• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
1 /*
2    drbd_worker.c
3 
4    This file is part of DRBD by Philipp Reisner and Lars Ellenberg.
5 
6    Copyright (C) 2001-2008, LINBIT Information Technologies GmbH.
7    Copyright (C) 1999-2008, Philipp Reisner <philipp.reisner@linbit.com>.
8    Copyright (C) 2002-2008, Lars Ellenberg <lars.ellenberg@linbit.com>.
9 
10    drbd is free software; you can redistribute it and/or modify
11    it under the terms of the GNU General Public License as published by
12    the Free Software Foundation; either version 2, or (at your option)
13    any later version.
14 
15    drbd is distributed in the hope that it will be useful,
16    but WITHOUT ANY WARRANTY; without even the implied warranty of
17    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
18    GNU General Public License for more details.
19 
20    You should have received a copy of the GNU General Public License
21    along with drbd; see the file COPYING.  If not, write to
22    the Free Software Foundation, 675 Mass Ave, Cambridge, MA 02139, USA.
23 
24  */
25 
26 #include <linux/module.h>
27 #include <linux/drbd.h>
28 #include <linux/sched.h>
29 #include <linux/wait.h>
30 #include <linux/mm.h>
31 #include <linux/memcontrol.h>
32 #include <linux/mm_inline.h>
33 #include <linux/slab.h>
34 #include <linux/random.h>
35 #include <linux/string.h>
36 #include <linux/scatterlist.h>
37 
38 #include "drbd_int.h"
39 #include "drbd_req.h"
40 
41 static int w_make_ov_request(struct drbd_conf *mdev, struct drbd_work *w, int cancel);
42 static int w_make_resync_request(struct drbd_conf *mdev,
43 				 struct drbd_work *w, int cancel);
44 
45 
46 
47 /* endio handlers:
48  *   drbd_md_io_complete (defined here)
49  *   drbd_endio_pri (defined here)
50  *   drbd_endio_sec (defined here)
51  *   bm_async_io_complete (defined in drbd_bitmap.c)
52  *
53  * For all these callbacks, note the following:
54  * The callbacks will be called in irq context by the IDE drivers,
55  * and in Softirqs/Tasklets/BH context by the SCSI drivers.
56  * Try to get the locking right :)
57  *
58  */
59 
60 
61 /* About the global_state_lock
62    Each state transition on an device holds a read lock. In case we have
63    to evaluate the sync after dependencies, we grab a write lock, because
64    we need stable states on all devices for that.  */
65 rwlock_t global_state_lock;
66 
67 /* used for synchronous meta data and bitmap IO
68  * submitted by drbd_md_sync_page_io()
69  */
drbd_md_io_complete(struct bio * bio,int error)70 void drbd_md_io_complete(struct bio *bio, int error)
71 {
72 	struct drbd_md_io *md_io;
73 
74 	md_io = (struct drbd_md_io *)bio->bi_private;
75 	md_io->error = error;
76 
77 	complete(&md_io->event);
78 }
79 
80 /* reads on behalf of the partner,
81  * "submitted" by the receiver
82  */
drbd_endio_read_sec_final(struct drbd_epoch_entry * e)83 void drbd_endio_read_sec_final(struct drbd_epoch_entry *e) __releases(local)
84 {
85 	unsigned long flags = 0;
86 	struct drbd_conf *mdev = e->mdev;
87 
88 	D_ASSERT(e->block_id != ID_VACANT);
89 
90 	spin_lock_irqsave(&mdev->req_lock, flags);
91 	mdev->read_cnt += e->size >> 9;
92 	list_del(&e->w.list);
93 	if (list_empty(&mdev->read_ee))
94 		wake_up(&mdev->ee_wait);
95 	if (test_bit(__EE_WAS_ERROR, &e->flags))
96 		__drbd_chk_io_error(mdev, false);
97 	spin_unlock_irqrestore(&mdev->req_lock, flags);
98 
99 	drbd_queue_work(&mdev->data.work, &e->w);
100 	put_ldev(mdev);
101 }
102 
103 /* writes on behalf of the partner, or resync writes,
104  * "submitted" by the receiver, final stage.  */
drbd_endio_write_sec_final(struct drbd_epoch_entry * e)105 static void drbd_endio_write_sec_final(struct drbd_epoch_entry *e) __releases(local)
106 {
107 	unsigned long flags = 0;
108 	struct drbd_conf *mdev = e->mdev;
109 	sector_t e_sector;
110 	int do_wake;
111 	int is_syncer_req;
112 	int do_al_complete_io;
113 
114 	D_ASSERT(e->block_id != ID_VACANT);
115 
116 	/* after we moved e to done_ee,
117 	 * we may no longer access it,
118 	 * it may be freed/reused already!
119 	 * (as soon as we release the req_lock) */
120 	e_sector = e->sector;
121 	do_al_complete_io = e->flags & EE_CALL_AL_COMPLETE_IO;
122 	is_syncer_req = is_syncer_block_id(e->block_id);
123 
124 	spin_lock_irqsave(&mdev->req_lock, flags);
125 	mdev->writ_cnt += e->size >> 9;
126 	list_del(&e->w.list); /* has been on active_ee or sync_ee */
127 	list_add_tail(&e->w.list, &mdev->done_ee);
128 
129 	/* No hlist_del_init(&e->collision) here, we did not send the Ack yet,
130 	 * neither did we wake possibly waiting conflicting requests.
131 	 * done from "drbd_process_done_ee" within the appropriate w.cb
132 	 * (e_end_block/e_end_resync_block) or from _drbd_clear_done_ee */
133 
134 	do_wake = is_syncer_req
135 		? list_empty(&mdev->sync_ee)
136 		: list_empty(&mdev->active_ee);
137 
138 	if (test_bit(__EE_WAS_ERROR, &e->flags))
139 		__drbd_chk_io_error(mdev, false);
140 	spin_unlock_irqrestore(&mdev->req_lock, flags);
141 
142 	if (is_syncer_req)
143 		drbd_rs_complete_io(mdev, e_sector);
144 
145 	if (do_wake)
146 		wake_up(&mdev->ee_wait);
147 
148 	if (do_al_complete_io)
149 		drbd_al_complete_io(mdev, e_sector);
150 
151 	wake_asender(mdev);
152 	put_ldev(mdev);
153 }
154 
155 /* writes on behalf of the partner, or resync writes,
156  * "submitted" by the receiver.
157  */
drbd_endio_sec(struct bio * bio,int error)158 void drbd_endio_sec(struct bio *bio, int error)
159 {
160 	struct drbd_epoch_entry *e = bio->bi_private;
161 	struct drbd_conf *mdev = e->mdev;
162 	int uptodate = bio_flagged(bio, BIO_UPTODATE);
163 	int is_write = bio_data_dir(bio) == WRITE;
164 
165 	if (error && __ratelimit(&drbd_ratelimit_state))
166 		dev_warn(DEV, "%s: error=%d s=%llus\n",
167 				is_write ? "write" : "read", error,
168 				(unsigned long long)e->sector);
169 	if (!error && !uptodate) {
170 		if (__ratelimit(&drbd_ratelimit_state))
171 			dev_warn(DEV, "%s: setting error to -EIO s=%llus\n",
172 					is_write ? "write" : "read",
173 					(unsigned long long)e->sector);
174 		/* strange behavior of some lower level drivers...
175 		 * fail the request by clearing the uptodate flag,
176 		 * but do not return any error?! */
177 		error = -EIO;
178 	}
179 
180 	if (error)
181 		set_bit(__EE_WAS_ERROR, &e->flags);
182 
183 	bio_put(bio); /* no need for the bio anymore */
184 	if (atomic_dec_and_test(&e->pending_bios)) {
185 		if (is_write)
186 			drbd_endio_write_sec_final(e);
187 		else
188 			drbd_endio_read_sec_final(e);
189 	}
190 }
191 
192 /* read, readA or write requests on R_PRIMARY coming from drbd_make_request
193  */
drbd_endio_pri(struct bio * bio,int error)194 void drbd_endio_pri(struct bio *bio, int error)
195 {
196 	unsigned long flags;
197 	struct drbd_request *req = bio->bi_private;
198 	struct drbd_conf *mdev = req->mdev;
199 	struct bio_and_error m;
200 	enum drbd_req_event what;
201 	int uptodate = bio_flagged(bio, BIO_UPTODATE);
202 
203 	if (!error && !uptodate) {
204 		dev_warn(DEV, "p %s: setting error to -EIO\n",
205 			 bio_data_dir(bio) == WRITE ? "write" : "read");
206 		/* strange behavior of some lower level drivers...
