[net-next-2.6.git] / arch / powerpc / kernel / machine_kexec_64.c

/*
 * PPC64 code to handle Linux booting another kernel.
 *
 * Copyright (C) 2004-2005, IBM Corp.
 *
 * Created by: Milton D Miller II
 *
 * This source code is licensed under the GNU General Public License,
 * Version 2.  See the file COPYING for more details.
 */


#include <linux/kexec.h>
#include <linux/smp.h>
#include <linux/thread_info.h>
#include <linux/init_task.h>
#include <linux/errno.h>

#include <asm/page.h>
#include <asm/current.h>
#include <asm/machdep.h>
#include <asm/cacheflush.h>
#include <asm/paca.h>
#include <asm/mmu.h>
#include <asm/sections.h>	/* _end */
#include <asm/prom.h>
#include <asm/smp.h>

int default_machine_kexec_prepare(struct kimage *image)
{
	int i;
	unsigned long begin, end;	/* limits of segment */
	unsigned long low, high;	/* limits of blocked memory range */
	struct device_node *node;
	const unsigned long *basep;
	const unsigned int *sizep;

	if (!ppc_md.hpte_clear_all)
		return -ENOENT;

	/*
	 * Since we use the kernel fault handlers and paging code to
	 * handle the virtual mode, we must make sure no destination
	 * overlaps kernel static data or bss.
	 */
	for (i = 0; i < image->nr_segments; i++)
		if (image->segment[i].mem < __pa(_end))
			return -ETXTBSY;

	/*
	 * For non-LPAR, we absolutely can not overwrite the mmu hash
	 * table, since we are still using the bolted entries in it to
	 * do the copy.  Check that here.
	 *
	 * It is safe if the end is below the start of the blocked
	 * region (end <= low), or if the beginning is after the
	 * end of the blocked region (begin >= high).  Use the
	 * boolean identity !(a || b)  === (!a && !b).
	 */
	if (htab_address) {
		low = __pa(htab_address);
		high = low + htab_size_bytes;

		for (i = 0; i < image->nr_segments; i++) {
			begin = image->segment[i].mem;
			end = begin + image->segment[i].memsz;

			if ((begin < high) && (end > low))
				return -ETXTBSY;
		}
	}

	/* We also should not overwrite the tce tables */
	for (node = of_find_node_by_type(NULL, "pci"); node != NULL;
			node = of_find_node_by_type(node, "pci")) {
		basep = of_get_property(node, "linux,tce-base", NULL);
		sizep = of_get_property(node, "linux,tce-size", NULL);
		if (basep == NULL || sizep == NULL)
			continue;

		low = *basep;
		high = low + (*sizep);

		for (i = 0; i < image->nr_segments; i++) {
			begin = image->segment[i].mem;
			end = begin + image->segment[i].memsz;

			if ((begin < high) && (end > low))
				return -ETXTBSY;
		}
	}

	return 0;
}

#define IND_FLAGS (IND_DESTINATION | IND_INDIRECTION | IND_DONE | IND_SOURCE)

static void copy_segments(unsigned long ind)
{
	unsigned long entry;
	unsigned long *ptr;
	void *dest;
	void *addr;

	/*
	 * We rely on kexec_load to create a lists that properly
	 * initializes these pointers before they are used.
	 * We will still crash if the list is wrong, but at least
	 * the compiler will be quiet.
	 */
	ptr = NULL;
	dest = NULL;

	for (entry = ind; !(entry & IND_DONE); entry = *ptr++) {
		addr = __va(entry & PAGE_MASK);

		switch (entry & IND_FLAGS) {
		case IND_DESTINATION:
			dest = addr;
			break;
		case IND_INDIRECTION:
			ptr = addr;
			break;
		case IND_SOURCE:
			copy_page(dest, addr);
			dest += PAGE_SIZE;
		}
	}
}

void kexec_copy_flush(struct kimage *image)
{
	long i, nr_segments = image->nr_segments;
	struct  kexec_segment ranges[KEXEC_SEGMENT_MAX];

