scsi_lib.c 62.1 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 22 23 24 25 26 27 28 29 30 31 32

#include <scsi/scsi.h>
#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/scsi_request.h>

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


33
#define SG_MEMPOOL_NR		ARRAY_SIZE(scsi_sg_pools)
Linus Torvalds's avatar
Linus Torvalds committed
34 35 36 37 38 39 40 41 42 43 44 45 46 47
#define SG_MEMPOOL_SIZE		32

struct scsi_host_sg_pool {
	size_t		size;
	char		*name; 
	kmem_cache_t	*slab;
	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 85 86 87 88 89

/*
 * 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;

	req->flags &= ~REQ_DONTPREP;
	req->special = (req->flags & REQ_SPECIAL) ? cmd->sc_request : NULL;

	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
161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219
	return 0;
}

/*
 * Function:    scsi_do_req
 *
 * Purpose:     Queue a SCSI request
 *
 * Arguments:   sreq	  - command descriptor.
 *              cmnd      - actual SCSI command to be performed.
 *              buffer    - data buffer.
 *              bufflen   - size of data buffer.
 *              done      - completion function to be run.
 *              timeout   - how long to let it run before timeout.
 *              retries   - number of retries we allow.
 *
 * Lock status: No locks held upon entry.
 *
 * Returns:     Nothing.
 *
 * Notes:	This function is only used for queueing requests for things
 *		like ioctls and character device requests - this is because
 *		we essentially just inject a request into the queue for the
 *		device.
 *
 *		In order to support the scsi_device_quiesce function, we
 *		now inject requests on the *head* of the device queue
 *		rather than the tail.
 */
void scsi_do_req(struct scsi_request *sreq, const void *cmnd,
		 void *buffer, unsigned bufflen,
		 void (*done)(struct scsi_cmnd *),
		 int timeout, int retries)
{
	/*
	 * If the upper level driver is reusing these things, then
	 * we should release the low-level block now.  Another one will
	 * be allocated later when this request is getting queued.
	 */
	__scsi_release_request(sreq);

	/*
	 * Our own function scsi_done (which marks the host as not busy,
	 * disables the timeout counter, etc) will be called by us or by the
	 * scsi_hosts[host].queuecommand() function needs to also call
	 * the completion function for the high level driver.
	 */
	memcpy(sreq->sr_cmnd, cmnd, sizeof(sreq->sr_cmnd));
	sreq->sr_bufflen = bufflen;
	sreq->sr_buffer = buffer;
	sreq->sr_allowed = retries;
	sreq->sr_done = done;
	sreq->sr_timeout_per_command = timeout;

	if (sreq->sr_cmd_len == 0)
		sreq->sr_cmd_len = COMMAND_SIZE(sreq->sr_cmnd[0]);

	/*
	 * head injection *required* here otherwise quiesce won't work
220 221 222
	 *
	 * Because users of this function are apt to reuse requests with no
	 * modification, we have to sanitise the request flags here
Linus Torvalds's avatar
Linus Torvalds committed
223
	 */
224 225 226
	sreq->sr_request->flags &= ~REQ_DONTPREP;
	blk_insert_request(sreq->sr_device->request_queue, sreq->sr_request,
		       	   1, sreq);
Linus Torvalds's avatar
Linus Torvalds committed
227 228 229
}
EXPORT_SYMBOL(scsi_do_req);

230
/**
231
 * scsi_execute - insert request and wait for the result
232 233 234 235 236 237 238 239
 * @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
240
 * @flags:	or into request flags;
241
 *
242 243
 * returns the req->errors value which is the the scsi_cmnd result
 * field.
244
 **/
245 246 247
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)
248 249 250 251 252 253 254 255 256 257 258 259 260 261 262
{
	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;
263
	req->retries = retries;
264
	req->timeout = timeout;
265
	req->flags |= flags | REQ_BLOCK_PC | REQ_SPECIAL | REQ_QUIET;
266 267 268 269 270 271 272 273 274 275 276 277

