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 int 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 -EINVAL;
P
Philipp Reisner 已提交
967
	}
968
	return 0;
969 970
}

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
	int err;
975

976 977
	err = drbd_recv(tconn, h, sizeof(*h));
	if (unlikely(err != sizeof(*h))) {
978
		if (!signal_pending(current))
979 980 981 982
			conn_warn(tconn, "short read expecting header on sock: r=%d\n", err);
		if (err >= 0)
			err = -EIO;
		return err;
983 984
	}

985
	err = decode_header(tconn, h, pi);
986
	tconn->last_received = jiffies;
P
Philipp Reisner 已提交
987

988
	return err;
P
Philipp Reisner 已提交
989 990
}

991
static void drbd_flush(struct drbd_conf *mdev)
P
Philipp Reisner 已提交
992 993 994 995
{
	int rv;

	if (mdev->write_ordering >= WO_bdev_flush && get_ldev(mdev)) {
996
		rv = blkdev_issue_flush(mdev->ldev->backing_bdev, GFP_KERNEL,
997
					NULL);
P
Philipp Reisner 已提交
998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018
		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)
{
1019
	int epoch_size;
P
Philipp Reisner 已提交
1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042
	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 &&
1043
		    test_bit(DE_HAVE_BARRIER_NUMBER, &epoch->flags)) {
P
Philipp Reisner 已提交
1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062
			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);
1063
				/* atomic_set(&epoch->active, 0); is already zero */
P
Philipp Reisner 已提交
1064 1065
				if (rv == FE_STILL_LIVE)
					rv = FE_RECYCLED;
1066
				wake_up(&mdev->ee_wait);
P
Philipp Reisner 已提交
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 1100 1101
			}
		}

		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;
1102
	if (pwo != mdev->write_ordering || wo == WO_bdev_flush)
P
Philipp Reisner 已提交
1103 1104 1105
		dev_info(DEV, "Method to ensure write ordering: %s\n", write_ordering_str[mdev->write_ordering]);
}

1106
/**
1107
 * drbd_submit_peer_request()
1108
 * @mdev:	DRBD device.
1109
 * @peer_req:	peer request
1110
 * @rw:		flag field, see bio->bi_rw
1111 1112 1113 1114 1115 1116 1117 1118 1119 1120
 *
 * 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.
1121 1122
 */
/* TODO allocate from our own bio_set. */
1123 1124 1125
int drbd_submit_peer_request(struct drbd_conf *mdev,
			     struct drbd_peer_request *peer_req,
			     const unsigned rw, const int fault_type)
1126 1127 1128
{
	struct bio *bios = NULL;
	struct bio *bio;
1129 1130 1131
	struct page *page = peer_req->pages;
	sector_t sector = peer_req->i.sector;
	unsigned ds = peer_req->i.size;
1132 1133
	unsigned n_bios = 0;
	unsigned nr_pages = (ds + PAGE_SIZE -1) >> PAGE_SHIFT;
1134
	int err = -ENOMEM;
1135 1136 1137 1138

	/* 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
1139 1140 1141 1142 1143
	 * 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.
	 */
1144 1145 1146 1147 1148 1149
next_bio:
	bio = bio_alloc(GFP_NOIO, nr_pages);
	if (!bio) {
		dev_err(DEV, "submit_ee: Allocation of a bio failed\n");
		goto fail;
	}
1150
	/* > peer_req->i.sector, unless this is the first bio */
1151 1152 1153
	bio->bi_sector = sector;
	bio->bi_bdev = mdev->ldev->backing_bdev;
	bio->bi_rw = rw;
1154
	bio->bi_private = peer_req;
1155
	bio->bi_end_io = drbd_peer_request_endio;
1156 1157 1158 1159 1160 1161 1162 1163

	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)) {
1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174
			/* 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;
			}
1175 1176 1177 1178 1179 1180 1181 1182 1183
			goto next_bio;
		}
		ds -= len;
		sector += len >> 9;
		--nr_pages;
	}
	D_ASSERT(page == NULL);
	D_ASSERT(ds == 0);

1184
	atomic_set(&peer_req->pending_bios, n_bios);
1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199
	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);
	}
1200
	return err;
1201 1202
}

1203
static void drbd_remove_epoch_entry_interval(struct drbd_conf *mdev,
1204
					     struct drbd_peer_request *peer_req)
1205
{
1206
	struct drbd_interval *i = &peer_req->i;
1207 1208 1209 1210

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

A
Andreas Gruenbacher 已提交
1211
	/* Wake up any processes waiting for this peer request to complete.  */
1212 1213 1214 1215
	if (i->waiting)
		wake_up(&mdev->misc_wait);
}

1216 1217
static int receive_Barrier(struct drbd_conf *mdev, enum drbd_packet cmd,
			   unsigned int data_size)
P
Philipp Reisner 已提交
1218
{
1219
	int rv;
1220
	struct p_barrier *p = &mdev->tconn->data.rbuf.barrier;
P
Philipp Reisner 已提交
1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235
	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)
1236
			return true;
1237 1238 1239 1240 1241 1242 1243 1244 1245

		/* 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
1246 1247 1248 1249

	case WO_bdev_flush:
	case WO_drain_io:
		drbd_wait_ee_list_empty(mdev, &mdev->active_ee);
1250 1251 1252 1253 1254 1255
		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
1256 1257
		}

1258 1259 1260 1261 1262
		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
1263

1264
		return true;
1265 1266
	default:
		dev_err(DEV, "Strangeness in mdev->write_ordering %d\n", mdev->write_ordering);
1267
		return false;
P
Philipp Reisner 已提交
1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284
	}

	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);

1285
	return true;
P
Philipp Reisner 已提交
1286 1287 1288 1289
}

/* used from receive_RSDataReply (recv_resync_read)
 * and from receive_Data */
1290 1291 1292
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
1293
{
1294
	const sector_t capacity = drbd_get_capacity(mdev->this_bdev);
1295
	struct drbd_peer_request *peer_req;
P
Philipp Reisner 已提交
1296
	struct page *page;
1297
	int dgs, ds, rr;
1298 1299
	void *dig_in = mdev->tconn->int_dig_in;
	void *dig_vv = mdev->tconn->int_dig_vv;
1300
	unsigned long *data;
P
Philipp Reisner 已提交
1301

1302 1303
	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
1304 1305

	if (dgs) {
1306
		rr = drbd_recv(mdev->tconn, dig_in, dgs);
P
Philipp Reisner 已提交
1307
		if (rr != dgs) {
1308 1309 1310 1311
			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
1312 1313 1314 1315 1316 1317
			return NULL;
		}
	}

	data_size -= dgs;

1318 1319 1320 1321 1322 1323
	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
1324

1325 1326 1327
	/* even though we trust out peer,
	 * we sometimes have to double check. */
	if (sector + (data_size>>9) > capacity) {
1328 1329
		dev_err(DEV, "request from peer beyond end of local disk: "
			"capacity: %llus < sector: %llus + size: %u\n",
1330 1331 1332 1333 1334
			(unsigned long long)capacity,
			(unsigned long long)sector, data_size);
		return NULL;
	}

P
Philipp Reisner 已提交
1335 1336 1337
	/* 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.  */
1338 1339
	peer_req = drbd_alloc_ee(mdev, id, sector, data_size, GFP_NOIO);
	if (!peer_req)
P
Philipp Reisner 已提交
1340
		return NULL;
1341

P
Philipp Reisner 已提交
1342
	ds = data_size;
1343
	page = peer_req->pages;
1344 1345
	page_chain_for_each(page) {
		unsigned len = min_t(int, ds, PAGE_SIZE);
1346
		data = kmap(page);
1347
		rr = drbd_recv(mdev->tconn, data, len);
1348
		if (drbd_insert_fault(mdev, DRBD_FAULT_RECEIVE)) {
1349 1350 1351
			dev_err(DEV, "Fault injection: Corrupting data on receive\n");
			data[0] = data[0] ^ (unsigned long)-1;
		}
P
Philipp Reisner 已提交
1352
		kunmap(page);
1353
		if (rr != len) {
1354
			drbd_free_ee(mdev, peer_req);
1355 1356 1357
			if (!signal_pending(current))
				dev_warn(DEV, "short read receiving data: read %d expected %d\n",
				rr, len);
P
Philipp Reisner 已提交
1358 1359 1360 1361 1362 1363
			return NULL;
		}
		ds -= rr;
	}

	if (dgs) {
1364
		drbd_csum_ee(mdev, mdev->tconn->integrity_r_tfm, peer_req, dig_vv);
P
Philipp Reisner 已提交
1365
		if (memcmp(dig_in, dig_vv, dgs)) {
1366 1367
			dev_err(DEV, "Digest integrity check FAILED: %llus +%u\n",
				(unsigned long long)sector, data_size);
1368
			drbd_free_ee(mdev, peer_req);
P
Philipp Reisner 已提交
1369 1370 1371 1372
			return NULL;
		}
	}
	mdev->recv_cnt += data_size>>9;
1373
	return peer_req;
P
Philipp Reisner 已提交
1374 1375 1376 1377 1378 1379 1380 1381 1382 1383 1384
}

/* 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;

1385
	if (!data_size)
1386
		return true;
1387

1388
	page = drbd_pp_alloc(mdev, 1, 1);
P
Philipp Reisner 已提交
1389 1390 1391

	data = kmap(page);
	while (data_size) {
1392
		rr = drbd_recv(mdev->tconn, data, min_t(int, data_size, PAGE_SIZE));
P
Philipp Reisner 已提交
1393 1394
		if (rr != min_t(int, data_size, PAGE_SIZE)) {
			rv = 0;
1395 1396 1397 1398
			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
1399 1400 1401 1402 1403
			break;
		}
		data_size -= rr;
	}
	kunmap(page);
1404
	drbd_pp_free(mdev, page, 0);
P
Philipp Reisner 已提交
1405 1406 1407 1408 1409 1410 1411 1412 1413
	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;
1414 1415
	void *dig_in = mdev->tconn->int_dig_in;
	void *dig_vv = mdev->tconn->int_dig_vv;
P
Philipp Reisner 已提交
1416

1417 1418
	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
1419 1420

	if (dgs) {
1421
		rr = drbd_recv(mdev->tconn, dig_in, dgs);
P
Philipp Reisner 已提交
1422
		if (rr != dgs) {
1423 1424 1425 1426
			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
1427 1428 1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440 1441
			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);
1442
		rr = drbd_recv(mdev->tconn,
P
Philipp Reisner 已提交
1443 1444 1445 1446
			     kmap(bvec->bv_page)+bvec->bv_offset,
			     expect);
		kunmap(bvec->bv_page);
		if (rr != expect) {
1447 1448 1449 1450
			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
1451 1452 1453 1454 1455 1456
			return 0;
		}
		data_size -= rr;
	}

	if (dgs) {
1457
		drbd_csum_bio(mdev, mdev->tconn->integrity_r_tfm, bio, dig_vv);
P
Philipp Reisner 已提交
1458 1459 1460 1461 1462 1463 1464 1465 1466 1467 1468 1469
		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 */
1470
static int e_end_resync_block(struct drbd_work *w, int unused)
P
Philipp Reisner 已提交
1471
{
1472 1473
	struct drbd_peer_request *peer_req =
		container_of(w, struct drbd_peer_request, w);
1474
	struct drbd_conf *mdev = w->mdev;
1475
	sector_t sector = peer_req->i.sector;
1476
	int err;
P
Philipp Reisner 已提交
1477

1478
	D_ASSERT(drbd_interval_empty(&peer_req->i));
P
Philipp Reisner 已提交
1479

1480 1481
	if (likely((peer_req->flags & EE_WAS_ERROR) == 0)) {
		drbd_set_in_sync(mdev, sector, peer_req->i.size);
1482
		err = drbd_send_ack(mdev, P_RS_WRITE_ACK, peer_req);
P
Philipp Reisner 已提交
1483 1484
	} else {
		/* Record failure to sync */
1485
		drbd_rs_failed_io(mdev, sector, peer_req->i.size);
P
Philipp Reisner 已提交
1486

1487
		err  = drbd_send_ack(mdev, P_NEG_ACK, peer_req);
P
Philipp Reisner 已提交
1488 1489 1490
	}
	dec_unacked(mdev);

1491
	return err;
P
Philipp Reisner 已提交
1492 1493 1494 1495
}

static int recv_resync_read(struct drbd_conf *mdev, sector_t sector, int data_size) __releases(local)
{
1496
	struct drbd_peer_request *peer_req;
P
Philipp Reisner 已提交
1497

1498 1499
	peer_req = read_in_block(mdev, ID_SYNCER, sector, data_size);
	if (!peer_req)
1500
		goto fail;
P
Philipp Reisner 已提交
1501 1502 1503 1504 1505 1506 1507

	dec_rs_pending(mdev);

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

1508
	peer_req->w.cb = e_end_resync_block;
1509

1510
	spin_lock_irq(&mdev->tconn->req_lock);
1511
	list_add(&peer_req->w.list, &mdev->sync_ee);
1512
	spin_unlock_irq(&mdev->tconn->req_lock);
P
Philipp Reisner 已提交
1513

1514
	atomic_add(data_size >> 9, &mdev->rs_sect_ev);
1515
	if (drbd_submit_peer_request(mdev, peer_req, WRITE, DRBD_FAULT_RS_WR) == 0)
1516
		return true;
P
Philipp Reisner 已提交
1517

1518 1519
	/* don't care for the reason here */
	dev_err(DEV, "submit failed, triggering re-connect\n");
1520
	spin_lock_irq(&mdev->tconn->req_lock);
1521
	list_del(&peer_req->w.list);
1522
	spin_unlock_irq(&mdev->tconn->req_lock);
1523

