux/in.h>
#include <linux/module.h>
#include <linux/net.h>
#include <linux/ipv6.h>
34
#include <linux/kthread.h>
35 36 37 38 39 40 41
#include <linux/errno.h>
#include <linux/kernel.h>
#include <linux/un.h>
#include <linux/uaccess.h>
#include <linux/inet.h>
#include <linux/idr.h>
#include <linux/file.h>
42
#include <linux/parser.h>
43
#include <linux/slab.h>
44
#include <net/9p/9p.h>
45
#include <net/9p/client.h>
46 47
#include <net/9p/transport.h>

Al Viro's avatar
Al Viro committed
48 49
#include <linux/syscalls.h> /* killme */

50
#define P9_PORT 564
51
#define MAX_SOCK_BUF (64*1024)
52
#define MAXPOLLWADDR	2
53

54 55 56 57 58 59 60 61
/**
 * struct p9_fd_opts - per-transport options
 * @rfd: file descriptor for reading (trans=fd)
 * @wfd: file descriptor for writing (trans=fd)
 * @port: port to connect to (trans=tcp)
 *
 */

62 63 64 65 66
struct p9_fd_opts {
	int rfd;
	int wfd;
	u16 port;
};
67

68 69 70 71 72 73 74 75
/**
 * struct p9_trans_fd - transport state
 * @rd: reference to file to read from
 * @wr: reference of file to write to
 * @conn: connection state reference
 *
 */

76 77 78
struct p9_trans_fd {
	struct file *rd;
	struct file *wr;
79
	struct p9_conn *conn;
80 81
};

82 83 84 85
/*
  * Option Parsing (code inspired by NFS code)
  *  - a little lazy - parse all fd-transport options
  */
86

87 88
enum {
	/* Options that take integer arguments */
89
	Opt_port, Opt_rfdno, Opt_wfdno, Opt_err,
90
};
91

92
static const match_table_t tokens = {
93 94 95
	{Opt_port, "port=%u"},
	{Opt_rfdno, "rfdno=%u"},
	{Opt_wfdno, "wfdno=%u"},
96
	{Opt_err, NULL},
97
};
98

99 100 101 102 103 104 105
enum {
	Rworksched = 1,		/* read work scheduled or running */
	Rpending = 2,		/* can read */
	Wworksched = 4,		/* write work scheduled or running */
	Wpending = 8,		/* can write */
};

Tejun Heo's avatar
Tejun Heo committed
106 107 108 109
struct p9_poll_wait {
	struct p9_conn *conn;
	wait_queue_t wait;
	wait_queue_head_t *wait_addr;
110 111 112 113 114
};

/**
 * struct p9_conn - fd mux connection state information
 * @mux_list: list link for mux to manage multiple connections (?)
115
 * @client: reference to client instance for this connection
116 117 118
 * @err: error state
 * @req_list: accounting for requests which have been sent
 * @unsent_req_list: accounting for requests that haven't been sent
119 120 121
 * @req: current request being processed (if any)
 * @tmp_buf: temporary buffer to read in header
 * @rsize: amount to read for current frame
122 123 124 125 126
 * @rpos: read position in current frame
 * @rbuf: current read buffer
 * @wpos: write position for current frame
 * @wsize: amount of data to write for current frame
 * @wbuf: current write buffer
Abhishek Kulkarni's avatar
Abhishek Kulkarni committed
127
 * @poll_pending_link: pending links to be polled per conn
128 129 130 131 132 133 134
 * @poll_wait: array of wait_q's for various worker threads
 * @pt: poll state
 * @rq: current read work
 * @wq: current write work
 * @wsched: ????
 *
 */
135 136 137

struct p9_conn {
	struct list_head mux_list;
138
	struct p9_client *client;
139 140 141
	int err;
	struct list_head req_list;
	struct list_head unsent_req_list;
142 143 144
	struct p9_req_t *req;
	char tmp_buf[7];
	int rsize;
145 146 147 148 149
	int rpos;
	char *rbuf;
	int wpos;
	int wsize;
	char *wbuf;
Tejun Heo's avatar
Tejun Heo committed
150 151
	struct list_head poll_pending_link;
	struct p9_poll_wait poll_wait[MAXPOLLWADDR];
152 153 154 155 156 157
	poll_table pt;
	struct work_struct rq;
	struct work_struct wq;
	unsigned long wsched;
};

158 159
static void p9_poll_workfn(struct work_struct *work);

