Merge remote branch 'origin/master' into lp-binning

Conflicts:
	src/gallium/drivers/llvmpipe/lp_quad.h
	src/gallium/drivers/llvmpipe/lp_setup.c

7 files changed, 316 insertions, 42 deletions

diff --git a/src/gallium/auxiliary/util/u_blitter.c b/src/gallium/auxiliary/util/u_blitter.c
index cef3b69e46d..249a0375fc5 100644
--- a/src/gallium/auxiliary/util/u_blitter.c
+++ b/src/gallium/auxiliary/util/u_blitter.c
@@ -163,6 +163,7 @@ struct blitter_context *util_blitter_create(struct pipe_context *pipe)
    rs_state.cull_mode = PIPE_WINDING_NONE;
    rs_state.bypass_vs_clip_and_viewport = 1;
    rs_state.gl_rasterization_rules = 1;
+   rs_state.flatshade = 1;
    ctx->rs_state = pipe->create_rasterizer_state(pipe, &rs_state);
 
    /* fragment shaders are created on-demand */
@@ -378,9 +379,16 @@ static void blitter_set_texcoords_cube(struct blitter_context_priv *ctx,
    float t1 = y1 / (float)surf->height;
    float s2 = x2 / (float)surf->width;
    float t2 = y2 / (float)surf->height;
-   const float st[4][2] = {
-      {s1, t1}, {s2, t1}, {s2, t2}, {s1, t2}
-   };
+   float st[4][2];
+
+   st[0][0] = s1;
+   st[0][1] = t1;
+   st[1][0] = s2;
+   st[1][1] = t1;
+   st[2][0] = s2;
+   st[2][1] = t2;
+   st[3][0] = s1;
+   st[3][1] = t2;
 
    util_map_texcoords2d_onto_cubemap(surf->face,
                                      /* pointer, stride in floats */
@@ -567,45 +575,29 @@ void util_blitter_clear(struct blitter_context *blitter,
    blitter_restore_CSOs(ctx);
 }
 
-void util_blitter_copy(struct blitter_context *blitter,
-                       struct pipe_surface *dst,
-                       unsigned dstx, unsigned dsty,
-                       struct pipe_surface *src,
-                       unsigned srcx, unsigned srcy,
-                       unsigned width, unsigned height,
-                       boolean ignore_stencil)
+static boolean
+is_overlap(unsigned sx1, unsigned sx2, unsigned sy1, unsigned sy2,
+           unsigned dx1, unsigned dx2, unsigned dy1, unsigned dy2)
+{
+    if (sx1 >= dx2 || sx2 <= dx1 || sy1 >= dy2 || sy2 <= dy1) {
+        return FALSE;
+    } else {
+        return TRUE;
+    }
+}
+
+static void util_blitter_do_copy(struct blitter_context *blitter,
+				 struct pipe_surface *dst,
+				 unsigned dstx, unsigned dsty,
+				 struct pipe_surface *src,
+				 unsigned srcx, unsigned srcy,
+				 unsigned width, unsigned height,
+				 boolean is_depth)
 {
    struct blitter_context_priv *ctx = (struct blitter_context_priv*)blitter;
    struct pipe_context *pipe = ctx->pipe;
-   struct pipe_screen *screen = pipe->screen;
    struct pipe_framebuffer_state fb_state;
-   boolean is_stencil, is_depth;
-   unsigned dst_tex_usage;
-
-   /* give up if textures are not set */
-   assert(dst->texture && src->texture);
-   if (!dst->texture || !src->texture)
-      return;
-
-   is_depth = util_format_get_component_bits(src->format, UTIL_FORMAT_COLORSPACE_ZS, 0) != 0;
-   is_stencil = util_format_get_component_bits(src->format, UTIL_FORMAT_COLORSPACE_ZS, 1) != 0;
-   dst_tex_usage = is_depth || is_stencil ? PIPE_TEXTURE_USAGE_DEPTH_STENCIL :
-                                            PIPE_TEXTURE_USAGE_RENDER_TARGET;
 
