scsi_lib.c 63.8 KB
Newer Older
Linus Torvalds's avatar
Linus Torvalds committed
1 2 3 4 5 6 7 8 9 10
/*
 *  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>
11
#include <linux/bitops.h>
Linus Torvalds's avatar
Linus Torvalds committed
12 13 14 15 16 17 18 19
#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>
20
#include <linux/hardirq.h>
Jens Axboe's avatar
Jens Axboe committed
21
#include <linux/scatterlist.h>
Linus Torvalds's avatar
Linus Torvalds committed
22 23

#include <scsi/scsi.h>
24
#include <scsi/scsi_cmnd.h>
Linus Torvalds's avatar
Linus Torvalds committed
25 26 27 28 29 30 31 32 33 34
#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"


35
#define SG_MEMPOOL_NR		ARRAY_SIZE(scsi_sg_pools)
36
#define SG_MEMPOOL_SIZE		2
Linus Torvalds's avatar
Linus Torvalds committed
37 38 39

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

45 46 47 48
#define SP(x) { x, "sgpool-" __stringify(x) }
#if (SCSI_MAX_SG_SEGMENTS < 32)
#error SCSI_MAX_SG_SEGMENTS is too small (must be 32 or greater)
#endif
49
static struct scsi_host_sg_pool scsi_sg_pools[] = {
Linus Torvalds's avatar
Linus Torvalds committed
50 51
	SP(8),
	SP(16),
52
#if (SCSI_MAX_SG_SEGMENTS > 32)
53
	SP(32),
54
#if (SCSI_MAX_SG_SEGMENTS > 64)
55 56
	SP(64),
#if (SCSI_MAX_SG_SEGMENTS > 128)
Linus Torvalds's avatar
Linus Torvalds committed
57
	SP(128),
58 59
#if (SCSI_MAX_SG_SEGMENTS > 256)
#error SCSI_MAX_SG_SEGMENTS is too large (256 MAX)
60 61 62
#endif
#endif
#endif
63 64
#endif
	SP(SCSI_MAX_SG_SEGMENTS)
65
};
Linus Torvalds's avatar
Linus Torvalds committed
66 67
#undef SP

68
static struct kmem_cache *scsi_sdb_cache;
69

70
static void scsi_run_queue(struct request_queue *q);
71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87

/*
 * 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;

88
	req->cmd_flags &= ~REQ_DONTPREP;
89
	req->special = NULL;
90 91 92

	scsi_put_command(cmd);
}
93

Linus Torvalds's avatar
Linus Torvalds committed
94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116
/*
 * 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;
117 118
	struct request_queue *q = device->request_queue;
	unsigned long flags;
Linus Torvalds's avatar
Linus Torvalds committed
119 120 121 122 123

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

	/*
124
	 * Set the appropriate busy bit for the device/host.
Linus Torvalds's avatar
Linus Torvalds committed
125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147
	 *
	 * 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);

	/*
148 149
	 * Requeue this command.  It will go before all other commands
	 * that are already in the queue.
Linus Torvalds's avatar
Linus Torvalds committed
150 151 152 153
	 *
	 * NOTE: there is magic here about the way the queue is plugged if
	 * we have no outstanding commands.
	 * 
154
	 * Although we *don't* plug the queue, we call the request
Linus Torvalds's avatar
Linus Torvalds committed
155 156
	 * function.  The SCSI request function detects the blocked condition
	 * and plugs the queue appropriately.
157 158
         */
	spin_lock_irqsave(q->queue_lock, flags);
159
	blk_requeue_request(q, cmd->request);
160 161 162 163
	spin_unlock_irqrestore(q->queue_lock, flags);

	scsi_run_queue(q);

Linus Torvalds's avatar
Linus Torvalds committed
164 165 166
	return 0;
}

167
/**
168
 * scsi_execute - insert request and wait for the result
169 170 171 172 173 174 175 176
 * @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
177
 * @flags:	or into request flags;
178
 *
179
 * returns the req->errors value which is the scsi_cmnd result
180
 * field.
181
 */
182 183 184
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)
185 186 187 188 189 190 191 192 193 194 195 196 197 198 199
{
	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;
200
	req->retries = retries;
201
	req->timeout = timeout;
202 203
	req->cmd_type = REQ_TYPE_BLOCK_PC;
	req->cmd_flags |= flags | REQ_QUIET | REQ_PREEMPT;
204 205 206 207 208 209