Tejun Heo's avatar
Tejun Heo committed
160 161
static DEFINE_SPINLOCK(p9_poll_lock);
static LIST_HEAD(p9_poll_pending_list);
162
static DECLARE_WORK(p9_poll_work, p9_poll_workfn);
163

Tejun Heo's avatar
Tejun Heo committed
164
static void p9_mux_poll_stop(struct p9_conn *m)
165
{
Tejun Heo's avatar
Tejun Heo committed
166 167
	unsigned long flags;
	int i;
168

Tejun Heo's avatar
Tejun Heo committed
169 170
	for (i = 0; i < ARRAY_SIZE(m->poll_wait); i++) {
		struct p9_poll_wait *pwait = &m->poll_wait[i];
171

Tejun Heo's avatar
Tejun Heo committed
172 173 174
		if (pwait->wait_addr) {
			remove_wait_queue(pwait->wait_addr, &pwait->wait);
			pwait->wait_addr = NULL;
175 176 177
		}
	}

Tejun Heo's avatar
Tejun Heo committed
178 179 180
	spin_lock_irqsave(&p9_poll_lock, flags);
	list_del_init(&m->poll_pending_link);
	spin_unlock_irqrestore(&p9_poll_lock, flags);
181 182 183
}

/**
184 185 186
 * p9_conn_cancel - cancel all pending requests with error
 * @m: mux data
 * @err: error code
187 188
 *
 */
189

190
static void p9_conn_cancel(struct p9_conn *m, int err)
191
{
192
	struct p9_req_t *req, *rtmp;
193
	unsigned long flags;
194
	LIST_HEAD(cancel_list);
195

196
	p9_debug(P9_DEBUG_ERROR, "mux %p err %d\n", m, err);
197

198
	spin_lock_irqsave(&m->client->lock, flags);
199 200 201 202 203 204 205 206

	if (m->err) {
		spin_unlock_irqrestore(&m->client->lock, flags);
		return;
	}

	m->err = err;

207
	list_for_each_entry_safe(req, rtmp, &m->req_list, req_list) {
208 209 210
		req->status = REQ_STATUS_ERROR;
		if (!req->t_err)
			req->t_err = err;
211 212 213
		list_move(&req->req_list, &cancel_list);
	}
	list_for_each_entry_safe(req, rtmp, &m->unsent_req_list, req_list) {
214 215 216
		req->status = REQ_STATUS_ERROR;
		if (!req->t_err)
			req->t_err = err;
217
		list_move(&req->req_list, &cancel_list);
218
	}
219
	spin_unlock_irqrestore(&m->client->lock, flags);
220

221
	list_for_each_entry_safe(req, rtmp, &cancel_list, req_list) {
222
		p9_debug(P9_DEBUG_ERROR, "call back req %p\n", req);
223
		list_del(&req->req_list);
224
		p9_client_cb(m->client, req);
225 226 227
	}
}

228
static int
229
p9_fd_poll(struct p9_client *client, struct poll_table_struct *pt)
230
{
231 232
	int ret, n;
	struct p9_trans_fd *ts = NULL;
233

234 235
	if (client && client->status == Connected)
		ts = client->trans;
236

237 238
	if (!ts)
		return -EREMOTEIO;
239

240 241
	if (!ts->rd->f_op || !ts->rd->f_op->poll)
		return -EIO;
242

243 244
	if (!ts->wr->f_op || !ts->wr->f_op->poll)
		return -EIO;
Tejun Heo's avatar
Tejun Heo committed
245

246 247 248
	ret = ts->rd->f_op->poll(ts->rd, pt);
	if (ret < 0)
		return ret;
Tejun Heo's avatar
Tejun Heo committed
249

250 251 252 253 254 255 256 257
	if (ts->rd != ts->wr) {
		n = ts->wr->f_op->poll(ts->wr, pt);
		if (n < 0)
			return n;
		ret = (ret & ~POLLOUT) | (n & ~POLLIN);
	}

	return ret;
Tejun Heo's avatar
Tejun Heo committed
258 259
}

260
/**
261 262 263 264
 * p9_fd_read- read from a fd
 * @client: client instance
 * @v: buffer to receive data into
 * @len: size of receive buffer
265
 *
266
 */
267

268
static int p9_fd_read(struct p9_client *client, void *v, int len)
269
{
270 271
	int ret;
	struct p9_trans_fd *ts = NULL;
272

273 274
	if (client && client->status != Disconnected)
		ts = client->trans;
275

276 277
	if (!ts)
		return -EREMOTEIO;
278

279
	if (!(ts->rd->f_flags & O_NONBLOCK))
280
		p9_debug(P9_DEBUG_ERROR, "blocking read ...\n");
281

