*/
#include <grub/symbol.h>
+#include <grub/i386/memory.h>
#ifdef BACKWARD
#define RELOCATOR_VARIABLE(x) VARIABLE(grub_relocator32_backward_ ## x)
#else
#define RELOCATOR_VARIABLE(x) VARIABLE(grub_relocator32_forward_ ## x)
#endif
+#ifdef __x86_64__
+#define RAX %rax
+#define RCX %rcx
+#define RDI %rdi
+#define RSI %rdi
+#else
+#define RAX %eax
+#define RCX %ecx
+#define RDI %edi
+#define RSI %esi
+#endif
+
+/* Apple's linker has a problem with 64-bit relocations. */
+#if defined (__apple__) || ! defined (__x86_64__)
+#define RSIA %esi
+#define RAXA %eax
+#else
+#define RSIA %rsi
+#define RAXA %rax
+#endif
+
+
.p2align 4 /* force 16-byte alignment */
RELOCATOR_VARIABLE(start)
/* mov imm32, %eax */
.byte 0xb8
RELOCATOR_VARIABLE(dest)
- .long 0
- mov %eax, %edi
+ .long 0
+ movl %eax, %edi
/* mov imm32, %eax */
.byte 0xb8
RELOCATOR_VARIABLE(src)
- .long 0
- mov %eax, %esi
+ .long 0
+ movl %eax, %esi
/* mov imm32, %ecx */
.byte 0xb9
RELOCATOR_VARIABLE(size)
- .long 0
- mov %edi, %eax
-#ifndef BACKWARD
- add %ecx, %eax
-#endif
- add $0x3, %ecx
- shr $2, %ecx
-
-#ifdef BACKWARD
- /* Backward movsl is implicitly off-by-four. compensate that. */
- subl $4, %esi
- subl $4, %edi
-
- /* Backward copy. */
- std
- addl %ecx, %esi
- addl %ecx, %edi
-
- rep
- movsl
-
-#else
- /* Forward copy. */
- cld
- rep
- movsl
-#endif
- /* %eax contains now our new base. */
- mov %eax, %esi
- add $(cont0 - base), %eax
- jmp *%eax
-cont0:
+ .long 0
#else
- xorq %rax, %rax
+ xorq %rax, %rax
/* mov imm32, %eax */
.byte 0xb8
RELOCATOR_VARIABLE(dest)
- .long 0
- mov %eax, %edi
+ .long 0
+ movq %rax, %rdi
/* mov imm64, %rax */
.byte 0x48
.byte 0xb8
RELOCATOR_VARIABLE(src)
- .long 0, 0
- mov %rax, %rsi
+ .long 0, 0
+ movq %rax, %rsi
- xorq %rcx, %rcx
+ xorq %rcx, %rcx
/* mov imm32, %ecx */
.byte 0xb9
RELOCATOR_VARIABLE(size)
- .long 0
+ .long 0
+
+#endif
+
+ mov RDI, RAX
+
+#ifdef BACKWARD
+ add RCX, RSI
+ add RDX, RDI
+#endif
- mov %rdi, %rax
#ifndef BACKWARD
- add %rcx, %rax
+ add RCX, RAX
#endif
- add $0x3, %rcx
- shr $2, %rcx
+ addq $0x3, RCX
+ shrq $2, RCX
+
#ifdef BACKWARD
/* Backward movsl is implicitly off-by-four. compensate that. */
- subq $4, %rsi
- subq $4, %rdi
+ subq $4, RSI
+ subq $4, RDI
/* Backward copy. */
std
- addq %rcx, %rsi
- addq %rcx, %rdi
rep
movsl
#endif
/* %rax contains now our new 'base'. */
- mov %rax, %rsi
-#ifdef APPLE_CC
- add $(cont0 - base), %eax
-#else
- add $(cont0 - base), %rax
-#endif
- jmp *%rax
+ mov RAX, RSI
+ add $(cont0 - base), RAXA
+ jmp *RAX
cont0:
-#ifdef APPLE_CC
- lea (cont1 - base) (%esi, 1), %eax
- mov %eax, (jump_vector - base) (%esi, 1)
-
- lea (gdt - base) (%esi, 1), %eax
- mov %eax, (gdt_addr - base) (%esi, 1)
-
- /* Switch to compatibility mode. */
-
- lgdt (gdtdesc - base) (%esi, 1)
+ lea (cont1 - base) (RSIA, 1), RAXA
+ movl %eax, (jump_vector - base) (RSIA, 1)
- /* Update %cs. Thanks to David Miller for pointing this mistake out. */
- ljmp *(jump_vector - base) (%esi,1)
-#else
- lea (cont1 - base) (%rsi, 1), %rax
- mov %eax, (jump_vector - base) (%rsi, 1)
-
- lea (gdt - base) (%rsi, 1), %rax
- mov %rax, (gdt_addr - base) (%rsi, 1)
+ lea (gdt - base) (RSIA, 1), RAXA
+ mov RAXA, (gdt_addr - base) (RSIA, 1)
/* Switch to compatibility mode. */
- lgdt (gdtdesc - base) (%rsi, 1)
