Blame view
block/blk-flush.c
16.4 KB
86db1e297
|
1 |
/* |
4fed947cb
|
2 |
* Functions to sequence FLUSH and FUA writes. |
ae1b15396
|
3 4 5 6 7 8 9 10 11 12 |
* * Copyright (C) 2011 Max Planck Institute for Gravitational Physics * Copyright (C) 2011 Tejun Heo <tj@kernel.org> * * This file is released under the GPLv2. * * REQ_{FLUSH|FUA} requests are decomposed to sequences consisted of three * optional steps - PREFLUSH, DATA and POSTFLUSH - according to the request * properties and hardware capability. * |
28a8f0d31
|
13 14 |
* If a request doesn't have data, only REQ_PREFLUSH makes sense, which * indicates a simple flush request. If there is data, REQ_PREFLUSH indicates |
ae1b15396
|
15 16 17 18 19 20 21 22 |
* that the device cache should be flushed before the data is executed, and * REQ_FUA means that the data must be on non-volatile media on request * completion. * * If the device doesn't have writeback cache, FLUSH and FUA don't make any * difference. The requests are either completed immediately if there's no * data or executed as normal requests otherwise. * |
28a8f0d31
|
23 |
* If the device has writeback cache and supports FUA, REQ_PREFLUSH is |
ae1b15396
|
24 25 |
* translated to PREFLUSH but REQ_FUA is passed down directly with DATA. * |
28a8f0d31
|
26 27 |
* If the device has writeback cache and doesn't support FUA, REQ_PREFLUSH * is translated to PREFLUSH and REQ_FUA to POSTFLUSH. |
ae1b15396
|
28 29 30 |
* * The actual execution of flush is double buffered. Whenever a request * needs to execute PRE or POSTFLUSH, it queues at |
7c94e1c15
|
31 |
* fq->flush_queue[fq->flush_pending_idx]. Once certain criteria are met, a |
3a5e02ced
|
32 |
* REQ_OP_FLUSH is issued and the pending_idx is toggled. When the flush |
ae1b15396
|
33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 |
* completes, all the requests which were pending are proceeded to the next * step. This allows arbitrary merging of different types of FLUSH/FUA * requests. * * Currently, the following conditions are used to determine when to issue * flush. * * C1. At any given time, only one flush shall be in progress. This makes * double buffering sufficient. * * C2. Flush is deferred if any request is executing DATA of its sequence. * This avoids issuing separate POSTFLUSHes for requests which shared * PREFLUSH. * * C3. The second condition is ignored if there is a request which has * waited longer than FLUSH_PENDING_TIMEOUT. This is to avoid * starvation in the unlikely case where there are continuous stream of * FUA (without FLUSH) requests. * * For devices which support FUA, it isn't clear whether C2 (and thus C3) * is beneficial. * * Note that a sequenced FLUSH/FUA request with DATA is completed twice. * Once while executing DATA and again after the whole sequence is * complete. The first completion updates the contained bio but doesn't * finish it so that the bio submitter is notified only after the whole |
e80640213
|
59 |
* sequence is complete. This is implemented by testing RQF_FLUSH_SEQ in |
ae1b15396
|
60 61 62 63 64 |
* req_bio_endio(). * * The above peculiarity requires that each FLUSH/FUA request has only one * bio attached to it, which is guaranteed as they aren't allowed to be * merged in the usual way. |
