| 1 | /* |
| 2 | * blk-integrity.c - Block layer data integrity extensions |
| 3 | * |
| 4 | * Copyright (C) 2007, 2008 Oracle Corporation |
| 5 | * Written by: Martin K. Petersen <martin.petersen@oracle.com> |
| 6 | * |
| 7 | * This program is free software; you can redistribute it and/or |
| 8 | * modify it under the terms of the GNU General Public License version |
| 9 | * 2 as published by the Free Software Foundation. |
| 10 | * |
| 11 | * This program is distributed in the hope that it will be useful, but |
| 12 | * WITHOUT ANY WARRANTY; without even the implied warranty of |
| 13 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU |
| 14 | * General Public License for more details. |
| 15 | * |
| 16 | * You should have received a copy of the GNU General Public License |
| 17 | * along with this program; see the file COPYING. If not, write to |
| 18 | * the Free Software Foundation, 675 Mass Ave, Cambridge, MA 02139, |
| 19 | * USA. |
| 20 | * |
| 21 | */ |
| 22 | |
| 23 | #include <linux/blkdev.h> |
| 24 | #include <linux/backing-dev.h> |
| 25 | #include <linux/mempool.h> |
| 26 | #include <linux/bio.h> |
| 27 | #include <linux/scatterlist.h> |
| 28 | #include <linux/export.h> |
| 29 | #include <linux/slab.h> |
| 30 | |
| 31 | #include "blk.h" |
| 32 | |
| 33 | /** |
| 34 | * blk_rq_count_integrity_sg - Count number of integrity scatterlist elements |
| 35 | * @q: request queue |
| 36 | * @bio: bio with integrity metadata attached |
| 37 | * |
| 38 | * Description: Returns the number of elements required in a |
| 39 | * scatterlist corresponding to the integrity metadata in a bio. |
| 40 | */ |
| 41 | int blk_rq_count_integrity_sg(struct request_queue *q, struct bio *bio) |
| 42 | { |
| 43 | struct bio_vec iv, ivprv = { NULL }; |
| 44 | unsigned int segments = 0; |
| 45 | unsigned int seg_size = 0; |
| 46 | struct bvec_iter iter; |
| 47 | int prev = 0; |
| 48 | |
| 49 | bio_for_each_integrity_vec(iv, bio, iter) { |
| 50 | |
| 51 | if (prev) { |
| 52 | if (!BIOVEC_PHYS_MERGEABLE(&ivprv, &iv)) |
| 53 | goto new_segment; |
| 54 | |
| 55 | if (!BIOVEC_SEG_BOUNDARY(q, &ivprv, &iv)) |
| 56 | goto new_segment; |
| 57 | |
| 58 | if (seg_size + iv.bv_len > queue_max_segment_size(q)) |
| 59 | goto new_segment; |
| 60 | |
| 61 | seg_size += iv.bv_len; |
| 62 | } else { |
| 63 | new_segment: |
| 64 | segments++; |
| 65 | seg_size = iv.bv_len; |
| 66 | } |
| 67 | |
| 68 | prev = 1; |
| 69 | ivprv = iv; |
| 70 | } |
| 71 | |
| 72 | return segments; |
| 73 | } |
| 74 | EXPORT_SYMBOL(blk_rq_count_integrity_sg); |
| 75 | |
| 76 | /** |
| 77 | * blk_rq_map_integrity_sg - Map integrity metadata into a scatterlist |
| 78 | * @q: request queue |
| 79 | * @bio: bio with integrity metadata attached |
| 80 | * @sglist: target scatterlist |
| 81 | * |
| 82 | * Description: Map the integrity vectors in request into a |
| 83 | * scatterlist. The scatterlist must be big enough to hold all |
| 84 | * elements. I.e. sized using blk_rq_count_integrity_sg(). |
| 85 | */ |
| 86 | int blk_rq_map_integrity_sg(struct request_queue *q, struct bio *bio, |
| 87 | struct scatterlist *sglist) |
| 88 | { |
| 89 | struct bio_vec iv, ivprv = { NULL }; |
| 90 | struct scatterlist *sg = NULL; |
| 91 | unsigned int segments = 0; |
| 92 | struct bvec_iter iter; |
| 93 | int prev = 0; |
