1 /****************************************************************************** 2 * 3 * Back-end of the driver for virtual block devices. This portion of the 4 * driver exports a 'unified' block-device interface that can be accessed 5 * by any operating system that implements a compatible front end. A 6 * reference front-end implementation can be found in: 7 * drivers/block/xen-blkfront.c 8 * 9 * Copyright (c) 2003-2004, Keir Fraser & Steve Hand 10 * Copyright (c) 2005, Christopher Clark 11 * 12 * This program is free software; you can redistribute it and/or 13 * modify it under the terms of the GNU General Public License version 2 14 * as published by the Free Software Foundation; or, when distributed 15 * separately from the Linux kernel or incorporated into other 16 * software packages, subject to the following license: 17 * 18 * Permission is hereby granted, free of charge, to any person obtaining a copy 19 * of this source file (the "Software"), to deal in the Software without 20 * restriction, including without limitation the rights to use, copy, modify, 21 * merge, publish, distribute, sublicense, and/or sell copies of the Software, 22 * and to permit persons to whom the Software is furnished to do so, subject to 23 * the following conditions: 24 * 25 * The above copyright notice and this permission notice shall be included in 26 * all copies or substantial portions of the Software. 27 * 28 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR 29 * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, 30 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE 31 * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER 32 * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING 33 * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS 34 * IN THE SOFTWARE. 35 */ 36 37 #include <linux/spinlock.h> 38 #include <linux/kthread.h> 39 #include <linux/list.h> 40 #include <linux/delay.h> 41 #include <linux/freezer.h> 42 #include <linux/bitmap.h> 43 44 #include <xen/events.h> 45 #include <xen/page.h> 46 #include <xen/xen.h> 47 #include <asm/xen/hypervisor.h> 48 #include <asm/xen/hypercall.h> 49 #include <xen/balloon.h> 50 #include "common.h" 51 52 /* 53 * Maximum number of unused free pages to keep in the internal buffer. 54 * Setting this to a value too low will reduce memory used in each backend, 55 * but can have a performance penalty. 56 * 57 * A sane value is xen_blkif_reqs * BLKIF_MAX_SEGMENTS_PER_REQUEST, but can 58 * be set to a lower value that might degrade performance on some intensive 59 * IO workloads. 60 */ 61 62 static int xen_blkif_max_buffer_pages = 1024; 63 module_param_named(max_buffer_pages, xen_blkif_max_buffer_pages, int, 0644); 64 MODULE_PARM_DESC(max_buffer_pages, 65 "Maximum number of free pages to keep in each block backend buffer"); 66 67 /* 68 * Maximum number of grants to map persistently in blkback. For maximum 69 * performance this should be the total numbers of grants that can be used 70 * to fill the ring, but since this might become too high, specially with 71 * the use of indirect descriptors, we set it to a value that provides good 72 * performance without using too much memory. 73 * 74 * When the list of persistent grants is full we clean it up using a LRU 75 * algorithm. 76 */ 77 78 static int xen_blkif_max_pgrants = 1056; 79 module_param_named(max_persistent_grants, xen_blkif_max_pgrants, int, 0644); 80 MODULE_PARM_DESC(max_persistent_grants, 81 "Maximum number of grants to map persistently"); 82 83 /* 84 * The LRU mechanism to clean the lists of persistent grants needs to 85 * be executed periodically. The time interval between consecutive executions 86 * of the purge mechanism is set in ms. 87 */ 88 #define LRU_INTERVAL 100 89 90 /* 91 * When the persistent grants list is full we will remove unused grants 92 * from the list. The percent number of grants to be removed at each LRU 93 * execution. 94 */ 95 #define LRU_PERCENT_CLEAN 5 96 97 /* Run-time switchable: /sys/module/blkback/parameters/ */ 98 static unsigned int log_stats; 99 module_param(log_stats, int, 0644); 100 101 #define BLKBACK_INVALID_HANDLE (~0) 102 103 /* Number of free pages to remove on each call to free_xenballooned_pages */ 104 #define NUM_BATCH_FREE_PAGES 10 105 106 static inline int get_free_page(struct xen_blkif *blkif, struct page **page) 107 { 108 unsigned long flags; 109 110 spin_lock_irqsave(&blkif->free_pages_lock, flags); 111 if (list_empty(&blkif->free_pages)) { 112 BUG_ON(blkif->free_pages_num != 0); 113 spin_unlock_irqrestore(&blkif->free_pages_lock, flags); 114 return alloc_xenballooned_pages(1, page, false); 115 } 116 BUG_ON(blkif->free_pages_num == 0); 117 page[0] = list_first_entry(&blkif->free_pages, struct page, lru); 118 list_del(&page[0]->lru); 119 blkif->free_pages_num--; 120 spin_unlock_irqrestore(&blkif->free_pages_lock, flags); 121 122 return 0; 123 } 124 125 static inline void put_free_pages(struct xen_blkif *blkif, struct page **page, 126 int num) 127 { 128 unsigned long flags; 129 int i; 130 131 spin_lock_irqsave(&blkif->free_pages_lock, flags); 132 for (i = 0; i < num; i++) 133 list_add(&page[i]->lru, &blkif->free_pages); 134 blkif->free_pages_num += num; 135 spin_unlock_irqrestore(&blkif->free_pages_lock, flags); 136 } 137 138 static inline void shrink_free_pagepool(struct xen_blkif *blkif, int num) 139 { 140 /* Remove requested pages in batches of NUM_BATCH_FREE_PAGES */ 141 struct page *page[NUM_BATCH_FREE_PAGES]; 142 unsigned int num_pages = 0; 143 unsigned long flags; 144 145 spin_lock_irqsave(&blkif->free_pages_lock, flags); 146 while (blkif->free_pages_num > num) { 147 BUG_ON(list_empty(&blkif->free_pages)); 148 page[num_pages] = list_first_entry(&blkif->free_pages, 149 struct page, lru); 150 list_del(&page[num_pages]->lru); 151 blkif->free_pages_num--; 152 if (++num_pages == NUM_BATCH_FREE_PAGES) { 153 spin_unlock_irqrestore(&blkif->free_pages_lock, flags); 154 free_xenballooned_pages(num_pages, page); 155 spin_lock_irqsave(&blkif->free_pages_lock, flags); 156 num_pages = 0; 157 } 158 } 159 spin_unlock_irqrestore(&blkif->free_pages_lock, flags); 160 if (num_pages != 0) 161 free_xenballooned_pages(num_pages, page); 162 } 163 164 #define vaddr(page) ((unsigned long)pfn_to_kaddr(page_to_pfn(page))) 165 166 static int do_block_io_op(struct xen_blkif *blkif); 167 static int dispatch_rw_block_io(struct xen_blkif *blkif, 168 struct blkif_request *req, 169 struct pending_req *pending_req); 170 static void make_response(struct xen_blkif *blkif, u64 id, 171 unsigned short op, int st); 172 173 #define foreach_grant_safe(pos, n, rbtree, node) \ 174 for ((pos) = container_of(rb_first((rbtree)), typeof(*(pos)), node), \ 175 (n) = (&(pos)->node != NULL) ? rb_next(&(pos)->node) : NULL; \ 176 &(pos)->node != NULL; \ 177 (pos) = container_of(n, typeof(*(pos)), node), \ 178 (n) = (&(pos)->node != NULL) ? rb_next(&(pos)->node) : NULL) 179 180 181 /* 182 * We don't need locking around the persistent grant helpers 183 * because blkback uses a single-thread for each backed, so we 184 * can be sure that this functions will never be called recursively. 185 * 186 * The only exception to that is put_persistent_grant, that can be called 187 * from interrupt context (by xen_blkbk_unmap), so we have to use atomic 188 * bit operations to modify the flags of a persistent grant and to count 189 * the number of used grants. 190 */ 191 static int add_persistent_gnt(struct xen_blkif *blkif, 192 struct persistent_gnt *persistent_gnt) 193 { 194 struct rb_node **new = NULL, *parent = NULL; 195 struct persistent_gnt *this; 196 197 if (blkif->persistent_gnt_c >= xen_blkif_max_pgrants) { 198 if (!blkif->vbd.overflow_max_grants) 199 blkif->vbd.overflow_max_grants = 1; 200 return -EBUSY; 201 } 202 /* Figure out where to put new node */ 203 new = &blkif->persistent_gnts.rb_node; 204 while (*new) { 205 this = container_of(*new, struct persistent_gnt, node); 206 207 parent = *new; 208 if (persistent_gnt->gnt < this->gnt) 209 new = &((*new)->rb_left); 210 else if (persistent_gnt->gnt > this->gnt) 211 new = &((*new)->rb_right); 212 else { 213 pr_alert_ratelimited(DRV_PFX " trying to add a gref that's already in the tree\n"); 214 return -EINVAL; 215 } 216 } 217 218 bitmap_zero(persistent_gnt->flags, PERSISTENT_GNT_FLAGS_SIZE); 219 set_bit(PERSISTENT_GNT_ACTIVE, persistent_gnt->flags); 220 /* Add new node and rebalance tree. */ 221 rb_link_node(&(persistent_gnt->node), parent, new); 222 rb_insert_color(&(persistent_gnt->node), &blkif->persistent_gnts); 223 blkif->persistent_gnt_c++; 224 atomic_inc(&blkif->persistent_gnt_in_use); 225 return 0; 226 } 227 228 static struct persistent_gnt *get_persistent_gnt(struct xen_blkif *blkif, 229 grant_ref_t gref) 230 { 231 struct persistent_gnt *data; 232 struct rb_node *node = NULL; 233 234 node = blkif->persistent_gnts.rb_node; 235 while (node) { 236 data = container_of(node, struct persistent_gnt, node); 237 238 if (gref < data->gnt) 239 node = node->rb_left; 240 else if (gref > data->gnt) 241 node = node->rb_right; 242 else { 243 if(test_bit(PERSISTENT_GNT_ACTIVE, data->flags)) { 244 pr_alert_ratelimited(DRV_PFX " requesting a grant already in use\n"); 245 return NULL; 246 } 247 set_bit(PERSISTENT_GNT_ACTIVE, data->flags); 248 atomic_inc(&blkif->persistent_gnt_in_use); 249 return data; 250 } 251 } 252 return NULL; 253 } 254 255 static void put_persistent_gnt(struct xen_blkif *blkif, 256 struct persistent_gnt *persistent_gnt) 257 { 258 if(!test_bit(PERSISTENT_GNT_ACTIVE, persistent_gnt->flags)) 259 pr_alert_ratelimited(DRV_PFX " freeing a grant already unused"); 260 set_bit(PERSISTENT_GNT_WAS_ACTIVE, persistent_gnt->flags); 261 clear_bit(PERSISTENT_GNT_ACTIVE, persistent_gnt->flags); 262 atomic_dec(&blkif->persistent_gnt_in_use); 263 } 264 265 static void free_persistent_gnts(struct xen_blkif *blkif, struct rb_root *root, 266 unsigned int num) 267 { 268 struct gnttab_unmap_grant_ref unmap[BLKIF_MAX_SEGMENTS_PER_REQUEST]; 269 struct page *pages[BLKIF_MAX_SEGMENTS_PER_REQUEST]; 270 struct persistent_gnt *persistent_gnt; 271 struct rb_node *n; 272 int ret = 0; 273 int segs_to_unmap = 0; 274 275 foreach_grant_safe(persistent_gnt, n, root, node) { 276 BUG_ON(persistent_gnt->handle == 277 BLKBACK_INVALID_HANDLE); 278 gnttab_set_unmap_op(&unmap[segs_to_unmap], 279 (unsigned long) pfn_to_kaddr(page_to_pfn( 280 persistent_gnt->page)), 281 GNTMAP_host_map, 282 persistent_gnt->handle); 283 284 pages[segs_to_unmap] = persistent_gnt->page; 285 286 if (++segs_to_unmap == BLKIF_MAX_SEGMENTS_PER_REQUEST || 287 !rb_next(&persistent_gnt->node)) { 288 ret = gnttab_unmap_refs(unmap, NULL, pages, 289 segs_to_unmap); 290 BUG_ON(ret); 291 put_free_pages(blkif, pages, segs_to_unmap); 292 segs_to_unmap = 0; 293 } 294 295 rb_erase(&persistent_gnt->node, root); 296 kfree(persistent_gnt); 297 num--; 298 } 299 BUG_ON(num != 0); 300 } 301 302 void