1 /* 2 * Copyright (c) 2009, Microsoft Corporation. 3 * 4 * This program is free software; you can redistribute it and/or modify it 5 * under the terms and conditions of the GNU General Public License, 6 * version 2, as published by the Free Software Foundation. 7 * 8 * This program is distributed in the hope it will be useful, but WITHOUT 9 * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or 10 * FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for 11 * more details. 12 * 13 * You should have received a copy of the GNU General Public License along with 14 * this program; if not, write to the Free Software Foundation, Inc., 59 Temple 15 * Place - Suite 330, Boston, MA 02111-1307 USA. 16 * 17 * Authors: 18 * Haiyang Zhang <haiyangz@microsoft.com> 19 * Hank Janssen <hjanssen@microsoft.com> 20 * K. Y. Srinivasan <kys@microsoft.com> 21 */ 22 23 #include <linux/kernel.h> 24 #include <linux/wait.h> 25 #include <linux/sched.h> 26 #include <linux/completion.h> 27 #include <linux/string.h> 28 #include <linux/mm.h> 29 #include <linux/delay.h> 30 #include <linux/init.h> 31 #include <linux/slab.h> 32 #include <linux/module.h> 33 #include <linux/device.h> 34 #include <linux/hyperv.h> 35 #include <linux/mempool.h> 36 #include <scsi/scsi.h> 37 #include <scsi/scsi_cmnd.h> 38 #include <scsi/scsi_host.h> 39 #include <scsi/scsi_device.h> 40 #include <scsi/scsi_tcq.h> 41 #include <scsi/scsi_eh.h> 42 #include <scsi/scsi_devinfo.h> 43 #include <scsi/scsi_dbg.h> 44 45 /* 46 * All wire protocol details (storage protocol between the guest and the host) 47 * are consolidated here. 48 * 49 * Begin protocol definitions. 50 */ 51 52 /* 53 * Version history: 54 * V1 Beta: 0.1 55 * V1 RC < 2008/1/31: 1.0 56 * V1 RC > 2008/1/31: 2.0 57 * Win7: 4.2 58 * Win8: 5.1 59 */ 60 61 62 #define VMSTOR_WIN7_MAJOR 4 63 #define VMSTOR_WIN7_MINOR 2 64 65 #define VMSTOR_WIN8_MAJOR 5 66 #define VMSTOR_WIN8_MINOR 1 67 68 69 /* Packet structure describing virtual storage requests. */ 70 enum vstor_packet_operation { 71 VSTOR_OPERATION_COMPLETE_IO = 1, 72 VSTOR_OPERATION_REMOVE_DEVICE = 2, 73 VSTOR_OPERATION_EXECUTE_SRB = 3, 74 VSTOR_OPERATION_RESET_LUN = 4, 75 VSTOR_OPERATION_RESET_ADAPTER = 5, 76 VSTOR_OPERATION_RESET_BUS = 6, 77 VSTOR_OPERATION_BEGIN_INITIALIZATION = 7, 78 VSTOR_OPERATION_END_INITIALIZATION = 8, 79 VSTOR_OPERATION_QUERY_PROTOCOL_VERSION = 9, 80 VSTOR_OPERATION_QUERY_PROPERTIES = 10, 81 VSTOR_OPERATION_ENUMERATE_BUS = 11, 82 VSTOR_OPERATION_FCHBA_DATA = 12, 83 VSTOR_OPERATION_CREATE_SUB_CHANNELS = 13, 84 VSTOR_OPERATION_MAXIMUM = 13 85 }; 86 87 /* 88 * WWN packet for Fibre Channel HBA 89 */ 90 91 struct hv_fc_wwn_packet { 92 bool primary_active; 93 u8 reserved1; 94 u8 reserved2; 95 u8 primary_port_wwn[8]; 96 u8 primary_node_wwn[8]; 97 u8 secondary_port_wwn[8]; 98 u8 secondary_node_wwn[8]; 99 }; 100 101 102 103 /* 104 * SRB Flag Bits 105 */ 106 107 #define SRB_FLAGS_QUEUE_ACTION_ENABLE 0x00000002 108 #define SRB_FLAGS_DISABLE_DISCONNECT 0x00000004 109 #define SRB_FLAGS_DISABLE_SYNCH_TRANSFER 0x00000008 110 #define SRB_FLAGS_BYPASS_FROZEN_QUEUE 0x00000010 111 #define SRB_FLAGS_DISABLE_AUTOSENSE 0x00000020 112 #define SRB_FLAGS_DATA_IN 0x00000040 113 #define SRB_FLAGS_DATA_OUT 0x00000080 114 #define SRB_FLAGS_NO_DATA_TRANSFER 0x00000000 115 #define SRB_FLAGS_UNSPECIFIED_DIRECTION (SRB_FLAGS_DATA_IN | SRB_FLAGS_DATA_OUT) 116 #define SRB_FLAGS_NO_QUEUE_FREEZE 0x00000100 117 #define SRB_FLAGS_ADAPTER_CACHE_ENABLE 0x00000200 118 #define SRB_FLAGS_FREE_SENSE_BUFFER 0x00000400 119 120 /* 121 * This flag indicates the request is part of the workflow for processing a D3. 122 */ 123 #define SRB_FLAGS_D3_PROCESSING 0x00000800 124 #define SRB_FLAGS_IS_ACTIVE 0x00010000 125 #define SRB_FLAGS_ALLOCATED_FROM_ZONE 0x00020000 126 #define SRB_FLAGS_SGLIST_FROM_POOL 0x00040000 127 #define SRB_FLAGS_BYPASS_LOCKED_QUEUE 0x00080000 128 #define SRB_FLAGS_NO_KEEP_AWAKE 0x00100000 129 #define SRB_FLAGS_PORT_DRIVER_ALLOCSENSE 0x00200000 130 #define SRB_FLAGS_PORT_DRIVER_SENSEHASPORT 0x00400000 131 #define SRB_FLAGS_DONT_START_NEXT_PACKET 0x00800000 132 #define SRB_FLAGS_PORT_DRIVER_RESERVED 0x0F000000 133 #define SRB_FLAGS_CLASS_DRIVER_RESERVED 0xF0000000 134 135 136 /* 137 * Platform neutral description of a scsi request - 138 * this remains the same across the write regardless of 32/64 bit 139 * note: it's patterned off the SCSI_PASS_THROUGH structure 140 */ 141 #define STORVSC_MAX_CMD_LEN 0x10 142 143 #define POST_WIN7_STORVSC_SENSE_BUFFER_SIZE 0x14 144 #define PRE_WIN8_STORVSC_SENSE_BUFFER_SIZE 0x12 145 146 #define STORVSC_SENSE_BUFFER_SIZE 0x14 147 #define STORVSC_MAX_BUF_LEN_WITH_PADDING 0x14 148 149 /* 150 * Sense buffer size changed in win8; have a run-time 151 * variable to track the size we should use. 152 */ 153 static int sense_buffer_size; 154 155 /* 156 * The size of the vmscsi_request has changed in win8. The 157 * additional size is because of new elements added to the 158 * structure. These elements are valid only when we are talking 159 * to a win8 host. 160 * Track the correction to size we need to apply. 161 */ 162 163 static int vmscsi_size_delta; 164 static int vmstor_current_major; 165 static int vmstor_current_minor; 166 167 struct vmscsi_win8_extension { 168 /* 169 * The following were added in Windows 8 170 */ 171 u16 reserve; 172 u8 queue_tag; 173 u8 queue_action; 174 u32 srb_flags; 175 u32 time_out_value; 176 u32 queue_sort_ey; 177 } __packed; 178 179 struct vmscsi_request { 180 u16 length; 181 u8 srb_status; 182 u8 scsi_status; 183 184 u8 port_number; 185 u8 path_id; 186 u8 target_id; 187 u8 lun; 188 189 u8 cdb_length; 190 u8 sense_info_length; 191 u8 data_in; 192 u8 reserved; 193 194 u32 data_transfer_length; 195 196 union { 197 u8 cdb[STORVSC_MAX_CMD_LEN]; 198 u8 sense_data[STORVSC_SENSE_BUFFER_SIZE]; 199 u8 reserved_array[STORVSC_MAX_BUF_LEN_WITH_PADDING]; 200 }; 201 /* 202 * The following was added in win8. 203 */ 204 struct vmscsi_win8_extension win8_extension; 205 206 } __attribute((packed)); 207 208 209 /* 210 * This structure is sent during the intialization phase to get the different 211 * properties of the channel. 212 */ 213 214 #define STORAGE_CHANNEL_SUPPORTS_MULTI_CHANNEL 0x1 215 216 struct vmstorage_channel_properties { 217 u32 reserved; 218 u16 max_channel_cnt; 219 u16 reserved1; 220 221 u32 flags; 222 u32 max_transfer_bytes; 223 224 u64 reserved2; 225 } __packed; 226 227 /* This structure is sent during the storage protocol negotiations. */ 228 struct vmstorage_protocol_version { 229 /* Major (MSW) and minor (LSW) version numbers. */ 230 u16 major_minor; 231 232 /* 233 * Revision number is auto-incremented whenever this file is changed 234 * (See FILL_VMSTOR_REVISION macro above). Mismatch does not 235 * definitely indicate incompatibility--but it does indicate mismatched 236 * builds. 