x86inc.asm

从服务器上下的x264编码器C源码……希望对大家有帮助……这个是09年4月的

    %endif
    ret
%endmacro

%macro REP_RET 0
    %if regs_used > 4 || xmm_regs_used > 6
        RET
    %else
        rep ret
    %endif
%endmacro

%elifdef ARCH_X86_64 ; *nix x64 ;=============================================

DECLARE_REG 0, rdi, edi, di,  dil, edi
DECLARE_REG 1, rsi, esi, si,  sil, esi
DECLARE_REG 2, rdx, edx, dx,  dl,  edx
DECLARE_REG 3, rcx, ecx, cx,  cl,  ecx
DECLARE_REG 4, r8,  r8d, r8w, r8b, r8d
DECLARE_REG 5, r9,  r9d, r9w, r9b, r9d
DECLARE_REG 6, rax, eax, ax,  al,  [rsp + stack_offset + 8]
%define r7m [rsp + stack_offset + 16]
%define r8m [rsp + stack_offset + 24]

%macro LOAD_IF_USED 2 ; reg_id, number_of_args
    %if %1 < %2
        mov r%1, [rsp - 40 + %1*8]
    %endif
%endmacro

%macro PROLOGUE 2-4+ ; #args, #regs, #xmm_regs, arg_names...
    ASSERT %2 >= %1
    ASSERT %2 <= 7
    LOAD_IF_USED 6, %1
    DEFINE_ARGS %4
%endmacro

%macro RET 0
    ret
%endmacro

%macro REP_RET 0
    rep ret
%endmacro

%else ; X86_32 ;==============================================================

DECLARE_REG 0, eax, eax, ax, al,   [esp + stack_offset + 4]
DECLARE_REG 1, ecx, ecx, cx, cl,   [esp + stack_offset + 8]
DECLARE_REG 2, edx, edx, dx, dl,   [esp + stack_offset + 12]
DECLARE_REG 3, ebx, ebx, bx, bl,   [esp + stack_offset + 16]
DECLARE_REG 4, esi, esi, si, null, [esp + stack_offset + 20]
DECLARE_REG 5, edi, edi, di, null, [esp + stack_offset + 24]
DECLARE_REG 6, ebp, ebp, bp, null, [esp + stack_offset + 28]
%define r7m [esp + stack_offset + 32]
%define r8m [esp + stack_offset + 36]
%define rsp esp

%macro PUSH_IF_USED 1 ; reg_id
    %if %1 < regs_used
        push r%1
        %assign stack_offset stack_offset+4
    %endif
%endmacro

%macro POP_IF_USED 1 ; reg_id
    %if %1 < regs_used
        pop r%1
    %endif
%endmacro

%macro LOAD_IF_USED 2 ; reg_id, number_of_args
    %if %1 < %2
        mov r%1, [esp + stack_offset + 4 + %1*4]
    %endif
%endmacro

%macro PROLOGUE 2-4+ ; #args, #regs, arg_names...
    ASSERT %2 >= %1
    %assign regs_used %2
    ASSERT regs_used <= 7
    PUSH_IF_USED 3
    PUSH_IF_USED 4
    PUSH_IF_USED 5
    PUSH_IF_USED 6
    LOAD_IF_USED 0, %1
    LOAD_IF_USED 1, %1
    LOAD_IF_USED 2, %1
    LOAD_IF_USED 3, %1
    LOAD_IF_USED 4, %1
    LOAD_IF_USED 5, %1
    LOAD_IF_USED 6, %1
    DEFINE_ARGS %4
%endmacro

%macro RET 0
    POP_IF_USED 6
    POP_IF_USED 5
    POP_IF_USED 4
    POP_IF_USED 3
    ret
%endmacro

%macro REP_RET 0
    %if regs_used > 3
        RET
    %else
        rep ret
    %endif
%endmacro

%endif ;======================================================================



;=============================================================================
; arch-independent part
;=============================================================================

%assign function_align 16

; Symbol prefix for C linkage
%macro cglobal 1-2+
    %ifdef PREFIX
        %xdefine %1.skip_prologue _%1.skip_prologue
        %xdefine %1 _%1
    %endif
    %ifidn __OUTPUT_FORMAT__,elf
        global %1:function hidden
    %else
        global %1
    %endif
    align function_align
    %1:
    RESET_MM_PERMUTATION ; not really needed, but makes disassembly somewhat nicer
    %assign stack_offset 0
    %if %0 > 1
        PROLOGUE %2
    %endif
%endmacro

%macro cextern 1
    %ifdef PREFIX
        %xdefine %1 _%1
    %endif
    extern %1
%endmacro

; This is needed for ELF, otherwise the GNU linker assumes the stack is
; executable by default.
%ifidn __OUTPUT_FORMAT__,elf
SECTION .note.GNU-stack noalloc noexec nowrite progbits
%endif