-   /* check if we can sample from and render to the surfaces */
-   /* (assuming copying a stencil buffer is not possible) */
-   if ((!ignore_stencil && is_stencil) ||
-       !screen->is_format_supported(screen, dst->format, dst->texture->target,
-                                    dst_tex_usage, 0) ||
-       !screen->is_format_supported(screen, src->format, src->texture->target,
-                                    PIPE_TEXTURE_USAGE_SAMPLER, 0)) {
-      util_surface_copy(pipe, FALSE, dst, dstx, dsty, src, srcx, srcy,
-                        width, height);
-      return;
-   }
-
-   /* check whether the states are properly saved */
-   blitter_check_saved_CSOs(ctx);
    assert(blitter->saved_fb_state.nr_cbufs != ~0);
    assert(blitter->saved_num_textures != ~0);
    assert(blitter->saved_num_sampler_states != ~0);
@@ -663,6 +655,108 @@ void util_blitter_copy(struct blitter_context *blitter,
 
    blitter_set_rectangle(ctx, dstx, dsty, dstx+width, dsty+height, 0);
    blitter_draw_quad(ctx);
+
+}
+
+static void util_blitter_overlap_copy(struct blitter_context *blitter,
+				      struct pipe_surface *dst,
+				      unsigned dstx, unsigned dsty,
+				      struct pipe_surface *src,
+				      unsigned srcx, unsigned srcy,
+				      unsigned width, unsigned height)
+{
+   struct blitter_context_priv *ctx = (struct blitter_context_priv*)blitter;
+   struct pipe_context *pipe = ctx->pipe;
+   struct pipe_screen *screen = pipe->screen;
+
+   struct pipe_texture texTemp;
+   struct pipe_texture *texture;
+   struct pipe_surface *tex_surf;
+
+   /* check whether the states are properly saved */
+   blitter_check_saved_CSOs(ctx);
+
+   memset(&texTemp, 0, sizeof(texTemp));
+   texTemp.target = PIPE_TEXTURE_2D;
+   texTemp.format = dst->texture->format; /* XXX verify supported by driver! */
+   texTemp.last_level = 0;
+   texTemp.width0 = width;
+   texTemp.height0 = height;
+   texTemp.depth0 = 1;
+
+   texture = screen->texture_create(screen, &texTemp);
+   if (!texture)
+      return;
+
+   tex_surf = screen->get_tex_surface(screen, texture, 0, 0, 0,
+				      PIPE_BUFFER_USAGE_GPU_READ | 
+				      PIPE_BUFFER_USAGE_GPU_WRITE);
+
+   /* blit from the src to the temp */
+   util_blitter_do_copy(blitter, tex_surf, 0, 0,
+			src, srcx, srcy,
+			width, height,
+			FALSE);
+   util_blitter_do_copy(blitter, dst, dstx, dsty,
+			tex_surf, 0, 0,
+			width, height,
+			FALSE);
+   pipe_surface_reference(&tex_surf, NULL);
+   pipe_texture_reference(&texture, NULL);
+   blitter_restore_CSOs(ctx);
+}
+
+void util_blitter_copy(struct blitter_context *blitter,
+                       struct pipe_surface *dst,
+                       unsigned dstx, unsigned dsty,
+                       struct pipe_surface *src,
+                       unsigned srcx, unsigned srcy,
+                       unsigned width, unsigned height,
+                       boolean ignore_stencil)
+{
+   struct blitter_context_priv *ctx = (struct blitter_context_priv*)blitter;
+   struct pipe_context *pipe = ctx->pipe;
+   struct pipe_screen *screen = pipe->screen;
+   boolean is_stencil, is_depth;
+   unsigned dst_tex_usage;
+
+   /* give up if textures are not set */
+   assert(dst->texture && src->texture);
+   if (!dst->texture || !src->texture)
+      return;
+
+   if (dst->texture == src->texture) {
+      if (is_overlap(srcx, srcx + width, srcy, srcy + height,
+		             dstx, dstx + width, dsty, dsty + height)) {
+         util_blitter_overlap_copy(blitter, dst, dstx, dsty, src, srcx, srcy,
+                                   width, height);
+         return;
+      }
+   }
+		   
+   is_depth = util_format_get_component_bits(src->format, UTIL_FORMAT_COLORSPACE_ZS, 0) != 0;
+   is_stencil = util_format_get_component_bits(src->format, UTIL_FORMAT_COLORSPACE_ZS, 1) != 0;
+   dst_tex_usage = is_depth || is_stencil ? PIPE_TEXTURE_USAGE_DEPTH_STENCIL :
+                                            PIPE_TEXTURE_USAGE_RENDER_TARGET;
+
+   /* check if we can sample from and render to the surfaces */
+   /* (assuming copying a stencil buffer is not possible) */
+   if ((!ignore_stencil && is_stencil) ||
+       !screen->is_format_supported(screen, dst->format, dst->texture->target,
+                                    dst_tex_usage, 0) ||
+       !screen->is_format_supported(screen, src->format, src->texture->target,
+                                    PIPE_TEXTURE_USAGE_SAMPLER, 0)) {
+      util_surface_copy(pipe, FALSE, dst, dstx, dsty, src, srcx, srcy,
+                        width, height);
+      return;
+   }
+
+   /* check whether the states are properly saved */
+   blitter_check_saved_CSOs(ctx);
+   util_blitter_do_copy(blitter,
+			dst, dstx, dsty,
+			src, srcx, srcy,
+			width, height, is_depth);
    blitter_restore_CSOs(ctx);
 }
 
