scsi_lib.c 56.7 KB
Newer Older
Linus Torvalds's avatar
Linus Torvalds committed
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
/*
 *  scsi_lib.c Copyright (C) 1999 Eric Youngdale
 *
 *  SCSI queueing library.
 *      Initial versions: Eric Youngdale (eric@andante.org).
 *                        Based upon conversations with large numbers
 *                        of people at Linux Expo.
 */

#include <linux/bio.h>
#include <linux/blkdev.h>
#include <linux/completion.h>
#include <linux/kernel.h>
#include <linux/mempool.h>
#include <linux/slab.h>
#include <linux/init.h>
#include <linux/pci.h>
#include <linux/delay.h>
19
#include <linux/hardirq.h>
Linus Torvalds's avatar
Linus Torvalds committed
20
21

#include <scsi/scsi.h>
22
#include <scsi/scsi_cmnd.h>
Linus Torvalds's avatar
Linus Torvalds committed
23
24
25
26
27
28
29
30
31
32
#include <scsi/scsi_dbg.h>
#include <scsi/scsi_device.h>
#include <scsi/scsi_driver.h>
#include <scsi/scsi_eh.h>
#include <scsi/scsi_host.h>

#include "scsi_priv.h"
#include "scsi_logging.h"


33
#define SG_MEMPOOL_NR		ARRAY_SIZE(scsi_sg_pools)
34
#define SG_MEMPOOL_SIZE		2
Linus Torvalds's avatar
Linus Torvalds committed
35
36
37
38

struct scsi_host_sg_pool {
	size_t		size;
	char		*name; 
39
	struct kmem_cache	*slab;
Linus Torvalds's avatar
Linus Torvalds committed
40
41
42
43
44
45
46
47
	mempool_t	*pool;
};

#if (SCSI_MAX_PHYS_SEGMENTS < 32)
#error SCSI_MAX_PHYS_SEGMENTS is too small
#endif

#define SP(x) { x, "sgpool-" #x } 
48
static struct scsi_host_sg_pool scsi_sg_pools[] = {
Linus Torvalds's avatar
Linus Torvalds committed
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
	SP(8),
	SP(16),
	SP(32),
#if (SCSI_MAX_PHYS_SEGMENTS > 32)
	SP(64),
#if (SCSI_MAX_PHYS_SEGMENTS > 64)
	SP(128),
#if (SCSI_MAX_PHYS_SEGMENTS > 128)
	SP(256),
#if (SCSI_MAX_PHYS_SEGMENTS > 256)
#error SCSI_MAX_PHYS_SEGMENTS is too large
#endif
#endif
#endif
#endif
}; 	
#undef SP

67
static void scsi_run_queue(struct request_queue *q);
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84

/*
 * Function:	scsi_unprep_request()
 *
 * Purpose:	Remove all preparation done for a request, including its
 *		associated scsi_cmnd, so that it can be requeued.
 *
 * Arguments:	req	- request to unprepare
 *
 * Lock status:	Assumed that no locks are held upon entry.
 *
 * Returns:	Nothing.
 */
static void scsi_unprep_request(struct request *req)
{
	struct scsi_cmnd *cmd = req->special;

85
	req->cmd_flags &= ~REQ_DONTPREP;
86
	req->special = NULL;
87
88
89

	scsi_put_command(cmd);
}
90

Linus Torvalds's avatar
Linus Torvalds committed
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
/*
 * Function:    scsi_queue_insert()
 *
 * Purpose:     Insert a command in the midlevel queue.
 *
 * Arguments:   cmd    - command that we are adding to queue.
 *              reason - why we are inserting command to queue.
 *
 * Lock status: Assumed that lock is not held upon entry.
 *
 * Returns:     Nothing.
 *
 * Notes:       We do this for one of two cases.  Either the host is busy
 *              and it cannot accept any more commands for the time being,
 *              or the device returned QUEUE_FULL and can accept no more
 *              commands.
 * Notes:       This could be called either from an interrupt context or a
 *              normal process context.
 */
int scsi_queue_insert(struct scsi_cmnd *cmd, int reason)
{
	struct Scsi_Host *host = cmd->device->host;
	struct scsi_device *device = cmd->device;
114
115
	struct request_queue *q = device->request_queue;
	unsigned long flags;
Linus Torvalds's avatar
Linus Torvalds committed
116
117
118
119
120

	SCSI_LOG_MLQUEUE(1,
		 printk("Inserting command %p into mlqueue\n", cmd));

	/*
121
	 * Set the appropriate busy bit for the device/host.
Linus Torvalds's avatar
Linus Torvalds committed
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
	 *
	 * If the host/device isn't busy, assume that something actually
	 * completed, and that we should be able to queue a command now.
	 *
	 * Note that the prior mid-layer assumption that any host could
	 * always queue at least one command is now broken.  The mid-layer
	 * will implement a user specifiable stall (see
	 * scsi_host.max_host_blocked and scsi_device.max_device_blocked)
	 * if a command is requeued with no other commands outstanding
	 * either for the device or for the host.
	 */
	if (reason == SCSI_MLQUEUE_HOST_BUSY)
		host->host_blocked = host->max_host_blocked;
	else if (reason == SCSI_MLQUEUE_DEVICE_BUSY)
		device->device_blocked = device->max_device_blocked;

	/*
	 * Decrement the counters, since these commands are no longer
	 * active on the host/device.
	 */
	scsi_device_unbusy(device);

	/*
145
146
	 * Requeue this command.  It will go before all other commands
	 * that are already in the queue.
Linus Torvalds's avatar
Linus Torvalds committed
147
148
149
150
	 *
	 * NOTE: there is magic here about the way the queue is plugged if
	 * we have no outstanding commands.
	 * 
151
	 * Although we *don't* plug the queue, we call the request
Linus Torvalds's avatar
Linus Torvalds committed
152
153
	 * function.  The SCSI request function detects the blocked condition
	 * and plugs the queue appropriately.
154
155
         */
	spin_lock_irqsave(q->queue_lock, flags);
156
	blk_requeue_request(q, cmd->request);
157
158
159
160
	spin_unlock_irqrestore(q->queue_lock, flags);

	scsi_run_queue(q);

Linus Torvalds's avatar
Linus Torvalds committed
161
162
163
	return 0;
}

164
/**
165
 * scsi_execute - insert request and wait for the result
166
167
168
169
170
171
172
173
 * @sdev:	scsi device
 * @cmd:	scsi command
 * @data_direction: data direction
 * @buffer:	data buffer
 * @bufflen:	len of buffer
 * @sense:	optional sense buffer
 * @timeout:	request timeout in seconds
 * @retries:	number of times to retry request
174
 * @flags:	or into request flags;
175
 *
176
 * returns the req->errors value which is the scsi_cmnd result
177
 * field.
178
 **/
179
180
181
int scsi_execute(struct scsi_device *sdev, const unsigned char *cmd,
		 int data_direction, void *buffer, unsigned bufflen,
		 unsigned char *sense, int timeout, int retries, int flags)
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
{
	struct request *req;
	int write = (data_direction == DMA_TO_DEVICE);
	int ret = DRIVER_ERROR << 24;

	req = blk_get_request(sdev->request_queue, write, __GFP_WAIT);

	if (bufflen &&	blk_rq_map_kern(sdev->request_queue, req,
					buffer, bufflen, __GFP_WAIT))
		goto out;

	req->cmd_len = COMMAND_SIZE(cmd[0]);
	memcpy(req->cmd, cmd, req->cmd_len);
	req->sense = sense;
	req->sense_len = 0;
197
	req->retries = retries;
198
	req->timeout = timeout;
199
200
	req->cmd_type = REQ_TYPE_BLOCK_PC;
	req->cmd_flags |= flags | REQ_QUIET | REQ_PREEMPT;
201
202
203
204
205
206
207
208
209
210
211
212

