radeon/r200: add draw/stencil buffer detiling

This moves the detiling to the fbo mapping, r200 depth is always tiled,
and we can't detile it with the blitter.

Signed-off-by: Dave Airlie <airlied@redhat.com>

src/mesa/drivers/dri/radeon/radeon_common_context.h

@@ -90,6 +90,7 @@ struct radeon_renderbuffer
 	GLbitfield map_mode;
 	int map_x, map_y, map_w, map_h;
 	int map_pitch;
+	void *map_buffer;
 
 	uint32_t draw_offset; /* FBO */
 	/* boo Xorg 6.8.2 compat */

src/mesa/drivers/dri/radeon/radeon_fbo.c

@@ -70,6 +70,66 @@ radeon_delete_renderbuffer(struct gl_renderbuffer *rb)
   free(rrb);
 }
 
+#if defined(RADEON_R200)
+static GLuint r200_depth_4byte(const struct radeon_renderbuffer * rrb,
+				 GLint x, GLint y)
+{
+    GLuint offset;
+    GLuint b;
+    offset = 0;
+    b = (((y & 0x7ff) >> 4) * (rrb->pitch >> 7) + (x >> 5));
+    offset += (b >> 1) << 12;
+    offset += (((rrb->pitch >> 7) & 0x1) ? (b & 0x1) : ((b & 0x1) ^ ((y >> 4) & 0x1))) << 11;
+    offset += ((y >> 2) & 0x3) << 9;
+    offset += ((x >> 2) & 0x1) << 8;
+    offset += ((x >> 3) & 0x3) << 6;
+    offset += ((y >> 1) & 0x1) << 5;
+    offset += ((x >> 1) & 0x1) << 4;
+    offset += (y & 0x1) << 3;
+    offset += (x & 0x1) << 2;
+
+    return offset;
+}
+
+static void
+radeon_map_renderbuffer_s8z24(struct gl_context *ctx,
+		       struct gl_renderbuffer *rb,
+		       GLuint x, GLuint y, GLuint w, GLuint h,
+		       GLbitfield mode,
+		       GLubyte **out_map,
+		       GLint *out_stride)
+{
+    struct radeon_renderbuffer *rrb = radeon_renderbuffer(rb);
+    uint32_t *untiled_s8z24_map, *tiled_s8z24_map;
+    int ret;
+    int y_flip = (rb->Name == 0) ? -1 : 1;
+    int y_bias = (rb->Name == 0) ? (rb->Height - 1) : 0;
+    uint32_t pitch = w * rrb->cpp;
+
+    rrb->map_pitch = pitch;
+
+    rrb->map_buffer = malloc(w * h * 4);
+    ret = radeon_bo_map(rrb->bo, !!(mode & GL_MAP_WRITE_BIT));
+
+    untiled_s8z24_map = rrb->map_buffer;
+    tiled_s8z24_map = rrb->bo->ptr;
+
+    for (uint32_t pix_y = 0; pix_y < h; ++ pix_y) {
+	for (uint32_t pix_x = 0; pix_x < w; ++pix_x) {
+	    uint32_t flipped_y = y_flip * (int32_t)(y + pix_y) + y_bias;
+	    uint32_t src_offset = r200_depth_4byte(rrb, x + pix_x, flipped_y);
+	    uint32_t dst_offset = pix_y * rrb->map_pitch + pix_x * rrb->cpp;
+	    untiled_s8z24_map[dst_offset/4] = tiled_s8z24_map[src_offset/4];
+	}
+    }
+
+    radeon_bo_unmap(rrb->bo);
+		   
+    *out_map = rrb->map_buffer;
+    *out_stride = rrb->map_pitch;
+}
+#endif
+
 static void
 radeon_map_renderbuffer(struct gl_context *ctx,
 		       struct gl_renderbuffer *rb,
@@ -153,6 +213,13 @@ radeon_map_renderbuffer(struct gl_context *ctx,
       radeon_firevertices(rmesa);
    }
 
+#if defined(RADEON_R200)
+   if (rb->Format == MESA_FORMAT_S8_Z24 && !rrb->has_surface) {
+     radeon_map_renderbuffer_s8z24(ctx, rb, x, y, w, h,
+				   mode, out_map, out_stride);
+     return;
+   }
+#endif
    ret = radeon_bo_map(rrb->bo, !!(mode & GL_MAP_WRITE_BIT));
    assert(!ret);
 
@@ -174,6 +241,42 @@ radeon_map_renderbuffer(struct gl_context *ctx,
    *out_stride = flip_stride;
 }
 
+#if defined(RADEON_R200)
+static void
+radeon_unmap_renderbuffer_s8z24(struct gl_context *ctx,
+			  struct gl_renderbuffer *rb)
+{
+   struct radeon_context *const rmesa = RADEON_CONTEXT(ctx);
+   struct radeon_renderbuffer *rrb = radeon_renderbuffer(rb);
+
+   if (!rrb->map_buffer)
+     return;
+
+   if (rrb->map_mode & GL_MAP_WRITE_BIT) {
+       uint32_t *untiled_s8z24_map = rrb->map_buffer;
+       uint32_t *tiled_s8z24_map;
+       int y_flip = (rb->Name == 0) ? -1 : 1;
+       int y_bias = (rb->Name == 0) ? (rb->Height - 1) : 0;
+
+       radeon_bo_map(rrb->bo, 1);
+       
+       tiled_s8z24_map = rrb->bo->ptr;
+
+       for (uint32_t pix_y = 0; pix_y < rrb->map_h; pix_y++) {
+	   for (uint32_t pix_x = 0; pix_x < rrb->map_w; pix_x++) {
+	       uint32_t flipped_y = y_flip * (int32_t)(pix_y + rrb->map_y) + y_bias;
+	       uint32_t dst_offset = r200_depth_4byte(rrb, rrb->map_x + pix_x, flipped_y);
+	       uint32_t src_offset = pix_y * rrb->map_pitch + pix_x * rrb->cpp;
+	       tiled_s8z24_map[dst_offset/4] = untiled_s8z24_map[src_offset/4];
+	   }
+       }
+       radeon_bo_unmap(rrb->bo);
+   }
+   free(rrb->map_buffer);
+   rrb->map_buffer = NULL;
+}
+#endif
+
 static void
 radeon_unmap_renderbuffer(struct gl_context *ctx,
 			  struct gl_renderbuffer *rb)
@@ -182,6 +285,13 @@ radeon_unmap_renderbuffer(struct gl_context *ctx,
    struct radeon_renderbuffer *rrb = radeon_renderbuffer(rb);
    GLboolean ok;
 
+#ifdef RADEON_R200
+   if (rb->Format == MESA_FORMAT_S8_Z24 && !rrb->has_surface) {
+       radeon_unmap_renderbuffer_s8z24(ctx, rb);
+       return;
+   }
+#endif
+
    if (!rrb->map_bo) {
 	   if (rrb->bo)
 		   radeon_bo_unmap(rrb->bo);