xen_blkbk_unmap_purged_grants(struct work_struct *work) 303 { 304 struct gnttab_unmap_grant_ref unmap[BLKIF_MAX_SEGMENTS_PER_REQUEST]; 305 struct page *pages[BLKIF_MAX_SEGMENTS_PER_REQUEST]; 306 struct persistent_gnt *persistent_gnt; 307 int ret, segs_to_unmap = 0; 308 struct xen_blkif *blkif = container_of(work, typeof(*blkif), persistent_purge_work); 309 310 while(!list_empty(&blkif->persistent_purge_list)) { 311 persistent_gnt = list_first_entry(&blkif->persistent_purge_list, 312 struct persistent_gnt, 313 remove_node); 314 list_del(&persistent_gnt->remove_node); 315 316 gnttab_set_unmap_op(&unmap[segs_to_unmap], 317 vaddr(persistent_gnt->page), 318 GNTMAP_host_map, 319 persistent_gnt->handle); 320 321 pages[segs_to_unmap] = persistent_gnt->page; 322 323 if (++segs_to_unmap == BLKIF_MAX_SEGMENTS_PER_REQUEST) { 324 ret = gnttab_unmap_refs(unmap, NULL, pages, 325 segs_to_unmap); 326 BUG_ON(ret); 327 put_free_pages(blkif, pages, segs_to_unmap); 328 segs_to_unmap = 0; 329 } 330 kfree(persistent_gnt); 331 } 332 if (segs_to_unmap > 0) { 333 ret = gnttab_unmap_refs(unmap, NULL, pages, segs_to_unmap); 334 BUG_ON(ret); 335 put_free_pages(blkif, pages, segs_to_unmap); 336 } 337 } 338 339 static void purge_persistent_gnt(struct xen_blkif *blkif) 340 { 341 struct persistent_gnt *persistent_gnt; 342 struct rb_node *n; 343 unsigned int num_clean, total; 344 bool scan_used = false, clean_used = false; 345 struct rb_root *root; 346 347 if (blkif->persistent_gnt_c < xen_blkif_max_pgrants || 348 (blkif->persistent_gnt_c == xen_blkif_max_pgrants && 349 !blkif->vbd.overflow_max_grants)) { 350 return; 351 } 352 353 if (work_pending(&blkif->persistent_purge_work)) { 354 pr_alert_ratelimited(DRV_PFX "Scheduled work from previous purge is still pending, cannot purge list\n"); 355 return; 356 } 357 358 num_clean = (xen_blkif_max_pgrants / 100) * LRU_PERCENT_CLEAN; 359 num_clean = blkif->persistent_gnt_c - xen_blkif_max_pgrants + num_clean; 360 num_clean = min(blkif->persistent_gnt_c, num_clean); 361 if ((num_clean == 0) || 362 (num_clean > (blkif->persistent_gnt_c - atomic_read(&blkif->persistent_gnt_in_use)))) 363 return; 364 365 /* 366 * At this point, we can assure that there will be no calls 367 * to get_persistent_grant (because we are executing this code from 368 * xen_blkif_schedule), there can only be calls to put_persistent_gnt, 369 * which means that the number of currently used grants will go down, 370 * but never up, so we will always be able to remove the requested 371 * number of grants. 372 */ 373 374 total = num_clean; 375 376 pr_debug(DRV_PFX "Going to purge %u persistent grants\n", num_clean); 377 378 BUG_ON(!list_empty(&blkif->persistent_purge_list)); 379 root = &blkif->persistent_gnts; 380 purge_list: 381 foreach_grant_safe(persistent_gnt, n, root, node) { 382 BUG_ON(persistent_gnt->handle == 383 BLKBACK_INVALID_HANDLE); 384 385 if (clean_used) { 386 clear_bit(PERSISTENT_GNT_WAS_ACTIVE, persistent_gnt->flags); 387 continue; 388 } 389 390 if (test_bit(PERSISTENT_GNT_ACTIVE, persistent_gnt->flags)) 391 continue; 392 if (!scan_used && 393 (test_bit(PERSISTENT_GNT_WAS_ACTIVE, persistent_gnt->flags))) 394 continue; 395 396 rb_erase(&persistent_gnt->node, root); 397 list_add(&persistent_gnt->remove_node, 398 &blkif->persistent_purge_list); 399 if (--num_clean == 0) 400 goto finished; 401 } 402 /* 403 * If we get here it means we also need to start cleaning 404 * grants that were used since last purge in order to cope 405 * with the requested num 406 */ 407 if (!scan_used && !clean_used) { 408 pr_debug(DRV_PFX "Still missing %u purged frames\n", num_clean); 409 scan_used = true; 410 goto purge_list; 411 } 412 finished: 413 if (!clean_used) { 414 pr_debug(DRV_PFX "Finished scanning for grants to clean, removing used flag\n"); 415 clean_used = true; 416 goto purge_list; 417 } 418 419 blkif->persistent_gnt_c -= (total - num_clean); 420 blkif->vbd.overflow_max_grants = 0; 421 422 /* We can defer this work */ 423 schedule_work(&blkif->persistent_purge_work); 424 pr_debug(DRV_PFX "Purged %u/%u\n", (total - num_clean), total); 425 return; 426 } 427 428 /* 429 * Retrieve from the 'pending_reqs' a free pending_req structure to be used. 430 */ 431 static struct pending_req *alloc_req(struct xen_blkif *blkif) 432 { 433 struct pending_req *req = NULL; 434 unsigned long flags; 435 436 spin_lock_irqsave(&blkif->pending_free_lock, flags); 437 if (!list_empty(&blkif->pending_free)) { 438 req = list_entry(blkif->pending_free.next, struct pending_req, 439 free_list); 440 list_del(&req->free_list); 441 } 442 spin_unlock_irqrestore(&blkif->pending_free_lock, flags); 443 return req; 444 } 445 446 /* 447 * Return the 'pending_req' structure back to the freepool. We also 448 * wake up the thread if it was waiting for a free page. 449 */ 450 static void free_req(struct xen_blkif *blkif, struct pending_req *req) 451 { 452 unsigned long flags; 453 int was_empty; 454 455 spin_lock_irqsave(&blkif->pending_free_lock, flags); 456 was_empty = list_empty(&blkif->pending_free); 457 list_add(&req->free_list, &blkif->pending_free); 458 spin_unlock_irqrestore(&blkif->pending_free_lock, flags); 459 if (was_empty) 460 wake_up(&blkif->pending_free_wq); 461 } 462 463 /* 464 * Routines for managing virtual block devices (vbds). 