	/*
	 * 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;
}
278
EXPORT_SYMBOL(scsi_execute);
279

280 281 282 283 284 285

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;
286 287
	int result;
	
288
	if (sshdr) {
289
		sense = kzalloc(SCSI_SENSE_BUFFERSIZE, GFP_NOIO);
290 291 292
		if (!sense)
			return DRIVER_ERROR << 24;
	}
293
	result = scsi_execute(sdev, cmd, data_direction, buffer, bufflen,
294
			      sense, timeout, retries, 0);
295
	if (sshdr)
296
		scsi_normalize_sense(sense, SCSI_SENSE_BUFFERSIZE, sshdr);
297 298 299 300 301 302

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

303 304 305 306 307 308
struct scsi_io_context {
	void *data;
	void (*done)(void *data, char *sense, int result, int resid);
	char sense[SCSI_SENSE_BUFFERSIZE];
};

309 310
static kmem_cache_t *scsi_io_context_cache;

311
static void scsi_end_async(struct request *req, int uptodate)
312 313 314 315 316 317
{
	struct scsi_io_context *sioc = req->end_io_data;

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

318
	kmem_cache_free(scsi_io_context_cache, sioc);
319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369
	__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);
	else if (!q->back_merge_fn(q, rq, bio))
		return -EINVAL;
	else {
		rq->biotail->bi_next = bio;
		rq->biotail = bio;
		rq->hard_nr_sectors += bio_sectors(bio);
		rq->nr_sectors = rq->hard_nr_sectors;
	}

	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;
370
	int nr_pages = (bufflen + sgl[0].offset + PAGE_SIZE - 1) >> PAGE_SHIFT;
371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438
	unsigned int data_len = 0, len, bytes, off;
	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;
		data_len += len;

		while (len > 0) {
			bytes = min_t(unsigned int, len, PAGE_SIZE - off);

			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;
			off = 0;
		}
	}

	rq->buffer = rq->data = NULL;
	rq->data_len = data_len;
	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
439
 * @cmd_len:	length of scsi cdb
440 441 442 443 444 445 446 447 448
 * @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,
449
		       int cmd_len, int data_direction, void *buffer, unsigned bufflen,
450 451 452 453 454 455 456 457
		       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);

458
	sioc = kmem_cache_alloc(scsi_io_context_cache, gfp);
459 460
	if (!sioc)
		return DRIVER_ERROR << 24;
461
	memset(sioc, 0, sizeof(*sioc));
462 463 464 465

	req = blk_get_request(sdev->request_queue, write, gfp);
	if (!req)
		goto free_sense;
466
	req->flags |= REQ_BLOCK_PC | REQ_QUIET;
467 468 469 470 471 472 473 474 475

	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;

476
	req->cmd_len = cmd_len;
477 478 479 480
	memcpy(req->cmd, cmd, req->cmd_len);
	req->sense = sioc->sense;
	req->sense_len = 0;
	req->timeout = timeout;
481
	req->retries = retries;
482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497
	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:
	kfree(sioc);
	return DRIVER_ERROR << 24;
}
EXPORT_SYMBOL_GPL(scsi_execute_async);

Linus Torvalds's avatar
Linus Torvalds committed
498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567
/*
 * Function:    scsi_init_cmd_errh()
 *
 * Purpose:     Initialize cmd fields related to error handling.
 *
 * Arguments:   cmd	- command that is ready to be queued.
 *
 * Returns:     Nothing
 *
 * 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.
 */
static int scsi_init_cmd_errh(struct scsi_cmnd *cmd)
{
	cmd->serial_number = 0;

	memset(cmd->sense_buffer, 0, sizeof cmd->sense_buffer);

	if (cmd->cmd_len == 0)
		cmd->cmd_len = COMMAND_SIZE(cmd->cmnd[0]);

