xf86-video-intel: 3 commits - src/sna/gen4_render.c src/sna/sna_io.c

[Chris Wilson]

 src/sna/gen4_render.c |   16 +++++++---------
 src/sna/sna_io.c      |   44 +++++++++++++++++++++++++++++++++-----------
 2 files changed, 40 insertions(+), 20 deletions(-)

New commits:
commit b6d2bb961517623d46aa6944307cb998ee125459
Author: Chris Wilson <chris at chris-wilson.co.uk>
Date:   Tue Nov 13 10:05:56 2012 +0000

    sna/gen4: Do not prefer inplace non-rectilinear spans
    
    As gen4 requires the per-rectangle vertex flush, emitting spans on the
    GPU is inefficient and so we prefer to composite the mask instead.
    
    Signed-off-by: Chris Wilson <chris at chris-wilson.co.uk>

diff --git a/src/sna/gen4_render.c b/src/sna/gen4_render.c
index 95a8363..3aa33db 100644
--- a/src/sna/gen4_render.c
+++ b/src/sna/gen4_render.c
@@ -49,6 +49,7 @@
  */
 #define PREFER_BLT 1
 #define FLUSH_EVERY_VERTEX 1
+#define FORCE_SPANS 0
 
 #define NO_COMPOSITE 0
 #define NO_COMPOSITE_SPANS 0
@@ -2624,16 +2625,11 @@ gen4_check_composite_spans(struct sna *sna,
 		return false;
 	}
 
-	if ((flags & (COMPOSITE_SPANS_RECTILINEAR | COMPOSITE_SPANS_INPLACE_HINT)) == 0) {
-		struct sna_pixmap *priv = sna_pixmap_from_drawable(dst->pDrawable);
-		assert(priv);
+	if (FORCE_SPANS)
+		return FORCE_SPANS > 0;
 
-		if ((priv->cpu_bo && kgem_bo_is_busy(priv->cpu_bo)) ||
-		    (priv->gpu_bo && kgem_bo_is_busy(priv->gpu_bo))) {
-			return true;
-		}
-
-		DBG(("%s: fallback, non-rectilinear spans to idle bo\n",
+	if ((flags & COMPOSITE_SPANS_RECTILINEAR) == 0) {
+		DBG(("%s: fallback, non-rectilinear spans\n",
 		     __FUNCTION__));
 		return false;
 	}
commit ae293609c7400cd3c753ed3762772264c4741df5
Author: Chris Wilson <chris at chris-wilson.co.uk>
Date:   Tue Nov 13 10:21:29 2012 +0000

    sna/gen4: Always initialise redirect
    
    Do not assume the caller cleared the composite-op structure for us.
    
    Signed-off-by: Chris Wilson <chris at chris-wilson.co.uk>

diff --git a/src/sna/gen4_render.c b/src/sna/gen4_render.c
index be97458..95a8363 100644
--- a/src/sna/gen4_render.c
+++ b/src/sna/gen4_render.c
@@ -2294,6 +2294,7 @@ gen4_render_composite(struct sna *sna,
 		return false;
 	sna_render_reduce_damage(tmp, dst_x, dst_y, width, height);
 
+	sna_render_composite_redirect_init(tmp);
 	if (too_large(tmp->dst.width, tmp->dst.height) &&
 	    !sna_render_composite_redirect(sna, tmp,
 					   dst_x, dst_y, width, height))
@@ -2669,6 +2670,7 @@ gen4_render_composite_spans(struct sna *sna,
 		return false;
 	sna_render_reduce_damage(&tmp->base, dst_x, dst_y, width, height);
 
