Merge remote branch 'origin/master' into nv50-compiler

Conflicts:
	src/gallium/drivers/nv50/nv50_program.c

1 files changed, 95 insertions, 35 deletions

diff --git a/src/gallium/auxiliary/util/u_blit.c b/src/gallium/auxiliary/util/u_blit.c
index 97fa99ec65d..dfb142b9e1c 100644
--- a/src/gallium/auxiliary/util/u_blit.c
+++ b/src/gallium/auxiliary/util/u_blit.c
@@ -42,6 +42,7 @@
 
 #include "util/u_blit.h"
 #include "util/u_draw_quad.h"
+#include "util/u_format.h"
 #include "util/u_math.h"
 #include "util/u_memory.h"
 #include "util/u_sampler.h"
@@ -56,15 +57,18 @@ struct blit_state
    struct cso_context *cso;
 
    struct pipe_blend_state blend;
-   struct pipe_depth_stencil_alpha_state depthstencil;
+   struct pipe_depth_stencil_alpha_state depthstencil_keep;
+   struct pipe_depth_stencil_alpha_state depthstencil_write;
    struct pipe_rasterizer_state rasterizer;
    struct pipe_sampler_state sampler;
    struct pipe_viewport_state viewport;
    struct pipe_clip_state clip;
    struct pipe_vertex_element velem[2];
+   enum pipe_texture_target internal_target;
 
    void *vs;
    void *fs[TGSI_WRITEMASK_XYZW + 1];
+   void *fs_depth;
 
    struct pipe_resource *vbuf;  /**< quad vertices */
    unsigned vbuf_slot;
@@ -95,7 +99,11 @@ util_create_blit(struct pipe_context *pipe, struct cso_context *cso)
    ctx->blend.rt[0].colormask = PIPE_MASK_RGBA;
 
    /* no-op depth/stencil/alpha */
-   memset(&ctx->depthstencil, 0, sizeof(ctx->depthstencil));
+   memset(&ctx->depthstencil_keep, 0, sizeof(ctx->depthstencil_keep));
+   memset(&ctx->depthstencil_write, 0, sizeof(ctx->depthstencil_write));
+   ctx->depthstencil_write.depth.enabled = 1;
+   ctx->depthstencil_write.depth.writemask = 1;
+   ctx->depthstencil_write.depth.func = PIPE_FUNC_ALWAYS;
 
    /* rasterizer */
    memset(&ctx->rasterizer, 0, sizeof(ctx->rasterizer));
@@ -110,7 +118,6 @@ util_create_blit(struct pipe_context *pipe, struct cso_context *cso)
    ctx->sampler.min_mip_filter = PIPE_TEX_MIPFILTER_NONE;
    ctx->sampler.min_img_filter = 0; /* set later */
    ctx->sampler.mag_img_filter = 0; /* set later */
-   ctx->sampler.normalized_coords = 1;
 
    /* vertex elements state */
    memset(&ctx->velem[0], 0, sizeof(ctx->velem[0]) * 2);
@@ -145,6 +152,11 @@ util_create_blit(struct pipe_context *pipe, struct cso_context *cso)
       ctx->vertices[i][1][3] = 1.0f; /* q */
    }
 
+   if(pipe->screen->get_param(pipe->screen, PIPE_CAP_NPOT_TEXTURES))
+      ctx->internal_target = PIPE_TEXTURE_2D;
+   else
+      ctx->internal_target = PIPE_TEXTURE_RECT;
+
    return ctx;
 }
 
@@ -164,6 +176,9 @@ util_destroy_blit(struct blit_state *ctx)
       if (ctx->fs[i])
          pipe->delete_fs_state(pipe, ctx->fs[i]);
 
+   if (ctx->fs_depth)
+      pipe->delete_fs_state(pipe, ctx->fs_depth);
+
    pipe_resource_reference(&ctx->vbuf, NULL);
 
    FREE(ctx);
@@ -271,7 +286,7 @@ regions_overlap(int srcX0, int srcY0,
  * \param writemask  controls which channels in the dest surface are sourced
  *                   from the src surface.  Disabled channels are sourced
  *                   from (0,0,0,1).
- * XXX need some control over blitting Z and/or stencil.
+ * XXX need some control over blitting stencil.
  */
 void
 util_blit_pixels_writemask(struct blit_state *ctx,
@@ -294,8 +309,9 @@ util_blit_pixels_writemask(struct blit_state *ctx,
    const int srcW = abs(srcX1 - srcX0);
    const int srcH = abs(srcY1 - srcY0);
    unsigned offset;
-   boolean overlap;
+   boolean overlap, dst_is_depth;
    float s0, t0, s1, t1;
+   boolean normalized;
 
    assert(filter == PIPE_TEX_MIPFILTER_NEAREST ||
           filter == PIPE_TEX_MIPFILTER_LINEAR);
@@ -335,7 +351,6 @@ util_blit_pixels_writemask(struct blit_state *ctx,
       return;
    }
 