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

210 211 212 213 214 215 216 217 218
	/*
	 * Some devices (USB mass-storage in particular) may transfer
	 * garbage data together with a residue indicating that the data
	 * is invalid.  Prevent the garbage from being misinterpreted
	 * and prevent security leaks by zeroing out the excess data.
	 */
	if (unlikely(req->data_len > 0 && req->data_len <= bufflen))
		memset(buffer + (bufflen - req->data_len), 0, req->data_len);

219 220 221 222 223 224
	ret = req->errors;
 out:
	blk_put_request(req);

	return ret;
}
225
EXPORT_SYMBOL(scsi_execute);
226

227 228 229 230 231 232

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;
233 234
	int result;
	
235
	if (sshdr) {
236
		sense = kzalloc(SCSI_SENSE_BUFFERSIZE, GFP_NOIO);
237 238 239
		if (!sense)
			return DRIVER_ERROR << 24;
	}
240
	result = scsi_execute(sdev, cmd, data_direction, buffer, bufflen,
241
			      sense, timeout, retries, 0);
242
	if (sshdr)
243
		scsi_normalize_sense(sense, SCSI_SENSE_BUFFERSIZE, sshdr);
244 245 246 247 248 249

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

250 251 252 253 254 255
struct scsi_io_context {
	void *data;
	void (*done)(void *data, char *sense, int result, int resid);
	char sense[SCSI_SENSE_BUFFERSIZE];
};

256
static struct kmem_cache *scsi_io_context_cache;
257

258
static void scsi_end_async(struct request *req, int uptodate)
259 260 261 262 263 264
{
	struct scsi_io_context *sioc = req->end_io_data;

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

265
	kmem_cache_free(scsi_io_context_cache, sioc);
266 267 268 269 270 271 272 273 274 275 276 277
	__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);

NeilBrown's avatar
NeilBrown committed
278
	return blk_rq_append_bio(q, rq, bio);
279 280
}

281
static void scsi_bi_endio(struct bio *bio, int error)
282 283 284 285 286 287 288
{
	bio_put(bio);
}

/**
 * scsi_req_map_sg - map a scatterlist into a request
 * @rq:		request to fill
289
 * @sgl:	scatterlist
290 291 292 293 294 295 296 297 298 299 300 301
 * @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;
302
	int nr_pages = (bufflen + sgl[0].offset + PAGE_SIZE - 1) >> PAGE_SHIFT;
303
	unsigned int data_len = bufflen, len, bytes, off;
Jens Axboe's avatar
Jens Axboe committed
304
	struct scatterlist *sg;
305 306 307 308
	struct page *page;
	struct bio *bio = NULL;
	int i, err, nr_vecs = 0;

Jens Axboe's avatar
Jens Axboe committed
309
	for_each_sg(sgl, sg, nsegs, i) {
310
		page = sg_page(sg);
Jens Axboe's avatar
Jens Axboe committed
311 312
		off = sg->offset;
		len = sg->length;
313

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

			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) {
345
					bio_endio(bio, 0);
346 347 348 349 350 351 352
					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
	return 0;

free_bios:
	while ((bio = rq->bio) != NULL) {
		rq->bio = bio->bi_next;
		/*
		 * call endio instead of bio_put incase it was bounced
		 */
368
		bio_endio(bio, 0);
369 370 371 372 373 374 375 376 377
	}

	return err;
}

/**
 * scsi_execute_async - insert request
 * @sdev:	scsi device
 * @cmd:	scsi command
378
 * @cmd_len:	length of scsi cdb
379
 * @data_direction: DMA_TO_DEVICE, DMA_FROM_DEVICE, or DMA_NONE
380 381 382 383 384
 * @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
385 386 387 388
 * @privdata:	data passed to done()
 * @done:	callback function when done
 * @gfp:	memory allocation flags
 */
389
int scsi_execute_async(struct scsi_device *sdev, const unsigned char *cmd,
390
		       int cmd_len, int data_direction, void *buffer, unsigned bufflen,
391 392 393 394 395 396 397 398
		       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);