282 283 284 285
	ret = kernel_read(ts->rd, ts->rd->f_pos, v, len);
	if (ret <= 0 && ret != -ERESTARTSYS && ret != -EAGAIN)
		client->status = Disconnected;
	return ret;
286 287 288
}

/**
289 290
 * p9_read_work - called when there is some data to be read from a transport
 * @work: container of work to be done
291
 *
292
 */
293

294
static void p9_read_work(struct work_struct *work)
295
{
296 297 298 299
	int n, err;
	struct p9_conn *m;

	m = container_of(work, struct p9_conn, rq);
300 301 302 303

	if (m->err < 0)
		return;

304
	p9_debug(P9_DEBUG_TRANS, "start mux %p pos %d\n", m, m->rpos);
305

306 307
	if (!m->rbuf) {
		m->rbuf = m->tmp_buf;
308
		m->rpos = 0;
309
		m->rsize = 7; /* start by reading header */
310 311
	}

312
	clear_bit(Rpending, &m->wsched);
313 314
	p9_debug(P9_DEBUG_TRANS, "read mux %p pos %d size: %d = %d\n",
		 m, m->rpos, m->rsize, m->rsize-m->rpos);
315
	err = p9_fd_read(m->client, m->rbuf + m->rpos,
316
						m->rsize - m->rpos);
317
	p9_debug(P9_DEBUG_TRANS, "mux %p got %d bytes\n", m, err);
318
	if (err == -EAGAIN) {
Simon Derr's avatar
Simon Derr committed
319
		goto end_clear;
320 321
	}

322 323 324 325
	if (err <= 0)
		goto error;

	m->rpos += err;
326 327 328

	if ((!m->req) && (m->rpos == m->rsize)) { /* header read in */
		u16 tag;
329
		p9_debug(P9_DEBUG_TRANS, "got new header\n");
330 331

		n = le32_to_cpu(*(__le32 *) m->rbuf); /* read packet size */
332
		if (n >= m->client->msize) {
333 334
			p9_debug(P9_DEBUG_ERROR,
				 "requested packet size too big: %d\n", n);
335 336 337 338
			err = -EIO;
			goto error;
		}

339
		tag = le16_to_cpu(*(__le16 *) (m->rbuf+5)); /* read tag */
340 341
		p9_debug(P9_DEBUG_TRANS,
			 "mux %p pkt: size: %d bytes tag: %d\n", m, n, tag);
342 343

		m->req = p9_tag_lookup(m->client, tag);
344 345
		if (!m->req || (m->req->status != REQ_STATUS_SENT &&
					m->req->status != REQ_STATUS_FLSH)) {
346 347
			p9_debug(P9_DEBUG_ERROR, "Unexpected packet tag %d\n",
				 tag);
348 349 350 351 352 353
			err = -EIO;
			goto error;
		}

		if (m->req->rc == NULL) {
			m->req->rc = kmalloc(sizeof(struct p9_fcall) +
354
						m->client->msize, GFP_NOFS);
355 356 357 358 359 360 361 362 363 364
			if (!m->req->rc) {
				m->req = NULL;
				err = -ENOMEM;
				goto error;
			}
		}
		m->rbuf = (char *)m->req->rc + sizeof(struct p9_fcall);
		memcpy(m->rbuf, m->tmp_buf, m->rsize);
		m->rsize = n;
	}
365

366 367
	/* not an else because some packets (like clunk) have no payload */
	if ((m->req) && (m->rpos == m->rsize)) { /* packet is read in */
368
		p9_debug(P9_DEBUG_TRANS, "got new packet\n");
369
		spin_lock(&m->client->lock);
370 371
		if (m->req->status != REQ_STATUS_ERROR)
			m->req->status = REQ_STATUS_RCVD;
372
		list_del(&m->req->req_list);
373
		spin_unlock(&m->client->lock);
374
		p9_client_cb(m->client, m->req);
375 376 377 378
		m->rbuf = NULL;
		m->rpos = 0;
		m->rsize = 0;
		m->req = NULL;
379 380
	}

Simon Derr's avatar
Simon Derr committed
381 382 383
end_clear:
	clear_bit(Rworksched, &m->wsched);

384 385 386 387 388 389
	if (!list_empty(&m->req_list)) {
		if (test_and_clear_bit(Rpending, &m->wsched))
			n = POLLIN;
		else
			n = p9_fd_poll(m->client, NULL);