	/*
	 * We need saved copies of a number of fields - this is because
	 * error handling may need to overwrite these with different values
	 * to run different commands, and once error handling is complete,
	 * we will need to restore these values prior to running the actual
	 * command.
	 */
	cmd->old_use_sg = cmd->use_sg;
	cmd->old_cmd_len = cmd->cmd_len;
	cmd->sc_old_data_direction = cmd->sc_data_direction;
	cmd->old_underflow = cmd->underflow;
	memcpy(cmd->data_cmnd, cmd->cmnd, sizeof(cmd->cmnd));
	cmd->buffer = cmd->request_buffer;
	cmd->bufflen = cmd->request_bufflen;

	return 1;
}

/*
 * Function:   scsi_setup_cmd_retry()
 *
 * Purpose:    Restore the command state for a retry
 *
 * Arguments:  cmd	- command to be restored
 *
 * Returns:    Nothing
 *
 * Notes:      Immediately prior to retrying a command, we need
 *             to restore certain fields that we saved above.
 */
void scsi_setup_cmd_retry(struct scsi_cmnd *cmd)
{
	memcpy(cmd->cmnd, cmd->data_cmnd, sizeof(cmd->data_cmnd));
	cmd->request_buffer = cmd->buffer;
	cmd->request_bufflen = cmd->bufflen;
	cmd->use_sg = cmd->old_use_sg;
	cmd->cmd_len = cmd->old_cmd_len;
	cmd->sc_data_direction = cmd->sc_old_data_direction;
	cmd->underflow = cmd->old_underflow;
}

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--;
568
	if (unlikely(scsi_host_in_recovery(shost) &&
Linus Torvalds's avatar
Linus Torvalds committed
569 570 571
		     shost->host_failed))
		scsi_eh_wakeup(shost);
	spin_unlock(shost->host_lock);
's avatar
committed
572
	spin_lock(sdev->request_queue->queue_lock);
Linus Torvalds's avatar
Linus Torvalds committed
573
	sdev->device_busy--;
's avatar
committed
574
	spin_unlock_irqrestore(sdev->request_queue->queue_lock, flags);
Linus Torvalds's avatar
Linus Torvalds committed
575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695
}

/*
 * 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);

		blk_run_queue(sdev->request_queue);

		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.
696
 * Notes:	Upon return, cmd is a stale pointer.
Linus Torvalds's avatar
Linus Torvalds committed
697 698 699
 */
static void scsi_requeue_command(struct request_queue *q, struct scsi_cmnd *cmd)
{
700
	struct request *req = cmd->request;
701 702
	unsigned long flags;

703
	scsi_unprep_request(req);
704
	spin_lock_irqsave(q->queue_lock, flags);
705
	blk_requeue_request(q, req);
706
	spin_unlock_irqrestore(q->queue_lock, flags);
Linus Torvalds's avatar
Linus Torvalds committed
707 708 709 710 711 712

	scsi_run_queue(q);
}

void scsi_next_command(struct scsi_cmnd *cmd)
{
713 714 715 716 717
	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
718 719 720

	scsi_put_command(cmd);
	scsi_run_queue(q);
721 722 723

	/* ok to remove device now */
	put_device(&sdev->sdev_gendev);
Linus Torvalds's avatar
Linus Torvalds committed
724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746
}

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.
 *
747
 * Returns:     cmd if requeue required, NULL otherwise.
Linus Torvalds's avatar
Linus Torvalds committed
748 749 750 751 752 753
 *
 * 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.
754
 * Notes:	If cmd was requeued, upon return it will be a stale pointer.
Linus Torvalds's avatar
Linus Torvalds committed
755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776
 */
static struct scsi_cmnd *scsi_end_request(struct scsi_cmnd *cmd, int uptodate,
					  int bytes, int requeue)
{
	request_queue_t *q = cmd->device->request_queue;
	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 {
777
			if (requeue) {
Linus Torvalds's avatar
Linus Torvalds committed
778 779 780 781 782 783
				/*
				 * Bleah.  Leftovers again.  Stick the
				 * leftovers in the front of the
				 * queue, and goose the queue again.
				 */
				scsi_requeue_command(q, cmd);
784 785
				cmd = NULL;
			}
Linus Torvalds's avatar
Linus Torvalds committed
786 787 788 789 790 791 792 793 794
			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);
795
	end_that_request_last(req, uptodate);
Linus Torvalds's avatar
Linus Torvalds committed
796 797 798 799 800 801 802 803 804 805
	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;
}