399
	sioc = kmem_cache_zalloc(scsi_io_context_cache, gfp);
400 401 402 403 404 405
	if (!sioc)
		return DRIVER_ERROR << 24;

	req = blk_get_request(sdev->request_queue, write, gfp);
	if (!req)
		goto free_sense;
406 407
	req->cmd_type = REQ_TYPE_BLOCK_PC;
	req->cmd_flags |= REQ_QUIET;
408 409 410 411 412 413 414 415 416

	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;

417
	req->cmd_len = cmd_len;
418
	memset(req->cmd, 0, BLK_MAX_CDB); /* ATAPI hates garbage after CDB */
419 420 421 422
	memcpy(req->cmd, cmd, req->cmd_len);
	req->sense = sioc->sense;
	req->sense_len = 0;
	req->timeout = timeout;
423
	req->retries = retries;
424 425 426 427 428 429 430 431 432 433 434
	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:
435
	kmem_cache_free(scsi_io_context_cache, sioc);
436 437 438 439
	return DRIVER_ERROR << 24;
}
EXPORT_SYMBOL_GPL(scsi_execute_async);

Linus Torvalds's avatar
Linus Torvalds committed
440 441 442 443 444 445 446 447 448 449 450
/*
 * 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.
 */
451
static void scsi_init_cmd_errh(struct scsi_cmnd *cmd)
Linus Torvalds's avatar
Linus Torvalds committed
452 453
{
	cmd->serial_number = 0;
454
	scsi_set_resid(cmd, 0);
455
	memset(cmd->sense_buffer, 0, SCSI_SENSE_BUFFERSIZE);
Linus Torvalds's avatar
Linus Torvalds committed
456
	if (cmd->cmd_len == 0)
457
		cmd->cmd_len = scsi_command_size(cmd->cmnd);
Linus Torvalds's avatar
Linus Torvalds committed
458 459 460 461 462 463 464 465 466
}

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--;
467
	if (unlikely(scsi_host_in_recovery(shost) &&
468
		     (shost->host_failed || shost->host_eh_scheduled)))
Linus Torvalds's avatar
Linus Torvalds committed
469 470
		scsi_eh_wakeup(shost);
	spin_unlock(shost->host_lock);
's avatar
committed
471
	spin_lock(sdev->request_queue->queue_lock);
Linus Torvalds's avatar
Linus Torvalds committed
472
	sdev->device_busy--;
's avatar
committed
473
	spin_unlock_irqrestore(sdev->request_queue->queue_lock, flags);
Linus Torvalds's avatar
Linus Torvalds committed
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
}

/*
 * 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;

540
	if (scsi_target(sdev)->single_lun)
Linus Torvalds's avatar
Linus Torvalds committed
541 542 543 544 545 546 547
		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))) {
548 549 550

		int flagset;

Linus Torvalds's avatar
Linus Torvalds committed
551 552 553 554 555 556 557 558 559 560 561 562 563
		/*
		 * 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);
564 565 566 567 568 569 570 571 572 573 574 575
		spin_unlock(shost->host_lock);

		spin_lock(sdev->request_queue->queue_lock);
		flagset = test_bit(QUEUE_FLAG_REENTER, &q->queue_flags) &&
				!test_bit(QUEUE_FLAG_REENTER,
					&sdev->request_queue->queue_flags);
		if (flagset)
			queue_flag_set(QUEUE_FLAG_REENTER, sdev->request_queue);
		__blk_run_queue(sdev->request_queue);
		if (flagset)
			queue_flag_clear(QUEUE_FLAG_REENTER, sdev->request_queue);
		spin_unlock(sdev->request_queue->queue_lock);
576

577
		spin_lock(shost->host_lock);
Linus Torvalds's avatar
Linus Torvalds committed
578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606
		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.
607
 * Notes:	Upon return, cmd is a stale pointer.
Linus Torvalds's avatar
Linus Torvalds committed
608 609 610
 */
static void scsi_requeue_command(struct request_queue *q, struct scsi_cmnd *cmd)
{
611
	struct request *req = cmd->request;
612 613
	unsigned long flags;

