drbd_receiver.c 135.8 KB
Newer Older
P
Philipp Reisner 已提交
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50
/*
   drbd_receiver.c

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

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

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

   drbd is distributed in the hope that it will be useful,
   but WITHOUT ANY WARRANTY; without even the implied warranty of
   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
   GNU General Public License for more details.

   You should have received a copy of the GNU General Public License
   along with drbd; see the file COPYING.  If not, write to
   the Free Software Foundation, 675 Mass Ave, Cambridge, MA 02139, USA.
 */


#include <linux/module.h>

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

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

#include "drbd_vli.h"

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

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

63
static int drbd_do_features(struct drbd_tconn *tconn);
64
static int drbd_do_auth(struct drbd_tconn *tconn);
65
static int drbd_disconnected(int vnr, void *p, void *data);
P
Philipp Reisner 已提交
66 67

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


#define GFP_TRY	(__GFP_HIGHMEM | __GFP_NOWARN)

73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90
/*
 * 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;
91 92 93 94

	if (!page)
		return NULL;

95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152
	while (page) {
		tmp = page_chain_next(page);
		if (--n == 0)
			break; /* found sufficient pages */
		if (tmp == NULL)
			/* insufficient pages, don't use any of them. */
			return NULL;
		page = tmp;
	}

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

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

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

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

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

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

	/* Yes, testing drbd_pp_vacant outside the lock is racy.
	 * So what. It saves a spin_lock. */
160
	if (drbd_pp_vacant >= number) {
P
Philipp Reisner 已提交
161
		spin_lock(&drbd_pp_lock);
162 163 164
		page = page_chain_del(&drbd_pp_pool, number);
		if (page)
			drbd_pp_vacant -= number;
P
Philipp Reisner 已提交
165
		spin_unlock(&drbd_pp_lock);
166 167
		if (page)
			return page;
P
Philipp Reisner 已提交
168
	}
169

P
Philipp Reisner 已提交
170 171 172
	/* 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.  */
173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194
	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
195 196 197 198
}

static void reclaim_net_ee(struct drbd_conf *mdev, struct list_head *to_be_freed)
{
199
	struct drbd_peer_request *peer_req;
P
Philipp Reisner 已提交
200 201 202 203 204 205 206 207
	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) {
208 209
		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
210 211 212 213 214 215 216 217
			break;
		list_move(le, to_be_freed);
	}
}

static void drbd_kick_lo_and_reclaim_net(struct drbd_conf *mdev)
{
	LIST_HEAD(reclaimed);
218
	struct drbd_peer_request *peer_req, *t;
P
Philipp Reisner 已提交
219

220
	spin_lock_irq(&mdev->tconn->req_lock);
P
Philipp Reisner 已提交
221
	reclaim_net_ee(mdev, &reclaimed);
222
	spin_unlock_irq(&mdev->tconn->req_lock);
P
Philipp Reisner 已提交
223

224 225
	list_for_each_entry_safe(peer_req, t, &reclaimed, w.list)
		drbd_free_net_ee(mdev, peer_req);
P
Philipp Reisner 已提交
226 227 228
}

/**
229
 * drbd_pp_alloc() - Returns @number pages, retries forever (or until signalled)
P
Philipp Reisner 已提交
230
 * @mdev:	DRBD device.
231 232 233 234 235 236
 * @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
237
 *
238
 * Returns a page chain linked via page->private.
P
Philipp Reisner 已提交
239
 */
240
static struct page *drbd_pp_alloc(struct drbd_conf *mdev, unsigned number, bool retry)
P
Philipp Reisner 已提交
241 242 243 244
{
	struct page *page = NULL;
	DEFINE_WAIT(wait);

245 246
	/* Yes, we may run up to @number over max_buffers. If we
	 * follow it strictly, the admin will get it wrong anyways. */
247
	if (atomic_read(&mdev->pp_in_use) < mdev->tconn->net_conf->max_buffers)
248
		page = drbd_pp_first_pages_or_try_alloc(mdev, number);
P
Philipp Reisner 已提交
249

250
	while (page == NULL) {
P
Philipp Reisner 已提交
251 252 253 254
		prepare_to_wait(&drbd_pp_wait, &wait, TASK_INTERRUPTIBLE);

		drbd_kick_lo_and_reclaim_net(mdev);

255
		if (atomic_read(&mdev->pp_in_use) < mdev->tconn->net_conf->max_buffers) {
256
			page = drbd_pp_first_pages_or_try_alloc(mdev, number);
P
Philipp Reisner 已提交
257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272
			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);

273 274
	if (page)
		atomic_add(number, &mdev->pp_in_use);
P
Philipp Reisner 已提交
275 276 277 278
	return page;
}

/* Must not be used from irq, as that may deadlock: see drbd_pp_alloc.
279
 * Is also used from inside an other spin_lock_irq(&mdev->tconn->req_lock);
280 281
 * Either links the page chain back to the global pool,
 * or returns all pages to the system. */
282
static void drbd_pp_free(struct drbd_conf *mdev, struct page *page, int is_net)
P
Philipp Reisner 已提交
283
{
284
	atomic_t *a = is_net ? &mdev->pp_in_use_by_net : &mdev->pp_in_use;
P
Philipp Reisner 已提交
285
	int i;
286

287
	if (drbd_pp_vacant > (DRBD_MAX_BIO_SIZE/PAGE_SIZE) * minor_count)
288 289 290 291 292 293 294 295
		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
296
	}
297
	i = atomic_sub_return(i, a);
298
	if (i < 0)
299 300
		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
301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318
	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()
*/

319 320 321
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
322
{
323
	struct drbd_peer_request *peer_req;
P
Philipp Reisner 已提交
324
	struct page *page;
325
	unsigned nr_pages = (data_size + PAGE_SIZE -1) >> PAGE_SHIFT;
P
Philipp Reisner 已提交
326

327
	if (drbd_insert_fault(mdev, DRBD_FAULT_AL_EE))
P
Philipp Reisner 已提交
328 329
		return NULL;

330 331
	peer_req = mempool_alloc(drbd_ee_mempool, gfp_mask & ~__GFP_HIGHMEM);
	if (!peer_req) {
P
Philipp Reisner 已提交
332 333 334 335 336
		if (!(gfp_mask & __GFP_NOWARN))
			dev_err(DEV, "alloc_ee: Allocation of an EE failed\n");
		return NULL;
	}

337 338 339
	page = drbd_pp_alloc(mdev, nr_pages, (gfp_mask & __GFP_WAIT));
	if (!page)
		goto fail;
P
Philipp Reisner 已提交
340

341 342 343 344 345 346 347
	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;
348
	peer_req->w.mdev = mdev;
349 350 351
	peer_req->pages = page;
	atomic_set(&peer_req->pending_bios, 0);
	peer_req->flags = 0;
352 353 354 355
	/*
	 * The block_id is opaque to the receiver.  It is not endianness
	 * converted, and sent back to the sender unchanged.
	 */
356
	peer_req->block_id = id;
P
Philipp Reisner 已提交
357

358
	return peer_req;
P
Philipp Reisner 已提交
359

360
 fail:
361
	mempool_free(peer_req, drbd_ee_mempool);
P
Philipp Reisner 已提交
362 363 364
	return NULL;
}

365
void drbd_free_some_ee(struct drbd_conf *mdev, struct drbd_peer_request *peer_req,
366
		       int is_net)
P
Philipp Reisner 已提交
367
{
368 369 370 371 372 373
	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
374 375 376 377 378
}

int drbd_release_ee(struct drbd_conf *mdev, struct list_head *list)
{
	LIST_HEAD(work_list);
379
	struct drbd_peer_request *peer_req, *t;
P
Philipp Reisner 已提交
380
	int count = 0;
381
	int is_net = list == &mdev->net_ee;
P
Philipp Reisner 已提交
382

383
	spin_lock_irq(&mdev->tconn->req_lock);
P
Philipp Reisner 已提交
384
	list_splice_init(list, &work_list);
385
	spin_unlock_irq(&mdev->tconn->req_lock);
P
Philipp Reisner 已提交
386

387 388
	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
389 390 391 392 393 394
		count++;
	}
	return count;
}


395
/* See also comments in _req_mod(,BARRIER_ACKED)
P
Philipp Reisner 已提交
396 397 398 399 400 401 402 403 404 405
 * 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);
406
	struct drbd_peer_request *peer_req, *t;
407
	int err = 0;
P
Philipp Reisner 已提交
408

409
	spin_lock_irq(&mdev->tconn->req_lock);
P
Philipp Reisner 已提交
410 411
	reclaim_net_ee(mdev, &reclaimed);
	list_splice_init(&mdev->done_ee, &work_list);
412
	spin_unlock_irq(&mdev->tconn->req_lock);
P
Philipp Reisner 已提交
413

414 415
	list_for_each_entry_safe(peer_req, t, &reclaimed, w.list)
		drbd_free_net_ee(mdev, peer_req);
P
Philipp Reisner 已提交
416 417

	/* possible callbacks here:
418
	 * e_end_block, and e_end_resync_block, e_send_discard_write.
P
Philipp Reisner 已提交
419 420
	 * all ignore the last argument.
	 */
421
	list_for_each_entry_safe(peer_req, t, &work_list, w.list) {
422 423
		int err2;

P
Philipp Reisner 已提交
424
		/* list_del not necessary, next/prev members not touched */
425 426 427
		err2 = peer_req->w.cb(&peer_req->w, !!err);
		if (!err)
			err = err2;
428
		drbd_free_ee(mdev, peer_req);
P
Philipp Reisner 已提交
429 430 431
	}
	wake_up(&mdev->ee_wait);

432
	return err;
P
Philipp Reisner 已提交
433 434 435 436 437 438 439 440 441 442
}

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);
443
		spin_unlock_irq(&mdev->tconn->req_lock);
J
Jens Axboe 已提交
444
		io_schedule();
P
Philipp Reisner 已提交
445
		finish_wait(&mdev->ee_wait, &wait);
446
		spin_lock_irq(&mdev->tconn->req_lock);
P
Philipp Reisner 已提交
447 448 449 450 451
	}
}

void drbd_wait_ee_list_empty(struct drbd_conf *mdev, struct list_head *head)
{
452
	spin_lock_irq(&mdev->tconn->req_lock);
P
Philipp Reisner 已提交
453
	_drbd_wait_ee_list_empty(mdev, head);
454
	spin_unlock_irq(&mdev->tconn->req_lock);
P
Philipp Reisner 已提交
455 456 457 458
}

/* see also kernel_accept; which is only present since 2.6.18.
 * also we want to log which part of it failed, exactly */
459
static int drbd_accept(const char **what, struct socket *sock, struct socket **newsock)
P
Philipp Reisner 已提交
460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487
{
	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;
}

488
static int drbd_recv_short(struct socket *sock, void *buf, size_t size, int flags)
P
Philipp Reisner 已提交
489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509
{
	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;
}

510
static int drbd_recv(struct drbd_tconn *tconn, void *buf, size_t size)
P
Philipp Reisner 已提交
511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527
{
	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 (;;) {
528
		rv = sock_recvmsg(tconn->data.socket, &msg, size, msg.msg_flags);
P
Philipp Reisner 已提交
529 530 531 532 533 534 535 536 537 538
		if (rv == size)
			break;

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

		if (rv < 0) {
			if (rv == -ECONNRESET)
539
				conn_info(tconn, "sock was reset by peer\n");
P
Philipp Reisner 已提交
540
			else if (rv != -ERESTARTSYS)
541
				conn_err(tconn, "sock_recvmsg returned %d\n", rv);
P
Philipp Reisner 已提交
542 543
			break;
		} else if (rv == 0) {
544
			conn_info(tconn, "sock was shut down by peer\n");
P
Philipp Reisner 已提交
545 546 547 548 549 550 551 552 553 554 555 556 557
			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)
558
		conn_request_state(tconn, NS(conn, C_BROKEN_PIPE), CS_HARD);
P
Philipp Reisner 已提交
559 560 561 562

	return rv;
}

563 564 565 566 567 568 569 570 571 572 573 574 575
static int drbd_recv_all(struct drbd_tconn *tconn, void *buf, size_t size)
{
	int err;

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

576 577 578 579 580 581 582 583 584 585
static int drbd_recv_all_warn(struct drbd_tconn *tconn, void *buf, size_t size)
{
	int err;

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

586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604
/* 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;
	}
}

605
static struct socket *drbd_try_connect(struct drbd_tconn *tconn)
P
Philipp Reisner 已提交
606 607 608 609 610 611 612
{
	const char *what;
	struct socket *sock;
	struct sockaddr_in6 src_in6;
	int err;
	int disconnect_on_error = 1;

613
	if (!get_net_conf(tconn))
P
Philipp Reisner 已提交
614 615 616
		return NULL;

	what = "sock_create_kern";
617
	err = sock_create_kern(((struct sockaddr *)tconn->net_conf->my_addr)->sa_family,
P
Philipp Reisner 已提交
618 619 620 621 622 623 624
		SOCK_STREAM, IPPROTO_TCP, &sock);
	if (err < 0) {
		sock = NULL;
		goto out;
	}

	sock->sk->sk_rcvtimeo =
625 626 627
	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
628 629 630 631 632 633 634 635

       /* 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.
	*/
636 637 638
	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
639 640 641 642 643 644 645
		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,
646
			      tconn->net_conf->my_addr_len);
P
Philipp Reisner 已提交
647 648 649 650 651 652 653 654
	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,
655 656
				 (struct sockaddr *)tconn->net_conf->peer_addr,
				 tconn->net_conf->peer_addr_len, 0);
P
Philipp Reisner 已提交
657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673

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:
674
			conn_err(tconn, "%s failed, err = %d\n", what, err);
P
Philipp Reisner 已提交
675 676
		}
		if (disconnect_on_error)
677
			conn_request_state(tconn, NS(conn, C_DISCONNECTING), CS_HARD);
P
Philipp Reisner 已提交
678
	}
679
	put_net_conf(tconn);
P
Philipp Reisner 已提交
680 681 682
	return sock;
}

683
static struct socket *drbd_wait_for_connect(struct drbd_tconn *tconn)
P
Philipp Reisner 已提交
684 685 686 687 688
{
	int timeo, err;
	struct socket *s_estab = NULL, *s_listen;
	const char *what;

689
	if (!get_net_conf(tconn))
P
Philipp Reisner 已提交
690 691 692
		return NULL;

	what = "sock_create_kern";
693
	err = sock_create_kern(((struct sockaddr *)tconn->net_conf->my_addr)->sa_family,
P
Philipp Reisner 已提交
694 695 696 697 698 699
		SOCK_STREAM, IPPROTO_TCP, &s_listen);
	if (err) {
		s_listen = NULL;
		goto out;
	}

700
	timeo = tconn->net_conf->try_connect_int * HZ;
P
Philipp Reisner 已提交
701 702 703 704 705
	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;
706 707
	drbd_setbufsize(s_listen, tconn->net_conf->sndbuf_size,
			tconn->net_conf->rcvbuf_size);
P
Philipp Reisner 已提交
708 709 710

	what = "bind before listen";
	err = s_listen->ops->bind(s_listen,
711 712
			      (struct sockaddr *) tconn->net_conf->my_addr,
			      tconn->net_conf->my_addr_len);
P
Philipp Reisner 已提交
713 714 715
	if (err < 0)
		goto out;

716
	err = drbd_accept(&what, s_listen, &s_estab);
P
Philipp Reisner 已提交
717 718 719 720 721 722

out:
	if (s_listen)
		sock_release(s_listen);
	if (err < 0) {
		if (err != -EAGAIN && err != -EINTR && err != -ERESTARTSYS) {
723
			conn_err(tconn, "%s failed, err = %d\n", what, err);
724
			conn_request_state(tconn, NS(conn, C_DISCONNECTING), CS_HARD);
P
Philipp Reisner 已提交
725 726
		}
	}
727
	put_net_conf(tconn);
P
Philipp Reisner 已提交
728 729 730 731

	return s_estab;
}

732
static int decode_header(struct drbd_tconn *, struct p_header *, struct packet_info *);
P
Philipp Reisner 已提交
733

734 735 736 737 738 739
static int send_first_packet(struct drbd_tconn *tconn, struct drbd_socket *sock,
			     enum drbd_packet cmd)
{
	if (!conn_prepare_command(tconn, sock))
		return -EIO;
	return conn_send_command(tconn, sock, cmd, sizeof(struct p_header), NULL, 0);
P
Philipp Reisner 已提交
740 741
}

742
static int receive_first_packet(struct drbd_tconn *tconn, struct socket *sock)
P
Philipp Reisner 已提交
743
{
744 745 746
	unsigned int header_size = drbd_header_size(tconn);
	struct packet_info pi;
	int err;
P
Philipp Reisner 已提交
747

748 749 750 751 752 753 754 755 756 757
	err = drbd_recv_short(sock, tconn->data.rbuf, header_size, 0);
	if (err != header_size) {
		if (err >= 0)
			err = -EIO;
		return err;
	}
	err = decode_header(tconn, tconn->data.rbuf, &pi);
	if (err)
		return err;
	return pi.cmd;
P
Philipp Reisner 已提交
758 759 760 761 762 763
}

/**
 * drbd_socket_okay() - Free the socket if its connection is not okay
 * @sock:	pointer to the pointer to the socket.
 */
764
static int drbd_socket_okay(struct socket **sock)
P
Philipp Reisner 已提交
765 766 767 768 769
{
	int rr;
	char tb[4];

	if (!*sock)
770
		return false;
P
Philipp Reisner 已提交
771

772
	rr = drbd_recv_short(*sock, tb, 4, MSG_DONTWAIT | MSG_PEEK);
P
Philipp Reisner 已提交
773 774

	if (rr > 0 || rr == -EAGAIN) {
775
		return true;
P
Philipp Reisner 已提交
776 777 778
	} else {
		sock_release(*sock);
		*sock = NULL;
779
		return false;
P
Philipp Reisner 已提交
780 781
	}
}
782 783 784
/* 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)
785 786
{
	struct drbd_conf *mdev = (struct drbd_conf *)p;
787
	int err;
788 789 790 791

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

792 793 794 795
	mdev->state_mutex = mdev->tconn->agreed_pro_version < 100 ?
		&mdev->tconn->cstate_mutex :
		&mdev->own_state_mutex;

