1// SPDX-License-Identifier: GPL-2.0-only 2/* 3 drbd_req.c 4 5 This file is part of DRBD by Philipp Reisner and Lars Ellenberg. 6 7 Copyright (C) 2001-2008, LINBIT Information Technologies GmbH. 8 Copyright (C) 1999-2008, Philipp Reisner <philipp.reisner@linbit.com>. 9 Copyright (C) 2002-2008, Lars Ellenberg <lars.ellenberg@linbit.com>. 10 11 12 */ 13 14#include <linux/module.h> 15 16#include <linux/slab.h> 17#include <linux/drbd.h> 18#include "drbd_int.h" 19#include "drbd_req.h" 20 21 22static bool drbd_may_do_local_read(struct drbd_device *device, sector_t sector, int size); 23 24static struct drbd_request *drbd_req_new(struct drbd_device *device, struct bio *bio_src) 25{ 26 struct drbd_request *req; 27 28 req = mempool_alloc(&drbd_request_mempool, GFP_NOIO); 29 if (!req) 30 return NULL; 31 memset(req, 0, sizeof(*req)); 32 33 req->rq_state = (bio_data_dir(bio_src) == WRITE ? RQ_WRITE : 0) 34 | (bio_op(bio_src) == REQ_OP_WRITE_ZEROES ? RQ_ZEROES : 0) 35 | (bio_op(bio_src) == REQ_OP_DISCARD ? RQ_UNMAP : 0); 36 req->device = device; 37 req->master_bio = bio_src; 38 req->epoch = 0; 39 40 drbd_clear_interval(&req->i); 41 req->i.sector = bio_src->bi_iter.bi_sector; 42 req->i.size = bio_src->bi_iter.bi_size; 43 req->i.local = true; 44 req->i.waiting = false; 45 46 INIT_LIST_HEAD(&req->tl_requests); 47 INIT_LIST_HEAD(&req->w.list); 48 INIT_LIST_HEAD(&req->req_pending_master_completion); 49 INIT_LIST_HEAD(&req->req_pending_local); 50 51 /* one reference to be put by __drbd_make_request */ 52 atomic_set(&req->completion_ref, 1); 53 /* one kref as long as completion_ref > 0 */ 54 kref_init(&req->kref); 55 return req; 56} 57 58static void drbd_remove_request_interval(struct rb_root *root, 59 struct drbd_request *req) 60{ 61 struct drbd_device *device = req->device; 62 struct drbd_interval *i = &req->i; 63 64 drbd_remove_interval(root, i); 65 66 /* Wake up any processes waiting for this request to complete. */ 67 if (i->waiting) 68 wake_up(&device->misc_wait); 69} 70 71void drbd_req_destroy(struct kref *kref) 72{ 73 struct drbd_request *req = container_of(kref, struct drbd_request, kref); 74 struct drbd_device *device = req->device; 75 const unsigned s = req->rq_state; 76 77 if ((req->master_bio && !(s & RQ_POSTPONED)) || 78 atomic_read(&req->completion_ref) || 79 (s & RQ_LOCAL_PENDING) || 80 ((s & RQ_NET_MASK) && !(s & RQ_NET_DONE))) { 81 drbd_err(device, "drbd_req_destroy: Logic BUG rq_state = 0x%x, completion_ref = %d\n", 82 s, atomic_read(&req->completion_ref)); 83 return; 84 } 85 86 /* If called from mod_rq_state (expected normal case) or 87 * drbd_send_and_submit (the less likely normal path), this holds the 88 * req_lock, and req->tl_requests will typicaly be on ->transfer_log, 89 * though it may be still empty (never added to the transfer log). 90 * 91 * If called from do_retry(), we do NOT hold the req_lock, but we are 92 * still allowed to unconditionally list_del(&req->tl_requests), 93 * because it will be on a local on-stack list only. */ 94 list_del_init(&req->tl_requests); 95 96 /* finally remove the request from the conflict detection 97 * respective block_id verification interval tree. */ 98 if (!drbd_interval_empty(&req->i)) { 99 struct rb_root *root; 100 101 if (s & RQ_WRITE) 102 root = &device->write_requests; 103 else 104 root = &device->read_requests; 105 drbd_remove_request_interval(root, req); 106 } else if (s & (RQ_NET_MASK & ~RQ_NET_DONE) && req->i.size != 0) 107 drbd_err(device, "drbd_req_destroy: Logic BUG: interval empty, but: rq_state=0x%x, sect=%llu, size=%u\n", 108 s, (unsigned long long)req->i.sector, req->i.size); 109 110 /* if it was a write, we may have to set the corresponding 111 * bit(s) out-of-sync first. If it had a local part, we need to 112 * release the reference to the activity log. */ 113 if (s & RQ_WRITE) { 114 /* Set out-of-sync unless both OK flags are set 115 * (local only or remote failed). 116 * Other places where we set out-of-sync: 117 * READ with local io-error */ 118 119 /* There is a special case: 120 * we may notice late that IO was suspended, 121 * and postpone, or schedule for retry, a write, 122 * before it even was submitted or sent. 123 * In that case we do not want to touch the bitmap at all. 124 */ 125 struct drbd_peer_device *peer_device = first_peer_device(device); 126 if ((s & (RQ_POSTPONED|RQ_LOCAL_MASK|RQ_NET_MASK)) != RQ_POSTPONED) { 127 if (!(s & RQ_NET_OK) || !(s & RQ_LOCAL_OK)) 128 drbd_set_out_of_sync(peer_device, req->i.sector, req->i.size); 129 130 if ((s & RQ_NET_OK) && (s & RQ_LOCAL_OK) && (s & RQ_NET_SIS)) 131 drbd_set_in_sync(peer_device, req->i.sector, req->i.size); 132 } 133 134 /* one might be tempted to move the drbd_al_complete_io 135 * to the local io completion callback drbd_request_endio. 136 * but, if this was a mirror write, we may only 137 * drbd_al_complete_io after this is RQ_NET_DONE, 138 * otherwise the extent could be dropped from the al 139 * before it has actually been written on the peer. 140 * if we crash before our peer knows about the request, 141 * but after the extent has been dropped from the al, 142 * we would forget to resync the corresponding extent. 143 */ 144 if (s & RQ_IN_ACT_LOG) { 145 if (get_ldev_if_state(device, D_FAILED)) { 146 drbd_al_complete_io(device, &req->i); 147 put_ldev(device); 148 } else if (drbd_ratelimit()) { 149 drbd_warn(device, "Should have called drbd_al_complete_io(, %llu, %u), " 150 "but my Disk seems to have failed :(\n", 151 (unsigned long long) req->i.sector, req->i.size); 152 } 153 } 154 } 155 156 mempool_free(req, &drbd_request_mempool); 157} 158 159static void wake_all_senders(struct drbd_connection *connection) 160{ 161 wake_up(&connection->sender_work.q_wait); 162} 163 164/* must hold resource->req_lock */ 165void start_new_tl_epoch(struct drbd_connection *connection) 166{ 167 /* no point closing an epoch, if it is empty, anyways. */ 168 if (connection->current_tle_writes == 0) 169 return; 170 171 connection->current_tle_writes = 0; 172 atomic_inc(&connection->current_tle_nr); 173 wake_all_senders(connection); 174} 175 176void complete_master_bio(struct drbd_device *device, 177 struct bio_and_error *m) 178{ 179 if (unlikely(m->error)) 180 m->bio->bi_status = errno_to_blk_status(m->error); 181 bio_endio(m->bio); 182 dec_ap_bio(device); 183} 184 185 186/* Helper for __req_mod(). 187 * Set m->bio to the master bio, if it is fit to be completed, 188 * or leave it alone (it is initialized to NULL in __req_mod), 189 * if it has already been completed, or cannot be completed yet. 190 * If m->bio is set, the error status to be returned is placed in m->error. 191 */ 192static 193void drbd_req_complete(struct drbd_request *req, struct bio_and_error *m) 194{ 195 const unsigned s = req->rq_state; 196 struct drbd_device *device = req->device; 197 int error, ok; 198 199 /* we must not complete the master bio, while it is 200 * still being processed by _drbd_send_zc_bio (drbd_send_dblock) 201 * not yet acknowledged by the peer 202 * not yet completed by the local io subsystem 203 * these flags may get cleared in any order by 204 * the worker, 205 * the receiver, 206 * the bio_endio completion callbacks. 207 */ 208 if ((s & RQ_LOCAL_PENDING && !(s & RQ_LOCAL_ABORTED)) || 209 (s & RQ_NET_QUEUED) || (s & RQ_NET_PENDING) || 210 (s & RQ_COMPLETION_SUSP)) { 211 drbd_err(device, "drbd_req_complete: Logic BUG rq_state = 0x%x\n", s); 212 return; 213 } 214 215 if (!req->master_bio) { 216 drbd_err(device, "drbd_req_complete: Logic BUG, master_bio == NULL!\n"); 217 return; 218 } 219 220 /* 221 * figure out whether to report success or failure. 222 * 223 * report success when at least one of the operations succeeded. 224 * or, to put the other way, 225 * only report failure, when both operations failed. 