Fixed MTP to work with TWRP

drivers/s390/cio/Makefile

@@ -0,0 +1,16 @@
+#
+# Makefile for the S/390 common i/o drivers
+#
+
+obj-y += airq.o blacklist.o chsc.o cio.o css.o chp.o idset.o isc.o \
+	fcx.o itcw.o crw.o ccwreq.o
+ccw_device-objs += device.o device_fsm.o device_ops.o
+ccw_device-objs += device_id.o device_pgid.o device_status.o
+obj-y += ccw_device.o cmf.o
+obj-$(CONFIG_CHSC_SCH) += chsc_sch.o
+obj-$(CONFIG_EADM_SCH) += eadm_sch.o
+obj-$(CONFIG_SCM_BUS) += scm.o
+obj-$(CONFIG_CCWGROUP) += ccwgroup.o
+
+qdio-objs := qdio_main.o qdio_thinint.o qdio_debug.o qdio_setup.o
+obj-$(CONFIG_QDIO) += qdio.o

drivers/s390/cio/airq.c

@@ -0,0 +1,275 @@
+/*
+ *    Support for adapter interruptions
+ *
+ *    Copyright IBM Corp. 1999, 2007
+ *    Author(s): Ingo Adlung <adlung@de.ibm.com>
+ *		 Cornelia Huck <cornelia.huck@de.ibm.com>
+ *		 Arnd Bergmann <arndb@de.ibm.com>
+ *		 Peter Oberparleiter <peter.oberparleiter@de.ibm.com>
+ */
+
+#include <linux/init.h>
+#include <linux/irq.h>
+#include <linux/kernel_stat.h>
+#include <linux/module.h>
+#include <linux/mutex.h>
+#include <linux/rculist.h>
+#include <linux/slab.h>
+
+#include <asm/airq.h>
+#include <asm/isc.h>
+
+#include "cio.h"
+#include "cio_debug.h"
+#include "ioasm.h"
+
+static DEFINE_SPINLOCK(airq_lists_lock);
+static struct hlist_head airq_lists[MAX_ISC+1];
+
+/**
+ * register_adapter_interrupt() - register adapter interrupt handler
+ * @airq: pointer to adapter interrupt descriptor
+ *
+ * Returns 0 on success, or -EINVAL.
+ */
+int register_adapter_interrupt(struct airq_struct *airq)
+{
+	char dbf_txt[32];
+
+	if (!airq->handler || airq->isc > MAX_ISC)
+		return -EINVAL;
+	if (!airq->lsi_ptr) {
+		airq->lsi_ptr = kzalloc(1, GFP_KERNEL);
+		if (!airq->lsi_ptr)
+			return -ENOMEM;
+		airq->flags |= AIRQ_PTR_ALLOCATED;
+	}
+	if (!airq->lsi_mask)
+		airq->lsi_mask = 0xff;
+	snprintf(dbf_txt, sizeof(dbf_txt), "rairq:%p", airq);
+	CIO_TRACE_EVENT(4, dbf_txt);
+	isc_register(airq->isc);
+	spin_lock(&airq_lists_lock);
+	hlist_add_head_rcu(&airq->list, &airq_lists[airq->isc]);
+	spin_unlock(&airq_lists_lock);
+	return 0;
+}
+EXPORT_SYMBOL(register_adapter_interrupt);
+
+/**
+ * unregister_adapter_interrupt - unregister adapter interrupt handler
+ * @airq: pointer to adapter interrupt descriptor
+ */
+void unregister_adapter_interrupt(struct airq_struct *airq)
+{
+	char dbf_txt[32];
+
+	if (hlist_unhashed(&airq->list))
+		return;
+	snprintf(dbf_txt, sizeof(dbf_txt), "urairq:%p", airq);
+	CIO_TRACE_EVENT(4, dbf_txt);
+	spin_lock(&airq_lists_lock);
+	hlist_del_rcu(&airq->list);
+	spin_unlock(&airq_lists_lock);
+	synchronize_rcu();
+	isc_unregister(airq->isc);
+	if (airq->flags & AIRQ_PTR_ALLOCATED) {
+		kfree(airq->lsi_ptr);
+		airq->lsi_ptr = NULL;
+		airq->flags &= ~AIRQ_PTR_ALLOCATED;
+	}
+}
+EXPORT_SYMBOL(unregister_adapter_interrupt);
+
+static irqreturn_t do_airq_interrupt(int irq, void *dummy)
+{
+	struct tpi_info *tpi_info;
+	struct airq_struct *airq;
+	struct hlist_head *head;
+
+	set_cpu_flag(CIF_NOHZ_DELAY);
+	tpi_info = (struct tpi_info *) &get_irq_regs()->int_code;
+	head = &airq_lists[tpi_info->isc];
+	rcu_read_lock();
+	hlist_for_each_entry_rcu(airq, head, list)
+		if ((*airq->lsi_ptr & airq->lsi_mask) != 0)
+			airq->handler(airq);
+	rcu_read_unlock();
+
+	return IRQ_HANDLED;
+}
+
+static struct irqaction airq_interrupt = {
+	.name	 = "AIO",
+	.handler = do_airq_interrupt,
+};
+
+void __init init_airq_interrupts(void)
+{
+	irq_set_chip_and_handler(THIN_INTERRUPT,
+				 &dummy_irq_chip, handle_percpu_irq);
+	setup_irq(THIN_INTERRUPT, &airq_interrupt);
+}
+
+/**
+ * airq_iv_create - create an interrupt vector
+ * @bits: number of bits in the interrupt vector
+ * @flags: allocation flags
+ *
+ * Returns a pointer to an interrupt vector structure
+ */
+struct airq_iv *airq_iv_create(unsigned long bits, unsigned long flags)
+{
+	struct airq_iv *iv;
+	unsigned long size;
+
+	iv = kzalloc(sizeof(*iv), GFP_KERNEL);
+	if (!iv)
+		goto out;
+	iv->bits = bits;
+	size = BITS_TO_LONGS(bits) * sizeof(unsigned long);
+	iv->vector = kzalloc(size, GFP_KERNEL);
+	if (!iv->vector)
+		goto out_free;
+	if (flags & AIRQ_IV_ALLOC) {
+		iv->avail = kmalloc(size, GFP_KERNEL);
+		if (!iv->avail)
+			goto out_free;
+		memset(iv->avail, 0xff, size);
+		iv->end = 0;
+	} else
+		iv->end = bits;
+	if (flags & AIRQ_IV_BITLOCK) {
+		iv->bitlock = kzalloc(size, GFP_KERNEL);
+		if (!iv->bitlock)
+			goto out_free;
+	}
+	if (flags & AIRQ_IV_PTR) {
+		size = bits * sizeof(unsigned long);
+		iv->ptr = kzalloc(size, GFP_KERNEL);
+		if (!iv->ptr)
+			goto out_free;
+	}
+	if (flags & AIRQ_IV_DATA) {
+		size = bits * sizeof(unsigned int);
+		iv->data = kzalloc(size, GFP_KERNEL);
+		if (!iv->data)
+			goto out_free;
+	}
+	spin_lock_init(&iv->lock);
+	return iv;
+
+out_free:
+	kfree(iv->ptr);
+	kfree(iv->bitlock);
+	kfree(iv->avail);
+	kfree(iv->vector);
+	kfree(iv);
+out:
+	return NULL;
+}
+EXPORT_SYMBOL(airq_iv_create);
+
+/**
+ * airq_iv_release - release an interrupt vector
+ * @iv: pointer to interrupt vector structure
+ */
+void airq_iv_release(struct airq_iv *iv)
+{
+	kfree(iv->data);
+	kfree(iv->ptr);
+	kfree(iv->bitlock);
+	kfree(iv->vector);
+	kfree(iv->avail);
+	kfree(iv);
+}
+EXPORT_SYMBOL(airq_iv_release);
+
+/**
+ * airq_iv_alloc - allocate irq bits from an interrupt vector
+ * @iv: pointer to an interrupt vector structure
+ * @num: number of consecutive irq bits to allocate
+ *
+ * Returns the bit number of the first irq in the allocated block of irqs,
+ * or -1UL if no bit is available or the AIRQ_IV_ALLOC flag has not been
+ * specified
+ */
+unsigned long airq_iv_alloc(struct airq_iv *iv, unsigned long num)
+{
+	unsigned long bit, i, flags;
+
+	if (!iv->avail || num == 0)
+		return -1UL;
+	spin_lock_irqsave(&iv->lock, flags);
+	bit = find_first_bit_inv(iv->avail, iv->bits);
+	while (bit + num <= iv->bits) {
+		for (i = 1; i < num; i++)
+			if (!test_bit_inv(bit + i, iv->avail))
+				break;
+		if (i >= num) {
+			/* Found a suitable block of irqs */
+			for (i = 0; i < num; i++)
+				clear_bit_inv(bit + i, iv->avail);
+			if (bit + num >= iv->end)
+				iv->end = bit + num + 1;
+			break;
+		}
+		bit = find_next_bit_inv(iv->avail, iv->bits, bit + i + 1);
+	}
+	if (bit + num > iv->bits)
+		bit = -1UL;
+	spin_unlock_irqrestore(&iv->lock, flags);
+	return bit;
+}
+EXPORT_SYMBOL(airq_iv_alloc);
+
+/**
+ * airq_iv_free - free irq bits of an interrupt vector
+ * @iv: pointer to interrupt vector structure
+ * @bit: number of the first irq bit to free
+ * @num: number of consecutive irq bits to free
+ */
+void airq_iv_free(struct airq_iv *iv, unsigned long bit, unsigned long num)
+{
+	unsigned long i, flags;
+
+	if (!iv->avail || num == 0)
+		return;
+	spin_lock_irqsave(&iv->lock, flags);
+	for (i = 0; i < num; i++) {
+		/* Clear (possibly left over) interrupt bit */
+		clear_bit_inv(bit + i, iv->vector);
+		/* Make the bit positions available again */
+		set_bit_inv(bit + i, iv->avail);
+	}
+	if (bit + num >= iv->end) {
+		/* Find new end of bit-field */
+		while (iv->end > 0 && !test_bit_inv(iv->end - 1, iv->avail))
+			iv->end--;
+	}
+	spin_unlock_irqrestore(&iv->lock, flags);
+}
+EXPORT_SYMBOL(airq_iv_free);
+
+/**
+ * airq_iv_scan - scan interrupt vector for non-zero bits
+ * @iv: pointer to interrupt vector structure
+ * @start: bit number to start the search
+ * @end: bit number to end the search
+ *
+ * Returns the bit number of the next non-zero interrupt bit, or
+ * -1UL if the scan completed without finding any more any non-zero bits.
+ */
+unsigned long airq_iv_scan(struct airq_iv *iv, unsigned long start,
+			   unsigned long end)
+{
+	unsigned long bit;
+
+	/* Find non-zero bit starting from 'ivs->next'. */
+	bit = find_next_bit_inv(iv->vector, end, start);
+	if (bit >= end)
+		return -1UL;
+	clear_bit_inv(bit, iv->vector);
+	return bit;
+}
+EXPORT_SYMBOL(airq_iv_scan);

drivers/s390/cio/blacklist.c

@@ -0,0 +1,421 @@
+/*
+ *   S/390 common I/O routines -- blacklisting of specific devices
+ *
+ *    Copyright IBM Corp. 1999, 2013
+ *    Author(s): Ingo Adlung (adlung@de.ibm.com)
+ *		 Cornelia Huck (cornelia.huck@de.ibm.com)
+ *		 Arnd Bergmann (arndb@de.ibm.com)
+ */
+
+#define KMSG_COMPONENT "cio"
+#define pr_fmt(fmt) KMSG_COMPONENT ": " fmt
+
+#include <linux/init.h>
+#include <linux/vmalloc.h>
+#include <linux/proc_fs.h>
+#include <linux/seq_file.h>
+#include <linux/ctype.h>
+#include <linux/device.h>
+
+#include <asm/uaccess.h>
+#include <asm/cio.h>
+#include <asm/ipl.h>
+
+#include "blacklist.h"
+#include "cio.h"
+#include "cio_debug.h"
+#include "css.h"
+#include "device.h"
+
+/*
+ * "Blacklisting" of certain devices:
+ * Device numbers given in the commandline as cio_ignore=... won't be known
+ * to Linux.
+ *
+ * These can be single devices or ranges of devices
+ */
+
+/* 65536 bits for each set to indicate if a devno is blacklisted or not */
+#define __BL_DEV_WORDS ((__MAX_SUBCHANNEL + (8*sizeof(long) - 1)) / \
+			 (8*sizeof(long)))
+static unsigned long bl_dev[__MAX_SSID + 1][__BL_DEV_WORDS];
+typedef enum {add, free} range_action;
+
+/*
+ * Function: blacklist_range
+ * (Un-)blacklist the devices from-to
+ */
+static int blacklist_range(range_action action, unsigned int from_ssid,
+			   unsigned int to_ssid, unsigned int from,
+			   unsigned int to, int msgtrigger)
+{
+	if ((from_ssid > to_ssid) || ((from_ssid == to_ssid) && (from > to))) {
+		if (msgtrigger)
+			pr_warning("0.%x.%04x to 0.%x.%04x is not a valid "
+				   "range for cio_ignore\n", from_ssid, from,
+				   to_ssid, to);
+
+		return 1;
+	}
+
+	while ((from_ssid < to_ssid) || ((from_ssid == to_ssid) &&
+	       (from <= to))) {
+		if (action == add)
+			set_bit(from, bl_dev[from_ssid]);
+		else
+			clear_bit(from, bl_dev[from_ssid]);
+		from++;
+		if (from > __MAX_SUBCHANNEL) {
+			from_ssid++;
+			from = 0;
+		}
+	}
+
+	return 0;
+}
+
+static int pure_hex(char **cp, unsigned int *val, int min_digit,
+		    int max_digit, int max_val)
+{
+	int diff;
+
+	diff = 0;
+	*val = 0;
+
+	while (diff <= max_digit) {
+		int value = hex_to_bin(**cp);
+
+		if (value < 0)
+			break;
+		*val = *val * 16 + value;
+		(*cp)++;
+		diff++;
+	}
+
+	if ((diff < min_digit) || (diff > max_digit) || (*val > max_val))
+		return 1;
+
+	return 0;
+}
+
+static int parse_busid(char *str, unsigned int *cssid, unsigned int *ssid,
+		       unsigned int *devno, int msgtrigger)
+{
+	char *str_work;
+	int val, rc, ret;
+
+	rc = 1;
+
+	if (*str == '\0')
+		goto out;
+
+	/* old style */
+	str_work = str;
+	val = simple_strtoul(str, &str_work, 16);
+
+	if (*str_work == '\0') {
+		if (val <= __MAX_SUBCHANNEL) {
+			*devno = val;
+			*ssid = 0;
+			*cssid = 0;
+			rc = 0;
+		}
+		goto out;
+	}
+
+	/* new style */
+	str_work = str;
+	ret = pure_hex(&str_work, cssid, 1, 2, __MAX_CSSID);
+	if (ret || (str_work[0] != '.'))
+		goto out;
+	str_work++;
+	ret = pure_hex(&str_work, ssid, 1, 1, __MAX_SSID);
+	if (ret || (str_work[0] != '.'))
+		goto out;
+	str_work++;
+	ret = pure_hex(&str_work, devno, 4, 4, __MAX_SUBCHANNEL);
+	if (ret || (str_work[0] != '\0'))
+		goto out;
+
+	rc = 0;
+out:
+	if (rc && msgtrigger)
+		pr_warning("%s is not a valid device for the cio_ignore "
+			   "kernel parameter\n", str);
+
+	return rc;
+}
+
+static int blacklist_parse_parameters(char *str, range_action action,
+				      int msgtrigger)
+{
+	unsigned int from_cssid, to_cssid, from_ssid, to_ssid, from, to;
+	int rc, totalrc;
+	char *parm;
+	range_action ra;
+
+	totalrc = 0;
+
+	while ((parm = strsep(&str, ","))) {
+		rc = 0;
+		ra = action;
+		if (*parm == '!') {
+			if (ra == add)
+				ra = free;
+			else
+				ra = add;
+			parm++;
+		}
+		if (strcmp(parm, "all") == 0) {
+			from_cssid = 0;
+			from_ssid = 0;
+			from = 0;
+			to_cssid = __MAX_CSSID;
+			to_ssid = __MAX_SSID;
+			to = __MAX_SUBCHANNEL;
+		} else if (strcmp(parm, "ipldev") == 0) {
+			if (ipl_info.type == IPL_TYPE_CCW) {
+				from_cssid = 0;
+				from_ssid = ipl_info.data.ccw.dev_id.ssid;
+				from = ipl_info.data.ccw.dev_id.devno;
+			} else if (ipl_info.type == IPL_TYPE_FCP ||
+				   ipl_info.type == IPL_TYPE_FCP_DUMP) {
+				from_cssid = 0;
+				from_ssid = ipl_info.data.fcp.dev_id.ssid;
+				from = ipl_info.data.fcp.dev_id.devno;
+			} else {
+				continue;
+			}
+			to_cssid = from_cssid;
+			to_ssid = from_ssid;
+			to = from;
+		} else if (strcmp(parm, "condev") == 0) {
+			if (console_devno == -1)
+				continue;
+
+			from_cssid = to_cssid = 0;
+			from_ssid = to_ssid = 0;
+			from = to = console_devno;
+		} else {
+			rc = parse_busid(strsep(&parm, "-"), &from_cssid,
+					 &from_ssid, &from, msgtrigger);
+			if (!rc) {
+				if (parm != NULL)
+					rc = parse_busid(parm, &to_cssid,
+							 &to_ssid, &to,
+							 msgtrigger);
+				else {
+					to_cssid = from_cssid;
+					to_ssid = from_ssid;
+					to = from;
+				}
+			}
+		}
+		if (!rc) {
+			rc = blacklist_range(ra, from_ssid, to_ssid, from, to,
+					     msgtrigger);
+			if (rc)
+				totalrc = -EINVAL;
+		} else
+			totalrc = -EINVAL;
+	}
+
+	return totalrc;
+}
+
+static int __init
+blacklist_setup (char *str)
+{
+	CIO_MSG_EVENT(6, "Reading blacklist parameters\n");
+	if (blacklist_parse_parameters(str, add, 1))
+		return 0;
+	return 1;
+}
+
+__setup ("cio_ignore=", blacklist_setup);
+
+/* Checking if devices are blacklisted */
+
+/*
+ * Function: is_blacklisted
+ * Returns 1 if the given devicenumber can be found in the blacklist,
+ * otherwise 0.
+ * Used by validate_subchannel()
+ */
+int
+is_blacklisted (int ssid, int devno)
+{
+	return test_bit (devno, bl_dev[ssid]);
+}
+
+#ifdef CONFIG_PROC_FS
+/*
+ * Function: blacklist_parse_proc_parameters
+ * parse the stuff which is piped to /proc/cio_ignore
+ */
+static int blacklist_parse_proc_parameters(char *buf)
+{
+	int rc;
+	char *parm;
+
+	parm = strsep(&buf, " ");
+
+	if (strcmp("free", parm) == 0) {
+		rc = blacklist_parse_parameters(buf, free, 0);
+		css_schedule_eval_all_unreg(0);
+	} else if (strcmp("add", parm) == 0)
+		rc = blacklist_parse_parameters(buf, add, 0);
+	else if (strcmp("purge", parm) == 0)
+		return ccw_purge_blacklisted();
+	else
+		return -EINVAL;
+
+
+	return rc;
+}
+
+/* Iterator struct for all devices. */
+struct ccwdev_iter {
+	int devno;
+	int ssid;
+	int in_range;
+};
+
+static void *
+cio_ignore_proc_seq_start(struct seq_file *s, loff_t *offset)
+{
+	struct ccwdev_iter *iter = s->private;
+
+	if (*offset >= (__MAX_SUBCHANNEL + 1) * (__MAX_SSID + 1))
+		return NULL;
+	memset(iter, 0, sizeof(*iter));
+	iter->ssid = *offset / (__MAX_SUBCHANNEL + 1);
+	iter->devno = *offset % (__MAX_SUBCHANNEL + 1);
+	return iter;
+}
+
+static void
+cio_ignore_proc_seq_stop(struct seq_file *s, void *it)
+{
+}
+
+static void *
+cio_ignore_proc_seq_next(struct seq_file *s, void *it, loff_t *offset)
+{
+	struct ccwdev_iter *iter;
+
+	if (*offset >= (__MAX_SUBCHANNEL + 1) * (__MAX_SSID + 1))
+		return NULL;
+	iter = it;
+	if (iter->devno == __MAX_SUBCHANNEL) {
+		iter->devno = 0;
+		iter->ssid++;
+		if (iter->ssid > __MAX_SSID)
+			return NULL;
+	} else
+		iter->devno++;
+	(*offset)++;
+	return iter;
+}
+
+static int
+cio_ignore_proc_seq_show(struct seq_file *s, void *it)
+{
+	struct ccwdev_iter *iter;
+
+	iter = it;
+	if (!is_blacklisted(iter->ssid, iter->devno))
+		/* Not blacklisted, nothing to output. */
+		return 0;
+	if (!iter->in_range) {
+		/* First device in range. */
+		if ((iter->devno == __MAX_SUBCHANNEL) ||
+		    !is_blacklisted(iter->ssid, iter->devno + 1))
+			/* Singular device. */
+			return seq_printf(s, "0.%x.%04x\n",
+					  iter->ssid, iter->devno);
+		iter->in_range = 1;
+		return seq_printf(s, "0.%x.%04x-", iter->ssid, iter->devno);
+	}
+	if ((iter->devno == __MAX_SUBCHANNEL) ||
+	    !is_blacklisted(iter->ssid, iter->devno + 1)) {
+		/* Last device in range. */
+		iter->in_range = 0;
+		return seq_printf(s, "0.%x.%04x\n", iter->ssid, iter->devno);
+	}
+	return 0;
+}
+
+static ssize_t
+cio_ignore_write(struct file *file, const char __user *user_buf,
+		 size_t user_len, loff_t *offset)
+{
+	char *buf;
+	ssize_t rc, ret, i;
+
+	if (*offset)
+		return -EINVAL;
+	if (user_len > 65536)
+		user_len = 65536;
+	buf = vzalloc(user_len + 1); /* maybe better use the stack? */
+	if (buf == NULL)
+		return -ENOMEM;
+
+	if (strncpy_from_user (buf, user_buf, user_len) < 0) {
+		rc = -EFAULT;
+		goto out_free;
+	}
+
+	i = user_len - 1;
+	while ((i >= 0) && (isspace(buf[i]) || (buf[i] == 0))) {
+		buf[i] = '\0';
+		i--;
+	}
+	ret = blacklist_parse_proc_parameters(buf);
+	if (ret)
+		rc = ret;
+	else
+		rc = user_len;
+
+out_free:
+	vfree (buf);
+	return rc;
+}
+
+static const struct seq_operations cio_ignore_proc_seq_ops = {
+	.start = cio_ignore_proc_seq_start,
+	.stop  = cio_ignore_proc_seq_stop,
+	.next  = cio_ignore_proc_seq_next,
+	.show  = cio_ignore_proc_seq_show,
+};
+
+static int
+cio_ignore_proc_open(struct inode *inode, struct file *file)
+{
+	return seq_open_private(file, &cio_ignore_proc_seq_ops,
+				sizeof(struct ccwdev_iter));
+}
+
+static const struct file_operations cio_ignore_proc_fops = {
+	.open    = cio_ignore_proc_open,
+	.read    = seq_read,
+	.llseek  = seq_lseek,
+	.release = seq_release_private,
+	.write   = cio_ignore_write,
+};
+
+static int
+cio_ignore_proc_init (void)
+{
+	struct proc_dir_entry *entry;
+
+	entry = proc_create("cio_ignore", S_IFREG | S_IRUGO | S_IWUSR, NULL,
+			    &cio_ignore_proc_fops);
+	if (!entry)
+		return -ENOENT;
+	return 0;
+}
+
+__initcall (cio_ignore_proc_init);
+
+#endif /* CONFIG_PROC_FS */

drivers/s390/cio/blacklist.h

@@ -0,0 +1,6 @@
+#ifndef S390_BLACKLIST_H
+#define S390_BLACKLIST_H
+
+extern int is_blacklisted (int ssid, int devno);
+
+#endif

drivers/s390/cio/ccwgroup.c

@@ -0,0 +1,641 @@
+/*
+ *  bus driver for ccwgroup
+ *
+ *  Copyright IBM Corp. 2002, 2012
+ *
+ *  Author(s): Arnd Bergmann (arndb@de.ibm.com)
+ *	       Cornelia Huck (cornelia.huck@de.ibm.com)
+ */
+#include <linux/module.h>
+#include <linux/errno.h>
+#include <linux/slab.h>
+#include <linux/list.h>
+#include <linux/device.h>
+#include <linux/init.h>
+#include <linux/ctype.h>
+#include <linux/dcache.h>
+
+#include <asm/cio.h>
+#include <asm/ccwdev.h>
+#include <asm/ccwgroup.h>
+
+#include "device.h"
+
+#define CCW_BUS_ID_SIZE		10
+
+/* In Linux 2.4, we had a channel device layer called "chandev"
+ * that did all sorts of obscure stuff for networking devices.
+ * This is another driver that serves as a replacement for just
+ * one of its functions, namely the translation of single subchannels
+ * to devices that use multiple subchannels.
+ */
+
+static struct bus_type ccwgroup_bus_type;
+
+static void __ccwgroup_remove_symlinks(struct ccwgroup_device *gdev)
+{
+	int i;
+	char str[8];
+
+	for (i = 0; i < gdev->count; i++) {
+		sprintf(str, "cdev%d", i);
+		sysfs_remove_link(&gdev->dev.kobj, str);
+		sysfs_remove_link(&gdev->cdev[i]->dev.kobj, "group_device");
+	}
+}
+
+/*
+ * Remove references from ccw devices to ccw group device and from
+ * ccw group device to ccw devices.
+ */
+static void __ccwgroup_remove_cdev_refs(struct ccwgroup_device *gdev)
+{
+	struct ccw_device *cdev;
+	int i;
+
+	for (i = 0; i < gdev->count; i++) {
+		cdev = gdev->cdev[i];
+		if (!cdev)
+			continue;
+		spin_lock_irq(cdev->ccwlock);
+		dev_set_drvdata(&cdev->dev, NULL);
+		spin_unlock_irq(cdev->ccwlock);
+		gdev->cdev[i] = NULL;
+		put_device(&cdev->dev);
+	}
+}
+
+/**
+ * ccwgroup_set_online() - enable a ccwgroup device
+ * @gdev: target ccwgroup device
+ *
+ * This function attempts to put the ccwgroup device into the online state.
+ * Returns:
+ *  %0 on success and a negative error value on failure.
+ */
+int ccwgroup_set_online(struct ccwgroup_device *gdev)
+{
+	struct ccwgroup_driver *gdrv = to_ccwgroupdrv(gdev->dev.driver);
+	int ret = -EINVAL;
+
+	if (atomic_cmpxchg(&gdev->onoff, 0, 1) != 0)
+		return -EAGAIN;
+	if (gdev->state == CCWGROUP_ONLINE)
+		goto out;
+	if (gdrv->set_online)
+		ret = gdrv->set_online(gdev);
+	if (ret)
+		goto out;
+
+	gdev->state = CCWGROUP_ONLINE;
+out:
+	atomic_set(&gdev->onoff, 0);
+	return ret;
+}
+EXPORT_SYMBOL(ccwgroup_set_online);
+
+/**
+ * ccwgroup_set_offline() - disable a ccwgroup device
+ * @gdev: target ccwgroup device
+ *
+ * This function attempts to put the ccwgroup device into the offline state.
+ * Returns:
+ *  %0 on success and a negative error value on failure.
+ */
+int ccwgroup_set_offline(struct ccwgroup_device *gdev)
+{
+	struct ccwgroup_driver *gdrv = to_ccwgroupdrv(gdev->dev.driver);
+	int ret = -EINVAL;
+
+	if (atomic_cmpxchg(&gdev->onoff, 0, 1) != 0)
+		return -EAGAIN;
+	if (gdev->state == CCWGROUP_OFFLINE)
+		goto out;
+	if (gdrv->set_offline)
+		ret = gdrv->set_offline(gdev);
+	if (ret)
+		goto out;
+
+	gdev->state = CCWGROUP_OFFLINE;
+out:
+	atomic_set(&gdev->onoff, 0);
+	return ret;
+}
+EXPORT_SYMBOL(ccwgroup_set_offline);
+
+static ssize_t ccwgroup_online_store(struct device *dev,
+				     struct device_attribute *attr,
+				     const char *buf, size_t count)
+{
+	struct ccwgroup_device *gdev = to_ccwgroupdev(dev);
+	unsigned long value;
+	int ret;
+
+	device_lock(dev);
+	if (!dev->driver) {
+		ret = -EINVAL;
+		goto out;
+	}
+
+	ret = kstrtoul(buf, 0, &value);
+	if (ret)
+		goto out;
+
+	if (value == 1)
+		ret = ccwgroup_set_online(gdev);
+	else if (value == 0)
+		ret = ccwgroup_set_offline(gdev);
+	else
+		ret = -EINVAL;
+out:
+	device_unlock(dev);
+	return (ret == 0) ? count : ret;
+}
+
+static ssize_t ccwgroup_online_show(struct device *dev,
+				    struct device_attribute *attr,
+				    char *buf)
+{
+	struct ccwgroup_device *gdev = to_ccwgroupdev(dev);
+	int online;
+
+	online = (gdev->state == CCWGROUP_ONLINE) ? 1 : 0;
+
+	return scnprintf(buf, PAGE_SIZE, "%d\n", online);
+}
+
+/*
+ * Provide an 'ungroup' attribute so the user can remove group devices no
+ * longer needed or accidentially created. Saves memory :)
+ */
+static void ccwgroup_ungroup(struct ccwgroup_device *gdev)
+{
+	mutex_lock(&gdev->reg_mutex);
+	if (device_is_registered(&gdev->dev)) {
+		__ccwgroup_remove_symlinks(gdev);
+		device_unregister(&gdev->dev);
+		__ccwgroup_remove_cdev_refs(gdev);
+	}
+	mutex_unlock(&gdev->reg_mutex);
+}
+
+static ssize_t ccwgroup_ungroup_store(struct device *dev,
+				      struct device_attribute *attr,
+				      const char *buf, size_t count)
+{
+	struct ccwgroup_device *gdev = to_ccwgroupdev(dev);
+	int rc = 0;
+
+	/* Prevent concurrent online/offline processing and ungrouping. */
+	if (atomic_cmpxchg(&gdev->onoff, 0, 1) != 0)
+		return -EAGAIN;
+	if (gdev->state != CCWGROUP_OFFLINE) {
+		rc = -EINVAL;
+		goto out;
+	}
+
+	if (device_remove_file_self(dev, attr))
+		ccwgroup_ungroup(gdev);
+	else
+		rc = -ENODEV;
+out:
+	if (rc) {
+		/* Release onoff "lock" when ungrouping failed. */
+		atomic_set(&gdev->onoff, 0);
+		return rc;
+	}
+	return count;
+}
+static DEVICE_ATTR(ungroup, 0200, NULL, ccwgroup_ungroup_store);
+static DEVICE_ATTR(online, 0644, ccwgroup_online_show, ccwgroup_online_store);
+
+static struct attribute *ccwgroup_attrs[] = {
+	&dev_attr_online.attr,
+	&dev_attr_ungroup.attr,
+	NULL,
+};
+static struct attribute_group ccwgroup_attr_group = {
+	.attrs = ccwgroup_attrs,
+};
+static const struct attribute_group *ccwgroup_attr_groups[] = {
+	&ccwgroup_attr_group,
+	NULL,
+};
+
+static void ccwgroup_ungroup_workfn(struct work_struct *work)
+{
+	struct ccwgroup_device *gdev =
+		container_of(work, struct ccwgroup_device, ungroup_work);
+
+	ccwgroup_ungroup(gdev);
+	put_device(&gdev->dev);
+}
+
+static void ccwgroup_release(struct device *dev)
+{
+	kfree(to_ccwgroupdev(dev));
+}
+
+static int __ccwgroup_create_symlinks(struct ccwgroup_device *gdev)
+{
+	char str[8];
+	int i, rc;
+
+	for (i = 0; i < gdev->count; i++) {
+		rc = sysfs_create_link(&gdev->cdev[i]->dev.kobj,
+				       &gdev->dev.kobj, "group_device");
+		if (rc) {
+			for (--i; i >= 0; i--)
+				sysfs_remove_link(&gdev->cdev[i]->dev.kobj,
+						  "group_device");
+			return rc;
+		}
+	}
+	for (i = 0; i < gdev->count; i++) {
+		sprintf(str, "cdev%d", i);
+		rc = sysfs_create_link(&gdev->dev.kobj,
+				       &gdev->cdev[i]->dev.kobj, str);
+		if (rc) {
+			for (--i; i >= 0; i--) {
+				sprintf(str, "cdev%d", i);
+				sysfs_remove_link(&gdev->dev.kobj, str);
+			}
+			for (i = 0; i < gdev->count; i++)
+				sysfs_remove_link(&gdev->cdev[i]->dev.kobj,
+						  "group_device");
+			return rc;
+		}
+	}
+	return 0;
+}
+
+static int __get_next_id(const char **buf, struct ccw_dev_id *id)
+{
+	unsigned int cssid, ssid, devno;
+	int ret = 0, len;
+	char *start, *end;
+
+	start = (char *)*buf;
+	end = strchr(start, ',');
+	if (!end) {
+		/* Last entry. Strip trailing newline, if applicable. */
+		end = strchr(start, '\n');
+		if (end)
+			*end = '\0';
+		len = strlen(start) + 1;
+	} else {
+		len = end - start + 1;
+		end++;
+	}
+	if (len <= CCW_BUS_ID_SIZE) {
+		if (sscanf(start, "%2x.%1x.%04x", &cssid, &ssid, &devno) != 3)
+			ret = -EINVAL;
+	} else
+		ret = -EINVAL;
+
+	if (!ret) {
+		id->ssid = ssid;
+		id->devno = devno;
+	}
+	*buf = end;
+	return ret;
+}
+
+/**
+ * ccwgroup_create_dev() - create and register a ccw group device
+ * @parent: parent device for the new device
+ * @gdrv: driver for the new group device
+ * @num_devices: number of slave devices
+ * @buf: buffer containing comma separated bus ids of slave devices
+ *
+ * Create and register a new ccw group device as a child of @parent. Slave
+ * devices are obtained from the list of bus ids given in @buf.
+ * Returns:
+ *  %0 on success and an error code on failure.
+ * Context:
+ *  non-atomic
+ */
+int ccwgroup_create_dev(struct device *parent, struct ccwgroup_driver *gdrv,
+			int num_devices, const char *buf)
+{
+	struct ccwgroup_device *gdev;
+	struct ccw_dev_id dev_id;
+	int rc, i;
+
+	gdev = kzalloc(sizeof(*gdev) + num_devices * sizeof(gdev->cdev[0]),
+		       GFP_KERNEL);
+	if (!gdev)
+		return -ENOMEM;
+
+	atomic_set(&gdev->onoff, 0);
+	mutex_init(&gdev->reg_mutex);
+	mutex_lock(&gdev->reg_mutex);
+	INIT_WORK(&gdev->ungroup_work, ccwgroup_ungroup_workfn);
+	gdev->count = num_devices;
+	gdev->dev.bus = &ccwgroup_bus_type;
+	gdev->dev.parent = parent;
+	gdev->dev.release = ccwgroup_release;
+	device_initialize(&gdev->dev);
+
+	for (i = 0; i < num_devices && buf; i++) {
+		rc = __get_next_id(&buf, &dev_id);
+		if (rc != 0)
+			goto error;
+		gdev->cdev[i] = get_ccwdev_by_dev_id(&dev_id);
+		/*
+		 * All devices have to be of the same type in
+		 * order to be grouped.
+		 */
+		if (!gdev->cdev[i] || !gdev->cdev[i]->drv ||
+		    gdev->cdev[i]->drv != gdev->cdev[0]->drv ||
+		    gdev->cdev[i]->id.driver_info !=
+		    gdev->cdev[0]->id.driver_info) {
+			rc = -EINVAL;
+			goto error;
+		}
+		/* Don't allow a device to belong to more than one group. */
+		spin_lock_irq(gdev->cdev[i]->ccwlock);
+		if (dev_get_drvdata(&gdev->cdev[i]->dev)) {
+			spin_unlock_irq(gdev->cdev[i]->ccwlock);
+			rc = -EINVAL;
+			goto error;
+		}
+		dev_set_drvdata(&gdev->cdev[i]->dev, gdev);
+		spin_unlock_irq(gdev->cdev[i]->ccwlock);
+	}
+	/* Check for sufficient number of bus ids. */
+	if (i < num_devices) {
+		rc = -EINVAL;
+		goto error;
+	}
+	/* Check for trailing stuff. */
+	if (i == num_devices && strlen(buf) > 0) {
+		rc = -EINVAL;
+		goto error;
+	}
+
+	dev_set_name(&gdev->dev, "%s", dev_name(&gdev->cdev[0]->dev));
+	gdev->dev.groups = ccwgroup_attr_groups;
+
+	if (gdrv) {
+		gdev->dev.driver = &gdrv->driver;
+		rc = gdrv->setup ? gdrv->setup(gdev) : 0;
+		if (rc)
+			goto error;
+	}
+	rc = device_add(&gdev->dev);
+	if (rc)
+		goto error;
+	rc = __ccwgroup_create_symlinks(gdev);
+	if (rc) {
+		device_del(&gdev->dev);
+		goto error;
+	}
+	mutex_unlock(&gdev->reg_mutex);
+	return 0;
+error:
+	for (i = 0; i < num_devices; i++)
+		if (gdev->cdev[i]) {
+			spin_lock_irq(gdev->cdev[i]->ccwlock);
+			if (dev_get_drvdata(&gdev->cdev[i]->dev) == gdev)
+				dev_set_drvdata(&gdev->cdev[i]->dev, NULL);
+			spin_unlock_irq(gdev->cdev[i]->ccwlock);
+			put_device(&gdev->cdev[i]->dev);
+			gdev->cdev[i] = NULL;
+		}
+	mutex_unlock(&gdev->reg_mutex);
+	put_device(&gdev->dev);
+	return rc;
+}
+EXPORT_SYMBOL(ccwgroup_create_dev);
+
+static int ccwgroup_notifier(struct notifier_block *nb, unsigned long action,
+			     void *data)
+{
+	struct ccwgroup_device *gdev = to_ccwgroupdev(data);
+
+	if (action == BUS_NOTIFY_UNBIND_DRIVER) {
+		get_device(&gdev->dev);
+		schedule_work(&gdev->ungroup_work);
+	}
+
+	return NOTIFY_OK;
+}
+
+static struct notifier_block ccwgroup_nb = {
+	.notifier_call = ccwgroup_notifier
+};
+
+static int __init init_ccwgroup(void)
+{
+	int ret;
+
+	ret = bus_register(&ccwgroup_bus_type);
+	if (ret)
+		return ret;
+
+	ret = bus_register_notifier(&ccwgroup_bus_type, &ccwgroup_nb);
+	if (ret)
+		bus_unregister(&ccwgroup_bus_type);
+
+	return ret;
+}
+
+static void __exit cleanup_ccwgroup(void)
+{
+	bus_unregister_notifier(&ccwgroup_bus_type, &ccwgroup_nb);
+	bus_unregister(&ccwgroup_bus_type);
+}
+
+module_init(init_ccwgroup);
+module_exit(cleanup_ccwgroup);
+
+/************************** driver stuff ******************************/
+
+static int ccwgroup_remove(struct device *dev)
+{
+	struct ccwgroup_device *gdev = to_ccwgroupdev(dev);
+	struct ccwgroup_driver *gdrv = to_ccwgroupdrv(dev->driver);
+
+	if (!dev->driver)
+		return 0;
+	if (gdrv->remove)
+		gdrv->remove(gdev);
+
+	return 0;
+}
+
+static void ccwgroup_shutdown(struct device *dev)
+{
+	struct ccwgroup_device *gdev = to_ccwgroupdev(dev);
+	struct ccwgroup_driver *gdrv = to_ccwgroupdrv(dev->driver);
+
+	if (!dev->driver)
+		return;
+	if (gdrv->shutdown)
+		gdrv->shutdown(gdev);
+}
+
+static int ccwgroup_pm_prepare(struct device *dev)
+{
+	struct ccwgroup_device *gdev = to_ccwgroupdev(dev);
+	struct ccwgroup_driver *gdrv = to_ccwgroupdrv(gdev->dev.driver);
+
+	/* Fail while device is being set online/offline. */
+	if (atomic_read(&gdev->onoff))
+		return -EAGAIN;
+
+	if (!gdev->dev.driver || gdev->state != CCWGROUP_ONLINE)
+		return 0;
+
+	return gdrv->prepare ? gdrv->prepare(gdev) : 0;
+}
+
+static void ccwgroup_pm_complete(struct device *dev)
+{
+	struct ccwgroup_device *gdev = to_ccwgroupdev(dev);
+	struct ccwgroup_driver *gdrv = to_ccwgroupdrv(dev->driver);
+
+	if (!gdev->dev.driver || gdev->state != CCWGROUP_ONLINE)
+		return;
+
+	if (gdrv->complete)
+		gdrv->complete(gdev);
+}
+
+static int ccwgroup_pm_freeze(struct device *dev)
+{
+	struct ccwgroup_device *gdev = to_ccwgroupdev(dev);
+	struct ccwgroup_driver *gdrv = to_ccwgroupdrv(gdev->dev.driver);
+
+	if (!gdev->dev.driver || gdev->state != CCWGROUP_ONLINE)
+		return 0;
+
+	return gdrv->freeze ? gdrv->freeze(gdev) : 0;
+}
+
+static int ccwgroup_pm_thaw(struct device *dev)
+{
+	struct ccwgroup_device *gdev = to_ccwgroupdev(dev);
+	struct ccwgroup_driver *gdrv = to_ccwgroupdrv(gdev->dev.driver);
+
+	if (!gdev->dev.driver || gdev->state != CCWGROUP_ONLINE)
+		return 0;
+
+	return gdrv->thaw ? gdrv->thaw(gdev) : 0;
+}
+
+static int ccwgroup_pm_restore(struct device *dev)
+{
+	struct ccwgroup_device *gdev = to_ccwgroupdev(dev);
+	struct ccwgroup_driver *gdrv = to_ccwgroupdrv(gdev->dev.driver);
+
+	if (!gdev->dev.driver || gdev->state != CCWGROUP_ONLINE)
+		return 0;
+
+	return gdrv->restore ? gdrv->restore(gdev) : 0;
+}
+
+static const struct dev_pm_ops ccwgroup_pm_ops = {
+	.prepare = ccwgroup_pm_prepare,
+	.complete = ccwgroup_pm_complete,
+	.freeze = ccwgroup_pm_freeze,
+	.thaw = ccwgroup_pm_thaw,
+	.restore = ccwgroup_pm_restore,
+};
+
+static struct bus_type ccwgroup_bus_type = {
+	.name   = "ccwgroup",
+	.remove = ccwgroup_remove,
+	.shutdown = ccwgroup_shutdown,
+	.pm = &ccwgroup_pm_ops,
+};
+
+/**
+ * ccwgroup_driver_register() - register a ccw group driver
+ * @cdriver: driver to be registered
+ *
+ * This function is mainly a wrapper around driver_register().
+ */
+int ccwgroup_driver_register(struct ccwgroup_driver *cdriver)
+{
+	/* register our new driver with the core */
+	cdriver->driver.bus = &ccwgroup_bus_type;
+
+	return driver_register(&cdriver->driver);
+}
+EXPORT_SYMBOL(ccwgroup_driver_register);
+
+static int __ccwgroup_match_all(struct device *dev, void *data)
+{
+	return 1;
+}
+
+/**
+ * ccwgroup_driver_unregister() - deregister a ccw group driver
+ * @cdriver: driver to be deregistered
+ *
+ * This function is mainly a wrapper around driver_unregister().
+ */
+void ccwgroup_driver_unregister(struct ccwgroup_driver *cdriver)
+{
+	struct device *dev;
+
+	/* We don't want ccwgroup devices to live longer than their driver. */
+	while ((dev = driver_find_device(&cdriver->driver, NULL, NULL,
+					 __ccwgroup_match_all))) {
+		struct ccwgroup_device *gdev = to_ccwgroupdev(dev);
+
+		ccwgroup_ungroup(gdev);
+		put_device(dev);
+	}
+	driver_unregister(&cdriver->driver);
+}
+EXPORT_SYMBOL(ccwgroup_driver_unregister);
+
+/**
+ * ccwgroup_probe_ccwdev() - probe function for slave devices
+ * @cdev: ccw device to be probed
+ *
+ * This is a dummy probe function for ccw devices that are slave devices in
+ * a ccw group device.
+ * Returns:
+ *  always %0
+ */
+int ccwgroup_probe_ccwdev(struct ccw_device *cdev)
+{
+	return 0;
+}
+EXPORT_SYMBOL(ccwgroup_probe_ccwdev);
+
+/**
+ * ccwgroup_remove_ccwdev() - remove function for slave devices
+ * @cdev: ccw device to be removed
+ *
+ * This is a remove function for ccw devices that are slave devices in a ccw
+ * group device. It sets the ccw device offline and also deregisters the
+ * embedding ccw group device.
+ */
+void ccwgroup_remove_ccwdev(struct ccw_device *cdev)
+{
+	struct ccwgroup_device *gdev;
+
+	/* Ignore offlining errors, device is gone anyway. */
+	ccw_device_set_offline(cdev);
+	/* If one of its devices is gone, the whole group is done for. */
+	spin_lock_irq(cdev->ccwlock);
+	gdev = dev_get_drvdata(&cdev->dev);
+	if (!gdev) {
+		spin_unlock_irq(cdev->ccwlock);
+		return;
+	}
+	/* Get ccwgroup device reference for local processing. */
+	get_device(&gdev->dev);
+	spin_unlock_irq(cdev->ccwlock);
+	/* Unregister group device. */
+	ccwgroup_ungroup(gdev);
+	/* Release ccwgroup device reference for local processing. */
+	put_device(&gdev->dev);
+}
+EXPORT_SYMBOL(ccwgroup_remove_ccwdev);
+MODULE_LICENSE("GPL");

