summaryrefslogtreecommitdiffstats
path: root/src/gallium/auxiliary/draw
diff options
context:
space:
mode:
Diffstat (limited to 'src/gallium/auxiliary/draw')
-rw-r--r--src/gallium/auxiliary/draw/Makefile4
-rw-r--r--src/gallium/auxiliary/draw/draw_context.c7
-rw-r--r--src/gallium/auxiliary/draw/draw_pipe_aaline.c17
-rw-r--r--src/gallium/auxiliary/draw/draw_pipe_aapoint.c1
-rw-r--r--src/gallium/auxiliary/draw/draw_pipe_pstipple.c20
-rw-r--r--src/gallium/auxiliary/draw/draw_pipe_vbuf.c95
-rw-r--r--src/gallium/auxiliary/draw/draw_pt.c2
-rw-r--r--src/gallium/auxiliary/draw/draw_pt.h4
-rw-r--r--src/gallium/auxiliary/draw/draw_pt_emit.c67
-rw-r--r--src/gallium/auxiliary/draw/draw_pt_fetch_emit.c69
-rw-r--r--src/gallium/auxiliary/draw/draw_pt_fetch_shade_emit.c107
-rw-r--r--src/gallium/auxiliary/draw/draw_pt_fetch_shade_pipeline.c23
-rw-r--r--src/gallium/auxiliary/draw/draw_pt_post_vs.c4
-rw-r--r--src/gallium/auxiliary/draw/draw_vbuf.h16
-rw-r--r--src/gallium/auxiliary/draw/draw_vs_aos.c55
-rw-r--r--src/gallium/auxiliary/draw/draw_vs_aos.h2
-rw-r--r--src/gallium/auxiliary/draw/draw_vs_aos_io.c4
17 files changed, 255 insertions, 242 deletions
diff --git a/src/gallium/auxiliary/draw/Makefile b/src/gallium/auxiliary/draw/Makefile
index bdbf5a08ede..5041dcc072b 100644
--- a/src/gallium/auxiliary/draw/Makefile
+++ b/src/gallium/auxiliary/draw/Makefile
@@ -43,8 +43,4 @@ C_SOURCES = \
draw_vs_ppc.c \
draw_vs_sse.c
-
include ../../Makefile.template
-
-symlinks:
-
diff --git a/src/gallium/auxiliary/draw/draw_context.c b/src/gallium/auxiliary/draw/draw_context.c
index 7bd4a2e2218..a4f1fcddc1a 100644
--- a/src/gallium/auxiliary/draw/draw_context.c
+++ b/src/gallium/auxiliary/draw/draw_context.c
@@ -127,7 +127,7 @@ void draw_set_rasterizer_state( struct draw_context *draw,
draw->rasterizer = raster;
draw->bypass_clipping =
- ((draw->rasterizer && draw->rasterizer->bypass_clipping) ||
+ ((draw->rasterizer && draw->rasterizer->bypass_vs_clip_and_viewport) ||
draw->driver.bypass_clipping);
}
@@ -138,8 +138,9 @@ void draw_set_driver_clipping( struct draw_context *draw,
draw_do_flush( draw, DRAW_FLUSH_STATE_CHANGE );
draw->driver.bypass_clipping = bypass_clipping;
- draw->bypass_clipping = (draw->rasterizer->bypass_clipping ||
- draw->driver.bypass_clipping);
+ draw->bypass_clipping =
+ ((draw->rasterizer && draw->rasterizer->bypass_vs_clip_and_viewport) ||
+ draw->driver.bypass_clipping);
}
diff --git a/src/gallium/auxiliary/draw/draw_pipe_aaline.c b/src/gallium/auxiliary/draw/draw_pipe_aaline.c
index 20841bb5d69..ca69f0b95e5 100644
--- a/src/gallium/auxiliary/draw/draw_pipe_aaline.c
+++ b/src/gallium/auxiliary/draw/draw_pipe_aaline.c
@@ -32,7 +32,6 @@
*/
-#include "pipe/p_inlines.h"
#include "pipe/p_context.h"
#include "pipe/p_defines.h"
#include "pipe/p_shader_tokens.h"
@@ -410,7 +409,7 @@ aaline_create_texture(struct aaline_stage *aaline)
* texels which are zero. Special case the 1x1 and 2x2 levels.
*/
for (level = 0; level <= MAX_TEXTURE_LEVEL; level++) {
- struct pipe_surface *surface;
+ struct pipe_transfer *transfer;
const uint size = aaline->texture->width[level];
ubyte *data;
uint i, j;
@@ -419,9 +418,9 @@ aaline_create_texture(struct aaline_stage *aaline)
/* This texture is new, no need to flush.