226 * 227 * what to do about the failures is handled elsewhere. 228 * what we need to do here is just: complete the master_bio. 229 * 230 * local completion error, if any, has been stored as ERR_PTR 231 * in private_bio within drbd_request_endio. 232 */ 233 ok = (s & RQ_LOCAL_OK) || (s & RQ_NET_OK); 234 error = PTR_ERR(req->private_bio); 235 236 /* Before we can signal completion to the upper layers, 237 * we may need to close the current transfer log epoch. 238 * We are within the request lock, so we can simply compare 239 * the request epoch number with the current transfer log 240 * epoch number. If they match, increase the current_tle_nr, 241 * and reset the transfer log epoch write_cnt. 242 */ 243 if (op_is_write(bio_op(req->master_bio)) && 244 req->epoch == atomic_read(&first_peer_device(device)->connection->current_tle_nr)) 245 start_new_tl_epoch(first_peer_device(device)->connection); 246 247 /* Update disk stats */ 248 bio_end_io_acct(req->master_bio, req->start_jif); 249 250 /* If READ failed, 251 * have it be pushed back to the retry work queue, 252 * so it will re-enter __drbd_make_request(), 253 * and be re-assigned to a suitable local or remote path, 254 * or failed if we do not have access to good data anymore. 255 * 256 * Unless it was failed early by __drbd_make_request(), 257 * because no path was available, in which case 258 * it was not even added to the transfer_log. 259 * 260 * read-ahead may fail, and will not be retried. 261 * 262 * WRITE should have used all available paths already. 263 */ 264 if (!ok && 265 bio_op(req->master_bio) == REQ_OP_READ && 266 !(req->master_bio->bi_opf & REQ_RAHEAD) && 267 !list_empty(&req->tl_requests)) 268 req->rq_state |= RQ_POSTPONED; 269 270 if (!(req->rq_state & RQ_POSTPONED)) { 271 m->error = ok ? 0 : (error ?: -EIO); 272 m->bio = req->master_bio; 273 req->master_bio = NULL; 274 /* We leave it in the tree, to be able to verify later 275 * write-acks in protocol != C during resync. 276 * But we mark it as "complete", so it won't be counted as 277 * conflict in a multi-primary setup. */ 278 req->i.completed = true; 279 } 280 281 if (req->i.waiting) 282 wake_up(&device->misc_wait); 283 284 /* Either we are about to complete to upper layers, 285 * or we will restart this request. 286 * In either case, the request object will be destroyed soon, 287 * so better remove it from all lists. */ 288 list_del_init(&req->req_pending_master_completion); 289} 290 291/* still holds resource->req_lock */ 292static void drbd_req_put_completion_ref(struct drbd_request *req, struct bio_and_error *m, int put) 293{ 294 struct drbd_device *device = req->device; 295 D_ASSERT(device, m || (req->rq_state & RQ_POSTPONED)); 296 297 if (!put) 298 return; 299 300 if (!atomic_sub_and_test(put, &req->completion_ref)) 301 return; 302 303 drbd_req_complete(req, m); 304 305 /* local completion may still come in later, 306 * we need to keep the req object around. */ 307 if (req->rq_state & RQ_LOCAL_ABORTED) 308 return; 309 310 if (req->rq_state & RQ_POSTPONED) { 311 /* don't destroy the req object just yet, 312 * but queue it for retry */ 313 drbd_restart_request(req); 314 return; 315 } 316 317 kref_put(&req->kref, drbd_req_destroy); 318} 319 320static void set_if_null_req_next(struct drbd_peer_device *peer_device, struct drbd_request *req) 321{ 322 struct drbd_connection *connection = peer_device ? peer_device->connection : NULL; 323 if (!connection) 324 return; 325 if (connection->req_next == NULL) 326 connection->req_next = req; 327} 328 329static void advance_conn_req_next(struct drbd_peer_device *peer_device, struct drbd_request *req) 330{ 331 struct drbd_connection *connection = peer_device ? peer_device->connection : NULL; 332 struct drbd_request *iter = req; 333 if (!connection) 334 return; 335 if (connection->req_next != req) 336 return; 337 338 req = NULL; 339 list_for_each_entry_continue(iter, &connection->transfer_log, tl_requests) { 340 const unsigned int s = iter->rq_state; 341 342 if (s & RQ_NET_QUEUED) { 343 req = iter; 344 break; 345 } 346 } 347 connection->req_next = req; 348} 349 350static void set_if_null_req_ack_pending(struct drbd_peer_device *peer_device, struct drbd_request *req) 351{ 352 struct drbd_connection *connection = peer_device ? peer_device->connection : NULL; 353 if (!connection) 354 return; 355 if (connection->req_ack_pending == NULL) 356 connection->req_ack_pending = req; 357} 358 359static void advance_conn_req_ack_pending(struct drbd_peer_device *peer_device, struct drbd_request *req) 360{ 361 struct drbd_connection *connection = peer_device ? peer_device->connection : NULL; 362 struct drbd_request *iter = req; 363 if (!connection) 364 return; 365 if (connection->req_ack_pending != req) 366 return; 367 368 req = NULL; 369 list_for_each_entry_continue(iter, &connection->transfer_log, tl_requests) { 370 const unsigned int s = iter->rq_state; 371 372 if ((s & RQ_NET_SENT) && (s & RQ_NET_PENDING)) { 373 req = iter; 374 break; 375 } 376 } 377 connection->req_ack_pending = req; 378} 379 380static void set_if_null_req_not_net_done(struct drbd_peer_device *peer_device, struct drbd_request *req) 381{ 382 struct drbd_connection *connection = peer_device ? peer_device->connection : NULL; 383 if (!connection) 384 return; 385 if (connection->req_not_net_done == NULL) 386 connection->req_not_net_done = req; 387} 388 389static void advance_conn_req_not_net_done(struct drbd_peer_device *peer_device, struct drbd_request *req) 390{ 391 struct drbd_connection *connection = peer_device ? peer_device->connection : NULL; 392 struct drbd_request *iter = req; 393 if (!connection) 394 return; 395 if (connection->req_not_net_done != req) 396 return; 397 398 req = NULL; 399 list_for_each_entry_continue(iter, &connection->transfer_log, tl_requests) { 400 const unsigned int s = iter->rq_state; 401 402 if ((s & RQ_NET_SENT) && !(s & RQ_NET_DONE)) { 403 req = iter; 404 break; 405 } 406 } 407 connection->req_not_net_done = req; 408} 409 410/* I'd like this to be the only place that manipulates 411 * req->completion_ref and req->kref. */ 412static void mod_rq_state(struct drbd_request *req, struct bio_and_error *m, 413 int clear, int set) 414{ 415 struct drbd_device *device = req->device; 416 struct drbd_peer_device *peer_device = first_peer_device(device); 417 unsigned s = req->rq_state; 418 int c_put = 0; 419 420 if (drbd_suspended(device) && !((s | clear) & RQ_COMPLETION_SUSP)) 421 set |= RQ_COMPLETION_SUSP; 422 423 /* apply */ 424 425 req->rq_state &= ~clear; 426 req->rq_state |= set; 427 428 /* no change? */ 429 if (req->rq_state == s) 430 return; 431 432 /* intent: get references */ 433 434 kref_get(&req->kref); 435 436 if (!(s & RQ_LOCAL_PENDING) && (set & RQ_LOCAL_PENDING)) 437 atomic_inc(&req->completion_ref); 438 439 if (!(s & RQ_NET_PENDING) && (set & RQ_NET_PENDING)) { 440 inc_ap_pending(device); 441 atomic_inc(&req->completion_ref); 442 } 443 444 if (!(s & RQ_NET_QUEUED) && (set & RQ_NET_QUEUED)) { 445 atomic_inc(&req->completion_ref); 446 set_if_null_req_next(peer_device, req); 447 } 448 449 if (!(s & RQ_EXP_BARR_ACK) && (set & RQ_EXP_BARR_ACK)) 450 kref_get(&req->kref); /* wait for the DONE */ 451 452 if (!(s & RQ_NET_SENT) && (set & RQ_NET_SENT)) { 453 /* potentially already completed in the ack_receiver thread */ 454 if (!(s & RQ_NET_DONE)) { 455 atomic_add(req->i.size >> 9, &device->ap_in_flight); 456 set_if_null_req_not_net_done(peer_device, req); 457 } 458 if (req->rq_state & RQ_NET_PENDING) 459 set_if_null_req_ack_pending(peer_device, req); 460 } 461 462 if (!(s & RQ_COMPLETION_SUSP) && (set & RQ_COMPLETION_SUSP)) 463 atomic_inc(&req->completion_ref); 464 465 /* progress: put references */ 466 467 if ((s & RQ_COMPLETION_SUSP) && (clear & RQ_COMPLETION_SUSP)) 468 ++c_put; 469 470 if (!