	/*
	 * head injection *required* here otherwise quiesce won't work
	 */
	blk_execute_rq(req->q, NULL, req, 1);

	ret = req->errors;
 out:
	blk_put_request(req);

	return ret;
}
213
EXPORT_SYMBOL(scsi_execute);
214

215
216
217
218
219
220

int scsi_execute_req(struct scsi_device *sdev, const unsigned char *cmd,
		     int data_direction, void *buffer, unsigned bufflen,
		     struct scsi_sense_hdr *sshdr, int timeout, int retries)
{
	char *sense = NULL;
221
222
	int result;
	
223
	if (sshdr) {
224
		sense = kzalloc(SCSI_SENSE_BUFFERSIZE, GFP_NOIO);
225
226
227
		if (!sense)
			return DRIVER_ERROR << 24;
	}
228
	result = scsi_execute(sdev, cmd, data_direction, buffer, bufflen,
229
			      sense, timeout, retries, 0);
230
	if (sshdr)
231
		scsi_normalize_sense(sense, SCSI_SENSE_BUFFERSIZE, sshdr);
232
233
234
235
236
237

	kfree(sense);
	return result;
}
EXPORT_SYMBOL(scsi_execute_req);

238
239
240
241
242
243
struct scsi_io_context {
	void *data;
	void (*done)(void *data, char *sense, int result, int resid);
	char sense[SCSI_SENSE_BUFFERSIZE];
};

244
static struct kmem_cache *scsi_io_context_cache;
245

246
static void scsi_end_async(struct request *req, int uptodate)
247
248
249
250
251
252
{
	struct scsi_io_context *sioc = req->end_io_data;

	if (sioc->done)
		sioc->done(sioc->data, sioc->sense, req->errors, req->data_len);

253
	kmem_cache_free(scsi_io_context_cache, sioc);
254
255
256
257
258
259
260
261
262
263
264
265
266
267
	__blk_put_request(req->q, req);
}

static int scsi_merge_bio(struct request *rq, struct bio *bio)
{
	struct request_queue *q = rq->q;

	bio->bi_flags &= ~(1 << BIO_SEG_VALID);
	if (rq_data_dir(rq) == WRITE)
		bio->bi_rw |= (1 << BIO_RW);
	blk_queue_bounce(q, &bio);

	if (!rq->bio)
		blk_rq_bio_prep(q, rq, bio);
268
	else if (!ll_back_merge_fn(q, rq, bio))
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
		return -EINVAL;
	else {
		rq->biotail->bi_next = bio;
		rq->biotail = bio;
	}

	return 0;
}

static int scsi_bi_endio(struct bio *bio, unsigned int bytes_done, int error)
{
	if (bio->bi_size)
		return 1;

	bio_put(bio);
	return 0;
}

/**
 * scsi_req_map_sg - map a scatterlist into a request
 * @rq:		request to fill
 * @sg:		scatterlist
 * @nsegs:	number of elements
 * @bufflen:	len of buffer
 * @gfp:	memory allocation flags
 *
 * scsi_req_map_sg maps a scatterlist into a request so that the
 * request can be sent to the block layer. We do not trust the scatterlist
 * sent to use, as some ULDs use that struct to only organize the pages.
 */
static int scsi_req_map_sg(struct request *rq, struct scatterlist *sgl,
			   int nsegs, unsigned bufflen, gfp_t gfp)
{
	struct request_queue *q = rq->q;
303
	int nr_pages = (bufflen + sgl[0].offset + PAGE_SIZE - 1) >> PAGE_SHIFT;
304
	unsigned int data_len = bufflen, len, bytes, off;
305
306
307
308
309
310
311
312
313
	struct page *page;
	struct bio *bio = NULL;
	int i, err, nr_vecs = 0;

	for (i = 0; i < nsegs; i++) {
		page = sgl[i].page;
		off = sgl[i].offset;
		len = sgl[i].length;

314
315
316
317
318
319
		while (len > 0 && data_len > 0) {
			/*
			 * sg sends a scatterlist that is larger than
			 * the data_len it wants transferred for certain
			 * IO sizes
			 */
320
			bytes = min_t(unsigned int, len, PAGE_SIZE - off);
321
			bytes = min(bytes, data_len);
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352

			if (!bio) {
				nr_vecs = min_t(int, BIO_MAX_PAGES, nr_pages);
				nr_pages -= nr_vecs;

				bio = bio_alloc(gfp, nr_vecs);
				if (!bio) {
					err = -ENOMEM;
					goto free_bios;
				}
				bio->bi_end_io = scsi_bi_endio;
			}

			if (bio_add_pc_page(q, bio, page, bytes, off) !=
			    bytes) {
				bio_put(bio);
				err = -EINVAL;
				goto free_bios;
			}

			if (bio->bi_vcnt >= nr_vecs) {
				err = scsi_merge_bio(rq, bio);
				if (err) {
					bio_endio(bio, bio->bi_size, 0);
					goto free_bios;
				}
				bio = NULL;
			}

			page++;
			len -= bytes;
353
			data_len -=bytes;
354
355
356
357
358
			off = 0;
		}
	}

	rq->buffer = rq->data = NULL;
359
	rq->data_len = bufflen;
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
	return 0;

free_bios:
	while ((bio = rq->bio) != NULL) {
		rq->bio = bio->bi_next;
		/*
		 * call endio instead of bio_put incase it was bounced
		 */
		bio_endio(bio, bio->bi_size, 0);
	}

	return err;
}

/**
 * scsi_execute_async - insert request
 * @sdev:	scsi device
 * @cmd:	scsi command
378
 * @cmd_len:	length of scsi cdb
379
380
381
382
383
384
385
386
387
 * @data_direction: data direction
 * @buffer:	data buffer (this can be a kernel buffer or scatterlist)
 * @bufflen:	len of buffer
 * @use_sg:	if buffer is a scatterlist this is the number of elements
 * @timeout:	request timeout in seconds
 * @retries:	number of times to retry request
 * @flags:	or into request flags
 **/
int scsi_execute_async(struct scsi_device *sdev, const unsigned char *cmd,
388
		       int cmd_len, int data_direction, void *buffer, unsigned bufflen,
389
390
391
392
393
394
395
396
		       int use_sg, int timeout, int retries, void *privdata,
		       void (*done)(void *, char *, int, int), gfp_t gfp)
{
	struct request *req;
	struct scsi_io_context *sioc;
	int err = 0;
	int write = (data_direction == DMA_TO_DEVICE);

397
	sioc = kmem_cache_zalloc(scsi_io_context_cache, gfp);
398
399
400
401
402
403
	if (!sioc)
		return DRIVER_ERROR << 24;

	req = blk_get_request(sdev->request_queue, write, gfp);
	if (!req)
		goto free_sense;
404
405
	req->cmd_type = REQ_TYPE_BLOCK_PC;
	req->cmd_flags |= REQ_QUIET;
406
407
408
409
410
411
412
413
414

	if (use_sg)
		err = scsi_req_map_sg(req, buffer, use_sg, bufflen, gfp);
	else if (bufflen)
		err = blk_rq_map_kern(req->q, req, buffer, bufflen, gfp);

	if (err)
		goto free_req;

415
	req->cmd_len = cmd_len;
416
	memset(req->cmd, 0, BLK_MAX_CDB); /* ATAPI hates garbage after CDB */
417
418
419
420
	memcpy(req->cmd, cmd, req->cmd_len);
	req->sense = sioc->sense;
	req->sense_len = 0;
	req->timeout = timeout;
421
	req->retries = retries;
422
423
424
425
426
427
428
429
430
431
432
	req->end_io_data = sioc;

	sioc->data = privdata;
	sioc->done = done;

	blk_execute_rq_nowait(req->q, NULL, req, 1, scsi_end_async);
	return 0;

free_req:
	blk_put_request(req);
free_sense:
433
	kmem_cache_free(scsi_io_context_cache, sioc);
434
435
436
437
	return DRIVER_ERROR << 24;
}
EXPORT_SYMBOL_GPL(scsi_execute_async);

