1 /* 2 * XenBSD block device driver 3 * 4 * Copyright (c) 2010-2013 Spectra Logic Corporation 5 * Copyright (c) 2009 Scott Long, Yahoo! 6 * Copyright (c) 2009 Frank Suchomel, Citrix 7 * Copyright (c) 2009 Doug F. Rabson, Citrix 8 * Copyright (c) 2005 Kip Macy 9 * Copyright (c) 2003-2004, Keir Fraser & Steve Hand 10 * Modifications by Mark A. Williamson are (c) Intel Research Cambridge 11 * 12 * 13 * Permission is hereby granted, free of charge, to any person obtaining a copy 14 * of this software and associated documentation files (the "Software"), to 15 * deal in the Software without restriction, including without limitation the 16 * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or 17 * sell copies of the Software, and to permit persons to whom the Software is 18 * furnished to do so, subject to the following conditions: 19 * 20 * The above copyright notice and this permission notice shall be included in 21 * all copies or substantial portions of the Software. 22 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR 23 * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, 24 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE 25 * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER 26 * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING 27 * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER 28 * DEALINGS IN THE SOFTWARE. 29 */ 30 31 #include <sys/cdefs.h> 32 __FBSDID("$FreeBSD$"); 33 34 #include <sys/param.h> 35 #include <sys/systm.h> 36 #include <sys/malloc.h> 37 #include <sys/kernel.h> 38 #include <vm/vm.h> 39 #include <vm/pmap.h> 40 41 #include <sys/bio.h> 42 #include <sys/bus.h> 43 #include <sys/conf.h> 44 #include <sys/module.h> 45 #include <sys/sysctl.h> 46 47 #include <machine/bus.h> 48 #include <sys/rman.h> 49 #include <machine/resource.h> 50 #include <machine/intr_machdep.h> 51 #include <machine/vmparam.h> 52 53 #include <xen/xen-os.h> 54 #include <xen/hypervisor.h> 55 #include <xen/xen_intr.h> 56 #include <xen/gnttab.h> 57 #include <xen/interface/grant_table.h> 58 #include <xen/interface/io/protocols.h> 59 #include <xen/xenbus/xenbusvar.h> 60 61 #include <machine/_inttypes.h> 62 63 #include <geom/geom_disk.h> 64 65 #include <dev/xen/blkfront/block.h> 66 67 #include "xenbus_if.h" 68 69 /*--------------------------- Forward Declarations ---------------------------*/ 70 static void xbd_closing(device_t); 71 static void xbd_startio(struct xbd_softc *sc); 72 73 /*---------------------------------- Macros ----------------------------------*/ 74 #if 0 75 #define DPRINTK(fmt, args...) printf("[XEN] %s:%d: " fmt ".\n", __func__, __LINE__, ##args) 76 #else 77 #define DPRINTK(fmt, args...) 78 #endif 79 80 #define XBD_SECTOR_SHFT 9 81 82 /*---------------------------- Global Static Data ----------------------------*/ 83 static MALLOC_DEFINE(M_XENBLOCKFRONT, "xbd", "Xen Block Front driver data"); 84 85 static int xbd_enable_indirect = 1; 86 SYSCTL_NODE(_hw, OID_AUTO, xbd, CTLFLAG_RD | CTLFLAG_MPSAFE, 0, 87 "xbd driver parameters"); 88 SYSCTL_INT(_hw_xbd, OID_AUTO, xbd_enable_indirect, CTLFLAG_RDTUN, 89 &xbd_enable_indirect, 0, "Enable xbd indirect segments"); 90 91 /*---------------------------- Command Processing ----------------------------*/ 92 static void 93 xbd_freeze(struct xbd_softc *sc, xbd_flag_t xbd_flag) 94 { 95 if (xbd_flag != XBDF_NONE && (sc->xbd_flags & xbd_flag) != 0) 96 return; 97 98 sc->xbd_flags |= xbd_flag; 99 sc->xbd_qfrozen_cnt++; 100 } 101 102 static void 103 xbd_thaw(struct xbd_softc *sc, xbd_flag_t xbd_flag) 104 { 105 if (xbd_flag != XBDF_NONE && (sc->xbd_flags & xbd_flag) == 0) 106 return; 107 108 if (sc->xbd_qfrozen_cnt == 0) 109 panic("%s: Thaw with flag 0x%x while not frozen.", 110 __func__, xbd_flag); 111 112 sc->xbd_flags &= ~xbd_flag; 113 sc->xbd_qfrozen_cnt--; 114 } 115 116 static void 117 xbd_cm_freeze(struct xbd_softc *sc, struct xbd_command *cm, xbdc_flag_t cm_flag) 118 { 119 if ((cm->cm_flags & XBDCF_FROZEN) != 0) 120 return; 121 122 cm->cm_flags |= XBDCF_FROZEN|cm_flag; 123 xbd_freeze(sc, XBDF_NONE); 124 } 125 126 static void 127 xbd_cm_thaw(struct xbd_softc *sc, struct xbd_command *cm) 128 { 129 if ((cm->cm_flags & XBDCF_FROZEN) == 0) 130 return; 131 132 cm->cm_flags &= ~XBDCF_FROZEN; 133 xbd_thaw(sc, XBDF_NONE); 134 } 135 136 static inline void 137 xbd_flush_requests(struct xbd_softc *sc) 138 { 139 int notify; 140 141 RING_PUSH_REQUESTS_AND_CHECK_NOTIFY(&sc->xbd_ring, notify); 142 143 if (notify) 144 xen_intr_signal(sc->xen_intr_handle); 145 } 146 147 static void 148 xbd_free_command(struct xbd_command *cm) 149 { 150 151 KASSERT((cm->cm_flags & XBDCF_Q_MASK) == XBD_Q_NONE, 152 ("Freeing command that is still on queue %d.", 153 cm->cm_flags & XBDCF_Q_MASK)); 154 155 cm->cm_flags = XBDCF_INITIALIZER; 156 cm->cm_bp = NULL; 157 cm->cm_complete = NULL; 158 xbd_enqueue_cm(cm, XBD_Q_FREE); 159 xbd_thaw(cm->cm_sc, XBDF_CM_SHORTAGE); 160 } 161 162 static void 163 xbd_mksegarray(bus_dma_segment_t *segs, int nsegs, 164 grant_ref_t * gref_head, int otherend_id, int readonly, 165 grant_ref_t * sg_ref, struct blkif_request_segment *sg) 166 { 167 struct blkif_request_segment *last_block_sg = sg + nsegs; 168 vm_paddr_t buffer_ma; 169 uint64_t fsect, lsect; 170 int ref; 171 172 while (sg < last_block_sg) { 173 KASSERT(segs->ds_addr % (1 << XBD_SECTOR_SHFT) == 0, 174 ("XEN disk driver I/O must be sector aligned")); 175 KASSERT(segs->ds_len % (1 << XBD_SECTOR_SHFT) == 0, 176 ("XEN disk driver I/Os must be a multiple of " 177 "the sector length")); 178 buffer_ma = segs->ds_addr; 179 fsect = (buffer_ma & PAGE_MASK) >> XBD_SECTOR_SHFT; 180 lsect = fsect + (segs->ds_len >> XBD_SECTOR_SHFT) - 1; 181 182 KASSERT(lsect <= 7, ("XEN disk driver data cannot " 183 "cross a page boundary")); 184 185 /* install a grant reference. */ 186 ref = gnttab_claim_grant_reference(gref_head); 187 188 /* 189 * GNTTAB_LIST_END == 0xffffffff, but it is private 190 * to gnttab.c. 191 */ 192 KASSERT(ref != ~0, ("grant_reference failed")); 193 194 gnttab_grant_foreign_access_ref( 195 ref, 196 otherend_id, 197 buffer_ma >> PAGE_SHIFT, 198 readonly); 199 200 *sg_ref = ref; 201 *sg = (struct blkif_request_segment) { 202 .gref = ref, 203 .first_sect = fsect, 204 .last_sect = lsect 205 }; 206 sg++; 207 sg_ref++; 208 segs++; 209 } 210 } 211 212 static void 213 xbd_queue_cb(void *arg, bus_dma_segment_t *segs, int nsegs, int error) 214 { 215 struct xbd_softc *sc; 216 struct xbd_command *cm; 217 int op; 218 219 cm = arg; 220 sc = cm->cm_sc; 221 222 if (error) { 223 cm->cm_bp->bio_error = EIO; 224 biodone(cm->cm_bp); 225 xbd_free_command(cm); 226 return; 227 } 228 229 KASSERT(nsegs <= sc->xbd_max_request_segments, 230 ("Too many segments in a blkfront I/O")); 231 232 if (nsegs <= BLKIF_MAX_SEGMENTS_PER_REQUEST) { 233 blkif_request_t *ring_req; 234 235 /* Fill out a blkif_request_t structure. */ 236 ring_req = (blkif_request_t *) 237 RING_GET_REQUEST(&sc->xbd_ring, sc->xbd_ring.req_prod_pvt); 238 sc->xbd_ring.req_prod_pvt++; 239 ring_req->id = cm->cm_id; 240 ring_req->operation = cm->cm_operation; 241 ring_req->sector_number = cm->cm_sector_number; 242 ring_req->handle = (blkif_vdev_t)(uintptr_t)sc->xbd_disk; 243 ring_req->nr_segments = nsegs; 244 cm->cm_nseg = nsegs; 245 xbd_mksegarray(segs, nsegs, &cm->cm_gref_head, 246 xenbus_get_otherend_id(sc->xbd_dev), 247 cm->cm_operation == BLKIF_OP_WRITE, 248 cm->cm_sg_refs, ring_req->seg); 249 } else { 250 blkif_request_indirect_t *ring_req; 251 252 /* Fill out a blkif_request_indirect_t structure. */ 253 ring_req = (blkif_request_indirect_t *) 254 RING_GET_REQUEST(&sc->xbd_ring, sc->xbd_ring.req_prod_pvt); 255 sc->xbd_ring.req_prod_pvt++; 256 ring_req->id = cm->cm_id; 257 ring_req->operation = BLKIF_OP_INDIRECT; 258 ring_req->indirect_op = cm->cm_operation; 259 ring_req->sector_number = cm->cm_sector_number; 260 ring_req->handle = (blkif_vdev_t)(uintptr_t)sc->xbd_disk; 261 ring_req->nr_segments = nsegs; 262 cm->cm_nseg = nsegs; 263 xbd_mksegarray(segs, nsegs, &cm->cm_gref_head, 264 xenbus_get_otherend_id(sc->xbd_dev), 265 cm->cm_operation == BLKIF_OP_WRITE, 266 cm->cm_sg_refs, cm->cm_indirectionpages); 267 memcpy(ring_req->indirect_grefs, &cm->cm_indirectionrefs, 268 sizeof(grant_ref_t) * sc->xbd_max_request_indirectpages); 269 } 270 271 if (cm->cm_operation == BLKIF_OP_READ) 272 op = BUS_DMASYNC_PREREAD; 273 else if (cm->cm_operation == BLKIF_OP_WRITE) 274 op = BUS_DMASYNC_PREWRITE; 275 else 276 op = 0; 277 bus_dmamap_sync(sc->xbd_io_dmat, cm->cm_map, op); 278 279 gnttab_free_grant_references(cm->cm_gref_head); 280 281 xbd_enqueue_cm(cm, XBD_Q_BUSY); 282 283 /* 284 * If bus dma had to asynchronously call us back to dispatch 285 * this command, we are no longer executing in the context of 286 * xbd_startio(). Thus we cannot rely on xbd_startio()'s call to 287 * xbd_flush_requests() to publish this command to the backend 288 * along with any other commands that it could batch. 289 */ 290 if ((cm->cm_flags & XBDCF_ASYNC_MAPPING) != 0) 291 xbd_flush_requests(sc); 292 293 return; 294 } 295 296 static int 297 xbd_queue_request(struct xbd_softc *sc, struct xbd_command *cm) 298 { 299 int error; 300 301 if (cm->cm_bp != NULL) 302 error = bus_dmamap_load_bio(sc->xbd_io_dmat, cm->cm_map, 303 cm->cm_bp, xbd_queue_cb, cm, 0); 304 else 305 error = bus_dmamap_load(sc->xbd_io_dmat, cm->cm_map, 306 cm->cm_data, cm->cm_datalen, xbd_queue_cb, cm, 0); 307 if (error == EINPROGRESS) { 308 /* 309 * Maintain queuing order by freezing the queue. The next 310 * command may not require as many resources as the command 311 * we just attempted to map, so we can't rely on bus dma 312 * blocking for it too. 313 */ 314 xbd_cm_freeze(sc, cm, XBDCF_ASYNC_MAPPING); 315 return (0); 316 } 317 318 return (error); 319 } 320 321 static void 322 xbd_restart_queue_callback(void *arg) 323 { 324 struct xbd_softc *sc = arg; 325 326 mtx_lock(&sc->xbd_io_lock); 327 328 xbd_thaw(sc, XBDF_GNT_SHORTAGE); 329 330 xbd_startio(sc); 331 332 mtx_unlock(&sc->xbd_io_lock); 333 } 334 335 static struct xbd_command * 336 xbd_bio_command(struct xbd_softc *sc) 337 { 338 struct xbd_command *cm; 339 struct bio *bp; 340 341 if (__predict_false(sc->xbd_state != XBD_STATE_CONNECTED)) 342 return (NULL); 343 344 bp = xbd_dequeue_bio(sc); 345 if (bp == NULL) 346 return (NULL); 347 348 if ((cm = xbd_dequeue_cm(sc, XBD_Q_FREE)) == NULL) { 349 xbd_freeze(sc, XBDF_CM_SHORTAGE); 350 xbd_requeue_bio(sc, bp); 351 return (NULL); 352 } 353 354 if (gnttab_alloc_grant_references(sc->xbd_max_request_segments, 355 &cm->cm_gref_head) != 0) { 356 gnttab_request_free_callback(&sc->xbd_callback, 357 xbd_restart_queue_callback, sc, 358 sc->xbd_max_request_segments); 359 xbd_freeze(sc, XBDF_GNT_SHORTAGE); 360 xbd_requeue_bio(sc, bp); 361 xbd_enqueue_cm(cm, XBD_Q_FREE); 362 return (NULL); 363 } 364 365 cm->cm_bp = bp; 366 cm->cm_sector_number = (blkif_sector_t)bp->bio_pblkno; 367 368 switch (bp->bio_cmd) { 369 case BIO_READ: 370 cm->cm_operation = BLKIF_OP_READ; 371 break; 372 case BIO_WRITE: 373 cm->cm_operation = BLKIF_OP_WRITE; 374 if ((bp->bio_flags & BIO_ORDERED) != 0) { 375 if ((sc->xbd_flags & XBDF_BARRIER) != 0) { 376 cm->cm_operation = BLKIF_OP_WRITE_BARRIER; 377 } else { 378 /* 379 * Single step this command. 380 */ 381 cm->cm_flags |= XBDCF_Q_FREEZE; 382 if (xbd_queue_length(sc, XBD_Q_BUSY) != 0) { 383 /* 384 * Wait for in-flight requests to 385 * finish. 