drivers/s390/cio/ccwreq.c

@@ -0,0 +1,367 @@
+/*
+ *  Handling of internal CCW device requests.
+ *
+ *    Copyright IBM Corp. 2009, 2011
+ *    Author(s): Peter Oberparleiter <peter.oberparleiter@de.ibm.com>
+ */
+
+#define KMSG_COMPONENT "cio"
+#define pr_fmt(fmt) KMSG_COMPONENT ": " fmt
+
+#include <linux/types.h>
+#include <linux/err.h>
+#include <asm/ccwdev.h>
+#include <asm/cio.h>
+
+#include "io_sch.h"
+#include "cio.h"
+#include "device.h"
+#include "cio_debug.h"
+
+/**
+ * lpm_adjust - adjust path mask
+ * @lpm: path mask to adjust
+ * @mask: mask of available paths
+ *
+ * Shift @lpm right until @lpm and @mask have at least one bit in common or
+ * until @lpm is zero. Return the resulting lpm.
+ */
+int lpm_adjust(int lpm, int mask)
+{
+	while (lpm && ((lpm & mask) == 0))
+		lpm >>= 1;
+	return lpm;
+}
+
+/*
+ * Adjust path mask to use next path and reset retry count. Return resulting
+ * path mask.
+ */
+static u16 ccwreq_next_path(struct ccw_device *cdev)
+{
+	struct ccw_request *req = &cdev->private->req;
+
+	if (!req->singlepath) {
+		req->mask = 0;
+		goto out;
+	}
+	req->retries	= req->maxretries;
+	req->mask	= lpm_adjust(req->mask >> 1, req->lpm);
+out:
+	return req->mask;
+}
+
+/*
+ * Clean up device state and report to callback.
+ */
+static void ccwreq_stop(struct ccw_device *cdev, int rc)
+{
+	struct ccw_request *req = &cdev->private->req;
+
+	if (req->done)
+		return;
+	req->done = 1;
+	ccw_device_set_timeout(cdev, 0);
+	memset(&cdev->private->irb, 0, sizeof(struct irb));
+	if (rc && rc != -ENODEV && req->drc)
+		rc = req->drc;
+	req->callback(cdev, req->data, rc);
+}
+
+/*
+ * (Re-)Start the operation until retries and paths are exhausted.
+ */
+static void ccwreq_do(struct ccw_device *cdev)
+{
+	struct ccw_request *req = &cdev->private->req;
+	struct subchannel *sch = to_subchannel(cdev->dev.parent);
+	struct ccw1 *cp = req->cp;
+	int rc = -EACCES;
+
+	while (req->mask) {
+		if (req->retries-- == 0) {
+			/* Retries exhausted, try next path. */
+			ccwreq_next_path(cdev);
+			continue;
+		}
+		/* Perform start function. */
+		memset(&cdev->private->irb, 0, sizeof(struct irb));
+		rc = cio_start(sch, cp, (u8) req->mask);
+		if (rc == 0) {
+			/* I/O started successfully. */
+			ccw_device_set_timeout(cdev, req->timeout);
+			return;
+		}
+		if (rc == -ENODEV) {
+			/* Permanent device error. */
+			break;
+		}
+		if (rc == -EACCES) {
+			/* Permant path error. */
+			ccwreq_next_path(cdev);
+			continue;
+		}
+		/* Temporary improper status. */
+		rc = cio_clear(sch);
+		if (rc)
+			break;
+		return;
+	}
+	ccwreq_stop(cdev, rc);
+}
+
+/**
+ * ccw_request_start - perform I/O request
+ * @cdev: ccw device
+ *
+ * Perform the I/O request specified by cdev->req.
+ */
+void ccw_request_start(struct ccw_device *cdev)
+{
+	struct ccw_request *req = &cdev->private->req;
+
+	if (req->singlepath) {
+		/* Try all paths twice to counter link flapping. */
+		req->mask = 0x8080;
+	} else
+		req->mask = req->lpm;
+
+	req->retries	= req->maxretries;
+	req->mask	= lpm_adjust(req->mask, req->lpm);
+	req->drc	= 0;
+	req->done	= 0;
+	req->cancel	= 0;
+	if (!req->mask)
+		goto out_nopath;
+	ccwreq_do(cdev);
+	return;
+
+out_nopath:
+	ccwreq_stop(cdev, -EACCES);
+}
+
+/**
+ * ccw_request_cancel - cancel running I/O request
+ * @cdev: ccw device
+ *
+ * Cancel the I/O request specified by cdev->req. Return non-zero if request
+ * has already finished, zero otherwise.
+ */
+int ccw_request_cancel(struct ccw_device *cdev)
+{
+	struct subchannel *sch = to_subchannel(cdev->dev.parent);
+	struct ccw_request *req = &cdev->private->req;
+	int rc;
+
+	if (req->done)
+		return 1;
+	req->cancel = 1;
+	rc = cio_clear(sch);
+	if (rc)
+		ccwreq_stop(cdev, rc);
+	return 0;
+}
+
+/*
+ * Return the status of the internal I/O started on the specified ccw device.
+ * Perform BASIC SENSE if required.
+ */
+static enum io_status ccwreq_status(struct ccw_device *cdev, struct irb *lcirb)
+{
+	struct irb *irb = &cdev->private->irb;
+	struct cmd_scsw *scsw = &irb->scsw.cmd;
+	enum uc_todo todo;
+
+	/* Perform BASIC SENSE if needed. */
+	if (ccw_device_accumulate_and_sense(cdev, lcirb))
+		return IO_RUNNING;
+	/* Check for halt/clear interrupt. */
+	if (scsw->fctl & (SCSW_FCTL_HALT_FUNC | SCSW_FCTL_CLEAR_FUNC))
+		return IO_KILLED;
+	/* Check for path error. */
+	if (scsw->cc == 3 || scsw->pno)
+		return IO_PATH_ERROR;
+	/* Handle BASIC SENSE data. */
+	if (irb->esw.esw0.erw.cons) {
+		CIO_TRACE_EVENT(2, "sensedata");
+		CIO_HEX_EVENT(2, &cdev->private->dev_id,
+			      sizeof(struct ccw_dev_id));
+		CIO_HEX_EVENT(2, &cdev->private->irb.ecw, SENSE_MAX_COUNT);
+		/* Check for command reject. */
+		if (irb->ecw[0] & SNS0_CMD_REJECT)
+			return IO_REJECTED;
+		/* Ask the driver what to do */
+		if (cdev->drv && cdev->drv->uc_handler) {
+			todo = cdev->drv->uc_handler(cdev, lcirb);
+			CIO_TRACE_EVENT(2, "uc_response");
+			CIO_HEX_EVENT(2, &todo, sizeof(todo));
+			switch (todo) {
+			case UC_TODO_RETRY:
+				return IO_STATUS_ERROR;
+			case UC_TODO_RETRY_ON_NEW_PATH:
+				return IO_PATH_ERROR;
+			case UC_TODO_STOP:
+				return IO_REJECTED;
+			default:
+				return IO_STATUS_ERROR;
+			}
+		}
+		/* Assume that unexpected SENSE data implies an error. */
+		return IO_STATUS_ERROR;
+	}
+	/* Check for channel errors. */
+	if (scsw->cstat != 0)
+		return IO_STATUS_ERROR;
+	/* Check for device errors. */
+	if (scsw->dstat & ~(DEV_STAT_CHN_END | DEV_STAT_DEV_END))
+		return IO_STATUS_ERROR;
+	/* Check for final state. */
+	if (!(scsw->dstat & DEV_STAT_DEV_END))
+		return IO_RUNNING;
+	/* Check for other improper status. */
+	if (scsw->cc == 1 && (scsw->stctl & SCSW_STCTL_ALERT_STATUS))
+		return IO_STATUS_ERROR;
+	return IO_DONE;
+}
+
+/*
+ * Log ccw request status.
+ */
+static void ccwreq_log_status(struct ccw_device *cdev, enum io_status status)
+{
+	struct ccw_request *req = &cdev->private->req;
+	struct {
+		struct ccw_dev_id dev_id;
+		u16 retries;
+		u8 lpm;
+		u8 status;
+	}  __attribute__ ((packed)) data;
+	data.dev_id	= cdev->private->dev_id;
+	data.retries	= req->retries;
+	data.lpm	= (u8) req->mask;
+	data.status	= (u8) status;
+	CIO_TRACE_EVENT(2, "reqstat");
+	CIO_HEX_EVENT(2, &data, sizeof(data));
+}
+
+/**
+ * ccw_request_handler - interrupt handler for I/O request procedure.
+ * @cdev: ccw device
+ *
+ * Handle interrupt during I/O request procedure.
+ */
+void ccw_request_handler(struct ccw_device *cdev)
+{
+	struct irb *irb = this_cpu_ptr(&cio_irb);
+	struct ccw_request *req = &cdev->private->req;
+	enum io_status status;
+	int rc = -EOPNOTSUPP;
+
+	/* Check status of I/O request. */
+	status = ccwreq_status(cdev, irb);
+	if (req->filter)
+		status = req->filter(cdev, req->data, irb, status);
+	if (status != IO_RUNNING)
+		ccw_device_set_timeout(cdev, 0);
+	if (status != IO_DONE && status != IO_RUNNING)
+		ccwreq_log_status(cdev, status);
+	switch (status) {
+	case IO_DONE:
+		break;
+	case IO_RUNNING:
+		return;
+	case IO_REJECTED:
+		goto err;
+	case IO_PATH_ERROR:
+		goto out_next_path;
+	case IO_STATUS_ERROR:
+		goto out_restart;
+	case IO_KILLED:
+		/* Check if request was cancelled on purpose. */
+		if (req->cancel) {
+			rc = -EIO;
+			goto err;
+		}
+		goto out_restart;
+	}
+	/* Check back with request initiator. */
+	if (!req->check)
+		goto out;
+	switch (req->check(cdev, req->data)) {
+	case 0:
+		break;
+	case -EAGAIN:
+		goto out_restart;
+	case -EACCES:
+		goto out_next_path;
+	default:
+		goto err;
+	}
+out:
+	ccwreq_stop(cdev, 0);
+	return;
+
+out_next_path:
+	/* Try next path and restart I/O. */
+	if (!ccwreq_next_path(cdev)) {
+		rc = -EACCES;
+		goto err;
+	}
+out_restart:
+	/* Restart. */
+	ccwreq_do(cdev);
+	return;
+err:
+	ccwreq_stop(cdev, rc);
+}
+
+
+/**
+ * ccw_request_timeout - timeout handler for I/O request procedure
+ * @cdev: ccw device
+ *
+ * Handle timeout during I/O request procedure.
+ */
+void ccw_request_timeout(struct ccw_device *cdev)
+{
+	struct subchannel *sch = to_subchannel(cdev->dev.parent);
+	struct ccw_request *req = &cdev->private->req;
+	int rc = -ENODEV, chp;
+
+	if (cio_update_schib(sch))
+		goto err;
+
+	for (chp = 0; chp < 8; chp++) {
+		if ((0x80 >> chp) & sch->schib.pmcw.lpum)
+			pr_warning("%s: No interrupt was received within %lus "
+				   "(CS=%02x, DS=%02x, CHPID=%x.%02x)\n",
+				   dev_name(&cdev->dev), req->timeout / HZ,
+				   scsw_cstat(&sch->schib.scsw),
+				   scsw_dstat(&sch->schib.scsw),
+				   sch->schid.cssid,
+				   sch->schib.pmcw.chpid[chp]);
+	}
+
+	if (!ccwreq_next_path(cdev)) {
+		/* set the final return code for this request */
+		req->drc = -ETIME;
+	}
+	rc = cio_clear(sch);
+	if (rc)
+		goto err;
+	return;
+
+err:
+	ccwreq_stop(cdev, rc);
+}
+
+/**
+ * ccw_request_notoper - notoper handler for I/O request procedure
+ * @cdev: ccw device
+ *
+ * Handle notoper during I/O request procedure.
+ */
+void ccw_request_notoper(struct ccw_device *cdev)
+{
+	ccwreq_stop(cdev, -ENODEV);
+}

drivers/s390/cio/chp.c

@@ -0,0 +1,792 @@
+/*
+ *    Copyright IBM Corp. 1999, 2010
+ *    Author(s): Cornelia Huck (cornelia.huck@de.ibm.com)
+ *		 Arnd Bergmann (arndb@de.ibm.com)
+ *		 Peter Oberparleiter <peter.oberparleiter@de.ibm.com>
+ */
+
+#include <linux/bug.h>
+#include <linux/workqueue.h>
+#include <linux/spinlock.h>
+#include <linux/export.h>
+#include <linux/sched.h>
+#include <linux/init.h>
+#include <linux/jiffies.h>
+#include <linux/wait.h>
+#include <linux/mutex.h>
+#include <linux/errno.h>
+#include <linux/slab.h>
+#include <asm/chpid.h>
+#include <asm/sclp.h>
+#include <asm/crw.h>
+
+#include "cio.h"
+#include "css.h"
+#include "ioasm.h"
+#include "cio_debug.h"
+#include "chp.h"
+
+#define to_channelpath(device) container_of(device, struct channel_path, dev)
+#define CHP_INFO_UPDATE_INTERVAL	1*HZ
+
+enum cfg_task_t {
+	cfg_none,
+	cfg_configure,
+	cfg_deconfigure
+};
+
+/* Map for pending configure tasks. */
+static enum cfg_task_t chp_cfg_task[__MAX_CSSID + 1][__MAX_CHPID + 1];
+static DEFINE_MUTEX(cfg_lock);
+static int cfg_busy;
+
+/* Map for channel-path status. */
+static struct sclp_chp_info chp_info;
+static DEFINE_MUTEX(info_lock);
+
+/* Time after which channel-path status may be outdated. */
+static unsigned long chp_info_expires;
+
+/* Workqueue to perform pending configure tasks. */
+static struct workqueue_struct *chp_wq;
+static struct work_struct cfg_work;
+
+/* Wait queue for configure completion events. */
+static wait_queue_head_t cfg_wait_queue;
+
+/* Set vary state for given chpid. */
+static void set_chp_logically_online(struct chp_id chpid, int onoff)
+{
+	chpid_to_chp(chpid)->state = onoff;
+}
+
+/* On success return 0 if channel-path is varied offline, 1 if it is varied
+ * online. Return -ENODEV if channel-path is not registered. */
+int chp_get_status(struct chp_id chpid)
+{
+	return (chpid_to_chp(chpid) ? chpid_to_chp(chpid)->state : -ENODEV);
+}
+
+/**
+ * chp_get_sch_opm - return opm for subchannel
+ * @sch: subchannel
+ *
+ * Calculate and return the operational path mask (opm) based on the chpids
+ * used by the subchannel and the status of the associated channel-paths.
+ */
+u8 chp_get_sch_opm(struct subchannel *sch)
+{
+	struct chp_id chpid;
+	int opm;
+	int i;
+
+	opm = 0;
+	chp_id_init(&chpid);
+	for (i = 0; i < 8; i++) {
+		opm <<= 1;
+		chpid.id = sch->schib.pmcw.chpid[i];
+		if (chp_get_status(chpid) != 0)
+			opm |= 1;
+	}
+	return opm;
+}
+EXPORT_SYMBOL_GPL(chp_get_sch_opm);
+
+/**
+ * chp_is_registered - check if a channel-path is registered
+ * @chpid: channel-path ID
+ *
+ * Return non-zero if a channel-path with the given chpid is registered,
+ * zero otherwise.
+ */
+int chp_is_registered(struct chp_id chpid)
+{
+	return chpid_to_chp(chpid) != NULL;
+}
+
+/*
+ * Function: s390_vary_chpid
+ * Varies the specified chpid online or offline
+ */
+static int s390_vary_chpid(struct chp_id chpid, int on)
+{
+	char dbf_text[15];
+	int status;
+
+	sprintf(dbf_text, on?"varyon%x.%02x":"varyoff%x.%02x", chpid.cssid,
+		chpid.id);
+	CIO_TRACE_EVENT(2, dbf_text);
+
+	status = chp_get_status(chpid);
+	if (!on && !status)
+		return 0;
+
+	set_chp_logically_online(chpid, on);
+	chsc_chp_vary(chpid, on);
+	return 0;
+}
+
+/*
+ * Channel measurement related functions
+ */
+static ssize_t chp_measurement_chars_read(struct file *filp,
+					  struct kobject *kobj,
+					  struct bin_attribute *bin_attr,
+					  char *buf, loff_t off, size_t count)
+{
+	struct channel_path *chp;
+	struct device *device;
+
+	device = container_of(kobj, struct device, kobj);
+	chp = to_channelpath(device);
+	if (!chp->cmg_chars)
+		return 0;
+
+	return memory_read_from_buffer(buf, count, &off,
+				chp->cmg_chars, sizeof(struct cmg_chars));
+}
+
+static struct bin_attribute chp_measurement_chars_attr = {
+	.attr = {
+		.name = "measurement_chars",
+		.mode = S_IRUSR,
+	},
+	.size = sizeof(struct cmg_chars),
+	.read = chp_measurement_chars_read,
+};
+
+static void chp_measurement_copy_block(struct cmg_entry *buf,
+				       struct channel_subsystem *css,
+				       struct chp_id chpid)
+{
+	void *area;
+	struct cmg_entry *entry, reference_buf;
+	int idx;
+
+	if (chpid.id < 128) {
+		area = css->cub_addr1;
+		idx = chpid.id;
+	} else {
+		area = css->cub_addr2;
+		idx = chpid.id - 128;
+	}
+	entry = area + (idx * sizeof(struct cmg_entry));
+	do {
+		memcpy(buf, entry, sizeof(*entry));
+		memcpy(&reference_buf, entry, sizeof(*entry));
+	} while (reference_buf.values[0] != buf->values[0]);
+}
+
+static ssize_t chp_measurement_read(struct file *filp, struct kobject *kobj,
+				    struct bin_attribute *bin_attr,
+				    char *buf, loff_t off, size_t count)
+{
+	struct channel_path *chp;
+	struct channel_subsystem *css;
+	struct device *device;
+	unsigned int size;
+
+	device = container_of(kobj, struct device, kobj);
+	chp = to_channelpath(device);
+	css = to_css(chp->dev.parent);
+
+	size = sizeof(struct cmg_entry);
+
+	/* Only allow single reads. */
+	if (off || count < size)
+		return 0;
+	chp_measurement_copy_block((struct cmg_entry *)buf, css, chp->chpid);
+	count = size;
+	return count;
+}
+
+static struct bin_attribute chp_measurement_attr = {
+	.attr = {
+		.name = "measurement",
+		.mode = S_IRUSR,
+	},
+	.size = sizeof(struct cmg_entry),
+	.read = chp_measurement_read,
+};
+
+void chp_remove_cmg_attr(struct channel_path *chp)
+{
+	device_remove_bin_file(&chp->dev, &chp_measurement_chars_attr);
+	device_remove_bin_file(&chp->dev, &chp_measurement_attr);
+}
+
+int chp_add_cmg_attr(struct channel_path *chp)
+{
+	int ret;
+
+	ret = device_create_bin_file(&chp->dev, &chp_measurement_chars_attr);
+	if (ret)
+		return ret;
+	ret = device_create_bin_file(&chp->dev, &chp_measurement_attr);
+	if (ret)
+		device_remove_bin_file(&chp->dev, &chp_measurement_chars_attr);
+	return ret;
+}
+
+/*
+ * Files for the channel path entries.
+ */
+static ssize_t chp_status_show(struct device *dev,
+			       struct device_attribute *attr, char *buf)
+{
+	struct channel_path *chp = to_channelpath(dev);
+	int status;
+
+	mutex_lock(&chp->lock);
+	status = chp->state;
+	mutex_unlock(&chp->lock);
+
+	return status ? sprintf(buf, "online\n") : sprintf(buf, "offline\n");
+}
+
+static ssize_t chp_status_write(struct device *dev,
+				struct device_attribute *attr,
+				const char *buf, size_t count)
+{
+	struct channel_path *cp = to_channelpath(dev);
+	char cmd[10];
+	int num_args;
+	int error;
+
+	num_args = sscanf(buf, "%5s", cmd);
+	if (!num_args)
+		return count;
+
+	if (!strncasecmp(cmd, "on", 2) || !strcmp(cmd, "1")) {
+		mutex_lock(&cp->lock);
+		error = s390_vary_chpid(cp->chpid, 1);
+		mutex_unlock(&cp->lock);
+	} else if (!strncasecmp(cmd, "off", 3) || !strcmp(cmd, "0")) {
+		mutex_lock(&cp->lock);
+		error = s390_vary_chpid(cp->chpid, 0);
+		mutex_unlock(&cp->lock);
+	} else
+		error = -EINVAL;
+
+	return error < 0 ? error : count;
+}
+
+static DEVICE_ATTR(status, 0644, chp_status_show, chp_status_write);
+
+static ssize_t chp_configure_show(struct device *dev,
+				  struct device_attribute *attr, char *buf)
+{
+	struct channel_path *cp;
+	int status;
+
+	cp = to_channelpath(dev);
+	status = chp_info_get_status(cp->chpid);
+	if (status < 0)
+		return status;
+
+	return snprintf(buf, PAGE_SIZE, "%d\n", status);
+}
+
+static int cfg_wait_idle(void);
+
+static ssize_t chp_configure_write(struct device *dev,
+				   struct device_attribute *attr,
+				   const char *buf, size_t count)
+{
+	struct channel_path *cp;
+	int val;
+	char delim;
+
+	if (sscanf(buf, "%d %c", &val, &delim) != 1)
+		return -EINVAL;
+	if (val != 0 && val != 1)
+		return -EINVAL;
+	cp = to_channelpath(dev);
+	chp_cfg_schedule(cp->chpid, val);
+	cfg_wait_idle();
+
+	return count;
+}
+
+static DEVICE_ATTR(configure, 0644, chp_configure_show, chp_configure_write);
+
+static ssize_t chp_type_show(struct device *dev, struct device_attribute *attr,
+			     char *buf)
+{
+	struct channel_path *chp = to_channelpath(dev);
+	u8 type;
+
+	mutex_lock(&chp->lock);
+	type = chp->desc.desc;
+	mutex_unlock(&chp->lock);
+	return sprintf(buf, "%x\n", type);
+}
+
+static DEVICE_ATTR(type, 0444, chp_type_show, NULL);
+
+static ssize_t chp_cmg_show(struct device *dev, struct device_attribute *attr,
+			    char *buf)
+{
+	struct channel_path *chp = to_channelpath(dev);
+
+	if (!chp)
+		return 0;
+	if (chp->cmg == -1) /* channel measurements not available */
+		return sprintf(buf, "unknown\n");
+	return sprintf(buf, "%x\n", chp->cmg);
+}
+
+static DEVICE_ATTR(cmg, 0444, chp_cmg_show, NULL);
+
+static ssize_t chp_shared_show(struct device *dev,
+			       struct device_attribute *attr, char *buf)
+{
+	struct channel_path *chp = to_channelpath(dev);
+
+	if (!chp)
+		return 0;
+	if (chp->shared == -1) /* channel measurements not available */
+		return sprintf(buf, "unknown\n");
+	return sprintf(buf, "%x\n", chp->shared);
+}
+
+static DEVICE_ATTR(shared, 0444, chp_shared_show, NULL);
+
+static ssize_t chp_chid_show(struct device *dev, struct device_attribute *attr,
+			     char *buf)
+{
+	struct channel_path *chp = to_channelpath(dev);
+	ssize_t rc;
+
+	mutex_lock(&chp->lock);
+	if (chp->desc_fmt1.flags & 0x10)
+		rc = sprintf(buf, "%04x\n", chp->desc_fmt1.chid);
+	else
+		rc = 0;
+	mutex_unlock(&chp->lock);
+
+	return rc;
+}
+static DEVICE_ATTR(chid, 0444, chp_chid_show, NULL);
+
+static ssize_t chp_chid_external_show(struct device *dev,
+				      struct device_attribute *attr, char *buf)
+{
+	struct channel_path *chp = to_channelpath(dev);
+	ssize_t rc;
+
+	mutex_lock(&chp->lock);
+	if (chp->desc_fmt1.flags & 0x10)
+		rc = sprintf(buf, "%x\n", chp->desc_fmt1.flags & 0x8 ? 1 : 0);
+	else
+		rc = 0;
+	mutex_unlock(&chp->lock);
+
+	return rc;
+}
+static DEVICE_ATTR(chid_external, 0444, chp_chid_external_show, NULL);
+
+static struct attribute *chp_attrs[] = {
+	&dev_attr_status.attr,
+	&dev_attr_configure.attr,
+	&dev_attr_type.attr,
+	&dev_attr_cmg.attr,
+	&dev_attr_shared.attr,
+	&dev_attr_chid.attr,
+	&dev_attr_chid_external.attr,
+	NULL,
+};
+static struct attribute_group chp_attr_group = {
+	.attrs = chp_attrs,
+};
+static const struct attribute_group *chp_attr_groups[] = {
+	&chp_attr_group,
+	NULL,
+};
+
+static void chp_release(struct device *dev)
+{
+	struct channel_path *cp;
+
+	cp = to_channelpath(dev);
+	kfree(cp);
+}
+
+/**
+ * chp_update_desc - update channel-path description
+ * @chp - channel-path
+ *
+ * Update the channel-path description of the specified channel-path.
+ * Return zero on success, non-zero otherwise.
+ */
+int chp_update_desc(struct channel_path *chp)
+{
+	int rc;
+
+	rc = chsc_determine_base_channel_path_desc(chp->chpid, &chp->desc);
+	if (rc)
+		return rc;
+
+	rc = chsc_determine_fmt1_channel_path_desc(chp->chpid, &chp->desc_fmt1);
+
+	return rc;
+}
+
+/**
+ * chp_new - register a new channel-path
+ * @chpid - channel-path ID
+ *
+ * Create and register data structure representing new channel-path. Return
+ * zero on success, non-zero otherwise.
+ */
+int chp_new(struct chp_id chpid)
+{
+	struct channel_path *chp;
+	int ret;
+
+	if (chp_is_registered(chpid))
+		return 0;
+	chp = kzalloc(sizeof(struct channel_path), GFP_KERNEL);
+	if (!chp)
+		return -ENOMEM;
+
+	/* fill in status, etc. */
+	chp->chpid = chpid;
+	chp->state = 1;
+	chp->dev.parent = &channel_subsystems[chpid.cssid]->device;
+	chp->dev.groups = chp_attr_groups;
+	chp->dev.release = chp_release;
+	mutex_init(&chp->lock);
+
+	/* Obtain channel path description and fill it in. */
+	ret = chp_update_desc(chp);
+	if (ret)
+		goto out_free;
+	if ((chp->desc.flags & 0x80) == 0) {
+		ret = -ENODEV;
+		goto out_free;
+	}
+	/* Get channel-measurement characteristics. */
+	if (css_chsc_characteristics.scmc && css_chsc_characteristics.secm) {
+		ret = chsc_get_channel_measurement_chars(chp);
+		if (ret)
+			goto out_free;
+	} else {
+		chp->cmg = -1;
+	}
+	dev_set_name(&chp->dev, "chp%x.%02x", chpid.cssid, chpid.id);
+
+	/* make it known to the system */
+	ret = device_register(&chp->dev);
+	if (ret) {
+		CIO_MSG_EVENT(0, "Could not register chp%x.%02x: %d\n",
+			      chpid.cssid, chpid.id, ret);
+		put_device(&chp->dev);
+		goto out;
+	}
+	mutex_lock(&channel_subsystems[chpid.cssid]->mutex);
+	if (channel_subsystems[chpid.cssid]->cm_enabled) {
+		ret = chp_add_cmg_attr(chp);
+		if (ret) {
+			device_unregister(&chp->dev);
+			mutex_unlock(&channel_subsystems[chpid.cssid]->mutex);
+			goto out;
+		}
+	}
+	channel_subsystems[chpid.cssid]->chps[chpid.id] = chp;
+	mutex_unlock(&channel_subsystems[chpid.cssid]->mutex);
+	goto out;
+out_free:
+	kfree(chp);
+out:
+	return ret;
+}
+
+/**
+ * chp_get_chp_desc - return newly allocated channel-path description
+ * @chpid: channel-path ID
+ *
+ * On success return a newly allocated copy of the channel-path description
+ * data associated with the given channel-path ID. Return %NULL on error.
+ */
+struct channel_path_desc *chp_get_chp_desc(struct chp_id chpid)
+{
+	struct channel_path *chp;
+	struct channel_path_desc *desc;
+
+	chp = chpid_to_chp(chpid);
+	if (!chp)
+		return NULL;
+	desc = kmalloc(sizeof(struct channel_path_desc), GFP_KERNEL);
+	if (!desc)
+		return NULL;
+
+	mutex_lock(&chp->lock);
+	memcpy(desc, &chp->desc, sizeof(struct channel_path_desc));
+	mutex_unlock(&chp->lock);
+	return desc;
+}
+
+/**
+ * chp_process_crw - process channel-path status change
+ * @crw0: channel report-word to handler
+ * @crw1: second channel-report word (always NULL)
+ * @overflow: crw overflow indication
+ *
+ * Handle channel-report-words indicating that the status of a channel-path
+ * has changed.
+ */
+static void chp_process_crw(struct crw *crw0, struct crw *crw1,
+			    int overflow)
+{
+	struct chp_id chpid;
+
+	if (overflow) {
+		css_schedule_eval_all();
+		return;
+	}
+	CIO_CRW_EVENT(2, "CRW 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);
+	/*
+	 * Check for solicited machine checks. These are
+	 * created by reset channel path and need not be
+	 * handled here.
+	 */
+	if (crw0->slct) {
+		CIO_CRW_EVENT(2, "solicited machine check for "
+			      "channel path %02X\n", crw0->rsid);
+		return;
+	}
+	chp_id_init(&chpid);
+	chpid.id = crw0->rsid;
+	switch (crw0->erc) {
+	case CRW_ERC_IPARM: /* Path has come. */
+		if (!chp_is_registered(chpid))
+			chp_new(chpid);
+		chsc_chp_online(chpid);
+		break;
+	case CRW_ERC_PERRI: /* Path has gone. */
+	case CRW_ERC_PERRN:
+		chsc_chp_offline(chpid);
+		break;
+	default:
+		CIO_CRW_EVENT(2, "Don't know how to handle erc=%x\n",
+			      crw0->erc);
+	}
+}
+
+int chp_ssd_get_mask(struct chsc_ssd_info *ssd, struct chp_link *link)
+{
+	int i;
+	int mask;
+
+	for (i = 0; i < 8; i++) {
+		mask = 0x80 >> i;
+		if (!(ssd->path_mask & mask))
+			continue;
+		if (!chp_id_is_equal(&ssd->chpid[i], &link->chpid))
+			continue;
+		if ((ssd->fla_valid_mask & mask) &&
+		    ((ssd->fla[i] & link->fla_mask) != link->fla))
+			continue;
+		return mask;
+	}
+	return 0;
+}
+EXPORT_SYMBOL_GPL(chp_ssd_get_mask);
+
+static inline int info_bit_num(struct chp_id id)
+{
+	return id.id + id.cssid * (__MAX_CHPID + 1);
+}
+
+/* Force chp_info refresh on next call to info_validate(). */
+static void info_expire(void)
+{
+	mutex_lock(&info_lock);
+	chp_info_expires = jiffies - 1;
+	mutex_unlock(&info_lock);
+}
+
+/* Ensure that chp_info is up-to-date. */
+static int info_update(void)
+{
+	int rc;
+
+	mutex_lock(&info_lock);
+	rc = 0;
+	if (time_after(jiffies, chp_info_expires)) {
+		/* Data is too old, update. */
+		rc = sclp_chp_read_info(&chp_info);
+		chp_info_expires = jiffies + CHP_INFO_UPDATE_INTERVAL ;
+	}
+	mutex_unlock(&info_lock);
+
+	return rc;
+}
+
+/**
+ * chp_info_get_status - retrieve configure status of a channel-path
+ * @chpid: channel-path ID
+ *
+ * On success, return 0 for standby, 1 for configured, 2 for reserved,
+ * 3 for not recognized. Return negative error code on error.
+ */
+int chp_info_get_status(struct chp_id chpid)
+{
+	int rc;
+	int bit;
+
+	rc = info_update();
+	if (rc)
+		return rc;
+
+	bit = info_bit_num(chpid);
+	mutex_lock(&info_lock);
+	if (!chp_test_bit(chp_info.recognized, bit))
+		rc = CHP_STATUS_NOT_RECOGNIZED;
+	else if (chp_test_bit(chp_info.configured, bit))
+		rc = CHP_STATUS_CONFIGURED;
+	else if (chp_test_bit(chp_info.standby, bit))
+		rc = CHP_STATUS_STANDBY;
+	else
+		rc = CHP_STATUS_RESERVED;
+	mutex_unlock(&info_lock);
+
+	return rc;
+}
+
+/* Return configure task for chpid. */
+static enum cfg_task_t cfg_get_task(struct chp_id chpid)
+{
+	return chp_cfg_task[chpid.cssid][chpid.id];
+}
+
+/* Set configure task for chpid. */
+static void cfg_set_task(struct chp_id chpid, enum cfg_task_t cfg)
+{
+	chp_cfg_task[chpid.cssid][chpid.id] = cfg;
+}
+
+/* Perform one configure/deconfigure request. Reschedule work function until
+ * last request. */
+static void cfg_func(struct work_struct *work)
+{
+	struct chp_id chpid;
+	enum cfg_task_t t;
+	int rc;
+
+	mutex_lock(&cfg_lock);
+	t = cfg_none;
+	chp_id_for_each(&chpid) {
+		t = cfg_get_task(chpid);
+		if (t != cfg_none) {
+			cfg_set_task(chpid, cfg_none);
+			break;
+		}
+	}
+	mutex_unlock(&cfg_lock);
+
+	switch (t) {
+	case cfg_configure:
+		rc = sclp_chp_configure(chpid);
+		if (rc)
+			CIO_MSG_EVENT(2, "chp: sclp_chp_configure(%x.%02x)="
+				      "%d\n", chpid.cssid, chpid.id, rc);
+		else {
+			info_expire();
+			chsc_chp_online(chpid);
+		}
+		break;
+	case cfg_deconfigure:
+		rc = sclp_chp_deconfigure(chpid);
+		if (rc)
+			CIO_MSG_EVENT(2, "chp: sclp_chp_deconfigure(%x.%02x)="
+				      "%d\n", chpid.cssid, chpid.id, rc);
+		else {
+			info_expire();
+			chsc_chp_offline(chpid);
+		}
+		break;
+	case cfg_none:
+		/* Get updated information after last change. */
+		info_update();
+		mutex_lock(&cfg_lock);
+		cfg_busy = 0;
+		mutex_unlock(&cfg_lock);
+		wake_up_interruptible(&cfg_wait_queue);
+		return;
+	}
+	queue_work(chp_wq, &cfg_work);
+}
+
+/**
+ * chp_cfg_schedule - schedule chpid configuration request
+ * @chpid - channel-path ID
+ * @configure - Non-zero for configure, zero for deconfigure
+ *
+ * Schedule a channel-path configuration/deconfiguration request.
+ */
+void chp_cfg_schedule(struct chp_id chpid, int configure)
+{
+	CIO_MSG_EVENT(2, "chp_cfg_sched%x.%02x=%d\n", chpid.cssid, chpid.id,
+		      configure);
+	mutex_lock(&cfg_lock);
+	cfg_set_task(chpid, configure ? cfg_configure : cfg_deconfigure);
+	cfg_busy = 1;
+	mutex_unlock(&cfg_lock);
+	queue_work(chp_wq, &cfg_work);
+}
+
+/**
+ * chp_cfg_cancel_deconfigure - cancel chpid deconfiguration request
+ * @chpid - channel-path ID
+ *
+ * Cancel an active channel-path deconfiguration request if it has not yet
+ * been performed.
+ */
+void chp_cfg_cancel_deconfigure(struct chp_id chpid)
+{
+	CIO_MSG_EVENT(2, "chp_cfg_cancel:%x.%02x\n", chpid.cssid, chpid.id);
+	mutex_lock(&cfg_lock);
+	if (cfg_get_task(chpid) == cfg_deconfigure)
+		cfg_set_task(chpid, cfg_none);
+	mutex_unlock(&cfg_lock);
+}
+
+static int cfg_wait_idle(void)
+{
+	if (wait_event_interruptible(cfg_wait_queue, !cfg_busy))
+		return -ERESTARTSYS;
+	return 0;
+}
+
+static int __init chp_init(void)
+{
+	struct chp_id chpid;
+	int ret;
+
+	ret = crw_register_handler(CRW_RSC_CPATH, chp_process_crw);
+	if (ret)
+		return ret;
+	chp_wq = create_singlethread_workqueue("cio_chp");
+	if (!chp_wq) {
+		crw_unregister_handler(CRW_RSC_CPATH);
+		return -ENOMEM;
+	}
+	INIT_WORK(&cfg_work, cfg_func);
+	init_waitqueue_head(&cfg_wait_queue);
+	if (info_update())
+		return 0;
+	/* Register available channel-paths. */
+	chp_id_for_each(&chpid) {
+		if (chp_info_get_status(chpid) != CHP_STATUS_NOT_RECOGNIZED)
+			chp_new(chpid);
+	}
+
+	return 0;
+}
+
+subsys_initcall(chp_init);

drivers/s390/cio/chp.h

@@ -0,0 +1,72 @@
+/*
+ *    Copyright IBM Corp. 2007, 2010
+ *    Author(s): Peter Oberparleiter <peter.oberparleiter@de.ibm.com>
+ */
+
+#ifndef S390_CHP_H
+#define S390_CHP_H S390_CHP_H
+
+#include <linux/types.h>
+#include <linux/device.h>
+#include <linux/mutex.h>
+#include <asm/chpid.h>
+#include "chsc.h"
+#include "css.h"
+
+#define CHP_STATUS_STANDBY		0
+#define CHP_STATUS_CONFIGURED		1
+#define CHP_STATUS_RESERVED		2
+#define CHP_STATUS_NOT_RECOGNIZED	3
+
+#define CHP_ONLINE 0
+#define CHP_OFFLINE 1
+#define CHP_VARY_ON 2
+#define CHP_VARY_OFF 3
+
+struct chp_link {
+	struct chp_id chpid;
+	u32 fla_mask;
+	u16 fla;
+};
+
+static inline int chp_test_bit(u8 *bitmap, int num)
+{
+	int byte = num >> 3;
+	int mask = 128 >> (num & 7);
+
+	return (bitmap[byte] & mask) ? 1 : 0;
+}
+
+
+struct channel_path {
+	struct device dev;
+	struct chp_id chpid;
+	struct mutex lock; /* Serialize access to below members. */
+	int state;
+	struct channel_path_desc desc;
+	struct channel_path_desc_fmt1 desc_fmt1;
+	/* Channel-measurement related stuff: */
+	int cmg;
+	int shared;
+	void *cmg_chars;
+};
+
+/* Return channel_path struct for given chpid. */
+static inline struct channel_path *chpid_to_chp(struct chp_id chpid)
+{
+	return channel_subsystems[chpid.cssid]->chps[chpid.id];
+}
+
+int chp_get_status(struct chp_id chpid);
+u8 chp_get_sch_opm(struct subchannel *sch);
+int chp_is_registered(struct chp_id chpid);
+struct channel_path_desc *chp_get_chp_desc(struct chp_id chpid);
+void chp_remove_cmg_attr(struct channel_path *chp);
+int chp_add_cmg_attr(struct channel_path *chp);
+int chp_update_desc(struct channel_path *chp);
+int chp_new(struct chp_id chpid);
+void chp_cfg_schedule(struct chp_id chpid, int configure);
+void chp_cfg_cancel_deconfigure(struct chp_id chpid);
+int chp_info_get_status(struct chp_id chpid);
+int chp_ssd_get_mask(struct chsc_ssd_info *, struct chp_link *);
+#endif /* S390_CHP_H */