	/* save the ranges on the stack to efficiently flush the icache */
	memcpy(ranges, image->segment, sizeof(ranges));

	/*
	 * After this call we may not use anything allocated in dynamic
	 * memory, including *image.
	 *
	 * Only globals and the stack are allowed.
	 */
	copy_segments(image->head);

	/*
	 * we need to clear the icache for all dest pages sometime,
	 * including ones that were in place on the original copy
	 */
	for (i = 0; i < nr_segments; i++)
		flush_icache_range((unsigned long)__va(ranges[i].mem),
			(unsigned long)__va(ranges[i].mem + ranges[i].memsz));
}

#ifdef CONFIG_SMP

static int kexec_all_irq_disabled = 0;

static void kexec_smp_down(void *arg)
{
	local_irq_disable();
	mb(); /* make sure our irqs are disabled before we say they are */
	get_paca()->kexec_state = KEXEC_STATE_IRQS_OFF;
	while(kexec_all_irq_disabled == 0)
		cpu_relax();
	mb(); /* make sure all irqs are disabled before this */
	/*
	 * Now every CPU has IRQs off, we can clear out any pending
	 * IPIs and be sure that no more will come in after this.
	 */
	if (ppc_md.kexec_cpu_down)
		ppc_md.kexec_cpu_down(0, 1);

	kexec_smp_wait();
	/* NOTREACHED */
}

static void kexec_prepare_cpus_wait(int wait_state)
{
	int my_cpu, i, notified=-1;

	my_cpu = get_cpu();
	/* Make sure each CPU has atleast made it to the state we need */
	for_each_online_cpu(i) {
		if (i == my_cpu)
			continue;

		while (paca[i].kexec_state < wait_state) {
			barrier();
			if (i != notified) {
				printk( "kexec: waiting for cpu %d (physical"
						" %d) to enter %i state\n",
					i, paca[i].hw_cpu_id, wait_state);
				notified = i;
			}
		}
	}
	mb();
}

static void kexec_prepare_cpus(void)
{

	smp_call_function(kexec_smp_down, NULL, /* wait */0);
	local_irq_disable();
	mb(); /* make sure IRQs are disabled before we say they are */
	get_paca()->kexec_state = KEXEC_STATE_IRQS_OFF;

	kexec_prepare_cpus_wait(KEXEC_STATE_IRQS_OFF);
	/* we are sure every CPU has IRQs off at this point */
	kexec_all_irq_disabled = 1;

	/* after we tell the others to go down */
	if (ppc_md.kexec_cpu_down)
		ppc_md.kexec_cpu_down(0, 0);

	/* Before removing MMU mapings make sure all CPUs have entered real mode */
	kexec_prepare_cpus_wait(KEXEC_STATE_REAL_MODE);

	put_cpu();
}

#else /* ! SMP */

static void kexec_prepare_cpus(void)
{
	/*
	 * move the secondarys to us so that we can copy
	 * the new kernel 0-0x100 safely
	 *
	 * do this if kexec in setup.c ?
	 *
	 * We need to release the cpus if we are ever going from an
	 * UP to an SMP kernel.
	 */
	smp_release_cpus();
	if (ppc_md.kexec_cpu_down)
		ppc_md.kexec_cpu_down(0, 0);
	local_irq_disable();
}

#endif /* SMP */

/*
 * kexec thread structure and stack.
 *
 * We need to make sure that this is 16384-byte aligned due to the
 * way process stacks are handled.  It also must be statically allocated
 * or allocated as part of the kimage, because everything else may be
 * overwritten when we copy the kexec image.  We piggyback on the
 * "init_task" linker section here to statically allocate a stack.
 *
 * We could use a smaller stack if we don't care about anything using
 * current, but that audit has not been performed.
 */
static union thread_union kexec_stack __init_task_data =
	{ };

/* Our assembly helper, in kexec_stub.S */
extern NORET_TYPE void kexec_sequence(void *newstack, unsigned long start,
					void *image, void *control,
					void (*clear_all)(void)) ATTRIB_NORET;

