drbd_receiver.c 135.3 KB
Newer Older
P
Philipp Reisner 已提交
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50
/*
   drbd_receiver.c

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

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

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

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

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


#include <linux/module.h>

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

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

#include "drbd_vli.h"

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

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

64
static int drbd_do_features(struct drbd_tconn *tconn);
65
static int drbd_do_auth(struct drbd_tconn *tconn);
66
static int drbd_disconnected(int vnr, void *p, void *data);
P
Philipp Reisner 已提交
67 68

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


#define GFP_TRY	(__GFP_HIGHMEM | __GFP_NOWARN)

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

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

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

	page = *head;
92 93 94 95

	if (!page)
		return NULL;

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

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

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

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

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

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

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

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

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

	if (i == number)
		return page;

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

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

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

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

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

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

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

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

249 250
	/* Yes, we may run up to @number over max_buffers. If we
	 * follow it strictly, the admin will get it wrong anyways. */
251
	if (atomic_read(&mdev->pp_in_use) < mdev->tconn->net_conf->max_buffers)
252
		page = __drbd_alloc_pages(mdev, number);
P
Philipp Reisner 已提交
253

254
	while (page == NULL) {
P
Philipp Reisner 已提交
255 256 257 258
		prepare_to_wait(&drbd_pp_wait, &wait, TASK_INTERRUPTIBLE);

		drbd_kick_lo_and_reclaim_net(mdev);

259
		if (atomic_read(&mdev->pp_in_use) < mdev->tconn->net_conf->max_buffers) {
260
			page = __drbd_alloc_pages(mdev, number);
P
Philipp Reisner 已提交
261 262 263 264 265 266 267 268
			if (page)
				break;
		}

		if (!retry)
			break;

		if (signal_pending(current)) {
269
			dev_warn(DEV, "drbd_alloc_pages interrupted!\n");
P
Philipp Reisner 已提交
270 271 272 273 274 275 276
			break;
		}

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

277 278
	if (page)
		atomic_add(number, &mdev->pp_in_use);
P
Philipp Reisner 已提交
279 280 281
	return page;
}

282
/* Must not be used from irq, as that may deadlock: see drbd_alloc_pages.
283
 * Is also used from inside an other spin_lock_irq(&mdev->tconn->req_lock);
284 285
 * Either links the page chain back to the global pool,
 * or returns all pages to the system. */
286
static void drbd_pp_free(struct drbd_conf *mdev, struct page *page, int is_net)
P
Philipp Reisner 已提交
287
{
288
	atomic_t *a = is_net ? &mdev->pp_in_use_by_net : &mdev->pp_in_use;
P
Philipp Reisner 已提交
289
	int i;
290

291
	if (drbd_pp_vacant > (DRBD_MAX_BIO_SIZE/PAGE_SIZE) * minor_count)
292 293 294 295 296 297 298 299
		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
300
	}
301
	i = atomic_sub_return(i, a);
302
	if (i < 0)
303 304
		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
305 306 307 308 309 310 311 312
	wake_up(&drbd_pp_wait);
}

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

You must not have the req_lock:
313
 drbd_free_peer_req()
314
 drbd_alloc_peer_req()
315
 drbd_free_peer_reqs()
P
Philipp Reisner 已提交
316
 drbd_ee_fix_bhs()
317
 drbd_finish_peer_reqs()
P
Philipp Reisner 已提交
318 319 320 321
 drbd_clear_done_ee()
 drbd_wait_ee_list_empty()
*/

322
struct drbd_peer_request *
323 324
drbd_alloc_peer_req(struct drbd_conf *mdev, u64 id, sector_t sector,
		    unsigned int data_size, gfp_t gfp_mask) __must_hold(local)
P
Philipp Reisner 已提交
325
{
326
	struct drbd_peer_request *peer_req;
P
Philipp Reisner 已提交
327
	struct page *page;
328
	unsigned nr_pages = (data_size + PAGE_SIZE -1) >> PAGE_SHIFT;
P
Philipp Reisner 已提交
329

330
	if (drbd_insert_fault(mdev, DRBD_FAULT_AL_EE))
P
Philipp Reisner 已提交
331 332
		return NULL;

333 334
	peer_req = mempool_alloc(drbd_ee_mempool, gfp_mask & ~__GFP_HIGHMEM);
	if (!peer_req) {
P
Philipp Reisner 已提交
335
		if (!(gfp_mask & __GFP_NOWARN))
336
			dev_err(DEV, "%s: allocation failed\n", __func__);
P
Philipp Reisner 已提交
337 338 339
		return NULL;
	}

340
	page = drbd_alloc_pages(mdev, nr_pages, (gfp_mask & __GFP_WAIT));
341 342
	if (!page)
		goto fail;
P
Philipp Reisner 已提交
343

344 345 346 347 348 349 350
	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;
351
	peer_req->w.mdev = mdev;
352 353 354
	peer_req->pages = page;
	atomic_set(&peer_req->pending_bios, 0);
	peer_req->flags = 0;
355 356 357 358
	/*
	 * The block_id is opaque to the receiver.  It is not endianness
	 * converted, and sent back to the sender unchanged.
	 */
359
	peer_req->block_id = id;
P
Philipp Reisner 已提交
360

361
	return peer_req;
P
Philipp Reisner 已提交
362

363
 fail:
364
	mempool_free(peer_req, drbd_ee_mempool);
P
Philipp Reisner 已提交
365 366 367
	return NULL;
}

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

379
int drbd_free_peer_reqs(struct drbd_conf *mdev, struct list_head *list)
P
Philipp Reisner 已提交
380 381
{
	LIST_HEAD(work_list);
382
	struct drbd_peer_request *peer_req, *t;
P
Philipp Reisner 已提交
383
	int count = 0;
384
	int is_net = list == &mdev->net_ee;
P
Philipp Reisner 已提交
385

386
	spin_lock_irq(&mdev->tconn->req_lock);
P
Philipp Reisner 已提交
387
	list_splice_init(list, &work_list);
388
	spin_unlock_irq(&mdev->tconn->req_lock);
P
Philipp Reisner 已提交
389

390
	list_for_each_entry_safe(peer_req, t, &work_list, w.list) {
391
		__drbd_free_peer_req(mdev, peer_req, is_net);
P
Philipp Reisner 已提交
392 393 394 395 396
		count++;
	}
	return count;
}

397 398
/*
 * See also comments in _req_mod(,BARRIER_ACKED) and receive_Barrier.
P
Philipp Reisner 已提交
399
 */
400
static int drbd_finish_peer_reqs(struct drbd_conf *mdev)
P
Philipp Reisner 已提交
401 402 403
{
	LIST_HEAD(work_list);
	LIST_HEAD(reclaimed);
404
	struct drbd_peer_request *peer_req, *t;
405
	int err = 0;
P
Philipp Reisner 已提交
406

407
	spin_lock_irq(&mdev->tconn->req_lock);
408
	reclaim_finished_net_peer_reqs(mdev, &reclaimed);
P
Philipp Reisner 已提交
409
	list_splice_init(&mdev->done_ee, &work_list);
410
	spin_unlock_irq(&mdev->tconn->req_lock);
P
Philipp Reisner 已提交
411

412
	list_for_each_entry_safe(peer_req, t, &reclaimed, w.list)
413
		drbd_free_net_peer_req(mdev, peer_req);
P
Philipp Reisner 已提交
414 415

	/* possible callbacks here:
416
	 * e_end_block, and e_end_resync_block, e_send_discard_write.
P
Philipp Reisner 已提交
417 418
	 * all ignore the last argument.
	 */
419
	list_for_each_entry_safe(peer_req, t, &work_list, w.list) {
420 421
		int err2;

P
Philipp Reisner 已提交
422
		/* list_del not necessary, next/prev members not touched */
423 424 425
		err2 = peer_req->w.cb(&peer_req->w, !!err);
		if (!err)
			err = err2;
426
		drbd_free_peer_req(mdev, peer_req);
P
Philipp Reisner 已提交
427 428 429
	}
	wake_up(&mdev->ee_wait);

430
	return err;
P
Philipp Reisner 已提交
431 432
}

433 434
static void _drbd_wait_ee_list_empty(struct drbd_conf *mdev,
				     struct list_head *head)
P
Philipp Reisner 已提交
435 436 437 438 439 440 441
{
	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);
442
		spin_unlock_irq(&mdev->tconn->req_lock);
J
Jens Axboe 已提交
443
		io_schedule();
P
Philipp Reisner 已提交
444
		finish_wait(&mdev->ee_wait, &wait);
445
		spin_lock_irq(&mdev->tconn->req_lock);
P
Philipp Reisner 已提交
446 447 448
	}
}

449 450
static void drbd_wait_ee_list_empty(struct drbd_conf *mdev,
				    struct list_head *head)
P
Philipp Reisner 已提交
451
{
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 *, void *, struct packet_info *);
P
Philipp Reisner 已提交
733

734 735 736 737 738
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;
739
	return conn_send_command(tconn, sock, cmd, 0, 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, void *header, struct packet_info *pi)
P
Philipp Reisner 已提交
982
{
983 984
	unsigned int header_size = drbd_header_size(tconn);

985 986 987 988 989 990 991 992 993 994 995 996
	if (header_size == sizeof(struct p_header100) &&
	    *(__be32 *)header == cpu_to_be32(DRBD_MAGIC_100)) {
		struct p_header100 *h = header;
		if (h->pad != 0) {
			conn_err(tconn, "Header padding is not zero\n");
			return -EINVAL;
		}
		pi->vnr = be16_to_cpu(h->volume);
		pi->cmd = be16_to_cpu(h->command);
		pi->size = be32_to_cpu(h->length);
	} else if (header_size == sizeof(struct p_header95) &&
		   *(__be16 *)header == cpu_to_be16(DRBD_MAGIC_BIG)) {
997 998
		struct p_header95 *h = header;
		pi->cmd = be16_to_cpu(h->command);
999 1000
		pi->size = be32_to_cpu(h->length);
		pi->vnr = 0;
1001 1002 1003 1004 1005
	} else if (header_size == sizeof(struct p_header80) &&
		   *(__be32 *)header == cpu_to_be32(DRBD_MAGIC)) {
		struct p_header80 *h = header;
		pi->cmd = be16_to_cpu(h->command);
		pi->size = be16_to_cpu(h->length);
1006
		pi->vnr = 0;
1007
	} else {
1008 1009 1010
		conn_err(tconn, "Wrong magic value 0x%08x in protocol version %d\n",
			 be32_to_cpu(*(__be32 *)header),
			 tconn->agreed_pro_version);
1011
		return -EINVAL;
P
Philipp Reisner 已提交
1012
	}
1013
	pi->data = header + header_size;
1014
	return 0;
1015 1016
}

1017
static int drbd_recv_header(struct drbd_tconn *tconn, struct packet_info *pi)
1018
{
1019
	void *buffer = tconn->data.rbuf;
1020
	int err;
1021

1022
	err = drbd_recv_all_warn(tconn, buffer, drbd_header_size(tconn));
1023
	if (err)
1024
		return err;
1025

1026
	err = decode_header(tconn, buffer, pi);
1027
	tconn->last_received = jiffies;
P
Philipp Reisner 已提交
1028

1029
	return err;
P
Philipp Reisner 已提交
1030 1031
}

1032
static void drbd_flush(struct drbd_conf *mdev)
P
Philipp Reisner 已提交
1033 1034 1035 1036
{
	int rv;

	if (mdev->write_ordering >= WO_bdev_flush && get_ldev(mdev)) {
1037
		rv = blkdev_issue_flush(mdev->ldev->backing_bdev, GFP_KERNEL,
1038
					NULL);
P
Philipp Reisner 已提交
1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059
		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)
{
1060
	int epoch_size;
P
Philipp Reisner 已提交
1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083
	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 &&
1084
		    test_bit(DE_HAVE_BARRIER_NUMBER, &epoch->flags)) {
P
Philipp Reisner 已提交
1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103
			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);
1104
				/* atomic_set(&epoch->active, 0); is already zero */
P
Philipp Reisner 已提交
1105 1106
				if (rv == FE_STILL_LIVE)
					rv = FE_RECYCLED;
1107
				wake_up(&mdev->ee_wait);
P
Philipp Reisner 已提交
1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142
			}
		}

		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;
1143
	if (pwo != mdev->write_ordering || wo == WO_bdev_flush)
P
Philipp Reisner 已提交
1144 1145 1146
		dev_info(DEV, "Method to ensure write ordering: %s\n", write_ordering_str[mdev->write_ordering]);
}

1147
/**
1148
 * drbd_submit_peer_request()
1149
 * @mdev:	DRBD device.
1150
 * @peer_req:	peer request
1151
 * @rw:		flag field, see bio->bi_rw
1152 1153 1154 1155 1156 1157 1158 1159 1160 1161
 *
 * 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.
1162 1163
 */
/* TODO allocate from our own bio_set. */
1164 1165 1166
int drbd_submit_peer_request(struct drbd_conf *mdev,
			     struct drbd_peer_request *peer_req,
			     const unsigned rw, const int fault_type)
1167 1168 1169
{
	struct bio *bios = NULL;
	struct bio *bio;
1170 1171 1172
	struct page *page = peer_req->pages;
	sector_t sector = peer_req->i.sector;
	unsigned ds = peer_req->i.size;
1173 1174
	unsigned n_bios = 0;
	unsigned nr_pages = (ds + PAGE_SIZE -1) >> PAGE_SHIFT;
1175
	int err = -ENOMEM;
1176 1177 1178 1179

	/* 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
1180 1181 1182 1183 1184
	 * 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.
	 */
1185 1186 1187 1188 1189 1190
next_bio:
	bio = bio_alloc(GFP_NOIO, nr_pages);
	if (!bio) {
		dev_err(DEV, "submit_ee: Allocation of a bio failed\n");
		goto fail;
	}
1191
	/* > peer_req->i.sector, unless this is the first bio */
1192 1193 1194
	bio->bi_sector = sector;
	bio->bi_bdev = mdev->ldev->backing_bdev;
	bio->bi_rw = rw;
1195
	bio->bi_private = peer_req;
1196
	bio->bi_end_io = drbd_peer_request_endio;
1197 1198 1199 1200 1201 1202 1203 1204

	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)) {
1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215
			/* 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;
			}
1216 1217 1218 1219 1220 1221 1222 1223 1224
			goto next_bio;
		}
		ds -= len;
		sector += len >> 9;
		--nr_pages;
	}
	D_ASSERT(page == NULL);
	D_ASSERT(ds == 0);

1225
	atomic_set(&peer_req->pending_bios, n_bios);
1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240
	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);
	}
1241
	return err;
1242 1243
}

1244
static void drbd_remove_epoch_entry_interval(struct drbd_conf *mdev,
1245
					     struct drbd_peer_request *peer_req)
1246
{
1247
	struct drbd_interval *i = &peer_req->i;
1248 1249 1250 1251

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

A
Andreas Gruenbacher 已提交
1252
	/* Wake up any processes waiting for this peer request to complete.  */
1253 1254 1255 1256
	if (i->waiting)
		wake_up(&mdev->misc_wait);
}

1257
static int receive_Barrier(struct drbd_tconn *tconn, struct packet_info *pi)
P
Philipp Reisner 已提交
1258
{
1259
	struct drbd_conf *mdev;
1260
	int rv;
1261
	struct p_barrier *p = pi->data;
P
Philipp Reisner 已提交
1262 1263
	struct drbd_epoch *epoch;

1264 1265 1266 1267
	mdev = vnr_to_mdev(tconn, pi->vnr);
	if (!mdev)
		return -EIO;

P
Philipp Reisner 已提交
1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280
	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)
1281
			return 0;
1282 1283 1284 1285 1286 1287 1288 1289 1290

		/* 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
1291 1292 1293 1294

	case WO_bdev_flush:
	case WO_drain_io:
		drbd_wait_ee_list_empty(mdev, &mdev->active_ee);
1295 1296 1297 1298 1299 1300
		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
1301 1302
		}

1303 1304 1305 1306 1307
		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
1308

