linux内核中的copy_to_user和copy_from_user(一)
Kernel version:2.6.14
CPU architecture:ARM920T
Author:ce123(http://blog.csdn.net/ce123)
在学习Linux内核驱动的时候,经常会碰到copy_from_user和copy_to_user这两个函数,设备驱动程序中的ioctl函数就经常会用到。这两个函数负责在用户空间和内核空间传递数据。首先看看它们的定义(linux/include/asm-arm/uaccess.h),先看copy_from_user:
static inline unsigned long copy_from_user(void *to, const void __user *from, unsigned long n)
{
if (access_ok(VERIFY_READ, from, n))
n = __arch_copy_from_user(to, from, n);
else /* security hole - plug it */
memzero(to, n);
return n;
}
先看函数的三个参数:*to是内核空间的指针,*from是用户空间指针,n表示从用户空间想内核空间拷贝数据的字节数。如果成功执行拷贝操作,则返回0,否则返回还没有完成拷贝的字节数。
这个函数从结构上来分析,其实都可以分为两个部分:access_ok用来对用户空间的地址指针from作某种有效性检验,这个宏和体系结构相关,在arm平台上为(linux/include/asm-arm/uaccess.h):
#define __range_ok(addr,size) ({ \
unsigned long flag, sum; \
__chk_user_ptr(addr); \
__asm__("adds %1, %2, %3; sbcccs %1, %1, %0; movcc %0, #0" \
: "=&r" (flag), "=&r" (sum) \
: "r" (addr), "Ir" (size), "0" (current_thread_info()->addr_limit) \
: "cc"); \
flag; })
#define access_ok(type,addr,size) (__range_ok(addr,size) == 0)
可以看到access_ok中第一个参数type并没有用到,__range_ok的作用在于判断addr+size之后是否还在进程的用户空间范围之内。下面我们具体看一下。这段代码涉及到GCC内联汇编,不懂的朋友可以先看看这篇博客(http://blog.csdn.net/ce123/article/details/8209702)。请看具体的定义(linux/compiler.h):
#ifdef __CHECKER__
...
extern void __chk_user_ptr(void __user *);
extern void __chk_io_ptr(void __iomem *);
#else
...
# define __chk_user_ptr(x) (void)0
# define __chk_io_ptr(x) (void)0
...
#endif
(3)接下来是汇编: 如果上面的加法指令进位了(C=1),则以下的指令都不执行,flag就为初始值current_thread_info()->addr_limit(非0),并返回。
如果没有进位(C=0),就执行下面的指令:
sbcccs %1, %1, %0
sum = sum - flag - 1,也就是(addr + size) - (current_thread_info()->addr_limit) - 1,操作影响符号位。
如果(addr + size) >= (current_thread_info()->addr_limit) - 1,则C=1
如果(addr + size) < (current_thread_info()->addr_limit) - 1,则C=0
当C=0的时候执行以下指令,否则跳过(flag非零)。
movcc %0, #0
flag = 0,给flag赋值0。
综上所述:__range_ok宏其实等价于:
#define USER(x...) \
9999: x; \
.section __ex_table,"a"; \
.align 3; \
.long 9999b,9001f; \
.previous
.long 9999b,9001f;
其中9999b对应标号9999处的指令,9001f是9001处的指令,是9999b处指令的修复指令。这样,当标号9999处发生缺页异常时,系统将调用do_page_fault提交物理页面,然后跳到9001继续执行。
/* Prototype: unsigned long __arch_copy_from_user(void *to,const void *from,unsigned long n);
* Purpose : copy a block from user memory to kernel memory
* Params : to - kernel memory
* : from - user memory
* : n - number of bytes to copy
* Returns : Number of bytes NOT copied.
*/
.cfu_dest_not_aligned:
rsb ip, ip, #4
cmp ip, #2
USER( ldrbt r3, [r1], #1) @ May fault
strb r3, [r0], #1
USER( ldrgebt r3, [r1], #1) @ May fault
strgeb r3, [r0], #1
USER( ldrgtbt r3, [r1], #1) @ May fault
strgtb r3, [r0], #1
sub r2, r2, ip
b .cfu_dest_aligned
ENTRY(__arch_copy_from_user)
stmfd sp!, {r0, r2, r4 - r7, lr}
cmp r2, #4
blt .cfu_not_enough
PLD( pld [r1, #0] )
PLD( pld [r0, #0] )
ands ip, r0, #3
bne .cfu_dest_not_aligned
.cfu_dest_aligned:
ands ip, r1, #3
bne .cfu_src_not_aligned
/*
* Seeing as there has to be at least 8 bytes to copy, we can
* copy one word, and force a user-mode page fault...
