head_64.S 10.9 KB
Newer Older
L
Linus Torvalds 已提交
1 2 3 4 5 6 7
/*
 *  linux/arch/x86_64/kernel/head.S -- start in 32bit and switch to 64bit
 *
 *  Copyright (C) 2000 Andrea Arcangeli <andrea@suse.de> SuSE
 *  Copyright (C) 2000 Pavel Machek <pavel@suse.cz>
 *  Copyright (C) 2000 Karsten Keil <kkeil@suse.de>
 *  Copyright (C) 2001,2002 Andi Kleen <ak@suse.de>
8
 *  Copyright (C) 2005 Eric Biederman <ebiederm@xmission.com>
L
Linus Torvalds 已提交
9 10 11 12 13
 */


#include <linux/linkage.h>
#include <linux/threads.h>
14
#include <linux/init.h>
L
Linus Torvalds 已提交
15 16
#include <asm/desc.h>
#include <asm/segment.h>
17
#include <asm/pgtable.h>
L
Linus Torvalds 已提交
18 19 20
#include <asm/page.h>
#include <asm/msr.h>
#include <asm/cache.h>
21

22 23 24 25 26 27 28
#ifdef CONFIG_PARAVIRT
#include <asm/asm-offsets.h>
#include <asm/paravirt.h>
#else
#define GET_CR2_INTO_RCX movq %cr2, %rcx
#endif

L
Linus Torvalds 已提交
29
/* we are not able to switch in one step to the final KERNEL ADRESS SPACE
30 31
 * because we need identity-mapped pages.
 *
L
Linus Torvalds 已提交
32 33 34
 */

	.text
35
	.section .text.head
36 37 38 39
	.code64
	.globl startup_64
startup_64:

L
Linus Torvalds 已提交
40
	/*
41 42 43 44 45 46 47 48 49 50 51 52 53 54 55
	 * At this point the CPU runs in 64bit mode CS.L = 1 CS.D = 1,
	 * and someone has loaded an identity mapped page table
	 * for us.  These identity mapped page tables map all of the
	 * kernel pages and possibly all of memory.
	 *
	 * %esi holds a physical pointer to real_mode_data.
	 *
	 * We come here either directly from a 64bit bootloader, or from
	 * arch/x86_64/boot/compressed/head.S.
	 *
	 * We only come here initially at boot nothing else comes here.
	 *
	 * Since we may be loaded at an address different from what we were
	 * compiled to run at we first fixup the physical addresses in our page
	 * tables and then reload them.
L
Linus Torvalds 已提交
56 57
	 */

58 59
	/* Compute the delta between the address I am compiled to run at and the
	 * address I am actually running at.
L
Linus Torvalds 已提交
60
	 */
61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76
	leaq	_text(%rip), %rbp
	subq	$_text - __START_KERNEL_map, %rbp

	/* Is the address not 2M aligned? */
	movq	%rbp, %rax
	andl	$~LARGE_PAGE_MASK, %eax
	testl	%eax, %eax
	jnz	bad_address

	/* Is the address too large? */
	leaq	_text(%rip), %rdx
	movq	$PGDIR_SIZE, %rax
	cmpq	%rax, %rdx
	jae	bad_address

	/* Fixup the physical addresses in the page table
L
Linus Torvalds 已提交
77
	 */
78 79 80 81 82
	addq	%rbp, init_level4_pgt + 0(%rip)
	addq	%rbp, init_level4_pgt + (258*8)(%rip)
	addq	%rbp, init_level4_pgt + (511*8)(%rip)

	addq	%rbp, level3_ident_pgt + 0(%rip)
E
Eric W. Biderman 已提交
83

84
	addq	%rbp, level3_kernel_pgt + (510*8)(%rip)
E
Eric W. Biderman 已提交
85 86 87
	addq	%rbp, level3_kernel_pgt + (511*8)(%rip)

	addq	%rbp, level2_fixmap_pgt + (506*8)(%rip)
88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124

