1 /****************************************************************************** 2 ******************************************************************************* 3 ** 4 ** Copyright (C) Sistina Software, Inc. 1997-2003 All rights reserved. 5 ** Copyright (C) 2005-2008 Red Hat, Inc. All rights reserved. 6 ** 7 ** This copyrighted material is made available to anyone wishing to use, 8 ** modify, copy, or redistribute it subject to the terms and conditions 9 ** of the GNU General Public License v.2. 10 ** 11 ******************************************************************************* 12 ******************************************************************************/ 13 14 #include "dlm_internal.h" 15 #include "lockspace.h" 16 #include "member.h" 17 #include "lowcomms.h" 18 #include "midcomms.h" 19 #include "rcom.h" 20 #include "recover.h" 21 #include "dir.h" 22 #include "config.h" 23 #include "memory.h" 24 #include "lock.h" 25 #include "util.h" 26 27 static int rcom_response(struct dlm_ls *ls) 28 { 29 return test_bit(LSFL_RCOM_READY, &ls->ls_flags); 30 } 31 32 static int create_rcom(struct dlm_ls *ls, int to_nodeid, int type, int len, 33 struct dlm_rcom **rc_ret, struct dlm_mhandle **mh_ret) 34 { 35 struct dlm_rcom *rc; 36 struct dlm_mhandle *mh; 37 char *mb; 38 int mb_len = sizeof(struct dlm_rcom) + len; 39 40 mh = dlm_lowcomms_get_buffer(to_nodeid, mb_len, GFP_NOFS, &mb); 41 if (!mh) { 42 log_print("create_rcom to %d type %d len %d ENOBUFS", 43 to_nodeid, type, len); 44 return -ENOBUFS; 45 } 46 memset(mb, 0, mb_len); 47 48 rc = (struct dlm_rcom *) mb; 49 50 rc->rc_header.h_version = (DLM_HEADER_MAJOR | DLM_HEADER_MINOR); 51 rc->rc_header.h_lockspace = ls->ls_global_id; 52 rc->rc_header.h_nodeid = dlm_our_nodeid(); 53 rc->rc_header.h_length = mb_len; 54 rc->rc_header.h_cmd = DLM_RCOM; 55 56 rc->rc_type = type; 57 58 spin_lock(&ls->ls_recover_lock); 59 rc->rc_seq = ls->ls_recover_seq; 60 spin_unlock(&ls->ls_recover_lock); 61 62 *mh_ret = mh; 63 *rc_ret = rc; 64 return 0; 65 } 66 67 static void send_rcom(struct dlm_ls *ls, struct dlm_mhandle *mh, 68 struct dlm_rcom *rc) 69 { 70 dlm_rcom_out(rc); 71 dlm_lowcomms_commit_buffer(mh); 72 } 73 74 static void set_rcom_status(struct dlm_ls *ls, struct rcom_status *rs, 75 uint32_t flags) 76 { 77 rs->rs_flags = cpu_to_le32(flags); 78 } 79 80 /* When replying to a status request, a node also sends back its 81 configuration values. The requesting node then checks that the remote 82 node is configured the same way as itself. */ 83 84 static void set_rcom_config(struct dlm_ls *ls, struct rcom_config *rf, 85 uint32_t num_slots) 86 { 87 rf->rf_lvblen = cpu_to_le32(ls->ls_lvblen); 88 rf->rf_lsflags = cpu_to_le32(ls->ls_exflags); 89 90 rf->rf_our_slot = cpu_to_le16(ls->ls_slot); 91 rf->rf_num_slots = cpu_to_le16(num_slots); 92 rf->rf_generation = cpu_to_le32(ls->ls_generation); 93 } 94 95 static int check_rcom_config(struct dlm_ls *ls, struct dlm_rcom *rc, int nodeid) 