*/
.cfu_0fupi: subs r2, r2, #4
addmi ip, r2, #4
bmi .cfu_0nowords
USER( ldrt r3, [r1], #4)
str r3, [r0], #4
mov ip, r1, lsl #32 - PAGE_SHIFT @ On each page, use a ld/st??t instruction
rsb ip, ip, #0
movs ip, ip, lsr #32 - PAGE_SHIFT
beq .cfu_0fupi
/*
* ip = max no. of bytes to copy before needing another "strt" insn
*/
cmp r2, ip
movlt ip, r2
sub r2, r2, ip
subs ip, ip, #32
blt .cfu_0rem8lp
PLD( pld [r1, #28] )
PLD( pld [r0, #28] )
PLD( subs ip, ip, #64 )
PLD( blt .cfu_0cpynopld )
PLD( pld [r1, #60] )
PLD( pld [r0, #60] )
.cfu_0cpy8lp:
PLD( pld [r1, #92] )
PLD( pld [r0, #92] )
.cfu_0cpynopld: ldmia r1!, {r3 - r6} @ Shouldnt fault
stmia r0!, {r3 - r6}
ldmia r1!, {r3 - r6} @ Shouldnt fault
subs ip, ip, #32
stmia r0!, {r3 - r6}
bpl .cfu_0cpy8lp
PLD( cmn ip, #64 )
PLD( bge .cfu_0cpynopld )
PLD( add ip, ip, #64 )
.cfu_0rem8lp: cmn ip, #16
ldmgeia r1!, {r3 - r6} @ Shouldnt fault
stmgeia r0!, {r3 - r6}
tst ip, #8
ldmneia r1!, {r3 - r4} @ Shouldnt fault
stmneia r0!, {r3 - r4}
tst ip, #4
ldrnet r3, [r1], #4 @ Shouldnt fault
strne r3, [r0], #4
ands ip, ip, #3
beq .cfu_0fupi
.cfu_0nowords: teq ip, #0
beq .cfu_finished
.cfu_nowords: cmp ip, #2
USER( ldrbt r3, [r1], #1) @ May fault
strb r3, [r0], #1
USER( ldrgebt r3, [r1], #1) @ May fault
strgeb r3, [r0], #1
USER( ldrgtbt r3, [r1], #1) @ May fault
strgtb r3, [r0], #1
b .cfu_finished
.cfu_not_enough:
movs ip, r2
bne .cfu_nowords
.cfu_finished: mov r0, #0
add sp, sp, #8
LOADREGS(fd,sp!,{r4 - r7, pc})
.cfu_src_not_aligned:
bic r1, r1, #3
USER( ldrt r7, [r1], #4) @ May fault
cmp ip, #2
bgt .cfu_3fupi
beq .cfu_2fupi
.cfu_1fupi: subs r2, r2, #4
addmi ip, r2, #4
bmi .cfu_1nowords
mov r3, r7, pull #8
USER( ldrt r7, [r1], #4) @ May fault
orr r3, r3, r7, push #24
str r3, [r0], #4
mov ip, r1, lsl #32 - PAGE_SHIFT
rsb ip, ip, #0
movs ip, ip, lsr #32 - PAGE_SHIFT
beq .cfu_1fupi
cmp r2, ip
movlt ip, r2
sub r2, r2, ip
subs ip, ip, #16
blt .cfu_1rem8lp
PLD( pld [r1, #12] )
PLD( pld [r0, #12] )
PLD( subs ip, ip, #32 )
PLD( blt .cfu_1cpynopld )
PLD( pld [r1, #28] )
PLD( pld [r0, #28] )
.cfu_1cpy8lp:
PLD( pld [r1, #44] )
PLD( pld [r0, #44] )
.cfu_1cpynopld: mov r3, r7, pull #8
ldmia r1!, {r4 - r7} @ Shouldnt fault
subs ip, ip, #16
orr r3, r3, r4, push #24
mov r4, r4, pull #8
orr r4, r4, r5, push #24
mov r5, r5, pull #8
orr r5, r5, r6, push #24
mov r6, r6, pull #8
orr r6, r6, r7, push #24
stmia r0!, {r3 - r6}
bpl .cfu_1cpy8lp
PLD( cmn ip, #32 )
PLD( bge .cfu_1cpynopld )
PLD( add ip, ip, #32 )
.cfu_1rem8lp: tst ip, #8
movne r3, r7, pull #8
ldmneia r1!, {r4, r7} @ Shouldnt fault
orrne r3, r3, r4, push #24
movne r4, r4, pull #8
orrne r4, r4, r7, push #24
stmneia r0!, {r3 - r4}
tst ip, #4
movne r3, r7, pull #8
USER( ldrnet r7, [r1], #4) @ May fault
orrne r3, r3, r7, push #24
strne r3, [r0], #4
ands ip, ip, #3
beq .cfu_1fupi
.cfu_1nowords: mov r3, r7, get_byte_1
teq ip, #0
beq .cfu_finished
cmp ip, #2
strb r3, [r0], #1
movge r3, r7, get_byte_2
strgeb r3, [r0], #1
movgt r3, r7, get_byte_3
strgtb r3, [r0], #1
b .cfu_finished
.cfu_2fupi: subs r2, r2, #4
addmi ip, r2, #4
bmi .cfu_2nowords
mov r3, r7, pull #16
USER( ldrt r7, [r1], #4) @ May fault