(s & RQ_LOCAL_ABORTED) && (set & RQ_LOCAL_ABORTED)) { 471 D_ASSERT(device, req->rq_state & RQ_LOCAL_PENDING); 472 ++c_put; 473 } 474 475 if ((s & RQ_LOCAL_PENDING) && (clear & RQ_LOCAL_PENDING)) { 476 if (req->rq_state & RQ_LOCAL_ABORTED) 477 kref_put(&req->kref, drbd_req_destroy); 478 else 479 ++c_put; 480 list_del_init(&req->req_pending_local); 481 } 482 483 if ((s & RQ_NET_PENDING) && (clear & RQ_NET_PENDING)) { 484 dec_ap_pending(device); 485 ++c_put; 486 req->acked_jif = jiffies; 487 advance_conn_req_ack_pending(peer_device, req); 488 } 489 490 if ((s & RQ_NET_QUEUED) && (clear & RQ_NET_QUEUED)) { 491 ++c_put; 492 advance_conn_req_next(peer_device, req); 493 } 494 495 if (!(s & RQ_NET_DONE) && (set & RQ_NET_DONE)) { 496 if (s & RQ_NET_SENT) 497 atomic_sub(req->i.size >> 9, &device->ap_in_flight); 498 if (s & RQ_EXP_BARR_ACK) 499 kref_put(&req->kref, drbd_req_destroy); 500 req->net_done_jif = jiffies; 501 502 /* in ahead/behind mode, or just in case, 503 * before we finally destroy this request, 504 * the caching pointers must not reference it anymore */ 505 advance_conn_req_next(peer_device, req); 506 advance_conn_req_ack_pending(peer_device, req); 507 advance_conn_req_not_net_done(peer_device, req); 508 } 509 510 /* potentially complete and destroy */ 511 512 /* If we made progress, retry conflicting peer requests, if any. */ 513 if (req->i.waiting) 514 wake_up(&device->misc_wait); 515 516 drbd_req_put_completion_ref(req, m, c_put); 517 kref_put(&req->kref, drbd_req_destroy); 518} 519 520static void drbd_report_io_error(struct drbd_device *device, struct drbd_request *req) 521{ 522 if (!drbd_ratelimit()) 523 return; 524 525 drbd_warn(device, "local %s IO error sector %llu+%u on %pg\n", 526 (req->rq_state & RQ_WRITE) ? "WRITE" : "READ", 527 (unsigned long long)req->i.sector, 528 req->i.size >> 9, 529 device->ldev->backing_bdev); 530} 531 532/* Helper for HANDED_OVER_TO_NETWORK. 533 * Is this a protocol A write (neither WRITE_ACK nor RECEIVE_ACK expected)? 534 * Is it also still "PENDING"? 535 * --> If so, clear PENDING and set NET_OK below. 536 * If it is a protocol A write, but not RQ_PENDING anymore, neg-ack was faster 537 * (and we must not set RQ_NET_OK) */ 538static inline bool is_pending_write_protocol_A(struct drbd_request *req) 539{ 540 return (req->rq_state & 541 (RQ_WRITE|RQ_NET_PENDING|RQ_EXP_WRITE_ACK|RQ_EXP_RECEIVE_ACK)) 542 == (RQ_WRITE|RQ_NET_PENDING); 543} 544 545/* obviously this could be coded as many single functions 546 * instead of one huge switch, 547 * or by putting the code directly in the respective locations 548 * (as it has been before). 549 * 550 * but having it this way 551 * enforces that it is all in this one place, where it is easier to audit, 552 * it makes it obvious that whatever "event" "happens" to a request should 553 * happen "atomically" within the req_lock, 554 * and it enforces that we have to think in a very structured manner 555 * about the "events" that may happen to a request during its life time ... 556 * 557 * 558 * peer_device == NULL means local disk 559 */ 560int __req_mod(struct drbd_request *req, enum drbd_req_event what, 561 struct drbd_peer_device *peer_device, 562 struct bio_and_error *m) 563{ 564 struct drbd_device *const device = req->device; 565 struct drbd_connection *const connection = peer_device ? peer_device->connection : NULL; 566 struct net_conf *nc; 567 int p, rv = 0; 568 569 if (m) 570 m->bio = NULL; 571 572 switch (what) { 573 default: 574 drbd_err(device, "LOGIC BUG in %s:%u\n", __FILE__ , __LINE__); 575 break; 576 577 /* does not happen... 578 * initialization done in drbd_req_new 579 case CREATED: 580 break; 581 */ 582 583 case TO_BE_SENT: /* via network */ 584 /* reached via __drbd_make_request 585 * and from w_read_retry_remote */ 586 D_ASSERT(device, !(req->rq_state & RQ_NET_MASK)); 587 rcu_read_lock(); 588 nc = rcu_dereference(connection->net_conf); 589 p = nc->wire_protocol; 590 rcu_read_unlock(); 591 req->rq_state |= 592 p == DRBD_PROT_C ? RQ_EXP_WRITE_ACK : 593 p == DRBD_PROT_B ? RQ_EXP_RECEIVE_ACK : 0; 594 mod_rq_state(req, m, 0, RQ_NET_PENDING); 595 break; 596 597 case TO_BE_SUBMITTED: /* locally */ 598 /* reached via __drbd_make_request */ 599 D_ASSERT(device, !(req->rq_state & RQ_LOCAL_MASK)); 600 mod_rq_state(req, m, 0, RQ_LOCAL_PENDING); 601 break; 602 603 case COMPLETED_OK: 604 if (req->rq_state & RQ_WRITE) 605 device->writ_cnt += req->i.size >> 9; 606 else 607 device->read_cnt += req->i.size >> 9; 608 609 mod_rq_state(req, m, RQ_LOCAL_PENDING, 610 RQ_LOCAL_COMPLETED|RQ_LOCAL_OK); 611 break; 612 613 case ABORT_DISK_IO: 614 mod_rq_state(req, m, 0, RQ_LOCAL_ABORTED); 615 break; 616 617 case WRITE_COMPLETED_WITH_ERROR: 618 drbd_report_io_error(device, req); 619 __drbd_chk_io_error(device, DRBD_WRITE_ERROR); 620 mod_rq_state(req, m, RQ_LOCAL_PENDING, RQ_LOCAL_COMPLETED); 621 break; 622 623 case READ_COMPLETED_WITH_ERROR: 624 drbd_set_out_of_sync(peer_device, req->i.sector, req->i.size); 625 drbd_report_io_error(device, req); 626 __drbd_chk_io_error(device, DRBD_READ_ERROR); 627 fallthrough; 628 case READ_AHEAD_COMPLETED_WITH_ERROR: 629 /* it is legal to fail read-ahead, no __drbd_chk_io_error in that case. */ 630 mod_rq_state(req, m, RQ_LOCAL_PENDING, RQ_LOCAL_COMPLETED); 631 break; 632 633 case DISCARD_COMPLETED_NOTSUPP: 634 case DISCARD_COMPLETED_WITH_ERROR: 635 /* I'd rather not detach from local disk just because it 636 * failed a REQ_OP_DISCARD. */ 637 mod_rq_state(req, m, RQ_LOCAL_PENDING, RQ_LOCAL_COMPLETED); 638 break; 639 640 case QUEUE_FOR_NET_READ: 641 /* READ, and 642 * no local disk, 643 * or target area marked as invalid, 644 * or just got an io-error. */ 645 /* from __drbd_make_request 646 * or from bio_endio during read io-error recovery */ 647 648 /* So we can verify the handle in the answer packet. 649 * Corresponding drbd_remove_request_interval is in 650 * drbd_req_complete() */ 651 D_ASSERT(device, drbd_interval_empty(&req->i)); 652 drbd_insert_interval(&device->read_requests, &req->i); 653 654 set_bit(UNPLUG_REMOTE, &device->flags); 655 656 D_ASSERT(device, req->rq_state & RQ_NET_PENDING); 657 D_ASSERT(device, (req->rq_state & RQ_LOCAL_MASK) == 0); 658 mod_rq_state(req, m, 0, RQ_NET_QUEUED); 659 req->w.cb = w_send_read_req; 660 drbd_queue_work(&connection->sender_work, 661 &req->w); 662 break; 663 664 case QUEUE_FOR_NET_WRITE: 665 /* assert something? */ 666 /* from __drbd_make_request only */ 667 668 /* Corresponding drbd_remove_request_interval is in 669 * drbd_req_complete() */ 670 D_ASSERT(device, drbd_interval_empty(&req->i)); 671 drbd_insert_interval(&device->write_requests, &req->i); 672 673 /* NOTE 674 * In case the req ended up on the transfer log before being 675 * queued on the worker, it could lead to this request being 676 * missed during cleanup after connection loss. 677 * So we have to do both operations here, 678 * within the same lock that protects the transfer log. 679 * 680 * _req_add_to_epoch(req); this has to be after the 681 * _maybe_start_new_epoch(req); which happened in 682 * __drbd_make_request, because we now may set the bit 683 * again ourselves to close the current epoch. 