drivers/s390/cio/css.c - kernel/msm - Gitiles

 /*
  *  drivers/s390/cio/css.c
  *  driver for channel subsystem
  *   $Revision: 1.85 $
  *
  *    Copyright (C) 2002 IBM Deutschland Entwicklung GmbH,
  *			 IBM Corporation
  *    Author(s): Arnd Bergmann (arndb@de.ibm.com)
  *		 Cornelia Huck (cohuck@de.ibm.com)
  */
 #include <linux/module.h>
 #include <linux/init.h>
 #include <linux/device.h>
 #include <linux/slab.h>
 #include <linux/errno.h>
 #include <linux/list.h>

 #include "css.h"
 #include "cio.h"
 #include "cio_debug.h"
 #include "ioasm.h"
 #include "chsc.h"

 unsigned int highest_subchannel;
 int need_rescan = 0;
 int css_init_done = 0;

 struct pgid global_pgid;
 int css_characteristics_avail = 0;

 struct device css_bus_device = {
 	.bus_id = "css0",
 };

 static struct subchannel *
 css_alloc_subchannel(int irq)
 {
 	struct subchannel *sch;
 	int ret;

 	sch = kmalloc (sizeof (*sch), GFP_KERNEL | GFP_DMA);
 	if (sch == NULL)
 		return ERR_PTR(-ENOMEM);
 	ret = cio_validate_subchannel (sch, irq);
 	if (ret < 0) {
 		kfree(sch);
 		return ERR_PTR(ret);
 	}
 	if (irq > highest_subchannel)
 		highest_subchannel = irq;

 	if (sch->st != SUBCHANNEL_TYPE_IO) {
 		/* For now we ignore all non-io subchannels. */
 		kfree(sch);
 		return ERR_PTR(-EINVAL);
 	}

 	/*
 	 * Set intparm to subchannel address.
 	 * This is fine even on 64bit since the subchannel is always located
 	 * under 2G.
 	 */
 	sch->schib.pmcw.intparm = (__u32)(unsigned long)sch;
 	ret = cio_modify(sch);
 	if (ret) {
 		kfree(sch);
 		return ERR_PTR(ret);
 	}
 	return sch;
 }

 static void
 css_free_subchannel(struct subchannel *sch)
 {
 	if (sch) {
 		/* Reset intparm to zeroes. */
 		sch->schib.pmcw.intparm = 0;
 		cio_modify(sch);
 		kfree(sch);
 	}

 }

 static void
 css_subchannel_release(struct device *dev)
 {
 	struct subchannel *sch;

 	sch = to_subchannel(dev);
 	if (!cio_is_console(sch->irq))
 		kfree(sch);
 }

 extern int css_get_ssd_info(struct subchannel *sch);

 static int
 css_register_subchannel(struct subchannel *sch)
 {
 	int ret;

 	/* Initialize the subchannel structure */
 	sch->dev.parent = &css_bus_device;
 	sch->dev.bus = &css_bus_type;
 	sch->dev.release = &css_subchannel_release;

 	/* make it known to the system */
 	ret = device_register(&sch->dev);
 	if (ret)
 		printk (KERN_WARNING "%s: could not register %s\n",
 			__func__, sch->dev.bus_id);
 	else
 		css_get_ssd_info(sch);
 	return ret;
 }

 int
 css_probe_device(int irq)
 {
 	int ret;
 	struct subchannel *sch;

 	sch = css_alloc_subchannel(irq);
 	if (IS_ERR(sch))
 		return PTR_ERR(sch);
 	ret = css_register_subchannel(sch);
 	if (ret)
 		css_free_subchannel(sch);
 	return ret;
 }

 struct subchannel *
 get_subchannel_by_schid(int irq)
 {
 	struct subchannel *sch;
 	struct list_head *entry;
 	struct device *dev;

 	if (!get_bus(&css_bus_type))
 		return NULL;
 	down_read(&css_bus_type.subsys.rwsem);
 	sch = NULL;
 	list_for_each(entry, &css_bus_type.devices.list) {
 		dev = get_device(container_of(entry,
 					      struct device, bus_list));
 		if (!dev)
 			continue;
 		sch = to_subchannel(dev);
 		if (sch->irq == irq)
 			break;
 		put_device(dev);
 		sch = NULL;
 	}
 	up_read(&css_bus_type.subsys.rwsem);
 	put_bus(&css_bus_type);