+ lgdt (gdtdesc - base) (RSIA, 1)
/* Update %cs. Thanks to David Miller for pointing this mistake out. */
- ljmp *(jump_vector - base) (%rsi, 1)
-#endif
+ ljmp *(jump_vector - base) (RSIA, 1)
cont1:
.code32
/* Update other registers. */
- mov $0x18, %eax
- mov %eax, %ds
- mov %eax, %es
- mov %eax, %fs
- mov %eax, %gs
- mov %eax, %ss
+ movl $GRUB_MEMORY_MACHINE_PROT_MODE_DSEG, %eax
+ movl %eax, %ds
+ movl %eax, %es
+ movl %eax, %fs
+ movl %eax, %gs
+ movl %eax, %ss
/* Disable paging. */
- mov %cr0, %eax
- and $0x7fffffff, %eax
- mov %eax, %cr0
+ movl %cr0, %eax
+ andl $0x7fffffff, %eax
+ movl %eax, %cr0
/* Disable amd64. */
- mov $0xc0000080, %ecx
+ movl $0xc0000080, %ecx
rdmsr
- and $0xfffffeff, %eax
+ andl $0xfffffeff, %eax
wrmsr
/* Turn off PAE. */
- movl %cr4, %eax
- and $0xffffffcf, %eax
- mov %eax, %cr4
+ movl %cr4, %eax
+ andl $0xffffffcf, %eax
+ movl %eax, %cr4
- jmp cont2
+ jmp cont2
cont2:
-#endif
.code32
/* mov imm32, %eax */
.byte 0xb8
RELOCATOR_VARIABLE (esp)
- .long 0
+ .long 0
- movl %eax, %esp
+ movl %eax, %esp
/* mov imm32, %eax */
.byte 0xb8
RELOCATOR_VARIABLE (eax)
- .long 0
+ .long 0
/* mov imm32, %ebx */
.byte 0xbb
RELOCATOR_VARIABLE (ebx)
- .long 0
+ .long 0
/* mov imm32, %ecx */
.byte 0xb9
RELOCATOR_VARIABLE (ecx)
- .long 0
+ .long 0
/* mov imm32, %edx */
.byte 0xba
RELOCATOR_VARIABLE (edx)
- .long 0
+ .long 0
/* Cleared direction flag is of no problem with any current
payload and makes this implementation easier. */
cld
- .byte 0xea
+ .byte 0xea
RELOCATOR_VARIABLE (eip)
- .long 0
-#ifdef __x86_64__
- .word 0x10
-#else
- .word 0x08
-#endif
+ .long 0
+ .word 0x08
-#ifdef __x86_64__
- /* GDT. Copied from loader/i386/linux.c. */
- .p2align 4
+ /* GDT. The same as is used in 32-bit GRUB. */
+ .p2align 4
gdt:
- /* NULL. */
- .byte 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00
-
- /* Reserved. */
- .byte 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00
-
- /* Code segment. */
- .byte 0xFF, 0xFF, 0x00, 0x00, 0x00, 0x9A, 0xCF, 0x00
-
- /* Data segment. */
- .byte 0xFF, 0xFF, 0x00, 0x00, 0x00, 0x92, 0xCF, 0x00
-
+ .word 0, 0
+ .byte 0, 0, 0, 0
+
+ /* -- code segment --
+ * base = 0x00000000, limit = 0xFFFFF (4 KiB Granularity), present
+ * type = 32bit code execute/read, DPL = 0
+ */
+ .word 0xFFFF, 0
+ .byte 0, 0x9A, 0xCF, 0
+
+ /* -- data segment --
+ * base = 0x00000000, limit 0xFFFFF (4 KiB Granularity), present
+ * type = 32 bit data read/write, DPL = 0
+ */
+ .word 0xFFFF, 0
+ .byte 0, 0x92, 0xCF, 0
+
+ /* -- 16 bit real mode CS --
+ * base = 0x00000000, limit 0x0FFFF (1 B Granularity), present
+ * type = 16 bit code execute/read only/conforming, DPL = 0
+ */
+ .word 0xFFFF, 0
+ .byte 0, 0x9E, 0, 0
+
+ /* -- 16 bit real mode DS --
+ * base = 0x00000000, limit 0x0FFFF (1 B Granularity), present
+ * type = 16 bit data read/write, DPL = 0
+ */
+ .word 0xFFFF, 0
+ .byte 0, 0x92, 0, 0
+
+ .p2align 4
gdtdesc:
- .word 31
+ .word 0x27
gdt_addr:
+#ifdef __x86_64__
/* Filled by the code. */
- .quad 0
-
- .p2align 4
+ .quad 0
+#else
+ /* Filled by the code. */
+ .long 0
+#endif
+
+ .p2align 4
jump_vector:
/* Jump location. Is filled by the code */
- .long 0
- .long 0x10
-#endif
+ .long 0
+ .long GRUB_MEMORY_MACHINE_PROT_MODE_CSEG
#ifndef BACKWARD
base: