nvfx: rewrite draw code and buffer code

This is a full rewrite of the drawing and buffer management logic.

It offers a lot of improvements:
1. A copy of buffers is now always kept in system memory. This is
   necessary to allow software processing of them, which is necessary
   or improves performance in many cases.
2. Support for pushing vertices on the FIFO, with index lookup if necessary.
3. "Smart" draw code that tries to intelligently choose the cheapest
  way to draw something: whether to use inline vertices or hardware
  vertex buffer, and whether to use hardware index buffers
4. Support for all vertex formats supported by the hardware
5. Usage of translate to push vertices, supporting all formats that are
   sensible to use as vertex formats
6. Support for base vertex
7. Usage of Ben Skeggs' primitive splitter originally for nv50, allowing
   correct splitting of line loops, triangle fans, etc.
8. Support for instancing
9. Precomputation using the vertex elements CSO

Thanks to Ben Skeggs for his primitive splitter originally for nv50.

Thanks to Christoph Bumiller for his nv50 push code, that was the basis
of this work, even though I changed his code dramatically, in particular
to replace his ad-hoc vertex data emitter with translate.

The changes could also go into nv50 too, but there are substantial
differences due to the additional nv50 hardware features.

1 files changed, 69 insertions, 22 deletions

diff --git a/src/gallium/drivers/nvfx/nvfx_resource.h b/src/gallium/drivers/nvfx/nvfx_resource.h
index ff86f6d9cb..583be4de2a 100644
--- a/src/gallium/drivers/nvfx/nvfx_resource.h
+++ b/src/gallium/drivers/nvfx/nvfx_resource.h
@@ -17,8 +17,23 @@ struct nvfx_resource {
 	struct nouveau_bo *bo;
 };
 
+static INLINE
+struct nvfx_resource *nvfx_resource(struct pipe_resource *resource)
+{
+	return (struct nvfx_resource *)resource;
+}
+
 #define NVFX_RESOURCE_FLAG_LINEAR (PIPE_RESOURCE_FLAG_DRV_PRIV << 0)
+#define NVFX_RESOURCE_FLAG_USER (PIPE_RESOURCE_FLAG_DRV_PRIV << 1)
+
+/* is resource mapped into the GPU's address space (i.e. VRAM or GART) ? */
+static INLINE boolean
+nvfx_resource_mapped_by_gpu(struct pipe_resource *resource)
+{
+   return nvfx_resource(resource)->bo->handle;
+}
 
+/* is resource in VRAM? */
 static inline int
 nvfx_resource_on_gpu(struct pipe_resource* pr)
 {
@@ -63,12 +78,6 @@ struct nvfx_surface {
 	struct nvfx_miptree* temp;
 };
 
-static INLINE 
-struct nvfx_resource *nvfx_resource(struct pipe_resource *resource)
-{
-	return (struct nvfx_resource *)resource;
-}
-
 static INLINE struct nouveau_bo *
 nvfx_surface_buffer(struct pipe_surface *surf)
 {
@@ -106,22 +115,6 @@ nvfx_miptree_from_handle(struct pipe_screen *pscreen,
 			 const struct pipe_resource *template,
 			 struct winsys_handle *whandle);
 
-struct pipe_resource *
-nvfx_buffer_create(struct pipe_screen *pscreen,
-		   const struct pipe_resource *template);
-
-void
-nvfx_buffer_destroy(struct pipe_screen *pscreen,
-                    struct pipe_resource *presource);
-
-struct pipe_resource *
-nvfx_user_buffer_create(struct pipe_screen *screen,
-			void *ptr,
-			unsigned bytes,
-			unsigned usage);
-
-
-
 void
 nvfx_miptree_surface_del(struct pipe_surface *ps);
 
@@ -173,4 +166,58 @@ nvfx_surface_create_temp(struct pipe_context* pipe, struct pipe_surface* surf);
 void
 nvfx_surface_flush(struct pipe_context* pipe, struct pipe_surface* surf);
 
+struct nvfx_buffer
+{
+	struct nvfx_resource base;
+	uint8_t* data;
+	unsigned size;
+
+	/* the range of data not yet uploaded to the GPU bo */
+	unsigned dirty_begin;
+	unsigned dirty_end;
+
+	/* whether all transfers were unsynchronized */
+	boolean dirty_unsynchronized;
+
+	/* whether it would have been profitable to upload
+	 * the latest updated data to the GPU immediately */
+	boolean last_update_static;
+
+	/* how many bytes we need to draw before we deem
+	 * the buffer to be static
+	 */
+	long long bytes_to_draw_until_static;
+};
+
+static inline struct nvfx_buffer* nvfx_buffer(struct pipe_resource* pr)
+{
+	return (struct nvfx_buffer*)pr;
+}
+
+/* this is an heuristic to determine whether we are better off uploading the
+ * buffer to the GPU, or just continuing pushing it on the FIFO
+ */
+static inline boolean nvfx_buffer_seems_static(struct nvfx_buffer* buffer)
+{
+	return buffer->last_update_static
+		|| buffer->bytes_to_draw_until_static < 0;
+}
+
+struct pipe_resource *
+nvfx_buffer_create(struct pipe_screen *pscreen,
+		   const struct pipe_resource *template);
+
+void
+nvfx_buffer_destroy(struct pipe_screen *pscreen,
+                    struct pipe_resource *presource);
+
+struct pipe_resource *
+nvfx_user_buffer_create(struct pipe_screen *screen,
+			void *ptr,
+			unsigned bytes,
+			unsigned usage);
+
+void
+nvfx_buffer_upload(struct nvfx_buffer* buffer);
+
 #endif