Blame view
drivers/md/raid10.c
92.8 KB
1da177e4c Linux-2.6.12-rc2 |
1 2 3 4 5 6 7 |
/* * raid10.c : Multiple Devices driver for Linux * * Copyright (C) 2000-2004 Neil Brown * * RAID-10 support for md. * |
25985edce Fix common misspe... |
8 |
* Base on code in raid1.c. See raid1.c for further copyright information. |
1da177e4c Linux-2.6.12-rc2 |
9 10 11 12 13 14 15 16 17 18 19 |
* * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by * the Free Software Foundation; either version 2, or (at your option) * any later version. * * You should have received a copy of the GNU General Public License * (for example /usr/src/linux/COPYING); if not, write to the Free * Software Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. */ |
5a0e3ad6a include cleanup: ... |
20 |
#include <linux/slab.h> |
255707274 md: build failure... |
21 |
#include <linux/delay.h> |
bff61975b md: move lots of ... |
22 |
#include <linux/blkdev.h> |
056075c76 md: Add module.h ... |
23 |
#include <linux/module.h> |
bff61975b md: move lots of ... |
24 |
#include <linux/seq_file.h> |
8bda470e8 md/raid: use prin... |
25 |
#include <linux/ratelimit.h> |
43b2e5d86 md: move md_k.h f... |
26 |
#include "md.h" |
ef740c372 md: move headers ... |
27 |
#include "raid10.h" |
dab8b2924 md: Add support f... |
28 |
#include "raid0.h" |
ef740c372 md: move headers ... |
29 |
#include "bitmap.h" |
1da177e4c Linux-2.6.12-rc2 |
30 31 32 33 34 35 36 37 |
/* * RAID10 provides a combination of RAID0 and RAID1 functionality. * The layout of data is defined by * chunk_size * raid_disks * near_copies (stored in low byte of layout) * far_copies (stored in second byte of layout) |
c93983bf5 [PATCH] md: suppo... |
38 |
* far_offset (stored in bit 16 of layout ) |
1da177e4c Linux-2.6.12-rc2 |
39 40 41 42 43 44 45 |
* * The data to be stored is divided into chunks using chunksize. * Each device is divided into far_copies sections. * In each section, chunks are laid out in a style similar to raid0, but * near_copies copies of each chunk is stored (each on a different drive). * The starting device for each section is offset near_copies from the starting * device of the previous section. |
c93983bf5 [PATCH] md: suppo... |
46 |
* Thus they are (near_copies*far_copies) of each chunk, and each is on a different |
1da177e4c Linux-2.6.12-rc2 |
47 48 49 |
* drive. * near_copies and far_copies must be at least one, and their product is at most * raid_disks. |
c93983bf5 [PATCH] md: suppo... |
50 51 52 53 |
* * If far_offset is true, then the far_copies are handled a bit differently. * The copies are still in different stripes, but instead of be very far apart * on disk, there are adjacent stripes. |
1da177e4c Linux-2.6.12-rc2 |
54 55 56 57 58 59 |
*/ /* * Number of guaranteed r10bios in case of extreme VM load: */ #define NR_RAID10_BIOS 256 |
34db0cd60 md: add proper wr... |
60 61 62 63 64 |
/* When there are this many requests queue to be written by * the raid10 thread, we become 'congested' to provide back-pressure * for writeback. */ static int max_queued_requests = 1024; |
e879a8793 md/raid10: typede... |
65 66 |
static void allow_barrier(struct r10conf *conf); static void lower_barrier(struct r10conf *conf); |
0a27ec96b [PATCH] md: impro... |
67 |
|
dd0fc66fb [PATCH] gfp flags... |
68 |
static void * r10bio_pool_alloc(gfp_t gfp_flags, void *data) |
1da177e4c Linux-2.6.12-rc2 |
69 |
{ |
e879a8793 md/raid10: typede... |
70 |
struct r10conf *conf = data; |
9f2c9d12b md: remove typede... |
71 |
int size = offsetof(struct r10bio, devs[conf->copies]); |
1da177e4c Linux-2.6.12-rc2 |
72 |
|
69335ef3b md/raid10: prepar... |
73 74 |
/* allocate a r10bio with room for raid_disks entries in the * bios array */ |
7eaceacca block: remove per... |
75 |
return kzalloc(size, gfp_flags); |
1da177e4c Linux-2.6.12-rc2 |
76 77 78 79 80 81 |
} static void r10bio_pool_free(void *r10_bio, void *data) { kfree(r10_bio); } |
0310fa216 Allow raid10 resy... |
82 |
/* Maximum size of each resync request */ |
1da177e4c Linux-2.6.12-rc2 |
83 |
#define RESYNC_BLOCK_SIZE (64*1024) |
1da177e4c Linux-2.6.12-rc2 |
84 |
#define RESYNC_PAGES ((RESYNC_BLOCK_SIZE + PAGE_SIZE-1) / PAGE_SIZE) |
0310fa216 Allow raid10 resy... |
85 86 87 88 |
/* amount of memory to reserve for resync requests */ #define RESYNC_WINDOW (1024*1024) /* maximum number of concurrent requests, memory permitting */ #define RESYNC_DEPTH (32*1024*1024/RESYNC_BLOCK_SIZE) |
1da177e4c Linux-2.6.12-rc2 |
89 90 91 92 93 94 95 96 |
/* * When performing a resync, we need to read and compare, so * we need as many pages are there are copies. * When performing a recovery, we need 2 bios, one for read, * one for write (we recover only one drive per r10buf) * */ |
dd0fc66fb [PATCH] gfp flags... |
97 |
static void * r10buf_pool_alloc(gfp_t gfp_flags, void *data) |
1da177e4c Linux-2.6.12-rc2 |
98 |
{ |
e879a8793 md/raid10: typede... |
99 |
struct r10conf *conf = data; |
1da177e4c Linux-2.6.12-rc2 |
100 |
struct page *page; |
9f2c9d12b md: remove typede... |
101 |
struct r10bio *r10_bio; |
1da177e4c Linux-2.6.12-rc2 |
102 103 104 105 106 |
struct bio *bio; int i, j; int nalloc; r10_bio = r10bio_pool_alloc(gfp_flags, conf); |
7eaceacca block: remove per... |
107 |
if (!r10_bio) |
1da177e4c Linux-2.6.12-rc2 |
108 |
return NULL; |
1da177e4c Linux-2.6.12-rc2 |
109 110 111 112 113 114 115 116 117 118 |
if (test_bit(MD_RECOVERY_SYNC, &conf->mddev->recovery)) nalloc = conf->copies; /* resync */ else nalloc = 2; /* recovery */ /* * Allocate bios. */ for (j = nalloc ; j-- ; ) { |
6746557f0 md: use bio_kmall... |
119 |
bio = bio_kmalloc(gfp_flags, RESYNC_PAGES); |
1da177e4c Linux-2.6.12-rc2 |
120 121 122 |
if (!bio) goto out_free_bio; r10_bio->devs[j].bio = bio; |
69335ef3b md/raid10: prepar... |
123 124 125 126 127 128 |
if (!conf->have_replacement) continue; bio = bio_kmalloc(gfp_flags, RESYNC_PAGES); if (!bio) goto out_free_bio; r10_bio->devs[j].repl_bio = bio; |
1da177e4c Linux-2.6.12-rc2 |
129 130 131 132 133 134 |
} /* * Allocate RESYNC_PAGES data pages and attach them * where needed. */ for (j = 0 ; j < nalloc; j++) { |
69335ef3b md/raid10: prepar... |
135 |
struct bio *rbio = r10_bio->devs[j].repl_bio; |
1da177e4c Linux-2.6.12-rc2 |
136 137 |
bio = r10_bio->devs[j].bio; for (i = 0; i < RESYNC_PAGES; i++) { |
c65060ad4 md/raid10: share ... |
138 139 140 141 142 143 144 145 |
if (j == 1 && !test_bit(MD_RECOVERY_SYNC, &conf->mddev->recovery)) { /* we can share bv_page's during recovery */ struct bio *rbio = r10_bio->devs[0].bio; page = rbio->bi_io_vec[i].bv_page; get_page(page); } else page = alloc_page(gfp_flags); |
1da177e4c Linux-2.6.12-rc2 |
146 147 148 149 |
if (unlikely(!page)) goto out_free_pages; bio->bi_io_vec[i].bv_page = page; |
69335ef3b md/raid10: prepar... |
150 151 |
if (rbio) rbio->bi_io_vec[i].bv_page = page; |
1da177e4c Linux-2.6.12-rc2 |
152 153 154 155 156 157 158 |
} } return r10_bio; out_free_pages: for ( ; i > 0 ; i--) |
1345b1d8a [PATCH] md: defin... |
159 |
safe_put_page(bio->bi_io_vec[i-1].bv_page); |
1da177e4c Linux-2.6.12-rc2 |
160 161 |
while (j--) for (i = 0; i < RESYNC_PAGES ; i++) |
1345b1d8a [PATCH] md: defin... |
162 |
safe_put_page(r10_bio->devs[j].bio->bi_io_vec[i].bv_page); |
1da177e4c Linux-2.6.12-rc2 |
163 164 |
j = -1; out_free_bio: |
69335ef3b md/raid10: prepar... |
165 |
while (++j < nalloc) { |
1da177e4c Linux-2.6.12-rc2 |
166 |
bio_put(r10_bio->devs[j].bio); |
69335ef3b md/raid10: prepar... |
167 168 169 |
if (r10_bio->devs[j].repl_bio) bio_put(r10_bio->devs[j].repl_bio); } |
1da177e4c Linux-2.6.12-rc2 |
170 171 172 173 174 175 176 |
r10bio_pool_free(r10_bio, conf); return NULL; } static void r10buf_pool_free(void *__r10_bio, void *data) { int i; |
e879a8793 md/raid10: typede... |
177 |
struct r10conf *conf = data; |
9f2c9d12b md: remove typede... |
178 |
struct r10bio *r10bio = __r10_bio; |
1da177e4c Linux-2.6.12-rc2 |
179 180 181 182 183 184 |
int j; for (j=0; j < conf->copies; j++) { struct bio *bio = r10bio->devs[j].bio; if (bio) { for (i = 0; i < RESYNC_PAGES; i++) { |
1345b1d8a [PATCH] md: defin... |
185 |
safe_put_page(bio->bi_io_vec[i].bv_page); |
1da177e4c Linux-2.6.12-rc2 |
186 187 188 189 |
bio->bi_io_vec[i].bv_page = NULL; } bio_put(bio); } |
69335ef3b md/raid10: prepar... |
190 191 192 |
bio = r10bio->devs[j].repl_bio; if (bio) bio_put(bio); |
1da177e4c Linux-2.6.12-rc2 |
193 194 195 |
} r10bio_pool_free(r10bio, conf); } |
e879a8793 md/raid10: typede... |
196 |
static void put_all_bios(struct r10conf *conf, struct r10bio *r10_bio) |
1da177e4c Linux-2.6.12-rc2 |
197 198 199 200 201 |
{ int i; for (i = 0; i < conf->copies; i++) { struct bio **bio = & r10_bio->devs[i].bio; |
749c55e94 md/raid10: clear ... |
202 |
if (!BIO_SPECIAL(*bio)) |
1da177e4c Linux-2.6.12-rc2 |
203 204 |
bio_put(*bio); *bio = NULL; |
69335ef3b md/raid10: prepar... |
205 206 207 208 |
bio = &r10_bio->devs[i].repl_bio; if (r10_bio->read_slot < 0 && !BIO_SPECIAL(*bio)) bio_put(*bio); *bio = NULL; |
1da177e4c Linux-2.6.12-rc2 |
209 210 |
} } |
9f2c9d12b md: remove typede... |
211 |
static void free_r10bio(struct r10bio *r10_bio) |
1da177e4c Linux-2.6.12-rc2 |
212 |
{ |
e879a8793 md/raid10: typede... |
213 |
struct r10conf *conf = r10_bio->mddev->private; |
1da177e4c Linux-2.6.12-rc2 |
214 |
|
1da177e4c Linux-2.6.12-rc2 |
215 216 217 |
put_all_bios(conf, r10_bio); mempool_free(r10_bio, conf->r10bio_pool); } |
9f2c9d12b md: remove typede... |
218 |
static void put_buf(struct r10bio *r10_bio) |
1da177e4c Linux-2.6.12-rc2 |
219 |
{ |
e879a8793 md/raid10: typede... |
220 |
struct r10conf *conf = r10_bio->mddev->private; |
1da177e4c Linux-2.6.12-rc2 |
221 222 |
mempool_free(r10_bio, conf->r10buf_pool); |
0a27ec96b [PATCH] md: impro... |
223 |
lower_barrier(conf); |
1da177e4c Linux-2.6.12-rc2 |
224 |
} |
9f2c9d12b md: remove typede... |
225 |
static void reschedule_retry(struct r10bio *r10_bio) |
1da177e4c Linux-2.6.12-rc2 |
226 227 |
{ unsigned long flags; |
fd01b88c7 md: remove typede... |
228 |
struct mddev *mddev = r10_bio->mddev; |
e879a8793 md/raid10: typede... |
229 |
struct r10conf *conf = mddev->private; |
1da177e4c Linux-2.6.12-rc2 |
230 231 232 |
spin_lock_irqsave(&conf->device_lock, flags); list_add(&r10_bio->retry_list, &conf->retry_list); |
4443ae10c [PATCH] md: auto-... |
233 |
conf->nr_queued ++; |
1da177e4c Linux-2.6.12-rc2 |
234 |
spin_unlock_irqrestore(&conf->device_lock, flags); |
388667bed md: raid10: wake ... |
235 236 |
/* wake up frozen array... */ wake_up(&conf->wait_barrier); |
1da177e4c Linux-2.6.12-rc2 |
237 238 239 240 241 242 243 244 |
md_wakeup_thread(mddev->thread); } /* * raid_end_bio_io() is called when we have finished servicing a mirrored * operation and are ready to return a success/failure code to the buffer * cache layer. */ |
9f2c9d12b md: remove typede... |
245 |
static void raid_end_bio_io(struct r10bio *r10_bio) |
1da177e4c Linux-2.6.12-rc2 |
246 247 |
{ struct bio *bio = r10_bio->master_bio; |
856e08e23 md/raid10: avoid ... |
248 |
int done; |
e879a8793 md/raid10: typede... |
249 |
struct r10conf *conf = r10_bio->mddev->private; |
1da177e4c Linux-2.6.12-rc2 |
250 |
|
856e08e23 md/raid10: avoid ... |
251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 |
if (bio->bi_phys_segments) { unsigned long flags; spin_lock_irqsave(&conf->device_lock, flags); bio->bi_phys_segments--; done = (bio->bi_phys_segments == 0); spin_unlock_irqrestore(&conf->device_lock, flags); } else done = 1; if (!test_bit(R10BIO_Uptodate, &r10_bio->state)) clear_bit(BIO_UPTODATE, &bio->bi_flags); if (done) { bio_endio(bio, 0); /* * Wake up any possible resync thread that waits for the device * to go idle. */ allow_barrier(conf); } |
1da177e4c Linux-2.6.12-rc2 |
269 270 271 272 273 274 |
free_r10bio(r10_bio); } /* * Update disk head position estimator based on IRQ completion info. */ |
9f2c9d12b md: remove typede... |
275 |
static inline void update_head_pos(int slot, struct r10bio *r10_bio) |
1da177e4c Linux-2.6.12-rc2 |
276 |
{ |
e879a8793 md/raid10: typede... |
277 |
struct r10conf *conf = r10_bio->mddev->private; |
1da177e4c Linux-2.6.12-rc2 |
278 279 280 281 |
conf->mirrors[r10_bio->devs[slot].devnum].head_position = r10_bio->devs[slot].addr + (r10_bio->sectors); } |
778ca0185 md/raid10: factor... |
282 283 284 |
/* * Find the disk number which triggered given bio */ |
e879a8793 md/raid10: typede... |
285 |
static int find_bio_disk(struct r10conf *conf, struct r10bio *r10_bio, |
69335ef3b md/raid10: prepar... |
286 |
struct bio *bio, int *slotp, int *replp) |
778ca0185 md/raid10: factor... |
287 288 |
{ int slot; |
69335ef3b md/raid10: prepar... |
289 |
int repl = 0; |
778ca0185 md/raid10: factor... |
290 |
|
69335ef3b md/raid10: prepar... |
291 |
for (slot = 0; slot < conf->copies; slot++) { |
778ca0185 md/raid10: factor... |
292 293 |
if (r10_bio->devs[slot].bio == bio) break; |
69335ef3b md/raid10: prepar... |
294 295 296 297 298 |
if (r10_bio->devs[slot].repl_bio == bio) { repl = 1; break; } } |
778ca0185 md/raid10: factor... |
299 300 301 |
BUG_ON(slot == conf->copies); update_head_pos(slot, r10_bio); |
749c55e94 md/raid10: clear ... |
302 303 |
if (slotp) *slotp = slot; |
69335ef3b md/raid10: prepar... |
304 305 |
if (replp) *replp = repl; |
778ca0185 md/raid10: factor... |
306 307 |
return r10_bio->devs[slot].devnum; } |
6712ecf8f Drop 'size' argum... |
308 |
static void raid10_end_read_request(struct bio *bio, int error) |
1da177e4c Linux-2.6.12-rc2 |
309 310 |
{ int uptodate = test_bit(BIO_UPTODATE, &bio->bi_flags); |
9f2c9d12b md: remove typede... |
311 |
struct r10bio *r10_bio = bio->bi_private; |
1da177e4c Linux-2.6.12-rc2 |
312 |
int slot, dev; |
abbf098e6 md/raid10: prefer... |
313 |
struct md_rdev *rdev; |
e879a8793 md/raid10: typede... |
314 |
struct r10conf *conf = r10_bio->mddev->private; |
1da177e4c Linux-2.6.12-rc2 |
315 |
|
1da177e4c Linux-2.6.12-rc2 |
316 317 318 |
slot = r10_bio->read_slot; dev = r10_bio->devs[slot].devnum; |
abbf098e6 md/raid10: prefer... |
319 |
rdev = r10_bio->devs[slot].rdev; |
1da177e4c Linux-2.6.12-rc2 |
320 321 322 |
/* * this branch is our 'one mirror IO has finished' event handler: */ |
4443ae10c [PATCH] md: auto-... |
323 324 325 |
update_head_pos(slot, r10_bio); if (uptodate) { |
1da177e4c Linux-2.6.12-rc2 |
326 327 328 329 330 331 332 333 334 335 |
/* * Set R10BIO_Uptodate in our master bio, so that * we will return a good error code to the higher * levels even if IO on some other mirrored buffer fails. * * The 'master' represents the composite IO operation to * user-side. So if something waits for IO, then it will * wait for the 'master' bio. */ set_bit(R10BIO_Uptodate, &r10_bio->state); |
1da177e4c Linux-2.6.12-rc2 |
336 |
raid_end_bio_io(r10_bio); |
abbf098e6 md/raid10: prefer... |
337 |
rdev_dec_pending(rdev, conf->mddev); |
4443ae10c [PATCH] md: auto-... |
338 |
} else { |
1da177e4c Linux-2.6.12-rc2 |
339 |
/* |
7c4e06ff2 md/raid10: some t... |
340 |
* oops, read error - keep the refcount on the rdev |
1da177e4c Linux-2.6.12-rc2 |
341 342 |
*/ char b[BDEVNAME_SIZE]; |
8bda470e8 md/raid: use prin... |
343 344 345 346 |
printk_ratelimited(KERN_ERR "md/raid10:%s: %s: rescheduling sector %llu ", mdname(conf->mddev), |
abbf098e6 md/raid10: prefer... |
347 |
bdevname(rdev->bdev, b), |
8bda470e8 md/raid: use prin... |
348 |
(unsigned long long)r10_bio->sector); |
856e08e23 md/raid10: avoid ... |
349 |
set_bit(R10BIO_ReadError, &r10_bio->state); |
1da177e4c Linux-2.6.12-rc2 |
350 351 |
reschedule_retry(r10_bio); } |
1da177e4c Linux-2.6.12-rc2 |
352 |
} |
9f2c9d12b md: remove typede... |
353 |
static void close_write(struct r10bio *r10_bio) |
bd870a16c md/raid10: Handl... |
354 355 356 357 358 359 360 361 |
{ /* clear the bitmap if all writes complete successfully */ bitmap_endwrite(r10_bio->mddev->bitmap, r10_bio->sector, r10_bio->sectors, !test_bit(R10BIO_Degraded, &r10_bio->state), 0); md_write_end(r10_bio->mddev); } |
9f2c9d12b md: remove typede... |
362 |
static void one_write_done(struct r10bio *r10_bio) |
19d5f834d md/raid10: unify ... |
363 364 365 366 367 368 369 370 371 372 373 374 375 |
{ if (atomic_dec_and_test(&r10_bio->remaining)) { if (test_bit(R10BIO_WriteError, &r10_bio->state)) reschedule_retry(r10_bio); else { close_write(r10_bio); if (test_bit(R10BIO_MadeGood, &r10_bio->state)) reschedule_retry(r10_bio); else raid_end_bio_io(r10_bio); } } } |
6712ecf8f Drop 'size' argum... |
376 |
static void raid10_end_write_request(struct bio *bio, int error) |
1da177e4c Linux-2.6.12-rc2 |
377 378 |
{ int uptodate = test_bit(BIO_UPTODATE, &bio->bi_flags); |
9f2c9d12b md: remove typede... |
379 |
struct r10bio *r10_bio = bio->bi_private; |
778ca0185 md/raid10: factor... |
380 |
int dev; |
749c55e94 md/raid10: clear ... |
381 |
int dec_rdev = 1; |
e879a8793 md/raid10: typede... |
382 |
struct r10conf *conf = r10_bio->mddev->private; |
475b0321a md/raid10: writes... |
383 |
int slot, repl; |
4ca40c2ce md/raid10: Allow ... |
384 |
struct md_rdev *rdev = NULL; |
1da177e4c Linux-2.6.12-rc2 |
385 |
|
475b0321a md/raid10: writes... |
386 |
dev = find_bio_disk(conf, r10_bio, bio, &slot, &repl); |
1da177e4c Linux-2.6.12-rc2 |
387 |
|
475b0321a md/raid10: writes... |
388 389 |
if (repl) rdev = conf->mirrors[dev].replacement; |
4ca40c2ce md/raid10: Allow ... |
390 391 392 |
if (!rdev) { smp_rmb(); repl = 0; |
475b0321a md/raid10: writes... |
393 |
rdev = conf->mirrors[dev].rdev; |
4ca40c2ce md/raid10: Allow ... |
394 |
} |
1da177e4c Linux-2.6.12-rc2 |
395 396 397 |
/* * this branch is our 'one mirror IO has finished' event handler: */ |
6cce3b23f [PATCH] md: write... |
398 |
if (!uptodate) { |
475b0321a md/raid10: writes... |
399 400 401 402 403 404 405 |
if (repl) /* Never record new bad blocks to replacement, * just fail it. */ md_error(rdev->mddev, rdev); else { set_bit(WriteErrorSeen, &rdev->flags); |
b7044d41b md/raid10: If the... |
406 407 408 |
