drbd_receiver.c 143.5 KB
Newer Older
P
Philipp Reisner 已提交
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50
/*
   drbd_receiver.c

   This file is part of DRBD by Philipp Reisner and Lars Ellenberg.

   Copyright (C) 2001-2008, LINBIT Information Technologies GmbH.
   Copyright (C) 1999-2008, Philipp Reisner <philipp.reisner@linbit.com>.
   Copyright (C) 2002-2008, Lars Ellenberg <lars.ellenberg@linbit.com>.

   drbd is free software; you can redistribute it and/or modify
   it under the terms of the GNU General Public License as published by
   the Free Software Foundation; either version 2, or (at your option)
   any later version.

   drbd 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 drbd; see the file COPYING.  If not, write to
   the Free Software Foundation, 675 Mass Ave, Cambridge, MA 02139, USA.
 */


#include <linux/module.h>

#include <asm/uaccess.h>
#include <net/sock.h>

#include <linux/drbd.h>
#include <linux/fs.h>
#include <linux/file.h>
#include <linux/in.h>
#include <linux/mm.h>
#include <linux/memcontrol.h>
#include <linux/mm_inline.h>
#include <linux/slab.h>
#include <linux/pkt_sched.h>
#define __KERNEL_SYSCALLS__
#include <linux/unistd.h>
#include <linux/vmalloc.h>
#include <linux/random.h>
#include <linux/string.h>
#include <linux/scatterlist.h>
#include "drbd_int.h"
#include "drbd_req.h"

#include "drbd_vli.h"

51 52
struct packet_info {
	enum drbd_packet cmd;
53 54
	unsigned int size;
	unsigned int vnr;
55
	void *data;
56 57
};

P
Philipp Reisner 已提交
58 59 60 61 62 63
enum finish_epoch {
	FE_STILL_LIVE,
	FE_DESTROYED,
	FE_RECYCLED,
};

64
static int drbd_do_features(struct drbd_tconn *tconn);
65
static int drbd_do_auth(struct drbd_tconn *tconn);
P
Philipp Reisner 已提交
66
static int drbd_disconnected(struct drbd_conf *mdev);
P
Philipp Reisner 已提交
67

68
static enum finish_epoch drbd_may_finish_epoch(struct drbd_tconn *, struct drbd_epoch *, enum epoch_event);
69
static int e_end_block(struct drbd_work *, int);
P
Philipp Reisner 已提交
70 71 72 73


#define GFP_TRY	(__GFP_HIGHMEM | __GFP_NOWARN)

74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91
/*
 * some helper functions to deal with single linked page lists,
 * page->private being our "next" pointer.
 */

/* If at least n pages are linked at head, get n pages off.
 * Otherwise, don't modify head, and return NULL.
 * Locking is the responsibility of the caller.
 */
static struct page *page_chain_del(struct page **head, int n)
{
	struct page *page;
	struct page *tmp;

	BUG_ON(!n);
	BUG_ON(!head);

	page = *head;
92 93 94 95

	if (!page)
		return NULL;

96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152
	while (page) {
		tmp = page_chain_next(page);
		if (--n == 0)
			break; /* found sufficient pages */
		if (tmp == NULL)
			/* insufficient pages, don't use any of them. */
			return NULL;
		page = tmp;
	}

	/* add end of list marker for the returned list */
	set_page_private(page, 0);
	/* actual return value, and adjustment of head */
	page = *head;
	*head = tmp;
	return page;
}

/* may be used outside of locks to find the tail of a (usually short)
 * "private" page chain, before adding it back to a global chain head
 * with page_chain_add() under a spinlock. */
static struct page *page_chain_tail(struct page *page, int *len)
{
	struct page *tmp;
	int i = 1;
	while ((tmp = page_chain_next(page)))
		++i, page = tmp;
	if (len)
		*len = i;
	return page;
}

static int page_chain_free(struct page *page)
{
	struct page *tmp;
	int i = 0;
	page_chain_for_each_safe(page, tmp) {
		put_page(page);
		++i;
	}
	return i;
}

static void page_chain_add(struct page **head,
		struct page *chain_first, struct page *chain_last)
{
#if 1
	struct page *tmp;
	tmp = page_chain_tail(chain_first, NULL);
	BUG_ON(tmp != chain_last);
#endif

	/* add chain to head */
	set_page_private(chain_last, (unsigned long)*head);
	*head = chain_first;
}

153 154
static struct page *__drbd_alloc_pages(struct drbd_conf *mdev,
				       unsigned int number)
P
Philipp Reisner 已提交
155 156
{
	struct page *page = NULL;
157
	struct page *tmp = NULL;
158
	unsigned int i = 0;
P
Philipp Reisner 已提交
159 160 161

	/* Yes, testing drbd_pp_vacant outside the lock is racy.
	 * So what. It saves a spin_lock. */
162
	if (drbd_pp_vacant >= number) {
P
Philipp Reisner 已提交
163
		spin_lock(&drbd_pp_lock);
164 165 166
		page = page_chain_del(&drbd_pp_pool, number);
		if (page)
			drbd_pp_vacant -= number;
P
Philipp Reisner 已提交
167
		spin_unlock(&drbd_pp_lock);
168 169
		if (page)
			return page;
P
Philipp Reisner 已提交
170
	}
171

P
Philipp Reisner 已提交
172 173 174
	/* GFP_TRY, because we must not cause arbitrary write-out: in a DRBD
	 * "criss-cross" setup, that might cause write-out on some other DRBD,
	 * which in turn might block on the other node at this very place.  */
175 176 177 178 179 180 181 182 183 184 185 186
	for (i = 0; i < number; i++) {
		tmp = alloc_page(GFP_TRY);
		if (!tmp)
			break;
		set_page_private(tmp, (unsigned long)page);
		page = tmp;
	}

	if (i == number)
		return page;

	/* Not enough pages immediately available this time.
187
	 * No need to jump around here, drbd_alloc_pages will retry this
188 189 190 191 192 193 194 195 196
	 * function "soon". */
	if (page) {
		tmp = page_chain_tail(page, NULL);
		spin_lock(&drbd_pp_lock);
		page_chain_add(&drbd_pp_pool, page, tmp);
		drbd_pp_vacant += i;
		spin_unlock(&drbd_pp_lock);
	}
	return NULL;
P
Philipp Reisner 已提交
197 198
}

199 200
static void reclaim_finished_net_peer_reqs(struct drbd_conf *mdev,
					   struct list_head *to_be_freed)
P
Philipp Reisner 已提交
201
{
202
	struct drbd_peer_request *peer_req;
P
Philipp Reisner 已提交
203 204 205 206 207 208 209 210
	struct list_head *le, *tle;

	/* The EEs are always appended to the end of the list. Since
	   they are sent in order over the wire, they have to finish
	   in order. As soon as we see the first not finished we can
	   stop to examine the list... */

	list_for_each_safe(le, tle, &mdev->net_ee) {
211
		peer_req = list_entry(le, struct drbd_peer_request, w.list);
212
		if (drbd_peer_req_has_active_page(peer_req))
P
Philipp Reisner 已提交
213 214 215 216 217 218 219 220
			break;
		list_move(le, to_be_freed);
	}
}

static void drbd_kick_lo_and_reclaim_net(struct drbd_conf *mdev)
{
	LIST_HEAD(reclaimed);
221
	struct drbd_peer_request *peer_req, *t;
P
Philipp Reisner 已提交
222

223
	spin_lock_irq(&mdev->tconn->req_lock);
224
	reclaim_finished_net_peer_reqs(mdev, &reclaimed);
225
	spin_unlock_irq(&mdev->tconn->req_lock);
P
Philipp Reisner 已提交
226

227
	list_for_each_entry_safe(peer_req, t, &reclaimed, w.list)
228
		drbd_free_net_peer_req(mdev, peer_req);
P
Philipp Reisner 已提交
229 230 231
}

/**
232
 * drbd_alloc_pages() - Returns @number pages, retries forever (or until signalled)
P
Philipp Reisner 已提交
233
 * @mdev:	DRBD device.
234 235 236 237 238 239
 * @number:	number of pages requested
 * @retry:	whether to retry, if not enough pages are available right now
 *
 * Tries to allocate number pages, first from our own page pool, then from
 * the kernel, unless this allocation would exceed the max_buffers setting.
 * Possibly retry until DRBD frees sufficient pages somewhere else.
P
Philipp Reisner 已提交
240
 *
241
 * Returns a page chain linked via page->private.
P
Philipp Reisner 已提交
242
 */
243 244
struct page *drbd_alloc_pages(struct drbd_conf *mdev, unsigned int number,
			      bool retry)
P
Philipp Reisner 已提交
245 246
{
	struct page *page = NULL;
247
	struct net_conf *nc;
P
Philipp Reisner 已提交
248
	DEFINE_WAIT(wait);
249
	int mxb;
P
Philipp Reisner 已提交
250

251 252
	/* Yes, we may run up to @number over max_buffers. If we
	 * follow it strictly, the admin will get it wrong anyways. */
253 254 255 256 257 258
	rcu_read_lock();
	nc = rcu_dereference(mdev->tconn->net_conf);
	mxb = nc ? nc->max_buffers : 1000000;
	rcu_read_unlock();

	if (atomic_read(&mdev->pp_in_use) < mxb)
259
		page = __drbd_alloc_pages(mdev, number);
P
Philipp Reisner 已提交
260

261
	while (page == NULL) {
P
Philipp Reisner 已提交
262 263 264 265
		prepare_to_wait(&drbd_pp_wait, &wait, TASK_INTERRUPTIBLE);

		drbd_kick_lo_and_reclaim_net(mdev);

266
		if (atomic_read(&mdev->pp_in_use) < mxb) {
267
			page = __drbd_alloc_pages(mdev, number);
P
Philipp Reisner 已提交
268 269 270 271 272 273 274 275
			if (page)
				break;
		}

		if (!retry)
			break;

		if (signal_pending(current)) {
276
			dev_warn(DEV, "drbd_alloc_pages interrupted!\n");
P
Philipp Reisner 已提交
277 278 279 280 281 282 283
			break;
		}

		schedule();
	}
	finish_wait(&drbd_pp_wait, &wait);

284 285
	if (page)
		atomic_add(number, &mdev->pp_in_use);
P
Philipp Reisner 已提交
286 287 288
	return page;
}

289
/* Must not be used from irq, as that may deadlock: see drbd_alloc_pages.
290
 * Is also used from inside an other spin_lock_irq(&mdev->tconn->req_lock);
291 292
 * Either links the page chain back to the global pool,
 * or returns all pages to the system. */
293
static void drbd_free_pages(struct drbd_conf *mdev, struct page *page, int is_net)
P
Philipp Reisner 已提交
294
{
295
	atomic_t *a = is_net ? &mdev->pp_in_use_by_net : &mdev->pp_in_use;
P
Philipp Reisner 已提交
296
	int i;
297

298
	if (drbd_pp_vacant > (DRBD_MAX_BIO_SIZE/PAGE_SIZE) * minor_count)
299 300 301 302 303 304 305 306
		i = page_chain_free(page);
	else {
		struct page *tmp;
		tmp = page_chain_tail(page, &i);
		spin_lock(&drbd_pp_lock);
		page_chain_add(&drbd_pp_pool, page, tmp);
		drbd_pp_vacant += i;
		spin_unlock(&drbd_pp_lock);
P
Philipp Reisner 已提交
307
	}
308
	i = atomic_sub_return(i, a);
309
	if (i < 0)
310 311
		dev_warn(DEV, "ASSERTION FAILED: %s: %d < 0\n",
			is_net ? "pp_in_use_by_net" : "pp_in_use", i);
P
Philipp Reisner 已提交
312 313 314 315 316 317 318 319
	wake_up(&drbd_pp_wait);
}

/*
You need to hold the req_lock:
 _drbd_wait_ee_list_empty()

You must not have the req_lock:
320
 drbd_free_peer_req()
321
 drbd_alloc_peer_req()
322
 drbd_free_peer_reqs()
P
Philipp Reisner 已提交
323
 drbd_ee_fix_bhs()
324
 drbd_finish_peer_reqs()
P
Philipp Reisner 已提交
325 326 327 328
 drbd_clear_done_ee()
 drbd_wait_ee_list_empty()
*/

329
struct drbd_peer_request *
330 331
drbd_alloc_peer_req(struct drbd_conf *mdev, u64 id, sector_t sector,
		    unsigned int data_size, gfp_t gfp_mask) __must_hold(local)
P
Philipp Reisner 已提交
332
{
333
	struct drbd_peer_request *peer_req;
P
Philipp Reisner 已提交
334
	struct page *page;
335
	unsigned nr_pages = (data_size + PAGE_SIZE -1) >> PAGE_SHIFT;
P
Philipp Reisner 已提交
336

337
	if (drbd_insert_fault(mdev, DRBD_FAULT_AL_EE))
P
Philipp Reisner 已提交
338 339
		return NULL;

340 341
	peer_req = mempool_alloc(drbd_ee_mempool, gfp_mask & ~__GFP_HIGHMEM);
	if (!peer_req) {
P
Philipp Reisner 已提交
342
		if (!(gfp_mask & __GFP_NOWARN))
343
			dev_err(DEV, "%s: allocation failed\n", __func__);
P
Philipp Reisner 已提交
344 345 346
		return NULL;
	}

347
	page = drbd_alloc_pages(mdev, nr_pages, (gfp_mask & __GFP_WAIT));
348 349
	if (!page)
		goto fail;
P
Philipp Reisner 已提交
350

351 352 353 354 355 356 357
	drbd_clear_interval(&peer_req->i);
	peer_req->i.size = data_size;
	peer_req->i.sector = sector;
	peer_req->i.local = false;
	peer_req->i.waiting = false;

	peer_req->epoch = NULL;
358
	peer_req->w.mdev = mdev;
359 360 361
	peer_req->pages = page;
	atomic_set(&peer_req->pending_bios, 0);
	peer_req->flags = 0;
362 363 364 365
	/*
	 * The block_id is opaque to the receiver.  It is not endianness
	 * converted, and sent back to the sender unchanged.
	 */
366
	peer_req->block_id = id;
P
Philipp Reisner 已提交
367

368
	return peer_req;
P
Philipp Reisner 已提交
369

370
 fail:
371
	mempool_free(peer_req, drbd_ee_mempool);
P
Philipp Reisner 已提交
372 373 374
	return NULL;
}

375
void __drbd_free_peer_req(struct drbd_conf *mdev, struct drbd_peer_request *peer_req,
376
		       int is_net)
P
Philipp Reisner 已提交
377
{
378 379
	if (peer_req->flags & EE_HAS_DIGEST)
		kfree(peer_req->digest);
380
	drbd_free_pages(mdev, peer_req->pages, is_net);
381 382 383
	D_ASSERT(atomic_read(&peer_req->pending_bios) == 0);
	D_ASSERT(drbd_interval_empty(&peer_req->i));
	mempool_free(peer_req, drbd_ee_mempool);
P
Philipp Reisner 已提交
384 385
}

386
int drbd_free_peer_reqs(struct drbd_conf *mdev, struct list_head *list)
P
Philipp Reisner 已提交
387 388
{
	LIST_HEAD(work_list);
389
	struct drbd_peer_request *peer_req, *t;
P
Philipp Reisner 已提交
390
	int count = 0;
391
	int is_net = list == &mdev->net_ee;
P
Philipp Reisner 已提交
392

393
	spin_lock_irq(&mdev->tconn->req_lock);
P
Philipp Reisner 已提交
394
	list_splice_init(list, &work_list);
395
	spin_unlock_irq(&mdev->tconn->req_lock);
P
Philipp Reisner 已提交
396

397
	list_for_each_entry_safe(peer_req, t, &work_list, w.list) {
398
		__drbd_free_peer_req(mdev, peer_req, is_net);
P
Philipp Reisner 已提交
399 400 401 402 403
		count++;
	}
	return count;
}

404 405
/*
 * See also comments in _req_mod(,BARRIER_ACKED) and receive_Barrier.
P
Philipp Reisner 已提交
406
 */
407
static int drbd_finish_peer_reqs(struct drbd_conf *mdev)
P
Philipp Reisner 已提交
408 409 410
{
	LIST_HEAD(work_list);
	LIST_HEAD(reclaimed);
411
	struct drbd_peer_request *peer_req, *t;
412
	int err = 0;
P
Philipp Reisner 已提交
413

414
	spin_lock_irq(&mdev->tconn->req_lock);
415
	reclaim_finished_net_peer_reqs(mdev, &reclaimed);
P
Philipp Reisner 已提交
416
	list_splice_init(&mdev->done_ee, &work_list);
417
	spin_unlock_irq(&mdev->tconn->req_lock);
P
Philipp Reisner 已提交
418

419
	list_for_each_entry_safe(peer_req, t, &reclaimed, w.list)
420
		drbd_free_net_peer_req(mdev, peer_req);
P
Philipp Reisner 已提交
421 422

	/* possible callbacks here:
423
	 * e_end_block, and e_end_resync_block, e_send_discard_write.
P
Philipp Reisner 已提交
424 425
	 * all ignore the last argument.
	 */
426
	list_for_each_entry_safe(peer_req, t, &work_list, w.list) {
427 428
		int err2;

P
Philipp Reisner 已提交
429
		/* list_del not necessary, next/prev members not touched */
430 431 432
		err2 = peer_req->w.cb(&peer_req->w, !!err);
		if (!err)
			err = err2;
433
		drbd_free_peer_req(mdev, peer_req);
P
Philipp Reisner 已提交
434 435 436
	}
	wake_up(&mdev->ee_wait);

437
	return err;
P
Philipp Reisner 已提交
438 439
}

440 441
static void _drbd_wait_ee_list_empty(struct drbd_conf *mdev,
				     struct list_head *head)
P
Philipp Reisner 已提交
442 443 444 445 446 447 448
{
	DEFINE_WAIT(wait);

	/* avoids spin_lock/unlock
	 * and calling prepare_to_wait in the fast path */
	while (!list_empty(head)) {
		prepare_to_wait(&mdev->ee_wait, &wait, TASK_UNINTERRUPTIBLE);
449
		spin_unlock_irq(&mdev->tconn->req_lock);
J
Jens Axboe 已提交
450
		io_schedule();
P
Philipp Reisner 已提交
451
		finish_wait(&mdev->ee_wait, &wait);
452
		spin_lock_irq(&mdev->tconn->req_lock);
P
Philipp Reisner 已提交
453 454 455
	}
}

456 457
static void drbd_wait_ee_list_empty(struct drbd_conf *mdev,
				    struct list_head *head)
P
Philipp Reisner 已提交
458
{
459
	spin_lock_irq(&mdev->tconn->req_lock);
P
Philipp Reisner 已提交
460
	_drbd_wait_ee_list_empty(mdev, head);
461
	spin_unlock_irq(&mdev->tconn->req_lock);
P
Philipp Reisner 已提交
462 463
}

464
static int drbd_recv_short(struct socket *sock, void *buf, size_t size, int flags)
P
Philipp Reisner 已提交
465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485
{
	mm_segment_t oldfs;
	struct kvec iov = {
		.iov_base = buf,
		.iov_len = size,
	};
	struct msghdr msg = {
		.msg_iovlen = 1,
		.msg_iov = (struct iovec *)&iov,
		.msg_flags = (flags ? flags : MSG_WAITALL | MSG_NOSIGNAL)
	};
	int rv;

	oldfs = get_fs();
	set_fs(KERNEL_DS);
	rv = sock_recvmsg(sock, &msg, size, msg.msg_flags);
	set_fs(oldfs);

	return rv;
}

486
static int drbd_recv(struct drbd_tconn *tconn, void *buf, size_t size)
P
Philipp Reisner 已提交
487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503
{
	mm_segment_t oldfs;
	struct kvec iov = {
		.iov_base = buf,
		.iov_len = size,
	};
	struct msghdr msg = {
		.msg_iovlen = 1,
		.msg_iov = (struct iovec *)&iov,
		.msg_flags = MSG_WAITALL | MSG_NOSIGNAL
	};
	int rv;

	oldfs = get_fs();
	set_fs(KERNEL_DS);

	for (;;) {
504
		rv = sock_recvmsg(tconn->data.socket, &msg, size, msg.msg_flags);
P
Philipp Reisner 已提交
505 506 507 508 509 510 511 512 513 514
		if (rv == size)
			break;

		/* Note:
		 * ECONNRESET	other side closed the connection
		 * ERESTARTSYS	(on  sock) we got a signal
		 */

		if (rv < 0) {
			if (rv == -ECONNRESET)
515
				conn_info(tconn, "sock was reset by peer\n");
P
Philipp Reisner 已提交
516
			else if (rv != -ERESTARTSYS)
517
				conn_err(tconn, "sock_recvmsg returned %d\n", rv);
P
Philipp Reisner 已提交
518 519
			break;
		} else if (rv == 0) {
520
			conn_info(tconn, "sock was shut down by peer\n");
P
Philipp Reisner 已提交
521 522 523 524 525 526 527 528 529 530 531 532 533
			break;
		} else	{
			/* signal came in, or peer/link went down,
			 * after we read a partial message
			 */
			/* D_ASSERT(signal_pending(current)); */
			break;
		}
	};

	set_fs(oldfs);

	if (rv != size)
534
		conn_request_state(tconn, NS(conn, C_BROKEN_PIPE), CS_HARD);
P
Philipp Reisner 已提交
535 536 537 538

	return rv;
}

539 540 541 542 543 544 545 546 547 548 549 550 551
static int drbd_recv_all(struct drbd_tconn *tconn, void *buf, size_t size)
{
	int err;

	err = drbd_recv(tconn, buf, size);
	if (err != size) {
		if (err >= 0)
			err = -EIO;
	} else
		err = 0;
	return err;
}

552 553 554 555 556 557 558 559 560 561
static int drbd_recv_all_warn(struct drbd_tconn *tconn, void *buf, size_t size)
{
	int err;

	err = drbd_recv_all(tconn, buf, size);
	if (err && !signal_pending(current))
		conn_warn(tconn, "short read (expected size %d)\n", (int)size);
	return err;
}

562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580
/* quoting tcp(7):
 *   On individual connections, the socket buffer size must be set prior to the
 *   listen(2) or connect(2) calls in order to have it take effect.
 * This is our wrapper to do so.
 */
static void drbd_setbufsize(struct socket *sock, unsigned int snd,
		unsigned int rcv)
{
	/* open coded SO_SNDBUF, SO_RCVBUF */
	if (snd) {
		sock->sk->sk_sndbuf = snd;
		sock->sk->sk_userlocks |= SOCK_SNDBUF_LOCK;
	}
	if (rcv) {
		sock->sk->sk_rcvbuf = rcv;
		sock->sk->sk_userlocks |= SOCK_RCVBUF_LOCK;
	}
}

581
static struct socket *drbd_try_connect(struct drbd_tconn *tconn)
P
Philipp Reisner 已提交
582 583 584 585
{
	const char *what;
	struct socket *sock;
	struct sockaddr_in6 src_in6;
586 587 588
	struct sockaddr_in6 peer_in6;
	struct net_conf *nc;
	int err, peer_addr_len, my_addr_len;
589
	int sndbuf_size, rcvbuf_size, connect_int;
P
Philipp Reisner 已提交
590 591
	int disconnect_on_error = 1;

592 593 594 595
	rcu_read_lock();
	nc = rcu_dereference(tconn->net_conf);
	if (!nc) {
		rcu_read_unlock();
P
Philipp Reisner 已提交
596
		return NULL;
597 598 599
	}
	sndbuf_size = nc->sndbuf_size;
	rcvbuf_size = nc->rcvbuf_size;
600
	connect_int = nc->connect_int;
601
	rcu_read_unlock();
602

603 604
	my_addr_len = min_t(int, tconn->my_addr_len, sizeof(src_in6));
	memcpy(&src_in6, &tconn->my_addr, my_addr_len);
605

606
	if (((struct sockaddr *)&tconn->my_addr)->sa_family == AF_INET6)
607 608 609 610
		src_in6.sin6_port = 0;
	else
		((struct sockaddr_in *)&src_in6)->sin_port = 0; /* AF_INET & AF_SCI */

611 612
	peer_addr_len = min_t(int, tconn->peer_addr_len, sizeof(src_in6));
	memcpy(&peer_in6, &tconn->peer_addr, peer_addr_len);
P
Philipp Reisner 已提交
613 614

	what = "sock_create_kern";
615 616
	err = sock_create_kern(((struct sockaddr *)&src_in6)->sa_family,
			       SOCK_STREAM, IPPROTO_TCP, &sock);
P
Philipp Reisner 已提交
617 618 619 620 621 622
	if (err < 0) {
		sock = NULL;
		goto out;
	}

	sock->sk->sk_rcvtimeo =
623
	sock->sk->sk_sndtimeo = connect_int * HZ;
624
	drbd_setbufsize(sock, sndbuf_size, rcvbuf_size);
P
Philipp Reisner 已提交
625 626 627 628 629 630 631 632 633

       /* explicitly bind to the configured IP as source IP
	*  for the outgoing connections.
	*  This is needed for multihomed hosts and to be
	*  able to use lo: interfaces for drbd.
	* Make sure to use 0 as port number, so linux selects
	*  a free one dynamically.
	*/
	what = "bind before connect";
634
	err = sock->ops->bind(sock, (struct sockaddr *) &src_in6, my_addr_len);
P
Philipp Reisner 已提交
635 636 637 638 639 640 641
	if (err < 0)
		goto out;

	/* connect may fail, peer not yet available.
	 * stay C_WF_CONNECTION, don't go Disconnecting! */
	disconnect_on_error = 0;
	what = "connect";
642
	err = sock->ops->connect(sock, (struct sockaddr *) &peer_in6, peer_addr_len, 0);
P
Philipp Reisner 已提交
643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659

out:
	if (err < 0) {
		if (sock) {
			sock_release(sock);
			sock = NULL;
		}
		switch (-err) {
			/* timeout, busy, signal pending */
		case ETIMEDOUT: case EAGAIN: case EINPROGRESS:
		case EINTR: case ERESTARTSYS:
			/* peer not (yet) available, network problem */
		case ECONNREFUSED: case ENETUNREACH:
		case EHOSTDOWN:    case EHOSTUNREACH:
			disconnect_on_error = 0;
			break;
		default:
660
			conn_err(tconn, "%s failed, err = %d\n", what, err);
P
Philipp Reisner 已提交
661 662
		}
		if (disconnect_on_error)
663
			conn_request_state(tconn, NS(conn, C_DISCONNECTING), CS_HARD);
P
Philipp Reisner 已提交
664
	}
665

P
Philipp Reisner 已提交
666 667 668
	return sock;
}

669
static struct socket *prepare_listen_socket(struct drbd_tconn *tconn)
P
Philipp Reisner 已提交
670
{
671
	int err, sndbuf_size, rcvbuf_size, my_addr_len;
672
	struct sockaddr_in6 my_addr;
673
	struct socket *s_listen;
674
	struct net_conf *nc;
P
Philipp Reisner 已提交
675 676
	const char *what;

677 678 679 680
	rcu_read_lock();
	nc = rcu_dereference(tconn->net_conf);
	if (!nc) {
		rcu_read_unlock();
P
Philipp Reisner 已提交
681
		return NULL;
682 683 684 685
	}
	sndbuf_size = nc->sndbuf_size;
	rcvbuf_size = nc->rcvbuf_size;
	rcu_read_unlock();
P
Philipp Reisner 已提交
686

687 688 689
	my_addr_len = min_t(int, tconn->my_addr_len, sizeof(struct sockaddr_in6));
	memcpy(&my_addr, &tconn->my_addr, my_addr_len);

P
Philipp Reisner 已提交
690
	what = "sock_create_kern";
691
	err = sock_create_kern(((struct sockaddr *)&my_addr)->sa_family,
692
			       SOCK_STREAM, IPPROTO_TCP, &s_listen);
P
Philipp Reisner 已提交
693 694 695 696 697
	if (err) {
		s_listen = NULL;
		goto out;
	}

698
	s_listen->sk->sk_reuse = 1; /* SO_REUSEADDR */
699
	drbd_setbufsize(s_listen, sndbuf_size, rcvbuf_size);
P
Philipp Reisner 已提交
700 701

	what = "bind before listen";
702
	err = s_listen->ops->bind(s_listen, (struct sockaddr *)&my_addr, my_addr_len);
P
Philipp Reisner 已提交
703 704 705
	if (err < 0)
		goto out;

706 707 708 709 710
	what = "listen";
	err = s_listen->ops->listen(s_listen, 5);
	if (err < 0)
		goto out;

711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750
	return s_listen;
out:
	if (s_listen)
		sock_release(s_listen);
	if (err < 0) {
		if (err != -EAGAIN && err != -EINTR && err != -ERESTARTSYS) {
			conn_err(tconn, "%s failed, err = %d\n", what, err);
			conn_request_state(tconn, NS(conn, C_DISCONNECTING), CS_HARD);
		}
	}

	return NULL;
}

static struct socket *drbd_wait_for_connect(struct drbd_tconn *tconn)
{
	int timeo, connect_int, err = 0;
	struct socket *s_estab = NULL;
	struct socket *s_listen;
	struct net_conf *nc;

	rcu_read_lock();
	nc = rcu_dereference(tconn->net_conf);
	if (!nc) {
		rcu_read_unlock();
		return NULL;
	}
	connect_int = nc->connect_int;
	rcu_read_unlock();

	timeo = connect_int * HZ;
	timeo += (random32() & 1) ? timeo / 7 : -timeo / 7; /* 28.5% random jitter */

	s_listen = prepare_listen_socket(tconn);
	if (!s_listen)
		goto out;

	s_listen->sk->sk_rcvtimeo = timeo;
	s_listen->sk->sk_sndtimeo = timeo;

751
	err = kernel_accept(s_listen, &s_estab, 0);
P
Philipp Reisner 已提交
752 753 754 755 756 757

out:
	if (s_listen)
		sock_release(s_listen);
	if (err < 0) {
		if (err != -EAGAIN && err != -EINTR && err != -ERESTARTSYS) {
758
			conn_err(tconn, "accept failed, err = %d\n", err);
759
			conn_request_state(tconn, NS(conn, C_DISCONNECTING), CS_HARD);
P
Philipp Reisner 已提交
760 761 762 763 764 765
		}
	}

	return s_estab;
}

766
static int decode_header(struct drbd_tconn *, void *, struct packet_info *);
P
Philipp Reisner 已提交
767

768 769 770 771 772
static int send_first_packet(struct drbd_tconn *tconn, struct drbd_socket *sock,
			     enum drbd_packet cmd)
{
	if (!conn_prepare_command(tconn, sock))
		return -EIO;
773
	return conn_send_command(tconn, sock, cmd, 0, NULL, 0);
P
Philipp Reisner 已提交
774 775
}

776
static int receive_first_packet(struct drbd_tconn *tconn, struct socket *sock)
P
Philipp Reisner 已提交
777
{
778 779 780
	unsigned int header_size = drbd_header_size(tconn);
	struct packet_info pi;
	int err;
P
Philipp Reisner 已提交
781

782 783 784 785 786 787 788 789 790 791
	err = drbd_recv_short(sock, tconn->data.rbuf, header_size, 0);
	if (err != header_size) {
		if (err >= 0)
			err = -EIO;
		return err;
	}
	err = decode_header(tconn, tconn->data.rbuf, &pi);
	if (err)
		return err;
	return pi.cmd;
P
Philipp Reisner 已提交
792 793 794 795 796 797
}

