[linux-2.6-block.git] / arch / ia64 / kernel / crash.c

/*
 * arch/ia64/kernel/crash.c
 *
 * Architecture specific (ia64) functions for kexec based crash dumps.
 *
 * Created by: Khalid Aziz <khalid.aziz@hp.com>
 * Copyright (C) 2005 Hewlett-Packard Development Company, L.P.
 * Copyright (C) 2005 Intel Corp	Zou Nan hai <nanhai.zou@intel.com>
 *
 */
#include <linux/smp.h>
#include <linux/delay.h>
#include <linux/crash_dump.h>
#include <linux/bootmem.h>
#include <linux/kexec.h>
#include <linux/elfcore.h>
#include <linux/sysctl.h>
#include <linux/init.h>
#include <linux/kdebug.h>

#include <asm/mca.h>

int kdump_status[NR_CPUS];
static atomic_t kdump_cpu_frozen;
atomic_t kdump_in_progress;
static int kdump_freeze_monarch;
static int kdump_on_init = 1;
static int kdump_on_fatal_mca = 1;

static inline Elf64_Word
*append_elf_note(Elf64_Word *buf, char *name, unsigned type, void *data,
		size_t data_len)
{
	struct elf_note *note = (struct elf_note *)buf;
	note->n_namesz = strlen(name) + 1;
	note->n_descsz = data_len;
	note->n_type   = type;
	buf += (sizeof(*note) + 3)/4;
	memcpy(buf, name, note->n_namesz);
	buf += (note->n_namesz + 3)/4;
	memcpy(buf, data, data_len);
	buf += (data_len + 3)/4;
	return buf;
}

static void
final_note(void *buf)
{
	memset(buf, 0, sizeof(struct elf_note));
}

extern void ia64_dump_cpu_regs(void *);

static DEFINE_PER_CPU(struct elf_prstatus, elf_prstatus);

void
crash_save_this_cpu(void)
{
	void *buf;
	unsigned long cfm, sof, sol;

	int cpu = smp_processor_id();
	struct elf_prstatus *prstatus = &per_cpu(elf_prstatus, cpu);

	elf_greg_t *dst = (elf_greg_t *)&(prstatus->pr_reg);
	memset(prstatus, 0, sizeof(*prstatus));
	prstatus->pr_pid = current->pid;

	ia64_dump_cpu_regs(dst);
	cfm = dst[43];
	sol = (cfm >> 7) & 0x7f;
	sof = cfm & 0x7f;
	dst[46] = (unsigned long)ia64_rse_skip_regs((unsigned long *)dst[46],
			sof - sol);

	buf = (u64 *) per_cpu_ptr(crash_notes, cpu);
	if (!buf)
		return;
	buf = append_elf_note(buf, KEXEC_CORE_NOTE_NAME, NT_PRSTATUS, prstatus,
			sizeof(*prstatus));
	final_note(buf);
}

#ifdef CONFIG_SMP
static int
kdump_wait_cpu_freeze(void)
{
	int cpu_num = num_online_cpus() - 1;
	int timeout = 1000;
	while(timeout-- > 0) {
		if (atomic_read(&kdump_cpu_frozen) == cpu_num)
			return 0;
		udelay(1000);
	}
	return 1;
}
#endif

void
machine_crash_shutdown(struct pt_regs *pt)
{
	/* This function is only called after the system
	 * has paniced or is otherwise in a critical state.
	 * The minimum amount of code to allow a kexec'd kernel
	 * to run successfully needs to happen here.
	 *
	 * In practice this means shooting down the other cpus in
	 * an SMP system.
	 */
	kexec_disable_iosapic();
#ifdef CONFIG_SMP
	/*
	 * If kdump_on_init is set and an INIT is asserted here, kdump will
	 * be started again via INIT monarch.
	 */
	local_irq_disable();
	ia64_set_psr_mc();	/* mask MCA/INIT */
	if (atomic_inc_return(&kdump_in_progress) != 1)
		unw_init_running(kdump_cpu_freeze, NULL);

	/*
	 * Now this cpu is ready for kdump.
	 * Stop all others by IPI or INIT.  They could receive INIT from
	 * outside and might be INIT monarch, but only thing they have to
	 * do is falling into kdump_cpu_freeze().
	 *
	 * If an INIT is asserted here:
	 * - All receivers might be slaves, since some of cpus could already
	 *   be frozen and INIT might be masked on monarch.  In this case,
	 *   all slaves will be frozen soon since kdump_in_progress will let
	 *   them into DIE_INIT_SLAVE_LEAVE.
	 * - One might be a monarch, but INIT rendezvous will fail since
	 *   at least this cpu already have INIT masked so it never join
	 *   to the rendezvous.  In this case, all slaves and monarch will
	 *   be frozen soon with no wait since the INIT rendezvous is skipped
	 *   by kdump_in_progress.
	 */
	kdump_smp_send_stop();
	/* not all cpu response to IPI, send INIT to freeze them */
	if (kdump_wait_cpu_freeze()) {
		kdump_smp_send_init();
		/* wait again, don't go ahead if possible */
		kdump_wait_cpu_freeze();
	}
#endif
}

static void
machine_kdump_on_init(void)
{
	crash_save_vmcoreinfo();
	local_irq_disable();
	kexec_disable_iosapic();
	machine_kexec(ia64_kimage);
}

void
kdump_cpu_freeze(struct unw_frame_info *info, void *arg)
{
	int cpuid;

	local_irq_disable();
	cpuid = smp_processor_id();
	crash_save_this_cpu();
	current->thread.ksp = (__u64)info->sw - 16;

	ia64_set_psr_mc();	/* mask MCA/INIT and stop reentrance */

	atomic_inc(&kdump_cpu_frozen);
	kdump_status[cpuid] = 1;
	mb();
	for (;;)
		cpu_relax();
}

static int
kdump_init_notifier(struct notifier_block *self, unsigned long val, void *data)
{
	struct ia64_mca_notify_die *nd;
	struct die_args *args = data;

	if (atomic_read(&kdump_in_progress)) {
		switch (val) {
		case DIE_INIT_MONARCH_LEAVE:
			if (!kdump_freeze_monarch)
				break;
			/* fall through */
		case DIE_INIT_SLAVE_LEAVE:
		case DIE_INIT_MONARCH_ENTER:
		case DIE_MCA_RENDZVOUS_LEAVE:
			unw_init_running(kdump_cpu_freeze, NULL);
			break;
		}
	}

	if (!kdump_on_init && !kdump_on_fatal_mca)
		return NOTIFY_DONE;

	if (!ia64_kimage) {
		if (val == DIE_INIT_MONARCH_LEAVE)
			ia64_mca_printk(KERN_NOTICE
					"%s: kdump not configured\n",
					__func__);
		return NOTIFY_DONE;
	}

	if (val != DIE_INIT_MONARCH_LEAVE &&
	    val != DIE_INIT_MONARCH_PROCESS &&
	    val != DIE_MCA_MONARCH_LEAVE)
		return NOTIFY_DONE;

	nd = (struct ia64_mca_notify_die *)args->err;

	switch (val) {
	case DIE_INIT_MONARCH_PROCESS:
		/* Reason code 1 means machine check rendezvous*/
		if (kdump_on_init && (nd->sos->rv_rc != 1)) {
			if (atomic_inc_return(&kdump_in_progress) != 1)
				kdump_freeze_monarch = 1;
		}
		break;
	case DIE_INIT_MONARCH_LEAVE:
		/* Reason code 1 means machine check rendezvous*/
		if (kdump_on_init && (nd->sos->rv_rc != 1))
			machine_kdump_on_init();
		break;
	case DIE_MCA_MONARCH_LEAVE:
		/* *(nd->data) indicate if MCA is recoverable */
		if (kdump_on_fatal_mca && !(*(nd->data))) {
			if (atomic_inc_return(&kdump_in_progress) == 1)
				machine_kdump_on_init();
			/* We got fatal MCA while kdump!? No way!! */
		}
		break;
	}
	return NOTIFY_DONE;
}

#ifdef CONFIG_SYSCTL
static struct ctl_table kdump_ctl_table[] = {
	{
		.procname = "kdump_on_init",
		.data = &kdump_on_init,
		.maxlen = sizeof(int),
		.mode = 0644,
		.proc_handler = proc_dointvec,
	},
	{
		.procname = "kdump_on_fatal_mca",
		.data = &kdump_on_fatal_mca,
		.maxlen = sizeof(int),
		.mode = 0644,
		.proc_handler = proc_dointvec,
	},
	{ }
};

static struct ctl_table sys_table[] = {
	{
	  .procname = "kernel",
	  .mode = 0555,
	  .child = kdump_ctl_table,
	},
	{ }
};
#endif

static int
machine_crash_setup(void)
{
	/* be notified before default_monarch_init_process */
	static struct notifier_block kdump_init_notifier_nb = {
		.notifier_call = kdump_init_notifier,
		.priority = 1,
	};
	int ret;
	if((ret = register_die_notifier(&kdump_init_notifier_nb)) != 0)
		return ret;
#ifdef CONFIG_SYSCTL
	register_sysctl_table(sys_table);
#endif
	return 0;
}