684 * 685 * Add req to the (now) current epoch (barrier). */ 686 687 /* otherwise we may lose an unplug, which may cause some remote 688 * io-scheduler timeout to expire, increasing maximum latency, 689 * hurting performance. */ 690 set_bit(UNPLUG_REMOTE, &device->flags); 691 692 /* queue work item to send data */ 693 D_ASSERT(device, req->rq_state & RQ_NET_PENDING); 694 mod_rq_state(req, m, 0, RQ_NET_QUEUED|RQ_EXP_BARR_ACK); 695 req->w.cb = w_send_dblock; 696 drbd_queue_work(&connection->sender_work, 697 &req->w); 698 699 /* close the epoch, in case it outgrew the limit */ 700 rcu_read_lock(); 701 nc = rcu_dereference(connection->net_conf); 702 p = nc->max_epoch_size; 703 rcu_read_unlock(); 704 if (connection->current_tle_writes >= p) 705 start_new_tl_epoch(connection); 706 707 break; 708 709 case QUEUE_FOR_SEND_OOS: 710 mod_rq_state(req, m, 0, RQ_NET_QUEUED); 711 req->w.cb = w_send_out_of_sync; 712 drbd_queue_work(&connection->sender_work, 713 &req->w); 714 break; 715 716 case READ_RETRY_REMOTE_CANCELED: 717 case SEND_CANCELED: 718 case SEND_FAILED: 719 /* real cleanup will be done from tl_clear. just update flags 720 * so it is no longer marked as on the worker queue */ 721 mod_rq_state(req, m, RQ_NET_QUEUED, 0); 722 break; 723 724 case HANDED_OVER_TO_NETWORK: 725 /* assert something? */ 726 if (is_pending_write_protocol_A(req)) 727 /* this is what is dangerous about protocol A: 728 * pretend it was successfully written on the peer. */ 729 mod_rq_state(req, m, RQ_NET_QUEUED|RQ_NET_PENDING, 730 RQ_NET_SENT|RQ_NET_OK); 731 else 732 mod_rq_state(req, m, RQ_NET_QUEUED, RQ_NET_SENT); 733 /* It is still not yet RQ_NET_DONE until the 734 * corresponding epoch barrier got acked as well, 735 * so we know what to dirty on connection loss. */ 736 break; 737 738 case OOS_HANDED_TO_NETWORK: 739 /* Was not set PENDING, no longer QUEUED, so is now DONE 740 * as far as this connection is concerned. */ 741 mod_rq_state(req, m, RQ_NET_QUEUED, RQ_NET_DONE); 742 break; 743 744 case CONNECTION_LOST_WHILE_PENDING: 745 /* transfer log cleanup after connection loss */ 746 mod_rq_state(req, m, 747 RQ_NET_OK|RQ_NET_PENDING|RQ_COMPLETION_SUSP, 748 RQ_NET_DONE); 749 break; 750 751 case CONFLICT_RESOLVED: 752 /* for superseded conflicting writes of multiple primaries, 753 * there is no need to keep anything in the tl, potential 754 * node crashes are covered by the activity log. 755 * 756 * If this request had been marked as RQ_POSTPONED before, 757 * it will actually not be completed, but "restarted", 758 * resubmitted from the retry worker context. */ 759 D_ASSERT(device, req->rq_state & RQ_NET_PENDING); 760 D_ASSERT(device, req->rq_state & RQ_EXP_WRITE_ACK); 761 mod_rq_state(req, m, RQ_NET_PENDING, RQ_NET_DONE|RQ_NET_OK); 762 break; 763 764 case WRITE_ACKED_BY_PEER_AND_SIS: 765 req->rq_state |= RQ_NET_SIS; 766 fallthrough; 767 case WRITE_ACKED_BY_PEER: 768 /* Normal operation protocol C: successfully written on peer. 769 * During resync, even in protocol != C, 770 * we requested an explicit write ack anyways. 771 * Which means we cannot even assert anything here. 772 * Nothing more to do here. 773 * We want to keep the tl in place for all protocols, to cater 774 * for volatile write-back caches on lower level devices. */ 775 goto ack_common; 776 case RECV_ACKED_BY_PEER: 777 D_ASSERT(device, req->rq_state & RQ_EXP_RECEIVE_ACK); 778 /* protocol B; pretends to be successfully written on peer. 779 * see also notes above in HANDED_OVER_TO_NETWORK about 780 * protocol != C */ 781 ack_common: 782 mod_rq_state(req, m, RQ_NET_PENDING, RQ_NET_OK); 783 break; 784 785 case POSTPONE_WRITE: 786 D_ASSERT(device, req->rq_state & RQ_EXP_WRITE_ACK); 787 /* If this node has already detected the write conflict, the 788 * worker will be waiting on misc_wait. Wake it up once this 789 * request has completed locally. 790 */ 791 D_ASSERT(device, req->rq_state & RQ_NET_PENDING); 792 req->rq_state |= RQ_POSTPONED; 793 if (req->i.waiting) 794 wake_up(&device->misc_wait); 795 /* Do not clear RQ_NET_PENDING. This request will make further 796 * progress via restart_conflicting_writes() or 797 * fail_postponed_requests(). Hopefully. */ 798 break; 799 800 case NEG_ACKED: 801 mod_rq_state(req, m, RQ_NET_OK|RQ_NET_PENDING, 0); 802 break; 803 804 case FAIL_FROZEN_DISK_IO: 805 if (!(req->rq_state & RQ_LOCAL_COMPLETED)) 806 break; 807 mod_rq_state(req, m, RQ_COMPLETION_SUSP, 0); 808 break; 809 810 case RESTART_FROZEN_DISK_IO: 811 if (!(req->rq_state & RQ_LOCAL_COMPLETED)) 812 break; 813 814 mod_rq_state(req, m, 815 RQ_COMPLETION_SUSP|RQ_LOCAL_COMPLETED, 816 RQ_LOCAL_PENDING); 817 818 rv = MR_READ; 819 if (bio_data_dir(req->master_bio) == WRITE) 820 rv = MR_WRITE; 821 822 get_ldev(device); /* always succeeds in this call path */ 823 req->w.cb = w_restart_disk_io; 824 drbd_queue_work(&connection->sender_work, 825 &req->w); 826 break; 827 828 case RESEND: 829 /* Simply complete (local only) READs. */ 830 if (!(req->rq_state & RQ_WRITE) && !req->w.cb) { 831 mod_rq_state(req, m, RQ_COMPLETION_SUSP, 0); 832 break; 833 } 834 835 /* If RQ_NET_OK is already set, we got a P_WRITE_ACK or P_RECV_ACK 836 before the connection loss (B&C only); only P_BARRIER_ACK 837 (or the local completion?) was missing when we suspended. 838 Throwing them out of the TL here by pretending we got a BARRIER_ACK. 839 During connection handshake, we ensure that the peer was not rebooted. */ 840 if (!(req->rq_state & RQ_NET_OK)) { 841 /* FIXME could this possibly be a req->dw.cb == w_send_out_of_sync? 842 * in that case we must not set RQ_NET_PENDING. */ 843 844 mod_rq_state(req, m, RQ_COMPLETION_SUSP, RQ_NET_QUEUED|RQ_NET_PENDING); 845 if (req->w.cb) { 846 /* w.cb expected to be w_send_dblock, or w_send_read_req */ 847 drbd_queue_work(&connection->sender_work, 848 &req->w); 849 rv = req->rq_state & RQ_WRITE ? MR_WRITE : MR_READ; 850 } /* else: FIXME can this happen? */ 851 break; 852 } 853 fallthrough; /* to BARRIER_ACKED */ 854 855 case BARRIER_ACKED: 856 /* barrier ack for READ requests does not make sense */ 857 if (!(req->rq_state & RQ_WRITE)) 858 break; 859 860 if (req->rq_state & RQ_NET_PENDING) { 861 /* barrier came in before all requests were acked. 862 * this is bad, because if the connection is lost now, 863 * we won't be able to clean them up... */ 864 drbd_err(device, "FIXME (BARRIER_ACKED but pending)\n"); 865 } 866 /* Allowed to complete requests, even while suspended. 867 * As this is called for all requests within a matching epoch, 868 * we need to filter, and only set RQ_NET_DONE for those that 869 * have actually been on the wire. */ 870 mod_rq_state(req, m, RQ_COMPLETION_SUSP, 871 (req->rq_state & RQ_NET_MASK) ? RQ_NET_DONE : 0); 872 break; 873 874 case DATA_RECEIVED: 875 D_ASSERT(device, req->rq_state & RQ_NET_PENDING); 876 mod_rq_state(req, m, RQ_NET_PENDING, RQ_NET_OK|RQ_NET_DONE); 877 break; 878 879 case QUEUE_AS_DRBD_BARRIER: 880 start_new_tl_epoch(connection); 881 mod_rq_state(req, m, 0, RQ_NET_OK|RQ_NET_DONE); 882 break; 883 } 884 885 return rv; 886} 887 888/* we may do a local read if: 889 * - we are consistent (of course), 890 * - or we are generally inconsistent, 891 * BUT we are still/already IN SYNC for this area. 892 * since size may be bigger than BM_BLOCK_SIZE, 893 * we may need to check several bits. 894 */ 895static bool drbd_may_do_local_read(struct drbd_device *device, sector_t sector, int size) 896{ 897 unsigned long sbnr, ebnr; 898 sector_t esector, nr_sectors; 899 900 if (device->state.disk == D_UP_TO_DATE) 901 return true; 902 if (device->state.disk != D_INCONSISTENT) 903 return false; 904 esector = sector + (size >> 9) - 1; 905 nr_sectors = get_capacity(device->vdisk); 906 D_ASSERT(device, sector < nr_sectors); 907 D_ASSERT(device, esector < nr_sectors); 908 909 sbnr = BM_SECT_TO_BIT(sector); 910 ebnr = BM_SECT_TO_BIT(esector); 911 912 return drbd_bm_count_bits(device, sbnr, ebnr) == 0; 913} 914 915static bool remote_due_to_read_balancing(struct drbd_device *device, sector_t sector, 916 enum drbd_read_balancing rbm) 917{ 918 int stripe_shift; 919 920 switch (rbm) { 921 case RB_CONGESTED_REMOTE: 922 return false; 923 case RB_LEAST_PENDING: 924 return atomic_read(&device->local_cnt) > 