diff --git a/src/gallium/auxiliary/util/u_draw_quad.c b/src/gallium/auxiliary/util/u_draw_quad.c
index 4110485fb19..e2e23c3cdd8 100644
--- a/src/gallium/auxiliary/util/u_draw_quad.c
+++ b/src/gallium/auxiliary/util/u_draw_quad.c
@@ -61,6 +61,7 @@ util_draw_vertex_buffer(struct pipe_context *pipe,
    /* tell pipe about the vertex attributes */
    for (i = 0; i < num_attribs; i++) {
       velements[i].src_offset = i * 4 * sizeof(float);
+      velements[i].instance_divisor = 0;
       velements[i].vertex_buffer_index = 0;
       velements[i].src_format = PIPE_FORMAT_R32G32B32A32_FLOAT;
       velements[i].nr_components = 4;
diff --git a/src/gallium/auxiliary/util/u_format.csv b/src/gallium/auxiliary/util/u_format.csv
index 9f16b42944e..01f7931aed1 100644
--- a/src/gallium/auxiliary/util/u_format.csv
+++ b/src/gallium/auxiliary/util/u_format.csv
@@ -62,10 +62,10 @@ PIPE_FORMAT_R16G16_SSCALED        , array , 1, 1, s16 , s16 ,     ,     , xy01,
 PIPE_FORMAT_R16G16B16_SSCALED     , array , 1, 1, s16 , s16 , s16 ,     , xyz1, rgb
 PIPE_FORMAT_R16G16B16A16_SSCALED  , array , 1, 1, s16 , s16 , s16 , s16 , xyzw, rgb
 PIPE_FORMAT_R8_UNORM              , array , 1, 1, un8 ,     ,     ,     , x001, rgb
-PIPE_FORMAT_R8G8_UNORM            , array , 1, 1, un8 , un8 ,     ,     , xy01, rgb
-PIPE_FORMAT_R8G8B8_UNORM          , array , 1, 1, un8 , un8 , un8 ,     , xyz1, rgb
-PIPE_FORMAT_R8G8B8A8_UNORM        , array , 1, 1, un8 , un8 , un8 , un8 , xyzw, rgb
-PIPE_FORMAT_R8G8B8X8_UNORM        , array , 1, 1, un8 , un8 , un8 , un8 , xyz1, rgb
+PIPE_FORMAT_R8G8_UNORM            , array , 1, 1, un8 , un8 ,     ,     , yx01, rgb
+PIPE_FORMAT_R8G8B8_UNORM          , array , 1, 1, un8 , un8 , un8 ,     , zyx1, rgb
+PIPE_FORMAT_R8G8B8A8_UNORM        , array , 1, 1, un8 , un8 , un8 , un8 , wzyx, rgb
+PIPE_FORMAT_R8G8B8X8_UNORM        , array , 1, 1, un8 , un8 , un8 , un8 , wzy1, rgb
 PIPE_FORMAT_R8_USCALED            , array , 1, 1, u8  ,     ,     ,     , x001, rgb
 PIPE_FORMAT_R8G8_USCALED          , array , 1, 1, u8  , u8  ,     ,     , xy01, rgb
 PIPE_FORMAT_R8G8B8_USCALED        , array , 1, 1, u8  , u8  , u8  ,     , xyz1, rgb
diff --git a/src/gallium/auxiliary/util/u_pack_color.h b/src/gallium/auxiliary/util/u_pack_color.h
index 43eb0153ee7..0ab53c75dd6 100644
--- a/src/gallium/auxiliary/util/u_pack_color.h
+++ b/src/gallium/auxiliary/util/u_pack_color.h
@@ -425,6 +425,8 @@ util_pack_z(enum pipe_format format, double z)
       if (z == 1.0)
          return 0xffffffff;
       return (uint) (z * 0xffffffff);
+   case PIPE_FORMAT_Z32_FLOAT:
+      return (uint)z;
    case PIPE_FORMAT_S8Z24_UNORM:
    case PIPE_FORMAT_X8Z24_UNORM:
       if (z == 1.0)
diff --git a/src/gallium/auxiliary/util/u_ringbuffer.c b/src/gallium/auxiliary/util/u_ringbuffer.c