/* too late to fail here */
void default_machine_kexec(struct kimage *image)
{
	/* prepare control code if any */

	/*
        * If the kexec boot is the normal one, need to shutdown other cpus
        * into our wait loop and quiesce interrupts.
        * Otherwise, in the case of crashed mode (crashing_cpu >= 0),
        * stopping other CPUs and collecting their pt_regs is done before
        * using debugger IPI.
        */

	if (crashing_cpu == -1)
		kexec_prepare_cpus();

	/* switch to a staticly allocated stack.  Based on irq stack code.
	 * XXX: the task struct will likely be invalid once we do the copy!
	 */
	kexec_stack.thread_info.task = current_thread_info()->task;
	kexec_stack.thread_info.flags = 0;

	/* Some things are best done in assembly.  Finding globals with
	 * a toc is easier in C, so pass in what we can.
	 */
	kexec_sequence(&kexec_stack, image->start, image,
			page_address(image->control_code_page),
			ppc_md.hpte_clear_all);
	/* NOTREACHED */
}

/* Values we need to export to the second kernel via the device tree. */
static unsigned long htab_base;

static struct property htab_base_prop = {
	.name = "linux,htab-base",
	.length = sizeof(unsigned long),
	.value = &htab_base,
};

static struct property htab_size_prop = {
	.name = "linux,htab-size",
	.length = sizeof(unsigned long),
	.value = &htab_size_bytes,
};

static int __init export_htab_values(void)
{
	struct device_node *node;
	struct property *prop;

	/* On machines with no htab htab_address is NULL */
	if (!htab_address)
		return -ENODEV;

	node = of_find_node_by_path("/chosen");
	if (!node)
		return -ENODEV;

	/* remove any stale propertys so ours can be found */
	prop = of_find_property(node, htab_base_prop.name, NULL);
	if (prop)
		prom_remove_property(node, prop);
	prop = of_find_property(node, htab_size_prop.name, NULL);
	if (prop)
		prom_remove_property(node, prop);

	htab_base = __pa(htab_address);
	prom_add_property(node, &htab_base_prop);
	prom_add_property(node, &htab_size_prop);