 	return sch;
 }

 static inline int
 css_get_subchannel_status(struct subchannel *sch, int schid)
 {
 	struct schib schib;
 	int cc;

 	cc = stsch(schid, &schib);
 	if (cc)
 		return CIO_GONE;
 	if (!schib.pmcw.dnv)
 		return CIO_GONE;
 	if (sch && sch->schib.pmcw.dnv &&
 	    (schib.pmcw.dev != sch->schib.pmcw.dev))
 		return CIO_REVALIDATE;
 	if (sch && !sch->lpm)
 		return CIO_NO_PATH;
 	return CIO_OPER;
 }

 static int
 css_evaluate_subchannel(int irq, int slow)
 {
 	int event, ret, disc;
 	struct subchannel *sch;
 	unsigned long flags;

 	sch = get_subchannel_by_schid(irq);
 	disc = sch ? device_is_disconnected(sch) : 0;
 	if (disc && slow) {
 		if (sch)
 			put_device(&sch->dev);
 		return 0; /* Already processed. */
 	}
 	/*
 	 * We've got a machine check, so running I/O won't get an interrupt.
 	 * Kill any pending timers.
 	 */
 	if (sch)
 		device_kill_pending_timer(sch);
 	if (!disc && !slow) {
 		if (sch)
 			put_device(&sch->dev);
 		return -EAGAIN; /* Will be done on the slow path. */
 	}
 	event = css_get_subchannel_status(sch, irq);
 	CIO_MSG_EVENT(4, "Evaluating schid %04x, event %d, %s, %s path.\n",
 		      irq, event, sch?(disc?"disconnected":"normal"):"unknown",
 		      slow?"slow":"fast");
 	switch (event) {
 	case CIO_NO_PATH:
 	case CIO_GONE:
 		if (!sch) {
 			/* Never used this subchannel. Ignore. */
 			ret = 0;
 			break;
 		}
 		if (disc && (event == CIO_NO_PATH)) {
 			/*
 			 * Uargh, hack again. Because we don't get a machine
 			 * check on configure on, our path bookkeeping can
 			 * be out of date here (it's fine while we only do
 			 * logical varying or get chsc machine checks). We
 			 * need to force reprobing or we might miss devices
 			 * coming operational again. It won't do harm in real
 			 * no path situations.
 			 */
 			spin_lock_irqsave(&sch->lock, flags);
 			device_trigger_reprobe(sch);
 			spin_unlock_irqrestore(&sch->lock, flags);
 			ret = 0;
 			break;
 		}
 		if (sch->driver && sch->driver->notify &&
 		    sch->driver->notify(&sch->dev, event)) {
 			cio_disable_subchannel(sch);
 			device_set_disconnected(sch);
 			ret = 0;
 			break;
 		}
 		/*
 		 * Unregister subchannel.
 		 * The device will be killed automatically.
 		 */
 		cio_disable_subchannel(sch);
 		device_unregister(&sch->dev);
 		/* Reset intparm to zeroes. */
 		sch->schib.pmcw.intparm = 0;
 		cio_modify(sch);
 		put_device(&sch->dev);
 		ret = 0;
 		break;
 	case CIO_REVALIDATE:
 		/*
 		 * Revalidation machine check. Sick.
 		 * We don't notify the driver since we have to throw the device
 		 * away in any case.
 		 */
 		if (!disc) {
 			device_unregister(&sch->dev);
 			/* Reset intparm to zeroes. */
 			sch->schib.pmcw.intparm = 0;
 			cio_modify(sch);
 			put_device(&sch->dev);
 			ret = css_probe_device(irq);
 		} else {
 			/*
 			 * We can't immediately deregister the disconnected
 			 * device since it might block.
 			 */
 			spin_lock_irqsave(&sch->lock, flags);
 			device_trigger_reprobe(sch);
 			spin_unlock_irqrestore(&sch->lock, flags);
 			ret = 0;
 		}
 		break;
 	case CIO_OPER:
 		if (disc) {
 			spin_lock_irqsave(&sch->lock, flags);
 			/* Get device operational again. */
 			device_trigger_reprobe(sch);
 			spin_unlock_irqrestore(&sch->lock, flags);
 		}
 		ret = sch ? 0 : css_probe_device(irq);
 		break;
 	default:
 		BUG();
 		ret = 0;
 	}
 	return ret;
 }

