avfilter/x86/vf_hflip : merge hflip byte and hflip short to one macro

libavfilter/x86/vf_hflip.asm

@@ -29,11 +29,16 @@ pb_flip_short: db 14,15,12,13,10,11,8,9,6,7,4,5,2,3,0,1
 
 SECTION .text
 
-INIT_XMM ssse3
+;%1 byte or short, %2 b or w, %3 size in byte (1 for byte, 2 for short)
-cglobal hflip_byte, 3, 5, 3, src, dst, w, r, x
+%macro HFLIP 3
-    mova    m0, [pb_flip_byte]
+cglobal hflip_%1, 3, 5, 3, src, dst, w, r, x
+    mova    m0, [pb_flip_%1]
     xor     xq, xq
+%if %3 == 1
     movsxdifnidn wq, wd
+%else ; short
+    add     wd, wd
+%endif
     mov     rq, wq
     and     rq, 2 * mmsize - 1
     cmp     wq, 2 * mmsize
@@ -42,8 +47,8 @@ cglobal hflip_byte, 3, 5, 3, src, dst, w, r, x
 
     .loop0:
         neg     xq
-        movu    m1, [srcq + xq -     mmsize + 1]
+        movu    m1, [srcq + xq -     mmsize + %3]
-        movu    m2, [srcq + xq - 2 * mmsize + 1]
+        movu    m2, [srcq + xq - 2 * mmsize + %3]
         pshufb  m1, m0
         pshufb  m2, m0
         neg     xq
@@ -59,49 +64,17 @@ cglobal hflip_byte, 3, 5, 3, src, dst, w, r, x
 
     .loop1:
         neg    xq
-        mov    rb, [srcq + xq]
+        mov    r%2, [srcq + xq]
         neg    xq
-        mov    [dstq + xq], rb
+        mov    [dstq + xq], r%2
-        add    xq, 1
+        add    xq, %3
         cmp    xq, wq
         jl .loop1
     .end:
 RET
+%endmacro
 
-cglobal hflip_short, 3, 5, 3, src, dst, w, r, x
+INIT_XMM ssse3
-    mova    m0, [pb_flip_short]
+HFLIP byte, b, 1
-    xor     xq, xq
+HFLIP short, w, 2
-    add     wd, wd
-    mov     rq, wq
-    and     rq, 2 * mmsize - 1
-    cmp     wq, 2 * mmsize
-    jl .loop1
-    sub     wq, rq
-
-    .loop0:
-        neg     xq
-        movu    m1, [srcq + xq -     mmsize + 2]
-        movu    m2, [srcq + xq - 2 * mmsize + 2]
-        pshufb  m1, m0
-        pshufb  m2, m0
-        neg     xq
-        movu    [dstq + xq         ], m1
-        movu    [dstq + xq + mmsize], m2
-        add     xq, mmsize * 2
-        cmp     xq, wq
-        jl .loop0
-
-        cmp    rq, 0
-        je .end
-        add    wq, rq
 
-    .loop1:
-        neg    xq
-        mov    rw, [srcq + xq]
-        neg    xq
-        mov    [dstq + xq], rw
-        add    xq, 2
-        cmp    xq, wq
-        jl .loop1
-    .end:
-RET