+	sna_render_composite_redirect_init(&tmp->base);
 	if (too_large(tmp->base.dst.width, tmp->base.dst.height)) {
 		if (!sna_render_composite_redirect(sna, &tmp->base,
 						   dst_x, dst_y, width, height))
commit 2954f15e2bcb590a90c2cb6077c0843ee25a4413
Author: Chris Wilson <chris at chris-wilson.co.uk>
Date:   Tue Nov 13 09:46:19 2012 +0000

    sna: Specialise the decision for inplace xor uploads
    
    Fixes a regression from
    
    commit 0be1d964713ca407f029278a8256d02d925dc9da
    Author: Chris Wilson <chris at chris-wilson.co.uk>
    Date:   Tue Sep 11 21:48:24 2012 +0100
    
        sna: Use inplace X tiling for LLC uploads
    
    which introduced the ability to swizzle into CPU maps, but also
    convinced the xorg path to the same - which for large images blows up.
    
    Reported-by: Michael Laß <bevan at bi-co.net>
    Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=57031
    Signed-off-by: Chris Wilson <chris at chris-wilson.co.uk>

diff --git a/src/sna/sna_io.c b/src/sna/sna_io.c
index 69d920c..2038e5d 100644
--- a/src/sna/sna_io.c
+++ b/src/sna/sna_io.c
@@ -579,19 +579,13 @@ static bool write_boxes_inplace(struct kgem *kgem,
 	return true;
 }
 
-static bool upload_inplace(struct kgem *kgem,
-			   struct kgem_bo *bo,
-			   const BoxRec *box,
-			   int n, int bpp)
+static bool __upload_inplace(struct kgem *kgem,
+			     struct kgem_bo *bo,
+			     const BoxRec *box,
+			     int n, int bpp)
 {
 	unsigned int bytes;
 
-	if (kgem->wedged)
-		return true;
-
-	if (!kgem_bo_can_map(kgem, bo) && !upload_inplace__tiled(kgem, bo))
-		return false;
-
 	if (FORCE_INPLACE)
 		return FORCE_INPLACE > 0;
 
@@ -610,6 +604,20 @@ static bool upload_inplace(struct kgem *kgem,
 		return bytes * bpp >> 12;
 }
 
+static bool upload_inplace(struct kgem *kgem,
+			   struct kgem_bo *bo,
+			   const BoxRec *box,
+			   int n, int bpp)
+{
+	if (kgem->wedged)
+		return true;
+
+	if (!kgem_bo_can_map(kgem, bo) && !upload_inplace__tiled(kgem, bo))
+		return false;
+
+	return __upload_inplace(kgem, bo, box, n,bpp);
+}
+
 bool sna_write_boxes(struct sna *sna, PixmapPtr dst,
 		     struct kgem_bo * const dst_bo, int16_t const dst_dx, int16_t const dst_dy,
 		     const void * const src, int const stride, int16_t const src_dx, int16_t const src_dy,
@@ -960,6 +968,20 @@ write_boxes_inplace__xor(struct kgem *kgem,
 	} while (--n);
 }
 
+static bool upload_inplace__xor(struct kgem *kgem,
+				struct kgem_bo *bo,
+				const BoxRec *box,
+				int n, int bpp)
+{
+	if (kgem->wedged)
+		return true;
+
+	if (!kgem_bo_can_map(kgem, bo))
+		return false;
+
+	return __upload_inplace(kgem, bo, box, n, bpp);
+}
+
 void sna_write_boxes__xor(struct sna *sna, PixmapPtr dst,
 			  struct kgem_bo *dst_bo, int16_t dst_dx, int16_t dst_dy,
 			  const void *src, int stride, int16_t src_dx, int16_t src_dy,
@@ -976,7 +998,7 @@ void sna_write_boxes__xor(struct sna *sna, PixmapPtr dst,
 
 	DBG(("%s x %d\n", __FUNCTION__, nbox));
 
-	if (upload_inplace(kgem, dst_bo, box, nbox, dst->drawable.bitsPerPixel)) {
+	if (upload_inplace__xor(kgem, dst_bo, box, nbox, dst->drawable.bitsPerPixel)) {
 fallback:
 		write_boxes_inplace__xor(kgem,
 					 src, stride, dst->drawable.bitsPerPixel, src_dx, src_dy,