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

Conflicts:
	Makefile
	src/gallium/auxiliary/util/u_surface.c
	src/gallium/drivers/llvmpipe/lp_flush.c
	src/gallium/drivers/llvmpipe/lp_setup.c
	src/gallium/drivers/llvmpipe/lp_state_derived.c
	src/gallium/drivers/llvmpipe/lp_state_fs.c
	src/gallium/drivers/llvmpipe/lp_state_surface.c
	src/gallium/drivers/llvmpipe/lp_tex_cache.c
	src/gallium/drivers/llvmpipe/lp_texture.c
	src/gallium/drivers/llvmpipe/lp_tile_cache.c
	src/mesa/state_tracker/st_cb_condrender.c

13 files changed, 95 insertions, 70 deletions

diff --git a/src/mesa/swrast/s_aatritemp.h b/src/mesa/swrast/s_aatritemp.h
index 0827b3db9eb..a8b22c548f3 100644
--- a/src/mesa/swrast/s_aatritemp.h
+++ b/src/mesa/swrast/s_aatritemp.h
@@ -220,11 +220,11 @@
 
 #if defined(DO_ATTRIBS)
          /* compute attributes at left-most fragment */
-         span.attrStart[FRAG_ATTRIB_WPOS][3] = solve_plane(ix + 0.5, iy + 0.5, wPlane);
+         span.attrStart[FRAG_ATTRIB_WPOS][3] = solve_plane(ix + 0.5F, iy + 0.5F, wPlane);
          ATTRIB_LOOP_BEGIN
             GLuint c;
             for (c = 0; c < 4; c++) {
-               span.attrStart[attr][c] = solve_plane(ix + 0.5, iy + 0.5, attrPlane[attr][c]);
+               span.attrStart[attr][c] = solve_plane(ix + 0.5F, iy + 0.5F, attrPlane[attr][c]);
             }
          ATTRIB_LOOP_END
 #endif
@@ -328,11 +328,11 @@
          
 #if defined(DO_ATTRIBS)
          /* compute attributes at left-most fragment */
-         span.attrStart[FRAG_ATTRIB_WPOS][3] = solve_plane(ix + 1.5, iy + 0.5, wPlane);
+         span.attrStart[FRAG_ATTRIB_WPOS][3] = solve_plane(ix + 1.5, iy + 0.5F, wPlane);
          ATTRIB_LOOP_BEGIN
             GLuint c;
             for (c = 0; c < 4; c++) {
-               span.attrStart[attr][c] = solve_plane(ix + 1.5, iy + 0.5, attrPlane[attr][c]);
+               span.attrStart[attr][c] = solve_plane(ix + 1.5, iy + 0.5F, attrPlane[attr][c]);
             }
          ATTRIB_LOOP_END
 #endif
diff --git a/src/mesa/swrast/s_accum.c b/src/mesa/swrast/s_accum.c
index cf53f01b7c1..2dd9ca6348b 100644
--- a/src/mesa/swrast/s_accum.c
+++ b/src/mesa/swrast/s_accum.c
@@ -37,7 +37,7 @@
 /* XXX this would have to change for accum buffers with more or less
  * than 16 bits per color channel.
  */
-#define ACCUM_SCALE16 32767.0
+#define ACCUM_SCALE16 32767.0F
 
