1 #include <linux/ceph/ceph_debug.h> 2 3 #include <linux/module.h> 4 #include <linux/err.h> 5 #include <linux/highmem.h> 6 #include <linux/mm.h> 7 #include <linux/pagemap.h> 8 #include <linux/slab.h> 9 #include <linux/uaccess.h> 10 #ifdef CONFIG_BLOCK 11 #include <linux/bio.h> 12 #endif 13 14 #include <linux/ceph/libceph.h> 15 #include <linux/ceph/osd_client.h> 16 #include <linux/ceph/messenger.h> 17 #include <linux/ceph/decode.h> 18 #include <linux/ceph/auth.h> 19 #include <linux/ceph/pagelist.h> 20 21 #define OSD_OP_FRONT_LEN 4096 22 #define OSD_OPREPLY_FRONT_LEN 512 23 24 static const struct ceph_connection_operations osd_con_ops; 25 26 static void send_queued(struct ceph_osd_client *osdc); 27 static int __reset_osd(struct ceph_osd_client *osdc, struct ceph_osd *osd); 28 static void __register_request(struct ceph_osd_client *osdc, 29 struct ceph_osd_request *req); 30 static void __unregister_linger_request(struct ceph_osd_client *osdc, 31 struct ceph_osd_request *req); 32 static int __send_request(struct ceph_osd_client *osdc, 33 struct ceph_osd_request *req); 34 35 static int op_needs_trail(int op) 36 { 37 switch (op) { 38 case CEPH_OSD_OP_GETXATTR: 39 case CEPH_OSD_OP_SETXATTR: 40 case CEPH_OSD_OP_CMPXATTR: 41 case CEPH_OSD_OP_CALL: 42 case CEPH_OSD_OP_NOTIFY: 43 return 1; 44 default: 45 return 0; 46 } 47 } 48 49 static int op_has_extent(int op) 50 { 51 return (op == CEPH_OSD_OP_READ || 52 op == CEPH_OSD_OP_WRITE); 53 } 54 55 void ceph_calc_raw_layout(struct ceph_osd_client *osdc, 56 struct ceph_file_layout *layout, 57 u64 snapid, 58 u64 off, u64 *plen, u64 *bno, 59 struct ceph_osd_request *req, 60 struct ceph_osd_req_op *op) 61 { 62 struct ceph_osd_request_head *reqhead = req->r_request->front.iov_base; 63 u64 orig_len = *plen; 64 u64 objoff, objlen; /* extent in object */ 65 66 reqhead->snapid = cpu_to_le64(snapid); 67 68 /* object extent? */ 69 ceph_calc_file_object_mapping(layout, off, plen, bno, 70 &objoff, &objlen); 71 if (*plen < orig_len) 72 dout(" skipping last %llu, final file extent %llu~%llu\n", 73 orig_len - *plen, off, *plen); 74 75 if (op_has_extent(op->op)) { 76 op->extent.offset = objoff; 77 op->extent.length = objlen; 78 } 79 req->r_num_pages = calc_pages_for(off, *plen); 80 req->r_page_alignment = off & ~PAGE_MASK; 81 if (op->op == CEPH_OSD_OP_WRITE) 82 op->payload_len = *plen; 83 84 dout("calc_layout bno=%llx %llu~%llu (%d pages)\n", 85 *bno, objoff, objlen, req->r_num_pages); 86 87 } 88 EXPORT_SYMBOL(ceph_calc_raw_layout); 89 90 /* 91 * Implement client access to distributed object storage cluster. 92 * 93 * All data objects are stored within a cluster/cloud of OSDs, or 94 * "object storage devices." (Note that Ceph OSDs have _nothing_ to 95 * do with the T10 OSD extensions to SCSI.) Ceph OSDs are simply 96 * remote daemons serving up and coordinating consistent and safe 97 * access to storage. 98 * 99 * Cluster membership and the mapping of data objects onto storage devices 100 * are described by the osd map. 101 * 102 * We keep track of pending OSD requests (read, write), resubmit 103 * requests to different OSDs when the cluster topology/data layout 104 * change, or retry the affected requests when the communications 105 * channel with an OSD is reset. 106 */ 107 108 /* 109 * calculate the mapping of a file extent onto an object, and fill out the 110 * request accordingly. shorten extent as necessary if it crosses an 111 * object boundary. 112 * 113 * fill osd op in request message. 114 */ 115 static void calc_layout(struct ceph_osd_client *osdc, 116 struct ceph_vino vino, 117 struct ceph_file_layout *layout, 118 u64 off, u64 *plen, 119 struct ceph_osd_request *req, 120 struct ceph_osd_req_op *op) 121 { 122 u64 bno; 123 124 ceph_calc_raw_layout(osdc, layout, vino.snap, off, 125 plen, &bno, req, op); 126 127 snprintf(req->r_oid, sizeof(req->r_oid), "%llx.%08llx", vino.ino, bno); 128 req->r_oid_len = strlen(req->r_oid); 129 } 130 131 /* 132 * requests 133 */ 134 void ceph_osdc_release_request(struct kref *kref) 135 { 136 struct ceph_osd_request *req = container_of(kref, 137 struct ceph_osd_request, 138 r_kref); 139 140 if (req->r_request) 141 ceph_msg_put(req->r_request); 142 if (req->r_reply) 143 ceph_msg_put(req->r_reply); 144 if (req->r_con_filling_msg) { 145 dout("release_request revoking pages %p from con %p\n", 146 req->r_pages, req->r_con_filling_msg); 147 ceph_con_revoke_message(req->r_con_filling_msg, 148 req->r_reply); 149 ceph_con_put(req->r_con_filling_msg); 150 } 151 if (req->r_own_pages) 152 ceph_release_page_vector(req->r_pages, 153 req->r_num_pages); 154 #ifdef CONFIG_BLOCK 155 if (req->r_bio) 156 bio_put(req->r_bio); 157 #endif 158 ceph_put_snap_context(req->r_snapc); 159 if (req->r_trail) { 160 ceph_pagelist_release(req->r_trail); 161 kfree(req->r_trail); 162 } 163 if (req->r_mempool) 164 mempool_free(req, req->r_osdc->req_mempool); 165 else 166 kfree(req); 167 } 168 EXPORT_SYMBOL(ceph_osdc_release_request); 169 170 static int get_num_ops(struct ceph_osd_req_op *ops, int *needs_trail) 171 { 172 int i = 0; 173 174 if (needs_trail) 175 *needs_trail = 0; 176 while (ops[i].op) { 177 if (needs_trail && op_needs_trail(ops[i].op)) 178 *needs_trail = 1; 179 i++; 180 } 181 182 return i; 183 } 184 185 struct ceph_osd_request *ceph_osdc_alloc_request(struct ceph_osd_client *osdc, 186 int flags, 187 struct ceph_snap_context *snapc, 188 struct ceph_osd_req_op *ops, 189 bool use_mempool, 190 gfp_t gfp_flags, 191 struct page **pages, 192 struct bio *bio) 193 { 194 struct ceph_osd_request *req; 195 struct ceph_msg *msg; 196 int needs_trail; 197 int num_op = get_num_ops(ops, &needs_trail); 198 size_t msg_size = sizeof(struct ceph_osd_request_head); 199 200 msg_size += num_op*sizeof(struct ceph_osd_op); 201 202 if (use_mempool) { 203 req = mempool_alloc(osdc->req_mempool, gfp_flags); 204 memset(req, 0, sizeof(*req)); 205 } else { 206 req = kzalloc(sizeof(*req), gfp_flags); 207 } 208 if (req == NULL) 209 return NULL; 210 211 req->r_osdc = osdc; 212 req->r_mempool = use_mempool; 213 214 kref_init(&req->r_kref); 215 init_completion(&req->r_completion); 216 init_completion(&req->r_safe_completion); 217 INIT_LIST_HEAD(&req->r_unsafe_item); 218 INIT_LIST_HEAD(&req->r_linger_item); 219 INIT_LIST_HEAD(&req->r_linger_osd); 220 req->r_flags = flags; 221 222 WARN_ON((flags & (CEPH_OSD_FLAG_READ|CEPH_OSD_FLAG_WRITE)) == 0); 223 224 /* create reply message */ 225 if (use_mempool) 226 msg = ceph_msgpool_get(&osdc->msgpool_op_reply, 0); 227 else 228 msg = ceph_msg_new(CEPH_MSG_OSD_OPREPLY, 229 OSD_OPREPLY_FRONT_LEN, gfp_flags); 230 if (!msg) { 231 ceph_osdc_put_request(req); 232 return NULL; 233 } 234 req->r_reply = msg; 235 236 /* allocate space for the trailing data */ 237 if (needs_trail) { 238 req->r_trail = kmalloc(sizeof(struct ceph_pagelist), gfp_flags); 239 if (!req->r_trail) { 240 ceph_osdc_put_request(req); 241 return NULL; 242 } 243 ceph_pagelist_init(req->r_trail); 244 } 245 /* create request message; allow space for oid */ 246 msg_size += 40; 247 if (snapc) 248 msg_size += sizeof(u64) * snapc->num_snaps; 249 if (use_mempool) 250 msg = ceph_msgpool_get(&osdc->msgpool_op, 0); 251 else 252 msg = ceph_msg_new(CEPH_MSG_OSD_OP, msg_size, gfp_flags); 253 if (!msg) { 254 ceph_osdc_put_request(req); 255 return NULL; 256 } 257 258 msg->hdr.type = cpu_to_le16(CEPH_MSG_OSD_OP); 259 memset(msg->front.iov_base, 0, msg->front.iov_len); 260 261 req->r_request = msg; 262 req->r_pages = pages; 263 #ifdef CONFIG_BLOCK 264 if (bio) { 265 req->r_bio = bio; 266 bio_get(req->r_bio); 267 } 268 #endif 269 270 return req; 271 } 272 EXPORT_SYMBOL(ceph_osdc_alloc_request); 273 274 static void osd_req_encode_op(struct ceph_osd_request *req, 275 struct ceph_osd_op *dst, 276 struct ceph_osd_req_op *src) 277 { 278 dst->op = cpu_to_le16(src->op); 279 280 switch (dst->op) { 281 case CEPH_OSD_OP_READ: 282 case CEPH_OSD_OP_WRITE: 