drbd_receiver.c 135.3 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 53 54 55 56
struct packet_info {
	enum drbd_packet cmd;
	int size;
	int vnr;
};

P
Philipp Reisner 已提交
57 58 59 60 61 62
enum finish_epoch {
	FE_STILL_LIVE,
	FE_DESTROYED,
	FE_RECYCLED,
};

63 64 65 66 67
enum mdev_or_conn {
	MDEV,
	CONN,
};

68
static int drbd_do_handshake(struct drbd_tconn *tconn);
69
static int drbd_do_auth(struct drbd_tconn *tconn);
70
static int drbd_disconnected(int vnr, void *p, void *data);
P
Philipp Reisner 已提交
71 72

static enum finish_epoch drbd_may_finish_epoch(struct drbd_conf *, struct drbd_epoch *, enum epoch_event);
73
static int e_end_block(struct drbd_work *, int);
P
Philipp Reisner 已提交
74 75 76 77


#define GFP_TRY	(__GFP_HIGHMEM | __GFP_NOWARN)

78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95
/*
 * 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;
96 97 98 99

	if (!page)
		return NULL;

100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157
	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;
}

static struct page *drbd_pp_first_pages_or_try_alloc(struct drbd_conf *mdev, int number)
P
Philipp Reisner 已提交
158 159
{
	struct page *page = NULL;
160 161
	struct page *tmp = NULL;
	int i = 0;
P
Philipp Reisner 已提交
162 163 164

	/* Yes, testing drbd_pp_vacant outside the lock is racy.
	 * So what. It saves a spin_lock. */
165
	if (drbd_pp_vacant >= number) {
P
Philipp Reisner 已提交
166
		spin_lock(&drbd_pp_lock);
167 168 169
		page = page_chain_del(&drbd_pp_pool, number);
		if (page)
			drbd_pp_vacant -= number;
P
Philipp Reisner 已提交
170
		spin_unlock(&drbd_pp_lock);
171 172
		if (page)
			return page;
P
Philipp Reisner 已提交
173
	}
174

P
Philipp Reisner 已提交
175 176 177
	/* 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.  */
178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199
	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.
	 * No need to jump around here, drbd_pp_alloc will retry this
	 * 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
200 201 202 203
}

static void reclaim_net_ee(struct drbd_conf *mdev, struct list_head *to_be_freed)
{
204
	struct drbd_peer_request *peer_req;
P
Philipp Reisner 已提交
205 206 207 208 209 210 211 212
	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) {
213 214
		peer_req = list_entry(le, struct drbd_peer_request, w.list);
		if (drbd_ee_has_active_page(peer_req))
P
Philipp Reisner 已提交
215 216 217 218 219 220 221 222
			break;
		list_move(le, to_be_freed);
	}
}

static void drbd_kick_lo_and_reclaim_net(struct drbd_conf *mdev)
{
	LIST_HEAD(reclaimed);
223
	struct drbd_peer_request *peer_req, *t;
P
Philipp Reisner 已提交
224

225
	spin_lock_irq(&mdev->tconn->req_lock);
P
Philipp Reisner 已提交
226
	reclaim_net_ee(mdev, &reclaimed);
227
	spin_unlock_irq(&mdev->tconn->req_lock);
P
Philipp Reisner 已提交
228

229 230
	list_for_each_entry_safe(peer_req, t, &reclaimed, w.list)
		drbd_free_net_ee(mdev, peer_req);
P
Philipp Reisner 已提交
231 232 233
}

/**
234
 * drbd_pp_alloc() - Returns @number pages, retries forever (or until signalled)
P
Philipp Reisner 已提交
235
 * @mdev:	DRBD device.
236 237 238 239 240 241
 * @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
242
 *
243
 * Returns a page chain linked via page->private.
P
Philipp Reisner 已提交
244
 */
245
static struct page *drbd_pp_alloc(struct drbd_conf *mdev, unsigned number, bool retry)
P
Philipp Reisner 已提交
246 247 248 249
{
	struct page *page = NULL;
	DEFINE_WAIT(wait);

250 251
	/* Yes, we may run up to @number over max_buffers. If we
	 * follow it strictly, the admin will get it wrong anyways. */
252
	if (atomic_read(&mdev->pp_in_use) < mdev->tconn->net_conf->max_buffers)
253
		page = drbd_pp_first_pages_or_try_alloc(mdev, number);
P
Philipp Reisner 已提交
254

255
	while (page == NULL) {
P
Philipp Reisner 已提交
256 257 258 259
		prepare_to_wait(&drbd_pp_wait, &wait, TASK_INTERRUPTIBLE);

		drbd_kick_lo_and_reclaim_net(mdev);

260
		if (atomic_read(&mdev->pp_in_use) < mdev->tconn->net_conf->max_buffers) {
261
			page = drbd_pp_first_pages_or_try_alloc(mdev, number);
P
Philipp Reisner 已提交
262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277
			if (page)
				break;
		}

		if (!retry)
			break;

		if (signal_pending(current)) {
			dev_warn(DEV, "drbd_pp_alloc interrupted!\n");
			break;
		}

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

278 279
	if (page)
		atomic_add(number, &mdev->pp_in_use);
P
Philipp Reisner 已提交
280 281 282 283
	return page;
}

/* Must not be used from irq, as that may deadlock: see drbd_pp_alloc.
284
 * Is also used from inside an other spin_lock_irq(&mdev->tconn->req_lock);
285 286
 * Either links the page chain back to the global pool,
 * or returns all pages to the system. */
287
static void drbd_pp_free(struct drbd_conf *mdev, struct page *page, int is_net)
P
Philipp Reisner 已提交
288
{
289
	atomic_t *a = is_net ? &mdev->pp_in_use_by_net : &mdev->pp_in_use;
P
Philipp Reisner 已提交
290
	int i;
291

292
	if (drbd_pp_vacant > (DRBD_MAX_BIO_SIZE/PAGE_SIZE) * minor_count)
293 294 295 296 297 298 299 300
		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
301
	}
302
	i = atomic_sub_return(i, a);
303
	if (i < 0)
304 305
		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
306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323
	wake_up(&drbd_pp_wait);
}

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

You must not have the req_lock:
 drbd_free_ee()
 drbd_alloc_ee()
 drbd_init_ee()
 drbd_release_ee()
 drbd_ee_fix_bhs()
 drbd_process_done_ee()
 drbd_clear_done_ee()
 drbd_wait_ee_list_empty()
*/

324 325 326
struct drbd_peer_request *
drbd_alloc_ee(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
327
{
328
	struct drbd_peer_request *peer_req;
P
Philipp Reisner 已提交
329
	struct page *page;
330
	unsigned nr_pages = (data_size + PAGE_SIZE -1) >> PAGE_SHIFT;
P
Philipp Reisner 已提交
331

332
	if (drbd_insert_fault(mdev, DRBD_FAULT_AL_EE))
P
Philipp Reisner 已提交
333 334
		return NULL;

335 336
	peer_req = mempool_alloc(drbd_ee_mempool, gfp_mask & ~__GFP_HIGHMEM);
	if (!peer_req) {
P
Philipp Reisner 已提交
337 338 339 340 341
		if (!(gfp_mask & __GFP_NOWARN))
			dev_err(DEV, "alloc_ee: Allocation of an EE failed\n");
		return NULL;
	}

342 343 344
	page = drbd_pp_alloc(mdev, nr_pages, (gfp_mask & __GFP_WAIT));
	if (!page)
		goto fail;
P
Philipp Reisner 已提交
345

346 347 348 349 350 351 352
	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;
353
	peer_req->w.mdev = mdev;
354 355 356
	peer_req->pages = page;
	atomic_set(&peer_req->pending_bios, 0);
	peer_req->flags = 0;
357 358 359 360
	/*
	 * The block_id is opaque to the receiver.  It is not endianness
	 * converted, and sent back to the sender unchanged.
	 */
361
	peer_req->block_id = id;
P
Philipp Reisner 已提交
362

363
	return peer_req;
P
Philipp Reisner 已提交
364

365
 fail:
366
	mempool_free(peer_req, drbd_ee_mempool);
P
Philipp Reisner 已提交
367 368 369
	return NULL;
}

370
void drbd_free_some_ee(struct drbd_conf *mdev, struct drbd_peer_request *peer_req,
371
		       int is_net)
P
Philipp Reisner 已提交
372
{
373 374 375 376 377 378
	if (peer_req->flags & EE_HAS_DIGEST)
		kfree(peer_req->digest);
	drbd_pp_free(mdev, peer_req->pages, is_net);
	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
379 380 381 382 383
}

int drbd_release_ee(struct drbd_conf *mdev, struct list_head *list)
{
	LIST_HEAD(work_list);
384
	struct drbd_peer_request *peer_req, *t;
P
Philipp Reisner 已提交
385
	int count = 0;
386
	int is_net = list == &mdev->net_ee;
P
Philipp Reisner 已提交
387

388
	spin_lock_irq(&mdev->tconn->req_lock);
P
Philipp Reisner 已提交
389
	list_splice_init(list, &work_list);
390
	spin_unlock_irq(&mdev->tconn->req_lock);
P
Philipp Reisner 已提交
391

392 393
	list_for_each_entry_safe(peer_req, t, &work_list, w.list) {
		drbd_free_some_ee(mdev, peer_req, is_net);
P
Philipp Reisner 已提交
394 395 396 397 398 399
		count++;
	}
	return count;
}


400
/* See also comments in _req_mod(,BARRIER_ACKED)
P
Philipp Reisner 已提交
401 402 403 404 405 406 407 408 409 410
 * and receive_Barrier.
 *
 * Move entries from net_ee to done_ee, if ready.
 * Grab done_ee, call all callbacks, free the entries.
 * The callbacks typically send out ACKs.
 */
static int drbd_process_done_ee(struct drbd_conf *mdev)
{
	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);
P
Philipp Reisner 已提交
415 416
	reclaim_net_ee(mdev, &reclaimed);
	list_splice_init(&mdev->done_ee, &work_list);
417
	spin_unlock_irq(&mdev->tconn->req_lock);
P
Philipp Reisner 已提交
418

419 420
	list_for_each_entry_safe(peer_req, t, &reclaimed, w.list)
		drbd_free_net_ee(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_ee(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
438 439 440 441 442 443 444 445 446 447
}

void _drbd_wait_ee_list_empty(struct drbd_conf *mdev, struct list_head *head)
{
	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);
448
		spin_unlock_irq(&mdev->tconn->req_lock);
J
Jens Axboe 已提交
449
		io_schedule();
P
Philipp Reisner 已提交
450
		finish_wait(&mdev->ee_wait, &wait);
451
		spin_lock_irq(&mdev->tconn->req_lock);
P
Philipp Reisner 已提交
452 453 454 455 456
	}
}

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

/* see also kernel_accept; which is only present since 2.6.18.
 * also we want to log which part of it failed, exactly */
464
static int drbd_accept(const char **what, struct socket *sock, struct socket **newsock)
P
Philipp Reisner 已提交
465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492
{
	struct sock *sk = sock->sk;
	int err = 0;

	*what = "listen";
	err = sock->ops->listen(sock, 5);
	if (err < 0)
		goto out;

	*what = "sock_create_lite";
	err = sock_create_lite(sk->sk_family, sk->sk_type, sk->sk_protocol,
			       newsock);
	if (err < 0)
		goto out;

	*what = "accept";
	err = sock->ops->accept(sock, *newsock, 0);
	if (err < 0) {
		sock_release(*newsock);
		*newsock = NULL;
		goto out;
	}
	(*newsock)->ops  = sock->ops;

out:
	return err;
}

493
static int drbd_recv_short(struct socket *sock, void *buf, size_t size, int flags)
P
Philipp Reisner 已提交
494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514
{
	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;
}

515
static int drbd_recv(struct drbd_tconn *tconn, void *buf, size_t size)
P
Philipp Reisner 已提交
516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532
{
	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 (;;) {
533
		rv = sock_recvmsg(tconn->data.socket, &msg, size, msg.msg_flags);
P
Philipp Reisner 已提交
534 535 536 537 538 539 540 541 542 543
		if (rv == size)
			break;

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

		if (rv < 0) {
			if (rv == -ECONNRESET)
544
				conn_info(tconn, "sock was reset by peer\n");
P
Philipp Reisner 已提交
545
			else if (rv != -ERESTARTSYS)
546
				conn_err(tconn, "sock_recvmsg returned %d\n", rv);
P
Philipp Reisner 已提交
547 548
			break;
		} else if (rv == 0) {
549
			conn_info(tconn, "sock was shut down by peer\n");
P
Philipp Reisner 已提交
550 551 552 553 554 555 556 557 558 559 560 561 562
			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)
563
		conn_request_state(tconn, NS(conn, C_BROKEN_PIPE), CS_HARD);
P
Philipp Reisner 已提交
564 565 566 567

	return rv;
}

568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586
/* 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;
	}
}

587
static struct socket *drbd_try_connect(struct drbd_tconn *tconn)
P
Philipp Reisner 已提交
588 589 590 591 592 593 594
{
	const char *what;
	struct socket *sock;
	struct sockaddr_in6 src_in6;
	int err;
	int disconnect_on_error = 1;

595
	if (!get_net_conf(tconn))
P
Philipp Reisner 已提交
596 597 598
		return NULL;

	what = "sock_create_kern";
599
	err = sock_create_kern(((struct sockaddr *)tconn->net_conf->my_addr)->sa_family,
P
Philipp Reisner 已提交
600 601 602 603 604 605 606
		SOCK_STREAM, IPPROTO_TCP, &sock);
	if (err < 0) {
		sock = NULL;
		goto out;
	}

	sock->sk->sk_rcvtimeo =
607 608 609
	sock->sk->sk_sndtimeo =  tconn->net_conf->try_connect_int*HZ;
	drbd_setbufsize(sock, tconn->net_conf->sndbuf_size,
			tconn->net_conf->rcvbuf_size);
P
Philipp Reisner 已提交
610 611 612 613 614 615 616 617

       /* 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.
	*/
618 619 620
	memcpy(&src_in6, tconn->net_conf->my_addr,
	       min_t(int, tconn->net_conf->my_addr_len, sizeof(src_in6)));
	if (((struct sockaddr *)tconn->net_conf->my_addr)->sa_family == AF_INET6)
P
Philipp Reisner 已提交
621 622 623 624 625 626 627
		src_in6.sin6_port = 0;
	else
		((struct sockaddr_in *)&src_in6)->sin_port = 0; /* AF_INET & AF_SCI */

	what = "bind before connect";
	err = sock->ops->bind(sock,
			      (struct sockaddr *) &src_in6,
628
			      tconn->net_conf->my_addr_len);
P
Philipp Reisner 已提交
629 630 631 632 633 634 635 636
	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";
	err = sock->ops->connect(sock,
637 638
				 (struct sockaddr *)tconn->net_conf->peer_addr,
				 tconn->net_conf->peer_addr_len, 0);
P
Philipp Reisner 已提交
639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655

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:
656
			conn_err(tconn, "%s failed, err = %d\n", what, err);
P
Philipp Reisner 已提交
657 658
		}
		if (disconnect_on_error)
659
			conn_request_state(tconn, NS(conn, C_DISCONNECTING), CS_HARD);
P
Philipp Reisner 已提交
660
	}
661
	put_net_conf(tconn);
P
Philipp Reisner 已提交
662 663 664
	return sock;
}

665
static struct socket *drbd_wait_for_connect(struct drbd_tconn *tconn)
P
Philipp Reisner 已提交
666 667 668 669 670
{
	int timeo, err;
	struct socket *s_estab = NULL, *s_listen;
	const char *what;

671
	if (!get_net_conf(tconn))
P
Philipp Reisner 已提交
672 673 674
		return NULL;

	what = "sock_create_kern";
675
	err = sock_create_kern(((struct sockaddr *)tconn->net_conf->my_addr)->sa_family,
P
Philipp Reisner 已提交
676 677 678 679 680 681
		SOCK_STREAM, IPPROTO_TCP, &s_listen);
	if (err) {
		s_listen = NULL;
		goto out;
	}

682
	timeo = tconn->net_conf->try_connect_int * HZ;
P
Philipp Reisner 已提交
683 684 685 686 687
	timeo += (random32() & 1) ? timeo / 7 : -timeo / 7; /* 28.5% random jitter */

	s_listen->sk->sk_reuse    = 1; /* SO_REUSEADDR */
	s_listen->sk->sk_rcvtimeo = timeo;
	s_listen->sk->sk_sndtimeo = timeo;
688 689
	drbd_setbufsize(s_listen, tconn->net_conf->sndbuf_size,
			tconn->net_conf->rcvbuf_size);
P
Philipp Reisner 已提交
690 691 692

	what = "bind before listen";
	err = s_listen->ops->bind(s_listen,
693 694
			      (struct sockaddr *) tconn->net_conf->my_addr,
			      tconn->net_conf->my_addr_len);
P
Philipp Reisner 已提交
695 696 697
	if (err < 0)
		goto out;

698
	err = drbd_accept(&what, s_listen, &s_estab);
P
Philipp Reisner 已提交
699 700 701 702 703 704

out:
	if (s_listen)
		sock_release(s_listen);
	if (err < 0) {
		if (err != -EAGAIN && err != -EINTR && err != -ERESTARTSYS) {
705
			conn_err(tconn, "%s failed, err = %d\n", what, err);
706
			conn_request_state(tconn, NS(conn, C_DISCONNECTING), CS_HARD);
P
Philipp Reisner 已提交
707 708
		}
	}
709
	put_net_conf(tconn);
P
Philipp Reisner 已提交
710 711 712 713

	return s_estab;
}

714
static int drbd_send_fp(struct drbd_tconn *tconn, struct socket *sock, enum drbd_packet cmd)
P
Philipp Reisner 已提交
715
{
716
	struct p_header *h = &tconn->data.sbuf.header;
P
Philipp Reisner 已提交
717

718
	return !_conn_send_cmd(tconn, 0, sock, cmd, h, sizeof(*h), 0);
P
Philipp Reisner 已提交
719 720
}

721
static enum drbd_packet drbd_recv_fp(struct drbd_tconn *tconn, struct socket *sock)
P
Philipp Reisner 已提交
722
{
723
	struct p_header80 *h = &tconn->data.rbuf.header.h80;
P
Philipp Reisner 已提交
724 725
	int rr;

726
	rr = drbd_recv_short(sock, h, sizeof(*h), 0);
P
Philipp Reisner 已提交
727

728
	if (rr == sizeof(*h) && h->magic == cpu_to_be32(DRBD_MAGIC))
P
Philipp Reisner 已提交
729 730 731 732 733 734 735 736 737
		return be16_to_cpu(h->command);

	return 0xffff;
}

/**
 * drbd_socket_okay() - Free the socket if its connection is not okay
 * @sock:	pointer to the pointer to the socket.
 */
738
static int drbd_socket_okay(struct socket **sock)
P
Philipp Reisner 已提交
739 740 741 742 743
{
	int rr;
	char tb[4];

	if (!*sock)
744
		return false;
P
Philipp Reisner 已提交
745

746
	rr = drbd_recv_short(*sock, tb, 4, MSG_DONTWAIT | MSG_PEEK);
P
Philipp Reisner 已提交
747 748

	if (rr > 0 || rr == -EAGAIN) {
749
		return true;
P
Philipp Reisner 已提交
750 751 752
	} else {
		sock_release(*sock);
		*sock = NULL;
753
		return false;
P
Philipp Reisner 已提交
754 755
	}
}
756 757 758
/* Gets called if a connection is established, or if a new minor gets created
   in a connection */
int drbd_connected(int vnr, void *p, void *data)
759 760 761 762 763 764 765
{
	struct drbd_conf *mdev = (struct drbd_conf *)p;
	int ok = 1;

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

766 767 768 769
	mdev->state_mutex = mdev->tconn->agreed_pro_version < 100 ?
		&mdev->tconn->cstate_mutex :
		&mdev->own_state_mutex;

770
	ok &= !drbd_send_sync_param(mdev);
771
	ok &= !drbd_send_sizes(mdev, 0, 0);
772
	ok &= !drbd_send_uuids(mdev);
773
	ok &= !drbd_send_state(mdev);
774 775 776
	clear_bit(USE_DEGR_WFC_T, &mdev->flags);
	clear_bit(RESIZE_PENDING, &mdev->flags);

777

778 779 780
	return !ok;
}

P
Philipp Reisner 已提交
781 782 783 784 785 786 787 788
/*
 * 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...
 */
789
static int drbd_connect(struct drbd_tconn *tconn)
P
Philipp Reisner 已提交
790 791 792 793
{
	struct socket *s, *sock, *msock;
	int try, h, ok;

794
	if (conn_request_state(tconn, NS(conn, C_WF_CONNECTION), CS_VERBOSE) < SS_SUCCESS)
P
Philipp Reisner 已提交
795 796
		return -2;