207 		 * fail the request by clearing the uptodate flag,
208 		 * but do not return any error?! */
209 		error = -EIO;
210 	}
211 
212 	/* to avoid recursion in __req_mod */
213 	if (unlikely(error)) {
214 		what = (bio_data_dir(bio) == WRITE)
215 			? write_completed_with_error
216 			: (bio_rw(bio) == READ)
217 			  ? read_completed_with_error
218 			  : read_ahead_completed_with_error;
219 	} else
220 		what = completed_ok;
221 
222 	bio_put(req->private_bio);
223 	req->private_bio = ERR_PTR(error);
224 
225 	/* not req_mod(), we need irqsave here! */
226 	spin_lock_irqsave(&mdev->req_lock, flags);
227 	__req_mod(req, what, &m);
228 	spin_unlock_irqrestore(&mdev->req_lock, flags);
229 
230 	if (m.bio)
231 		complete_master_bio(mdev, &m);
232 }
233 
w_read_retry_remote(struct drbd_conf * mdev,struct drbd_work * w,int cancel)234 int w_read_retry_remote(struct drbd_conf *mdev, struct drbd_work *w, int cancel)
235 {
236 	struct drbd_request *req = container_of(w, struct drbd_request, w);
237 
238 	/* We should not detach for read io-error,
239 	 * but try to WRITE the P_DATA_REPLY to the failed location,
240 	 * to give the disk the chance to relocate that block */
241 
242 	spin_lock_irq(&mdev->req_lock);
243 	if (cancel || mdev->state.pdsk != D_UP_TO_DATE) {
244 		_req_mod(req, read_retry_remote_canceled);
245 		spin_unlock_irq(&mdev->req_lock);
246 		return 1;
247 	}
248 	spin_unlock_irq(&mdev->req_lock);
249 
250 	return w_send_read_req(mdev, w, 0);
251 }
252 
drbd_csum_ee(struct drbd_conf * mdev,struct crypto_hash * tfm,struct drbd_epoch_entry * e,void * digest)253 void drbd_csum_ee(struct drbd_conf *mdev, struct crypto_hash *tfm, struct drbd_epoch_entry *e, void *digest)
254 {
255 	struct hash_desc desc;
256 	struct scatterlist sg;
257 	struct page *page = e->pages;
258 	struct page *tmp;
259 	unsigned len;
260 
261 	desc.tfm = tfm;
262 	desc.flags = 0;
263 
264 	sg_init_table(&sg, 1);
265 	crypto_hash_init(&desc);
266 
267 	while ((tmp = page_chain_next(page))) {
268 		/* all but the last page will be fully used */
269 		sg_set_page(&sg, page, PAGE_SIZE, 0);
270 		crypto_hash_update(&desc, &sg, sg.length);
271 		page = tmp;
272 	}
273 	/* and now the last, possibly only partially used page */
274 	len = e->size & (PAGE_SIZE - 1);
275 	sg_set_page(&sg, page, len ?: PAGE_SIZE, 0);
276 	crypto_hash_update(&desc, &sg, sg.length);
277 	crypto_hash_final(&desc, digest);
278 }
279 
drbd_csum_bio(struct drbd_conf * mdev,struct crypto_hash * tfm,struct bio * bio,void * digest)280 void drbd_csum_bio(struct drbd_conf *mdev, struct crypto_hash *tfm, struct bio *bio, void *digest)
281 {
282 	struct hash_desc desc;
283 	struct scatterlist sg;
284 	struct bio_vec *bvec;
285 	int i;
286 
287 	desc.tfm = tfm;
288 	desc.flags = 0;
289 
290 	sg_init_table(&sg, 1);
291 	crypto_hash_init(&desc);
292 
293 	__bio_for_each_segment(bvec, bio, i, 0) {
294 		sg_set_page(&sg, bvec->bv_page, bvec->bv_len, bvec->bv_offset);
295 		crypto_hash_update(&desc, &sg, sg.length);
296 	}
297 	crypto_hash_final(&desc, digest);
298 }
299 
300 /* TODO merge common code with w_e_end_ov_req */
w_e_send_csum(struct drbd_conf * mdev,struct drbd_work * w,int cancel)301 int w_e_send_csum(struct drbd_conf *mdev, struct drbd_work *w, int cancel)
302 {
303 	struct drbd_epoch_entry *e = container_of(w, struct drbd_epoch_entry, w);
304 	int digest_size;
305 	void *digest;
306 	int ok = 1;
307 
308 	D_ASSERT(e->block_id == DRBD_MAGIC + 0xbeef);
309 
310 	if (unlikely(cancel))
311 		goto out;
312 
313 	if (likely((e->flags & EE_WAS_ERROR) != 0))
314 		goto out;
315 
316 	digest_size = crypto_hash_digestsize(mdev->csums_tfm);
317 	digest = kmalloc(digest_size, GFP_NOIO);
318 	if (digest) {
319 		sector_t sector = e->sector;
320 		unsigned int size = e->size;
321 		drbd_csum_ee(mdev, mdev->csums_tfm, e, digest);
322 		/* Free e and pages before send.
323 		 * In case we block on congestion, we could otherwise run into
324 		 * some distributed deadlock, if the other side blocks on
325 		 * congestion as well, because our receiver blocks in
326 		 * drbd_pp_alloc due to pp_in_use > max_buffers. */
327 		drbd_free_ee(mdev, e);
328 		e = NULL;
329 		inc_rs_pending(mdev);
330 		ok = drbd_send_drequest_csum(mdev, sector, size,
331 					     digest, digest_size,
332 					     P_CSUM_RS_REQUEST);
333 		kfree(digest);
334 	} else {
335 		dev_err(DEV, "kmalloc() of digest failed.\n");
336 		ok = 0;
337 	}
338 
339 out:
340 	if (e)
341 		drbd_free_ee(mdev, e);
342 
343 	if (unlikely(!ok))
344 		dev_err(DEV, "drbd_send_drequest(..., csum) failed\n");
345 	return ok;
346 }
347 
348 #define GFP_TRY	(__GFP_HIGHMEM | __GFP_NOWARN)
349 
read_for_csum(struct drbd_conf * mdev,sector_t sector,int size)350 static int read_for_csum(struct drbd_conf *mdev, sector_t sector, int size)
351 {
352 	struct drbd_epoch_entry *e;
353 
354 	if (!get_ldev(mdev))
355 		return -EIO;
356 
357 	if (drbd_rs_should_slow_down(mdev, sector))
358 		goto defer;
359 
360 	/* GFP_TRY, because if there is no memory available right now, this may
361 	 * be rescheduled for later. It is "only" background resync, after all. */
362 	e = drbd_alloc_ee(mdev, DRBD_MAGIC+0xbeef, sector, size, GFP_TRY);
363 	if (!e)
364 		goto defer;
365 
366 	e->w.cb = w_e_send_csum;
367 	spin_lock_irq(&mdev->req_lock);
368 	list_add(&e->w.list, &mdev->read_ee);
369 	spin_unlock_irq(&mdev->req_lock);
370 
371 	atomic_add(size >> 9, &mdev->rs_sect_ev);
372 	if (drbd_submit_ee(mdev, e, READ, DRBD_FAULT_RS_RD) == 0)
373 		return 0;
374 
375 	/* If it failed because of ENOMEM, retry should help.  If it failed
376 	 * because bio_add_page failed (probably broken lower level driver),
377 	 * retry may or may not help.