	/* Add an Identity mapping if I am above 1G */
	leaq	_text(%rip), %rdi
	andq	$LARGE_PAGE_MASK, %rdi

	movq	%rdi, %rax
	shrq	$PUD_SHIFT, %rax
	andq	$(PTRS_PER_PUD - 1), %rax
	jz	ident_complete

	leaq	(level2_spare_pgt - __START_KERNEL_map + _KERNPG_TABLE)(%rbp), %rdx
	leaq	level3_ident_pgt(%rip), %rbx
	movq	%rdx, 0(%rbx, %rax, 8)

	movq	%rdi, %rax
	shrq	$PMD_SHIFT, %rax
	andq	$(PTRS_PER_PMD - 1), %rax
	leaq	__PAGE_KERNEL_LARGE_EXEC(%rdi), %rdx
	leaq	level2_spare_pgt(%rip), %rbx
	movq	%rdx, 0(%rbx, %rax, 8)
ident_complete:

	/* Fixup the kernel text+data virtual addresses
	 */
	leaq	level2_kernel_pgt(%rip), %rdi
	leaq	4096(%rdi), %r8
	/* See if it is a valid page table entry */
1:	testq	$1, 0(%rdi)
	jz	2f
	addq	%rbp, 0(%rdi)
	/* Go to the next page */
2:	addq	$8, %rdi
	cmp	%r8, %rdi
	jne	1b

	/* Fixup phys_base */
	addq	%rbp, phys_base(%rip)
L
Linus Torvalds 已提交
125

126 127 128 129
#ifdef CONFIG_SMP
	addq	%rbp, trampoline_level4_pgt + 0(%rip)
	addq	%rbp, trampoline_level4_pgt + (511*8)(%rip)
#endif
L
Len Brown 已提交
130
#ifdef CONFIG_ACPI_SLEEP
131 132 133
	addq	%rbp, wakeup_level4_pgt + 0(%rip)
	addq	%rbp, wakeup_level4_pgt + (511*8)(%rip)
#endif
L
Linus Torvalds 已提交
134

135 136 137
	/* Due to ENTRY(), sometimes the empty space gets filled with
	 * zeros. Better take a jmp than relying on empty space being
	 * filled with 0x90 (nop)
L
Linus Torvalds 已提交
138
	 */
139
	jmp secondary_startup_64
140
ENTRY(secondary_startup_64)
141 142 143 144 145 146 147 148 149 150 151 152
	/*
	 * At this point the CPU runs in 64bit mode CS.L = 1 CS.D = 1,
	 * and someone has loaded a mapped page table.
	 *
	 * %esi holds a physical pointer to real_mode_data.
	 *
	 * We come here either from startup_64 (using physical addresses)
	 * or from trampoline.S (using virtual addresses).
	 *
	 * Using virtual addresses from trampoline.S removes the need
	 * to have any identity mapped pages in the kernel page table
	 * after the boot processor executes this code.
L
Linus Torvalds 已提交
153 154 155 156 157 158 159 160 161
	 */

	/* Enable PAE mode and PGE */
	xorq	%rax, %rax
	btsq	$5, %rax
	btsq	$7, %rax
	movq	%rax, %cr4

	/* Setup early boot stage 4 level pagetables. */
162
	movq	$(init_level4_pgt - __START_KERNEL_map), %rax
163
	addq	phys_base(%rip), %rax
L
Linus Torvalds 已提交
164 165
	movq	%rax, %cr3

166 167 168 169 170
	/* Ensure I am executing from virtual addresses */
	movq	$1f, %rax
	jmp	*%rax
1:

L
Linus Torvalds 已提交
171 172 173 174 175 176 177 178
	/* Check if nx is implemented */
	movl	$0x80000001, %eax
	cpuid
	movl	%edx,%edi

