machine_kexec_64.c 6.5 KB
Newer Older
1
/*
Dave Jones's avatar
Dave Jones committed
2
 * handle transition of Linux booting another kernel
3 4 5 6 7 8 9 10 11 12 13 14 15 16
 * 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/string.h>
#include <linux/reboot.h>
#include <asm/pgtable.h>
#include <asm/tlbflush.h>
#include <asm/mmu_context.h>
#include <asm/io.h>
17

18 19 20 21 22 23 24 25 26
#define PAGE_ALIGNED __attribute__ ((__aligned__(PAGE_SIZE)))
static u64 kexec_pgd[512] PAGE_ALIGNED;
static u64 kexec_pud0[512] PAGE_ALIGNED;
static u64 kexec_pmd0[512] PAGE_ALIGNED;
static u64 kexec_pte0[512] PAGE_ALIGNED;
static u64 kexec_pud1[512] PAGE_ALIGNED;
static u64 kexec_pmd1[512] PAGE_ALIGNED;
static u64 kexec_pte1[512] PAGE_ALIGNED;

27
static void init_level2_page(pmd_t *level2p, unsigned long addr)
28 29
{
	unsigned long end_addr;
Maneesh Soni's avatar
Maneesh Soni committed
30

31
	addr &= PAGE_MASK;
32
	end_addr = addr + PUD_SIZE;
Maneesh Soni's avatar
Maneesh Soni committed
33
	while (addr < end_addr) {
34 35
		set_pmd(level2p++, __pmd(addr | __PAGE_KERNEL_LARGE_EXEC));
		addr += PMD_SIZE;
36 37 38
	}
}

39
static int init_level3_page(struct kimage *image, pud_t *level3p,
Maneesh Soni's avatar
Maneesh Soni committed
40
				unsigned long addr, unsigned long last_addr)
41 42 43
{
	unsigned long end_addr;
	int result;
Maneesh Soni's avatar
Maneesh Soni committed
44

45 46
	result = 0;
	addr &= PAGE_MASK;
47
	end_addr = addr + PGDIR_SIZE;
Maneesh Soni's avatar
Maneesh Soni committed
48
	while ((addr < last_addr) && (addr < end_addr)) {
49
		struct page *page;
50
		pmd_t *level2p;
Maneesh Soni's avatar
Maneesh Soni committed
51

52 53 54 55 56
		page = kimage_alloc_control_pages(image, 0);
		if (!page) {
			result = -ENOMEM;
			goto out;
		}
57
		level2p = (pmd_t *)page_address(page);
58
		init_level2_page(level2p, addr);
59 60
		set_pud(level3p++, __pud(__pa(level2p) | _KERNPG_TABLE));
		addr += PUD_SIZE;
61 62
	}
	/* clear the unused entries */
Maneesh Soni's avatar
Maneesh Soni committed
63
	while (addr < end_addr) {
64 65
		pud_clear(level3p++);
		addr += PUD_SIZE;
66 67 68 69 70 71
	}
out:
	return result;
}


72
static int init_level4_page(struct kimage *image, pgd_t *level4p,
Maneesh Soni's avatar
Maneesh Soni committed
73
				unsigned long addr, unsigned long last_addr)
74 75 76
{
	unsigned long end_addr;
	int result;
Maneesh Soni's avatar
Maneesh Soni committed
77

78 79
	result = 0;
	addr &= PAGE_MASK;
80
	end_addr = addr + (PTRS_PER_PGD * PGDIR_SIZE);
Maneesh Soni's avatar
Maneesh Soni committed
81
	while ((addr < last_addr) && (addr < end_addr)) {
82
		struct page *page;
83
		pud_t *level3p;
Maneesh Soni's avatar
Maneesh Soni committed
84

85 86 87 88 89
		page = kimage_alloc_control_pages(image, 0);
		if (!page) {
			result = -ENOMEM;
			goto out;
		}
90
		level3p = (pud_t *)page_address(page);
91 92 93 94
		result = init_level3_page(image, level3p, addr, last_addr);
		if (result) {
			goto out;
		}
95 96
		set_pgd(level4p++, __pgd(__pa(level3p) | _KERNPG_TABLE));
		addr += PGDIR_SIZE;
97 98
	}
	/* clear the unused entries */
Maneesh Soni's avatar
Maneesh Soni committed
99
	while (addr < end_addr) {
100 101
		pgd_clear(level4p++);
		addr += PGDIR_SIZE;
102
	}
Maneesh Soni's avatar
Maneesh Soni committed
103
out:
104 105 106 107 108 109
	return result;
}


static int init_pgtable(struct kimage *image, unsigned long start_pgtable)
{
110 111
	pgd_t *level4p;
	level4p = (pgd_t *)__va(start_pgtable);
Maneesh Soni's avatar
Maneesh Soni committed
112
 	return init_level4_page(image, level4p, 0, end_pfn << PAGE_SHIFT);
113 114 115 116
}

static void set_idt(void *newidt, u16 limit)
{
117
	struct desc_ptr curidt;
118 119