Simon Derr's avatar
Simon Derr committed
390
		if ((n & POLLIN) && !test_and_set_bit(Rworksched, &m->wsched)) {
391
			p9_debug(P9_DEBUG_TRANS, "sched read work %p\n", m);
392
			schedule_work(&m->rq);
Simon Derr's avatar
Simon Derr committed
393 394
		}
	}
395 396 397 398 399 400 401 402 403 404 405 406

	return;
error:
	p9_conn_cancel(m, err);
	clear_bit(Rworksched, &m->wsched);
}

/**
 * p9_fd_write - write to a socket
 * @client: client instance
 * @v: buffer to send data from
 * @len: size of send buffer
407
 *
408
 */
409

410
static int p9_fd_write(struct p9_client *client, void *v, int len)
411
{
412 413 414
	int ret;
	mm_segment_t oldfs;
	struct p9_trans_fd *ts = NULL;
415

416 417
	if (client && client->status != Disconnected)
		ts = client->trans;
418

419 420
	if (!ts)
		return -EREMOTEIO;
421

422
	if (!(ts->wr->f_flags & O_NONBLOCK))
423
		p9_debug(P9_DEBUG_ERROR, "blocking write ...\n");
Tejun Heo's avatar
Tejun Heo committed
424

425 426 427
	oldfs = get_fs();
	set_fs(get_ds());
	/* The cast to a user pointer is valid due to the set_fs() */
428
	ret = vfs_write(ts->wr, (__force void __user *)v, len, &ts->wr->f_pos);
429
	set_fs(oldfs);
Tejun Heo's avatar
Tejun Heo committed
430

431 432 433
	if (ret <= 0 && ret != -ERESTARTSYS && ret != -EAGAIN)
		client->status = Disconnected;
	return ret;
434 435 436 437
}

/**
 * p9_write_work - called when a transport can send some data
438 439
 * @work: container for work to be done
 *
440
 */
441

442 443 444 445
static void p9_write_work(struct work_struct *work)
{
	int n, err;
	struct p9_conn *m;
446
	struct p9_req_t *req;
447 448 449 450 451 452 453 454 455

	m = container_of(work, struct p9_conn, wq);

	if (m->err < 0) {
		clear_bit(Wworksched, &m->wsched);
		return;
	}

	if (!m->wsize) {
456
		spin_lock(&m->client->lock);
457 458
		if (list_empty(&m->unsent_req_list)) {
			clear_bit(Wworksched, &m->wsched);
459
			spin_unlock(&m->client->lock);
460 461 462
			return;
		}

463
		req = list_entry(m->unsent_req_list.next, struct p9_req_t,
464
			       req_list);
465
		req->status = REQ_STATUS_SENT;
466
		p9_debug(P9_DEBUG_TRANS, "move req %p\n", req);
467 468
		list_move_tail(&req->req_list, &m->req_list);

469 470
		m->wbuf = req->tc->sdata;
		m->wsize = req->tc->size;
471
		m->wpos = 0;
472
		spin_unlock(&m->client->lock);
473 474
	}

475 476
	p9_debug(P9_DEBUG_TRANS, "mux %p pos %d size %d\n",
		 m, m->wpos, m->wsize);
477
	clear_bit(Wpending, &m->wsched);
478
	err = p9_fd_write(m->client, m->wbuf + m->wpos, m->wsize - m->wpos);
479
	p9_debug(P9_DEBUG_TRANS, "mux %p sent %d bytes\n", m, err);
Simon Derr's avatar
Simon Derr committed
480 481 482
	if (err == -EAGAIN)
		goto end_clear;

483 484 485 486 487 488 489 490 491 492 493 494

	if (err < 0)
		goto error;
	else if (err == 0) {
		err = -EREMOTEIO;
		goto error;
	}

	m->wpos += err;
	if (m->wpos == m->wsize)
		m->wpos = m->wsize = 0;

Simon Derr's avatar
Simon Derr committed
495 496 497
end_clear:
	clear_bit(Wworksched, &m->wsched);

498
	if (m->wsize || !list_empty(&m->unsent_req_list)) {
499 500 501
		if (test_and_clear_bit(Wpending, &m->wsched))
			n = POLLOUT;
		else
502
			n = p9_fd_poll(m->client, NULL);
503