 static void
 css_rescan_devices(void)
 {
 	int irq, ret;

 	for (irq = 0; irq < __MAX_SUBCHANNELS; irq++) {
 		ret = css_evaluate_subchannel(irq, 1);
 		/* No more memory. It doesn't make sense to continue. No
 		 * panic because this can happen in midflight and just
 		 * because we can't use a new device is no reason to crash
 		 * the system. */
 		if (ret == -ENOMEM)
 			break;
 		/* -ENXIO indicates that there are no more subchannels. */
 		if (ret == -ENXIO)
 			break;
 	}
 }

 struct slow_subchannel {
 	struct list_head slow_list;
 	unsigned long schid;
 };

 static LIST_HEAD(slow_subchannels_head);
 static DEFINE_SPINLOCK(slow_subchannel_lock);

 static void
 css_trigger_slow_path(void)
 {
 	CIO_TRACE_EVENT(4, "slowpath");

 	if (need_rescan) {
 		need_rescan = 0;
 		css_rescan_devices();
 		return;
 	}

 	spin_lock_irq(&slow_subchannel_lock);
 	while (!list_empty(&slow_subchannels_head)) {
 		struct slow_subchannel *slow_sch =
 			list_entry(slow_subchannels_head.next,
 				   struct slow_subchannel, slow_list);

 		list_del_init(slow_subchannels_head.next);
 		spin_unlock_irq(&slow_subchannel_lock);
 		css_evaluate_subchannel(slow_sch->schid, 1);
 		spin_lock_irq(&slow_subchannel_lock);
 		kfree(slow_sch);
 	}
 	spin_unlock_irq(&slow_subchannel_lock);
 }

 typedef void (*workfunc)(void *);
 DECLARE_WORK(slow_path_work, (workfunc)css_trigger_slow_path, NULL);
 struct workqueue_struct *slow_path_wq;

 /*
  * Rescan for new devices. FIXME: This is slow.
  * This function is called when we have lost CRWs due to overflows and we have
  * to do subchannel housekeeping.
  */
 void
 css_reiterate_subchannels(void)
 {
 	css_clear_subchannel_slow_list();
 	need_rescan = 1;
 }

 /*
  * Called from the machine check handler for subchannel report words.
  */
 int
 css_process_crw(int irq)
 {
 	int ret;

 	CIO_CRW_EVENT(2, "source is subchannel %04X\n", irq);

 	if (need_rescan)
 		/* We need to iterate all subchannels anyway. */
 		return -EAGAIN;
 	/*
 	 * Since we are always presented with IPI in the CRW, we have to
 	 * use stsch() to find out if the subchannel in question has come
 	 * or gone.
 	 */
 	ret = css_evaluate_subchannel(irq, 0);
 	if (ret == -EAGAIN) {
 		if (css_enqueue_subchannel_slow(irq)) {
 			css_clear_subchannel_slow_list();
 			need_rescan = 1;
 		}
 	}
 	return ret;
 }

 static void __init
 css_generate_pgid(void)
 {
 	/* Let's build our path group ID here. */
 	if (css_characteristics_avail && css_general_characteristics.mcss)
 		global_pgid.cpu_addr = 0x8000;
 	else {
 #ifdef CONFIG_SMP
 		global_pgid.cpu_addr = hard_smp_processor_id();
 #else
 		global_pgid.cpu_addr = 0;
 #endif
 	}
 	global_pgid.cpu_id = ((cpuid_t *) __LC_CPUID)->ident;
 	global_pgid.cpu_model = ((cpuid_t *) __LC_CPUID)->machine;
 	global_pgid.tod_high = (__u32) (get_clock() >> 32);
 }