new file mode 100644
index 00000000000..3f43a19e018
--- /dev/null
+++ b/src/gallium/auxiliary/util/u_ringbuffer.c
@@ -0,0 +1,145 @@
+
+#include "pipe/p_thread.h"
+#include "pipe/p_defines.h"
+#include "util/u_ringbuffer.h"
+#include "util/u_math.h"
+#include "util/u_memory.h"
+
+/* Generic ringbuffer: 
+ */
+struct util_ringbuffer 
+{
+   struct util_packet *buf;
+   unsigned mask;
+
+   /* Can this be done with atomic variables??
+    */
+   unsigned head;
+   unsigned tail;
+   pipe_condvar change;
+   pipe_mutex mutex;
+};
+
+
+struct util_ringbuffer *util_ringbuffer_create( unsigned dwords )
+{
+   struct util_ringbuffer *ring = CALLOC_STRUCT(util_ringbuffer);
+   if (ring == NULL)
+      return NULL;
+
+   assert(util_is_power_of_two(dwords));
+   
+   ring->buf = MALLOC( dwords * sizeof(unsigned) );
+   if (ring->buf == NULL)
+      goto fail;
+
+   ring->mask = dwords - 1;
+
+   pipe_condvar_init(ring->change);
+   pipe_mutex_init(ring->mutex);
+   return ring;
+
+fail:
+   FREE(ring->buf);
+   FREE(ring);
+   return NULL;
+}
+
+void util_ringbuffer_destroy( struct util_ringbuffer *ring )
+{
+   pipe_condvar_destroy(ring->change);
+   pipe_mutex_destroy(ring->mutex);
+   FREE(ring->buf);
+   FREE(ring);
+}
+
+static INLINE unsigned util_ringbuffer_space( const struct util_ringbuffer *ring )
+{
+   return (ring->tail - (ring->head + 1)) & ring->mask;
+}
+
+void util_ringbuffer_enqueue( struct util_ringbuffer *ring,
+                              const struct util_packet *packet )
+{
+   unsigned i;
+
+   /* XXX: over-reliance on mutexes, etc:
+    */
+   pipe_mutex_lock(ring->mutex);
+
+   /* Wait for free space:
+    */
+   while (util_ringbuffer_space(ring) < packet->dwords)
+      pipe_condvar_wait(ring->change, ring->mutex);
+
+   /* Copy data to ring:
+    */
+   for (i = 0; i < packet->dwords; i++) {
+
+      /* Copy all dwords of the packet.  Note we're abusing the
+       * typesystem a little - we're being passed a pointer to
+       * something, but probably not an array of packet structs:
+       */
+      ring->buf[ring->head] = packet[i];
+      ring->head++;
+      ring->head &= ring->mask;
+   }
+
+   /* Signal change:
+    */
+   pipe_condvar_signal(ring->change);
+   pipe_mutex_unlock(ring->mutex);
+}
+
+enum pipe_error util_ringbuffer_dequeue( struct util_ringbuffer *ring,
+                                         struct util_packet *packet,
+                                         unsigned max_dwords,
+                                         boolean wait )
+{
+   const struct util_packet *ring_packet;
+   unsigned i;
+   int ret = PIPE_OK;
+
+   /* XXX: over-reliance on mutexes, etc:
+    */
+   pipe_mutex_lock(ring->mutex);
