#undef TRACE_SYSTEM
#define TRACE_SYSTEM ras
#define TRACE_INCLUDE_FILE ras_event

#if !defined(_TRACE_HW_EVENT_MC_H) || defined(TRACE_HEADER_MULTI_READ)
#define _TRACE_HW_EVENT_MC_H

#include <linux/tracepoint.h>
#include <linux/edac.h>
#include <linux/ktime.h>
#include <linux/pci.h>
#include <linux/aer.h>
#include <linux/cper.h>
#include <linux/mm.h>

/*
 * MCE Extended Error Log trace event
 *
 * These events are generated when hardware detects a corrected or
 * uncorrected event.
 */

/* memory trace event */

#if defined(CONFIG_ACPI_EXTLOG) || defined(CONFIG_ACPI_EXTLOG_MODULE)
TRACE_EVENT(extlog_mem_event,
	TP_PROTO(struct cper_sec_mem_err *mem,
		 u32 err_seq,
		 const uuid_le *fru_id,
		 const char *fru_text,
		 u8 sev),

	TP_ARGS(mem, err_seq, fru_id, fru_text, sev),

	TP_STRUCT__entry(
		__field(u32, err_seq)
		__field(u8, etype)
		__field(u8, sev)
		__field(u64, pa)
		__field(u8, pa_mask_lsb)
		__field_struct(uuid_le, fru_id)
		__string(fru_text, fru_text)
		__field_struct(struct cper_mem_err_compact, data)
	),

	TP_fast_assign(
		__entry->err_seq = err_seq;
		if (mem->validation_bits & CPER_MEM_VALID_ERROR_TYPE)
			__entry->etype = mem->error_type;
		else
			__entry->etype = ~0;
		__entry->sev = sev;
		if (mem->validation_bits & CPER_MEM_VALID_PA)
			__entry->pa = mem->physical_addr;
		else
			__entry->pa = ~0ull;

		if (mem->validation_bits & CPER_MEM_VALID_PA_MASK)
			__entry->pa_mask_lsb = (u8)__ffs64(mem->physical_addr_mask);
		else
			__entry->pa_mask_lsb = ~0;
		__entry->fru_id = *fru_id;
		__assign_str(fru_text, fru_text);
		cper_mem_err_pack(mem, &__entry->data);
	),

	TP_printk("{%d} %s error: %s physical addr: %016llx (mask lsb: %x) %sFRU: %pUl %.20s",
		  __entry->err_seq,
		  cper_severity_str(__entry->sev),
		  cper_mem_err_type_str(__entry->etype),
		  __entry->pa,
		  __entry->pa_mask_lsb,
		  cper_mem_err_unpack(p, &__entry->data),
		  &__entry->fru_id,
		  __get_str(fru_text))
);
#endif

/*
 * Hardware Events Report
 *
 * Those events are generated when hardware detected a corrected or
 * uncorrected event, and are meant to replace the current API to report
 * errors defined on both EDAC and MCE subsystems.
 *
 * FIXME: Add events for handling memory errors originated from the
 *        MCE subsystem.
 */

/*
 * Hardware-independent Memory Controller specific events
 */

/*
 * Default error mechanisms for Memory Controller errors (CE and UE)
 */
TRACE_EVENT(mc_event,

	TP_PROTO(const unsigned int err_type,
		 const char *error_msg,
		 const char *label,
		 const int error_count,
		 const u8 mc_index,
		 const s8 top_layer,
		 const s8 mid_layer,
		 const s8 low_layer,
		 unsigned long address,
		 const u8 grain_bits,
		 unsigned long syndrome,
		 const char *driver_detail),

	TP_ARGS(err_type, error_msg, label, error_count, mc_index,
		top_layer, mid_layer, low_layer, address, grain_bits,
		syndrome, driver_detail),

	TP_STRUCT__entry(
		__field(	unsigned int,	error_type		)
		__string(	msg,		error_msg		)
		__string(	label,		label			)
		__field(	u16,		error_count		)
		__field(	u8,		mc_index		)
		__field(	s8,		top_layer		)
		__field(	s8,		middle_layer		)
		__field(	s8,		lower_layer		)
		__field(	long,		address			)
		__field(	u8,		grain_bits		)
		__field(	long,		syndrome		)
		__string(	driver_detail,	driver_detail		)
	),

	TP_fast_assign(
		__entry->error_type		= err_type;
		__assign_str(msg, error_msg);
		__assign_str(label, label);
		__entry->error_count		= error_count;
		__entry->mc_index		= mc_index;
		__entry->top_layer		= top_layer;
		__entry->middle_layer		= mid_layer;
		__entry->lower_layer		= low_layer;
		__entry->address		= address;
		__entry->grain_bits		= grain_bits;
		__entry->syndrome		= syndrome;
		__assign_str(driver_detail, driver_detail);
	),

	TP_printk("%d %s error%s:%s%s on %s (mc:%d location:%d:%d:%d address:0x%08lx grain:%d syndrome:0x%08lx%s%s)",
		  __entry->error_count,
		  mc_event_error_type(__entry->error_type),
		  __entry->error_count > 1 ? "s" : "",
		  ((char *)__get_str(msg))[0] ? " " : "",
		  __get_str(msg),
		  __get_str(label),
		  __entry->mc_index,
		  __entry->top_layer,
		  __entry->middle_layer,
		  __entry->lower_layer,
		  __entry->address,
		  1 << __entry->grain_bits,
		  __entry->syndrome,
		  ((char *)__get_str(driver_detail))[0] ? " " : "",
		  __get_str(driver_detail))
);

/*
 * PCIe AER Trace event
 *
 * These events are generated when hardware detects a corrected or
 * uncorrected event on a PCIe device. The event report has
 * the following structure:
 *
 * char * dev_name -	The name of the slot where the device resides
 *			([domain:]bus:device.function).
 * u32 status -		Either the correctable or uncorrectable register
 *			indicating what error or errors have been seen
 * u8 severity -	error severity 0:NONFATAL 1:FATAL 2:CORRECTED
 */