465 */ 466 static int xen_vbd_translate(struct phys_req *req, struct xen_blkif *blkif, 467 int operation) 468 { 469 struct xen_vbd *vbd = &blkif->vbd; 470 int rc = -EACCES; 471 472 if ((operation != READ) && vbd->readonly) 473 goto out; 474 475 if (likely(req->nr_sects)) { 476 blkif_sector_t end = req->sector_number + req->nr_sects; 477 478 if (unlikely(end < req->sector_number)) 479 goto out; 480 if (unlikely(end > vbd_sz(vbd))) 481 goto out; 482 } 483 484 req->dev = vbd->pdevice; 485 req->bdev = vbd->bdev; 486 rc = 0; 487 488 out: 489 return rc; 490 } 491 492 static void xen_vbd_resize(struct xen_blkif *blkif) 493 { 494 struct xen_vbd *vbd = &blkif->vbd; 495 struct xenbus_transaction xbt; 496 int err; 497 struct xenbus_device *dev = xen_blkbk_xenbus(blkif->be); 498 unsigned long long new_size = vbd_sz(vbd); 499 500 pr_info(DRV_PFX "VBD Resize: Domid: %d, Device: (%d, %d)\n", 501 blkif->domid, MAJOR(vbd->pdevice), MINOR(vbd->pdevice)); 502 pr_info(DRV_PFX "VBD Resize: new size %llu\n", new_size); 503 vbd->size = new_size; 504 again: 505 err = xenbus_transaction_start(&xbt); 506 if (err) { 507 pr_warn(DRV_PFX "Error starting transaction"); 508 return; 509 } 510 err = xenbus_printf(xbt, dev->nodename, "sectors", "%llu", 511 (unsigned long long)vbd_sz(vbd)); 512 if (err) { 513 pr_warn(DRV_PFX "Error writing new size"); 514 goto abort; 515 } 516 /* 517 * Write the current state; we will use this to synchronize 518 * the front-end. If the current state is "connected" the 519 * front-end will get the new size information online. 520 */ 521 err = xenbus_printf(xbt, dev->nodename, "state", "%d", dev->state); 522 if (err) { 523 pr_warn(DRV_PFX "Error writing the state"); 524 goto abort; 525 } 526 527 err = xenbus_transaction_end(xbt, 0); 528 if (err == -EAGAIN) 529 goto again; 530 if (err) 531 pr_warn(DRV_PFX "Error ending transaction"); 532 return; 533 abort: 534 xenbus_transaction_end(xbt, 1); 535 } 536 537 /* 538 * Notification from the guest OS. 539 */ 540 static void blkif_notify_work(struct xen_blkif *blkif) 541 { 542 blkif->waiting_reqs = 1; 543 wake_up(&blkif->wq); 544 } 545 546 irqreturn_t xen_blkif_be_int(int irq, void *dev_id) 547 { 548 blkif_notify_work(dev_id); 549 return IRQ_HANDLED; 550 } 551 552 /* 553 * SCHEDULER FUNCTIONS 554 */ 555 556 static void print_stats(struct xen_blkif *blkif) 557 { 558 pr_info("xen-blkback (%s): oo %3llu | rd %4llu | wr %4llu | f %4llu" 559 " | ds %4llu | pg: %4u/%4d\n", 560 current->comm, blkif->st_oo_req, 561 blkif->st_rd_req, blkif->st_wr_req, 562 blkif->st_f_req, blkif->st_ds_req, 563 blkif->persistent_gnt_c, 564 xen_blkif_max_pgrants); 565 blkif->st_print = jiffies + msecs_to_jiffies(10 * 1000); 566 blkif->st_rd_req = 0; 567 blkif->st_wr_req = 0; 568 blkif->st_oo_req = 0; 569 blkif->st_ds_req = 0; 570 } 571 572 int xen_blkif_schedule(void *arg) 573 { 574 struct xen_blkif *blkif = arg; 575 struct xen_vbd *vbd = &blkif->vbd; 576 unsigned long timeout; 577 int ret; 578 579 xen_blkif_get(blkif); 580 581 while (!kthread_should_stop()) { 582 if (try_to_freeze()) 583 continue; 584 if (unlikely(vbd->size != vbd_sz(vbd))) 585 xen_vbd_resize(blkif); 586 587 timeout = msecs_to_jiffies(LRU_INTERVAL); 588 589 timeout = wait_event_interruptible_timeout( 590 blkif->wq, 591 blkif->waiting_reqs || kthread_should_stop(), 592 timeout); 593 if (timeout == 0) 594 goto purge_gnt_list; 595 timeout = wait_event_interruptible_timeout( 596 blkif->pending_free_wq, 597 !list_empty(&blkif->pending_free) || 598 kthread_should_stop(), 599 timeout); 600 if (timeout == 0) 601 goto purge_gnt_list; 602 603 blkif->waiting_reqs = 0; 604 smp_mb(); /* clear flag *before* checking for work */ 605 606 ret = do_block_io_op(blkif); 607 if (ret > 0) 608 blkif->waiting_reqs = 1; 609 if (ret == -EACCES) 610 wait_event_interruptible(blkif->shutdown_wq, 611 kthread_should_stop()); 612 613 purge_gnt_list: 614 if (blkif->vbd.feature_gnt_persistent && 615 time_after(jiffies, blkif->next_lru)) { 616 purge_persistent_gnt(blkif); 617 blkif->next_lru = jiffies + msecs_to_jiffies(LRU_INTERVAL); 618 } 619 620 /* Shrink if we have more than xen_blkif_max_buffer_pages */ 621 shrink_free_pagepool(blkif, xen_blkif_max_buffer_pages); 622 623 if (log_stats && time_after(jiffies, blkif->st_print)) 624 print_stats(blkif); 625 } 626 627 /* Drain pending purge work */ 628 flush_work(&blkif->persistent_purge_work); 629 630 if (log_stats) 631 print_stats(blkif); 632 633 blkif->xenblkd = NULL; 634 xen_blkif_put(blkif); 635 636 return 0; 637 } 638 639 /* 640 * Remove persistent grants and empty the pool of free pages 641 */ 642 void xen_blkbk_free_caches(struct xen_blkif *blkif) 643 { 644 /* Free all persistent grant pages */ 645 if (!RB_EMPTY_ROOT(&blkif->persistent_gnts)) 646 free_persistent_gnts(blkif, &blkif->persistent_gnts, 647 blkif->persistent_gnt_c); 648 649 BUG_ON(!RB_EMPTY_ROOT(&blkif->persistent_gnts)); 650 blkif->persistent_gnt_c = 0; 651 652 /* Since we are shutting down remove all pages from the buffer */ 653 shrink_free_pagepool(blkif, 0 /* All */); 654 } 655 656 /* 657 * Unmap the grant references, and also remove the M2P over-rides 658 * used in the 'pending_req'. 