Simon Derr's avatar
Simon Derr committed
504 505
		if ((n & POLLOUT) &&
		   !test_and_set_bit(Wworksched, &m->wsched)) {
506
			p9_debug(P9_DEBUG_TRANS, "sched write work %p\n", m);
507
			schedule_work(&m->wq);
Simon Derr's avatar
Simon Derr committed
508 509
		}
	}
510 511 512 513 514 515 516 517

	return;

error:
	p9_conn_cancel(m, err);
	clear_bit(Wworksched, &m->wsched);
}

518
static int p9_pollwake(wait_queue_t *wait, unsigned int mode, int sync, void *key)
519
{
520 521 522 523
	struct p9_poll_wait *pwait =
		container_of(wait, struct p9_poll_wait, wait);
	struct p9_conn *m = pwait->conn;
	unsigned long flags;
524

525 526 527 528
	spin_lock_irqsave(&p9_poll_lock, flags);
	if (list_empty(&m->poll_pending_link))
		list_add_tail(&m->poll_pending_link, &p9_poll_pending_list);
	spin_unlock_irqrestore(&p9_poll_lock, flags);
529

530 531
	schedule_work(&p9_poll_work);
	return 1;
532 533 534
}

/**
535 536 537 538
 * p9_pollwait - add poll task to the wait queue
 * @filp: file pointer being polled
 * @wait_address: wait_q to block on
 * @p: poll state
539
 *
540
 * called by files poll operation to add v9fs-poll task to files wait queue
541
 */
542

543 544
static void
p9_pollwait(struct file *filp, wait_queue_head_t *wait_address, poll_table *p)
545
{
546 547 548
	struct p9_conn *m = container_of(p, struct p9_conn, pt);
	struct p9_poll_wait *pwait = NULL;
	int i;
549

550 551 552 553
	for (i = 0; i < ARRAY_SIZE(m->poll_wait); i++) {
		if (m->poll_wait[i].wait_addr == NULL) {
			pwait = &m->poll_wait[i];
			break;
554 555 556
		}
	}

557
	if (!pwait) {
558
		p9_debug(P9_DEBUG_ERROR, "not enough wait_address slots\n");
559 560 561
		return;
	}

562 563 564 565 566
	pwait->conn = m;
	pwait->wait_addr = wait_address;
	init_waitqueue_func_entry(&pwait->wait, p9_pollwake);
	add_wait_queue(wait_address, &pwait->wait);
}
567

568 569 570 571 572 573
/**
 * p9_conn_create - allocate and initialize the per-session mux data
 * @client: client instance
 *
 * Note: Creates the polling task if this is the first session.
 */
574

575 576
static struct p9_conn *p9_conn_create(struct p9_client *client)
{
577
	int n;
578
	struct p9_conn *m;
579

580
	p9_debug(P9_DEBUG_TRANS, "client %p msize %d\n", client, client->msize);
581 582 583
	m = kzalloc(sizeof(struct p9_conn), GFP_KERNEL);
	if (!m)
		return ERR_PTR(-ENOMEM);
584

585 586
	INIT_LIST_HEAD(&m->mux_list);
	m->client = client;
587

588 589 590 591 592 593
	INIT_LIST_HEAD(&m->req_list);
	INIT_LIST_HEAD(&m->unsent_req_list);
	INIT_WORK(&m->rq, p9_read_work);
	INIT_WORK(&m->wq, p9_write_work);
	INIT_LIST_HEAD(&m->poll_pending_link);
	init_poll_funcptr(&m->pt, p9_pollwait);
594

595 596
	n = p9_fd_poll(client, &m->pt);
	if (n & POLLIN) {
597
		p9_debug(P9_DEBUG_TRANS, "mux %p can read\n", m);
598 599
		set_bit(Rpending, &m->wsched);
	}
600

601
	if (n & POLLOUT) {
602
		p9_debug(P9_DEBUG_TRANS, "mux %p can write\n", m);
603 604 605 606 607
		set_bit(Wpending, &m->wsched);
	}

	return m;
}
608

609 610 611 612 613 614 615 616 617 618 619 620 621 622 623
/**
 * p9_poll_mux - polls a mux and schedules read or write works if necessary
 * @m: connection to poll
 *
 */

static void p9_poll_mux(struct p9_conn *m)
{
	int n;

	if (m->err < 0)
		return;

	n = p9_fd_poll(m->client, NULL);
	if (n < 0 || n & (POLLERR | POLLHUP | POLLNVAL)) {
624
		p9_debug(P9_DEBUG_TRANS, "error mux %p err %d\n", m, n);
625 626 627 628 629 630 631
		if (n >= 0)
			n = -ECONNRESET;
		p9_conn_cancel(m, n);
	}

	if (n & POLLIN) {
		set_bit(Rpending, &m->wsched);
632
		p9_debug(P9_DEBUG_TRANS, "mux %p can read\n", m);
633
		if (!test_and_set_bit(Rworksched, &m->wsched)) {
634
			p9_debug(P9_DEBUG_TRANS, "sched read work %p\n", m);
635
			schedule_work(&m->rq);
636 637
		}
	}
638