if (!test_and_set_bit(WantReplacement, &rdev->flags)) set_bit(MD_RECOVERY_NEEDED, &rdev->mddev->recovery); |
475b0321a md/raid10: writes... |
409 410 411 |
set_bit(R10BIO_WriteError, &r10_bio->state); dec_rdev = 0; } |
749c55e94 md/raid10: clear ... |
412 |
} else { |
1da177e4c Linux-2.6.12-rc2 |
413 414 415 416 417 418 419 420 421 |
/* * Set R10BIO_Uptodate in our master bio, so that * we will return a good error code for to the higher * levels even if IO on some other mirrored buffer fails. * * The 'master' represents the composite IO operation to * user-side. So if something waits for IO, then it will * wait for the 'master' bio. */ |
749c55e94 md/raid10: clear ... |
422 423 |
sector_t first_bad; int bad_sectors; |
1da177e4c Linux-2.6.12-rc2 |
424 |
set_bit(R10BIO_Uptodate, &r10_bio->state); |
749c55e94 md/raid10: clear ... |
425 |
/* Maybe we can clear some bad blocks. */ |
475b0321a md/raid10: writes... |
426 |
if (is_badblock(rdev, |
749c55e94 md/raid10: clear ... |
427 428 429 430 |
r10_bio->devs[slot].addr, r10_bio->sectors, &first_bad, &bad_sectors)) { bio_put(bio); |
475b0321a md/raid10: writes... |
431 432 433 434 |
if (repl) r10_bio->devs[slot].repl_bio = IO_MADE_GOOD; else r10_bio->devs[slot].bio = IO_MADE_GOOD; |
749c55e94 md/raid10: clear ... |
435 436 437 438 |
dec_rdev = 0; set_bit(R10BIO_MadeGood, &r10_bio->state); } } |
1da177e4c Linux-2.6.12-rc2 |
439 440 441 442 443 |
/* * * Let's see if all mirrored write operations have finished * already. */ |
19d5f834d md/raid10: unify ... |
444 |
one_write_done(r10_bio); |
749c55e94 md/raid10: clear ... |
445 446 |
if (dec_rdev) rdev_dec_pending(conf->mirrors[dev].rdev, conf->mddev); |
1da177e4c Linux-2.6.12-rc2 |
447 |
} |
1da177e4c Linux-2.6.12-rc2 |
448 449 |
/* * RAID10 layout manager |
25985edce Fix common misspe... |
450 |
* As well as the chunksize and raid_disks count, there are two |
1da177e4c Linux-2.6.12-rc2 |
451 452 453 454 455 456 |
* parameters: near_copies and far_copies. * near_copies * far_copies must be <= raid_disks. * Normally one of these will be 1. * If both are 1, we get raid0. * If near_copies == raid_disks, we get raid1. * |
25985edce Fix common misspe... |
457 |
* Chunks are laid out in raid0 style with near_copies copies of the |
1da177e4c Linux-2.6.12-rc2 |
458 459 460 461 462 463 464 465 466 |
* first chunk, followed by near_copies copies of the next chunk and * so on. * If far_copies > 1, then after 1/far_copies of the array has been assigned * as described above, we start again with a device offset of near_copies. * So we effectively have another copy of the whole array further down all * the drives, but with blocks on different drives. * With this layout, and block is never stored twice on the one device. * * raid10_find_phys finds the sector offset of a given virtual sector |
c93983bf5 [PATCH] md: suppo... |
467 |
* on each device that it is on. |
1da177e4c Linux-2.6.12-rc2 |
468 469 470 471 |
* * raid10_find_virt does the reverse mapping, from a device and a * sector offset to a virtual address */ |
e879a8793 md/raid10: typede... |
472 |
static void raid10_find_phys(struct r10conf *conf, struct r10bio *r10bio) |
1da177e4c Linux-2.6.12-rc2 |
473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 |
{ int n,f; sector_t sector; sector_t chunk; sector_t stripe; int dev; int slot = 0; /* now calculate first sector/dev */ chunk = r10bio->sector >> conf->chunk_shift; sector = r10bio->sector & conf->chunk_mask; chunk *= conf->near_copies; stripe = chunk; dev = sector_div(stripe, conf->raid_disks); |
c93983bf5 [PATCH] md: suppo... |
489 490 |
if (conf->far_offset) stripe *= conf->far_copies; |
1da177e4c Linux-2.6.12-rc2 |
491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 |
sector += stripe << conf->chunk_shift; /* and calculate all the others */ for (n=0; n < conf->near_copies; n++) { int d = dev; sector_t s = sector; r10bio->devs[slot].addr = sector; r10bio->devs[slot].devnum = d; slot++; for (f = 1; f < conf->far_copies; f++) { d += conf->near_copies; if (d >= conf->raid_disks) d -= conf->raid_disks; s += conf->stride; r10bio->devs[slot].devnum = d; r10bio->devs[slot].addr = s; slot++; } dev++; if (dev >= conf->raid_disks) { dev = 0; sector += (conf->chunk_mask + 1); } } BUG_ON(slot != conf->copies); } |
e879a8793 md/raid10: typede... |
519 |
static sector_t raid10_find_virt(struct r10conf *conf, sector_t sector, int dev) |
1da177e4c Linux-2.6.12-rc2 |
520 521 |
{ sector_t offset, chunk, vchunk; |
1da177e4c Linux-2.6.12-rc2 |
522 |
offset = sector & conf->chunk_mask; |
c93983bf5 [PATCH] md: suppo... |
523 524 525 526 527 528 529 530 |
if (conf->far_offset) { int fc; chunk = sector >> conf->chunk_shift; fc = sector_div(chunk, conf->far_copies); dev -= fc * conf->near_copies; if (dev < 0) dev += conf->raid_disks; } else { |
64a742bc6 [PATCH] md: fix r... |
531 |
while (sector >= conf->stride) { |
c93983bf5 [PATCH] md: suppo... |
532 533 534 535 536 537 538 539 |
sector -= conf->stride; if (dev < conf->near_copies) dev += conf->raid_disks - conf->near_copies; else dev -= conf->near_copies; } chunk = sector >> conf->chunk_shift; } |
1da177e4c Linux-2.6.12-rc2 |
540 541 542 543 544 545 546 547 |
vchunk = chunk * conf->raid_disks + dev; sector_div(vchunk, conf->near_copies); return (vchunk << conf->chunk_shift) + offset; } /** * raid10_mergeable_bvec -- tell bio layer if a two requests can be merged * @q: request queue |
cc371e66e Add bvec_merge_da... |
548 |
* @bvm: properties of new bio |
1da177e4c Linux-2.6.12-rc2 |
549 550 551 552 553 554 |
* @biovec: the request that could be merged to it. * * Return amount of bytes we can accept at this offset * If near_copies == raid_disk, there are no striping issues, * but in that case, the function isn't called at all. */ |
cc371e66e Add bvec_merge_da... |
555 556 557 |
static int raid10_mergeable_bvec(struct request_queue *q, struct bvec_merge_data *bvm, struct bio_vec *biovec) |
1da177e4c Linux-2.6.12-rc2 |
558 |
{ |
fd01b88c7 md: remove typede... |
559 |
struct mddev *mddev = q->queuedata; |
cc371e66e Add bvec_merge_da... |
560 |
sector_t sector = bvm->bi_sector + get_start_sect(bvm->bi_bdev); |
1da177e4c Linux-2.6.12-rc2 |
561 |
int max; |
9d8f03636 md: Make mddev->c... |
562 |
unsigned int chunk_sectors = mddev->chunk_sectors; |
cc371e66e Add bvec_merge_da... |
563 |
unsigned int bio_sectors = bvm->bi_size >> 9; |
1da177e4c Linux-2.6.12-rc2 |
564 565 566 |
max = (chunk_sectors - ((sector & (chunk_sectors - 1)) + bio_sectors)) << 9; if (max < 0) max = 0; /* bio_add cannot handle a negative return */ |
cc371e66e Add bvec_merge_da... |
567 568 |
if (max <= biovec->bv_len && bio_sectors == 0) return biovec->bv_len; |
1da177e4c Linux-2.6.12-rc2 |
569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 |
else return max; } /* * This routine returns the disk from which the requested read should * be done. There is a per-array 'next expected sequential IO' sector * number - if this matches on the next IO then we use the last disk. * There is also a per-disk 'last know head position' sector that is * maintained from IRQ contexts, both the normal and the resync IO * completion handlers update this position correctly. If there is no * perfect sequential match then we pick the disk whose head is closest. * * If there are 2 mirrors in the same 2 devices, performance degrades * because position is mirror, not device based. * * The rdev for the device selected will have nr_pending incremented. */ /* * FIXME: possibly should rethink readbalancing and do it differently * depending on near_copies / far_copies geometry. */ |
96c3fd1f3 md/raid10: chang... |
592 593 594 |
static struct md_rdev *read_balance(struct r10conf *conf, struct r10bio *r10_bio, int *max_sectors) |
1da177e4c Linux-2.6.12-rc2 |
595 |
{ |
af3a2cd6b md: Fix read bala... |
596 |
const sector_t this_sector = r10_bio->sector; |
56d991210 md: simplify raid... |
597 |
int disk, slot; |
856e08e23 md/raid10: avoid ... |
598 599 |
int sectors = r10_bio->sectors; int best_good_sectors; |
56d991210 md: simplify raid... |
600 |
sector_t new_distance, best_dist; |
abbf098e6 md/raid10: prefer... |
601 |
struct md_rdev *rdev, *best_rdev; |
56d991210 md: simplify raid... |
602 603 |
int do_balance; int best_slot; |
1da177e4c Linux-2.6.12-rc2 |
604 605 606 |
raid10_find_phys(conf, r10_bio); rcu_read_lock(); |
56d991210 md: simplify raid... |
607 |
retry: |
856e08e23 md/raid10: avoid ... |
608 |
sectors = r10_bio->sectors; |
56d991210 md: simplify raid... |
609 |
best_slot = -1; |
abbf098e6 md/raid10: prefer... |
610 |
best_rdev = NULL; |
56d991210 md: simplify raid... |
611 |
best_dist = MaxSector; |
856e08e23 md/raid10: avoid ... |
612 |
best_good_sectors = 0; |
56d991210 md: simplify raid... |
613 |
do_balance = 1; |
1da177e4c Linux-2.6.12-rc2 |
614 615 |
/* * Check if we can balance. We can balance on the whole |
6cce3b23f [PATCH] md: write... |
616 617 618 |
* device if no resync is going on (recovery is ok), or below * the resync window. We take the first readable disk when * above the resync window. |
1da177e4c Linux-2.6.12-rc2 |
619 620 |
*/ if (conf->mddev->recovery_cp < MaxSector |
56d991210 md: simplify raid... |
621 622 |
&& (this_sector + sectors >= conf->next_resync)) do_balance = 0; |
1da177e4c Linux-2.6.12-rc2 |
623 |
|
56d991210 md: simplify raid... |
624 |
for (slot = 0; slot < conf->copies ; slot++) { |
856e08e23 md/raid10: avoid ... |
625 626 627 |
sector_t first_bad; int bad_sectors; sector_t dev_sector; |
56d991210 md: simplify raid... |
628 629 |
if (r10_bio->devs[slot].bio == IO_BLOCKED) continue; |
1da177e4c Linux-2.6.12-rc2 |
630 |
disk = r10_bio->devs[slot].devnum; |
abbf098e6 md/raid10: prefer... |
631 632 633 634 |
rdev = rcu_dereference(conf->mirrors[disk].replacement); if (rdev == NULL || test_bit(Faulty, &rdev->flags) || r10_bio->devs[slot].addr + sectors > rdev->recovery_offset) rdev = rcu_dereference(conf->mirrors[disk].rdev); |
56d991210 md: simplify raid... |
635 |
if (rdev == NULL) |
1da177e4c Linux-2.6.12-rc2 |
636 |
continue; |
abbf098e6 md/raid10: prefer... |
637 638 639 640 |
if (test_bit(Faulty, &rdev->flags)) continue; if (!test_bit(In_sync, &rdev->flags) && r10_bio->devs[slot].addr + sectors > rdev->recovery_offset) |
56d991210 md: simplify raid... |
641 |
continue; |
856e08e23 md/raid10: avoid ... |
642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 |
dev_sector = r10_bio->devs[slot].addr; if (is_badblock(rdev, dev_sector, sectors, &first_bad, &bad_sectors)) { if (best_dist < MaxSector) /* Already have a better slot */ continue; if (first_bad <= dev_sector) { /* Cannot read here. If this is the * 'primary' device, then we must not read * beyond 'bad_sectors' from another device. */ bad_sectors -= (dev_sector - first_bad); if (!do_balance && sectors > bad_sectors) sectors = bad_sectors; if (best_good_sectors > sectors) best_good_sectors = sectors; } else { sector_t good_sectors = first_bad - dev_sector; if (good_sectors > best_good_sectors) { best_good_sectors = good_sectors; best_slot = slot; |
abbf098e6 md/raid10: prefer... |
664 |
best_rdev = rdev; |
856e08e23 md/raid10: avoid ... |
665 666 667 668 669 670 671 672 |
} if (!do_balance) /* Must read from here */ break; } continue; } else best_good_sectors = sectors; |
56d991210 md: simplify raid... |
673 674 |
if (!do_balance) break; |
1da177e4c Linux-2.6.12-rc2 |
675 |
|
22dfdf521 [PATCH] md: impro... |
676 677 678 679 |
/* This optimisation is debatable, and completely destroys * sequential read speed for 'far copies' arrays. So only * keep it for 'near' arrays, and review those later. */ |
56d991210 md: simplify raid... |
680 |
if (conf->near_copies > 1 && !atomic_read(&rdev->nr_pending)) |
1da177e4c Linux-2.6.12-rc2 |
681 |
break; |
8ed3a1956 md: don't attempt... |
682 683 684 |
/* for far > 1 always use the lowest address */ if (conf->far_copies > 1) |
56d991210 md: simplify raid... |
685 |
new_distance = r10_bio->devs[slot].addr; |
8ed3a1956 md: don't attempt... |
686 |
else |
56d991210 md: simplify raid... |
687 688 689 690 691 |
new_distance = abs(r10_bio->devs[slot].addr - conf->mirrors[disk].head_position); if (new_distance < best_dist) { best_dist = new_distance; best_slot = slot; |
abbf098e6 md/raid10: prefer... |
692 |
best_rdev = rdev; |
1da177e4c Linux-2.6.12-rc2 |
693 694 |
} } |
abbf098e6 md/raid10: prefer... |
695 |
if (slot >= conf->copies) { |
56d991210 md: simplify raid... |
696 |
slot = best_slot; |
abbf098e6 md/raid10: prefer... |
697 698 |
rdev = best_rdev; } |
1da177e4c Linux-2.6.12-rc2 |
699 |
|
56d991210 md: simplify raid... |
700 |
if (slot >= 0) { |
56d991210 md: simplify raid... |
701 702 703 704 705 706 707 708 709 710 |
atomic_inc(&rdev->nr_pending); if (test_bit(Faulty, &rdev->flags)) { /* Cannot risk returning a device that failed * before we inc'ed nr_pending */ rdev_dec_pending(rdev, conf->mddev); goto retry; } r10_bio->read_slot = slot; } else |
96c3fd1f3 md/raid10: chang... |
711 |
rdev = NULL; |
1da177e4c Linux-2.6.12-rc2 |
712 |
rcu_read_unlock(); |
856e08e23 md/raid10: avoid ... |
713 |
*max_sectors = best_good_sectors; |
1da177e4c Linux-2.6.12-rc2 |
714 |
|
96c3fd1f3 md/raid10: chang... |
715 |
return rdev; |
1da177e4c Linux-2.6.12-rc2 |
716 |
} |
0d1292282 [PATCH] md: defin... |
717 718 |
static int raid10_congested(void *data, int bits) { |
fd01b88c7 md: remove typede... |
719 |
struct mddev *mddev = data; |
e879a8793 md/raid10: typede... |
720 |
struct r10conf *conf = mddev->private; |
0d1292282 [PATCH] md: defin... |
721 |
int i, ret = 0; |
34db0cd60 md: add proper wr... |
722 723 724 |
if ((bits & (1 << BDI_async_congested)) && conf->pending_count >= max_queued_requests) return 1; |
3fa841d7e md: report device... |
725 726 |
if (mddev_congested(mddev, bits)) return 1; |
0d1292282 [PATCH] md: defin... |
727 |
rcu_read_lock(); |
84707f38e md: don't use mdd... |
728 |
for (i = 0; i < conf->raid_disks && ret == 0; i++) { |
3cb030020 md: removing type... |
729 |
struct md_rdev *rdev = rcu_dereference(conf->mirrors[i].rdev); |
0d1292282 [PATCH] md: defin... |
730 |
if (rdev && !test_bit(Faulty, &rdev->flags)) { |
165125e1e [BLOCK] Get rid o... |
731 |
struct request_queue *q = bdev_get_queue(rdev->bdev); |
0d1292282 [PATCH] md: defin... |
732 733 734 735 736 737 738 |
ret |= bdi_congested(&q->backing_dev_info, bits); } } rcu_read_unlock(); return ret; } |
e879a8793 md/raid10: typede... |
739 |
static void flush_pending_writes(struct r10conf *conf) |
a35e63efa md: fix deadlock ... |
740 741 742 |
{ /* Any writes that have been queued but are awaiting * bitmap updates get flushed here. |
a35e63efa md: fix deadlock ... |
743 |
*/ |
a35e63efa md: fix deadlock ... |
744 745 746 747 748 |
spin_lock_irq(&conf->device_lock); if (conf->pending_bio_list.head) { struct bio *bio; bio = bio_list_get(&conf->pending_bio_list); |
34db0cd60 md: add proper wr... |
749 |
conf->pending_count = 0; |
a35e63efa md: fix deadlock ... |
750 751 752 753 |
spin_unlock_irq(&conf->device_lock); /* flush any pending bitmap writes to disk * before proceeding w/ I/O */ bitmap_unplug(conf->mddev->bitmap); |
34db0cd60 md: add proper wr... |
754 |
wake_up(&conf->wait_barrier); |
a35e63efa md: fix deadlock ... |
755 756 757 758 759 760 761 |
while (bio) { /* submit pending writes */ struct bio *next = bio->bi_next; bio->bi_next = NULL; generic_make_request(bio); bio = next; } |
a35e63efa md: fix deadlock ... |
762 763 |
} else spin_unlock_irq(&conf->device_lock); |
a35e63efa md: fix deadlock ... |
764 |
} |
7eaceacca block: remove per... |
765 |
|
0a27ec96b [PATCH] md: impro... |
766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 |
/* Barriers.... * Sometimes we need to suspend IO while we do something else, * either some resync/recovery, or reconfigure the array. * To do this we raise a 'barrier'. * The 'barrier' is a counter that can be raised multiple times * to count how many activities are happening which preclude * normal IO. * We can only raise the barrier if there is no pending IO. * i.e. if nr_pending == 0. * We choose only to raise the barrier if no-one is waiting for the * barrier to go down. This means that as soon as an IO request * is ready, no other operations which require a barrier will start * until the IO request has had a chance. * * So: regular IO calls 'wait_barrier'. When that returns there * is no backgroup IO happening, It must arrange to call * allow_barrier when it has finished its IO. * backgroup IO calls must call raise_barrier. Once that returns * there is no normal IO happeing. It must arrange to call * lower_barrier when the particular background IO completes. |
1da177e4c Linux-2.6.12-rc2 |
786 |
*/ |
1da177e4c Linux-2.6.12-rc2 |
787 |
|
e879a8793 md/raid10: typede... |
788 |
static void raise_barrier(struct r10conf *conf, int force) |
1da177e4c Linux-2.6.12-rc2 |
789 |
{ |
6cce3b23f [PATCH] md: write... |
790 |
BUG_ON(force && !conf->barrier); |
1da177e4c Linux-2.6.12-rc2 |
791 |
spin_lock_irq(&conf->resync_lock); |
0a27ec96b [PATCH] md: impro... |
792 |
|
6cce3b23f [PATCH] md: write... |
793 794 |
/* Wait until no block IO is waiting (unless 'force') */ wait_event_lock_irq(conf->wait_barrier, force || !conf->nr_waiting, |
c3b328ac8 md: fix up raid1/... |
795 |
conf->resync_lock, ); |
0a27ec96b [PATCH] md: impro... |
796 797 798 |
/* block any new IO from starting */ conf->barrier++; |
c3b328ac8 md: fix up raid1/... |
799 |
/* Now wait for all pending IO to complete */ |
0a27ec96b [PATCH] md: impro... |
800 801 |
wait_event_lock_irq(conf->wait_barrier, !conf->nr_pending && conf->barrier < RESYNC_DEPTH, |
c3b328ac8 md: fix up raid1/... |
802 |
conf->resync_lock, ); |
0a27ec96b [PATCH] md: impro... |
803 804 805 |
spin_unlock_irq(&conf->resync_lock); } |
e879a8793 md/raid10: typede... |
806 |
static void lower_barrier(struct r10conf *conf) |
0a27ec96b [PATCH] md: impro... |
807 808 809 810 811 812 813 |
{ unsigned long flags; spin_lock_irqsave(&conf->resync_lock, flags); conf->barrier--; spin_unlock_irqrestore(&conf->resync_lock, flags); wake_up(&conf->wait_barrier); } |
e879a8793 md/raid10: typede... |
814 |
static void wait_barrier(struct r10conf *conf) |
0a27ec96b [PATCH] md: impro... |
815 816 817 818 819 820 |
{ spin_lock_irq(&conf->resync_lock); if (conf->barrier) { conf->nr_waiting++; wait_event_lock_irq(conf->wait_barrier, !conf->barrier, conf->resync_lock, |
c3b328ac8 md: fix up raid1/... |
821 |
); |
0a27ec96b [PATCH] md: impro... |
822 |
conf->nr_waiting--; |
1da177e4c Linux-2.6.12-rc2 |
823 |
} |
0a27ec96b [PATCH] md: impro... |
824 |
conf->nr_pending++; |
1da177e4c Linux-2.6.12-rc2 |
825 826 |
spin_unlock_irq(&conf->resync_lock); } |
e879a8793 md/raid10: typede... |
827 |
static void allow_barrier(struct r10conf *conf) |
0a27ec96b [PATCH] md: impro... |
828 829 830 831 832 833 834 |
{ unsigned long flags; spin_lock_irqsave(&conf->resync_lock, flags); conf->nr_pending--; spin_unlock_irqrestore(&conf->resync_lock, flags); wake_up(&conf->wait_barrier); } |
e879a8793 md/raid10: typede... |
835 |
static void freeze_array(struct r10conf *conf) |
4443ae10c [PATCH] md: auto-... |
836 837 |
{ /* stop syncio and normal IO and wait for everything to |
f188593ee [PATCH] md: fix t... |
838 |
* go quiet. |
4443ae10c [PATCH] md: auto-... |
839 |
* We increment barrier and nr_waiting, and then |
1c830532f md: fix possible ... |
840 841 842 843 844 845 846 847 |
* wait until nr_pending match nr_queued+1 * This is called in the context of one normal IO request * that has failed. Thus any sync request that might be pending * will be blocked by nr_pending, and we need to wait for * pending IO requests to complete or be queued for re-try. * Thus the number queued (nr_queued) plus this request (1) * must match the number of pending IOs (nr_pending) before * we continue. |
4443ae10c [PATCH] md: auto-... |
848 849 850 851 852 |
*/ spin_lock_irq(&conf->resync_lock); conf->barrier++; conf->nr_waiting++; wait_event_lock_irq(conf->wait_barrier, |
1c830532f md: fix possible ... |
853 |
conf->nr_pending == conf->nr_queued+1, |
4443ae10c [PATCH] md: auto-... |
854 |
conf->resync_lock, |
c3b328ac8 md: fix up raid1/... |
855 |
flush_pending_writes(conf)); |
4443ae10c [PATCH] md: auto-... |
856 857 |
spin_unlock_irq(&conf->resync_lock); } |
e879a8793 md/raid10: typede... |
858 |
static void unfreeze_array(struct r10conf *conf) |
4443ae10c [PATCH] md: auto-... |
859 860 861 862 863 864 865 866 |
{ /* reverse the effect of the freeze */ spin_lock_irq(&conf->resync_lock); conf->barrier--; conf->nr_waiting--; wake_up(&conf->wait_barrier); spin_unlock_irq(&conf->resync_lock); } |
b4fdcb02f Merge branch 'for... |
867 |
static void make_request(struct mddev *mddev, struct bio * bio) |
1da177e4c Linux-2.6.12-rc2 |
868 |
{ |
e879a8793 md/raid10: typede... |
869 |
struct r10conf *conf = mddev->private; |
9f2c9d12b md: remove typede... |
870 |
struct r10bio *r10_bio; |
1da177e4c Linux-2.6.12-rc2 |
871 872 873 |
struct bio *read_bio; int i; int chunk_sects = conf->chunk_mask + 1; |
a362357b6 [BLOCK] Unify the... |
874 |
const int rw = bio_data_dir(bio); |
2c7d46ec1 md raid-1/10 Fix ... |
875 |
const unsigned long do_sync = (bio->bi_rw & REQ_SYNC); |
e9c7469bb md: implment REQ_... |
876 |
const unsigned long do_fua = (bio->bi_rw & REQ_FUA); |
6cce3b23f [PATCH] md: write... |
877 |
unsigned long flags; |
3cb030020 md: removing type... |
878 |
struct md_rdev *blocked_rdev; |
c3b328ac8 md: fix up raid1/... |
879 |
int plugged; |
d4432c23b md/raid10: avoid ... |
880 881 |
int sectors_handled; int max_sectors; |
1da177e4c Linux-2.6.12-rc2 |
882 |
|
e9c7469bb md: implment REQ_... |
883 884 |
if (unlikely(bio->bi_rw & REQ_FLUSH)) { md_flush_request(mddev, bio); |
5a7bbad27 block: remove sup... |
885 |
return; |
e5dcdd80a [PATCH] md: fail ... |
886 |
} |
1da177e4c Linux-2.6.12-rc2 |
887 888 889 890 891 892 893 894 895 896 897 898 899 900 |
/* If this request crosses a chunk boundary, we need to * split it. This will only happen for 1 PAGE (or less) requests. */ if (unlikely( (bio->bi_sector & conf->chunk_mask) + (bio->bi_size >> 9) > chunk_sects && conf->near_copies < conf->raid_disks)) { struct bio_pair *bp; /* Sanity check -- queue functions should prevent this happening */ if (bio->bi_vcnt != 1 || bio->bi_idx != 0) goto bad_map; /* This is a one page bio that upper layers * refuse to split for us, so we need to split it. */ |
6feef531f block: mark bio_s... |
901 |
bp = bio_split(bio, |
1da177e4c Linux-2.6.12-rc2 |
902 |
chunk_sects - (bio->bi_sector & (chunk_sects - 1)) ); |
51e9ac770 md/raid10: fix de... |
903 904 905 906 907 908 909 910 911 912 913 914 |
/* Each of these 'make_request' calls will call 'wait_barrier'. * If the first succeeds but the second blocks due to the resync * thread raising the barrier, we will deadlock because the * IO to the underlying device will be queued in generic_make_request * and will never complete, so will never reduce nr_pending. * So increment nr_waiting here so no new raise_barriers will * succeed, and so the second wait_barrier cannot block. */ spin_lock_irq(&conf->resync_lock); conf->nr_waiting++; spin_unlock_irq(&conf->resync_lock); |
5a7bbad27 block: remove sup... |
915 916 |
make_request(mddev, &bp->bio1); make_request(mddev, &bp->bio2); |
1da177e4c Linux-2.6.12-rc2 |
917 |
|
51e9ac770 md/raid10: fix de... |
918 919 920 921 |
spin_lock_irq(&conf->resync_lock); conf->nr_waiting--; wake_up(&conf->wait_barrier); spin_unlock_irq(&conf->resync_lock); |
1da177e4c Linux-2.6.12-rc2 |
922 |
bio_pair_release(bp); |
5a7bbad27 block: remove sup... |
923 |
return; |
1da177e4c Linux-2.6.12-rc2 |
924 |
bad_map: |
128595ed6 md/raid10: tidy u... |
925 926 927 |
printk("md/raid10:%s: make_request bug: can't convert block across chunks" " or bigger than %dk %llu %d ", mdname(mddev), chunk_sects/2, |
1da177e4c Linux-2.6.12-rc2 |
928 |
(unsigned long long)bio->bi_sector, bio->bi_size >> 10); |
6712ecf8f Drop 'size' argum... |
929 |
bio_io_error(bio); |
5a7bbad27 block: remove sup... |
930 |
return; |
1da177e4c Linux-2.6.12-rc2 |
931 |
} |
3d310eb7b [PATCH] md: fix d... |
932 |
md_write_start(mddev, bio); |
06d91a5fe [PATCH] md: impro... |
933 |
|
1da177e4c Linux-2.6.12-rc2 |
934 935 936 937 938 |
/* * Register the new request and wait if the reconstruction * thread has put up a bar for new requests. * Continue immediately if no resync is active currently. */ |
0a27ec96b [PATCH] md: impro... |
939 |
wait_barrier(conf); |
1da177e4c Linux-2.6.12-rc2 |
940 |
|
1da177e4c Linux-2.6.12-rc2 |
941 942 943 944 945 946 947 |
r10_bio = mempool_alloc(conf->r10bio_pool, GFP_NOIO); r10_bio->master_bio = bio; r10_bio->sectors = bio->bi_size >> 9; r10_bio->mddev = mddev; r10_bio->sector = bio->bi_sector; |
6cce3b23f [PATCH] md: write... |
948 |
r10_bio->state = 0; |
1da177e4c Linux-2.6.12-rc2 |
949 |
|
856e08e23 md/raid10: avoid ... |
950 951 952 953 954 955 956 957 958 |
/* We might need to issue multiple reads to different * devices if there are bad blocks around, so we keep * track of the number of reads in bio->bi_phys_segments. * If this is 0, there is only one r10_bio and no locking * will be needed when the request completes. If it is * non-zero, then it is the number of not-completed requests. */ bio->bi_phys_segments = 0; clear_bit(BIO_SEG_VALID, &bio->bi_flags); |
a362357b6 [BLOCK] Unify the... |
959 |
if (rw == READ) { |
1da177e4c Linux-2.6.12-rc2 |
960 961 962 |
/* * read balancing logic: */ |
96c3fd1f3 md/raid10: chang... |
963 |
struct md_rdev *rdev; |
856e08e23 md/raid10: avoid ... |
964 965 966 |
int slot; read_again: |
96c3fd1f3 md/raid10: chang... |
967 968 |
rdev = read_balance(conf, r10_bio, &max_sectors); if (!rdev) { |
1da177e4c Linux-2.6.12-rc2 |
969 |
raid_end_bio_io(r10_bio); |
5a7bbad27 block: remove sup... |
970 |
return; |
1da177e4c Linux-2.6.12-rc2 |
971 |
} |
96c3fd1f3 md/raid10: chang... |
972 |
slot = r10_bio->read_slot; |
1da177e4c Linux-2.6.12-rc2 |
973 |
|
a167f6632 md: use separate ... |
974 |
read_bio = bio_clone_mddev(bio, GFP_NOIO, mddev); |
856e08e23 md/raid10: avoid ... |
975 976 |
md_trim_bio(read_bio, r10_bio->sector - bio->bi_sector, max_sectors); |
1da177e4c Linux-2.6.12-rc2 |
977 978 |
r10_bio->devs[slot].bio = read_bio; |
abbf098e6 md/raid10: prefer... |
979 |
r10_bio->devs[slot].rdev = rdev; |
1da177e4c Linux-2.6.12-rc2 |
980 981 |
read_bio->bi_sector = r10_bio->devs[slot].addr + |
96c3fd1f3 md/raid10: chang... |
982 983 |
rdev->data_offset; read_bio->bi_bdev = rdev->bdev; |
1da177e4c Linux-2.6.12-rc2 |
984 |
read_bio->bi_end_io = raid10_end_read_request; |
7b6d91dae block: unify flag... |
985 |
read_bio->bi_rw = READ | do_sync; |
1da177e4c Linux-2.6.12-rc2 |
986 |
read_bio->bi_private = r10_bio; |
856e08e23 md/raid10: avoid ... |
987 988 989 990 |
if (max_sectors < r10_bio->sectors) { /* Could not read all from this device, so we will * need another r10_bio. */ |
856e08e23 md/raid10: avoid ... |
991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 |
sectors_handled = (r10_bio->sectors + max_sectors - bio->bi_sector); r10_bio->sectors = max_sectors; spin_lock_irq(&conf->device_lock); if (bio->bi_phys_segments == 0) bio->bi_phys_segments = 2; else bio->bi_phys_segments++; spin_unlock(&conf->device_lock); /* Cannot call generic_make_request directly * as that will be queued in __generic_make_request * and subsequent mempool_alloc might block * waiting for it. so hand bio over to raid10d. */ reschedule_retry(r10_bio); r10_bio = mempool_alloc(conf->r10bio_pool, GFP_NOIO); r10_bio->master_bio = bio; r10_bio->sectors = ((bio->bi_size >> 9) - sectors_handled); r10_bio->state = 0; r10_bio->mddev = mddev; r10_bio->sector = bio->bi_sector + sectors_handled; goto read_again; } else generic_make_request(read_bio); |
5a7bbad27 block: remove sup... |
1018 |
return; |
1da177e4c Linux-2.6.12-rc2 |
1019 1020 1021 1022 1023 |
} /* * WRITE: */ |
34db0cd60 md: add proper wr... |
1024 1025 1026 1027 1028 |
if (conf->pending_count >= max_queued_requests) { md_wakeup_thread(mddev->thread); wait_event(conf->wait_barrier, conf->pending_count < max_queued_requests); } |
6bfe0b499 md: support block... |
1029 |
/* first select target devices under rcu_lock and |
1da177e4c Linux-2.6.12-rc2 |
1030 1031 |
* inc refcount on their rdev. Record them by setting * bios[x] to bio |
d4432c23b md/raid10: avoid ... |
1032 1033 1034 1035 1036 1037 1038 |
* If there are known/acknowledged bad blocks on any device * on which we have seen a write error, we want to avoid * writing to those blocks. This potentially requires several * writes to write around the bad blocks. Each set of writes * gets its own r10_bio with a set of bios attached. The number * of r10_bios is recored in bio->bi_phys_segments just as with * the read case. |
1da177e4c Linux-2.6.12-rc2 |
1039 |
*/ |
c3b328ac8 md: fix up raid1/... |
1040 |
plugged = mddev_check_plugged(mddev); |
69335ef3b md/raid10: prepar... |
1041 |
r10_bio->read_slot = -1; /* make sure repl_bio gets freed */ |
1da177e4c Linux-2.6.12-rc2 |
1042 |
raid10_find_phys(conf, r10_bio); |
d4432c23b md/raid10: avoid ... |
1043 |
retry_write: |
cb6969e8c misc: fix integer... |
1044 |
blocked_rdev = NULL; |
1da177e4c Linux-2.6.12-rc2 |
1045 |
rcu_read_lock(); |
d4432c23b md/raid10: avoid ... |
1046 |
max_sectors = r10_bio->sectors; |
1da177e4c Linux-2.6.12-rc2 |
1047 1048 |
for (i = 0; i < conf->copies; i++) { int d = r10_bio->devs[i].devnum; |
3cb030020 md: removing type... |
1049 |
struct md_rdev *rdev = rcu_dereference(conf->mirrors[d].rdev); |
475b0321a md/raid10: writes... |
1050 1051 |
struct md_rdev *rrdev = rcu_dereference( conf->mirrors[d].replacement); |
4ca40c2ce md/raid10: Allow ... |
1052 1053 |
if (rdev == rrdev) rrdev = NULL; |
6bfe0b499 md: support block... |
1054 1055 1056 1057 1058 |
if (rdev && unlikely(test_bit(Blocked, &rdev->flags))) { atomic_inc(&rdev->nr_pending); blocked_rdev = rdev; break; } |
475b0321a md/raid10: writes... |
1059 1060 1061 1062 1063 1064 1065 |
if (rrdev && unlikely(test_bit(Blocked, &rrdev->flags))) { atomic_inc(&rrdev->nr_pending); blocked_rdev = rrdev; break; } if (rrdev && test_bit(Faulty, &rrdev->flags)) rrdev = NULL; |
d4432c23b md/raid10: avoid ... |
1066 |
r10_bio->devs[i].bio = NULL; |
475b0321a md/raid10: writes... |
1067 |
r10_bio->devs[i].repl_bio = NULL; |
d4432c23b md/raid10: avoid ... |
1068 |
if (!rdev || test_bit(Faulty, &rdev->flags)) { |
6cce3b23f [PATCH] md: write... |
1069 |
set_bit(R10BIO_Degraded, &r10_bio->state); |
d4432c23b md/raid10: avoid ... |
1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 |
continue; } if (test_bit(WriteErrorSeen, &rdev->flags)) { sector_t first_bad; sector_t dev_sector = r10_bio->devs[i].addr; int bad_sectors; int is_bad; is_bad = is_badblock(rdev, dev_sector, max_sectors, &first_bad, &bad_sectors); if (is_bad < 0) { /* Mustn't write here until the bad block * is acknowledged */ atomic_inc(&rdev->nr_pending); set_bit(BlockedBadBlocks, &rdev->flags); blocked_rdev = rdev; break; } if (is_bad && first_bad <= dev_sector) { /* Cannot write here at all */ bad_sectors -= (dev_sector - first_bad); if (bad_sectors < max_sectors) /* Mustn't write more than bad_sectors * to other devices yet */ max_sectors = bad_sectors; /* We don't set R10BIO_Degraded as that * only applies if the disk is missing, * so it might be re-added, and we want to * know to recover this chunk. * In this case the device is here, and the * fact that this chunk is not in-sync is * recorded in the bad block log. */ continue; } if (is_bad) { int good_sectors = first_bad - dev_sector; if (good_sectors < max_sectors) max_sectors = good_sectors; } |
6cce3b23f [PATCH] md: write... |
1113 |
} |
d4432c23b md/raid10: avoid ... |
1114 1115 |
r10_bio->devs[i].bio = bio; atomic_inc(&rdev->nr_pending); |
475b0321a md/raid10: writes... |
1116 1117 1118 1119 |
if (rrdev) { r10_bio->devs[i].repl_bio = bio; atomic_inc(&rrdev->nr_pending); } |
1da177e4c Linux-2.6.12-rc2 |
1120 1121 |
} rcu_read_unlock(); |
6bfe0b499 md: support block... |
1122 1123 1124 1125 |
if (unlikely(blocked_rdev)) { /* Have to wait for this device to get unblocked, then retry */ int j; int d; |
475b0321a md/raid10: writes... |
1126 |
for (j = 0; j < i; j++) { |
6bfe0b499 md: support block... |
1127 1128 1129 1130 |
if (r10_bio->devs[j].bio) { d = r10_bio->devs[j].devnum; rdev_dec_pending(conf->mirrors[d].rdev, mddev); } |
475b0321a md/raid10: writes... |
1131 |
if (r10_bio->devs[j].repl_bio) { |
4ca40c2ce md/raid10: Allow ... |
1132 |
struct md_rdev *rdev; |
475b0321a md/raid10: writes... |
1133 |
d = r10_bio->devs[j].devnum; |
4ca40c2ce md/raid10: Allow ... |
1134 1135 1136 1137 1138 1139 1140 |
rdev = conf->mirrors[d].replacement; if (!rdev) { /* Race with remove_disk */ smp_mb(); rdev = conf->mirrors[d].rdev; } rdev_dec_pending(rdev, mddev); |
475b0321a md/raid10: writes... |
1141 1142 |
} } |
6bfe0b499 md: support block... |
1143 1144 1145 1146 1147 |
allow_barrier(conf); md_wait_for_blocked_rdev(blocked_rdev, mddev); wait_barrier(conf); goto retry_write; } |
d4432c23b md/raid10: avoid ... |
1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 |
if (max_sectors < r10_bio->sectors) { /* We are splitting this into multiple parts, so * we need to prepare for allocating another r10_bio. */ r10_bio->sectors = max_sectors; spin_lock_irq(&conf->device_lock); if (bio->bi_phys_segments == 0) bio->bi_phys_segments = 2; else bio->bi_phys_segments++; spin_unlock_irq(&conf->device_lock); } sectors_handled = r10_bio->sector + max_sectors - bio->bi_sector; |
4e78064f4 md: Fix possible ... |
1161 |
atomic_set(&r10_bio->remaining, 1); |
d4432c23b md/raid10: avoid ... |
1162 |
bitmap_startwrite(mddev->bitmap, r10_bio->sector, r10_bio->sectors, 0); |
06d91a5fe [PATCH] md: impro... |
1163 |
|
1da177e4c Linux-2.6.12-rc2 |
1164 1165 1166 1167 1168 |
for (i = 0; i < conf->copies; i++) { struct bio *mbio; int d = r10_bio->devs[i].devnum; if (!r10_bio->devs[i].bio) continue; |
a167f6632 md: use separate ... |
1169 |
mbio = bio_clone_mddev(bio, GFP_NOIO, mddev); |
d4432c23b md/raid10: avoid ... |
1170 1171 |
md_trim_bio(mbio, r10_bio->sector - bio->bi_sector, max_sectors); |
1da177e4c Linux-2.6.12-rc2 |
1172 |
r10_bio->devs[i].bio = mbio; |
d4432c23b md/raid10: avoid ... |
1173 1174 |
mbio->bi_sector = (r10_bio->devs[i].addr+ conf->mirrors[d].rdev->data_offset); |
1da177e4c Linux-2.6.12-rc2 |
1175 1176 |
mbio->bi_bdev = conf->mirrors[d].rdev->bdev; mbio->bi_end_io = raid10_end_write_request; |
e9c7469bb md: implment REQ_... |
1177 |
mbio->bi_rw = WRITE | do_sync | do_fua; |
1da177e4c Linux-2.6.12-rc2 |
1178 1179 1180 |
mbio->bi_private = r10_bio; atomic_inc(&r10_bio->remaining); |
4e78064f4 md: Fix possible ... |
1181 1182 |
spin_lock_irqsave(&conf->device_lock, flags); bio_list_add(&conf->pending_bio_list, mbio); |
34db0cd60 md: add proper wr... |
1183 |
conf->pending_count++; |
4e78064f4 md: Fix possible ... |
1184 |
spin_unlock_irqrestore(&conf->device_lock, flags); |
475b0321a md/raid10: writes... |
1185 1186 1187 1188 1189 1190 1191 1192 |
if (!r10_bio->devs[i].repl_bio) continue; mbio = bio_clone_mddev(bio, GFP_NOIO, mddev); md_trim_bio(mbio, r10_bio->sector - bio->bi_sector, max_sectors); r10_bio->devs[i].repl_bio = mbio; |
4ca40c2ce md/raid10: Allow ... |
1193 1194 1195 1196 |
/* We are actively writing to the original device * so it cannot disappear, so the replacement cannot * become NULL here */ |
475b0321a md/raid10: writes... |
1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 |
mbio->bi_sector = (r10_bio->devs[i].addr+ conf->mirrors[d].replacement->data_offset); mbio->bi_bdev = conf->mirrors[d].replacement->bdev; mbio->bi_end_io = raid10_end_write_request; mbio->bi_rw = WRITE | do_sync | do_fua; mbio->bi_private = r10_bio; atomic_inc(&r10_bio->remaining); spin_lock_irqsave(&conf->device_lock, flags); bio_list_add(&conf->pending_bio_list, mbio); conf->pending_count++; spin_unlock_irqrestore(&conf->device_lock, flags); |
