[people/arne_f/kernel.git] / arch / powerpc / platforms / pseries / rtasd.c

/*
 * Copyright (C) 2001 Anton Blanchard <anton@au.ibm.com>, IBM
 *
 * This program is free software; you can redistribute it and/or
 * modify it under the terms of the GNU General Public License
 * as published by the Free Software Foundation; either version
 * 2 of the License, or (at your option) any later version.
 *
 * Communication to userspace based on kernel/printk.c
 */

#include <linux/types.h>
#include <linux/errno.h>
#include <linux/sched.h>
#include <linux/kernel.h>
#include <linux/poll.h>
#include <linux/proc_fs.h>
#include <linux/init.h>
#include <linux/vmalloc.h>
#include <linux/spinlock.h>
#include <linux/cpu.h>
#include <linux/delay.h>

#include <asm/uaccess.h>
#include <asm/io.h>
#include <asm/rtas.h>
#include <asm/prom.h>
#include <asm/nvram.h>
#include <asm/atomic.h>
#include <asm/machdep.h>

#if 0
#define DEBUG(A...)	printk(KERN_ERR A)
#else
#define DEBUG(A...)
#endif

static DEFINE_SPINLOCK(rtasd_log_lock);

DECLARE_WAIT_QUEUE_HEAD(rtas_log_wait);

static char *rtas_log_buf;
static unsigned long rtas_log_start;
static unsigned long rtas_log_size;

static int surveillance_timeout = -1;
static unsigned int rtas_error_log_max;
static unsigned int rtas_error_log_buffer_max;

/* RTAS service tokens */
static unsigned int event_scan;
static unsigned int rtas_event_scan_rate;

static int full_rtas_msgs = 0;

/* Stop logging to nvram after first fatal error */
static int no_more_logging;

static int error_log_cnt;

/*
 * Since we use 32 bit RTAS, the physical address of this must be below
 * 4G or else bad things happen. Allocate this in the kernel data and
 * make it big enough.
 */
static unsigned char logdata[RTAS_ERROR_LOG_MAX];

static char *rtas_type[] = {
	"Unknown", "Retry", "TCE Error", "Internal Device Failure",
	"Timeout", "Data Parity", "Address Parity", "Cache Parity",
	"Address Invalid", "ECC Uncorrected", "ECC Corrupted",
};

static char *rtas_event_type(int type)
{
	if ((type > 0) && (type < 11))
		return rtas_type[type];

	switch (type) {
		case RTAS_TYPE_EPOW:
			return "EPOW";
		case RTAS_TYPE_PLATFORM:
			return "Platform Error";
		case RTAS_TYPE_IO:
			return "I/O Event";
		case RTAS_TYPE_INFO:
			return "Platform Information Event";
		case RTAS_TYPE_DEALLOC:
			return "Resource Deallocation Event";
		case RTAS_TYPE_DUMP:
			return "Dump Notification Event";
	}

	return rtas_type[0];
}

/* To see this info, grep RTAS /var/log/messages and each entry
 * will be collected together with obvious begin/end.
 * There will be a unique identifier on the begin and end lines.
 * This will persist across reboots.
 *
 * format of error logs returned from RTAS:
 * bytes	(size)	: contents
 * --------------------------------------------------------
 * 0-7		(8)	: rtas_error_log
 * 8-47		(40)	: extended info
 * 48-51	(4)	: vendor id
 * 52-1023 (vendor specific) : location code and debug data
 */
static void printk_log_rtas(char *buf, int len)
{

	int i,j,n = 0;
	int perline = 16;
	char buffer[64];
	char * str = "RTAS event";

	if (full_rtas_msgs) {
		printk(RTAS_DEBUG "%d -------- %s begin --------\n",
		       error_log_cnt, str);

		/*
		 * Print perline bytes on each line, each line will start
		 * with RTAS and a changing number, so syslogd will
		 * print lines that are otherwise the same.  Separate every
		 * 4 bytes with a space.
		 */
		for (i = 0; i < len; i++) {
			j = i % perline;
			if (j == 0) {
				memset(buffer, 0, sizeof(buffer));
				n = sprintf(buffer, "RTAS %d:", i/perline);
			}

			if ((i % 4) == 0)
				n += sprintf(buffer+n, " ");

			n += sprintf(buffer+n, "%02x", (unsigned char)buf[i]);

			if (j == (perline-1))
				printk(KERN_DEBUG "%s\n", buffer);
		}
		if ((i % perline) != 0)
			printk(KERN_DEBUG "%s\n", buffer);

		printk(RTAS_DEBUG "%d -------- %s end ----------\n",
		       error_log_cnt, str);
	} else {
		struct rtas_error_log *errlog = (struct rtas_error_log *)buf;

		printk(RTAS_DEBUG "event: %d, Type: %s, Severity: %d\n",
		       error_log_cnt, rtas_event_type(errlog->type),
		       errlog->severity);
	}
}

static int log_rtas_len(char * buf)
{
	int len;
	struct rtas_error_log *err;

	/* rtas fixed header */
	len = 8;
	err = (struct rtas_error_log *)buf;
	if (err->extended_log_length) {

		/* extended header */
		len += err->extended_log_length;
	}

	if (rtas_error_log_max == 0)
		rtas_error_log_max = rtas_get_error_log_max();

	if (len > rtas_error_log_max)
		len = rtas_error_log_max;

	return len;
}

/*
 * First write to nvram, if fatal error, that is the only
 * place we log the info.  The error will be picked up
 * on the next reboot by rtasd.  If not fatal, run the
 * method for the type of error.  Currently, only RTAS
 * errors have methods implemented, but in the future
 * there might be a need to store data in nvram before a
 * call to panic().
 *
 * XXX We write to nvram periodically, to indicate error has
 * been written and sync'd, but there is a possibility
 * that if we don't shutdown correctly, a duplicate error
 * record will be created on next reboot.
 */
void pSeries_log_error(char *buf, unsigned int err_type, int fatal)
{
	unsigned long offset;
	unsigned long s;
	int len = 0;

	DEBUG("logging event\n");
	if (buf == NULL)
		return;

	spin_lock_irqsave(&rtasd_log_lock, s);

	/* get length and increase count */
	switch (err_type & ERR_TYPE_MASK) {
	case ERR_TYPE_RTAS_LOG:
		len = log_rtas_len(buf);
		if (!(err_type & ERR_FLAG_BOOT))
			error_log_cnt++;
		break;
	case ERR_TYPE_KERNEL_PANIC:
	default:
		spin_unlock_irqrestore(&rtasd_log_lock, s);
		return;
	}

	/* Write error to NVRAM */
	if (!no_more_logging && !(err_type & ERR_FLAG_BOOT))
		nvram_write_error_log(buf, len, err_type, error_log_cnt);

	/*
	 * rtas errors can occur during boot, and we do want to capture
	 * those somewhere, even if nvram isn't ready (why not?), and even
	 * if rtasd isn't ready. Put them into the boot log, at least.
	 */
	if ((err_type & ERR_TYPE_MASK) == ERR_TYPE_RTAS_LOG)
		printk_log_rtas(buf, len);

	/* Check to see if we need to or have stopped logging */
	if (fatal || no_more_logging) {
		no_more_logging = 1;
		spin_unlock_irqrestore(&rtasd_log_lock, s);
		return;
	}

	/* call type specific method for error */
	switch (err_type & ERR_TYPE_MASK) {
	case ERR_TYPE_RTAS_LOG:
		offset = rtas_error_log_buffer_max *
			((rtas_log_start+rtas_log_size) & LOG_NUMBER_MASK);

		/* First copy over sequence number */
		memcpy(&rtas_log_buf[offset], (void *) &error_log_cnt, sizeof(int));

		/* Second copy over error log data */
		offset += sizeof(int);
		memcpy(&rtas_log_buf[offset], buf, len);

		if (rtas_log_size < LOG_NUMBER)
			rtas_log_size += 1;
		else
			rtas_log_start += 1;

		spin_unlock_irqrestore(&rtasd_log_lock, s);
		wake_up_interruptible(&rtas_log_wait);
		break;
	case ERR_TYPE_KERNEL_PANIC:
	default:
		spin_unlock_irqrestore(&rtasd_log_lock, s);
		return;
	}

}


static int rtas_log_open(struct inode * inode, struct file * file)
{
	return 0;
}

static int rtas_log_release(struct inode * inode, struct file * file)
{
	return 0;
}

