[linux.git] / arch / powerpc / platforms / pseries / ras.c

/*
 * Copyright (C) 2001 Dave Engebretsen IBM Corporation
 *
 * 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.
 *
 * This program is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 * GNU General Public License for more details.
 *
 * You should have received a copy of the GNU General Public License
 * along with this program; if not, write to the Free Software
 * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307 USA
 */

/* Change Activity:
 * 2001/09/21 : engebret : Created with minimal EPOW and HW exception support.
 * End Change Activity
 */

#include <linux/errno.h>
#include <linux/threads.h>
#include <linux/kernel_stat.h>
#include <linux/signal.h>
#include <linux/sched.h>
#include <linux/ioport.h>
#include <linux/interrupt.h>
#include <linux/timex.h>
#include <linux/init.h>
#include <linux/slab.h>
#include <linux/pci.h>
#include <linux/delay.h>
#include <linux/irq.h>
#include <linux/random.h>
#include <linux/sysrq.h>
#include <linux/bitops.h>

#include <asm/uaccess.h>
#include <asm/system.h>
#include <asm/io.h>
#include <asm/pgtable.h>
#include <asm/irq.h>
#include <asm/cache.h>
#include <asm/prom.h>
#include <asm/ptrace.h>
#include <asm/machdep.h>
#include <asm/rtas.h>
#include <asm/udbg.h>

static unsigned char ras_log_buf[RTAS_ERROR_LOG_MAX];
static DEFINE_SPINLOCK(ras_log_buf_lock);

char mce_data_buf[RTAS_ERROR_LOG_MAX]
;
/* This is true if we are using the firmware NMI handler (typically LPAR) */
extern int fwnmi_active;

static int ras_get_sensor_state_token;
static int ras_check_exception_token;

#define EPOW_SENSOR_TOKEN	9
#define EPOW_SENSOR_INDEX	0
#define RAS_VECTOR_OFFSET	0x500

static irqreturn_t ras_epow_interrupt(int irq, void *dev_id,
					struct pt_regs * regs);
static irqreturn_t ras_error_interrupt(int irq, void *dev_id,
					struct pt_regs * regs);

/* #define DEBUG */

static void request_ras_irqs(struct device_node *np, char *propname,
			irqreturn_t (*handler)(int, void *, struct pt_regs *),
			const char *name)
{
	unsigned int *ireg, len, i;
	int virq, n_intr;

	ireg = (unsigned int *)get_property(np, propname, &len);
	if (ireg == NULL)
		return;
	n_intr = prom_n_intr_cells(np);
	len /= n_intr * sizeof(*ireg);

	for (i = 0; i < len; i++) {
		virq = virt_irq_create_mapping(*ireg);
		if (virq == NO_IRQ) {
			printk(KERN_ERR "Unable to allocate interrupt "
			       "number for %s\n", np->full_name);
			return;
		}
		if (request_irq(irq_offset_up(virq), handler, 0, name, NULL)) {
			printk(KERN_ERR "Unable to request interrupt %d for "
			       "%s\n", irq_offset_up(virq), np->full_name);
			return;
		}
		ireg += n_intr;
	}
}

/*
 * Initialize handlers for the set of interrupts caused by hardware errors
 * and power system events.
 */
static int __init init_ras_IRQ(void)
{
	struct device_node *np;

	ras_get_sensor_state_token = rtas_token("get-sensor-state");
	ras_check_exception_token = rtas_token("check-exception");

	/* Internal Errors */
	np = of_find_node_by_path("/event-sources/internal-errors");
	if (np != NULL) {
		request_ras_irqs(np, "open-pic-interrupt", ras_error_interrupt,
				 "RAS_ERROR");
		request_ras_irqs(np, "interrupts", ras_error_interrupt,
				 "RAS_ERROR");
		of_node_put(np);
	}

	/* EPOW Events */
	np = of_find_node_by_path("/event-sources/epow-events");
	if (np != NULL) {
		request_ras_irqs(np, "open-pic-interrupt", ras_epow_interrupt,
				 "RAS_EPOW");
		request_ras_irqs(np, "interrupts", ras_epow_interrupt,
				 "RAS_EPOW");
		of_node_put(np);
	}

	return 1;
}
__initcall(init_ras_IRQ);