639 640
	if (n & POLLOUT) {
		set_bit(Wpending, &m->wsched);
641
		p9_debug(P9_DEBUG_TRANS, "mux %p can write\n", m);
642 643
		if ((m->wsize || !list_empty(&m->unsent_req_list)) &&
		    !test_and_set_bit(Wworksched, &m->wsched)) {
644
			p9_debug(P9_DEBUG_TRANS, "sched write work %p\n", m);
645
			schedule_work(&m->wq);
646 647
		}
	}
648 649 650
}

/**
651
 * p9_fd_request - send 9P request
652 653
 * The function can sleep until the request is scheduled for sending.
 * The function can be interrupted. Return from the function is not
654
 * a guarantee that the request is sent successfully.
655
 *
656 657
 * @client: client instance
 * @req: request to be sent
658
 *
659
 */
660

661
static int p9_fd_request(struct p9_client *client, struct p9_req_t *req)
662 663
{
	int n;
664 665
	struct p9_trans_fd *ts = client->trans;
	struct p9_conn *m = ts->conn;
666

667 668
	p9_debug(P9_DEBUG_TRANS, "mux %p task %p tcall %p id %d\n",
		 m, current, req->tc, req->tc->id);
669
	if (m->err < 0)
670
		return m->err;
671

672
	spin_lock(&client->lock);
673
	req->status = REQ_STATUS_UNSENT;
674
	list_add_tail(&req->req_list, &m->unsent_req_list);
675
	spin_unlock(&client->lock);
676 677 678 679

	if (test_and_clear_bit(Wpending, &m->wsched))
		n = POLLOUT;
	else
680
		n = p9_fd_poll(m->client, NULL);
681 682

	if (n & POLLOUT && !test_and_set_bit(Wworksched, &m->wsched))
683
		schedule_work(&m->wq);
684

685
	return 0;
686 687
}

688
static int p9_fd_cancel(struct p9_client *client, struct p9_req_t *req)
689
{
690
	int ret = 1;
691

692
	p9_debug(P9_DEBUG_TRANS, "client %p req %p\n", client, req);
693

694 695 696
	spin_lock(&client->lock);

	if (req->status == REQ_STATUS_UNSENT) {
697
		list_del(&req->req_list);
698
		req->status = REQ_STATUS_FLSHD;
699
		ret = 0;
700 701
	} else if (req->status == REQ_STATUS_SENT)
		req->status = REQ_STATUS_FLSH;
702

703 704 705
	spin_unlock(&client->lock);

	return ret;
706 707
}

708
/**
Abhishek Kulkarni's avatar
Abhishek Kulkarni committed
709 710 711
 * parse_opts - parse mount options into p9_fd_opts structure
 * @params: options string passed from mount
 * @opts: fd transport-specific structure to parse options into
712
 *
713
 * Returns 0 upon success, -ERRNO upon failure
714
 */
715

716
static int parse_opts(char *params, struct p9_fd_opts *opts)
717
{
718 719 720
	char *p;
	substring_t args[MAX_OPT_ARGS];
	int option;
Eric Van Hensbergen's avatar
Eric Van Hensbergen committed
721
	char *options, *tmp_options;
722

723 724 725
	opts->port = P9_PORT;
	opts->rfd = ~0;
	opts->wfd = ~0;
726

727 728 729
	if (!params)
		return 0;

Eric Van Hensbergen's avatar
Eric Van Hensbergen committed
730 731
	tmp_options = kstrdup(params, GFP_KERNEL);
	if (!tmp_options) {
732 733
		p9_debug(P9_DEBUG_ERROR,
			 "failed to allocate copy of option string\n");
734 735
		return -ENOMEM;
	}
Eric Van Hensbergen's avatar
Eric Van Hensbergen committed
736
	options = tmp_options;
737