	/* x86-64 supports unaliged loads & stores */
120 121
	curidt.size    = limit;
	curidt.address = (unsigned long)newidt;
122 123

	__asm__ __volatile__ (
124 125
		"lidtq %0\n"
		: : "m" (curidt)
126 127 128 129 130 131
		);
};


static void set_gdt(void *newgdt, u16 limit)
{
132
	struct desc_ptr curgdt;
133 134

	/* x86-64 supports unaligned loads & stores */
135 136
	curgdt.size    = limit;
	curgdt.address = (unsigned long)newgdt;
137 138

	__asm__ __volatile__ (
139 140
		"lgdtq %0\n"
		: : "m" (curgdt)
141 142 143 144 145 146
		);
};

static void load_segments(void)
{
	__asm__ __volatile__ (
147 148 149 150 151
		"\tmovl %0,%%ds\n"
		"\tmovl %0,%%es\n"
		"\tmovl %0,%%ss\n"
		"\tmovl %0,%%fs\n"
		"\tmovl %0,%%gs\n"
Michael Matz's avatar
Michael Matz committed
152
		: : "a" (__KERNEL_DS) : "memory"
153 154 155 156 157
		);
}

int machine_kexec_prepare(struct kimage *image)
{
158
	unsigned long start_pgtable;
159 160 161
	int result;

	/* Calculate the offsets */
Maneesh Soni's avatar
Maneesh Soni committed
162
	start_pgtable = page_to_pfn(image->control_code_page) << PAGE_SHIFT;
163 164 165

	/* Setup the identity mapped 64bit page table */
	result = init_pgtable(image, start_pgtable);
Maneesh Soni's avatar
Maneesh Soni committed
166
	if (result)
167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182
		return result;

	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)
{
183 184
	unsigned long page_list[PAGES_NR];
	void *control_page;
185 186 187 188

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

189 190 191
	control_page = page_address(image->control_code_page) + PAGE_SIZE;
	memcpy(control_page, relocate_kernel, PAGE_SIZE);

192
	page_list[PA_CONTROL_PAGE] = virt_to_phys(control_page);
193
	page_list[VA_CONTROL_PAGE] = (unsigned long)relocate_kernel;
194
	page_list[PA_PGD] = virt_to_phys(&kexec_pgd);
195
	page_list[VA_PGD] = (unsigned long)kexec_pgd;
196
	page_list[PA_PUD_0] = virt_to_phys(&kexec_pud0);
197
	page_list[VA_PUD_0] = (unsigned long)kexec_pud0;
198
	page_list[PA_PMD_0] = virt_to_phys(&kexec_pmd0);
199
	page_list[VA_PMD_0] = (unsigned long)kexec_pmd0;
200
	page_list[PA_PTE_0] = virt_to_phys(&kexec_pte0);
201
	page_list[VA_PTE_0] = (unsigned long)kexec_pte0;
202
	page_list[PA_PUD_1] = virt_to_phys(&kexec_pud1);
203
	page_list[VA_PUD_1] = (unsigned long)kexec_pud1;
204
	page_list[PA_PMD_1] = virt_to_phys(&kexec_pmd1);
205
	page_list[VA_PMD_1] = (unsigned long)kexec_pmd1;
206
	page_list[PA_PTE_1] = virt_to_phys(&kexec_pte1);
207 208 209 210
	page_list[VA_PTE_1] = (unsigned long)kexec_pte1;

	page_list[PA_TABLE_PAGE] =
	  (unsigned long)__pa(page_address(image->control_code_page));
211

212 213 214 215 216
	/* The segment registers are funny things, they have both a
	 * visible and an invisible part.  Whenever the visible part is
	 * set to a specific selector, the invisible part is loaded
	 * with from a table in memory.  At no other time is the
	 * descriptor table in memory accessed.
217 218 219 220 221 222 223 224 225 226
	 *
	 * 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);
227

228
	/* now call it */
229 230
	relocate_kernel((unsigned long)image->head, (unsigned long)page_list,
			image->start);
231
}
232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259

/* crashkernel=size@addr specifies the location to reserve for
 * a crash kernel.  By reserving this memory we guarantee
 * that linux never set's it up as a DMA target.
 * Useful for holding code to do something appropriate
 * after a kernel panic.
 */
static int __init setup_crashkernel(char *arg)
{
	unsigned long size, base;
	char *p;
	if (!arg)
		return -EINVAL;
	size = memparse(arg, &p);
	if (arg == p)
		return -EINVAL;
	if (*p == '@') {
		base = memparse(p+1, &p);
		/* FIXME: Do I want a sanity check to validate the
		 * memory range?  Yes you do, but it's too early for
		 * e820 -AK */
		crashk_res.start = base;
		crashk_res.end   = base + size - 1;
	}
	return 0;
}
early_param("crashkernel", setup_crashkernel);