 
 /*
diff --git a/src/mesa/swrast/s_alpha.c b/src/mesa/swrast/s_alpha.c
index 5761bb00b4e..509477433a5 100644
--- a/src/mesa/swrast/s_alpha.c
+++ b/src/mesa/swrast/s_alpha.c
@@ -146,8 +146,8 @@ _swrast_alpha_test(const GLcontext *ctx, SWspan *span)
          ALPHA_TEST(FixedToInt(alpha), alpha += alphaStep);
       }
       else {
-         const GLfloat alphaStep = span->alphaStep;
-         GLfloat alpha = span->alpha;
+         const GLfloat alphaStep = FixedToFloat(span->alphaStep);
+         GLfloat alpha = FixedToFloat(span->alpha);
          const GLfloat ref = ctx->Color.AlphaRef;
          ALPHA_TEST(alpha, alpha += alphaStep);
       }
diff --git a/src/mesa/swrast/s_atifragshader.c b/src/mesa/swrast/s_atifragshader.c
index 353e9999d61..05da64de3ac 100644
--- a/src/mesa/swrast/s_atifragshader.c
+++ b/src/mesa/swrast/s_atifragshader.c
@@ -82,10 +82,11 @@ apply_swizzle(GLfloat values[4], GLuint swizzle)
       break;
    case GL_SWIZZLE_STQ_DQ_ATI:
 /* make sure q is not 0 to avoid problems later with infinite values (texture lookup)? */
-      if (q == 0.0F) q = 0.000000001;
+      if (q == 0.0F)
+         q = 0.000000001F;
       values[0] = s / q;
       values[1] = t / q;
-      values[2] = 1 / q;
+      values[2] = 1.0 / q;
       break;
    }
    values[3] = 0.0;
@@ -171,27 +172,27 @@ apply_dst_mod(GLuint optype, GLuint mod, GLfloat * val)
 	 val[i] = 8 * val[i];
 	 break;
       case GL_HALF_BIT_ATI:
-	 val[i] = val[i] * 0.5;
+	 val[i] = val[i] * 0.5F;
 	 break;
       case GL_QUARTER_BIT_ATI:
-	 val[i] = val[i] * 0.25;
+	 val[i] = val[i] * 0.25F;
 	 break;
       case GL_EIGHTH_BIT_ATI:
-	 val[i] = val[i] * 0.125;
+	 val[i] = val[i] * 0.125F;
 	 break;
       }
 
       if (has_sat) {
-	 if (val[i] < 0.0)
-	    val[i] = 0;
-	 else if (val[i] > 1.0)
-	    val[i] = 1.0;
+	 if (val[i] < 0.0F)
+	    val[i] = 0.0F;
+	 else if (val[i] > 1.0F)
+	    val[i] = 1.0F;
       }
       else {
-	 if (val[i] < -8.0)
-	    val[i] = -8.0;
-	 else if (val[i] > 8.0)
-	    val[i] = 8.0;
+	 if (val[i] < -8.0F)
+	    val[i] = -8.0F;
+	 else if (val[i] > 8.0F)
+	    val[i] = 8.0F;
       }
    }
 }
diff --git a/src/mesa/swrast/s_context.c b/src/mesa/swrast/s_context.c
index f9092c215a7..0d4680db9f7 100644
--- a/src/mesa/swrast/s_context.c
+++ b/src/mesa/swrast/s_context.c
@@ -149,26 +149,26 @@ _swrast_update_polygon( GLcontext *ctx )
    if (ctx->Polygon.CullFlag) {
       switch (ctx->Polygon.CullFaceMode) {
       case GL_BACK:
-         backface_sign = -1.0;
+         backface_sign = -1.0F;
 	 break;
       case GL_FRONT:
-         backface_sign = 1.0;
+         backface_sign = 1.0F;
 	 break;
       case GL_FRONT_AND_BACK:
          /* fallthrough */
       default:
-	 backface_sign = 0.0;
+	 backface_sign = 0.0F;
       }
    }
    else {
-      backface_sign = 0.0;
+      backface_sign = 0.0F;
    }
 
    SWRAST_CONTEXT(ctx)->_BackfaceCullSign = backface_sign;
 
    /* This is for front/back-face determination, but not for culling */
    SWRAST_CONTEXT(ctx)->_BackfaceSign
-      = (ctx->Polygon.FrontFace == GL_CW) ? -1.0 : 1.0;
+      = (ctx->Polygon.FrontFace == GL_CW) ? -1.0F : 1.0F;
 }
 
 
diff --git a/src/mesa/swrast/s_depth.c b/src/mesa/swrast/s_depth.c
index 0b6bb7e3ec9..ac5dae2148b 100644
--- a/src/mesa/swrast/s_depth.c
+++ b/src/mesa/swrast/s_depth.c
@@ -497,17 +497,24 @@ depth_test_span32( GLcontext *ctx, GLuint n,
    return passed;
 }
 