796 797 798 799 800 801 802
	err = drbd_send_sync_param(mdev);
	if (!err)
		err = drbd_send_sizes(mdev, 0, 0);
	if (!err)
		err = drbd_send_uuids(mdev);
	if (!err)
		err = drbd_send_state(mdev);
803 804
	clear_bit(USE_DEGR_WFC_T, &mdev->flags);
	clear_bit(RESIZE_PENDING, &mdev->flags);
805
	mod_timer(&mdev->request_timer, jiffies + HZ); /* just start it here. */
806
	return err;
807 808
}

P
Philipp Reisner 已提交
809 810 811 812 813 814 815 816
/*
 * 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...
 */
817
static int drbd_connect(struct drbd_tconn *tconn)
P
Philipp Reisner 已提交
818
{
819
	struct socket *sock, *msock;
P
Philipp Reisner 已提交
820 821
	int try, h, ok;

822
	if (conn_request_state(tconn, NS(conn, C_WF_CONNECTION), CS_VERBOSE) < SS_SUCCESS)
P
Philipp Reisner 已提交
823 824
		return -2;

825
	clear_bit(DISCARD_CONCURRENT, &tconn->flags);
826 827 828

	/* Assume that the peer only understands protocol 80 until we know better.  */
	tconn->agreed_pro_version = 80;
P
Philipp Reisner 已提交
829 830

	do {
831 832
		struct socket *s;

P
Philipp Reisner 已提交
833 834
		for (try = 0;;) {
			/* 3 tries, this should take less than a second! */
835
			s = drbd_try_connect(tconn);
P
Philipp Reisner 已提交
836 837 838
			if (s || ++try >= 3)
				break;
			/* give the other side time to call bind() & listen() */
839
			schedule_timeout_interruptible(HZ / 10);
P
Philipp Reisner 已提交
840 841 842
		}

		if (s) {
843 844
			if (!tconn->data.socket) {
				tconn->data.socket = s;
845
				send_first_packet(tconn, &tconn->data, P_INITIAL_DATA);
846 847
			} else if (!tconn->meta.socket) {
				tconn->meta.socket = s;
848
				send_first_packet(tconn, &tconn->meta, P_INITIAL_META);
P
Philipp Reisner 已提交
849
			} else {
850
				conn_err(tconn, "Logic error in drbd_connect()\n");
P
Philipp Reisner 已提交
851 852 853 854
				goto out_release_sockets;
			}
		}

855
		if (tconn->data.socket && tconn->meta.socket) {
856
			schedule_timeout_interruptible(tconn->net_conf->ping_timeo*HZ/10);
857 858
			ok = drbd_socket_okay(&tconn->data.socket);
			ok = drbd_socket_okay(&tconn->meta.socket) && ok;
P
Philipp Reisner 已提交
859 860 861 862 863
			if (ok)
				break;
		}

retry:
864
		s = drbd_wait_for_connect(tconn);
P
Philipp Reisner 已提交
865
		if (s) {
866
			try = receive_first_packet(tconn, s);
867 868
			drbd_socket_okay(&tconn->data.socket);
			drbd_socket_okay(&tconn->meta.socket);
P
Philipp Reisner 已提交
869
			switch (try) {
870
			case P_INITIAL_DATA:
871
				if (tconn->data.socket) {
872
					conn_warn(tconn, "initial packet S crossed\n");
873
					sock_release(tconn->data.socket);
P
Philipp Reisner 已提交
874
				}
875
				tconn->data.socket = s;
P
Philipp Reisner 已提交
876
				break;
877
			case P_INITIAL_META:
878
				if (tconn->meta.socket) {
879
					conn_warn(tconn, "initial packet M crossed\n");
880
					sock_release(tconn->meta.socket);
P
Philipp Reisner 已提交
881
				}
882
				tconn->meta.socket = s;
883
				set_bit(DISCARD_CONCURRENT, &tconn->flags);
P
Philipp Reisner 已提交
884 885
				break;
			default:
886
				conn_warn(tconn, "Error receiving initial packet\n");
P
Philipp Reisner 已提交
887 888 889 890 891 892
				sock_release(s);
				if (random32() & 1)
					goto retry;
			}
		}

893
		if (tconn->cstate <= C_DISCONNECTING)
P
Philipp Reisner 已提交
894 895 896 897
			goto out_release_sockets;
		if (signal_pending(current)) {
			flush_signals(current);
			smp_rmb();
898
			if (get_t_state(&tconn->receiver) == EXITING)
P
Philipp Reisner 已提交
899 900 901
				goto out_release_sockets;
		}

902 903 904
		if (tconn->data.socket && &tconn->meta.socket) {
			ok = drbd_socket_okay(&tconn->data.socket);
			ok = drbd_socket_okay(&tconn->meta.socket) && ok;
P
Philipp Reisner 已提交
905 906 907 908 909
			if (ok)
				break;
		}
	} while (1);

910 911 912
	sock  = tconn->data.socket;
	msock = tconn->meta.socket;

P
Philipp Reisner 已提交
913 914 915 916 917 918 919 920 921 922
	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 ...
923
	 * sock->sk->sk_sndtimeo = tconn->net_conf->timeout*HZ/10;
P
Philipp Reisner 已提交
924
	 * sock->sk->sk_rcvtimeo = MAX_SCHEDULE_TIMEOUT;
925
	 * first set it to the P_CONNECTION_FEATURES timeout,
P
Philipp Reisner 已提交
926 927
	 * which we set to 4x the configured ping_timeout. */
	sock->sk->sk_sndtimeo =
928
	sock->sk->sk_rcvtimeo = tconn->net_conf->ping_timeo*4*HZ/10;
P
Philipp Reisner 已提交
929

930 931
	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
932 933

	/* we don't want delays.
L
Lucas De Marchi 已提交
934
	 * we use TCP_CORK where appropriate, though */
P
Philipp Reisner 已提交
935 936 937
	drbd_tcp_nodelay(sock);
	drbd_tcp_nodelay(msock);

938
	tconn->last_received = jiffies;
P
Philipp Reisner 已提交
939

940
	h = drbd_do_features(tconn);
P
Philipp Reisner 已提交
941 942 943
	if (h <= 0)
		return h;

944
	if (tconn->cram_hmac_tfm) {
P
Philipp Reisner 已提交
945
		/* drbd_request_state(mdev, NS(conn, WFAuth)); */
946
		switch (drbd_do_auth(tconn)) {
947
		case -1:
948
			conn_err(tconn, "Authentication of peer failed\n");
P
Philipp Reisner 已提交
949
			return -1;
950
		case 0:
951
			conn_err(tconn, "Authentication of peer failed, trying again.\n");
952
			return 0;
P
Philipp Reisner 已提交
953 954 955
		}
	}

956
	if (conn_request_state(tconn, NS(conn, C_WF_REPORT_PARAMS), CS_VERBOSE) < SS_SUCCESS)
P
Philipp Reisner 已提交
957 958
		return 0;

959
	sock->sk->sk_sndtimeo = tconn->net_conf->timeout*HZ/10;
P
Philipp Reisner 已提交
960 961
	sock->sk->sk_rcvtimeo = MAX_SCHEDULE_TIMEOUT;

962
	drbd_thread_start(&tconn->asender);
P
Philipp Reisner 已提交
963

964
	if (drbd_send_protocol(tconn) == -EOPNOTSUPP)
965
		return -1;
P
Philipp Reisner 已提交
966

967
	return !idr_for_each(&tconn->volumes, drbd_connected, tconn);
P
Philipp Reisner 已提交
968 969

out_release_sockets:
970 971 972 973 974 975 976 977
	if (tconn->data.socket) {
		sock_release(tconn->data.socket);
		tconn->data.socket = NULL;
	}
	if (tconn->meta.socket) {
		sock_release(tconn->meta.socket);
		tconn->meta.socket = NULL;
	}
P
Philipp Reisner 已提交
978 979 980
	return -1;
}

981
static int decode_header(struct drbd_tconn *tconn, struct p_header *h, struct packet_info *pi)
P
Philipp Reisner 已提交
982
{
P
Philipp Reisner 已提交
983
	if (h->h80.magic == cpu_to_be32(DRBD_MAGIC)) {
984 985
		pi->cmd = be16_to_cpu(h->h80.command);
		pi->size = be16_to_cpu(h->h80.length);
986
		pi->vnr = 0;
987
	} else if (h->h95.magic == cpu_to_be16(DRBD_MAGIC_BIG)) {
988 989 990
		pi->cmd = be16_to_cpu(h->h95.command);
		pi->size = be32_to_cpu(h->h95.length) & 0x00ffffff;
		pi->vnr = 0;
991
	} else {
992
		conn_err(tconn, "magic?? on data m: 0x%08x c: %d l: %d\n",
993 994 995
		    be32_to_cpu(h->h80.magic),
		    be16_to_cpu(h->h80.command),
		    be16_to_cpu(h->h80.length));
996
		return -EINVAL;
P
Philipp Reisner 已提交
997
	}
998
	return 0;
999 1000
}

1001
static int drbd_recv_header(struct drbd_tconn *tconn, struct packet_info *pi)
1002
{
1003
	struct p_header *h = tconn->data.rbuf;
1004
	int err;
1005

1006
	err = drbd_recv_all_warn(tconn, h, drbd_header_size(tconn));
1007
	if (err)
1008
		return err;
1009

1010
	err = decode_header(tconn, h, pi);
1011
	tconn->last_received = jiffies;
P
Philipp Reisner 已提交
1012

1013
	return err;
P
Philipp Reisner 已提交
1014 1015
}

1016
static void drbd_flush(struct drbd_conf *mdev)
P
Philipp Reisner 已提交
1017 1018 1019 1020
{
	int rv;

	if (mdev->write_ordering >= WO_bdev_flush && get_ldev(mdev)) {
1021
		rv = blkdev_issue_flush(mdev->ldev->backing_bdev, GFP_KERNEL,
1022
					NULL);
P
Philipp Reisner 已提交
1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043
		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)
{
1044
	int epoch_size;
P
Philipp Reisner 已提交
1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067
	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 &&
1068
		    test_bit(DE_HAVE_BARRIER_NUMBER, &epoch->flags)) {
P
Philipp Reisner 已提交
1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087
			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);
1088
				/* atomic_set(&epoch->active, 0); is already zero */
P
Philipp Reisner 已提交
1089 1090
				if (rv == FE_STILL_LIVE)
					rv = FE_RECYCLED;
1091
				wake_up(&mdev->ee_wait);
P
Philipp Reisner 已提交
1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126
			}
		}

		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;
1127
	if (pwo != mdev->write_ordering || wo == WO_bdev_flush)
P
Philipp Reisner 已提交
1128 1129 1130
		dev_info(DEV, "Method to ensure write ordering: %s\n", write_ordering_str[mdev->write_ordering]);
}

1131
/**
1132
 * drbd_submit_peer_request()
1133
 * @mdev:	DRBD device.
1134
 * @peer_req:	peer request
1135
 * @rw:		flag field, see bio->bi_rw
1136 1137 1138 1139 1140 1141 1142 1143 1144 1145
 *
 * 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.
1146 1147
 */
/* TODO allocate from our own bio_set. */
1148 1149 1150
int drbd_submit_peer_request(struct drbd_conf *mdev,
			     struct drbd_peer_request *peer_req,
			     const unsigned rw, const int fault_type)
1151 1152 1153
{
	struct bio *bios = NULL;
	struct bio *bio;
1154 1155 1156
	struct page *page = peer_req->pages;
	sector_t sector = peer_req->i.sector;
	unsigned ds = peer_req->i.size;
1157 1158
	unsigned n_bios = 0;
	unsigned nr_pages = (ds + PAGE_SIZE -1) >> PAGE_SHIFT;
1159
	int err = -ENOMEM;
1160 1161 1162 1163

	/* 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
1164 1165 1166 1167 1168
	 * 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.
	 */
1169 1170 1171 1172 1173 1174
next_bio:
	bio = bio_alloc(GFP_NOIO, nr_pages);
	if (!bio) {
		dev_err(DEV, "submit_ee: Allocation of a bio failed\n");
		goto fail;
	}
1175
	/* > peer_req->i.sector, unless this is the first bio */
1176 1177 1178
	bio->bi_sector = sector;
	bio->bi_bdev = mdev->ldev->backing_bdev;
	bio->bi_rw = rw;
1179
	bio->bi_private = peer_req;
1180
	bio->bi_end_io = drbd_peer_request_endio;
1181 1182 1183 1184 1185 1186 1187 1188

	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)) {
1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199
			/* 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;
			}
1200 1201 1202 1203 1204 1205 1206 1207 1208
			goto next_bio;
		}
		ds -= len;
		sector += len >> 9;
		--nr_pages;
	}
	D_ASSERT(page == NULL);
	D_ASSERT(ds == 0);

1209
	atomic_set(&peer_req->pending_bios, n_bios);
1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224
	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);
	}
1225
	return err;
1226 1227
}

1228
static void drbd_remove_epoch_entry_interval(struct drbd_conf *mdev,
1229
					     struct drbd_peer_request *peer_req)
1230
{
1231
	struct drbd_interval *i = &peer_req->i;
1232 1233 1234 1235

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

A
Andreas Gruenbacher 已提交
1236
	/* Wake up any processes waiting for this peer request to complete.  */
1237 1238 1239 1240
	if (i->waiting)
		wake_up(&mdev->misc_wait);
}

1241
static int receive_Barrier(struct drbd_tconn *tconn, struct packet_info *pi)
P
Philipp Reisner 已提交
1242
{
1243
	struct drbd_conf *mdev;
1244
	int rv;
1245
	struct p_barrier *p = tconn->data.rbuf;
P
Philipp Reisner 已提交
1246 1247
	struct drbd_epoch *epoch;

1248 1249 1250 1251
	mdev = vnr_to_mdev(tconn, pi->vnr);
	if (!mdev)
		return -EIO;

P
Philipp Reisner 已提交
1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264
	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)
1265
			return 0;
1266 1267 1268 1269 1270 1271 1272 1273 1274

		/* 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
1275 1276 1277 1278

	case WO_bdev_flush:
	case WO_drain_io:
		drbd_wait_ee_list_empty(mdev, &mdev->active_ee);
1279 1280 1281 1282 1283 1284
		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
1285 1286
		}

1287 1288 1289 1290 1291
		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
1292

1293
		return 0;
1294 1295
	default:
		dev_err(DEV, "Strangeness in mdev->write_ordering %d\n", mdev->write_ordering);
1296
		return -EIO;
P
Philipp Reisner 已提交
1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313
	}

	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);

1314
	return 0;
P
Philipp Reisner 已提交
1315 1316 1317 1318
}

/* used from receive_RSDataReply (recv_resync_read)
 * and from receive_Data */
1319 1320 1321
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
1322
{
1323
	const sector_t capacity = drbd_get_capacity(mdev->this_bdev);
1324
	struct drbd_peer_request *peer_req;
P
Philipp Reisner 已提交
1325
	struct page *page;
1326
	int dgs, ds, err;
1327 1328
	void *dig_in = mdev->tconn->int_dig_in;
	void *dig_vv = mdev->tconn->int_dig_vv;
1329
	unsigned long *data;
P
Philipp Reisner 已提交
1330

1331 1332
	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
1333 1334

	if (dgs) {
1335 1336 1337 1338
		/*
		 * FIXME: Receive the incoming digest into the receive buffer
		 *	  here, together with its struct p_data?
		 */
1339 1340
		err = drbd_recv_all_warn(mdev->tconn, dig_in, dgs);
		if (err)
P
Philipp Reisner 已提交
1341 1342 1343 1344 1345
			return NULL;
	}

	data_size -= dgs;

1346 1347 1348 1349 1350 1351
	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
1352

1353 1354 1355
	/* even though we trust out peer,
	 * we sometimes have to double check. */
	if (sector + (data_size>>9) > capacity) {
1356 1357
		dev_err(DEV, "request from peer beyond end of local disk: "
			"capacity: %llus < sector: %llus + size: %u\n",
1358 1359 1360 1361 1362
			(unsigned long long)capacity,
			(unsigned long long)sector, data_size);
		return NULL;
	}

P
Philipp Reisner 已提交
1363 1364 1365
	/* 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.  */
1366 1367
	peer_req = drbd_alloc_ee(mdev, id, sector, data_size, GFP_NOIO);
	if (!peer_req)
P
Philipp Reisner 已提交
1368
		return NULL;
1369

P
Philipp Reisner 已提交
1370
	ds = data_size;
1371
	page = peer_req->pages;
1372 1373
	page_chain_for_each(page) {
		unsigned len = min_t(int, ds, PAGE_SIZE);
1374
		data = kmap(page);
1375
		err = drbd_recv_all_warn(mdev->tconn, data, len);
1376
		if (drbd_insert_fault(mdev, DRBD_FAULT_RECEIVE)) {
1377 1378 1379
			dev_err(DEV, "Fault injection: Corrupting data on receive\n");
			data[0] = data[0] ^ (unsigned long)-1;
		}
P
Philipp Reisner 已提交
1380
		kunmap(page);
1381
		if (err) {
1382
			drbd_free_ee(mdev, peer_req);
P
Philipp Reisner 已提交
1383 1384
			return NULL;
		}
1385
		ds -= len;
P
Philipp Reisner 已提交
1386 1387 1388
	}

	if (dgs) {
1389
		drbd_csum_ee(mdev, mdev->tconn->integrity_r_tfm, peer_req, dig_vv);
P
Philipp Reisner 已提交
1390
		if (memcmp(dig_in, dig_vv, dgs)) {
1391 1392
			dev_err(DEV, "Digest integrity check FAILED: %llus +%u\n",
				(unsigned long long)sector, data_size);
1393
			drbd_free_ee(mdev, peer_req);
P
Philipp Reisner 已提交
1394 1395 1396 1397
			return NULL;
		}
	}
	mdev->recv_cnt += data_size>>9;
1398
	return peer_req;
P
Philipp Reisner 已提交
1399 1400 1401 1402 1403 1404 1405 1406
}

/* 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;
1407
	int err = 0;
P
Philipp Reisner 已提交
1408 1409
	void *data;

1410
	if (!data_size)
1411
		return 0;
1412

1413
	page = drbd_pp_alloc(mdev, 1, 1);
P
Philipp Reisner 已提交
1414 1415 1416

	data = kmap(page);
	while (data_size) {
1417 1418
		unsigned int len = min_t(int, data_size, PAGE_SIZE);

1419 1420
		err = drbd_recv_all_warn(mdev->tconn, data, len);
		if (err)
P
Philipp Reisner 已提交
1421
			break;
1422
		data_size -= len;
P
Philipp Reisner 已提交
1423 1424
	}
	kunmap(page);
1425
	drbd_pp_free(mdev, page, 0);
1426
	return err;
P
Philipp Reisner 已提交
1427 1428 1429 1430 1431 1432 1433
}