Al Viro's avatar
Al Viro committed
806
static struct scatterlist *scsi_alloc_sgtable(struct scsi_cmnd *cmd, gfp_t gfp_mask)
Linus Torvalds's avatar
Linus Torvalds committed
807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850
{
	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;
}

static void scsi_free_sgtable(struct scatterlist *sgl, int index)
{
	struct scsi_host_sg_pool *sgp;

851
	BUG_ON(index >= SG_MEMPOOL_NR);
Linus Torvalds's avatar
Linus Torvalds committed
852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009

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

/*
 * 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)
{
	struct request *req = cmd->request;

	/*
	 * Free up any indirection buffers we allocated for DMA purposes. 
	 */
	if (cmd->use_sg)
		scsi_free_sgtable(cmd->request_buffer, cmd->sglist_len);
	else if (cmd->request_buffer != req->buffer)
		kfree(cmd->request_buffer);

	/*
	 * Zero these out.  They now point to freed memory, and it is
	 * dangerous to hang onto the pointers.
	 */
	cmd->buffer  = NULL;
	cmd->bufflen = 0;
	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.
 */
void scsi_io_completion(struct scsi_cmnd *cmd, unsigned int good_bytes,
			unsigned int block_bytes)
{
	int result = cmd->result;
	int this_count = cmd->bufflen;
	request_queue_t *q = cmd->device->request_queue;
	struct request *req = cmd->request;
	int clear_errors = 1;
	struct scsi_sense_hdr sshdr;
	int sense_valid = 0;
	int sense_deferred = 0;

	/*
	 * Free up any indirection buffers we allocated for DMA purposes. 
	 * For the case of a READ, we need to copy the data out of the
	 * bounce buffer and into the real buffer.
	 */
	if (cmd->use_sg)
		scsi_free_sgtable(cmd->buffer, cmd->sglist_len);
	else if (cmd->buffer != req->buffer) {
		if (rq_data_dir(req) == READ) {
			unsigned long flags;
			char *to = bio_kmap_irq(req->bio, &flags);
			memcpy(to, cmd->buffer, cmd->bufflen);
			bio_kunmap_irq(to, &flags);
		}
		kfree(cmd->buffer);
	}

	if (result) {
		sense_valid = scsi_command_normalize_sense(cmd, &sshdr);
		if (sense_valid)
			sense_deferred = scsi_sense_is_deferred(&sshdr);
	}
	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;
			}
		} else
			req->data_len = cmd->resid;
	}

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

	/*
	 * Next deal with any sectors which we were able to correctly
	 * handle.
	 */
	if (good_bytes >= 0) {
		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;
		/*
		 * If multiple sectors are requested in one buffer, then
		 * they will have been finished off by the first command.
		 * If not, then we have a multi-buffer command.
		 *
		 * If block_bytes != 0, it means we had a medium error
		 * of some sort, and that we want to mark some number of
		 * sectors as not uptodate.  Thus we want to inhibit
		 * requeueing right here - we will requeue down below
		 * when we handle the bad sectors.
		 */