797 798
	clear_bit(DISCARD_CONCURRENT, &tconn->flags);
	tconn->agreed_pro_version = 99;
P
Philipp Reisner 已提交
799 800
	/* agreed_pro_version must be smaller than 100 so we send the old
	   header (h80) in the first packet and in the handshake packet. */
P
Philipp Reisner 已提交
801 802 803 804 805 806 807

	sock  = NULL;
	msock = NULL;

	do {
		for (try = 0;;) {
			/* 3 tries, this should take less than a second! */
808
			s = drbd_try_connect(tconn);
P
Philipp Reisner 已提交
809 810 811
			if (s || ++try >= 3)
				break;
			/* give the other side time to call bind() & listen() */
812
			schedule_timeout_interruptible(HZ / 10);
P
Philipp Reisner 已提交
813 814 815 816
		}

		if (s) {
			if (!sock) {
817
				drbd_send_fp(tconn, s, P_HAND_SHAKE_S);
P
Philipp Reisner 已提交
818 819 820
				sock = s;
				s = NULL;
			} else if (!msock) {
821
				drbd_send_fp(tconn, s, P_HAND_SHAKE_M);
P
Philipp Reisner 已提交
822 823 824
				msock = s;
				s = NULL;
			} else {
825
				conn_err(tconn, "Logic error in drbd_connect()\n");
P
Philipp Reisner 已提交
826 827 828 829 830
				goto out_release_sockets;
			}
		}

		if (sock && msock) {
831
			schedule_timeout_interruptible(tconn->net_conf->ping_timeo*HZ/10);
832 833
			ok = drbd_socket_okay(&sock);
			ok = drbd_socket_okay(&msock) && ok;
P
Philipp Reisner 已提交
834 835 836 837 838
			if (ok)
				break;
		}

retry:
839
		s = drbd_wait_for_connect(tconn);
P
Philipp Reisner 已提交
840
		if (s) {
841
			try = drbd_recv_fp(tconn, s);
842 843
			drbd_socket_okay(&sock);
			drbd_socket_okay(&msock);
P
Philipp Reisner 已提交
844 845 846
			switch (try) {
			case P_HAND_SHAKE_S:
				if (sock) {
847
					conn_warn(tconn, "initial packet S crossed\n");
P
Philipp Reisner 已提交
848 849 850 851 852 853
					sock_release(sock);
				}
				sock = s;
				break;
			case P_HAND_SHAKE_M:
				if (msock) {
854
					conn_warn(tconn, "initial packet M crossed\n");
P
Philipp Reisner 已提交
855 856 857
					sock_release(msock);
				}
				msock = s;
858
				set_bit(DISCARD_CONCURRENT, &tconn->flags);
P
Philipp Reisner 已提交
859 860
				break;
			default:
861
				conn_warn(tconn, "Error receiving initial packet\n");
P
Philipp Reisner 已提交
862 863 864 865 866 867
				sock_release(s);
				if (random32() & 1)
					goto retry;
			}
		}

868
		if (tconn->cstate <= C_DISCONNECTING)
P
Philipp Reisner 已提交
869 870 871 872
			goto out_release_sockets;
		if (signal_pending(current)) {
			flush_signals(current);
			smp_rmb();
873
			if (get_t_state(&tconn->receiver) == EXITING)
P
Philipp Reisner 已提交
874 875 876 877
				goto out_release_sockets;
		}

		if (sock && msock) {
878 879
			ok = drbd_socket_okay(&sock);
			ok = drbd_socket_okay(&msock) && ok;
P
Philipp Reisner 已提交
880 881 882 883 884 885 886 887 888 889 890 891 892 893 894
			if (ok)
				break;
		}
	} while (1);

	msock->sk->sk_reuse = 1; /* SO_REUSEADDR */
	sock->sk->sk_reuse = 1; /* SO_REUSEADDR */

	sock->sk->sk_allocation = GFP_NOIO;
	msock->sk->sk_allocation = GFP_NOIO;

	sock->sk->sk_priority = TC_PRIO_INTERACTIVE_BULK;
	msock->sk->sk_priority = TC_PRIO_INTERACTIVE;

	/* NOT YET ...
895
	 * sock->sk->sk_sndtimeo = tconn->net_conf->timeout*HZ/10;
P
Philipp Reisner 已提交
896 897 898 899
	 * sock->sk->sk_rcvtimeo = MAX_SCHEDULE_TIMEOUT;
	 * first set it to the P_HAND_SHAKE timeout,
	 * which we set to 4x the configured ping_timeout. */
	sock->sk->sk_sndtimeo =
900
	sock->sk->sk_rcvtimeo = tconn->net_conf->ping_timeo*4*HZ/10;
P
Philipp Reisner 已提交
901

902 903
	msock->sk->sk_sndtimeo = tconn->net_conf->timeout*HZ/10;
	msock->sk->sk_rcvtimeo = tconn->net_conf->ping_int*HZ;
P
Philipp Reisner 已提交
904 905

	/* we don't want delays.
L
Lucas De Marchi 已提交
906
	 * we use TCP_CORK where appropriate, though */
P
Philipp Reisner 已提交
907 908 909
	drbd_tcp_nodelay(sock);
	drbd_tcp_nodelay(msock);

910 911 912
	tconn->data.socket = sock;
	tconn->meta.socket = msock;
	tconn->last_received = jiffies;
P
Philipp Reisner 已提交
913

914
	h = drbd_do_handshake(tconn);
P
Philipp Reisner 已提交
915 916 917
	if (h <= 0)
		return h;

918
	if (tconn->cram_hmac_tfm) {
P
Philipp Reisner 已提交
919
		/* drbd_request_state(mdev, NS(conn, WFAuth)); */
920
		switch (drbd_do_auth(tconn)) {
921
		case -1:
922
			conn_err(tconn, "Authentication of peer failed\n");
P
Philipp Reisner 已提交
923
			return -1;
924
		case 0:
925
			conn_err(tconn, "Authentication of peer failed, trying again.\n");
926
			return 0;
P
Philipp Reisner 已提交
927 928 929
		}
	}

930
	if (conn_request_state(tconn, NS(conn, C_WF_REPORT_PARAMS), CS_VERBOSE) < SS_SUCCESS)
P
Philipp Reisner 已提交
931 932
		return 0;

933
	sock->sk->sk_sndtimeo = tconn->net_conf->timeout*HZ/10;
P
Philipp Reisner 已提交
934 935
	sock->sk->sk_rcvtimeo = MAX_SCHEDULE_TIMEOUT;

936
	drbd_thread_start(&tconn->asender);
P
Philipp Reisner 已提交
937

938
	if (drbd_send_protocol(tconn) == -EOPNOTSUPP)
939
		return -1;
P
Philipp Reisner 已提交
940

941
	return !idr_for_each(&tconn->volumes, drbd_connected, tconn);
P
Philipp Reisner 已提交
942 943 944 945 946 947 948 949 950

out_release_sockets:
	if (sock)
		sock_release(sock);
	if (msock)
		sock_release(msock);
	return -1;
}

951
static bool decode_header(struct drbd_tconn *tconn, struct p_header *h, struct packet_info *pi)
P
Philipp Reisner 已提交
952
{
P
Philipp Reisner 已提交
953
	if (h->h80.magic == cpu_to_be32(DRBD_MAGIC)) {
954 955
		pi->cmd = be16_to_cpu(h->h80.command);
		pi->size = be16_to_cpu(h->h80.length);
956
		pi->vnr = 0;
957
	} else if (h->h95.magic == cpu_to_be16(DRBD_MAGIC_BIG)) {
958 959 960
		pi->cmd = be16_to_cpu(h->h95.command);
		pi->size = be32_to_cpu(h->h95.length) & 0x00ffffff;
		pi->vnr = 0;
961
	} else {
962
		conn_err(tconn, "magic?? on data m: 0x%08x c: %d l: %d\n",
963 964 965
		    be32_to_cpu(h->h80.magic),
		    be16_to_cpu(h->h80.command),
		    be16_to_cpu(h->h80.length));
966
		return false;
P
Philipp Reisner 已提交
967
	}
968 969 970
	return true;
}

971
static int drbd_recv_header(struct drbd_tconn *tconn, struct packet_info *pi)
972
{
973
	struct p_header *h = &tconn->data.rbuf.header;
974 975
	int r;

976
	r = drbd_recv(tconn, h, sizeof(*h));
977 978
	if (unlikely(r != sizeof(*h))) {
		if (!signal_pending(current))
979
			conn_warn(tconn, "short read expecting header on sock: r=%d\n", r);
980 981 982
		return false;
	}

983 984
	r = decode_header(tconn, h, pi);
	tconn->last_received = jiffies;
P
Philipp Reisner 已提交
985

986
	return r;
P
Philipp Reisner 已提交
987 988
}

989
static void drbd_flush(struct drbd_conf *mdev)
P
Philipp Reisner 已提交
990 991 992 993
{
	int rv;

	if (mdev->write_ordering >= WO_bdev_flush && get_ldev(mdev)) {
994
		rv = blkdev_issue_flush(mdev->ldev->backing_bdev, GFP_KERNEL,
995
					NULL);
P
Philipp Reisner 已提交
996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016
		if (rv) {
			dev_err(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(mdev, WO_drain_io);
		}
		put_ldev(mdev);
	}
}

/**
 * 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.
 */
static enum finish_epoch drbd_may_finish_epoch(struct drbd_conf *mdev,
					       struct drbd_epoch *epoch,
					       enum epoch_event ev)
{
1017
	int epoch_size;
P
Philipp Reisner 已提交
1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040
	struct drbd_epoch *next_epoch;
	enum finish_epoch rv = FE_STILL_LIVE;

	spin_lock(&mdev->epoch_lock);
	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 &&
1041
		    test_bit(DE_HAVE_BARRIER_NUMBER, &epoch->flags)) {
P
Philipp Reisner 已提交
1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060
			if (!(ev & EV_CLEANUP)) {
				spin_unlock(&mdev->epoch_lock);
				drbd_send_b_ack(mdev, epoch->barrier_nr, epoch_size);
				spin_lock(&mdev->epoch_lock);
			}
			dec_unacked(mdev);

			if (mdev->current_epoch != epoch) {
				next_epoch = list_entry(epoch->list.next, struct drbd_epoch, list);
				list_del(&epoch->list);
				ev = EV_BECAME_LAST | (ev & EV_CLEANUP);
				mdev->epochs--;
				kfree(epoch);

				if (rv == FE_STILL_LIVE)
					rv = FE_DESTROYED;
			} else {
				epoch->flags = 0;
				atomic_set(&epoch->epoch_size, 0);
1061
				/* atomic_set(&epoch->active, 0); is already zero */
P
Philipp Reisner 已提交
1062 1063
				if (rv == FE_STILL_LIVE)
					rv = FE_RECYCLED;
1064
				wake_up(&mdev->ee_wait);
P
Philipp Reisner 已提交
1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099
			}
		}

		if (!next_epoch)
			break;

		epoch = next_epoch;
	} while (1);

	spin_unlock(&mdev->epoch_lock);

	return rv;
}

/**
 * drbd_bump_write_ordering() - Fall back to an other write ordering method
 * @mdev:	DRBD device.
 * @wo:		Write ordering method to try.
 */
void drbd_bump_write_ordering(struct drbd_conf *mdev, enum write_ordering_e wo) __must_hold(local)
{
	enum write_ordering_e pwo;
	static char *write_ordering_str[] = {
		[WO_none] = "none",
		[WO_drain_io] = "drain",
		[WO_bdev_flush] = "flush",
	};

	pwo = mdev->write_ordering;
	wo = min(pwo, wo);
	if (wo == WO_bdev_flush && mdev->ldev->dc.no_disk_flush)
		wo = WO_drain_io;
	if (wo == WO_drain_io && mdev->ldev->dc.no_disk_drain)
		wo = WO_none;
	mdev->write_ordering = wo;
1100
	if (pwo != mdev->write_ordering || wo == WO_bdev_flush)
P
Philipp Reisner 已提交
1101 1102 1103
		dev_info(DEV, "Method to ensure write ordering: %s\n", write_ordering_str[mdev->write_ordering]);
}

1104
/**
1105
 * drbd_submit_peer_request()
1106
 * @mdev:	DRBD device.
1107
 * @peer_req:	peer request
1108
 * @rw:		flag field, see bio->bi_rw
1109 1110 1111 1112 1113 1114 1115 1116 1117 1118
 *
 * 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.
1119 1120
 */
/* TODO allocate from our own bio_set. */
1121 1122 1123
int drbd_submit_peer_request(struct drbd_conf *mdev,
			     struct drbd_peer_request *peer_req,
			     const unsigned rw, const int fault_type)
1124 1125 1126
{
	struct bio *bios = NULL;
	struct bio *bio;
1127 1128 1129
	struct page *page = peer_req->pages;
	sector_t sector = peer_req->i.sector;
	unsigned ds = peer_req->i.size;
1130 1131
	unsigned n_bios = 0;
	unsigned nr_pages = (ds + PAGE_SIZE -1) >> PAGE_SHIFT;
1132
	int err = -ENOMEM;
1133 1134 1135 1136

	/* 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
1137 1138 1139 1140 1141
	 * 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.
	 */
1142 1143 1144 1145 1146 1147
next_bio:
	bio = bio_alloc(GFP_NOIO, nr_pages);
	if (!bio) {
		dev_err(DEV, "submit_ee: Allocation of a bio failed\n");
		goto fail;
	}
1148
	/* > peer_req->i.sector, unless this is the first bio */
1149 1150 1151
	bio->bi_sector = sector;
	bio->bi_bdev = mdev->ldev->backing_bdev;
	bio->bi_rw = rw;
1152
	bio->bi_private = peer_req;
1153
	bio->bi_end_io = drbd_peer_request_endio;
1154 1155 1156 1157 1158 1159 1160 1161

	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)) {
1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172
			/* 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;
			}
1173 1174 1175 1176 1177 1178 1179 1180 1181
			goto next_bio;
		}
		ds -= len;
		sector += len >> 9;
		--nr_pages;
	}
	D_ASSERT(page == NULL);
	D_ASSERT(ds == 0);

1182
	atomic_set(&peer_req->pending_bios, n_bios);
1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197
	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);
	}
1198
	return err;
1199 1200
}

1201
static void drbd_remove_epoch_entry_interval(struct drbd_conf *mdev,
1202
					     struct drbd_peer_request *peer_req)
1203
{
1204
	struct drbd_interval *i = &peer_req->i;
1205 1206 1207 1208

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

A
Andreas Gruenbacher 已提交
1209
	/* Wake up any processes waiting for this peer request to complete.  */
1210 1211 1212 1213
	if (i->waiting)
		wake_up(&mdev->misc_wait);
}

1214 1215
static int receive_Barrier(struct drbd_conf *mdev, enum drbd_packet cmd,
			   unsigned int data_size)
P
Philipp Reisner 已提交
1216
{
1217
	int rv;
1218
	struct p_barrier *p = &mdev->tconn->data.rbuf.barrier;
P
Philipp Reisner 已提交
1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233
	struct drbd_epoch *epoch;

	inc_unacked(mdev);

	mdev->current_epoch->barrier_nr = p->barrier;
	rv = drbd_may_finish_epoch(mdev, mdev->current_epoch, EV_GOT_BARRIER_NR);

	/* 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. */
	switch (mdev->write_ordering) {
	case WO_none:
		if (rv == FE_RECYCLED)
1234
			return true;
1235 1236 1237 1238 1239 1240 1241 1242 1243

		/* 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
			dev_warn(DEV, "Allocation of an epoch failed, slowing down\n");
			/* Fall through */
P
Philipp Reisner 已提交
1244 1245 1246 1247

	case WO_bdev_flush:
	case WO_drain_io:
		drbd_wait_ee_list_empty(mdev, &mdev->active_ee);
1248 1249 1250 1251 1252 1253
		drbd_flush(mdev);

		if (atomic_read(&mdev->current_epoch->epoch_size)) {
			epoch = kmalloc(sizeof(struct drbd_epoch), GFP_NOIO);
			if (epoch)
				break;
P
Philipp Reisner 已提交
1254 1255
		}

1256 1257 1258 1259 1260
		epoch = mdev->current_epoch;
		wait_event(mdev->ee_wait, atomic_read(&epoch->epoch_size) == 0);

		D_ASSERT(atomic_read(&epoch->active) == 0);
		D_ASSERT(epoch->flags == 0);
P
Philipp Reisner 已提交
1261

1262
		return true;
1263 1264
	default:
		dev_err(DEV, "Strangeness in mdev->write_ordering %d\n", mdev->write_ordering);
1265
		return false;
P
Philipp Reisner 已提交
1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282
	}

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

	spin_lock(&mdev->epoch_lock);
	if (atomic_read(&mdev->current_epoch->epoch_size)) {
		list_add(&epoch->list, &mdev->current_epoch->list);
		mdev->current_epoch = epoch;
		mdev->epochs++;
	} else {
		/* The current_epoch got recycled while we allocated this one... */
		kfree(epoch);
	}
	spin_unlock(&mdev->epoch_lock);

1283
	return true;
P
Philipp Reisner 已提交
1284 1285 1286 1287
}

/* used from receive_RSDataReply (recv_resync_read)
 * and from receive_Data */
1288 1289 1290
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
1291
{
1292
	const sector_t capacity = drbd_get_capacity(mdev->this_bdev);
1293
	struct drbd_peer_request *peer_req;
P
Philipp Reisner 已提交
1294
	struct page *page;
1295
	int dgs, ds, rr;
1296 1297
	void *dig_in = mdev->tconn->int_dig_in;
	void *dig_vv = mdev->tconn->int_dig_vv;
1298
	unsigned long *data;
P
Philipp Reisner 已提交
1299

1300 1301
	dgs = (mdev->tconn->agreed_pro_version >= 87 && mdev->tconn->integrity_r_tfm) ?
		crypto_hash_digestsize(mdev->tconn->integrity_r_tfm) : 0;
P
Philipp Reisner 已提交
1302 1303

	if (dgs) {
1304
		rr = drbd_recv(mdev->tconn, dig_in, dgs);
P
Philipp Reisner 已提交
1305
		if (rr != dgs) {
1306 1307 1308 1309
			if (!signal_pending(current))
				dev_warn(DEV,
					"short read receiving data digest: read %d expected %d\n",
					rr, dgs);
P
Philipp Reisner 已提交
1310 1311 1312 1313 1314 1315
			return NULL;
		}
	}

	data_size -= dgs;

1316 1317 1318 1319 1320 1321
	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
1322

1323 1324 1325
	/* even though we trust out peer,
	 * we sometimes have to double check. */
	if (sector + (data_size>>9) > capacity) {
1326 1327
		dev_err(DEV, "request from peer beyond end of local disk: "
			"capacity: %llus < sector: %llus + size: %u\n",
1328 1329 1330 1331 1332
			(unsigned long long)capacity,
			(unsigned long long)sector, data_size);
		return NULL;
	}

P
Philipp Reisner 已提交
1333 1334 1335
	/* 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.  */
1336 1337
	peer_req = drbd_alloc_ee(mdev, id, sector, data_size, GFP_NOIO);
	if (!peer_req)
P
Philipp Reisner 已提交
1338
		return NULL;
1339

P
Philipp Reisner 已提交
1340
	ds = data_size;
1341
	page = peer_req->pages;
1342 1343
	page_chain_for_each(page) {
		unsigned len = min_t(int, ds, PAGE_SIZE);
1344
		data = kmap(page);
1345
		rr = drbd_recv(mdev->tconn, data, len);
1346
		if (drbd_insert_fault(mdev, DRBD_FAULT_RECEIVE)) {
1347 1348 1349
			dev_err(DEV, "Fault injection: Corrupting data on receive\n");
			data[0] = data[0] ^ (unsigned long)-1;
		}
P
Philipp Reisner 已提交
1350
		kunmap(page);
1351
		if (rr != len) {
1352
			drbd_free_ee(mdev, peer_req);
1353 1354 1355
			if (!signal_pending(current))
				dev_warn(DEV, "short read receiving data: read %d expected %d\n",
				rr, len);
P
Philipp Reisner 已提交
1356 1357 1358 1359 1360 1361
			return NULL;
		}
		ds -= rr;
	}

	if (dgs) {
1362
		drbd_csum_ee(mdev, mdev->tconn->integrity_r_tfm, peer_req, dig_vv);
P
Philipp Reisner 已提交
1363
		if (memcmp(dig_in, dig_vv, dgs)) {
1364 1365
			dev_err(DEV, "Digest integrity check FAILED: %llus +%u\n",
				(unsigned long long)sector, data_size);
1366
			drbd_free_ee(mdev, peer_req);
P
Philipp Reisner 已提交
1367 1368 1369 1370
			return NULL;
		}
	}
	mdev->recv_cnt += data_size>>9;
1371
	return peer_req;
P
Philipp Reisner 已提交
1372 1373 1374 1375 1376 1377 1378 1379 1380 1381 1382
}