925 atomic_read(&device->ap_pending_cnt) + atomic_read(&device->rs_pending_cnt); 926 case RB_32K_STRIPING: /* stripe_shift = 15 */ 927 case RB_64K_STRIPING: 928 case RB_128K_STRIPING: 929 case RB_256K_STRIPING: 930 case RB_512K_STRIPING: 931 case RB_1M_STRIPING: /* stripe_shift = 20 */ 932 stripe_shift = (rbm - RB_32K_STRIPING + 15); 933 return (sector >> (stripe_shift - 9)) & 1; 934 case RB_ROUND_ROBIN: 935 return test_and_change_bit(READ_BALANCE_RR, &device->flags); 936 case RB_PREFER_REMOTE: 937 return true; 938 case RB_PREFER_LOCAL: 939 default: 940 return false; 941 } 942} 943 944/* 945 * complete_conflicting_writes - wait for any conflicting write requests 946 * 947 * The write_requests tree contains all active write requests which we 948 * currently know about. Wait for any requests to complete which conflict with 949 * the new one. 950 * 951 * Only way out: remove the conflicting intervals from the tree. 952 */ 953static void complete_conflicting_writes(struct drbd_request *req) 954{ 955 DEFINE_WAIT(wait); 956 struct drbd_device *device = req->device; 957 struct drbd_interval *i; 958 sector_t sector = req->i.sector; 959 int size = req->i.size; 960 961 for (;;) { 962 drbd_for_each_overlap(i, &device->write_requests, sector, size) { 963 /* Ignore, if already completed to upper layers. */ 964 if (i->completed) 965 continue; 966 /* Handle the first found overlap. After the schedule 967 * we have to restart the tree walk. */ 968 break; 969 } 970 if (!i) /* if any */ 971 break; 972 973 /* Indicate to wake up device->misc_wait on progress. */ 974 prepare_to_wait(&device->misc_wait, &wait, TASK_UNINTERRUPTIBLE); 975 i->waiting = true; 976 spin_unlock_irq(&device->resource->req_lock); 977 schedule(); 978 spin_lock_irq(&device->resource->req_lock); 979 } 980 finish_wait(&device->misc_wait, &wait); 981} 982 983/* called within req_lock */ 984static void maybe_pull_ahead(struct drbd_device *device) 985{ 986 struct drbd_connection *connection = first_peer_device(device)->connection; 987 struct net_conf *nc; 988 bool congested = false; 989 enum drbd_on_congestion on_congestion; 990 991 rcu_read_lock(); 992 nc = rcu_dereference(connection->net_conf); 993 on_congestion = nc ? nc->on_congestion : OC_BLOCK; 994 rcu_read_unlock(); 995 if (on_congestion == OC_BLOCK || 996 connection->agreed_pro_version < 96) 997 return; 998 999 if (on_congestion == OC_PULL_AHEAD && device->state.conn == C_AHEAD) 1000 return; /* nothing to do ... */ 1001 1002 /* If I don't even have good local storage, we can not reasonably try 1003 * to pull ahead of the peer. We also need the local reference to make 1004 * sure device->act_log is there. 1005 */ 1006 if (!get_ldev_if_state(device, D_UP_TO_DATE)) 1007 return; 1008 1009 if (nc->cong_fill && 1010 atomic_read(&device->ap_in_flight) >= nc->cong_fill) { 1011 drbd_info(device, "Congestion-fill threshold reached\n"); 1012 congested = true; 1013 } 1014 1015 if (device->act_log->used >= nc->cong_extents) { 1016 drbd_info(device, "Congestion-extents threshold reached\n"); 1017 congested = true; 1018 } 1019 1020 if (congested) { 1021 /* start a new epoch for non-mirrored writes */ 1022 start_new_tl_epoch(first_peer_device(device)->connection); 1023 1024 if (on_congestion == OC_PULL_AHEAD) 1025 _drbd_set_state(_NS(device, conn, C_AHEAD), 0, NULL); 1026 else /*nc->on_congestion == OC_DISCONNECT */ 1027 _drbd_set_state(_NS(device, conn, C_DISCONNECTING), 0, NULL); 1028 } 1029 put_ldev(device); 1030} 1031 1032/* If this returns false, and req->private_bio is still set, 1033 * this should be submitted locally. 1034 * 1035 * If it returns false, but req->private_bio is not set, 1036 * we do not have access to good data :( 1037 * 1038 * Otherwise, this destroys req->private_bio, if any, 1039 * and returns true. 1040 */ 1041static bool do_remote_read(struct drbd_request *req) 1042{ 1043 struct drbd_device *device = req->device; 1044 enum drbd_read_balancing rbm; 1045 1046 if (req->private_bio) { 1047 if (!drbd_may_do_local_read(device, 1048 req->i.sector, req->i.size)) { 1049 bio_put(req->private_bio); 1050 req->private_bio = NULL; 1051 put_ldev(device); 1052 } 1053 } 1054 1055 if (device->state.pdsk != D_UP_TO_DATE) 1056 return false; 1057 1058 if (req->private_bio == NULL) 1059 return true; 1060 1061 /* TODO: improve read balancing decisions, take into account drbd 1062 * protocol, pending requests etc. */ 1063 1064 rcu_read_lock(); 1065 rbm = rcu_dereference(device->ldev->disk_conf)->read_balancing; 1066 rcu_read_unlock(); 1067 1068 if (rbm == RB_PREFER_LOCAL && req->private_bio) 1069 return false; /* submit locally */ 1070 1071 if (remote_due_to_read_balancing(device, req->i.sector, rbm)) { 1072 if (req->private_bio) { 1073 bio_put(req->private_bio); 1074 req->private_bio = NULL; 1075 put_ldev(device); 1076 } 1077 return true; 1078 } 1079 1080 return false; 1081} 1082 1083bool drbd_should_do_remote(union drbd_dev_state s) 1084{ 1085 return s.pdsk == D_UP_TO_DATE || 1086 (s.pdsk >= D_INCONSISTENT && 1087 s.conn >= C_WF_BITMAP_T && 1088 s.conn < C_AHEAD); 1089 /* Before proto 96 that was >= CONNECTED instead of >= C_WF_BITMAP_T. 1090 That is equivalent since before 96 IO was frozen in the C_WF_BITMAP* 1091 states. */ 1092} 1093 1094static bool drbd_should_send_out_of_sync(union drbd_dev_state s) 1095{ 1096 return s.conn == C_AHEAD || s.conn == C_WF_BITMAP_S; 1097 /* pdsk = D_INCONSISTENT as a consequence. Protocol 96 check not necessary 1098 since we enter state C_AHEAD only if proto >= 96 */ 1099} 1100 1101/* returns number of connections (== 1, for drbd 8.4) 1102 * expected to actually write this data, 1103 * which does NOT include those that we are L_AHEAD for. */ 1104static int drbd_process_write_request(struct drbd_request *req) 1105{ 1106 struct drbd_device *device = req->device; 1107 struct drbd_peer_device *peer_device = first_peer_device(device); 1108 int remote, send_oos; 1109 1110 remote = drbd_should_do_remote(device->state); 1111 send_oos = drbd_should_send_out_of_sync(device->state); 1112 1113 /* Need to replicate writes. Unless it is an empty flush, 1114 * which is better mapped to a DRBD P_BARRIER packet, 1115 * also for drbd wire protocol compatibility reasons. 1116 * If this was a flush, just start a new epoch. 1117 * Unless the current epoch was empty anyways, or we are not currently 1118 * replicating, in which case there is no point. */ 1119 if (unlikely(req->i.size == 0)) { 1120 /* The only size==0 bios we expect are empty flushes. */ 1121 D_ASSERT(device, req->master_bio->bi_opf & REQ_PREFLUSH); 1122 if (remote) 1123 _req_mod(req, QUEUE_AS_DRBD_BARRIER, peer_device); 1124 return remote; 1125 } 1126 1127 if (!remote && !send_oos) 1128 return 0; 1129 1130 D_ASSERT(device, !(remote && send_oos)); 1131 1132 if (remote) { 1133 _req_mod(req, TO_BE_SENT, peer_device); 1134 _req_mod(req, QUEUE_FOR_NET_WRITE, peer_device); 1135 } else if (drbd_set_out_of_sync(peer_device, req->i.sector, req->i.size)) 1136 _req_mod(req, QUEUE_FOR_SEND_OOS, peer_device); 1137 1138 return remote; 1139} 1140 1141static void drbd_process_discard_or_zeroes_req(struct drbd_request *req, int flags) 1142{ 1143 int err = drbd_issue_discard_or_zero_out(req->device, 1144 req->i.sector, req->i.size >> 9, flags); 1145 if (err) 1146 req->private_bio->bi_status = BLK_STS_IOERR; 1147 bio_endio(req->private_bio); 1148} 1149 1150static void 1151drbd_submit_req_private_bio(struct drbd_request *req) 1152{ 1153 struct drbd_device *device = req->device; 1154 struct bio *bio = req->private_bio; 1155 unsigned int type; 1156 1157 if (bio_op(bio) != REQ_OP_READ) 1158 type = DRBD_FAULT_DT_WR; 1159 else if (bio->bi_opf & REQ_RAHEAD) 1160 type = DRBD_FAULT_DT_RA; 1161 else 1162 type = DRBD_FAULT_DT_RD; 1163 1164 /* State may have changed since we grabbed our reference on the 1165 * ->ldev member. Double check, and short-circuit to endio. 