		/*
1010 1011
		 * If the command completed without error, then either
		 * finish off the rest of the command, or start a new one.
Linus Torvalds's avatar
Linus Torvalds committed
1012
		 */
1013
		if (scsi_end_request(cmd, 1, good_bytes, result == 0) == NULL)
Linus Torvalds's avatar
Linus Torvalds committed
1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028
			return;
	}
	/*
	 * Now, if we were good little boys and girls, Santa left us a request
	 * sense buffer.  We can extract information from this, so we
	 * can choose a block to remap, etc.
	 */
	if (sense_valid && !sense_deferred) {
		switch (sshdr.sense_key) {
		case UNIT_ATTENTION:
			if (cmd->device->removable) {
				/* detected disc change.  set a bit 
				 * and quietly refuse further access.
				 */
				cmd->device->changed = 1;
1029
				scsi_end_request(cmd, 0,
Linus Torvalds's avatar
Linus Torvalds committed
1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051
						this_count, 1);
				return;
			} else {
				/*
				* 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.  
				*/
				scsi_requeue_command(q, cmd);
				return;
			}
			break;
		case ILLEGAL_REQUEST:
			/*
		 	* 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.
		 	*/
1052 1053
			if ((cmd->device->use_10_for_rw &&
			    sshdr.asc == 0x20 && sshdr.ascq == 0x00) &&
Linus Torvalds's avatar
Linus Torvalds committed
1054 1055 1056 1057 1058 1059 1060 1061 1062 1063
			    (cmd->cmnd[0] == READ_10 ||
			     cmd->cmnd[0] == WRITE_10)) {
				cmd->device->use_10_for_rw = 0;
				/*
				 * This will cause a retry with a 6-byte
				 * command.
				 */
				scsi_requeue_command(q, cmd);
				result = 0;
			} else {
1064
				scsi_end_request(cmd, 0, this_count, 1);
Linus Torvalds's avatar
Linus Torvalds committed
1065 1066 1067 1068 1069
				return;
			}
			break;
		case NOT_READY:
			/*
1070 1071
			 * If the device is in the process of becoming
			 * ready, or has a temporary blockage, retry.
Linus Torvalds's avatar
Linus Torvalds committed
1072
			 */
1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086
			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
1087
			}
1088
			if (!(req->flags & REQ_QUIET)) {
1089
				scmd_printk(KERN_INFO, cmd,
1090 1091 1092
					   "Device not ready: ");
				scsi_print_sense_hdr("", &sshdr);
			}
1093
			scsi_end_request(cmd, 0, this_count, 1);
Linus Torvalds's avatar
Linus Torvalds committed
1094 1095
			return;
		case VOLUME_OVERFLOW:
1096
			if (!(req->flags & REQ_QUIET)) {
1097 1098
				scmd_printk(KERN_INFO, cmd,
					   "Volume overflow, CDB: ");
1099 1100 1101
				__scsi_print_command(cmd->data_cmnd);
				scsi_print_sense("", cmd);
			}
1102
			scsi_end_request(cmd, 0, block_bytes, 1);
Linus Torvalds's avatar
Linus Torvalds committed
1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117
			return;
		default:
			break;
		}
	}			/* driver byte != 0 */
	if (host_byte(result) == DID_RESET) {
		/*
		 * Third party bus reset or reset for error
		 * recovery reasons.  Just retry the request
		 * and see what happens.  
		 */
		scsi_requeue_command(q, cmd);
		return;
	}
	if (result) {
1118
		if (!(req->flags & REQ_QUIET)) {
1119 1120
			scmd_printk(KERN_INFO, cmd,
				   "SCSI error: return code = 0x%x\n", result);
1121 1122 1123 1124

			if (driver_byte(result) & DRIVER_SENSE)
				scsi_print_sense("", cmd);
		}
Linus Torvalds's avatar
Linus Torvalds committed
1125 1126 1127 1128 1129 1130 1131 1132
		/*
		 * Mark a single buffer as not uptodate.  Queue the remainder.
		 * We sometimes get this cruft in the event that a medium error
		 * isn't properly reported.
		 */
		block_bytes = req->hard_cur_sectors << 9;
		if (!block_bytes)
			block_bytes = req->data_len;
1133
		scsi_end_request(cmd, 0, block_bytes, 1);
Linus Torvalds's avatar
Linus Torvalds committed
1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176
	}
}
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;