 /*
  * Now that the driver core is running, we can setup our channel subsystem.
  * The struct subchannel's are created during probing (except for the
  * static console subchannel).
  */
 static int __init
 init_channel_subsystem (void)
 {
 	int ret, irq;

 	if (chsc_determine_css_characteristics() == 0)
 		css_characteristics_avail = 1;

 	css_generate_pgid();

 	if ((ret = bus_register(&css_bus_type)))
 		goto out;
 	if ((ret = device_register (&css_bus_device)))
 		goto out_bus;

 	css_init_done = 1;

 	ctl_set_bit(6, 28);

 	for (irq = 0; irq < __MAX_SUBCHANNELS; irq++) {
 		struct subchannel *sch;

 		if (cio_is_console(irq))
 			sch = cio_get_console_subchannel();
 		else {
 			sch = css_alloc_subchannel(irq);
 			if (IS_ERR(sch))
 				ret = PTR_ERR(sch);
 			else
 				ret = 0;
 			if (ret == -ENOMEM)
 				panic("Out of memory in "
 				      "init_channel_subsystem\n");
 			/* -ENXIO: no more subchannels. */
 			if (ret == -ENXIO)
 				break;
 			if (ret)
 				continue;
 		}
 		/*
 		 * We register ALL valid subchannels in ioinfo, even those
 		 * that have been present before init_channel_subsystem.
 		 * These subchannels can't have been registered yet (kmalloc
 		 * not working) so we do it now. This is true e.g. for the
 		 * console subchannel.
 		 */
 		css_register_subchannel(sch);
 	}
 	return 0;

 out_bus:
 	bus_unregister(&css_bus_type);
 out:
 	return ret;
 }

 /*
  * find a driver for a subchannel. They identify by the subchannel
  * type with the exception that the console subchannel driver has its own
  * subchannel type although the device is an i/o subchannel
  */
 static int
 css_bus_match (struct device *dev, struct device_driver *drv)
 {
 	struct subchannel *sch = container_of (dev, struct subchannel, dev);
 	struct css_driver *driver = container_of (drv, struct css_driver, drv);

 	if (sch->st == driver->subchannel_type)
 		return 1;

 	return 0;
 }

 struct bus_type css_bus_type = {
 	.name  = "css",
 	.match = &css_bus_match,
 };

 subsys_initcall(init_channel_subsystem);

 /*
  * Register root devices for some drivers. The release function must not be
  * in the device drivers, so we do it here.
  */
 static void
 s390_root_dev_release(struct device *dev)
 {
 	kfree(dev);
 }

 struct device *
 s390_root_dev_register(const char *name)
 {
 	struct device *dev;
 	int ret;

 	if (!strlen(name))
 		return ERR_PTR(-EINVAL);
 	dev = kmalloc(sizeof(struct device), GFP_KERNEL);
 	if (!dev)
 		return ERR_PTR(-ENOMEM);
 	memset(dev, 0, sizeof(struct device));
 	strncpy(dev->bus_id, name, min(strlen(name), (size_t)BUS_ID_SIZE));
 	dev->release = s390_root_dev_release;
 	ret = device_register(dev);
 	if (ret) {
 		kfree(dev);
 		return ERR_PTR(ret);
 	}
 	return dev;
 }

 void
 s390_root_dev_unregister(struct device *dev)
 {
 	if (dev)
 		device_unregister(dev);
 }

 int
 css_enqueue_subchannel_slow(unsigned long schid)
 {
 	struct slow_subchannel *new_slow_sch;
 	unsigned long flags;

 	new_slow_sch = kmalloc(sizeof(struct slow_subchannel), GFP_ATOMIC);
 	if (!new_slow_sch)
 		return -ENOMEM;
 	memset(new_slow_sch, 0, sizeof(struct slow_subchannel));
 	new_slow_sch->schid = schid;
 	spin_lock_irqsave(&slow_subchannel_lock, flags);
 	list_add_tail(&new_slow_sch->slow_list, &slow_subchannels_head);
 	spin_unlock_irqrestore(&slow_subchannel_lock, flags);
 	return 0;
 }

 void
 css_clear_subchannel_slow_list(void)
 {
 	unsigned long flags;

 	spin_lock_irqsave(&slow_subchannel_lock, flags);
 	while (!list_empty(&slow_subchannels_head)) {
 		struct slow_subchannel *slow_sch =
 			list_entry(slow_subchannels_head.next,
 				   struct slow_subchannel, slow_list);

 		list_del_init(slow_subchannels_head.next);
 		kfree(slow_sch);
 	}
 	spin_unlock_irqrestore(&slow_subchannel_lock, flags);
 }


 int
 css_slow_subchannels_exist(void)
 {
 	return (!list_empty(&slow_subchannels_head));
 }