/*
 * Handle power subsystem events (EPOW).
 *
 * Presently we just log the event has occurred.  This should be fixed
 * to examine the type of power failure and take appropriate action where
 * the time horizon permits something useful to be done.
 */
static irqreturn_t
ras_epow_interrupt(int irq, void *dev_id, struct pt_regs * regs)
{
	int status = 0xdeadbeef;
	int state = 0;
	int critical;

	status = rtas_call(ras_get_sensor_state_token, 2, 2, &state,
			   EPOW_SENSOR_TOKEN, EPOW_SENSOR_INDEX);

	if (state > 3)
		critical = 1;  /* Time Critical */
	else
		critical = 0;

	spin_lock(&ras_log_buf_lock);

	status = rtas_call(ras_check_exception_token, 6, 1, NULL,
			   RAS_VECTOR_OFFSET,
			   virt_irq_to_real(irq_offset_down(irq)),
			   RTAS_EPOW_WARNING | RTAS_POWERMGM_EVENTS,
			   critical, __pa(&ras_log_buf),
				rtas_get_error_log_max());

	udbg_printf("EPOW <0x%lx 0x%x 0x%x>\n",
		    *((unsigned long *)&ras_log_buf), status, state);
	printk(KERN_WARNING "EPOW <0x%lx 0x%x 0x%x>\n",
	       *((unsigned long *)&ras_log_buf), status, state);

	/* format and print the extended information */
	log_error(ras_log_buf, ERR_TYPE_RTAS_LOG, 0);

	spin_unlock(&ras_log_buf_lock);
	return IRQ_HANDLED;
}

/*
 * Handle hardware error interrupts.
 *
 * RTAS check-exception is called to collect data on the exception.  If
 * the error is deemed recoverable, we log a warning and return.
 * For nonrecoverable errors, an error is logged and we stop all processing
 * as quickly as possible in order to prevent propagation of the failure.
 */
static irqreturn_t
ras_error_interrupt(int irq, void *dev_id, struct pt_regs * regs)
{
	struct rtas_error_log *rtas_elog;
	int status = 0xdeadbeef;
	int fatal;

	spin_lock(&ras_log_buf_lock);

	status = rtas_call(ras_check_exception_token, 6, 1, NULL,
			   RAS_VECTOR_OFFSET,
			   virt_irq_to_real(irq_offset_down(irq)),
			   RTAS_INTERNAL_ERROR, 1 /*Time Critical */,
			   __pa(&ras_log_buf),
				rtas_get_error_log_max());

	rtas_elog = (struct rtas_error_log *)ras_log_buf;

	if ((status == 0) && (rtas_elog->severity >= RTAS_SEVERITY_ERROR_SYNC))
		fatal = 1;
	else
		fatal = 0;

	/* format and print the extended information */
	log_error(ras_log_buf, ERR_TYPE_RTAS_LOG, fatal);

	if (fatal) {
		udbg_printf("Fatal HW Error <0x%lx 0x%x>\n",
			    *((unsigned long *)&ras_log_buf), status);
		printk(KERN_EMERG "Error: Fatal hardware error <0x%lx 0x%x>\n",
		       *((unsigned long *)&ras_log_buf), status);

#ifndef DEBUG
		/* Don't actually power off when debugging so we can test
		 * without actually failing while injecting errors.
		 * Error data will not be logged to syslog.
		 */
		ppc_md.power_off();
#endif
	} else {
		udbg_printf("Recoverable HW Error <0x%lx 0x%x>\n",
			    *((unsigned long *)&ras_log_buf), status);
		printk(KERN_WARNING
		       "Warning: Recoverable hardware error <0x%lx 0x%x>\n",
		       *((unsigned long *)&ras_log_buf), status);
	}

	spin_unlock(&ras_log_buf_lock);
	return IRQ_HANDLED;
}

/* Get the error information for errors coming through the
 * FWNMI vectors.  The pt_regs' r3 will be updated to reflect
 * the actual r3 if possible, and a ptr to the error log entry
 * will be returned if found.
 *
 * The mce_data_buf does not have any locks or protection around it,
 * if a second machine check comes in, or a system reset is done
 * before we have logged the error, then we will get corruption in the
 * error log.  This is preferable over holding off on calling
 * ibm,nmi-interlock which would result in us checkstopping if a
 * second machine check did come in.
 */
static struct rtas_error_log *fwnmi_get_errinfo(struct pt_regs *regs)
{
	unsigned long errdata = regs->gpr[3];
	struct rtas_error_log *errhdr = NULL;
	unsigned long *savep;

	if ((errdata >= 0x7000 && errdata < 0x7fff0) ||
	    (errdata >= rtas.base && errdata < rtas.base + rtas.size - 16)) {
		savep = __va(errdata);
		regs->gpr[3] = savep[0];	/* restore original r3 */
		memset(mce_data_buf, 0, RTAS_ERROR_LOG_MAX);
		memcpy(mce_data_buf, (char *)(savep + 1), RTAS_ERROR_LOG_MAX);
		errhdr = (struct rtas_error_log *)mce_data_buf;
	} else {
		printk("FWNMI: corrupt r3\n");
	}
	return errhdr;
}