1309
		return 0;
1310 1311
	default:
		dev_err(DEV, "Strangeness in mdev->write_ordering %d\n", mdev->write_ordering);
1312
		return -EIO;
P
Philipp Reisner 已提交
1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329
	}

	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);

1330
	return 0;
P
Philipp Reisner 已提交
1331 1332 1333 1334
}

/* used from receive_RSDataReply (recv_resync_read)
 * and from receive_Data */
1335 1336 1337
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
1338
{
1339
	const sector_t capacity = drbd_get_capacity(mdev->this_bdev);
1340
	struct drbd_peer_request *peer_req;
P
Philipp Reisner 已提交
1341
	struct page *page;
1342
	int dgs, ds, err;
1343 1344
	void *dig_in = mdev->tconn->int_dig_in;
	void *dig_vv = mdev->tconn->int_dig_vv;
1345
	unsigned long *data;
P
Philipp Reisner 已提交
1346

1347 1348
	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
1349 1350

	if (dgs) {
1351 1352 1353 1354
		/*
		 * FIXME: Receive the incoming digest into the receive buffer
		 *	  here, together with its struct p_data?
		 */
1355 1356
		err = drbd_recv_all_warn(mdev->tconn, dig_in, dgs);
		if (err)
P
Philipp Reisner 已提交
1357 1358 1359 1360 1361
			return NULL;
	}

	data_size -= dgs;

1362 1363 1364 1365 1366 1367
	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
1368

1369 1370 1371
	/* even though we trust out peer,
	 * we sometimes have to double check. */
	if (sector + (data_size>>9) > capacity) {
1372 1373
		dev_err(DEV, "request from peer beyond end of local disk: "
			"capacity: %llus < sector: %llus + size: %u\n",
1374 1375 1376 1377 1378
			(unsigned long long)capacity,
			(unsigned long long)sector, data_size);
		return NULL;
	}

P
Philipp Reisner 已提交
1379 1380 1381
	/* 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.  */
1382
	peer_req = drbd_alloc_peer_req(mdev, id, sector, data_size, GFP_NOIO);
1383
	if (!peer_req)
P
Philipp Reisner 已提交
1384
		return NULL;
1385

P
Philipp Reisner 已提交
1386
	ds = data_size;
1387
	page = peer_req->pages;
1388 1389
	page_chain_for_each(page) {
		unsigned len = min_t(int, ds, PAGE_SIZE);
1390
		data = kmap(page);
1391
		err = drbd_recv_all_warn(mdev->tconn, data, len);
1392
		if (drbd_insert_fault(mdev, DRBD_FAULT_RECEIVE)) {
1393 1394 1395
			dev_err(DEV, "Fault injection: Corrupting data on receive\n");
			data[0] = data[0] ^ (unsigned long)-1;
		}
P
Philipp Reisner 已提交
1396
		kunmap(page);
1397
		if (err) {
1398
			drbd_free_peer_req(mdev, peer_req);
P
Philipp Reisner 已提交
1399 1400
			return NULL;
		}
1401
		ds -= len;
P
Philipp Reisner 已提交
1402 1403 1404
	}

	if (dgs) {
1405
		drbd_csum_ee(mdev, mdev->tconn->integrity_r_tfm, peer_req, dig_vv);
P
Philipp Reisner 已提交
1406
		if (memcmp(dig_in, dig_vv, dgs)) {
1407 1408
			dev_err(DEV, "Digest integrity check FAILED: %llus +%u\n",
				(unsigned long long)sector, data_size);
1409
			drbd_free_peer_req(mdev, peer_req);
P
Philipp Reisner 已提交
1410 1411 1412 1413
			return NULL;
		}
	}
	mdev->recv_cnt += data_size>>9;
1414
	return peer_req;
P
Philipp Reisner 已提交
1415 1416 1417 1418 1419 1420 1421 1422
}

/* 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;
1423
	int err = 0;
P
Philipp Reisner 已提交
1424 1425
	void *data;

1426
	if (!data_size)
1427
		return 0;
1428

1429
	page = drbd_alloc_pages(mdev, 1, 1);
P
Philipp Reisner 已提交
1430 1431 1432

	data = kmap(page);
	while (data_size) {
1433 1434
		unsigned int len = min_t(int, data_size, PAGE_SIZE);

1435 1436
		err = drbd_recv_all_warn(mdev->tconn, data, len);
		if (err)
P
Philipp Reisner 已提交
1437
			break;
1438
		data_size -= len;
P
Philipp Reisner 已提交
1439 1440
	}
	kunmap(page);
1441
	drbd_pp_free(mdev, page, 0);
1442
	return err;
P
Philipp Reisner 已提交
1443 1444 1445 1446 1447 1448 1449
}

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;
1450
	int dgs, err, i, expect;
1451 1452
	void *dig_in = mdev->tconn->int_dig_in;
	void *dig_vv = mdev->tconn->int_dig_vv;
P
Philipp Reisner 已提交
1453

1454 1455
	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
1456 1457

	if (dgs) {
1458 1459 1460
		err = drbd_recv_all_warn(mdev->tconn, dig_in, dgs);
		if (err)
			return err;
P
Philipp Reisner 已提交
1461 1462 1463 1464 1465 1466 1467 1468 1469 1470 1471 1472
	}

	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) {
1473
		void *mapped = kmap(bvec->bv_page) + bvec->bv_offset;
P
Philipp Reisner 已提交
1474
		expect = min_t(int, data_size, bvec->bv_len);
1475
		err = drbd_recv_all_warn(mdev->tconn, mapped, expect);
P
Philipp Reisner 已提交
1476
		kunmap(bvec->bv_page);
1477 1478 1479
		if (err)
			return err;
		data_size -= expect;
P
Philipp Reisner 已提交
1480 1481 1482
	}

	if (dgs) {
1483
		drbd_csum_bio(mdev, mdev->tconn->integrity_r_tfm, bio, dig_vv);
P
Philipp Reisner 已提交
1484 1485
		if (memcmp(dig_in, dig_vv, dgs)) {
			dev_err(DEV, "Digest integrity check FAILED. Broken NICs?\n");
1486
			return -EINVAL;
P
Philipp Reisner 已提交
1487 1488 1489 1490
		}
	}

	D_ASSERT(data_size == 0);
1491
	return 0;
P
Philipp Reisner 已提交
1492 1493
}

1494 1495 1496 1497
/*
 * e_end_resync_block() is called in asender context via
 * drbd_finish_peer_reqs().
 */
1498
static int e_end_resync_block(struct drbd_work *w, int unused)
P
Philipp Reisner 已提交
1499
{
1500 1501
	struct drbd_peer_request *peer_req =
		container_of(w, struct drbd_peer_request, w);
1502
	struct drbd_conf *mdev = w->mdev;
1503
	sector_t sector = peer_req->i.sector;
1504
	int err;
P
Philipp Reisner 已提交
1505

1506
	D_ASSERT(drbd_interval_empty(&peer_req->i));
P
Philipp Reisner 已提交
1507

1508 1509
	if (likely((peer_req->flags & EE_WAS_ERROR) == 0)) {
		drbd_set_in_sync(mdev, sector, peer_req->i.size);
1510
		err = drbd_send_ack(mdev, P_RS_WRITE_ACK, peer_req);
P
Philipp Reisner 已提交
1511 1512
	} else {
		/* Record failure to sync */
1513
		drbd_rs_failed_io(mdev, sector, peer_req->i.size);
P
Philipp Reisner 已提交
1514

1515
		err  = drbd_send_ack(mdev, P_NEG_ACK, peer_req);
P
Philipp Reisner 已提交
1516 1517 1518
	}
	dec_unacked(mdev);

1519
	return err;
P
Philipp Reisner 已提交
1520 1521 1522 1523
}

static int recv_resync_read(struct drbd_conf *mdev, sector_t sector, int data_size) __releases(local)
{
1524
	struct drbd_peer_request *peer_req;
P
Philipp Reisner 已提交
1525

1526 1527
	peer_req = read_in_block(mdev, ID_SYNCER, sector, data_size);
	if (!peer_req)
1528
		goto fail;
P
Philipp Reisner 已提交
1529 1530 1531 1532 1533 1534 1535

	dec_rs_pending(mdev);

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

1536
	peer_req->w.cb = e_end_resync_block;
1537

1538
	spin_lock_irq(&mdev->tconn->req_lock);
1539
	list_add(&peer_req->w.list, &mdev->sync_ee);
1540
	spin_unlock_irq(&mdev->tconn->req_lock);
P
Philipp Reisner 已提交
1541

1542
	atomic_add(data_size >> 9, &mdev->rs_sect_ev);
1543
	if (drbd_submit_peer_request(mdev, peer_req, WRITE, DRBD_FAULT_RS_WR) == 0)
1544
		return 0;
P
Philipp Reisner 已提交
1545

1546 1547
	/* don't care for the reason here */
	dev_err(DEV, "submit failed, triggering re-connect\n");
1548
	spin_lock_irq(&mdev->tconn->req_lock);
1549
	list_del(&peer_req->w.list);
1550
	spin_unlock_irq(&mdev->tconn->req_lock);
1551

1552
	drbd_free_peer_req(mdev, peer_req);
1553 1554
fail:
	put_ldev(mdev);
1555
	return -EIO;
P
Philipp Reisner 已提交
1556 1557
}

1558
static struct drbd_request *
1559 1560
find_request(struct drbd_conf *mdev, struct rb_root *root, u64 id,
	     sector_t sector, bool missing_ok, const char *func)
1561 1562 1563
{
	struct drbd_request *req;

1564 1565
	/* Request object according to our peer */
	req = (struct drbd_request *)(unsigned long)id;
1566
	if (drbd_contains_interval(root, sector, &req->i) && req->i.local)
1567
		return req;
1568 1569 1570 1571
	if (!missing_ok) {
		dev_err(DEV, "%s: failed to find request %lu, sector %llus\n", func,
			(unsigned long)id, (unsigned long long)sector);
	}
1572 1573 1574
	return NULL;
}

1575
static int receive_DataReply(struct drbd_tconn *tconn, struct packet_info *pi)
P
Philipp Reisner 已提交
1576
{
1577
	struct drbd_conf *mdev;
P
Philipp Reisner 已提交
1578 1579
	struct drbd_request *req;
	sector_t sector;
1580
	int err;
1581
	struct p_data *p = pi->data;
1582 1583 1584 1585

	mdev = vnr_to_mdev(tconn, pi->vnr);
	if (!mdev)
		return -EIO;
P
Philipp Reisner 已提交
1586 1587 1588

	sector = be64_to_cpu(p->sector);

1589
	spin_lock_irq(&mdev->tconn->req_lock);
1590
	req = find_request(mdev, &mdev->read_requests, p->block_id, sector, false, __func__);
1591
	spin_unlock_irq(&mdev->tconn->req_lock);
1592
	if (unlikely(!req))
1593
		return -EIO;
P
Philipp Reisner 已提交
1594

B
Bart Van Assche 已提交
1595
	/* hlist_del(&req->collision) is done in _req_may_be_done, to avoid
P
Philipp Reisner 已提交
1596 1597
	 * special casing it there for the various failure cases.
	 * still no race with drbd_fail_pending_reads */
1598
	err = recv_dless_read(mdev, req, sector, pi->size);
1599
	if (!err)
1600
		req_mod(req, DATA_RECEIVED);
P
Philipp Reisner 已提交
1601 1602 1603 1604
	/* else: nothing. handled from drbd_disconnect...
	 * I don't think we may complete this just yet
	 * in case we are "on-disconnect: freeze" */

1605
	return err;
P
Philipp Reisner 已提交
1606 1607
}

1608
static int receive_RSDataReply(struct drbd_tconn *tconn, struct packet_info *pi)
P
Philipp Reisner 已提交
1609
{
1610
	struct drbd_conf *mdev;
P
Philipp Reisner 已提交
1611
	sector_t sector;
1612
	int err;
1613
	struct p_data *p = pi->data;
1614 1615 1616 1617

	mdev = vnr_to_mdev(tconn, pi->vnr);
	if (!mdev)
		return -EIO;
P
Philipp Reisner 已提交
1618 1619 1620 1621 1622 1623 1624

	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,
1625
		 * or in drbd_peer_request_endio. */
1626
		err = recv_resync_read(mdev, sector, pi->size);
P
Philipp Reisner 已提交
1627 1628 1629 1630
	} else {
		if (__ratelimit(&drbd_ratelimit_state))
			dev_err(DEV, "Can not write resync data to local disk.\n");

1631
		err = drbd_drain_block(mdev, pi->size);
P
Philipp Reisner 已提交
1632

1633
		drbd_send_ack_dp(mdev, P_NEG_ACK, p, pi->size);
P
Philipp Reisner 已提交
1634 1635
	}

1636
	atomic_add(pi->size >> 9, &mdev->rs_sect_in);
1637

1638
	return err;
P
Philipp Reisner 已提交
1639 1640
}

1641
static int w_restart_write(struct drbd_work *w, int cancel)
1642 1643 1644 1645 1646 1647 1648 1649 1650 1651
{
	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);
1652
		return -EIO;
1653 1654 1655 1656 1657 1658 1659 1660 1661
	}
	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 */ ;
1662
	return 0;
1663 1664 1665 1666 1667 1668 1669 1670 1671 1672 1673 1674 1675 1676 1677 1678 1679 1680 1681 1682 1683 1684 1685
}

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);
		}
	}
}

1686 1687
/*
 * e_end_block() is called in asender context via drbd_finish_peer_reqs().
P
Philipp Reisner 已提交
1688
 */
1689
static int e_end_block(struct drbd_work *w, int cancel)
P
Philipp Reisner 已提交
1690
{
1691 1692
	struct drbd_peer_request *peer_req =
		container_of(w, struct drbd_peer_request, w);
1693
	struct drbd_conf *mdev = w->mdev;
1694
	sector_t sector = peer_req->i.sector;
1695
	int err = 0, pcmd;
P
Philipp Reisner 已提交
1696

1697
	if (mdev->tconn->net_conf->wire_protocol == DRBD_PROT_C) {
1698
		if (likely((peer_req->flags & EE_WAS_ERROR) == 0)) {
P
Philipp Reisner 已提交
1699 1700
			pcmd = (mdev->state.conn >= C_SYNC_SOURCE &&
				mdev->state.conn <= C_PAUSED_SYNC_T &&
1701
				peer_req->flags & EE_MAY_SET_IN_SYNC) ?
P
Philipp Reisner 已提交
1702
				P_RS_WRITE_ACK : P_WRITE_ACK;
1703
			err = drbd_send_ack(mdev, pcmd, peer_req);
P
Philipp Reisner 已提交
1704
			if (pcmd == P_RS_WRITE_ACK)
1705
				drbd_set_in_sync(mdev, sector, peer_req->i.size);
P
Philipp Reisner 已提交
1706
		} else {
1707
			err = drbd_send_ack(mdev, P_NEG_ACK, peer_req);
P
Philipp Reisner 已提交
1708 1709 1710 1711 1712 1713 1714
			/* 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.  */
1715
	if (mdev->tconn->net_conf->two_primaries) {
1716
		spin_lock_irq(&mdev->tconn->req_lock);
1717 1718
		D_ASSERT(!drbd_interval_empty(&peer_req->i));
		drbd_remove_epoch_entry_interval(mdev, peer_req);
1719 1720
		if (peer_req->flags & EE_RESTART_REQUESTS)
			restart_conflicting_writes(mdev, sector, peer_req->i.size);
1721
		spin_unlock_irq(&mdev->tconn->req_lock);
1722
	} else
1723
		D_ASSERT(drbd_interval_empty(&peer_req->i));
P
Philipp Reisner 已提交
1724

1725
	drbd_may_finish_epoch(mdev, peer_req->epoch, EV_PUT + (cancel ? EV_CLEANUP : 0));
P
Philipp Reisner 已提交
1726

1727
	return err;
P
Philipp Reisner 已提交
1728 1729
}

1730
static int e_send_ack(struct drbd_work *w, enum drbd_packet ack)
P
Philipp Reisner 已提交
1731
{
1732
	struct drbd_conf *mdev = w->mdev;
1733 1734
	struct drbd_peer_request *peer_req =
		container_of(w, struct drbd_peer_request, w);
1735
	int err;
P
Philipp Reisner 已提交
1736

1737
	err = drbd_send_ack(mdev, ack, peer_req);
P
Philipp Reisner 已提交
1738 1739
	dec_unacked(mdev);

1740
	return err;
P
Philipp Reisner 已提交
1741 1742
}

1743
static int e_send_discard_write(struct drbd_work *w, int unused)
1744 1745 1746 1747
{
	return e_send_ack(w, P_DISCARD_WRITE);
}

1748
static int e_send_retry_write(struct drbd_work *w, int unused)
1749 1750 1751 1752 1753 1754 1755
{
	struct drbd_tconn *tconn = w->mdev->tconn;

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

1756 1757 1758 1759 1760 1761 1762 1763 1764 1765 1766 1767 1768 1769 1770
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;
}

1771 1772 1773 1774 1775 1776 1777 1778 1779 1780 1781 1782 1783
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);
}

1784
static void update_peer_seq(struct drbd_conf *mdev, unsigned int peer_seq)
1785
{
1786
	unsigned int newest_peer_seq;
1787

1788 1789
	if (need_peer_seq(mdev)) {
		spin_lock(&mdev->peer_seq_lock);
1790 1791
		newest_peer_seq = seq_max(mdev->peer_seq, peer_seq);
		mdev->peer_seq = newest_peer_seq;
1792
		spin_unlock(&mdev->peer_seq_lock);
1793 1794
		/* wake up only if we actually changed mdev->peer_seq */
		if (peer_seq == newest_peer_seq)
1795 1796
			wake_up(&mdev->seq_wait);
	}
1797 1798
}

P
Philipp Reisner 已提交
1799 1800 1801 1802 1803 1804 1805 1806 1807 1808 1809 1810 1811 1812 1813 1814 1815 1816 1817 1818 1819
/* 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). */
1820
static int wait_for_and_update_peer_seq(struct drbd_conf *mdev, const u32 peer_seq)
P
Philipp Reisner 已提交
1821 1822 1823
{
	DEFINE_WAIT(wait);
	long timeout;
1824 1825 1826 1827 1828
	int ret;

	if (!need_peer_seq(mdev))
		return 0;

P
Philipp Reisner 已提交
1829 1830
	spin_lock(&mdev->peer_seq_lock);
	for (;;) {
1831 1832 1833
		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
1834
			break;
1835
		}
P
Philipp Reisner 已提交
1836 1837 1838 1839
		if (signal_pending(current)) {
			ret = -ERESTARTSYS;
			break;
		}
1840
		prepare_to_wait(&mdev->seq_wait, &wait, TASK_INTERRUPTIBLE);
P
Philipp Reisner 已提交
1841
		spin_unlock(&mdev->peer_seq_lock);
1842 1843
		timeout = mdev->tconn->net_conf->ping_timeo*HZ/10;
		timeout = schedule_timeout(timeout);
P
Philipp Reisner 已提交
1844
		spin_lock(&mdev->peer_seq_lock);
1845
		if (!timeout) {
P
Philipp Reisner 已提交
1846
			ret = -ETIMEDOUT;
1847
			dev_err(DEV, "Timed out waiting for missing ack packets; disconnecting\n");
P
Philipp Reisner 已提交
1848 1849 1850 1851
			break;
		}
	}
	spin_unlock(&mdev->peer_seq_lock);
1852
	finish_wait(&mdev->seq_wait, &wait);
P
Philipp Reisner 已提交
1853 1854 1855
	return ret;
}

1856 1857 1858 1859
/* 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)
1860
{
1861 1862 1863 1864
	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);
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 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999
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
2000
/* mirrored write */
2001
static int receive_Data(struct drbd_tconn *tconn, struct packet_info *pi)
P
Philipp Reisner 已提交
2002
{
2003
	struct drbd_conf *mdev;
P
Philipp Reisner 已提交
2004
	sector_t sector;
2005
	struct drbd_peer_request *peer_req;
2006
	struct p_data *p = pi->data;
2007
	u32 peer_seq = be32_to_cpu(p->seq_num);
P
Philipp Reisner 已提交
2008 2009
	int rw = WRITE;
	u32 dp_flags;
2010
	int err;
P
Philipp Reisner 已提交
2011

2012 2013 2014 2015
	mdev = vnr_to_mdev(tconn, pi->vnr);
	if (!mdev)
		return -EIO;

2016
	if (!get_ldev(mdev)) {
2017 2018
		int err2;

2019
		err = wait_for_and_update_peer_seq(mdev, peer_seq);
2020
		drbd_send_ack_dp(mdev, P_NEG_ACK, p, pi->size);
P
Philipp Reisner 已提交
2021
		atomic_inc(&mdev->current_epoch->epoch_size);
2022
		err2 = drbd_drain_block(mdev, pi->size);
2023 2024 2025
		if (!err)
			err = err2;
		return err;
P
Philipp Reisner 已提交
2026 2027
	}

2028 2029 2030 2031 2032
	/*
	 * 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
2033 2034

	sector = be64_to_cpu(p->sector);
2035
	peer_req = read_in_block(mdev, p->block_id, sector, pi->size);
2036
	if (!peer_req) {
P
Philipp Reisner 已提交
2037
		put_ldev(mdev);
2038
		return -EIO;
P
Philipp Reisner 已提交
2039 2040
	}

2041
	peer_req->w.cb = e_end_block;
P
Philipp Reisner 已提交
2042

2043 2044 2045 2046
	dp_flags = be32_to_cpu(p->dp_flags);
	rw |= wire_flags_to_bio(mdev, dp_flags);

	if (dp_flags & DP_MAY_SET_IN_SYNC)
2047
		peer_req->flags |= EE_MAY_SET_IN_SYNC;
2048

P
Philipp Reisner 已提交
2049
	spin_lock(&mdev->epoch_lock);
2050 2051 2052
	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
2053 2054
	spin_unlock(&mdev->epoch_lock);

2055 2056 2057
	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
2058
			goto out_interrupted;
2059
		spin_lock_irq(&mdev->tconn->req_lock);
2060 2061 2062 2063
		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
2064
				put_ldev(mdev);
2065
				return 0;
P
Philipp Reisner 已提交
2066
			}
2067
			goto out_interrupted;
P
Philipp Reisner 已提交
2068
		}
2069 2070
	} else
		spin_lock_irq(&mdev->tconn->req_lock);
2071
	list_add(&peer_req->w.list, &mdev->active_ee);
2072
	spin_unlock_irq(&mdev->tconn->req_lock);
P
Philipp Reisner 已提交
2073

2074
	switch (mdev->tconn->net_conf->wire_protocol) {
P
Philipp Reisner 已提交
2075 2076 2077 2078 2079 2080 2081 2082
	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 */
2083
		drbd_send_ack(mdev, P_RECV_ACK, peer_req);
P
Philipp Reisner 已提交
2084 2085 2086 2087 2088 2089
		break;
	case DRBD_PROT_A:
		/* nothing to do */
		break;
	}

2090
	if (mdev->state.pdsk < D_INCONSISTENT) {
P
Philipp Reisner 已提交
2091
		/* In case we have the only disk of the cluster, */
2092 2093 2094
		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;
2095
		drbd_al_begin_io(mdev, &peer_req->i);
P
Philipp Reisner 已提交
2096 2097
	}

2098 2099 2100
	err = drbd_submit_peer_request(mdev, peer_req, rw, DRBD_FAULT_DT_WR);
	if (!err)
		return 0;
P
Philipp Reisner 已提交
2101

2102 2103
	/* don't care for the reason here */
	dev_err(DEV, "submit failed, triggering re-connect\n");
2104
	spin_lock_irq(&mdev->tconn->req_lock);
2105 2106
	list_del(&peer_req->w.list);
	drbd_remove_epoch_entry_interval(mdev, peer_req);
2107
	spin_unlock_irq(&mdev->tconn->req_lock);
2108
	if (peer_req->flags & EE_CALL_AL_COMPLETE_IO)
2109
		drbd_al_complete_io(mdev, &peer_req->i);
2110

P
Philipp Reisner 已提交
2111
out_interrupted:
2112
	drbd_may_finish_epoch(mdev, peer_req->epoch, EV_PUT + EV_CLEANUP);
P
Philipp Reisner 已提交
2113
	put_ldev(mdev);
2114
	drbd_free_peer_req(mdev, peer_req);
2115
	return err;
P
Philipp Reisner 已提交
2116 2117
}

2118 2119 2120 2121 2122 2123 2124 2125 2126 2127 2128
/* 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.
 */
2129
int drbd_rs_should_slow_down(struct drbd_conf *mdev, sector_t sector)
2130 2131 2132
{
	struct gendisk *disk = mdev->ldev->backing_bdev->bd_contains->bd_disk;
	unsigned long db, dt, dbdt;
2133
	struct lc_element *tmp;
2134 2135 2136 2137
	int curr_events;
	int throttle = 0;

	/* feature disabled? */
2138
	if (mdev->ldev->dc.c_min_rate == 0)
2139 2140
		return 0;

2141 2142 2143 2144 2145 2146 2147 2148 2149 2150 2151 2152
	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);

2153 2154 2155
	curr_events = (int)part_stat_read(&disk->part0, sectors[0]) +
		      (int)part_stat_read(&disk->part0, sectors[1]) -
			atomic_read(&mdev->rs_sect_ev);
2156

2157 2158 2159 2160 2161 2162 2163 2164
	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. */
2165 2166 2167 2168 2169 2170
		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;
2171 2172 2173 2174 2175 2176 2177

		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);

2178
		if (dbdt > mdev->ldev->dc.c_min_rate)
2179 2180 2181 2182 2183 2184
			throttle = 1;
	}
	return throttle;
}


2185
static int receive_DataRequest(struct drbd_tconn *tconn, struct packet_info *pi)
P
Philipp Reisner 已提交
2186
{
2187
	struct drbd_conf *mdev;
P
Philipp Reisner 已提交
2188
	sector_t sector;
2189
	sector_t capacity;
2190
	struct drbd_peer_request *peer_req;
P
Philipp Reisner 已提交
2191
	struct digest_info *di = NULL;
2192
	int size, verb;
P
Philipp Reisner 已提交
2193
	unsigned int fault_type;
2194
	struct p_block_req *p =	pi->data;
2195 2196 2197 2198 2199

	mdev = vnr_to_mdev(tconn, pi->vnr);
	if (!mdev)
		return -EIO;
	capacity = drbd_get_capacity(mdev->this_bdev);
P
Philipp Reisner 已提交
2200 2201 2202 2203

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

2204
	if (size <= 0 || !IS_ALIGNED(size, 512) || size > DRBD_MAX_BIO_SIZE) {
P
Philipp Reisner 已提交
2205 2206
		dev_err(DEV, "%s:%d: sector: %llus, size: %u\n", __FILE__, __LINE__,
				(unsigned long long)sector, size);
2207
		return -EINVAL;
P
Philipp Reisner 已提交
2208 2209 2210 2211
	}
	if (sector + (size>>9) > capacity) {
		dev_err(DEV, "%s:%d: sector: %llus, size: %u\n", __FILE__, __LINE__,
				(unsigned long long)sector, size);
2212
		return -EINVAL;
P
Philipp Reisner 已提交
2213 2214 2215
	}

	if (!get_ldev_if_state(mdev, D_UP_TO_DATE)) {
2216
		verb = 1;
2217
		switch (pi->cmd) {
2218 2219 2220 2221 2222 2223 2224 2225 2226 2227 2228 2229 2230 2231
		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:
2232
			BUG();
2233 2234
		}
		if (verb && __ratelimit(&drbd_ratelimit_state))
P
Philipp Reisner 已提交
2235 2236
			dev_err(DEV, "Can not satisfy peer's read request, "
			    "no local data.\n");
2237

L
Lars Ellenberg 已提交
2238
		/* drain possibly payload */
2239
		return drbd_drain_block(mdev, pi->size);
P
Philipp Reisner 已提交
2240 2241 2242 2243 2244
	}

	/* 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.  */
2245
	peer_req = drbd_alloc_peer_req(mdev, p->block_id, sector, size, GFP_NOIO);
2246
	if (!peer_req) {
P
Philipp Reisner 已提交
2247
		put_ldev(mdev);
2248
		return -ENOMEM;
P
Philipp Reisner 已提交
2249 2250
	}

2251
	switch (pi->cmd) {
P
Philipp Reisner 已提交
2252
	case P_DATA_REQUEST:
2253
		peer_req->w.cb = w_e_end_data_req;
P
Philipp Reisner 已提交
2254
		fault_type = DRBD_FAULT_DT_RD;
2255 2256 2257
		/* application IO, don't drbd_rs_begin_io */
		goto submit;

P
Philipp Reisner 已提交
2258
	case P_RS_DATA_REQUEST:
2259
		peer_req->w.cb = w_e_end_rsdata_req;
P
Philipp Reisner 已提交
2260
		fault_type = DRBD_FAULT_RS_RD;
2261 2262
		/* used in the sector offset progress display */
		mdev->bm_resync_fo = BM_SECT_TO_BIT(sector);
P
Philipp Reisner 已提交
2263 2264 2265 2266 2267
		break;

	case P_OV_REPLY:
	case P_CSUM_RS_REQUEST:
		fault_type = DRBD_FAULT_RS_RD;
2268
		di = kmalloc(sizeof(*di) + pi->size, GFP_NOIO);
P
Philipp Reisner 已提交
2269 2270 2271
		if (!di)
			goto out_free_e;

2272
		di->digest_size = pi->size;
P
Philipp Reisner 已提交
2273 2274
		di->digest = (((char *)di)+sizeof(struct digest_info));

2275 2276
		peer_req->digest = di;
		peer_req->flags |= EE_HAS_DIGEST;
2277

2278
		if (drbd_recv_all(mdev->tconn, di->digest, pi->size))
P
Philipp Reisner 已提交
2279 2280
			goto out_free_e;

2281
		if (pi->cmd == P_CSUM_RS_REQUEST) {
2282
			D_ASSERT(mdev->tconn->agreed_pro_version >= 89);
2283
			peer_req->w.cb = w_e_end_csum_rs_req;
2284 2285
			/* used in the sector offset progress display */
			mdev->bm_resync_fo = BM_SECT_TO_BIT(sector);
2286
		} else if (pi->cmd == P_OV_REPLY) {
2287 2288
			/* track progress, we may need to throttle */
			atomic_add(size >> 9, &mdev->rs_sect_in);
2289
			peer_req->w.cb = w_e_end_ov_reply;
P
Philipp Reisner 已提交
2290
			dec_rs_pending(mdev);
2291 2292 2293
			/* 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
2294 2295 2296 2297 2298
		}
		break;

	case P_OV_REQUEST:
		if (mdev->ov_start_sector == ~(sector_t)0 &&
2299
		    mdev->tconn->agreed_pro_version >= 90) {
2300 2301
			unsigned long now = jiffies;
			int i;
P
Philipp Reisner 已提交
2302 2303
			mdev->ov_start_sector = sector;
			mdev->ov_position = sector;
2304 2305
			mdev->ov_left = drbd_bm_bits(mdev) - BM_SECT_TO_BIT(sector);
			mdev->rs_total = mdev->ov_left;
2306 2307 2308 2309
			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
2310 2311 2312
			dev_info(DEV, "Online Verify start sector: %llu\n",
					(unsigned long long)sector);
		}
2313
		peer_req->w.cb = w_e_end_ov_req;
P
Philipp Reisner 已提交
2314 2315 2316 2317
		fault_type = DRBD_FAULT_RS_RD;
		break;

	default:
2318
		BUG();
P
Philipp Reisner 已提交
2319 2320
	}

2321 2322 2323 2324 2325 2326 2327 2328 2329 2330 2331 2332 2333 2334 2335 2336 2337 2338 2339 2340 2341 2342
	/* 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.
	 */
2343 2344 2345
	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))
2346
		goto out_free_e;
P
Philipp Reisner 已提交
2347