386 */ 387 xbd_freeze(sc, XBDF_WAIT_IDLE); 388 xbd_requeue_cm(cm, XBD_Q_READY); 389 return (NULL); 390 } 391 } 392 } 393 break; 394 case BIO_FLUSH: 395 if ((sc->xbd_flags & XBDF_FLUSH) != 0) 396 cm->cm_operation = BLKIF_OP_FLUSH_DISKCACHE; 397 else if ((sc->xbd_flags & XBDF_BARRIER) != 0) 398 cm->cm_operation = BLKIF_OP_WRITE_BARRIER; 399 else 400 panic("flush request, but no flush support available"); 401 break; 402 default: 403 biofinish(bp, NULL, EOPNOTSUPP); 404 xbd_enqueue_cm(cm, XBD_Q_FREE); 405 return (NULL); 406 } 407 408 return (cm); 409 } 410 411 /* 412 * Dequeue buffers and place them in the shared communication ring. 413 * Return when no more requests can be accepted or all buffers have 414 * been queued. 415 * 416 * Signal XEN once the ring has been filled out. 417 */ 418 static void 419 xbd_startio(struct xbd_softc *sc) 420 { 421 struct xbd_command *cm; 422 int error, queued = 0; 423 424 mtx_assert(&sc->xbd_io_lock, MA_OWNED); 425 426 if (sc->xbd_state != XBD_STATE_CONNECTED) 427 return; 428 429 while (!RING_FULL(&sc->xbd_ring)) { 430 if (sc->xbd_qfrozen_cnt != 0) 431 break; 432 433 cm = xbd_dequeue_cm(sc, XBD_Q_READY); 434 435 if (cm == NULL) 436 cm = xbd_bio_command(sc); 437 438 if (cm == NULL) 439 break; 440 441 if ((cm->cm_flags & XBDCF_Q_FREEZE) != 0) { 442 /* 443 * Single step command. Future work is 444 * held off until this command completes. 445 */ 446 xbd_cm_freeze(sc, cm, XBDCF_Q_FREEZE); 447 } 448 449 if ((error = xbd_queue_request(sc, cm)) != 0) { 450 printf("xbd_queue_request returned %d\n", error); 451 break; 452 } 453 queued++; 454 } 455 456 if (queued != 0) 457 xbd_flush_requests(sc); 458 } 459 460 static void 461 xbd_bio_complete(struct xbd_softc *sc, struct xbd_command *cm) 462 { 463 struct bio *bp; 464 465 bp = cm->cm_bp; 466 467 if (__predict_false(cm->cm_status != BLKIF_RSP_OKAY)) { 468 disk_err(bp, "disk error" , -1, 0); 469 printf(" status: %x\n", cm->cm_status); 470 bp->bio_flags |= BIO_ERROR; 471 } 472 473 if (bp->bio_flags & BIO_ERROR) 474 bp->bio_error = EIO; 475 else 476 bp->bio_resid = 0; 477 478 xbd_free_command(cm); 479 biodone(bp); 480 } 481 482 static void 483 xbd_int(void *xsc) 484 { 485 struct xbd_softc *sc = xsc; 486 struct xbd_command *cm; 487 blkif_response_t *bret; 488 RING_IDX i, rp; 489 int op; 490 491 mtx_lock(&sc->xbd_io_lock); 492 493 if (__predict_false(sc->xbd_state == XBD_STATE_DISCONNECTED)) { 494 mtx_unlock(&sc->xbd_io_lock); 495 return; 496 } 497 498 again: 499 rp = sc->xbd_ring.sring->rsp_prod; 500 rmb(); /* Ensure we see queued responses up to 'rp'. */ 501 502 for (i = sc->xbd_ring.rsp_cons; i != rp;) { 503 bret = RING_GET_RESPONSE(&sc->xbd_ring, i); 504 cm = &sc->xbd_shadow[bret->id]; 505 506 xbd_remove_cm(cm, XBD_Q_BUSY); 507 gnttab_end_foreign_access_references(cm->cm_nseg, 508 cm->cm_sg_refs); 509 i++; 510 511 if (cm->cm_operation == BLKIF_OP_READ) 512 op = BUS_DMASYNC_POSTREAD; 513 else if (cm->cm_operation == BLKIF_OP_WRITE || 514 cm->cm_operation == BLKIF_OP_WRITE_BARRIER) 515 op = BUS_DMASYNC_POSTWRITE; 516 else 517 op = 0; 518 bus_dmamap_sync(sc->xbd_io_dmat, cm->cm_map, op); 519 bus_dmamap_unload(sc->xbd_io_dmat, cm->cm_map); 520 521 /* 522 * Release any hold this command has on future command 523 * dispatch. 524 */ 525 xbd_cm_thaw(sc, cm); 526 527 /* 528 * Directly call the i/o complete routine to save an 529 * an indirection in the common case. 530 */ 531 cm->cm_status = bret->status; 532 if (cm->cm_bp) 533 xbd_bio_complete(sc, cm); 534 else if (cm->cm_complete != NULL) 535 cm->cm_complete(cm); 536 else 537 xbd_free_command(cm); 538 } 539 540 sc->xbd_ring.rsp_cons = i; 541 542 if (i != sc->xbd_ring.req_prod_pvt) { 543 int more_to_do; 544 RING_FINAL_CHECK_FOR_RESPONSES(&sc->xbd_ring, more_to_do); 545 if (more_to_do) 546 goto again; 547 } else { 548 sc->xbd_ring.sring->rsp_event = i + 1; 549 } 550 551 if (xbd_queue_length(sc, XBD_Q_BUSY) == 0) 552 xbd_thaw(sc, XBDF_WAIT_IDLE); 553 554 xbd_startio(sc); 555 556 if (__predict_false(sc->xbd_state == XBD_STATE_SUSPENDED)) 557 wakeup(&sc->xbd_cm_q[XBD_Q_BUSY]); 558 559 mtx_unlock(&sc->xbd_io_lock); 560 } 561 562 /*------------------------------- Dump Support -------------------------------*/ 563 /** 564 * Quiesce the disk writes for a dump file before allowing the next buffer. 565 */ 566 static void 567 xbd_quiesce(struct xbd_softc *sc) 568 { 569 int mtd; 570 571 // While there are outstanding requests 572 while (xbd_queue_length(sc, XBD_Q_BUSY) != 0) { 573 RING_FINAL_CHECK_FOR_RESPONSES(&sc->xbd_ring, mtd); 574 if (mtd) { 575 /* Received request completions, update queue. */ 576 xbd_int(sc); 577 } 578 if (xbd_queue_length(sc, XBD_Q_BUSY) != 0) { 579 /* 580 * Still pending requests, wait for the disk i/o 581 * to complete. 582 */ 583 HYPERVISOR_yield(); 584 } 585 } 586 } 587 588 /* Kernel dump function for a paravirtualized disk device */ 589 static void 590 xbd_dump_complete(struct xbd_command *cm) 591 { 592 593 xbd_enqueue_cm(cm, XBD_Q_COMPLETE); 594 } 595 596 static int 597 xbd_dump(void *arg, void *virtual, vm_offset_t physical, off_t offset, 598 size_t length) 599 { 600 struct disk *dp = arg; 601 struct xbd_softc *sc = dp->d_drv1; 602 struct xbd_command *cm; 603 size_t chunk; 604 int sbp; 605 int rc = 0; 606 607 if (length == 0) 608 return (0); 609 610 xbd_quiesce(sc); /* All quiet on the western front. */ 611 612 /* 613 * If this lock is held, then this module is failing, and a 614 * successful kernel dump is highly unlikely anyway. 615 */ 616 mtx_lock(&sc->xbd_io_lock); 617 618 /* Split the 64KB block as needed */ 619 for (sbp=0; length > 0; sbp++) { 620 cm = xbd_dequeue_cm(sc, XBD_Q_FREE); 621 if (cm == NULL) { 622 mtx_unlock(&sc->xbd_io_lock); 623 device_printf(sc->xbd_dev, "dump: no more commands?\n"); 624 return (EBUSY); 625 } 626 627 if (gnttab_alloc_grant_references(sc->xbd_max_request_segments, 628 &cm->cm_gref_head) != 0) { 629 xbd_free_command(cm); 630 mtx_unlock(&sc->xbd_io_lock); 631 device_printf(sc->xbd_dev, "no more grant allocs?\n"); 632 return (EBUSY); 633 } 634 635 chunk = length > sc->xbd_max_request_size ? 