sm501: Optimise 1 pixel 2d ops

Some guests do 1x1 blits which is faster to do directly than calling a
function for it so avoid overhead in this case.

Signed-off-by: BALATON Zoltan <balaton@eik.bme.hu>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Message-id: 7cccc302d7b4c5c313bad7681ac4686417143c3e.1592686588.git.balaton@eik.bme.hu
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>

hw/display/sm501.c

@@ -794,6 +794,14 @@ static void sm501_2d_operation(SM501State *s)
                 src_x == dst_x && src_y == dst_y) {
                 break;
             }
+            /* Some clients also do 1 pixel blits, avoid overhead for these */
+            if (width == 1 && height == 1) {
+                unsigned int si = (src_x + src_y * src_pitch) * bypp;
+                unsigned int di = (dst_x + dst_y * dst_pitch) * bypp;
+                stn_he_p(&s->local_mem[dst_base + di], bypp,
+                         ldn_he_p(&s->local_mem[src_base + si], bypp));
+                break;
+            }
             /* Check for overlaps, this could be made more exact */
             uint32_t sb, se, db, de;
             sb = src_base + src_x + src_y * (width + src_pitch);
@@ -842,9 +850,14 @@ static void sm501_2d_operation(SM501State *s)
             color = cpu_to_le16(color);
         }
 
-        pixman_fill((uint32_t *)&s->local_mem[dst_base],
-                    dst_pitch * bypp / sizeof(uint32_t),
-                    8 * bypp, dst_x, dst_y, width, height, color);
+        if (width == 1 && height == 1) {
+            unsigned int i = (dst_x + dst_y * dst_pitch) * bypp;
+            stn_he_p(&s->local_mem[dst_base + i], bypp, color);
+        } else {
+            pixman_fill((uint32_t *)&s->local_mem[dst_base],
+                        dst_pitch * bypp / sizeof(uint32_t),
+                        8 * bypp, dst_x, dst_y, width, height, color);
+        }
         break;
     }
     default: