drbd_main.c 95.2 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
/*
   drbd.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>.

   Thanks to Carter Burden, Bart Grantham and Gennadiy Nerubayev
   from Logicworks, Inc. for making SDP replication support possible.

   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 <linux/drbd.h>
#include <asm/uaccess.h>
#include <asm/types.h>
#include <net/sock.h>
#include <linux/ctype.h>
35
#include <linux/mutex.h>
P
Philipp Reisner 已提交
36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58
#include <linux/fs.h>
#include <linux/file.h>
#include <linux/proc_fs.h>
#include <linux/init.h>
#include <linux/mm.h>
#include <linux/memcontrol.h>
#include <linux/mm_inline.h>
#include <linux/slab.h>
#include <linux/random.h>
#include <linux/reboot.h>
#include <linux/notifier.h>
#include <linux/kthread.h>

#define __KERNEL_SYSCALLS__
#include <linux/unistd.h>
#include <linux/vmalloc.h>

#include <linux/drbd_limits.h>
#include "drbd_int.h"
#include "drbd_req.h" /* only for _req_mod in tl_release and tl_clear */

#include "drbd_vli.h"

59
static DEFINE_MUTEX(drbd_main_mutex);
P
Philipp Reisner 已提交
60 61 62 63 64 65 66
int drbdd_init(struct drbd_thread *);
int drbd_worker(struct drbd_thread *);
int drbd_asender(struct drbd_thread *);

int drbd_init(void);
static int drbd_open(struct block_device *bdev, fmode_t mode);
static int drbd_release(struct gendisk *gd, fmode_t mode);
67
static int w_md_sync(struct drbd_work *w, int unused);
P
Philipp Reisner 已提交
68
static void md_sync_timer_fn(unsigned long data);
69 70
static int w_bitmap_io(struct drbd_work *w, int unused);
static int w_go_diskless(struct drbd_work *w, int unused);
P
Philipp Reisner 已提交
71 72 73 74 75 76

MODULE_AUTHOR("Philipp Reisner <phil@linbit.com>, "
	      "Lars Ellenberg <lars@linbit.com>");
MODULE_DESCRIPTION("drbd - Distributed Replicated Block Device v" REL_VERSION);
MODULE_VERSION(REL_VERSION);
MODULE_LICENSE("GPL");
77
MODULE_PARM_DESC(minor_count, "Approximate number of drbd devices ("
78
		 __stringify(DRBD_MINOR_COUNT_MIN) "-" __stringify(DRBD_MINOR_COUNT_MAX) ")");
P
Philipp Reisner 已提交
79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106
MODULE_ALIAS_BLOCKDEV_MAJOR(DRBD_MAJOR);

#include <linux/moduleparam.h>
/* allow_open_on_secondary */
MODULE_PARM_DESC(allow_oos, "DONT USE!");
/* thanks to these macros, if compiled into the kernel (not-module),
 * this becomes the boot parameter drbd.minor_count */
module_param(minor_count, uint, 0444);
module_param(disable_sendpage, bool, 0644);
module_param(allow_oos, bool, 0);
module_param(proc_details, int, 0644);

#ifdef CONFIG_DRBD_FAULT_INJECTION
int enable_faults;
int fault_rate;
static int fault_count;
int fault_devs;
/* bitmap of enabled faults */
module_param(enable_faults, int, 0664);
/* fault rate % value - applies to all enabled faults */
module_param(fault_rate, int, 0664);
/* count of faults inserted */
module_param(fault_count, int, 0664);
/* bitmap of devices to insert faults on */
module_param(fault_devs, int, 0644);
#endif

/* module parameter, defined */
107
unsigned int minor_count = DRBD_MINOR_COUNT_DEF;
P
Philipp Reisner 已提交
108 109 110 111 112 113 114 115 116 117 118 119 120
int disable_sendpage;
int allow_oos;
int proc_details;       /* Detail level in proc drbd*/

/* Module parameter for setting the user mode helper program
 * to run. Default is /sbin/drbdadm */
char usermode_helper[80] = "/sbin/drbdadm";

module_param_string(usermode_helper, usermode_helper, sizeof(usermode_helper), 0644);

/* in 2.6.x, our device mapping and config info contains our virtual gendisks
 * as member "struct gendisk *vdisk;"
 */
121
struct idr minors;
122
struct list_head drbd_tconns;  /* list of struct drbd_tconn */
P
Philipp Reisner 已提交
123 124

struct kmem_cache *drbd_request_cache;
A
Andreas Gruenbacher 已提交
125
struct kmem_cache *drbd_ee_cache;	/* peer requests */
P
Philipp Reisner 已提交
126 127 128 129
struct kmem_cache *drbd_bm_ext_cache;	/* bitmap extents */
struct kmem_cache *drbd_al_ext_cache;	/* activity log extents */
mempool_t *drbd_request_mempool;
mempool_t *drbd_ee_mempool;
130
mempool_t *drbd_md_io_page_pool;
131
struct bio_set *drbd_md_io_bio_set;
P
Philipp Reisner 已提交
132 133 134 135 136 137 138 139 140 141 142 143 144 145

/* I do not use a standard mempool, because:
   1) I want to hand out the pre-allocated objects first.
   2) I want to be able to interrupt sleeping allocation with a signal.
   Note: This is a single linked list, the next pointer is the private
	 member of struct page.
 */
struct page *drbd_pp_pool;
spinlock_t   drbd_pp_lock;
int          drbd_pp_vacant;
wait_queue_head_t drbd_pp_wait;

DEFINE_RATELIMIT_STATE(drbd_ratelimit_state, 5 * HZ, 5);

146
static const struct block_device_operations drbd_ops = {
P
Philipp Reisner 已提交
147 148 149 150 151
	.owner =   THIS_MODULE,
	.open =    drbd_open,
	.release = drbd_release,
};

152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170
static void bio_destructor_drbd(struct bio *bio)
{
	bio_free(bio, drbd_md_io_bio_set);
}

struct bio *bio_alloc_drbd(gfp_t gfp_mask)
{
	struct bio *bio;

	if (!drbd_md_io_bio_set)
		return bio_alloc(gfp_mask, 1);

	bio = bio_alloc_bioset(gfp_mask, 1, drbd_md_io_bio_set);
	if (!bio)
		return NULL;
	bio->bi_destructor = bio_destructor_drbd;
	return bio;
}

P
Philipp Reisner 已提交
171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193
#ifdef __CHECKER__
/* When checking with sparse, and this is an inline function, sparse will
   give tons of false positives. When this is a real functions sparse works.
 */
int _get_ldev_if_state(struct drbd_conf *mdev, enum drbd_disk_state mins)
{
	int io_allowed;

	atomic_inc(&mdev->local_cnt);
	io_allowed = (mdev->state.disk >= mins);
	if (!io_allowed) {
		if (atomic_dec_and_test(&mdev->local_cnt))
			wake_up(&mdev->misc_wait);
	}
	return io_allowed;
}

#endif

/**
 * DOC: The transfer log
 *
 * The transfer log is a single linked list of &struct drbd_tl_epoch objects.
194
 * mdev->tconn->newest_tle points to the head, mdev->tconn->oldest_tle points to the tail
P
Philipp Reisner 已提交
195 196 197 198 199
 * of the list. There is always at least one &struct drbd_tl_epoch object.
 *
 * Each &struct drbd_tl_epoch has a circular double linked list of requests
 * attached.
 */
200
static int tl_init(struct drbd_tconn *tconn)
P
Philipp Reisner 已提交
201 202 203 204 205 206 207 208 209 210 211
{
	struct drbd_tl_epoch *b;

	/* during device minor initialization, we may well use GFP_KERNEL */
	b = kmalloc(sizeof(struct drbd_tl_epoch), GFP_KERNEL);
	if (!b)
		return 0;
	INIT_LIST_HEAD(&b->requests);
	INIT_LIST_HEAD(&b->w.list);
	b->next = NULL;
	b->br_number = 4711;
212
	b->n_writes = 0;
P
Philipp Reisner 已提交
213 214
	b->w.cb = NULL; /* if this is != NULL, we need to dec_ap_pending in tl_clear */

215 216 217
	tconn->oldest_tle = b;
	tconn->newest_tle = b;
	INIT_LIST_HEAD(&tconn->out_of_sequence_requests);
218
	INIT_LIST_HEAD(&tconn->barrier_acked_requests);
P
Philipp Reisner 已提交
219 220 221 222

	return 1;
}

223
static void tl_cleanup(struct drbd_tconn *tconn)
P
Philipp Reisner 已提交
224
{
225 226 227 228 229 230 231 232
	if (tconn->oldest_tle != tconn->newest_tle)
		conn_err(tconn, "ASSERT FAILED: oldest_tle == newest_tle\n");
	if (!list_empty(&tconn->out_of_sequence_requests))
		conn_err(tconn, "ASSERT FAILED: list_empty(out_of_sequence_requests)\n");
	kfree(tconn->oldest_tle);
	tconn->oldest_tle = NULL;
	kfree(tconn->unused_spare_tle);
	tconn->unused_spare_tle = NULL;
233 234
}

P
Philipp Reisner 已提交
235 236 237 238 239 240 241
/**
 * _tl_add_barrier() - Adds a barrier to the transfer log
 * @mdev:	DRBD device.
 * @new:	Barrier to be added before the current head of the TL.
 *
 * The caller must hold the req_lock.
 */
242
void _tl_add_barrier(struct drbd_tconn *tconn, struct drbd_tl_epoch *new)
P
Philipp Reisner 已提交
243 244 245 246 247 248 249
{
	struct drbd_tl_epoch *newest_before;

	INIT_LIST_HEAD(&new->requests);
	INIT_LIST_HEAD(&new->w.list);
	new->w.cb = NULL; /* if this is != NULL, we need to dec_ap_pending in tl_clear */
	new->next = NULL;
250
	new->n_writes = 0;
P
Philipp Reisner 已提交
251

252
	newest_before = tconn->newest_tle;
P
Philipp Reisner 已提交
253 254 255
	/* never send a barrier number == 0, because that is special-cased
	 * when using TCQ for our write ordering code */
	new->br_number = (newest_before->br_number+1) ?: 1;
256 257 258
	if (tconn->newest_tle != new) {
		tconn->newest_tle->next = new;
		tconn->newest_tle = new;
P
Philipp Reisner 已提交
259 260 261 262 263 264 265 266 267 268 269 270 271
	}
}

/**
 * tl_release() - Free or recycle the oldest &struct drbd_tl_epoch object of the TL
 * @mdev:	DRBD device.
 * @barrier_nr:	Expected identifier of the DRBD write barrier packet.
 * @set_size:	Expected number of requests before that barrier.
 *
 * In case the passed barrier_nr or set_size does not match the oldest
 * &struct drbd_tl_epoch objects this function will cause a termination
 * of the connection.
 */
272 273
void tl_release(struct drbd_tconn *tconn, unsigned int barrier_nr,
		unsigned int set_size)
P
Philipp Reisner 已提交
274
{
275
	struct drbd_conf *mdev;
P
Philipp Reisner 已提交
276 277 278 279
	struct drbd_tl_epoch *b, *nob; /* next old barrier */
	struct list_head *le, *tle;
	struct drbd_request *r;

280
	spin_lock_irq(&tconn->req_lock);
P
Philipp Reisner 已提交
281

282
	b = tconn->oldest_tle;
P
Philipp Reisner 已提交
283 284 285

	/* first some paranoia code */
	if (b == NULL) {
286 287
		conn_err(tconn, "BAD! BarrierAck #%u received, but no epoch in tl!?\n",
			 barrier_nr);
P
Philipp Reisner 已提交
288 289 290
		goto bail;
	}
	if (b->br_number != barrier_nr) {
291 292
		conn_err(tconn, "BAD! BarrierAck #%u received, expected #%u!\n",
			 barrier_nr, b->br_number);
P
Philipp Reisner 已提交
293 294
		goto bail;
	}
295
	if (b->n_writes != set_size) {
296 297
		conn_err(tconn, "BAD! BarrierAck #%u received with n_writes=%u, expected n_writes=%u!\n",
			 barrier_nr, set_size, b->n_writes);
P
Philipp Reisner 已提交
298 299 300 301 302 303
		goto bail;
	}

	/* Clean up list of requests processed during current epoch */
	list_for_each_safe(le, tle, &b->requests) {
		r = list_entry(le, struct drbd_request, tl_requests);
304
		_req_mod(r, BARRIER_ACKED);
P
Philipp Reisner 已提交
305 306 307 308 309 310 311 312 313
	}
	/* There could be requests on the list waiting for completion
	   of the write to the local disk. To avoid corruptions of
	   slab's data structures we have to remove the lists head.

	   Also there could have been a barrier ack out of sequence, overtaking
	   the write acks - which would be a bug and violating write ordering.
	   To not deadlock in case we lose connection while such requests are
	   still pending, we need some way to find them for the
314
	   _req_mode(CONNECTION_LOST_WHILE_PENDING).
P
Philipp Reisner 已提交
315 316

	   These have been list_move'd to the out_of_sequence_requests list in
317
	   _req_mod(, BARRIER_ACKED) above.
P
Philipp Reisner 已提交
318
	   */
319
	list_splice_init(&b->requests, &tconn->barrier_acked_requests);
320
	mdev = b->w.mdev;
P
Philipp Reisner 已提交
321 322 323

	nob = b->next;
	if (test_and_clear_bit(CREATE_BARRIER, &mdev->flags)) {
324
		_tl_add_barrier(tconn, b);
P
Philipp Reisner 已提交
325
		if (nob)
326
			tconn->oldest_tle = nob;
P
Philipp Reisner 已提交
327
		/* if nob == NULL b was the only barrier, and becomes the new
328
		   barrier. Therefore tconn->oldest_tle points already to b */
P
Philipp Reisner 已提交
329 330
	} else {
		D_ASSERT(nob != NULL);
331
		tconn->oldest_tle = nob;
P
Philipp Reisner 已提交
332 333 334
		kfree(b);
	}

335
	spin_unlock_irq(&tconn->req_lock);
P
Philipp Reisner 已提交
336 337 338 339 340
	dec_ap_pending(mdev);

	return;

bail:
341 342
	spin_unlock_irq(&tconn->req_lock);
	conn_request_state(tconn, NS(conn, C_PROTOCOL_ERROR), CS_HARD);
P
Philipp Reisner 已提交
343 344
}

345

P
Philipp Reisner 已提交
346
/**
347
 * _tl_restart() - Walks the transfer log, and applies an action to all requests
P
Philipp Reisner 已提交
348
 * @mdev:	DRBD device.
349
 * @what:       The action/event to perform with all request objects
P
Philipp Reisner 已提交
350
 *
351 352
 * @what might be one of CONNECTION_LOST_WHILE_PENDING, RESEND, FAIL_FROZEN_DISK_IO,
 * RESTART_FROZEN_DISK_IO.
P
Philipp Reisner 已提交
353
 */
354
void _tl_restart(struct drbd_tconn *tconn, enum drbd_req_event what)
P
Philipp Reisner 已提交
355
{
356
	struct drbd_tl_epoch *b, *tmp, **pn;
357
	struct list_head *le, *tle, carry_reads;
358 359
	struct drbd_request *req;
	int rv, n_writes, n_reads;
P
Philipp Reisner 已提交
360

361 362
	b = tconn->oldest_tle;
	pn = &tconn->oldest_tle;
P
Philipp Reisner 已提交
363
	while (b) {
364 365
		n_writes = 0;
		n_reads = 0;
366
		INIT_LIST_HEAD(&carry_reads);
P
Philipp Reisner 已提交
367
		list_for_each_safe(le, tle, &b->requests) {
368 369 370 371 372
			req = list_entry(le, struct drbd_request, tl_requests);
			rv = _req_mod(req, what);

			n_writes += (rv & MR_WRITE) >> MR_WRITE_SHIFT;
			n_reads  += (rv & MR_READ) >> MR_READ_SHIFT;
P
Philipp Reisner 已提交
373 374 375
		}
		tmp = b->next;

376
		if (n_writes) {
377
			if (what == RESEND) {
378 379 380
				b->n_writes = n_writes;
				if (b->w.cb == NULL) {
					b->w.cb = w_send_barrier;
381 382
					inc_ap_pending(b->w.mdev);
					set_bit(CREATE_BARRIER, &b->w.mdev->flags);
383 384
				}

385
				drbd_queue_work(&tconn->data.work, &b->w);
386 387 388
			}
			pn = &b->next;
		} else {
389 390
			if (n_reads)
				list_add(&carry_reads, &b->requests);
391 392 393 394 395 396 397 398
			/* there could still be requests on that ring list,
			 * in case local io is still pending */
			list_del(&b->requests);

			/* dec_ap_pending corresponding to queue_barrier.
			 * the newest barrier may not have been queued yet,
			 * in which case w.cb is still NULL. */
			if (b->w.cb != NULL)
399
				dec_ap_pending(b->w.mdev);
400

401
			if (b == tconn->newest_tle) {
402
				/* recycle, but reinit! */
403 404
				if (tmp != NULL)
					conn_err(tconn, "ASSERT FAILED tmp == NULL");
405
				INIT_LIST_HEAD(&b->requests);
406
				list_splice(&carry_reads, &b->requests);
407 408 409 410 411 412 413 414 415 416
				INIT_LIST_HEAD(&b->w.list);
				b->w.cb = NULL;
				b->br_number = net_random();
				b->n_writes = 0;

				*pn = b;
				break;
			}
			*pn = tmp;
			kfree(b);
P
Philipp Reisner 已提交
417 418
		}
		b = tmp;
419
		list_splice(&carry_reads, &b->requests);
P
Philipp Reisner 已提交
420
	}
421

422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437
	/* Actions operating on the disk state, also want to work on
	   requests that got barrier acked. */
	switch (what) {
	case FAIL_FROZEN_DISK_IO:
	case RESTART_FROZEN_DISK_IO:
		list_for_each_safe(le, tle, &tconn->barrier_acked_requests) {
			req = list_entry(le, struct drbd_request, tl_requests);
			_req_mod(req, what);
		}
	case CONNECTION_LOST_WHILE_PENDING:
	case RESEND:
		break;
	default:
		conn_err(tconn, "what = %d in _tl_restart()\n", what);
	}
}
P
Philipp Reisner 已提交
438 439 440 441 442 443 444 445 446

/**
 * tl_clear() - Clears all requests and &struct drbd_tl_epoch objects out of the TL
 * @mdev:	DRBD device.
 *
 * This is called after the connection to the peer was lost. The storage covered
 * by the requests on the transfer gets marked as our of sync. Called from the
 * receiver thread and the worker thread.
 */
447
void tl_clear(struct drbd_tconn *tconn)
P
Philipp Reisner 已提交
448
{
449
	struct drbd_conf *mdev;
P
Philipp Reisner 已提交
450 451
	struct list_head *le, *tle;
	struct drbd_request *r;
P
Philipp Reisner 已提交
452
	int vnr;
P
Philipp Reisner 已提交
453

454
	spin_lock_irq(&tconn->req_lock);
P
Philipp Reisner 已提交
455

456
	_tl_restart(tconn, CONNECTION_LOST_WHILE_PENDING);
P
Philipp Reisner 已提交
457 458

	/* we expect this list to be empty. */
459 460
	if (!list_empty(&tconn->out_of_sequence_requests))
		conn_err(tconn, "ASSERT FAILED list_empty(&out_of_sequence_requests)\n");
P
Philipp Reisner 已提交
461 462

	/* but just in case, clean it up anyways! */
463
	list_for_each_safe(le, tle, &tconn->out_of_sequence_requests) {
P
Philipp Reisner 已提交
464 465 466
		r = list_entry(le, struct drbd_request, tl_requests);
		/* It would be nice to complete outside of spinlock.
		 * But this is easier for now. */
467
		_req_mod(r, CONNECTION_LOST_WHILE_PENDING);
P
Philipp Reisner 已提交
468 469 470
	}

	/* ensure bit indicating barrier is required is clear */
471
	rcu_read_lock();
P
Philipp Reisner 已提交
472
	idr_for_each_entry(&tconn->volumes, mdev, vnr)
473
		clear_bit(CREATE_BARRIER, &mdev->flags);
474
	rcu_read_unlock();
P
Philipp Reisner 已提交
475