237 * This is only used on the windows side. Just set it to 0. 238 */ 239 u16 revision; 240 } __packed; 241 242 /* Channel Property Flags */ 243 #define STORAGE_CHANNEL_REMOVABLE_FLAG 0x1 244 #define STORAGE_CHANNEL_EMULATED_IDE_FLAG 0x2 245 246 struct vstor_packet { 247 /* Requested operation type */ 248 enum vstor_packet_operation operation; 249 250 /* Flags - see below for values */ 251 u32 flags; 252 253 /* Status of the request returned from the server side. */ 254 u32 status; 255 256 /* Data payload area */ 257 union { 258 /* 259 * Structure used to forward SCSI commands from the 260 * client to the server. 261 */ 262 struct vmscsi_request vm_srb; 263 264 /* Structure used to query channel properties. */ 265 struct vmstorage_channel_properties storage_channel_properties; 266 267 /* Used during version negotiations. */ 268 struct vmstorage_protocol_version version; 269 270 /* Fibre channel address packet */ 271 struct hv_fc_wwn_packet wwn_packet; 272 273 /* Number of sub-channels to create */ 274 u16 sub_channel_count; 275 276 /* This will be the maximum of the union members */ 277 u8 buffer[0x34]; 278 }; 279 } __packed; 280 281 /* 282 * Packet Flags: 283 * 284 * This flag indicates that the server should send back a completion for this 285 * packet. 286 */ 287 288 #define REQUEST_COMPLETION_FLAG 0x1 289 290 /* Matches Windows-end */ 291 enum storvsc_request_type { 292 WRITE_TYPE = 0, 293 READ_TYPE, 294 UNKNOWN_TYPE, 295 }; 296 297 /* 298 * SRB status codes and masks; a subset of the codes used here. 299 */ 300 301 #define SRB_STATUS_AUTOSENSE_VALID 0x80 302 #define SRB_STATUS_INVALID_LUN 0x20 303 #define SRB_STATUS_SUCCESS 0x01 304 #define SRB_STATUS_ABORTED 0x02 305 #define SRB_STATUS_ERROR 0x04 306 307 /* 308 * This is the end of Protocol specific defines. 309 */ 310 311 312 /* 313 * We setup a mempool to allocate request structures for this driver 314 * on a per-lun basis. The following define specifies the number of 315 * elements in the pool. 316 */ 317 318 #define STORVSC_MIN_BUF_NR 64 319 static int storvsc_ringbuffer_size = (20 * PAGE_SIZE); 320 321 module_param(storvsc_ringbuffer_size, int, S_IRUGO); 322 MODULE_PARM_DESC(storvsc_ringbuffer_size, "Ring buffer size (bytes)"); 323 324 /* 325 * Timeout in seconds for all devices managed by this driver. 326 */ 327 static int storvsc_timeout = 180; 328 329 #define STORVSC_MAX_IO_REQUESTS 200 330 331 static void storvsc_on_channel_callback(void *context); 332 333 /* 334 * In Hyper-V, each port/path/target maps to 1 scsi host adapter. In 335 * reality, the path/target is not used (ie always set to 0) so our 336 * scsi host adapter essentially has 1 bus with 1 target that contains 337 * up to 256 luns. 338 */ 339 #define STORVSC_MAX_LUNS_PER_TARGET 64 340 #define STORVSC_MAX_TARGETS 1 341 #define STORVSC_MAX_CHANNELS 1 342 343 344 345 struct storvsc_cmd_request { 346 struct list_head entry; 347 struct scsi_cmnd *cmd; 348 349 unsigned int bounce_sgl_count; 350 struct scatterlist *bounce_sgl; 351 352 struct hv_device *device; 353 354 /* Synchronize the request/response if needed */ 355 struct completion wait_event; 356 357 unsigned char *sense_buffer; 358 struct hv_multipage_buffer data_buffer; 359 struct vstor_packet vstor_packet; 360 }; 361 362 363 /* A storvsc device is a device object that contains a vmbus channel */ 364 struct storvsc_device { 365 struct hv_device *device; 366 367 bool destroy; 368 bool drain_notify; 369 bool open_sub_channel; 370 atomic_t num_outstanding_req; 371 struct Scsi_Host *host; 372 373 wait_queue_head_t waiting_to_drain; 374 375 /* 376 * Each unique Port/Path/Target represents 1 channel ie scsi 377 * controller. In reality, the pathid, targetid is always 0 378 * and the port is set by us 379 */ 380 unsigned int port_number; 381 unsigned char path_id; 382 unsigned char target_id; 383 384 /* Used for vsc/vsp channel reset process */ 385 struct storvsc_cmd_request init_request; 386 struct storvsc_cmd_request reset_request; 387 }; 388 389 struct stor_mem_pools { 390 struct kmem_cache *request_pool; 391 mempool_t *request_mempool; 392 }; 393 394 struct hv_host_device { 395 struct hv_device *dev; 396 unsigned int port; 397 unsigned char path; 398 unsigned char target; 399 }; 400 401 struct storvsc_scan_work { 402 struct work_struct work; 403 struct Scsi_Host *host; 404 uint lun; 405 }; 406 407 static void storvsc_device_scan(struct work_struct *work) 408 { 409 struct storvsc_scan_work *wrk; 410 uint lun; 411 struct scsi_device *sdev; 412 413 wrk = container_of(work, struct storvsc_scan_work, work); 414 lun = wrk->lun; 415 416 sdev = scsi_device_lookup(wrk->host, 0, 0, lun); 417 if (!sdev) 418 goto done; 419 scsi_rescan_device(&sdev->sdev_gendev); 420 scsi_device_put(sdev); 421 422 done: 423 kfree(wrk); 424 } 425 426 static void storvsc_bus_scan(struct work_struct *work) 427 { 428 struct storvsc_scan_work *wrk; 429 int id, order_id; 430 431 wrk = container_of(work, struct storvsc_scan_work, work); 432 for (id = 0; id < wrk->host->max_id; ++id) { 433 if (wrk->host->reverse_ordering) 434 order_id = wrk->host->max_id - id - 1; 435 else 436 order_id = id; 437 438 scsi_scan_target(&wrk->host->shost_gendev, 0, 439 order_id, SCAN_WILD_CARD, 1); 440 } 441 kfree(wrk); 442 } 443 444 static void storvsc_remove_lun(struct work_struct *work) 445 { 446 struct storvsc_scan_work *wrk; 447 struct scsi_device *sdev; 448 449 wrk = container_of(work, struct storvsc_scan_work, work); 450 if (!scsi_host_get(wrk->host)) 451 goto done; 452 453 sdev = scsi_device_lookup(wrk->host, 0, 0, wrk->lun); 454 455 if (sdev) { 456 scsi_remove_device(sdev); 457 scsi_device_put(sdev); 458 } 459 scsi_host_put(wrk->host); 460 461 done: 462 kfree(wrk); 463 } 464 465 /* 466 * Major/minor macros. Minor version is in LSB, meaning that earlier flat 467 * version numbers will be interpreted as "0.x" (i.e., 1 becomes 0.1). 468 */ 469 470 static inline u16 storvsc_get_version(u8 major, u8 minor) 471 { 472 u16 version; 473 474 version = ((major << 8) | minor); 475 return version; 476 } 477 478 /* 479 * We can get incoming messages from the host that are not in response to 480 * messages that we have sent out. An example of this would be messages 481 * received by the guest to notify dynamic addition/removal of LUNs. To 482 * deal with potential race conditions where the driver may be in the 483 * midst of being unloaded when we might receive an unsolicited message 484 * from the host, we have implemented a mechanism to gurantee sequential 485 * consistency: 486 * 487 * 1) Once the device is marked as being destroyed, we will fail all 488 * outgoing messages. 