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;
1434
	int dgs, err, i, expect;
1435 1436
	void *dig_in = mdev->tconn->int_dig_in;
	void *dig_vv = mdev->tconn->int_dig_vv;
P
Philipp Reisner 已提交
1437

1438 1439
	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
1440 1441

	if (dgs) {
1442 1443 1444
		err = drbd_recv_all_warn(mdev->tconn, dig_in, dgs);
		if (err)
			return err;
P
Philipp Reisner 已提交
1445 1446 1447 1448 1449 1450 1451 1452 1453 1454 1455 1456
	}

	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) {
1457
		void *mapped = kmap(bvec->bv_page) + bvec->bv_offset;
P
Philipp Reisner 已提交
1458
		expect = min_t(int, data_size, bvec->bv_len);
1459
		err = drbd_recv_all_warn(mdev->tconn, mapped, expect);
P
Philipp Reisner 已提交
1460
		kunmap(bvec->bv_page);
1461 1462 1463
		if (err)
			return err;
		data_size -= expect;
P
Philipp Reisner 已提交
1464 1465 1466
	}

	if (dgs) {
1467
		drbd_csum_bio(mdev, mdev->tconn->integrity_r_tfm, bio, dig_vv);
P
Philipp Reisner 已提交
1468 1469
		if (memcmp(dig_in, dig_vv, dgs)) {
			dev_err(DEV, "Digest integrity check FAILED. Broken NICs?\n");
1470
			return -EINVAL;
P
Philipp Reisner 已提交
1471 1472 1473 1474
		}
	}

	D_ASSERT(data_size == 0);
1475
	return 0;
P
Philipp Reisner 已提交
1476 1477 1478 1479
}

/* e_end_resync_block() is called via
 * drbd_process_done_ee() by asender only */
1480
static int e_end_resync_block(struct drbd_work *w, int unused)
P
Philipp Reisner 已提交
1481
{
1482 1483
	struct drbd_peer_request *peer_req =
		container_of(w, struct drbd_peer_request, w);
1484
	struct drbd_conf *mdev = w->mdev;
1485
	sector_t sector = peer_req->i.sector;
1486
	int err;
P
Philipp Reisner 已提交
1487

1488
	D_ASSERT(drbd_interval_empty(&peer_req->i));
P
Philipp Reisner 已提交
1489

1490 1491
	if (likely((peer_req->flags & EE_WAS_ERROR) == 0)) {
		drbd_set_in_sync(mdev, sector, peer_req->i.size);
1492
		err = drbd_send_ack(mdev, P_RS_WRITE_ACK, peer_req);
P
Philipp Reisner 已提交
1493 1494
	} else {
		/* Record failure to sync */
1495
		drbd_rs_failed_io(mdev, sector, peer_req->i.size);
P
Philipp Reisner 已提交
1496

1497
		err  = drbd_send_ack(mdev, P_NEG_ACK, peer_req);
P
Philipp Reisner 已提交
1498 1499 1500
	}
	dec_unacked(mdev);

1501
	return err;
P
Philipp Reisner 已提交
1502 1503 1504 1505
}

static int recv_resync_read(struct drbd_conf *mdev, sector_t sector, int data_size) __releases(local)
{
1506
	struct drbd_peer_request *peer_req;
P
Philipp Reisner 已提交
1507

1508 1509
	peer_req = read_in_block(mdev, ID_SYNCER, sector, data_size);
	if (!peer_req)
1510
		goto fail;
P
Philipp Reisner 已提交
1511 1512 1513 1514 1515 1516 1517

	dec_rs_pending(mdev);

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

1518
	peer_req->w.cb = e_end_resync_block;
1519

1520
	spin_lock_irq(&mdev->tconn->req_lock);
1521
	list_add(&peer_req->w.list, &mdev->sync_ee);
1522
	spin_unlock_irq(&mdev->tconn->req_lock);
P
Philipp Reisner 已提交
1523

1524
	atomic_add(data_size >> 9, &mdev->rs_sect_ev);
1525
	if (drbd_submit_peer_request(mdev, peer_req, WRITE, DRBD_FAULT_RS_WR) == 0)
1526
		return 0;
P
Philipp Reisner 已提交
1527

1528 1529
	/* don't care for the reason here */
	dev_err(DEV, "submit failed, triggering re-connect\n");
1530
	spin_lock_irq(&mdev->tconn->req_lock);
1531
	list_del(&peer_req->w.list);
1532
	spin_unlock_irq(&mdev->tconn->req_lock);
1533

1534
	drbd_free_ee(mdev, peer_req);
1535 1536
fail:
	put_ldev(mdev);
1537
	return -EIO;
P
Philipp Reisner 已提交
1538 1539
}

1540
static struct drbd_request *
1541 1542
find_request(struct drbd_conf *mdev, struct rb_root *root, u64 id,
	     sector_t sector, bool missing_ok, const char *func)
1543 1544 1545
{
	struct drbd_request *req;

1546 1547
	/* Request object according to our peer */
	req = (struct drbd_request *)(unsigned long)id;
1548
	if (drbd_contains_interval(root, sector, &req->i) && req->i.local)
1549
		return req;
1550 1551 1552 1553
	if (!missing_ok) {
		dev_err(DEV, "%s: failed to find request %lu, sector %llus\n", func,
			(unsigned long)id, (unsigned long long)sector);
	}
1554 1555 1556
	return NULL;
}

1557
static int receive_DataReply(struct drbd_tconn *tconn, struct packet_info *pi)
P
Philipp Reisner 已提交
1558
{
1559
	struct drbd_conf *mdev;
P
Philipp Reisner 已提交
1560 1561
	struct drbd_request *req;
	sector_t sector;
1562
	int err;
1563 1564 1565 1566 1567
	struct p_data *p = tconn->data.rbuf;

	mdev = vnr_to_mdev(tconn, pi->vnr);
	if (!mdev)
		return -EIO;
P
Philipp Reisner 已提交
1568 1569 1570

	sector = be64_to_cpu(p->sector);

1571
	spin_lock_irq(&mdev->tconn->req_lock);
1572
	req = find_request(mdev, &mdev->read_requests, p->block_id, sector, false, __func__);
1573
	spin_unlock_irq(&mdev->tconn->req_lock);
1574
	if (unlikely(!req))
1575
		return -EIO;
P
Philipp Reisner 已提交
1576

B
Bart Van Assche 已提交
1577
	/* hlist_del(&req->collision) is done in _req_may_be_done, to avoid
P
Philipp Reisner 已提交
1578 1579
	 * special casing it there for the various failure cases.
	 * still no race with drbd_fail_pending_reads */
1580
	err = recv_dless_read(mdev, req, sector, pi->size);
1581
	if (!err)
1582
		req_mod(req, DATA_RECEIVED);
P
Philipp Reisner 已提交
1583 1584 1585 1586
	/* else: nothing. handled from drbd_disconnect...
	 * I don't think we may complete this just yet
	 * in case we are "on-disconnect: freeze" */

1587
	return err;
P
Philipp Reisner 已提交
1588 1589
}

1590
static int receive_RSDataReply(struct drbd_tconn *tconn, struct packet_info *pi)
P
Philipp Reisner 已提交
1591
{
1592
	struct drbd_conf *mdev;
P
Philipp Reisner 已提交
1593
	sector_t sector;
1594
	int err;
1595 1596 1597 1598 1599
	struct p_data *p = tconn->data.rbuf;

	mdev = vnr_to_mdev(tconn, pi->vnr);
	if (!mdev)
		return -EIO;
P
Philipp Reisner 已提交
1600 1601 1602 1603 1604 1605 1606

	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,
1607
		 * or in drbd_peer_request_endio. */
1608
		err = recv_resync_read(mdev, sector, pi->size);
P
Philipp Reisner 已提交
1609 1610 1611 1612
	} else {
		if (__ratelimit(&drbd_ratelimit_state))
			dev_err(DEV, "Can not write resync data to local disk.\n");

1613
		err = drbd_drain_block(mdev, pi->size);
P
Philipp Reisner 已提交
1614

1615
		drbd_send_ack_dp(mdev, P_NEG_ACK, p, pi->size);
P
Philipp Reisner 已提交
1616 1617
	}

1618
	atomic_add(pi->size >> 9, &mdev->rs_sect_in);
1619

1620
	return err;
P
Philipp Reisner 已提交
1621 1622
}

1623
static int w_restart_write(struct drbd_work *w, int cancel)
1624 1625 1626 1627 1628 1629 1630 1631 1632 1633
{
	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);
1634
		return -EIO;
1635 1636 1637 1638 1639 1640 1641 1642 1643
	}
	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 */ ;
1644
	return 0;
1645 1646 1647 1648 1649 1650 1651 1652 1653 1654 1655 1656 1657 1658 1659 1660 1661 1662 1663 1664 1665 1666 1667
}

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
1668 1669 1670
/* e_end_block() is called via drbd_process_done_ee().
 * this means this function only runs in the asender thread
 */
1671
static int e_end_block(struct drbd_work *w, int cancel)
P
Philipp Reisner 已提交
1672
{
1673 1674
	struct drbd_peer_request *peer_req =
		container_of(w, struct drbd_peer_request, w);
1675
	struct drbd_conf *mdev = w->mdev;
1676
	sector_t sector = peer_req->i.sector;
1677
	int err = 0, pcmd;
P
Philipp Reisner 已提交
1678

1679
	if (mdev->tconn->net_conf->wire_protocol == DRBD_PROT_C) {
1680
		if (likely((peer_req->flags & EE_WAS_ERROR) == 0)) {
P
Philipp Reisner 已提交
1681 1682
			pcmd = (mdev->state.conn >= C_SYNC_SOURCE &&
				mdev->state.conn <= C_PAUSED_SYNC_T &&
1683
				peer_req->flags & EE_MAY_SET_IN_SYNC) ?
P
Philipp Reisner 已提交
1684
				P_RS_WRITE_ACK : P_WRITE_ACK;
1685
			err = drbd_send_ack(mdev, pcmd, peer_req);
P
Philipp Reisner 已提交
1686
			if (pcmd == P_RS_WRITE_ACK)
1687
				drbd_set_in_sync(mdev, sector, peer_req->i.size);
P
Philipp Reisner 已提交
1688
		} else {
1689
			err = drbd_send_ack(mdev, P_NEG_ACK, peer_req);
P
Philipp Reisner 已提交
1690 1691 1692 1693 1694 1695 1696
			/* 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.  */
1697
	if (mdev->tconn->net_conf->two_primaries) {
1698
		spin_lock_irq(&mdev->tconn->req_lock);
1699 1700
		D_ASSERT(!drbd_interval_empty(&peer_req->i));
		drbd_remove_epoch_entry_interval(mdev, peer_req);
1701 1702
		if (peer_req->flags & EE_RESTART_REQUESTS)
			restart_conflicting_writes(mdev, sector, peer_req->i.size);
1703
		spin_unlock_irq(&mdev->tconn->req_lock);
1704
	} else
1705
		D_ASSERT(drbd_interval_empty(&peer_req->i));
P
Philipp Reisner 已提交
1706

1707
	drbd_may_finish_epoch(mdev, peer_req->epoch, EV_PUT + (cancel ? EV_CLEANUP : 0));
P
Philipp Reisner 已提交
1708

1709
	return err;
P
Philipp Reisner 已提交
1710 1711
}

1712
static int e_send_ack(struct drbd_work *w, enum drbd_packet ack)
P
Philipp Reisner 已提交
1713
{
1714
	struct drbd_conf *mdev = w->mdev;
1715 1716
	struct drbd_peer_request *peer_req =
		container_of(w, struct drbd_peer_request, w);
1717
	int err;
P
Philipp Reisner 已提交
1718

1719
	err = drbd_send_ack(mdev, ack, peer_req);
P
Philipp Reisner 已提交
1720 1721
	dec_unacked(mdev);

1722
	return err;
P
Philipp Reisner 已提交
1723 1724
}

1725
static int e_send_discard_write(struct drbd_work *w, int unused)
1726 1727 1728 1729
{
	return e_send_ack(w, P_DISCARD_WRITE);
}

1730
static int e_send_retry_write(struct drbd_work *w, int unused)
1731 1732 1733 1734 1735 1736 1737
{
	struct drbd_tconn *tconn = w->mdev->tconn;

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

1738 1739 1740 1741 1742 1743 1744 1745 1746 1747 1748 1749 1750 1751 1752
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;
}

1753 1754 1755 1756 1757 1758 1759 1760 1761 1762 1763 1764 1765
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);
}

1766
static void update_peer_seq(struct drbd_conf *mdev, unsigned int peer_seq)
1767
{
1768
	unsigned int newest_peer_seq;
1769

1770 1771
	if (need_peer_seq(mdev)) {
		spin_lock(&mdev->peer_seq_lock);
1772 1773
		newest_peer_seq = seq_max(mdev->peer_seq, peer_seq);
		mdev->peer_seq = newest_peer_seq;
1774
		spin_unlock(&mdev->peer_seq_lock);
1775 1776
		/* wake up only if we actually changed mdev->peer_seq */
		if (peer_seq == newest_peer_seq)
1777 1778
			wake_up(&mdev->seq_wait);
	}
1779 1780
}

P
Philipp Reisner 已提交
1781 1782 1783 1784 1785 1786 1787 1788 1789 1790 1791 1792 1793 1794 1795 1796 1797 1798 1799 1800 1801
/* 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). */
1802
static int wait_for_and_update_peer_seq(struct drbd_conf *mdev, const u32 peer_seq)
P
Philipp Reisner 已提交
1803 1804 1805
{
	DEFINE_WAIT(wait);
	long timeout;
1806 1807 1808 1809 1810
	int ret;

	if (!need_peer_seq(mdev))
		return 0;

P
Philipp Reisner 已提交
1811 1812
	spin_lock(&mdev->peer_seq_lock);
	for (;;) {
1813 1814 1815
		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
1816
			break;
1817
		}
P
Philipp Reisner 已提交
1818 1819 1820 1821
		if (signal_pending(current)) {
			ret = -ERESTARTSYS;
			break;
		}
1822
		prepare_to_wait(&mdev->seq_wait, &wait, TASK_INTERRUPTIBLE);
P
Philipp Reisner 已提交
1823
		spin_unlock(&mdev->peer_seq_lock);
1824 1825
		timeout = mdev->tconn->net_conf->ping_timeo*HZ/10;
		timeout = schedule_timeout(timeout);
P
Philipp Reisner 已提交
1826
		spin_lock(&mdev->peer_seq_lock);
1827
		if (!timeout) {
P
Philipp Reisner 已提交
1828
			ret = -ETIMEDOUT;
1829
			dev_err(DEV, "Timed out waiting for missing ack packets; disconnecting\n");
P
Philipp Reisner 已提交
1830 1831 1832 1833
			break;
		}
	}
	spin_unlock(&mdev->peer_seq_lock);
1834
	finish_wait(&mdev->seq_wait, &wait);
P
Philipp Reisner 已提交
1835 1836 1837
	return ret;
}

1838 1839 1840 1841
/* 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)
1842
{
1843 1844 1845 1846
	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);
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 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981
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
1982
/* mirrored write */
1983
static int receive_Data(struct drbd_tconn *tconn, struct packet_info *pi)
P
Philipp Reisner 已提交
1984
{
1985
	struct drbd_conf *mdev;
P
Philipp Reisner 已提交
1986
	sector_t sector;
1987
	struct drbd_peer_request *peer_req;
1988
	struct p_data *p = tconn->data.rbuf;
1989
	u32 peer_seq = be32_to_cpu(p->seq_num);
P
Philipp Reisner 已提交
1990 1991
	int rw = WRITE;
	u32 dp_flags;
1992
	int err;
P
Philipp Reisner 已提交
1993

1994 1995 1996 1997
	mdev = vnr_to_mdev(tconn, pi->vnr);
	if (!mdev)
		return -EIO;

1998
	if (!get_ldev(mdev)) {
1999 2000
		int err2;

2001
		err = wait_for_and_update_peer_seq(mdev, peer_seq);
2002
		drbd_send_ack_dp(mdev, P_NEG_ACK, p, pi->size);
P
Philipp Reisner 已提交
2003
		atomic_inc(&mdev->current_epoch->epoch_size);
2004
		err2 = drbd_drain_block(mdev, pi->size);
2005 2006 2007
		if (!err)
			err = err2;
		return err;
P
Philipp Reisner 已提交
2008 2009
	}

2010 2011 2012 2013 2014
	/*
	 * 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
2015 2016

	sector = be64_to_cpu(p->sector);
2017
	peer_req = read_in_block(mdev, p->block_id, sector, pi->size);
2018
	if (!peer_req) {
P
Philipp Reisner 已提交
2019
		put_ldev(mdev);
2020
		return -EIO;
P
Philipp Reisner 已提交
2021 2022
	}

2023
	peer_req->w.cb = e_end_block;
P
Philipp Reisner 已提交
2024

2025 2026 2027 2028
	dp_flags = be32_to_cpu(p->dp_flags);
	rw |= wire_flags_to_bio(mdev, dp_flags);

	if (dp_flags & DP_MAY_SET_IN_SYNC)
2029
		peer_req->flags |= EE_MAY_SET_IN_SYNC;
2030

P
Philipp Reisner 已提交
2031
	spin_lock(&mdev->epoch_lock);
2032 2033 2034
	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
2035 2036
	spin_unlock(&mdev->epoch_lock);

2037 2038 2039
	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
2040
			goto out_interrupted;
2041
		spin_lock_irq(&mdev->tconn->req_lock);
2042 2043 2044 2045
		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
2046
				put_ldev(mdev);
2047
				return 0;
P
Philipp Reisner 已提交
2048
			}
2049
			goto out_interrupted;
P
Philipp Reisner 已提交
2050
		}
2051 2052
	} else
		spin_lock_irq(&mdev->tconn->req_lock);
2053
	list_add(&peer_req->w.list, &mdev->active_ee);
2054
	spin_unlock_irq(&mdev->tconn->req_lock);
P
Philipp Reisner 已提交
2055

2056
	switch (mdev->tconn->net_conf->wire_protocol) {
P
Philipp Reisner 已提交
2057 2058 2059 2060 2061 2062 2063 2064
	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 */
2065
		drbd_send_ack(mdev, P_RECV_ACK, peer_req);
P
Philipp Reisner 已提交
2066 2067 2068 2069 2070 2071
		break;
	case DRBD_PROT_A:
		/* nothing to do */
		break;
	}