__initcall(machine_crash_setup);
Commit	Line	Data
a7956113 ZN	1	/*
	2	* arch/ia64/kernel/crash.c
	3	*
	4	* Architecture specific (ia64) functions for kexec based crash dumps.
	5	*
	6	* Created by: Khalid Aziz <khalid.aziz@hp.com>
	7	* Copyright (C) 2005 Hewlett-Packard Development Company, L.P.
	8	* Copyright (C) 2005 Intel Corp Zou Nan hai <nanhai.zou@intel.com>
	9	*
	10	*/
	11	#include <linux/smp.h>
	12	#include <linux/delay.h>
	13	#include <linux/crash_dump.h>
	14	#include <linux/bootmem.h>
	15	#include <linux/kexec.h>
	16	#include <linux/elfcore.h>
	17	#include <linux/sysctl.h>
	18	#include <linux/init.h>
1eeb66a1	19	#include <linux/kdebug.h>
a7956113	20
a7956113	21	#include <asm/mca.h>
a7956113 ZN	22
a7956113 ZN	23	int kdump_status[NR_CPUS];
0ac1faca	24	static atomic_t kdump_cpu_frozen;
a7956113	25	atomic_t kdump_in_progress;
1726b088	26	static int kdump_freeze_monarch;
0ac1faca	27	static int kdump_on_init = 1;
b0247a55	28	static int kdump_on_fatal_mca = 1;
a7956113 ZN	29
	30	static inline Elf64_Word
	31	append_elf_note(Elf64_Word buf, char name, unsigned type, void data,
	32	size_t data_len)
	33	{
	34	struct elf_note note = (struct elf_note )buf;
	35	note->n_namesz = strlen(name) + 1;
	36	note->n_descsz = data_len;
	37	note->n_type = type;
	38	buf += (sizeof(*note) + 3)/4;
	39	memcpy(buf, name, note->n_namesz);
	40	buf += (note->n_namesz + 3)/4;
	41	memcpy(buf, data, data_len);
	42	buf += (data_len + 3)/4;
	43	return buf;
	44	}
	45
	46	static void
	47	final_note(void *buf)
	48	{
	49	memset(buf, 0, sizeof(struct elf_note));
	50	}
	51
	52	extern void ia64_dump_cpu_regs(void *);
	53
	54	static DEFINE_PER_CPU(struct elf_prstatus, elf_prstatus);
	55
	56	void
ccbebdac	57	crash_save_this_cpu(void)
a7956113 ZN	58	{
	59	void *buf;
	60	unsigned long cfm, sof, sol;
	61
	62	int cpu = smp_processor_id();
	63	struct elf_prstatus *prstatus = &per_cpu(elf_prstatus, cpu);
	64
	65	elf_greg_t dst = (elf_greg_t )&(prstatus->pr_reg);
	66	memset(prstatus, 0, sizeof(*prstatus));
	67	prstatus->pr_pid = current->pid;
	68
	69	ia64_dump_cpu_regs(dst);
	70	cfm = dst[43];
	71	sol = (cfm >> 7) & 0x7f;
	72	sof = cfm & 0x7f;
	73	dst[46] = (unsigned long)ia64_rse_skip_regs((unsigned long *)dst[46],
	74	sof - sol);
	75
	76	buf = (u64 *) per_cpu_ptr(crash_notes, cpu);
	77	if (!buf)
	78	return;
6672f76a	79	buf = append_elf_note(buf, KEXEC_CORE_NOTE_NAME, NT_PRSTATUS, prstatus,
a7956113 ZN	80	sizeof(*prstatus));
	81	final_note(buf);
	82	}
	83
bcb9b99d	84	#ifdef CONFIG_SMP
a7956113 ZN	85	static int
	86	kdump_wait_cpu_freeze(void)
	87	{
	88	int cpu_num = num_online_cpus() - 1;
	89	int timeout = 1000;
	90	while(timeout-- > 0) {
0ac1faca	91	if (atomic_read(&kdump_cpu_frozen) == cpu_num)
a7956113 ZN	92	return 0;
	93	udelay(1000);
	94	}
	95	return 1;
	96	}
bcb9b99d	97	#endif
a7956113 ZN	98
	99	void
	100	machine_crash_shutdown(struct pt_regs *pt)
	101	{
	102	/* This function is only called after the system
	103	* has paniced or is otherwise in a critical state.