Linus Torvalds's avatar
Linus Torvalds committed
438
439
440
441
442
443
444
445
446
447
448
/*
 * Function:    scsi_init_cmd_errh()
 *
 * Purpose:     Initialize cmd fields related to error handling.
 *
 * Arguments:   cmd	- command that is ready to be queued.
 *
 * Notes:       This function has the job of initializing a number of
 *              fields related to error handling.   Typically this will
 *              be called once for each command, as required.
 */
449
static void scsi_init_cmd_errh(struct scsi_cmnd *cmd)
Linus Torvalds's avatar
Linus Torvalds committed
450
451
{
	cmd->serial_number = 0;
452
	cmd->resid = 0;
Linus Torvalds's avatar
Linus Torvalds committed
453
454
455
456
457
458
459
460
461
462
463
464
	memset(cmd->sense_buffer, 0, sizeof cmd->sense_buffer);
	if (cmd->cmd_len == 0)
		cmd->cmd_len = COMMAND_SIZE(cmd->cmnd[0]);
}

void scsi_device_unbusy(struct scsi_device *sdev)
{
	struct Scsi_Host *shost = sdev->host;
	unsigned long flags;

	spin_lock_irqsave(shost->host_lock, flags);
	shost->host_busy--;
465
	if (unlikely(scsi_host_in_recovery(shost) &&
466
		     (shost->host_failed || shost->host_eh_scheduled)))
Linus Torvalds's avatar
Linus Torvalds committed
467
468
		scsi_eh_wakeup(shost);
	spin_unlock(shost->host_lock);
's avatar
committed
469
	spin_lock(sdev->request_queue->queue_lock);
Linus Torvalds's avatar
Linus Torvalds committed
470
	sdev->device_busy--;
's avatar
committed
471
	spin_unlock_irqrestore(sdev->request_queue->queue_lock, flags);
Linus Torvalds's avatar
Linus Torvalds committed
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
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
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
}

/*
 * Called for single_lun devices on IO completion. Clear starget_sdev_user,
 * and call blk_run_queue for all the scsi_devices on the target -
 * including current_sdev first.
 *
 * Called with *no* scsi locks held.
 */
static void scsi_single_lun_run(struct scsi_device *current_sdev)
{
	struct Scsi_Host *shost = current_sdev->host;
	struct scsi_device *sdev, *tmp;
	struct scsi_target *starget = scsi_target(current_sdev);
	unsigned long flags;

	spin_lock_irqsave(shost->host_lock, flags);
	starget->starget_sdev_user = NULL;
	spin_unlock_irqrestore(shost->host_lock, flags);

	/*
	 * Call blk_run_queue for all LUNs on the target, starting with
	 * current_sdev. We race with others (to set starget_sdev_user),
	 * but in most cases, we will be first. Ideally, each LU on the
	 * target would get some limited time or requests on the target.
	 */
	blk_run_queue(current_sdev->request_queue);

	spin_lock_irqsave(shost->host_lock, flags);
	if (starget->starget_sdev_user)
		goto out;
	list_for_each_entry_safe(sdev, tmp, &starget->devices,
			same_target_siblings) {
		if (sdev == current_sdev)
			continue;
		if (scsi_device_get(sdev))
			continue;

		spin_unlock_irqrestore(shost->host_lock, flags);
		blk_run_queue(sdev->request_queue);
		spin_lock_irqsave(shost->host_lock, flags);
	
		scsi_device_put(sdev);
	}
 out:
	spin_unlock_irqrestore(shost->host_lock, flags);
}

/*
 * Function:	scsi_run_queue()
 *
 * Purpose:	Select a proper request queue to serve next
 *
 * Arguments:	q	- last request's queue
 *
 * Returns:     Nothing
 *
 * Notes:	The previous command was completely finished, start
 *		a new one if possible.
 */
static void scsi_run_queue(struct request_queue *q)
{
	struct scsi_device *sdev = q->queuedata;
	struct Scsi_Host *shost = sdev->host;
	unsigned long flags;

	if (sdev->single_lun)
		scsi_single_lun_run(sdev);

	spin_lock_irqsave(shost->host_lock, flags);
	while (!list_empty(&shost->starved_list) &&
	       !shost->host_blocked && !shost->host_self_blocked &&
		!((shost->can_queue > 0) &&
		  (shost->host_busy >= shost->can_queue))) {
		/*
		 * As long as shost is accepting commands and we have
		 * starved queues, call blk_run_queue. scsi_request_fn
		 * drops the queue_lock and can add us back to the
		 * starved_list.
		 *
		 * host_lock protects the starved_list and starved_entry.
		 * scsi_request_fn must get the host_lock before checking
		 * or modifying starved_list or starved_entry.
		 */
		sdev = list_entry(shost->starved_list.next,
					  struct scsi_device, starved_entry);
		list_del_init(&sdev->starved_entry);
		spin_unlock_irqrestore(shost->host_lock, flags);

561
562
563
564
565
566
567
568
569

		if (test_bit(QUEUE_FLAG_REENTER, &q->queue_flags) &&
		    !test_and_set_bit(QUEUE_FLAG_REENTER,
				      &sdev->request_queue->queue_flags)) {
			blk_run_queue(sdev->request_queue);
			clear_bit(QUEUE_FLAG_REENTER,
				  &sdev->request_queue->queue_flags);
		} else
			blk_run_queue(sdev->request_queue);
Linus Torvalds's avatar
Linus Torvalds committed
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600

		spin_lock_irqsave(shost->host_lock, flags);
		if (unlikely(!list_empty(&sdev->starved_entry)))
			/*
			 * sdev lost a race, and was put back on the
			 * starved list. This is unlikely but without this
			 * in theory we could loop forever.
			 */
			break;
	}
	spin_unlock_irqrestore(shost->host_lock, flags);

	blk_run_queue(q);
}

/*
 * Function:	scsi_requeue_command()
 *
 * Purpose:	Handle post-processing of completed commands.
 *
 * Arguments:	q	- queue to operate on
 *		cmd	- command that may need to be requeued.
 *
 * Returns:	Nothing
 *
 * Notes:	After command completion, there may be blocks left
 *		over which weren't finished by the previous command
 *		this can be for a number of reasons - the main one is
 *		I/O errors in the middle of the request, in which case
 *		we need to request the blocks that come after the bad
 *		sector.
601
 * Notes:	Upon return, cmd is a stale pointer.
Linus Torvalds's avatar
Linus Torvalds committed
602
603
604
 */
static void scsi_requeue_command(struct request_queue *q, struct scsi_cmnd *cmd)
{
605
	struct request *req = cmd->request;
606
607
	unsigned long flags;

608
	scsi_unprep_request(req);
609
	spin_lock_irqsave(q->queue_lock, flags);
610
	blk_requeue_request(q, req);
611
	spin_unlock_irqrestore(q->queue_lock, flags);
Linus Torvalds's avatar
Linus Torvalds committed
612
613
614
615
616
617

	scsi_run_queue(q);
}

void scsi_next_command(struct scsi_cmnd *cmd)
{
618
619
620
621
622
	struct scsi_device *sdev = cmd->device;
	struct request_queue *q = sdev->request_queue;

	/* need to hold a reference on the device before we let go of the cmd */
	get_device(&sdev->sdev_gendev);
Linus Torvalds's avatar
Linus Torvalds committed
623
624
625

	scsi_put_command(cmd);
	scsi_run_queue(q);
626
627
628

	/* ok to remove device now */
	put_device(&sdev->sdev_gendev);
Linus Torvalds's avatar
Linus Torvalds committed
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
}

void scsi_run_host_queues(struct Scsi_Host *shost)
{
	struct scsi_device *sdev;

	shost_for_each_device(sdev, shost)
		scsi_run_queue(sdev->request_queue);
}

/*
 * Function:    scsi_end_request()
 *
 * Purpose:     Post-processing of completed commands (usually invoked at end
 *		of upper level post-processing and scsi_io_completion).
 *
 * Arguments:   cmd	 - command that is complete.
 *              uptodate - 1 if I/O indicates success, <= 0 for I/O error.
 *              bytes    - number of bytes of completed I/O
 *		requeue  - indicates whether we should requeue leftovers.
 *
 * Lock status: Assumed that lock is not held upon entry.
 *
652
 * Returns:     cmd if requeue required, NULL otherwise.
Linus Torvalds's avatar
Linus Torvalds committed
653
654
655
656
657
658
 *
 * Notes:       This is called for block device requests in order to
 *              mark some number of sectors as complete.
 * 
 *		We are guaranteeing that the request queue will be goosed
 *		at some point during this call.
659
 * Notes:	If cmd was requeued, upon return it will be a stale pointer.
Linus Torvalds's avatar
Linus Torvalds committed
660
661
662
663
 */
static struct scsi_cmnd *scsi_end_request(struct scsi_cmnd *cmd, int uptodate,
					  int bytes, int requeue)
{
664
	struct request_queue *q = cmd->device->request_queue;
Linus Torvalds's avatar
Linus Torvalds committed
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
	struct request *req = cmd->request;
	unsigned long flags;

