]>
Commit | Line | Data |
---|---|---|
1da177e4 LT |
1 | #ifndef _RAID5_H |
2 | #define _RAID5_H | |
3 | ||
1da177e4 | 4 | #include <linux/raid/xor.h> |
ad283ea4 | 5 | #include <linux/dmaengine.h> |
1da177e4 LT |
6 | |
7 | /* | |
8 | * | |
c4c1663b | 9 | * Each stripe contains one buffer per device. Each buffer can be in |
1da177e4 | 10 | * one of a number of states stored in "flags". Changes between |
c4c1663b N |
11 | * these states happen *almost* exclusively under the protection of the |
12 | * STRIPE_ACTIVE flag. Some very specific changes can happen in bi_end_io, and | |
13 | * these are not protected by STRIPE_ACTIVE. | |
1da177e4 LT |
14 | * |
15 | * The flag bits that are used to represent these states are: | |
16 | * R5_UPTODATE and R5_LOCKED | |
17 | * | |
18 | * State Empty == !UPTODATE, !LOCK | |
19 | * We have no data, and there is no active request | |
20 | * State Want == !UPTODATE, LOCK | |
21 | * A read request is being submitted for this block | |
22 | * State Dirty == UPTODATE, LOCK | |
23 | * Some new data is in this buffer, and it is being written out | |
24 | * State Clean == UPTODATE, !LOCK | |
25 | * We have valid data which is the same as on disc | |
26 | * | |
27 | * The possible state transitions are: | |
28 | * | |
29 | * Empty -> Want - on read or write to get old data for parity calc | |
30 | * Empty -> Dirty - on compute_parity to satisfy write/sync request.(RECONSTRUCT_WRITE) | |
31 | * Empty -> Clean - on compute_block when computing a block for failed drive | |
32 | * Want -> Empty - on failed read | |
33 | * Want -> Clean - on successful completion of read request | |
34 | * Dirty -> Clean - on successful completion of write request | |
35 | * Dirty -> Clean - on failed write | |
36 | * Clean -> Dirty - on compute_parity to satisfy write/sync (RECONSTRUCT or RMW) | |
37 | * | |
38 | * The Want->Empty, Want->Clean, Dirty->Clean, transitions | |
39 | * all happen in b_end_io at interrupt time. | |
40 | * Each sets the Uptodate bit before releasing the Lock bit. | |
41 | * This leaves one multi-stage transition: | |
42 | * Want->Dirty->Clean | |
43 | * This is safe because thinking that a Clean buffer is actually dirty | |
44 | * will at worst delay some action, and the stripe will be scheduled | |
45 | * for attention after the transition is complete. | |
46 | * | |
47 | * There is one possibility that is not covered by these states. That | |
48 | * is if one drive has failed and there is a spare being rebuilt. We | |
49 | * can't distinguish between a clean block that has been generated | |
50 | * from parity calculations, and a clean block that has been | |
51 | * successfully written to the spare ( or to parity when resyncing). | |
52 | * To distingush these states we have a stripe bit STRIPE_INSYNC that | |
53 | * is set whenever a write is scheduled to the spare, or to the parity | |
54 | * disc if there is no spare. A sync request clears this bit, and | |
55 | * when we find it set with no buffers locked, we know the sync is | |
56 | * complete. | |
57 | * | |
58 | * Buffers for the md device that arrive via make_request are attached | |
59 | * to the appropriate stripe in one of two lists linked on b_reqnext. | |
60 | * One list (bh_read) for read requests, one (bh_write) for write. | |
61 | * There should never be more than one buffer on the two lists | |
62 | * together, but we are not guaranteed of that so we allow for more. | |
63 | * | |
64 | * If a buffer is on the read list when the associated cache buffer is | |
65 | * Uptodate, the data is copied into the read buffer and it's b_end_io | |