489 * 2) We permit incoming messages when the device is being destroyed, 490 * only to properly account for messages already sent out. 491 */ 492 493 static inline struct storvsc_device *get_out_stor_device( 494 struct hv_device *device) 495 { 496 struct storvsc_device *stor_device; 497 498 stor_device = hv_get_drvdata(device); 499 500 if (stor_device && stor_device->destroy) 501 stor_device = NULL; 502 503 return stor_device; 504 } 505 506 507 static inline void storvsc_wait_to_drain(struct storvsc_device *dev) 508 { 509 dev->drain_notify = true; 510 wait_event(dev->waiting_to_drain, 511 atomic_read(&dev->num_outstanding_req) == 0); 512 dev->drain_notify = false; 513 } 514 515 static inline struct storvsc_device *get_in_stor_device( 516 struct hv_device *device) 517 { 518 struct storvsc_device *stor_device; 519 520 stor_device = hv_get_drvdata(device); 521 522 if (!stor_device) 523 goto get_in_err; 524 525 /* 526 * If the device is being destroyed; allow incoming 527 * traffic only to cleanup outstanding requests. 528 */ 529 530 if (stor_device->destroy && 531 (atomic_read(&stor_device->num_outstanding_req) == 0)) 532 stor_device = NULL; 533 534 get_in_err: 535 return stor_device; 536 537 } 538 539 static void destroy_bounce_buffer(struct scatterlist *sgl, 540 unsigned int sg_count) 541 { 542 int i; 543 struct page *page_buf; 544 545 for (i = 0; i < sg_count; i++) { 546 page_buf = sg_page((&sgl[i])); 547 if (page_buf != NULL) 548 __free_page(page_buf); 549 } 550 551 kfree(sgl); 552 } 553 554 static int do_bounce_buffer(struct scatterlist *sgl, unsigned int sg_count) 555 { 556 int i; 557 558 /* No need to check */ 559 if (sg_count < 2) 560 return -1; 561 562 /* We have at least 2 sg entries */ 563 for (i = 0; i < sg_count; i++) { 564 if (i == 0) { 565 /* make sure 1st one does not have hole */ 566 if (sgl[i].offset + sgl[i].length != PAGE_SIZE) 567 return i; 568 } else if (i == sg_count - 1) { 569 /* make sure last one does not have hole */ 570 if (sgl[i].offset != 0) 571 return i; 572 } else { 573 /* make sure no hole in the middle */ 574 if (sgl[i].length != PAGE_SIZE || sgl[i].offset != 0) 575 return i; 576 } 577 } 578 return -1; 579 } 580 581 static struct scatterlist *create_bounce_buffer(struct scatterlist *sgl, 582 unsigned int sg_count, 583 unsigned int len, 584 int write) 585 { 586 int i; 587 int num_pages; 588 struct scatterlist *bounce_sgl; 589 struct page *page_buf; 590 unsigned int buf_len = ((write == WRITE_TYPE) ? 0 : PAGE_SIZE); 591 592 num_pages = ALIGN(len, PAGE_SIZE) >> PAGE_SHIFT; 593 594 bounce_sgl = kcalloc(num_pages, sizeof(struct scatterlist), GFP_ATOMIC); 595 if (!bounce_sgl) 596 return NULL; 597 598 sg_init_table(bounce_sgl, num_pages); 599 for (i = 0; i < num_pages; i++) { 600 page_buf = alloc_page(GFP_ATOMIC); 601 if (!page_buf) 602 goto cleanup; 603 sg_set_page(&bounce_sgl[i], page_buf, buf_len, 0); 604 } 605 606 return bounce_sgl; 607 608 cleanup: 609 destroy_bounce_buffer(bounce_sgl, num_pages); 610 return NULL; 611 } 612 613 /* Disgusting wrapper functions */ 614 static inline unsigned long sg_kmap_atomic(struct scatterlist *sgl, int idx) 615 { 616 void *addr = kmap_atomic(sg_page(sgl + idx)); 617 return (unsigned long)addr; 618 } 619 620 static inline void sg_kunmap_atomic(unsigned long addr) 621 { 622 kunmap_atomic((void *)addr); 623 } 624 625 626 /* Assume the original sgl has enough room */ 627 static unsigned int copy_from_bounce_buffer(struct scatterlist *orig_sgl, 628 struct scatterlist *bounce_sgl, 629 unsigned int orig_sgl_count, 630 unsigned int bounce_sgl_count) 631 { 632 int i; 633 int j = 0; 634 unsigned long src, dest; 635 unsigned int srclen, destlen, copylen; 636 unsigned int total_copied = 0; 637 unsigned long bounce_addr = 0; 638 unsigned long dest_addr = 0; 639 unsigned long flags; 640 641 local_irq_save(flags); 642 643 for (i = 0; i < orig_sgl_count; i++) { 644 dest_addr = sg_kmap_atomic(orig_sgl,i) + orig_sgl[i].offset; 645 dest = dest_addr; 646 destlen = orig_sgl[i].length; 647 648 if (bounce_addr == 0) 649 bounce_addr = sg_kmap_atomic(bounce_sgl,j); 650 651 while (destlen) { 652 src = bounce_addr + bounce_sgl[j].offset; 653 srclen = bounce_sgl[j].length - bounce_sgl[j].offset; 654 655 copylen = min(srclen, destlen); 656 memcpy((void *)dest, (void *)src, copylen); 657 658 total_copied += copylen; 659 bounce_sgl[j].offset += copylen; 660 destlen -= copylen; 661 dest += copylen; 662 663 if (bounce_sgl[j].offset == bounce_sgl[j].length) { 664 /* full */ 665 sg_kunmap_atomic(bounce_addr); 666 j++; 667 668 /* 669 * It is possible that the number of elements 670 * in the bounce buffer may not be equal to 671 * the number of elements in the original 672 * scatter list. Handle this correctly. 673 */ 674 675 if (j == bounce_sgl_count) { 676 /* 677 * We are done; cleanup and return. 678 */ 679 sg_kunmap_atomic(dest_addr - orig_sgl[i].offset); 680 local_irq_restore(flags); 681 return total_copied; 682 } 683 684 /* if we need to use another bounce buffer */ 685 if (destlen || i != orig_sgl_count - 1) 686 bounce_addr = sg_kmap_atomic(bounce_sgl,j); 687 } else if (destlen == 0 && i == orig_sgl_count - 1) { 688 /* unmap the last bounce that is < PAGE_SIZE */ 689 sg_kunmap_atomic(bounce_addr); 690 } 691 } 692 693 sg_kunmap_atomic(dest_addr - orig_sgl[i].offset); 694 } 695 696 local_irq_restore(flags); 697 698 return total_copied; 699 } 700 701 /* Assume the bounce_sgl has enough room ie using the create_bounce_buffer() */ 702 static unsigned int copy_to_bounce_buffer(struct scatterlist *orig_sgl, 703 struct scatterlist *bounce_sgl, 704 unsigned int orig_sgl_count) 705 { 706 int i; 707 int j = 0; 708 unsigned long src, dest; 709 unsigned int srclen, destlen, copylen; 710 unsigned int total_copied = 0; 711 unsigned long bounce_addr = 0; 712 unsigned long src_addr = 0; 713 unsigned long flags; 714 715 local_irq_save(flags); 716 717 for (i = 0; i < orig_sgl_count; i++) { 718 src_addr = sg_kmap_atomic(orig_sgl,i) + orig_sgl[i].offset; 719 src = src_addr; 720 srclen = orig_sgl[i].length; 721 722 if (bounce_addr == 0) 723 bounce_addr = sg_kmap_atomic(bounce_sgl,j); 724 725 while (srclen) { 726 /* assume bounce offset always == 0 */ 727 dest = bounce_addr + bounce_sgl[j].length; 728 destlen = PAGE_SIZE - bounce_sgl[j].length; 729 730 copylen = min(srclen, destlen); 731 