#define aer_correctable_errors					\
	{PCI_ERR_COR_RCVR,	"Receiver Error"},		\
	{PCI_ERR_COR_BAD_TLP,	"Bad TLP"},			\
	{PCI_ERR_COR_BAD_DLLP,	"Bad DLLP"},			\
	{PCI_ERR_COR_REP_ROLL,	"RELAY_NUM Rollover"},		\
	{PCI_ERR_COR_REP_TIMER,	"Replay Timer Timeout"},	\
	{PCI_ERR_COR_ADV_NFAT,	"Advisory Non-Fatal Error"},	\
	{PCI_ERR_COR_INTERNAL,	"Corrected Internal Error"},	\
	{PCI_ERR_COR_LOG_OVER,	"Header Log Overflow"}

#define aer_uncorrectable_errors				\
	{PCI_ERR_UNC_UND,	"Undefined"},			\
	{PCI_ERR_UNC_DLP,	"Data Link Protocol Error"},	\
	{PCI_ERR_UNC_SURPDN,	"Surprise Down Error"},		\
	{PCI_ERR_UNC_POISON_TLP,"Poisoned TLP"},		\
	{PCI_ERR_UNC_FCP,	"Flow Control Protocol Error"},	\
	{PCI_ERR_UNC_COMP_TIME,	"Completion Timeout"},		\
	{PCI_ERR_UNC_COMP_ABORT,"Completer Abort"},		\
	{PCI_ERR_UNC_UNX_COMP,	"Unexpected Completion"},	\
	{PCI_ERR_UNC_RX_OVER,	"Receiver Overflow"},		\
	{PCI_ERR_UNC_MALF_TLP,	"Malformed TLP"},		\
	{PCI_ERR_UNC_ECRC,	"ECRC Error"},			\
	{PCI_ERR_UNC_UNSUP,	"Unsupported Request Error"},	\
	{PCI_ERR_UNC_ACSV,	"ACS Violation"},		\
	{PCI_ERR_UNC_INTN,	"Uncorrectable Internal Error"},\
	{PCI_ERR_UNC_MCBTLP,	"MC Blocked TLP"},		\
	{PCI_ERR_UNC_ATOMEG,	"AtomicOp Egress Blocked"},	\
	{PCI_ERR_UNC_TLPPRE,	"TLP Prefix Blocked Error"}

TRACE_EVENT(aer_event,
	TP_PROTO(const char *dev_name,
		 const u32 status,
		 const u8 severity),

	TP_ARGS(dev_name, status, severity),

	TP_STRUCT__entry(
		__string(	dev_name,	dev_name	)
		__field(	u32,		status		)
		__field(	u8,		severity	)
	),

	TP_fast_assign(
		__assign_str(dev_name, dev_name);
		__entry->status		= status;
		__entry->severity	= severity;
	),

	TP_printk("%s PCIe Bus Error: severity=%s, %s\n",
		__get_str(dev_name),
		__entry->severity == AER_CORRECTABLE ? "Corrected" :
			__entry->severity == AER_FATAL ?
			"Fatal" : "Uncorrected, non-fatal",
		__entry->severity == AER_CORRECTABLE ?
		__print_flags(__entry->status, "|", aer_correctable_errors) :
		__print_flags(__entry->status, "|", aer_uncorrectable_errors))
);

/*
 * memory-failure recovery action result event
 *
 * unsigned long pfn -	Page Frame Number of the corrupted page
 * int type	-	Page types of the corrupted page
 * int result	-	Result of recovery action
 */

#ifdef CONFIG_MEMORY_FAILURE
#define MF_ACTION_RESULT	\
	EM ( MF_IGNORED, "Ignored" )	\
	EM ( MF_FAILED,  "Failed" )	\
	EM ( MF_DELAYED, "Delayed" )	\
	EMe ( MF_RECOVERED, "Recovered" )

#define MF_PAGE_TYPE		\
	EM ( MF_MSG_KERNEL, "reserved kernel page" )			\
	EM ( MF_MSG_KERNEL_HIGH_ORDER, "high-order kernel page" )	\
	EM ( MF_MSG_SLAB, "kernel slab page" )				\
	EM ( MF_MSG_DIFFERENT_COMPOUND, "different compound page after locking" ) \
	EM ( MF_MSG_POISONED_HUGE, "huge page already hardware poisoned" )	\
	EM ( MF_MSG_HUGE, "huge page" )					\
	EM ( MF_MSG_FREE_HUGE, "free huge page" )			\
	EM ( MF_MSG_UNMAP_FAILED, "unmapping failed page" )		\
	EM ( MF_MSG_DIRTY_SWAPCACHE, "dirty swapcache page" )		\
	EM ( MF_MSG_CLEAN_SWAPCACHE, "clean swapcache page" )		\
	EM ( MF_MSG_DIRTY_MLOCKED_LRU, "dirty mlocked LRU page" )	\
	EM ( MF_MSG_CLEAN_MLOCKED_LRU, "clean mlocked LRU page" )	\
	EM ( MF_MSG_DIRTY_UNEVICTABLE_LRU, "dirty unevictable LRU page" )	\
	EM ( MF_MSG_CLEAN_UNEVICTABLE_LRU, "clean unevictable LRU page" )	\
	EM ( MF_MSG_DIRTY_LRU, "dirty LRU page" )			\
	EM ( MF_MSG_CLEAN_LRU, "clean LRU page" )			\
	EM ( MF_MSG_TRUNCATED_LRU, "already truncated LRU page" )	\
	EM ( MF_MSG_BUDDY, "free buddy page" )				\
	EM ( MF_MSG_BUDDY_2ND, "free buddy page (2nd try)" )		\
	EMe ( MF_MSG_UNKNOWN, "unknown page" )

/*
 * First define the enums in MM_ACTION_RESULT to be exported to userspace
 * via TRACE_DEFINE_ENUM().
 */
#undef EM
#undef EMe
#define EM(a, b) TRACE_DEFINE_ENUM(a);
#define EMe(a, b)	TRACE_DEFINE_ENUM(a);

MF_ACTION_RESULT
MF_PAGE_TYPE

/*
 * driver for channel subsystem
 *
 * Copyright IBM Corp. 2002, 2010
 *
 * Author(s): Arnd Bergmann (arndb@de.ibm.com)
 *	      Cornelia Huck (cornelia.huck@de.ibm.com)
 */