66 | * routine is called. This may happen in the end_request routine only | |
67 | * if the buffer has just successfully been read. end_request should | |
68 | * remove the buffers from the list and then set the Uptodate bit on | |
69 | * the buffer. Other threads may do this only if they first check | |
70 | * that the Uptodate bit is set. Once they have checked that they may | |
71 | * take buffers off the read queue. | |
72 | * | |
73 | * When a buffer on the write list is committed for write it is copied | |
74 | * into the cache buffer, which is then marked dirty, and moved onto a | |
75 | * third list, the written list (bh_written). Once both the parity | |
76 | * block and the cached buffer are successfully written, any buffer on | |
77 | * a written list can be returned with b_end_io. | |
78 | * | |
c4c1663b N |
79 | * The write list and read list both act as fifos. The read list, |
80 | * write list and written list are protected by the device_lock. | |
81 | * The device_lock is only for list manipulations and will only be | |
82 | * held for a very short time. It can be claimed from interrupts. | |
1da177e4 LT |
83 | * |
84 | * | |
85 | * Stripes in the stripe cache can be on one of two lists (or on | |
86 | * neither). The "inactive_list" contains stripes which are not | |
87 | * currently being used for any request. They can freely be reused | |
88 | * for another stripe. The "handle_list" contains stripes that need | |
89 | * to be handled in some way. Both of these are fifo queues. Each | |
90 | * stripe is also (potentially) linked to a hash bucket in the hash | |
91 | * table so that it can be found by sector number. Stripes that are | |
92 | * not hashed must be on the inactive_list, and will normally be at | |
93 | * the front. All stripes start life this way. | |
94 | * | |
95 | * The inactive_list, handle_list and hash bucket lists are all protected by the | |
96 | * device_lock. | |
1da177e4 LT |
97 | * - stripes have a reference counter. If count==0, they are on a list. |
98 | * - If a stripe might need handling, STRIPE_HANDLE is set. | |
99 | * - When refcount reaches zero, then if STRIPE_HANDLE it is put on | |
100 | * handle_list else inactive_list | |
101 | * | |
102 | * This, combined with the fact that STRIPE_HANDLE is only ever | |
103 | * cleared while a stripe has a non-zero count means that if the | |
104 | * refcount is 0 and STRIPE_HANDLE is set, then it is on the | |
105 | * handle_list and if recount is 0 and STRIPE_HANDLE is not set, then | |
106 | * the stripe is on inactive_list. | |
107 | * | |
108 | * The possible transitions are: | |
109 | * activate an unhashed/inactive stripe (get_active_stripe()) | |
110 | * lockdev check-hash unlink-stripe cnt++ clean-stripe hash-stripe unlockdev | |
111 | * activate a hashed, possibly active stripe (get_active_stripe()) | |
112 | * lockdev check-hash if(!cnt++)unlink-stripe unlockdev | |
113 | * attach a request to an active stripe (add_stripe_bh()) | |
114 | * lockdev attach-buffer unlockdev | |
115 | * handle a stripe (handle_stripe()) | |
c4c1663b | 116 | * setSTRIPE_ACTIVE, clrSTRIPE_HANDLE ... |
91c00924 DW |
117 | * (lockdev check-buffers unlockdev) .. |
118 | * change-state .. | |
c4c1663b | 119 | * record io/ops needed clearSTRIPE_ACTIVE schedule io/ops |
1da177e4 LT |
120 | * release an active stripe (release_stripe()) |
121 | * lockdev if (!--cnt) { if STRIPE_HANDLE, add to handle_list else add to inactive-list } unlockdev | |
122 | * | |
123 | * The refcount counts each thread that have activated the stripe, | |
124 | * plus raid5d if it is handling it, plus one for each active request | |