 MODULE_LICENSE("GPL");
 EXPORT_SYMBOL(css_bus_type);
 EXPORT_SYMBOL(s390_root_dev_register);
 EXPORT_SYMBOL(s390_root_dev_unregister);
 EXPORT_SYMBOL_GPL(css_characteristics_avail);
	/*
	* drivers/s390/cio/css.c
	* driver for channel subsystem
	* $Revision: 1.85 $
	*
	* Copyright (C) 2002 IBM Deutschland Entwicklung GmbH,
	* IBM Corporation
	* Author(s): Arnd Bergmann (arndb@de.ibm.com)
	* Cornelia Huck (cohuck@de.ibm.com)
	*/
	#include <linux/module.h>
	#include <linux/init.h>
	#include <linux/device.h>
	#include <linux/slab.h>
	#include <linux/errno.h>
	#include <linux/list.h>

	#include "css.h"
	#include "cio.h"
	#include "cio_debug.h"
	#include "ioasm.h"
	#include "chsc.h"

	unsigned int highest_subchannel;
	int need_rescan = 0;
	int css_init_done = 0;

	struct pgid global_pgid;
	int css_characteristics_avail = 0;

	struct device css_bus_device = {
	.bus_id = "css0",
	};

	static struct subchannel *
	css_alloc_subchannel(int irq)
	{
	struct subchannel *sch;
	int ret;

	sch = kmalloc (sizeof (*sch), GFP_KERNEL \| GFP_DMA);
	if (sch == NULL)
	return ERR_PTR(-ENOMEM);
	ret = cio_validate_subchannel (sch, irq);
	if (ret < 0) {
	kfree(sch);
	return ERR_PTR(ret);
	}
	if (irq > highest_subchannel)
	highest_subchannel = irq;

	if (sch->st != SUBCHANNEL_TYPE_IO) {
	/* For now we ignore all non-io subchannels. */
	kfree(sch);
	return ERR_PTR(-EINVAL);
	}

	/*
	* Set intparm to subchannel address.
	* This is fine even on 64bit since the subchannel is always located
	* under 2G.
	*/
	sch->schib.pmcw.intparm = (__u32)(unsigned long)sch;
	ret = cio_modify(sch);
	if (ret) {
	kfree(sch);
	return ERR_PTR(ret);
	}
	return sch;
	}

	static void
	css_free_subchannel(struct subchannel *sch)
	{
	if (sch) {
	/* Reset intparm to zeroes. */
	sch->schib.pmcw.intparm = 0;
	cio_modify(sch);
	kfree(sch);
	}

	}

	static void
	css_subchannel_release(struct device *dev)
	{
	struct subchannel *sch;

	sch = to_subchannel(dev);
	if (!cio_is_console(sch->irq))
	kfree(sch);
	}

	extern int css_get_ssd_info(struct subchannel *sch);

	static int
	css_register_subchannel(struct subchannel *sch)
	{
	int ret;

	/* Initialize the subchannel structure */
	sch->dev.parent = &css_bus_device;
	sch->dev.bus = &css_bus_type;
	sch->dev.release = &css_subchannel_release;

	/* make it known to the system */
	ret = device_register(&sch->dev);
	if (ret)
	printk (KERN_WARNING "%s: could not register %s\n",
	__func__, sch->dev.bus_id);
	else
	css_get_ssd_info(sch);
	return ret;
	}

	int
	css_probe_device(int irq)
	{
	int ret;
	struct subchannel *sch;

	sch = css_alloc_subchannel(irq);
	if (IS_ERR(sch))
	return PTR_ERR(sch);
	ret = css_register_subchannel(sch);
	if (ret)
	css_free_subchannel(sch);
	return ret;
	}

	struct subchannel *
	get_subchannel_by_schid(int irq)
	{
	struct subchannel *sch;
	struct list_head *entry;
	struct device *dev;