614
	scsi_unprep_request(req);
615
	spin_lock_irqsave(q->queue_lock, flags);
616
	blk_requeue_request(q, req);
617
	spin_unlock_irqrestore(q->queue_lock, flags);
Linus Torvalds's avatar
Linus Torvalds committed
618 619 620 621 622 623

	scsi_run_queue(q);
}

void scsi_next_command(struct scsi_cmnd *cmd)
{
624 625 626 627 628
	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
629 630 631

	scsi_put_command(cmd);
	scsi_run_queue(q);
632 633 634

	/* ok to remove device now */
	put_device(&sdev->sdev_gendev);
Linus Torvalds's avatar
Linus Torvalds committed
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.
652
 *              error    - 0 if I/O indicates success, < 0 for I/O error.
Linus Torvalds's avatar
Linus Torvalds committed
653 654 655 656 657
 *              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.
 *
658
 * Returns:     cmd if requeue required, NULL otherwise.
Linus Torvalds's avatar
Linus Torvalds committed
659 660 661 662 663 664
 *
 * 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.
665
 * Notes:	If cmd was requeued, upon return it will be a stale pointer.
Linus Torvalds's avatar
Linus Torvalds committed
666
 */
667
static struct scsi_cmnd *scsi_end_request(struct scsi_cmnd *cmd, int error,
Linus Torvalds's avatar
Linus Torvalds committed
668 669
					  int bytes, int requeue)
{
670
	struct request_queue *q = cmd->device->request_queue;
Linus Torvalds's avatar
Linus Torvalds committed
671 672 673 674 675 676
	struct request *req = cmd->request;

	/*
	 * If there are blocks left over at the end, set up the command
	 * to queue the remainder of them.
	 */
677
	if (blk_end_request(req, error, bytes)) {
Linus Torvalds's avatar
Linus Torvalds committed
678 679 680 681 682 683
		int leftover = (req->hard_nr_sectors << 9);

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

		/* kill remainder if no retrys */
684 685
		if (error && blk_noretry_request(req))
			blk_end_request(req, error, leftover);
Linus Torvalds's avatar
Linus Torvalds committed
686
		else {
687
			if (requeue) {
Linus Torvalds's avatar
Linus Torvalds committed
688 689 690 691 692 693
				/*
				 * Bleah.  Leftovers again.  Stick the
				 * leftovers in the front of the
				 * queue, and goose the queue again.
				 */
				scsi_requeue_command(q, cmd);
694 695
				cmd = NULL;
			}
Linus Torvalds's avatar
Linus Torvalds committed
696 697 698 699 700 701 702 703 704 705 706 707
			return cmd;
		}
	}

	/*
	 * 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;
}

708 709 710 711
static inline unsigned int scsi_sgtable_index(unsigned short nents)
{
	unsigned int index;

712 713 714
	BUG_ON(nents > SCSI_MAX_SG_SEGMENTS);

	if (nents <= 8)
715
		index = 0;
716 717
	else
		index = get_count_order(nents) - 3;
Linus Torvalds's avatar
Linus Torvalds committed
718

719 720 721
	return index;
}

722
static void scsi_sg_free(struct scatterlist *sgl, unsigned int nents)
723 724 725
{
	struct scsi_host_sg_pool *sgp;

726 727 728
	sgp = scsi_sg_pools + scsi_sgtable_index(nents);
	mempool_free(sgl, sgp->pool);
}
729

730 731 732
static struct scatterlist *scsi_sg_alloc(unsigned int nents, gfp_t gfp_mask)
{
	struct scsi_host_sg_pool *sgp;
733

734 735 736
	sgp = scsi_sg_pools + scsi_sgtable_index(nents);
	return mempool_alloc(sgp->pool, gfp_mask);
}
737

738 739
static int scsi_alloc_sgtable(struct scsi_data_buffer *sdb, int nents,
			      gfp_t gfp_mask)
740 741
{
	int ret;
742

743
	BUG_ON(!nents);
744

745 746
	ret = __sg_alloc_table(&sdb->table, nents, SCSI_MAX_SG_SEGMENTS,
			       gfp_mask, scsi_sg_alloc);
747
	if (unlikely(ret))
748
		__sg_free_table(&sdb->table, SCSI_MAX_SG_SEGMENTS,
749
				scsi_sg_free);
750

