machine_kexec_64.c 8.26 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
 * 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>
Ken'ichi Ohmichi's avatar
Ken'ichi Ohmichi committed
13
#include <linux/numa.h>
Ingo Molnar's avatar
Ingo Molnar committed
14
#include <linux/ftrace.h>
15
#include <linux/io.h>
16
#include <linux/suspend.h>
Ingo Molnar's avatar
Ingo Molnar committed
17

18 19 20
#include <asm/pgtable.h>
#include <asm/tlbflush.h>
#include <asm/mmu_context.h>
21
#include <asm/debugreg.h>
22

23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57
static int init_one_level2_page(struct kimage *image, pgd_t *pgd,
				unsigned long addr)
{
	pud_t *pud;
	pmd_t *pmd;
	struct page *page;
	int result = -ENOMEM;

	addr &= PMD_MASK;
	pgd += pgd_index(addr);
	if (!pgd_present(*pgd)) {
		page = kimage_alloc_control_pages(image, 0);
		if (!page)
			goto out;
		pud = (pud_t *)page_address(page);
		memset(pud, 0, PAGE_SIZE);
		set_pgd(pgd, __pgd(__pa(pud) | _KERNPG_TABLE));
	}
	pud = pud_offset(pgd, addr);
	if (!pud_present(*pud)) {
		page = kimage_alloc_control_pages(image, 0);
		if (!page)
			goto out;
		pmd = (pmd_t *)page_address(page);
		memset(pmd, 0, PAGE_SIZE);
		set_pud(pud, __pud(__pa(pmd) | _KERNPG_TABLE));
	}
	pmd = pmd_offset(pud, addr);
	if (!pmd_present(*pmd))
		set_pmd(pmd, __pmd(addr | __PAGE_KERNEL_LARGE_EXEC));
	result = 0;
out:
	return result;
}

58
static void init_level2_page(pmd_t *level2p, unsigned long addr)
59 60
{
	unsigned long end_addr;
Maneesh Soni's avatar
Maneesh Soni committed
61

62
	addr &= PAGE_MASK;
63
	end_addr = addr + PUD_SIZE;
Maneesh Soni's avatar
Maneesh Soni committed
64
	while (addr < end_addr) {
65 66
		set_pmd(level2p++, __pmd(addr | __PAGE_KERNEL_LARGE_EXEC));
		addr += PMD_SIZE;
67 68 69
	}
}

70
static int init_level3_page(struct kimage *image, pud_t *level3p,
Maneesh Soni's avatar
Maneesh Soni committed
71
				unsigned long addr, unsigned long last_addr)
72 73 74
{
	unsigned long end_addr;
	int result;
Maneesh Soni's avatar
Maneesh Soni committed
75

76 77
	result = 0;
	addr &= PAGE_MASK;
78
	end_addr = addr + PGDIR_SIZE;
Maneesh Soni's avatar
Maneesh Soni committed
79
	while ((addr < last_addr) && (addr < end_addr)) {
80
		struct page *page;
81
		pmd_t *level2p;
Maneesh Soni's avatar
Maneesh Soni committed
82

83 84 85 86 87
		page = kimage_alloc_control_pages(image, 0);
		if (!page) {
			result = -ENOMEM;
			goto out;
		}
88
		level2p = (pmd_t *)page_address(page);
89
		init_level2_page(level2p, addr);
90 91
		set_pud(level3p++, __pud(__pa(level2p) | _KERNPG_TABLE));
		addr += PUD_SIZE;
92 93
	}
	/* clear the unused entries */
Maneesh Soni's avatar
Maneesh Soni committed
94
	while (addr < end_addr) {
95 96
		pud_clear(level3p++);
		addr += PUD_SIZE;
97 98 99 100 101 102
	}
out:
	return result;
}


103
static int init_level4_page(struct kimage *image, pgd_t *level4p,
Maneesh Soni's avatar
Maneesh Soni committed
104
				unsigned long addr, unsigned long last_addr)
105 106 107
{
	unsigned long end_addr;
	int result;
Maneesh Soni's avatar
Maneesh Soni committed
108

109 110
	result = 0;
	addr &= PAGE_MASK;
111
	end_addr = addr + (PTRS_PER_PGD * PGDIR_SIZE);
Maneesh Soni's avatar
Maneesh Soni committed
112
	while ((addr < last_addr) && (addr < end_addr)) {
113
		struct page *page;
114
		pud_t *level3p;
Maneesh Soni's avatar
Maneesh Soni committed
115