283 dst->extent.offset = 284 cpu_to_le64(src->extent.offset); 285 dst->extent.length = 286 cpu_to_le64(src->extent.length); 287 dst->extent.truncate_size = 288 cpu_to_le64(src->extent.truncate_size); 289 dst->extent.truncate_seq = 290 cpu_to_le32(src->extent.truncate_seq); 291 break; 292 293 case CEPH_OSD_OP_GETXATTR: 294 case CEPH_OSD_OP_SETXATTR: 295 case CEPH_OSD_OP_CMPXATTR: 296 BUG_ON(!req->r_trail); 297 298 dst->xattr.name_len = cpu_to_le32(src->xattr.name_len); 299 dst->xattr.value_len = cpu_to_le32(src->xattr.value_len); 300 dst->xattr.cmp_op = src->xattr.cmp_op; 301 dst->xattr.cmp_mode = src->xattr.cmp_mode; 302 ceph_pagelist_append(req->r_trail, src->xattr.name, 303 src->xattr.name_len); 304 ceph_pagelist_append(req->r_trail, src->xattr.val, 305 src->xattr.value_len); 306 break; 307 case CEPH_OSD_OP_CALL: 308 BUG_ON(!req->r_trail); 309 310 dst->cls.class_len = src->cls.class_len; 311 dst->cls.method_len = src->cls.method_len; 312 dst->cls.indata_len = cpu_to_le32(src->cls.indata_len); 313 314 ceph_pagelist_append(req->r_trail, src->cls.class_name, 315 src->cls.class_len); 316 ceph_pagelist_append(req->r_trail, src->cls.method_name, 317 src->cls.method_len); 318 ceph_pagelist_append(req->r_trail, src->cls.indata, 319 src->cls.indata_len); 320 break; 321 case CEPH_OSD_OP_ROLLBACK: 322 dst->snap.snapid = cpu_to_le64(src->snap.snapid); 323 break; 324 case CEPH_OSD_OP_STARTSYNC: 325 break; 326 case CEPH_OSD_OP_NOTIFY: 327 { 328 __le32 prot_ver = cpu_to_le32(src->watch.prot_ver); 329 __le32 timeout = cpu_to_le32(src->watch.timeout); 330 331 BUG_ON(!req->r_trail); 332 333 ceph_pagelist_append(req->r_trail, 334 &prot_ver, sizeof(prot_ver)); 335 ceph_pagelist_append(req->r_trail, 336 &timeout, sizeof(timeout)); 337 } 338 case CEPH_OSD_OP_NOTIFY_ACK: 339 case CEPH_OSD_OP_WATCH: 340 dst->watch.cookie = cpu_to_le64(src->watch.cookie); 341 dst->watch.ver = cpu_to_le64(src->watch.ver); 342 dst->watch.flag = src->watch.flag; 343 break; 344 default: 345 pr_err("unrecognized osd opcode %d\n", dst->op); 346 WARN_ON(1); 347 break; 348 } 349 dst->payload_len = cpu_to_le32(src->payload_len); 350 } 351 352 /* 353 * build new request AND message 354 * 355 */ 356 void ceph_osdc_build_request(struct ceph_osd_request *req, 357 u64 off, u64 *plen, 358 struct ceph_osd_req_op *src_ops, 359 struct ceph_snap_context *snapc, 360 struct timespec *mtime, 361 const char *oid, 362 int oid_len) 363 { 364 struct ceph_msg *msg = req->r_request; 365 struct ceph_osd_request_head *head; 366 struct ceph_osd_req_op *src_op; 367 struct ceph_osd_op *op; 368 void *p; 369 int num_op = get_num_ops(src_ops, NULL); 370 size_t msg_size = sizeof(*head) + num_op*sizeof(*op); 371 int flags = req->r_flags; 372 u64 data_len = 0; 373 int i; 374 375 head = msg->front.iov_base; 376 op = (void *)(head + 1); 377 p = (void *)(op + num_op); 378 379 req->r_snapc = ceph_get_snap_context(snapc); 380 381 head->client_inc = cpu_to_le32(1); /* always, for now. */ 382 head->flags = cpu_to_le32(flags); 383 if (flags & CEPH_OSD_FLAG_WRITE) 384 ceph_encode_timespec(&head->mtime, mtime); 385 head->num_ops = cpu_to_le16(num_op); 386 387 388 /* fill in oid */ 389 head->object_len = cpu_to_le32(oid_len); 390 memcpy(p, oid, oid_len); 391 p += oid_len; 392 393 src_op = src_ops; 394 while (src_op->op) { 395 osd_req_encode_op(req, op, src_op); 396 src_op++; 397 op++; 398 } 399 400 if (req->r_trail) 401 data_len += req->r_trail->length; 402 403 if (snapc) { 404 head->snap_seq = cpu_to_le64(snapc->seq); 405 head->num_snaps = cpu_to_le32(snapc->num_snaps); 406 for (i = 0; i < snapc->num_snaps; i++) { 407 put_unaligned_le64(snapc->snaps[i], p); 408 p += sizeof(u64); 409 } 410 } 411 412 if (flags & CEPH_OSD_FLAG_WRITE) { 413 req->r_request->hdr.data_off = cpu_to_le16(off); 414 req->r_request->hdr.data_len = cpu_to_le32(*plen + data_len); 415 } else if (data_len) { 416 req->r_request->hdr.data_off = 0; 417 req->r_request->hdr.data_len = cpu_to_le32(data_len); 418 } 419 420 req->r_request->page_alignment = req->r_page_alignment; 421 422 BUG_ON(p > msg->front.iov_base + msg->front.iov_len); 423 msg_size = p - msg->front.iov_base; 424 msg->front.iov_len = msg_size; 425 msg->hdr.front_len = cpu_to_le32(msg_size); 426 return; 427 } 428 EXPORT_SYMBOL(ceph_osdc_build_request); 429 430 /* 431 * build new request AND message, calculate layout, and adjust file 432 * extent as needed. 433 * 434 * if the file was recently truncated, we include information about its 435 * old and new size so that the object can be updated appropriately. (we 436 * avoid synchronously deleting truncated objects because it's slow.) 437 * 438 * if @do_sync, include a 'startsync' command so that the osd will flush 439 * data quickly. 440 */ 441 struct ceph_osd_request *ceph_osdc_new_request(struct ceph_osd_client *osdc, 442 struct ceph_file_layout *layout, 443 struct ceph_vino vino, 444 u64 off, u64 *plen, 445 int opcode, int flags, 446 struct ceph_snap_context *snapc, 447 int do_sync, 448 u32 truncate_seq, 449 u64 truncate_size, 450 struct timespec *mtime, 451 bool use_mempool, int num_reply, 452 int page_align) 453 { 454 struct ceph_osd_req_op ops[3]; 455 struct ceph_osd_request *req; 456 457 ops[0].op = opcode; 458 ops[0].extent.truncate_seq = truncate_seq; 459 ops[0].extent.truncate_size = truncate_size; 460 ops[0].payload_len = 0; 461 462 if (do_sync) { 463 ops[1].op = CEPH_OSD_OP_STARTSYNC; 464 ops[1].payload_len = 0; 465 ops[2].op = 0; 466 } else 467 ops[1].op = 0; 468 469 req = ceph_osdc_alloc_request(osdc, flags, 470 snapc, ops, 471 use_mempool, 472 GFP_NOFS, NULL, NULL); 473 if (!req) 474 return NULL; 475 476 /* calculate max write size */ 477 calc_layout(osdc, vino, layout, off, plen, req, ops); 478 req->r_file_layout = *layout; /* keep a copy */ 479 480 /* in case it differs from natural (file) alignment that 481 calc_layout filled in for us */ 482 req->r_num_pages = calc_pages_for(page_align, *plen); 483 req->r_page_alignment = page_align; 484 485 ceph_osdc_build_request(req, off, plen, ops, 486 snapc, 487 mtime, 488 req->r_oid, req->r_oid_len); 489 490 return req; 491 } 492 EXPORT_SYMBOL(ceph_osdc_new_request); 493 494 /* 495 * We keep osd requests in an rbtree, sorted by ->r_tid. 496 */ 497 static void __insert_request(struct ceph_osd_client *osdc, 498 struct ceph_osd_request *new) 499 { 500 struct rb_node **p = &osdc->requests.rb_node; 501 struct rb_node *parent = NULL; 502 struct ceph_osd_request *req = NULL; 503 504 while (*p) { 505 parent = *p; 506 req = rb_entry(parent, struct ceph_osd_request, r_node); 507 if (new->r_tid < req->r_tid) 508 p = &(*p)->rb_left; 509 else if (new->r_tid > req->r_tid) 510 p = &(*p)->rb_right; 511 else 512 BUG(); 513 } 514 515 rb_link_node(&new->r_node, parent, p); 516 rb_insert_color(&new->r_node, &osdc->requests); 517 } 518 519 static struct ceph_osd_request *__lookup_request(struct ceph_osd_client *osdc, 520 u64 tid) 521 { 522 struct ceph_osd_request *req; 523 struct rb_node *n = osdc->requests.rb_node; 524 525 while (n) { 526 req = rb_entry(n, struct ceph_osd_request, r_node); 527 if (tid < req->r_tid) 528 n = n->rb_left; 529 else if (tid > req->r_tid) 530 n = n->rb_right; 531 else 532 return req; 533 } 534 return NULL; 535 } 536 537 static struct ceph_osd_request * 538 __lookup_request_ge(struct ceph_osd_client *osdc, 539 u64 tid) 540 { 541 struct ceph_osd_request *req; 542 struct rb_node *n = osdc->requests.rb_node; 543 544 while (n) { 545 req = rb_entry(n, struct ceph_osd_request, r_node); 546 if (tid < req->r_tid) { 547 if (!n->rb_left) 548 return req; 549 n = n->rb_left; 550 } else if (tid > req->r_tid) { 551 n = n->rb_right; 552 } else { 553 return req; 554 } 555 } 556 return NULL; 557 } 558 559 /* 560 * Resubmit requests pending on the given osd. 561 */ 562 static void __kick_osd_requests(struct ceph_osd_client *osdc, 563 struct ceph_osd *osd) 564 { 565 struct ceph_osd_request *req, *nreq; 566 int err; 567 568 dout("__kick_osd_requests osd%d\n", osd->o_osd); 569 err = __reset_osd(osdc, osd); 570 if (err == -EAGAIN) 571 return; 572 573 list_for_each_entry(req, &osd->o_requests, r_osd_item) { 574 list_move(&req->r_req_lru_item, &osdc->req_unsent); 575 dout("requeued %p tid %llu osd%d\n", req, req->r_tid, 576 osd->o_osd); 577 if (!req->r_linger) 578 req->r_flags |= CEPH_OSD_FLAG_RETRY; 579 } 580 581 list_for_each_entry_safe(req, nreq, &osd->o_linger_requests, 582 r_linger_osd) { 583 /* 584 * reregister request prior to unregistering linger so 585 * that r_osd is preserved. 