96 { 97 struct rcom_config *rf = (struct rcom_config *) rc->rc_buf; 98 99 if ((rc->rc_header.h_version & 0xFFFF0000) != DLM_HEADER_MAJOR) { 100 log_error(ls, "version mismatch: %x nodeid %d: %x", 101 DLM_HEADER_MAJOR | DLM_HEADER_MINOR, nodeid, 102 rc->rc_header.h_version); 103 return -EPROTO; 104 } 105 106 if (le32_to_cpu(rf->rf_lvblen) != ls->ls_lvblen || 107 le32_to_cpu(rf->rf_lsflags) != ls->ls_exflags) { 108 log_error(ls, "config mismatch: %d,%x nodeid %d: %d,%x", 109 ls->ls_lvblen, ls->ls_exflags, nodeid, 110 le32_to_cpu(rf->rf_lvblen), 111 le32_to_cpu(rf->rf_lsflags)); 112 return -EPROTO; 113 } 114 return 0; 115 } 116 117 static void allow_sync_reply(struct dlm_ls *ls, uint64_t *new_seq) 118 { 119 spin_lock(&ls->ls_rcom_spin); 120 *new_seq = ++ls->ls_rcom_seq; 121 set_bit(LSFL_RCOM_WAIT, &ls->ls_flags); 122 spin_unlock(&ls->ls_rcom_spin); 123 } 124 125 static void disallow_sync_reply(struct dlm_ls *ls) 126 { 127 spin_lock(&ls->ls_rcom_spin); 128 clear_bit(LSFL_RCOM_WAIT, &ls->ls_flags); 129 clear_bit(LSFL_RCOM_READY, &ls->ls_flags); 130 spin_unlock(&ls->ls_rcom_spin); 131 } 132 133 /* 134 * low nodeid gathers one slot value at a time from each node. 135 * it sets need_slots=0, and saves rf_our_slot returned from each 136 * rcom_config. 137 * 138 * other nodes gather all slot values at once from the low nodeid. 139 * they set need_slots=1, and ignore the rf_our_slot returned from each 140 * rcom_config. they use the rf_num_slots returned from the low 141 * node's rcom_config. 142 */ 143 144 int dlm_rcom_status(struct dlm_ls *ls, int nodeid, uint32_t status_flags) 145 { 146 struct dlm_rcom *rc; 147 struct dlm_mhandle *mh; 148 int error = 0; 149 150 ls->ls_recover_nodeid = nodeid; 151 152 if (nodeid == dlm_our_nodeid()) { 153 rc = ls->ls_recover_buf; 154 rc->rc_result = dlm_recover_status(ls); 155 goto out; 156 } 157 158 retry: 159 error = create_rcom(ls, nodeid, DLM_RCOM_STATUS, 160 sizeof(struct rcom_status), &rc, &mh); 161 if (error) 162 goto out; 163 164 set_rcom_status(ls, (struct rcom_status *)rc->rc_buf, status_flags); 165 166 allow_sync_reply(ls, &rc->rc_id); 167 memset(ls->ls_recover_buf, 0, dlm_config.ci_buffer_size); 168 169 send_rcom(ls, mh, rc); 170 171 error = dlm_wait_function(ls, &rcom_response); 172 disallow_sync_reply(ls); 173 if (error == -ETIMEDOUT) 174 goto retry; 175 if (error) 176 goto out; 177 178 rc = ls->ls_recover_buf; 179 180 if (rc->rc_result == -ESRCH) { 181 /* we pretend the remote lockspace exists with 0 status */ 182 log_debug(ls, "remote node %d not ready", nodeid); 183 rc->rc_result = 0; 184 error = 0; 185 } else { 186 error = check_rcom_config(ls, rc, nodeid); 187 } 188 189 /* the caller looks at rc_result for the remote recovery status */ 190 out: 191 return error; 192 } 193 194 static void receive_rcom_status(struct dlm_ls *ls, struct dlm_rcom *rc_in) 195 { 196 struct dlm_rcom *rc; 197 struct dlm_mhandle *mh; 198 struct rcom_status *rs; 199 uint32_t status; 200 int nodeid = rc_in->rc_header.h_nodeid; 201 int len = sizeof(struct rcom_config); 202 int num_slots = 0; 203 int error; 204 205 if (!dlm_slots_version(&rc_in->rc_header)) { 206 status = dlm_recover_status(ls); 207 goto do_create; 208 } 209 210 rs = (struct rcom_status *)rc_in->rc_buf; 211 212 if (!