/* 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;
	int rr, rv = 1;
	void *data;

1383
	if (!data_size)
1384
		return true;
1385

1386
	page = drbd_pp_alloc(mdev, 1, 1);
P
Philipp Reisner 已提交
1387 1388 1389

	data = kmap(page);
	while (data_size) {
1390
		rr = drbd_recv(mdev->tconn, data, min_t(int, data_size, PAGE_SIZE));
P
Philipp Reisner 已提交
1391 1392
		if (rr != min_t(int, data_size, PAGE_SIZE)) {
			rv = 0;
1393 1394 1395 1396
			if (!signal_pending(current))
				dev_warn(DEV,
					"short read receiving data: read %d expected %d\n",
					rr, min_t(int, data_size, PAGE_SIZE));
P
Philipp Reisner 已提交
1397 1398 1399 1400 1401
			break;
		}
		data_size -= rr;
	}
	kunmap(page);
1402
	drbd_pp_free(mdev, page, 0);
P
Philipp Reisner 已提交
1403 1404 1405 1406 1407 1408 1409 1410 1411
	return rv;
}

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;
	int dgs, rr, i, expect;
1412 1413
	void *dig_in = mdev->tconn->int_dig_in;
	void *dig_vv = mdev->tconn->int_dig_vv;
P
Philipp Reisner 已提交
1414

1415 1416
	dgs = (mdev->tconn->agreed_pro_version >= 87 && mdev->tconn->integrity_r_tfm) ?
		crypto_hash_digestsize(mdev->tconn->integrity_r_tfm) : 0;
P
Philipp Reisner 已提交
1417 1418

	if (dgs) {
1419
		rr = drbd_recv(mdev->tconn, dig_in, dgs);
P
Philipp Reisner 已提交
1420
		if (rr != dgs) {
1421 1422 1423 1424
			if (!signal_pending(current))
				dev_warn(DEV,
					"short read receiving data reply digest: read %d expected %d\n",
					rr, dgs);
P
Philipp Reisner 已提交
1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 1439
			return 0;
		}
	}

	data_size -= dgs;

	/* 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) {
		expect = min_t(int, data_size, bvec->bv_len);
1440
		rr = drbd_recv(mdev->tconn,
P
Philipp Reisner 已提交
1441 1442 1443 1444
			     kmap(bvec->bv_page)+bvec->bv_offset,
			     expect);
		kunmap(bvec->bv_page);
		if (rr != expect) {
1445 1446 1447 1448
			if (!signal_pending(current))
				dev_warn(DEV, "short read receiving data reply: "
					"read %d expected %d\n",
					rr, expect);
P
Philipp Reisner 已提交
1449 1450 1451 1452 1453 1454
			return 0;
		}
		data_size -= rr;
	}

	if (dgs) {
1455
		drbd_csum_bio(mdev, mdev->tconn->integrity_r_tfm, bio, dig_vv);
P
Philipp Reisner 已提交
1456 1457 1458 1459 1460 1461 1462 1463 1464 1465 1466 1467
		if (memcmp(dig_in, dig_vv, dgs)) {
			dev_err(DEV, "Digest integrity check FAILED. Broken NICs?\n");
			return 0;
		}
	}

	D_ASSERT(data_size == 0);
	return 1;
}

/* e_end_resync_block() is called via
 * drbd_process_done_ee() by asender only */
1468
static int e_end_resync_block(struct drbd_work *w, int unused)
P
Philipp Reisner 已提交
1469
{
1470 1471
	struct drbd_peer_request *peer_req =
		container_of(w, struct drbd_peer_request, w);
1472
	struct drbd_conf *mdev = w->mdev;
1473
	sector_t sector = peer_req->i.sector;
1474
	int err;
P
Philipp Reisner 已提交
1475

1476
	D_ASSERT(drbd_interval_empty(&peer_req->i));
P
Philipp Reisner 已提交
1477

1478 1479
	if (likely((peer_req->flags & EE_WAS_ERROR) == 0)) {
		drbd_set_in_sync(mdev, sector, peer_req->i.size);
1480
		err = drbd_send_ack(mdev, P_RS_WRITE_ACK, peer_req);
P
Philipp Reisner 已提交
1481 1482
	} else {
		/* Record failure to sync */
1483
		drbd_rs_failed_io(mdev, sector, peer_req->i.size);
P
Philipp Reisner 已提交
1484

1485
		err  = drbd_send_ack(mdev, P_NEG_ACK, peer_req);
P
Philipp Reisner 已提交
1486 1487 1488
	}
	dec_unacked(mdev);

1489
	return err;
P
Philipp Reisner 已提交
1490 1491 1492 1493
}

static int recv_resync_read(struct drbd_conf *mdev, sector_t sector, int data_size) __releases(local)
{
1494
	struct drbd_peer_request *peer_req;
P
Philipp Reisner 已提交
1495

1496 1497
	peer_req = read_in_block(mdev, ID_SYNCER, sector, data_size);
	if (!peer_req)
1498
		goto fail;
P
Philipp Reisner 已提交
1499 1500 1501 1502 1503 1504 1505

	dec_rs_pending(mdev);

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

1506
	peer_req->w.cb = e_end_resync_block;
1507

1508
	spin_lock_irq(&mdev->tconn->req_lock);
1509
	list_add(&peer_req->w.list, &mdev->sync_ee);
1510
	spin_unlock_irq(&mdev->tconn->req_lock);
P
Philipp Reisner 已提交
1511

1512
	atomic_add(data_size >> 9, &mdev->rs_sect_ev);
1513
	if (drbd_submit_peer_request(mdev, peer_req, WRITE, DRBD_FAULT_RS_WR) == 0)
1514
		return true;
P
Philipp Reisner 已提交
1515

1516 1517
	/* don't care for the reason here */
	dev_err(DEV, "submit failed, triggering re-connect\n");
1518
	spin_lock_irq(&mdev->tconn->req_lock);
1519
	list_del(&peer_req->w.list);
1520
	spin_unlock_irq(&mdev->tconn->req_lock);
1521

1522
	drbd_free_ee(mdev, peer_req);
1523 1524
fail:
	put_ldev(mdev);
1525
	return false;
P
Philipp Reisner 已提交
1526 1527
}

1528
static struct drbd_request *
1529 1530
find_request(struct drbd_conf *mdev, struct rb_root *root, u64 id,
	     sector_t sector, bool missing_ok, const char *func)
1531 1532 1533
{
	struct drbd_request *req;

1534 1535
	/* Request object according to our peer */
	req = (struct drbd_request *)(unsigned long)id;
1536
	if (drbd_contains_interval(root, sector, &req->i) && req->i.local)
1537
		return req;
1538 1539 1540 1541
	if (!missing_ok) {
		dev_err(DEV, "%s: failed to find request %lu, sector %llus\n", func,
			(unsigned long)id, (unsigned long long)sector);
	}
1542 1543 1544
	return NULL;
}

1545 1546
static int receive_DataReply(struct drbd_conf *mdev, enum drbd_packet cmd,
			     unsigned int data_size)
P
Philipp Reisner 已提交
1547 1548 1549 1550
{
	struct drbd_request *req;
	sector_t sector;
	int ok;
1551
	struct p_data *p = &mdev->tconn->data.rbuf.data;
P
Philipp Reisner 已提交
1552 1553 1554

	sector = be64_to_cpu(p->sector);

1555
	spin_lock_irq(&mdev->tconn->req_lock);
1556
	req = find_request(mdev, &mdev->read_requests, p->block_id, sector, false, __func__);
1557
	spin_unlock_irq(&mdev->tconn->req_lock);
1558
	if (unlikely(!req))
1559
		return false;
P
Philipp Reisner 已提交
1560

B
Bart Van Assche 已提交
1561
	/* hlist_del(&req->collision) is done in _req_may_be_done, to avoid
P
Philipp Reisner 已提交
1562 1563 1564 1565 1566
	 * special casing it there for the various failure cases.
	 * still no race with drbd_fail_pending_reads */
	ok = recv_dless_read(mdev, req, sector, data_size);

	if (ok)
1567
		req_mod(req, DATA_RECEIVED);
P
Philipp Reisner 已提交
1568 1569 1570 1571 1572 1573 1574
	/* else: nothing. handled from drbd_disconnect...
	 * I don't think we may complete this just yet
	 * in case we are "on-disconnect: freeze" */

	return ok;
}

1575 1576
static int receive_RSDataReply(struct drbd_conf *mdev, enum drbd_packet cmd,
			       unsigned int data_size)
P
Philipp Reisner 已提交
1577 1578 1579
{
	sector_t sector;
	int ok;
1580
	struct p_data *p = &mdev->tconn->data.rbuf.data;
P
Philipp Reisner 已提交
1581 1582 1583 1584 1585 1586 1587

	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,
1588
		 * or in drbd_peer_request_endio. */
P
Philipp Reisner 已提交
1589 1590 1591 1592 1593 1594 1595
		ok = recv_resync_read(mdev, sector, data_size);
	} else {
		if (__ratelimit(&drbd_ratelimit_state))
			dev_err(DEV, "Can not write resync data to local disk.\n");

		ok = drbd_drain_block(mdev, data_size);

1596
		drbd_send_ack_dp(mdev, P_NEG_ACK, p, data_size);
P
Philipp Reisner 已提交
1597 1598
	}

1599 1600
	atomic_add(data_size >> 9, &mdev->rs_sect_in);

P
Philipp Reisner 已提交
1601 1602 1603
	return ok;
}

1604
static int w_restart_write(struct drbd_work *w, int cancel)
1605 1606 1607 1608 1609 1610 1611 1612 1613 1614
{
	struct drbd_request *req = container_of(w, struct drbd_request, w);
	struct drbd_conf *mdev = w->mdev;
	struct bio *bio;
	unsigned long start_time;
	unsigned long flags;

	spin_lock_irqsave(&mdev->tconn->req_lock, flags);
	if (!expect(req->rq_state & RQ_POSTPONED)) {
		spin_unlock_irqrestore(&mdev->tconn->req_lock, flags);
1615
		return -EIO;
1616 1617 1618 1619 1620 1621 1622 1623 1624
	}
	bio = req->master_bio;
	start_time = req->start_time;
	/* Postponed requests will not have their master_bio completed!  */
	__req_mod(req, DISCARD_WRITE, NULL);
	spin_unlock_irqrestore(&mdev->tconn->req_lock, flags);

	while (__drbd_make_request(mdev, bio, start_time))
		/* retry */ ;
1625
	return 0;
1626 1627 1628 1629 1630 1631 1632 1633 1634 1635 1636 1637 1638 1639 1640 1641 1642 1643 1644 1645 1646 1647 1648
}

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;
		if (expect(list_empty(&req->w.list))) {
			req->w.mdev = mdev;
			req->w.cb = w_restart_write;
			drbd_queue_work(&mdev->tconn->data.work, &req->w);
		}
	}
}

P
Philipp Reisner 已提交
1649 1650 1651
/* e_end_block() is called via drbd_process_done_ee().
 * this means this function only runs in the asender thread
 */
1652
static int e_end_block(struct drbd_work *w, int cancel)
P
Philipp Reisner 已提交
1653
{
1654 1655
	struct drbd_peer_request *peer_req =
		container_of(w, struct drbd_peer_request, w);
1656
	struct drbd_conf *mdev = w->mdev;
1657
	sector_t sector = peer_req->i.sector;
1658
	int err = 0, pcmd;
P
Philipp Reisner 已提交
1659

1660
	if (mdev->tconn->net_conf->wire_protocol == DRBD_PROT_C) {
1661
		if (likely((peer_req->flags & EE_WAS_ERROR) == 0)) {
P
Philipp Reisner 已提交
1662 1663
			pcmd = (mdev->state.conn >= C_SYNC_SOURCE &&
				mdev->state.conn <= C_PAUSED_SYNC_T &&
1664
				peer_req->flags & EE_MAY_SET_IN_SYNC) ?
P
Philipp Reisner 已提交
1665
				P_RS_WRITE_ACK : P_WRITE_ACK;
1666
			err = drbd_send_ack(mdev, pcmd, peer_req);
P
Philipp Reisner 已提交
1667
			if (pcmd == P_RS_WRITE_ACK)
1668
				drbd_set_in_sync(mdev, sector, peer_req->i.size);
P
Philipp Reisner 已提交
1669
		} else {
1670
			err = drbd_send_ack(mdev, P_NEG_ACK, peer_req);
P
Philipp Reisner 已提交
1671 1672 1673 1674 1675 1676 1677
			/* 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.  */
1678
	if (mdev->tconn->net_conf->two_primaries) {
1679
		spin_lock_irq(&mdev->tconn->req_lock);
1680 1681
		D_ASSERT(!drbd_interval_empty(&peer_req->i));
		drbd_remove_epoch_entry_interval(mdev, peer_req);
1682 1683
		if (peer_req->flags & EE_RESTART_REQUESTS)
			restart_conflicting_writes(mdev, sector, peer_req->i.size);
1684
		spin_unlock_irq(&mdev->tconn->req_lock);
1685
	} else
1686
		D_ASSERT(drbd_interval_empty(&peer_req->i));
P
Philipp Reisner 已提交
1687

1688
	drbd_may_finish_epoch(mdev, peer_req->epoch, EV_PUT + (cancel ? EV_CLEANUP : 0));
P
Philipp Reisner 已提交
1689

1690
	return err;
P
Philipp Reisner 已提交
1691 1692
}

1693
static int e_send_ack(struct drbd_work *w, enum drbd_packet ack)
P
Philipp Reisner 已提交
1694
{
1695
	struct drbd_conf *mdev = w->mdev;
1696 1697
	struct drbd_peer_request *peer_req =
		container_of(w, struct drbd_peer_request, w);
1698
	int err;
P
Philipp Reisner 已提交
1699

1700
	err = drbd_send_ack(mdev, ack, peer_req);
P
Philipp Reisner 已提交
1701 1702
	dec_unacked(mdev);

1703
	return err;
P
Philipp Reisner 已提交
1704 1705
}

1706
static int e_send_discard_write(struct drbd_work *w, int unused)
1707 1708 1709 1710
{
	return e_send_ack(w, P_DISCARD_WRITE);
}

1711
static int e_send_retry_write(struct drbd_work *w, int unused)
1712 1713 1714 1715 1716 1717 1718
{
	struct drbd_tconn *tconn = w->mdev->tconn;

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

1719 1720 1721 1722 1723 1724 1725 1726 1727 1728 1729 1730 1731 1732 1733
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;
}

1734 1735 1736 1737 1738 1739 1740 1741 1742 1743 1744 1745 1746
static bool need_peer_seq(struct drbd_conf *mdev)
{
	struct drbd_tconn *tconn = mdev->tconn;

	/*
	 * 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().
	 */
	return tconn->net_conf->two_primaries &&
	       test_bit(DISCARD_CONCURRENT, &tconn->flags);
}

1747
static void update_peer_seq(struct drbd_conf *mdev, unsigned int peer_seq)
1748
{
1749
	unsigned int newest_peer_seq;
1750

1751 1752
	if (need_peer_seq(mdev)) {
		spin_lock(&mdev->peer_seq_lock);
1753 1754
		newest_peer_seq = seq_max(mdev->peer_seq, peer_seq);
		mdev->peer_seq = newest_peer_seq;
1755
		spin_unlock(&mdev->peer_seq_lock);
1756 1757
		/* wake up only if we actually changed mdev->peer_seq */
		if (peer_seq == newest_peer_seq)
1758 1759
			wake_up(&mdev->seq_wait);
	}
1760 1761
}

P
Philipp Reisner 已提交
1762 1763 1764 1765 1766 1767 1768 1769 1770 1771 1772 1773 1774 1775 1776 1777 1778 1779 1780 1781 1782
/* 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). */
1783
static int wait_for_and_update_peer_seq(struct drbd_conf *mdev, const u32 peer_seq)
P
Philipp Reisner 已提交
1784 1785 1786
{
	DEFINE_WAIT(wait);
	long timeout;
1787 1788 1789 1790 1791
	int ret;

	if (!need_peer_seq(mdev))
		return 0;

P
Philipp Reisner 已提交
1792 1793
	spin_lock(&mdev->peer_seq_lock);
	for (;;) {
1794 1795 1796
		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
1797
			break;
1798
		}
P
Philipp Reisner 已提交
1799 1800 1801 1802
		if (signal_pending(current)) {
			ret = -ERESTARTSYS;
			break;
		}
1803
		prepare_to_wait(&mdev->seq_wait, &wait, TASK_INTERRUPTIBLE);
P
Philipp Reisner 已提交
1804
		spin_unlock(&mdev->peer_seq_lock);
1805 1806
		timeout = mdev->tconn->net_conf->ping_timeo*HZ/10;
		timeout = schedule_timeout(timeout);
P
Philipp Reisner 已提交
1807
		spin_lock(&mdev->peer_seq_lock);
1808
		if (!timeout) {
P
Philipp Reisner 已提交
1809
			ret = -ETIMEDOUT;
1810
			dev_err(DEV, "Timed out waiting for missing ack packets; disconnecting\n");
P
Philipp Reisner 已提交
1811 1812 1813 1814
			break;
		}
	}
	spin_unlock(&mdev->peer_seq_lock);
1815
	finish_wait(&mdev->seq_wait, &wait);
P
Philipp Reisner 已提交
1816 1817 1818
	return ret;
}

1819 1820 1821 1822
/* 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)
1823
{
1824 1825 1826 1827
	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);
1828 1829
}

1830 1831 1832 1833 1834 1835 1836 1837 1838 1839 1840 1841 1842 1843 1844 1845 1846 1847 1848 1849 1850 1851 1852 1853 1854 1855 1856 1857 1858 1859 1860 1861 1862 1863 1864 1865 1866 1867 1868 1869 1870 1871 1872 1873 1874 1875 1876 1877 1878 1879 1880 1881 1882 1883 1884 1885 1886 1887 1888 1889 1890 1891 1892 1893 1894 1895 1896 1897 1898 1899 1900 1901 1902 1903 1904 1905 1906 1907 1908 1909 1910 1911 1912 1913 1914 1915 1916 1917 1918 1919 1920 1921 1922 1923 1924 1925 1926 1927 1928 1929 1930 1931 1932 1933 1934 1935 1936 1937 1938 1939 1940 1941 1942 1943 1944 1945 1946 1947 1948 1949 1950 1951 1952 1953 1954 1955 1956 1957 1958 1959 1960 1961 1962
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
1963
/* mirrored write */
1964 1965
static int receive_Data(struct drbd_conf *mdev, enum drbd_packet cmd,
			unsigned int data_size)
P
Philipp Reisner 已提交
1966 1967
{
	sector_t sector;
1968
	struct drbd_peer_request *peer_req;
1969
	struct p_data *p = &mdev->tconn->data.rbuf.data;
1970
	u32 peer_seq = be32_to_cpu(p->seq_num);
P
Philipp Reisner 已提交
1971 1972
	int rw = WRITE;
	u32 dp_flags;
1973
	int err;
P
Philipp Reisner 已提交
1974 1975


1976 1977
	if (!get_ldev(mdev)) {
		err = wait_for_and_update_peer_seq(mdev, peer_seq);
1978
		drbd_send_ack_dp(mdev, P_NEG_ACK, p, data_size);
P
Philipp Reisner 已提交
1979
		atomic_inc(&mdev->current_epoch->epoch_size);
1980
		return drbd_drain_block(mdev, data_size) && err == 0;
P
Philipp Reisner 已提交
1981 1982
	}

1983 1984 1985 1986 1987
	/*
	 * 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
1988 1989

	sector = be64_to_cpu(p->sector);
1990 1991
	peer_req = read_in_block(mdev, p->block_id, sector, data_size);
	if (!peer_req) {
P
Philipp Reisner 已提交
1992
		put_ldev(mdev);
1993
		return false;
P
Philipp Reisner 已提交
1994 1995
	}

1996
	peer_req->w.cb = e_end_block;
P
Philipp Reisner 已提交
1997

1998 1999 2000 2001
	dp_flags = be32_to_cpu(p->dp_flags);
	rw |= wire_flags_to_bio(mdev, dp_flags);

	if (dp_flags & DP_MAY_SET_IN_SYNC)
2002
		peer_req->flags |= EE_MAY_SET_IN_SYNC;
2003

P
Philipp Reisner 已提交
2004
	spin_lock(&mdev->epoch_lock);
2005 2006 2007
	peer_req->epoch = mdev->current_epoch;
	atomic_inc(&peer_req->epoch->epoch_size);
	atomic_inc(&peer_req->epoch->active);
P
Philipp Reisner 已提交
2008 2009
	spin_unlock(&mdev->epoch_lock);

