[linux-2.6-block.git] / drivers / xen / mcelog.c

/******************************************************************************
 * mcelog.c
 * Driver for receiving and transferring machine check error infomation
 *
 * Copyright (c) 2012 Intel Corporation
 * Author: Liu, Jinsong <jinsong.liu@intel.com>
 * Author: Jiang, Yunhong <yunhong.jiang@intel.com>
 * Author: Ke, Liping <liping.ke@intel.com>
 *
 * 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.
 */

#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 <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>

#define XEN_MCELOG "xen_mcelog: "

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

static DEFINE_SPINLOCK(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 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;

	spin_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:
	spin_unlock(&mcelog_lock);

	return err ? err : buf - ubuf;
}

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,
	.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_warning(XEN_MCELOG "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_warning(XEN_MCELOG "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_warning(XEN_MCELOG "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(XEN_MCELOG "Failed to fetch %s error log\n",
			       (flags == XEN_MC_URGENT) ?
			       "urgnet" : "nonurgent");
			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_warning(XEN_MCELOG
					   "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(XEN_MCELOG
				       "Failed to ack previous error log\n");
				break;
			}
		}
	} while (1);

	return ret;
}

/* virq handler for machine check error info*/
static irqreturn_t xen_mce_interrupt(int irq, void *dev_id)
{
	int err;
	unsigned long tmp;

	spin_lock_irqsave(&mcelog_lock, tmp);

	/* urgent mc_info */
	err = mc_queue_handle(XEN_MC_URGENT);
	if (err)
		pr_err(XEN_MCELOG
		       "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(XEN_MCELOG
		       "Failed to handle nonurgent mc_info queue.\n");

	spin_unlock_irqrestore(&mcelog_lock, tmp);