#define KMSG_COMPONENT "cio"
#define pr_fmt(fmt) KMSG_COMPONENT ": " fmt

#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 <linux/reboot.h>
#include <linux/suspend.h>
#include <linux/proc_fs.h>
#include <asm/isc.h>
#include <asm/crw.h>

#include "css.h"
#include "cio.h"
#include "cio_debug.h"
#include "ioasm.h"
#include "chsc.h"
#include "device.h"
#include "idset.h"
#include "chp.h"

int css_init_done = 0;
int max_ssid;

struct channel_subsystem *channel_subsystems[__MAX_CSSID + 1];
static struct bus_type css_bus_type;

int
for_each_subchannel(int(*fn)(struct subchannel_id, void *), void *data)
{
	struct subchannel_id schid;
	int ret;

	init_subchannel_id(&schid);
	do {
		do {
			ret = fn(schid, data);
			if (ret)
				break;
		} while (schid.sch_no++ < __MAX_SUBCHANNEL);
		schid.sch_no = 0;
	} while (schid.ssid++ < max_ssid);
	return ret;
}

struct cb_data {
	void *data;
	struct idset *set;
	int (*fn_known_sch)(struct subchannel *, void *);
	int (*fn_unknown_sch)(struct subchannel_id, void *);
};

static int call_fn_known_sch(struct device *dev, void *data)
{
	struct subchannel *sch = to_subchannel(dev);
	struct cb_data *cb = data;
	int rc = 0;

	if (cb->set)
		idset_sch_del(cb->set, sch->schid);
	if (cb->fn_known_sch)
		rc = cb->fn_known_sch(sch, cb->data);
	return rc;
}

static int call_fn_unknown_sch(struct subchannel_id schid, void *data)
{
	struct cb_data *cb = data;
	int rc = 0;

	if (idset_sch_contains(cb->set, schid))
		rc = cb->fn_unknown_sch(schid, cb->data);
	return rc;
}

static int call_fn_all_sch(struct subchannel_id schid, void *data)
{
	struct cb_data *cb = data;
	struct subchannel *sch;
	int rc = 0;

	sch = get_subchannel_by_schid(schid);
	if (sch) {
		if (cb->fn_known_sch)
			rc = cb->fn_known_sch(sch, cb->data);
		put_device(&sch->dev);
	} else {
		if (cb->fn_unknown_sch)
			rc = cb->fn_unknown_sch(schid, cb->data);
	}

	return rc;
}

int for_each_subchannel_staged(int (*fn_known)(struct subchannel *, void *),
			       int (*fn_unknown)(struct subchannel_id,
			       void *), void *data)
{
	struct cb_data cb;
	int rc;

	cb.data = data;
	cb.fn_known_sch = fn_known;
	cb.fn_unknown_sch = fn_unknown;

	if (fn_known && !fn_unknown) {
		/* Skip idset allocation in case of known-only loop. */
		cb.set = NULL;
		return bus_for_each_dev(&css_bus_type, NULL, &cb,
					call_fn_known_sch);
	}

	cb.set = idset_sch_new();
	if (!cb.set)
		/* fall back to brute force scanning in case of oom */
		return for_each_subchannel(call_fn_all_sch, &cb);

	idset_fill(cb.set);

	/* Process registered subchannels. */
	rc = bus_for_each_dev(&css_bus_type, NULL, &cb, call_fn_known_sch);
	if (rc)
		goto out;
	/* Process unregistered subchannels. */
	if (fn_unknown)
		rc = for_each_subchannel(call_fn_unknown_sch, &cb);
out:
	idset_free(cb.set);

	return rc;
}

static void css_sch_todo(struct work_struct *work);

static int css_sch_create_locks(struct subchannel *sch)
{
	sch->lock = kmalloc(sizeof(*sch->lock), GFP_KERNEL);
	if (!sch->lock)
		return -ENOMEM;

	spin_lock_init(sch->lock);
	mutex_init(&sch->reg_mutex);

	return 0;
}

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

	sch->config.intparm = 0;
	cio_commit_config(sch);
	kfree(sch->lock);
	kfree(sch);
}

struct subchannel *css_alloc_subchannel(struct subchannel_id schid)
{
	struct subchannel *sch;
	int ret;

	sch = kzalloc(sizeof(*sch), GFP_KERNEL | GFP_DMA);
	if (!sch)
		return ERR_PTR(-ENOMEM);

	ret = cio_validate_subchannel(sch, schid);
	if (ret < 0)
		goto err;

	ret = css_sch_create_locks(sch);
	if (ret)
		goto err;

	INIT_WORK(&sch->todo_work, css_sch_todo);
	sch->dev.release = &css_subchannel_release;
	device_initialize(&sch->dev);
	return sch;

err:
	kfree(sch);
	return ERR_PTR(ret);
}

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

	mutex_lock(&sch->reg_mutex);
	dev_set_name(&sch->dev, "0.%x.%04x", sch->schid.ssid,
		     sch->schid.sch_no);
	ret = device_add(&sch->dev);
	mutex_unlock(&sch->reg_mutex);
	return ret;
}

/**
 * css_sch_device_unregister - unregister a subchannel
 * @sch: subchannel to be unregistered
 */
void css_sch_device_unregister(struct subchannel *sch)
{
	mutex_lock(&sch->reg_mutex);
	if (device_is_registered(&sch->dev))
		device_unregister(&sch->dev);
	mutex_unlock(&sch->reg_mutex);
}
EXPORT_SYMBOL_GPL(css_sch_device_unregister);

static void ssd_from_pmcw(struct chsc_ssd_info *ssd, struct pmcw *pmcw)
{
	int i;
	int mask;

	memset(ssd, 0, sizeof(struct chsc_ssd_info));
	ssd->path_mask = pmcw->pim;
	for (i = 0; i < 8; i++) {
		mask = 0x80 >> i;
		if (pmcw->pim & mask) {
			chp_id_init(&ssd->chpid[i]);
			ssd->chpid[i].id = pmcw->chpid[i];
		}
	}
}

static void ssd_register_chpids(struct chsc_ssd_info *ssd)
{
	int i;
	int mask;