/**
 * drbd_socket_okay() - Free the socket if its connection is not okay
 * @sock:	pointer to the pointer to the socket.
 */
798
static int drbd_socket_okay(struct socket **sock)
P
Philipp Reisner 已提交
799 800 801 802 803
{
	int rr;
	char tb[4];

	if (!*sock)
804
		return false;
P
Philipp Reisner 已提交
805

806
	rr = drbd_recv_short(*sock, tb, 4, MSG_DONTWAIT | MSG_PEEK);
P
Philipp Reisner 已提交
807 808

	if (rr > 0 || rr == -EAGAIN) {
809
		return true;
P
Philipp Reisner 已提交
810 811 812
	} else {
		sock_release(*sock);
		*sock = NULL;
813
		return false;
P
Philipp Reisner 已提交
814 815
	}
}
816 817
/* Gets called if a connection is established, or if a new minor gets created
   in a connection */
P
Philipp Reisner 已提交
818
int drbd_connected(struct drbd_conf *mdev)
819
{
820
	int err;
821 822 823 824

	atomic_set(&mdev->packet_seq, 0);
	mdev->peer_seq = 0;

825 826 827 828
	mdev->state_mutex = mdev->tconn->agreed_pro_version < 100 ?
		&mdev->tconn->cstate_mutex :
		&mdev->own_state_mutex;

829 830 831 832 833 834
	err = drbd_send_sync_param(mdev);
	if (!err)
		err = drbd_send_sizes(mdev, 0, 0);
	if (!err)
		err = drbd_send_uuids(mdev);
	if (!err)
835
		err = drbd_send_current_state(mdev);
836 837
	clear_bit(USE_DEGR_WFC_T, &mdev->flags);
	clear_bit(RESIZE_PENDING, &mdev->flags);
838
	mod_timer(&mdev->request_timer, jiffies + HZ); /* just start it here. */
839
	return err;
840 841
}

P
Philipp Reisner 已提交
842 843 844 845 846 847 848 849
/*
 * return values:
 *   1 yes, we have a valid connection
 *   0 oops, did not work out, please try again
 *  -1 peer talks different language,
 *     no point in trying again, please go standalone.
 *  -2 We do not have a network config...
 */
850
static int conn_connect(struct drbd_tconn *tconn)
P
Philipp Reisner 已提交
851
{
852
	struct drbd_socket sock, msock;
P
Philipp Reisner 已提交
853
	struct drbd_conf *mdev;
854
	struct net_conf *nc;
P
Philipp Reisner 已提交
855
	int vnr, timeout, try, h, ok;
856
	bool discard_my_data;
857
	enum drbd_state_rv rv;
P
Philipp Reisner 已提交
858

859
	if (conn_request_state(tconn, NS(conn, C_WF_CONNECTION), CS_VERBOSE) < SS_SUCCESS)
P
Philipp Reisner 已提交
860 861
		return -2;

862 863 864 865 866 867 868 869 870
	mutex_init(&sock.mutex);
	sock.sbuf = tconn->data.sbuf;
	sock.rbuf = tconn->data.rbuf;
	sock.socket = NULL;
	mutex_init(&msock.mutex);
	msock.sbuf = tconn->meta.sbuf;
	msock.rbuf = tconn->meta.rbuf;
	msock.socket = NULL;

871
	clear_bit(DISCARD_CONCURRENT, &tconn->flags);
872 873 874

	/* Assume that the peer only understands protocol 80 until we know better.  */
	tconn->agreed_pro_version = 80;
P
Philipp Reisner 已提交
875 876

	do {
877 878
		struct socket *s;

P
Philipp Reisner 已提交
879 880
		for (try = 0;;) {
			/* 3 tries, this should take less than a second! */
881
			s = drbd_try_connect(tconn);
P
Philipp Reisner 已提交
882 883 884
			if (s || ++try >= 3)
				break;
			/* give the other side time to call bind() & listen() */
885
			schedule_timeout_interruptible(HZ / 10);
P
Philipp Reisner 已提交
886 887 888
		}

		if (s) {
889 890 891 892 893 894
			if (!sock.socket) {
				sock.socket = s;
				send_first_packet(tconn, &sock, P_INITIAL_DATA);
			} else if (!msock.socket) {
				msock.socket = s;
				send_first_packet(tconn, &msock, P_INITIAL_META);
P
Philipp Reisner 已提交
895
			} else {
896
				conn_err(tconn, "Logic error in conn_connect()\n");
P
Philipp Reisner 已提交
897 898 899 900
				goto out_release_sockets;
			}
		}

901 902 903 904 905 906 907 908
		if (sock.socket && msock.socket) {
			rcu_read_lock();
			nc = rcu_dereference(tconn->net_conf);
			timeout = nc->ping_timeo * HZ / 10;
			rcu_read_unlock();
			schedule_timeout_interruptible(timeout);
			ok = drbd_socket_okay(&sock.socket);
			ok = drbd_socket_okay(&msock.socket) && ok;
P
Philipp Reisner 已提交
909 910 911 912 913
			if (ok)
				break;
		}

retry:
914
		s = drbd_wait_for_connect(tconn);
P
Philipp Reisner 已提交
915
		if (s) {
916
			try = receive_first_packet(tconn, s);
917 918
			drbd_socket_okay(&sock.socket);
			drbd_socket_okay(&msock.socket);
P
Philipp Reisner 已提交
919
			switch (try) {
920
			case P_INITIAL_DATA:
921
				if (sock.socket) {
922
					conn_warn(tconn, "initial packet S crossed\n");
923
					sock_release(sock.socket);
P
Philipp Reisner 已提交
924
				}
925
				sock.socket = s;
P
Philipp Reisner 已提交
926
				break;
927
			case P_INITIAL_META:
928
				if (msock.socket) {
929
					conn_warn(tconn, "initial packet M crossed\n");
930
					sock_release(msock.socket);
P
Philipp Reisner 已提交
931
				}
932
				msock.socket = s;
933
				set_bit(DISCARD_CONCURRENT, &tconn->flags);
P
Philipp Reisner 已提交
934 935
				break;
			default:
936
				conn_warn(tconn, "Error receiving initial packet\n");
P
Philipp Reisner 已提交
937 938 939 940 941 942
				sock_release(s);
				if (random32() & 1)
					goto retry;
			}
		}

943
		if (tconn->cstate <= C_DISCONNECTING)
P
Philipp Reisner 已提交
944 945 946 947
			goto out_release_sockets;
		if (signal_pending(current)) {
			flush_signals(current);
			smp_rmb();
948
			if (get_t_state(&tconn->receiver) == EXITING)
P
Philipp Reisner 已提交
949 950 951
				goto out_release_sockets;
		}

952 953 954
		if (sock.socket && &msock.socket) {
			ok = drbd_socket_okay(&sock.socket);
			ok = drbd_socket_okay(&msock.socket) && ok;
P
Philipp Reisner 已提交
955 956 957 958 959
			if (ok)
				break;
		}
	} while (1);

960 961
	sock.socket->sk->sk_reuse = 1; /* SO_REUSEADDR */
	msock.socket->sk->sk_reuse = 1; /* SO_REUSEADDR */
P
Philipp Reisner 已提交
962

963 964
	sock.socket->sk->sk_allocation = GFP_NOIO;
	msock.socket->sk->sk_allocation = GFP_NOIO;
P
Philipp Reisner 已提交
965

966 967
	sock.socket->sk->sk_priority = TC_PRIO_INTERACTIVE_BULK;
	msock.socket->sk->sk_priority = TC_PRIO_INTERACTIVE;
P
Philipp Reisner 已提交
968 969

	/* NOT YET ...
970 971
	 * sock.socket->sk->sk_sndtimeo = tconn->net_conf->timeout*HZ/10;
	 * sock.socket->sk->sk_rcvtimeo = MAX_SCHEDULE_TIMEOUT;
972
	 * first set it to the P_CONNECTION_FEATURES timeout,
P
Philipp Reisner 已提交
973
	 * which we set to 4x the configured ping_timeout. */
974 975 976
	rcu_read_lock();
	nc = rcu_dereference(tconn->net_conf);

977 978
	sock.socket->sk->sk_sndtimeo =
	sock.socket->sk->sk_rcvtimeo = nc->ping_timeo*4*HZ/10;
979

980
	msock.socket->sk->sk_rcvtimeo = nc->ping_int*HZ;
981
	timeout = nc->timeout * HZ / 10;
982
	discard_my_data = nc->discard_my_data;
983
	rcu_read_unlock();
P
Philipp Reisner 已提交
984

985
	msock.socket->sk->sk_sndtimeo = timeout;
P
Philipp Reisner 已提交
986 987

	/* we don't want delays.
L
Lucas De Marchi 已提交
988
	 * we use TCP_CORK where appropriate, though */
989 990
	drbd_tcp_nodelay(sock.socket);
	drbd_tcp_nodelay(msock.socket);
P
Philipp Reisner 已提交
991

992 993
	tconn->data.socket = sock.socket;
	tconn->meta.socket = msock.socket;
994
	tconn->last_received = jiffies;
P
Philipp Reisner 已提交
995

996
	h = drbd_do_features(tconn);
P
Philipp Reisner 已提交
997 998 999
	if (h <= 0)
		return h;

1000
	if (tconn->cram_hmac_tfm) {
P
Philipp Reisner 已提交
1001
		/* drbd_request_state(mdev, NS(conn, WFAuth)); */
1002
		switch (drbd_do_auth(tconn)) {
1003
		case -1:
1004
			conn_err(tconn, "Authentication of peer failed\n");
P
Philipp Reisner 已提交
1005
			return -1;
1006
		case 0:
1007
			conn_err(tconn, "Authentication of peer failed, trying again.\n");
1008
			return 0;
P
Philipp Reisner 已提交
1009 1010 1011
		}
	}

1012 1013
	tconn->data.socket->sk->sk_sndtimeo = timeout;
	tconn->data.socket->sk->sk_rcvtimeo = MAX_SCHEDULE_TIMEOUT;
P
Philipp Reisner 已提交
1014

1015
	if (drbd_send_protocol(tconn) == -EOPNOTSUPP)
1016
		return -1;
P
Philipp Reisner 已提交
1017

1018 1019
	set_bit(STATE_SENT, &tconn->flags);

P
Philipp Reisner 已提交
1020 1021 1022 1023
	rcu_read_lock();
	idr_for_each_entry(&tconn->volumes, mdev, vnr) {
		kref_get(&mdev->kref);
		rcu_read_unlock();
1024 1025 1026 1027 1028 1029

		if (discard_my_data)
			set_bit(DISCARD_MY_DATA, &mdev->flags);
		else
			clear_bit(DISCARD_MY_DATA, &mdev->flags);

P
Philipp Reisner 已提交
1030 1031 1032 1033 1034 1035
		drbd_connected(mdev);
		kref_put(&mdev->kref, &drbd_minor_destroy);
		rcu_read_lock();
	}
	rcu_read_unlock();

1036 1037 1038
	rv = conn_request_state(tconn, NS(conn, C_WF_REPORT_PARAMS), CS_VERBOSE);
	if (rv < SS_SUCCESS) {
		clear_bit(STATE_SENT, &tconn->flags);
1039
		return 0;
1040
	}
1041 1042 1043

	drbd_thread_start(&tconn->asender);

1044 1045 1046 1047 1048 1049 1050 1051
	mutex_lock(&tconn->conf_update);
	/* The discard_my_data flag is a single-shot modifier to the next
	 * connection attempt, the handshake of which is now well underway.
	 * No need for rcu style copying of the whole struct
	 * just to clear a single value. */
	tconn->net_conf->discard_my_data = 0;
	mutex_unlock(&tconn->conf_update);

1052
	return h;
P
Philipp Reisner 已提交
1053 1054

out_release_sockets:
1055 1056 1057 1058
	if (sock.socket)
		sock_release(sock.socket);
	if (msock.socket)
		sock_release(msock.socket);
P
Philipp Reisner 已提交
1059 1060 1061
	return -1;
}

1062
static int decode_header(struct drbd_tconn *tconn, void *header, struct packet_info *pi)
P
Philipp Reisner 已提交
1063
{
1064 1065
	unsigned int header_size = drbd_header_size(tconn);

1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077
	if (header_size == sizeof(struct p_header100) &&
	    *(__be32 *)header == cpu_to_be32(DRBD_MAGIC_100)) {
		struct p_header100 *h = header;
		if (h->pad != 0) {
			conn_err(tconn, "Header padding is not zero\n");
			return -EINVAL;
		}
		pi->vnr = be16_to_cpu(h->volume);
		pi->cmd = be16_to_cpu(h->command);
		pi->size = be32_to_cpu(h->length);
	} else if (header_size == sizeof(struct p_header95) &&
		   *(__be16 *)header == cpu_to_be16(DRBD_MAGIC_BIG)) {
1078 1079
		struct p_header95 *h = header;
		pi->cmd = be16_to_cpu(h->command);
1080 1081
		pi->size = be32_to_cpu(h->length);
		pi->vnr = 0;
1082 1083 1084 1085 1086
	} else if (header_size == sizeof(struct p_header80) &&
		   *(__be32 *)header == cpu_to_be32(DRBD_MAGIC)) {
		struct p_header80 *h = header;
		pi->cmd = be16_to_cpu(h->command);
		pi->size = be16_to_cpu(h->length);
1087
		pi->vnr = 0;
1088
	} else {
1089 1090 1091
		conn_err(tconn, "Wrong magic value 0x%08x in protocol version %d\n",
			 be32_to_cpu(*(__be32 *)header),
			 tconn->agreed_pro_version);
1092
		return -EINVAL;
P
Philipp Reisner 已提交
1093
	}
1094
	pi->data = header + header_size;
1095
	return 0;
1096 1097
}

1098
static int drbd_recv_header(struct drbd_tconn *tconn, struct packet_info *pi)
1099
{
1100
	void *buffer = tconn->data.rbuf;
1101
	int err;
1102

1103
	err = drbd_recv_all_warn(tconn, buffer, drbd_header_size(tconn));
1104
	if (err)
1105
		return err;
1106

1107
	err = decode_header(tconn, buffer, pi);
1108
	tconn->last_received = jiffies;
P
Philipp Reisner 已提交
1109

1110
	return err;
P
Philipp Reisner 已提交
1111 1112
}

1113
static void drbd_flush(struct drbd_tconn *tconn)
P
Philipp Reisner 已提交
1114 1115
{
	int rv;
1116 1117 1118 1119
	struct drbd_conf *mdev;
	int vnr;

	if (tconn->write_ordering >= WO_bdev_flush) {
1120
		rcu_read_lock();
1121
		idr_for_each_entry(&tconn->volumes, mdev, vnr) {
1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134
			if (!get_ldev(mdev))
				continue;
			kref_get(&mdev->kref);
			rcu_read_unlock();

			rv = blkdev_issue_flush(mdev->ldev->backing_bdev,
					GFP_NOIO, NULL);
			if (rv) {
				dev_info(DEV, "local disk flush failed with status %d\n", rv);
				/* would rather check on EOPNOTSUPP, but that is not reliable.
				 * don't try again for ANY return value != 0
				 * if (rv == -EOPNOTSUPP) */
				drbd_bump_write_ordering(tconn, WO_drain_io);
1135
			}
1136 1137 1138 1139 1140 1141
			put_ldev(mdev);
			kref_put(&mdev->kref, &drbd_minor_destroy);

			rcu_read_lock();
			if (rv)
				break;
P
Philipp Reisner 已提交
1142
		}
1143
		rcu_read_unlock();
P
Philipp Reisner 已提交
1144 1145 1146 1147 1148 1149 1150 1151 1152
	}
}

/**
 * drbd_may_finish_epoch() - Applies an epoch_event to the epoch's state, eventually finishes it.
 * @mdev:	DRBD device.
 * @epoch:	Epoch object.
 * @ev:		Epoch event.
 */
1153
static enum finish_epoch drbd_may_finish_epoch(struct drbd_tconn *tconn,
P
Philipp Reisner 已提交
1154 1155 1156
					       struct drbd_epoch *epoch,
					       enum epoch_event ev)
{
1157
	int epoch_size;
P
Philipp Reisner 已提交
1158 1159 1160
	struct drbd_epoch *next_epoch;
	enum finish_epoch rv = FE_STILL_LIVE;

1161
	spin_lock(&tconn->epoch_lock);
P
Philipp Reisner 已提交
1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180
	do {
		next_epoch = NULL;

		epoch_size = atomic_read(&epoch->epoch_size);

		switch (ev & ~EV_CLEANUP) {
		case EV_PUT:
			atomic_dec(&epoch->active);
			break;
		case EV_GOT_BARRIER_NR:
			set_bit(DE_HAVE_BARRIER_NUMBER, &epoch->flags);
			break;
		case EV_BECAME_LAST:
			/* nothing to do*/
			break;
		}

		if (epoch_size != 0 &&
		    atomic_read(&epoch->active) == 0 &&
1181
		    (test_bit(DE_HAVE_BARRIER_NUMBER, &epoch->flags) || ev & EV_CLEANUP)) {
P
Philipp Reisner 已提交
1182
			if (!(ev & EV_CLEANUP)) {
1183
				spin_unlock(&tconn->epoch_lock);
1184
				drbd_send_b_ack(epoch->tconn, epoch->barrier_nr, epoch_size);
1185
				spin_lock(&tconn->epoch_lock);
P
Philipp Reisner 已提交
1186
			}
1187 1188 1189
#if 0
			/* FIXME: dec unacked on connection, once we have
			 * something to count pending connection packets in. */
1190
			if (test_bit(DE_HAVE_BARRIER_NUMBER, &epoch->flags))
1191 1192
				dec_unacked(epoch->tconn);
#endif
P
Philipp Reisner 已提交
1193

1194
			if (tconn->current_epoch != epoch) {
P
Philipp Reisner 已提交
1195 1196 1197
				next_epoch = list_entry(epoch->list.next, struct drbd_epoch, list);
				list_del(&epoch->list);
				ev = EV_BECAME_LAST | (ev & EV_CLEANUP);
1198
				tconn->epochs--;
P
Philipp Reisner 已提交
1199 1200 1201 1202 1203 1204 1205
				kfree(epoch);

				if (rv == FE_STILL_LIVE)
					rv = FE_DESTROYED;
			} else {
				epoch->flags = 0;
				atomic_set(&epoch->epoch_size, 0);
1206
				/* atomic_set(&epoch->active, 0); is already zero */
P
Philipp Reisner 已提交
1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217
				if (rv == FE_STILL_LIVE)
					rv = FE_RECYCLED;
			}
		}

		if (!next_epoch)
			break;

		epoch = next_epoch;
	} while (1);

1218
	spin_unlock(&tconn->epoch_lock);
P
Philipp Reisner 已提交
1219 1220 1221 1222 1223 1224

	return rv;
}

/**
 * drbd_bump_write_ordering() - Fall back to an other write ordering method
1225
 * @tconn:	DRBD connection.
P
Philipp Reisner 已提交
1226 1227
 * @wo:		Write ordering method to try.
 */
1228
void drbd_bump_write_ordering(struct drbd_tconn *tconn, enum write_ordering_e wo)
P
Philipp Reisner 已提交
1229
{
P
Philipp Reisner 已提交
1230
	struct disk_conf *dc;
1231
	struct drbd_conf *mdev;
P
Philipp Reisner 已提交
1232
	enum write_ordering_e pwo;
1233
	int vnr;
P
Philipp Reisner 已提交
1234 1235 1236 1237 1238 1239
	static char *write_ordering_str[] = {
		[WO_none] = "none",
		[WO_drain_io] = "drain",
		[WO_bdev_flush] = "flush",
	};

1240
	pwo = tconn->write_ordering;
P
Philipp Reisner 已提交
1241
	wo = min(pwo, wo);
P
Philipp Reisner 已提交
1242
	rcu_read_lock();
1243
	idr_for_each_entry(&tconn->volumes, mdev, vnr) {
1244
		if (!get_ldev_if_state(mdev, D_ATTACHING))
1245 1246 1247 1248 1249 1250 1251 1252 1253
			continue;
		dc = rcu_dereference(mdev->ldev->disk_conf);

		if (wo == WO_bdev_flush && !dc->disk_flushes)
			wo = WO_drain_io;
		if (wo == WO_drain_io && !dc->disk_drain)
			wo = WO_none;
		put_ldev(mdev);
	}
P
Philipp Reisner 已提交
1254
	rcu_read_unlock();
1255 1256 1257
	tconn->write_ordering = wo;
	if (pwo != tconn->write_ordering || wo == WO_bdev_flush)
		conn_info(tconn, "Method to ensure write ordering: %s\n", write_ordering_str[tconn->write_ordering]);
P
Philipp Reisner 已提交
1258 1259
}

1260
/**
1261
 * drbd_submit_peer_request()
1262
 * @mdev:	DRBD device.
1263
 * @peer_req:	peer request
1264
 * @rw:		flag field, see bio->bi_rw
1265 1266 1267 1268 1269 1270 1271 1272 1273 1274
 *
 * May spread the pages to multiple bios,
 * depending on bio_add_page restrictions.
 *
 * Returns 0 if all bios have been submitted,
 * -ENOMEM if we could not allocate enough bios,
 * -ENOSPC (any better suggestion?) if we have not been able to bio_add_page a
 *  single page to an empty bio (which should never happen and likely indicates
 *  that the lower level IO stack is in some way broken). This has been observed
 *  on certain Xen deployments.
1275 1276
 */
/* TODO allocate from our own bio_set. */
1277 1278 1279
int drbd_submit_peer_request(struct drbd_conf *mdev,
			     struct drbd_peer_request *peer_req,
			     const unsigned rw, const int fault_type)
1280 1281 1282
{
	struct bio *bios = NULL;
	struct bio *bio;
1283 1284 1285
	struct page *page = peer_req->pages;
	sector_t sector = peer_req->i.sector;
	unsigned ds = peer_req->i.size;
1286 1287
	unsigned n_bios = 0;
	unsigned nr_pages = (ds + PAGE_SIZE -1) >> PAGE_SHIFT;
1288
	int err = -ENOMEM;
1289 1290 1291 1292

	/* In most cases, we will only need one bio.  But in case the lower
	 * level restrictions happen to be different at this offset on this
	 * side than those of the sending peer, we may need to submit the
1293 1294 1295 1296 1297
	 * request in more than one bio.
	 *
	 * Plain bio_alloc is good enough here, this is no DRBD internally
	 * generated bio, but a bio allocated on behalf of the peer.
	 */
1298 1299 1300 1301 1302 1303
next_bio:
	bio = bio_alloc(GFP_NOIO, nr_pages);
	if (!bio) {
		dev_err(DEV, "submit_ee: Allocation of a bio failed\n");
		goto fail;
	}
1304
	/* > peer_req->i.sector, unless this is the first bio */
1305 1306 1307
	bio->bi_sector = sector;
	bio->bi_bdev = mdev->ldev->backing_bdev;
	bio->bi_rw = rw;
1308
	bio->bi_private = peer_req;
1309
	bio->bi_end_io = drbd_peer_request_endio;
1310 1311 1312 1313 1314 1315 1316 1317

	bio->bi_next = bios;
	bios = bio;
	++n_bios;

	page_chain_for_each(page) {
		unsigned len = min_t(unsigned, ds, PAGE_SIZE);
		if (!bio_add_page(bio, page, len, 0)) {
1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328
			/* A single page must always be possible!
			 * But in case it fails anyways,
			 * we deal with it, and complain (below). */
			if (bio->bi_vcnt == 0) {
				dev_err(DEV,
					"bio_add_page failed for len=%u, "
					"bi_vcnt=0 (bi_sector=%llu)\n",
					len, (unsigned long long)bio->bi_sector);
				err = -ENOSPC;
				goto fail;
			}
1329 1330 1331 1332 1333 1334 1335 1336 1337
			goto next_bio;
		}
		ds -= len;
		sector += len >> 9;
		--nr_pages;
	}
	D_ASSERT(page == NULL);
	D_ASSERT(ds == 0);

1338
	atomic_set(&peer_req->pending_bios, n_bios);
1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353
	do {
		bio = bios;
		bios = bios->bi_next;
		bio->bi_next = NULL;

		drbd_generic_make_request(mdev, fault_type, bio);
	} while (bios);
	return 0;

fail:
	while (bios) {
		bio = bios;
		bios = bios->bi_next;
		bio_put(bio);
	}
1354
	return err;
1355 1356
}

1357
static void drbd_remove_epoch_entry_interval(struct drbd_conf *mdev,
1358
					     struct drbd_peer_request *peer_req)
1359
{
1360
	struct drbd_interval *i = &peer_req->i;
1361 1362 1363 1364

	drbd_remove_interval(&mdev->write_requests, i);
	drbd_clear_interval(i);

A
Andreas Gruenbacher 已提交
1365
	/* Wake up any processes waiting for this peer request to complete.  */
1366 1367 1368 1369
	if (i->waiting)
		wake_up(&mdev->misc_wait);
}

1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381 1382 1383 1384 1385
void conn_wait_active_ee_empty(struct drbd_tconn *tconn)
{
	struct drbd_conf *mdev;
	int vnr;

	rcu_read_lock();
	idr_for_each_entry(&tconn->volumes, mdev, vnr) {
		kref_get(&mdev->kref);
		rcu_read_unlock();
		drbd_wait_ee_list_empty(mdev, &mdev->active_ee);
		kref_put(&mdev->kref, &drbd_minor_destroy);
		rcu_read_lock();
	}
	rcu_read_unlock();
}

1386
static int receive_Barrier(struct drbd_tconn *tconn, struct packet_info *pi)
P
Philipp Reisner 已提交
1387
{
1388
	int rv;
1389
	struct p_barrier *p = pi->data;
P
Philipp Reisner 已提交
1390 1391
	struct drbd_epoch *epoch;

1392 1393 1394
	/* FIXME these are unacked on connection,
	 * not a specific (peer)device.
	 */
1395
	tconn->current_epoch->barrier_nr = p->barrier;
1396
	tconn->current_epoch->tconn = tconn;
1397
	rv = drbd_may_finish_epoch(tconn, tconn->current_epoch, EV_GOT_BARRIER_NR);
P
Philipp Reisner 已提交
1398 1399 1400 1401 1402 1403

	/* P_BARRIER_ACK may imply that the corresponding extent is dropped from
	 * the activity log, which means it would not be resynced in case the
	 * R_PRIMARY crashes now.
	 * Therefore we must send the barrier_ack after the barrier request was
	 * completed. */
1404
	switch (tconn->write_ordering) {
P
Philipp Reisner 已提交
1405 1406
	case WO_none:
		if (rv == FE_RECYCLED)
1407
			return 0;
1408 1409 1410 1411 1412 1413 1414

		/* receiver context, in the writeout path of the other node.
		 * avoid potential distributed deadlock */
		epoch = kmalloc(sizeof(struct drbd_epoch), GFP_NOIO);
		if (epoch)
			break;
		else
1415
			conn_warn(tconn, "Allocation of an epoch failed, slowing down\n");
1416
			/* Fall through */
P
Philipp Reisner 已提交
1417 1418 1419

	case WO_bdev_flush:
	case WO_drain_io:
1420
		conn_wait_active_ee_empty(tconn);
1421
		drbd_flush(tconn);
1422

1423
		if (atomic_read(&tconn->current_epoch->epoch_size)) {
1424 1425 1426
			epoch = kmalloc(sizeof(struct drbd_epoch), GFP_NOIO);
			if (epoch)
				break;
P
Philipp Reisner 已提交
1427 1428
		}

1429
		return 0;
1430
	default:
1431
		conn_err(tconn, "Strangeness in tconn->write_ordering %d\n", tconn->write_ordering);
1432
		return -EIO;
P
Philipp Reisner 已提交
1433 1434 1435 1436 1437 1438
	}

	epoch->flags = 0;
	atomic_set(&epoch->epoch_size, 0);
	atomic_set(&epoch->active, 0);

1439 1440 1441 1442 1443
	spin_lock(&tconn->epoch_lock);
	if (atomic_read(&tconn->current_epoch->epoch_size)) {
		list_add(&epoch->list, &tconn->current_epoch->list);
		tconn->current_epoch = epoch;
		tconn->epochs++;
P
Philipp Reisner 已提交
1444 1445 1446 1447
	} else {
		/* The current_epoch got recycled while we allocated this one... */
		kfree(epoch);
	}
1448
	spin_unlock(&tconn->epoch_lock);
P
Philipp Reisner 已提交
1449

1450
	return 0;
P
Philipp Reisner 已提交
1451 1452 1453 1454
}

/* used from receive_RSDataReply (recv_resync_read)
 * and from receive_Data */
1455 1456 1457
static struct drbd_peer_request *
read_in_block(struct drbd_conf *mdev, u64 id, sector_t sector,
	      int data_size) __must_hold(local)
P
Philipp Reisner 已提交
1458
{
1459
	const sector_t capacity = drbd_get_capacity(mdev->this_bdev);
1460
	struct drbd_peer_request *peer_req;
P
Philipp Reisner 已提交
1461
	struct page *page;
1462
	int dgs, ds, err;
1463 1464
	void *dig_in = mdev->tconn->int_dig_in;
	void *dig_vv = mdev->tconn->int_dig_vv;
1465
	unsigned long *data;
P
Philipp Reisner 已提交
1466

1467 1468 1469
	dgs = 0;
	if (mdev->tconn->peer_integrity_tfm) {
		dgs = crypto_hash_digestsize(mdev->tconn->peer_integrity_tfm);
1470 1471 1472 1473
		/*
		 * FIXME: Receive the incoming digest into the receive buffer
		 *	  here, together with its struct p_data?
		 */
1474 1475
		err = drbd_recv_all_warn(mdev->tconn, dig_in, dgs);
		if (err)
P
Philipp Reisner 已提交
1476
			return NULL;
1477
		data_size -= dgs;
P
Philipp Reisner 已提交
1478 1479
	}

1480 1481 1482 1483 1484 1485
	if (!expect(data_size != 0))
		return NULL;
	if (!expect(IS_ALIGNED(data_size, 512)))
		return NULL;
	if (!expect(data_size <= DRBD_MAX_BIO_SIZE))
		return NULL;
P
Philipp Reisner 已提交
1486

1487 1488 1489
	/* even though we trust out peer,
	 * we sometimes have to double check. */
	if (sector + (data_size>>9) > capacity) {
1490 1491
		dev_err(DEV, "request from peer beyond end of local disk: "
			"capacity: %llus < sector: %llus + size: %u\n",
1492 1493 1494 1495 1496
			(unsigned long long)capacity,
			(unsigned long long)sector, data_size);
		return NULL;
	}

P
Philipp Reisner 已提交
1497 1498 1499
	/* GFP_NOIO, because we must not cause arbitrary write-out: in a DRBD
	 * "criss-cross" setup, that might cause write-out on some other DRBD,
	 * which in turn might block on the other node at this very place.  */
1500
	peer_req = drbd_alloc_peer_req(mdev, id, sector, data_size, GFP_NOIO);
1501
	if (!peer_req)
P
Philipp Reisner 已提交
1502
		return NULL;
1503