2072
	if (mdev->state.pdsk < D_INCONSISTENT) {
P
Philipp Reisner 已提交
2073
		/* In case we have the only disk of the cluster, */
2074 2075 2076
		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;
2077
		drbd_al_begin_io(mdev, &peer_req->i);
P
Philipp Reisner 已提交
2078 2079
	}

2080 2081 2082
	err = drbd_submit_peer_request(mdev, peer_req, rw, DRBD_FAULT_DT_WR);
	if (!err)
		return 0;
P
Philipp Reisner 已提交
2083

2084 2085
	/* don't care for the reason here */
	dev_err(DEV, "submit failed, triggering re-connect\n");
2086
	spin_lock_irq(&mdev->tconn->req_lock);
2087 2088
	list_del(&peer_req->w.list);
	drbd_remove_epoch_entry_interval(mdev, peer_req);
2089
	spin_unlock_irq(&mdev->tconn->req_lock);
2090
	if (peer_req->flags & EE_CALL_AL_COMPLETE_IO)
2091
		drbd_al_complete_io(mdev, &peer_req->i);
2092

P
Philipp Reisner 已提交
2093
out_interrupted:
2094
	drbd_may_finish_epoch(mdev, peer_req->epoch, EV_PUT + EV_CLEANUP);
P
Philipp Reisner 已提交
2095
	put_ldev(mdev);
2096
	drbd_free_ee(mdev, peer_req);
2097
	return err;
P
Philipp Reisner 已提交
2098 2099
}

2100 2101 2102 2103 2104 2105 2106 2107 2108 2109 2110
/* 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.
 */
2111
int drbd_rs_should_slow_down(struct drbd_conf *mdev, sector_t sector)
2112 2113 2114
{
	struct gendisk *disk = mdev->ldev->backing_bdev->bd_contains->bd_disk;
	unsigned long db, dt, dbdt;
2115
	struct lc_element *tmp;
2116 2117 2118 2119
	int curr_events;
	int throttle = 0;

	/* feature disabled? */
2120
	if (mdev->ldev->dc.c_min_rate == 0)
2121 2122
		return 0;

2123 2124 2125 2126 2127 2128 2129 2130 2131 2132 2133 2134
	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);

2135 2136 2137
	curr_events = (int)part_stat_read(&disk->part0, sectors[0]) +
		      (int)part_stat_read(&disk->part0, sectors[1]) -
			atomic_read(&mdev->rs_sect_ev);
2138

2139 2140 2141 2142 2143 2144 2145 2146
	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. */
2147 2148 2149 2150 2151 2152
		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;
2153 2154 2155 2156 2157 2158 2159

		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);

2160
		if (dbdt > mdev->ldev->dc.c_min_rate)
2161 2162 2163 2164 2165 2166
			throttle = 1;
	}
	return throttle;
}


2167
static int receive_DataRequest(struct drbd_tconn *tconn, struct packet_info *pi)
P
Philipp Reisner 已提交
2168
{
2169
	struct drbd_conf *mdev;
P
Philipp Reisner 已提交
2170
	sector_t sector;
2171
	sector_t capacity;
2172
	struct drbd_peer_request *peer_req;
P
Philipp Reisner 已提交
2173
	struct digest_info *di = NULL;
2174
	int size, verb;
P
Philipp Reisner 已提交
2175
	unsigned int fault_type;
2176 2177 2178 2179 2180 2181
	struct p_block_req *p =	tconn->data.rbuf;

	mdev = vnr_to_mdev(tconn, pi->vnr);
	if (!mdev)
		return -EIO;
	capacity = drbd_get_capacity(mdev->this_bdev);
P
Philipp Reisner 已提交
2182 2183 2184 2185

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

2186
	if (size <= 0 || !IS_ALIGNED(size, 512) || size > DRBD_MAX_BIO_SIZE) {
P
Philipp Reisner 已提交
2187 2188
		dev_err(DEV, "%s:%d: sector: %llus, size: %u\n", __FILE__, __LINE__,
				(unsigned long long)sector, size);
2189
		return -EINVAL;
P
Philipp Reisner 已提交
2190 2191 2192 2193
	}
	if (sector + (size>>9) > capacity) {
		dev_err(DEV, "%s:%d: sector: %llus, size: %u\n", __FILE__, __LINE__,
				(unsigned long long)sector, size);
2194
		return -EINVAL;
P
Philipp Reisner 已提交
2195 2196 2197
	}

	if (!get_ldev_if_state(mdev, D_UP_TO_DATE)) {
2198
		verb = 1;
2199
		switch (pi->cmd) {
2200 2201 2202 2203 2204 2205 2206 2207 2208 2209 2210 2211 2212 2213
		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:
2214
			BUG();
2215 2216
		}
		if (verb && __ratelimit(&drbd_ratelimit_state))
P
Philipp Reisner 已提交
2217 2218
			dev_err(DEV, "Can not satisfy peer's read request, "
			    "no local data.\n");
2219

L
Lars Ellenberg 已提交
2220
		/* drain possibly payload */
2221
		return drbd_drain_block(mdev, pi->size);
P
Philipp Reisner 已提交
2222 2223 2224 2225 2226
	}

	/* 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.  */
2227 2228
	peer_req = drbd_alloc_ee(mdev, p->block_id, sector, size, GFP_NOIO);
	if (!peer_req) {
P
Philipp Reisner 已提交
2229
		put_ldev(mdev);
2230
		return -ENOMEM;
P
Philipp Reisner 已提交
2231 2232
	}

2233
	switch (pi->cmd) {
P
Philipp Reisner 已提交
2234
	case P_DATA_REQUEST:
2235
		peer_req->w.cb = w_e_end_data_req;
P
Philipp Reisner 已提交
2236
		fault_type = DRBD_FAULT_DT_RD;
2237 2238 2239
		/* application IO, don't drbd_rs_begin_io */
		goto submit;

P
Philipp Reisner 已提交
2240
	case P_RS_DATA_REQUEST:
2241
		peer_req->w.cb = w_e_end_rsdata_req;
P
Philipp Reisner 已提交
2242
		fault_type = DRBD_FAULT_RS_RD;
2243 2244
		/* used in the sector offset progress display */
		mdev->bm_resync_fo = BM_SECT_TO_BIT(sector);
P
Philipp Reisner 已提交
2245 2246 2247 2248 2249
		break;

	case P_OV_REPLY:
	case P_CSUM_RS_REQUEST:
		fault_type = DRBD_FAULT_RS_RD;
2250
		di = kmalloc(sizeof(*di) + pi->size, GFP_NOIO);
P
Philipp Reisner 已提交
2251 2252 2253
		if (!di)
			goto out_free_e;

2254
		di->digest_size = pi->size;
P
Philipp Reisner 已提交
2255 2256
		di->digest = (((char *)di)+sizeof(struct digest_info));

2257 2258
		peer_req->digest = di;
		peer_req->flags |= EE_HAS_DIGEST;
2259

2260
		if (drbd_recv_all(mdev->tconn, di->digest, pi->size))
P
Philipp Reisner 已提交
2261 2262
			goto out_free_e;

2263
		if (pi->cmd == P_CSUM_RS_REQUEST) {
2264
			D_ASSERT(mdev->tconn->agreed_pro_version >= 89);
2265
			peer_req->w.cb = w_e_end_csum_rs_req;
2266 2267
			/* used in the sector offset progress display */
			mdev->bm_resync_fo = BM_SECT_TO_BIT(sector);
2268
		} else if (pi->cmd == P_OV_REPLY) {
2269 2270
			/* track progress, we may need to throttle */
			atomic_add(size >> 9, &mdev->rs_sect_in);
2271
			peer_req->w.cb = w_e_end_ov_reply;
P
Philipp Reisner 已提交
2272
			dec_rs_pending(mdev);
2273 2274 2275
			/* 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
2276 2277 2278 2279 2280
		}
		break;

	case P_OV_REQUEST:
		if (mdev->ov_start_sector == ~(sector_t)0 &&
2281
		    mdev->tconn->agreed_pro_version >= 90) {
2282 2283
			unsigned long now = jiffies;
			int i;
P
Philipp Reisner 已提交
2284 2285
			mdev->ov_start_sector = sector;
			mdev->ov_position = sector;
2286 2287
			mdev->ov_left = drbd_bm_bits(mdev) - BM_SECT_TO_BIT(sector);
			mdev->rs_total = mdev->ov_left;
2288 2289 2290 2291
			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
2292 2293 2294
			dev_info(DEV, "Online Verify start sector: %llu\n",
					(unsigned long long)sector);
		}
2295
		peer_req->w.cb = w_e_end_ov_req;
P
Philipp Reisner 已提交
2296 2297 2298 2299
		fault_type = DRBD_FAULT_RS_RD;
		break;

	default:
2300
		BUG();
P
Philipp Reisner 已提交
2301 2302
	}

2303 2304 2305 2306 2307 2308 2309 2310 2311 2312 2313 2314 2315 2316 2317 2318 2319 2320 2321 2322 2323 2324
	/* 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.
	 */
2325 2326 2327
	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))
2328
		goto out_free_e;
P
Philipp Reisner 已提交
2329

2330 2331 2332
submit_for_resync:
	atomic_add(size >> 9, &mdev->rs_sect_ev);

2333
submit:
P
Philipp Reisner 已提交
2334
	inc_unacked(mdev);
2335
	spin_lock_irq(&mdev->tconn->req_lock);
2336
	list_add_tail(&peer_req->w.list, &mdev->read_ee);
2337
	spin_unlock_irq(&mdev->tconn->req_lock);
P
Philipp Reisner 已提交
2338

2339
	if (drbd_submit_peer_request(mdev, peer_req, READ, fault_type) == 0)
2340
		return 0;
P
Philipp Reisner 已提交
2341

2342 2343
	/* don't care for the reason here */
	dev_err(DEV, "submit failed, triggering re-connect\n");
2344
	spin_lock_irq(&mdev->tconn->req_lock);
2345
	list_del(&peer_req->w.list);
2346
	spin_unlock_irq(&mdev->tconn->req_lock);
2347 2348
	/* no drbd_rs_complete_io(), we are dropping the connection anyways */

P
Philipp Reisner 已提交
2349 2350
out_free_e:
	put_ldev(mdev);
2351
	drbd_free_ee(mdev, peer_req);
2352
	return -EIO;
P
Philipp Reisner 已提交
2353 2354 2355 2356 2357 2358 2359 2360 2361 2362 2363 2364 2365
}

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;

2366
	switch (mdev->tconn->net_conf->after_sb_0p) {
P
Philipp Reisner 已提交
2367 2368 2369 2370 2371 2372 2373 2374 2375 2376 2377 2378 2379 2380 2381 2382 2383 2384 2385 2386 2387 2388 2389 2390 2391 2392 2393
	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
2394
		dev_warn(DEV, "Discard younger/older primary did not find a decision\n"
P
Philipp Reisner 已提交
2395 2396 2397
		     "Using discard-least-changes instead\n");
	case ASB_DISCARD_ZERO_CHG:
		if (ch_peer == 0 && ch_self == 0) {
2398
			rv = test_bit(DISCARD_CONCURRENT, &mdev->tconn->flags)
P
Philipp Reisner 已提交
2399 2400 2401 2402 2403 2404
				? -1 : 1;
			break;
		} else {
			if (ch_peer == 0) { rv =  1; break; }
			if (ch_self == 0) { rv = -1; break; }
		}
2405
		if (mdev->tconn->net_conf->after_sb_0p == ASB_DISCARD_ZERO_CHG)
P
Philipp Reisner 已提交
2406 2407 2408 2409 2410 2411 2412 2413
			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. */
2414
			rv = test_bit(DISCARD_CONCURRENT, &mdev->tconn->flags)
P
Philipp Reisner 已提交
2415 2416 2417 2418 2419 2420 2421 2422 2423 2424 2425 2426 2427 2428
				? -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)
{
2429
	int hg, rv = -100;
P
Philipp Reisner 已提交
2430

2431
	switch (mdev->tconn->net_conf->after_sb_1p) {
P
Philipp Reisner 已提交
2432 2433 2434 2435 2436 2437 2438 2439 2440 2441 2442 2443 2444 2445 2446 2447 2448 2449 2450 2451 2452 2453 2454 2455
	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) {
2456 2457 2458
			enum drbd_state_rv rv2;

			drbd_set_role(mdev, R_SECONDARY, 0);
P
Philipp Reisner 已提交
2459 2460 2461
			 /* 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. */
2462 2463
			rv2 = drbd_change_state(mdev, CS_VERBOSE, NS(role, R_SECONDARY));
			if (rv2 != SS_SUCCESS) {
P
Philipp Reisner 已提交
2464 2465 2466 2467 2468 2469 2470 2471 2472 2473 2474 2475 2476 2477
				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)
{
2478
	int hg, rv = -100;
P
Philipp Reisner 已提交
2479

2480
	switch (mdev->tconn->net_conf->after_sb_2p) {
P
Philipp Reisner 已提交
2481 2482 2483 2484 2485 2486 2487 2488 2489 2490 2491 2492 2493 2494 2495 2496 2497
	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) {
2498 2499
			enum drbd_state_rv rv2;

P
Philipp Reisner 已提交
2500 2501 2502
			 /* 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. */
2503 2504
			rv2 = drbd_change_state(mdev, CS_VERBOSE, NS(role, R_SECONDARY));
			if (rv2 != SS_SUCCESS) {
P
Philipp Reisner 已提交
2505 2506 2507 2508 2509 2510 2511 2512 2513 2514 2515 2516 2517 2518 2519 2520 2521 2522 2523 2524 2525 2526 2527 2528 2529 2530 2531 2532 2533 2534 2535 2536 2537 2538 2539 2540 2541 2542
				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
2543 2544
-1091   requires proto 91
-1096   requires proto 96
P
Philipp Reisner 已提交
2545 2546 2547 2548 2549 2550 2551 2552 2553 2554 2555 2556 2557 2558 2559 2560 2561 2562 2563 2564 2565 2566 2567 2568 2569 2570 2571 2572
 */
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) {

2573
			if (mdev->tconn->agreed_pro_version < 91)
2574
				return -1091;
P
Philipp Reisner 已提交
2575 2576 2577 2578 2579 2580 2581 2582 2583 2584 2585 2586 2587 2588 2589 2590 2591 2592 2593

			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) {

2594
			if (mdev->tconn->agreed_pro_version < 91)
2595
				return -1091;
P
Philipp Reisner 已提交
2596 2597 2598 2599 2600 2601 2602 2603 2604 2605 2606 2607 2608 2609 2610 2611 2612 2613 2614 2615 2616 2617 2618 2619 2620 2621 2622 2623 2624 2625 2626

			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 */
2627
			dc = test_bit(DISCARD_CONCURRENT, &mdev->tconn->flags);
P
Philipp Reisner 已提交
2628 2629 2630 2631 2632 2633 2634 2635 2636 2637 2638 2639
			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) {
2640
		if (mdev->tconn->agreed_pro_version < 96 ?
2641 2642 2643
		    (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
2644 2645 2646
			/* The last P_SYNC_UUID did not get though. Undo the last start of
			   resync as sync source modifications of the peer's UUIDs. */

2647
			if (mdev->tconn->agreed_pro_version < 91)
2648
				return -1091;
P
Philipp Reisner 已提交
2649 2650 2651

			mdev->p_uuid[UI_BITMAP] = mdev->p_uuid[UI_HISTORY_START];
			mdev->p_uuid[UI_HISTORY_START] = mdev->p_uuid[UI_HISTORY_START + 1];
2652 2653 2654 2655

			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
2656 2657 2658 2659 2660 2661 2662 2663 2664 2665 2666 2667 2668 2669 2670 2671 2672 2673 2674 2675 2676
			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) {
2677
		if (mdev->tconn->agreed_pro_version < 96 ?
2678 2679 2680
		    (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
2681 2682 2683
			/* The last P_SYNC_UUID did not get though. Undo the last start of
			   resync as sync source modifications of our UUIDs. */

2684
			if (mdev->tconn->agreed_pro_version < 91)
2685
				return -1091;
P
Philipp Reisner 已提交
2686 2687 2688 2689

			_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]);

2690
			dev_info(DEV, "Last syncUUID did not get through, corrected:\n");
P
Philipp Reisner 已提交
2691 2692 2693 2694 2695 2696 2697 2698 2699
			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;
2700
	peer = mdev->p_uuid[UI_CURRENT] & ~((u64)1);
P
Philipp Reisner 已提交
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 2726 2727 2728 2729 2730 2731 2732 2733 2734 2735 2736 2737 2738 2739 2740 2741 2742 2743 2744 2745 2746 2747 2748 2749 2750 2751 2752
	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;
	}
2753 2754
	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
2755 2756 2757 2758 2759 2760 2761 2762 2763 2764 2765 2766 2767
		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");
	}

2768 2769 2770
	if (abs(hg) == 100)
		drbd_khelper(mdev, "initial-split-brain");

2771
	if (hg == 100 || (hg == -100 && mdev->tconn->net_conf->always_asbp)) {
P
Philipp Reisner 已提交
2772 2773 2774 2775 2776 2777 2778 2779 2780 2781 2782 2783 2784 2785 2786 2787 2788 2789 2790 2791 2792 2793 2794 2795 2796 2797 2798 2799
		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) {
2800
		if (mdev->tconn->net_conf->want_lose && !(mdev->p_uuid[UI_FLAGS]&1))
P
Philipp Reisner 已提交
2801
			hg = -1;
2802
		if (!mdev->tconn->net_conf->want_lose && (mdev->p_uuid[UI_FLAGS]&1))
P
Philipp Reisner 已提交
2803 2804 2805 2806 2807 2808 2809 2810 2811
			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) {
2812 2813 2814 2815
		/* 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... */
2816
		dev_alert(DEV, "Split-Brain detected but unresolved, dropping connection!\n");
P
Philipp Reisner 已提交
2817 2818 2819 2820 2821 2822 2823 2824 2825 2826 2827
		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) {
2828
		switch (mdev->tconn->net_conf->rr_conflict) {
P
Philipp Reisner 已提交
2829 2830 2831 2832 2833 2834 2835 2836 2837 2838 2839 2840
		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");
		}
	}

2841
	if (mdev->tconn->net_conf->dry_run || test_bit(CONN_DRY_RUN, &mdev->tconn->flags)) {
2842 2843 2844 2845 2846 2847 2848 2849 2850
		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
2851 2852
	if (abs(hg) >= 2) {
		dev_info(DEV, "Writing the whole bitmap, full sync required after drbd_sync_handshake.\n");
2853 2854
		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
2855 2856 2857 2858 2859 2860 2861 2862 2863 2864 2865 2866 2867 2868 2869 2870 2871 2872 2873 2874 2875 2876 2877 2878 2879 2880 2881 2882 2883 2884 2885 2886 2887 2888 2889 2890 2891 2892 2893
			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;
}

