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/slab.h>
  28
  29 #include "blk.h"
  30
  31 static struct kmem_cache *integrity_cachep;
  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         unsigned int i = 0;
  47
  48         bio_for_each_integrity_vec(iv, bio, i) {
  49
  50                 if (ivprv) {
  51                         if (!BIOVEC_PHYS_MERGEABLE(ivprv, iv))
  52                                 goto new_segment;
  53
  54                         if (!BIOVEC_SEG_BOUNDARY(q, ivprv, iv))
  55                                 goto new_segment;
  56
  57                         if (seg_size + iv->bv_len > queue_max_segment_size(q))
  58                                 goto new_segment;
  59
  60                         seg_size += iv->bv_len;
  61                 } else {
  62 new_segment:
  63                         segments++;
  64                         seg_size = iv->bv_len;
  65                 }
  66
  67                 ivprv = iv;
  68         }
  69
  70         return segments;
  71 }
  72 EXPORT_SYMBOL(blk_rq_count_integrity_sg);
  73
  74 /**
  75  * blk_rq_map_integrity_sg - Map integrity metadata into a scatterlist
  76  * @q:          request queue
  77  * @bio:        bio with integrity metadata attached
  78  * @sglist:     target scatterlist
  79  *
  80  * Description: Map the integrity vectors in request into a
  81  * scatterlist.  The scatterlist must be big enough to hold all
  82  * elements.  I.e. sized using blk_rq_count_integrity_sg().
  83  */
  84 int blk_rq_map_integrity_sg(struct request_queue *q, struct bio *bio,
  85                             struct scatterlist *sglist)
  86 {
  87         struct bio_vec *iv, *ivprv = NULL;
  88         struct scatterlist *sg = NULL;
  89         unsigned int segments = 0;
  90         unsigned int i = 0;
  91
  92         bio_for_each_integrity_vec(iv, bio, i) {
  93
  94                 if (ivprv) {
  95                         if (!BIOVEC_PHYS_MERGEABLE(ivprv, iv))
  96                                 goto new_segment;
  97
  98                         if (!BIOVEC_SEG_BOUNDARY(q, ivprv, iv))
  99                                 goto new_segment;
 100
 101                         if (sg->length + iv->bv_len > queue_max_segment_size(q))
 102                                 goto new_segment;
 103
 104                         sg->length += iv->bv_len;
 105                 } else {
 106 new_segment:
 107                         if (!sg)
 108                                 sg = sglist;
 109                         else {
 110                                 sg->page_link &= ~0x02;
 111                                 sg = sg_next(sg);
 112                         }
 113
 114                         sg_set_page(sg, iv->bv_page, iv->bv_len, iv->bv_offset);
 115                         segments++;
 116                 }
 117
 118                 ivprv = iv;
 119         }
 120
 121         if (sg)
 122                 sg_mark_end(sg);
 123
 124         return segments;
 125 }
 126 EXPORT_SYMBOL(blk_rq_map_integrity_sg);
 127
 128 /**
 129  * blk_integrity_compare - Compare integrity profile of two disks
 130  * @gd1:        Disk to compare
 131  * @gd2:        Disk to compare
 132  *
 133  * Description: Meta-devices like DM and MD need to verify that all
 134  * sub-devices use the same integrity format before advertising to
 135  * upper layers that they can send/receive integrity metadata.  This
 136  * function can be used to check whether two gendisk devices have
 137  * compatible integrity formats.
 138  */
 139 int blk_integrity_compare(struct gendisk *gd1, struct gendisk *gd2)
 140 {
 141         struct blk_integrity *b1 = gd1->integrity;
 142         struct blk_integrity *b2 = gd2->integrity;
 143
 144         if (!b1 && !b2)
 145                 return 0;
 146
 147         if (!b1 || !b2)
 148                 return -1;
 149
 150         if (b1->sector_size != b2->sector_size) {
 151                 printk(KERN_ERR "%s: %s/%s sector sz %u != %u\n", __func__,
 152                        gd1->disk_name, gd2->disk_name,
 153                        b1->sector_size, b2->sector_size);
 154                 return -1;
 155         }
 156
 157         if (b1->tuple_size != b2->tuple_size) {
 158                 printk(KERN_ERR "%s: %s/%s tuple sz %u != %u\n", __func__,
 159                        gd1->disk_name, gd2->disk_name,
 160                        b1->tuple_size, b2->tuple_size);
 161                 return -1;
 162         }
 163
 164         if (b1->tag_size && b2->tag_size && (b1->tag_size != b2->tag_size)) {
 165                 printk(KERN_ERR "%s: %s/%s tag sz %u != %u\n", __func__,
 166                        gd1->disk_name, gd2->disk_name,
 167                        b1->tag_size, b2->tag_size);
 168                 return -1;
 169         }
 170
 171         if (strcmp(b1->name, b2->name)) {