| 94 | |
| 95 | bio_for_each_integrity_vec(iv, bio, iter) { |
| 96 | |
| 97 | if (prev) { |
| 98 | if (!BIOVEC_PHYS_MERGEABLE(&ivprv, &iv)) |
| 99 | goto new_segment; |
| 100 | |
| 101 | if (!BIOVEC_SEG_BOUNDARY(q, &ivprv, &iv)) |
| 102 | goto new_segment; |
| 103 | |
| 104 | if (sg->length + iv.bv_len > queue_max_segment_size(q)) |
| 105 | goto new_segment; |
| 106 | |
| 107 | sg->length += iv.bv_len; |
| 108 | } else { |
| 109 | new_segment: |
| 110 | if (!sg) |
| 111 | sg = sglist; |
| 112 | else { |
| 113 | sg_unmark_end(sg); |
| 114 | sg = sg_next(sg); |
| 115 | } |
| 116 | |
| 117 | sg_set_page(sg, iv.bv_page, iv.bv_len, iv.bv_offset); |
| 118 | segments++; |
| 119 | } |
| 120 | |
| 121 | prev = 1; |
| 122 | ivprv = iv; |
| 123 | } |
| 124 | |
| 125 | if (sg) |
| 126 | sg_mark_end(sg); |
| 127 | |
| 128 | return segments; |
| 129 | } |
| 130 | EXPORT_SYMBOL(blk_rq_map_integrity_sg); |
| 131 | |
| 132 | /** |
| 133 | * blk_integrity_compare - Compare integrity profile of two disks |
| 134 | * @gd1: Disk to compare |
| 135 | * @gd2: Disk to compare |
| 136 | * |
| 137 | * Description: Meta-devices like DM and MD need to verify that all |
| 138 | * sub-devices use the same integrity format before advertising to |
| 139 | * upper layers that they can send/receive integrity metadata. This |
| 140 | * function can be used to check whether two gendisk devices have |
| 141 | * compatible integrity formats. |
| 142 | */ |
| 143 | int blk_integrity_compare(struct gendisk *gd1, struct gendisk *gd2) |
| 144 | { |
| 145 | struct blk_integrity *b1 = &gd1->queue->integrity; |
| 146 | struct blk_integrity *b2 = &gd2->queue->integrity; |
| 147 | |
| 148 | if (!b1->profile && !b2->profile) |
| 149 | return 0; |
| 150 | |
| 151 | if (!b1->profile || !b2->profile) |
| 152 | return -1; |
| 153 | |
| 154 | if (b1->interval_exp != b2->interval_exp) { |
| 155 | pr_err("%s: %s/%s protection interval %u != %u\n", |
| 156 | __func__, gd1->disk_name, gd2->disk_name, |
| 157 | 1 << b1->interval_exp, 1 << b2->interval_exp); |
| 158 | return -1; |
| 159 | } |
| 160 | |
| 161 | if (b1->tuple_size != b2->tuple_size) { |
| 162 | pr_err("%s: %s/%s tuple sz %u != %u\n", __func__, |
| 163 | gd1->disk_name, gd2->disk_name, |
| 164 | b1->tuple_size, b2->tuple_size); |
| 165 | return -1; |
| 166 | } |
| 167 | |
| 168 | if (b1->tag_size && b2->tag_size && (b1->tag_size != b2->tag_size)) { |
| 169 | pr_err("%s: %s/%s tag sz %u != %u\n", __func__, |
| 170 | gd1->disk_name, gd2->disk_name, |
| 171 | b1->tag_size, b2->tag_size); |
| 172 | return -1; |
| 173 | } |
| 174 | |
| 175 | if (b1->profile != b2->profile) { |
| 176 | pr_err("%s: %s/%s type %s != %s\n", __func__, |
| 177 | gd1->disk_name, gd2->disk_name, |
| 178 | b1->profile->name, b2->profile->name); |
| 179 | return -1; |
| 180 | } |
| 181 | |
| 182 | return 0; |
| 183 | } |
| 184 | EXPORT_SYMBOL(blk_integrity_compare); |
| 185 | |
| 186 | bool blk_integrity_merge_rq(struct request_queue *q, struct request *req, |
| 187 | struct request *next) |
| 188 | { |
| 189 | if (blk_integrity_rq(req) == 0 && blk_integrity_rq(next) == 0) |
| 190 | return true; |
| 191 | |
| 192 | if (blk_integrity_rq(req) == 0 || blk_integrity_rq(next) == 0) |
| 193 | return false; |
| 194 | |
| 195 | if (bio_integrity(req->bio)->bip_flags != |