drivers/s390/cio/chsc.h

@@ -0,0 +1,238 @@
+#ifndef S390_CHSC_H
+#define S390_CHSC_H
+
+#include <linux/types.h>
+#include <linux/device.h>
+#include <asm/css_chars.h>
+#include <asm/chpid.h>
+#include <asm/chsc.h>
+#include <asm/schid.h>
+#include <asm/qdio.h>
+
+#define CHSC_SDA_OC_MSS   0x2
+
+#define NR_MEASUREMENT_CHARS 5
+struct cmg_chars {
+	u32 values[NR_MEASUREMENT_CHARS];
+} __attribute__ ((packed));
+
+#define NR_MEASUREMENT_ENTRIES 8
+struct cmg_entry {
+	u32 values[NR_MEASUREMENT_ENTRIES];
+} __attribute__ ((packed));
+
+struct channel_path_desc_fmt1 {
+	u8 flags;
+	u8 lsn;
+	u8 desc;
+	u8 chpid;
+	u32:24;
+	u8 chpp;
+	u32 unused[2];
+	u16 chid;
+	u32:16;
+	u16 mdc;
+	u16:13;
+	u8 r:1;
+	u8 s:1;
+	u8 f:1;
+	u32 zeros[2];
+} __attribute__ ((packed));
+
+struct channel_path;
+
+struct css_chsc_char {
+	u64 res;
+	u64 : 20;
+	u32 secm : 1; /* bit 84 */
+	u32 : 1;
+	u32 scmc : 1; /* bit 86 */
+	u32 : 20;
+	u32 scssc : 1;  /* bit 107 */
+	u32 scsscf : 1; /* bit 108 */
+	u32:7;
+	u32 pnso:1; /* bit 116 */
+	u32:11;
+}__attribute__((packed));
+
+extern struct css_chsc_char css_chsc_characteristics;
+
+struct chsc_ssd_info {
+	u8 path_mask;
+	u8 fla_valid_mask;
+	struct chp_id chpid[8];
+	u16 fla[8];
+};
+
+struct chsc_ssqd_area {
+	struct chsc_header request;
+	u16:10;
+	u8 ssid:2;
+	u8 fmt:4;
+	u16 first_sch;
+	u16:16;
+	u16 last_sch;
+	u32:32;
+	struct chsc_header response;
+	u32:32;
+	struct qdio_ssqd_desc qdio_ssqd;
+} __packed;
+
+struct chsc_scssc_area {
+	struct chsc_header request;
+	u16 operation_code;
+	u16:16;
+	u32:32;
+	u32:32;
+	u64 summary_indicator_addr;
+	u64 subchannel_indicator_addr;
+	u32 ks:4;
+	u32 kc:4;
+	u32:21;
+	u32 isc:3;
+	u32 word_with_d_bit;
+	u32:32;
+	struct subchannel_id schid;
+	u32 reserved[1004];
+	struct chsc_header response;
+	u32:32;
+} __packed;
+
+struct chsc_scpd {
+	struct chsc_header request;
+	u32:2;
+	u32 m:1;
+	u32 c:1;
+	u32 fmt:4;
+	u32 cssid:8;
+	u32:4;
+	u32 rfmt:4;
+	u32 first_chpid:8;
+	u32:24;
+	u32 last_chpid:8;
+	u32 zeroes1;
+	struct chsc_header response;
+	u8 data[PAGE_SIZE - 20];
+} __attribute__ ((packed));
+
+
+extern int chsc_get_ssd_info(struct subchannel_id schid,
+			     struct chsc_ssd_info *ssd);
+extern int chsc_determine_css_characteristics(void);
+extern int chsc_init(void);
+extern void chsc_init_cleanup(void);
+
+extern int chsc_enable_facility(int);
+struct channel_subsystem;
+extern int chsc_secm(struct channel_subsystem *, int);
+int __chsc_do_secm(struct channel_subsystem *css, int enable);
+
+int chsc_chp_vary(struct chp_id chpid, int on);
+int chsc_determine_channel_path_desc(struct chp_id chpid, int fmt, int rfmt,
+				     int c, int m, void *page);
+int chsc_determine_base_channel_path_desc(struct chp_id chpid,
+					  struct channel_path_desc *desc);
+int chsc_determine_fmt1_channel_path_desc(struct chp_id chpid,
+					  struct channel_path_desc_fmt1 *desc);
+void chsc_chp_online(struct chp_id chpid);
+void chsc_chp_offline(struct chp_id chpid);
+int chsc_get_channel_measurement_chars(struct channel_path *chp);
+int chsc_ssqd(struct subchannel_id schid, struct chsc_ssqd_area *ssqd);
+int chsc_sadc(struct subchannel_id schid, struct chsc_scssc_area *scssc,
+	      u64 summary_indicator_addr, u64 subchannel_indicator_addr);
+int chsc_error_from_response(int response);
+
+int chsc_siosl(struct subchannel_id schid);
+
+/* Functions and definitions to query storage-class memory. */
+struct sale {
+	u64 sa;
+	u32 p:4;
+	u32 op_state:4;
+	u32 data_state:4;
+	u32 rank:4;
+	u32 r:1;
+	u32:7;
+	u32 rid:8;
+	u32:32;
+} __packed;
+
+struct chsc_scm_info {
+	struct chsc_header request;
+	u32:32;
+	u64 reqtok;
+	u32 reserved1[4];
+	struct chsc_header response;
+	u64:56;
+	u8 rq;
+	u32 mbc;
+	u64 msa;
+	u16 is;
+	u16 mmc;
+	u32 mci;
+	u64 nr_scm_ini;
+	u64 nr_scm_unini;
+	u32 reserved2[10];
+	u64 restok;
+	struct sale scmal[248];
+} __packed;
+
+int chsc_scm_info(struct chsc_scm_info *scm_area, u64 token);
+
+struct chsc_brinfo_resume_token {
+	u64 t1;
+	u64 t2;
+} __packed;
+
+struct chsc_brinfo_naihdr {
+	struct chsc_brinfo_resume_token resume_token;
+	u32:32;
+	u32 instance;
+	u32:24;
+	u8 naids;
+	u32 reserved[3];
+} __packed;
+
+struct chsc_pnso_area {
+	struct chsc_header request;
+	u8:2;
+	u8 m:1;
+	u8:5;
+	u8:2;
+	u8 ssid:2;
+	u8 fmt:4;
+	u16 sch;
+	u8:8;
+	u8 cssid;
+	u16:16;
+	u8 oc;
+	u32:24;
+	struct chsc_brinfo_resume_token resume_token;
+	u32 n:1;
+	u32:31;
+	u32 reserved[3];
+	struct chsc_header response;
+	u32:32;
+	struct chsc_brinfo_naihdr naihdr;
+	union {
+		struct qdio_brinfo_entry_l3_ipv6 l3_ipv6[0];
+		struct qdio_brinfo_entry_l3_ipv4 l3_ipv4[0];
+		struct qdio_brinfo_entry_l2	 l2[0];
+	} entries;
+} __packed;
+
+int chsc_pnso_brinfo(struct subchannel_id schid,
+		struct chsc_pnso_area *brinfo_area,
+		struct chsc_brinfo_resume_token resume_token,
+		int cnc);
+
+#ifdef CONFIG_SCM_BUS
+int scm_update_information(void);
+int scm_process_availability_information(void);
+#else /* CONFIG_SCM_BUS */
+static inline int scm_update_information(void) { return 0; }
+static inline int scm_process_availability_information(void) { return 0; }
+#endif /* CONFIG_SCM_BUS */
+
+
+#endif

drivers/s390/cio/chsc_sch.h

@@ -0,0 +1,13 @@
+#ifndef _CHSC_SCH_H
+#define _CHSC_SCH_H
+
+struct chsc_request {
+	struct completion completion;
+	struct irb irb;
+};
+
+struct chsc_private {
+	struct chsc_request *request;
+};
+
+#endif

drivers/s390/cio/cio.h

@@ -0,0 +1,137 @@
+#ifndef S390_CIO_H
+#define S390_CIO_H
+
+#include <linux/mutex.h>
+#include <linux/device.h>
+#include <linux/mod_devicetable.h>
+#include <asm/chpid.h>
+#include <asm/cio.h>
+#include <asm/fcx.h>
+#include <asm/schid.h>
+#include "chsc.h"
+
+/*
+ * path management control word
+ */
+struct pmcw {
+	u32 intparm;		/* interruption parameter */
+	u32 qf	 : 1;		/* qdio facility */
+	u32 w	 : 1;
+	u32 isc  : 3;		/* interruption sublass */
+	u32 res5 : 3;		/* reserved zeros */
+	u32 ena  : 1;		/* enabled */
+	u32 lm	 : 2;		/* limit mode */
+	u32 mme  : 2;		/* measurement-mode enable */
+	u32 mp	 : 1;		/* multipath mode */
+	u32 tf	 : 1;		/* timing facility */
+	u32 dnv  : 1;		/* device number valid */
+	u32 dev  : 16;		/* device number */
+	u8  lpm;		/* logical path mask */
+	u8  pnom;		/* path not operational mask */
+	u8  lpum;		/* last path used mask */
+	u8  pim;		/* path installed mask */
+	u16 mbi;		/* measurement-block index */
+	u8  pom;		/* path operational mask */
+	u8  pam;		/* path available mask */
+	u8  chpid[8];		/* CHPID 0-7 (if available) */
+	u32 unused1 : 8;	/* reserved zeros */
+	u32 st	    : 3;	/* subchannel type */
+	u32 unused2 : 18;	/* reserved zeros */
+	u32 mbfc    : 1;	/* measurement block format control */
+	u32 xmwme   : 1;	/* extended measurement word mode enable */
+	u32 csense  : 1;	/* concurrent sense; can be enabled ...*/
+				/*  ... per MSCH, however, if facility */
+				/*  ... is not installed, this results */
+				/*  ... in an operand exception.       */
+} __attribute__ ((packed));
+
+/* Target SCHIB configuration. */
+struct schib_config {
+	u64 mba;
+	u32 intparm;
+	u16 mbi;
+	u32 isc:3;
+	u32 ena:1;
+	u32 mme:2;
+	u32 mp:1;
+	u32 csense:1;
+	u32 mbfc:1;
+} __attribute__ ((packed));
+
+/*
+ * subchannel information block
+ */
+struct schib {
+	struct pmcw pmcw;	 /* path management control word */
+	union scsw scsw;	 /* subchannel status word */
+	__u64 mba;               /* measurement block address */
+	__u8 mda[4];		 /* model dependent area */
+} __attribute__ ((packed,aligned(4)));
+
+/*
+ * When rescheduled, todo's with higher values will overwrite those
+ * with lower values.
+ */
+enum sch_todo {
+	SCH_TODO_NOTHING,
+	SCH_TODO_EVAL,
+	SCH_TODO_UNREG,
+};
+
+/* subchannel data structure used by I/O subroutines */
+struct subchannel {
+	struct subchannel_id schid;
+	spinlock_t *lock;	/* subchannel lock */
+	struct mutex reg_mutex;
+	enum {
+		SUBCHANNEL_TYPE_IO = 0,
+		SUBCHANNEL_TYPE_CHSC = 1,
+		SUBCHANNEL_TYPE_MSG = 2,
+		SUBCHANNEL_TYPE_ADM = 3,
+	} st;			/* subchannel type */
+	__u8 vpm;		/* verified path mask */
+	__u8 lpm;		/* logical path mask */
+	__u8 opm;               /* operational path mask */
+	struct schib schib;	/* subchannel information block */
+	int isc; /* desired interruption subclass */
+	struct chsc_ssd_info ssd_info;	/* subchannel description */
+	struct device dev;	/* entry in device tree */
+	struct css_driver *driver;
+	enum sch_todo todo;
+	struct work_struct todo_work;
+	struct schib_config config;
+} __attribute__ ((aligned(8)));
+
+DECLARE_PER_CPU(struct irb, cio_irb);
+
+#define to_subchannel(n) container_of(n, struct subchannel, dev)
+
+extern int cio_validate_subchannel (struct subchannel *, struct subchannel_id);
+extern int cio_enable_subchannel(struct subchannel *, u32);
+extern int cio_disable_subchannel (struct subchannel *);
+extern int cio_cancel (struct subchannel *);
+extern int cio_clear (struct subchannel *);
+extern int cio_resume (struct subchannel *);
+extern int cio_halt (struct subchannel *);
+extern int cio_start (struct subchannel *, struct ccw1 *, __u8);
+extern int cio_start_key (struct subchannel *, struct ccw1 *, __u8, __u8);
+extern int cio_cancel (struct subchannel *);
+extern int cio_set_options (struct subchannel *, int);
+extern int cio_update_schib(struct subchannel *sch);
+extern int cio_commit_config(struct subchannel *sch);
+
+int cio_tm_start_key(struct subchannel *sch, struct tcw *tcw, u8 lpm, u8 key);
+int cio_tm_intrg(struct subchannel *sch);
+
+/* Use with care. */
+#ifdef CONFIG_CCW_CONSOLE
+extern struct subchannel *cio_probe_console(void);
+extern int cio_is_console(struct subchannel_id);
+extern void cio_register_early_subchannels(void);
+extern void cio_tsch(struct subchannel *sch);
+#else
+#define cio_is_console(schid) 0
+static inline void cio_register_early_subchannels(void) {}
+#endif
+
+#endif

drivers/s390/cio/cio_debug.h

@@ -0,0 +1,34 @@
+#ifndef CIO_DEBUG_H
+#define CIO_DEBUG_H
+
+#include <asm/debug.h>
+
+/* for use of debug feature */
+extern debug_info_t *cio_debug_msg_id;
+extern debug_info_t *cio_debug_trace_id;
+extern debug_info_t *cio_debug_crw_id;
+
+#define CIO_TRACE_EVENT(imp, txt) do {				\
+		debug_text_event(cio_debug_trace_id, imp, txt); \
+	} while (0)
+
+#define CIO_MSG_EVENT(imp, args...) do {				\
+		debug_sprintf_event(cio_debug_msg_id, imp , ##args);	\
+	} while (0)
+
+#define CIO_CRW_EVENT(imp, args...) do {				\
+		debug_sprintf_event(cio_debug_crw_id, imp , ##args);	\
+	} while (0)
+
+static inline void CIO_HEX_EVENT(int level, void *data, int length)
+{
+	if (unlikely(!cio_debug_trace_id))
+		return;
+	while (length > 0) {
+		debug_event(cio_debug_trace_id, level, data, length);
+		length -= cio_debug_trace_id->buf_size;
+		data += cio_debug_trace_id->buf_size;
+	}
+}
+
+#endif

drivers/s390/cio/crw.c

@@ -0,0 +1,161 @@
+/*
+ *   Channel report handling code
+ *
+ *    Copyright IBM Corp. 2000, 2009
+ *    Author(s): Ingo Adlung <adlung@de.ibm.com>,
+ *		 Martin Schwidefsky <schwidefsky@de.ibm.com>,
+ *		 Cornelia Huck <cornelia.huck@de.ibm.com>,
+ *		 Heiko Carstens <heiko.carstens@de.ibm.com>,
+ */
+
+#include <linux/mutex.h>
+#include <linux/kthread.h>
+#include <linux/init.h>
+#include <linux/wait.h>
+#include <asm/crw.h>
+#include <asm/ctl_reg.h>
+
+static DEFINE_MUTEX(crw_handler_mutex);
+static crw_handler_t crw_handlers[NR_RSCS];
+static atomic_t crw_nr_req = ATOMIC_INIT(0);
+static DECLARE_WAIT_QUEUE_HEAD(crw_handler_wait_q);
+
+/**
+ * crw_register_handler() - register a channel report word handler
+ * @rsc: reporting source code to handle
+ * @handler: handler to be registered
+ *
+ * Returns %0 on success and a negative error value otherwise.
+ */
+int crw_register_handler(int rsc, crw_handler_t handler)
+{
+	int rc = 0;
+
+	if ((rsc < 0) || (rsc >= NR_RSCS))
+		return -EINVAL;
+	mutex_lock(&crw_handler_mutex);
+	if (crw_handlers[rsc])
+		rc = -EBUSY;
+	else
+		crw_handlers[rsc] = handler;
+	mutex_unlock(&crw_handler_mutex);
+	return rc;
+}
+
+/**
+ * crw_unregister_handler() - unregister a channel report word handler
+ * @rsc: reporting source code to handle
+ */
+void crw_unregister_handler(int rsc)
+{
+	if ((rsc < 0) || (rsc >= NR_RSCS))
+		return;
+	mutex_lock(&crw_handler_mutex);
+	crw_handlers[rsc] = NULL;
+	mutex_unlock(&crw_handler_mutex);
+}
+
+/*
+ * Retrieve CRWs and call function to handle event.
+ */
+static int crw_collect_info(void *unused)
+{
+	struct crw crw[2];
+	int ccode, signal;
+	unsigned int chain;
+
+repeat:
+	signal = wait_event_interruptible(crw_handler_wait_q,
+					  atomic_read(&crw_nr_req) > 0);
+	if (unlikely(signal))
+		atomic_inc(&crw_nr_req);
+	chain = 0;
+	while (1) {
+		crw_handler_t handler;
+
+		if (unlikely(chain > 1)) {
+			struct crw tmp_crw;
+
+			printk(KERN_WARNING"%s: Code does not support more "
+			       "than two chained crws; please report to "
+			       "linux390@de.ibm.com!\n", __func__);
+			ccode = stcrw(&tmp_crw);
+			printk(KERN_WARNING"%s: crw reports slct=%d, oflw=%d, "
+			       "chn=%d, rsc=%X, anc=%d, erc=%X, rsid=%X\n",
+			       __func__, tmp_crw.slct, tmp_crw.oflw,
+			       tmp_crw.chn, tmp_crw.rsc, tmp_crw.anc,
+			       tmp_crw.erc, tmp_crw.rsid);
+			printk(KERN_WARNING"%s: This was crw number %x in the "
+			       "chain\n", __func__, chain);
+			if (ccode != 0)
+				break;
+			chain = tmp_crw.chn ? chain + 1 : 0;
+			continue;
+		}
+		ccode = stcrw(&crw[chain]);
+		if (ccode != 0)
+			break;
+		printk(KERN_DEBUG "crw_info : CRW reports slct=%d, oflw=%d, "
+		       "chn=%d, rsc=%X, anc=%d, erc=%X, rsid=%X\n",
+		       crw[chain].slct, crw[chain].oflw, crw[chain].chn,
+		       crw[chain].rsc, crw[chain].anc, crw[chain].erc,
+		       crw[chain].rsid);
+		/* Check for overflows. */
+		if (crw[chain].oflw) {
+			int i;
+
+			pr_debug("%s: crw overflow detected!\n", __func__);
+			mutex_lock(&crw_handler_mutex);
+			for (i = 0; i < NR_RSCS; i++) {
+				if (crw_handlers[i])
+					crw_handlers[i](NULL, NULL, 1);
+			}
+			mutex_unlock(&crw_handler_mutex);
+			chain = 0;
+			continue;
+		}
+		if (crw[0].chn && !chain) {
+			chain++;
+			continue;
+		}
+		mutex_lock(&crw_handler_mutex);
+		handler = crw_handlers[crw[chain].rsc];
+		if (handler)
+			handler(&crw[0], chain ? &crw[1] : NULL, 0);
+		mutex_unlock(&crw_handler_mutex);
+		/* chain is always 0 or 1 here. */
+		chain = crw[chain].chn ? chain + 1 : 0;
+	}
+	if (atomic_dec_and_test(&crw_nr_req))
+		wake_up(&crw_handler_wait_q);
+	goto repeat;
+	return 0;
+}
+
+void crw_handle_channel_report(void)
+{
+	atomic_inc(&crw_nr_req);
+	wake_up(&crw_handler_wait_q);
+}
+
+void crw_wait_for_channel_report(void)
+{
+	crw_handle_channel_report();
+	wait_event(crw_handler_wait_q, atomic_read(&crw_nr_req) == 0);
+}
+
+/*
+ * Machine checks for the channel subsystem must be enabled
+ * after the channel subsystem is initialized
+ */
+static int __init crw_machine_check_init(void)
+{
+	struct task_struct *task;
+
+	task = kthread_run(crw_collect_info, NULL, "kmcheck");
+	if (IS_ERR(task))
+		return PTR_ERR(task);
+	ctl_set_bit(14, 28);	/* enable channel report MCH */
+	return 0;
+}
+device_initcall(crw_machine_check_init);

drivers/s390/cio/css.h

@@ -0,0 +1,146 @@
+#ifndef _CSS_H
+#define _CSS_H
+
+#include <linux/mutex.h>
+#include <linux/wait.h>
+#include <linux/workqueue.h>
+#include <linux/device.h>
+#include <linux/types.h>
+
+#include <asm/cio.h>
+#include <asm/chpid.h>
+#include <asm/schid.h>
+
+#include "cio.h"
+
+/*
+ * path grouping stuff
+ */
+#define SPID_FUNC_SINGLE_PATH	   0x00
+#define SPID_FUNC_MULTI_PATH	   0x80
+#define SPID_FUNC_ESTABLISH	   0x00
+#define SPID_FUNC_RESIGN	   0x40
+#define SPID_FUNC_DISBAND	   0x20
+
+#define SNID_STATE1_RESET	   0
+#define SNID_STATE1_UNGROUPED	   2
+#define SNID_STATE1_GROUPED	   3
+
+#define SNID_STATE2_NOT_RESVD	   0
+#define SNID_STATE2_RESVD_ELSE	   2
+#define SNID_STATE2_RESVD_SELF	   3
+
+#define SNID_STATE3_MULTI_PATH	   1
+#define SNID_STATE3_SINGLE_PATH	   0
+
+struct path_state {
+	__u8  state1 : 2;	/* path state value 1 */
+	__u8  state2 : 2;	/* path state value 2 */
+	__u8  state3 : 1;	/* path state value 3 */
+	__u8  resvd  : 3;	/* reserved */
+} __attribute__ ((packed));
+
+struct extended_cssid {
+	u8 version;
+	u8 cssid;
+} __attribute__ ((packed));
+
+struct pgid {
+	union {
+		__u8 fc;   	/* SPID function code */
+		struct path_state ps;	/* SNID path state */
+	} __attribute__ ((packed)) inf;
+	union {
+		__u32 cpu_addr	: 16;	/* CPU address */
+		struct extended_cssid ext_cssid;
+	} __attribute__ ((packed)) pgid_high;
+	__u32 cpu_id	: 24;	/* CPU identification */
+	__u32 cpu_model : 16;	/* CPU model */
+	__u32 tod_high;		/* high word TOD clock */
+} __attribute__ ((packed));
+
+struct subchannel;
+struct chp_link;
+/**
+ * struct css_driver - device driver for subchannels
+ * @subchannel_type: subchannel type supported by this driver
+ * @drv: embedded device driver structure
+ * @irq: called on interrupts
+ * @chp_event: called for events affecting a channel path
+ * @sch_event: called for events affecting the subchannel
+ * @probe: function called on probe
+ * @remove: function called on remove
+ * @shutdown: called at device shutdown
+ * @prepare: prepare for pm state transition
+ * @complete: undo work done in @prepare
+ * @freeze: callback for freezing during hibernation snapshotting
+ * @thaw: undo work done in @freeze
+ * @restore: callback for restoring after hibernation
+ * @settle: wait for asynchronous work to finish
+ */
+struct css_driver {
+	struct css_device_id *subchannel_type;
+	struct device_driver drv;
+	void (*irq)(struct subchannel *);
+	int (*chp_event)(struct subchannel *, struct chp_link *, int);
+	int (*sch_event)(struct subchannel *, int);
+	int (*probe)(struct subchannel *);
+	int (*remove)(struct subchannel *);
+	void (*shutdown)(struct subchannel *);
+	int (*prepare) (struct subchannel *);
+	void (*complete) (struct subchannel *);
+	int (*freeze)(struct subchannel *);
+	int (*thaw) (struct subchannel *);
+	int (*restore)(struct subchannel *);
+	int (*settle)(void);
+};
+
+#define to_cssdriver(n) container_of(n, struct css_driver, drv)
+
+extern int css_driver_register(struct css_driver *);
+extern void css_driver_unregister(struct css_driver *);
+
+extern void css_sch_device_unregister(struct subchannel *);
+extern int css_register_subchannel(struct subchannel *);
+extern struct subchannel *css_alloc_subchannel(struct subchannel_id);
+extern struct subchannel *get_subchannel_by_schid(struct subchannel_id);
+extern int css_init_done;
+extern int max_ssid;
+int for_each_subchannel_staged(int (*fn_known)(struct subchannel *, void *),
+			       int (*fn_unknown)(struct subchannel_id,
+			       void *), void *data);
+extern int for_each_subchannel(int(*fn)(struct subchannel_id, void *), void *);
+void css_update_ssd_info(struct subchannel *sch);
+
+struct channel_subsystem {
+	u8 cssid;
+	int valid;
+	struct channel_path *chps[__MAX_CHPID + 1];
+	struct device device;
+	struct pgid global_pgid;
+	struct mutex mutex;
+	/* channel measurement related */
+	int cm_enabled;
+	void *cub_addr1;
+	void *cub_addr2;
+	/* for orphaned ccw devices */
+	struct subchannel *pseudo_subchannel;
+};
+#define to_css(dev) container_of(dev, struct channel_subsystem, device)
+
+extern struct channel_subsystem *channel_subsystems[];
+
+/* Helper functions to build lists for the slow path. */
+void css_schedule_eval(struct subchannel_id schid);
+void css_schedule_eval_all(void);
+void css_schedule_eval_all_unreg(unsigned long delay);
+int css_complete_work(void);
+
+int sch_is_pseudo_sch(struct subchannel *);
+struct schib;
+int css_sch_is_valid(struct schib *);
+
+extern struct workqueue_struct *cio_work_q;
+void css_wait_for_slow_path(void);
+void css_sched_sch_todo(struct subchannel *sch, enum sch_todo todo);
+#endif

drivers/s390/cio/device.h

@@ -0,0 +1,150 @@
+#ifndef S390_DEVICE_H
+#define S390_DEVICE_H
+
+#include <asm/ccwdev.h>
+#include <linux/atomic.h>
+#include <linux/wait.h>
+#include <linux/notifier.h>
+#include <linux/kernel_stat.h>
+#include "io_sch.h"
+
+/*
+ * states of the device statemachine
+ */
+enum dev_state {
+	DEV_STATE_NOT_OPER,
+	DEV_STATE_SENSE_PGID,
+	DEV_STATE_SENSE_ID,
+	DEV_STATE_OFFLINE,
+	DEV_STATE_VERIFY,
+	DEV_STATE_ONLINE,
+	DEV_STATE_W4SENSE,
+	DEV_STATE_DISBAND_PGID,
+	DEV_STATE_BOXED,
+	/* states to wait for i/o completion before doing something */
+	DEV_STATE_TIMEOUT_KILL,
+	DEV_STATE_QUIESCE,
+	/* special states for devices gone not operational */
+	DEV_STATE_DISCONNECTED,
+	DEV_STATE_DISCONNECTED_SENSE_ID,
+	DEV_STATE_CMFCHANGE,
+	DEV_STATE_CMFUPDATE,
+	DEV_STATE_STEAL_LOCK,
+	/* last element! */
+	NR_DEV_STATES
+};
+
+/*
+ * asynchronous events of the device statemachine
+ */
+enum dev_event {
+	DEV_EVENT_NOTOPER,
+	DEV_EVENT_INTERRUPT,
+	DEV_EVENT_TIMEOUT,
+	DEV_EVENT_VERIFY,
+	/* last element! */
+	NR_DEV_EVENTS
+};
+
+struct ccw_device;
+
+/*
+ * action called through jumptable
+ */
+typedef void (fsm_func_t)(struct ccw_device *, enum dev_event);
+extern fsm_func_t *dev_jumptable[NR_DEV_STATES][NR_DEV_EVENTS];
+
+static inline void
+dev_fsm_event(struct ccw_device *cdev, enum dev_event dev_event)
+{
+	int state = cdev->private->state;
+
+	if (dev_event == DEV_EVENT_INTERRUPT) {
+		if (state == DEV_STATE_ONLINE)
+			inc_irq_stat(cdev->private->int_class);
+		else if (state != DEV_STATE_CMFCHANGE &&
+			 state != DEV_STATE_CMFUPDATE)
+			inc_irq_stat(IRQIO_CIO);
+	}
+	dev_jumptable[state][dev_event](cdev, dev_event);
+}
+
+/*
+ * Delivers 1 if the device state is final.
+ */
+static inline int
+dev_fsm_final_state(struct ccw_device *cdev)
+{
+	return (cdev->private->state == DEV_STATE_NOT_OPER ||
+		cdev->private->state == DEV_STATE_OFFLINE ||
+		cdev->private->state == DEV_STATE_ONLINE ||
+		cdev->private->state == DEV_STATE_BOXED);
+}
+
+int __init io_subchannel_init(void);
+
+void io_subchannel_recog_done(struct ccw_device *cdev);
+void io_subchannel_init_config(struct subchannel *sch);
+
+int ccw_device_cancel_halt_clear(struct ccw_device *);
+
+int ccw_device_is_orphan(struct ccw_device *);
+
+void ccw_device_recognition(struct ccw_device *);
+int ccw_device_online(struct ccw_device *);
+int ccw_device_offline(struct ccw_device *);
+void ccw_device_update_sense_data(struct ccw_device *);
+int ccw_device_test_sense_data(struct ccw_device *);
+void ccw_device_schedule_sch_unregister(struct ccw_device *);
+int ccw_purge_blacklisted(void);
+void ccw_device_sched_todo(struct ccw_device *cdev, enum cdev_todo todo);
+struct ccw_device *get_ccwdev_by_dev_id(struct ccw_dev_id *dev_id);
+
+/* Function prototypes for device status and basic sense stuff. */
+void ccw_device_accumulate_irb(struct ccw_device *, struct irb *);
+void ccw_device_accumulate_basic_sense(struct ccw_device *, struct irb *);
+int ccw_device_accumulate_and_sense(struct ccw_device *, struct irb *);
+int ccw_device_do_sense(struct ccw_device *, struct irb *);
+
+/* Function prototype for internal request handling. */
+int lpm_adjust(int lpm, int mask);
+void ccw_request_start(struct ccw_device *);
+int ccw_request_cancel(struct ccw_device *cdev);
+void ccw_request_handler(struct ccw_device *cdev);
+void ccw_request_timeout(struct ccw_device *cdev);
+void ccw_request_notoper(struct ccw_device *cdev);
+
+/* Function prototypes for sense id stuff. */
+void ccw_device_sense_id_start(struct ccw_device *);
+void ccw_device_sense_id_done(struct ccw_device *, int);
+
+/* Function prototypes for path grouping stuff. */
+void ccw_device_verify_start(struct ccw_device *);
+void ccw_device_verify_done(struct ccw_device *, int);
+
+void ccw_device_disband_start(struct ccw_device *);
+void ccw_device_disband_done(struct ccw_device *, int);
+
+void ccw_device_stlck_start(struct ccw_device *, void *, void *, void *);
+void ccw_device_stlck_done(struct ccw_device *, void *, int);
+
+int ccw_device_call_handler(struct ccw_device *);
+
+int ccw_device_stlck(struct ccw_device *);
+
+/* Helper function for machine check handling. */
+void ccw_device_trigger_reprobe(struct ccw_device *);
+void ccw_device_kill_io(struct ccw_device *);
+int ccw_device_notify(struct ccw_device *, int);
+void ccw_device_set_disconnected(struct ccw_device *cdev);
+void ccw_device_set_notoper(struct ccw_device *cdev);
+
+void ccw_device_set_timeout(struct ccw_device *, int);
+
+/* Channel measurement facility related */
+void retry_set_schib(struct ccw_device *cdev);
+void cmf_retry_copy_block(struct ccw_device *);
+int cmf_reenable(struct ccw_device *);
+int ccw_set_cmf(struct ccw_device *cdev, int enable);
+extern struct device_attribute dev_attr_cmb_enable;
+#endif

drivers/s390/cio/device_id.c

@@ -0,0 +1,222 @@
+/*
+ *  CCW device SENSE ID I/O handling.
+ *
+ *    Copyright IBM Corp. 2002, 2009
+ *    Author(s): Cornelia Huck <cornelia.huck@de.ibm.com>
+ *		 Martin Schwidefsky <schwidefsky@de.ibm.com>
+ *		 Peter Oberparleiter <peter.oberparleiter@de.ibm.com>
+ */
+
+#include <linux/kernel.h>
+#include <linux/string.h>
+#include <linux/types.h>
+#include <linux/errno.h>
+#include <asm/ccwdev.h>
+#include <asm/setup.h>
+#include <asm/cio.h>
+#include <asm/diag.h>
+
+#include "cio.h"
+#include "cio_debug.h"
+#include "device.h"
+#include "io_sch.h"
+
+#define SENSE_ID_RETRIES	256
+#define SENSE_ID_TIMEOUT	(10 * HZ)
+#define SENSE_ID_MIN_LEN	4
+#define SENSE_ID_BASIC_LEN	7
+
+/**
+ * diag210_to_senseid - convert diag 0x210 data to sense id information
+ * @senseid: sense id
+ * @diag: diag 0x210 data
+ *
+ * Return 0 on success, non-zero otherwise.
+ */
+static int diag210_to_senseid(struct senseid *senseid, struct diag210 *diag)
+{
+	static struct {
+		int class, type, cu_type;
+	} vm_devices[] = {
+		{ 0x08, 0x01, 0x3480 },
+		{ 0x08, 0x02, 0x3430 },
+		{ 0x08, 0x10, 0x3420 },
+		{ 0x08, 0x42, 0x3424 },
+		{ 0x08, 0x44, 0x9348 },
+		{ 0x08, 0x81, 0x3490 },
+		{ 0x08, 0x82, 0x3422 },
+		{ 0x10, 0x41, 0x1403 },
+		{ 0x10, 0x42, 0x3211 },
+		{ 0x10, 0x43, 0x3203 },
+		{ 0x10, 0x45, 0x3800 },
+		{ 0x10, 0x47, 0x3262 },
+		{ 0x10, 0x48, 0x3820 },
+		{ 0x10, 0x49, 0x3800 },
+		{ 0x10, 0x4a, 0x4245 },
+		{ 0x10, 0x4b, 0x4248 },
+		{ 0x10, 0x4d, 0x3800 },
+		{ 0x10, 0x4e, 0x3820 },
+		{ 0x10, 0x4f, 0x3820 },
+		{ 0x10, 0x82, 0x2540 },
+		{ 0x10, 0x84, 0x3525 },
+		{ 0x20, 0x81, 0x2501 },
+		{ 0x20, 0x82, 0x2540 },
+		{ 0x20, 0x84, 0x3505 },
+		{ 0x40, 0x01, 0x3278 },
+		{ 0x40, 0x04, 0x3277 },
+		{ 0x40, 0x80, 0x2250 },
+		{ 0x40, 0xc0, 0x5080 },
+		{ 0x80, 0x00, 0x3215 },
+	};
+	int i;
+
+	/* Special case for osa devices. */
+	if (diag->vrdcvcla == 0x02 && diag->vrdcvtyp == 0x20) {
+		senseid->cu_type = 0x3088;
+		senseid->cu_model = 0x60;
+		senseid->reserved = 0xff;
+		return 0;
+	}
+	for (i = 0; i < ARRAY_SIZE(vm_devices); i++) {
+		if (diag->vrdcvcla == vm_devices[i].class &&
+		    diag->vrdcvtyp == vm_devices[i].type) {
+			senseid->cu_type = vm_devices[i].cu_type;
+			senseid->reserved = 0xff;
+			return 0;
+		}
+	}
+
+	return -ENODEV;
+}
+
+/**
+ * diag_get_dev_info - retrieve device information via diag 0x210
+ * @cdev: ccw device
+ *
+ * Returns zero on success, non-zero otherwise.
+ */
+static int diag210_get_dev_info(struct ccw_device *cdev)
+{
+	struct ccw_dev_id *dev_id = &cdev->private->dev_id;
+	struct senseid *senseid = &cdev->private->senseid;
+	struct diag210 diag_data;
+	int rc;
+
+	if (dev_id->ssid != 0)
+		return -ENODEV;
+	memset(&diag_data, 0, sizeof(diag_data));
+	diag_data.vrdcdvno	= dev_id->devno;
+	diag_data.vrdclen	= sizeof(diag_data);
+	rc = diag210(&diag_data);
+	CIO_TRACE_EVENT(4, "diag210");
+	CIO_HEX_EVENT(4, &rc, sizeof(rc));
+	CIO_HEX_EVENT(4, &diag_data, sizeof(diag_data));
+	if (rc != 0 && rc != 2)
+		goto err_failed;
+	if (diag210_to_senseid(senseid, &diag_data))
+		goto err_unknown;
+	return 0;
+
+err_unknown:
+	CIO_MSG_EVENT(0, "snsid: device 0.%x.%04x: unknown diag210 data\n",
+		      dev_id->ssid, dev_id->devno);
+	return -ENODEV;
+err_failed:
+	CIO_MSG_EVENT(0, "snsid: device 0.%x.%04x: diag210 failed (rc=%d)\n",
+		      dev_id->ssid, dev_id->devno, rc);
+	return -ENODEV;
+}
+
+/*
+ * Initialize SENSE ID data.
+ */
+static void snsid_init(struct ccw_device *cdev)
+{
+	cdev->private->flags.esid = 0;
+	memset(&cdev->private->senseid, 0, sizeof(cdev->private->senseid));
+	cdev->private->senseid.cu_type = 0xffff;
+}
+
+/*
+ * Check for complete SENSE ID data.
+ */
+static int snsid_check(struct ccw_device *cdev, void *data)
+{
+	struct cmd_scsw *scsw = &cdev->private->irb.scsw.cmd;
+	int len = sizeof(struct senseid) - scsw->count;
+
+	/* Check for incomplete SENSE ID data. */
+	if (len < SENSE_ID_MIN_LEN)
+		goto out_restart;
+	if (cdev->private->senseid.cu_type == 0xffff)
+		goto out_restart;
+	/* Check for incompatible SENSE ID data. */
+	if (cdev->private->senseid.reserved != 0xff)
+		return -EOPNOTSUPP;
+	/* Check for extended-identification information. */
+	if (len > SENSE_ID_BASIC_LEN)
+		cdev->private->flags.esid = 1;
+	return 0;
+
+out_restart:
+	snsid_init(cdev);
+	return -EAGAIN;
+}
+
+/*
+ * Process SENSE ID request result.
+ */
+static void snsid_callback(struct ccw_device *cdev, void *data, int rc)
+{
+	struct ccw_dev_id *id = &cdev->private->dev_id;
+	struct senseid *senseid = &cdev->private->senseid;
+	int vm = 0;
+
+	if (rc && MACHINE_IS_VM) {
+		/* Try diag 0x210 fallback on z/VM. */
+		snsid_init(cdev);
+		if (diag210_get_dev_info(cdev) == 0) {
+			rc = 0;
+			vm = 1;
+		}
+	}
+	CIO_MSG_EVENT(2, "snsid: device 0.%x.%04x: rc=%d %04x/%02x "
+		      "%04x/%02x%s\n", id->ssid, id->devno, rc,
+		      senseid->cu_type, senseid->cu_model, senseid->dev_type,
+		      senseid->dev_model, vm ? " (diag210)" : "");
+	ccw_device_sense_id_done(cdev, rc);
+}
+
+/**
+ * ccw_device_sense_id_start - perform SENSE ID
+ * @cdev: ccw device
+ *
+ * Execute a SENSE ID channel program on @cdev to update its sense id
+ * information. When finished, call ccw_device_sense_id_done with a
+ * return code specifying the result.
+ */
+void ccw_device_sense_id_start(struct ccw_device *cdev)
+{
+	struct subchannel *sch = to_subchannel(cdev->dev.parent);
+	struct ccw_request *req = &cdev->private->req;
+	struct ccw1 *cp = cdev->private->iccws;
+
+	CIO_TRACE_EVENT(4, "snsid");
+	CIO_HEX_EVENT(4, &cdev->private->dev_id, sizeof(cdev->private->dev_id));
+	/* Data setup. */
+	snsid_init(cdev);
+	/* Channel program setup. */
+	cp->cmd_code	= CCW_CMD_SENSE_ID;
+	cp->cda		= (u32) (addr_t) &cdev->private->senseid;
+	cp->count	= sizeof(struct senseid);
+	cp->flags	= CCW_FLAG_SLI;
+	/* Request setup. */
+	memset(req, 0, sizeof(*req));
+	req->cp		= cp;
+	req->timeout	= SENSE_ID_TIMEOUT;
+	req->maxretries	= SENSE_ID_RETRIES;
+	req->lpm	= sch->schib.pmcw.pam & sch->opm;
+	req->check	= snsid_check;
+	req->callback	= snsid_callback;
+	ccw_request_start(cdev);
+}