476
	spin_unlock_irq(&tconn->req_lock);
P
Philipp Reisner 已提交
477 478
}

479
void tl_restart(struct drbd_tconn *tconn, enum drbd_req_event what)
480
{
481 482 483
	spin_lock_irq(&tconn->req_lock);
	_tl_restart(tconn, what);
	spin_unlock_irq(&tconn->req_lock);
P
Philipp Reisner 已提交
484 485
}

486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521
/**
 * tl_apply() - Applies an event to all requests for a certain mdev in the TL
 * @mdev:	DRBD device.
 * @what:       The action/event to perform with all request objects
 *
 * @what might ony be ABORT_DISK_IO.
 */
void tl_apply(struct drbd_conf *mdev, enum drbd_req_event what)
{
	struct drbd_tconn *tconn = mdev->tconn;
	struct drbd_tl_epoch *b;
	struct list_head *le, *tle;
	struct drbd_request *req;

	D_ASSERT(what == ABORT_DISK_IO);

	spin_lock_irq(&tconn->req_lock);
	b = tconn->oldest_tle;
	while (b) {
		list_for_each_safe(le, tle, &b->requests) {
			req = list_entry(le, struct drbd_request, tl_requests);
			if (req->w.mdev == mdev)
				_req_mod(req, what);
		}
		b = b->next;
	}

	list_for_each_safe(le, tle, &tconn->barrier_acked_requests) {
		req = list_entry(le, struct drbd_request, tl_requests);
		if (req->w.mdev == mdev)
			_req_mod(req, what);
	}

	spin_unlock_irq(&tconn->req_lock);
}

P
Philipp Reisner 已提交
522 523 524
static int drbd_thread_setup(void *arg)
{
	struct drbd_thread *thi = (struct drbd_thread *) arg;
525
	struct drbd_tconn *tconn = thi->tconn;
P
Philipp Reisner 已提交
526 527 528
	unsigned long flags;
	int retval;

529
	snprintf(current->comm, sizeof(current->comm), "drbd_%c_%s",
530
		 thi->name[0], thi->tconn->name);
531

P
Philipp Reisner 已提交
532 533 534 535 536
restart:
	retval = thi->function(thi);

	spin_lock_irqsave(&thi->t_lock, flags);

537
	/* if the receiver has been "EXITING", the last thing it did
P
Philipp Reisner 已提交
538 539 540
	 * was set the conn state to "StandAlone",
	 * if now a re-connect request comes in, conn state goes C_UNCONNECTED,
	 * and receiver thread will be "started".
541
	 * drbd_thread_start needs to set "RESTARTING" in that case.
P
Philipp Reisner 已提交
542
	 * t_state check and assignment needs to be within the same spinlock,
543 544
	 * so either thread_start sees EXITING, and can remap to RESTARTING,
	 * or thread_start see NONE, and can proceed as normal.
P
Philipp Reisner 已提交
545 546
	 */

547
	if (thi->t_state == RESTARTING) {
548
		conn_info(tconn, "Restarting %s thread\n", thi->name);
549
		thi->t_state = RUNNING;
P
Philipp Reisner 已提交
550 551 552 553 554
		spin_unlock_irqrestore(&thi->t_lock, flags);
		goto restart;
	}

	thi->task = NULL;
555
	thi->t_state = NONE;
P
Philipp Reisner 已提交
556
	smp_mb();
L
Lars Ellenberg 已提交
557
	complete_all(&thi->stop);
P
Philipp Reisner 已提交
558 559
	spin_unlock_irqrestore(&thi->t_lock, flags);

560
	conn_info(tconn, "Terminating %s\n", current->comm);
P
Philipp Reisner 已提交
561 562

	/* Release mod reference taken when thread was started */
563 564

	kref_put(&tconn->kref, &conn_destroy);
P
Philipp Reisner 已提交
565 566 567 568
	module_put(THIS_MODULE);
	return retval;
}

569
static void drbd_thread_init(struct drbd_tconn *tconn, struct drbd_thread *thi,
570
			     int (*func) (struct drbd_thread *), char *name)
P
Philipp Reisner 已提交
571 572 573
{
	spin_lock_init(&thi->t_lock);
	thi->task    = NULL;
574
	thi->t_state = NONE;
P
Philipp Reisner 已提交
575
	thi->function = func;
576
	thi->tconn = tconn;
577
	strncpy(thi->name, name, ARRAY_SIZE(thi->name));
P
Philipp Reisner 已提交
578 579 580 581
}

int drbd_thread_start(struct drbd_thread *thi)
{
582
	struct drbd_tconn *tconn = thi->tconn;
P
Philipp Reisner 已提交
583 584 585 586 587 588 589 590
	struct task_struct *nt;
	unsigned long flags;

	/* is used from state engine doing drbd_thread_stop_nowait,
	 * while holding the req lock irqsave */
	spin_lock_irqsave(&thi->t_lock, flags);

	switch (thi->t_state) {
591
	case NONE:
592
		conn_info(tconn, "Starting %s thread (from %s [%d])\n",
593
			 thi->name, current->comm, current->pid);
P
Philipp Reisner 已提交
594 595 596

		/* Get ref on module for thread - this is released when thread exits */
		if (!try_module_get(THIS_MODULE)) {
597
			conn_err(tconn, "Failed to get module reference in drbd_thread_start\n");
P
Philipp Reisner 已提交
598
			spin_unlock_irqrestore(&thi->t_lock, flags);
599
			return false;
P
Philipp Reisner 已提交
600 601
		}

602 603
		kref_get(&thi->tconn->kref);

P
Philipp Reisner 已提交
604 605
		init_completion(&thi->stop);
		thi->reset_cpu_mask = 1;
606
		thi->t_state = RUNNING;
P
Philipp Reisner 已提交
607 608 609 610
		spin_unlock_irqrestore(&thi->t_lock, flags);
		flush_signals(current); /* otherw. may get -ERESTARTNOINTR */

		nt = kthread_create(drbd_thread_setup, (void *) thi,
611
				    "drbd_%c_%s", thi->name[0], thi->tconn->name);
P
Philipp Reisner 已提交
612 613

		if (IS_ERR(nt)) {
614
			conn_err(tconn, "Couldn't start thread\n");
P
Philipp Reisner 已提交
615

616
			kref_put(&tconn->kref, &conn_destroy);
P
Philipp Reisner 已提交
617
			module_put(THIS_MODULE);
618
			return false;
P
Philipp Reisner 已提交
619 620 621
		}
		spin_lock_irqsave(&thi->t_lock, flags);
		thi->task = nt;
622
		thi->t_state = RUNNING;
P
Philipp Reisner 已提交
623 624 625
		spin_unlock_irqrestore(&thi->t_lock, flags);
		wake_up_process(nt);
		break;
626 627
	case EXITING:
		thi->t_state = RESTARTING;
628
		conn_info(tconn, "Restarting %s thread (from %s [%d])\n",
629
				thi->name, current->comm, current->pid);
P
Philipp Reisner 已提交
630
		/* fall through */
631 632
	case RUNNING:
	case RESTARTING:
P
Philipp Reisner 已提交
633 634 635 636 637
	default:
		spin_unlock_irqrestore(&thi->t_lock, flags);
		break;
	}

638
	return true;
P
Philipp Reisner 已提交
639 640 641 642 643 644 645
}


void _drbd_thread_stop(struct drbd_thread *thi, int restart, int wait)
{
	unsigned long flags;

646
	enum drbd_thread_state ns = restart ? RESTARTING : EXITING;
P
Philipp Reisner 已提交
647 648 649 650

	/* may be called from state engine, holding the req lock irqsave */
	spin_lock_irqsave(&thi->t_lock, flags);

651
	if (thi->t_state == NONE) {
P
Philipp Reisner 已提交
652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676
		spin_unlock_irqrestore(&thi->t_lock, flags);
		if (restart)
			drbd_thread_start(thi);
		return;
	}

	if (thi->t_state != ns) {
		if (thi->task == NULL) {
			spin_unlock_irqrestore(&thi->t_lock, flags);
			return;
		}

		thi->t_state = ns;
		smp_mb();
		init_completion(&thi->stop);
		if (thi->task != current)
			force_sig(DRBD_SIGKILL, thi->task);
	}

	spin_unlock_irqrestore(&thi->t_lock, flags);

	if (wait)
		wait_for_completion(&thi->stop);
}

677
static struct drbd_thread *drbd_task_to_thread(struct drbd_tconn *tconn, struct task_struct *task)
678 679 680 681 682 683 684 685 686
{
	struct drbd_thread *thi =
		task == tconn->receiver.task ? &tconn->receiver :
		task == tconn->asender.task  ? &tconn->asender :
		task == tconn->worker.task   ? &tconn->worker : NULL;

	return thi;
}

687
char *drbd_task_to_thread_name(struct drbd_tconn *tconn, struct task_struct *task)
688
{
689
	struct drbd_thread *thi = drbd_task_to_thread(tconn, task);
690 691 692
	return thi ? thi->name : task->comm;
}

693
int conn_lowest_minor(struct drbd_tconn *tconn)
694
{
P
Philipp Reisner 已提交
695
	struct drbd_conf *mdev;
696
	int vnr = 0, m;
697

698
	rcu_read_lock();
P
Philipp Reisner 已提交
699
	mdev = idr_get_next(&tconn->volumes, &vnr);
700 701 702 703
	m = mdev ? mdev_to_minor(mdev) : -1;
	rcu_read_unlock();

	return m;
704
}
705 706

#ifdef CONFIG_SMP
P
Philipp Reisner 已提交
707 708 709 710 711 712 713
/**
 * drbd_calc_cpu_mask() - Generate CPU masks, spread over all CPUs
 * @mdev:	DRBD device.
 *
 * Forces all threads of a device onto the same CPU. This is beneficial for
 * DRBD's performance. May be overwritten by user's configuration.
 */
714
void drbd_calc_cpu_mask(struct drbd_tconn *tconn)
P
Philipp Reisner 已提交
715 716 717 718
{
	int ord, cpu;

	/* user override. */
719
	if (cpumask_weight(tconn->cpu_mask))
P
Philipp Reisner 已提交
720 721
		return;

722
	ord = conn_lowest_minor(tconn) % cpumask_weight(cpu_online_mask);
P
Philipp Reisner 已提交
723 724
	for_each_online_cpu(cpu) {
		if (ord-- == 0) {
725
			cpumask_set_cpu(cpu, tconn->cpu_mask);
P
Philipp Reisner 已提交
726 727 728 729
			return;
		}
	}
	/* should not be reached */
730
	cpumask_setall(tconn->cpu_mask);
P
Philipp Reisner 已提交
731 732 733 734 735
}

/**
 * drbd_thread_current_set_cpu() - modifies the cpu mask of the _current_ thread
 * @mdev:	DRBD device.
736
 * @thi:	drbd_thread object
P
Philipp Reisner 已提交
737 738 739 740
 *
 * call in the "main loop" of _all_ threads, no need for any mutex, current won't die
 * prematurely.
 */
741
void drbd_thread_current_set_cpu(struct drbd_thread *thi)
P
Philipp Reisner 已提交
742 743
{
	struct task_struct *p = current;
744

P
Philipp Reisner 已提交
745 746 747
	if (!thi->reset_cpu_mask)
		return;
	thi->reset_cpu_mask = 0;
748
	set_cpus_allowed_ptr(p, thi->tconn->cpu_mask);
P
Philipp Reisner 已提交
749 750 751
}
#endif

752 753 754 755 756 757 758 759 760
/**
 * drbd_header_size  -  size of a packet header
 *
 * The header size is a multiple of 8, so any payload following the header is
 * word aligned on 64-bit architectures.  (The bitmap send and receive code
 * relies on this.)
 */
unsigned int drbd_header_size(struct drbd_tconn *tconn)
{
761 762 763 764 765 766 767 768 769
	if (tconn->agreed_pro_version >= 100) {
		BUILD_BUG_ON(!IS_ALIGNED(sizeof(struct p_header100), 8));
		return sizeof(struct p_header100);
	} else {
		BUILD_BUG_ON(sizeof(struct p_header80) !=
			     sizeof(struct p_header95));
		BUILD_BUG_ON(!IS_ALIGNED(sizeof(struct p_header80), 8));
		return sizeof(struct p_header80);
	}
770 771
}

772
static unsigned int prepare_header80(struct p_header80 *h, enum drbd_packet cmd, int size)
P
Philipp Reisner 已提交
773 774 775 776
{
	h->magic   = cpu_to_be32(DRBD_MAGIC);
	h->command = cpu_to_be16(cmd);
	h->length  = cpu_to_be16(size);
777
	return sizeof(struct p_header80);
P
Philipp Reisner 已提交
778 779
}

780
static unsigned int prepare_header95(struct p_header95 *h, enum drbd_packet cmd, int size)
P
Philipp Reisner 已提交
781 782 783
{
	h->magic   = cpu_to_be16(DRBD_MAGIC_BIG);
	h->command = cpu_to_be16(cmd);
784
	h->length = cpu_to_be32(size);
785
	return sizeof(struct p_header95);
P
Philipp Reisner 已提交
786 787
}

788 789 790 791 792 793 794 795 796 797 798 799 800
static unsigned int prepare_header100(struct p_header100 *h, enum drbd_packet cmd,
				      int size, int vnr)
{
	h->magic = cpu_to_be32(DRBD_MAGIC_100);
	h->volume = cpu_to_be16(vnr);
	h->command = cpu_to_be16(cmd);
	h->length = cpu_to_be32(size);
	h->pad = 0;
	return sizeof(struct p_header100);
}

static unsigned int prepare_header(struct drbd_tconn *tconn, int vnr,
				   void *buffer, enum drbd_packet cmd, int size)
801
{
802 803 804 805
	if (tconn->agreed_pro_version >= 100)
		return prepare_header100(buffer, cmd, size, vnr);
	else if (tconn->agreed_pro_version >= 95 &&
		 size > DRBD_MAX_SIZE_H80_PACKET)
806
		return prepare_header95(buffer, cmd, size);
807
	else
808
		return prepare_header80(buffer, cmd, size);
809 810
}

811 812 813 814 815 816 817 818
static void *__conn_prepare_command(struct drbd_tconn *tconn,
				    struct drbd_socket *sock)
{
	if (!sock->socket)
		return NULL;
	return sock->sbuf + drbd_header_size(tconn);
}

819 820
void *conn_prepare_command(struct drbd_tconn *tconn, struct drbd_socket *sock)
{
821 822
	void *p;

823
	mutex_lock(&sock->mutex);
824 825
	p = __conn_prepare_command(tconn, sock);
	if (!p)
826
		mutex_unlock(&sock->mutex);
827 828

	return p;
829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852
}

void *drbd_prepare_command(struct drbd_conf *mdev, struct drbd_socket *sock)
{
	return conn_prepare_command(mdev->tconn, sock);
}

static int __send_command(struct drbd_tconn *tconn, int vnr,
			  struct drbd_socket *sock, enum drbd_packet cmd,
			  unsigned int header_size, void *data,
			  unsigned int size)
{
	int msg_flags;
	int err;

	/*
	 * Called with @data == NULL and the size of the data blocks in @size
	 * for commands that send data blocks.  For those commands, omit the
	 * MSG_MORE flag: this will increase the likelihood that data blocks
	 * which are page aligned on the sender will end up page aligned on the
	 * receiver.
	 */
	msg_flags = data ? MSG_MORE : 0;

853 854
	header_size += prepare_header(tconn, vnr, sock->sbuf, cmd,
				      header_size + size);
855 856 857 858 859 860 861
	err = drbd_send_all(tconn, sock->socket, sock->sbuf, header_size,
			    msg_flags);
	if (data && !err)
		err = drbd_send_all(tconn, sock->socket, data, size, 0);
	return err;
}

862 863 864 865 866 867 868
static int __conn_send_command(struct drbd_tconn *tconn, struct drbd_socket *sock,
			       enum drbd_packet cmd, unsigned int header_size,
			       void *data, unsigned int size)
{
	return __send_command(tconn, 0, sock, cmd, header_size, data, size);
}

869 870 871 872 873 874
int conn_send_command(struct drbd_tconn *tconn, struct drbd_socket *sock,
		      enum drbd_packet cmd, unsigned int header_size,
		      void *data, unsigned int size)
{
	int err;

875
	err = __conn_send_command(tconn, sock, cmd, header_size, data, size);
876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891
	mutex_unlock(&sock->mutex);
	return err;
}

int drbd_send_command(struct drbd_conf *mdev, struct drbd_socket *sock,
		      enum drbd_packet cmd, unsigned int header_size,
		      void *data, unsigned int size)
{
	int err;

	err = __send_command(mdev->tconn, mdev->vnr, sock, cmd, header_size,
			     data, size);
	mutex_unlock(&sock->mutex);
	return err;
}

892 893
int drbd_send_ping(struct drbd_tconn *tconn)
{
894 895 896 897 898
	struct drbd_socket *sock;

	sock = &tconn->meta;
	if (!conn_prepare_command(tconn, sock))
		return -EIO;
899
	return conn_send_command(tconn, sock, P_PING, 0, NULL, 0);
900 901 902 903
}

int drbd_send_ping_ack(struct drbd_tconn *tconn)
{
904 905 906 907 908
	struct drbd_socket *sock;

	sock = &tconn->meta;
	if (!conn_prepare_command(tconn, sock))
		return -EIO;
909
	return conn_send_command(tconn, sock, P_PING_ACK, 0, NULL, 0);
910 911
}

912
int drbd_send_sync_param(struct drbd_conf *mdev)
P
Philipp Reisner 已提交
913
{
914
	struct drbd_socket *sock;
915 916
	struct p_rs_param_95 *p;
	int size;
917
	const int apv = mdev->tconn->agreed_pro_version;
918
	enum drbd_packet cmd;
919
	struct net_conf *nc;
P
Philipp Reisner 已提交
920
	struct disk_conf *dc;
921 922 923 924 925

	sock = &mdev->tconn->data;
	p = drbd_prepare_command(mdev, sock);
	if (!p)
		return -EIO;
P
Philipp Reisner 已提交
926

927 928 929
	rcu_read_lock();
	nc = rcu_dereference(mdev->tconn->net_conf);

P
Philipp Reisner 已提交
930 931
	size = apv <= 87 ? sizeof(struct p_rs_param)
		: apv == 88 ? sizeof(struct p_rs_param)
932
			+ strlen(nc->verify_alg) + 1
933 934
		: apv <= 94 ? sizeof(struct p_rs_param_89)
		: /* apv >= 95 */ sizeof(struct p_rs_param_95);
P
Philipp Reisner 已提交
935

936
	cmd = apv >= 89 ? P_SYNC_PARAM89 : P_SYNC_PARAM;
P
Philipp Reisner 已提交
937

938 939
	/* initialize verify_alg and csums_alg */
	memset(p->verify_alg, 0, 2 * SHARED_SECRET_MAX);
P
Philipp Reisner 已提交
940

941
	if (get_ldev(mdev)) {
P
Philipp Reisner 已提交
942
		dc = rcu_dereference(mdev->ldev->disk_conf);
943
		p->resync_rate = cpu_to_be32(dc->resync_rate);
P
Philipp Reisner 已提交
944 945 946 947
		p->c_plan_ahead = cpu_to_be32(dc->c_plan_ahead);
		p->c_delay_target = cpu_to_be32(dc->c_delay_target);
		p->c_fill_target = cpu_to_be32(dc->c_fill_target);
		p->c_max_rate = cpu_to_be32(dc->c_max_rate);
948 949
		put_ldev(mdev);
	} else {
950
		p->resync_rate = cpu_to_be32(DRBD_RESYNC_RATE_DEF);
951 952 953 954 955
		p->c_plan_ahead = cpu_to_be32(DRBD_C_PLAN_AHEAD_DEF);
		p->c_delay_target = cpu_to_be32(DRBD_C_DELAY_TARGET_DEF);
		p->c_fill_target = cpu_to_be32(DRBD_C_FILL_TARGET_DEF);
		p->c_max_rate = cpu_to_be32(DRBD_C_MAX_RATE_DEF);
	}
P
Philipp Reisner 已提交
956

957
	if (apv >= 88)
958
		strcpy(p->verify_alg, nc->verify_alg);
959
	if (apv >= 89)
960 961
		strcpy(p->csums_alg, nc->csums_alg);
	rcu_read_unlock();
P
Philipp Reisner 已提交
962

963
	return drbd_send_command(mdev, sock, cmd, size, NULL, 0);
P
Philipp Reisner 已提交
964 965
}

966
int __drbd_send_protocol(struct drbd_tconn *tconn, enum drbd_packet cmd)
P
Philipp Reisner 已提交
967
{
968
	struct drbd_socket *sock;
P
Philipp Reisner 已提交
969
	struct p_protocol *p;
970
	struct net_conf *nc;
971
	int size, cf;
P
Philipp Reisner 已提交
972

973
	sock = &tconn->data;
974
	p = __conn_prepare_command(tconn, sock);
975 976 977
	if (!p)
		return -EIO;

978 979 980
	rcu_read_lock();
	nc = rcu_dereference(tconn->net_conf);

981
	if (nc->tentative && tconn->agreed_pro_version < 92) {
982 983 984 985 986 987
		rcu_read_unlock();
		mutex_unlock(&sock->mutex);
		conn_err(tconn, "--dry-run is not supported by peer");
		return -EOPNOTSUPP;
	}

988
	size = sizeof(*p);
989
	if (tconn->agreed_pro_version >= 87)
990
		size += strlen(nc->integrity_alg) + 1;
P
Philipp Reisner 已提交
991

992 993 994 995 996
	p->protocol      = cpu_to_be32(nc->wire_protocol);
	p->after_sb_0p   = cpu_to_be32(nc->after_sb_0p);
	p->after_sb_1p   = cpu_to_be32(nc->after_sb_1p);
	p->after_sb_2p   = cpu_to_be32(nc->after_sb_2p);
	p->two_primaries = cpu_to_be32(nc->two_primaries);
997
	cf = 0;
998 999
	if (nc->discard_my_data)
		cf |= CF_DISCARD_MY_DATA;
1000
	if (nc->tentative)
1001
		cf |= CF_DRY_RUN;
1002 1003
	p->conn_flags    = cpu_to_be32(cf);

1004
	if (tconn->agreed_pro_version >= 87)
1005 1006 1007
		strcpy(p->integrity_alg, nc->integrity_alg);
	rcu_read_unlock();

1008
	return __conn_send_command(tconn, sock, cmd, size, NULL, 0);
1009 1010 1011 1012 1013 1014 1015
}

int drbd_send_protocol(struct drbd_tconn *tconn)
{
	int err;

	mutex_lock(&tconn->data.mutex);
1016
	err = __drbd_send_protocol(tconn, P_PROTOCOL);
1017 1018 1019
	mutex_unlock(&tconn->data.mutex);

	return err;
P
Philipp Reisner 已提交
1020 1021 1022 1023
}

int _drbd_send_uuids(struct drbd_conf *mdev, u64 uuid_flags)
{
1024 1025
	struct drbd_socket *sock;
	struct p_uuids *p;
P
Philipp Reisner 已提交
1026 1027 1028
	int i;

	if (!get_ldev_if_state(mdev, D_NEGOTIATING))
1029
		return 0;
P
Philipp Reisner 已提交
1030

1031 1032 1033 1034 1035 1036
	sock = &mdev->tconn->data;
	p = drbd_prepare_command(mdev, sock);
	if (!p) {
		put_ldev(mdev);
		return -EIO;
	}
P
Philipp Reisner 已提交
1037
	for (i = UI_CURRENT; i < UI_SIZE; i++)
1038
		p->uuid[i] = mdev->ldev ? cpu_to_be64(mdev->ldev->md.uuid[i]) : 0;
P
Philipp Reisner 已提交
1039 1040

	mdev->comm_bm_set = drbd_bm_total_weight(mdev);
1041
	p->uuid[UI_SIZE] = cpu_to_be64(mdev->comm_bm_set);
1042
	rcu_read_lock();
1043
	uuid_flags |= rcu_dereference(mdev->tconn->net_conf)->discard_my_data ? 1 : 0;
1044
	rcu_read_unlock();
P
Philipp Reisner 已提交
1045 1046
	uuid_flags |= test_bit(CRASHED_PRIMARY, &mdev->flags) ? 2 : 0;
	uuid_flags |= mdev->new_state_tmp.disk == D_INCONSISTENT ? 4 : 0;
1047
	p->uuid[UI_FLAGS] = cpu_to_be64(uuid_flags);
P
Philipp Reisner 已提交
1048 1049

	put_ldev(mdev);
1050
	return drbd_send_command(mdev, sock, P_UUIDS, sizeof(*p), NULL, 0);
P
Philipp Reisner 已提交
1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062
}

int drbd_send_uuids(struct drbd_conf *mdev)
{
	return _drbd_send_uuids(mdev, 0);
}

int drbd_send_uuids_skip_initial_sync(struct drbd_conf *mdev)
{
	return _drbd_send_uuids(mdev, 8);
}

1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080
void drbd_print_uuids(struct drbd_conf *mdev, const char *text)
{
	if (get_ldev_if_state(mdev, D_NEGOTIATING)) {
		u64 *uuid = mdev->ldev->md.uuid;
		dev_info(DEV, "%s %016llX:%016llX:%016llX:%016llX\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]);
		put_ldev(mdev);
	} else {
		dev_info(DEV, "%s effective data uuid: %016llX\n",
				text,
				(unsigned long long)mdev->ed_uuid);
	}
}

1081
void drbd_gen_and_send_sync_uuid(struct drbd_conf *mdev)
P
Philipp Reisner 已提交
1082
{
1083 1084
	struct drbd_socket *sock;
	struct p_rs_uuid *p;
1085 1086 1087
	u64 uuid;

	D_ASSERT(mdev->state.disk == D_UP_TO_DATE);
P
Philipp Reisner 已提交
1088

1089
	uuid = mdev->ldev->md.uuid[UI_BITMAP] + UUID_NEW_BM_OFFSET;
1090
	drbd_uuid_set(mdev, UI_BITMAP, uuid);
1091
	drbd_print_uuids(mdev, "updated sync UUID");
1092
	drbd_md_sync(mdev);
P
Philipp Reisner 已提交
1093

1094 1095 1096 1097 1098 1099
	sock = &mdev->tconn->data;
	p = drbd_prepare_command(mdev, sock);
	if (p) {
		p->uuid = cpu_to_be64(uuid);
		drbd_send_command(mdev, sock, P_SYNC_UUID, sizeof(*p), NULL, 0);
	}
P
Philipp Reisner 已提交
1100 1101
}

1102
int drbd_send_sizes(struct drbd_conf *mdev, int trigger_reply, enum dds_flags flags)
P
Philipp Reisner 已提交
1103
{
1104 1105
	struct drbd_socket *sock;
	struct p_sizes *p;
P
Philipp Reisner 已提交
1106
	sector_t d_size, u_size;
1107
	int q_order_type, max_bio_size;
P
Philipp Reisner 已提交
1108 1109 1110 1111

	if (get_ldev_if_state(mdev, D_NEGOTIATING)) {
		D_ASSERT(mdev->ldev->backing_bdev);
		d_size = drbd_get_max_capacity(mdev->ldev);
P
Philipp Reisner 已提交
1112 1113 1114
		rcu_read_lock();
		u_size = rcu_dereference(mdev->ldev->disk_conf)->disk_size;
		rcu_read_unlock();
P
Philipp Reisner 已提交
1115
		q_order_type = drbd_queue_order_type(mdev);
1116 1117
		max_bio_size = queue_max_hw_sectors(mdev->ldev->backing_bdev->bd_disk->queue) << 9;
		max_bio_size = min_t(int, max_bio_size, DRBD_MAX_BIO_SIZE);
P
Philipp Reisner 已提交
1118 1119 1120 1121 1122
		put_ldev(mdev);
	} else {
		d_size = 0;
		u_size = 0;
		q_order_type = QUEUE_ORDERED_NONE;
1123
		max_bio_size = DRBD_MAX_BIO_SIZE; /* ... multiple BIOs per peer_request */
P
Philipp Reisner 已提交
1124 1125
	}

1126 1127 1128 1129
	sock = &mdev->tconn->data;
	p = drbd_prepare_command(mdev, sock);
	if (!p)
		return -EIO;
1130 1131 1132 1133 1134 1135

	if (mdev->tconn->agreed_pro_version <= 94)
		max_bio_size = min_t(int, max_bio_size, DRBD_MAX_SIZE_H80_PACKET);
	else if (mdev->tconn->agreed_pro_version < 100)
		max_bio_size = min_t(int, max_bio_size, DRBD_MAX_BIO_SIZE_P95);

1136 1137 1138 1139 1140 1141 1142
	p->d_size = cpu_to_be64(d_size);
	p->u_size = cpu_to_be64(u_size);
	p->c_size = cpu_to_be64(trigger_reply ? 0 : drbd_get_capacity(mdev->this_bdev));
	p->max_bio_size = cpu_to_be32(max_bio_size);
	p->queue_order_type = cpu_to_be16(q_order_type);
	p->dds_flags = cpu_to_be16(flags);
	return drbd_send_command(mdev, sock, P_SIZES, sizeof(*p), NULL, 0);
P
Philipp Reisner 已提交
1143 1144 1145 1146 1147 1148 1149 1150
}

/**
 * drbd_send_state() - Sends the drbd state to the peer
 * @mdev:	DRBD device.
 */
int drbd_send_state(struct drbd_conf *mdev)
{
1151
	struct drbd_socket *sock;
1152
	struct p_state *p;
P
Philipp Reisner 已提交
1153

1154
	sock = &mdev->tconn->data;
1155 1156 1157 1158 1159 1160
	p = drbd_prepare_command(mdev, sock);
	if (!p)
		return -EIO;
	p->state = cpu_to_be32(mdev->state.i); /* Within the send mutex */
	return drbd_send_command(mdev, sock, P_STATE, sizeof(*p), NULL, 0);
}
P
Philipp Reisner 已提交
1161

1162 1163 1164 1165
int drbd_send_state_req(struct drbd_conf *mdev, union drbd_state mask, union drbd_state val)
{
	struct drbd_socket *sock;
	struct p_req_state *p;
P
Philipp Reisner 已提交
1166

1167 1168 1169 1170 1171 1172 1173
	sock = &mdev->tconn->data;
	p = drbd_prepare_command(mdev, sock);
	if (!p)
		return -EIO;
	p->mask = cpu_to_be32(mask.i);
	p->val = cpu_to_be32(val.i);
	return drbd_send_command(mdev, sock, P_STATE_CHG_REQ, sizeof(*p), NULL, 0);
P
Philipp Reisner 已提交
1174 1175 1176

}

1177
int conn_send_state_req(struct drbd_tconn *tconn, union drbd_state mask, union drbd_state val)
P
Philipp Reisner 已提交
1178
{
1179 1180 1181
	enum drbd_packet cmd;
	struct drbd_socket *sock;
	struct p_req_state *p;
P
Philipp Reisner 已提交
1182

1183 1184 1185 1186 1187 1188 1189 1190
	cmd = tconn->agreed_pro_version < 100 ? P_STATE_CHG_REQ : P_CONN_ST_CHG_REQ;
	sock = &tconn->data;
	p = conn_prepare_command(tconn, sock);
	if (!p)
		return -EIO;
	p->mask = cpu_to_be32(mask.i);
	p->val = cpu_to_be32(val.i);
	return conn_send_command(tconn, sock, cmd, sizeof(*p), NULL, 0);
P
Philipp Reisner 已提交
1191 1192
}

1193
void drbd_send_sr_reply(struct drbd_conf *mdev, enum drbd_state_rv retcode)
P
Philipp Reisner 已提交
1194
{
1195 1196
	struct drbd_socket *sock;
	struct p_req_state_reply *p;
P
Philipp Reisner 已提交
1197

1198 1199 1200 1201 1202 1203
	sock = &mdev->tconn->meta;
	p = drbd_prepare_command(mdev, sock);
	if (p) {
		p->retcode = cpu_to_be32(retcode);
		drbd_send_command(mdev, sock, P_STATE_CHG_REPLY, sizeof(*p), NULL, 0);
	}
P
Philipp Reisner 已提交
1204 1205
}

1206
void conn_send_sr_reply(struct drbd_tconn *tconn, enum drbd_state_rv retcode)
1207
{
1208 1209
	struct drbd_socket *sock;
	struct p_req_state_reply *p;
1210 1211
	enum drbd_packet cmd = tconn->agreed_pro_version < 100 ? P_STATE_CHG_REPLY : P_CONN_ST_CHG_REPLY;

1212 1213 1214 1215 1216 1217
	sock = &tconn->meta;
	p = conn_prepare_command(tconn, sock);
	if (p) {
		p->retcode = cpu_to_be32(retcode);
		conn_send_command(tconn, sock, cmd, sizeof(*p), NULL, 0);
	}
1218 1219
}

1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236
static void dcbp_set_code(struct p_compressed_bm *p, enum drbd_bitmap_code code)
{
	BUG_ON(code & ~0xf);
	p->encoding = (p->encoding & ~0xf) | code;
}

static void dcbp_set_start(struct p_compressed_bm *p, int set)
{
	p->encoding = (p->encoding & ~0x80) | (set ? 0x80 : 0);
}

static void dcbp_set_pad_bits(struct p_compressed_bm *p, int n)
{
	BUG_ON(n & ~0x7);
	p->encoding = (p->encoding & (~0x7 << 4)) | (n << 4);
}

P
Philipp Reisner 已提交
1237
int fill_bitmap_rle_bits(struct drbd_conf *mdev,
1238 1239 1240
			 struct p_compressed_bm *p,
			 unsigned int size,
			 struct bm_xfer_ctx *c)
P
Philipp Reisner 已提交
1241 1242 1243 1244 1245 1246 1247
{
	struct bitstream bs;
	unsigned long plain_bits;
	unsigned long tmp;
	unsigned long rl;
	unsigned len;
	unsigned toggle;
1248
	int bits, use_rle;
P
Philipp Reisner 已提交
1249 1250

	/* may we use this feature? */
1251 1252 1253 1254 1255
	rcu_read_lock();
	use_rle = rcu_dereference(mdev->tconn->net_conf)->use_rle;
	rcu_read_unlock();
	if (!use_rle || mdev->tconn->agreed_pro_version < 90)
		return 0;
P
Philipp Reisner 已提交
1256 1257 1258 1259 1260

	if (c->bit_offset >= c->bm_bits)
		return 0; /* nothing to do. */

	/* use at most thus many bytes */
1261 1262
	bitstream_init(&bs, p->code, size, 0);
	memset(p->code, 0, size);
P
Philipp Reisner 已提交
1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283
	/* plain bits covered in this code string */
	plain_bits = 0;

	/* p->encoding & 0x80 stores whether the first run length is set.
	 * bit offset is implicit.
	 * start with toggle == 2 to be able to tell the first iteration */
	toggle = 2;

	/* see how much plain bits we can stuff into one packet
	 * using RLE and VLI. */
	do {
		tmp = (toggle == 0) ? _drbd_bm_find_next_zero(mdev, c->bit_offset)
				    : _drbd_bm_find_next(mdev, c->bit_offset);
		if (tmp == -1UL)
			tmp = c->bm_bits;
		rl = tmp - c->bit_offset;

		if (toggle == 2) { /* first iteration */
			if (rl == 0) {
				/* the first checked bit was set,
				 * store start value, */
1284
				dcbp_set_start(p, 1);
P
Philipp Reisner 已提交
1285 1286 1287 1288
				/* but skip encoding of zero run length */
				toggle = !toggle;
				continue;
			}
1289
			dcbp_set_start(p, 0);
P
Philipp Reisner 已提交
1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328
		}

		/* paranoia: catch zero runlength.
		 * can only happen if bitmap is modified while we scan it. */
		if (rl == 0) {
			dev_err(DEV, "unexpected zero runlength while encoding bitmap "
			    "t:%u bo:%lu\n", toggle, c->bit_offset);
			return -1;
		}

		bits = vli_encode_bits(&bs, rl);
		if (bits == -ENOBUFS) /* buffer full */
			break;
		if (bits <= 0) {
			dev_err(DEV, "error while encoding bitmap: %d\n", bits);
			return 0;
		}

		toggle = !toggle;
		plain_bits += rl;
		c->bit_offset = tmp;
	} while (c->bit_offset < c->bm_bits);

	len = bs.cur.b - p->code + !!bs.cur.bit;

	if (plain_bits < (len << 3)) {
		/* incompressible with this method.
		 * we need to rewind both word and bit position. */
		c->bit_offset -= plain_bits;
		bm_xfer_ctx_bit_to_word_offset(c);
		c->bit_offset = c->word_offset * BITS_PER_LONG;
		return 0;
	}

	/* RLE + VLI was able to compress it just fine.
	 * update c->word_offset. */
	bm_xfer_ctx_bit_to_word_offset(c);

	/* store pad_bits */
1329
	dcbp_set_pad_bits(p, (8 - bs.cur.bit) & 0x7);
P
Philipp Reisner 已提交
1330 1331 1332 1333

	return len;
}

1334 1335 1336 1337 1338 1339 1340
/**
 * send_bitmap_rle_or_plain
 *
 * Return 0 when done, 1 when another iteration is needed, and a negative error
 * code upon failure.
 */
static int
1341
send_bitmap_rle_or_plain(struct drbd_conf *mdev, struct bm_xfer_ctx *c)
P
Philipp Reisner 已提交
1342
{
1343
	struct drbd_socket *sock = &mdev->tconn->data;
1344
	unsigned int header_size = drbd_header_size(mdev->tconn);
1345
	struct p_compressed_bm *p = sock->sbuf + header_size;
1346
	int len, err;
P
Philipp Reisner 已提交
1347

1348 1349
	len = fill_bitmap_rle_bits(mdev, p,
			DRBD_SOCKET_BUFFER_SIZE - header_size - sizeof(*p), c);
P
Philipp Reisner 已提交
1350
	if (len < 0)
1351
		return -EIO;
P
Philipp Reisner 已提交
1352 1353

	if (len) {
1354
		dcbp_set_code(p, RLE_VLI_Bits);
1355 1356 1357
		err = __send_command(mdev->tconn, mdev->vnr, sock,
				     P_COMPRESSED_BITMAP, sizeof(*p) + len,
				     NULL, 0);
P
Philipp Reisner 已提交
1358
		c->packets[0]++;
1359
		c->bytes[0] += header_size + sizeof(*p) + len;
P
Philipp Reisner 已提交
1360 1361 1362 1363 1364 1365

		if (c->bit_offset >= c->bm_bits)
			len = 0; /* DONE */
	} else {
		/* was not compressible.
		 * send a buffer full of plain text bits instead. */
1366 1367
		unsigned int data_size;
		unsigned long num_words;
1368
		unsigned long *p = sock->sbuf + header_size;
1369 1370

		data_size = DRBD_SOCKET_BUFFER_SIZE - header_size;
1371
		num_words = min_t(size_t, data_size / sizeof(*p),
1372
				  c->bm_words - c->word_offset);
1373
		len = num_words * sizeof(*p);
P
Philipp Reisner 已提交
1374
		if (len)
1375 1376
			drbd_bm_get_lel(mdev, c->word_offset, num_words, p);
		err = __send_command(mdev->tconn, mdev->vnr, sock, P_BITMAP, len, NULL, 0);
P
Philipp Reisner 已提交
1377 1378 1379 1380
		c->word_offset += num_words;
		c->bit_offset = c->word_offset * BITS_PER_LONG;

		c->packets[1]++;
1381
		c->bytes[1] += header_size + len;
P
Philipp Reisner 已提交
1382 1383 1384 1385

		if (c->bit_offset > c->bm_bits)
			c->bit_offset = c->bm_bits;
	}
1386
	if (!err) {
1387 1388 1389 1390 1391 1392 1393
		if (len == 0) {
			INFO_bm_xfer_stats(mdev, "send", c);
			return 0;
		} else
			return 1;
	}
	return -EIO;
P
Philipp Reisner 已提交
1394 1395 1396
}

/* See the comment at receive_bitmap() */
1397
static int _drbd_send_bitmap(struct drbd_conf *mdev)
P
Philipp Reisner 已提交
1398 1399
{
	struct bm_xfer_ctx c;
1400
	int err;
P
Philipp Reisner 已提交
1401