586 */ 587 BUG_ON(!list_empty(&req->r_req_lru_item)); 588 __register_request(osdc, req); 589 list_add(&req->r_req_lru_item, &osdc->req_unsent); 590 list_add(&req->r_osd_item, &req->r_osd->o_requests); 591 __unregister_linger_request(osdc, req); 592 dout("requeued lingering %p tid %llu osd%d\n", req, req->r_tid, 593 osd->o_osd); 594 } 595 } 596 597 static void kick_osd_requests(struct ceph_osd_client *osdc, 598 struct ceph_osd *kickosd) 599 { 600 mutex_lock(&osdc->request_mutex); 601 __kick_osd_requests(osdc, kickosd); 602 mutex_unlock(&osdc->request_mutex); 603 } 604 605 /* 606 * If the osd connection drops, we need to resubmit all requests. 607 */ 608 static void osd_reset(struct ceph_connection *con) 609 { 610 struct ceph_osd *osd = con->private; 611 struct ceph_osd_client *osdc; 612 613 if (!osd) 614 return; 615 dout("osd_reset osd%d\n", osd->o_osd); 616 osdc = osd->o_osdc; 617 down_read(&osdc->map_sem); 618 kick_osd_requests(osdc, osd); 619 send_queued(osdc); 620 up_read(&osdc->map_sem); 621 } 622 623 /* 624 * Track open sessions with osds. 625 */ 626 static struct ceph_osd *create_osd(struct ceph_osd_client *osdc) 627 { 628 struct ceph_osd *osd; 629 630 osd = kzalloc(sizeof(*osd), GFP_NOFS); 631 if (!osd) 632 return NULL; 633 634 atomic_set(&osd->o_ref, 1); 635 osd->o_osdc = osdc; 636 INIT_LIST_HEAD(&osd->o_requests); 637 INIT_LIST_HEAD(&osd->o_linger_requests); 638 INIT_LIST_HEAD(&osd->o_osd_lru); 639 osd->o_incarnation = 1; 640 641 ceph_con_init(osdc->client->msgr, &osd->o_con); 642 osd->o_con.private = osd; 643 osd->o_con.ops = &osd_con_ops; 644 osd->o_con.peer_name.type = CEPH_ENTITY_TYPE_OSD; 645 646 INIT_LIST_HEAD(&osd->o_keepalive_item); 647 return osd; 648 } 649 650 static struct ceph_osd *get_osd(struct ceph_osd *osd) 651 { 652 if (atomic_inc_not_zero(&osd->o_ref)) { 653 dout("get_osd %p %d -> %d\n", osd, atomic_read(&osd->o_ref)-1, 654 atomic_read(&osd->o_ref)); 655 return osd; 656 } else { 657 dout("get_osd %p FAIL\n", osd); 658 return NULL; 659 } 660 } 661 662 static void put_osd(struct ceph_osd *osd) 663 { 664 dout("put_osd %p %d -> %d\n", osd, atomic_read(&osd->o_ref), 665 atomic_read(&osd->o_ref) - 1); 666 if (atomic_dec_and_test(&osd->o_ref)) { 667 struct ceph_auth_client *ac = osd->o_osdc->client->monc.auth; 668 669 if (osd->o_authorizer) 670 ac->ops->destroy_authorizer(ac, osd->o_authorizer); 671 kfree(osd); 672 } 673 } 674 675 /* 676 * remove an osd from our map 677 */ 678 static void __remove_osd(struct ceph_osd_client *osdc, struct ceph_osd *osd) 679 { 680 dout("__remove_osd %p\n", osd); 681 BUG_ON(!list_empty(&osd->o_requests)); 682 rb_erase(&osd->o_node, &osdc->osds); 683 list_del_init(&osd->o_osd_lru); 684 ceph_con_close(&osd->o_con); 685 put_osd(osd); 686 } 687 688 static void __move_osd_to_lru(struct ceph_osd_client *osdc, 689 struct ceph_osd *osd) 690 { 691 dout("__move_osd_to_lru %p\n", osd); 692 BUG_ON(!list_empty(&osd->o_osd_lru)); 693 list_add_tail(&osd->o_osd_lru, &osdc->osd_lru); 694 osd->lru_ttl = jiffies + osdc->client->options->osd_idle_ttl * HZ; 695 } 696 697 static void __remove_osd_from_lru(struct ceph_osd *osd) 698 { 699 dout("__remove_osd_from_lru %p\n", osd); 700 if (!list_empty(&osd->o_osd_lru)) 701 list_del_init(&osd->o_osd_lru); 702 } 703 704 static void remove_old_osds(struct ceph_osd_client *osdc, int remove_all) 705 { 706 struct ceph_osd *osd, *nosd; 707 708 dout("__remove_old_osds %p\n", osdc); 709 mutex_lock(&osdc->request_mutex); 710 list_for_each_entry_safe(osd, nosd, &osdc->osd_lru, o_osd_lru) { 711 if (!remove_all && time_before(jiffies, osd->lru_ttl)) 712 break; 713 __remove_osd(osdc, osd); 714 } 715 mutex_unlock(&osdc->request_mutex); 716 } 717 718 /* 719 * reset osd connect 720 */ 721 static int __reset_osd(struct ceph_osd_client *osdc, struct ceph_osd *osd) 722 { 723 struct ceph_osd_request *req; 724 int ret = 0; 725 726 dout("__reset_osd %p osd%d\n", osd, osd->o_osd); 727 if (list_empty(&osd->o_requests) && 728 list_empty(&osd->o_linger_requests)) { 729 __remove_osd(osdc, osd); 730 } else if (memcmp(&osdc->osdmap->osd_addr[osd->o_osd], 731 &osd->o_con.peer_addr, 732 sizeof(osd->o_con.peer_addr)) == 0 && 733 !ceph_con_opened(&osd->o_con)) { 734 dout(" osd addr hasn't changed and connection never opened," 735 " letting msgr retry"); 736 /* touch each r_stamp for handle_timeout()'s benfit */ 737 list_for_each_entry(req, &osd->o_requests, r_osd_item) 738 req->r_stamp = jiffies; 739 ret = -EAGAIN; 740 } else { 741 ceph_con_close(&osd->o_con); 742 ceph_con_open(&osd->o_con, &osdc->osdmap->osd_addr[osd->o_osd]); 743 osd->o_incarnation++; 744 } 745 return ret; 746 } 747 748 static void __insert_osd(struct ceph_osd_client *osdc, struct ceph_osd *new) 749 { 750 struct rb_node **p = &osdc->osds.rb_node; 751 struct rb_node *parent = NULL; 752 struct ceph_osd *osd = NULL; 753 754 while (*p) { 755 parent = *p; 756 osd = rb_entry(parent, struct ceph_osd, o_node); 757 if (new->o_osd < osd->o_osd) 758 p = &(*p)->rb_left; 759 else if (new->o_osd > osd->o_osd) 760 p = &(*p)->rb_right; 761 else 762 BUG(); 763 } 764 765 rb_link_node(&new->o_node, parent, p); 766 rb_insert_color(&new->o_node, &osdc->osds); 767 } 768 769 static struct ceph_osd *__lookup_osd(struct ceph_osd_client *osdc, int o) 770 { 771 struct ceph_osd *osd; 772 struct rb_node *n = osdc->osds.rb_node; 773 774 while (n) { 775 osd = rb_entry(n, struct ceph_osd, o_node); 776 if (o < osd->o_osd) 777 n = n->rb_left; 778 else if (o > osd->o_osd) 779 n = n->rb_right; 780 else 781 return osd; 782 } 783 return NULL; 784 } 785 786 static void __schedule_osd_timeout(struct ceph_osd_client *osdc) 787 { 788 schedule_delayed_work(&osdc->timeout_work, 789 osdc->client->options->osd_keepalive_timeout * HZ); 790 } 791 792 static void __cancel_osd_timeout(struct ceph_osd_client *osdc) 793 { 794 cancel_delayed_work(&osdc->timeout_work); 795 } 796 797 /* 798 * Register request, assign tid. If this is the first request, set up 799 * the timeout event. 800 */ 801 static void __register_request(struct ceph_osd_client *osdc, 802 struct ceph_osd_request *req) 803 { 804 req->r_tid = ++osdc->last_tid; 805 req->r_request->hdr.tid = cpu_to_le64(req->r_tid); 806 INIT_LIST_HEAD(&req->r_req_lru_item); 807 808 dout("__register_request %p tid %lld\n", req, req->r_tid); 809 __insert_request(osdc, req); 810 ceph_osdc_get_request(req); 811 osdc->num_requests++; 812 813 if (osdc->num_requests == 1) { 814 dout(" first request, scheduling timeout\n"); 815 __schedule_osd_timeout(osdc); 816 } 817 } 818 819 static void register_request(struct ceph_osd_client *osdc, 820 struct ceph_osd_request *req) 821 { 822 mutex_lock(&osdc->request_mutex); 823 __register_request(osdc, req); 824 mutex_unlock(&osdc->request_mutex); 825 } 826 827 /* 828 * called under osdc->request_mutex 829 */ 830 static void __unregister_request(struct ceph_osd_client *osdc, 831 struct ceph_osd_request *req) 832 { 833 dout("__unregister_request %p tid %lld\n", req, req->r_tid); 834 rb_erase(&req->r_node, &osdc->requests); 835 osdc->num_requests--; 836 837 if (req->r_osd) { 838 /* make sure the original request isn't in flight. */ 839 ceph_con_revoke(&req->r_osd->o_con, req->r_request); 840 841 list_del_init(&req->r_osd_item); 842 if (list_empty(&req->r_osd->o_requests) && 843 list_empty(&req->r_osd->o_linger_requests)) { 844 dout("moving osd to %p lru\n", req->r_osd); 845 __move_osd_to_lru(osdc, req->r_osd); 846 } 847 if (list_empty(&req->r_linger_item)) 848 req->r_osd = NULL; 849 } 850 851 ceph_osdc_put_request(req); 852 853 list_del_init(&req->r_req_lru_item); 854 if (osdc->num_requests == 0) { 855 dout(" no requests, canceling timeout\n"); 856 __cancel_osd_timeout(osdc); 857 } 858 } 859 860 /* 861 * Cancel a previously queued request message 862 */ 863 static void __cancel_request(struct ceph_osd_request *req) 864 { 865 if (req->r_sent && req->r_osd) { 866 ceph_con_revoke(&req->r_osd->o_con, req->r_request); 867 req->r_sent = 0; 868 } 869 } 870 871 static void __register_linger_request(struct ceph_osd_client *osdc, 872 struct ceph_osd_request *req) 873 { 874 dout("__register_linger_request %p\n", req); 875 list_add_tail(&req->r_linger_item, &osdc->req_linger); 876 list_add_tail(&req->r_linger_osd, &req->r_osd->o_linger_requests); 877 } 878 879 static void __unregister_linger_request(struct ceph_osd_client *osdc, 880 struct ceph_osd_request *req) 881 { 882 dout("__unregister_linger_request %p\n", req); 883 if (req->r_osd) { 884 list_del_init(&req->r_linger_item); 885 list_del_init(&req->r_linger_osd); 886 887 if (list_empty(&req->r_osd->o_requests) && 888 list_empty(&req->r_osd->o_linger_requests)) { 889 dout("moving osd to %p lru\n", req->r_osd); 890 __move_osd_to_lru(osdc, req->r_osd); 891 } 892 if (list_empty(&req->r_osd_item)) 893 req->r_osd = NULL; 894 } 895 } 896 897 void ceph_osdc_unregister_linger_request(struct ceph_osd_client *osdc, 898 struct ceph_osd_request *req) 899 { 900 mutex_lock(&osdc->request_mutex); 901 if (req->r_linger) { 902 __unregister_linger_request(osdc, req); 903 ceph_osdc_put_request(req); 904 } 905 mutex_unlock(&osdc->request_mutex); 906 } 907 EXPORT_SYMBOL(ceph_osdc_unregister_linger_request); 908 909 void ceph_osdc_set_request_linger(struct ceph_osd_client *osdc, 910 struct ceph_osd_request *req) 911 { 912 if (!req->r_linger) { 913 dout("set_request_linger %p\n", req); 914 req->r_linger = 1; 915 /* 916 * caller is now responsible for calling 917 * unregister_linger_request 918 */ 919 ceph_osdc_get_request(req); 920 } 921 } 922 EXPORT_SYMBOL(ceph_osdc_set_request_linger); 923 924 /* 925 * Pick an osd (the first 'up' osd in the pg), allocate the osd struct 926 * (as needed), and set the request r_osd appropriately. If there is 927 * no up osd, set r_osd to NULL. Move the request to the appropriate list 928 * (unsent, homeless) or leave on in-flight lru. 929 * 930 * Return 0 if unchanged, 1 if changed, or negative on error. 931 * 932 * Caller should hold map_sem for read and request_mutex. 933 */ 934 static int __map_request(struct ceph_osd_client *osdc, 935 struct ceph_osd_request *req) 936 { 937 struct ceph_osd_request_head *reqhead = req->r_request->front.iov_base; 938 struct ceph_pg pgid; 939 int acting[CEPH_PG_MAX_SIZE]; 940 int o = -1, num = 0; 941 int err; 942 943 dout("map_request %p tid %lld\n", req, req->r_tid); 944 err = ceph_calc_object_layout(&reqhead->layout, req->r_oid, 945 &req->r_file_layout, osdc->osdmap); 946 if (err) { 947 list_move(&req->r_req_lru_item, &osdc->req_notarget); 948 return err; 949 } 950 pgid = reqhead->layout.ol_pgid; 951 req->r_pgid = pgid; 952 953 err = ceph_calc_pg_acting(osdc->osdmap, pgid, acting); 954 if (err > 0) { 955 o = acting[0]; 956 num = err; 957 } 958 959 if ((req->r_osd && req->r_osd->o_osd == o && 960 req->r_sent >= req->r_osd->o_incarnation && 961 req->r_num_pg_osds == num && 962 memcmp(req->r_pg_osds, acting, sizeof(acting[0])*num) == 0) || 963 (req->r_osd == NULL && o == -1)) 964 return 0; /* no change */ 965 966 dout("map_request tid %llu pgid %d.%x osd%d (was osd%d)\n", 967 req->r_tid, le32_to_cpu(pgid.pool), le16_to_cpu(pgid.ps), o, 968 req->r_osd ? req->r_osd->o_osd : -1); 969 970 /* record full pg acting set */ 971 memcpy(req->r_pg_osds, acting, sizeof(acting[0]) * num); 972 req->r_num_pg_osds = num; 973 974 if (req->r_osd) { 975 __cancel_request(req); 976 list_del_init(&req->r_osd_item); 977 req->r_osd = NULL; 978 } 979 980 req->r_osd = __lookup_osd(osdc, o); 981 if (!req->r_osd && o >= 0) { 982 err = -ENOMEM; 983 req->r_osd = create_osd(osdc); 984 if (!req->r_osd) { 985 list_move(&req->r_req_lru_item, &osdc->req_notarget); 986 goto out; 987 } 988 989 dout("map_request osd %p is osd%d\n", req->r_osd, o); 990 req->r_osd->o_osd = o; 991 req->r_osd->o_con.peer_name.num = cpu_to_le64(o); 992 __insert_osd(osdc, req->r_osd); 993 994 ceph_con_open(&req->r_osd->o_con, &osdc->osdmap->osd_addr[o]); 995 } 996 997 if (req->r_osd) { 998 __remove_osd_from_lru(req->r_osd); 999 list_add(&req->r_osd_item, &req->r_osd->o_requests); 1000 list_move(&req->r_req_lru_item, &osdc->req_unsent); 1001 } else { 1002 list_move(&req->r_req_lru_item, &osdc->req_notarget); 1003 } 1004 err = 1; /* osd or pg changed */ 1005 1006 out: 1007 return err; 1008 } 1009 1010 /* 1011 * caller should hold map_sem (for read) and request_mutex 1012 */ 1013 static int __send_request(struct ceph_osd_client *osdc, 1014 struct ceph_osd_request *req) 1015 { 1016 struct ceph_osd_request_head *reqhead; 1017 1018 dout("send_request %p tid %llu to osd%d flags %d\n", 1019 req, req->r_tid, req->r_osd->o_osd, req->r_flags); 1020 1021 reqhead = req->r_request->front.iov_base; 1022 reqhead->osdmap_epoch = cpu_to_le32(osdc->osdmap->epoch); 1023 reqhead->flags |= cpu_to_le32(req->r_flags); /* e.g., RETRY */ 1024 reqhead->reassert_version = req->r_reassert_version; 1025 1026 req->r_stamp = jiffies; 1027 list_move_tail(&req->r_req_lru_item, &osdc->req_lru); 1028 1029 ceph_msg_get(req->r_request); /* send consumes a ref */ 1030 ceph_con_send(&req->r_osd->o_con, req->r_request); 1031 req->r_sent = req->r_osd->o_incarnation; 1032 return 0; 1033 } 1034 1035 /* 1036 * Send any requests in the queue (req_unsent). 1037 */ 1038 static void send_queued(struct ceph_osd_client *osdc) 1039 { 1040 struct ceph_osd_request *req, *tmp; 1041 1042 dout("send_queued\n"); 1043 mutex_lock(&osdc->request_mutex); 1044 list_for_each_entry_safe(req, tmp, &osdc->req_unsent, r_req_lru_item) { 1045 __send_request(osdc, req); 1046 } 1047 mutex_unlock(&osdc->request_mutex); 1048 } 1049 1050 /* 1051 * Timeout callback, called every N seconds when 1 or more osd 1052 * requests has been active for more than N seconds. When this 1053 * happens, we ping all OSDs with requests who have timed out to 1054 * ensure any communications channel reset is detected. Reset the 1055 * request timeouts another N seconds in the future as we go. 1056 * Reschedule the timeout event another N seconds in future (unless 1057 * there are no open requests). 1058 */ 1059 static void handle_timeout(struct work_struct *work) 1060 { 1061 struct ceph_osd_client *osdc = 1062 container_of(work, struct ceph_osd_client, timeout_work.work); 1063 struct ceph_osd_request *req, *last_req = NULL; 1064 struct ceph_osd *osd; 1065 unsigned long timeout = osdc->client->options->osd_timeout * HZ; 1066 unsigned long keepalive = 1067 osdc->client->options->osd_keepalive_timeout * HZ; 1068 unsigned long last_stamp = 0; 1069 struct list_head slow_osds; 1070 dout("timeout\n"); 1071 down_read(&osdc->map_sem); 1072 1073 ceph_monc_request_next_osdmap(&osdc->client->monc); 1074 1075 mutex_lock(&osdc->request_mutex); 1076 1077 /* 1078 * reset osds that appear to be _really_ unresponsive. this 1079 * is a failsafe measure.. we really shouldn't be getting to 1080 * this point if the system is working properly. the monitors 1081 * should mark the osd as failed and we should find out about 1082 * it from an updated osd map. 1083 */ 1084 while (timeout && !list_empty(&osdc->req_lru)) { 1085 req = list_entry(osdc->req_lru.next, struct ceph_osd_request, 1086 r_req_lru_item); 1087 1088 /* hasn't been long enough since we sent it? */ 1089 if (time_before(jiffies, req->r_stamp + timeout)) 1090 break; 1091 1092 /* hasn't been long enough since it was acked? */ 1093 if (req->r_request->ack_stamp == 0 || 1094 time_before(jiffies, req->r_request->ack_stamp + timeout)) 1095 break; 1096 1097 BUG_ON(req == last_req && req->r_stamp == last_stamp); 1098 last_req = req; 1099 last_stamp = req->r_stamp; 1100 1101 osd = req->r_osd; 1102 BUG_ON(!osd); 1103 pr_warning(" tid %llu timed out on osd%d, will reset osd\n", 1104 req->r_tid, osd->o_osd); 1105 __kick_osd_requests(osdc, osd); 1106 } 1107 1108 /* 1109 * ping osds that are a bit slow. this ensures that if there 1110 * is a break in the TCP connection we will notice, and reopen 1111 * a connection with that osd (from the fault callback). 