 172                 printk(KERN_ERR "%s: %s/%s type %s != %s\n", __func__,
 173                        gd1->disk_name, gd2->disk_name,
 174                        b1->name, b2->name);
 175                 return -1;
 176         }
 177
 178         return 0;
 179 }
 180 EXPORT_SYMBOL(blk_integrity_compare);
 181
 182 int blk_integrity_merge_rq(struct request_queue *q, struct request *req,
 183                            struct request *next)
 184 {
 185         if (blk_integrity_rq(req) != blk_integrity_rq(next))
 186                 return -1;
 187
 188         if (req->nr_integrity_segments + next->nr_integrity_segments >
 189             q->limits.max_integrity_segments)
 190                 return -1;
 191
 192         return 0;
 193 }
 194 EXPORT_SYMBOL(blk_integrity_merge_rq);
 195
 196 int blk_integrity_merge_bio(struct request_queue *q, struct request *req,
 197                             struct bio *bio)
 198 {
 199         int nr_integrity_segs;
 200         struct bio *next = bio->bi_next;
 201
 202         bio->bi_next = NULL;
 203         nr_integrity_segs = blk_rq_count_integrity_sg(q, bio);
 204         bio->bi_next = next;
 205
 206         if (req->nr_integrity_segments + nr_integrity_segs >
 207             q->limits.max_integrity_segments)
 208                 return -1;
 209
 210         req->nr_integrity_segments += nr_integrity_segs;
 211
 212         return 0;
 213 }
 214 EXPORT_SYMBOL(blk_integrity_merge_bio);
 215
 216 struct integrity_sysfs_entry {
 217         struct attribute attr;
 218         ssize_t (*show)(struct blk_integrity *, char *);
 219         ssize_t (*store)(struct blk_integrity *, const char *, size_t);
 220 };
 221
 222 static ssize_t integrity_attr_show(struct kobject *kobj, struct attribute *attr,
 223                                    char *page)
 224 {
 225         struct blk_integrity *bi =
 226                 container_of(kobj, struct blk_integrity, kobj);
 227         struct integrity_sysfs_entry *entry =
 228                 container_of(attr, struct integrity_sysfs_entry, attr);
 229
 230         return entry->show(bi, page);
 231 }
 232
 233 static ssize_t integrity_attr_store(struct kobject *kobj,
 234                                     struct attribute *attr, const char *page,
 235                                     size_t count)
 236 {
 237         struct blk_integrity *bi =
 238                 container_of(kobj, struct blk_integrity, kobj);
 239         struct integrity_sysfs_entry *entry =
 240                 container_of(attr, struct integrity_sysfs_entry, attr);
 241         ssize_t ret = 0;
 242
 243         if (entry->store)
 244                 ret = entry->store(bi, page, count);
 245
 246         return ret;
 247 }
 248
 249 static ssize_t integrity_format_show(struct blk_integrity *bi, char *page)
 250 {
 251         if (bi != NULL && bi->name != NULL)
 252                 return sprintf(page, "%s\n", bi->name);
 253         else
 254                 return sprintf(page, "none\n");
 255 }
 256
 257 static ssize_t integrity_tag_size_show(struct blk_integrity *bi, char *page)
 258 {
 259         if (bi != NULL)
 260                 return sprintf(page, "%u\n", bi->tag_size);
 261         else
 262                 return sprintf(page, "0\n");
 263 }
 264
 265 static ssize_t integrity_read_store(struct blk_integrity *bi,
 266                                     const char *page, size_t count)
 267 {
 268         char *p = (char *) page;
 269         unsigned long val = simple_strtoul(p, &p, 10);
 270
 271         if (val)
 272                 bi->flags |= INTEGRITY_FLAG_READ;
 273         else
 274                 bi->flags &= ~INTEGRITY_FLAG_READ;
 275
 276         return count;
 277 }
 278
 279 static ssize_t integrity_read_show(struct blk_integrity *bi, char *page)
 280 {
 281         return sprintf(page, "%d\n", (bi->flags & INTEGRITY_FLAG_READ) != 0);
 282 }
 283
 284 static ssize_t integrity_write_store(struct blk_integrity *bi,
 285                                      const char *page, size_t count)
 286 {
 287         char *p = (char *) page;
 288         unsigned long val = simple_strtoul(p, &p, 10);
 289
 290         if (val)
 291                 bi->flags |= INTEGRITY_FLAG_WRITE;
 292         else
 293                 bi->flags &= ~INTEGRITY_FLAG_WRITE;
 294
 295         return count;
 296 }
 297
 298 static ssize_t integrity_write_show(struct blk_integrity *bi, char *page)
 299 {
 300         return sprintf(page, "%d\n", (bi->flags & INTEGRITY_FLAG_WRITE) != 0);
 301 }
 302
 303 static struct integrity_sysfs_entry integrity_format_entry = {
 304         .attr = { .name = "format", .mode = S_IRUGO },
 305         .show = integrity_format_show,
 306 };
 307
 308 static struct integrity_sysfs_entry integrity_tag_size_entry = {
 309         .attr = { .name = "tag_size", .mode = S_IRUGO },
 310         .show = integrity_tag_size_show,