2348 2349 2350
submit_for_resync:
	atomic_add(size >> 9, &mdev->rs_sect_ev);

2351
submit:
P
Philipp Reisner 已提交
2352
	inc_unacked(mdev);
2353
	spin_lock_irq(&mdev->tconn->req_lock);
2354
	list_add_tail(&peer_req->w.list, &mdev->read_ee);
2355
	spin_unlock_irq(&mdev->tconn->req_lock);
P
Philipp Reisner 已提交
2356

2357
	if (drbd_submit_peer_request(mdev, peer_req, READ, fault_type) == 0)
2358
		return 0;
P
Philipp Reisner 已提交
2359

2360 2361
	/* don't care for the reason here */
	dev_err(DEV, "submit failed, triggering re-connect\n");
2362
	spin_lock_irq(&mdev->tconn->req_lock);
2363
	list_del(&peer_req->w.list);
2364
	spin_unlock_irq(&mdev->tconn->req_lock);
2365 2366
	/* no drbd_rs_complete_io(), we are dropping the connection anyways */

P
Philipp Reisner 已提交
2367 2368
out_free_e:
	put_ldev(mdev);
2369
	drbd_free_peer_req(mdev, peer_req);
2370
	return -EIO;
P
Philipp Reisner 已提交
2371 2372 2373 2374 2375 2376 2377 2378 2379 2380 2381 2382 2383
}

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;

2384
	switch (mdev->tconn->net_conf->after_sb_0p) {
P
Philipp Reisner 已提交
2385 2386 2387 2388 2389 2390 2391 2392 2393 2394 2395 2396 2397 2398 2399 2400 2401 2402 2403 2404 2405 2406 2407 2408 2409 2410 2411
	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
2412
		dev_warn(DEV, "Discard younger/older primary did not find a decision\n"
P
Philipp Reisner 已提交
2413 2414 2415
		     "Using discard-least-changes instead\n");
	case ASB_DISCARD_ZERO_CHG:
		if (ch_peer == 0 && ch_self == 0) {
2416
			rv = test_bit(DISCARD_CONCURRENT, &mdev->tconn->flags)
P
Philipp Reisner 已提交
2417 2418 2419 2420 2421 2422
				? -1 : 1;
			break;
		} else {
			if (ch_peer == 0) { rv =  1; break; }
			if (ch_self == 0) { rv = -1; break; }
		}
2423
		if (mdev->tconn->net_conf->after_sb_0p == ASB_DISCARD_ZERO_CHG)
P
Philipp Reisner 已提交
2424 2425 2426 2427 2428 2429 2430 2431
			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. */
2432
			rv = test_bit(DISCARD_CONCURRENT, &mdev->tconn->flags)
P
Philipp Reisner 已提交
2433 2434 2435 2436 2437 2438 2439 2440 2441 2442 2443 2444 2445 2446
				? -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)
{
2447
	int hg, rv = -100;
P
Philipp Reisner 已提交
2448

2449
	switch (mdev->tconn->net_conf->after_sb_1p) {
P
Philipp Reisner 已提交
2450 2451 2452 2453 2454 2455 2456 2457 2458 2459 2460 2461 2462 2463 2464 2465 2466 2467 2468 2469 2470 2471 2472 2473
	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) {
2474 2475 2476
			enum drbd_state_rv rv2;

			drbd_set_role(mdev, R_SECONDARY, 0);
P
Philipp Reisner 已提交
2477 2478 2479
			 /* 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. */
2480 2481
			rv2 = drbd_change_state(mdev, CS_VERBOSE, NS(role, R_SECONDARY));
			if (rv2 != SS_SUCCESS) {
P
Philipp Reisner 已提交
2482 2483 2484 2485 2486 2487 2488 2489 2490 2491 2492 2493 2494 2495
				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)
{
2496
	int hg, rv = -100;
P
Philipp Reisner 已提交
2497

2498
	switch (mdev->tconn->net_conf->after_sb_2p) {
P
Philipp Reisner 已提交
2499 2500 2501 2502 2503 2504 2505 2506 2507 2508 2509 2510 2511 2512 2513 2514 2515
	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) {
2516 2517
			enum drbd_state_rv rv2;

P
Philipp Reisner 已提交
2518 2519 2520
			 /* 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. */
2521 2522
			rv2 = drbd_change_state(mdev, CS_VERBOSE, NS(role, R_SECONDARY));
			if (rv2 != SS_SUCCESS) {
P
Philipp Reisner 已提交
2523 2524 2525 2526 2527 2528 2529 2530 2531 2532 2533 2534 2535 2536 2537 2538 2539 2540 2541 2542 2543 2544 2545 2546 2547 2548 2549 2550 2551 2552 2553 2554 2555 2556 2557 2558 2559 2560
				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
2561 2562
-1091   requires proto 91
-1096   requires proto 96
P
Philipp Reisner 已提交
2563 2564 2565 2566 2567 2568 2569 2570 2571 2572 2573 2574 2575 2576 2577 2578 2579 2580 2581 2582 2583 2584 2585 2586 2587 2588 2589 2590
 */
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) {

2591
			if (mdev->tconn->agreed_pro_version < 91)
2592
				return -1091;
P
Philipp Reisner 已提交
2593 2594 2595 2596 2597 2598 2599 2600 2601 2602 2603 2604 2605 2606 2607 2608 2609 2610 2611

			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) {

2612
			if (mdev->tconn->agreed_pro_version < 91)
2613
				return -1091;
P
Philipp Reisner 已提交
2614 2615 2616 2617 2618 2619 2620 2621 2622 2623 2624 2625 2626 2627 2628 2629 2630 2631 2632 2633 2634 2635 2636 2637 2638 2639 2640 2641 2642 2643 2644

			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 */
2645
			dc = test_bit(DISCARD_CONCURRENT, &mdev->tconn->flags);
P
Philipp Reisner 已提交
2646 2647 2648 2649 2650 2651 2652 2653 2654 2655 2656 2657
			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) {
2658
		if (mdev->tconn->agreed_pro_version < 96 ?
2659 2660 2661
		    (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
2662 2663 2664
			/* The last P_SYNC_UUID did not get though. Undo the last start of
			   resync as sync source modifications of the peer's UUIDs. */

2665
			if (mdev->tconn->agreed_pro_version < 91)
2666
				return -1091;
P
Philipp Reisner 已提交
2667 2668 2669

			mdev->p_uuid[UI_BITMAP] = mdev->p_uuid[UI_HISTORY_START];
			mdev->p_uuid[UI_HISTORY_START] = mdev->p_uuid[UI_HISTORY_START + 1];
2670 2671 2672 2673

			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
2674 2675 2676 2677 2678 2679 2680 2681 2682 2683 2684 2685 2686 2687 2688 2689 2690 2691 2692 2693 2694
			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) {
2695
		if (mdev->tconn->agreed_pro_version < 96 ?
2696 2697 2698
		    (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
2699 2700 2701
			/* The last P_SYNC_UUID did not get though. Undo the last start of
			   resync as sync source modifications of our UUIDs. */

2702
			if (mdev->tconn->agreed_pro_version < 91)
2703
				return -1091;
P
Philipp Reisner 已提交
2704 2705 2706 2707

			_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]);

2708
			dev_info(DEV, "Last syncUUID did not get through, corrected:\n");
P
Philipp Reisner 已提交
2709 2710 2711 2712 2713 2714 2715 2716 2717
			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;
2718
	peer = mdev->p_uuid[UI_CURRENT] & ~((u64)1);
P
Philipp Reisner 已提交
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 2753 2754 2755 2756 2757 2758 2759 2760 2761 2762 2763 2764 2765 2766 2767 2768 2769 2770
	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;
	}
2771 2772
	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
2773 2774 2775 2776 2777 2778 2779 2780 2781 2782 2783 2784 2785
		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");
	}

2786 2787 2788
	if (abs(hg) == 100)
		drbd_khelper(mdev, "initial-split-brain");

2789
	if (hg == 100 || (hg == -100 && mdev->tconn->net_conf->always_asbp)) {
P
Philipp Reisner 已提交
2790 2791 2792 2793 2794 2795 2796 2797 2798 2799 2800 2801 2802 2803 2804 2805 2806 2807 2808 2809 2810 2811 2812 2813 2814 2815 2816 2817
		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) {
2818
		if (mdev->tconn->net_conf->want_lose && !(mdev->p_uuid[UI_FLAGS]&1))
P
Philipp Reisner 已提交
2819
			hg = -1;
2820
		if (!mdev->tconn->net_conf->want_lose && (mdev->p_uuid[UI_FLAGS]&1))
P
Philipp Reisner 已提交
2821 2822 2823 2824 2825 2826 2827 2828 2829
			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) {
2830 2831 2832 2833
		/* 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... */
2834
		dev_alert(DEV, "Split-Brain detected but unresolved, dropping connection!\n");
P
Philipp Reisner 已提交
2835 2836 2837 2838 2839 2840 2841 2842 2843 2844 2845
		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) {
2846
		switch (mdev->tconn->net_conf->rr_conflict) {
P
Philipp Reisner 已提交
2847 2848 2849 2850 2851 2852 2853 2854 2855 2856 2857 2858
		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");
		}
	}

2859
	if (mdev->tconn->net_conf->dry_run || test_bit(CONN_DRY_RUN, &mdev->tconn->flags)) {
2860 2861 2862 2863 2864 2865 2866 2867 2868
		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
2869 2870
	if (abs(hg) >= 2) {
		dev_info(DEV, "Writing the whole bitmap, full sync required after drbd_sync_handshake.\n");
2871 2872
		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
2873 2874 2875 2876 2877 2878 2879 2880 2881 2882 2883 2884 2885 2886 2887 2888 2889 2890 2891 2892 2893 2894 2895 2896 2897 2898 2899 2900 2901 2902 2903 2904 2905 2906 2907 2908 2909 2910 2911
			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;
}

2912
static int receive_protocol(struct drbd_tconn *tconn, struct packet_info *pi)
P
Philipp Reisner 已提交
2913
{
2914
	struct p_protocol *p = pi->data;
P
Philipp Reisner 已提交
2915
	int p_proto, p_after_sb_0p, p_after_sb_1p, p_after_sb_2p;
2916
	int p_want_lose, p_two_primaries, cf;
P
Philipp Reisner 已提交
2917 2918 2919 2920 2921 2922 2923
	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);
2924 2925 2926
	cf		= be32_to_cpu(p->conn_flags);
	p_want_lose = cf & CF_WANT_LOSE;

2927
	clear_bit(CONN_DRY_RUN, &tconn->flags);
2928 2929

	if (cf & CF_DRY_RUN)
2930
		set_bit(CONN_DRY_RUN, &tconn->flags);
P
Philipp Reisner 已提交
2931

2932 2933
	if (p_proto != tconn->net_conf->wire_protocol) {
		conn_err(tconn, "incompatible communication protocols\n");
P
Philipp Reisner 已提交
2934 2935 2936
		goto disconnect;
	}

2937 2938
	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
2939 2940 2941
		goto disconnect;
	}

2942 2943
	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
2944 2945 2946
		goto disconnect;
	}

2947 2948
	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
2949 2950 2951
		goto disconnect;
	}

2952 2953
	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
2954 2955 2956
		goto disconnect;
	}

2957 2958
	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
2959 2960 2961
		goto disconnect;
	}

2962 2963
	if (tconn->agreed_pro_version >= 87) {
		unsigned char *my_alg = tconn->net_conf->integrity_alg;
2964
		int err;
P
Philipp Reisner 已提交
2965

2966
		err = drbd_recv_all(tconn, p_integrity_alg, pi->size);
2967 2968
		if (err)
			return err;
P
Philipp Reisner 已提交
2969 2970 2971

		p_integrity_alg[SHARED_SECRET_MAX-1] = 0;
		if (strcmp(p_integrity_alg, my_alg)) {
2972
			conn_err(tconn, "incompatible setting of the data-integrity-alg\n");
P
Philipp Reisner 已提交
2973 2974
			goto disconnect;
		}
2975
		conn_info(tconn, "data-integrity-alg: %s\n",
P
Philipp Reisner 已提交
2976 2977 2978
		     my_alg[0] ? my_alg : (unsigned char *)"<not-used>");
	}

2979
	return 0;
P
Philipp Reisner 已提交
2980 2981

disconnect:
2982
	conn_request_state(tconn, NS(conn, C_DISCONNECTING), CS_HARD);
2983
	return -EIO;
P
Philipp Reisner 已提交
2984 2985 2986 2987 2988 2989 2990 2991 2992 2993 2994 2995 2996 2997 2998 2999 3000 3001 3002 3003 3004 3005 3006 3007 3008 3009 3010 3011 3012
}

/* 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;
}

3013 3014 3015 3016 3017 3018 3019 3020 3021 3022 3023 3024 3025 3026 3027 3028 3029 3030 3031 3032 3033 3034 3035 3036 3037 3038 3039 3040 3041 3042 3043 3044 3045 3046 3047 3048 3049 3050 3051
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
3052
{
3053
	struct drbd_conf *mdev;
3054
	struct p_rs_param_95 *p;
P
Philipp Reisner 已提交
3055 3056 3057
	unsigned int header_size, data_size, exp_max_sz;
	struct crypto_hash *verify_tfm = NULL;
	struct crypto_hash *csums_tfm = NULL;
3058
	const int apv = tconn->agreed_pro_version;
3059 3060
	int *rs_plan_s = NULL;
	int fifo_size = 0;
3061
	int err;
P
Philipp Reisner 已提交
3062

3063 3064 3065 3066
	mdev = vnr_to_mdev(tconn, pi->vnr);
	if (!mdev)
		return config_unknown_volume(tconn, pi);

P
Philipp Reisner 已提交
3067 3068 3069
	exp_max_sz  = apv <= 87 ? sizeof(struct p_rs_param)
		    : apv == 88 ? sizeof(struct p_rs_param)
					+ SHARED_SECRET_MAX
3070 3071
		    : apv <= 94 ? sizeof(struct p_rs_param_89)
		    : /* apv >= 95 */ sizeof(struct p_rs_param_95);
P
Philipp Reisner 已提交
3072

3073
	if (pi->size > exp_max_sz) {
P
Philipp Reisner 已提交
3074
		dev_err(DEV, "SyncParam packet too long: received %u, expected <= %u bytes\n",
3075
		    pi->size, exp_max_sz);
3076
		return -EIO;
P
Philipp Reisner 已提交
3077 3078 3079
	}

	if (apv <= 88) {
3080
		header_size = sizeof(struct p_rs_param);
3081
		data_size = pi->size - header_size;
3082
	} else if (apv <= 94) {
3083
		header_size = sizeof(struct p_rs_param_89);
3084
		data_size = pi->size - header_size;
P
Philipp Reisner 已提交
3085
		D_ASSERT(data_size == 0);
3086
	} else {
3087
		header_size = sizeof(struct p_rs_param_95);
3088
		data_size = pi->size - header_size;
P
Philipp Reisner 已提交
3089 3090 3091 3092
		D_ASSERT(data_size == 0);
	}

	/* initialize verify_alg and csums_alg */
3093
	p = pi->data;
P
Philipp Reisner 已提交
3094 3095
	memset(p->verify_alg, 0, 2 * SHARED_SECRET_MAX);

3096
	err = drbd_recv_all(mdev->tconn, p, header_size);
3097 3098
	if (err)
		return err;
P
Philipp Reisner 已提交
3099

3100 3101 3102 3103
	if (get_ldev(mdev)) {
		mdev->ldev->dc.resync_rate = be32_to_cpu(p->rate);
		put_ldev(mdev);
	}
P
Philipp Reisner 已提交
3104 3105 3106 3107 3108 3109 3110

	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);
3111
				return -EIO;
P
Philipp Reisner 已提交
3112 3113
			}

3114 3115 3116
			err = drbd_recv_all(mdev->tconn, p->verify_alg, data_size);
			if (err)
				return err;
P
Philipp Reisner 已提交
3117 3118 3119 3120 3121 3122 3123 3124 3125 3126 3127 3128 3129 3130 3131

			/* 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;
		}

3132
		if (strcmp(mdev->tconn->net_conf->verify_alg, p->verify_alg)) {
P
Philipp Reisner 已提交
3133 3134
			if (mdev->state.conn == C_WF_REPORT_PARAMS) {
				dev_err(DEV, "Different verify-alg settings. me=\"%s\" peer=\"%s\"\n",
3135
				    mdev->tconn->net_conf->verify_alg, p->verify_alg);
P
Philipp Reisner 已提交
3136 3137 3138 3139 3140 3141 3142 3143 3144 3145
				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;
			}
		}

3146
		if (apv >= 89 && strcmp(mdev->tconn->net_conf->csums_alg, p->csums_alg)) {
P
Philipp Reisner 已提交
3147 3148
			if (mdev->state.conn == C_WF_REPORT_PARAMS) {
				dev_err(DEV, "Different csums-alg settings. me=\"%s\" peer=\"%s\"\n",
3149
				    mdev->tconn->net_conf->csums_alg, p->csums_alg);
P
Philipp Reisner 已提交
3150 3151 3152 3153 3154 3155 3156 3157 3158 3159
				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;
			}
		}

3160 3161 3162 3163 3164 3165
		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);
3166

3167
			fifo_size = (mdev->ldev->dc.c_plan_ahead * 10 * SLEEP_TIME) / HZ;
3168 3169 3170 3171
			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");
3172
					put_ldev(mdev);
3173 3174 3175
					goto disconnect;
				}
			}
3176
			put_ldev(mdev);
3177
		}
P
Philipp Reisner 已提交
3178 3179 3180 3181

		spin_lock(&mdev->peer_seq_lock);
		/* lock against drbd_nl_syncer_conf() */
		if (verify_tfm) {
3182 3183 3184 3185
			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
3186 3187 3188
			dev_info(DEV, "using verify-alg: \"%s\"\n", p->verify_alg);
		}
		if (csums_tfm) {
3189 3190 3191 3192
			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
3193 3194
			dev_info(DEV, "using csums-alg: \"%s\"\n", p->csums_alg);
		}
3195 3196 3197 3198 3199 3200
		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
3201 3202
		spin_unlock(&mdev->peer_seq_lock);
	}
3203
	return 0;
P
Philipp Reisner 已提交
3204 3205 3206 3207 3208 3209 3210

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);
3211
	conn_request_state(mdev->tconn, NS(conn, C_DISCONNECTING), CS_HARD);
3212
	return -EIO;
P
Philipp Reisner 已提交
3213 3214 3215 3216 3217 3218 3219 3220 3221 3222 3223 3224 3225 3226 3227
}

/* 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);
}

3228
static int receive_sizes(struct drbd_tconn *tconn, struct packet_info *pi)
P
Philipp Reisner 已提交
3229
{
3230
	struct drbd_conf *mdev;
3231
	struct p_sizes *p = pi->data;
P
Philipp Reisner 已提交
3232 3233 3234
	enum determine_dev_size dd = unchanged;
	sector_t p_size, p_usize, my_usize;
	int ldsc = 0; /* local disk size changed */
3235
	enum dds_flags ddsf;
P
Philipp Reisner 已提交
3236

3237 3238 3239 3240
	mdev = vnr_to_mdev(tconn, pi->vnr);
	if (!mdev)
		return config_unknown_volume(tconn, pi);

P
Philipp Reisner 已提交
3241 3242 3243 3244 3245 3246 3247 3248 3249 3250 3251 3252 3253 3254 3255 3256 3257 3258 3259 3260 3261 3262 3263 3264 3265 3266 3267 3268 3269
	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. */
3270
		if (drbd_new_dev_size(mdev, mdev->ldev, 0) <
P
Philipp Reisner 已提交
3271 3272 3273 3274
		   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");
3275
			conn_request_state(mdev->tconn, NS(conn, C_DISCONNECTING), CS_HARD);
P
Philipp Reisner 已提交
3276 3277
			mdev->ldev->dc.disk_size = my_usize;
			put_ldev(mdev);
3278
			return -EIO;
P
Philipp Reisner 已提交
3279 3280 3281 3282
		}
		put_ldev(mdev);
	}

3283
	ddsf = be16_to_cpu(p->dds_flags);
P
Philipp Reisner 已提交
3284
	if (get_ldev(mdev)) {
B
Bart Van Assche 已提交
3285
		dd = drbd_determine_dev_size(mdev, ddsf);
P
Philipp Reisner 已提交
3286 3287
		put_ldev(mdev);
		if (dd == dev_size_error)
3288
			return -EIO;
P
Philipp Reisner 已提交
3289 3290 3291 3292 3293 3294
		drbd_md_sync(mdev);
	} else {
		/* I am diskless, need to accept the peer's size. */
		drbd_set_my_capacity(mdev, p_size);
	}

3295 3296 3297
	mdev->peer_max_bio_size = be32_to_cpu(p->max_bio_size);
	drbd_reconsider_max_bio_size(mdev);

P
Philipp Reisner 已提交
3298 3299 3300 3301 3302 3303 3304 3305 3306 3307 3308 3309 3310 3311
	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... */
3312
			drbd_send_sizes(mdev, 0, ddsf);
P
Philipp Reisner 已提交
3313 3314 3315 3316
		}
		if (test_and_clear_bit(RESIZE_PENDING, &mdev->flags) ||
		    (dd == grew && mdev->state.conn == C_CONNECTED)) {
			if (mdev->state.pdsk >= D_INCONSISTENT &&
3317 3318 3319 3320 3321 3322
			    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
3323 3324 3325 3326
				set_bit(RESYNC_AFTER_NEG, &mdev->flags);
		}
	}

3327
	return 0;
P
Philipp Reisner 已提交
3328 3329
}

3330
static int receive_uuids(struct drbd_tconn *tconn, struct packet_info *pi)
P
Philipp Reisner 已提交
3331
{
3332
	struct drbd_conf *mdev;
3333
	struct p_uuids *p = pi->data;
P
Philipp Reisner 已提交
3334
	u64 *p_uuid;
3335
	int i, updated_uuids = 0;
P
Philipp Reisner 已提交
3336

3337 3338 3339 3340
	mdev = vnr_to_mdev(tconn, pi->vnr);
	if (!mdev)
		return config_unknown_volume(tconn, pi);

P
Philipp Reisner 已提交
3341 3342 3343 3344 3345 3346 3347 3348 3349 3350 3351 3352 3353 3354
	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);
3355
		conn_request_state(mdev->tconn, NS(conn, C_DISCONNECTING), CS_HARD);
3356
		return -EIO;
P
Philipp Reisner 已提交
3357 3358 3359 3360 3361
	}

	if (get_ldev(mdev)) {
		int skip_initial_sync =
			mdev->state.conn == C_CONNECTED &&
3362
			mdev->tconn->agreed_pro_version >= 90 &&
P
Philipp Reisner 已提交
3363 3364 3365 3366 3367
			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,
3368 3369
					"clear_n_write from receive_uuids",
					BM_LOCKED_TEST_ALLOWED);
P
Philipp Reisner 已提交
3370 3371 3372 3373 3374
			_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);
3375
			updated_uuids = 1;
P
Philipp Reisner 已提交
3376 3377
		}
		put_ldev(mdev);
3378 3379 3380 3381
	} 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. */
3382
		updated_uuids = drbd_set_ed_uuid(mdev, p_uuid[UI_CURRENT]);
P
Philipp Reisner 已提交
3383 3384 3385 3386 3387 3388
	}

	/* 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... */
3389 3390
	mutex_lock(mdev->state_mutex);
	mutex_unlock(mdev->state_mutex);
P
Philipp Reisner 已提交
3391
	if (mdev->state.conn >= C_CONNECTED && mdev->state.disk < D_INCONSISTENT)
3392 3393 3394 3395
		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
3396

3397
	return 0;
P
Philipp Reisner 已提交
3398 3399 3400 3401 3402 3403 3404 3405 3406 3407 3408 3409 3410 3411 3412 3413 3414 3415 3416 3417 3418 3419 3420 3421 3422 3423 3424 3425 3426 3427 3428 3429
}

/**
 * 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;
}

3430
static int receive_req_state(struct drbd_tconn *tconn, struct packet_info *pi)
P
Philipp Reisner 已提交
3431
{
3432
	struct drbd_conf *mdev;
3433
	struct p_req_state *p = pi->data;
P
Philipp Reisner 已提交
3434
	union drbd_state mask, val;
3435
	enum drbd_state_rv rv;
P
Philipp Reisner 已提交
3436

3437 3438 3439 3440
	mdev = vnr_to_mdev(tconn, pi->vnr);
	if (!mdev)
		return -EIO;

P
Philipp Reisner 已提交
3441 3442 3443
	mask.i = be32_to_cpu(p->mask);
	val.i = be32_to_cpu(p->val);

3444
	if (test_bit(DISCARD_CONCURRENT, &mdev->tconn->flags) &&
3445
	    mutex_is_locked(mdev->state_mutex)) {
P
Philipp Reisner 已提交
3446
		drbd_send_sr_reply(mdev, SS_CONCURRENT_ST_CHG);
3447
		return 0;
P
Philipp Reisner 已提交
3448 3449 3450 3451 3452
	}

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

3453 3454
	rv = drbd_change_state(mdev, CS_VERBOSE, mask, val);
	drbd_send_sr_reply(mdev, rv);
P
Philipp Reisner 已提交
3455 3456 3457

	drbd_md_sync(mdev);

3458
	return 0;
P
Philipp Reisner 已提交
3459 3460
}

3461
static int receive_req_conn_state(struct drbd_tconn *tconn, struct packet_info *pi)
3462
{
3463
	struct p_req_state *p = pi->data;
3464 3465 3466 3467 3468 3469 3470 3471 3472
	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);
3473
		return 0;
3474 3475 3476 3477 3478
	}

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

3479
	rv = conn_request_state(tconn, mask, val, CS_VERBOSE | CS_LOCAL_ONLY | CS_IGN_OUTD_FAIL);
3480 3481
	conn_send_sr_reply(tconn, rv);

3482
	return 0;
3483 3484
}

3485
static int receive_state(struct drbd_tconn *tconn, struct packet_info *pi)
P
Philipp Reisner 已提交
3486
{
3487
	struct drbd_conf *mdev;
3488
	struct p_state *p = pi->data;
3489
	union drbd_state os, ns, peer_state;
P
Philipp Reisner 已提交
3490
	enum drbd_disk_state real_peer_disk;
3491
	enum chg_state_flags cs_flags;
P
Philipp Reisner 已提交
3492 3493
	int rv;

3494 3495 3496 3497
	mdev = vnr_to_mdev(tconn, pi->vnr);
	if (!mdev)
		return config_unknown_volume(tconn, pi);

P
Philipp Reisner 已提交
3498 3499 3500 3501 3502 3503 3504 3505
	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));
	}

3506
	spin_lock_irq(&mdev->tconn->req_lock);
P
Philipp Reisner 已提交
3507
 retry:
3508
	os = ns = drbd_read_state(mdev);
3509
	spin_unlock_irq(&mdev->tconn->req_lock);
P
Philipp Reisner 已提交
3510

3511 3512 3513 3514 3515 3516 3517 3518 3519 3520 3521 3522 3523 3524 3525 3526 3527 3528 3529 3530 3531
	/* 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);
3532
			return 0;
3533 3534 3535 3536 3537 3538 3539 3540 3541 3542 3543 3544
		}
	}

	/* 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;

3545 3546
	if (ns.conn == C_WF_REPORT_PARAMS)
		ns.conn = C_CONNECTED;
P
Philipp Reisner 已提交
3547

3548 3549 3550
	if (peer_state.conn == C_AHEAD)
		ns.conn = C_BEHIND;

P
Philipp Reisner 已提交
3551 3552 3553 3554 3555
	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 */
3556
		cr  = (os.conn < C_CONNECTED);
P
Philipp Reisner 已提交
3557 3558
		/* if we had an established connection
		 * and one of the nodes newly attaches a disk */
3559
		cr |= (os.conn == C_CONNECTED &&
P
Philipp Reisner 已提交
3560
		       (peer_state.disk == D_NEGOTIATING ||
3561
			os.disk == D_NEGOTIATING));
P
Philipp Reisner 已提交
3562 3563 3564 3565 3566
		/* 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" */
3567
		cr |= (os.conn == C_CONNECTED &&
P
Philipp Reisner 已提交
3568 3569 3570 3571
				(peer_state.conn >= C_STARTING_SYNC_S &&
				 peer_state.conn <= C_WF_BITMAP_T));

		if (cr)
3572
			ns.conn = drbd_sync_handshake(mdev, peer_state.role, real_peer_disk);
P
Philipp Reisner 已提交
3573 3574

		put_ldev(mdev);
3575 3576
		if (ns.conn == C_MASK) {
			ns.conn = C_CONNECTED;
P
Philipp Reisner 已提交
3577
			if (mdev->state.disk == D_NEGOTIATING) {
3578
				drbd_force_state(mdev, NS(disk, D_FAILED));
P
Philipp Reisner 已提交
3579 3580 3581
			} 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;
3582
				real_peer_disk = D_DISKLESS;
P
Philipp Reisner 已提交
3583
			} else {
3584
				if (test_and_clear_bit(CONN_DRY_RUN, &mdev->tconn->flags))
3585
					return -EIO;
3586
				D_ASSERT(os.conn == C_WF_REPORT_PARAMS);
3587
				conn_request_state(mdev->tconn, NS(conn, C_DISCONNECTING), CS_HARD);
3588
				return -EIO;
P
Philipp Reisner 已提交
3589 3590 3591 3592
			}
		}
	}

3593
	spin_lock_irq(&mdev->tconn->req_lock);
3594
	if (os.i != drbd_read_state(mdev).i)
P
Philipp Reisner 已提交
3595 3596 3597 3598 3599
		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);
3600
	if ((ns.conn == C_CONNECTED || ns.conn == C_WF_BITMAP_S) && ns.disk == D_NEGOTIATING)
P
Philipp Reisner 已提交
3601
		ns.disk = mdev->new_state_tmp.disk;
3602
	cs_flags = CS_VERBOSE + (os.conn < C_CONNECTED && ns.conn >= C_CONNECTED ? 0 : CS_HARD);
3603
	if (ns.pdsk == D_CONSISTENT && drbd_suspended(mdev) && ns.conn == C_CONNECTED && os.conn < C_CONNECTED &&
3604
	    test_bit(NEW_CUR_UUID, &mdev->flags)) {
3605
		/* Do not allow tl_restart(RESEND) for a rebooted peer. We can only allow this
3606
		   for temporal network outages! */
3607
		spin_unlock_irq(&mdev->tconn->req_lock);
3608
		dev_err(DEV, "Aborting Connect, can not thaw IO with an only Consistent peer\n");
3609
		tl_clear(mdev->tconn);
3610 3611
		drbd_uuid_new_current(mdev);
		clear_bit(NEW_CUR_UUID, &mdev->flags);
3612
		conn_request_state(mdev->tconn, NS2(conn, C_PROTOCOL_ERROR, susp, 0), CS_HARD);
3613
		return -EIO;
3614
	}
3615
	rv = _drbd_set_state(mdev, ns, cs_flags, NULL);
3616
	ns = drbd_read_state(mdev);
3617
	spin_unlock_irq(&mdev->tconn->req_lock);
P
Philipp Reisner 已提交
3618 3619

	if (rv < SS_SUCCESS) {
3620
		conn_request_state(mdev->tconn, NS(conn, C_DISCONNECTING), CS_HARD);
3621
		return -EIO;
P
Philipp Reisner 已提交
3622 3623
	}

3624 3625
	if (os.conn > C_WF_REPORT_PARAMS) {
		if (ns.conn > C_CONNECTED && peer_state.conn <= C_CONNECTED &&
P
Philipp Reisner 已提交
3626 3627 3628 3629 3630 3631 3632 3633 3634
		    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);
		}
	}