	of_node_put(node);
	return 0;
}
late_initcall(export_htab_values);
Commit	Line	Data
fce0d574	1	/*
3d1229d6	2	* PPC64 code to handle Linux booting another kernel.
fce0d574 S	3	*
	4	* Copyright (C) 2004-2005, IBM Corp.
	5	*
	6	* Created by: Milton D Miller II
	7	*
	8	* This source code is licensed under the GNU General Public License,
	9	* Version 2. See the file COPYING for more details.
	10	*/
	11
	12
fce0d574 S	13	#include <linux/kexec.h>
	14	#include <linux/smp.h>
	15	#include <linux/thread_info.h>
d200c922	16	#include <linux/init_task.h>
fce0d574 S	17	#include <linux/errno.h>
	18
	19	#include <asm/page.h>
	20	#include <asm/current.h>
	21	#include <asm/machdep.h>
	22	#include <asm/cacheflush.h>
	23	#include <asm/paca.h>
	24	#include <asm/mmu.h>
	25	#include <asm/sections.h> /* _end */
	26	#include <asm/prom.h>
2249ca9d	27	#include <asm/smp.h>
fce0d574	28
3d1229d6	29	int default_machine_kexec_prepare(struct kimage *image)
fce0d574 S	30	{
	31	int i;
	32	unsigned long begin, end; /* limits of segment */
	33	unsigned long low, high; /* limits of blocked memory range */
	34	struct device_node *node;
a7f67bdf JK	35	const unsigned long *basep;
a7f67bdf JK	36	const unsigned int *sizep;
fce0d574 S	37
	38	if (!ppc_md.hpte_clear_all)
	39	return -ENOENT;
	40
	41	/*
	42	* Since we use the kernel fault handlers and paging code to
	43	* handle the virtual mode, we must make sure no destination
	44	* overlaps kernel static data or bss.
	45	*/
72414d3f	46	for (i = 0; i < image->nr_segments; i++)
fce0d574 S	47	if (image->segment[i].mem < __pa(_end))
	48	return -ETXTBSY;
	49
	50	/*
	51	* For non-LPAR, we absolutely can not overwrite the mmu hash
	52	* table, since we are still using the bolted entries in it to
	53	* do the copy. Check that here.
	54	*
	55	* It is safe if the end is below the start of the blocked
	56	* region (end <= low), or if the beginning is after the
	57	* end of the blocked region (begin >= high). Use the
	58	* boolean identity !(a \|\| b) === (!a && !b).
	59	*/
	60	if (htab_address) {
	61	low = __pa(htab_address);
337a7128	62	high = low + htab_size_bytes;
fce0d574	63
72414d3f	64	for (i = 0; i < image->nr_segments; i++) {
fce0d574 S	65	begin = image->segment[i].mem;
	66	end = begin + image->segment[i].memsz;
	67
	68	if ((begin < high) && (end > low))
	69	return -ETXTBSY;
	70	}
	71	}
	72
	73	/* We also should not overwrite the tce tables */
	74	for (node = of_find_node_by_type(NULL, "pci"); node != NULL;
	75	node = of_find_node_by_type(node, "pci")) {
e2eb6392 SR	76	basep = of_get_property(node, "linux,tce-base", NULL);
e2eb6392 SR	77	sizep = of_get_property(node, "linux,tce-size", NULL);
fce0d574 S	78	if (basep == NULL \|\| sizep == NULL)
	79	continue;
	80
	81	low = *basep;
	82	high = low + (*sizep);
	83
72414d3f	84	for (i = 0; i < image->nr_segments; i++) {
fce0d574 S	85	begin = image->segment[i].mem;
	86	end = begin + image->segment[i].memsz;
	87
	88	if ((begin < high) && (end > low))
	89	return -ETXTBSY;
	90	}
	91	}
	92
	93	return 0;
	94	}
	95
fce0d574 S	96	#define IND_FLAGS (IND_DESTINATION \| IND_INDIRECTION \| IND_DONE \| IND_SOURCE)
	97
	98	static void copy_segments(unsigned long ind)
	99	{
	100	unsigned long entry;
	101	unsigned long *ptr;
	102	void *dest;
	103	void *addr;
	104
	105	/*
	106	* We rely on kexec_load to create a lists that properly
	107	* initializes these pointers before they are used.
	108	* We will still crash if the list is wrong, but at least
	109	* the compiler will be quiet.
	110	*/
	111	ptr = NULL;
	112	dest = NULL;
	113
	114	for (entry = ind; !(entry & IND_DONE); entry = *ptr++) {
	115	addr = __va(entry & PAGE_MASK);
	116
	117	switch (entry & IND_FLAGS) {
	118	case IND_DESTINATION:
	119	dest = addr;
	120	break;
