ffmpeg source update for sep 09

source/patched-ffmpeg-mt/libavcodec/x86/vp8dsp.asm

 ;******************************************************************************
 ;* VP8 MMXEXT optimizations
 ;* Copyright (c) 2010 Ronald S. Bultje <rsbultje@gmail.com>
 ;* Copyright (c) 2010 Jason Garrett-Glaser <darkshikari@gmail.com>
 ;*
 ;* This file is part of FFmpeg.
 ;*
 ;* FFmpeg is free software; you can redistribute it and/or
 ;* modify it under the terms of the GNU Lesser General Public
 ;* License as published by the Free Software Foundation; either
@@ skipping 193 matching lines @@
     paddsw    m0, m1
     paddsw    m0, m2
     paddsw    m0, [pw_64]
     psraw     m0, 7
     packuswb  m0, m0
     movh    [r0], m0        ; store
 
     ; go to next line
     add       r0, r1
     add       r2, r3
-    dec       r4            ; next row
+    dec      r4d            ; next row
     jg .nextrow
     REP_RET
 
 cglobal put_vp8_epel%1_h4_ssse3, 6, 6, %3
     shl      r5d, 4
     mova      m2, [pw_64]
     mova      m3, [filter_h2_shuf]
     mova      m4, [filter_h4_shuf]
 %ifdef PIC
     lea      r11, [fourtap_filter_hb_m]
@@ skipping 10 matching lines @@
     pmaddubsw m1, m6
     paddsw    m0, m2
     paddsw    m0, m1
     psraw     m0, 7
     packuswb  m0, m0
     movh    [r0], m0        ; store
 
     ; go to next line
     add       r0, r1
     add       r2, r3
-    dec       r4            ; next row
+    dec      r4d            ; next row
     jg .nextrow
     REP_RET
 
 cglobal put_vp8_epel%1_v4_ssse3, 7, 7, %2
     shl      r6d, 4
 %ifdef PIC
     lea      r11, [fourtap_filter_hb_m]
 %endif
     mova      m5, [fourtap_filter_hb+r6-16]
     mova      m6, [fourtap_filter_hb+r6]
@@ skipping 18 matching lines @@
     paddsw    m4, m2
     mova      m2, m3
     paddsw    m4, m7
     psraw     m4, 7
     packuswb  m4, m4
     movh    [r0], m4
 
     ; go to next line
     add        r0, r1
     add        r2, r3
-    dec        r4                          ; next row
+    dec       r4d                          ; next row
     jg .nextrow
     REP_RET
 
 cglobal put_vp8_epel%1_v6_ssse3, 7, 7, %2
     lea      r6d, [r6*3]
 %ifdef PIC
     lea      r11, [sixtap_filter_hb_m]
 %endif
     lea       r6, [sixtap_filter_hb+r6*8]
 
@@ skipping 26 matching lines @@
     mova      m2, m3
     psraw     m6, 7
     mova      m3, m4
     packuswb  m6, m6
     mova      m4, m5
     movh    [r0], m6
 
     ; go to next line
     add        r0, r1
     add        r2, r3
-    dec        r4                          ; next row
+    dec       r4d                          ; next row
     jg .nextrow
     REP_RET
 %endmacro
 
 INIT_MMX
 FILTER_SSSE3 4, 0, 0
 INIT_XMM
 FILTER_SSSE3 8, 8, 7
 
 ; 4x4 block, H-only 4-tap filter
@@ skipping 32 matching lines @@
     ; merge two sets of 2 pixels into one set of 4, round/clip/store
     packssdw  mm3, mm0                     ; merge dword->word (4px)
     paddsw    mm3, mm7                     ; rounding
     psraw     mm3, 7
     packuswb  mm3, mm6                     ; clip and word->bytes
     movd     [r0], mm3                     ; store
 
     ; go to next line
     add        r0, r1
     add        r2, r3
-    dec        r4                          ; next row
+    dec       r4d                          ; next row
     jg .nextrow
     REP_RET
 
 ; 4x4 block, H-only 6-tap filter
 cglobal put_vp8_epel4_h6_mmxext, 6, 6
     lea       r5d, [r5*3]
 %ifdef PIC
     lea       r11, [sixtap_filter_hw_m]
 %endif
     movq      mm4, [sixtap_filter_hw+r5*8-48] ; set up 4tap filter in words
@@ skipping 36 matching lines @@
     ; merge two sets of 2 pixels into one set of 4, round/clip/store
     packssdw  mm1, mm0                     ; merge dword->word (4px)
     paddsw    mm1, mm7                     ; rounding
     psraw     mm1, 7
     packuswb  mm1, mm3                     ; clip and word->bytes
     movd     [r0], mm1                     ; store
 