378 	 * If it does not, you may need to force disconnect. */
379 	spin_lock_irq(&mdev->req_lock);
380 	list_del(&e->w.list);
381 	spin_unlock_irq(&mdev->req_lock);
382 
383 	drbd_free_ee(mdev, e);
384 defer:
385 	put_ldev(mdev);
386 	return -EAGAIN;
387 }
388 
w_resync_timer(struct drbd_conf * mdev,struct drbd_work * w,int cancel)389 int w_resync_timer(struct drbd_conf *mdev, struct drbd_work *w, int cancel)
390 {
391 	switch (mdev->state.conn) {
392 	case C_VERIFY_S:
393 		w_make_ov_request(mdev, w, cancel);
394 		break;
395 	case C_SYNC_TARGET:
396 		w_make_resync_request(mdev, w, cancel);
397 		break;
398 	}
399 
400 	return 1;
401 }
402 
resync_timer_fn(unsigned long data)403 void resync_timer_fn(unsigned long data)
404 {
405 	struct drbd_conf *mdev = (struct drbd_conf *) data;
406 
407 	if (list_empty(&mdev->resync_work.list))
408 		drbd_queue_work(&mdev->data.work, &mdev->resync_work);
409 }
410 
fifo_set(struct fifo_buffer * fb,int value)411 static void fifo_set(struct fifo_buffer *fb, int value)
412 {
413 	int i;
414 
415 	for (i = 0; i < fb->size; i++)
416 		fb->values[i] = value;
417 }
418 
fifo_push(struct fifo_buffer * fb,int value)419 static int fifo_push(struct fifo_buffer *fb, int value)
420 {
421 	int ov;
422 
423 	ov = fb->values[fb->head_index];
424 	fb->values[fb->head_index++] = value;
425 
426 	if (fb->head_index >= fb->size)
427 		fb->head_index = 0;
428 
429 	return ov;
430 }
431 
fifo_add_val(struct fifo_buffer * fb,int value)432 static void fifo_add_val(struct fifo_buffer *fb, int value)
433 {
434 	int i;
435 
436 	for (i = 0; i < fb->size; i++)
437 		fb->values[i] += value;
438 }
439 
drbd_rs_controller(struct drbd_conf * mdev)440 static int drbd_rs_controller(struct drbd_conf *mdev)
441 {
442 	unsigned int sect_in;  /* Number of sectors that came in since the last turn */
443 	unsigned int want;     /* The number of sectors we want in the proxy */
444 	int req_sect; /* Number of sectors to request in this turn */
445 	int correction; /* Number of sectors more we need in the proxy*/
446 	int cps; /* correction per invocation of drbd_rs_controller() */
447 	int steps; /* Number of time steps to plan ahead */
448 	int curr_corr;
449 	int max_sect;
450 
451 	sect_in = atomic_xchg(&mdev->rs_sect_in, 0); /* Number of sectors that came in */
452 	mdev->rs_in_flight -= sect_in;
453 
454 	spin_lock(&mdev->peer_seq_lock); /* get an atomic view on mdev->rs_plan_s */
455 
456 	steps = mdev->rs_plan_s.size; /* (mdev->sync_conf.c_plan_ahead * 10 * SLEEP_TIME) / HZ; */
457 
458 	if (mdev->rs_in_flight + sect_in == 0) { /* At start of resync */
459 		want = ((mdev->sync_conf.rate * 2 * SLEEP_TIME) / HZ) * steps;
460 	} else { /* normal path */
461 		want = mdev->sync_conf.c_fill_target ? mdev->sync_conf.c_fill_target :
462 			sect_in * mdev->sync_conf.c_delay_target * HZ / (SLEEP_TIME * 10);
463 	}
464 
465 	correction = want - mdev->rs_in_flight - mdev->rs_planed;
466 
467 	/* Plan ahead */
468 	cps = correction / steps;
469 	fifo_add_val(&mdev->rs_plan_s, cps);
470 	mdev->rs_planed += cps * steps;
471 
472 	/* What we do in this step */
473 	curr_corr = fifo_push(&mdev->rs_plan_s, 0);
474 	spin_unlock(&mdev->peer_seq_lock);
475 	mdev->rs_planed -= curr_corr;
476 
477 	req_sect = sect_in + curr_corr;
478 	if (req_sect < 0)
479 		req_sect = 0;
480 
481 	max_sect = (mdev->sync_conf.c_max_rate * 2 * SLEEP_TIME) / HZ;
482 	if (req_sect > max_sect)
483 		req_sect = max_sect;
484 
485 	/*
486 	dev_warn(DEV, "si=%u if=%d wa=%u co=%d st=%d cps=%d pl=%d cc=%d rs=%d\n",
487 		 sect_in, mdev->rs_in_flight, want, correction,
488 		 steps, cps, mdev->rs_planed, curr_corr, req_sect);
489 	*/
490 
491 	return req_sect;
492 }
493 
drbd_rs_number_requests(struct drbd_conf * mdev)494 static int drbd_rs_number_requests(struct drbd_conf *mdev)
495 {
496 	int number;
497 	if (mdev->rs_plan_s.size) { /* mdev->sync_conf.c_plan_ahead */
498 		number = drbd_rs_controller(mdev) >> (BM_BLOCK_SHIFT - 9);
499 		mdev->c_sync_rate = number * HZ * (BM_BLOCK_SIZE / 1024) / SLEEP_TIME;
500 	} else {
501 		mdev->c_sync_rate = mdev->sync_conf.rate;
502 		number = SLEEP_TIME * mdev->c_sync_rate  / ((BM_BLOCK_SIZE / 1024) * HZ);
503 	}
504 
505 	/* ignore the amount of pending requests, the resync controller should
506 	 * throttle down to incoming reply rate soon enough anyways. */
507 	return number;
508 }
509 
w_make_resync_request(struct drbd_conf * mdev,struct drbd_work * w,int cancel)510 static int w_make_resync_request(struct drbd_conf *mdev,
511 				 struct drbd_work *w, int cancel)
512 {
513 	unsigned long bit;
514 	sector_t sector;
515 	const sector_t capacity = drbd_get_capacity(mdev->this_bdev);
516 	int max_bio_size;
517 	int number, rollback_i, size;
518 	int align, queued, sndbuf;
519 	int i = 0;
520 
521 	if (unlikely(cancel))
522 		return 1;
523 
524 	if (mdev->rs_total == 0) {
525 		/* empty resync? */
526 		drbd_resync_finished(mdev);
527 		return 1;
528 	}
529 
530 	if (!get_ldev(mdev)) {
531 		/* Since we only need to access mdev->rsync a
532 		   get_ldev_if_state(mdev,D_FAILED) would be sufficient, but
533 		   to continue resync with a broken disk makes no sense at
534 		   all */
535 		dev_err(DEV, "Disk broke down during resync!\n");
536 		return 1;
537 	}
538 
539 	max_bio_size = queue_max_hw_sectors(mdev->rq_queue) << 9;
540 	number = drbd_rs_number_requests(mdev);
541 	if (number == 0)
542 		goto requeue;
543 
544 	for (i = 0; i < number; i++) {
545 		/* Stop generating RS requests, when half of the send buffer is filled */
546 		mutex_lock(&mdev->data.mutex);
547 		if (mdev->data.socket) {
548 			queued = mdev->data.socket->sk->sk_wmem_queued;
549 			sndbuf = mdev->data.socket->sk->sk_sndbuf;
550 		} else {
551 			queued = 1;
552 			sndbuf = 0;
553 		}
554 		mutex_unlock(&mdev->data.mutex);
555 		if (queued > sndbuf / 2)
556 			goto requeue;
557 
558 next_sector:
559 		size = BM_BLOCK_SIZE;
560 		bit  = drbd_bm_find_next(mdev, mdev->bm_resync_fo);
561 
562 		if (bit == DRBD_END_OF_BITMAP) {
563 			mdev->bm_resync_fo = drbd_bm_bits(mdev);
564 			put_ldev(mdev);
565 			return 1;
566 		}
567 
568 		sector = BM_BIT_TO_SECT(bit);
569 
570 		if (drbd_rs_should_slow_down(mdev, sector) ||
571 		    drbd_try_rs_begin_io(mdev, sector)) {
572 			mdev->bm_resync_fo = bit;
573 			goto requeue;
574 		}
575 		mdev->bm_resync_fo = bit + 1;
576 
577 		if (unlikely(drbd_bm_test_bit(mdev, bit) == 0)) {
578 			drbd_rs_complete_io(mdev, sector);
579 			goto next_sector;
580 		}
581 
582 #if DRBD_MAX_BIO_SIZE > BM_BLOCK_SIZE
583 		/* try to find some adjacent bits.
584 		 * we stop if we have already the maximum req size.
585 		 *
586 		 * Additionally always align bigger requests, in order to
587 		 * be prepared for all stripe sizes of software RAIDs.
588 		 */
589 		align = 1;
590 		rollback_i = i;
591 		for (;;) {
592 			if (size + BM_BLOCK_SIZE > max_bio_size)
593 				break;
594 
595 			/* Be always aligned */
596 			if (sector & ((1<<(align+3))-1))
597 				break;
598 
599 			/* do not cross extent boundaries */
600 			if (((bit+1) & BM_BLOCKS_PER_BM_EXT_MASK) == 0)
601 				break;
602 			/* now, is it actually dirty, after all?
603 			 * caution, drbd_bm_test_bit is tri-state for some
604 			 * obscure reason; ( b == 0 ) would get the out-of-band
605 			 * only accidentally right because of the "oddly sized"
606 			 * adjustment below */
607 			if (drbd_bm_test_bit(mdev, bit+1) != 1)
608 				break;
609 			bit++;
610 			size += BM_BLOCK_SIZE;
611 			if ((BM_BLOCK_SIZE << align) <= size)
612 				align++;
613 			i++;
614 		}
615 		/* if we merged some,
616 		 * reset the offset to start the next drbd_bm_find_next from */
617 		if (size > BM_BLOCK_SIZE)
618 			mdev->bm_resync_fo = bit + 1;
619 #endif
620 
621 		/* adjust very last sectors, in case we are oddly sized */
622 		if (sector + (size>>9) > capacity)
623 			size = (capacity-sector)<<9;
624 		if (mdev->agreed_pro_version >= 89 && mdev->csums_tfm) {
625 			switch (read_for_csum(mdev, sector, size)) {
626 			case -EIO: /* Disk failure */
627 				put_ldev(mdev);
628 				return 0;
629 			case -EAGAIN: /* allocation failed, or ldev busy */
630 				drbd_rs_complete_io(mdev, sector);
631 				mdev->bm_resync_fo = BM_SECT_TO_BIT(sector);
632 				i = rollback_i;
633 				goto requeue;
634 			case 0:
635 				/* everything ok */
636 				break;
637 			default:
638 				BUG();
639 			}
640 		} else {
641 			inc_rs_pending(mdev);
642 			if (!drbd_send_drequest(mdev, P_RS_DATA_REQUEST,
643 					       sector, size, ID_SYNCER)) {
644 				dev_err(DEV, "drbd_send_drequest() failed, aborting...\n");
645 				dec_rs_pending(mdev);
646 				put_ldev(mdev);
647 				return 0;
648 			}
649 		}
650 	}
651 
652 	if (mdev->bm_resync_fo >= drbd_bm_bits(mdev)) {
653 		/* last syncer _request_ was sent,
654 		 * but the P_RS_DATA_REPLY not yet received.  sync will end (and
655 		 * next sync group will resume), as soon as we receive the last
656 		 * resync data block, and the last bit is cleared.
657 		 * until then resync "work" is "inactive" ...
