ide-io.c

来自「Linux Kernel 2.6.9 for OMAP1710」· C语言 代码 · 共 1,604 行 · 第 1/4 页

		goto kill_rq;

	block    = rq->sector;
	if (blk_fs_request(rq) &&
	    (drive->media == ide_disk || drive->media == ide_floppy)) {
		block += drive->sect0;
	}
	/* Yecch - this will shift the entire interval,
	   possibly killing some innocent following sector */
	if (block == 0 && drive->remap_0_to_1 == 1)
		block = 1;  /* redirect MBR access to EZ-Drive partn table */

	if (blk_pm_suspend_request(rq) &&
	    rq->pm->pm_step == ide_pm_state_start_suspend)
		/* Mark drive blocked when starting the suspend sequence. */
		drive->blocked = 1;
	else if (blk_pm_resume_request(rq) &&
		 rq->pm->pm_step == ide_pm_state_start_resume) {
		/* 
		 * The first thing we do on wakeup is to wait for BSY bit to
		 * go away (with a looong timeout) as a drive on this hwif may
		 * just be POSTing itself.
		 * We do that before even selecting as the "other" device on
		 * the bus may be broken enough to walk on our toes at this
		 * point.
		 */
		int rc;
#ifdef DEBUG_PM
		printk("%s: Wakeup request inited, waiting for !BSY...\n", drive->name);
#endif
		rc = ide_wait_not_busy(HWIF(drive), 35000);
		if (rc)
			printk(KERN_WARNING "%s: bus not ready on wakeup\n", drive->name);
		SELECT_DRIVE(drive);
		HWIF(drive)->OUTB(8, HWIF(drive)->io_ports[IDE_CONTROL_OFFSET]);
		rc = ide_wait_not_busy(HWIF(drive), 10000);
		if (rc)
			printk(KERN_WARNING "%s: drive not ready on wakeup\n", drive->name);
	}

	SELECT_DRIVE(drive);
	if (ide_wait_stat(&startstop, drive, drive->ready_stat, BUSY_STAT|DRQ_STAT, WAIT_READY)) {
		printk(KERN_ERR "%s: drive not ready for command\n", drive->name);
		return startstop;
	}
	if (!drive->special.all) {
		if (rq->flags & (REQ_DRIVE_CMD | REQ_DRIVE_TASK))
			return execute_drive_cmd(drive, rq);
		else if (rq->flags & REQ_DRIVE_TASKFILE)
			return execute_drive_cmd(drive, rq);
		else if (blk_pm_request(rq)) {
#ifdef DEBUG_PM
			printk("%s: start_power_step(step: %d)\n",
				drive->name, rq->pm->pm_step);
#endif
			startstop = DRIVER(drive)->start_power_step(drive, rq);
			if (startstop == ide_stopped &&
			    rq->pm->pm_step == ide_pm_state_completed)
				ide_complete_pm_request(drive, rq);
			return startstop;
		}
		return (DRIVER(drive)->do_request(drive, rq, block));
	}
	return do_special(drive);
kill_rq:
	DRIVER(drive)->end_request(drive, 0, 0);
	return ide_stopped;
}

EXPORT_SYMBOL(start_request);

/**
 *	ide_stall_queue		-	pause an IDE device
 *	@drive: drive to stall
 *	@timeout: time to stall for (jiffies)
 *
 *	ide_stall_queue() can be used by a drive to give excess bandwidth back
 *	to the hwgroup by sleeping for timeout jiffies.
 */
 
void ide_stall_queue (ide_drive_t *drive, unsigned long timeout)
{
	if (timeout > WAIT_WORSTCASE)
		timeout = WAIT_WORSTCASE;
	drive->sleep = timeout + jiffies;
}

EXPORT_SYMBOL(ide_stall_queue);

#define WAKEUP(drive)	((drive)->service_start + 2 * (drive)->service_time)

/**
 *	choose_drive		-	select a drive to service
 *	@hwgroup: hardware group to select on
 *
 *	choose_drive() selects the next drive which will be serviced.
 *	This is necessary because the IDE layer can't issue commands
 *	to both drives on the same cable, unlike SCSI.
 */
 
static inline ide_drive_t *choose_drive (ide_hwgroup_t *hwgroup)
{
	ide_drive_t *drive, *best;

repeat:	
	best = NULL;
	drive = hwgroup->drive;