1166 * In case the last activity log transaction failed to get on 1167 * stable storage, and this is a WRITE, we may not even submit 1168 * this bio. */ 1169 if (get_ldev(device)) { 1170 if (drbd_insert_fault(device, type)) 1171 bio_io_error(bio); 1172 else if (bio_op(bio) == REQ_OP_WRITE_ZEROES) 1173 drbd_process_discard_or_zeroes_req(req, EE_ZEROOUT | 1174 ((bio->bi_opf & REQ_NOUNMAP) ? 0 : EE_TRIM)); 1175 else if (bio_op(bio) == REQ_OP_DISCARD) 1176 drbd_process_discard_or_zeroes_req(req, EE_TRIM); 1177 else 1178 submit_bio_noacct(bio); 1179 put_ldev(device); 1180 } else 1181 bio_io_error(bio); 1182} 1183 1184static void drbd_queue_write(struct drbd_device *device, struct drbd_request *req) 1185{ 1186 spin_lock_irq(&device->resource->req_lock); 1187 list_add_tail(&req->tl_requests, &device->submit.writes); 1188 list_add_tail(&req->req_pending_master_completion, 1189 &device->pending_master_completion[1 /* WRITE */]); 1190 spin_unlock_irq(&device->resource->req_lock); 1191 queue_work(device->submit.wq, &device->submit.worker); 1192 /* do_submit() may sleep internally on al_wait, too */ 1193 wake_up(&device->al_wait); 1194} 1195 1196/* returns the new drbd_request pointer, if the caller is expected to 1197 * drbd_send_and_submit() it (to save latency), or NULL if we queued the 1198 * request on the submitter thread. 1199 * Returns ERR_PTR(-ENOMEM) if we cannot allocate a drbd_request. 1200 */ 1201static struct drbd_request * 1202drbd_request_prepare(struct drbd_device *device, struct bio *bio) 1203{ 1204 const int rw = bio_data_dir(bio); 1205 struct drbd_request *req; 1206 1207 /* allocate outside of all locks; */ 1208 req = drbd_req_new(device, bio); 1209 if (!req) { 1210 dec_ap_bio(device); 1211 /* only pass the error to the upper layers. 1212 * if user cannot handle io errors, that's not our business. */ 1213 drbd_err(device, "could not kmalloc() req\n"); 1214 bio->bi_status = BLK_STS_RESOURCE; 1215 bio_endio(bio); 1216 return ERR_PTR(-ENOMEM); 1217 } 1218 1219 /* Update disk stats */ 1220 req->start_jif = bio_start_io_acct(req->master_bio); 1221 1222 if (get_ldev(device)) { 1223 req->private_bio = bio_alloc_clone(device->ldev->backing_bdev, 1224 bio, GFP_NOIO, 1225 &drbd_io_bio_set); 1226 req->private_bio->bi_private = req; 1227 req->private_bio->bi_end_io = drbd_request_endio; 1228 } 1229 1230 /* process discards always from our submitter thread */ 1231 if (bio_op(bio) == REQ_OP_WRITE_ZEROES || 1232 bio_op(bio) == REQ_OP_DISCARD) 1233 goto queue_for_submitter_thread; 1234 1235 if (rw == WRITE && req->private_bio && req->i.size 1236 && !test_bit(AL_SUSPENDED, &device->flags)) { 1237 if (!drbd_al_begin_io_fastpath(device, &req->i)) 1238 goto queue_for_submitter_thread; 1239 req->rq_state |= RQ_IN_ACT_LOG; 1240 req->in_actlog_jif = jiffies; 1241 } 1242 return req; 1243 1244 queue_for_submitter_thread: 1245 atomic_inc(&device->ap_actlog_cnt); 1246 drbd_queue_write(device, req); 1247 return NULL; 1248} 1249 1250/* Require at least one path to current data. 1251 * We don't want to allow writes on C_STANDALONE D_INCONSISTENT: 1252 * We would not allow to read what was written, 1253 * we would not have bumped the data generation uuids, 1254 * we would cause data divergence for all the wrong reasons. 1255 * 1256 * If we don't see at least one D_UP_TO_DATE, we will fail this request, 1257 * which either returns EIO, or, if OND_SUSPEND_IO is set, suspends IO, 1258 * and queues for retry later. 1259 */ 1260static bool may_do_writes(struct drbd_device *device) 1261{ 1262 const union drbd_dev_state s = device->state; 1263 return s.disk == D_UP_TO_DATE || s.pdsk == D_UP_TO_DATE; 1264} 1265 1266struct drbd_plug_cb { 1267 struct blk_plug_cb cb; 1268 struct drbd_request *most_recent_req; 1269 /* do we need more? */ 1270}; 1271 1272static void drbd_unplug(struct blk_plug_cb *cb, bool from_schedule) 1273{ 1274 struct drbd_plug_cb *plug = container_of(cb, struct drbd_plug_cb, cb); 1275 struct drbd_resource *resource = plug->cb.data; 1276 struct drbd_request *req = plug->most_recent_req; 1277 1278 kfree(cb); 1279 if (!req) 1280 return; 1281 1282 spin_lock_irq(&resource->req_lock); 1283 /* In case the sender did not process it yet, raise the flag to 1284 * have it followed with P_UNPLUG_REMOTE just after. */ 1285 req->rq_state |= RQ_UNPLUG; 1286 /* but also queue a generic unplug */ 1287 drbd_queue_unplug(req->device); 1288 kref_put(&req->kref, drbd_req_destroy); 1289 spin_unlock_irq(&resource->req_lock); 1290} 1291 1292static struct drbd_plug_cb* drbd_check_plugged(struct drbd_resource *resource) 1293{ 1294 /* A lot of text to say 1295 * return (struct drbd_plug_cb*)blk_check_plugged(); */ 1296 struct drbd_plug_cb *plug; 1297 struct blk_plug_cb *cb = blk_check_plugged(drbd_unplug, resource, sizeof(*plug)); 1298 1299 if (cb) 1300 plug = container_of(cb, struct drbd_plug_cb, cb); 1301 else 1302 plug = NULL; 1303 return plug; 1304} 1305 1306static void drbd_update_plug(struct drbd_plug_cb *plug, struct drbd_request *req) 1307{ 1308 struct drbd_request *tmp = plug->most_recent_req; 1309 /* Will be sent to some peer. 1310 * Remember to tag it with UNPLUG_REMOTE on unplug */ 1311 kref_get(&req->kref); 1312 plug->most_recent_req = req; 1313 if (tmp) 1314 kref_put(&tmp->kref, drbd_req_destroy); 1315} 1316 1317static void drbd_send_and_submit(struct drbd_device *device, struct drbd_request *req) 1318{ 1319 struct drbd_resource *resource = device->resource; 1320 struct drbd_peer_device *peer_device = first_peer_device(device); 1321 const int rw = bio_data_dir(req->master_bio); 1322 struct bio_and_error m = { NULL, }; 1323 bool no_remote = false; 1324 bool submit_private_bio = false; 1325 1326 spin_lock_irq(&resource->req_lock); 1327 if (rw == WRITE) { 1328 /* This may temporarily give up the req_lock, 1329 * but will re-aquire it before it returns here. 1330 * Needs to be before the check on drbd_suspended() */ 1331 complete_conflicting_writes(req); 1332 /* no more giving up req_lock from now on! */ 1333 1334 /* check for congestion, and potentially stop sending 1335 * full data updates, but start sending "dirty bits" only. */ 1336 maybe_pull_ahead(device); 1337 } 1338 1339 1340 if (drbd_suspended(device)) { 1341 /* push back and retry: */ 1342 req->rq_state |= RQ_POSTPONED; 1343 if (req->private_bio) { 1344 bio_put(req->private_bio); 1345 req->private_bio = NULL; 1346 put_ldev(device); 1347 } 1348 goto out; 1349 } 1350 1351 /* We fail READ early, if we can not serve it. 1352 * We must do this before req is registered on any lists. 1353 * Otherwise, drbd_req_complete() will queue failed READ for retry. */ 1354 if (rw != WRITE) { 1355 if (!do_remote_read(req) && !req->private_bio) 1356 goto nodata; 1357 } 1358 1359 /* which transfer log epoch does this belong to? */ 1360 req->epoch = atomic_read(&first_peer_device(device)->connection->current_tle_nr); 1361 1362 /* no point in adding empty flushes to the transfer log, 1363 * they are mapped to drbd barriers already. */ 1364 if (likely(req->i.size!=0)) { 1365 if (rw == WRITE) 1366 first_peer_device(device)->connection->current_tle_writes++; 1367 1368 list_add_tail(&req->tl_requests, &first_peer_device(device)->connection->transfer_log); 1369 } 1370 1371 if (rw == WRITE) { 1372 if (req->private_bio && !may_do_writes(device)) { 1373 bio_put(req->private_bio); 1374 req->private_bio = NULL; 1375 put_ldev(device); 1376 goto nodata; 1377 } 1378 if (!drbd_process_write_request(req)) 1379 no_remote = true; 1380 } else { 1381 /* We either have a private_bio, or we can read from remote. 