*/
- surface = screen->get_tex_surface(screen, aaline->texture, 0, level, 0,
- PIPE_BUFFER_USAGE_CPU_WRITE);
- data = screen->surface_map(screen, surface, PIPE_BUFFER_USAGE_CPU_WRITE);
+ transfer = screen->get_tex_transfer(screen, aaline->texture, 0, level, 0,
+ PIPE_TRANSFER_WRITE, 0, 0, size, size);
+ data = screen->transfer_map(screen, transfer);
if (data == NULL)
return FALSE;
@@ -440,13 +439,13 @@ aaline_create_texture(struct aaline_stage *aaline)
else {
d = 255;
}
- data[i * surface->stride + j] = d;
+ data[i * transfer->stride + j] = d;
}
}
/* unmap */
- screen->surface_unmap(screen, surface);
- screen->tex_surface_release(screen, &surface);
+ screen->transfer_unmap(screen, transfer);
+ screen->tex_transfer_destroy(transfer);
}
return TRUE;
}
@@ -728,7 +727,7 @@ aaline_destroy(struct draw_stage *stage)
aaline->pipe->delete_sampler_state(aaline->pipe, aaline->sampler_cso);
if (aaline->texture)
- pipe_texture_release(&aaline->texture);
+ pipe_texture_reference(&aaline->texture, NULL);
draw_free_temp_verts( stage );
diff --git a/src/gallium/auxiliary/draw/draw_pipe_aapoint.c b/src/gallium/auxiliary/draw/draw_pipe_aapoint.c
index 2c1cacbdb47..5008086cf26 100644
--- a/src/gallium/auxiliary/draw/draw_pipe_aapoint.c
+++ b/src/gallium/auxiliary/draw/draw_pipe_aapoint.c
@@ -38,7 +38,6 @@
*/
-#include "pipe/p_inlines.h"
#include "pipe/p_context.h"
#include "pipe/p_defines.h"
#include "pipe/p_shader_tokens.h"
diff --git a/src/gallium/auxiliary/draw/draw_pipe_pstipple.c b/src/gallium/auxiliary/draw/draw_pipe_pstipple.c
index a0f9716dac2..04e59152c5c 100644
--- a/src/gallium/auxiliary/draw/draw_pipe_pstipple.c
+++ b/src/gallium/auxiliary/draw/draw_pipe_pstipple.c
@@ -34,7 +34,6 @@
*/
-#include "pipe/p_inlines.h"
#include "pipe/p_context.h"
#include "pipe/p_defines.h"
#include "pipe/p_shader_tokens.h"
@@ -372,7 +371,7 @@ pstip_update_texture(struct pstip_stage *pstip)
static const uint bit31 = 1 << 31;
struct pipe_context *pipe = pstip->pipe;
struct pipe_screen *screen = pipe->screen;
- struct pipe_surface *surface;
+ struct pipe_transfer *transfer;
const uint *stipple = pstip->state.stipple->stipple;
uint i, j;
ubyte *data;
@@ -381,10 +380,9 @@ pstip_update_texture(struct pstip_stage *pstip)
*/
pipe->flush( pipe, PIPE_FLUSH_TEXTURE_CACHE, NULL );
- surface = screen->get_tex_surface(screen, pstip->texture, 0, 0, 0,
- PIPE_BUFFER_USAGE_CPU_WRITE);
- data = screen->surface_map(screen, surface,
- PIPE_BUFFER_USAGE_CPU_WRITE);
+ transfer = screen->get_tex_transfer(screen, pstip->texture, 0, 0, 0,
+ PIPE_TRANSFER_WRITE, 0, 0, 32, 32);
+ data = screen->transfer_map(screen, transfer);
/*
* Load alpha texture.
@@ -396,18 +394,18 @@ pstip_update_texture(struct pstip_stage *pstip)
for (j = 0; j < 32; j++) {
if (stipple[i] & (bit31 >> j)) {
/* fragment "on" */
- data[i * surface->stride + j] = 0;
+ data[i * transfer->stride + j] = 0;
}
else {
/* fragment "off" */
- data[i * surface->stride + j] = 255;
+ data[i * transfer->stride + j] = 255;
}
}
}
/* unmap */
- screen->surface_unmap(screen, surface);
- screen->tex_surface_release(screen, &surface);
+ screen->transfer_unmap(screen, transfer);
+ screen->tex_transfer_destroy(transfer);
}
@@ -573,7 +571,7 @@ pstip_destroy(struct draw_stage *stage)
pstip->pipe->delete_sampler_state(pstip->pipe, pstip->sampler_cso);
- pipe_texture_release(&pstip->texture);
+ pipe_texture_reference(&pstip->texture, NULL);
draw_free_temp_verts( stage );
FREE( stage );
diff --git a/src/gallium/auxiliary/draw/draw_pipe_vbuf.c b/src/gallium/auxiliary/draw/draw_pipe_vbuf.c
index 9153bc2f865..12325d30d61 100644
--- a/src/gallium/auxiliary/draw/draw_pipe_vbuf.c
+++ b/src/gallium/auxiliary/draw/draw_pipe_vbuf.c
@@ -29,12 +29,12 @@
* \file
* Vertex buffer drawing stage.