91c00924 DW |
125 | * on a cached buffer, and plus one if the stripe is undergoing stripe |
126 | * operations. | |
127 | * | |
c4c1663b | 128 | * The stripe operations are: |
91c00924 DW |
129 | * -copying data between the stripe cache and user application buffers |
130 | * -computing blocks to save a disk access, or to recover a missing block | |
131 | * -updating the parity on a write operation (reconstruct write and | |
132 | * read-modify-write) | |
133 | * -checking parity correctness | |
134 | * -running i/o to disk | |
135 | * These operations are carried out by raid5_run_ops which uses the async_tx | |
136 | * api to (optionally) offload operations to dedicated hardware engines. | |
137 | * When requesting an operation handle_stripe sets the pending bit for the | |
138 | * operation and increments the count. raid5_run_ops is then run whenever | |
139 | * the count is non-zero. | |
140 | * There are some critical dependencies between the operations that prevent some | |
141 | * from being requested while another is in flight. | |
142 | * 1/ Parity check operations destroy the in cache version of the parity block, | |
143 | * so we prevent parity dependent operations like writes and compute_blocks | |
144 | * from starting while a check is in progress. Some dma engines can perform | |
145 | * the check without damaging the parity block, in these cases the parity | |
146 | * block is re-marked up to date (assuming the check was successful) and is | |
147 | * not re-read from disk. | |
148 | * 2/ When a write operation is requested we immediately lock the affected | |
149 | * blocks, and mark them as not up to date. This causes new read requests | |
150 | * to be held off, as well as parity checks and compute block operations. | |
151 | * 3/ Once a compute block operation has been requested handle_stripe treats | |
152 | * that block as if it is up to date. raid5_run_ops guaruntees that any | |
153 | * operation that is dependent on the compute block result is initiated after | |
154 | * the compute block completes. | |
1da177e4 LT |
155 | */ |
156 | ||
ecc65c9b | 157 | /* |
c4c1663b N |
158 | * Operations state - intermediate states that are visible outside of |
159 | * STRIPE_ACTIVE. | |
ecc65c9b DW |
160 | * In general _idle indicates nothing is running, _run indicates a data |
161 | * processing operation is active, and _result means the data processing result | |
162 | * is stable and can be acted upon. For simple operations like biofill and | |
163 | * compute that only have an _idle and _run state they are indicated with | |
164 | * sh->state flags (STRIPE_BIOFILL_RUN and STRIPE_COMPUTE_RUN) | |
165 | */ | |
166 | /** | |
167 | * enum check_states - handles syncing / repairing a stripe | |
168 | * @check_state_idle - check operations are quiesced | |
169 | * @check_state_run - check operation is running | |
170 | * @check_state_result - set outside lock when check result is valid | |
171 | * @check_state_compute_run - check failed and we are repairing | |
172 | * @check_state_compute_result - set outside lock when compute result is valid | |
173 | */ | |
174 | enum check_states { | |
175 | check_state_idle = 0, | |
ac6b53b6 DW |
176 | check_state_run, /* xor parity check */ |
177 | check_state_run_q, /* q-parity check */ | |
178 | check_state_run_pq, /* pq dual parity check */ | |
ecc65c9b DW |
179 | check_state_check_result, |
180 | check_state_compute_run, /* parity repair */ | |
181 | check_state_compute_result, | |
182 | }; | |
183 | ||
184 | /** | |
185 | * enum reconstruct_states - handles writing or expanding a stripe | |