memcpy((void *)dest, (void *)src, copylen); 732 733 total_copied += copylen; 734 bounce_sgl[j].length += copylen; 735 srclen -= copylen; 736 src += copylen; 737 738 if (bounce_sgl[j].length == PAGE_SIZE) { 739 /* full..move to next entry */ 740 sg_kunmap_atomic(bounce_addr); 741 j++; 742 743 /* if we need to use another bounce buffer */ 744 if (srclen || i != orig_sgl_count - 1) 745 bounce_addr = sg_kmap_atomic(bounce_sgl,j); 746 747 } else if (srclen == 0 && i == orig_sgl_count - 1) { 748 /* unmap the last bounce that is < PAGE_SIZE */ 749 sg_kunmap_atomic(bounce_addr); 750 } 751 } 752 753 sg_kunmap_atomic(src_addr - orig_sgl[i].offset); 754 } 755 756 local_irq_restore(flags); 757 758 return total_copied; 759 } 760 761 static void handle_sc_creation(struct vmbus_channel *new_sc) 762 { 763 struct hv_device *device = new_sc->primary_channel->device_obj; 764 struct storvsc_device *stor_device; 765 struct vmstorage_channel_properties props; 766 767 stor_device = get_out_stor_device(device); 768 if (!stor_device) 769 return; 770 771 if (stor_device->open_sub_channel == false) 772 return; 773 774 memset(&props, 0, sizeof(struct vmstorage_channel_properties)); 775 776 vmbus_open(new_sc, 777 storvsc_ringbuffer_size, 778 storvsc_ringbuffer_size, 779 (void *)&props, 780 sizeof(struct vmstorage_channel_properties), 781 storvsc_on_channel_callback, new_sc); 782 } 783 784 static void handle_multichannel_storage(struct hv_device *device, int max_chns) 785 { 786 struct storvsc_device *stor_device; 787 int num_cpus = num_online_cpus(); 788 int num_sc; 789 struct storvsc_cmd_request *request; 790 struct vstor_packet *vstor_packet; 791 int ret, t; 792 793 num_sc = ((max_chns > num_cpus) ? num_cpus : max_chns); 794 stor_device = get_out_stor_device(device); 795 if (!stor_device) 796 return; 797 798 request = &stor_device->init_request; 799 vstor_packet = &request->vstor_packet; 800 801 stor_device->open_sub_channel = true; 802 /* 803 * Establish a handler for dealing with subchannels. 804 */ 805 vmbus_set_sc_create_callback(device->channel, handle_sc_creation); 806 807 /* 808 * Check to see if sub-channels have already been created. This 809 * can happen when this driver is re-loaded after unloading. 810 */ 811 812 if (vmbus_are_subchannels_present(device->channel)) 813 return; 814 815 stor_device->open_sub_channel = false; 816 /* 817 * Request the host to create sub-channels. 818 */ 819 memset(request, 0, sizeof(struct storvsc_cmd_request)); 820 init_completion(&request->wait_event); 821 vstor_packet->operation = VSTOR_OPERATION_CREATE_SUB_CHANNELS; 822 vstor_packet->flags = REQUEST_COMPLETION_FLAG; 823 vstor_packet->sub_channel_count = num_sc; 824 825 ret = vmbus_sendpacket(device->channel, vstor_packet, 826 (sizeof(struct vstor_packet) - 827 vmscsi_size_delta), 828 (unsigned long)request, 829 VM_PKT_DATA_INBAND, 830 VMBUS_DATA_PACKET_FLAG_COMPLETION_REQUESTED); 831 832 if (ret != 0) 833 return; 834 835 t = wait_for_completion_timeout(&request->wait_event, 10*HZ); 836 if (t == 0) 837 return; 838 839 if (vstor_packet->operation != VSTOR_OPERATION_COMPLETE_IO || 840 vstor_packet->status != 0) 841 return; 842 843 /* 844 * Now that we created the sub-channels, invoke the check; this 845 * may trigger the callback. 846 */ 847 stor_device->open_sub_channel = true; 848 vmbus_are_subchannels_present(device->channel); 849 } 850 851 static int storvsc_channel_init(struct hv_device *device) 852 { 853 struct storvsc_device *stor_device; 854 struct storvsc_cmd_request *request; 855 struct vstor_packet *vstor_packet; 856 int ret, t; 857 int max_chns; 858 bool process_sub_channels = false; 859 860 stor_device = get_out_stor_device(device); 861 if (!stor_device) 862 return -ENODEV; 863 864 request = &stor_device->init_request; 865 vstor_packet = &request->vstor_packet; 866 867 /* 868 * Now, initiate the vsc/vsp initialization protocol on the open 869 * channel 870 */ 871 memset(request, 0, sizeof(struct storvsc_cmd_request)); 872 init_completion(&request->wait_event); 873 vstor_packet->operation = VSTOR_OPERATION_BEGIN_INITIALIZATION; 874 vstor_packet->flags = REQUEST_COMPLETION_FLAG; 875 876 ret = vmbus_sendpacket(device->channel, vstor_packet, 877 (sizeof(struct vstor_packet) - 878 vmscsi_size_delta), 879 (unsigned long)request, 880 VM_PKT_DATA_INBAND, 881 VMBUS_DATA_PACKET_FLAG_COMPLETION_REQUESTED); 882 if (ret != 0) 883 goto cleanup; 884 885 t = wait_for_completion_timeout(&request->wait_event, 5*HZ); 886 if (t == 0) { 887 ret = -ETIMEDOUT; 888 goto cleanup; 889 } 890 891 if (vstor_packet->operation != VSTOR_OPERATION_COMPLETE_IO || 892 vstor_packet->status != 0) 893 goto cleanup; 894 895 896 /* reuse the packet for version range supported */ 897 memset(vstor_packet, 0, sizeof(struct vstor_packet)); 898 vstor_packet->operation = VSTOR_OPERATION_QUERY_PROTOCOL_VERSION; 899 vstor_packet->flags = REQUEST_COMPLETION_FLAG; 900 901 vstor_packet->version.major_minor = 902 storvsc_get_version(vmstor_current_major, vmstor_current_minor); 903 904 /* 905 * The revision number is only used in Windows; set it to 0. 906 */ 907 vstor_packet->version.revision = 0; 908 909 ret = vmbus_sendpacket(device->channel, vstor_packet, 910 (sizeof(struct vstor_packet) - 911 vmscsi_size_delta), 912 (unsigned long)request, 913 VM_PKT_DATA_INBAND, 914 VMBUS_DATA_PACKET_FLAG_COMPLETION_REQUESTED); 915 if (ret != 0) 916 goto cleanup; 917 918 t = wait_for_completion_timeout(&request->wait_event, 5*HZ); 919 if (t == 0) { 920 ret = -ETIMEDOUT; 921 goto cleanup; 922 } 923 924 if (vstor_packet->operation != VSTOR_OPERATION_COMPLETE_IO || 925 vstor_packet->status != 0) 926 goto cleanup; 927 928 929 memset(vstor_packet, 0, sizeof(struct vstor_packet)); 930 vstor_packet->operation = VSTOR_OPERATION_QUERY_PROPERTIES; 931 vstor_packet->flags = REQUEST_COMPLETION_FLAG; 932 933 ret = vmbus_sendpacket(device->channel, vstor_packet, 934 (sizeof(struct vstor_packet) - 935 vmscsi_size_delta), 936 (unsigned long)request, 937 VM_PKT_DATA_INBAND, 938 VMBUS_DATA_PACKET_FLAG_COMPLETION_REQUESTED); 939 940 if (ret != 0) 941 goto cleanup; 942 943 t = wait_for_completion_timeout(&request->wait_event, 5*HZ); 944 if (t == 0) { 945 ret = -ETIMEDOUT; 946 goto cleanup; 947 } 948 949 if (vstor_packet->operation != VSTOR_OPERATION_COMPLETE_IO || 950 vstor_packet->status != 0) 951 goto cleanup; 952 953 /* 954 * Check to see if multi-channel support is there. 955 * Hosts that implement protocol version of 5.1 and above 956 * support multi-channel. 