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, 14 insertions, 45 deletions

diff --git a/src/gallium/drivers/nvfx/nvfx_draw.c b/src/gallium/drivers/nvfx/nvfx_draw.c
index 22cff370b77..331e28418ad 100644
--- a/src/gallium/drivers/nvfx/nvfx_draw.c
+++ b/src/gallium/drivers/nvfx/nvfx_draw.c
@@ -9,6 +9,7 @@
 #include "draw/draw_pipe.h"
 
 #include "nvfx_context.h"
+#include "nvfx_resource.h"
 
 /* Simple, but crappy, swtnl path, hopefully we wont need to hit this very
  * often at all.  Uses "quadro style" vertex submission + a fixed vertex
@@ -39,30 +40,21 @@ nvfx_render_vertex(struct nvfx_context *nvfx, const struct vertex_header *v)
 		unsigned idx = nvfx->swtnl.draw[i];
 		unsigned hw = nvfx->swtnl.hw[i];
 
+		WAIT_RING(chan, 5);
 		switch (nvfx->swtnl.emit[i]) {
 		case EMIT_OMIT:
 			break;
 		case EMIT_1F:
-			BEGIN_RING(chan, eng3d, NV34TCL_VTX_ATTR_1F(hw), 1);
-			OUT_RING  (chan, fui(v->data[idx][0]));
+			nvfx_emit_vtx_attr(chan, hw, v->data[idx], 1);
 			break;
 		case EMIT_2F:
-			BEGIN_RING(chan, eng3d, NV34TCL_VTX_ATTR_2F_X(hw), 2);
-			OUT_RING  (chan, fui(v->data[idx][0]));
-			OUT_RING  (chan, fui(v->data[idx][1]));
+			nvfx_emit_vtx_attr(chan, hw, v->data[idx], 2);
 			break;
 		case EMIT_3F:
-			BEGIN_RING(chan, eng3d, NV34TCL_VTX_ATTR_3F_X(hw), 3);
-			OUT_RING  (chan, fui(v->data[idx][0]));
-			OUT_RING  (chan, fui(v->data[idx][1]));
-			OUT_RING  (chan, fui(v->data[idx][2]));
+			nvfx_emit_vtx_attr(chan, hw, v->data[idx], 3);
 			break;
 		case EMIT_4F:
-			BEGIN_RING(chan, eng3d, NV34TCL_VTX_ATTR_4F_X(hw), 4);
-			OUT_RING  (chan, fui(v->data[idx][0]));
-			OUT_RING  (chan, fui(v->data[idx][1]));
-			OUT_RING  (chan, fui(v->data[idx][2]));
-			OUT_RING  (chan, fui(v->data[idx][3]));
+			nvfx_emit_vtx_attr(chan, hw, v->data[idx], 4);
 			break;
 		case 0xff:
 			BEGIN_RING(chan, eng3d, NV34TCL_VTX_ATTR_4F_X(hw), 4);
@@ -231,15 +223,9 @@ nvfx_draw_render_stage(struct nvfx_context *nvfx)
 }
 
 void
-nvfx_draw_elements_swtnl(struct pipe_context *pipe,
-			 struct pipe_resource *idxbuf,
-			 unsigned idxbuf_size, int idxbuf_bias,
-			 unsigned mode, unsigned start, unsigned count)
+nvfx_draw_vbo_swtnl(struct pipe_context *pipe, const struct pipe_draw_info* info)
 {
 	struct nvfx_context *nvfx = nvfx_context(pipe);
-	struct pipe_transfer *vb_transfer[PIPE_MAX_ATTRIBS];
-	struct pipe_transfer *ib_transfer = NULL;
-	struct pipe_transfer *cb_transfer = NULL;
 	unsigned i;
 	void *map;
 
@@ -247,18 +233,15 @@ nvfx_draw_elements_swtnl(struct pipe_context *pipe,
 		return;
 	nvfx_state_emit(nvfx);
 
+	/* these must be passed without adding the offsets */
 	for (i = 0; i < nvfx->vtxbuf_nr; i++) {
-		map = pipe_buffer_map(pipe, nvfx->vtxbuf[i].buffer,
-                                      PIPE_TRANSFER_READ,
-				      &vb_transfer[i]);
+		map = nvfx_buffer(nvfx->vtxbuf[i].buffer)->data;
 		draw_set_mapped_vertex_buffer(nvfx->draw, i, map);
 	}
 
-	if (idxbuf) {
-		map = pipe_buffer_map(pipe, idxbuf,
-				      PIPE_TRANSFER_READ,
-				      &ib_transfer);
-		draw_set_mapped_element_buffer(nvfx->draw, idxbuf_size, idxbuf_bias, map);
+	if (info->indexed) {
+		map = nvfx_buffer(nvfx->idxbuf.buffer)->data + nvfx->idxbuf.offset;
+		draw_set_mapped_element_buffer_range(nvfx->draw, nvfx->idxbuf.index_size, info->index_bias, info->min_index, info->max_index, map);
 	} else {
 		draw_set_mapped_element_buffer(nvfx->draw, 0, 0, NULL);
 	}
@@ -266,28 +249,14 @@ nvfx_draw_elements_swtnl(struct pipe_context *pipe,
 	if (nvfx->constbuf[PIPE_SHADER_VERTEX]) {
 		const unsigned nr = nvfx->constbuf_nr[PIPE_SHADER_VERTEX];
 
-		map = pipe_buffer_map(pipe,
-				      nvfx->constbuf[PIPE_SHADER_VERTEX],
-				      PIPE_TRANSFER_READ,
-				      &cb_transfer);
+		map = nvfx_buffer(nvfx->constbuf[PIPE_SHADER_VERTEX])->data;
 		draw_set_mapped_constant_buffer(nvfx->draw, PIPE_SHADER_VERTEX, 0,
                                                 map, nr);
 	}
 
-	draw_arrays(nvfx->draw, mode, start, count);
-
-	for (i = 0; i < nvfx->vtxbuf_nr; i++)
-		pipe_buffer_unmap(pipe, nvfx->vtxbuf[i].buffer, vb_transfer[i]);
-
-	if (idxbuf)
-		pipe_buffer_unmap(pipe, idxbuf, ib_transfer);
-
-	if (nvfx->constbuf[PIPE_SHADER_VERTEX])
-		pipe_buffer_unmap(pipe, nvfx->constbuf[PIPE_SHADER_VERTEX],
-				  cb_transfer);
+	draw_arrays_instanced(nvfx->draw, info->mode, info->start, info->count, info->start_instance, info->instance_count);
 
 	draw_flush(nvfx->draw);
-	pipe->flush(pipe, 0, NULL);
 }
 
 static INLINE void