ARM汇编伪指令.word

 ARM汇编伪指令 .word

    经常碰到那些以“.”打头的一些令人头疼的伪指令，

    至于.globl _start .balign .align .data .text等等就算了，最最bt的如下：

    _undefined_instruction: .word undefined_instruction

    这个.word令人费解。网上的技术人员都不屑回答，说请参考GNU ASM。我去看了，对于.word解释如下：

    http://tigcc.ticalc.org/doc/gnuasm.html#SEC49

    .word

    Syntax: .word expressions

    This directive expects zero or more expressions, of any section, separated by commas. For each expression, as emits a 16-bit number for this target.

    以及as.info文档：

    7.92 .word expressions

    This directive expects zero or more expressions, of any section, separated by commas.

    The size of the number emitted, and its byte order, depend on what target computer

    the assembly is for.

    Warning: Special Treatment to support Compilers

    Machines with a 32-bit address space, but that do less than 32-bit addressing, require

    the following special treatment. If the machine of interest to you does 32-bit addressing

    (or doesn’t require it; see Chapter 8 [Machine Dependencies], page 61), you can ignore this

    issue.

    In order to assemble compiler output into something that works, as occasionally does

    strange things to ‘.word’ directives. Directives of the form ‘.word sym1-sym2’ are often

    emitted by compilers as part of jump tables. Therefore, when as assembles a directive of

    the form ‘.word sym1-sym2’, and the difference between sym1 and sym2 does not fit in 16

    bits, as creates a secondary jump table, immediately before the next label. This secondary

    jump table is preceded by a short-jump to the first byte after the secondary table. This

    short-jump prevents the flow of control from accidentally falling into the new table. Inside

    the table is a long-jump to sym2. The original ‘.word’ contains sym1 minus the address of

    the long-jump to sym2.

    If there were several occurrences of ‘.word sym1-sym2’ before the secondary jump table,

    all of them are adjusted. If there was a ‘.word sym3-sym4’, that also did not fit in sixteen

    bits, a long-jump to sym4 is included in the secondary jump table, and the .word directives

    are adjusted to contain sym3 minus the address of the long-jump to sym4; and so on, for as

    many entries in the original jump table as necessary.

    看了以后仍然一头雾水。

    我把bin文件反汇编，想通过这种方法来找找这个.word究竟干什么。

    原汇编程序：(start.S)

    .globl _start

    _start: b reset

    ldr pc, _undefined_instruction

    ldr pc, _software_interrupt

    ldr pc, _prefetch_abort

    ldr pc, _data_abort

    ldr pc, _not_used

    ldr pc, _irq

    ldr pc, _fiq

    _undefined_instruction: .word undefined_instruction

    _software_interrupt: .word software_interrupt

    _prefetch_abort: .word prefetch_abort

    _data_abort: .word data_abort

    _not_used: .word not_used

    _irq: .word irq

    _fiq: .word fiq

    .balignl 16,0xdeadbeef

    _TEXT_BASE:

    .word TEXT_BASE

    .globl _armboot_start

    _armboot_start:

    .word _start

    .globl _bss_start

    _bss_start:

    .word __bss_start

    .globl _bss_end

    _bss_end:

    .word _end

    reset:

    /*

    * set the cpu to SVC32 mode

    */

    mrs r0,cpsr

    bic r0,r0,#0x1f

    orr r0,r0,#0xd3

    msr cpsr,r0

    对应的反汇编：

    00000000 [0xea000012] b 0x50

    00000004 [0xe59ff014] ldr pc,0x00000020 ; = #0x33f80140

    00000008 [0xe59ff014] ldr pc,0x00000024 ; = #0x33f801a0

    0000000c [0xe59ff014] ldr pc,0x00000028 ; = #0x33f80200

    00000010 [0xe59ff014] ldr pc,0x0000002c ; = #0x33f80260

    00000014 [0xe59ff014] ldr pc,0x00000030 ; = #0x33f802c0

    00000018 [0xe59ff014] ldr pc,0x00000034 ; = #0x33f80320

    0000001c [0xe59ff014] ldr pc,0x00000038 ; = #0x33f80380

    00000020 [0x33f80140] mvnccs r0,#0x10 ; ? rn = 0x8

    00000024 [0x33f801a0] mvnccs r0,#0x28 ; ? rn = 0x8

    00000028 [0x33f80200] mvnccs r0,#0, 4 ; ? rn = 0x8

    0000002c [0x33f80260] mvnccs r0,#6 ; ? rn = 0x8

    00000030 [0x33f802c0] mvnccs r0,#0xc ; ? rn = 0x8

    00000034 [0x33f80320] mvnccs r0,#0x80000000 ; ? rn = 0x8

    00000038 [0x33f80380] mvnccs r0,#2 ; ? rn = 0x8

    0000003c [0xdeadbeef] cdple p14,0xa,c11,c13,c15,7

    00000040 [0x33f80000] mvnccs r0,#0 ; ? rn = 0x8

    00000044 [0x33f80000] mvnccs r0,#0 ; ? rn = 0x8

    00000048 [0x33f96650] mvnccs r6,#0x5000000 ; ? rn = 0x9

    0000004c [0x33f9ab80] mvnccs r10,#0x20000 ; ? rn = 0x9

    00000050 [0xe10f0000] mrs r0,cpsr

    00000054 [0xe3c0001f] bic r0,r0,#0x1f

    00000058 [0xe38000d3] orr r0,r0,#0xd3

    0000005c [0xe129f000] msr cpsr_cf,r0

    这么看来，

    _undefined_instruction: .word undefined_instruction

    这句对应的反汇编是：

    mvnccs r0,#0x10 ;

    这么一来我又更糊涂了。

    到ChinaUnix求助。幸好碰到一位热心的网友wheelz，详细地给我解答了。

    帖子链接如下：

    http://www.linuxforum.net/forum/showflat.p...K&Number=563178

    现在总结wheelz的回答，说说这个.word的作用。

    word expression就是在当前位置放一个word型的值，这个值就是expression

    举例来说，

    _rWTCON:

    .word 0x15300000

    就是在当前地址，即_rWTCON处放一个值0x15300000

    翻译成intel的汇编语句就是：

    _rWTCON dw 0x15300000

    就是在当前位置放个expression的值。 原来如此啊。

    PS:

    贴一个##的作用。

    #define _syscall0(type,name) \

    type name(void) \

    { \

    long __res; \

    __asm__ volatile ("int $0x80" \

    : "=a" (__res) \

    : "0" (__NR_##name)); \

    if (__res >= 0) \

    return (type) __res; \

    errno = -__res; \

    return -1; \

    }

    __NR_##name是系统调用号，##指的是两次宏展开．即用实际的系统调用名字代替"name",然后再把__NR_...展开．如name == ioctl，则为__NR_ioctl。