	for (i = 0; i < 8; i++) {
		mask = 0x80 >> i;
		if (ssd->path_mask & mask)
			if (!chp_is_registered(ssd->chpid[i]))
				chp_new(ssd->chpid[i]);
	}
}

void css_update_ssd_info(struct subchannel *sch)
{
	int ret;

	ret = chsc_get_ssd_info(sch->schid, &sch->ssd_info);
	if (ret)
		ssd_from_pmcw(&sch->ssd_info, &sch->schib.pmcw);

	ssd_register_chpids(&sch->ssd_info);
}

static ssize_t type_show(struct device *dev, struct device_attribute *attr,
			 char *buf)
{
	struct subchannel *sch = to_subchannel(dev);

	return sprintf(buf, "%01x\n", sch->st);
}

static DEVICE_ATTR(type, 0444, type_show, NULL);

static ssize_t modalias_show(struct device *dev, struct device_attribute *attr,
			     char *buf)
{
	struct subchannel *sch = to_subchannel(dev);

	return sprintf(buf, "css:t%01X\n", sch->st);
}

static DEVICE_ATTR(modalias, 0444, modalias_show, NULL);

static struct attribute *subch_attrs[] = {
	&dev_attr_type.attr,
	&dev_attr_modalias.attr,
	NULL,
};

static struct attribute_group subch_attr_group = {
	.attrs = subch_attrs,
};

static const struct attribute_group *default_subch_attr_groups[] = {
	&subch_attr_group,
	NULL,
};

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

	/* Initialize the subchannel structure */
	sch->dev.parent = &channel_subsystems[0]->device;
	sch->dev.bus = &css_bus_type;
	sch->dev.groups = default_subch_attr_groups;
	/*
	 * We don't want to generate uevents for I/O subchannels that don't
	 * have a working ccw device behind them since they will be
	 * unregistered before they can be used anyway, so we delay the add
	 * uevent until after device recognition was successful.
	 * Note that we suppress the uevent for all subchannel types;
	 * the subchannel driver can decide itself when it wants to inform
	 * userspace of its existence.
	 */
	dev_set_uevent_suppress(&sch->dev, 1);
	css_update_ssd_info(sch);
	/* make it known to the system */
	ret = css_sch_device_register(sch);
	if (ret) {
		CIO_MSG_EVENT(0, "Could not register sch 0.%x.%04x: %d\n",
			      sch->schid.ssid, sch->schid.sch_no, ret);
		return ret;
	}
	if (!sch->driver) {
		/*
		 * No driver matched. Generate the uevent now so that
		 * a fitting driver module may be loaded based on the
		 * modalias.
		 */
		dev_set_uevent_suppress(&sch->dev, 0);
		kobject_uevent(&sch->dev.kobj, KOBJ_ADD);
	}
	return ret;
}

static int css_probe_device(struct subchannel_id schid)
{
	struct subchannel *sch;
	int ret;

	sch = css_alloc_subchannel(schid);
	if (IS_ERR(sch))
		return PTR_ERR(sch);

	ret = css_register_subchannel(sch);
	if (ret)
		put_device(&sch->dev);

	return ret;
}

static int
check_subchannel(struct device * dev, void * data)
{
	struct subchannel *sch;
	struct subchannel_id *schid = data;

	sch = to_subchannel(dev);
	return schid_equal(&sch->schid, schid);
}

struct subchannel *
get_subchannel_by_schid(struct subchannel_id schid)
{
	struct device *dev;

	dev = bus_find_device(&css_bus_type, NULL,
			      &schid, check_subchannel);

	return dev ? to_subchannel(dev) : NULL;
}

/**
 * css_sch_is_valid() - check if a subchannel is valid
 * @schib: subchannel information block for the subchannel
 */
int css_sch_is_valid(struct schib *schib)
{
	if ((schib->pmcw.st == SUBCHANNEL_TYPE_IO) && !schib->pmcw.dnv)
		return 0;
	if ((schib->pmcw.st == SUBCHANNEL_TYPE_MSG) && !schib->pmcw.w)
		return 0;
	return 1;
}
EXPORT_SYMBOL_GPL(css_sch_is_valid);

static int css_evaluate_new_subchannel(struct subchannel_id schid, int slow)
{
	struct schib schib;

	if (!slow) {
		/* Will be done on the slow path. */
		return -EAGAIN;
	}
	if (stsch_err(schid, &schib)) {
		/* Subchannel is not provided. */
		return -ENXIO;
	}
	if (!css_sch_is_valid(&schib)) {
		/* Unusable - ignore. */
		return 0;
	}
	CIO_MSG_EVENT(4, "event: sch 0.%x.%04x, new\n", schid.ssid,
		      schid.sch_no);

	return css_probe_device(schid);
}

static int css_evaluate_known_subchannel(struct subchannel *sch, int slow)
{
	int ret = 0;

	if (sch->driver) {
		if (sch->driver->sch_event)
			ret = sch->driver->sch_event(sch, slow);
		else
			dev_dbg(&sch->dev,
				"Got subchannel machine check but "
				"no sch_event handler provided.\n");
	}
	if (ret != 0 && ret != -EAGAIN) {
		CIO_MSG_EVENT(2, "eval: sch 0.%x.%04x, rc=%d\n",
			      sch->schid.ssid, sch->schid.sch_no, ret);
	}
	return ret;
}

static void css_evaluate_subchannel(struct subchannel_id schid, int slow)
{
	struct subchannel *sch;
	int ret;

	sch = get_subchannel_by_schid(schid);
	if (sch) {
		ret = css_evaluate_known_subchannel(sch, slow);
		put_device(&sch->dev);
	} else
		ret = css_evaluate_new_subchannel(schid, slow);
	if (ret == -EAGAIN)
		css_schedule_eval(schid);
}