86db1e297
|
65 |
*/ |
ae1b15396
|
66 |
|
86db1e297
|
67 68 69 70 |
#include <linux/kernel.h> #include <linux/module.h> #include <linux/bio.h> #include <linux/blkdev.h> |
5a0e3ad6a
|
71 |
#include <linux/gfp.h> |
320ae51fe
|
72 |
#include <linux/blk-mq.h> |
86db1e297
|
73 74 |
#include "blk.h" |
320ae51fe
|
75 |
#include "blk-mq.h" |
0048b4837
|
76 |
#include "blk-mq-tag.h" |
86db1e297
|
77 |
|
4fed947cb
|
78 79 |
/* FLUSH/FUA sequences */ enum { |
ae1b15396
|
80 81 82 83 84 85 86 87 88 89 90 91 92 |
REQ_FSEQ_PREFLUSH = (1 << 0), /* pre-flushing in progress */ REQ_FSEQ_DATA = (1 << 1), /* data write in progress */ REQ_FSEQ_POSTFLUSH = (1 << 2), /* post-flushing in progress */ REQ_FSEQ_DONE = (1 << 3), REQ_FSEQ_ACTIONS = REQ_FSEQ_PREFLUSH | REQ_FSEQ_DATA | REQ_FSEQ_POSTFLUSH, /* * If flush has been pending longer than the following timeout, * it's issued even if flush_data requests are still in flight. */ FLUSH_PENDING_TIMEOUT = 5 * HZ, |
4fed947cb
|
93 |
}; |
0bae352da
|
94 95 |
static bool blk_kick_flush(struct request_queue *q, struct blk_flush_queue *fq); |
28e7d1845
|
96 |
|
c888a8f95
|
97 |
static unsigned int blk_flush_policy(unsigned long fflags, struct request *rq) |
86db1e297
|
98 |
{ |
ae1b15396
|
99 |
unsigned int policy = 0; |
86db1e297
|
100 |
|
fa1bf42ff
|
101 102 |
if (blk_rq_sectors(rq)) policy |= REQ_FSEQ_DATA; |
c888a8f95
|
103 |
if (fflags & (1UL << QUEUE_FLAG_WC)) { |
28a8f0d31
|
104 |
if (rq->cmd_flags & REQ_PREFLUSH) |
ae1b15396
|
105 |
policy |= REQ_FSEQ_PREFLUSH; |
c888a8f95
|
106 107 |
if (!(fflags & (1UL << QUEUE_FLAG_FUA)) && (rq->cmd_flags & REQ_FUA)) |
ae1b15396
|
108 |
policy |= REQ_FSEQ_POSTFLUSH; |
28e7d1845
|
109 |
} |
ae1b15396
|
110 |
return policy; |
86db1e297
|
111 |
} |
ae1b15396
|
112 |
static unsigned int blk_flush_cur_seq(struct request *rq) |
47f70d5a6
|
113 |
{ |
ae1b15396
|
114 115 |
return 1 << ffz(rq->flush.seq); } |
47f70d5a6
|
116 |
|
ae1b15396
|
117 118 |
static void blk_flush_restore_request(struct request *rq) { |
47f70d5a6
|
119 |
/* |
ae1b15396
|
120 121 122 |
* After flush data completion, @rq->bio is %NULL but we need to * complete the bio again. @rq->biotail is guaranteed to equal the * original @rq->bio. Restore it. |
47f70d5a6
|
123 |
*/ |
ae1b15396
|
124 125 126 |
rq->bio = rq->biotail; /* make @rq a normal request */ |
e80640213
|
127 |
rq->rq_flags &= ~RQF_FLUSH_SEQ; |
4853abaae
|
128 |
rq->end_io = rq->flush.saved_end_io; |
320ae51fe
|
129 |
} |
10beafc19
|
130 |
static bool blk_flush_queue_rq(struct request *rq, bool add_front) |
320ae51fe
|
131 |
{ |
18741986a
|
132 |
if (rq->q->mq_ops) { |
2b053aca7
|
133 |
blk_mq_add_to_requeue_list(rq, add_front, true); |
18741986a
|
134 135 |
return false; } else { |
10beafc19
|
136 137 138 139 |
if (add_front) list_add(&rq->queuelist, &rq->q->queue_head); else list_add_tail(&rq->queuelist, &rq->q->queue_head); |
18741986a
|
140 141 |
return true; } |
47f70d5a6
|
142 |
} |
ae1b15396
|
143 144 145 |
/** * blk_flush_complete_seq - complete flush sequence * @rq: FLUSH/FUA request being sequenced |
0bae352da
|
146 |
* @fq: flush queue |
ae1b15396
|
147 148 149 150 151 152 153 |