drivers/s390/cio/device_ops.c

@@ -0,0 +1,799 @@
+/*
+ * Copyright IBM Corp. 2002, 2009
+ *
+ * Author(s): Martin Schwidefsky (schwidefsky@de.ibm.com)
+ *	      Cornelia Huck (cornelia.huck@de.ibm.com)
+ */
+#include <linux/module.h>
+#include <linux/init.h>
+#include <linux/errno.h>
+#include <linux/slab.h>
+#include <linux/list.h>
+#include <linux/device.h>
+#include <linux/delay.h>
+#include <linux/completion.h>
+
+#include <asm/ccwdev.h>
+#include <asm/idals.h>
+#include <asm/chpid.h>
+#include <asm/fcx.h>
+
+#include "cio.h"
+#include "cio_debug.h"
+#include "css.h"
+#include "chsc.h"
+#include "device.h"
+#include "chp.h"
+
+/**
+ * ccw_device_set_options_mask() - set some options and unset the rest
+ * @cdev: device for which the options are to be set
+ * @flags: options to be set
+ *
+ * All flags specified in @flags are set, all flags not specified in @flags
+ * are cleared.
+ * Returns:
+ *   %0 on success, -%EINVAL on an invalid flag combination.
+ */
+int ccw_device_set_options_mask(struct ccw_device *cdev, unsigned long flags)
+{
+       /*
+	* The flag usage is mutal exclusive ...
+	*/
+	if ((flags & CCWDEV_EARLY_NOTIFICATION) &&
+	    (flags & CCWDEV_REPORT_ALL))
+		return -EINVAL;
+	cdev->private->options.fast = (flags & CCWDEV_EARLY_NOTIFICATION) != 0;
+	cdev->private->options.repall = (flags & CCWDEV_REPORT_ALL) != 0;
+	cdev->private->options.pgroup = (flags & CCWDEV_DO_PATHGROUP) != 0;
+	cdev->private->options.force = (flags & CCWDEV_ALLOW_FORCE) != 0;
+	cdev->private->options.mpath = (flags & CCWDEV_DO_MULTIPATH) != 0;
+	return 0;
+}
+
+/**
+ * ccw_device_set_options() - set some options
+ * @cdev: device for which the options are to be set
+ * @flags: options to be set
+ *
+ * All flags specified in @flags are set, the remainder is left untouched.
+ * Returns:
+ *   %0 on success, -%EINVAL if an invalid flag combination would ensue.
+ */
+int ccw_device_set_options(struct ccw_device *cdev, unsigned long flags)
+{
+       /*
+	* The flag usage is mutal exclusive ...
+	*/
+	if (((flags & CCWDEV_EARLY_NOTIFICATION) &&
+	    (flags & CCWDEV_REPORT_ALL)) ||
+	    ((flags & CCWDEV_EARLY_NOTIFICATION) &&
+	     cdev->private->options.repall) ||
+	    ((flags & CCWDEV_REPORT_ALL) &&
+	     cdev->private->options.fast))
+		return -EINVAL;
+	cdev->private->options.fast |= (flags & CCWDEV_EARLY_NOTIFICATION) != 0;
+	cdev->private->options.repall |= (flags & CCWDEV_REPORT_ALL) != 0;
+	cdev->private->options.pgroup |= (flags & CCWDEV_DO_PATHGROUP) != 0;
+	cdev->private->options.force |= (flags & CCWDEV_ALLOW_FORCE) != 0;
+	cdev->private->options.mpath |= (flags & CCWDEV_DO_MULTIPATH) != 0;
+	return 0;
+}
+
+/**
+ * ccw_device_clear_options() - clear some options
+ * @cdev: device for which the options are to be cleared
+ * @flags: options to be cleared
+ *
+ * All flags specified in @flags are cleared, the remainder is left untouched.
+ */
+void ccw_device_clear_options(struct ccw_device *cdev, unsigned long flags)
+{
+	cdev->private->options.fast &= (flags & CCWDEV_EARLY_NOTIFICATION) == 0;
+	cdev->private->options.repall &= (flags & CCWDEV_REPORT_ALL) == 0;
+	cdev->private->options.pgroup &= (flags & CCWDEV_DO_PATHGROUP) == 0;
+	cdev->private->options.force &= (flags & CCWDEV_ALLOW_FORCE) == 0;
+	cdev->private->options.mpath &= (flags & CCWDEV_DO_MULTIPATH) == 0;
+}
+
+/**
+ * ccw_device_is_pathgroup - determine if paths to this device are grouped
+ * @cdev: ccw device
+ *
+ * Return non-zero if there is a path group, zero otherwise.
+ */
+int ccw_device_is_pathgroup(struct ccw_device *cdev)
+{
+	return cdev->private->flags.pgroup;
+}
+EXPORT_SYMBOL(ccw_device_is_pathgroup);
+
+/**
+ * ccw_device_is_multipath - determine if device is operating in multipath mode
+ * @cdev: ccw device
+ *
+ * Return non-zero if device is operating in multipath mode, zero otherwise.
+ */
+int ccw_device_is_multipath(struct ccw_device *cdev)
+{
+	return cdev->private->flags.mpath;
+}
+EXPORT_SYMBOL(ccw_device_is_multipath);
+
+/**
+ * ccw_device_clear() - terminate I/O request processing
+ * @cdev: target ccw device
+ * @intparm: interruption parameter; value is only used if no I/O is
+ *	     outstanding, otherwise the intparm associated with the I/O request
+ *	     is returned
+ *
+ * ccw_device_clear() calls csch on @cdev's subchannel.
+ * Returns:
+ *  %0 on success,
+ *  -%ENODEV on device not operational,
+ *  -%EINVAL on invalid device state.
+ * Context:
+ *  Interrupts disabled, ccw device lock held
+ */
+int ccw_device_clear(struct ccw_device *cdev, unsigned long intparm)
+{
+	struct subchannel *sch;
+	int ret;
+
+	if (!cdev || !cdev->dev.parent)
+		return -ENODEV;
+	sch = to_subchannel(cdev->dev.parent);
+	if (!sch->schib.pmcw.ena)
+		return -EINVAL;
+	if (cdev->private->state == DEV_STATE_NOT_OPER)
+		return -ENODEV;
+	if (cdev->private->state != DEV_STATE_ONLINE &&
+	    cdev->private->state != DEV_STATE_W4SENSE)
+		return -EINVAL;
+
+	ret = cio_clear(sch);
+	if (ret == 0)
+		cdev->private->intparm = intparm;
+	return ret;
+}
+
+/**
+ * ccw_device_start_key() - start a s390 channel program with key
+ * @cdev: target ccw device
+ * @cpa: logical start address of channel program
+ * @intparm: user specific interruption parameter; will be presented back to
+ *	     @cdev's interrupt handler. Allows a device driver to associate
+ *	     the interrupt with a particular I/O request.
+ * @lpm: defines the channel path to be used for a specific I/O request. A
+ *	 value of 0 will make cio use the opm.
+ * @key: storage key to be used for the I/O
+ * @flags: additional flags; defines the action to be performed for I/O
+ *	   processing.
+ *
+ * Start a S/390 channel program. When the interrupt arrives, the
+ * IRQ handler is called, either immediately, delayed (dev-end missing,
+ * or sense required) or never (no IRQ handler registered).
+ * Returns:
+ *  %0, if the operation was successful;
+ *  -%EBUSY, if the device is busy, or status pending;
+ *  -%EACCES, if no path specified in @lpm is operational;
+ *  -%ENODEV, if the device is not operational.
+ * Context:
+ *  Interrupts disabled, ccw device lock held
+ */
+int ccw_device_start_key(struct ccw_device *cdev, struct ccw1 *cpa,
+			 unsigned long intparm, __u8 lpm, __u8 key,
+			 unsigned long flags)
+{
+	struct subchannel *sch;
+	int ret;
+
+	if (!cdev || !cdev->dev.parent)
+		return -ENODEV;
+	sch = to_subchannel(cdev->dev.parent);
+	if (!sch->schib.pmcw.ena)
+		return -EINVAL;
+	if (cdev->private->state == DEV_STATE_NOT_OPER)
+		return -ENODEV;
+	if (cdev->private->state == DEV_STATE_VERIFY) {
+		/* Remember to fake irb when finished. */
+		if (!cdev->private->flags.fake_irb) {
+			cdev->private->flags.fake_irb = FAKE_CMD_IRB;
+			cdev->private->intparm = intparm;
+			return 0;
+		} else
+			/* There's already a fake I/O around. */
+			return -EBUSY;
+	}
+	if (cdev->private->state != DEV_STATE_ONLINE ||
+	    ((sch->schib.scsw.cmd.stctl & SCSW_STCTL_PRIM_STATUS) &&
+	     !(sch->schib.scsw.cmd.stctl & SCSW_STCTL_SEC_STATUS)) ||
+	    cdev->private->flags.doverify)
+		return -EBUSY;
+	ret = cio_set_options (sch, flags);
+	if (ret)
+		return ret;
+	/* Adjust requested path mask to exclude unusable paths. */
+	if (lpm) {
+		lpm &= sch->lpm;
+		if (lpm == 0)
+			return -EACCES;
+	}
+	ret = cio_start_key (sch, cpa, lpm, key);
+	switch (ret) {
+	case 0:
+		cdev->private->intparm = intparm;
+		break;
+	case -EACCES:
+	case -ENODEV:
+		dev_fsm_event(cdev, DEV_EVENT_VERIFY);
+		break;
+	}
+	return ret;
+}
+
+/**
+ * ccw_device_start_timeout_key() - start a s390 channel program with timeout and key
+ * @cdev: target ccw device
+ * @cpa: logical start address of channel program
+ * @intparm: user specific interruption parameter; will be presented back to
+ *	     @cdev's interrupt handler. Allows a device driver to associate
+ *	     the interrupt with a particular I/O request.
+ * @lpm: defines the channel path to be used for a specific I/O request. A
+ *	 value of 0 will make cio use the opm.
+ * @key: storage key to be used for the I/O
+ * @flags: additional flags; defines the action to be performed for I/O
+ *	   processing.
+ * @expires: timeout value in jiffies
+ *
+ * Start a S/390 channel program. When the interrupt arrives, the
+ * IRQ handler is called, either immediately, delayed (dev-end missing,
+ * or sense required) or never (no IRQ handler registered).
+ * This function notifies the device driver if the channel program has not
+ * completed during the time specified by @expires. If a timeout occurs, the
+ * channel program is terminated via xsch, hsch or csch, and the device's
+ * interrupt handler will be called with an irb containing ERR_PTR(-%ETIMEDOUT).
+ * Returns:
+ *  %0, if the operation was successful;
+ *  -%EBUSY, if the device is busy, or status pending;
+ *  -%EACCES, if no path specified in @lpm is operational;
+ *  -%ENODEV, if the device is not operational.
+ * Context:
+ *  Interrupts disabled, ccw device lock held
+ */
+int ccw_device_start_timeout_key(struct ccw_device *cdev, struct ccw1 *cpa,
+				 unsigned long intparm, __u8 lpm, __u8 key,
+				 unsigned long flags, int expires)
+{
+	int ret;
+
+	if (!cdev)
+		return -ENODEV;
+	ccw_device_set_timeout(cdev, expires);
+	ret = ccw_device_start_key(cdev, cpa, intparm, lpm, key, flags);
+	if (ret != 0)
+		ccw_device_set_timeout(cdev, 0);
+	return ret;
+}
+
+/**
+ * ccw_device_start() - start a s390 channel program
+ * @cdev: target ccw device
+ * @cpa: logical start address of channel program
+ * @intparm: user specific interruption parameter; will be presented back to
+ *	     @cdev's interrupt handler. Allows a device driver to associate
+ *	     the interrupt with a particular I/O request.
+ * @lpm: defines the channel path to be used for a specific I/O request. A
+ *	 value of 0 will make cio use the opm.
+ * @flags: additional flags; defines the action to be performed for I/O
+ *	   processing.
+ *
+ * Start a S/390 channel program. When the interrupt arrives, the
+ * IRQ handler is called, either immediately, delayed (dev-end missing,
+ * or sense required) or never (no IRQ handler registered).
+ * Returns:
+ *  %0, if the operation was successful;
+ *  -%EBUSY, if the device is busy, or status pending;
+ *  -%EACCES, if no path specified in @lpm is operational;
+ *  -%ENODEV, if the device is not operational.
+ * Context:
+ *  Interrupts disabled, ccw device lock held
+ */
+int ccw_device_start(struct ccw_device *cdev, struct ccw1 *cpa,
+		     unsigned long intparm, __u8 lpm, unsigned long flags)
+{
+	return ccw_device_start_key(cdev, cpa, intparm, lpm,
+				    PAGE_DEFAULT_KEY, flags);
+}
+
+/**
+ * ccw_device_start_timeout() - start a s390 channel program with timeout
+ * @cdev: target ccw device
+ * @cpa: logical start address of channel program
+ * @intparm: user specific interruption parameter; will be presented back to
+ *	     @cdev's interrupt handler. Allows a device driver to associate
+ *	     the interrupt with a particular I/O request.
+ * @lpm: defines the channel path to be used for a specific I/O request. A
+ *	 value of 0 will make cio use the opm.
+ * @flags: additional flags; defines the action to be performed for I/O
+ *	   processing.
+ * @expires: timeout value in jiffies
+ *
+ * Start a S/390 channel program. When the interrupt arrives, the
+ * IRQ handler is called, either immediately, delayed (dev-end missing,
+ * or sense required) or never (no IRQ handler registered).
+ * This function notifies the device driver if the channel program has not
+ * completed during the time specified by @expires. If a timeout occurs, the
+ * channel program is terminated via xsch, hsch or csch, and the device's
+ * interrupt handler will be called with an irb containing ERR_PTR(-%ETIMEDOUT).
+ * Returns:
+ *  %0, if the operation was successful;
+ *  -%EBUSY, if the device is busy, or status pending;
+ *  -%EACCES, if no path specified in @lpm is operational;
+ *  -%ENODEV, if the device is not operational.
+ * Context:
+ *  Interrupts disabled, ccw device lock held
+ */
+int ccw_device_start_timeout(struct ccw_device *cdev, struct ccw1 *cpa,
+			     unsigned long intparm, __u8 lpm,
+			     unsigned long flags, int expires)
+{
+	return ccw_device_start_timeout_key(cdev, cpa, intparm, lpm,
+					    PAGE_DEFAULT_KEY, flags,
+					    expires);
+}
+
+
+/**
+ * ccw_device_halt() - halt I/O request processing
+ * @cdev: target ccw device
+ * @intparm: interruption parameter; value is only used if no I/O is
+ *	     outstanding, otherwise the intparm associated with the I/O request
+ *	     is returned
+ *
+ * ccw_device_halt() calls hsch on @cdev's subchannel.
+ * Returns:
+ *  %0 on success,
+ *  -%ENODEV on device not operational,
+ *  -%EINVAL on invalid device state,
+ *  -%EBUSY on device busy or interrupt pending.
+ * Context:
+ *  Interrupts disabled, ccw device lock held
+ */
+int ccw_device_halt(struct ccw_device *cdev, unsigned long intparm)
+{
+	struct subchannel *sch;
+	int ret;
+
+	if (!cdev || !cdev->dev.parent)
+		return -ENODEV;
+	sch = to_subchannel(cdev->dev.parent);
+	if (!sch->schib.pmcw.ena)
+		return -EINVAL;
+	if (cdev->private->state == DEV_STATE_NOT_OPER)
+		return -ENODEV;
+	if (cdev->private->state != DEV_STATE_ONLINE &&
+	    cdev->private->state != DEV_STATE_W4SENSE)
+		return -EINVAL;
+
+	ret = cio_halt(sch);
+	if (ret == 0)
+		cdev->private->intparm = intparm;
+	return ret;
+}
+
+/**
+ * ccw_device_resume() - resume channel program execution
+ * @cdev: target ccw device
+ *
+ * ccw_device_resume() calls rsch on @cdev's subchannel.
+ * Returns:
+ *  %0 on success,
+ *  -%ENODEV on device not operational,
+ *  -%EINVAL on invalid device state,
+ *  -%EBUSY on device busy or interrupt pending.
+ * Context:
+ *  Interrupts disabled, ccw device lock held
+ */
+int ccw_device_resume(struct ccw_device *cdev)
+{
+	struct subchannel *sch;
+
+	if (!cdev || !cdev->dev.parent)
+		return -ENODEV;
+	sch = to_subchannel(cdev->dev.parent);
+	if (!sch->schib.pmcw.ena)
+		return -EINVAL;
+	if (cdev->private->state == DEV_STATE_NOT_OPER)
+		return -ENODEV;
+	if (cdev->private->state != DEV_STATE_ONLINE ||
+	    !(sch->schib.scsw.cmd.actl & SCSW_ACTL_SUSPENDED))
+		return -EINVAL;
+	return cio_resume(sch);
+}
+
+/*
+ * Pass interrupt to device driver.
+ */
+int
+ccw_device_call_handler(struct ccw_device *cdev)
+{
+	unsigned int stctl;
+	int ending_status;
+
+	/*
+	 * we allow for the device action handler if .
+	 *  - we received ending status
+	 *  - the action handler requested to see all interrupts
+	 *  - we received an intermediate status
+	 *  - fast notification was requested (primary status)
+	 *  - unsolicited interrupts
+	 */
+	stctl = scsw_stctl(&cdev->private->irb.scsw);
+	ending_status = (stctl & SCSW_STCTL_SEC_STATUS) ||
+		(stctl == (SCSW_STCTL_ALERT_STATUS | SCSW_STCTL_STATUS_PEND)) ||
+		(stctl == SCSW_STCTL_STATUS_PEND);
+	if (!ending_status &&
+	    !cdev->private->options.repall &&
+	    !(stctl & SCSW_STCTL_INTER_STATUS) &&
+	    !(cdev->private->options.fast &&
+	      (stctl & SCSW_STCTL_PRIM_STATUS)))
+		return 0;
+
+	/* Clear pending timers for device driver initiated I/O. */
+	if (ending_status)
+		ccw_device_set_timeout(cdev, 0);
+	/*
+	 * Now we are ready to call the device driver interrupt handler.
+	 */
+	if (cdev->handler)
+		cdev->handler(cdev, cdev->private->intparm,
+			      &cdev->private->irb);
+
+	/*
+	 * Clear the old and now useless interrupt response block.
+	 */
+	memset(&cdev->private->irb, 0, sizeof(struct irb));
+
+	return 1;
+}
+
+/**
+ * ccw_device_get_ciw() - Search for CIW command in extended sense data.
+ * @cdev: ccw device to inspect
+ * @ct: command type to look for
+ *
+ * During SenseID, command information words (CIWs) describing special
+ * commands available to the device may have been stored in the extended
+ * sense data. This function searches for CIWs of a specified command
+ * type in the extended sense data.
+ * Returns:
+ *  %NULL if no extended sense data has been stored or if no CIW of the
+ *  specified command type could be found,
+ *  else a pointer to the CIW of the specified command type.
+ */
+struct ciw *ccw_device_get_ciw(struct ccw_device *cdev, __u32 ct)
+{
+	int ciw_cnt;
+
+	if (cdev->private->flags.esid == 0)
+		return NULL;
+	for (ciw_cnt = 0; ciw_cnt < MAX_CIWS; ciw_cnt++)
+		if (cdev->private->senseid.ciw[ciw_cnt].ct == ct)
+			return cdev->private->senseid.ciw + ciw_cnt;
+	return NULL;
+}
+
+/**
+ * ccw_device_get_path_mask() - get currently available paths
+ * @cdev: ccw device to be queried
+ * Returns:
+ *  %0 if no subchannel for the device is available,
+ *  else the mask of currently available paths for the ccw device's subchannel.
+ */
+__u8 ccw_device_get_path_mask(struct ccw_device *cdev)
+{
+	struct subchannel *sch;
+
+	if (!cdev->dev.parent)
+		return 0;
+
+	sch = to_subchannel(cdev->dev.parent);
+	return sch->lpm;
+}
+
+struct stlck_data {
+	struct completion done;
+	int rc;
+};
+
+void ccw_device_stlck_done(struct ccw_device *cdev, void *data, int rc)
+{
+	struct stlck_data *sdata = data;
+
+	sdata->rc = rc;
+	complete(&sdata->done);
+}
+
+/*
+ * Perform unconditional reserve + release.
+ */
+int ccw_device_stlck(struct ccw_device *cdev)
+{
+	struct subchannel *sch = to_subchannel(cdev->dev.parent);
+	struct stlck_data data;
+	u8 *buffer;
+	int rc;
+
+	/* Check if steal lock operation is valid for this device. */
+	if (cdev->drv) {
+		if (!cdev->private->options.force)
+			return -EINVAL;
+	}
+	buffer = kzalloc(64, GFP_DMA | GFP_KERNEL);
+	if (!buffer)
+		return -ENOMEM;
+	init_completion(&data.done);
+	data.rc = -EIO;
+	spin_lock_irq(sch->lock);
+	rc = cio_enable_subchannel(sch, (u32) (addr_t) sch);
+	if (rc)
+		goto out_unlock;
+	/* Perform operation. */
+	cdev->private->state = DEV_STATE_STEAL_LOCK,
+	ccw_device_stlck_start(cdev, &data, &buffer[0], &buffer[32]);
+	spin_unlock_irq(sch->lock);
+	/* Wait for operation to finish. */
+	if (wait_for_completion_interruptible(&data.done)) {
+		/* Got a signal. */
+		spin_lock_irq(sch->lock);
+		ccw_request_cancel(cdev);
+		spin_unlock_irq(sch->lock);
+		wait_for_completion(&data.done);
+	}
+	rc = data.rc;
+	/* Check results. */
+	spin_lock_irq(sch->lock);
+	cio_disable_subchannel(sch);
+	cdev->private->state = DEV_STATE_BOXED;
+out_unlock:
+	spin_unlock_irq(sch->lock);
+	kfree(buffer);
+
+	return rc;
+}
+
+/**
+ * chp_get_chp_desc - return newly allocated channel-path descriptor
+ * @cdev: device to obtain the descriptor for
+ * @chp_idx: index of the channel path
+ *
+ * On success return a newly allocated copy of the channel-path description
+ * data associated with the given channel path. Return %NULL on error.
+ */
+struct channel_path_desc *ccw_device_get_chp_desc(struct ccw_device *cdev,
+						  int chp_idx)
+{
+	struct subchannel *sch;
+	struct chp_id chpid;
+
+	sch = to_subchannel(cdev->dev.parent);
+	chp_id_init(&chpid);
+	chpid.id = sch->schib.pmcw.chpid[chp_idx];
+	return chp_get_chp_desc(chpid);
+}
+
+/**
+ * ccw_device_get_id - obtain a ccw device id
+ * @cdev: device to obtain the id for
+ * @dev_id: where to fill in the values
+ */
+void ccw_device_get_id(struct ccw_device *cdev, struct ccw_dev_id *dev_id)
+{
+	*dev_id = cdev->private->dev_id;
+}
+EXPORT_SYMBOL(ccw_device_get_id);
+
+/**
+ * ccw_device_tm_start_key - perform start function
+ * @cdev: ccw device on which to perform the start function
+ * @tcw: transport-command word to be started
+ * @intparm: user defined parameter to be passed to the interrupt handler
+ * @lpm: mask of paths to use
+ * @key: storage key to use for storage access
+ *
+ * Start the tcw on the given ccw device. Return zero on success, non-zero
+ * otherwise.
+ */
+int ccw_device_tm_start_key(struct ccw_device *cdev, struct tcw *tcw,
+			    unsigned long intparm, u8 lpm, u8 key)
+{
+	struct subchannel *sch;
+	int rc;
+
+	sch = to_subchannel(cdev->dev.parent);
+	if (!sch->schib.pmcw.ena)
+		return -EINVAL;
+	if (cdev->private->state == DEV_STATE_VERIFY) {
+		/* Remember to fake irb when finished. */
+		if (!cdev->private->flags.fake_irb) {
+			cdev->private->flags.fake_irb = FAKE_TM_IRB;
+			cdev->private->intparm = intparm;
+			return 0;
+		} else
+			/* There's already a fake I/O around. */
+			return -EBUSY;
+	}
+	if (cdev->private->state != DEV_STATE_ONLINE)
+		return -EIO;
+	/* Adjust requested path mask to exclude unusable paths. */
+	if (lpm) {
+		lpm &= sch->lpm;
+		if (lpm == 0)
+			return -EACCES;
+	}
+	rc = cio_tm_start_key(sch, tcw, lpm, key);
+	if (rc == 0)
+		cdev->private->intparm = intparm;
+	return rc;
+}
+EXPORT_SYMBOL(ccw_device_tm_start_key);
+
+/**
+ * ccw_device_tm_start_timeout_key - perform start function
+ * @cdev: ccw device on which to perform the start function
+ * @tcw: transport-command word to be started
+ * @intparm: user defined parameter to be passed to the interrupt handler
+ * @lpm: mask of paths to use
+ * @key: storage key to use for storage access
+ * @expires: time span in jiffies after which to abort request
+ *
+ * Start the tcw on the given ccw device. Return zero on success, non-zero
+ * otherwise.
+ */
+int ccw_device_tm_start_timeout_key(struct ccw_device *cdev, struct tcw *tcw,
+				    unsigned long intparm, u8 lpm, u8 key,
+				    int expires)
+{
+	int ret;
+
+	ccw_device_set_timeout(cdev, expires);
+	ret = ccw_device_tm_start_key(cdev, tcw, intparm, lpm, key);
+	if (ret != 0)
+		ccw_device_set_timeout(cdev, 0);
+	return ret;
+}
+EXPORT_SYMBOL(ccw_device_tm_start_timeout_key);
+
+/**
+ * ccw_device_tm_start - perform start function
+ * @cdev: ccw device on which to perform the start function
+ * @tcw: transport-command word to be started
+ * @intparm: user defined parameter to be passed to the interrupt handler
+ * @lpm: mask of paths to use
+ *
+ * Start the tcw on the given ccw device. Return zero on success, non-zero
+ * otherwise.
+ */
+int ccw_device_tm_start(struct ccw_device *cdev, struct tcw *tcw,
+			unsigned long intparm, u8 lpm)
+{
+	return ccw_device_tm_start_key(cdev, tcw, intparm, lpm,
+				       PAGE_DEFAULT_KEY);
+}
+EXPORT_SYMBOL(ccw_device_tm_start);
+
+/**
+ * ccw_device_tm_start_timeout - perform start function
+ * @cdev: ccw device on which to perform the start function
+ * @tcw: transport-command word to be started
+ * @intparm: user defined parameter to be passed to the interrupt handler
+ * @lpm: mask of paths to use
+ * @expires: time span in jiffies after which to abort request
+ *
+ * Start the tcw on the given ccw device. Return zero on success, non-zero
+ * otherwise.
+ */
+int ccw_device_tm_start_timeout(struct ccw_device *cdev, struct tcw *tcw,
+			       unsigned long intparm, u8 lpm, int expires)
+{
+	return ccw_device_tm_start_timeout_key(cdev, tcw, intparm, lpm,
+					       PAGE_DEFAULT_KEY, expires);
+}
+EXPORT_SYMBOL(ccw_device_tm_start_timeout);
+
+/**
+ * ccw_device_get_mdc - accumulate max data count
+ * @cdev: ccw device for which the max data count is accumulated
+ * @mask: mask of paths to use
+ *
+ * Return the number of 64K-bytes blocks all paths at least support
+ * for a transport command. Return values <= 0 indicate failures.
+ */
+int ccw_device_get_mdc(struct ccw_device *cdev, u8 mask)
+{
+	struct subchannel *sch = to_subchannel(cdev->dev.parent);
+	struct channel_path *chp;
+	struct chp_id chpid;
+	int mdc = 0, i;
+
+	/* Adjust requested path mask to excluded varied off paths. */
+	if (mask)
+		mask &= sch->lpm;
+	else
+		mask = sch->lpm;
+
+	chp_id_init(&chpid);
+	for (i = 0; i < 8; i++) {
+		if (!(mask & (0x80 >> i)))
+			continue;
+		chpid.id = sch->schib.pmcw.chpid[i];
+		chp = chpid_to_chp(chpid);
+		if (!chp)
+			continue;
+
+		mutex_lock(&chp->lock);
+		if (!chp->desc_fmt1.f) {
+			mutex_unlock(&chp->lock);
+			return 0;
+		}
+		if (!chp->desc_fmt1.r)
+			mdc = 1;
+		mdc = mdc ? min_t(int, mdc, chp->desc_fmt1.mdc) :
+			    chp->desc_fmt1.mdc;
+		mutex_unlock(&chp->lock);
+	}
+
+	return mdc;
+}
+EXPORT_SYMBOL(ccw_device_get_mdc);
+
+/**
+ * ccw_device_tm_intrg - perform interrogate function
+ * @cdev: ccw device on which to perform the interrogate function
+ *
+ * Perform an interrogate function on the given ccw device. Return zero on
+ * success, non-zero otherwise.
+ */
+int ccw_device_tm_intrg(struct ccw_device *cdev)
+{
+	struct subchannel *sch = to_subchannel(cdev->dev.parent);
+
+	if (!sch->schib.pmcw.ena)
+		return -EINVAL;
+	if (cdev->private->state != DEV_STATE_ONLINE)
+		return -EIO;
+	if (!scsw_is_tm(&sch->schib.scsw) ||
+	    !(scsw_actl(&sch->schib.scsw) & SCSW_ACTL_START_PEND))
+		return -EINVAL;
+	return cio_tm_intrg(sch);
+}
+EXPORT_SYMBOL(ccw_device_tm_intrg);
+
+/**
+ * ccw_device_get_schid - obtain a subchannel id
+ * @cdev: device to obtain the id for
+ * @schid: where to fill in the values
+ */
+void ccw_device_get_schid(struct ccw_device *cdev, struct subchannel_id *schid)
+{
+	struct subchannel *sch = to_subchannel(cdev->dev.parent);
+
+	*schid = sch->schid;
+}
+EXPORT_SYMBOL_GPL(ccw_device_get_schid);
+
+MODULE_LICENSE("GPL");
+EXPORT_SYMBOL(ccw_device_set_options_mask);
+EXPORT_SYMBOL(ccw_device_set_options);
+EXPORT_SYMBOL(ccw_device_clear_options);
+EXPORT_SYMBOL(ccw_device_clear);
+EXPORT_SYMBOL(ccw_device_halt);
+EXPORT_SYMBOL(ccw_device_resume);
+EXPORT_SYMBOL(ccw_device_start_timeout);
+EXPORT_SYMBOL(ccw_device_start);
+EXPORT_SYMBOL(ccw_device_start_timeout_key);
+EXPORT_SYMBOL(ccw_device_start_key);
+EXPORT_SYMBOL(ccw_device_get_ciw);
+EXPORT_SYMBOL(ccw_device_get_path_mask);
+EXPORT_SYMBOL_GPL(ccw_device_get_chp_desc);

drivers/s390/cio/device_pgid.c

@@ -0,0 +1,669 @@
+/*
+ *  CCW device PGID and path verification I/O handling.
+ *
+ *    Copyright IBM Corp. 2002, 2009
+ *    Author(s): Cornelia Huck <cornelia.huck@de.ibm.com>
+ *		 Martin Schwidefsky <schwidefsky@de.ibm.com>
+ *		 Peter Oberparleiter <peter.oberparleiter@de.ibm.com>
+ */
+
+#include <linux/kernel.h>
+#include <linux/string.h>
+#include <linux/types.h>
+#include <linux/errno.h>
+#include <linux/bitops.h>
+#include <asm/ccwdev.h>
+#include <asm/cio.h>
+
+#include "cio.h"
+#include "cio_debug.h"
+#include "device.h"
+#include "io_sch.h"
+
+#define PGID_RETRIES	256
+#define PGID_TIMEOUT	(10 * HZ)
+
+static void verify_start(struct ccw_device *cdev);
+
+/*
+ * Process path verification data and report result.
+ */
+static void verify_done(struct ccw_device *cdev, int rc)
+{
+	struct subchannel *sch = to_subchannel(cdev->dev.parent);
+	struct ccw_dev_id *id = &cdev->private->dev_id;
+	int mpath = cdev->private->flags.mpath;
+	int pgroup = cdev->private->flags.pgroup;
+
+	if (rc)
+		goto out;
+	/* Ensure consistent multipathing state at device and channel. */
+	if (sch->config.mp != mpath) {
+		sch->config.mp = mpath;
+		rc = cio_commit_config(sch);
+	}
+out:
+	CIO_MSG_EVENT(2, "vrfy: device 0.%x.%04x: rc=%d pgroup=%d mpath=%d "
+			 "vpm=%02x\n", id->ssid, id->devno, rc, pgroup, mpath,
+			 sch->vpm);
+	ccw_device_verify_done(cdev, rc);
+}
+
+/*
+ * Create channel program to perform a NOOP.
+ */
+static void nop_build_cp(struct ccw_device *cdev)
+{
+	struct ccw_request *req = &cdev->private->req;
+	struct ccw1 *cp = cdev->private->iccws;
+
+	cp->cmd_code	= CCW_CMD_NOOP;
+	cp->cda		= 0;
+	cp->count	= 0;
+	cp->flags	= CCW_FLAG_SLI;
+	req->cp		= cp;
+}
+
+/*
+ * Perform NOOP on a single path.
+ */
+static void nop_do(struct ccw_device *cdev)
+{
+	struct subchannel *sch = to_subchannel(cdev->dev.parent);
+	struct ccw_request *req = &cdev->private->req;
+
+	req->lpm = lpm_adjust(req->lpm, sch->schib.pmcw.pam & sch->opm &
+			      ~cdev->private->path_noirq_mask);
+	if (!req->lpm)
+		goto out_nopath;
+	nop_build_cp(cdev);
+	ccw_request_start(cdev);
+	return;
+
+out_nopath:
+	verify_done(cdev, sch->vpm ? 0 : -EACCES);
+}
+
+/*
+ * Adjust NOOP I/O status.
+ */
+static enum io_status nop_filter(struct ccw_device *cdev, void *data,
+				 struct irb *irb, enum io_status status)
+{
+	/* Only subchannel status might indicate a path error. */
+	if (status == IO_STATUS_ERROR && irb->scsw.cmd.cstat == 0)
+		return IO_DONE;
+	return status;
+}
+
+/*
+ * Process NOOP request result for a single path.
+ */
+static void nop_callback(struct ccw_device *cdev, void *data, int rc)
+{
+	struct subchannel *sch = to_subchannel(cdev->dev.parent);
+	struct ccw_request *req = &cdev->private->req;
+
+	switch (rc) {
+	case 0:
+		sch->vpm |= req->lpm;
+		break;
+	case -ETIME:
+		cdev->private->path_noirq_mask |= req->lpm;
+		break;
+	case -EACCES:
+		cdev->private->path_notoper_mask |= req->lpm;
+		break;
+	default:
+		goto err;
+	}
+	/* Continue on the next path. */
+	req->lpm >>= 1;
+	nop_do(cdev);
+	return;
+
+err:
+	verify_done(cdev, rc);
+}
+
+/*
+ * Create channel program to perform SET PGID on a single path.
+ */
+static void spid_build_cp(struct ccw_device *cdev, u8 fn)
+{
+	struct ccw_request *req = &cdev->private->req;
+	struct ccw1 *cp = cdev->private->iccws;
+	int i = 8 - ffs(req->lpm);
+	struct pgid *pgid = &cdev->private->pgid[i];
+
+	pgid->inf.fc	= fn;
+	cp->cmd_code	= CCW_CMD_SET_PGID;
+	cp->cda		= (u32) (addr_t) pgid;
+	cp->count	= sizeof(*pgid);
+	cp->flags	= CCW_FLAG_SLI;
+	req->cp		= cp;
+}
+
+static void pgid_wipeout_callback(struct ccw_device *cdev, void *data, int rc)
+{
+	if (rc) {
+		/* We don't know the path groups' state. Abort. */
+		verify_done(cdev, rc);
+		return;
+	}
+	/*
+	 * Path groups have been reset. Restart path verification but
+	 * leave paths in path_noirq_mask out.
+	 */
+	cdev->private->flags.pgid_unknown = 0;
+	verify_start(cdev);
+}
+
+/*
+ * Reset pathgroups and restart path verification, leave unusable paths out.
+ */
+static void pgid_wipeout_start(struct ccw_device *cdev)
+{
+	struct subchannel *sch = to_subchannel(cdev->dev.parent);
+	struct ccw_dev_id *id = &cdev->private->dev_id;
+	struct ccw_request *req = &cdev->private->req;
+	u8 fn;
+
+	CIO_MSG_EVENT(2, "wipe: device 0.%x.%04x: pvm=%02x nim=%02x\n",
+		      id->ssid, id->devno, cdev->private->pgid_valid_mask,
+		      cdev->private->path_noirq_mask);
+
+	/* Initialize request data. */
+	memset(req, 0, sizeof(*req));
+	req->timeout	= PGID_TIMEOUT;
+	req->maxretries	= PGID_RETRIES;
+	req->lpm	= sch->schib.pmcw.pam;
+	req->callback	= pgid_wipeout_callback;
+	fn = SPID_FUNC_DISBAND;
+	if (cdev->private->flags.mpath)
+		fn |= SPID_FUNC_MULTI_PATH;
+	spid_build_cp(cdev, fn);
+	ccw_request_start(cdev);
+}
+
+/*
+ * Perform establish/resign SET PGID on a single path.
+ */
+static void spid_do(struct ccw_device *cdev)
+{
+	struct subchannel *sch = to_subchannel(cdev->dev.parent);
+	struct ccw_request *req = &cdev->private->req;
+	u8 fn;
+
+	/* Use next available path that is not already in correct state. */
+	req->lpm = lpm_adjust(req->lpm, cdev->private->pgid_todo_mask);
+	if (!req->lpm)
+		goto out_nopath;
+	/* Channel program setup. */
+	if (req->lpm & sch->opm)
+		fn = SPID_FUNC_ESTABLISH;
+	else
+		fn = SPID_FUNC_RESIGN;
+	if (cdev->private->flags.mpath)
+		fn |= SPID_FUNC_MULTI_PATH;
+	spid_build_cp(cdev, fn);
+	ccw_request_start(cdev);
+	return;
+
+out_nopath:
+	if (cdev->private->flags.pgid_unknown) {
+		/* At least one SPID could be partially done. */
+		pgid_wipeout_start(cdev);
+		return;
+	}
+	verify_done(cdev, sch->vpm ? 0 : -EACCES);
+}
+
+/*
+ * Process SET PGID request result for a single path.
+ */
+static void spid_callback(struct ccw_device *cdev, void *data, int rc)
+{
+	struct subchannel *sch = to_subchannel(cdev->dev.parent);
+	struct ccw_request *req = &cdev->private->req;
+
+	switch (rc) {
+	case 0:
+		sch->vpm |= req->lpm & sch->opm;
+		break;
+	case -ETIME:
+		cdev->private->flags.pgid_unknown = 1;
+		cdev->private->path_noirq_mask |= req->lpm;
+		break;
+	case -EACCES:
+		cdev->private->path_notoper_mask |= req->lpm;
+		break;
+	case -EOPNOTSUPP:
+		if (cdev->private->flags.mpath) {
+			/* Try without multipathing. */
+			cdev->private->flags.mpath = 0;
+			goto out_restart;
+		}
+		/* Try without pathgrouping. */
+		cdev->private->flags.pgroup = 0;
+		goto out_restart;
+	default:
+		goto err;
+	}
+	req->lpm >>= 1;
+	spid_do(cdev);
+	return;
+
+out_restart:
+	verify_start(cdev);
+	return;
+err:
+	verify_done(cdev, rc);
+}
+
+static void spid_start(struct ccw_device *cdev)
+{
+	struct ccw_request *req = &cdev->private->req;
+
+	/* Initialize request data. */
+	memset(req, 0, sizeof(*req));
+	req->timeout	= PGID_TIMEOUT;
+	req->maxretries	= PGID_RETRIES;
+	req->lpm	= 0x80;
+	req->singlepath	= 1;
+	req->callback	= spid_callback;
+	spid_do(cdev);
+}
+
+static int pgid_is_reset(struct pgid *p)
+{
+	char *c;
+
+	for (c = (char *)p + 1; c < (char *)(p + 1); c++) {
+		if (*c != 0)
+			return 0;
+	}
+	return 1;
+}
+
+static int pgid_cmp(struct pgid *p1, struct pgid *p2)
+{
+	return memcmp((char *) p1 + 1, (char *) p2 + 1,
+		      sizeof(struct pgid) - 1);
+}
+
+/*
+ * Determine pathgroup state from PGID data.
+ */
+static void pgid_analyze(struct ccw_device *cdev, struct pgid **p,
+			 int *mismatch, u8 *reserved, u8 *reset)
+{
+	struct pgid *pgid = &cdev->private->pgid[0];
+	struct pgid *first = NULL;
+	int lpm;
+	int i;
+
+	*mismatch = 0;
+	*reserved = 0;
+	*reset = 0;
+	for (i = 0, lpm = 0x80; i < 8; i++, pgid++, lpm >>= 1) {
+		if ((cdev->private->pgid_valid_mask & lpm) == 0)
+			continue;
+		if (pgid->inf.ps.state2 == SNID_STATE2_RESVD_ELSE)
+			*reserved |= lpm;
+		if (pgid_is_reset(pgid)) {
+			*reset |= lpm;
+			continue;
+		}
+		if (!first) {
+			first = pgid;
+			continue;
+		}
+		if (pgid_cmp(pgid, first) != 0)
+			*mismatch = 1;
+	}
+	if (!first)
+		first = &channel_subsystems[0]->global_pgid;
+	*p = first;
+}
+
+static u8 pgid_to_donepm(struct ccw_device *cdev)
+{
+	struct subchannel *sch = to_subchannel(cdev->dev.parent);
+	struct pgid *pgid;
+	int i;
+	int lpm;
+	u8 donepm = 0;
+
+	/* Set bits for paths which are already in the target state. */
+	for (i = 0; i < 8; i++) {
+		lpm = 0x80 >> i;
+		if ((cdev->private->pgid_valid_mask & lpm) == 0)
+			continue;
+		pgid = &cdev->private->pgid[i];
+		if (sch->opm & lpm) {
+			if (pgid->inf.ps.state1 != SNID_STATE1_GROUPED)
+				continue;
+		} else {
+			if (pgid->inf.ps.state1 != SNID_STATE1_UNGROUPED)
+				continue;
+		}
+		if (cdev->private->flags.mpath) {
+			if (pgid->inf.ps.state3 != SNID_STATE3_MULTI_PATH)
+				continue;
+		} else {
+			if (pgid->inf.ps.state3 != SNID_STATE3_SINGLE_PATH)
+				continue;
+		}
+		donepm |= lpm;
+	}
+
+	return donepm;
+}
+
+static void pgid_fill(struct ccw_device *cdev, struct pgid *pgid)
+{
+	int i;
+
+	for (i = 0; i < 8; i++)
+		memcpy(&cdev->private->pgid[i], pgid, sizeof(struct pgid));
+}
+
+/*
+ * Process SENSE PGID data and report result.
+ */
+static void snid_done(struct ccw_device *cdev, int rc)
+{
+	struct ccw_dev_id *id = &cdev->private->dev_id;
+	struct subchannel *sch = to_subchannel(cdev->dev.parent);
+	struct pgid *pgid;
+	int mismatch = 0;
+	u8 reserved = 0;
+	u8 reset = 0;
+	u8 donepm;
+
+	if (rc)
+		goto out;
+	pgid_analyze(cdev, &pgid, &mismatch, &reserved, &reset);
+	if (reserved == cdev->private->pgid_valid_mask)
+		rc = -EUSERS;
+	else if (mismatch)
+		rc = -EOPNOTSUPP;
+	else {
+		donepm = pgid_to_donepm(cdev);
+		sch->vpm = donepm & sch->opm;
+		cdev->private->pgid_reset_mask |= reset;
+		cdev->private->pgid_todo_mask &=
+			~(donepm | cdev->private->path_noirq_mask);
+		pgid_fill(cdev, pgid);
+	}
+out:
+	CIO_MSG_EVENT(2, "snid: device 0.%x.%04x: rc=%d pvm=%02x vpm=%02x "
+		      "todo=%02x mism=%d rsvd=%02x reset=%02x\n", id->ssid,
+		      id->devno, rc, cdev->private->pgid_valid_mask, sch->vpm,
+		      cdev->private->pgid_todo_mask, mismatch, reserved, reset);
+	switch (rc) {
+	case 0:
+		if (cdev->private->flags.pgid_unknown) {
+			pgid_wipeout_start(cdev);
+			return;
+		}
+		/* Anything left to do? */
+		if (cdev->private->pgid_todo_mask == 0) {
+			verify_done(cdev, sch->vpm == 0 ? -EACCES : 0);
+			return;
+		}
+		/* Perform path-grouping. */
+		spid_start(cdev);
+		break;
+	case -EOPNOTSUPP:
+		/* Path-grouping not supported. */
+		cdev->private->flags.pgroup = 0;
+		cdev->private->flags.mpath = 0;
+		verify_start(cdev);
+		break;
+	default:
+		verify_done(cdev, rc);
+	}
+}
+
+/*
+ * Create channel program to perform a SENSE PGID on a single path.
+ */
+static void snid_build_cp(struct ccw_device *cdev)
+{
+	struct ccw_request *req = &cdev->private->req;
+	struct ccw1 *cp = cdev->private->iccws;
+	int i = 8 - ffs(req->lpm);
+
+	/* Channel program setup. */
+	cp->cmd_code	= CCW_CMD_SENSE_PGID;
+	cp->cda		= (u32) (addr_t) &cdev->private->pgid[i];
+	cp->count	= sizeof(struct pgid);
+	cp->flags	= CCW_FLAG_SLI;
+	req->cp		= cp;
+}
+
+/*
+ * Perform SENSE PGID on a single path.
+ */
+static void snid_do(struct ccw_device *cdev)
+{
+	struct subchannel *sch = to_subchannel(cdev->dev.parent);
+	struct ccw_request *req = &cdev->private->req;
+	int ret;
+
+	req->lpm = lpm_adjust(req->lpm, sch->schib.pmcw.pam &
+			      ~cdev->private->path_noirq_mask);
+	if (!req->lpm)
+		goto out_nopath;
+	snid_build_cp(cdev);
+	ccw_request_start(cdev);
+	return;
+
+out_nopath:
+	if (cdev->private->pgid_valid_mask)
+		ret = 0;
+	else if (cdev->private->path_noirq_mask)
+		ret = -ETIME;
+	else
+		ret = -EACCES;
+	snid_done(cdev, ret);
+}
+
+/*
+ * Process SENSE PGID request result for single path.
+ */
+static void snid_callback(struct ccw_device *cdev, void *data, int rc)
+{
+	struct ccw_request *req = &cdev->private->req;
+
+	switch (rc) {
+	case 0:
+		cdev->private->pgid_valid_mask |= req->lpm;
+		break;
+	case -ETIME:
+		cdev->private->flags.pgid_unknown = 1;
+		cdev->private->path_noirq_mask |= req->lpm;
+		break;
+	case -EACCES:
+		cdev->private->path_notoper_mask |= req->lpm;
+		break;
+	default:
+		goto err;
+	}
+	/* Continue on the next path. */
+	req->lpm >>= 1;
+	snid_do(cdev);
+	return;
+
+err:
+	snid_done(cdev, rc);
+}
+
+/*
+ * Perform path verification.
+ */
+static void verify_start(struct ccw_device *cdev)
+{
+	struct subchannel *sch = to_subchannel(cdev->dev.parent);
+	struct ccw_request *req = &cdev->private->req;
+	struct ccw_dev_id *devid = &cdev->private->dev_id;
+
+	sch->vpm = 0;
+	sch->lpm = sch->schib.pmcw.pam;
+
+	/* Initialize PGID data. */
+	memset(cdev->private->pgid, 0, sizeof(cdev->private->pgid));
+	cdev->private->pgid_valid_mask = 0;
+	cdev->private->pgid_todo_mask = sch->schib.pmcw.pam;
+	cdev->private->path_notoper_mask = 0;
+
+	/* Initialize request data. */
+	memset(req, 0, sizeof(*req));
+	req->timeout	= PGID_TIMEOUT;
+	req->maxretries	= PGID_RETRIES;
+	req->lpm	= 0x80;
+	req->singlepath	= 1;
+	if (cdev->private->flags.pgroup) {
+		CIO_TRACE_EVENT(4, "snid");
+		CIO_HEX_EVENT(4, devid, sizeof(*devid));
+		req->callback	= snid_callback;
+		snid_do(cdev);
+	} else {
+		CIO_TRACE_EVENT(4, "nop");
+		CIO_HEX_EVENT(4, devid, sizeof(*devid));
+		req->filter	= nop_filter;
+		req->callback	= nop_callback;
+		nop_do(cdev);
+	}
+}
+
+/**
+ * ccw_device_verify_start - perform path verification
+ * @cdev: ccw device
+ *
+ * Perform an I/O on each available channel path to @cdev to determine which
+ * paths are operational. The resulting path mask is stored in sch->vpm.
+ * If device options specify pathgrouping, establish a pathgroup for the
+ * operational paths. When finished, call ccw_device_verify_done with a
+ * return code specifying the result.
+ */
+void ccw_device_verify_start(struct ccw_device *cdev)
+{
+	CIO_TRACE_EVENT(4, "vrfy");
+	CIO_HEX_EVENT(4, &cdev->private->dev_id, sizeof(cdev->private->dev_id));
+	/*
+	 * Initialize pathgroup and multipath state with target values.
+	 * They may change in the course of path verification.
+	 */
+	cdev->private->flags.pgroup = cdev->private->options.pgroup;
+	cdev->private->flags.mpath = cdev->private->options.mpath;
+	cdev->private->flags.doverify = 0;
+	cdev->private->path_noirq_mask = 0;
+	verify_start(cdev);
+}
+
+/*
+ * Process disband SET PGID request result.
+ */
+static void disband_callback(struct ccw_device *cdev, void *data, int rc)
+{
+	struct subchannel *sch = to_subchannel(cdev->dev.parent);
+	struct ccw_dev_id *id = &cdev->private->dev_id;
+
+	if (rc)
+		goto out;
+	/* Ensure consistent multipathing state at device and channel. */
+	cdev->private->flags.mpath = 0;
+	if (sch->config.mp) {
+		sch->config.mp = 0;
+		rc = cio_commit_config(sch);
+	}
+out:
+	CIO_MSG_EVENT(0, "disb: device 0.%x.%04x: rc=%d\n", id->ssid, id->devno,
+		      rc);
+	ccw_device_disband_done(cdev, rc);
+}
+
+/**
+ * ccw_device_disband_start - disband pathgroup
+ * @cdev: ccw device
+ *
+ * Execute a SET PGID channel program on @cdev to disband a previously
+ * established pathgroup. When finished, call ccw_device_disband_done with
+ * a return code specifying the result.
+ */
+void ccw_device_disband_start(struct ccw_device *cdev)
+{
+	struct subchannel *sch = to_subchannel(cdev->dev.parent);
+	struct ccw_request *req = &cdev->private->req;
+	u8 fn;
+
+	CIO_TRACE_EVENT(4, "disb");
+	CIO_HEX_EVENT(4, &cdev->private->dev_id, sizeof(cdev->private->dev_id));
+	/* Request setup. */
+	memset(req, 0, sizeof(*req));
+	req->timeout	= PGID_TIMEOUT;
+	req->maxretries	= PGID_RETRIES;
+	req->lpm	= sch->schib.pmcw.pam & sch->opm;
+	req->singlepath	= 1;
+	req->callback	= disband_callback;
+	fn = SPID_FUNC_DISBAND;
+	if (cdev->private->flags.mpath)
+		fn |= SPID_FUNC_MULTI_PATH;
+	spid_build_cp(cdev, fn);
+	ccw_request_start(cdev);
+}
+
+static void stlck_build_cp(struct ccw_device *cdev, void *buf1, void *buf2)
+{
+	struct ccw_request *req = &cdev->private->req;
+	struct ccw1 *cp = cdev->private->iccws;
+
+	cp[0].cmd_code = CCW_CMD_STLCK;
+	cp[0].cda = (u32) (addr_t) buf1;
+	cp[0].count = 32;
+	cp[0].flags = CCW_FLAG_CC;
+	cp[1].cmd_code = CCW_CMD_RELEASE;
+	cp[1].cda = (u32) (addr_t) buf2;
+	cp[1].count = 32;
+	cp[1].flags = 0;
+	req->cp = cp;
+}
+
+static void stlck_callback(struct ccw_device *cdev, void *data, int rc)
+{
+	ccw_device_stlck_done(cdev, data, rc);
+}
+
+/**
+ * ccw_device_stlck_start - perform unconditional release
+ * @cdev: ccw device
+ * @data: data pointer to be passed to ccw_device_stlck_done
+ * @buf1: data pointer used in channel program
+ * @buf2: data pointer used in channel program
+ *
+ * Execute a channel program on @cdev to release an existing PGID reservation.
+ * When finished, call ccw_device_stlck_done with a return code specifying the
+ * result.
+ */
+void ccw_device_stlck_start(struct ccw_device *cdev, void *data, void *buf1,
+			    void *buf2)
+{
+	struct subchannel *sch = to_subchannel(cdev->dev.parent);
+	struct ccw_request *req = &cdev->private->req;
+
+	CIO_TRACE_EVENT(4, "stlck");
+	CIO_HEX_EVENT(4, &cdev->private->dev_id, sizeof(cdev->private->dev_id));
+	/* Request setup. */
+	memset(req, 0, sizeof(*req));
+	req->timeout	= PGID_TIMEOUT;
+	req->maxretries	= PGID_RETRIES;
+	req->lpm	= sch->schib.pmcw.pam & sch->opm;
+	req->data	= data;
+	req->callback	= stlck_callback;
+	stlck_build_cp(cdev, buf1, buf2);
+	ccw_request_start(cdev);
+}
+