/* This will check if all events are logged, if they are then, we
 * know that we can safely clear the events in NVRAM.
 * Next we'll sit and wait for something else to log.
 */
static ssize_t rtas_log_read(struct file * file, char __user * buf,
			 size_t count, loff_t *ppos)
{
	int error;
	char *tmp;
	unsigned long s;
	unsigned long offset;

	if (!buf || count < rtas_error_log_buffer_max)
		return -EINVAL;

	count = rtas_error_log_buffer_max;

	if (!access_ok(VERIFY_WRITE, buf, count))
		return -EFAULT;

	tmp = kmalloc(count, GFP_KERNEL);
	if (!tmp)
		return -ENOMEM;


	spin_lock_irqsave(&rtasd_log_lock, s);
	/* if it's 0, then we know we got the last one (the one in NVRAM) */
	if (rtas_log_size == 0 && !no_more_logging)
		nvram_clear_error_log();
	spin_unlock_irqrestore(&rtasd_log_lock, s);


	error = wait_event_interruptible(rtas_log_wait, rtas_log_size);
	if (error)
		goto out;

	spin_lock_irqsave(&rtasd_log_lock, s);
	offset = rtas_error_log_buffer_max * (rtas_log_start & LOG_NUMBER_MASK);
	memcpy(tmp, &rtas_log_buf[offset], count);

	rtas_log_start += 1;
	rtas_log_size -= 1;
	spin_unlock_irqrestore(&rtasd_log_lock, s);

	error = copy_to_user(buf, tmp, count) ? -EFAULT : count;
out:
	kfree(tmp);
	return error;
}

static unsigned int rtas_log_poll(struct file *file, poll_table * wait)
{
	poll_wait(file, &rtas_log_wait, wait);
	if (rtas_log_size)
		return POLLIN | POLLRDNORM;
	return 0;
}

const struct file_operations proc_rtas_log_operations = {
	.read =		rtas_log_read,
	.poll =		rtas_log_poll,
	.open =		rtas_log_open,
	.release =	rtas_log_release,
};

static int enable_surveillance(int timeout)
{
	int error;

	error = rtas_set_indicator(SURVEILLANCE_TOKEN, 0, timeout);

	if (error == 0)
		return 0;

	if (error == -EINVAL) {
		printk(KERN_DEBUG "rtasd: surveillance not supported\n");
		return 0;
	}

	printk(KERN_ERR "rtasd: could not update surveillance\n");
	return -1;
}

static void do_event_scan(void)
{
	int error;
	do {
		memset(logdata, 0, rtas_error_log_max);
		error = rtas_call(event_scan, 4, 1, NULL,
				  RTAS_EVENT_SCAN_ALL_EVENTS, 0,
				  __pa(logdata), rtas_error_log_max);
		if (error == -1) {
			printk(KERN_ERR "event-scan failed\n");
			break;
		}

		if (error == 0)
			pSeries_log_error(logdata, ERR_TYPE_RTAS_LOG, 0);

	} while(error == 0);
}

static void do_event_scan_all_cpus(long delay)
{
	int cpu;

	lock_cpu_hotplug();
	cpu = first_cpu(cpu_online_map);
	for (;;) {
		set_cpus_allowed(current, cpumask_of_cpu(cpu));
		do_event_scan();
		set_cpus_allowed(current, CPU_MASK_ALL);

		/* Drop hotplug lock, and sleep for the specified delay */
		unlock_cpu_hotplug();
		msleep_interruptible(delay);
		lock_cpu_hotplug();

		cpu = next_cpu(cpu, cpu_online_map);
		if (cpu == NR_CPUS)
			break;
	}
	unlock_cpu_hotplug();
}

static int rtasd(void *unused)
{
	unsigned int err_type;
	int rc;

	daemonize("rtasd");

	printk(KERN_DEBUG "RTAS daemon started\n");
	DEBUG("will sleep for %d milliseconds\n", (30000/rtas_event_scan_rate));

	/* See if we have any error stored in NVRAM */
	memset(logdata, 0, rtas_error_log_max);
	rc = nvram_read_error_log(logdata, rtas_error_log_max,
	                          &err_type, &error_log_cnt);

	if (!rc) {
		if (err_type != ERR_FLAG_ALREADY_LOGGED) {
			pSeries_log_error(logdata, err_type | ERR_FLAG_BOOT, 0);
		}
	}

