[linux.git] / drivers / xen / mcelog.c

/******************************************************************************
 * mcelog.c
 * Driver for receiving and transferring machine check error infomation
 *
 * Copyright (c) 2012 Intel Corporation
 * Author: Liu, Jinsong <[email protected]>
 * Author: Jiang, Yunhong <[email protected]>
 * Author: Ke, Liping <[email protected]>
 *
 * This program is free software; you can redistribute it and/or
 * modify it under the terms of the GNU General Public License version 2
 * as published by the Free Software Foundation; or, when distributed
 * separately from the Linux kernel or incorporated into other
 * software packages, subject to the following license:
 *
 * Permission is hereby granted, free of charge, to any person obtaining a copy
 * of this source file (the "Software"), to deal in the Software without
 * restriction, including without limitation the rights to use, copy, modify,
 * merge, publish, distribute, sublicense, and/or sell copies of the Software,
 * and to permit persons to whom the Software is furnished to do so, subject to
 * the following conditions:
 *
 * The above copyright notice and this permission notice shall be included in
 * all copies or substantial portions of the Software.
 *
 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
 * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
 * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
 * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
 * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
 * IN THE SOFTWARE.
 */

#define pr_fmt(fmt) "xen_mcelog: " fmt

#include <linux/init.h>
#include <linux/types.h>
#include <linux/kernel.h>
#include <linux/slab.h>
#include <linux/fs.h>
#include <linux/device.h>
#include <linux/miscdevice.h>
#include <linux/uaccess.h>
#include <linux/capability.h>
#include <linux/poll.h>
#include <linux/sched.h>

#include <xen/interface/xen.h>
#include <xen/events.h>
#include <xen/interface/vcpu.h>
#include <xen/xen.h>
#include <asm/xen/hypercall.h>
#include <asm/xen/hypervisor.h>

static struct mc_info g_mi;
static struct mcinfo_logical_cpu *g_physinfo;
static uint32_t ncpus;

static DEFINE_MUTEX(mcelog_lock);

static struct xen_mce_log xen_mcelog = {
	.signature	= XEN_MCE_LOG_SIGNATURE,
	.len		= XEN_MCE_LOG_LEN,
	.recordlen	= sizeof(struct xen_mce),
};

static DEFINE_SPINLOCK(xen_mce_chrdev_state_lock);
static int xen_mce_chrdev_open_count;	/* #times opened */
static int xen_mce_chrdev_open_exclu;	/* already open exclusive? */

static DECLARE_WAIT_QUEUE_HEAD(xen_mce_chrdev_wait);

static int xen_mce_chrdev_open(struct inode *inode, struct file *file)
{
	spin_lock(&xen_mce_chrdev_state_lock);

	if (xen_mce_chrdev_open_exclu ||
	    (xen_mce_chrdev_open_count && (file->f_flags & O_EXCL))) {
		spin_unlock(&xen_mce_chrdev_state_lock);

		return -EBUSY;
	}

	if (file->f_flags & O_EXCL)
		xen_mce_chrdev_open_exclu = 1;
	xen_mce_chrdev_open_count++;

	spin_unlock(&xen_mce_chrdev_state_lock);

	return nonseekable_open(inode, file);
}

static int xen_mce_chrdev_release(struct inode *inode, struct file *file)
{
	spin_lock(&xen_mce_chrdev_state_lock);

	xen_mce_chrdev_open_count--;
	xen_mce_chrdev_open_exclu = 0;

	spin_unlock(&xen_mce_chrdev_state_lock);

	return 0;
}

static ssize_t xen_mce_chrdev_read(struct file *filp, char __user *ubuf,
				size_t usize, loff_t *off)
{
	char __user *buf = ubuf;
	unsigned num;
	int i, err;

	mutex_lock(&mcelog_lock);

	num = xen_mcelog.next;

	/* Only supports full reads right now */
	err = -EINVAL;
	if (*off != 0 || usize < XEN_MCE_LOG_LEN*sizeof(struct xen_mce))
		goto out;

	err = 0;
	for (i = 0; i < num; i++) {
		struct xen_mce *m = &xen_mcelog.entry[i];

		err |= copy_to_user(buf, m, sizeof(*m));
		buf += sizeof(*m);
	}

	memset(xen_mcelog.entry, 0, num * sizeof(struct xen_mce));
	xen_mcelog.next = 0;

	if (err)
		err = -EFAULT;

out:
	mutex_unlock(&mcelog_lock);

	return err ? err : buf - ubuf;
}

static unsigned int xen_mce_chrdev_poll(struct file *file, poll_table *wait)
{
	poll_wait(file, &xen_mce_chrdev_wait, wait);

	if (xen_mcelog.next)
		return POLLIN | POLLRDNORM;

	return 0;
}

static long xen_mce_chrdev_ioctl(struct file *f, unsigned int cmd,
				unsigned long arg)
{
	int __user *p = (int __user *)arg;

	if (!capable(CAP_SYS_ADMIN))
		return -EPERM;

	switch (cmd) {
	case MCE_GET_RECORD_LEN:
		return put_user(sizeof(struct xen_mce), p);
	case MCE_GET_LOG_LEN:
		return put_user(XEN_MCE_LOG_LEN, p);
	case MCE_GETCLEAR_FLAGS: {
		unsigned flags;

		do {
			flags = xen_mcelog.flags;
		} while (cmpxchg(&xen_mcelog.flags, flags, 0) != flags);

		return put_user(flags, p);
	}
	default:
		return -ENOTTY;
	}
}

static const struct file_operations xen_mce_chrdev_ops = {
	.open			= xen_mce_chrdev_open,
	.release		= xen_mce_chrdev_release,
	.read			= xen_mce_chrdev_read,
	.poll			= xen_mce_chrdev_poll,
	.unlocked_ioctl		= xen_mce_chrdev_ioctl,
	.llseek			= no_llseek,
};

static struct miscdevice xen_mce_chrdev_device = {
	MISC_MCELOG_MINOR,
	"mcelog",
	&xen_mce_chrdev_ops,
};

/*
 * Caller should hold the mcelog_lock
 */
static void xen_mce_log(struct xen_mce *mce)
{
	unsigned entry;

	entry = xen_mcelog.next;

	/*
	 * When the buffer fills up discard new entries.
	 * Assume that the earlier errors are the more
	 * interesting ones:
	 */
	if (entry >= XEN_MCE_LOG_LEN) {
		set_bit(XEN_MCE_OVERFLOW,
			(unsigned long *)&xen_mcelog.flags);
		return;
	}

	memcpy(xen_mcelog.entry + entry, mce, sizeof(struct xen_mce));

	xen_mcelog.next++;
}

static int convert_log(struct mc_info *mi)
{
	struct mcinfo_common *mic;
	struct mcinfo_global *mc_global;
	struct mcinfo_bank *mc_bank;
	struct xen_mce m;
	uint32_t i;

	mic = NULL;
	x86_mcinfo_lookup(&mic, mi, MC_TYPE_GLOBAL);
	if (unlikely(!mic)) {
		pr_warn("Failed to find global error info\n");
		return -ENODEV;
	}

	memset(&m, 0, sizeof(struct xen_mce));

	mc_global = (struct mcinfo_global *)mic;
	m.mcgstatus = mc_global->mc_gstatus;
	m.apicid = mc_global->mc_apicid;

	for (i = 0; i < ncpus; i++)
		if (g_physinfo[i].mc_apicid == m.apicid)
			break;
	if (unlikely(i == ncpus)) {
		pr_warn("Failed to match cpu with apicid %d\n", m.apicid);
		return -ENODEV;
	}

	m.socketid = g_physinfo[i].mc_chipid;
	m.cpu = m.extcpu = g_physinfo[i].mc_cpunr;
	m.cpuvendor = (__u8)g_physinfo[i].mc_vendor;
	m.mcgcap = g_physinfo[i].mc_msrvalues[__MC_MSR_MCGCAP].value;

	mic = NULL;
	x86_mcinfo_lookup(&mic, mi, MC_TYPE_BANK);
	if (unlikely(!mic)) {
		pr_warn("Fail to find bank error info\n");
		return -ENODEV;
	}

	do {
		if ((!mic) || (mic->size == 0) ||
		    (mic->type != MC_TYPE_GLOBAL   &&
		     mic->type != MC_TYPE_BANK     &&
		     mic->type != MC_TYPE_EXTENDED &&
		     mic->type != MC_TYPE_RECOVERY))
			break;