-/* Apply ARB_depth_clamp to span of fragments. */
+
+
+/**
+ * Clamp fragment Z values to the depth near/far range (glDepthRange()).
+ * This is used when GL_ARB_depth_clamp/GL_DEPTH_CLAMP is turned on.
+ * In that case, vertexes are not clipped against the near/far planes
+ * so rasterization will produce fragment Z values outside the usual
+ * [0,1] range.
+ */
 void
 _swrast_depth_clamp_span( GLcontext *ctx, SWspan *span )
 {
    struct gl_framebuffer *fb = ctx->DrawBuffer;
-   struct gl_renderbuffer *rb = fb->_DepthBuffer;
    const GLuint count = span->end;
-   GLuint *zValues = span->array->z;
-   GLuint min, max;
+   GLint *zValues = (GLint *) span->array->z; /* sign change */
+   GLint min, max;
    GLfloat min_f, max_f;
-   int i;
+   GLuint i;
 
    if (ctx->Viewport.Near < ctx->Viewport.Far) {
       min_f = ctx->Viewport.Near;
@@ -517,15 +524,21 @@ _swrast_depth_clamp_span( GLcontext *ctx, SWspan *span )
       max_f = ctx->Viewport.Near;
    }
 
-   if (rb->DataType == GL_UNSIGNED_SHORT) {
-      CLAMPED_FLOAT_TO_USHORT(min, min_f);
-      CLAMPED_FLOAT_TO_USHORT(max, max_f);
-   } else {
-      assert(rb->DataType == GL_UNSIGNED_INT);
-      min = FLOAT_TO_UINT(min_f);
-      max = FLOAT_TO_UINT(max_f);
-   }
-
+   /* Convert floating point values in [0,1] to device Z coordinates in
+    * [0, DepthMax].
+    * ex: If the the Z buffer has 24 bits, DepthMax = 0xffffff.
+    * 
+    * XXX this all falls apart if we have 31 or more bits of Z because
+    * the triangle rasterization code produces unsigned Z values.  Negative
+    * vertex Z values come out as large fragment Z uints.
+    */
+   min = (GLint) (min_f * fb->_DepthMaxF);
+   max = (GLint) (max_f * fb->_DepthMaxF);
+   if (max < 0)
+      max = 0x7fffffff; /* catch over flow for 30-bit z */
+
+   /* Note that we do the comparisons here using signed integers.
+    */
    for (i = 0; i < count; i++) {
       if (zValues[i] < min)
 	 zValues[i] = min;
diff --git a/src/mesa/swrast/s_fog.c b/src/mesa/swrast/s_fog.c
index 77ed0cfef96..0472bbf553c 100644
--- a/src/mesa/swrast/s_fog.c
+++ b/src/mesa/swrast/s_fog.c
@@ -172,14 +172,14 @@ _swrast_fog_rgba_span( const GLcontext *ctx, SWspan *span )
 
    /* compute (scaled) fog color */
    if (span->array->ChanType == GL_UNSIGNED_BYTE) {
-      rFog = ctx->Fog.Color[RCOMP] * 255.0;
-      gFog = ctx->Fog.Color[GCOMP] * 255.0;
-      bFog = ctx->Fog.Color[BCOMP] * 255.0;
+      rFog = ctx->Fog.Color[RCOMP] * 255.0F;
+      gFog = ctx->Fog.Color[GCOMP] * 255.0F;
+      bFog = ctx->Fog.Color[BCOMP] * 255.0F;
    }
    else if (span->array->ChanType == GL_UNSIGNED_SHORT) {
-      rFog = ctx->Fog.Color[RCOMP] * 65535.0;
-      gFog = ctx->Fog.Color[GCOMP] * 65535.0;
-      bFog = ctx->Fog.Color[BCOMP] * 65535.0;
+      rFog = ctx->Fog.Color[RCOMP] * 65535.0F;
+      gFog = ctx->Fog.Color[GCOMP] * 65535.0F;
+      bFog = ctx->Fog.Color[BCOMP] * 65535.0F;
    }
    else {
       rFog = ctx->Fog.Color[RCOMP];
diff --git a/src/mesa/swrast/s_fragprog.c b/src/mesa/swrast/s_fragprog.c
index 9ac33a26a65..d31da4c4022 100644
--- a/src/mesa/swrast/s_fragprog.c
+++ b/src/mesa/swrast/s_fragprog.c
@@ -144,12 +144,22 @@ init_machine(GLcontext *ctx, struct gl_program_machine *machine,
              const struct gl_fragment_program *program,
              const SWspan *span, GLuint col)
 {
+   GLfloat *wpos = span->array->attribs[FRAG_ATTRIB_WPOS][col];
+
    if (program->Base.Target == GL_FRAGMENT_PROGRAM_NV) {
       /* Clear temporary registers (undefined for ARB_f_p) */
       _mesa_bzero(machine->Temporaries,
                   MAX_PROGRAM_TEMPS * 4 * sizeof(GLfloat));
    }
 