	/* Setup EFER (Extended Feature Enable Register) */
	movl	$MSR_EFER, %ecx
	rdmsr
179 180
	btsl	$_EFER_SCE, %eax	/* Enable System Call */
	btl	$20,%edi		/* No Execute supported? */
L
Linus Torvalds 已提交
181 182
	jnc     1f
	btsl	$_EFER_NX, %eax
183
1:	wrmsr				/* Make changes effective */
L
Linus Torvalds 已提交
184 185

	/* Setup cr0 */
186 187 188 189 190 191 192 193
#define CR0_PM				1		/* protected mode */
#define CR0_MP				(1<<1)
#define CR0_ET				(1<<4)
#define CR0_NE				(1<<5)
#define CR0_WP				(1<<16)
#define CR0_AM				(1<<18)
#define CR0_PAGING 			(1<<31)
	movl $CR0_PM|CR0_MP|CR0_ET|CR0_NE|CR0_WP|CR0_AM|CR0_PAGING,%eax
L
Linus Torvalds 已提交
194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209
	/* Make changes effective */
	movq	%rax, %cr0

	/* Setup a boot time stack */
	movq init_rsp(%rip),%rsp

	/* zero EFLAGS after setting rsp */
	pushq $0
	popfq

	/*
	 * We must switch to a new descriptor in kernel space for the GDT
	 * because soon the kernel won't have access anymore to the userspace
	 * addresses where we're currently running on. We have to do that here
	 * because in 32bit we couldn't load a 64bit linear address.
	 */
210
	lgdt	cpu_gdt_descr(%rip)
L
Linus Torvalds 已提交
211

212 213 214 215 216 217 218 219 220 221 222 223 224 225
	/* set up data segments. actually 0 would do too */
	movl $__KERNEL_DS,%eax
	movl %eax,%ds
	movl %eax,%ss
	movl %eax,%es

	/*
	 * We don't really need to load %fs or %gs, but load them anyway
	 * to kill any stale realmode selectors.  This allows execution
	 * under VT hardware.
	 */
	movl %eax,%fs
	movl %eax,%gs

L
Linus Torvalds 已提交
226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241
	/* 
	 * Setup up a dummy PDA. this is just for some early bootup code
	 * that does in_interrupt() 
	 */ 
	movl	$MSR_GS_BASE,%ecx
	movq	$empty_zero_page,%rax
	movq    %rax,%rdx
	shrq	$32,%rdx
	wrmsr	

	/* esi is pointer to real mode structure with interesting info.
	   pass it to C */
	movl	%esi, %edi
	
	/* Finally jump to run C code and to be on real kernel address
	 * Since we are running on identity-mapped space we have to jump
242 243 244
	 * to the full 64bit address, this is only possible as indirect
	 * jump.  In addition we need to ensure %cs is set so we make this
	 * a far return.
L
Linus Torvalds 已提交
245 246
	 */
	movq	initial_code(%rip),%rax
247 248 249 250
	pushq	$0		# fake return address to stop unwinder
	pushq	$__KERNEL_CS	# set correct cs
	pushq	%rax		# target address in negative space
	lretq
L
Linus Torvalds 已提交
251

252
	/* SMP bootup changes these two */
253 254 255
#ifndef CONFIG_HOTPLUG_CPU
	.pushsection .init.data
#endif
256
	.align	8
L
Linus Torvalds 已提交
257 258 259
	.globl	initial_code
initial_code:
	.quad	x86_64_start_kernel
260 261 262
#ifndef CONFIG_HOTPLUG_CPU
	.popsection
#endif
L
Linus Torvalds 已提交
263 264 265 266
	.globl init_rsp
init_rsp:
	.quad  init_thread_union+THREAD_SIZE-8