1402 1403
	if (!expect(mdev->bitmap))
		return false;
P
Philipp Reisner 已提交
1404 1405 1406 1407 1408 1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 1419 1420 1421 1422 1423 1424 1425 1426 1427

	if (get_ldev(mdev)) {
		if (drbd_md_test_flag(mdev->ldev, MDF_FULL_SYNC)) {
			dev_info(DEV, "Writing the whole bitmap, MDF_FullSync was set.\n");
			drbd_bm_set_all(mdev);
			if (drbd_bm_write(mdev)) {
				/* write_bm did fail! Leave full sync flag set in Meta P_DATA
				 * but otherwise process as per normal - need to tell other
				 * side that a full resync is required! */
				dev_err(DEV, "Failed to write bitmap to disk!\n");
			} else {
				drbd_md_clear_flag(mdev, MDF_FULL_SYNC);
				drbd_md_sync(mdev);
			}
		}
		put_ldev(mdev);
	}

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

	do {
1428
		err = send_bitmap_rle_or_plain(mdev, &c);
1429
	} while (err > 0);
P
Philipp Reisner 已提交
1430

1431
	return err == 0;
P
Philipp Reisner 已提交
1432 1433 1434 1435
}

int drbd_send_bitmap(struct drbd_conf *mdev)
{
1436 1437
	struct drbd_socket *sock = &mdev->tconn->data;
	int err = -1;
P
Philipp Reisner 已提交
1438

1439 1440 1441 1442
	mutex_lock(&sock->mutex);
	if (sock->socket)
		err = !_drbd_send_bitmap(mdev);
	mutex_unlock(&sock->mutex);
P
Philipp Reisner 已提交
1443 1444
	return err;
}
1445

1446
void drbd_send_b_ack(struct drbd_conf *mdev, u32 barrier_nr, u32 set_size)
P
Philipp Reisner 已提交
1447
{
1448 1449
	struct drbd_socket *sock;
	struct p_barrier_ack *p;
P
Philipp Reisner 已提交
1450

1451 1452
	if (mdev->state.conn < C_CONNECTED)
		return;
P
Philipp Reisner 已提交
1453

1454 1455 1456 1457 1458 1459 1460
	sock = &mdev->tconn->meta;
	p = drbd_prepare_command(mdev, sock);
	if (!p)
		return;
	p->barrier = barrier_nr;
	p->set_size = cpu_to_be32(set_size);
	drbd_send_command(mdev, sock, P_BARRIER_ACK, sizeof(*p), NULL, 0);
P
Philipp Reisner 已提交
1461 1462 1463 1464 1465 1466 1467 1468 1469 1470
}

/**
 * _drbd_send_ack() - Sends an ack packet
 * @mdev:	DRBD device.
 * @cmd:	Packet command code.
 * @sector:	sector, needs to be in big endian byte order
 * @blksize:	size in byte, needs to be in big endian byte order
 * @block_id:	Id, big endian byte order
 */
1471 1472
static int _drbd_send_ack(struct drbd_conf *mdev, enum drbd_packet cmd,
			  u64 sector, u32 blksize, u64 block_id)
P
Philipp Reisner 已提交
1473
{
1474 1475
	struct drbd_socket *sock;
	struct p_block_ack *p;
P
Philipp Reisner 已提交
1476

1477 1478
	if (mdev->state.conn < C_CONNECTED)
		return -EIO;
P
Philipp Reisner 已提交
1479

1480 1481 1482
	sock = &mdev->tconn->meta;
	p = drbd_prepare_command(mdev, sock);
	if (!p)
1483
		return -EIO;
1484 1485 1486 1487 1488
	p->sector = sector;
	p->block_id = block_id;
	p->blksize = blksize;
	p->seq_num = cpu_to_be32(atomic_inc_return(&mdev->packet_seq));
	return drbd_send_command(mdev, sock, cmd, sizeof(*p), NULL, 0);
P
Philipp Reisner 已提交
1489 1490
}

1491 1492 1493
/* dp->sector and dp->block_id already/still in network byte order,
 * data_size is payload size according to dp->head,
 * and may need to be corrected for digest size. */
1494 1495
void drbd_send_ack_dp(struct drbd_conf *mdev, enum drbd_packet cmd,
		      struct p_data *dp, int data_size)
P
Philipp Reisner 已提交
1496
{
1497 1498
	if (mdev->tconn->peer_integrity_tfm)
		data_size -= crypto_hash_digestsize(mdev->tconn->peer_integrity_tfm);
1499 1500
	_drbd_send_ack(mdev, cmd, dp->sector, cpu_to_be32(data_size),
		       dp->block_id);
P
Philipp Reisner 已提交
1501 1502
}

1503 1504
void drbd_send_ack_rp(struct drbd_conf *mdev, enum drbd_packet cmd,
		      struct p_block_req *rp)
P
Philipp Reisner 已提交
1505
{
1506
	_drbd_send_ack(mdev, cmd, rp->sector, rp->blksize, rp->block_id);
P
Philipp Reisner 已提交
1507 1508 1509 1510
}

/**
 * drbd_send_ack() - Sends an ack packet
1511 1512 1513
 * @mdev:	DRBD device
 * @cmd:	packet command code
 * @peer_req:	peer request
P
Philipp Reisner 已提交
1514
 */
1515
int drbd_send_ack(struct drbd_conf *mdev, enum drbd_packet cmd,
1516
		  struct drbd_peer_request *peer_req)
P
Philipp Reisner 已提交
1517
{
1518 1519 1520 1521
	return _drbd_send_ack(mdev, cmd,
			      cpu_to_be64(peer_req->i.sector),
			      cpu_to_be32(peer_req->i.size),
			      peer_req->block_id);
P
Philipp Reisner 已提交
1522 1523 1524 1525
}

/* This function misuses the block_id field to signal if the blocks
 * are is sync or not. */
1526
int drbd_send_ack_ex(struct drbd_conf *mdev, enum drbd_packet cmd,
P
Philipp Reisner 已提交
1527 1528
		     sector_t sector, int blksize, u64 block_id)
{
1529 1530 1531 1532
	return _drbd_send_ack(mdev, cmd,
			      cpu_to_be64(sector),
			      cpu_to_be32(blksize),
			      cpu_to_be64(block_id));
P
Philipp Reisner 已提交
1533 1534 1535 1536 1537
}

int drbd_send_drequest(struct drbd_conf *mdev, int cmd,
		       sector_t sector, int size, u64 block_id)
{
1538 1539
	struct drbd_socket *sock;
	struct p_block_req *p;
P
Philipp Reisner 已提交
1540

1541 1542 1543 1544 1545 1546 1547 1548
	sock = &mdev->tconn->data;
	p = drbd_prepare_command(mdev, sock);
	if (!p)
		return -EIO;
	p->sector = cpu_to_be64(sector);
	p->block_id = block_id;
	p->blksize = cpu_to_be32(size);
	return drbd_send_command(mdev, sock, cmd, sizeof(*p), NULL, 0);
P
Philipp Reisner 已提交
1549 1550
}

1551 1552
int drbd_send_drequest_csum(struct drbd_conf *mdev, sector_t sector, int size,
			    void *digest, int digest_size, enum drbd_packet cmd)
P
Philipp Reisner 已提交
1553
{
1554 1555
	struct drbd_socket *sock;
	struct p_block_req *p;
P
Philipp Reisner 已提交
1556

1557
	/* FIXME: Put the digest into the preallocated socket buffer.  */
P
Philipp Reisner 已提交
1558

1559 1560 1561 1562 1563 1564 1565 1566 1567
	sock = &mdev->tconn->data;
	p = drbd_prepare_command(mdev, sock);
	if (!p)
		return -EIO;
	p->sector = cpu_to_be64(sector);
	p->block_id = ID_SYNCER /* unused */;
	p->blksize = cpu_to_be32(size);
	return drbd_send_command(mdev, sock, cmd, sizeof(*p),
				 digest, digest_size);
P
Philipp Reisner 已提交
1568 1569 1570 1571
}

int drbd_send_ov_request(struct drbd_conf *mdev, sector_t sector, int size)
{
1572 1573
	struct drbd_socket *sock;
	struct p_block_req *p;
P
Philipp Reisner 已提交
1574

1575 1576 1577 1578 1579 1580 1581 1582
	sock = &mdev->tconn->data;
	p = drbd_prepare_command(mdev, sock);
	if (!p)
		return -EIO;
	p->sector = cpu_to_be64(sector);
	p->block_id = ID_SYNCER /* unused */;
	p->blksize = cpu_to_be32(size);
	return drbd_send_command(mdev, sock, P_OV_REQUEST, sizeof(*p), NULL, 0);
P
Philipp Reisner 已提交
1583 1584 1585
}

/* called on sndtimeo
1586 1587
 * returns false if we should retry,
 * true if we think connection is dead
P
Philipp Reisner 已提交
1588
 */
1589
static int we_should_drop_the_connection(struct drbd_tconn *tconn, struct socket *sock)
P
Philipp Reisner 已提交
1590 1591 1592 1593
{
	int drop_it;
	/* long elapsed = (long)(jiffies - mdev->last_received); */

1594 1595 1596
	drop_it =   tconn->meta.socket == sock
		|| !tconn->asender.task
		|| get_t_state(&tconn->asender) != RUNNING
1597
		|| tconn->cstate < C_WF_REPORT_PARAMS;
P
Philipp Reisner 已提交
1598 1599

	if (drop_it)
1600
		return true;
P
Philipp Reisner 已提交
1601

1602
	drop_it = !--tconn->ko_count;
P
Philipp Reisner 已提交
1603
	if (!drop_it) {
1604 1605 1606
		conn_err(tconn, "[%s/%d] sock_sendmsg time expired, ko = %u\n",
			 current->comm, current->pid, tconn->ko_count);
		request_ping(tconn);
P
Philipp Reisner 已提交
1607 1608 1609 1610 1611
	}

	return drop_it; /* && (mdev->state == R_PRIMARY) */;
}

1612
static void drbd_update_congested(struct drbd_tconn *tconn)
1613
{
1614
	struct sock *sk = tconn->data.socket->sk;
1615
	if (sk->sk_wmem_queued > sk->sk_sndbuf * 4 / 5)
1616
		set_bit(NET_CONGESTED, &tconn->flags);
1617 1618
}

P
Philipp Reisner 已提交
1619 1620 1621 1622 1623 1624 1625 1626 1627 1628 1629 1630 1631 1632 1633 1634 1635 1636 1637 1638 1639 1640
/* The idea of sendpage seems to be to put some kind of reference
 * to the page into the skb, and to hand it over to the NIC. In
 * this process get_page() gets called.
 *
 * As soon as the page was really sent over the network put_page()
 * gets called by some part of the network layer. [ NIC driver? ]
 *
 * [ get_page() / put_page() increment/decrement the count. If count
 *   reaches 0 the page will be freed. ]
 *
 * This works nicely with pages from FSs.
 * But this means that in protocol A we might signal IO completion too early!
 *
 * In order not to corrupt data during a resync we must make sure
 * that we do not reuse our own buffer pages (EEs) to early, therefore
 * we have the net_ee list.
 *
 * XFS seems to have problems, still, it submits pages with page_count == 0!
 * As a workaround, we disable sendpage on pages
 * with page_count == 0 or PageSlab.
 */
static int _drbd_no_send_page(struct drbd_conf *mdev, struct page *page,
1641
			      int offset, size_t size, unsigned msg_flags)
P
Philipp Reisner 已提交
1642
{
1643 1644 1645 1646 1647 1648 1649
	struct socket *socket;
	void *addr;
	int err;

	socket = mdev->tconn->data.socket;
	addr = kmap(page) + offset;
	err = drbd_send_all(mdev->tconn, socket, addr, size, msg_flags);
P
Philipp Reisner 已提交
1650
	kunmap(page);
1651 1652 1653
	if (!err)
		mdev->send_cnt += size >> 9;
	return err;
P
Philipp Reisner 已提交
1654 1655 1656
}

static int _drbd_send_page(struct drbd_conf *mdev, struct page *page,
L
Lars Ellenberg 已提交
1657
		    int offset, size_t size, unsigned msg_flags)
P
Philipp Reisner 已提交
1658
{
1659
	struct socket *socket = mdev->tconn->data.socket;
P
Philipp Reisner 已提交
1660 1661
	mm_segment_t oldfs = get_fs();
	int len = size;
1662
	int err = -EIO;
P
Philipp Reisner 已提交
1663 1664 1665 1666 1667 1668 1669 1670

	/* e.g. XFS meta- & log-data is in slab pages, which have a
	 * page_count of 0 and/or have PageSlab() set.
	 * we cannot use send_page for those, as that does get_page();
	 * put_page(); and would cause either a VM_BUG directly, or
	 * __page_cache_release a page that would actually still be referenced
	 * by someone, leading to some obscure delayed Oops somewhere else. */
	if (disable_sendpage || (page_count(page) < 1) || PageSlab(page))
1671
		return _drbd_no_send_page(mdev, page, offset, size, msg_flags);
P
Philipp Reisner 已提交
1672

L
Lars Ellenberg 已提交
1673
	msg_flags |= MSG_NOSIGNAL;
1674
	drbd_update_congested(mdev->tconn);
P
Philipp Reisner 已提交
1675 1676
	set_fs(KERNEL_DS);
	do {
1677 1678 1679
		int sent;

		sent = socket->ops->sendpage(socket, page, offset, len, msg_flags);
P
Philipp Reisner 已提交
1680
		if (sent <= 0) {
1681 1682 1683 1684 1685
			if (sent == -EAGAIN) {
				if (we_should_drop_the_connection(mdev->tconn, socket))
					break;
				continue;
			}
P
Philipp Reisner 已提交
1686 1687
			dev_warn(DEV, "%s: size=%d len=%d sent=%d\n",
			     __func__, (int)size, len, sent);
1688 1689
			if (sent < 0)
				err = sent;
P
Philipp Reisner 已提交
1690 1691 1692 1693 1694 1695
			break;
		}
		len    -= sent;
		offset += sent;
	} while (len > 0 /* THINK && mdev->cstate >= C_CONNECTED*/);
	set_fs(oldfs);
1696
	clear_bit(NET_CONGESTED, &mdev->tconn->flags);
P
Philipp Reisner 已提交
1697

1698 1699 1700 1701 1702
	if (len == 0) {
		err = 0;
		mdev->send_cnt += size >> 9;
	}
	return err;
P
Philipp Reisner 已提交
1703 1704 1705 1706 1707 1708
}

static int _drbd_send_bio(struct drbd_conf *mdev, struct bio *bio)
{
	struct bio_vec *bvec;
	int i;
L
Lars Ellenberg 已提交
1709
	/* hint all but last page with MSG_MORE */
P
Philipp Reisner 已提交
1710
	__bio_for_each_segment(bvec, bio, i, 0) {
1711 1712 1713 1714 1715 1716 1717
		int err;

		err = _drbd_no_send_page(mdev, bvec->bv_page,
					 bvec->bv_offset, bvec->bv_len,
					 i == bio->bi_vcnt - 1 ? 0 : MSG_MORE);
		if (err)
			return err;
P
Philipp Reisner 已提交
1718
	}
1719
	return 0;
P
Philipp Reisner 已提交
1720 1721 1722 1723 1724 1725
}

static int _drbd_send_zc_bio(struct drbd_conf *mdev, struct bio *bio)
{
	struct bio_vec *bvec;
	int i;
L
Lars Ellenberg 已提交
1726
	/* hint all but last page with MSG_MORE */
P
Philipp Reisner 已提交
1727
	__bio_for_each_segment(bvec, bio, i, 0) {
1728 1729 1730 1731 1732 1733 1734
		int err;

		err = _drbd_send_page(mdev, bvec->bv_page,
				      bvec->bv_offset, bvec->bv_len,
				      i == bio->bi_vcnt - 1 ? 0 : MSG_MORE);
		if (err)
			return err;
P
Philipp Reisner 已提交
1735
	}
1736
	return 0;
P
Philipp Reisner 已提交
1737 1738
}

1739 1740
static int _drbd_send_zc_ee(struct drbd_conf *mdev,
			    struct drbd_peer_request *peer_req)
1741
{
1742 1743
	struct page *page = peer_req->pages;
	unsigned len = peer_req->i.size;
1744
	int err;
1745

L
Lars Ellenberg 已提交
1746
	/* hint all but last page with MSG_MORE */
1747 1748
	page_chain_for_each(page) {
		unsigned l = min_t(unsigned, len, PAGE_SIZE);
1749 1750 1751 1752 1753

		err = _drbd_send_page(mdev, page, 0, l,
				      page_chain_next(page) ? MSG_MORE : 0);
		if (err)
			return err;
1754 1755
		len -= l;
	}
1756
	return 0;
1757 1758
}

1759 1760
static u32 bio_flags_to_wire(struct drbd_conf *mdev, unsigned long bi_rw)
{
1761
	if (mdev->tconn->agreed_pro_version >= 95)
1762 1763 1764 1765 1766
		return  (bi_rw & REQ_SYNC ? DP_RW_SYNC : 0) |
			(bi_rw & REQ_FUA ? DP_FUA : 0) |
			(bi_rw & REQ_FLUSH ? DP_FLUSH : 0) |
			(bi_rw & REQ_DISCARD ? DP_DISCARD : 0);
	else
J
Jens Axboe 已提交
1767
		return bi_rw & REQ_SYNC ? DP_RW_SYNC : 0;
1768 1769
}

P
Philipp Reisner 已提交
1770 1771 1772 1773 1774
/* Used to send write requests
 * R_PRIMARY -> Peer	(P_DATA)
 */
int drbd_send_dblock(struct drbd_conf *mdev, struct drbd_request *req)
{
1775 1776
	struct drbd_socket *sock;
	struct p_data *p;
P
Philipp Reisner 已提交
1777 1778
	unsigned int dp_flags = 0;
	int dgs;
1779
	int err;
P
Philipp Reisner 已提交
1780

1781 1782
	sock = &mdev->tconn->data;
	p = drbd_prepare_command(mdev, sock);
1783 1784
	dgs = (mdev->tconn->agreed_pro_version >= 87 && mdev->tconn->integrity_tfm) ?
		crypto_hash_digestsize(mdev->tconn->integrity_tfm) : 0;
P
Philipp Reisner 已提交
1785

1786 1787 1788 1789 1790
	if (!p)
		return -EIO;
	p->sector = cpu_to_be64(req->i.sector);
	p->block_id = (unsigned long)req;
	p->seq_num = cpu_to_be32(req->seq_num = atomic_inc_return(&mdev->packet_seq));
1791
	dp_flags = bio_flags_to_wire(mdev, req->master_bio->bi_rw);
P
Philipp Reisner 已提交
1792 1793 1794
	if (mdev->state.conn >= C_SYNC_SOURCE &&
	    mdev->state.conn <= C_PAUSED_SYNC_T)
		dp_flags |= DP_MAY_SET_IN_SYNC;
1795 1796 1797 1798 1799 1800
	if (mdev->tconn->agreed_pro_version >= 100) {
		if (req->rq_state & RQ_EXP_RECEIVE_ACK)
			dp_flags |= DP_SEND_RECEIVE_ACK;
		if (req->rq_state & RQ_EXP_WRITE_ACK)
			dp_flags |= DP_SEND_WRITE_ACK;
	}
1801 1802
	p->dp_flags = cpu_to_be32(dp_flags);
	if (dgs)
1803
		drbd_csum_bio(mdev, mdev->tconn->integrity_tfm, req->master_bio, p + 1);
1804
	err = __send_command(mdev->tconn, mdev->vnr, sock, P_DATA, sizeof(*p) + dgs, NULL, req->i.size);
1805
	if (!err) {
1806 1807 1808 1809 1810 1811 1812 1813 1814 1815 1816
		/* For protocol A, we have to memcpy the payload into
		 * socket buffers, as we may complete right away
		 * as soon as we handed it over to tcp, at which point the data
		 * pages may become invalid.
		 *
		 * For data-integrity enabled, we copy it as well, so we can be
		 * sure that even if the bio pages may still be modified, it
		 * won't change the data on the wire, thus if the digest checks
		 * out ok after sending on this side, but does not fit on the
		 * receiving side, we sure have detected corruption elsewhere.
		 */
1817
		if (!(req->rq_state & (RQ_EXP_RECEIVE_ACK | RQ_EXP_WRITE_ACK)) || dgs)
1818
			err = _drbd_send_bio(mdev, req->master_bio);
P
Philipp Reisner 已提交
1819
		else
1820
			err = _drbd_send_zc_bio(mdev, req->master_bio);
1821 1822 1823