* @seq: sequences to complete (mask of %REQ_FSEQ_*, can be zero) * @error: whether an error occurred * * @rq just completed @seq part of its flush sequence, record the * completion and trigger the next step. * * CONTEXT: |
7c94e1c15
|
154 |
* spin_lock_irq(q->queue_lock or fq->mq_flush_lock) |
ae1b15396
|
155 156 157 158 |
* * RETURNS: * %true if requests were added to the dispatch queue, %false otherwise. */ |
0bae352da
|
159 160 161 |
static bool blk_flush_complete_seq(struct request *rq, struct blk_flush_queue *fq, unsigned int seq, int error) |
86db1e297
|
162 |
{ |
ae1b15396
|
163 |
struct request_queue *q = rq->q; |
7c94e1c15
|
164 |
struct list_head *pending = &fq->flush_queue[fq->flush_pending_idx]; |
320ae51fe
|
165 |
bool queued = false, kicked; |
ae1b15396
|
166 167 168 169 170 171 172 173 174 175 176 177 178 179 |
BUG_ON(rq->flush.seq & seq); rq->flush.seq |= seq; if (likely(!error)) seq = blk_flush_cur_seq(rq); else seq = REQ_FSEQ_DONE; switch (seq) { case REQ_FSEQ_PREFLUSH: case REQ_FSEQ_POSTFLUSH: /* queue for flush */ if (list_empty(pending)) |
7c94e1c15
|
180 |
fq->flush_pending_since = jiffies; |
ae1b15396
|
181 182 183 184 |
list_move_tail(&rq->flush.list, pending); break; case REQ_FSEQ_DATA: |
7c94e1c15
|
185 |
list_move_tail(&rq->flush.list, &fq->flush_data_in_flight); |
10beafc19
|
186 |
queued = blk_flush_queue_rq(rq, true); |
ae1b15396
|
187 188 189 190 191 192 193 194 195 196 197 198 |
break; case REQ_FSEQ_DONE: /* * @rq was previously adjusted by blk_flush_issue() for * flush sequencing and may already have gone through the * flush data request completion path. Restore @rq for * normal completion and end it. */ BUG_ON(!list_empty(&rq->queuelist)); list_del_init(&rq->flush.list); blk_flush_restore_request(rq); |
320ae51fe
|
199 |
if (q->mq_ops) |
c8a446ad6
|
200 |
blk_mq_end_request(rq, error); |
320ae51fe
|
201 202 |
else __blk_end_request_all(rq, error); |
ae1b15396
|
203 204 205 206 207 |
break; default: BUG(); } |
0bae352da
|
208 |
kicked = blk_kick_flush(q, fq); |
320ae51fe
|
209 |
return kicked | queued; |
86db1e297
|
210 |
} |
ae1b15396
|
211 |
static void flush_end_io(struct request *flush_rq, int error) |
86db1e297
|
212 |
{ |
ae1b15396
|
213 |
struct request_queue *q = flush_rq->q; |
320ae51fe
|
214 |
struct list_head *running; |
ae1b15396
|
215 216 |
bool queued = false; struct request *rq, *n; |
320ae51fe
|
217 |
unsigned long flags = 0; |
e97c293cd
|
218 |
struct blk_flush_queue *fq = blk_get_flush_queue(q, flush_rq->mq_ctx); |
ae1b15396
|
219 |
|
223023750
|
220 |
if (q->mq_ops) { |
0048b4837
|
221 222 223 |
struct blk_mq_hw_ctx *hctx; /* release the tag's ownership to the req cloned from */ |
7c94e1c15
|
224 |
spin_lock_irqsave(&fq->mq_flush_lock, flags); |
7d7e0f90b
|
225 |
hctx = blk_mq_map_queue(q, flush_rq->mq_ctx->cpu); |
0048b4837
|
226 |
blk_mq_tag_set_rq(hctx, flush_rq->tag, fq->orig_rq); |
7ddab5de5
|
227 |
flush_rq->tag = -1; |
223023750
|
228 |
} |
18741986a
|
229 |
|
7c94e1c15
|
230 231 |
running = &fq->flush_queue[fq->flush_running_idx]; BUG_ON(fq->flush_pending_idx == fq->flush_running_idx); |
ae1b15396
|
232 233 |
/* account completion of the flush request */ |
7c94e1c15
|
234 |
fq->flush_running_idx ^= 1; |
