This patch adds support for Zoned Block Devices (ZBDs) to the kernel virtio-blk driver. The patch accompanies the virtio-blk ZBD support draft that is now being proposed for standardization. The latest version of the draft is linked at https://github.com/oasis-tcs/virtio-spec/issues/143 . The QEMU zoned device code that implements these protocol extensions has been developed by Sam Li and it is currently in review at the QEMU mailing list. A number of virtblk request structure changes has been introduced to accommodate the functionality that is specific to zoned block devices and, most importantly, make room for carrying the Zoned Append sector value from the device back to the driver along with the request status. The zone-specific code in the patch is heavily influenced by NVMe ZNS code in drivers/nvme/host/zns.c, but it is simpler because the proposed virtio ZBD draft only covers the zoned device features that are relevant to the zoned functionality provided by Linux block layer. includes the following fixup: virtio-blk: fix probe without CONFIG_BLK_DEV_ZONED When building without CONFIG_BLK_DEV_ZONED, VIRTIO_BLK_F_ZONED is excluded from array of driver features. As a result virtio_has_feature panics in virtio_check_driver_offered_feature since that by design verifies that a feature we are checking for is listed in the feature array. To fix, replace the call to virtio_has_feature with a stub.
Unified · Split
| @@ -15,6 +15,7 @@ · 14 unmodified lines | |||
| 15 | #include <linux/blk-mq.h> | 15 | #include <linux/blk-mq.h> |
| 16 | #include <linux/blk-mq-virtio.h> | 16 | #include <linux/blk-mq-virtio.h> |
| 17 | #include <linux/numa.h> | 17 | #include <linux/numa.h> |
| 18 | #include <linux/vmalloc.h> | ||
| 18 | #include <uapi/linux/virtio_ring.h> | 19 | #include <uapi/linux/virtio_ring.h> |
| 19 | 20 | ||
| 20 | #define PART_BITS 4 | 21 | #define PART_BITS 4 |
| @@ -80,22 +81,51 @@ struct virtio_blk { · 59 unmodified lines | |||
| 80 | int num_vqs; | 81 | int num_vqs; |
| 81 | int io_queues[HCTX_MAX_TYPES]; | 82 | int io_queues[HCTX_MAX_TYPES]; |
| 82 | struct virtio_blk_vq *vqs; | 83 | struct virtio_blk_vq *vqs; |
| 84 | |||
| 85 | /* For zoned device */ | ||
| 86 | unsigned int zone_sectors; | ||
| 83 | }; | 87 | }; |
| 84 | 88 | ||
| 85 | struct virtblk_req { | 89 | struct virtblk_req { |
| 90 | /* Out header */ | ||
| 86 | struct virtio_blk_outhdr out_hdr; | 91 | struct virtio_blk_outhdr out_hdr; |
| 87 | u8 status; | 92 | |
| 93 | /* In header */ | ||
| 94 | union { | ||
| 95 | u8 status; | ||
| 96 | |||
| 97 | /* | ||
| 98 | * The zone append command has an extended in header. | ||
| 99 | * The status field in zone_append_in_hdr must have | ||
| 100 | * the same offset in virtblk_req as the non-zoned | ||
| 101 | * status field above. | ||
| 102 | */ | ||
| 103 | struct { | ||
| 104 | u8 status; | ||
| 105 | u8 reserved[7]; | ||
| 106 | u64 append_sector; | ||
| 107 | } zone_append_in_hdr; | ||
| 108 | }; | ||
| 109 | |||
| 110 | size_t in_hdr_len; | ||
| 111 | |||
| 88 | struct sg_table sg_table; | 112 | struct sg_table sg_table; |
| 89 | struct scatterlist sg[]; | 113 | struct scatterlist sg[]; |
| 90 | }; | 114 | }; |
| 91 | 115 | ||
| 92 | static inline blk_status_t virtblk_result(struct virtblk_req *vbr) | 116 | static inline blk_status_t virtblk_result(u8 status) |
| 93 | { | 117 | { |
| 94 | switch (vbr->status) { | 118 | switch (status) { |
| 95 | case VIRTIO_BLK_S_OK: | 119 | case VIRTIO_BLK_S_OK: |
| 96 | return BLK_STS_OK; | 120 | return BLK_STS_OK; |
| 97 | case VIRTIO_BLK_S_UNSUPP: | 121 | case VIRTIO_BLK_S_UNSUPP: |
| 98 | return BLK_STS_NOTSUPP; | 122 | return BLK_STS_NOTSUPP; |
| 123 | case VIRTIO_BLK_S_ZONE_OPEN_RESOURCE: | ||
| 124 | return BLK_STS_ZONE_OPEN_RESOURCE; | ||
| 125 | case VIRTIO_BLK_S_ZONE_ACTIVE_RESOURCE: | ||
| 126 | return BLK_STS_ZONE_ACTIVE_RESOURCE; | ||
| 127 | case VIRTIO_BLK_S_IOERR: | ||