	/*
	 * If there are blocks left over at the end, set up the command
	 * to queue the remainder of them.
	 */
	if (end_that_request_chunk(req, uptodate, bytes)) {
		int leftover = (req->hard_nr_sectors << 9);

		if (blk_pc_request(req))
			leftover = req->data_len;

		/* kill remainder if no retrys */
		if (!uptodate && blk_noretry_request(req))
			end_that_request_chunk(req, 0, leftover);
		else {
682
			if (requeue) {
Linus Torvalds's avatar
Linus Torvalds committed
683
684
685
686
687
688
				/*
				 * Bleah.  Leftovers again.  Stick the
				 * leftovers in the front of the
				 * queue, and goose the queue again.
				 */
				scsi_requeue_command(q, cmd);
689
690
				cmd = NULL;
			}
Linus Torvalds's avatar
Linus Torvalds committed
691
692
693
694
695
696
697
698
699
			return cmd;
		}
	}

	add_disk_randomness(req->rq_disk);

	spin_lock_irqsave(q->queue_lock, flags);
	if (blk_rq_tagged(req))
		blk_queue_end_tag(q, req);
700
	end_that_request_last(req, uptodate);
Linus Torvalds's avatar
Linus Torvalds committed
701
702
703
704
705
706
707
708
709
710
	spin_unlock_irqrestore(q->queue_lock, flags);

	/*
	 * This will goose the queue request function at the end, so we don't
	 * need to worry about launching another command.
	 */
	scsi_next_command(cmd);
	return NULL;
}

711
struct scatterlist *scsi_alloc_sgtable(struct scsi_cmnd *cmd, gfp_t gfp_mask)
Linus Torvalds's avatar
Linus Torvalds committed
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
{
	struct scsi_host_sg_pool *sgp;
	struct scatterlist *sgl;

	BUG_ON(!cmd->use_sg);

	switch (cmd->use_sg) {
	case 1 ... 8:
		cmd->sglist_len = 0;
		break;
	case 9 ... 16:
		cmd->sglist_len = 1;
		break;
	case 17 ... 32:
		cmd->sglist_len = 2;
		break;
#if (SCSI_MAX_PHYS_SEGMENTS > 32)
	case 33 ... 64:
		cmd->sglist_len = 3;
		break;
#if (SCSI_MAX_PHYS_SEGMENTS > 64)
	case 65 ... 128:
		cmd->sglist_len = 4;
		break;
#if (SCSI_MAX_PHYS_SEGMENTS  > 128)
	case 129 ... 256:
		cmd->sglist_len = 5;
		break;
#endif
#endif
#endif
	default:
		return NULL;
	}

	sgp = scsi_sg_pools + cmd->sglist_len;
	sgl = mempool_alloc(sgp->pool, gfp_mask);
	return sgl;
}

752
753
754
EXPORT_SYMBOL(scsi_alloc_sgtable);

void scsi_free_sgtable(struct scatterlist *sgl, int index)
Linus Torvalds's avatar
Linus Torvalds committed
755
756
757
{
	struct scsi_host_sg_pool *sgp;

758
	BUG_ON(index >= SG_MEMPOOL_NR);
Linus Torvalds's avatar
Linus Torvalds committed
759
760
761
762
763

	sgp = scsi_sg_pools + index;
	mempool_free(sgl, sgp->pool);
}

764
765
EXPORT_SYMBOL(scsi_free_sgtable);

Linus Torvalds's avatar
Linus Torvalds committed
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
/*
 * Function:    scsi_release_buffers()
 *
 * Purpose:     Completion processing for block device I/O requests.
 *
 * Arguments:   cmd	- command that we are bailing.
 *
 * Lock status: Assumed that no lock is held upon entry.
 *
 * Returns:     Nothing
 *
 * Notes:       In the event that an upper level driver rejects a
 *		command, we must release resources allocated during
 *		the __init_io() function.  Primarily this would involve
 *		the scatter-gather table, and potentially any bounce
 *		buffers.
 */
static void scsi_release_buffers(struct scsi_cmnd *cmd)
{
	if (cmd->use_sg)
		scsi_free_sgtable(cmd->request_buffer, cmd->sglist_len);

	/*
	 * Zero these out.  They now point to freed memory, and it is
	 * dangerous to hang onto the pointers.
	 */
	cmd->request_buffer = NULL;
	cmd->request_bufflen = 0;
}

/*
 * Function:    scsi_io_completion()
 *
 * Purpose:     Completion processing for block device I/O requests.
 *
 * Arguments:   cmd   - command that is finished.
 *
 * Lock status: Assumed that no lock is held upon entry.
 *
 * Returns:     Nothing
 *
 * Notes:       This function is matched in terms of capabilities to
 *              the function that created the scatter-gather list.
 *              In other words, if there are no bounce buffers
 *              (the normal case for most drivers), we don't need
 *              the logic to deal with cleaning up afterwards.
 *
 *		We must do one of several things here:
 *
 *		a) Call scsi_end_request.  This will finish off the
 *		   specified number of sectors.  If we are done, the
 *		   command block will be released, and the queue
 *		   function will be goosed.  If we are not done, then
 *		   scsi_end_request will directly goose the queue.
 *
 *		b) We can just use scsi_requeue_command() here.  This would
 *		   be used if we just wanted to retry, for example.
 */
824
void scsi_io_completion(struct scsi_cmnd *cmd, unsigned int good_bytes)
Linus Torvalds's avatar
Linus Torvalds committed
825
826
{
	int result = cmd->result;
827
	int this_count = cmd->request_bufflen;
828
	struct request_queue *q = cmd->device->request_queue;
Linus Torvalds's avatar
Linus Torvalds committed
829
830
831
832
833
834
	struct request *req = cmd->request;
	int clear_errors = 1;
	struct scsi_sense_hdr sshdr;
	int sense_valid = 0;
	int sense_deferred = 0;

835
	scsi_release_buffers(cmd);
Linus Torvalds's avatar
Linus Torvalds committed
836
837
838
839
840
841

	if (result) {
		sense_valid = scsi_command_normalize_sense(cmd, &sshdr);
		if (sense_valid)
			sense_deferred = scsi_sense_is_deferred(&sshdr);
	}
842

Linus Torvalds's avatar
Linus Torvalds committed
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
	if (blk_pc_request(req)) { /* SG_IO ioctl from block level */
		req->errors = result;
		if (result) {
			clear_errors = 0;
			if (sense_valid && req->sense) {
				/*
				 * SG_IO wants current and deferred errors
				 */
				int len = 8 + cmd->sense_buffer[7];

				if (len > SCSI_SENSE_BUFFERSIZE)
					len = SCSI_SENSE_BUFFERSIZE;
				memcpy(req->sense, cmd->sense_buffer,  len);
				req->sense_len = len;
			}
858
859
		}
		req->data_len = cmd->resid;
Linus Torvalds's avatar
Linus Torvalds committed
860
861
862
863
864
865
	}

	/*
	 * Next deal with any sectors which we were able to correctly
	 * handle.
	 */
866
867
868
869
870
871
872
873
874
875
876
877
878
879
	SCSI_LOG_HLCOMPLETE(1, printk("%ld sectors total, "
				      "%d bytes done.\n",
				      req->nr_sectors, good_bytes));
	SCSI_LOG_HLCOMPLETE(1, printk("use_sg is %d\n", cmd->use_sg));

	if (clear_errors)
		req->errors = 0;