1112 */ 1113 INIT_LIST_HEAD(&slow_osds); 1114 list_for_each_entry(req, &osdc->req_lru, r_req_lru_item) { 1115 if (time_before(jiffies, req->r_stamp + keepalive)) 1116 break; 1117 1118 osd = req->r_osd; 1119 BUG_ON(!osd); 1120 dout(" tid %llu is slow, will send keepalive on osd%d\n", 1121 req->r_tid, osd->o_osd); 1122 list_move_tail(&osd->o_keepalive_item, &slow_osds); 1123 } 1124 while (!list_empty(&slow_osds)) { 1125 osd = list_entry(slow_osds.next, struct ceph_osd, 1126 o_keepalive_item); 1127 list_del_init(&osd->o_keepalive_item); 1128 ceph_con_keepalive(&osd->o_con); 1129 } 1130 1131 __schedule_osd_timeout(osdc); 1132 mutex_unlock(&osdc->request_mutex); 1133 send_queued(osdc); 1134 up_read(&osdc->map_sem); 1135 } 1136 1137 static void handle_osds_timeout(struct work_struct *work) 1138 { 1139 struct ceph_osd_client *osdc = 1140 container_of(work, struct ceph_osd_client, 1141 osds_timeout_work.work); 1142 unsigned long delay = 1143 osdc->client->options->osd_idle_ttl * HZ >> 2; 1144 1145 dout("osds timeout\n"); 1146 down_read(&osdc->map_sem); 1147 remove_old_osds(osdc, 0); 1148 up_read(&osdc->map_sem); 1149 1150 schedule_delayed_work(&osdc->osds_timeout_work, 1151 round_jiffies_relative(delay)); 1152 } 1153 1154 static void complete_request(struct ceph_osd_request *req) 1155 { 1156 if (req->r_safe_callback) 1157 req->r_safe_callback(req, NULL); 1158 complete_all(&req->r_safe_completion); /* fsync waiter */ 1159 } 1160 1161 /* 1162 * handle osd op reply. either call the callback if it is specified, 1163 * or do the completion to wake up the waiting thread. 1164 */ 1165 static void handle_reply(struct ceph_osd_client *osdc, struct ceph_msg *msg, 1166 struct ceph_connection *con) 1167 { 1168 struct ceph_osd_reply_head *rhead = msg->front.iov_base; 1169 struct ceph_osd_request *req; 1170 u64 tid; 1171 int numops, object_len, flags; 1172 s32 result; 1173 1174 tid = le64_to_cpu(msg->hdr.tid); 1175 if (msg->front.iov_len < sizeof(*rhead)) 1176 goto bad; 1177 numops = le32_to_cpu(rhead->num_ops); 1178 object_len = le32_to_cpu(rhead->object_len); 1179 result = le32_to_cpu(rhead->result); 1180 if (msg->front.iov_len != sizeof(*rhead) + object_len + 1181 numops * sizeof(struct ceph_osd_op)) 1182 goto bad; 1183 dout("handle_reply %p tid %llu result %d\n", msg, tid, (int)result); 1184 /* lookup */ 1185 mutex_lock(&osdc->request_mutex); 1186 req = __lookup_request(osdc, tid); 1187 if (req == NULL) { 1188 dout("handle_reply tid %llu dne\n", tid); 1189 mutex_unlock(&osdc->request_mutex); 1190 return; 1191 } 1192 ceph_osdc_get_request(req); 1193 flags = le32_to_cpu(rhead->flags); 1194 1195 /* 1196 * if this connection filled our message, drop our reference now, to 1197 * avoid a (safe but slower) revoke later. 1198 */ 1199 if (req->r_con_filling_msg == con && req->r_reply == msg) { 1200 dout(" dropping con_filling_msg ref %p\n", con); 1201 req->r_con_filling_msg = NULL; 1202 ceph_con_put(con); 1203 } 1204 1205 if (!req->r_got_reply) { 1206 unsigned bytes; 1207 1208 req->r_result = le32_to_cpu(rhead->result); 1209 bytes = le32_to_cpu(msg->hdr.data_len); 1210 dout("handle_reply result %d bytes %d\n", req->r_result, 1211 bytes); 1212 if (req->r_result == 0) 1213 req->r_result = bytes; 1214 1215 /* in case this is a write and we need to replay, */ 1216 req->r_reassert_version = rhead->reassert_version; 1217 1218 req->r_got_reply = 1; 1219 } else if ((flags & CEPH_OSD_FLAG_ONDISK) == 0) { 1220 dout("handle_reply tid %llu dup ack\n", tid); 1221 mutex_unlock(&osdc->request_mutex); 1222 goto done; 1223 } 1224 1225 dout("handle_reply tid %llu flags %d\n", tid, flags); 1226 1227 if (req->r_linger && (flags & CEPH_OSD_FLAG_ONDISK)) 1228 __register_linger_request(osdc, req); 1229 1230 /* either this is a read, or we got the safe response */ 1231 if (result < 0 || 1232 (flags & CEPH_OSD_FLAG_ONDISK) || 1233 ((flags & CEPH_OSD_FLAG_WRITE) == 0)) 1234 __unregister_request(osdc, req); 1235 1236 mutex_unlock(&osdc->request_mutex); 1237 1238 if (req->r_callback) 1239 req->r_callback(req, msg); 1240 else 1241 complete_all(&req->r_completion); 1242 1243 if (flags & CEPH_OSD_FLAG_ONDISK) 1244 complete_request(req); 1245 1246 done: 1247 dout("req=%p req->r_linger=%d\n", req, req->r_linger); 1248 ceph_osdc_put_request(req); 1249 return; 1250 1251 bad: 1252 pr_err("corrupt osd_op_reply got %d %d expected %d\n", 1253 (int)msg->front.iov_len, le32_to_cpu(msg->hdr.front_len), 1254 (int)sizeof(*rhead)); 1255 ceph_msg_dump(msg); 1256 } 1257 1258 static void reset_changed_osds(struct ceph_osd_client *osdc) 1259 { 1260 struct rb_node *p, *n; 1261 1262 for (p = rb_first(&osdc->osds); p; p = n) { 1263 struct ceph_osd *osd = rb_entry(p, struct ceph_osd, o_node); 1264 1265 n = rb_next(p); 1266 if (!ceph_osd_is_up(osdc->osdmap, osd->o_osd) || 1267 memcmp(&osd->o_con.peer_addr, 1268 ceph_osd_addr(osdc->osdmap, 1269 osd->o_osd), 1270 sizeof(struct ceph_entity_addr)) != 0) 1271 __reset_osd(osdc, osd); 1272 } 1273 } 1274 1275 /* 1276 * Requeue requests whose mapping to an OSD has changed. If requests map to 1277 * no osd, request a new map. 1278 * 1279 * Caller should hold map_sem for read and request_mutex. 1280 */ 1281 static void kick_requests(struct ceph_osd_client *osdc) 1282 { 1283 struct ceph_osd_request *req, *nreq; 1284 struct rb_node *p; 1285 int needmap = 0; 1286 int err; 1287 1288 dout("kick_requests\n"); 1289 mutex_lock(&osdc->request_mutex); 1290 for (p = rb_first(&osdc->requests); p; p = rb_next(p)) { 1291 req = rb_entry(p, struct ceph_osd_request, r_node); 1292 err = __map_request(osdc, req); 1293 if (err < 0) 1294 continue; /* error */ 1295 if (req->r_osd == NULL) { 1296 dout("%p tid %llu maps to no osd\n", req, req->r_tid); 1297 needmap++; /* request a newer map */ 1298 } else if (err > 0) { 1299 dout("%p tid %llu requeued on osd%d\n", req, req->r_tid, 1300 req->r_osd ? req->r_osd->o_osd : -1); 1301 if (!req->r_linger) 1302 req->r_flags |= CEPH_OSD_FLAG_RETRY; 1303 } 1304 } 1305 1306 list_for_each_entry_safe(req, nreq, &osdc->req_linger, 1307 r_linger_item) { 1308 dout("linger req=%p req->r_osd=%p\n", req, req->r_osd); 1309 1310 err = __map_request(osdc, req); 1311 if (err == 0) 1312 continue; /* no change and no osd was specified */ 1313 if (err < 0) 1314 continue; /* hrm! */ 1315 if (req->r_osd == NULL) { 1316 dout("tid %llu maps to no valid osd\n", req->r_tid); 1317 needmap++; /* request a newer map */ 1318 continue; 1319 } 1320 1321 dout("kicking lingering %p tid %llu osd%d\n", req, req->r_tid, 1322 req->r_osd ? req->r_osd->o_osd : -1); 1323 __unregister_linger_request(osdc, req); 1324 __register_request(osdc, req); 1325 } 1326 mutex_unlock(&osdc->request_mutex); 1327 1328 if (needmap) { 1329 dout("%d requests for down osds, need new map\n", needmap); 1330 ceph_monc_request_next_osdmap(&osdc->client->monc); 1331 } 1332 } 1333 1334 1335 /* 1336 * Process updated osd map. 1337 * 1338 * The message contains any number of incremental and full maps, normally 1339 * indicating some sort of topology change in the cluster. Kick requests 1340 * off to different OSDs as needed. 