Merge remote-tracking branch 'qatar/master'

[ffmpeg] / libavcodec / x86 / dsputil_yasm.asm

diff --git a/libavcodec/x86/dsputil_yasm.asm b/libavcodec/x86/dsputil_yasm.asm
index dcd6fa422391caba28a809f3eb09ee8465847472..6c52ac1ffbc5623aba0b28d6a5c4c3574026126f 100644 (file)
--- a/libavcodec/x86/dsputil_yasm.asm
+++ b/libavcodec/x86/dsputil_yasm.asm
@@ -138,7 +138,7 @@ align 16
 %endif
     %define t0  [v1q + orderq]
     %define t1  [v1q + orderq + mmsize]
-%ifdef ARCH_X86_64
+%if ARCH_X86_64
     mova    m8, t0
     mova    m9, t1
     %define t0  m8
@@ -474,7 +474,7 @@ cglobal scalarproduct_float_sse, 3,3,2, v1, v2, offset
     movss   xmm1, xmm0
     shufps  xmm0, xmm0, 1
     addss   xmm0, xmm1
-%ifndef ARCH_X86_64
+%if ARCH_X86_64 == 0
     movd    r0m,  xmm0
     fld     dword r0m
 %endif
@@ -498,7 +498,7 @@ cglobal scalarproduct_float_sse, 3,3,2, v1, v2, offset
 ; function implementations. Fast are fixed-width, slow is variable-width
 
 %macro EMU_EDGE_FUNC 0
-%ifdef ARCH_X86_64
+%if ARCH_X86_64
 %define w_reg r10
 cglobal emu_edge_core, 6, 7, 1
     mov        r11, r5          ; save block_h
@@ -513,14 +513,14 @@ cglobal emu_edge_core, 2, 7, 0
     mov      w_reg, r7m
     sub      w_reg, r6m         ; w = start_x - end_x
     sub         r5, r4
-%ifdef ARCH_X86_64
+%if ARCH_X86_64
     sub         r4, r3
 %else
     sub         r4, dword r3m
 %endif
     cmp      w_reg, 22
     jg .slow_v_extend_loop
-%ifdef ARCH_X86_32
+%if ARCH_X86_32
     mov         r2, r2m         ; linesize
 %endif
     sal      w_reg, 7           ; w * 128
@@ -536,7 +536,7 @@ cglobal emu_edge_core, 2, 7, 0
     ; horizontal extend (left/right)
     mov      w_reg, r6m         ; start_x
     sub         r0, w_reg
-%ifdef ARCH_X86_64
+%if ARCH_X86_64
     mov         r3, r0          ; backup of buf+block_h*linesize
     mov         r5, r11
 %else
@@ -564,7 +564,7 @@ cglobal emu_edge_core, 2, 7, 0
 
     ; now r3(64)/r0(32)=buf,r2=linesize,r11/r5=block_h,r6/r3=val, r10/r6=end_x, r1=block_w
 .right_extend:
-%ifdef ARCH_X86_32
+%if ARCH_X86_32
     mov         r0, r0m
     mov         r5, r5m
 %endif
@@ -589,13 +589,13 @@ cglobal emu_edge_core, 2, 7, 0
 .h_extend_end:
     RET
 
-%ifdef ARCH_X86_64
+%if ARCH_X86_64
 %define vall  al
 %define valh  ah
 %define valw  ax
 %define valw2 r10w
 %define valw3 r3w
-%ifdef WIN64
+%if WIN64
 %define valw4 r4w
 %else ; unix64
 %define valw4 r3w
@@ -643,7 +643,7 @@ cglobal emu_edge_core, 2, 7, 0
 %endrep ; %2/16
 %endif
 
-%ifdef ARCH_X86_64
+%if ARCH_X86_64
 %if (%2-%%src_off) == 8
     mov           rax, [r1+%%src_off]
 %assign %%src_off %%src_off+8
@@ -692,7 +692,7 @@ cglobal emu_edge_core, 2, 7, 0
 %endrep ; %2/16
 %endif
 
-%ifdef ARCH_X86_64
+%if ARCH_X86_64
 %if (%2-%%dst_off) == 8
     mov    [r0+%%dst_off], rax
 %assign %%dst_off %%dst_off+8
@@ -740,7 +740,7 @@ cglobal emu_edge_core, 2, 7, 0
 ALIGN 128
 .emuedge_v_extend_ %+ %%n:
     ; extend pixels above body
-%ifdef ARCH_X86_64
+%if ARCH_X86_64
     test           r3 , r3                   ; if (!start_y)
     jz .emuedge_copy_body_ %+ %%n %+ _loop   ;   goto body
 %else ; ARCH_X86_32
@@ -751,7 +751,7 @@ ALIGN 128
 .emuedge_extend_top_ %+ %%n %+ _loop:        ; do {
     WRITE_NUM_BYTES top,    %%n              ;   write bytes
     add            r0 , r2                   ;   dst += linesize
-%ifdef ARCH_X86_64
+%if ARCH_X86_64
     dec            r3d
 %else ; ARCH_X86_32
     dec      dword r3m
@@ -779,7 +779,7 @@ ALIGN 128
     jnz .emuedge_extend_bottom_ %+ %%n %+ _loop ; } while (--block_h)
 