orr r3, r3, r7, push #16
str r3, [r0], #4
mov ip, r1, lsl #32 - PAGE_SHIFT
rsb ip, ip, #0
movs ip, ip, lsr #32 - PAGE_SHIFT
beq .cfu_2fupi
cmp r2, ip
movlt ip, r2
sub r2, r2, ip
subs ip, ip, #16
blt .cfu_2rem8lp
PLD( pld [r1, #12] )
PLD( pld [r0, #12] )
PLD( subs ip, ip, #32 )
PLD( blt .cfu_2cpynopld )
PLD( pld [r1, #28] )
PLD( pld [r0, #28] )
.cfu_2cpy8lp:
PLD( pld [r1, #44] )
PLD( pld [r0, #44] )
.cfu_2cpynopld: mov r3, r7, pull #16
ldmia r1!, {r4 - r7} @ Shouldnt fault
subs ip, ip, #16
orr r3, r3, r4, push #16
mov r4, r4, pull #16
orr r4, r4, r5, push #16
mov r5, r5, pull #16
orr r5, r5, r6, push #16
mov r6, r6, pull #16
orr r6, r6, r7, push #16
stmia r0!, {r3 - r6}
bpl .cfu_2cpy8lp
PLD( cmn ip, #32 )
PLD( bge .cfu_2cpynopld )
PLD( add ip, ip, #32 )
.cfu_2rem8lp: tst ip, #8
movne r3, r7, pull #16
ldmneia r1!, {r4, r7} @ Shouldnt fault
orrne r3, r3, r4, push #16
movne r4, r4, pull #16
orrne r4, r4, r7, push #16
stmneia r0!, {r3 - r4}
tst ip, #4
movne r3, r7, pull #16
USER( ldrnet r7, [r1], #4) @ May fault
orrne r3, r3, r7, push #16
strne r3, [r0], #4
ands ip, ip, #3
beq .cfu_2fupi
.cfu_2nowords: mov r3, r7, get_byte_2
teq ip, #0
beq .cfu_finished
cmp ip, #2
strb r3, [r0], #1
movge r3, r7, get_byte_3
strgeb r3, [r0], #1
USER( ldrgtbt r3, [r1], #0) @ May fault
strgtb r3, [r0], #1
b .cfu_finished
.cfu_3fupi: subs r2, r2, #4
addmi ip, r2, #4
bmi .cfu_3nowords
mov r3, r7, pull #24
USER( ldrt r7, [r1], #4) @ May fault
orr r3, r3, r7, push #8
str r3, [r0], #4
mov ip, r1, lsl #32 - PAGE_SHIFT
rsb ip, ip, #0
movs ip, ip, lsr #32 - PAGE_SHIFT
beq .cfu_3fupi
cmp r2, ip
movlt ip, r2
sub r2, r2, ip
subs ip, ip, #16
blt .cfu_3rem8lp
PLD( pld [r1, #12] )
PLD( pld [r0, #12] )
PLD( subs ip, ip, #32 )
PLD( blt .cfu_3cpynopld )
PLD( pld [r1, #28] )
PLD( pld [r0, #28] )
.cfu_3cpy8lp:
PLD( pld [r1, #44] )
PLD( pld [r0, #44] )
.cfu_3cpynopld: mov r3, r7, pull #24
ldmia r1!, {r4 - r7} @ Shouldnt fault
orr r3, r3, r4, push #8
mov r4, r4, pull #24
orr r4, r4, r5, push #8
mov r5, r5, pull #24
orr r5, r5, r6, push #8
mov r6, r6, pull #24
orr r6, r6, r7, push #8
stmia r0!, {r3 - r6}
subs ip, ip, #16
bpl .cfu_3cpy8lp
PLD( cmn ip, #32 )
PLD( bge .cfu_3cpynopld )
PLD( add ip, ip, #32 )
.cfu_3rem8lp: tst ip, #8
movne r3, r7, pull #24
ldmneia r1!, {r4, r7} @ Shouldnt fault
orrne r3, r3, r4, push #8
movne r4, r4, pull #24
orrne r4, r4, r7, push #8
stmneia r0!, {r3 - r4}
tst ip, #4
movne r3, r7, pull #24
USER( ldrnet r7, [r1], #4) @ May fault
orrne r3, r3, r7, push #8
strne r3, [r0], #4
ands ip, ip, #3
beq .cfu_3fupi
.cfu_3nowords: mov r3, r7, get_byte_3
teq ip, #0
beq .cfu_finished
cmp ip, #2
strb r3, [r0], #1
USER( ldrgebt r3, [r1], #1) @ May fault
strgeb r3, [r0], #1
USER( ldrgtbt r3, [r1], #1) @ May fault
strgtb r3, [r0], #1
b .cfu_finished
.section .fixup,"ax"
.align 0
/*
* We took an exception. r0 contains a pointer to
* the byte not copied.
*/
9001: ldr r2, [sp], #4 @ void *to
sub r2, r0, r2 @ bytes copied
ldr r1, [sp], #4 @ unsigned long count
subs r4, r1, r2 @ bytes left to copy
movne r1, r4
blne __memzero
mov r0, r4
LOADREGS(fd,sp!, {r4 - r7, pc})
.previous
我们将在另一篇博文中详细分析该函数。