1524
	drbd_free_ee(mdev, peer_req);
1525 1526
fail:
	put_ldev(mdev);
1527
	return false;
P
Philipp Reisner 已提交
1528 1529
}

1530
static struct drbd_request *
1531 1532
find_request(struct drbd_conf *mdev, struct rb_root *root, u64 id,
	     sector_t sector, bool missing_ok, const char *func)
1533 1534 1535
{
	struct drbd_request *req;

1536 1537
	/* Request object according to our peer */
	req = (struct drbd_request *)(unsigned long)id;
1538
	if (drbd_contains_interval(root, sector, &req->i) && req->i.local)
1539
		return req;
1540 1541 1542 1543
	if (!missing_ok) {
		dev_err(DEV, "%s: failed to find request %lu, sector %llus\n", func,
			(unsigned long)id, (unsigned long long)sector);
	}
1544 1545 1546
	return NULL;
}

1547 1548
static int receive_DataReply(struct drbd_conf *mdev, enum drbd_packet cmd,
			     unsigned int data_size)
P
Philipp Reisner 已提交
1549 1550 1551 1552
{
	struct drbd_request *req;
	sector_t sector;
	int ok;
1553
	struct p_data *p = &mdev->tconn->data.rbuf.data;
P
Philipp Reisner 已提交
1554 1555 1556

	sector = be64_to_cpu(p->sector);

1557
	spin_lock_irq(&mdev->tconn->req_lock);
1558
	req = find_request(mdev, &mdev->read_requests, p->block_id, sector, false, __func__);
1559
	spin_unlock_irq(&mdev->tconn->req_lock);
1560
	if (unlikely(!req))
1561
		return false;
P
Philipp Reisner 已提交
1562

B
Bart Van Assche 已提交
1563
	/* hlist_del(&req->collision) is done in _req_may_be_done, to avoid
P
Philipp Reisner 已提交
1564 1565 1566 1567 1568
	 * 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)
1569
		req_mod(req, DATA_RECEIVED);
P
Philipp Reisner 已提交
1570 1571 1572 1573 1574 1575 1576
	/* 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;
}

1577 1578
static int receive_RSDataReply(struct drbd_conf *mdev, enum drbd_packet cmd,
			       unsigned int data_size)
P
Philipp Reisner 已提交
1579 1580 1581
{
	sector_t sector;
	int ok;
1582
	struct p_data *p = &mdev->tconn->data.rbuf.data;
P
Philipp Reisner 已提交
1583 1584 1585 1586 1587 1588 1589

	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,
1590
		 * or in drbd_peer_request_endio. */
P
Philipp Reisner 已提交
1591 1592 1593 1594 1595 1596 1597
		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);

1598
		drbd_send_ack_dp(mdev, P_NEG_ACK, p, data_size);
P
Philipp Reisner 已提交
1599 1600
	}

1601 1602
	atomic_add(data_size >> 9, &mdev->rs_sect_in);

P
Philipp Reisner 已提交
1603 1604 1605
	return ok;
}

1606
static int w_restart_write(struct drbd_work *w, int cancel)
1607 1608 1609 1610 1611 1612 1613 1614 1615 1616
{
	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);
1617
		return -EIO;
1618 1619 1620 1621 1622 1623 1624 1625 1626
	}
	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 */ ;
1627
	return 0;
1628 1629 1630 1631 1632 1633 1634 1635 1636 1637 1638 1639 1640 1641 1642 1643 1644 1645 1646 1647 1648 1649 1650
}

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
1651 1652 1653
/* e_end_block() is called via drbd_process_done_ee().
 * this means this function only runs in the asender thread
 */
1654
static int e_end_block(struct drbd_work *w, int cancel)
P
Philipp Reisner 已提交
1655
{
1656 1657
	struct drbd_peer_request *peer_req =
		container_of(w, struct drbd_peer_request, w);
1658
	struct drbd_conf *mdev = w->mdev;
1659
	sector_t sector = peer_req->i.sector;
1660
	int err = 0, pcmd;
P
Philipp Reisner 已提交
1661

1662
	if (mdev->tconn->net_conf->wire_protocol == DRBD_PROT_C) {
1663
		if (likely((peer_req->flags & EE_WAS_ERROR) == 0)) {
P
Philipp Reisner 已提交
1664 1665
			pcmd = (mdev->state.conn >= C_SYNC_SOURCE &&
				mdev->state.conn <= C_PAUSED_SYNC_T &&
1666
				peer_req->flags & EE_MAY_SET_IN_SYNC) ?
P
Philipp Reisner 已提交
1667
				P_RS_WRITE_ACK : P_WRITE_ACK;
1668
			err = drbd_send_ack(mdev, pcmd, peer_req);
P
Philipp Reisner 已提交
1669
			if (pcmd == P_RS_WRITE_ACK)
1670
				drbd_set_in_sync(mdev, sector, peer_req->i.size);
P
Philipp Reisner 已提交
1671
		} else {
1672
			err = drbd_send_ack(mdev, P_NEG_ACK, peer_req);
P
Philipp Reisner 已提交
1673 1674 1675 1676 1677 1678 1679
			/* 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.  */
1680
	if (mdev->tconn->net_conf->two_primaries) {
1681
		spin_lock_irq(&mdev->tconn->req_lock);
1682 1683
		D_ASSERT(!drbd_interval_empty(&peer_req->i));
		drbd_remove_epoch_entry_interval(mdev, peer_req);
1684 1685
		if (peer_req->flags & EE_RESTART_REQUESTS)
			restart_conflicting_writes(mdev, sector, peer_req->i.size);
1686
		spin_unlock_irq(&mdev->tconn->req_lock);
1687
	} else
1688
		D_ASSERT(drbd_interval_empty(&peer_req->i));
P
Philipp Reisner 已提交
1689

1690
	drbd_may_finish_epoch(mdev, peer_req->epoch, EV_PUT + (cancel ? EV_CLEANUP : 0));
P
Philipp Reisner 已提交
1691

1692
	return err;
P
Philipp Reisner 已提交
1693 1694
}

1695
static int e_send_ack(struct drbd_work *w, enum drbd_packet ack)
P
Philipp Reisner 已提交
1696
{
1697
	struct drbd_conf *mdev = w->mdev;
1698 1699
	struct drbd_peer_request *peer_req =
		container_of(w, struct drbd_peer_request, w);
1700
	int err;
P
Philipp Reisner 已提交
1701

1702
	err = drbd_send_ack(mdev, ack, peer_req);
P
Philipp Reisner 已提交
1703 1704
	dec_unacked(mdev);

1705
	return err;
P
Philipp Reisner 已提交
1706 1707
}

1708
static int e_send_discard_write(struct drbd_work *w, int unused)
1709 1710 1711 1712
{
	return e_send_ack(w, P_DISCARD_WRITE);
}

1713
static int e_send_retry_write(struct drbd_work *w, int unused)
1714 1715 1716 1717 1718 1719 1720
{
	struct drbd_tconn *tconn = w->mdev->tconn;

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

1721 1722 1723 1724 1725 1726 1727 1728 1729 1730 1731 1732 1733 1734 1735
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;
}

1736 1737 1738 1739 1740 1741 1742 1743 1744 1745 1746 1747 1748
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);
}

1749
static void update_peer_seq(struct drbd_conf *mdev, unsigned int peer_seq)
1750
{
1751
	unsigned int newest_peer_seq;
1752

1753 1754
	if (need_peer_seq(mdev)) {
		spin_lock(&mdev->peer_seq_lock);
1755 1756
		newest_peer_seq = seq_max(mdev->peer_seq, peer_seq);
		mdev->peer_seq = newest_peer_seq;
1757
		spin_unlock(&mdev->peer_seq_lock);
1758 1759
		/* wake up only if we actually changed mdev->peer_seq */
		if (peer_seq == newest_peer_seq)
1760 1761
			wake_up(&mdev->seq_wait);
	}
1762 1763
}

P
Philipp Reisner 已提交
1764 1765 1766 1767 1768 1769 1770 1771 1772 1773 1774 1775 1776 1777 1778 1779 1780 1781 1782 1783 1784
/* 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). */
1785
static int wait_for_and_update_peer_seq(struct drbd_conf *mdev, const u32 peer_seq)
P
Philipp Reisner 已提交
1786 1787 1788
{
	DEFINE_WAIT(wait);
	long timeout;
1789 1790 1791 1792 1793
	int ret;

	if (!need_peer_seq(mdev))
		return 0;

P
Philipp Reisner 已提交
1794 1795
	spin_lock(&mdev->peer_seq_lock);
	for (;;) {
1796 1797 1798
		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
1799
			break;
1800
		}
P
Philipp Reisner 已提交
1801 1802 1803 1804
		if (signal_pending(current)) {
			ret = -ERESTARTSYS;
			break;
		}
1805
		prepare_to_wait(&mdev->seq_wait, &wait, TASK_INTERRUPTIBLE);
P
Philipp Reisner 已提交
1806
		spin_unlock(&mdev->peer_seq_lock);
1807 1808
		timeout = mdev->tconn->net_conf->ping_timeo*HZ/10;
		timeout = schedule_timeout(timeout);
P
Philipp Reisner 已提交
1809
		spin_lock(&mdev->peer_seq_lock);
1810
		if (!timeout) {
P
Philipp Reisner 已提交
1811
			ret = -ETIMEDOUT;
1812
			dev_err(DEV, "Timed out waiting for missing ack packets; disconnecting\n");
P
Philipp Reisner 已提交
1813 1814 1815 1816
			break;
		}
	}
	spin_unlock(&mdev->peer_seq_lock);
1817
	finish_wait(&mdev->seq_wait, &wait);
P
Philipp Reisner 已提交
1818 1819 1820
	return ret;
}

1821 1822 1823 1824
/* 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)
1825
{
1826 1827 1828 1829
	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);
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 1963 1964
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
1965
/* mirrored write */
1966 1967
static int receive_Data(struct drbd_conf *mdev, enum drbd_packet cmd,
			unsigned int data_size)
P
Philipp Reisner 已提交
1968 1969
{
	sector_t sector;
1970
	struct drbd_peer_request *peer_req;
1971
	struct p_data *p = &mdev->tconn->data.rbuf.data;
1972
	u32 peer_seq = be32_to_cpu(p->seq_num);
P
Philipp Reisner 已提交
1973 1974
	int rw = WRITE;
	u32 dp_flags;
1975
	int err;
P
Philipp Reisner 已提交
1976 1977


1978 1979
	if (!get_ldev(mdev)) {
		err = wait_for_and_update_peer_seq(mdev, peer_seq);
1980
		drbd_send_ack_dp(mdev, P_NEG_ACK, p, data_size);
P
Philipp Reisner 已提交
1981
		atomic_inc(&mdev->current_epoch->epoch_size);
1982
		return drbd_drain_block(mdev, data_size) && err == 0;
P
Philipp Reisner 已提交
1983 1984
	}

1985 1986 1987 1988 1989
	/*
	 * 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
1990 1991

	sector = be64_to_cpu(p->sector);
1992 1993
	peer_req = read_in_block(mdev, p->block_id, sector, data_size);
	if (!peer_req) {
P
Philipp Reisner 已提交
1994
		put_ldev(mdev);
1995
		return false;
P
Philipp Reisner 已提交
1996 1997
	}

1998
	peer_req->w.cb = e_end_block;
P
Philipp Reisner 已提交
1999

2000 2001 2002 2003
	dp_flags = be32_to_cpu(p->dp_flags);
	rw |= wire_flags_to_bio(mdev, dp_flags);

	if (dp_flags & DP_MAY_SET_IN_SYNC)
2004
		peer_req->flags |= EE_MAY_SET_IN_SYNC;
2005

P
Philipp Reisner 已提交
2006
	spin_lock(&mdev->epoch_lock);
2007 2008 2009
	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
2010 2011
	spin_unlock(&mdev->epoch_lock);

2012 2013 2014
	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
2015
			goto out_interrupted;
2016
		spin_lock_irq(&mdev->tconn->req_lock);
2017 2018 2019 2020
		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
2021
				put_ldev(mdev);
2022
				return true;
P
Philipp Reisner 已提交
2023
			}
2024
			goto out_interrupted;
P
Philipp Reisner 已提交
2025
		}
2026 2027
	} else
		spin_lock_irq(&mdev->tconn->req_lock);
2028
	list_add(&peer_req->w.list, &mdev->active_ee);
2029
	spin_unlock_irq(&mdev->tconn->req_lock);
P
Philipp Reisner 已提交
2030

2031
	switch (mdev->tconn->net_conf->wire_protocol) {
P
Philipp Reisner 已提交
2032 2033 2034 2035 2036 2037 2038 2039
	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 */
2040
		drbd_send_ack(mdev, P_RECV_ACK, peer_req);
P
Philipp Reisner 已提交
2041 2042 2043 2044 2045 2046
		break;
	case DRBD_PROT_A:
		/* nothing to do */
		break;
	}

2047
	if (mdev->state.pdsk < D_INCONSISTENT) {
P
Philipp Reisner 已提交
2048
		/* In case we have the only disk of the cluster, */
2049 2050 2051 2052
		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
2053 2054
	}

2055
	if (drbd_submit_peer_request(mdev, peer_req, rw, DRBD_FAULT_DT_WR) == 0)
2056
		return true;
P
Philipp Reisner 已提交
2057