	/* First pass. */
	do_event_scan_all_cpus(1000);

	if (surveillance_timeout != -1) {
		DEBUG("enabling surveillance\n");
		enable_surveillance(surveillance_timeout);
		DEBUG("surveillance enabled\n");
	}

	/* Delay should be at least one second since some
	 * machines have problems if we call event-scan too
	 * quickly. */
	for (;;)
		do_event_scan_all_cpus(30000/rtas_event_scan_rate);

	return -EINVAL;
}

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

	if (!machine_is(pseries))
		return 0;

	/* No RTAS */
	event_scan = rtas_token("event-scan");
	if (event_scan == RTAS_UNKNOWN_SERVICE) {
		printk(KERN_DEBUG "rtasd: no event-scan on system\n");
		return -ENODEV;
	}

	rtas_event_scan_rate = rtas_token("rtas-event-scan-rate");
	if (rtas_event_scan_rate == RTAS_UNKNOWN_SERVICE) {
		printk(KERN_ERR "rtasd: no rtas-event-scan-rate on system\n");
		return -ENODEV;
	}

	/* Make room for the sequence number */
	rtas_error_log_max = rtas_get_error_log_max();
	rtas_error_log_buffer_max = rtas_error_log_max + sizeof(int);

	rtas_log_buf = vmalloc(rtas_error_log_buffer_max*LOG_NUMBER);
	if (!rtas_log_buf) {
		printk(KERN_ERR "rtasd: no memory\n");
		return -ENOMEM;
	}

	entry = create_proc_entry("ppc64/rtas/error_log", S_IRUSR, NULL);
	if (entry)
		entry->proc_fops = &proc_rtas_log_operations;
	else
		printk(KERN_ERR "Failed to create error_log proc entry\n");

	if (kernel_thread(rtasd, NULL, CLONE_FS) < 0)
		printk(KERN_ERR "Failed to start RTAS daemon\n");

	return 0;
}

static int __init surveillance_setup(char *str)
{
	int i;

	if (get_option(&str,&i)) {
		if (i >= 0 && i <= 255)
			surveillance_timeout = i;
	}

	return 1;
}

static int __init rtasmsgs_setup(char *str)
{
	if (strcmp(str, "on") == 0)
		full_rtas_msgs = 1;
	else if (strcmp(str, "off") == 0)
		full_rtas_msgs = 0;