P
Philipp Reisner 已提交
1504
	ds = data_size;
1505
	page = peer_req->pages;
1506 1507
	page_chain_for_each(page) {
		unsigned len = min_t(int, ds, PAGE_SIZE);
1508
		data = kmap(page);
1509
		err = drbd_recv_all_warn(mdev->tconn, data, len);
1510
		if (drbd_insert_fault(mdev, DRBD_FAULT_RECEIVE)) {
1511 1512 1513
			dev_err(DEV, "Fault injection: Corrupting data on receive\n");
			data[0] = data[0] ^ (unsigned long)-1;
		}
P
Philipp Reisner 已提交
1514
		kunmap(page);
1515
		if (err) {
1516
			drbd_free_peer_req(mdev, peer_req);
P
Philipp Reisner 已提交
1517 1518
			return NULL;
		}
1519
		ds -= len;
P
Philipp Reisner 已提交
1520 1521 1522
	}

	if (dgs) {
1523
		drbd_csum_ee(mdev, mdev->tconn->peer_integrity_tfm, peer_req, dig_vv);
P
Philipp Reisner 已提交
1524
		if (memcmp(dig_in, dig_vv, dgs)) {
1525 1526
			dev_err(DEV, "Digest integrity check FAILED: %llus +%u\n",
				(unsigned long long)sector, data_size);
1527
			drbd_free_peer_req(mdev, peer_req);
P
Philipp Reisner 已提交
1528 1529 1530 1531
			return NULL;
		}
	}
	mdev->recv_cnt += data_size>>9;
1532
	return peer_req;
P
Philipp Reisner 已提交
1533 1534 1535 1536 1537 1538 1539 1540
}

/* drbd_drain_block() just takes a data block
 * out of the socket input buffer, and discards it.
 */
static int drbd_drain_block(struct drbd_conf *mdev, int data_size)
{
	struct page *page;
1541
	int err = 0;
P
Philipp Reisner 已提交
1542 1543
	void *data;

1544
	if (!data_size)
1545
		return 0;
1546

1547
	page = drbd_alloc_pages(mdev, 1, 1);
P
Philipp Reisner 已提交
1548 1549 1550

	data = kmap(page);
	while (data_size) {
1551 1552
		unsigned int len = min_t(int, data_size, PAGE_SIZE);

1553 1554
		err = drbd_recv_all_warn(mdev->tconn, data, len);
		if (err)
P
Philipp Reisner 已提交
1555
			break;
1556
		data_size -= len;
P
Philipp Reisner 已提交
1557 1558
	}
	kunmap(page);
1559
	drbd_free_pages(mdev, page, 0);
1560
	return err;
P
Philipp Reisner 已提交
1561 1562 1563 1564 1565 1566 1567
}

static int recv_dless_read(struct drbd_conf *mdev, struct drbd_request *req,
			   sector_t sector, int data_size)
{
	struct bio_vec *bvec;
	struct bio *bio;
1568
	int dgs, err, i, expect;
1569 1570
	void *dig_in = mdev->tconn->int_dig_in;
	void *dig_vv = mdev->tconn->int_dig_vv;
P
Philipp Reisner 已提交
1571

1572 1573 1574
	dgs = 0;
	if (mdev->tconn->peer_integrity_tfm) {
		dgs = crypto_hash_digestsize(mdev->tconn->peer_integrity_tfm);
1575 1576 1577
		err = drbd_recv_all_warn(mdev->tconn, dig_in, dgs);
		if (err)
			return err;
1578
		data_size -= dgs;
P
Philipp Reisner 已提交
1579 1580 1581 1582 1583 1584 1585 1586 1587 1588
	}

	/* optimistically update recv_cnt.  if receiving fails below,
	 * we disconnect anyways, and counters will be reset. */
	mdev->recv_cnt += data_size>>9;

	bio = req->master_bio;
	D_ASSERT(sector == bio->bi_sector);

	bio_for_each_segment(bvec, bio, i) {
1589
		void *mapped = kmap(bvec->bv_page) + bvec->bv_offset;
P
Philipp Reisner 已提交
1590
		expect = min_t(int, data_size, bvec->bv_len);
1591
		err = drbd_recv_all_warn(mdev->tconn, mapped, expect);
P
Philipp Reisner 已提交
1592
		kunmap(bvec->bv_page);
1593 1594 1595
		if (err)
			return err;
		data_size -= expect;
P
Philipp Reisner 已提交
1596 1597 1598
	}

	if (dgs) {
1599
		drbd_csum_bio(mdev, mdev->tconn->peer_integrity_tfm, bio, dig_vv);
P
Philipp Reisner 已提交
1600 1601
		if (memcmp(dig_in, dig_vv, dgs)) {
			dev_err(DEV, "Digest integrity check FAILED. Broken NICs?\n");
1602
			return -EINVAL;
P
Philipp Reisner 已提交
1603 1604 1605 1606
		}
	}

	D_ASSERT(data_size == 0);
1607
	return 0;
P
Philipp Reisner 已提交
1608 1609
}

1610 1611 1612 1613
/*
 * e_end_resync_block() is called in asender context via
 * drbd_finish_peer_reqs().
 */
1614
static int e_end_resync_block(struct drbd_work *w, int unused)
P
Philipp Reisner 已提交
1615
{
1616 1617
	struct drbd_peer_request *peer_req =
		container_of(w, struct drbd_peer_request, w);
1618
	struct drbd_conf *mdev = w->mdev;
1619
	sector_t sector = peer_req->i.sector;
1620
	int err;
P
Philipp Reisner 已提交
1621

1622
	D_ASSERT(drbd_interval_empty(&peer_req->i));
P
Philipp Reisner 已提交
1623

1624 1625
	if (likely((peer_req->flags & EE_WAS_ERROR) == 0)) {
		drbd_set_in_sync(mdev, sector, peer_req->i.size);
1626
		err = drbd_send_ack(mdev, P_RS_WRITE_ACK, peer_req);
P
Philipp Reisner 已提交
1627 1628
	} else {
		/* Record failure to sync */
1629
		drbd_rs_failed_io(mdev, sector, peer_req->i.size);
P
Philipp Reisner 已提交
1630

1631
		err  = drbd_send_ack(mdev, P_NEG_ACK, peer_req);
P
Philipp Reisner 已提交
1632 1633 1634
	}
	dec_unacked(mdev);

1635
	return err;
P
Philipp Reisner 已提交
1636 1637 1638 1639
}

static int recv_resync_read(struct drbd_conf *mdev, sector_t sector, int data_size) __releases(local)
{
1640
	struct drbd_peer_request *peer_req;
P
Philipp Reisner 已提交
1641

1642 1643
	peer_req = read_in_block(mdev, ID_SYNCER, sector, data_size);
	if (!peer_req)
1644
		goto fail;
P
Philipp Reisner 已提交
1645 1646 1647 1648 1649 1650 1651

	dec_rs_pending(mdev);

	inc_unacked(mdev);
	/* corresponding dec_unacked() in e_end_resync_block()
	 * respective _drbd_clear_done_ee */

1652
	peer_req->w.cb = e_end_resync_block;
1653

1654
	spin_lock_irq(&mdev->tconn->req_lock);
1655
	list_add(&peer_req->w.list, &mdev->sync_ee);
1656
	spin_unlock_irq(&mdev->tconn->req_lock);
P
Philipp Reisner 已提交
1657

1658
	atomic_add(data_size >> 9, &mdev->rs_sect_ev);
1659
	if (drbd_submit_peer_request(mdev, peer_req, WRITE, DRBD_FAULT_RS_WR) == 0)
1660
		return 0;
P
Philipp Reisner 已提交
1661

1662 1663
	/* don't care for the reason here */
	dev_err(DEV, "submit failed, triggering re-connect\n");
1664
	spin_lock_irq(&mdev->tconn->req_lock);
1665
	list_del(&peer_req->w.list);
1666
	spin_unlock_irq(&mdev->tconn->req_lock);
1667

1668
	drbd_free_peer_req(mdev, peer_req);
1669 1670
fail:
	put_ldev(mdev);
1671
	return -EIO;
P
Philipp Reisner 已提交
1672 1673
}

1674
static struct drbd_request *
1675 1676
find_request(struct drbd_conf *mdev, struct rb_root *root, u64 id,
	     sector_t sector, bool missing_ok, const char *func)
1677 1678 1679
{
	struct drbd_request *req;

1680 1681
	/* Request object according to our peer */
	req = (struct drbd_request *)(unsigned long)id;
1682
	if (drbd_contains_interval(root, sector, &req->i) && req->i.local)
1683
		return req;
1684
	if (!missing_ok) {
1685
		dev_err(DEV, "%s: failed to find request 0x%lx, sector %llus\n", func,
1686 1687
			(unsigned long)id, (unsigned long long)sector);
	}
1688 1689 1690
	return NULL;
}

1691
static int receive_DataReply(struct drbd_tconn *tconn, struct packet_info *pi)
P
Philipp Reisner 已提交
1692
{
1693
	struct drbd_conf *mdev;
P
Philipp Reisner 已提交
1694 1695
	struct drbd_request *req;
	sector_t sector;
1696
	int err;
1697
	struct p_data *p = pi->data;
1698 1699 1700 1701

	mdev = vnr_to_mdev(tconn, pi->vnr);
	if (!mdev)
		return -EIO;
P
Philipp Reisner 已提交
1702 1703 1704

	sector = be64_to_cpu(p->sector);

1705
	spin_lock_irq(&mdev->tconn->req_lock);
1706
	req = find_request(mdev, &mdev->read_requests, p->block_id, sector, false, __func__);
1707
	spin_unlock_irq(&mdev->tconn->req_lock);
1708
	if (unlikely(!req))
1709
		return -EIO;
P
Philipp Reisner 已提交
1710

B
Bart Van Assche 已提交
1711
	/* hlist_del(&req->collision) is done in _req_may_be_done, to avoid
P
Philipp Reisner 已提交
1712 1713
	 * special casing it there for the various failure cases.
	 * still no race with drbd_fail_pending_reads */
1714
	err = recv_dless_read(mdev, req, sector, pi->size);
1715
	if (!err)
1716
		req_mod(req, DATA_RECEIVED);
P
Philipp Reisner 已提交
1717 1718 1719 1720
	/* else: nothing. handled from drbd_disconnect...
	 * I don't think we may complete this just yet
	 * in case we are "on-disconnect: freeze" */

1721
	return err;
P
Philipp Reisner 已提交
1722 1723
}

1724
static int receive_RSDataReply(struct drbd_tconn *tconn, struct packet_info *pi)
P
Philipp Reisner 已提交
1725
{
1726
	struct drbd_conf *mdev;
P
Philipp Reisner 已提交
1727
	sector_t sector;
1728
	int err;
1729
	struct p_data *p = pi->data;
1730 1731 1732 1733

	mdev = vnr_to_mdev(tconn, pi->vnr);
	if (!mdev)
		return -EIO;
P
Philipp Reisner 已提交
1734 1735 1736 1737 1738 1739 1740

	sector = be64_to_cpu(p->sector);
	D_ASSERT(p->block_id == ID_SYNCER);

	if (get_ldev(mdev)) {
		/* data is submitted to disk within recv_resync_read.
		 * corresponding put_ldev done below on error,
1741
		 * or in drbd_peer_request_endio. */
1742
		err = recv_resync_read(mdev, sector, pi->size);
P
Philipp Reisner 已提交
1743 1744 1745 1746
	} else {
		if (__ratelimit(&drbd_ratelimit_state))
			dev_err(DEV, "Can not write resync data to local disk.\n");

1747
		err = drbd_drain_block(mdev, pi->size);
P
Philipp Reisner 已提交
1748

1749
		drbd_send_ack_dp(mdev, P_NEG_ACK, p, pi->size);
P
Philipp Reisner 已提交
1750 1751
	}

1752
	atomic_add(pi->size >> 9, &mdev->rs_sect_in);
1753

1754
	return err;
P
Philipp Reisner 已提交
1755 1756
}

1757 1758 1759 1760 1761 1762 1763 1764 1765 1766 1767 1768 1769
static void restart_conflicting_writes(struct drbd_conf *mdev,
				       sector_t sector, int size)
{
	struct drbd_interval *i;
	struct drbd_request *req;

	drbd_for_each_overlap(i, &mdev->write_requests, sector, size) {
		if (!i->local)
			continue;
		req = container_of(i, struct drbd_request, i);
		if (req->rq_state & RQ_LOCAL_PENDING ||
		    !(req->rq_state & RQ_POSTPONED))
			continue;
1770 1771 1772
		/* as it is RQ_POSTPONED, this will cause it to
		 * be queued on the retry workqueue. */
		__req_mod(req, DISCARD_WRITE, NULL);
1773 1774 1775
	}
}

1776 1777
/*
 * e_end_block() is called in asender context via drbd_finish_peer_reqs().
P
Philipp Reisner 已提交
1778
 */
1779
static int e_end_block(struct drbd_work *w, int cancel)
P
Philipp Reisner 已提交
1780
{
1781 1782
	struct drbd_peer_request *peer_req =
		container_of(w, struct drbd_peer_request, w);
1783
	struct drbd_conf *mdev = w->mdev;
1784
	sector_t sector = peer_req->i.sector;
1785
	int err = 0, pcmd;
P
Philipp Reisner 已提交
1786

1787
	if (peer_req->flags & EE_SEND_WRITE_ACK) {
1788
		if (likely((peer_req->flags & EE_WAS_ERROR) == 0)) {
P
Philipp Reisner 已提交
1789 1790
			pcmd = (mdev->state.conn >= C_SYNC_SOURCE &&
				mdev->state.conn <= C_PAUSED_SYNC_T &&
1791
				peer_req->flags & EE_MAY_SET_IN_SYNC) ?
P
Philipp Reisner 已提交
1792
				P_RS_WRITE_ACK : P_WRITE_ACK;
1793
			err = drbd_send_ack(mdev, pcmd, peer_req);
P
Philipp Reisner 已提交
1794
			if (pcmd == P_RS_WRITE_ACK)
1795
				drbd_set_in_sync(mdev, sector, peer_req->i.size);
P
Philipp Reisner 已提交
1796
		} else {
1797
			err = drbd_send_ack(mdev, P_NEG_ACK, peer_req);
P
Philipp Reisner 已提交
1798 1799 1800 1801 1802 1803 1804
			/* we expect it to be marked out of sync anyways...
			 * maybe assert this?  */
		}
		dec_unacked(mdev);
	}
	/* we delete from the conflict detection hash _after_ we sent out the
	 * P_WRITE_ACK / P_NEG_ACK, to get the sequence number right.  */
1805
	if (peer_req->flags & EE_IN_INTERVAL_TREE) {
1806
		spin_lock_irq(&mdev->tconn->req_lock);
1807 1808
		D_ASSERT(!drbd_interval_empty(&peer_req->i));
		drbd_remove_epoch_entry_interval(mdev, peer_req);
1809 1810
		if (peer_req->flags & EE_RESTART_REQUESTS)
			restart_conflicting_writes(mdev, sector, peer_req->i.size);
1811
		spin_unlock_irq(&mdev->tconn->req_lock);
1812
	} else
1813
		D_ASSERT(drbd_interval_empty(&peer_req->i));
P
Philipp Reisner 已提交
1814

1815
	drbd_may_finish_epoch(mdev->tconn, peer_req->epoch, EV_PUT + (cancel ? EV_CLEANUP : 0));
P
Philipp Reisner 已提交
1816

1817
	return err;
P
Philipp Reisner 已提交
1818 1819
}

1820
static int e_send_ack(struct drbd_work *w, enum drbd_packet ack)
P
Philipp Reisner 已提交
1821
{
1822
	struct drbd_conf *mdev = w->mdev;
1823 1824
	struct drbd_peer_request *peer_req =
		container_of(w, struct drbd_peer_request, w);
1825
	int err;
P
Philipp Reisner 已提交
1826

1827
	err = drbd_send_ack(mdev, ack, peer_req);
P
Philipp Reisner 已提交
1828 1829
	dec_unacked(mdev);

1830
	return err;
P
Philipp Reisner 已提交
1831 1832
}

1833
static int e_send_discard_write(struct drbd_work *w, int unused)
1834 1835 1836 1837
{
	return e_send_ack(w, P_DISCARD_WRITE);
}

1838
static int e_send_retry_write(struct drbd_work *w, int unused)
1839 1840 1841 1842 1843 1844 1845
{
	struct drbd_tconn *tconn = w->mdev->tconn;

	return e_send_ack(w, tconn->agreed_pro_version >= 100 ?
			     P_RETRY_WRITE : P_DISCARD_WRITE);
}

1846 1847 1848 1849 1850 1851 1852 1853 1854 1855 1856 1857 1858 1859 1860
static bool seq_greater(u32 a, u32 b)
{
	/*
	 * We assume 32-bit wrap-around here.
	 * For 24-bit wrap-around, we would have to shift:
	 *  a <<= 8; b <<= 8;
	 */
	return (s32)a - (s32)b > 0;
}

static u32 seq_max(u32 a, u32 b)
{
	return seq_greater(a, b) ? a : b;
}

1861 1862 1863
static bool need_peer_seq(struct drbd_conf *mdev)
{
	struct drbd_tconn *tconn = mdev->tconn;
1864
	int tp;
1865 1866 1867 1868 1869 1870

	/*
	 * We only need to keep track of the last packet_seq number of our peer
	 * if we are in dual-primary mode and we have the discard flag set; see
	 * handle_write_conflicts().
	 */
1871 1872 1873 1874 1875 1876

	rcu_read_lock();
	tp = rcu_dereference(mdev->tconn->net_conf)->two_primaries;
	rcu_read_unlock();

	return tp && test_bit(DISCARD_CONCURRENT, &tconn->flags);
1877 1878
}

1879
static void update_peer_seq(struct drbd_conf *mdev, unsigned int peer_seq)
1880
{
1881
	unsigned int newest_peer_seq;
1882

1883 1884
	if (need_peer_seq(mdev)) {
		spin_lock(&mdev->peer_seq_lock);
1885 1886
		newest_peer_seq = seq_max(mdev->peer_seq, peer_seq);
		mdev->peer_seq = newest_peer_seq;
1887
		spin_unlock(&mdev->peer_seq_lock);
1888 1889
		/* wake up only if we actually changed mdev->peer_seq */
		if (peer_seq == newest_peer_seq)
1890 1891
			wake_up(&mdev->seq_wait);
	}
1892 1893
}

1894 1895 1896 1897 1898 1899
static inline int overlaps(sector_t s1, int l1, sector_t s2, int l2)
{
	return !((s1 + (l1>>9) <= s2) || (s1 >= s2 + (l2>>9)));
}

/* maybe change sync_ee into interval trees as well? */
1900
static bool overlapping_resync_write(struct drbd_conf *mdev, struct drbd_peer_request *peer_req)
1901 1902 1903 1904 1905 1906 1907 1908 1909 1910 1911 1912 1913 1914 1915 1916 1917
{
	struct drbd_peer_request *rs_req;
	bool rv = 0;

	spin_lock_irq(&mdev->tconn->req_lock);
	list_for_each_entry(rs_req, &mdev->sync_ee, w.list) {
		if (overlaps(peer_req->i.sector, peer_req->i.size,
			     rs_req->i.sector, rs_req->i.size)) {
			rv = 1;
			break;
		}
	}
	spin_unlock_irq(&mdev->tconn->req_lock);

	return rv;
}

P
Philipp Reisner 已提交
1918 1919 1920 1921 1922 1923 1924 1925 1926 1927 1928 1929 1930 1931 1932 1933 1934 1935 1936 1937 1938
/* Called from receive_Data.
 * Synchronize packets on sock with packets on msock.
 *
 * This is here so even when a P_DATA packet traveling via sock overtook an Ack
 * packet traveling on msock, they are still processed in the order they have
 * been sent.
 *
 * Note: we don't care for Ack packets overtaking P_DATA packets.
 *
 * In case packet_seq is larger than mdev->peer_seq number, there are
 * outstanding packets on the msock. We wait for them to arrive.
 * In case we are the logically next packet, we update mdev->peer_seq
 * ourselves. Correctly handles 32bit wrap around.
 *
 * Assume we have a 10 GBit connection, that is about 1<<30 byte per second,
 * about 1<<21 sectors per second. So "worst" case, we have 1<<3 == 8 seconds
 * for the 24bit wrap (historical atomic_t guarantee on some archs), and we have
 * 1<<9 == 512 seconds aka ages for the 32bit wrap around...
 *
 * returns 0 if we may process the packet,
 * -ERESTARTSYS if we were interrupted (by disconnect signal). */
1939
static int wait_for_and_update_peer_seq(struct drbd_conf *mdev, const u32 peer_seq)
P
Philipp Reisner 已提交
1940 1941 1942
{
	DEFINE_WAIT(wait);
	long timeout;
1943 1944 1945 1946 1947
	int ret;

	if (!need_peer_seq(mdev))
		return 0;

P
Philipp Reisner 已提交
1948 1949
	spin_lock(&mdev->peer_seq_lock);
	for (;;) {
1950 1951 1952
		if (!seq_greater(peer_seq - 1, mdev->peer_seq)) {
			mdev->peer_seq = seq_max(mdev->peer_seq, peer_seq);
			ret = 0;
P
Philipp Reisner 已提交
1953
			break;
1954
		}
P
Philipp Reisner 已提交
1955 1956 1957 1958
		if (signal_pending(current)) {
			ret = -ERESTARTSYS;
			break;
		}
1959
		prepare_to_wait(&mdev->seq_wait, &wait, TASK_INTERRUPTIBLE);
P
Philipp Reisner 已提交
1960
		spin_unlock(&mdev->peer_seq_lock);
1961 1962 1963
		rcu_read_lock();
		timeout = rcu_dereference(mdev->tconn->net_conf)->ping_timeo*HZ/10;
		rcu_read_unlock();
1964
		timeout = schedule_timeout(timeout);
P
Philipp Reisner 已提交
1965
		spin_lock(&mdev->peer_seq_lock);
1966
		if (!timeout) {
P
Philipp Reisner 已提交
1967
			ret = -ETIMEDOUT;
1968
			dev_err(DEV, "Timed out waiting for missing ack packets; disconnecting\n");
P
Philipp Reisner 已提交
1969 1970 1971 1972
			break;
		}
	}
	spin_unlock(&mdev->peer_seq_lock);
1973
	finish_wait(&mdev->seq_wait, &wait);
P
Philipp Reisner 已提交
1974 1975 1976
	return ret;
}

1977 1978 1979 1980
/* see also bio_flags_to_wire()
 * DRBD_REQ_*, because we need to semantically map the flags to data packet
 * flags and back. We may replicate to other kernel versions. */
static unsigned long wire_flags_to_bio(struct drbd_conf *mdev, u32 dpf)
1981
{
1982 1983 1984 1985
	return  (dpf & DP_RW_SYNC ? REQ_SYNC : 0) |
		(dpf & DP_FUA ? REQ_FUA : 0) |
		(dpf & DP_FLUSH ? REQ_FLUSH : 0) |
		(dpf & DP_DISCARD ? REQ_DISCARD : 0);
1986 1987
}

1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2036 2037 2038 2039 2040 2041 2042 2043 2044 2045 2046 2047 2048 2049 2050 2051 2052 2053 2054 2055 2056 2057 2058 2059 2060 2061 2062 2063 2064 2065 2066 2067 2068 2069 2070 2071 2072 2073 2074 2075 2076 2077 2078 2079 2080 2081 2082 2083 2084 2085 2086 2087 2088 2089 2090 2091 2092 2093 2094 2095 2096 2097 2098 2099 2100 2101 2102 2103 2104 2105 2106 2107 2108 2109 2110 2111 2112 2113 2114 2115 2116 2117 2118 2119 2120
static void fail_postponed_requests(struct drbd_conf *mdev, sector_t sector,
				    unsigned int size)
{
	struct drbd_interval *i;

    repeat:
	drbd_for_each_overlap(i, &mdev->write_requests, sector, size) {
		struct drbd_request *req;
		struct bio_and_error m;

		if (!i->local)
			continue;
		req = container_of(i, struct drbd_request, i);
		if (!(req->rq_state & RQ_POSTPONED))
			continue;
		req->rq_state &= ~RQ_POSTPONED;
		__req_mod(req, NEG_ACKED, &m);
		spin_unlock_irq(&mdev->tconn->req_lock);
		if (m.bio)
			complete_master_bio(mdev, &m);
		spin_lock_irq(&mdev->tconn->req_lock);
		goto repeat;
	}
}

static int handle_write_conflicts(struct drbd_conf *mdev,
				  struct drbd_peer_request *peer_req)
{
	struct drbd_tconn *tconn = mdev->tconn;
	bool resolve_conflicts = test_bit(DISCARD_CONCURRENT, &tconn->flags);
	sector_t sector = peer_req->i.sector;
	const unsigned int size = peer_req->i.size;
	struct drbd_interval *i;
	bool equal;
	int err;

	/*
	 * Inserting the peer request into the write_requests tree will prevent
	 * new conflicting local requests from being added.
	 */
	drbd_insert_interval(&mdev->write_requests, &peer_req->i);

    repeat:
	drbd_for_each_overlap(i, &mdev->write_requests, sector, size) {
		if (i == &peer_req->i)
			continue;

		if (!i->local) {
			/*
			 * Our peer has sent a conflicting remote request; this
			 * should not happen in a two-node setup.  Wait for the
			 * earlier peer request to complete.
			 */
			err = drbd_wait_misc(mdev, i);
			if (err)
				goto out;
			goto repeat;
		}

		equal = i->sector == sector && i->size == size;
		if (resolve_conflicts) {
			/*
			 * If the peer request is fully contained within the
			 * overlapping request, it can be discarded; otherwise,
			 * it will be retried once all overlapping requests
			 * have completed.
			 */
			bool discard = i->sector <= sector && i->sector +
				       (i->size >> 9) >= sector + (size >> 9);

			if (!equal)
				dev_alert(DEV, "Concurrent writes detected: "
					       "local=%llus +%u, remote=%llus +%u, "
					       "assuming %s came first\n",
					  (unsigned long long)i->sector, i->size,
					  (unsigned long long)sector, size,
					  discard ? "local" : "remote");

			inc_unacked(mdev);
			peer_req->w.cb = discard ? e_send_discard_write :
						   e_send_retry_write;
			list_add_tail(&peer_req->w.list, &mdev->done_ee);
			wake_asender(mdev->tconn);

			err = -ENOENT;
			goto out;
		} else {
			struct drbd_request *req =
				container_of(i, struct drbd_request, i);

			if (!equal)
				dev_alert(DEV, "Concurrent writes detected: "
					       "local=%llus +%u, remote=%llus +%u\n",
					  (unsigned long long)i->sector, i->size,
					  (unsigned long long)sector, size);

			if (req->rq_state & RQ_LOCAL_PENDING ||
			    !(req->rq_state & RQ_POSTPONED)) {
				/*
				 * Wait for the node with the discard flag to
				 * decide if this request will be discarded or
				 * retried.  Requests that are discarded will
				 * disappear from the write_requests tree.
				 *
				 * In addition, wait for the conflicting
				 * request to finish locally before submitting
				 * the conflicting peer request.
				 */
				err = drbd_wait_misc(mdev, &req->i);
				if (err) {
					_conn_request_state(mdev->tconn,
							    NS(conn, C_TIMEOUT),
							    CS_HARD);
					fail_postponed_requests(mdev, sector, size);
					goto out;
				}
				goto repeat;
			}
			/*
			 * Remember to restart the conflicting requests after
			 * the new peer request has completed.
			 */
			peer_req->flags |= EE_RESTART_REQUESTS;
		}
	}
	err = 0;

    out:
	if (err)
		drbd_remove_epoch_entry_interval(mdev, peer_req);
	return err;
}

P
Philipp Reisner 已提交
2121
/* mirrored write */
2122
static int receive_Data(struct drbd_tconn *tconn, struct packet_info *pi)
P
Philipp Reisner 已提交
2123
{
2124
	struct drbd_conf *mdev;
P
Philipp Reisner 已提交
2125
	sector_t sector;
2126
	struct drbd_peer_request *peer_req;
2127
	struct p_data *p = pi->data;
2128
	u32 peer_seq = be32_to_cpu(p->seq_num);
P
Philipp Reisner 已提交
2129 2130
	int rw = WRITE;
	u32 dp_flags;
2131
	int err, tp;
P
Philipp Reisner 已提交
2132

2133 2134 2135 2136
	mdev = vnr_to_mdev(tconn, pi->vnr);
	if (!mdev)
		return -EIO;

2137
	if (!get_ldev(mdev)) {
2138 2139
		int err2;

2140
		err = wait_for_and_update_peer_seq(mdev, peer_seq);
2141
		drbd_send_ack_dp(mdev, P_NEG_ACK, p, pi->size);
2142
		atomic_inc(&tconn->current_epoch->epoch_size);
2143
		err2 = drbd_drain_block(mdev, pi->size);
2144 2145 2146
		if (!err)
			err = err2;
		return err;
P
Philipp Reisner 已提交
2147 2148
	}

2149 2150 2151 2152 2153
	/*
	 * Corresponding put_ldev done either below (on various errors), or in
	 * drbd_peer_request_endio, if we successfully submit the data at the
	 * end of this function.
	 */
P
Philipp Reisner 已提交
2154 2155

	sector = be64_to_cpu(p->sector);
2156
	peer_req = read_in_block(mdev, p->block_id, sector, pi->size);
2157
	if (!peer_req) {
P
Philipp Reisner 已提交
2158
		put_ldev(mdev);
2159
		return -EIO;
P
Philipp Reisner 已提交
2160 2161
	}

2162
	peer_req->w.cb = e_end_block;
P
Philipp Reisner 已提交
2163

2164 2165 2166 2167
	dp_flags = be32_to_cpu(p->dp_flags);
	rw |= wire_flags_to_bio(mdev, dp_flags);

	if (dp_flags & DP_MAY_SET_IN_SYNC)
2168
		peer_req->flags |= EE_MAY_SET_IN_SYNC;
2169

2170 2171
	spin_lock(&tconn->epoch_lock);
	peer_req->epoch = tconn->current_epoch;
2172 2173
	atomic_inc(&peer_req->epoch->epoch_size);
	atomic_inc(&peer_req->epoch->active);
2174
	spin_unlock(&tconn->epoch_lock);
P
Philipp Reisner 已提交
2175

2176 2177 2178 2179 2180
	rcu_read_lock();
	tp = rcu_dereference(mdev->tconn->net_conf)->two_primaries;
	rcu_read_unlock();
	if (tp) {
		peer_req->flags |= EE_IN_INTERVAL_TREE;
2181 2182
		err = wait_for_and_update_peer_seq(mdev, peer_seq);
		if (err)
P
Philipp Reisner 已提交
2183
			goto out_interrupted;
2184
		spin_lock_irq(&mdev->tconn->req_lock);
2185 2186 2187 2188
		err = handle_write_conflicts(mdev, peer_req);
		if (err) {
			spin_unlock_irq(&mdev->tconn->req_lock);
			if (err == -ENOENT) {
P
Philipp Reisner 已提交
2189
				put_ldev(mdev);
2190
				return 0;
P
Philipp Reisner 已提交
2191
			}
2192
			goto out_interrupted;
P
Philipp Reisner 已提交
2193
		}
2194 2195
	} else
		spin_lock_irq(&mdev->tconn->req_lock);
2196
	list_add(&peer_req->w.list, &mdev->active_ee);
2197
	spin_unlock_irq(&mdev->tconn->req_lock);
P
Philipp Reisner 已提交
2198