1341 */ 1342 void ceph_osdc_handle_map(struct ceph_osd_client *osdc, struct ceph_msg *msg) 1343 { 1344 void *p, *end, *next; 1345 u32 nr_maps, maplen; 1346 u32 epoch; 1347 struct ceph_osdmap *newmap = NULL, *oldmap; 1348 int err; 1349 struct ceph_fsid fsid; 1350 1351 dout("handle_map have %u\n", osdc->osdmap ? osdc->osdmap->epoch : 0); 1352 p = msg->front.iov_base; 1353 end = p + msg->front.iov_len; 1354 1355 /* verify fsid */ 1356 ceph_decode_need(&p, end, sizeof(fsid), bad); 1357 ceph_decode_copy(&p, &fsid, sizeof(fsid)); 1358 if (ceph_check_fsid(osdc->client, &fsid) < 0) 1359 return; 1360 1361 down_write(&osdc->map_sem); 1362 1363 /* incremental maps */ 1364 ceph_decode_32_safe(&p, end, nr_maps, bad); 1365 dout(" %d inc maps\n", nr_maps); 1366 while (nr_maps > 0) { 1367 ceph_decode_need(&p, end, 2*sizeof(u32), bad); 1368 epoch = ceph_decode_32(&p); 1369 maplen = ceph_decode_32(&p); 1370 ceph_decode_need(&p, end, maplen, bad); 1371 next = p + maplen; 1372 if (osdc->osdmap && osdc->osdmap->epoch+1 == epoch) { 1373 dout("applying incremental map %u len %d\n", 1374 epoch, maplen); 1375 newmap = osdmap_apply_incremental(&p, next, 1376 osdc->osdmap, 1377 osdc->client->msgr); 1378 if (IS_ERR(newmap)) { 1379 err = PTR_ERR(newmap); 1380 goto bad; 1381 } 1382 BUG_ON(!newmap); 1383 if (newmap != osdc->osdmap) { 1384 ceph_osdmap_destroy(osdc->osdmap); 1385 osdc->osdmap = newmap; 1386 } 1387 kick_requests(osdc); 1388 reset_changed_osds(osdc); 1389 } else { 1390 dout("ignoring incremental map %u len %d\n", 1391 epoch, maplen); 1392 } 1393 p = next; 1394 nr_maps--; 1395 } 1396 if (newmap) 1397 goto done; 1398 1399 /* full maps */ 1400 ceph_decode_32_safe(&p, end, nr_maps, bad); 1401 dout(" %d full maps\n", nr_maps); 1402 while (nr_maps) { 1403 ceph_decode_need(&p, end, 2*sizeof(u32), bad); 1404 epoch = ceph_decode_32(&p); 1405 maplen = ceph_decode_32(&p); 1406 ceph_decode_need(&p, end, maplen, bad); 1407 if (nr_maps > 1) { 1408 dout("skipping non-latest full map %u len %d\n", 1409 epoch, maplen); 1410 } else if (osdc->osdmap && osdc->osdmap->epoch >= epoch) { 1411 dout("skipping full map %u len %d, " 1412 "older than our %u\n", epoch, maplen, 1413 osdc->osdmap->epoch); 1414 } else { 1415 dout("taking full map %u len %d\n", epoch, maplen); 1416 newmap = osdmap_decode(&p, p+maplen); 1417 if (IS_ERR(newmap)) { 1418 err = PTR_ERR(newmap); 1419 goto bad; 1420 } 1421 BUG_ON(!newmap); 1422 oldmap = osdc->osdmap; 1423 osdc->osdmap = newmap; 1424 if (oldmap) 1425 ceph_osdmap_destroy(oldmap); 1426 kick_requests(osdc); 1427 } 1428 p += maplen; 1429 nr_maps--; 1430 } 1431 1432 done: 1433 downgrade_write(&osdc->map_sem); 1434 ceph_monc_got_osdmap(&osdc->client->monc, osdc->osdmap->epoch); 1435 1436 /* 1437 * subscribe to subsequent osdmap updates if full to ensure 1438 * we find out when we are no longer full and stop returning 1439 * ENOSPC. 1440 */ 1441 if (ceph_osdmap_flag(osdc->osdmap, CEPH_OSDMAP_FULL)) 1442 ceph_monc_request_next_osdmap(&osdc->client->monc); 1443 1444 send_queued(osdc); 1445 up_read(&osdc->map_sem); 1446 wake_up_all(&osdc->client->auth_wq); 1447 return; 1448 1449 bad: 1450 pr_err("osdc handle_map corrupt msg\n"); 1451 ceph_msg_dump(msg); 1452 up_write(&osdc->map_sem); 1453 return; 1454 } 1455 1456 /* 1457 * watch/notify callback event infrastructure 1458 * 1459 * These callbacks are used both for watch and notify operations. 1460 */ 1461 static void __release_event(struct kref *kref) 1462 { 1463 struct ceph_osd_event *event = 1464 container_of(kref, struct ceph_osd_event, kref); 1465 1466 dout("__release_event %p\n", event); 1467 kfree(event); 1468 } 1469 1470 static void get_event(struct ceph_osd_event *event) 1471 { 1472 kref_get(&event->kref); 1473 } 1474 1475 void ceph_osdc_put_event(struct ceph_osd_event *event) 1476 { 1477 kref_put(&event->kref, __release_event); 1478 } 1479 EXPORT_SYMBOL(ceph_osdc_put_event); 1480 1481 static void __insert_event(struct ceph_osd_client *osdc, 1482 struct ceph_osd_event *new) 1483 { 1484 struct rb_node **p = &osdc->event_tree.rb_node; 1485 struct rb_node *parent = NULL; 1486 struct ceph_osd_event *event = NULL; 1487 1488 while (*p) { 1489 parent = *p; 1490 event = rb_entry(parent, struct ceph_osd_event, node); 1491 if (new->cookie < event->cookie) 1492 p = &(*p)->rb_left; 1493 else if (new->cookie > event->cookie) 1494 p = &(*p)->rb_right; 1495 else 1496 BUG(); 1497 } 1498 1499 rb_link_node(&new->node, parent, p); 1500 rb_insert_color(&new->node, &osdc->event_tree); 1501 } 1502 1503 static struct ceph_osd_event *__find_event(struct ceph_osd_client *osdc, 1504 u64 cookie) 1505 { 1506 struct rb_node **p = &osdc->event_tree.rb_node; 1507 struct rb_node *parent = NULL; 1508 struct ceph_osd_event *event = NULL; 1509 1510 while (*p) { 1511 parent = *p; 1512 event = rb_entry(parent, struct ceph_osd_event, node); 1513 if (cookie < event->cookie) 1514 p = &(*p)->rb_left; 1515 else if (cookie > event->cookie) 1516 p = &(*p)->rb_right; 1517 else 1518 return event; 1519 } 1520 return NULL; 1521 } 1522 1523 static void __remove_event(struct ceph_osd_event *event) 1524 { 1525 struct ceph_osd_client *osdc = event->osdc; 1526 1527 if (!RB_EMPTY_NODE(&event->node)) { 1528 dout("__remove_event removed %p\n", event); 1529 rb_erase(&event->node, &osdc->event_tree); 1530 ceph_osdc_put_event(event); 1531 } else { 1532 dout("__remove_event didn't remove %p\n", event); 1533 } 1534 } 1535 1536 int ceph_osdc_create_event(struct ceph_osd_client *osdc, 1537 void (*event_cb)(u64, u64, u8, void *), 1538 int one_shot, void *data, 1539 struct ceph_osd_event **pevent) 1540 { 1541 struct ceph_osd_event *event; 1542 1543 event = kmalloc(sizeof(*event), GFP_NOIO); 1544 if (!event) 1545 return -ENOMEM; 1546 1547 dout("create_event %p\n", event); 1548 event->cb = event_cb; 1549 event->one_shot = one_shot; 1550 event->data = data; 1551 event->osdc = osdc; 1552 INIT_LIST_HEAD(&event->osd_node); 1553 kref_init(&event->kref); /* one ref for us */ 1554 kref_get(&event->kref); /* one ref for the caller */ 1555 init_completion(&event->completion); 1556 1557 spin_lock(&osdc->event_lock); 1558 event->cookie = ++osdc->event_count; 1559 __insert_event(osdc, event); 1560 spin_unlock(&osdc->event_lock); 1561 1562 *pevent = event; 1563 return 0; 1564 } 1565 EXPORT_SYMBOL(ceph_osdc_create_event); 1566 1567 void ceph_osdc_cancel_event(struct ceph_osd_event *event) 1568 { 1569 struct ceph_osd_client *osdc = event->osdc; 1570 1571 dout("cancel_event %p\n", event); 1572 spin_lock(&osdc->event_lock); 1573 __remove_event(event); 1574 spin_unlock(&osdc->event_lock); 1575 ceph_osdc_put_event(event); /* caller's */ 1576 } 1577 EXPORT_SYMBOL(ceph_osdc_cancel_event); 1578 1579 1580 static void do_event_work(struct work_struct *work) 1581 { 1582 struct ceph_osd_event_work *event_work = 1583 container_of(work, struct ceph_osd_event_work, work); 1584 struct ceph_osd_event *event = event_work->event; 1585 u64 ver = event_work->ver; 1586 u64 notify_id = event_work->notify_id; 1587 u8 opcode = event_work->opcode; 1588 1589 dout("do_event_work completing %p\n", event); 1590 event->cb(ver, notify_id, opcode, event->data); 1591 complete(&event->completion); 1592 dout("do_event_work completed %p\n", event); 1593 ceph_osdc_put_event(event); 1594 kfree(event_work); 1595 } 1596 1597 1598 /* 1599 * Process osd watch notifications 1600 */ 1601 void handle_watch_notify(struct ceph_osd_client *osdc, struct ceph_msg *msg) 1602 { 1603 void *p, *end; 1604 u8 proto_ver; 1605 u64 cookie, ver, notify_id; 1606 u8 opcode; 1607 struct ceph_osd_event *event; 1608 struct ceph_osd_event_work *event_work; 1609 1610 p = msg->front.iov_base; 1611 end = p + msg->front.iov_len; 1612 1613 ceph_decode_8_safe(&p, end, proto_ver, bad); 1614 ceph_decode_8_safe(&p, end, opcode, bad); 1615 ceph_decode_64_safe(&p, end, cookie, bad); 1616 ceph_decode_64_safe(&p, end, ver, bad); 1617 ceph_decode_64_safe(&p, end, notify_id, bad); 1618 1619 