block/blk-integrity.c

   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/mempool.h>
  25 #include <linux/bio.h>
  26 #include <linux/scatterlist.h>
  27 #include <linux/export.h>
  28 #include <linux/slab.h>
  29
  30 #include "blk.h"
  31
  32 static struct kmem_cache *integrity_cachep;
  33
  34 static const char *bi_unsupported_name = "unsupported";
  35
  36 /**
  37  * blk_rq_count_integrity_sg - Count number of integrity scatterlist elements
  38  * @q:          request queue
  39  * @bio:        bio with integrity metadata attached
  40  *
  41  * Description: Returns the number of elements required in a
  42  * scatterlist corresponding to the integrity metadata in a bio.
  43  */
  44 int blk_rq_count_integrity_sg(struct request_queue *q, struct bio *bio)
  45 {
  46         struct bio_vec *iv, *ivprv = NULL;
  47         unsigned int segments = 0;
  48         unsigned int seg_size = 0;
  49         unsigned int i = 0;
  50
  51         bio_for_each_integrity_vec(iv, bio, i) {
  52
  53                 if (ivprv) {
  54                         if (!BIOVEC_PHYS_MERGEABLE(ivprv, iv))
  55                                 goto new_segment;
  56
  57                         if (!BIOVEC_SEG_BOUNDARY(q, ivprv, iv))
  58                                 goto new_segment;
  59
  60                         if (seg_size + iv->bv_len > queue_max_segment_size(q))
  61                                 goto new_segment;
  62
  63                         seg_size += iv->bv_len;
  64                 } else {
  65 new_segment:
  66                         segments++;
  67                         seg_size = iv->bv_len;
  68                 }
  69
  70                 ivprv = iv;
  71         }
  72
  73         return segments;
  74 }
  75 EXPORT_SYMBOL(blk_rq_count_integrity_sg);
  76
  77 /**
  78  * blk_rq_map_integrity_sg - Map integrity metadata into a scatterlist
  79  * @q:          request queue
  80  * @bio:        bio with integrity metadata attached
  81  * @sglist:     target scatterlist
  82  *
  83  * Description: Map the integrity vectors in request into a
  84  * scatterlist.  The scatterlist must be big enough to hold all
  85  * elements.  I.e. sized using blk_rq_count_integrity_sg().
  86  */
  87 int blk_rq_map_integrity_sg(struct request_queue *q, struct bio *bio,
  88                             struct scatterlist *sglist)
  89 {
  90         struct bio_vec *iv, *ivprv = NULL;
  91         struct scatterlist *sg = NULL;
  92         unsigned int segments = 0;
  93         unsigned int i = 0;
  94
  95         bio_for_each_integrity_vec(iv, bio, i) {
  96
  97                 if (ivprv) {
  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->page_link &= ~0x02;
 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                 ivprv = iv;
 122         }
 123
 124         if (sg)
 125                 sg_mark_end(sg);
 126
 127         return segments;
 128 }
 129 EXPORT_SYMBOL(blk_rq_map_integrity_sg);
 130
 131 /**
 132  * blk_integrity_compare - Compare integrity profile of two disks
 133  * @gd1:        Disk to compare
 134  * @gd2:        Disk to compare
 135  *
 136  * Description: Meta-devices like DM and MD need to verify that all
 137  * sub-devices use the same integrity format before advertising to
 138  * upper layers that they can send/receive integrity metadata.  This
 139  * function can be used to check whether two gendisk devices have
 140  * compatible integrity formats.