/* Call this when done with the data returned by FWNMI_get_errinfo.
 * It will release the saved data area for other CPUs in the
 * partition to receive FWNMI errors.
 */
static void fwnmi_release_errinfo(void)
{
	int ret = rtas_call(rtas_token("ibm,nmi-interlock"), 0, 1, NULL);
	if (ret != 0)
		printk("FWNMI: nmi-interlock failed: %d\n", ret);
}

void pSeries_system_reset_exception(struct pt_regs *regs)
{
	if (fwnmi_active) {
		struct rtas_error_log *errhdr = fwnmi_get_errinfo(regs);
		if (errhdr) {
			/* XXX Should look at FWNMI information */
		}
		fwnmi_release_errinfo();
	}
}

/*
 * See if we can recover from a machine check exception.
 * This is only called on power4 (or above) and only via
 * the Firmware Non-Maskable Interrupts (fwnmi) handler
 * which provides the error analysis for us.
 *
 * Return 1 if corrected (or delivered a signal).
 * Return 0 if there is nothing we can do.
 */
static int recover_mce(struct pt_regs *regs, struct rtas_error_log * err)
{
	int nonfatal = 0;

	if (err->disposition == RTAS_DISP_FULLY_RECOVERED) {
		/* Platform corrected itself */
		nonfatal = 1;
	} else if ((regs->msr & MSR_RI) &&
		   user_mode(regs) &&
		   err->severity == RTAS_SEVERITY_ERROR_SYNC &&
		   err->disposition == RTAS_DISP_NOT_RECOVERED &&
		   err->target == RTAS_TARGET_MEMORY &&
		   err->type == RTAS_TYPE_ECC_UNCORR &&
		   !(current->pid == 0 || current->pid == 1)) {
		/* Kill off a user process with an ECC error */
		printk(KERN_ERR "MCE: uncorrectable ecc error for pid %d\n",
		       current->pid);
		/* XXX something better for ECC error? */
		_exception(SIGBUS, regs, BUS_ADRERR, regs->nip);
		nonfatal = 1;
	}

	log_error((char *)err, ERR_TYPE_RTAS_LOG, !nonfatal);

	return nonfatal;
}

/*
 * Handle a machine check.
 *
 * Note that on Power 4 and beyond Firmware Non-Maskable Interrupts (fwnmi)
 * should be present.  If so the handler which called us tells us if the
 * error was recovered (never true if RI=0).
 *
 * On hardware prior to Power 4 these exceptions were asynchronous which
 * means we can't tell exactly where it occurred and so we can't recover.
 */
int pSeries_machine_check_exception(struct pt_regs *regs)
{
	struct rtas_error_log *errp;

	if (fwnmi_active) {
		errp = fwnmi_get_errinfo(regs);
		fwnmi_release_errinfo();
		if (errp && recover_mce(regs, errp))
			return 1;
	}