2894
static int receive_protocol(struct drbd_tconn *tconn, struct packet_info *pi)
P
Philipp Reisner 已提交
2895
{
2896
	struct p_protocol *p = tconn->data.rbuf;
P
Philipp Reisner 已提交
2897
	int p_proto, p_after_sb_0p, p_after_sb_1p, p_after_sb_2p;
2898
	int p_want_lose, p_two_primaries, cf;
P
Philipp Reisner 已提交
2899 2900 2901 2902 2903 2904 2905
	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);
2906 2907 2908
	cf		= be32_to_cpu(p->conn_flags);
	p_want_lose = cf & CF_WANT_LOSE;

2909
	clear_bit(CONN_DRY_RUN, &tconn->flags);
2910 2911

	if (cf & CF_DRY_RUN)
2912
		set_bit(CONN_DRY_RUN, &tconn->flags);
P
Philipp Reisner 已提交
2913

2914 2915
	if (p_proto != tconn->net_conf->wire_protocol) {
		conn_err(tconn, "incompatible communication protocols\n");
P
Philipp Reisner 已提交
2916 2917 2918
		goto disconnect;
	}

2919 2920
	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
2921 2922 2923
		goto disconnect;
	}

2924 2925
	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
2926 2927 2928
		goto disconnect;
	}

2929 2930
	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
2931 2932 2933
		goto disconnect;
	}

2934 2935
	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
2936 2937 2938
		goto disconnect;
	}

2939 2940
	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
2941 2942 2943
		goto disconnect;
	}

2944 2945
	if (tconn->agreed_pro_version >= 87) {
		unsigned char *my_alg = tconn->net_conf->integrity_alg;
2946
		int err;
P
Philipp Reisner 已提交
2947

2948
		err = drbd_recv_all(tconn, p_integrity_alg, pi->size);
2949 2950
		if (err)
			return err;
P
Philipp Reisner 已提交
2951 2952 2953

		p_integrity_alg[SHARED_SECRET_MAX-1] = 0;
		if (strcmp(p_integrity_alg, my_alg)) {
2954
			conn_err(tconn, "incompatible setting of the data-integrity-alg\n");
P
Philipp Reisner 已提交
2955 2956
			goto disconnect;
		}
2957
		conn_info(tconn, "data-integrity-alg: %s\n",
P
Philipp Reisner 已提交
2958 2959 2960
		     my_alg[0] ? my_alg : (unsigned char *)"<not-used>");
	}

2961
	return 0;
P
Philipp Reisner 已提交
2962 2963

disconnect:
2964
	conn_request_state(tconn, NS(conn, C_DISCONNECTING), CS_HARD);
2965
	return -EIO;
P
Philipp Reisner 已提交
2966 2967 2968 2969 2970 2971 2972 2973 2974 2975 2976 2977 2978 2979 2980 2981 2982 2983 2984 2985 2986 2987 2988 2989 2990 2991 2992 2993 2994
}

/* 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;
}

2995 2996 2997 2998 2999 3000 3001 3002 3003 3004 3005 3006 3007 3008 3009 3010 3011 3012 3013 3014 3015 3016 3017 3018 3019 3020 3021 3022 3023 3024 3025 3026 3027 3028 3029 3030 3031 3032 3033
static int ignore_remaining_packet(struct drbd_tconn *tconn, struct packet_info *pi)
{
	void *buffer = tconn->data.rbuf;
	int size = pi->size;

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

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

static int receive_SyncParam(struct drbd_tconn *tconn, struct packet_info *pi)
P
Philipp Reisner 已提交
3034
{
3035 3036
	struct drbd_conf *mdev;
	struct p_rs_param_95 *p = tconn->data.rbuf;
P
Philipp Reisner 已提交
3037 3038 3039
	unsigned int header_size, data_size, exp_max_sz;
	struct crypto_hash *verify_tfm = NULL;
	struct crypto_hash *csums_tfm = NULL;
3040
	const int apv = tconn->agreed_pro_version;
3041 3042
	int *rs_plan_s = NULL;
	int fifo_size = 0;
3043
	int err;
P
Philipp Reisner 已提交
3044

3045 3046 3047 3048
	mdev = vnr_to_mdev(tconn, pi->vnr);
	if (!mdev)
		return config_unknown_volume(tconn, pi);

P
Philipp Reisner 已提交
3049 3050 3051
	exp_max_sz  = apv <= 87 ? sizeof(struct p_rs_param)
		    : apv == 88 ? sizeof(struct p_rs_param)
					+ SHARED_SECRET_MAX
3052 3053
		    : apv <= 94 ? sizeof(struct p_rs_param_89)
		    : /* apv >= 95 */ sizeof(struct p_rs_param_95);
P
Philipp Reisner 已提交
3054

3055
	if (pi->size > exp_max_sz) {
P
Philipp Reisner 已提交
3056
		dev_err(DEV, "SyncParam packet too long: received %u, expected <= %u bytes\n",
3057
		    pi->size, exp_max_sz);
3058
		return -EIO;
P
Philipp Reisner 已提交
3059 3060 3061
	}

	if (apv <= 88) {
3062
		header_size = sizeof(struct p_rs_param) - sizeof(struct p_header);
3063
		data_size = pi->size - header_size;
3064
	} else if (apv <= 94) {
3065
		header_size = sizeof(struct p_rs_param_89) - sizeof(struct p_header);
3066
		data_size = pi->size - header_size;
P
Philipp Reisner 已提交
3067
		D_ASSERT(data_size == 0);
3068
	} else {
3069
		header_size = sizeof(struct p_rs_param_95) - sizeof(struct p_header);
3070
		data_size = pi->size - header_size;
P
Philipp Reisner 已提交
3071 3072 3073 3074 3075 3076
		D_ASSERT(data_size == 0);
	}

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

3077 3078 3079
	err = drbd_recv_all(mdev->tconn, &p->head.payload, header_size);
	if (err)
		return err;
P
Philipp Reisner 已提交
3080

3081 3082 3083 3084
	if (get_ldev(mdev)) {
		mdev->ldev->dc.resync_rate = be32_to_cpu(p->rate);
		put_ldev(mdev);
	}
P
Philipp Reisner 已提交
3085 3086 3087 3088 3089 3090 3091

	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);
3092
				return -EIO;
P
Philipp Reisner 已提交
3093 3094
			}

3095 3096 3097
			err = drbd_recv_all(mdev->tconn, p->verify_alg, data_size);
			if (err)
				return err;
P
Philipp Reisner 已提交
3098 3099 3100 3101 3102 3103 3104 3105 3106 3107 3108 3109 3110 3111 3112

			/* 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;
		}

3113
		if (strcmp(mdev->tconn->net_conf->verify_alg, p->verify_alg)) {
P
Philipp Reisner 已提交
3114 3115
			if (mdev->state.conn == C_WF_REPORT_PARAMS) {
				dev_err(DEV, "Different verify-alg settings. me=\"%s\" peer=\"%s\"\n",
3116
				    mdev->tconn->net_conf->verify_alg, p->verify_alg);
P
Philipp Reisner 已提交
3117 3118 3119 3120 3121 3122 3123 3124 3125 3126
				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;
			}
		}

3127
		if (apv >= 89 && strcmp(mdev->tconn->net_conf->csums_alg, p->csums_alg)) {
P
Philipp Reisner 已提交
3128 3129
			if (mdev->state.conn == C_WF_REPORT_PARAMS) {
				dev_err(DEV, "Different csums-alg settings. me=\"%s\" peer=\"%s\"\n",
3130
				    mdev->tconn->net_conf->csums_alg, p->csums_alg);
P
Philipp Reisner 已提交
3131 3132 3133 3134 3135 3136 3137 3138 3139 3140
				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;
			}
		}

3141 3142 3143 3144 3145 3146
		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);
3147

3148
			fifo_size = (mdev->ldev->dc.c_plan_ahead * 10 * SLEEP_TIME) / HZ;
3149 3150 3151 3152
			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");
3153
					put_ldev(mdev);
3154 3155 3156
					goto disconnect;
				}
			}
3157
			put_ldev(mdev);
3158
		}
P
Philipp Reisner 已提交
3159 3160 3161 3162

		spin_lock(&mdev->peer_seq_lock);
		/* lock against drbd_nl_syncer_conf() */
		if (verify_tfm) {
3163 3164 3165 3166
			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
3167 3168 3169
			dev_info(DEV, "using verify-alg: \"%s\"\n", p->verify_alg);
		}
		if (csums_tfm) {
3170 3171 3172 3173
			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
3174 3175
			dev_info(DEV, "using csums-alg: \"%s\"\n", p->csums_alg);
		}
3176 3177 3178 3179 3180 3181
		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
3182 3183
		spin_unlock(&mdev->peer_seq_lock);
	}
3184
	return 0;
P
Philipp Reisner 已提交
3185 3186 3187 3188 3189 3190 3191

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);
3192
	conn_request_state(mdev->tconn, NS(conn, C_DISCONNECTING), CS_HARD);
3193
	return -EIO;
P
Philipp Reisner 已提交
3194 3195 3196 3197 3198 3199 3200 3201 3202 3203 3204 3205 3206 3207 3208
}

/* 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);
}

3209
static int receive_sizes(struct drbd_tconn *tconn, struct packet_info *pi)
P
Philipp Reisner 已提交
3210
{
3211 3212
	struct drbd_conf *mdev;
	struct p_sizes *p = tconn->data.rbuf;
P
Philipp Reisner 已提交
3213 3214 3215
	enum determine_dev_size dd = unchanged;
	sector_t p_size, p_usize, my_usize;
	int ldsc = 0; /* local disk size changed */
3216
	enum dds_flags ddsf;
P
Philipp Reisner 已提交
3217

3218 3219 3220 3221
	mdev = vnr_to_mdev(tconn, pi->vnr);
	if (!mdev)
		return config_unknown_volume(tconn, pi);

P
Philipp Reisner 已提交
3222 3223 3224 3225 3226 3227 3228 3229 3230 3231 3232 3233 3234 3235 3236 3237 3238 3239 3240 3241 3242 3243 3244 3245 3246 3247 3248 3249 3250
	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. */
3251
		if (drbd_new_dev_size(mdev, mdev->ldev, 0) <
P
Philipp Reisner 已提交
3252 3253 3254 3255
		   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");
3256
			conn_request_state(mdev->tconn, NS(conn, C_DISCONNECTING), CS_HARD);
P
Philipp Reisner 已提交
3257 3258
			mdev->ldev->dc.disk_size = my_usize;
			put_ldev(mdev);
3259
			return -EIO;
P
Philipp Reisner 已提交
3260 3261 3262 3263
		}
		put_ldev(mdev);
	}

3264
	ddsf = be16_to_cpu(p->dds_flags);
P
Philipp Reisner 已提交
3265
	if (get_ldev(mdev)) {
B
Bart Van Assche 已提交
3266
		dd = drbd_determine_dev_size(mdev, ddsf);
P
Philipp Reisner 已提交
3267 3268
		put_ldev(mdev);
		if (dd == dev_size_error)
3269
			return -EIO;
P
Philipp Reisner 已提交
3270 3271 3272 3273 3274 3275
		drbd_md_sync(mdev);
	} else {
		/* I am diskless, need to accept the peer's size. */
		drbd_set_my_capacity(mdev, p_size);
	}

3276 3277 3278
	mdev->peer_max_bio_size = be32_to_cpu(p->max_bio_size);
	drbd_reconsider_max_bio_size(mdev);

P
Philipp Reisner 已提交
3279 3280 3281 3282 3283 3284 3285 3286 3287 3288 3289 3290 3291 3292
	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... */
3293
			drbd_send_sizes(mdev, 0, ddsf);
P
Philipp Reisner 已提交
3294 3295 3296 3297
		}
		if (test_and_clear_bit(RESIZE_PENDING, &mdev->flags) ||
		    (dd == grew && mdev->state.conn == C_CONNECTED)) {
			if (mdev->state.pdsk >= D_INCONSISTENT &&
3298 3299 3300 3301 3302 3303
			    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
3304 3305 3306 3307
				set_bit(RESYNC_AFTER_NEG, &mdev->flags);
		}
	}

3308
	return 0;
P
Philipp Reisner 已提交
3309 3310
}

3311
static int receive_uuids(struct drbd_tconn *tconn, struct packet_info *pi)
P
Philipp Reisner 已提交
3312
{
3313 3314
	struct drbd_conf *mdev;
	struct p_uuids *p = tconn->data.rbuf;
P
Philipp Reisner 已提交
3315
	u64 *p_uuid;
3316
	int i, updated_uuids = 0;
P
Philipp Reisner 已提交
3317

3318 3319 3320 3321
	mdev = vnr_to_mdev(tconn, pi->vnr);
	if (!mdev)
		return config_unknown_volume(tconn, pi);

P
Philipp Reisner 已提交
3322 3323 3324 3325 3326 3327 3328 3329 3330 3331 3332 3333 3334 3335
	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);
3336
		conn_request_state(mdev->tconn, NS(conn, C_DISCONNECTING), CS_HARD);
3337
		return -EIO;
P
Philipp Reisner 已提交
3338 3339 3340 3341 3342
	}

	if (get_ldev(mdev)) {
		int skip_initial_sync =
			mdev->state.conn == C_CONNECTED &&
3343
			mdev->tconn->agreed_pro_version >= 90 &&
P
Philipp Reisner 已提交
3344 3345 3346 3347 3348
			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,
3349 3350
					"clear_n_write from receive_uuids",
					BM_LOCKED_TEST_ALLOWED);
P
Philipp Reisner 已提交
3351 3352 3353 3354 3355
			_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);
3356
			updated_uuids = 1;
P
Philipp Reisner 已提交
3357 3358
		}
		put_ldev(mdev);
3359 3360 3361 3362
	} 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. */
3363
		updated_uuids = drbd_set_ed_uuid(mdev, p_uuid[UI_CURRENT]);
P
Philipp Reisner 已提交
3364 3365 3366 3367 3368 3369
	}

	/* 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... */
3370 3371
	mutex_lock(mdev->state_mutex);
	mutex_unlock(mdev->state_mutex);
P
Philipp Reisner 已提交
3372
	if (mdev->state.conn >= C_CONNECTED && mdev->state.disk < D_INCONSISTENT)
3373 3374 3375 3376
		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
3377

3378
	return 0;
P
Philipp Reisner 已提交
3379 3380 3381 3382 3383 3384 3385 3386 3387 3388 3389 3390 3391 3392 3393 3394 3395 3396 3397 3398 3399 3400 3401 3402 3403 3404 3405 3406 3407 3408 3409 3410
}

/**
 * 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;
}

3411
static int receive_req_state(struct drbd_tconn *tconn, struct packet_info *pi)
P
Philipp Reisner 已提交
3412
{
3413 3414
	struct drbd_conf *mdev;
	struct p_req_state *p = tconn->data.rbuf;
P
Philipp Reisner 已提交
3415
	union drbd_state mask, val;
3416
	enum drbd_state_rv rv;
P
Philipp Reisner 已提交
3417

3418 3419 3420 3421
	mdev = vnr_to_mdev(tconn, pi->vnr);
	if (!mdev)
		return -EIO;

P
Philipp Reisner 已提交
3422 3423 3424
	mask.i = be32_to_cpu(p->mask);
	val.i = be32_to_cpu(p->val);

3425
	if (test_bit(DISCARD_CONCURRENT, &mdev->tconn->flags) &&
3426
	    mutex_is_locked(mdev->state_mutex)) {
P
Philipp Reisner 已提交
3427
		drbd_send_sr_reply(mdev, SS_CONCURRENT_ST_CHG);
3428
		return 0;
P
Philipp Reisner 已提交
3429 3430 3431 3432 3433
	}

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

3434 3435
	rv = drbd_change_state(mdev, CS_VERBOSE, mask, val);
	drbd_send_sr_reply(mdev, rv);
P
Philipp Reisner 已提交
3436 3437 3438

	drbd_md_sync(mdev);

3439
	return 0;
P
Philipp Reisner 已提交
3440 3441
}

3442
static int receive_req_conn_state(struct drbd_tconn *tconn, struct packet_info *pi)
3443
{
3444
	struct p_req_state *p = tconn->data.rbuf;
3445 3446 3447 3448 3449 3450 3451 3452 3453
	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);
3454
		return 0;
3455 3456 3457 3458 3459
	}

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

3460
	rv = conn_request_state(tconn, mask, val, CS_VERBOSE | CS_LOCAL_ONLY | CS_IGN_OUTD_FAIL);
3461 3462
	conn_send_sr_reply(tconn, rv);

3463
	return 0;
3464 3465
}

3466
static int receive_state(struct drbd_tconn *tconn, struct packet_info *pi)
P
Philipp Reisner 已提交
3467
{
3468 3469
	struct drbd_conf *mdev;
	struct p_state *p = tconn->data.rbuf;
3470
	union drbd_state os, ns, peer_state;
P
Philipp Reisner 已提交
3471
	enum drbd_disk_state real_peer_disk;
3472
	enum chg_state_flags cs_flags;
P
Philipp Reisner 已提交
3473 3474
	int rv;

3475 3476 3477 3478
	mdev = vnr_to_mdev(tconn, pi->vnr);
	if (!mdev)
		return config_unknown_volume(tconn, pi);

P
Philipp Reisner 已提交
3479 3480 3481 3482 3483 3484 3485 3486
	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));
	}

3487
	spin_lock_irq(&mdev->tconn->req_lock);
P
Philipp Reisner 已提交
3488
 retry:
3489
	os = ns = drbd_read_state(mdev);
3490
	spin_unlock_irq(&mdev->tconn->req_lock);
P
Philipp Reisner 已提交
3491

3492 3493 3494 3495 3496 3497 3498 3499 3500 3501 3502 3503 3504 3505 3506 3507 3508 3509 3510 3511 3512
	/* 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);
3513
			return 0;
3514 3515 3516 3517 3518 3519 3520 3521 3522 3523 3524 3525
		}
	}

	/* 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;

3526 3527
	if (ns.conn == C_WF_REPORT_PARAMS)
		ns.conn = C_CONNECTED;
P
Philipp Reisner 已提交
3528

3529 3530 3531
	if (peer_state.conn == C_AHEAD)
		ns.conn = C_BEHIND;