 141  */
 142 int blk_integrity_compare(struct gendisk *gd1, struct gendisk *gd2)
 143 {
 144         struct blk_integrity *b1 = gd1->integrity;
 145         struct blk_integrity *b2 = gd2->integrity;
 146
 147         if (!b1 && !b2)
 148                 return 0;
 149
 150         if (!b1 || !b2)
 151                 return -1;
 152
 153         if (b1->sector_size != b2->sector_size) {
 154                 printk(KERN_ERR "%s: %s/%s sector sz %u != %u\n", __func__,
 155                        gd1->disk_name, gd2->disk_name,
 156                        b1->sector_size, b2->sector_size);
 157                 return -1;
 158         }
 159
 160         if (b1->tuple_size != b2->tuple_size) {
 161                 printk(KERN_ERR "%s: %s/%s tuple sz %u != %u\n", __func__,
 162                        gd1->disk_name, gd2->disk_name,
 163                        b1->tuple_size, b2->tuple_size);
 164                 return -1;
 165         }
 166
 167         if (b1->tag_size && b2->tag_size && (b1->tag_size != b2->tag_size)) {
 168                 printk(KERN_ERR "%s: %s/%s tag sz %u != %u\n", __func__,
 169                        gd1->disk_name, gd2->disk_name,
 170                        b1->tag_size, b2->tag_size);
 171                 return -1;
 172         }
 173
 174         if (strcmp(b1->name, b2->name)) {
 175                 printk(KERN_ERR "%s: %s/%s type %s != %s\n", __func__,
 176                        gd1->disk_name, gd2->disk_name,
 177                        b1->name, b2->name);
 178                 return -1;
 179         }
 180
 181         return 0;
 182 }
 183 EXPORT_SYMBOL(blk_integrity_compare);
 184
 185 int blk_integrity_merge_rq(struct request_queue *q, struct request *req,
 186                            struct request *next)
 187 {
 188         if (blk_integrity_rq(req) != blk_integrity_rq(next))
 189                 return -1;
 190
 191         if (req->nr_integrity_segments + next->nr_integrity_segments >
 192             q->limits.max_integrity_segments)
 193                 return -1;
 194
 195         return 0;
 196 }
 197 EXPORT_SYMBOL(blk_integrity_merge_rq);
 198
 199 int blk_integrity_merge_bio(struct request_queue *q, struct request *req,
 200                             struct bio *bio)
 201 {
 202         int nr_integrity_segs;
 203         struct bio *next = bio->bi_next;
 204
 205         bio->bi_next = NULL;
 206         nr_integrity_segs = blk_rq_count_integrity_sg(q, bio);
 207         bio->bi_next = next;
 208
 209         if (req->nr_integrity_segments + nr_integrity_segs >
 210             q->limits.max_integrity_segments)
 211                 return -1;
 212
 213         req->nr_integrity_segments += nr_integrity_segs;
 214
 215         return 0;
 216 }
 217 EXPORT_SYMBOL(blk_integrity_merge_bio);
 218
 219 struct integrity_sysfs_entry {
 220         struct attribute attr;
 221         ssize_t (*show)(struct blk_integrity *, char *);
 222         ssize_t (*store)(struct blk_integrity *, const char *, size_t);
 223 };
 224
 225 static ssize_t integrity_attr_show(struct kobject *kobj, struct attribute *attr,
 226                                    char *page)
 227 {
 228         struct blk_integrity *bi =
 229                 container_of(kobj, struct blk_integrity, kobj);
 230         struct integrity_sysfs_entry *entry =
 231                 container_of(attr, struct integrity_sysfs_entry, attr);
 232
 233         return entry->show(bi, page);
 234 }
 235
 236 static ssize_t integrity_attr_store(struct kobject *kobj,
 237                                     struct attribute *attr, const char *page,
 238                                     size_t count)
 239 {
 240         struct blk_integrity *bi =
 241                 container_of(kobj, struct blk_integrity, kobj);
 242         struct integrity_sysfs_entry *entry =
 243                 container_of(attr, struct integrity_sysfs_entry, attr);
 244         ssize_t ret = 0;
 245
 246         if (entry->store)
 247                 ret = entry->store(bi, page, count);
 248
 249         return ret;
 250 }
 251
 252 static ssize_t integrity_format_show(struct blk_integrity *bi, char *page)
 253 {
 254         if (bi != NULL && bi->name != NULL)
 255                 return sprintf(page, "%s\n", bi->name);
 256         else
 257                 return sprintf(page, "none\n");
 258 }
 259
 260 static ssize_t integrity_tag_size_show(struct blk_integrity *bi, char *page)
 261 {