2010 2011 2012
	if (mdev->tconn->net_conf->two_primaries) {
		err = wait_for_and_update_peer_seq(mdev, peer_seq);
		if (err)
P
Philipp Reisner 已提交
2013
			goto out_interrupted;
2014
		spin_lock_irq(&mdev->tconn->req_lock);
2015 2016 2017 2018
		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
2019
				put_ldev(mdev);
2020
				return true;
P
Philipp Reisner 已提交
2021
			}
2022
			goto out_interrupted;
P
Philipp Reisner 已提交
2023
		}
2024 2025
	} else
		spin_lock_irq(&mdev->tconn->req_lock);
2026
	list_add(&peer_req->w.list, &mdev->active_ee);
2027
	spin_unlock_irq(&mdev->tconn->req_lock);
P
Philipp Reisner 已提交
2028

2029
	switch (mdev->tconn->net_conf->wire_protocol) {
P
Philipp Reisner 已提交
2030 2031 2032 2033 2034 2035 2036 2037
	case DRBD_PROT_C:
		inc_unacked(mdev);
		/* corresponding dec_unacked() in e_end_block()
		 * respective _drbd_clear_done_ee */
		break;
	case DRBD_PROT_B:
		/* I really don't like it that the receiver thread
		 * sends on the msock, but anyways */
2038
		drbd_send_ack(mdev, P_RECV_ACK, peer_req);
P
Philipp Reisner 已提交
2039 2040 2041 2042 2043 2044
		break;
	case DRBD_PROT_A:
		/* nothing to do */
		break;
	}

2045
	if (mdev->state.pdsk < D_INCONSISTENT) {
P
Philipp Reisner 已提交
2046
		/* In case we have the only disk of the cluster, */
2047 2048 2049 2050
		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;
		drbd_al_begin_io(mdev, peer_req->i.sector);
P
Philipp Reisner 已提交
2051 2052
	}

2053
	if (drbd_submit_peer_request(mdev, peer_req, rw, DRBD_FAULT_DT_WR) == 0)
2054
		return true;
P
Philipp Reisner 已提交
2055

2056 2057
	/* don't care for the reason here */
	dev_err(DEV, "submit failed, triggering re-connect\n");
2058
	spin_lock_irq(&mdev->tconn->req_lock);
2059 2060
	list_del(&peer_req->w.list);
	drbd_remove_epoch_entry_interval(mdev, peer_req);
2061
	spin_unlock_irq(&mdev->tconn->req_lock);
2062 2063
	if (peer_req->flags & EE_CALL_AL_COMPLETE_IO)
		drbd_al_complete_io(mdev, peer_req->i.sector);
2064

P
Philipp Reisner 已提交
2065
out_interrupted:
2066
	drbd_may_finish_epoch(mdev, peer_req->epoch, EV_PUT + EV_CLEANUP);
P
Philipp Reisner 已提交
2067
	put_ldev(mdev);
2068
	drbd_free_ee(mdev, peer_req);
2069
	return false;
P
Philipp Reisner 已提交
2070 2071
}

2072 2073 2074 2075 2076 2077 2078 2079 2080 2081 2082
/* 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.
 */
2083
int drbd_rs_should_slow_down(struct drbd_conf *mdev, sector_t sector)
2084 2085 2086
{
	struct gendisk *disk = mdev->ldev->backing_bdev->bd_contains->bd_disk;
	unsigned long db, dt, dbdt;
2087
	struct lc_element *tmp;
2088 2089 2090 2091
	int curr_events;
	int throttle = 0;

	/* feature disabled? */
2092
	if (mdev->ldev->dc.c_min_rate == 0)
2093 2094
		return 0;

2095 2096 2097 2098 2099 2100 2101 2102 2103 2104 2105 2106
	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);

2107 2108 2109
	curr_events = (int)part_stat_read(&disk->part0, sectors[0]) +
		      (int)part_stat_read(&disk->part0, sectors[1]) -
			atomic_read(&mdev->rs_sect_ev);
2110

2111 2112 2113 2114 2115 2116 2117 2118
	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. */
2119 2120 2121 2122 2123 2124
		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;
2125 2126 2127 2128 2129 2130 2131

		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);

2132
		if (dbdt > mdev->ldev->dc.c_min_rate)
2133 2134 2135 2136 2137 2138
			throttle = 1;
	}
	return throttle;
}


2139 2140
static int receive_DataRequest(struct drbd_conf *mdev, enum drbd_packet cmd,
			       unsigned int digest_size)
P
Philipp Reisner 已提交
2141 2142 2143
{
	sector_t sector;
	const sector_t capacity = drbd_get_capacity(mdev->this_bdev);
2144
	struct drbd_peer_request *peer_req;
P
Philipp Reisner 已提交
2145
	struct digest_info *di = NULL;
2146
	int size, verb;
P
Philipp Reisner 已提交
2147
	unsigned int fault_type;
2148
	struct p_block_req *p =	&mdev->tconn->data.rbuf.block_req;
P
Philipp Reisner 已提交
2149 2150 2151 2152

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

2153
	if (size <= 0 || !IS_ALIGNED(size, 512) || size > DRBD_MAX_BIO_SIZE) {
P
Philipp Reisner 已提交
2154 2155
		dev_err(DEV, "%s:%d: sector: %llus, size: %u\n", __FILE__, __LINE__,
				(unsigned long long)sector, size);
2156
		return false;
P
Philipp Reisner 已提交
2157 2158 2159 2160
	}
	if (sector + (size>>9) > capacity) {
		dev_err(DEV, "%s:%d: sector: %llus, size: %u\n", __FILE__, __LINE__,
				(unsigned long long)sector, size);
2161
		return false;
P
Philipp Reisner 已提交
2162 2163 2164
	}

	if (!get_ldev_if_state(mdev, D_UP_TO_DATE)) {
2165 2166 2167 2168 2169 2170 2171 2172 2173 2174 2175 2176 2177 2178 2179 2180 2181 2182 2183 2184
		verb = 1;
		switch (cmd) {
		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:
			dev_err(DEV, "unexpected command (%s) in receive_DataRequest\n",
				cmdname(cmd));
		}
		if (verb && __ratelimit(&drbd_ratelimit_state))
P
Philipp Reisner 已提交
2185 2186
			dev_err(DEV, "Can not satisfy peer's read request, "
			    "no local data.\n");
2187

L
Lars Ellenberg 已提交
2188 2189
		/* drain possibly payload */
		return drbd_drain_block(mdev, digest_size);
P
Philipp Reisner 已提交
2190 2191 2192 2193 2194
	}

	/* 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.  */
2195 2196
	peer_req = drbd_alloc_ee(mdev, p->block_id, sector, size, GFP_NOIO);
	if (!peer_req) {
P
Philipp Reisner 已提交
2197
		put_ldev(mdev);
2198
		return false;
P
Philipp Reisner 已提交
2199 2200
	}

2201
	switch (cmd) {
P
Philipp Reisner 已提交
2202
	case P_DATA_REQUEST:
2203
		peer_req->w.cb = w_e_end_data_req;
P
Philipp Reisner 已提交
2204
		fault_type = DRBD_FAULT_DT_RD;
2205 2206 2207
		/* application IO, don't drbd_rs_begin_io */
		goto submit;

P
Philipp Reisner 已提交
2208
	case P_RS_DATA_REQUEST:
2209
		peer_req->w.cb = w_e_end_rsdata_req;
P
Philipp Reisner 已提交
2210
		fault_type = DRBD_FAULT_RS_RD;
2211 2212
		/* used in the sector offset progress display */
		mdev->bm_resync_fo = BM_SECT_TO_BIT(sector);
P
Philipp Reisner 已提交
2213 2214 2215 2216 2217 2218 2219 2220 2221 2222 2223 2224
		break;

	case P_OV_REPLY:
	case P_CSUM_RS_REQUEST:
		fault_type = DRBD_FAULT_RS_RD;
		di = kmalloc(sizeof(*di) + digest_size, GFP_NOIO);
		if (!di)
			goto out_free_e;

		di->digest_size = digest_size;
		di->digest = (((char *)di)+sizeof(struct digest_info));

2225 2226
		peer_req->digest = di;
		peer_req->flags |= EE_HAS_DIGEST;
2227

2228
		if (drbd_recv(mdev->tconn, di->digest, digest_size) != digest_size)
P
Philipp Reisner 已提交
2229 2230
			goto out_free_e;

2231
		if (cmd == P_CSUM_RS_REQUEST) {
2232
			D_ASSERT(mdev->tconn->agreed_pro_version >= 89);
2233
			peer_req->w.cb = w_e_end_csum_rs_req;
2234 2235
			/* used in the sector offset progress display */
			mdev->bm_resync_fo = BM_SECT_TO_BIT(sector);
2236
		} else if (cmd == P_OV_REPLY) {
2237 2238
			/* track progress, we may need to throttle */
			atomic_add(size >> 9, &mdev->rs_sect_in);
2239
			peer_req->w.cb = w_e_end_ov_reply;
P
Philipp Reisner 已提交
2240
			dec_rs_pending(mdev);
2241 2242 2243
			/* 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
2244 2245 2246 2247 2248
		}
		break;

	case P_OV_REQUEST:
		if (mdev->ov_start_sector == ~(sector_t)0 &&
2249
		    mdev->tconn->agreed_pro_version >= 90) {
2250 2251
			unsigned long now = jiffies;
			int i;
P
Philipp Reisner 已提交
2252 2253
			mdev->ov_start_sector = sector;
			mdev->ov_position = sector;
2254 2255
			mdev->ov_left = drbd_bm_bits(mdev) - BM_SECT_TO_BIT(sector);
			mdev->rs_total = mdev->ov_left;
2256 2257 2258 2259
			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
2260 2261 2262
			dev_info(DEV, "Online Verify start sector: %llu\n",
					(unsigned long long)sector);
		}
2263
		peer_req->w.cb = w_e_end_ov_req;
P
Philipp Reisner 已提交
2264 2265 2266 2267 2268
		fault_type = DRBD_FAULT_RS_RD;
		break;

	default:
		dev_err(DEV, "unexpected command (%s) in receive_DataRequest\n",
2269
		    cmdname(cmd));
P
Philipp Reisner 已提交
2270
		fault_type = DRBD_FAULT_MAX;
2271
		goto out_free_e;
P
Philipp Reisner 已提交
2272 2273
	}

2274 2275 2276 2277 2278 2279 2280 2281 2282 2283 2284 2285 2286 2287 2288 2289 2290 2291 2292 2293 2294 2295
	/* 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.
	 */
2296 2297 2298
	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))
2299
		goto out_free_e;
P
Philipp Reisner 已提交
2300

2301 2302 2303
submit_for_resync:
	atomic_add(size >> 9, &mdev->rs_sect_ev);

2304
submit:
P
Philipp Reisner 已提交
2305
	inc_unacked(mdev);
2306
	spin_lock_irq(&mdev->tconn->req_lock);
2307
	list_add_tail(&peer_req->w.list, &mdev->read_ee);
2308
	spin_unlock_irq(&mdev->tconn->req_lock);
P
Philipp Reisner 已提交
2309

2310
	if (drbd_submit_peer_request(mdev, peer_req, READ, fault_type) == 0)
2311
		return true;
P
Philipp Reisner 已提交
2312

2313 2314
	/* don't care for the reason here */
	dev_err(DEV, "submit failed, triggering re-connect\n");
2315
	spin_lock_irq(&mdev->tconn->req_lock);
2316
	list_del(&peer_req->w.list);
2317
	spin_unlock_irq(&mdev->tconn->req_lock);
2318 2319
	/* no drbd_rs_complete_io(), we are dropping the connection anyways */

P
Philipp Reisner 已提交
2320 2321
out_free_e:
	put_ldev(mdev);
2322
	drbd_free_ee(mdev, peer_req);
2323
	return false;
P
Philipp Reisner 已提交
2324 2325 2326 2327 2328 2329 2330 2331 2332 2333 2334 2335 2336
}

static int drbd_asb_recover_0p(struct drbd_conf *mdev) __must_hold(local)
{
	int self, peer, rv = -100;
	unsigned long ch_self, ch_peer;

	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;

2337
	switch (mdev->tconn->net_conf->after_sb_0p) {
P
Philipp Reisner 已提交
2338 2339 2340 2341 2342 2343 2344 2345 2346 2347 2348 2349 2350 2351 2352 2353 2354 2355 2356 2357 2358 2359 2360 2361 2362 2363 2364
	case ASB_CONSENSUS:
	case ASB_DISCARD_SECONDARY:
	case ASB_CALL_HELPER:
		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
2365
		dev_warn(DEV, "Discard younger/older primary did not find a decision\n"
P
Philipp Reisner 已提交
2366 2367 2368
		     "Using discard-least-changes instead\n");
	case ASB_DISCARD_ZERO_CHG:
		if (ch_peer == 0 && ch_self == 0) {
2369
			rv = test_bit(DISCARD_CONCURRENT, &mdev->tconn->flags)
P
Philipp Reisner 已提交
2370 2371 2372 2373 2374 2375
				? -1 : 1;
			break;
		} else {
			if (ch_peer == 0) { rv =  1; break; }
			if (ch_self == 0) { rv = -1; break; }
		}
2376
		if (mdev->tconn->net_conf->after_sb_0p == ASB_DISCARD_ZERO_CHG)
P
Philipp Reisner 已提交
2377 2378 2379 2380 2381 2382 2383 2384
			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. */
2385
			rv = test_bit(DISCARD_CONCURRENT, &mdev->tconn->flags)
P
Philipp Reisner 已提交
2386 2387 2388 2389 2390 2391 2392 2393 2394 2395 2396 2397 2398 2399
				? -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)
{
2400
	int hg, rv = -100;
P
Philipp Reisner 已提交
2401

2402
	switch (mdev->tconn->net_conf->after_sb_1p) {
P
Philipp Reisner 已提交
2403 2404 2405 2406 2407 2408 2409 2410 2411 2412 2413 2414 2415 2416 2417 2418 2419 2420 2421 2422 2423 2424 2425 2426
	case ASB_DISCARD_YOUNGER_PRI:
	case ASB_DISCARD_OLDER_PRI:
	case ASB_DISCARD_LEAST_CHG:
	case ASB_DISCARD_LOCAL:
	case ASB_DISCARD_REMOTE:
		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) {
2427 2428 2429
			enum drbd_state_rv rv2;

			drbd_set_role(mdev, R_SECONDARY, 0);
P
Philipp Reisner 已提交
2430 2431 2432
			 /* 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. */
2433 2434
			rv2 = drbd_change_state(mdev, CS_VERBOSE, NS(role, R_SECONDARY));
			if (rv2 != SS_SUCCESS) {
P
Philipp Reisner 已提交
2435 2436 2437 2438 2439 2440 2441 2442 2443 2444 2445 2446 2447 2448
				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)
{
2449
	int hg, rv = -100;
P
Philipp Reisner 已提交
2450

2451
	switch (mdev->tconn->net_conf->after_sb_2p) {
P
Philipp Reisner 已提交
2452 2453 2454 2455 2456 2457 2458 2459 2460 2461 2462 2463 2464 2465 2466 2467 2468
	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:
		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) {
2469 2470
			enum drbd_state_rv rv2;

P
Philipp Reisner 已提交
2471 2472 2473
			 /* 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. */
2474 2475
			rv2 = drbd_change_state(mdev, CS_VERBOSE, NS(role, R_SECONDARY));
			if (rv2 != SS_SUCCESS) {
P
Philipp Reisner 已提交
2476 2477 2478 2479 2480 2481 2482 2483 2484 2485 2486 2487 2488 2489 2490 2491 2492 2493 2494 2495 2496 2497 2498 2499 2500 2501 2502 2503 2504 2505 2506 2507 2508 2509 2510 2511 2512 2513
				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
2514 2515
-1091   requires proto 91
-1096   requires proto 96
P
Philipp Reisner 已提交
2516 2517 2518 2519 2520 2521 2522 2523 2524 2525 2526 2527 2528 2529 2530 2531 2532 2533 2534 2535 2536 2537 2538 2539 2540 2541 2542 2543
 */
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) {

2544
			if (mdev->tconn->agreed_pro_version < 91)
2545
				return -1091;
P
Philipp Reisner 已提交
2546 2547 2548 2549 2550 2551 2552 2553 2554 2555 2556 2557 2558 2559 2560 2561 2562 2563 2564

			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) {

2565
			if (mdev->tconn->agreed_pro_version < 91)
2566
				return -1091;
P
Philipp Reisner 已提交
2567 2568 2569 2570 2571 2572 2573 2574 2575 2576 2577 2578 2579 2580 2581 2582 2583 2584 2585 2586 2587 2588 2589 2590 2591 2592 2593 2594 2595 2596 2597

			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 */
2598
			dc = test_bit(DISCARD_CONCURRENT, &mdev->tconn->flags);
P
Philipp Reisner 已提交
2599 2600 2601 2602 2603 2604 2605 2606 2607 2608 2609 2610
			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) {
2611
		if (mdev->tconn->agreed_pro_version < 96 ?
2612 2613 2614
		    (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
2615 2616 2617
			/* The last P_SYNC_UUID did not get though. Undo the last start of
			   resync as sync source modifications of the peer's UUIDs. */

2618
			if (mdev->tconn->agreed_pro_version < 91)
2619
				return -1091;
P
Philipp Reisner 已提交
2620 2621 2622

			mdev->p_uuid[UI_BITMAP] = mdev->p_uuid[UI_HISTORY_START];
			mdev->p_uuid[UI_HISTORY_START] = mdev->p_uuid[UI_HISTORY_START + 1];
2623 2624 2625 2626

			dev_info(DEV, "Did not got last syncUUID packet, corrected:\n");
			drbd_uuid_dump(mdev, "peer", mdev->p_uuid, mdev->p_uuid[UI_SIZE], mdev->p_uuid[UI_FLAGS]);

P
Philipp Reisner 已提交
2627 2628 2629 2630 2631 2632 2633 2634 2635 2636 2637 2638 2639 2640 2641 2642 2643 2644 2645 2646 2647
			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) {
2648
		if (mdev->tconn->agreed_pro_version < 96 ?
2649 2650 2651
		    (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
2652 2653 2654
			/* The last P_SYNC_UUID did not get though. Undo the last start of
			   resync as sync source modifications of our UUIDs. */

2655
			if (mdev->tconn->agreed_pro_version < 91)
2656
				return -1091;
P
Philipp Reisner 已提交
2657 2658 2659 2660

			_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]);

2661
			dev_info(DEV, "Last syncUUID did not get through, corrected:\n");
P
Philipp Reisner 已提交
2662 2663 2664 2665 2666 2667 2668 2669 2670
			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;
2671
	peer = mdev->p_uuid[UI_CURRENT] & ~((u64)1);
P
Philipp Reisner 已提交
2672 2673 2674 2675 2676 2677 2678 2679 2680 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 2719 2720 2721 2722 2723
	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)
{
	int hg, rule_nr;
	enum drbd_conns rv = C_MASK;
	enum drbd_disk_state mydisk;

	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;
	}
2724 2725
	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
2726 2727 2728 2729 2730 2731 2732 2733 2734 2735 2736 2737 2738
		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");
	}

2739 2740 2741
	if (abs(hg) == 100)
		drbd_khelper(mdev, "initial-split-brain");

2742
	if (hg == 100 || (hg == -100 && mdev->tconn->net_conf->always_asbp)) {
P
Philipp Reisner 已提交
2743 2744 2745 2746 2747 2748 2749 2750 2751 2752 2753 2754 2755 2756 2757 2758 2759 2760 2761 2762 2763 2764 2765 2766 2767 2768 2769 2770
		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) {
2771
		if (mdev->tconn->net_conf->want_lose && !(mdev->p_uuid[UI_FLAGS]&1))
P
Philipp Reisner 已提交
2772
			hg = -1;
2773
		if (!mdev->tconn->net_conf->want_lose && (mdev->p_uuid[UI_FLAGS]&1))
P
Philipp Reisner 已提交
2774 2775 2776 2777 2778 2779 2780 2781 2782
			hg = 1;

		if (abs(hg) < 100)
			dev_warn(DEV, "Split-Brain detected, manually solved. "
			     "Sync from %s node\n",
			     (hg < 0) ? "peer" : "this");
	}

	if (hg == -100) {
2783 2784 2785 2786
		/* 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... */
2787
		dev_alert(DEV, "Split-Brain detected but unresolved, dropping connection!\n");
P
Philipp Reisner 已提交
2788 2789 2790 2791 2792 2793 2794 2795 2796 2797 2798
		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) {
2799
		switch (mdev->tconn->net_conf->rr_conflict) {
P
Philipp Reisner 已提交
2800 2801 2802 2803 2804 2805 2806 2807 2808 2809 2810 2811
		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");
		}
	}

