[deliverable/linux.git] / arch / x86_64 / kernel / machine_kexec.c

/*
 * machine_kexec.c - handle transition of Linux booting another kernel
 * Copyright (C) 2002-2005 Eric Biederman  <ebiederm@xmission.com>
 *
 * This source code is licensed under the GNU General Public License,
 * Version 2.  See the file COPYING for more details.
 */

#include <linux/mm.h>
#include <linux/kexec.h>
#include <linux/delay.h>
#include <linux/string.h>
#include <linux/reboot.h>
#include <asm/pda.h>
#include <asm/pgtable.h>
#include <asm/pgalloc.h>
#include <asm/tlbflush.h>
#include <asm/mmu_context.h>
#include <asm/io.h>
#include <asm/apic.h>
#include <asm/cpufeature.h>
#include <asm/hw_irq.h>

#define LEVEL0_SIZE (1UL << 12UL)
#define LEVEL1_SIZE (1UL << 21UL)
#define LEVEL2_SIZE (1UL << 30UL)
#define LEVEL3_SIZE (1UL << 39UL)
#define LEVEL4_SIZE (1UL << 48UL)

#define L0_ATTR (_PAGE_PRESENT | _PAGE_RW | _PAGE_ACCESSED | _PAGE_DIRTY)
#define L1_ATTR (_PAGE_PRESENT | _PAGE_RW | _PAGE_ACCESSED | _PAGE_DIRTY | _PAGE_PSE)
#define L2_ATTR (_PAGE_PRESENT | _PAGE_RW | _PAGE_ACCESSED | _PAGE_DIRTY)
#define L3_ATTR (_PAGE_PRESENT | _PAGE_RW | _PAGE_ACCESSED | _PAGE_DIRTY)

static void init_level2_page(u64 *level2p, unsigned long addr)
{
	unsigned long end_addr;

	addr &= PAGE_MASK;
	end_addr = addr + LEVEL2_SIZE;
	while (addr < end_addr) {
		*(level2p++) = addr | L1_ATTR;
		addr += LEVEL1_SIZE;
	}
}

static int init_level3_page(struct kimage *image, u64 *level3p,
				unsigned long addr, unsigned long last_addr)
{
	unsigned long end_addr;
	int result;

	result = 0;
	addr &= PAGE_MASK;
	end_addr = addr + LEVEL3_SIZE;
	while ((addr < last_addr) && (addr < end_addr)) {
		struct page *page;
		u64 *level2p;

		page = kimage_alloc_control_pages(image, 0);
		if (!page) {
			result = -ENOMEM;
			goto out;
		}
		level2p = (u64 *)page_address(page);
		init_level2_page(level2p, addr);
		*(level3p++) = __pa(level2p) | L2_ATTR;
		addr += LEVEL2_SIZE;
	}
	/* clear the unused entries */
	while (addr < end_addr) {
		*(level3p++) = 0;
		addr += LEVEL2_SIZE;
	}
out:
	return result;
}


static int init_level4_page(struct kimage *image, u64 *level4p,
				unsigned long addr, unsigned long last_addr)
{
	unsigned long end_addr;
	int result;

	result = 0;
	addr &= PAGE_MASK;
	end_addr = addr + LEVEL4_SIZE;
	while ((addr < last_addr) && (addr < end_addr)) {
		struct page *page;
		u64 *level3p;

		page = kimage_alloc_control_pages(image, 0);
		if (!page) {
			result = -ENOMEM;
			goto out;
		}
		level3p = (u64 *)page_address(page);
		result = init_level3_page(image, level3p, addr, last_addr);
		if (result) {
			goto out;
		}
		*(level4p++) = __pa(level3p) | L3_ATTR;
		addr += LEVEL3_SIZE;
	}
	/* clear the unused entries */
	while (addr < end_addr) {
		*(level4p++) = 0;
		addr += LEVEL3_SIZE;
	}
out:
	return result;
}


static int init_pgtable(struct kimage *image, unsigned long start_pgtable)
{
	u64 *level4p;
	level4p = (u64 *)__va(start_pgtable);
 	return init_level4_page(image, level4p, 0, end_pfn << PAGE_SHIFT);
}

static void set_idt(void *newidt, u16 limit)
{
	struct desc_ptr curidt;

	/* x86-64 supports unaliged loads & stores */
	curidt.size    = limit;
	curidt.address = (unsigned long)newidt;

	__asm__ __volatile__ (
		"lidtq %0\n"
		: : "m" (curidt)
		);
};


static void set_gdt(void *newgdt, u16 limit)
{
	struct desc_ptr curgdt;

	/* x86-64 supports unaligned loads & stores */
	curgdt.size    = limit;
	curgdt.address = (unsigned long)newgdt;