957 */ 958 max_chns = vstor_packet->storage_channel_properties.max_channel_cnt; 959 if ((vmbus_proto_version != VERSION_WIN7) && 960 (vmbus_proto_version != VERSION_WS2008)) { 961 if (vstor_packet->storage_channel_properties.flags & 962 STORAGE_CHANNEL_SUPPORTS_MULTI_CHANNEL) 963 process_sub_channels = true; 964 } 965 966 memset(vstor_packet, 0, sizeof(struct vstor_packet)); 967 vstor_packet->operation = VSTOR_OPERATION_END_INITIALIZATION; 968 vstor_packet->flags = REQUEST_COMPLETION_FLAG; 969 970 ret = vmbus_sendpacket(device->channel, vstor_packet, 971 (sizeof(struct vstor_packet) - 972 vmscsi_size_delta), 973 (unsigned long)request, 974 VM_PKT_DATA_INBAND, 975 VMBUS_DATA_PACKET_FLAG_COMPLETION_REQUESTED); 976 977 if (ret != 0) 978 goto cleanup; 979 980 t = wait_for_completion_timeout(&request->wait_event, 5*HZ); 981 if (t == 0) { 982 ret = -ETIMEDOUT; 983 goto cleanup; 984 } 985 986 if (vstor_packet->operation != VSTOR_OPERATION_COMPLETE_IO || 987 vstor_packet->status != 0) 988 goto cleanup; 989 990 if (process_sub_channels) 991 handle_multichannel_storage(device, max_chns); 992 993 994 cleanup: 995 return ret; 996 } 997 998 static void storvsc_handle_error(struct vmscsi_request *vm_srb, 999 struct scsi_cmnd *scmnd, 1000 struct Scsi_Host *host, 1001 u8 asc, u8 ascq) 1002 { 1003 struct storvsc_scan_work *wrk; 1004 void (*process_err_fn)(struct work_struct *work); 1005 bool do_work = false; 1006 1007 switch (vm_srb->srb_status) { 1008 case SRB_STATUS_ERROR: 1009 /* 1010 * If there is an error; offline the device since all 1011 * error recovery strategies would have already been 1012 * deployed on the host side. However, if the command 1013 * were a pass-through command deal with it appropriately. 1014 */ 1015 switch (scmnd->cmnd[0]) { 1016 case ATA_16: 1017 case ATA_12: 1018 set_host_byte(scmnd, DID_PASSTHROUGH); 1019 break; 1020 default: 1021 set_host_byte(scmnd, DID_TARGET_FAILURE); 1022 } 1023 break; 1024 case SRB_STATUS_INVALID_LUN: 1025 do_work = true; 1026 process_err_fn = storvsc_remove_lun; 1027 break; 1028 case (SRB_STATUS_ABORTED | SRB_STATUS_AUTOSENSE_VALID): 1029 if ((asc == 0x2a) && (ascq == 0x9)) { 1030 do_work = true; 1031 process_err_fn = storvsc_device_scan; 1032 /* 1033 * Retry the I/O that trigerred this. 1034 */ 1035 set_host_byte(scmnd, DID_REQUEUE); 1036 } 1037 break; 1038 } 1039 1040 if (!do_work) 1041 return; 1042 1043 /* 1044 * We need to schedule work to process this error; schedule it. 1045 */ 1046 wrk = kmalloc(sizeof(struct storvsc_scan_work), GFP_ATOMIC); 1047 if (!wrk) { 1048 set_host_byte(scmnd, DID_TARGET_FAILURE); 1049 return; 1050 } 1051 1052 wrk->host = host; 1053 wrk->lun = vm_srb->lun; 1054 INIT_WORK(&wrk->work, process_err_fn); 1055 schedule_work(&wrk->work); 1056 } 1057 1058 1059 static void storvsc_command_completion(struct storvsc_cmd_request *cmd_request) 1060 { 1061 struct scsi_cmnd *scmnd = cmd_request->cmd; 1062 struct hv_host_device *host_dev = shost_priv(scmnd->device->host); 1063 void (*scsi_done_fn)(struct scsi_cmnd *); 1064 struct scsi_sense_hdr sense_hdr; 1065 struct vmscsi_request *vm_srb; 1066 struct stor_mem_pools *memp = scmnd->device->hostdata; 1067 struct Scsi_Host *host; 1068 struct storvsc_device *stor_dev; 1069 struct hv_device *dev = host_dev->dev; 1070 1071 stor_dev = get_in_stor_device(dev); 1072 host = stor_dev->host; 1073 1074 vm_srb = &cmd_request->vstor_packet.vm_srb; 1075 if (cmd_request->bounce_sgl_count) { 1076 if (vm_srb->data_in == READ_TYPE) 1077 copy_from_bounce_buffer(scsi_sglist(scmnd), 1078 cmd_request->bounce_sgl, 1079 scsi_sg_count(scmnd), 1080 cmd_request->bounce_sgl_count); 1081 destroy_bounce_buffer(cmd_request->bounce_sgl, 1082 cmd_request->bounce_sgl_count); 1083 } 1084 1085 scmnd->result = vm_srb->scsi_status; 1086 1087 if (scmnd->result) { 1088 if (scsi_normalize_sense(scmnd->sense_buffer, 1089 SCSI_SENSE_BUFFERSIZE, &sense_hdr)) 1090 scsi_print_sense_hdr("storvsc", &sense_hdr); 1091 } 1092 1093 if (vm_srb->srb_status != SRB_STATUS_SUCCESS) 1094 storvsc_handle_error(vm_srb, scmnd, host, sense_hdr.asc, 1095 sense_hdr.ascq); 1096 1097 scsi_set_resid(scmnd, 1098 cmd_request->data_buffer.len - 1099 vm_srb->data_transfer_length); 1100 1101 scsi_done_fn = scmnd->scsi_done; 1102 1103 scmnd->host_scribble = NULL; 1104 scmnd->scsi_done = NULL; 1105 1106 scsi_done_fn(scmnd); 1107 1108 mempool_free(cmd_request, memp->request_mempool); 1109 } 1110 1111 static void storvsc_on_io_completion(struct hv_device *device, 1112 struct vstor_packet *vstor_packet, 1113 struct storvsc_cmd_request *request) 1114 { 1115 struct storvsc_device *stor_device; 1116 struct vstor_packet *stor_pkt; 1117 1118 stor_device = hv_get_drvdata(device); 1119 stor_pkt = &request->vstor_packet; 1120 1121 /* 1122 * The current SCSI handling on the host side does 1123 * not correctly handle: 1124 * INQUIRY command with page code parameter set to 0x80 1125 * MODE_SENSE command with cmd[2] == 0x1c 1126 * 1127 * Setup srb and scsi status so this won't be fatal. 1128 * We do this so we can distinguish truly fatal failues 1129 * (srb status == 0x4) and off-line the device in that case. 1130 */ 1131 1132 if ((stor_pkt->vm_srb.cdb[0] == INQUIRY) || 1133 (stor_pkt->vm_srb.cdb[0] == MODE_SENSE)) { 1134 vstor_packet->vm_srb.scsi_status = 0; 1135 vstor_packet->vm_srb.srb_status = SRB_STATUS_SUCCESS; 1136 } 1137 1138 1139 /* Copy over the status...etc */ 1140 stor_pkt->vm_srb.scsi_status = vstor_packet->vm_srb.scsi_status; 1141 stor_pkt->vm_srb.srb_status = vstor_packet->vm_srb.srb_status; 1142 stor_pkt->vm_srb.sense_info_length = 1143 vstor_packet->vm_srb.sense_info_length; 1144 1145 if (vstor_packet->vm_srb.scsi_status != 0 || 1146 vstor_packet->vm_srb.srb_status != SRB_STATUS_SUCCESS){ 1147 dev_warn(&device->device, 1148 "cmd 0x%x scsi status 0x%x srb status 0x%x\n", 1149 stor_pkt->vm_srb.cdb[0], 1150 vstor_packet->vm_srb.scsi_status, 1151 vstor_packet->vm_srb.srb_status); 1152 } 1153 1154 if ((vstor_packet->vm_srb.scsi_status & 0xFF) == 0x02) { 1155 /* CHECK_CONDITION */ 1156 if (vstor_packet->vm_srb.srb_status & 1157 SRB_STATUS_AUTOSENSE_VALID) { 1158 /* autosense data available */ 1159 dev_warn(&device->device, 1160 "stor pkt %p autosense data valid - len %d\n", 1161 request, 1162 vstor_packet->vm_srb.sense_info_length); 1163 1164 memcpy(request->sense_buffer, 1165 vstor_packet->vm_srb.sense_data, 1166 vstor_packet->vm_srb.sense_info_length); 1167 1168 } 1169 } 1170 1171 stor_pkt->vm_srb.data_transfer_length = 1172 vstor_packet->vm_srb.data_transfer_length; 1173 1174 storvsc_command_completion(request); 1175 1176 if (atomic_dec_and_test(&stor_device->num_outstanding_req) && 1177 stor_device->drain_notify) 1178 