| 196 | bio_integrity(next->bio)->bip_flags) |
| 197 | return false; |
| 198 | |
| 199 | if (req->nr_integrity_segments + next->nr_integrity_segments > |
| 200 | q->limits.max_integrity_segments) |
| 201 | return false; |
| 202 | |
| 203 | if (integrity_req_gap_back_merge(req, next->bio)) |
| 204 | return false; |
| 205 | |
| 206 | return true; |
| 207 | } |
| 208 | EXPORT_SYMBOL(blk_integrity_merge_rq); |
| 209 | |
| 210 | bool blk_integrity_merge_bio(struct request_queue *q, struct request *req, |
| 211 | struct bio *bio) |
| 212 | { |
| 213 | int nr_integrity_segs; |
| 214 | struct bio *next = bio->bi_next; |
| 215 | |
| 216 | if (blk_integrity_rq(req) == 0 && bio_integrity(bio) == NULL) |
| 217 | return true; |
| 218 | |
| 219 | if (blk_integrity_rq(req) == 0 || bio_integrity(bio) == NULL) |
| 220 | return false; |
| 221 | |
| 222 | if (bio_integrity(req->bio)->bip_flags != bio_integrity(bio)->bip_flags) |
| 223 | return false; |
| 224 | |
| 225 | bio->bi_next = NULL; |
| 226 | nr_integrity_segs = blk_rq_count_integrity_sg(q, bio); |
| 227 | bio->bi_next = next; |
| 228 | |
| 229 | if (req->nr_integrity_segments + nr_integrity_segs > |
| 230 | q->limits.max_integrity_segments) |
| 231 | return false; |
| 232 | |
| 233 | req->nr_integrity_segments += nr_integrity_segs; |
| 234 | |
| 235 | return true; |
| 236 | } |
| 237 | EXPORT_SYMBOL(blk_integrity_merge_bio); |
| 238 | |
| 239 | struct integrity_sysfs_entry { |
| 240 | struct attribute attr; |
| 241 | ssize_t (*show)(struct blk_integrity *, char *); |
| 242 | ssize_t (*store)(struct blk_integrity *, const char *, size_t); |
| 243 | }; |
| 244 | |
| 245 | static ssize_t integrity_attr_show(struct kobject *kobj, struct attribute *attr, |
| 246 | char *page) |
| 247 | { |
| 248 | struct gendisk *disk = container_of(kobj, struct gendisk, integrity_kobj); |
| 249 | struct blk_integrity *bi = &disk->queue->integrity; |
| 250 | struct integrity_sysfs_entry *entry = |
| 251 | container_of(attr, struct integrity_sysfs_entry, attr); |
| 252 | |
| 253 | return entry->show(bi, page); |
| 254 | } |
| 255 | |
| 256 | static ssize_t integrity_attr_store(struct kobject *kobj, |
| 257 | struct attribute *attr, const char *page, |
| 258 | size_t count) |
| 259 | { |
| 260 | struct gendisk *disk = container_of(kobj, struct gendisk, integrity_kobj); |
| 261 | struct blk_integrity *bi = &disk->queue->integrity; |
| 262 | struct integrity_sysfs_entry *entry = |
| 263 | container_of(attr, struct integrity_sysfs_entry, attr); |
| 264 | ssize_t ret = 0; |
| 265 | |
| 266 | if (entry->store) |
| 267 | ret = entry->store(bi, page, count); |
| 268 | |
| 269 | return ret; |
| 270 | } |
| 271 | |
| 272 | static ssize_t integrity_format_show(struct blk_integrity *bi, char *page) |
| 273 | { |
| 274 | if (bi->profile && bi->profile->name) |
| 275 | return sprintf(page, "%s\n", bi->profile->name); |
| 276 | else |
| 277 | return sprintf(page, "none\n"); |
| 278 | } |
| 279 | |
| 280 | static ssize_t integrity_tag_size_show(struct blk_integrity *bi, char *page) |
| 281 | { |
| 282 | return sprintf(page, "%u\n", bi->tag_size); |
| 283 | } |
| 284 | |
| 285 | static ssize_t integrity_interval_show(struct blk_integrity *bi, char *page) |
| 286 | { |
| 287 | return sprintf(page, "%u\n", |
| 288 | bi->interval_exp ? 1 << bi->interval_exp : 0); |