2812
	if (mdev->tconn->net_conf->dry_run || test_bit(CONN_DRY_RUN, &mdev->tconn->flags)) {
2813 2814 2815 2816 2817 2818 2819 2820 2821
		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
2822 2823
	if (abs(hg) >= 2) {
		dev_info(DEV, "Writing the whole bitmap, full sync required after drbd_sync_handshake.\n");
2824 2825
		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
2826 2827 2828 2829 2830 2831 2832 2833 2834 2835 2836 2837 2838 2839 2840 2841 2842 2843 2844 2845 2846 2847 2848 2849 2850 2851 2852 2853 2854 2855 2856 2857 2858 2859 2860 2861 2862 2863 2864
			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;
}

/* returns 1 if invalid */
static int cmp_after_sb(enum drbd_after_sb_p peer, enum drbd_after_sb_p self)
{
	/* ASB_DISCARD_REMOTE - ASB_DISCARD_LOCAL is valid */
	if ((peer == ASB_DISCARD_REMOTE && self == ASB_DISCARD_LOCAL) ||
	    (self == ASB_DISCARD_REMOTE && peer == ASB_DISCARD_LOCAL))
		return 0;

	/* any other things with ASB_DISCARD_REMOTE or ASB_DISCARD_LOCAL are invalid */
	if (peer == ASB_DISCARD_REMOTE || peer == ASB_DISCARD_LOCAL ||
	    self == ASB_DISCARD_REMOTE || self == ASB_DISCARD_LOCAL)
		return 1;

	/* everything else is valid if they are equal on both sides. */
	if (peer == self)
		return 0;

	/* everything es is invalid. */
	return 1;
}

2865
static int receive_protocol(struct drbd_tconn *tconn, enum drbd_packet cmd,
2866
			    unsigned int data_size)
P
Philipp Reisner 已提交
2867
{
2868
	struct p_protocol *p = &tconn->data.rbuf.protocol;
P
Philipp Reisner 已提交
2869
	int p_proto, p_after_sb_0p, p_after_sb_1p, p_after_sb_2p;
2870
	int p_want_lose, p_two_primaries, cf;
P
Philipp Reisner 已提交
2871 2872 2873 2874 2875 2876 2877
	char p_integrity_alg[SHARED_SECRET_MAX] = "";

	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);
2878 2879 2880
	cf		= be32_to_cpu(p->conn_flags);
	p_want_lose = cf & CF_WANT_LOSE;

2881
	clear_bit(CONN_DRY_RUN, &tconn->flags);
2882 2883

	if (cf & CF_DRY_RUN)
2884
		set_bit(CONN_DRY_RUN, &tconn->flags);
P
Philipp Reisner 已提交
2885

2886 2887
	if (p_proto != tconn->net_conf->wire_protocol) {
		conn_err(tconn, "incompatible communication protocols\n");
P
Philipp Reisner 已提交
2888 2889 2890
		goto disconnect;
	}

2891 2892
	if (cmp_after_sb(p_after_sb_0p, tconn->net_conf->after_sb_0p)) {
		conn_err(tconn, "incompatible after-sb-0pri settings\n");
P
Philipp Reisner 已提交
2893 2894 2895
		goto disconnect;
	}

2896 2897
	if (cmp_after_sb(p_after_sb_1p, tconn->net_conf->after_sb_1p)) {
		conn_err(tconn, "incompatible after-sb-1pri settings\n");
P
Philipp Reisner 已提交
2898 2899 2900
		goto disconnect;
	}

2901 2902
	if (cmp_after_sb(p_after_sb_2p, tconn->net_conf->after_sb_2p)) {
		conn_err(tconn, "incompatible after-sb-2pri settings\n");
P
Philipp Reisner 已提交
2903 2904 2905
		goto disconnect;
	}

2906 2907
	if (p_want_lose && tconn->net_conf->want_lose) {
		conn_err(tconn, "both sides have the 'want_lose' flag set\n");
P
Philipp Reisner 已提交
2908 2909 2910
		goto disconnect;
	}

2911 2912
	if (p_two_primaries != tconn->net_conf->two_primaries) {
		conn_err(tconn, "incompatible setting of the two-primaries options\n");
P
Philipp Reisner 已提交
2913 2914 2915
		goto disconnect;
	}

2916 2917
	if (tconn->agreed_pro_version >= 87) {
		unsigned char *my_alg = tconn->net_conf->integrity_alg;
P
Philipp Reisner 已提交
2918

2919
		if (drbd_recv(tconn, p_integrity_alg, data_size) != data_size)
2920
			return false;
P
Philipp Reisner 已提交
2921 2922 2923

		p_integrity_alg[SHARED_SECRET_MAX-1] = 0;
		if (strcmp(p_integrity_alg, my_alg)) {
2924
			conn_err(tconn, "incompatible setting of the data-integrity-alg\n");
P
Philipp Reisner 已提交
2925 2926
			goto disconnect;
		}
2927
		conn_info(tconn, "data-integrity-alg: %s\n",
P
Philipp Reisner 已提交
2928 2929 2930
		     my_alg[0] ? my_alg : (unsigned char *)"<not-used>");
	}

2931
	return true;
P
Philipp Reisner 已提交
2932 2933

disconnect:
2934
	conn_request_state(tconn, NS(conn, C_DISCONNECTING), CS_HARD);
2935
	return false;
P
Philipp Reisner 已提交
2936 2937 2938 2939 2940 2941 2942 2943 2944 2945 2946 2947 2948 2949 2950 2951 2952 2953 2954 2955 2956 2957 2958 2959 2960 2961 2962 2963 2964
}

/* 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;
	}
	if (!drbd_crypto_is_hash(crypto_hash_tfm(tfm))) {
		crypto_free_hash(tfm);
		dev_err(DEV, "\"%s\" is not a digest (%s)\n", alg, name);
		return ERR_PTR(-EINVAL);
	}
	return tfm;
}

2965 2966
static int receive_SyncParam(struct drbd_conf *mdev, enum drbd_packet cmd,
			     unsigned int packet_size)
P
Philipp Reisner 已提交
2967
{
2968
	int ok = true;
2969
	struct p_rs_param_95 *p = &mdev->tconn->data.rbuf.rs_param_95;
P
Philipp Reisner 已提交
2970 2971 2972
	unsigned int header_size, data_size, exp_max_sz;
	struct crypto_hash *verify_tfm = NULL;
	struct crypto_hash *csums_tfm = NULL;
2973
	const int apv = mdev->tconn->agreed_pro_version;
2974 2975
	int *rs_plan_s = NULL;
	int fifo_size = 0;
P
Philipp Reisner 已提交
2976 2977 2978 2979

	exp_max_sz  = apv <= 87 ? sizeof(struct p_rs_param)
		    : apv == 88 ? sizeof(struct p_rs_param)
					+ SHARED_SECRET_MAX
2980 2981
		    : apv <= 94 ? sizeof(struct p_rs_param_89)
		    : /* apv >= 95 */ sizeof(struct p_rs_param_95);
P
Philipp Reisner 已提交
2982

2983
	if (packet_size > exp_max_sz) {
P
Philipp Reisner 已提交
2984
		dev_err(DEV, "SyncParam packet too long: received %u, expected <= %u bytes\n",
2985
		    packet_size, exp_max_sz);
2986
		return false;
P
Philipp Reisner 已提交
2987 2988 2989
	}

	if (apv <= 88) {
2990
		header_size = sizeof(struct p_rs_param) - sizeof(struct p_header);
2991
		data_size   = packet_size  - header_size;
2992
	} else if (apv <= 94) {
2993
		header_size = sizeof(struct p_rs_param_89) - sizeof(struct p_header);
2994
		data_size   = packet_size  - header_size;
P
Philipp Reisner 已提交
2995
		D_ASSERT(data_size == 0);
2996
	} else {
2997
		header_size = sizeof(struct p_rs_param_95) - sizeof(struct p_header);
2998
		data_size   = packet_size  - header_size;
P
Philipp Reisner 已提交
2999 3000 3001 3002 3003 3004
		D_ASSERT(data_size == 0);
	}

	/* initialize verify_alg and csums_alg */
	memset(p->verify_alg, 0, 2 * SHARED_SECRET_MAX);

3005
	if (drbd_recv(mdev->tconn, &p->head.payload, header_size) != header_size)
3006
		return false;
P
Philipp Reisner 已提交
3007

3008 3009 3010 3011
	if (get_ldev(mdev)) {
		mdev->ldev->dc.resync_rate = be32_to_cpu(p->rate);
		put_ldev(mdev);
	}
P
Philipp Reisner 已提交
3012 3013 3014 3015 3016 3017 3018

	if (apv >= 88) {
		if (apv == 88) {
			if (data_size > SHARED_SECRET_MAX) {
				dev_err(DEV, "verify-alg too long, "
				    "peer wants %u, accepting only %u byte\n",
						data_size, SHARED_SECRET_MAX);
3019
				return false;
P
Philipp Reisner 已提交
3020 3021
			}

3022
			if (drbd_recv(mdev->tconn, p->verify_alg, data_size) != data_size)
3023
				return false;
P
Philipp Reisner 已提交
3024 3025 3026 3027 3028 3029 3030 3031 3032 3033 3034 3035 3036 3037 3038

			/* 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;
		}

3039
		if (strcmp(mdev->tconn->net_conf->verify_alg, p->verify_alg)) {
P
Philipp Reisner 已提交
3040 3041
			if (mdev->state.conn == C_WF_REPORT_PARAMS) {
				dev_err(DEV, "Different verify-alg settings. me=\"%s\" peer=\"%s\"\n",
3042
				    mdev->tconn->net_conf->verify_alg, p->verify_alg);
P
Philipp Reisner 已提交
3043 3044 3045 3046 3047 3048 3049 3050 3051 3052
				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;
			}
		}

3053
		if (apv >= 89 && strcmp(mdev->tconn->net_conf->csums_alg, p->csums_alg)) {
P
Philipp Reisner 已提交
3054 3055
			if (mdev->state.conn == C_WF_REPORT_PARAMS) {
				dev_err(DEV, "Different csums-alg settings. me=\"%s\" peer=\"%s\"\n",
3056
				    mdev->tconn->net_conf->csums_alg, p->csums_alg);
P
Philipp Reisner 已提交
3057 3058 3059 3060 3061 3062 3063 3064 3065 3066
				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;
			}
		}

3067 3068 3069 3070 3071 3072
		if (apv > 94 && get_ldev(mdev)) {
			mdev->ldev->dc.resync_rate = be32_to_cpu(p->rate);
			mdev->ldev->dc.c_plan_ahead = be32_to_cpu(p->c_plan_ahead);
			mdev->ldev->dc.c_delay_target = be32_to_cpu(p->c_delay_target);
			mdev->ldev->dc.c_fill_target = be32_to_cpu(p->c_fill_target);
			mdev->ldev->dc.c_max_rate = be32_to_cpu(p->c_max_rate);
3073

3074
			fifo_size = (mdev->ldev->dc.c_plan_ahead * 10 * SLEEP_TIME) / HZ;
3075 3076 3077 3078
			if (fifo_size != mdev->rs_plan_s.size && fifo_size > 0) {
				rs_plan_s   = kzalloc(sizeof(int) * fifo_size, GFP_KERNEL);
				if (!rs_plan_s) {
					dev_err(DEV, "kmalloc of fifo_buffer failed");
3079
					put_ldev(mdev);
3080 3081 3082
					goto disconnect;
				}
			}
3083
			put_ldev(mdev);
3084
		}
P
Philipp Reisner 已提交
3085 3086 3087 3088

		spin_lock(&mdev->peer_seq_lock);
		/* lock against drbd_nl_syncer_conf() */
		if (verify_tfm) {
3089 3090 3091 3092
			strcpy(mdev->tconn->net_conf->verify_alg, p->verify_alg);
			mdev->tconn->net_conf->verify_alg_len = strlen(p->verify_alg) + 1;
			crypto_free_hash(mdev->tconn->verify_tfm);
			mdev->tconn->verify_tfm = verify_tfm;
P
Philipp Reisner 已提交
3093 3094 3095
			dev_info(DEV, "using verify-alg: \"%s\"\n", p->verify_alg);
		}
		if (csums_tfm) {
3096 3097 3098 3099
			strcpy(mdev->tconn->net_conf->csums_alg, p->csums_alg);
			mdev->tconn->net_conf->csums_alg_len = strlen(p->csums_alg) + 1;
			crypto_free_hash(mdev->tconn->csums_tfm);
			mdev->tconn->csums_tfm = csums_tfm;
P
Philipp Reisner 已提交
3100 3101
			dev_info(DEV, "using csums-alg: \"%s\"\n", p->csums_alg);
		}
3102 3103 3104 3105 3106 3107
		if (fifo_size != mdev->rs_plan_s.size) {
			kfree(mdev->rs_plan_s.values);
			mdev->rs_plan_s.values = rs_plan_s;
			mdev->rs_plan_s.size   = fifo_size;
			mdev->rs_planed = 0;
		}
P
Philipp Reisner 已提交
3108 3109 3110 3111 3112 3113 3114 3115 3116 3117
		spin_unlock(&mdev->peer_seq_lock);
	}

	return ok;
disconnect:
	/* 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);
3118
	conn_request_state(mdev->tconn, NS(conn, C_DISCONNECTING), CS_HARD);
3119
	return false;
P
Philipp Reisner 已提交
3120 3121 3122 3123 3124 3125 3126 3127 3128 3129 3130 3131 3132 3133 3134
}

/* 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);
}

3135 3136
static int receive_sizes(struct drbd_conf *mdev, enum drbd_packet cmd,
			 unsigned int data_size)
P
Philipp Reisner 已提交
3137
{
3138
	struct p_sizes *p = &mdev->tconn->data.rbuf.sizes;
P
Philipp Reisner 已提交
3139 3140 3141
	enum determine_dev_size dd = unchanged;
	sector_t p_size, p_usize, my_usize;
	int ldsc = 0; /* local disk size changed */
3142
	enum dds_flags ddsf;
P
Philipp Reisner 已提交
3143 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

	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)) {
		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_usize, mdev->ldev->dc.disk_size);

		/* if this is the first connect, or an otherwise expected
		 * param exchange, choose the minimum */
		if (mdev->state.conn == C_WF_REPORT_PARAMS)
			p_usize = min_not_zero((sector_t)mdev->ldev->dc.disk_size,
					     p_usize);

		my_usize = mdev->ldev->dc.disk_size;

		if (mdev->ldev->dc.disk_size != p_usize) {
			mdev->ldev->dc.disk_size = p_usize;
			dev_info(DEV, "Peer sets u_size to %lu sectors\n",
			     (unsigned long)mdev->ldev->dc.disk_size);
		}

		/* Never shrink a device with usable data during connect.
		   But allow online shrinking if we are connected. */
3173
		if (drbd_new_dev_size(mdev, mdev->ldev, 0) <
P
Philipp Reisner 已提交
3174 3175 3176 3177
		   drbd_get_capacity(mdev->this_bdev) &&
		   mdev->state.disk >= D_OUTDATED &&
		   mdev->state.conn < C_CONNECTED) {
			dev_err(DEV, "The peer's disk size is too small!\n");
3178
			conn_request_state(mdev->tconn, NS(conn, C_DISCONNECTING), CS_HARD);
P
Philipp Reisner 已提交
3179 3180
			mdev->ldev->dc.disk_size = my_usize;
			put_ldev(mdev);
3181
			return false;
P
Philipp Reisner 已提交
3182 3183 3184 3185
		}
		put_ldev(mdev);
	}

3186
	ddsf = be16_to_cpu(p->dds_flags);
P
Philipp Reisner 已提交
3187
	if (get_ldev(mdev)) {
B
Bart Van Assche 已提交
3188
		dd = drbd_determine_dev_size(mdev, ddsf);
P
Philipp Reisner 已提交
3189 3190
		put_ldev(mdev);
		if (dd == dev_size_error)
3191
			return false;
P
Philipp Reisner 已提交
3192 3193 3194 3195 3196 3197
		drbd_md_sync(mdev);
	} else {
		/* I am diskless, need to accept the peer's size. */
		drbd_set_my_capacity(mdev, p_size);
	}

3198 3199 3200
	mdev->peer_max_bio_size = be32_to_cpu(p->max_bio_size);
	drbd_reconsider_max_bio_size(mdev);

P
Philipp Reisner 已提交
3201 3202 3203 3204 3205 3206 3207 3208 3209 3210 3211 3212 3213 3214
	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... */
3215
			drbd_send_sizes(mdev, 0, ddsf);
P
Philipp Reisner 已提交
3216 3217 3218 3219
		}
		if (test_and_clear_bit(RESIZE_PENDING, &mdev->flags) ||
		    (dd == grew && mdev->state.conn == C_CONNECTED)) {
			if (mdev->state.pdsk >= D_INCONSISTENT &&
3220 3221 3222 3223 3224 3225
			    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
3226 3227 3228 3229
				set_bit(RESYNC_AFTER_NEG, &mdev->flags);
		}
	}

3230
	return true;
P
Philipp Reisner 已提交
3231 3232
}

3233 3234
static int receive_uuids(struct drbd_conf *mdev, enum drbd_packet cmd,
			 unsigned int data_size)
P
Philipp Reisner 已提交
3235
{
3236
	struct p_uuids *p = &mdev->tconn->data.rbuf.uuids;
P
Philipp Reisner 已提交
3237
	u64 *p_uuid;
3238
	int i, updated_uuids = 0;
P
Philipp Reisner 已提交
3239 3240 3241 3242 3243 3244 3245 3246 3247 3248 3249 3250 3251 3252 3253

	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);
3254
		conn_request_state(mdev->tconn, NS(conn, C_DISCONNECTING), CS_HARD);
3255
		return false;
P
Philipp Reisner 已提交
3256 3257 3258 3259 3260
	}

	if (get_ldev(mdev)) {
		int skip_initial_sync =
			mdev->state.conn == C_CONNECTED &&
3261
			mdev->tconn->agreed_pro_version >= 90 &&
P
Philipp Reisner 已提交
3262 3263 3264 3265 3266
			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,
3267 3268
					"clear_n_write from receive_uuids",
					BM_LOCKED_TEST_ALLOWED);
P
Philipp Reisner 已提交
3269 3270 3271 3272 3273
			_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);
3274
			updated_uuids = 1;
P
Philipp Reisner 已提交
3275 3276
		}
		put_ldev(mdev);
3277 3278 3279 3280
	} 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. */
3281
		updated_uuids = drbd_set_ed_uuid(mdev, p_uuid[UI_CURRENT]);
P
Philipp Reisner 已提交
3282 3283 3284 3285 3286 3287
	}

	/* 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... */
3288 3289
	mutex_lock(mdev->state_mutex);
	mutex_unlock(mdev->state_mutex);
P
Philipp Reisner 已提交
3290
	if (mdev->state.conn >= C_CONNECTED && mdev->state.disk < D_INCONSISTENT)
3291 3292 3293 3294
		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
3295

3296
	return true;
P
Philipp Reisner 已提交
3297 3298 3299 3300 3301 3302 3303 3304 3305 3306 3307 3308 3309 3310 3311 3312 3313 3314 3315 3316 3317 3318 3319 3320 3321 3322 3323 3324 3325 3326 3327 3328
}

/**
 * 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[] = {
		[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;
}

3329 3330
static int receive_req_state(struct drbd_conf *mdev, enum drbd_packet cmd,
			     unsigned int data_size)
P
Philipp Reisner 已提交
3331
{
3332
	struct p_req_state *p = &mdev->tconn->data.rbuf.req_state;
P
Philipp Reisner 已提交
3333
	union drbd_state mask, val;
3334
	enum drbd_state_rv rv;
P
Philipp Reisner 已提交
3335 3336 3337 3338

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

3339
	if (test_bit(DISCARD_CONCURRENT, &mdev->tconn->flags) &&
3340
	    mutex_is_locked(mdev->state_mutex)) {
P
Philipp Reisner 已提交
3341
		drbd_send_sr_reply(mdev, SS_CONCURRENT_ST_CHG);
3342
		return true;
P
Philipp Reisner 已提交
3343 3344 3345 3346 3347
	}

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

3348 3349
	rv = drbd_change_state(mdev, CS_VERBOSE, mask, val);
	drbd_send_sr_reply(mdev, rv);
P
Philipp Reisner 已提交
3350 3351 3352

	drbd_md_sync(mdev);

3353
	return true;
P
Philipp Reisner 已提交
3354 3355
}

3356 3357 3358 3359 3360 3361 3362 3363 3364 3365 3366 3367 3368 3369 3370 3371 3372 3373 3374 3375 3376 3377 3378 3379 3380
static int receive_req_conn_state(struct drbd_tconn *tconn, enum drbd_packet cmd,
				  unsigned int data_size)
{
	struct p_req_state *p = &tconn->data.rbuf.req_state;
	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);
		return true;
	}

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

	rv = conn_request_state(tconn, mask, val, CS_VERBOSE | CS_LOCAL_ONLY);
	conn_send_sr_reply(tconn, rv);

	return true;
}

3381 3382
static int receive_state(struct drbd_conf *mdev, enum drbd_packet cmd,
			 unsigned int data_size)
P
Philipp Reisner 已提交
3383
{
3384
	struct p_state *p = &mdev->tconn->data.rbuf.state;
3385
	union drbd_state os, ns, peer_state;
P
Philipp Reisner 已提交
3386
	enum drbd_disk_state real_peer_disk;
3387
	enum chg_state_flags cs_flags;
P
Philipp Reisner 已提交
3388 3389 3390 3391 3392 3393 3394 3395 3396 3397
	int rv;

	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));
	}

3398
	spin_lock_irq(&mdev->tconn->req_lock);
P
Philipp Reisner 已提交
3399
 retry:
3400
	os = ns = mdev->state;
3401
	spin_unlock_irq(&mdev->tconn->req_lock);
P
Philipp Reisner 已提交
3402

3403 3404 3405 3406 3407 3408 3409 3410 3411 3412 3413 3414 3415 3416 3417 3418 3419 3420 3421 3422 3423
	/* peer says his disk is uptodate, while we think it is inconsistent,
	 * and this happens while we think we have a sync going on. */
	if (os.pdsk == D_INCONSISTENT && real_peer_disk == D_UP_TO_DATE &&
	    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);
3424
			return true;
3425 3426 3427 3428 3429 3430 3431 3432 3433 3434 3435 3436
		}
	}

	/* 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;

3437 3438
	if (ns.conn == C_WF_REPORT_PARAMS)
		ns.conn = C_CONNECTED;
P
Philipp Reisner 已提交
3439

3440 3441 3442
	if (peer_state.conn == C_AHEAD)
		ns.conn = C_BEHIND;

P
Philipp Reisner 已提交
3443 3444 3445 3446 3447
	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 */
3448
		cr  = (os.conn < C_CONNECTED);
P
Philipp Reisner 已提交
3449 3450
		/* if we had an established connection
		 * and one of the nodes newly attaches a disk */
3451
		cr |= (os.conn == C_CONNECTED &&
P
Philipp Reisner 已提交
3452
		       (peer_state.disk == D_NEGOTIATING ||
3453
			os.disk == D_NEGOTIATING));
P
Philipp Reisner 已提交
3454 3455 3456 3457 3458
		/* 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" */
3459
		cr |= (os.conn == C_CONNECTED &&
P
Philipp Reisner 已提交
3460 3461 3462 3463
				(peer_state.conn >= C_STARTING_SYNC_S &&
				 peer_state.conn <= C_WF_BITMAP_T));

		if (cr)
3464
			ns.conn = drbd_sync_handshake(mdev, peer_state.role, real_peer_disk);
P
Philipp Reisner 已提交
3465 3466

		put_ldev(mdev);
3467 3468
		if (ns.conn == C_MASK) {
			ns.conn = C_CONNECTED;
P
Philipp Reisner 已提交
3469
			if (mdev->state.disk == D_NEGOTIATING) {
3470
				drbd_force_state(mdev, NS(disk, D_FAILED));
P
Philipp Reisner 已提交
3471 3472 3473
			} 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;
3474
				real_peer_disk = D_DISKLESS;
P
Philipp Reisner 已提交
3475
			} else {
3476
				if (test_and_clear_bit(CONN_DRY_RUN, &mdev->tconn->flags))
3477
					return false;
3478
				D_ASSERT(os.conn == C_WF_REPORT_PARAMS);
3479
				conn_request_state(mdev->tconn, NS(conn, C_DISCONNECTING), CS_HARD);
3480
				return false;
P
Philipp Reisner 已提交
3481 3482 3483 3484
			}
		}
	}

3485
	spin_lock_irq(&mdev->tconn->req_lock);
3486
	if (mdev->state.i != os.i)
P
Philipp Reisner 已提交
3487 3488 3489 3490 3491
		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);
3492
	if ((ns.conn == C_CONNECTED || ns.conn == C_WF_BITMAP_S) && ns.disk == D_NEGOTIATING)
P
Philipp Reisner 已提交
3493
		ns.disk = mdev->new_state_tmp.disk;
3494 3495
	cs_flags = CS_VERBOSE + (os.conn < C_CONNECTED && ns.conn >= C_CONNECTED ? 0 : CS_HARD);
	if (ns.pdsk == D_CONSISTENT && is_susp(ns) && ns.conn == C_CONNECTED && os.conn < C_CONNECTED &&
3496
	    test_bit(NEW_CUR_UUID, &mdev->flags)) {
3497
		/* Do not allow tl_restart(RESEND) for a rebooted peer. We can only allow this
3498
		   for temporal network outages! */
3499
		spin_unlock_irq(&mdev->tconn->req_lock);
3500
		dev_err(DEV, "Aborting Connect, can not thaw IO with an only Consistent peer\n");
3501
		tl_clear(mdev->tconn);
3502 3503
		drbd_uuid_new_current(mdev);
		clear_bit(NEW_CUR_UUID, &mdev->flags);
3504
		conn_request_state(mdev->tconn, NS2(conn, C_PROTOCOL_ERROR, susp, 0), CS_HARD);
3505
		return false;
3506
	}
3507
	rv = _drbd_set_state(mdev, ns, cs_flags, NULL);
P
Philipp Reisner 已提交
3508
	ns = mdev->state;
3509
	spin_unlock_irq(&mdev->tconn->req_lock);
P
Philipp Reisner 已提交
3510 3511

	if (rv < SS_SUCCESS) {
3512
		conn_request_state(mdev->tconn, NS(conn, C_DISCONNECTING), CS_HARD);
3513
		return false;
P
Philipp Reisner 已提交
3514 3515
	}

3516 3517
	if (os.conn > C_WF_REPORT_PARAMS) {
		if (ns.conn > C_CONNECTED && peer_state.conn <= C_CONNECTED &&
P
Philipp Reisner 已提交
3518 3519 3520 3521 3522 3523 3524 3525 3526
		    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);
			drbd_send_state(mdev);
		}
	}

3527
	mdev->tconn->net_conf->want_lose = 0;
P
Philipp Reisner 已提交
3528 3529 3530

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

3531
	return true;
P
Philipp Reisner 已提交
3532 3533
}

3534 3535
static int receive_sync_uuid(struct drbd_conf *mdev, enum drbd_packet cmd,
			     unsigned int data_size)
P
Philipp Reisner 已提交
3536
{
3537
	struct p_rs_uuid *p = &mdev->tconn->data.rbuf.rs_uuid;
P
Philipp Reisner 已提交
3538 3539 3540

	wait_event(mdev->misc_wait,
		   mdev->state.conn == C_WF_SYNC_UUID ||
3541
		   mdev->state.conn == C_BEHIND ||
P
Philipp Reisner 已提交
3542 3543 3544 3545 3546 3547 3548 3549 3550 3551 3552
		   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);

3553
		drbd_print_uuids(mdev, "updated sync uuid");
P
Philipp Reisner 已提交
3554 3555 3556 3557 3558 3559
		drbd_start_resync(mdev, C_SYNC_TARGET);

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

3560
	return true;
P
Philipp Reisner 已提交
3561 3562
}

3563 3564 3565 3566 3567 3568 3569
/**
 * receive_bitmap_plain
 *
 * Return 0 when done, 1 when another iteration is needed, and a negative error
 * code upon failure.
 */
static int
3570 3571
receive_bitmap_plain(struct drbd_conf *mdev, unsigned int data_size,
		     unsigned long *buffer, struct bm_xfer_ctx *c)
P
Philipp Reisner 已提交
3572 3573 3574
{
	unsigned num_words = min_t(size_t, BM_PACKET_WORDS, c->bm_words - c->word_offset);
	unsigned want = num_words * sizeof(long);
3575
	int err;
P
Philipp Reisner 已提交
3576

3577 3578
	if (want != data_size) {
		dev_err(DEV, "%s:want (%u) != data_size (%u)\n", __func__, want, data_size);
3579
		return -EIO;
P
Philipp Reisner 已提交
3580 3581
	}
	if (want == 0)
3582
		return 0;
3583
	err = drbd_recv(mdev->tconn, buffer, want);
3584 3585 3586 3587 3588
	if (err != want) {
		if (err >= 0)
			err = -EIO;
		return err;
	}
P
Philipp Reisner 已提交
3589 3590 3591 3592 3593 3594 3595 3596

	drbd_bm_merge_lel(mdev, c->word_offset, num_words, buffer);

	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;

3597
	return 1;
P
Philipp Reisner 已提交
3598 3599
}

3600 3601 3602 3603 3604 3605 3606
/**
 * 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
3607 3608
recv_bm_rle_bits(struct drbd_conf *mdev,
		struct p_compressed_bm *p,
3609 3610
		 struct bm_xfer_ctx *c,
		 unsigned int len)
P
Philipp Reisner 已提交
3611 3612 3613 3614 3615 3616 3617 3618 3619 3620 3621 3622 3623 3624 3625
{
	struct bitstream bs;
	u64 look_ahead;
	u64 rl;
	u64 tmp;
	unsigned long s = c->bit_offset;
	unsigned long e;
	int toggle = DCBP_get_start(p);
	int have;
	int bits;

	bitstream_init(&bs, p->code, len, DCBP_get_pad_bits(p));

	bits = bitstream_get_bits(&bs, &look_ahead, 64);
	if (bits < 0)
3626
		return -EIO;
P
Philipp Reisner 已提交
3627 3628 3629 3630

	for (have = bits; have > 0; s += rl, toggle = !toggle) {
		bits = vli_decode_bits(&rl, look_ahead);
		if (bits <= 0)
3631
			return -EIO;
P
Philipp Reisner 已提交
3632 3633 3634 3635 3636

		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);
3637
				return -EIO;
P
Philipp Reisner 已提交
3638 3639 3640 3641 3642 3643 3644 3645 3646
			}
			_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);
3647
			return -EIO;
P
Philipp Reisner 已提交
3648 3649 3650 3651 3652 3653
		}
		look_ahead >>= bits;
		have -= bits;

		bits = bitstream_get_bits(&bs, &tmp, 64 - have);
		if (bits < 0)
3654
			return -EIO;
P
Philipp Reisner 已提交
3655 3656 3657 3658 3659 3660 3661
		look_ahead |= tmp << have;
		have += bits;
	}

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

3662
	return (s != c->bm_bits);
P
Philipp Reisner 已提交
3663 3664
}

3665 3666 3667 3668 3669 3670 3671
/**
 * 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
3672 3673
decode_bitmap_c(struct drbd_conf *mdev,
		struct p_compressed_bm *p,
3674 3675
		struct bm_xfer_ctx *c,
		unsigned int len)
P
Philipp Reisner 已提交
3676 3677
{
	if (DCBP_get_code(p) == RLE_VLI_Bits)
3678
		return recv_bm_rle_bits(mdev, p, c, len);
P
Philipp Reisner 已提交
3679 3680 3681 3682 3683 3684

	/* 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);
3685
	conn_request_state(mdev->tconn, NS(conn, C_PROTOCOL_ERROR), CS_HARD);
3686
	return -EIO;
P
Philipp Reisner 已提交
3687 3688 3689 3690 3691 3692
}

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" */
3693
	unsigned plain = sizeof(struct p_header) *
P
Philipp Reisner 已提交
3694 3695 3696 3697 3698 3699 3700 3701 3702 3703 3704 3705 3706 3707 3708 3709 3710 3711 3712 3713 3714 3715 3716 3717 3718 3719 3720 3721 3722 3723 3724 3725 3726 3727 3728 3729 3730
		((c->bm_words+BM_PACKET_WORDS-1)/BM_PACKET_WORDS+1)
		+ c->bm_words * sizeof(long);
	unsigned total = c->bytes[0] + c->bytes[1];
	unsigned r;

	/* 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. */
3731 3732
static int receive_bitmap(struct drbd_conf *mdev, enum drbd_packet cmd,
			  unsigned int data_size)
P
Philipp Reisner 已提交
3733 3734 3735
{
	struct bm_xfer_ctx c;
	void *buffer;
3736
	int err;
3737
	int ok = false;
3738
	struct p_header *h = &mdev->tconn->data.rbuf.header;
3739
	struct packet_info pi;
P
Philipp Reisner 已提交
3740

3741 3742 3743
	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
3744 3745 3746 3747 3748 3749 3750 3751 3752 3753 3754 3755 3756 3757

	/* maybe we should use some per thread scratch page,
	 * and allocate that during initial device creation? */
	buffer	 = (unsigned long *) __get_free_page(GFP_NOIO);
	if (!buffer) {
		dev_err(DEV, "failed to allocate one page buffer in %s\n", __func__);
		goto out;
	}

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

3758
	for(;;) {
3759
		if (cmd == P_BITMAP) {
3760
			err = receive_bitmap_plain(mdev, data_size, buffer, &c);
3761
		} else if (cmd == P_COMPRESSED_BITMAP) {
P
Philipp Reisner 已提交
3762 3763 3764 3765
			/* MAYBE: sanity check that we speak proto >= 90,
			 * and the feature is enabled! */
			struct p_compressed_bm *p;

3766
			if (data_size > BM_PACKET_PAYLOAD_BYTES) {
P
Philipp Reisner 已提交
3767 3768 3769 3770 3771 3772
				dev_err(DEV, "ReportCBitmap packet too large\n");
				goto out;
			}
			/* use the page buff */
			p = buffer;
			memcpy(p, h, sizeof(*h));
3773
			if (drbd_recv(mdev->tconn, p->head.payload, data_size) != data_size)
P
Philipp Reisner 已提交
3774
				goto out;
3775 3776
			if (data_size <= (sizeof(*p) - sizeof(p->head))) {
				dev_err(DEV, "ReportCBitmap packet too small (l:%u)\n", data_size);
3777
				goto out;
P
Philipp Reisner 已提交
3778
			}
3779
			err = decode_bitmap_c(mdev, p, &c, data_size);
P
Philipp Reisner 已提交
3780
		} else {
3781
			dev_warn(DEV, "receive_bitmap: cmd neither ReportBitMap nor ReportCBitMap (is 0x%x)", cmd);
P
Philipp Reisner 已提交
3782 3783 3784
			goto out;
		}

3785
		c.packets[cmd == P_BITMAP]++;
3786
		c.bytes[cmd == P_BITMAP] += sizeof(struct p_header) + data_size;
P
Philipp Reisner 已提交
3787

3788 3789 3790
		if (err <= 0) {
			if (err < 0)
				goto out;
P
Philipp Reisner 已提交
3791
			break;
3792
		}
3793
		if (!drbd_recv_header(mdev->tconn, &pi))
P
Philipp Reisner 已提交
3794
			goto out;
3795 3796
		cmd = pi.cmd;
		data_size = pi.size;
3797
	}
P
Philipp Reisner 已提交
3798 3799 3800 3801

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

	if (mdev->state.conn == C_WF_BITMAP_T) {
3802 3803
		enum drbd_state_rv rv;

P
Philipp Reisner 已提交
3804 3805 3806 3807
		ok = !drbd_send_bitmap(mdev);
		if (!ok)
			goto out;
		/* Omit CS_ORDERED with this state transition to avoid deadlocks. */
3808 3809
		rv = _drbd_request_state(mdev, NS(conn, C_WF_SYNC_UUID), CS_VERBOSE);
		D_ASSERT(rv == SS_SUCCESS);
P
Philipp Reisner 已提交
3810 3811 3812 3813 3814 3815 3816
	} 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));
	}

3817
	ok = true;
P
Philipp Reisner 已提交
3818
 out:
3819
	drbd_bm_unlock(mdev);
P
Philipp Reisner 已提交
3820 3821 3822 3823 3824 3825
	if (ok && mdev->state.conn == C_WF_BITMAP_S)
		drbd_start_resync(mdev, C_SYNC_SOURCE);
	free_page((unsigned long) buffer);
	return ok;
}

3826
static int _tconn_receive_skip(struct drbd_tconn *tconn, unsigned int data_size)
P
Philipp Reisner 已提交
3827 3828 3829 3830 3831
{
	/* TODO zero copy sink :) */
	static char sink[128];
	int size, want, r;

3832
	size = data_size;
P
Philipp Reisner 已提交
3833 3834
	while (size > 0) {
		want = min_t(int, size, sizeof(sink));
3835 3836
		r = drbd_recv(tconn, sink, want);
		if (r <= 0)
3837
			break;
P
Philipp Reisner 已提交
3838 3839 3840 3841 3842
		size -= r;
	}
	return size == 0;
}

3843 3844 3845 3846 3847 3848 3849 3850 3851 3852 3853 3854 3855 3856 3857 3858 3859
static int receive_skip(struct drbd_conf *mdev, enum drbd_packet cmd,
			unsigned int data_size)
{
	dev_warn(DEV, "skipping unknown optional packet type %d, l: %d!\n",
		 cmd, data_size);

	return _tconn_receive_skip(mdev->tconn, data_size);
}

static int tconn_receive_skip(struct drbd_tconn *tconn, enum drbd_packet cmd, unsigned int data_size)
{
	conn_warn(tconn, "skipping packet for non existing volume type %d, l: %d!\n",
		  cmd, data_size);

	return _tconn_receive_skip(tconn, data_size);
}

3860 3861
static int receive_UnplugRemote(struct drbd_conf *mdev, enum drbd_packet cmd,
				unsigned int data_size)
3862
{
3863 3864
	/* Make sure we've acked all the TCP data associated
	 * with the data requests being unplugged */
3865
	drbd_tcp_quickack(mdev->tconn->data.socket);
3866

3867
	return true;
3868 3869
}

3870 3871
static int receive_out_of_sync(struct drbd_conf *mdev, enum drbd_packet cmd,
			       unsigned int data_size)
3872
{
3873
	struct p_block_desc *p = &mdev->tconn->data.rbuf.block_desc;
3874

3875 3876 3877 3878 3879 3880 3881 3882 3883 3884
	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));
	}

3885 3886
	drbd_set_out_of_sync(mdev, be64_to_cpu(p->sector), be32_to_cpu(p->blksize));

3887
	return true;
3888 3889
}

3890 3891 3892
struct data_cmd {
	int expect_payload;
	size_t pkt_size;
3893
	enum mdev_or_conn fa_type; /* first argument's type */
3894 3895 3896 3897 3898 3899
	union {
		int (*mdev_fn)(struct drbd_conf *, enum drbd_packet cmd,
				  unsigned int to_receive);
		int (*conn_fn)(struct drbd_tconn *, enum drbd_packet cmd,
				  unsigned int to_receive);
	};
3900 3901 3902
};

static struct data_cmd drbd_cmd_handler[] = {
3903 3904 3905 3906 3907 3908 3909 3910 3911 3912 3913
	[P_DATA]	    = { 1, sizeof(struct p_data), MDEV, { receive_Data } },
	[P_DATA_REPLY]	    = { 1, sizeof(struct p_data), MDEV, { receive_DataReply } },
	[P_RS_DATA_REPLY]   = { 1, sizeof(struct p_data), MDEV, { receive_RSDataReply } } ,
	[P_BARRIER]	    = { 0, sizeof(struct p_barrier), MDEV, { receive_Barrier } } ,
	[P_BITMAP]	    = { 1, sizeof(struct p_header), MDEV, { receive_bitmap } } ,
	[P_COMPRESSED_BITMAP] = { 1, sizeof(struct p_header), MDEV, { receive_bitmap } } ,
	[P_UNPLUG_REMOTE]   = { 0, sizeof(struct p_header), MDEV, { receive_UnplugRemote } },
	[P_DATA_REQUEST]    = { 0, sizeof(struct p_block_req), MDEV, { receive_DataRequest } },
	[P_RS_DATA_REQUEST] = { 0, sizeof(struct p_block_req), MDEV, { receive_DataRequest } },
	[P_SYNC_PARAM]	    = { 1, sizeof(struct p_header), MDEV, { receive_SyncParam } },
	[P_SYNC_PARAM89]    = { 1, sizeof(struct p_header), MDEV, { receive_SyncParam } },
3914
	[P_PROTOCOL]        = { 1, sizeof(struct p_protocol), CONN, { .conn_fn = receive_protocol } },
3915 3916 3917 3918 3919 3920 3921 3922 3923 3924
	[P_UUIDS]	    = { 0, sizeof(struct p_uuids), MDEV, { receive_uuids } },
	[P_SIZES]	    = { 0, sizeof(struct p_sizes), MDEV, { receive_sizes } },
	[P_STATE]	    = { 0, sizeof(struct p_state), MDEV, { receive_state } },
	[P_STATE_CHG_REQ]   = { 0, sizeof(struct p_req_state), MDEV, { receive_req_state } },
	[P_SYNC_UUID]       = { 0, sizeof(struct p_rs_uuid), MDEV, { receive_sync_uuid } },
	[P_OV_REQUEST]      = { 0, sizeof(struct p_block_req), MDEV, { receive_DataRequest } },
	[P_OV_REPLY]        = { 1, sizeof(struct p_block_req), MDEV, { receive_DataRequest } },
	[P_CSUM_RS_REQUEST] = { 1, sizeof(struct p_block_req), MDEV, { receive_DataRequest } },
	[P_DELAY_PROBE]     = { 0, sizeof(struct p_delay_probe93), MDEV, { receive_skip } },
	[P_OUT_OF_SYNC]     = { 0, sizeof(struct p_block_desc), MDEV, { receive_out_of_sync } },
3925
	[P_CONN_ST_CHG_REQ] = { 0, sizeof(struct p_req_state), CONN, { .conn_fn = receive_req_conn_state } },
P
Philipp Reisner 已提交
3926 3927
};

