Skip to content

K
Kernel

openeuler / Kernel
1 年多 前同步成功

通知 8
Star 0
Fork 0
K
Kernel
提交 8d379dad 编写于 作者: J Jan Beulich 提交者： Andi Kleen
浏览文件
操作

[PATCH] annotate arch/x86_64/lib/*.S 

Add unwind annotations to arch/x86_64/lib/*.S, and also use the macros
provided by linux/linkage.h where-ever possible.

Some of the alternative instructions handling needed to be adjusted so
that the replacement code would also have valid unwind information.
Signed-off-by: NJan Beulich <jbeulich@novell.com>
Signed-off-by: NAndi Kleen <ak@suse.de>
上级 fb2e2848
隐藏空白更改
内联 并排
Showing with 244 addition and 143 deletion
arch/x86_64/lib/clear_page.S
浏览文件 @ 8d379dad
#include <linux/linkage.h>
#include <asm/dwarf2.h>
/* /*
* Zero a page. * Zero a page.
* rdi page * rdi page
*/ */
.globl clear_page ALIGN
.p2align 4 clear_page_c:
clear_page: CFI_STARTPROC
movl $4096/8,%ecx
xorl %eax,%eax
rep stosq
ret
CFI_ENDPROC
ENDPROC(clear_page)
ENTRY(clear_page)
CFI_STARTPROC
xorl %eax,%eax xorl %eax,%eax
movl $4096/64,%ecx movl $4096/64,%ecx
.p2align 4 .p2align 4
...@@ -23,28 +35,25 @@ clear_page: ...@@ -23,28 +35,25 @@ clear_page:
jnz .Lloop jnz .Lloop
nop nop
ret ret
clear_page_end: CFI_ENDPROC
.Lclear_page_end:
ENDPROC(clear_page)
/* Some CPUs run faster using the string instructions. /* Some CPUs run faster using the string instructions.
It is also a lot simpler. Use this when possible */ It is also a lot simpler. Use this when possible */
#include <asm/cpufeature.h> #include <asm/cpufeature.h>
.section .altinstr_replacement,"ax"
1: .byte 0xeb /* jmp <disp8> */
.byte (clear_page_c - clear_page) - (2f - 1b) /* offset */
2:
.previous
.section .altinstructions,"a" .section .altinstructions,"a"
.align 8 .align 8
.quad clear_page .quad clear_page
.quad clear_page_c .quad 1b
.byte X86_FEATURE_REP_GOOD .byte X86_FEATURE_REP_GOOD
.byte clear_page_end-clear_page .byte .Lclear_page_end - clear_page
.byte clear_page_c_end-clear_page_c .byte 2b - 1b
.previous
.section .altinstr_replacement,"ax"
clear_page_c:
movl $4096/8,%ecx
xorl %eax,%eax
rep
stosq
ret
clear_page_c_end:
.previous .previous
arch/x86_64/lib/copy_page.S
浏览文件 @ 8d379dad
/* Written 2003 by Andi Kleen, based on a kernel by Evandro Menezes */ /* Written 2003 by Andi Kleen, based on a kernel by Evandro Menezes */
#include <linux/config.h>
#include <linux/linkage.h>
#include <asm/dwarf2.h>
ALIGN
copy_page_c:
CFI_STARTPROC
movl $4096/8,%ecx
rep movsq
ret
CFI_ENDPROC
ENDPROC(copy_page_c)
/* Don't use streaming store because it's better when the target /* Don't use streaming store because it's better when the target
ends up in cache. */ ends up in cache. */
/* Could vary the prefetch distance based on SMP/UP */ /* Could vary the prefetch distance based on SMP/UP */
.globl copy_page ENTRY(copy_page)
.p2align 4 CFI_STARTPROC
copy_page:
subq $3*8,%rsp subq $3*8,%rsp
CFI_ADJUST_CFA_OFFSET 3*8
movq %rbx,(%rsp) movq %rbx,(%rsp)
CFI_REL_OFFSET rbx, 0
movq %r12,1*8(%rsp) movq %r12,1*8(%rsp)
CFI_REL_OFFSET r12, 1*8
movq %r13,2*8(%rsp) movq %r13,2*8(%rsp)
CFI_REL_OFFSET r13, 2*8
movl $(4096/64)-5,%ecx movl $(4096/64)-5,%ecx
.p2align 4 .p2align 4
...@@ -72,30 +88,33 @@ copy_page: ...@@ -72,30 +88,33 @@ copy_page:
jnz .Loop2 jnz .Loop2
movq (%rsp),%rbx movq (%rsp),%rbx
CFI_RESTORE rbx
movq 1*8(%rsp),%r12 movq 1*8(%rsp),%r12
CFI_RESTORE r12
movq 2*8(%rsp),%r13 movq 2*8(%rsp),%r13
CFI_RESTORE r13
addq $3*8,%rsp addq $3*8,%rsp
CFI_ADJUST_CFA_OFFSET -3*8
ret ret
.Lcopy_page_end:
CFI_ENDPROC
ENDPROC(copy_page)
/* Some CPUs run faster using the string copy instructions. /* Some CPUs run faster using the string copy instructions.