	/*
	 * if this is a rq->data based REQ_BLOCK_PC, setup for a non-sg xfer
	 */
	if ((req->flags & REQ_BLOCK_PC) && !req->bio) {
		cmd->request_bufflen = req->data_len;
		cmd->request_buffer = req->data;
		req->buffer = req->data;
		cmd->use_sg = 0;
		return 0;
	}

	/*
	 * we used to not use scatter-gather for single segment request,
	 * but now we do (it makes highmem I/O easier to support without
	 * kmapping pages)
	 */
	cmd->use_sg = req->nr_phys_segments;

	/*
	 * if sg table allocation fails, requeue request later.
	 */
	sgpnt = scsi_alloc_sgtable(cmd, GFP_ATOMIC);
Alan Stern's avatar
Alan Stern committed
1177 1178
	if (unlikely(!sgpnt)) {
		scsi_unprep_request(req);
Linus Torvalds's avatar
Linus Torvalds committed
1179
		return BLKPREP_DEFER;
Alan Stern's avatar
Alan Stern committed
1180
	}
Linus Torvalds's avatar
Linus Torvalds committed
1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228

	cmd->request_buffer = (char *) sgpnt;
	cmd->request_bufflen = req->nr_sectors << 9;
	if (blk_pc_request(req))
		cmd->request_bufflen = req->data_len;
	req->buffer = NULL;

	/* 
	 * 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);

	/*
	 * mapped well, send it off
	 */
	if (likely(count <= cmd->use_sg)) {
		cmd->use_sg = count;
		return 0;
	}

	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);

	/* release the command and kill it */
	scsi_release_buffers(cmd);
	scsi_put_command(cmd);
	return BLKPREP_KILL;
}

static int scsi_issue_flush_fn(request_queue_t *q, struct gendisk *disk,
			       sector_t *error_sector)
{
	struct scsi_device *sdev = q->queuedata;
	struct scsi_driver *drv;

	if (sdev->sdev_state != SDEV_RUNNING)
		return -ENXIO;

	drv = *(struct scsi_driver **) disk->private_data;
	if (drv->issue_flush)
		return drv->issue_flush(&sdev->sdev_gendev, error_sector);

	return -EOPNOTSUPP;
}

1229
static void scsi_blk_pc_done(struct scsi_cmnd *cmd)
1230 1231
{
	BUG_ON(!blk_pc_request(cmd->request));
1232 1233 1234 1235 1236 1237 1238
	/*
	 * 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
	 */
	scsi_io_completion(cmd, cmd->bufflen, 0);
1239 1240
}

1241
static void scsi_setup_blk_pc_cmnd(struct scsi_cmnd *cmd)
James Bottomley's avatar
James Bottomley committed
1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257
{
	struct request *req = cmd->request;

	BUG_ON(sizeof(req->cmd) > sizeof(cmd->cmnd));
	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;
1258
	cmd->done = scsi_blk_pc_done;
James Bottomley's avatar
James Bottomley committed
1259 1260
}

Linus Torvalds's avatar
Linus Torvalds committed
1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272
static int scsi_prep_fn(struct request_queue *q, struct request *req)
{
	struct scsi_device *sdev = q->queuedata;
	struct scsi_cmnd *cmd;
	int specials_only = 0;

	/*
	 * Just check to see if the device is online.  If it isn't, we
	 * refuse to process any commands.  The device must be brought
	 * online before trying any recovery commands
	 */
	if (unlikely(!scsi_device_online(sdev))) {
1273 1274
		sdev_printk(KERN_ERR, sdev,
			    "rejecting I/O to offline device\n");
1275
		goto kill;
Linus Torvalds's avatar
Linus Torvalds committed
1276 1277 1278 1279 1280 1281 1282
	}
	if (unlikely(sdev->sdev_state != SDEV_RUNNING)) {
		/* OK, we're not in a running state don't prep
		 * user commands */
		if (sdev->sdev_state == SDEV_DEL) {
			/* Device is fully deleted, no commands
			 * at all allowed down */
1283 1284
			sdev_printk(KERN_ERR, sdev,
				    "rejecting I/O to dead device\n");
1285
			goto kill;
Linus Torvalds's avatar
Linus Torvalds committed
1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301
		}
		/* OK, we only allow special commands (i.e. not
		 * user initiated ones */
		specials_only = sdev->sdev_state;
	}