636 sc->xbd_max_request_size : length; 637 cm->cm_data = virtual; 638 cm->cm_datalen = chunk; 639 cm->cm_operation = BLKIF_OP_WRITE; 640 cm->cm_sector_number = offset / dp->d_sectorsize; 641 cm->cm_complete = xbd_dump_complete; 642 643 xbd_enqueue_cm(cm, XBD_Q_READY); 644 645 length -= chunk; 646 offset += chunk; 647 virtual = (char *) virtual + chunk; 648 } 649 650 /* Tell DOM0 to do the I/O */ 651 xbd_startio(sc); 652 mtx_unlock(&sc->xbd_io_lock); 653 654 /* Poll for the completion. */ 655 xbd_quiesce(sc); /* All quite on the eastern front */ 656 657 /* If there were any errors, bail out... */ 658 while ((cm = xbd_dequeue_cm(sc, XBD_Q_COMPLETE)) != NULL) { 659 if (cm->cm_status != BLKIF_RSP_OKAY) { 660 device_printf(sc->xbd_dev, 661 "Dump I/O failed at sector %jd\n", 662 cm->cm_sector_number); 663 rc = EIO; 664 } 665 xbd_free_command(cm); 666 } 667 668 return (rc); 669 } 670 671 /*----------------------------- Disk Entrypoints -----------------------------*/ 672 static int 673 xbd_open(struct disk *dp) 674 { 675 struct xbd_softc *sc = dp->d_drv1; 676 677 if (sc == NULL) { 678 printf("xbd%d: not found", dp->d_unit); 679 return (ENXIO); 680 } 681 682 sc->xbd_flags |= XBDF_OPEN; 683 sc->xbd_users++; 684 return (0); 685 } 686 687 static int 688 xbd_close(struct disk *dp) 689 { 690 struct xbd_softc *sc = dp->d_drv1; 691 692 if (sc == NULL) 693 return (ENXIO); 694 sc->xbd_flags &= ~XBDF_OPEN; 695 if (--(sc->xbd_users) == 0) { 696 /* 697 * Check whether we have been instructed to close. We will 698 * have ignored this request initially, as the device was 699 * still mounted. 700 */ 701 if (xenbus_get_otherend_state(sc->xbd_dev) == 702 XenbusStateClosing) 703 xbd_closing(sc->xbd_dev); 704 } 705 return (0); 706 } 707 708 static int 709 xbd_ioctl(struct disk *dp, u_long cmd, void *addr, int flag, struct thread *td) 710 { 711 struct xbd_softc *sc = dp->d_drv1; 712 713 if (sc == NULL) 714 return (ENXIO); 715 716 return (ENOTTY); 717 } 718 719 /* 720 * Read/write routine for a buffer. Finds the proper unit, place it on 721 * the sortq and kick the controller. 722 */ 723 static void 724 xbd_strategy(struct bio *bp) 725 { 726 struct xbd_softc *sc = bp->bio_disk->d_drv1; 727 728 /* bogus disk? */ 729 if (sc == NULL) { 730 bp->bio_error = EINVAL; 731 bp->bio_flags |= BIO_ERROR; 732 bp->bio_resid = bp->bio_bcount; 733 biodone(bp); 734 return; 735 } 736 737 /* 738 * Place it in the queue of disk activities for this disk 739 */ 740 mtx_lock(&sc->xbd_io_lock); 741 742 xbd_enqueue_bio(sc, bp); 743 xbd_startio(sc); 744 745 mtx_unlock(&sc->xbd_io_lock); 746 return; 747 } 748 749 /*------------------------------ Ring Management -----------------------------*/ 750 static int 751 xbd_alloc_ring(struct xbd_softc *sc) 752 { 753 blkif_sring_t *sring; 754 uintptr_t sring_page_addr; 755 int error; 756 int i; 757 758 sring = malloc(sc->xbd_ring_pages * PAGE_SIZE, M_XENBLOCKFRONT, 759 M_NOWAIT|M_ZERO); 760 if (sring == NULL) { 761 xenbus_dev_fatal(sc->xbd_dev, ENOMEM, "allocating shared ring"); 762 return (ENOMEM); 763 } 764 SHARED_RING_INIT(sring); 765 FRONT_RING_INIT(&sc->xbd_ring, sring, sc->xbd_ring_pages * PAGE_SIZE); 766 767 for (i = 0, sring_page_addr = (uintptr_t)sring; 768 i < sc->xbd_ring_pages; 769 i++, sring_page_addr += PAGE_SIZE) { 770 error = xenbus_grant_ring(sc->xbd_dev, 771 (vtophys(sring_page_addr) >> PAGE_SHIFT), 772 &sc->xbd_ring_ref[i]); 773 if (error) { 774 xenbus_dev_fatal(sc->xbd_dev, error, 775 "granting ring_ref(%d)", i); 776 return (error); 777 } 778 } 779 if (sc->xbd_ring_pages == 1) { 780 error = xs_printf(XST_NIL, xenbus_get_node(sc->xbd_dev), 781 "ring-ref", "%u", sc->xbd_ring_ref[0]); 782 if (error) { 783 xenbus_dev_fatal(sc->xbd_dev, error, 784 "writing %s/ring-ref", 785 xenbus_get_node(sc->xbd_dev)); 786 return (error); 787 } 788 } else { 789 for (i = 0; i < sc->xbd_ring_pages; i++) { 790 char ring_ref_name[]= "ring_refXX"; 791 792 snprintf(ring_ref_name, sizeof(ring_ref_name), 793 "ring-ref%u", i); 794 error = xs_printf(XST_NIL, xenbus_get_node(sc->xbd_dev), 795 ring_ref_name, "%u", sc->xbd_ring_ref[i]); 796 if (error) { 797 xenbus_dev_fatal(sc->xbd_dev, error, 798 "writing %s/%s", 799 xenbus_get_node(sc->xbd_dev), 800 ring_ref_name); 801 return (error); 802 } 803 } 804 } 805 806 error = xen_intr_alloc_and_bind_local_port(sc->xbd_dev, 807 xenbus_get_otherend_id(sc->xbd_dev), NULL, xbd_int, sc, 808 INTR_TYPE_BIO | INTR_MPSAFE, &sc->xen_intr_handle); 809 if (error) { 810 xenbus_dev_fatal(sc->xbd_dev, error, 811 "xen_intr_alloc_and_bind_local_port failed"); 812 return (error); 813 } 814 815 return (0); 816 } 817 818 static void 819 xbd_free_ring(struct xbd_softc *sc) 820 { 821 int i; 822 823 if (sc->xbd_ring.sring == NULL) 824 return; 825 826 for (i = 0; i < sc->xbd_ring_pages; i++) { 827 if (sc->xbd_ring_ref[i] != GRANT_REF_INVALID) { 828 gnttab_end_foreign_access_ref(sc->xbd_ring_ref[i]); 829 sc->xbd_ring_ref[i] = GRANT_REF_INVALID; 830 } 831 } 832 free(sc->xbd_ring.sring, M_XENBLOCKFRONT); 833 sc->xbd_ring.sring = NULL; 834 } 835 836 /*-------------------------- Initialization/Teardown -------------------------*/ 837 static int 838 xbd_feature_string(struct xbd_softc *sc, char *features, size_t len) 839 { 840 struct sbuf sb; 841 int feature_cnt; 842 843 sbuf_new(&sb, features, len, SBUF_FIXEDLEN); 844 845 feature_cnt = 0; 846 if ((sc->xbd_flags & XBDF_FLUSH) != 0) { 847 sbuf_printf(&sb, "flush"); 848 feature_cnt++; 849 } 850 851 if ((sc->xbd_flags & XBDF_BARRIER) != 0) { 852 if (feature_cnt != 0) 853 sbuf_printf(&sb, ", "); 854 sbuf_printf(&sb, "write_barrier"); 855 feature_cnt++; 856 } 857 858 if ((sc->xbd_flags & XBDF_DISCARD) != 0) { 859 if (feature_cnt != 0) 860 sbuf_printf(&sb, ", "); 861 sbuf_printf(&sb, "discard"); 862 feature_cnt++; 863 } 864 865 if ((sc->xbd_flags & XBDF_PERSISTENT) != 0) { 866 if (feature_cnt != 0) 867 sbuf_printf(&sb, ", "); 