1da177e4c Linux-2.6.12-rc2 |
1209 |
} |
079fa166a md/raid1,10: Remo... |
1210 1211 1212 |
/* Don't remove the bias on 'remaining' (one_write_done) until * after checking if we need to go around again. */ |
a35e63efa md: fix deadlock ... |
1213 |
|
d4432c23b md/raid10: avoid ... |
1214 |
if (sectors_handled < (bio->bi_size >> 9)) { |
079fa166a md/raid1,10: Remo... |
1215 |
one_write_done(r10_bio); |
5e5702898 md/raid10: Handle... |
1216 |
/* We need another r10_bio. It has already been counted |
d4432c23b md/raid10: avoid ... |
1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 |
* in bio->bi_phys_segments. */ r10_bio = mempool_alloc(conf->r10bio_pool, GFP_NOIO); r10_bio->master_bio = bio; r10_bio->sectors = (bio->bi_size >> 9) - sectors_handled; r10_bio->mddev = mddev; r10_bio->sector = bio->bi_sector + sectors_handled; r10_bio->state = 0; goto retry_write; } |
079fa166a md/raid1,10: Remo... |
1229 1230 1231 1232 |
one_write_done(r10_bio); /* In case raid10d snuck in to freeze_array */ wake_up(&conf->wait_barrier); |
d4432c23b md/raid10: avoid ... |
1233 |
|
c3b328ac8 md: fix up raid1/... |
1234 |
if (do_sync || !mddev->bitmap || !plugged) |
e3881a681 [PATCH] md: pass ... |
1235 |
md_wakeup_thread(mddev->thread); |
1da177e4c Linux-2.6.12-rc2 |
1236 |
} |
fd01b88c7 md: remove typede... |
1237 |
static void status(struct seq_file *seq, struct mddev *mddev) |
1da177e4c Linux-2.6.12-rc2 |
1238 |
{ |
e879a8793 md/raid10: typede... |
1239 |
struct r10conf *conf = mddev->private; |
1da177e4c Linux-2.6.12-rc2 |
1240 1241 1242 |
int i; if (conf->near_copies < conf->raid_disks) |
9d8f03636 md: Make mddev->c... |
1243 |
seq_printf(seq, " %dK chunks", mddev->chunk_sectors / 2); |
1da177e4c Linux-2.6.12-rc2 |
1244 1245 |
if (conf->near_copies > 1) seq_printf(seq, " %d near-copies", conf->near_copies); |
c93983bf5 [PATCH] md: suppo... |
1246 1247 1248 1249 1250 1251 |
if (conf->far_copies > 1) { if (conf->far_offset) seq_printf(seq, " %d offset-copies", conf->far_copies); else seq_printf(seq, " %d far-copies", conf->far_copies); } |
1da177e4c Linux-2.6.12-rc2 |
1252 |
seq_printf(seq, " [%d/%d] [", conf->raid_disks, |
76186dd8b [PATCH] md: remov... |
1253 |
conf->raid_disks - mddev->degraded); |
1da177e4c Linux-2.6.12-rc2 |
1254 1255 1256 |
for (i = 0; i < conf->raid_disks; i++) seq_printf(seq, "%s", conf->mirrors[i].rdev && |
b2d444d7a [PATCH] md: conve... |
1257 |
test_bit(In_sync, &conf->mirrors[i].rdev->flags) ? "U" : "_"); |
1da177e4c Linux-2.6.12-rc2 |
1258 1259 |
seq_printf(seq, "]"); } |
700c72138 md/raid10: Improv... |
1260 1261 1262 1263 1264 |
/* check if there are enough drives for * every block to appear on atleast one. * Don't consider the device numbered 'ignore' * as we might be about to remove it. */ |
e879a8793 md/raid10: typede... |
1265 |
static int enough(struct r10conf *conf, int ignore) |
700c72138 md/raid10: Improv... |
1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 |
{ int first = 0; do { int n = conf->copies; int cnt = 0; while (n--) { if (conf->mirrors[first].rdev && first != ignore) cnt++; first = (first+1) % conf->raid_disks; } if (cnt == 0) return 0; } while (first != 0); return 1; } |
fd01b88c7 md: remove typede... |
1283 |
static void error(struct mddev *mddev, struct md_rdev *rdev) |
1da177e4c Linux-2.6.12-rc2 |
1284 1285 |
{ char b[BDEVNAME_SIZE]; |
e879a8793 md/raid10: typede... |
1286 |
struct r10conf *conf = mddev->private; |
1da177e4c Linux-2.6.12-rc2 |
1287 1288 1289 1290 1291 1292 1293 |
/* * If it is not operational, then we have already marked it as dead * else if it is the last working disks, ignore the error, let the * next level up know. * else mark the drive as failed */ |
b2d444d7a [PATCH] md: conve... |
1294 |
if (test_bit(In_sync, &rdev->flags) |
700c72138 md/raid10: Improv... |
1295 |
&& !enough(conf, rdev->raid_disk)) |
1da177e4c Linux-2.6.12-rc2 |
1296 1297 |
/* * Don't fail the drive, just return an IO error. |
1da177e4c Linux-2.6.12-rc2 |
1298 1299 |
*/ return; |
c04be0aa8 [PATCH] md: Impro... |
1300 1301 1302 |
if (test_and_clear_bit(In_sync, &rdev->flags)) { unsigned long flags; spin_lock_irqsave(&conf->device_lock, flags); |
1da177e4c Linux-2.6.12-rc2 |
1303 |
mddev->degraded++; |
c04be0aa8 [PATCH] md: Impro... |
1304 |
spin_unlock_irqrestore(&conf->device_lock, flags); |
1da177e4c Linux-2.6.12-rc2 |
1305 1306 1307 |
/* * if recovery is running, make sure it aborts. */ |
dfc706450 md: restart recov... |
1308 |
set_bit(MD_RECOVERY_INTR, &mddev->recovery); |
1da177e4c Linux-2.6.12-rc2 |
1309 |
} |
de393cdea md: make it easie... |
1310 |
set_bit(Blocked, &rdev->flags); |
b2d444d7a [PATCH] md: conve... |
1311 |
set_bit(Faulty, &rdev->flags); |
850b2b420 [PATCH] md: repla... |
1312 |
set_bit(MD_CHANGE_DEVS, &mddev->flags); |
067032bc6 md: Fix single pr... |
1313 1314 1315 1316 1317 |
printk(KERN_ALERT "md/raid10:%s: Disk failure on %s, disabling device. " "md/raid10:%s: Operation continuing on %d devices. ", |
128595ed6 md/raid10: tidy u... |
1318 1319 |
mdname(mddev), bdevname(rdev->bdev, b), mdname(mddev), conf->raid_disks - mddev->degraded); |
1da177e4c Linux-2.6.12-rc2 |
1320 |
} |
e879a8793 md/raid10: typede... |
1321 |
static void print_conf(struct r10conf *conf) |
1da177e4c Linux-2.6.12-rc2 |
1322 1323 |
{ int i; |
0f6d02d58 md: remove typede... |
1324 |
struct mirror_info *tmp; |
1da177e4c Linux-2.6.12-rc2 |
1325 |
|
128595ed6 md/raid10: tidy u... |
1326 1327 |
printk(KERN_DEBUG "RAID10 conf printout: "); |
1da177e4c Linux-2.6.12-rc2 |
1328 |
if (!conf) { |
128595ed6 md/raid10: tidy u... |
1329 1330 |
printk(KERN_DEBUG "(!conf) "); |
1da177e4c Linux-2.6.12-rc2 |
1331 1332 |
return; } |
128595ed6 md/raid10: tidy u... |
1333 1334 |
printk(KERN_DEBUG " --- wd:%d rd:%d ", conf->raid_disks - conf->mddev->degraded, |
1da177e4c Linux-2.6.12-rc2 |
1335 1336 1337 1338 1339 1340 |
conf->raid_disks); for (i = 0; i < conf->raid_disks; i++) { char b[BDEVNAME_SIZE]; tmp = conf->mirrors + i; if (tmp->rdev) |
128595ed6 md/raid10: tidy u... |
1341 1342 |
printk(KERN_DEBUG " disk %d, wo:%d, o:%d, dev:%s ", |
b2d444d7a [PATCH] md: conve... |
1343 1344 |
i, !test_bit(In_sync, &tmp->rdev->flags), !test_bit(Faulty, &tmp->rdev->flags), |
1da177e4c Linux-2.6.12-rc2 |
1345 1346 1347 |
bdevname(tmp->rdev->bdev,b)); } } |
e879a8793 md/raid10: typede... |
1348 |
static void close_sync(struct r10conf *conf) |
1da177e4c Linux-2.6.12-rc2 |
1349 |
{ |
0a27ec96b [PATCH] md: impro... |
1350 1351 |
wait_barrier(conf); allow_barrier(conf); |
1da177e4c Linux-2.6.12-rc2 |
1352 1353 1354 1355 |
mempool_destroy(conf->r10buf_pool); conf->r10buf_pool = NULL; } |
fd01b88c7 md: remove typede... |
1356 |
static int raid10_spare_active(struct mddev *mddev) |
1da177e4c Linux-2.6.12-rc2 |
1357 1358 |
{ int i; |
e879a8793 md/raid10: typede... |
1359 |
struct r10conf *conf = mddev->private; |
0f6d02d58 md: remove typede... |
1360 |
struct mirror_info *tmp; |
6b9656205 md: provide appro... |
1361 1362 |
int count = 0; unsigned long flags; |
1da177e4c Linux-2.6.12-rc2 |
1363 1364 1365 1366 1367 1368 1369 |
/* * Find all non-in_sync disks within the RAID10 configuration * and mark them in_sync */ for (i = 0; i < conf->raid_disks; i++) { tmp = conf->mirrors + i; |
4ca40c2ce md/raid10: Allow ... |
1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390 |
if (tmp->replacement && tmp->replacement->recovery_offset == MaxSector && !test_bit(Faulty, &tmp->replacement->flags) && !test_and_set_bit(In_sync, &tmp->replacement->flags)) { /* Replacement has just become active */ if (!tmp->rdev || !test_and_clear_bit(In_sync, &tmp->rdev->flags)) count++; if (tmp->rdev) { /* Replaced device not technically faulty, * but we need to be sure it gets removed * and never re-added. */ set_bit(Faulty, &tmp->rdev->flags); sysfs_notify_dirent_safe( tmp->rdev->sysfs_state); } sysfs_notify_dirent_safe(tmp->replacement->sysfs_state); } else if (tmp->rdev && !test_bit(Faulty, &tmp->rdev->flags) && !test_and_set_bit(In_sync, &tmp->rdev->flags)) { |
6b9656205 md: provide appro... |
1391 |
count++; |
e6ffbcb6c md: Notify sysfs ... |
1392 |
sysfs_notify_dirent(tmp->rdev->sysfs_state); |
1da177e4c Linux-2.6.12-rc2 |
1393 1394 |
} } |
6b9656205 md: provide appro... |
1395 1396 1397 |
spin_lock_irqsave(&conf->device_lock, flags); mddev->degraded -= count; spin_unlock_irqrestore(&conf->device_lock, flags); |
1da177e4c Linux-2.6.12-rc2 |
1398 1399 |
print_conf(conf); |
6b9656205 md: provide appro... |
1400 |
return count; |
1da177e4c Linux-2.6.12-rc2 |
1401 |
} |
fd01b88c7 md: remove typede... |
1402 |
static int raid10_add_disk(struct mddev *mddev, struct md_rdev *rdev) |
1da177e4c Linux-2.6.12-rc2 |
1403 |
{ |
e879a8793 md/raid10: typede... |
1404 |
struct r10conf *conf = mddev->private; |
199050ea1 rationalise retur... |
1405 |
int err = -EEXIST; |
1da177e4c Linux-2.6.12-rc2 |
1406 |
int mirror; |
6c2fce2ef Support adding a ... |
1407 |
int first = 0; |
84707f38e md: don't use mdd... |
1408 |
int last = conf->raid_disks - 1; |
1da177e4c Linux-2.6.12-rc2 |
1409 1410 1411 1412 1413 |
if (mddev->recovery_cp < MaxSector) /* only hot-add to in-sync arrays, as recovery is * very different from resync */ |
199050ea1 rationalise retur... |
1414 |
return -EBUSY; |
700c72138 md/raid10: Improv... |
1415 |
if (!enough(conf, -1)) |
199050ea1 rationalise retur... |
1416 |
return -EINVAL; |
1da177e4c Linux-2.6.12-rc2 |
1417 |
|
a53a6c857 md: fix bug in ra... |
1418 |
if (rdev->raid_disk >= 0) |
6c2fce2ef Support adding a ... |
1419 |
first = last = rdev->raid_disk; |
1da177e4c Linux-2.6.12-rc2 |
1420 |
|
2c4193df3 md/raid10: get ri... |
1421 |
if (rdev->saved_raid_disk >= first && |
6cce3b23f [PATCH] md: write... |
1422 1423 1424 |
conf->mirrors[rdev->saved_raid_disk].rdev == NULL) mirror = rdev->saved_raid_disk; else |
6c2fce2ef Support adding a ... |
1425 |
mirror = first; |
2bb77736a md/raid10: Make u... |
1426 |
for ( ; mirror <= last ; mirror++) { |
0f6d02d58 md: remove typede... |
1427 |
struct mirror_info *p = &conf->mirrors[mirror]; |
2bb77736a md/raid10: Make u... |
1428 1429 |
if (p->recovery_disabled == mddev->recovery_disabled) continue; |
b7044d41b md/raid10: If the... |
1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440 1441 1442 1443 1444 1445 1446 1447 1448 |
if (p->rdev) { if (!test_bit(WantReplacement, &p->rdev->flags) || p->replacement != NULL) continue; clear_bit(In_sync, &rdev->flags); set_bit(Replacement, &rdev->flags); rdev->raid_disk = mirror; err = 0; disk_stack_limits(mddev->gendisk, rdev->bdev, rdev->data_offset << 9); if (rdev->bdev->bd_disk->queue->merge_bvec_fn) { blk_queue_max_segments(mddev->queue, 1); blk_queue_segment_boundary(mddev->queue, PAGE_CACHE_SIZE - 1); } conf->fullsync = 1; rcu_assign_pointer(p->replacement, rdev); break; } |
1da177e4c Linux-2.6.12-rc2 |
1449 |
|
2bb77736a md/raid10: Make u... |
1450 1451 1452 1453 1454 1455 1456 1457 1458 1459 1460 1461 |
disk_stack_limits(mddev->gendisk, rdev->bdev, rdev->data_offset << 9); /* as we don't honour merge_bvec_fn, we must * never risk violating it, so limit * ->max_segments to one lying with a single * page, as a one page request is never in * violation. */ if (rdev->bdev->bd_disk->queue->merge_bvec_fn) { blk_queue_max_segments(mddev->queue, 1); blk_queue_segment_boundary(mddev->queue, PAGE_CACHE_SIZE - 1); |
1da177e4c Linux-2.6.12-rc2 |
1462 |
} |
2bb77736a md/raid10: Make u... |
1463 |
p->head_position = 0; |
d890fa2b0 md: Fix some bugs... |
1464 |
p->recovery_disabled = mddev->recovery_disabled - 1; |
2bb77736a md/raid10: Make u... |
1465 1466 1467 1468 1469 1470 1471 |
rdev->raid_disk = mirror; err = 0; if (rdev->saved_raid_disk != mirror) conf->fullsync = 1; rcu_assign_pointer(p->rdev, rdev); break; } |
ac5e7113e md: Push down dat... |
1472 |
md_integrity_add_rdev(rdev, mddev); |
1da177e4c Linux-2.6.12-rc2 |
1473 |
print_conf(conf); |
199050ea1 rationalise retur... |
1474 |
return err; |
1da177e4c Linux-2.6.12-rc2 |
1475 |
} |
b8321b68d md: change hot_re... |
1476 |
static int raid10_remove_disk(struct mddev *mddev, struct md_rdev *rdev) |
1da177e4c Linux-2.6.12-rc2 |
1477 |
{ |
e879a8793 md/raid10: typede... |
1478 |
struct r10conf *conf = mddev->private; |
1da177e4c Linux-2.6.12-rc2 |
1479 |
int err = 0; |
b8321b68d md: change hot_re... |
1480 |
int number = rdev->raid_disk; |
c8ab903ea md/raid10: allow ... |
1481 1482 |
struct md_rdev **rdevp; struct mirror_info *p = conf->mirrors + number; |
1da177e4c Linux-2.6.12-rc2 |
1483 1484 |
print_conf(conf); |
c8ab903ea md/raid10: allow ... |
1485 1486 1487 1488 1489 1490 1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 |
if (rdev == p->rdev) rdevp = &p->rdev; else if (rdev == p->replacement) rdevp = &p->replacement; else return 0; if (test_bit(In_sync, &rdev->flags) || atomic_read(&rdev->nr_pending)) { err = -EBUSY; goto abort; } /* Only remove faulty devices if recovery * is not possible. */ if (!test_bit(Faulty, &rdev->flags) && mddev->recovery_disabled != p->recovery_disabled && |
4ca40c2ce md/raid10: Allow ... |
1502 |
(!p->replacement || p->replacement == rdev) && |
c8ab903ea md/raid10: allow ... |
1503 1504 1505 |
enough(conf, -1)) { err = -EBUSY; goto abort; |
1da177e4c Linux-2.6.12-rc2 |
1506 |
} |
c8ab903ea md/raid10: allow ... |
1507 1508 1509 1510 1511 1512 1513 |
*rdevp = NULL; synchronize_rcu(); if (atomic_read(&rdev->nr_pending)) { /* lost the race, try later */ err = -EBUSY; *rdevp = rdev; goto abort; |
4ca40c2ce md/raid10: Allow ... |
1514 1515 1516 1517 1518 1519 1520 1521 1522 1523 1524 1525 1526 1527 |
} else if (p->replacement) { /* We must have just cleared 'rdev' */ p->rdev = p->replacement; clear_bit(Replacement, &p->replacement->flags); smp_mb(); /* Make sure other CPUs may see both as identical * but will never see neither -- if they are careful. */ p->replacement = NULL; clear_bit(WantReplacement, &rdev->flags); } else /* We might have just remove the Replacement as faulty * Clear the flag just in case */ clear_bit(WantReplacement, &rdev->flags); |
c8ab903ea md/raid10: allow ... |
1528 |
err = md_integrity_register(mddev); |
1da177e4c Linux-2.6.12-rc2 |
1529 1530 1531 1532 1533 |
abort: print_conf(conf); return err; } |
6712ecf8f Drop 'size' argum... |
1534 |
static void end_sync_read(struct bio *bio, int error) |
1da177e4c Linux-2.6.12-rc2 |
1535 |
{ |
9f2c9d12b md: remove typede... |
1536 |
struct r10bio *r10_bio = bio->bi_private; |
e879a8793 md/raid10: typede... |
1537 |
struct r10conf *conf = r10_bio->mddev->private; |
778ca0185 md/raid10: factor... |
1538 |
int d; |
1da177e4c Linux-2.6.12-rc2 |
1539 |
|
69335ef3b md/raid10: prepar... |
1540 |
d = find_bio_disk(conf, r10_bio, bio, NULL, NULL); |
0eb3ff12a [PATCH] md: raid1... |
1541 1542 1543 |
if (test_bit(BIO_UPTODATE, &bio->bi_flags)) set_bit(R10BIO_Uptodate, &r10_bio->state); |
e684e41db md/raid10: simpli... |
1544 1545 1546 1547 |
else /* The write handler will notice the lack of * R10BIO_Uptodate and record any errors etc */ |
4dbcdc751 [PATCH] md: count... |
1548 1549 |
atomic_add(r10_bio->sectors, &conf->mirrors[d].rdev->corrected_errors); |
1da177e4c Linux-2.6.12-rc2 |
1550 1551 1552 1553 |
/* for reconstruct, we always reschedule after a read. * for resync, only after all reads */ |
73d5c38a9 md: avoid races w... |
1554 |
rdev_dec_pending(conf->mirrors[d].rdev, conf->mddev); |
1da177e4c Linux-2.6.12-rc2 |
1555 1556 1557 1558 1559 1560 1561 |
if (test_bit(R10BIO_IsRecover, &r10_bio->state) || atomic_dec_and_test(&r10_bio->remaining)) { /* we have read all the blocks, * do the comparison in process context in raid10d */ reschedule_retry(r10_bio); } |
1da177e4c Linux-2.6.12-rc2 |
1562 |
} |
9f2c9d12b md: remove typede... |
1563 |
static void end_sync_request(struct r10bio *r10_bio) |
1da177e4c Linux-2.6.12-rc2 |
1564 |
{ |
fd01b88c7 md: remove typede... |
1565 |
struct mddev *mddev = r10_bio->mddev; |
dfc706450 md: restart recov... |
1566 |
|
1da177e4c Linux-2.6.12-rc2 |
1567 1568 1569 |
while (atomic_dec_and_test(&r10_bio->remaining)) { if (r10_bio->master_bio == NULL) { /* the primary of several recovery bios */ |
73d5c38a9 md: avoid races w... |
1570 |
sector_t s = r10_bio->sectors; |
1a0b7cd82 md/raid10: record... |
1571 1572 |
if (test_bit(R10BIO_MadeGood, &r10_bio->state) || test_bit(R10BIO_WriteError, &r10_bio->state)) |
749c55e94 md/raid10: clear ... |
1573 1574 1575 |
reschedule_retry(r10_bio); else put_buf(r10_bio); |
73d5c38a9 md: avoid races w... |
1576 |
md_done_sync(mddev, s, 1); |
1da177e4c Linux-2.6.12-rc2 |
1577 1578 |
break; } else { |
9f2c9d12b md: remove typede... |
1579 |
struct r10bio *r10_bio2 = (struct r10bio *)r10_bio->master_bio; |
1a0b7cd82 md/raid10: record... |
1580 1581 |
if (test_bit(R10BIO_MadeGood, &r10_bio->state) || test_bit(R10BIO_WriteError, &r10_bio->state)) |
749c55e94 md/raid10: clear ... |
1582 1583 1584 |
reschedule_retry(r10_bio); else put_buf(r10_bio); |
1da177e4c Linux-2.6.12-rc2 |
1585 1586 1587 |
r10_bio = r10_bio2; } } |
1da177e4c Linux-2.6.12-rc2 |
1588 |
} |
5e5702898 md/raid10: Handle... |
1589 1590 1591 |
static void end_sync_write(struct bio *bio, int error) { int uptodate = test_bit(BIO_UPTODATE, &bio->bi_flags); |
9f2c9d12b md: remove typede... |
1592 |
struct r10bio *r10_bio = bio->bi_private; |
fd01b88c7 md: remove typede... |
1593 |
struct mddev *mddev = r10_bio->mddev; |
e879a8793 md/raid10: typede... |
1594 |
struct r10conf *conf = mddev->private; |
5e5702898 md/raid10: Handle... |
1595 1596 1597 1598 |
int d; sector_t first_bad; int bad_sectors; int slot; |
9ad1aefc8 md/raid10: Handl... |
1599 |
int repl; |
4ca40c2ce md/raid10: Allow ... |
1600 |
struct md_rdev *rdev = NULL; |
5e5702898 md/raid10: Handle... |
1601 |
|
9ad1aefc8 md/raid10: Handl... |
1602 1603 1604 |