267 268 269
bad_address:
	jmp bad_address

L
Linus Torvalds 已提交
270
ENTRY(early_idt_handler)
271 272 273
	cmpl $2,early_recursion_flag(%rip)
	jz  1f
	incl early_recursion_flag(%rip)
L
Linus Torvalds 已提交
274 275 276
	xorl %eax,%eax
	movq 8(%rsp),%rsi	# get rip
	movq (%rsp),%rdx
277
	GET_CR2_INTO_RCX
L
Linus Torvalds 已提交
278 279
	leaq early_idt_msg(%rip),%rdi
	call early_printk
280 281 282
	cmpl $2,early_recursion_flag(%rip)
	jz  1f
	call dump_stack
283 284 285 286 287
#ifdef CONFIG_KALLSYMS	
	leaq early_idt_ripmsg(%rip),%rdi
	movq 8(%rsp),%rsi	# get rip again
	call __print_symbol
#endif
L
Linus Torvalds 已提交
288 289
1:	hlt
	jmp 1b
290 291
early_recursion_flag:
	.long 0
L
Linus Torvalds 已提交
292 293 294

early_idt_msg:
	.asciz "PANIC: early exception rip %lx error %lx cr2 %lx\n"
295 296
early_idt_ripmsg:
	.asciz "RIP %s\n"
L
Linus Torvalds 已提交
297

298
.balign PAGE_SIZE
L
Linus Torvalds 已提交
299

300
#define NEXT_PAGE(name) \
301
	.balign	PAGE_SIZE; \
302 303
ENTRY(name)

304 305 306 307 308 309 310 311
/* Automate the creation of 1 to 1 mapping pmd entries */
#define PMDS(START, PERM, COUNT)		\
	i = 0 ;					\
	.rept (COUNT) ;				\
	.quad	(START) + (i << 21) + (PERM) ;	\
	i = i + 1 ;				\
	.endr

312 313 314 315 316 317
	/*
	 * This default setting generates an ident mapping at address 0x100000
	 * and a mapping for the kernel that precisely maps virtual address
	 * 0xffffffff80000000 to physical address 0x000000. (always using
	 * 2Mbyte large pages provided by PAE mode)
	 */
318
NEXT_PAGE(init_level4_pgt)
319 320 321 322 323 324
	.quad	level3_ident_pgt - __START_KERNEL_map + _KERNPG_TABLE
	.fill	257,8,0
	.quad	level3_ident_pgt - __START_KERNEL_map + _KERNPG_TABLE
	.fill	252,8,0
	/* (2^48-(2*1024*1024*1024))/(2^39) = 511 */
	.quad	level3_kernel_pgt - __START_KERNEL_map + _PAGE_TABLE
L
Linus Torvalds 已提交
325

326
NEXT_PAGE(level3_ident_pgt)
327
	.quad	level2_ident_pgt - __START_KERNEL_map + _KERNPG_TABLE
L
Linus Torvalds 已提交
328 329
	.fill	511,8,0

330
NEXT_PAGE(level3_kernel_pgt)
L
Linus Torvalds 已提交
331 332
	.fill	510,8,0
	/* (2^48-(2*1024*1024*1024)-((2^39)*511))/(2^30) = 510 */
333
	.quad	level2_kernel_pgt - __START_KERNEL_map + _KERNPG_TABLE
E
Eric W. Biderman 已提交
334 335 336 337 338 339 340 341 342 343
	.quad	level2_fixmap_pgt - __START_KERNEL_map + _PAGE_TABLE

NEXT_PAGE(level2_fixmap_pgt)
	.fill	506,8,0
	.quad	level1_fixmap_pgt - __START_KERNEL_map + _PAGE_TABLE
	/* 8MB reserved for vsyscalls + a 2MB hole = 4 + 1 entries */
	.fill	5,8,0

NEXT_PAGE(level1_fixmap_pgt)
	.fill	512,8,0
L
Linus Torvalds 已提交
344

345
NEXT_PAGE(level2_ident_pgt)
346 347 348 349
	/* Since I easily can, map the first 1G.
	 * Don't set NX because code runs from these pages.
	 */
	PMDS(0x0000000000000000, __PAGE_KERNEL_LARGE_EXEC, PTRS_PER_PMD)
350