2058 2059
	/* don't care for the reason here */
	dev_err(DEV, "submit failed, triggering re-connect\n");
2060
	spin_lock_irq(&mdev->tconn->req_lock);
2061 2062
	list_del(&peer_req->w.list);
	drbd_remove_epoch_entry_interval(mdev, peer_req);
2063
	spin_unlock_irq(&mdev->tconn->req_lock);
2064 2065
	if (peer_req->flags & EE_CALL_AL_COMPLETE_IO)
		drbd_al_complete_io(mdev, peer_req->i.sector);
2066

P
Philipp Reisner 已提交
2067
out_interrupted:
2068
	drbd_may_finish_epoch(mdev, peer_req->epoch, EV_PUT + EV_CLEANUP);
P
Philipp Reisner 已提交
2069
	put_ldev(mdev);
2070
	drbd_free_ee(mdev, peer_req);
2071
	return false;
P
Philipp Reisner 已提交
2072 2073
}

2074 2075 2076 2077 2078 2079 2080 2081 2082 2083 2084
/* 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.
 */
2085
int drbd_rs_should_slow_down(struct drbd_conf *mdev, sector_t sector)
2086 2087 2088
{
	struct gendisk *disk = mdev->ldev->backing_bdev->bd_contains->bd_disk;
	unsigned long db, dt, dbdt;
2089
	struct lc_element *tmp;
2090 2091 2092 2093
	int curr_events;
	int throttle = 0;

	/* feature disabled? */
2094
	if (mdev->ldev->dc.c_min_rate == 0)
2095 2096
		return 0;

2097 2098 2099 2100 2101 2102 2103 2104 2105 2106 2107 2108
	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);

2109 2110 2111
	curr_events = (int)part_stat_read(&disk->part0, sectors[0]) +
		      (int)part_stat_read(&disk->part0, sectors[1]) -
			atomic_read(&mdev->rs_sect_ev);
2112

2113 2114 2115 2116 2117 2118 2119 2120
	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. */
2121 2122 2123 2124 2125 2126
		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;
2127 2128 2129 2130 2131 2132 2133

		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);

2134
		if (dbdt > mdev->ldev->dc.c_min_rate)
2135 2136 2137 2138 2139 2140
			throttle = 1;
	}
	return throttle;
}


2141 2142
static int receive_DataRequest(struct drbd_conf *mdev, enum drbd_packet cmd,
			       unsigned int digest_size)
P
Philipp Reisner 已提交
2143 2144 2145
{
	sector_t sector;
	const sector_t capacity = drbd_get_capacity(mdev->this_bdev);
2146
	struct drbd_peer_request *peer_req;
P
Philipp Reisner 已提交
2147
	struct digest_info *di = NULL;
2148
	int size, verb;
P
Philipp Reisner 已提交
2149
	unsigned int fault_type;
2150
	struct p_block_req *p =	&mdev->tconn->data.rbuf.block_req;
P
Philipp Reisner 已提交
2151 2152 2153 2154

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

2155
	if (size <= 0 || !IS_ALIGNED(size, 512) || size > DRBD_MAX_BIO_SIZE) {
P
Philipp Reisner 已提交
2156 2157
		dev_err(DEV, "%s:%d: sector: %llus, size: %u\n", __FILE__, __LINE__,
				(unsigned long long)sector, size);
2158
		return false;
P
Philipp Reisner 已提交
2159 2160 2161 2162
	}
	if (sector + (size>>9) > capacity) {
		dev_err(DEV, "%s:%d: sector: %llus, size: %u\n", __FILE__, __LINE__,
				(unsigned long long)sector, size);
2163
		return false;
P
Philipp Reisner 已提交
2164 2165 2166
	}

	if (!get_ldev_if_state(mdev, D_UP_TO_DATE)) {
2167 2168 2169 2170 2171 2172 2173 2174 2175 2176 2177 2178 2179 2180 2181 2182 2183 2184 2185 2186
		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
2187 2188
			dev_err(DEV, "Can not satisfy peer's read request, "
			    "no local data.\n");
2189

L
Lars Ellenberg 已提交
2190 2191
		/* drain possibly payload */
		return drbd_drain_block(mdev, digest_size);
P
Philipp Reisner 已提交
2192 2193 2194 2195 2196
	}

	/* 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.  */
2197 2198
	peer_req = drbd_alloc_ee(mdev, p->block_id, sector, size, GFP_NOIO);
	if (!peer_req) {
P
Philipp Reisner 已提交
2199
		put_ldev(mdev);
2200
		return false;
P
Philipp Reisner 已提交
2201 2202
	}

2203
	switch (cmd) {
P
Philipp Reisner 已提交
2204
	case P_DATA_REQUEST:
2205
		peer_req->w.cb = w_e_end_data_req;
P
Philipp Reisner 已提交
2206
		fault_type = DRBD_FAULT_DT_RD;
2207 2208 2209
		/* application IO, don't drbd_rs_begin_io */
		goto submit;

P
Philipp Reisner 已提交
2210
	case P_RS_DATA_REQUEST:
2211
		peer_req->w.cb = w_e_end_rsdata_req;
P
Philipp Reisner 已提交
2212
		fault_type = DRBD_FAULT_RS_RD;
2213 2214
		/* used in the sector offset progress display */
		mdev->bm_resync_fo = BM_SECT_TO_BIT(sector);
P
Philipp Reisner 已提交
2215 2216 2217 2218 2219 2220 2221 2222 2223 2224 2225 2226
		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));

2227 2228
		peer_req->digest = di;
		peer_req->flags |= EE_HAS_DIGEST;
2229

2230
		if (drbd_recv(mdev->tconn, di->digest, digest_size) != digest_size)
P
Philipp Reisner 已提交
2231 2232
			goto out_free_e;

2233
		if (cmd == P_CSUM_RS_REQUEST) {
2234
			D_ASSERT(mdev->tconn->agreed_pro_version >= 89);
2235
			peer_req->w.cb = w_e_end_csum_rs_req;
2236 2237
			/* used in the sector offset progress display */
			mdev->bm_resync_fo = BM_SECT_TO_BIT(sector);
2238
		} else if (cmd == P_OV_REPLY) {
2239 2240
			/* track progress, we may need to throttle */
			atomic_add(size >> 9, &mdev->rs_sect_in);
2241
			peer_req->w.cb = w_e_end_ov_reply;
P
Philipp Reisner 已提交
2242
			dec_rs_pending(mdev);
2243 2244 2245
			/* 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
2246 2247 2248 2249 2250
		}
		break;

	case P_OV_REQUEST:
		if (mdev->ov_start_sector == ~(sector_t)0 &&
2251
		    mdev->tconn->agreed_pro_version >= 90) {
2252 2253
			unsigned long now = jiffies;
			int i;
P
Philipp Reisner 已提交
2254 2255
			mdev->ov_start_sector = sector;
			mdev->ov_position = sector;
2256 2257
			mdev->ov_left = drbd_bm_bits(mdev) - BM_SECT_TO_BIT(sector);
			mdev->rs_total = mdev->ov_left;
2258 2259 2260 2261
			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
2262 2263 2264
			dev_info(DEV, "Online Verify start sector: %llu\n",
					(unsigned long long)sector);
		}
2265
		peer_req->w.cb = w_e_end_ov_req;
P
Philipp Reisner 已提交
2266 2267 2268 2269 2270
		fault_type = DRBD_FAULT_RS_RD;
		break;

	default:
		dev_err(DEV, "unexpected command (%s) in receive_DataRequest\n",
2271
		    cmdname(cmd));
P
Philipp Reisner 已提交
2272
		fault_type = DRBD_FAULT_MAX;
2273
		goto out_free_e;
P
Philipp Reisner 已提交
2274 2275
	}

2276 2277 2278 2279 2280 2281 2282 2283 2284 2285 2286 2287 2288 2289 2290 2291 2292 2293 2294 2295 2296 2297
	/* 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.
	 */
2298 2299 2300
	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))
2301
		goto out_free_e;
P
Philipp Reisner 已提交
2302

2303 2304 2305
submit_for_resync:
	atomic_add(size >> 9, &mdev->rs_sect_ev);

2306
submit:
P
Philipp Reisner 已提交
2307
	inc_unacked(mdev);
2308
	spin_lock_irq(&mdev->tconn->req_lock);
2309
	list_add_tail(&peer_req->w.list, &mdev->read_ee);
2310
	spin_unlock_irq(&mdev->tconn->req_lock);
P
Philipp Reisner 已提交
2311

2312
	if (drbd_submit_peer_request(mdev, peer_req, READ, fault_type) == 0)
2313
		return true;
P
Philipp Reisner 已提交
2314

2315 2316
	/* don't care for the reason here */
	dev_err(DEV, "submit failed, triggering re-connect\n");
2317
	spin_lock_irq(&mdev->tconn->req_lock);
2318
	list_del(&peer_req->w.list);
2319
	spin_unlock_irq(&mdev->tconn->req_lock);
2320 2321
	/* no drbd_rs_complete_io(), we are dropping the connection anyways */

P
Philipp Reisner 已提交
2322 2323
out_free_e:
	put_ldev(mdev);
2324
	drbd_free_ee(mdev, peer_req);
2325
	return false;
P
Philipp Reisner 已提交
2326 2327 2328 2329 2330 2331 2332 2333 2334 2335 2336 2337 2338
}

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;

2339
	switch (mdev->tconn->net_conf->after_sb_0p) {
P
Philipp Reisner 已提交
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 2365 2366
	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
2367
		dev_warn(DEV, "Discard younger/older primary did not find a decision\n"
P
Philipp Reisner 已提交
2368 2369 2370
		     "Using discard-least-changes instead\n");
	case ASB_DISCARD_ZERO_CHG:
		if (ch_peer == 0 && ch_self == 0) {
2371
			rv = test_bit(DISCARD_CONCURRENT, &mdev->tconn->flags)
P
Philipp Reisner 已提交
2372 2373 2374 2375 2376 2377
				? -1 : 1;
			break;
		} else {
			if (ch_peer == 0) { rv =  1; break; }
			if (ch_self == 0) { rv = -1; break; }
		}
2378
		if (mdev->tconn->net_conf->after_sb_0p == ASB_DISCARD_ZERO_CHG)
P
Philipp Reisner 已提交
2379 2380 2381 2382 2383 2384 2385 2386
			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. */
2387
			rv = test_bit(DISCARD_CONCURRENT, &mdev->tconn->flags)
P
Philipp Reisner 已提交
2388 2389 2390 2391 2392 2393 2394 2395 2396 2397 2398 2399 2400 2401
				? -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)
{
2402
	int hg, rv = -100;
P
Philipp Reisner 已提交
2403

2404
	switch (mdev->tconn->net_conf->after_sb_1p) {
P
Philipp Reisner 已提交
2405 2406 2407 2408 2409 2410 2411 2412 2413 2414 2415 2416 2417 2418 2419 2420 2421 2422 2423 2424 2425 2426 2427 2428
	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) {
2429 2430 2431
			enum drbd_state_rv rv2;

			drbd_set_role(mdev, R_SECONDARY, 0);
P
Philipp Reisner 已提交
2432 2433 2434
			 /* 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. */
2435 2436
			rv2 = drbd_change_state(mdev, CS_VERBOSE, NS(role, R_SECONDARY));
			if (rv2 != SS_SUCCESS) {
P
Philipp Reisner 已提交
2437 2438 2439 2440 2441 2442 2443 2444 2445 2446 2447 2448 2449 2450
				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)
{
2451
	int hg, rv = -100;
P
Philipp Reisner 已提交
2452

2453
	switch (mdev->tconn->net_conf->after_sb_2p) {
P
Philipp Reisner 已提交
2454 2455 2456 2457 2458 2459 2460 2461 2462 2463 2464 2465 2466 2467 2468 2469 2470
	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) {
2471 2472
			enum drbd_state_rv rv2;

P
Philipp Reisner 已提交
2473 2474 2475
			 /* 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. */
2476 2477
			rv2 = drbd_change_state(mdev, CS_VERBOSE, NS(role, R_SECONDARY));
			if (rv2 != SS_SUCCESS) {
P
Philipp Reisner 已提交
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 2514 2515
				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
2516 2517
-1091   requires proto 91
-1096   requires proto 96
P
Philipp Reisner 已提交
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 2544 2545
 */
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) {

2546
			if (mdev->tconn->agreed_pro_version < 91)
2547
				return -1091;
P
Philipp Reisner 已提交
2548 2549 2550 2551 2552 2553 2554 2555 2556 2557 2558 2559 2560 2561 2562 2563 2564 2565 2566

			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) {

2567
			if (mdev->tconn->agreed_pro_version < 91)
2568
				return -1091;
P
Philipp Reisner 已提交
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 2598 2599

			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 */
2600
			dc = test_bit(DISCARD_CONCURRENT, &mdev->tconn->flags);
P
Philipp Reisner 已提交
2601 2602 2603 2604 2605 2606 2607 2608 2609 2610 2611 2612
			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) {
2613
		if (mdev->tconn->agreed_pro_version < 96 ?
2614 2615 2616
		    (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
2617 2618 2619
			/* The last P_SYNC_UUID did not get though. Undo the last start of
			   resync as sync source modifications of the peer's UUIDs. */

2620
			if (mdev->tconn->agreed_pro_version < 91)
2621
				return -1091;
P
Philipp Reisner 已提交
2622 2623 2624

			mdev->p_uuid[UI_BITMAP] = mdev->p_uuid[UI_HISTORY_START];
			mdev->p_uuid[UI_HISTORY_START] = mdev->p_uuid[UI_HISTORY_START + 1];
2625 2626 2627 2628

			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
2629 2630 2631 2632 2633 2634 2635 2636 2637 2638 2639 2640 2641 2642 2643 2644 2645 2646 2647 2648 2649
			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) {
2650
		if (mdev->tconn->agreed_pro_version < 96 ?
2651 2652 2653
		    (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
2654 2655 2656
			/* The last P_SYNC_UUID did not get though. Undo the last start of
			   resync as sync source modifications of our UUIDs. */

2657
			if (mdev->tconn->agreed_pro_version < 91)
2658
				return -1091;
P
Philipp Reisner 已提交
2659 2660 2661 2662

			_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]);

2663
			dev_info(DEV, "Last syncUUID did not get through, corrected:\n");
P
Philipp Reisner 已提交
2664 2665 2666 2667 2668 2669 2670 2671 2672
			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;
2673
	peer = mdev->p_uuid[UI_CURRENT] & ~((u64)1);
P
Philipp Reisner 已提交
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 2724 2725
	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;
	}
2726 2727
	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
2728 2729 2730 2731 2732 2733 2734 2735 2736 2737 2738 2739 2740
		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");
	}

2741 2742 2743
	if (abs(hg) == 100)
		drbd_khelper(mdev, "initial-split-brain");

2744
	if (hg == 100 || (hg == -100 && mdev->tconn->net_conf->always_asbp)) {
P
Philipp Reisner 已提交
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 2771 2772
		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) {
2773
		if (mdev->tconn->net_conf->want_lose && !(mdev->p_uuid[UI_FLAGS]&1))
P
Philipp Reisner 已提交
2774
			hg = -1;
2775
		if (!mdev->tconn->net_conf->want_lose && (mdev->p_uuid[UI_FLAGS]&1))
P
Philipp Reisner 已提交
2776 2777 2778 2779 2780 2781 2782 2783 2784
			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) {
2785 2786 2787 2788
		/* 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... */
2789
		dev_alert(DEV, "Split-Brain detected but unresolved, dropping connection!\n");
P
Philipp Reisner 已提交
2790 2791 2792 2793 2794 2795 2796 2797 2798 2799 2800
		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) {
2801
		switch (mdev->tconn->net_conf->rr_conflict) {
P
Philipp Reisner 已提交
2802 2803 2804 2805 2806 2807 2808 2809 2810 2811 2812 2813
		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");
		}
	}

2814
	if (mdev->tconn->net_conf->dry_run || test_bit(CONN_DRY_RUN, &mdev->tconn->flags)) {
2815 2816 2817 2818 2819 2820 2821 2822 2823
		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
2824 2825
	if (abs(hg) >= 2) {
		dev_info(DEV, "Writing the whole bitmap, full sync required after drbd_sync_handshake.\n");
2826 2827
		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
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 2865 2866
			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;
}

2867
static int receive_protocol(struct drbd_tconn *tconn, enum drbd_packet cmd,
2868
			    unsigned int data_size)
P
Philipp Reisner 已提交
2869
{
2870
	struct p_protocol *p = &tconn->data.rbuf.protocol;
P
Philipp Reisner 已提交
2871
	int p_proto, p_after_sb_0p, p_after_sb_1p, p_after_sb_2p;
2872
	int p_want_lose, p_two_primaries, cf;
P
Philipp Reisner 已提交
2873 2874 2875 2876 2877 2878 2879
	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);
2880 2881 2882
	cf		= be32_to_cpu(p->conn_flags);
	p_want_lose = cf & CF_WANT_LOSE;

2883
	clear_bit(CONN_DRY_RUN, &tconn->flags);
2884 2885

	if (cf & CF_DRY_RUN)
2886
		set_bit(CONN_DRY_RUN, &tconn->flags);
P
Philipp Reisner 已提交
2887

2888 2889
	if (p_proto != tconn->net_conf->wire_protocol) {
		conn_err(tconn, "incompatible communication protocols\n");
P
Philipp Reisner 已提交
2890 2891 2892
		goto disconnect;
	}

2893 2894
	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
2895 2896 2897
		goto disconnect;
	}

2898 2899
	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
2900 2901 2902
		goto disconnect;
	}

2903 2904
	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
2905 2906 2907
		goto disconnect;
	}

2908 2909
	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
2910 2911 2912
		goto disconnect;
	}

2913 2914
	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
2915 2916 2917
		goto disconnect;
	}

2918 2919
	if (tconn->agreed_pro_version >= 87) {
		unsigned char *my_alg = tconn->net_conf->integrity_alg;
P
Philipp Reisner 已提交
2920

2921
		if (drbd_recv(tconn, p_integrity_alg, data_size) != data_size)
2922
			return false;
P
Philipp Reisner 已提交
2923 2924 2925

		p_integrity_alg[SHARED_SECRET_MAX-1] = 0;
		if (strcmp(p_integrity_alg, my_alg)) {
2926
			conn_err(tconn, "incompatible setting of the data-integrity-alg\n");
P
Philipp Reisner 已提交
2927 2928
			goto disconnect;
		}
2929
		conn_info(tconn, "data-integrity-alg: %s\n",
P
Philipp Reisner 已提交
2930 2931 2932
		     my_alg[0] ? my_alg : (unsigned char *)"<not-used>");
	}

2933
	return true;
P
Philipp Reisner 已提交
2934 2935

disconnect:
2936
	conn_request_state(tconn, NS(conn, C_DISCONNECTING), CS_HARD);
2937
	return false;
P
Philipp Reisner 已提交
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 2965 2966
}

/* 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;
}

2967 2968
static int receive_SyncParam(struct drbd_conf *mdev, enum drbd_packet cmd,
			     unsigned int packet_size)
P
Philipp Reisner 已提交
2969
{
2970
	int ok = true;
2971
	struct p_rs_param_95 *p = &mdev->tconn->data.rbuf.rs_param_95;
P
Philipp Reisner 已提交
2972 2973 2974
	unsigned int header_size, data_size, exp_max_sz;
	struct crypto_hash *verify_tfm = NULL;
	struct crypto_hash *csums_tfm = NULL;
2975
	const int apv = mdev->tconn->agreed_pro_version;
2976 2977
	int *rs_plan_s = NULL;
	int fifo_size = 0;
P
Philipp Reisner 已提交
2978 2979 2980 2981

	exp_max_sz  = apv <= 87 ? sizeof(struct p_rs_param)
		    : apv == 88 ? sizeof(struct p_rs_param)
					+ SHARED_SECRET_MAX
2982 2983
		    : apv <= 94 ? sizeof(struct p_rs_param_89)
		    : /* apv >= 95 */ sizeof(struct p_rs_param_95);
P
Philipp Reisner 已提交
2984

2985
	if (packet_size > exp_max_sz) {
P
Philipp Reisner 已提交
2986
		dev_err(DEV, "SyncParam packet too long: received %u, expected <= %u bytes\n",
2987
		    packet_size, exp_max_sz);
2988
		return false;
P
Philipp Reisner 已提交
2989 2990 2991
	}

	if (apv <= 88) {
2992
		header_size = sizeof(struct p_rs_param) - sizeof(struct p_header);
2993
		data_size   = packet_size  - header_size;
2994
	} else if (apv <= 94) {
2995
		header_size = sizeof(struct p_rs_param_89) - sizeof(struct p_header);
2996
		data_size   = packet_size  - header_size;
P
Philipp Reisner 已提交
2997
		D_ASSERT(data_size == 0);
2998
	} else {
2999
		header_size = sizeof(struct p_rs_param_95) - sizeof(struct p_header);
3000
		data_size   = packet_size  - header_size;
P
Philipp Reisner 已提交
3001 3002 3003 3004 3005 3006
		D_ASSERT(data_size == 0);
	}

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

3007
	if (drbd_recv(mdev->tconn, &p->head.payload, header_size) != header_size)
3008
		return false;
P
Philipp Reisner 已提交
3009

3010 3011 3012 3013
	if (get_ldev(mdev)) {
		mdev->ldev->dc.resync_rate = be32_to_cpu(p->rate);
		put_ldev(mdev);
	}
P
Philipp Reisner 已提交
3014 3015 3016 3017 3018 3019 3020

	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);
3021
				return false;
P
Philipp Reisner 已提交
3022 3023
			}

3024
			if (drbd_recv(mdev->tconn, p->verify_alg, data_size) != data_size)
3025
				return false;
P
Philipp Reisner 已提交
3026 3027 3028 3029 3030 3031 3032 3033 3034 3035 3036 3037 3038 3039 3040

			/* 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;
		}

3041
		if (strcmp(mdev->tconn->net_conf->verify_alg, p->verify_alg)) {
P
Philipp Reisner 已提交
3042 3043
			if (mdev->state.conn == C_WF_REPORT_PARAMS) {
				dev_err(DEV, "Different verify-alg settings. me=\"%s\" peer=\"%s\"\n",
3044
				    mdev->tconn->net_conf->verify_alg, p->verify_alg);
P
Philipp Reisner 已提交
3045 3046 3047 3048 3049 3050 3051 3052 3053 3054
				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;
			}
		}

3055
		if (apv >= 89 && strcmp(mdev->tconn->net_conf->csums_alg, p->csums_alg)) {
P
Philipp Reisner 已提交
3056 3057
			if (mdev->state.conn == C_WF_REPORT_PARAMS) {
				dev_err(DEV, "Different csums-alg settings. me=\"%s\" peer=\"%s\"\n",
3058
				    mdev->tconn->net_conf->csums_alg, p->csums_alg);
P
Philipp Reisner 已提交
3059 3060 3061 3062 3063 3064 3065 3066 3067 3068
				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;
			}
		}

3069 3070 3071 3072 3073 3074
		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);
3075

3076
			fifo_size = (mdev->ldev->dc.c_plan_ahead * 10 * SLEEP_TIME) / HZ;
3077 3078 3079 3080
			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");
3081
					put_ldev(mdev);
3082 3083 3084
					goto disconnect;
				}
			}
3085
			put_ldev(mdev);
3086
		}
P
Philipp Reisner 已提交
3087 3088 3089 3090

		spin_lock(&mdev->peer_seq_lock);
		/* lock against drbd_nl_syncer_conf() */
		if (verify_tfm) {
3091 3092 3093 3094
			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
3095 3096 3097
			dev_info(DEV, "using verify-alg: \"%s\"\n", p->verify_alg);
		}
		if (csums_tfm) {
3098 3099 3100 3101
			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
3102 3103
			dev_info(DEV, "using csums-alg: \"%s\"\n", p->csums_alg);
		}
3104 3105 3106 3107 3108 3109
		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
3110 3111 3112 3113 3114 3115 3116 3117 3118 3119
		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);
3120
	conn_request_state(mdev->tconn, NS(conn, C_DISCONNECTING), CS_HARD);
3121
	return false;
P
Philipp Reisner 已提交
3122 3123 3124 3125 3126 3127 3128 3129 3130 3131 3132 3133 3134 3135 3136
}

/* 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);
}

3137 3138
static int receive_sizes(struct drbd_conf *mdev, enum drbd_packet cmd,
			 unsigned int data_size)
P
Philipp Reisner 已提交
3139
{
3140
	struct p_sizes *p = &mdev->tconn->data.rbuf.sizes;
P
Philipp Reisner 已提交
3141 3142 3143
	enum determine_dev_size dd = unchanged;
	sector_t p_size, p_usize, my_usize;
	int ldsc = 0; /* local disk size changed */
3144
	enum dds_flags ddsf;
P
Philipp Reisner 已提交
3145 3146 3147 3148 3149 3150 3151 3152 3153 3154 3155 3156 3157 3158 3159 3160 3161 3162 3163 3164 3165 3166 3167 3168 3169 3170 3171 3172 3173 3174

	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. */
3175
		if (drbd_new_dev_size(mdev, mdev->ldev, 0) <
P
Philipp Reisner 已提交
3176 3177 3178 3179
		   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");
3180
			conn_request_state(mdev->tconn, NS(conn, C_DISCONNECTING), CS_HARD);
P
Philipp Reisner 已提交
3181 3182
			mdev->ldev->dc.disk_size = my_usize;
			put_ldev(mdev);
3183
			return false;
P
Philipp Reisner 已提交
3184 3185 3186 3187
		}
		put_ldev(mdev);
	}

3188
	ddsf = be16_to_cpu(p->dds_flags);
P
Philipp Reisner 已提交
3189
	if (get_ldev(mdev)) {
B
Bart Van Assche 已提交
3190
		dd = drbd_determine_dev_size(mdev, ddsf);
P
Philipp Reisner 已提交
3191 3192
		put_ldev(mdev);
		if (dd == dev_size_error)
3193
			return false;
P
Philipp Reisner 已提交
3194 3195 3196 3197 3198 3199
		drbd_md_sync(mdev);
	} else {
		/* I am diskless, need to accept the peer's size. */
		drbd_set_my_capacity(mdev, p_size);
	}

3200 3201 3202
	mdev->peer_max_bio_size = be32_to_cpu(p->max_bio_size);
	drbd_reconsider_max_bio_size(mdev);

P
Philipp Reisner 已提交
3203 3204 3205 3206 3207 3208 3209 3210 3211 3212 3213 3214 3215 3216
	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... */
3217
			drbd_send_sizes(mdev, 0, ddsf);
P
Philipp Reisner 已提交
3218 3219 3220 3221
		}
		if (test_and_clear_bit(RESIZE_PENDING, &mdev->flags) ||
		    (dd == grew && mdev->state.conn == C_CONNECTED)) {
			if (mdev->state.pdsk >= D_INCONSISTENT &&
3222 3223 3224 3225 3226 3227
			    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
3228 3229 3230 3231
				set_bit(RESYNC_AFTER_NEG, &mdev->flags);
		}
	}

3232
	return true;
P
Philipp Reisner 已提交
3233 3234
}

3235 3236
static int receive_uuids(struct drbd_conf *mdev, enum drbd_packet cmd,
			 unsigned int data_size)
P
Philipp Reisner 已提交
3237
{
3238
	struct p_uuids *p = &mdev->tconn->data.rbuf.uuids;
P
Philipp Reisner 已提交
3239
	u64 *p_uuid;
3240
	int i, updated_uuids = 0;
P
Philipp Reisner 已提交
3241 3242 3243 3244 3245 3246 3247 3248 3249 3250 3251 3252 3253 3254 3255

	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);
3256
		conn_request_state(mdev->tconn, NS(conn, C_DISCONNECTING), CS_HARD);
3257
		return false;
P
Philipp Reisner 已提交
3258 3259 3260 3261 3262
	}

	if (get_ldev(mdev)) {
		int skip_initial_sync =
			mdev->state.conn == C_CONNECTED &&
3263
			mdev->tconn->agreed_pro_version >= 90 &&
P
Philipp Reisner 已提交
3264 3265 3266 3267 3268
			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,
3269 3270
					"clear_n_write from receive_uuids",
					BM_LOCKED_TEST_ALLOWED);
P
Philipp Reisner 已提交
3271 3272 3273 3274 3275
			_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);
3276
			updated_uuids = 1;
P
Philipp Reisner 已提交
3277 3278
		}
		put_ldev(mdev);
3279 3280 3281 3282
	} 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. */
3283
		updated_uuids = drbd_set_ed_uuid(mdev, p_uuid[UI_CURRENT]);
P
Philipp Reisner 已提交
3284 3285 3286 3287 3288 3289
	}

	/* 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... */
3290 3291
	mutex_lock(mdev->state_mutex);
	mutex_unlock(mdev->state_mutex);
P
Philipp Reisner 已提交
3292
	if (mdev->state.conn >= C_CONNECTED && mdev->state.disk < D_INCONSISTENT)
3293 3294 3295 3296
		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
3297

3298
	return true;
P
Philipp Reisner 已提交
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 3329 3330
}

/**
 * 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;
}

3331 3332
static int receive_req_state(struct drbd_conf *mdev, enum drbd_packet cmd,
			     unsigned int data_size)
P
Philipp Reisner 已提交
3333
{
3334
	struct p_req_state *p = &mdev->tconn->data.rbuf.req_state;
P
Philipp Reisner 已提交
3335
	union drbd_state mask, val;
3336
	enum drbd_state_rv rv;
P
Philipp Reisner 已提交
3337 3338 3339 3340

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

3341
	if (test_bit(DISCARD_CONCURRENT, &mdev->tconn->flags) &&
3342
	    mutex_is_locked(mdev->state_mutex)) {
P
Philipp Reisner 已提交
3343
		drbd_send_sr_reply(mdev, SS_CONCURRENT_ST_CHG);
3344
		return true;
P
Philipp Reisner 已提交
3345 3346 3347 3348 3349
	}

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

3350 3351
	rv = drbd_change_state(mdev, CS_VERBOSE, mask, val);
	drbd_send_sr_reply(mdev, rv);
P
Philipp Reisner 已提交
3352 3353 3354

	drbd_md_sync(mdev);

3355
	return true;
P
Philipp Reisner 已提交
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 3381 3382
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;
}

3383 3384
static int receive_state(struct drbd_conf *mdev, enum drbd_packet cmd,
			 unsigned int data_size)
P
Philipp Reisner 已提交
3385
{
3386
	struct p_state *p = &mdev->tconn->data.rbuf.state;
3387
	union drbd_state os, ns, peer_state;
P
Philipp Reisner 已提交
3388
	enum drbd_disk_state real_peer_disk;
3389
	enum chg_state_flags cs_flags;
P
Philipp Reisner 已提交
3390 3391 3392 3393 3394 3395 3396 3397 3398 3399
	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));
	}

3400
	spin_lock_irq(&mdev->tconn->req_lock);
P
Philipp Reisner 已提交
3401
 retry:
3402
	os = ns = mdev->state;
3403
	spin_unlock_irq(&mdev->tconn->req_lock);
P
Philipp Reisner 已提交
3404

3405 3406 3407 3408 3409 3410 3411 3412 3413 3414 3415 3416 3417 3418 3419 3420 3421 3422 3423 3424 3425
	/* 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);
3426
			return true;
3427 3428 3429 3430 3431 3432 3433 3434 3435 3436 3437 3438
		}
	}

	/* 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;

3439 3440
	if (ns.conn == C_WF_REPORT_PARAMS)
		ns.conn = C_CONNECTED;
P
Philipp Reisner 已提交
3441

3442 3443 3444
	if (peer_state.conn == C_AHEAD)
		ns.conn = C_BEHIND;

P
Philipp Reisner 已提交
3445 3446 3447 3448 3449
	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 */
3450
		cr  = (os.conn < C_CONNECTED);
P
Philipp Reisner 已提交
3451 3452
		/* if we had an established connection
		 * and one of the nodes newly attaches a disk */
3453
		cr |= (os.conn == C_CONNECTED &&
P
Philipp Reisner 已提交
3454
		       (peer_state.disk == D_NEGOTIATING ||
3455
			os.disk == D_NEGOTIATING));
P
Philipp Reisner 已提交
3456 3457 3458 3459 3460
		/* 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" */
3461
		cr |= (os.conn == C_CONNECTED &&
P
Philipp Reisner 已提交
3462 3463 3464 3465
				(peer_state.conn >= C_STARTING_SYNC_S &&
				 peer_state.conn <= C_WF_BITMAP_T));

		if (cr)
3466
			ns.conn = drbd_sync_handshake(mdev, peer_state.role, real_peer_disk);
P
Philipp Reisner 已提交
3467 3468

		put_ldev(mdev);
3469 3470
		if (ns.conn == C_MASK) {
			ns.conn = C_CONNECTED;
P
Philipp Reisner 已提交
3471
			if (mdev->state.disk == D_NEGOTIATING) {
3472
				drbd_force_state(mdev, NS(disk, D_FAILED));
P
Philipp Reisner 已提交
3473 3474 3475
			} 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;
3476
				real_peer_disk = D_DISKLESS;
P
Philipp Reisner 已提交
3477
			} else {
3478
				if (test_and_clear_bit(CONN_DRY_RUN, &mdev->tconn->flags))
3479
					return false;
3480
				D_ASSERT(os.conn == C_WF_REPORT_PARAMS);
3481
				conn_request_state(mdev->tconn, NS(conn, C_DISCONNECTING), CS_HARD);
3482
				return false;
P
Philipp Reisner 已提交
3483 3484 3485 3486
			}
		}
	}

3487
	spin_lock_irq(&mdev->tconn->req_lock);
3488
	if (mdev->state.i != os.i)
P
Philipp Reisner 已提交
3489 3490 3491 3492 3493
		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);
3494
	if ((ns.conn == C_CONNECTED || ns.conn == C_WF_BITMAP_S) && ns.disk == D_NEGOTIATING)
P
Philipp Reisner 已提交
3495
		ns.disk = mdev->new_state_tmp.disk;
3496 3497
	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 &&
3498
	    test_bit(NEW_CUR_UUID, &mdev->flags)) {
3499
		/* Do not allow tl_restart(RESEND) for a rebooted peer. We can only allow this
3500
		   for temporal network outages! */
3501
		spin_unlock_irq(&mdev->tconn->req_lock);
3502
		dev_err(DEV, "Aborting Connect, can not thaw IO with an only Consistent peer\n");
3503
		tl_clear(mdev->tconn);
3504 3505
		drbd_uuid_new_current(mdev);
		clear_bit(NEW_CUR_UUID, &mdev->flags);
3506
		conn_request_state(mdev->tconn, NS2(conn, C_PROTOCOL_ERROR, susp, 0), CS_HARD);
3507
		return false;
3508
	}
3509
	rv = _drbd_set_state(mdev, ns, cs_flags, NULL);
P
Philipp Reisner 已提交
3510
	ns = mdev->state;
3511
	spin_unlock_irq(&mdev->tconn->req_lock);
P
Philipp Reisner 已提交
3512 3513

	if (rv < SS_SUCCESS) {
3514
		conn_request_state(mdev->tconn, NS(conn, C_DISCONNECTING), CS_HARD);
3515
		return false;
P
Philipp Reisner 已提交
3516 3517
	}

3518 3519
	if (os.conn > C_WF_REPORT_PARAMS) {
		if (ns.conn > C_CONNECTED && peer_state.conn <= C_CONNECTED &&
P
Philipp Reisner 已提交
3520 3521 3522 3523 3524 3525 3526 3527 3528
		    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);
		}
	}

3529
	mdev->tconn->net_conf->want_lose = 0;
P
Philipp Reisner 已提交
3530 3531 3532

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

3533
	return true;
P
Philipp Reisner 已提交
3534 3535
}

3536 3537
static int receive_sync_uuid(struct drbd_conf *mdev, enum drbd_packet cmd,
			     unsigned int data_size)
P
Philipp Reisner 已提交
3538
{
3539
	struct p_rs_uuid *p = &mdev->tconn->data.rbuf.rs_uuid;
P
Philipp Reisner 已提交
3540 3541 3542

	wait_event(mdev->misc_wait,
		   mdev->state.conn == C_WF_SYNC_UUID ||
3543
		   mdev->state.conn == C_BEHIND ||
P
Philipp Reisner 已提交
3544 3545 3546 3547 3548 3549 3550 3551 3552 3553 3554
		   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);

3555
		drbd_print_uuids(mdev, "updated sync uuid");
P
Philipp Reisner 已提交
3556 3557 3558 3559 3560 3561
		drbd_start_resync(mdev, C_SYNC_TARGET);

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

3562
	return true;
P
Philipp Reisner 已提交
3563 3564
}

3565 3566 3567 3568 3569 3570 3571
/**
 * receive_bitmap_plain
 *
 * Return 0 when done, 1 when another iteration is needed, and a negative error
 * code upon failure.
 */
static int
3572 3573
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
3574 3575 3576
{
	unsigned num_words = min_t(size_t, BM_PACKET_WORDS, c->bm_words - c->word_offset);
	unsigned want = num_words * sizeof(long);
3577
	int err;
P
Philipp Reisner 已提交
3578

3579 3580
	if (want != data_size) {
		dev_err(DEV, "%s:want (%u) != data_size (%u)\n", __func__, want, data_size);
3581
		return -EIO;
P
Philipp Reisner 已提交
3582 3583
	}
	if (want == 0)
3584
		return 0;
3585
	err = drbd_recv(mdev->tconn, buffer, want);
3586 3587 3588 3589 3590
	if (err != want) {
		if (err >= 0)
			err = -EIO;
		return err;
	}
P
Philipp Reisner 已提交
3591 3592 3593 3594 3595 3596 3597 3598

	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;

3599
	return 1;
P
Philipp Reisner 已提交
3600 3601
}

3602 3603 3604 3605 3606 3607 3608
/**
 * 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
3609 3610
recv_bm_rle_bits(struct drbd_conf *mdev,
		struct p_compressed_bm *p,
3611 3612
		 struct bm_xfer_ctx *c,
		 unsigned int len)
P
Philipp Reisner 已提交
3613 3614 3615 3616 3617 3618 3619 3620 3621 3622 3623 3624 3625 3626 3627
{
	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)
3628
		return -EIO;
P
Philipp Reisner 已提交
3629 3630 3631 3632

	for (have = bits; have > 0; s += rl, toggle = !toggle) {
		bits = vli_decode_bits(&rl, look_ahead);
		if (bits <= 0)
3633
			return -EIO;
P
Philipp Reisner 已提交
3634 3635 3636 3637 3638

		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);
3639
				return -EIO;
P
Philipp Reisner 已提交
3640 3641 3642 3643 3644 3645 3646 3647 3648
			}
			_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);
3649
			return -EIO;
P
Philipp Reisner 已提交
3650 3651 3652 3653 3654 3655
		}
		look_ahead >>= bits;
		have -= bits;

		bits = bitstream_get_bits(&bs, &tmp, 64 - have);
		if (bits < 0)
3656
			return -EIO;
P
Philipp Reisner 已提交
3657 3658 3659 3660 3661 3662 3663
		look_ahead |= tmp << have;
		have += bits;
	}

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

3664
	return (s != c->bm_bits);
P
Philipp Reisner 已提交
3665 3666
}

3667 3668 3669 3670 3671 3672 3673
/**
 * 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
3674 3675
decode_bitmap_c(struct drbd_conf *mdev,
		struct p_compressed_bm *p,
3676 3677
		struct bm_xfer_ctx *c,
		unsigned int len)
P
Philipp Reisner 已提交
3678 3679
{
	if (DCBP_get_code(p) == RLE_VLI_Bits)
3680
		return recv_bm_rle_bits(mdev, p, c, len);
P
Philipp Reisner 已提交
3681 3682 3683 3684 3685 3686

	/* 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);
3687
	conn_request_state(mdev->tconn, NS(conn, C_PROTOCOL_ERROR), CS_HARD);
3688
	return -EIO;
P
Philipp Reisner 已提交
3689 3690 3691 3692 3693 3694
}

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" */
3695
	unsigned plain = sizeof(struct p_header) *
P
Philipp Reisner 已提交
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 3731 3732
		((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. */
3733 3734
static int receive_bitmap(struct drbd_conf *mdev, enum drbd_packet cmd,
			  unsigned int data_size)
P
Philipp Reisner 已提交
3735 3736 3737
{
	struct bm_xfer_ctx c;
	void *buffer;
3738
	int err;
3739
	int ok = false;
3740
	struct p_header *h = &mdev->tconn->data.rbuf.header;
3741
	struct packet_info pi;
P
Philipp Reisner 已提交
3742