868 sbuf_printf(&sb, "persistent_grants"); 869 feature_cnt++; 870 } 871 872 (void) sbuf_finish(&sb); 873 return (sbuf_len(&sb)); 874 } 875 876 static int 877 xbd_sysctl_features(SYSCTL_HANDLER_ARGS) 878 { 879 char features[80]; 880 struct xbd_softc *sc = arg1; 881 int error; 882 int len; 883 884 error = sysctl_wire_old_buffer(req, 0); 885 if (error != 0) 886 return (error); 887 888 len = xbd_feature_string(sc, features, sizeof(features)); 889 890 /* len is -1 on error, which will make the SYSCTL_OUT a no-op. */ 891 return (SYSCTL_OUT(req, features, len + 1/*NUL*/)); 892 } 893 894 static void 895 xbd_setup_sysctl(struct xbd_softc *xbd) 896 { 897 struct sysctl_ctx_list *sysctl_ctx = NULL; 898 struct sysctl_oid *sysctl_tree = NULL; 899 struct sysctl_oid_list *children; 900 901 sysctl_ctx = device_get_sysctl_ctx(xbd->xbd_dev); 902 if (sysctl_ctx == NULL) 903 return; 904 905 sysctl_tree = device_get_sysctl_tree(xbd->xbd_dev); 906 if (sysctl_tree == NULL) 907 return; 908 909 children = SYSCTL_CHILDREN(sysctl_tree); 910 SYSCTL_ADD_UINT(sysctl_ctx, children, OID_AUTO, 911 "max_requests", CTLFLAG_RD, &xbd->xbd_max_requests, -1, 912 "maximum outstanding requests (negotiated)"); 913 914 SYSCTL_ADD_UINT(sysctl_ctx, children, OID_AUTO, 915 "max_request_segments", CTLFLAG_RD, 916 &xbd->xbd_max_request_segments, 0, 917 "maximum number of pages per requests (negotiated)"); 918 919 SYSCTL_ADD_UINT(sysctl_ctx, children, OID_AUTO, 920 "max_request_size", CTLFLAG_RD, &xbd->xbd_max_request_size, 0, 921 "maximum size in bytes of a request (negotiated)"); 922 923 SYSCTL_ADD_UINT(sysctl_ctx, children, OID_AUTO, 924 "ring_pages", CTLFLAG_RD, &xbd->xbd_ring_pages, 0, 925 "communication channel pages (negotiated)"); 926 927 SYSCTL_ADD_PROC(sysctl_ctx, children, OID_AUTO, 928 "features", CTLTYPE_STRING | CTLFLAG_RD | CTLFLAG_NEEDGIANT, xbd, 929 0, xbd_sysctl_features, "A", "protocol features (negotiated)"); 930 } 931 932 /* 933 * Translate Linux major/minor to an appropriate name and unit 934 * number. For HVM guests, this allows us to use the same drive names 935 * with blkfront as the emulated drives, easing transition slightly. 936 */ 937 static void 938 xbd_vdevice_to_unit(uint32_t vdevice, int *unit, const char **name) 939 { 940 static struct vdev_info { 941 int major; 942 int shift; 943 int base; 944 const char *name; 945 } info[] = { 946 {3, 6, 0, "ada"}, /* ide0 */ 947 {22, 6, 2, "ada"}, /* ide1 */ 948 {33, 6, 4, "ada"}, /* ide2 */ 949 {34, 6, 6, "ada"}, /* ide3 */ 950 {56, 6, 8, "ada"}, /* ide4 */ 951 {57, 6, 10, "ada"}, /* ide5 */ 952 {88, 6, 12, "ada"}, /* ide6 */ 953 {89, 6, 14, "ada"}, /* ide7 */ 954 {90, 6, 16, "ada"}, /* ide8 */ 955 {91, 6, 18, "ada"}, /* ide9 */ 956 957 {8, 4, 0, "da"}, /* scsi disk0 */ 958 {65, 4, 16, "da"}, /* scsi disk1 */ 959 {66, 4, 32, "da"}, /* scsi disk2 */ 960 {67, 4, 48, "da"}, /* scsi disk3 */ 961 {68, 4, 64, "da"}, /* scsi disk4 */ 962 {69, 4, 80, "da"}, /* scsi disk5 */ 963 {70, 4, 96, "da"}, /* scsi disk6 */ 964 {71, 4, 112, "da"}, /* scsi disk7 */ 965 {128, 4, 128, "da"}, /* scsi disk8 */ 966 {129, 4, 144, "da"}, /* scsi disk9 */ 967 {130, 4, 160, "da"}, /* scsi disk10 */ 968 {131, 4, 176, "da"}, /* scsi disk11 */ 969 {132, 4, 192, "da"}, /* scsi disk12 */ 970 {133, 4, 208, "da"}, /* scsi disk13 */ 971 {134, 4, 224, "da"}, /* scsi disk14 */ 972 {135, 4, 240, "da"}, /* scsi disk15 */ 973 974 {202, 4, 0, "xbd"}, /* xbd */ 975 976 {0, 0, 0, NULL}, 977 }; 978 int major = vdevice >> 8; 979 int minor = vdevice & 0xff; 980 int i; 981 982 if (vdevice & (1 << 28)) { 983 *unit = (vdevice & ((1 << 28) - 1)) >> 8; 984 *name = "xbd"; 985 return; 986 } 987 988 for (i = 0; info[i].major; i++) { 989 if (info[i].major == major) { 990 *unit = info[i].base + (minor >> info[i].shift); 991 *name = info[i].name; 992 return; 993 } 994 } 995 996 *unit = minor >> 4; 997 *name = "xbd"; 998 } 999 1000 int 1001 xbd_instance_create(struct xbd_softc *sc, blkif_sector_t sectors, 1002 int vdevice, uint16_t vdisk_info, unsigned long sector_size, 1003 unsigned long phys_sector_size) 1004 { 1005 char features[80]; 1006 int unit, error = 0; 1007 const char *name; 1008 1009 xbd_vdevice_to_unit(vdevice, &unit, &name); 1010 1011 sc->xbd_unit = unit; 1012 1013 if (strcmp(name, "xbd") != 0) 1014 device_printf(sc->xbd_dev, "attaching as %s%d\n", name, unit); 1015 1016 if (xbd_feature_string(sc, features, sizeof(features)) > 0) { 1017 device_printf(sc->xbd_dev, "features: %s\n", 1018 features); 1019 } 1020 1021 sc->xbd_disk = disk_alloc(); 1022 sc->xbd_disk->d_unit = sc->xbd_unit; 1023 sc->xbd_disk->d_open = xbd_open; 1024 sc->xbd_disk->d_close = xbd_close; 1025 sc->xbd_disk->d_ioctl = xbd_ioctl; 1026 sc->xbd_disk->d_strategy = xbd_strategy; 1027 sc->xbd_disk->d_dump = xbd_dump; 1028 sc->xbd_disk->d_name = name; 1029 sc->xbd_disk->d_drv1 = sc; 1030 sc->xbd_disk->d_sectorsize = sector_size; 1031 sc->xbd_disk->d_stripesize = phys_sector_size; 1032 sc->xbd_disk->d_stripeoffset = 0; 1033 1034 sc->xbd_disk->d_mediasize = sectors * sector_size; 1035 sc->xbd_disk->d_maxsize = sc->xbd_max_request_size; 1036 sc->xbd_disk->d_flags = DISKFLAG_UNMAPPED_BIO; 1037 if ((sc->xbd_flags & (XBDF_FLUSH|XBDF_BARRIER)) != 0) { 1038 sc->xbd_disk->d_flags |= DISKFLAG_CANFLUSHCACHE; 1039 device_printf(sc->xbd_dev, 1040 "synchronize cache commands enabled.