It is also a lot simpler. Use this when possible */ It is also a lot simpler. Use this when possible */
#include <asm/cpufeature.h> #include <asm/cpufeature.h>
.section .altinstr_replacement,"ax"
1: .byte 0xeb /* jmp <disp8> */
.byte (copy_page_c - copy_page) - (2f - 1b) /* offset */
2:
.previous
.section .altinstructions,"a" .section .altinstructions,"a"
.align 8 .align 8
.quad copy_page .quad copy_page
.quad copy_page_c .quad 1b
.byte X86_FEATURE_REP_GOOD .byte X86_FEATURE_REP_GOOD
.byte copy_page_c_end-copy_page_c .byte .Lcopy_page_end - copy_page
.byte copy_page_c_end-copy_page_c .byte 2b - 1b
.previous
.section .altinstr_replacement,"ax"
copy_page_c:
movl $4096/8,%ecx
rep
movsq
ret
copy_page_c_end:
.previous .previous
arch/x86_64/lib/copy_user.S
浏览文件 @ 8d379dad
...@@ -4,6 +4,9 @@ ...@@ -4,6 +4,9 @@
* Functions to copy from and to user space. * Functions to copy from and to user space.
*/ */
#include <linux/linkage.h>
#include <asm/dwarf2.h>
#define FIX_ALIGNMENT 1 #define FIX_ALIGNMENT 1
#include <asm/current.h> #include <asm/current.h>
...@@ -12,9 +15,8 @@ ...@@ -12,9 +15,8 @@
#include <asm/cpufeature.h> #include <asm/cpufeature.h>
/* Standard copy_to_user with segment limit checking */ /* Standard copy_to_user with segment limit checking */
.globl copy_to_user ENTRY(copy_to_user)
.p2align 4 CFI_STARTPROC
copy_to_user:
GET_THREAD_INFO(%rax) GET_THREAD_INFO(%rax)
movq %rdi,%rcx movq %rdi,%rcx
addq %rdx,%rcx addq %rdx,%rcx
...@@ -25,9 +27,11 @@ copy_to_user: ...@@ -25,9 +27,11 @@ copy_to_user:
.byte 0xe9 /* 32bit jump */ .byte 0xe9 /* 32bit jump */
.long .Lcug-1f .long .Lcug-1f
1: 1:
CFI_ENDPROC
ENDPROC(copy_to_user)
.section .altinstr_replacement,"ax" .section .altinstr_replacement,"ax"
3: .byte 0xe9 /* replacement jmp with 8 bit immediate */ 3: .byte 0xe9 /* replacement jmp with 32 bit immediate */
.long copy_user_generic_c-1b /* offset */ .long copy_user_generic_c-1b /* offset */
.previous .previous
.section .altinstructions,"a" .section .altinstructions,"a"
...@@ -40,9 +44,8 @@ copy_to_user: ...@@ -40,9 +44,8 @@ copy_to_user:
.previous .previous
/* Standard copy_from_user with segment limit checking */ /* Standard copy_from_user with segment limit checking */
.globl copy_from_user ENTRY(copy_from_user)
.p2align 4 CFI_STARTPROC
copy_from_user:
GET_THREAD_INFO(%rax) GET_THREAD_INFO(%rax)
movq %rsi,%rcx movq %rsi,%rcx
addq %rdx,%rcx addq %rdx,%rcx
...@@ -50,10 +53,13 @@ copy_from_user: ...@@ -50,10 +53,13 @@ copy_from_user:
cmpq threadinfo_addr_limit(%rax),%rcx cmpq threadinfo_addr_limit(%rax),%rcx
jae bad_from_user jae bad_from_user
/* FALL THROUGH to copy_user_generic */ /* FALL THROUGH to copy_user_generic */
CFI_ENDPROC
ENDPROC(copy_from_user)
.section .fixup,"ax" .section .fixup,"ax"
/* must zero dest */ /* must zero dest */
bad_from_user: bad_from_user:
CFI_STARTPROC
movl %edx,%ecx movl %edx,%ecx
xorl %eax,%eax xorl %eax,%eax
rep rep
...@@ -61,6 +67,8 @@ bad_from_user: ...@@ -61,6 +67,8 @@ bad_from_user:
bad_to_user: bad_to_user:
movl %edx,%eax movl %edx,%eax
ret ret
CFI_ENDPROC
END(bad_from_user)
.previous .previous
...@@ -75,9 +83,8 @@ bad_to_user: ...@@ -75,9 +83,8 @@ bad_to_user:
* Output: * Output:
* eax uncopied bytes or 0 if successful. * eax uncopied bytes or 0 if successful.