+   /* ARB_fragment_coord_conventions */
+   if (program->OriginUpperLeft)
+      wpos[1] = ctx->DrawBuffer->Height - 1 - wpos[1];
+   if (!program->PixelCenterInteger) {
+      wpos[0] += 0.5F;
+      wpos[1] += 0.5F;
+   }
+
    /* Setup pointer to input attributes */
    machine->Attribs = span->array->attribs;
 
@@ -162,7 +172,7 @@ init_machine(GLcontext *ctx, struct gl_program_machine *machine,
    /* if running a GLSL program (not ARB_fragment_program) */
    if (ctx->Shader.CurrentProgram) {
       /* Store front/back facing value */
-      machine->Attribs[FRAG_ATTRIB_FACE][col][0] = 1.0 - span->facing;
+      machine->Attribs[FRAG_ATTRIB_FACE][col][0] = 1.0F - span->facing;
    }
 
    machine->CurElement = col;
diff --git a/src/mesa/swrast/s_points.c b/src/mesa/swrast/s_points.c
index 6b955429e94..a9a704a16ee 100644
--- a/src/mesa/swrast/s_points.c
+++ b/src/mesa/swrast/s_points.c
@@ -125,16 +125,16 @@ sprite_point(GLcontext *ctx, const SWvertex *vert)
       GLfloat s, r, dsdx;
 
       /* texcoord / pointcoord interpolants */
-      s = 0.0;
-      dsdx = 1.0 / size;
+      s = 0.0F;
+      dsdx = 1.0F / size;
       if (ctx->Point.SpriteOrigin == GL_LOWER_LEFT) {
-         dtdy = 1.0 / size;
-         t0 = 0.5 * dtdy;
+         dtdy = 1.0F / size;
+         t0 = 0.5F * dtdy;
       }
       else {
          /* GL_UPPER_LEFT */
-         dtdy = -1.0 / size;
-         t0 = 1.0 + 0.5 * dtdy;
+         dtdy = -1.0F / size;
+         t0 = 1.0F + 0.5F * dtdy;
       }
 
       ATTRIB_LOOP_BEGIN
diff --git a/src/mesa/swrast/s_span.c b/src/mesa/swrast/s_span.c
index 4ea9547cd92..905cf3d5501 100644
--- a/src/mesa/swrast/s_span.c
+++ b/src/mesa/swrast/s_span.c
@@ -645,7 +645,7 @@ interpolate_wpos(GLcontext *ctx, SWspan *span)
 {
    GLfloat (*wpos)[4] = span->array->attribs[FRAG_ATTRIB_WPOS];
    GLuint i;
-   const GLfloat zScale = 1.0 / ctx->DrawBuffer->_DepthMaxF;
+   const GLfloat zScale = 1.0F / ctx->DrawBuffer->_DepthMaxF;
    GLfloat w, dw;
 
    if (span->arrayMask & SPAN_XY) {
@@ -1316,6 +1316,13 @@ _swrast_write_rgba_span( GLcontext *ctx, SWspan *span)
 
    ASSERT(span->end <= MAX_WIDTH);
 