	if (!get_bus(&css_bus_type))
	return NULL;
	down_read(&css_bus_type.subsys.rwsem);
	sch = NULL;
	list_for_each(entry, &css_bus_type.devices.list) {
	dev = get_device(container_of(entry,
	struct device, bus_list));
	if (!dev)
	continue;
	sch = to_subchannel(dev);
	if (sch->irq == irq)
	break;
	put_device(dev);
	sch = NULL;
	}
	up_read(&css_bus_type.subsys.rwsem);
	put_bus(&css_bus_type);

	return sch;
	}

	static inline int
	css_get_subchannel_status(struct subchannel *sch, int schid)
	{
	struct schib schib;
	int cc;

	cc = stsch(schid, &schib);
	if (cc)
	return CIO_GONE;
	if (!schib.pmcw.dnv)
	return CIO_GONE;
	if (sch && sch->schib.pmcw.dnv &&
	(schib.pmcw.dev != sch->schib.pmcw.dev))
	return CIO_REVALIDATE;
	if (sch && !sch->lpm)
	return CIO_NO_PATH;
	return CIO_OPER;
	}

	static int
	css_evaluate_subchannel(int irq, int slow)
	{
	int event, ret, disc;
	struct subchannel *sch;
	unsigned long flags;

	sch = get_subchannel_by_schid(irq);
	disc = sch ? device_is_disconnected(sch) : 0;
	if (disc && slow) {
	if (sch)
	put_device(&sch->dev);
	return 0; /* Already processed. */
	}
	/*
	* We've got a machine check, so running I/O won't get an interrupt.
	* Kill any pending timers.
	*/
	if (sch)
	device_kill_pending_timer(sch);
	if (!disc && !slow) {
	if (sch)
	put_device(&sch->dev);
	return -EAGAIN; /* Will be done on the slow path. */
	}
	event = css_get_subchannel_status(sch, irq);
	CIO_MSG_EVENT(4, "Evaluating schid %04x, event %d, %s, %s path.\n",
	irq, event, sch?(disc?"disconnected":"normal"):"unknown",
	slow?"slow":"fast");
	switch (event) {
	case CIO_NO_PATH:
	case CIO_GONE:
	if (!sch) {
	/* Never used this subchannel. Ignore. */
	ret = 0;
	break;
	}
	if (disc && (event == CIO_NO_PATH)) {
	/*
	* Uargh, hack again. Because we don't get a machine
	* check on configure on, our path bookkeeping can
	* be out of date here (it's fine while we only do
	* logical varying or get chsc machine checks). We
	* need to force reprobing or we might miss devices
	* coming operational again. It won't do harm in real
	* no path situations.
	*/
	spin_lock_irqsave(&sch->lock, flags);
	device_trigger_reprobe(sch);
	spin_unlock_irqrestore(&sch->lock, flags);
	ret = 0;
	break;
	}
	if (sch->driver && sch->driver->notify &&
	sch->driver->notify(&sch->dev, event)) {
	cio_disable_subchannel(sch);
	device_set_disconnected(sch);
	ret = 0;
	break;
	}
	/*
	* Unregister subchannel.
	* The device will be killed automatically.
	*/
	cio_disable_subchannel(sch);
	device_unregister(&sch->dev);
	/* Reset intparm to zeroes. */
	sch->schib.pmcw.intparm = 0;
	cio_modify(sch);
	put_device(&sch->dev);
	ret = 0;
	break;
	case CIO_REVALIDATE:
	/*
	* Revalidation machine check. Sick.
	* We don't notify the driver since we have to throw the device
	* away in any case.
	*/
	if (!disc) {
	device_unregister(&sch->dev);
	/* Reset intparm to zeroes. */
	sch->schib.pmcw.intparm = 0;
	cio_modify(sch);
	put_device(&sch->dev);
	ret = css_probe_device(irq);
	} else {
	/*
	* We can't immediately deregister the disconnected
	* device since it might block.
	*/
	spin_lock_irqsave(&sch->lock, flags);
	device_trigger_reprobe(sch);
	spin_unlock_irqrestore(&sch->lock, flags);
	ret = 0;
	}
	break;
	case CIO_OPER:
	if (disc) {
	spin_lock_irqsave(&sch->lock, flags);
	/* Get device operational again. */
	device_trigger_reprobe(sch);
	spin_unlock_irqrestore(&sch->lock, flags);
	}
	ret = sch ? 0 : css_probe_device(irq);
	break;
	default:
	BUG();
	ret = 0;
	}
	return ret;
	}