d = find_bio_disk(conf, r10_bio, bio, &slot, &repl); if (repl) rdev = conf->mirrors[d].replacement; |
4ca40c2ce md/raid10: Allow ... |
1605 1606 |
if (!rdev) { smp_mb(); |
9ad1aefc8 md/raid10: Handl... |
1607 |
rdev = conf->mirrors[d].rdev; |
4ca40c2ce md/raid10: Allow ... |
1608 |
} |
5e5702898 md/raid10: Handle... |
1609 1610 |
if (!uptodate) { |
9ad1aefc8 md/raid10: Handl... |
1611 1612 1613 1614 |
if (repl) md_error(mddev, rdev); else { set_bit(WriteErrorSeen, &rdev->flags); |
b7044d41b md/raid10: If the... |
1615 1616 1617 |
if (!test_and_set_bit(WantReplacement, &rdev->flags)) set_bit(MD_RECOVERY_NEEDED, &rdev->mddev->recovery); |
9ad1aefc8 md/raid10: Handl... |
1618 1619 1620 |
set_bit(R10BIO_WriteError, &r10_bio->state); } } else if (is_badblock(rdev, |
5e5702898 md/raid10: Handle... |
1621 1622 1623 1624 |
r10_bio->devs[slot].addr, r10_bio->sectors, &first_bad, &bad_sectors)) set_bit(R10BIO_MadeGood, &r10_bio->state); |
9ad1aefc8 md/raid10: Handl... |
1625 |
rdev_dec_pending(rdev, mddev); |
5e5702898 md/raid10: Handle... |
1626 1627 1628 |
end_sync_request(r10_bio); } |
1da177e4c Linux-2.6.12-rc2 |
1629 1630 1631 1632 1633 1634 1635 1636 1637 1638 1639 1640 1641 1642 1643 1644 |
/* * Note: sync and recover and handled very differently for raid10 * This code is for resync. * For resync, we read through virtual addresses and read all blocks. * If there is any error, we schedule a write. The lowest numbered * drive is authoritative. * However requests come for physical address, so we need to map. * For every physical address there are raid_disks/copies virtual addresses, * which is always are least one, but is not necessarly an integer. * This means that a physical address can span multiple chunks, so we may * have to submit multiple io requests for a single sync request. */ /* * We check if all blocks are in-sync and only write to blocks that * aren't in sync */ |
9f2c9d12b md: remove typede... |
1645 |
static void sync_request_write(struct mddev *mddev, struct r10bio *r10_bio) |
1da177e4c Linux-2.6.12-rc2 |
1646 |
{ |
e879a8793 md/raid10: typede... |
1647 |
struct r10conf *conf = mddev->private; |
1da177e4c Linux-2.6.12-rc2 |
1648 1649 1650 1651 1652 1653 1654 1655 1656 1657 1658 1659 1660 1661 1662 1663 1664 |
int i, first; struct bio *tbio, *fbio; atomic_set(&r10_bio->remaining, 1); /* find the first device with a block */ for (i=0; i<conf->copies; i++) if (test_bit(BIO_UPTODATE, &r10_bio->devs[i].bio->bi_flags)) break; if (i == conf->copies) goto done; first = i; fbio = r10_bio->devs[i].bio; /* now find blocks with errors */ |
0eb3ff12a [PATCH] md: raid1... |
1665 1666 1667 |
for (i=0 ; i < conf->copies ; i++) { int j, d; int vcnt = r10_bio->sectors >> (PAGE_SHIFT-9); |
1da177e4c Linux-2.6.12-rc2 |
1668 |
|
1da177e4c Linux-2.6.12-rc2 |
1669 |
tbio = r10_bio->devs[i].bio; |
0eb3ff12a [PATCH] md: raid1... |
1670 1671 1672 1673 |
if (tbio->bi_end_io != end_sync_read) continue; if (i == first) |
1da177e4c Linux-2.6.12-rc2 |
1674 |
continue; |
0eb3ff12a [PATCH] md: raid1... |
1675 1676 1677 1678 1679 1680 1681 1682 1683 1684 1685 1686 1687 |
if (test_bit(BIO_UPTODATE, &r10_bio->devs[i].bio->bi_flags)) { /* We know that the bi_io_vec layout is the same for * both 'first' and 'i', so we just compare them. * All vec entries are PAGE_SIZE; */ for (j = 0; j < vcnt; j++) if (memcmp(page_address(fbio->bi_io_vec[j].bv_page), page_address(tbio->bi_io_vec[j].bv_page), PAGE_SIZE)) break; if (j == vcnt) continue; mddev->resync_mismatches += r10_bio->sectors; |
f84ee364d md/raid10: attem... |
1688 1689 1690 |
if (test_bit(MD_RECOVERY_CHECK, &mddev->recovery)) /* Don't fix anything. */ continue; |
0eb3ff12a [PATCH] md: raid1... |
1691 |
} |
f84ee364d md/raid10: attem... |
1692 1693 |
/* Ok, we need to write this bio, either to correct an * inconsistency or to correct an unreadable block. |
1da177e4c Linux-2.6.12-rc2 |
1694 1695 1696 1697 1698 1699 1700 |
* First we need to fixup bv_offset, bv_len and * bi_vecs, as the read request might have corrupted these */ tbio->bi_vcnt = vcnt; tbio->bi_size = r10_bio->sectors << 9; tbio->bi_idx = 0; tbio->bi_phys_segments = 0; |
1da177e4c Linux-2.6.12-rc2 |
1701 1702 1703 1704 1705 1706 1707 1708 1709 1710 1711 1712 1713 1714 1715 1716 1717 1718 1719 1720 1721 1722 1723 1724 1725 1726 |
tbio->bi_flags &= ~(BIO_POOL_MASK - 1); tbio->bi_flags |= 1 << BIO_UPTODATE; tbio->bi_next = NULL; tbio->bi_rw = WRITE; tbio->bi_private = r10_bio; tbio->bi_sector = r10_bio->devs[i].addr; for (j=0; j < vcnt ; j++) { tbio->bi_io_vec[j].bv_offset = 0; tbio->bi_io_vec[j].bv_len = PAGE_SIZE; memcpy(page_address(tbio->bi_io_vec[j].bv_page), page_address(fbio->bi_io_vec[j].bv_page), PAGE_SIZE); } tbio->bi_end_io = end_sync_write; d = r10_bio->devs[i].devnum; atomic_inc(&conf->mirrors[d].rdev->nr_pending); atomic_inc(&r10_bio->remaining); md_sync_acct(conf->mirrors[d].rdev->bdev, tbio->bi_size >> 9); tbio->bi_sector += conf->mirrors[d].rdev->data_offset; tbio->bi_bdev = conf->mirrors[d].rdev->bdev; generic_make_request(tbio); } |
9ad1aefc8 md/raid10: Handl... |
1727 1728 1729 1730 1731 1732 1733 1734 1735 1736 1737 1738 1739 1740 1741 1742 1743 1744 1745 1746 1747 1748 |
/* Now write out to any replacement devices * that are active */ for (i = 0; i < conf->copies; i++) { int j, d; int vcnt = r10_bio->sectors >> (PAGE_SHIFT-9); tbio = r10_bio->devs[i].repl_bio; if (!tbio || !tbio->bi_end_io) continue; if (r10_bio->devs[i].bio->bi_end_io != end_sync_write && r10_bio->devs[i].bio != fbio) for (j = 0; j < vcnt; j++) memcpy(page_address(tbio->bi_io_vec[j].bv_page), page_address(fbio->bi_io_vec[j].bv_page), PAGE_SIZE); d = r10_bio->devs[i].devnum; atomic_inc(&r10_bio->remaining); md_sync_acct(conf->mirrors[d].replacement->bdev, tbio->bi_size >> 9); generic_make_request(tbio); } |
1da177e4c Linux-2.6.12-rc2 |
1749 1750 1751 1752 1753 1754 1755 1756 1757 1758 1759 1760 1761 1762 1763 1764 1765 |
done: if (atomic_dec_and_test(&r10_bio->remaining)) { md_done_sync(mddev, r10_bio->sectors, 1); put_buf(r10_bio); } } /* * Now for the recovery code. * Recovery happens across physical sectors. * We recover all non-is_sync drives by finding the virtual address of * each, and then choose a working drive that also has that virt address. * There is a separate r10_bio for each non-in_sync drive. * Only the first two slots are in use. The first for reading, * The second for writing. * */ |
9f2c9d12b md: remove typede... |
1766 |
static void fix_recovery_read_error(struct r10bio *r10_bio) |
5e5702898 md/raid10: Handle... |
1767 1768 1769 1770 1771 1772 1773 1774 |
{ /* We got a read error during recovery. * We repeat the read in smaller page-sized sections. * If a read succeeds, write it to the new device or record * a bad block if we cannot. * If a read fails, record a bad block on both old and * new devices. */ |
fd01b88c7 md: remove typede... |
1775 |
struct mddev *mddev = r10_bio->mddev; |
e879a8793 md/raid10: typede... |
1776 |
struct r10conf *conf = mddev->private; |
5e5702898 md/raid10: Handle... |
1777 1778 1779 1780 1781 1782 1783 1784 1785 |
struct bio *bio = r10_bio->devs[0].bio; sector_t sect = 0; int sectors = r10_bio->sectors; int idx = 0; int dr = r10_bio->devs[0].devnum; int dw = r10_bio->devs[1].devnum; while (sectors) { int s = sectors; |
3cb030020 md: removing type... |
1786 |
struct md_rdev *rdev; |
5e5702898 md/raid10: Handle... |
1787 1788 1789 1790 1791 1792 1793 1794 1795 1796 1797 1798 1799 1800 1801 1802 1803 1804 1805 1806 1807 |
sector_t addr; int ok; if (s > (PAGE_SIZE>>9)) s = PAGE_SIZE >> 9; rdev = conf->mirrors[dr].rdev; addr = r10_bio->devs[0].addr + sect, ok = sync_page_io(rdev, addr, s << 9, bio->bi_io_vec[idx].bv_page, READ, false); if (ok) { rdev = conf->mirrors[dw].rdev; addr = r10_bio->devs[1].addr + sect; ok = sync_page_io(rdev, addr, s << 9, bio->bi_io_vec[idx].bv_page, WRITE, false); |
b7044d41b md/raid10: If the... |
1808 |
if (!ok) { |
5e5702898 md/raid10: Handle... |
1809 |
set_bit(WriteErrorSeen, &rdev->flags); |
b7044d41b md/raid10: If the... |
1810 1811 1812 1813 1814 |
if (!test_and_set_bit(WantReplacement, &rdev->flags)) set_bit(MD_RECOVERY_NEEDED, &rdev->mddev->recovery); } |
5e5702898 md/raid10: Handle... |
1815 1816 1817 1818 1819 1820 1821 1822 1823 1824 |
} if (!ok) { /* We don't worry if we cannot set a bad block - * it really is bad so there is no loss in not * recording it yet */ rdev_set_badblocks(rdev, addr, s, 0); if (rdev != conf->mirrors[dw].rdev) { /* need bad block on destination too */ |
3cb030020 md: removing type... |
1825 |
struct md_rdev *rdev2 = conf->mirrors[dw].rdev; |
5e5702898 md/raid10: Handle... |
1826 1827 1828 1829 1830 1831 1832 1833 1834 1835 1836 1837 1838 1839 1840 1841 1842 1843 1844 1845 1846 1847 1848 1849 |
addr = r10_bio->devs[1].addr + sect; ok = rdev_set_badblocks(rdev2, addr, s, 0); if (!ok) { /* just abort the recovery */ printk(KERN_NOTICE "md/raid10:%s: recovery aborted" " due to read error ", mdname(mddev)); conf->mirrors[dw].recovery_disabled = mddev->recovery_disabled; set_bit(MD_RECOVERY_INTR, &mddev->recovery); break; } } } sectors -= s; sect += s; idx++; } } |
1da177e4c Linux-2.6.12-rc2 |
1850 |
|
9f2c9d12b md: remove typede... |
1851 |
static void recovery_request_write(struct mddev *mddev, struct r10bio *r10_bio) |
1da177e4c Linux-2.6.12-rc2 |
1852 |
{ |
e879a8793 md/raid10: typede... |
1853 |
struct r10conf *conf = mddev->private; |
c65060ad4 md/raid10: share ... |
1854 |
int d; |
24afd80d9 md/raid10: handle... |
1855 |
struct bio *wbio, *wbio2; |
1da177e4c Linux-2.6.12-rc2 |
1856 |
|
5e5702898 md/raid10: Handle... |
1857 1858 1859 1860 1861 |
if (!test_bit(R10BIO_Uptodate, &r10_bio->state)) { fix_recovery_read_error(r10_bio); end_sync_request(r10_bio); return; } |
c65060ad4 md/raid10: share ... |
1862 1863 |
/* * share the pages with the first bio |
1da177e4c Linux-2.6.12-rc2 |
1864 1865 |
* and submit the write request */ |
1da177e4c Linux-2.6.12-rc2 |
1866 |
d = r10_bio->devs[1].devnum; |
24afd80d9 md/raid10: handle... |
1867 1868 1869 1870 1871 1872 1873 1874 1875 1876 1877 1878 1879 |
wbio = r10_bio->devs[1].bio; wbio2 = r10_bio->devs[1].repl_bio; if (wbio->bi_end_io) { atomic_inc(&conf->mirrors[d].rdev->nr_pending); md_sync_acct(conf->mirrors[d].rdev->bdev, wbio->bi_size >> 9); generic_make_request(wbio); } if (wbio2 && wbio2->bi_end_io) { atomic_inc(&conf->mirrors[d].replacement->nr_pending); md_sync_acct(conf->mirrors[d].replacement->bdev, wbio2->bi_size >> 9); generic_make_request(wbio2); } |
1da177e4c Linux-2.6.12-rc2 |
1880 1881 1882 1883 |
} /* |
1e50915fe raid: improve MD/... |
1884 1885 1886 1887 1888 |
* Used by fix_read_error() to decay the per rdev read_errors. * We halve the read error count for every hour that has elapsed * since the last recorded read error. * */ |
fd01b88c7 md: remove typede... |
1889 |
static void check_decay_read_errors(struct mddev *mddev, struct md_rdev *rdev) |
1e50915fe raid: improve MD/... |
1890 1891 1892 1893 1894 1895 1896 1897 1898 1899 1900 1901 1902 1903 1904 1905 1906 1907 1908 1909 1910 1911 1912 1913 1914 1915 1916 1917 1918 |
{ struct timespec cur_time_mon; unsigned long hours_since_last; unsigned int read_errors = atomic_read(&rdev->read_errors); ktime_get_ts(&cur_time_mon); if (rdev->last_read_error.tv_sec == 0 && rdev->last_read_error.tv_nsec == 0) { /* first time we've seen a read error */ rdev->last_read_error = cur_time_mon; return; } hours_since_last = (cur_time_mon.tv_sec - rdev->last_read_error.tv_sec) / 3600; rdev->last_read_error = cur_time_mon; /* * if hours_since_last is > the number of bits in read_errors * just set read errors to 0. We do this to avoid * overflowing the shift of read_errors by hours_since_last. */ if (hours_since_last >= 8 * sizeof(read_errors)) atomic_set(&rdev->read_errors, 0); else atomic_set(&rdev->read_errors, read_errors >> hours_since_last); } |
3cb030020 md: removing type... |
1919 |
static int r10_sync_page_io(struct md_rdev *rdev, sector_t sector, |
58c54fcca md/raid10: handle... |
1920 1921 1922 1923 1924 1925 1926 1927 1928 1929 1930 |
int sectors, struct page *page, int rw) { sector_t first_bad; int bad_sectors; if (is_badblock(rdev, sector, sectors, &first_bad, &bad_sectors) && (rw == READ || test_bit(WriteErrorSeen, &rdev->flags))) return -1; if (sync_page_io(rdev, sector, sectors << 9, page, rw, false)) /* success */ return 1; |
b7044d41b md/raid10: If the... |
1931 |
if (rw == WRITE) { |
58c54fcca md/raid10: handle... |
1932 |
set_bit(WriteErrorSeen, &rdev->flags); |
b7044d41b md/raid10: If the... |
1933 1934 1935 1936 |
if (!test_and_set_bit(WantReplacement, &rdev->flags)) set_bit(MD_RECOVERY_NEEDED, &rdev->mddev->recovery); } |
58c54fcca md/raid10: handle... |
1937 1938 1939 1940 1941 |
/* need to record an error - either for the block or the device */ if (!rdev_set_badblocks(rdev, sector, sectors, 0)) md_error(rdev->mddev, rdev); return 0; } |
1e50915fe raid: improve MD/... |
1942 |
/* |
1da177e4c Linux-2.6.12-rc2 |
1943 1944 1945 1946 |
* This is a kernel thread which: * * 1. Retries failed read operations on working mirrors. * 2. Updates the raid superblock when problems encounter. |
6814d5368 [PATCH] md: facto... |
1947 |
* 3. Performs writes following reads for array synchronising. |
1da177e4c Linux-2.6.12-rc2 |
1948 |
*/ |
e879a8793 md/raid10: typede... |
1949 |
static void fix_read_error(struct r10conf *conf, struct mddev *mddev, struct r10bio *r10_bio) |
6814d5368 [PATCH] md: facto... |
1950 1951 1952 |
{ int sect = 0; /* Offset from r10_bio->sector */ int sectors = r10_bio->sectors; |
3cb030020 md: removing type... |
1953 |
struct md_rdev*rdev; |
1e50915fe raid: improve MD/... |
1954 |
int max_read_errors = atomic_read(&mddev->max_corr_read_errors); |
0544a21db md: raid10: Fix n... |
1955 |
int d = r10_bio->devs[r10_bio->read_slot].devnum; |
1e50915fe raid: improve MD/... |
1956 |
|
7c4e06ff2 md/raid10: some t... |
1957 1958 1959 1960 |
/* still own a reference to this rdev, so it cannot * have been cleared recently. */ rdev = conf->mirrors[d].rdev; |
1e50915fe raid: improve MD/... |
1961 |
|
7c4e06ff2 md/raid10: some t... |
1962 1963 1964 1965 |
if (test_bit(Faulty, &rdev->flags)) /* drive has already been failed, just ignore any more fix_read_error() attempts */ return; |
1e50915fe raid: improve MD/... |
1966 |
|
7c4e06ff2 md/raid10: some t... |
1967 1968 1969 1970 1971 |
check_decay_read_errors(mddev, rdev); atomic_inc(&rdev->read_errors); if (atomic_read(&rdev->read_errors) > max_read_errors) { char b[BDEVNAME_SIZE]; bdevname(rdev->bdev, b); |
1e50915fe raid: improve MD/... |
1972 |
|
7c4e06ff2 md/raid10: some t... |
1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 |
printk(KERN_NOTICE "md/raid10:%s: %s: Raid device exceeded " "read_error threshold [cur %d:max %d] ", mdname(mddev), b, atomic_read(&rdev->read_errors), max_read_errors); printk(KERN_NOTICE "md/raid10:%s: %s: Failing raid device ", mdname(mddev), b); md_error(mddev, conf->mirrors[d].rdev); return; |
1e50915fe raid: improve MD/... |
1985 |
} |
1e50915fe raid: improve MD/... |
1986 |
|
6814d5368 [PATCH] md: facto... |
1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 |
while(sectors) { int s = sectors; int sl = r10_bio->read_slot; int success = 0; int start; if (s > (PAGE_SIZE>>9)) s = PAGE_SIZE >> 9; rcu_read_lock(); do { |
8dbed5ceb md/raid10 - avoid... |
1998 1999 |
sector_t first_bad; int bad_sectors; |
0544a21db md: raid10: Fix n... |
2000 |
d = r10_bio->devs[sl].devnum; |
6814d5368 [PATCH] md: facto... |
2001 2002 |
rdev = rcu_dereference(conf->mirrors[d].rdev); if (rdev && |
8dbed5ceb md/raid10 - avoid... |
2003 2004 2005 |
test_bit(In_sync, &rdev->flags) && is_badblock(rdev, r10_bio->devs[sl].addr + sect, s, &first_bad, &bad_sectors) == 0) { |
6814d5368 [PATCH] md: facto... |
2006 2007 |
atomic_inc(&rdev->nr_pending); rcu_read_unlock(); |
2b193363e md: change type o... |
2008 |
success = sync_page_io(rdev, |
6814d5368 [PATCH] md: facto... |
2009 |
r10_bio->devs[sl].addr + |
ccebd4c41 md-new-param-to_s... |
2010 |
sect, |
6814d5368 [PATCH] md: facto... |
2011 |
s<<9, |
ccebd4c41 md-new-param-to_s... |
2012 |
conf->tmppage, READ, false); |
6814d5368 [PATCH] md: facto... |
2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 |
rdev_dec_pending(rdev, mddev); rcu_read_lock(); if (success) break; } sl++; if (sl == conf->copies) sl = 0; } while (!success && sl != r10_bio->read_slot); rcu_read_unlock(); if (!success) { |
58c54fcca md/raid10: handle... |
2025 2026 2027 2028 |
/* Cannot read from anywhere, just mark the block * as bad on the first device to discourage future * reads. */ |
6814d5368 [PATCH] md: facto... |
2029 |
int dn = r10_bio->devs[r10_bio->read_slot].devnum; |
58c54fcca md/raid10: handle... |
2030 2031 2032 2033 2034 2035 2036 2037 |
rdev = conf->mirrors[dn].rdev; if (!rdev_set_badblocks( rdev, r10_bio->devs[r10_bio->read_slot].addr + sect, s, 0)) md_error(mddev, rdev); |
6814d5368 [PATCH] md: facto... |
2038 2039 2040 2041 2042 2043 2044 |
break; } start = sl; /* write it back and re-read */ rcu_read_lock(); while (sl != r10_bio->read_slot) { |
67b8dc4b0 md/raid10: print ... |
2045 |
char b[BDEVNAME_SIZE]; |
0544a21db md: raid10: Fix n... |
2046 |
|
6814d5368 [PATCH] md: facto... |
2047 2048 2049 2050 2051 |
if (sl==0) sl = conf->copies; sl--; d = r10_bio->devs[sl].devnum; rdev = rcu_dereference(conf->mirrors[d].rdev); |
1294b9c97 md/raid10: simpli... |
2052 2053 2054 2055 2056 2057 |
if (!rdev || !test_bit(In_sync, &rdev->flags)) continue; atomic_inc(&rdev->nr_pending); rcu_read_unlock(); |
58c54fcca md/raid10: handle... |
2058 2059 2060 2061 |
if (r10_sync_page_io(rdev, r10_bio->devs[sl].addr + sect, s<<9, conf->tmppage, WRITE) |
1294b9c97 md/raid10: simpli... |
2062 2063 2064 2065 2066 2067 2068 2069 2070 2071 2072 2073 2074 2075 2076 2077 |
== 0) { /* Well, this device is dead */ printk(KERN_NOTICE "md/raid10:%s: read correction " "write failed" " (%d sectors at %llu on %s) ", mdname(mddev), s, (unsigned long long)( sect + rdev->data_offset), bdevname(rdev->bdev, b)); printk(KERN_NOTICE "md/raid10:%s: %s: failing " "drive ", mdname(mddev), bdevname(rdev->bdev, b)); |
6814d5368 [PATCH] md: facto... |
2078 |
} |
1294b9c97 md/raid10: simpli... |
2079 2080 |
rdev_dec_pending(rdev, mddev); rcu_read_lock(); |
6814d5368 [PATCH] md: facto... |
2081 2082 2083 |
} sl = start; while (sl != r10_bio->read_slot) { |
1294b9c97 md/raid10: simpli... |
2084 |
char b[BDEVNAME_SIZE]; |
0544a21db md: raid10: Fix n... |
2085 |
|
6814d5368 [PATCH] md: facto... |
2086 2087 2088 2089 2090 |
if (sl==0) sl = conf->copies; sl--; d = r10_bio->devs[sl].devnum; rdev = rcu_dereference(conf->mirrors[d].rdev); |
1294b9c97 md/raid10: simpli... |
2091 2092 2093 |
if (!rdev || !test_bit(In_sync, &rdev->flags)) continue; |
6814d5368 [PATCH] md: facto... |
2094 |
|
1294b9c97 md/raid10: simpli... |
2095 2096 |
atomic_inc(&rdev->nr_pending); rcu_read_unlock(); |
58c54fcca md/raid10: handle... |
2097 2098 2099 2100 2101 2102 |
switch (r10_sync_page_io(rdev, r10_bio->devs[sl].addr + sect, s<<9, conf->tmppage, READ)) { case 0: |
1294b9c97 md/raid10: simpli... |
2103 2104 2105 2106 2107 2108 2109 2110 2111 2112 2113 2114 2115 2116 2117 |
/* Well, this device is dead */ printk(KERN_NOTICE "md/raid10:%s: unable to read back " "corrected sectors" " (%d sectors at %llu on %s) ", mdname(mddev), s, (unsigned long long)( sect + rdev->data_offset), bdevname(rdev->bdev, b)); printk(KERN_NOTICE "md/raid10:%s: %s: failing " "drive ", mdname(mddev), bdevname(rdev->bdev, b)); |
58c54fcca md/raid10: handle... |
2118 2119 |
break; case 1: |
1294b9c97 md/raid10: simpli... |
2120 2121 2122 2123 2124 2125 2126 2127 2128 |
printk(KERN_INFO "md/raid10:%s: read error corrected" " (%d sectors at %llu on %s) ", mdname(mddev), s, (unsigned long long)( sect + rdev->data_offset), bdevname(rdev->bdev, b)); atomic_add(s, &rdev->corrected_errors); |
6814d5368 [PATCH] md: facto... |
2129 |
} |
1294b9c97 md/raid10: simpli... |
2130 2131 2132 |
rdev_dec_pending(rdev, mddev); rcu_read_lock(); |
6814d5368 [PATCH] md: facto... |
2133 2134 2135 2136 2137 2138 2139 |
} rcu_read_unlock(); sectors -= s; sect += s; } } |
bd870a16c md/raid10: Handl... |
2140 2141 2142 2143 2144 2145 2146 2147 2148 2149 2150 2151 2152 2153 2154 2155 2156 2157 |
static void bi_complete(struct bio *bio, int error) { complete((struct completion *)bio->bi_private); } static int submit_bio_wait(int rw, struct bio *bio) { struct completion event; rw |= REQ_SYNC; init_completion(&event); bio->bi_private = &event; bio->bi_end_io = bi_complete; submit_bio(rw, bio); wait_for_completion(&event); return test_bit(BIO_UPTODATE, &bio->bi_flags); } |
9f2c9d12b md: remove typede... |
2158 |
static int narrow_write_error(struct r10bio *r10_bio, int i) |
bd870a16c md/raid10: Handl... |
2159 2160 |
{ struct bio *bio = r10_bio->master_bio; |
fd01b88c7 md: remove typede... |
2161 |
struct mddev *mddev = r10_bio->mddev; |
e879a8793 md/raid10: typede... |
2162 |
struct r10conf *conf = mddev->private; |
3cb030020 md: removing type... |
2163 |
struct md_rdev *rdev = conf->mirrors[r10_bio->devs[i].devnum].rdev; |
bd870a16c md/raid10: Handl... |
2164 2165 2166 2167 2168 2169 2170 2171 2172 2173 2174 2175 2176 2177 2178 2179 2180 2181 2182 2183 2184 2185 2186 2187 2188 2189 2190 2191 2192 2193 2194 2195 2196 2197 2198 2199 2200 2201 2202 2203 2204 2205 2206 2207 2208 2209 2210 2211 2212 2213 |
/* bio has the data to be written to slot 'i' where * we just recently had a write error. * We repeatedly clone the bio and trim down to one block, * then try the write. Where the write fails we record * a bad block. * It is conceivable that the bio doesn't exactly align with * blocks. We must handle this. * * We currently own a reference to the rdev. */ int block_sectors; sector_t sector; int sectors; int sect_to_write = r10_bio->sectors; int ok = 1; if (rdev->badblocks.shift < 0) return 0; block_sectors = 1 << rdev->badblocks.shift; sector = r10_bio->sector; sectors = ((r10_bio->sector + block_sectors) & ~(sector_t)(block_sectors - 1)) - sector; while (sect_to_write) { struct bio *wbio; if (sectors > sect_to_write) sectors = sect_to_write; /* Write at 'sector' for 'sectors' */ wbio = bio_clone_mddev(bio, GFP_NOIO, mddev); md_trim_bio(wbio, sector - bio->bi_sector, sectors); wbio->bi_sector = (r10_bio->devs[i].addr+ rdev->data_offset+ (sector - r10_bio->sector)); wbio->bi_bdev = rdev->bdev; if (submit_bio_wait(WRITE, wbio) == 0) /* Failure! */ ok = rdev_set_badblocks(rdev, sector, sectors, 0) && ok; bio_put(wbio); sect_to_write -= sectors; sector += sectors; sectors = block_sectors; } return ok; } |
9f2c9d12b md: remove typede... |
2214 |
static void handle_read_error(struct mddev *mddev, struct r10bio *r10_bio) |
560f8e553 md/raid10: Split ... |
2215 2216 |
{ int slot = r10_bio->read_slot; |
560f8e553 md/raid10: Split ... |
2217 |
struct bio *bio; |
e879a8793 md/raid10: typede... |
2218 |
struct r10conf *conf = mddev->private; |
abbf098e6 md/raid10: prefer... |
2219 |
struct md_rdev *rdev = r10_bio->devs[slot].rdev; |
560f8e553 md/raid10: Split ... |
2220 2221 |
char b[BDEVNAME_SIZE]; unsigned long do_sync; |
856e08e23 md/raid10: avoid ... |
2222 |
int max_sectors; |
560f8e553 md/raid10: Split ... |
2223 2224 2225 2226 2227 2228 2229 2230 2231 2232 2233 2234 2235 2236 |
/* we got a read error. Maybe the drive is bad. Maybe just * the block and we can fix it. * We freeze all other IO, and try reading the block from * other devices. When we find one, we re-write * and check it that fixes the read error. * This is all done synchronously while the array is * frozen. */ if (mddev->ro == 0) { freeze_array(conf); fix_read_error(conf, mddev, r10_bio); unfreeze_array(conf); } |
abbf098e6 md/raid10: prefer... |
2237 |
rdev_dec_pending(rdev, mddev); |
560f8e553 md/raid10: Split ... |
2238 2239 |
bio = r10_bio->devs[slot].bio; |
7399c31bc md/raid10: avoid ... |
2240 |
bdevname(bio->bi_bdev, b); |
560f8e553 md/raid10: Split ... |
2241 2242 |
r10_bio->devs[slot].bio = mddev->ro ? IO_BLOCKED : NULL; |
7399c31bc md/raid10: avoid ... |
2243 |
read_more: |
96c3fd1f3 md/raid10: chang... |
2244 2245 |
rdev = read_balance(conf, r10_bio, &max_sectors); if (rdev == NULL) { |
560f8e553 md/raid10: Split ... |
2246 2247 2248 |
printk(KERN_ALERT "md/raid10:%s: %s: unrecoverable I/O" " read error for block %llu ", |
7399c31bc md/raid10: avoid ... |
2249 |
mdname(mddev), b, |
560f8e553 md/raid10: Split ... |
2250 2251 2252 2253 2254 2255 2256 |
(unsigned long long)r10_bio->sector); raid_end_bio_io(r10_bio); bio_put(bio); return; } do_sync = (r10_bio->master_bio->bi_rw & REQ_SYNC); |
7399c31bc md/raid10: avoid ... |
2257 2258 |
if (bio) bio_put(bio); |
560f8e553 md/raid10: Split ... |
2259 |
slot = r10_bio->read_slot; |
560f8e553 md/raid10: Split ... |
2260 2261 2262 2263 2264 2265 2266 2267 2268 2269 |
printk_ratelimited( KERN_ERR "md/raid10:%s: %s: redirecting" "sector %llu to another mirror ", mdname(mddev), bdevname(rdev->bdev, b), (unsigned long long)r10_bio->sector); bio = bio_clone_mddev(r10_bio->master_bio, GFP_NOIO, mddev); |
7399c31bc md/raid10: avoid ... |
2270 2271 2272 |
md_trim_bio(bio, r10_bio->sector - bio->bi_sector, max_sectors); |
560f8e553 md/raid10: Split ... |
2273 |
r10_bio->devs[slot].bio = bio; |
abbf098e6 md/raid10: prefer... |
2274 |
r10_bio->devs[slot].rdev = rdev; |
560f8e553 md/raid10: Split ... |
2275 2276 2277 2278 2279 2280 |
bio->bi_sector = r10_bio->devs[slot].addr + rdev->data_offset; bio->bi_bdev = rdev->bdev; bio->bi_rw = READ | do_sync; bio->bi_private = r10_bio; bio->bi_end_io = raid10_end_read_request; |
7399c31bc md/raid10: avoid ... |
2281 2282 2283 2284 2285 2286 2287 2288 2289 2290 2291 2292 2293 2294 2295 2296 2297 2298 2299 2300 2301 2302 2303 2304 2305 2306 2307 2308 2309 2310 2311 |
if (max_sectors < r10_bio->sectors) { /* Drat - have to split this up more */ struct bio *mbio = r10_bio->master_bio; int sectors_handled = r10_bio->sector + max_sectors - mbio->bi_sector; r10_bio->sectors = max_sectors; spin_lock_irq(&conf->device_lock); if (mbio->bi_phys_segments == 0) mbio->bi_phys_segments = 2; else mbio->bi_phys_segments++; spin_unlock_irq(&conf->device_lock); generic_make_request(bio); bio = NULL; r10_bio = mempool_alloc(conf->r10bio_pool, GFP_NOIO); r10_bio->master_bio = mbio; r10_bio->sectors = (mbio->bi_size >> 9) - sectors_handled; r10_bio->state = 0; set_bit(R10BIO_ReadError, &r10_bio->state); r10_bio->mddev = mddev; r10_bio->sector = mbio->bi_sector + sectors_handled; goto read_more; } else generic_make_request(bio); |
560f8e553 md/raid10: Split ... |
2312 |
} |
e879a8793 md/raid10: typede... |
2313 |
static void handle_write_completed(struct r10conf *conf, struct r10bio *r10_bio) |
749c55e94 md/raid10: clear ... |
2314 2315 2316 2317 |
{ /* Some sort of write request has finished and it * succeeded in writing where we thought there was a * bad block. So forget the bad block. |
1a0b7cd82 md/raid10: record... |
2318 2319 |
* Or possibly if failed and we need to record * a bad block. |
749c55e94 md/raid10: clear ... |
2320 2321 |
*/ int m; |
3cb030020 md: removing type... |
2322 |
struct md_rdev *rdev; |
749c55e94 md/raid10: clear ... |
2323 2324 2325 |
if (test_bit(R10BIO_IsSync, &r10_bio->state) || test_bit(R10BIO_IsRecover, &r10_bio->state)) { |
1a0b7cd82 md/raid10: record... |
2326 2327 2328 2329 2330 2331 |
for (m = 0; m < conf->copies; m++) { int dev = r10_bio->devs[m].devnum; rdev = conf->mirrors[dev].rdev; if (r10_bio->devs[m].bio == NULL) continue; if (test_bit(BIO_UPTODATE, |
749c55e94 md/raid10: clear ... |
2332 |
&r10_bio->devs[m].bio->bi_flags)) { |
749c55e94 md/raid10: clear ... |
2333 2334 2335 2336 |
rdev_clear_badblocks( rdev, r10_bio->devs[m].addr, r10_bio->sectors); |
1a0b7cd82 md/raid10: record... |
2337 2338 2339 2340 2341 2342 |
} else { if (!rdev_set_badblocks( rdev, r10_bio->devs[m].addr, r10_bio->sectors, 0)) md_error(conf->mddev, rdev); |
749c55e94 md/raid10: clear ... |
2343 |
} |
9ad1aefc8 md/raid10: Handl... |
2344 2345 2346 2347 2348 2349 2350 2351 2352 2353 2354 2355 2356 2357 2358 2359 |
rdev = conf->mirrors[dev].replacement; if (r10_bio->devs[m].repl_bio == NULL) continue; if (test_bit(BIO_UPTODATE, &r10_bio->devs[m].repl_bio->bi_flags)) { rdev_clear_badblocks( rdev, r10_bio->devs[m].addr, r10_bio->sectors); } else { if (!rdev_set_badblocks( rdev, r10_bio->devs[m].addr, r10_bio->sectors, 0)) md_error(conf->mddev, rdev); } |
1a0b7cd82 md/raid10: record... |
2360 |
} |
749c55e94 md/raid10: clear ... |
2361 2362 |
put_buf(r10_bio); } else { |
bd870a16c md/raid10: Handl... |
2363 2364 2365 2366 2367 |
for (m = 0; m < conf->copies; m++) { int dev = r10_bio->devs[m].devnum; struct bio *bio = r10_bio->devs[m].bio; rdev = conf->mirrors[dev].rdev; if (bio == IO_MADE_GOOD) { |
749c55e94 md/raid10: clear ... |
2368 2369 2370 2371 2372 |
rdev_clear_badblocks( rdev, r10_bio->devs[m].addr, r10_bio->sectors); rdev_dec_pending(rdev, conf->mddev); |
bd870a16c md/raid10: Handl... |
2373 2374 2375 2376 2377 2378 2379 2380 |
} else if (bio != NULL && !test_bit(BIO_UPTODATE, &bio->bi_flags)) { if (!narrow_write_error(r10_bio, m)) { md_error(conf->mddev, rdev); set_bit(R10BIO_Degraded, &r10_bio->state); } rdev_dec_pending(rdev, conf->mddev); |
749c55e94 md/raid10: clear ... |
2381 |
} |
475b0321a md/raid10: writes... |
2382 2383 |
bio = r10_bio->devs[m].repl_bio; rdev = conf->mirrors[dev].replacement; |
4ca40c2ce md/raid10: Allow ... |
2384 |
if (rdev && bio == IO_MADE_GOOD) { |
475b0321a md/raid10: writes... |
2385 2386 2387 2388 2389 2390 |
rdev_clear_badblocks( rdev, r10_bio->devs[m].addr, r10_bio->sectors); rdev_dec_pending(rdev, conf->mddev); } |
bd870a16c md/raid10: Handl... |
2391 2392 2393 2394 |
} if (test_bit(R10BIO_WriteError, &r10_bio->state)) close_write(r10_bio); |
749c55e94 md/raid10: clear ... |
2395 2396 2397 |
raid_end_bio_io(r10_bio); } } |
fd01b88c7 md: remove typede... |
2398 |
static void raid10d(struct mddev *mddev) |
1da177e4c Linux-2.6.12-rc2 |
2399 |
{ |
9f2c9d12b md: remove typede... |
2400 |
struct r10bio *r10_bio; |
1da177e4c Linux-2.6.12-rc2 |
2401 |
unsigned long flags; |
e879a8793 md/raid10: typede... |
2402 |
struct r10conf *conf = mddev->private; |
1da177e4c Linux-2.6.12-rc2 |
2403 |
struct list_head *head = &conf->retry_list; |
e1dfa0a29 md: use new plugg... |
2404 |
struct blk_plug plug; |
1da177e4c Linux-2.6.12-rc2 |
2405 2406 |
md_check_recovery(mddev); |
1da177e4c Linux-2.6.12-rc2 |
2407 |
|
e1dfa0a29 md: use new plugg... |
2408 |
blk_start_plug(&plug); |
1da177e4c Linux-2.6.12-rc2 |
2409 |
for (;;) { |
6cce3b23f [PATCH] md: write... |
2410 |
|
7eaceacca block: remove per... |
2411 |
flush_pending_writes(conf); |
6cce3b23f [PATCH] md: write... |
2412 |
|
a35e63efa md: fix deadlock ... |
2413 2414 2415 |
spin_lock_irqsave(&conf->device_lock, flags); if (list_empty(head)) { spin_unlock_irqrestore(&conf->device_lock, flags); |
1da177e4c Linux-2.6.12-rc2 |
2416 |
break; |
a35e63efa md: fix deadlock ... |
2417 |
} |
9f2c9d12b md: remove typede... |
2418 |
r10_bio = list_entry(head->prev, struct r10bio, retry_list); |
1da177e4c Linux-2.6.12-rc2 |
2419 |
list_del(head->prev); |
4443ae10c [PATCH] md: auto-... |
2420 |
conf->nr_queued--; |
1da177e4c Linux-2.6.12-rc2 |
2421 2422 2423 |
spin_unlock_irqrestore(&conf->device_lock, flags); mddev = r10_bio->mddev; |
070ec55d0 md: remove mddev_... |
2424 |
conf = mddev->private; |
bd870a16c md/raid10: Handl... |
2425 2426 |
if (test_bit(R10BIO_MadeGood, &r10_bio->state) || test_bit(R10BIO_WriteError, &r10_bio->state)) |
749c55e94 md/raid10: clear ... |
2427 2428 |
handle_write_completed(conf, r10_bio); else if (test_bit(R10BIO_IsSync, &r10_bio->state)) |
1da177e4c Linux-2.6.12-rc2 |
2429 |
sync_request_write(mddev, r10_bio); |
7eaceacca block: remove per... |
2430 |
else if (test_bit(R10BIO_IsRecover, &r10_bio->state)) |
1da177e4c Linux-2.6.12-rc2 |
2431 |
recovery_request_write(mddev, r10_bio); |
856e08e23 md/raid10: avoid ... |
2432 |
else if (test_bit(R10BIO_ReadError, &r10_bio->state)) |
560f8e553 md/raid10: Split ... |
2433 |
handle_read_error(mddev, r10_bio); |
856e08e23 md/raid10: avoid ... |
2434 2435 2436 2437 2438 2439 2440 |
else { /* just a partial read to be scheduled from a * separate context */ int slot = r10_bio->read_slot; generic_make_request(r10_bio->devs[slot].bio); } |
560f8e553 md/raid10: Split ... |
2441 |
|
1d9d52416 md/raid1/raid10: ... |
2442 |
cond_resched(); |
de393cdea md: make it easie... |
2443 2444 |
if (mddev->flags & ~(1<<MD_CHANGE_PENDING)) md_check_recovery(mddev); |
1da177e4c Linux-2.6.12-rc2 |
2445 |
} |
e1dfa0a29 md: use new plugg... |
2446 |
blk_finish_plug(&plug); |
1da177e4c Linux-2.6.12-rc2 |
2447 |
} |
e879a8793 md/raid10: typede... |
2448 |
static int init_resync(struct r10conf *conf) |
1da177e4c Linux-2.6.12-rc2 |
2449 2450 |
{ int buffs; |
69335ef3b md/raid10: prepar... |
2451 |
int i; |
1da177e4c Linux-2.6.12-rc2 |
2452 2453 |
buffs = RESYNC_WINDOW / RESYNC_BLOCK_SIZE; |
b63854838 BUG_ON() Conversi... |
2454 |
BUG_ON(conf->r10buf_pool); |
69335ef3b md/raid10: prepar... |
2455 2456 2457 2458 |
conf->have_replacement = 0; for (i = 0; i < conf->raid_disks; i++) if (conf->mirrors[i].replacement) conf->have_replacement = 1; |
1da177e4c Linux-2.6.12-rc2 |
2459 2460 2461 2462 2463 2464 2465 2466 2467 2468 2469 2470 2471 2472 2473 2474 2475 2476 2477 2478 2479 2480 2481 2482 2483 2484 2485 2486 2487 2488 2489 2490 2491 2492 2493 2494 2495 2496 |
conf->r10buf_pool = mempool_create(buffs, r10buf_pool_alloc, r10buf_pool_free, conf); if (!conf->r10buf_pool) return -ENOMEM; conf->next_resync = 0; return 0; } /* * perform a "sync" on one "block" * * We need to make sure that no normal I/O request - particularly write * requests - conflict with active sync requests. * * This is achieved by tracking pending requests and a 'barrier' concept * that can be installed to exclude normal IO requests. * * Resync and recovery are handled very differently. * We differentiate by looking at MD_RECOVERY_SYNC in mddev->recovery. * * For resync, we iterate over virtual addresses, read all copies, * and update if there are differences. If only one copy is live, * skip it. * For recovery, we iterate over physical addresses, read a good * value for each non-in_sync drive, and over-write. * * So, for recovery we may have several outstanding complex requests for a * given address, one for each out-of-sync device. We model this by allocating * a number of r10_bio structures, one for each out-of-sync device. * As we setup these structures, we collect all bio's together into a list * which we then process collectively to add pages, and then process again * to pass to generic_make_request. * * The r10_bio structures are linked using a borrowed master_bio pointer. * This link is counted in ->remaining. When the r10_bio that points to NULL * has its remaining count decremented to 0, the whole complex operation * is complete. * */ |
fd01b88c7 md: remove typede... |
2497 |
static sector_t sync_request(struct mddev *mddev, sector_t sector_nr, |
ab9d47e99 md/raid10: refor... |
2498 |
int *skipped, int go_faster) |
1da177e4c Linux-2.6.12-rc2 |
2499 |
{ |
e879a8793 md/raid10: typede... |
2500 |
struct r10conf *conf = mddev->private; |
9f2c9d12b md: remove typede... |
2501 |
struct r10bio *r10_bio; |
1da177e4c Linux-2.6.12-rc2 |
2502 2503 |
struct bio *biolist = NULL, *bio; sector_t max_sector, nr_sectors; |
1da177e4c Linux-2.6.12-rc2 |
2504 |
int i; |
6cce3b23f [PATCH] md: write... |
2505 |
int max_sync; |
57dab0bdf md: use sector_t ... |
2506 |
sector_t sync_blocks; |
1da177e4c Linux-2.6.12-rc2 |
2507 2508 2509 2510 2511 |
sector_t sectors_skipped = 0; int chunks_skipped = 0; if (!conf->r10buf_pool) if (init_resync(conf)) |
57afd89f9 [PATCH] md: impro... |
2512 |
return 0; |
1da177e4c Linux-2.6.12-rc2 |
2513 2514 |
skipped: |
58c0fed40 md: Make mddev->s... |
2515 |
max_sector = mddev->dev_sectors; |
1da177e4c Linux-2.6.12-rc2 |
2516 2517 2518 |
if (test_bit(MD_RECOVERY_SYNC, &mddev->recovery)) max_sector = mddev->resync_max_sectors; if (sector_nr >= max_sector) { |
6cce3b23f [PATCH] md: write... |
2519 2520 2521 2522 2523 2524 2525 2526 2527 2528 2529 2530 2531 2532 2533 2534 2535 2536 2537 |
/* If we aborted, we need to abort the * sync on the 'current' bitmap chucks (there can * be several when recovering multiple devices). * as we may have started syncing it but not finished. * We can find the current address in * mddev->curr_resync, but for recovery, * we need to convert that to several * virtual addresses. */ if (mddev->curr_resync < max_sector) { /* aborted */ if (test_bit(MD_RECOVERY_SYNC, &mddev->recovery)) bitmap_end_sync(mddev->bitmap, mddev->curr_resync, &sync_blocks, 1); else for (i=0; i<conf->raid_disks; i++) { sector_t sect = raid10_find_virt(conf, mddev->curr_resync, i); bitmap_end_sync(mddev->bitmap, sect, &sync_blocks, 1); } |
9ad1aefc8 md/raid10: Handl... |
2538 2539 2540 2541 2542 2543 2544 2545 2546 2547 2548 2549 2550 2551 |
} else { /* completed sync */ if ((!mddev->bitmap || conf->fullsync) && conf->have_replacement && test_bit(MD_RECOVERY_SYNC, &mddev->recovery)) { /* Completed a full sync so the replacements * are now fully recovered. */ for (i = 0; i < conf->raid_disks; i++) if (conf->mirrors[i].replacement) conf->mirrors[i].replacement ->recovery_offset = MaxSector; } |
6cce3b23f [PATCH] md: write... |
2552 |
conf->fullsync = 0; |
9ad1aefc8 md/raid10: Handl... |
2553 |
} |
6cce3b23f [PATCH] md: write... |
2554 |
bitmap_close_sync(mddev->bitmap); |
1da177e4c Linux-2.6.12-rc2 |
2555 |
close_sync(conf); |
57afd89f9 [PATCH] md: impro... |
2556 |
*skipped = 1; |
1da177e4c Linux-2.6.12-rc2 |
2557 2558 2559 2560 2561 2562 |
return sectors_skipped; } if (chunks_skipped >= conf->raid_disks) { /* if there has been nothing to do on any drive, * then there is nothing to do at all.. */ |
57afd89f9 [PATCH] md: impro... |
2563 2564 |
*skipped = 1; return (max_sector - sector_nr) + sectors_skipped; |
1da177e4c Linux-2.6.12-rc2 |
2565 |
} |
c62072777 md: allow a maxim... |
2566 2567 |
if (max_sector > mddev->resync_max) max_sector = mddev->resync_max; /* Don't do IO beyond here */ |
1da177e4c Linux-2.6.12-rc2 |
2568 2569 2570 2571 2572 2573 2574 2575 2576 2577 |
/* make sure whole request will fit in a chunk - if chunks * are meaningful */ if (conf->near_copies < conf->raid_disks && max_sector > (sector_nr | conf->chunk_mask)) max_sector = (sector_nr | conf->chunk_mask) + 1; /* * If there is non-resync activity waiting for us then * put in a delay to throttle resync. */ |
0a27ec96b [PATCH] md: impro... |
2578 |
if (!go_faster && conf->nr_waiting) |
1da177e4c Linux-2.6.12-rc2 |
2579 |
msleep_interruptible(1000); |
1da177e4c Linux-2.6.12-rc2 |
2580 2581 2582 2583 2584 2585 2586 2587 2588 2589 2590 2591 2592 2593 2594 |
/* Again, very different code for resync and recovery. * Both must result in an r10bio with a list of bios that * have bi_end_io, bi_sector, bi_bdev set, * and bi_private set to the r10bio. * For recovery, we may actually create several r10bios * with 2 bios in each, that correspond to the bios in the main one. * In this case, the subordinate r10bios link back through a * borrowed master_bio pointer, and the counter in the master * includes a ref from each subordinate. */ /* First, we decide what to do and set ->bi_end_io * To end_sync_read if we want to read, and * end_sync_write if we will want to write. */ |
6cce3b23f [PATCH] md: write... |
2595 |
max_sync = RESYNC_PAGES << (PAGE_SHIFT-9); |
1da177e4c Linux-2.6.12-rc2 |
2596 2597 |
if (!test_bit(MD_RECOVERY_SYNC, &mddev->recovery)) { /* recovery... the complicated one */ |
e875ecea2 md/raid10 record ... |
2598 |
int j; |
1da177e4c Linux-2.6.12-rc2 |
2599 |
r10_bio = NULL; |
ab9d47e99 md/raid10: refor... |
2600 2601 |
for (i=0 ; i<conf->raid_disks; i++) { int still_degraded; |
9f2c9d12b md: remove typede... |
2602 |
struct r10bio *rb2; |
ab9d47e99 md/raid10: refor... |
2603 2604 |
sector_t sect; int must_sync; |
e875ecea2 md/raid10 record ... |
2605 |
int any_working; |
24afd80d9 md/raid10: handle... |
2606 2607 2608 2609 2610 2611 2612 2613 |
struct mirror_info *mirror = &conf->mirrors[i]; if ((mirror->rdev == NULL || test_bit(In_sync, &mirror->rdev->flags)) && (mirror->replacement == NULL || test_bit(Faulty, &mirror->replacement->flags))) |
ab9d47e99 md/raid10: refor... |
2614 |
continue; |
1da177e4c Linux-2.6.12-rc2 |
2615 |
|
ab9d47e99 md/raid10: refor... |
2616 2617 2618 2619 |
still_degraded = 0; /* want to reconstruct this device */ rb2 = r10_bio; sect = raid10_find_virt(conf, sector_nr, i); |
24afd80d9 md/raid10: handle... |
2620 2621 2622 |
/* Unless we are doing a full sync, or a replacement * we only need to recover the block if it is set in * the bitmap |
ab9d47e99 md/raid10: refor... |
2623 2624 2625 2626 2627 2628 |
*/ must_sync = bitmap_start_sync(mddev->bitmap, sect, &sync_blocks, 1); if (sync_blocks < max_sync) max_sync = sync_blocks; if (!must_sync && |
24afd80d9 md/raid10: handle... |
2629 |
mirror->replacement == NULL && |
ab9d47e99 md/raid10: refor... |
2630 2631 2632 2633 2634 2635 2636 |
!conf->fullsync) { /* yep, skip the sync_blocks here, but don't assume * that there will never be anything to do here */ chunks_skipped = -1; continue; } |
6cce3b23f [PATCH] md: write... |
2637 |
|
ab9d47e99 md/raid10: refor... |
2638 2639 2640 |
r10_bio = mempool_alloc(conf->r10buf_pool, GFP_NOIO); raise_barrier(conf, rb2 != NULL); atomic_set(&r10_bio->remaining, 0); |
180555691 md/raid10: don't ... |
2641 |
|
ab9d47e99 md/raid10: refor... |
2642 2643 2644 2645 2646 2647 |
r10_bio->master_bio = (struct bio*)rb2; if (rb2) atomic_inc(&rb2->remaining); r10_bio->mddev = mddev; set_bit(R10BIO_IsRecover, &r10_bio->state); r10_bio->sector = sect; |
1da177e4c Linux-2.6.12-rc2 |
2648 |
|
ab9d47e99 md/raid10: refor... |
2649 2650 2651 2652 2653 2654 2655 2656 2657 |
raid10_find_phys(conf, r10_bio); /* Need to check if the array will still be * degraded */ for (j=0; j<conf->raid_disks; j++) if (conf->mirrors[j].rdev == NULL || test_bit(Faulty, &conf->mirrors[j].rdev->flags)) { still_degraded = 1; |
87fc767b8 [PATCH] md: fix B... |
2658 |
break; |
1da177e4c Linux-2.6.12-rc2 |
2659 |
} |
ab9d47e99 md/raid10: refor... |
2660 2661 2662 |
must_sync = bitmap_start_sync(mddev->bitmap, sect, &sync_blocks, still_degraded); |
e875ecea2 md/raid10 record ... |
2663 |
any_working = 0; |
ab9d47e99 md/raid10: refor... |
2664 |
for (j=0; j<conf->copies;j++) { |
e875ecea2 md/raid10 record ... |
2665 |
int k; |
ab9d47e99 md/raid10: refor... |
2666 |
int d = r10_bio->devs[j].devnum; |
5e5702898 md/raid10: Handle... |
2667 |
sector_t from_addr, to_addr; |
3cb030020 md: removing type... |
2668 |
struct md_rdev *rdev; |
40c356ce5 md/raid10: avoid ... |
2669 2670 |
sector_t sector, first_bad; int bad_sectors; |
ab9d47e99 md/raid10: refor... |
2671 2672 2673 2674 |
if (!conf->mirrors[d].rdev || !test_bit(In_sync, &conf->mirrors[d].rdev->flags)) continue; /* This is where we read from */ |
e875ecea2 md/raid10 record ... |
2675 |
any_working = 1; |
40c356ce5 md/raid10: avoid ... |
2676 2677 2678 2679 2680 2681 2682 2683 2684 2685 2686 2687 2688 2689 2690 |
rdev = conf->mirrors[d].rdev; sector = r10_bio->devs[j].addr; if (is_badblock(rdev, sector, max_sync, &first_bad, &bad_sectors)) { if (first_bad > sector) max_sync = first_bad - sector; else { bad_sectors -= (sector - first_bad); if (max_sync > bad_sectors) max_sync = bad_sectors; continue; } } |
ab9d47e99 md/raid10: refor... |
2691 2692 2693 2694 2695 2696 |
bio = r10_bio->devs[0].bio; bio->bi_next = biolist; biolist = bio; bio->bi_private = r10_bio; bio->bi_end_io = end_sync_read; bio->bi_rw = READ; |
5e5702898 md/raid10: Handle... |
2697 |
from_addr = r10_bio->devs[j].addr; |
24afd80d9 md/raid10: handle... |
2698 2699 2700 2701 |
bio->bi_sector = from_addr + rdev->data_offset; bio->bi_bdev = rdev->bdev; atomic_inc(&rdev->nr_pending); /* and we write to 'i' (if not in_sync) */ |
ab9d47e99 md/raid10: refor... |
2702 2703 2704 2705 2706 |
for (k=0; k<conf->copies; k++) if (r10_bio->devs[k].devnum == i) break; BUG_ON(k == conf->copies); |
5e5702898 md/raid10: Handle... |
2707 |
to_addr = r10_bio->devs[k].addr; |
ab9d47e99 md/raid10: refor... |
2708 |
r10_bio->devs[0].devnum = d; |
5e5702898 md/raid10: Handle... |
2709 |
r10_bio->devs[0].addr = from_addr; |
ab9d47e99 md/raid10: refor... |
2710 |
r10_bio->devs[1].devnum = i; |
5e5702898 md/raid10: Handle... |
2711 |
r10_bio->devs[1].addr = to_addr; |
ab9d47e99 md/raid10: refor... |
2712 |
|
24afd80d9 md/raid10: handle... |
2713 2714 2715 2716 2717 2718 2719 2720 2721 2722 2723 2724 2725 2726 2727 2728 2729 2730 2731 2732 2733 2734 2735 2736 2737 2738 2739 2740 2741 2742 2743 2744 2745 2746 2747 2748 2749 2750 2751 |
rdev = mirror->rdev; if (!test_bit(In_sync, &rdev->flags)) { bio = r10_bio->devs[1].bio; bio->bi_next = biolist; biolist = bio; bio->bi_private = r10_bio; bio->bi_end_io = end_sync_write; bio->bi_rw = WRITE; bio->bi_sector = to_addr + rdev->data_offset; bio->bi_bdev = rdev->bdev; atomic_inc(&r10_bio->remaining); } else r10_bio->devs[1].bio->bi_end_io = NULL; /* and maybe write to replacement */ bio = r10_bio->devs[1].repl_bio; if (bio) bio->bi_end_io = NULL; rdev = mirror->replacement; /* Note: if rdev != NULL, then bio * cannot be NULL as r10buf_pool_alloc will * have allocated it. * So the second test here is pointless. * But it keeps semantic-checkers happy, and * this comment keeps human reviewers * happy. */ if (rdev == NULL || bio == NULL || test_bit(Faulty, &rdev->flags)) break; bio->bi_next = biolist; biolist = bio; bio->bi_private = r10_bio; bio->bi_end_io = end_sync_write; bio->bi_rw = WRITE; bio->bi_sector = to_addr + rdev->data_offset; bio->bi_bdev = rdev->bdev; atomic_inc(&r10_bio->remaining); |
ab9d47e99 md/raid10: refor... |
2752 2753 2754 |
break; } if (j == conf->copies) { |
e875ecea2 md/raid10 record ... |
2755 2756 |
/* Cannot recover, so abort the recovery or * record a bad block */ |
ab9d47e99 md/raid10: refor... |
2757 2758 2759 2760 |
put_buf(r10_bio); if (rb2) atomic_dec(&rb2->remaining); r10_bio = rb2; |
e875ecea2 md/raid10 record ... |
2761 2762 2763 2764 2765 2766 2767 2768 |
if (any_working) { /* problem is that there are bad blocks * on other device(s) */ int k; for (k = 0; k < conf->copies; k++) if (r10_bio->devs[k].devnum == i) break; |
24afd80d9 md/raid10: handle... |
2769 2770 2771 2772 2773 2774 2775 2776 2777 2778 |
if (!test_bit(In_sync, &mirror->rdev->flags) && !rdev_set_badblocks( mirror->rdev, r10_bio->devs[k].addr, max_sync, 0)) any_working = 0; if (mirror->replacement && !rdev_set_badblocks( mirror->replacement, |
e875ecea2 md/raid10 record ... |
2779 2780 2781 2782 2783 2784 2785 2786 2787 2788 2789 |
r10_bio->devs[k].addr, max_sync, 0)) any_working = 0; } if (!any_working) { if (!test_and_set_bit(MD_RECOVERY_INTR, &mddev->recovery)) printk(KERN_INFO "md/raid10:%s: insufficient " "working devices for recovery. ", mdname(mddev)); |
24afd80d9 md/raid10: handle... |
2790 |
mirror->recovery_disabled |
e875ecea2 md/raid10 record ... |
2791 2792 |
= mddev->recovery_disabled; } |
ab9d47e99 md/raid10: refor... |
2793 |
break; |
1da177e4c Linux-2.6.12-rc2 |
2794 |
} |
ab9d47e99 md/raid10: refor... |
2795 |
} |
1da177e4c Linux-2.6.12-rc2 |
2796 2797 |
if (biolist == NULL) { while (r10_bio) { |
9f2c9d12b md: remove typede... |
2798 2799 |
struct r10bio *rb2 = r10_bio; r10_bio = (struct r10bio*) rb2->master_bio; |
1da177e4c Linux-2.6.12-rc2 |
2800 2801 2802 2803 2804 2805 2806 2807 |
rb2->master_bio = NULL; put_buf(rb2); } goto giveup; } } else { /* resync. Schedule a read for every block at this virt offset */ int count = 0; |
6cce3b23f [PATCH] md: write... |
2808 |
|
78200d45c md/raid10: Don't... |
2809 |
bitmap_cond_end_sync(mddev->bitmap, sector_nr); |
6cce3b23f [PATCH] md: write... |
2810 2811 |
if (!bitmap_start_sync(mddev->bitmap, sector_nr, &sync_blocks, mddev->degraded) && |
ab9d47e99 md/raid10: refor... |
2812 2813 |
!conf->fullsync && !test_bit(MD_RECOVERY_REQUESTED, &mddev->recovery)) { |
6cce3b23f [PATCH] md: write... |
2814 2815 2816 2817 2818 2819 |
/* We can skip this block */ *skipped = 1; return sync_blocks + sectors_skipped; } if (sync_blocks < max_sync) max_sync = sync_blocks; |
1da177e4c Linux-2.6.12-rc2 |
2820 |
r10_bio = mempool_alloc(conf->r10buf_pool, GFP_NOIO); |
1da177e4c Linux-2.6.12-rc2 |
2821 2822 |
r10_bio->mddev = mddev; atomic_set(&r10_bio->remaining, 0); |
6cce3b23f [PATCH] md: write... |
2823 2824 |
raise_barrier(conf, 0); conf->next_resync = sector_nr; |
1da177e4c Linux-2.6.12-rc2 |
2825 2826 2827 2828 2829 2830 2831 2832 2833 |
r10_bio->master_bio = NULL; r10_bio->sector = sector_nr; set_bit(R10BIO_IsSync, &r10_bio->state); raid10_find_phys(conf, r10_bio); r10_bio->sectors = (sector_nr | conf->chunk_mask) - sector_nr +1; for (i=0; i<conf->copies; i++) { int d = r10_bio->devs[i].devnum; |
40c356ce5 md/raid10: avoid ... |
2834 2835 |
sector_t first_bad, sector; int bad_sectors; |
9ad1aefc8 md/raid10: Handl... |
2836 2837 |
if (r10_bio->devs[i].repl_bio) r10_bio->devs[i].repl_bio->bi_end_io = NULL; |
1da177e4c Linux-2.6.12-rc2 |
2838 2839 |
bio = r10_bio->devs[i].bio; bio->bi_end_io = NULL; |
af03b8e4e md: fix two raid1... |
2840 |
clear_bit(BIO_UPTODATE, &bio->bi_flags); |
1da177e4c Linux-2.6.12-rc2 |
2841 |
if (conf->mirrors[d].rdev == NULL || |
b2d444d7a [PATCH] md: conve... |
2842 |
test_bit(Faulty, &conf->mirrors[d].rdev->flags)) |
1da177e4c Linux-2.6.12-rc2 |
2843 |
continue; |
40c356ce5 md/raid10: avoid ... |
2844 2845 2846 2847 2848 2849 2850 2851 2852 2853 2854 2855 2856 |
sector = r10_bio->devs[i].addr; if (is_badblock(conf->mirrors[d].rdev, sector, max_sync, &first_bad, &bad_sectors)) { if (first_bad > sector) max_sync = first_bad - sector; else { bad_sectors -= (sector - first_bad); if (max_sync > bad_sectors) max_sync = max_sync; continue; } } |
1da177e4c Linux-2.6.12-rc2 |
2857 2858 2859 2860 2861 2862 |
atomic_inc(&conf->mirrors[d].rdev->nr_pending); atomic_inc(&r10_bio->remaining); bio->bi_next = biolist; biolist = bio; bio->bi_private = r10_bio; bio->bi_end_io = end_sync_read; |
802ba064c [PATCH] md: Don't... |
2863 |
bio->bi_rw = READ; |
40c356ce5 md/raid10: avoid ... |
2864 |
bio->bi_sector = sector + |
1da177e4c Linux-2.6.12-rc2 |
2865 2866 2867 |
conf->mirrors[d].rdev->data_offset; bio->bi_bdev = conf->mirrors[d].rdev->bdev; count++; |
9ad1aefc8 md/raid10: Handl... |
2868 2869 2870 2871 2872 2873 2874 2875 2876 2877 2878 2879 2880 2881 2882 2883 2884 2885 2886 2887 2888 |
if (conf->mirrors[d].replacement == NULL || test_bit(Faulty, &conf->mirrors[d].replacement->flags)) continue; /* Need to set up for writing to the replacement */ bio = r10_bio->devs[i].repl_bio; clear_bit(BIO_UPTODATE, &bio->bi_flags); sector = r10_bio->devs[i].addr; atomic_inc(&conf->mirrors[d].rdev->nr_pending); bio->bi_next = biolist; biolist = bio; bio->bi_private = r10_bio; bio->bi_end_io = end_sync_write; bio->bi_rw = WRITE; bio->bi_sector = sector + conf->mirrors[d].replacement->data_offset; bio->bi_bdev = conf->mirrors[d].replacement->bdev; count++; |
1da177e4c Linux-2.6.12-rc2 |
2889 2890 2891 2892 2893 2894 |
} if (count < 2) { for (i=0; i<conf->copies; i++) { int d = r10_bio->devs[i].devnum; if (r10_bio->devs[i].bio->bi_end_io) |
ab9d47e99 md/raid10: refor... |
2895 2896 |
rdev_dec_pending(conf->mirrors[d].rdev, mddev); |
9ad1aefc8 md/raid10: Handl... |
2897 2898 2899 2900 2901 |
if (r10_bio->devs[i].repl_bio && r10_bio->devs[i].repl_bio->bi_end_io) rdev_dec_pending( conf->mirrors[d].replacement, mddev); |
1da177e4c Linux-2.6.12-rc2 |
2902 2903 2904 2905 2906 2907 2908 2909 2910 2911 2912 2913 2914 2915 2916 |
} put_buf(r10_bio); biolist = NULL; goto giveup; } } for (bio = biolist; bio ; bio=bio->bi_next) { bio->bi_flags &= ~(BIO_POOL_MASK - 1); if (bio->bi_end_io) bio->bi_flags |= 1 << BIO_UPTODATE; bio->bi_vcnt = 0; bio->bi_idx = 0; bio->bi_phys_segments = 0; |
1da177e4c Linux-2.6.12-rc2 |
2917 2918 2919 2920 |
bio->bi_size = 0; } nr_sectors = 0; |
6cce3b23f [PATCH] md: write... |
2921 2922 |
if (sector_nr + max_sync < max_sector) max_sector = sector_nr + max_sync; |
1da177e4c Linux-2.6.12-rc2 |
2923 2924 2925 |
do { struct page *page; int len = PAGE_SIZE; |
1da177e4c Linux-2.6.12-rc2 |
2926 2927 2928 2929 2930 |
if (sector_nr + (len>>9) > max_sector) len = (max_sector - sector_nr) << 9; if (len == 0) break; for (bio= biolist ; bio ; bio=bio->bi_next) { |
ab9d47e99 md/raid10: refor... |
2931 |
struct bio *bio2; |
1da177e4c Linux-2.6.12-rc2 |
2932 |
page = bio->bi_io_vec[bio->bi_vcnt].bv_page; |
ab9d47e99 md/raid10: refor... |
2933 2934 2935 2936 2937 2938 2939 2940 2941 2942 2943 2944 |
if (bio_add_page(bio, page, len, 0)) continue; /* stop here */ bio->bi_io_vec[bio->bi_vcnt].bv_page = page; for (bio2 = biolist; bio2 && bio2 != bio; bio2 = bio2->bi_next) { /* remove last page from this bio */ bio2->bi_vcnt--; bio2->bi_size -= len; bio2->bi_flags &= ~(1<< BIO_SEG_VALID); |
1da177e4c Linux-2.6.12-rc2 |
2945 |
} |
ab9d47e99 md/raid10: refor... |
2946 |
goto bio_full; |
1da177e4c Linux-2.6.12-rc2 |
2947 2948 2949 2950 2951 2952 2953 2954 2955 2956 2957 2958 2959 2960 2961 2962 2963 2964 2965 2966 |
} nr_sectors += len>>9; sector_nr += len>>9; } while (biolist->bi_vcnt < RESYNC_PAGES); bio_full: r10_bio->sectors = nr_sectors; while (biolist) { bio = biolist; biolist = biolist->bi_next; bio->bi_next = NULL; r10_bio = bio->bi_private; r10_bio->sectors = nr_sectors; if (bio->bi_end_io == end_sync_read) { md_sync_acct(bio->bi_bdev, nr_sectors); generic_make_request(bio); } } |
57afd89f9 [PATCH] md: impro... |