1382 * Otherwise we had done the goto nodata above. */ 1383 if (req->private_bio == NULL) { 1384 _req_mod(req, TO_BE_SENT, peer_device); 1385 _req_mod(req, QUEUE_FOR_NET_READ, peer_device); 1386 } else 1387 no_remote = true; 1388 } 1389 1390 if (no_remote == false) { 1391 struct drbd_plug_cb *plug = drbd_check_plugged(resource); 1392 if (plug) 1393 drbd_update_plug(plug, req); 1394 } 1395 1396 /* If it took the fast path in drbd_request_prepare, add it here. 1397 * The slow path has added it already. */ 1398 if (list_empty(&req->req_pending_master_completion)) 1399 list_add_tail(&req->req_pending_master_completion, 1400 &device->pending_master_completion[rw == WRITE]); 1401 if (req->private_bio) { 1402 /* needs to be marked within the same spinlock */ 1403 req->pre_submit_jif = jiffies; 1404 list_add_tail(&req->req_pending_local, 1405 &device->pending_completion[rw == WRITE]); 1406 _req_mod(req, TO_BE_SUBMITTED, NULL); 1407 /* but we need to give up the spinlock to submit */ 1408 submit_private_bio = true; 1409 } else if (no_remote) { 1410nodata: 1411 if (drbd_ratelimit()) 1412 drbd_err(device, "IO ERROR: neither local nor remote data, sector %llu+%u\n", 1413 (unsigned long long)req->i.sector, req->i.size >> 9); 1414 /* A write may have been queued for send_oos, however. 1415 * So we can not simply free it, we must go through drbd_req_put_completion_ref() */ 1416 } 1417 1418out: 1419 drbd_req_put_completion_ref(req, &m, 1); 1420 spin_unlock_irq(&resource->req_lock); 1421 1422 /* Even though above is a kref_put(), this is safe. 1423 * As long as we still need to submit our private bio, 1424 * we hold a completion ref, and the request cannot disappear. 1425 * If however this request did not even have a private bio to submit 1426 * (e.g. remote read), req may already be invalid now. 1427 * That's why we cannot check on req->private_bio. */ 1428 if (submit_private_bio) 1429 drbd_submit_req_private_bio(req); 1430 if (m.bio) 1431 complete_master_bio(device, &m); 1432} 1433 1434void __drbd_make_request(struct drbd_device *device, struct bio *bio) 1435{ 1436 struct drbd_request *req = drbd_request_prepare(device, bio); 1437 if (IS_ERR_OR_NULL(req)) 1438 return; 1439 drbd_send_and_submit(device, req); 1440} 1441 1442static void submit_fast_path(struct drbd_device *device, struct list_head *incoming) 1443{ 1444 struct blk_plug plug; 1445 struct drbd_request *req, *tmp; 1446 1447 blk_start_plug(&plug); 1448 list_for_each_entry_safe(req, tmp, incoming, tl_requests) { 1449 const int rw = bio_data_dir(req->master_bio); 1450 1451 if (rw == WRITE /* rw != WRITE should not even end up here! */ 1452 && req->private_bio && req->i.size 1453 && !test_bit(AL_SUSPENDED, &device->flags)) { 1454 if (!drbd_al_begin_io_fastpath(device, &req->i)) 1455 continue; 1456 1457 req->rq_state |= RQ_IN_ACT_LOG; 1458 req->in_actlog_jif = jiffies; 1459 atomic_dec(&device->ap_actlog_cnt); 1460 } 1461 1462 list_del_init(&req->tl_requests); 1463 drbd_send_and_submit(device, req); 1464 } 1465 blk_finish_plug(&plug); 1466} 1467 1468static bool prepare_al_transaction_nonblock(struct drbd_device *device, 1469 struct list_head *incoming, 1470 struct list_head *pending, 1471 struct list_head *later) 1472{ 1473 struct drbd_request *req; 1474 int wake = 0; 1475 int err; 1476 1477 spin_lock_irq(&device->al_lock); 1478 while ((req = list_first_entry_or_null(incoming, struct drbd_request, tl_requests))) { 1479 err = drbd_al_begin_io_nonblock(device, &req->i); 1480 if (err == -ENOBUFS) 1481 break; 1482 if (err == -EBUSY) 1483 wake = 1; 1484 if (err) 1485 list_move_tail(&req->tl_requests, later); 1486 else 1487 list_move_tail(&req->tl_requests, pending); 1488 } 1489 spin_unlock_irq(&device->al_lock); 1490 if (wake) 1491 wake_up(&device->al_wait); 1492 return !list_empty(pending); 1493} 1494 1495static void send_and_submit_pending(struct drbd_device *device, struct list_head *pending) 1496{ 1497 struct blk_plug plug; 1498 struct drbd_request *req; 1499 1500 blk_start_plug(&plug); 1501 while ((req = list_first_entry_or_null(pending, struct drbd_request, tl_requests))) { 1502 req->rq_state |= RQ_IN_ACT_LOG; 1503 req->in_actlog_jif = jiffies; 1504 atomic_dec(&device->ap_actlog_cnt); 1505 list_del_init(&req->tl_requests); 1506 drbd_send_and_submit(device, req); 1507 } 1508 blk_finish_plug(&plug); 1509} 1510 1511void do_submit(struct work_struct *ws) 1512{ 1513 struct drbd_device *device = container_of(ws, struct drbd_device, submit.worker); 1514 LIST_HEAD(incoming); /* from drbd_make_request() */ 1515 LIST_HEAD(pending); /* to be submitted after next AL-transaction commit */ 1516 LIST_HEAD(busy); /* blocked by resync requests */ 1517 1518 /* grab new incoming requests */ 1519 spin_lock_irq(&device->resource->req_lock); 1520 list_splice_tail_init(&device->submit.writes, &incoming); 1521 spin_unlock_irq(&device->resource->req_lock); 1522 1523 for (;;) { 1524 DEFINE_WAIT(wait); 1525 1526 /* move used-to-be-busy back to front of incoming */ 1527 list_splice_init(&busy, &incoming); 1528 submit_fast_path(device, &incoming); 1529 if (list_empty(&incoming)) 1530 break; 1531 1532 for (;;) { 1533 prepare_to_wait(&device->al_wait, &wait, TASK_UNINTERRUPTIBLE); 1534 1535 list_splice_init(&busy, &incoming); 1536 prepare_al_transaction_nonblock(device, &incoming, &pending, &busy); 1537 if (!list_empty(&pending)) 1538 break; 1539 1540 schedule(); 1541 1542 /* If all currently "hot" activity log extents are kept busy by 1543 * incoming requests, we still must not totally starve new 1544 * requests to "cold" extents. 1545 * Something left on &incoming means there had not been 1546 * enough update slots available, and the activity log 1547 * has been marked as "starving". 1548 * 1549 * Try again now, without looking for new requests, 1550 * effectively blocking all new requests until we made 1551 * at least _some_ progress with what we currently have. 1552 */ 1553 if (!list_empty(&incoming)) 1554 continue; 1555 1556 /* Nothing moved to pending, but nothing left 1557 * on incoming: all moved to busy! 1558 * Grab new and iterate. */ 1559 spin_lock_irq(&device->resource->req_lock); 1560 list_splice_tail_init(&device->submit.writes, &incoming); 1561 spin_unlock_irq(&device->resource->req_lock); 1562 } 1563 finish_wait(&device->al_wait, &wait); 1564 1565 /* If the transaction was full, before all incoming requests 1566 * had been processed, skip ahead to commit, and iterate 1567 * without splicing in more incoming requests from upper layers. 1568 * 1569 * Else, if all incoming have been processed, 1570 * they have become either "pending" (to be submitted after 1571 * next transaction commit) or "busy" (blocked by resync). 1572 * 1573 * Maybe more was queued, while we prepared the transaction? 1574 * Try to stuff those into this transaction as well. 1575 * Be strictly non-blocking here, 1576 * we already have something to commit. 1577 * 1578 * Commit if we don't make any more progres. 