	return 1;
}
__initcall(rtas_init);
__setup("surveillance=", surveillance_setup);
__setup("rtasmsgs=", rtasmsgs_setup);
Commit	Line	Data
1da177e4 LT	1	/*
	2	* Copyright (C) 2001 Anton Blanchard <anton@au.ibm.com>, IBM
	3	*
	4	* This program is free software; you can redistribute it and/or
	5	* modify it under the terms of the GNU General Public License
	6	* as published by the Free Software Foundation; either version
	7	* 2 of the License, or (at your option) any later version.
	8	*
	9	* Communication to userspace based on kernel/printk.c
	10	*/
	11
	12	#include <linux/types.h>
	13	#include <linux/errno.h>
	14	#include <linux/sched.h>
	15	#include <linux/kernel.h>
	16	#include <linux/poll.h>
	17	#include <linux/proc_fs.h>
	18	#include <linux/init.h>
	19	#include <linux/vmalloc.h>
	20	#include <linux/spinlock.h>
	21	#include <linux/cpu.h>
0287ebed	22	#include <linux/delay.h>
1da177e4 LT	23
	24	#include <asm/uaccess.h>
	25	#include <asm/io.h>
	26	#include <asm/rtas.h>
	27	#include <asm/prom.h>
	28	#include <asm/nvram.h>
	29	#include <asm/atomic.h>
e8222502	30	#include <asm/machdep.h>
1da177e4 LT	31
	32	#if 0
	33	#define DEBUG(A...) printk(KERN_ERR A)
	34	#else
	35	#define DEBUG(A...)
	36	#endif
	37
	38	static DEFINE_SPINLOCK(rtasd_log_lock);
	39
	40	DECLARE_WAIT_QUEUE_HEAD(rtas_log_wait);
	41
	42	static char *rtas_log_buf;
	43	static unsigned long rtas_log_start;
	44	static unsigned long rtas_log_size;
	45
	46	static int surveillance_timeout = -1;
1da177e4 LT	47	static unsigned int rtas_error_log_max;
	48	static unsigned int rtas_error_log_buffer_max;
	49
a4fc3a3c LV	50	/* RTAS service tokens */
	51	static unsigned int event_scan;
	52	static unsigned int rtas_event_scan_rate;
	53
1da177e4 LT	54	static int full_rtas_msgs = 0;
1da177e4 LT	55
79c0108d LV	56	/* Stop logging to nvram after first fatal error */
79c0108d LV	57	static int no_more_logging;
1da177e4	58
0f2342c8	59	static int error_log_cnt;
1da177e4 LT	60
	61	/*
	62	* Since we use 32 bit RTAS, the physical address of this must be below
	63	* 4G or else bad things happen. Allocate this in the kernel data and
	64	* make it big enough.
	65	*/
	66	static unsigned char logdata[RTAS_ERROR_LOG_MAX];
	67
1da177e4 LT	68	static char *rtas_type[] = {
	69	"Unknown", "Retry", "TCE Error", "Internal Device Failure",
	70	"Timeout", "Data Parity", "Address Parity", "Cache Parity",
	71	"Address Invalid", "ECC Uncorrected", "ECC Corrupted",
	72	};
	73
	74	static char *rtas_event_type(int type)
	75	{
	76	if ((type > 0) && (type < 11))
	77	return rtas_type[type];
	78
	79	switch (type) {
	80	case RTAS_TYPE_EPOW:
	81	return "EPOW";
	82	case RTAS_TYPE_PLATFORM:
	83	return "Platform Error";
	84	case RTAS_TYPE_IO:
	85	return "I/O Event";
	86	case RTAS_TYPE_INFO:
	87	return "Platform Information Event";
	88	case RTAS_TYPE_DEALLOC:
	89	return "Resource Deallocation Event";
	90	case RTAS_TYPE_DUMP:
	91	return "Dump Notification Event";
	92	}
	93
	94	return rtas_type[0];
	95	}
	96
	97	/* To see this info, grep RTAS /var/log/messages and each entry
	98	* will be collected together with obvious begin/end.
	99	* There will be a unique identifier on the begin and end lines.
	100	* This will persist across reboots.
	101	*
	102	* format of error logs returned from RTAS:
	103	* bytes (size) : contents
	104	* --------------------------------------------------------
	105	* 0-7 (8) : rtas_error_log
	106	* 8-47 (40) : extended info
	107	* 48-51 (4) : vendor id
	108	* 52-1023 (vendor specific) : location code and debug data
	109	*/
	110	static void printk_log_rtas(char *buf, int len)
	111	{
	112
	113	int i,j,n = 0;
	114	int perline = 16;
	115	char buffer[64];
	116	char * str = "RTAS event";
	117