2967 2968 2969 2970 2971 |
if (sectors_skipped) /* pretend they weren't skipped, it makes * no important difference in this case */ md_done_sync(mddev, sectors_skipped, 1); |
1da177e4c Linux-2.6.12-rc2 |
2972 2973 2974 |
return sectors_skipped + nr_sectors; giveup: /* There is nowhere to write, so all non-sync |
e875ecea2 md/raid10 record ... |
2975 2976 |
* drives must be failed or in resync, all drives * have a bad block, so try the next chunk... |
1da177e4c Linux-2.6.12-rc2 |
2977 |
*/ |
09b4068a7 md/raid10: Don't... |
2978 2979 2980 2981 |
if (sector_nr + max_sync < max_sector) max_sector = sector_nr + max_sync; sectors_skipped += (max_sector - sector_nr); |
1da177e4c Linux-2.6.12-rc2 |
2982 2983 |
chunks_skipped ++; sector_nr = max_sector; |
1da177e4c Linux-2.6.12-rc2 |
2984 |
goto skipped; |
1da177e4c Linux-2.6.12-rc2 |
2985 |
} |
80c3a6ce4 md: add 'size' as... |
2986 |
static sector_t |
fd01b88c7 md: remove typede... |
2987 |
raid10_size(struct mddev *mddev, sector_t sectors, int raid_disks) |
80c3a6ce4 md: add 'size' as... |
2988 2989 |
{ sector_t size; |
e879a8793 md/raid10: typede... |
2990 |
struct r10conf *conf = mddev->private; |
80c3a6ce4 md: add 'size' as... |
2991 2992 |
if (!raid_disks) |
84707f38e md: don't use mdd... |
2993 |
raid_disks = conf->raid_disks; |
80c3a6ce4 md: add 'size' as... |
2994 |
if (!sectors) |
dab8b2924 md: Add support f... |
2995 |
sectors = conf->dev_sectors; |
80c3a6ce4 md: add 'size' as... |
2996 2997 2998 2999 3000 3001 3002 3003 |
size = sectors >> conf->chunk_shift; sector_div(size, conf->far_copies); size = size * raid_disks; sector_div(size, conf->near_copies); return size << conf->chunk_shift; } |
dab8b2924 md: Add support f... |
3004 |
|
e879a8793 md/raid10: typede... |
3005 |
static struct r10conf *setup_conf(struct mddev *mddev) |
1da177e4c Linux-2.6.12-rc2 |
3006 |
{ |
e879a8793 md/raid10: typede... |
3007 |
struct r10conf *conf = NULL; |
c93983bf5 [PATCH] md: suppo... |
3008 |
int nc, fc, fo; |
1da177e4c Linux-2.6.12-rc2 |
3009 |
sector_t stride, size; |
dab8b2924 md: Add support f... |
3010 |
int err = -EINVAL; |
1da177e4c Linux-2.6.12-rc2 |
3011 |
|
f73ea8737 md: fix raid10 ta... |
3012 3013 |
if (mddev->new_chunk_sectors < (PAGE_SIZE >> 9) || !is_power_of_2(mddev->new_chunk_sectors)) { |
128595ed6 md/raid10: tidy u... |
3014 3015 3016 3017 |
printk(KERN_ERR "md/raid10:%s: chunk size must be " "at least PAGE_SIZE(%ld) and be a power of 2. ", mdname(mddev), PAGE_SIZE); |
dab8b2924 md: Add support f... |
3018 |
goto out; |
1da177e4c Linux-2.6.12-rc2 |
3019 |
} |
2604b703b [PATCH] md: remov... |
3020 |
|
f73ea8737 md: fix raid10 ta... |
3021 3022 3023 |
nc = mddev->new_layout & 255; fc = (mddev->new_layout >> 8) & 255; fo = mddev->new_layout & (1<<16); |
dab8b2924 md: Add support f... |
3024 |
|
1da177e4c Linux-2.6.12-rc2 |
3025 |
if ((nc*fc) <2 || (nc*fc) > mddev->raid_disks || |
f73ea8737 md: fix raid10 ta... |
3026 |
(mddev->new_layout >> 17)) { |
128595ed6 md/raid10: tidy u... |
3027 3028 |
printk(KERN_ERR "md/raid10:%s: unsupported raid10 layout: 0x%8x ", |
f73ea8737 md: fix raid10 ta... |
3029 |
mdname(mddev), mddev->new_layout); |
1da177e4c Linux-2.6.12-rc2 |
3030 3031 |
goto out; } |
dab8b2924 md: Add support f... |
3032 3033 |
err = -ENOMEM; |
e879a8793 md/raid10: typede... |
3034 |
conf = kzalloc(sizeof(struct r10conf), GFP_KERNEL); |
dab8b2924 md: Add support f... |
3035 |
if (!conf) |
1da177e4c Linux-2.6.12-rc2 |
3036 |
goto out; |
dab8b2924 md: Add support f... |
3037 |
|
4443ae10c [PATCH] md: auto-... |
3038 |
conf->mirrors = kzalloc(sizeof(struct mirror_info)*mddev->raid_disks, |
dab8b2924 md: Add support f... |
3039 3040 3041 |
GFP_KERNEL); if (!conf->mirrors) goto out; |
4443ae10c [PATCH] md: auto-... |
3042 3043 3044 |
conf->tmppage = alloc_page(GFP_KERNEL); if (!conf->tmppage) |
dab8b2924 md: Add support f... |
3045 |
goto out; |
1da177e4c Linux-2.6.12-rc2 |
3046 |
|
64a742bc6 [PATCH] md: fix r... |
3047 |
conf->raid_disks = mddev->raid_disks; |
1da177e4c Linux-2.6.12-rc2 |
3048 3049 3050 |
conf->near_copies = nc; conf->far_copies = fc; conf->copies = nc*fc; |
c93983bf5 [PATCH] md: suppo... |
3051 |
conf->far_offset = fo; |
dab8b2924 md: Add support f... |
3052 3053 3054 3055 3056 3057 3058 |
conf->chunk_mask = mddev->new_chunk_sectors - 1; conf->chunk_shift = ffz(~mddev->new_chunk_sectors); conf->r10bio_pool = mempool_create(NR_RAID10_BIOS, r10bio_pool_alloc, r10bio_pool_free, conf); if (!conf->r10bio_pool) goto out; |
58c0fed40 md: Make mddev->s... |
3059 |
size = mddev->dev_sectors >> conf->chunk_shift; |
64a742bc6 [PATCH] md: fix r... |
3060 3061 3062 3063 3064 3065 |
sector_div(size, fc); size = size * conf->raid_disks; sector_div(size, nc); /* 'size' is now the number of chunks in the array */ /* calculate "used chunks per device" in 'stride' */ stride = size * conf->copies; |
af03b8e4e md: fix two raid1... |
3066 3067 3068 3069 3070 |
/* We need to round up when dividing by raid_disks to * get the stride size. */ stride += conf->raid_disks - 1; |
64a742bc6 [PATCH] md: fix r... |
3071 |
sector_div(stride, conf->raid_disks); |
dab8b2924 md: Add support f... |
3072 3073 |
conf->dev_sectors = stride << conf->chunk_shift; |
64a742bc6 [PATCH] md: fix r... |
3074 |
|
c93983bf5 [PATCH] md: suppo... |
3075 |
if (fo) |
64a742bc6 [PATCH] md: fix r... |
3076 3077 |
stride = 1; else |
c93983bf5 [PATCH] md: suppo... |
3078 |
sector_div(stride, fc); |
64a742bc6 [PATCH] md: fix r... |
3079 |
conf->stride = stride << conf->chunk_shift; |
1da177e4c Linux-2.6.12-rc2 |
3080 |
|
e7e72bf64 Remove blkdev war... |
3081 |
spin_lock_init(&conf->device_lock); |
dab8b2924 md: Add support f... |
3082 3083 3084 3085 3086 3087 3088 3089 |
INIT_LIST_HEAD(&conf->retry_list); spin_lock_init(&conf->resync_lock); init_waitqueue_head(&conf->wait_barrier); conf->thread = md_register_thread(raid10d, mddev, NULL); if (!conf->thread) goto out; |
dab8b2924 md: Add support f... |
3090 3091 3092 3093 |
conf->mddev = mddev; return conf; out: |
128595ed6 md/raid10: tidy u... |
3094 3095 |
printk(KERN_ERR "md/raid10:%s: couldn't allocate memory. ", |
dab8b2924 md: Add support f... |
3096 3097 3098 3099 3100 3101 3102 3103 3104 3105 |
mdname(mddev)); if (conf) { if (conf->r10bio_pool) mempool_destroy(conf->r10bio_pool); kfree(conf->mirrors); safe_put_page(conf->tmppage); kfree(conf); } return ERR_PTR(err); } |
fd01b88c7 md: remove typede... |
3106 |
static int run(struct mddev *mddev) |
dab8b2924 md: Add support f... |
3107 |
{ |
e879a8793 md/raid10: typede... |
3108 |
struct r10conf *conf; |
dab8b2924 md: Add support f... |
3109 |
int i, disk_idx, chunk_size; |
0f6d02d58 md: remove typede... |
3110 |
struct mirror_info *disk; |
3cb030020 md: removing type... |
3111 |
struct md_rdev *rdev; |
dab8b2924 md: Add support f... |
3112 3113 3114 3115 3116 3117 3118 3119 3120 3121 3122 3123 3124 3125 3126 3127 3128 |
sector_t size; /* * copy the already verified devices into our private RAID10 * bookkeeping area. [whatever we allocate in run(), * should be freed in stop()] */ if (mddev->private == NULL) { conf = setup_conf(mddev); if (IS_ERR(conf)) return PTR_ERR(conf); mddev->private = conf; } conf = mddev->private; if (!conf) goto out; |
dab8b2924 md: Add support f... |
3129 3130 |
mddev->thread = conf->thread; conf->thread = NULL; |
8f6c2e4b3 md: Use new topol... |
3131 3132 3133 3134 3135 3136 3137 |
chunk_size = mddev->chunk_sectors << 9; blk_queue_io_min(mddev->queue, chunk_size); if (conf->raid_disks % conf->near_copies) blk_queue_io_opt(mddev->queue, chunk_size * conf->raid_disks); else blk_queue_io_opt(mddev->queue, chunk_size * (conf->raid_disks / conf->near_copies)); |
159ec1fc0 md: use list_for_... |
3138 |
list_for_each_entry(rdev, &mddev->disks, same_set) { |
34b343cff md: don't allow a... |
3139 |
|
1da177e4c Linux-2.6.12-rc2 |
3140 |
disk_idx = rdev->raid_disk; |
84707f38e md: don't use mdd... |
3141 |
if (disk_idx >= conf->raid_disks |
1da177e4c Linux-2.6.12-rc2 |
3142 3143 3144 |
|| disk_idx < 0) continue; disk = conf->mirrors + disk_idx; |
56a2559bb md/raid10: recogn... |
3145 3146 3147 3148 3149 3150 3151 3152 3153 |
if (test_bit(Replacement, &rdev->flags)) { if (disk->replacement) goto out_free_conf; disk->replacement = rdev; } else { if (disk->rdev) goto out_free_conf; disk->rdev = rdev; } |
1da177e4c Linux-2.6.12-rc2 |
3154 |
disk->rdev = rdev; |
8f6c2e4b3 md: Use new topol... |
3155 3156 |
disk_stack_limits(mddev->gendisk, rdev->bdev, rdev->data_offset << 9); |
1da177e4c Linux-2.6.12-rc2 |
3157 |
/* as we don't honour merge_bvec_fn, we must never risk |
627a2d3c2 md: deal with mer... |
3158 3159 |
* violating it, so limit max_segments to 1 lying * within a single page. |
1da177e4c Linux-2.6.12-rc2 |
3160 |
*/ |
627a2d3c2 md: deal with mer... |
3161 3162 3163 3164 3165 |
if (rdev->bdev->bd_disk->queue->merge_bvec_fn) { blk_queue_max_segments(mddev->queue, 1); blk_queue_segment_boundary(mddev->queue, PAGE_CACHE_SIZE - 1); } |
1da177e4c Linux-2.6.12-rc2 |
3166 3167 |
disk->head_position = 0; |
1da177e4c Linux-2.6.12-rc2 |
3168 |
} |
6d508242b [PATCH] md: fix r... |
3169 |
/* need to check that every block has at least one working mirror */ |
700c72138 md/raid10: Improv... |
3170 |
if (!enough(conf, -1)) { |
128595ed6 md/raid10: tidy u... |
3171 3172 |
printk(KERN_ERR "md/raid10:%s: not enough operational mirrors. ", |
6d508242b [PATCH] md: fix r... |
3173 |
mdname(mddev)); |
1da177e4c Linux-2.6.12-rc2 |
3174 3175 3176 3177 3178 3179 3180 |
goto out_free_conf; } mddev->degraded = 0; for (i = 0; i < conf->raid_disks; i++) { disk = conf->mirrors + i; |
56a2559bb md/raid10: recogn... |
3181 3182 3183 3184 3185 3186 |
if (!disk->rdev && disk->replacement) { /* The replacement is all we have - use it */ disk->rdev = disk->replacement; disk->replacement = NULL; clear_bit(Replacement, &disk->rdev->flags); } |
5fd6c1dce [PATCH] md: allow... |
3187 |
if (!disk->rdev || |
2e333e898 [PATCH] md: fix c... |
3188 |
!test_bit(In_sync, &disk->rdev->flags)) { |
1da177e4c Linux-2.6.12-rc2 |
3189 3190 |
disk->head_position = 0; mddev->degraded++; |
8c2e870a6 Ensure interrupte... |
3191 3192 |
if (disk->rdev) conf->fullsync = 1; |
1da177e4c Linux-2.6.12-rc2 |
3193 |
} |
d890fa2b0 md: Fix some bugs... |
3194 |
disk->recovery_disabled = mddev->recovery_disabled - 1; |
1da177e4c Linux-2.6.12-rc2 |
3195 |
} |
8c6ac868b md: Push down rec... |
3196 |
if (mddev->recovery_cp != MaxSector) |
128595ed6 md/raid10: tidy u... |
3197 |
printk(KERN_NOTICE "md/raid10:%s: not clean" |
8c6ac868b md: Push down rec... |
3198 3199 3200 |
" -- starting background reconstruction ", mdname(mddev)); |
1da177e4c Linux-2.6.12-rc2 |
3201 |
printk(KERN_INFO |
128595ed6 md/raid10: tidy u... |
3202 3203 |
"md/raid10:%s: active with %d out of %d devices ", |
84707f38e md: don't use mdd... |
3204 3205 |
mdname(mddev), conf->raid_disks - mddev->degraded, conf->raid_disks); |
1da177e4c Linux-2.6.12-rc2 |
3206 3207 3208 |
/* * Ok, everything is just fine now */ |
dab8b2924 md: Add support f... |
3209 3210 3211 3212 |
mddev->dev_sectors = conf->dev_sectors; size = raid10_size(mddev, 0, 0); md_set_array_sectors(mddev, size); mddev->resync_max_sectors = size; |
1da177e4c Linux-2.6.12-rc2 |
3213 |
|
0d1292282 [PATCH] md: defin... |
3214 3215 |
mddev->queue->backing_dev_info.congested_fn = raid10_congested; mddev->queue->backing_dev_info.congested_data = mddev; |
7a5febe9f [PATCH] md: set t... |
3216 |
|
1da177e4c Linux-2.6.12-rc2 |
3217 3218 3219 3220 3221 |
/* Calculate max read-ahead size. * We need to readahead at least twice a whole stripe.... * maybe... */ { |
9d8f03636 md: Make mddev->c... |
3222 3223 |
int stripe = conf->raid_disks * ((mddev->chunk_sectors << 9) / PAGE_SIZE); |
1da177e4c Linux-2.6.12-rc2 |
3224 3225 3226 3227 |
stripe /= conf->near_copies; if (mddev->queue->backing_dev_info.ra_pages < 2* stripe) mddev->queue->backing_dev_info.ra_pages = 2* stripe; } |
84707f38e md: don't use mdd... |
3228 |
if (conf->near_copies < conf->raid_disks) |
1da177e4c Linux-2.6.12-rc2 |
3229 |
blk_queue_merge_bvec(mddev->queue, raid10_mergeable_bvec); |
a91a2785b block: Require su... |
3230 3231 3232 |
if (md_integrity_register(mddev)) goto out_free_conf; |
1da177e4c Linux-2.6.12-rc2 |
3233 3234 3235 |
return 0; out_free_conf: |
01f96c0a9 md: Avoid waking ... |
3236 |
md_unregister_thread(&mddev->thread); |
1da177e4c Linux-2.6.12-rc2 |
3237 3238 |
if (conf->r10bio_pool) mempool_destroy(conf->r10bio_pool); |
1345b1d8a [PATCH] md: defin... |
3239 |
safe_put_page(conf->tmppage); |
990a8baf5 [PATCH] md: remov... |
3240 |
kfree(conf->mirrors); |
1da177e4c Linux-2.6.12-rc2 |
3241 3242 3243 3244 3245 |
kfree(conf); mddev->private = NULL; out: return -EIO; } |
fd01b88c7 md: remove typede... |
3246 |
static int stop(struct mddev *mddev) |
1da177e4c Linux-2.6.12-rc2 |
3247 |
{ |
e879a8793 md/raid10: typede... |
3248 |
struct r10conf *conf = mddev->private; |
1da177e4c Linux-2.6.12-rc2 |
3249 |
|
409c57f38 md: enable suspen... |
3250 3251 |
raise_barrier(conf, 0); lower_barrier(conf); |
01f96c0a9 md: Avoid waking ... |
3252 |
md_unregister_thread(&mddev->thread); |
1da177e4c Linux-2.6.12-rc2 |
3253 3254 3255 |
blk_sync_queue(mddev->queue); /* the unplug fn references 'conf'*/ if (conf->r10bio_pool) mempool_destroy(conf->r10bio_pool); |
990a8baf5 [PATCH] md: remov... |
3256 |
kfree(conf->mirrors); |
1da177e4c Linux-2.6.12-rc2 |
3257 3258 3259 3260 |
kfree(conf); mddev->private = NULL; return 0; } |
fd01b88c7 md: remove typede... |
3261 |
static void raid10_quiesce(struct mddev *mddev, int state) |
6cce3b23f [PATCH] md: write... |
3262 |
{ |
e879a8793 md/raid10: typede... |
3263 |
struct r10conf *conf = mddev->private; |
6cce3b23f [PATCH] md: write... |
3264 3265 3266 3267 3268 3269 3270 3271 3272 |
switch(state) { case 1: raise_barrier(conf, 0); break; case 0: lower_barrier(conf); break; } |
6cce3b23f [PATCH] md: write... |
3273 |
} |
1da177e4c Linux-2.6.12-rc2 |
3274 |
|
fd01b88c7 md: remove typede... |
3275 |
static void *raid10_takeover_raid0(struct mddev *mddev) |
dab8b2924 md: Add support f... |
3276 |
{ |
3cb030020 md: removing type... |
3277 |
struct md_rdev *rdev; |
e879a8793 md/raid10: typede... |
3278 |
struct r10conf *conf; |
dab8b2924 md: Add support f... |
3279 3280 |
if (mddev->degraded > 0) { |
128595ed6 md/raid10: tidy u... |
3281 3282 3283 |
printk(KERN_ERR "md/raid10:%s: Error: degraded raid0! ", mdname(mddev)); |
dab8b2924 md: Add support f... |
3284 3285 |
return ERR_PTR(-EINVAL); } |
dab8b2924 md: Add support f... |
3286 3287 3288 3289 3290 3291 |
/* Set new parameters */ mddev->new_level = 10; /* new layout: far_copies = 1, near_copies = 2 */ mddev->new_layout = (1<<8) + 2; mddev->new_chunk_sectors = mddev->chunk_sectors; mddev->delta_disks = mddev->raid_disks; |
dab8b2924 md: Add support f... |
3292 3293 3294 3295 3296 |
mddev->raid_disks *= 2; /* make sure it will be not marked as dirty */ mddev->recovery_cp = MaxSector; conf = setup_conf(mddev); |
02214dc54 FIX: md: process ... |
3297 |
if (!IS_ERR(conf)) { |
e93f68a1f md: fix handling ... |
3298 3299 3300 |
list_for_each_entry(rdev, &mddev->disks, same_set) if (rdev->raid_disk >= 0) rdev->new_raid_disk = rdev->raid_disk * 2; |
02214dc54 FIX: md: process ... |
3301 3302 |
conf->barrier = 1; } |
dab8b2924 md: Add support f... |
3303 3304 |
return conf; } |
fd01b88c7 md: remove typede... |
3305 |
static void *raid10_takeover(struct mddev *mddev) |
dab8b2924 md: Add support f... |
3306 |
{ |
e373ab109 md/raid0: typedef... |
3307 |
struct r0conf *raid0_conf; |
dab8b2924 md: Add support f... |
3308 3309 3310 3311 3312 3313 |
/* raid10 can take over: * raid0 - providing it has only two drives */ if (mddev->level == 0) { /* for raid0 takeover only one zone is supported */ |
e373ab109 md/raid0: typedef... |
3314 3315 |
raid0_conf = mddev->private; if (raid0_conf->nr_strip_zones > 1) { |
128595ed6 md/raid10: tidy u... |
3316 3317 3318 3319 |
printk(KERN_ERR "md/raid10:%s: cannot takeover raid 0" " with more than one zone. ", mdname(mddev)); |
dab8b2924 md: Add support f... |
3320 3321 3322 3323 3324 3325 |
return ERR_PTR(-EINVAL); } return raid10_takeover_raid0(mddev); } return ERR_PTR(-EINVAL); } |
84fc4b56d md: rename "mdk_p... |
3326 |
static struct md_personality raid10_personality = |
1da177e4c Linux-2.6.12-rc2 |
3327 3328 |
{ .name = "raid10", |
2604b703b [PATCH] md: remov... |
3329 |
.level = 10, |
1da177e4c Linux-2.6.12-rc2 |
3330 3331 3332 3333 3334 3335 3336 3337 3338 3339 |
.owner = THIS_MODULE, .make_request = make_request, .run = run, .stop = stop, .status = status, .error_handler = error, .hot_add_disk = raid10_add_disk, .hot_remove_disk= raid10_remove_disk, .spare_active = raid10_spare_active, .sync_request = sync_request, |
6cce3b23f [PATCH] md: write... |
3340 |
.quiesce = raid10_quiesce, |
80c3a6ce4 md: add 'size' as... |
3341 |
.size = raid10_size, |
dab8b2924 md: Add support f... |
3342 |
.takeover = raid10_takeover, |
1da177e4c Linux-2.6.12-rc2 |
3343 3344 3345 3346 |
}; static int __init raid_init(void) { |
2604b703b [PATCH] md: remov... |
3347 |
return register_md_personality(&raid10_personality); |
1da177e4c Linux-2.6.12-rc2 |
3348 3349 3350 3351 |
} static void raid_exit(void) { |
2604b703b [PATCH] md: remov... |
3352 |
unregister_md_personality(&raid10_personality); |
1da177e4c Linux-2.6.12-rc2 |
3353 3354 3355 3356 3357 |
} module_init(raid_init); module_exit(raid_exit); MODULE_LICENSE("GPL"); |
0efb9e619 md: add MODULE_DE... |
3358 |
MODULE_DESCRIPTION("RAID10 (striped mirror) personality for MD"); |
1da177e4c Linux-2.6.12-rc2 |
3359 |
MODULE_ALIAS("md-personality-9"); /* RAID10 */ |
d9d166c2a [PATCH] md: allow... |
3360 |
MODULE_ALIAS("md-raid10"); |
2604b703b [PATCH] md: remov... |
3361 |
MODULE_ALIAS("md-level-10"); |
34db0cd60 md: add proper wr... |
3362 3363 |
module_param(max_queued_requests, int, S_IRUGO|S_IWUSR); |