3928
/* All handler functions that expect a sub-header get that sub-heder in
3929
   mdev->tconn->data.rbuf.header.head.payload.
3930

3931
   Usually in mdev->tconn->data.rbuf.header.head the callback can find the usual
3932 3933
   p_header, but they may not rely on that. Since there is also p_header95 !
 */
P
Philipp Reisner 已提交
3934

3935
static void drbdd(struct drbd_tconn *tconn)
P
Philipp Reisner 已提交
3936
{
3937
	struct p_header *header = &tconn->data.rbuf.header;
3938
	struct packet_info pi;
3939 3940
	size_t shs; /* sub header size */
	int rv;
P
Philipp Reisner 已提交
3941

3942 3943 3944
	while (get_t_state(&tconn->receiver) == RUNNING) {
		drbd_thread_current_set_cpu(&tconn->receiver);
		if (!drbd_recv_header(tconn, &pi))
3945
			goto err_out;
P
Philipp Reisner 已提交
3946

A
Andreas Gruenbacher 已提交
3947
		if (unlikely(pi.cmd >= ARRAY_SIZE(drbd_cmd_handler) ||
3948
		    !drbd_cmd_handler[pi.cmd].mdev_fn)) {
3949
			conn_err(tconn, "unknown packet type %d, l: %d!\n", pi.cmd, pi.size);
3950
			goto err_out;
3951
		}
P
Philipp Reisner 已提交
3952

3953 3954
		shs = drbd_cmd_handler[pi.cmd].pkt_size - sizeof(struct p_header);
		if (pi.size - shs > 0 && !drbd_cmd_handler[pi.cmd].expect_payload) {
3955
			conn_err(tconn, "No payload expected %s l:%d\n", cmdname(pi.cmd), pi.size);
3956
			goto err_out;
P
Philipp Reisner 已提交
3957 3958
		}

3959
		if (shs) {
3960
			rv = drbd_recv(tconn, &header->payload, shs);
3961
			if (unlikely(rv != shs)) {
3962
				if (!signal_pending(current))
3963
					conn_warn(tconn, "short read while reading sub header: rv=%d\n", rv);
3964 3965 3966 3967
				goto err_out;
			}
		}

3968
		if (drbd_cmd_handler[pi.cmd].fa_type == CONN) {
3969 3970 3971 3972 3973 3974 3975
			rv = drbd_cmd_handler[pi.cmd].conn_fn(tconn, pi.cmd, pi.size - shs);
		} else {
			struct drbd_conf *mdev = vnr_to_mdev(tconn, pi.vnr);
			rv = mdev ?
				drbd_cmd_handler[pi.cmd].mdev_fn(mdev, pi.cmd, pi.size - shs) :
				tconn_receive_skip(tconn, pi.cmd, pi.size - shs);
		}
P
Philipp Reisner 已提交
3976

3977
		if (unlikely(!rv)) {
3978
			conn_err(tconn, "error receiving %s, l: %d!\n",
3979
			    cmdname(pi.cmd), pi.size);
3980
			goto err_out;
P
Philipp Reisner 已提交
3981 3982 3983
		}
	}

3984 3985
	if (0) {
	err_out:
3986
		conn_request_state(tconn, NS(conn, C_PROTOCOL_ERROR), CS_HARD);
3987
	}
P
Philipp Reisner 已提交
3988 3989
}

3990
void conn_flush_workqueue(struct drbd_tconn *tconn)
P
Philipp Reisner 已提交
3991 3992 3993 3994
{
	struct drbd_wq_barrier barr;

	barr.w.cb = w_prev_work_done;
3995
	barr.w.tconn = tconn;
P
Philipp Reisner 已提交
3996
	init_completion(&barr.done);
3997
	drbd_queue_work(&tconn->data.work, &barr.w);
P
Philipp Reisner 已提交
3998 3999 4000
	wait_for_completion(&barr.done);
}

4001
static void drbd_disconnect(struct drbd_tconn *tconn)
P
Philipp Reisner 已提交
4002
{
4003
	enum drbd_conns oc;
P
Philipp Reisner 已提交
4004 4005
	int rv = SS_UNKNOWN_ERROR;

4006
	if (tconn->cstate == C_STANDALONE)
P
Philipp Reisner 已提交
4007 4008 4009
		return;

	/* asender does not clean up anything. it must not interfere, either */
4010 4011 4012 4013 4014 4015 4016 4017
	drbd_thread_stop(&tconn->asender);
	drbd_free_sock(tconn);

	idr_for_each(&tconn->volumes, drbd_disconnected, tconn);

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

	spin_lock_irq(&tconn->req_lock);
4018 4019 4020 4021
	oc = tconn->cstate;
	if (oc >= C_UNCONNECTED)
		rv = _conn_request_state(tconn, NS(conn, C_UNCONNECTED), CS_VERBOSE);

4022 4023
	spin_unlock_irq(&tconn->req_lock);

4024
	if (oc == C_DISCONNECTING) {
4025 4026 4027 4028 4029 4030 4031
		wait_event(tconn->net_cnt_wait, atomic_read(&tconn->net_cnt) == 0);

		crypto_free_hash(tconn->cram_hmac_tfm);
		tconn->cram_hmac_tfm = NULL;

		kfree(tconn->net_conf);
		tconn->net_conf = NULL;
4032
		conn_request_state(tconn, NS(conn, C_STANDALONE), CS_VERBOSE);
4033 4034 4035 4036 4037 4038 4039 4040
	}
}

static int drbd_disconnected(int vnr, void *p, void *data)
{
	struct drbd_conf *mdev = (struct drbd_conf *)p;
	enum drbd_fencing_p fp;
	unsigned int i;
P
Philipp Reisner 已提交
4041

4042
	/* wait for current activity to cease. */
4043
	spin_lock_irq(&mdev->tconn->req_lock);
P
Philipp Reisner 已提交
4044 4045 4046
	_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);
4047
	spin_unlock_irq(&mdev->tconn->req_lock);
P
Philipp Reisner 已提交
4048 4049 4050 4051 4052 4053 4054 4055 4056 4057 4058 4059 4060 4061 4062 4063 4064

	/* 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);

4065 4066
	del_timer(&mdev->request_timer);

P
Philipp Reisner 已提交
4067 4068 4069 4070 4071 4072
	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" */
4073
	drbd_flush_workqueue(mdev);
P
Philipp Reisner 已提交
4074 4075 4076 4077 4078 4079 4080 4081

	/* This also does reclaim_net_ee().  If we do this too early, we might
	 * miss some resync ee and pages.*/
	drbd_process_done_ee(mdev);

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

4082
	if (!is_susp(mdev->state))
4083
		tl_clear(mdev->tconn);
P
Philipp Reisner 已提交
4084 4085 4086 4087 4088 4089 4090 4091 4092

	drbd_md_sync(mdev);

	fp = FP_DONT_CARE;
	if (get_ldev(mdev)) {
		fp = mdev->ldev->dc.fencing;
		put_ldev(mdev);
	}

4093 4094
	if (mdev->state.role == R_PRIMARY && fp >= FP_RESOURCE && mdev->state.pdsk >= D_UNKNOWN)
		drbd_try_outdate_peer_async(mdev);
P
Philipp Reisner 已提交
4095

4096 4097 4098 4099
	/* 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
4100 4101 4102 4103 4104 4105 4106 4107 4108 4109
	/* 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.
	 */
	i = drbd_release_ee(mdev, &mdev->net_ee);
	if (i)
		dev_info(DEV, "net_ee not empty, killed %u entries\n", i);
4110 4111 4112
	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
4113 4114
	i = atomic_read(&mdev->pp_in_use);
	if (i)
4115
		dev_info(DEV, "pp_in_use = %d, expected 0\n", i);
P
Philipp Reisner 已提交
4116 4117 4118 4119 4120 4121 4122 4123 4124

	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));

	/* ok, no more ee's on the fly, it is safe to reset the epoch_size */
	atomic_set(&mdev->current_epoch->epoch_size, 0);
	D_ASSERT(list_empty(&mdev->current_epoch->list));
4125 4126

	return 0;
P
Philipp Reisner 已提交
4127 4128 4129 4130 4131 4132 4133 4134 4135 4136 4137
}

/*
 * 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.
 */
4138
static int drbd_send_handshake(struct drbd_tconn *tconn)
P
Philipp Reisner 已提交
4139
{
4140
	/* ASSERT current == mdev->tconn->receiver ... */
4141
	struct p_handshake *p = &tconn->data.sbuf.handshake;
4142
	int err;
P
Philipp Reisner 已提交
4143

4144 4145
	if (mutex_lock_interruptible(&tconn->data.mutex)) {
		conn_err(tconn, "interrupted during initial handshake\n");
4146
		return -EINTR;
P
Philipp Reisner 已提交
4147 4148
	}

4149 4150
	if (tconn->data.socket == NULL) {
		mutex_unlock(&tconn->data.mutex);
4151
		return -EIO;
P
Philipp Reisner 已提交
4152 4153 4154 4155 4156
	}

	memset(p, 0, sizeof(*p));
	p->protocol_min = cpu_to_be32(PRO_VERSION_MIN);
	p->protocol_max = cpu_to_be32(PRO_VERSION_MAX);
4157
	err = _conn_send_cmd(tconn, 0, tconn->data.socket, P_HAND_SHAKE,
4158
			     &p->head, sizeof(*p), 0);
4159
	mutex_unlock(&tconn->data.mutex);
4160
	return err;
P
Philipp Reisner 已提交
4161 4162 4163 4164 4165 4166 4167 4168 4169
}

/*
 * 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.
 */
4170
static int drbd_do_handshake(struct drbd_tconn *tconn)
P
Philipp Reisner 已提交
4171
{
4172 4173
	/* ASSERT current == tconn->receiver ... */
	struct p_handshake *p = &tconn->data.rbuf.handshake;
4174
	const int expect = sizeof(struct p_handshake) - sizeof(struct p_header80);
4175
	struct packet_info pi;
4176
	int err, rv;
P
Philipp Reisner 已提交
4177

4178 4179
	err = drbd_send_handshake(tconn);
	if (err)
P
Philipp Reisner 已提交
4180 4181
		return 0;

4182
	rv = drbd_recv_header(tconn, &pi);
P
Philipp Reisner 已提交
4183 4184 4185
	if (!rv)
		return 0;

4186
	if (pi.cmd != P_HAND_SHAKE) {
4187
		conn_err(tconn, "expected HandShake packet, received: %s (0x%04x)\n",
4188
		     cmdname(pi.cmd), pi.cmd);
P
Philipp Reisner 已提交
4189 4190 4191
		return -1;
	}

4192
	if (pi.size != expect) {
4193
		conn_err(tconn, "expected HandShake length: %u, received: %u\n",
4194
		     expect, pi.size);
P
Philipp Reisner 已提交
4195 4196 4197
		return -1;
	}

4198
	rv = drbd_recv(tconn, &p->head.payload, expect);
P
Philipp Reisner 已提交
4199 4200

	if (rv != expect) {
4201
		if (!signal_pending(current))
4202
			conn_warn(tconn, "short read receiving handshake packet: l=%u\n", rv);
P
Philipp Reisner 已提交
4203 4204 4205 4206 4207 4208 4209 4210 4211 4212 4213 4214
		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;

4215
	tconn->agreed_pro_version = min_t(int, PRO_VERSION_MAX, p->protocol_max);
P
Philipp Reisner 已提交
4216

4217 4218
	conn_info(tconn, "Handshake successful: "
	     "Agreed network protocol version %d\n", tconn->agreed_pro_version);
P
Philipp Reisner 已提交
4219 4220 4221 4222

	return 1;

 incompat:
4223
	conn_err(tconn, "incompatible DRBD dialects: "
P
Philipp Reisner 已提交
4224 4225 4226 4227 4228 4229 4230
	    "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)
4231
static int drbd_do_auth(struct drbd_tconn *tconn)
P
Philipp Reisner 已提交
4232 4233 4234
{
	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");
4235
	return -1;
P
Philipp Reisner 已提交
4236 4237 4238
}
#else
#define CHALLENGE_LEN 64
4239 4240 4241 4242 4243 4244 4245

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

4246
static int drbd_do_auth(struct drbd_tconn *tconn)
P
Philipp Reisner 已提交
4247 4248 4249 4250 4251 4252
{
	char my_challenge[CHALLENGE_LEN];  /* 64 Bytes... */
	struct scatterlist sg;
	char *response = NULL;
	char *right_response = NULL;
	char *peers_ch = NULL;
4253
	unsigned int key_len = strlen(tconn->net_conf->shared_secret);
P
Philipp Reisner 已提交
4254 4255
	unsigned int resp_size;
	struct hash_desc desc;
4256
	struct packet_info pi;
P
Philipp Reisner 已提交
4257 4258
	int rv;

4259
	desc.tfm = tconn->cram_hmac_tfm;
P
Philipp Reisner 已提交
4260 4261
	desc.flags = 0;

4262 4263
	rv = crypto_hash_setkey(tconn->cram_hmac_tfm,
				(u8 *)tconn->net_conf->shared_secret, key_len);
P
Philipp Reisner 已提交
4264
	if (rv) {
4265
		conn_err(tconn, "crypto_hash_setkey() failed with %d\n", rv);
4266
		rv = -1;
P
Philipp Reisner 已提交
4267 4268 4269 4270 4271
		goto fail;
	}

	get_random_bytes(my_challenge, CHALLENGE_LEN);

4272
	rv = !conn_send_cmd2(tconn, P_AUTH_CHALLENGE, my_challenge, CHALLENGE_LEN);
P
Philipp Reisner 已提交
4273 4274 4275
	if (!rv)
		goto fail;

4276
	rv = drbd_recv_header(tconn, &pi);
P
Philipp Reisner 已提交
4277 4278 4279
	if (!rv)
		goto fail;

4280
	if (pi.cmd != P_AUTH_CHALLENGE) {
4281
		conn_err(tconn, "expected AuthChallenge packet, received: %s (0x%04x)\n",
4282
		    cmdname(pi.cmd), pi.cmd);
P
Philipp Reisner 已提交
4283 4284 4285 4286
		rv = 0;
		goto fail;
	}

4287
	if (pi.size > CHALLENGE_LEN * 2) {
4288
		conn_err(tconn, "expected AuthChallenge payload too big.\n");
4289
		rv = -1;
P
Philipp Reisner 已提交
4290 4291 4292
		goto fail;
	}

4293
	peers_ch = kmalloc(pi.size, GFP_NOIO);
P
Philipp Reisner 已提交
4294
	if (peers_ch == NULL) {
4295
		conn_err(tconn, "kmalloc of peers_ch failed\n");
4296
		rv = -1;
P
Philipp Reisner 已提交
4297 4298 4299
		goto fail;
	}

4300
	rv = drbd_recv(tconn, peers_ch, pi.size);
P
Philipp Reisner 已提交
4301

4302
	if (rv != pi.size) {
4303
		if (!signal_pending(current))
4304
			conn_warn(tconn, "short read AuthChallenge: l=%u\n", rv);
P
Philipp Reisner 已提交
4305 4306 4307 4308
		rv = 0;
		goto fail;
	}

4309
	resp_size = crypto_hash_digestsize(tconn->cram_hmac_tfm);
P
Philipp Reisner 已提交
4310 4311
	response = kmalloc(resp_size, GFP_NOIO);
	if (response == NULL) {
4312
		conn_err(tconn, "kmalloc of response failed\n");
4313
		rv = -1;
P
Philipp Reisner 已提交
4314 4315 4316 4317
		goto fail;
	}

	sg_init_table(&sg, 1);
4318
	sg_set_buf(&sg, peers_ch, pi.size);
P
Philipp Reisner 已提交
4319 4320 4321

	rv = crypto_hash_digest(&desc, &sg, sg.length, response);
	if (rv) {
4322
		conn_err(tconn, "crypto_hash_digest() failed with %d\n", rv);
4323
		rv = -1;
P
Philipp Reisner 已提交
4324 4325 4326
		goto fail;
	}

4327
	rv = !conn_send_cmd2(tconn, P_AUTH_RESPONSE, response, resp_size);
P
Philipp Reisner 已提交
4328 4329 4330
	if (!rv)
		goto fail;

4331
	rv = drbd_recv_header(tconn, &pi);
P
Philipp Reisner 已提交
4332 4333 4334
	if (!rv)
		goto fail;

4335
	if (pi.cmd != P_AUTH_RESPONSE) {
4336
		conn_err(tconn, "expected AuthResponse packet, received: %s (0x%04x)\n",
4337
			cmdname(pi.cmd), pi.cmd);
P
Philipp Reisner 已提交
4338 4339 4340 4341
		rv = 0;
		goto fail;
	}

4342
	if (pi.size != resp_size) {
4343
		conn_err(tconn, "expected AuthResponse payload of wrong size\n");
P
Philipp Reisner 已提交
4344 4345 4346 4347
		rv = 0;
		goto fail;
	}

4348
	rv = drbd_recv(tconn, response , resp_size);
P
Philipp Reisner 已提交
4349 4350

	if (rv != resp_size) {
4351
		if (!signal_pending(current))
4352
			conn_warn(tconn, "short read receiving AuthResponse: l=%u\n", rv);
P
Philipp Reisner 已提交
4353 4354 4355 4356 4357
		rv = 0;
		goto fail;
	}

	right_response = kmalloc(resp_size, GFP_NOIO);
4358
	if (right_response == NULL) {
4359
		conn_err(tconn, "kmalloc of right_response failed\n");
4360
		rv = -1;
P
Philipp Reisner 已提交
4361 4362 4363 4364 4365 4366 4367
		goto fail;
	}

	sg_set_buf(&sg, my_challenge, CHALLENGE_LEN);

	rv = crypto_hash_digest(&desc, &sg, sg.length, right_response);
	if (rv) {
4368
		conn_err(tconn, "crypto_hash_digest() failed with %d\n", rv);
4369
		rv = -1;
P
Philipp Reisner 已提交
4370 4371 4372 4373 4374 4375
		goto fail;
	}

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

	if (rv)
4376 4377
		conn_info(tconn, "Peer authenticated using %d bytes of '%s' HMAC\n",
		     resp_size, tconn->net_conf->cram_hmac_alg);
4378 4379
	else
		rv = -1;
P
Philipp Reisner 已提交
4380 4381 4382 4383 4384 4385 4386 4387 4388 4389 4390 4391

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

	return rv;
}
#endif

int drbdd_init(struct drbd_thread *thi)
{
4392
	struct drbd_tconn *tconn = thi->tconn;
P
Philipp Reisner 已提交
4393 4394
	int h;

4395
	conn_info(tconn, "receiver (re)started\n");
P
Philipp Reisner 已提交
4396 4397

	do {
4398
		h = drbd_connect(tconn);
P
Philipp Reisner 已提交
4399
		if (h == 0) {
4400
			drbd_disconnect(tconn);
4401
			schedule_timeout_interruptible(HZ);
P
Philipp Reisner 已提交
4402 4403
		}
		if (h == -1) {
4404
			conn_warn(tconn, "Discarding network configuration.\n");
4405
			conn_request_state(tconn, NS(conn, C_DISCONNECTING), CS_HARD);
P
Philipp Reisner 已提交
4406 4407 4408 4409
		}
	} while (h == 0);

	if (h > 0) {
4410 4411 4412
		if (get_net_conf(tconn)) {
			drbdd(tconn);
			put_net_conf(tconn);
P
Philipp Reisner 已提交
4413 4414 4415
		}
	}