drivers/s390/cio/device_status.c

@@ -0,0 +1,397 @@
+/*
+ *    Copyright IBM Corp. 2002
+ *    Author(s): Cornelia Huck (cornelia.huck@de.ibm.com)
+ *		 Martin Schwidefsky (schwidefsky@de.ibm.com)
+ *
+ * Status accumulation and basic sense functions.
+ */
+
+#include <linux/module.h>
+#include <linux/init.h>
+
+#include <asm/ccwdev.h>
+#include <asm/cio.h>
+
+#include "cio.h"
+#include "cio_debug.h"
+#include "css.h"
+#include "device.h"
+#include "ioasm.h"
+#include "io_sch.h"
+
+/*
+ * Check for any kind of channel or interface control check but don't
+ * issue the message for the console device
+ */
+static void
+ccw_device_msg_control_check(struct ccw_device *cdev, struct irb *irb)
+{
+	char dbf_text[15];
+
+	if (!scsw_is_valid_cstat(&irb->scsw) ||
+	    !(scsw_cstat(&irb->scsw) & (SCHN_STAT_CHN_DATA_CHK |
+	      SCHN_STAT_CHN_CTRL_CHK | SCHN_STAT_INTF_CTRL_CHK)))
+		return;
+	CIO_MSG_EVENT(0, "Channel-Check or Interface-Control-Check "
+		      "received"
+		      " ... device %04x on subchannel 0.%x.%04x, dev_stat "
+		      ": %02X sch_stat : %02X\n",
+		      cdev->private->dev_id.devno, cdev->private->schid.ssid,
+		      cdev->private->schid.sch_no,
+		      scsw_dstat(&irb->scsw), scsw_cstat(&irb->scsw));
+	sprintf(dbf_text, "chk%x", cdev->private->schid.sch_no);
+	CIO_TRACE_EVENT(0, dbf_text);
+	CIO_HEX_EVENT(0, irb, sizeof(struct irb));
+}
+
+/*
+ * Some paths became not operational (pno bit in scsw is set).
+ */
+static void
+ccw_device_path_notoper(struct ccw_device *cdev)
+{
+	struct subchannel *sch;
+
+	sch = to_subchannel(cdev->dev.parent);
+	if (cio_update_schib(sch))
+		goto doverify;
+
+	CIO_MSG_EVENT(0, "%s(0.%x.%04x) - path(s) %02x are "
+		      "not operational \n", __func__,
+		      sch->schid.ssid, sch->schid.sch_no,
+		      sch->schib.pmcw.pnom);
+
+	sch->lpm &= ~sch->schib.pmcw.pnom;
+doverify:
+	cdev->private->flags.doverify = 1;
+}
+
+/*
+ * Copy valid bits from the extended control word to device irb.
+ */
+static void
+ccw_device_accumulate_ecw(struct ccw_device *cdev, struct irb *irb)
+{
+	/*
+	 * Copy extended control bit if it is valid... yes there
+	 * are condition that have to be met for the extended control
+	 * bit to have meaning. Sick.
+	 */
+	cdev->private->irb.scsw.cmd.ectl = 0;
+	if ((irb->scsw.cmd.stctl & SCSW_STCTL_ALERT_STATUS) &&
+	    !(irb->scsw.cmd.stctl & SCSW_STCTL_INTER_STATUS))
+		cdev->private->irb.scsw.cmd.ectl = irb->scsw.cmd.ectl;
+	/* Check if extended control word is valid. */
+	if (!cdev->private->irb.scsw.cmd.ectl)
+		return;
+	/* Copy concurrent sense / model dependent information. */
+	memcpy (&cdev->private->irb.ecw, irb->ecw, sizeof (irb->ecw));
+}
+
+/*
+ * Check if extended status word is valid.
+ */
+static int
+ccw_device_accumulate_esw_valid(struct irb *irb)
+{
+	if (!irb->scsw.cmd.eswf &&
+	    (irb->scsw.cmd.stctl == SCSW_STCTL_STATUS_PEND))
+		return 0;
+	if (irb->scsw.cmd.stctl ==
+			(SCSW_STCTL_INTER_STATUS|SCSW_STCTL_STATUS_PEND) &&
+	    !(irb->scsw.cmd.actl & SCSW_ACTL_SUSPENDED))
+		return 0;
+	return 1;
+}
+
+/*
+ * Copy valid bits from the extended status word to device irb.
+ */
+static void
+ccw_device_accumulate_esw(struct ccw_device *cdev, struct irb *irb)
+{
+	struct irb *cdev_irb;
+	struct sublog *cdev_sublog, *sublog;
+
+	if (!ccw_device_accumulate_esw_valid(irb))
+		return;
+
+	cdev_irb = &cdev->private->irb;
+
+	/* Copy last path used mask. */
+	cdev_irb->esw.esw1.lpum = irb->esw.esw1.lpum;
+
+	/* Copy subchannel logout information if esw is of format 0. */
+	if (irb->scsw.cmd.eswf) {
+		cdev_sublog = &cdev_irb->esw.esw0.sublog;
+		sublog = &irb->esw.esw0.sublog;
+		/* Copy extended status flags. */
+		cdev_sublog->esf = sublog->esf;
+		/*
+		 * Copy fields that have a meaning for channel data check
+		 * channel control check and interface control check.
+		 */
+		if (irb->scsw.cmd.cstat & (SCHN_STAT_CHN_DATA_CHK |
+				       SCHN_STAT_CHN_CTRL_CHK |
+				       SCHN_STAT_INTF_CTRL_CHK)) {
+			/* Copy ancillary report bit. */
+			cdev_sublog->arep = sublog->arep;
+			/* Copy field-validity-flags. */
+			cdev_sublog->fvf = sublog->fvf;
+			/* Copy storage access code. */
+			cdev_sublog->sacc = sublog->sacc;
+			/* Copy termination code. */
+			cdev_sublog->termc = sublog->termc;
+			/* Copy sequence code. */
+			cdev_sublog->seqc = sublog->seqc;
+		}
+		/* Copy device status check. */
+		cdev_sublog->devsc = sublog->devsc;
+		/* Copy secondary error. */
+		cdev_sublog->serr = sublog->serr;
+		/* Copy i/o-error alert. */
+		cdev_sublog->ioerr = sublog->ioerr;
+		/* Copy channel path timeout bit. */
+		if (irb->scsw.cmd.cstat & SCHN_STAT_INTF_CTRL_CHK)
+			cdev_irb->esw.esw0.erw.cpt = irb->esw.esw0.erw.cpt;
+		/* Copy failing storage address validity flag. */
+		cdev_irb->esw.esw0.erw.fsavf = irb->esw.esw0.erw.fsavf;
+		if (cdev_irb->esw.esw0.erw.fsavf) {
+			/* ... and copy the failing storage address. */
+			memcpy(cdev_irb->esw.esw0.faddr, irb->esw.esw0.faddr,
+			       sizeof (irb->esw.esw0.faddr));
+			/* ... and copy the failing storage address format. */
+			cdev_irb->esw.esw0.erw.fsaf = irb->esw.esw0.erw.fsaf;
+		}
+		/* Copy secondary ccw address validity bit. */
+		cdev_irb->esw.esw0.erw.scavf = irb->esw.esw0.erw.scavf;
+		if (irb->esw.esw0.erw.scavf)
+			/* ... and copy the secondary ccw address. */
+			cdev_irb->esw.esw0.saddr = irb->esw.esw0.saddr;
+		
+	}
+	/* FIXME: DCTI for format 2? */
+
+	/* Copy authorization bit. */
+	cdev_irb->esw.esw0.erw.auth = irb->esw.esw0.erw.auth;
+	/* Copy path verification required flag. */
+	cdev_irb->esw.esw0.erw.pvrf = irb->esw.esw0.erw.pvrf;
+	if (irb->esw.esw0.erw.pvrf)
+		cdev->private->flags.doverify = 1;
+	/* Copy concurrent sense bit. */
+	cdev_irb->esw.esw0.erw.cons = irb->esw.esw0.erw.cons;
+	if (irb->esw.esw0.erw.cons)
+		cdev_irb->esw.esw0.erw.scnt = irb->esw.esw0.erw.scnt;
+}
+
+/*
+ * Accumulate status from irb to devstat.
+ */
+void
+ccw_device_accumulate_irb(struct ccw_device *cdev, struct irb *irb)
+{
+	struct irb *cdev_irb;
+
+	/*
+	 * Check if the status pending bit is set in stctl.
+	 * If not, the remaining bit have no meaning and we must ignore them.
+	 * The esw is not meaningful as well...
+	 */
+	if (!(scsw_stctl(&irb->scsw) & SCSW_STCTL_STATUS_PEND))
+		return;
+
+	/* Check for channel checks and interface control checks. */
+	ccw_device_msg_control_check(cdev, irb);
+
+	/* Check for path not operational. */
+	if (scsw_is_valid_pno(&irb->scsw) && scsw_pno(&irb->scsw))
+		ccw_device_path_notoper(cdev);
+	/* No irb accumulation for transport mode irbs. */
+	if (scsw_is_tm(&irb->scsw)) {
+		memcpy(&cdev->private->irb, irb, sizeof(struct irb));
+		return;
+	}
+	/*
+	 * Don't accumulate unsolicited interrupts.
+	 */
+	if (!scsw_is_solicited(&irb->scsw))
+		return;
+
+	cdev_irb = &cdev->private->irb;
+
+	/*
+	 * If the clear function had been performed, all formerly pending
+	 * status at the subchannel has been cleared and we must not pass
+	 * intermediate accumulated status to the device driver.
+	 */
+	if (irb->scsw.cmd.fctl & SCSW_FCTL_CLEAR_FUNC)
+		memset(&cdev->private->irb, 0, sizeof(struct irb));
+
+	/* Copy bits which are valid only for the start function. */
+	if (irb->scsw.cmd.fctl & SCSW_FCTL_START_FUNC) {
+		/* Copy key. */
+		cdev_irb->scsw.cmd.key = irb->scsw.cmd.key;
+		/* Copy suspend control bit. */
+		cdev_irb->scsw.cmd.sctl = irb->scsw.cmd.sctl;
+		/* Accumulate deferred condition code. */
+		cdev_irb->scsw.cmd.cc |= irb->scsw.cmd.cc;
+		/* Copy ccw format bit. */
+		cdev_irb->scsw.cmd.fmt = irb->scsw.cmd.fmt;
+		/* Copy prefetch bit. */
+		cdev_irb->scsw.cmd.pfch = irb->scsw.cmd.pfch;
+		/* Copy initial-status-interruption-control. */
+		cdev_irb->scsw.cmd.isic = irb->scsw.cmd.isic;
+		/* Copy address limit checking control. */
+		cdev_irb->scsw.cmd.alcc = irb->scsw.cmd.alcc;
+		/* Copy suppress suspend bit. */
+		cdev_irb->scsw.cmd.ssi = irb->scsw.cmd.ssi;
+	}
+
+	/* Take care of the extended control bit and extended control word. */
+	ccw_device_accumulate_ecw(cdev, irb);
+	    
+	/* Accumulate function control. */
+	cdev_irb->scsw.cmd.fctl |= irb->scsw.cmd.fctl;
+	/* Copy activity control. */
+	cdev_irb->scsw.cmd.actl = irb->scsw.cmd.actl;
+	/* Accumulate status control. */
+	cdev_irb->scsw.cmd.stctl |= irb->scsw.cmd.stctl;
+	/*
+	 * Copy ccw address if it is valid. This is a bit simplified
+	 * but should be close enough for all practical purposes.
+	 */
+	if ((irb->scsw.cmd.stctl & SCSW_STCTL_PRIM_STATUS) ||
+	    ((irb->scsw.cmd.stctl ==
+	      (SCSW_STCTL_INTER_STATUS|SCSW_STCTL_STATUS_PEND)) &&
+	     (irb->scsw.cmd.actl & SCSW_ACTL_DEVACT) &&
+	     (irb->scsw.cmd.actl & SCSW_ACTL_SCHACT)) ||
+	    (irb->scsw.cmd.actl & SCSW_ACTL_SUSPENDED))
+		cdev_irb->scsw.cmd.cpa = irb->scsw.cmd.cpa;
+	/* Accumulate device status, but not the device busy flag. */
+	cdev_irb->scsw.cmd.dstat &= ~DEV_STAT_BUSY;
+	/* dstat is not always valid. */
+	if (irb->scsw.cmd.stctl &
+	    (SCSW_STCTL_PRIM_STATUS | SCSW_STCTL_SEC_STATUS
+	     | SCSW_STCTL_INTER_STATUS | SCSW_STCTL_ALERT_STATUS))
+		cdev_irb->scsw.cmd.dstat |= irb->scsw.cmd.dstat;
+	/* Accumulate subchannel status. */
+	cdev_irb->scsw.cmd.cstat |= irb->scsw.cmd.cstat;
+	/* Copy residual count if it is valid. */
+	if ((irb->scsw.cmd.stctl & SCSW_STCTL_PRIM_STATUS) &&
+	    (irb->scsw.cmd.cstat & ~(SCHN_STAT_PCI | SCHN_STAT_INCORR_LEN))
+	     == 0)
+		cdev_irb->scsw.cmd.count = irb->scsw.cmd.count;
+
+	/* Take care of bits in the extended status word. */
+	ccw_device_accumulate_esw(cdev, irb);
+
+	/*
+	 * Check whether we must issue a SENSE CCW ourselves if there is no
+	 * concurrent sense facility installed for the subchannel.
+	 * No sense is required if no delayed sense is pending
+	 * and we did not get a unit check without sense information.
+	 *
+	 * Note: We should check for ioinfo[irq]->flags.consns but VM
+	 *	 violates the ESA/390 architecture and doesn't present an
+	 *	 operand exception for virtual devices without concurrent
+	 *	 sense facility available/supported when enabling the
+	 *	 concurrent sense facility.
+	 */
+	if ((cdev_irb->scsw.cmd.dstat & DEV_STAT_UNIT_CHECK) &&
+	    !(cdev_irb->esw.esw0.erw.cons))
+		cdev->private->flags.dosense = 1;
+}
+
+/*
+ * Do a basic sense.
+ */
+int
+ccw_device_do_sense(struct ccw_device *cdev, struct irb *irb)
+{
+	struct subchannel *sch;
+	struct ccw1 *sense_ccw;
+	int rc;
+
+	sch = to_subchannel(cdev->dev.parent);
+
+	/* A sense is required, can we do it now ? */
+	if (scsw_actl(&irb->scsw) & (SCSW_ACTL_DEVACT | SCSW_ACTL_SCHACT))
+		/*
+		 * we received an Unit Check but we have no final
+		 *  status yet, therefore we must delay the SENSE
+		 *  processing. We must not report this intermediate
+		 *  status to the device interrupt handler.
+		 */
+		return -EBUSY;
+
+	/*
+	 * We have ending status but no sense information. Do a basic sense.
+	 */
+	sense_ccw = &to_io_private(sch)->sense_ccw;
+	sense_ccw->cmd_code = CCW_CMD_BASIC_SENSE;
+	sense_ccw->cda = (__u32) __pa(cdev->private->irb.ecw);
+	sense_ccw->count = SENSE_MAX_COUNT;
+	sense_ccw->flags = CCW_FLAG_SLI;
+
+	rc = cio_start(sch, sense_ccw, 0xff);
+	if (rc == -ENODEV || rc == -EACCES)
+		dev_fsm_event(cdev, DEV_EVENT_VERIFY);
+	return rc;
+}
+
+/*
+ * Add information from basic sense to devstat.
+ */
+void
+ccw_device_accumulate_basic_sense(struct ccw_device *cdev, struct irb *irb)
+{
+	/*
+	 * Check if the status pending bit is set in stctl.
+	 * If not, the remaining bit have no meaning and we must ignore them.
+	 * The esw is not meaningful as well...
+	 */
+	if (!(scsw_stctl(&irb->scsw) & SCSW_STCTL_STATUS_PEND))
+		return;
+
+	/* Check for channel checks and interface control checks. */
+	ccw_device_msg_control_check(cdev, irb);
+
+	/* Check for path not operational. */
+	if (scsw_is_valid_pno(&irb->scsw) && scsw_pno(&irb->scsw))
+		ccw_device_path_notoper(cdev);
+
+	if (!(irb->scsw.cmd.dstat & DEV_STAT_UNIT_CHECK) &&
+	    (irb->scsw.cmd.dstat & DEV_STAT_CHN_END)) {
+		cdev->private->irb.esw.esw0.erw.cons = 1;
+		cdev->private->flags.dosense = 0;
+	}
+	/* Check if path verification is required. */
+	if (ccw_device_accumulate_esw_valid(irb) &&
+	    irb->esw.esw0.erw.pvrf)
+		cdev->private->flags.doverify = 1;
+}
+
+/*
+ * This function accumulates the status into the private devstat and
+ * starts a basic sense if one is needed.
+ */
+int
+ccw_device_accumulate_and_sense(struct ccw_device *cdev, struct irb *irb)
+{
+	ccw_device_accumulate_irb(cdev, irb);
+	if ((irb->scsw.cmd.actl  & (SCSW_ACTL_DEVACT | SCSW_ACTL_SCHACT)) != 0)
+		return -EBUSY;
+	/* Check for basic sense. */
+	if (cdev->private->flags.dosense &&
+	    !(irb->scsw.cmd.dstat & DEV_STAT_UNIT_CHECK)) {
+		cdev->private->irb.esw.esw0.erw.cons = 1;
+		cdev->private->flags.dosense = 0;
+		return 0;
+	}
+	if (cdev->private->flags.dosense) {
+		ccw_device_do_sense(cdev, irb);
+		return -EBUSY;
+	}
+	return 0;
+}
+

drivers/s390/cio/eadm_sch.c

@@ -0,0 +1,418 @@
+/*
+ * Driver for s390 eadm subchannels
+ *
+ * Copyright IBM Corp. 2012
+ * Author(s): Sebastian Ott <sebott@linux.vnet.ibm.com>
+ */
+
+#include <linux/kernel_stat.h>
+#include <linux/completion.h>
+#include <linux/workqueue.h>
+#include <linux/spinlock.h>
+#include <linux/device.h>
+#include <linux/module.h>
+#include <linux/timer.h>
+#include <linux/slab.h>
+#include <linux/list.h>
+
+#include <asm/css_chars.h>
+#include <asm/debug.h>
+#include <asm/isc.h>
+#include <asm/cio.h>
+#include <asm/scsw.h>
+#include <asm/eadm.h>
+
+#include "eadm_sch.h"
+#include "ioasm.h"
+#include "cio.h"
+#include "css.h"
+#include "orb.h"
+
+MODULE_DESCRIPTION("driver for s390 eadm subchannels");
+MODULE_LICENSE("GPL");
+
+#define EADM_TIMEOUT (5 * HZ)
+static DEFINE_SPINLOCK(list_lock);
+static LIST_HEAD(eadm_list);
+
+static debug_info_t *eadm_debug;
+
+#define EADM_LOG(imp, txt) do {					\
+		debug_text_event(eadm_debug, imp, txt);		\
+	} while (0)
+
+static void EADM_LOG_HEX(int level, void *data, int length)
+{
+	if (!debug_level_enabled(eadm_debug, level))
+		return;
+	while (length > 0) {
+		debug_event(eadm_debug, level, data, length);
+		length -= eadm_debug->buf_size;
+		data += eadm_debug->buf_size;
+	}
+}
+
+static void orb_init(union orb *orb)
+{
+	memset(orb, 0, sizeof(union orb));
+	orb->eadm.compat1 = 1;
+	orb->eadm.compat2 = 1;
+	orb->eadm.fmt = 1;
+	orb->eadm.x = 1;
+}
+
+static int eadm_subchannel_start(struct subchannel *sch, struct aob *aob)
+{
+	union orb *orb = &get_eadm_private(sch)->orb;
+	int cc;
+
+	orb_init(orb);
+	orb->eadm.aob = (u32)__pa(aob);
+	orb->eadm.intparm = (u32)(addr_t)sch;
+	orb->eadm.key = PAGE_DEFAULT_KEY >> 4;
+
+	EADM_LOG(6, "start");
+	EADM_LOG_HEX(6, &sch->schid, sizeof(sch->schid));
+
+	cc = ssch(sch->schid, orb);
+	switch (cc) {
+	case 0:
+		sch->schib.scsw.eadm.actl |= SCSW_ACTL_START_PEND;
+		break;
+	case 1:		/* status pending */
+	case 2:		/* busy */
+		return -EBUSY;
+	case 3:		/* not operational */
+		return -ENODEV;
+	}
+	return 0;
+}
+
+static int eadm_subchannel_clear(struct subchannel *sch)
+{
+	int cc;
+
+	cc = csch(sch->schid);
+	if (cc)
+		return -ENODEV;
+
+	sch->schib.scsw.eadm.actl |= SCSW_ACTL_CLEAR_PEND;
+	return 0;
+}
+
+static void eadm_subchannel_timeout(unsigned long data)
+{
+	struct subchannel *sch = (struct subchannel *) data;
+
+	spin_lock_irq(sch->lock);
+	EADM_LOG(1, "timeout");
+	EADM_LOG_HEX(1, &sch->schid, sizeof(sch->schid));
+	if (eadm_subchannel_clear(sch))
+		EADM_LOG(0, "clear failed");
+	spin_unlock_irq(sch->lock);
+}
+
+static void eadm_subchannel_set_timeout(struct subchannel *sch, int expires)
+{
+	struct eadm_private *private = get_eadm_private(sch);
+
+	if (expires == 0) {
+		del_timer(&private->timer);
+		return;
+	}
+	if (timer_pending(&private->timer)) {
+		if (mod_timer(&private->timer, jiffies + expires))
+			return;
+	}
+	private->timer.function = eadm_subchannel_timeout;
+	private->timer.data = (unsigned long) sch;
+	private->timer.expires = jiffies + expires;
+	add_timer(&private->timer);
+}
+
+static void eadm_subchannel_irq(struct subchannel *sch)
+{
+	struct eadm_private *private = get_eadm_private(sch);
+	struct eadm_scsw *scsw = &sch->schib.scsw.eadm;
+	struct irb *irb = this_cpu_ptr(&cio_irb);
+	int error = 0;
+
+	EADM_LOG(6, "irq");
+	EADM_LOG_HEX(6, irb, sizeof(*irb));
+
+	inc_irq_stat(IRQIO_ADM);
+
+	if ((scsw->stctl & (SCSW_STCTL_ALERT_STATUS | SCSW_STCTL_STATUS_PEND))
+	    && scsw->eswf == 1 && irb->esw.eadm.erw.r)
+		error = -EIO;
+
+	if (scsw->fctl & SCSW_FCTL_CLEAR_FUNC)
+		error = -ETIMEDOUT;
+
+	eadm_subchannel_set_timeout(sch, 0);
+
+	if (private->state != EADM_BUSY) {
+		EADM_LOG(1, "irq unsol");
+		EADM_LOG_HEX(1, irb, sizeof(*irb));
+		private->state = EADM_NOT_OPER;
+		css_sched_sch_todo(sch, SCH_TODO_EVAL);
+		return;
+	}
+	scm_irq_handler((struct aob *)(unsigned long)scsw->aob, error);
+	private->state = EADM_IDLE;
+
+	if (private->completion)
+		complete(private->completion);
+}
+
+static struct subchannel *eadm_get_idle_sch(void)
+{
+	struct eadm_private *private;
+	struct subchannel *sch;
+	unsigned long flags;
+
+	spin_lock_irqsave(&list_lock, flags);
+	list_for_each_entry(private, &eadm_list, head) {
+		sch = private->sch;
+		spin_lock(sch->lock);
+		if (private->state == EADM_IDLE) {
+			private->state = EADM_BUSY;
+			list_move_tail(&private->head, &eadm_list);
+			spin_unlock(sch->lock);
+			spin_unlock_irqrestore(&list_lock, flags);
+
+			return sch;
+		}
+		spin_unlock(sch->lock);
+	}
+	spin_unlock_irqrestore(&list_lock, flags);
+
+	return NULL;
+}
+
+int eadm_start_aob(struct aob *aob)
+{
+	struct eadm_private *private;
+	struct subchannel *sch;
+	unsigned long flags;
+	int ret;
+
+	sch = eadm_get_idle_sch();
+	if (!sch)
+		return -EBUSY;
+
+	spin_lock_irqsave(sch->lock, flags);
+	eadm_subchannel_set_timeout(sch, EADM_TIMEOUT);
+	ret = eadm_subchannel_start(sch, aob);
+	if (!ret)
+		goto out_unlock;
+
+	/* Handle start subchannel failure. */
+	eadm_subchannel_set_timeout(sch, 0);
+	private = get_eadm_private(sch);
+	private->state = EADM_NOT_OPER;
+	css_sched_sch_todo(sch, SCH_TODO_EVAL);
+
+out_unlock:
+	spin_unlock_irqrestore(sch->lock, flags);
+
+	return ret;
+}
+EXPORT_SYMBOL_GPL(eadm_start_aob);
+
+static int eadm_subchannel_probe(struct subchannel *sch)
+{
+	struct eadm_private *private;
+	int ret;
+
+	private = kzalloc(sizeof(*private), GFP_KERNEL | GFP_DMA);
+	if (!private)
+		return -ENOMEM;
+
+	INIT_LIST_HEAD(&private->head);
+	init_timer(&private->timer);
+
+	spin_lock_irq(sch->lock);
+	set_eadm_private(sch, private);
+	private->state = EADM_IDLE;
+	private->sch = sch;
+	sch->isc = EADM_SCH_ISC;
+	ret = cio_enable_subchannel(sch, (u32)(unsigned long)sch);
+	if (ret) {
+		set_eadm_private(sch, NULL);
+		spin_unlock_irq(sch->lock);
+		kfree(private);
+		goto out;
+	}
+	spin_unlock_irq(sch->lock);
+
+	spin_lock_irq(&list_lock);
+	list_add(&private->head, &eadm_list);
+	spin_unlock_irq(&list_lock);
+
+	if (dev_get_uevent_suppress(&sch->dev)) {
+		dev_set_uevent_suppress(&sch->dev, 0);
+		kobject_uevent(&sch->dev.kobj, KOBJ_ADD);
+	}
+out:
+	return ret;
+}
+
+static void eadm_quiesce(struct subchannel *sch)
+{
+	struct eadm_private *private = get_eadm_private(sch);
+	DECLARE_COMPLETION_ONSTACK(completion);
+	int ret;
+
+	spin_lock_irq(sch->lock);
+	if (private->state != EADM_BUSY)
+		goto disable;
+
+	if (eadm_subchannel_clear(sch))
+		goto disable;
+
+	private->completion = &completion;
+	spin_unlock_irq(sch->lock);
+
+	wait_for_completion_io(&completion);
+
+	spin_lock_irq(sch->lock);
+	private->completion = NULL;
+
+disable:
+	eadm_subchannel_set_timeout(sch, 0);
+	do {
+		ret = cio_disable_subchannel(sch);
+	} while (ret == -EBUSY);
+
+	spin_unlock_irq(sch->lock);
+}
+
+static int eadm_subchannel_remove(struct subchannel *sch)
+{
+	struct eadm_private *private = get_eadm_private(sch);
+
+	spin_lock_irq(&list_lock);
+	list_del(&private->head);
+	spin_unlock_irq(&list_lock);
+
+	eadm_quiesce(sch);
+
+	spin_lock_irq(sch->lock);
+	set_eadm_private(sch, NULL);
+	spin_unlock_irq(sch->lock);
+
+	kfree(private);
+
+	return 0;
+}
+
+static void eadm_subchannel_shutdown(struct subchannel *sch)
+{
+	eadm_quiesce(sch);
+}
+
+static int eadm_subchannel_freeze(struct subchannel *sch)
+{
+	return cio_disable_subchannel(sch);
+}
+
+static int eadm_subchannel_restore(struct subchannel *sch)
+{
+	return cio_enable_subchannel(sch, (u32)(unsigned long)sch);
+}
+
+/**
+ * eadm_subchannel_sch_event - process subchannel event
+ * @sch: subchannel
+ * @process: non-zero if function is called in process context
+ *
+ * An unspecified event occurred for this subchannel. Adjust data according
+ * to the current operational state of the subchannel. Return zero when the
+ * event has been handled sufficiently or -EAGAIN when this function should
+ * be called again in process context.
+ */
+static int eadm_subchannel_sch_event(struct subchannel *sch, int process)
+{
+	struct eadm_private *private;
+	unsigned long flags;
+	int ret = 0;
+
+	spin_lock_irqsave(sch->lock, flags);
+	if (!device_is_registered(&sch->dev))
+		goto out_unlock;
+
+	if (work_pending(&sch->todo_work))
+		goto out_unlock;
+
+	if (cio_update_schib(sch)) {
+		css_sched_sch_todo(sch, SCH_TODO_UNREG);
+		goto out_unlock;
+	}
+	private = get_eadm_private(sch);
+	if (private->state == EADM_NOT_OPER)
+		private->state = EADM_IDLE;
+
+out_unlock:
+	spin_unlock_irqrestore(sch->lock, flags);
+
+	return ret;
+}
+
+static struct css_device_id eadm_subchannel_ids[] = {
+	{ .match_flags = 0x1, .type = SUBCHANNEL_TYPE_ADM, },
+	{ /* end of list */ },
+};
+MODULE_DEVICE_TABLE(css, eadm_subchannel_ids);
+
+static struct css_driver eadm_subchannel_driver = {
+	.drv = {
+		.name = "eadm_subchannel",
+		.owner = THIS_MODULE,
+	},
+	.subchannel_type = eadm_subchannel_ids,
+	.irq = eadm_subchannel_irq,
+	.probe = eadm_subchannel_probe,
+	.remove = eadm_subchannel_remove,
+	.shutdown = eadm_subchannel_shutdown,
+	.sch_event = eadm_subchannel_sch_event,
+	.freeze = eadm_subchannel_freeze,
+	.thaw = eadm_subchannel_restore,
+	.restore = eadm_subchannel_restore,
+};
+
+static int __init eadm_sch_init(void)
+{
+	int ret;
+
+	if (!css_general_characteristics.eadm)
+		return -ENXIO;
+
+	eadm_debug = debug_register("eadm_log", 16, 1, 16);
+	if (!eadm_debug)
+		return -ENOMEM;
+
+	debug_register_view(eadm_debug, &debug_hex_ascii_view);
+	debug_set_level(eadm_debug, 2);
+
+	isc_register(EADM_SCH_ISC);
+	ret = css_driver_register(&eadm_subchannel_driver);
+	if (ret)
+		goto cleanup;
+
+	return ret;
+
+cleanup:
+	isc_unregister(EADM_SCH_ISC);
+	debug_unregister(eadm_debug);
+	return ret;
+}
+
+static void __exit eadm_sch_exit(void)
+{
+	css_driver_unregister(&eadm_subchannel_driver);
+	isc_unregister(EADM_SCH_ISC);
+	debug_unregister(eadm_debug);
+}
+module_init(eadm_sch_init);
+module_exit(eadm_sch_exit);

drivers/s390/cio/eadm_sch.h

@@ -0,0 +1,22 @@
+#ifndef EADM_SCH_H
+#define EADM_SCH_H
+
+#include <linux/completion.h>
+#include <linux/device.h>
+#include <linux/timer.h>
+#include <linux/list.h>
+#include "orb.h"
+
+struct eadm_private {
+	union orb orb;
+	enum {EADM_IDLE, EADM_BUSY, EADM_NOT_OPER} state;
+	struct completion *completion;
+	struct subchannel *sch;
+	struct timer_list timer;
+	struct list_head head;
+} __aligned(8);
+
+#define get_eadm_private(n) ((struct eadm_private *)dev_get_drvdata(&n->dev))
+#define set_eadm_private(n, p) (dev_set_drvdata(&n->dev, p))
+
+#endif