		/* double check digest, sometimes buffers have been modified in flight. */
		if (dgs > 0 && dgs <= 64) {
B
Bart Van Assche 已提交
1824
			/* 64 byte, 512 bit, is the largest digest size
1825 1826
			 * currently supported in kernel crypto. */
			unsigned char digest[64];
1827
			drbd_csum_bio(mdev, mdev->tconn->integrity_tfm, req->master_bio, digest);
1828
			if (memcmp(p + 1, digest, dgs)) {
1829 1830
				dev_warn(DEV,
					"Digest mismatch, buffer modified by upper layers during write: %llus +%u\n",
1831
					(unsigned long long)req->i.sector, req->i.size);
1832 1833 1834 1835
			}
		} /* else if (dgs > 64) {
		     ... Be noisy about digest too large ...
		} */
P
Philipp Reisner 已提交
1836
	}
1837
	mutex_unlock(&sock->mutex);  /* locked by drbd_prepare_command() */
1838

1839
	return err;
P
Philipp Reisner 已提交
1840 1841 1842 1843 1844 1845
}

/* answer packet, used to send data back for read requests:
 *  Peer       -> (diskless) R_PRIMARY   (P_DATA_REPLY)
 *  C_SYNC_SOURCE -> C_SYNC_TARGET         (P_RS_DATA_REPLY)
 */
1846
int drbd_send_block(struct drbd_conf *mdev, enum drbd_packet cmd,
1847
		    struct drbd_peer_request *peer_req)
P
Philipp Reisner 已提交
1848
{
1849 1850
	struct drbd_socket *sock;
	struct p_data *p;
1851
	int err;
P
Philipp Reisner 已提交
1852 1853
	int dgs;

1854 1855 1856
	sock = &mdev->tconn->data;
	p = drbd_prepare_command(mdev, sock);

1857 1858
	dgs = (mdev->tconn->agreed_pro_version >= 87 && mdev->tconn->integrity_tfm) ?
		crypto_hash_digestsize(mdev->tconn->integrity_tfm) : 0;
P
Philipp Reisner 已提交
1859

1860 1861 1862 1863 1864 1865
	if (!p)
		return -EIO;
	p->sector = cpu_to_be64(peer_req->i.sector);
	p->block_id = peer_req->block_id;
	p->seq_num = 0;  /* unused */
	if (dgs)
1866
		drbd_csum_ee(mdev, mdev->tconn->integrity_tfm, peer_req, p + 1);
1867
	err = __send_command(mdev->tconn, mdev->vnr, sock, cmd, sizeof(*p) + dgs, NULL, peer_req->i.size);
1868 1869
	if (!err)
		err = _drbd_send_zc_ee(mdev, peer_req);
1870
	mutex_unlock(&sock->mutex);  /* locked by drbd_prepare_command() */
1871

1872
	return err;
P
Philipp Reisner 已提交
1873 1874
}

1875
int drbd_send_out_of_sync(struct drbd_conf *mdev, struct drbd_request *req)
1876
{
1877 1878
	struct drbd_socket *sock;
	struct p_block_desc *p;
1879

1880 1881 1882 1883 1884 1885 1886
	sock = &mdev->tconn->data;
	p = drbd_prepare_command(mdev, sock);
	if (!p)
		return -EIO;
	p->sector = cpu_to_be64(req->i.sector);
	p->blksize = cpu_to_be32(req->i.size);
	return drbd_send_command(mdev, sock, P_OUT_OF_SYNC, sizeof(*p), NULL, 0);
1887 1888
}

P
Philipp Reisner 已提交
1889 1890 1891 1892 1893 1894 1895 1896 1897 1898 1899 1900 1901 1902 1903 1904
/*
  drbd_send distinguishes two cases:

  Packets sent via the data socket "sock"
  and packets sent via the meta data socket "msock"

		    sock                      msock
  -----------------+-------------------------+------------------------------
  timeout           conf.timeout / 2          conf.timeout / 2
  timeout action    send a ping via msock     Abort communication
					      and close all sockets
*/

/*
 * you must have down()ed the appropriate [m]sock_mutex elsewhere!
 */
1905
int drbd_send(struct drbd_tconn *tconn, struct socket *sock,
P
Philipp Reisner 已提交
1906 1907 1908 1909 1910 1911 1912
	      void *buf, size_t size, unsigned msg_flags)
{
	struct kvec iov;
	struct msghdr msg;
	int rv, sent = 0;

	if (!sock)
1913
		return -EBADR;
P
Philipp Reisner 已提交
1914 1915 1916 1917 1918 1919 1920 1921 1922 1923 1924 1925

	/* THINK  if (signal_pending) return ... ? */

	iov.iov_base = buf;
	iov.iov_len  = size;

	msg.msg_name       = NULL;
	msg.msg_namelen    = 0;
	msg.msg_control    = NULL;
	msg.msg_controllen = 0;
	msg.msg_flags      = msg_flags | MSG_NOSIGNAL;

1926
	if (sock == tconn->data.socket) {
1927 1928 1929
		rcu_read_lock();
		tconn->ko_count = rcu_dereference(tconn->net_conf)->ko_count;
		rcu_read_unlock();
1930
		drbd_update_congested(tconn);
P
Philipp Reisner 已提交
1931 1932 1933 1934 1935 1936 1937 1938 1939 1940 1941 1942 1943
	}
	do {
		/* STRANGE
		 * tcp_sendmsg does _not_ use its size parameter at all ?
		 *
		 * -EAGAIN on timeout, -EINTR on signal.
		 */
/* THINK
 * do we need to block DRBD_SIG if sock == &meta.socket ??
 * otherwise wake_asender() might interrupt some send_*Ack !
 */
		rv = kernel_sendmsg(sock, &msg, &iov, 1, size);
		if (rv == -EAGAIN) {
1944
			if (we_should_drop_the_connection(tconn, sock))
P
Philipp Reisner 已提交
1945 1946 1947 1948 1949 1950 1951 1952 1953 1954 1955 1956 1957 1958 1959
				break;
			else
				continue;
		}
		if (rv == -EINTR) {
			flush_signals(current);
			rv = 0;
		}
		if (rv < 0)
			break;
		sent += rv;
		iov.iov_base += rv;
		iov.iov_len  -= rv;
	} while (sent < size);

1960 1961
	if (sock == tconn->data.socket)
		clear_bit(NET_CONGESTED, &tconn->flags);
P
Philipp Reisner 已提交
1962 1963 1964

	if (rv <= 0) {
		if (rv != -EAGAIN) {
1965 1966 1967
			conn_err(tconn, "%s_sendmsg returned %d\n",
				 sock == tconn->meta.socket ? "msock" : "sock",
				 rv);
1968
			conn_request_state(tconn, NS(conn, C_BROKEN_PIPE), CS_HARD);
P
Philipp Reisner 已提交
1969
		} else
1970
			conn_request_state(tconn, NS(conn, C_TIMEOUT), CS_HARD);
P
Philipp Reisner 已提交
1971 1972 1973 1974 1975
	}

	return sent;
}

1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993
/**
 * drbd_send_all  -  Send an entire buffer
 *
 * Returns 0 upon success and a negative error value otherwise.
 */
int drbd_send_all(struct drbd_tconn *tconn, struct socket *sock, void *buffer,
		  size_t size, unsigned msg_flags)
{
	int err;

	err = drbd_send(tconn, sock, buffer, size, msg_flags);
	if (err < 0)
		return err;
	if (err != size)
		return -EIO;
	return 0;
}

P
Philipp Reisner 已提交
1994 1995 1996 1997 1998 1999
static int drbd_open(struct block_device *bdev, fmode_t mode)
{
	struct drbd_conf *mdev = bdev->bd_disk->private_data;
	unsigned long flags;
	int rv = 0;

2000
	mutex_lock(&drbd_main_mutex);
2001
	spin_lock_irqsave(&mdev->tconn->req_lock, flags);
P
Philipp Reisner 已提交
2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013
	/* to have a stable mdev->state.role
	 * and no race with updating open_cnt */

	if (mdev->state.role != R_PRIMARY) {
		if (mode & FMODE_WRITE)
			rv = -EROFS;
		else if (!allow_oos)
			rv = -EMEDIUMTYPE;
	}

	if (!rv)
		mdev->open_cnt++;
2014
	spin_unlock_irqrestore(&mdev->tconn->req_lock, flags);
2015
	mutex_unlock(&drbd_main_mutex);
P
Philipp Reisner 已提交
2016 2017 2018 2019 2020 2021 2022

	return rv;
}

static int drbd_release(struct gendisk *gd, fmode_t mode)
{
	struct drbd_conf *mdev = gd->private_data;
2023
	mutex_lock(&drbd_main_mutex);
P
Philipp Reisner 已提交
2024
	mdev->open_cnt--;
2025
	mutex_unlock(&drbd_main_mutex);
P
Philipp Reisner 已提交
2026 2027 2028 2029 2030
	return 0;
}

static void drbd_set_defaults(struct drbd_conf *mdev)
{
2031 2032
	/* Beware! The actual layout differs
	 * between big endian and little endian */
2033
	mdev->state = (union drbd_dev_state) {
P
Philipp Reisner 已提交
2034 2035 2036 2037 2038 2039 2040 2041 2042 2043 2044 2045 2046 2047 2048 2049 2050 2051 2052 2053
		{ .role = R_SECONDARY,
		  .peer = R_UNKNOWN,
		  .conn = C_STANDALONE,
		  .disk = D_DISKLESS,
		  .pdsk = D_UNKNOWN,
		} };
}

void drbd_init_set_defaults(struct drbd_conf *mdev)
{
	/* the memset(,0,) did most of this.
	 * note: only assignments, no allocation in here */

	drbd_set_defaults(mdev);

	atomic_set(&mdev->ap_bio_cnt, 0);
	atomic_set(&mdev->ap_pending_cnt, 0);
	atomic_set(&mdev->rs_pending_cnt, 0);
	atomic_set(&mdev->unacked_cnt, 0);
	atomic_set(&mdev->local_cnt, 0);
2054
	atomic_set(&mdev->pp_in_use_by_net, 0);
2055
	atomic_set(&mdev->rs_sect_in, 0);
2056
	atomic_set(&mdev->rs_sect_ev, 0);
2057
	atomic_set(&mdev->ap_in_flight, 0);
2058
	atomic_set(&mdev->md_io_in_use, 0);
P
Philipp Reisner 已提交
2059

2060 2061
	mutex_init(&mdev->own_state_mutex);
	mdev->state_mutex = &mdev->own_state_mutex;
P
Philipp Reisner 已提交
2062 2063 2064 2065 2066 2067 2068 2069 2070 2071 2072 2073 2074

	spin_lock_init(&mdev->al_lock);
	spin_lock_init(&mdev->peer_seq_lock);
	spin_lock_init(&mdev->epoch_lock);

	INIT_LIST_HEAD(&mdev->active_ee);
	INIT_LIST_HEAD(&mdev->sync_ee);
	INIT_LIST_HEAD(&mdev->done_ee);
	INIT_LIST_HEAD(&mdev->read_ee);
	INIT_LIST_HEAD(&mdev->net_ee);
	INIT_LIST_HEAD(&mdev->resync_reads);
	INIT_LIST_HEAD(&mdev->resync_work.list);
	INIT_LIST_HEAD(&mdev->unplug_work.list);
2075
	INIT_LIST_HEAD(&mdev->go_diskless.list);
P
Philipp Reisner 已提交
2076
	INIT_LIST_HEAD(&mdev->md_sync_work.list);
2077
	INIT_LIST_HEAD(&mdev->start_resync_work.list);
P
Philipp Reisner 已提交
2078
	INIT_LIST_HEAD(&mdev->bm_io_work.w.list);
2079

2080
	mdev->resync_work.cb  = w_resync_timer;
P
Philipp Reisner 已提交
2081
	mdev->unplug_work.cb  = w_send_write_hint;
2082
	mdev->go_diskless.cb  = w_go_diskless;
P
Philipp Reisner 已提交
2083 2084
	mdev->md_sync_work.cb = w_md_sync;
	mdev->bm_io_work.w.cb = w_bitmap_io;
2085
	mdev->start_resync_work.cb = w_start_resync;
2086 2087 2088 2089 2090 2091 2092 2093

	mdev->resync_work.mdev  = mdev;
	mdev->unplug_work.mdev  = mdev;
	mdev->go_diskless.mdev  = mdev;
	mdev->md_sync_work.mdev = mdev;
	mdev->bm_io_work.w.mdev = mdev;
	mdev->start_resync_work.mdev = mdev;

P
Philipp Reisner 已提交
2094 2095
	init_timer(&mdev->resync_timer);
	init_timer(&mdev->md_sync_timer);
2096
	init_timer(&mdev->start_resync_timer);
2097
	init_timer(&mdev->request_timer);
P
Philipp Reisner 已提交
2098 2099 2100 2101
	mdev->resync_timer.function = resync_timer_fn;
	mdev->resync_timer.data = (unsigned long) mdev;
	mdev->md_sync_timer.function = md_sync_timer_fn;
	mdev->md_sync_timer.data = (unsigned long) mdev;
2102 2103
	mdev->start_resync_timer.function = start_resync_timer_fn;
	mdev->start_resync_timer.data = (unsigned long) mdev;
2104 2105
	mdev->request_timer.function = request_timer_fn;
	mdev->request_timer.data = (unsigned long) mdev;
P
Philipp Reisner 已提交
2106 2107 2108 2109 2110 2111 2112

	init_waitqueue_head(&mdev->misc_wait);
	init_waitqueue_head(&mdev->state_wait);
	init_waitqueue_head(&mdev->ee_wait);
	init_waitqueue_head(&mdev->al_wait);
	init_waitqueue_head(&mdev->seq_wait);

2113
	mdev->write_ordering = WO_bdev_flush;
P
Philipp Reisner 已提交
2114
	mdev->resync_wenr = LC_FREE;
2115 2116
	mdev->peer_max_bio_size = DRBD_MAX_BIO_SIZE_SAFE;
	mdev->local_max_bio_size = DRBD_MAX_BIO_SIZE_SAFE;
P
Philipp Reisner 已提交
2117 2118 2119 2120
}

void drbd_mdev_cleanup(struct drbd_conf *mdev)
{
2121
	int i;
2122
	if (mdev->tconn->receiver.t_state != NONE)
P
Philipp Reisner 已提交
2123
		dev_err(DEV, "ASSERT FAILED: receiver t_state == %d expected 0.\n",
2124
				mdev->tconn->receiver.t_state);
P
Philipp Reisner 已提交
2125 2126 2127 2128 2129 2130 2131 2132 2133 2134 2135 2136 2137

	/* no need to lock it, I'm the only thread alive */
	if (atomic_read(&mdev->current_epoch->epoch_size) !=  0)
		dev_err(DEV, "epoch_size:%d\n", atomic_read(&mdev->current_epoch->epoch_size));
	mdev->al_writ_cnt  =
	mdev->bm_writ_cnt  =
	mdev->read_cnt     =
	mdev->recv_cnt     =
	mdev->send_cnt     =
	mdev->writ_cnt     =
	mdev->p_size       =
	mdev->rs_start     =
	mdev->rs_total     =
2138 2139
	mdev->rs_failed    = 0;
	mdev->rs_last_events = 0;
2140
	mdev->rs_last_sect_ev = 0;
2141 2142 2143 2144
	for (i = 0; i < DRBD_SYNC_MARKS; i++) {
		mdev->rs_mark_left[i] = 0;
		mdev->rs_mark_time[i] = 0;
	}
2145
	D_ASSERT(mdev->tconn->net_conf == NULL);
P
Philipp Reisner 已提交
2146 2147 2148 2149

	drbd_set_my_capacity(mdev, 0);
	if (mdev->bitmap) {
		/* maybe never allocated. */
2150
		drbd_bm_resize(mdev, 0, 1);
P
Philipp Reisner 已提交
2151 2152 2153
		drbd_bm_cleanup(mdev);
	}

2154 2155 2156
	drbd_free_bc(mdev->ldev);
	mdev->ldev = NULL;

2157
	clear_bit(AL_SUSPENDED, &mdev->flags);
P
Philipp Reisner 已提交
2158 2159 2160 2161 2162 2163 2164

	D_ASSERT(list_empty(&mdev->active_ee));
	D_ASSERT(list_empty(&mdev->sync_ee));
	D_ASSERT(list_empty(&mdev->done_ee));
	D_ASSERT(list_empty(&mdev->read_ee));
	D_ASSERT(list_empty(&mdev->net_ee));
	D_ASSERT(list_empty(&mdev->resync_reads));
2165 2166
	D_ASSERT(list_empty(&mdev->tconn->data.work.q));
	D_ASSERT(list_empty(&mdev->tconn->meta.work.q));
P
Philipp Reisner 已提交
2167 2168
	D_ASSERT(list_empty(&mdev->resync_work.list));
	D_ASSERT(list_empty(&mdev->unplug_work.list));
2169
	D_ASSERT(list_empty(&mdev->go_diskless.list));
2170 2171

	drbd_set_defaults(mdev);
P
Philipp Reisner 已提交
2172 2173 2174 2175 2176 2177 2178 2179 2180 2181 2182 2183 2184 2185 2186 2187
}


static void drbd_destroy_mempools(void)
{
	struct page *page;

	while (drbd_pp_pool) {
		page = drbd_pp_pool;
		drbd_pp_pool = (struct page *)page_private(page);
		__free_page(page);
		drbd_pp_vacant--;
	}

	/* D_ASSERT(atomic_read(&drbd_pp_vacant)==0); */

2188 2189
	if (drbd_md_io_bio_set)
		bioset_free(drbd_md_io_bio_set);
2190 2191
	if (drbd_md_io_page_pool)
		mempool_destroy(drbd_md_io_page_pool);
P
Philipp Reisner 已提交
2192 2193 2194 2195 2196 2197 2198 2199 2200 2201 2202 2203 2204
	if (drbd_ee_mempool)
		mempool_destroy(drbd_ee_mempool);
	if (drbd_request_mempool)
		mempool_destroy(drbd_request_mempool);
	if (drbd_ee_cache)
		kmem_cache_destroy(drbd_ee_cache);
	if (drbd_request_cache)
		kmem_cache_destroy(drbd_request_cache);
	if (drbd_bm_ext_cache)
		kmem_cache_destroy(drbd_bm_ext_cache);
	if (drbd_al_ext_cache)
		kmem_cache_destroy(drbd_al_ext_cache);

2205
	drbd_md_io_bio_set   = NULL;
2206
	drbd_md_io_page_pool = NULL;
P
Philipp Reisner 已提交
2207 2208 2209 2210 2211 2212 2213 2214 2215 2216 2217 2218 2219
	drbd_ee_mempool      = NULL;
	drbd_request_mempool = NULL;
	drbd_ee_cache        = NULL;
	drbd_request_cache   = NULL;
	drbd_bm_ext_cache    = NULL;
	drbd_al_ext_cache    = NULL;

	return;
}

static int drbd_create_mempools(void)
{
	struct page *page;
2220
	const int number = (DRBD_MAX_BIO_SIZE/PAGE_SIZE) * minor_count;
P
Philipp Reisner 已提交
2221 2222 2223 2224 2225 2226 2227 2228 2229
	int i;

	/* prepare our caches and mempools */
	drbd_request_mempool = NULL;
	drbd_ee_cache        = NULL;
	drbd_request_cache   = NULL;
	drbd_bm_ext_cache    = NULL;
	drbd_al_ext_cache    = NULL;
	drbd_pp_pool         = NULL;
2230
	drbd_md_io_page_pool = NULL;
2231
	drbd_md_io_bio_set   = NULL;
P
Philipp Reisner 已提交
2232 2233 2234 2235 2236 2237 2238 2239

	/* caches */
	drbd_request_cache = kmem_cache_create(
		"drbd_req", sizeof(struct drbd_request), 0, 0, NULL);
	if (drbd_request_cache == NULL)
		goto Enomem;

	drbd_ee_cache = kmem_cache_create(
2240
		"drbd_ee", sizeof(struct drbd_peer_request), 0, 0, NULL);
P
Philipp Reisner 已提交
2241 2242 2243 2244 2245 2246 2247 2248 2249 2250 2251 2252 2253 2254
	if (drbd_ee_cache == NULL)
		goto Enomem;

	drbd_bm_ext_cache = kmem_cache_create(
		"drbd_bm", sizeof(struct bm_extent), 0, 0, NULL);
	if (drbd_bm_ext_cache == NULL)
		goto Enomem;

	drbd_al_ext_cache = kmem_cache_create(
		"drbd_al", sizeof(struct lc_element), 0, 0, NULL);
	if (drbd_al_ext_cache == NULL)
		goto Enomem;

