Fix horizontal/horizontal_up 8x8l intra prediction x86/simd functions.

The original functions did not work correctly for edge pixels, e.g.
when CODEC_FLAG_EMU_EDGE is set, leading to corrupt output in e.g. VLC.
Based on a patch by Daniel Kang <daniel d kang gmail com>.

Signed-off-by: Ronald S. Bultje <rsbultje gmail com>
(cherry picked from commit b9c7f66e)

libavcodec/x86/h264_intrapred.asm

@@ -1249,7 +1249,10 @@ cglobal pred8x8l_horizontal_%1, 4,4
     sub          r0, r3
     lea          r2, [r0+r3*2]
     movq        mm0, [r0+r3*1-8]
-    punpckhbw   mm0, [r0+r3*0-8]
+    test         r1, r1
+    lea          r1, [r0+r3]
+    cmovnz       r1, r0
+    punpckhbw   mm0, [r1+r3*0-8]
     movq        mm1, [r2+r3*1-8]
     punpckhbw   mm1, [r0+r3*2-8]
     mov          r2, r0
@@ -1264,21 +1267,12 @@ cglobal pred8x8l_horizontal_%1, 4,4
     punpckhdq   mm3, mm1
     lea          r0, [r0+r3*2]
     movq        mm0, [r0+r3*0-8]
-    movq        mm1, [r2]
+    movq        mm1, [r1+r3*0-8]
     mov          r0, r2
     movq        mm4, mm3
     movq        mm2, mm3
     PALIGNR     mm4, mm0, 7, mm0
     PALIGNR     mm1, mm2, 1, mm2
-    test        r1, r1 ; top_left
-    jnz .do_left
-.fix_lt_1:
-    movq        mm5, mm3
-    pxor        mm5, mm4
-    psrlq       mm5, 56
-    psllq       mm5, 48
-    pxor        mm1, mm5
-.do_left:
     movq        mm0, mm4
     PRED4x4_LOWPASS mm2, mm1, mm4, mm3, mm5
     movq        mm4, mm0
@@ -2153,7 +2147,10 @@ cglobal pred8x8l_horizontal_up_%1, 4,4
     sub          r0, r3
     lea          r2, [r0+r3*2]
     movq        mm0, [r0+r3*1-8]
-    punpckhbw   mm0, [r0+r3*0-8]
+    test         r1, r1
+    lea          r1, [r0+r3]
+    cmovnz       r1, r0
+    punpckhbw   mm0, [r1+r3*0-8]
     movq        mm1, [r2+r3*1-8]
     punpckhbw   mm1, [r0+r3*2-8]
     mov          r2, r0
@@ -2168,21 +2165,12 @@ cglobal pred8x8l_horizontal_up_%1, 4,4
     punpckhdq   mm3, mm1
     lea          r0, [r0+r3*2]
     movq        mm0, [r0+r3*0-8]
-    movq        mm1, [r2]
+    movq        mm1, [r1+r3*0-8]
     mov          r0, r2
     movq        mm4, mm3
     movq        mm2, mm3
     PALIGNR     mm4, mm0, 7, mm0
     PALIGNR     mm1, mm2, 1, mm2
-    test        r1, r1
-    jnz .do_left
-.fix_lt_1:
-    movq        mm5, mm3
-    pxor        mm5, mm4
-    psrlq       mm5, 56
-    psllq       mm5, 48
-    pxor        mm1, mm5
-.do_left:
     movq       mm0, mm4
     PRED4x4_LOWPASS mm2, mm1, mm4, mm3, mm5
     movq       mm4, mm0