	/*
	 * Find the actual device driver associated with this command.
	 * The SPECIAL requests are things like character device or
	 * ioctls, which did not originate from ll_rw_blk.  Note that
	 * the special field is also used to indicate the cmd for
	 * the remainder of a partially fulfilled request that can 
	 * come up when there is a medium error.  We have to treat
	 * these two cases differently.  We differentiate by looking
	 * at request->cmd, as this tells us the real story.
	 */
1302
	if (req->flags & REQ_SPECIAL && req->special) {
Linus Torvalds's avatar
Linus Torvalds committed
1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313
		struct scsi_request *sreq = req->special;

		if (sreq->sr_magic == SCSI_REQ_MAGIC) {
			cmd = scsi_get_command(sreq->sr_device, GFP_ATOMIC);
			if (unlikely(!cmd))
				goto defer;
			scsi_init_cmd_from_req(cmd, sreq);
		} else
			cmd = req->special;
	} else if (req->flags & (REQ_CMD | REQ_BLOCK_PC)) {

1314
		if(unlikely(specials_only) && !(req->flags & REQ_SPECIAL)) {
Linus Torvalds's avatar
Linus Torvalds committed
1315 1316
			if(specials_only == SDEV_QUIESCE ||
					specials_only == SDEV_BLOCK)
1317
				goto defer;
Linus Torvalds's avatar
Linus Torvalds committed
1318
			
1319 1320
			sdev_printk(KERN_ERR, sdev,
				    "rejecting I/O to device being removed\n");
1321
			goto kill;
Linus Torvalds's avatar
Linus Torvalds committed
1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338
		}
			
			
		/*
		 * Now try and find a command block that we can use.
		 */
		if (!req->special) {
			cmd = scsi_get_command(sdev, GFP_ATOMIC);
			if (unlikely(!cmd))
				goto defer;
		} else
			cmd = req->special;
		
		/* pull a tag out of the request if we have one */
		cmd->tag = req->tag;
	} else {
		blk_dump_rq_flags(req, "SCSI bad req");
1339
		goto kill;
Linus Torvalds's avatar
Linus Torvalds committed
1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375
	}
	
	/* note the overloading of req->special.  When the tag
	 * is active it always means cmd.  If the tag goes
	 * back for re-queueing, it may be reset */
	req->special = cmd;
	cmd->request = req;
	
	/*
	 * FIXME: drop the lock here because the functions below
	 * expect to be called without the queue lock held.  Also,
	 * previously, we dequeued the request before dropping the
	 * lock.  We hope REQ_STARTED prevents anything untoward from
	 * happening now.
	 */
	if (req->flags & (REQ_CMD | REQ_BLOCK_PC)) {
		int ret;

		/*
		 * This will do a couple of things:
		 *  1) Fill in the actual SCSI command.
		 *  2) Fill in any other upper-level specific fields
		 * (timeout).
		 *
		 * If this returns 0, it means that the request failed
		 * (reading past end of disk, reading offline device,
		 * etc).   This won't actually talk to the device, but
		 * some kinds of consistency checking may cause the	
		 * request to be rejected immediately.
		 */

		/* 
		 * This sets up the scatter-gather table (allocating if
		 * required).
		 */
		ret = scsi_init_io(cmd);
1376
		switch(ret) {
Alan Stern's avatar
Alan Stern committed
1377
			/* For BLKPREP_KILL/DEFER the cmd was released */
1378 1379 1380 1381 1382
		case BLKPREP_KILL:
			goto kill;
		case BLKPREP_DEFER:
			goto defer;
		}
Linus Torvalds's avatar
Linus Torvalds committed
1383 1384 1385 1386
		