116 117 118 119 120
		page = kimage_alloc_control_pages(image, 0);
		if (!page) {
			result = -ENOMEM;
			goto out;
		}
121
		level3p = (pud_t *)page_address(page);
122
		result = init_level3_page(image, level3p, addr, last_addr);
123
		if (result)
124
			goto out;
125 126
		set_pgd(level4p++, __pgd(__pa(level3p) | _KERNPG_TABLE));
		addr += PGDIR_SIZE;
127 128
	}
	/* clear the unused entries */
Maneesh Soni's avatar
Maneesh Soni committed
129
	while (addr < end_addr) {
130 131
		pgd_clear(level4p++);
		addr += PGDIR_SIZE;
132
	}
Maneesh Soni's avatar
Maneesh Soni committed
133
out:
134 135 136
	return result;
}

137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185
static void free_transition_pgtable(struct kimage *image)
{
	free_page((unsigned long)image->arch.pud);
	free_page((unsigned long)image->arch.pmd);
	free_page((unsigned long)image->arch.pte);
}

static int init_transition_pgtable(struct kimage *image, pgd_t *pgd)
{
	pud_t *pud;
	pmd_t *pmd;
	pte_t *pte;
	unsigned long vaddr, paddr;
	int result = -ENOMEM;

	vaddr = (unsigned long)relocate_kernel;
	paddr = __pa(page_address(image->control_code_page)+PAGE_SIZE);
	pgd += pgd_index(vaddr);
	if (!pgd_present(*pgd)) {
		pud = (pud_t *)get_zeroed_page(GFP_KERNEL);
		if (!pud)
			goto err;
		image->arch.pud = pud;
		set_pgd(pgd, __pgd(__pa(pud) | _KERNPG_TABLE));
	}
	pud = pud_offset(pgd, vaddr);
	if (!pud_present(*pud)) {
		pmd = (pmd_t *)get_zeroed_page(GFP_KERNEL);
		if (!pmd)
			goto err;
		image->arch.pmd = pmd;
		set_pud(pud, __pud(__pa(pmd) | _KERNPG_TABLE));
	}
	pmd = pmd_offset(pud, vaddr);
	if (!pmd_present(*pmd)) {
		pte = (pte_t *)get_zeroed_page(GFP_KERNEL);
		if (!pte)
			goto err;
		image->arch.pte = pte;
		set_pmd(pmd, __pmd(__pa(pte) | _KERNPG_TABLE));
	}
	pte = pte_offset_kernel(pmd, vaddr);
	set_pte(pte, pfn_pte(paddr >> PAGE_SHIFT, PAGE_KERNEL_EXEC));
	return 0;
err:
	free_transition_pgtable(image);
	return result;
}

186 187 188

static int init_pgtable(struct kimage *image, unsigned long start_pgtable)
{
189
	pgd_t *level4p;
190
	int result;
191
	level4p = (pgd_t *)__va(start_pgtable);
192
	result = init_level4_page(image, level4p, 0, max_pfn << PAGE_SHIFT);
193 194 195 196 197 198 199
	if (result)
		return result;
	/*
	 * image->start may be outside 0 ~ max_pfn, for example when
	 * jump back to original kernel from kexeced kernel
	 */
	result = init_one_level2_page(image, level4p, image->start);
200 201 202
	if (result)
		return result;
	return init_transition_pgtable(image, level4p);
203 204 205 206
}

static void set_idt(void *newidt, u16 limit)
{
207
	struct desc_ptr curidt;
208 209

	/* x86-64 supports unaliged loads & stores */
210 211
	curidt.size    = limit;
	curidt.address = (unsigned long)newidt;
212 213

	__asm__ __volatile__ (
214 215
		"lidtq %0\n"
		: : "m" (curidt)
216 217 218 219 220 221
		);
};


static void set_gdt(void *newgdt, u16 limit)
{
222
	struct desc_ptr curgdt;
223 224

	/* x86-64 supports unaligned loads & stores */
225 226
	curgdt.size    = limit;
	curgdt.address = (unsigned long)newgdt;
227 228