	/* A number of bytes were successfully read.  If there
	 * are leftovers and there is some kind of error
	 * (result != 0), retry the rest.
	 */
	if (scsi_end_request(cmd, 1, good_bytes, result == 0) == NULL)
		return;
880
881
882

	/* good_bytes = 0, or (inclusive) there were leftovers and
	 * result = 0, so scsi_end_request couldn't retry.
Linus Torvalds's avatar
Linus Torvalds committed
883
884
885
886
887
	 */
	if (sense_valid && !sense_deferred) {
		switch (sshdr.sense_key) {
		case UNIT_ATTENTION:
			if (cmd->device->removable) {
888
				/* Detected disc change.  Set a bit
Linus Torvalds's avatar
Linus Torvalds committed
889
890
891
				 * and quietly refuse further access.
				 */
				cmd->device->changed = 1;
892
				scsi_end_request(cmd, 0, this_count, 1);
Linus Torvalds's avatar
Linus Torvalds committed
893
894
				return;
			} else {
895
896
897
898
899
				/* Must have been a power glitch, or a
				 * bus reset.  Could not have been a
				 * media change, so we just retry the
				 * request and see what happens.
				 */
Linus Torvalds's avatar
Linus Torvalds committed
900
901
902
903
904
				scsi_requeue_command(q, cmd);
				return;
			}
			break;
		case ILLEGAL_REQUEST:
905
906
907
908
909
910
911
912
			/* If we had an ILLEGAL REQUEST returned, then
			 * we may have performed an unsupported
			 * command.  The only thing this should be
			 * would be a ten byte read where only a six
			 * byte read was supported.  Also, on a system
			 * where READ CAPACITY failed, we may have
			 * read past the end of the disk.
			 */
913
914
			if ((cmd->device->use_10_for_rw &&
			    sshdr.asc == 0x20 && sshdr.ascq == 0x00) &&
Linus Torvalds's avatar
Linus Torvalds committed
915
916
917
			    (cmd->cmnd[0] == READ_10 ||
			     cmd->cmnd[0] == WRITE_10)) {
				cmd->device->use_10_for_rw = 0;
918
919
				/* This will cause a retry with a
				 * 6-byte command.
Linus Torvalds's avatar
Linus Torvalds committed
920
921
				 */
				scsi_requeue_command(q, cmd);
922
				return;
Linus Torvalds's avatar
Linus Torvalds committed
923
			} else {
924
				scsi_end_request(cmd, 0, this_count, 1);
Linus Torvalds's avatar
Linus Torvalds committed
925
926
927
928
				return;
			}
			break;
		case NOT_READY:
929
			/* If the device is in the process of becoming
930
			 * ready, or has a temporary blockage, retry.
Linus Torvalds's avatar
Linus Torvalds committed
931
			 */
932
933
934
935
936
937
938
939
940
941
942
943
944
945
			if (sshdr.asc == 0x04) {
				switch (sshdr.ascq) {
				case 0x01: /* becoming ready */
				case 0x04: /* format in progress */
				case 0x05: /* rebuild in progress */
				case 0x06: /* recalculation in progress */
				case 0x07: /* operation in progress */
				case 0x08: /* Long write in progress */
				case 0x09: /* self test in progress */
					scsi_requeue_command(q, cmd);
					return;
				default:
					break;
				}
Linus Torvalds's avatar
Linus Torvalds committed
946
			}
947
			if (!(req->cmd_flags & REQ_QUIET)) {
948
				scmd_printk(KERN_INFO, cmd,
949
					    "Device not ready: ");
950
951
				scsi_print_sense_hdr("", &sshdr);
			}
952
			scsi_end_request(cmd, 0, this_count, 1);
Linus Torvalds's avatar
Linus Torvalds committed
953
954
			return;
		case VOLUME_OVERFLOW:
955
			if (!(req->cmd_flags & REQ_QUIET)) {
956
				scmd_printk(KERN_INFO, cmd,
957
					    "Volume overflow, CDB: ");
958
				__scsi_print_command(cmd->cmnd);
959
960
				scsi_print_sense("", cmd);
			}
961
962
			/* See SSC3rXX or current. */
			scsi_end_request(cmd, 0, this_count, 1);
Linus Torvalds's avatar
Linus Torvalds committed
963
964
965
966
			return;
		default:
			break;
		}
967
	}
Linus Torvalds's avatar
Linus Torvalds committed
968
	if (host_byte(result) == DID_RESET) {
969
970
971
		/* Third party bus reset or reset for error recovery
		 * reasons.  Just retry the request and see what
		 * happens.
Linus Torvalds's avatar
Linus Torvalds committed
972
973
974
975
976
		 */
		scsi_requeue_command(q, cmd);
		return;
	}
	if (result) {
977
		if (!(req->cmd_flags & REQ_QUIET)) {
978
			scsi_print_result(cmd);
979
980
981
			if (driver_byte(result) & DRIVER_SENSE)
				scsi_print_sense("", cmd);
		}
Linus Torvalds's avatar
Linus Torvalds committed
982
	}
983
	scsi_end_request(cmd, 0, this_count, !result);
Linus Torvalds's avatar
Linus Torvalds committed
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
1001
1002
1003
1004
}
EXPORT_SYMBOL(scsi_io_completion);

/*
 * Function:    scsi_init_io()
 *
 * Purpose:     SCSI I/O initialize function.
 *
 * Arguments:   cmd   - Command descriptor we wish to initialize
 *
 * Returns:     0 on success
 *		BLKPREP_DEFER if the failure is retryable
 *		BLKPREP_KILL if the failure is fatal
 */
static int scsi_init_io(struct scsi_cmnd *cmd)
{
	struct request     *req = cmd->request;
	struct scatterlist *sgpnt;
	int		   count;

	/*
1005
	 * We used to not use scatter-gather for single segment request,
Linus Torvalds's avatar
Linus Torvalds committed
1006
1007
1008
1009
1010
1011
	 * but now we do (it makes highmem I/O easier to support without
	 * kmapping pages)
	 */
	cmd->use_sg = req->nr_phys_segments;

	/*
1012
	 * If sg table allocation fails, requeue request later.
Linus Torvalds's avatar
Linus Torvalds committed
1013
1014
	 */
	sgpnt = scsi_alloc_sgtable(cmd, GFP_ATOMIC);
Alan Stern's avatar
Alan Stern committed
1015
1016
	if (unlikely(!sgpnt)) {
		scsi_unprep_request(req);
Linus Torvalds's avatar
Linus Torvalds committed
1017
		return BLKPREP_DEFER;
Alan Stern's avatar
Alan Stern committed
1018
	}
Linus Torvalds's avatar
Linus Torvalds committed
1019

1020
	req->buffer = NULL;
Linus Torvalds's avatar
Linus Torvalds committed
1021
1022
1023
	cmd->request_buffer = (char *) sgpnt;
	if (blk_pc_request(req))
		cmd->request_bufflen = req->data_len;
1024
1025
	else
		cmd->request_bufflen = req->nr_sectors << 9;
Linus Torvalds's avatar
Linus Torvalds committed
1026
1027
1028
1029
1030
1031
1032
1033

	/* 
	 * Next, walk the list, and fill in the addresses and sizes of
	 * each segment.
	 */
	count = blk_rq_map_sg(req->q, req, cmd->request_buffer);
	if (likely(count <= cmd->use_sg)) {
		cmd->use_sg = count;
1034
		return BLKPREP_OK;
Linus Torvalds's avatar
Linus Torvalds committed
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
	}

	printk(KERN_ERR "Incorrect number of segments after building list\n");
	printk(KERN_ERR "counted %d, received %d\n", count, cmd->use_sg);
	printk(KERN_ERR "req nr_sec %lu, cur_nr_sec %u\n", req->nr_sectors,
			req->current_nr_sectors);

	return BLKPREP_KILL;
}

1045
1046
1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
static struct scsi_cmnd *scsi_get_cmd_from_req(struct scsi_device *sdev,
		struct request *req)
{
	struct scsi_cmnd *cmd;

	if (!req->special) {
		cmd = scsi_get_command(sdev, GFP_ATOMIC);
		if (unlikely(!cmd))
			return NULL;
		req->special = cmd;
	} else {
		cmd = req->special;
	}