		/*
		 * Initialize the actual SCSI command for this request.
		 */
1387 1388 1389 1390 1391
		if (req->flags & REQ_BLOCK_PC) {
			scsi_setup_blk_pc_cmnd(cmd);
		} else if (req->rq_disk) {
			struct scsi_driver *drv;

1392 1393 1394 1395
			drv = *(struct scsi_driver **)req->rq_disk->private_data;
			if (unlikely(!drv->init_command(cmd))) {
				scsi_release_buffers(cmd);
				scsi_put_command(cmd);
1396
				goto kill;
1397
			}
Linus Torvalds's avatar
Linus Torvalds committed
1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413
		}
	}

	/*
	 * The request is now prepped, no need to come back here
	 */
	req->flags |= REQ_DONTPREP;
	return BLKPREP_OK;

 defer:
	/* 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. */
	if (sdev->device_busy == 0)
		blk_plug_device(q);
	return BLKPREP_DEFER;
1414 1415 1416
 kill:
	req->errors = DID_NO_CONNECT << 16;
	return BLKPREP_KILL;
Linus Torvalds's avatar
Linus Torvalds committed
1417 1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435
}

/*
 * 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,
1436 1437
				   sdev_printk(KERN_INFO, sdev,
				   "unblocking device at zero depth\n"));
Linus Torvalds's avatar
Linus Torvalds committed
1438 1439 1440 1441 1442 1443 1444 1445 1446 1447 1448 1449 1450 1451 1452 1453 1454 1455 1456 1457 1458 1459
		} 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)
{
1460
	if (scsi_host_in_recovery(shost))
Linus Torvalds's avatar
Linus Torvalds committed
1461 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
		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;
}

/*
1490
 * Kill a request for a dead device
Linus Torvalds's avatar
Linus Torvalds committed
1491
 */
1492
static void scsi_kill_request(struct request *req, request_queue_t *q)
Linus Torvalds's avatar
Linus Torvalds committed
1493
{
1494
	struct scsi_cmnd *cmd = req->special;
1495 1496
	struct scsi_device *sdev = cmd->device;
	struct Scsi_Host *shost = sdev->host;
Linus Torvalds's avatar
Linus Torvalds committed
1497

1498 1499
	blkdev_dequeue_request(req);

1500 1501 1502 1503
	if (unlikely(cmd == NULL)) {
		printk(KERN_CRIT "impossible request in %s.\n",
				 __FUNCTION__);
		BUG();
Linus Torvalds's avatar
Linus Torvalds committed
1504
	}
1505 1506 1507 1508

	scsi_init_cmd_errh(cmd);
	cmd->result = DID_NO_CONNECT << 16;
	atomic_inc(&cmd->device->iorequest_cnt);
1509 1510 1511 1512 1513 1514 1515 1516 1517 1518 1519 1520 1521

	/*
	 * 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);

1522
	__scsi_done(cmd);
Linus Torvalds's avatar
Linus Torvalds committed
1523 1524
}

1525 1526 1527
static void scsi_softirq_done(struct request *rq)
{
	struct scsi_cmnd *cmd = rq->completion_data;
1528
	unsigned long wait_for = (cmd->allowed + 1) * cmd->timeout_per_command;
1529 1530 1531 1532 1533 1534 1535 1536 1537 1538 1539 1540 1541 1542 1543 1544 1545 1546 1547 1548 1549 1550 1551 1552 1553 1554 1555 1556 1557 1558 1559
	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:
			scsi_retry_command(cmd);
			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
1560 1561 1562 1563 1564 1565 1566 1567 1568 1569 1570 1571 1572 1573 1574 1575 1576 1577 1578 1579
/*
 * 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");
1580 1581
		while ((req = elv_next_request(q)) != NULL)
			scsi_kill_request(req, q);
Linus Torvalds's avatar
Linus Torvalds committed
1582 1583 1584 1585 1586 1587 1588 1589 1590 1591 1592 1593 1594