| 128 | case VIRTIO_BLK_S_ZONE_UNALIGNED_WP: | ||
| 99 | default: | 129 | default: |
| 100 | return BLK_STS_IOERR; | 130 | return BLK_STS_IOERR; |
| 101 | } | 131 | } |
| @@ -111,11 +141,11 @@ static inline struct virtio_blk_vq *get_virtio_blk_vq(struct blk_mq_hw_ctx *hctx · 9 unmodified lines | |||
| 111 | 141 | ||
| 112 | static int virtblk_add_req(struct virtqueue *vq, struct virtblk_req *vbr) | 142 | static int virtblk_add_req(struct virtqueue *vq, struct virtblk_req *vbr) |
| 113 | { | 143 | { |
| 114 | struct scatterlist hdr, status, *sgs[3]; | 144 | struct scatterlist out_hdr, in_hdr, *sgs[3]; |
| 115 | unsigned int num_out = 0, num_in = 0; | 145 | unsigned int num_out = 0, num_in = 0; |
| 116 | 146 | ||
| 117 | sg_init_one(&hdr, &vbr->out_hdr, sizeof(vbr->out_hdr)); | 147 | sg_init_one(&out_hdr, &vbr->out_hdr, sizeof(vbr->out_hdr)); |
| 118 | sgs[num_out++] = &hdr; | 148 | sgs[num_out++] = &out_hdr; |
| 119 | 149 | ||
| 120 | if (vbr->sg_table.nents) { | 150 | if (vbr->sg_table.nents) { |
| 121 | if (vbr->out_hdr.type & cpu_to_virtio32(vq->vdev, VIRTIO_BLK_T_OUT)) | 151 | if (vbr->out_hdr.type & cpu_to_virtio32(vq->vdev, VIRTIO_BLK_T_OUT)) |
| @@ -124,8 +154,8 @@ static int virtblk_add_req(struct virtqueue *vq, struct virtblk_req *vbr) · 2 unmodified lines | |||
| 124 | sgs[num_out + num_in++] = vbr->sg_table.sgl; | 154 | sgs[num_out + num_in++] = vbr->sg_table.sgl; |
| 125 | } | 155 | } |
| 126 | 156 | ||
| 127 | sg_init_one(&status, &vbr->status, sizeof(vbr->status)); | 157 | sg_init_one(&in_hdr, &vbr->status, vbr->in_hdr_len); |
| 128 | sgs[num_out + num_in++] = &status; | 158 | sgs[num_out + num_in++] = &in_hdr; |
| 129 | 159 | ||
| 130 | return virtqueue_add_sgs(vq, sgs, num_out, num_in, vbr, GFP_ATOMIC); | 160 | return virtqueue_add_sgs(vq, sgs, num_out, num_in, vbr, GFP_ATOMIC); |
| 131 | } | 161 | } |
| @@ -214,21 +244,22 @@ static blk_status_t virtblk_setup_cmd(struct virtio_device *vdev, · 82 unmodified lines | |||
| 214 | struct request *req, | 244 | struct request *req, |
| 215 | struct virtblk_req *vbr) | 245 | struct virtblk_req *vbr) |
| 216 | { | 246 | { |
| 247 | size_t in_hdr_len = sizeof(vbr->status); | ||
| 217 | bool unmap = false; | 248 | bool unmap = false; |
| 218 | u32 type; | 249 | u32 type; |
| 250 | u64 sector = 0; | ||
| 219 | 251 | ||
| 220 | vbr->out_hdr.sector = 0; | 252 | /* Set fields for all request types */ |
| 253 | vbr->out_hdr.ioprio = cpu_to_virtio32(vdev, req_get_ioprio(req)); | ||
| 221 | 254 | ||
| 222 | switch (req_op(req)) { | 255 | switch (req_op(req)) { |
| 223 | case REQ_OP_READ: | 256 | case REQ_OP_READ: |
| 224 | type = VIRTIO_BLK_T_IN; | 257 | type = VIRTIO_BLK_T_IN; |
| 225 | vbr->out_hdr.sector = cpu_to_virtio64(vdev, | 258 | sector = blk_rq_pos(req); |
| 226 | blk_rq_pos(req)); | ||
| 227 | break; | 259 | break; |
| 228 | case REQ_OP_WRITE: | 260 | case REQ_OP_WRITE: |
| 229 | type = VIRTIO_BLK_T_OUT; | 261 | type = VIRTIO_BLK_T_OUT; |
| 230 | vbr->out_hdr.sector = cpu_to_virtio64(vdev, | 262 | sector = blk_rq_pos(req); |
| 231 | blk_rq_pos(req)); | ||
| 232 | break; | 263 | break; |
| 233 | case REQ_OP_FLUSH: | 264 | case REQ_OP_FLUSH: |
| 234 | type = VIRTIO_BLK_T_FLUSH; | 265 | type = VIRTIO_BLK_T_FLUSH; |
| @@ -243,16 +274,42 @@ static blk_status_t virtblk_setup_cmd(struct virtio_device *vdev, · 8 unmodified lines | |||
| 243 | case REQ_OP_SECURE_ERASE: | 274 | case REQ_OP_SECURE_ERASE: |
| 244 | type = VIRTIO_BLK_T_SECURE_ERASE; | 275 | type = VIRTIO_BLK_T_SECURE_ERASE; |
| 245 | break; | 276 | break; |
| 246 | case REQ_OP_DRV_IN: | 277 | case REQ_OP_ZONE_OPEN: |
| 247 | type = VIRTIO_BLK_T_GET_ID; | 278 | type = VIRTIO_BLK_T_ZONE_OPEN; |
| 279 | sector = blk_rq_pos(req); | ||
| 280 | break; | ||
| 281 | case REQ_OP_ZONE_CLOSE: | ||
| 282 | type = VIRTIO_BLK_T_ZONE_CLOSE; | ||
| 283 | sector = blk_rq_pos(req); | ||
| 284 | break; | ||
| 285 | case REQ_OP_ZONE_FINISH: | ||
| 286 | type = VIRTIO_BLK_T_ZONE_FINISH; | ||
| 287 | sector = blk_rq_pos(req); | ||
| 288 | break; | ||
| 289 | case REQ_OP_ZONE_APPEND: | ||
| 290 | type = VIRTIO_BLK_T_ZONE_APPEND; | ||
| 291 | sector = blk_rq_pos(req); | ||
| 292 | in_hdr_len = sizeof(vbr->zone_append_in_hdr); | ||
| 248 | break; | 293 | break; |
| 294 | case REQ_OP_ZONE_RESET: | ||
| 295 | type = VIRTIO_BLK_T_ZONE_RESET; | ||
| 296 | sector = blk_rq_pos(req); | ||
| 297 | break; | ||
| 298 | case REQ_OP_ZONE_RESET_ALL: | ||
| 299 | type = VIRTIO_BLK_T_ZONE_RESET_ALL; | ||
| 300 | break; | ||
| 301 | case REQ_OP_DRV_IN: | ||
| 302 | /* Out header already filled in, nothing to do */ | ||
| 303 | return 0; | ||
| 249 | default: | 304 | default: |
| 250 | WARN_ON_ONCE(1); | 305 | WARN_ON_ONCE(1); |
| 251 | return BLK_STS_IOERR; | 306 | return BLK_STS_IOERR; |
| 252 | } | 307 | } |
| 253 | 308 | ||
| 309 | /* Set fields for non-REQ_OP_DRV_IN request types */ | ||
| 310 | vbr->in_hdr_len = in_hdr_len; | ||
| 254 | vbr->out_hdr.type = cpu_to_virtio32(vdev, type); | 311 | vbr->out_hdr.type = cpu_to_virtio32(vdev, type); |
| 255 | vbr->out_hdr.ioprio = cpu_to_virtio32(vdev, req_get_ioprio(req)); | 312 | vbr->out_hdr.sector = cpu_to_virtio64(vdev, sector); |
| 256 | 313 | ||
| 257 | if (type == VIRTIO_BLK_T_DISCARD || type == VIRTIO_BLK_T_WRITE_ZEROES || | 314 | if (type == VIRTIO_BLK_T_DISCARD || type == VIRTIO_BLK_T_WRITE_ZEROES || |
| 258 | type == VIRTIO_BLK_T_SECURE_ERASE) { | 315 | type == VIRTIO_BLK_T_SECURE_ERASE) { |
| @@ -266,10 +323,15 @@ static blk_status_t virtblk_setup_cmd(struct virtio_device *vdev, · 7 unmodified lines | |||
| 266 | static inline void virtblk_request_done(struct request *req) | 323 | static inline void virtblk_request_done(struct request *req) |
| 267 | { | 324 | { |
| 268 | struct virtblk_req *vbr = blk_mq_rq_to_pdu(req); | 325 | struct virtblk_req *vbr = blk_mq_rq_to_pdu(req); |
| 326 | int status = virtblk_result(vbr->status); | ||
| 269 | 327 | ||
| 270 | virtblk_unmap_data(req, vbr); | 328 | virtblk_unmap_data(req, vbr); |
| 271 | virtblk_cleanup_cmd(req); | 329 | virtblk_cleanup_cmd(req); |
| 272 | blk_mq_end_request(req, virtblk_result(vbr)); | 330 | |
| 331 | if (req_op(req) == REQ_OP_ZONE_APPEND) | ||
| 332 | req->__sector = le64_to_cpu(vbr->zone_append_in_hdr.append_sector); | ||
| 333 | |||
| 334 | blk_mq_end_request(req, status); | ||
| 273 | } | 335 | } |
| 274 | 336 | ||
| 275 | static void virtblk_done(struct virtqueue *vq) | 337 | static void virtblk_done(struct virtqueue *vq) |
| @@ -457,6 +519,275 @@ static void virtio_queue_rqs(struct request **rqlist) · 181 unmodified lines | |||
| 457 | *rqlist = requeue_list; | 519 | *rqlist = requeue_list; |
| 458 | } | 520 | } |
| 459 | 521 | ||
| 522 | #ifdef CONFIG_BLK_DEV_ZONED | ||
| 523 | static void *virtblk_alloc_report_buffer(struct virtio_blk *vblk, | ||
| 524 | unsigned int nr_zones, | ||
| 525 | unsigned int zone_sectors, | ||
| 526 | size_t *buflen) | ||
| 527 | { | ||
| 528 | struct request_queue *q = vblk->disk->queue; | ||
| 529 | size_t bufsize; | ||
| 530 | void *buf; | ||
| 531 | |||
| 532 | nr_zones = min_t(unsigned int, nr_zones, | ||
| 533 | get_capacity(vblk->disk) >> ilog2(zone_sectors)); | ||
| 534 | |||
| 535 | bufsize = sizeof(struct virtio_blk_zone_report) + | ||
| 536 | nr_zones * sizeof(struct virtio_blk_zone_descriptor); | ||
| 537 | bufsize = min_t(size_t, bufsize, | ||
| 538 | queue_max_hw_sectors(q) << SECTOR_SHIFT); | ||
| 539 | bufsize = min_t(size_t, bufsize, queue_max_segments(q) << PAGE_SHIFT); | ||
| 540 | |||
| 541 | while (bufsize >= sizeof(struct virtio_blk_zone_report)) { | ||
| 542 | buf = __vmalloc(bufsize, GFP_KERNEL | __GFP_NORETRY); | ||
| 543 | if (buf) { | ||
| 544 | *buflen = bufsize; | ||
| 545 | return buf; | ||
| 546 | } | ||
| 547 | bufsize >>= 1; | ||
| 548 | } | ||
| 549 | |||
| 550 | return NULL; | ||
| 551 | } | ||
| 552 | |||