659 */ 660 static void xen_blkbk_unmap(struct xen_blkif *blkif, 661 struct grant_page *pages[], 662 int num) 663 { 664 struct gnttab_unmap_grant_ref unmap[BLKIF_MAX_SEGMENTS_PER_REQUEST]; 665 struct page *unmap_pages[BLKIF_MAX_SEGMENTS_PER_REQUEST]; 666 unsigned int i, invcount = 0; 667 int ret; 668 669 for (i = 0; i < num; i++) { 670 if (pages[i]->persistent_gnt != NULL) { 671 put_persistent_gnt(blkif, pages[i]->persistent_gnt); 672 continue; 673 } 674 if (pages[i]->handle == BLKBACK_INVALID_HANDLE) 675 continue; 676 unmap_pages[invcount] = pages[i]->page; 677 gnttab_set_unmap_op(&unmap[invcount], vaddr(pages[i]->page), 678 GNTMAP_host_map, pages[i]->handle); 679 pages[i]->handle = BLKBACK_INVALID_HANDLE; 680 if (++invcount == BLKIF_MAX_SEGMENTS_PER_REQUEST) { 681 ret = gnttab_unmap_refs(unmap, NULL, unmap_pages, 682 invcount); 683 BUG_ON(ret); 684 put_free_pages(blkif, unmap_pages, invcount); 685 invcount = 0; 686 } 687 } 688 if (invcount) { 689 ret = gnttab_unmap_refs(unmap, NULL, unmap_pages, invcount); 690 BUG_ON(ret); 691 put_free_pages(blkif, unmap_pages, invcount); 692 } 693 } 694 695 static int xen_blkbk_map(struct xen_blkif *blkif, 696 struct grant_page *pages[], 697 int num, bool ro) 698 { 699 struct gnttab_map_grant_ref map[BLKIF_MAX_SEGMENTS_PER_REQUEST]; 700 struct page *pages_to_gnt[BLKIF_MAX_SEGMENTS_PER_REQUEST]; 701 struct persistent_gnt *persistent_gnt = NULL; 702 phys_addr_t addr = 0; 703 int i, seg_idx, new_map_idx; 704 int segs_to_map = 0; 705 int ret = 0; 706 int last_map = 0, map_until = 0; 707 int use_persistent_gnts; 708 709 use_persistent_gnts = (blkif->vbd.feature_gnt_persistent); 710 711 /* 712 * Fill out preq.nr_sects with proper amount of sectors, and setup 713 * assign map[..] with the PFN of the page in our domain with the 714 * corresponding grant reference for each page. 715 */ 716 again: 717 for (i = map_until; i < num; i++) { 718 uint32_t flags; 719 720 if (use_persistent_gnts) 721 persistent_gnt = get_persistent_gnt( 722 blkif, 723 pages[i]->gref); 724 725 if (persistent_gnt) { 726 /* 727 * We are using persistent grants and 728 * the grant is already mapped 729 */ 730 pages[i]->page = persistent_gnt->page; 731 pages[i]->persistent_gnt = persistent_gnt; 732 } else { 733 if (get_free_page(blkif, &pages[i]->page)) 734 goto out_of_memory; 735 addr = vaddr(pages[i]->page); 736 pages_to_gnt[segs_to_map] = pages[i]->page; 737 pages[i]->persistent_gnt = NULL; 738 flags = GNTMAP_host_map; 739 if (!use_persistent_gnts && ro) 740 flags |= GNTMAP_readonly; 741 gnttab_set_map_op(&map[segs_to_map++], addr, 742 flags, pages[i]->gref, 743 blkif->domid); 744 } 745 map_until = i + 1; 746 if (segs_to_map == BLKIF_MAX_SEGMENTS_PER_REQUEST) 747 break; 748 } 749 750 if (segs_to_map) { 751 ret = gnttab_map_refs(map, NULL, pages_to_gnt, segs_to_map); 752 BUG_ON(ret); 753 } 754 755 /* 756 * Now swizzle the MFN in our domain with the MFN from the other domain 757 * so that when we access vaddr(pending_req,i) it has the contents of 758 * the page from the other domain. 759 */ 760 for (seg_idx = last_map, new_map_idx = 0; seg_idx < map_until; seg_idx++) { 761 if (!pages[seg_idx]->persistent_gnt) { 762 /* This is a newly mapped grant */ 763 BUG_ON(new_map_idx >= segs_to_map); 764 if (unlikely(map[new_map_idx].status != 0)) { 765 pr_debug(DRV_PFX "invalid buffer -- could not remap it\n"); 766 put_free_pages(blkif, &pages[seg_idx]->page, 1); 767 pages[seg_idx]->handle = BLKBACK_INVALID_HANDLE; 768 ret |= 1; 769 goto next; 770 } 771 pages[seg_idx]->handle = map[new_map_idx].handle; 772 } else { 773 continue; 774 } 775 if (use_persistent_gnts && 776 blkif->persistent_gnt_c < xen_blkif_max_pgrants) { 777 /* 778 * We are using persistent grants, the grant is 779 * not mapped but we might have room for it. 780 */ 781 persistent_gnt = kmalloc(sizeof(struct persistent_gnt), 782 GFP_KERNEL); 783 if (!persistent_gnt) { 784 /* 785 * If we don't have enough memory to 786 * allocate the persistent_gnt struct 787 * map this grant non-persistenly 788 */ 789 goto next; 790 } 791 persistent_gnt->gnt = map[new_map_idx].ref; 792 persistent_gnt->handle = map[new_map_idx].handle; 793 persistent_gnt->page = pages[seg_idx]->page; 794 if (add_persistent_gnt(blkif, 795 persistent_gnt)) { 796 kfree(persistent_gnt); 797 persistent_gnt = NULL; 798 goto next; 799 } 800 pages[seg_idx]->persistent_gnt = persistent_gnt; 801 pr_debug(DRV_PFX " grant %u added to the tree of persistent grants, using %u/%u\n", 802 persistent_gnt->gnt, blkif->persistent_gnt_c, 803 xen_blkif_max_pgrants); 804 goto next; 805 } 806 if (use_persistent_gnts && !blkif->vbd.overflow_max_grants) { 807 blkif->vbd.overflow_max_grants = 1; 808 pr_debug(DRV_PFX " domain %u, device %#x is using maximum number of persistent grants\n", 809 blkif->domid, blkif->vbd.handle); 810 } 811 /* 812 * We could not map this grant persistently, so use it as 813 * a non-persistent grant. 814 */ 815 next: 816 new_map_idx++; 817 } 818 segs_to_map = 0; 819 last_map = map_until; 820 if (map_until != num) 821 goto again; 822 823 return ret; 824 825 out_of_memory: 826 pr_alert(DRV_PFX "%s: out of memory\n", __func__); 827 put_free_pages(blkif, pages_to_gnt, segs_to_map); 828 return -ENOMEM; 829 } 830 831 static int xen_blkbk_map_seg(struct pending_req *pending_req) 832 { 833 int rc; 834 835 rc = xen_blkbk_map(pending_req->blkif, pending_req->segments, 836 pending_req->nr_pages, 837 (pending_req->operation != BLKIF_OP_READ)); 838 839 return rc; 840 } 841 842 static int xen_blkbk_parse_indirect(struct blkif_request *req, 843 struct pending_req *pending_req, 844 struct seg_buf seg[], 845 struct phys_req *preq) 846 { 847 struct grant_page **pages = pending_req->indirect_pages; 848 struct xen_blkif *blkif = pending_req->blkif; 849 int indirect_grefs, rc, n, nseg, i; 850 struct blkif_request_segment *segments = NULL; 851 852 nseg = pending_req->nr_pages; 853 