 311 };
 312
 313 static struct integrity_sysfs_entry integrity_read_entry = {
 314         .attr = { .name = "read_verify", .mode = S_IRUGO | S_IWUSR },
 315         .show = integrity_read_show,
 316         .store = integrity_read_store,
 317 };
 318
 319 static struct integrity_sysfs_entry integrity_write_entry = {
 320         .attr = { .name = "write_generate", .mode = S_IRUGO | S_IWUSR },
 321         .show = integrity_write_show,
 322         .store = integrity_write_store,
 323 };
 324
 325 static struct attribute *integrity_attrs[] = {
 326         &integrity_format_entry.attr,
 327         &integrity_tag_size_entry.attr,
 328         &integrity_read_entry.attr,
 329         &integrity_write_entry.attr,
 330         NULL,
 331 };
 332
 333 static const struct sysfs_ops integrity_ops = {
 334         .show   = &integrity_attr_show,
 335         .store  = &integrity_attr_store,
 336 };
 337
 338 static int __init blk_dev_integrity_init(void)
 339 {
 340         integrity_cachep = kmem_cache_create("blkdev_integrity",
 341                                              sizeof(struct blk_integrity),
 342                                              0, SLAB_PANIC, NULL);
 343         return 0;
 344 }
 345 subsys_initcall(blk_dev_integrity_init);
 346
 347 static void blk_integrity_release(struct kobject *kobj)
 348 {
 349         struct blk_integrity *bi =
 350                 container_of(kobj, struct blk_integrity, kobj);
 351
 352         kmem_cache_free(integrity_cachep, bi);
 353 }
 354
 355 static struct kobj_type integrity_ktype = {
 356         .default_attrs  = integrity_attrs,
 357         .sysfs_ops      = &integrity_ops,
 358         .release        = blk_integrity_release,
 359 };
 360
 361 /**
 362  * blk_integrity_register - Register a gendisk as being integrity-capable
 363  * @disk:       struct gendisk pointer to make integrity-aware
 364  * @template:   optional integrity profile to register
 365  *
 366  * Description: When a device needs to advertise itself as being able
 367  * to send/receive integrity metadata it must use this function to
 368  * register the capability with the block layer.  The template is a
 369  * blk_integrity struct with values appropriate for the underlying
 370  * hardware.  If template is NULL the new profile is allocated but
 371  * not filled out. See Documentation/block/data-integrity.txt.
 372  */
 373 int blk_integrity_register(struct gendisk *disk, struct blk_integrity *template)
 374 {
 375         struct blk_integrity *bi;
 376
 377         BUG_ON(disk == NULL);
 378
 379         if (disk->integrity == NULL) {
 380                 bi = kmem_cache_alloc(integrity_cachep,
 381                                       GFP_KERNEL | __GFP_ZERO);
 382                 if (!bi)
 383                         return -1;
 384
 385                 if (kobject_init_and_add(&bi->kobj, &integrity_ktype,
 386                                          &disk_to_dev(disk)->kobj,
 387                                          "%s", "integrity")) {
 388                         kmem_cache_free(integrity_cachep, bi);
 389                         return -1;
 390                 }
 391
 392                 kobject_uevent(&bi->kobj, KOBJ_ADD);
 393
 394                 bi->flags |= INTEGRITY_FLAG_READ | INTEGRITY_FLAG_WRITE;
 395                 bi->sector_size = queue_logical_block_size(disk->queue);
 396                 disk->integrity = bi;
 397         } else
 398                 bi = disk->integrity;
 399
 400         /* Use the provided profile as template */
 401         if (template != NULL) {
 402                 bi->name = template->name;
 403                 bi->generate_fn = template->generate_fn;
 404                 bi->verify_fn = template->verify_fn;
 405                 bi->tuple_size = template->tuple_size;
 406                 bi->set_tag_fn = template->set_tag_fn;
 407                 bi->get_tag_fn = template->get_tag_fn;
 408                 bi->tag_size = template->tag_size;
 409         } else
 410                 bi->name = "unsupported";
 411
 412         return 0;
 413 }
 414 EXPORT_SYMBOL(blk_integrity_register);
 415
 416 /**
 417  * blk_integrity_unregister - Remove block integrity profile
 418  * @disk:       disk whose integrity profile to deallocate
 419  *
 420  * Description: This function frees all memory used by the block
 421  * integrity profile.  To be called at device teardown.
 422  */
 423 void blk_integrity_unregister(struct gendisk *disk)
 424 {
 425         struct blk_integrity *bi;
 426
 427         if (!disk || !disk->integrity)
 428                 return;
 429
 430         bi = disk->integrity;
 431
 432         kobject_uevent(&bi->kobj, KOBJ_REMOVE);
 433         kobject_del(&bi->kobj);
 434         kobject_put(&bi->kobj);
 435         disk->integrity = NULL;
 436 }
 437 EXPORT_SYMBOL(blk_integrity_unregister);