| 553 | static int virtblk_submit_zone_report(struct virtio_blk *vblk, | ||
| 554 | char *report_buf, size_t report_len, | ||
| 555 | sector_t sector) | ||
| 556 | { | ||
| 557 | struct request_queue *q = vblk->disk->queue; | ||
| 558 | struct request *req; | ||
| 559 | struct virtblk_req *vbr; | ||
| 560 | int err; | ||
| 561 | |||
| 562 | req = blk_mq_alloc_request(q, REQ_OP_DRV_IN, 0); | ||
| 563 | if (IS_ERR(req)) | ||
| 564 | return PTR_ERR(req); | ||
| 565 | |||
| 566 | vbr = blk_mq_rq_to_pdu(req); | ||
| 567 | vbr->in_hdr_len = sizeof(vbr->status); | ||
| 568 | vbr->out_hdr.type = cpu_to_virtio32(vblk->vdev, VIRTIO_BLK_T_ZONE_REPORT); | ||
| 569 | vbr->out_hdr.sector = cpu_to_virtio64(vblk->vdev, sector); | ||
| 570 | |||
| 571 | err = blk_rq_map_kern(q, req, report_buf, report_len, GFP_KERNEL); | ||
| 572 | if (err) | ||
| 573 | goto out; | ||
| 574 | |||
| 575 | blk_execute_rq(req, false); | ||
| 576 | err = blk_status_to_errno(virtblk_result(vbr->status)); | ||
| 577 | out: | ||
| 578 | blk_mq_free_request(req); | ||
| 579 | return err; | ||
| 580 | } | ||
| 581 | |||
| 582 | static int virtblk_parse_zone(struct virtio_blk *vblk, | ||
| 583 | struct virtio_blk_zone_descriptor *entry, | ||
| 584 | unsigned int idx, unsigned int zone_sectors, | ||
| 585 | report_zones_cb cb, void *data) | ||
| 586 | { | ||
| 587 | struct blk_zone zone = { }; | ||
| 588 | |||
| 589 | if (entry->z_type != VIRTIO_BLK_ZT_SWR && | ||
| 590 | entry->z_type != VIRTIO_BLK_ZT_SWP && | ||
| 591 | entry->z_type != VIRTIO_BLK_ZT_CONV) { | ||
| 592 | dev_err(&vblk->vdev->dev, "invalid zone type %#x\n", | ||
| 593 | entry->z_type); | ||
| 594 | return -EINVAL; | ||
| 595 | } | ||
| 596 | |||
| 597 | zone.type = entry->z_type; | ||
| 598 | zone.cond = entry->z_state; | ||
| 599 | zone.len = zone_sectors; | ||
| 600 | zone.capacity = le64_to_cpu(entry->z_cap); | ||
| 601 | zone.start = le64_to_cpu(entry->z_start); | ||
| 602 | if (zone.cond == BLK_ZONE_COND_FULL) | ||
| 603 | zone.wp = zone.start + zone.len; | ||
| 604 | else | ||
| 605 | zone.wp = le64_to_cpu(entry->z_wp); | ||
| 606 | |||
| 607 | return cb(&zone, idx, data); | ||
| 608 | } | ||
| 609 | |||
| 610 | static int virtblk_report_zones(struct gendisk *disk, sector_t sector, | ||
| 611 | unsigned int nr_zones, report_zones_cb cb, | ||
| 612 | void *data) | ||
| 613 | { | ||
| 614 | struct virtio_blk *vblk = disk->private_data; | ||
| 615 | struct virtio_blk_zone_report *report; | ||
| 616 | unsigned int zone_sectors = vblk->zone_sectors; | ||
| 617 | unsigned int nz, i; | ||
| 618 | int ret, zone_idx = 0; | ||
| 619 | size_t buflen; | ||
| 620 | |||
| 621 | if (WARN_ON_ONCE(!vblk->zone_sectors)) | ||
| 622 | return -EOPNOTSUPP; | ||
| 623 | |||
| 624 | report = virtblk_alloc_report_buffer(vblk, nr_zones, | ||
| 625 | zone_sectors, &buflen); | ||
| 626 | if (!report) | ||
| 627 | return -ENOMEM; | ||
| 628 | |||
| 629 | while (zone_idx < nr_zones && sector < get_capacity(vblk->disk)) { | ||
| 630 | memset(report, 0, buflen); | ||
| 631 | |||
| 632 | ret = virtblk_submit_zone_report(vblk, (char *)report, | ||
| 633 | buflen, sector); | ||
| 634 | if (ret) { | ||
| 635 | if (ret > 0) | ||
| 636 | ret = -EIO; | ||
| 637 | goto out_free; | ||
| 638 | } | ||
| 639 | nz = min((unsigned int)le64_to_cpu(report->nr_zones), nr_zones); | ||
| 640 | if (!nz) | ||
| 641 | break; | ||
| 642 | |||
| 643 | for (i = 0; i < nz && zone_idx < nr_zones; i++) { | ||
| 644 | ret = virtblk_parse_zone(vblk, &report->zones[i], | ||
| 645 | zone_idx, zone_sectors, cb, data); | ||
| 646 | if (ret) | ||
| 647 | goto out_free; | ||
| 648 | sector = le64_to_cpu(report->zones[i].z_start) + zone_sectors; | ||
| 649 | zone_idx++; | ||
| 650 | } | ||
| 651 | } | ||
| 652 | |||
| 653 | if (zone_idx > 0) | ||
| 654 | ret = zone_idx; | ||
| 655 | else | ||
| 656 | ret = -EINVAL; | ||
| 657 | out_free: | ||
| 658 | kvfree(report); | ||
| 659 | return ret; | ||
| 660 | } | ||
| 661 | |||
| 662 | static void virtblk_revalidate_zones(struct virtio_blk *vblk) | ||
| 663 | { | ||
| 664 | u8 model; | ||
| 665 | |||
| 666 | if (!vblk->zone_sectors) | ||
| 667 | return; | ||
| 668 | |||
| 669 | virtio_cread(vblk->vdev, struct virtio_blk_config, | ||
| 670 | zoned.model, &model); | ||
| 671 | if (!blk_revalidate_disk_zones(vblk->disk, NULL)) | ||
| 672 | set_capacity_and_notify(vblk->disk, 0); | ||
| 673 | } | ||
| 674 | |||
| 675 | static int virtblk_probe_zoned_device(struct virtio_device *vdev, | ||
| 676 | struct virtio_blk *vblk, | ||
| 677 | struct request_queue *q) | ||
| 678 | { | ||
| 679 | u32 v; | ||
| 680 | u8 model; | ||
| 681 | int ret; | ||
| 682 | |||
| 683 | virtio_cread(vdev, struct virtio_blk_config, | ||
| 684 | zoned.model, &model); | ||
| 685 | |||
| 686 | switch (model) { | ||
| 687 | case VIRTIO_BLK_Z_NONE: | ||
| 688 | return 0; | ||
| 689 | case VIRTIO_BLK_Z_HM: | ||
| 690 | break; | ||
| 691 | case VIRTIO_BLK_Z_HA: | ||
| 692 | /* | ||
| 693 | * Present the host-aware device as a regular drive. | ||
| 694 | * TODO It is possible to add an option to make it appear | ||
| 695 | * in the system as a zoned drive. | ||
| 696 | */ | ||
| 697 | return 0; | ||
| 698 | default: | ||
| 699 | dev_err(&vdev->dev, "unsupported zone model %d\n", model); | ||
| 700 | return -EINVAL; | ||
| 701 | } | ||
| 702 | |||
| 703 | dev_dbg(&vdev->dev, "probing host-managed zoned device\n"); | ||
| 704 | |||
| 705 | disk_set_zoned(vblk->disk, BLK_ZONED_HM); | ||
| 706 | blk_queue_flag_set(QUEUE_FLAG_ZONE_RESETALL, q); | ||
| 707 | |||
| 708 | virtio_cread(vdev, struct virtio_blk_config, | ||
| 709 | zoned.max_open_zones, &v); | ||
| 710 | disk_set_max_open_zones(vblk->disk, le32_to_cpu(v)); | ||
| 711 | |||
| 712 | dev_dbg(&vdev->dev, "max open zones = %u\n", le32_to_cpu(v)); | ||
| 713 | |||
| 714 | virtio_cread(vdev, struct virtio_blk_config, | ||
| 715 | zoned.max_active_zones, &v); | ||
| 716 | disk_set_max_active_zones(vblk->disk, le32_to_cpu(v)); | ||
| 717 | dev_dbg(&vdev->dev, "max active zones = %u\n", le32_to_cpu(v)); | ||
| 718 | |||
| 719 | virtio_cread(vdev, struct virtio_blk_config, | ||
| 720 | zoned.write_granularity, &v); | ||
| 721 | if (!v) { | ||
| 722 | dev_warn(&vdev->dev, "zero write granularity reported\n"); | ||
| 723 | return -ENODEV; | ||
| 724 | } | ||
| 725 | blk_queue_physical_block_size(q, le32_to_cpu(v)); | ||
| 726 | blk_queue_io_min(q, le32_to_cpu(v)); | ||
| 727 | |||
| 728 | dev_dbg(&vdev->dev, "write granularity = %u\n", le32_to_cpu(v)); | ||
| 729 | |||
| 730 | /* | ||
| 731 | * virtio ZBD specification doesn't require zones to be a power of | ||
| 732 | * two sectors in size, but the code in this driver expects that. | ||
| 733 | */ | ||
| 734 | virtio_cread(vdev, struct virtio_blk_config, zoned.zone_sectors, &v); | ||
| 735 | vblk->zone_sectors = le32_to_cpu(v); | ||
| 736 | if (vblk->zone_sectors == 0 || !is_power_of_2(vblk->zone_sectors)) { | ||
| 737 | dev_err(&vdev->dev, | ||
| 738 | "zoned device with non power of two zone size %u\n", | ||
| 739 | vblk->zone_sectors); | ||
| 740 | return -ENODEV; | ||
| 741 | } | ||
| 742 | dev_dbg(&vdev->dev, "zone sectors = %u\n", vblk->zone_sectors); | ||
| 743 | |||
| 744 | if (virtio_has_feature(vdev, VIRTIO_BLK_F_DISCARD)) { | ||
| 745 | dev_warn(&vblk->vdev->dev, | ||
| 746 | "ignoring negotiated F_DISCARD for zoned device\n"); | ||
| 747 | blk_queue_max_discard_sectors(q, 0); | ||
| 748 | } | ||
| 749 | |||
| 750 | ret = blk_revalidate_disk_zones(vblk->disk, NULL); | ||
| 751 | if (!ret) { | ||
| 752 | virtio_cread(vdev, struct virtio_blk_config, | ||
| 753 | zoned.max_append_sectors, &v); | ||
| 754 | if (!v) { | ||
| 755 | dev_warn(&vdev->dev, "zero max_append_sectors reported\n"); | ||
| 756 | return -ENODEV; | ||
| 757 | } | ||
| 758 | blk_queue_max_zone_append_sectors(q, le32_to_cpu(v)); | ||
| 759 | dev_dbg(&vdev->dev, "max append sectors = %u\n", le32_to_cpu(v)); | ||