%assign FENC_STRIDE 16
%assign FDEC_STRIDE 32

; merge mmx and sse*

%macro CAT_XDEFINE 3
    %xdefine %1%2 %3
%endmacro

%macro CAT_UNDEF 2
    %undef %1%2
%endmacro

%macro INIT_MMX 0
    %define RESET_MM_PERMUTATION INIT_MMX
    %define mmsize 8
    %define num_mmregs 8
    %define mova movq
    %define movu movq
    %define movh movd
    %define movnt movntq
    %assign %%i 0
    %rep 8
    CAT_XDEFINE m, %%i, mm %+ %%i
    CAT_XDEFINE nmm, %%i, %%i
    %assign %%i %%i+1
    %endrep
    %rep 8
    CAT_UNDEF m, %%i
    CAT_UNDEF nmm, %%i
    %assign %%i %%i+1
    %endrep
%endmacro

%macro INIT_XMM 0
    %define RESET_MM_PERMUTATION INIT_XMM
    %define mmsize 16
    %define num_mmregs 8
    %ifdef ARCH_X86_64
    %define num_mmregs 16
    %endif
    %define mova movdqa
    %define movu movdqu
    %define movh movq
    %define movnt movntdq
    %assign %%i 0
    %rep num_mmregs
    CAT_XDEFINE m, %%i, xmm %+ %%i
    CAT_XDEFINE nxmm, %%i, %%i
    %assign %%i %%i+1
    %endrep
%endmacro

INIT_MMX

; I often want to use macros that permute their arguments. e.g. there's no
; efficient way to implement butterfly or transpose or dct without swapping some
; arguments.
;
; I would like to not have to manually keep track of the permutations:
; If I insert a permutation in the middle of a function, it should automatically
; change everything that follows. For more complex macros I may also have multiple
; implementations, e.g. the SSE2 and SSSE3 versions may have different permutations.
;
; Hence these macros. Insert a PERMUTE or some SWAPs at the end of a macro that
; permutes its arguments. It's equivalent to exchanging the contents of the
; registers, except that this way you exchange the register names instead, so it
; doesn't cost any cycles.

%macro PERMUTE 2-* ; takes a list of pairs to swap
%rep %0/2
    %xdefine tmp%2 m%2
    %xdefine ntmp%2 nm%2
    %rotate 2
%endrep
%rep %0/2
    %xdefine m%1 tmp%2
    %xdefine nm%1 ntmp%2
    %undef tmp%2
    %undef ntmp%2
    %rotate 2
%endrep
%endmacro

%macro SWAP 2-* ; swaps a single chain (sometimes more concise than pairs)
%rep %0-1
%ifdef m%1
    %xdefine tmp m%1
    %xdefine m%1 m%2
    %xdefine m%2 tmp
    CAT_XDEFINE n, m%1, %1
    CAT_XDEFINE n, m%2, %2
%else
    ; If we were called as "SWAP m0,m1" rather than "SWAP 0,1" infer the original numbers here.
    ; Be careful using this mode in nested macros though, as in some cases there may be
    ; other copies of m# that have already been dereferenced and don't get updated correctly.
    %xdefine %%n1 n %+ %1
    %xdefine %%n2 n %+ %2
    %xdefine tmp m %+ %%n1
    CAT_XDEFINE m, %%n1, m %+ %%n2
    CAT_XDEFINE m, %%n2, tmp
    CAT_XDEFINE n, m %+ %%n1, %%n1
    CAT_XDEFINE n, m %+ %%n2, %%n2
%endif
    %undef tmp
    %rotate 1
%endrep
%endmacro

%macro SAVE_MM_PERMUTATION 1
    %assign %%i 0
    %rep num_mmregs
    CAT_XDEFINE %1_m, %%i, m %+ %%i
    %assign %%i %%i+1
    %endrep
%endmacro

%macro LOAD_MM_PERMUTATION 1
    %assign %%i 0
    %rep num_mmregs
    CAT_XDEFINE m, %%i, %1_m %+ %%i
    CAT_XDEFINE n, m %+ %%i, %%i
    %assign %%i %%i+1
    %endrep
%endmacro

%macro call 1
    call %1
    %ifdef %1_m0
        LOAD_MM_PERMUTATION %1
    %endif
%endmacro

;Substitutions that reduce instruction size but are functionally equivalent
%macro add 2
    %ifnum %2
        %if %2==128
            sub %1, -128
        %else
            add %1, %2
        %endif
    %else
        add %1, %2
    %endif
%endmacro

%macro sub 2
    %ifnum %2
        %if %2==128
            add %1, -128
        %else
            sub %1, %2
        %endif
    %else
        sub %1, %2
    %endif
%endmacro