/**
 * css_sched_sch_todo - schedule a subchannel operation
 * @sch: subchannel
 * @todo: todo
 *
 * Schedule the operation identified by @todo to be performed on the slow path
 * workqueue. Do nothing if another operation with higher priority is already
 * scheduled. Needs to be called with subchannel lock held.
 */
void css_sched_sch_todo(struct subchannel *sch, enum sch_todo todo)
{
	CIO_MSG_EVENT(4, "sch_todo: sched sch=0.%x.%04x todo=%d\n",
		      sch->schid.ssid, sch->schid.sch_no, todo);
	if (sch->todo >= todo)
		return;
	/* Get workqueue ref. */
	if (!get_device(&sch->dev))
		return;
	sch->todo = todo;
	if (!queue_work(cio_work_q, &sch->todo_work)) {
		/* Already queued, release workqueue ref. */
		put_device(&sch->dev);
	}
}
EXPORT_SYMBOL_GPL(css_sched_sch_todo);

static void css_sch_todo(struct work_struct *work)
{
	struct subchannel *sch;
	enum sch_todo todo;
	int ret;

	sch = container_of(work, struct subchannel, todo_work);
	/* Find out todo. */
	spin_lock_irq(sch->lock);
	todo = sch->todo;
	CIO_MSG_EVENT(4, "sch_todo: sch=0.%x.%04x, todo=%d\n", sch->schid.ssid,
		      sch->schid.sch_no, todo);
	sch->todo = SCH_TODO_NOTHING;
	spin_unlock_irq(sch->lock);
	/* Perform todo. */
	switch (todo) {
	case SCH_TODO_NOTHING:
		break;
	case SCH_TODO_EVAL:
		ret = css_evaluate_known_subchannel(sch, 1);
		if (ret == -EAGAIN) {
			spin_lock_irq(sch->lock);
			css_sched_sch_todo(sch, todo);
			spin_unlock_irq(sch->lock);
		}
		break;
	case SCH_TODO_UNREG:
		css_sch_device_unregister(sch);
		break;
	}
	/* Release workqueue ref. */
	put_device(&sch->dev);
}

static struct idset *slow_subchannel_set;
static spinlock_t slow_subchannel_lock;
static wait_queue_head_t css_eval_wq;
static atomic_t css_eval_scheduled;

static int __init slow_subchannel_init(void)
{
	spin_lock_init(&slow_subchannel_lock);
	atomic_set(&css_eval_scheduled, 0);
	init_waitqueue_head(&css_eval_wq);
	slow_subchannel_set = idset_sch_new();
	if (!slow_subchannel_set) {
		CIO_MSG_EVENT(0, "could not allocate slow subchannel set\n");
		return -ENOMEM;
	}
	return 0;
}

static int slow_eval_known_fn(struct subchannel *sch, void *data)
{
	int eval;
	int rc;

	spin_lock_irq(&slow_subchannel_lock);
	eval = idset_sch_contains(slow_subchannel_set, sch->schid);
	idset_sch_del(slow_subchannel_set, sch->schid);
	spin_unlock_irq(&slow_subchannel_lock);
	if (eval) {
		rc = css_evaluate_known_subchannel(sch, 1);
		if (rc == -EAGAIN)
			css_schedule_eval(sch->schid);
	}
	return 0;
}

static int slow_eval_unknown_fn(struct subchannel_id schid, void *data)
{
	int eval;
	int rc = 0;

	spin_lock_irq(&slow_subchannel_lock);
	eval = idset_sch_contains(slow_subchannel_set, schid);
	idset_sch_del(slow_subchannel_set, schid);
	spin_unlock_irq(&slow_subchannel_lock);
	if (eval) {
		rc = css_evaluate_new_subchannel(schid, 1);
		switch (rc) {
		case -EAGAIN:
			css_schedule_eval(schid);
			rc = 0;
			break;
		case -ENXIO:
		case -ENOMEM:
		case -EIO:
			/* These should abort looping */
			spin_lock_irq(&slow_subchannel_lock);
			idset_sch_del_subseq(slow_subchannel_set, schid);
			spin_unlock_irq(&slow_subchannel_lock);
			break;
		default:
			rc = 0;
		}
		/* Allow scheduling here since the containing loop might
		 * take a while.  */
		cond_resched();
	}
	return rc;
}

static void css_slow_path_func(struct work_struct *unused)
{
	unsigned long flags;

	CIO_TRACE_EVENT(4, "slowpath");
	for_each_subchannel_staged(slow_eval_known_fn, slow_eval_unknown_fn,
				   NULL);
	spin_lock_irqsave(&slow_subchannel_lock, flags);
	if (idset_is_empty(slow_subchannel_set)) {
		atomic_set(&css_eval_scheduled, 0);
		wake_up(&css_eval_wq);
	}
	spin_unlock_irqrestore(&slow_subchannel_lock, flags);
}

static DECLARE_DELAYED_WORK(slow_path_work, css_slow_path_func);
struct workqueue_struct *cio_work_q;

void css_schedule_eval(struct subchannel_id schid)
{
	unsigned long flags;

	spin_lock_irqsave(&slow_subchannel_lock, flags);
	idset_sch_add(slow_subchannel_set, schid);
	atomic_set(&css_eval_scheduled, 1);
	queue_delayed_work(cio_work_q, &slow_path_work, 0);
	spin_unlock_irqrestore(&slow_subchannel_lock, flags);
}

void css_schedule_eval_all(void)
{
	unsigned long flags;

	spin_lock_irqsave(&slow_subchannel_lock, flags);
	idset_fill(slow_subchannel_set);
	atomic_set(&css_eval_scheduled, 1);
	queue_delayed_work(cio_work_q, &slow_path_work, 0);
	spin_unlock_irqrestore(&slow_subchannel_lock, flags);
}

static int __unset_registered(struct device *dev, void *data)
{
	struct idset *set = data;
	struct subchannel *sch = to_subchannel(dev);

	idset_sch_del(set, sch->schid);
	return 0;
}

void css_schedule_eval_all_unreg(unsigned long delay)
{
	unsigned long flags;
	struct idset *unreg_set;

	/* Find unregistered subchannels. */
	unreg_set = idset_sch_new();
	if (!unreg_set) {
		/* Fallback. */
		css_schedule_eval_all();
		return;
	}
	idset_fill(unreg_set);
	bus_for_each_dev(&css_bus_type, NULL, unreg_set, __unset_registered);
	/* Apply to slow_subchannel_set. */
	spin_lock_irqsave(&slow_subchannel_lock, flags);
	idset_add_set(slow_subchannel_set, unreg_set);
	atomic_set(&css_eval_scheduled, 1);
	queue_delayed_work(cio_work_q, &slow_path_work, delay);
	spin_unlock_irqrestore(&slow_subchannel_lock, flags);
	idset_free(unreg_set);
}

void css_wait_for_slow_path(void)
{
	flush_workqueue(cio_work_q);
}

/* Schedule reprobing of all unregistered subchannels. */
void css_schedule_reprobe(void)
{
	/* Schedule with a delay to allow merging of subsequent calls. */
	css_schedule_eval_all_unreg(1 * HZ);
}
EXPORT_SYMBOL_GPL(css_schedule_reprobe);

/*
 * Called from the machine check handler for subchannel report words.
 */
static void css_process_crw(struct crw *crw0, struct crw *crw1, int overflow)
{
	struct subchannel_id mchk_schid;
	struct subchannel *sch;

	if (overflow) {
		css_schedule_eval_all();
		return;
	}
	CIO_CRW_EVENT(2, "CRW0 reports slct=%d, oflw=%d, "
		      "chn=%d, rsc=%X, anc=%d, erc=%X, rsid=%X\n",
		      crw0->slct, crw0->oflw, crw0->chn, crw0->rsc, crw0->anc,
		      crw0->erc, crw0->rsid);
	if (crw1)
		CIO_CRW_EVENT(2, "CRW1 reports slct=%d, oflw=%d, "
			      "chn=%d, rsc=%X, anc=%d, erc=%X, rsid=%X\n",
			      crw1->slct, crw1->oflw, crw1->chn, crw1->rsc,
			      crw1->anc, crw1->erc, crw1->rsid);
	init_subchannel_id(&mchk_schid);
	mchk_schid.sch_no = crw0->rsid;
	if (crw1)
		mchk_schid.ssid = (crw1->rsid >> 4) & 3;

	if (crw0->erc == CRW_ERC_PMOD) {
		sch = get_subchannel_by_schid(mchk_schid);
		if (sch) {
			css_update_ssd_info(sch);
			put_device(&sch->dev);
		}
	}
	/*
	 * 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.
	 */
	css_evaluate_subchannel(mchk_schid, 0);
}

static void __init
css_generate_pgid(struct channel_subsystem *css, u32 tod_high)
{
	struct cpuid cpu_id;

	if (css_general_characteristics.mcss) {
		css->global_pgid.pgid_high.ext_cssid.version = 0x80;
		css->global_pgid.pgid_high.ext_cssid.cssid = css->cssid;
	} else {
		css->global_pgid.pgid_high.cpu_addr = stap();
	}
	get_cpu_id(&cpu_id);
	css->global_pgid.cpu_id = cpu_id.ident;
	css->global_pgid.cpu_model = cpu_id.machine;
	css->global_pgid.tod_high = tod_high;
}

static void
channel_subsystem_release(struct device *dev)
{
	struct channel_subsystem *css;

	css = to_css(dev);
	mutex_destroy(&css->mutex);
	if (css->pseudo_subchannel) {
		/* Implies that it has been generated but never registered. */
		css_subchannel_release(&css->pseudo_subchannel->dev);
		css->pseudo_subchannel = NULL;
	}
	kfree(css);
}

static ssize_t
css_cm_enable_show(struct device *dev, struct device_attribute *attr,
		   char *buf)
{
	struct channel_subsystem *css = to_css(dev);
	int ret;

	if (!css)
		return 0;
	mutex_lock(&css->mutex);
	ret = sprintf(buf, "%x\n", css->cm_enabled);
	mutex_unlock(&css->mutex);
	return ret;
}

static ssize_t
css_cm_enable_store(struct device *dev, struct device_attribute *attr,
		    const char *buf, size_t count)
{
	struct channel_subsystem *css = to_css(dev);
	int ret;
	unsigned long val;

	ret = kstrtoul(buf, 16, &val);
	if (ret)
		return ret;
	mutex_lock(&css->mutex);
	switch (val) {
	case 0:
		ret = css->cm_enabled ? chsc_secm(css, 0) : 0;
		break;
	case 1:
		ret = css->cm_enabled ? 0 : chsc_secm(css, 1);
		break;
	default:
		ret = -EINVAL;
	}
	mutex_unlock(&css->mutex);
	return ret < 0 ? ret : count;
}

static DEVICE_ATTR(cm_enable, 0644, css_cm_enable_show, css_cm_enable_store);

static int __init setup_css(int nr)
{
	u32 tod_high;
	int ret;
	struct channel_subsystem *css;

	css = channel_subsystems[nr];
	memset(css, 0, sizeof(struct channel_subsystem));
	css->pseudo_subchannel =
		kzalloc(sizeof(*css->pseudo_subchannel), GFP_KERNEL);
	if (!css->pseudo_subchannel)
		return -ENOMEM;
	css->pseudo_subchannel->dev.parent = &css->device;
	css->pseudo_subchannel->dev.release = css_subchannel_release;
	dev_set_name(&css->pseudo_subchannel->dev, "defunct");
	mutex_init(&css->pseudo_subchannel->reg_mutex);
	ret = css_sch_create_locks(css->pseudo_subchannel);
	if (ret) {
		kfree(css->pseudo_subchannel);
		return ret;
	}
	mutex_init(&css->mutex);
	css->valid = 1;
	css->cssid = nr;
	dev_set_name(&css->device, "css%x", nr);
	css->device.release = channel_subsystem_release;
	tod_high = (u32) (get_tod_clock() >> 32);
	css_generate_pgid(css, tod_high);
	return 0;
}

static int css_reboot_event(struct notifier_block *this,
			    unsigned long event,
			    void *ptr)
{
	int ret, i;

	ret = NOTIFY_DONE;
	for (i = 0; i <= __MAX_CSSID; i++) {
		struct channel_subsystem *css;

		css = channel_subsystems[i];
		mutex_lock(&css->mutex);
		if (css->cm_enabled)
			if (chsc_secm(css, 0))
				ret = NOTIFY_BAD;
		mutex_unlock(&css->mutex);
	}

	return ret;
}

static struct notifier_block css_reboot_notifier = {
	.notifier_call = css_reboot_event,
};