		if (mic->type == MC_TYPE_BANK) {
			mc_bank = (struct mcinfo_bank *)mic;
			m.misc = mc_bank->mc_misc;
			m.status = mc_bank->mc_status;
			m.addr = mc_bank->mc_addr;
			m.tsc = mc_bank->mc_tsc;
			m.bank = mc_bank->mc_bank;
			m.finished = 1;
			/*log this record*/
			xen_mce_log(&m);
		}
		mic = x86_mcinfo_next(mic);
	} while (1);

	return 0;
}

static int mc_queue_handle(uint32_t flags)
{
	struct xen_mc mc_op;
	int ret = 0;

	mc_op.cmd = XEN_MC_fetch;
	mc_op.interface_version = XEN_MCA_INTERFACE_VERSION;
	set_xen_guest_handle(mc_op.u.mc_fetch.data, &g_mi);
	do {
		mc_op.u.mc_fetch.flags = flags;
		ret = HYPERVISOR_mca(&mc_op);
		if (ret) {
			pr_err("Failed to fetch %surgent error log\n",
			       flags == XEN_MC_URGENT ? "" : "non");
			break;
		}

		if (mc_op.u.mc_fetch.flags & XEN_MC_NODATA ||
		    mc_op.u.mc_fetch.flags & XEN_MC_FETCHFAILED)
			break;
		else {
			ret = convert_log(&g_mi);
			if (ret)
				pr_warn("Failed to convert this error log, continue acking it anyway\n");

			mc_op.u.mc_fetch.flags = flags | XEN_MC_ACK;
			ret = HYPERVISOR_mca(&mc_op);
			if (ret) {
				pr_err("Failed to ack previous error log\n");
				break;
			}
		}
	} while (1);

	return ret;
}

/* virq handler for machine check error info*/
static void xen_mce_work_fn(struct work_struct *work)
{
	int err;

	mutex_lock(&mcelog_lock);

	/* urgent mc_info */
	err = mc_queue_handle(XEN_MC_URGENT);
	if (err)
		pr_err("Failed to handle urgent mc_info queue, continue handling nonurgent mc_info queue anyway\n");

	/* nonurgent mc_info */
	err = mc_queue_handle(XEN_MC_NONURGENT);
	if (err)
		pr_err("Failed to handle nonurgent mc_info queue\n");

	/* wake processes polling /dev/mcelog */
	wake_up_interruptible(&xen_mce_chrdev_wait);

	mutex_unlock(&mcelog_lock);
}
static DECLARE_WORK(xen_mce_work, xen_mce_work_fn);

static irqreturn_t xen_mce_interrupt(int irq, void *dev_id)
{
	schedule_work(&xen_mce_work);
	return IRQ_HANDLED;
}

static int bind_virq_for_mce(void)
{
	int ret;
	struct xen_mc mc_op;

	memset(&mc_op, 0, sizeof(struct xen_mc));

	/* Fetch physical CPU Numbers */
	mc_op.cmd = XEN_MC_physcpuinfo;
	mc_op.interface_version = XEN_MCA_INTERFACE_VERSION;
	set_xen_guest_handle(mc_op.u.mc_physcpuinfo.info, g_physinfo);
	ret = HYPERVISOR_mca(&mc_op);
	if (ret) {
		pr_err("Failed to get CPU numbers\n");
		return ret;
	}

	/* Fetch each CPU Physical Info for later reference*/
	ncpus = mc_op.u.mc_physcpuinfo.ncpus;
	g_physinfo = kcalloc(ncpus, sizeof(struct mcinfo_logical_cpu),
			     GFP_KERNEL);
	if (!g_physinfo)
		return -ENOMEM;
	set_xen_guest_handle(mc_op.u.mc_physcpuinfo.info, g_physinfo);
	ret = HYPERVISOR_mca(&mc_op);
	if (ret) {
		pr_err("Failed to get CPU info\n");
		kfree(g_physinfo);
		return ret;
	}

	ret  = bind_virq_to_irqhandler(VIRQ_MCA, 0,
				       xen_mce_interrupt, 0, "mce", NULL);
	if (ret < 0) {
		pr_err("Failed to bind virq\n");
		kfree(g_physinfo);
		return ret;
	}