| 289 | } |
| 290 | |
| 291 | static ssize_t integrity_verify_store(struct blk_integrity *bi, |
| 292 | const char *page, size_t count) |
| 293 | { |
| 294 | char *p = (char *) page; |
| 295 | unsigned long val = simple_strtoul(p, &p, 10); |
| 296 | |
| 297 | if (val) |
| 298 | bi->flags |= BLK_INTEGRITY_VERIFY; |
| 299 | else |
| 300 | bi->flags &= ~BLK_INTEGRITY_VERIFY; |
| 301 | |
| 302 | return count; |
| 303 | } |
| 304 | |
| 305 | static ssize_t integrity_verify_show(struct blk_integrity *bi, char *page) |
| 306 | { |
| 307 | return sprintf(page, "%d\n", (bi->flags & BLK_INTEGRITY_VERIFY) != 0); |
| 308 | } |
| 309 | |
| 310 | static ssize_t integrity_generate_store(struct blk_integrity *bi, |
| 311 | const char *page, size_t count) |
| 312 | { |
| 313 | char *p = (char *) page; |
| 314 | unsigned long val = simple_strtoul(p, &p, 10); |
| 315 | |
| 316 | if (val) |
| 317 | bi->flags |= BLK_INTEGRITY_GENERATE; |
| 318 | else |
| 319 | bi->flags &= ~BLK_INTEGRITY_GENERATE; |
| 320 | |
| 321 | return count; |
| 322 | } |
| 323 | |
| 324 | static ssize_t integrity_generate_show(struct blk_integrity *bi, char *page) |
| 325 | { |
| 326 | return sprintf(page, "%d\n", (bi->flags & BLK_INTEGRITY_GENERATE) != 0); |
| 327 | } |
| 328 | |
| 329 | static ssize_t integrity_device_show(struct blk_integrity *bi, char *page) |
| 330 | { |
| 331 | return sprintf(page, "%u\n", |
| 332 | (bi->flags & BLK_INTEGRITY_DEVICE_CAPABLE) != 0); |
| 333 | } |
| 334 | |
| 335 | static struct integrity_sysfs_entry integrity_format_entry = { |
| 336 | .attr = { .name = "format", .mode = S_IRUGO }, |
| 337 | .show = integrity_format_show, |
| 338 | }; |
| 339 | |
| 340 | static struct integrity_sysfs_entry integrity_tag_size_entry = { |
| 341 | .attr = { .name = "tag_size", .mode = S_IRUGO }, |
| 342 | .show = integrity_tag_size_show, |
| 343 | }; |
| 344 | |
| 345 | static struct integrity_sysfs_entry integrity_interval_entry = { |
| 346 | .attr = { .name = "protection_interval_bytes", .mode = S_IRUGO }, |
| 347 | .show = integrity_interval_show, |
| 348 | }; |
| 349 | |
| 350 | static struct integrity_sysfs_entry integrity_verify_entry = { |
| 351 | .attr = { .name = "read_verify", .mode = S_IRUGO | S_IWUSR }, |
| 352 | .show = integrity_verify_show, |
| 353 | .store = integrity_verify_store, |
| 354 | }; |
| 355 | |
| 356 | static struct integrity_sysfs_entry integrity_generate_entry = { |
| 357 | .attr = { .name = "write_generate", .mode = S_IRUGO | S_IWUSR }, |
| 358 | .show = integrity_generate_show, |
| 359 | .store = integrity_generate_store, |
| 360 | }; |
| 361 | |
| 362 | static struct integrity_sysfs_entry integrity_device_entry = { |
| 363 | .attr = { .name = "device_is_integrity_capable", .mode = S_IRUGO }, |
| 364 | .show = integrity_device_show, |
| 365 | }; |
| 366 | |
| 367 | static struct attribute *integrity_attrs[] = { |
| 368 | &integrity_format_entry.attr, |
| 369 | &integrity_tag_size_entry.attr, |
| 370 | &integrity_interval_entry.attr, |
| 371 | &integrity_verify_entry.attr, |
| 372 | &integrity_generate_entry.attr, |
| 373 | &integrity_device_entry.attr, |
| 374 | NULL, |
| 375 | }; |
| 376 | |
| 377 | static const struct sysfs_ops integrity_ops = { |
| 378 | .show = &integrity_attr_show, |
| 379 | .store = &integrity_attr_store, |
| 380 | }; |