2199
	if (mdev->state.conn == C_SYNC_TARGET)
2200
		wait_event(mdev->ee_wait, !overlapping_resync_write(mdev, peer_req));
2201

2202
	if (mdev->tconn->agreed_pro_version < 100) {
2203 2204
		rcu_read_lock();
		switch (rcu_dereference(mdev->tconn->net_conf)->wire_protocol) {
2205 2206 2207 2208 2209 2210 2211
		case DRBD_PROT_C:
			dp_flags |= DP_SEND_WRITE_ACK;
			break;
		case DRBD_PROT_B:
			dp_flags |= DP_SEND_RECEIVE_ACK;
			break;
		}
2212
		rcu_read_unlock();
2213 2214 2215 2216
	}

	if (dp_flags & DP_SEND_WRITE_ACK) {
		peer_req->flags |= EE_SEND_WRITE_ACK;
P
Philipp Reisner 已提交
2217 2218 2219
		inc_unacked(mdev);
		/* corresponding dec_unacked() in e_end_block()
		 * respective _drbd_clear_done_ee */
2220 2221 2222
	}

	if (dp_flags & DP_SEND_RECEIVE_ACK) {
P
Philipp Reisner 已提交
2223 2224
		/* I really don't like it that the receiver thread
		 * sends on the msock, but anyways */
2225
		drbd_send_ack(mdev, P_RECV_ACK, peer_req);
P
Philipp Reisner 已提交
2226 2227
	}

2228
	if (mdev->state.pdsk < D_INCONSISTENT) {
P
Philipp Reisner 已提交
2229
		/* In case we have the only disk of the cluster, */
2230 2231 2232
		drbd_set_out_of_sync(mdev, peer_req->i.sector, peer_req->i.size);
		peer_req->flags |= EE_CALL_AL_COMPLETE_IO;
		peer_req->flags &= ~EE_MAY_SET_IN_SYNC;
2233
		drbd_al_begin_io(mdev, &peer_req->i);
P
Philipp Reisner 已提交
2234 2235
	}

2236 2237 2238
	err = drbd_submit_peer_request(mdev, peer_req, rw, DRBD_FAULT_DT_WR);
	if (!err)
		return 0;
P
Philipp Reisner 已提交
2239

2240 2241
	/* don't care for the reason here */
	dev_err(DEV, "submit failed, triggering re-connect\n");
2242
	spin_lock_irq(&mdev->tconn->req_lock);
2243 2244
	list_del(&peer_req->w.list);
	drbd_remove_epoch_entry_interval(mdev, peer_req);
2245
	spin_unlock_irq(&mdev->tconn->req_lock);
2246
	if (peer_req->flags & EE_CALL_AL_COMPLETE_IO)
2247
		drbd_al_complete_io(mdev, &peer_req->i);
2248

P
Philipp Reisner 已提交
2249
out_interrupted:
2250
	drbd_may_finish_epoch(tconn, peer_req->epoch, EV_PUT + EV_CLEANUP);
P
Philipp Reisner 已提交
2251
	put_ldev(mdev);
2252
	drbd_free_peer_req(mdev, peer_req);
2253
	return err;
P
Philipp Reisner 已提交
2254 2255
}

2256 2257 2258 2259 2260 2261 2262 2263 2264 2265 2266
/* We may throttle resync, if the lower device seems to be busy,
 * and current sync rate is above c_min_rate.
 *
 * To decide whether or not the lower device is busy, we use a scheme similar
 * to MD RAID is_mddev_idle(): if the partition stats reveal "significant"
 * (more than 64 sectors) of activity we cannot account for with our own resync
 * activity, it obviously is "busy".
 *
 * The current sync rate used here uses only the most recent two step marks,
 * to have a short time average so we can react faster.
 */
2267
int drbd_rs_should_slow_down(struct drbd_conf *mdev, sector_t sector)
2268 2269 2270
{
	struct gendisk *disk = mdev->ldev->backing_bdev->bd_contains->bd_disk;
	unsigned long db, dt, dbdt;
2271
	struct lc_element *tmp;
2272 2273
	int curr_events;
	int throttle = 0;
P
Philipp Reisner 已提交
2274 2275 2276 2277 2278
	unsigned int c_min_rate;

	rcu_read_lock();
	c_min_rate = rcu_dereference(mdev->ldev->disk_conf)->c_min_rate;
	rcu_read_unlock();
2279 2280

	/* feature disabled? */
P
Philipp Reisner 已提交
2281
	if (c_min_rate == 0)
2282 2283
		return 0;

2284 2285 2286 2287 2288 2289 2290 2291 2292 2293 2294 2295
	spin_lock_irq(&mdev->al_lock);
	tmp = lc_find(mdev->resync, BM_SECT_TO_EXT(sector));
	if (tmp) {
		struct bm_extent *bm_ext = lc_entry(tmp, struct bm_extent, lce);
		if (test_bit(BME_PRIORITY, &bm_ext->flags)) {
			spin_unlock_irq(&mdev->al_lock);
			return 0;
		}
		/* Do not slow down if app IO is already waiting for this extent */
	}
	spin_unlock_irq(&mdev->al_lock);

2296 2297 2298
	curr_events = (int)part_stat_read(&disk->part0, sectors[0]) +
		      (int)part_stat_read(&disk->part0, sectors[1]) -
			atomic_read(&mdev->rs_sect_ev);
2299

2300 2301 2302 2303 2304 2305 2306 2307
	if (!mdev->rs_last_events || curr_events - mdev->rs_last_events > 64) {
		unsigned long rs_left;
		int i;

		mdev->rs_last_events = curr_events;

		/* sync speed average over the last 2*DRBD_SYNC_MARK_STEP,
		 * approx. */
2308 2309 2310 2311 2312 2313
		i = (mdev->rs_last_mark + DRBD_SYNC_MARKS-1) % DRBD_SYNC_MARKS;

		if (mdev->state.conn == C_VERIFY_S || mdev->state.conn == C_VERIFY_T)
			rs_left = mdev->ov_left;
		else
			rs_left = drbd_bm_total_weight(mdev) - mdev->rs_failed;
2314 2315 2316 2317 2318 2319 2320

		dt = ((long)jiffies - (long)mdev->rs_mark_time[i]) / HZ;
		if (!dt)
			dt++;
		db = mdev->rs_mark_left[i] - rs_left;
		dbdt = Bit2KB(db/dt);

P
Philipp Reisner 已提交
2321
		if (dbdt > c_min_rate)
2322 2323 2324 2325 2326 2327
			throttle = 1;
	}
	return throttle;
}


2328
static int receive_DataRequest(struct drbd_tconn *tconn, struct packet_info *pi)
P
Philipp Reisner 已提交
2329
{
2330
	struct drbd_conf *mdev;
P
Philipp Reisner 已提交
2331
	sector_t sector;
2332
	sector_t capacity;
2333
	struct drbd_peer_request *peer_req;
P
Philipp Reisner 已提交
2334
	struct digest_info *di = NULL;
2335
	int size, verb;
P
Philipp Reisner 已提交
2336
	unsigned int fault_type;
2337
	struct p_block_req *p =	pi->data;
2338 2339 2340 2341 2342

	mdev = vnr_to_mdev(tconn, pi->vnr);
	if (!mdev)
		return -EIO;
	capacity = drbd_get_capacity(mdev->this_bdev);
P
Philipp Reisner 已提交
2343 2344 2345 2346

	sector = be64_to_cpu(p->sector);
	size   = be32_to_cpu(p->blksize);

2347
	if (size <= 0 || !IS_ALIGNED(size, 512) || size > DRBD_MAX_BIO_SIZE) {
P
Philipp Reisner 已提交
2348 2349
		dev_err(DEV, "%s:%d: sector: %llus, size: %u\n", __FILE__, __LINE__,
				(unsigned long long)sector, size);
2350
		return -EINVAL;
P
Philipp Reisner 已提交
2351 2352 2353 2354
	}
	if (sector + (size>>9) > capacity) {
		dev_err(DEV, "%s:%d: sector: %llus, size: %u\n", __FILE__, __LINE__,
				(unsigned long long)sector, size);
2355
		return -EINVAL;
P
Philipp Reisner 已提交
2356 2357 2358
	}

	if (!get_ldev_if_state(mdev, D_UP_TO_DATE)) {
2359
		verb = 1;
2360
		switch (pi->cmd) {
2361 2362 2363 2364 2365 2366 2367 2368 2369 2370 2371 2372 2373 2374
		case P_DATA_REQUEST:
			drbd_send_ack_rp(mdev, P_NEG_DREPLY, p);
			break;
		case P_RS_DATA_REQUEST:
		case P_CSUM_RS_REQUEST:
		case P_OV_REQUEST:
			drbd_send_ack_rp(mdev, P_NEG_RS_DREPLY , p);
			break;
		case P_OV_REPLY:
			verb = 0;
			dec_rs_pending(mdev);
			drbd_send_ack_ex(mdev, P_OV_RESULT, sector, size, ID_IN_SYNC);
			break;
		default:
2375
			BUG();
2376 2377
		}
		if (verb && __ratelimit(&drbd_ratelimit_state))
P
Philipp Reisner 已提交
2378 2379
			dev_err(DEV, "Can not satisfy peer's read request, "
			    "no local data.\n");
2380

L
Lars Ellenberg 已提交
2381
		/* drain possibly payload */
2382
		return drbd_drain_block(mdev, pi->size);
P
Philipp Reisner 已提交
2383 2384 2385 2386 2387
	}

	/* GFP_NOIO, because we must not cause arbitrary write-out: in a DRBD
	 * "criss-cross" setup, that might cause write-out on some other DRBD,
	 * which in turn might block on the other node at this very place.  */
2388
	peer_req = drbd_alloc_peer_req(mdev, p->block_id, sector, size, GFP_NOIO);
2389
	if (!peer_req) {
P
Philipp Reisner 已提交
2390
		put_ldev(mdev);
2391
		return -ENOMEM;
P
Philipp Reisner 已提交
2392 2393
	}

2394
	switch (pi->cmd) {
P
Philipp Reisner 已提交
2395
	case P_DATA_REQUEST:
2396
		peer_req->w.cb = w_e_end_data_req;
P
Philipp Reisner 已提交
2397
		fault_type = DRBD_FAULT_DT_RD;
2398 2399 2400
		/* application IO, don't drbd_rs_begin_io */
		goto submit;

P
Philipp Reisner 已提交
2401
	case P_RS_DATA_REQUEST:
2402
		peer_req->w.cb = w_e_end_rsdata_req;
P
Philipp Reisner 已提交
2403
		fault_type = DRBD_FAULT_RS_RD;
2404 2405
		/* used in the sector offset progress display */
		mdev->bm_resync_fo = BM_SECT_TO_BIT(sector);
P
Philipp Reisner 已提交
2406 2407 2408 2409 2410
		break;

	case P_OV_REPLY:
	case P_CSUM_RS_REQUEST:
		fault_type = DRBD_FAULT_RS_RD;
2411
		di = kmalloc(sizeof(*di) + pi->size, GFP_NOIO);
P
Philipp Reisner 已提交
2412 2413 2414
		if (!di)
			goto out_free_e;

2415
		di->digest_size = pi->size;
P
Philipp Reisner 已提交
2416 2417
		di->digest = (((char *)di)+sizeof(struct digest_info));

2418 2419
		peer_req->digest = di;
		peer_req->flags |= EE_HAS_DIGEST;
2420

2421
		if (drbd_recv_all(mdev->tconn, di->digest, pi->size))
P
Philipp Reisner 已提交
2422 2423
			goto out_free_e;

2424
		if (pi->cmd == P_CSUM_RS_REQUEST) {
2425
			D_ASSERT(mdev->tconn->agreed_pro_version >= 89);
2426
			peer_req->w.cb = w_e_end_csum_rs_req;
2427 2428
			/* used in the sector offset progress display */
			mdev->bm_resync_fo = BM_SECT_TO_BIT(sector);
2429
		} else if (pi->cmd == P_OV_REPLY) {
2430 2431
			/* track progress, we may need to throttle */
			atomic_add(size >> 9, &mdev->rs_sect_in);
2432
			peer_req->w.cb = w_e_end_ov_reply;
P
Philipp Reisner 已提交
2433
			dec_rs_pending(mdev);
2434 2435 2436
			/* drbd_rs_begin_io done when we sent this request,
			 * but accounting still needs to be done. */
			goto submit_for_resync;
P
Philipp Reisner 已提交
2437 2438 2439 2440 2441
		}
		break;

	case P_OV_REQUEST:
		if (mdev->ov_start_sector == ~(sector_t)0 &&
2442
		    mdev->tconn->agreed_pro_version >= 90) {
2443 2444
			unsigned long now = jiffies;
			int i;
P
Philipp Reisner 已提交
2445 2446
			mdev->ov_start_sector = sector;
			mdev->ov_position = sector;
2447 2448
			mdev->ov_left = drbd_bm_bits(mdev) - BM_SECT_TO_BIT(sector);
			mdev->rs_total = mdev->ov_left;
2449 2450 2451 2452
			for (i = 0; i < DRBD_SYNC_MARKS; i++) {
				mdev->rs_mark_left[i] = mdev->ov_left;
				mdev->rs_mark_time[i] = now;
			}
P
Philipp Reisner 已提交
2453 2454 2455
			dev_info(DEV, "Online Verify start sector: %llu\n",
					(unsigned long long)sector);
		}
2456
		peer_req->w.cb = w_e_end_ov_req;
P
Philipp Reisner 已提交
2457 2458 2459 2460
		fault_type = DRBD_FAULT_RS_RD;
		break;

	default:
2461
		BUG();
P
Philipp Reisner 已提交
2462 2463
	}

2464 2465 2466 2467 2468 2469 2470 2471 2472 2473 2474 2475 2476 2477 2478 2479 2480 2481 2482 2483 2484 2485
	/* Throttle, drbd_rs_begin_io and submit should become asynchronous
	 * wrt the receiver, but it is not as straightforward as it may seem.
	 * Various places in the resync start and stop logic assume resync
	 * requests are processed in order, requeuing this on the worker thread
	 * introduces a bunch of new code for synchronization between threads.
	 *
	 * Unlimited throttling before drbd_rs_begin_io may stall the resync
	 * "forever", throttling after drbd_rs_begin_io will lock that extent
	 * for application writes for the same time.  For now, just throttle
	 * here, where the rest of the code expects the receiver to sleep for
	 * a while, anyways.
	 */

	/* Throttle before drbd_rs_begin_io, as that locks out application IO;
	 * this defers syncer requests for some time, before letting at least
	 * on request through.  The resync controller on the receiving side
	 * will adapt to the incoming rate accordingly.
	 *
	 * We cannot throttle here if remote is Primary/SyncTarget:
	 * we would also throttle its application reads.
	 * In that case, throttling is done on the SyncTarget only.
	 */
2486 2487 2488
	if (mdev->state.peer != R_PRIMARY && drbd_rs_should_slow_down(mdev, sector))
		schedule_timeout_uninterruptible(HZ/10);
	if (drbd_rs_begin_io(mdev, sector))
2489
		goto out_free_e;
P
Philipp Reisner 已提交
2490

2491 2492 2493
submit_for_resync:
	atomic_add(size >> 9, &mdev->rs_sect_ev);

2494
submit:
P
Philipp Reisner 已提交
2495
	inc_unacked(mdev);
2496
	spin_lock_irq(&mdev->tconn->req_lock);
2497
	list_add_tail(&peer_req->w.list, &mdev->read_ee);
2498
	spin_unlock_irq(&mdev->tconn->req_lock);
P
Philipp Reisner 已提交
2499

2500
	if (drbd_submit_peer_request(mdev, peer_req, READ, fault_type) == 0)
2501
		return 0;
P
Philipp Reisner 已提交
2502

2503 2504
	/* don't care for the reason here */
	dev_err(DEV, "submit failed, triggering re-connect\n");
2505
	spin_lock_irq(&mdev->tconn->req_lock);
2506
	list_del(&peer_req->w.list);
2507
	spin_unlock_irq(&mdev->tconn->req_lock);
2508 2509
	/* no drbd_rs_complete_io(), we are dropping the connection anyways */

P
Philipp Reisner 已提交
2510 2511
out_free_e:
	put_ldev(mdev);
2512
	drbd_free_peer_req(mdev, peer_req);
2513
	return -EIO;
P
Philipp Reisner 已提交
2514 2515 2516 2517 2518 2519
}

static int drbd_asb_recover_0p(struct drbd_conf *mdev) __must_hold(local)
{
	int self, peer, rv = -100;
	unsigned long ch_self, ch_peer;
2520
	enum drbd_after_sb_p after_sb_0p;
P
Philipp Reisner 已提交
2521 2522 2523 2524 2525 2526 2527

	self = mdev->ldev->md.uuid[UI_BITMAP] & 1;
	peer = mdev->p_uuid[UI_BITMAP] & 1;

	ch_peer = mdev->p_uuid[UI_SIZE];
	ch_self = mdev->comm_bm_set;

2528 2529 2530 2531
	rcu_read_lock();
	after_sb_0p = rcu_dereference(mdev->tconn->net_conf)->after_sb_0p;
	rcu_read_unlock();
	switch (after_sb_0p) {
P
Philipp Reisner 已提交
2532 2533 2534
	case ASB_CONSENSUS:
	case ASB_DISCARD_SECONDARY:
	case ASB_CALL_HELPER:
2535
	case ASB_VIOLENTLY:
P
Philipp Reisner 已提交
2536 2537 2538 2539 2540 2541 2542 2543 2544 2545 2546 2547 2548 2549 2550 2551 2552 2553 2554 2555 2556 2557 2558 2559
		dev_err(DEV, "Configuration error.\n");
		break;
	case ASB_DISCONNECT:
		break;
	case ASB_DISCARD_YOUNGER_PRI:
		if (self == 0 && peer == 1) {
			rv = -1;
			break;
		}
		if (self == 1 && peer == 0) {
			rv =  1;
			break;
		}
		/* Else fall through to one of the other strategies... */
	case ASB_DISCARD_OLDER_PRI:
		if (self == 0 && peer == 1) {
			rv = 1;
			break;
		}
		if (self == 1 && peer == 0) {
			rv = -1;
			break;
		}
		/* Else fall through to one of the other strategies... */
L
Lars Ellenberg 已提交
2560
		dev_warn(DEV, "Discard younger/older primary did not find a decision\n"
P
Philipp Reisner 已提交
2561 2562 2563
		     "Using discard-least-changes instead\n");
	case ASB_DISCARD_ZERO_CHG:
		if (ch_peer == 0 && ch_self == 0) {
2564
			rv = test_bit(DISCARD_CONCURRENT, &mdev->tconn->flags)
P
Philipp Reisner 已提交
2565 2566 2567 2568 2569 2570
				? -1 : 1;
			break;
		} else {
			if (ch_peer == 0) { rv =  1; break; }
			if (ch_self == 0) { rv = -1; break; }
		}
2571
		if (after_sb_0p == ASB_DISCARD_ZERO_CHG)
P
Philipp Reisner 已提交
2572 2573 2574 2575 2576 2577 2578 2579
			break;
	case ASB_DISCARD_LEAST_CHG:
		if	(ch_self < ch_peer)
			rv = -1;
		else if (ch_self > ch_peer)
			rv =  1;
		else /* ( ch_self == ch_peer ) */
		     /* Well, then use something else. */
2580
			rv = test_bit(DISCARD_CONCURRENT, &mdev->tconn->flags)
P
Philipp Reisner 已提交
2581 2582 2583 2584 2585 2586 2587 2588 2589 2590 2591 2592 2593 2594
				? -1 : 1;
		break;
	case ASB_DISCARD_LOCAL:
		rv = -1;
		break;
	case ASB_DISCARD_REMOTE:
		rv =  1;
	}

	return rv;
}

static int drbd_asb_recover_1p(struct drbd_conf *mdev) __must_hold(local)
{
2595
	int hg, rv = -100;
2596
	enum drbd_after_sb_p after_sb_1p;
P
Philipp Reisner 已提交
2597

2598 2599 2600 2601
	rcu_read_lock();
	after_sb_1p = rcu_dereference(mdev->tconn->net_conf)->after_sb_1p;
	rcu_read_unlock();
	switch (after_sb_1p) {
P
Philipp Reisner 已提交
2602 2603 2604 2605 2606
	case ASB_DISCARD_YOUNGER_PRI:
	case ASB_DISCARD_OLDER_PRI:
	case ASB_DISCARD_LEAST_CHG:
	case ASB_DISCARD_LOCAL:
	case ASB_DISCARD_REMOTE:
2607
	case ASB_DISCARD_ZERO_CHG:
P
Philipp Reisner 已提交
2608 2609 2610 2611 2612 2613 2614 2615 2616 2617 2618 2619 2620 2621 2622 2623 2624 2625 2626
		dev_err(DEV, "Configuration error.\n");
		break;
	case ASB_DISCONNECT:
		break;
	case ASB_CONSENSUS:
		hg = drbd_asb_recover_0p(mdev);
		if (hg == -1 && mdev->state.role == R_SECONDARY)
			rv = hg;
		if (hg == 1  && mdev->state.role == R_PRIMARY)
			rv = hg;
		break;
	case ASB_VIOLENTLY:
		rv = drbd_asb_recover_0p(mdev);
		break;
	case ASB_DISCARD_SECONDARY:
		return mdev->state.role == R_PRIMARY ? 1 : -1;
	case ASB_CALL_HELPER:
		hg = drbd_asb_recover_0p(mdev);
		if (hg == -1 && mdev->state.role == R_PRIMARY) {
2627 2628 2629
			enum drbd_state_rv rv2;

			drbd_set_role(mdev, R_SECONDARY, 0);
P
Philipp Reisner 已提交
2630 2631 2632
			 /* drbd_change_state() does not sleep while in SS_IN_TRANSIENT_STATE,
			  * we might be here in C_WF_REPORT_PARAMS which is transient.
			  * we do not need to wait for the after state change work either. */
2633 2634
			rv2 = drbd_change_state(mdev, CS_VERBOSE, NS(role, R_SECONDARY));
			if (rv2 != SS_SUCCESS) {
P
Philipp Reisner 已提交
2635 2636 2637 2638 2639 2640 2641 2642 2643 2644 2645 2646 2647 2648
				drbd_khelper(mdev, "pri-lost-after-sb");
			} else {
				dev_warn(DEV, "Successfully gave up primary role.\n");
				rv = hg;
			}
		} else
			rv = hg;
	}

	return rv;
}

static int drbd_asb_recover_2p(struct drbd_conf *mdev) __must_hold(local)
{
2649
	int hg, rv = -100;
2650
	enum drbd_after_sb_p after_sb_2p;
P
Philipp Reisner 已提交
2651

2652 2653 2654 2655
	rcu_read_lock();
	after_sb_2p = rcu_dereference(mdev->tconn->net_conf)->after_sb_2p;
	rcu_read_unlock();
	switch (after_sb_2p) {
P
Philipp Reisner 已提交
2656 2657 2658 2659 2660 2661 2662
	case ASB_DISCARD_YOUNGER_PRI:
	case ASB_DISCARD_OLDER_PRI:
	case ASB_DISCARD_LEAST_CHG:
	case ASB_DISCARD_LOCAL:
	case ASB_DISCARD_REMOTE:
	case ASB_CONSENSUS:
	case ASB_DISCARD_SECONDARY:
2663
	case ASB_DISCARD_ZERO_CHG:
P
Philipp Reisner 已提交
2664 2665 2666 2667 2668 2669 2670 2671 2672 2673
		dev_err(DEV, "Configuration error.\n");
		break;
	case ASB_VIOLENTLY:
		rv = drbd_asb_recover_0p(mdev);
		break;
	case ASB_DISCONNECT:
		break;
	case ASB_CALL_HELPER:
		hg = drbd_asb_recover_0p(mdev);
		if (hg == -1) {
2674 2675
			enum drbd_state_rv rv2;

P
Philipp Reisner 已提交
2676 2677 2678
			 /* drbd_change_state() does not sleep while in SS_IN_TRANSIENT_STATE,
			  * we might be here in C_WF_REPORT_PARAMS which is transient.
			  * we do not need to wait for the after state change work either. */
2679 2680
			rv2 = drbd_change_state(mdev, CS_VERBOSE, NS(role, R_SECONDARY));
			if (rv2 != SS_SUCCESS) {
P
Philipp Reisner 已提交
2681 2682 2683 2684 2685 2686 2687 2688 2689 2690 2691 2692 2693 2694 2695 2696 2697 2698 2699 2700 2701 2702 2703 2704 2705 2706 2707 2708 2709 2710 2711 2712 2713 2714 2715 2716 2717 2718
				drbd_khelper(mdev, "pri-lost-after-sb");
			} else {
				dev_warn(DEV, "Successfully gave up primary role.\n");
				rv = hg;
			}
		} else
			rv = hg;
	}

	return rv;
}

static void drbd_uuid_dump(struct drbd_conf *mdev, char *text, u64 *uuid,
			   u64 bits, u64 flags)
{
	if (!uuid) {
		dev_info(DEV, "%s uuid info vanished while I was looking!\n", text);
		return;
	}
	dev_info(DEV, "%s %016llX:%016llX:%016llX:%016llX bits:%llu flags:%llX\n",
	     text,
	     (unsigned long long)uuid[UI_CURRENT],
	     (unsigned long long)uuid[UI_BITMAP],
	     (unsigned long long)uuid[UI_HISTORY_START],
	     (unsigned long long)uuid[UI_HISTORY_END],
	     (unsigned long long)bits,
	     (unsigned long long)flags);
}

/*
  100	after split brain try auto recover
    2	C_SYNC_SOURCE set BitMap
    1	C_SYNC_SOURCE use BitMap
    0	no Sync
   -1	C_SYNC_TARGET use BitMap
   -2	C_SYNC_TARGET set BitMap
 -100	after split brain, disconnect
-1000	unrelated data
2719 2720
-1091   requires proto 91
-1096   requires proto 96
P
Philipp Reisner 已提交
2721 2722 2723 2724 2725 2726 2727 2728 2729 2730 2731 2732 2733 2734 2735 2736 2737 2738 2739 2740 2741 2742 2743 2744 2745 2746 2747 2748
 */
static int drbd_uuid_compare(struct drbd_conf *mdev, int *rule_nr) __must_hold(local)
{
	u64 self, peer;
	int i, j;

	self = mdev->ldev->md.uuid[UI_CURRENT] & ~((u64)1);
	peer = mdev->p_uuid[UI_CURRENT] & ~((u64)1);

	*rule_nr = 10;
	if (self == UUID_JUST_CREATED && peer == UUID_JUST_CREATED)
		return 0;

	*rule_nr = 20;
	if ((self == UUID_JUST_CREATED || self == (u64)0) &&
	     peer != UUID_JUST_CREATED)
		return -2;

	*rule_nr = 30;
	if (self != UUID_JUST_CREATED &&
	    (peer == UUID_JUST_CREATED || peer == (u64)0))
		return 2;

	if (self == peer) {
		int rct, dc; /* roles at crash time */

		if (mdev->p_uuid[UI_BITMAP] == (u64)0 && mdev->ldev->md.uuid[UI_BITMAP] != (u64)0) {

2749
			if (mdev->tconn->agreed_pro_version < 91)
2750
				return -1091;
P
Philipp Reisner 已提交
2751 2752 2753 2754 2755 2756 2757 2758 2759 2760 2761 2762 2763 2764 2765 2766 2767 2768 2769

			if ((mdev->ldev->md.uuid[UI_BITMAP] & ~((u64)1)) == (mdev->p_uuid[UI_HISTORY_START] & ~((u64)1)) &&
			    (mdev->ldev->md.uuid[UI_HISTORY_START] & ~((u64)1)) == (mdev->p_uuid[UI_HISTORY_START + 1] & ~((u64)1))) {
				dev_info(DEV, "was SyncSource, missed the resync finished event, corrected myself:\n");
				drbd_uuid_set_bm(mdev, 0UL);

				drbd_uuid_dump(mdev, "self", mdev->ldev->md.uuid,
					       mdev->state.disk >= D_NEGOTIATING ? drbd_bm_total_weight(mdev) : 0, 0);
				*rule_nr = 34;
			} else {
				dev_info(DEV, "was SyncSource (peer failed to write sync_uuid)\n");
				*rule_nr = 36;
			}

			return 1;
		}

		if (mdev->ldev->md.uuid[UI_BITMAP] == (u64)0 && mdev->p_uuid[UI_BITMAP] != (u64)0) {

2770
			if (mdev->tconn->agreed_pro_version < 91)
2771
				return -1091;
P
Philipp Reisner 已提交
2772 2773 2774 2775 2776 2777 2778 2779 2780 2781 2782 2783 2784 2785 2786 2787 2788 2789 2790 2791 2792 2793 2794 2795 2796 2797 2798 2799 2800 2801 2802

			if ((mdev->ldev->md.uuid[UI_HISTORY_START] & ~((u64)1)) == (mdev->p_uuid[UI_BITMAP] & ~((u64)1)) &&
			    (mdev->ldev->md.uuid[UI_HISTORY_START + 1] & ~((u64)1)) == (mdev->p_uuid[UI_HISTORY_START] & ~((u64)1))) {
				dev_info(DEV, "was SyncTarget, peer missed the resync finished event, corrected peer:\n");

				mdev->p_uuid[UI_HISTORY_START + 1] = mdev->p_uuid[UI_HISTORY_START];
				mdev->p_uuid[UI_HISTORY_START] = mdev->p_uuid[UI_BITMAP];
				mdev->p_uuid[UI_BITMAP] = 0UL;

				drbd_uuid_dump(mdev, "peer", mdev->p_uuid, mdev->p_uuid[UI_SIZE], mdev->p_uuid[UI_FLAGS]);
				*rule_nr = 35;
			} else {
				dev_info(DEV, "was SyncTarget (failed to write sync_uuid)\n");
				*rule_nr = 37;
			}

			return -1;
		}

		/* Common power [off|failure] */
		rct = (test_bit(CRASHED_PRIMARY, &mdev->flags) ? 1 : 0) +
			(mdev->p_uuid[UI_FLAGS] & 2);
		/* lowest bit is set when we were primary,
		 * next bit (weight 2) is set when peer was primary */
		*rule_nr = 40;

		switch (rct) {
		case 0: /* !self_pri && !peer_pri */ return 0;
		case 1: /*  self_pri && !peer_pri */ return 1;
		case 2: /* !self_pri &&  peer_pri */ return -1;
		case 3: /*  self_pri &&  peer_pri */
2803
			dc = test_bit(DISCARD_CONCURRENT, &mdev->tconn->flags);
P
Philipp Reisner 已提交
2804 2805 2806 2807 2808 2809 2810 2811 2812 2813 2814 2815
			return dc ? -1 : 1;
		}
	}

	*rule_nr = 50;
	peer = mdev->p_uuid[UI_BITMAP] & ~((u64)1);
	if (self == peer)
		return -1;

	*rule_nr = 51;
	peer = mdev->p_uuid[UI_HISTORY_START] & ~((u64)1);
	if (self == peer) {
2816
		if (mdev->tconn->agreed_pro_version < 96 ?
2817 2818 2819
		    (mdev->ldev->md.uuid[UI_HISTORY_START] & ~((u64)1)) ==
		    (mdev->p_uuid[UI_HISTORY_START + 1] & ~((u64)1)) :
		    peer + UUID_NEW_BM_OFFSET == (mdev->p_uuid[UI_BITMAP] & ~((u64)1))) {
P
Philipp Reisner 已提交
2820 2821 2822
			/* The last P_SYNC_UUID did not get though. Undo the last start of
			   resync as sync source modifications of the peer's UUIDs. */

