From 1500be13f204acb7e74dac4325ef0052576fa2a9 Mon Sep 17 00:00:00 2001
From: Alexander Strange <astrange@ithinksw.com>
Date: Sat, 26 Mar 2011 17:31:13 -0400
Subject: [PATCH] dsputil: allow to skip drawing of top/bottom edges.

---
 libavcodec/dsputil.c         | 25 +++++++---------
 libavcodec/dsputil.h         |  4 ++-
 libavcodec/mpegvideo.c       | 12 ++++++--
 libavcodec/snow.c            | 12 ++++++--
 libavcodec/x86/dsputil_mmx.c | 69 ++++++++++++++++++++++++--------------------
 5 files changed, 70 insertions(+), 52 deletions(-)

diff --git a/libavcodec/dsputil.c b/libavcodec/dsputil.c
index 15925f6..33fc78a 100644
--- a/libavcodec/dsputil.c
+++ b/libavcodec/dsputil.c
@@ -298,17 +298,11 @@ static int sse16_c(void *v, uint8_t *pix1, uint8_t *pix2, int line_size, int h)
 
 /* draw the edges of width 'w' of an image of size width, height */
 //FIXME check that this is ok for mpeg4 interlaced
-static void draw_edges_c(uint8_t *buf, int wrap, int width, int height, int w)
+static void draw_edges_c(uint8_t *buf, int wrap, int width, int height, int w, int sides)
 {
     uint8_t *ptr, *last_line;
     int i;
 
-    last_line = buf + (height - 1) * wrap;
-    for(i=0;i<w;i++) {
-        /* top and bottom */
-        memcpy(buf - (i + 1) * wrap, buf, width);
-        memcpy(last_line + (i + 1) * wrap, last_line, width);
-    }
     /* left and right */
     ptr = buf;
     for(i=0;i<height;i++) {
@@ -316,13 +310,16 @@ static void draw_edges_c(uint8_t *buf, int wrap, int width, int height, int w)
         memset(ptr + width, ptr[width-1], w);
         ptr += wrap;
     }
-    /* corners */
-    for(i=0;i<w;i++) {
-        memset(buf - (i + 1) * wrap - w, buf[0], w); /* top left */
-        memset(buf - (i + 1) * wrap + width, buf[width-1], w); /* top right */
-        memset(last_line + (i + 1) * wrap - w, last_line[0], w); /* top left */
-        memset(last_line + (i + 1) * wrap + width, last_line[width-1], w); /* top right */
-    }
+
+    /* top and bottom + corners */
+    buf -= w;
+    last_line = buf + (height - 1) * wrap;
+    if (sides & EDGE_TOP)
+        for(i = 0; i < w; i++)
+            memcpy(buf - (i + 1) * wrap, buf, width + w + w); // top
+    if (sides & EDGE_BOTTOM)
+        for (i = 0; i < w; i++)
+            memcpy(last_line + (i + 1) * wrap, last_line, width + w + w); // bottom
 }
 
 /**
diff --git a/libavcodec/dsputil.h b/libavcodec/dsputil.h
index a5ae68a..99b3283 100644
--- a/libavcodec/dsputil.h
+++ b/libavcodec/dsputil.h
@@ -492,8 +492,10 @@ typedef struct DSPContext {
 #define BASIS_SHIFT 16
 #define RECON_SHIFT 6
 
