[PATCH xserver 1/2] fb: Handle ZPixmap planemask in GetImage the other way around

[Adam Jackson]

Formerly we'd zero the image data and then pull out a plane at a time.
It's faster to apply the planemask after the fact, since that turns the
GetImage into a memcpy:

  100000.0  101000.0 (1.010) (copy 0xaaaaaaaa) ShmGetImage 10x10 square
   42400.0   59400.0 (1.401) (copy 0xaaaaaaaa) ShmGetImage 100x100 square
    3040.0    5280.0 (1.737) (copy 0xaaaaaaaa) ShmGetImage 500x500 square
   96100.0   95200.0 (0.991) (0xaaaaaaaa) GetImage 10x10 square
   29600.0   36800.0 (1.243) (0xaaaaaaaa) GetImage 100x100 square
    1850.0    2620.0 (1.416) (0xaaaaaaaa) GetImage 500x500 square

Measured with Xvfb at depth 24 on Skylake i7-6560U.

Signed-off-by: Adam Jackson <ajax at redhat.com>
---
 fb/fbimage.c | 9 ++++++---
 1 file changed, 6 insertions(+), 3 deletions(-)

diff --git a/fb/fbimage.c b/fb/fbimage.c
index 59daa21..8f5f3dc 100644
--- a/fb/fbimage.c
+++ b/fb/fbimage.c
@@ -250,13 +250,16 @@ fbGetImage(DrawablePtr pDrawable,
 
         pm = fbReplicatePixel(planeMask, srcBpp);
         dstStride = PixmapBytePad(w, pDrawable->depth);
-        if (pm != FB_ALLONES)
-            memset(d, 0, dstStride * h);
         dstStride /= sizeof(FbStip);
         fbBltStip((FbStip *) (src + (y + srcYoff) * srcStride),
                   FbBitsStrideToStipStride(srcStride),
                   (x + srcXoff) * srcBpp,
-                  dst, dstStride, 0, w * srcBpp, h, GXcopy, pm, srcBpp);
+                  dst, dstStride, 0, w * srcBpp, h, GXcopy, FB_ALLONES, srcBpp);
+
+        if (pm != FB_ALLONES) {
+            for (int i = 0; i < dstStride * h; i++)
+                dst[i] &= pm;
+        }
     }
     else {
         dstStride = BitmapBytePad(w) / sizeof(FbStip);
-- 
2.9.3