indirect_grefs = INDIRECT_PAGES(nseg); 854 BUG_ON(indirect_grefs > BLKIF_MAX_INDIRECT_PAGES_PER_REQUEST); 855 856 for (i = 0; i < indirect_grefs; i++) 857 pages[i]->gref = req->u.indirect.indirect_grefs[i]; 858 859 rc = xen_blkbk_map(blkif, pages, indirect_grefs, true); 860 if (rc) 861 goto unmap; 862 863 for (n = 0, i = 0; n < nseg; n++) { 864 if ((n % SEGS_PER_INDIRECT_FRAME) == 0) { 865 /* Map indirect segments */ 866 if (segments) 867 kunmap_atomic(segments); 868 segments = kmap_atomic(pages[n/SEGS_PER_INDIRECT_FRAME]->page); 869 } 870 i = n % SEGS_PER_INDIRECT_FRAME; 871 pending_req->segments[n]->gref = segments[i].gref; 872 seg[n].nsec = segments[i].last_sect - 873 segments[i].first_sect + 1; 874 seg[n].offset = (segments[i].first_sect << 9); 875 if ((segments[i].last_sect >= (PAGE_SIZE >> 9)) || 876 (segments[i].last_sect < segments[i].first_sect)) { 877 rc = -EINVAL; 878 goto unmap; 879 } 880 preq->nr_sects += seg[n].nsec; 881 } 882 883 unmap: 884 if (segments) 885 kunmap_atomic(segments); 886 xen_blkbk_unmap(blkif, pages, indirect_grefs); 887 return rc; 888 } 889 890 static int dispatch_discard_io(struct xen_blkif *blkif, 891 struct blkif_request *req) 892 { 893 int err = 0; 894 int status = BLKIF_RSP_OKAY; 895 struct block_device *bdev = blkif->vbd.bdev; 896 unsigned long secure; 897 struct phys_req preq; 898 899 xen_blkif_get(blkif); 900 901 preq.sector_number = req->u.discard.sector_number; 902 preq.nr_sects = req->u.discard.nr_sectors; 903 904 err = xen_vbd_translate(&preq, blkif, WRITE); 905 if (err) { 906 pr_warn(DRV_PFX "access denied: DISCARD [%llu->%llu] on dev=%04x\n", 907 preq.sector_number, 908 preq.sector_number + preq.nr_sects, blkif->vbd.pdevice); 909 goto fail_response; 910 } 911 blkif->st_ds_req++; 912 913 secure = (blkif->vbd.discard_secure && 914 (req->u.discard.flag & BLKIF_DISCARD_SECURE)) ? 915 BLKDEV_DISCARD_SECURE : 0; 916 917 err = blkdev_issue_discard(bdev, req->u.discard.sector_number, 918 req->u.discard.nr_sectors, 919 GFP_KERNEL, secure); 920 fail_response: 921 if (err == -EOPNOTSUPP) { 922 pr_debug(DRV_PFX "discard op failed, not supported\n"); 923 status = BLKIF_RSP_EOPNOTSUPP; 924 } else if (err) 925 status = BLKIF_RSP_ERROR; 926 927 make_response(blkif, req->u.discard.id, req->operation, status); 928 xen_blkif_put(blkif); 929 return err; 930 } 931 932 static int dispatch_other_io(struct xen_blkif *blkif, 933 struct blkif_request *req, 934 struct pending_req *pending_req) 935 { 936 free_req(blkif, pending_req); 937 make_response(blkif, req->u.other.id, req->operation, 938 BLKIF_RSP_EOPNOTSUPP); 939 return -EIO; 940 } 941 942 static void xen_blk_drain_io(struct xen_blkif *blkif) 943 { 944 atomic_set(&blkif->drain, 1); 945 do { 946 if (atomic_read(&blkif->inflight) == 0) 947 break; 948 wait_for_completion_interruptible_timeout( 949 &blkif->drain_complete, HZ); 950 951 if (!atomic_read(&blkif->drain)) 952 break; 953 } while (!kthread_should_stop()); 954 atomic_set(&blkif->drain, 0); 955 } 956 957 /* 958 * Completion callback on the bio's. Called as bh->b_end_io() 959 */ 960 961 static void __end_block_io_op(struct pending_req *pending_req, int error) 962 { 963 /* An error fails the entire request. */ 964 if ((pending_req->operation == BLKIF_OP_FLUSH_DISKCACHE) && 965 (error == -EOPNOTSUPP)) { 966 pr_debug(DRV_PFX "flush diskcache op failed, not supported\n"); 967 xen_blkbk_flush_diskcache(XBT_NIL, pending_req->blkif->be, 0); 968 pending_req->status = BLKIF_RSP_EOPNOTSUPP; 969 } else if ((pending_req->operation == BLKIF_OP_WRITE_BARRIER) && 970 (error == -EOPNOTSUPP)) { 971 pr_debug(DRV_PFX "write barrier op failed, not supported\n"); 972 xen_blkbk_barrier(XBT_NIL, pending_req->blkif->be, 0); 973 pending_req->status = BLKIF_RSP_EOPNOTSUPP; 974 } else if (error) { 975 pr_debug(DRV_PFX "Buffer not up-to-date at end of operation," 976 " error=%d\n", error); 977 pending_req->status = BLKIF_RSP_ERROR; 978 } 979 980 /* 981 * If all of the bio's have completed it is time to unmap 982 * the grant references associated with 'request' and provide 983 * the proper response on the ring. 984 */ 985 if (atomic_dec_and_test(&pending_req->pendcnt)) { 986 struct xen_blkif *blkif = pending_req->blkif; 987 988 xen_blkbk_unmap(blkif, 989 pending_req->segments, 990 pending_req->nr_pages); 991 make_response(blkif, pending_req->id, 992 pending_req->operation, pending_req->status); 993 free_req(blkif, pending_req); 994 /* 995 * Make sure the request is freed before releasing blkif, 996 * or there could be a race between free_req and the 997 * cleanup done in xen_blkif_free during shutdown. 998 * 999 * NB: The fact that we might try to wake up pending_free_wq 1000 * before drain_complete (in case there's a drain going on) 1001 * it's not a problem with our current implementation 1002 * because we can assure there's no thread waiting on 1003 * pending_free_wq if there's a drain going on, but it has 1004 * to be taken into account if the current model is changed. 1005 */ 1006 if (atomic_dec_and_test(&blkif->inflight) && atomic_read(&blkif->drain)) { 1007 complete(&blkif->drain_complete); 1008 } 1009 xen_blkif_put(blkif); 1010 } 1011 } 1012 1013 /* 1014 * bio callback. 