	104	* The minimum amount of code to allow a kexec'd kernel
	105	* to run successfully needs to happen here.
	106	*
	107	* In practice this means shooting down the other cpus in
	108	* an SMP system.
	109	*/
	110	kexec_disable_iosapic();
	111	#ifdef CONFIG_SMP
1726b088 HS	112	/*
	113	* If kdump_on_init is set and an INIT is asserted here, kdump will
	114	* be started again via INIT monarch.
	115	*/
	116	local_irq_disable();
	117	ia64_set_psr_mc(); /* mask MCA/INIT */
	118	if (atomic_inc_return(&kdump_in_progress) != 1)
	119	unw_init_running(kdump_cpu_freeze, NULL);
	120
	121	/*
	122	* Now this cpu is ready for kdump.
	123	* Stop all others by IPI or INIT. They could receive INIT from
	124	* outside and might be INIT monarch, but only thing they have to
	125	* do is falling into kdump_cpu_freeze().
	126	*
	127	* If an INIT is asserted here:
	128	* - All receivers might be slaves, since some of cpus could already
	129	* be frozen and INIT might be masked on monarch. In this case,
0cced40e HS	130	* all slaves will be frozen soon since kdump_in_progress will let
0cced40e HS	131	* them into DIE_INIT_SLAVE_LEAVE.
1726b088 HS	132	* - One might be a monarch, but INIT rendezvous will fail since
	133	* at least this cpu already have INIT masked so it never join
	134	* to the rendezvous. In this case, all slaves and monarch will
0cced40e HS	135	* be frozen soon with no wait since the INIT rendezvous is skipped
0cced40e HS	136	* by kdump_in_progress.
1726b088	137	*/
a7956113	138	kdump_smp_send_stop();
0ac1faca	139	/* not all cpu response to IPI, send INIT to freeze them */
5959906e	140	if (kdump_wait_cpu_freeze()) {
a7956113	141	kdump_smp_send_init();
5959906e HS	142	/* wait again, don't go ahead if possible */
5959906e HS	143	kdump_wait_cpu_freeze();
a7956113 ZN	144	}
	145	#endif
	146	}
	147
	148	static void
	149	machine_kdump_on_init(void)
	150	{
072f042d	151	crash_save_vmcoreinfo();
a7956113 ZN	152	local_irq_disable();
	153	kexec_disable_iosapic();
	154	machine_kexec(ia64_kimage);
	155	}
	156
	157	void
	158	kdump_cpu_freeze(struct unw_frame_info info, void arg)
	159	{
	160	int cpuid;
4295ab34	161
a7956113 ZN	162	local_irq_disable();
	163	cpuid = smp_processor_id();
	164	crash_save_this_cpu();
	165	current->thread.ksp = (__u64)info->sw - 16;
4295ab34 HS	166
	167	ia64_set_psr_mc(); /* mask MCA/INIT and stop reentrance */
	168
0ac1faca	169	atomic_inc(&kdump_cpu_frozen);
a7956113 ZN	170	kdump_status[cpuid] = 1;
a7956113 ZN	171	mb();
bcb9b99d MD	172	for (;;)
bcb9b99d MD	173	cpu_relax();
a7956113 ZN	174	}
	175
	176	static int
	177	kdump_init_notifier(struct notifier_block self, unsigned long val, void data)
	178	{
	179	struct ia64_mca_notify_die *nd;
	180	struct die_args *args = data;
	181
5959906e HS	182	if (atomic_read(&kdump_in_progress)) {
	183	switch (val) {
	184	case DIE_INIT_MONARCH_LEAVE:
	185	if (!kdump_freeze_monarch)
	186	break;
	187	/* fall through */
	188	case DIE_INIT_SLAVE_LEAVE:
0cced40e	189	case DIE_INIT_MONARCH_ENTER:
5959906e HS	190	case DIE_MCA_RENDZVOUS_LEAVE:
	191	unw_init_running(kdump_cpu_freeze, NULL);
	192	break;
	193	}
	194	}
	195
b0247a55	196	if (!kdump_on_init && !kdump_on_fatal_mca)
a7956113 ZN	197	return NOTIFY_DONE;
a7956113 ZN	198
2010d7fe KK	199	if (!ia64_kimage) {
	200	if (val == DIE_INIT_MONARCH_LEAVE)
	201	ia64_mca_printk(KERN_NOTICE
	202	"%s: kdump not configured\n",
d4ed8084	203	__func__);
2010d7fe KK	204	return NOTIFY_DONE;
	205	}
	206
311f594d	207	if (val != DIE_INIT_MONARCH_LEAVE &&
311f594d	208	val != DIE_INIT_MONARCH_PROCESS &&
a7956113 ZN	209	val != DIE_MCA_MONARCH_LEAVE)
	210	return NOTIFY_DONE;
	211
	212	nd = (struct ia64_mca_notify_die *)args->err;
a7956113 ZN	213
a7956113 ZN	214	switch (val) {
3975afff	215	case DIE_INIT_MONARCH_PROCESS:
5959906e HS	216	/* Reason code 1 means machine check rendezvous*/
5959906e HS	217	if (kdump_on_init && (nd->sos->rv_rc != 1)) {
1726b088 HS	218	if (atomic_inc_return(&kdump_in_progress) != 1)
1726b088 HS	219	kdump_freeze_monarch = 1;
3975afff HS	220	}
	221	break;
	222	case DIE_INIT_MONARCH_LEAVE:
5959906e HS	223	/* Reason code 1 means machine check rendezvous*/
	224	if (kdump_on_init && (nd->sos->rv_rc != 1))
	225	machine_kdump_on_init();
3975afff HS	226	break;
3975afff HS	227	case DIE_MCA_MONARCH_LEAVE:
4fa2f0e6 HS	228	/* (nd->data) indicate if MCA is recoverable /
4fa2f0e6 HS	229	if (kdump_on_fatal_mca && !(*(nd->data))) {
0cced40e	230	if (atomic_inc_return(&kdump_in_progress) == 1)
1726b088	231	machine_kdump_on_init();
1726b088	232	/* We got fatal MCA while kdump!? No way!! */
3975afff HS	233	}
3975afff HS	234	break;
a7956113 ZN	235	}
	236	return NOTIFY_DONE;
	237	}
	238
	239	#ifdef CONFIG_SYSCTL
2841efa6	240	static struct ctl_table kdump_ctl_table[] = {
a7956113	241	{
a7956113 ZN	242	.procname = "kdump_on_init",
	243	.data = &kdump_on_init,
	244	.maxlen = sizeof(int),
	245	.mode = 0644,
6d456111	246	.proc_handler = proc_dointvec,
a7956113	247	},
b0247a55	248	{
b0247a55 HS	249	.procname = "kdump_on_fatal_mca",
	250	.data = &kdump_on_fatal_mca,
	251	.maxlen = sizeof(int),
	252	.mode = 0644,
6d456111	253	.proc_handler = proc_dointvec,
b0247a55	254	},
d00faf81	255	{ }
a7956113 ZN	256	};
a7956113 ZN	257
2841efa6	258	static struct ctl_table sys_table[] = {
a7956113	259	{
a7956113 ZN	260	.procname = "kernel",
a7956113 ZN	261	.mode = 0555,
b0247a55	262	.child = kdump_ctl_table,
a7956113	263	},
d00faf81	264	{ }
a7956113 ZN	265	};
	266	#endif
	267
	268	static int
	269	machine_crash_setup(void)
	270	{
311f594d	271	/* be notified before default_monarch_init_process */
a7956113 ZN	272	static struct notifier_block kdump_init_notifier_nb = {
a7956113 ZN	273	.notifier_call = kdump_init_notifier,
311f594d	274	.priority = 1,
a7956113 ZN	275	};
a7956113 ZN	276	int ret;
a7956113 ZN	277	if((ret = register_die_notifier(&kdump_init_notifier_nb)) != 0)
	278	return ret;
	279	#ifdef CONFIG_SYSCTL
0b4d4147	280	register_sysctl_table(sys_table);
a7956113 ZN	281	#endif
	282	return 0;
	283	}
	284
	285	__initcall(machine_crash_setup);
	286