*/ */
.globl copy_user_generic ENTRY(copy_user_generic)
.p2align 4 CFI_STARTPROC
copy_user_generic:
.byte 0x66,0x66,0x90 /* 5 byte nop for replacement jump */ .byte 0x66,0x66,0x90 /* 5 byte nop for replacement jump */
.byte 0x66,0x90 .byte 0x66,0x90
1: 1:
...@@ -95,6 +102,8 @@ copy_user_generic: ...@@ -95,6 +102,8 @@ copy_user_generic:
.previous .previous
.Lcug: .Lcug:
pushq %rbx pushq %rbx
CFI_ADJUST_CFA_OFFSET 8
CFI_REL_OFFSET rbx, 0
xorl %eax,%eax /*zero for the exception handler */ xorl %eax,%eax /*zero for the exception handler */
#ifdef FIX_ALIGNMENT #ifdef FIX_ALIGNMENT
...@@ -168,9 +177,13 @@ copy_user_generic: ...@@ -168,9 +177,13 @@ copy_user_generic:
decl %ecx decl %ecx
jnz .Lloop_1 jnz .Lloop_1
CFI_REMEMBER_STATE
.Lende: .Lende:
popq %rbx popq %rbx
CFI_ADJUST_CFA_OFFSET -8
CFI_RESTORE rbx
ret ret
CFI_RESTORE_STATE
#ifdef FIX_ALIGNMENT #ifdef FIX_ALIGNMENT
/* align destination */ /* align destination */
...@@ -261,6 +274,9 @@ copy_user_generic: ...@@ -261,6 +274,9 @@ copy_user_generic:
.Le_zero: .Le_zero:
movq %rdx,%rax movq %rdx,%rax
jmp .Lende jmp .Lende
CFI_ENDPROC
ENDPROC(copy_user_generic)
/* Some CPUs run faster using the string copy instructions. /* Some CPUs run faster using the string copy instructions.
This is also a lot simpler. Use them when possible. This is also a lot simpler. Use them when possible.
...@@ -282,6 +298,7 @@ copy_user_generic: ...@@ -282,6 +298,7 @@ copy_user_generic:
* this please consider this. * this please consider this.
*/ */
copy_user_generic_c: copy_user_generic_c:
CFI_STARTPROC
movl %edx,%ecx movl %edx,%ecx
shrl $3,%ecx shrl $3,%ecx
andl $7,%edx andl $7,%edx
...@@ -294,6 +311,8 @@ copy_user_generic_c: ...@@ -294,6 +311,8 @@ copy_user_generic_c:
ret ret
3: lea (%rdx,%rcx,8),%rax 3: lea (%rdx,%rcx,8),%rax
ret ret
CFI_ENDPROC
END(copy_user_generic_c)
.section __ex_table,"a" .section __ex_table,"a"
.quad 1b,3b .quad 1b,3b
......
arch/x86_64/lib/csum-copy.S
浏览文件 @ 8d379dad
...@@ -5,8 +5,9 @@ ...@@ -5,8 +5,9 @@
* License. See the file COPYING in the main directory of this archive * License. See the file COPYING in the main directory of this archive
* for more details. No warranty for anything given at all. * for more details. No warranty for anything given at all.