	/* pull a tag out of the request if we have one */
	cmd->tag = req->tag;
	cmd->request = req;

	return cmd;
}

1066
static void scsi_blk_pc_done(struct scsi_cmnd *cmd)
1067
1068
{
	BUG_ON(!blk_pc_request(cmd->request));
1069
1070
1071
1072
1073
1074
	/*
	 * This will complete the whole command with uptodate=1 so
	 * as far as the block layer is concerned the command completed
	 * successfully. Since this is a REQ_BLOCK_PC command the
	 * caller should check the request's errors value
	 */
1075
	scsi_io_completion(cmd, cmd->request_bufflen);
1076
1077
}

1078
int scsi_setup_blk_pc_cmnd(struct scsi_device *sdev, struct request *req)
James Bottomley's avatar
James Bottomley committed
1079
{
1080
	struct scsi_cmnd *cmd;
1081
1082
1083
1084
	int ret = scsi_prep_state_check(sdev, req);

	if (ret != BLKPREP_OK)
		return ret;
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

	cmd = scsi_get_cmd_from_req(sdev, req);
	if (unlikely(!cmd))
		return BLKPREP_DEFER;

	/*
	 * BLOCK_PC requests may transfer data, in which case they must
	 * a bio attached to them.  Or they might contain a SCSI command
	 * that does not transfer data, in which case they may optionally
	 * submit a request without an attached bio.
	 */
	if (req->bio) {
		int ret;

		BUG_ON(!req->nr_phys_segments);

		ret = scsi_init_io(cmd);
		if (unlikely(ret))
			return ret;
	} else {
		BUG_ON(req->data_len);
		BUG_ON(req->data);

		cmd->request_bufflen = 0;
		cmd->request_buffer = NULL;
		cmd->use_sg = 0;
		req->buffer = NULL;
	}
James Bottomley's avatar
James Bottomley committed
1113

1114
	BUILD_BUG_ON(sizeof(req->cmd) > sizeof(cmd->cmnd));
James Bottomley's avatar
James Bottomley committed
1115
1116
1117
1118
1119
1120
1121
1122
1123
1124
1125
1126
	memcpy(cmd->cmnd, req->cmd, sizeof(cmd->cmnd));
	cmd->cmd_len = req->cmd_len;
	if (!req->data_len)
		cmd->sc_data_direction = DMA_NONE;
	else if (rq_data_dir(req) == WRITE)
		cmd->sc_data_direction = DMA_TO_DEVICE;
	else
		cmd->sc_data_direction = DMA_FROM_DEVICE;
	
	cmd->transfersize = req->data_len;
	cmd->allowed = req->retries;
	cmd->timeout_per_command = req->timeout;
1127
	cmd->done = scsi_blk_pc_done;
1128
	return BLKPREP_OK;
James Bottomley's avatar
James Bottomley committed
1129
}
1130
EXPORT_SYMBOL(scsi_setup_blk_pc_cmnd);
James Bottomley's avatar
James Bottomley committed
1131

1132
1133
1134
1135
1136
/*
 * Setup a REQ_TYPE_FS command.  These are simple read/write request
 * from filesystems that still need to be translated to SCSI CDBs from
 * the ULD.
 */
1137
int scsi_setup_fs_cmnd(struct scsi_device *sdev, struct request *req)
Linus Torvalds's avatar
Linus Torvalds committed
1138
1139
{
	struct scsi_cmnd *cmd;
1140
	int ret = scsi_prep_state_check(sdev, req);
Linus Torvalds's avatar
Linus Torvalds committed
1141

1142
1143
	if (ret != BLKPREP_OK)
		return ret;
Linus Torvalds's avatar
Linus Torvalds committed
1144
	/*
1145
	 * Filesystem requests must transfer data.
Linus Torvalds's avatar
Linus Torvalds committed
1146
	 */
1147
1148
1149
1150
1151
1152
	BUG_ON(!req->nr_phys_segments);

	cmd = scsi_get_cmd_from_req(sdev, req);
	if (unlikely(!cmd))
		return BLKPREP_DEFER;

1153
	return scsi_init_io(cmd);
1154
}
1155
EXPORT_SYMBOL(scsi_setup_fs_cmnd);
1156

1157
int scsi_prep_state_check(struct scsi_device *sdev, struct request *req)
1158
1159
1160
{
	int ret = BLKPREP_OK;

Linus Torvalds's avatar
Linus Torvalds committed
1161
	/*
1162
1163
	 * If the device is not in running state we will reject some
	 * or all commands.
Linus Torvalds's avatar
Linus Torvalds committed
1164
	 */
1165
1166
1167
1168
1169
1170
1171
1172
1173
1174
1175
1176
1177
1178
1179
1180
1181
	if (unlikely(sdev->sdev_state != SDEV_RUNNING)) {
		switch (sdev->sdev_state) {
		case SDEV_OFFLINE:
			/*
			 * If the device is offline we refuse to process any
			 * commands.  The device must be brought online
			 * before trying any recovery commands.
			 */
			sdev_printk(KERN_ERR, sdev,
				    "rejecting I/O to offline device\n");
			ret = BLKPREP_KILL;
			break;
		case SDEV_DEL:
			/*
			 * If the device is fully deleted, we refuse to
			 * process any commands as well.
			 */
1182
			sdev_printk(KERN_ERR, sdev,
1183
1184
1185
1186
1187
1188
1189
1190
1191
1192
1193
1194
1195
1196
1197
1198
1199
1200
1201
1202
				    "rejecting I/O to dead device\n");
			ret = BLKPREP_KILL;
			break;
		case SDEV_QUIESCE:
		case SDEV_BLOCK:
			/*
			 * If the devices is blocked we defer normal commands.
			 */
			if (!(req->cmd_flags & REQ_PREEMPT))
				ret = BLKPREP_DEFER;
			break;
		default:
			/*
			 * For any other not fully online state we only allow
			 * special commands.  In particular any user initiated
			 * command is not allowed.
			 */
			if (!(req->cmd_flags & REQ_PREEMPT))
				ret = BLKPREP_KILL;
			break;
Linus Torvalds's avatar
Linus Torvalds committed
1203
1204
		}
	}
1205
1206
1207
	return ret;
}
EXPORT_SYMBOL(scsi_prep_state_check);
Linus Torvalds's avatar
Linus Torvalds committed
1208

1209
1210
1211
int scsi_prep_return(struct request_queue *q, struct request *req, int ret)
{
	struct scsi_device *sdev = q->queuedata;
Linus Torvalds's avatar
Linus Torvalds committed
1212

1213
1214
1215
	switch (ret) {
	case BLKPREP_KILL:
		req->errors = DID_NO_CONNECT << 16;
1216
1217
1218
1219
1220
1221
1222
		/* release the command and kill it */
		if (req->special) {
			struct scsi_cmnd *cmd = req->special;
			scsi_release_buffers(cmd);
			scsi_put_command(cmd);
			req->special = NULL;
		}
1223
1224
		break;
	case BLKPREP_DEFER:
Linus Torvalds's avatar
Linus Torvalds committed
1225
		/*
1226
1227
1228
		 * If we defer, the elv_next_request() returns NULL, but the
		 * queue must be restarted, so we plug here if no returning
		 * command will automatically do that.
Linus Torvalds's avatar
Linus Torvalds committed
1229
		 */
1230
1231
1232
1233
1234
		if (sdev->device_busy == 0)
			blk_plug_device(q);
		break;
	default:
		req->cmd_flags |= REQ_DONTPREP;
Linus Torvalds's avatar
Linus Torvalds committed
1235
1236
	}

1237
	return ret;
Linus Torvalds's avatar
Linus Torvalds committed
1238
}
1239
1240
1241
1242
1243
1244
1245
1246
1247
1248
1249
EXPORT_SYMBOL(scsi_prep_return);

static int scsi_prep_fn(struct request_queue *q, struct request *req)
{
	struct scsi_device *sdev = q->queuedata;
	int ret = BLKPREP_KILL;

	if (req->cmd_type == REQ_TYPE_BLOCK_PC)
		ret = scsi_setup_blk_pc_cmnd(sdev, req);
	return scsi_prep_return(q, req, ret);
}
Linus Torvalds's avatar
Linus Torvalds committed
1250
1251
1252
1253
1254
1255
1256
1257
1258
1259
1260
1261
1262
1263
1264
1265
1266
1267