     ; go to next line
     add        r0, r1
     add        r2, r3
-    dec        r4                          ; next row
+    dec       r4d                          ; next row
     jg .nextrow
     REP_RET
 
 INIT_XMM
 cglobal put_vp8_epel8_h4_sse2, 6, 6, 10
     shl      r5d, 5
 %ifdef PIC
     lea      r11, [fourtap_filter_v_m]
 %endif
     lea       r5, [fourtap_filter_v+r5-32]
@@ skipping 27 matching lines @@
     paddsw    m2, m3
     paddsw    m0, m2
     paddsw    m0, m4
     psraw     m0, 7
     packuswb  m0, m7
     movh    [r0], m0        ; store
 
     ; go to next line
     add       r0, r1
     add       r2, r3
-    dec       r4            ; next row
+    dec      r4d            ; next row
     jg .nextrow
     REP_RET
 
 cglobal put_vp8_epel8_h6_sse2, 6, 6, 14
     lea      r5d, [r5*3]
     shl      r5d, 4
 %ifdef PIC
     lea      r11, [sixtap_filter_v_m]
 %endif
     lea       r5, [sixtap_filter_v+r5-96]
@@ skipping 41 matching lines @@
     paddsw    m0, m3
     paddsw    m0, m1
     paddsw    m0, m6
     psraw     m0, 7
     packuswb  m0, m7
     movh    [r0], m0        ; store
 
     ; go to next line
     add       r0, r1
     add       r2, r3
-    dec       r4            ; next row
+    dec      r4d            ; next row
     jg .nextrow
     REP_RET
 
 %macro FILTER_V 3
 ; 4x4 block, V-only 4-tap filter
 cglobal put_vp8_epel%2_v4_%1, 7, 7, %3
     shl      r6d, 5
 %ifdef PIC
     lea      r11, [fourtap_filter_v_m]
 %endif
@@ skipping 32 matching lines @@
 
     ; round/clip/store
     paddsw    m4, m6
     psraw     m4, 7
     packuswb  m4, m7
     movh    [r0], m4
 
     ; go to next line
     add       r0, r1
     add       r2, r3
-    dec       r4                           ; next row
+    dec      r4d                           ; next row
     jg .nextrow
     REP_RET
 
 
 ; 4x4 block, V-only 6-tap filter
 cglobal put_vp8_epel%2_v6_%1, 7, 7, %3
     shl      r6d, 4
     lea       r6, [r6*3]
 %ifdef PIC
     lea      r11, [sixtap_filter_v_m]
@@ skipping 44 matching lines @@
 
     ; round/clip/store
     paddsw    m6, [pw_64]
     psraw     m6, 7
     packuswb  m6, m7
     movh    [r0], m6
 
     ; go to next line
     add       r0, r1
     add       r2, r3
-    dec       r4                           ; next row
+    dec      r4d                           ; next row
     jg .nextrow
     REP_RET
 %endmacro
 
 INIT_MMX
 FILTER_V mmxext, 4, 0
 INIT_XMM
 FILTER_V sse2,   8, 8
 
 %macro FILTER_BILINEAR 3
@@ skipping 31 matching lines @@
     movh [r0+r1*0], m0
     movh [r0+r1*1], m2
 %else
     packuswb  m0, m2
     movh   [r0+r1*0], m0
     movhps [r0+r1*1], m0
 %endif
 
     lea       r0, [r0+r1*2]
     lea       r2, [r2+r3*2]
-    sub       r4, 2
+    sub      r4d, 2
     jg .nextrow
     REP_RET
 
 cglobal put_vp8_bilinear%2_h_%1, 7,7,%3
     mov      r6d, 8*16
     shl      r5d, 4
     sub      r6d, r5d
 %ifdef PIC
     lea      r11, [bilinear_filter_vw_m]
 %endif
@@ skipping 25 matching lines @@
     movh [r0+r1*0], m0
     movh [r0+r1*1], m2
 %else
     packuswb  m0, m2
     movh   [r0+r1*0], m0
     movhps [r0+r1*1], m0
 %endif
 
     lea       r0, [r0+r1*2]
     lea       r2, [r2+r3*2]
-    sub       r4, 2
+    sub      r4d, 2
     jg .nextrow
     REP_RET
 %endmacro
 