	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(XEN_MCELOG "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(XEN_MCELOG "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(XEN_MCELOG "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 <jinsong.liu@intel.com>
	7	* Author: Jiang, Yunhong <yunhong.jiang@intel.com>
	8	* Author: Ke, Liping <liping.ke@intel.com>
	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
	35	#include <linux/init.h>
	36	#include <linux/types.h>
	37	#include <linux/kernel.h>
	38	#include <linux/slab.h>
	39	#include <linux/fs.h>
	40	#include <linux/device.h>
	41	#include <linux/miscdevice.h>
	42	#include <linux/uaccess.h>
	43	#include <linux/capability.h>
	44
	45	#include <xen/interface/xen.h>
	46	#include <xen/events.h>
	47	#include <xen/interface/vcpu.h>
	48	#include <xen/xen.h>
	49	#include <asm/xen/hypercall.h>
	50	#include <asm/xen/hypervisor.h>
	51
	52	#define XEN_MCELOG "xen_mcelog: "
	53
	54	static struct mc_info g_mi;
	55	static struct mcinfo_logical_cpu *g_physinfo;
	56	static uint32_t ncpus;
	57
	58	static DEFINE_SPINLOCK(mcelog_lock);
	59
	60	static struct xen_mce_log xen_mcelog = {
	61	.signature = XEN_MCE_LOG_SIGNATURE,
	62	.len = XEN_MCE_LOG_LEN,
	63	.recordlen = sizeof(struct xen_mce),
	64	};
65
66	static DEFINE_SPINLOCK(xen_mce_chrdev_state_lock);
67	static int xen_mce_chrdev_open_count; /* #times opened */
68	static int xen_mce_chrdev_open_exclu; /* already open exclusive? */
69
70	static int xen_mce_chrdev_open(struct inode inode, struct file file)
71	{
72	spin_lock(&xen_mce_chrdev_state_lock);
73
74	if (xen_mce_chrdev_open_exclu \|\|
75	(xen_mce_chrdev_open_count && (file->f_flags & O_EXCL))) {
76	spin_unlock(&xen_mce_chrdev_state_lock);
77
78	return -EBUSY;
79	}
80
81	if (file->f_flags & O_EXCL)
82	xen_mce_chrdev_open_exclu = 1;
83	xen_mce_chrdev_open_count++;
84
85	spin_unlock(&xen_mce_chrdev_state_lock);
86
87	return nonseekable_open(inode, file);
88	}
89
90	static int xen_mce_chrdev_release(struct inode inode, struct file file)
91	{
92	spin_lock(&xen_mce_chrdev_state_lock);
93
94	xen_mce_chrdev_open_count--;
95	xen_mce_chrdev_open_exclu = 0;
96
97	spin_unlock(&xen_mce_chrdev_state_lock);
98
99	return 0;
100	}
101
102	static ssize_t xen_mce_chrdev_read(struct file filp, char __user ubuf,
103	size_t usize, loff_t *off)
104	{
105	char __user *buf = ubuf;
106	unsigned num;
107	int i, err;
108
109	spin_lock(&mcelog_lock);
110
111	num = xen_mcelog.next;
112
113	/* Only supports full reads right now */
114	err = -EINVAL;
115	if (off != 0 \|\| usize < XEN_MCE_LOG_LENsizeof(struct xen_mce))
116	goto out;
117
118	err = 0;
119	for (i = 0; i < num; i++) {
120	struct xen_mce *m = &xen_mcelog.entry[i];
121
122	err \|= copy_to_user(buf, m, sizeof(*m));
123	buf += sizeof(*m);
124	}
125
126	memset(xen_mcelog.entry, 0, num * sizeof(struct xen_mce));
127	xen_mcelog.next = 0;
128
129	if (err)
130	err = -EFAULT;
131
132	out:
133	spin_unlock(&mcelog_lock);
134
135	return err ? err : buf - ubuf;
136	}
137
138	static long xen_mce_chrdev_ioctl(struct file *f, unsigned int cmd,
139	unsigned long arg)
140	{
141	int __user p = (int __user )arg;
142
143	if (!capable(CAP_SYS_ADMIN))
144	return -EPERM;
145
146	switch (cmd) {
147	case MCE_GET_RECORD_LEN:
148	return put_user(sizeof(struct xen_mce), p);
149	case MCE_GET_LOG_LEN:
150	return put_user(XEN_MCE_LOG_LEN, p);
151	case MCE_GETCLEAR_FLAGS: {
152	unsigned flags;
153
154	do {
155	flags = xen_mcelog.flags;
156	} while (cmpxchg(&xen_mcelog.flags, flags, 0) != flags);
157
158	return put_user(flags, p);
159	}
160	default:
161	return -ENOTTY;
162	}
163	}
164
165	static const struct file_operations xen_mce_chrdev_ops = {
166	.open = xen_mce_chrdev_open,
167	.release = xen_mce_chrdev_release,
168	.read = xen_mce_chrdev_read,
169	.unlocked_ioctl = xen_mce_chrdev_ioctl,
170	.llseek = no_llseek,
171	};
172
173	static struct miscdevice xen_mce_chrdev_device = {
174	MISC_MCELOG_MINOR,
175	"mcelog",
176	&xen_mce_chrdev_ops,
177	};
178
179	/*
180	* Caller should hold the mcelog_lock
181	*/
182	static void xen_mce_log(struct xen_mce *mce)
183	{
184	unsigned entry;
185
186	entry = xen_mcelog.next;
187
188	/*
189	* When the buffer fills up discard new entries.
190	* Assume that the earlier errors are the more
191	* interesting ones:
192	*/
193	if (entry >= XEN_MCE_LOG_LEN) {
194	set_bit(XEN_MCE_OVERFLOW,
195	(unsigned long *)&xen_mcelog.flags);
196	return;
197	}
198
199	memcpy(xen_mcelog.entry + entry, mce, sizeof(struct xen_mce));
200
201	xen_mcelog.next++;
202	}
203
204	static int convert_log(struct mc_info *mi)
205	{
206	struct mcinfo_common *mic;
207	struct mcinfo_global *mc_global;
208	struct mcinfo_bank *mc_bank;
209	struct xen_mce m;
210	uint32_t i;
211