	__asm__ __volatile__ (
		"lgdtq %0\n"
		: : "m" (curgdt)
		);
};

static void load_segments(void)
{
	__asm__ __volatile__ (
		"\tmovl %0,%%ds\n"
		"\tmovl %0,%%es\n"
		"\tmovl %0,%%ss\n"
		"\tmovl %0,%%fs\n"
		"\tmovl %0,%%gs\n"
		: : "a" (__KERNEL_DS)
		);
}

typedef NORET_TYPE void (*relocate_new_kernel_t)(unsigned long indirection_page,
					unsigned long control_code_buffer,
					unsigned long start_address,
					unsigned long pgtable) ATTRIB_NORET;

const extern unsigned char relocate_new_kernel[];
const extern unsigned long relocate_new_kernel_size;

int machine_kexec_prepare(struct kimage *image)
{
	unsigned long start_pgtable, control_code_buffer;
	int result;

	/* Calculate the offsets */
	start_pgtable = page_to_pfn(image->control_code_page) << PAGE_SHIFT;
	control_code_buffer = start_pgtable + 4096UL;

	/* Setup the identity mapped 64bit page table */
	result = init_pgtable(image, start_pgtable);
	if (result)
		return result;

	/* Place the code in the reboot code buffer */
	memcpy(__va(control_code_buffer), relocate_new_kernel,
						relocate_new_kernel_size);

	return 0;
}

void machine_kexec_cleanup(struct kimage *image)
{
	return;
}

/*
 * Do not allocate memory (or fail in any way) in machine_kexec().
 * We are past the point of no return, committed to rebooting now.
 */
NORET_TYPE void machine_kexec(struct kimage *image)
{
	unsigned long page_list;
	unsigned long control_code_buffer;
	unsigned long start_pgtable;
	relocate_new_kernel_t rnk;

	/* Interrupts aren't acceptable while we reboot */
	local_irq_disable();

	/* Calculate the offsets */
	page_list = image->head;
	start_pgtable = page_to_pfn(image->control_code_page) << PAGE_SHIFT;
	control_code_buffer = start_pgtable + 4096UL;

	/* Set the low half of the page table to my identity mapped
	 * page table for kexec.  Leave the high half pointing at the
	 * kernel pages.   Don't bother to flush the global pages
	 * as that will happen when I fully switch to my identity mapped
	 * page table anyway.
	 */
	memcpy(__va(read_cr3()), __va(start_pgtable), PAGE_SIZE/2);
	__flush_tlb();