/*
 * scsi_dev_queue_ready: if we can send requests to sdev, return 1 else
 * return 0.
 *
 * Called with the queue_lock held.
 */
static inline int scsi_dev_queue_ready(struct request_queue *q,
				  struct scsi_device *sdev)
{
	if (sdev->device_busy >= sdev->queue_depth)
		return 0;
	if (sdev->device_busy == 0 && sdev->device_blocked) {
		/*
		 * unblock after device_blocked iterates to zero
		 */
		if (--sdev->device_blocked == 0) {
			SCSI_LOG_MLQUEUE(3,
1268
1269
				   sdev_printk(KERN_INFO, sdev,
				   "unblocking device at zero depth\n"));
Linus Torvalds's avatar
Linus Torvalds committed
1270
1271
1272
1273
1274
1275
1276
1277
1278
1279
1280
1281
1282
1283
1284
1285
1286
1287
1288
1289
1290
1291
		} else {
			blk_plug_device(q);
			return 0;
		}
	}
	if (sdev->device_blocked)
		return 0;

	return 1;
}

/*
 * scsi_host_queue_ready: if we can send requests to shost, return 1 else
 * return 0. We must end up running the queue again whenever 0 is
 * returned, else IO can hang.
 *
 * Called with host_lock held.
 */
static inline int scsi_host_queue_ready(struct request_queue *q,
				   struct Scsi_Host *shost,
				   struct scsi_device *sdev)
{
1292
	if (scsi_host_in_recovery(shost))
Linus Torvalds's avatar
Linus Torvalds committed
1293
1294
1295
1296
1297
1298
1299
1300
1301
1302
1303
1304
1305
1306
1307
1308
1309
1310
1311
1312
1313
1314
1315
1316
1317
1318
1319
1320
1321
		return 0;
	if (shost->host_busy == 0 && shost->host_blocked) {
		/*
		 * unblock after host_blocked iterates to zero
		 */
		if (--shost->host_blocked == 0) {
			SCSI_LOG_MLQUEUE(3,
				printk("scsi%d unblocking host at zero depth\n",
					shost->host_no));
		} else {
			blk_plug_device(q);
			return 0;
		}
	}
	if ((shost->can_queue > 0 && shost->host_busy >= shost->can_queue) ||
	    shost->host_blocked || shost->host_self_blocked) {
		if (list_empty(&sdev->starved_entry))
			list_add_tail(&sdev->starved_entry, &shost->starved_list);
		return 0;
	}

	/* We're OK to process the command, so we can't be starved */
	if (!list_empty(&sdev->starved_entry))
		list_del_init(&sdev->starved_entry);

	return 1;
}

/*
1322
 * Kill a request for a dead device
Linus Torvalds's avatar
Linus Torvalds committed
1323
 */
1324
static void scsi_kill_request(struct request *req, struct request_queue *q)
Linus Torvalds's avatar
Linus Torvalds committed
1325
{
1326
	struct scsi_cmnd *cmd = req->special;
1327
1328
	struct scsi_device *sdev = cmd->device;
	struct Scsi_Host *shost = sdev->host;
Linus Torvalds's avatar
Linus Torvalds committed
1329

1330
1331
	blkdev_dequeue_request(req);

1332
1333
1334
1335
	if (unlikely(cmd == NULL)) {
		printk(KERN_CRIT "impossible request in %s.\n",
				 __FUNCTION__);
		BUG();
Linus Torvalds's avatar
Linus Torvalds committed
1336
	}
1337
1338
1339
1340

	scsi_init_cmd_errh(cmd);
	cmd->result = DID_NO_CONNECT << 16;
	atomic_inc(&cmd->device->iorequest_cnt);
1341
1342
1343
1344
1345
1346
1347
1348
1349
1350
1351
1352
1353

	/*
	 * SCSI request completion path will do scsi_device_unbusy(),
	 * bump busy counts.  To bump the counters, we need to dance
	 * with the locks as normal issue path does.
	 */
	sdev->device_busy++;
	spin_unlock(sdev->request_queue->queue_lock);
	spin_lock(shost->host_lock);
	shost->host_busy++;
	spin_unlock(shost->host_lock);
	spin_lock(sdev->request_queue->queue_lock);

1354
	__scsi_done(cmd);
Linus Torvalds's avatar
Linus Torvalds committed
1355
1356
}

1357
1358
1359
static void scsi_softirq_done(struct request *rq)
{
	struct scsi_cmnd *cmd = rq->completion_data;
1360
	unsigned long wait_for = (cmd->allowed + 1) * cmd->timeout_per_command;
1361
1362
1363
1364
1365
1366
1367
1368
1369
1370
1371
1372
1373
1374
1375
1376
1377
1378
1379
1380
	int disposition;

	INIT_LIST_HEAD(&cmd->eh_entry);

	disposition = scsi_decide_disposition(cmd);
	if (disposition != SUCCESS &&
	    time_before(cmd->jiffies_at_alloc + wait_for, jiffies)) {
		sdev_printk(KERN_ERR, cmd->device,
			    "timing out command, waited %lus\n",
			    wait_for/HZ);
		disposition = SUCCESS;
	}
			
	scsi_log_completion(cmd, disposition);

	switch (disposition) {
		case SUCCESS:
			scsi_finish_command(cmd);
			break;
		case NEEDS_RETRY:
1381
			scsi_queue_insert(cmd, SCSI_MLQUEUE_EH_RETRY);
1382
1383
1384
1385
1386
1387
1388
1389
1390
1391
			break;
		case ADD_TO_MLQUEUE:
			scsi_queue_insert(cmd, SCSI_MLQUEUE_DEVICE_BUSY);
			break;
		default:
			if (!scsi_eh_scmd_add(cmd, 0))
				scsi_finish_command(cmd);
	}
}

Linus Torvalds's avatar
Linus Torvalds committed
1392
1393
1394
1395
1396
1397
1398
1399
1400
1401
1402
1403
1404
1405
1406
1407
1408
1409
1410
1411
/*
 * Function:    scsi_request_fn()
 *
 * Purpose:     Main strategy routine for SCSI.
 *
 * Arguments:   q       - Pointer to actual queue.
 *
 * Returns:     Nothing
 *
 * Lock status: IO request lock assumed to be held when called.
 */
static void scsi_request_fn(struct request_queue *q)
{
	struct scsi_device *sdev = q->queuedata;
	struct Scsi_Host *shost;
	struct scsi_cmnd *cmd;
	struct request *req;

	if (!sdev) {
		printk("scsi: killing requests for dead queue\n");
1412
1413
		while ((req = elv_next_request(q)) != NULL)
			scsi_kill_request(req, q);
Linus Torvalds's avatar
Linus Torvalds committed
1414
1415
1416
1417
1418
1419
1420
1421
1422
1423
1424
1425
1426
1427
1428
1429
1430
1431
1432
1433
1434
1435
1436
1437
		return;
	}

	if(!get_device(&sdev->sdev_gendev))
		/* We must be tearing the block queue down already */
		return;

	/*
	 * To start with, we keep looping until the queue is empty, or until
	 * the host is no longer able to accept any more requests.
	 */
	shost = sdev->host;
	while (!blk_queue_plugged(q)) {
		int rtn;
		/*
		 * get next queueable request.  We do this early to make sure
		 * that the request is fully prepared even if we cannot 
		 * accept it.
		 */
		req = elv_next_request(q);
		if (!req || !scsi_dev_queue_ready(q, sdev))
			break;

		if (unlikely(!scsi_device_online(sdev))) {
1438
1439
			sdev_printk(KERN_ERR, sdev,
				    "rejecting I/O to offline device\n");
1440
			scsi_kill_request(req, q);
Linus Torvalds's avatar
Linus Torvalds committed
1441
1442
1443
1444
1445
1446
1447
1448
1449
1450
1451
1452
			continue;
		}