\n"); 1041 } 1042 disk_create(sc->xbd_disk, DISK_VERSION); 1043 1044 return error; 1045 } 1046 1047 static void 1048 xbd_free(struct xbd_softc *sc) 1049 { 1050 int i; 1051 1052 /* Prevent new requests being issued until we fix things up. */ 1053 mtx_lock(&sc->xbd_io_lock); 1054 sc->xbd_state = XBD_STATE_DISCONNECTED; 1055 mtx_unlock(&sc->xbd_io_lock); 1056 1057 /* Free resources associated with old device channel. */ 1058 xbd_free_ring(sc); 1059 if (sc->xbd_shadow) { 1060 for (i = 0; i < sc->xbd_max_requests; i++) { 1061 struct xbd_command *cm; 1062 1063 cm = &sc->xbd_shadow[i]; 1064 if (cm->cm_sg_refs != NULL) { 1065 free(cm->cm_sg_refs, M_XENBLOCKFRONT); 1066 cm->cm_sg_refs = NULL; 1067 } 1068 1069 if (cm->cm_indirectionpages != NULL) { 1070 gnttab_end_foreign_access_references( 1071 sc->xbd_max_request_indirectpages, 1072 &cm->cm_indirectionrefs[0]); 1073 contigfree(cm->cm_indirectionpages, PAGE_SIZE * 1074 sc->xbd_max_request_indirectpages, 1075 M_XENBLOCKFRONT); 1076 cm->cm_indirectionpages = NULL; 1077 } 1078 1079 bus_dmamap_destroy(sc->xbd_io_dmat, cm->cm_map); 1080 } 1081 free(sc->xbd_shadow, M_XENBLOCKFRONT); 1082 sc->xbd_shadow = NULL; 1083 1084 bus_dma_tag_destroy(sc->xbd_io_dmat); 1085 1086 xbd_initq_cm(sc, XBD_Q_FREE); 1087 xbd_initq_cm(sc, XBD_Q_READY); 1088 xbd_initq_cm(sc, XBD_Q_COMPLETE); 1089 } 1090 1091 xen_intr_unbind(&sc->xen_intr_handle); 1092 1093 } 1094 1095 /*--------------------------- State Change Handlers --------------------------*/ 1096 static void 1097 xbd_initialize(struct xbd_softc *sc) 1098 { 1099 const char *otherend_path; 1100 const char *node_path; 1101 uint32_t max_ring_page_order; 1102 int error; 1103 1104 if (xenbus_get_state(sc->xbd_dev) != XenbusStateInitialising) { 1105 /* Initialization has already been performed. */ 1106 return; 1107 } 1108 1109 /* 1110 * Protocol defaults valid even if negotiation for a 1111 * setting fails. 1112 */ 1113 max_ring_page_order = 0; 1114 sc->xbd_ring_pages = 1; 1115 1116 /* 1117 * Protocol negotiation. 1118 * 1119 * \note xs_gather() returns on the first encountered error, so 1120 * we must use independent calls in order to guarantee 1121 * we don't miss information in a sparsly populated back-end 1122 * tree. 1123 * 1124 * \note xs_scanf() does not update variables for unmatched 1125 * fields. 1126 */ 1127 otherend_path = xenbus_get_otherend_path(sc->xbd_dev); 1128 node_path = xenbus_get_node(sc->xbd_dev); 1129 1130 /* Support both backend schemes for relaying ring page limits. */ 1131 (void)xs_scanf(XST_NIL, otherend_path, 1132 "max-ring-page-order", NULL, "%" PRIu32, 1133 &max_ring_page_order); 1134 sc->xbd_ring_pages = 1 << max_ring_page_order; 1135 (void)xs_scanf(XST_NIL, otherend_path, 1136 "max-ring-pages", NULL, "%" PRIu32, 1137 &sc->xbd_ring_pages); 1138 if (sc->xbd_ring_pages < 1) 1139 sc->xbd_ring_pages = 1; 1140 1141 if (sc->xbd_ring_pages > XBD_MAX_RING_PAGES) { 1142 device_printf(sc->xbd_dev, 1143 "Back-end specified ring-pages of %u " 1144 "limited to front-end limit of %u.\n", 1145 sc->xbd_ring_pages, XBD_MAX_RING_PAGES); 1146 sc->xbd_ring_pages = XBD_MAX_RING_PAGES; 1147 } 1148 1149 if (powerof2(sc->xbd_ring_pages) == 0) { 1150 uint32_t new_page_limit; 1151 1152 new_page_limit = 0x01 << (fls(sc->xbd_ring_pages) - 1); 1153 device_printf(sc->xbd_dev, 1154 "Back-end specified ring-pages of %u " 1155 "is not a power of 2. Limited to %u.\n", 1156 sc->xbd_ring_pages, new_page_limit); 1157 sc->xbd_ring_pages = new_page_limit; 1158 } 1159 1160 sc->xbd_max_requests = 1161 BLKIF_MAX_RING_REQUESTS(sc->xbd_ring_pages * PAGE_SIZE); 1162 if (sc->xbd_max_requests > XBD_MAX_REQUESTS) { 1163 device_printf(sc->xbd_dev, 1164 "Back-end specified max_requests of %u " 1165 "limited to front-end limit of %zu.\n", 1166 sc->xbd_max_requests, XBD_MAX_REQUESTS); 1167 sc->xbd_max_requests = XBD_MAX_REQUESTS; 1168 } 1169 1170 if (xbd_alloc_ring(sc) != 0) 1171 return; 1172 1173 /* Support both backend schemes for relaying ring page limits. */ 1174 if (sc->xbd_ring_pages > 1) { 1175 error = xs_printf(XST_NIL, node_path, 1176 "num-ring-pages","%u", 1177 sc->xbd_ring_pages); 1178 if (error) { 1179 xenbus_dev_fatal(sc->xbd_dev, error, 1180 "writing %s/num-ring-pages", 1181 node_path); 1182 return; 1183 } 1184 1185 error = xs_printf(XST_NIL, node_path, 1186 "ring-page-order", "%u", 1187 fls(sc->xbd_ring_pages) - 1); 1188 if (error) { 1189 xenbus_dev_fatal(sc->xbd_dev, error, 1190 "writing %s/ring-page-order", 1191 node_path); 1192 return; 1193 } 1194 } 1195 1196 error = xs_printf(XST_NIL, node_path, "event-channel", 1197 "%u", xen_intr_port(sc->xen_intr_handle)); 1198 if (error) { 1199 xenbus_dev_fatal(sc->xbd_dev, error, 1200 "writing %s/event-channel", 1201 node_path); 1202 return; 1203 } 1204 1205 error = xs_printf(XST_NIL, node_path, "protocol", 1206 "%s", XEN_IO_PROTO_ABI_NATIVE); 1207 if (error) { 1208 xenbus_dev_fatal(sc->xbd_dev, error, 1209 "writing %s/protocol", 1210 node_path); 1211 return; 1212 } 1213 1214 xenbus_set_state(sc->xbd_dev, XenbusStateInitialised); 1215 } 1216 1217 /* 1218 * Invoked when the backend is finally 'ready' (and has published 1219 * the details about the physical device - #sectors, size, etc). 1220 */ 1221 static void 1222 xbd_connect(struct xbd_softc *sc) 1223 { 1224 device_t dev = sc->xbd_dev; 1225 blkif_sector_t sectors; 1226 unsigned long sector_size, phys_sector_size; 1227 unsigned int binfo; 1228 int err, feature_barrier, feature_flush; 1229 int i, j; 1230 1231 DPRINTK("blkfront.c:connect:%s.