[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/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

/* FIXME: we should schedule this function to be called on all cpus based
 * on calling the interrupts, but we would like to call it off irq level
 * so that the interrupt controller is clean.
 */
static void kexec_smp_down(void *arg)
{
	if (ppc_md.kexec_cpu_down)
		ppc_md.kexec_cpu_down(0, 1);

	local_irq_disable();
	kexec_smp_wait();
	/* NOTREACHED */
}

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

	smp_call_function(kexec_smp_down, NULL, /* wait */0);
	my_cpu = get_cpu();

	/* check the others cpus are now down (via paca hw cpu id == -1) */
	for (i=0; i < NR_CPUS; i++) {
		if (i == my_cpu)
			continue;

		while (paca[i].hw_cpu_id != -1) {
			barrier();
			if (!cpu_possible(i)) {
				printk("kexec: cpu %d hw_cpu_id %d is not"
						" possible, ignoring\n",
						i, paca[i].hw_cpu_id);
				break;
			}
			if (!cpu_online(i)) {
				/* Fixme: this can be spinning in
				 * pSeries_secondary_wait with a paca
				 * waiting for it to go online.
				 */
				printk("kexec: cpu %d hw_cpu_id %d is not"
						" online, ignoring\n",
						i, paca[i].hw_cpu_id);
				break;
			}
			if (i != notified) {
				printk( "kexec: waiting for cpu %d (physical"
						" %d) to go down\n",
						i, paca[i].hw_cpu_id);
				notified = i;
			}
		}
	}

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

	put_cpu();

	local_irq_disable();
}

#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
	__attribute__((__section__(".data.init_task"))) = { };

/* 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, kernel_end;

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 struct property kernel_end_prop = {
	.name = "linux,kernel-end",
	.length = sizeof(unsigned long),
	.value = &kernel_end,
};

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

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

	kernel_end = __pa(_end);
	prom_add_property(node, &kernel_end_prop);

	/* On machines with no htab htab_address is NULL */
	if (NULL == htab_address)
		goto out;

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

 out:
	of_node_put(node);
}

static struct property crashk_base_prop = {
	.name = "linux,crashkernel-base",
	.length = sizeof(unsigned long),
	.value = &crashk_res.start,
};

static unsigned long crashk_size;

static struct property crashk_size_prop = {
	.name = "linux,crashkernel-size",
	.length = sizeof(unsigned long),
	.value = &crashk_size,
};

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

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

	/* There might be existing crash kernel properties, but we can't
	 * be sure what's in them, so remove them. */
	prop = of_find_property(node, "linux,crashkernel-base", NULL);
	if (prop)
		prom_remove_property(node, prop);

	prop = of_find_property(node, "linux,crashkernel-size", NULL);
	if (prop)
		prom_remove_property(node, prop);

	if (crashk_res.start != 0) {
		prom_add_property(node, &crashk_base_prop);
		crashk_size = crashk_res.end - crashk_res.start + 1;
		prom_add_property(node, &crashk_size_prop);
	}