		/*
		 * Remove the request from the request list.
		 */
		if (!(blk_queue_tagged(q) && !blk_queue_start_tag(q, req)))
			blkdev_dequeue_request(req);
		sdev->device_busy++;

		spin_unlock(q->queue_lock);
1453
1454
1455
1456
		cmd = req->special;
		if (unlikely(cmd == NULL)) {
			printk(KERN_CRIT "impossible request in %s.\n"
					 "please mail a stack trace to "
1457
					 "linux-scsi@vger.kernel.org\n",
1458
					 __FUNCTION__);
1459
			blk_dump_rq_flags(req, "foo");
1460
1461
			BUG();
		}
Linus Torvalds's avatar
Linus Torvalds committed
1462
1463
1464
1465
1466
1467
1468
1469
1470
1471
1472
1473
1474
1475
1476
1477
1478
1479
1480
1481
1482
1483
1484
1485
1486
1487
1488
1489
1490
1491
1492
1493
1494
1495
1496
1497
1498
1499
1500
1501
1502
1503
1504
1505
1506
1507
1508
1509
1510
1511
1512
1513
1514
1515
1516
1517
1518
1519
1520
1521
1522
1523
1524
1525
1526
1527
1528
1529
1530
1531
1532
1533
1534
1535
1536
1537
1538
1539
1540
1541
1542
1543
1544
1545
1546
1547
1548
1549
		spin_lock(shost->host_lock);

		if (!scsi_host_queue_ready(q, shost, sdev))
			goto not_ready;
		if (sdev->single_lun) {
			if (scsi_target(sdev)->starget_sdev_user &&
			    scsi_target(sdev)->starget_sdev_user != sdev)
				goto not_ready;
			scsi_target(sdev)->starget_sdev_user = sdev;
		}
		shost->host_busy++;

		/*
		 * XXX(hch): This is rather suboptimal, scsi_dispatch_cmd will
		 *		take the lock again.
		 */
		spin_unlock_irq(shost->host_lock);

		/*
		 * Finally, initialize any error handling parameters, and set up
		 * the timers for timeouts.
		 */
		scsi_init_cmd_errh(cmd);

		/*
		 * Dispatch the command to the low-level driver.
		 */
		rtn = scsi_dispatch_cmd(cmd);
		spin_lock_irq(q->queue_lock);
		if(rtn) {
			/* we're refusing the command; because of
			 * the way locks get dropped, we need to 
			 * check here if plugging is required */
			if(sdev->device_busy == 0)
				blk_plug_device(q);

			break;
		}
	}

	goto out;

 not_ready:
	spin_unlock_irq(shost->host_lock);

	/*
	 * lock q, handle tag, requeue req, and decrement device_busy. We
	 * must return with queue_lock held.
	 *
	 * Decrementing device_busy without checking it is OK, as all such
	 * cases (host limits or settings) should run the queue at some
	 * later time.
	 */
	spin_lock_irq(q->queue_lock);
	blk_requeue_request(q, req);
	sdev->device_busy--;
	if(sdev->device_busy == 0)
		blk_plug_device(q);
 out:
	/* must be careful here...if we trigger the ->remove() function
	 * we cannot be holding the q lock */
	spin_unlock_irq(q->queue_lock);
	put_device(&sdev->sdev_gendev);
	spin_lock_irq(q->queue_lock);
}

u64 scsi_calculate_bounce_limit(struct Scsi_Host *shost)
{
	struct device *host_dev;
	u64 bounce_limit = 0xffffffff;

	if (shost->unchecked_isa_dma)
		return BLK_BOUNCE_ISA;
	/*
	 * Platforms with virtual-DMA translation
	 * hardware have no practical limit.
	 */
	if (!PCI_DMA_BUS_IS_PHYS)
		return BLK_BOUNCE_ANY;

	host_dev = scsi_get_device(shost);
	if (host_dev && host_dev->dma_mask)
		bounce_limit = *host_dev->dma_mask;

	return bounce_limit;
}
EXPORT_SYMBOL(scsi_calculate_bounce_limit);

1550
1551
struct request_queue *__scsi_alloc_queue(struct Scsi_Host *shost,
					 request_fn_proc *request_fn)
Linus Torvalds's avatar
Linus Torvalds committed
1552
1553
1554
{
	struct request_queue *q;

1555
	q = blk_init_queue(request_fn, NULL);
Linus Torvalds's avatar
Linus Torvalds committed
1556
1557
1558
1559
1560
1561
1562
1563
1564
1565
1566
1567
1568
	if (!q)
		return NULL;

	blk_queue_max_hw_segments(q, shost->sg_tablesize);
	blk_queue_max_phys_segments(q, SCSI_MAX_PHYS_SEGMENTS);
	blk_queue_max_sectors(q, shost->max_sectors);
	blk_queue_bounce_limit(q, scsi_calculate_bounce_limit(shost));
	blk_queue_segment_boundary(q, shost->dma_boundary);

	if (!shost->use_clustering)
		clear_bit(QUEUE_FLAG_CLUSTER, &q->queue_flags);
	return q;
}
1569
1570
1571
1572
1573
1574
1575
1576
1577
1578
1579
1580
1581
1582
EXPORT_SYMBOL(__scsi_alloc_queue);

struct request_queue *scsi_alloc_queue(struct scsi_device *sdev)
{
	struct request_queue *q;

	q = __scsi_alloc_queue(sdev->host, scsi_request_fn);
	if (!q)
		return NULL;

	blk_queue_prep_rq(q, scsi_prep_fn);
	blk_queue_softirq_done(q, scsi_softirq_done);
	return q;
}
Linus Torvalds's avatar
Linus Torvalds committed
1583
1584
1585
1586
1587
1588
1589
1590
1591
1592
1593
1594
1595
1596
1597
1598
1599
1600
1601
1602
1603
1604
1605
1606
1607
1608
1609
1610
1611
1612
1613
1614
1615
1616
1617
1618
1619
1620
1621
1622
1623
1624
1625
1626
1627
1628
1629
1630
1631
1632
1633
1634
1635
1636
1637
1638
1639
1640
1641

void scsi_free_queue(struct request_queue *q)
{
	blk_cleanup_queue(q);
}

/*
 * Function:    scsi_block_requests()
 *
 * Purpose:     Utility function used by low-level drivers to prevent further
 *		commands from being queued to the device.
 *
 * Arguments:   shost       - Host in question
 *
 * Returns:     Nothing
 *
 * Lock status: No locks are assumed held.
 *
 * Notes:       There is no timer nor any other means by which the requests
 *		get unblocked other than the low-level driver calling
 *		scsi_unblock_requests().
 */
void scsi_block_requests(struct Scsi_Host *shost)
{
	shost->host_self_blocked = 1;
}
EXPORT_SYMBOL(scsi_block_requests);

/*
 * Function:    scsi_unblock_requests()
 *
 * Purpose:     Utility function used by low-level drivers to allow further
 *		commands from being queued to the device.
 *
 * Arguments:   shost       - Host in question
 *
 * Returns:     Nothing
 *
 * Lock status: No locks are assumed held.
 *
 * Notes:       There is no timer nor any other means by which the requests
 *		get unblocked other than the low-level driver calling
 *		scsi_unblock_requests().
 *
 *		This is done as an API function so that changes to the
 *		internals of the scsi mid-layer won't require wholesale
 *		changes to drivers that use this feature.
 */
void scsi_unblock_requests(struct Scsi_Host *shost)
{
	shost->host_self_blocked = 0;
	scsi_run_host_queues(shost);
}
EXPORT_SYMBOL(scsi_unblock_requests);

int __init scsi_init_queue(void)
{
	int i;

1642
1643
	scsi_io_context_cache = kmem_cache_create("scsi_io_context",
					sizeof(struct scsi_io_context),
1644
					0, 0, NULL);
1645
1646
1647
1648
1649
	if (!scsi_io_context_cache) {
		printk(KERN_ERR "SCSI: can't init scsi io context cache\n");
		return -ENOMEM;
	}

Linus Torvalds's avatar
Linus Torvalds committed