4416
	drbd_disconnect(tconn);
P
Philipp Reisner 已提交
4417

4418
	conn_info(tconn, "receiver terminated\n");
P
Philipp Reisner 已提交
4419 4420 4421 4422 4423
	return 0;
}

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

4424 4425 4426 4427 4428 4429 4430 4431 4432 4433 4434 4435 4436 4437 4438 4439 4440
static int got_conn_RqSReply(struct drbd_tconn *tconn, enum drbd_packet cmd)
{
	struct p_req_state_reply *p = &tconn->meta.rbuf.req_state_reply;
	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);

	return true;
}

4441
static int got_RqSReply(struct drbd_conf *mdev, enum drbd_packet cmd)
P
Philipp Reisner 已提交
4442
{
4443
	struct p_req_state_reply *p = &mdev->tconn->meta.rbuf.req_state_reply;
P
Philipp Reisner 已提交
4444 4445
	int retcode = be32_to_cpu(p->retcode);

4446 4447 4448 4449 4450 4451
	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
4452
	}
4453 4454
	wake_up(&mdev->state_wait);

4455
	return true;
P
Philipp Reisner 已提交
4456 4457
}

4458
static int got_Ping(struct drbd_tconn *tconn, enum drbd_packet cmd)
P
Philipp Reisner 已提交
4459
{
4460
	return drbd_send_ping_ack(tconn);
P
Philipp Reisner 已提交
4461 4462 4463

}

4464
static int got_PingAck(struct drbd_tconn *tconn, enum drbd_packet cmd)
P
Philipp Reisner 已提交
4465 4466
{
	/* restore idle timeout */
4467 4468 4469
	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
4470

4471
	return true;
P
Philipp Reisner 已提交
4472 4473
}

4474
static int got_IsInSync(struct drbd_conf *mdev, enum drbd_packet cmd)
P
Philipp Reisner 已提交
4475
{
4476
	struct p_block_ack *p = &mdev->tconn->meta.rbuf.block_ack;
P
Philipp Reisner 已提交
4477 4478 4479
	sector_t sector = be64_to_cpu(p->sector);
	int blksize = be32_to_cpu(p->blksize);

4480
	D_ASSERT(mdev->tconn->agreed_pro_version >= 89);
P
Philipp Reisner 已提交
4481 4482 4483

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

4484 4485 4486 4487 4488 4489 4490
	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
4491
	dec_rs_pending(mdev);
4492
	atomic_add(blksize >> 9, &mdev->rs_sect_in);
P
Philipp Reisner 已提交
4493

4494
	return true;
P
Philipp Reisner 已提交
4495 4496
}

4497 4498 4499 4500
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
4501 4502 4503 4504
{
	struct drbd_request *req;
	struct bio_and_error m;

4505
	spin_lock_irq(&mdev->tconn->req_lock);
4506
	req = find_request(mdev, root, id, sector, missing_ok, func);
P
Philipp Reisner 已提交
4507
	if (unlikely(!req)) {
4508
		spin_unlock_irq(&mdev->tconn->req_lock);
4509
		return false;
P
Philipp Reisner 已提交
4510 4511
	}
	__req_mod(req, what, &m);
4512
	spin_unlock_irq(&mdev->tconn->req_lock);
P
Philipp Reisner 已提交
4513 4514 4515

	if (m.bio)
		complete_master_bio(mdev, &m);
4516
	return true;
P
Philipp Reisner 已提交
4517 4518
}

4519
static int got_BlockAck(struct drbd_conf *mdev, enum drbd_packet cmd)
P
Philipp Reisner 已提交
4520
{
4521
	struct p_block_ack *p = &mdev->tconn->meta.rbuf.block_ack;
P
Philipp Reisner 已提交
4522 4523 4524 4525 4526 4527
	sector_t sector = be64_to_cpu(p->sector);
	int blksize = be32_to_cpu(p->blksize);
	enum drbd_req_event what;

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

4528
	if (p->block_id == ID_SYNCER) {
P
Philipp Reisner 已提交
4529 4530
		drbd_set_in_sync(mdev, sector, blksize);
		dec_rs_pending(mdev);
4531
		return true;
P
Philipp Reisner 已提交
4532
	}
4533
	switch (cmd) {
P
Philipp Reisner 已提交
4534
	case P_RS_WRITE_ACK:
4535
		D_ASSERT(mdev->tconn->net_conf->wire_protocol == DRBD_PROT_C);
4536
		what = WRITE_ACKED_BY_PEER_AND_SIS;
P
Philipp Reisner 已提交
4537 4538
		break;
	case P_WRITE_ACK:
4539
		D_ASSERT(mdev->tconn->net_conf->wire_protocol == DRBD_PROT_C);
4540
		what = WRITE_ACKED_BY_PEER;
P
Philipp Reisner 已提交
4541 4542
		break;
	case P_RECV_ACK:
4543
		D_ASSERT(mdev->tconn->net_conf->wire_protocol == DRBD_PROT_B);
4544
		what = RECV_ACKED_BY_PEER;
P
Philipp Reisner 已提交
4545
		break;
4546
	case P_DISCARD_WRITE:
4547
		D_ASSERT(mdev->tconn->net_conf->wire_protocol == DRBD_PROT_C);
4548 4549 4550 4551 4552
		what = DISCARD_WRITE;
		break;
	case P_RETRY_WRITE:
		D_ASSERT(mdev->tconn->net_conf->wire_protocol == DRBD_PROT_C);
		what = POSTPONE_WRITE;
P
Philipp Reisner 已提交
4553 4554 4555
		break;
	default:
		D_ASSERT(0);
4556
		return false;
P
Philipp Reisner 已提交
4557 4558 4559
	}

	return validate_req_change_req_state(mdev, p->block_id, sector,
4560 4561
					     &mdev->write_requests, __func__,
					     what, false);
P
Philipp Reisner 已提交
4562 4563
}

4564
static int got_NegAck(struct drbd_conf *mdev, enum drbd_packet cmd)
P
Philipp Reisner 已提交
4565
{
4566
	struct p_block_ack *p = &mdev->tconn->meta.rbuf.block_ack;
P
Philipp Reisner 已提交
4567
	sector_t sector = be64_to_cpu(p->sector);
4568
	int size = be32_to_cpu(p->blksize);
4569 4570
	bool missing_ok = mdev->tconn->net_conf->wire_protocol == DRBD_PROT_A ||
			  mdev->tconn->net_conf->wire_protocol == DRBD_PROT_B;
4571
	bool found;
P
Philipp Reisner 已提交
4572 4573 4574

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

4575
	if (p->block_id == ID_SYNCER) {
P
Philipp Reisner 已提交
4576 4577
		dec_rs_pending(mdev);
		drbd_rs_failed_io(mdev, sector, size);
4578
		return true;
P
Philipp Reisner 已提交
4579
	}
4580

4581
	found = validate_req_change_req_state(mdev, p->block_id, sector,
4582
					      &mdev->write_requests, __func__,
4583
					      NEG_ACKED, missing_ok);
4584 4585 4586 4587 4588 4589 4590
	if (!found) {
		/* 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. */
		if (!missing_ok)
4591
			return false;
4592
		drbd_set_out_of_sync(mdev, sector, size);
4593 4594
	}
	return true;
P
Philipp Reisner 已提交
4595 4596
}

4597
static int got_NegDReply(struct drbd_conf *mdev, enum drbd_packet cmd)
P
Philipp Reisner 已提交
4598
{
4599
	struct p_block_ack *p = &mdev->tconn->meta.rbuf.block_ack;
P
Philipp Reisner 已提交
4600 4601 4602
	sector_t sector = be64_to_cpu(p->sector);

	update_peer_seq(mdev, be32_to_cpu(p->seq_num));
4603

P
Philipp Reisner 已提交
4604 4605 4606 4607
	dev_err(DEV, "Got NegDReply; Sector %llus, len %u; Fail original request.\n",
	    (unsigned long long)sector, be32_to_cpu(p->blksize));

	return validate_req_change_req_state(mdev, p->block_id, sector,
4608
					     &mdev->read_requests, __func__,
4609
					     NEG_ACKED, false);
P
Philipp Reisner 已提交
4610 4611
}

4612
static int got_NegRSDReply(struct drbd_conf *mdev, enum drbd_packet cmd)
P
Philipp Reisner 已提交
4613 4614 4615
{
	sector_t sector;
	int size;
4616
	struct p_block_ack *p = &mdev->tconn->meta.rbuf.block_ack;
P
Philipp Reisner 已提交
4617 4618 4619 4620 4621 4622 4623 4624 4625 4626

	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);
4627
		switch (cmd) {
4628 4629 4630 4631 4632 4633 4634 4635 4636
		case P_NEG_RS_DREPLY:
			drbd_rs_failed_io(mdev, sector, size);
		case P_RS_CANCEL:
			break;
		default:
			D_ASSERT(0);
			put_ldev(mdev);
			return false;
		}
P
Philipp Reisner 已提交
4637 4638 4639
		put_ldev(mdev);
	}

4640
	return true;
P
Philipp Reisner 已提交
4641 4642
}

4643
static int got_BarrierAck(struct drbd_conf *mdev, enum drbd_packet cmd)
P
Philipp Reisner 已提交
4644
{
4645
	struct p_barrier_ack *p = &mdev->tconn->meta.rbuf.barrier_ack;
P
Philipp Reisner 已提交
4646

4647
	tl_release(mdev->tconn, p->barrier, be32_to_cpu(p->set_size));
P
Philipp Reisner 已提交
4648

4649 4650
	if (mdev->state.conn == C_AHEAD &&
	    atomic_read(&mdev->ap_in_flight) == 0 &&
4651 4652 4653
	    !test_and_set_bit(AHEAD_TO_SYNC_SOURCE, &mdev->current_epoch->flags)) {
		mdev->start_resync_timer.expires = jiffies + HZ;
		add_timer(&mdev->start_resync_timer);
4654 4655
	}

4656
	return true;
P
Philipp Reisner 已提交
4657 4658
}

4659
static int got_OVResult(struct drbd_conf *mdev, enum drbd_packet cmd)
P
Philipp Reisner 已提交
4660
{
4661
	struct p_block_ack *p = &mdev->tconn->meta.rbuf.block_ack;
P
Philipp Reisner 已提交
4662 4663 4664 4665 4666 4667 4668 4669 4670 4671 4672 4673 4674 4675
	struct drbd_work *w;
	sector_t sector;
	int size;

	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)
		drbd_ov_oos_found(mdev, sector, size);
	else
		ov_oos_print(mdev);

4676
	if (!get_ldev(mdev))
4677
		return true;
4678

P
Philipp Reisner 已提交
4679 4680 4681
	drbd_rs_complete_io(mdev, sector);
	dec_rs_pending(mdev);

4682 4683 4684 4685 4686 4687 4688
	--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
4689 4690 4691
		w = kmalloc(sizeof(*w), GFP_NOIO);
		if (w) {
			w->cb = w_ov_finished;
4692
			w->mdev = mdev;
4693
			drbd_queue_work_front(&mdev->tconn->data.work, w);
P
Philipp Reisner 已提交
4694 4695 4696 4697 4698 4699
		} else {
			dev_err(DEV, "kmalloc(w) failed.");
			ov_oos_print(mdev);
			drbd_resync_finished(mdev);
		}
	}
4700
	put_ldev(mdev);
4701
	return true;
P
Philipp Reisner 已提交
4702 4703
}

4704
static int got_skip(struct drbd_conf *mdev, enum drbd_packet cmd)
4705
{
4706
	return true;
4707 4708
}

4709 4710
static int tconn_process_done_ee(struct drbd_tconn *tconn)
{
4711 4712
	struct drbd_conf *mdev;
	int i, not_empty = 0;
4713 4714 4715 4716

	do {
		clear_bit(SIGNAL_ASENDER, &tconn->flags);
		flush_signals(current);
4717
		idr_for_each_entry(&tconn->volumes, mdev, i) {
4718
			if (drbd_process_done_ee(mdev))
4719 4720
				return 1; /* error */
		}
4721
		set_bit(SIGNAL_ASENDER, &tconn->flags);
4722 4723 4724 4725 4726 4727 4728 4729

		spin_lock_irq(&tconn->req_lock);
		idr_for_each_entry(&tconn->volumes, mdev, i) {
			not_empty = !list_empty(&mdev->done_ee);
			if (not_empty)
				break;
		}
		spin_unlock_irq(&tconn->req_lock);
4730 4731 4732 4733 4734
	} while (not_empty);

	return 0;
}

4735 4736
struct asender_cmd {
	size_t pkt_size;
4737 4738 4739 4740 4741
	enum mdev_or_conn fa_type; /* first argument's type */
	union {
		int (*mdev_fn)(struct drbd_conf *mdev, enum drbd_packet cmd);
		int (*conn_fn)(struct drbd_tconn *tconn, enum drbd_packet cmd);
	};
4742 4743 4744
};

static struct asender_cmd asender_tbl[] = {
4745 4746
	[P_PING]	    = { sizeof(struct p_header), CONN, { .conn_fn = got_Ping } },
	[P_PING_ACK]	    = { sizeof(struct p_header), CONN, { .conn_fn = got_PingAck } },
4747 4748 4749 4750 4751 4752 4753 4754 4755 4756 4757 4758 4759
	[P_RECV_ACK]	    = { sizeof(struct p_block_ack), MDEV, { got_BlockAck } },
	[P_WRITE_ACK]	    = { sizeof(struct p_block_ack), MDEV, { got_BlockAck } },
	[P_RS_WRITE_ACK]    = { sizeof(struct p_block_ack), MDEV, { got_BlockAck } },
	[P_DISCARD_WRITE]   = { sizeof(struct p_block_ack), MDEV, { got_BlockAck } },
	[P_NEG_ACK]	    = { sizeof(struct p_block_ack), MDEV, { got_NegAck } },
	[P_NEG_DREPLY]	    = { sizeof(struct p_block_ack), MDEV, { got_NegDReply } },
	[P_NEG_RS_DREPLY]   = { sizeof(struct p_block_ack), MDEV, { got_NegRSDReply } },
	[P_OV_RESULT]	    = { sizeof(struct p_block_ack), MDEV, { got_OVResult } },
	[P_BARRIER_ACK]	    = { sizeof(struct p_barrier_ack), MDEV, { got_BarrierAck } },
	[P_STATE_CHG_REPLY] = { sizeof(struct p_req_state_reply), MDEV, { got_RqSReply } },
	[P_RS_IS_IN_SYNC]   = { sizeof(struct p_block_ack), MDEV, { got_IsInSync } },
	[P_DELAY_PROBE]     = { sizeof(struct p_delay_probe93), MDEV, { got_skip } },
	[P_RS_CANCEL]       = { sizeof(struct p_block_ack), MDEV, { got_NegRSDReply } },
4760
	[P_CONN_ST_CHG_REPLY]={ sizeof(struct p_req_state_reply), CONN, {.conn_fn = got_conn_RqSReply}},
4761
	[P_RETRY_WRITE]	    = { sizeof(struct p_block_ack), MDEV, { got_BlockAck } },
4762 4763
};

P
Philipp Reisner 已提交
4764 4765
int drbd_asender(struct drbd_thread *thi)
{
4766
	struct drbd_tconn *tconn = thi->tconn;
4767
	struct p_header *h = &tconn->meta.rbuf.header;
P
Philipp Reisner 已提交
4768
	struct asender_cmd *cmd = NULL;
4769
	struct packet_info pi;
4770
	int rv;
P
Philipp Reisner 已提交
4771 4772
	void *buf    = h;
	int received = 0;
4773
	int expect   = sizeof(struct p_header);
4774
	int ping_timeout_active = 0;
P
Philipp Reisner 已提交
4775 4776 4777 4778

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

4779
	while (get_t_state(thi) == RUNNING) {
4780
		drbd_thread_current_set_cpu(thi);
4781
		if (test_and_clear_bit(SEND_PING, &tconn->flags)) {
4782
			if (!drbd_send_ping(tconn)) {
4783
				conn_err(tconn, "drbd_send_ping has failed\n");
4784 4785
				goto reconnect;
			}
4786 4787
			tconn->meta.socket->sk->sk_rcvtimeo =
				tconn->net_conf->ping_timeo*HZ/10;
4788
			ping_timeout_active = 1;
P
Philipp Reisner 已提交
4789 4790
		}

4791 4792 4793 4794
		/* TODO: conditionally cork; it may hurt latency if we cork without
		   much to send */
		if (!tconn->net_conf->no_cork)
			drbd_tcp_cork(tconn->meta.socket);
4795 4796
		if (tconn_process_done_ee(tconn)) {
			conn_err(tconn, "tconn_process_done_ee() failed\n");
4797
			goto reconnect;
4798
		}
P
Philipp Reisner 已提交
4799
		/* but unconditionally uncork unless disabled */
4800 4801
		if (!tconn->net_conf->no_cork)
			drbd_tcp_uncork(tconn->meta.socket);
P
Philipp Reisner 已提交
4802 4803 4804 4805 4806

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

4807 4808
		rv = drbd_recv_short(tconn->meta.socket, buf, expect-received, 0);
		clear_bit(SIGNAL_ASENDER, &tconn->flags);
P
Philipp Reisner 已提交
4809 4810 4811 4812 4813 4814 4815 4816 4817 4818 4819 4820 4821 4822 4823 4824 4825

		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) {
4826
			conn_err(tconn, "meta connection shut down by peer.\n");
P
Philipp Reisner 已提交
4827 4828
			goto reconnect;
		} else if (rv == -EAGAIN) {
4829 4830
			/* If the data socket received something meanwhile,
			 * that is good enough: peer is still alive. */
4831 4832
			if (time_after(tconn->last_received,
				jiffies - tconn->meta.socket->sk->sk_rcvtimeo))
4833
				continue;
4834
			if (ping_timeout_active) {
4835
				conn_err(tconn, "PingAck did not arrive in time.\n");
P
Philipp Reisner 已提交
4836 4837
				goto reconnect;
			}
4838
			set_bit(SEND_PING, &tconn->flags);
P
Philipp Reisner 已提交
4839 4840 4841 4842
			continue;
		} else if (rv == -EINTR) {
			continue;
		} else {
4843
			conn_err(tconn, "sock_recvmsg returned %d\n", rv);
P
Philipp Reisner 已提交
4844 4845 4846 4847
			goto reconnect;
		}

		if (received == expect && cmd == NULL) {
4848
			if (!decode_header(tconn, h, &pi))
P
Philipp Reisner 已提交
4849
				goto reconnect;
4850 4851
			cmd = &asender_tbl[pi.cmd];
			if (pi.cmd >= ARRAY_SIZE(asender_tbl) || !cmd) {
4852
				conn_err(tconn, "unknown command %d on meta (l: %d)\n",
4853
					pi.cmd, pi.size);
P
Philipp Reisner 已提交
4854 4855 4856
				goto disconnect;
			}
			expect = cmd->pkt_size;
4857
			if (pi.size != expect - sizeof(struct p_header)) {
4858
				conn_err(tconn, "Wrong packet size on meta (c: %d, l: %d)\n",
4859
					pi.cmd, pi.size);
P
Philipp Reisner 已提交
4860
				goto reconnect;
4861
			}
P
Philipp Reisner 已提交
4862 4863
		}
		if (received == expect) {
4864 4865 4866 4867 4868 4869 4870 4871 4872 4873
			bool rv;

			if (cmd->fa_type == CONN) {
				rv = cmd->conn_fn(tconn, pi.cmd);
			} else {
				struct drbd_conf *mdev = vnr_to_mdev(tconn, pi.vnr);
				rv = cmd->mdev_fn(mdev, pi.cmd);
			}

			if (!rv)
P
Philipp Reisner 已提交
4874 4875
				goto reconnect;

4876 4877
			tconn->last_received = jiffies;

4878 4879
			/* the idle_timeout (ping-int)
			 * has been restored in got_PingAck() */
4880
			if (cmd == &asender_tbl[P_PING_ACK])
4881 4882
				ping_timeout_active = 0;

P
Philipp Reisner 已提交
4883 4884
			buf	 = h;
			received = 0;
4885
			expect	 = sizeof(struct p_header);
P
Philipp Reisner 已提交
4886 4887 4888 4889 4890 4891
			cmd	 = NULL;
		}
	}

	if (0) {
reconnect:
4892
		conn_request_state(tconn, NS(conn, C_NETWORK_FAILURE), CS_HARD);
P
Philipp Reisner 已提交
4893 4894 4895
	}
	if (0) {
disconnect:
4896
		conn_request_state(tconn, NS(conn, C_DISCONNECTING), CS_HARD);
P
Philipp Reisner 已提交
4897
	}
4898
	clear_bit(SIGNAL_ASENDER, &tconn->flags);
P
Philipp Reisner 已提交
4899

4900
	conn_info(tconn, "asender terminated\n");
P
Philipp Reisner 已提交
4901 4902 4903

	return 0;
}