/*
 * Since the css devices are neither on a bus nor have a class
 * nor have a special device type, we cannot stop/restart channel
 * path measurements via the normal suspend/resume callbacks, but have
 * to use notifiers.
 */
static int css_power_event(struct notifier_block *this, unsigned long event,
			   void *ptr)
{
	int ret, i;

	switch (event) {
	case PM_HIBERNATION_PREPARE:
	case PM_SUSPEND_PREPARE:
		ret = NOTIFY_DONE;
		for (i = 0; i <= __MAX_CSSID; i++) {
			struct channel_subsystem *css;

			css = channel_subsystems[i];
			mutex_lock(&css->mutex);
			if (!css->cm_enabled) {
				mutex_unlock(&css->mutex);
				continue;
			}
			ret = __chsc_do_secm(css, 0);
			ret = notifier_from_errno(ret);
			mutex_unlock(&css->mutex);
		}
		break;
	case PM_POST_HIBERNATION:
	case PM_POST_SUSPEND:
		ret = NOTIFY_DONE;
		for (i = 0; i <= __MAX_CSSID; i++) {
			struct channel_subsystem *css;

			css = channel_subsystems[i];
			mutex_lock(&css->mutex);
			if (!css->cm_enabled) {
				mutex_unlock(&css->mutex);
				continue;
			}
			ret = __chsc_do_secm(css, 1);
			ret = notifier_from_errno(ret);
			mutex_unlock(&css->mutex);
		}
		/* search for subchannels, which appeared during hibernation */
		css_schedule_reprobe();
		break;
	default:
		ret = NOTIFY_DONE;
	}
	return ret;

}
static struct notifier_block css_power_notifier = {
	.notifier_call = css_power_event,
};

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

	ret = chsc_init();
	if (ret)
		return ret;

	chsc_determine_css_characteristics();
	/* Try to enable MSS. */
	ret = chsc_enable_facility(CHSC_SDA_OC_MSS);
	if (ret)
		max_ssid = 0;
	else /* Success. */
		max_ssid = __MAX_SSID;

	ret = slow_subchannel_init();
	if (ret)
		goto out;

	ret = crw_register_handler(CRW_RSC_SCH, css_process_crw);
	if (ret)
		goto out;

	if ((ret = bus_register(&css_bus_type)))
		goto out;

	/* Setup css structure. */
	for (i = 0; i <= __MAX_CSSID; i++) {
		struct channel_subsystem *css;

		css = kmalloc(sizeof(struct channel_subsystem), GFP_KERNEL);
		if (!css) {
			ret = -ENOMEM;
			goto out_unregister;
		}
		channel_subsystems[i] = css;
		ret = setup_css(i);
		if (ret) {
			kfree(channel_subsystems[i]);
			goto out_unregister;
		}
		ret = device_register(&css->device);
		if (ret) {
			put_device(&css->device);
			goto out_unregister;
		}
		if (css_chsc_characteristics.secm) {
			ret = device_create_file(&css->device,
						 &dev_attr_cm_enable);
			if (ret)
				goto out_device;
		}
		ret = device_register(&css->pseudo_subchannel->dev);
		if (ret) {
			put_device(&css->pseudo_subchannel->dev);
			goto out_file;
		}
	}
	ret = register_reboot_notifier(&css_reboot_notifier);
	if (ret)
		goto out_unregister;
	ret = register_pm_notifier(&css_power_notifier);
	if (ret) {
		unregister_reboot_notifier(&css_reboot_notifier);
		goto out_unregister;
	}
	css_init_done = 1;

	/* Enable default isc for I/O subchannels. */
	isc_register(IO_SCH_ISC);

	return 0;
out_file:
	if (css_chsc_characteristics.secm)
		device_remove_file(&channel_subsystems[i]->device,
				   &dev_attr_cm_enable);
out_device:
	device_unregister(&channel_subsystems[i]->device);
out_unregister:
	while (i > 0) {
		struct channel_subsystem *css;

		i--;
		css = channel_subsystems[i];
		device_unregister(&css->pseudo_subchannel->dev);
		css->pseudo_subchannel = NULL;
		if (css_chsc_characteristics.secm)
			device_remove_file(&css->device,
					   &dev_attr_cm_enable);
		device_unregister(&css->device);
	}
	bus_unregister(&css_bus_type);
out:
	crw_unregister_handler(CRW_RSC_SCH);
	idset_free(slow_subchannel_set);
	chsc_init_cleanup();
	pr_alert("The CSS device driver initialization failed with "
		 "errno=%d\n", ret);
	return ret;
}

static void __init css_bus_cleanup(void)
{
	struct channel_subsystem *css;
	int i;

	for (i = 0; i <= __MAX_CSSID; i++) {
		css = channel_subsystems[i];
		device_unregister(&css->pseudo_subchannel->dev);
		css->pseudo_subchannel = NULL;
		if (css_chsc_characteristics.secm)
			device_remove_file(&css->device, &dev_attr_cm_enable);
		device_unregister(&css->device);
	}
	bus_unregister(&css_bus_type);
	crw_unregister_handler(CRW_RSC_SCH);
	idset_free(slow_subchannel_set);
	chsc_init_cleanup();
	isc_unregister(IO_SCH_ISC);
}

static int __init channel_subsystem_init(void)
{
	int ret;

	ret = css_bus_init();
	if (ret)
		return ret;
	cio_work_q = create_singlethread_workqueue("cio");
	if (!cio_work_q) {
		ret = -ENOMEM;
		goto out_bus;
	}
	ret = io_subchannel_init();
	if (ret)
		goto out_wq;

	return ret;
out_wq:
	destroy_workqueue(cio_work_q);
out_bus:
	css_bus_cleanup();
	return ret;
}
subsys_initcall(channel_subsystem_init);

static int css_settle(struct device_driver *drv, void *unused)
{
	struct css_driver *cssdrv = to_cssdriver(drv);

	if (cssdrv->settle)
		return cssdrv->settle();
	return 0;
}

int css_complete_work(void)
{
	int ret;

	/* Wait for the evaluation of subchannels to finish. */
	ret = wait_event_interruptible(css_eval_wq,
				       atomic_read(&css_eval_scheduled) == 0);
	if (ret)
		return -EINTR;
	flush_workqueue(cio_work_q);
	/* Wait for the subchannel type specific initialization to finish */
	return bus_for_each_drv(&css_bus_type, NULL, NULL, css_settle);
}