320ae51fe
|
235 236 237 |
if (!q->mq_ops) elv_completed_request(q, flush_rq); |
ae1b15396
|
238 239 240 241 242 243 |
/* and push the waiting requests to the next stage */ list_for_each_entry_safe(rq, n, running, flush.list) { unsigned int seq = blk_flush_cur_seq(rq); BUG_ON(seq != REQ_FSEQ_PREFLUSH && seq != REQ_FSEQ_POSTFLUSH); |
0bae352da
|
244 |
queued |= blk_flush_complete_seq(rq, fq, seq, error); |
ae1b15396
|
245 |
} |
47f70d5a6
|
246 |
/* |
3ac0cc450
|
247 248 249 250 251 252 253 254 255 |
* Kick the queue to avoid stall for two cases: * 1. Moving a request silently to empty queue_head may stall the * queue. * 2. When flush request is running in non-queueable queue, the * queue is hold. Restart the queue after flush request is finished * to avoid stall. * This function is called from request completion path and calling * directly into request_fn may confuse the driver. Always use * kblockd. |
47f70d5a6
|
256 |
*/ |
7c94e1c15
|
257 |
if (queued || fq->flush_queue_delayed) { |
18741986a
|
258 259 |
WARN_ON(q->mq_ops); blk_run_queue_async(q); |
320ae51fe
|
260 |
} |
7c94e1c15
|
261 |
fq->flush_queue_delayed = 0; |
320ae51fe
|
262 |
if (q->mq_ops) |
7c94e1c15
|
263 |
spin_unlock_irqrestore(&fq->mq_flush_lock, flags); |
320ae51fe
|
264 |
} |
ae1b15396
|
265 266 267 |
/** * blk_kick_flush - consider issuing flush request * @q: request_queue being kicked |
0bae352da
|
268 |
* @fq: flush queue |
ae1b15396
|
269 270 271 272 273 |
* * Flush related states of @q have changed, consider issuing flush request. * Please read the comment at the top of this file for more info. * * CONTEXT: |
7c94e1c15
|
274 |
* spin_lock_irq(q->queue_lock or fq->mq_flush_lock) |
ae1b15396
|
275 276 277 278 |
* * RETURNS: * %true if flush was issued, %false otherwise. */ |
0bae352da
|
279 |
static bool blk_kick_flush(struct request_queue *q, struct blk_flush_queue *fq) |
86db1e297
|
280 |
{ |
7c94e1c15
|
281 |
struct list_head *pending = &fq->flush_queue[fq->flush_pending_idx]; |
ae1b15396
|
282 283 |
struct request *first_rq = list_first_entry(pending, struct request, flush.list); |
7c94e1c15
|
284 |
struct request *flush_rq = fq->flush_rq; |
ae1b15396
|
285 286 |
/* C1 described at the top of this file */ |
7c94e1c15
|
287 |
if (fq->flush_pending_idx != fq->flush_running_idx || list_empty(pending)) |
ae1b15396
|
288 289 290 |
return false; /* C2 and C3 */ |
7c94e1c15
|
291 |
if (!list_empty(&fq->flush_data_in_flight) && |
ae1b15396
|
292 |
time_before(jiffies, |
7c94e1c15
|
293 |
fq->flush_pending_since + FLUSH_PENDING_TIMEOUT)) |
ae1b15396
|
294 295 296 297 298 299 |
return false; /* * Issue flush and toggle pending_idx. This makes pending_idx * different from running_idx, which means flush is in flight. */ |
7c94e1c15
|
300 |
fq->flush_pending_idx ^= 1; |
18741986a
|
301 |
|
7ddab5de5
|
302 |
blk_rq_init(q, flush_rq); |
f70ced091
|
303 304 305 |
/* * Borrow tag from the first request since they can't |
0048b4837
|
306 307 |
* be in flight at the same time. And acquire the tag's * ownership for flush req. |
f70ced091
|
308 309 |
*/ if (q->mq_ops) { |
0048b4837
|
310 |
struct blk_mq_hw_ctx *hctx; |
f70ced091
|
311 312 |
flush_rq->mq_ctx = first_rq->mq_ctx; flush_rq->tag = first_rq->tag; |