\n", xenbus_get_otherend_path(dev)); 1232 1233 if (sc->xbd_state == XBD_STATE_SUSPENDED) { 1234 return; 1235 } 1236 1237 if (sc->xbd_state == XBD_STATE_CONNECTED) { 1238 struct disk *disk; 1239 1240 disk = sc->xbd_disk; 1241 if (disk == NULL) { 1242 return; 1243 } 1244 err = xs_gather(XST_NIL, xenbus_get_otherend_path(dev), 1245 "sectors", "%"PRIu64, §ors, NULL); 1246 if (err != 0) { 1247 xenbus_dev_error(dev, err, 1248 "reading sectors at %s", 1249 xenbus_get_otherend_path(dev)); 1250 return; 1251 } 1252 disk->d_mediasize = disk->d_sectorsize * sectors; 1253 err = disk_resize(disk, M_NOWAIT); 1254 if (err) { 1255 xenbus_dev_error(dev, err, 1256 "unable to resize disk %s%u", 1257 disk->d_name, disk->d_unit); 1258 return; 1259 } 1260 device_printf(sc->xbd_dev, 1261 "changed capacity to %jd\n", 1262 (intmax_t)disk->d_mediasize); 1263 return; 1264 } 1265 1266 err = xs_gather(XST_NIL, xenbus_get_otherend_path(dev), 1267 "sectors", "%"PRIu64, §ors, 1268 "info", "%u", &binfo, 1269 "sector-size", "%lu", §or_size, 1270 NULL); 1271 if (err) { 1272 xenbus_dev_fatal(dev, err, 1273 "reading backend fields at %s", 1274 xenbus_get_otherend_path(dev)); 1275 return; 1276 } 1277 if ((sectors == 0) || (sector_size == 0)) { 1278 xenbus_dev_fatal(dev, 0, 1279 "invalid parameters from %s:" 1280 " sectors = %"PRIu64", sector_size = %lu", 1281 xenbus_get_otherend_path(dev), 1282 sectors, sector_size); 1283 return; 1284 } 1285 err = xs_gather(XST_NIL, xenbus_get_otherend_path(dev), 1286 "physical-sector-size", "%lu", &phys_sector_size, 1287 NULL); 1288 if (err || phys_sector_size <= sector_size) 1289 phys_sector_size = 0; 1290 err = xs_gather(XST_NIL, xenbus_get_otherend_path(dev), 1291 "feature-barrier", "%d", &feature_barrier, 1292 NULL); 1293 if (err == 0 && feature_barrier != 0) 1294 sc->xbd_flags |= XBDF_BARRIER; 1295 1296 err = xs_gather(XST_NIL, xenbus_get_otherend_path(dev), 1297 "feature-flush-cache", "%d", &feature_flush, 1298 NULL); 1299 if (err == 0 && feature_flush != 0) 1300 sc->xbd_flags |= XBDF_FLUSH; 1301 1302 err = xs_gather(XST_NIL, xenbus_get_otherend_path(dev), 1303 "feature-max-indirect-segments", "%" PRIu32, 1304 &sc->xbd_max_request_segments, NULL); 1305 if ((err != 0) || (xbd_enable_indirect == 0)) 1306 sc->xbd_max_request_segments = 0; 1307 if (sc->xbd_max_request_segments > XBD_MAX_INDIRECT_SEGMENTS) 1308 sc->xbd_max_request_segments = XBD_MAX_INDIRECT_SEGMENTS; 1309 if (sc->xbd_max_request_segments > XBD_SIZE_TO_SEGS(maxphys)) 1310 sc->xbd_max_request_segments = XBD_SIZE_TO_SEGS(maxphys); 1311 sc->xbd_max_request_indirectpages = 1312 XBD_INDIRECT_SEGS_TO_PAGES(sc->xbd_max_request_segments); 1313 if (sc->xbd_max_request_segments < BLKIF_MAX_SEGMENTS_PER_REQUEST) 1314 sc->xbd_max_request_segments = BLKIF_MAX_SEGMENTS_PER_REQUEST; 1315 sc->xbd_max_request_size = 1316 XBD_SEGS_TO_SIZE(sc->xbd_max_request_segments); 1317 1318 /* Allocate datastructures based on negotiated values. */ 1319 err = bus_dma_tag_create( 1320 bus_get_dma_tag(sc->xbd_dev), /* parent */ 1321 512, PAGE_SIZE, /* algnmnt, boundary */ 1322 BUS_SPACE_MAXADDR, /* lowaddr */ 1323 BUS_SPACE_MAXADDR, /* highaddr */ 1324 NULL, NULL, /* filter, filterarg */ 1325 sc->xbd_max_request_size, 1326 sc->xbd_max_request_segments, 1327 PAGE_SIZE, /* maxsegsize */ 1328 BUS_DMA_ALLOCNOW, /* flags */ 1329 busdma_lock_mutex, /* lockfunc */ 1330 &sc->xbd_io_lock, /* lockarg */ 1331 &sc->xbd_io_dmat); 1332 if (err != 0) { 1333 xenbus_dev_fatal(sc->xbd_dev, err, 1334 "Cannot allocate parent DMA tag\n"); 1335 return; 1336 } 1337 1338 /* Per-transaction data allocation. */ 1339 sc->xbd_shadow = malloc(sizeof(*sc->xbd_shadow) * sc->xbd_max_requests, 1340 M_XENBLOCKFRONT, M_NOWAIT|M_ZERO); 1341 if (sc->xbd_shadow == NULL) { 1342 bus_dma_tag_destroy(sc->xbd_io_dmat); 1343 xenbus_dev_fatal(sc->xbd_dev, ENOMEM, 1344 "Cannot allocate request structures\n"); 1345 return; 1346 } 1347 1348 for (i = 0; i < sc->xbd_max_requests; i++) { 1349 struct xbd_command *cm; 1350 void * indirectpages; 1351 1352 cm = &sc->xbd_shadow[i]; 1353 cm->cm_sg_refs = malloc( 1354 sizeof(grant_ref_t) * sc->xbd_max_request_segments, 1355 M_XENBLOCKFRONT, M_NOWAIT); 1356 if (cm->cm_sg_refs == NULL) 1357 break; 1358 cm->cm_id = i; 1359 cm->cm_flags = XBDCF_INITIALIZER; 1360 cm->cm_sc = sc; 1361 if (bus_dmamap_create(sc->xbd_io_dmat, 0, &cm->cm_map) != 0) 1362 break; 1363 if (sc->xbd_max_request_indirectpages > 0) { 1364 indirectpages = contigmalloc( 1365 PAGE_SIZE * sc->xbd_max_request_indirectpages, 1366 M_XENBLOCKFRONT, M_ZERO | M_NOWAIT, 0, ~0, 1367 PAGE_SIZE, 0); 1368 if (indirectpages == NULL) 1369 sc->xbd_max_request_indirectpages = 0; 1370 } else { 1371 indirectpages = NULL; 1372 } 1373 for (j = 0; j < sc->xbd_max_request_indirectpages; j++) { 1374 if (gnttab_grant_foreign_access( 1375 xenbus_get_otherend_id(sc->xbd_dev), 1376 (vtophys(indirectpages) >> PAGE_SHIFT) + j, 1377 1 /* grant read-only access */, 1378 &cm->cm_indirectionrefs[j])) 1379 break; 1380 } 1381 if (j < sc->xbd_max_request_indirectpages) { 1382 contigfree(indirectpages, 1383 PAGE_SIZE * sc->xbd_max_request_indirectpages, 1384 M_XENBLOCKFRONT); 1385 break; 1386 } 1387 cm->cm_indirectionpages = indirectpages; 1388 xbd_free_command(cm); 1389 } 1390 1391 if (sc->xbd_disk == NULL) { 1392 device_printf(dev, "%juMB <%s> at %s", 1393 (uintmax_t) sectors / (1048576 / sector_size), 1394 device_get_desc(dev), 1395 xenbus_get_node(dev)); 1396 bus_print_child_footer(device_get_parent(dev), dev); 1397 1398 xbd_instance_create(sc, sectors, sc->xbd_vdevice, binfo, 1399 sector_size, phys_sector_size); 1400 } 1401 1402 (void)xenbus_set_state(dev, XenbusStateConnected); 1403 1404 /* Kick pending requests. */ 1405 mtx_lock(&sc->xbd_io_lock); 1406 sc->xbd_state = XBD_STATE_CONNECTED; 1407 xbd_startio(sc); 1408 sc->xbd_flags |= XBDF_READY; 1409 mtx_unlock(&sc->xbd_io_lock); 1410 } 1411 1412 /** 1413 * Handle the change of state of the backend to Closing. We must delete our 1414 * device-layer structures now, to ensure that writes are flushed through to 1415 * the backend. Once this is done, we can switch to Closed in 1416 * acknowledgement. 1417 */ 1418 static void 1419 xbd_closing(device_t dev) 1420 { 1421 struct xbd_softc *sc = device_get_softc(dev); 1422 1423 xenbus_set_state(dev, XenbusStateClosing); 1424 1425 DPRINTK("xbd_closing: %s removed\n", xenbus_get_node(dev)); 1426 1427 if (sc->xbd_disk != NULL) { 1428 disk_destroy(sc->xbd_disk); 1429 sc->xbd_disk = NULL; 1430 } 1431 1432 xenbus_set_state(dev, XenbusStateClosed); 1433 } 1434 1435 /*---------------------------- NewBus Entrypoints ----------------------------*/ 1436 static int 1437 xbd_probe(device_t dev) 1438 { 1439 if (strcmp(xenbus_get_type(dev), "vbd") != 0) 1440 return (ENXIO); 1441 1442 if (xen_pv_disks_disabled()) 1443 return (ENXIO); 1444 1445 if (xen_hvm_domain()) { 1446 int error; 1447 char *type; 1448 1449 /* 1450 * When running in an HVM domain, IDE disk emulation is 1451 * disabled early in boot so that native drivers will 1452 * not see emulated hardware. However, CDROM device 1453 * emulation cannot be disabled. 