spin_lock(&osdc->event_lock); 1620 event = __find_event(osdc, cookie); 1621 if (event) { 1622 get_event(event); 1623 if (event->one_shot) 1624 __remove_event(event); 1625 } 1626 spin_unlock(&osdc->event_lock); 1627 dout("handle_watch_notify cookie %lld ver %lld event %p\n", 1628 cookie, ver, event); 1629 if (event) { 1630 event_work = kmalloc(sizeof(*event_work), GFP_NOIO); 1631 if (!event_work) { 1632 dout("ERROR: could not allocate event_work\n"); 1633 goto done_err; 1634 } 1635 INIT_WORK(&event_work->work, do_event_work); 1636 event_work->event = event; 1637 event_work->ver = ver; 1638 event_work->notify_id = notify_id; 1639 event_work->opcode = opcode; 1640 if (!queue_work(osdc->notify_wq, &event_work->work)) { 1641 dout("WARNING: failed to queue notify event work\n"); 1642 goto done_err; 1643 } 1644 } 1645 1646 return; 1647 1648 done_err: 1649 complete(&event->completion); 1650 ceph_osdc_put_event(event); 1651 return; 1652 1653 bad: 1654 pr_err("osdc handle_watch_notify corrupt msg\n"); 1655 return; 1656 } 1657 1658 int ceph_osdc_wait_event(struct ceph_osd_event *event, unsigned long timeout) 1659 { 1660 int err; 1661 1662 dout("wait_event %p\n", event); 1663 err = wait_for_completion_interruptible_timeout(&event->completion, 1664 timeout * HZ); 1665 ceph_osdc_put_event(event); 1666 if (err > 0) 1667 err = 0; 1668 dout("wait_event %p returns %d\n", event, err); 1669 return err; 1670 } 1671 EXPORT_SYMBOL(ceph_osdc_wait_event); 1672 1673 /* 1674 * Register request, send initial attempt. 1675 */ 1676 int ceph_osdc_start_request(struct ceph_osd_client *osdc, 1677 struct ceph_osd_request *req, 1678 bool nofail) 1679 { 1680 int rc = 0; 1681 1682 req->r_request->pages = req->r_pages; 1683 req->r_request->nr_pages = req->r_num_pages; 1684 #ifdef CONFIG_BLOCK 1685 req->r_request->bio = req->r_bio; 1686 #endif 1687 req->r_request->trail = req->r_trail; 1688 1689 register_request(osdc, req); 1690 1691 down_read(&osdc->map_sem); 1692 mutex_lock(&osdc->request_mutex); 1693 /* 1694 * a racing kick_requests() may have sent the message for us 1695 * while we dropped request_mutex above, so only send now if 1696 * the request still han't been touched yet. 1697 */ 1698 if (req->r_sent == 0) { 1699 rc = __map_request(osdc, req); 1700 if (rc < 0) { 1701 if (nofail) { 1702 dout("osdc_start_request failed map, " 1703 " will retry %lld\n", req->r_tid); 1704 rc = 0; 1705 } 1706 goto out_unlock; 1707 } 1708 if (req->r_osd == NULL) { 1709 dout("send_request %p no up osds in pg\n", req); 1710 ceph_monc_request_next_osdmap(&osdc->client->monc); 1711 } else { 1712 rc = __send_request(osdc, req); 1713 if (rc) { 1714 if (nofail) { 1715 dout("osdc_start_request failed send, " 1716 " will retry %lld\n", req->r_tid); 1717 rc = 0; 1718 } else { 1719 __unregister_request(osdc, req); 1720 } 1721 } 1722 } 1723 } 1724 1725 out_unlock: 1726 mutex_unlock(&osdc->request_mutex); 1727 up_read(&osdc->map_sem); 1728 return rc; 1729 } 1730 EXPORT_SYMBOL(ceph_osdc_start_request); 1731 1732 /* 1733 * wait for a request to complete 1734 */ 1735 int ceph_osdc_wait_request(struct ceph_osd_client *osdc, 1736 struct ceph_osd_request *req) 1737 { 1738 int rc; 1739 1740 rc = wait_for_completion_interruptible(&req->r_completion); 1741 if (rc < 0) { 1742 mutex_lock(&osdc->request_mutex); 1743 __cancel_request(req); 1744 __unregister_request(osdc, req); 1745 mutex_unlock(&osdc->request_mutex); 1746 complete_request(req); 1747 dout("wait_request tid %llu canceled/timed out\n", req->r_tid); 1748 return rc; 1749 } 1750 1751 dout("wait_request tid %llu result %d\n", req->r_tid, req->r_result); 1752 return req->r_result; 1753 } 1754 EXPORT_SYMBOL(ceph_osdc_wait_request); 1755 1756 /* 1757 * sync - wait for all in-flight requests to flush. avoid starvation. 1758 */ 1759 void ceph_osdc_sync(struct ceph_osd_client *osdc) 1760 { 1761 struct ceph_osd_request *req; 1762 u64 last_tid, next_tid = 0; 1763 1764 mutex_lock(&osdc->request_mutex); 1765 last_tid = osdc->last_tid; 1766 while (1) { 1767 req = __lookup_request_ge(osdc, next_tid); 1768 if (!req) 1769 break; 1770 if (req->r_tid > last_tid) 1771 break; 1772 1773 next_tid = req->r_tid + 1; 1774 if ((req->r_flags & CEPH_OSD_FLAG_WRITE) == 0) 1775 continue; 1776 1777 ceph_osdc_get_request(req); 1778 mutex_unlock(&osdc->request_mutex); 1779 dout("sync waiting on tid %llu (last is %llu)\n", 1780 req->r_tid, last_tid); 1781 wait_for_completion(&req->r_safe_completion); 1782 mutex_lock(&osdc->request_mutex); 1783 ceph_osdc_put_request(req); 1784 } 1785 mutex_unlock(&osdc->request_mutex); 1786 dout("sync done (thru tid %llu)\n", last_tid); 1787 } 1788 EXPORT_SYMBOL(ceph_osdc_sync); 1789 1790 /* 1791 * init, shutdown 1792 */ 1793 int ceph_osdc_init(struct ceph_osd_client *osdc, struct ceph_client *client) 1794 { 1795 int err; 1796 1797 dout("init\n"); 1798 osdc->client = client; 1799 osdc->osdmap = NULL; 1800 init_rwsem(&osdc->map_sem); 1801 init_completion(&osdc->map_waiters); 1802 osdc->last_requested_map = 0; 1803 mutex_init(&osdc->request_mutex); 1804 osdc->last_tid = 0; 1805 osdc->osds = RB_ROOT; 1806 INIT_LIST_HEAD(&osdc->osd_lru); 1807 osdc->requests = RB_ROOT; 1808 INIT_LIST_HEAD(&osdc->req_lru); 1809 INIT_LIST_HEAD(&osdc->req_unsent); 1810 INIT_LIST_HEAD(&osdc->req_notarget); 1811 INIT_LIST_HEAD(&osdc->req_linger); 1812 osdc->num_requests = 0; 1813 INIT_DELAYED_WORK(&osdc->timeout_work, handle_timeout); 1814 INIT_DELAYED_WORK(&osdc->osds_timeout_work, handle_osds_timeout); 1815 spin_lock_init(&osdc->event_lock); 1816 osdc->event_tree = RB_ROOT; 1817 osdc->event_count = 0; 1818 1819 schedule_delayed_work(&osdc->osds_timeout_work, 1820 round_jiffies_relative(osdc->client->options->osd_idle_ttl * HZ)); 1821 1822 err = -ENOMEM; 1823 osdc->req_mempool = mempool_create_kmalloc_pool(10, 1824 sizeof(struct ceph_osd_request)); 1825 if (!osdc->req_mempool) 1826 goto out; 1827 1828 err = ceph_msgpool_init(&osdc->msgpool_op, OSD_OP_FRONT_LEN, 10, true, 1829 "osd_op"); 1830 if (err < 0) 1831 goto out_mempool; 1832 err = ceph_msgpool_init(&osdc->msgpool_op_reply, 1833 OSD_OPREPLY_FRONT_LEN, 10, true, 1834 "osd_op_reply"); 1835 if (err < 0) 1836 goto out_msgpool; 1837 1838 osdc->notify_wq = create_singlethread_workqueue("ceph-watch-notify"); 1839 if (IS_ERR(osdc->notify_wq)) { 1840 err = PTR_ERR(osdc->notify_wq); 1841 osdc->notify_wq = NULL; 1842 goto out_msgpool; 1843 } 1844 return 0; 1845 1846 out_msgpool: 1847 ceph_msgpool_destroy(&osdc->msgpool_op); 1848 out_mempool: 1849 mempool_destroy(osdc->req_mempool); 1850 out: 1851 return err; 1852 } 1853 EXPORT_SYMBOL(ceph_osdc_init); 1854 1855 void ceph_osdc_stop(struct ceph_osd_client *osdc) 1856 { 1857 flush_workqueue(osdc->notify_wq); 1858 destroy_workqueue(osdc->notify_wq); 1859 cancel_delayed_work_sync(&osdc->timeout_work); 1860 cancel_delayed_work_sync(&osdc->osds_timeout_work); 1861 if (osdc->osdmap) { 1862 ceph_osdmap_destroy(osdc->osdmap); 1863 osdc->osdmap = NULL; 1864 } 1865 remove_old_osds(osdc, 1); 1866 WARN_ON(!RB_EMPTY_ROOT(&osdc->osds)); 1867 mempool_destroy(osdc->req_mempool); 1868 ceph_msgpool_destroy(&osdc->msgpool_op); 1869 ceph_msgpool_destroy(&osdc->msgpool_op_reply); 1870 } 1871 EXPORT_SYMBOL(ceph_osdc_stop); 1872 1873 /* 1874 * Read some contiguous pages. If we cross a stripe boundary, shorten 1875 * *plen. Return number of bytes read, or error. 1876 */ 1877 int ceph_osdc_readpages(struct ceph_osd_client *osdc, 1878 struct ceph_vino vino, struct ceph_file_layout *layout, 1879 u64 off, u64 *plen, 1880 u32 truncate_seq, u64 truncate_size, 1881 struct page **pages, int num_pages, int page_align) 1882 { 1883 struct ceph_osd_request *req; 1884 int rc = 0; 1885 1886 dout("readpages on ino %llx.