3743 3744 3745
	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
3746 3747 3748 3749 3750 3751 3752 3753 3754 3755 3756 3757 3758 3759

	/* 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),
	};

3760
	for(;;) {
3761
		if (cmd == P_BITMAP) {
3762
			err = receive_bitmap_plain(mdev, data_size, buffer, &c);
3763
		} else if (cmd == P_COMPRESSED_BITMAP) {
P
Philipp Reisner 已提交
3764 3765 3766 3767
			/* MAYBE: sanity check that we speak proto >= 90,
			 * and the feature is enabled! */
			struct p_compressed_bm *p;

3768
			if (data_size > BM_PACKET_PAYLOAD_BYTES) {
P
Philipp Reisner 已提交
3769 3770 3771 3772 3773 3774
				dev_err(DEV, "ReportCBitmap packet too large\n");
				goto out;
			}
			/* use the page buff */
			p = buffer;
			memcpy(p, h, sizeof(*h));
3775
			if (drbd_recv(mdev->tconn, p->head.payload, data_size) != data_size)
P
Philipp Reisner 已提交
3776
				goto out;
3777 3778
			if (data_size <= (sizeof(*p) - sizeof(p->head))) {
				dev_err(DEV, "ReportCBitmap packet too small (l:%u)\n", data_size);
3779
				goto out;
P
Philipp Reisner 已提交
3780
			}
3781
			err = decode_bitmap_c(mdev, p, &c, data_size);
P
Philipp Reisner 已提交
3782
		} else {
3783
			dev_warn(DEV, "receive_bitmap: cmd neither ReportBitMap nor ReportCBitMap (is 0x%x)", cmd);
P
Philipp Reisner 已提交
3784 3785 3786
			goto out;
		}

3787
		c.packets[cmd == P_BITMAP]++;
3788
		c.bytes[cmd == P_BITMAP] += sizeof(struct p_header) + data_size;
P
Philipp Reisner 已提交
3789

3790 3791 3792
		if (err <= 0) {
			if (err < 0)
				goto out;
P
Philipp Reisner 已提交
3793
			break;
3794
		}
3795
		if (drbd_recv_header(mdev->tconn, &pi))
P
Philipp Reisner 已提交
3796
			goto out;
3797 3798
		cmd = pi.cmd;
		data_size = pi.size;
3799
	}
P
Philipp Reisner 已提交
3800 3801 3802 3803

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

	if (mdev->state.conn == C_WF_BITMAP_T) {
3804 3805
		enum drbd_state_rv rv;

P
Philipp Reisner 已提交
3806 3807 3808 3809
		ok = !drbd_send_bitmap(mdev);
		if (!ok)
			goto out;
		/* Omit CS_ORDERED with this state transition to avoid deadlocks. */
3810 3811
		rv = _drbd_request_state(mdev, NS(conn, C_WF_SYNC_UUID), CS_VERBOSE);
		D_ASSERT(rv == SS_SUCCESS);
P
Philipp Reisner 已提交
3812 3813 3814 3815 3816 3817 3818
	} 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));
	}

3819
	ok = true;
P
Philipp Reisner 已提交
3820
 out:
3821
	drbd_bm_unlock(mdev);
P
Philipp Reisner 已提交
3822 3823 3824 3825 3826 3827
	if (ok && mdev->state.conn == C_WF_BITMAP_S)
		drbd_start_resync(mdev, C_SYNC_SOURCE);
	free_page((unsigned long) buffer);
	return ok;
}

3828
static int _tconn_receive_skip(struct drbd_tconn *tconn, unsigned int data_size)
P
Philipp Reisner 已提交
3829 3830 3831 3832 3833
{
	/* TODO zero copy sink :) */
	static char sink[128];
	int size, want, r;

3834
	size = data_size;
P
Philipp Reisner 已提交
3835 3836
	while (size > 0) {
		want = min_t(int, size, sizeof(sink));
3837 3838
		r = drbd_recv(tconn, sink, want);
		if (r <= 0)
3839
			break;
P
Philipp Reisner 已提交
3840 3841 3842 3843 3844
		size -= r;
	}
	return size == 0;
}

3845 3846 3847 3848 3849 3850 3851 3852 3853 3854 3855 3856 3857 3858 3859 3860 3861
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);
}

3862 3863
static int receive_UnplugRemote(struct drbd_conf *mdev, enum drbd_packet cmd,
				unsigned int data_size)
3864
{
3865 3866
	/* Make sure we've acked all the TCP data associated
	 * with the data requests being unplugged */
3867
	drbd_tcp_quickack(mdev->tconn->data.socket);
3868

3869
	return true;
3870 3871
}

3872 3873
static int receive_out_of_sync(struct drbd_conf *mdev, enum drbd_packet cmd,
			       unsigned int data_size)
3874
{
3875
	struct p_block_desc *p = &mdev->tconn->data.rbuf.block_desc;
3876

3877 3878 3879 3880 3881 3882 3883 3884 3885 3886
	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));
	}

3887 3888
	drbd_set_out_of_sync(mdev, be64_to_cpu(p->sector), be32_to_cpu(p->blksize));

3889
	return true;
3890 3891
}

3892 3893 3894
struct data_cmd {
	int expect_payload;
	size_t pkt_size;
3895
	enum mdev_or_conn fa_type; /* first argument's type */
3896 3897 3898 3899 3900 3901
	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);
	};
3902 3903 3904
};

static struct data_cmd drbd_cmd_handler[] = {
3905 3906 3907 3908 3909 3910 3911 3912 3913 3914 3915
	[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 } },
3916
	[P_PROTOCOL]        = { 1, sizeof(struct p_protocol), CONN, { .conn_fn = receive_protocol } },
3917 3918 3919 3920 3921 3922 3923 3924 3925 3926
	[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 } },
3927
	[P_CONN_ST_CHG_REQ] = { 0, sizeof(struct p_req_state), CONN, { .conn_fn = receive_req_conn_state } },
P
Philipp Reisner 已提交
3928 3929
};

3930
/* All handler functions that expect a sub-header get that sub-heder in
3931
   mdev->tconn->data.rbuf.header.head.payload.
3932

3933
   Usually in mdev->tconn->data.rbuf.header.head the callback can find the usual
3934 3935
   p_header, but they may not rely on that. Since there is also p_header95 !
 */
P
Philipp Reisner 已提交
3936

3937
static void drbdd(struct drbd_tconn *tconn)
P
Philipp Reisner 已提交
3938
{
3939
	struct p_header *header = &tconn->data.rbuf.header;
3940
	struct packet_info pi;
3941 3942
	size_t shs; /* sub header size */
	int rv;
P
Philipp Reisner 已提交
3943

3944 3945
	while (get_t_state(&tconn->receiver) == RUNNING) {
		drbd_thread_current_set_cpu(&tconn->receiver);
3946
		if (drbd_recv_header(tconn, &pi))
3947
			goto err_out;
P
Philipp Reisner 已提交
3948

A
Andreas Gruenbacher 已提交
3949
		if (unlikely(pi.cmd >= ARRAY_SIZE(drbd_cmd_handler) ||
3950
		    !drbd_cmd_handler[pi.cmd].mdev_fn)) {
3951
			conn_err(tconn, "unknown packet type %d, l: %d!\n", pi.cmd, pi.size);
3952
			goto err_out;
3953
		}
P
Philipp Reisner 已提交
3954

3955 3956
		shs = drbd_cmd_handler[pi.cmd].pkt_size - sizeof(struct p_header);
		if (pi.size - shs > 0 && !drbd_cmd_handler[pi.cmd].expect_payload) {
3957
			conn_err(tconn, "No payload expected %s l:%d\n", cmdname(pi.cmd), pi.size);
3958
			goto err_out;
P
Philipp Reisner 已提交
3959 3960
		}

3961
		if (shs) {
3962
			rv = drbd_recv(tconn, &header->payload, shs);
3963
			if (unlikely(rv != shs)) {
3964
				if (!signal_pending(current))
3965
					conn_warn(tconn, "short read while reading sub header: rv=%d\n", rv);
3966 3967 3968 3969
				goto err_out;
			}
		}

3970
		if (drbd_cmd_handler[pi.cmd].fa_type == CONN) {
3971 3972 3973 3974 3975 3976 3977
			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
3978

3979
		if (unlikely(!rv)) {
3980
			conn_err(tconn, "error receiving %s, l: %d!\n",
3981
			    cmdname(pi.cmd), pi.size);
3982
			goto err_out;
P
Philipp Reisner 已提交
3983 3984 3985
		}
	}

3986 3987
	if (0) {
	err_out:
3988
		conn_request_state(tconn, NS(conn, C_PROTOCOL_ERROR), CS_HARD);
3989
	}
P
Philipp Reisner 已提交
3990 3991
}

3992
void conn_flush_workqueue(struct drbd_tconn *tconn)
P
Philipp Reisner 已提交
3993 3994 3995 3996
{
	struct drbd_wq_barrier barr;

	barr.w.cb = w_prev_work_done;
3997
	barr.w.tconn = tconn;
P
Philipp Reisner 已提交
3998
	init_completion(&barr.done);
3999
	drbd_queue_work(&tconn->data.work, &barr.w);
P
Philipp Reisner 已提交
4000 4001 4002
	wait_for_completion(&barr.done);
}

4003
static void drbd_disconnect(struct drbd_tconn *tconn)
P
Philipp Reisner 已提交
4004
{
4005
	enum drbd_conns oc;
P
Philipp Reisner 已提交
4006 4007
	int rv = SS_UNKNOWN_ERROR;

4008
	if (tconn->cstate == C_STANDALONE)
P
Philipp Reisner 已提交
4009 4010 4011
		return;

	/* asender does not clean up anything. it must not interfere, either */
4012 4013 4014 4015 4016 4017 4018 4019
	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);
4020 4021 4022 4023
	oc = tconn->cstate;
	if (oc >= C_UNCONNECTED)
		rv = _conn_request_state(tconn, NS(conn, C_UNCONNECTED), CS_VERBOSE);

4024 4025
	spin_unlock_irq(&tconn->req_lock);

4026
	if (oc == C_DISCONNECTING) {
4027 4028 4029 4030 4031 4032 4033
		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;
4034
		conn_request_state(tconn, NS(conn, C_STANDALONE), CS_VERBOSE);
4035 4036 4037 4038 4039 4040 4041 4042
	}
}

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
4043

4044
	/* wait for current activity to cease. */
4045
	spin_lock_irq(&mdev->tconn->req_lock);
P
Philipp Reisner 已提交
4046 4047 4048
	_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);
4049
	spin_unlock_irq(&mdev->tconn->req_lock);
P
Philipp Reisner 已提交
4050 4051 4052 4053 4054 4055 4056 4057 4058 4059 4060 4061 4062 4063 4064 4065 4066

	/* 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);

4067 4068
	del_timer(&mdev->request_timer);

P
Philipp Reisner 已提交
4069 4070 4071 4072 4073 4074
	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" */
4075
	drbd_flush_workqueue(mdev);
P
Philipp Reisner 已提交
4076 4077 4078 4079 4080 4081 4082 4083

	/* 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;

4084
	if (!is_susp(mdev->state))
4085
		tl_clear(mdev->tconn);
P
Philipp Reisner 已提交
4086 4087 4088 4089 4090 4091 4092 4093 4094

	drbd_md_sync(mdev);

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

4095 4096
	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
4097

4098 4099 4100 4101
	/* 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
4102 4103 4104 4105 4106 4107 4108 4109 4110 4111
	/* 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);
4112 4113 4114
	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
4115 4116
	i = atomic_read(&mdev->pp_in_use);
	if (i)
4117
		dev_info(DEV, "pp_in_use = %d, expected 0\n", i);
P
Philipp Reisner 已提交
4118 4119 4120 4121 4122 4123 4124 4125 4126

	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));
4127 4128

	return 0;
P
Philipp Reisner 已提交
4129 4130 4131 4132 4133 4134 4135 4136 4137 4138 4139
}

/*
 * 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.
 */
4140
static int drbd_send_handshake(struct drbd_tconn *tconn)
P
Philipp Reisner 已提交
4141
{
4142
	/* ASSERT current == mdev->tconn->receiver ... */
4143
	struct p_handshake *p = &tconn->data.sbuf.handshake;
4144
	int err;
P
Philipp Reisner 已提交
4145

4146 4147
	if (mutex_lock_interruptible(&tconn->data.mutex)) {
		conn_err(tconn, "interrupted during initial handshake\n");
4148
		return -EINTR;
P
Philipp Reisner 已提交
4149 4150
	}

4151 4152
	if (tconn->data.socket == NULL) {
		mutex_unlock(&tconn->data.mutex);
4153
		return -EIO;
P
Philipp Reisner 已提交
4154 4155 4156 4157 4158
	}

	memset(p, 0, sizeof(*p));
	p->protocol_min = cpu_to_be32(PRO_VERSION_MIN);
	p->protocol_max = cpu_to_be32(PRO_VERSION_MAX);
4159
	err = _conn_send_cmd(tconn, 0, tconn->data.socket, P_HAND_SHAKE,
4160
			     &p->head, sizeof(*p), 0);
4161
	mutex_unlock(&tconn->data.mutex);
4162
	return err;
P
Philipp Reisner 已提交
4163 4164 4165 4166 4167 4168 4169 4170 4171
}