+
+   /* Wait for free space:
+    */
+   if (wait) {
+      while (util_ringbuffer_space(ring) == 0)
+         pipe_condvar_wait(ring->change, ring->mutex);
+   }
+   else {
+      if (util_ringbuffer_space(ring) == 0) {
+         ret = PIPE_ERROR_OUT_OF_MEMORY;
+         goto out;
+      }
+   }
+
+   ring_packet = &ring->buf[ring->tail];
+
+   /* Both of these are considered bugs.  Raise an assert on debug builds.
+    */
+   if (ring_packet->dwords > ring->mask + 1 - util_ringbuffer_space(ring) ||
+       ring_packet->dwords > max_dwords) {
+      assert(0);
+      ret = PIPE_ERROR_BAD_INPUT;
+      goto out;
+   }
+
+   /* Copy data from ring:
+    */
+   for (i = 0; i < ring_packet->dwords; i++) {
+      packet[i] = ring->buf[ring->tail];
+      ring->tail++;
+      ring->tail &= ring->mask;
+   }
+
+out:
+   /* Signal change:
+    */
+   pipe_condvar_signal(ring->change);
+   pipe_mutex_unlock(ring->mutex);
+   return ret;
+}
diff --git a/src/gallium/auxiliary/util/u_ringbuffer.h b/src/gallium/auxiliary/util/u_ringbuffer.h
new file mode 100644
index 00000000000..85f0ad6c1f6
--- /dev/null
+++ b/src/gallium/auxiliary/util/u_ringbuffer.h
@@ -0,0 +1,29 @@
+
+#ifndef UTIL_RINGBUFFER_H
+#define UTIL_RINGBUFFER_H
+
+#include "pipe/p_compiler.h"
+#include "pipe/p_defines.h"       /* only for pipe_error! */
+
+/* Generic header
+ */
+struct util_packet {
+   unsigned dwords:8;
+   unsigned data24:24;
+};
+
+struct util_ringbuffer;
+
+struct util_ringbuffer *util_ringbuffer_create( unsigned dwords );
+
+void util_ringbuffer_destroy( struct util_ringbuffer *ring );
+
+void util_ringbuffer_enqueue( struct util_ringbuffer *ring,
+                              const struct util_packet *packet );
+
+enum pipe_error util_ringbuffer_dequeue( struct util_ringbuffer *ring,
+                                         struct util_packet *packet,
+                                         unsigned max_dwords,
+                                         boolean wait );
+
+#endif
diff --git a/src/gallium/auxiliary/util/u_tile.c b/src/gallium/auxiliary/util/u_tile.c
index 1ba82bb21f0..f9936eb1cb2 100644
--- a/src/gallium/auxiliary/util/u_tile.c
+++ b/src/gallium/auxiliary/util/u_tile.c
@@ -1357,7 +1357,10 @@ pipe_put_tile_rgba(struct pipe_transfer *pt,
       /*z24s8_put_tile_rgba((unsigned *) packed, w, h, p, src_stride);*/
       break;
    default:
-      debug_printf("%s: unsupported format %s\n", __FUNCTION__, pf_name(format));
+      util_format_write_4f(format,
+                           p, src_stride * sizeof(float),
+                           packed, util_format_get_stride(format, w),
+                           0, 0, w, h);
    }
 
    pipe_put_tile_raw(pt, x, y, w, h, packed, 0);