	118	if (full_rtas_msgs) {
	119	printk(RTAS_DEBUG "%d -------- %s begin --------\n",
	120	error_log_cnt, str);
	121
	122	/*
	123	* Print perline bytes on each line, each line will start
	124	* with RTAS and a changing number, so syslogd will
	125	* print lines that are otherwise the same. Separate every
	126	* 4 bytes with a space.
	127	*/
	128	for (i = 0; i < len; i++) {
	129	j = i % perline;
	130	if (j == 0) {
	131	memset(buffer, 0, sizeof(buffer));
132	n = sprintf(buffer, "RTAS %d:", i/perline);
133	}
134
135	if ((i % 4) == 0)
136	n += sprintf(buffer+n, " ");
137
138	n += sprintf(buffer+n, "%02x", (unsigned char)buf[i]);
139
140	if (j == (perline-1))
141	printk(KERN_DEBUG "%s\n", buffer);
142	}
143	if ((i % perline) != 0)
144	printk(KERN_DEBUG "%s\n", buffer);
145
146	printk(RTAS_DEBUG "%d -------- %s end ----------\n",
147	error_log_cnt, str);
148	} else {
149	struct rtas_error_log errlog = (struct rtas_error_log )buf;
150
151	printk(RTAS_DEBUG "event: %d, Type: %s, Severity: %d\n",
152	error_log_cnt, rtas_event_type(errlog->type),
153	errlog->severity);
154	}
155	}
156
157	static int log_rtas_len(char * buf)
158	{
159	int len;
160	struct rtas_error_log *err;
161
162	/* rtas fixed header */
163	len = 8;
164	err = (struct rtas_error_log *)buf;
165	if (err->extended_log_length) {
166
167	/* extended header */
168	len += err->extended_log_length;
169	}
170
4511dad4 LV	171	if (rtas_error_log_max == 0)
	172	rtas_error_log_max = rtas_get_error_log_max();
	173
1da177e4 LT	174	if (len > rtas_error_log_max)
	175	len = rtas_error_log_max;
	176
	177	return len;
	178	}
	179
	180	/*
	181	* First write to nvram, if fatal error, that is the only
	182	* place we log the info. The error will be picked up
	183	* on the next reboot by rtasd. If not fatal, run the
	184	* method for the type of error. Currently, only RTAS
	185	* errors have methods implemented, but in the future
	186	* there might be a need to store data in nvram before a
	187	* call to panic().
	188	*
	189	* XXX We write to nvram periodically, to indicate error has
	190	* been written and sync'd, but there is a possibility
	191	* that if we don't shutdown correctly, a duplicate error
	192	* record will be created on next reboot.
	193	*/
	194	void pSeries_log_error(char *buf, unsigned int err_type, int fatal)
	195	{
	196	unsigned long offset;
	197	unsigned long s;
	198	int len = 0;
	199
	200	DEBUG("logging event\n");
	201	if (buf == NULL)
	202	return;
	203
	204	spin_lock_irqsave(&rtasd_log_lock, s);
	205
	206	/* get length and increase count */
	207	switch (err_type & ERR_TYPE_MASK) {
	208	case ERR_TYPE_RTAS_LOG:
	209	len = log_rtas_len(buf);
	210	if (!(err_type & ERR_FLAG_BOOT))
	211	error_log_cnt++;
	212	break;
	213	case ERR_TYPE_KERNEL_PANIC:
	214	default:
	215	spin_unlock_irqrestore(&rtasd_log_lock, s);
	216	return;
	217	}
	218
	219	/* Write error to NVRAM */
79c0108d	220	if (!no_more_logging && !(err_type & ERR_FLAG_BOOT))
0f2342c8	221	nvram_write_error_log(buf, len, err_type, error_log_cnt);
1da177e4 LT	222
	223	/*
	224	* rtas errors can occur during boot, and we do want to capture
	225	* those somewhere, even if nvram isn't ready (why not?), and even
	226	* if rtasd isn't ready. Put them into the boot log, at least.
	227	*/
	228	if ((err_type & ERR_TYPE_MASK) == ERR_TYPE_RTAS_LOG)
	229	printk_log_rtas(buf, len);
	230
	231	/* Check to see if we need to or have stopped logging */
79c0108d LV	232	if (fatal \|\| no_more_logging) {
79c0108d LV	233	no_more_logging = 1;
1da177e4 LT	234	spin_unlock_irqrestore(&rtasd_log_lock, s);
	235	return;
	236	}
	237
	238	/* call type specific method for error */
	239	switch (err_type & ERR_TYPE_MASK) {
	240	case ERR_TYPE_RTAS_LOG:
	241	offset = rtas_error_log_buffer_max *
	242	((rtas_log_start+rtas_log_size) & LOG_NUMBER_MASK);
	243
	244	/* First copy over sequence number */
	245	memcpy(&rtas_log_buf[offset], (void *) &error_log_cnt, sizeof(int));
	246
	247	/* Second copy over error log data */
	248	offset += sizeof(int);
	249	memcpy(&rtas_log_buf[offset], buf, len);
	250
	251	if (rtas_log_size < LOG_NUMBER)
	252	rtas_log_size += 1;
	253	else
	254	rtas_log_start += 1;
	255
	256	spin_unlock_irqrestore(&rtasd_log_lock, s);
	257	wake_up_interruptible(&rtas_log_wait);
	258	break;
	259	case ERR_TYPE_KERNEL_PANIC:
	260	default:
	261	spin_unlock_irqrestore(&rtasd_log_lock, s);
	262	return;
	263	}
	264
	265	}
	266
	267
	268	static int rtas_log_open(struct inode * inode, struct file * file)
	269	{
	270	return 0;
	271	}
	272
	273	static int rtas_log_release(struct inode * inode, struct file * file)
	274	{
	275	return 0;
	276	}
	277
	278	/* This will check if all events are logged, if they are then, we
	279	* know that we can safely clear the events in NVRAM.
	280	* Next we'll sit and wait for something else to log.
	281	*/
	282	static ssize_t rtas_log_read(struct file * file, char __user * buf,
	283	size_t count, loff_t *ppos)
	284	{
	285	int error;
	286	char *tmp;
	287	unsigned long s;
	288	unsigned long offset;
	289
	290	if (!buf \|\| count < rtas_error_log_buffer_max)
	291	return -EINVAL;
	292
	293	count = rtas_error_log_buffer_max;
	294
	295	if (!access_ok(VERIFY_WRITE, buf, count))
	296	return -EFAULT;
	297
298	tmp = kmalloc(count, GFP_KERNEL);
299	if (!tmp)
300	return -ENOMEM;
301
302
303	spin_lock_irqsave(&rtasd_log_lock, s);
304	/* if it's 0, then we know we got the last one (the one in NVRAM) */
79c0108d	305	if (rtas_log_size == 0 && !no_more_logging)
1da177e4 LT	306	nvram_clear_error_log();
	307	spin_unlock_irqrestore(&rtasd_log_lock, s);
	308
	309
	310	error = wait_event_interruptible(rtas_log_wait, rtas_log_size);
	311	if (error)
	312	goto out;
	313
	314	spin_lock_irqsave(&rtasd_log_lock, s);
	315	offset = rtas_error_log_buffer_max * (rtas_log_start & LOG_NUMBER_MASK);
	316	memcpy(tmp, &rtas_log_buf[offset], count);
	317
	318	rtas_log_start += 1;
	319	rtas_log_size -= 1;
	320	spin_unlock_irqrestore(&rtasd_log_lock, s);
	321
	322	error = copy_to_user(buf, tmp, count) ? -EFAULT : count;
	323	out:
	324	kfree(tmp);
	325	return error;
	326	}
	327
	328	static unsigned int rtas_log_poll(struct file file, poll_table wait)
	329	{
	330	poll_wait(file, &rtas_log_wait, wait);
	331	if (rtas_log_size)
	332	return POLLIN \| POLLRDNORM;
	333	return 0;
	334	}
	335
5dfe4c96	336	const struct file_operations proc_rtas_log_operations = {
1da177e4 LT	337	.read = rtas_log_read,
	338	.poll = rtas_log_poll,
	339	.open = rtas_log_open,
	340	.release = rtas_log_release,
	341	};
	342
	343	static int enable_surveillance(int timeout)
	344	{
	345	int error;
	346
	347	error = rtas_set_indicator(SURVEILLANCE_TOKEN, 0, timeout);
	348
	349	if (error == 0)
	350	return 0;
	351
	352	if (error == -EINVAL) {
90ddfebe	353	printk(KERN_DEBUG "rtasd: surveillance not supported\n");
1da177e4 LT	354	return 0;
	355	}
	356
	357	printk(KERN_ERR "rtasd: could not update surveillance\n");
	358	return -1;
	359	}
	360
a4fc3a3c	361	static void do_event_scan(void)
1da177e4 LT	362	{
	363	int error;
	364	do {
	365	memset(logdata, 0, rtas_error_log_max);
	366	error = rtas_call(event_scan, 4, 1, NULL,
	367	RTAS_EVENT_SCAN_ALL_EVENTS, 0,
	368	__pa(logdata), rtas_error_log_max);
	369	if (error == -1) {
	370	printk(KERN_ERR "event-scan failed\n");
	371	break;
	372	}
	373
	374	if (error == 0)
	375	pSeries_log_error(logdata, ERR_TYPE_RTAS_LOG, 0);
	376
	377	} while(error == 0);
	378	}
	379