212	mic = NULL;
213	x86_mcinfo_lookup(&mic, mi, MC_TYPE_GLOBAL);
214	if (unlikely(!mic)) {
215	pr_warning(XEN_MCELOG "Failed to find global error info\n");
216	return -ENODEV;
217	}
218
219	memset(&m, 0, sizeof(struct xen_mce));
220
221	mc_global = (struct mcinfo_global *)mic;
222	m.mcgstatus = mc_global->mc_gstatus;
223	m.apicid = mc_global->mc_apicid;
224
225	for (i = 0; i < ncpus; i++)
226	if (g_physinfo[i].mc_apicid == m.apicid)
227	break;
228	if (unlikely(i == ncpus)) {
229	pr_warning(XEN_MCELOG "Failed to match cpu with apicid %d\n",
230	m.apicid);
231	return -ENODEV;
232	}
233
234	m.socketid = g_physinfo[i].mc_chipid;
235	m.cpu = m.extcpu = g_physinfo[i].mc_cpunr;
236	m.cpuvendor = (__u8)g_physinfo[i].mc_vendor;
237	m.mcgcap = g_physinfo[i].mc_msrvalues[__MC_MSR_MCGCAP].value;
238
239	mic = NULL;
240	x86_mcinfo_lookup(&mic, mi, MC_TYPE_BANK);
241	if (unlikely(!mic)) {
242	pr_warning(XEN_MCELOG "Fail to find bank error info\n");
243	return -ENODEV;
244	}
245
246	do {
247	if ((!mic) \|\| (mic->size == 0) \|\|
248	(mic->type != MC_TYPE_GLOBAL &&
249	mic->type != MC_TYPE_BANK &&
250	mic->type != MC_TYPE_EXTENDED &&
251	mic->type != MC_TYPE_RECOVERY))
252	break;
253
254	if (mic->type == MC_TYPE_BANK) {
255	mc_bank = (struct mcinfo_bank *)mic;
256	m.misc = mc_bank->mc_misc;
257	m.status = mc_bank->mc_status;
258	m.addr = mc_bank->mc_addr;
259	m.tsc = mc_bank->mc_tsc;
260	m.bank = mc_bank->mc_bank;
261	m.finished = 1;
262	/log this record/
263	xen_mce_log(&m);
264	}
265	mic = x86_mcinfo_next(mic);
266	} while (1);
267
268	return 0;
269	}
270
271	static int mc_queue_handle(uint32_t flags)
272	{
273	struct xen_mc mc_op;
274	int ret = 0;
275
276	mc_op.cmd = XEN_MC_fetch;
277	mc_op.interface_version = XEN_MCA_INTERFACE_VERSION;
278	set_xen_guest_handle(mc_op.u.mc_fetch.data, &g_mi);
279	do {
280	mc_op.u.mc_fetch.flags = flags;
281	ret = HYPERVISOR_mca(&mc_op);
282	if (ret) {
283	pr_err(XEN_MCELOG "Failed to fetch %s error log\n",
284	(flags == XEN_MC_URGENT) ?
285	"urgnet" : "nonurgent");
286	break;
287	}
288
289	if (mc_op.u.mc_fetch.flags & XEN_MC_NODATA \|\|
290	mc_op.u.mc_fetch.flags & XEN_MC_FETCHFAILED)
291	break;
292	else {
293	ret = convert_log(&g_mi);
294	if (ret)
295	pr_warning(XEN_MCELOG
296	"Failed to convert this error log, "
297	"continue acking it anyway\n");
298
299	mc_op.u.mc_fetch.flags = flags \| XEN_MC_ACK;
300	ret = HYPERVISOR_mca(&mc_op);
301	if (ret) {
302	pr_err(XEN_MCELOG
303	"Failed to ack previous error log\n");
304	break;
305	}
306	}
307	} while (1);
308
309	return ret;
310	}
311
312	/* virq handler for machine check error info*/
313	static irqreturn_t xen_mce_interrupt(int irq, void *dev_id)
314	{
315	int err;
316	unsigned long tmp;
317
318	spin_lock_irqsave(&mcelog_lock, tmp);
319
320	/* urgent mc_info */
321	err = mc_queue_handle(XEN_MC_URGENT);
322	if (err)
323	pr_err(XEN_MCELOG
324	"Failed to handle urgent mc_info queue, "
325	"continue handling nonurgent mc_info queue anyway.\n");
326
327	/* nonurgent mc_info */
328	err = mc_queue_handle(XEN_MC_NONURGENT);
329	if (err)
330	pr_err(XEN_MCELOG
331	"Failed to handle nonurgent mc_info queue.\n");
332
333	spin_unlock_irqrestore(&mcelog_lock, tmp);
334
335	return IRQ_HANDLED;
336	}
337
338	static int bind_virq_for_mce(void)
339	{
340	int ret;
341	struct xen_mc mc_op;
342
343	memset(&mc_op, 0, sizeof(struct xen_mc));
344
345	/* Fetch physical CPU Numbers */
346	mc_op.cmd = XEN_MC_physcpuinfo;
347	mc_op.interface_version = XEN_MCA_INTERFACE_VERSION;
348	set_xen_guest_handle(mc_op.u.mc_physcpuinfo.info, g_physinfo);
349	ret = HYPERVISOR_mca(&mc_op);
350	if (ret) {
351	pr_err(XEN_MCELOG "Failed to get CPU numbers\n");
352	return ret;
353	}
354
355	/* Fetch each CPU Physical Info for later reference*/
356	ncpus = mc_op.u.mc_physcpuinfo.ncpus;
357	g_physinfo = kcalloc(ncpus, sizeof(struct mcinfo_logical_cpu),
358	GFP_KERNEL);
359	if (!g_physinfo)
360	return -ENOMEM;
361	set_xen_guest_handle(mc_op.u.mc_physcpuinfo.info, g_physinfo);
362	ret = HYPERVISOR_mca(&mc_op);
363	if (ret) {
364	pr_err(XEN_MCELOG "Failed to get CPU info\n");
365	kfree(g_physinfo);
366	return ret;
367	}
368
369	ret = bind_virq_to_irqhandler(VIRQ_MCA, 0,
370	xen_mce_interrupt, 0, "mce", NULL);
371	if (ret < 0) {
372	pr_err(XEN_MCELOG "Failed to bind virq\n");
373	kfree(g_physinfo);
374	return ret;
375	}
376
377	return 0;
378	}
379
380	static int __init xen_late_init_mcelog(void)
381	{
382	/* Only DOM0 is responsible for MCE logging */
383	if (xen_initial_domain()) {
384	/* register character device /dev/mcelog for xen mcelog */
385	if (misc_register(&xen_mce_chrdev_device))
386	return -ENODEV;
387	return bind_virq_for_mce();
388	}
389
390	return -ENODEV;
391	}
392	device_initcall(xen_late_init_mcelog);