2823
			if (mdev->tconn->agreed_pro_version < 91)
2824
				return -1091;
P
Philipp Reisner 已提交
2825 2826 2827

			mdev->p_uuid[UI_BITMAP] = mdev->p_uuid[UI_HISTORY_START];
			mdev->p_uuid[UI_HISTORY_START] = mdev->p_uuid[UI_HISTORY_START + 1];
2828

L
Lars Ellenberg 已提交
2829
			dev_info(DEV, "Lost last syncUUID packet, corrected:\n");
2830 2831
			drbd_uuid_dump(mdev, "peer", mdev->p_uuid, mdev->p_uuid[UI_SIZE], mdev->p_uuid[UI_FLAGS]);

P
Philipp Reisner 已提交
2832 2833 2834 2835 2836 2837 2838 2839 2840 2841 2842 2843 2844 2845 2846 2847 2848 2849 2850 2851 2852
			return -1;
		}
	}

	*rule_nr = 60;
	self = mdev->ldev->md.uuid[UI_CURRENT] & ~((u64)1);
	for (i = UI_HISTORY_START; i <= UI_HISTORY_END; i++) {
		peer = mdev->p_uuid[i] & ~((u64)1);
		if (self == peer)
			return -2;
	}

	*rule_nr = 70;
	self = mdev->ldev->md.uuid[UI_BITMAP] & ~((u64)1);
	peer = mdev->p_uuid[UI_CURRENT] & ~((u64)1);
	if (self == peer)
		return 1;

	*rule_nr = 71;
	self = mdev->ldev->md.uuid[UI_HISTORY_START] & ~((u64)1);
	if (self == peer) {
2853
		if (mdev->tconn->agreed_pro_version < 96 ?
2854 2855 2856
		    (mdev->ldev->md.uuid[UI_HISTORY_START + 1] & ~((u64)1)) ==
		    (mdev->p_uuid[UI_HISTORY_START] & ~((u64)1)) :
		    self + UUID_NEW_BM_OFFSET == (mdev->ldev->md.uuid[UI_BITMAP] & ~((u64)1))) {
P
Philipp Reisner 已提交
2857 2858 2859
			/* The last P_SYNC_UUID did not get though. Undo the last start of
			   resync as sync source modifications of our UUIDs. */

2860
			if (mdev->tconn->agreed_pro_version < 91)
2861
				return -1091;
P
Philipp Reisner 已提交
2862 2863 2864 2865

			_drbd_uuid_set(mdev, UI_BITMAP, mdev->ldev->md.uuid[UI_HISTORY_START]);
			_drbd_uuid_set(mdev, UI_HISTORY_START, mdev->ldev->md.uuid[UI_HISTORY_START + 1]);

2866
			dev_info(DEV, "Last syncUUID did not get through, corrected:\n");
P
Philipp Reisner 已提交
2867 2868 2869 2870 2871 2872 2873 2874 2875
			drbd_uuid_dump(mdev, "self", mdev->ldev->md.uuid,
				       mdev->state.disk >= D_NEGOTIATING ? drbd_bm_total_weight(mdev) : 0, 0);

			return 1;
		}
	}


	*rule_nr = 80;
2876
	peer = mdev->p_uuid[UI_CURRENT] & ~((u64)1);
P
Philipp Reisner 已提交
2877 2878 2879 2880 2881 2882 2883 2884 2885 2886 2887 2888 2889 2890 2891 2892 2893 2894 2895 2896 2897 2898 2899 2900 2901 2902 2903 2904 2905 2906 2907 2908 2909
	for (i = UI_HISTORY_START; i <= UI_HISTORY_END; i++) {
		self = mdev->ldev->md.uuid[i] & ~((u64)1);
		if (self == peer)
			return 2;
	}

	*rule_nr = 90;
	self = mdev->ldev->md.uuid[UI_BITMAP] & ~((u64)1);
	peer = mdev->p_uuid[UI_BITMAP] & ~((u64)1);
	if (self == peer && self != ((u64)0))
		return 100;

	*rule_nr = 100;
	for (i = UI_HISTORY_START; i <= UI_HISTORY_END; i++) {
		self = mdev->ldev->md.uuid[i] & ~((u64)1);
		for (j = UI_HISTORY_START; j <= UI_HISTORY_END; j++) {
			peer = mdev->p_uuid[j] & ~((u64)1);
			if (self == peer)
				return -100;
		}
	}

	return -1000;
}

/* drbd_sync_handshake() returns the new conn state on success, or
   CONN_MASK (-1) on failure.
 */
static enum drbd_conns drbd_sync_handshake(struct drbd_conf *mdev, enum drbd_role peer_role,
					   enum drbd_disk_state peer_disk) __must_hold(local)
{
	enum drbd_conns rv = C_MASK;
	enum drbd_disk_state mydisk;
2910
	struct net_conf *nc;
2911
	int hg, rule_nr, rr_conflict, tentative;
P
Philipp Reisner 已提交
2912 2913 2914 2915 2916 2917 2918 2919 2920 2921 2922 2923 2924 2925 2926 2927 2928 2929

	mydisk = mdev->state.disk;
	if (mydisk == D_NEGOTIATING)
		mydisk = mdev->new_state_tmp.disk;

	dev_info(DEV, "drbd_sync_handshake:\n");
	drbd_uuid_dump(mdev, "self", mdev->ldev->md.uuid, mdev->comm_bm_set, 0);
	drbd_uuid_dump(mdev, "peer", mdev->p_uuid,
		       mdev->p_uuid[UI_SIZE], mdev->p_uuid[UI_FLAGS]);

	hg = drbd_uuid_compare(mdev, &rule_nr);

	dev_info(DEV, "uuid_compare()=%d by rule %d\n", hg, rule_nr);

	if (hg == -1000) {
		dev_alert(DEV, "Unrelated data, aborting!\n");
		return C_MASK;
	}
2930 2931
	if (hg < -1000) {
		dev_alert(DEV, "To resolve this both sides have to support at least protocol %d\n", -hg - 1000);
P
Philipp Reisner 已提交
2932 2933 2934 2935 2936 2937 2938 2939 2940 2941 2942 2943 2944
		return C_MASK;
	}

	if    ((mydisk == D_INCONSISTENT && peer_disk > D_INCONSISTENT) ||
	    (peer_disk == D_INCONSISTENT && mydisk    > D_INCONSISTENT)) {
		int f = (hg == -100) || abs(hg) == 2;
		hg = mydisk > D_INCONSISTENT ? 1 : -1;
		if (f)
			hg = hg*2;
		dev_info(DEV, "Becoming sync %s due to disk states.\n",
		     hg > 0 ? "source" : "target");
	}

2945 2946 2947
	if (abs(hg) == 100)
		drbd_khelper(mdev, "initial-split-brain");

2948 2949 2950 2951
	rcu_read_lock();
	nc = rcu_dereference(mdev->tconn->net_conf);

	if (hg == 100 || (hg == -100 && nc->always_asbp)) {
P
Philipp Reisner 已提交
2952 2953 2954 2955 2956 2957 2958 2959 2960 2961 2962 2963 2964 2965 2966 2967 2968 2969 2970 2971 2972 2973 2974 2975 2976 2977 2978 2979
		int pcount = (mdev->state.role == R_PRIMARY)
			   + (peer_role == R_PRIMARY);
		int forced = (hg == -100);

		switch (pcount) {
		case 0:
			hg = drbd_asb_recover_0p(mdev);
			break;
		case 1:
			hg = drbd_asb_recover_1p(mdev);
			break;
		case 2:
			hg = drbd_asb_recover_2p(mdev);
			break;
		}
		if (abs(hg) < 100) {
			dev_warn(DEV, "Split-Brain detected, %d primaries, "
			     "automatically solved. Sync from %s node\n",
			     pcount, (hg < 0) ? "peer" : "this");
			if (forced) {
				dev_warn(DEV, "Doing a full sync, since"
				     " UUIDs where ambiguous.\n");
				hg = hg*2;
			}
		}
	}

	if (hg == -100) {
2980
		if (test_bit(DISCARD_MY_DATA, &mdev->flags) && !(mdev->p_uuid[UI_FLAGS]&1))
P
Philipp Reisner 已提交
2981
			hg = -1;
2982
		if (!test_bit(DISCARD_MY_DATA, &mdev->flags) && (mdev->p_uuid[UI_FLAGS]&1))
P
Philipp Reisner 已提交
2983 2984 2985 2986 2987 2988 2989
			hg = 1;

		if (abs(hg) < 100)
			dev_warn(DEV, "Split-Brain detected, manually solved. "
			     "Sync from %s node\n",
			     (hg < 0) ? "peer" : "this");
	}
2990
	rr_conflict = nc->rr_conflict;
2991
	tentative = nc->tentative;
2992
	rcu_read_unlock();
P
Philipp Reisner 已提交
2993 2994

	if (hg == -100) {
2995 2996 2997 2998
		/* FIXME this log message is not correct if we end up here
		 * after an attempted attach on a diskless node.
		 * We just refuse to attach -- well, we drop the "connection"
		 * to that disk, in a way... */
2999
		dev_alert(DEV, "Split-Brain detected but unresolved, dropping connection!\n");
P
Philipp Reisner 已提交
3000 3001 3002 3003 3004 3005 3006 3007 3008 3009 3010
		drbd_khelper(mdev, "split-brain");
		return C_MASK;
	}

	if (hg > 0 && mydisk <= D_INCONSISTENT) {
		dev_err(DEV, "I shall become SyncSource, but I am inconsistent!\n");
		return C_MASK;
	}

	if (hg < 0 && /* by intention we do not use mydisk here. */
	    mdev->state.role == R_PRIMARY && mdev->state.disk >= D_CONSISTENT) {
3011
		switch (rr_conflict) {
P
Philipp Reisner 已提交
3012 3013 3014 3015 3016 3017 3018 3019 3020 3021 3022 3023
		case ASB_CALL_HELPER:
			drbd_khelper(mdev, "pri-lost");
			/* fall through */
		case ASB_DISCONNECT:
			dev_err(DEV, "I shall become SyncTarget, but I am primary!\n");
			return C_MASK;
		case ASB_VIOLENTLY:
			dev_warn(DEV, "Becoming SyncTarget, violating the stable-data"
			     "assumption\n");
		}
	}

3024
	if (tentative || test_bit(CONN_DRY_RUN, &mdev->tconn->flags)) {
3025 3026 3027 3028 3029 3030 3031 3032 3033
		if (hg == 0)
			dev_info(DEV, "dry-run connect: No resync, would become Connected immediately.\n");
		else
			dev_info(DEV, "dry-run connect: Would become %s, doing a %s resync.",
				 drbd_conn_str(hg > 0 ? C_SYNC_SOURCE : C_SYNC_TARGET),
				 abs(hg) >= 2 ? "full" : "bit-map based");
		return C_MASK;
	}

P
Philipp Reisner 已提交
3034 3035
	if (abs(hg) >= 2) {
		dev_info(DEV, "Writing the whole bitmap, full sync required after drbd_sync_handshake.\n");
3036 3037
		if (drbd_bitmap_io(mdev, &drbd_bmio_set_n_write, "set_n_write from sync_handshake",
					BM_LOCKED_SET_ALLOWED))
P
Philipp Reisner 已提交
3038 3039 3040 3041 3042 3043 3044 3045 3046 3047 3048 3049 3050 3051 3052 3053 3054 3055
			return C_MASK;
	}

	if (hg > 0) { /* become sync source. */
		rv = C_WF_BITMAP_S;
	} else if (hg < 0) { /* become sync target */
		rv = C_WF_BITMAP_T;
	} else {
		rv = C_CONNECTED;
		if (drbd_bm_total_weight(mdev)) {
			dev_info(DEV, "No resync, but %lu bits in bitmap!\n",
			     drbd_bm_total_weight(mdev));
		}
	}

	return rv;
}

3056
static enum drbd_after_sb_p convert_after_sb(enum drbd_after_sb_p peer)
P
Philipp Reisner 已提交
3057 3058
{
	/* ASB_DISCARD_REMOTE - ASB_DISCARD_LOCAL is valid */
3059 3060
	if (peer == ASB_DISCARD_REMOTE)
		return ASB_DISCARD_LOCAL;
P
Philipp Reisner 已提交
3061 3062

	/* any other things with ASB_DISCARD_REMOTE or ASB_DISCARD_LOCAL are invalid */
3063 3064
	if (peer == ASB_DISCARD_LOCAL)
		return ASB_DISCARD_REMOTE;
P
Philipp Reisner 已提交
3065 3066

	/* everything else is valid if they are equal on both sides. */
3067
	return peer;
P
Philipp Reisner 已提交
3068 3069
}

3070
static int receive_protocol(struct drbd_tconn *tconn, struct packet_info *pi)
P
Philipp Reisner 已提交
3071
{
3072
	struct p_protocol *p = pi->data;
3073 3074 3075 3076
	enum drbd_after_sb_p p_after_sb_0p, p_after_sb_1p, p_after_sb_2p;
	int p_proto, p_discard_my_data, p_two_primaries, cf;
	struct net_conf *nc, *old_net_conf, *new_net_conf = NULL;
	char integrity_alg[SHARED_SECRET_MAX] = "";
3077
	struct crypto_hash *peer_integrity_tfm = NULL;
3078
	void *int_dig_in = NULL, *int_dig_vv = NULL;
P
Philipp Reisner 已提交
3079 3080 3081 3082 3083 3084

	p_proto		= be32_to_cpu(p->protocol);
	p_after_sb_0p	= be32_to_cpu(p->after_sb_0p);
	p_after_sb_1p	= be32_to_cpu(p->after_sb_1p);
	p_after_sb_2p	= be32_to_cpu(p->after_sb_2p);
	p_two_primaries = be32_to_cpu(p->two_primaries);
3085
	cf		= be32_to_cpu(p->conn_flags);
3086
	p_discard_my_data = cf & CF_DISCARD_MY_DATA;
3087

3088 3089 3090
	if (tconn->agreed_pro_version >= 87) {
		int err;

3091
		if (pi->size > sizeof(integrity_alg))
3092
			return -EIO;
3093
		err = drbd_recv_all(tconn, integrity_alg, pi->size);
3094 3095
		if (err)
			return err;
3096 3097
		integrity_alg[SHARED_SECRET_MAX - 1] = 0;
	}
3098

3099
	if (pi->cmd != P_PROTOCOL_UPDATE) {
3100
		clear_bit(CONN_DRY_RUN, &tconn->flags);
3101

3102 3103
		if (cf & CF_DRY_RUN)
			set_bit(CONN_DRY_RUN, &tconn->flags);
3104

3105 3106
		rcu_read_lock();
		nc = rcu_dereference(tconn->net_conf);
P
Philipp Reisner 已提交
3107

3108
		if (p_proto != nc->wire_protocol) {
3109
			conn_err(tconn, "incompatible %s settings\n", "protocol");
3110 3111
			goto disconnect_rcu_unlock;
		}
3112

3113
		if (convert_after_sb(p_after_sb_0p) != nc->after_sb_0p) {
3114
			conn_err(tconn, "incompatible %s settings\n", "after-sb-0pri");
3115 3116
			goto disconnect_rcu_unlock;
		}
P
Philipp Reisner 已提交
3117

3118
		if (convert_after_sb(p_after_sb_1p) != nc->after_sb_1p) {
3119
			conn_err(tconn, "incompatible %s settings\n", "after-sb-1pri");
3120 3121
			goto disconnect_rcu_unlock;
		}
P
Philipp Reisner 已提交
3122

3123
		if (convert_after_sb(p_after_sb_2p) != nc->after_sb_2p) {
3124
			conn_err(tconn, "incompatible %s settings\n", "after-sb-2pri");
3125 3126
			goto disconnect_rcu_unlock;
		}
P
Philipp Reisner 已提交
3127

3128
		if (p_discard_my_data && nc->discard_my_data) {
3129
			conn_err(tconn, "incompatible %s settings\n", "discard-my-data");
3130 3131
			goto disconnect_rcu_unlock;
		}
P
Philipp Reisner 已提交
3132

3133
		if (p_two_primaries != nc->two_primaries) {
3134
			conn_err(tconn, "incompatible %s settings\n", "allow-two-primaries");
3135 3136
			goto disconnect_rcu_unlock;
		}
P
Philipp Reisner 已提交
3137

3138
		if (strcmp(integrity_alg, nc->integrity_alg)) {
3139
			conn_err(tconn, "incompatible %s settings\n", "data-integrity-alg");
3140 3141
			goto disconnect_rcu_unlock;
		}
P
Philipp Reisner 已提交
3142

3143
		rcu_read_unlock();
3144
	}
3145 3146 3147 3148 3149 3150 3151 3152 3153 3154 3155 3156 3157 3158 3159 3160 3161 3162 3163 3164 3165 3166 3167 3168 3169 3170 3171 3172 3173 3174 3175 3176 3177 3178 3179 3180 3181 3182 3183 3184 3185 3186 3187 3188 3189 3190 3191 3192 3193 3194 3195 3196 3197 3198 3199 3200 3201 3202 3203 3204 3205 3206 3207

	if (integrity_alg[0]) {
		int hash_size;

		/*
		 * We can only change the peer data integrity algorithm
		 * here.  Changing our own data integrity algorithm
		 * requires that we send a P_PROTOCOL_UPDATE packet at
		 * the same time; otherwise, the peer has no way to
		 * tell between which packets the algorithm should
		 * change.
		 */

		peer_integrity_tfm = crypto_alloc_hash(integrity_alg, 0, CRYPTO_ALG_ASYNC);
		if (!peer_integrity_tfm) {
			conn_err(tconn, "peer data-integrity-alg %s not supported\n",
				 integrity_alg);
			goto disconnect;
		}

		hash_size = crypto_hash_digestsize(peer_integrity_tfm);
		int_dig_in = kmalloc(hash_size, GFP_KERNEL);
		int_dig_vv = kmalloc(hash_size, GFP_KERNEL);
		if (!(int_dig_in && int_dig_vv)) {
			conn_err(tconn, "Allocation of buffers for data integrity checking failed\n");
			goto disconnect;
		}
	}

	new_net_conf = kmalloc(sizeof(struct net_conf), GFP_KERNEL);
	if (!new_net_conf) {
		conn_err(tconn, "Allocation of new net_conf failed\n");
		goto disconnect;
	}

	mutex_lock(&tconn->data.mutex);
	mutex_lock(&tconn->conf_update);
	old_net_conf = tconn->net_conf;
	*new_net_conf = *old_net_conf;

	new_net_conf->wire_protocol = p_proto;
	new_net_conf->after_sb_0p = convert_after_sb(p_after_sb_0p);
	new_net_conf->after_sb_1p = convert_after_sb(p_after_sb_1p);
	new_net_conf->after_sb_2p = convert_after_sb(p_after_sb_2p);
	new_net_conf->two_primaries = p_two_primaries;

	rcu_assign_pointer(tconn->net_conf, new_net_conf);
	mutex_unlock(&tconn->conf_update);
	mutex_unlock(&tconn->data.mutex);

	crypto_free_hash(tconn->peer_integrity_tfm);
	kfree(tconn->int_dig_in);
	kfree(tconn->int_dig_vv);
	tconn->peer_integrity_tfm = peer_integrity_tfm;
	tconn->int_dig_in = int_dig_in;
	tconn->int_dig_vv = int_dig_vv;

	if (strcmp(old_net_conf->integrity_alg, integrity_alg))
		conn_info(tconn, "peer data-integrity-alg: %s\n",
			  integrity_alg[0] ? integrity_alg : "(none)");

	synchronize_rcu();
	kfree(old_net_conf);
3208
	return 0;
P
Philipp Reisner 已提交
3209

3210 3211
disconnect_rcu_unlock:
	rcu_read_unlock();
P
Philipp Reisner 已提交
3212
disconnect:
3213
	crypto_free_hash(peer_integrity_tfm);
3214 3215
	kfree(int_dig_in);
	kfree(int_dig_vv);
3216
	conn_request_state(tconn, NS(conn, C_DISCONNECTING), CS_HARD);
3217
	return -EIO;
P
Philipp Reisner 已提交
3218 3219 3220 3221 3222 3223 3224 3225 3226 3227 3228 3229 3230 3231 3232 3233 3234 3235 3236 3237 3238 3239 3240 3241
}

/* helper function
 * input: alg name, feature name
 * return: NULL (alg name was "")
 *         ERR_PTR(error) if something goes wrong
 *         or the crypto hash ptr, if it worked out ok. */
struct crypto_hash *drbd_crypto_alloc_digest_safe(const struct drbd_conf *mdev,
		const char *alg, const char *name)
{
	struct crypto_hash *tfm;

	if (!alg[0])
		return NULL;

	tfm = crypto_alloc_hash(alg, 0, CRYPTO_ALG_ASYNC);
	if (IS_ERR(tfm)) {
		dev_err(DEV, "Can not allocate \"%s\" as %s (reason: %ld)\n",
			alg, name, PTR_ERR(tfm));
		return tfm;
	}
	return tfm;
}

3242 3243 3244 3245 3246 3247 3248 3249 3250 3251 3252 3253 3254 3255 3256 3257 3258 3259 3260 3261 3262 3263 3264 3265 3266 3267 3268 3269 3270 3271 3272 3273 3274
static int ignore_remaining_packet(struct drbd_tconn *tconn, struct packet_info *pi)
{
	void *buffer = tconn->data.rbuf;
	int size = pi->size;

	while (size) {
		int s = min_t(int, size, DRBD_SOCKET_BUFFER_SIZE);
		s = drbd_recv(tconn, buffer, s);
		if (s <= 0) {
			if (s < 0)
				return s;
			break;
		}
		size -= s;
	}
	if (size)
		return -EIO;
	return 0;
}

/*
 * config_unknown_volume  -  device configuration command for unknown volume
 *
 * When a device is added to an existing connection, the node on which the
 * device is added first will send configuration commands to its peer but the
 * peer will not know about the device yet.  It will warn and ignore these
 * commands.  Once the device is added on the second node, the second node will
 * send the same device configuration commands, but in the other direction.
 *
 * (We can also end up here if drbd is misconfigured.)
 */
static int config_unknown_volume(struct drbd_tconn *tconn, struct packet_info *pi)
{
3275 3276
	conn_warn(tconn, "%s packet received for volume %u, which is not configured locally\n",
		  cmdname(pi->cmd), pi->vnr);
3277 3278 3279 3280
	return ignore_remaining_packet(tconn, pi);
}

static int receive_SyncParam(struct drbd_tconn *tconn, struct packet_info *pi)
P
Philipp Reisner 已提交
3281
{
3282
	struct drbd_conf *mdev;
3283
	struct p_rs_param_95 *p;
P
Philipp Reisner 已提交
3284 3285 3286
	unsigned int header_size, data_size, exp_max_sz;
	struct crypto_hash *verify_tfm = NULL;
	struct crypto_hash *csums_tfm = NULL;
3287
	struct net_conf *old_net_conf, *new_net_conf = NULL;
P
Philipp Reisner 已提交
3288
	struct disk_conf *old_disk_conf = NULL, *new_disk_conf = NULL;
3289
	const int apv = tconn->agreed_pro_version;
P
Philipp Reisner 已提交
3290
	struct fifo_buffer *old_plan = NULL, *new_plan = NULL;
3291
	int fifo_size = 0;
3292
	int err;
P
Philipp Reisner 已提交
3293

3294 3295 3296 3297
	mdev = vnr_to_mdev(tconn, pi->vnr);
	if (!mdev)
		return config_unknown_volume(tconn, pi);

P
Philipp Reisner 已提交
3298 3299 3300
	exp_max_sz  = apv <= 87 ? sizeof(struct p_rs_param)
		    : apv == 88 ? sizeof(struct p_rs_param)
					+ SHARED_SECRET_MAX
3301 3302
		    : apv <= 94 ? sizeof(struct p_rs_param_89)
		    : /* apv >= 95 */ sizeof(struct p_rs_param_95);
P
Philipp Reisner 已提交
3303

3304
	if (pi->size > exp_max_sz) {
P
Philipp Reisner 已提交
3305
		dev_err(DEV, "SyncParam packet too long: received %u, expected <= %u bytes\n",
3306
		    pi->size, exp_max_sz);
3307
		return -EIO;
P
Philipp Reisner 已提交
3308 3309 3310
	}

	if (apv <= 88) {
3311
		header_size = sizeof(struct p_rs_param);
3312
		data_size = pi->size - header_size;
3313
	} else if (apv <= 94) {
3314
		header_size = sizeof(struct p_rs_param_89);
3315
		data_size = pi->size - header_size;
P
Philipp Reisner 已提交
3316
		D_ASSERT(data_size == 0);
3317
	} else {
3318
		header_size = sizeof(struct p_rs_param_95);
3319
		data_size = pi->size - header_size;
P
Philipp Reisner 已提交
3320 3321 3322 3323
		D_ASSERT(data_size == 0);
	}

	/* initialize verify_alg and csums_alg */
3324
	p = pi->data;
P
Philipp Reisner 已提交
3325 3326
	memset(p->verify_alg, 0, 2 * SHARED_SECRET_MAX);

3327
	err = drbd_recv_all(mdev->tconn, p, header_size);
3328 3329
	if (err)
		return err;
P
Philipp Reisner 已提交
3330

P
Philipp Reisner 已提交
3331 3332
	mutex_lock(&mdev->tconn->conf_update);
	old_net_conf = mdev->tconn->net_conf;
P
Philipp Reisner 已提交
3333 3334 3335 3336 3337 3338 3339 3340
	if (get_ldev(mdev)) {
		new_disk_conf = kzalloc(sizeof(struct disk_conf), GFP_KERNEL);
		if (!new_disk_conf) {
			put_ldev(mdev);
			mutex_unlock(&mdev->tconn->conf_update);
			dev_err(DEV, "Allocation of new disk_conf failed\n");
			return -ENOMEM;
		}
P
Philipp Reisner 已提交
3341

P
Philipp Reisner 已提交
3342 3343 3344
		old_disk_conf = mdev->ldev->disk_conf;
		*new_disk_conf = *old_disk_conf;

3345
		new_disk_conf->resync_rate = be32_to_cpu(p->resync_rate);
P
Philipp Reisner 已提交
3346
	}
P
Philipp Reisner 已提交
3347

P
Philipp Reisner 已提交
3348 3349
	if (apv >= 88) {
		if (apv == 88) {
3350 3351 3352 3353
			if (data_size > SHARED_SECRET_MAX || data_size == 0) {
				dev_err(DEV, "verify-alg of wrong size, "
					"peer wants %u, accepting only up to %u byte\n",
					data_size, SHARED_SECRET_MAX);
P
Philipp Reisner 已提交
3354 3355
				err = -EIO;
				goto reconnect;
P
Philipp Reisner 已提交
3356 3357
			}

3358
			err = drbd_recv_all(mdev->tconn, p->verify_alg, data_size);
P
Philipp Reisner 已提交
3359 3360
			if (err)
				goto reconnect;
P
Philipp Reisner 已提交
3361 3362 3363 3364 3365 3366 3367 3368 3369 3370 3371 3372 3373 3374
			/* we expect NUL terminated string */
			/* but just in case someone tries to be evil */
			D_ASSERT(p->verify_alg[data_size-1] == 0);
			p->verify_alg[data_size-1] = 0;

		} else /* apv >= 89 */ {
			/* we still expect NUL terminated strings */
			/* but just in case someone tries to be evil */
			D_ASSERT(p->verify_alg[SHARED_SECRET_MAX-1] == 0);
			D_ASSERT(p->csums_alg[SHARED_SECRET_MAX-1] == 0);
			p->verify_alg[SHARED_SECRET_MAX-1] = 0;
			p->csums_alg[SHARED_SECRET_MAX-1] = 0;
		}

3375
		if (strcmp(old_net_conf->verify_alg, p->verify_alg)) {
P
Philipp Reisner 已提交
3376 3377
			if (mdev->state.conn == C_WF_REPORT_PARAMS) {
				dev_err(DEV, "Different verify-alg settings. me=\"%s\" peer=\"%s\"\n",
3378
				    old_net_conf->verify_alg, p->verify_alg);
P
Philipp Reisner 已提交
3379 3380 3381 3382 3383 3384 3385 3386 3387 3388
				goto disconnect;
			}
			verify_tfm = drbd_crypto_alloc_digest_safe(mdev,
					p->verify_alg, "verify-alg");
			if (IS_ERR(verify_tfm)) {
				verify_tfm = NULL;
				goto disconnect;
			}
		}

3389
		if (apv >= 89 && strcmp(old_net_conf->csums_alg, p->csums_alg)) {
P
Philipp Reisner 已提交
3390 3391
			if (mdev->state.conn == C_WF_REPORT_PARAMS) {
				dev_err(DEV, "Different csums-alg settings. me=\"%s\" peer=\"%s\"\n",
3392
				    old_net_conf->csums_alg, p->csums_alg);
P
Philipp Reisner 已提交
3393 3394 3395 3396 3397 3398 3399 3400 3401 3402
				goto disconnect;
			}
			csums_tfm = drbd_crypto_alloc_digest_safe(mdev,
					p->csums_alg, "csums-alg");
			if (IS_ERR(csums_tfm)) {
				csums_tfm = NULL;
				goto disconnect;
			}
		}

P
Philipp Reisner 已提交
3403
		if (apv > 94 && new_disk_conf) {
P
Philipp Reisner 已提交
3404 3405 3406 3407
			new_disk_conf->c_plan_ahead = be32_to_cpu(p->c_plan_ahead);
			new_disk_conf->c_delay_target = be32_to_cpu(p->c_delay_target);
			new_disk_conf->c_fill_target = be32_to_cpu(p->c_fill_target);
			new_disk_conf->c_max_rate = be32_to_cpu(p->c_max_rate);
3408

P
Philipp Reisner 已提交
3409
			fifo_size = (new_disk_conf->c_plan_ahead * 10 * SLEEP_TIME) / HZ;
3410
			if (fifo_size != mdev->rs_plan_s->size) {
P
Philipp Reisner 已提交
3411 3412
				new_plan = fifo_alloc(fifo_size);
				if (!new_plan) {
3413
					dev_err(DEV, "kmalloc of fifo_buffer failed");
3414
					put_ldev(mdev);
3415 3416 3417
					goto disconnect;
				}
			}
3418
		}
P
Philipp Reisner 已提交
3419

3420
		if (verify_tfm || csums_tfm) {
3421 3422
			new_net_conf = kzalloc(sizeof(struct net_conf), GFP_KERNEL);
			if (!new_net_conf) {
3423 3424 3425 3426
				dev_err(DEV, "Allocation of new net_conf failed\n");
				goto disconnect;
			}