	121	case IND_INDIRECTION:
	122	ptr = addr;
	123	break;
	124	case IND_SOURCE:
	125	copy_page(dest, addr);
	126	dest += PAGE_SIZE;
	127	}
	128	}
	129	}
	130
	131	void kexec_copy_flush(struct kimage *image)
	132	{
	133	long i, nr_segments = image->nr_segments;
	134	struct kexec_segment ranges[KEXEC_SEGMENT_MAX];
	135
	136	/* save the ranges on the stack to efficiently flush the icache */
	137	memcpy(ranges, image->segment, sizeof(ranges));
	138
	139	/*
	140	* After this call we may not use anything allocated in dynamic
	141	* memory, including *image.
	142	*
	143	* Only globals and the stack are allowed.
	144	*/
	145	copy_segments(image->head);
	146
	147	/*
	148	* we need to clear the icache for all dest pages sometime,
	149	* including ones that were in place on the original copy
	150	*/
	151	for (i = 0; i < nr_segments; i++)
b5666f70 ME	152	flush_icache_range((unsigned long)__va(ranges[i].mem),
b5666f70 ME	153	(unsigned long)__va(ranges[i].mem + ranges[i].memsz));
fce0d574 S	154	}
	155
	156	#ifdef CONFIG_SMP
	157
1fc711f7 MN	158	static int kexec_all_irq_disabled = 0;
1fc711f7 MN	159
1c21a293	160	static void kexec_smp_down(void *arg)
fce0d574	161	{
1fc711f7 MN	162	local_irq_disable();
	163	mb(); /* make sure our irqs are disabled before we say they are */
	164	get_paca()->kexec_state = KEXEC_STATE_IRQS_OFF;
	165	while(kexec_all_irq_disabled == 0)
	166	cpu_relax();
	167	mb(); /* make sure all irqs are disabled before this */
	168	/*
	169	* Now every CPU has IRQs off, we can clear out any pending
	170	* IPIs and be sure that no more will come in after this.
	171	*/
c5e24354 ME	172	if (ppc_md.kexec_cpu_down)
c5e24354 ME	173	ppc_md.kexec_cpu_down(0, 1);
fce0d574	174
fce0d574 S	175	kexec_smp_wait();
	176	/* NOTREACHED */
	177	}
	178
1fc711f7	179	static void kexec_prepare_cpus_wait(int wait_state)
fce0d574 S	180	{
	181	int my_cpu, i, notified=-1;
	182
fce0d574	183	my_cpu = get_cpu();
1fc711f7	184	/* Make sure each CPU has atleast made it to the state we need */
b636f137	185	for_each_online_cpu(i) {
fce0d574 S	186	if (i == my_cpu)
	187	continue;
	188
1fc711f7	189	while (paca[i].kexec_state < wait_state) {
b3ca8093	190	barrier();
fce0d574 S	191	if (i != notified) {
fce0d574 S	192	printk( "kexec: waiting for cpu %d (physical"
1fc711f7 MN	193	" %d) to enter %i state\n",
1fc711f7 MN	194	i, paca[i].hw_cpu_id, wait_state);
fce0d574 S	195	notified = i;
	196	}
	197	}
	198	}
1fc711f7 MN	199	mb();
	200	}
	201
	202	static void kexec_prepare_cpus(void)
	203	{
	204
	205	smp_call_function(kexec_smp_down, NULL, /* wait */0);
	206	local_irq_disable();
	207	mb(); /* make sure IRQs are disabled before we say they are */
	208	get_paca()->kexec_state = KEXEC_STATE_IRQS_OFF;
	209
	210	kexec_prepare_cpus_wait(KEXEC_STATE_IRQS_OFF);
	211	/* we are sure every CPU has IRQs off at this point */
	212	kexec_all_irq_disabled = 1;
fce0d574 S	213
fce0d574 S	214	/* after we tell the others to go down */
c5e24354 ME	215	if (ppc_md.kexec_cpu_down)
c5e24354 ME	216	ppc_md.kexec_cpu_down(0, 0);
fce0d574	217
1fc711f7 MN	218	/* Before removing MMU mapings make sure all CPUs have entered real mode */
1fc711f7 MN	219	kexec_prepare_cpus_wait(KEXEC_STATE_REAL_MODE);
fce0d574	220
1fc711f7	221	put_cpu();
fce0d574 S	222	}
	223
	224	#else /* ! SMP */
	225
	226	static void kexec_prepare_cpus(void)
	227	{
	228	/*
	229	* move the secondarys to us so that we can copy
	230	* the new kernel 0-0x100 safely
	231	*
	232	* do this if kexec in setup.c ?
75eedfed OJ	233	*
	234	* We need to release the cpus if we are ever going from an
	235	* UP to an SMP kernel.
fce0d574	236	*/
75eedfed	237	smp_release_cpus();
c5e24354 ME	238	if (ppc_md.kexec_cpu_down)
c5e24354 ME	239	ppc_md.kexec_cpu_down(0, 0);