-
    /* Create a temporary texture when src and dest alias or when src
     * is anything other than a 2d texture.
     * XXX should just use appropriate shader to access 1d / 3d slice / cube face,
@@ -347,7 +362,8 @@ util_blit_pixels_writemask(struct blit_state *ctx,
        dst->face == srcsub.face &&
        dst->level == srcsub.level &&
        dst->zslice == srcZ0) ||
-       src_tex->target != PIPE_TEXTURE_2D)
+       (src_tex->target != PIPE_TEXTURE_2D &&
+       src_tex->target != PIPE_TEXTURE_RECT))
    {
       struct pipe_resource texTemp;
       struct pipe_resource *tex;
@@ -372,7 +388,7 @@ util_blit_pixels_writemask(struct blit_state *ctx,
 
       /* create temp texture */
       memset(&texTemp, 0, sizeof(texTemp));
-      texTemp.target = PIPE_TEXTURE_2D;
+      texTemp.target = ctx->internal_target;
       texTemp.format = src_tex->format;
       texTemp.last_level = 0;
       texTemp.width0 = srcW;
@@ -392,10 +408,19 @@ util_blit_pixels_writemask(struct blit_state *ctx,
                                  src_tex, srcsub, srcLeft, srcTop, srcZ0, /* src */
                                  srcW, srcH);     /* size */
 
-      s0 = 0.0f; 
-      s1 = 1.0f;
-      t0 = 0.0f;
-      t1 = 1.0f;
+      normalized = tex->target != PIPE_TEXTURE_RECT;
+      if(normalized) {
+         s0 = 0.0f;
+         s1 = 1.0f;
+         t0 = 0.0f;
+         t1 = 1.0f;
+      }
+      else {
+         s0 = 0;
+         s1 = srcW;
+         t0 = 0;
+         t1 = srcH;
+      }
 
       u_sampler_view_default_template(&sv_templ, tex, tex->format);
       sampler_view = pipe->create_sampler_view(pipe, tex, &sv_templ);
@@ -415,20 +440,29 @@ util_blit_pixels_writemask(struct blit_state *ctx,
          return;
       }
 
-      s0 = srcX0 / (float)(u_minify(sampler_view->texture->width0, srcsub.level));
-      s1 = srcX1 / (float)(u_minify(sampler_view->texture->width0, srcsub.level));
-      t0 = srcY0 / (float)(u_minify(sampler_view->texture->height0, srcsub.level));
-      t1 = srcY1 / (float)(u_minify(sampler_view->texture->height0, srcsub.level));
+      s0 = srcX0;
+      s1 = srcX1;
+      t0 = srcY0;
+      t1 = srcY1;
+      normalized = sampler_view->texture->target != PIPE_TEXTURE_RECT;
+      if(normalized)
+      {
+         s0 /= (float)(u_minify(sampler_view->texture->width0, srcsub.level));
+         s1 /= (float)(u_minify(sampler_view->texture->width0, srcsub.level));
+         t0 /= (float)(u_minify(sampler_view->texture->height0, srcsub.level));
+         t1 /= (float)(u_minify(sampler_view->texture->height0, srcsub.level));
+      }
    }
 
+   dst_is_depth = util_format_is_depth_or_stencil(dst->format);
 
-   assert(screen->is_format_supported(screen, sampler_view->format, PIPE_TEXTURE_2D,
+   assert(screen->is_format_supported(screen, sampler_view->format, ctx->internal_target,
                                       sampler_view->texture->nr_samples,
                                       PIPE_BIND_SAMPLER_VIEW, 0));
-   assert(screen->is_format_supported(screen, dst->format, PIPE_TEXTURE_2D,
+   assert(screen->is_format_supported(screen, dst->format, ctx->internal_target,
                                       dst->texture->nr_samples,
-                                      PIPE_BIND_RENDER_TARGET, 0));
-
+                                      dst_is_depth ? PIPE_BIND_DEPTH_STENCIL :
+                                                     PIPE_BIND_RENDER_TARGET, 0));
    /* save state (restored below) */
    cso_save_blend(ctx->cso);
    cso_save_depth_stencil_alpha(ctx->cso);
@@ -444,12 +478,15 @@ util_blit_pixels_writemask(struct blit_state *ctx,
 
    /* set misc state we care about */
    cso_set_blend(ctx->cso, &ctx->blend);
-   cso_set_depth_stencil_alpha(ctx->cso, &ctx->depthstencil);
+   cso_set_depth_stencil_alpha(ctx->cso,
+                               dst_is_depth ? &ctx->depthstencil_write :
+                                              &ctx->depthstencil_keep);
    cso_set_rasterizer(ctx->cso, &ctx->rasterizer);
    cso_set_clip(ctx->cso, &ctx->clip);
    cso_set_vertex_elements(ctx->cso, 2, ctx->velem);
 
    /* sampler */
+   ctx->sampler.normalized_coords = normalized;
    ctx->sampler.min_img_filter = filter;
    ctx->sampler.mag_img_filter = filter;
    /* we've limited this already with the sampler view but you never know... */
@@ -472,22 +509,35 @@ util_blit_pixels_writemask(struct blit_state *ctx,
    /* texture */
    cso_set_fragment_sampler_views(ctx->cso, 1, &sampler_view);
 
-   if (ctx->fs[writemask] == NULL)
-      ctx->fs[writemask] =
-         util_make_fragment_tex_shader_writemask(pipe, TGSI_TEXTURE_2D,
-                                                 TGSI_INTERPOLATE_LINEAR,
-                                                 writemask);
-
    /* shaders */
-   cso_set_fragment_shader_handle(ctx->cso, ctx->fs[writemask]);
+   if (dst_is_depth) {
+      if (ctx->fs_depth == NULL)
+         ctx->fs_depth =
+            util_make_fragment_tex_shader_writedepth(pipe, TGSI_TEXTURE_2D,
+                                                     TGSI_INTERPOLATE_LINEAR);
+
+      cso_set_fragment_shader_handle(ctx->cso, ctx->fs_depth);
+   } else {
+      if (ctx->fs[writemask] == NULL)
+         ctx->fs[writemask] =
+            util_make_fragment_tex_shader_writemask(pipe, TGSI_TEXTURE_2D,
+                                                    TGSI_INTERPOLATE_LINEAR,
+                                                    writemask);
+
+      cso_set_fragment_shader_handle(ctx->cso, ctx->fs[writemask]);
+   }
    cso_set_vertex_shader_handle(ctx->cso, ctx->vs);
 
    /* drawing dest */
    memset(&fb, 0, sizeof(fb));
    fb.width = dst->width;
    fb.height = dst->height;
-   fb.nr_cbufs = 1;
-   fb.cbufs[0] = dst;
+   if (dst_is_depth) {
+      fb.zsbuf = dst;
+   } else {
+      fb.nr_cbufs = 1;
+      fb.cbufs[0] = dst;
+   }
    cso_set_framebuffer(ctx->cso, &fb);
 
    /* draw quad */
@@ -574,6 +624,7 @@ util_blit_pixels_tex(struct blit_state *ctx,
                      int dstX1, int dstY1,
                      float z, uint filter)
 {
+   boolean normalized = src_sampler_view->texture->target != PIPE_TEXTURE_RECT;
    struct pipe_framebuffer_state fb;
    float s0, t0, s1, t1;
    unsigned offset;
@@ -586,10 +637,18 @@ util_blit_pixels_tex(struct blit_state *ctx,
    assert(tex->width0 != 0);
    assert(tex->height0 != 0);
 
-   s0 = srcX0 / (float)tex->width0;
-   s1 = srcX1 / (float)tex->width0;
-   t0 = srcY0 / (float)tex->height0;
-   t1 = srcY1 / (float)tex->height0;
+   s0 = srcX0;
+   s1 = srcX1;
+   t0 = srcY0;
+   t1 = srcY1;
+
+   if(normalized)
+   {
+      s0 /= (float)tex->width0;
+      s1 /= (float)tex->width0;
+      t0 /= (float)tex->height0;
+      t1 /= (float)tex->height0;
+   }
 
    assert(ctx->pipe->screen->is_format_supported(ctx->pipe->screen, dst->format,
                                                  PIPE_TEXTURE_2D,
@@ -611,12 +670,13 @@ util_blit_pixels_tex(struct blit_state *ctx,
 
    /* set misc state we care about */
    cso_set_blend(ctx->cso, &ctx->blend);
-   cso_set_depth_stencil_alpha(ctx->cso, &ctx->depthstencil);
+   cso_set_depth_stencil_alpha(ctx->cso, &ctx->depthstencil_keep);
    cso_set_rasterizer(ctx->cso, &ctx->rasterizer);
    cso_set_clip(ctx->cso, &ctx->clip);
    cso_set_vertex_elements(ctx->cso, 2, ctx->velem);
 
    /* sampler */
+   ctx->sampler.normalized_coords = normalized;
    ctx->sampler.min_img_filter = filter;
    ctx->sampler.mag_img_filter = filter;
    cso_single_sampler(ctx->cso, 0, &ctx->sampler);