%llx on %llu~%llu\n", vino.ino, 1887 vino.snap, off, *plen); 1888 req = ceph_osdc_new_request(osdc, layout, vino, off, plen, 1889 CEPH_OSD_OP_READ, CEPH_OSD_FLAG_READ, 1890 NULL, 0, truncate_seq, truncate_size, NULL, 1891 false, 1, page_align); 1892 if (!req) 1893 return -ENOMEM; 1894 1895 /* it may be a short read due to an object boundary */ 1896 req->r_pages = pages; 1897 1898 dout("readpages final extent is %llu~%llu (%d pages align %d)\n", 1899 off, *plen, req->r_num_pages, page_align); 1900 1901 rc = ceph_osdc_start_request(osdc, req, false); 1902 if (!rc) 1903 rc = ceph_osdc_wait_request(osdc, req); 1904 1905 ceph_osdc_put_request(req); 1906 dout("readpages result %d\n", rc); 1907 return rc; 1908 } 1909 EXPORT_SYMBOL(ceph_osdc_readpages); 1910 1911 /* 1912 * do a synchronous write on N pages 1913 */ 1914 int ceph_osdc_writepages(struct ceph_osd_client *osdc, struct ceph_vino vino, 1915 struct ceph_file_layout *layout, 1916 struct ceph_snap_context *snapc, 1917 u64 off, u64 len, 1918 u32 truncate_seq, u64 truncate_size, 1919 struct timespec *mtime, 1920 struct page **pages, int num_pages, 1921 int flags, int do_sync, bool nofail) 1922 { 1923 struct ceph_osd_request *req; 1924 int rc = 0; 1925 int page_align = off & ~PAGE_MASK; 1926 1927 BUG_ON(vino.snap != CEPH_NOSNAP); 1928 req = ceph_osdc_new_request(osdc, layout, vino, off, &len, 1929 CEPH_OSD_OP_WRITE, 1930 flags | CEPH_OSD_FLAG_ONDISK | 1931 CEPH_OSD_FLAG_WRITE, 1932 snapc, do_sync, 1933 truncate_seq, truncate_size, mtime, 1934 nofail, 1, page_align); 1935 if (!req) 1936 return -ENOMEM; 1937 1938 /* it may be a short write due to an object boundary */ 1939 req->r_pages = pages; 1940 dout("writepages %llu~%llu (%d pages)\n", off, len, 1941 req->r_num_pages); 1942 1943 rc = ceph_osdc_start_request(osdc, req, nofail); 1944 if (!rc) 1945 rc = ceph_osdc_wait_request(osdc, req); 1946 1947 ceph_osdc_put_request(req); 1948 if (rc == 0) 1949 rc = len; 1950 dout("writepages result %d\n", rc); 1951 return rc; 1952 } 1953 EXPORT_SYMBOL(ceph_osdc_writepages); 1954 1955 /* 1956 * handle incoming message 1957 */ 1958 static void dispatch(struct ceph_connection *con, struct ceph_msg *msg) 1959 { 1960 struct ceph_osd *osd = con->private; 1961 struct ceph_osd_client *osdc; 1962 int type = le16_to_cpu(msg->hdr.type); 1963 1964 if (!osd) 1965 goto out; 1966 osdc = osd->o_osdc; 1967 1968 switch (type) { 1969 case CEPH_MSG_OSD_MAP: 1970 ceph_osdc_handle_map(osdc, msg); 1971 break; 1972 case CEPH_MSG_OSD_OPREPLY: 1973 handle_reply(osdc, msg, con); 1974 break; 1975 case CEPH_MSG_WATCH_NOTIFY: 1976 handle_watch_notify(osdc, msg); 1977 break; 1978 1979 default: 1980 pr_err("received unknown message type %d %s\n", type, 1981 ceph_msg_type_name(type)); 1982 } 1983 out: 1984 ceph_msg_put(msg); 1985 } 1986 1987 /* 1988 * lookup and return message for incoming reply. set up reply message 1989 * pages. 1990 */ 1991 static struct ceph_msg *get_reply(struct ceph_connection *con, 1992 struct ceph_msg_header *hdr, 1993 int *skip) 1994 { 1995 struct ceph_osd *osd = con->private; 1996 struct ceph_osd_client *osdc = osd->o_osdc; 1997 struct ceph_msg *m; 1998 struct ceph_osd_request *req; 1999 int front = le32_to_cpu(hdr->front_len); 2000 int data_len = le32_to_cpu(hdr->data_len); 2001 u64 tid; 2002 2003 tid = le64_to_cpu(hdr->tid); 2004 mutex_lock(&osdc->request_mutex); 2005 req = __lookup_request(osdc, tid); 2006 if (!req) { 2007 *skip = 1; 2008 m = NULL; 2009 pr_info("get_reply unknown tid %llu from osd%d\n", tid, 2010 osd->o_osd); 2011 goto out; 2012 } 2013 2014 if (req->r_con_filling_msg) { 2015 dout("get_reply revoking msg %p from old con %p\n", 2016 req->r_reply, req->r_con_filling_msg); 2017 ceph_con_revoke_message(req->r_con_filling_msg, req->r_reply); 2018 ceph_con_put(req->r_con_filling_msg); 2019 req->r_con_filling_msg = NULL; 2020 } 2021 2022 if (front > req->r_reply->front.iov_len) { 2023 pr_warning("get_reply front %d > preallocated %d\n", 2024 front, (int)req->r_reply->front.iov_len); 2025 m = ceph_msg_new(CEPH_MSG_OSD_OPREPLY, front, GFP_NOFS); 2026 if (!m) 2027 goto out; 2028 ceph_msg_put(req->r_reply); 2029 req->r_reply = m; 2030 } 2031 m = ceph_msg_get(req->r_reply); 2032 2033 if (data_len > 0) { 2034 int want = calc_pages_for(req->r_page_alignment, data_len); 2035 2036 if (unlikely(req->r_num_pages < want)) { 2037 pr_warning("tid %lld reply has %d bytes %d pages, we" 2038 " had only %d pages ready\n", tid, data_len, 2039 want, req->r_num_pages); 2040 *skip = 1; 2041 ceph_msg_put(m); 2042 m = NULL; 2043 goto out; 2044 } 2045 m->pages = req->r_pages; 2046 m->nr_pages = req->r_num_pages; 2047 m->page_alignment = req->r_page_alignment; 2048 #ifdef CONFIG_BLOCK 2049 m->bio = req->r_bio; 2050 #endif 2051 } 2052 *skip = 0; 2053 req->r_con_filling_msg = ceph_con_get(con); 2054 dout("get_reply tid %lld %p\n", tid, m); 2055 2056 out: 2057 mutex_unlock(&osdc->request_mutex); 2058 return m; 2059 2060 } 2061 2062 static struct ceph_msg *alloc_msg(struct ceph_connection *con, 2063 struct ceph_msg_header *hdr, 2064 int *skip) 2065 { 2066 struct ceph_osd *osd = con->private; 2067 int type = le16_to_cpu(hdr->type); 2068 int front = le32_to_cpu(hdr->front_len); 2069 2070 switch (type) { 2071 case CEPH_MSG_OSD_MAP: 2072 case CEPH_MSG_WATCH_NOTIFY: 2073 return ceph_msg_new(type, front, GFP_NOFS); 2074 case CEPH_MSG_OSD_OPREPLY: 2075 return get_reply(con, hdr, skip); 2076 default: 2077 pr_info("alloc_msg unexpected msg type %d from osd%d\n", type, 2078 osd->o_osd); 2079 *skip = 1; 2080 return NULL; 2081 } 2082 } 2083 2084 /* 2085 * Wrappers to refcount containing ceph_osd struct 2086 */ 2087 static struct ceph_connection *get_osd_con(struct ceph_connection *con) 2088 { 2089 struct ceph_osd *osd = con->private; 2090 if (get_osd(osd)) 2091 return con; 2092 return NULL; 2093 } 2094 2095 static void put_osd_con(struct ceph_connection *con) 2096 { 2097 struct ceph_osd *osd = con->private; 2098 put_osd(osd); 2099 } 2100 2101 /* 2102 * authentication 2103 */ 2104 static int get_authorizer(struct ceph_connection *con, 2105 void **buf, int *len, int *proto, 2106 void **reply_buf, int *reply_len, int force_new) 2107 { 2108 struct ceph_osd *o = con->private; 2109 struct ceph_osd_client *osdc = o->o_osdc; 2110 struct ceph_auth_client *ac = osdc->client->monc.auth; 2111 int ret = 0; 2112 2113 if (force_new && o->o_authorizer) { 2114 ac->ops->destroy_authorizer(ac, o->o_authorizer); 2115 o->o_authorizer = NULL; 2116 } 2117 if (o->o_authorizer == NULL) { 2118 ret = ac->ops->create_authorizer( 2119 ac, CEPH_ENTITY_TYPE_OSD, 2120 &o->o_authorizer, 2121 &o->o_authorizer_buf, 2122 &o->o_authorizer_buf_len, 2123 &o->o_authorizer_reply_buf, 2124 &o->o_authorizer_reply_buf_len); 2125 if (ret) 2126 return ret; 2127 } 2128 2129 *proto = ac->protocol; 2130 *buf = o->o_authorizer_buf; 2131 *len = o->o_authorizer_buf_len; 2132 *reply_buf = o->o_authorizer_reply_buf; 2133 *reply_len = o->o_authorizer_reply_buf_len; 2134 return 0; 2135 } 2136 2137 2138 static int verify_authorizer_reply(struct ceph_connection *con, int len) 2139 { 2140 struct ceph_osd *o = con->private; 2141 struct ceph_osd_client *osdc = o->o_osdc; 2142 struct ceph_auth_client *ac = osdc->client->monc.auth; 2143 2144 return ac->ops->verify_authorizer_reply(ac, o->o_authorizer, len); 2145 } 2146 2147 static int invalidate_authorizer(struct ceph_connection *con) 2148 { 2149 struct ceph_osd *o = con->private; 2150 struct ceph_osd_client *osdc = o->o_osdc; 2151 struct ceph_auth_client *ac = osdc->client->monc.auth; 2152 2153 if (ac->ops->invalidate_authorizer) 2154 ac->ops->invalidate_authorizer(ac, CEPH_ENTITY_TYPE_OSD); 2155 2156 return ceph_monc_validate_auth(&osdc->client->monc); 2157 } 2158 2159 static const struct ceph_connection_operations osd_con_ops = { 2160 .get = get_osd_con, 2161 .put = put_osd_con, 2162 .dispatch = dispatch, 2163 .get_authorizer = get_authorizer, 2164 .verify_authorizer_reply = verify_authorizer_reply, 2165 .invalidate_authorizer = invalidate_authorizer, 2166 .alloc_msg = alloc_msg, 2167 .fault = osd_reset, 2168 }; 2169