	return 0;
}
Commit	Line	Data
1da177e4	1	/*
1da177e4	2	* Copyright (C) 2001 Dave Engebretsen IBM Corporation
d9953105	3	*
1da177e4 LT	4	* This program is free software; you can redistribute it and/or modify
	5	* it under the terms of the GNU General Public License as published by
	6	* the Free Software Foundation; either version 2 of the License, or
	7	* (at your option) any later version.
d9953105	8	*
1da177e4 LT	9	* This program is distributed in the hope that it will be useful,
	10	* but WITHOUT ANY WARRANTY; without even the implied warranty of
	11	* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
	12	* GNU General Public License for more details.
d9953105	13	*
1da177e4 LT	14	* You should have received a copy of the GNU General Public License
	15	* along with this program; if not, write to the Free Software
	16	* Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
	17	*/
	18
	19	/* Change Activity:
	20	* 2001/09/21 : engebret : Created with minimal EPOW and HW exception support.
d9953105	21	* End Change Activity
1da177e4 LT	22	*/
	23
	24	#include <linux/errno.h>
	25	#include <linux/threads.h>
	26	#include <linux/kernel_stat.h>
	27	#include <linux/signal.h>
	28	#include <linux/sched.h>
	29	#include <linux/ioport.h>
	30	#include <linux/interrupt.h>
	31	#include <linux/timex.h>
	32	#include <linux/init.h>
	33	#include <linux/slab.h>
	34	#include <linux/pci.h>
	35	#include <linux/delay.h>
	36	#include <linux/irq.h>
	37	#include <linux/random.h>
	38	#include <linux/sysrq.h>
	39	#include <linux/bitops.h>
	40
	41	#include <asm/uaccess.h>
	42	#include <asm/system.h>
	43	#include <asm/io.h>
	44	#include <asm/pgtable.h>
	45	#include <asm/irq.h>
	46	#include <asm/cache.h>
	47	#include <asm/prom.h>
	48	#include <asm/ptrace.h>
1da177e4 LT	49	#include <asm/machdep.h>
1da177e4 LT	50	#include <asm/rtas.h>
dcad47fc	51	#include <asm/udbg.h>
1da177e4 LT	52
	53	static unsigned char ras_log_buf[RTAS_ERROR_LOG_MAX];
	54	static DEFINE_SPINLOCK(ras_log_buf_lock);
	55
	56	char mce_data_buf[RTAS_ERROR_LOG_MAX]
	57	;
	58	/* This is true if we are using the firmware NMI handler (typically LPAR) */
	59	extern int fwnmi_active;
	60
1da177e4 LT	61	static int ras_get_sensor_state_token;
	62	static int ras_check_exception_token;
	63
	64	#define EPOW_SENSOR_TOKEN 9
	65	#define EPOW_SENSOR_INDEX 0
	66	#define RAS_VECTOR_OFFSET 0x500
	67
	68	static irqreturn_t ras_epow_interrupt(int irq, void *dev_id,
	69	struct pt_regs * regs);
	70	static irqreturn_t ras_error_interrupt(int irq, void *dev_id,
	71	struct pt_regs * regs);
	72
	73	/* #define DEBUG */
	74
	75	static void request_ras_irqs(struct device_node np, char propname,
	76	irqreturn_t (handler)(int, void , struct pt_regs *),
	77	const char *name)
	78	{
	79	unsigned int *ireg, len, i;
	80	int virq, n_intr;
	81
	82	ireg = (unsigned int *)get_property(np, propname, &len);
	83	if (ireg == NULL)
	84	return;
	85	n_intr = prom_n_intr_cells(np);
	86	len /= n_intr * sizeof(*ireg);
	87
	88	for (i = 0; i < len; i++) {
	89	virq = virt_irq_create_mapping(*ireg);
	90	if (virq == NO_IRQ) {
	91	printk(KERN_ERR "Unable to allocate interrupt "
	92	"number for %s\n", np->full_name);
	93	return;
	94	}
	95	if (request_irq(irq_offset_up(virq), handler, 0, name, NULL)) {
	96	printk(KERN_ERR "Unable to request interrupt %d for "
	97	"%s\n", irq_offset_up(virq), np->full_name);
	98	return;
	99	}
	100	ireg += n_intr;
	101	}
	102	}
	103
	104	/*
	105	* Initialize handlers for the set of interrupts caused by hardware errors
	106	* and power system events.
	107	*/
	108	static int __init init_ras_IRQ(void)
	109	{
	110	struct device_node *np;
	111
	112	ras_get_sensor_state_token = rtas_token("get-sensor-state");
	113	ras_check_exception_token = rtas_token("check-exception");
	114
	115	/* Internal Errors */
	116	np = of_find_node_by_path("/event-sources/internal-errors");