186 | */ | |
187 | enum reconstruct_states { | |
188 | reconstruct_state_idle = 0, | |
d8ee0728 | 189 | reconstruct_state_prexor_drain_run, /* prexor-write */ |
ecc65c9b DW |
190 | reconstruct_state_drain_run, /* write */ |
191 | reconstruct_state_run, /* expand */ | |
d8ee0728 | 192 | reconstruct_state_prexor_drain_result, |
ecc65c9b DW |
193 | reconstruct_state_drain_result, |
194 | reconstruct_state_result, | |
195 | }; | |
196 | ||
1da177e4 | 197 | struct stripe_head { |
fccddba0 | 198 | struct hlist_node hash; |
d0dabf7e N |
199 | struct list_head lru; /* inactive_list or handle_list */ |
200 | struct raid5_private_data *raid_conf; | |
86b42c71 N |
201 | short generation; /* increments with every |
202 | * reshape */ | |
d0dabf7e N |
203 | sector_t sector; /* sector of this row */ |
204 | short pd_idx; /* parity disk index */ | |
205 | short qd_idx; /* 'Q' disk index for raid6 */ | |
67cc2b81 | 206 | short ddf_layout;/* use DDF ordering to calculate Q */ |
d0dabf7e N |
207 | unsigned long state; /* state flags */ |
208 | atomic_t count; /* nr of active thread/requests */ | |
72626685 | 209 | int bm_seq; /* sequence number for bitmap flushes */ |
d0dabf7e | 210 | int disks; /* disks in stripe */ |
ecc65c9b | 211 | enum check_states check_state; |
600aa109 | 212 | enum reconstruct_states reconstruct_state; |
417b8d4a DW |
213 | /** |
214 | * struct stripe_operations | |
91c00924 | 215 | * @target - STRIPE_OP_COMPUTE_BLK target |
417b8d4a DW |
216 | * @target2 - 2nd compute target in the raid6 case |
217 | * @zero_sum_result - P and Q verification flags | |
218 | * @request - async service request flags for raid_run_ops | |
91c00924 DW |
219 | */ |
220 | struct stripe_operations { | |
ac6b53b6 | 221 | int target, target2; |
ad283ea4 | 222 | enum sum_check_flags zero_sum_result; |
417b8d4a DW |
223 | #ifdef CONFIG_MULTICORE_RAID456 |
224 | unsigned long request; | |
225 | wait_queue_head_t wait_for_ops; | |
226 | #endif | |
91c00924 | 227 | } ops; |
1da177e4 LT |
228 | struct r5dev { |
229 | struct bio req; | |
230 | struct bio_vec vec; | |
231 | struct page *page; | |
91c00924 | 232 | struct bio *toread, *read, *towrite, *written; |
1da177e4 LT |
233 | sector_t sector; /* sector of this page */ |
234 | unsigned long flags; | |
235 | } dev[1]; /* allocated with extra space depending of RAID geometry */ | |
236 | }; | |
a4456856 DW |
237 | |
238 | /* stripe_head_state - collects and tracks the dynamic state of a stripe_head | |
c4c1663b | 239 | * for handle_stripe. |
a4456856 DW |
240 | */ |
241 | struct stripe_head_state { | |
242 | int syncing, expanding, expanded; | |
243 | int locked, uptodate, to_read, to_write, failed, written; | |
b5e98d65 | 244 | int to_fill, compute, req_compute, non_overwrite; |
f2b3b44d | 245 | int failed_num[2]; |
f2b3b44d | 246 | int p_failed, q_failed; |
c5709ef6 N |
247 | int dec_preread_active; |
248 | unsigned long ops_request; | |
249 | ||
250 | struct bio *return_bi; | |
251 | mdk_rdev_t *blocked_rdev; | |
a4456856 DW |
252 | }; |
253 | ||
1da177e4 LT |
254 | /* Flags */ |
255 | #define R5_UPTODATE 0 /* page contains current data */ | |
256 | #define R5_LOCKED 1 /* IO has been submitted on "req" */ | |
257 | #define R5_OVERWRITE 2 /* towrite covers whole page */ | |
258 | /* and some that are internal to handle_stripe */ | |
259 | #define R5_Insync 3 /* rdev && rdev->in_sync at start */ | |
260 | #define R5_Wantread 4 /* want to schedule a read */ | |
261 | #define R5_Wantwrite 5 | |
1da177e4 | 262 | #define R5_Overlap 7 /* There is a pending overlapping request on this block */ |
4e5314b5 N |
263 | #define R5_ReadError 8 /* seen a read error here recently */ |
264 | #define R5_ReWrite 9 /* have tried to over-write the readerror */ | |
1da177e4 | 265 | |
ccfcc3c1 | 266 | #define R5_Expanded 10 /* This block now has post-expand data */ |
91c00924 DW |
267 | #define R5_Wantcompute 11 /* compute_block in progress treat as |
268 | * uptodate | |
269 | */ | |
270 | #define R5_Wantfill 12 /* dev->toread contains a bio that needs | |
271 | * filling | |
272 | */ | |
d8ee0728 | 273 | #define R5_Wantdrain 13 /* dev->towrite needs to be drained */ |
e9c7469b | 274 | #define R5_WantFUA 14 /* Write should be FUA */ |
1da177e4 LT |
275 | /* |
276 | * Write method | |
277 | */ | |
278 | #define RECONSTRUCT_WRITE 1 | |
279 | #define READ_MODIFY_WRITE 2 | |
280 | /* not a write method, but a compute_parity mode */ | |
281 | #define CHECK_PARITY 3 | |
f701d589 DW |
282 | /* Additional compute_parity mode -- updates the parity w/o LOCKING */ |
283 | #define UPDATE_PARITY 4 | |
1da177e4 LT |
284 | |
285 | /* | |
286 | * Stripe state | |
287 | */ | |
83206d66 | 288 | enum { |
c4c1663b | 289 | STRIPE_ACTIVE, |
83206d66 N |
290 | STRIPE_HANDLE, |
291 | STRIPE_SYNC_REQUESTED, | |
292 | STRIPE_SYNCING, | |
293 | STRIPE_INSYNC, | |
294 | STRIPE_PREREAD_ACTIVE, | |
295 | STRIPE_DELAYED, | |
296 | STRIPE_DEGRADED, | |
297 | STRIPE_BIT_DELAY, | |
298 | STRIPE_EXPANDING, | |
299 | STRIPE_EXPAND_SOURCE, | |
300 | STRIPE_EXPAND_READY, | |
301 | STRIPE_IO_STARTED, /* do not count towards 'bypass_count' */ | |
302 | STRIPE_FULL_WRITE, /* all blocks are set to be overwritten */ | |
303 | STRIPE_BIOFILL_RUN, | |
304 | STRIPE_COMPUTE_RUN, | |
305 | STRIPE_OPS_REQ_PENDING, | |
306 | }; | |
417b8d4a | 307 | |
91c00924 | 308 | /* |
ecc65c9b | 309 | * Operation request flags |
91c00924 DW |
310 | */ |
311 | #define STRIPE_OP_BIOFILL 0 | |
312 | #define STRIPE_OP_COMPUTE_BLK 1 | |
313 | #define STRIPE_OP_PREXOR 2 | |
314 | #define STRIPE_OP_BIODRAIN 3 | |
ac6b53b6 | 315 | #define STRIPE_OP_RECONSTRUCT 4 |
91c00924 | 316 | #define STRIPE_OP_CHECK 5 |
91c00924 | 317 | |
1da177e4 LT |
318 | /* |
319 | * Plugging: | |
320 | * | |
321 | * To improve write throughput, we need to delay the handling of some | |
322 | * stripes until there has been a chance that several write requests | |
323 | * for the one stripe have all been collected. | |
324 | * In particular, any write request that would require pre-reading | |
325 | * is put on a "delayed" queue until there are no stripes currently | |
326 | * in a pre-read phase. Further, if the "delayed" queue is empty when | |
327 | * a stripe is put on it then we "plug" the queue and do not process it | |
328 | * until an unplug call is made. (the unplug_io_fn() is called). | |
329 | * | |
330 | * When preread is initiated on a stripe, we set PREREAD_ACTIVE and add | |
331 | * it to the count of prereading stripes. | |
332 | * When write is initiated, or the stripe refcnt == 0 (just in case) we | |
333 | * clear the PREREAD_ACTIVE flag and decrement the count | |
b5c124af N |
334 | * Whenever the 'handle' queue is empty and the device is not plugged, we |
335 | * move any strips from delayed to handle and clear the DELAYED flag and set | |
336 | * PREREAD_ACTIVE. | |
1da177e4 LT |
337 | * In stripe_handle, if we find pre-reading is necessary, we do it if |
338 | * PREREAD_ACTIVE is set, else we set DELAYED which will send it to the delayed queue. | |
c4c1663b | 339 | * HANDLE gets cleared if stripe_handle leaves nothing locked. |
1da177e4 | 340 | */ |
ef740c37 | 341 | |
1da177e4 LT |
342 | |
343 | struct disk_info { | |
344 | mdk_rdev_t *rdev; | |