738 739
	while ((p = strsep(&options, ",")) != NULL) {
		int token;
740
		int r;
741 742 743
		if (!*p)
			continue;
		token = match_token(p, tokens, args);
744 745 746
		if (token != Opt_err) {
			r = match_int(&args[0], &option);
			if (r < 0) {
747 748
				p9_debug(P9_DEBUG_ERROR,
					 "integer field, but no integer?\n");
749 750
				continue;
			}
751 752 753 754 755 756 757 758 759 760 761 762 763 764
		}
		switch (token) {
		case Opt_port:
			opts->port = option;
			break;
		case Opt_rfdno:
			opts->rfd = option;
			break;
		case Opt_wfdno:
			opts->wfd = option;
			break;
		default:
			continue;
		}
765
	}
Eric Van Hensbergen's avatar
Eric Van Hensbergen committed
766 767

	kfree(tmp_options);
768
	return 0;
769 770
}

771
static int p9_fd_open(struct p9_client *client, int rfd, int wfd)
772
{
773 774 775 776
	struct p9_trans_fd *ts = kmalloc(sizeof(struct p9_trans_fd),
					   GFP_KERNEL);
	if (!ts)
		return -ENOMEM;
777

778 779 780 781 782 783 784 785 786
	ts->rd = fget(rfd);
	ts->wr = fget(wfd);
	if (!ts->rd || !ts->wr) {
		if (ts->rd)
			fput(ts->rd);
		if (ts->wr)
			fput(ts->wr);
		kfree(ts);
		return -EIO;
787 788
	}

789 790
	client->trans = ts;
	client->status = Connected;
791

792
	return 0;
793 794
}

795
static int p9_socket_open(struct p9_client *client, struct socket *csocket)
796
{
Al Viro's avatar
Al Viro committed
797
	struct p9_trans_fd *p;
798 799
	struct file *file;
	int ret;
Al Viro's avatar
Al Viro committed
800 801 802 803

	p = kmalloc(sizeof(struct p9_trans_fd), GFP_KERNEL);
	if (!p)
		return -ENOMEM;
804 805

	csocket->sk->sk_allocation = GFP_NOIO;
806
	file = sock_alloc_file(csocket, 0, NULL);
807
	if (IS_ERR(file)) {
808 809
		pr_err("%s (%d): failed to map fd\n",
		       __func__, task_pid_nr(current));
Al Viro's avatar
Al Viro committed
810 811
		sock_release(csocket);
		kfree(p);
812
		return PTR_ERR(file);
813 814
	}

815 816
	get_file(file);
	p->wr = p->rd = file;
Al Viro's avatar
Al Viro committed
817 818 819 820 821 822 823 824 825 826 827
	client->trans = p;
	client->status = Connected;

	p->rd->f_flags |= O_NONBLOCK;

	p->conn = p9_conn_create(client);
	if (IS_ERR(p->conn)) {
		ret = PTR_ERR(p->conn);
		p->conn = NULL;
		kfree(p);
		sockfd_put(csocket);
828 829 830 831 832 833 834
		sockfd_put(csocket);
		return ret;
	}
	return 0;
}

/**
835 836
 * p9_mux_destroy - cancels all pending requests and frees mux resources
 * @m: mux to destroy
837 838
 *
 */
839

840
static void p9_conn_destroy(struct p9_conn *m)
841
{
842 843
	p9_debug(P9_DEBUG_TRANS, "mux %p prev %p next %p\n",
		 m, m->mux_list.prev, m->mux_list.next);
844

845 846 847
	p9_mux_poll_stop(m);
	cancel_work_sync(&m->rq);
	cancel_work_sync(&m->wq);
848

849
	p9_conn_cancel(m, -ECONNRESET);
850

851 852
	m->client = NULL;
	kfree(m);
853 854 855
}

/**
856 857
 * p9_fd_close - shutdown file descriptor transport
 * @client: client instance
858 859
 *
 */
860

861
static void p9_fd_close(struct p9_client *client)
862 863 864
{
	struct p9_trans_fd *ts;

865
	if (!client)
866 867
		return;

868
	ts = client->trans;
869 870 871
	if (!ts)
		return;

872 873
	client->status = Disconnected;

874 875
	p9_conn_destroy(ts->conn);

876 877 878 879
	if (ts->rd)
		fput(ts->rd);
	if (ts->wr)
		fput(ts->wr);
880

881 882 883
	kfree(ts);
}

884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900
/*
 * stolen from NFS - maybe should be made a generic function?
 */
static inline int valid_ipaddr4(const char *buf)
{
	int rc, count, in[4];

	rc = sscanf(buf, "%d.%d.%d.%d", &in[0], &in[1], &in[2], &in[3]);
	if (rc != 4)
		return -EINVAL;
	for (count = 0; count < 4; count++) {
		if (in[count] > 255)
			return -EINVAL;
	}
	return 0;
}

901 902
static int
p9_fd_create_tcp(struct p9_client *client, const char *addr, char *args)
903 904 905 906 907 908
{
	int err;
	struct socket *csocket;
	struct sockaddr_in sin_server;
	struct p9_fd_opts opts;