| 381 | |
| 382 | static struct kobj_type integrity_ktype = { |
| 383 | .default_attrs = integrity_attrs, |
| 384 | .sysfs_ops = &integrity_ops, |
| 385 | }; |
| 386 | |
| 387 | static int blk_integrity_nop_fn(struct blk_integrity_iter *iter) |
| 388 | { |
| 389 | return 0; |
| 390 | } |
| 391 | |
| 392 | static struct blk_integrity_profile nop_profile = { |
| 393 | .name = "nop", |
| 394 | .generate_fn = blk_integrity_nop_fn, |
| 395 | .verify_fn = blk_integrity_nop_fn, |
| 396 | }; |
| 397 | |
| 398 | /** |
| 399 | * blk_integrity_register - Register a gendisk as being integrity-capable |
| 400 | * @disk: struct gendisk pointer to make integrity-aware |
| 401 | * @template: block integrity profile to register |
| 402 | * |
| 403 | * Description: When a device needs to advertise itself as being able to |
| 404 | * send/receive integrity metadata it must use this function to register |
| 405 | * the capability with the block layer. The template is a blk_integrity |
| 406 | * struct with values appropriate for the underlying hardware. See |
| 407 | * Documentation/block/data-integrity.txt. |
| 408 | */ |
| 409 | void blk_integrity_register(struct gendisk *disk, struct blk_integrity *template) |
| 410 | { |
| 411 | struct blk_integrity *bi = &disk->queue->integrity; |
| 412 | |
| 413 | bi->flags = BLK_INTEGRITY_VERIFY | BLK_INTEGRITY_GENERATE | |
| 414 | template->flags; |
| 415 | bi->interval_exp = ilog2(queue_logical_block_size(disk->queue)); |
| 416 | bi->profile = template->profile ? template->profile : &nop_profile; |
| 417 | bi->tuple_size = template->tuple_size; |
| 418 | bi->tag_size = template->tag_size; |
| 419 | |
| 420 | blk_integrity_revalidate(disk); |
| 421 | } |
| 422 | EXPORT_SYMBOL(blk_integrity_register); |
| 423 | |
| 424 | /** |
| 425 | * blk_integrity_unregister - Unregister block integrity profile |
| 426 | * @disk: disk whose integrity profile to unregister |
| 427 | * |
| 428 | * Description: This function unregisters the integrity capability from |
| 429 | * a block device. |
| 430 | */ |
| 431 | void blk_integrity_unregister(struct gendisk *disk) |
| 432 | { |
| 433 | blk_integrity_revalidate(disk); |
| 434 | memset(&disk->queue->integrity, 0, sizeof(struct blk_integrity)); |
| 435 | } |
| 436 | EXPORT_SYMBOL(blk_integrity_unregister); |
| 437 | |
| 438 | void blk_integrity_revalidate(struct gendisk *disk) |
| 439 | { |
| 440 | struct blk_integrity *bi = &disk->queue->integrity; |
| 441 | |
| 442 | if (!(disk->flags & GENHD_FL_UP)) |
| 443 | return; |
| 444 | |
| 445 | if (bi->profile) |
| 446 | disk->queue->backing_dev_info.capabilities |= |
| 447 | BDI_CAP_STABLE_WRITES; |
| 448 | else |
| 449 | disk->queue->backing_dev_info.capabilities &= |
| 450 | ~BDI_CAP_STABLE_WRITES; |
| 451 | } |
| 452 | |
| 453 | void blk_integrity_add(struct gendisk *disk) |
| 454 | { |
| 455 | if (kobject_init_and_add(&disk->integrity_kobj, &integrity_ktype, |
| 456 | &disk_to_dev(disk)->kobj, "%s", "integrity")) |
| 457 | return; |
| 458 | |
| 459 | kobject_uevent(&disk->integrity_kobj, KOBJ_ADD); |
| 460 | } |
| 461 | |
| 462 | void blk_integrity_del(struct gendisk *disk) |
| 463 | { |
| 464 | kobject_uevent(&disk->integrity_kobj, KOBJ_REMOVE); |
| 465 | kobject_del(&disk->integrity_kobj); |
| 466 | kobject_put(&disk->integrity_kobj); |
| 467 | } |