	380	static void do_event_scan_all_cpus(long delay)
	381	{
	382	int cpu;
	383
	384	lock_cpu_hotplug();
	385	cpu = first_cpu(cpu_online_map);
	386	for (;;) {
	387	set_cpus_allowed(current, cpumask_of_cpu(cpu));
a4fc3a3c	388	do_event_scan();
1da177e4 LT	389	set_cpus_allowed(current, CPU_MASK_ALL);
	390
	391	/* Drop hotplug lock, and sleep for the specified delay */
	392	unlock_cpu_hotplug();
0287ebed	393	msleep_interruptible(delay);
1da177e4 LT	394	lock_cpu_hotplug();
	395
	396	cpu = next_cpu(cpu, cpu_online_map);
	397	if (cpu == NR_CPUS)
	398	break;
	399	}
	400	unlock_cpu_hotplug();
	401	}
	402
	403	static int rtasd(void *unused)
	404	{
	405	unsigned int err_type;
1da177e4 LT	406	int rc;
	407
	408	daemonize("rtasd");
	409
90ddfebe	410	printk(KERN_DEBUG "RTAS daemon started\n");
0287ebed	411	DEBUG("will sleep for %d milliseconds\n", (30000/rtas_event_scan_rate));
1da177e4 LT	412
	413	/* See if we have any error stored in NVRAM */
	414	memset(logdata, 0, rtas_error_log_max);
0f2342c8 LV	415	rc = nvram_read_error_log(logdata, rtas_error_log_max,
0f2342c8 LV	416	&err_type, &error_log_cnt);
1da177e4	417
1da177e4 LT	418	if (!rc) {
	419	if (err_type != ERR_FLAG_ALREADY_LOGGED) {
	420	pSeries_log_error(logdata, err_type \| ERR_FLAG_BOOT, 0);
	421	}
	422	}
	423
	424	/* First pass. */
0287ebed	425	do_event_scan_all_cpus(1000);
1da177e4 LT	426
	427	if (surveillance_timeout != -1) {
	428	DEBUG("enabling surveillance\n");
	429	enable_surveillance(surveillance_timeout);
	430	DEBUG("surveillance enabled\n");
	431	}
	432
	433	/* Delay should be at least one second since some
	434	* machines have problems if we call event-scan too
	435	* quickly. */
	436	for (;;)
0287ebed	437	do_event_scan_all_cpus(30000/rtas_event_scan_rate);
1da177e4	438
1da177e4 LT	439	return -EINVAL;
	440	}
	441
	442	static int __init rtas_init(void)
	443	{
	444	struct proc_dir_entry *entry;
	445
e8222502	446	if (!machine_is(pseries))
799d6046 PM	447	return 0;
	448
	449	/* No RTAS */
a4fc3a3c LV	450	event_scan = rtas_token("event-scan");
a4fc3a3c LV	451	if (event_scan == RTAS_UNKNOWN_SERVICE) {
90ddfebe	452	printk(KERN_DEBUG "rtasd: no event-scan on system\n");
49c28e4e	453	return -ENODEV;
1da177e4 LT	454	}
1da177e4 LT	455
4511dad4 LV	456	rtas_event_scan_rate = rtas_token("rtas-event-scan-rate");
	457	if (rtas_event_scan_rate == RTAS_UNKNOWN_SERVICE) {
	458	printk(KERN_ERR "rtasd: no rtas-event-scan-rate on system\n");
	459	return -ENODEV;
	460	}
	461
	462	/* Make room for the sequence number */
	463	rtas_error_log_max = rtas_get_error_log_max();
	464	rtas_error_log_buffer_max = rtas_error_log_max + sizeof(int);
	465
	466	rtas_log_buf = vmalloc(rtas_error_log_buffer_max*LOG_NUMBER);
	467	if (!rtas_log_buf) {
	468	printk(KERN_ERR "rtasd: no memory\n");
	469	return -ENOMEM;
	470	}
	471
1da177e4 LT	472	entry = create_proc_entry("ppc64/rtas/error_log", S_IRUSR, NULL);
	473	if (entry)
	474	entry->proc_fops = &proc_rtas_log_operations;
	475	else
	476	printk(KERN_ERR "Failed to create error_log proc entry\n");
	477
	478	if (kernel_thread(rtasd, NULL, CLONE_FS) < 0)
	479	printk(KERN_ERR "Failed to start RTAS daemon\n");
	480
	481	return 0;
	482	}
	483
	484	static int __init surveillance_setup(char *str)
	485	{
	486	int i;
	487
	488	if (get_option(&str,&i)) {
	489	if (i >= 0 && i <= 255)
	490	surveillance_timeout = i;
	491	}
	492
	493	return 1;
	494	}
	495
	496	static int __init rtasmsgs_setup(char *str)
	497	{
	498	if (strcmp(str, "on") == 0)
	499	full_rtas_msgs = 1;
	500	else if (strcmp(str, "off") == 0)
	501	full_rtas_msgs = 0;
	502
	503	return 1;
	504	}
	505	__initcall(rtas_init);
	506	__setup("surveillance=", surveillance_setup);
	507	__setup("rtasmsgs=", rtasmsgs_setup);