1579 */ 1580 1581 while (list_empty(&incoming)) { 1582 LIST_HEAD(more_pending); 1583 LIST_HEAD(more_incoming); 1584 bool made_progress; 1585 1586 /* It is ok to look outside the lock, 1587 * it's only an optimization anyways */ 1588 if (list_empty(&device->submit.writes)) 1589 break; 1590 1591 spin_lock_irq(&device->resource->req_lock); 1592 list_splice_tail_init(&device->submit.writes, &more_incoming); 1593 spin_unlock_irq(&device->resource->req_lock); 1594 1595 if (list_empty(&more_incoming)) 1596 break; 1597 1598 made_progress = prepare_al_transaction_nonblock(device, &more_incoming, &more_pending, &busy); 1599 1600 list_splice_tail_init(&more_pending, &pending); 1601 list_splice_tail_init(&more_incoming, &incoming); 1602 if (!made_progress) 1603 break; 1604 } 1605 1606 drbd_al_begin_io_commit(device); 1607 send_and_submit_pending(device, &pending); 1608 } 1609} 1610 1611void drbd_submit_bio(struct bio *bio) 1612{ 1613 struct drbd_device *device = bio->bi_bdev->bd_disk->private_data; 1614 1615 bio = bio_split_to_limits(bio); 1616 if (!bio) 1617 return; 1618 1619 /* 1620 * what we "blindly" assume: 1621 */ 1622 D_ASSERT(device, IS_ALIGNED(bio->bi_iter.bi_size, 512)); 1623 1624 inc_ap_bio(device); 1625 __drbd_make_request(device, bio); 1626} 1627 1628static bool net_timeout_reached(struct drbd_request *net_req, 1629 struct drbd_connection *connection, 1630 unsigned long now, unsigned long ent, 1631 unsigned int ko_count, unsigned int timeout) 1632{ 1633 struct drbd_device *device = net_req->device; 1634 1635 if (!time_after(now, net_req->pre_send_jif + ent)) 1636 return false; 1637 1638 if (time_in_range(now, connection->last_reconnect_jif, connection->last_reconnect_jif + ent)) 1639 return false; 1640 1641 if (net_req->rq_state & RQ_NET_PENDING) { 1642 drbd_warn(device, "Remote failed to finish a request within %ums > ko-count (%u) * timeout (%u * 0.1s)\n", 1643 jiffies_to_msecs(now - net_req->pre_send_jif), ko_count, timeout); 1644 return true; 1645 } 1646 1647 /* We received an ACK already (or are using protocol A), 1648 * but are waiting for the epoch closing barrier ack. 1649 * Check if we sent the barrier already. We should not blame the peer 1650 * for being unresponsive, if we did not even ask it yet. */ 1651 if (net_req->epoch == connection->send.current_epoch_nr) { 1652 drbd_warn(device, 1653 "We did not send a P_BARRIER for %ums > ko-count (%u) * timeout (%u * 0.1s); drbd kernel thread blocked?\n", 1654 jiffies_to_msecs(now - net_req->pre_send_jif), ko_count, timeout); 1655 return false; 1656 } 1657 1658 /* Worst case: we may have been blocked for whatever reason, then 1659 * suddenly are able to send a lot of requests (and epoch separating 1660 * barriers) in quick succession. 1661 * The timestamp of the net_req may be much too old and not correspond 1662 * to the sending time of the relevant unack'ed barrier packet, so 1663 * would trigger a spurious timeout. The latest barrier packet may 1664 * have a too recent timestamp to trigger the timeout, potentially miss 1665 * a timeout. Right now we don't have a place to conveniently store 1666 * these timestamps. 1667 * But in this particular situation, the application requests are still 1668 * completed to upper layers, DRBD should still "feel" responsive. 1669 * No need yet to kill this connection, it may still recover. 1670 * If not, eventually we will have queued enough into the network for 1671 * us to block. From that point of view, the timestamp of the last sent 1672 * barrier packet is relevant enough. 1673 */ 1674 if (time_after(now, connection->send.last_sent_barrier_jif + ent)) { 1675 drbd_warn(device, "Remote failed to answer a P_BARRIER (sent at %lu jif; now=%lu jif) within %ums > ko-count (%u) * timeout (%u * 0.1s)\n", 1676 connection->send.last_sent_barrier_jif, now, 1677 jiffies_to_msecs(now - connection->send.last_sent_barrier_jif), ko_count, timeout); 1678 return true; 1679 } 1680 return false; 1681} 1682 1683/* A request is considered timed out, if 1684 * - we have some effective timeout from the configuration, 1685 * with some state restrictions applied, 1686 * - the oldest request is waiting for a response from the network 1687 * resp. the local disk, 1688 * - the oldest request is in fact older than the effective timeout, 1689 * - the connection was established (resp. disk was attached) 1690 * for longer than the timeout already. 1691 * Note that for 32bit jiffies and very stable connections/disks, 1692 * we may have a wrap around, which is catched by 1693 * !time_in_range(now, last_..._jif, last_..._jif + timeout). 1694 * 1695 * Side effect: once per 32bit wrap-around interval, which means every 1696 * ~198 days with 250 HZ, we have a window where the timeout would need 1697 * to expire twice (worst case) to become effective. Good enough. 1698 */ 1699 1700void request_timer_fn(struct timer_list *t) 1701{ 1702 struct drbd_device *device = from_timer(device, t, request_timer); 1703 struct drbd_connection *connection = first_peer_device(device)->connection; 1704 struct drbd_request *req_read, *req_write, *req_peer; /* oldest request */ 1705 struct net_conf *nc; 1706 unsigned long oldest_submit_jif; 1707 unsigned long ent = 0, dt = 0, et, nt; /* effective timeout = ko_count * timeout */ 1708 unsigned long now; 1709 unsigned int ko_count = 0, timeout = 0; 1710 1711 rcu_read_lock(); 1712 nc = rcu_dereference(connection->net_conf); 1713 if (nc && device->state.conn >= C_WF_REPORT_PARAMS) { 1714 ko_count = nc->ko_count; 1715 timeout = nc->timeout; 1716 } 1717 1718 if (get_ldev(device)) { /* implicit state.disk >= D_INCONSISTENT */ 1719 dt = rcu_dereference(device->ldev->disk_conf)->disk_timeout * HZ / 10; 1720 put_ldev(device); 1721 } 1722 rcu_read_unlock(); 1723 1724 1725 ent = timeout * HZ/10 * ko_count; 1726 et = min_not_zero(dt, ent); 1727 1728 if (!et) 1729 return; /* Recurring timer stopped */ 1730 1731 now = jiffies; 1732 nt = now + et; 1733 1734 spin_lock_irq(&device->resource->req_lock); 1735 req_read = list_first_entry_or_null(&device->pending_completion[0], struct drbd_request, req_pending_local); 1736 req_write = list_first_entry_or_null(&device->pending_completion[1], struct drbd_request, req_pending_local); 1737 1738 /* maybe the oldest request waiting for the peer is in fact still 1739 * blocking in tcp sendmsg. That's ok, though, that's handled via the 1740 * socket send timeout, requesting a ping, and bumping ko-count in 1741 * we_should_drop_the_connection(). 1742 */ 1743 1744 /* check the oldest request we did successfully sent, 1745 * but which is still waiting for an ACK. */ 1746 req_peer = connection->req_ack_pending; 1747 1748 /* if we don't have such request (e.g. protocoll A) 1749 * check the oldest requests which is still waiting on its epoch 1750 * closing barrier ack. */ 1751 if (!req_peer) 1752 req_peer = connection->req_not_net_done; 1753 1754 /* evaluate the oldest peer request only in one timer! */ 1755 if (req_peer && req_peer->device != device) 1756 req_peer = NULL; 1757 1758 /* do we have something to evaluate? */ 1759 if (req_peer == NULL && req_write == NULL && req_read == NULL) 1760 goto out; 1761 1762 oldest_submit_jif = 1763 (req_write && req_read) 1764 ? ( time_before(req_write->pre_submit_jif, req_read->pre_submit_jif) 1765 ? req_write->pre_submit_jif : req_read->pre_submit_jif ) 1766 : req_write ? req_write->pre_submit_jif 1767 : req_read ? req_read->pre_submit_jif : now; 1768 1769 if (ent && req_peer && net_timeout_reached(req_peer, connection, now, ent, ko_count, timeout)) 1770 _conn_request_state(connection, NS(conn, C_TIMEOUT), CS_VERBOSE | CS_HARD); 1771 1772 if (dt && oldest_submit_jif != now && 1773 time_after(now, oldest_submit_jif + dt) && 1774 !time_in_range(now, device->last_reattach_jif, device->last_reattach_jif + dt)) { 1775 drbd_warn(device, "Local backing device failed to meet the disk-timeout\n"); 1776 __drbd_chk_io_error(device, DRBD_FORCE_DETACH); 1777 } 1778 1779 /* Reschedule timer for the nearest not already expired timeout. 1780 * Fallback to now + min(effective network timeout, disk timeout). */ 1781 ent = (ent && req_peer && time_before(now, req_peer->pre_send_jif + ent)) 1782 ? req_peer->pre_send_jif + ent : now + et; 1783 dt = (dt && oldest_submit_jif != now && time_before(now, oldest_submit_jif + dt)) 1784 ? oldest_submit_jif + dt : now + et; 1785 nt = time_before(ent, dt) ? ent : dt; 1786out: 1787 spin_unlock_irq(&device->resource->req_lock); 1788 mod_timer(&device->request_timer, nt); 1789} 1790