909 910
	err = parse_opts(args, &opts);
	if (err < 0)
911
		return err;
912

913
	if (valid_ipaddr4(addr) < 0)
914
		return -EINVAL;
915

916 917 918 919 920
	csocket = NULL;

	sin_server.sin_family = AF_INET;
	sin_server.sin_addr.s_addr = in_aton(addr);
	sin_server.sin_port = htons(opts.port);
921 922
	err = __sock_create(read_pnet(&current->nsproxy->net_ns), PF_INET,
			    SOCK_STREAM, IPPROTO_TCP, &csocket, 1);
Al Viro's avatar
Al Viro committed
923
	if (err) {
924 925
		pr_err("%s (%d): problem creating socket\n",
		       __func__, task_pid_nr(current));
Al Viro's avatar
Al Viro committed
926
		return err;
927 928 929 930 931 932
	}

	err = csocket->ops->connect(csocket,
				    (struct sockaddr *)&sin_server,
				    sizeof(struct sockaddr_in), 0);
	if (err < 0) {
933 934
		pr_err("%s (%d): problem connecting socket to %s\n",
		       __func__, task_pid_nr(current), addr);
935
		sock_release(csocket);
Al Viro's avatar
Al Viro committed
936 937
		return err;
	}
938

Al Viro's avatar
Al Viro committed
939
	return p9_socket_open(client, csocket);
940 941
}

942 943
static int
p9_fd_create_unix(struct p9_client *client, const char *addr, char *args)
944 945 946 947 948 949 950
{
	int err;
	struct socket *csocket;
	struct sockaddr_un sun_server;

	csocket = NULL;

951
	if (strlen(addr) >= UNIX_PATH_MAX) {
952 953
		pr_err("%s (%d): address too long: %s\n",
		       __func__, task_pid_nr(current), addr);
Al Viro's avatar
Al Viro committed
954
		return -ENAMETOOLONG;
955 956 957 958
	}

	sun_server.sun_family = PF_UNIX;
	strcpy(sun_server.sun_path, addr);
959 960
	err = __sock_create(read_pnet(&current->nsproxy->net_ns), PF_UNIX,
			    SOCK_STREAM, 0, &csocket, 1);
Al Viro's avatar
Al Viro committed
961
	if (err < 0) {
962 963 964
		pr_err("%s (%d): problem creating socket\n",
		       __func__, task_pid_nr(current));

Al Viro's avatar
Al Viro committed
965 966
		return err;
	}
967 968 969
	err = csocket->ops->connect(csocket, (struct sockaddr *)&sun_server,
			sizeof(struct sockaddr_un) - 1, 0);
	if (err < 0) {
970 971
		pr_err("%s (%d): problem connecting socket: %s: %d\n",
		       __func__, task_pid_nr(current), addr, err);
972
		sock_release(csocket);
Al Viro's avatar
Al Viro committed
973 974
		return err;
	}
975

Al Viro's avatar
Al Viro committed
976
	return p9_socket_open(client, csocket);
977 978
}

979 980
static int
p9_fd_create(struct p9_client *client, const char *addr, char *args)
981 982 983
{
	int err;
	struct p9_fd_opts opts;
Al Viro's avatar
Al Viro committed
984
	struct p9_trans_fd *p;
985 986 987 988

	parse_opts(args, &opts);

	if (opts.rfd == ~0 || opts.wfd == ~0) {
989
		pr_err("Insufficient options for proto=fd\n");
990
		return -ENOPROTOOPT;
991 992
	}

993
	err = p9_fd_open(client, opts.rfd, opts.wfd);
994
	if (err < 0)
Al Viro's avatar
Al Viro committed
995
		return err;
996

997 998
	p = (struct p9_trans_fd *) client->trans;
	p->conn = p9_conn_create(client);
999 1000 1001
	if (IS_ERR(p->conn)) {
		err = PTR_ERR(p->conn);
		p->conn = NULL;
Al Viro's avatar
Al Viro committed
1002 1003 1004
		fput(p->rd);
		fput(p->wr);
		return err;
1005 1006
	}

1007
	return 0;
1008 1009 1010 1011 1012 1013
}

static struct p9_trans_module p9_tcp_trans = {
	.name = "tcp",
	.maxsize = MAX_SOCK_BUF,
	.def = 1,
1014 1015
	.create = p9_fd_create_tcp,
	.close = p9_fd_close,
1016 1017
	.request = p9_fd_request,
	.cancel = p9_fd_cancel,
1018
	.owner = THIS_MODULE,