drivers/s390/cio/fcx.c

@@ -0,0 +1,350 @@
+/*
+ *  Functions for assembling fcx enabled I/O control blocks.
+ *
+ *    Copyright IBM Corp. 2008
+ *    Author(s): Peter Oberparleiter <peter.oberparleiter@de.ibm.com>
+ */
+
+#include <linux/kernel.h>
+#include <linux/types.h>
+#include <linux/string.h>
+#include <linux/errno.h>
+#include <linux/err.h>
+#include <linux/module.h>
+#include <asm/fcx.h>
+#include "cio.h"
+
+/**
+ * tcw_get_intrg - return pointer to associated interrogate tcw
+ * @tcw: pointer to the original tcw
+ *
+ * Return a pointer to the interrogate tcw associated with the specified tcw
+ * or %NULL if there is no associated interrogate tcw.
+ */
+struct tcw *tcw_get_intrg(struct tcw *tcw)
+{
+	return (struct tcw *) ((addr_t) tcw->intrg);
+}
+EXPORT_SYMBOL(tcw_get_intrg);
+
+/**
+ * tcw_get_data - return pointer to input/output data associated with tcw
+ * @tcw: pointer to the tcw
+ *
+ * Return the input or output data address specified in the tcw depending
+ * on whether the r-bit or the w-bit is set. If neither bit is set, return
+ * %NULL.
+ */
+void *tcw_get_data(struct tcw *tcw)
+{
+	if (tcw->r)
+		return (void *) ((addr_t) tcw->input);
+	if (tcw->w)
+		return (void *) ((addr_t) tcw->output);
+	return NULL;
+}
+EXPORT_SYMBOL(tcw_get_data);
+
+/**
+ * tcw_get_tccb - return pointer to tccb associated with tcw
+ * @tcw: pointer to the tcw
+ *
+ * Return pointer to the tccb associated with this tcw.
+ */
+struct tccb *tcw_get_tccb(struct tcw *tcw)
+{
+	return (struct tccb *) ((addr_t) tcw->tccb);
+}
+EXPORT_SYMBOL(tcw_get_tccb);
+
+/**
+ * tcw_get_tsb - return pointer to tsb associated with tcw
+ * @tcw: pointer to the tcw
+ *
+ * Return pointer to the tsb associated with this tcw.
+ */
+struct tsb *tcw_get_tsb(struct tcw *tcw)
+{
+	return (struct tsb *) ((addr_t) tcw->tsb);
+}
+EXPORT_SYMBOL(tcw_get_tsb);
+
+/**
+ * tcw_init - initialize tcw data structure
+ * @tcw: pointer to the tcw to be initialized
+ * @r: initial value of the r-bit
+ * @w: initial value of the w-bit
+ *
+ * Initialize all fields of the specified tcw data structure with zero and
+ * fill in the format, flags, r and w fields.
+ */
+void tcw_init(struct tcw *tcw, int r, int w)
+{
+	memset(tcw, 0, sizeof(struct tcw));
+	tcw->format = TCW_FORMAT_DEFAULT;
+	tcw->flags = TCW_FLAGS_TIDAW_FORMAT(TCW_TIDAW_FORMAT_DEFAULT);
+	if (r)
+		tcw->r = 1;
+	if (w)
+		tcw->w = 1;
+}
+EXPORT_SYMBOL(tcw_init);
+
+static inline size_t tca_size(struct tccb *tccb)
+{
+	return tccb->tcah.tcal - 12;
+}
+
+static u32 calc_dcw_count(struct tccb *tccb)
+{
+	int offset;
+	struct dcw *dcw;
+	u32 count = 0;
+	size_t size;
+
+	size = tca_size(tccb);
+	for (offset = 0; offset < size;) {
+		dcw = (struct dcw *) &tccb->tca[offset];
+		count += dcw->count;
+		if (!(dcw->flags & DCW_FLAGS_CC))
+			break;
+		offset += sizeof(struct dcw) + ALIGN((int) dcw->cd_count, 4);
+	}
+	return count;
+}
+
+static u32 calc_cbc_size(struct tidaw *tidaw, int num)
+{
+	int i;
+	u32 cbc_data;
+	u32 cbc_count = 0;
+	u64 data_count = 0;
+
+	for (i = 0; i < num; i++) {
+		if (tidaw[i].flags & TIDAW_FLAGS_LAST)
+			break;
+		/* TODO: find out if padding applies to total of data
+		 * transferred or data transferred by this tidaw. Assumption:
+		 * applies to total. */
+		data_count += tidaw[i].count;
+		if (tidaw[i].flags & TIDAW_FLAGS_INSERT_CBC) {
+			cbc_data = 4 + ALIGN(data_count, 4) - data_count;
+			cbc_count += cbc_data;
+			data_count += cbc_data;
+		}
+	}
+	return cbc_count;
+}
+
+/**
+ * tcw_finalize - finalize tcw length fields and tidaw list
+ * @tcw: pointer to the tcw
+ * @num_tidaws: the number of tidaws used to address input/output data or zero
+ * if no tida is used
+ *
+ * Calculate the input-/output-count and tccbl field in the tcw, add a
+ * tcat the tccb and terminate the data tidaw list if used.
+ *
+ * Note: in case input- or output-tida is used, the tidaw-list must be stored
+ * in contiguous storage (no ttic). The tcal field in the tccb must be
+ * up-to-date.
+ */
+void tcw_finalize(struct tcw *tcw, int num_tidaws)
+{
+	struct tidaw *tidaw;
+	struct tccb *tccb;
+	struct tccb_tcat *tcat;
+	u32 count;
+
+	/* Terminate tidaw list. */
+	tidaw = tcw_get_data(tcw);
+	if (num_tidaws > 0)
+		tidaw[num_tidaws - 1].flags |= TIDAW_FLAGS_LAST;
+	/* Add tcat to tccb. */
+	tccb = tcw_get_tccb(tcw);
+	tcat = (struct tccb_tcat *) &tccb->tca[tca_size(tccb)];
+	memset(tcat, 0, sizeof(*tcat));
+	/* Calculate tcw input/output count and tcat transport count. */
+	count = calc_dcw_count(tccb);
+	if (tcw->w && (tcw->flags & TCW_FLAGS_OUTPUT_TIDA))
+		count += calc_cbc_size(tidaw, num_tidaws);
+	if (tcw->r)
+		tcw->input_count = count;
+	else if (tcw->w)
+		tcw->output_count = count;
+	tcat->count = ALIGN(count, 4) + 4;
+	/* Calculate tccbl. */
+	tcw->tccbl = (sizeof(struct tccb) + tca_size(tccb) +
+		      sizeof(struct tccb_tcat) - 20) >> 2;
+}
+EXPORT_SYMBOL(tcw_finalize);
+
+/**
+ * tcw_set_intrg - set the interrogate tcw address of a tcw
+ * @tcw: the tcw address
+ * @intrg_tcw: the address of the interrogate tcw
+ *
+ * Set the address of the interrogate tcw in the specified tcw.
+ */
+void tcw_set_intrg(struct tcw *tcw, struct tcw *intrg_tcw)
+{
+	tcw->intrg = (u32) ((addr_t) intrg_tcw);
+}
+EXPORT_SYMBOL(tcw_set_intrg);
+
+/**
+ * tcw_set_data - set data address and tida flag of a tcw
+ * @tcw: the tcw address
+ * @data: the data address
+ * @use_tidal: zero of the data address specifies a contiguous block of data,
+ * non-zero if it specifies a list if tidaws.
+ *
+ * Set the input/output data address of a tcw (depending on the value of the
+ * r-flag and w-flag). If @use_tidal is non-zero, the corresponding tida flag
+ * is set as well.
+ */
+void tcw_set_data(struct tcw *tcw, void *data, int use_tidal)
+{
+	if (tcw->r) {
+		tcw->input = (u64) ((addr_t) data);
+		if (use_tidal)
+			tcw->flags |= TCW_FLAGS_INPUT_TIDA;
+	} else if (tcw->w) {
+		tcw->output = (u64) ((addr_t) data);
+		if (use_tidal)
+			tcw->flags |= TCW_FLAGS_OUTPUT_TIDA;
+	}
+}
+EXPORT_SYMBOL(tcw_set_data);
+
+/**
+ * tcw_set_tccb - set tccb address of a tcw
+ * @tcw: the tcw address
+ * @tccb: the tccb address
+ *
+ * Set the address of the tccb in the specified tcw.
+ */
+void tcw_set_tccb(struct tcw *tcw, struct tccb *tccb)
+{
+	tcw->tccb = (u64) ((addr_t) tccb);
+}
+EXPORT_SYMBOL(tcw_set_tccb);
+
+/**
+ * tcw_set_tsb - set tsb address of a tcw
+ * @tcw: the tcw address
+ * @tsb: the tsb address
+ *
+ * Set the address of the tsb in the specified tcw.
+ */
+void tcw_set_tsb(struct tcw *tcw, struct tsb *tsb)
+{
+	tcw->tsb = (u64) ((addr_t) tsb);
+}
+EXPORT_SYMBOL(tcw_set_tsb);
+
+/**
+ * tccb_init - initialize tccb
+ * @tccb: the tccb address
+ * @size: the maximum size of the tccb
+ * @sac: the service-action-code to be user
+ *
+ * Initialize the header of the specified tccb by resetting all values to zero
+ * and filling in defaults for format, sac and initial tcal fields.
+ */
+void tccb_init(struct tccb *tccb, size_t size, u32 sac)
+{
+	memset(tccb, 0, size);
+	tccb->tcah.format = TCCB_FORMAT_DEFAULT;
+	tccb->tcah.sac = sac;
+	tccb->tcah.tcal = 12;
+}
+EXPORT_SYMBOL(tccb_init);
+
+/**
+ * tsb_init - initialize tsb
+ * @tsb: the tsb address
+ *
+ * Initialize the specified tsb by resetting all values to zero.
+ */
+void tsb_init(struct tsb *tsb)
+{
+	memset(tsb, 0, sizeof(*tsb));
+}
+EXPORT_SYMBOL(tsb_init);
+
+/**
+ * tccb_add_dcw - add a dcw to the tccb
+ * @tccb: the tccb address
+ * @tccb_size: the maximum tccb size
+ * @cmd: the dcw command
+ * @flags: flags for the dcw
+ * @cd: pointer to control data for this dcw or NULL if none is required
+ * @cd_count: number of control data bytes for this dcw
+ * @count: number of data bytes for this dcw
+ *
+ * Add a new dcw to the specified tccb by writing the dcw information specified
+ * by @cmd, @flags, @cd, @cd_count and @count to the tca of the tccb. Return
+ * a pointer to the newly added dcw on success or -%ENOSPC if the new dcw
+ * would exceed the available space as defined by @tccb_size.
+ *
+ * Note: the tcal field of the tccb header will be updates to reflect added
+ * content.
+ */
+struct dcw *tccb_add_dcw(struct tccb *tccb, size_t tccb_size, u8 cmd, u8 flags,
+			 void *cd, u8 cd_count, u32 count)
+{
+	struct dcw *dcw;
+	int size;
+	int tca_offset;
+
+	/* Check for space. */
+	tca_offset = tca_size(tccb);
+	size = ALIGN(sizeof(struct dcw) + cd_count, 4);
+	if (sizeof(struct tccb_tcah) + tca_offset + size +
+	    sizeof(struct tccb_tcat) > tccb_size)
+		return ERR_PTR(-ENOSPC);
+	/* Add dcw to tca. */
+	dcw = (struct dcw *) &tccb->tca[tca_offset];
+	memset(dcw, 0, size);
+	dcw->cmd = cmd;
+	dcw->flags = flags;
+	dcw->count = count;
+	dcw->cd_count = cd_count;
+	if (cd)
+		memcpy(&dcw->cd[0], cd, cd_count);
+	tccb->tcah.tcal += size;
+	return dcw;
+}
+EXPORT_SYMBOL(tccb_add_dcw);
+
+/**
+ * tcw_add_tidaw - add a tidaw to a tcw
+ * @tcw: the tcw address
+ * @num_tidaws: the current number of tidaws
+ * @flags: flags for the new tidaw
+ * @addr: address value for the new tidaw
+ * @count: count value for the new tidaw
+ *
+ * Add a new tidaw to the input/output data tidaw-list of the specified tcw
+ * (depending on the value of the r-flag and w-flag) and return a pointer to
+ * the new tidaw.
+ *
+ * Note: the tidaw-list is assumed to be contiguous with no ttics. The caller
+ * must ensure that there is enough space for the new tidaw. The last-tidaw
+ * flag for the last tidaw in the list will be set by tcw_finalize.
+ */
+struct tidaw *tcw_add_tidaw(struct tcw *tcw, int num_tidaws, u8 flags,
+			    void *addr, u32 count)
+{
+	struct tidaw *tidaw;
+
+	/* Add tidaw to tidaw-list. */
+	tidaw = ((struct tidaw *) tcw_get_data(tcw)) + num_tidaws;
+	memset(tidaw, 0, sizeof(struct tidaw));
+	tidaw->flags = flags;
+	tidaw->count = count;
+	tidaw->addr = (u64) ((addr_t) addr);
+	return tidaw;
+}
+EXPORT_SYMBOL(tcw_add_tidaw);

drivers/s390/cio/idset.c

@@ -0,0 +1,131 @@
+/*
+ *    Copyright IBM Corp. 2007, 2012
+ *    Author(s): Peter Oberparleiter <peter.oberparleiter@de.ibm.com>
+ */
+
+#include <linux/vmalloc.h>
+#include <linux/bitmap.h>
+#include <linux/bitops.h>
+#include "idset.h"
+#include "css.h"
+
+struct idset {
+	int num_ssid;
+	int num_id;
+	unsigned long bitmap[0];
+};
+
+static inline unsigned long bitmap_size(int num_ssid, int num_id)
+{
+	return BITS_TO_LONGS(num_ssid * num_id) * sizeof(unsigned long);
+}
+
+static struct idset *idset_new(int num_ssid, int num_id)
+{
+	struct idset *set;
+
+	set = vmalloc(sizeof(struct idset) + bitmap_size(num_ssid, num_id));
+	if (set) {
+		set->num_ssid = num_ssid;
+		set->num_id = num_id;
+		memset(set->bitmap, 0, bitmap_size(num_ssid, num_id));
+	}
+	return set;
+}
+
+void idset_free(struct idset *set)
+{
+	vfree(set);
+}
+
+void idset_clear(struct idset *set)
+{
+	memset(set->bitmap, 0, bitmap_size(set->num_ssid, set->num_id));
+}
+
+void idset_fill(struct idset *set)
+{
+	memset(set->bitmap, 0xff, bitmap_size(set->num_ssid, set->num_id));
+}
+
+static inline void idset_add(struct idset *set, int ssid, int id)
+{
+	set_bit(ssid * set->num_id + id, set->bitmap);
+}
+
+static inline void idset_del(struct idset *set, int ssid, int id)
+{
+	clear_bit(ssid * set->num_id + id, set->bitmap);
+}
+
+static inline int idset_contains(struct idset *set, int ssid, int id)
+{
+	return test_bit(ssid * set->num_id + id, set->bitmap);
+}
+
+static inline int idset_get_first(struct idset *set, int *ssid, int *id)
+{
+	int bitnum;
+
+	bitnum = find_first_bit(set->bitmap, set->num_ssid * set->num_id);
+	if (bitnum >= set->num_ssid * set->num_id)
+		return 0;
+	*ssid = bitnum / set->num_id;
+	*id = bitnum % set->num_id;
+	return 1;
+}
+
+struct idset *idset_sch_new(void)
+{
+	return idset_new(max_ssid + 1, __MAX_SUBCHANNEL + 1);
+}
+
+void idset_sch_add(struct idset *set, struct subchannel_id schid)
+{
+	idset_add(set, schid.ssid, schid.sch_no);
+}
+
+void idset_sch_del(struct idset *set, struct subchannel_id schid)
+{
+	idset_del(set, schid.ssid, schid.sch_no);
+}
+
+/* Clear ids starting from @schid up to end of subchannel set. */
+void idset_sch_del_subseq(struct idset *set, struct subchannel_id schid)
+{
+	int pos = schid.ssid * set->num_id + schid.sch_no;
+
+	bitmap_clear(set->bitmap, pos, set->num_id - schid.sch_no);
+}
+
+int idset_sch_contains(struct idset *set, struct subchannel_id schid)
+{
+	return idset_contains(set, schid.ssid, schid.sch_no);
+}
+
+int idset_sch_get_first(struct idset *set, struct subchannel_id *schid)
+{
+	int ssid = 0;
+	int id = 0;
+	int rc;
+
+	rc = idset_get_first(set, &ssid, &id);
+	if (rc) {
+		init_subchannel_id(schid);
+		schid->ssid = ssid;
+		schid->sch_no = id;
+	}
+	return rc;
+}
+
+int idset_is_empty(struct idset *set)
+{
+	return bitmap_empty(set->bitmap, set->num_ssid * set->num_id);
+}
+
+void idset_add_set(struct idset *to, struct idset *from)
+{
+	int len = min(to->num_ssid * to->num_id, from->num_ssid * from->num_id);
+
+	bitmap_or(to->bitmap, to->bitmap, from->bitmap, len);
+}

drivers/s390/cio/idset.h

@@ -0,0 +1,26 @@
+/*
+ *    Copyright IBM Corp. 2007, 2012
+ *    Author(s): Peter Oberparleiter <peter.oberparleiter@de.ibm.com>
+ */
+
+#ifndef S390_IDSET_H
+#define S390_IDSET_H S390_IDSET_H
+
+#include <asm/schid.h>
+
+struct idset;
+
+void idset_free(struct idset *set);
+void idset_clear(struct idset *set);
+void idset_fill(struct idset *set);
+
+struct idset *idset_sch_new(void);
+void idset_sch_add(struct idset *set, struct subchannel_id id);
+void idset_sch_del(struct idset *set, struct subchannel_id id);
+void idset_sch_del_subseq(struct idset *set, struct subchannel_id schid);
+int idset_sch_contains(struct idset *set, struct subchannel_id id);
+int idset_sch_get_first(struct idset *set, struct subchannel_id *id);
+int idset_is_empty(struct idset *set);
+void idset_add_set(struct idset *to, struct idset *from);
+
+#endif /* S390_IDSET_H */

drivers/s390/cio/io_sch.h

@@ -0,0 +1,217 @@
+#ifndef S390_IO_SCH_H
+#define S390_IO_SCH_H
+
+#include <linux/types.h>
+#include <asm/schid.h>
+#include <asm/ccwdev.h>
+#include <asm/irq.h>
+#include "css.h"
+#include "orb.h"
+
+struct io_subchannel_private {
+	union orb orb;		/* operation request block */
+	struct ccw1 sense_ccw;	/* static ccw for sense command */
+	struct ccw_device *cdev;/* pointer to the child ccw device */
+	struct {
+		unsigned int suspend:1;	/* allow suspend */
+		unsigned int prefetch:1;/* deny prefetch */
+		unsigned int inter:1;	/* suppress intermediate interrupts */
+	} __packed options;
+} __aligned(8);
+
+#define to_io_private(n) ((struct io_subchannel_private *) \
+			  dev_get_drvdata(&(n)->dev))
+#define set_io_private(n, p) (dev_set_drvdata(&(n)->dev, p))
+
+static inline struct ccw_device *sch_get_cdev(struct subchannel *sch)
+{
+	struct io_subchannel_private *priv = to_io_private(sch);
+	return priv ? priv->cdev : NULL;
+}
+
+static inline void sch_set_cdev(struct subchannel *sch,
+				struct ccw_device *cdev)
+{
+	struct io_subchannel_private *priv = to_io_private(sch);
+	if (priv)
+		priv->cdev = cdev;
+}
+
+#define MAX_CIWS 8
+
+/*
+ * Possible status values for a CCW request's I/O.
+ */
+enum io_status {
+	IO_DONE,
+	IO_RUNNING,
+	IO_STATUS_ERROR,
+	IO_PATH_ERROR,
+	IO_REJECTED,
+	IO_KILLED
+};
+
+/**
+ * ccw_request - Internal CCW request.
+ * @cp: channel program to start
+ * @timeout: maximum allowable time in jiffies between start I/O and interrupt
+ * @maxretries: number of retries per I/O operation and path
+ * @lpm: mask of paths to use
+ * @check: optional callback that determines if results are final
+ * @filter: optional callback to adjust request status based on IRB data
+ * @callback: final callback
+ * @data: user-defined pointer passed to all callbacks
+ * @singlepath: if set, use only one path from @lpm per start I/O
+ * @cancel: non-zero if request was cancelled
+ * @done: non-zero if request was finished
+ * @mask: current path mask
+ * @retries: current number of retries
+ * @drc: delayed return code
+ */
+struct ccw_request {
+	struct ccw1 *cp;
+	unsigned long timeout;
+	u16 maxretries;
+	u8 lpm;
+	int (*check)(struct ccw_device *, void *);
+	enum io_status (*filter)(struct ccw_device *, void *, struct irb *,
+				 enum io_status);
+	void (*callback)(struct ccw_device *, void *, int);
+	void *data;
+	unsigned int singlepath:1;
+	/* These fields are used internally. */
+	unsigned int cancel:1;
+	unsigned int done:1;
+	u16 mask;
+	u16 retries;
+	int drc;
+} __attribute__((packed));
+
+/*
+ * sense-id response buffer layout
+ */
+struct senseid {
+	/* common part */
+	u8  reserved;	/* always 0x'FF' */
+	u16 cu_type;	/* control unit type */
+	u8  cu_model;	/* control unit model */
+	u16 dev_type;	/* device type */
+	u8  dev_model;	/* device model */
+	u8  unused;	/* padding byte */
+	/* extended part */
+	struct ciw ciw[MAX_CIWS];	/* variable # of CIWs */
+}  __attribute__ ((packed, aligned(4)));
+
+enum cdev_todo {
+	CDEV_TODO_NOTHING,
+	CDEV_TODO_ENABLE_CMF,
+	CDEV_TODO_REBIND,
+	CDEV_TODO_REGISTER,
+	CDEV_TODO_UNREG,
+	CDEV_TODO_UNREG_EVAL,
+};
+
+#define FAKE_CMD_IRB	1
+#define FAKE_TM_IRB	2
+
+struct ccw_device_private {
+	struct ccw_device *cdev;
+	struct subchannel *sch;
+	int state;		/* device state */
+	atomic_t onoff;
+	struct ccw_dev_id dev_id;	/* device id */
+	struct subchannel_id schid;	/* subchannel number */
+	struct ccw_request req;		/* internal I/O request */
+	int iretry;
+	u8 pgid_valid_mask;	/* mask of valid PGIDs */
+	u8 pgid_todo_mask;	/* mask of PGIDs to be adjusted */
+	u8 pgid_reset_mask;	/* mask of PGIDs which were reset */
+	u8 path_noirq_mask;	/* mask of paths for which no irq was
+				   received */
+	u8 path_notoper_mask;	/* mask of paths which were found
+				   not operable */
+	u8 path_gone_mask;	/* mask of paths, that became unavailable */
+	u8 path_new_mask;	/* mask of paths, that became available */
+	struct {
+		unsigned int fast:1;	/* post with "channel end" */
+		unsigned int repall:1;	/* report every interrupt status */
+		unsigned int pgroup:1;	/* do path grouping */
+		unsigned int force:1;	/* allow forced online */
+		unsigned int mpath:1;	/* do multipathing */
+	} __attribute__ ((packed)) options;
+	struct {
+		unsigned int esid:1;	    /* Ext. SenseID supported by HW */
+		unsigned int dosense:1;	    /* delayed SENSE required */
+		unsigned int doverify:1;    /* delayed path verification */
+		unsigned int donotify:1;    /* call notify function */
+		unsigned int recog_done:1;  /* dev. recog. complete */
+		unsigned int fake_irb:2;    /* deliver faked irb */
+		unsigned int resuming:1;    /* recognition while resume */
+		unsigned int pgroup:1;	    /* pathgroup is set up */
+		unsigned int mpath:1;	    /* multipathing is set up */
+		unsigned int pgid_unknown:1;/* unknown pgid state */
+		unsigned int initialized:1; /* set if initial reference held */
+	} __attribute__((packed)) flags;
+	unsigned long intparm;	/* user interruption parameter */
+	struct qdio_irq *qdio_data;
+	struct irb irb;		/* device status */
+	struct senseid senseid;	/* SenseID info */
+	struct pgid pgid[8];	/* path group IDs per chpid*/
+	struct ccw1 iccws[2];	/* ccws for SNID/SID/SPGID commands */
+	struct work_struct todo_work;
+	enum cdev_todo todo;
+	wait_queue_head_t wait_q;
+	struct timer_list timer;
+	void *cmb;			/* measurement information */
+	struct list_head cmb_list;	/* list of measured devices */
+	u64 cmb_start_time;		/* clock value of cmb reset */
+	void *cmb_wait;			/* deferred cmb enable/disable */
+	enum interruption_class int_class;
+};
+
+static inline int rsch(struct subchannel_id schid)
+{
+	register struct subchannel_id reg1 asm("1") = schid;
+	int ccode;
+
+	asm volatile(
+		"	rsch\n"
+		"	ipm	%0\n"
+		"	srl	%0,28"
+		: "=d" (ccode)
+		: "d" (reg1)
+		: "cc", "memory");
+	return ccode;
+}
+
+static inline int hsch(struct subchannel_id schid)
+{
+	register struct subchannel_id reg1 asm("1") = schid;
+	int ccode;
+
+	asm volatile(
+		"	hsch\n"
+		"	ipm	%0\n"
+		"	srl	%0,28"
+		: "=d" (ccode)
+		: "d" (reg1)
+		: "cc");
+	return ccode;
+}
+
+static inline int xsch(struct subchannel_id schid)
+{
+	register struct subchannel_id reg1 asm("1") = schid;
+	int ccode;
+
+	asm volatile(
+		"	.insn	rre,0xb2760000,%1,0\n"
+		"	ipm	%0\n"
+		"	srl	%0,28"
+		: "=d" (ccode)
+		: "d" (reg1)
+		: "cc");
+	return ccode;
+}
+
+#endif

drivers/s390/cio/ioasm.h

@@ -0,0 +1,167 @@
+#ifndef S390_CIO_IOASM_H
+#define S390_CIO_IOASM_H
+
+#include <asm/chpid.h>
+#include <asm/schid.h>
+#include "orb.h"
+#include "cio.h"
+
+/*
+ * TPI info structure
+ */
+struct tpi_info {
+	struct subchannel_id schid;
+	__u32 intparm;		 /* interruption parameter */
+	__u32 adapter_IO : 1;
+	__u32 reserved2	 : 1;
+	__u32 isc	 : 3;
+	__u32 reserved3	 : 12;
+	__u32 int_type	 : 3;
+	__u32 reserved4	 : 12;
+} __attribute__ ((packed));
+
+
+/*
+ * Some S390 specific IO instructions as inline
+ */
+
+static inline int stsch_err(struct subchannel_id schid, struct schib *addr)
+{
+	register struct subchannel_id reg1 asm ("1") = schid;
+	int ccode = -EIO;
+
+	asm volatile(
+		"	stsch	0(%3)\n"
+		"0:	ipm	%0\n"
+		"	srl	%0,28\n"
+		"1:\n"
+		EX_TABLE(0b,1b)
+		: "+d" (ccode), "=m" (*addr)
+		: "d" (reg1), "a" (addr)
+		: "cc");
+	return ccode;
+}
+
+static inline int msch(struct subchannel_id schid, struct schib *addr)
+{
+	register struct subchannel_id reg1 asm ("1") = schid;
+	int ccode;
+
+	asm volatile(
+		"	msch	0(%2)\n"
+		"	ipm	%0\n"
+		"	srl	%0,28"
+		: "=d" (ccode)
+		: "d" (reg1), "a" (addr), "m" (*addr)
+		: "cc");
+	return ccode;
+}
+
+static inline int msch_err(struct subchannel_id schid, struct schib *addr)
+{
+	register struct subchannel_id reg1 asm ("1") = schid;
+	int ccode = -EIO;
+
+	asm volatile(
+		"	msch	0(%2)\n"
+		"0:	ipm	%0\n"
+		"	srl	%0,28\n"
+		"1:\n"
+		EX_TABLE(0b,1b)
+		: "+d" (ccode)
+		: "d" (reg1), "a" (addr), "m" (*addr)
+		: "cc");
+	return ccode;
+}
+
+static inline int tsch(struct subchannel_id schid, struct irb *addr)
+{
+	register struct subchannel_id reg1 asm ("1") = schid;
+	int ccode;
+
+	asm volatile(
+		"	tsch	0(%3)\n"
+		"	ipm	%0\n"
+		"	srl	%0,28"
+		: "=d" (ccode), "=m" (*addr)
+		: "d" (reg1), "a" (addr)
+		: "cc");
+	return ccode;
+}
+
+static inline int ssch(struct subchannel_id schid, union orb *addr)
+{
+	register struct subchannel_id reg1 asm("1") = schid;
+	int ccode = -EIO;
+
+	asm volatile(
+		"	ssch	0(%2)\n"
+		"0:	ipm	%0\n"
+		"	srl	%0,28\n"
+		"1:\n"
+		EX_TABLE(0b, 1b)
+		: "+d" (ccode)
+		: "d" (reg1), "a" (addr), "m" (*addr)
+		: "cc", "memory");
+	return ccode;
+}
+
+static inline int csch(struct subchannel_id schid)
+{
+	register struct subchannel_id reg1 asm("1") = schid;
+	int ccode;
+
+	asm volatile(
+		"	csch\n"
+		"	ipm	%0\n"
+		"	srl	%0,28"
+		: "=d" (ccode)
+		: "d" (reg1)
+		: "cc");
+	return ccode;
+}
+
+static inline int tpi(struct tpi_info *addr)
+{
+	int ccode;
+
+	asm volatile(
+		"	tpi	0(%2)\n"
+		"	ipm	%0\n"
+		"	srl	%0,28"
+		: "=d" (ccode), "=m" (*addr)
+		: "a" (addr)
+		: "cc");
+	return ccode;
+}
+
+static inline int chsc(void *chsc_area)
+{
+	typedef struct { char _[4096]; } addr_type;
+	int cc;
+
+	asm volatile(
+		"	.insn	rre,0xb25f0000,%2,0\n"
+		"	ipm	%0\n"
+		"	srl	%0,28\n"
+		: "=d" (cc), "=m" (*(addr_type *) chsc_area)
+		: "d" (chsc_area), "m" (*(addr_type *) chsc_area)
+		: "cc");
+	return cc;
+}
+
+static inline int rchp(struct chp_id chpid)
+{
+	register struct chp_id reg1 asm ("1") = chpid;
+	int ccode;
+
+	asm volatile(
+		"	lr	1,%1\n"
+		"	rchp\n"
+		"	ipm	%0\n"
+		"	srl	%0,28"
+		: "=d" (ccode) : "d" (reg1) : "cc");
+	return ccode;
+}
+
+#endif

drivers/s390/cio/isc.c

@@ -0,0 +1,68 @@
+/*
+ * Functions for registration of I/O interruption subclasses on s390.
+ *
+ * Copyright IBM Corp. 2008
+ * Authors: Sebastian Ott <sebott@linux.vnet.ibm.com>
+ */
+
+#include <linux/spinlock.h>
+#include <linux/module.h>
+#include <asm/isc.h>
+
+static unsigned int isc_refs[MAX_ISC + 1];
+static DEFINE_SPINLOCK(isc_ref_lock);
+
+
+/**
+ * isc_register - register an I/O interruption subclass.
+ * @isc: I/O interruption subclass to register
+ *
+ * The number of users for @isc is increased. If this is the first user to
+ * register @isc, the corresponding I/O interruption subclass mask is enabled.
+ *
+ * Context:
+ *   This function must not be called in interrupt context.
+ */
+void isc_register(unsigned int isc)
+{
+	if (isc > MAX_ISC) {
+		WARN_ON(1);
+		return;
+	}
+
+	spin_lock(&isc_ref_lock);
+	if (isc_refs[isc] == 0)
+		ctl_set_bit(6, 31 - isc);
+	isc_refs[isc]++;
+	spin_unlock(&isc_ref_lock);
+}
+EXPORT_SYMBOL_GPL(isc_register);
+
+/**
+ * isc_unregister - unregister an I/O interruption subclass.
+ * @isc: I/O interruption subclass to unregister
+ *
+ * The number of users for @isc is decreased. If this is the last user to
+ * unregister @isc, the corresponding I/O interruption subclass mask is
+ * disabled.
+ * Note: This function must not be called if isc_register() hasn't been called
+ * before by the driver for @isc.
+ *
+ * Context:
+ *   This function must not be called in interrupt context.
+ */
+void isc_unregister(unsigned int isc)
+{
+	spin_lock(&isc_ref_lock);
+	/* check for misuse */
+	if (isc > MAX_ISC || isc_refs[isc] == 0) {
+		WARN_ON(1);
+		goto out_unlock;
+	}
+	if (isc_refs[isc] == 1)
+		ctl_clear_bit(6, 31 - isc);
+	isc_refs[isc]--;
+out_unlock:
+	spin_unlock(&isc_ref_lock);
+}
+EXPORT_SYMBOL_GPL(isc_unregister);

drivers/s390/cio/itcw.c

@@ -0,0 +1,369 @@
+/*
+ *  Functions for incremental construction of fcx enabled I/O control blocks.
+ *
+ *    Copyright IBM Corp. 2008
+ *    Author(s): Peter Oberparleiter <peter.oberparleiter@de.ibm.com>
+ */
+
+#include <linux/kernel.h>
+#include <linux/types.h>
+#include <linux/string.h>
+#include <linux/errno.h>
+#include <linux/err.h>
+#include <linux/module.h>
+#include <asm/fcx.h>
+#include <asm/itcw.h>
+
+/**
+ * struct itcw - incremental tcw helper data type
+ *
+ * This structure serves as a handle for the incremental construction of a
+ * tcw and associated tccb, tsb, data tidaw-list plus an optional interrogate
+ * tcw and associated data. The data structures are contained inside a single
+ * contiguous buffer provided by the user.
+ *
+ * The itcw construction functions take care of overall data integrity:
+ * - reset unused fields to zero
+ * - fill in required pointers
+ * - ensure required alignment for data structures
+ * - prevent data structures to cross 4k-byte boundary where required
+ * - calculate tccb-related length fields
+ * - optionally provide ready-made interrogate tcw and associated structures
+ *
+ * Restrictions apply to the itcws created with these construction functions:
+ * - tida only supported for data address, not for tccb
+ * - only contiguous tidaw-lists (no ttic)
+ * - total number of bytes required per itcw may not exceed 4k bytes
+ * - either read or write operation (may not work with r=0 and w=0)
+ *
+ * Example:
+ * struct itcw *itcw;
+ * void *buffer;
+ * size_t size;
+ *
+ * size = itcw_calc_size(1, 2, 0);
+ * buffer = kmalloc(size, GFP_KERNEL | GFP_DMA);
+ * if (!buffer)
+ *	return -ENOMEM;
+ * itcw = itcw_init(buffer, size, ITCW_OP_READ, 1, 2, 0);
+ * if (IS_ERR(itcw))
+ *	return PTR_ER(itcw);
+ * itcw_add_dcw(itcw, 0x2, 0, NULL, 0, 72);
+ * itcw_add_tidaw(itcw, 0, 0x30000, 20);
+ * itcw_add_tidaw(itcw, 0, 0x40000, 52);
+ * itcw_finalize(itcw);
+ *
+ */
+struct itcw {
+	struct tcw *tcw;
+	struct tcw *intrg_tcw;
+	int num_tidaws;
+	int max_tidaws;
+	int intrg_num_tidaws;
+	int intrg_max_tidaws;
+};
+
+/**
+ * itcw_get_tcw - return pointer to tcw associated with the itcw
+ * @itcw: address of the itcw
+ *
+ * Return pointer to the tcw associated with the itcw.
+ */
+struct tcw *itcw_get_tcw(struct itcw *itcw)
+{
+	return itcw->tcw;
+}
+EXPORT_SYMBOL(itcw_get_tcw);
+
+/**
+ * itcw_calc_size - return the size of an itcw with the given parameters
+ * @intrg: if non-zero, add an interrogate tcw
+ * @max_tidaws: maximum number of tidaws to be used for data addressing or zero
+ * if no tida is to be used.
+ * @intrg_max_tidaws: maximum number of tidaws to be used for data addressing
+ * by the interrogate tcw, if specified
+ *
+ * Calculate and return the number of bytes required to hold an itcw with the
+ * given parameters and assuming tccbs with maximum size.
+ *
+ * Note that the resulting size also contains bytes needed for alignment
+ * padding as well as padding to ensure that data structures don't cross a
+ * 4k-boundary where required.
+ */
+size_t itcw_calc_size(int intrg, int max_tidaws, int intrg_max_tidaws)
+{
+	size_t len;
+	int cross_count;
+
+	/* Main data. */
+	len = sizeof(struct itcw);
+	len += /* TCW */ sizeof(struct tcw) + /* TCCB */ TCCB_MAX_SIZE +
+	       /* TSB */ sizeof(struct tsb) +
+	       /* TIDAL */ max_tidaws * sizeof(struct tidaw);
+	/* Interrogate data. */
+	if (intrg) {
+		len += /* TCW */ sizeof(struct tcw) + /* TCCB */ TCCB_MAX_SIZE +
+		       /* TSB */ sizeof(struct tsb) +
+		       /* TIDAL */ intrg_max_tidaws * sizeof(struct tidaw);
+	}
+
+	/* Maximum required alignment padding. */
+	len += /* Initial TCW */ 63 + /* Interrogate TCCB */ 7;
+
+	/* TIDAW lists may not cross a 4k boundary. To cross a
+	 * boundary we need to add a TTIC TIDAW. We need to reserve
+	 * one additional TIDAW for a TTIC that we may need to add due
+	 * to the placement of the data chunk in memory, and a further
+	 * TIDAW for each page boundary that the TIDAW list may cross
+	 * due to it's own size.
+	 */
+	if (max_tidaws) {
+		cross_count = 1 + ((max_tidaws * sizeof(struct tidaw) - 1)
+				   >> PAGE_SHIFT);
+		len += cross_count * sizeof(struct tidaw);
+	}
+	if (intrg_max_tidaws) {
+		cross_count = 1 + ((intrg_max_tidaws * sizeof(struct tidaw) - 1)
+				   >> PAGE_SHIFT);
+		len += cross_count * sizeof(struct tidaw);
+	}
+	return len;
+}
+EXPORT_SYMBOL(itcw_calc_size);
+
+#define CROSS4K(x, l)	(((x) & ~4095) != ((x + l) & ~4095))
+
+static inline void *fit_chunk(addr_t *start, addr_t end, size_t len,
+			      int align, int check_4k)
+{
+	addr_t addr;
+
+	addr = ALIGN(*start, align);
+	if (check_4k && CROSS4K(addr, len)) {
+		addr = ALIGN(addr, 4096);
+		addr = ALIGN(addr, align);
+	}
+	if (addr + len > end)
+		return ERR_PTR(-ENOSPC);
+	*start = addr + len;
+	return (void *) addr;
+}
+
+/**
+ * itcw_init - initialize incremental tcw data structure
+ * @buffer: address of buffer to use for data structures
+ * @size: number of bytes in buffer
+ * @op: %ITCW_OP_READ for a read operation tcw, %ITCW_OP_WRITE for a write
+ * operation tcw
+ * @intrg: if non-zero, add and initialize an interrogate tcw
+ * @max_tidaws: maximum number of tidaws to be used for data addressing or zero
+ * if no tida is to be used.
+ * @intrg_max_tidaws: maximum number of tidaws to be used for data addressing
+ * by the interrogate tcw, if specified
+ *
+ * Prepare the specified buffer to be used as an incremental tcw, i.e. a
+ * helper data structure that can be used to construct a valid tcw by
+ * successive calls to other helper functions. Note: the buffer needs to be
+ * located below the 2G address limit. The resulting tcw has the following
+ * restrictions:
+ *  - no tccb tidal
+ *  - input/output tidal is contiguous (no ttic)
+ *  - total data should not exceed 4k
+ *  - tcw specifies either read or write operation
+ *
+ * On success, return pointer to the resulting incremental tcw data structure,
+ * ERR_PTR otherwise.
+ */
+struct itcw *itcw_init(void *buffer, size_t size, int op, int intrg,
+		       int max_tidaws, int intrg_max_tidaws)
+{
+	struct itcw *itcw;
+	void *chunk;
+	addr_t start;
+	addr_t end;
+	int cross_count;
+
+	/* Check for 2G limit. */
+	start = (addr_t) buffer;
+	end = start + size;
+	if (end > (1 << 31))
+		return ERR_PTR(-EINVAL);
+	memset(buffer, 0, size);
+	/* ITCW. */
+	chunk = fit_chunk(&start, end, sizeof(struct itcw), 1, 0);
+	if (IS_ERR(chunk))
+		return chunk;
+	itcw = chunk;
+	/* allow for TTIC tidaws that may be needed to cross a page boundary */
+	cross_count = 0;
+	if (max_tidaws)
+		cross_count = 1 + ((max_tidaws * sizeof(struct tidaw) - 1)
+				   >> PAGE_SHIFT);
+	itcw->max_tidaws = max_tidaws + cross_count;
+	cross_count = 0;
+	if (intrg_max_tidaws)
+		cross_count = 1 + ((intrg_max_tidaws * sizeof(struct tidaw) - 1)
+				   >> PAGE_SHIFT);
+	itcw->intrg_max_tidaws = intrg_max_tidaws + cross_count;
+	/* Main TCW. */
+	chunk = fit_chunk(&start, end, sizeof(struct tcw), 64, 0);
+	if (IS_ERR(chunk))
+		return chunk;
+	itcw->tcw = chunk;
+	tcw_init(itcw->tcw, (op == ITCW_OP_READ) ? 1 : 0,
+		 (op == ITCW_OP_WRITE) ? 1 : 0);
+	/* Interrogate TCW. */
+	if (intrg) {
+		chunk = fit_chunk(&start, end, sizeof(struct tcw), 64, 0);
+		if (IS_ERR(chunk))
+			return chunk;
+		itcw->intrg_tcw = chunk;
+		tcw_init(itcw->intrg_tcw, 1, 0);
+		tcw_set_intrg(itcw->tcw, itcw->intrg_tcw);
+	}
+	/* Data TIDAL. */
+	if (max_tidaws > 0) {
+		chunk = fit_chunk(&start, end, sizeof(struct tidaw) *
+				  itcw->max_tidaws, 16, 0);
+		if (IS_ERR(chunk))
+			return chunk;
+		tcw_set_data(itcw->tcw, chunk, 1);
+	}
+	/* Interrogate data TIDAL. */
+	if (intrg && (intrg_max_tidaws > 0)) {
+		chunk = fit_chunk(&start, end, sizeof(struct tidaw) *
+				  itcw->intrg_max_tidaws, 16, 0);
+		if (IS_ERR(chunk))
+			return chunk;
+		tcw_set_data(itcw->intrg_tcw, chunk, 1);
+	}
+	/* TSB. */
+	chunk = fit_chunk(&start, end, sizeof(struct tsb), 8, 0);
+	if (IS_ERR(chunk))
+		return chunk;
+	tsb_init(chunk);
+	tcw_set_tsb(itcw->tcw, chunk);
+	/* Interrogate TSB. */
+	if (intrg) {
+		chunk = fit_chunk(&start, end, sizeof(struct tsb), 8, 0);
+		if (IS_ERR(chunk))
+			return chunk;
+		tsb_init(chunk);
+		tcw_set_tsb(itcw->intrg_tcw, chunk);
+	}
+	/* TCCB. */
+	chunk = fit_chunk(&start, end, TCCB_MAX_SIZE, 8, 0);
+	if (IS_ERR(chunk))
+		return chunk;
+	tccb_init(chunk, TCCB_MAX_SIZE, TCCB_SAC_DEFAULT);
+	tcw_set_tccb(itcw->tcw, chunk);
+	/* Interrogate TCCB. */
+	if (intrg) {
+		chunk = fit_chunk(&start, end, TCCB_MAX_SIZE, 8, 0);
+		if (IS_ERR(chunk))
+			return chunk;
+		tccb_init(chunk, TCCB_MAX_SIZE, TCCB_SAC_INTRG);
+		tcw_set_tccb(itcw->intrg_tcw, chunk);
+		tccb_add_dcw(chunk, TCCB_MAX_SIZE, DCW_CMD_INTRG, 0, NULL,
+			     sizeof(struct dcw_intrg_data), 0);
+		tcw_finalize(itcw->intrg_tcw, 0);
+	}
+	return itcw;
+}
+EXPORT_SYMBOL(itcw_init);
+
+/**
+ * itcw_add_dcw - add a dcw to the itcw
+ * @itcw: address of the itcw
+ * @cmd: the dcw command
+ * @flags: flags for the dcw
+ * @cd: address of control data for this dcw or NULL if none is required
+ * @cd_count: number of control data bytes for this dcw
+ * @count: number of data bytes for this dcw
+ *
+ * Add a new dcw to the specified itcw by writing the dcw information specified
+ * by @cmd, @flags, @cd, @cd_count and @count to the tca of the tccb. Return
+ * a pointer to the newly added dcw on success or -%ENOSPC if the new dcw
+ * would exceed the available space.
+ *
+ * Note: the tcal field of the tccb header will be updated to reflect added
+ * content.
+ */
+struct dcw *itcw_add_dcw(struct itcw *itcw, u8 cmd, u8 flags, void *cd,
+			 u8 cd_count, u32 count)
+{
+	return tccb_add_dcw(tcw_get_tccb(itcw->tcw), TCCB_MAX_SIZE, cmd,
+			    flags, cd, cd_count, count);
+}
+EXPORT_SYMBOL(itcw_add_dcw);
+
+/**
+ * itcw_add_tidaw - add a tidaw to the itcw
+ * @itcw: address of the itcw
+ * @flags: flags for the new tidaw
+ * @addr: address value for the new tidaw
+ * @count: count value for the new tidaw
+ *
+ * Add a new tidaw to the input/output data tidaw-list of the specified itcw
+ * (depending on the value of the r-flag and w-flag). Return a pointer to
+ * the new tidaw on success or -%ENOSPC if the new tidaw would exceed the
+ * available space.
+ *
+ * Note: TTIC tidaws are automatically added when needed, so explicitly calling
+ * this interface with the TTIC flag is not supported. The last-tidaw flag
+ * for the last tidaw in the list will be set by itcw_finalize.
+ */
+struct tidaw *itcw_add_tidaw(struct itcw *itcw, u8 flags, void *addr, u32 count)
+{
+	struct tidaw *following;
+
+	if (itcw->num_tidaws >= itcw->max_tidaws)
+		return ERR_PTR(-ENOSPC);
+	/*
+	 * Is the tidaw, which follows the one we are about to fill, on the next
+	 * page? Then we have to insert a TTIC tidaw first, that points to the
+	 * tidaw on the new page.
+	 */
+	following = ((struct tidaw *) tcw_get_data(itcw->tcw))
+		+ itcw->num_tidaws + 1;
+	if (itcw->num_tidaws && !((unsigned long) following & ~PAGE_MASK)) {
+		tcw_add_tidaw(itcw->tcw, itcw->num_tidaws++,
+			      TIDAW_FLAGS_TTIC, following, 0);
+		if (itcw->num_tidaws >= itcw->max_tidaws)
+			return ERR_PTR(-ENOSPC);
+	}
+	return tcw_add_tidaw(itcw->tcw, itcw->num_tidaws++, flags, addr, count);
+}
+EXPORT_SYMBOL(itcw_add_tidaw);
+
+/**
+ * itcw_set_data - set data address and tida flag of the itcw
+ * @itcw: address of the itcw
+ * @addr: the data address
+ * @use_tidal: zero of the data address specifies a contiguous block of data,
+ * non-zero if it specifies a list if tidaws.
+ *
+ * Set the input/output data address of the itcw (depending on the value of the
+ * r-flag and w-flag). If @use_tidal is non-zero, the corresponding tida flag
+ * is set as well.
+ */
+void itcw_set_data(struct itcw *itcw, void *addr, int use_tidal)
+{
+	tcw_set_data(itcw->tcw, addr, use_tidal);
+}
+EXPORT_SYMBOL(itcw_set_data);
+
+/**
+ * itcw_finalize - calculate length and count fields of the itcw
+ * @itcw: address of the itcw
+ *
+ * Calculate tcw input-/output-count and tccbl fields and add a tcat the tccb.
+ * In case input- or output-tida is used, the tidaw-list must be stored in
+ * continuous storage (no ttic). The tcal field in the tccb must be
+ * up-to-date.
+ */
+void itcw_finalize(struct itcw *itcw)
+{
+	tcw_finalize(itcw->tcw, itcw->num_tidaws);
+}
+EXPORT_SYMBOL(itcw_finalize);

drivers/s390/cio/orb.h

@@ -0,0 +1,91 @@
+/*
+ * Orb related data structures.
+ *
+ * Copyright IBM Corp. 2007, 2011
+ *
+ * Author(s): Cornelia Huck <cornelia.huck@de.ibm.com>
+ *	      Peter Oberparleiter <peter.oberparleiter@de.ibm.com>
+ *	      Sebastian Ott <sebott@linux.vnet.ibm.com>
+ */
+
+#ifndef S390_ORB_H
+#define S390_ORB_H
+
+/*
+ * Command-mode operation request block
+ */
+struct cmd_orb {
+	u32 intparm;	/* interruption parameter */
+	u32 key:4;	/* flags, like key, suspend control, etc. */
+	u32 spnd:1;	/* suspend control */
+	u32 res1:1;	/* reserved */
+	u32 mod:1;	/* modification control */
+	u32 sync:1;	/* synchronize control */
+	u32 fmt:1;	/* format control */
+	u32 pfch:1;	/* prefetch control */
+	u32 isic:1;	/* initial-status-interruption control */
+	u32 alcc:1;	/* address-limit-checking control */
+	u32 ssic:1;	/* suppress-suspended-interr. control */
+	u32 res2:1;	/* reserved */
+	u32 c64:1;	/* IDAW/QDIO 64 bit control  */
+	u32 i2k:1;	/* IDAW 2/4kB block size control */
+	u32 lpm:8;	/* logical path mask */
+	u32 ils:1;	/* incorrect length */
+	u32 zero:6;	/* reserved zeros */
+	u32 orbx:1;	/* ORB extension control */
+	u32 cpa;	/* channel program address */
+}  __packed __aligned(4);
+
+/*
+ * Transport-mode operation request block
+ */
+struct tm_orb {
+	u32 intparm;
+	u32 key:4;
+	u32:9;
+	u32 b:1;
+	u32:2;
+	u32 lpm:8;
+	u32:7;
+	u32 x:1;
+	u32 tcw;
+	u32 prio:8;
+	u32:8;
+	u32 rsvpgm:8;
+	u32:8;
+	u32:32;
+	u32:32;
+	u32:32;
+	u32:32;
+}  __packed __aligned(4);
+
+/*
+ * eadm operation request block
+ */
+struct eadm_orb {
+	u32 intparm;
+	u32 key:4;
+	u32:4;
+	u32 compat1:1;
+	u32 compat2:1;
+	u32:21;
+	u32 x:1;
+	u32 aob;
+	u32 css_prio:8;
+	u32:8;
+	u32 scm_prio:8;
+	u32:8;
+	u32:29;
+	u32 fmt:3;
+	u32:32;
+	u32:32;
+	u32:32;
+}  __packed __aligned(4);
+
+union orb {
+	struct cmd_orb cmd;
+	struct tm_orb tm;
+	struct eadm_orb eadm;
+}  __packed __aligned(4);
+
+#endif /* S390_ORB_H */

drivers/s390/cio/qdio.h

@@ -0,0 +1,425 @@
+/*
+ * Copyright IBM Corp. 2000, 2009
+ * Author(s): Utz Bacher <utz.bacher@de.ibm.com>
+ *	      Jan Glauber <jang@linux.vnet.ibm.com>
+ */
+#ifndef _CIO_QDIO_H
+#define _CIO_QDIO_H
+
+#include <asm/page.h>
+#include <asm/schid.h>
+#include <asm/debug.h>
+#include "chsc.h"
+
+#define QDIO_BUSY_BIT_PATIENCE		(100 << 12)	/* 100 microseconds */
+#define QDIO_BUSY_BIT_RETRY_DELAY	10		/* 10 milliseconds */
+#define QDIO_BUSY_BIT_RETRIES		1000		/* = 10s retry time */
+#define QDIO_INPUT_THRESHOLD		(500 << 12)	/* 500 microseconds */
+
+enum qdio_irq_states {
+	QDIO_IRQ_STATE_INACTIVE,
+	QDIO_IRQ_STATE_ESTABLISHED,
+	QDIO_IRQ_STATE_ACTIVE,
+	QDIO_IRQ_STATE_STOPPED,
+	QDIO_IRQ_STATE_CLEANUP,
+	QDIO_IRQ_STATE_ERR,
+	NR_QDIO_IRQ_STATES,
+};
+
+/* used as intparm in do_IO */
+#define QDIO_DOING_ESTABLISH	1
+#define QDIO_DOING_ACTIVATE	2
+#define QDIO_DOING_CLEANUP	3
+
+#define SLSB_STATE_NOT_INIT	0x0
+#define SLSB_STATE_EMPTY	0x1
+#define SLSB_STATE_PRIMED	0x2
+#define SLSB_STATE_PENDING	0x3
+#define SLSB_STATE_HALTED	0xe
+#define SLSB_STATE_ERROR	0xf
+#define SLSB_TYPE_INPUT		0x0
+#define SLSB_TYPE_OUTPUT	0x20
+#define SLSB_OWNER_PROG		0x80
+#define SLSB_OWNER_CU		0x40
+
+#define SLSB_P_INPUT_NOT_INIT	\
+	(SLSB_OWNER_PROG | SLSB_TYPE_INPUT | SLSB_STATE_NOT_INIT)  /* 0x80 */
+#define SLSB_P_INPUT_ACK	\
+	(SLSB_OWNER_PROG | SLSB_TYPE_INPUT | SLSB_STATE_EMPTY)	   /* 0x81 */
+#define SLSB_CU_INPUT_EMPTY	\
+	(SLSB_OWNER_CU | SLSB_TYPE_INPUT | SLSB_STATE_EMPTY)	   /* 0x41 */
+#define SLSB_P_INPUT_PRIMED	\
+	(SLSB_OWNER_PROG | SLSB_TYPE_INPUT | SLSB_STATE_PRIMED)	   /* 0x82 */
+#define SLSB_P_INPUT_HALTED	\
+	(SLSB_OWNER_PROG | SLSB_TYPE_INPUT | SLSB_STATE_HALTED)	   /* 0x8e */
+#define SLSB_P_INPUT_ERROR	\
+	(SLSB_OWNER_PROG | SLSB_TYPE_INPUT | SLSB_STATE_ERROR)	   /* 0x8f */
+#define SLSB_P_OUTPUT_NOT_INIT	\
+	(SLSB_OWNER_PROG | SLSB_TYPE_OUTPUT | SLSB_STATE_NOT_INIT) /* 0xa0 */
+#define SLSB_P_OUTPUT_EMPTY	\
+	(SLSB_OWNER_PROG | SLSB_TYPE_OUTPUT | SLSB_STATE_EMPTY)	   /* 0xa1 */
+#define SLSB_P_OUTPUT_PENDING \
+	(SLSB_OWNER_PROG | SLSB_TYPE_OUTPUT | SLSB_STATE_PENDING)  /* 0xa3 */
+#define SLSB_CU_OUTPUT_PRIMED	\
+	(SLSB_OWNER_CU | SLSB_TYPE_OUTPUT | SLSB_STATE_PRIMED)	   /* 0x62 */
+#define SLSB_P_OUTPUT_HALTED	\
+	(SLSB_OWNER_PROG | SLSB_TYPE_OUTPUT | SLSB_STATE_HALTED)   /* 0xae */
+#define SLSB_P_OUTPUT_ERROR	\
+	(SLSB_OWNER_PROG | SLSB_TYPE_OUTPUT | SLSB_STATE_ERROR)	   /* 0xaf */
+
+#define SLSB_ERROR_DURING_LOOKUP  0xff
+
+/* additional CIWs returned by extended Sense-ID */
+#define CIW_TYPE_EQUEUE			0x3 /* establish QDIO queues */
+#define CIW_TYPE_AQUEUE			0x4 /* activate QDIO queues */
+
+/* flags for st qdio sch data */
+#define CHSC_FLAG_QDIO_CAPABILITY	0x80
+#define CHSC_FLAG_VALIDITY		0x40
+
+/* SIGA flags */
+#define QDIO_SIGA_WRITE		0x00
+#define QDIO_SIGA_READ		0x01
+#define QDIO_SIGA_SYNC		0x02
+#define QDIO_SIGA_WRITEQ	0x04
+#define QDIO_SIGA_QEBSM_FLAG	0x80
+
+#ifdef CONFIG_64BIT
+static inline int do_sqbs(u64 token, unsigned char state, int queue,
+			  int *start, int *count)
+{
+	register unsigned long _ccq asm ("0") = *count;
+	register unsigned long _token asm ("1") = token;
+	unsigned long _queuestart = ((unsigned long)queue << 32) | *start;
+
+	asm volatile(
+		"	.insn	rsy,0xeb000000008A,%1,0,0(%2)"
+		: "+d" (_ccq), "+d" (_queuestart)
+		: "d" ((unsigned long)state), "d" (_token)
+		: "memory", "cc");
+	*count = _ccq & 0xff;
+	*start = _queuestart & 0xff;
+
+	return (_ccq >> 32) & 0xff;
+}
+
+static inline int do_eqbs(u64 token, unsigned char *state, int queue,
+			  int *start, int *count, int ack)
+{
+	register unsigned long _ccq asm ("0") = *count;
+	register unsigned long _token asm ("1") = token;
+	unsigned long _queuestart = ((unsigned long)queue << 32) | *start;
+	unsigned long _state = (unsigned long)ack << 63;
+
+	asm volatile(
+		"	.insn	rrf,0xB99c0000,%1,%2,0,0"
+		: "+d" (_ccq), "+d" (_queuestart), "+d" (_state)
+		: "d" (_token)
+		: "memory", "cc");
+	*count = _ccq & 0xff;
+	*start = _queuestart & 0xff;
+	*state = _state & 0xff;
+
+	return (_ccq >> 32) & 0xff;
+}
+#else
+static inline int do_sqbs(u64 token, unsigned char state, int queue,
+			  int *start, int *count) { return 0; }
+static inline int do_eqbs(u64 token, unsigned char *state, int queue,
+			  int *start, int *count, int ack) { return 0; }
+#endif /* CONFIG_64BIT */
+
+struct qdio_irq;
+
+struct siga_flag {
+	u8 input:1;
+	u8 output:1;
+	u8 sync:1;
+	u8 sync_after_ai:1;
+	u8 sync_out_after_pci:1;
+	u8:3;
+} __attribute__ ((packed));
+
+struct qdio_dev_perf_stat {
+	unsigned int adapter_int;
+	unsigned int qdio_int;
+	unsigned int pci_request_int;
+
+	unsigned int tasklet_inbound;
+	unsigned int tasklet_inbound_resched;
+	unsigned int tasklet_inbound_resched2;
+	unsigned int tasklet_outbound;
+
+	unsigned int siga_read;
+	unsigned int siga_write;
+	unsigned int siga_sync;
+
+	unsigned int inbound_call;
+	unsigned int inbound_handler;
+	unsigned int stop_polling;
+	unsigned int inbound_queue_full;
+	unsigned int outbound_call;
+	unsigned int outbound_handler;
+	unsigned int outbound_queue_full;
+	unsigned int fast_requeue;
+	unsigned int target_full;
+	unsigned int eqbs;
+	unsigned int eqbs_partial;
+	unsigned int sqbs;
+	unsigned int sqbs_partial;
+	unsigned int int_discarded;
+} ____cacheline_aligned;
+
+struct qdio_queue_perf_stat {
+	/*
+	 * Sorted into order-2 buckets: 1, 2-3, 4-7, ... 64-127, 128.
+	 * Since max. 127 SBALs are scanned reuse entry for 128 as queue full
+	 * aka 127 SBALs found.
+	 */
+	unsigned int nr_sbals[8];
+	unsigned int nr_sbal_error;
+	unsigned int nr_sbal_nop;
+	unsigned int nr_sbal_total;
+};
+
+enum qdio_queue_irq_states {
+	QDIO_QUEUE_IRQS_DISABLED,
+};
+
+struct qdio_input_q {
+	/* input buffer acknowledgement flag */
+	int polling;
+	/* first ACK'ed buffer */
+	int ack_start;
+	/* how much sbals are acknowledged with qebsm */
+	int ack_count;
+	/* last time of noticing incoming data */
+	u64 timestamp;
+	/* upper-layer polling flag */
+	unsigned long queue_irq_state;
+	/* callback to start upper-layer polling */
+	void (*queue_start_poll) (struct ccw_device *, int, unsigned long);
+};
+
+struct qdio_output_q {
+	/* PCIs are enabled for the queue */
+	int pci_out_enabled;
+	/* cq: use asynchronous output buffers */
+	int use_cq;
+	/* cq: aobs used for particual SBAL */
+	struct qaob **aobs;
+	/* cq: sbal state related to asynchronous operation */
+	struct qdio_outbuf_state *sbal_state;
+	/* timer to check for more outbound work */
+	struct timer_list timer;
+	/* used SBALs before tasklet schedule */
+	int scan_threshold;
+};
+
+/*
+ * Note on cache alignment: grouped slsb and write mostly data at the beginning
+ * sbal[] is read-only and starts on a new cacheline followed by read mostly.
+ */
+struct qdio_q {
+	struct slsb slsb;
+
+	union {
+		struct qdio_input_q in;
+		struct qdio_output_q out;
+	} u;
+
+	/*
+	 * inbound: next buffer the program should check for
+	 * outbound: next buffer to check if adapter processed it
+	 */
+	int first_to_check;
+
+	/* first_to_check of the last time */
+	int last_move;
+
+	/* beginning position for calling the program */
+	int first_to_kick;
+
+	/* number of buffers in use by the adapter */
+	atomic_t nr_buf_used;
+
+	/* error condition during a data transfer */
+	unsigned int qdio_error;
+
+	/* last scan of the queue */
+	u64 timestamp;
+
+	struct tasklet_struct tasklet;
+	struct qdio_queue_perf_stat q_stats;
+
+	struct qdio_buffer *sbal[QDIO_MAX_BUFFERS_PER_Q] ____cacheline_aligned;
+
+	/* queue number */
+	int nr;
+
+	/* bitmask of queue number */
+	int mask;
+
+	/* input or output queue */
+	int is_input_q;
+
+	/* list of thinint input queues */
+	struct list_head entry;
+
+	/* upper-layer program handler */
+	qdio_handler_t (*handler);
+
+	struct dentry *debugfs_q;
+	struct qdio_irq *irq_ptr;
+	struct sl *sl;
+	/*
+	 * A page is allocated under this pointer and used for slib and sl.
+	 * slib is 2048 bytes big and sl points to offset PAGE_SIZE / 2.
+	 */
+	struct slib *slib;
+} __attribute__ ((aligned(256)));
+
+struct qdio_irq {
+	struct qib qib;
+	u32 *dsci;		/* address of device state change indicator */
+	struct ccw_device *cdev;
+	struct dentry *debugfs_dev;
+	struct dentry *debugfs_perf;
+
+	unsigned long int_parm;
+	struct subchannel_id schid;
+	unsigned long sch_token;	/* QEBSM facility */
+
+	enum qdio_irq_states state;
+
+	struct siga_flag siga_flag;	/* siga sync information from qdioac */
+
+	int nr_input_qs;
+	int nr_output_qs;
+
+	struct ccw1 ccw;
+	struct ciw equeue;
+	struct ciw aqueue;
+
+	struct qdio_ssqd_desc ssqd_desc;
+	void (*orig_handler) (struct ccw_device *, unsigned long, struct irb *);
+
+	int perf_stat_enabled;
+
+	struct qdr *qdr;
+	unsigned long chsc_page;
+
+	struct qdio_q *input_qs[QDIO_MAX_QUEUES_PER_IRQ];
+	struct qdio_q *output_qs[QDIO_MAX_QUEUES_PER_IRQ];
+
+	debug_info_t *debug_area;
+	struct mutex setup_mutex;
+	struct qdio_dev_perf_stat perf_stat;
+};
+
+/* helper functions */
+#define queue_type(q)	q->irq_ptr->qib.qfmt
+#define SCH_NO(q)	(q->irq_ptr->schid.sch_no)
+
+#define is_thinint_irq(irq) \
+	(irq->qib.qfmt == QDIO_IQDIO_QFMT || \
+	 css_general_characteristics.aif_osa)
+
+#define qperf(__qdev, __attr)	((__qdev)->perf_stat.(__attr))
+
+#define qperf_inc(__q, __attr)						\
+({									\
+	struct qdio_irq *qdev = (__q)->irq_ptr;				\
+	if (qdev->perf_stat_enabled)					\
+		(qdev->perf_stat.__attr)++;				\
+})
+
+static inline void account_sbals_error(struct qdio_q *q, int count)
+{
+	q->q_stats.nr_sbal_error += count;
+	q->q_stats.nr_sbal_total += count;
+}
+
+/* the highest iqdio queue is used for multicast */
+static inline int multicast_outbound(struct qdio_q *q)
+{
+	return (q->irq_ptr->nr_output_qs > 1) &&
+	       (q->nr == q->irq_ptr->nr_output_qs - 1);
+}
+
+#define pci_out_supported(q) \
+	(q->irq_ptr->qib.ac & QIB_AC_OUTBOUND_PCI_SUPPORTED)
+#define is_qebsm(q)			(q->irq_ptr->sch_token != 0)
+
+#define need_siga_in(q)			(q->irq_ptr->siga_flag.input)
+#define need_siga_out(q)		(q->irq_ptr->siga_flag.output)
+#define need_siga_sync(q)		(unlikely(q->irq_ptr->siga_flag.sync))
+#define need_siga_sync_after_ai(q)	\
+	(unlikely(q->irq_ptr->siga_flag.sync_after_ai))
+#define need_siga_sync_out_after_pci(q)	\
+	(unlikely(q->irq_ptr->siga_flag.sync_out_after_pci))
+
+#define for_each_input_queue(irq_ptr, q, i)		\
+	for (i = 0; i < irq_ptr->nr_input_qs &&		\
+		({ q = irq_ptr->input_qs[i]; 1; }); i++)
+#define for_each_output_queue(irq_ptr, q, i)		\
+	for (i = 0; i < irq_ptr->nr_output_qs &&	\
+		({ q = irq_ptr->output_qs[i]; 1; }); i++)
+
+#define prev_buf(bufnr)	\
+	((bufnr + QDIO_MAX_BUFFERS_MASK) & QDIO_MAX_BUFFERS_MASK)
+#define next_buf(bufnr)	\
+	((bufnr + 1) & QDIO_MAX_BUFFERS_MASK)
+#define add_buf(bufnr, inc) \
+	((bufnr + inc) & QDIO_MAX_BUFFERS_MASK)
+#define sub_buf(bufnr, dec) \
+	((bufnr - dec) & QDIO_MAX_BUFFERS_MASK)
+
+#define queue_irqs_enabled(q)			\
+	(test_bit(QDIO_QUEUE_IRQS_DISABLED, &q->u.in.queue_irq_state) == 0)
+#define queue_irqs_disabled(q)			\
+	(test_bit(QDIO_QUEUE_IRQS_DISABLED, &q->u.in.queue_irq_state) != 0)
+
+extern u64 last_ai_time;
+
+/* prototypes for thin interrupt */
+void qdio_setup_thinint(struct qdio_irq *irq_ptr);
+int qdio_establish_thinint(struct qdio_irq *irq_ptr);
+void qdio_shutdown_thinint(struct qdio_irq *irq_ptr);
+void tiqdio_add_input_queues(struct qdio_irq *irq_ptr);
+void tiqdio_remove_input_queues(struct qdio_irq *irq_ptr);
+void tiqdio_inbound_processing(unsigned long q);
+int tiqdio_allocate_memory(void);
+void tiqdio_free_memory(void);
+int tiqdio_register_thinints(void);
+void tiqdio_unregister_thinints(void);
+void clear_nonshared_ind(struct qdio_irq *);
+int test_nonshared_ind(struct qdio_irq *);
+
+/* prototypes for setup */
+void qdio_inbound_processing(unsigned long data);
+void qdio_outbound_processing(unsigned long data);
+void qdio_outbound_timer(unsigned long data);
+void qdio_int_handler(struct ccw_device *cdev, unsigned long intparm,
+		      struct irb *irb);
+int qdio_allocate_qs(struct qdio_irq *irq_ptr, int nr_input_qs,
+		     int nr_output_qs);
+void qdio_setup_ssqd_info(struct qdio_irq *irq_ptr);
+int qdio_setup_get_ssqd(struct qdio_irq *irq_ptr,
+			struct subchannel_id *schid,
+			struct qdio_ssqd_desc *data);
+int qdio_setup_irq(struct qdio_initialize *init_data);
+void qdio_print_subchannel_info(struct qdio_irq *irq_ptr,
+				struct ccw_device *cdev);
+void qdio_release_memory(struct qdio_irq *irq_ptr);
+int qdio_setup_create_sysfs(struct ccw_device *cdev);
+void qdio_setup_destroy_sysfs(struct ccw_device *cdev);
+int qdio_setup_init(void);
+void qdio_setup_exit(void);
+int qdio_enable_async_operation(struct qdio_output_q *q);
+void qdio_disable_async_operation(struct qdio_output_q *q);
+struct qaob *qdio_allocate_aob(void);
+
+int debug_get_buf_state(struct qdio_q *q, unsigned int bufnr,
+			unsigned char *state);
+#endif /* _CIO_QDIO_H */

drivers/s390/cio/qdio_debug.c

@@ -0,0 +1,373 @@
+/*
+ *  Copyright IBM Corp. 2008, 2009
+ *
+ *  Author: Jan Glauber (jang@linux.vnet.ibm.com)
+ */
+#include <linux/seq_file.h>
+#include <linux/debugfs.h>
+#include <linux/uaccess.h>
+#include <linux/export.h>
+#include <linux/slab.h>
+#include <asm/debug.h>
+#include "qdio_debug.h"
+#include "qdio.h"
+
+debug_info_t *qdio_dbf_setup;
+debug_info_t *qdio_dbf_error;
+
+static struct dentry *debugfs_root;
+#define QDIO_DEBUGFS_NAME_LEN	10
+#define QDIO_DBF_NAME_LEN	20
+
+struct qdio_dbf_entry {
+	char dbf_name[QDIO_DBF_NAME_LEN];
+	debug_info_t *dbf_info;
+	struct list_head dbf_list;
+};
+
+static LIST_HEAD(qdio_dbf_list);
+static DEFINE_MUTEX(qdio_dbf_list_mutex);
+
+static debug_info_t *qdio_get_dbf_entry(char *name)
+{
+	struct qdio_dbf_entry *entry;
+	debug_info_t *rc = NULL;
+
+	mutex_lock(&qdio_dbf_list_mutex);
+	list_for_each_entry(entry, &qdio_dbf_list, dbf_list) {
+		if (strcmp(entry->dbf_name, name) == 0) {
+			rc = entry->dbf_info;
+			break;
+		}
+	}
+	mutex_unlock(&qdio_dbf_list_mutex);
+	return rc;
+}
+
+static void qdio_clear_dbf_list(void)
+{
+	struct qdio_dbf_entry *entry, *tmp;
+
+	mutex_lock(&qdio_dbf_list_mutex);
+	list_for_each_entry_safe(entry, tmp, &qdio_dbf_list, dbf_list) {
+		list_del(&entry->dbf_list);
+		debug_unregister(entry->dbf_info);
+		kfree(entry);
+	}
+	mutex_unlock(&qdio_dbf_list_mutex);
+}
+
+int qdio_allocate_dbf(struct qdio_initialize *init_data,
+		       struct qdio_irq *irq_ptr)
+{
+	char text[QDIO_DBF_NAME_LEN];
+	struct qdio_dbf_entry *new_entry;
+
+	DBF_EVENT("qfmt:%1d", init_data->q_format);
+	DBF_HEX(init_data->adapter_name, 8);
+	DBF_EVENT("qpff%4x", init_data->qib_param_field_format);
+	DBF_HEX(&init_data->qib_param_field, sizeof(void *));
+	DBF_HEX(&init_data->input_slib_elements, sizeof(void *));
+	DBF_HEX(&init_data->output_slib_elements, sizeof(void *));
+	DBF_EVENT("niq:%1d noq:%1d", init_data->no_input_qs,
+		  init_data->no_output_qs);
+	DBF_HEX(&init_data->input_handler, sizeof(void *));
+	DBF_HEX(&init_data->output_handler, sizeof(void *));
+	DBF_HEX(&init_data->int_parm, sizeof(long));
+	DBF_HEX(&init_data->input_sbal_addr_array, sizeof(void *));
+	DBF_HEX(&init_data->output_sbal_addr_array, sizeof(void *));
+	DBF_EVENT("irq:%8lx", (unsigned long)irq_ptr);
+
+	/* allocate trace view for the interface */
+	snprintf(text, QDIO_DBF_NAME_LEN, "qdio_%s",
+					dev_name(&init_data->cdev->dev));
+	irq_ptr->debug_area = qdio_get_dbf_entry(text);
+	if (irq_ptr->debug_area)
+		DBF_DEV_EVENT(DBF_ERR, irq_ptr, "dbf reused");
+	else {
+		irq_ptr->debug_area = debug_register(text, 2, 1, 16);
+		if (!irq_ptr->debug_area)
+			return -ENOMEM;
+		if (debug_register_view(irq_ptr->debug_area,
+						&debug_hex_ascii_view)) {
+			debug_unregister(irq_ptr->debug_area);
+			return -ENOMEM;
+		}
+		debug_set_level(irq_ptr->debug_area, DBF_WARN);
+		DBF_DEV_EVENT(DBF_ERR, irq_ptr, "dbf created");
+		new_entry = kzalloc(sizeof(struct qdio_dbf_entry), GFP_KERNEL);
+		if (!new_entry) {
+			debug_unregister(irq_ptr->debug_area);
+			return -ENOMEM;
+		}
+		strlcpy(new_entry->dbf_name, text, QDIO_DBF_NAME_LEN);
+		new_entry->dbf_info = irq_ptr->debug_area;
+		mutex_lock(&qdio_dbf_list_mutex);
+		list_add(&new_entry->dbf_list, &qdio_dbf_list);
+		mutex_unlock(&qdio_dbf_list_mutex);
+	}
+	return 0;
+}
+
+static int qstat_show(struct seq_file *m, void *v)
+{
+	unsigned char state;
+	struct qdio_q *q = m->private;
+	int i;
+
+	if (!q)
+		return 0;
+
+	seq_printf(m, "Timestamp: %Lx  Last AI: %Lx\n",
+		   q->timestamp, last_ai_time);
+	seq_printf(m, "nr_used: %d  ftc: %d  last_move: %d\n",
+		   atomic_read(&q->nr_buf_used),
+		   q->first_to_check, q->last_move);
+	if (q->is_input_q) {
+		seq_printf(m, "polling: %d  ack start: %d  ack count: %d\n",
+			   q->u.in.polling, q->u.in.ack_start,
+			   q->u.in.ack_count);
+		seq_printf(m, "DSCI: %d   IRQs disabled: %u\n",
+			   *(u32 *)q->irq_ptr->dsci,
+			   test_bit(QDIO_QUEUE_IRQS_DISABLED,
+			   &q->u.in.queue_irq_state));
+	}
+	seq_printf(m, "SBAL states:\n");
+	seq_printf(m, "|0      |8      |16     |24     |32     |40     |48     |56  63|\n");
+
+	for (i = 0; i < QDIO_MAX_BUFFERS_PER_Q; i++) {
+		debug_get_buf_state(q, i, &state);
+		switch (state) {
+		case SLSB_P_INPUT_NOT_INIT:
+		case SLSB_P_OUTPUT_NOT_INIT:
+			seq_printf(m, "N");
+			break;
+		case SLSB_P_OUTPUT_PENDING:
+			seq_printf(m, "P");
+			break;
+		case SLSB_P_INPUT_PRIMED:
+		case SLSB_CU_OUTPUT_PRIMED:
+			seq_printf(m, "+");
+			break;
+		case SLSB_P_INPUT_ACK:
+			seq_printf(m, "A");
+			break;
+		case SLSB_P_INPUT_ERROR:
+		case SLSB_P_OUTPUT_ERROR:
+			seq_printf(m, "x");
+			break;
+		case SLSB_CU_INPUT_EMPTY:
+		case SLSB_P_OUTPUT_EMPTY:
+			seq_printf(m, "-");
+			break;
+		case SLSB_P_INPUT_HALTED:
+		case SLSB_P_OUTPUT_HALTED:
+			seq_printf(m, ".");
+			break;
+		default:
+			seq_printf(m, "?");
+		}
+		if (i == 63)
+			seq_printf(m, "\n");
+	}
+	seq_printf(m, "\n");
+	seq_printf(m, "|64     |72     |80     |88     |96     |104    |112    |   127|\n");
+
+	seq_printf(m, "\nSBAL statistics:");
+	if (!q->irq_ptr->perf_stat_enabled) {
+		seq_printf(m, " disabled\n");
+		return 0;
+	}
+
+	seq_printf(m, "\n1          2..        4..        8..        "
+		   "16..       32..       64..       127\n");
+	for (i = 0; i < ARRAY_SIZE(q->q_stats.nr_sbals); i++)
+		seq_printf(m, "%-10u ", q->q_stats.nr_sbals[i]);
+	seq_printf(m, "\nError      NOP        Total\n%-10u %-10u %-10u\n\n",
+		   q->q_stats.nr_sbal_error, q->q_stats.nr_sbal_nop,
+		   q->q_stats.nr_sbal_total);
+	return 0;
+}
+
+static int qstat_seq_open(struct inode *inode, struct file *filp)
+{
+	return single_open(filp, qstat_show,
+			   file_inode(filp)->i_private);
+}
+
+static const struct file_operations debugfs_fops = {
+	.owner	 = THIS_MODULE,
+	.open	 = qstat_seq_open,
+	.read	 = seq_read,
+	.llseek  = seq_lseek,
+	.release = single_release,
+};
+
+static char *qperf_names[] = {
+	"Assumed adapter interrupts",
+	"QDIO interrupts",
+	"Requested PCIs",
+	"Inbound tasklet runs",
+	"Inbound tasklet resched",
+	"Inbound tasklet resched2",
+	"Outbound tasklet runs",
+	"SIGA read",
+	"SIGA write",
+	"SIGA sync",
+	"Inbound calls",
+	"Inbound handler",
+	"Inbound stop_polling",
+	"Inbound queue full",
+	"Outbound calls",
+	"Outbound handler",
+	"Outbound queue full",
+	"Outbound fast_requeue",
+	"Outbound target_full",
+	"QEBSM eqbs",
+	"QEBSM eqbs partial",
+	"QEBSM sqbs",
+	"QEBSM sqbs partial",
+	"Discarded interrupts"
+};
+
+static int qperf_show(struct seq_file *m, void *v)
+{
+	struct qdio_irq *irq_ptr = m->private;
+	unsigned int *stat;
+	int i;
+
+	if (!irq_ptr)
+		return 0;
+	if (!irq_ptr->perf_stat_enabled) {
+		seq_printf(m, "disabled\n");
+		return 0;
+	}
+	stat = (unsigned int *)&irq_ptr->perf_stat;
+
+	for (i = 0; i < ARRAY_SIZE(qperf_names); i++)
+		seq_printf(m, "%26s:\t%u\n",
+			   qperf_names[i], *(stat + i));
+	return 0;
+}
+
+static ssize_t qperf_seq_write(struct file *file, const char __user *ubuf,
+			       size_t count, loff_t *off)
+{
+	struct seq_file *seq = file->private_data;
+	struct qdio_irq *irq_ptr = seq->private;
+	struct qdio_q *q;
+	unsigned long val;
+	int ret, i;
+
+	if (!irq_ptr)
+		return 0;
+
+	ret = kstrtoul_from_user(ubuf, count, 10, &val);
+	if (ret)
+		return ret;
+
+	switch (val) {
+	case 0:
+		irq_ptr->perf_stat_enabled = 0;
+		memset(&irq_ptr->perf_stat, 0, sizeof(irq_ptr->perf_stat));
+		for_each_input_queue(irq_ptr, q, i)
+			memset(&q->q_stats, 0, sizeof(q->q_stats));
+		for_each_output_queue(irq_ptr, q, i)
+			memset(&q->q_stats, 0, sizeof(q->q_stats));
+		break;
+	case 1:
+		irq_ptr->perf_stat_enabled = 1;
+		break;
+	}
+	return count;
+}
+
+static int qperf_seq_open(struct inode *inode, struct file *filp)
+{
+	return single_open(filp, qperf_show,
+			   file_inode(filp)->i_private);
+}
+
+static const struct file_operations debugfs_perf_fops = {
+	.owner	 = THIS_MODULE,
+	.open	 = qperf_seq_open,
+	.read	 = seq_read,
+	.write	 = qperf_seq_write,
+	.llseek  = seq_lseek,
+	.release = single_release,
+};
+
+static void setup_debugfs_entry(struct qdio_q *q)
+{
+	char name[QDIO_DEBUGFS_NAME_LEN];
+
+	snprintf(name, QDIO_DEBUGFS_NAME_LEN, "%s_%d",
+		 q->is_input_q ? "input" : "output",
+		 q->nr);
+	q->debugfs_q = debugfs_create_file(name, S_IFREG | S_IRUGO | S_IWUSR,
+				q->irq_ptr->debugfs_dev, q, &debugfs_fops);
+	if (IS_ERR(q->debugfs_q))
+		q->debugfs_q = NULL;
+}
+
+void qdio_setup_debug_entries(struct qdio_irq *irq_ptr, struct ccw_device *cdev)
+{
+	struct qdio_q *q;
+	int i;
+
+	irq_ptr->debugfs_dev = debugfs_create_dir(dev_name(&cdev->dev),
+						  debugfs_root);
+	if (IS_ERR(irq_ptr->debugfs_dev))
+		irq_ptr->debugfs_dev = NULL;
+
+	irq_ptr->debugfs_perf = debugfs_create_file("statistics",
+				S_IFREG | S_IRUGO | S_IWUSR,
+				irq_ptr->debugfs_dev, irq_ptr,
+				&debugfs_perf_fops);
+	if (IS_ERR(irq_ptr->debugfs_perf))
+		irq_ptr->debugfs_perf = NULL;
+
+	for_each_input_queue(irq_ptr, q, i)
+		setup_debugfs_entry(q);
+	for_each_output_queue(irq_ptr, q, i)
+		setup_debugfs_entry(q);
+}
+
+void qdio_shutdown_debug_entries(struct qdio_irq *irq_ptr)
+{
+	struct qdio_q *q;
+	int i;
+
+	for_each_input_queue(irq_ptr, q, i)
+		debugfs_remove(q->debugfs_q);
+	for_each_output_queue(irq_ptr, q, i)
+		debugfs_remove(q->debugfs_q);
+	debugfs_remove(irq_ptr->debugfs_perf);
+	debugfs_remove(irq_ptr->debugfs_dev);
+}
+
+int __init qdio_debug_init(void)
+{
+	debugfs_root = debugfs_create_dir("qdio", NULL);
+
+	qdio_dbf_setup = debug_register("qdio_setup", 16, 1, 16);
+	debug_register_view(qdio_dbf_setup, &debug_hex_ascii_view);
+	debug_set_level(qdio_dbf_setup, DBF_INFO);
+	DBF_EVENT("dbf created\n");
+
+	qdio_dbf_error = debug_register("qdio_error", 4, 1, 16);
+	debug_register_view(qdio_dbf_error, &debug_hex_ascii_view);
+	debug_set_level(qdio_dbf_error, DBF_INFO);
+	DBF_ERROR("dbf created\n");
+	return 0;
+}
+
+void qdio_debug_exit(void)
+{
+	qdio_clear_dbf_list();
+	debugfs_remove(debugfs_root);
+	if (qdio_dbf_setup)
+		debug_unregister(qdio_dbf_setup);
+	if (qdio_dbf_error)
+		debug_unregister(qdio_dbf_error);
+}

drivers/s390/cio/qdio_debug.h

@@ -0,0 +1,86 @@
+/*
+ *  Copyright IBM Corp. 2008
+ *
+ *  Author: Jan Glauber (jang@linux.vnet.ibm.com)
+ */
+#ifndef QDIO_DEBUG_H
+#define QDIO_DEBUG_H
+
+#include <asm/debug.h>
+#include <asm/qdio.h>
+#include "qdio.h"
+
+/* that gives us 15 characters in the text event views */
+#define QDIO_DBF_LEN	16
+
+extern debug_info_t *qdio_dbf_setup;
+extern debug_info_t *qdio_dbf_error;
+
+#define DBF_ERR		3	/* error conditions	*/
+#define DBF_WARN	4	/* warning conditions	*/
+#define DBF_INFO	6	/* informational	*/
+
+#undef DBF_EVENT
+#undef DBF_ERROR
+#undef DBF_DEV_EVENT
+
+#define DBF_EVENT(text...) \
+	do { \
+		char debug_buffer[QDIO_DBF_LEN]; \
+		snprintf(debug_buffer, QDIO_DBF_LEN, text); \
+		debug_text_event(qdio_dbf_setup, DBF_ERR, debug_buffer); \
+	} while (0)
+
+static inline void DBF_HEX(void *addr, int len)
+{
+	while (len > 0) {
+		debug_event(qdio_dbf_setup, DBF_ERR, addr, len);
+		len -= qdio_dbf_setup->buf_size;
+		addr += qdio_dbf_setup->buf_size;
+	}
+}
+
+#define DBF_ERROR(text...) \
+	do { \
+		char debug_buffer[QDIO_DBF_LEN]; \
+		snprintf(debug_buffer, QDIO_DBF_LEN, text); \
+		debug_text_event(qdio_dbf_error, DBF_ERR, debug_buffer); \
+	} while (0)
+
+static inline void DBF_ERROR_HEX(void *addr, int len)
+{
+	while (len > 0) {
+		debug_event(qdio_dbf_error, DBF_ERR, addr, len);
+		len -= qdio_dbf_error->buf_size;
+		addr += qdio_dbf_error->buf_size;
+	}
+}
+
+#define DBF_DEV_EVENT(level, device, text...) \
+	do { \
+		char debug_buffer[QDIO_DBF_LEN]; \
+		if (debug_level_enabled(device->debug_area, level)) { \
+			snprintf(debug_buffer, QDIO_DBF_LEN, text); \
+			debug_text_event(device->debug_area, level, debug_buffer); \
+		} \
+	} while (0)
+
+static inline void DBF_DEV_HEX(struct qdio_irq *dev, void *addr,
+			       int len, int level)
+{
+	while (len > 0) {
+		debug_event(dev->debug_area, level, addr, len);
+		len -= dev->debug_area->buf_size;
+		addr += dev->debug_area->buf_size;
+	}
+}
+
+int qdio_allocate_dbf(struct qdio_initialize *init_data,
+		       struct qdio_irq *irq_ptr);
+void qdio_setup_debug_entries(struct qdio_irq *irq_ptr,
+			      struct ccw_device *cdev);
+void qdio_shutdown_debug_entries(struct qdio_irq *irq_ptr);
+int qdio_debug_init(void);
+void qdio_debug_exit(void);
+
+#endif