 INIT_MMX
 FILTER_BILINEAR mmxext, 4, 0
 INIT_XMM
 FILTER_BILINEAR   sse2, 8, 7
 
 %macro FILTER_BILINEAR_SSSE3 1
@@ skipping 22 matching lines @@
     movh [r0+r1*0], m0
     movh [r0+r1*1], m1
 %else
     packuswb  m0, m1
     movh   [r0+r1*0], m0
     movhps [r0+r1*1], m0
 %endif
 
     lea       r0, [r0+r1*2]
     lea       r2, [r2+r3*2]
-    sub       r4, 2
+    sub      r4d, 2
     jg .nextrow
     REP_RET
 
 cglobal put_vp8_bilinear%1_h_ssse3, 7,7
     shl      r5d, 4
 %ifdef PIC
     lea      r11, [bilinear_filter_vb_m]
 %endif
     pxor      m4, m4
     mova      m2, [filter_h2_shuf]
@@ skipping 15 matching lines @@
     movh [r0+r1*0], m0
     movh [r0+r1*1], m1
 %else
     packuswb  m0, m1
     movh   [r0+r1*0], m0
     movhps [r0+r1*1], m0
 %endif
 
     lea       r0, [r0+r1*2]
     lea       r2, [r2+r3*2]
-    sub       r4, 2
+    sub      r4d, 2
     jg .nextrow
     REP_RET
 %endmacro
 
 INIT_MMX
 FILTER_BILINEAR_SSSE3 4
 INIT_XMM
 FILTER_BILINEAR_SSSE3 8
 
 cglobal put_vp8_pixels8_mmx, 5,5
@@ skipping 478 matching lines @@
 %if %10 == 16
     movd    [%6+%9*4], m%3
 %endif
     movd      [%7+%9], m%4
 
     ; write dwords 2
     psrldq        m%1, 4
     psrldq        m%2, 4
 %if %10 == 8
     movd    [%5+%8*2], m%1
-    movd           %5, m%3
+    movd          %5d, m%3
 %endif
     psrldq        m%3, 4
     psrldq        m%4, 4
 %if %10 == 16
     movd    [%5+%8*2], m%1
 %endif
     movd      [%6+%9], m%2
     movd    [%7+%8*2], m%3
     movd    [%7+%9*2], m%4
     add            %7, %9
@@ skipping 16 matching lines @@
 
 ; write 4 or 8 words in the mmx/xmm registers as 8 lines
 ; 1 and 2 are the registers to write, this can be the same (for SSE2)
 ; for pre-SSE4:
 ; 3 is a general-purpose register that we will clobber
 ; for SSE4:
 ; 3 is a pointer to the destination's 5th line
 ; 4 is a pointer to the destination's 4th line
 ; 5/6 is -stride and +stride
 %macro WRITE_2x4W 6
-    movd             %3, %1
+    movd            %3d, %1
     punpckhdq        %1, %1
     mov       [%4+%5*4], %3w
     shr              %3, 16
     add              %4, %6
     mov       [%4+%5*4], %3w
 
-    movd             %3, %1
+    movd            %3d, %1
     add              %4, %5
     mov       [%4+%5*2], %3w
     shr              %3, 16
     mov       [%4+%5  ], %3w
 
-    movd             %3, %2
+    movd            %3d, %2
     punpckhdq        %2, %2
     mov       [%4     ], %3w
     shr              %3, 16
     mov       [%4+%6  ], %3w
 
-    movd             %3, %2
+    movd            %3d, %2
     add              %4, %6
     mov       [%4+%6  ], %3w
     shr              %3, 16
     mov       [%4+%6*2], %3w
     add              %4, %5
 %endmacro
 
 %macro WRITE_8W_SSE2 5
-    movd             %2, %1
+    movd            %2d, %1
     psrldq           %1, 4
     mov       [%3+%4*4], %2w
     shr              %2, 16
     add              %3, %5
     mov       [%3+%4*4], %2w
 
-    movd             %2, %1
+    movd            %2d, %1
     psrldq           %1, 4
     add              %3, %4
     mov       [%3+%4*2], %2w
     shr              %2, 16
     mov       [%3+%4  ], %2w
 
-    movd             %2, %1
+    movd            %2d, %1
     psrldq           %1, 4
     mov       [%3     ], %2w
     shr              %2, 16
     mov       [%3+%5  ], %2w
 
-    movd             %2, %1
+    movd            %2d, %1
     add              %3, %5
     mov       [%3+%5  ], %2w
     shr              %2, 16
     mov       [%3+%5*2], %2w
 %endmacro
 
