--- trunk/xvidcore/src/image/x86_asm/postprocessing_sse2.asm	2008/11/26 01:04:34	1795
+++ trunk/xvidcore/src/image/x86_asm/postprocessing_sse2.asm	2008/12/04 14:41:50	1844
@@ -37,7 +37,7 @@
 ; Code
 ;=============================================================================
 
-SECTION .rotext align=SECTION_ALIGN
+TEXT
 
 cglobal image_brightness_sse2
 
@@ -66,8 +66,11 @@
 
 ALIGN SECTION_ALIGN
 image_brightness_sse2:
-
-  mov eax, prm5d  ; brightness offset value	
+%ifdef ARCH_IS_X86_64
+  XVID_MOVSX _EAX, prm5d
+%else
+  mov eax, prm5   ; brightness offset value	
+%endif
   mov TMP1, prm1  ; Dst
   mov TMP0, prm2  ; stride
 
@@ -75,14 +78,14 @@
   push _EDI    ; 8 bytes offset for push
   sub _ESP, 32 ; 32 bytes for local data (16bytes will be used, 16bytes more to align correctly mod 16)
 
-  movdqa xmm6, [xmm_0x80]
+  movdqa xmm2, [xmm_0x80]
 
   ; Create a offset...offset vector
-  mov TMP1, _ESP          ; TMP1 will be esp aligned mod 16
-  add TMP1, 15            ; TMP1 = esp + 15
-  and TMP1, ~15           ; TMP1 = (esp + 15)&(~15)
-  CREATE_OFFSET_VECTOR TMP1, al
-  movdqa xmm7, [TMP1]
+  mov _ESI, _ESP          ; TMP1 will be esp aligned mod 16
+  add _ESI, 15            ; TMP1 = esp + 15
+  and _ESI, ~15           ; TMP1 = (esp + 15)&(~15)
+  CREATE_OFFSET_VECTOR _ESI, al
+  movdqa xmm3, [_ESI]
 
 %ifdef ARCH_IS_X86_64
   mov _ESI, prm3
@@ -99,12 +102,12 @@
   movdqa xmm0, [TMP1 + _EAX]
   movdqa xmm1, [TMP1 + _EAX + 16] ; xmm0 = [dst]
 
-  paddb xmm0, xmm6              ; unsigned -> signed domain
-  paddb xmm1, xmm6
-  paddsb xmm0, xmm7
-  paddsb xmm1, xmm7             ; xmm0 += offset
-  psubb xmm0, xmm6
-  psubb xmm1, xmm6              ; signed -> unsigned domain
+  paddb xmm0, xmm2              ; unsigned -> signed domain
+  paddb xmm1, xmm2
+  paddsb xmm0, xmm3
+  paddsb xmm1, xmm3             ; xmm0 += offset
+  psubb xmm0, xmm2
+  psubb xmm1, xmm2              ; signed -> unsigned domain
 
   movdqa [TMP1 + _EAX], xmm0
   movdqa [TMP1 + _EAX + 16], xmm1 ; [dst] = xmm0
@@ -114,7 +117,7 @@
   jl .xloop
 
   add TMP1, TMP0                  ; dst += stride
-  sub _EDI, 1
+  dec _EDI
   jg .yloop
 
   add _ESP, 32