+   /* Depth bounds test */
+   if (ctx->Depth.BoundsTest && fb->Visual.depthBits > 0) {
+      if (!_swrast_depth_bounds_test(ctx, span)) {
+         return;
+      }
+   }
+
 #ifdef DEBUG
    /* Make sure all fragments are within window bounds */
    if (span->arrayMask & SPAN_XY) {
diff --git a/src/mesa/swrast/s_texcombine.c b/src/mesa/swrast/s_texcombine.c
index 594b71a03c6..95e2c082ba7 100644
--- a/src/mesa/swrast/s_texcombine.c
+++ b/src/mesa/swrast/s_texcombine.c
@@ -316,18 +316,18 @@ texture_combine( GLcontext *ctx, GLuint unit, GLuint n,
             /* (a * b) + (c * d) - 0.5 */
             for (i = 0; i < n; i++) {
                rgba[i][RCOMP] = (arg0[i][RCOMP] * arg1[i][RCOMP] +
-                                 arg2[i][RCOMP] * arg3[i][RCOMP] - 0.5) * scaleRGB;
+                                 arg2[i][RCOMP] * arg3[i][RCOMP] - 0.5F) * scaleRGB;
                rgba[i][GCOMP] = (arg0[i][GCOMP] * arg1[i][GCOMP] +
-                                 arg2[i][GCOMP] * arg3[i][GCOMP] - 0.5) * scaleRGB;
+                                 arg2[i][GCOMP] * arg3[i][GCOMP] - 0.5F) * scaleRGB;
                rgba[i][BCOMP] = (arg0[i][BCOMP] * arg1[i][BCOMP] +
-                                 arg2[i][BCOMP] * arg3[i][BCOMP] - 0.5) * scaleRGB;
+                                 arg2[i][BCOMP] * arg3[i][BCOMP] - 0.5F) * scaleRGB;
             }
          }
          else {
             for (i = 0; i < n; i++) {
-               rgba[i][RCOMP] = (arg0[i][RCOMP] + arg1[i][RCOMP] - 0.5) * scaleRGB;
-               rgba[i][GCOMP] = (arg0[i][GCOMP] + arg1[i][GCOMP] - 0.5) * scaleRGB;
-               rgba[i][BCOMP] = (arg0[i][BCOMP] + arg1[i][BCOMP] - 0.5) * scaleRGB;
+               rgba[i][RCOMP] = (arg0[i][RCOMP] + arg1[i][RCOMP] - 0.5F) * scaleRGB;
+               rgba[i][GCOMP] = (arg0[i][GCOMP] + arg1[i][GCOMP] - 0.5F) * scaleRGB;
+               rgba[i][BCOMP] = (arg0[i][BCOMP] + arg1[i][BCOMP] - 0.5F) * scaleRGB;
             }
          }
          break;
@@ -385,11 +385,11 @@ texture_combine( GLcontext *ctx, GLuint unit, GLuint n,
       case GL_MODULATE_SIGNED_ADD_ATI:
          for (i = 0; i < n; i++) {
             rgba[i][RCOMP] = ((arg0[i][RCOMP] * arg2[i][RCOMP]) +
-                              arg1[i][RCOMP] - 0.5) * scaleRGB;
+                              arg1[i][RCOMP] - 0.5F) * scaleRGB;
             rgba[i][GCOMP] = ((arg0[i][GCOMP] * arg2[i][GCOMP]) +
-                              arg1[i][GCOMP] - 0.5) * scaleRGB;
+                              arg1[i][GCOMP] - 0.5F) * scaleRGB;
             rgba[i][BCOMP] = ((arg0[i][BCOMP] * arg2[i][BCOMP]) +
-                              arg1[i][BCOMP] - 0.5) * scaleRGB;
+                              arg1[i][BCOMP] - 0.5F) * scaleRGB;
 	 }
          break;
       case GL_MODULATE_SUBTRACT_ATI:
@@ -456,7 +456,7 @@ texture_combine( GLcontext *ctx, GLuint unit, GLuint n,
             for (i = 0; i < n; i++) {
                rgba[i][ACOMP] = (arg0[i][ACOMP] * arg1[i][ACOMP] +
                                  arg2[i][ACOMP] * arg3[i][ACOMP] -
-                                 0.5) * scaleA;
+                                 0.5F) * scaleA;
             }
          }
          else {
diff --git a/src/mesa/swrast/s_texfilter.c b/src/mesa/swrast/s_texfilter.c
index 76b65cc755e..ff7deecc392 100644
--- a/src/mesa/swrast/s_texfilter.c
+++ b/src/mesa/swrast/s_texfilter.c
@@ -445,7 +445,7 @@ clamp_rect_coord_linear(GLenum wrapMode, GLfloat coord, GLint max,
    switch (wrapMode) {
    case GL_CLAMP:
       /* Not exactly what the spec says, but it matches NVIDIA output */
-      fcol = CLAMP(coord - 0.5F, 0.0, max-1);
+      fcol = CLAMP(coord - 0.5F, 0.0F, max - 1);
       i0 = IFLOOR(fcol);
       i1 = i0 + 1;
       break;
diff --git a/src/mesa/swrast/swrast.h b/src/mesa/swrast/swrast.h
index c183b315b67..c01cf7d1f0b 100644
--- a/src/mesa/swrast/swrast.h
+++ b/src/mesa/swrast/swrast.h
@@ -75,12 +75,6 @@ typedef struct {
 } SWvertex;
 
 
-/**
- * Fixed point data type.
- */
-typedef int GLfixed;
-
-
 #define FRAG_ATTRIB_CI FRAG_ATTRIB_COL0