3427
			*new_net_conf = *old_net_conf;
3428 3429

			if (verify_tfm) {
3430 3431
				strcpy(new_net_conf->verify_alg, p->verify_alg);
				new_net_conf->verify_alg_len = strlen(p->verify_alg) + 1;
3432 3433 3434 3435 3436
				crypto_free_hash(mdev->tconn->verify_tfm);
				mdev->tconn->verify_tfm = verify_tfm;
				dev_info(DEV, "using verify-alg: \"%s\"\n", p->verify_alg);
			}
			if (csums_tfm) {
3437 3438
				strcpy(new_net_conf->csums_alg, p->csums_alg);
				new_net_conf->csums_alg_len = strlen(p->csums_alg) + 1;
3439 3440 3441 3442
				crypto_free_hash(mdev->tconn->csums_tfm);
				mdev->tconn->csums_tfm = csums_tfm;
				dev_info(DEV, "using csums-alg: \"%s\"\n", p->csums_alg);
			}
3443
			rcu_assign_pointer(tconn->net_conf, new_net_conf);
P
Philipp Reisner 已提交
3444
		}
P
Philipp Reisner 已提交
3445
	}
3446

P
Philipp Reisner 已提交
3447 3448 3449 3450 3451 3452 3453 3454
	if (new_disk_conf) {
		rcu_assign_pointer(mdev->ldev->disk_conf, new_disk_conf);
		put_ldev(mdev);
	}

	if (new_plan) {
		old_plan = mdev->rs_plan_s;
		rcu_assign_pointer(mdev->rs_plan_s, new_plan);
P
Philipp Reisner 已提交
3455
	}
P
Philipp Reisner 已提交
3456 3457 3458 3459 3460 3461

	mutex_unlock(&mdev->tconn->conf_update);
	synchronize_rcu();
	if (new_net_conf)
		kfree(old_net_conf);
	kfree(old_disk_conf);
P
Philipp Reisner 已提交
3462
	kfree(old_plan);
P
Philipp Reisner 已提交
3463

3464
	return 0;
P
Philipp Reisner 已提交
3465

P
Philipp Reisner 已提交
3466 3467 3468 3469 3470 3471 3472 3473
reconnect:
	if (new_disk_conf) {
		put_ldev(mdev);
		kfree(new_disk_conf);
	}
	mutex_unlock(&mdev->tconn->conf_update);
	return -EIO;

P
Philipp Reisner 已提交
3474
disconnect:
P
Philipp Reisner 已提交
3475 3476 3477 3478 3479
	kfree(new_plan);
	if (new_disk_conf) {
		put_ldev(mdev);
		kfree(new_disk_conf);
	}
3480
	mutex_unlock(&mdev->tconn->conf_update);
P
Philipp Reisner 已提交
3481 3482 3483 3484 3485
	/* just for completeness: actually not needed,
	 * as this is not reached if csums_tfm was ok. */
	crypto_free_hash(csums_tfm);
	/* but free the verify_tfm again, if csums_tfm did not work out */
	crypto_free_hash(verify_tfm);
3486
	conn_request_state(mdev->tconn, NS(conn, C_DISCONNECTING), CS_HARD);
3487
	return -EIO;
P
Philipp Reisner 已提交
3488 3489 3490 3491 3492 3493 3494 3495 3496 3497 3498 3499 3500 3501 3502
}

/* warn if the arguments differ by more than 12.5% */
static void warn_if_differ_considerably(struct drbd_conf *mdev,
	const char *s, sector_t a, sector_t b)
{
	sector_t d;
	if (a == 0 || b == 0)
		return;
	d = (a > b) ? (a - b) : (b - a);
	if (d > (a>>3) || d > (b>>3))
		dev_warn(DEV, "Considerable difference in %s: %llus vs. %llus\n", s,
		     (unsigned long long)a, (unsigned long long)b);
}

3503
static int receive_sizes(struct drbd_tconn *tconn, struct packet_info *pi)
P
Philipp Reisner 已提交
3504
{
3505
	struct drbd_conf *mdev;
3506
	struct p_sizes *p = pi->data;
P
Philipp Reisner 已提交
3507 3508 3509
	enum determine_dev_size dd = unchanged;
	sector_t p_size, p_usize, my_usize;
	int ldsc = 0; /* local disk size changed */
3510
	enum dds_flags ddsf;
P
Philipp Reisner 已提交
3511

3512 3513 3514 3515
	mdev = vnr_to_mdev(tconn, pi->vnr);
	if (!mdev)
		return config_unknown_volume(tconn, pi);

P
Philipp Reisner 已提交
3516 3517 3518 3519 3520 3521 3522 3523
	p_size = be64_to_cpu(p->d_size);
	p_usize = be64_to_cpu(p->u_size);

	/* just store the peer's disk size for now.
	 * we still need to figure out whether we accept that. */
	mdev->p_size = p_size;

	if (get_ldev(mdev)) {
P
Philipp Reisner 已提交
3524 3525 3526 3527
		rcu_read_lock();
		my_usize = rcu_dereference(mdev->ldev->disk_conf)->disk_size;
		rcu_read_unlock();

P
Philipp Reisner 已提交
3528 3529 3530
		warn_if_differ_considerably(mdev, "lower level device sizes",
			   p_size, drbd_get_max_capacity(mdev->ldev));
		warn_if_differ_considerably(mdev, "user requested size",
P
Philipp Reisner 已提交
3531
					    p_usize, my_usize);
P
Philipp Reisner 已提交
3532 3533 3534 3535

		/* if this is the first connect, or an otherwise expected
		 * param exchange, choose the minimum */
		if (mdev->state.conn == C_WF_REPORT_PARAMS)
P
Philipp Reisner 已提交
3536
			p_usize = min_not_zero(my_usize, p_usize);
P
Philipp Reisner 已提交
3537 3538 3539

		/* Never shrink a device with usable data during connect.
		   But allow online shrinking if we are connected. */
3540
		if (drbd_new_dev_size(mdev, mdev->ldev, p_usize, 0) <
P
Philipp Reisner 已提交
3541 3542 3543
		    drbd_get_capacity(mdev->this_bdev) &&
		    mdev->state.disk >= D_OUTDATED &&
		    mdev->state.conn < C_CONNECTED) {
P
Philipp Reisner 已提交
3544
			dev_err(DEV, "The peer's disk size is too small!\n");
3545
			conn_request_state(mdev->tconn, NS(conn, C_DISCONNECTING), CS_HARD);
P
Philipp Reisner 已提交
3546
			put_ldev(mdev);
3547
			return -EIO;
P
Philipp Reisner 已提交
3548
		}
P
Philipp Reisner 已提交
3549 3550 3551 3552 3553 3554 3555 3556 3557 3558 3559 3560 3561 3562 3563 3564 3565 3566 3567 3568 3569 3570 3571 3572 3573

		if (my_usize != p_usize) {
			struct disk_conf *old_disk_conf, *new_disk_conf = NULL;

			new_disk_conf = kzalloc(sizeof(struct disk_conf), GFP_KERNEL);
			if (!new_disk_conf) {
				dev_err(DEV, "Allocation of new disk_conf failed\n");
				put_ldev(mdev);
				return -ENOMEM;
			}

			mutex_lock(&mdev->tconn->conf_update);
			old_disk_conf = mdev->ldev->disk_conf;
			*new_disk_conf = *old_disk_conf;
			new_disk_conf->disk_size = p_usize;

			rcu_assign_pointer(mdev->ldev->disk_conf, new_disk_conf);
			mutex_unlock(&mdev->tconn->conf_update);
			synchronize_rcu();
			kfree(old_disk_conf);

			dev_info(DEV, "Peer sets u_size to %lu sectors\n",
				 (unsigned long)my_usize);
		}

P
Philipp Reisner 已提交
3574 3575 3576
		put_ldev(mdev);
	}

3577
	ddsf = be16_to_cpu(p->dds_flags);
P
Philipp Reisner 已提交
3578
	if (get_ldev(mdev)) {
B
Bart Van Assche 已提交
3579
		dd = drbd_determine_dev_size(mdev, ddsf);
P
Philipp Reisner 已提交
3580 3581
		put_ldev(mdev);
		if (dd == dev_size_error)
3582
			return -EIO;
P
Philipp Reisner 已提交
3583 3584 3585 3586 3587 3588
		drbd_md_sync(mdev);
	} else {
		/* I am diskless, need to accept the peer's size. */
		drbd_set_my_capacity(mdev, p_size);
	}

3589 3590 3591
	mdev->peer_max_bio_size = be32_to_cpu(p->max_bio_size);
	drbd_reconsider_max_bio_size(mdev);

P
Philipp Reisner 已提交
3592 3593 3594 3595 3596 3597 3598 3599 3600 3601 3602 3603 3604 3605
	if (get_ldev(mdev)) {
		if (mdev->ldev->known_size != drbd_get_capacity(mdev->ldev->backing_bdev)) {
			mdev->ldev->known_size = drbd_get_capacity(mdev->ldev->backing_bdev);
			ldsc = 1;
		}

		put_ldev(mdev);
	}

	if (mdev->state.conn > C_WF_REPORT_PARAMS) {
		if (be64_to_cpu(p->c_size) !=
		    drbd_get_capacity(mdev->this_bdev) || ldsc) {
			/* we have different sizes, probably peer
			 * needs to know my new size... */
3606
			drbd_send_sizes(mdev, 0, ddsf);
P
Philipp Reisner 已提交
3607 3608 3609 3610
		}
		if (test_and_clear_bit(RESIZE_PENDING, &mdev->flags) ||
		    (dd == grew && mdev->state.conn == C_CONNECTED)) {
			if (mdev->state.pdsk >= D_INCONSISTENT &&
3611 3612 3613 3614 3615 3616
			    mdev->state.disk >= D_INCONSISTENT) {
				if (ddsf & DDSF_NO_RESYNC)
					dev_info(DEV, "Resync of new storage suppressed with --assume-clean\n");
				else
					resync_after_online_grow(mdev);
			} else
P
Philipp Reisner 已提交
3617 3618 3619 3620
				set_bit(RESYNC_AFTER_NEG, &mdev->flags);
		}
	}

3621
	return 0;
P
Philipp Reisner 已提交
3622 3623
}

3624
static int receive_uuids(struct drbd_tconn *tconn, struct packet_info *pi)
P
Philipp Reisner 已提交
3625
{
3626
	struct drbd_conf *mdev;
3627
	struct p_uuids *p = pi->data;
P
Philipp Reisner 已提交
3628
	u64 *p_uuid;
3629
	int i, updated_uuids = 0;
P
Philipp Reisner 已提交
3630

3631 3632 3633 3634
	mdev = vnr_to_mdev(tconn, pi->vnr);
	if (!mdev)
		return config_unknown_volume(tconn, pi);

P
Philipp Reisner 已提交
3635 3636 3637 3638 3639 3640 3641 3642 3643 3644 3645 3646 3647 3648
	p_uuid = kmalloc(sizeof(u64)*UI_EXTENDED_SIZE, GFP_NOIO);

	for (i = UI_CURRENT; i < UI_EXTENDED_SIZE; i++)
		p_uuid[i] = be64_to_cpu(p->uuid[i]);

	kfree(mdev->p_uuid);
	mdev->p_uuid = p_uuid;

	if (mdev->state.conn < C_CONNECTED &&
	    mdev->state.disk < D_INCONSISTENT &&
	    mdev->state.role == R_PRIMARY &&
	    (mdev->ed_uuid & ~((u64)1)) != (p_uuid[UI_CURRENT] & ~((u64)1))) {
		dev_err(DEV, "Can only connect to data with current UUID=%016llX\n",
		    (unsigned long long)mdev->ed_uuid);
3649
		conn_request_state(mdev->tconn, NS(conn, C_DISCONNECTING), CS_HARD);
3650
		return -EIO;
P
Philipp Reisner 已提交
3651 3652 3653 3654 3655
	}

	if (get_ldev(mdev)) {
		int skip_initial_sync =
			mdev->state.conn == C_CONNECTED &&
3656
			mdev->tconn->agreed_pro_version >= 90 &&
P
Philipp Reisner 已提交
3657 3658 3659 3660 3661
			mdev->ldev->md.uuid[UI_CURRENT] == UUID_JUST_CREATED &&
			(p_uuid[UI_FLAGS] & 8);
		if (skip_initial_sync) {
			dev_info(DEV, "Accepted new current UUID, preparing to skip initial sync\n");
			drbd_bitmap_io(mdev, &drbd_bmio_clear_n_write,
3662 3663
					"clear_n_write from receive_uuids",
					BM_LOCKED_TEST_ALLOWED);
P
Philipp Reisner 已提交
3664 3665 3666 3667 3668
			_drbd_uuid_set(mdev, UI_CURRENT, p_uuid[UI_CURRENT]);
			_drbd_uuid_set(mdev, UI_BITMAP, 0);
			_drbd_set_state(_NS2(mdev, disk, D_UP_TO_DATE, pdsk, D_UP_TO_DATE),
					CS_VERBOSE, NULL);
			drbd_md_sync(mdev);
3669
			updated_uuids = 1;
P
Philipp Reisner 已提交
3670 3671
		}
		put_ldev(mdev);
3672 3673 3674 3675
	} else if (mdev->state.disk < D_INCONSISTENT &&
		   mdev->state.role == R_PRIMARY) {
		/* I am a diskless primary, the peer just created a new current UUID
		   for me. */
3676
		updated_uuids = drbd_set_ed_uuid(mdev, p_uuid[UI_CURRENT]);
P
Philipp Reisner 已提交
3677 3678 3679 3680 3681 3682
	}

	/* Before we test for the disk state, we should wait until an eventually
	   ongoing cluster wide state change is finished. That is important if
	   we are primary and are detaching from our disk. We need to see the
	   new disk state... */
3683 3684
	mutex_lock(mdev->state_mutex);
	mutex_unlock(mdev->state_mutex);
P
Philipp Reisner 已提交
3685
	if (mdev->state.conn >= C_CONNECTED && mdev->state.disk < D_INCONSISTENT)
3686 3687 3688 3689
		updated_uuids |= drbd_set_ed_uuid(mdev, p_uuid[UI_CURRENT]);

	if (updated_uuids)
		drbd_print_uuids(mdev, "receiver updated UUIDs to");
P
Philipp Reisner 已提交
3690

3691
	return 0;
P
Philipp Reisner 已提交
3692 3693 3694 3695 3696 3697 3698 3699 3700 3701 3702
}

/**
 * convert_state() - Converts the peer's view of the cluster state to our point of view
 * @ps:		The state as seen by the peer.
 */
static union drbd_state convert_state(union drbd_state ps)
{
	union drbd_state ms;

	static enum drbd_conns c_tab[] = {
3703
		[C_WF_REPORT_PARAMS] = C_WF_REPORT_PARAMS,
P
Philipp Reisner 已提交
3704 3705 3706 3707 3708 3709 3710 3711 3712 3713 3714 3715 3716 3717 3718 3719 3720 3721 3722 3723 3724
		[C_CONNECTED] = C_CONNECTED,

		[C_STARTING_SYNC_S] = C_STARTING_SYNC_T,
		[C_STARTING_SYNC_T] = C_STARTING_SYNC_S,
		[C_DISCONNECTING] = C_TEAR_DOWN, /* C_NETWORK_FAILURE, */
		[C_VERIFY_S]       = C_VERIFY_T,
		[C_MASK]   = C_MASK,
	};

	ms.i = ps.i;

	ms.conn = c_tab[ps.conn];
	ms.peer = ps.role;
	ms.role = ps.peer;
	ms.pdsk = ps.disk;
	ms.disk = ps.pdsk;
	ms.peer_isp = (ps.aftr_isp | ps.user_isp);

	return ms;
}

3725
static int receive_req_state(struct drbd_tconn *tconn, struct packet_info *pi)
P
Philipp Reisner 已提交
3726
{
3727
	struct drbd_conf *mdev;
3728
	struct p_req_state *p = pi->data;
P
Philipp Reisner 已提交
3729
	union drbd_state mask, val;
3730
	enum drbd_state_rv rv;
P
Philipp Reisner 已提交
3731

3732 3733 3734 3735
	mdev = vnr_to_mdev(tconn, pi->vnr);
	if (!mdev)
		return -EIO;

P
Philipp Reisner 已提交
3736 3737 3738
	mask.i = be32_to_cpu(p->mask);
	val.i = be32_to_cpu(p->val);

3739
	if (test_bit(DISCARD_CONCURRENT, &mdev->tconn->flags) &&
3740
	    mutex_is_locked(mdev->state_mutex)) {
P
Philipp Reisner 已提交
3741
		drbd_send_sr_reply(mdev, SS_CONCURRENT_ST_CHG);
3742
		return 0;
P
Philipp Reisner 已提交
3743 3744 3745 3746 3747
	}

	mask = convert_state(mask);
	val = convert_state(val);

3748 3749
	rv = drbd_change_state(mdev, CS_VERBOSE, mask, val);
	drbd_send_sr_reply(mdev, rv);
P
Philipp Reisner 已提交
3750 3751 3752

	drbd_md_sync(mdev);

3753
	return 0;
P
Philipp Reisner 已提交
3754 3755
}

3756
static int receive_req_conn_state(struct drbd_tconn *tconn, struct packet_info *pi)
3757
{
3758
	struct p_req_state *p = pi->data;
3759 3760 3761 3762 3763 3764 3765 3766 3767
	union drbd_state mask, val;
	enum drbd_state_rv rv;

	mask.i = be32_to_cpu(p->mask);
	val.i = be32_to_cpu(p->val);

	if (test_bit(DISCARD_CONCURRENT, &tconn->flags) &&
	    mutex_is_locked(&tconn->cstate_mutex)) {
		conn_send_sr_reply(tconn, SS_CONCURRENT_ST_CHG);
3768
		return 0;
3769 3770 3771 3772 3773
	}

	mask = convert_state(mask);
	val = convert_state(val);

3774
	rv = conn_request_state(tconn, mask, val, CS_VERBOSE | CS_LOCAL_ONLY | CS_IGN_OUTD_FAIL);
3775 3776
	conn_send_sr_reply(tconn, rv);

3777
	return 0;
3778 3779
}

3780
static int receive_state(struct drbd_tconn *tconn, struct packet_info *pi)
P
Philipp Reisner 已提交
3781
{
3782
	struct drbd_conf *mdev;
3783
	struct p_state *p = pi->data;
3784
	union drbd_state os, ns, peer_state;
P
Philipp Reisner 已提交
3785
	enum drbd_disk_state real_peer_disk;
3786
	enum chg_state_flags cs_flags;
P
Philipp Reisner 已提交
3787 3788
	int rv;

3789 3790 3791 3792
	mdev = vnr_to_mdev(tconn, pi->vnr);
	if (!mdev)
		return config_unknown_volume(tconn, pi);

P
Philipp Reisner 已提交
3793 3794 3795 3796 3797 3798 3799 3800
	peer_state.i = be32_to_cpu(p->state);

	real_peer_disk = peer_state.disk;
	if (peer_state.disk == D_NEGOTIATING) {
		real_peer_disk = mdev->p_uuid[UI_FLAGS] & 4 ? D_INCONSISTENT : D_CONSISTENT;
		dev_info(DEV, "real peer disk state = %s\n", drbd_disk_str(real_peer_disk));
	}

3801
	spin_lock_irq(&mdev->tconn->req_lock);
P
Philipp Reisner 已提交
3802
 retry:
3803
	os = ns = drbd_read_state(mdev);
3804
	spin_unlock_irq(&mdev->tconn->req_lock);
P
Philipp Reisner 已提交
3805

3806 3807 3808 3809 3810 3811
	/* If some other part of the code (asender thread, timeout)
	 * already decided to close the connection again,
	 * we must not "re-establish" it here. */
	if (os.conn <= C_TEAR_DOWN)
		return false;

P
Philipp Reisner 已提交
3812 3813 3814 3815 3816 3817 3818 3819
	/* If this is the "end of sync" confirmation, usually the peer disk
	 * transitions from D_INCONSISTENT to D_UP_TO_DATE. For empty (0 bits
	 * set) resync started in PausedSyncT, or if the timing of pause-/
	 * unpause-sync events has been "just right", the peer disk may
	 * transition from D_CONSISTENT to D_UP_TO_DATE as well.
	 */
	if ((os.pdsk == D_INCONSISTENT || os.pdsk == D_CONSISTENT) &&
	    real_peer_disk == D_UP_TO_DATE &&
3820 3821 3822 3823 3824 3825 3826 3827 3828 3829 3830 3831 3832 3833 3834 3835 3836 3837
	    os.conn > C_CONNECTED && os.disk == D_UP_TO_DATE) {
		/* If we are (becoming) SyncSource, but peer is still in sync
		 * preparation, ignore its uptodate-ness to avoid flapping, it
		 * will change to inconsistent once the peer reaches active
		 * syncing states.
		 * It may have changed syncer-paused flags, however, so we
		 * cannot ignore this completely. */
		if (peer_state.conn > C_CONNECTED &&
		    peer_state.conn < C_SYNC_SOURCE)
			real_peer_disk = D_INCONSISTENT;

		/* if peer_state changes to connected at the same time,
		 * it explicitly notifies us that it finished resync.
		 * Maybe we should finish it up, too? */
		else if (os.conn >= C_SYNC_SOURCE &&
			 peer_state.conn == C_CONNECTED) {
			if (drbd_bm_total_weight(mdev) <= mdev->rs_failed)
				drbd_resync_finished(mdev);
3838
			return 0;
3839 3840 3841 3842 3843 3844 3845 3846 3847 3848 3849 3850
		}
	}

	/* peer says his disk is inconsistent, while we think it is uptodate,
	 * and this happens while the peer still thinks we have a sync going on,
	 * but we think we are already done with the sync.
	 * We ignore this to avoid flapping pdsk.
	 * This should not happen, if the peer is a recent version of drbd. */
	if (os.pdsk == D_UP_TO_DATE && real_peer_disk == D_INCONSISTENT &&
	    os.conn == C_CONNECTED && peer_state.conn > C_SYNC_SOURCE)
		real_peer_disk = D_UP_TO_DATE;

3851 3852
	if (ns.conn == C_WF_REPORT_PARAMS)
		ns.conn = C_CONNECTED;
P
Philipp Reisner 已提交
3853

3854 3855 3856
	if (peer_state.conn == C_AHEAD)
		ns.conn = C_BEHIND;

P
Philipp Reisner 已提交
3857 3858 3859 3860 3861
	if (mdev->p_uuid && peer_state.disk >= D_NEGOTIATING &&
	    get_ldev_if_state(mdev, D_NEGOTIATING)) {
		int cr; /* consider resync */

		/* if we established a new connection */
3862
		cr  = (os.conn < C_CONNECTED);
P
Philipp Reisner 已提交
3863 3864
		/* if we had an established connection
		 * and one of the nodes newly attaches a disk */
3865
		cr |= (os.conn == C_CONNECTED &&
P
Philipp Reisner 已提交
3866
		       (peer_state.disk == D_NEGOTIATING ||
3867
			os.disk == D_NEGOTIATING));
P
Philipp Reisner 已提交
3868 3869 3870 3871 3872
		/* if we have both been inconsistent, and the peer has been
		 * forced to be UpToDate with --overwrite-data */
		cr |= test_bit(CONSIDER_RESYNC, &mdev->flags);
		/* if we had been plain connected, and the admin requested to
		 * start a sync by "invalidate" or "invalidate-remote" */
3873
		cr |= (os.conn == C_CONNECTED &&
P
Philipp Reisner 已提交
3874 3875 3876 3877
				(peer_state.conn >= C_STARTING_SYNC_S &&
				 peer_state.conn <= C_WF_BITMAP_T));

		if (cr)
3878
			ns.conn = drbd_sync_handshake(mdev, peer_state.role, real_peer_disk);
P
Philipp Reisner 已提交
3879 3880

		put_ldev(mdev);
3881 3882
		if (ns.conn == C_MASK) {
			ns.conn = C_CONNECTED;
P
Philipp Reisner 已提交
3883
			if (mdev->state.disk == D_NEGOTIATING) {
3884
				drbd_force_state(mdev, NS(disk, D_FAILED));
P
Philipp Reisner 已提交
3885 3886 3887
			} else if (peer_state.disk == D_NEGOTIATING) {
				dev_err(DEV, "Disk attach process on the peer node was aborted.\n");
				peer_state.disk = D_DISKLESS;
3888
				real_peer_disk = D_DISKLESS;
P
Philipp Reisner 已提交
3889
			} else {
3890
				if (test_and_clear_bit(CONN_DRY_RUN, &mdev->tconn->flags))
3891
					return -EIO;
3892
				D_ASSERT(os.conn == C_WF_REPORT_PARAMS);
3893
				conn_request_state(mdev->tconn, NS(conn, C_DISCONNECTING), CS_HARD);
3894
				return -EIO;
P
Philipp Reisner 已提交
3895 3896 3897 3898
			}
		}
	}

3899
	spin_lock_irq(&mdev->tconn->req_lock);
3900
	if (os.i != drbd_read_state(mdev).i)
P
Philipp Reisner 已提交
3901 3902 3903 3904 3905
		goto retry;
	clear_bit(CONSIDER_RESYNC, &mdev->flags);
	ns.peer = peer_state.role;
	ns.pdsk = real_peer_disk;
	ns.peer_isp = (peer_state.aftr_isp | peer_state.user_isp);
3906
	if ((ns.conn == C_CONNECTED || ns.conn == C_WF_BITMAP_S) && ns.disk == D_NEGOTIATING)
P
Philipp Reisner 已提交
3907
		ns.disk = mdev->new_state_tmp.disk;
3908
	cs_flags = CS_VERBOSE + (os.conn < C_CONNECTED && ns.conn >= C_CONNECTED ? 0 : CS_HARD);
3909
	if (ns.pdsk == D_CONSISTENT && drbd_suspended(mdev) && ns.conn == C_CONNECTED && os.conn < C_CONNECTED &&
3910
	    test_bit(NEW_CUR_UUID, &mdev->flags)) {
3911
		/* Do not allow tl_restart(RESEND) for a rebooted peer. We can only allow this
3912
		   for temporal network outages! */
3913
		spin_unlock_irq(&mdev->tconn->req_lock);
3914
		dev_err(DEV, "Aborting Connect, can not thaw IO with an only Consistent peer\n");
3915
		tl_clear(mdev->tconn);
3916 3917
		drbd_uuid_new_current(mdev);
		clear_bit(NEW_CUR_UUID, &mdev->flags);
3918
		conn_request_state(mdev->tconn, NS2(conn, C_PROTOCOL_ERROR, susp, 0), CS_HARD);
3919
		return -EIO;
3920
	}
3921
	rv = _drbd_set_state(mdev, ns, cs_flags, NULL);
3922
	ns = drbd_read_state(mdev);
3923
	spin_unlock_irq(&mdev->tconn->req_lock);
P
Philipp Reisner 已提交
3924 3925

	if (rv < SS_SUCCESS) {
3926
		conn_request_state(mdev->tconn, NS(conn, C_DISCONNECTING), CS_HARD);
3927
		return -EIO;
P
Philipp Reisner 已提交
3928 3929
	}

3930 3931
	if (os.conn > C_WF_REPORT_PARAMS) {
		if (ns.conn > C_CONNECTED && peer_state.conn <= C_CONNECTED &&
P
Philipp Reisner 已提交
3932 3933 3934 3935 3936
		    peer_state.disk != D_NEGOTIATING ) {
			/* we want resync, peer has not yet decided to sync... */
			/* Nowadays only used when forcing a node into primary role and
			   setting its disk to UpToDate with that */
			drbd_send_uuids(mdev);
3937
			drbd_send_current_state(mdev);
P
Philipp Reisner 已提交
3938 3939 3940
		}
	}

3941
	clear_bit(DISCARD_MY_DATA, &mdev->flags);
P
Philipp Reisner 已提交
3942 3943 3944

	drbd_md_sync(mdev); /* update connected indicator, la_size, ... */

3945
	return 0;
P
Philipp Reisner 已提交
3946 3947
}

3948
static int receive_sync_uuid(struct drbd_tconn *tconn, struct packet_info *pi)
P
Philipp Reisner 已提交
3949
{
3950
	struct drbd_conf *mdev;
3951
	struct p_rs_uuid *p = pi->data;
3952 3953 3954 3955

	mdev = vnr_to_mdev(tconn, pi->vnr);
	if (!mdev)
		return -EIO;
P
Philipp Reisner 已提交
3956 3957 3958

	wait_event(mdev->misc_wait,
		   mdev->state.conn == C_WF_SYNC_UUID ||
3959
		   mdev->state.conn == C_BEHIND ||
P
Philipp Reisner 已提交
3960 3961 3962 3963 3964 3965 3966 3967 3968 3969 3970
		   mdev->state.conn < C_CONNECTED ||
		   mdev->state.disk < D_NEGOTIATING);

	/* D_ASSERT( mdev->state.conn == C_WF_SYNC_UUID ); */

	/* Here the _drbd_uuid_ functions are right, current should
	   _not_ be rotated into the history */
	if (get_ldev_if_state(mdev, D_NEGOTIATING)) {
		_drbd_uuid_set(mdev, UI_CURRENT, be64_to_cpu(p->uuid));
		_drbd_uuid_set(mdev, UI_BITMAP, 0UL);

3971
		drbd_print_uuids(mdev, "updated sync uuid");
P
Philipp Reisner 已提交
3972 3973 3974 3975 3976 3977
		drbd_start_resync(mdev, C_SYNC_TARGET);

		put_ldev(mdev);
	} else
		dev_err(DEV, "Ignoring SyncUUID packet!\n");

3978
	return 0;
P
Philipp Reisner 已提交
3979 3980
}

3981 3982 3983 3984 3985 3986 3987
/**
 * receive_bitmap_plain
 *
 * Return 0 when done, 1 when another iteration is needed, and a negative error
 * code upon failure.
 */
static int
3988
receive_bitmap_plain(struct drbd_conf *mdev, unsigned int size,
3989
		     unsigned long *p, struct bm_xfer_ctx *c)
P
Philipp Reisner 已提交
3990
{
3991 3992
	unsigned int data_size = DRBD_SOCKET_BUFFER_SIZE -
				 drbd_header_size(mdev->tconn);
3993
	unsigned int num_words = min_t(size_t, data_size / sizeof(*p),
3994
				       c->bm_words - c->word_offset);
3995
	unsigned int want = num_words * sizeof(*p);
3996
	int err;
P
Philipp Reisner 已提交
3997

3998 3999
	if (want != size) {
		dev_err(DEV, "%s:want (%u) != size (%u)\n", __func__, want, size);
4000
		return -EIO;
P
Philipp Reisner 已提交
4001 4002
	}
	if (want == 0)
4003
		return 0;
4004
	err = drbd_recv_all(mdev->tconn, p, want);
4005
	if (err)
4006
		return err;
P
Philipp Reisner 已提交
4007

4008
	drbd_bm_merge_lel(mdev, c->word_offset, num_words, p);
P
Philipp Reisner 已提交
4009 4010 4011 4012 4013 4014

	c->word_offset += num_words;
	c->bit_offset = c->word_offset * BITS_PER_LONG;
	if (c->bit_offset > c->bm_bits)
		c->bit_offset = c->bm_bits;

4015
	return 1;
P
Philipp Reisner 已提交
4016 4017
}

4018 4019 4020 4021 4022 4023 4024 4025 4026 4027 4028 4029 4030 4031 4032
static enum drbd_bitmap_code dcbp_get_code(struct p_compressed_bm *p)
{
	return (enum drbd_bitmap_code)(p->encoding & 0x0f);
}

static int dcbp_get_start(struct p_compressed_bm *p)
{
	return (p->encoding & 0x80) != 0;
}

static int dcbp_get_pad_bits(struct p_compressed_bm *p)
{
	return (p->encoding >> 4) & 0x7;
}