	return 0;
}

static int __init xen_late_init_mcelog(void)
{
	/* Only DOM0 is responsible for MCE logging */
	if (xen_initial_domain()) {
		/* register character device /dev/mcelog for xen mcelog */
		if (misc_register(&xen_mce_chrdev_device))
			return -ENODEV;
		return bind_virq_for_mce();
	}

	return -ENODEV;
}
device_initcall(xen_late_init_mcelog);
Commit	Line	Data
cef12ee5 LJ	1	/******************************************************************************
	2	* mcelog.c
	3	* Driver for receiving and transferring machine check error infomation
	4	*
	5	* Copyright (c) 2012 Intel Corporation
	6	* Author: Liu, Jinsong <[email protected]>
	7	* Author: Jiang, Yunhong <[email protected]>
	8	* Author: Ke, Liping <[email protected]>
	9	*
	10	* This program is free software; you can redistribute it and/or
	11	* modify it under the terms of the GNU General Public License version 2
	12	* as published by the Free Software Foundation; or, when distributed
	13	* separately from the Linux kernel or incorporated into other
	14	* software packages, subject to the following license:
	15	*
	16	* Permission is hereby granted, free of charge, to any person obtaining a copy
	17	* of this source file (the "Software"), to deal in the Software without
	18	* restriction, including without limitation the rights to use, copy, modify,
	19	* merge, publish, distribute, sublicense, and/or sell copies of the Software,
	20	* and to permit persons to whom the Software is furnished to do so, subject to
	21	* the following conditions:
	22	*
	23	* The above copyright notice and this permission notice shall be included in
	24	* all copies or substantial portions of the Software.
	25	*
	26	* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
	27	* IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
	28	* FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
	29	* AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
	30	* LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
	31	* FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
	32	* IN THE SOFTWARE.
	33	*/
	34
283c0972 JP	35	#define pr_fmt(fmt) "xen_mcelog: " fmt
283c0972 JP	36
cef12ee5 LJ	37	#include <linux/init.h>
	38	#include <linux/types.h>
	39	#include <linux/kernel.h>
	40	#include <linux/slab.h>
	41	#include <linux/fs.h>
	42	#include <linux/device.h>
	43	#include <linux/miscdevice.h>
	44	#include <linux/uaccess.h>
	45	#include <linux/capability.h>
a867e5d6 LJ	46	#include <linux/poll.h>
a867e5d6 LJ	47	#include <linux/sched.h>
cef12ee5 LJ	48
	49	#include <xen/interface/xen.h>
	50	#include <xen/events.h>
	51	#include <xen/interface/vcpu.h>
	52	#include <xen/xen.h>
	53	#include <asm/xen/hypercall.h>
	54	#include <asm/xen/hypervisor.h>
	55
cef12ee5 LJ	56	static struct mc_info g_mi;
	57	static struct mcinfo_logical_cpu *g_physinfo;
	58	static uint32_t ncpus;
	59
1b2a0551	60	static DEFINE_MUTEX(mcelog_lock);
cef12ee5 LJ	61
	62	static struct xen_mce_log xen_mcelog = {
	63	.signature = XEN_MCE_LOG_SIGNATURE,
	64	.len = XEN_MCE_LOG_LEN,
	65	.recordlen = sizeof(struct xen_mce),
	66	};
	67