	/*
	 * drive is doing pre-flush, ordered write, post-flush sequence. even
	 * though that is 3 requests, it must be seen as a single transaction.
	 * we must not preempt this drive until that is complete
	 */
	if (drive->doing_barrier) {
		/*
		 * small race where queue could get replugged during
		 * the 3-request flush cycle, just yank the plug since
		 * we want it to finish asap
		 */
		blk_remove_plug(drive->queue);
		return drive;
	}

	do {
		if ((!drive->sleep || time_after_eq(jiffies, drive->sleep))
		    && !elv_queue_empty(drive->queue)) {
			if (!best
			 || (drive->sleep && (!best->sleep || 0 < (signed long)(best->sleep - drive->sleep)))
			 || (!best->sleep && 0 < (signed long)(WAKEUP(best) - WAKEUP(drive))))
			{
				if (!blk_queue_plugged(drive->queue))
					best = drive;
			}
		}
	} while ((drive = drive->next) != hwgroup->drive);
	if (best && best->nice1 && !best->sleep && best != hwgroup->drive && best->service_time > WAIT_MIN_SLEEP) {
		long t = (signed long)(WAKEUP(best) - jiffies);
		if (t >= WAIT_MIN_SLEEP) {
		/*
		 * We *may* have some time to spare, but first let's see if
		 * someone can potentially benefit from our nice mood today..
		 */
			drive = best->next;
			do {
				if (!drive->sleep
				/* FIXME: use time_before */
				 && 0 < (signed long)(WAKEUP(drive) - (jiffies - best->service_time))
				 && 0 < (signed long)((jiffies + t) - WAKEUP(drive)))
				{
					ide_stall_queue(best, min_t(long, t, 10 * WAIT_MIN_SLEEP));
					goto repeat;
				}
			} while ((drive = drive->next) != best);
		}
	}
	return best;
}

/*
 * Issue a new request to a drive from hwgroup
 * Caller must have already done spin_lock_irqsave(&ide_lock, ..);
 *
 * A hwgroup is a serialized group of IDE interfaces.  Usually there is
 * exactly one hwif (interface) per hwgroup, but buggy controllers (eg. CMD640)
 * may have both interfaces in a single hwgroup to "serialize" access.
 * Or possibly multiple ISA interfaces can share a common IRQ by being grouped
 * together into one hwgroup for serialized access.
 *
 * Note also that several hwgroups can end up sharing a single IRQ,
 * possibly along with many other devices.  This is especially common in
 * PCI-based systems with off-board IDE controller cards.
 *
 * The IDE driver uses the single global ide_lock spinlock to protect
 * access to the request queues, and to protect the hwgroup->busy flag.
 *
 * The first thread into the driver for a particular hwgroup sets the
 * hwgroup->busy flag to indicate that this hwgroup is now active,
 * and then initiates processing of the top request from the request queue.
 *
 * Other threads attempting entry notice the busy setting, and will simply
 * queue their new requests and exit immediately.  Note that hwgroup->busy
 * remains set even when the driver is merely awaiting the next interrupt.
 * Thus, the meaning is "this hwgroup is busy processing a request".
 *
 * When processing of a request completes, the completing thread or IRQ-handler
 * will start the next request from the queue.  If no more work remains,
 * the driver will clear the hwgroup->busy flag and exit.
 *
 * The ide_lock (spinlock) is used to protect all access to the
 * hwgroup->busy flag, but is otherwise not needed for most processing in
 * the driver.  This makes the driver much more friendlier to shared IRQs
 * than previous designs, while remaining 100% (?) SMP safe and capable.
 */
/* --BenH: made non-static as ide-pmac.c uses it to kick the hwgroup back
 *         into life on wakeup from machine sleep.
 */ 
void ide_do_request (ide_hwgroup_t *hwgroup, int masked_irq)
{
	ide_drive_t	*drive;
	ide_hwif_t	*hwif;
	struct request	*rq;
	ide_startstop_t	startstop;

	/* for atari only: POSSIBLY BROKEN HERE(?) */
	ide_get_lock(ide_intr, hwgroup);

	/* caller must own ide_lock */
	BUG_ON(!irqs_disabled());

	while (!hwgroup->busy) {
		hwgroup->busy = 1;
		drive = choose_drive(hwgroup);
		if (drive == NULL) {
			unsigned long sleep = 0;
			hwgroup->rq = NULL;
			drive = hwgroup->drive;
			do {
				if (drive->sleep && (!sleep || 0 < (signed long)(sleep - drive->sleep)))
					sleep = drive->sleep;
			} while ((drive = drive->next) != hwgroup->drive);
			if (sleep) {
		/*
		 * Take a short snooze, and then wake up this hwgroup again.
		 * This gives other hwgroups on the same a chance to
		 * play fairly with us, just in case there are big differences
		 * in relative throughputs.. don't want to hog the cpu too much.
		 */
				if (time_before(sleep, jiffies + WAIT_MIN_SLEEP))
					sleep = jiffies + WAIT_MIN_SLEEP;
#if 1
				if (timer_pending(&hwgroup->timer))
					printk(KERN_CRIT "ide_set_handler: timer already active\n");
#endif
				/* so that ide_timer_expiry knows what to do */
				hwgroup->sleeping = 1;
				mod_timer(&hwgroup->timer, sleep);
				/* we purposely leave hwgroup->busy==1
				 * while sleeping */
			} else {
				/* Ugly, but how can we sleep for the lock
				 * otherwise? perhaps from tq_disk?
				 */