/*
 * 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.
 */
4172
static int drbd_do_handshake(struct drbd_tconn *tconn)
P
Philipp Reisner 已提交
4173
{
4174 4175
	/* ASSERT current == tconn->receiver ... */
	struct p_handshake *p = &tconn->data.rbuf.handshake;
4176
	const int expect = sizeof(struct p_handshake) - sizeof(struct p_header80);
4177
	struct packet_info pi;
4178
	int err, rv;
P
Philipp Reisner 已提交
4179

4180 4181
	err = drbd_send_handshake(tconn);
	if (err)
P
Philipp Reisner 已提交
4182 4183
		return 0;

4184 4185
	err = drbd_recv_header(tconn, &pi);
	if (err)
P
Philipp Reisner 已提交
4186 4187
		return 0;

4188
	if (pi.cmd != P_HAND_SHAKE) {
4189
		conn_err(tconn, "expected HandShake packet, received: %s (0x%04x)\n",
4190
		     cmdname(pi.cmd), pi.cmd);
P
Philipp Reisner 已提交
4191 4192 4193
		return -1;
	}

4194
	if (pi.size != expect) {
4195
		conn_err(tconn, "expected HandShake length: %u, received: %u\n",
4196
		     expect, pi.size);
P
Philipp Reisner 已提交
4197 4198 4199
		return -1;
	}

4200
	rv = drbd_recv(tconn, &p->head.payload, expect);
P
Philipp Reisner 已提交
4201 4202

	if (rv != expect) {
4203
		if (!signal_pending(current))
4204
			conn_warn(tconn, "short read receiving handshake packet: l=%u\n", rv);
P
Philipp Reisner 已提交
4205 4206 4207 4208 4209 4210 4211 4212 4213 4214 4215 4216
		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;

4217
	tconn->agreed_pro_version = min_t(int, PRO_VERSION_MAX, p->protocol_max);
P
Philipp Reisner 已提交
4218

4219 4220
	conn_info(tconn, "Handshake successful: "
	     "Agreed network protocol version %d\n", tconn->agreed_pro_version);
P
Philipp Reisner 已提交
4221 4222 4223 4224

	return 1;

 incompat:
4225
	conn_err(tconn, "incompatible DRBD dialects: "
P
Philipp Reisner 已提交
4226 4227 4228 4229 4230 4231 4232
	    "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)
4233
static int drbd_do_auth(struct drbd_tconn *tconn)
P
Philipp Reisner 已提交
4234 4235 4236
{
	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");
4237
	return -1;
P
Philipp Reisner 已提交
4238 4239 4240
}
#else
#define CHALLENGE_LEN 64
4241 4242 4243 4244 4245 4246 4247

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

4248
static int drbd_do_auth(struct drbd_tconn *tconn)
P
Philipp Reisner 已提交
4249 4250 4251 4252 4253 4254
{
	char my_challenge[CHALLENGE_LEN];  /* 64 Bytes... */
	struct scatterlist sg;
	char *response = NULL;
	char *right_response = NULL;
	char *peers_ch = NULL;
4255
	unsigned int key_len = strlen(tconn->net_conf->shared_secret);
P
Philipp Reisner 已提交
4256 4257
	unsigned int resp_size;
	struct hash_desc desc;
4258
	struct packet_info pi;
4259
	int err, rv;
P
Philipp Reisner 已提交
4260

4261
	desc.tfm = tconn->cram_hmac_tfm;
P
Philipp Reisner 已提交
4262 4263
	desc.flags = 0;

4264 4265
	rv = crypto_hash_setkey(tconn->cram_hmac_tfm,
				(u8 *)tconn->net_conf->shared_secret, key_len);
P
Philipp Reisner 已提交
4266
	if (rv) {
4267
		conn_err(tconn, "crypto_hash_setkey() failed with %d\n", rv);
4268
		rv = -1;
P
Philipp Reisner 已提交
4269 4270 4271 4272 4273
		goto fail;
	}

	get_random_bytes(my_challenge, CHALLENGE_LEN);

4274
	rv = !conn_send_cmd2(tconn, P_AUTH_CHALLENGE, my_challenge, CHALLENGE_LEN);
P
Philipp Reisner 已提交
4275 4276 4277
	if (!rv)
		goto fail;

4278 4279 4280
	err = drbd_recv_header(tconn, &pi);
	if (err) {
		rv = 0;
P
Philipp Reisner 已提交
4281
		goto fail;
4282
	}
P
Philipp Reisner 已提交
4283

4284
	if (pi.cmd != P_AUTH_CHALLENGE) {
4285
		conn_err(tconn, "expected AuthChallenge packet, received: %s (0x%04x)\n",
4286
		    cmdname(pi.cmd), pi.cmd);
P
Philipp Reisner 已提交
4287 4288 4289 4290
		rv = 0;
		goto fail;
	}

4291
	if (pi.size > CHALLENGE_LEN * 2) {
4292
		conn_err(tconn, "expected AuthChallenge payload too big.\n");
4293
		rv = -1;
P
Philipp Reisner 已提交
4294 4295 4296
		goto fail;
	}

4297
	peers_ch = kmalloc(pi.size, GFP_NOIO);
P
Philipp Reisner 已提交
4298
	if (peers_ch == NULL) {
4299
		conn_err(tconn, "kmalloc of peers_ch failed\n");
4300
		rv = -1;
P
Philipp Reisner 已提交
4301 4302 4303
		goto fail;
	}

4304
	rv = drbd_recv(tconn, peers_ch, pi.size);
P
Philipp Reisner 已提交
4305

4306
	if (rv != pi.size) {
4307
		if (!signal_pending(current))
4308
			conn_warn(tconn, "short read AuthChallenge: l=%u\n", rv);
P
Philipp Reisner 已提交
4309 4310 4311 4312
		rv = 0;
		goto fail;
	}

4313
	resp_size = crypto_hash_digestsize(tconn->cram_hmac_tfm);
P
Philipp Reisner 已提交
4314 4315
	response = kmalloc(resp_size, GFP_NOIO);
	if (response == NULL) {
4316
		conn_err(tconn, "kmalloc of response failed\n");
4317
		rv = -1;
P
Philipp Reisner 已提交
4318 4319 4320 4321
		goto fail;
	}

	sg_init_table(&sg, 1);
4322
	sg_set_buf(&sg, peers_ch, pi.size);
P
Philipp Reisner 已提交
4323 4324 4325

	rv = crypto_hash_digest(&desc, &sg, sg.length, response);
	if (rv) {
4326
		conn_err(tconn, "crypto_hash_digest() failed with %d\n", rv);
4327
		rv = -1;
P
Philipp Reisner 已提交
4328 4329 4330
		goto fail;
	}

4331
	rv = !conn_send_cmd2(tconn, P_AUTH_RESPONSE, response, resp_size);
P
Philipp Reisner 已提交
4332 4333 4334
	if (!rv)
		goto fail;

4335 4336 4337
	err = drbd_recv_header(tconn, &pi);
	if (err) {
		rv = 0;
P
Philipp Reisner 已提交
4338
		goto fail;
4339
	}
P
Philipp Reisner 已提交
4340

4341
	if (pi.cmd != P_AUTH_RESPONSE) {
4342
		conn_err(tconn, "expected AuthResponse packet, received: %s (0x%04x)\n",
4343
			cmdname(pi.cmd), pi.cmd);
P
Philipp Reisner 已提交
4344 4345 4346 4347
		rv = 0;
		goto fail;
	}

4348
	if (pi.size != resp_size) {
4349
		conn_err(tconn, "expected AuthResponse payload of wrong size\n");
P
Philipp Reisner 已提交
4350 4351 4352 4353
		rv = 0;
		goto fail;
	}

4354
	rv = drbd_recv(tconn, response , resp_size);
P
Philipp Reisner 已提交
4355 4356

	if (rv != resp_size) {
4357
		if (!signal_pending(current))
4358
			conn_warn(tconn, "short read receiving AuthResponse: l=%u\n", rv);
P
Philipp Reisner 已提交
4359 4360 4361 4362 4363
		rv = 0;
		goto fail;
	}

	right_response = kmalloc(resp_size, GFP_NOIO);
4364
	if (right_response == NULL) {
4365
		conn_err(tconn, "kmalloc of right_response failed\n");
4366
		rv = -1;
P
Philipp Reisner 已提交
4367 4368 4369 4370 4371 4372 4373
		goto fail;
	}

	sg_set_buf(&sg, my_challenge, CHALLENGE_LEN);

	rv = crypto_hash_digest(&desc, &sg, sg.length, right_response);
	if (rv) {
4374
		conn_err(tconn, "crypto_hash_digest() failed with %d\n", rv);
4375
		rv = -1;
P
Philipp Reisner 已提交
4376 4377 4378 4379 4380 4381
		goto fail;
	}

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

	if (rv)
4382 4383
		conn_info(tconn, "Peer authenticated using %d bytes of '%s' HMAC\n",
		     resp_size, tconn->net_conf->cram_hmac_alg);
4384 4385
	else
		rv = -1;
P
Philipp Reisner 已提交
4386 4387 4388 4389 4390 4391 4392 4393 4394 4395 4396 4397

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

	return rv;
}
#endif

int drbdd_init(struct drbd_thread *thi)
{
4398
	struct drbd_tconn *tconn = thi->tconn;
P
Philipp Reisner 已提交
4399 4400
	int h;

4401
	conn_info(tconn, "receiver (re)started\n");
P
Philipp Reisner 已提交
4402 4403

	do {
4404
		h = drbd_connect(tconn);
P
Philipp Reisner 已提交
4405
		if (h == 0) {
4406
			drbd_disconnect(tconn);
4407
			schedule_timeout_interruptible(HZ);
P
Philipp Reisner 已提交
4408 4409
		}
		if (h == -1) {
4410
			conn_warn(tconn, "Discarding network configuration.\n");
4411
			conn_request_state(tconn, NS(conn, C_DISCONNECTING), CS_HARD);
P
Philipp Reisner 已提交
4412 4413 4414 4415
		}
	} while (h == 0);

	if (h > 0) {
4416 4417 4418
		if (get_net_conf(tconn)) {
			drbdd(tconn);
			put_net_conf(tconn);
P
Philipp Reisner 已提交
4419 4420 4421
		}
	}

4422
	drbd_disconnect(tconn);
P
Philipp Reisner 已提交
4423

4424
	conn_info(tconn, "receiver terminated\n");
P
Philipp Reisner 已提交
4425 4426 4427 4428 4429
	return 0;
}

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

4430 4431 4432 4433 4434 4435 4436 4437 4438 4439 4440 4441 4442 4443 4444 4445 4446
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;
}

4447
static int got_RqSReply(struct drbd_conf *mdev, enum drbd_packet cmd)
P
Philipp Reisner 已提交
4448
{
4449
	struct p_req_state_reply *p = &mdev->tconn->meta.rbuf.req_state_reply;
P
Philipp Reisner 已提交
4450 4451
	int retcode = be32_to_cpu(p->retcode);

4452 4453 4454 4455 4456 4457
	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
4458
	}
4459 4460
	wake_up(&mdev->state_wait);

4461
	return true;
P
Philipp Reisner 已提交
4462 4463
}

4464
static int got_Ping(struct drbd_tconn *tconn, enum drbd_packet cmd)
P
Philipp Reisner 已提交
4465
{
4466
	return drbd_send_ping_ack(tconn);
P
Philipp Reisner 已提交
4467 4468 4469

}

4470
static int got_PingAck(struct drbd_tconn *tconn, enum drbd_packet cmd)
P
Philipp Reisner 已提交
4471 4472
{
	/* restore idle timeout */
4473 4474 4475
	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
4476

4477
	return true;
P
Philipp Reisner 已提交
4478 4479
}

4480
static int got_IsInSync(struct drbd_conf *mdev, enum drbd_packet cmd)
P
Philipp Reisner 已提交
4481
{
4482
	struct p_block_ack *p = &mdev->tconn->meta.rbuf.block_ack;
P
Philipp Reisner 已提交
4483 4484 4485
	sector_t sector = be64_to_cpu(p->sector);
	int blksize = be32_to_cpu(p->blksize);

4486
	D_ASSERT(mdev->tconn->agreed_pro_version >= 89);
P
Philipp Reisner 已提交
4487 4488 4489

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

4490 4491 4492 4493 4494 4495 4496
	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
4497
	dec_rs_pending(mdev);
4498
	atomic_add(blksize >> 9, &mdev->rs_sect_in);
P
Philipp Reisner 已提交
4499

4500
	return true;
P
Philipp Reisner 已提交
4501 4502
}

4503 4504 4505 4506
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
4507 4508 4509 4510
{
	struct drbd_request *req;
	struct bio_and_error m;

4511
	spin_lock_irq(&mdev->tconn->req_lock);
4512
	req = find_request(mdev, root, id, sector, missing_ok, func);
P
Philipp Reisner 已提交
4513
	if (unlikely(!req)) {
4514
		spin_unlock_irq(&mdev->tconn->req_lock);
4515
		return false;
P
Philipp Reisner 已提交
4516 4517
	}
	__req_mod(req, what, &m);
4518
	spin_unlock_irq(&mdev->tconn->req_lock);
P
Philipp Reisner 已提交
4519 4520 4521

	if (m.bio)
		complete_master_bio(mdev, &m);
4522
	return true;
P
Philipp Reisner 已提交
4523 4524
}

4525
static int got_BlockAck(struct drbd_conf *mdev, enum drbd_packet cmd)
P
Philipp Reisner 已提交
4526
{
4527
	struct p_block_ack *p = &mdev->tconn->meta.rbuf.block_ack;
P
Philipp Reisner 已提交
4528 4529 4530 4531 4532 4533
	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));