*/ */
#include <linux/linkage.h> #include <linux/linkage.h>
#include <asm/errno.h> #include <asm/dwarf2.h>
#include <asm/errno.h>
/* /*
* Checksum copy with exception handling. * Checksum copy with exception handling.
...@@ -53,19 +54,24 @@ ...@@ -53,19 +54,24 @@
.endm .endm
.globl csum_partial_copy_generic ENTRY(csum_partial_copy_generic)
.p2align 4 CFI_STARTPROC
csum_partial_copy_generic:
cmpl $3*64,%edx cmpl $3*64,%edx
jle .Lignore jle .Lignore
.Lignore: .Lignore:
subq $7*8,%rsp subq $7*8,%rsp
CFI_ADJUST_CFA_OFFSET 7*8
movq %rbx,2*8(%rsp) movq %rbx,2*8(%rsp)
CFI_REL_OFFSET rbx, 2*8
movq %r12,3*8(%rsp) movq %r12,3*8(%rsp)
CFI_REL_OFFSET r12, 3*8
movq %r14,4*8(%rsp) movq %r14,4*8(%rsp)
CFI_REL_OFFSET r14, 4*8
movq %r13,5*8(%rsp) movq %r13,5*8(%rsp)
CFI_REL_OFFSET r13, 5*8
movq %rbp,6*8(%rsp) movq %rbp,6*8(%rsp)
CFI_REL_OFFSET rbp, 6*8
movq %r8,(%rsp) movq %r8,(%rsp)
movq %r9,1*8(%rsp) movq %r9,1*8(%rsp)
...@@ -208,14 +214,22 @@ csum_partial_copy_generic: ...@@ -208,14 +214,22 @@ csum_partial_copy_generic:
addl %ebx,%eax addl %ebx,%eax
adcl %r9d,%eax /* carry */ adcl %r9d,%eax /* carry */
CFI_REMEMBER_STATE
.Lende: .Lende:
movq 2*8(%rsp),%rbx movq 2*8(%rsp),%rbx
CFI_RESTORE rbx
movq 3*8(%rsp),%r12 movq 3*8(%rsp),%r12
CFI_RESTORE r12
movq 4*8(%rsp),%r14 movq 4*8(%rsp),%r14
CFI_RESTORE r14
movq 5*8(%rsp),%r13 movq 5*8(%rsp),%r13
CFI_RESTORE r13
movq 6*8(%rsp),%rbp movq 6*8(%rsp),%rbp
CFI_RESTORE rbp
addq $7*8,%rsp addq $7*8,%rsp
CFI_ADJUST_CFA_OFFSET -7*8
ret ret
CFI_RESTORE_STATE
/* Exception handlers. Very simple, zeroing is done in the wrappers */ /* Exception handlers. Very simple, zeroing is done in the wrappers */
.Lbad_source: .Lbad_source:
...@@ -231,3 +245,5 @@ csum_partial_copy_generic: ...@@ -231,3 +245,5 @@ csum_partial_copy_generic:
jz .Lende jz .Lende
movl $-EFAULT,(%rax) movl $-EFAULT,(%rax)
jmp .Lende jmp .Lende
CFI_ENDPROC
ENDPROC(csum_partial_copy_generic)
arch/x86_64/lib/getuser.S
浏览文件 @ 8d379dad
...@@ -27,25 +27,26 @@ ...@@ -27,25 +27,26 @@
*/ */
#include <linux/linkage.h> #include <linux/linkage.h>
#include <asm/dwarf2.h>
#include <asm/page.h> #include <asm/page.h>
#include <asm/errno.h> #include <asm/errno.h>
#include <asm/asm-offsets.h> #include <asm/asm-offsets.h>
#include <asm/thread_info.h> #include <asm/thread_info.h>
.text .text
.p2align 4 ENTRY(__get_user_1)
.globl __get_user_1 CFI_STARTPROC
__get_user_1:
GET_THREAD_INFO(%r8) GET_THREAD_INFO(%r8)
cmpq threadinfo_addr_limit(%r8),%rcx cmpq threadinfo_addr_limit(%r8),%rcx
jae bad_get_user jae bad_get_user
1: movzb (%rcx),%edx 1: movzb (%rcx),%edx
xorl %eax,%eax xorl %eax,%eax
ret ret
CFI_ENDPROC
ENDPROC(__get_user_1)
.p2align 4 ENTRY(__get_user_2)
.globl __get_user_2 CFI_STARTPROC
__get_user_2:
GET_THREAD_INFO(%r8) GET_THREAD_INFO(%r8)
addq $1,%rcx addq $1,%rcx
jc 20f jc 20f