	__asm__ __volatile__ (
229 230
		"lgdtq %0\n"
		: : "m" (curgdt)
231 232 233 234 235 236
		);
};

static void load_segments(void)
{
	__asm__ __volatile__ (
237 238 239 240 241
		"\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
242
		: : "a" (__KERNEL_DS) : "memory"
243 244 245 246 247
		);
}

int machine_kexec_prepare(struct kimage *image)
{
248
	unsigned long start_pgtable;
249 250 251
	int result;

	/* Calculate the offsets */
Maneesh Soni's avatar
Maneesh Soni committed
252
	start_pgtable = page_to_pfn(image->control_code_page) << PAGE_SHIFT;
253 254 255

	/* Setup the identity mapped 64bit page table */
	result = init_pgtable(image, start_pgtable);
Maneesh Soni's avatar
Maneesh Soni committed
256
	if (result)
257 258 259 260 261 262 263
		return result;

	return 0;
}

void machine_kexec_cleanup(struct kimage *image)
{
264
	free_transition_pgtable(image);
265 266 267 268 269 270
}

/*
 * Do not allocate memory (or fail in any way) in machine_kexec().
 * We are past the point of no return, committed to rebooting now.
 */
Huang Ying's avatar
Huang Ying committed
271
void machine_kexec(struct kimage *image)
272
{
273 274
	unsigned long page_list[PAGES_NR];
	void *control_page;
275
	int save_ftrace_enabled;
276

277
#ifdef CONFIG_KEXEC_JUMP
278
	if (image->preserve_context)
279 280 281 282
		save_processor_state();
#endif

	save_ftrace_enabled = __ftrace_enabled_save();
Ingo Molnar's avatar
Ingo Molnar committed
283

284 285
	/* Interrupts aren't acceptable while we reboot */
	local_irq_disable();
286
	hw_breakpoint_disable();
287

288 289 290 291 292 293 294 295 296 297 298 299 300
	if (image->preserve_context) {
#ifdef CONFIG_X86_IO_APIC
		/*
		 * We need to put APICs in legacy mode so that we can
		 * get timer interrupts in second kernel. kexec/kdump
		 * paths already have calls to disable_IO_APIC() in
		 * one form or other. kexec jump path also need
		 * one.
		 */
		disable_IO_APIC();
#endif
	}

301
	control_page = page_address(image->control_code_page) + PAGE_SIZE;
302
	memcpy(control_page, relocate_kernel, KEXEC_CONTROL_CODE_MAX_SIZE);
303

304
	page_list[PA_CONTROL_PAGE] = virt_to_phys(control_page);
305
	page_list[VA_CONTROL_PAGE] = (unsigned long)control_page;
306 307
	page_list[PA_TABLE_PAGE] =
	  (unsigned long)__pa(page_address(image->control_code_page));
308

309 310 311 312
	if (image->type == KEXEC_TYPE_DEFAULT)
		page_list[PA_SWAP_PAGE] = (page_to_pfn(image->swap_page)
						<< PAGE_SHIFT);

313 314
	/*
	 * The segment registers are funny things, they have both a
315 316 317 318
	 * 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.
319 320 321 322 323
	 *
	 * I take advantage of this here by force loading the
	 * segments, before I zap the gdt with an invalid value.
	 */
	load_segments();
324 325
	/*
	 * The gdt & idt are now invalid.
326 327
	 * If you want to load them you must set up your own idt & gdt.
	 */
328 329
	set_gdt(phys_to_virt(0), 0);
	set_idt(phys_to_virt(0), 0);
330

331
	/* now call it */
332 333 334 335 336 337
	image->start = relocate_kernel((unsigned long)image->head,
				       (unsigned long)page_list,
				       image->start,
				       image->preserve_context);

#ifdef CONFIG_KEXEC_JUMP
338
	if (image->preserve_context)
339 340 341 342
		restore_processor_state();
#endif

	__ftrace_enabled_restore(save_ftrace_enabled);
343
}
344

Ken'ichi Ohmichi's avatar
Ken'ichi Ohmichi committed
345 346
void arch_crash_save_vmcoreinfo(void)
{
347
	VMCOREINFO_SYMBOL(phys_base);
348
	VMCOREINFO_SYMBOL(init_level4_pgt);
349 350 351 352 353

#ifdef CONFIG_NUMA
	VMCOREINFO_SYMBOL(node_data);
	VMCOREINFO_LENGTH(node_data, MAX_NUMNODES);
#endif
Ken'ichi Ohmichi's avatar
Ken'ichi Ohmichi committed
354 355
}