ide.c

讲述linux的初始化过程

}

/*
 * Issue a new request to a drive from hwgroup
 * Caller must have already done spin_lock_irqsave(&io_request_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 io_request_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 io_request_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.
 */
static void ide_do_request(ide_hwgroup_t *hwgroup, int masked_irq)
{
	ide_drive_t	*drive;
	ide_hwif_t	*hwif;
	ide_startstop_t	startstop;

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

	__cli();	/* necessary paranoia: ensure IRQs are masked on local CPU */

	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 (0 < (signed long)(jiffies + WAIT_MIN_SLEEP - sleep)) 
					sleep = jiffies + WAIT_MIN_SLEEP;
#if 1
				if (timer_pending(&hwgroup->timer))
					printk("ide_set_handler: timer already active\n");
#endif
				hwgroup->sleeping = 1;	/* so that ide_timer_expiry knows what to do */
				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? */
				ide_release_lock(&ide_lock);	/* for atari only */
				hwgroup->busy = 0;
			}
			return;		/* no more work for this hwgroup (for now) */
		}
		hwif = HWIF(drive);
		if (hwgroup->hwif->sharing_irq && hwif != hwgroup->hwif && hwif->io_ports[IDE_CONTROL_OFFSET]) {
			/* set nIEN for previous hwif */
			SELECT_INTERRUPT(hwif, drive);
		}
		hwgroup->hwif = hwif;
		hwgroup->drive = drive;
		drive->sleep = 0;
		drive->service_start = jiffies;

		if ( drive->queue.plugged )	/* paranoia */
			printk("%s: Huh? nuking plugged queue\n", drive->name);
		hwgroup->rq = blkdev_entry_next_request(&drive->queue.queue_head);
		/*
		 * 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 (masked_irq && hwif->irq != masked_irq)
			disable_irq_nosync(hwif->irq);
		spin_unlock(&io_request_lock);
		ide__sti();	/* allow other IRQs while we start this request */
		startstop = start_request(drive);
		spin_lock_irq(&io_request_lock);
		if (masked_irq && hwif->irq != masked_irq)
			enable_irq(hwif->irq);
		if (startstop == ide_stopped)
			hwgroup->busy = 0;
	}
}

/*
 * ide_get_queue() returns the queue which corresponds to a given device.
 */
request_queue_t *ide_get_queue (kdev_t dev)
{
	ide_hwif_t *hwif = (ide_hwif_t *)blk_dev[MAJOR(dev)].data;

	return &hwif->drives[DEVICE_NR(dev) & 1].queue;
}

/*
 * Passes the stuff to ide_do_request
 */
void do_ide_request(request_queue_t *q)
{
	ide_do_request(q->queuedata, 0);
}

/*
 * ide_timer_expiry() is our timeout function for all drive operations.
 * But note that it can also be invoked as a result of a "sleep" operation
 * triggered by the mod_timer() call in ide_do_request.
 */
void ide_timer_expiry (unsigned long data)
{
	ide_hwgroup_t	*hwgroup = (ide_hwgroup_t *) data;
	ide_handler_t	*handler;
	ide_expiry_t	*expiry;
 	unsigned long	flags;
	unsigned long	wait;

	spin_lock_irqsave(&io_request_lock, flags);
	del_timer(&hwgroup->timer);

	if ((handler = hwgroup->handler) == NULL) {
		/*
		 * Either a marginal timeout occured
		 * (got the interrupt just as timer expired),
		 * or we were "sleeping" to give other devices a chance.
		 * Either way, we don't really want to complain about anything.
		 */
		if (hwgroup->sleeping) {
			hwgroup->sleeping = 0;
			hwgroup->busy = 0;
		}
	} else {
		ide_drive_t *drive = hwgroup->drive;
		if (!drive) {
			printk("ide_timer_expiry: hwgroup->drive was NULL\n");
			hwgroup->handler = NULL;
		} else {
			ide_hwif_t *hwif;
			ide_startstop_t startstop;
			if (!hwgroup->busy) {
				hwgroup->busy = 1;	/* paranoia */
				printk("%s: ide_timer_expiry: hwgroup->busy was 0 ??\n", drive->name);
			}
			if ((expiry = hwgroup->expiry) != NULL) {
				/* continue */
				if ((wait = expiry(drive)) != 0) {
					/* reset timer */
					hwgroup->timer.expires  = jiffies + wait;
					add_timer(&hwgroup->timer);
					spin_unlock_irqrestore(&io_request_lock, flags);
					return;
				}
			}
			hwgroup->handler = NULL;
			/*
			 * We need to simulate a real interrupt when invoking
			 * the handler() function, which means we need to globally
			 * mask the specific IRQ:
			 */
			spin_unlock(&io_request_lock);
			hwif  = HWIF(drive);
#if DISABLE_IRQ_NOSYNC
			disable_irq_nosync(hwif->irq);
#else
			disable_irq(hwif->irq);	/* disable_irq_nosync ?? */
#endif /* DISABLE_IRQ_NOSYNC */
			__cli();	/* local CPU only, as if we were handling an interrupt */
			if (hwgroup->poll_timeout != 0) {
				startstop = handler(drive);
			} else if (drive_is_ready(drive)) {
				if (drive->waiting_for_dma)
					(void) hwgroup->hwif->dmaproc(ide_dma_lostirq, drive);
				(void)ide_ack_intr(hwif);
				printk("%s: lost interrupt\n", drive->name);
				startstop = handler(drive);
			} else {
				if (drive->waiting_for_dma) {
					(void) hwgroup->hwif->dmaproc(ide_dma_end, drive);
					printk("%s: timeout waiting for DMA\n", drive->name);
					(void) hwgroup->hwif->dmaproc(ide_dma_timeout, drive);
				}
				startstop = ide_error(drive, "irq timeout", GET_STAT());
			}
			set_recovery_timer(hwif);
			drive->service_time = jiffies - drive->service_start;
			enable_irq(hwif->irq);
			spin_lock_irq(&io_request_lock);
			if (startstop == ide_stopped)
				hwgroup->busy = 0;
		}
	}
	ide_do_request(hwgroup, 0);
	spin_unlock_irqrestore(&io_request_lock, flags);
}

/*
 * There's nothing really useful we can do with an unexpected interrupt,
 * other than reading the status register (to clear it), and logging it.
 * There should be no way that an irq can happen before we're ready for it,
 * so we needn't worry much about losing an "important" interrupt here.
 *
 * On laptops (and "green" PCs), an unexpected interrupt occurs whenever the
 * drive enters "idle", "standby", or "sleep" mode, so if the status looks
 * "good", we just ignore the interrupt completely.
 *
 * This routine assumes __cli() is in effect when called.
 *
 * If an unexpected interrupt happens on irq15 while we are handling irq14
 * and if the two interfaces are "serialized" (CMD640), then it looks like
 * we could screw up by interfering with a new request being set up for irq15.
 *
 * In reality, this is a non-issue.  The new command is not sent unless the
 * drive is ready to accept one, in which case we know the drive is not
 * trying to interrupt us.  And ide_set_handler() is always invoked before
 * completing the issuance of any new drive command, so we will not be
 * accidently invoked as a result of any valid command completion interrupt.
 *
 */
static void unexpected_intr (int irq, ide_hwgroup_t *hwgroup)
{
	byte stat;
	ide_hwif_t *hwif = hwgroup->hwif;

	/*
	 * handle the unexpected interrupt
	 */
	do {
		if (hwif->irq == irq) {
			stat = IN_BYTE(hwif->io_ports[IDE_STATUS_OFFSET]);
			if (!OK_STAT(stat, READY_STAT, BAD_STAT)) {
				/* Try to not flood the console with msgs */
				static unsigned long last_msgtime, count;
				++count;
				if (0 < (signed long)(jiffies - (last_msgtime + HZ))) {
					last_msgtime = jiffies;
					printk("%s%s: unexpected interrupt, status=0x%02x, count=%ld\n",
					 hwif->name, (hwif->next == hwgroup->hwif) ? "" : "(?)", stat, count);
				}
			}
		}
	} while ((hwif = hwif->next) != hwgroup->hwif);
}

/*
 * entry point for all interrupts, caller does __cli() for us
 */
void ide_intr (int irq, void *dev_id, struct pt_regs *regs)
{
	unsigned long flags;
	ide_hwgroup_t *hwgroup = (ide_hwgroup_t *)dev_id;
	ide_hwif_t *hwif;
	ide_drive_t *drive;
	ide_handler_t *handler;
	ide_startstop_t startstop;

	spin_lock_irqsave(&io_request_lock, flags);
	hwif = hwgroup->hwif;

	if (!ide_ack_intr(hwif)) {
		spin_unlock_irqrestore(&io_request_lock, flags);
		return;
	}

	if ((handler = hwgroup->handler) == NULL || hwgroup->poll_timeout != 0) {
		/*
		 * Not expecting an interrupt from this drive.
		 * That means this could be:
		 *	(1) an interrupt from another PCI device
		 *	sharing the same PCI INT# as us.
		 * or	(2) a drive just entered sleep or standby mode,
		 *	and is interrupting to let us know.
		 * or	(3) a spurious interrupt of unknown origin.
		 *
		 * For PCI, we cannot tell the difference,
		 * so in that case we just ignore it and hope it goes away.
		 */
#ifdef CONFIG_BLK_DEV_IDEPCI
		if (IDE_PCI_DEVID_EQ(hwif->pci_devid, IDE_PCI_DEVID_NULL))
#endif	/* CONFIG_BLK_DEV_IDEPCI */
		{
			/*
			 * Probably not a shared PCI interrupt,
			 * so we can safely try to do something about it:
			 */
			unexpected_intr(irq, hwgroup);
#ifdef CONFIG_BLK_DEV_IDEPCI
		} else {
			/*
			 * Whack the status register, just in case we have a leftover pending IRQ.
			 */
			(void) IN_BYTE(hwif->io_ports[IDE_STATUS_OFFSET]);
#endif /* CONFIG_BLK_DEV_IDEPCI */
		}
		spin_unlock_irqrestore(&io_request_lock, flags);
		return;
	}
	drive = hwgroup->drive;
	if (!drive) {
		/*
		 * This should NEVER happen, and there isn't much we could do about it here.
		 */
		spin_unlock_irqrestore(&io_request_lock, flags);
		return;
	}
	if (!drive_is_ready(drive)) {
		/*
		 * This happens regularly when we share a PCI IRQ with another device.
		 * Unfortunately, it can also happen with some buggy drives that trigger
		 * the IRQ before their status register is up to date.  Hopefully we have
		 * enough advance overhead that the latter isn't a problem.
		 */
		spin_unlock_irqrestore(&io_request_lock, flags);
		return;
	}
	if (!hwgroup->busy) {
		hwgroup->busy = 1;	/* paranoia */
		printk("%s: ide_intr: hwgroup->busy was 0 ??\n", drive->name);
	}
	hwgroup->handler = NULL;
	del_timer(&hwgroup->timer);
	spin_unlock(&io_request_lock);

	if (drive->unmask)
		ide__sti();	/* local CPU only */
	startstop = handler(drive);		/* service this interrupt, may set handler for next interrupt */
	spin_lock_irq(&io_request_lock);

	/*
	 * Note that handler() may have set things up for another
	 * interrupt to occur soon, but it cannot happen until
	 * we exit from this routine, because it will be the
	 * same irq as is currently being serviced here, and Linux
	 * won't allow another of the same (on any CPU) until we return.
	 */
	set_recovery_timer(HWIF(drive));
	drive->service_time = jiffies - drive->service_start;
	if (startstop == ide_stopped) {
		if (hwgroup->handler == NULL) {	/* paranoia */
			hwgroup->busy = 0;
			ide_do_request(hwgroup, hwif->irq);
		} else {
			printk("%s: ide_intr: huh? expected NULL handler on exit\n", drive->name);
		}
	}
	spin_unlock_irqrestore(&io_request_lock, flags);
}

/*
 * get_info_ptr() returns the (ide_drive_t *) for a given device number.
 * It returns NULL if the given device number does not match any present drives.
 */
ide_drive_t *get_info_ptr (kdev_t i_rdev)
{
	int		major = MAJOR(i_rdev);
#if 0
	int		minor = MINOR(i_rdev) & PARTN_MASK;
#endif
	unsigned int	h;

	for (h = 0; h < MAX_HWIFS; ++h) {
		ide_hwif_t  *hwif = &ide_hwifs[h];
		if (hwif->present && major == hwif->major) {
			unsigned unit = DEVICE_NR(i_rdev);
			if (unit < MAX_DRIVES) {
				ide_drive_t *drive = &hwif->drives[unit];
#if 0
				if ((drive->present) && (drive->part[minor].nr_sects))
#else
				if (drive->present)
#endif
					return drive;
			}
			break;
		}
	}
	return NULL;
}

/*
 * This function is intended to be used prior to invoking ide_do_drive_cmd().
 */
void ide_init_drive_cmd (struct request *rq)
{
	memset(rq, 0, sizeof(*rq));
	rq->cmd = IDE_DRIVE_CMD;
}

/*