drivers/s390/cio/qdio_setup.c

@@ -0,0 +1,600 @@
+/*
+ * qdio queue initialization
+ *
+ * Copyright IBM Corp. 2008
+ * Author(s): Jan Glauber <jang@linux.vnet.ibm.com>
+ */
+#include <linux/kernel.h>
+#include <linux/slab.h>
+#include <linux/export.h>
+#include <asm/qdio.h>
+
+#include "cio.h"
+#include "css.h"
+#include "device.h"
+#include "ioasm.h"
+#include "chsc.h"
+#include "qdio.h"
+#include "qdio_debug.h"
+
+#define QBUFF_PER_PAGE (PAGE_SIZE / sizeof(struct qdio_buffer))
+
+static struct kmem_cache *qdio_q_cache;
+static struct kmem_cache *qdio_aob_cache;
+
+struct qaob *qdio_allocate_aob(void)
+{
+	return kmem_cache_zalloc(qdio_aob_cache, GFP_ATOMIC);
+}
+EXPORT_SYMBOL_GPL(qdio_allocate_aob);
+
+void qdio_release_aob(struct qaob *aob)
+{
+	kmem_cache_free(qdio_aob_cache, aob);
+}
+EXPORT_SYMBOL_GPL(qdio_release_aob);
+
+/**
+ * qdio_free_buffers() - free qdio buffers
+ * @buf: array of pointers to qdio buffers
+ * @count: number of qdio buffers to free
+ */
+void qdio_free_buffers(struct qdio_buffer **buf, unsigned int count)
+{
+	int pos;
+
+	for (pos = 0; pos < count; pos += QBUFF_PER_PAGE)
+		free_page((unsigned long) buf[pos]);
+}
+EXPORT_SYMBOL_GPL(qdio_free_buffers);
+
+/**
+ * qdio_alloc_buffers() - allocate qdio buffers
+ * @buf: array of pointers to qdio buffers
+ * @count: number of qdio buffers to allocate
+ */
+int qdio_alloc_buffers(struct qdio_buffer **buf, unsigned int count)
+{
+	int pos;
+
+	for (pos = 0; pos < count; pos += QBUFF_PER_PAGE) {
+		buf[pos] = (void *) get_zeroed_page(GFP_KERNEL);
+		if (!buf[pos]) {
+			qdio_free_buffers(buf, count);
+			return -ENOMEM;
+		}
+	}
+	for (pos = 0; pos < count; pos++)
+		if (pos % QBUFF_PER_PAGE)
+			buf[pos] = buf[pos - 1] + 1;
+	return 0;
+}
+EXPORT_SYMBOL_GPL(qdio_alloc_buffers);
+
+/**
+ * qdio_reset_buffers() - reset qdio buffers
+ * @buf: array of pointers to qdio buffers
+ * @count: number of qdio buffers that will be zeroed
+ */
+void qdio_reset_buffers(struct qdio_buffer **buf, unsigned int count)
+{
+	int pos;
+
+	for (pos = 0; pos < count; pos++)
+		memset(buf[pos], 0, sizeof(struct qdio_buffer));
+}
+EXPORT_SYMBOL_GPL(qdio_reset_buffers);
+
+/*
+ * qebsm is only available under 64bit but the adapter sets the feature
+ * flag anyway, so we manually override it.
+ */
+static inline int qebsm_possible(void)
+{
+#ifdef CONFIG_64BIT
+	return css_general_characteristics.qebsm;
+#endif
+	return 0;
+}
+
+/*
+ * qib_param_field: pointer to 128 bytes or NULL, if no param field
+ * nr_input_qs: pointer to nr_queues*128 words of data or NULL
+ */
+static void set_impl_params(struct qdio_irq *irq_ptr,
+			    unsigned int qib_param_field_format,
+			    unsigned char *qib_param_field,
+			    unsigned long *input_slib_elements,
+			    unsigned long *output_slib_elements)
+{
+	struct qdio_q *q;
+	int i, j;
+
+	if (!irq_ptr)
+		return;
+
+	irq_ptr->qib.pfmt = qib_param_field_format;
+	if (qib_param_field)
+		memcpy(irq_ptr->qib.parm, qib_param_field,
+		       QDIO_MAX_BUFFERS_PER_Q);
+
+	if (!input_slib_elements)
+		goto output;
+
+	for_each_input_queue(irq_ptr, q, i) {
+		for (j = 0; j < QDIO_MAX_BUFFERS_PER_Q; j++)
+			q->slib->slibe[j].parms =
+				input_slib_elements[i * QDIO_MAX_BUFFERS_PER_Q + j];
+	}
+output:
+	if (!output_slib_elements)
+		return;
+
+	for_each_output_queue(irq_ptr, q, i) {
+		for (j = 0; j < QDIO_MAX_BUFFERS_PER_Q; j++)
+			q->slib->slibe[j].parms =
+				output_slib_elements[i * QDIO_MAX_BUFFERS_PER_Q + j];
+	}
+}
+
+static int __qdio_allocate_qs(struct qdio_q **irq_ptr_qs, int nr_queues)
+{
+	struct qdio_q *q;
+	int i;
+
+	for (i = 0; i < nr_queues; i++) {
+		q = kmem_cache_alloc(qdio_q_cache, GFP_KERNEL);
+		if (!q)
+			return -ENOMEM;
+
+		q->slib = (struct slib *) __get_free_page(GFP_KERNEL);
+		if (!q->slib) {
+			kmem_cache_free(qdio_q_cache, q);
+			return -ENOMEM;
+		}
+		irq_ptr_qs[i] = q;
+	}
+	return 0;
+}
+
+int qdio_allocate_qs(struct qdio_irq *irq_ptr, int nr_input_qs, int nr_output_qs)
+{
+	int rc;
+
+	rc = __qdio_allocate_qs(irq_ptr->input_qs, nr_input_qs);
+	if (rc)
+		return rc;
+	rc = __qdio_allocate_qs(irq_ptr->output_qs, nr_output_qs);
+	return rc;
+}
+
+static void setup_queues_misc(struct qdio_q *q, struct qdio_irq *irq_ptr,
+			      qdio_handler_t *handler, int i)
+{
+	struct slib *slib = q->slib;
+
+	/* queue must be cleared for qdio_establish */
+	memset(q, 0, sizeof(*q));
+	memset(slib, 0, PAGE_SIZE);
+	q->slib = slib;
+	q->irq_ptr = irq_ptr;
+	q->mask = 1 << (31 - i);
+	q->nr = i;
+	q->handler = handler;
+}
+
+static void setup_storage_lists(struct qdio_q *q, struct qdio_irq *irq_ptr,
+				void **sbals_array, int i)
+{
+	struct qdio_q *prev;
+	int j;
+
+	DBF_HEX(&q, sizeof(void *));
+	q->sl = (struct sl *)((char *)q->slib + PAGE_SIZE / 2);
+
+	/* fill in sbal */
+	for (j = 0; j < QDIO_MAX_BUFFERS_PER_Q; j++)
+		q->sbal[j] = *sbals_array++;
+
+	/* fill in slib */
+	if (i > 0) {
+		prev = (q->is_input_q) ? irq_ptr->input_qs[i - 1]
+			: irq_ptr->output_qs[i - 1];
+		prev->slib->nsliba = (unsigned long)q->slib;
+	}
+
+	q->slib->sla = (unsigned long)q->sl;
+	q->slib->slsba = (unsigned long)&q->slsb.val[0];
+
+	/* fill in sl */
+	for (j = 0; j < QDIO_MAX_BUFFERS_PER_Q; j++)
+		q->sl->element[j].sbal = (unsigned long)q->sbal[j];
+}
+
+static void setup_queues(struct qdio_irq *irq_ptr,
+			 struct qdio_initialize *qdio_init)
+{
+	struct qdio_q *q;
+	void **input_sbal_array = qdio_init->input_sbal_addr_array;
+	void **output_sbal_array = qdio_init->output_sbal_addr_array;
+	struct qdio_outbuf_state *output_sbal_state_array =
+				  qdio_init->output_sbal_state_array;
+	int i;
+
+	for_each_input_queue(irq_ptr, q, i) {
+		DBF_EVENT("inq:%1d", i);
+		setup_queues_misc(q, irq_ptr, qdio_init->input_handler, i);
+
+		q->is_input_q = 1;
+		q->u.in.queue_start_poll = qdio_init->queue_start_poll_array ?
+				qdio_init->queue_start_poll_array[i] : NULL;
+
+		setup_storage_lists(q, irq_ptr, input_sbal_array, i);
+		input_sbal_array += QDIO_MAX_BUFFERS_PER_Q;
+
+		if (is_thinint_irq(irq_ptr)) {
+			tasklet_init(&q->tasklet, tiqdio_inbound_processing,
+				     (unsigned long) q);
+		} else {
+			tasklet_init(&q->tasklet, qdio_inbound_processing,
+				     (unsigned long) q);
+		}
+	}
+
+	for_each_output_queue(irq_ptr, q, i) {
+		DBF_EVENT("outq:%1d", i);
+		setup_queues_misc(q, irq_ptr, qdio_init->output_handler, i);
+
+		q->u.out.sbal_state = output_sbal_state_array;
+		output_sbal_state_array += QDIO_MAX_BUFFERS_PER_Q;
+
+		q->is_input_q = 0;
+		q->u.out.scan_threshold = qdio_init->scan_threshold;
+		setup_storage_lists(q, irq_ptr, output_sbal_array, i);
+		output_sbal_array += QDIO_MAX_BUFFERS_PER_Q;
+
+		tasklet_init(&q->tasklet, qdio_outbound_processing,
+			     (unsigned long) q);
+		setup_timer(&q->u.out.timer, (void(*)(unsigned long))
+			    &qdio_outbound_timer, (unsigned long)q);
+	}
+}
+
+static void process_ac_flags(struct qdio_irq *irq_ptr, unsigned char qdioac)
+{
+	if (qdioac & AC1_SIGA_INPUT_NEEDED)
+		irq_ptr->siga_flag.input = 1;
+	if (qdioac & AC1_SIGA_OUTPUT_NEEDED)
+		irq_ptr->siga_flag.output = 1;
+	if (qdioac & AC1_SIGA_SYNC_NEEDED)
+		irq_ptr->siga_flag.sync = 1;
+	if (!(qdioac & AC1_AUTOMATIC_SYNC_ON_THININT))
+		irq_ptr->siga_flag.sync_after_ai = 1;
+	if (!(qdioac & AC1_AUTOMATIC_SYNC_ON_OUT_PCI))
+		irq_ptr->siga_flag.sync_out_after_pci = 1;
+}
+
+static void check_and_setup_qebsm(struct qdio_irq *irq_ptr,
+				  unsigned char qdioac, unsigned long token)
+{
+	if (!(irq_ptr->qib.rflags & QIB_RFLAGS_ENABLE_QEBSM))
+		goto no_qebsm;
+	if (!(qdioac & AC1_SC_QEBSM_AVAILABLE) ||
+	    (!(qdioac & AC1_SC_QEBSM_ENABLED)))
+		goto no_qebsm;
+
+	irq_ptr->sch_token = token;
+
+	DBF_EVENT("V=V:1");
+	DBF_EVENT("%8lx", irq_ptr->sch_token);
+	return;
+
+no_qebsm:
+	irq_ptr->sch_token = 0;
+	irq_ptr->qib.rflags &= ~QIB_RFLAGS_ENABLE_QEBSM;
+	DBF_EVENT("noV=V");
+}
+
+/*
+ * If there is a qdio_irq we use the chsc_page and store the information
+ * in the qdio_irq, otherwise we copy it to the specified structure.
+ */
+int qdio_setup_get_ssqd(struct qdio_irq *irq_ptr,
+			struct subchannel_id *schid,
+			struct qdio_ssqd_desc *data)
+{
+	struct chsc_ssqd_area *ssqd;
+	int rc;
+
+	DBF_EVENT("getssqd:%4x", schid->sch_no);
+	if (!irq_ptr) {
+		ssqd = (struct chsc_ssqd_area *)__get_free_page(GFP_KERNEL);
+		if (!ssqd)
+			return -ENOMEM;
+	} else {
+		ssqd = (struct chsc_ssqd_area *)irq_ptr->chsc_page;
+	}
+
+	rc = chsc_ssqd(*schid, ssqd);
+	if (rc)
+		goto out;
+
+	if (!(ssqd->qdio_ssqd.flags & CHSC_FLAG_QDIO_CAPABILITY) ||
+	    !(ssqd->qdio_ssqd.flags & CHSC_FLAG_VALIDITY) ||
+	    (ssqd->qdio_ssqd.sch != schid->sch_no))
+		rc = -EINVAL;
+
+	if (!rc)
+		memcpy(data, &ssqd->qdio_ssqd, sizeof(*data));
+
+out:
+	if (!irq_ptr)
+		free_page((unsigned long)ssqd);
+
+	return rc;
+}
+
+void qdio_setup_ssqd_info(struct qdio_irq *irq_ptr)
+{
+	unsigned char qdioac;
+	int rc;
+
+	rc = qdio_setup_get_ssqd(irq_ptr, &irq_ptr->schid, &irq_ptr->ssqd_desc);
+	if (rc) {
+		DBF_ERROR("%4x ssqd ERR", irq_ptr->schid.sch_no);
+		DBF_ERROR("rc:%x", rc);
+		/* all flags set, worst case */
+		qdioac = AC1_SIGA_INPUT_NEEDED | AC1_SIGA_OUTPUT_NEEDED |
+			 AC1_SIGA_SYNC_NEEDED;
+	} else
+		qdioac = irq_ptr->ssqd_desc.qdioac1;
+
+	check_and_setup_qebsm(irq_ptr, qdioac, irq_ptr->ssqd_desc.sch_token);
+	process_ac_flags(irq_ptr, qdioac);
+	DBF_EVENT("ac 1:%2x 2:%4x", qdioac, irq_ptr->ssqd_desc.qdioac2);
+	DBF_EVENT("3:%4x qib:%4x", irq_ptr->ssqd_desc.qdioac3, irq_ptr->qib.ac);
+}
+
+void qdio_release_memory(struct qdio_irq *irq_ptr)
+{
+	struct qdio_q *q;
+	int i;
+
+	/*
+	 * Must check queue array manually since irq_ptr->nr_input_queues /
+	 * irq_ptr->nr_input_queues may not yet be set.
+	 */
+	for (i = 0; i < QDIO_MAX_QUEUES_PER_IRQ; i++) {
+		q = irq_ptr->input_qs[i];
+		if (q) {
+			free_page((unsigned long) q->slib);
+			kmem_cache_free(qdio_q_cache, q);
+		}
+	}
+	for (i = 0; i < QDIO_MAX_QUEUES_PER_IRQ; i++) {
+		q = irq_ptr->output_qs[i];
+		if (q) {
+			if (q->u.out.use_cq) {
+				int n;
+
+				for (n = 0; n < QDIO_MAX_BUFFERS_PER_Q; ++n) {
+					struct qaob *aob = q->u.out.aobs[n];
+					if (aob) {
+						qdio_release_aob(aob);
+						q->u.out.aobs[n] = NULL;
+					}
+				}
+
+				qdio_disable_async_operation(&q->u.out);
+			}
+			free_page((unsigned long) q->slib);
+			kmem_cache_free(qdio_q_cache, q);
+		}
+	}
+	free_page((unsigned long) irq_ptr->qdr);
+	free_page(irq_ptr->chsc_page);
+	free_page((unsigned long) irq_ptr);
+}
+
+static void __qdio_allocate_fill_qdr(struct qdio_irq *irq_ptr,
+				     struct qdio_q **irq_ptr_qs,
+				     int i, int nr)
+{
+	irq_ptr->qdr->qdf0[i + nr].sliba =
+		(unsigned long)irq_ptr_qs[i]->slib;
+
+	irq_ptr->qdr->qdf0[i + nr].sla =
+		(unsigned long)irq_ptr_qs[i]->sl;
+
+	irq_ptr->qdr->qdf0[i + nr].slsba =
+		(unsigned long)&irq_ptr_qs[i]->slsb.val[0];
+
+	irq_ptr->qdr->qdf0[i + nr].akey = PAGE_DEFAULT_KEY >> 4;
+	irq_ptr->qdr->qdf0[i + nr].bkey = PAGE_DEFAULT_KEY >> 4;
+	irq_ptr->qdr->qdf0[i + nr].ckey = PAGE_DEFAULT_KEY >> 4;
+	irq_ptr->qdr->qdf0[i + nr].dkey = PAGE_DEFAULT_KEY >> 4;
+}
+
+static void setup_qdr(struct qdio_irq *irq_ptr,
+		      struct qdio_initialize *qdio_init)
+{
+	int i;
+
+	irq_ptr->qdr->qfmt = qdio_init->q_format;
+	irq_ptr->qdr->ac = qdio_init->qdr_ac;
+	irq_ptr->qdr->iqdcnt = qdio_init->no_input_qs;
+	irq_ptr->qdr->oqdcnt = qdio_init->no_output_qs;
+	irq_ptr->qdr->iqdsz = sizeof(struct qdesfmt0) / 4; /* size in words */
+	irq_ptr->qdr->oqdsz = sizeof(struct qdesfmt0) / 4;
+	irq_ptr->qdr->qiba = (unsigned long)&irq_ptr->qib;
+	irq_ptr->qdr->qkey = PAGE_DEFAULT_KEY >> 4;
+
+	for (i = 0; i < qdio_init->no_input_qs; i++)
+		__qdio_allocate_fill_qdr(irq_ptr, irq_ptr->input_qs, i, 0);
+
+	for (i = 0; i < qdio_init->no_output_qs; i++)
+		__qdio_allocate_fill_qdr(irq_ptr, irq_ptr->output_qs, i,
+					 qdio_init->no_input_qs);
+}
+
+static void setup_qib(struct qdio_irq *irq_ptr,
+		      struct qdio_initialize *init_data)
+{
+	if (qebsm_possible())
+		irq_ptr->qib.rflags |= QIB_RFLAGS_ENABLE_QEBSM;
+
+	irq_ptr->qib.rflags |= init_data->qib_rflags;
+
+	irq_ptr->qib.qfmt = init_data->q_format;
+	if (init_data->no_input_qs)
+		irq_ptr->qib.isliba =
+			(unsigned long)(irq_ptr->input_qs[0]->slib);
+	if (init_data->no_output_qs)
+		irq_ptr->qib.osliba =
+			(unsigned long)(irq_ptr->output_qs[0]->slib);
+	memcpy(irq_ptr->qib.ebcnam, init_data->adapter_name, 8);
+}
+
+int qdio_setup_irq(struct qdio_initialize *init_data)
+{
+	struct ciw *ciw;
+	struct qdio_irq *irq_ptr = init_data->cdev->private->qdio_data;
+	int rc;
+
+	memset(&irq_ptr->qib, 0, sizeof(irq_ptr->qib));
+	memset(&irq_ptr->siga_flag, 0, sizeof(irq_ptr->siga_flag));
+	memset(&irq_ptr->ccw, 0, sizeof(irq_ptr->ccw));
+	memset(&irq_ptr->ssqd_desc, 0, sizeof(irq_ptr->ssqd_desc));
+	memset(&irq_ptr->perf_stat, 0, sizeof(irq_ptr->perf_stat));
+
+	irq_ptr->debugfs_dev = irq_ptr->debugfs_perf = NULL;
+	irq_ptr->sch_token = irq_ptr->state = irq_ptr->perf_stat_enabled = 0;
+
+	/* wipes qib.ac, required by ar7063 */
+	memset(irq_ptr->qdr, 0, sizeof(struct qdr));
+
+	irq_ptr->int_parm = init_data->int_parm;
+	irq_ptr->nr_input_qs = init_data->no_input_qs;
+	irq_ptr->nr_output_qs = init_data->no_output_qs;
+	irq_ptr->cdev = init_data->cdev;
+	ccw_device_get_schid(irq_ptr->cdev, &irq_ptr->schid);
+	setup_queues(irq_ptr, init_data);
+
+	setup_qib(irq_ptr, init_data);
+	qdio_setup_thinint(irq_ptr);
+	set_impl_params(irq_ptr, init_data->qib_param_field_format,
+			init_data->qib_param_field,
+			init_data->input_slib_elements,
+			init_data->output_slib_elements);
+
+	/* fill input and output descriptors */
+	setup_qdr(irq_ptr, init_data);
+
+	/* qdr, qib, sls, slsbs, slibs, sbales are filled now */
+
+	/* get qdio commands */
+	ciw = ccw_device_get_ciw(init_data->cdev, CIW_TYPE_EQUEUE);
+	if (!ciw) {
+		DBF_ERROR("%4x NO EQ", irq_ptr->schid.sch_no);
+		rc = -EINVAL;
+		goto out_err;
+	}
+	irq_ptr->equeue = *ciw;
+
+	ciw = ccw_device_get_ciw(init_data->cdev, CIW_TYPE_AQUEUE);
+	if (!ciw) {
+		DBF_ERROR("%4x NO AQ", irq_ptr->schid.sch_no);
+		rc = -EINVAL;
+		goto out_err;
+	}
+	irq_ptr->aqueue = *ciw;
+
+	/* set new interrupt handler */
+	irq_ptr->orig_handler = init_data->cdev->handler;
+	init_data->cdev->handler = qdio_int_handler;
+	return 0;
+out_err:
+	qdio_release_memory(irq_ptr);
+	return rc;
+}
+
+void qdio_print_subchannel_info(struct qdio_irq *irq_ptr,
+				struct ccw_device *cdev)
+{
+	char s[80];
+
+	snprintf(s, 80, "qdio: %s %s on SC %x using "
+		 "AI:%d QEBSM:%d PRI:%d TDD:%d SIGA:%s%s%s%s%s\n",
+		 dev_name(&cdev->dev),
+		 (irq_ptr->qib.qfmt == QDIO_QETH_QFMT) ? "OSA" :
+			((irq_ptr->qib.qfmt == QDIO_ZFCP_QFMT) ? "ZFCP" : "HS"),
+		 irq_ptr->schid.sch_no,
+		 is_thinint_irq(irq_ptr),
+		 (irq_ptr->sch_token) ? 1 : 0,
+		 (irq_ptr->qib.ac & QIB_AC_OUTBOUND_PCI_SUPPORTED) ? 1 : 0,
+		 css_general_characteristics.aif_tdd,
+		 (irq_ptr->siga_flag.input) ? "R" : " ",
+		 (irq_ptr->siga_flag.output) ? "W" : " ",
+		 (irq_ptr->siga_flag.sync) ? "S" : " ",
+		 (irq_ptr->siga_flag.sync_after_ai) ? "A" : " ",
+		 (irq_ptr->siga_flag.sync_out_after_pci) ? "P" : " ");
+	printk(KERN_INFO "%s", s);
+}
+
+int qdio_enable_async_operation(struct qdio_output_q *outq)
+{
+	outq->aobs = kzalloc(sizeof(struct qaob *) * QDIO_MAX_BUFFERS_PER_Q,
+			     GFP_ATOMIC);
+	if (!outq->aobs) {
+		outq->use_cq = 0;
+		return -ENOMEM;
+	}
+	outq->use_cq = 1;
+	return 0;
+}
+
+void qdio_disable_async_operation(struct qdio_output_q *q)
+{
+	kfree(q->aobs);
+	q->aobs = NULL;
+	q->use_cq = 0;
+}
+
+int __init qdio_setup_init(void)
+{
+	int rc;
+
+	qdio_q_cache = kmem_cache_create("qdio_q", sizeof(struct qdio_q),
+					 256, 0, NULL);
+	if (!qdio_q_cache)
+		return -ENOMEM;
+
+	qdio_aob_cache = kmem_cache_create("qdio_aob",
+					sizeof(struct qaob),
+					sizeof(struct qaob),
+					0,
+					NULL);
+	if (!qdio_aob_cache) {
+		rc = -ENOMEM;
+		goto free_qdio_q_cache;
+	}
+
+	/* Check for OSA/FCP thin interrupts (bit 67). */
+	DBF_EVENT("thinint:%1d",
+		  (css_general_characteristics.aif_osa) ? 1 : 0);
+
+	/* Check for QEBSM support in general (bit 58). */
+	DBF_EVENT("cssQEBSM:%1d", (qebsm_possible()) ? 1 : 0);
+	rc = 0;
+out:
+	return rc;
+free_qdio_q_cache:
+	kmem_cache_destroy(qdio_q_cache);
+	goto out;
+}
+
+void qdio_setup_exit(void)
+{
+	kmem_cache_destroy(qdio_aob_cache);
+	kmem_cache_destroy(qdio_q_cache);
+}

drivers/s390/cio/qdio_thinint.c

@@ -0,0 +1,299 @@
+/*
+ * Copyright IBM Corp. 2000, 2009
+ * Author(s): Utz Bacher <utz.bacher@de.ibm.com>
+ *	      Cornelia Huck <cornelia.huck@de.ibm.com>
+ *	      Jan Glauber <jang@linux.vnet.ibm.com>
+ */
+#include <linux/io.h>
+#include <linux/slab.h>
+#include <linux/kernel_stat.h>
+#include <linux/atomic.h>
+#include <asm/debug.h>
+#include <asm/qdio.h>
+#include <asm/airq.h>
+#include <asm/isc.h>
+
+#include "cio.h"
+#include "ioasm.h"
+#include "qdio.h"
+#include "qdio_debug.h"
+
+/*
+ * Restriction: only 63 iqdio subchannels would have its own indicator,
+ * after that, subsequent subchannels share one indicator
+ */
+#define TIQDIO_NR_NONSHARED_IND		63
+#define TIQDIO_NR_INDICATORS		(TIQDIO_NR_NONSHARED_IND + 1)
+#define TIQDIO_SHARED_IND		63
+
+/* device state change indicators */
+struct indicator_t {
+	u32 ind;	/* u32 because of compare-and-swap performance */
+	atomic_t count; /* use count, 0 or 1 for non-shared indicators */
+};
+
+/* list of thin interrupt input queues */
+static LIST_HEAD(tiq_list);
+static DEFINE_MUTEX(tiq_list_lock);
+
+/* Adapter interrupt definitions */
+static void tiqdio_thinint_handler(struct airq_struct *airq);
+
+static struct airq_struct tiqdio_airq = {
+	.handler = tiqdio_thinint_handler,
+	.isc = QDIO_AIRQ_ISC,
+};
+
+static struct indicator_t *q_indicators;
+
+u64 last_ai_time;
+
+/* returns addr for the device state change indicator */
+static u32 *get_indicator(void)
+{
+	int i;
+
+	for (i = 0; i < TIQDIO_NR_NONSHARED_IND; i++)
+		if (!atomic_read(&q_indicators[i].count)) {
+			atomic_set(&q_indicators[i].count, 1);
+			return &q_indicators[i].ind;
+		}
+
+	/* use the shared indicator */
+	atomic_inc(&q_indicators[TIQDIO_SHARED_IND].count);
+	return &q_indicators[TIQDIO_SHARED_IND].ind;
+}
+
+static void put_indicator(u32 *addr)
+{
+	int i;
+
+	if (!addr)
+		return;
+	i = ((unsigned long)addr - (unsigned long)q_indicators) /
+		sizeof(struct indicator_t);
+	atomic_dec(&q_indicators[i].count);
+}
+
+void tiqdio_add_input_queues(struct qdio_irq *irq_ptr)
+{
+	mutex_lock(&tiq_list_lock);
+	list_add_rcu(&irq_ptr->input_qs[0]->entry, &tiq_list);
+	mutex_unlock(&tiq_list_lock);
+	xchg(irq_ptr->dsci, 1 << 7);
+}
+
+void tiqdio_remove_input_queues(struct qdio_irq *irq_ptr)
+{
+	struct qdio_q *q;
+
+	q = irq_ptr->input_qs[0];
+	/* if establish triggered an error */
+	if (!q || !q->entry.prev || !q->entry.next)
+		return;
+
+	mutex_lock(&tiq_list_lock);
+	list_del_rcu(&q->entry);
+	mutex_unlock(&tiq_list_lock);
+	synchronize_rcu();
+}
+
+static inline int has_multiple_inq_on_dsci(struct qdio_irq *irq_ptr)
+{
+	return irq_ptr->nr_input_qs > 1;
+}
+
+static inline int references_shared_dsci(struct qdio_irq *irq_ptr)
+{
+	return irq_ptr->dsci == &q_indicators[TIQDIO_SHARED_IND].ind;
+}
+
+static inline int shared_ind(struct qdio_irq *irq_ptr)
+{
+	return references_shared_dsci(irq_ptr) ||
+		has_multiple_inq_on_dsci(irq_ptr);
+}
+
+void clear_nonshared_ind(struct qdio_irq *irq_ptr)
+{
+	if (!is_thinint_irq(irq_ptr))
+		return;
+	if (shared_ind(irq_ptr))
+		return;
+	xchg(irq_ptr->dsci, 0);
+}
+
+int test_nonshared_ind(struct qdio_irq *irq_ptr)
+{
+	if (!is_thinint_irq(irq_ptr))
+		return 0;
+	if (shared_ind(irq_ptr))
+		return 0;
+	if (*irq_ptr->dsci)
+		return 1;
+	else
+		return 0;
+}
+
+static inline u32 clear_shared_ind(void)
+{
+	if (!atomic_read(&q_indicators[TIQDIO_SHARED_IND].count))
+		return 0;
+	return xchg(&q_indicators[TIQDIO_SHARED_IND].ind, 0);
+}
+
+static inline void tiqdio_call_inq_handlers(struct qdio_irq *irq)
+{
+	struct qdio_q *q;
+	int i;
+
+	for_each_input_queue(irq, q, i) {
+		if (!references_shared_dsci(irq) &&
+		    has_multiple_inq_on_dsci(irq))
+			xchg(q->irq_ptr->dsci, 0);
+
+		if (q->u.in.queue_start_poll) {
+			/* skip if polling is enabled or already in work */
+			if (test_and_set_bit(QDIO_QUEUE_IRQS_DISABLED,
+					     &q->u.in.queue_irq_state)) {
+				qperf_inc(q, int_discarded);
+				continue;
+			}
+
+			/* avoid dsci clear here, done after processing */
+			q->u.in.queue_start_poll(q->irq_ptr->cdev, q->nr,
+						 q->irq_ptr->int_parm);
+		} else {
+			if (!shared_ind(q->irq_ptr))
+				xchg(q->irq_ptr->dsci, 0);
+
+			/*
+			 * Call inbound processing but not directly
+			 * since that could starve other thinint queues.
+			 */
+			tasklet_schedule(&q->tasklet);
+		}
+	}
+}
+
+/**
+ * tiqdio_thinint_handler - thin interrupt handler for qdio
+ * @alsi: pointer to adapter local summary indicator
+ * @data: NULL
+ */
+static void tiqdio_thinint_handler(struct airq_struct *airq)
+{
+	u32 si_used = clear_shared_ind();
+	struct qdio_q *q;
+
+	last_ai_time = S390_lowcore.int_clock;
+	inc_irq_stat(IRQIO_QAI);
+
+	/* protect tiq_list entries, only changed in activate or shutdown */
+	rcu_read_lock();
+
+	/* check for work on all inbound thinint queues */
+	list_for_each_entry_rcu(q, &tiq_list, entry) {
+		struct qdio_irq *irq;
+
+		/* only process queues from changed sets */
+		irq = q->irq_ptr;
+		if (unlikely(references_shared_dsci(irq))) {
+			if (!si_used)
+				continue;
+		} else if (!*irq->dsci)
+			continue;
+
+		tiqdio_call_inq_handlers(irq);
+
+		qperf_inc(q, adapter_int);
+	}
+	rcu_read_unlock();
+}
+
+static int set_subchannel_ind(struct qdio_irq *irq_ptr, int reset)
+{
+	struct chsc_scssc_area *scssc = (void *)irq_ptr->chsc_page;
+	u64 summary_indicator_addr, subchannel_indicator_addr;
+	int rc;
+
+	if (reset) {
+		summary_indicator_addr = 0;
+		subchannel_indicator_addr = 0;
+	} else {
+		summary_indicator_addr = virt_to_phys(tiqdio_airq.lsi_ptr);
+		subchannel_indicator_addr = virt_to_phys(irq_ptr->dsci);
+	}
+
+	rc = chsc_sadc(irq_ptr->schid, scssc, summary_indicator_addr,
+		       subchannel_indicator_addr);
+	if (rc) {
+		DBF_ERROR("%4x SSI r:%4x", irq_ptr->schid.sch_no,
+			  scssc->response.code);
+		goto out;
+	}
+
+	DBF_EVENT("setscind");
+	DBF_HEX(&summary_indicator_addr, sizeof(summary_indicator_addr));
+	DBF_HEX(&subchannel_indicator_addr, sizeof(subchannel_indicator_addr));
+out:
+	return rc;
+}
+
+/* allocate non-shared indicators and shared indicator */
+int __init tiqdio_allocate_memory(void)
+{
+	q_indicators = kzalloc(sizeof(struct indicator_t) * TIQDIO_NR_INDICATORS,
+			     GFP_KERNEL);
+	if (!q_indicators)
+		return -ENOMEM;
+	return 0;
+}
+
+void tiqdio_free_memory(void)
+{
+	kfree(q_indicators);
+}
+
+int __init tiqdio_register_thinints(void)
+{
+	int rc;
+
+	rc = register_adapter_interrupt(&tiqdio_airq);
+	if (rc) {
+		DBF_EVENT("RTI:%x", rc);
+		return rc;
+	}
+	return 0;
+}
+
+int qdio_establish_thinint(struct qdio_irq *irq_ptr)
+{
+	if (!is_thinint_irq(irq_ptr))
+		return 0;
+	return set_subchannel_ind(irq_ptr, 0);
+}
+
+void qdio_setup_thinint(struct qdio_irq *irq_ptr)
+{
+	if (!is_thinint_irq(irq_ptr))
+		return;
+	irq_ptr->dsci = get_indicator();
+	DBF_HEX(&irq_ptr->dsci, sizeof(void *));
+}
+
+void qdio_shutdown_thinint(struct qdio_irq *irq_ptr)
+{
+	if (!is_thinint_irq(irq_ptr))
+		return;
+
+	/* reset adapter interrupt indicators */
+	set_subchannel_ind(irq_ptr, 1);
+	put_indicator(irq_ptr->dsci);
+}
+
+void __exit tiqdio_unregister_thinints(void)
+{
+	WARN_ON(!list_empty(&tiq_list));
+	unregister_adapter_interrupt(&tiqdio_airq);
+}

drivers/s390/cio/scm.c

@@ -0,0 +1,288 @@
+/*
+ * Recognize and maintain s390 storage class memory.
+ *
+ * Copyright IBM Corp. 2012
+ * Author(s): Sebastian Ott <sebott@linux.vnet.ibm.com>
+ */
+
+#include <linux/device.h>
+#include <linux/module.h>
+#include <linux/mutex.h>
+#include <linux/slab.h>
+#include <linux/init.h>
+#include <linux/err.h>
+#include <asm/eadm.h>
+#include "chsc.h"
+
+static struct device *scm_root;
+
+#define to_scm_dev(n) container_of(n, struct scm_device, dev)
+#define	to_scm_drv(d) container_of(d, struct scm_driver, drv)
+
+static int scmdev_probe(struct device *dev)
+{
+	struct scm_device *scmdev = to_scm_dev(dev);
+	struct scm_driver *scmdrv = to_scm_drv(dev->driver);
+
+	return scmdrv->probe ? scmdrv->probe(scmdev) : -ENODEV;
+}
+
+static int scmdev_remove(struct device *dev)
+{
+	struct scm_device *scmdev = to_scm_dev(dev);
+	struct scm_driver *scmdrv = to_scm_drv(dev->driver);
+
+	return scmdrv->remove ? scmdrv->remove(scmdev) : -ENODEV;
+}
+
+static int scmdev_uevent(struct device *dev, struct kobj_uevent_env *env)
+{
+	return add_uevent_var(env, "MODALIAS=scm:scmdev");
+}
+
+static struct bus_type scm_bus_type = {
+	.name  = "scm",
+	.probe = scmdev_probe,
+	.remove = scmdev_remove,
+	.uevent = scmdev_uevent,
+};
+
+/**
+ * scm_driver_register() - register a scm driver
+ * @scmdrv: driver to be registered
+ */
+int scm_driver_register(struct scm_driver *scmdrv)
+{
+	struct device_driver *drv = &scmdrv->drv;
+
+	drv->bus = &scm_bus_type;
+
+	return driver_register(drv);
+}
+EXPORT_SYMBOL_GPL(scm_driver_register);
+
+/**
+ * scm_driver_unregister() - deregister a scm driver
+ * @scmdrv: driver to be deregistered
+ */
+void scm_driver_unregister(struct scm_driver *scmdrv)
+{
+	driver_unregister(&scmdrv->drv);
+}
+EXPORT_SYMBOL_GPL(scm_driver_unregister);
+
+void scm_irq_handler(struct aob *aob, int error)
+{
+	struct aob_rq_header *aobrq = (void *) aob->request.data;
+	struct scm_device *scmdev = aobrq->scmdev;
+	struct scm_driver *scmdrv = to_scm_drv(scmdev->dev.driver);
+
+	scmdrv->handler(scmdev, aobrq->data, error);
+}
+EXPORT_SYMBOL_GPL(scm_irq_handler);
+
+#define scm_attr(name)							\
+static ssize_t show_##name(struct device *dev,				\
+	       struct device_attribute *attr, char *buf)		\
+{									\
+	struct scm_device *scmdev = to_scm_dev(dev);			\
+	int ret;							\
+									\
+	device_lock(dev);						\
+	ret = sprintf(buf, "%u\n", scmdev->attrs.name);			\
+	device_unlock(dev);						\
+									\
+	return ret;							\
+}									\
+static DEVICE_ATTR(name, S_IRUGO, show_##name, NULL);
+
+scm_attr(persistence);
+scm_attr(oper_state);
+scm_attr(data_state);
+scm_attr(rank);
+scm_attr(release);
+scm_attr(res_id);
+
+static struct attribute *scmdev_attrs[] = {
+	&dev_attr_persistence.attr,
+	&dev_attr_oper_state.attr,
+	&dev_attr_data_state.attr,
+	&dev_attr_rank.attr,
+	&dev_attr_release.attr,
+	&dev_attr_res_id.attr,
+	NULL,
+};
+
+static struct attribute_group scmdev_attr_group = {
+	.attrs = scmdev_attrs,
+};
+
+static const struct attribute_group *scmdev_attr_groups[] = {
+	&scmdev_attr_group,
+	NULL,
+};
+
+static void scmdev_release(struct device *dev)
+{
+	struct scm_device *scmdev = to_scm_dev(dev);
+
+	kfree(scmdev);
+}
+
+static void scmdev_setup(struct scm_device *scmdev, struct sale *sale,
+			 unsigned int size, unsigned int max_blk_count)
+{
+	dev_set_name(&scmdev->dev, "%016llx", (unsigned long long) sale->sa);
+	scmdev->nr_max_block = max_blk_count;
+	scmdev->address = sale->sa;
+	scmdev->size = 1UL << size;
+	scmdev->attrs.rank = sale->rank;
+	scmdev->attrs.persistence = sale->p;
+	scmdev->attrs.oper_state = sale->op_state;
+	scmdev->attrs.data_state = sale->data_state;
+	scmdev->attrs.rank = sale->rank;
+	scmdev->attrs.release = sale->r;
+	scmdev->attrs.res_id = sale->rid;
+	scmdev->dev.parent = scm_root;
+	scmdev->dev.bus = &scm_bus_type;
+	scmdev->dev.release = scmdev_release;
+	scmdev->dev.groups = scmdev_attr_groups;
+}
+
+/*
+ * Check for state-changes, notify the driver and userspace.
+ */
+static void scmdev_update(struct scm_device *scmdev, struct sale *sale)
+{
+	struct scm_driver *scmdrv;
+	bool changed;
+
+	device_lock(&scmdev->dev);
+	changed = scmdev->attrs.rank != sale->rank ||
+		  scmdev->attrs.oper_state != sale->op_state;
+	scmdev->attrs.rank = sale->rank;
+	scmdev->attrs.oper_state = sale->op_state;
+	if (!scmdev->dev.driver)
+		goto out;
+	scmdrv = to_scm_drv(scmdev->dev.driver);
+	if (changed && scmdrv->notify)
+		scmdrv->notify(scmdev, SCM_CHANGE);
+out:
+	device_unlock(&scmdev->dev);
+	if (changed)
+		kobject_uevent(&scmdev->dev.kobj, KOBJ_CHANGE);
+}
+
+static int check_address(struct device *dev, void *data)
+{
+	struct scm_device *scmdev = to_scm_dev(dev);
+	struct sale *sale = data;
+
+	return scmdev->address == sale->sa;
+}
+
+static struct scm_device *scmdev_find(struct sale *sale)
+{
+	struct device *dev;
+
+	dev = bus_find_device(&scm_bus_type, NULL, sale, check_address);
+
+	return dev ? to_scm_dev(dev) : NULL;
+}
+
+static int scm_add(struct chsc_scm_info *scm_info, size_t num)
+{
+	struct sale *sale, *scmal = scm_info->scmal;
+	struct scm_device *scmdev;
+	int ret;
+
+	for (sale = scmal; sale < scmal + num; sale++) {
+		scmdev = scmdev_find(sale);
+		if (scmdev) {
+			scmdev_update(scmdev, sale);
+			/* Release reference from scm_find(). */
+			put_device(&scmdev->dev);
+			continue;
+		}
+		scmdev = kzalloc(sizeof(*scmdev), GFP_KERNEL);
+		if (!scmdev)
+			return -ENODEV;
+		scmdev_setup(scmdev, sale, scm_info->is, scm_info->mbc);
+		ret = device_register(&scmdev->dev);
+		if (ret) {
+			/* Release reference from device_initialize(). */
+			put_device(&scmdev->dev);
+			return ret;
+		}
+	}
+
+	return 0;
+}
+
+int scm_update_information(void)
+{
+	struct chsc_scm_info *scm_info;
+	u64 token = 0;
+	size_t num;
+	int ret;
+
+	scm_info = (void *)__get_free_page(GFP_KERNEL | GFP_DMA);
+	if (!scm_info)
+		return -ENOMEM;
+
+	do {
+		ret = chsc_scm_info(scm_info, token);
+		if (ret)
+			break;
+
+		num = (scm_info->response.length -
+		       (offsetof(struct chsc_scm_info, scmal) -
+			offsetof(struct chsc_scm_info, response))
+		      ) / sizeof(struct sale);
+
+		ret = scm_add(scm_info, num);
+		if (ret)
+			break;
+
+		token = scm_info->restok;
+	} while (token);
+
+	free_page((unsigned long)scm_info);
+
+	return ret;
+}
+
+static int scm_dev_avail(struct device *dev, void *unused)
+{
+	struct scm_driver *scmdrv = to_scm_drv(dev->driver);
+	struct scm_device *scmdev = to_scm_dev(dev);
+
+	if (dev->driver && scmdrv->notify)
+		scmdrv->notify(scmdev, SCM_AVAIL);
+
+	return 0;
+}
+
+int scm_process_availability_information(void)
+{
+	return bus_for_each_dev(&scm_bus_type, NULL, NULL, scm_dev_avail);
+}
+
+static int __init scm_init(void)
+{
+	int ret;
+
+	ret = bus_register(&scm_bus_type);
+	if (ret)
+		return ret;
+
+	scm_root = root_device_register("scm");
+	if (IS_ERR(scm_root)) {
+		bus_unregister(&scm_bus_type);
+		return PTR_ERR(scm_root);
+	}
+
+	scm_update_information();
+	return 0;
+}
+subsys_initcall_sync(scm_init);