| 760 | } | ||
| 761 | |||
| 762 | return ret; | ||
| 763 | } | ||
| 764 | |||
| 765 | static inline bool virtblk_has_zoned_feature(struct virtio_device *vdev) | ||
| 766 | { | ||
| 767 | return virtio_has_feature(vdev, VIRTIO_BLK_F_ZONED); | ||
| 768 | } | ||
| 769 | #else | ||
| 770 | |||
| 771 | /* | ||
| 772 | * Zoned block device support is not configured in this kernel. | ||
| 773 | * We only need to define a few symbols to avoid compilation errors. | ||
| 774 | */ | ||
| 775 | #define virtblk_report_zones NULL | ||
| 776 | static inline void virtblk_revalidate_zones(struct virtio_blk *vblk) | ||
| 777 | { | ||
| 778 | } | ||
| 779 | static inline int virtblk_probe_zoned_device(struct virtio_device *vdev, | ||
| 780 | struct virtio_blk *vblk, struct request_queue *q) | ||
| 781 | { | ||
| 782 | return -EOPNOTSUPP; | ||
| 783 | } | ||
| 784 | |||
| 785 | static inline bool virtblk_has_zoned_feature(struct virtio_device *vdev) | ||
| 786 | { | ||
| 787 | return false; | ||
| 788 | } | ||
| 789 | #endif /* CONFIG_BLK_DEV_ZONED */ | ||
| 790 | |||
| 460 | /* return id (s/n) string for *disk to *id_str | 791 | /* return id (s/n) string for *disk to *id_str |
| 461 | */ | 792 | */ |
| 462 | static int virtblk_get_id(struct gendisk *disk, char *id_str) | 793 | static int virtblk_get_id(struct gendisk *disk, char *id_str) |
| @@ -464,18 +795,24 @@ static int virtblk_get_id(struct gendisk *disk, char *id_str) · 1 unmodified line | |||
| 464 | struct virtio_blk *vblk = disk->private_data; | 795 | struct virtio_blk *vblk = disk->private_data; |
| 465 | struct request_queue *q = vblk->disk->queue; | 796 | struct request_queue *q = vblk->disk->queue; |
| 466 | struct request *req; | 797 | struct request *req; |
| 798 | struct virtblk_req *vbr; | ||
| 467 | int err; | 799 | int err; |
| 468 | 800 | ||
| 469 | req = blk_mq_alloc_request(q, REQ_OP_DRV_IN, 0); | 801 | req = blk_mq_alloc_request(q, REQ_OP_DRV_IN, 0); |
| 470 | if (IS_ERR(req)) | 802 | if (IS_ERR(req)) |
| 471 | return PTR_ERR(req); | 803 | return PTR_ERR(req); |
| 472 | 804 | ||
| 805 | vbr = blk_mq_rq_to_pdu(req); | ||
| 806 | vbr->in_hdr_len = sizeof(vbr->status); | ||
| 807 | vbr->out_hdr.type = cpu_to_virtio32(vblk->vdev, VIRTIO_BLK_T_GET_ID); | ||
| 808 | vbr->out_hdr.sector = 0; | ||
| 809 | |||
| 473 | err = blk_rq_map_kern(q, req, id_str, VIRTIO_BLK_ID_BYTES, GFP_KERNEL); | 810 | err = blk_rq_map_kern(q, req, id_str, VIRTIO_BLK_ID_BYTES, GFP_KERNEL); |
| 474 | if (err) | 811 | if (err) |
| 475 | goto out; | 812 | goto out; |
| 476 | 813 | ||
| 477 | blk_execute_rq(req, false); | 814 | blk_execute_rq(req, false); |
| 478 | err = blk_status_to_errno(virtblk_result(blk_mq_rq_to_pdu(req))); | 815 | err = blk_status_to_errno(virtblk_result(vbr->status)); |
| 479 | out: | 816 | out: |
| 480 | blk_mq_free_request(req); | 817 | blk_mq_free_request(req); |
| 481 | return err; | 818 | return err; |
| @@ -526,6 +863,7 @@ static const struct block_device_operations virtblk_fops = { · 44 unmodified lines | |||
| 526 | .owner = THIS_MODULE, | 863 | .owner = THIS_MODULE, |
| 527 | .getgeo = virtblk_getgeo, | 864 | .getgeo = virtblk_getgeo, |
| 528 | .free_disk = virtblk_free_disk, | 865 | .free_disk = virtblk_free_disk, |
| 866 | .report_zones = virtblk_report_zones, | ||
| 529 | }; | 867 | }; |
| 530 | 868 | ||
| 531 | static int index_to_minor(int index) | 869 | static int index_to_minor(int index) |
| @@ -596,6 +934,7 @@ static void virtblk_config_changed_work(struct work_struct *work) · 64 unmodified lines | |||
| 596 | struct virtio_blk *vblk = | 934 | struct virtio_blk *vblk = |
| 597 | container_of(work, struct virtio_blk, config_work); | 935 | container_of(work, struct virtio_blk, config_work); |
| 598 | 936 | ||
| 937 | virtblk_revalidate_zones(vblk); | ||
| 599 | virtblk_update_capacity(vblk, true); | 938 | virtblk_update_capacity(vblk, true); |
| 600 | } | 939 | } |