1015 */ 1016 static void end_block_io_op(struct bio *bio, int error) 1017 { 1018 __end_block_io_op(bio->bi_private, error); 1019 bio_put(bio); 1020 } 1021 1022 1023 1024 /* 1025 * Function to copy the from the ring buffer the 'struct blkif_request' 1026 * (which has the sectors we want, number of them, grant references, etc), 1027 * and transmute it to the block API to hand it over to the proper block disk. 1028 */ 1029 static int 1030 __do_block_io_op(struct xen_blkif *blkif) 1031 { 1032 union blkif_back_rings *blk_rings = &blkif->blk_rings; 1033 struct blkif_request req; 1034 struct pending_req *pending_req; 1035 RING_IDX rc, rp; 1036 int more_to_do = 0; 1037 1038 rc = blk_rings->common.req_cons; 1039 rp = blk_rings->common.sring->req_prod; 1040 rmb(); /* Ensure we see queued requests up to 'rp'. */ 1041 1042 if (RING_REQUEST_PROD_OVERFLOW(&blk_rings->common, rp)) { 1043 rc = blk_rings->common.rsp_prod_pvt; 1044 pr_warn(DRV_PFX "Frontend provided bogus ring requests (%d - %d = %d). Halting ring processing on dev=%04x\n", 1045 rp, rc, rp - rc, blkif->vbd.pdevice); 1046 return -EACCES; 1047 } 1048 while (rc != rp) { 1049 1050 if (RING_REQUEST_CONS_OVERFLOW(&blk_rings->common, rc)) 1051 break; 1052 1053 if (kthread_should_stop()) { 1054 more_to_do = 1; 1055 break; 1056 } 1057 1058 pending_req = alloc_req(blkif); 1059 if (NULL == pending_req) { 1060 blkif->st_oo_req++; 1061 more_to_do = 1; 1062 break; 1063 } 1064 1065 switch (blkif->blk_protocol) { 1066 case BLKIF_PROTOCOL_NATIVE: 1067 memcpy(&req, RING_GET_REQUEST(&blk_rings->native, rc), sizeof(req)); 1068 break; 1069 case BLKIF_PROTOCOL_X86_32: 1070 blkif_get_x86_32_req(&req, RING_GET_REQUEST(&blk_rings->x86_32, rc)); 1071 break; 1072 case BLKIF_PROTOCOL_X86_64: 1073 blkif_get_x86_64_req(&req, RING_GET_REQUEST(&blk_rings->x86_64, rc)); 1074 break; 1075 default: 1076 BUG(); 1077 } 1078 blk_rings->common.req_cons = ++rc; /* before make_response() */ 1079 1080 /* Apply all sanity checks to /private copy/ of request. */ 1081 barrier(); 1082 1083 switch (req.operation) { 1084 case BLKIF_OP_READ: 1085 case BLKIF_OP_WRITE: 1086 case BLKIF_OP_WRITE_BARRIER: 1087 case BLKIF_OP_FLUSH_DISKCACHE: 1088 case BLKIF_OP_INDIRECT: 1089 if (dispatch_rw_block_io(blkif, &req, pending_req)) 1090 goto done; 1091 break; 1092 case BLKIF_OP_DISCARD: 1093 free_req(blkif, pending_req); 1094 if (dispatch_discard_io(blkif, &req)) 1095 goto done; 1096 break; 1097 default: 1098 if (dispatch_other_io(blkif, &req, pending_req)) 1099 goto done; 1100 break; 1101 } 1102 1103 /* Yield point for this unbounded loop. */ 1104 cond_resched(); 1105 } 1106 done: 1107 return more_to_do; 1108 } 1109 1110 static int 1111 do_block_io_op(struct xen_blkif *blkif) 1112 { 1113 union blkif_back_rings *blk_rings = &blkif->blk_rings; 1114 int more_to_do; 1115 1116 do { 1117 more_to_do = __do_block_io_op(blkif); 1118 if (more_to_do) 1119 break; 1120 1121 RING_FINAL_CHECK_FOR_REQUESTS(&blk_rings->common, more_to_do); 1122 } while (more_to_do); 1123 1124 return more_to_do; 1125 } 1126 /* 1127 * Transmutation of the 'struct blkif_request' to a proper 'struct bio' 1128 * and call the 'submit_bio' to pass it to the underlying storage. 1129 */ 1130 static int dispatch_rw_block_io(struct xen_blkif *blkif, 1131 struct blkif_request *req, 1132 struct pending_req *pending_req) 1133 { 1134 struct phys_req preq; 1135 struct seg_buf *seg = pending_req->seg; 1136 unsigned int nseg; 1137 struct bio *bio = NULL; 1138 struct bio **biolist = pending_req->biolist; 1139 int i, nbio = 0; 1140 int operation; 1141 struct blk_plug plug; 1142 bool drain = false; 1143 struct grant_page **pages = pending_req->segments; 1144 unsigned short req_operation; 1145 1146 req_operation = req->operation == BLKIF_OP_INDIRECT ? 1147 req->u.indirect.indirect_op : req->operation; 1148 if ((req->operation == BLKIF_OP_INDIRECT) && 1149 (req_operation != BLKIF_OP_READ) && 1150 (req_operation != BLKIF_OP_WRITE)) { 1151 pr_debug(DRV_PFX "Invalid indirect operation (%u)\n", 1152 req_operation); 1153 goto fail_response; 1154 } 1155 1156 switch (req_operation) { 1157 case BLKIF_OP_READ: 1158 blkif->st_rd_req++; 1159 operation = READ; 1160 break; 1161 case BLKIF_OP_WRITE: 1162 blkif->st_wr_req++; 1163 operation = WRITE_ODIRECT; 1164 break; 1165 case BLKIF_OP_WRITE_BARRIER: 1166 drain = true; 1167 case BLKIF_OP_FLUSH_DISKCACHE: 1168 blkif->st_f_req++; 1169 operation = WRITE_FLUSH; 1170 break; 1171 default: 1172 operation = 0; /* make gcc happy */ 1173 goto fail_response; 1174 break; 1175 } 1176 1177 /* Check that the number of segments is sane. */ 1178 nseg = req->operation == BLKIF_OP_INDIRECT ? 1179 req->u.indirect.nr_segments : req->u.rw.nr_segments; 1180 1181 if (unlikely(nseg == 0 && operation != WRITE_FLUSH) || 1182 unlikely((req->operation != BLKIF_OP_INDIRECT) && 1183 (nseg > BLKIF_MAX_SEGMENTS_PER_REQUEST)) || 1184 unlikely((req->operation == BLKIF_OP_INDIRECT) && 1185 (nseg > MAX_INDIRECT_SEGMENTS))) { 1186 pr_debug(DRV_PFX "Bad number of segments in request (%d)\n", 1187 nseg); 1188 /* Haven't submitted any bio's yet. */ 1189 goto fail_response; 1190 } 1191 1192 preq.nr_sects = 0; 1193 1194 pending_req->blkif = blkif; 1195 pending_req->id = req->u.rw.id; 1196 pending_req->operation = req_operation; 1197 pending_req->status = BLKIF_RSP_OKAY; 1198 pending_req->nr_pages = nseg; 1199 1200 if (req->operation != BLKIF_OP_INDIRECT) { 1201 preq.dev = req->u.rw.handle; 1202 preq.sector_number = req->u.rw.sector_number; 1203 for (i = 0; i < nseg; i++) { 1204 pages[i]->gref = req->u.rw.seg[i].gref; 1205 seg[i].nsec = req->u.rw.seg[i].last_sect - 1206 req->u.rw.seg[i].first_sect + 1; 1207 seg[i].offset = (req->u.rw.seg[i].first_sect << 9); 1208 if ((req->u.rw.seg[i].last_sect >= (PAGE_SIZE >> 9)) || 1209 (req->u.rw.seg[i].last_sect < 1210 req->u.rw.seg[i].first_sect)) 1211 goto fail_response; 1212 preq.nr_sects += seg[i].nsec; 1213 } 1214 } else { 1215 preq.dev = req->u.indirect.handle; 1216 preq.sector_number = req->u.indirect.sector_number; 1217 if (xen_blkbk_parse_indirect(req, pending_req, seg, &preq)) 1218 goto fail_response; 1219 } 1220 1221 if (xen_vbd_translate(&preq, blkif, operation) != 0) { 1222 pr_debug(DRV_PFX "access denied: %s of [%llu,%llu] on dev=%04x\n", 1223 operation == READ ? "read" : "write", 1224 preq.sector_number, 1225 preq.sector_number + preq.nr_sects, 1226 blkif->vbd.pdevice); 1227 goto fail_response; 1228 } 1229 1230 /* 1231 * This check _MUST_ be done after xen_vbd_translate as the preq.bdev 1232 * is set there. 