wake_up(&stor_device->waiting_to_drain); 1179 1180 1181 } 1182 1183 static void storvsc_on_receive(struct hv_device *device, 1184 struct vstor_packet *vstor_packet, 1185 struct storvsc_cmd_request *request) 1186 { 1187 struct storvsc_scan_work *work; 1188 struct storvsc_device *stor_device; 1189 1190 switch (vstor_packet->operation) { 1191 case VSTOR_OPERATION_COMPLETE_IO: 1192 storvsc_on_io_completion(device, vstor_packet, request); 1193 break; 1194 1195 case VSTOR_OPERATION_REMOVE_DEVICE: 1196 case VSTOR_OPERATION_ENUMERATE_BUS: 1197 stor_device = get_in_stor_device(device); 1198 work = kmalloc(sizeof(struct storvsc_scan_work), GFP_ATOMIC); 1199 if (!work) 1200 return; 1201 1202 INIT_WORK(&work->work, storvsc_bus_scan); 1203 work->host = stor_device->host; 1204 schedule_work(&work->work); 1205 break; 1206 1207 default: 1208 break; 1209 } 1210 } 1211 1212 static void storvsc_on_channel_callback(void *context) 1213 { 1214 struct vmbus_channel *channel = (struct vmbus_channel *)context; 1215 struct hv_device *device; 1216 struct storvsc_device *stor_device; 1217 u32 bytes_recvd; 1218 u64 request_id; 1219 unsigned char packet[ALIGN(sizeof(struct vstor_packet), 8)]; 1220 struct storvsc_cmd_request *request; 1221 int ret; 1222 1223 if (channel->primary_channel != NULL) 1224 device = channel->primary_channel->device_obj; 1225 else 1226 device = channel->device_obj; 1227 1228 stor_device = get_in_stor_device(device); 1229 if (!stor_device) 1230 return; 1231 1232 do { 1233 ret = vmbus_recvpacket(channel, packet, 1234 ALIGN((sizeof(struct vstor_packet) - 1235 vmscsi_size_delta), 8), 1236 &bytes_recvd, &request_id); 1237 if (ret == 0 && bytes_recvd > 0) { 1238 1239 request = (struct storvsc_cmd_request *) 1240 (unsigned long)request_id; 1241 1242 if ((request == &stor_device->init_request) || 1243 (request == &stor_device->reset_request)) { 1244 1245 memcpy(&request->vstor_packet, packet, 1246 (sizeof(struct vstor_packet) - 1247 vmscsi_size_delta)); 1248 complete(&request->wait_event); 1249 } else { 1250 storvsc_on_receive(device, 1251 (struct vstor_packet *)packet, 1252 request); 1253 } 1254 } else { 1255 break; 1256 } 1257 } while (1); 1258 1259 return; 1260 } 1261 1262 static int storvsc_connect_to_vsp(struct hv_device *device, u32 ring_size) 1263 { 1264 struct vmstorage_channel_properties props; 1265 int ret; 1266 1267 memset(&props, 0, sizeof(struct vmstorage_channel_properties)); 1268 1269 ret = vmbus_open(device->channel, 1270 ring_size, 1271 ring_size, 1272 (void *)&props, 1273 sizeof(struct vmstorage_channel_properties), 1274 storvsc_on_channel_callback, device->channel); 1275 1276 if (ret != 0) 1277 return ret; 1278 1279 ret = storvsc_channel_init(device); 1280 1281 return ret; 1282 } 1283 1284 static int storvsc_dev_remove(struct hv_device *device) 1285 { 1286 struct storvsc_device *stor_device; 1287 unsigned long flags; 1288 1289 stor_device = hv_get_drvdata(device); 1290 1291 spin_lock_irqsave(&device->channel->inbound_lock, flags); 1292 stor_device->destroy = true; 1293 spin_unlock_irqrestore(&device->channel->inbound_lock, flags); 1294 1295 /* 1296 * At this point, all outbound traffic should be disable. We 1297 * only allow inbound traffic (responses) to proceed so that 1298 * outstanding requests can be completed. 1299 */ 1300 1301 storvsc_wait_to_drain(stor_device); 1302 1303 /* 1304 * Since we have already drained, we don't need to busy wait 1305 * as was done in final_release_stor_device() 1306 * Note that we cannot set the ext pointer to NULL until 1307 * we have drained - to drain the outgoing packets, we need to 1308 * allow incoming packets. 1309 */ 1310 spin_lock_irqsave(&device->channel->inbound_lock, flags); 1311 hv_set_drvdata(device, NULL); 1312 spin_unlock_irqrestore(&device->channel->inbound_lock, flags); 1313 1314 /* Close the channel */ 1315 vmbus_close(device->channel); 1316 1317 kfree(stor_device); 1318 return 0; 1319 } 1320 1321 static int storvsc_do_io(struct hv_device *device, 1322 struct storvsc_cmd_request *request) 1323 { 1324 struct storvsc_device *stor_device; 1325 struct vstor_packet *vstor_packet; 1326 struct vmbus_channel *outgoing_channel; 1327 int ret = 0; 1328 1329 vstor_packet = &request->vstor_packet; 1330 stor_device = get_out_stor_device(device); 1331 1332 if (!stor_device) 1333 return -ENODEV; 1334 1335 1336 request->device = device; 1337 /* 1338 * Select an an appropriate channel to send the request out. 1339 */ 1340 1341 outgoing_channel = vmbus_get_outgoing_channel(device->channel); 1342 1343 1344 vstor_packet->flags |= REQUEST_COMPLETION_FLAG; 1345 1346 vstor_packet->vm_srb.length = (sizeof(struct vmscsi_request) - 1347 vmscsi_size_delta); 1348 1349 1350 vstor_packet->vm_srb.sense_info_length = sense_buffer_size; 1351 1352 1353 vstor_packet->vm_srb.data_transfer_length = 1354 request->data_buffer.len; 1355 1356 vstor_packet->operation = VSTOR_OPERATION_EXECUTE_SRB; 1357 1358 if (request->data_buffer.len) { 1359 ret = vmbus_sendpacket_multipagebuffer(outgoing_channel, 1360 &request->data_buffer, 1361 vstor_packet, 1362 (sizeof(struct vstor_packet) - 1363 vmscsi_size_delta), 1364 (unsigned long)request); 1365 } else { 1366 ret = vmbus_sendpacket(device->channel, vstor_packet, 1367 (sizeof(struct vstor_packet) - 1368 vmscsi_size_delta), 1369 (unsigned long)request, 1370 VM_PKT_DATA_INBAND, 1371 VMBUS_DATA_PACKET_FLAG_COMPLETION_REQUESTED); 1372 } 1373 1374 if (ret != 0) 1375 return ret; 1376 1377 atomic_inc(&stor_device->num_outstanding_req); 1378 1379 return ret; 1380 } 1381 1382 static int storvsc_device_alloc(struct scsi_device *sdevice) 1383 { 1384 struct stor_mem_pools *memp; 1385 int number = STORVSC_MIN_BUF_NR; 1386 1387 memp = kzalloc(sizeof(struct stor_mem_pools), GFP_KERNEL); 1388 if (!memp) 1389 return -ENOMEM; 1390 1391 memp->request_pool = 1392 kmem_cache_create(dev_name(&sdevice->sdev_dev), 1393 sizeof(struct storvsc_cmd_request), 0, 1394 SLAB_HWCACHE_ALIGN, NULL); 1395 1396 if (!memp->request_pool) 1397 goto err0; 1398 1399 memp->request_mempool = mempool_create(number, mempool_alloc_slab, 1400 mempool_free_slab, 1401 memp->request_pool); 1402 1403 if (!memp->request_mempool) 1404 goto err1; 1405 1406 sdevice->hostdata = memp; 1407 1408 return 0; 1409 1410 err1: 1411 kmem_cache_destroy(memp->request_pool); 1412 1413 err0: 1414 kfree(memp); 1415 return -ENOMEM; 1416 } 1417 1418 static void storvsc_device_destroy(struct scsi_device *sdevice) 1419 { 1420 struct stor_mem_pools *memp = sdevice->hostdata; 1421 1422 mempool_destroy(memp->request_mempool); 1423 kmem_cache_destroy(memp->request_pool); 1424 kfree(memp); 1425 sdevice->hostdata = NULL; 1426 } 1427 1428 static int storvsc_device_configure(struct scsi_device *sdevice) 1429 { 1430 scsi_adjust_queue_depth(sdevice, MSG_SIMPLE_TAG, 1431 STORVSC_MAX_IO_REQUESTS); 1432 1433 blk_queue_max_segment_size(sdevice->request_queue, PAGE_SIZE); 1434 1435 blk_queue_bounce_limit(sdevice->request_queue, BLK_BOUNCE_ANY); 1436 1437 blk_queue_rq_timeout(sdevice->request_queue, (storvsc_timeout * HZ)); 1438 1439 sdevice->no_write_same = 1; 1440 1441 return 0; 1442 } 1443 1444 static int storvsc_get_chs(struct scsi_device *sdev, struct block_device * bdev, 1445 sector_t capacity, int *info) 1446 { 1447 sector_t nsect = capacity; 1448 sector_t cylinders = nsect; 1449 int heads, sectors_pt; 1450 1451 /* 1452 * We are making up these values; let us keep it simple. 