4534
	if (p->block_id == ID_SYNCER) {
P
Philipp Reisner 已提交
4535 4536
		drbd_set_in_sync(mdev, sector, blksize);
		dec_rs_pending(mdev);
4537
		return true;
P
Philipp Reisner 已提交
4538
	}
4539
	switch (cmd) {
P
Philipp Reisner 已提交
4540
	case P_RS_WRITE_ACK:
4541
		D_ASSERT(mdev->tconn->net_conf->wire_protocol == DRBD_PROT_C);
4542
		what = WRITE_ACKED_BY_PEER_AND_SIS;
P
Philipp Reisner 已提交
4543 4544
		break;
	case P_WRITE_ACK:
4545
		D_ASSERT(mdev->tconn->net_conf->wire_protocol == DRBD_PROT_C);
4546
		what = WRITE_ACKED_BY_PEER;
P
Philipp Reisner 已提交
4547 4548
		break;
	case P_RECV_ACK:
4549
		D_ASSERT(mdev->tconn->net_conf->wire_protocol == DRBD_PROT_B);
4550
		what = RECV_ACKED_BY_PEER;
P
Philipp Reisner 已提交
4551
		break;
4552
	case P_DISCARD_WRITE:
4553
		D_ASSERT(mdev->tconn->net_conf->wire_protocol == DRBD_PROT_C);
4554 4555 4556 4557 4558
		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
4559 4560 4561
		break;
	default:
		D_ASSERT(0);
4562
		return false;
P
Philipp Reisner 已提交
4563 4564 4565
	}

	return validate_req_change_req_state(mdev, p->block_id, sector,
4566 4567
					     &mdev->write_requests, __func__,
					     what, false);
P
Philipp Reisner 已提交
4568 4569
}

4570
static int got_NegAck(struct drbd_conf *mdev, enum drbd_packet cmd)
P
Philipp Reisner 已提交
4571
{
4572
	struct p_block_ack *p = &mdev->tconn->meta.rbuf.block_ack;
P
Philipp Reisner 已提交
4573
	sector_t sector = be64_to_cpu(p->sector);
4574
	int size = be32_to_cpu(p->blksize);
4575 4576
	bool missing_ok = mdev->tconn->net_conf->wire_protocol == DRBD_PROT_A ||
			  mdev->tconn->net_conf->wire_protocol == DRBD_PROT_B;
4577
	bool found;
P
Philipp Reisner 已提交
4578 4579 4580

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

4581
	if (p->block_id == ID_SYNCER) {
P
Philipp Reisner 已提交
4582 4583
		dec_rs_pending(mdev);
		drbd_rs_failed_io(mdev, sector, size);
4584
		return true;
P
Philipp Reisner 已提交
4585
	}
4586

4587
	found = validate_req_change_req_state(mdev, p->block_id, sector,
4588
					      &mdev->write_requests, __func__,
4589
					      NEG_ACKED, missing_ok);
4590 4591 4592 4593 4594 4595 4596
	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)
4597
			return false;
4598
		drbd_set_out_of_sync(mdev, sector, size);
4599 4600
	}
	return true;
P
Philipp Reisner 已提交
4601 4602
}

4603
static int got_NegDReply(struct drbd_conf *mdev, enum drbd_packet cmd)
P
Philipp Reisner 已提交
4604
{
4605
	struct p_block_ack *p = &mdev->tconn->meta.rbuf.block_ack;
P
Philipp Reisner 已提交
4606 4607 4608
	sector_t sector = be64_to_cpu(p->sector);

	update_peer_seq(mdev, be32_to_cpu(p->seq_num));
4609

P
Philipp Reisner 已提交
4610 4611 4612 4613
	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,
4614
					     &mdev->read_requests, __func__,
4615
					     NEG_ACKED, false);
P
Philipp Reisner 已提交
4616 4617
}

4618
static int got_NegRSDReply(struct drbd_conf *mdev, enum drbd_packet cmd)
P
Philipp Reisner 已提交
4619 4620 4621
{
	sector_t sector;
	int size;
4622
	struct p_block_ack *p = &mdev->tconn->meta.rbuf.block_ack;
P
Philipp Reisner 已提交
4623 4624 4625 4626 4627 4628 4629 4630 4631 4632

	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);
4633
		switch (cmd) {
4634 4635 4636 4637 4638 4639 4640 4641 4642
		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
4643 4644 4645
		put_ldev(mdev);
	}

4646
	return true;
P
Philipp Reisner 已提交
4647 4648
}

4649
static int got_BarrierAck(struct drbd_conf *mdev, enum drbd_packet cmd)
P
Philipp Reisner 已提交
4650
{
4651
	struct p_barrier_ack *p = &mdev->tconn->meta.rbuf.barrier_ack;
P
Philipp Reisner 已提交
4652

4653
	tl_release(mdev->tconn, p->barrier, be32_to_cpu(p->set_size));
P
Philipp Reisner 已提交
4654

4655 4656
	if (mdev->state.conn == C_AHEAD &&
	    atomic_read(&mdev->ap_in_flight) == 0 &&
4657 4658 4659
	    !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);
4660 4661
	}

4662
	return true;
P
Philipp Reisner 已提交
4663 4664
}

4665
static int got_OVResult(struct drbd_conf *mdev, enum drbd_packet cmd)
P
Philipp Reisner 已提交
4666
{
4667
	struct p_block_ack *p = &mdev->tconn->meta.rbuf.block_ack;
P
Philipp Reisner 已提交
4668 4669 4670 4671 4672 4673 4674 4675 4676 4677 4678 4679 4680 4681
	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);

4682
	if (!get_ldev(mdev))
4683
		return true;
4684

P
Philipp Reisner 已提交
4685 4686 4687
	drbd_rs_complete_io(mdev, sector);
	dec_rs_pending(mdev);

4688 4689 4690 4691 4692 4693 4694
	--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
4695 4696 4697
		w = kmalloc(sizeof(*w), GFP_NOIO);
		if (w) {
			w->cb = w_ov_finished;
4698
			w->mdev = mdev;
4699
			drbd_queue_work_front(&mdev->tconn->data.work, w);
P
Philipp Reisner 已提交
4700 4701 4702 4703 4704 4705
		} else {
			dev_err(DEV, "kmalloc(w) failed.");
			ov_oos_print(mdev);
			drbd_resync_finished(mdev);
		}
	}
4706
	put_ldev(mdev);
4707
	return true;
P
Philipp Reisner 已提交
4708 4709
}

4710
static int got_skip(struct drbd_conf *mdev, enum drbd_packet cmd)
4711
{
4712
	return true;
4713 4714
}

4715 4716
static int tconn_process_done_ee(struct drbd_tconn *tconn)
{
4717 4718
	struct drbd_conf *mdev;
	int i, not_empty = 0;
4719 4720 4721 4722

	do {
		clear_bit(SIGNAL_ASENDER, &tconn->flags);
		flush_signals(current);
4723
		idr_for_each_entry(&tconn->volumes, mdev, i) {
4724
			if (drbd_process_done_ee(mdev))
4725 4726
				return 1; /* error */
		}
4727
		set_bit(SIGNAL_ASENDER, &tconn->flags);
4728 4729 4730 4731 4732 4733 4734 4735

		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);
4736 4737 4738 4739 4740
	} while (not_empty);

	return 0;
}

4741 4742
struct asender_cmd {
	size_t pkt_size;
4743 4744 4745 4746 4747
	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);
	};
4748 4749 4750
};

static struct asender_cmd asender_tbl[] = {
4751 4752
	[P_PING]	    = { sizeof(struct p_header), CONN, { .conn_fn = got_Ping } },
	[P_PING_ACK]	    = { sizeof(struct p_header), CONN, { .conn_fn = got_PingAck } },
4753 4754 4755 4756 4757 4758 4759 4760 4761 4762 4763 4764 4765
	[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 } },
4766
	[P_CONN_ST_CHG_REPLY]={ sizeof(struct p_req_state_reply), CONN, {.conn_fn = got_conn_RqSReply}},
4767
	[P_RETRY_WRITE]	    = { sizeof(struct p_block_ack), MDEV, { got_BlockAck } },
4768 4769
};

P
Philipp Reisner 已提交
4770 4771
int drbd_asender(struct drbd_thread *thi)
{
4772
	struct drbd_tconn *tconn = thi->tconn;
4773
	struct p_header *h = &tconn->meta.rbuf.header;
P
Philipp Reisner 已提交
4774
	struct asender_cmd *cmd = NULL;
4775
	struct packet_info pi;
4776
	int rv;
P
Philipp Reisner 已提交
4777 4778
	void *buf    = h;
	int received = 0;
4779
	int expect   = sizeof(struct p_header);
4780
	int ping_timeout_active = 0;
P
Philipp Reisner 已提交
4781 4782 4783 4784

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

4785
	while (get_t_state(thi) == RUNNING) {
4786
		drbd_thread_current_set_cpu(thi);
4787
		if (test_and_clear_bit(SEND_PING, &tconn->flags)) {
4788
			if (!drbd_send_ping(tconn)) {
4789
				conn_err(tconn, "drbd_send_ping has failed\n");
4790 4791
				goto reconnect;
			}
4792 4793
			tconn->meta.socket->sk->sk_rcvtimeo =
				tconn->net_conf->ping_timeo*HZ/10;
4794
			ping_timeout_active = 1;
P
Philipp Reisner 已提交
4795 4796
		}

4797 4798 4799 4800
		/* 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);
4801 4802
		if (tconn_process_done_ee(tconn)) {
			conn_err(tconn, "tconn_process_done_ee() failed\n");
4803
			goto reconnect;
4804
		}
P
Philipp Reisner 已提交
4805
		/* but unconditionally uncork unless disabled */
4806 4807
		if (!tconn->net_conf->no_cork)
			drbd_tcp_uncork(tconn->meta.socket);
P
Philipp Reisner 已提交
4808 4809 4810 4811 4812

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

4813 4814
		rv = drbd_recv_short(tconn->meta.socket, buf, expect-received, 0);
		clear_bit(SIGNAL_ASENDER, &tconn->flags);
P
Philipp Reisner 已提交
4815 4816 4817 4818 4819 4820 4821 4822 4823 4824 4825 4826 4827 4828 4829 4830 4831

		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) {
4832
			conn_err(tconn, "meta connection shut down by peer.\n");
P
Philipp Reisner 已提交
4833 4834
			goto reconnect;
		} else if (rv == -EAGAIN) {
4835 4836
			/* If the data socket received something meanwhile,
			 * that is good enough: peer is still alive. */
4837 4838
			if (time_after(tconn->last_received,
				jiffies - tconn->meta.socket->sk->sk_rcvtimeo))
4839
				continue;
4840
			if (ping_timeout_active) {
4841
				conn_err(tconn, "PingAck did not arrive in time.\n");
P
Philipp Reisner 已提交
4842 4843
				goto reconnect;
			}
4844
			set_bit(SEND_PING, &tconn->flags);
P
Philipp Reisner 已提交
4845 4846 4847 4848
			continue;
		} else if (rv == -EINTR) {
			continue;
		} else {
4849
			conn_err(tconn, "sock_recvmsg returned %d\n", rv);
P
Philipp Reisner 已提交
4850 4851 4852 4853
			goto reconnect;
		}

		if (received == expect && cmd == NULL) {
4854
			if (decode_header(tconn, h, &pi))
P
Philipp Reisner 已提交
4855
				goto reconnect;
4856 4857
			cmd = &asender_tbl[pi.cmd];
			if (pi.cmd >= ARRAY_SIZE(asender_tbl) || !cmd) {
4858
				conn_err(tconn, "unknown command %d on meta (l: %d)\n",
4859
					pi.cmd, pi.size);
P
Philipp Reisner 已提交
4860 4861 4862
				goto disconnect;
			}
			expect = cmd->pkt_size;
4863
			if (pi.size != expect - sizeof(struct p_header)) {
4864
				conn_err(tconn, "Wrong packet size on meta (c: %d, l: %d)\n",
4865
					pi.cmd, pi.size);
P
Philipp Reisner 已提交
4866
				goto reconnect;
4867
			}
P
Philipp Reisner 已提交
4868 4869
		}
		if (received == expect) {
4870 4871 4872 4873 4874 4875 4876 4877 4878 4879
			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
4880 4881
				goto reconnect;

4882 4883
			tconn->last_received = jiffies;

4884 4885
			/* the idle_timeout (ping-int)
			 * has been restored in got_PingAck() */
4886
			if (cmd == &asender_tbl[P_PING_ACK])
4887 4888
				ping_timeout_active = 0;

P
Philipp Reisner 已提交
4889 4890
			buf	 = h;
			received = 0;
4891
			expect	 = sizeof(struct p_header);
P
Philipp Reisner 已提交
4892 4893 4894 4895 4896 4897
			cmd	 = NULL;
		}
	}

	if (0) {
reconnect:
4898
		conn_request_state(tconn, NS(conn, C_NETWORK_FAILURE), CS_HARD);
P
Philipp Reisner 已提交
4899 4900 4901
	}
	if (0) {
disconnect:
4902
		conn_request_state(tconn, NS(conn, C_DISCONNECTING), CS_HARD);
P
Philipp Reisner 已提交
4903
	}
4904
	clear_bit(SIGNAL_ASENDER, &tconn->flags);
P
Philipp Reisner 已提交
4905

4906
	conn_info(tconn, "asender terminated\n");
P
Philipp Reisner 已提交
4907 4908 4909

	return 0;
}