...@@ -57,10 +58,11 @@ __get_user_2: ...@@ -57,10 +58,11 @@ __get_user_2:
ret ret
20: decq %rcx 20: decq %rcx
jmp bad_get_user jmp bad_get_user
CFI_ENDPROC
ENDPROC(__get_user_2)
.p2align 4 ENTRY(__get_user_4)
.globl __get_user_4 CFI_STARTPROC
__get_user_4:
GET_THREAD_INFO(%r8) GET_THREAD_INFO(%r8)
addq $3,%rcx addq $3,%rcx
jc 30f jc 30f
...@@ -72,10 +74,11 @@ __get_user_4: ...@@ -72,10 +74,11 @@ __get_user_4:
ret ret
30: subq $3,%rcx 30: subq $3,%rcx
jmp bad_get_user jmp bad_get_user
CFI_ENDPROC
ENDPROC(__get_user_4)
.p2align 4 ENTRY(__get_user_8)
.globl __get_user_8 CFI_STARTPROC
__get_user_8:
GET_THREAD_INFO(%r8) GET_THREAD_INFO(%r8)
addq $7,%rcx addq $7,%rcx
jc 40f jc 40f
...@@ -87,11 +90,16 @@ __get_user_8: ...@@ -87,11 +90,16 @@ __get_user_8:
ret ret
40: subq $7,%rcx 40: subq $7,%rcx
jmp bad_get_user jmp bad_get_user
CFI_ENDPROC
ENDPROC(__get_user_8)
bad_get_user: bad_get_user:
CFI_STARTPROC
xorl %edx,%edx xorl %edx,%edx
movq $(-EFAULT),%rax movq $(-EFAULT),%rax
ret ret
CFI_ENDPROC
END(bad_get_user)
.section __ex_table,"a" .section __ex_table,"a"
.quad 1b,bad_get_user .quad 1b,bad_get_user
......
arch/x86_64/lib/iomap_copy.S
浏览文件 @ 8d379dad
...@@ -15,12 +15,16 @@ ...@@ -15,12 +15,16 @@
* Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301, USA. * Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301, USA.
*/ */
#include <linux/linkage.h>
#include <asm/dwarf2.h>
/* /*
* override generic version in lib/iomap_copy.c * override generic version in lib/iomap_copy.c
*/ */
.globl __iowrite32_copy ENTRY(__iowrite32_copy)
.p2align 4 CFI_STARTPROC
__iowrite32_copy:
movl %edx,%ecx movl %edx,%ecx
rep movsd rep movsd
ret ret
CFI_ENDPROC
ENDPROC(__iowrite32_copy)
arch/x86_64/lib/memcpy.S
浏览文件 @ 8d379dad
/* Copyright 2002 Andi Kleen */ /* Copyright 2002 Andi Kleen */
#include <asm/cpufeature.h> #include <linux/config.h>
#include <linux/linkage.h>
#include <asm/dwarf2.h>
#include <asm/cpufeature.h>
/* /*
* memcpy - Copy a memory block. * memcpy - Copy a memory block.
* *
...@@ -13,12 +17,26 @@ ...@@ -13,12 +17,26 @@
* rax original destination * rax original destination
*/ */
.globl __memcpy ALIGN
.globl memcpy memcpy_c:
.p2align 4 CFI_STARTPROC
__memcpy: movq %rdi,%rax
memcpy: movl %edx,%ecx
shrl $3,%ecx
andl $7,%edx
rep movsq
movl %edx,%ecx
rep movsb
ret
CFI_ENDPROC
ENDPROC(memcpy_c)
ENTRY(__memcpy)
ENTRY(memcpy)
CFI_STARTPROC
pushq %rbx pushq %rbx
CFI_ADJUST_CFA_OFFSET 8
CFI_REL_OFFSET rbx, 0
movq %rdi,%rax movq %rdi,%rax
movl %edx,%ecx movl %edx,%ecx
...@@ -86,36 +104,27 @@ memcpy: ...@@ -86,36 +104,27 @@ memcpy:
.Lende: .Lende:
popq %rbx popq %rbx
CFI_ADJUST_CFA_OFFSET -8
CFI_RESTORE rbx
ret ret
.Lfinal: .Lfinal:
CFI_ENDPROC
ENDPROC(memcpy)
ENDPROC(__memcpy)
/* Some CPUs run faster using the string copy instructions. /* Some CPUs run faster using the string copy instructions.