	/* mempools */
2255 2256 2257 2258
	drbd_md_io_bio_set = bioset_create(DRBD_MIN_POOL_PAGES, 0);
	if (drbd_md_io_bio_set == NULL)
		goto Enomem;

2259 2260 2261 2262
	drbd_md_io_page_pool = mempool_create_page_pool(DRBD_MIN_POOL_PAGES, 0);
	if (drbd_md_io_page_pool == NULL)
		goto Enomem;

P
Philipp Reisner 已提交
2263 2264 2265 2266 2267 2268 2269
	drbd_request_mempool = mempool_create(number,
		mempool_alloc_slab, mempool_free_slab, drbd_request_cache);
	if (drbd_request_mempool == NULL)
		goto Enomem;

	drbd_ee_mempool = mempool_create(number,
		mempool_alloc_slab, mempool_free_slab, drbd_ee_cache);
2270
	if (drbd_ee_mempool == NULL)
P
Philipp Reisner 已提交
2271 2272 2273 2274 2275 2276 2277 2278 2279 2280 2281 2282 2283 2284 2285 2286 2287 2288 2289 2290 2291 2292 2293 2294 2295 2296 2297 2298 2299 2300 2301 2302 2303 2304 2305
		goto Enomem;

	/* drbd's page pool */
	spin_lock_init(&drbd_pp_lock);

	for (i = 0; i < number; i++) {
		page = alloc_page(GFP_HIGHUSER);
		if (!page)
			goto Enomem;
		set_page_private(page, (unsigned long)drbd_pp_pool);
		drbd_pp_pool = page;
	}
	drbd_pp_vacant = number;

	return 0;

Enomem:
	drbd_destroy_mempools(); /* in case we allocated some */
	return -ENOMEM;
}

static int drbd_notify_sys(struct notifier_block *this, unsigned long code,
	void *unused)
{
	/* just so we have it.  you never know what interesting things we
	 * might want to do here some day...
	 */

	return NOTIFY_DONE;
}

static struct notifier_block drbd_notifier = {
	.notifier_call = drbd_notify_sys,
};

2306
static void drbd_release_all_peer_reqs(struct drbd_conf *mdev)
P
Philipp Reisner 已提交
2307 2308 2309
{
	int rr;

2310
	rr = drbd_free_peer_reqs(mdev, &mdev->active_ee);
P
Philipp Reisner 已提交
2311 2312 2313
	if (rr)
		dev_err(DEV, "%d EEs in active list found!\n", rr);

2314
	rr = drbd_free_peer_reqs(mdev, &mdev->sync_ee);
P
Philipp Reisner 已提交
2315 2316 2317
	if (rr)
		dev_err(DEV, "%d EEs in sync list found!\n", rr);

2318
	rr = drbd_free_peer_reqs(mdev, &mdev->read_ee);
P
Philipp Reisner 已提交
2319 2320 2321
	if (rr)
		dev_err(DEV, "%d EEs in read list found!\n", rr);

2322
	rr = drbd_free_peer_reqs(mdev, &mdev->done_ee);
P
Philipp Reisner 已提交
2323 2324 2325
	if (rr)
		dev_err(DEV, "%d EEs in done list found!\n", rr);

2326
	rr = drbd_free_peer_reqs(mdev, &mdev->net_ee);
P
Philipp Reisner 已提交
2327 2328 2329 2330
	if (rr)
		dev_err(DEV, "%d EEs in net list found!\n", rr);
}

2331
/* caution. no locking. */
2332
void drbd_minor_destroy(struct kref *kref)
P
Philipp Reisner 已提交
2333
{
2334
	struct drbd_conf *mdev = container_of(kref, struct drbd_conf, kref);
2335 2336
	struct drbd_tconn *tconn = mdev->tconn;

2337 2338
	del_timer_sync(&mdev->request_timer);

P
Philipp Reisner 已提交
2339
	/* paranoia asserts */
2340
	D_ASSERT(mdev->open_cnt == 0);
2341
	D_ASSERT(list_empty(&mdev->tconn->data.work.q));
P
Philipp Reisner 已提交
2342 2343 2344 2345 2346 2347 2348 2349
	/* end paranoia asserts */

	/* cleanup stuff that may have been allocated during
	 * device (re-)configuration or state changes */

	if (mdev->this_bdev)
		bdput(mdev->this_bdev);

2350 2351
	drbd_free_bc(mdev->ldev);
	mdev->ldev = NULL;
P
Philipp Reisner 已提交
2352

2353
	drbd_release_all_peer_reqs(mdev);
P
Philipp Reisner 已提交
2354 2355 2356 2357 2358 2359 2360

	lc_destroy(mdev->act_log);
	lc_destroy(mdev->resync);

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

2361 2362 2363 2364 2365 2366
	kfree(mdev->current_epoch);
	if (mdev->bitmap) /* should no longer be there. */
		drbd_bm_cleanup(mdev);
	__free_page(mdev->md_io_page);
	put_disk(mdev->vdisk);
	blk_cleanup_queue(mdev->rq_queue);
2367
	kfree(mdev->rs_plan_s);
2368
	kfree(mdev);
2369 2370

	kref_put(&tconn->kref, &conn_destroy);
P
Philipp Reisner 已提交
2371 2372 2373 2374 2375
}

static void drbd_cleanup(void)
{
	unsigned int i;
2376
	struct drbd_conf *mdev;
2377
	struct drbd_tconn *tconn, *tmp;
P
Philipp Reisner 已提交
2378 2379 2380

	unregister_reboot_notifier(&drbd_notifier);

2381 2382 2383 2384 2385 2386 2387 2388 2389 2390 2391
	/* first remove proc,
	 * drbdsetup uses it's presence to detect
	 * whether DRBD is loaded.
	 * If we would get stuck in proc removal,
	 * but have netlink already deregistered,
	 * some drbdsetup commands may wait forever
	 * for an answer.
	 */
	if (drbd_proc)
		remove_proc_entry("drbd", NULL);

2392
	drbd_genl_unregister();
P
Philipp Reisner 已提交
2393

2394 2395 2396 2397
	idr_for_each_entry(&minors, mdev, i) {
		idr_remove(&minors, mdev_to_minor(mdev));
		idr_remove(&mdev->tconn->volumes, mdev->vnr);
		del_gendisk(mdev->vdisk);
P
Philipp Reisner 已提交
2398
		/* synchronize_rcu(); No other threads running at this point */
2399 2400 2401
		kref_put(&mdev->kref, &drbd_minor_destroy);
	}

P
Philipp Reisner 已提交
2402
	/* not _rcu since, no other updater anymore. Genl already unregistered */
2403
	list_for_each_entry_safe(tconn, tmp, &drbd_tconns, all_tconn) {
P
Philipp Reisner 已提交
2404 2405
		list_del(&tconn->all_tconn); /* not _rcu no proc, not other threads */
		/* synchronize_rcu(); */
2406 2407
		kref_put(&tconn->kref, &conn_destroy);
	}
2408

2409
	drbd_destroy_mempools();
P
Philipp Reisner 已提交
2410 2411
	unregister_blkdev(DRBD_MAJOR, "drbd");

2412 2413
	idr_destroy(&minors);

P
Philipp Reisner 已提交
2414 2415 2416 2417 2418 2419 2420 2421 2422 2423 2424 2425 2426 2427 2428 2429 2430
	printk(KERN_INFO "drbd: module cleanup done.\n");
}

/**
 * drbd_congested() - Callback for pdflush
 * @congested_data:	User data
 * @bdi_bits:		Bits pdflush is currently interested in
 *
 * Returns 1<<BDI_async_congested and/or 1<<BDI_sync_congested if we are congested.
 */
static int drbd_congested(void *congested_data, int bdi_bits)
{
	struct drbd_conf *mdev = congested_data;
	struct request_queue *q;
	char reason = '-';
	int r = 0;

2431
	if (!may_inc_ap_bio(mdev)) {
P
Philipp Reisner 已提交
2432 2433 2434 2435 2436 2437 2438 2439 2440 2441 2442 2443 2444 2445
		/* DRBD has frozen IO */
		r = bdi_bits;
		reason = 'd';
		goto out;
	}

	if (get_ldev(mdev)) {
		q = bdev_get_queue(mdev->ldev->backing_bdev);
		r = bdi_congested(&q->backing_dev_info, bdi_bits);
		put_ldev(mdev);
		if (r)
			reason = 'b';
	}

2446
	if (bdi_bits & (1 << BDI_async_congested) && test_bit(NET_CONGESTED, &mdev->tconn->flags)) {
P
Philipp Reisner 已提交
2447 2448 2449 2450 2451 2452 2453 2454 2455
		r |= (1 << BDI_async_congested);
		reason = reason == 'b' ? 'a' : 'n';
	}

out:
	mdev->congestion_reason = reason;
	return r;
}

2456 2457 2458 2459 2460 2461 2462
static void drbd_init_workqueue(struct drbd_work_queue* wq)
{
	sema_init(&wq->s, 0);
	spin_lock_init(&wq->q_lock);
	INIT_LIST_HEAD(&wq->q);
}

2463
struct drbd_tconn *conn_get_by_name(const char *name)
2464 2465 2466
{
	struct drbd_tconn *tconn;

2467 2468 2469
	if (!name || !name[0])
		return NULL;

P
Philipp Reisner 已提交
2470
	rcu_read_lock();
2471
	list_for_each_entry_rcu(tconn, &drbd_tconns, all_tconn) {
2472 2473
		if (!strcmp(tconn->name, name)) {
			kref_get(&tconn->kref);
2474
			goto found;
2475
		}
2476 2477 2478
	}
	tconn = NULL;
found:
P
Philipp Reisner 已提交
2479
	rcu_read_unlock();
2480 2481 2482
	return tconn;
}

2483 2484 2485 2486 2487 2488 2489 2490 2491 2492 2493 2494 2495 2496 2497 2498 2499 2500 2501 2502 2503
struct drbd_tconn *conn_get_by_addrs(void *my_addr, int my_addr_len,
				     void *peer_addr, int peer_addr_len)
{
	struct drbd_tconn *tconn;

	rcu_read_lock();
	list_for_each_entry_rcu(tconn, &drbd_tconns, all_tconn) {
		if (tconn->my_addr_len == my_addr_len &&
		    tconn->peer_addr_len == peer_addr_len &&
		    !memcmp(&tconn->my_addr, my_addr, my_addr_len) &&
		    !memcmp(&tconn->peer_addr, peer_addr, peer_addr_len)) {
			kref_get(&tconn->kref);
			goto found;
		}
	}
	tconn = NULL;
found:
	rcu_read_unlock();
	return tconn;
}

2504 2505 2506 2507 2508
static int drbd_alloc_socket(struct drbd_socket *socket)
{
	socket->rbuf = (void *) __get_free_page(GFP_KERNEL);
	if (!socket->rbuf)
		return -ENOMEM;
2509 2510 2511
	socket->sbuf = (void *) __get_free_page(GFP_KERNEL);
	if (!socket->sbuf)
		return -ENOMEM;
2512 2513 2514 2515 2516
	return 0;
}

static void drbd_free_socket(struct drbd_socket *socket)
{
2517
	free_page((unsigned long) socket->sbuf);
2518 2519 2520
	free_page((unsigned long) socket->rbuf);
}

2521 2522
void conn_free_crypto(struct drbd_tconn *tconn)
{
2523 2524 2525 2526
	drbd_free_sock(tconn);

	crypto_free_hash(tconn->csums_tfm);
	crypto_free_hash(tconn->verify_tfm);
2527
	crypto_free_hash(tconn->cram_hmac_tfm);
2528
	crypto_free_hash(tconn->integrity_tfm);
2529
	crypto_free_hash(tconn->peer_integrity_tfm);
2530 2531
	kfree(tconn->int_dig_in);
	kfree(tconn->int_dig_vv);
2532 2533 2534

	tconn->csums_tfm = NULL;
	tconn->verify_tfm = NULL;
2535
	tconn->cram_hmac_tfm = NULL;
2536
	tconn->integrity_tfm = NULL;
2537
	tconn->peer_integrity_tfm = NULL;
2538 2539 2540 2541
	tconn->int_dig_in = NULL;
	tconn->int_dig_vv = NULL;
}

2542 2543 2544 2545 2546 2547 2548 2549 2550 2551 2552 2553 2554 2555 2556 2557 2558 2559 2560 2561 2562 2563 2564 2565 2566 2567 2568 2569 2570 2571 2572 2573 2574 2575 2576 2577 2578 2579 2580
int set_resource_options(struct drbd_tconn *tconn, struct res_opts *res_opts)
{
	cpumask_var_t new_cpu_mask;
	int err;

	if (!zalloc_cpumask_var(&new_cpu_mask, GFP_KERNEL))
		return -ENOMEM;
		/*
		retcode = ERR_NOMEM;
		drbd_msg_put_info("unable to allocate cpumask");
		*/

	/* silently ignore cpu mask on UP kernel */
	if (nr_cpu_ids > 1 && res_opts->cpu_mask[0] != 0) {
		/* FIXME: Get rid of constant 32 here */
		err = __bitmap_parse(res_opts->cpu_mask, 32, 0,
				cpumask_bits(new_cpu_mask), nr_cpu_ids);
		if (err) {
			conn_warn(tconn, "__bitmap_parse() failed with %d\n", err);
			/* retcode = ERR_CPU_MASK_PARSE; */
			goto fail;
		}
	}
	tconn->res_opts = *res_opts;
	if (!cpumask_equal(tconn->cpu_mask, new_cpu_mask)) {
		cpumask_copy(tconn->cpu_mask, new_cpu_mask);
		drbd_calc_cpu_mask(tconn);
		tconn->receiver.reset_cpu_mask = 1;
		tconn->asender.reset_cpu_mask = 1;
		tconn->worker.reset_cpu_mask = 1;
	}
	err = 0;

fail:
	free_cpumask_var(new_cpu_mask);
	return err;

}

2581
/* caller must be under genl_lock() */
2582
struct drbd_tconn *conn_create(const char *name, struct res_opts *res_opts)
2583 2584 2585 2586 2587 2588 2589 2590 2591 2592 2593
{
	struct drbd_tconn *tconn;

	tconn = kzalloc(sizeof(struct drbd_tconn), GFP_KERNEL);
	if (!tconn)
		return NULL;

	tconn->name = kstrdup(name, GFP_KERNEL);
	if (!tconn->name)
		goto fail;

2594 2595 2596 2597 2598
	if (drbd_alloc_socket(&tconn->data))
		goto fail;
	if (drbd_alloc_socket(&tconn->meta))
		goto fail;

2599 2600 2601
	if (!zalloc_cpumask_var(&tconn->cpu_mask, GFP_KERNEL))
		goto fail;

2602 2603 2604
	if (set_resource_options(tconn, res_opts))
		goto fail;

2605 2606 2607
	if (!tl_init(tconn))
		goto fail;

2608
	tconn->cstate = C_STANDALONE;
2609
	mutex_init(&tconn->cstate_mutex);
2610
	spin_lock_init(&tconn->req_lock);
2611
	mutex_init(&tconn->conf_update);
2612
	init_waitqueue_head(&tconn->ping_wait);
2613
	idr_init(&tconn->volumes);
2614

2615 2616 2617 2618 2619 2620
	drbd_init_workqueue(&tconn->data.work);
	mutex_init(&tconn->data.mutex);

	drbd_init_workqueue(&tconn->meta.work);
	mutex_init(&tconn->meta.mutex);

2621 2622 2623 2624
	drbd_thread_init(tconn, &tconn->receiver, drbdd_init, "receiver");
	drbd_thread_init(tconn, &tconn->worker, drbd_worker, "worker");
	drbd_thread_init(tconn, &tconn->asender, drbd_asender, "asender");

2625
	kref_init(&tconn->kref);
2626
	list_add_tail_rcu(&tconn->all_tconn, &drbd_tconns);
2627 2628 2629 2630

	return tconn;

fail:
2631
	tl_cleanup(tconn);
2632
	free_cpumask_var(tconn->cpu_mask);
2633 2634
	drbd_free_socket(&tconn->meta);
	drbd_free_socket(&tconn->data);
2635 2636 2637 2638 2639 2640
	kfree(tconn->name);
	kfree(tconn);

	return NULL;
}

2641
void conn_destroy(struct kref *kref)
2642
{
2643 2644
	struct drbd_tconn *tconn = container_of(kref, struct drbd_tconn, kref);

2645
	idr_destroy(&tconn->volumes);
2646

2647
	free_cpumask_var(tconn->cpu_mask);
2648 2649
	drbd_free_socket(&tconn->meta);
	drbd_free_socket(&tconn->data);
2650
	kfree(tconn->name);
2651 2652
	kfree(tconn->int_dig_in);
	kfree(tconn->int_dig_vv);
2653 2654 2655
	kfree(tconn);
}

2656
enum drbd_ret_code conn_new_minor(struct drbd_tconn *tconn, unsigned int minor, int vnr)
P
Philipp Reisner 已提交
2657 2658 2659 2660
{
	struct drbd_conf *mdev;
	struct gendisk *disk;
	struct request_queue *q;
2661
	int vnr_got = vnr;
2662
	int minor_got = minor;
2663
	enum drbd_ret_code err = ERR_NOMEM;
2664 2665 2666 2667

	mdev = minor_to_mdev(minor);
	if (mdev)
		return ERR_MINOR_EXISTS;
P
Philipp Reisner 已提交
2668 2669 2670 2671

	/* GFP_KERNEL, we are outside of all write-out paths */
	mdev = kzalloc(sizeof(struct drbd_conf), GFP_KERNEL);
	if (!mdev)
2672 2673
		return ERR_NOMEM;

2674
	kref_get(&tconn->kref);
2675
	mdev->tconn = tconn;
2676

P
Philipp Reisner 已提交
2677
	mdev->minor = minor;
2678
	mdev->vnr = vnr;
P
Philipp Reisner 已提交
2679 2680 2681 2682 2683 2684 2685 2686 2687 2688 2689 2690 2691 2692

	drbd_init_set_defaults(mdev);

	q = blk_alloc_queue(GFP_KERNEL);
	if (!q)
		goto out_no_q;
	mdev->rq_queue = q;
	q->queuedata   = mdev;

	disk = alloc_disk(1);
	if (!disk)
		goto out_no_disk;
	mdev->vdisk = disk;

2693
	set_disk_ro(disk, true);
P
Philipp Reisner 已提交
2694 2695 2696 2697 2698 2699 2700 2701 2702 2703 2704 2705 2706 2707 2708

	disk->queue = q;
	disk->major = DRBD_MAJOR;
	disk->first_minor = minor;
	disk->fops = &drbd_ops;
	sprintf(disk->disk_name, "drbd%d", minor);
	disk->private_data = mdev;

	mdev->this_bdev = bdget(MKDEV(DRBD_MAJOR, minor));
	/* we have no partitions. we contain only ourselves. */
	mdev->this_bdev->bd_contains = mdev->this_bdev;

	q->backing_dev_info.congested_fn = drbd_congested;
	q->backing_dev_info.congested_data = mdev;

2709
	blk_queue_make_request(q, drbd_make_request);
2710 2711 2712
	/* Setting the max_hw_sectors to an odd value of 8kibyte here
	   This triggers a max_bio_size message upon first attach or connect */
	blk_queue_max_hw_sectors(q, DRBD_MAX_BIO_SIZE_SAFE >> 8);
P
Philipp Reisner 已提交
2713 2714
	blk_queue_bounce_limit(q, BLK_BOUNCE_ANY);
	blk_queue_merge_bvec(q, drbd_merge_bvec);
2715
	q->queue_lock = &mdev->tconn->req_lock; /* needed since we use */
P
Philipp Reisner 已提交
2716 2717 2718 2719 2720 2721 2722

	mdev->md_io_page = alloc_page(GFP_KERNEL);
	if (!mdev->md_io_page)
		goto out_no_io_page;

	if (drbd_bm_init(mdev))
		goto out_no_bitmap;
2723
	mdev->read_requests = RB_ROOT;
2724
	mdev->write_requests = RB_ROOT;
P
Philipp Reisner 已提交
2725 2726 2727 2728 2729 2730 2731 2732

	mdev->current_epoch = kzalloc(sizeof(struct drbd_epoch), GFP_KERNEL);
	if (!mdev->current_epoch)
		goto out_no_epoch;