0048b4837
|
313 |
fq->orig_rq = first_rq; |
7d7e0f90b
|
314 |
hctx = blk_mq_map_queue(q, first_rq->mq_ctx->cpu); |
0048b4837
|
315 |
blk_mq_tag_set_rq(hctx, first_rq->tag, flush_rq); |
f70ced091
|
316 |
} |
320ae51fe
|
317 |
|
7ddab5de5
|
318 |
flush_rq->cmd_type = REQ_TYPE_FS; |
70fd76140
|
319 |
flush_rq->cmd_flags = REQ_OP_FLUSH | REQ_PREFLUSH; |
e80640213
|
320 |
flush_rq->rq_flags |= RQF_FLUSH_SEQ; |
7ddab5de5
|
321 322 |
flush_rq->rq_disk = first_rq->rq_disk; flush_rq->end_io = flush_end_io; |
ae1b15396
|
323 |
|
7ddab5de5
|
324 |
return blk_flush_queue_rq(flush_rq, false); |
86db1e297
|
325 |
} |
ae1b15396
|
326 |
static void flush_data_end_io(struct request *rq, int error) |
86db1e297
|
327 |
{ |
ae1b15396
|
328 |
struct request_queue *q = rq->q; |
e97c293cd
|
329 |
struct blk_flush_queue *fq = blk_get_flush_queue(q, NULL); |
ae1b15396
|
330 |
|
e83a46bbb
|
331 |
/* |
94d7dea44
|
332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 |
* Updating q->in_flight[] here for making this tag usable * early. Because in blk_queue_start_tag(), * q->in_flight[BLK_RW_ASYNC] is used to limit async I/O and * reserve tags for sync I/O. * * More importantly this way can avoid the following I/O * deadlock: * * - suppose there are 40 fua requests comming to flush queue * and queue depth is 31 * - 30 rqs are scheduled then blk_queue_start_tag() can't alloc * tag for async I/O any more * - all the 30 rqs are completed before FLUSH_PENDING_TIMEOUT * and flush_data_end_io() is called * - the other rqs still can't go ahead if not updating * q->in_flight[BLK_RW_ASYNC] here, meantime these rqs * are held in flush data queue and make no progress of * handling post flush rq * - only after the post flush rq is handled, all these rqs * can be completed */ elv_completed_request(q, rq); /* for avoiding double accounting */ |
36869cb93
|
357 |
rq->rq_flags &= ~RQF_STARTED; |
94d7dea44
|
358 359 |
/* |
e83a46bbb
|
360 361 362 |
* After populating an empty queue, kick it to avoid stall. Read * the comment in flush_end_io(). */ |
0bae352da
|
363 |
if (blk_flush_complete_seq(rq, fq, REQ_FSEQ_DATA, error)) |
24ecfbe27
|
364 |
blk_run_queue_async(q); |
86db1e297
|
365 |
} |
320ae51fe
|
366 367 368 369 |
static void mq_flush_data_end_io(struct request *rq, int error) { struct request_queue *q = rq->q; struct blk_mq_hw_ctx *hctx; |
e97c293cd
|
370 |
struct blk_mq_ctx *ctx = rq->mq_ctx; |
320ae51fe
|
371 |
unsigned long flags; |
e97c293cd
|
372 |
struct blk_flush_queue *fq = blk_get_flush_queue(q, ctx); |
320ae51fe
|
373 |
|
7d7e0f90b
|
374 |
hctx = blk_mq_map_queue(q, ctx->cpu); |
320ae51fe
|
375 376 377 378 379 |
/* * After populating an empty queue, kick it to avoid stall. Read * the comment in flush_end_io(). */ |
7c94e1c15
|
380 |
spin_lock_irqsave(&fq->mq_flush_lock, flags); |
0bae352da
|
381 |
if (blk_flush_complete_seq(rq, fq, REQ_FSEQ_DATA, error)) |
320ae51fe
|
382 |
blk_mq_run_hw_queue(hctx, true); |
7c94e1c15
|
383 |
spin_unlock_irqrestore(&fq->mq_flush_lock, flags); |
320ae51fe
|
384 |
} |
ae1b15396
|
385 386 387 388 |
/** * blk_insert_flush - insert a new FLUSH/FUA request * @rq: request to insert * |
b710a4805
|
389 |