P
Philipp Reisner 已提交
3532 3533 3534 3535 3536
	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 */
3537
		cr  = (os.conn < C_CONNECTED);
P
Philipp Reisner 已提交
3538 3539
		/* if we had an established connection
		 * and one of the nodes newly attaches a disk */
3540
		cr |= (os.conn == C_CONNECTED &&
P
Philipp Reisner 已提交
3541
		       (peer_state.disk == D_NEGOTIATING ||
3542
			os.disk == D_NEGOTIATING));
P
Philipp Reisner 已提交
3543 3544 3545 3546 3547
		/* 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" */
3548
		cr |= (os.conn == C_CONNECTED &&
P
Philipp Reisner 已提交
3549 3550 3551 3552
				(peer_state.conn >= C_STARTING_SYNC_S &&
				 peer_state.conn <= C_WF_BITMAP_T));

		if (cr)
3553
			ns.conn = drbd_sync_handshake(mdev, peer_state.role, real_peer_disk);
P
Philipp Reisner 已提交
3554 3555

		put_ldev(mdev);
3556 3557
		if (ns.conn == C_MASK) {
			ns.conn = C_CONNECTED;
P
Philipp Reisner 已提交
3558
			if (mdev->state.disk == D_NEGOTIATING) {
3559
				drbd_force_state(mdev, NS(disk, D_FAILED));
P
Philipp Reisner 已提交
3560 3561 3562
			} 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;
3563
				real_peer_disk = D_DISKLESS;
P
Philipp Reisner 已提交
3564
			} else {
3565
				if (test_and_clear_bit(CONN_DRY_RUN, &mdev->tconn->flags))
3566
					return -EIO;
3567
				D_ASSERT(os.conn == C_WF_REPORT_PARAMS);
3568
				conn_request_state(mdev->tconn, NS(conn, C_DISCONNECTING), CS_HARD);
3569
				return -EIO;
P
Philipp Reisner 已提交
3570 3571 3572 3573
			}
		}
	}

3574
	spin_lock_irq(&mdev->tconn->req_lock);
3575
	if (os.i != drbd_read_state(mdev).i)
P
Philipp Reisner 已提交
3576 3577 3578 3579 3580
		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);
3581
	if ((ns.conn == C_CONNECTED || ns.conn == C_WF_BITMAP_S) && ns.disk == D_NEGOTIATING)
P
Philipp Reisner 已提交
3582
		ns.disk = mdev->new_state_tmp.disk;
3583
	cs_flags = CS_VERBOSE + (os.conn < C_CONNECTED && ns.conn >= C_CONNECTED ? 0 : CS_HARD);
3584
	if (ns.pdsk == D_CONSISTENT && drbd_suspended(mdev) && ns.conn == C_CONNECTED && os.conn < C_CONNECTED &&
3585
	    test_bit(NEW_CUR_UUID, &mdev->flags)) {
3586
		/* Do not allow tl_restart(RESEND) for a rebooted peer. We can only allow this
3587
		   for temporal network outages! */
3588
		spin_unlock_irq(&mdev->tconn->req_lock);
3589
		dev_err(DEV, "Aborting Connect, can not thaw IO with an only Consistent peer\n");
3590
		tl_clear(mdev->tconn);
3591 3592
		drbd_uuid_new_current(mdev);
		clear_bit(NEW_CUR_UUID, &mdev->flags);
3593
		conn_request_state(mdev->tconn, NS2(conn, C_PROTOCOL_ERROR, susp, 0), CS_HARD);
3594
		return -EIO;
3595
	}
3596
	rv = _drbd_set_state(mdev, ns, cs_flags, NULL);
3597
	ns = drbd_read_state(mdev);
3598
	spin_unlock_irq(&mdev->tconn->req_lock);
P
Philipp Reisner 已提交
3599 3600

	if (rv < SS_SUCCESS) {
3601
		conn_request_state(mdev->tconn, NS(conn, C_DISCONNECTING), CS_HARD);
3602
		return -EIO;
P
Philipp Reisner 已提交
3603 3604
	}

3605 3606
	if (os.conn > C_WF_REPORT_PARAMS) {
		if (ns.conn > C_CONNECTED && peer_state.conn <= C_CONNECTED &&
P
Philipp Reisner 已提交
3607 3608 3609 3610 3611 3612 3613 3614 3615
		    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);
		}
	}

3616
	mdev->tconn->net_conf->want_lose = 0;
P
Philipp Reisner 已提交
3617 3618 3619

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

3620
	return 0;
P
Philipp Reisner 已提交
3621 3622
}

3623
static int receive_sync_uuid(struct drbd_tconn *tconn, struct packet_info *pi)
P
Philipp Reisner 已提交
3624
{
3625 3626 3627 3628 3629 3630
	struct drbd_conf *mdev;
	struct p_rs_uuid *p = tconn->data.rbuf;

	mdev = vnr_to_mdev(tconn, pi->vnr);
	if (!mdev)
		return -EIO;
P
Philipp Reisner 已提交
3631 3632 3633

	wait_event(mdev->misc_wait,
		   mdev->state.conn == C_WF_SYNC_UUID ||
3634
		   mdev->state.conn == C_BEHIND ||
P
Philipp Reisner 已提交
3635 3636 3637 3638 3639 3640 3641 3642 3643 3644 3645
		   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);

3646
		drbd_print_uuids(mdev, "updated sync uuid");
P
Philipp Reisner 已提交
3647 3648 3649 3650 3651 3652
		drbd_start_resync(mdev, C_SYNC_TARGET);

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

3653
	return 0;
P
Philipp Reisner 已提交
3654 3655
}

3656 3657 3658 3659 3660 3661 3662
/**
 * receive_bitmap_plain
 *
 * Return 0 when done, 1 when another iteration is needed, and a negative error
 * code upon failure.
 */
static int
3663
receive_bitmap_plain(struct drbd_conf *mdev, unsigned int size,
3664
		     struct p_header *h, struct bm_xfer_ctx *c)
P
Philipp Reisner 已提交
3665
{
3666
	unsigned long *buffer = (unsigned long *)h->payload;
3667 3668 3669 3670 3671
	unsigned int data_size = DRBD_SOCKET_BUFFER_SIZE -
				 drbd_header_size(mdev->tconn);
	unsigned int num_words = min_t(size_t, data_size / sizeof(unsigned long),
				       c->bm_words - c->word_offset);
	unsigned int want = num_words * sizeof(unsigned long);
3672
	int err;
P
Philipp Reisner 已提交
3673

3674 3675
	if (want != size) {
		dev_err(DEV, "%s:want (%u) != size (%u)\n", __func__, want, size);
3676
		return -EIO;
P
Philipp Reisner 已提交
3677 3678
	}
	if (want == 0)
3679
		return 0;
3680 3681
	err = drbd_recv_all(mdev->tconn, buffer, want);
	if (err)
3682
		return err;
P
Philipp Reisner 已提交
3683 3684 3685 3686 3687 3688 3689 3690

	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;

3691
	return 1;
P
Philipp Reisner 已提交
3692 3693
}

3694 3695 3696 3697 3698 3699 3700 3701 3702 3703 3704 3705 3706 3707 3708
static enum drbd_bitmap_code dcbp_get_code(struct p_compressed_bm *p)
{
	return (enum drbd_bitmap_code)(p->encoding & 0x0f);
}

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

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

3709 3710 3711 3712 3713 3714 3715
/**
 * 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
3716 3717
recv_bm_rle_bits(struct drbd_conf *mdev,
		struct p_compressed_bm *p,
3718 3719
		 struct bm_xfer_ctx *c,
		 unsigned int len)
P
Philipp Reisner 已提交
3720 3721 3722 3723 3724 3725 3726
{
	struct bitstream bs;
	u64 look_ahead;
	u64 rl;
	u64 tmp;
	unsigned long s = c->bit_offset;
	unsigned long e;
3727
	int toggle = dcbp_get_start(p);
P
Philipp Reisner 已提交
3728 3729 3730
	int have;
	int bits;

3731
	bitstream_init(&bs, p->code, len, dcbp_get_pad_bits(p));
P
Philipp Reisner 已提交
3732 3733 3734

	bits = bitstream_get_bits(&bs, &look_ahead, 64);
	if (bits < 0)
3735
		return -EIO;
P
Philipp Reisner 已提交
3736 3737 3738 3739

	for (have = bits; have > 0; s += rl, toggle = !toggle) {
		bits = vli_decode_bits(&rl, look_ahead);
		if (bits <= 0)
3740
			return -EIO;
P
Philipp Reisner 已提交
3741 3742 3743 3744 3745

		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);
3746
				return -EIO;
P
Philipp Reisner 已提交
3747 3748 3749 3750 3751 3752 3753 3754 3755
			}
			_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);
3756
			return -EIO;
P
Philipp Reisner 已提交
3757 3758 3759 3760 3761 3762
		}
		look_ahead >>= bits;
		have -= bits;

		bits = bitstream_get_bits(&bs, &tmp, 64 - have);
		if (bits < 0)
3763
			return -EIO;
P
Philipp Reisner 已提交
3764 3765 3766 3767 3768 3769 3770
		look_ahead |= tmp << have;
		have += bits;
	}

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

3771
	return (s != c->bm_bits);
P
Philipp Reisner 已提交
3772 3773
}

3774 3775 3776 3777 3778 3779 3780
/**
 * 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
3781 3782
decode_bitmap_c(struct drbd_conf *mdev,
		struct p_compressed_bm *p,
3783 3784
		struct bm_xfer_ctx *c,
		unsigned int len)
P
Philipp Reisner 已提交
3785
{
3786
	if (dcbp_get_code(p) == RLE_VLI_Bits)
3787
		return recv_bm_rle_bits(mdev, p, c, len);
P
Philipp Reisner 已提交
3788 3789 3790 3791 3792 3793

	/* 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);
3794
	conn_request_state(mdev->tconn, NS(conn, C_PROTOCOL_ERROR), CS_HARD);
3795
	return -EIO;
P
Philipp Reisner 已提交
3796 3797 3798 3799 3800 3801
}

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" */
3802 3803 3804 3805 3806 3807 3808
	unsigned int header_size = drbd_header_size(mdev->tconn);
	unsigned int data_size = DRBD_SOCKET_BUFFER_SIZE - header_size;
	unsigned int plain =
		header_size * (DIV_ROUND_UP(c->bm_words, data_size) + 1) +
		c->bm_words * sizeof(unsigned long);
	unsigned int total = c->bytes[0] + c->bytes[1];
	unsigned int r;
P
Philipp Reisner 已提交
3809 3810 3811 3812 3813 3814 3815 3816 3817 3818 3819 3820 3821 3822 3823 3824 3825 3826 3827 3828 3829 3830 3831 3832 3833 3834 3835 3836 3837 3838 3839 3840 3841

	/* 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. */
3842
static int receive_bitmap(struct drbd_tconn *tconn, struct packet_info *pi)
P
Philipp Reisner 已提交
3843
{
3844
	struct drbd_conf *mdev;
P
Philipp Reisner 已提交
3845
	struct bm_xfer_ctx c;
3846
	int err;
3847 3848 3849 3850 3851
	struct p_header *h = tconn->data.rbuf;

	mdev = vnr_to_mdev(tconn, pi->vnr);
	if (!mdev)
		return -EIO;
P
Philipp Reisner 已提交
3852

3853 3854 3855
	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
3856 3857 3858 3859 3860 3861

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

3862
	for(;;) {
3863 3864 3865
		if (pi->cmd == P_BITMAP) {
			err = receive_bitmap_plain(mdev, pi->size, h, &c);
		} else if (pi->cmd == P_COMPRESSED_BITMAP) {
P
Philipp Reisner 已提交
3866 3867 3868 3869
			/* MAYBE: sanity check that we speak proto >= 90,
			 * and the feature is enabled! */
			struct p_compressed_bm *p;

3870
			if (pi->size > DRBD_SOCKET_BUFFER_SIZE - drbd_header_size(tconn)) {
P
Philipp Reisner 已提交
3871
				dev_err(DEV, "ReportCBitmap packet too large\n");
3872
				err = -EIO;
P
Philipp Reisner 已提交
3873 3874
				goto out;
			}
3875 3876

			p = mdev->tconn->data.rbuf;
3877
			err = drbd_recv_all(mdev->tconn, p->head.payload, pi->size);
3878 3879
			if (err)
			       goto out;
3880 3881
			if (pi->size <= (sizeof(*p) - sizeof(p->head))) {
				dev_err(DEV, "ReportCBitmap packet too small (l:%u)\n", pi->size);
3882
				err = -EIO;
3883
				goto out;
P
Philipp Reisner 已提交
3884
			}
3885
			err = decode_bitmap_c(mdev, p, &c, pi->size);
P
Philipp Reisner 已提交
3886
		} else {
3887
			dev_warn(DEV, "receive_bitmap: cmd neither ReportBitMap nor ReportCBitMap (is 0x%x)", pi->cmd);
3888
			err = -EIO;
P
Philipp Reisner 已提交
3889 3890 3891
			goto out;
		}

3892
		c.packets[pi->cmd == P_BITMAP]++;
3893
		c.bytes[pi->cmd == P_BITMAP] += drbd_header_size(tconn) + pi->size;
P
Philipp Reisner 已提交
3894

3895 3896 3897
		if (err <= 0) {
			if (err < 0)
				goto out;
P
Philipp Reisner 已提交
3898
			break;
3899
		}
3900
		err = drbd_recv_header(mdev->tconn, pi);
3901
		if (err)
P
Philipp Reisner 已提交
3902
			goto out;
3903
	}
P
Philipp Reisner 已提交
3904 3905 3906 3907

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

	if (mdev->state.conn == C_WF_BITMAP_T) {
3908 3909
		enum drbd_state_rv rv;

3910 3911
		err = drbd_send_bitmap(mdev);
		if (err)
P
Philipp Reisner 已提交
3912 3913
			goto out;
		/* Omit CS_ORDERED with this state transition to avoid deadlocks. */
3914 3915
		rv = _drbd_request_state(mdev, NS(conn, C_WF_SYNC_UUID), CS_VERBOSE);
		D_ASSERT(rv == SS_SUCCESS);
P
Philipp Reisner 已提交
3916 3917 3918 3919 3920 3921
	} 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));
	}
3922
	err = 0;
P
Philipp Reisner 已提交
3923 3924

 out:
3925
	drbd_bm_unlock(mdev);
3926
	if (!err && mdev->state.conn == C_WF_BITMAP_S)
P
Philipp Reisner 已提交
3927
		drbd_start_resync(mdev, C_SYNC_SOURCE);
3928
	return err;
P
Philipp Reisner 已提交
3929 3930
}

3931
static int receive_skip(struct drbd_tconn *tconn, struct packet_info *pi)
P
Philipp Reisner 已提交
3932
{
3933
	conn_warn(tconn, "skipping unknown optional packet type %d, l: %d!\n",
3934
		 pi->cmd, pi->size);
3935

3936
	return ignore_remaining_packet(tconn, pi);
3937 3938
}

3939
static int receive_UnplugRemote(struct drbd_tconn *tconn, struct packet_info *pi)
3940
{
3941 3942
	/* Make sure we've acked all the TCP data associated
	 * with the data requests being unplugged */
3943
	drbd_tcp_quickack(tconn->data.socket);
3944

3945
	return 0;
3946 3947
}

3948
static int receive_out_of_sync(struct drbd_tconn *tconn, struct packet_info *pi)
3949
{
3950 3951 3952 3953 3954 3955
	struct drbd_conf *mdev;
	struct p_block_desc *p = tconn->data.rbuf;

	mdev = vnr_to_mdev(tconn, pi->vnr);
	if (!mdev)
		return -EIO;
3956

3957 3958 3959 3960 3961 3962 3963 3964 3965 3966
	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));
	}

3967 3968
	drbd_set_out_of_sync(mdev, be64_to_cpu(p->sector), be32_to_cpu(p->blksize));

3969
	return 0;
3970 3971
}

3972 3973 3974
struct data_cmd {
	int expect_payload;
	size_t pkt_size;
3975
	int (*fn)(struct drbd_tconn *, struct packet_info *);
3976 3977 3978
};

static struct data_cmd drbd_cmd_handler[] = {
3979 3980 3981 3982 3983 3984 3985 3986 3987 3988 3989 3990 3991 3992 3993 3994 3995 3996 3997 3998 3999 4000 4001
	[P_DATA]	    = { 1, sizeof(struct p_data), receive_Data },
	[P_DATA_REPLY]	    = { 1, sizeof(struct p_data), receive_DataReply },
	[P_RS_DATA_REPLY]   = { 1, sizeof(struct p_data), receive_RSDataReply } ,
	[P_BARRIER]	    = { 0, sizeof(struct p_barrier), receive_Barrier } ,
	[P_BITMAP]	    = { 1, sizeof(struct p_header), receive_bitmap } ,
	[P_COMPRESSED_BITMAP] = { 1, sizeof(struct p_header), receive_bitmap } ,
	[P_UNPLUG_REMOTE]   = { 0, sizeof(struct p_header), receive_UnplugRemote },
	[P_DATA_REQUEST]    = { 0, sizeof(struct p_block_req), receive_DataRequest },
	[P_RS_DATA_REQUEST] = { 0, sizeof(struct p_block_req), receive_DataRequest },
	[P_SYNC_PARAM]	    = { 1, sizeof(struct p_header), receive_SyncParam },
	[P_SYNC_PARAM89]    = { 1, sizeof(struct p_header), receive_SyncParam },
	[P_PROTOCOL]        = { 1, sizeof(struct p_protocol), receive_protocol },
	[P_UUIDS]	    = { 0, sizeof(struct p_uuids), receive_uuids },
	[P_SIZES]	    = { 0, sizeof(struct p_sizes), receive_sizes },
	[P_STATE]	    = { 0, sizeof(struct p_state), receive_state },
	[P_STATE_CHG_REQ]   = { 0, sizeof(struct p_req_state), receive_req_state },
	[P_SYNC_UUID]       = { 0, sizeof(struct p_rs_uuid), receive_sync_uuid },
	[P_OV_REQUEST]      = { 0, sizeof(struct p_block_req), receive_DataRequest },
	[P_OV_REPLY]        = { 1, sizeof(struct p_block_req), receive_DataRequest },
	[P_CSUM_RS_REQUEST] = { 1, sizeof(struct p_block_req), receive_DataRequest },
	[P_DELAY_PROBE]     = { 0, sizeof(struct p_delay_probe93), receive_skip },
	[P_OUT_OF_SYNC]     = { 0, sizeof(struct p_block_desc), receive_out_of_sync },
	[P_CONN_ST_CHG_REQ] = { 0, sizeof(struct p_req_state), receive_req_conn_state },
P
Philipp Reisner 已提交
4002 4003
};