				/* for atari only */
				ide_release_lock();
				hwgroup->busy = 0;
			}

			/* no more work for this hwgroup (for now) */
			return;
		}
		hwif = HWIF(drive);
		if (hwgroup->hwif->sharing_irq &&
		    hwif != hwgroup->hwif &&
		    hwif->io_ports[IDE_CONTROL_OFFSET]) {
			/* set nIEN for previous hwif */
			SELECT_INTERRUPT(drive);
		}
		hwgroup->hwif = hwif;
		hwgroup->drive = drive;
		drive->sleep = 0;
		drive->service_start = jiffies;

		if (blk_queue_plugged(drive->queue)) {
			printk(KERN_ERR "ide: huh? queue was plugged!\n");
			break;
		}

		/*
		 * we know that the queue isn't empty, but this can happen
		 * if the q->prep_rq_fn() decides to kill a request
		 */
		rq = elv_next_request(drive->queue);
		if (!rq) {
			hwgroup->busy = 0;
			break;
		}

		/*
		 * if rq is a barrier write, issue pre cache flush if not
		 * already done
		 */
		if (blk_barrier_rq(rq) && !blk_barrier_preflush(rq))
			rq = ide_queue_flush_cmd(drive, rq, 0);

		/*
		 * Sanity: don't accept a request that isn't a PM request
		 * if we are currently power managed. This is very important as
		 * blk_stop_queue() doesn't prevent the elv_next_request()
		 * above to return us whatever is in the queue. Since we call
		 * ide_do_request() ourselves, we end up taking requests while
		 * the queue is blocked...
		 * 
		 * We let requests forced at head of queue with ide-preempt
		 * though. I hope that doesn't happen too much, hopefully not
		 * unless the subdriver triggers such a thing in its own PM
		 * state machine.
		 */
		if (drive->blocked && !blk_pm_request(rq) && !(rq->flags & REQ_PREEMPT)) {
			/* We clear busy, there should be no pending ATA command at this point. */
			hwgroup->busy = 0;
			break;
		}

		hwgroup->rq = rq;

		/*
		 * Some systems have trouble with IDE IRQs arriving while
		 * the driver is still setting things up.  So, here we disable
		 * the IRQ used by this interface while the request is being started.
		 * This may look bad at first, but pretty much the same thing
		 * happens anyway when any interrupt comes in, IDE or otherwise
		 *  -- the kernel masks the IRQ while it is being handled.
		 */
		if (hwif->irq != masked_irq)
			disable_irq_nosync(hwif->irq);
		spin_unlock(&ide_lock);
		local_irq_enable();
			/* allow other IRQs while we start this request */
		startstop = start_request(drive, rq);
		spin_lock_irq(&ide_lock);
		if (hwif->irq != masked_irq)
			enable_irq(hwif->irq);
		if (startstop == ide_stopped)
			hwgroup->busy = 0;
	}
}

EXPORT_SYMBOL(ide_do_request);

/*
 * Passes the stuff to ide_do_request
 */
void do_ide_request(request_queue_t *q)
{
	ide_drive_t *drive = q->queuedata;

	ide_do_request(HWGROUP(drive), IDE_NO_IRQ);
}

/*
 * un-busy the hwgroup etc, and clear any pending DMA status. we want to
 * retry the current request in pio mode instead of risking tossing it
 * all away
 */
static ide_startstop_t ide_dma_timeout_retry(ide_drive_t *drive, int error)
{
	ide_hwif_t *hwif = HWIF(drive);
	struct request *rq;
	ide_startstop_t ret = ide_stopped;

	/*
	 * end current dma transaction
	 */

	if (error < 0) {
		printk(KERN_WARNING "%s: DMA timeout error\n", drive->name);
		(void)HWIF(drive)->ide_dma_end(drive);
		ret = DRIVER(drive)->error(drive, "dma timeout error",
						hwif->INB(IDE_STATUS_REG));
	} else {
		printk(KERN_WARNING "%s: DMA timeout retry\n", drive->name);
		(void) hwif->ide_dma_timeout(drive);
	}

	/*
	 * disable dma for now, but remember that we did so because of
	 * a timeout -- we'll reenable after we finish this next request
	 * (or rather the first chunk of it) in pio.
	 */
	drive->retry_pio++;
	drive->state = DMA_PIO_RETRY;
	(void) hwif->ide_dma_off_quietly(drive);

	/*
	 * un-busy drive etc (hwgroup->busy is cleared on return) and
	 * make sure request is sane
	 */
	rq = HWGROUP(drive)->rq;
	HWGROUP(drive)->rq = NULL;

	rq->errors = 0;
	rq->sector = rq->bio->bi_sector;
	rq->current_nr_sectors = bio_iovec(rq->bio)->bv_len >> 9;
	rq->hard_cur_sectors = rq->current_nr_sectors;
	if (rq->bio)
		rq->buffer = NULL;

	return ret;
}

/**
 *	ide_timer_expiry	-	handle lack of an IDE interrupt
 *	@data: timer callback magic (hwgroup)
 *
 *	An IDE command has timed out before the expected drive return
 *	occurred. At this point we attempt to clean up the current
 *	mess. If the current handler includes an expiry handler then
 *	we invoke the expiry handler, and providing it is happy the
 *	work is done. If that fails we apply generic recovery rules
 *	invoking the handler and checking the drive DMA status. We