	of_node_put(node);
}

static int __init kexec_setup(void)
{
	export_htab_values();
	export_crashk_values();
	return 0;
}
__initcall(kexec_setup);
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>
	16	#include <linux/errno.h>
	17
	18	#include <asm/page.h>
	19	#include <asm/current.h>
	20	#include <asm/machdep.h>
	21	#include <asm/cacheflush.h>
	22	#include <asm/paca.h>
	23	#include <asm/mmu.h>
	24	#include <asm/sections.h> /* _end */
	25	#include <asm/prom.h>
2249ca9d	26	#include <asm/smp.h>
fce0d574	27
3d1229d6	28	int default_machine_kexec_prepare(struct kimage *image)
fce0d574 S	29	{
	30	int i;
	31	unsigned long begin, end; /* limits of segment */
	32	unsigned long low, high; /* limits of blocked memory range */
	33	struct device_node *node;
a7f67bdf JK	34	const unsigned long *basep;
a7f67bdf JK	35	const unsigned int *sizep;
fce0d574 S	36
	37	if (!ppc_md.hpte_clear_all)
	38	return -ENOENT;
	39
	40	/*
	41	* Since we use the kernel fault handlers and paging code to
	42	* handle the virtual mode, we must make sure no destination
	43	* overlaps kernel static data or bss.
	44	*/
72414d3f	45	for (i = 0; i < image->nr_segments; i++)
fce0d574 S	46	if (image->segment[i].mem < __pa(_end))
	47	return -ETXTBSY;
	48
	49	/*
	50	* For non-LPAR, we absolutely can not overwrite the mmu hash
	51	* table, since we are still using the bolted entries in it to
	52	* do the copy. Check that here.
	53	*
	54	* It is safe if the end is below the start of the blocked
	55	* region (end <= low), or if the beginning is after the
	56	* end of the blocked region (begin >= high). Use the
	57	* boolean identity !(a \|\| b) === (!a && !b).
	58	*/
	59	if (htab_address) {
	60	low = __pa(htab_address);
337a7128	61	high = low + htab_size_bytes;
fce0d574	62
72414d3f	63	for (i = 0; i < image->nr_segments; i++) {
fce0d574 S	64	begin = image->segment[i].mem;
	65	end = begin + image->segment[i].memsz;
	66
	67	if ((begin < high) && (end > low))
	68	return -ETXTBSY;
	69	}
	70	}
	71
	72	/* We also should not overwrite the tce tables */
	73	for (node = of_find_node_by_type(NULL, "pci"); node != NULL;
	74	node = of_find_node_by_type(node, "pci")) {
e2eb6392 SR	75	basep = of_get_property(node, "linux,tce-base", NULL);
e2eb6392 SR	76	sizep = of_get_property(node, "linux,tce-size", NULL);
fce0d574 S	77	if (basep == NULL \|\| sizep == NULL)
	78	continue;
	79
	80	low = *basep;
	81	high = low + (*sizep);
	82
72414d3f	83	for (i = 0; i < image->nr_segments; i++) {
fce0d574 S	84	begin = image->segment[i].mem;
	85	end = begin + image->segment[i].memsz;
	86
	87	if ((begin < high) && (end > low))
	88	return -ETXTBSY;
	89	}
	90	}
	91
	92	return 0;
	93	}
	94
fce0d574 S	95	#define IND_FLAGS (IND_DESTINATION \| IND_INDIRECTION \| IND_DONE \| IND_SOURCE)
	96
	97	static void copy_segments(unsigned long ind)
	98	{
	99	unsigned long entry;
	100	unsigned long *ptr;
	101	void *dest;
	102	void *addr;
	103
	104	/*
	105	* We rely on kexec_load to create a lists that properly
	106	* initializes these pointers before they are used.
	107	* We will still crash if the list is wrong, but at least
	108	* the compiler will be quiet.
	109	*/
	110	ptr = NULL;
	111	dest = NULL;
	112
	113	for (entry = ind; !(entry & IND_DONE); entry = *ptr++) {
	114	addr = __va(entry & PAGE_MASK);
	115
	116	switch (entry & IND_FLAGS) {
	117	case IND_DESTINATION:
	118	dest = addr;
	119	break;
	120	case IND_INDIRECTION:
	121	ptr = addr;
	122	break;
	123	case IND_SOURCE:
	124	copy_page(dest, addr);
	125	dest += PAGE_SIZE;
	126	}
	127	}
	128	}
	129
	130	void kexec_copy_flush(struct kimage *image)
	131	{
	132	long i, nr_segments = image->nr_segments;
	133	struct kexec_segment ranges[KEXEC_SEGMENT_MAX];
	134
	135	/* save the ranges on the stack to efficiently flush the icache */
	136	memcpy(ranges, image->segment, sizeof(ranges));