(le32_to_cpu(rs->rs_flags) & DLM_RSF_NEED_SLOTS)) { 213 status = dlm_recover_status(ls); 214 goto do_create; 215 } 216 217 spin_lock(&ls->ls_recover_lock); 218 status = ls->ls_recover_status; 219 num_slots = ls->ls_num_slots; 220 spin_unlock(&ls->ls_recover_lock); 221 len += num_slots * sizeof(struct rcom_slot); 222 223 do_create: 224 error = create_rcom(ls, nodeid, DLM_RCOM_STATUS_REPLY, 225 len, &rc, &mh); 226 if (error) 227 return; 228 229 rc->rc_id = rc_in->rc_id; 230 rc->rc_seq_reply = rc_in->rc_seq; 231 rc->rc_result = status; 232 233 set_rcom_config(ls, (struct rcom_config *)rc->rc_buf, num_slots); 234 235 if (!num_slots) 236 goto do_send; 237 238 spin_lock(&ls->ls_recover_lock); 239 if (ls->ls_num_slots != num_slots) { 240 spin_unlock(&ls->ls_recover_lock); 241 log_debug(ls, "receive_rcom_status num_slots %d to %d", 242 num_slots, ls->ls_num_slots); 243 rc->rc_result = 0; 244 set_rcom_config(ls, (struct rcom_config *)rc->rc_buf, 0); 245 goto do_send; 246 } 247 248 dlm_slots_copy_out(ls, rc); 249 spin_unlock(&ls->ls_recover_lock); 250 251 do_send: 252 send_rcom(ls, mh, rc); 253 } 254 255 static void receive_sync_reply(struct dlm_ls *ls, struct dlm_rcom *rc_in) 256 { 257 spin_lock(&ls->ls_rcom_spin); 258 if (!test_bit(LSFL_RCOM_WAIT, &ls->ls_flags) || 259 rc_in->rc_id != ls->ls_rcom_seq) { 260 log_debug(ls, "reject reply %d from %d seq %llx expect %llx", 261 rc_in->rc_type, rc_in->rc_header.h_nodeid, 262 (unsigned long long)rc_in->rc_id, 263 (unsigned long long)ls->ls_rcom_seq); 264 goto out; 265 } 266 memcpy(ls->ls_recover_buf, rc_in, rc_in->rc_header.h_length); 267 set_bit(LSFL_RCOM_READY, &ls->ls_flags); 268 clear_bit(LSFL_RCOM_WAIT, &ls->ls_flags); 269 wake_up(&ls->ls_wait_general); 270 out: 271 spin_unlock(&ls->ls_rcom_spin); 272 } 273 274 int dlm_rcom_names(struct dlm_ls *ls, int nodeid, char *last_name, int last_len) 275 { 276 struct dlm_rcom *rc; 277 struct dlm_mhandle *mh; 278 int error = 0; 279 280 ls->ls_recover_nodeid = nodeid; 281 282 retry: 283 error = create_rcom(ls, nodeid, DLM_RCOM_NAMES, last_len, &rc, &mh); 284 if (error) 285 goto out; 286 memcpy(rc->rc_buf, last_name, last_len); 287 288 allow_sync_reply(ls, &rc->rc_id); 289 memset(ls->ls_recover_buf, 0, dlm_config.ci_buffer_size); 290 291 send_rcom(ls, mh, rc); 292 293 error = dlm_wait_function(ls, &rcom_response); 294 disallow_sync_reply(ls); 295 if (error == -ETIMEDOUT) 296 goto retry; 297 out: 298 return error; 299 } 300 301 static void receive_rcom_names(struct dlm_ls *ls, struct dlm_rcom *rc_in) 302 { 303 struct dlm_rcom *rc; 304 struct dlm_mhandle *mh; 305 int error, inlen, outlen, nodeid; 306 307 nodeid = rc_in->rc_header.h_nodeid; 