	68	static DEFINE_SPINLOCK(xen_mce_chrdev_state_lock);
	69	static int xen_mce_chrdev_open_count; /* #times opened */
	70	static int xen_mce_chrdev_open_exclu; /* already open exclusive? */
	71
a867e5d6 LJ	72	static DECLARE_WAIT_QUEUE_HEAD(xen_mce_chrdev_wait);
a867e5d6 LJ	73
cef12ee5 LJ	74	static int xen_mce_chrdev_open(struct inode inode, struct file file)
	75	{
	76	spin_lock(&xen_mce_chrdev_state_lock);
	77
	78	if (xen_mce_chrdev_open_exclu \|\|
	79	(xen_mce_chrdev_open_count && (file->f_flags & O_EXCL))) {
	80	spin_unlock(&xen_mce_chrdev_state_lock);
	81
	82	return -EBUSY;
	83	}
	84
	85	if (file->f_flags & O_EXCL)
	86	xen_mce_chrdev_open_exclu = 1;
	87	xen_mce_chrdev_open_count++;
	88
	89	spin_unlock(&xen_mce_chrdev_state_lock);
	90
	91	return nonseekable_open(inode, file);
	92	}
	93
	94	static int xen_mce_chrdev_release(struct inode inode, struct file file)
	95	{
	96	spin_lock(&xen_mce_chrdev_state_lock);
	97
	98	xen_mce_chrdev_open_count--;
	99	xen_mce_chrdev_open_exclu = 0;
	100
	101	spin_unlock(&xen_mce_chrdev_state_lock);
	102
	103	return 0;
	104	}
	105
	106	static ssize_t xen_mce_chrdev_read(struct file filp, char __user ubuf,
	107	size_t usize, loff_t *off)
	108	{
	109	char __user *buf = ubuf;
	110	unsigned num;
	111	int i, err;
	112
1b2a0551	113	mutex_lock(&mcelog_lock);
cef12ee5 LJ	114
	115	num = xen_mcelog.next;
	116
	117	/* Only supports full reads right now */
	118	err = -EINVAL;
	119	if (off != 0 \|\| usize < XEN_MCE_LOG_LENsizeof(struct xen_mce))
	120	goto out;
	121
	122	err = 0;
	123	for (i = 0; i < num; i++) {
	124	struct xen_mce *m = &xen_mcelog.entry[i];
	125
	126	err \|= copy_to_user(buf, m, sizeof(*m));
	127	buf += sizeof(*m);
	128	}
	129
	130	memset(xen_mcelog.entry, 0, num * sizeof(struct xen_mce));
	131	xen_mcelog.next = 0;
	132
	133	if (err)
	134	err = -EFAULT;
	135
	136	out:
1b2a0551	137	mutex_unlock(&mcelog_lock);
cef12ee5 LJ	138
	139	return err ? err : buf - ubuf;
	140	}
	141
a867e5d6 LJ	142	static unsigned int xen_mce_chrdev_poll(struct file file, poll_table wait)
	143	{
	144	poll_wait(file, &xen_mce_chrdev_wait, wait);
	145
	146	if (xen_mcelog.next)
	147	return POLLIN \| POLLRDNORM;
	148
	149	return 0;
	150	}
	151
cef12ee5 LJ	152	static long xen_mce_chrdev_ioctl(struct file *f, unsigned int cmd,
	153	unsigned long arg)
	154	{
	155	int __user p = (int __user )arg;
	156
	157	if (!capable(CAP_SYS_ADMIN))
	158	return -EPERM;
	159
	160	switch (cmd) {
	161	case MCE_GET_RECORD_LEN:
	162	return put_user(sizeof(struct xen_mce), p);
	163	case MCE_GET_LOG_LEN:
	164	return put_user(XEN_MCE_LOG_LEN, p);
	165	case MCE_GETCLEAR_FLAGS: {
	166	unsigned flags;
	167
	168	do {
	169	flags = xen_mcelog.flags;
	170	} while (cmpxchg(&xen_mcelog.flags, flags, 0) != flags);
	171
	172	return put_user(flags, p);
	173	}
	174	default:
	175	return -ENOTTY;
	176	}
	177	}
	178
	179	static const struct file_operations xen_mce_chrdev_ops = {
	180	.open = xen_mce_chrdev_open,
	181	.release = xen_mce_chrdev_release,
	182	.read = xen_mce_chrdev_read,
a867e5d6	183	.poll = xen_mce_chrdev_poll,
cef12ee5 LJ	184	.unlocked_ioctl = xen_mce_chrdev_ioctl,
	185	.llseek = no_llseek,