345 | }; | |
346 | ||
347 | struct raid5_private_data { | |
fccddba0 | 348 | struct hlist_head *stripe_hashtbl; |
1da177e4 LT |
349 | mddev_t *mddev; |
350 | struct disk_info *spare; | |
09c9e5fa AN |
351 | int chunk_sectors; |
352 | int level, algorithm; | |
16a53ecc | 353 | int max_degraded; |
02c2de8c | 354 | int raid_disks; |
1da177e4 LT |
355 | int max_nr_stripes; |
356 | ||
fef9c61f N |
357 | /* reshape_progress is the leading edge of a 'reshape' |
358 | * It has value MaxSector when no reshape is happening | |
359 | * If delta_disks < 0, it is the last sector we started work on, | |
360 | * else is it the next sector to work on. | |
361 | */ | |
362 | sector_t reshape_progress; | |
363 | /* reshape_safe is the trailing edge of a reshape. We know that | |
364 | * before (or after) this address, all reshape has completed. | |
365 | */ | |
366 | sector_t reshape_safe; | |
7ecaa1e6 | 367 | int previous_raid_disks; |
09c9e5fa AN |
368 | int prev_chunk_sectors; |
369 | int prev_algo; | |
86b42c71 | 370 | short generation; /* increments with every reshape */ |
c8f517c4 N |
371 | unsigned long reshape_checkpoint; /* Time we last updated |
372 | * metadata */ | |
7ecaa1e6 | 373 | |
1da177e4 | 374 | struct list_head handle_list; /* stripes needing handling */ |
8b3e6cdc | 375 | struct list_head hold_list; /* preread ready stripes */ |
1da177e4 | 376 | struct list_head delayed_list; /* stripes that have plugged requests */ |
72626685 | 377 | struct list_head bitmap_list; /* stripes delaying awaiting bitmap update */ |
46031f9a RBJ |
378 | struct bio *retry_read_aligned; /* currently retrying aligned bios */ |
379 | struct bio *retry_read_aligned_list; /* aligned bios retry list */ | |
1da177e4 | 380 | atomic_t preread_active_stripes; /* stripes with scheduled io */ |
46031f9a | 381 | atomic_t active_aligned_reads; |
8b3e6cdc DW |
382 | atomic_t pending_full_writes; /* full write backlog */ |
383 | int bypass_count; /* bypassed prereads */ | |
384 | int bypass_threshold; /* preread nice */ | |
385 | struct list_head *last_hold; /* detect hold_list promotions */ | |
1da177e4 | 386 | |
f6705578 | 387 | atomic_t reshape_stripes; /* stripes with pending writes for reshape */ |
ad01c9e3 N |
388 | /* unfortunately we need two cache names as we temporarily have |
389 | * two caches. | |
390 | */ | |
391 | int active_name; | |
f4be6b43 | 392 | char cache_name[2][32]; |
e18b890b | 393 | struct kmem_cache *slab_cache; /* for allocating stripes */ |
72626685 N |
394 | |
395 | int seq_flush, seq_write; | |
396 | int quiesce; | |
397 | ||
398 | int fullsync; /* set to 1 if a full sync is needed, | |
399 | * (fresh device added). | |
400 | * Cleared when a sync completes. | |
401 | */ | |
2ac87401 | 402 | |
36d1c647 DW |
403 | /* per cpu variables */ |
404 | struct raid5_percpu { | |
405 | struct page *spare_page; /* Used when checking P/Q in raid6 */ | |
d6f38f31 DW |
406 | void *scribble; /* space for constructing buffer |
407 | * lists and performing address | |
408 | * conversions | |
409 | */ | |
a29d8b8e | 410 | } __percpu *percpu; |
d6f38f31 DW |
411 | size_t scribble_len; /* size of scribble region must be |
412 | * associated with conf to handle | |
413 | * cpu hotplug while reshaping | |
414 | */ | |
36d1c647 DW |
415 | #ifdef CONFIG_HOTPLUG_CPU |
416 | struct notifier_block cpu_notify; | |
417 | #endif | |
ca65b73b | 418 | |
1da177e4 LT |
419 | /* |
420 | * Free stripes pool | |
421 | */ | |
422 | atomic_t active_stripes; | |
423 | struct list_head inactive_list; | |
424 | wait_queue_head_t wait_for_stripe; | |
425 | wait_queue_head_t wait_for_overlap; | |
426 | int inactive_blocked; /* release of inactive stripes blocked, | |
427 | * waiting for 25% to be free | |
ad01c9e3 N |
428 | */ |
429 | int pool_size; /* number of disks in stripeheads in pool */ | |
1da177e4 | 430 | spinlock_t device_lock; |
b55e6bfc | 431 | struct disk_info *disks; |
91adb564 N |
432 | |
433 | /* When taking over an array from a different personality, we store | |
434 | * the new thread here until we fully activate the array. | |
435 | */ | |
436 | struct mdk_thread_s *thread; | |
1da177e4 LT |
437 | }; |
438 | ||
439 | typedef struct raid5_private_data raid5_conf_t; | |
440 | ||
1da177e4 LT |
441 | /* |
442 | * Our supported algorithms | |
443 | */ | |
99c0fb5f N |
444 | #define ALGORITHM_LEFT_ASYMMETRIC 0 /* Rotating Parity N with Data Restart */ |
445 | #define ALGORITHM_RIGHT_ASYMMETRIC 1 /* Rotating Parity 0 with Data Restart */ | |
446 | #define ALGORITHM_LEFT_SYMMETRIC 2 /* Rotating Parity N with Data Continuation */ | |
447 | #define ALGORITHM_RIGHT_SYMMETRIC 3 /* Rotating Parity 0 with Data Continuation */ | |
1da177e4 | 448 | |
99c0fb5f N |
449 | /* Define non-rotating (raid4) algorithms. These allow |
450 | * conversion of raid4 to raid5. | |
451 | */ | |
452 | #define ALGORITHM_PARITY_0 4 /* P or P,Q are initial devices */ | |
453 | #define ALGORITHM_PARITY_N 5 /* P or P,Q are final devices. */ | |
454 | ||
455 | /* DDF RAID6 layouts differ from md/raid6 layouts in two ways. | |
456 | * Firstly, the exact positioning of the parity block is slightly | |
457 | * different between the 'LEFT_*' modes of md and the "_N_*" modes | |
458 | * of DDF. | |
459 | * Secondly, or order of datablocks over which the Q syndrome is computed | |
460 | * is different. | |
461 | * Consequently we have different layouts for DDF/raid6 than md/raid6. | |
462 | * These layouts are from the DDFv1.2 spec. | |
463 | * Interestingly DDFv1.2-Errata-A does not specify N_CONTINUE but | |
464 | * leaves RLQ=3 as 'Vendor Specific' | |
465 | */ | |
466 | ||
467 | #define ALGORITHM_ROTATING_ZERO_RESTART 8 /* DDF PRL=6 RLQ=1 */ | |
468 | #define ALGORITHM_ROTATING_N_RESTART 9 /* DDF PRL=6 RLQ=2 */ | |
469 | #define ALGORITHM_ROTATING_N_CONTINUE 10 /*DDF PRL=6 RLQ=3 */ | |
470 | ||
471 | ||
472 | /* For every RAID5 algorithm we define a RAID6 algorithm | |
473 | * with exactly the same layout for data and parity, and | |
474 | * with the Q block always on the last device (N-1). | |
475 | * This allows trivial conversion from RAID5 to RAID6 | |
476 | */ | |
477 | #define ALGORITHM_LEFT_ASYMMETRIC_6 16 | |
478 | #define ALGORITHM_RIGHT_ASYMMETRIC_6 17 | |
479 | #define ALGORITHM_LEFT_SYMMETRIC_6 18 | |
480 | #define ALGORITHM_RIGHT_SYMMETRIC_6 19 | |
481 | #define ALGORITHM_PARITY_0_6 20 | |
482 | #define ALGORITHM_PARITY_N_6 ALGORITHM_PARITY_N | |
483 | ||
484 | static inline int algorithm_valid_raid5(int layout) | |
485 | { | |
486 | return (layout >= 0) && | |
487 | (layout <= 5); | |
488 | } | |
489 | static inline int algorithm_valid_raid6(int layout) | |
490 | { | |
491 | return (layout >= 0 && layout <= 5) | |
492 | || | |
e4424fee | 493 | (layout >= 8 && layout <= 10) |
99c0fb5f N |
494 | || |
495 | (layout >= 16 && layout <= 20); | |
496 | } | |
497 | ||
498 | static inline int algorithm_is_DDF(int layout) | |
499 | { | |
500 | return layout >= 8 && layout <= 10; | |
501 | } | |
11d8a6e3 N |
502 | |
503 | extern int md_raid5_congested(mddev_t *mddev, int bits); | |
7eaceacc | 504 | extern void md_raid5_kick_device(raid5_conf_t *conf); |
c41d4ac4 | 505 | extern int raid5_set_cache_size(mddev_t *mddev, int size); |
1da177e4 | 506 | #endif |