 %macro WRITE_8W_SSE4 5
     pextrw    [%3+%4*4], %1, 0
     pextrw    [%2+%4*4], %1, 1
     pextrw    [%3+%4*2], %1, 2
     pextrw    [%3+%4  ], %1, 3
     pextrw    [%3     ], %1, 4
     pextrw    [%2     ], %1, 5
     pextrw    [%2+%5  ], %1, 6
     pextrw    [%2+%5*2], %1, 7
 %endmacro
 
 %macro SPLATB_REG_MMX 2-3
-    movd           %1, %2
+    movd           %1, %2d
     punpcklbw      %1, %1
     punpcklwd      %1, %1
     punpckldq      %1, %1
 %endmacro
 
 %macro SPLATB_REG_MMXEXT 2-3
-    movd           %1, %2
+    movd           %1, %2d
     punpcklbw      %1, %1
     pshufw         %1, %1, 0x0
 %endmacro
 
 %macro SPLATB_REG_SSE2 2-3
-    movd           %1, %2
+    movd           %1, %2d
     punpcklbw      %1, %1
     pshuflw        %1, %1, 0x0
     punpcklqdq     %1, %1
 %endmacro
 
 %macro SPLATB_REG_SSSE3 3
-    movd           %1, %2
+    movd           %1, %2d
     pshufb         %1, %3
 %endmacro
 
-%macro SIMPLE_LOOPFILTER 3
-cglobal vp8_%2_loop_filter_simple_%1, 3, %3
+%macro SIMPLE_LOOPFILTER 4
+cglobal vp8_%2_loop_filter_simple_%1, 3, %3, %4
 %if mmsize == 8 ; mmx/mmxext
     mov            r3, 2
 %endif
 %ifnidn %1, sse2
 %if mmsize == 16
     pxor           m0, m0
 %endif
 %endif
     SPLATB_REG     m7, r2, m0       ; splat "flim" into register
 
@@ skipping 120 matching lines @@
     dec            r3
     jg .next8px
     REP_RET
 %else ; sse2
     RET
 %endif
 %endmacro
 
 INIT_MMX
 %define SPLATB_REG SPLATB_REG_MMX
-SIMPLE_LOOPFILTER mmx,    v, 4
-SIMPLE_LOOPFILTER mmx,    h, 5
+SIMPLE_LOOPFILTER mmx,    v, 4, 0
+SIMPLE_LOOPFILTER mmx,    h, 5, 0
 %define SPLATB_REG SPLATB_REG_MMXEXT
-SIMPLE_LOOPFILTER mmxext, v, 4
-SIMPLE_LOOPFILTER mmxext, h, 5
+SIMPLE_LOOPFILTER mmxext, v, 4, 0
+SIMPLE_LOOPFILTER mmxext, h, 5, 0
 INIT_XMM
 %define SPLATB_REG SPLATB_REG_SSE2
 %define WRITE_8W   WRITE_8W_SSE2
-SIMPLE_LOOPFILTER sse2,   v, 3
-SIMPLE_LOOPFILTER sse2,   h, 5
+SIMPLE_LOOPFILTER sse2,   v, 3, 8
+SIMPLE_LOOPFILTER sse2,   h, 5, 8
 %define SPLATB_REG SPLATB_REG_SSSE3
-SIMPLE_LOOPFILTER ssse3,  v, 3
-SIMPLE_LOOPFILTER ssse3,  h, 5
+SIMPLE_LOOPFILTER ssse3,  v, 3, 8
+SIMPLE_LOOPFILTER ssse3,  h, 5, 8
 %define WRITE_8W   WRITE_8W_SSE4
-SIMPLE_LOOPFILTER sse4,   h, 5
+SIMPLE_LOOPFILTER sse4,   h, 5, 8
 
 ;-----------------------------------------------------------------------------
 ; void vp8_h/v_loop_filter<size>_inner_<opt>(uint8_t *dst, [uint8_t *v,] int stride,
 ;                                            int flimE, int flimI, int hev_thr);
 ;-----------------------------------------------------------------------------
 
 %macro INNER_LOOPFILTER 5
 %if %4 == 8 ; chroma
 cglobal vp8_%2_loop_filter8uv_inner_%1, 6, %3, %5
 %define dst8_reg    r1
@@ skipping 1209 matching lines @@
 MBEDGE_LOOPFILTER ssse3,  v, 6,  8, 15
 MBEDGE_LOOPFILTER ssse3,  h, 6,  8, 15
 
 %define WRITE_8W   WRITE_8W_SSE4
 %ifdef m8
 MBEDGE_LOOPFILTER sse4,   h, 5, 16, 15
 %else
 MBEDGE_LOOPFILTER sse4,   h, 6, 16, 15
 %endif
 MBEDGE_LOOPFILTER sse4,   h, 6,  8, 15