351
NEXT_PAGE(level2_kernel_pgt)
L
Linus Torvalds 已提交
352 353 354
	/* 40MB kernel mapping. The kernel code cannot be bigger than that.
	   When you change this change KERNEL_TEXT_SIZE in page.h too. */
	/* (2^48-(2*1024*1024*1024)-((2^39)*511)-((2^30)*510)) = 0 */
355
	PMDS(0x0000000000000000, __PAGE_KERNEL_LARGE_EXEC|_PAGE_GLOBAL, KERNEL_TEXT_SIZE/PMD_SIZE)
L
Linus Torvalds 已提交
356
	/* Module mapping starts here */
357
	.fill	(PTRS_PER_PMD - (KERNEL_TEXT_SIZE/PMD_SIZE)),8,0
L
Linus Torvalds 已提交
358

359 360 361
NEXT_PAGE(level2_spare_pgt)
	.fill   512,8,0

362
#undef PMDS
363
#undef NEXT_PAGE
L
Linus Torvalds 已提交
364

365
	.data
L
Linus Torvalds 已提交
366 367 368
	.align 16
	.globl cpu_gdt_descr
cpu_gdt_descr:
369
	.word	gdt_end-cpu_gdt_table-1
L
Linus Torvalds 已提交
370 371 372 373 374 375 376 377 378
gdt:
	.quad	cpu_gdt_table
#ifdef CONFIG_SMP
	.rept	NR_CPUS-1
	.word	0
	.quad	0
	.endr
#endif

379 380 381 382
ENTRY(phys_base)
	/* This must match the first entry in level2_kernel_pgt */
	.quad   0x0000000000000000

L
Linus Torvalds 已提交
383 384 385 386 387
/* We need valid kernel segments for data and code in long mode too
 * IRET will check the segment types  kkeil 2000/10/28
 * Also sysret mandates a special GDT layout 
 */
		 		
388 389
	.section .data.page_aligned, "aw"
	.align PAGE_SIZE
L
Linus Torvalds 已提交
390 391 392 393 394 395

/* The TLS descriptors are currently at a different place compared to i386.
   Hopefully nobody expects them at a fixed place (Wine?) */
	
ENTRY(cpu_gdt_table)
	.quad	0x0000000000000000	/* NULL descriptor */
V
Vivek Goyal 已提交
396 397 398 399 400 401
	.quad	0x00cf9b000000ffff	/* __KERNEL32_CS */
	.quad	0x00af9b000000ffff	/* __KERNEL_CS */
	.quad	0x00cf93000000ffff	/* __KERNEL_DS */
	.quad	0x00cffb000000ffff	/* __USER32_CS */
	.quad	0x00cff3000000ffff	/* __USER_DS, __USER32_DS  */
	.quad	0x00affb000000ffff	/* __USER_CS */
402
	.quad	0x0			/* unused */
L
Linus Torvalds 已提交
403 404 405
	.quad	0,0			/* TSS */
	.quad	0,0			/* LDT */
	.quad   0,0,0			/* three TLS descriptors */ 
406
	.quad	0x0000f40000000000	/* node/CPU stored in limit */
L
Linus Torvalds 已提交
407 408 409
gdt_end:	
	/* asm/segment.h:GDT_ENTRIES must match this */	
	/* This should be a multiple of the cache line size */
410 411 412 413
	/* GDTs of other CPUs are now dynamically allocated */

	/* zero the remaining page */
	.fill PAGE_SIZE / 8 - GDT_ENTRIES,8,0
L
Linus Torvalds 已提交
414

415 416 417 418
	.section .bss, "aw", @nobits
	.align L1_CACHE_BYTES
ENTRY(idt_table)
	.skip 256 * 16
L
Linus Torvalds 已提交
419

420 421 422 423
	.section .bss.page_aligned, "aw", @nobits
	.align PAGE_SIZE
ENTRY(empty_zero_page)
	.skip PAGE_SIZE