	137
	138	/*
	139	* After this call we may not use anything allocated in dynamic
	140	* memory, including *image.
	141	*
	142	* Only globals and the stack are allowed.
	143	*/
	144	copy_segments(image->head);
	145
	146	/*
	147	* we need to clear the icache for all dest pages sometime,
	148	* including ones that were in place on the original copy
	149	*/
	150	for (i = 0; i < nr_segments; i++)
b5666f70 ME	151	flush_icache_range((unsigned long)__va(ranges[i].mem),
b5666f70 ME	152	(unsigned long)__va(ranges[i].mem + ranges[i].memsz));
fce0d574 S	153	}
	154
	155	#ifdef CONFIG_SMP
	156
	157	/* FIXME: we should schedule this function to be called on all cpus based
	158	* on calling the interrupts, but we would like to call it off irq level
	159	* so that the interrupt controller is clean.
	160	*/
1c21a293	161	static void kexec_smp_down(void *arg)
fce0d574	162	{
c5e24354 ME	163	if (ppc_md.kexec_cpu_down)
c5e24354 ME	164	ppc_md.kexec_cpu_down(0, 1);
fce0d574 S	165
	166	local_irq_disable();
	167	kexec_smp_wait();
	168	/* NOTREACHED */
	169	}
	170
	171	static void kexec_prepare_cpus(void)
	172	{
	173	int my_cpu, i, notified=-1;
	174
392096e9	175	smp_call_function(kexec_smp_down, NULL, /* wait */0);
fce0d574 S	176	my_cpu = get_cpu();
	177
	178	/* check the others cpus are now down (via paca hw cpu id == -1) */
	179	for (i=0; i < NR_CPUS; i++) {
	180	if (i == my_cpu)
	181	continue;
	182
	183	while (paca[i].hw_cpu_id != -1) {
b3ca8093	184	barrier();
fce0d574 S	185	if (!cpu_possible(i)) {
	186	printk("kexec: cpu %d hw_cpu_id %d is not"
	187	" possible, ignoring\n",
	188	i, paca[i].hw_cpu_id);
	189	break;
	190	}
	191	if (!cpu_online(i)) {
	192	/* Fixme: this can be spinning in
	193	* pSeries_secondary_wait with a paca
	194	* waiting for it to go online.
	195	*/
	196	printk("kexec: cpu %d hw_cpu_id %d is not"
	197	" online, ignoring\n",
	198	i, paca[i].hw_cpu_id);
	199	break;
	200	}
	201	if (i != notified) {
	202	printk( "kexec: waiting for cpu %d (physical"
	203	" %d) to go down\n",
	204	i, paca[i].hw_cpu_id);
	205	notified = i;
	206	}
	207	}
	208	}
	209
	210	/* after we tell the others to go down */
c5e24354 ME	211	if (ppc_md.kexec_cpu_down)
c5e24354 ME	212	ppc_md.kexec_cpu_down(0, 0);
fce0d574 S	213
	214	put_cpu();
	215
	216	local_irq_disable();
	217	}
	218
	219	#else /* ! SMP */
	220
	221	static void kexec_prepare_cpus(void)
	222	{
	223	/*
	224	* move the secondarys to us so that we can copy
	225	* the new kernel 0-0x100 safely
	226	*
	227	* do this if kexec in setup.c ?
75eedfed OJ	228	*
	229	* We need to release the cpus if we are ever going from an
	230	* UP to an SMP kernel.
fce0d574	231	*/
75eedfed	232	smp_release_cpus();
c5e24354 ME	233	if (ppc_md.kexec_cpu_down)
c5e24354 ME	234	ppc_md.kexec_cpu_down(0, 0);
fce0d574 S	235	local_irq_disable();
	236	}
	237
	238	#endif /* SMP */
	239
	240	/*
	241	* kexec thread structure and stack.
	242	*
	243	* We need to make sure that this is 16384-byte aligned due to the
	244	* way process stacks are handled. It also must be statically allocated
	245	* or allocated as part of the kimage, because everything else may be
	246	* overwritten when we copy the kexec image. We piggyback on the
	247	* "init_task" linker section here to statically allocate a stack.
	248	*
	249	* We could use a smaller stack if we don't care about anything using
	250	* current, but that audit has not been performed.
	251	*/
1c21a293	252	static union thread_union kexec_stack
fce0d574 S	253	__attribute__((__section__(".data.init_task"))) = { };
	254
	255	/* Our assembly helper, in kexec_stub.S */
	256	extern NORET_TYPE void kexec_sequence(void *newstack, unsigned long start,
72414d3f MS	257	void image, void control,
72414d3f MS	258	void (*clear_all)(void)) ATTRIB_NORET;
fce0d574 S	259
fce0d574 S	260	/* too late to fail here */