4033 4034 4035 4036 4037 4038 4039
/**
 * recv_bm_rle_bits
 *
 * Return 0 when done, 1 when another iteration is needed, and a negative error
 * code upon failure.
 */
static int
P
Philipp Reisner 已提交
4040 4041
recv_bm_rle_bits(struct drbd_conf *mdev,
		struct p_compressed_bm *p,
4042 4043
		 struct bm_xfer_ctx *c,
		 unsigned int len)
P
Philipp Reisner 已提交
4044 4045 4046 4047 4048 4049 4050
{
	struct bitstream bs;
	u64 look_ahead;
	u64 rl;
	u64 tmp;
	unsigned long s = c->bit_offset;
	unsigned long e;
4051
	int toggle = dcbp_get_start(p);
P
Philipp Reisner 已提交
4052 4053 4054
	int have;
	int bits;

4055
	bitstream_init(&bs, p->code, len, dcbp_get_pad_bits(p));
P
Philipp Reisner 已提交
4056 4057 4058

	bits = bitstream_get_bits(&bs, &look_ahead, 64);
	if (bits < 0)
4059
		return -EIO;
P
Philipp Reisner 已提交
4060 4061 4062 4063

	for (have = bits; have > 0; s += rl, toggle = !toggle) {
		bits = vli_decode_bits(&rl, look_ahead);
		if (bits <= 0)
4064
			return -EIO;
P
Philipp Reisner 已提交
4065 4066 4067 4068 4069

		if (toggle) {
			e = s + rl -1;
			if (e >= c->bm_bits) {
				dev_err(DEV, "bitmap overflow (e:%lu) while decoding bm RLE packet\n", e);
4070
				return -EIO;
P
Philipp Reisner 已提交
4071 4072 4073 4074 4075 4076 4077 4078 4079
			}
			_drbd_bm_set_bits(mdev, s, e);
		}

		if (have < bits) {
			dev_err(DEV, "bitmap decoding error: h:%d b:%d la:0x%08llx l:%u/%u\n",
				have, bits, look_ahead,
				(unsigned int)(bs.cur.b - p->code),
				(unsigned int)bs.buf_len);
4080
			return -EIO;
P
Philipp Reisner 已提交
4081 4082 4083 4084 4085 4086
		}
		look_ahead >>= bits;
		have -= bits;

		bits = bitstream_get_bits(&bs, &tmp, 64 - have);
		if (bits < 0)
4087
			return -EIO;
P
Philipp Reisner 已提交
4088 4089 4090 4091 4092 4093 4094
		look_ahead |= tmp << have;
		have += bits;
	}

	c->bit_offset = s;
	bm_xfer_ctx_bit_to_word_offset(c);

4095
	return (s != c->bm_bits);
P
Philipp Reisner 已提交
4096 4097
}

4098 4099 4100 4101 4102 4103 4104
/**
 * decode_bitmap_c
 *
 * Return 0 when done, 1 when another iteration is needed, and a negative error
 * code upon failure.
 */
static int
P
Philipp Reisner 已提交
4105 4106
decode_bitmap_c(struct drbd_conf *mdev,
		struct p_compressed_bm *p,
4107 4108
		struct bm_xfer_ctx *c,
		unsigned int len)
P
Philipp Reisner 已提交
4109
{
4110
	if (dcbp_get_code(p) == RLE_VLI_Bits)
4111
		return recv_bm_rle_bits(mdev, p, c, len - sizeof(*p));
P
Philipp Reisner 已提交
4112 4113 4114 4115 4116 4117

	/* other variants had been implemented for evaluation,
	 * but have been dropped as this one turned out to be "best"
	 * during all our tests. */

	dev_err(DEV, "receive_bitmap_c: unknown encoding %u\n", p->encoding);
4118
	conn_request_state(mdev->tconn, NS(conn, C_PROTOCOL_ERROR), CS_HARD);
4119
	return -EIO;
P
Philipp Reisner 已提交
4120 4121 4122 4123 4124 4125
}

void INFO_bm_xfer_stats(struct drbd_conf *mdev,
		const char *direction, struct bm_xfer_ctx *c)
{
	/* what would it take to transfer it "plaintext" */
4126 4127 4128 4129 4130 4131 4132
	unsigned int header_size = drbd_header_size(mdev->tconn);
	unsigned int data_size = DRBD_SOCKET_BUFFER_SIZE - header_size;
	unsigned int plain =
		header_size * (DIV_ROUND_UP(c->bm_words, data_size) + 1) +
		c->bm_words * sizeof(unsigned long);
	unsigned int total = c->bytes[0] + c->bytes[1];
	unsigned int r;
P
Philipp Reisner 已提交
4133 4134 4135 4136 4137 4138 4139 4140 4141 4142 4143 4144 4145 4146 4147 4148 4149 4150 4151 4152 4153 4154 4155 4156 4157 4158 4159 4160 4161 4162 4163 4164 4165

	/* total can not be zero. but just in case: */
	if (total == 0)
		return;

	/* don't report if not compressed */
	if (total >= plain)
		return;

	/* total < plain. check for overflow, still */
	r = (total > UINT_MAX/1000) ? (total / (plain/1000))
		                    : (1000 * total / plain);

	if (r > 1000)
		r = 1000;

	r = 1000 - r;
	dev_info(DEV, "%s bitmap stats [Bytes(packets)]: plain %u(%u), RLE %u(%u), "
	     "total %u; compression: %u.%u%%\n",
			direction,
			c->bytes[1], c->packets[1],
			c->bytes[0], c->packets[0],
			total, r/10, r % 10);
}

/* Since we are processing the bitfield from lower addresses to higher,
   it does not matter if the process it in 32 bit chunks or 64 bit
   chunks as long as it is little endian. (Understand it as byte stream,
   beginning with the lowest byte...) If we would use big endian
   we would need to process it from the highest address to the lowest,
   in order to be agnostic to the 32 vs 64 bits issue.

   returns 0 on failure, 1 if we successfully received it. */
4166
static int receive_bitmap(struct drbd_tconn *tconn, struct packet_info *pi)
P
Philipp Reisner 已提交
4167
{
4168
	struct drbd_conf *mdev;
P
Philipp Reisner 已提交
4169
	struct bm_xfer_ctx c;
4170
	int err;
4171 4172 4173 4174

	mdev = vnr_to_mdev(tconn, pi->vnr);
	if (!mdev)
		return -EIO;
P
Philipp Reisner 已提交
4175

4176 4177 4178
	drbd_bm_lock(mdev, "receive bitmap", BM_LOCKED_SET_ALLOWED);
	/* you are supposed to send additional out-of-sync information
	 * if you actually set bits during this phase */
P
Philipp Reisner 已提交
4179 4180 4181 4182 4183 4184

	c = (struct bm_xfer_ctx) {
		.bm_bits = drbd_bm_bits(mdev),
		.bm_words = drbd_bm_words(mdev),
	};

4185
	for(;;) {
4186 4187 4188
		if (pi->cmd == P_BITMAP)
			err = receive_bitmap_plain(mdev, pi->size, pi->data, &c);
		else if (pi->cmd == P_COMPRESSED_BITMAP) {
P
Philipp Reisner 已提交
4189 4190
			/* MAYBE: sanity check that we speak proto >= 90,
			 * and the feature is enabled! */
4191
			struct p_compressed_bm *p = pi->data;
P
Philipp Reisner 已提交
4192

4193
			if (pi->size > DRBD_SOCKET_BUFFER_SIZE - drbd_header_size(tconn)) {
P
Philipp Reisner 已提交
4194
				dev_err(DEV, "ReportCBitmap packet too large\n");
4195
				err = -EIO;
P
Philipp Reisner 已提交
4196 4197
				goto out;
			}
4198
			if (pi->size <= sizeof(*p)) {
4199
				dev_err(DEV, "ReportCBitmap packet too small (l:%u)\n", pi->size);
4200
				err = -EIO;
4201
				goto out;
P
Philipp Reisner 已提交
4202
			}
4203 4204 4205
			err = drbd_recv_all(mdev->tconn, p, pi->size);
			if (err)
			       goto out;
4206
			err = decode_bitmap_c(mdev, p, &c, pi->size);
P
Philipp Reisner 已提交
4207
		} else {
4208
			dev_warn(DEV, "receive_bitmap: cmd neither ReportBitMap nor ReportCBitMap (is 0x%x)", pi->cmd);
4209
			err = -EIO;
P
Philipp Reisner 已提交
4210 4211 4212
			goto out;
		}

4213
		c.packets[pi->cmd == P_BITMAP]++;
4214
		c.bytes[pi->cmd == P_BITMAP] += drbd_header_size(tconn) + pi->size;
P
Philipp Reisner 已提交
4215

4216 4217 4218
		if (err <= 0) {
			if (err < 0)
				goto out;
P
Philipp Reisner 已提交
4219
			break;
4220
		}
4221
		err = drbd_recv_header(mdev->tconn, pi);
4222
		if (err)
P
Philipp Reisner 已提交
4223
			goto out;
4224
	}
P
Philipp Reisner 已提交
4225 4226 4227 4228

	INFO_bm_xfer_stats(mdev, "receive", &c);

	if (mdev->state.conn == C_WF_BITMAP_T) {
4229 4230
		enum drbd_state_rv rv;

4231 4232
		err = drbd_send_bitmap(mdev);
		if (err)
P
Philipp Reisner 已提交
4233 4234
			goto out;
		/* Omit CS_ORDERED with this state transition to avoid deadlocks. */
4235 4236
		rv = _drbd_request_state(mdev, NS(conn, C_WF_SYNC_UUID), CS_VERBOSE);
		D_ASSERT(rv == SS_SUCCESS);
P
Philipp Reisner 已提交
4237 4238 4239 4240 4241 4242
	} else if (mdev->state.conn != C_WF_BITMAP_S) {
		/* admin may have requested C_DISCONNECTING,
		 * other threads may have noticed network errors */
		dev_info(DEV, "unexpected cstate (%s) in receive_bitmap\n",
		    drbd_conn_str(mdev->state.conn));
	}
4243
	err = 0;
P
Philipp Reisner 已提交
4244 4245

 out:
4246
	drbd_bm_unlock(mdev);
4247
	if (!err && mdev->state.conn == C_WF_BITMAP_S)
P
Philipp Reisner 已提交
4248
		drbd_start_resync(mdev, C_SYNC_SOURCE);
4249
	return err;
P
Philipp Reisner 已提交
4250 4251
}

4252
static int receive_skip(struct drbd_tconn *tconn, struct packet_info *pi)
P
Philipp Reisner 已提交
4253
{
4254
	conn_warn(tconn, "skipping unknown optional packet type %d, l: %d!\n",
4255
		 pi->cmd, pi->size);
4256

4257
	return ignore_remaining_packet(tconn, pi);
4258 4259
}

4260
static int receive_UnplugRemote(struct drbd_tconn *tconn, struct packet_info *pi)
4261
{
4262 4263
	/* Make sure we've acked all the TCP data associated
	 * with the data requests being unplugged */
4264
	drbd_tcp_quickack(tconn->data.socket);
4265

4266
	return 0;
4267 4268
}

4269
static int receive_out_of_sync(struct drbd_tconn *tconn, struct packet_info *pi)
4270
{
4271
	struct drbd_conf *mdev;
4272
	struct p_block_desc *p = pi->data;
4273 4274 4275 4276

	mdev = vnr_to_mdev(tconn, pi->vnr);
	if (!mdev)
		return -EIO;
4277

4278 4279 4280 4281 4282 4283 4284 4285 4286 4287
	switch (mdev->state.conn) {
	case C_WF_SYNC_UUID:
	case C_WF_BITMAP_T:
	case C_BEHIND:
			break;
	default:
		dev_err(DEV, "ASSERT FAILED cstate = %s, expected: WFSyncUUID|WFBitMapT|Behind\n",
				drbd_conn_str(mdev->state.conn));
	}

4288 4289
	drbd_set_out_of_sync(mdev, be64_to_cpu(p->sector), be32_to_cpu(p->blksize));

4290
	return 0;
4291 4292
}

4293 4294 4295
struct data_cmd {
	int expect_payload;
	size_t pkt_size;
4296
	int (*fn)(struct drbd_tconn *, struct packet_info *);
4297 4298 4299
};

static struct data_cmd drbd_cmd_handler[] = {
4300 4301 4302 4303
	[P_DATA]	    = { 1, sizeof(struct p_data), receive_Data },
	[P_DATA_REPLY]	    = { 1, sizeof(struct p_data), receive_DataReply },
	[P_RS_DATA_REPLY]   = { 1, sizeof(struct p_data), receive_RSDataReply } ,
	[P_BARRIER]	    = { 0, sizeof(struct p_barrier), receive_Barrier } ,
4304 4305 4306
	[P_BITMAP]	    = { 1, 0, receive_bitmap } ,
	[P_COMPRESSED_BITMAP] = { 1, 0, receive_bitmap } ,
	[P_UNPLUG_REMOTE]   = { 0, 0, receive_UnplugRemote },
4307 4308
	[P_DATA_REQUEST]    = { 0, sizeof(struct p_block_req), receive_DataRequest },
	[P_RS_DATA_REQUEST] = { 0, sizeof(struct p_block_req), receive_DataRequest },
4309 4310
	[P_SYNC_PARAM]	    = { 1, 0, receive_SyncParam },
	[P_SYNC_PARAM89]    = { 1, 0, receive_SyncParam },
4311 4312 4313 4314 4315 4316 4317 4318 4319 4320 4321 4322
	[P_PROTOCOL]        = { 1, sizeof(struct p_protocol), receive_protocol },
	[P_UUIDS]	    = { 0, sizeof(struct p_uuids), receive_uuids },
	[P_SIZES]	    = { 0, sizeof(struct p_sizes), receive_sizes },
	[P_STATE]	    = { 0, sizeof(struct p_state), receive_state },
	[P_STATE_CHG_REQ]   = { 0, sizeof(struct p_req_state), receive_req_state },
	[P_SYNC_UUID]       = { 0, sizeof(struct p_rs_uuid), receive_sync_uuid },
	[P_OV_REQUEST]      = { 0, sizeof(struct p_block_req), receive_DataRequest },
	[P_OV_REPLY]        = { 1, sizeof(struct p_block_req), receive_DataRequest },
	[P_CSUM_RS_REQUEST] = { 1, sizeof(struct p_block_req), receive_DataRequest },
	[P_DELAY_PROBE]     = { 0, sizeof(struct p_delay_probe93), receive_skip },
	[P_OUT_OF_SYNC]     = { 0, sizeof(struct p_block_desc), receive_out_of_sync },
	[P_CONN_ST_CHG_REQ] = { 0, sizeof(struct p_req_state), receive_req_conn_state },
4323
	[P_PROTOCOL_UPDATE] = { 1, sizeof(struct p_protocol), receive_protocol },
P
Philipp Reisner 已提交
4324 4325
};

4326
static void drbdd(struct drbd_tconn *tconn)
P
Philipp Reisner 已提交
4327
{
4328
	struct packet_info pi;
4329
	size_t shs; /* sub header size */
4330
	int err;
P
Philipp Reisner 已提交
4331

4332
	while (get_t_state(&tconn->receiver) == RUNNING) {
4333 4334
		struct data_cmd *cmd;

4335
		drbd_thread_current_set_cpu(&tconn->receiver);
4336
		if (drbd_recv_header(tconn, &pi))
4337
			goto err_out;
P
Philipp Reisner 已提交
4338

4339
		cmd = &drbd_cmd_handler[pi.cmd];
4340
		if (unlikely(pi.cmd >= ARRAY_SIZE(drbd_cmd_handler) || !cmd->fn)) {
4341 4342
			conn_err(tconn, "Unexpected data packet %s (0x%04x)",
				 cmdname(pi.cmd), pi.cmd);
4343
			goto err_out;
4344
		}
P
Philipp Reisner 已提交
4345

4346 4347
		shs = cmd->pkt_size;
		if (pi.size > shs && !cmd->expect_payload) {
4348 4349
			conn_err(tconn, "No payload expected %s l:%d\n",
				 cmdname(pi.cmd), pi.size);
4350
			goto err_out;
P
Philipp Reisner 已提交
4351 4352
		}

4353
		if (shs) {
4354
			err = drbd_recv_all_warn(tconn, pi.data, shs);
4355
			if (err)
4356
				goto err_out;
4357
			pi.size -= shs;
4358 4359
		}

4360 4361
		err = cmd->fn(tconn, &pi);
		if (err) {
4362 4363
			conn_err(tconn, "error receiving %s, e: %d l: %d!\n",
				 cmdname(pi.cmd), err, pi.size);
4364
			goto err_out;
P
Philipp Reisner 已提交
4365 4366
		}
	}
4367
	return;
P
Philipp Reisner 已提交
4368

4369 4370
    err_out:
	conn_request_state(tconn, NS(conn, C_PROTOCOL_ERROR), CS_HARD);
P
Philipp Reisner 已提交
4371 4372
}

4373
void conn_flush_workqueue(struct drbd_tconn *tconn)
P
Philipp Reisner 已提交
4374 4375 4376 4377
{
	struct drbd_wq_barrier barr;

	barr.w.cb = w_prev_work_done;
4378
	barr.w.tconn = tconn;
P
Philipp Reisner 已提交
4379
	init_completion(&barr.done);
4380
	drbd_queue_work(&tconn->data.work, &barr.w);
P
Philipp Reisner 已提交
4381 4382 4383
	wait_for_completion(&barr.done);
}

4384
static void conn_disconnect(struct drbd_tconn *tconn)
P
Philipp Reisner 已提交
4385
{
P
Philipp Reisner 已提交
4386
	struct drbd_conf *mdev;
4387
	enum drbd_conns oc;
P
Philipp Reisner 已提交
4388
	int vnr;
P
Philipp Reisner 已提交
4389

4390
	if (tconn->cstate == C_STANDALONE)
P
Philipp Reisner 已提交
4391 4392
		return;

4393 4394 4395 4396 4397 4398 4399
	/* We are about to start the cleanup after connection loss.
	 * Make sure drbd_make_request knows about that.
	 * Usually we should be in some network failure state already,
	 * but just in case we are not, we fix it up here.
	 */
	conn_request_state(tconn, NS(conn, C_NETWORK_FAILURE), CS_HARD);

P
Philipp Reisner 已提交
4400
	/* asender does not clean up anything. it must not interfere, either */
4401 4402 4403
	drbd_thread_stop(&tconn->asender);
	drbd_free_sock(tconn);

P
Philipp Reisner 已提交
4404 4405 4406 4407 4408 4409 4410 4411 4412 4413
	rcu_read_lock();
	idr_for_each_entry(&tconn->volumes, mdev, vnr) {
		kref_get(&mdev->kref);
		rcu_read_unlock();
		drbd_disconnected(mdev);
		kref_put(&mdev->kref, &drbd_minor_destroy);
		rcu_read_lock();
	}
	rcu_read_unlock();

4414 4415 4416 4417 4418
	if (!list_empty(&tconn->current_epoch->list))
		conn_err(tconn, "ASSERTION FAILED: tconn->current_epoch->list not empty\n");
	/* ok, no more ee's on the fly, it is safe to reset the epoch_size */
	atomic_set(&tconn->current_epoch->epoch_size, 0);

4419 4420
	conn_info(tconn, "Connection closed\n");

4421 4422 4423
	if (conn_highest_role(tconn) == R_PRIMARY && conn_highest_pdsk(tconn) >= D_UNKNOWN)
		conn_try_outdate_peer_async(tconn);

4424
	spin_lock_irq(&tconn->req_lock);
4425 4426
	oc = tconn->cstate;
	if (oc >= C_UNCONNECTED)
P
Philipp Reisner 已提交
4427
		_conn_request_state(tconn, NS(conn, C_UNCONNECTED), CS_VERBOSE);
4428

4429 4430
	spin_unlock_irq(&tconn->req_lock);

4431
	if (oc == C_DISCONNECTING)
4432
		conn_request_state(tconn, NS(conn, C_STANDALONE), CS_VERBOSE | CS_HARD);
4433 4434
}

P
Philipp Reisner 已提交
4435
static int drbd_disconnected(struct drbd_conf *mdev)
4436 4437
{
	unsigned int i;
P
Philipp Reisner 已提交
4438

4439
	/* wait for current activity to cease. */
4440
	spin_lock_irq(&mdev->tconn->req_lock);
P
Philipp Reisner 已提交
4441 4442 4443
	_drbd_wait_ee_list_empty(mdev, &mdev->active_ee);
	_drbd_wait_ee_list_empty(mdev, &mdev->sync_ee);
	_drbd_wait_ee_list_empty(mdev, &mdev->read_ee);
4444
	spin_unlock_irq(&mdev->tconn->req_lock);
P
Philipp Reisner 已提交
4445 4446 4447 4448 4449 4450 4451 4452 4453 4454 4455 4456 4457 4458 4459 4460 4461 4462 4463 4464 4465 4466 4467

	/* We do not have data structures that would allow us to
	 * get the rs_pending_cnt down to 0 again.
	 *  * On C_SYNC_TARGET we do not have any data structures describing
	 *    the pending RSDataRequest's we have sent.
	 *  * On C_SYNC_SOURCE there is no data structure that tracks
	 *    the P_RS_DATA_REPLY blocks that we sent to the SyncTarget.
	 *  And no, it is not the sum of the reference counts in the
	 *  resync_LRU. The resync_LRU tracks the whole operation including
	 *  the disk-IO, while the rs_pending_cnt only tracks the blocks
	 *  on the fly. */
	drbd_rs_cancel_all(mdev);
	mdev->rs_total = 0;
	mdev->rs_failed = 0;
	atomic_set(&mdev->rs_pending_cnt, 0);
	wake_up(&mdev->misc_wait);

	del_timer_sync(&mdev->resync_timer);
	resync_timer_fn((unsigned long)mdev);

	/* wait for all w_e_end_data_req, w_e_end_rsdata_req, w_send_barrier,
	 * w_make_resync_request etc. which may still be on the worker queue
	 * to be "canceled" */
4468
	drbd_flush_workqueue(mdev);
P
Philipp Reisner 已提交
4469

4470
	drbd_finish_peer_reqs(mdev);
P
Philipp Reisner 已提交
4471

4472 4473 4474 4475 4476
	/* This second workqueue flush is necessary, since drbd_finish_peer_reqs()
	   might have issued a work again. The one before drbd_finish_peer_reqs() is
	   necessary to reclain net_ee in drbd_finish_peer_reqs(). */
	drbd_flush_workqueue(mdev);

P
Philipp Reisner 已提交
4477 4478 4479
	kfree(mdev->p_uuid);
	mdev->p_uuid = NULL;

4480
	if (!drbd_suspended(mdev))
4481
		tl_clear(mdev->tconn);
P
Philipp Reisner 已提交
4482 4483 4484

	drbd_md_sync(mdev);

4485 4486 4487 4488
	/* serialize with bitmap writeout triggered by the state change,
	 * if any. */
	wait_event(mdev->misc_wait, !test_bit(BITMAP_IO, &mdev->flags));

P
Philipp Reisner 已提交
4489 4490 4491 4492 4493 4494 4495
	/* tcp_close and release of sendpage pages can be deferred.  I don't
	 * want to use SO_LINGER, because apparently it can be deferred for
	 * more than 20 seconds (longest time I checked).
	 *
	 * Actually we don't care for exactly when the network stack does its
	 * put_page(), but release our reference on these pages right here.
	 */
4496
	i = drbd_free_peer_reqs(mdev, &mdev->net_ee);
P
Philipp Reisner 已提交
4497 4498
	if (i)
		dev_info(DEV, "net_ee not empty, killed %u entries\n", i);
4499 4500 4501
	i = atomic_read(&mdev->pp_in_use_by_net);
	if (i)
		dev_info(DEV, "pp_in_use_by_net = %d, expected 0\n", i);
P
Philipp Reisner 已提交
4502 4503
	i = atomic_read(&mdev->pp_in_use);
	if (i)
4504
		dev_info(DEV, "pp_in_use = %d, expected 0\n", i);
P
Philipp Reisner 已提交
4505 4506 4507 4508 4509 4510

	D_ASSERT(list_empty(&mdev->read_ee));
	D_ASSERT(list_empty(&mdev->active_ee));
	D_ASSERT(list_empty(&mdev->sync_ee));
	D_ASSERT(list_empty(&mdev->done_ee));

4511
	return 0;
P
Philipp Reisner 已提交
4512 4513 4514 4515 4516 4517 4518 4519 4520 4521 4522
}

/*
 * We support PRO_VERSION_MIN to PRO_VERSION_MAX. The protocol version
 * we can agree on is stored in agreed_pro_version.
 *
 * feature flags and the reserved array should be enough room for future
 * enhancements of the handshake protocol, and possible plugins...
 *
 * for now, they are expected to be zero, but ignored.
 */
4523
static int drbd_send_features(struct drbd_tconn *tconn)
P
Philipp Reisner 已提交
4524
{
4525 4526
	struct drbd_socket *sock;
	struct p_connection_features *p;
P
Philipp Reisner 已提交
4527

4528 4529 4530
	sock = &tconn->data;
	p = conn_prepare_command(tconn, sock);
	if (!p)
4531
		return -EIO;
P
Philipp Reisner 已提交
4532 4533 4534
	memset(p, 0, sizeof(*p));
	p->protocol_min = cpu_to_be32(PRO_VERSION_MIN);
	p->protocol_max = cpu_to_be32(PRO_VERSION_MAX);
4535
	return conn_send_command(tconn, sock, P_CONNECTION_FEATURES, sizeof(*p), NULL, 0);
P
Philipp Reisner 已提交
4536 4537 4538 4539 4540 4541 4542 4543 4544
}

/*
 * return values:
 *   1 yes, we have a valid connection
 *   0 oops, did not work out, please try again
 *  -1 peer talks different language,
 *     no point in trying again, please go standalone.
 */
4545
static int drbd_do_features(struct drbd_tconn *tconn)
P
Philipp Reisner 已提交
4546
{
4547
	/* ASSERT current == tconn->receiver ... */
4548 4549
	struct p_connection_features *p;
	const int expect = sizeof(struct p_connection_features);
4550
	struct packet_info pi;
4551
	int err;
P
Philipp Reisner 已提交
4552

4553
	err = drbd_send_features(tconn);
4554
	if (err)
P
Philipp Reisner 已提交
4555 4556
		return 0;

4557 4558
	err = drbd_recv_header(tconn, &pi);
	if (err)
P
Philipp Reisner 已提交
4559 4560
		return 0;

4561 4562
	if (pi.cmd != P_CONNECTION_FEATURES) {
		conn_err(tconn, "expected ConnectionFeatures packet, received: %s (0x%04x)\n",
4563
			 cmdname(pi.cmd), pi.cmd);
P
Philipp Reisner 已提交
4564 4565 4566
		return -1;
	}

4567
	if (pi.size != expect) {
4568
		conn_err(tconn, "expected ConnectionFeatures length: %u, received: %u\n",
4569
		     expect, pi.size);
P
Philipp Reisner 已提交
4570 4571 4572
		return -1;
	}

4573 4574
	p = pi.data;
	err = drbd_recv_all_warn(tconn, p, expect);
4575
	if (err)
P
Philipp Reisner 已提交
4576 4577 4578 4579 4580 4581 4582 4583 4584 4585 4586
		return 0;

	p->protocol_min = be32_to_cpu(p->protocol_min);
	p->protocol_max = be32_to_cpu(p->protocol_max);
	if (p->protocol_max == 0)
		p->protocol_max = p->protocol_min;

	if (PRO_VERSION_MAX < p->protocol_min ||
	    PRO_VERSION_MIN > p->protocol_max)
		goto incompat;

4587
	tconn->agreed_pro_version = min_t(int, PRO_VERSION_MAX, p->protocol_max);
P
Philipp Reisner 已提交
4588

4589 4590
	conn_info(tconn, "Handshake successful: "
	     "Agreed network protocol version %d\n", tconn->agreed_pro_version);
P
Philipp Reisner 已提交
4591 4592 4593 4594

	return 1;

 incompat:
4595
	conn_err(tconn, "incompatible DRBD dialects: "
P
Philipp Reisner 已提交
4596 4597 4598 4599 4600 4601 4602
	    "I support %d-%d, peer supports %d-%d\n",
	    PRO_VERSION_MIN, PRO_VERSION_MAX,
	    p->protocol_min, p->protocol_max);
	return -1;
}

#if !defined(CONFIG_CRYPTO_HMAC) && !defined(CONFIG_CRYPTO_HMAC_MODULE)
4603
static int drbd_do_auth(struct drbd_tconn *tconn)
P
Philipp Reisner 已提交
4604 4605 4606
{
	dev_err(DEV, "This kernel was build without CONFIG_CRYPTO_HMAC.\n");
	dev_err(DEV, "You need to disable 'cram-hmac-alg' in drbd.conf.\n");
4607
	return -1;
P
Philipp Reisner 已提交
4608 4609 4610
}
#else
#define CHALLENGE_LEN 64
4611 4612 4613 4614 4615 4616 4617

/* Return value:
	1 - auth succeeded,
	0 - failed, try again (network error),
	-1 - auth failed, don't try again.
*/

4618
static int drbd_do_auth(struct drbd_tconn *tconn)
P
Philipp Reisner 已提交
4619
{
4620
	struct drbd_socket *sock;
P
Philipp Reisner 已提交
4621 4622 4623 4624 4625
	char my_challenge[CHALLENGE_LEN];  /* 64 Bytes... */
	struct scatterlist sg;
	char *response = NULL;
	char *right_response = NULL;
	char *peers_ch = NULL;
4626 4627
	unsigned int key_len;
	char secret[SHARED_SECRET_MAX]; /* 64 byte */
P
Philipp Reisner 已提交
4628 4629
	unsigned int resp_size;
	struct hash_desc desc;
4630
	struct packet_info pi;
4631
	struct net_conf *nc;
4632
	int err, rv;
P
Philipp Reisner 已提交
4633

4634 4635
	/* FIXME: Put the challenge/response into the preallocated socket buffer.  */

4636 4637 4638 4639 4640 4641
	rcu_read_lock();
	nc = rcu_dereference(tconn->net_conf);
	key_len = strlen(nc->shared_secret);
	memcpy(secret, nc->shared_secret, key_len);
	rcu_read_unlock();

4642
	desc.tfm = tconn->cram_hmac_tfm;
P
Philipp Reisner 已提交
4643 4644
	desc.flags = 0;

4645
	rv = crypto_hash_setkey(tconn->cram_hmac_tfm, (u8 *)secret, key_len);
P
Philipp Reisner 已提交
4646
	if (rv) {
4647
		conn_err(tconn, "crypto_hash_setkey() failed with %d\n", rv);
4648
		rv = -1;
P
Philipp Reisner 已提交
4649 4650 4651 4652 4653
		goto fail;
	}

	get_random_bytes(my_challenge, CHALLENGE_LEN);

4654 4655 4656 4657 4658
	sock = &tconn->data;
	if (!conn_prepare_command(tconn, sock)) {
		rv = 0;
		goto fail;
	}
4659
	rv = !conn_send_command(tconn, sock, P_AUTH_CHALLENGE, 0,
4660
				my_challenge, CHALLENGE_LEN);
P
Philipp Reisner 已提交
4661 4662 4663
	if (!rv)
		goto fail;