3635
	mdev->tconn->net_conf->want_lose = 0;
P
Philipp Reisner 已提交
3636 3637 3638

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

3639
	return 0;
P
Philipp Reisner 已提交
3640 3641
}

3642
static int receive_sync_uuid(struct drbd_tconn *tconn, struct packet_info *pi)
P
Philipp Reisner 已提交
3643
{
3644
	struct drbd_conf *mdev;
3645
	struct p_rs_uuid *p = pi->data;
3646 3647 3648 3649

	mdev = vnr_to_mdev(tconn, pi->vnr);
	if (!mdev)
		return -EIO;
P
Philipp Reisner 已提交
3650 3651 3652

	wait_event(mdev->misc_wait,
		   mdev->state.conn == C_WF_SYNC_UUID ||
3653
		   mdev->state.conn == C_BEHIND ||
P
Philipp Reisner 已提交
3654 3655 3656 3657 3658 3659 3660 3661 3662 3663 3664
		   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);

3665
		drbd_print_uuids(mdev, "updated sync uuid");
P
Philipp Reisner 已提交
3666 3667 3668 3669 3670 3671
		drbd_start_resync(mdev, C_SYNC_TARGET);

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

3672
	return 0;
P
Philipp Reisner 已提交
3673 3674
}

3675 3676 3677 3678 3679 3680 3681
/**
 * receive_bitmap_plain
 *
 * Return 0 when done, 1 when another iteration is needed, and a negative error
 * code upon failure.
 */
static int
3682
receive_bitmap_plain(struct drbd_conf *mdev, unsigned int size,
3683
		     unsigned long *p, struct bm_xfer_ctx *c)
P
Philipp Reisner 已提交
3684
{
3685 3686
	unsigned int data_size = DRBD_SOCKET_BUFFER_SIZE -
				 drbd_header_size(mdev->tconn);
3687
	unsigned int num_words = min_t(size_t, data_size / sizeof(*p),
3688
				       c->bm_words - c->word_offset);
3689
	unsigned int want = num_words * sizeof(*p);
3690
	int err;
P
Philipp Reisner 已提交
3691

3692 3693
	if (want != size) {
		dev_err(DEV, "%s:want (%u) != size (%u)\n", __func__, want, size);
3694
		return -EIO;
P
Philipp Reisner 已提交
3695 3696
	}
	if (want == 0)
3697
		return 0;
3698
	err = drbd_recv_all(mdev->tconn, p, want);
3699
	if (err)
3700
		return err;
P
Philipp Reisner 已提交
3701

3702
	drbd_bm_merge_lel(mdev, c->word_offset, num_words, p);
P
Philipp Reisner 已提交
3703 3704 3705 3706 3707 3708

	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;

3709
	return 1;
P
Philipp Reisner 已提交
3710 3711
}

3712 3713 3714 3715 3716 3717 3718 3719 3720 3721 3722 3723 3724 3725 3726
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;
}

3727 3728 3729 3730 3731 3732 3733
/**
 * 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
3734 3735
recv_bm_rle_bits(struct drbd_conf *mdev,
		struct p_compressed_bm *p,
3736 3737
		 struct bm_xfer_ctx *c,
		 unsigned int len)
P
Philipp Reisner 已提交
3738 3739 3740 3741 3742 3743 3744
{
	struct bitstream bs;
	u64 look_ahead;
	u64 rl;
	u64 tmp;
	unsigned long s = c->bit_offset;
	unsigned long e;
3745
	int toggle = dcbp_get_start(p);
P
Philipp Reisner 已提交
3746 3747 3748
	int have;
	int bits;

3749
	bitstream_init(&bs, p->code, len, dcbp_get_pad_bits(p));
P
Philipp Reisner 已提交
3750 3751 3752

	bits = bitstream_get_bits(&bs, &look_ahead, 64);
	if (bits < 0)
3753
		return -EIO;
P
Philipp Reisner 已提交
3754 3755 3756 3757

	for (have = bits; have > 0; s += rl, toggle = !toggle) {
		bits = vli_decode_bits(&rl, look_ahead);
		if (bits <= 0)
3758
			return -EIO;
P
Philipp Reisner 已提交
3759 3760 3761 3762 3763

		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);
3764
				return -EIO;
P
Philipp Reisner 已提交
3765 3766 3767 3768 3769 3770 3771 3772 3773
			}
			_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);
3774
			return -EIO;
P
Philipp Reisner 已提交
3775 3776 3777 3778 3779 3780
		}
		look_ahead >>= bits;
		have -= bits;

		bits = bitstream_get_bits(&bs, &tmp, 64 - have);
		if (bits < 0)
3781
			return -EIO;
P
Philipp Reisner 已提交
3782 3783 3784 3785 3786 3787 3788
		look_ahead |= tmp << have;
		have += bits;
	}

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

3789
	return (s != c->bm_bits);
P
Philipp Reisner 已提交
3790 3791
}

3792 3793 3794 3795 3796 3797 3798
/**
 * 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
3799 3800
decode_bitmap_c(struct drbd_conf *mdev,
		struct p_compressed_bm *p,
3801 3802
		struct bm_xfer_ctx *c,
		unsigned int len)
P
Philipp Reisner 已提交
3803
{
3804
	if (dcbp_get_code(p) == RLE_VLI_Bits)
3805
		return recv_bm_rle_bits(mdev, p, c, len - sizeof(*p));
P
Philipp Reisner 已提交
3806 3807 3808 3809 3810 3811

	/* 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);
3812
	conn_request_state(mdev->tconn, NS(conn, C_PROTOCOL_ERROR), CS_HARD);
3813
	return -EIO;
P
Philipp Reisner 已提交
3814 3815 3816 3817 3818 3819
}

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" */
3820 3821 3822 3823 3824 3825 3826
	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
3827 3828 3829 3830 3831 3832 3833 3834 3835 3836 3837 3838 3839 3840 3841 3842 3843 3844 3845 3846 3847 3848 3849 3850 3851 3852 3853 3854 3855 3856 3857 3858 3859

	/* 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. */
3860
static int receive_bitmap(struct drbd_tconn *tconn, struct packet_info *pi)
P
Philipp Reisner 已提交
3861
{
3862
	struct drbd_conf *mdev;
P
Philipp Reisner 已提交
3863
	struct bm_xfer_ctx c;
3864
	int err;
3865 3866 3867 3868

	mdev = vnr_to_mdev(tconn, pi->vnr);
	if (!mdev)
		return -EIO;
P
Philipp Reisner 已提交
3869

3870 3871 3872
	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
3873 3874 3875 3876 3877 3878

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

3879
	for(;;) {
3880 3881 3882
		if (pi->cmd == P_BITMAP)
			err = receive_bitmap_plain(mdev, pi->size, pi->data, &c);
		else if (pi->cmd == P_COMPRESSED_BITMAP) {
P
Philipp Reisner 已提交
3883 3884
			/* MAYBE: sanity check that we speak proto >= 90,
			 * and the feature is enabled! */
3885
			struct p_compressed_bm *p = pi->data;
P
Philipp Reisner 已提交
3886

3887
			if (pi->size > DRBD_SOCKET_BUFFER_SIZE - drbd_header_size(tconn)) {
P
Philipp Reisner 已提交
3888
				dev_err(DEV, "ReportCBitmap packet too large\n");
3889
				err = -EIO;
P
Philipp Reisner 已提交
3890 3891
				goto out;
			}
3892
			if (pi->size <= sizeof(*p)) {
3893
				dev_err(DEV, "ReportCBitmap packet too small (l:%u)\n", pi->size);
3894
				err = -EIO;
3895
				goto out;
P
Philipp Reisner 已提交
3896
			}
3897 3898 3899
			err = drbd_recv_all(mdev->tconn, p, pi->size);
			if (err)
			       goto out;
3900
			err = decode_bitmap_c(mdev, p, &c, pi->size);
P
Philipp Reisner 已提交
3901
		} else {
3902
			dev_warn(DEV, "receive_bitmap: cmd neither ReportBitMap nor ReportCBitMap (is 0x%x)", pi->cmd);
3903
			err = -EIO;
P
Philipp Reisner 已提交
3904 3905 3906
			goto out;
		}

3907
		c.packets[pi->cmd == P_BITMAP]++;
3908
		c.bytes[pi->cmd == P_BITMAP] += drbd_header_size(tconn) + pi->size;
P
Philipp Reisner 已提交
3909

3910 3911 3912
		if (err <= 0) {
			if (err < 0)
				goto out;
P
Philipp Reisner 已提交
3913
			break;
3914
		}
3915
		err = drbd_recv_header(mdev->tconn, pi);
3916
		if (err)
P
Philipp Reisner 已提交
3917
			goto out;
3918
	}
P
Philipp Reisner 已提交
3919 3920 3921 3922

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

	if (mdev->state.conn == C_WF_BITMAP_T) {
3923 3924
		enum drbd_state_rv rv;

3925 3926
		err = drbd_send_bitmap(mdev);
		if (err)
P
Philipp Reisner 已提交
3927 3928
			goto out;
		/* Omit CS_ORDERED with this state transition to avoid deadlocks. */
3929 3930
		rv = _drbd_request_state(mdev, NS(conn, C_WF_SYNC_UUID), CS_VERBOSE);
		D_ASSERT(rv == SS_SUCCESS);
P
Philipp Reisner 已提交
3931 3932 3933 3934 3935 3936
	} 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));
	}
3937
	err = 0;
P
Philipp Reisner 已提交
3938 3939

 out:
3940
	drbd_bm_unlock(mdev);
3941
	if (!err && mdev->state.conn == C_WF_BITMAP_S)
P
Philipp Reisner 已提交
3942
		drbd_start_resync(mdev, C_SYNC_SOURCE);
3943
	return err;
P
Philipp Reisner 已提交
3944 3945
}

3946
static int receive_skip(struct drbd_tconn *tconn, struct packet_info *pi)
P
Philipp Reisner 已提交
3947
{
3948
	conn_warn(tconn, "skipping unknown optional packet type %d, l: %d!\n",
3949
		 pi->cmd, pi->size);
3950

3951
	return ignore_remaining_packet(tconn, pi);
3952 3953
}

3954
static int receive_UnplugRemote(struct drbd_tconn *tconn, struct packet_info *pi)
3955
{
3956 3957
	/* Make sure we've acked all the TCP data associated
	 * with the data requests being unplugged */
3958
	drbd_tcp_quickack(tconn->data.socket);
3959

3960
	return 0;
3961 3962
}

3963
static int receive_out_of_sync(struct drbd_tconn *tconn, struct packet_info *pi)
3964
{
3965
	struct drbd_conf *mdev;
3966
	struct p_block_desc *p = pi->data;
3967 3968 3969 3970

	mdev = vnr_to_mdev(tconn, pi->vnr);
	if (!mdev)
		return -EIO;
3971

3972 3973 3974 3975 3976 3977 3978 3979 3980 3981
	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));
	}

3982 3983
	drbd_set_out_of_sync(mdev, be64_to_cpu(p->sector), be32_to_cpu(p->blksize));

3984
	return 0;
3985 3986
}

3987 3988 3989
struct data_cmd {
	int expect_payload;
	size_t pkt_size;
3990
	int (*fn)(struct drbd_tconn *, struct packet_info *);
3991 3992 3993
};

static struct data_cmd drbd_cmd_handler[] = {
3994 3995 3996 3997
	[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 } ,
3998 3999 4000
	[P_BITMAP]	    = { 1, 0, receive_bitmap } ,
	[P_COMPRESSED_BITMAP] = { 1, 0, receive_bitmap } ,
	[P_UNPLUG_REMOTE]   = { 0, 0, receive_UnplugRemote },
4001 4002
	[P_DATA_REQUEST]    = { 0, sizeof(struct p_block_req), receive_DataRequest },
	[P_RS_DATA_REQUEST] = { 0, sizeof(struct p_block_req), receive_DataRequest },
4003 4004
	[P_SYNC_PARAM]	    = { 1, 0, receive_SyncParam },
	[P_SYNC_PARAM89]    = { 1, 0, receive_SyncParam },
4005 4006 4007 4008 4009 4010 4011 4012 4013 4014 4015 4016
	[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
4017 4018
};

4019
static void drbdd(struct drbd_tconn *tconn)
P
Philipp Reisner 已提交
4020
{
4021
	struct packet_info pi;
4022
	size_t shs; /* sub header size */
4023
	int err;
P
Philipp Reisner 已提交
4024

4025
	while (get_t_state(&tconn->receiver) == RUNNING) {
4026 4027
		struct data_cmd *cmd;

4028
		drbd_thread_current_set_cpu(&tconn->receiver);
4029
		if (drbd_recv_header(tconn, &pi))
4030
			goto err_out;
P
Philipp Reisner 已提交
4031

4032
		cmd = &drbd_cmd_handler[pi.cmd];
4033
		if (unlikely(pi.cmd >= ARRAY_SIZE(drbd_cmd_handler) || !cmd->fn)) {
4034
			conn_err(tconn, "unknown packet type %d, l: %d!\n", pi.cmd, pi.size);
4035
			goto err_out;
4036
		}
P
Philipp Reisner 已提交
4037

4038 4039
		shs = cmd->pkt_size;
		if (pi.size > shs && !cmd->expect_payload) {
4040
			conn_err(tconn, "No payload expected %s l:%d\n", cmdname(pi.cmd), pi.size);
4041
			goto err_out;
P
Philipp Reisner 已提交
4042 4043
		}

4044
		if (shs) {
4045
			err = drbd_recv_all_warn(tconn, pi.data, shs);
4046
			if (err)
4047
				goto err_out;
4048
			pi.size -= shs;
4049 4050
		}

4051 4052
		err = cmd->fn(tconn, &pi);
		if (err) {
4053 4054
			conn_err(tconn, "error receiving %s, e: %d l: %d!\n",
				 cmdname(pi.cmd), err, pi.size);
4055
			goto err_out;
P
Philipp Reisner 已提交
4056 4057
		}
	}
4058
	return;
P
Philipp Reisner 已提交
4059

4060 4061
    err_out:
	conn_request_state(tconn, NS(conn, C_PROTOCOL_ERROR), CS_HARD);
P
Philipp Reisner 已提交
4062 4063
}

4064
void conn_flush_workqueue(struct drbd_tconn *tconn)
P
Philipp Reisner 已提交
4065 4066 4067 4068
{
	struct drbd_wq_barrier barr;

	barr.w.cb = w_prev_work_done;
4069
	barr.w.tconn = tconn;
P
Philipp Reisner 已提交
4070
	init_completion(&barr.done);
4071
	drbd_queue_work(&tconn->data.work, &barr.w);
P
Philipp Reisner 已提交
4072 4073 4074
	wait_for_completion(&barr.done);
}

4075
static void drbd_disconnect(struct drbd_tconn *tconn)
P
Philipp Reisner 已提交
4076
{
4077
	enum drbd_conns oc;
P
Philipp Reisner 已提交
4078 4079
	int rv = SS_UNKNOWN_ERROR;

4080
	if (tconn->cstate == C_STANDALONE)
P
Philipp Reisner 已提交
4081 4082 4083
		return;

	/* asender does not clean up anything. it must not interfere, either */
4084 4085 4086 4087 4088 4089
	drbd_thread_stop(&tconn->asender);
	drbd_free_sock(tconn);

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

4090 4091 4092
	if (conn_highest_role(tconn) == R_PRIMARY && conn_highest_pdsk(tconn) >= D_UNKNOWN)
		conn_try_outdate_peer_async(tconn);

4093
	spin_lock_irq(&tconn->req_lock);
4094 4095 4096 4097
	oc = tconn->cstate;
	if (oc >= C_UNCONNECTED)
		rv = _conn_request_state(tconn, NS(conn, C_UNCONNECTED), CS_VERBOSE);

4098 4099
	spin_unlock_irq(&tconn->req_lock);

4100
	if (oc == C_DISCONNECTING) {
4101 4102 4103 4104 4105 4106 4107
		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;
4108
		conn_request_state(tconn, NS(conn, C_STANDALONE), CS_VERBOSE);
4109 4110 4111 4112 4113 4114 4115 4116
	}
}