1454 * 1455 * Through use of FreeBSD's vm_guest and xen_hvm_domain() 1456 * APIs, we could modify the native CDROM driver to fail its 1457 * probe when running under Xen. Unfortunatlely, the PV 1458 * CDROM support in XenServer (up through at least version 1459 * 6.2) isn't functional, so we instead rely on the emulated 1460 * CDROM instance, and fail to attach the PV one here in 1461 * the blkfront driver. 1462 */ 1463 error = xs_read(XST_NIL, xenbus_get_node(dev), 1464 "device-type", NULL, (void **) &type); 1465 if (error) 1466 return (ENXIO); 1467 1468 if (strncmp(type, "cdrom", 5) == 0) { 1469 free(type, M_XENSTORE); 1470 return (ENXIO); 1471 } 1472 free(type, M_XENSTORE); 1473 } 1474 1475 device_set_desc(dev, "Virtual Block Device"); 1476 device_quiet(dev); 1477 return (0); 1478 } 1479 1480 /* 1481 * Setup supplies the backend dir, virtual device. We place an event 1482 * channel and shared frame entries. We watch backend to wait if it's 1483 * ok. 1484 */ 1485 static int 1486 xbd_attach(device_t dev) 1487 { 1488 struct xbd_softc *sc; 1489 const char *name; 1490 uint32_t vdevice; 1491 int error; 1492 int i; 1493 int unit; 1494 1495 /* FIXME: Use dynamic device id if this is not set. */ 1496 error = xs_scanf(XST_NIL, xenbus_get_node(dev), 1497 "virtual-device", NULL, "%" PRIu32, &vdevice); 1498 if (error) 1499 error = xs_scanf(XST_NIL, xenbus_get_node(dev), 1500 "virtual-device-ext", NULL, "%" PRIu32, &vdevice); 1501 if (error) { 1502 xenbus_dev_fatal(dev, error, "reading virtual-device"); 1503 device_printf(dev, "Couldn't determine virtual device.\n"); 1504 return (error); 1505 } 1506 1507 xbd_vdevice_to_unit(vdevice, &unit, &name); 1508 if (!strcmp(name, "xbd")) 1509 device_set_unit(dev, unit); 1510 1511 sc = device_get_softc(dev); 1512 mtx_init(&sc->xbd_io_lock, "blkfront i/o lock", NULL, MTX_DEF); 1513 xbd_initqs(sc); 1514 for (i = 0; i < XBD_MAX_RING_PAGES; i++) 1515 sc->xbd_ring_ref[i] = GRANT_REF_INVALID; 1516 1517 sc->xbd_dev = dev; 1518 sc->xbd_vdevice = vdevice; 1519 sc->xbd_state = XBD_STATE_DISCONNECTED; 1520 1521 xbd_setup_sysctl(sc); 1522 1523 /* Wait for backend device to publish its protocol capabilities. */ 1524 xenbus_set_state(dev, XenbusStateInitialising); 1525 1526 return (0); 1527 } 1528 1529 static int 1530 xbd_detach(device_t dev) 1531 { 1532 struct xbd_softc *sc = device_get_softc(dev); 1533 1534 DPRINTK("%s: %s removed\n", __func__, xenbus_get_node(dev)); 1535 1536 xbd_free(sc); 1537 mtx_destroy(&sc->xbd_io_lock); 1538 1539 return 0; 1540 } 1541 1542 static int 1543 xbd_suspend(device_t dev) 1544 { 1545 struct xbd_softc *sc = device_get_softc(dev); 1546 int retval; 1547 int saved_state; 1548 1549 /* Prevent new requests being issued until we fix things up. */ 1550 mtx_lock(&sc->xbd_io_lock); 1551 saved_state = sc->xbd_state; 1552 sc->xbd_state = XBD_STATE_SUSPENDED; 1553 1554 /* Wait for outstanding I/O to drain. */ 1555 retval = 0; 1556 while (xbd_queue_length(sc, XBD_Q_BUSY) != 0) { 1557 if (msleep(&sc->xbd_cm_q[XBD_Q_BUSY], &sc->xbd_io_lock, 1558 PRIBIO, "blkf_susp", 30 * hz) == EWOULDBLOCK) { 1559 retval = EBUSY; 1560 break; 1561 } 1562 } 1563 mtx_unlock(&sc->xbd_io_lock); 1564 1565 if (retval != 0) 1566 sc->xbd_state = saved_state; 1567 1568 return (retval); 1569 } 1570 1571 static int 1572 xbd_resume(device_t dev) 1573 { 1574 struct xbd_softc *sc = device_get_softc(dev); 1575 1576 if (xen_suspend_cancelled) { 1577 sc->xbd_state = XBD_STATE_CONNECTED; 1578 return (0); 1579 } 1580 1581 DPRINTK("xbd_resume: %s\n", xenbus_get_node(dev)); 1582 1583 xbd_free(sc); 1584 xbd_initialize(sc); 1585 return (0); 1586 } 1587 1588 /** 1589 * Callback received when the backend's state changes. 1590 */ 1591 static void 1592 xbd_backend_changed(device_t dev, XenbusState backend_state) 1593 { 1594 struct xbd_softc *sc = device_get_softc(dev); 1595 1596 DPRINTK("backend_state=%d\n", backend_state); 1597 1598 switch (backend_state) { 1599 case XenbusStateUnknown: 1600 case XenbusStateInitialising: 1601 case XenbusStateReconfigured: 1602 case XenbusStateReconfiguring: 1603 case XenbusStateClosed: 1604 break; 1605 1606 case XenbusStateInitWait: 1607 case XenbusStateInitialised: 1608 xbd_initialize(sc); 1609 break; 1610 1611 case XenbusStateConnected: 1612 xbd_initialize(sc); 1613 xbd_connect(sc); 1614 break; 1615 1616 case XenbusStateClosing: 1617 if (sc->xbd_users > 0) { 1618 device_printf(dev, "detaching with pending users\n"); 1619 KASSERT(sc->xbd_disk != NULL, 1620 ("NULL disk with pending users\n")); 1621 disk_gone(sc->xbd_disk); 1622 } else { 1623 xbd_closing(dev); 1624 } 1625 break; 1626 } 1627 } 1628 1629 /*---------------------------- NewBus Registration ---------------------------*/ 1630 static device_method_t xbd_methods[] = { 1631 /* Device interface */ 1632 DEVMETHOD(device_probe, xbd_probe), 1633 DEVMETHOD(device_attach, xbd_attach), 1634 DEVMETHOD(device_detach, xbd_detach), 1635 DEVMETHOD(device_shutdown, bus_generic_shutdown), 1636 DEVMETHOD(device_suspend, xbd_suspend), 1637 DEVMETHOD(device_resume, xbd_resume), 1638 1639 /* Xenbus interface */ 1640 DEVMETHOD(xenbus_otherend_changed, xbd_backend_changed), 1641 1642 { 0, 0 } 1643 }; 1644 1645 static driver_t xbd_driver = { 1646 "xbd", 1647 xbd_methods, 1648 sizeof(struct xbd_softc), 1649 }; 1650 devclass_t xbd_devclass; 1651 1652 DRIVER_MODULE(xbd, xenbusb_front, xbd_driver, xbd_devclass, 0, 0); 1653