3d1229d6	261	void default_machine_kexec(struct kimage *image)
fce0d574	262	{
fce0d574 S	263	/* prepare control code if any */
fce0d574 S	264
cc532915 ME	265	/*
	266	* If the kexec boot is the normal one, need to shutdown other cpus
	267	* into our wait loop and quiesce interrupts.
	268	* Otherwise, in the case of crashed mode (crashing_cpu >= 0),
	269	* stopping other CPUs and collecting their pt_regs is done before
	270	* using debugger IPI.
	271	*/
	272
	273	if (crashing_cpu == -1)
	274	kexec_prepare_cpus();
fce0d574 S	275
	276	/* switch to a staticly allocated stack. Based on irq stack code.
	277	* XXX: the task struct will likely be invalid once we do the copy!
	278	*/
	279	kexec_stack.thread_info.task = current_thread_info()->task;
	280	kexec_stack.thread_info.flags = 0;
	281
	282	/* Some things are best done in assembly. Finding globals with
	283	* a toc is easier in C, so pass in what we can.
	284	*/
	285	kexec_sequence(&kexec_stack, image->start, image,
	286	page_address(image->control_code_page),
	287	ppc_md.hpte_clear_all);
	288	/* NOTREACHED */
	289	}
593e537b ME	290
593e537b ME	291	/* Values we need to export to the second kernel via the device tree. */
337a7128	292	static unsigned long htab_base, kernel_end;
593e537b ME	293
	294	static struct property htab_base_prop = {
	295	.name = "linux,htab-base",
	296	.length = sizeof(unsigned long),
1a38147e	297	.value = &htab_base,
593e537b ME	298	};
	299
	300	static struct property htab_size_prop = {
	301	.name = "linux,htab-size",
	302	.length = sizeof(unsigned long),
1a38147e	303	.value = &htab_size_bytes,
593e537b ME	304	};
	305
	306	static struct property kernel_end_prop = {
	307	.name = "linux,kernel-end",
	308	.length = sizeof(unsigned long),
1a38147e	309	.value = &kernel_end,
593e537b ME	310	};
	311
	312	static void __init export_htab_values(void)
	313	{
	314	struct device_node *node;
	315
	316	node = of_find_node_by_path("/chosen");
	317	if (!node)
	318	return;
	319
	320	kernel_end = __pa(_end);
	321	prom_add_property(node, &kernel_end_prop);
	322
	323	/* On machines with no htab htab_address is NULL */
	324	if (NULL == htab_address)
	325	goto out;
	326
	327	htab_base = __pa(htab_address);
	328	prom_add_property(node, &htab_base_prop);
593e537b ME	329	prom_add_property(node, &htab_size_prop);
	330
	331	out:
	332	of_node_put(node);
	333	}
	334
35dd5432 ME	335	static struct property crashk_base_prop = {
	336	.name = "linux,crashkernel-base",
	337	.length = sizeof(unsigned long),
1a38147e	338	.value = &crashk_res.start,
35dd5432 ME	339	};
	340
	341	static unsigned long crashk_size;
	342
	343	static struct property crashk_size_prop = {
	344	.name = "linux,crashkernel-size",
	345	.length = sizeof(unsigned long),
1a38147e	346	.value = &crashk_size,
35dd5432 ME	347	};
	348
	349	static void __init export_crashk_values(void)
	350	{
	351	struct device_node *node;
	352	struct property *prop;
	353
	354	node = of_find_node_by_path("/chosen");
	355	if (!node)
	356	return;
	357
	358	/* There might be existing crash kernel properties, but we can't
	359	* be sure what's in them, so remove them. */
	360	prop = of_find_property(node, "linux,crashkernel-base", NULL);
	361	if (prop)
	362	prom_remove_property(node, prop);
	363
	364	prop = of_find_property(node, "linux,crashkernel-size", NULL);
	365	if (prop)
	366	prom_remove_property(node, prop);
	367
	368	if (crashk_res.start != 0) {
	369	prom_add_property(node, &crashk_base_prop);
	370	crashk_size = crashk_res.end - crashk_res.start + 1;
	371	prom_add_property(node, &crashk_size_prop);
	372	}
	373
	374	of_node_put(node);
	375	}
	376
aa98c50d	377	static int __init kexec_setup(void)
593e537b ME	378	{
593e537b ME	379	export_htab_values();
35dd5432	380	export_crashk_values();
aa98c50d	381	return 0;
35dd5432	382	}
aa98c50d	383	__initcall(kexec_setup);