308 inlen = rc_in->rc_header.h_length - sizeof(struct dlm_rcom); 309 outlen = dlm_config.ci_buffer_size - sizeof(struct dlm_rcom); 310 311 error = create_rcom(ls, nodeid, DLM_RCOM_NAMES_REPLY, outlen, &rc, &mh); 312 if (error) 313 return; 314 rc->rc_id = rc_in->rc_id; 315 rc->rc_seq_reply = rc_in->rc_seq; 316 317 dlm_copy_master_names(ls, rc_in->rc_buf, inlen, rc->rc_buf, outlen, 318 nodeid); 319 send_rcom(ls, mh, rc); 320 } 321 322 int dlm_send_rcom_lookup(struct dlm_rsb *r, int dir_nodeid) 323 { 324 struct dlm_rcom *rc; 325 struct dlm_mhandle *mh; 326 struct dlm_ls *ls = r->res_ls; 327 int error; 328 329 error = create_rcom(ls, dir_nodeid, DLM_RCOM_LOOKUP, r->res_length, 330 &rc, &mh); 331 if (error) 332 goto out; 333 memcpy(rc->rc_buf, r->res_name, r->res_length); 334 rc->rc_id = (unsigned long) r->res_id; 335 336 send_rcom(ls, mh, rc); 337 out: 338 return error; 339 } 340 341 static void receive_rcom_lookup(struct dlm_ls *ls, struct dlm_rcom *rc_in) 342 { 343 struct dlm_rcom *rc; 344 struct dlm_mhandle *mh; 345 int error, ret_nodeid, nodeid = rc_in->rc_header.h_nodeid; 346 int len = rc_in->rc_header.h_length - sizeof(struct dlm_rcom); 347 348 error = create_rcom(ls, nodeid, DLM_RCOM_LOOKUP_REPLY, 0, &rc, &mh); 349 if (error) 350 return; 351 352 /* Old code would send this special id to trigger a debug dump. */ 353 if (rc_in->rc_id == 0xFFFFFFFF) { 354 log_error(ls, "receive_rcom_lookup dump from %d", nodeid); 355 dlm_dump_rsb_name(ls, rc_in->rc_buf, len); 356 return; 357 } 358 359 error = dlm_master_lookup(ls, nodeid, rc_in->rc_buf, len, 360 DLM_LU_RECOVER_MASTER, &ret_nodeid, NULL); 361 if (error) 362 ret_nodeid = error; 363 rc->rc_result = ret_nodeid; 364 rc->rc_id = rc_in->rc_id; 365 rc->rc_seq_reply = rc_in->rc_seq; 366 367 send_rcom(ls, mh, rc); 368 } 369 370 static void receive_rcom_lookup_reply(struct dlm_ls *ls, struct dlm_rcom *rc_in) 371 { 372 dlm_recover_master_reply(ls, rc_in); 373 } 374 375 static void pack_rcom_lock(struct dlm_rsb *r, struct dlm_lkb *lkb, 376 struct rcom_lock *rl) 377 { 378 memset(rl, 0, sizeof(*rl)); 379 380 rl->rl_ownpid = cpu_to_le32(lkb->lkb_ownpid); 381 rl->rl_lkid = cpu_to_le32(lkb->lkb_id); 382 rl->rl_exflags = cpu_to_le32(lkb->lkb_exflags); 383 rl->rl_flags = cpu_to_le32(lkb->lkb_flags); 384 rl->rl_lvbseq = cpu_to_le32(lkb->lkb_lvbseq); 385 rl->rl_rqmode = lkb->lkb_rqmode; 386 rl->rl_grmode = lkb->lkb_grmode; 387 rl->rl_status = lkb->lkb_status; 388 rl->rl_wait_type = cpu_to_le16(lkb->lkb_wait_type); 389 390 if (lkb->lkb_bastfn) 391 rl->rl_asts |= DLM_CB_BAST; 392 if (lkb->lkb_astfn) 393 rl->rl_asts |= DLM_CB_CAST; 394 395 rl->rl_namelen = cpu_to_le16(r->res_length); 396 memcpy(rl->rl_name, r->res_name, r->res_length); 397 398 /* FIXME: might we have an lvb without DLM_LKF_VALBLK set ? 399 If so, receive_rcom_lock_args() won't take this copy. */ 400 401 if (lkb->lkb_lvbptr) 402 memcpy(rl->rl_lvb, lkb->lkb_lvbptr, r->res_ls->ls_lvblen); 403 } 404 405 int dlm_send_rcom_lock(struct dlm_rsb *r, struct dlm_lkb *lkb) 406 { 407 struct dlm_ls *ls = r->res_ls; 408 struct dlm_rcom *rc; 409 struct dlm_mhandle *mh; 410 struct rcom_lock *rl; 411 int error, len = sizeof(struct rcom_lock); 412 413 if (lkb->lkb_lvbptr) 414 len += ls->ls_lvblen; 415 416 error = create_rcom(ls, r->res_nodeid, DLM_RCOM_LOCK, len, &rc, &mh); 417 if (error) 418 goto out; 419 420 rl = (struct rcom_lock *) rc->rc_buf; 421 pack_rcom_lock(r, lkb, rl); 422 rc->rc_id = (unsigned long) r; 423 424 send_rcom(ls, mh, rc); 425 out: 426 return error; 427 } 428 429 /* needs at least dlm_rcom + rcom_lock */ 430 static void receive_rcom_lock(struct dlm_ls *ls, struct dlm_rcom *rc_in) 431 { 432 struct dlm_rcom *rc; 433 struct dlm_mhandle *mh; 434 int error, nodeid = rc_in->rc_header.h_nodeid; 435 436 dlm_recover_master_copy(ls, rc_in); 437 438 error = create_rcom(ls, nodeid, DLM_RCOM_LOCK_REPLY, 439 sizeof(struct rcom_lock), &rc, &mh); 440 if (error) 441 return; 442 443 /* We send back the same rcom_lock struct we received, but 444 dlm_recover_master_copy() has filled in rl_remid and rl_result */ 445 446 memcpy(rc->rc_buf, rc_in->rc_buf, sizeof(struct rcom_lock)); 447 rc->rc_id = rc_in->rc_id; 448 rc->rc_seq_reply = rc_in->rc_seq; 449 450 send_rcom(ls, mh, rc); 451 } 452 453 /* If the lockspace doesn't exist then still send a status message 454 back; it's possible that it just doesn't have its global_id yet. */ 455 456 int dlm_send_ls_not_ready(int nodeid, struct dlm_rcom *rc_in) 457 { 458 struct dlm_rcom *rc; 459 struct rcom_config *rf; 460 struct dlm_mhandle *mh; 461 char *mb; 462 int mb_len = sizeof(struct dlm_rcom) + sizeof(struct rcom_config); 463 464 mh = dlm_lowcomms_get_buffer(nodeid, mb_len, GFP_NOFS, &mb); 465 if (!mh) 466 return -ENOBUFS; 467 memset(mb, 0, mb_len); 468 469 rc = (struct dlm_rcom *) mb; 470 471 rc->rc_header.h_version = (DLM_HEADER_MAJOR | DLM_HEADER_MINOR); 472 rc->rc_header.h_lockspace = rc_in->rc_header.h_lockspace; 473 rc->rc_header.h_nodeid = dlm_our_nodeid(); 474 rc->rc_header.h_length = mb_len; 475 rc->rc_header.h_cmd = DLM_RCOM; 476 477 rc->rc_type = DLM_RCOM_STATUS_REPLY; 478 rc->rc_id = rc_in->rc_id; 479 rc->rc_seq_reply = rc_in->rc_seq; 480 rc->rc_result = -ESRCH; 481 482 rf = (struct rcom_config *) rc->rc_buf; 483 rf->rf_lvblen = cpu_to_le32(~0U); 484 485 dlm_rcom_out(rc); 486 dlm_lowcomms_commit_buffer(mh); 487 488 return 0; 489 } 490 491 /* 492 * Ignore messages for stage Y before we set 493 * recover_status bit for stage X: 494 * 495 * recover_status = 0 496 * 497 * dlm_recover_members() 498 * - send nothing 499 * - recv nothing 500 * - ignore NAMES, NAMES_REPLY 501 * - ignore LOOKUP, LOOKUP_REPLY 502 * - ignore LOCK, LOCK_REPLY 503 * 504 * recover_status |= NODES 505 * 506 * dlm_recover_members_wait() 507 * 508 * dlm_recover_directory() 509 * - send NAMES 510 * - recv NAMES_REPLY 511 * - ignore LOOKUP, LOOKUP_REPLY 512 * - ignore LOCK, LOCK_REPLY 513 * 514 * recover_status |= DIR 515 * 516 * dlm_recover_directory_wait() 517 * 518 * dlm_recover_masters() 519 * - send LOOKUP 520 * - recv LOOKUP_REPLY 521 * 522 * dlm_recover_locks() 523 * - send LOCKS 524 * - recv LOCKS_REPLY 525 * 526 * recover_status |= LOCKS 527 * 528 * dlm_recover_locks_wait() 529 * 530 * recover_status |= DONE 531 */ 532 533 /* Called by dlm_recv; corresponds to dlm_receive_message() but special 534 recovery-only comms are sent through here. */ 535 536 void dlm_receive_rcom(struct dlm_ls *ls, struct dlm_rcom *rc, int nodeid) 537 { 538 int lock_size = sizeof(struct dlm_rcom) + sizeof(struct rcom_lock); 539 int stop, reply = 0, names = 0, lookup = 0, lock = 0; 540 uint32_t status; 541 uint64_t seq; 542 543 switch (rc->rc_type) { 544 case DLM_RCOM_STATUS_REPLY: 545 reply = 1; 546 break; 547 case DLM_RCOM_NAMES: 548 names = 1; 549 break; 550 case DLM_RCOM_NAMES_REPLY: 551 names = 1; 552 reply = 1; 553 break; 554 case DLM_RCOM_LOOKUP: 555 lookup = 1; 556 break; 557 case DLM_RCOM_LOOKUP_REPLY: 558 lookup = 1; 559 reply = 1; 560 break; 561 case DLM_RCOM_LOCK: 562 lock = 1; 563 break; 564 case DLM_RCOM_LOCK_REPLY: 565 lock = 1; 566 reply = 1; 567 break; 568 }; 569 570 spin_lock(&ls->ls_recover_lock); 571 status = ls->ls_recover_status; 572 stop = test_bit(LSFL_RECOVER_STOP, &ls->ls_flags); 573 seq = ls->ls_recover_seq; 574 spin_unlock(&ls->ls_recover_lock); 575 576 if (stop && (rc->rc_type != DLM_RCOM_STATUS)) 577 goto ignore; 578 579 if (reply && (rc->rc_seq_reply != seq)) 580 goto ignore; 581 582 if (!(status & DLM_RS_NODES) && (names || lookup || lock)) 583 goto ignore; 584 585 if (!(status & DLM_RS_DIR) && (lookup || lock)) 586 goto ignore; 587 588 switch (rc->rc_type) { 589 case DLM_RCOM_STATUS: 590 receive_rcom_status(ls, rc); 591 break; 592 593 case DLM_RCOM_NAMES: 594 receive_rcom_names(ls, rc); 595 break; 596 597 case DLM_RCOM_LOOKUP: 598 receive_rcom_lookup(ls, rc); 599 break; 600 601 case DLM_RCOM_LOCK: 602 if (rc->rc_header.h_length < lock_size) 603 goto Eshort; 604 receive_rcom_lock(ls, rc); 605 break; 606 607 case DLM_RCOM_STATUS_REPLY: 608 receive_sync_reply(ls, rc); 609 break; 610 611 case DLM_RCOM_NAMES_REPLY: 612 receive_sync_reply(ls, rc); 613 break; 614 615 case DLM_RCOM_LOOKUP_REPLY: 616 receive_rcom_lookup_reply(ls, rc); 617 break; 618 619 case DLM_RCOM_LOCK_REPLY: 620 if (rc->rc_header.h_length < lock_size) 621 goto Eshort; 622 dlm_recover_process_copy(ls, rc); 623 break; 624 625 default: 626 log_error(ls, "receive_rcom bad type %d", rc->rc_type); 627 } 628 return; 629 630 ignore: 631 log_limit(ls, "dlm_receive_rcom ignore msg %d " 632 "from %d %llu %llu recover seq %llu sts %x gen %u", 633 rc->rc_type, 634 nodeid, 635 (unsigned long long)rc->rc_seq, 636 (unsigned long long)rc->rc_seq_reply, 637 (unsigned long long)seq, 638 status, ls->ls_generation); 639 return; 640 Eshort: 641 log_error(ls, "recovery message %d from %d is too short", 642 rc->rc_type, nodeid); 643 } 644 645