658 		 */
659 		put_ldev(mdev);
660 		return 1;
661 	}
662 
663  requeue:
664 	mdev->rs_in_flight += (i << (BM_BLOCK_SHIFT - 9));
665 	mod_timer(&mdev->resync_timer, jiffies + SLEEP_TIME);
666 	put_ldev(mdev);
667 	return 1;
668 }
669 
w_make_ov_request(struct drbd_conf * mdev,struct drbd_work * w,int cancel)670 static int w_make_ov_request(struct drbd_conf *mdev, struct drbd_work *w, int cancel)
671 {
672 	int number, i, size;
673 	sector_t sector;
674 	const sector_t capacity = drbd_get_capacity(mdev->this_bdev);
675 
676 	if (unlikely(cancel))
677 		return 1;
678 
679 	number = drbd_rs_number_requests(mdev);
680 
681 	sector = mdev->ov_position;
682 	for (i = 0; i < number; i++) {
683 		if (sector >= capacity) {
684 			return 1;
685 		}
686 
687 		size = BM_BLOCK_SIZE;
688 
689 		if (drbd_rs_should_slow_down(mdev, sector) ||
690 		    drbd_try_rs_begin_io(mdev, sector)) {
691 			mdev->ov_position = sector;
692 			goto requeue;
693 		}
694 
695 		if (sector + (size>>9) > capacity)
696 			size = (capacity-sector)<<9;
697 
698 		inc_rs_pending(mdev);
699 		if (!drbd_send_ov_request(mdev, sector, size)) {
700 			dec_rs_pending(mdev);
701 			return 0;
702 		}
703 		sector += BM_SECT_PER_BIT;
704 	}
705 	mdev->ov_position = sector;
706 
707  requeue:
708 	mdev->rs_in_flight += (i << (BM_BLOCK_SHIFT - 9));
709 	mod_timer(&mdev->resync_timer, jiffies + SLEEP_TIME);
710 	return 1;
711 }
712 
713 
start_resync_timer_fn(unsigned long data)714 void start_resync_timer_fn(unsigned long data)
715 {
716 	struct drbd_conf *mdev = (struct drbd_conf *) data;
717 
718 	drbd_queue_work(&mdev->data.work, &mdev->start_resync_work);
719 }
720 
w_start_resync(struct drbd_conf * mdev,struct drbd_work * w,int cancel)721 int w_start_resync(struct drbd_conf *mdev, struct drbd_work *w, int cancel)
722 {
723 	if (atomic_read(&mdev->unacked_cnt) || atomic_read(&mdev->rs_pending_cnt)) {
724 		dev_warn(DEV, "w_start_resync later...\n");
725 		mdev->start_resync_timer.expires = jiffies + HZ/10;
726 		add_timer(&mdev->start_resync_timer);
727 		return 1;
728 	}
729 
730 	drbd_start_resync(mdev, C_SYNC_SOURCE);
731 	clear_bit(AHEAD_TO_SYNC_SOURCE, &mdev->current_epoch->flags);
732 	return 1;
733 }
734 
w_ov_finished(struct drbd_conf * mdev,struct drbd_work * w,int cancel)735 int w_ov_finished(struct drbd_conf *mdev, struct drbd_work *w, int cancel)
736 {
737 	kfree(w);
738 	ov_oos_print(mdev);
739 	drbd_resync_finished(mdev);
740 
741 	return 1;
742 }
743 
w_resync_finished(struct drbd_conf * mdev,struct drbd_work * w,int cancel)744 static int w_resync_finished(struct drbd_conf *mdev, struct drbd_work *w, int cancel)
745 {
746 	kfree(w);
747 
748 	drbd_resync_finished(mdev);
749 
750 	return 1;
751 }
752 
ping_peer(struct drbd_conf * mdev)753 static void ping_peer(struct drbd_conf *mdev)
754 {
755 	clear_bit(GOT_PING_ACK, &mdev->flags);
756 	request_ping(mdev);
757 	wait_event(mdev->misc_wait,
758 		   test_bit(GOT_PING_ACK, &mdev->flags) || mdev->state.conn < C_CONNECTED);
759 }
760 
drbd_resync_finished(struct drbd_conf * mdev)761 int drbd_resync_finished(struct drbd_conf *mdev)
762 {
763 	unsigned long db, dt, dbdt;
764 	unsigned long n_oos;
765 	union drbd_state os, ns;
766 	struct drbd_work *w;
767 	char *khelper_cmd = NULL;
768 	int verify_done = 0;
769 
770 	/* Remove all elements from the resync LRU. Since future actions
771 	 * might set bits in the (main) bitmap, then the entries in the
772 	 * resync LRU would be wrong. */
773 	if (drbd_rs_del_all(mdev)) {
774 		/* In case this is not possible now, most probably because
775 		 * there are P_RS_DATA_REPLY Packets lingering on the worker's
776 		 * queue (or even the read operations for those packets
777 		 * is not finished by now).   Retry in 100ms. */
778 
779 		schedule_timeout_interruptible(HZ / 10);
780 		w = kmalloc(sizeof(struct drbd_work), GFP_ATOMIC);
781 		if (w) {
782 			w->cb = w_resync_finished;
783 			drbd_queue_work(&mdev->data.work, w);
784 			return 1;
785 		}
786 		dev_err(DEV, "Warn failed to drbd_rs_del_all() and to kmalloc(w).\n");
787 	}
788 
789 	dt = (jiffies - mdev->rs_start - mdev->rs_paused) / HZ;
790 	if (dt <= 0)
791 		dt = 1;
792 	db = mdev->rs_total;
793 	dbdt = Bit2KB(db/dt);
794 	mdev->rs_paused /= HZ;
795 
796 	if (!get_ldev(mdev))
797 		goto out;
798 
799 	ping_peer(mdev);
800 
801 	spin_lock_irq(&mdev->req_lock);
802 	os = mdev->state;
803 
804 	verify_done = (os.conn == C_VERIFY_S || os.conn == C_VERIFY_T);
805 
806 	/* This protects us against multiple calls (that can happen in the presence
807 	   of application IO), and against connectivity loss just before we arrive here. */
808 	if (os.conn <= C_CONNECTED)
809 		goto out_unlock;
810 
811 	ns = os;
812 	ns.conn = C_CONNECTED;
813 
814 	dev_info(DEV, "%s done (total %lu sec; paused %lu sec; %lu K/sec)\n",
815 	     verify_done ? "Online verify " : "Resync",
816 	     dt + mdev->rs_paused, mdev->rs_paused, dbdt);
817 
818 	n_oos = drbd_bm_total_weight(mdev);
819 
820 	if (os.conn == C_VERIFY_S || os.conn == C_VERIFY_T) {
821 		if (n_oos) {
822 			dev_alert(DEV, "Online verify found %lu %dk block out of sync!\n",
823 			      n_oos, Bit2KB(1));
824 			khelper_cmd = "out-of-sync";
825 		}
826 	} else {
827 		D_ASSERT((n_oos - mdev->rs_failed) == 0);
828 
829 		if (os.conn == C_SYNC_TARGET || os.conn == C_PAUSED_SYNC_T)
830 			khelper_cmd = "after-resync-target";
831 
832 		if (mdev->csums_tfm && mdev->rs_total) {
833 			const unsigned long s = mdev->rs_same_csum;
834 			const unsigned long t = mdev->rs_total;
835 			const int ratio =
836 				(t == 0)     ? 0 :
837 			(t < 100000) ? ((s*100)/t) : (s/(t/100));
838 			dev_info(DEV, "%u %% had equal checksums, eliminated: %luK; "
839 			     "transferred %luK total %luK\n",
840 			     ratio,
841 			     Bit2KB(mdev->rs_same_csum),
842 			     Bit2KB(mdev->rs_total - mdev->rs_same_csum),
843 			     Bit2KB(mdev->rs_total));
844 		}
845 	}
846 
847 	if (mdev->rs_failed) {
848 		dev_info(DEV, "            %lu failed blocks\n", mdev->rs_failed);
849 
850 		if (os.conn == C_SYNC_TARGET || os.conn == C_PAUSED_SYNC_T) {
851 			ns.disk = D_INCONSISTENT;
852 			ns.pdsk = D_UP_TO_DATE;
853 		} else {
854 			ns.disk = D_UP_TO_DATE;
855 			ns.pdsk = D_INCONSISTENT;
856 		}
857 	} else {
858 		ns.disk = D_UP_TO_DATE;
859 		ns.pdsk = D_UP_TO_DATE;
860 
861 		if (os.conn == C_SYNC_TARGET || os.conn == C_PAUSED_SYNC_T) {
862 			if (mdev->p_uuid) {
863 				int i;
864 				for (i = UI_BITMAP ; i <= UI_HISTORY_END ; i++)
865 					_drbd_uuid_set(mdev, i, mdev->p_uuid[i]);
866 				drbd_uuid_set(mdev, UI_BITMAP, mdev->ldev->md.uuid[UI_CURRENT]);
867 				_drbd_uuid_set(mdev, UI_CURRENT, mdev->p_uuid[UI_CURRENT]);
868 			} else {
869 				dev_err(DEV, "mdev->p_uuid is NULL! BUG\n");
870 			}
871 		}
872 
873 		if (!(os.conn == C_VERIFY_S || os.conn == C_VERIFY_T)) {
874 			/* for verify runs, we don't update uuids here,
875 			 * so there would be nothing to report. */
876 			drbd_uuid_set_bm(mdev, 0UL);
877 			drbd_print_uuids(mdev, "updated UUIDs");
878 			if (mdev->p_uuid) {
879 				/* Now the two UUID sets are equal, update what we
880 				 * know of the peer. */
881 				int i;
882 				for (i = UI_CURRENT ; i <= UI_HISTORY_END ; i++)
883 					mdev->p_uuid[i] = mdev->ldev->md.uuid[i];
884 			}
885 		}
886 	}
887 
888 	_drbd_set_state(mdev, ns, CS_VERBOSE, NULL);
889 out_unlock:
890 	spin_unlock_irq(&mdev->req_lock);
891 	put_ldev(mdev);
892 out:
893 	mdev->rs_total  = 0;
894 	mdev->rs_failed = 0;
895 	mdev->rs_paused = 0;
896 	if (verify_done)
897 		mdev->ov_start_sector = 0;
898 
899 	drbd_md_sync(mdev);
900 
901 	if (khelper_cmd)
902 		drbd_khelper(mdev, khelper_cmd);
903 
904 	return 1;
905 }
906 
907 /* helper */
move_to_net_ee_or_free(struct drbd_conf * mdev,struct drbd_epoch_entry * e)908 static void move_to_net_ee_or_free(struct drbd_conf *mdev, struct drbd_epoch_entry *e)
909 {
910 	if (drbd_ee_has_active_page(e)) {
911 		/* This might happen if sendpage() has not finished */
912 		int i = (e->size + PAGE_SIZE -1) >> PAGE_SHIFT;
913 		atomic_add(i, &mdev->pp_in_use_by_net);
914 		atomic_sub(i, &mdev->pp_in_use);
915 		spin_lock_irq(&mdev->req_lock);
916 		list_add_tail(&e->w.list, &mdev->net_ee);
917 		spin_unlock_irq(&mdev->req_lock);
918 		wake_up(&drbd_pp_wait);
919 	} else
920 		drbd_free_ee(mdev, e);
921 }
922 
923 /**
924  * w_e_end_data_req() - Worker callback, to send a P_DATA_REPLY packet in response to a P_DATA_REQUEST
925  * @mdev:	DRBD device.