 .emuedge_v_extend_end_ %+ %%n:
-%ifdef ARCH_X86_64
+%if ARCH_X86_64
     ret
 %else ; ARCH_X86_32
     rep ret
@@ -841,7 +841,7 @@ ALIGN 64
     WRITE_V_PIXEL %%n, r0              ;   write pixels
     dec         r5
     jnz .emuedge_extend_left_ %+ %%n   ; } while (--block_h)
-%ifdef ARCH_X86_64
+%if ARCH_X86_64
     ret
 %else ; ARCH_X86_32
     rep ret
@@ -856,7 +856,7 @@ ALIGN 64
 %rep 11
 ALIGN 64
 .emuedge_extend_right_ %+ %%n:          ; do {
-%ifdef ARCH_X86_64
+%if ARCH_X86_64
     sub        r3, r2                   ;   dst -= linesize
     READ_V_PIXEL  %%n, [r3+w_reg-1]     ;   read pixels
     WRITE_V_PIXEL %%n, r3+r4-%%n        ;   write pixels
@@ -868,7 +868,7 @@ ALIGN 64
     dec     r5
 %endif ; ARCH_X86_64/32
     jnz .emuedge_extend_right_ %+ %%n   ; } while (--block_h)
-%ifdef ARCH_X86_64
+%if ARCH_X86_64
     ret
 %else ; ARCH_X86_32
     rep ret
@@ -876,7 +876,7 @@ ALIGN 64
 %assign %%n %%n+2
 %endrep
 
-%ifdef ARCH_X86_32
+%if ARCH_X86_32
 %define stack_offset 0x10
 %endif
 %endmacro ; RIGHT_EXTEND
@@ -916,7 +916,7 @@ ALIGN 64
     V_COPY_NPX %1,  mm0, movq,    8, 0xFFFFFFF8
 %else ; sse
     V_COPY_NPX %1, xmm0, movups, 16, 0xFFFFFFF0
-%ifdef ARCH_X86_64
+%if ARCH_X86_64
 %define linesize r2
     V_COPY_NPX %1, rax , mov,     8
 %else ; ARCH_X86_32
@@ -940,7 +940,7 @@ ALIGN 64
 .slow_v_extend_loop:
 ; r0=buf,r1=src,r2(64)/r2m(32)=linesize,r3(64)/r3m(32)=start_x,r4=end_y,r5=block_h
 ; r11(64)/r3(later-64)/r2(32)=cnt_reg,r6(64)/r3(32)=val_reg,r10(64)/r6(32)=w=end_x-start_x
-%ifdef ARCH_X86_64
+%if ARCH_X86_64
     push       r11              ; save old value of block_h
     test        r3, r3
 %define cnt_reg r11
@@ -956,18 +956,18 @@ ALIGN 64
 .do_body_copy:
     V_COPY_ROW body, r4
 
-%ifdef ARCH_X86_64
+%if ARCH_X86_64
     pop        r11              ; restore old value of block_h
 %define cnt_reg r3
 %endif
     test        r5, r5
-%ifdef ARCH_X86_64
+%if ARCH_X86_64
     jz .v_extend_end
 %else
     jz .skip_bottom_extend
 %endif
     V_COPY_ROW bottom, r5
-%ifdef ARCH_X86_32
+%if ARCH_X86_32
 .skip_bottom_extend:
     mov         r2, r2m
 %endif
@@ -996,7 +996,7 @@ ALIGN 64
 .left_extend_loop_end:
     dec         r5
     jnz .slow_left_extend_loop
-%ifdef ARCH_X86_32
+%if ARCH_X86_32
     mov         r2, r2m
 %endif
     jmp .right_extend
@@ -1006,7 +1006,7 @@ ALIGN 64
 .slow_right_extend_loop:
 ; r3(64)/r0(32)=buf+block_h*linesize,r2=linesize,r4=block_w,r11(64)/r5(32)=block_h,
 ; r10(64)/r6(32)=end_x,r6/r3=val,r1=cntr
-%ifdef ARCH_X86_64
+%if ARCH_X86_64
 %define buf_reg r3
 %define bh_reg r11
 %else
@@ -1047,7 +1047,7 @@ SLOW_RIGHT_EXTEND
 %endmacro
 
 emu_edge sse
-%ifdef ARCH_X86_32
+%if ARCH_X86_32
 emu_edge mmx
 %endif
 
@@ -1138,7 +1138,7 @@ VECTOR_CLIP_INT32 6, 1, 0, 0
 
 %macro BUTTERFLIES_FLOAT_INTERLEAVE 0
 cglobal butterflies_float_interleave, 4,4,3, dst, src0, src1, len
-%ifdef ARCH_X86_64
+%if ARCH_X86_64
     movsxd    lenq, lend
 %endif
     test      lenq, lenq