1453 */ 1454 heads = 0xff; 1455 sectors_pt = 0x3f; /* Sectors per track */ 1456 sector_div(cylinders, heads * sectors_pt); 1457 if ((sector_t)(cylinders + 1) * heads * sectors_pt < nsect) 1458 cylinders = 0xffff; 1459 1460 info[0] = heads; 1461 info[1] = sectors_pt; 1462 info[2] = (int)cylinders; 1463 1464 return 0; 1465 } 1466 1467 static int storvsc_host_reset_handler(struct scsi_cmnd *scmnd) 1468 { 1469 struct hv_host_device *host_dev = shost_priv(scmnd->device->host); 1470 struct hv_device *device = host_dev->dev; 1471 1472 struct storvsc_device *stor_device; 1473 struct storvsc_cmd_request *request; 1474 struct vstor_packet *vstor_packet; 1475 int ret, t; 1476 1477 1478 stor_device = get_out_stor_device(device); 1479 if (!stor_device) 1480 return FAILED; 1481 1482 request = &stor_device->reset_request; 1483 vstor_packet = &request->vstor_packet; 1484 1485 init_completion(&request->wait_event); 1486 1487 vstor_packet->operation = VSTOR_OPERATION_RESET_BUS; 1488 vstor_packet->flags = REQUEST_COMPLETION_FLAG; 1489 vstor_packet->vm_srb.path_id = stor_device->path_id; 1490 1491 ret = vmbus_sendpacket(device->channel, vstor_packet, 1492 (sizeof(struct vstor_packet) - 1493 vmscsi_size_delta), 1494 (unsigned long)&stor_device->reset_request, 1495 VM_PKT_DATA_INBAND, 1496 VMBUS_DATA_PACKET_FLAG_COMPLETION_REQUESTED); 1497 if (ret != 0) 1498 return FAILED; 1499 1500 t = wait_for_completion_timeout(&request->wait_event, 5*HZ); 1501 if (t == 0) 1502 return TIMEOUT_ERROR; 1503 1504 1505 /* 1506 * At this point, all outstanding requests in the adapter 1507 * should have been flushed out and return to us 1508 * There is a potential race here where the host may be in 1509 * the process of responding when we return from here. 1510 * Just wait for all in-transit packets to be accounted for 1511 * before we return from here. 1512 */ 1513 storvsc_wait_to_drain(stor_device); 1514 1515 return SUCCESS; 1516 } 1517 1518 static bool storvsc_scsi_cmd_ok(struct scsi_cmnd *scmnd) 1519 { 1520 bool allowed = true; 1521 u8 scsi_op = scmnd->cmnd[0]; 1522 1523 switch (scsi_op) { 1524 /* the host does not handle WRITE_SAME, log accident usage */ 1525 case WRITE_SAME: 1526 /* 1527 * smartd sends this command and the host does not handle 1528 * this. So, don't send it. 1529 */ 1530 case SET_WINDOW: 1531 scmnd->result = ILLEGAL_REQUEST << 16; 1532 allowed = false; 1533 break; 1534 default: 1535 break; 1536 } 1537 return allowed; 1538 } 1539 1540 static int storvsc_queuecommand(struct Scsi_Host *host, struct scsi_cmnd *scmnd) 1541 { 1542 int ret; 1543 struct hv_host_device *host_dev = shost_priv(host); 1544 struct hv_device *dev = host_dev->dev; 1545 struct storvsc_cmd_request *cmd_request; 1546 unsigned int request_size = 0; 1547 int i; 1548 struct scatterlist *sgl; 1549 unsigned int sg_count = 0; 1550 struct vmscsi_request *vm_srb; 1551 struct stor_mem_pools *memp = scmnd->device->hostdata; 1552 1553 if (!storvsc_scsi_cmd_ok(scmnd)) { 1554 scmnd->scsi_done(scmnd); 1555 return 0; 1556 } 1557 1558 request_size = sizeof(struct storvsc_cmd_request); 1559 1560 cmd_request = mempool_alloc(memp->request_mempool, 1561 GFP_ATOMIC); 1562 1563 /* 1564 * We might be invoked in an interrupt context; hence 1565 * mempool_alloc() can fail. 1566 */ 1567 if (!cmd_request) 1568 return SCSI_MLQUEUE_DEVICE_BUSY; 1569 1570 memset(cmd_request, 0, sizeof(struct storvsc_cmd_request)); 1571 1572 /* Setup the cmd request */ 1573 cmd_request->cmd = scmnd; 1574 1575 scmnd->host_scribble = (unsigned char *)cmd_request; 1576 1577 vm_srb = &cmd_request->vstor_packet.vm_srb; 1578 vm_srb->win8_extension.time_out_value = 60; 1579 1580 1581 /* Build the SRB */ 1582 switch (scmnd->sc_data_direction) { 1583 case DMA_TO_DEVICE: 1584 vm_srb->data_in = WRITE_TYPE; 1585 vm_srb->win8_extension.srb_flags |= SRB_FLAGS_DATA_OUT; 1586 vm_srb->win8_extension.srb_flags |= 1587 (SRB_FLAGS_QUEUE_ACTION_ENABLE | 1588 SRB_FLAGS_DISABLE_SYNCH_TRANSFER); 1589 break; 1590 case DMA_FROM_DEVICE: 1591 vm_srb->data_in = READ_TYPE; 1592 vm_srb->win8_extension.srb_flags |= SRB_FLAGS_DATA_IN; 1593 vm_srb->win8_extension.srb_flags |= 1594 (SRB_FLAGS_QUEUE_ACTION_ENABLE | 1595 SRB_FLAGS_DISABLE_SYNCH_TRANSFER); 1596 break; 1597 default: 1598 vm_srb->data_in = UNKNOWN_TYPE; 1599 vm_srb->win8_extension.srb_flags = 0; 1600 break; 1601 } 1602 1603 1604 vm_srb->port_number = host_dev->port; 1605 vm_srb->path_id = scmnd->device->channel; 1606 vm_srb->target_id = scmnd->device->id; 1607 vm_srb->lun = scmnd->device->lun; 1608 1609 vm_srb->cdb_length = scmnd->cmd_len; 1610 1611 memcpy(vm_srb->cdb, scmnd->cmnd, vm_srb->cdb_length); 1612 1613 cmd_request->sense_buffer = scmnd->sense_buffer; 1614 1615 1616 cmd_request->data_buffer.len = scsi_bufflen(scmnd); 1617 if (scsi_sg_count(scmnd)) { 1618 sgl = (struct scatterlist *)scsi_sglist(scmnd); 1619 sg_count = scsi_sg_count(scmnd); 1620 1621 /* check if we need to bounce the sgl */ 1622 if (do_bounce_buffer(sgl, scsi_sg_count(scmnd)) != -1) { 1623 cmd_request->bounce_sgl = 1624 create_bounce_buffer(sgl, scsi_sg_count(scmnd), 1625 scsi_bufflen(scmnd), 1626 vm_srb->data_in); 1627 if (!cmd_request->bounce_sgl) { 1628 ret = SCSI_MLQUEUE_HOST_BUSY; 1629 goto queue_error; 1630 } 1631 1632 cmd_request->bounce_sgl_count = 1633 ALIGN(scsi_bufflen(scmnd), PAGE_SIZE) >> 1634 PAGE_SHIFT; 1635 1636 if (vm_srb->data_in == WRITE_TYPE) 1637 copy_to_bounce_buffer(sgl, 1638 cmd_request->bounce_sgl, 1639 scsi_sg_count(scmnd)); 1640 1641 sgl = cmd_request->bounce_sgl; 1642 sg_count = cmd_request->bounce_sgl_count; 1643 } 1644 1645 cmd_request->data_buffer.offset = sgl[0].offset; 1646 1647 for (i = 0; i < sg_count; i++) 1648 cmd_request->data_buffer.pfn_array[i] = 1649 page_to_pfn(sg_page((&sgl[i]))); 1650 1651 } else if (scsi_sglist(scmnd)) { 1652 cmd_request->data_buffer.offset = 1653 virt_to_phys(scsi_sglist(scmnd)) & (PAGE_SIZE-1); 1654 cmd_request->data_buffer.pfn_array[0] = 1655 virt_to_phys(scsi_sglist(scmnd)) >> PAGE_SHIFT; 1656 } 1657 1658 /* Invokes the vsc to start an IO */ 1659 ret = storvsc_do_io(dev, cmd_request); 1660 1661 if (ret == -EAGAIN) { 1662 /* no more space */ 1663 1664 if (cmd_request->bounce_sgl_count) { 1665 destroy_bounce_buffer(cmd_request->bounce_sgl, 1666 cmd_request->bounce_sgl_count); 1667 1668 ret = SCSI_MLQUEUE_DEVICE_BUSY; 1669 goto queue_error; 1670 } 1671 } 1672 1673 return 0; 1674 1675 queue_error: 1676 mempool_free(cmd_request, memp->request_mempool); 1677 scmnd->host_scribble = NULL; 1678 return ret; 1679 } 1680 1681 static struct scsi_host_template scsi_driver = { 1682 .module = THIS_MODULE, 1683 .name = "storvsc_host_t", 1684 .bios_param = storvsc_get_chs, 1685 .queuecommand = storvsc_queuecommand, 1686 .eh_host_reset_handler = storvsc_host_reset_handler, 1687 .slave_alloc = storvsc_device_alloc, 1688 .slave_destroy = storvsc_device_destroy, 1689 .slave_configure = storvsc_device_configure, 1690 .cmd_per_lun = 1, 1691 /* 64 max_queue * 1 target */ 1692 .can_queue = STORVSC_MAX_IO_REQUESTS*STORVSC_MAX_TARGETS, 1693 .this_id = -1, 1694 /* no use setting to 0 since ll_blk_rw reset it to 1 */ 1695 /* currently 32 */ 1696 .sg_tablesize = MAX_MULTIPAGE_BUFFER_COUNT, 1697 .use_clustering = DISABLE_CLUSTERING, 1698 /* Make sure we dont get a sg segment crosses a page boundary */ 1699 .dma_boundary = PAGE_SIZE-1, 1700 }; 1701 1702 enum { 1703 SCSI_GUID, 1704 IDE_GUID, 1705 SFC_GUID, 1706 }; 1707 1708 static const struct hv_vmbus_device_id id_table[] = { 1709 /* SCSI guid */ 1710 { HV_SCSI_GUID, 1711 .driver_data = SCSI_GUID 1712 }, 1713 /* IDE guid */ 1714 { HV_IDE_GUID, 1715 .driver_data = IDE_GUID 1716 }, 1717 /* Fibre Channel GUID */ 1718 { 1719 HV_SYNTHFC_GUID, 1720 .driver_data = SFC_GUID 1721 }, 1722 { }, 1723 }; 1724 1725 MODULE_DEVICE_TABLE(vmbus, id_table); 1726 1727 static int storvsc_probe(struct hv_device *device, 1728 const struct hv_vmbus_device_id *dev_id) 1729 { 1730 int ret; 1731 struct Scsi_Host *host; 1732 struct hv_host_device *host_dev; 1733 bool dev_is_ide = ((dev_id->driver_data == IDE_GUID) ? true : false); 1734 int target = 0; 1735 struct storvsc_device *stor_device; 1736 1737 /* 1738 * Based on the windows host we are running on, 1739 * set state to properly communicate with the host. 1740 */ 1741 1742 if (vmbus_proto_version == VERSION_WIN8) { 1743 sense_buffer_size = POST_WIN7_STORVSC_SENSE_BUFFER_SIZE; 1744 vmscsi_size_delta = 0; 1745 vmstor_current_major = VMSTOR_WIN8_MAJOR; 1746 vmstor_current_minor = VMSTOR_WIN8_MINOR; 1747 } else { 1748 sense_buffer_size = PRE_WIN8_STORVSC_SENSE_BUFFER_SIZE; 1749 vmscsi_size_delta = sizeof(struct vmscsi_win8_extension); 1750 vmstor_current_major = VMSTOR_WIN7_MAJOR; 1751 vmstor_current_minor = VMSTOR_WIN7_MINOR; 1752 } 1753 1754 1755 host = scsi_host_alloc(&scsi_driver, 1756 sizeof(struct hv_host_device)); 1757 if (!host) 1758 return -ENOMEM; 1759 1760 host_dev = shost_priv(host); 1761 memset(host_dev, 0, sizeof(struct hv_host_device)); 1762 1763 host_dev->port = host->host_no; 1764 host_dev->dev = device; 1765 1766 1767 stor_device = kzalloc(sizeof(struct storvsc_device), GFP_KERNEL); 1768 if (!stor_device) { 1769 ret = -ENOMEM; 1770 goto err_out0; 1771 } 1772 1773 stor_device->destroy = false; 1774 stor_device->open_sub_channel = false; 1775 init_waitqueue_head(&stor_device->waiting_to_drain); 1776 stor_device->device = device; 1777 stor_device->host = host; 1778 hv_set_drvdata(device, stor_device); 1779 1780 stor_device->port_number = host->host_no; 1781 ret = storvsc_connect_to_vsp(device, storvsc_ringbuffer_size); 1782 if (ret) 1783 goto err_out1; 1784 1785 host_dev->path = stor_device->path_id; 1786 host_dev->target = stor_device->target_id; 1787 1788 /* max # of devices per target */ 1789 host->max_lun = STORVSC_MAX_LUNS_PER_TARGET; 1790 /* max # of targets per channel */ 1791 host->max_id = STORVSC_MAX_TARGETS; 1792 /* max # of channels */ 1793 host->max_channel = STORVSC_MAX_CHANNELS - 1; 1794 /* max cmd length */ 1795 host->max_cmd_len = STORVSC_MAX_CMD_LEN; 1796 1797 /* Register the HBA and start the scsi bus scan */ 1798 ret = scsi_add_host(host, &device->device); 1799 if (ret != 0) 1800 goto err_out2; 1801 1802 if (!dev_is_ide) { 1803 scsi_scan_host(host); 1804 } else { 1805 target = (device->dev_instance.b[5] << 8 | 1806 device->dev_instance.b[4]); 1807 ret = scsi_add_device(host, 0, target, 0); 1808 if (ret) { 1809 scsi_remove_host(host); 1810 goto err_out2; 1811 } 1812 } 1813 return 0; 1814 1815 err_out2: 1816 /* 1817 * Once we have connected with the host, we would need to 1818 * to invoke storvsc_dev_remove() to rollback this state and 1819 * this call also frees up the stor_device; hence the jump around 1820 * err_out1 label. 1821 */ 1822 storvsc_dev_remove(device); 1823 goto err_out0; 1824 1825 err_out1: 1826 kfree(stor_device); 1827 1828 err_out0: 1829 scsi_host_put(host); 1830 return ret; 1831 } 1832 1833 static int storvsc_remove(struct hv_device *dev) 1834 { 1835 struct storvsc_device *stor_device = hv_get_drvdata(dev); 1836 struct Scsi_Host *host = stor_device->host; 1837 1838 scsi_remove_host(host); 1839 storvsc_dev_remove(dev); 1840 scsi_host_put(host); 1841 1842 return 0; 1843 } 1844 1845 static struct hv_driver storvsc_drv = { 1846 .name = KBUILD_MODNAME, 1847 .id_table = id_table, 1848 .probe = storvsc_probe, 1849 .remove = storvsc_remove, 1850 }; 1851 1852 static int __init storvsc_drv_init(void) 1853 { 1854 u32 max_outstanding_req_per_channel; 1855 1856 /* 1857 * Divide the ring buffer data size (which is 1 page less 1858 * than the ring buffer size since that page is reserved for 1859 * the ring buffer indices) by the max request size (which is 1860 * vmbus_channel_packet_multipage_buffer + struct vstor_packet + u64) 1861 */ 1862 max_outstanding_req_per_channel = 1863 ((storvsc_ringbuffer_size - PAGE_SIZE) / 1864 ALIGN(MAX_MULTIPAGE_BUFFER_PACKET + 1865 sizeof(struct vstor_packet) + sizeof(u64) - 1866 vmscsi_size_delta, 1867 sizeof(u64))); 1868 1869 if (max_outstanding_req_per_channel < 1870 STORVSC_MAX_IO_REQUESTS) 1871 return -EINVAL; 1872 1873 return vmbus_driver_register(&storvsc_drv); 1874 } 1875 1876 static void __exit storvsc_drv_exit(void) 1877 { 1878 vmbus_driver_unregister(&storvsc_drv); 1879 } 1880 1881 MODULE_LICENSE("GPL"); 1882 MODULE_DESCRIPTION("Microsoft Hyper-V virtual storage driver"); 1883 module_init(storvsc_drv_init); 1884 module_exit(storvsc_drv_exit); 1885