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
4117

4118
	/* wait for current activity to cease. */
4119
	spin_lock_irq(&mdev->tconn->req_lock);
P
Philipp Reisner 已提交
4120 4121 4122
	_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);
4123
	spin_unlock_irq(&mdev->tconn->req_lock);
P
Philipp Reisner 已提交
4124 4125 4126 4127 4128 4129 4130 4131 4132 4133 4134 4135 4136 4137 4138 4139 4140

	/* 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);

4141 4142
	del_timer(&mdev->request_timer);

P
Philipp Reisner 已提交
4143 4144 4145 4146 4147 4148
	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" */
4149
	drbd_flush_workqueue(mdev);
P
Philipp Reisner 已提交
4150

4151
	drbd_finish_peer_reqs(mdev);
P
Philipp Reisner 已提交
4152 4153 4154 4155

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

4156
	if (!drbd_suspended(mdev))
4157
		tl_clear(mdev->tconn);
P
Philipp Reisner 已提交
4158 4159 4160 4161 4162 4163 4164 4165 4166

	drbd_md_sync(mdev);

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

4167 4168 4169 4170
	/* 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
4171 4172 4173 4174 4175 4176 4177
	/* 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.
	 */
4178
	i = drbd_free_peer_reqs(mdev, &mdev->net_ee);
P
Philipp Reisner 已提交
4179 4180
	if (i)
		dev_info(DEV, "net_ee not empty, killed %u entries\n", i);
4181 4182 4183
	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
4184 4185
	i = atomic_read(&mdev->pp_in_use);
	if (i)
4186
		dev_info(DEV, "pp_in_use = %d, expected 0\n", i);
P
Philipp Reisner 已提交
4187 4188 4189 4190 4191 4192 4193 4194 4195

	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));
4196 4197

	return 0;
P
Philipp Reisner 已提交
4198 4199 4200 4201 4202 4203 4204 4205 4206 4207 4208
}

/*
 * 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.
 */
4209
static int drbd_send_features(struct drbd_tconn *tconn)
P
Philipp Reisner 已提交
4210
{
4211 4212
	struct drbd_socket *sock;
	struct p_connection_features *p;
P
Philipp Reisner 已提交
4213

4214 4215 4216
	sock = &tconn->data;
	p = conn_prepare_command(tconn, sock);
	if (!p)
4217
		return -EIO;
P
Philipp Reisner 已提交
4218 4219 4220
	memset(p, 0, sizeof(*p));
	p->protocol_min = cpu_to_be32(PRO_VERSION_MIN);
	p->protocol_max = cpu_to_be32(PRO_VERSION_MAX);
4221
	return conn_send_command(tconn, sock, P_CONNECTION_FEATURES, sizeof(*p), NULL, 0);
P
Philipp Reisner 已提交
4222 4223 4224 4225 4226 4227 4228 4229 4230
}

/*
 * 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.
 */
4231
static int drbd_do_features(struct drbd_tconn *tconn)
P
Philipp Reisner 已提交
4232
{
4233
	/* ASSERT current == tconn->receiver ... */
4234 4235
	struct p_connection_features *p;
	const int expect = sizeof(struct p_connection_features);
4236
	struct packet_info pi;
4237
	int err;
P
Philipp Reisner 已提交
4238

4239
	err = drbd_send_features(tconn);
4240
	if (err)
P
Philipp Reisner 已提交
4241 4242
		return 0;

4243 4244
	err = drbd_recv_header(tconn, &pi);
	if (err)
P
Philipp Reisner 已提交
4245 4246
		return 0;

4247 4248
	if (pi.cmd != P_CONNECTION_FEATURES) {
		conn_err(tconn, "expected ConnectionFeatures packet, received: %s (0x%04x)\n",
4249
		     cmdname(pi.cmd), pi.cmd);
P
Philipp Reisner 已提交
4250 4251 4252
		return -1;
	}

4253
	if (pi.size != expect) {
4254
		conn_err(tconn, "expected ConnectionFeatures length: %u, received: %u\n",
4255
		     expect, pi.size);
P
Philipp Reisner 已提交
4256 4257 4258
		return -1;
	}

4259 4260
	p = pi.data;
	err = drbd_recv_all_warn(tconn, p, expect);
4261
	if (err)
P
Philipp Reisner 已提交
4262 4263 4264 4265 4266 4267 4268 4269 4270 4271 4272
		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;

4273
	tconn->agreed_pro_version = min_t(int, PRO_VERSION_MAX, p->protocol_max);
P
Philipp Reisner 已提交
4274

4275 4276
	conn_info(tconn, "Handshake successful: "
	     "Agreed network protocol version %d\n", tconn->agreed_pro_version);
P
Philipp Reisner 已提交
4277 4278 4279 4280

	return 1;

 incompat:
4281
	conn_err(tconn, "incompatible DRBD dialects: "
P
Philipp Reisner 已提交
4282 4283 4284 4285 4286 4287 4288
	    "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)
4289
static int drbd_do_auth(struct drbd_tconn *tconn)
P
Philipp Reisner 已提交
4290 4291 4292
{
	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");
4293
	return -1;
P
Philipp Reisner 已提交
4294 4295 4296
}
#else
#define CHALLENGE_LEN 64
4297 4298 4299 4300 4301 4302 4303

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

4304
static int drbd_do_auth(struct drbd_tconn *tconn)
P
Philipp Reisner 已提交
4305
{
4306
	struct drbd_socket *sock;
P
Philipp Reisner 已提交
4307 4308 4309 4310 4311
	char my_challenge[CHALLENGE_LEN];  /* 64 Bytes... */
	struct scatterlist sg;
	char *response = NULL;
	char *right_response = NULL;
	char *peers_ch = NULL;
4312
	unsigned int key_len = strlen(tconn->net_conf->shared_secret);
P
Philipp Reisner 已提交
4313 4314
	unsigned int resp_size;
	struct hash_desc desc;
4315
	struct packet_info pi;
4316
	int err, rv;
P
Philipp Reisner 已提交
4317

4318 4319
	/* FIXME: Put the challenge/response into the preallocated socket buffer.  */

4320
	desc.tfm = tconn->cram_hmac_tfm;
P
Philipp Reisner 已提交
4321 4322
	desc.flags = 0;

4323 4324
	rv = crypto_hash_setkey(tconn->cram_hmac_tfm,
				(u8 *)tconn->net_conf->shared_secret, key_len);
P
Philipp Reisner 已提交
4325
	if (rv) {
4326
		conn_err(tconn, "crypto_hash_setkey() failed with %d\n", rv);
4327
		rv = -1;
P
Philipp Reisner 已提交
4328 4329 4330 4331 4332
		goto fail;
	}

	get_random_bytes(my_challenge, CHALLENGE_LEN);

4333 4334 4335 4336 4337
	sock = &tconn->data;
	if (!conn_prepare_command(tconn, sock)) {
		rv = 0;
		goto fail;
	}
4338
	rv = !conn_send_command(tconn, sock, P_AUTH_CHALLENGE, 0,
4339
				my_challenge, CHALLENGE_LEN);
P
Philipp Reisner 已提交
4340 4341 4342
	if (!rv)
		goto fail;

4343 4344 4345
	err = drbd_recv_header(tconn, &pi);
	if (err) {
		rv = 0;
P
Philipp Reisner 已提交
4346
		goto fail;
4347
	}
P
Philipp Reisner 已提交
4348

4349
	if (pi.cmd != P_AUTH_CHALLENGE) {
4350
		conn_err(tconn, "expected AuthChallenge packet, received: %s (0x%04x)\n",
4351
		    cmdname(pi.cmd), pi.cmd);
P
Philipp Reisner 已提交
4352 4353 4354 4355
		rv = 0;
		goto fail;
	}

4356
	if (pi.size > CHALLENGE_LEN * 2) {
4357
		conn_err(tconn, "expected AuthChallenge payload too big.\n");
4358
		rv = -1;
P
Philipp Reisner 已提交
4359 4360 4361
		goto fail;
	}

4362
	peers_ch = kmalloc(pi.size, GFP_NOIO);
P
Philipp Reisner 已提交
4363
	if (peers_ch == NULL) {
4364
		conn_err(tconn, "kmalloc of peers_ch failed\n");
4365
		rv = -1;
P
Philipp Reisner 已提交
4366 4367 4368
		goto fail;
	}

4369 4370
	err = drbd_recv_all_warn(tconn, peers_ch, pi.size);
	if (err) {
P
Philipp Reisner 已提交
4371 4372 4373 4374
		rv = 0;
		goto fail;
	}

4375
	resp_size = crypto_hash_digestsize(tconn->cram_hmac_tfm);
P
Philipp Reisner 已提交
4376 4377
	response = kmalloc(resp_size, GFP_NOIO);
	if (response == NULL) {
4378
		conn_err(tconn, "kmalloc of response failed\n");
4379
		rv = -1;
P
Philipp Reisner 已提交
4380 4381 4382 4383
		goto fail;
	}

	sg_init_table(&sg, 1);
4384
	sg_set_buf(&sg, peers_ch, pi.size);
P
Philipp Reisner 已提交
4385 4386 4387

	rv = crypto_hash_digest(&desc, &sg, sg.length, response);
	if (rv) {
4388
		conn_err(tconn, "crypto_hash_digest() failed with %d\n", rv);
4389
		rv = -1;
P
Philipp Reisner 已提交
4390 4391 4392
		goto fail;
	}

4393 4394 4395 4396
	if (!conn_prepare_command(tconn, sock)) {
		rv = 0;
		goto fail;
	}
4397
	rv = !conn_send_command(tconn, sock, P_AUTH_RESPONSE, 0,
4398
				response, resp_size);
P
Philipp Reisner 已提交
4399 4400 4401
	if (!rv)
		goto fail;

4402 4403 4404
	err = drbd_recv_header(tconn, &pi);
	if (err) {
		rv = 0;
P
Philipp Reisner 已提交
4405
		goto fail;
4406
	}
P
Philipp Reisner 已提交
4407

4408
	if (pi.cmd != P_AUTH_RESPONSE) {
4409
		conn_err(tconn, "expected AuthResponse packet, received: %s (0x%04x)\n",
4410
			cmdname(pi.cmd), pi.cmd);
P
Philipp Reisner 已提交
4411 4412 4413 4414
		rv = 0;
		goto fail;
	}

4415
	if (pi.size != resp_size) {
4416
		conn_err(tconn, "expected AuthResponse payload of wrong size\n");
P
Philipp Reisner 已提交
4417 4418 4419 4420
		rv = 0;
		goto fail;
	}

4421 4422
	err = drbd_recv_all_warn(tconn, response , resp_size);
	if (err) {
P
Philipp Reisner 已提交
4423 4424 4425 4426 4427
		rv = 0;
		goto fail;
	}

	right_response = kmalloc(resp_size, GFP_NOIO);
4428
	if (right_response == NULL) {
4429
		conn_err(tconn, "kmalloc of right_response failed\n");
4430
		rv = -1;
P
Philipp Reisner 已提交
4431 4432 4433 4434 4435 4436 4437
		goto fail;
	}

	sg_set_buf(&sg, my_challenge, CHALLENGE_LEN);

	rv = crypto_hash_digest(&desc, &sg, sg.length, right_response);
	if (rv) {
4438
		conn_err(tconn, "crypto_hash_digest() failed with %d\n", rv);
4439
		rv = -1;
P
Philipp Reisner 已提交
4440 4441 4442 4443 4444 4445
		goto fail;
	}

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

	if (rv)
4446 4447
		conn_info(tconn, "Peer authenticated using %d bytes of '%s' HMAC\n",
		     resp_size, tconn->net_conf->cram_hmac_alg);
4448 4449
	else
		rv = -1;
P
Philipp Reisner 已提交
4450 4451 4452 4453 4454 4455 4456 4457 4458 4459 4460 4461

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

	return rv;
}
#endif

int drbdd_init(struct drbd_thread *thi)
{
4462
	struct drbd_tconn *tconn = thi->tconn;
P
Philipp Reisner 已提交
4463 4464
	int h;

4465
	conn_info(tconn, "receiver (re)started\n");
P
Philipp Reisner 已提交
4466 4467

	do {
4468
		h = drbd_connect(tconn);
P
Philipp Reisner 已提交
4469
		if (h == 0) {
4470
			drbd_disconnect(tconn);
4471
			schedule_timeout_interruptible(HZ);
P
Philipp Reisner 已提交
4472 4473
		}
		if (h == -1) {
4474
			conn_warn(tconn, "Discarding network configuration.\n");
4475
			conn_request_state(tconn, NS(conn, C_DISCONNECTING), CS_HARD);
P
Philipp Reisner 已提交
4476 4477 4478 4479
		}
	} while (h == 0);

	if (h > 0) {
4480 4481 4482
		if (get_net_conf(tconn)) {
			drbdd(tconn);
			put_net_conf(tconn);
P
Philipp Reisner 已提交
4483 4484 4485
		}
	}

4486
	drbd_disconnect(tconn);
P
Philipp Reisner 已提交
4487

4488
	conn_info(tconn, "receiver terminated\n");
P
Philipp Reisner 已提交
4489 4490 4491 4492 4493
	return 0;
}

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

4494
static int got_conn_RqSReply(struct drbd_tconn *tconn, struct packet_info *pi)
4495
{
4496
	struct p_req_state_reply *p = pi->data;
4497 4498 4499 4500 4501 4502 4503 4504 4505 4506 4507
	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);

4508
	return 0;
4509 4510
}

4511
static int got_RqSReply(struct drbd_tconn *tconn, struct packet_info *pi)
P
Philipp Reisner 已提交
4512
{
4513
	struct drbd_conf *mdev;
4514
	struct p_req_state_reply *p = pi->data;
P
Philipp Reisner 已提交
4515 4516
	int retcode = be32_to_cpu(p->retcode);

4517 4518
	mdev = vnr_to_mdev(tconn, pi->vnr);
	if (!mdev)
4519
		return -EIO;
4520

4521 4522 4523 4524 4525 4526
	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
4527
	}
4528 4529
	wake_up(&mdev->state_wait);

4530
	return 0;
P
Philipp Reisner 已提交
4531 4532
}

4533
static int got_Ping(struct drbd_tconn *tconn, struct packet_info *pi)
P
Philipp Reisner 已提交
4534
{
4535
	return drbd_send_ping_ack(tconn);
P
Philipp Reisner 已提交
4536 4537 4538

}

4539
static int got_PingAck(struct drbd_tconn *tconn, struct packet_info *pi)
P
Philipp Reisner 已提交
4540 4541
{
	/* restore idle timeout */
4542 4543 4544
	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
4545

4546
	return 0;
P
Philipp Reisner 已提交
4547 4548
}

4549
static int got_IsInSync(struct drbd_tconn *tconn, struct packet_info *pi)
P
Philipp Reisner 已提交
4550
{
4551
	struct drbd_conf *mdev;
4552
	struct p_block_ack *p = pi->data;
P
Philipp Reisner 已提交
4553 4554 4555
	sector_t sector = be64_to_cpu(p->sector);
	int blksize = be32_to_cpu(p->blksize);

4556 4557
	mdev = vnr_to_mdev(tconn, pi->vnr);
	if (!mdev)
4558
		return -EIO;
4559

4560
	D_ASSERT(mdev->tconn->agreed_pro_version >= 89);
P
Philipp Reisner 已提交
4561 4562 4563

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

4564 4565 4566 4567 4568 4569 4570
	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
4571
	dec_rs_pending(mdev);
4572
	atomic_add(blksize >> 9, &mdev->rs_sect_in);
P
Philipp Reisner 已提交
4573

4574
	return 0;
P
Philipp Reisner 已提交
4575 4576
}

4577 4578 4579 4580
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
4581 4582 4583 4584
{
	struct drbd_request *req;
	struct bio_and_error m;

4585
	spin_lock_irq(&mdev->tconn->req_lock);
4586
	req = find_request(mdev, root, id, sector, missing_ok, func);
P
Philipp Reisner 已提交
4587
	if (unlikely(!req)) {
4588
		spin_unlock_irq(&mdev->tconn->req_lock);
4589
		return -EIO;
P
Philipp Reisner 已提交
4590 4591
	}
	__req_mod(req, what, &m);
4592
	spin_unlock_irq(&mdev->tconn->req_lock);
P
Philipp Reisner 已提交
4593 4594 4595

	if (m.bio)
		complete_master_bio(mdev, &m);
4596
	return 0;
P
Philipp Reisner 已提交
4597 4598
}

4599
static int got_BlockAck(struct drbd_tconn *tconn, struct packet_info *pi)
P
Philipp Reisner 已提交
4600
{
4601
	struct drbd_conf *mdev;
4602
	struct p_block_ack *p = pi->data;
P
Philipp Reisner 已提交
4603 4604 4605 4606
	sector_t sector = be64_to_cpu(p->sector);
	int blksize = be32_to_cpu(p->blksize);
	enum drbd_req_event what;

4607 4608
	mdev = vnr_to_mdev(tconn, pi->vnr);
	if (!mdev)
4609
		return -EIO;
4610

P
Philipp Reisner 已提交
4611 4612
	update_peer_seq(mdev, be32_to_cpu(p->seq_num));

4613
	if (p->block_id == ID_SYNCER) {
P
Philipp Reisner 已提交
4614 4615
		drbd_set_in_sync(mdev, sector, blksize);
		dec_rs_pending(mdev);
4616
		return 0;
P
Philipp Reisner 已提交
4617
	}
4618
	switch (pi->cmd) {
P
Philipp Reisner 已提交
4619
	case P_RS_WRITE_ACK:
4620
		D_ASSERT(mdev->tconn->net_conf->wire_protocol == DRBD_PROT_C);
4621
		what = WRITE_ACKED_BY_PEER_AND_SIS;
P
Philipp Reisner 已提交
4622 4623
		break;
	case P_WRITE_ACK:
4624
		D_ASSERT(mdev->tconn->net_conf->wire_protocol == DRBD_PROT_C);
4625
		what = WRITE_ACKED_BY_PEER;
P
Philipp Reisner 已提交
4626 4627
		break;
	case P_RECV_ACK:
4628
		D_ASSERT(mdev->tconn->net_conf->wire_protocol == DRBD_PROT_B);
4629
		what = RECV_ACKED_BY_PEER;
P
Philipp Reisner 已提交
4630
		break;
4631
	case P_DISCARD_WRITE:
4632
		D_ASSERT(mdev->tconn->net_conf->wire_protocol == DRBD_PROT_C);
4633 4634 4635 4636 4637
		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
4638 4639
		break;
	default:
4640
		BUG();
P
Philipp Reisner 已提交
4641 4642
	}

4643 4644 4645
	return validate_req_change_req_state(mdev, p->block_id, sector,
					     &mdev->write_requests, __func__,
					     what, false);
P
Philipp Reisner 已提交
4646 4647
}

4648
static int got_NegAck(struct drbd_tconn *tconn, struct packet_info *pi)
P
Philipp Reisner 已提交
4649
{
4650
	struct drbd_conf *mdev;
4651
	struct p_block_ack *p = pi->data;
P
Philipp Reisner 已提交
4652
	sector_t sector = be64_to_cpu(p->sector);
4653
	int size = be32_to_cpu(p->blksize);
4654 4655
	bool missing_ok = tconn->net_conf->wire_protocol == DRBD_PROT_A ||
			  tconn->net_conf->wire_protocol == DRBD_PROT_B;
4656
	int err;
P
Philipp Reisner 已提交
4657

4658 4659
	mdev = vnr_to_mdev(tconn, pi->vnr);
	if (!mdev)
4660
		return -EIO;
4661

P
Philipp Reisner 已提交
4662 4663
	update_peer_seq(mdev, be32_to_cpu(p->seq_num));

4664
	if (p->block_id == ID_SYNCER) {
P
Philipp Reisner 已提交
4665 4666
		dec_rs_pending(mdev);
		drbd_rs_failed_io(mdev, sector, size);
4667
		return 0;
P
Philipp Reisner 已提交
4668
	}
4669

4670 4671 4672 4673
	err = validate_req_change_req_state(mdev, p->block_id, sector,
					    &mdev->write_requests, __func__,
					    NEG_ACKED, missing_ok);
	if (err) {
4674 4675 4676 4677 4678 4679
		/* 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)
4680
			return err;
4681
		drbd_set_out_of_sync(mdev, sector, size);
4682
	}
4683
	return 0;
P
Philipp Reisner 已提交
4684 4685
}

4686
static int got_NegDReply(struct drbd_tconn *tconn, struct packet_info *pi)
P
Philipp Reisner 已提交
4687
{
4688
	struct drbd_conf *mdev;
4689
	struct p_block_ack *p = pi->data;
P
Philipp Reisner 已提交
4690 4691
	sector_t sector = be64_to_cpu(p->sector);

4692 4693
	mdev = vnr_to_mdev(tconn, pi->vnr);
	if (!mdev)
4694
		return -EIO;
4695

P
Philipp Reisner 已提交
4696
	update_peer_seq(mdev, be32_to_cpu(p->seq_num));
4697

P
Philipp Reisner 已提交
4698 4699 4700
	dev_err(DEV, "Got NegDReply; Sector %llus, len %u; Fail original request.\n",
	    (unsigned long long)sector, be32_to_cpu(p->blksize));

4701 4702 4703
	return validate_req_change_req_state(mdev, p->block_id, sector,
					     &mdev->read_requests, __func__,
					     NEG_ACKED, false);
P
Philipp Reisner 已提交
4704 4705
}

4706
static int got_NegRSDReply(struct drbd_tconn *tconn, struct packet_info *pi)
P
Philipp Reisner 已提交
4707
{
4708
	struct drbd_conf *mdev;
P
Philipp Reisner 已提交
4709 4710
	sector_t sector;
	int size;
4711
	struct p_block_ack *p = pi->data;
4712 4713 4714

	mdev = vnr_to_mdev(tconn, pi->vnr);
	if (!mdev)
4715
		return -EIO;
P
Philipp Reisner 已提交
4716 4717 4718 4719 4720 4721 4722 4723 4724 4725

	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);
4726
		switch (pi->cmd) {
4727 4728 4729 4730 4731
		case P_NEG_RS_DREPLY:
			drbd_rs_failed_io(mdev, sector, size);
		case P_RS_CANCEL:
			break;
		default:
4732
			BUG();
4733
		}
P
Philipp Reisner 已提交
4734 4735 4736
		put_ldev(mdev);
	}

4737
	return 0;
P
Philipp Reisner 已提交
4738 4739
}

4740
static int got_BarrierAck(struct drbd_tconn *tconn, struct packet_info *pi)
P
Philipp Reisner 已提交
4741
{
4742
	struct drbd_conf *mdev;
4743
	struct p_barrier_ack *p = pi->data;
4744 4745 4746

	mdev = vnr_to_mdev(tconn, pi->vnr);
	if (!mdev)
4747
		return -EIO;
P
Philipp Reisner 已提交
4748

4749
	tl_release(mdev->tconn, p->barrier, be32_to_cpu(p->set_size));
P
Philipp Reisner 已提交
4750

4751 4752
	if (mdev->state.conn == C_AHEAD &&
	    atomic_read(&mdev->ap_in_flight) == 0 &&
4753 4754 4755
	    !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);
4756 4757
	}

4758
	return 0;
P
Philipp Reisner 已提交
4759 4760
}

4761
static int got_OVResult(struct drbd_tconn *tconn, struct packet_info *pi)
P
Philipp Reisner 已提交
4762
{
4763
	struct drbd_conf *mdev;
4764
	struct p_block_ack *p = pi->data;
P
Philipp Reisner 已提交
4765 4766 4767 4768
	struct drbd_work *w;
	sector_t sector;
	int size;

4769 4770
	mdev = vnr_to_mdev(tconn, pi->vnr);
	if (!mdev)
4771
		return -EIO;
4772

P
Philipp Reisner 已提交
4773 4774 4775 4776 4777 4778
	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)
4779
		drbd_ov_out_of_sync_found(mdev, sector, size);
P
Philipp Reisner 已提交
4780
	else
4781
		ov_out_of_sync_print(mdev);
P
Philipp Reisner 已提交
4782

4783
	if (!get_ldev(mdev))
4784
		return 0;
4785

P
Philipp Reisner 已提交
4786 4787 4788
	drbd_rs_complete_io(mdev, sector);
	dec_rs_pending(mdev);

4789 4790 4791 4792 4793 4794 4795
	--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
4796 4797 4798
		w = kmalloc(sizeof(*w), GFP_NOIO);
		if (w) {
			w->cb = w_ov_finished;
4799
			w->mdev = mdev;
4800
			drbd_queue_work_front(&mdev->tconn->data.work, w);
P
Philipp Reisner 已提交
4801 4802
		} else {
			dev_err(DEV, "kmalloc(w) failed.");
4803
			ov_out_of_sync_print(mdev);
P
Philipp Reisner 已提交
4804 4805 4806
			drbd_resync_finished(mdev);
		}
	}
4807
	put_ldev(mdev);
4808
	return 0;
P
Philipp Reisner 已提交
4809 4810
}

4811
static int got_skip(struct drbd_tconn *tconn, struct packet_info *pi)
4812
{
4813
	return 0;
4814 4815
}

4816
static int tconn_finish_peer_reqs(struct drbd_tconn *tconn)
4817
{
4818 4819
	struct drbd_conf *mdev;
	int i, not_empty = 0;
4820 4821 4822 4823

	do {
		clear_bit(SIGNAL_ASENDER, &tconn->flags);
		flush_signals(current);
4824
		idr_for_each_entry(&tconn->volumes, mdev, i) {
4825
			if (drbd_finish_peer_reqs(mdev))
4826 4827
				return 1; /* error */
		}
4828
		set_bit(SIGNAL_ASENDER, &tconn->flags);
4829 4830 4831 4832 4833 4834 4835 4836

		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);
4837 4838 4839 4840 4841
	} while (not_empty);

	return 0;
}

4842 4843
struct asender_cmd {
	size_t pkt_size;
4844
	int (*fn)(struct drbd_tconn *tconn, struct packet_info *);
4845 4846 4847
};

static struct asender_cmd asender_tbl[] = {
4848 4849
	[P_PING]	    = { 0, got_Ping },
	[P_PING_ACK]	    = { 0, got_PingAck },
4850 4851 4852 4853 4854 4855 4856 4857 4858 4859 4860 4861 4862 4863 4864
	[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 },
4865 4866
};

P
Philipp Reisner 已提交
4867 4868
int drbd_asender(struct drbd_thread *thi)
{
4869
	struct drbd_tconn *tconn = thi->tconn;
P
Philipp Reisner 已提交
4870
	struct asender_cmd *cmd = NULL;
4871
	struct packet_info pi;
4872
	int rv;
4873
	void *buf    = tconn->meta.rbuf;
P
Philipp Reisner 已提交
4874
	int received = 0;
4875 4876
	unsigned int header_size = drbd_header_size(tconn);
	int expect   = header_size;
4877
	int ping_timeout_active = 0;
P
Philipp Reisner 已提交
4878 4879 4880 4881

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

4882
	while (get_t_state(thi) == RUNNING) {
4883
		drbd_thread_current_set_cpu(thi);
4884
		if (test_and_clear_bit(SEND_PING, &tconn->flags)) {
4885
			if (drbd_send_ping(tconn)) {
4886
				conn_err(tconn, "drbd_send_ping has failed\n");
4887 4888
				goto reconnect;
			}
4889 4890
			tconn->meta.socket->sk->sk_rcvtimeo =
				tconn->net_conf->ping_timeo*HZ/10;
4891
			ping_timeout_active = 1;
P
Philipp Reisner 已提交
4892 4893
		}

4894 4895 4896 4897
		/* 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);
4898 4899
		if (tconn_finish_peer_reqs(tconn)) {
			conn_err(tconn, "tconn_finish_peer_reqs() failed\n");
4900
			goto reconnect;
4901
		}
P
Philipp Reisner 已提交
4902
		/* but unconditionally uncork unless disabled */
4903 4904
		if (!tconn->net_conf->no_cork)
			drbd_tcp_uncork(tconn->meta.socket);
P
Philipp Reisner 已提交
4905 4906 4907 4908 4909

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

4910 4911
		rv = drbd_recv_short(tconn->meta.socket, buf, expect-received, 0);
		clear_bit(SIGNAL_ASENDER, &tconn->flags);
P
Philipp Reisner 已提交
4912 4913 4914 4915 4916 4917 4918 4919 4920 4921 4922 4923 4924 4925 4926 4927 4928

		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) {
4929
			conn_err(tconn, "meta connection shut down by peer.\n");
P
Philipp Reisner 已提交
4930 4931
			goto reconnect;
		} else if (rv == -EAGAIN) {
4932 4933
			/* If the data socket received something meanwhile,
			 * that is good enough: peer is still alive. */
4934 4935
			if (time_after(tconn->last_received,
				jiffies - tconn->meta.socket->sk->sk_rcvtimeo))
4936
				continue;
4937
			if (ping_timeout_active) {
4938
				conn_err(tconn, "PingAck did not arrive in time.\n");
P
Philipp Reisner 已提交
4939 4940
				goto reconnect;
			}
4941
			set_bit(SEND_PING, &tconn->flags);
P
Philipp Reisner 已提交
4942 4943 4944 4945
			continue;
		} else if (rv == -EINTR) {
			continue;
		} else {
4946
			conn_err(tconn, "sock_recvmsg returned %d\n", rv);
P
Philipp Reisner 已提交
4947 4948 4949 4950
			goto reconnect;
		}

		if (received == expect && cmd == NULL) {
4951
			if (decode_header(tconn, tconn->meta.rbuf, &pi))
P
Philipp Reisner 已提交
4952
				goto reconnect;
4953
			cmd = &asender_tbl[pi.cmd];
4954
			if (pi.cmd >= ARRAY_SIZE(asender_tbl) || !cmd->fn) {
4955
				conn_err(tconn, "unknown command %d on meta (l: %d)\n",
4956
					pi.cmd, pi.size);
P
Philipp Reisner 已提交
4957 4958
				goto disconnect;
			}
4959
			expect = header_size + cmd->pkt_size;
4960
			if (pi.size != expect - header_size) {
4961
				conn_err(tconn, "Wrong packet size on meta (c: %d, l: %d)\n",
4962
					pi.cmd, pi.size);
P
Philipp Reisner 已提交
4963
				goto reconnect;
4964
			}
P
Philipp Reisner 已提交
4965 4966
		}
		if (received == expect) {
4967
			bool err;
4968

4969 4970
			err = cmd->fn(tconn, &pi);
			if (err) {
4971
				conn_err(tconn, "%pf failed\n", cmd->fn);
P
Philipp Reisner 已提交
4972
				goto reconnect;
4973
			}
P
Philipp Reisner 已提交
4974

4975 4976
			tconn->last_received = jiffies;

4977 4978
			/* the idle_timeout (ping-int)
			 * has been restored in got_PingAck() */
4979
			if (cmd == &asender_tbl[P_PING_ACK])
4980 4981
				ping_timeout_active = 0;

4982
			buf	 = tconn->meta.rbuf;
P
Philipp Reisner 已提交
4983
			received = 0;
4984
			expect	 = header_size;
P
Philipp Reisner 已提交
4985 4986 4987 4988 4989 4990
			cmd	 = NULL;
		}
	}

	if (0) {
reconnect:
4991
		conn_request_state(tconn, NS(conn, C_NETWORK_FAILURE), CS_HARD);
P
Philipp Reisner 已提交
4992 4993 4994
	}
	if (0) {
disconnect:
4995
		conn_request_state(tconn, NS(conn, C_DISCONNECTING), CS_HARD);
P
Philipp Reisner 已提交
4996
	}
4997
	clear_bit(SIGNAL_ASENDER, &tconn->flags);
P
Philipp Reisner 已提交
4998

4999
	conn_info(tconn, "asender terminated\n");
P
Philipp Reisner 已提交
5000 5001 5002

	return 0;
}