4004
static void drbdd(struct drbd_tconn *tconn)
P
Philipp Reisner 已提交
4005
{
4006
	struct p_header *header = tconn->data.rbuf;
4007
	struct packet_info pi;
4008
	size_t shs; /* sub header size */
4009
	int err;
P
Philipp Reisner 已提交
4010

4011
	while (get_t_state(&tconn->receiver) == RUNNING) {
4012 4013
		struct data_cmd *cmd;

4014
		drbd_thread_current_set_cpu(&tconn->receiver);
4015
		if (drbd_recv_header(tconn, &pi))
4016
			goto err_out;
P
Philipp Reisner 已提交
4017

4018
		cmd = &drbd_cmd_handler[pi.cmd];
4019
		if (unlikely(pi.cmd >= ARRAY_SIZE(drbd_cmd_handler) || !cmd->fn)) {
4020
			conn_err(tconn, "unknown packet type %d, l: %d!\n", pi.cmd, pi.size);
4021
			goto err_out;
4022
		}
P
Philipp Reisner 已提交
4023

4024 4025
		shs = cmd->pkt_size - sizeof(struct p_header);
		if (pi.size - shs > 0 && !cmd->expect_payload) {
4026
			conn_err(tconn, "No payload expected %s l:%d\n", cmdname(pi.cmd), pi.size);
4027
			goto err_out;
P
Philipp Reisner 已提交
4028 4029
		}

4030
		if (shs) {
4031 4032
			err = drbd_recv_all_warn(tconn, &header->payload, shs);
			if (err)
4033
				goto err_out;
4034
			pi.size -= shs;
4035 4036
		}

4037 4038
		err = cmd->fn(tconn, &pi);
		if (err) {
4039 4040
			conn_err(tconn, "error receiving %s, e: %d l: %d!\n",
				 cmdname(pi.cmd), err, pi.size);
4041
			goto err_out;
P
Philipp Reisner 已提交
4042 4043
		}
	}
4044
	return;
P
Philipp Reisner 已提交
4045

4046 4047
    err_out:
	conn_request_state(tconn, NS(conn, C_PROTOCOL_ERROR), CS_HARD);
P
Philipp Reisner 已提交
4048 4049
}

4050
void conn_flush_workqueue(struct drbd_tconn *tconn)
P
Philipp Reisner 已提交
4051 4052 4053 4054
{
	struct drbd_wq_barrier barr;

	barr.w.cb = w_prev_work_done;
4055
	barr.w.tconn = tconn;
P
Philipp Reisner 已提交
4056
	init_completion(&barr.done);
4057
	drbd_queue_work(&tconn->data.work, &barr.w);
P
Philipp Reisner 已提交
4058 4059 4060
	wait_for_completion(&barr.done);
}

4061
static void drbd_disconnect(struct drbd_tconn *tconn)
P
Philipp Reisner 已提交
4062
{
4063
	enum drbd_conns oc;
P
Philipp Reisner 已提交
4064 4065
	int rv = SS_UNKNOWN_ERROR;

4066
	if (tconn->cstate == C_STANDALONE)
P
Philipp Reisner 已提交
4067 4068 4069
		return;

	/* asender does not clean up anything. it must not interfere, either */
4070 4071 4072 4073 4074 4075
	drbd_thread_stop(&tconn->asender);
	drbd_free_sock(tconn);

	idr_for_each(&tconn->volumes, drbd_disconnected, tconn);
	conn_info(tconn, "Connection closed\n");

4076 4077 4078
	if (conn_highest_role(tconn) == R_PRIMARY && conn_highest_pdsk(tconn) >= D_UNKNOWN)
		conn_try_outdate_peer_async(tconn);

4079
	spin_lock_irq(&tconn->req_lock);
4080 4081 4082 4083
	oc = tconn->cstate;
	if (oc >= C_UNCONNECTED)
		rv = _conn_request_state(tconn, NS(conn, C_UNCONNECTED), CS_VERBOSE);

4084 4085
	spin_unlock_irq(&tconn->req_lock);

4086
	if (oc == C_DISCONNECTING) {
4087 4088 4089 4090 4091 4092 4093
		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;
4094
		conn_request_state(tconn, NS(conn, C_STANDALONE), CS_VERBOSE);
4095 4096 4097 4098 4099 4100 4101 4102
	}
}

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
4103

4104
	/* wait for current activity to cease. */
4105
	spin_lock_irq(&mdev->tconn->req_lock);
P
Philipp Reisner 已提交
4106 4107 4108
	_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);
4109
	spin_unlock_irq(&mdev->tconn->req_lock);
P
Philipp Reisner 已提交
4110 4111 4112 4113 4114 4115 4116 4117 4118 4119 4120 4121 4122 4123 4124 4125 4126

	/* 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);

4127 4128
	del_timer(&mdev->request_timer);

P
Philipp Reisner 已提交
4129 4130 4131 4132 4133 4134
	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" */
4135
	drbd_flush_workqueue(mdev);
P
Philipp Reisner 已提交
4136 4137 4138 4139 4140 4141 4142 4143

	/* 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;

4144
	if (!drbd_suspended(mdev))
4145
		tl_clear(mdev->tconn);
P
Philipp Reisner 已提交
4146 4147 4148 4149 4150 4151 4152 4153 4154

	drbd_md_sync(mdev);

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

4155 4156 4157 4158
	/* 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
4159 4160 4161 4162 4163 4164 4165 4166 4167 4168
	/* 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);
4169 4170 4171
	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
4172 4173
	i = atomic_read(&mdev->pp_in_use);
	if (i)
4174
		dev_info(DEV, "pp_in_use = %d, expected 0\n", i);
P
Philipp Reisner 已提交
4175 4176 4177 4178 4179 4180 4181 4182 4183

	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));
4184 4185

	return 0;
P
Philipp Reisner 已提交
4186 4187 4188 4189 4190 4191 4192 4193 4194 4195 4196
}

/*
 * 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.
 */
4197
static int drbd_send_features(struct drbd_tconn *tconn)
P
Philipp Reisner 已提交
4198
{
4199 4200
	struct drbd_socket *sock;
	struct p_connection_features *p;
P
Philipp Reisner 已提交
4201

4202 4203 4204
	sock = &tconn->data;
	p = conn_prepare_command(tconn, sock);
	if (!p)
4205
		return -EIO;
P
Philipp Reisner 已提交
4206 4207 4208
	memset(p, 0, sizeof(*p));
	p->protocol_min = cpu_to_be32(PRO_VERSION_MIN);
	p->protocol_max = cpu_to_be32(PRO_VERSION_MAX);
4209
	return conn_send_command(tconn, sock, P_CONNECTION_FEATURES, sizeof(*p), NULL, 0);
P
Philipp Reisner 已提交
4210 4211 4212 4213 4214 4215 4216 4217 4218
}

/*
 * 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.
 */
4219
static int drbd_do_features(struct drbd_tconn *tconn)
P
Philipp Reisner 已提交
4220
{
4221
	/* ASSERT current == tconn->receiver ... */
4222 4223
	struct p_connection_features *p = tconn->data.rbuf;
	const int expect = sizeof(struct p_connection_features) - sizeof(struct p_header80);
4224
	struct packet_info pi;
4225
	int err;
P
Philipp Reisner 已提交
4226

4227
	err = drbd_send_features(tconn);
4228
	if (err)
P
Philipp Reisner 已提交
4229 4230
		return 0;

4231 4232
	err = drbd_recv_header(tconn, &pi);
	if (err)
P
Philipp Reisner 已提交
4233 4234
		return 0;

4235 4236
	if (pi.cmd != P_CONNECTION_FEATURES) {
		conn_err(tconn, "expected ConnectionFeatures packet, received: %s (0x%04x)\n",
4237
		     cmdname(pi.cmd), pi.cmd);
P
Philipp Reisner 已提交
4238 4239 4240
		return -1;
	}

4241
	if (pi.size != expect) {
4242
		conn_err(tconn, "expected ConnectionFeatures length: %u, received: %u\n",
4243
		     expect, pi.size);
P
Philipp Reisner 已提交
4244 4245 4246
		return -1;
	}

4247 4248
	err = drbd_recv_all_warn(tconn, &p->head.payload, expect);
	if (err)
P
Philipp Reisner 已提交
4249 4250 4251 4252 4253 4254 4255 4256 4257 4258 4259
		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;

4260
	tconn->agreed_pro_version = min_t(int, PRO_VERSION_MAX, p->protocol_max);
P
Philipp Reisner 已提交
4261

4262 4263
	conn_info(tconn, "Handshake successful: "
	     "Agreed network protocol version %d\n", tconn->agreed_pro_version);
P
Philipp Reisner 已提交
4264 4265 4266 4267

	return 1;

 incompat:
4268
	conn_err(tconn, "incompatible DRBD dialects: "
P
Philipp Reisner 已提交
4269 4270 4271 4272 4273 4274 4275
	    "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)
4276
static int drbd_do_auth(struct drbd_tconn *tconn)
P
Philipp Reisner 已提交
4277 4278 4279
{
	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");
4280
	return -1;
P
Philipp Reisner 已提交
4281 4282 4283
}
#else
#define CHALLENGE_LEN 64
4284 4285 4286 4287 4288 4289 4290

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

4291
static int drbd_do_auth(struct drbd_tconn *tconn)
P
Philipp Reisner 已提交
4292
{
4293
	struct drbd_socket *sock;
P
Philipp Reisner 已提交
4294 4295 4296 4297 4298
	char my_challenge[CHALLENGE_LEN];  /* 64 Bytes... */
	struct scatterlist sg;
	char *response = NULL;
	char *right_response = NULL;
	char *peers_ch = NULL;
4299
	unsigned int key_len = strlen(tconn->net_conf->shared_secret);
P
Philipp Reisner 已提交
4300 4301
	unsigned int resp_size;
	struct hash_desc desc;
4302
	struct packet_info pi;
4303
	int err, rv;
P
Philipp Reisner 已提交
4304

4305 4306
	/* FIXME: Put the challenge/response into the preallocated socket buffer.  */

4307
	desc.tfm = tconn->cram_hmac_tfm;
P
Philipp Reisner 已提交
4308 4309
	desc.flags = 0;

4310 4311
	rv = crypto_hash_setkey(tconn->cram_hmac_tfm,
				(u8 *)tconn->net_conf->shared_secret, key_len);
P
Philipp Reisner 已提交
4312
	if (rv) {
4313
		conn_err(tconn, "crypto_hash_setkey() failed with %d\n", rv);
4314
		rv = -1;
P
Philipp Reisner 已提交
4315 4316 4317 4318 4319
		goto fail;
	}

	get_random_bytes(my_challenge, CHALLENGE_LEN);

4320 4321 4322 4323 4324 4325 4326 4327
	sock = &tconn->data;
	if (!conn_prepare_command(tconn, sock)) {
		rv = 0;
		goto fail;
	}
	rv = !conn_send_command(tconn, sock, P_AUTH_CHALLENGE,
				sizeof(struct p_header),
				my_challenge, CHALLENGE_LEN);
P
Philipp Reisner 已提交
4328 4329 4330
	if (!rv)
		goto fail;

4331 4332 4333
	err = drbd_recv_header(tconn, &pi);
	if (err) {
		rv = 0;
P
Philipp Reisner 已提交
4334
		goto fail;
4335
	}
P
Philipp Reisner 已提交
4336

4337
	if (pi.cmd != P_AUTH_CHALLENGE) {
4338
		conn_err(tconn, "expected AuthChallenge packet, received: %s (0x%04x)\n",
4339
		    cmdname(pi.cmd), pi.cmd);
P
Philipp Reisner 已提交
4340 4341 4342 4343
		rv = 0;
		goto fail;
	}

4344
	if (pi.size > CHALLENGE_LEN * 2) {
4345
		conn_err(tconn, "expected AuthChallenge payload too big.\n");
4346
		rv = -1;
P
Philipp Reisner 已提交
4347 4348 4349
		goto fail;
	}

4350
	peers_ch = kmalloc(pi.size, GFP_NOIO);
P
Philipp Reisner 已提交
4351
	if (peers_ch == NULL) {
4352
		conn_err(tconn, "kmalloc of peers_ch failed\n");
4353
		rv = -1;
P
Philipp Reisner 已提交
4354 4355 4356
		goto fail;
	}

4357 4358
	err = drbd_recv_all_warn(tconn, peers_ch, pi.size);
	if (err) {
P
Philipp Reisner 已提交
4359 4360 4361 4362
		rv = 0;
		goto fail;
	}

4363
	resp_size = crypto_hash_digestsize(tconn->cram_hmac_tfm);
P
Philipp Reisner 已提交
4364 4365
	response = kmalloc(resp_size, GFP_NOIO);
	if (response == NULL) {
4366
		conn_err(tconn, "kmalloc of response failed\n");
4367
		rv = -1;
P
Philipp Reisner 已提交
4368 4369 4370 4371
		goto fail;
	}

	sg_init_table(&sg, 1);
4372
	sg_set_buf(&sg, peers_ch, pi.size);
P
Philipp Reisner 已提交
4373 4374 4375

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

4381 4382 4383 4384 4385 4386 4387
	if (!conn_prepare_command(tconn, sock)) {
		rv = 0;
		goto fail;
	}
	rv = !conn_send_command(tconn, sock, P_AUTH_RESPONSE,
				sizeof(struct p_header),
				response, resp_size);
P
Philipp Reisner 已提交
4388 4389 4390
	if (!rv)
		goto fail;

4391 4392 4393
	err = drbd_recv_header(tconn, &pi);
	if (err) {
		rv = 0;
P
Philipp Reisner 已提交
4394
		goto fail;
4395
	}
P
Philipp Reisner 已提交
4396

4397
	if (pi.cmd != P_AUTH_RESPONSE) {
4398
		conn_err(tconn, "expected AuthResponse packet, received: %s (0x%04x)\n",
4399
			cmdname(pi.cmd), pi.cmd);
P
Philipp Reisner 已提交
4400 4401 4402 4403
		rv = 0;
		goto fail;
	}

4404
	if (pi.size != resp_size) {
4405
		conn_err(tconn, "expected AuthResponse payload of wrong size\n");
P
Philipp Reisner 已提交
4406 4407 4408 4409
		rv = 0;
		goto fail;
	}

4410 4411
	err = drbd_recv_all_warn(tconn, response , resp_size);
	if (err) {
P
Philipp Reisner 已提交
4412 4413 4414 4415 4416
		rv = 0;
		goto fail;
	}

	right_response = kmalloc(resp_size, GFP_NOIO);
4417
	if (right_response == NULL) {
4418
		conn_err(tconn, "kmalloc of right_response failed\n");
4419
		rv = -1;
P
Philipp Reisner 已提交
4420 4421 4422 4423 4424 4425 4426
		goto fail;
	}

	sg_set_buf(&sg, my_challenge, CHALLENGE_LEN);

	rv = crypto_hash_digest(&desc, &sg, sg.length, right_response);
	if (rv) {
4427
		conn_err(tconn, "crypto_hash_digest() failed with %d\n", rv);
4428
		rv = -1;
P
Philipp Reisner 已提交
4429 4430 4431 4432 4433 4434
		goto fail;
	}

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

	if (rv)
4435 4436
		conn_info(tconn, "Peer authenticated using %d bytes of '%s' HMAC\n",
		     resp_size, tconn->net_conf->cram_hmac_alg);
4437 4438
	else
		rv = -1;
P
Philipp Reisner 已提交
4439 4440 4441 4442 4443 4444 4445 4446 4447 4448 4449 4450

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

	return rv;
}
#endif

int drbdd_init(struct drbd_thread *thi)
{
4451
	struct drbd_tconn *tconn = thi->tconn;
P
Philipp Reisner 已提交
4452 4453
	int h;

4454
	conn_info(tconn, "receiver (re)started\n");
P
Philipp Reisner 已提交
4455 4456

	do {
4457
		h = drbd_connect(tconn);
P
Philipp Reisner 已提交
4458
		if (h == 0) {
4459
			drbd_disconnect(tconn);
4460
			schedule_timeout_interruptible(HZ);
P
Philipp Reisner 已提交
4461 4462
		}
		if (h == -1) {
4463
			conn_warn(tconn, "Discarding network configuration.\n");
4464
			conn_request_state(tconn, NS(conn, C_DISCONNECTING), CS_HARD);
P
Philipp Reisner 已提交
4465 4466 4467 4468
		}
	} while (h == 0);

	if (h > 0) {
4469 4470 4471
		if (get_net_conf(tconn)) {
			drbdd(tconn);
			put_net_conf(tconn);
P
Philipp Reisner 已提交
4472 4473 4474
		}
	}

4475
	drbd_disconnect(tconn);
P
Philipp Reisner 已提交
4476

4477
	conn_info(tconn, "receiver terminated\n");
P
Philipp Reisner 已提交
4478 4479 4480 4481 4482
	return 0;
}

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

4483
static int got_conn_RqSReply(struct drbd_tconn *tconn, struct packet_info *pi)
4484
{
4485
	struct p_req_state_reply *p = tconn->meta.rbuf;
4486 4487 4488 4489 4490 4491 4492 4493 4494 4495 4496 4497 4498 4499
	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;
}

4500
static int got_RqSReply(struct drbd_tconn *tconn, struct packet_info *pi)
P
Philipp Reisner 已提交
4501
{
4502 4503
	struct drbd_conf *mdev;
	struct p_req_state_reply *p = tconn->meta.rbuf;
P
Philipp Reisner 已提交
4504 4505
	int retcode = be32_to_cpu(p->retcode);

4506 4507 4508 4509
	mdev = vnr_to_mdev(tconn, pi->vnr);
	if (!mdev)
		return false;

4510 4511 4512 4513 4514 4515
	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
4516
	}
4517 4518
	wake_up(&mdev->state_wait);

4519
	return true;
P
Philipp Reisner 已提交
4520 4521
}

4522
static int got_Ping(struct drbd_tconn *tconn, struct packet_info *pi)
P
Philipp Reisner 已提交
4523
{
4524
	return !drbd_send_ping_ack(tconn);
P
Philipp Reisner 已提交
4525 4526 4527

}

4528
static int got_PingAck(struct drbd_tconn *tconn, struct packet_info *pi)
P
Philipp Reisner 已提交
4529 4530
{
	/* restore idle timeout */
4531 4532 4533
	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
4534