751
	return ret;
Linus Torvalds's avatar
Linus Torvalds committed
752 753
}

754
static void scsi_free_sgtable(struct scsi_data_buffer *sdb)
Linus Torvalds's avatar
Linus Torvalds committed
755
{
756
	__sg_free_table(&sdb->table, SCSI_MAX_SG_SEGMENTS, scsi_sg_free);
Linus Torvalds's avatar
Linus Torvalds committed
757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775
}

/*
 * 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.
 */
776
void scsi_release_buffers(struct scsi_cmnd *cmd)
Linus Torvalds's avatar
Linus Torvalds committed
777
{
778 779
	if (cmd->sdb.table.nents)
		scsi_free_sgtable(&cmd->sdb);
Linus Torvalds's avatar
Linus Torvalds committed
780

781
	memset(&cmd->sdb, 0, sizeof(cmd->sdb));
782 783 784 785 786

	if (scsi_bidi_cmnd(cmd)) {
		struct scsi_data_buffer *bidi_sdb =
			cmd->request->next_rq->special;
		scsi_free_sgtable(bidi_sdb);
787
		kmem_cache_free(scsi_sdb_cache, bidi_sdb);
788 789
		cmd->request->next_rq->special = NULL;
	}
Linus Torvalds's avatar
Linus Torvalds committed
790
}
791
EXPORT_SYMBOL(scsi_release_buffers);
Linus Torvalds's avatar
Linus Torvalds committed
792

793 794 795 796 797 798 799
/*
 * Bidi commands Must be complete as a whole, both sides at once.
 * If part of the bytes were written and lld returned
 * scsi_in()->resid and/or scsi_out()->resid this information will be left
 * in req->data_len and req->next_rq->data_len. The upper-layer driver can
 * decide what to do with this information.
 */
800
static void scsi_end_bidi_request(struct scsi_cmnd *cmd)
801
{
802 803 804 805 806 807 808 809 810 811
	struct request *req = cmd->request;
	unsigned int dlen = req->data_len;
	unsigned int next_dlen = req->next_rq->data_len;

	req->data_len = scsi_out(cmd)->resid;
	req->next_rq->data_len = scsi_in(cmd)->resid;

	/* The req and req->next_rq have not been completed */
	BUG_ON(blk_end_bidi_request(req, 0, dlen, next_dlen));

812 813 814 815 816 817 818 819 820
	scsi_release_buffers(cmd);

	/*
	 * 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);
}

Linus Torvalds's avatar
Linus Torvalds committed
821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848
/*
 * 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.
 */
849
void scsi_io_completion(struct scsi_cmnd *cmd, unsigned int good_bytes)
Linus Torvalds's avatar
Linus Torvalds committed
850 851
{
	int result = cmd->result;
852
	int this_count = scsi_bufflen(cmd);
853
	struct request_queue *q = cmd->device->request_queue;
Linus Torvalds's avatar
Linus Torvalds committed
854
	struct request *req = cmd->request;
855
	int error = 0;
Linus Torvalds's avatar
Linus Torvalds committed
856 857 858 859 860 861 862 863 864
	struct scsi_sense_hdr sshdr;
	int sense_valid = 0;
	int sense_deferred = 0;

	if (result) {
		sense_valid = scsi_command_normalize_sense(cmd, &sshdr);
		if (sense_valid)
			sense_deferred = scsi_sense_is_deferred(&sshdr);
	}
865

Linus Torvalds's avatar
Linus Torvalds committed
866 867 868 869 870 871 872 873 874 875 876 877 878 879
	if (blk_pc_request(req)) { /* SG_IO ioctl from block level */
		req->errors = result;
		if (result) {
			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;
			}
880 881
			if (!sense_deferred)
				error = -EIO;
882
		}
883 884 885 886 887
		if (scsi_bidi_cmnd(cmd)) {
			/* will also release_buffers */
			scsi_end_bidi_request(cmd);
			return;
		}
888
		req->data_len = scsi_get_resid(cmd);
Linus Torvalds's avatar
Linus Torvalds committed
889 890
	}

891
	BUG_ON(blk_bidi_rq(req)); /* bidi not support for !blk_pc_request yet */
892 893
	scsi_release_buffers(cmd);

Linus Torvalds's avatar
Linus Torvalds committed
894 895 896 897
	/*
	 * Next deal with any sectors which we were able to correctly
	 * handle.
	 */
898 899 900 901 902 903 904 905
	SCSI_LOG_HLCOMPLETE(1, printk("%ld sectors total, "
				      "%d bytes done.\n",
				      req->nr_sectors, good_bytes));