	186	};
	187
	188	static struct miscdevice xen_mce_chrdev_device = {
	189	MISC_MCELOG_MINOR,
	190	"mcelog",
	191	&xen_mce_chrdev_ops,
	192	};
	193
	194	/*
	195	* Caller should hold the mcelog_lock
	196	*/
	197	static void xen_mce_log(struct xen_mce *mce)
	198	{
	199	unsigned entry;
	200
	201	entry = xen_mcelog.next;
	202
	203	/*
	204	* When the buffer fills up discard new entries.
	205	* Assume that the earlier errors are the more
	206	* interesting ones:
	207	*/
	208	if (entry >= XEN_MCE_LOG_LEN) {
	209	set_bit(XEN_MCE_OVERFLOW,
	210	(unsigned long *)&xen_mcelog.flags);
	211	return;
	212	}
	213
	214	memcpy(xen_mcelog.entry + entry, mce, sizeof(struct xen_mce));
	215
	216	xen_mcelog.next++;
	217	}
	218
	219	static int convert_log(struct mc_info *mi)
	220	{
	221	struct mcinfo_common *mic;
	222	struct mcinfo_global *mc_global;
	223	struct mcinfo_bank *mc_bank;
	224	struct xen_mce m;
	225	uint32_t i;
	226
	227	mic = NULL;
	228	x86_mcinfo_lookup(&mic, mi, MC_TYPE_GLOBAL);
	229	if (unlikely(!mic)) {
283c0972	230	pr_warn("Failed to find global error info\n");
cef12ee5 LJ	231	return -ENODEV;
	232	}
	233
	234	memset(&m, 0, sizeof(struct xen_mce));
	235
	236	mc_global = (struct mcinfo_global *)mic;
	237	m.mcgstatus = mc_global->mc_gstatus;
	238	m.apicid = mc_global->mc_apicid;
	239
	240	for (i = 0; i < ncpus; i++)
	241	if (g_physinfo[i].mc_apicid == m.apicid)
	242	break;
	243	if (unlikely(i == ncpus)) {
283c0972	244	pr_warn("Failed to match cpu with apicid %d\n", m.apicid);
cef12ee5 LJ	245	return -ENODEV;
	246	}
	247
	248	m.socketid = g_physinfo[i].mc_chipid;
	249	m.cpu = m.extcpu = g_physinfo[i].mc_cpunr;
	250	m.cpuvendor = (__u8)g_physinfo[i].mc_vendor;
	251	m.mcgcap = g_physinfo[i].mc_msrvalues[__MC_MSR_MCGCAP].value;
	252
	253	mic = NULL;
	254	x86_mcinfo_lookup(&mic, mi, MC_TYPE_BANK);
	255	if (unlikely(!mic)) {
283c0972	256	pr_warn("Fail to find bank error info\n");
cef12ee5 LJ	257	return -ENODEV;
	258	}
	259
	260	do {
	261	if ((!mic) \|\| (mic->size == 0) \|\|
	262	(mic->type != MC_TYPE_GLOBAL &&
	263	mic->type != MC_TYPE_BANK &&
	264	mic->type != MC_TYPE_EXTENDED &&
	265	mic->type != MC_TYPE_RECOVERY))
	266	break;
	267
	268	if (mic->type == MC_TYPE_BANK) {
	269	mc_bank = (struct mcinfo_bank *)mic;
	270	m.misc = mc_bank->mc_misc;
	271	m.status = mc_bank->mc_status;
	272	m.addr = mc_bank->mc_addr;
	273	m.tsc = mc_bank->mc_tsc;
	274	m.bank = mc_bank->mc_bank;
	275	m.finished = 1;
	276	/log this record/
	277	xen_mce_log(&m);
	278	}
	279	mic = x86_mcinfo_next(mic);
	280	} while (1);
	281
	282	return 0;
	283	}
	284
	285	static int mc_queue_handle(uint32_t flags)
	286	{
	287	struct xen_mc mc_op;
	288	int ret = 0;
	289
	290	mc_op.cmd = XEN_MC_fetch;
	291	mc_op.interface_version = XEN_MCA_INTERFACE_VERSION;
	292	set_xen_guest_handle(mc_op.u.mc_fetch.data, &g_mi);
	293	do {
	294	mc_op.u.mc_fetch.flags = flags;
	295	ret = HYPERVISOR_mca(&mc_op);
	296	if (ret) {
283c0972 JP	297	pr_err("Failed to fetch %surgent error log\n",
283c0972 JP	298	flags == XEN_MC_URGENT ? "" : "non");
cef12ee5 LJ	299	break;
	300	}
	301
	302	if (mc_op.u.mc_fetch.flags & XEN_MC_NODATA \|\|