It is also a lot simpler. Use this when possible */ It is also a lot simpler. Use this when possible */
.section .altinstr_replacement,"ax"
1: .byte 0xeb /* jmp <disp8> */
.byte (memcpy_c - memcpy) - (2f - 1b) /* offset */
2:
.previous
.section .altinstructions,"a" .section .altinstructions,"a"
.align 8 .align 8
.quad memcpy .quad memcpy
.quad memcpy_c .quad 1b
.byte X86_FEATURE_REP_GOOD .byte X86_FEATURE_REP_GOOD
.byte .Lfinal-memcpy .byte .Lfinal - memcpy
.byte memcpy_c_end-memcpy_c .byte 2b - 1b
.previous
.section .altinstr_replacement,"ax"
/* rdi destination
* rsi source
* rdx count
*/
memcpy_c:
movq %rdi,%rax
movl %edx,%ecx
shrl $3,%ecx
andl $7,%edx
rep
movsq
movl %edx,%ecx
rep
movsb
ret
memcpy_c_end:
.previous .previous
arch/x86_64/lib/memset.S
浏览文件 @ 8d379dad
/* Copyright 2002 Andi Kleen, SuSE Labs */ /* Copyright 2002 Andi Kleen, SuSE Labs */
#include <linux/config.h>
#include <linux/linkage.h>
#include <asm/dwarf2.h>
/* /*
* ISO C memset - set a memory block to a byte value. * ISO C memset - set a memory block to a byte value.
* *
...@@ -8,11 +13,29 @@ ...@@ -8,11 +13,29 @@
* *
* rax original destination * rax original destination
*/ */
.globl __memset ALIGN
.globl memset memset_c:
.p2align 4 CFI_STARTPROC
memset: movq %rdi,%r9
__memset: movl %edx,%r8d
andl $7,%r8d
movl %edx,%ecx
shrl $3,%ecx
/* expand byte value */
movzbl %sil,%esi
movabs $0x0101010101010101,%rax
mulq %rsi /* with rax, clobbers rdx */
rep stosq
movl %r8d,%ecx
rep stosb
movq %r9,%rax
ret
CFI_ENDPROC
ENDPROC(memset_c)
ENTRY(memset)
ENTRY(__memset)
CFI_STARTPROC
movq %rdi,%r10 movq %rdi,%r10
movq %rdx,%r11 movq %rdx,%r11
...@@ -25,6 +48,7 @@ __memset: ...@@ -25,6 +48,7 @@ __memset:
movl %edi,%r9d movl %edi,%r9d
andl $7,%r9d andl $7,%r9d
jnz .Lbad_alignment jnz .Lbad_alignment
CFI_REMEMBER_STATE
.Lafter_bad_alignment: .Lafter_bad_alignment:
movl %r11d,%ecx movl %r11d,%ecx
...@@ -75,6 +99,7 @@ __memset: ...@@ -75,6 +99,7 @@ __memset:
movq %r10,%rax movq %r10,%rax
ret ret
CFI_RESTORE_STATE
.Lbad_alignment: .Lbad_alignment:
cmpq $7,%r11 cmpq $7,%r11
jbe .Lhandle_7 jbe .Lhandle_7
...@@ -84,42 +109,26 @@ __memset: ...@@ -84,42 +109,26 @@ __memset:
addq %r8,%rdi addq %r8,%rdi
subq %r8,%r11 subq %r8,%r11
jmp .Lafter_bad_alignment jmp .Lafter_bad_alignment
.Lfinal:
CFI_ENDPROC
ENDPROC(memset)
ENDPROC(__memset)
/* Some CPUs run faster using the string instructions. /* Some CPUs run faster using the string instructions.