	INIT_LIST_HEAD(&mdev->current_epoch->list);
	mdev->epochs = 1;

2733
	if (!idr_pre_get(&minors, GFP_KERNEL))
2734 2735 2736
		goto out_no_minor_idr;
	if (idr_get_new_above(&minors, mdev, minor, &minor_got))
		goto out_no_minor_idr;
2737
	if (minor_got != minor) {
2738 2739
		err = ERR_MINOR_EXISTS;
		drbd_msg_put_info("requested minor exists already");
2740
		goto out_idr_remove_minor;
2741
	}
2742 2743 2744 2745 2746 2747 2748 2749 2750 2751

	if (!idr_pre_get(&tconn->volumes, GFP_KERNEL))
		goto out_idr_remove_minor;
	if (idr_get_new_above(&tconn->volumes, mdev, vnr, &vnr_got))
		goto out_idr_remove_minor;
	if (vnr_got != vnr) {
		err = ERR_INVALID_REQUEST;
		drbd_msg_put_info("requested volume exists already");
		goto out_idr_remove_vol;
	}
2752
	add_disk(disk);
2753
	kref_init(&mdev->kref); /* one ref for both idrs and the the add_disk */
2754

2755 2756 2757
	/* inherit the connection state */
	mdev->state.conn = tconn->cstate;
	if (mdev->state.conn == C_WF_REPORT_PARAMS)
P
Philipp Reisner 已提交
2758
		drbd_connected(mdev);
2759

2760
	return NO_ERROR;
P
Philipp Reisner 已提交
2761

2762 2763
out_idr_remove_vol:
	idr_remove(&tconn->volumes, vnr_got);
2764 2765
out_idr_remove_minor:
	idr_remove(&minors, minor_got);
2766
	synchronize_rcu();
2767
out_no_minor_idr:
2768
	kfree(mdev->current_epoch);
P
Philipp Reisner 已提交
2769 2770 2771 2772 2773 2774 2775 2776 2777 2778
out_no_epoch:
	drbd_bm_cleanup(mdev);
out_no_bitmap:
	__free_page(mdev->md_io_page);
out_no_io_page:
	put_disk(disk);
out_no_disk:
	blk_cleanup_queue(q);
out_no_q:
	kfree(mdev);
2779
	kref_put(&tconn->kref, &conn_destroy);
2780
	return err;
P
Philipp Reisner 已提交
2781 2782 2783 2784 2785 2786
}

int __init drbd_init(void)
{
	int err;

2787
	if (minor_count < DRBD_MINOR_COUNT_MIN || minor_count > DRBD_MINOR_COUNT_MAX) {
P
Philipp Reisner 已提交
2788
		printk(KERN_ERR
2789
		       "drbd: invalid minor_count (%d)\n", minor_count);
P
Philipp Reisner 已提交
2790 2791 2792
#ifdef MODULE
		return -EINVAL;
#else
2793
		minor_count = DRBD_MINOR_COUNT_DEF;
P
Philipp Reisner 已提交
2794 2795 2796 2797 2798 2799 2800 2801 2802 2803 2804
#endif
	}

	err = register_blkdev(DRBD_MAJOR, "drbd");
	if (err) {
		printk(KERN_ERR
		       "drbd: unable to register block device major %d\n",
		       DRBD_MAJOR);
		return err;
	}

2805 2806 2807 2808 2809 2810 2811
	err = drbd_genl_register();
	if (err) {
		printk(KERN_ERR "drbd: unable to register generic netlink family\n");
		goto fail;
	}


P
Philipp Reisner 已提交
2812 2813 2814 2815 2816 2817 2818 2819 2820 2821
	register_reboot_notifier(&drbd_notifier);

	/*
	 * allocate all necessary structs
	 */
	err = -ENOMEM;

	init_waitqueue_head(&drbd_pp_wait);

	drbd_proc = NULL; /* play safe for drbd_cleanup */
2822
	idr_init(&minors);
P
Philipp Reisner 已提交
2823 2824 2825

	err = drbd_create_mempools();
	if (err)
2826
		goto fail;
P
Philipp Reisner 已提交
2827

2828
	drbd_proc = proc_create_data("drbd", S_IFREG | S_IRUGO , NULL, &drbd_proc_fops, NULL);
P
Philipp Reisner 已提交
2829 2830
	if (!drbd_proc)	{
		printk(KERN_ERR "drbd: unable to register proc file\n");
2831
		goto fail;
P
Philipp Reisner 已提交
2832 2833 2834
	}

	rwlock_init(&global_state_lock);
2835
	INIT_LIST_HEAD(&drbd_tconns);
P
Philipp Reisner 已提交
2836 2837 2838 2839 2840 2841 2842 2843 2844 2845

	printk(KERN_INFO "drbd: initialized. "
	       "Version: " REL_VERSION " (api:%d/proto:%d-%d)\n",
	       API_VERSION, PRO_VERSION_MIN, PRO_VERSION_MAX);
	printk(KERN_INFO "drbd: %s\n", drbd_buildtag());
	printk(KERN_INFO "drbd: registered as block device major %d\n",
		DRBD_MAJOR);

	return 0; /* Success! */

2846
fail:
P
Philipp Reisner 已提交
2847 2848 2849 2850 2851 2852 2853 2854 2855 2856 2857 2858 2859 2860
	drbd_cleanup();
	if (err == -ENOMEM)
		/* currently always the case */
		printk(KERN_ERR "drbd: ran out of memory\n");
	else
		printk(KERN_ERR "drbd: initialization failure\n");
	return err;
}

void drbd_free_bc(struct drbd_backing_dev *ldev)
{
	if (ldev == NULL)
		return;

2861 2862
	blkdev_put(ldev->backing_bdev, FMODE_READ | FMODE_WRITE | FMODE_EXCL);
	blkdev_put(ldev->md_bdev, FMODE_READ | FMODE_WRITE | FMODE_EXCL);
P
Philipp Reisner 已提交
2863 2864 2865 2866

	kfree(ldev);
}

2867 2868 2869 2870 2871 2872 2873 2874
void drbd_free_sock(struct drbd_tconn *tconn)
{
	if (tconn->data.socket) {
		mutex_lock(&tconn->data.mutex);
		kernel_sock_shutdown(tconn->data.socket, SHUT_RDWR);
		sock_release(tconn->data.socket);
		tconn->data.socket = NULL;
		mutex_unlock(&tconn->data.mutex);
P
Philipp Reisner 已提交
2875
	}
2876 2877 2878 2879 2880 2881
	if (tconn->meta.socket) {
		mutex_lock(&tconn->meta.mutex);
		kernel_sock_shutdown(tconn->meta.socket, SHUT_RDWR);
		sock_release(tconn->meta.socket);
		tconn->meta.socket = NULL;
		mutex_unlock(&tconn->meta.mutex);
P
Philipp Reisner 已提交
2882 2883 2884 2885 2886 2887 2888 2889 2890 2891 2892 2893 2894 2895 2896
	}
}

/* meta data management */

struct meta_data_on_disk {
	u64 la_size;           /* last agreed size. */
	u64 uuid[UI_SIZE];   /* UUIDs. */
	u64 device_uuid;
	u64 reserved_u64_1;
	u32 flags;             /* MDF */
	u32 magic;
	u32 md_size_sect;
	u32 al_offset;         /* offset to this block */
	u32 al_nr_extents;     /* important for restoring the AL */
2897
	      /* `-- act_log->nr_elements <-- ldev->dc.al_extents */
P
Philipp Reisner 已提交
2898 2899
	u32 bm_offset;         /* offset to the bitmap, from here */
	u32 bm_bytes_per_bit;  /* BM_BLOCK_SIZE */
2900 2901
	u32 la_peer_max_bio_size;   /* last peer max_bio_size */
	u32 reserved_u32[3];
P
Philipp Reisner 已提交
2902 2903 2904 2905 2906 2907 2908 2909 2910 2911 2912 2913 2914

} __packed;

/**
 * drbd_md_sync() - Writes the meta data super block if the MD_DIRTY flag bit is set
 * @mdev:	DRBD device.
 */
void drbd_md_sync(struct drbd_conf *mdev)
{
	struct meta_data_on_disk *buffer;
	sector_t sector;
	int i;

2915 2916
	del_timer(&mdev->md_sync_timer);
	/* timer may be rearmed by drbd_md_mark_dirty() now. */
P
Philipp Reisner 已提交
2917 2918 2919 2920 2921 2922 2923 2924
	if (!test_and_clear_bit(MD_DIRTY, &mdev->flags))
		return;

	/* We use here D_FAILED and not D_ATTACHING because we try to write
	 * metadata even if we detach due to a disk failure! */
	if (!get_ldev_if_state(mdev, D_FAILED))
		return;

2925 2926 2927 2928
	buffer = drbd_md_get_buffer(mdev);
	if (!buffer)
		goto out;

P
Philipp Reisner 已提交
2929 2930 2931 2932 2933 2934 2935 2936 2937 2938 2939 2940 2941 2942 2943
	memset(buffer, 0, 512);

	buffer->la_size = cpu_to_be64(drbd_get_capacity(mdev->this_bdev));
	for (i = UI_CURRENT; i < UI_SIZE; i++)
		buffer->uuid[i] = cpu_to_be64(mdev->ldev->md.uuid[i]);
	buffer->flags = cpu_to_be32(mdev->ldev->md.flags);
	buffer->magic = cpu_to_be32(DRBD_MD_MAGIC);

	buffer->md_size_sect  = cpu_to_be32(mdev->ldev->md.md_size_sect);
	buffer->al_offset     = cpu_to_be32(mdev->ldev->md.al_offset);
	buffer->al_nr_extents = cpu_to_be32(mdev->act_log->nr_elements);
	buffer->bm_bytes_per_bit = cpu_to_be32(BM_BLOCK_SIZE);
	buffer->device_uuid = cpu_to_be64(mdev->ldev->md.device_uuid);

	buffer->bm_offset = cpu_to_be32(mdev->ldev->md.bm_offset);
2944
	buffer->la_peer_max_bio_size = cpu_to_be32(mdev->peer_max_bio_size);
P
Philipp Reisner 已提交
2945 2946 2947 2948

	D_ASSERT(drbd_md_ss__(mdev, mdev->ldev) == mdev->ldev->md.md_offset);
	sector = mdev->ldev->md.md_offset;

2949
	if (drbd_md_sync_page_io(mdev, mdev->ldev, sector, WRITE)) {
P
Philipp Reisner 已提交
2950 2951
		/* this was a try anyways ... */
		dev_err(DEV, "meta data update failed!\n");
2952
		drbd_chk_io_error(mdev, 1, true);
P
Philipp Reisner 已提交
2953 2954 2955 2956 2957 2958
	}

	/* Update mdev->ldev->md.la_size_sect,
	 * since we updated it on metadata. */
	mdev->ldev->md.la_size_sect = drbd_get_capacity(mdev->this_bdev);

2959 2960
	drbd_md_put_buffer(mdev);
out:
P
Philipp Reisner 已提交
2961 2962 2963 2964 2965 2966 2967 2968
	put_ldev(mdev);
}

/**
 * drbd_md_read() - Reads in the meta data super block
 * @mdev:	DRBD device.
 * @bdev:	Device from which the meta data should be read in.
 *
2969
 * Return 0 (NO_ERROR) on success, and an enum drbd_ret_code in case
P
Philipp Reisner 已提交
2970 2971 2972 2973 2974 2975 2976 2977 2978 2979
 * something goes wrong.  Currently only: ERR_IO_MD_DISK, ERR_MD_INVALID.
 */
int drbd_md_read(struct drbd_conf *mdev, struct drbd_backing_dev *bdev)
{
	struct meta_data_on_disk *buffer;
	int i, rv = NO_ERROR;

	if (!get_ldev_if_state(mdev, D_ATTACHING))
		return ERR_IO_MD_DISK;

2980 2981 2982
	buffer = drbd_md_get_buffer(mdev);
	if (!buffer)
		goto out;
P
Philipp Reisner 已提交
2983

2984
	if (drbd_md_sync_page_io(mdev, bdev, bdev->md.md_offset, READ)) {
L
Lucas De Marchi 已提交
2985
		/* NOTE: can't do normal error processing here as this is
P
Philipp Reisner 已提交
2986 2987 2988 2989 2990 2991
		   called BEFORE disk is attached */
		dev_err(DEV, "Error while reading metadata.\n");
		rv = ERR_IO_MD_DISK;
		goto err;
	}

2992
	if (buffer->magic != cpu_to_be32(DRBD_MD_MAGIC)) {
P
Philipp Reisner 已提交
2993 2994 2995 2996 2997 2998 2999 3000 3001 3002 3003 3004 3005 3006 3007 3008 3009 3010 3011 3012 3013 3014 3015 3016 3017 3018 3019 3020 3021 3022 3023 3024 3025 3026 3027 3028
		dev_err(DEV, "Error while reading metadata, magic not found.\n");
		rv = ERR_MD_INVALID;
		goto err;
	}
	if (be32_to_cpu(buffer->al_offset) != bdev->md.al_offset) {
		dev_err(DEV, "unexpected al_offset: %d (expected %d)\n",
		    be32_to_cpu(buffer->al_offset), bdev->md.al_offset);
		rv = ERR_MD_INVALID;
		goto err;
	}
	if (be32_to_cpu(buffer->bm_offset) != bdev->md.bm_offset) {
		dev_err(DEV, "unexpected bm_offset: %d (expected %d)\n",
		    be32_to_cpu(buffer->bm_offset), bdev->md.bm_offset);
		rv = ERR_MD_INVALID;
		goto err;
	}
	if (be32_to_cpu(buffer->md_size_sect) != bdev->md.md_size_sect) {
		dev_err(DEV, "unexpected md_size: %u (expected %u)\n",
		    be32_to_cpu(buffer->md_size_sect), bdev->md.md_size_sect);
		rv = ERR_MD_INVALID;
		goto err;
	}

	if (be32_to_cpu(buffer->bm_bytes_per_bit) != BM_BLOCK_SIZE) {
		dev_err(DEV, "unexpected bm_bytes_per_bit: %u (expected %u)\n",
		    be32_to_cpu(buffer->bm_bytes_per_bit), BM_BLOCK_SIZE);
		rv = ERR_MD_INVALID;
		goto err;
	}

	bdev->md.la_size_sect = be64_to_cpu(buffer->la_size);
	for (i = UI_CURRENT; i < UI_SIZE; i++)
		bdev->md.uuid[i] = be64_to_cpu(buffer->uuid[i]);
	bdev->md.flags = be32_to_cpu(buffer->flags);
	bdev->md.device_uuid = be64_to_cpu(buffer->device_uuid);

3029
	spin_lock_irq(&mdev->tconn->req_lock);
3030 3031 3032 3033 3034 3035
	if (mdev->state.conn < C_CONNECTED) {
		int peer;
		peer = be32_to_cpu(buffer->la_peer_max_bio_size);
		peer = max_t(int, peer, DRBD_MAX_BIO_SIZE_SAFE);
		mdev->peer_max_bio_size = peer;
	}
3036
	spin_unlock_irq(&mdev->tconn->req_lock);
3037

P
Philipp Reisner 已提交
3038 3039 3040 3041
	/* This blocks wants to be get removed... */
	bdev->disk_conf->al_extents = be32_to_cpu(buffer->al_nr_extents);
	if (bdev->disk_conf->al_extents < DRBD_AL_EXTENTS_MIN)
		bdev->disk_conf->al_extents = DRBD_AL_EXTENTS_DEF;
P
Philipp Reisner 已提交
3042 3043

 err:
3044 3045
	drbd_md_put_buffer(mdev);
 out:
P
Philipp Reisner 已提交
3046 3047 3048 3049 3050 3051 3052 3053 3054 3055 3056 3057 3058
	put_ldev(mdev);

	return rv;
}

/**
 * drbd_md_mark_dirty() - Mark meta data super block as dirty
 * @mdev:	DRBD device.
 *
 * Call this function if you change anything that should be written to
 * the meta-data super block. This function sets MD_DIRTY, and starts a
 * timer that ensures that within five seconds you have to call drbd_md_sync().
 */
3059
#ifdef DEBUG
3060 3061 3062 3063 3064 3065 3066 3067 3068
void drbd_md_mark_dirty_(struct drbd_conf *mdev, unsigned int line, const char *func)
{
	if (!test_and_set_bit(MD_DIRTY, &mdev->flags)) {
		mod_timer(&mdev->md_sync_timer, jiffies + HZ);
		mdev->last_md_mark_dirty.line = line;
		mdev->last_md_mark_dirty.func = func;
	}
}
#else
P
Philipp Reisner 已提交
3069 3070
void drbd_md_mark_dirty(struct drbd_conf *mdev)
{
3071
	if (!test_and_set_bit(MD_DIRTY, &mdev->flags))
3072
		mod_timer(&mdev->md_sync_timer, jiffies + 5*HZ);
P
Philipp Reisner 已提交
3073
}
3074
#endif
P
Philipp Reisner 已提交
3075 3076 3077 3078 3079

static void drbd_uuid_move_history(struct drbd_conf *mdev) __must_hold(local)
{
	int i;

3080
	for (i = UI_HISTORY_START; i < UI_HISTORY_END; i++)
P
Philipp Reisner 已提交
3081 3082 3083 3084 3085 3086 3087 3088 3089 3090 3091 3092 3093 3094 3095 3096 3097 3098 3099 3100 3101 3102 3103 3104 3105 3106 3107 3108 3109 3110 3111 3112 3113 3114 3115 3116 3117 3118
		mdev->ldev->md.uuid[i+1] = mdev->ldev->md.uuid[i];
}

void _drbd_uuid_set(struct drbd_conf *mdev, int idx, u64 val) __must_hold(local)
{
	if (idx == UI_CURRENT) {
		if (mdev->state.role == R_PRIMARY)
			val |= 1;
		else
			val &= ~((u64)1);

		drbd_set_ed_uuid(mdev, val);
	}

	mdev->ldev->md.uuid[idx] = val;
	drbd_md_mark_dirty(mdev);
}


void drbd_uuid_set(struct drbd_conf *mdev, int idx, u64 val) __must_hold(local)
{
	if (mdev->ldev->md.uuid[idx]) {
		drbd_uuid_move_history(mdev);
		mdev->ldev->md.uuid[UI_HISTORY_START] = mdev->ldev->md.uuid[idx];
	}
	_drbd_uuid_set(mdev, idx, val);
}

/**
 * drbd_uuid_new_current() - Creates a new current UUID
 * @mdev:	DRBD device.
 *
 * Creates a new current UUID, and rotates the old current UUID into
 * the bitmap slot. Causes an incremental resync upon next connect.
 */
void drbd_uuid_new_current(struct drbd_conf *mdev) __must_hold(local)
{
	u64 val;
3119 3120 3121 3122
	unsigned long long bm_uuid = mdev->ldev->md.uuid[UI_BITMAP];

	if (bm_uuid)
		dev_warn(DEV, "bm UUID was already set: %llX\n", bm_uuid);
P
Philipp Reisner 已提交
3123 3124 3125 3126 3127

	mdev->ldev->md.uuid[UI_BITMAP] = mdev->ldev->md.uuid[UI_CURRENT];

	get_random_bytes(&val, sizeof(u64));
	_drbd_uuid_set(mdev, UI_CURRENT, val);
3128
	drbd_print_uuids(mdev, "new current UUID");
3129 3130
	/* get it to stable storage _now_ */
	drbd_md_sync(mdev);
P
Philipp Reisner 已提交
3131 3132 3133 3134 3135 3136 3137 3138 3139 3140 3141 3142
}

void drbd_uuid_set_bm(struct drbd_conf *mdev, u64 val) __must_hold(local)
{
	if (mdev->ldev->md.uuid[UI_BITMAP] == 0 && val == 0)
		return;

	if (val == 0) {
		drbd_uuid_move_history(mdev);
		mdev->ldev->md.uuid[UI_HISTORY_START] = mdev->ldev->md.uuid[UI_BITMAP];
		mdev->ldev->md.uuid[UI_BITMAP] = 0;
	} else {
3143 3144 3145
		unsigned long long bm_uuid = mdev->ldev->md.uuid[UI_BITMAP];
		if (bm_uuid)
			dev_warn(DEV, "bm UUID was already set: %llX\n", bm_uuid);
P
Philipp Reisner 已提交
3146

3147
		mdev->ldev->md.uuid[UI_BITMAP] = val & ~((u64)1);
P
Philipp Reisner 已提交
3148 3149 3150 3151 3152 3153 3154 3155 3156 3157 3158 3159 3160 3161 3162 3163 3164 3165 3166 3167 3168 3169 3170 3171 3172 3173 3174 3175 3176 3177 3178 3179 3180 3181 3182 3183 3184 3185 3186 3187 3188 3189
	}
	drbd_md_mark_dirty(mdev);
}

/**
 * drbd_bmio_set_n_write() - io_fn for drbd_queue_bitmap_io() or drbd_bitmap_io()
 * @mdev:	DRBD device.
 *
 * Sets all bits in the bitmap and writes the whole bitmap to stable storage.
 */
int drbd_bmio_set_n_write(struct drbd_conf *mdev)
{
	int rv = -EIO;

	if (get_ldev_if_state(mdev, D_ATTACHING)) {
		drbd_md_set_flag(mdev, MDF_FULL_SYNC);
		drbd_md_sync(mdev);
		drbd_bm_set_all(mdev);

		rv = drbd_bm_write(mdev);

		if (!rv) {
			drbd_md_clear_flag(mdev, MDF_FULL_SYNC);
			drbd_md_sync(mdev);
		}

		put_ldev(mdev);
	}

	return rv;
}

/**
 * drbd_bmio_clear_n_write() - io_fn for drbd_queue_bitmap_io() or drbd_bitmap_io()
 * @mdev:	DRBD device.
 *
 * Clears all bits in the bitmap and writes the whole bitmap to stable storage.
 */
int drbd_bmio_clear_n_write(struct drbd_conf *mdev)
{
	int rv = -EIO;

3190
	drbd_resume_al(mdev);
P
Philipp Reisner 已提交
3191 3192 3193 3194 3195 3196 3197 3198 3199
	if (get_ldev_if_state(mdev, D_ATTACHING)) {
		drbd_bm_clear_all(mdev);
		rv = drbd_bm_write(mdev);
		put_ldev(mdev);
	}

	return rv;
}

3200
static int w_bitmap_io(struct drbd_work *w, int unused)
P
Philipp Reisner 已提交
3201 3202
{
	struct bm_io_work *work = container_of(w, struct bm_io_work, w);
3203
	struct drbd_conf *mdev = w->mdev;
3204
	int rv = -EIO;
P
Philipp Reisner 已提交
3205 3206 3207

	D_ASSERT(atomic_read(&mdev->ap_bio_cnt) == 0);

3208
	if (get_ldev(mdev)) {
3209
		drbd_bm_lock(mdev, work->why, work->flags);
3210 3211 3212 3213
		rv = work->io_fn(mdev);
		drbd_bm_unlock(mdev);
		put_ldev(mdev);
	}
P
Philipp Reisner 已提交
3214

3215
	clear_bit_unlock(BITMAP_IO, &mdev->flags);
P
Philipp Reisner 已提交
3216 3217 3218 3219 3220 3221 3222
	wake_up(&mdev->misc_wait);

	if (work->done)
		work->done(mdev, rv);

	clear_bit(BITMAP_IO_QUEUED, &mdev->flags);
	work->why = NULL;
3223
	work->flags = 0;
P
Philipp Reisner 已提交
3224

3225
	return 0;
P
Philipp Reisner 已提交
3226 3227
}

3228 3229 3230 3231 3232 3233 3234 3235 3236 3237 3238 3239 3240
void drbd_ldev_destroy(struct drbd_conf *mdev)
{
	lc_destroy(mdev->resync);
	mdev->resync = NULL;
	lc_destroy(mdev->act_log);
	mdev->act_log = NULL;
	__no_warn(local,
		drbd_free_bc(mdev->ldev);
		mdev->ldev = NULL;);

	clear_bit(GO_DISKLESS, &mdev->flags);
}

3241
static int w_go_diskless(struct drbd_work *w, int unused)
3242
{
3243 3244
	struct drbd_conf *mdev = w->mdev;

3245
	D_ASSERT(mdev->state.disk == D_FAILED);
3246 3247
	/* we cannot assert local_cnt == 0 here, as get_ldev_if_state will
	 * inc/dec it frequently. Once we are D_DISKLESS, no one will touch
3248 3249
	 * the protected members anymore, though, so once put_ldev reaches zero
	 * again, it will be safe to free them. */
3250
	drbd_force_state(mdev, NS(disk, D_DISKLESS));
3251
	return 0;
3252 3253 3254 3255 3256 3257
}

void drbd_go_diskless(struct drbd_conf *mdev)
{
	D_ASSERT(mdev->state.disk == D_FAILED);
	if (!test_and_set_bit(GO_DISKLESS, &mdev->flags))
3258
		drbd_queue_work(&mdev->tconn->data.work, &mdev->go_diskless);
3259 3260
}

P
Philipp Reisner 已提交
3261 3262 3263 3264 3265 3266 3267 3268 3269 3270 3271 3272 3273 3274 3275
/**
 * drbd_queue_bitmap_io() - Queues an IO operation on the whole bitmap
 * @mdev:	DRBD device.
 * @io_fn:	IO callback to be called when bitmap IO is possible
 * @done:	callback to be called after the bitmap IO was performed
 * @why:	Descriptive text of the reason for doing the IO
 *
 * While IO on the bitmap happens we freeze application IO thus we ensure
 * that drbd_set_out_of_sync() can not be called. This function MAY ONLY be
 * called from worker context. It MUST NOT be used while a previous such
 * work is still pending!
 */
void drbd_queue_bitmap_io(struct drbd_conf *mdev,
			  int (*io_fn)(struct drbd_conf *),
			  void (*done)(struct drbd_conf *, int),
3276
			  char *why, enum bm_flag flags)
P
Philipp Reisner 已提交
3277
{
3278
	D_ASSERT(current == mdev->tconn->worker.task);
P
Philipp Reisner 已提交
3279 3280 3281 3282 3283 3284 3285 3286 3287 3288 3289

	D_ASSERT(!test_bit(BITMAP_IO_QUEUED, &mdev->flags));
	D_ASSERT(!test_bit(BITMAP_IO, &mdev->flags));
	D_ASSERT(list_empty(&mdev->bm_io_work.w.list));
	if (mdev->bm_io_work.why)
		dev_err(DEV, "FIXME going to queue '%s' but '%s' still pending?\n",
			why, mdev->bm_io_work.why);

	mdev->bm_io_work.io_fn = io_fn;
	mdev->bm_io_work.done = done;
	mdev->bm_io_work.why = why;
3290
	mdev->bm_io_work.flags = flags;
P
Philipp Reisner 已提交
3291

3292
	spin_lock_irq(&mdev->tconn->req_lock);
P
Philipp Reisner 已提交
3293 3294
	set_bit(BITMAP_IO, &mdev->flags);
	if (atomic_read(&mdev->ap_bio_cnt) == 0) {
3295
		if (!test_and_set_bit(BITMAP_IO_QUEUED, &mdev->flags))
3296
			drbd_queue_work(&mdev->tconn->data.work, &mdev->bm_io_work.w);
P
Philipp Reisner 已提交
3297
	}
3298
	spin_unlock_irq(&mdev->tconn->req_lock);
P
Philipp Reisner 已提交
3299 3300 3301 3302 3303 3304 3305 3306 3307 3308 3309
}

/**
 * drbd_bitmap_io() -  Does an IO operation on the whole bitmap
 * @mdev:	DRBD device.
 * @io_fn:	IO callback to be called when bitmap IO is possible
 * @why:	Descriptive text of the reason for doing the IO
 *
 * freezes application IO while that the actual IO operations runs. This
 * functions MAY NOT be called from worker context.
 */
3310 3311
int drbd_bitmap_io(struct drbd_conf *mdev, int (*io_fn)(struct drbd_conf *),
		char *why, enum bm_flag flags)
P
Philipp Reisner 已提交
3312 3313 3314
{
	int rv;

3315
	D_ASSERT(current != mdev->tconn->worker.task);
P
Philipp Reisner 已提交
3316

3317 3318
	if ((flags & BM_LOCKED_SET_ALLOWED) == 0)
		drbd_suspend_io(mdev);
P
Philipp Reisner 已提交
3319

3320
	drbd_bm_lock(mdev, why, flags);
P
Philipp Reisner 已提交
3321 3322 3323
	rv = io_fn(mdev);
	drbd_bm_unlock(mdev);

3324 3325
	if ((flags & BM_LOCKED_SET_ALLOWED) == 0)
		drbd_resume_io(mdev);
P
Philipp Reisner 已提交
3326 3327 3328 3329 3330 3331 3332 3333 3334 3335 3336 3337 3338 3339 3340 3341 3342 3343 3344 3345 3346 3347 3348 3349 3350 3351 3352 3353

	return rv;
}

void drbd_md_set_flag(struct drbd_conf *mdev, int flag) __must_hold(local)
{
	if ((mdev->ldev->md.flags & flag) != flag) {
		drbd_md_mark_dirty(mdev);
		mdev->ldev->md.flags |= flag;
	}
}

void drbd_md_clear_flag(struct drbd_conf *mdev, int flag) __must_hold(local)
{
	if ((mdev->ldev->md.flags & flag) != 0) {
		drbd_md_mark_dirty(mdev);
		mdev->ldev->md.flags &= ~flag;
	}
}
int drbd_md_test_flag(struct drbd_backing_dev *bdev, int flag)
{
	return (bdev->md.flags & flag) != 0;
}

static void md_sync_timer_fn(unsigned long data)
{
	struct drbd_conf *mdev = (struct drbd_conf *) data;

3354
	drbd_queue_work_front(&mdev->tconn->data.work, &mdev->md_sync_work);
P
Philipp Reisner 已提交
3355 3356
}

3357
static int w_md_sync(struct drbd_work *w, int unused)
P
Philipp Reisner 已提交
3358
{
3359 3360
	struct drbd_conf *mdev = w->mdev;

P
Philipp Reisner 已提交
3361
	dev_warn(DEV, "md_sync_timer expired! Worker calls drbd_md_sync().\n");
3362 3363 3364 3365
#ifdef DEBUG
	dev_warn(DEV, "last md_mark_dirty: %s:%u\n",
		mdev->last_md_mark_dirty.func, mdev->last_md_mark_dirty.line);
#endif
P
Philipp Reisner 已提交
3366
	drbd_md_sync(mdev);
3367
	return 0;
P
Philipp Reisner 已提交
3368 3369
}

3370
const char *cmdname(enum drbd_packet cmd)
3371 3372 3373 3374 3375 3376 3377 3378 3379 3380 3381 3382 3383 3384 3385 3386 3387 3388 3389 3390 3391 3392 3393 3394 3395 3396 3397 3398 3399
{
	/* THINK may need to become several global tables
	 * when we want to support more than
	 * one PRO_VERSION */
	static const char *cmdnames[] = {
		[P_DATA]	        = "Data",
		[P_DATA_REPLY]	        = "DataReply",
		[P_RS_DATA_REPLY]	= "RSDataReply",
		[P_BARRIER]	        = "Barrier",
		[P_BITMAP]	        = "ReportBitMap",
		[P_BECOME_SYNC_TARGET]  = "BecomeSyncTarget",
		[P_BECOME_SYNC_SOURCE]  = "BecomeSyncSource",
		[P_UNPLUG_REMOTE]	= "UnplugRemote",
		[P_DATA_REQUEST]	= "DataRequest",
		[P_RS_DATA_REQUEST]     = "RSDataRequest",
		[P_SYNC_PARAM]	        = "SyncParam",
		[P_SYNC_PARAM89]	= "SyncParam89",
		[P_PROTOCOL]            = "ReportProtocol",
		[P_UUIDS]	        = "ReportUUIDs",
		[P_SIZES]	        = "ReportSizes",
		[P_STATE]	        = "ReportState",
		[P_SYNC_UUID]           = "ReportSyncUUID",
		[P_AUTH_CHALLENGE]      = "AuthChallenge",
		[P_AUTH_RESPONSE]	= "AuthResponse",
		[P_PING]		= "Ping",
		[P_PING_ACK]	        = "PingAck",
		[P_RECV_ACK]	        = "RecvAck",
		[P_WRITE_ACK]	        = "WriteAck",
		[P_RS_WRITE_ACK]	= "RSWriteAck",
3400
		[P_DISCARD_WRITE]        = "DiscardWrite",
3401 3402 3403 3404 3405 3406 3407 3408 3409 3410 3411 3412 3413 3414
		[P_NEG_ACK]	        = "NegAck",
		[P_NEG_DREPLY]	        = "NegDReply",
		[P_NEG_RS_DREPLY]	= "NegRSDReply",
		[P_BARRIER_ACK]	        = "BarrierAck",
		[P_STATE_CHG_REQ]       = "StateChgRequest",
		[P_STATE_CHG_REPLY]     = "StateChgReply",
		[P_OV_REQUEST]          = "OVRequest",
		[P_OV_REPLY]            = "OVReply",
		[P_OV_RESULT]           = "OVResult",
		[P_CSUM_RS_REQUEST]     = "CsumRSRequest",
		[P_RS_IS_IN_SYNC]	= "CsumRSIsInSync",
		[P_COMPRESSED_BITMAP]   = "CBitmap",
		[P_DELAY_PROBE]         = "DelayProbe",
		[P_OUT_OF_SYNC]		= "OutOfSync",
3415
		[P_RETRY_WRITE]		= "RetryWrite",
3416 3417 3418
		[P_RS_CANCEL]		= "RSCancel",
		[P_CONN_ST_CHG_REQ]	= "conn_st_chg_req",
		[P_CONN_ST_CHG_REPLY]	= "conn_st_chg_reply",
3419 3420
		[P_RETRY_WRITE]		= "retry_write",
		[P_PROTOCOL_UPDATE]	= "protocol_update",
3421 3422 3423 3424 3425

		/* enum drbd_packet, but not commands - obsoleted flags:
		 *	P_MAY_IGNORE
		 *	P_MAX_OPT_CMD
		 */
3426 3427
	};

3428
	/* too big for the array: 0xfffX */
3429 3430 3431 3432
	if (cmd == P_INITIAL_META)
		return "InitialMeta";
	if (cmd == P_INITIAL_DATA)
		return "InitialData";
3433 3434
	if (cmd == P_CONNECTION_FEATURES)
		return "ConnectionFeatures";
A
Andreas Gruenbacher 已提交
3435
	if (cmd >= ARRAY_SIZE(cmdnames))
3436 3437 3438 3439
		return "Unknown";
	return cmdnames[cmd];
}

3440 3441 3442 3443 3444 3445 3446 3447
/**
 * drbd_wait_misc  -  wait for a request to make progress
 * @mdev:	device associated with the request
 * @i:		the struct drbd_interval embedded in struct drbd_request or
 *		struct drbd_peer_request
 */
int drbd_wait_misc(struct drbd_conf *mdev, struct drbd_interval *i)
{
3448
	struct net_conf *nc;
3449 3450 3451
	DEFINE_WAIT(wait);
	long timeout;

3452 3453 3454 3455
	rcu_read_lock();
	nc = rcu_dereference(mdev->tconn->net_conf);
	if (!nc) {
		rcu_read_unlock();
3456
		return -ETIMEDOUT;
3457 3458 3459
	}
	timeout = nc->ko_count ? nc->timeout * HZ / 10 * nc->ko_count : MAX_SCHEDULE_TIMEOUT;
	rcu_read_unlock();
3460 3461 3462 3463 3464 3465 3466 3467 3468 3469 3470 3471 3472 3473 3474

	/* Indicate to wake up mdev->misc_wait on progress.  */
	i->waiting = true;
	prepare_to_wait(&mdev->misc_wait, &wait, TASK_INTERRUPTIBLE);
	spin_unlock_irq(&mdev->tconn->req_lock);
	timeout = schedule_timeout(timeout);
	finish_wait(&mdev->misc_wait, &wait);
	spin_lock_irq(&mdev->tconn->req_lock);
	if (!timeout || mdev->state.conn < C_CONNECTED)
		return -ETIMEDOUT;
	if (signal_pending(current))
		return -ERESTARTSYS;
	return 0;
}

P
Philipp Reisner 已提交
3475 3476 3477 3478 3479 3480 3481 3482 3483 3484 3485 3486 3487 3488 3489 3490 3491 3492 3493 3494 3495
#ifdef CONFIG_DRBD_FAULT_INJECTION
/* Fault insertion support including random number generator shamelessly
 * stolen from kernel/rcutorture.c */
struct fault_random_state {
	unsigned long state;
	unsigned long count;
};

#define FAULT_RANDOM_MULT 39916801  /* prime */
#define FAULT_RANDOM_ADD	479001701 /* prime */
#define FAULT_RANDOM_REFRESH 10000

/*
 * Crude but fast random-number generator.  Uses a linear congruential
 * generator, with occasional help from get_random_bytes().
 */
static unsigned long
_drbd_fault_random(struct fault_random_state *rsp)
{
	long refresh;

3496
	if (!rsp->count--) {
P
Philipp Reisner 已提交
3497 3498 3499 3500 3501 3502 3503 3504 3505 3506 3507 3508 3509 3510 3511 3512 3513 3514 3515
		get_random_bytes(&refresh, sizeof(refresh));
		rsp->state += refresh;
		rsp->count = FAULT_RANDOM_REFRESH;
	}
	rsp->state = rsp->state * FAULT_RANDOM_MULT + FAULT_RANDOM_ADD;
	return swahw32(rsp->state);
}

static char *
_drbd_fault_str(unsigned int type) {
	static char *_faults[] = {
		[DRBD_FAULT_MD_WR] = "Meta-data write",
		[DRBD_FAULT_MD_RD] = "Meta-data read",
		[DRBD_FAULT_RS_WR] = "Resync write",
		[DRBD_FAULT_RS_RD] = "Resync read",
		[DRBD_FAULT_DT_WR] = "Data write",
		[DRBD_FAULT_DT_RD] = "Data read",
		[DRBD_FAULT_DT_RA] = "Data read ahead",
		[DRBD_FAULT_BM_ALLOC] = "BM allocation",
3516 3517
		[DRBD_FAULT_AL_EE] = "EE allocation",
		[DRBD_FAULT_RECEIVE] = "receive data corruption",
P
Philipp Reisner 已提交
3518 3519 3520 3521 3522 3523 3524 3525 3526 3527 3528 3529 3530 3531 3532 3533 3534 3535
	};

	return (type < DRBD_FAULT_MAX) ? _faults[type] : "**Unknown**";
}

unsigned int
_drbd_insert_fault(struct drbd_conf *mdev, unsigned int type)
{
	static struct fault_random_state rrs = {0, 0};

	unsigned int ret = (
		(fault_devs == 0 ||
			((1 << mdev_to_minor(mdev)) & fault_devs) != 0) &&
		(((_drbd_fault_random(&rrs) % 100) + 1) <= fault_rate));

	if (ret) {
		fault_count++;

3536
		if (__ratelimit(&drbd_ratelimit_state))
P
Philipp Reisner 已提交
3537 3538 3539 3540 3541 3542 3543 3544 3545 3546 3547 3548 3549 3550 3551 3552 3553 3554 3555 3556 3557 3558 3559 3560 3561 3562 3563 3564 3565 3566 3567 3568 3569 3570
			dev_warn(DEV, "***Simulating %s failure\n",
				_drbd_fault_str(type));
	}

	return ret;
}
#endif

const char *drbd_buildtag(void)
{
	/* DRBD built from external sources has here a reference to the
	   git hash of the source code. */

	static char buildtag[38] = "\0uilt-in";

	if (buildtag[0] == 0) {
#ifdef CONFIG_MODULES
		if (THIS_MODULE != NULL)
			sprintf(buildtag, "srcversion: %-24s", THIS_MODULE->srcversion);
		else
#endif
			buildtag[0] = 'b';
	}

	return buildtag;
}

module_init(drbd_init)
module_exit(drbd_cleanup)

EXPORT_SYMBOL(drbd_conn_str);
EXPORT_SYMBOL(drbd_role_str);
EXPORT_SYMBOL(drbd_disk_str);
EXPORT_SYMBOL(drbd_set_st_err_str);