* To be called from __elv_add_request() for %ELEVATOR_INSERT_FLUSH insertions. |
320ae51fe
|
390 |
* or __blk_mq_run_hw_queue() to dispatch request. |
ae1b15396
|
391 392 393 394 |
* @rq is being submitted. Analyze what needs to be done and put it on the * right queue. * * CONTEXT: |
320ae51fe
|
395 |
* spin_lock_irq(q->queue_lock) in !mq case |
ae1b15396
|
396 397 |
*/ void blk_insert_flush(struct request *rq) |
86db1e297
|
398 |
{ |
ae1b15396
|
399 |
struct request_queue *q = rq->q; |
c888a8f95
|
400 |
unsigned long fflags = q->queue_flags; /* may change, cache */ |
ae1b15396
|
401 |
unsigned int policy = blk_flush_policy(fflags, rq); |
e97c293cd
|
402 |
struct blk_flush_queue *fq = blk_get_flush_queue(q, rq->mq_ctx); |
86db1e297
|
403 |
|
ae1b15396
|
404 405 |
/* * @policy now records what operations need to be done. Adjust |
28a8f0d31
|
406 |
* REQ_PREFLUSH and FUA for the driver. |
ae1b15396
|
407 |
*/ |
28a8f0d31
|
408 |
rq->cmd_flags &= ~REQ_PREFLUSH; |
c888a8f95
|
409 |
if (!(fflags & (1UL << QUEUE_FLAG_FUA))) |
ae1b15396
|
410 411 412 |
rq->cmd_flags &= ~REQ_FUA; /* |
ae5b2ec8a
|
413 414 415 416 417 418 419 |
* REQ_PREFLUSH|REQ_FUA implies REQ_SYNC, so if we clear any * of those flags, we have to set REQ_SYNC to avoid skewing * the request accounting. */ rq->cmd_flags |= REQ_SYNC; /* |
4853abaae
|
420 421 422 423 424 425 |
* An empty flush handed down from a stacking driver may * translate into nothing if the underlying device does not * advertise a write-back cache. In this case, simply * complete the request. */ if (!policy) { |
320ae51fe
|
426 |
if (q->mq_ops) |
c8a446ad6
|
427 |
blk_mq_end_request(rq, 0); |
320ae51fe
|
428 429 |
else __blk_end_bidi_request(rq, 0, 0, 0); |
4853abaae
|
430 431 |
return; } |
834f9f61a
|
432 |
BUG_ON(rq->bio != rq->biotail); /*assumes zero or single bio rq */ |
4853abaae
|
433 434 |
/* |
ae1b15396
|
435 436 437 438 439 440 |
* If there's data but flush is not necessary, the request can be * processed directly without going through flush machinery. Queue * for normal execution. */ if ((policy & REQ_FSEQ_DATA) && !(policy & (REQ_FSEQ_PREFLUSH | REQ_FSEQ_POSTFLUSH))) { |
320ae51fe
|
441 |
if (q->mq_ops) { |
c8e52ba5e
|
442 |
blk_mq_insert_request(rq, false, true, false); |
320ae51fe
|
443 |
} else |
dcd8376c3
|
444 |
list_add_tail(&rq->queuelist, &q->queue_head); |
ae1b15396
|
445 |
return; |
28e7d1845
|
446 |
} |
cde4c406d
|
447 |
|
ae1b15396
|
448 449 450 451 452 453 |
/* * @rq should go through flush machinery. Mark it part of flush * sequence and submit for further processing. */ memset(&rq->flush, 0, sizeof(rq->flush)); INIT_LIST_HEAD(&rq->flush.list); |
e80640213
|
454 |
rq->rq_flags |= RQF_FLUSH_SEQ; |
4853abaae
|
455 |
rq->flush.saved_end_io = rq->end_io; /* Usually NULL */ |
320ae51fe
|
456 457 |
if (q->mq_ops) { rq->end_io = mq_flush_data_end_io; |
7c94e1c15
|
458 |
spin_lock_irq(&fq->mq_flush_lock); |
0bae352da
|
459 |
blk_flush_complete_seq(rq, fq, REQ_FSEQ_ACTIONS & ~policy, 0); |
7c94e1c15
|
460 |
spin_unlock_irq(&fq->mq_flush_lock); |
320ae51fe
|
461 462 |
return; } |
ae1b15396
|
463 |
rq->end_io = flush_data_end_io; |
0bae352da
|
464 |
blk_flush_complete_seq(rq, fq, REQ_FSEQ_ACTIONS & ~policy, 0); |
86db1e297