4535
	return true;
P
Philipp Reisner 已提交
4536 4537
}

4538
static int got_IsInSync(struct drbd_tconn *tconn, struct packet_info *pi)
P
Philipp Reisner 已提交
4539
{
4540 4541
	struct drbd_conf *mdev;
	struct p_block_ack *p = tconn->meta.rbuf;
P
Philipp Reisner 已提交
4542 4543 4544
	sector_t sector = be64_to_cpu(p->sector);
	int blksize = be32_to_cpu(p->blksize);

4545 4546 4547 4548
	mdev = vnr_to_mdev(tconn, pi->vnr);
	if (!mdev)
		return false;

4549
	D_ASSERT(mdev->tconn->agreed_pro_version >= 89);
P
Philipp Reisner 已提交
4550 4551 4552

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

4553 4554 4555 4556 4557 4558 4559
	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
4560
	dec_rs_pending(mdev);
4561
	atomic_add(blksize >> 9, &mdev->rs_sect_in);
P
Philipp Reisner 已提交
4562

4563
	return true;
P
Philipp Reisner 已提交
4564 4565
}

4566 4567 4568 4569
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
4570 4571 4572 4573
{
	struct drbd_request *req;
	struct bio_and_error m;

4574
	spin_lock_irq(&mdev->tconn->req_lock);
4575
	req = find_request(mdev, root, id, sector, missing_ok, func);
P
Philipp Reisner 已提交
4576
	if (unlikely(!req)) {
4577
		spin_unlock_irq(&mdev->tconn->req_lock);
4578
		return false;
P
Philipp Reisner 已提交
4579 4580
	}
	__req_mod(req, what, &m);
4581
	spin_unlock_irq(&mdev->tconn->req_lock);
P
Philipp Reisner 已提交
4582 4583 4584

	if (m.bio)
		complete_master_bio(mdev, &m);
4585
	return true;
P
Philipp Reisner 已提交
4586 4587
}

4588
static int got_BlockAck(struct drbd_tconn *tconn, struct packet_info *pi)
P
Philipp Reisner 已提交
4589
{
4590 4591
	struct drbd_conf *mdev;
	struct p_block_ack *p = tconn->meta.rbuf;
P
Philipp Reisner 已提交
4592 4593 4594 4595
	sector_t sector = be64_to_cpu(p->sector);
	int blksize = be32_to_cpu(p->blksize);
	enum drbd_req_event what;

4596 4597 4598 4599
	mdev = vnr_to_mdev(tconn, pi->vnr);
	if (!mdev)
		return false;

P
Philipp Reisner 已提交
4600 4601
	update_peer_seq(mdev, be32_to_cpu(p->seq_num));

4602
	if (p->block_id == ID_SYNCER) {
P
Philipp Reisner 已提交
4603 4604
		drbd_set_in_sync(mdev, sector, blksize);
		dec_rs_pending(mdev);
4605
		return true;
P
Philipp Reisner 已提交
4606
	}
4607
	switch (pi->cmd) {
P
Philipp Reisner 已提交
4608
	case P_RS_WRITE_ACK:
4609
		D_ASSERT(mdev->tconn->net_conf->wire_protocol == DRBD_PROT_C);
4610
		what = WRITE_ACKED_BY_PEER_AND_SIS;
P
Philipp Reisner 已提交
4611 4612
		break;
	case P_WRITE_ACK:
4613
		D_ASSERT(mdev->tconn->net_conf->wire_protocol == DRBD_PROT_C);
4614
		what = WRITE_ACKED_BY_PEER;
P
Philipp Reisner 已提交
4615 4616
		break;
	case P_RECV_ACK:
4617
		D_ASSERT(mdev->tconn->net_conf->wire_protocol == DRBD_PROT_B);
4618
		what = RECV_ACKED_BY_PEER;
P
Philipp Reisner 已提交
4619
		break;
4620
	case P_DISCARD_WRITE:
4621
		D_ASSERT(mdev->tconn->net_conf->wire_protocol == DRBD_PROT_C);
4622 4623 4624 4625 4626
		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
4627 4628 4629
		break;
	default:
		D_ASSERT(0);
4630
		return false;
P
Philipp Reisner 已提交
4631 4632 4633
	}

	return validate_req_change_req_state(mdev, p->block_id, sector,
4634 4635
					     &mdev->write_requests, __func__,
					     what, false);
P
Philipp Reisner 已提交
4636 4637
}

4638
static int got_NegAck(struct drbd_tconn *tconn, struct packet_info *pi)
P
Philipp Reisner 已提交
4639
{
4640 4641
	struct drbd_conf *mdev;
	struct p_block_ack *p = tconn->meta.rbuf;
P
Philipp Reisner 已提交
4642
	sector_t sector = be64_to_cpu(p->sector);
4643
	int size = be32_to_cpu(p->blksize);
4644 4645
	bool missing_ok = tconn->net_conf->wire_protocol == DRBD_PROT_A ||
			  tconn->net_conf->wire_protocol == DRBD_PROT_B;
4646
	bool found;
P
Philipp Reisner 已提交
4647

4648 4649 4650 4651
	mdev = vnr_to_mdev(tconn, pi->vnr);
	if (!mdev)
		return false;

P
Philipp Reisner 已提交
4652 4653
	update_peer_seq(mdev, be32_to_cpu(p->seq_num));

4654
	if (p->block_id == ID_SYNCER) {
P
Philipp Reisner 已提交
4655 4656
		dec_rs_pending(mdev);
		drbd_rs_failed_io(mdev, sector, size);
4657
		return true;
P
Philipp Reisner 已提交
4658
	}
4659

4660
	found = validate_req_change_req_state(mdev, p->block_id, sector,
4661
					      &mdev->write_requests, __func__,
4662
					      NEG_ACKED, missing_ok);
4663 4664 4665 4666 4667 4668 4669
	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)
4670
			return false;
4671
		drbd_set_out_of_sync(mdev, sector, size);
4672 4673
	}
	return true;
P
Philipp Reisner 已提交
4674 4675
}

4676
static int got_NegDReply(struct drbd_tconn *tconn, struct packet_info *pi)
P
Philipp Reisner 已提交
4677
{
4678 4679
	struct drbd_conf *mdev;
	struct p_block_ack *p = tconn->meta.rbuf;
P
Philipp Reisner 已提交
4680 4681
	sector_t sector = be64_to_cpu(p->sector);

4682 4683 4684 4685
	mdev = vnr_to_mdev(tconn, pi->vnr);
	if (!mdev)
		return false;

P
Philipp Reisner 已提交
4686
	update_peer_seq(mdev, be32_to_cpu(p->seq_num));
4687

P
Philipp Reisner 已提交
4688 4689 4690 4691
	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,
4692
					     &mdev->read_requests, __func__,
4693
					     NEG_ACKED, false);
P
Philipp Reisner 已提交
4694 4695
}

4696
static int got_NegRSDReply(struct drbd_tconn *tconn, struct packet_info *pi)
P
Philipp Reisner 已提交
4697
{
4698
	struct drbd_conf *mdev;
P
Philipp Reisner 已提交
4699 4700
	sector_t sector;
	int size;
4701 4702 4703 4704 4705
	struct p_block_ack *p = tconn->meta.rbuf;

	mdev = vnr_to_mdev(tconn, pi->vnr);
	if (!mdev)
		return false;
P
Philipp Reisner 已提交
4706 4707 4708 4709 4710 4711 4712 4713 4714 4715

	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);
4716
		switch (pi->cmd) {
4717 4718 4719 4720 4721 4722 4723 4724 4725
		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
4726 4727 4728
		put_ldev(mdev);
	}

4729
	return true;
P
Philipp Reisner 已提交
4730 4731
}

4732
static int got_BarrierAck(struct drbd_tconn *tconn, struct packet_info *pi)
P
Philipp Reisner 已提交
4733
{
4734 4735 4736 4737 4738 4739
	struct drbd_conf *mdev;
	struct p_barrier_ack *p = tconn->meta.rbuf;

	mdev = vnr_to_mdev(tconn, pi->vnr);
	if (!mdev)
		return false;
P
Philipp Reisner 已提交
4740

4741
	tl_release(mdev->tconn, p->barrier, be32_to_cpu(p->set_size));
P
Philipp Reisner 已提交
4742

4743 4744
	if (mdev->state.conn == C_AHEAD &&
	    atomic_read(&mdev->ap_in_flight) == 0 &&
4745 4746 4747
	    !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);
4748 4749
	}

4750
	return true;
P
Philipp Reisner 已提交
4751 4752
}

4753
static int got_OVResult(struct drbd_tconn *tconn, struct packet_info *pi)
P
Philipp Reisner 已提交
4754
{
4755 4756
	struct drbd_conf *mdev;
	struct p_block_ack *p = tconn->meta.rbuf;
P
Philipp Reisner 已提交
4757 4758 4759 4760
	struct drbd_work *w;
	sector_t sector;
	int size;

4761 4762 4763 4764
	mdev = vnr_to_mdev(tconn, pi->vnr);
	if (!mdev)
		return false;

P
Philipp Reisner 已提交
4765 4766 4767 4768 4769 4770
	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)
4771
		drbd_ov_out_of_sync_found(mdev, sector, size);
P
Philipp Reisner 已提交
4772
	else
4773
		ov_out_of_sync_print(mdev);
P
Philipp Reisner 已提交
4774

4775
	if (!get_ldev(mdev))
4776
		return true;
4777

P
Philipp Reisner 已提交
4778 4779 4780
	drbd_rs_complete_io(mdev, sector);
	dec_rs_pending(mdev);

4781 4782 4783 4784 4785 4786 4787
	--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
4788 4789 4790
		w = kmalloc(sizeof(*w), GFP_NOIO);
		if (w) {
			w->cb = w_ov_finished;
4791
			w->mdev = mdev;
4792
			drbd_queue_work_front(&mdev->tconn->data.work, w);
P
Philipp Reisner 已提交
4793 4794
		} else {
			dev_err(DEV, "kmalloc(w) failed.");
4795
			ov_out_of_sync_print(mdev);
P
Philipp Reisner 已提交
4796 4797 4798
			drbd_resync_finished(mdev);
		}
	}
4799
	put_ldev(mdev);
4800
	return true;
P
Philipp Reisner 已提交
4801 4802
}

4803
static int got_skip(struct drbd_tconn *tconn, struct packet_info *pi)
4804
{
4805
	return true;
4806 4807
}

4808 4809
static int tconn_process_done_ee(struct drbd_tconn *tconn)
{
4810 4811
	struct drbd_conf *mdev;
	int i, not_empty = 0;
4812 4813 4814 4815

	do {
		clear_bit(SIGNAL_ASENDER, &tconn->flags);
		flush_signals(current);
4816
		idr_for_each_entry(&tconn->volumes, mdev, i) {
4817
			if (drbd_process_done_ee(mdev))
4818 4819
				return 1; /* error */
		}
4820
		set_bit(SIGNAL_ASENDER, &tconn->flags);
4821 4822 4823 4824 4825 4826 4827 4828

		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);
4829 4830 4831 4832 4833
	} while (not_empty);

	return 0;
}

4834 4835
struct asender_cmd {
	size_t pkt_size;
4836
	int (*fn)(struct drbd_tconn *tconn, struct packet_info *);
4837 4838 4839
};

static struct asender_cmd asender_tbl[] = {
4840 4841 4842 4843 4844 4845 4846 4847 4848 4849 4850 4851 4852 4853 4854 4855 4856
	[P_PING]	    = { sizeof(struct p_header), got_Ping },
	[P_PING_ACK]	    = { sizeof(struct p_header), got_PingAck },
	[P_RECV_ACK]	    = { sizeof(struct p_block_ack), got_BlockAck },
	[P_WRITE_ACK]	    = { sizeof(struct p_block_ack), got_BlockAck },
	[P_RS_WRITE_ACK]    = { sizeof(struct p_block_ack), got_BlockAck },
	[P_DISCARD_WRITE]   = { sizeof(struct p_block_ack), got_BlockAck },
	[P_NEG_ACK]	    = { sizeof(struct p_block_ack), got_NegAck },
	[P_NEG_DREPLY]	    = { sizeof(struct p_block_ack), got_NegDReply },
	[P_NEG_RS_DREPLY]   = { sizeof(struct p_block_ack), got_NegRSDReply },
	[P_OV_RESULT]	    = { sizeof(struct p_block_ack), got_OVResult },
	[P_BARRIER_ACK]	    = { sizeof(struct p_barrier_ack), got_BarrierAck },
	[P_STATE_CHG_REPLY] = { sizeof(struct p_req_state_reply), got_RqSReply },
	[P_RS_IS_IN_SYNC]   = { sizeof(struct p_block_ack), got_IsInSync },
	[P_DELAY_PROBE]     = { sizeof(struct p_delay_probe93), got_skip },
	[P_RS_CANCEL]       = { sizeof(struct p_block_ack), got_NegRSDReply },
	[P_CONN_ST_CHG_REPLY]={ sizeof(struct p_req_state_reply), got_conn_RqSReply },
	[P_RETRY_WRITE]	    = { sizeof(struct p_block_ack), got_BlockAck },
4857 4858
};

P
Philipp Reisner 已提交
4859 4860
int drbd_asender(struct drbd_thread *thi)
{
4861
	struct drbd_tconn *tconn = thi->tconn;
4862
	struct p_header *h = tconn->meta.rbuf;
P
Philipp Reisner 已提交
4863
	struct asender_cmd *cmd = NULL;
4864
	struct packet_info pi;
4865
	int rv;
P
Philipp Reisner 已提交
4866 4867
	void *buf    = h;
	int received = 0;
4868 4869
	unsigned int header_size = drbd_header_size(tconn);
	int expect   = header_size;
4870
	int ping_timeout_active = 0;
P
Philipp Reisner 已提交
4871 4872 4873 4874

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

4875
	while (get_t_state(thi) == RUNNING) {
4876
		drbd_thread_current_set_cpu(thi);
4877
		if (test_and_clear_bit(SEND_PING, &tconn->flags)) {
4878
			if (drbd_send_ping(tconn)) {
4879
				conn_err(tconn, "drbd_send_ping has failed\n");
4880 4881
				goto reconnect;
			}
4882 4883
			tconn->meta.socket->sk->sk_rcvtimeo =
				tconn->net_conf->ping_timeo*HZ/10;
4884
			ping_timeout_active = 1;
P
Philipp Reisner 已提交
4885 4886
		}

4887 4888 4889 4890
		/* 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);
4891 4892
		if (tconn_process_done_ee(tconn)) {
			conn_err(tconn, "tconn_process_done_ee() failed\n");
4893
			goto reconnect;
4894
		}
P
Philipp Reisner 已提交
4895
		/* but unconditionally uncork unless disabled */
4896 4897
		if (!tconn->net_conf->no_cork)
			drbd_tcp_uncork(tconn->meta.socket);
P
Philipp Reisner 已提交
4898 4899 4900 4901 4902

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

4903 4904
		rv = drbd_recv_short(tconn->meta.socket, buf, expect-received, 0);
		clear_bit(SIGNAL_ASENDER, &tconn->flags);
P
Philipp Reisner 已提交
4905 4906 4907 4908 4909 4910 4911 4912 4913 4914 4915 4916 4917 4918 4919 4920 4921

		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) {
4922
			conn_err(tconn, "meta connection shut down by peer.\n");
P
Philipp Reisner 已提交
4923 4924
			goto reconnect;
		} else if (rv == -EAGAIN) {
4925 4926
			/* If the data socket received something meanwhile,
			 * that is good enough: peer is still alive. */
4927 4928
			if (time_after(tconn->last_received,
				jiffies - tconn->meta.socket->sk->sk_rcvtimeo))
4929
				continue;
4930
			if (ping_timeout_active) {
4931
				conn_err(tconn, "PingAck did not arrive in time.\n");
P
Philipp Reisner 已提交
4932 4933
				goto reconnect;
			}
4934
			set_bit(SEND_PING, &tconn->flags);
P
Philipp Reisner 已提交
4935 4936 4937 4938
			continue;
		} else if (rv == -EINTR) {
			continue;
		} else {
4939
			conn_err(tconn, "sock_recvmsg returned %d\n", rv);
P
Philipp Reisner 已提交
4940 4941 4942 4943
			goto reconnect;
		}

		if (received == expect && cmd == NULL) {
4944
			if (decode_header(tconn, h, &pi))
P
Philipp Reisner 已提交
4945
				goto reconnect;
4946
			cmd = &asender_tbl[pi.cmd];
4947
			if (pi.cmd >= ARRAY_SIZE(asender_tbl) || !cmd->fn) {
4948
				conn_err(tconn, "unknown command %d on meta (l: %d)\n",
4949
					pi.cmd, pi.size);
P
Philipp Reisner 已提交
4950 4951 4952
				goto disconnect;
			}
			expect = cmd->pkt_size;
4953
			if (pi.size != expect - header_size) {
4954
				conn_err(tconn, "Wrong packet size on meta (c: %d, l: %d)\n",
4955
					pi.cmd, pi.size);
P
Philipp Reisner 已提交
4956
				goto reconnect;
4957
			}
P
Philipp Reisner 已提交
4958 4959
		}
		if (received == expect) {
4960 4961
			bool rv;

4962 4963 4964
			rv = cmd->fn(tconn, &pi);
			if (!rv) {
				conn_err(tconn, "%pf failed\n", cmd->fn);
P
Philipp Reisner 已提交
4965
				goto reconnect;
4966
			}
P
Philipp Reisner 已提交
4967

4968 4969
			tconn->last_received = jiffies;

4970 4971
			/* the idle_timeout (ping-int)
			 * has been restored in got_PingAck() */
4972
			if (cmd == &asender_tbl[P_PING_ACK])
4973 4974
				ping_timeout_active = 0;

P
Philipp Reisner 已提交
4975 4976
			buf	 = h;
			received = 0;
4977
			expect	 = header_size;
P
Philipp Reisner 已提交
4978 4979 4980 4981 4982 4983
			cmd	 = NULL;
		}
	}

	if (0) {
reconnect:
4984
		conn_request_state(tconn, NS(conn, C_NETWORK_FAILURE), CS_HARD);
P
Philipp Reisner 已提交
4985 4986 4987
	}
	if (0) {
disconnect:
4988
		conn_request_state(tconn, NS(conn, C_DISCONNECTING), CS_HARD);
P
Philipp Reisner 已提交
4989
	}
4990
	clear_bit(SIGNAL_ASENDER, &tconn->flags);
P
Philipp Reisner 已提交
4991

4992
	conn_info(tconn, "asender terminated\n");
P
Philipp Reisner 已提交
4993 4994 4995

	return 0;
}