/*
 * Wait for the initialization of devices to finish, to make sure we are
 * done with our setup if the search for the root device starts.
 */
static int __init channel_subsystem_init_sync(void)
{
	/* Register subchannels which are already in use. */
	cio_register_early_subchannels();
	/* Start initial subchannel evaluation. */
	css_schedule_eval_all();
	css_complete_work();
	return 0;
}
subsys_initcall_sync(channel_subsystem_init_sync);

void channel_subsystem_reinit(void)
{
	struct channel_path *chp;
	struct chp_id chpid;

	chsc_enable_facility(CHSC_SDA_OC_MSS);
	chp_id_for_each(&chpid) {
		chp = chpid_to_chp(chpid);
		if (chp)
			chp_update_desc(chp);
	}
	cmf_reactivate();
}

#ifdef CONFIG_PROC_FS
static ssize_t cio_settle_write(struct file *file, const char __user *buf,
				size_t count, loff_t *ppos)
{
	int ret;

	/* Handle pending CRW's. */
	crw_wait_for_channel_report();
	ret = css_complete_work();

	return ret ? ret : count;
}

static const struct file_operations cio_settle_proc_fops = {
	.open = nonseekable_open,
	.write = cio_settle_write,
	.llseek = no_llseek,
};

static int __init cio_settle_init(void)
{
	struct proc_dir_entry *entry;

	entry = proc_create("cio_settle", S_IWUSR, NULL,
			    &cio_settle_proc_fops);
	if (!entry)
		return -ENOMEM;
	return 0;
}
device_initcall(cio_settle_init);
#endif /*CONFIG_PROC_FS*/

int sch_is_pseudo_sch(struct subchannel *sch)
{
	return sch == to_css(sch->dev.parent)->pseudo_subchannel;
}

static int css_bus_match(struct device *dev, struct device_driver *drv)
{
	struct subchannel *sch = to_subchannel(dev);
	struct css_driver *driver = to_cssdriver(drv);
	struct css_device_id *id;

	for (id = driver->subchannel_type; id->match_flags; id++) {
		if (sch->st == id->type)
			return 1;
	}

	return 0;
}

static int css_probe(struct device *dev)
{
	struct subchannel *sch;
	int ret;

	sch = to_subchannel(dev);
	sch->driver = to_cssdriver(dev->driver);
	ret = sch->driver->probe ? sch->driver->probe(sch) : 0;
	if (ret)
		sch->driver = NULL;
	return ret;
}

static int css_remove(struct device *dev)
{
	struct subchannel *sch;
	int ret;

	sch = to_subchannel(dev);
	ret = sch->driver->remove ? sch->driver->remove(sch) : 0;
	sch->driver = NULL;
	return ret;
}

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

	sch = to_subchannel(dev);
	if (sch->driver && sch->driver->shutdown)
		sch->driver->shutdown(sch);
}

static int css_uevent(struct device *dev, struct kobj_uevent_env *env)
{
	struct subchannel *sch = to_subchannel(dev);
	int ret;

	ret = add_uevent_var(env, "ST=%01X", sch->st);
	if (ret)
		return ret;
	ret = add_uevent_var(env, "MODALIAS=css:t%01X", sch->st);
	return ret;
}

static int css_pm_prepare(struct device *dev)
{
	struct subchannel *sch = to_subchannel(dev);
	struct css_driver *drv;

	if (mutex_is_locked(&sch->reg_mutex))
		return -EAGAIN;
	if (!sch->dev.driver)
		return 0;
	drv = to_cssdriver(sch->dev.driver);
	/* Notify drivers that they may not register children. */
	return drv->prepare ? drv->prepare(sch) : 0;
}

static void css_pm_complete(struct device *dev)
{
	struct subchannel *sch = to_subchannel(dev);
	struct css_driver *drv;

	if (!sch->dev.driver)
		return;
	drv = to_cssdriver(sch->dev.driver);
	if (drv->complete)
		drv->complete(sch);
}

static int css_pm_freeze(struct device *dev)
{
	struct subchannel *sch = to_subchannel(dev);
	struct css_driver *drv;

	if (!sch->dev.driver)
		return 0;
	drv = to_cssdriver(sch->dev.driver);
	return drv->freeze ? drv->freeze(sch) : 0;
}

static int css_pm_thaw(struct device *dev)
{
	struct subchannel *sch = to_subchannel(dev);
	struct css_driver *drv;

	if (!sch->dev.driver)
		return 0;
	drv = to_cssdriver(sch->dev.driver);
	return drv->thaw ? drv->thaw(sch) : 0;
}

static int css_pm_restore(struct device *dev)
{
	struct subchannel *sch = to_subchannel(dev);
	struct css_driver *drv;

	css_update_ssd_info(sch);
	if (!sch->dev.driver)
		return 0;
	drv = to_cssdriver(sch->dev.driver);
	return drv->restore ? drv->restore(sch) : 0;
}

static const struct dev_pm_ops css_pm_ops = {
	.prepare = css_pm_prepare,
	.complete = css_pm_complete,
	.freeze = css_pm_freeze,
	.thaw = css_pm_thaw,
	.restore = css_pm_restore,
};

static struct bus_type css_bus_type = {
	.name     = "css",
	.match    = css_bus_match,
	.probe    = css_probe,
	.remove   = css_remove,
	.shutdown = css_shutdown,
	.uevent   = css_uevent,
	.pm = &css_pm_ops,
};

/**
 * css_driver_register - register a css driver
 * @cdrv: css driver to register
 *
 * This is mainly a wrapper around driver_register that sets name
 * and bus_type in the embedded struct device_driver correctly.
 */
int css_driver_register(struct css_driver *cdrv)
{
	cdrv->drv.bus = &css_bus_type;
	return driver_register(&cdrv->drv);
}
EXPORT_SYMBOL_GPL(css_driver_register);

/**
 * css_driver_unregister - unregister a css driver
 * @cdrv: css driver to unregister
 *
 * This is a wrapper around driver_unregister.
 */
void css_driver_unregister(struct css_driver *cdrv)
{
	driver_unregister(&cdrv->drv);
}
EXPORT_SYMBOL_GPL(css_driver_unregister);

MODULE_LICENSE("GPL");