av_memcpy_backptr: optimise some special cases

- Add special cases for offsets of 2, 3, or 4 bytes.  This means the
  offset is always >4 in the generic case, allowing 32-bit copies to
  be used there.
- Don't use memcpy() for sizes less than 16 bytes.
Signed-off-by: Mans Rullgard <mans@mansr.com>

libavutil/mem.c

@@ -180,29 +180,93 @@ char *av_strdup(const char *s)
     return ptr;
 }
 
+static void fill16(uint8_t *dst, int len)
+{
+    uint32_t v = AV_RN16(dst - 2);
+
+    v |= v << 16;
+
+    while (len >= 4) {
+        AV_WN32(dst, v);
+        dst += 4;
+        len -= 4;
+    }
+
+    while (len--) {
+        *dst = dst[-2];
+        dst++;
+    }
+}
+
+static void fill24(uint8_t *dst, int len)
+{
+#if HAVE_BIGENDIAN
+    uint32_t v = AV_RB24(dst - 3);
+    uint32_t a = v << 8  | v >> 16;
+    uint32_t b = v << 16 | v >> 8;
+    uint32_t c = v << 24 | v;
+#else
+    uint32_t v = AV_RL24(dst - 3);
+    uint32_t a = v       | v << 24;
+    uint32_t b = v >> 8  | v << 16;
+    uint32_t c = v >> 16 | v << 8;
+#endif
+
+    while (len >= 12) {
+        AV_WN32(dst,     a);
+        AV_WN32(dst + 4, b);
+        AV_WN32(dst + 8, c);
+        dst += 12;
+        len -= 12;
+    }
+
+    if (len >= 4) {
+        AV_WN32(dst, a);
+        dst += 4;
+        len -= 4;
+    }
+
+    if (len >= 4) {
+        AV_WN32(dst, b);
+        dst += 4;
+        len -= 4;
+    }
+
+    while (len--) {
+        *dst = dst[-3];
+        dst++;
+    }
+}
+
+static void fill32(uint8_t *dst, int len)
+{
+    uint32_t v = AV_RN32(dst - 4);
+
+    while (len >= 4) {
+        AV_WN32(dst, v);
+        dst += 4;
+        len -= 4;
+    }
+
+    while (len--) {
+        *dst = dst[-4];
+        dst++;
+    }
+}
+
 void av_memcpy_backptr(uint8_t *dst, int back, int cnt)
 {
     const uint8_t *src = &dst[-back];
     if (back == 1) {
         memset(dst, *src, cnt);
+    } else if (back == 2) {
+        fill16(dst, cnt);
+    } else if (back == 3) {
+        fill24(dst, cnt);
+    } else if (back == 4) {
+        fill32(dst, cnt);
     } else {
-        if (cnt >= 4) {
-            AV_COPY16U(dst,     src);
-            AV_COPY16U(dst + 2, src + 2);
-            src += 4;
-            dst += 4;
-            cnt -= 4;
-        }
-        if (cnt >= 8) {
-            AV_COPY16U(dst,     src);
-            AV_COPY16U(dst + 2, src + 2);
-            AV_COPY16U(dst + 4, src + 4);
-            AV_COPY16U(dst + 6, src + 6);
-            src += 8;
-            dst += 8;
-            cnt -= 8;
-        }
-        if (cnt > 0) {
+        if (cnt >= 16) {
             int blocklen = back;
             while (cnt > blocklen) {
                 memcpy(dst, src, blocklen);
@@ -211,6 +275,28 @@ void av_memcpy_backptr(uint8_t *dst, int back, int cnt)
                 blocklen <<= 1;
             }
             memcpy(dst, src, cnt);
+            return;
+        }
+        if (cnt >= 8) {
+            AV_COPY32U(dst,     src);
+            AV_COPY32U(dst + 4, src + 4);
+            src += 8;
+            dst += 8;
+            cnt -= 8;
+        }
+        if (cnt >= 4) {
+            AV_COPY32U(dst, src);
+            src += 4;
+            dst += 4;
+            cnt -= 4;
+        }
+        if (cnt >= 2) {
+            AV_COPY16U(dst, src);
+            src += 2;
+            dst += 2;
+            cnt -= 2;
         }
+        if (cnt)
+            *dst = *src;
     }
 }