926  * @w:		work object.
927  * @cancel:	The connection will be closed anyways
928  */
w_e_end_data_req(struct drbd_conf * mdev,struct drbd_work * w,int cancel)929 int w_e_end_data_req(struct drbd_conf *mdev, struct drbd_work *w, int cancel)
930 {
931 	struct drbd_epoch_entry *e = container_of(w, struct drbd_epoch_entry, w);
932 	int ok;
933 
934 	if (unlikely(cancel)) {
935 		drbd_free_ee(mdev, e);
936 		dec_unacked(mdev);
937 		return 1;
938 	}
939 
940 	if (likely((e->flags & EE_WAS_ERROR) == 0)) {
941 		ok = drbd_send_block(mdev, P_DATA_REPLY, e);
942 	} else {
943 		if (__ratelimit(&drbd_ratelimit_state))
944 			dev_err(DEV, "Sending NegDReply. sector=%llus.\n",
945 			    (unsigned long long)e->sector);
946 
947 		ok = drbd_send_ack(mdev, P_NEG_DREPLY, e);
948 	}
949 
950 	dec_unacked(mdev);
951 
952 	move_to_net_ee_or_free(mdev, e);
953 
954 	if (unlikely(!ok))
955 		dev_err(DEV, "drbd_send_block() failed\n");
956 	return ok;
957 }
958 
959 /**
960  * w_e_end_rsdata_req() - Worker callback to send a P_RS_DATA_REPLY packet in response to a P_RS_DATA_REQUESTRS
961  * @mdev:	DRBD device.
962  * @w:		work object.
963  * @cancel:	The connection will be closed anyways
964  */
w_e_end_rsdata_req(struct drbd_conf * mdev,struct drbd_work * w,int cancel)965 int w_e_end_rsdata_req(struct drbd_conf *mdev, struct drbd_work *w, int cancel)
966 {
967 	struct drbd_epoch_entry *e = container_of(w, struct drbd_epoch_entry, w);
968 	int ok;
969 
970 	if (unlikely(cancel)) {
971 		drbd_free_ee(mdev, e);
972 		dec_unacked(mdev);
973 		return 1;
974 	}
975 
976 	if (get_ldev_if_state(mdev, D_FAILED)) {
977 		drbd_rs_complete_io(mdev, e->sector);
978 		put_ldev(mdev);
979 	}
980 
981 	if (mdev->state.conn == C_AHEAD) {
982 		ok = drbd_send_ack(mdev, P_RS_CANCEL, e);
983 	} else if (likely((e->flags & EE_WAS_ERROR) == 0)) {
984 		if (likely(mdev->state.pdsk >= D_INCONSISTENT)) {
985 			inc_rs_pending(mdev);
986 			ok = drbd_send_block(mdev, P_RS_DATA_REPLY, e);
987 		} else {
988 			if (__ratelimit(&drbd_ratelimit_state))
989 				dev_err(DEV, "Not sending RSDataReply, "
990 				    "partner DISKLESS!\n");
991 			ok = 1;
992 		}
993 	} else {
994 		if (__ratelimit(&drbd_ratelimit_state))
995 			dev_err(DEV, "Sending NegRSDReply. sector %llus.\n",
996 			    (unsigned long long)e->sector);
997 
998 		ok = drbd_send_ack(mdev, P_NEG_RS_DREPLY, e);
999 
1000 		/* update resync data with failure */
1001 		drbd_rs_failed_io(mdev, e->sector, e->size);
1002 	}
1003 
1004 	dec_unacked(mdev);
1005 
1006 	move_to_net_ee_or_free(mdev, e);
1007 
1008 	if (unlikely(!ok))
1009 		dev_err(DEV, "drbd_send_block() failed\n");
1010 	return ok;
1011 }
1012 
w_e_end_csum_rs_req(struct drbd_conf * mdev,struct drbd_work * w,int cancel)1013 int w_e_end_csum_rs_req(struct drbd_conf *mdev, struct drbd_work *w, int cancel)
1014 {
1015 	struct drbd_epoch_entry *e = container_of(w, struct drbd_epoch_entry, w);
1016 	struct digest_info *di;
1017 	int digest_size;
1018 	void *digest = NULL;
1019 	int ok, eq = 0;
1020 
1021 	if (unlikely(cancel)) {
1022 		drbd_free_ee(mdev, e);
1023 		dec_unacked(mdev);
1024 		return 1;
1025 	}
1026 
1027 	if (get_ldev(mdev)) {
1028 		drbd_rs_complete_io(mdev, e->sector);
1029 		put_ldev(mdev);
1030 	}
1031 
1032 	di = e->digest;
1033 
1034 	if (likely((e->flags & EE_WAS_ERROR) == 0)) {
1035 		/* quick hack to try to avoid a race against reconfiguration.
1036 		 * a real fix would be much more involved,
1037 		 * introducing more locking mechanisms */
1038 		if (mdev->csums_tfm) {
1039 			digest_size = crypto_hash_digestsize(mdev->csums_tfm);
1040 			D_ASSERT(digest_size == di->digest_size);
1041 			digest = kmalloc(digest_size, GFP_NOIO);
1042 		}
1043 		if (digest) {
1044 			drbd_csum_ee(mdev, mdev->csums_tfm, e, digest);
1045 			eq = !memcmp(digest, di->digest, digest_size);
1046 			kfree(digest);
1047 		}
1048 
1049 		if (eq) {
1050 			drbd_set_in_sync(mdev, e->sector, e->size);
1051 			/* rs_same_csums unit is BM_BLOCK_SIZE */
1052 			mdev->rs_same_csum += e->size >> BM_BLOCK_SHIFT;
1053 			ok = drbd_send_ack(mdev, P_RS_IS_IN_SYNC, e);
1054 		} else {
1055 			inc_rs_pending(mdev);
1056 			e->block_id = ID_SYNCER; /* By setting block_id, digest pointer becomes invalid! */
1057 			e->flags &= ~EE_HAS_DIGEST; /* This e no longer has a digest pointer */
1058 			kfree(di);
1059 			ok = drbd_send_block(mdev, P_RS_DATA_REPLY, e);
1060 		}
1061 	} else {
1062 		ok = drbd_send_ack(mdev, P_NEG_RS_DREPLY, e);
1063 		if (__ratelimit(&drbd_ratelimit_state))
1064 			dev_err(DEV, "Sending NegDReply. I guess it gets messy.\n");
1065 	}
1066 
1067 	dec_unacked(mdev);
1068 	move_to_net_ee_or_free(mdev, e);
1069 
1070 	if (unlikely(!ok))
1071 		dev_err(DEV, "drbd_send_block/ack() failed\n");
1072 	return ok;
1073 }
1074 
1075 /* TODO merge common code with w_e_send_csum */
w_e_end_ov_req(struct drbd_conf * mdev,struct drbd_work * w,int cancel)1076 int w_e_end_ov_req(struct drbd_conf *mdev, struct drbd_work *w, int cancel)
1077 {
1078 	struct drbd_epoch_entry *e = container_of(w, struct drbd_epoch_entry, w);
1079 	sector_t sector = e->sector;
1080 	unsigned int size = e->size;
1081 	int digest_size;
1082 	void *digest;
1083 	int ok = 1;
1084 
1085 	if (unlikely(cancel))
1086 		goto out;
1087 
1088 	digest_size = crypto_hash_digestsize(mdev->verify_tfm);
1089 	digest = kmalloc(digest_size, GFP_NOIO);
1090 	if (!digest) {
1091 		ok = 0;	/* terminate the connection in case the allocation failed */
1092 		goto out;
1093 	}
1094 
1095 	if (likely(!(e->flags & EE_WAS_ERROR)))
1096 		drbd_csum_ee(mdev, mdev->verify_tfm, e, digest);
1097 	else
1098 		memset(digest, 0, digest_size);
1099 
1100 	/* Free e and pages before send.
1101 	 * In case we block on congestion, we could otherwise run into
1102 	 * some distributed deadlock, if the other side blocks on
1103 	 * congestion as well, because our receiver blocks in
1104 	 * drbd_pp_alloc due to pp_in_use > max_buffers. */
1105 	drbd_free_ee(mdev, e);
1106 	e = NULL;
1107 	inc_rs_pending(mdev);
1108 	ok = drbd_send_drequest_csum(mdev, sector, size,
1109 				     digest, digest_size,
1110 				     P_OV_REPLY);
1111 	if (!ok)
1112 		dec_rs_pending(mdev);
1113 	kfree(digest);
1114 
1115 out:
1116 	if (e)
1117 		drbd_free_ee(mdev, e);
1118 	dec_unacked(mdev);
1119 	return ok;
1120 }
1121 
drbd_ov_oos_found(struct drbd_conf * mdev,sector_t sector,int size)1122 void drbd_ov_oos_found(struct drbd_conf *mdev, sector_t sector, int size)
1123 {
1124 	if (mdev->ov_last_oos_start + mdev->ov_last_oos_size == sector) {
1125 		mdev->ov_last_oos_size += size>>9;
1126 	} else {
1127 		mdev->ov_last_oos_start = sector;
1128 		mdev->ov_last_oos_size = size>>9;
1129 	}
1130 	drbd_set_out_of_sync(mdev, sector, size);
1131 }
1132 
w_e_end_ov_reply(struct drbd_conf * mdev,struct drbd_work * w,int cancel)1133 int w_e_end_ov_reply(struct drbd_conf *mdev, struct drbd_work *w, int cancel)
1134 {
1135 	struct drbd_epoch_entry *e = container_of(w, struct drbd_epoch_entry, w);
1136 	struct digest_info *di;
1137 	void *digest;
1138 	sector_t sector = e->sector;
1139 	unsigned int size = e->size;
1140 	int digest_size;
1141 	int ok, eq = 0;
1142 
1143 	if (unlikely(cancel)) {
1144 		drbd_free_ee(mdev, e);
1145 		dec_unacked(mdev);
1146 		return 1;
1147 	}
1148 
1149 	/* after "cancel", because after drbd_disconnect/drbd_rs_cancel_all
1150 	 * the resync lru has been cleaned up already */
1151 	if (get_ldev(mdev)) {
1152 		drbd_rs_complete_io(mdev, e->sector);
1153 		put_ldev(mdev);
1154 	}
1155 
1156 	di = e->digest;
1157 
1158 	if (likely((e->flags & EE_WAS_ERROR) == 0)) {
1159 		digest_size = crypto_hash_digestsize(mdev->verify_tfm);
1160 		digest = kmalloc(digest_size, GFP_NOIO);
1161 		if (digest) {
1162 			drbd_csum_ee(mdev, mdev->verify_tfm, e, digest);
1163 
1164 			D_ASSERT(digest_size == di->digest_size);
1165 			eq = !memcmp(digest, di->digest, digest_size);
1166 			kfree(digest);
1167 		}
1168 	}
1169 
1170 		/* Free e and pages before send.
1171 		 * In case we block on congestion, we could otherwise run into
1172 		 * some distributed deadlock, if the other side blocks on
1173 		 * congestion as well, because our receiver blocks in
1174 		 * drbd_pp_alloc due to pp_in_use > max_buffers. */
1175 	drbd_free_ee(mdev, e);
1176 	if (!eq)
1177 		drbd_ov_oos_found(mdev, sector, size);
1178 	else
1179 		ov_oos_print(mdev);
1180 
1181 	ok = drbd_send_ack_ex(mdev, P_OV_RESULT, sector, size,
1182 			      eq ? ID_IN_SYNC : ID_OUT_OF_SYNC);
1183 
1184 	dec_unacked(mdev);
1185 
1186 	--mdev->ov_left;
1187 
1188 	/* let's advance progress step marks only for every other megabyte */
1189 	if ((mdev->ov_left & 0x200) == 0x200)
1190 		drbd_advance_rs_marks(mdev, mdev->ov_left);
1191 
1192 	if (mdev->ov_left == 0) {
1193 		ov_oos_print(mdev);
1194 		drbd_resync_finished(mdev);
1195 	}
1196 
1197 	return ok;
1198 }
1199 
w_prev_work_done(struct drbd_conf * mdev,struct drbd_work * w,int cancel)1200 int w_prev_work_done(struct drbd_conf *mdev, struct drbd_work *w, int cancel)
1201 {
1202 	struct drbd_wq_barrier *b = container_of(w, struct drbd_wq_barrier, w);
1203 	complete(&b->done);
1204 	return 1;
1205 }
1206 
w_send_barrier(struct drbd_conf * mdev,struct drbd_work * w,int cancel)1207 int w_send_barrier(struct drbd_conf *mdev, struct drbd_work *w, int cancel)
1208 {
1209 	struct drbd_tl_epoch *b = container_of(w, struct drbd_tl_epoch, w);
1210 	struct p_barrier *p = &mdev->data.sbuf.barrier;
1211 	int ok = 1;
1212 
1213 	/* really avoid racing with tl_clear.  w.cb may have been referenced
1214 	 * just before it was reassigned and re-queued, so double check that.
1215 	 * actually, this race was harmless, since we only try to send the
1216 	 * barrier packet here, and otherwise do nothing with the object.
1217 	 * but compare with the head of w_clear_epoch */
1218 	spin_lock_irq(&mdev->req_lock);
1219 	if (w->cb != w_send_barrier || mdev->state.conn < C_CONNECTED)
1220 		cancel = 1;
1221 	spin_unlock_irq(&mdev->req_lock);
1222 	if (cancel)
1223 		return 1;
1224 
1225 	if (!drbd_get_data_sock(mdev))
1226 		return 0;
1227 	p->barrier = b->br_number;
1228 	/* inc_ap_pending was done where this was queued.
1229 	 * dec_ap_pending will be done in got_BarrierAck
1230 	 * or (on connection loss) in w_clear_epoch.  */
1231 	ok = _drbd_send_cmd(mdev, mdev->data.socket, P_BARRIER,
1232 				(struct p_header80 *)p, sizeof(*p), 0);
1233 	drbd_put_data_sock(mdev);
1234 
1235 	return ok;
1236 }
1237 
w_send_write_hint(struct drbd_conf * mdev,struct drbd_work * w,int cancel)1238 int w_send_write_hint(struct drbd_conf *mdev, struct drbd_work *w, int cancel)
1239 {
1240 	if (cancel)
1241 		return 1;
1242 	return drbd_send_short_cmd(mdev, P_UNPLUG_REMOTE);
1243 }
1244 
w_send_oos(struct drbd_conf * mdev,struct drbd_work * w,int cancel)1245 int w_send_oos(struct drbd_conf *mdev, struct drbd_work *w, int cancel)
1246 {
1247 	struct drbd_request *req = container_of(w, struct drbd_request, w);
1248 	int ok;
1249 
1250 	if (unlikely(cancel)) {
1251 		req_mod(req, send_canceled);
1252 		return 1;
1253 	}
1254 
1255 	ok = drbd_send_oos(mdev, req);
1256 	req_mod(req, oos_handed_to_network);
1257 
1258 	return ok;
1259 }
1260 
1261 /**
1262  * w_send_dblock() - Worker callback to send a P_DATA packet in order to mirror a write request
1263  * @mdev:	DRBD device.
1264  * @w:		work object.
1265  * @cancel:	The connection will be closed anyways
1266  */
w_send_dblock(struct drbd_conf * mdev,struct drbd_work * w,int cancel)1267 int w_send_dblock(struct drbd_conf *mdev, struct drbd_work *w, int cancel)
1268 {
1269 	struct drbd_request *req = container_of(w, struct drbd_request, w);
1270 	int ok;
1271 
1272 	if (unlikely(cancel)) {
1273 		req_mod(req, send_canceled);
1274 		return 1;
1275 	}
1276 
1277 	ok = drbd_send_dblock(mdev, req);
1278 	req_mod(req, ok ? handed_over_to_network : send_failed);
1279 
1280 	return ok;
1281 }
1282 
1283 /**
1284  * w_send_read_req() - Worker callback to send a read request (P_DATA_REQUEST) packet
1285  * @mdev:	DRBD device.
1286  * @w:		work object.
1287  * @cancel:	The connection will be closed anyways
1288  */
w_send_read_req(struct drbd_conf * mdev,struct drbd_work * w,int cancel)1289 int w_send_read_req(struct drbd_conf *mdev, struct drbd_work *w, int cancel)
1290 {
1291 	struct drbd_request *req = container_of(w, struct drbd_request, w);
1292 	int ok;
1293 
1294 	if (unlikely(cancel)) {
1295 		req_mod(req, send_canceled);
1296 		return 1;
1297 	}
1298 
1299 	ok = drbd_send_drequest(mdev, P_DATA_REQUEST, req->sector, req->size,
1300 				(unsigned long)req);
1301 
1302 	if (!ok) {
1303 		/* ?? we set C_TIMEOUT or C_BROKEN_PIPE in drbd_send();
1304 		 * so this is probably redundant */
1305 		if (mdev->state.conn >= C_CONNECTED)
1306 			drbd_force_state(mdev, NS(conn, C_NETWORK_FAILURE));
1307 	}
1308 	req_mod(req, ok ? handed_over_to_network : send_failed);
1309 
1310 	return ok;
1311 }
1312 
w_restart_disk_io(struct drbd_conf * mdev,struct drbd_work * w,int cancel)1313 int w_restart_disk_io(struct drbd_conf *mdev, struct drbd_work *w, int cancel)
1314 {
1315 	struct drbd_request *req = container_of(w, struct drbd_request, w);
1316 
1317 	if (bio_data_dir(req->master_bio) == WRITE && req->rq_state & RQ_IN_ACT_LOG)
1318 		drbd_al_begin_io(mdev, req->sector);
1319 	/* Calling drbd_al_begin_io() out of the worker might deadlocks
1320 	   theoretically. Practically it can not deadlock, since this is
1321 	   only used when unfreezing IOs. All the extents of the requests
1322 	   that made it into the TL are already active */
1323 
1324 	drbd_req_make_private_bio(req, req->master_bio);
1325 	req->private_bio->bi_bdev = mdev->ldev->backing_bdev;
1326 	generic_make_request(req->private_bio);
1327 
1328 	return 1;
1329 }
1330 
_drbd_may_sync_now(struct drbd_conf * mdev)1331 static int _drbd_may_sync_now(struct drbd_conf *mdev)
1332 {
1333 	struct drbd_conf *odev = mdev;
1334 
1335 	while (1) {
1336 		if (odev->sync_conf.after == -1)
1337 			return 1;
1338 		odev = minor_to_mdev(odev->sync_conf.after);
1339 		ERR_IF(!odev) return 1;
1340 		if ((odev->state.conn >= C_SYNC_SOURCE &&
1341 		     odev->state.conn <= C_PAUSED_SYNC_T) ||
1342 		    odev->state.aftr_isp || odev->state.peer_isp ||
1343 		    odev->state.user_isp)
1344 			return 0;
1345 	}
1346 }
1347 
1348 /**
1349  * _drbd_pause_after() - Pause resync on all devices that may not resync now
1350  * @mdev:	DRBD device.
1351  *
1352  * Called from process context only (admin command and after_state_ch).
1353  */
_drbd_pause_after(struct drbd_conf * mdev)1354 static int _drbd_pause_after(struct drbd_conf *mdev)
1355 {
1356 	struct drbd_conf *odev;
1357 	int i, rv = 0;
1358 
1359 	for (i = 0; i < minor_count; i++) {
1360 		odev = minor_to_mdev(i);
1361 		if (!odev)
1362 			continue;
1363 		if (odev->state.conn == C_STANDALONE && odev->state.disk == D_DISKLESS)
1364 			continue;
1365 		if (!_drbd_may_sync_now(odev))
1366 			rv |= (__drbd_set_state(_NS(odev, aftr_isp, 1), CS_HARD, NULL)
1367 			       != SS_NOTHING_TO_DO);
1368 	}
1369 
1370 	return rv;
1371 }
1372 
1373 /**
1374  * _drbd_resume_next() - Resume resync on all devices that may resync now
1375  * @mdev:	DRBD device.
1376  *
1377  * Called from process context only (admin command and worker).
1378  */
_drbd_resume_next(struct drbd_conf * mdev)1379 static int _drbd_resume_next(struct drbd_conf *mdev)
1380 {
1381 	struct drbd_conf *odev;
1382 	int i, rv = 0;
1383 
1384 	for (i = 0; i < minor_count; i++) {
1385 		odev = minor_to_mdev(i);
1386 		if (!odev)
1387 			continue;
1388 		if (odev->state.conn == C_STANDALONE && odev->state.disk == D_DISKLESS)
1389 			continue;
1390 		if (odev->state.aftr_isp) {
1391 			if (_drbd_may_sync_now(odev))
1392 				rv |= (__drbd_set_state(_NS(odev, aftr_isp, 0),
1393 							CS_HARD, NULL)
1394 				       != SS_NOTHING_TO_DO) ;
1395 		}
1396 	}
1397 	return rv;
1398 }
1399 
resume_next_sg(struct drbd_conf * mdev)1400 void resume_next_sg(struct drbd_conf *mdev)
1401 {
1402 	write_lock_irq(&global_state_lock);
1403 	_drbd_resume_next(mdev);
1404 	write_unlock_irq(&global_state_lock);
1405 }
1406 
suspend_other_sg(struct drbd_conf * mdev)1407 void suspend_other_sg(struct drbd_conf *mdev)
1408 {
1409 	write_lock_irq(&global_state_lock);
1410 	_drbd_pause_after(mdev);
1411 	write_unlock_irq(&global_state_lock);
1412 }
1413 
sync_after_error(struct drbd_conf * mdev,int o_minor)1414 static int sync_after_error(struct drbd_conf *mdev, int o_minor)
1415 {
1416 	struct drbd_conf *odev;
1417 
1418 	if (o_minor == -1)
1419 		return NO_ERROR;
1420 	if (o_minor < -1 || minor_to_mdev(o_minor) == NULL)
1421 		return ERR_SYNC_AFTER;
1422 
1423 	/* check for loops */
1424 	odev = minor_to_mdev(o_minor);
1425 	while (1) {
1426 		if (odev == mdev)
1427 			return ERR_SYNC_AFTER_CYCLE;
1428 
1429 		/* dependency chain ends here, no cycles. */
1430 		if (odev->sync_conf.after == -1)
1431 			return NO_ERROR;
1432 
1433 		/* follow the dependency chain */
1434 		odev = minor_to_mdev(odev->sync_conf.after);
1435 	}
1436 }
1437 
drbd_alter_sa(struct drbd_conf * mdev,int na)1438 int drbd_alter_sa(struct drbd_conf *mdev, int na)
1439 {
1440 	int changes;
1441 	int retcode;
1442 
1443 	write_lock_irq(&global_state_lock);
1444 	retcode = sync_after_error(mdev, na);
1445 	if (retcode == NO_ERROR) {
1446 		mdev->sync_conf.after = na;
1447 		do {
1448 			changes  = _drbd_pause_after(mdev);
1449 			changes |= _drbd_resume_next(mdev);
1450 		} while (changes);
1451 	}
1452 	write_unlock_irq(&global_state_lock);
1453 	return retcode;
1454 }
1455 
drbd_rs_controller_reset(struct drbd_conf * mdev)1456 void drbd_rs_controller_reset(struct drbd_conf *mdev)
1457 {
1458 	atomic_set(&mdev->rs_sect_in, 0);
1459 	atomic_set(&mdev->rs_sect_ev, 0);
1460 	mdev->rs_in_flight = 0;
1461 	mdev->rs_planed = 0;
1462 	spin_lock(&mdev->peer_seq_lock);
1463 	fifo_set(&mdev->rs_plan_s, 0);
1464 	spin_unlock(&mdev->peer_seq_lock);
1465 }
1466 
1467 /**
1468  * drbd_start_resync() - Start the resync process
1469  * @mdev:	DRBD device.
1470  * @side:	Either C_SYNC_SOURCE or C_SYNC_TARGET
1471  *
1472  * This function might bring you directly into one of the
1473  * C_PAUSED_SYNC_* states.