| 601 | 940 | ||
| @@ -1152,6 +1491,15 @@ static int virtblk_probe(struct virtio_device *vdev) · 550 unmodified lines | |||
| 1152 | virtblk_update_capacity(vblk, false); | 1491 | virtblk_update_capacity(vblk, false); |
| 1153 | virtio_device_ready(vdev); | 1492 | virtio_device_ready(vdev); |
| 1154 | 1493 | ||
| 1494 | if (virtblk_has_zoned_feature(vdev)) { | ||
| 1495 | err = virtblk_probe_zoned_device(vdev, vblk, q); | ||
| 1496 | if (err) | ||
| 1497 | goto out_cleanup_disk; | ||
| 1498 | } | ||
| 1499 | |||
| 1500 | dev_info(&vdev->dev, "blk config size: %zu\n", | ||
| 1501 | sizeof(struct virtio_blk_config)); | ||
| 1502 | |||
| 1155 | err = device_add_disk(&vdev->dev, vblk->disk, virtblk_attr_groups); | 1503 | err = device_add_disk(&vdev->dev, vblk->disk, virtblk_attr_groups); |
| 1156 | if (err) | 1504 | if (err) |
| 1157 | goto out_cleanup_disk; | 1505 | goto out_cleanup_disk; |
| @@ -1253,6 +1601,9 @@ static unsigned int features[] = { · 95 unmodified lines | |||
| 1253 | VIRTIO_BLK_F_FLUSH, VIRTIO_BLK_F_TOPOLOGY, VIRTIO_BLK_F_CONFIG_WCE, | 1601 | VIRTIO_BLK_F_FLUSH, VIRTIO_BLK_F_TOPOLOGY, VIRTIO_BLK_F_CONFIG_WCE, |
| 1254 | VIRTIO_BLK_F_MQ, VIRTIO_BLK_F_DISCARD, VIRTIO_BLK_F_WRITE_ZEROES, | 1602 | VIRTIO_BLK_F_MQ, VIRTIO_BLK_F_DISCARD, VIRTIO_BLK_F_WRITE_ZEROES, |
| 1255 | VIRTIO_BLK_F_SECURE_ERASE, | 1603 | VIRTIO_BLK_F_SECURE_ERASE, |
| 1604 | #ifdef CONFIG_BLK_DEV_ZONED | ||
| 1605 | VIRTIO_BLK_F_ZONED, | ||
| 1606 | #endif /* CONFIG_BLK_DEV_ZONED */ | ||
| 1256 | }; | 1607 | }; |
| 1257 | 1608 | ||
| 1258 | static struct virtio_driver virtio_blk = { | 1609 | static struct virtio_driver virtio_blk = { |
| @@ -41,6 +41,7 @@ · 40 unmodified lines | |||
| 41 | #define VIRTIO_BLK_F_DISCARD 13 /* DISCARD is supported */ | 41 | #define VIRTIO_BLK_F_DISCARD 13 /* DISCARD is supported */ |
| 42 | #define VIRTIO_BLK_F_WRITE_ZEROES 14 /* WRITE ZEROES is supported */ | 42 | #define VIRTIO_BLK_F_WRITE_ZEROES 14 /* WRITE ZEROES is supported */ |
| 43 | #define VIRTIO_BLK_F_SECURE_ERASE 16 /* Secure Erase is supported */ | 43 | #define VIRTIO_BLK_F_SECURE_ERASE 16 /* Secure Erase is supported */ |
| 44 | #define VIRTIO_BLK_F_ZONED 17 /* Zoned block device */ | ||
| 44 | 45 | ||
| 45 | /* Legacy feature bits */ | 46 | /* Legacy feature bits */ |
| 46 | #ifndef VIRTIO_BLK_NO_LEGACY | 47 | #ifndef VIRTIO_BLK_NO_LEGACY |
| @@ -137,6 +138,16 @@ struct virtio_blk_config { · 90 unmodified lines | |||
| 137 | /* Secure erase commands must be aligned to this number of sectors. */ | 138 | /* Secure erase commands must be aligned to this number of sectors. */ |
| 138 | __virtio32 secure_erase_sector_alignment; | 139 | __virtio32 secure_erase_sector_alignment; |
| 139 | 140 | ||
| 141 | /* Zoned block device characteristics (if VIRTIO_BLK_F_ZONED) */ | ||
| 142 | struct virtio_blk_zoned_characteristics { | ||
| 143 | __virtio32 zone_sectors; | ||
| 144 | __virtio32 max_open_zones; | ||
| 145 | __virtio32 max_active_zones; | ||
| 146 | __virtio32 max_append_sectors; | ||
| 147 | __virtio32 write_granularity; | ||
| 148 | __u8 model; | ||
| 149 | __u8 unused2[3]; | ||
| 150 | } zoned; | ||
| 140 | } __attribute__((packed)); | 151 | } __attribute__((packed)); |
| 141 | 152 | ||
| 142 | /* | 153 | /* |
| @@ -174,6 +185,27 @@ struct virtio_blk_config { · 31 unmodified lines | |||
| 174 | /* Secure erase command */ | 185 | /* Secure erase command */ |
| 175 | #define VIRTIO_BLK_T_SECURE_ERASE 14 | 186 | #define VIRTIO_BLK_T_SECURE_ERASE 14 |
| 176 | 187 | ||
| 188 | /* Zone append command */ | ||
| 189 | #define VIRTIO_BLK_T_ZONE_APPEND 15 | ||
| 190 | |||
| 191 | /* Report zones command */ | ||
| 192 | #define VIRTIO_BLK_T_ZONE_REPORT 16 | ||
| 193 | |||
| 194 | /* Open zone command */ | ||
| 195 | #define VIRTIO_BLK_T_ZONE_OPEN 18 | ||
| 196 | |||
| 197 | /* Close zone command */ | ||
| 198 | #define VIRTIO_BLK_T_ZONE_CLOSE 20 | ||
| 199 | |||