	/* A number of bytes were successfully read.  If there
	 * are leftovers and there is some kind of error
	 * (result != 0), retry the rest.
	 */
906
	if (scsi_end_request(cmd, error, good_bytes, result == 0) == NULL)
907
		return;
908 909 910

	/* 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
911 912 913 914 915
	 */
	if (sense_valid && !sense_deferred) {
		switch (sshdr.sense_key) {
		case UNIT_ATTENTION:
			if (cmd->device->removable) {
916
				/* Detected disc change.  Set a bit
Linus Torvalds's avatar
Linus Torvalds committed
917 918 919
				 * and quietly refuse further access.
				 */
				cmd->device->changed = 1;
920
				scsi_end_request(cmd, -EIO, this_count, 1);
Linus Torvalds's avatar
Linus Torvalds committed
921 922
				return;
			} else {
923 924 925 926 927
				/* 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
928 929 930 931 932
				scsi_requeue_command(q, cmd);
				return;
			}
			break;
		case ILLEGAL_REQUEST:
933 934 935 936 937 938 939 940
			/* 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.
			 */
941 942
			if ((cmd->device->use_10_for_rw &&
			    sshdr.asc == 0x20 && sshdr.ascq == 0x00) &&
Linus Torvalds's avatar
Linus Torvalds committed
943 944 945
			    (cmd->cmnd[0] == READ_10 ||
			     cmd->cmnd[0] == WRITE_10)) {
				cmd->device->use_10_for_rw = 0;
946 947
				/* This will cause a retry with a
				 * 6-byte command.
Linus Torvalds's avatar
Linus Torvalds committed
948 949
				 */
				scsi_requeue_command(q, cmd);
950
				return;
Linus Torvalds's avatar
Linus Torvalds committed
951
			} else {
952
				scsi_end_request(cmd, -EIO, this_count, 1);
Linus Torvalds's avatar
Linus Torvalds committed
953 954 955 956
				return;
			}
			break;
		case NOT_READY:
957
			/* If the device is in the process of becoming
958
			 * ready, or has a temporary blockage, retry.
Linus Torvalds's avatar
Linus Torvalds committed
959
			 */
960 961 962 963 964 965 966 967 968 969 970 971 972 973
			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
974
			}
975 976 977 978 979
			if (!(req->cmd_flags & REQ_QUIET))
				scsi_cmd_print_sense_hdr(cmd,
							 "Device not ready",
							 &sshdr);

980
			scsi_end_request(cmd, -EIO, this_count, 1);
Linus Torvalds's avatar
Linus Torvalds committed
981 982
			return;
		case VOLUME_OVERFLOW:
983
			if (!(req->cmd_flags & REQ_QUIET)) {
984
				scmd_printk(KERN_INFO, cmd,
985
					    "Volume overflow, CDB: ");
986
				__scsi_print_command(cmd->cmnd);
987 988
				scsi_print_sense("", cmd);
			}
989
			/* See SSC3rXX or current. */
990
			scsi_end_request(cmd, -EIO, this_count, 1);
Linus Torvalds's avatar
Linus Torvalds committed
991 992 993 994
			return;
		default:
			break;
		}
995
	}
Linus Torvalds's avatar
Linus Torvalds committed
996
	if (host_byte(result) == DID_RESET) {
997 998 999
		/* 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
1000 1001 1002 1003 1004
		 */
		scsi_requeue_command(q, cmd);
		return;
	}
	if (result) {
1005
		if (!(req->cmd_flags & REQ_QUIET)) {
1006
			scsi_print_result(cmd);
1007 1008 1009
			if (driver_byte(result) & DRIVER_SENSE)
				scsi_print_sense("", cmd);
		}
Linus Torvalds's avatar
Linus Torvalds committed
1010
	}
1011
	scsi_end_request(cmd, -EIO, this_count, !result);
Linus Torvalds's avatar
Linus Torvalds committed
1012 1013
}