1474  */
drbd_start_resync(struct drbd_conf * mdev,enum drbd_conns side)1475 void drbd_start_resync(struct drbd_conf *mdev, enum drbd_conns side)
1476 {
1477 	union drbd_state ns;
1478 	int r;
1479 
1480 	if (mdev->state.conn >= C_SYNC_SOURCE && mdev->state.conn < C_AHEAD) {
1481 		dev_err(DEV, "Resync already running!\n");
1482 		return;
1483 	}
1484 
1485 	if (mdev->state.conn < C_AHEAD) {
1486 		/* In case a previous resync run was aborted by an IO error/detach on the peer. */
1487 		drbd_rs_cancel_all(mdev);
1488 		/* This should be done when we abort the resync. We definitely do not
1489 		   want to have this for connections going back and forth between
1490 		   Ahead/Behind and SyncSource/SyncTarget */
1491 	}
1492 
1493 	if (side == C_SYNC_TARGET) {
1494 		/* Since application IO was locked out during C_WF_BITMAP_T and
1495 		   C_WF_SYNC_UUID we are still unmodified. Before going to C_SYNC_TARGET
1496 		   we check that we might make the data inconsistent. */
1497 		r = drbd_khelper(mdev, "before-resync-target");
1498 		r = (r >> 8) & 0xff;
1499 		if (r > 0) {
1500 			dev_info(DEV, "before-resync-target handler returned %d, "
1501 			     "dropping connection.\n", r);
1502 			drbd_force_state(mdev, NS(conn, C_DISCONNECTING));
1503 			return;
1504 		}
1505 	} else /* C_SYNC_SOURCE */ {
1506 		r = drbd_khelper(mdev, "before-resync-source");
1507 		r = (r >> 8) & 0xff;
1508 		if (r > 0) {
1509 			if (r == 3) {
1510 				dev_info(DEV, "before-resync-source handler returned %d, "
1511 					 "ignoring. Old userland tools?", r);
1512 			} else {
1513 				dev_info(DEV, "before-resync-source handler returned %d, "
1514 					 "dropping connection.\n", r);
1515 				drbd_force_state(mdev, NS(conn, C_DISCONNECTING));
1516 				return;
1517 			}
1518 		}
1519 	}
1520 
1521 	drbd_state_lock(mdev);
1522 
1523 	if (!get_ldev_if_state(mdev, D_NEGOTIATING)) {
1524 		drbd_state_unlock(mdev);
1525 		return;
1526 	}
1527 
1528 	write_lock_irq(&global_state_lock);
1529 	ns = mdev->state;
1530 
1531 	ns.aftr_isp = !_drbd_may_sync_now(mdev);
1532 
1533 	ns.conn = side;
1534 
1535 	if (side == C_SYNC_TARGET)
1536 		ns.disk = D_INCONSISTENT;
1537 	else /* side == C_SYNC_SOURCE */
1538 		ns.pdsk = D_INCONSISTENT;
1539 
1540 	r = __drbd_set_state(mdev, ns, CS_VERBOSE, NULL);
1541 	ns = mdev->state;
1542 
1543 	if (ns.conn < C_CONNECTED)
1544 		r = SS_UNKNOWN_ERROR;
1545 
1546 	if (r == SS_SUCCESS) {
1547 		unsigned long tw = drbd_bm_total_weight(mdev);
1548 		unsigned long now = jiffies;
1549 		int i;
1550 
1551 		mdev->rs_failed    = 0;
1552 		mdev->rs_paused    = 0;
1553 		mdev->rs_same_csum = 0;
1554 		mdev->rs_last_events = 0;
1555 		mdev->rs_last_sect_ev = 0;
1556 		mdev->rs_total     = tw;
1557 		mdev->rs_start     = now;
1558 		for (i = 0; i < DRBD_SYNC_MARKS; i++) {
1559 			mdev->rs_mark_left[i] = tw;
1560 			mdev->rs_mark_time[i] = now;
1561 		}
1562 		_drbd_pause_after(mdev);
1563 	}
1564 	write_unlock_irq(&global_state_lock);
1565 
1566 	if (r == SS_SUCCESS) {
1567 		dev_info(DEV, "Began resync as %s (will sync %lu KB [%lu bits set]).\n",
1568 		     drbd_conn_str(ns.conn),
1569 		     (unsigned long) mdev->rs_total << (BM_BLOCK_SHIFT-10),
1570 		     (unsigned long) mdev->rs_total);
1571 		if (side == C_SYNC_TARGET)
1572 			mdev->bm_resync_fo = 0;
1573 
1574 		/* Since protocol 96, we must serialize drbd_gen_and_send_sync_uuid
1575 		 * with w_send_oos, or the sync target will get confused as to
1576 		 * how much bits to resync.  We cannot do that always, because for an
1577 		 * empty resync and protocol < 95, we need to do it here, as we call
1578 		 * drbd_resync_finished from here in that case.
1579 		 * We drbd_gen_and_send_sync_uuid here for protocol < 96,
1580 		 * and from after_state_ch otherwise. */
1581 		if (side == C_SYNC_SOURCE && mdev->agreed_pro_version < 96)
1582 			drbd_gen_and_send_sync_uuid(mdev);
1583 
1584 		if (mdev->agreed_pro_version < 95 && mdev->rs_total == 0) {
1585 			/* This still has a race (about when exactly the peers
1586 			 * detect connection loss) that can lead to a full sync
1587 			 * on next handshake. In 8.3.9 we fixed this with explicit
1588 			 * resync-finished notifications, but the fix
1589 			 * introduces a protocol change.  Sleeping for some
1590 			 * time longer than the ping interval + timeout on the
1591 			 * SyncSource, to give the SyncTarget the chance to
1592 			 * detect connection loss, then waiting for a ping
1593 			 * response (implicit in drbd_resync_finished) reduces
1594 			 * the race considerably, but does not solve it. */
1595 			if (side == C_SYNC_SOURCE)
1596 				schedule_timeout_interruptible(
1597 					mdev->net_conf->ping_int * HZ +
1598 					mdev->net_conf->ping_timeo*HZ/9);
1599 			drbd_resync_finished(mdev);
1600 		}
1601 
1602 		drbd_rs_controller_reset(mdev);
1603 		/* ns.conn may already be != mdev->state.conn,
1604 		 * we may have been paused in between, or become paused until
1605 		 * the timer triggers.
1606 		 * No matter, that is handled in resync_timer_fn() */
1607 		if (ns.conn == C_SYNC_TARGET)
1608 			mod_timer(&mdev->resync_timer, jiffies);
1609 
1610 		drbd_md_sync(mdev);
1611 	}
1612 	put_ldev(mdev);
1613 	drbd_state_unlock(mdev);
1614 }
1615 
drbd_worker(struct drbd_thread * thi)1616 int drbd_worker(struct drbd_thread *thi)
1617 {
1618 	struct drbd_conf *mdev = thi->mdev;
1619 	struct drbd_work *w = NULL;
1620 	LIST_HEAD(work_list);
1621 	int intr = 0, i;
1622 
1623 	sprintf(current->comm, "drbd%d_worker", mdev_to_minor(mdev));
1624 
1625 	while (get_t_state(thi) == Running) {
1626 		drbd_thread_current_set_cpu(mdev);
1627 
1628 		if (down_trylock(&mdev->data.work.s)) {
1629 			mutex_lock(&mdev->data.mutex);
1630 			if (mdev->data.socket && !mdev->net_conf->no_cork)
1631 				drbd_tcp_uncork(mdev->data.socket);
1632 			mutex_unlock(&mdev->data.mutex);
1633 
1634 			intr = down_interruptible(&mdev->data.work.s);
1635 
1636 			mutex_lock(&mdev->data.mutex);
1637 			if (mdev->data.socket  && !mdev->net_conf->no_cork)
1638 				drbd_tcp_cork(mdev->data.socket);
1639 			mutex_unlock(&mdev->data.mutex);
1640 		}
1641 
1642 		if (intr) {
1643 			D_ASSERT(intr == -EINTR);
1644 			flush_signals(current);
1645 			ERR_IF (get_t_state(thi) == Running)
1646 				continue;
1647 			break;
1648 		}
1649 
1650 		if (get_t_state(thi) != Running)
1651 			break;
1652 		/* With this break, we have done a down() but not consumed
1653 		   the entry from the list. The cleanup code takes care of
1654 		   this...   */
1655 
1656 		w = NULL;
1657 		spin_lock_irq(&mdev->data.work.q_lock);
1658 		ERR_IF(list_empty(&mdev->data.work.q)) {
1659 			/* something terribly wrong in our logic.
1660 			 * we were able to down() the semaphore,
1661 			 * but the list is empty... doh.
1662 			 *
1663 			 * what is the best thing to do now?
1664 			 * try again from scratch, restarting the receiver,
1665 			 * asender, whatnot? could break even more ugly,
1666 			 * e.g. when we are primary, but no good local data.
1667 			 *
1668 			 * I'll try to get away just starting over this loop.
1669 			 */
1670 			spin_unlock_irq(&mdev->data.work.q_lock);
1671 			continue;
1672 		}
1673 		w = list_entry(mdev->data.work.q.next, struct drbd_work, list);
1674 		list_del_init(&w->list);
1675 		spin_unlock_irq(&mdev->data.work.q_lock);
1676 
1677 		if (!w->cb(mdev, w, mdev->state.conn < C_CONNECTED)) {
1678 			/* dev_warn(DEV, "worker: a callback failed! \n"); */
1679 			if (mdev->state.conn >= C_CONNECTED)
1680 				drbd_force_state(mdev,
1681 						NS(conn, C_NETWORK_FAILURE));
1682 		}
1683 	}
1684 	D_ASSERT(test_bit(DEVICE_DYING, &mdev->flags));
1685 	D_ASSERT(test_bit(CONFIG_PENDING, &mdev->flags));
1686 
1687 	spin_lock_irq(&mdev->data.work.q_lock);
1688 	i = 0;
1689 	while (!list_empty(&mdev->data.work.q)) {
1690 		list_splice_init(&mdev->data.work.q, &work_list);
1691 		spin_unlock_irq(&mdev->data.work.q_lock);
1692 
1693 		while (!list_empty(&work_list)) {
1694 			w = list_entry(work_list.next, struct drbd_work, list);
1695 			list_del_init(&w->list);
1696 			w->cb(mdev, w, 1);
1697 			i++; /* dead debugging code */
1698 		}
1699 
1700 		spin_lock_irq(&mdev->data.work.q_lock);
1701 	}
1702 	sema_init(&mdev->data.work.s, 0);
1703 	/* DANGEROUS race: if someone did queue his work within the spinlock,
1704 	 * but up() ed outside the spinlock, we could get an up() on the
1705 	 * semaphore without corresponding list entry.
1706 	 * So don't do that.
1707 	 */
1708 	spin_unlock_irq(&mdev->data.work.q_lock);
1709 
1710 	D_ASSERT(mdev->state.disk == D_DISKLESS && mdev->state.conn == C_STANDALONE);
1711 	/* _drbd_set_state only uses stop_nowait.
1712 	 * wait here for the Exiting receiver. */
1713 	drbd_thread_stop(&mdev->receiver);
1714 	drbd_mdev_cleanup(mdev);
1715 
1716 	dev_info(DEV, "worker terminated\n");
1717 
1718 	clear_bit(DEVICE_DYING, &mdev->flags);
1719 	clear_bit(CONFIG_PENDING, &mdev->flags);
1720 	wake_up(&mdev->state_wait);
1721 
1722 	return 0;
1723 }
1724