 262         if (bi != NULL)
 263                 return sprintf(page, "%u\n", bi->tag_size);
 264         else
 265                 return sprintf(page, "0\n");
 266 }
 267
 268 static ssize_t integrity_read_store(struct blk_integrity *bi,
 269                                     const char *page, size_t count)
 270 {
 271         char *p = (char *) page;
 272         unsigned long val = simple_strtoul(p, &p, 10);
 273
 274         if (val)
 275                 bi->flags |= INTEGRITY_FLAG_READ;
 276         else
 277                 bi->flags &= ~INTEGRITY_FLAG_READ;
 278
 279         return count;
 280 }
 281
 282 static ssize_t integrity_read_show(struct blk_integrity *bi, char *page)
 283 {
 284         return sprintf(page, "%d\n", (bi->flags & INTEGRITY_FLAG_READ) != 0);
 285 }
 286
 287 static ssize_t integrity_write_store(struct blk_integrity *bi,
 288                                      const char *page, size_t count)
 289 {
 290         char *p = (char *) page;
 291         unsigned long val = simple_strtoul(p, &p, 10);
 292
 293         if (val)
 294                 bi->flags |= INTEGRITY_FLAG_WRITE;
 295         else
 296                 bi->flags &= ~INTEGRITY_FLAG_WRITE;
 297
 298         return count;
 299 }
 300
 301 static ssize_t integrity_write_show(struct blk_integrity *bi, char *page)
 302 {
 303         return sprintf(page, "%d\n", (bi->flags & INTEGRITY_FLAG_WRITE) != 0);
 304 }
 305
 306 static struct integrity_sysfs_entry integrity_format_entry = {
 307         .attr = { .name = "format", .mode = S_IRUGO },
 308         .show = integrity_format_show,
 309 };
 310
 311 static struct integrity_sysfs_entry integrity_tag_size_entry = {
 312         .attr = { .name = "tag_size", .mode = S_IRUGO },
 313         .show = integrity_tag_size_show,
 314 };
 315
 316 static struct integrity_sysfs_entry integrity_read_entry = {
 317         .attr = { .name = "read_verify", .mode = S_IRUGO | S_IWUSR },
 318         .show = integrity_read_show,
 319         .store = integrity_read_store,
 320 };
 321
 322 static struct integrity_sysfs_entry integrity_write_entry = {
 323         .attr = { .name = "write_generate", .mode = S_IRUGO | S_IWUSR },
 324         .show = integrity_write_show,
 325         .store = integrity_write_store,
 326 };
 327
 328 static struct attribute *integrity_attrs[] = {
 329         &integrity_format_entry.attr,
 330         &integrity_tag_size_entry.attr,
 331         &integrity_read_entry.attr,
 332         &integrity_write_entry.attr,
 333         NULL,
 334 };
 335
 336 static const struct sysfs_ops integrity_ops = {
 337         .show   = &integrity_attr_show,
 338         .store  = &integrity_attr_store,
 339 };
 340
 341 static int __init blk_dev_integrity_init(void)
 342 {
 343         integrity_cachep = kmem_cache_create("blkdev_integrity",
 344                                              sizeof(struct blk_integrity),
 345                                              0, SLAB_PANIC, NULL);
 346         return 0;
 347 }
 348 subsys_initcall(blk_dev_integrity_init);
 349
 350 static void blk_integrity_release(struct kobject *kobj)
 351 {
 352         struct blk_integrity *bi =
 353                 container_of(kobj, struct blk_integrity, kobj);
 354
 355         kmem_cache_free(integrity_cachep, bi);
 356 }
 357
 358 static struct kobj_type integrity_ktype = {
 359         .default_attrs  = integrity_attrs,
 360         .sysfs_ops      = &integrity_ops,
 361         .release        = blk_integrity_release,
 362 };
 363
 364 bool blk_integrity_is_initialized(struct gendisk *disk)
 365 {
 366         struct blk_integrity *bi = blk_get_integrity(disk);
 367
 368         return (bi && bi->name && strcmp(bi->name, bi_unsupported_name) != 0);
 369 }
 370 EXPORT_SYMBOL(blk_integrity_is_initialized);
 371
 372 /**
 373  * blk_integrity_register - Register a gendisk as being integrity-capable
 374  * @disk:       struct gendisk pointer to make integrity-aware
 375  * @template:   optional integrity profile to register
 376  *
 377  * Description: When a device needs to advertise itself as being able
 378  * to send/receive integrity metadata it must use this function to
 379  * register the capability with the block layer.  The template is a
 380  * blk_integrity struct with values appropriate for the underlying
 381  * hardware.  If template is NULL the new profile is allocated but
 382  * not filled out. See Documentation/block/data-integrity.txt.