1233 */ 1234 for (i = 0; i < nseg; i++) { 1235 if (((int)preq.sector_number|(int)seg[i].nsec) & 1236 ((bdev_logical_block_size(preq.bdev) >> 9) - 1)) { 1237 pr_debug(DRV_PFX "Misaligned I/O request from domain %d", 1238 blkif->domid); 1239 goto fail_response; 1240 } 1241 } 1242 1243 /* Wait on all outstanding I/O's and once that has been completed 1244 * issue the WRITE_FLUSH. 1245 */ 1246 if (drain) 1247 xen_blk_drain_io(pending_req->blkif); 1248 1249 /* 1250 * If we have failed at this point, we need to undo the M2P override, 1251 * set gnttab_set_unmap_op on all of the grant references and perform 1252 * the hypercall to unmap the grants - that is all done in 1253 * xen_blkbk_unmap. 1254 */ 1255 if (xen_blkbk_map_seg(pending_req)) 1256 goto fail_flush; 1257 1258 /* 1259 * This corresponding xen_blkif_put is done in __end_block_io_op, or 1260 * below (in "!bio") if we are handling a BLKIF_OP_DISCARD. 1261 */ 1262 xen_blkif_get(blkif); 1263 atomic_inc(&blkif->inflight); 1264 1265 for (i = 0; i < nseg; i++) { 1266 while ((bio == NULL) || 1267 (bio_add_page(bio, 1268 pages[i]->page, 1269 seg[i].nsec << 9, 1270 seg[i].offset) == 0)) { 1271 1272 int nr_iovecs = min_t(int, (nseg-i), BIO_MAX_PAGES); 1273 bio = bio_alloc(GFP_KERNEL, nr_iovecs); 1274 if (unlikely(bio == NULL)) 1275 goto fail_put_bio; 1276 1277 biolist[nbio++] = bio; 1278 bio->bi_bdev = preq.bdev; 1279 bio->bi_private = pending_req; 1280 bio->bi_end_io = end_block_io_op; 1281 bio->bi_iter.bi_sector = preq.sector_number; 1282 } 1283 1284 preq.sector_number += seg[i].nsec; 1285 } 1286 1287 /* This will be hit if the operation was a flush or discard. */ 1288 if (!bio) { 1289 BUG_ON(operation != WRITE_FLUSH); 1290 1291 bio = bio_alloc(GFP_KERNEL, 0); 1292 if (unlikely(bio == NULL)) 1293 goto fail_put_bio; 1294 1295 biolist[nbio++] = bio; 1296 bio->bi_bdev = preq.bdev; 1297 bio->bi_private = pending_req; 1298 bio->bi_end_io = end_block_io_op; 1299 } 1300 1301 atomic_set(&pending_req->pendcnt, nbio); 1302 blk_start_plug(&plug); 1303 1304 for (i = 0; i < nbio; i++) 1305 submit_bio(operation, biolist[i]); 1306 1307 /* Let the I/Os go.. */ 1308 blk_finish_plug(&plug); 1309 1310 if (operation == READ) 1311 blkif->st_rd_sect += preq.nr_sects; 1312 else if (operation & WRITE) 1313 blkif->st_wr_sect += preq.nr_sects; 1314 1315 return 0; 1316 1317 fail_flush: 1318 xen_blkbk_unmap(blkif, pending_req->segments, 1319 pending_req->nr_pages); 1320 fail_response: 1321 /* Haven't submitted any bio's yet. */ 1322 make_response(blkif, req->u.rw.id, req_operation, BLKIF_RSP_ERROR); 1323 free_req(blkif, pending_req); 1324 msleep(1); /* back off a bit */ 1325 return -EIO; 1326 1327 fail_put_bio: 1328 for (i = 0; i < nbio; i++) 1329 bio_put(biolist[i]); 1330 atomic_set(&pending_req->pendcnt, 1); 1331 __end_block_io_op(pending_req, -EINVAL); 1332 msleep(1); /* back off a bit */ 1333 return -EIO; 1334 } 1335 1336 1337 1338 /* 1339 * Put a response on the ring on how the operation fared. 1340 */ 1341 static void make_response(struct xen_blkif *blkif, u64 id, 1342 unsigned short op, int st) 1343 { 1344 struct blkif_response resp; 1345 unsigned long flags; 1346 union blkif_back_rings *blk_rings = &blkif->blk_rings; 1347 int notify; 1348 1349 resp.id = id; 1350 resp.operation = op; 1351 resp.status = st; 1352 1353 spin_lock_irqsave(&blkif->blk_ring_lock, flags); 1354 /* Place on the response ring for the relevant domain. */ 1355 switch (blkif->blk_protocol) { 1356 case BLKIF_PROTOCOL_NATIVE: 1357 memcpy(RING_GET_RESPONSE(&blk_rings->native, blk_rings->native.rsp_prod_pvt), 1358 &resp, sizeof(resp)); 1359 break; 1360 case BLKIF_PROTOCOL_X86_32: 1361 memcpy(RING_GET_RESPONSE(&blk_rings->x86_32, blk_rings->x86_32.rsp_prod_pvt), 1362 &resp, sizeof(resp)); 1363 break; 1364 case BLKIF_PROTOCOL_X86_64: 1365 memcpy(RING_GET_RESPONSE(&blk_rings->x86_64, blk_rings->x86_64.rsp_prod_pvt), 1366 &resp, sizeof(resp)); 1367 break; 1368 default: 1369 BUG(); 1370 } 1371 blk_rings->common.rsp_prod_pvt++; 1372 RING_PUSH_RESPONSES_AND_CHECK_NOTIFY(&blk_rings->common, notify); 1373 spin_unlock_irqrestore(&blkif->blk_ring_lock, flags); 1374 if (notify) 1375 notify_remote_via_irq(blkif->irq); 1376 } 1377 1378 static int __init xen_blkif_init(void) 1379 { 1380 int rc = 0; 1381 1382 if (!xen_domain()) 1383 return -ENODEV; 1384 1385 rc = xen_blkif_interface_init(); 1386 if (rc) 1387 goto failed_init; 1388 1389 rc = xen_blkif_xenbus_init(); 1390 if (rc) 1391 goto failed_init; 1392 1393 failed_init: 1394 return rc; 1395 } 1396 1397 module_init(xen_blkif_init); 1398 1399 MODULE_LICENSE("Dual BSD/GPL"); 1400 MODULE_ALIAS("xen-backend:vbd"); 1401