	static void
	css_rescan_devices(void)
	{
	int irq, ret;

	for (irq = 0; irq < __MAX_SUBCHANNELS; irq++) {
	ret = css_evaluate_subchannel(irq, 1);
	/* No more memory. It doesn't make sense to continue. No
	* panic because this can happen in midflight and just
	* because we can't use a new device is no reason to crash
	* the system. */
	if (ret == -ENOMEM)
	break;
	/* -ENXIO indicates that there are no more subchannels. */
	if (ret == -ENXIO)
	break;
	}
	}

	struct slow_subchannel {
	struct list_head slow_list;
	unsigned long schid;
	};

	static LIST_HEAD(slow_subchannels_head);
	static DEFINE_SPINLOCK(slow_subchannel_lock);

	static void
	css_trigger_slow_path(void)
	{
	CIO_TRACE_EVENT(4, "slowpath");

	if (need_rescan) {
	need_rescan = 0;
	css_rescan_devices();
	return;
	}

	spin_lock_irq(&slow_subchannel_lock);
	while (!list_empty(&slow_subchannels_head)) {
	struct slow_subchannel *slow_sch =
	list_entry(slow_subchannels_head.next,
	struct slow_subchannel, slow_list);

	list_del_init(slow_subchannels_head.next);
	spin_unlock_irq(&slow_subchannel_lock);
	css_evaluate_subchannel(slow_sch->schid, 1);
	spin_lock_irq(&slow_subchannel_lock);
	kfree(slow_sch);
	}
	spin_unlock_irq(&slow_subchannel_lock);
	}

	typedef void (workfunc)(void );
	DECLARE_WORK(slow_path_work, (workfunc)css_trigger_slow_path, NULL);
	struct workqueue_struct *slow_path_wq;

	/*
	* Rescan for new devices. FIXME: This is slow.
	* This function is called when we have lost CRWs due to overflows and we have
	* to do subchannel housekeeping.
	*/
	void
	css_reiterate_subchannels(void)
	{
	css_clear_subchannel_slow_list();
	need_rescan = 1;
	}

	/*
	* Called from the machine check handler for subchannel report words.
	*/
	int
	css_process_crw(int irq)
	{
	int ret;

	CIO_CRW_EVENT(2, "source is subchannel %04X\n", irq);

	if (need_rescan)
	/* We need to iterate all subchannels anyway. */
	return -EAGAIN;
	/*
	* Since we are always presented with IPI in the CRW, we have to
	* use stsch() to find out if the subchannel in question has come
	* or gone.
	*/
	ret = css_evaluate_subchannel(irq, 0);
	if (ret == -EAGAIN) {
	if (css_enqueue_subchannel_slow(irq)) {
	css_clear_subchannel_slow_list();
	need_rescan = 1;
	}
	}
	return ret;
	}

	static void __init
	css_generate_pgid(void)
	{
	/* Let's build our path group ID here. */
	if (css_characteristics_avail && css_general_characteristics.mcss)
	global_pgid.cpu_addr = 0x8000;
	else {
	#ifdef CONFIG_SMP
	global_pgid.cpu_addr = hard_smp_processor_id();
	#else
	global_pgid.cpu_addr = 0;
	#endif
	}
	global_pgid.cpu_id = ((cpuid_t *) __LC_CPUID)->ident;
	global_pgid.cpu_model = ((cpuid_t *) __LC_CPUID)->machine;
	global_pgid.tod_high = (__u32) (get_clock() >> 32);
	}