fce0d574 S	240	local_irq_disable();
	241	}
	242
	243	#endif /* SMP */
	244
	245	/*
	246	* kexec thread structure and stack.
	247	*
	248	* We need to make sure that this is 16384-byte aligned due to the
	249	* way process stacks are handled. It also must be statically allocated
	250	* or allocated as part of the kimage, because everything else may be
	251	* overwritten when we copy the kexec image. We piggyback on the
	252	* "init_task" linker section here to statically allocate a stack.
	253	*
	254	* We could use a smaller stack if we don't care about anything using
	255	* current, but that audit has not been performed.
	256	*/
d200c922 JP	257	static union thread_union kexec_stack __init_task_data =
d200c922 JP	258	{ };
fce0d574 S	259
	260	/* Our assembly helper, in kexec_stub.S */
	261	extern NORET_TYPE void kexec_sequence(void *newstack, unsigned long start,
72414d3f	262	void image, void control,
1767c8f3	263	void (*clear_all)(void)) ATTRIB_NORET;
fce0d574 S	264
fce0d574 S	265	/* too late to fail here */
3d1229d6	266	void default_machine_kexec(struct kimage *image)
fce0d574	267	{
fce0d574 S	268	/* prepare control code if any */
fce0d574 S	269
cc532915 ME	270	/*
	271	* If the kexec boot is the normal one, need to shutdown other cpus
	272	* into our wait loop and quiesce interrupts.
	273	* Otherwise, in the case of crashed mode (crashing_cpu >= 0),
	274	* stopping other CPUs and collecting their pt_regs is done before
	275	* using debugger IPI.
	276	*/
	277
54622f10 MK	278	if (crashing_cpu == -1)
54622f10 MK	279	kexec_prepare_cpus();
fce0d574 S	280
	281	/* switch to a staticly allocated stack. Based on irq stack code.
	282	* XXX: the task struct will likely be invalid once we do the copy!
	283	*/
	284	kexec_stack.thread_info.task = current_thread_info()->task;
	285	kexec_stack.thread_info.flags = 0;
	286
	287	/* Some things are best done in assembly. Finding globals with
	288	* a toc is easier in C, so pass in what we can.
	289	*/
	290	kexec_sequence(&kexec_stack, image->start, image,
	291	page_address(image->control_code_page),
1767c8f3	292	ppc_md.hpte_clear_all);
fce0d574 S	293	/* NOTREACHED */
fce0d574 S	294	}
593e537b ME	295
593e537b ME	296	/* Values we need to export to the second kernel via the device tree. */
2e8e4f5b	297	static unsigned long htab_base;
593e537b ME	298
	299	static struct property htab_base_prop = {
	300	.name = "linux,htab-base",
	301	.length = sizeof(unsigned long),
1a38147e	302	.value = &htab_base,
593e537b ME	303	};
	304
	305	static struct property htab_size_prop = {
	306	.name = "linux,htab-size",
	307	.length = sizeof(unsigned long),
1a38147e	308	.value = &htab_size_bytes,
593e537b ME	309	};
593e537b ME	310
6f29c329	311	static int __init export_htab_values(void)
593e537b ME	312	{
593e537b ME	313	struct device_node *node;
ed7b2144	314	struct property *prop;
593e537b	315
2e8e4f5b DF	316	/* On machines with no htab htab_address is NULL */
2e8e4f5b DF	317	if (!htab_address)
6f29c329	318	return -ENODEV;
2e8e4f5b	319
593e537b ME	320	node = of_find_node_by_path("/chosen");
593e537b ME	321	if (!node)
6f29c329	322	return -ENODEV;
593e537b	323
ed7b2144	324	/* remove any stale propertys so ours can be found */
ed7b2144 MM	325	prop = of_find_property(node, htab_base_prop.name, NULL);
	326	if (prop)
	327	prom_remove_property(node, prop);
	328	prop = of_find_property(node, htab_size_prop.name, NULL);
	329	if (prop)
	330	prom_remove_property(node, prop);
	331
593e537b ME	332	htab_base = __pa(htab_address);
593e537b ME	333	prom_add_property(node, &htab_base_prop);
593e537b ME	334	prom_add_property(node, &htab_size_prop);
593e537b ME	335
593e537b	336	of_node_put(node);
aa98c50d	337	return 0;
35dd5432	338	}
6f29c329	339	late_initcall(export_htab_values);