4664 4665 4666
	err = drbd_recv_header(tconn, &pi);
	if (err) {
		rv = 0;
P
Philipp Reisner 已提交
4667
		goto fail;
4668
	}
P
Philipp Reisner 已提交
4669

4670
	if (pi.cmd != P_AUTH_CHALLENGE) {
4671
		conn_err(tconn, "expected AuthChallenge packet, received: %s (0x%04x)\n",
4672
			 cmdname(pi.cmd), pi.cmd);
P
Philipp Reisner 已提交
4673 4674 4675 4676
		rv = 0;
		goto fail;
	}

4677
	if (pi.size > CHALLENGE_LEN * 2) {
4678
		conn_err(tconn, "expected AuthChallenge payload too big.\n");
4679
		rv = -1;
P
Philipp Reisner 已提交
4680 4681 4682
		goto fail;
	}

4683
	peers_ch = kmalloc(pi.size, GFP_NOIO);
P
Philipp Reisner 已提交
4684
	if (peers_ch == NULL) {
4685
		conn_err(tconn, "kmalloc of peers_ch failed\n");
4686
		rv = -1;
P
Philipp Reisner 已提交
4687 4688 4689
		goto fail;
	}

4690 4691
	err = drbd_recv_all_warn(tconn, peers_ch, pi.size);
	if (err) {
P
Philipp Reisner 已提交
4692 4693 4694 4695
		rv = 0;
		goto fail;
	}

4696
	resp_size = crypto_hash_digestsize(tconn->cram_hmac_tfm);
P
Philipp Reisner 已提交
4697 4698
	response = kmalloc(resp_size, GFP_NOIO);
	if (response == NULL) {
4699
		conn_err(tconn, "kmalloc of response failed\n");
4700
		rv = -1;
P
Philipp Reisner 已提交
4701 4702 4703 4704
		goto fail;
	}

	sg_init_table(&sg, 1);
4705
	sg_set_buf(&sg, peers_ch, pi.size);
P
Philipp Reisner 已提交
4706 4707 4708

	rv = crypto_hash_digest(&desc, &sg, sg.length, response);
	if (rv) {
4709
		conn_err(tconn, "crypto_hash_digest() failed with %d\n", rv);
4710
		rv = -1;
P
Philipp Reisner 已提交
4711 4712 4713
		goto fail;
	}

4714 4715 4716 4717
	if (!conn_prepare_command(tconn, sock)) {
		rv = 0;
		goto fail;
	}
4718
	rv = !conn_send_command(tconn, sock, P_AUTH_RESPONSE, 0,
4719
				response, resp_size);
P
Philipp Reisner 已提交
4720 4721 4722
	if (!rv)
		goto fail;

4723 4724 4725
	err = drbd_recv_header(tconn, &pi);
	if (err) {
		rv = 0;
P
Philipp Reisner 已提交
4726
		goto fail;
4727
	}
P
Philipp Reisner 已提交
4728

4729
	if (pi.cmd != P_AUTH_RESPONSE) {
4730
		conn_err(tconn, "expected AuthResponse packet, received: %s (0x%04x)\n",
4731
			 cmdname(pi.cmd), pi.cmd);
P
Philipp Reisner 已提交
4732 4733 4734 4735
		rv = 0;
		goto fail;
	}

4736
	if (pi.size != resp_size) {
4737
		conn_err(tconn, "expected AuthResponse payload of wrong size\n");
P
Philipp Reisner 已提交
4738 4739 4740 4741
		rv = 0;
		goto fail;
	}

4742 4743
	err = drbd_recv_all_warn(tconn, response , resp_size);
	if (err) {
P
Philipp Reisner 已提交
4744 4745 4746 4747 4748
		rv = 0;
		goto fail;
	}

	right_response = kmalloc(resp_size, GFP_NOIO);
4749
	if (right_response == NULL) {
4750
		conn_err(tconn, "kmalloc of right_response failed\n");
4751
		rv = -1;
P
Philipp Reisner 已提交
4752 4753 4754 4755 4756 4757 4758
		goto fail;
	}

	sg_set_buf(&sg, my_challenge, CHALLENGE_LEN);

	rv = crypto_hash_digest(&desc, &sg, sg.length, right_response);
	if (rv) {
4759
		conn_err(tconn, "crypto_hash_digest() failed with %d\n", rv);
4760
		rv = -1;
P
Philipp Reisner 已提交
4761 4762 4763 4764 4765 4766
		goto fail;
	}

	rv = !memcmp(response, right_response, resp_size);

	if (rv)
4767 4768
		conn_info(tconn, "Peer authenticated using %d bytes HMAC\n",
		     resp_size);
4769 4770
	else
		rv = -1;
P
Philipp Reisner 已提交
4771 4772 4773 4774 4775 4776 4777 4778 4779 4780 4781 4782

 fail:
	kfree(peers_ch);
	kfree(response);
	kfree(right_response);

	return rv;
}
#endif

int drbdd_init(struct drbd_thread *thi)
{
4783
	struct drbd_tconn *tconn = thi->tconn;
P
Philipp Reisner 已提交
4784 4785
	int h;

4786
	conn_info(tconn, "receiver (re)started\n");
P
Philipp Reisner 已提交
4787 4788

	do {
4789
		h = conn_connect(tconn);
P
Philipp Reisner 已提交
4790
		if (h == 0) {
4791
			conn_disconnect(tconn);
4792
			schedule_timeout_interruptible(HZ);
P
Philipp Reisner 已提交
4793 4794
		}
		if (h == -1) {
4795
			conn_warn(tconn, "Discarding network configuration.\n");
4796
			conn_request_state(tconn, NS(conn, C_DISCONNECTING), CS_HARD);
P
Philipp Reisner 已提交
4797 4798 4799
		}
	} while (h == 0);

4800 4801
	if (h > 0)
		drbdd(tconn);
P
Philipp Reisner 已提交
4802

4803
	conn_disconnect(tconn);
P
Philipp Reisner 已提交
4804

4805
	conn_info(tconn, "receiver terminated\n");
P
Philipp Reisner 已提交
4806 4807 4808 4809 4810
	return 0;
}

/* ********* acknowledge sender ******** */

4811
static int got_conn_RqSReply(struct drbd_tconn *tconn, struct packet_info *pi)
4812
{
4813
	struct p_req_state_reply *p = pi->data;
4814 4815 4816 4817 4818 4819 4820 4821 4822 4823 4824
	int retcode = be32_to_cpu(p->retcode);

	if (retcode >= SS_SUCCESS) {
		set_bit(CONN_WD_ST_CHG_OKAY, &tconn->flags);
	} else {
		set_bit(CONN_WD_ST_CHG_FAIL, &tconn->flags);
		conn_err(tconn, "Requested state change failed by peer: %s (%d)\n",
			 drbd_set_st_err_str(retcode), retcode);
	}
	wake_up(&tconn->ping_wait);

4825
	return 0;
4826 4827
}

4828
static int got_RqSReply(struct drbd_tconn *tconn, struct packet_info *pi)
P
Philipp Reisner 已提交
4829
{
4830
	struct drbd_conf *mdev;
4831
	struct p_req_state_reply *p = pi->data;
P
Philipp Reisner 已提交
4832 4833
	int retcode = be32_to_cpu(p->retcode);

4834 4835
	mdev = vnr_to_mdev(tconn, pi->vnr);
	if (!mdev)
4836
		return -EIO;
4837

4838 4839 4840 4841 4842
	if (test_bit(CONN_WD_ST_CHG_REQ, &tconn->flags)) {
		D_ASSERT(tconn->agreed_pro_version < 100);
		return got_conn_RqSReply(tconn, pi);
	}

4843 4844 4845 4846 4847 4848
	if (retcode >= SS_SUCCESS) {
		set_bit(CL_ST_CHG_SUCCESS, &mdev->flags);
	} else {
		set_bit(CL_ST_CHG_FAIL, &mdev->flags);
		dev_err(DEV, "Requested state change failed by peer: %s (%d)\n",
			drbd_set_st_err_str(retcode), retcode);
P
Philipp Reisner 已提交
4849
	}
4850 4851
	wake_up(&mdev->state_wait);

4852
	return 0;
P
Philipp Reisner 已提交
4853 4854
}

4855
static int got_Ping(struct drbd_tconn *tconn, struct packet_info *pi)
P
Philipp Reisner 已提交
4856
{
4857
	return drbd_send_ping_ack(tconn);
P
Philipp Reisner 已提交
4858 4859 4860

}

4861
static int got_PingAck(struct drbd_tconn *tconn, struct packet_info *pi)
P
Philipp Reisner 已提交
4862 4863
{
	/* restore idle timeout */
4864 4865 4866
	tconn->meta.socket->sk->sk_rcvtimeo = tconn->net_conf->ping_int*HZ;
	if (!test_and_set_bit(GOT_PING_ACK, &tconn->flags))
		wake_up(&tconn->ping_wait);
P
Philipp Reisner 已提交
4867

4868
	return 0;
P
Philipp Reisner 已提交
4869 4870
}

4871
static int got_IsInSync(struct drbd_tconn *tconn, struct packet_info *pi)
P
Philipp Reisner 已提交
4872
{
4873
	struct drbd_conf *mdev;
4874
	struct p_block_ack *p = pi->data;
P
Philipp Reisner 已提交
4875 4876 4877
	sector_t sector = be64_to_cpu(p->sector);
	int blksize = be32_to_cpu(p->blksize);

4878 4879
	mdev = vnr_to_mdev(tconn, pi->vnr);
	if (!mdev)
4880
		return -EIO;
4881

4882
	D_ASSERT(mdev->tconn->agreed_pro_version >= 89);
P
Philipp Reisner 已提交
4883 4884 4885

	update_peer_seq(mdev, be32_to_cpu(p->seq_num));

4886 4887 4888 4889 4890 4891 4892
	if (get_ldev(mdev)) {
		drbd_rs_complete_io(mdev, sector);
		drbd_set_in_sync(mdev, sector, blksize);
		/* rs_same_csums is supposed to count in units of BM_BLOCK_SIZE */
		mdev->rs_same_csum += (blksize >> BM_BLOCK_SHIFT);
		put_ldev(mdev);
	}
P
Philipp Reisner 已提交
4893
	dec_rs_pending(mdev);
4894
	atomic_add(blksize >> 9, &mdev->rs_sect_in);
P
Philipp Reisner 已提交
4895

4896
	return 0;
P
Philipp Reisner 已提交
4897 4898
}

4899 4900 4901 4902
static int
validate_req_change_req_state(struct drbd_conf *mdev, u64 id, sector_t sector,
			      struct rb_root *root, const char *func,
			      enum drbd_req_event what, bool missing_ok)
P
Philipp Reisner 已提交
4903 4904 4905 4906
{
	struct drbd_request *req;
	struct bio_and_error m;

4907
	spin_lock_irq(&mdev->tconn->req_lock);
4908
	req = find_request(mdev, root, id, sector, missing_ok, func);
P
Philipp Reisner 已提交
4909
	if (unlikely(!req)) {
4910
		spin_unlock_irq(&mdev->tconn->req_lock);
4911
		return -EIO;
P
Philipp Reisner 已提交
4912 4913
	}
	__req_mod(req, what, &m);
4914
	spin_unlock_irq(&mdev->tconn->req_lock);
P
Philipp Reisner 已提交
4915 4916 4917

	if (m.bio)
		complete_master_bio(mdev, &m);
4918
	return 0;
P
Philipp Reisner 已提交
4919 4920
}

4921
static int got_BlockAck(struct drbd_tconn *tconn, struct packet_info *pi)
P
Philipp Reisner 已提交
4922
{
4923
	struct drbd_conf *mdev;
4924
	struct p_block_ack *p = pi->data;
P
Philipp Reisner 已提交
4925 4926 4927 4928
	sector_t sector = be64_to_cpu(p->sector);
	int blksize = be32_to_cpu(p->blksize);
	enum drbd_req_event what;

4929 4930
	mdev = vnr_to_mdev(tconn, pi->vnr);
	if (!mdev)
4931
		return -EIO;
4932

P
Philipp Reisner 已提交
4933 4934
	update_peer_seq(mdev, be32_to_cpu(p->seq_num));

4935
	if (p->block_id == ID_SYNCER) {
P
Philipp Reisner 已提交
4936 4937
		drbd_set_in_sync(mdev, sector, blksize);
		dec_rs_pending(mdev);
4938
		return 0;
P
Philipp Reisner 已提交
4939
	}
4940
	switch (pi->cmd) {
P
Philipp Reisner 已提交
4941
	case P_RS_WRITE_ACK:
4942
		what = WRITE_ACKED_BY_PEER_AND_SIS;
P
Philipp Reisner 已提交
4943 4944
		break;
	case P_WRITE_ACK:
4945
		what = WRITE_ACKED_BY_PEER;
P
Philipp Reisner 已提交
4946 4947
		break;
	case P_RECV_ACK:
4948
		what = RECV_ACKED_BY_PEER;
P
Philipp Reisner 已提交
4949
		break;
4950 4951 4952 4953 4954
	case P_DISCARD_WRITE:
		what = DISCARD_WRITE;
		break;
	case P_RETRY_WRITE:
		what = POSTPONE_WRITE;
P
Philipp Reisner 已提交
4955 4956
		break;
	default:
4957
		BUG();
P
Philipp Reisner 已提交
4958 4959
	}

4960 4961 4962
	return validate_req_change_req_state(mdev, p->block_id, sector,
					     &mdev->write_requests, __func__,
					     what, false);
P
Philipp Reisner 已提交
4963 4964
}

4965
static int got_NegAck(struct drbd_tconn *tconn, struct packet_info *pi)
P
Philipp Reisner 已提交
4966
{
4967
	struct drbd_conf *mdev;
4968
	struct p_block_ack *p = pi->data;
P
Philipp Reisner 已提交
4969
	sector_t sector = be64_to_cpu(p->sector);
4970
	int size = be32_to_cpu(p->blksize);
4971
	int err;
P
Philipp Reisner 已提交
4972

4973 4974
	mdev = vnr_to_mdev(tconn, pi->vnr);
	if (!mdev)
4975
		return -EIO;
4976

P
Philipp Reisner 已提交
4977 4978
	update_peer_seq(mdev, be32_to_cpu(p->seq_num));

4979
	if (p->block_id == ID_SYNCER) {
P
Philipp Reisner 已提交
4980 4981
		dec_rs_pending(mdev);
		drbd_rs_failed_io(mdev, sector, size);
4982
		return 0;
P
Philipp Reisner 已提交
4983
	}
4984

4985 4986
	err = validate_req_change_req_state(mdev, p->block_id, sector,
					    &mdev->write_requests, __func__,
4987
					    NEG_ACKED, true);
4988
	if (err) {
4989 4990 4991 4992 4993 4994
		/* Protocol A has no P_WRITE_ACKs, but has P_NEG_ACKs.
		   The master bio might already be completed, therefore the
		   request is no longer in the collision hash. */
		/* In Protocol B we might already have got a P_RECV_ACK
		   but then get a P_NEG_ACK afterwards. */
		drbd_set_out_of_sync(mdev, sector, size);
4995
	}
4996
	return 0;
P
Philipp Reisner 已提交
4997 4998
}

4999
static int got_NegDReply(struct drbd_tconn *tconn, struct packet_info *pi)
P
Philipp Reisner 已提交
5000
{
5001
	struct drbd_conf *mdev;
5002
	struct p_block_ack *p = pi->data;
P
Philipp Reisner 已提交
5003 5004
	sector_t sector = be64_to_cpu(p->sector);

5005 5006
	mdev = vnr_to_mdev(tconn, pi->vnr);
	if (!mdev)
5007
		return -EIO;
5008

P
Philipp Reisner 已提交
5009
	update_peer_seq(mdev, be32_to_cpu(p->seq_num));
5010

5011
	dev_err(DEV, "Got NegDReply; Sector %llus, len %u.\n",
P
Philipp Reisner 已提交
5012 5013
	    (unsigned long long)sector, be32_to_cpu(p->blksize));

5014 5015 5016
	return validate_req_change_req_state(mdev, p->block_id, sector,
					     &mdev->read_requests, __func__,
					     NEG_ACKED, false);
P
Philipp Reisner 已提交
5017 5018
}

5019
static int got_NegRSDReply(struct drbd_tconn *tconn, struct packet_info *pi)
P
Philipp Reisner 已提交
5020
{
5021
	struct drbd_conf *mdev;
P
Philipp Reisner 已提交
5022 5023
	sector_t sector;
	int size;
5024
	struct p_block_ack *p = pi->data;
5025 5026 5027

	mdev = vnr_to_mdev(tconn, pi->vnr);
	if (!mdev)
5028
		return -EIO;
P
Philipp Reisner 已提交
5029 5030 5031 5032 5033 5034 5035 5036 5037 5038

	sector = be64_to_cpu(p->sector);
	size = be32_to_cpu(p->blksize);

	update_peer_seq(mdev, be32_to_cpu(p->seq_num));

	dec_rs_pending(mdev);

	if (get_ldev_if_state(mdev, D_FAILED)) {
		drbd_rs_complete_io(mdev, sector);
5039
		switch (pi->cmd) {
5040 5041 5042 5043 5044
		case P_NEG_RS_DREPLY:
			drbd_rs_failed_io(mdev, sector, size);
		case P_RS_CANCEL:
			break;
		default:
5045
			BUG();
5046
		}
P
Philipp Reisner 已提交
5047 5048 5049
		put_ldev(mdev);
	}

5050
	return 0;
P
Philipp Reisner 已提交
5051 5052
}

5053
static int got_BarrierAck(struct drbd_tconn *tconn, struct packet_info *pi)
P
Philipp Reisner 已提交
5054
{
5055
	struct p_barrier_ack *p = pi->data;
5056 5057
	struct drbd_conf *mdev;
	int vnr;
5058

5059
	tl_release(tconn, p->barrier, be32_to_cpu(p->set_size));
P
Philipp Reisner 已提交
5060

5061 5062 5063 5064 5065 5066 5067 5068
	rcu_read_lock();
	idr_for_each_entry(&tconn->volumes, mdev, vnr) {
		if (mdev->state.conn == C_AHEAD &&
		    atomic_read(&mdev->ap_in_flight) == 0 &&
		    !test_and_set_bit(AHEAD_TO_SYNC_SOURCE, &mdev->flags)) {
			mdev->start_resync_timer.expires = jiffies + HZ;
			add_timer(&mdev->start_resync_timer);
		}
5069
	}
5070
	rcu_read_unlock();
5071

5072
	return 0;
P
Philipp Reisner 已提交
5073 5074
}

5075
static int got_OVResult(struct drbd_tconn *tconn, struct packet_info *pi)
P
Philipp Reisner 已提交
5076
{
5077
	struct drbd_conf *mdev;
5078
	struct p_block_ack *p = pi->data;
P
Philipp Reisner 已提交
5079 5080 5081 5082
	struct drbd_work *w;
	sector_t sector;
	int size;

5083 5084
	mdev = vnr_to_mdev(tconn, pi->vnr);
	if (!mdev)
5085
		return -EIO;
5086

P
Philipp Reisner 已提交
5087 5088 5089 5090 5091 5092
	sector = be64_to_cpu(p->sector);
	size = be32_to_cpu(p->blksize);

	update_peer_seq(mdev, be32_to_cpu(p->seq_num));

	if (be64_to_cpu(p->block_id) == ID_OUT_OF_SYNC)
5093
		drbd_ov_out_of_sync_found(mdev, sector, size);
P
Philipp Reisner 已提交
5094
	else
5095
		ov_out_of_sync_print(mdev);
P
Philipp Reisner 已提交
5096

5097
	if (!get_ldev(mdev))
5098
		return 0;
5099

P
Philipp Reisner 已提交
5100 5101 5102
	drbd_rs_complete_io(mdev, sector);
	dec_rs_pending(mdev);

5103 5104 5105 5106 5107 5108 5109
	--mdev->ov_left;

	/* let's advance progress step marks only for every other megabyte */
	if ((mdev->ov_left & 0x200) == 0x200)
		drbd_advance_rs_marks(mdev, mdev->ov_left);

	if (mdev->ov_left == 0) {
P
Philipp Reisner 已提交
5110 5111 5112
		w = kmalloc(sizeof(*w), GFP_NOIO);
		if (w) {
			w->cb = w_ov_finished;
5113
			w->mdev = mdev;
5114
			drbd_queue_work_front(&mdev->tconn->data.work, w);
P
Philipp Reisner 已提交
5115 5116
		} else {
			dev_err(DEV, "kmalloc(w) failed.");
5117
			ov_out_of_sync_print(mdev);
P
Philipp Reisner 已提交
5118 5119 5120
			drbd_resync_finished(mdev);
		}
	}
5121
	put_ldev(mdev);
5122
	return 0;
P
Philipp Reisner 已提交
5123 5124
}

5125
static int got_skip(struct drbd_tconn *tconn, struct packet_info *pi)
5126
{
5127
	return 0;
5128 5129
}

5130
static int tconn_finish_peer_reqs(struct drbd_tconn *tconn)
5131
{
5132
	struct drbd_conf *mdev;
P
Philipp Reisner 已提交
5133
	int vnr, not_empty = 0;
5134 5135 5136 5137

	do {
		clear_bit(SIGNAL_ASENDER, &tconn->flags);
		flush_signals(current);
P
Philipp Reisner 已提交
5138 5139 5140 5141 5142

		rcu_read_lock();
		idr_for_each_entry(&tconn->volumes, mdev, vnr) {
			kref_get(&mdev->kref);
			rcu_read_unlock();
5143
			if (drbd_finish_peer_reqs(mdev)) {
P
Philipp Reisner 已提交
5144 5145
				kref_put(&mdev->kref, &drbd_minor_destroy);
				return 1;
5146
			}
P
Philipp Reisner 已提交
5147 5148
			kref_put(&mdev->kref, &drbd_minor_destroy);
			rcu_read_lock();
5149
		}
5150
		set_bit(SIGNAL_ASENDER, &tconn->flags);
5151 5152

		spin_lock_irq(&tconn->req_lock);
P
Philipp Reisner 已提交
5153
		idr_for_each_entry(&tconn->volumes, mdev, vnr) {
5154 5155 5156 5157 5158
			not_empty = !list_empty(&mdev->done_ee);
			if (not_empty)
				break;
		}
		spin_unlock_irq(&tconn->req_lock);
P
Philipp Reisner 已提交
5159
		rcu_read_unlock();
5160 5161 5162 5163 5164
	} while (not_empty);

	return 0;
}

5165 5166
struct asender_cmd {
	size_t pkt_size;
5167
	int (*fn)(struct drbd_tconn *tconn, struct packet_info *);
5168 5169 5170
};

static struct asender_cmd asender_tbl[] = {
5171 5172
	[P_PING]	    = { 0, got_Ping },
	[P_PING_ACK]	    = { 0, got_PingAck },
5173 5174 5175 5176 5177 5178 5179 5180 5181 5182 5183 5184 5185 5186 5187
	[P_RECV_ACK]	    = { sizeof(struct p_block_ack), got_BlockAck },
	[P_WRITE_ACK]	    = { sizeof(struct p_block_ack), got_BlockAck },
	[P_RS_WRITE_ACK]    = { sizeof(struct p_block_ack), got_BlockAck },
	[P_DISCARD_WRITE]   = { sizeof(struct p_block_ack), got_BlockAck },
	[P_NEG_ACK]	    = { sizeof(struct p_block_ack), got_NegAck },
	[P_NEG_DREPLY]	    = { sizeof(struct p_block_ack), got_NegDReply },
	[P_NEG_RS_DREPLY]   = { sizeof(struct p_block_ack), got_NegRSDReply },
	[P_OV_RESULT]	    = { sizeof(struct p_block_ack), got_OVResult },
	[P_BARRIER_ACK]	    = { sizeof(struct p_barrier_ack), got_BarrierAck },
	[P_STATE_CHG_REPLY] = { sizeof(struct p_req_state_reply), got_RqSReply },
	[P_RS_IS_IN_SYNC]   = { sizeof(struct p_block_ack), got_IsInSync },
	[P_DELAY_PROBE]     = { sizeof(struct p_delay_probe93), got_skip },
	[P_RS_CANCEL]       = { sizeof(struct p_block_ack), got_NegRSDReply },
	[P_CONN_ST_CHG_REPLY]={ sizeof(struct p_req_state_reply), got_conn_RqSReply },
	[P_RETRY_WRITE]	    = { sizeof(struct p_block_ack), got_BlockAck },
5188 5189
};

P
Philipp Reisner 已提交
5190 5191
int drbd_asender(struct drbd_thread *thi)
{
5192
	struct drbd_tconn *tconn = thi->tconn;
P
Philipp Reisner 已提交
5193
	struct asender_cmd *cmd = NULL;
5194
	struct packet_info pi;
5195
	int rv;
5196
	void *buf    = tconn->meta.rbuf;
P
Philipp Reisner 已提交
5197
	int received = 0;
5198 5199
	unsigned int header_size = drbd_header_size(tconn);
	int expect   = header_size;
5200 5201
	bool ping_timeout_active = false;
	struct net_conf *nc;
5202
	int ping_timeo, tcp_cork, ping_int;
P
Philipp Reisner 已提交
5203 5204 5205 5206

	current->policy = SCHED_RR;  /* Make this a realtime task! */
	current->rt_priority = 2;    /* more important than all other tasks */

5207
	while (get_t_state(thi) == RUNNING) {
5208
		drbd_thread_current_set_cpu(thi);
5209 5210 5211 5212

		rcu_read_lock();
		nc = rcu_dereference(tconn->net_conf);
		ping_timeo = nc->ping_timeo;
5213
		tcp_cork = nc->tcp_cork;
5214 5215 5216
		ping_int = nc->ping_int;
		rcu_read_unlock();

5217
		if (test_and_clear_bit(SEND_PING, &tconn->flags)) {
5218
			if (drbd_send_ping(tconn)) {
5219
				conn_err(tconn, "drbd_send_ping has failed\n");
5220 5221
				goto reconnect;
			}
5222 5223
			tconn->meta.socket->sk->sk_rcvtimeo = ping_timeo * HZ / 10;
			ping_timeout_active = true;
P
Philipp Reisner 已提交
5224 5225
		}

5226 5227
		/* TODO: conditionally cork; it may hurt latency if we cork without
		   much to send */
5228
		if (tcp_cork)
5229
			drbd_tcp_cork(tconn->meta.socket);
5230 5231
		if (tconn_finish_peer_reqs(tconn)) {
			conn_err(tconn, "tconn_finish_peer_reqs() failed\n");
5232
			goto reconnect;
5233
		}
P
Philipp Reisner 已提交
5234
		/* but unconditionally uncork unless disabled */
5235
		if (tcp_cork)
5236
			drbd_tcp_uncork(tconn->meta.socket);
P
Philipp Reisner 已提交
5237 5238 5239 5240 5241

		/* short circuit, recv_msg would return EINTR anyways. */
		if (signal_pending(current))
			continue;

5242 5243
		rv = drbd_recv_short(tconn->meta.socket, buf, expect-received, 0);
		clear_bit(SIGNAL_ASENDER, &tconn->flags);
P
Philipp Reisner 已提交
5244 5245 5246 5247 5248 5249 5250 5251 5252 5253 5254 5255 5256 5257 5258 5259 5260

		flush_signals(current);

		/* Note:
		 * -EINTR	 (on meta) we got a signal
		 * -EAGAIN	 (on meta) rcvtimeo expired
		 * -ECONNRESET	 other side closed the connection
		 * -ERESTARTSYS  (on data) we got a signal
		 * rv <  0	 other than above: unexpected error!
		 * rv == expected: full header or command
		 * rv <  expected: "woken" by signal during receive
		 * rv == 0	 : "connection shut down by peer"
		 */
		if (likely(rv > 0)) {
			received += rv;
			buf	 += rv;
		} else if (rv == 0) {
5261
			conn_err(tconn, "meta connection shut down by peer.\n");
P
Philipp Reisner 已提交
5262 5263
			goto reconnect;
		} else if (rv == -EAGAIN) {
5264 5265
			/* If the data socket received something meanwhile,
			 * that is good enough: peer is still alive. */
5266 5267
			if (time_after(tconn->last_received,
				jiffies - tconn->meta.socket->sk->sk_rcvtimeo))
5268
				continue;
5269
			if (ping_timeout_active) {
5270
				conn_err(tconn, "PingAck did not arrive in time.\n");
P
Philipp Reisner 已提交
5271 5272
				goto reconnect;
			}
5273
			set_bit(SEND_PING, &tconn->flags);
P
Philipp Reisner 已提交
5274 5275 5276 5277
			continue;
		} else if (rv == -EINTR) {
			continue;
		} else {
5278
			conn_err(tconn, "sock_recvmsg returned %d\n", rv);
P
Philipp Reisner 已提交
5279 5280 5281 5282
			goto reconnect;
		}

		if (received == expect && cmd == NULL) {
5283
			if (decode_header(tconn, tconn->meta.rbuf, &pi))
P
Philipp Reisner 已提交
5284
				goto reconnect;
5285
			cmd = &asender_tbl[pi.cmd];
5286
			if (pi.cmd >= ARRAY_SIZE(asender_tbl) || !cmd->fn) {
5287 5288
				conn_err(tconn, "Unexpected meta packet %s (0x%04x)\n",
					 cmdname(pi.cmd), pi.cmd);
P
Philipp Reisner 已提交
5289 5290
				goto disconnect;
			}
5291
			expect = header_size + cmd->pkt_size;
5292
			if (pi.size != expect - header_size) {
5293
				conn_err(tconn, "Wrong packet size on meta (c: %d, l: %d)\n",
5294
					pi.cmd, pi.size);
P
Philipp Reisner 已提交
5295
				goto reconnect;
5296
			}
P
Philipp Reisner 已提交
5297 5298
		}
		if (received == expect) {
5299
			bool err;
5300

5301 5302
			err = cmd->fn(tconn, &pi);
			if (err) {
5303
				conn_err(tconn, "%pf failed\n", cmd->fn);
P
Philipp Reisner 已提交
5304
				goto reconnect;
5305
			}
P
Philipp Reisner 已提交
5306

5307 5308
			tconn->last_received = jiffies;

5309 5310 5311 5312 5313
			if (cmd == &asender_tbl[P_PING_ACK]) {
				/* restore idle timeout */
				tconn->meta.socket->sk->sk_rcvtimeo = ping_int * HZ;
				ping_timeout_active = false;
			}
5314

5315
			buf	 = tconn->meta.rbuf;
P
Philipp Reisner 已提交
5316
			received = 0;
5317
			expect	 = header_size;
P
Philipp Reisner 已提交
5318 5319 5320 5321 5322 5323
			cmd	 = NULL;
		}
	}

	if (0) {
reconnect:
5324
		conn_request_state(tconn, NS(conn, C_NETWORK_FAILURE), CS_HARD);
P
Philipp Reisner 已提交
5325 5326 5327
	}
	if (0) {
disconnect:
5328
		conn_request_state(tconn, NS(conn, C_DISCONNECTING), CS_HARD);
P
Philipp Reisner 已提交
5329
	}
5330
	clear_bit(SIGNAL_ASENDER, &tconn->flags);
P
Philipp Reisner 已提交
5331

5332
	conn_info(tconn, "asender terminated\n");
P
Philipp Reisner 已提交
5333 5334 5335

	return 0;
}