	/*
	* Now that the driver core is running, we can setup our channel subsystem.
	* The struct subchannel's are created during probing (except for the
	* static console subchannel).
	*/
	static int __init
	init_channel_subsystem (void)
	{
	int ret, irq;

	if (chsc_determine_css_characteristics() == 0)
	css_characteristics_avail = 1;

	css_generate_pgid();

	if ((ret = bus_register(&css_bus_type)))
	goto out;
	if ((ret = device_register (&css_bus_device)))
	goto out_bus;

	css_init_done = 1;

	ctl_set_bit(6, 28);

	for (irq = 0; irq < __MAX_SUBCHANNELS; irq++) {
	struct subchannel *sch;

	if (cio_is_console(irq))
	sch = cio_get_console_subchannel();
	else {
	sch = css_alloc_subchannel(irq);
	if (IS_ERR(sch))
	ret = PTR_ERR(sch);
	else
	ret = 0;
	if (ret == -ENOMEM)
	panic("Out of memory in "
	"init_channel_subsystem\n");
	/* -ENXIO: no more subchannels. */
	if (ret == -ENXIO)
	break;
	if (ret)
	continue;
	}
	/*
	* We register ALL valid subchannels in ioinfo, even those
	* that have been present before init_channel_subsystem.
	* These subchannels can't have been registered yet (kmalloc
	* not working) so we do it now. This is true e.g. for the
	* console subchannel.
	*/
	css_register_subchannel(sch);
	}
	return 0;

	out_bus:
	bus_unregister(&css_bus_type);
	out:
	return ret;
	}

	/*
	* find a driver for a subchannel. They identify by the subchannel
	* type with the exception that the console subchannel driver has its own
	* subchannel type although the device is an i/o subchannel
	*/
	static int
	css_bus_match (struct device dev, struct device_driver drv)
	{
	struct subchannel *sch = container_of (dev, struct subchannel, dev);
	struct css_driver *driver = container_of (drv, struct css_driver, drv);

	if (sch->st == driver->subchannel_type)
	return 1;

	return 0;
	}

	struct bus_type css_bus_type = {
	.name = "css",
	.match = &css_bus_match,
	};

	subsys_initcall(init_channel_subsystem);

	/*
	* Register root devices for some drivers. The release function must not be
	* in the device drivers, so we do it here.
	*/
	static void
	s390_root_dev_release(struct device *dev)
	{
	kfree(dev);
	}

	struct device *
	s390_root_dev_register(const char *name)
	{
	struct device *dev;
	int ret;

	if (!strlen(name))
	return ERR_PTR(-EINVAL);
	dev = kmalloc(sizeof(struct device), GFP_KERNEL);
	if (!dev)
	return ERR_PTR(-ENOMEM);
	memset(dev, 0, sizeof(struct device));
	strncpy(dev->bus_id, name, min(strlen(name), (size_t)BUS_ID_SIZE));
	dev->release = s390_root_dev_release;
	ret = device_register(dev);
	if (ret) {
	kfree(dev);
	return ERR_PTR(ret);
	}
	return dev;
	}

	void
	s390_root_dev_unregister(struct device *dev)
	{
	if (dev)
	device_unregister(dev);
	}

	int
	css_enqueue_subchannel_slow(unsigned long schid)
	{
	struct slow_subchannel *new_slow_sch;
	unsigned long flags;

	new_slow_sch = kmalloc(sizeof(struct slow_subchannel), GFP_ATOMIC);
	if (!new_slow_sch)
	return -ENOMEM;
	memset(new_slow_sch, 0, sizeof(struct slow_subchannel));
	new_slow_sch->schid = schid;
	spin_lock_irqsave(&slow_subchannel_lock, flags);
	list_add_tail(&new_slow_sch->slow_list, &slow_subchannels_head);
	spin_unlock_irqrestore(&slow_subchannel_lock, flags);
	return 0;
	}

	void
	css_clear_subchannel_slow_list(void)
	{
	unsigned long flags;

	spin_lock_irqsave(&slow_subchannel_lock, flags);
	while (!list_empty(&slow_subchannels_head)) {
	struct slow_subchannel *slow_sch =
	list_entry(slow_subchannels_head.next,
	struct slow_subchannel, slow_list);

	list_del_init(slow_subchannels_head.next);
	kfree(slow_sch);
	}
	spin_unlock_irqrestore(&slow_subchannel_lock, flags);
	}



	int
	css_slow_subchannels_exist(void)
	{
	return (!list_empty(&slow_subchannels_head));
	}

	MODULE_LICENSE("GPL");
	EXPORT_SYMBOL(css_bus_type);
	EXPORT_SYMBOL(s390_root_dev_register);
	EXPORT_SYMBOL(s390_root_dev_unregister);
	EXPORT_SYMBOL_GPL(css_characteristics_avail);