1014 1015
static int scsi_init_sgtable(struct request *req, struct scsi_data_buffer *sdb,
			     gfp_t gfp_mask)
Linus Torvalds's avatar
Linus Torvalds committed
1016
{
1017
	int count;
Linus Torvalds's avatar
Linus Torvalds committed
1018 1019

	/*
1020
	 * If sg table allocation fails, requeue request later.
Linus Torvalds's avatar
Linus Torvalds committed
1021
	 */
1022 1023
	if (unlikely(scsi_alloc_sgtable(sdb, req->nr_phys_segments,
					gfp_mask))) {
Linus Torvalds's avatar
Linus Torvalds committed
1024
		return BLKPREP_DEFER;
Alan Stern's avatar
Alan Stern committed
1025
	}
Linus Torvalds's avatar
Linus Torvalds committed
1026

1027
	req->buffer = NULL;
Linus Torvalds's avatar
Linus Torvalds committed
1028 1029 1030 1031 1032

	/* 
	 * Next, walk the list, and fill in the addresses and sizes of
	 * each segment.
	 */
1033 1034 1035
	count = blk_rq_map_sg(req->q, req, sdb->table.sgl);
	BUG_ON(count > sdb->table.nents);
	sdb->table.nents = count;
Tejun Heo's avatar
Tejun Heo committed
1036 1037 1038 1039
	if (blk_pc_request(req))
		sdb->length = req->data_len;
	else
		sdb->length = req->nr_sectors << 9;
1040
	return BLKPREP_OK;
Linus Torvalds's avatar
Linus Torvalds committed
1041
}
1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061

/*
 * 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
 */
int scsi_init_io(struct scsi_cmnd *cmd, gfp_t gfp_mask)
{
	int error = scsi_init_sgtable(cmd->request, &cmd->sdb, gfp_mask);
	if (error)
		goto err_exit;

	if (blk_bidi_rq(cmd->request)) {
		struct scsi_data_buffer *bidi_sdb = kmem_cache_zalloc(
1062
			scsi_sdb_cache, GFP_ATOMIC);
1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085
		if (!bidi_sdb) {
			error = BLKPREP_DEFER;
			goto err_exit;
		}

		cmd->request->next_rq->special = bidi_sdb;
		error = scsi_init_sgtable(cmd->request->next_rq, bidi_sdb,
								    GFP_ATOMIC);
		if (error)
			goto err_exit;
	}

	return BLKPREP_OK ;

err_exit:
	scsi_release_buffers(cmd);
	if (error == BLKPREP_KILL)
		scsi_put_command(cmd);
	else /* BLKPREP_DEFER */
		scsi_unprep_request(cmd->request);

	return error;
}
1086
EXPORT_SYMBOL(scsi_init_io);
Linus Torvalds's avatar
Linus Torvalds committed
1087

1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105
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;

1106 1107
	cmd->cmnd = req->cmd;

1108 1109 1110
	return cmd;
}

1111
int scsi_setup_blk_pc_cmnd(struct scsi_device *sdev, struct request *req)
James Bottomley's avatar
James Bottomley committed
1112
{
1113
	struct scsi_cmnd *cmd;
1114 1115 1116 1117
	int ret = scsi_prep_state_check(sdev, req);

	if (ret != BLKPREP_OK)
		return ret;
1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133

	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);

1134
		ret = scsi_init_io(cmd, GFP_ATOMIC);
1135 1136 1137 1138 1139 1140
		if (unlikely(ret))
			return ret;
	} else {
		BUG_ON(req->data_len);
		BUG_ON(req->data);

1141
		memset(&cmd->sdb, 0, sizeof(cmd->sdb));
1142 1143
		req->buffer = NULL;
	}
James Bottomley's avatar
James Bottomley committed
1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155

	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;
1156
	return BLKPREP_OK;
James Bottomley's avatar
James Bottomley committed
1157
}
1158
EXPORT_SYMBOL(scsi_setup_blk_pc_cmnd);
James Bottomley's avatar
James Bottomley committed
1159

1160 1161 1162 1163 1164
/*
 * 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.
 */
1165
int scsi_setup_fs_cmnd(struct scsi_device *sdev, struct request *req)
Linus Torvalds's avatar
Linus Torvalds committed
1166 1167
{
	struct scsi_cmnd *cmd;