-    void (*draw_edges)(uint8_t *buf, int wrap, int width, int height, int w);
+    void (*draw_edges)(uint8_t *buf, int wrap, int width, int height, int w, int sides);
 #define EDGE_WIDTH 16
+#define EDGE_TOP    1
+#define EDGE_BOTTOM 2
 
     void (*prefetch)(void *mem, int stride, int h);
 
diff --git a/libavcodec/mpegvideo.c b/libavcodec/mpegvideo.c
index 66aba1b..1c5ff27 100644
--- a/libavcodec/mpegvideo.c
+++ b/libavcodec/mpegvideo.c
@@ -1067,9 +1067,15 @@ void MPV_frame_end(MpegEncContext *s)
        && s->current_picture.reference
        && !s->intra_only
        && !(s->flags&CODEC_FLAG_EMU_EDGE)) {
-            s->dsp.draw_edges(s->current_picture.data[0], s->linesize  , s->h_edge_pos   , s->v_edge_pos   , EDGE_WIDTH  );
-            s->dsp.draw_edges(s->current_picture.data[1], s->uvlinesize, s->h_edge_pos>>1, s->v_edge_pos>>1, EDGE_WIDTH/2);
-            s->dsp.draw_edges(s->current_picture.data[2], s->uvlinesize, s->h_edge_pos>>1, s->v_edge_pos>>1, EDGE_WIDTH/2);
+            s->dsp.draw_edges(s->current_picture.data[0], s->linesize  ,
+                              s->h_edge_pos   , s->v_edge_pos   ,
+                              EDGE_WIDTH  , EDGE_TOP | EDGE_BOTTOM);
+            s->dsp.draw_edges(s->current_picture.data[1], s->uvlinesize,
+                              s->h_edge_pos>>1, s->v_edge_pos>>1,
+                              EDGE_WIDTH/2, EDGE_TOP | EDGE_BOTTOM);
+            s->dsp.draw_edges(s->current_picture.data[2], s->uvlinesize,
+                              s->h_edge_pos>>1, s->v_edge_pos>>1,
+                              EDGE_WIDTH/2, EDGE_TOP | EDGE_BOTTOM);
     }
     emms_c();
 
diff --git a/libavcodec/snow.c b/libavcodec/snow.c
index de5d2dc..42145f5 100644
--- a/libavcodec/snow.c
+++ b/libavcodec/snow.c
@@ -1976,9 +1976,15 @@ static int frame_start(SnowContext *s){
    int h= s->avctx->height;
 
     if(s->current_picture.data[0]){
-        s->dsp.draw_edges(s->current_picture.data[0], s->current_picture.linesize[0], w   , h   , EDGE_WIDTH  );
-        s->dsp.draw_edges(s->current_picture.data[1], s->current_picture.linesize[1], w>>1, h>>1, EDGE_WIDTH/2);
-        s->dsp.draw_edges(s->current_picture.data[2], s->current_picture.linesize[2], w>>1, h>>1, EDGE_WIDTH/2);
+        s->dsp.draw_edges(s->current_picture.data[0],
+                          s->current_picture.linesize[0], w   , h   ,
+                          EDGE_WIDTH  , EDGE_TOP | EDGE_BOTTOM);
+        s->dsp.draw_edges(s->current_picture.data[1],
+                          s->current_picture.linesize[1], w>>1, h>>1,
+                          EDGE_WIDTH/2, EDGE_TOP | EDGE_BOTTOM);
+        s->dsp.draw_edges(s->current_picture.data[2],
+                          s->current_picture.linesize[2], w>>1, h>>1,
+                          EDGE_WIDTH/2, EDGE_TOP | EDGE_BOTTOM);
     }
 
     release_buffer(s->avctx);
diff --git a/libavcodec/x86/dsputil_mmx.c b/libavcodec/x86/dsputil_mmx.c
index 4d1a305..f98e6ae 100644
--- a/libavcodec/x86/dsputil_mmx.c
+++ b/libavcodec/x86/dsputil_mmx.c
@@ -783,7 +783,7 @@ static void h263_h_loop_filter_mmx(uint8_t *src, int stride, int qscale){
 
 /* draw the edges of width 'w' of an image of size width, height
    this mmx version can only handle w==8 || w==16 */
-static void draw_edges_mmx(uint8_t *buf, int wrap, int width, int height, int w)
+static void draw_edges_mmx(uint8_t *buf, int wrap, int width, int height, int w, int sides)
 {
     uint8_t *ptr, *last_line;
     int i;
@@ -836,36 +836,43 @@ static void draw_edges_mmx(uint8_t *buf, int wrap, int width, int height, int w)
         );
     }
 
-    for(i=0;i<w;i+=4) {
-        /* top and bottom (and hopefully also the corners) */
-        ptr= buf - (i + 1) * wrap - w;
-        __asm__ volatile(
-                "1:                             \n\t"
-                "movq (%1, %0), %%mm0           \n\t"
-                "movq %%mm0, (%0)               \n\t"
-                "movq %%mm0, (%0, %2)           \n\t"
-                "movq %%mm0, (%0, %2, 2)        \n\t"
-                "movq %%mm0, (%0, %3)           \n\t"
-                "add $8, %0                     \n\t"
-                "cmp %4, %0                     \n\t"
-                " jb 1b                         \n\t"
-                : "+r" (ptr)
-                : "r" ((x86_reg)buf - (x86_reg)ptr - w), "r" ((x86_reg)-wrap), "r" ((x86_reg)-wrap*3), "r" (ptr+width+2*w)
-        );
-        ptr= last_line + (i + 1) * wrap - w;
-        __asm__ volatile(
-                "1:                             \n\t"
-                "movq (%1, %0), %%mm0           \n\t"
-                "movq %%mm0, (%0)               \n\t"
-                "movq %%mm0, (%0, %2)           \n\t"
-                "movq %%mm0, (%0, %2, 2)        \n\t"
-                "movq %%mm0, (%0, %3)           \n\t"
-                "add $8, %0                     \n\t"
-                "cmp %4, %0                     \n\t"
-                " jb 1b                         \n\t"
-                : "+r" (ptr)
-                : "r" ((x86_reg)last_line - (x86_reg)ptr - w), "r" ((x86_reg)wrap), "r" ((x86_reg)wrap*3), "r" (ptr+width+2*w)
-        );
+    /* top and bottom (and hopefully also the corners) */
+    if (sides&EDGE_TOP) {
+        for(i = 0; i < w; i += 4) {
+            ptr= buf - (i + 1) * wrap - w;
+            __asm__ volatile(
+                    "1:                             \n\t"
+                    "movq (%1, %0), %%mm0           \n\t"
+                    "movq %%mm0, (%0)               \n\t"
+                    "movq %%mm0, (%0, %2)           \n\t"
+                    "movq %%mm0, (%0, %2, 2)        \n\t"
+                    "movq %%mm0, (%0, %3)           \n\t"
+                    "add $8, %0                     \n\t"
+                    "cmp %4, %0                     \n\t"
+                    " jb 1b                         \n\t"
+                    : "+r" (ptr)
+                    : "r" ((x86_reg)buf - (x86_reg)ptr - w), "r" ((x86_reg)-wrap), "r" ((x86_reg)-wrap*3), "r" (ptr+width+2*w)
+            );
+        }
+    }
+
+    if (sides&EDGE_BOTTOM) {
+        for(i = 0; i < w; i += 4) {
+            ptr= last_line + (i + 1) * wrap - w;
+            __asm__ volatile(
+                    "1:                             \n\t"
+                    "movq (%1, %0), %%mm0           \n\t"
+                    "movq %%mm0, (%0)               \n\t"
+                    "movq %%mm0, (%0, %2)           \n\t"
+                    "movq %%mm0, (%0, %2, 2)        \n\t"
+                    "movq %%mm0, (%0, %3)           \n\t"
+                    "add $8, %0                     \n\t"
+                    "cmp %4, %0                     \n\t"
+                    " jb 1b                         \n\t"
+                    : "+r" (ptr)
+                    : "r" ((x86_reg)last_line - (x86_reg)ptr - w), "r" ((x86_reg)wrap), "r" ((x86_reg)wrap*3), "r" (ptr+width+2*w)
+            );
+        }
     }
 }
 
-- 
2.7.4