	117	if (np != NULL) {
	118	request_ras_irqs(np, "open-pic-interrupt", ras_error_interrupt,
	119	"RAS_ERROR");
	120	request_ras_irqs(np, "interrupts", ras_error_interrupt,
	121	"RAS_ERROR");
	122	of_node_put(np);
	123	}
	124
125	/* EPOW Events */
126	np = of_find_node_by_path("/event-sources/epow-events");
127	if (np != NULL) {
128	request_ras_irqs(np, "open-pic-interrupt", ras_epow_interrupt,
129	"RAS_EPOW");
130	request_ras_irqs(np, "interrupts", ras_epow_interrupt,
131	"RAS_EPOW");
132	of_node_put(np);
133	}
134
135	return 1;
136	}
137	__initcall(init_ras_IRQ);
138
139	/*
140	* Handle power subsystem events (EPOW).
141	*
142	* Presently we just log the event has occurred. This should be fixed
143	* to examine the type of power failure and take appropriate action where
144	* the time horizon permits something useful to be done.
145	*/
146	static irqreturn_t
147	ras_epow_interrupt(int irq, void dev_id, struct pt_regs regs)
148	{
149	int status = 0xdeadbeef;
150	int state = 0;
151	int critical;
152
153	status = rtas_call(ras_get_sensor_state_token, 2, 2, &state,
154	EPOW_SENSOR_TOKEN, EPOW_SENSOR_INDEX);
155
156	if (state > 3)
157	critical = 1; /* Time Critical */
158	else
159	critical = 0;
160
161	spin_lock(&ras_log_buf_lock);
162
163	status = rtas_call(ras_check_exception_token, 6, 1, NULL,
164	RAS_VECTOR_OFFSET,
165	virt_irq_to_real(irq_offset_down(irq)),
166	RTAS_EPOW_WARNING \| RTAS_POWERMGM_EVENTS,
167	critical, __pa(&ras_log_buf),
168	rtas_get_error_log_max());
169
170	udbg_printf("EPOW <0x%lx 0x%x 0x%x>\n",
171	((unsigned long )&ras_log_buf), status, state);
172	printk(KERN_WARNING "EPOW <0x%lx 0x%x 0x%x>\n",
173	((unsigned long )&ras_log_buf), status, state);
174
175	/* format and print the extended information */
176	log_error(ras_log_buf, ERR_TYPE_RTAS_LOG, 0);
177
178	spin_unlock(&ras_log_buf_lock);
179	return IRQ_HANDLED;
180	}
181
182	/*
183	* Handle hardware error interrupts.
184	*
185	* RTAS check-exception is called to collect data on the exception. If
186	* the error is deemed recoverable, we log a warning and return.
187	* For nonrecoverable errors, an error is logged and we stop all processing
188	* as quickly as possible in order to prevent propagation of the failure.
189	*/
190	static irqreturn_t
191	ras_error_interrupt(int irq, void dev_id, struct pt_regs regs)
192	{
193	struct rtas_error_log *rtas_elog;
194	int status = 0xdeadbeef;
195	int fatal;
196
197	spin_lock(&ras_log_buf_lock);
198
199	status = rtas_call(ras_check_exception_token, 6, 1, NULL,
200	RAS_VECTOR_OFFSET,
201	virt_irq_to_real(irq_offset_down(irq)),
202	RTAS_INTERNAL_ERROR, 1 /Time Critical /,
203	__pa(&ras_log_buf),
204	rtas_get_error_log_max());
205
206	rtas_elog = (struct rtas_error_log *)ras_log_buf;
207
208	if ((status == 0) && (rtas_elog->severity >= RTAS_SEVERITY_ERROR_SYNC))
209	fatal = 1;
210	else
211	fatal = 0;
212
213	/* format and print the extended information */
214	log_error(ras_log_buf, ERR_TYPE_RTAS_LOG, fatal);
215
216	if (fatal) {
217	udbg_printf("Fatal HW Error <0x%lx 0x%x>\n",
218	((unsigned long )&ras_log_buf), status);
219	printk(KERN_EMERG "Error: Fatal hardware error <0x%lx 0x%x>\n",
220	((unsigned long )&ras_log_buf), status);
221
222	#ifndef DEBUG
223	/* Don't actually power off when debugging so we can test
224	* without actually failing while injecting errors.
225	* Error data will not be logged to syslog.
226	*/
227	ppc_md.power_off();
228	#endif
229	} else {
230	udbg_printf("Recoverable HW Error <0x%lx 0x%x>\n",
231	((unsigned long )&ras_log_buf), status);
232	printk(KERN_WARNING
233	"Warning: Recoverable hardware error <0x%lx 0x%x>\n",
234	((unsigned long )&ras_log_buf), status);
235	}
236
237	spin_unlock(&ras_log_buf_lock);
238	return IRQ_HANDLED;
239	}
240
241	/* Get the error information for errors coming through the