	303	mc_op.u.mc_fetch.flags & XEN_MC_FETCHFAILED)
	304	break;
	305	else {
	306	ret = convert_log(&g_mi);
	307	if (ret)
283c0972	308	pr_warn("Failed to convert this error log, continue acking it anyway\n");
cef12ee5 LJ	309
	310	mc_op.u.mc_fetch.flags = flags \| XEN_MC_ACK;
	311	ret = HYPERVISOR_mca(&mc_op);
	312	if (ret) {
283c0972	313	pr_err("Failed to ack previous error log\n");
cef12ee5 LJ	314	break;
	315	}
	316	}
	317	} while (1);
	318
	319	return ret;
	320	}
	321
	322	/* virq handler for machine check error info*/
1b2a0551	323	static void xen_mce_work_fn(struct work_struct *work)
cef12ee5 LJ	324	{
cef12ee5 LJ	325	int err;
cef12ee5	326
1b2a0551	327	mutex_lock(&mcelog_lock);
cef12ee5 LJ	328
	329	/* urgent mc_info */
	330	err = mc_queue_handle(XEN_MC_URGENT);
	331	if (err)
283c0972	332	pr_err("Failed to handle urgent mc_info queue, continue handling nonurgent mc_info queue anyway\n");
cef12ee5 LJ	333
	334	/* nonurgent mc_info */
	335	err = mc_queue_handle(XEN_MC_NONURGENT);
	336	if (err)
283c0972	337	pr_err("Failed to handle nonurgent mc_info queue\n");
cef12ee5	338
a867e5d6 LJ	339	/* wake processes polling /dev/mcelog */
	340	wake_up_interruptible(&xen_mce_chrdev_wait);
	341
1b2a0551 LJ	342	mutex_unlock(&mcelog_lock);
	343	}
	344	static DECLARE_WORK(xen_mce_work, xen_mce_work_fn);
cef12ee5	345
1b2a0551 LJ	346	static irqreturn_t xen_mce_interrupt(int irq, void *dev_id)
	347	{
	348	schedule_work(&xen_mce_work);
cef12ee5 LJ	349	return IRQ_HANDLED;
	350	}
	351
	352	static int bind_virq_for_mce(void)
	353	{
	354	int ret;
	355	struct xen_mc mc_op;
	356
	357	memset(&mc_op, 0, sizeof(struct xen_mc));
	358
	359	/* Fetch physical CPU Numbers */
	360	mc_op.cmd = XEN_MC_physcpuinfo;
	361	mc_op.interface_version = XEN_MCA_INTERFACE_VERSION;
	362	set_xen_guest_handle(mc_op.u.mc_physcpuinfo.info, g_physinfo);
	363	ret = HYPERVISOR_mca(&mc_op);
	364	if (ret) {
283c0972	365	pr_err("Failed to get CPU numbers\n");
cef12ee5 LJ	366	return ret;
	367	}
	368
	369	/* Fetch each CPU Physical Info for later reference*/
	370	ncpus = mc_op.u.mc_physcpuinfo.ncpus;
	371	g_physinfo = kcalloc(ncpus, sizeof(struct mcinfo_logical_cpu),
	372	GFP_KERNEL);
	373	if (!g_physinfo)
	374	return -ENOMEM;
	375	set_xen_guest_handle(mc_op.u.mc_physcpuinfo.info, g_physinfo);
	376	ret = HYPERVISOR_mca(&mc_op);
	377	if (ret) {
283c0972	378	pr_err("Failed to get CPU info\n");
cef12ee5 LJ	379	kfree(g_physinfo);
	380	return ret;
	381	}
	382
	383	ret = bind_virq_to_irqhandler(VIRQ_MCA, 0,
	384	xen_mce_interrupt, 0, "mce", NULL);
	385	if (ret < 0) {
283c0972	386	pr_err("Failed to bind virq\n");
cef12ee5 LJ	387	kfree(g_physinfo);
	388	return ret;
	389	}
	390
	391	return 0;
	392	}
	393
	394	static int __init xen_late_init_mcelog(void)
	395	{
	396	/* Only DOM0 is responsible for MCE logging */
	397	if (xen_initial_domain()) {
	398	/* register character device /dev/mcelog for xen mcelog */
	399	if (misc_register(&xen_mce_chrdev_device))
	400	return -ENODEV;
	401	return bind_virq_for_mce();
	402	}
	403
	404	return -ENODEV;
	405	}
	406	device_initcall(xen_late_init_mcelog);