 383  */
 384 int blk_integrity_register(struct gendisk *disk, struct blk_integrity *template)
 385 {
 386         struct blk_integrity *bi;
 387
 388         BUG_ON(disk == NULL);
 389
 390         if (disk->integrity == NULL) {
 391                 bi = kmem_cache_alloc(integrity_cachep,
 392                                       GFP_KERNEL | __GFP_ZERO);
 393                 if (!bi)
 394                         return -1;
 395
 396                 if (kobject_init_and_add(&bi->kobj, &integrity_ktype,
 397                                          &disk_to_dev(disk)->kobj,
 398                                          "%s", "integrity")) {
 399                         kmem_cache_free(integrity_cachep, bi);
 400                         return -1;
 401                 }
 402
 403                 kobject_uevent(&bi->kobj, KOBJ_ADD);
 404
 405                 bi->flags |= INTEGRITY_FLAG_READ | INTEGRITY_FLAG_WRITE;
 406                 bi->sector_size = queue_logical_block_size(disk->queue);
 407                 disk->integrity = bi;
 408         } else
 409                 bi = disk->integrity;
 410
 411         /* Use the provided profile as template */
 412         if (template != NULL) {
 413                 bi->name = template->name;
 414                 bi->generate_fn = template->generate_fn;
 415                 bi->verify_fn = template->verify_fn;
 416                 bi->tuple_size = template->tuple_size;
 417                 bi->set_tag_fn = template->set_tag_fn;
 418                 bi->get_tag_fn = template->get_tag_fn;
 419                 bi->tag_size = template->tag_size;
 420         } else
 421                 bi->name = bi_unsupported_name;
 422
 423         disk->queue->backing_dev_info.capabilities |= BDI_CAP_STABLE_WRITES;
 424
 425         return 0;
 426 }
 427 EXPORT_SYMBOL(blk_integrity_register);
 428
 429 /**
 430  * blk_integrity_unregister - Remove block integrity profile
 431  * @disk:       disk whose integrity profile to deallocate
 432  *
 433  * Description: This function frees all memory used by the block
 434  * integrity profile.  To be called at device teardown.
 435  */
 436 void blk_integrity_unregister(struct gendisk *disk)
 437 {
 438         struct blk_integrity *bi;
 439
 440         if (!disk || !disk->integrity)
 441                 return;
 442
 443         disk->queue->backing_dev_info.capabilities &= ~BDI_CAP_STABLE_WRITES;
 444
 445         bi = disk->integrity;
 446
 447         kobject_uevent(&bi->kobj, KOBJ_REMOVE);
 448         kobject_del(&bi->kobj);
 449         kobject_put(&bi->kobj);
 450         disk->integrity = NULL;
 451 }
 452 EXPORT_SYMBOL(blk_integrity_unregister);