It is also a lot simpler. Use this when possible */ It is also a lot simpler. Use this when possible */
#include <asm/cpufeature.h> #include <asm/cpufeature.h>
.section .altinstr_replacement,"ax"
1: .byte 0xeb /* jmp <disp8> */
.byte (memset_c - memset) - (2f - 1b) /* offset */
2:
.previous
.section .altinstructions,"a" .section .altinstructions,"a"
.align 8 .align 8
.quad memset .quad memset
.quad memset_c .quad 1b
.byte X86_FEATURE_REP_GOOD .byte X86_FEATURE_REP_GOOD
.byte memset_c_end-memset_c .byte .Lfinal - memset
.byte memset_c_end-memset_c .byte 2b - 1b
.previous
.section .altinstr_replacement,"ax"
/* rdi destination
* rsi value
* rdx count
*/
memset_c:
movq %rdi,%r9
movl %edx,%r8d
andl $7,%r8d
movl %edx,%ecx
shrl $3,%ecx
/* expand byte value */
movzbl %sil,%esi
movabs $0x0101010101010101,%rax
mulq %rsi /* with rax, clobbers rdx */
rep
stosq
movl %r8d,%ecx
rep
stosb
movq %r9,%rax
ret
memset_c_end:
.previous .previous
arch/x86_64/lib/putuser.S
浏览文件 @ 8d379dad
...@@ -25,25 +25,26 @@ ...@@ -25,25 +25,26 @@
*/ */
#include <linux/linkage.h> #include <linux/linkage.h>
#include <asm/dwarf2.h>
#include <asm/page.h> #include <asm/page.h>
#include <asm/errno.h> #include <asm/errno.h>
#include <asm/asm-offsets.h> #include <asm/asm-offsets.h>
#include <asm/thread_info.h> #include <asm/thread_info.h>
.text .text
.p2align 4 ENTRY(__put_user_1)
.globl __put_user_1 CFI_STARTPROC
__put_user_1:
GET_THREAD_INFO(%r8) GET_THREAD_INFO(%r8)
cmpq threadinfo_addr_limit(%r8),%rcx cmpq threadinfo_addr_limit(%r8),%rcx
jae bad_put_user jae bad_put_user
1: movb %dl,(%rcx) 1: movb %dl,(%rcx)
xorl %eax,%eax xorl %eax,%eax
ret ret
CFI_ENDPROC
ENDPROC(__put_user_1)
.p2align 4 ENTRY(__put_user_2)
.globl __put_user_2 CFI_STARTPROC
__put_user_2:
GET_THREAD_INFO(%r8) GET_THREAD_INFO(%r8)
addq $1,%rcx addq $1,%rcx
jc 20f jc 20f
...@@ -55,10 +56,11 @@ __put_user_2: ...@@ -55,10 +56,11 @@ __put_user_2:
ret ret
20: decq %rcx 20: decq %rcx
jmp bad_put_user jmp bad_put_user
CFI_ENDPROC
ENDPROC(__put_user_2)
.p2align 4 ENTRY(__put_user_4)
.globl __put_user_4 CFI_STARTPROC
__put_user_4:
GET_THREAD_INFO(%r8) GET_THREAD_INFO(%r8)
addq $3,%rcx addq $3,%rcx
jc 30f jc 30f
...@@ -70,10 +72,11 @@ __put_user_4: ...@@ -70,10 +72,11 @@ __put_user_4:
ret ret
30: subq $3,%rcx 30: subq $3,%rcx
jmp bad_put_user jmp bad_put_user
CFI_ENDPROC
ENDPROC(__put_user_4)
.p2align 4 ENTRY(__put_user_8)
.globl __put_user_8 CFI_STARTPROC
__put_user_8:
GET_THREAD_INFO(%r8) GET_THREAD_INFO(%r8)
addq $7,%rcx addq $7,%rcx
jc 40f jc 40f
...@@ -85,10 +88,15 @@ __put_user_8: ...@@ -85,10 +88,15 @@ __put_user_8:
ret ret
40: subq $7,%rcx 40: subq $7,%rcx
jmp bad_put_user jmp bad_put_user
CFI_ENDPROC
ENDPROC(__put_user_8)
bad_put_user: bad_put_user:
CFI_STARTPROC
movq $(-EFAULT),%rax movq $(-EFAULT),%rax
ret ret
CFI_ENDPROC
END(bad_put_user)
.section __ex_table,"a" .section __ex_table,"a"
.quad 1b,bad_put_user .quad 1b,bad_put_user
......
Markdown is supported
0% 或 .
You are about to add 0 people to the discussion. Proceed with caution.
先完成此消息的编辑！
想要评论请 注册 或