| 200 | /* Finish zone command */ | ||
| 201 | #define VIRTIO_BLK_T_ZONE_FINISH 22 | ||
| 202 | |||
| 203 | /* Reset zone command */ | ||
| 204 | #define VIRTIO_BLK_T_ZONE_RESET 24 | ||
| 205 | |||
| 206 | /* Reset All zones command */ | ||
| 207 | #define VIRTIO_BLK_T_ZONE_RESET_ALL 26 | ||
| 208 | |||
| 177 | #ifndef VIRTIO_BLK_NO_LEGACY | 209 | #ifndef VIRTIO_BLK_NO_LEGACY |
| 178 | /* Barrier before this op. */ | 210 | /* Barrier before this op. */ |
| 179 | #define VIRTIO_BLK_T_BARRIER 0x80000000 | 211 | #define VIRTIO_BLK_T_BARRIER 0x80000000 |
| @@ -193,6 +225,72 @@ struct virtio_blk_outhdr { · 13 unmodified lines | |||
| 193 | __virtio64 sector; | 225 | __virtio64 sector; |
| 194 | }; | 226 | }; |
| 195 | 227 | ||
| 228 | /* | ||
| 229 | * Supported zoned device models. | ||
| 230 | */ | ||
| 231 | |||
| 232 | /* Regular block device */ | ||
| 233 | #define VIRTIO_BLK_Z_NONE 0 | ||
| 234 | /* Host-managed zoned device */ | ||
| 235 | #define VIRTIO_BLK_Z_HM 1 | ||
| 236 | /* Host-aware zoned device */ | ||
| 237 | #define VIRTIO_BLK_Z_HA 2 | ||
| 238 | |||
| 239 | /* | ||
| 240 | * Zone descriptor. A part of VIRTIO_BLK_T_ZONE_REPORT command reply. | ||
| 241 | */ | ||
| 242 | struct virtio_blk_zone_descriptor { | ||
| 243 | /* Zone capacity */ | ||
| 244 | __virtio64 z_cap; | ||
| 245 | /* The starting sector of the zone */ | ||
| 246 | __virtio64 z_start; | ||
| 247 | /* Zone write pointer position in sectors */ | ||
| 248 | __virtio64 z_wp; | ||
| 249 | /* Zone type */ | ||
| 250 | __u8 z_type; | ||
| 251 | /* Zone state */ | ||
| 252 | __u8 z_state; | ||
| 253 | __u8 reserved[38]; | ||
| 254 | }; | ||
| 255 | |||
| 256 | struct virtio_blk_zone_report { | ||
| 257 | __virtio64 nr_zones; | ||
| 258 | __u8 reserved[56]; | ||
| 259 | struct virtio_blk_zone_descriptor zones[]; | ||
| 260 | }; | ||
| 261 | |||
| 262 | /* | ||
| 263 | * Supported zone types. | ||
| 264 | */ | ||
| 265 | |||
| 266 | /* Conventional zone */ | ||
| 267 | #define VIRTIO_BLK_ZT_CONV 1 | ||
| 268 | /* Sequential Write Required zone */ | ||
| 269 | #define VIRTIO_BLK_ZT_SWR 2 | ||
| 270 | /* Sequential Write Preferred zone */ | ||
| 271 | #define VIRTIO_BLK_ZT_SWP 3 | ||
| 272 | |||
| 273 | /* | ||
| 274 | * Zone states that are available for zones of all types. | ||
| 275 | */ | ||
| 276 | |||
| 277 | /* Not a write pointer (conventional zones only) */ | ||
| 278 | #define VIRTIO_BLK_ZS_NOT_WP 0 | ||
| 279 | /* Empty */ | ||
| 280 | #define VIRTIO_BLK_ZS_EMPTY 1 | ||
| 281 | /* Implicitly Open */ | ||
| 282 | #define VIRTIO_BLK_ZS_IOPEN 2 | ||
| 283 | /* Explicitly Open */ | ||
| 284 | #define VIRTIO_BLK_ZS_EOPEN 3 | ||
| 285 | /* Closed */ | ||
| 286 | #define VIRTIO_BLK_ZS_CLOSED 4 | ||
| 287 | /* Read-Only */ | ||
| 288 | #define VIRTIO_BLK_ZS_RDONLY 13 | ||
| 289 | /* Full */ | ||
| 290 | #define VIRTIO_BLK_ZS_FULL 14 | ||
| 291 | /* Offline */ | ||
| 292 | #define VIRTIO_BLK_ZS_OFFLINE 15 | ||
| 293 | |||
| 196 | /* Unmap this range (only valid for write zeroes command) */ | 294 | /* Unmap this range (only valid for write zeroes command) */ |
| 197 | #define VIRTIO_BLK_WRITE_ZEROES_FLAG_UNMAP 0x00000001 | 295 | #define VIRTIO_BLK_WRITE_ZEROES_FLAG_UNMAP 0x00000001 |
| 198 | 296 | ||
| @@ -219,4 +317,11 @@ struct virtio_scsi_inhdr { · 20 unmodified lines | |||
| 219 | #define VIRTIO_BLK_S_OK 0 | 317 | #define VIRTIO_BLK_S_OK 0 |
| 220 | #define VIRTIO_BLK_S_IOERR 1 | 318 | #define VIRTIO_BLK_S_IOERR 1 |
| 221 | #define VIRTIO_BLK_S_UNSUPP 2 | 319 | #define VIRTIO_BLK_S_UNSUPP 2 |
| 320 | |||
| 321 | /* Error codes that are specific to zoned block devices */ | ||
| 322 | #define VIRTIO_BLK_S_ZONE_INVALID_CMD 3 | ||
| 323 | #define VIRTIO_BLK_S_ZONE_UNALIGNED_WP 4 | ||
| 324 | #define VIRTIO_BLK_S_ZONE_OPEN_RESOURCE 5 | ||
| 325 | #define VIRTIO_BLK_S_ZONE_ACTIVE_RESOURCE 6 | ||
| 326 | |||
| 222 | #endif /* _LINUX_VIRTIO_BLK_H */ | 327 | #endif /* _LINUX_VIRTIO_BLK_H */ |