|
465 |
} |
ae1b15396
|
466 |
/** |
86db1e297
|
467 468 |
* blkdev_issue_flush - queue a flush * @bdev: blockdev to issue flush for |
fbd9b09a1
|
469 |
* @gfp_mask: memory allocation flags (for bio_alloc) |
86db1e297
|
470 471 472 473 474 |
* @error_sector: error sector * * Description: * Issue a flush for the block device in question. Caller can supply * room for storing the error offset in case of a flush error, if they |
f17e232e9
|
475 476 |
* wish to. If WAIT flag is not passed then caller may check only what * request was pushed in some internal queue for later handling. |
86db1e297
|
477 |
*/ |
fbd9b09a1
|
478 |
int blkdev_issue_flush(struct block_device *bdev, gfp_t gfp_mask, |
dd3932edd
|
479 |
sector_t *error_sector) |
86db1e297
|
480 |
{ |
86db1e297
|
481 482 |
struct request_queue *q; struct bio *bio; |
fbd9b09a1
|
483 |
int ret = 0; |
86db1e297
|
484 485 486 487 488 489 490 |
if (bdev->bd_disk == NULL) return -ENXIO; q = bdev_get_queue(bdev); if (!q) return -ENXIO; |
f10d9f617
|
491 492 493 494 |
/* * some block devices may not have their queue correctly set up here * (e.g. loop device without a backing file) and so issuing a flush * here will panic. Ensure there is a request function before issuing |
d391a2dda
|
495 |
* the flush. |
f10d9f617
|
496 497 498 |
*/ if (!q->make_request_fn) return -ENXIO; |
fbd9b09a1
|
499 |
bio = bio_alloc(gfp_mask, 0); |
86db1e297
|
500 |
bio->bi_bdev = bdev; |
70fd76140
|
501 |
bio->bi_opf = REQ_OP_WRITE | REQ_PREFLUSH; |
86db1e297
|
502 |
|
4e49ea4a3
|
503 |
ret = submit_bio_wait(bio); |
dd3932edd
|
504 505 506 507 508 509 510 |
/* * The driver must store the error location in ->bi_sector, if * it supports it. For non-stacked drivers, this should be * copied from blk_rq_pos(rq). */ if (error_sector) |
4f024f379
|
511 |
*error_sector = bio->bi_iter.bi_sector; |
86db1e297
|
512 |
|
86db1e297
|
513 514 515 |
bio_put(bio); return ret; } |
86db1e297
|
516 |
EXPORT_SYMBOL(blkdev_issue_flush); |
320ae51fe
|
517 |
|
f70ced091
|
518 519 |
struct blk_flush_queue *blk_alloc_flush_queue(struct request_queue *q, int node, int cmd_size) |
320ae51fe
|
520 |
{ |
7c94e1c15
|
521 522 |
struct blk_flush_queue *fq; int rq_sz = sizeof(struct request); |
1bcb1eada
|
523 |
|
f70ced091
|
524 |
fq = kzalloc_node(sizeof(*fq), GFP_KERNEL, node); |
7c94e1c15
|
525 526 |
if (!fq) goto fail; |
1bcb1eada
|
527 |
|
7c94e1c15
|
528 529 |
if (q->mq_ops) { spin_lock_init(&fq->mq_flush_lock); |
f70ced091
|
530 |
rq_sz = round_up(rq_sz + cmd_size, cache_line_size()); |
7c94e1c15
|
531 |
} |
f70ced091
|
532 |
fq->flush_rq = kzalloc_node(rq_sz, GFP_KERNEL, node); |
7c94e1c15
|
533 534 535 536 537 538 539 540 541 542 543 544 545 |
if (!fq->flush_rq) goto fail_rq; INIT_LIST_HEAD(&fq->flush_queue[0]); INIT_LIST_HEAD(&fq->flush_queue[1]); INIT_LIST_HEAD(&fq->flush_data_in_flight); return fq; fail_rq: kfree(fq); fail: return NULL; |
320ae51fe
|
546 |
} |
f35526557
|
547 |
|
ba483388e
|
548 |
void blk_free_flush_queue(struct blk_flush_queue *fq) |
f35526557
|
549 |
{ |
7c94e1c15
|
550 551 552 |
/* bio based request queue hasn't flush queue */ if (!fq) return; |
3c09676c1
|
553 |
|
7c94e1c15
|
554 555 556 |
kfree(fq->flush_rq); kfree(fq); } |