	/* The segment registers are funny things, they are
	 * automatically loaded from a table, in memory wherever you
	 * set them to a specific selector, but this table is never
	 * accessed again unless you set the segment to a different selector.
	 *
	 * The more common model are caches where the behide
	 * the scenes work is done, but is also dropped at arbitrary
	 * times.
	 *
	 * I take advantage of this here by force loading the
	 * segments, before I zap the gdt with an invalid value.
	 */
	load_segments();
	/* The gdt & idt are now invalid.
	 * If you want to load them you must set up your own idt & gdt.
	 */
	set_gdt(phys_to_virt(0),0);
	set_idt(phys_to_virt(0),0);
	/* now call it */
	rnk = (relocate_new_kernel_t) control_code_buffer;
	(*rnk)(page_list, control_code_buffer, image->start, start_pgtable);
}
Commit	Line	Data
5234f5eb EB	1	/*
	2	* machine_kexec.c - handle transition of Linux booting another kernel
	3	* Copyright (C) 2002-2005 Eric Biederman <ebiederm@xmission.com>
	4	*
	5	* This source code is licensed under the GNU General Public License,
	6	* Version 2. See the file COPYING for more details.
	7	*/
	8
	9	#include <linux/mm.h>
	10	#include <linux/kexec.h>
	11	#include <linux/delay.h>
	12	#include <linux/string.h>
	13	#include <linux/reboot.h>
	14	#include <asm/pda.h>
	15	#include <asm/pgtable.h>
	16	#include <asm/pgalloc.h>
	17	#include <asm/tlbflush.h>
	18	#include <asm/mmu_context.h>
	19	#include <asm/io.h>
	20	#include <asm/apic.h>
	21	#include <asm/cpufeature.h>
	22	#include <asm/hw_irq.h>
	23
	24	#define LEVEL0_SIZE (1UL << 12UL)
	25	#define LEVEL1_SIZE (1UL << 21UL)
	26	#define LEVEL2_SIZE (1UL << 30UL)
	27	#define LEVEL3_SIZE (1UL << 39UL)
	28	#define LEVEL4_SIZE (1UL << 48UL)
	29
	30	#define L0_ATTR (_PAGE_PRESENT \| _PAGE_RW \| _PAGE_ACCESSED \| _PAGE_DIRTY)
	31	#define L1_ATTR (_PAGE_PRESENT \| _PAGE_RW \| _PAGE_ACCESSED \| _PAGE_DIRTY \| _PAGE_PSE)
	32	#define L2_ATTR (_PAGE_PRESENT \| _PAGE_RW \| _PAGE_ACCESSED \| _PAGE_DIRTY)
	33	#define L3_ATTR (_PAGE_PRESENT \| _PAGE_RW \| _PAGE_ACCESSED \| _PAGE_DIRTY)
	34
72414d3f	35	static void init_level2_page(u64 *level2p, unsigned long addr)
5234f5eb EB	36	{
5234f5eb EB	37	unsigned long end_addr;
72414d3f	38
5234f5eb EB	39	addr &= PAGE_MASK;
5234f5eb EB	40	end_addr = addr + LEVEL2_SIZE;
72414d3f	41	while (addr < end_addr) {
5234f5eb EB	42	*(level2p++) = addr \| L1_ATTR;
	43	addr += LEVEL1_SIZE;
	44	}
	45	}
	46
72414d3f MS	47	static int init_level3_page(struct kimage image, u64 level3p,
72414d3f MS	48	unsigned long addr, unsigned long last_addr)
5234f5eb EB	49	{
	50	unsigned long end_addr;
	51	int result;
72414d3f	52
5234f5eb EB	53	result = 0;
	54	addr &= PAGE_MASK;
	55	end_addr = addr + LEVEL3_SIZE;
72414d3f	56	while ((addr < last_addr) && (addr < end_addr)) {
5234f5eb EB	57	struct page *page;
5234f5eb EB	58	u64 *level2p;
72414d3f	59
5234f5eb EB	60	page = kimage_alloc_control_pages(image, 0);
	61	if (!page) {
	62	result = -ENOMEM;
	63	goto out;
	64	}
	65	level2p = (u64 *)page_address(page);
	66	init_level2_page(level2p, addr);
	67	*(level3p++) = __pa(level2p) \| L2_ATTR;
	68	addr += LEVEL2_SIZE;
	69	}
	70	/* clear the unused entries */
72414d3f	71	while (addr < end_addr) {
5234f5eb EB	72	*(level3p++) = 0;
	73	addr += LEVEL2_SIZE;
	74	}
	75	out:
	76	return result;
	77	}
	78
	79
72414d3f MS	80	static int init_level4_page(struct kimage image, u64 level4p,
72414d3f MS	81	unsigned long addr, unsigned long last_addr)
5234f5eb EB	82	{
	83	unsigned long end_addr;
	84	int result;
72414d3f	85
5234f5eb EB	86	result = 0;
	87	addr &= PAGE_MASK;
	88	end_addr = addr + LEVEL4_SIZE;
72414d3f	89	while ((addr < last_addr) && (addr < end_addr)) {
5234f5eb EB	90	struct page *page;
5234f5eb EB	91	u64 *level3p;
72414d3f	92
5234f5eb EB	93	page = kimage_alloc_control_pages(image, 0);
	94	if (!page) {
	95	result = -ENOMEM;
	96	goto out;
	97	}
	98	level3p = (u64 *)page_address(page);
	99	result = init_level3_page(image, level3p, addr, last_addr);
	100	if (result) {
	101	goto out;
	102	}
	103	*(level4p++) = __pa(level3p) \| L3_ATTR;
	104	addr += LEVEL3_SIZE;
	105	}
	106	/* clear the unused entries */
72414d3f	107	while (addr < end_addr) {
5234f5eb EB	108	*(level4p++) = 0;
	109	addr += LEVEL3_SIZE;
	110	}
72414d3f	111	out:
5234f5eb EB	112	return result;
	113	}
	114
	115
	116	static int init_pgtable(struct kimage *image, unsigned long start_pgtable)
	117	{
	118	u64 *level4p;
	119	level4p = (u64 *)__va(start_pgtable);
72414d3f	120	return init_level4_page(image, level4p, 0, end_pfn << PAGE_SHIFT);
5234f5eb EB	121	}
	122
	123	static void set_idt(void *newidt, u16 limit)
	124	{
36c4fd23	125	struct desc_ptr curidt;
5234f5eb EB	126
5234f5eb EB	127	/* x86-64 supports unaliged loads & stores */
36c4fd23 EB	128	curidt.size = limit;
36c4fd23 EB	129	curidt.address = (unsigned long)newidt;
5234f5eb EB	130
5234f5eb EB	131	__asm__ __volatile__ (
36c4fd23 EB	132	"lidtq %0\n"
36c4fd23 EB	133	: : "m" (curidt)
5234f5eb EB	134	);
	135	};
	136
	137
	138	static void set_gdt(void *newgdt, u16 limit)
	139	{
36c4fd23	140	struct desc_ptr curgdt;
5234f5eb EB	141
5234f5eb EB	142	/* x86-64 supports unaligned loads & stores */
36c4fd23 EB	143	curgdt.size = limit;
36c4fd23 EB	144	curgdt.address = (unsigned long)newgdt;
5234f5eb EB	145
5234f5eb EB	146	__asm__ __volatile__ (
36c4fd23 EB	147	"lgdtq %0\n"
36c4fd23 EB	148	: : "m" (curgdt)
5234f5eb EB	149	);
	150	};
	151
	152	static void load_segments(void)
	153	{
	154	__asm__ __volatile__ (
36c4fd23 EB	155	"\tmovl %0,%%ds\n"
	156	"\tmovl %0,%%es\n"
	157	"\tmovl %0,%%ss\n"
	158	"\tmovl %0,%%fs\n"
	159	"\tmovl %0,%%gs\n"
	160	: : "a" (__KERNEL_DS)
5234f5eb	161	);
5234f5eb EB	162	}
5234f5eb EB	163
72414d3f MS	164	typedef NORET_TYPE void (*relocate_new_kernel_t)(unsigned long indirection_page,
	165	unsigned long control_code_buffer,
	166	unsigned long start_address,
	167	unsigned long pgtable) ATTRIB_NORET;
5234f5eb EB	168
	169	const extern unsigned char relocate_new_kernel[];
	170	const extern unsigned long relocate_new_kernel_size;
	171
	172	int machine_kexec_prepare(struct kimage *image)
	173	{
	174	unsigned long start_pgtable, control_code_buffer;
	175	int result;
	176
	177	/* Calculate the offsets */
72414d3f	178	start_pgtable = page_to_pfn(image->control_code_page) << PAGE_SHIFT;
5234f5eb EB	179	control_code_buffer = start_pgtable + 4096UL;
	180
	181	/* Setup the identity mapped 64bit page table */
	182	result = init_pgtable(image, start_pgtable);
72414d3f	183	if (result)
5234f5eb	184	return result;
5234f5eb EB	185
5234f5eb EB	186	/* Place the code in the reboot code buffer */
72414d3f MS	187	memcpy(__va(control_code_buffer), relocate_new_kernel,
72414d3f MS	188	relocate_new_kernel_size);
5234f5eb EB	189
	190	return 0;
	191	}
	192
	193	void machine_kexec_cleanup(struct kimage *image)
	194	{
	195	return;
	196	}
	197
	198	/*
	199	* Do not allocate memory (or fail in any way) in machine_kexec().
	200	* We are past the point of no return, committed to rebooting now.
	201	*/
	202	NORET_TYPE void machine_kexec(struct kimage *image)
	203	{
	204	unsigned long page_list;
	205	unsigned long control_code_buffer;
	206	unsigned long start_pgtable;
	207	relocate_new_kernel_t rnk;
	208
	209	/* Interrupts aren't acceptable while we reboot */
	210	local_irq_disable();
	211
	212	/* Calculate the offsets */
72414d3f MS	213	page_list = image->head;
72414d3f MS	214	start_pgtable = page_to_pfn(image->control_code_page) << PAGE_SHIFT;
5234f5eb EB	215	control_code_buffer = start_pgtable + 4096UL;
	216
	217	/* Set the low half of the page table to my identity mapped
	218	* page table for kexec. Leave the high half pointing at the
	219	* kernel pages. Don't bother to flush the global pages
	220	* as that will happen when I fully switch to my identity mapped
	221	* page table anyway.
	222	*/
	223	memcpy(__va(read_cr3()), __va(start_pgtable), PAGE_SIZE/2);
	224	__flush_tlb();
	225
	226
	227	/* The segment registers are funny things, they are
	228	* automatically loaded from a table, in memory wherever you
	229	* set them to a specific selector, but this table is never
	230	* accessed again unless you set the segment to a different selector.
	231	*
	232	* The more common model are caches where the behide
	233	* the scenes work is done, but is also dropped at arbitrary
	234	* times.
	235	*
	236	* I take advantage of this here by force loading the
	237	* segments, before I zap the gdt with an invalid value.
	238	*/
	239	load_segments();
	240	/* The gdt & idt are now invalid.
	241	* If you want to load them you must set up your own idt & gdt.
	242	*/
	243	set_gdt(phys_to_virt(0),0);
	244	set_idt(phys_to_virt(0),0);
	245	/* now call it */
	246	rnk = (relocate_new_kernel_t) control_code_buffer;
	247	(*rnk)(page_list, control_code_buffer, image->start, start_pgtable);
	248	}