242	* FWNMI vectors. The pt_regs' r3 will be updated to reflect
243	* the actual r3 if possible, and a ptr to the error log entry
244	* will be returned if found.
245	*
246	* The mce_data_buf does not have any locks or protection around it,
247	* if a second machine check comes in, or a system reset is done
248	* before we have logged the error, then we will get corruption in the
249	* error log. This is preferable over holding off on calling
250	* ibm,nmi-interlock which would result in us checkstopping if a
251	* second machine check did come in.
252	*/
253	static struct rtas_error_log fwnmi_get_errinfo(struct pt_regs regs)
254	{
255	unsigned long errdata = regs->gpr[3];
256	struct rtas_error_log *errhdr = NULL;
257	unsigned long *savep;
258
259	if ((errdata >= 0x7000 && errdata < 0x7fff0) \|\|
260	(errdata >= rtas.base && errdata < rtas.base + rtas.size - 16)) {
261	savep = __va(errdata);
262	regs->gpr[3] = savep[0]; /* restore original r3 */
263	memset(mce_data_buf, 0, RTAS_ERROR_LOG_MAX);
264	memcpy(mce_data_buf, (char *)(savep + 1), RTAS_ERROR_LOG_MAX);
265	errhdr = (struct rtas_error_log *)mce_data_buf;
266	} else {
267	printk("FWNMI: corrupt r3\n");
268	}
269	return errhdr;
270	}
271
272	/* Call this when done with the data returned by FWNMI_get_errinfo.
273	* It will release the saved data area for other CPUs in the
274	* partition to receive FWNMI errors.
275	*/
276	static void fwnmi_release_errinfo(void)
277	{
278	int ret = rtas_call(rtas_token("ibm,nmi-interlock"), 0, 1, NULL);
279	if (ret != 0)
280	printk("FWNMI: nmi-interlock failed: %d\n", ret);
281	}
282
283	void pSeries_system_reset_exception(struct pt_regs *regs)
284	{
285	if (fwnmi_active) {
286	struct rtas_error_log *errhdr = fwnmi_get_errinfo(regs);
287	if (errhdr) {
288	/* XXX Should look at FWNMI information */
289	}
290	fwnmi_release_errinfo();
291	}
292	}
293
294	/*
295	* See if we can recover from a machine check exception.
296	* This is only called on power4 (or above) and only via
297	* the Firmware Non-Maskable Interrupts (fwnmi) handler
298	* which provides the error analysis for us.
299	*
300	* Return 1 if corrected (or delivered a signal).
301	* Return 0 if there is nothing we can do.
302	*/
303	static int recover_mce(struct pt_regs regs, struct rtas_error_log err)
304	{
305	int nonfatal = 0;
306
307	if (err->disposition == RTAS_DISP_FULLY_RECOVERED) {
308	/* Platform corrected itself */
309	nonfatal = 1;
310	} else if ((regs->msr & MSR_RI) &&
311	user_mode(regs) &&
312	err->severity == RTAS_SEVERITY_ERROR_SYNC &&
313	err->disposition == RTAS_DISP_NOT_RECOVERED &&
314	err->target == RTAS_TARGET_MEMORY &&
315	err->type == RTAS_TYPE_ECC_UNCORR &&
316	!(current->pid == 0 \|\| current->pid == 1)) {
317	/* Kill off a user process with an ECC error */
318	printk(KERN_ERR "MCE: uncorrectable ecc error for pid %d\n",
319	current->pid);
320	/* XXX something better for ECC error? */
321	_exception(SIGBUS, regs, BUS_ADRERR, regs->nip);
322	nonfatal = 1;
323	}
324
d9953105	325	log_error((char *)err, ERR_TYPE_RTAS_LOG, !nonfatal);
1da177e4 LT	326
	327	return nonfatal;
	328	}
	329
	330	/*
	331	* Handle a machine check.
	332	*
	333	* Note that on Power 4 and beyond Firmware Non-Maskable Interrupts (fwnmi)
	334	* should be present. If so the handler which called us tells us if the
	335	* error was recovered (never true if RI=0).
	336	*
	337	* On hardware prior to Power 4 these exceptions were asynchronous which
	338	* means we can't tell exactly where it occurred and so we can't recover.
	339	*/
	340	int pSeries_machine_check_exception(struct pt_regs *regs)
	341	{
	342	struct rtas_error_log *errp;
	343
	344	if (fwnmi_active) {
	345	errp = fwnmi_get_errinfo(regs);
	346	fwnmi_release_errinfo();
	347	if (errp && recover_mce(regs, errp))
	348	return 1;
	349	}
	350
	351	return 0;
	352	}