*
- * \author Jos� Fonseca <[email protected]>
+ * \author Jose Fonseca <[email protected]>
* \author Keith Whitwell <[email protected]>
*/
-#include "pipe/p_debug.h"
+#include "util/u_debug.h"
#include "util/u_math.h"
#include "util/u_memory.h"
@@ -93,7 +93,6 @@ vbuf_stage( struct draw_stage *stage )
}
-static void vbuf_flush_indices( struct vbuf_stage *vbuf );
static void vbuf_flush_vertices( struct vbuf_stage *vbuf );
static void vbuf_alloc_vertices( struct vbuf_stage *vbuf );
@@ -109,13 +108,12 @@ overflow( void *map, void *ptr, unsigned bytes, unsigned bufsz )
static INLINE void
check_space( struct vbuf_stage *vbuf, unsigned nr )
{
- if (vbuf->nr_vertices + nr > vbuf->max_vertices ) {
- vbuf_flush_vertices(vbuf);
- vbuf_alloc_vertices(vbuf);
+ if (vbuf->nr_vertices + nr > vbuf->max_vertices ||
+ vbuf->nr_indices + nr > vbuf->max_indices)
+ {
+ vbuf_flush_vertices( vbuf );
+ vbuf_alloc_vertices( vbuf );
}
-
- if (vbuf->nr_indices + nr > vbuf->max_indices )
- vbuf_flush_indices(vbuf);
}
@@ -202,7 +200,7 @@ vbuf_point( struct draw_stage *stage,
* will be flushed if needed and a new one allocated.
*/
static void
-vbuf_set_prim( struct vbuf_stage *vbuf, uint prim )
+vbuf_start_prim( struct vbuf_stage *vbuf, uint prim )
{
struct translate_key hw_key;
unsigned dst_offset;
@@ -217,11 +215,7 @@ vbuf_set_prim( struct vbuf_stage *vbuf, uint prim )
* state change.
*/
vbuf->vinfo = vbuf->render->get_vertex_info(vbuf->render);
-
- if (vbuf->vertex_size != vbuf->vinfo->size * sizeof(float)) {
- vbuf_flush_vertices(vbuf);
- vbuf->vertex_size = vbuf->vinfo->size * sizeof(float);
- }
+ vbuf->vertex_size = vbuf->vinfo->size * sizeof(float);
/* Translate from pipeline vertices to hw vertices.
*/
@@ -294,8 +288,8 @@ vbuf_set_prim( struct vbuf_stage *vbuf, uint prim )
/* Allocate new buffer?
*/
- if (!vbuf->vertices)
- vbuf_alloc_vertices(vbuf);
+ assert(vbuf->vertices == NULL);
+ vbuf_alloc_vertices(vbuf);
}
@@ -305,9 +299,9 @@ vbuf_first_tri( struct draw_stage *stage,
{
struct vbuf_stage *vbuf = vbuf_stage( stage );
- vbuf_flush_indices( vbuf );
+ vbuf_flush_vertices( vbuf );
+ vbuf_start_prim(vbuf, PIPE_PRIM_TRIANGLES);
stage->tri = vbuf_tri;
- vbuf_set_prim(vbuf, PIPE_PRIM_TRIANGLES);
stage->tri( stage, prim );
}
@@ -318,9 +312,9 @@ vbuf_first_line( struct draw_stage *stage,
{
struct vbuf_stage *vbuf = vbuf_stage( stage );
- vbuf_flush_indices( vbuf );
+ vbuf_flush_vertices( vbuf );
+ vbuf_start_prim(vbuf, PIPE_PRIM_LINES);
stage->line = vbuf_line;
- vbuf_set_prim(vbuf, PIPE_PRIM_LINES);
stage->line( stage, prim );
}
@@ -331,53 +325,42 @@ vbuf_first_point( struct draw_stage *stage,
{
struct vbuf_stage *vbuf = vbuf_stage( stage );
- vbuf_flush_indices( vbuf );
+ vbuf_flush_vertices(vbuf);
+ vbuf_start_prim(vbuf, PIPE_PRIM_POINTS);
stage->point = vbuf_point;
- vbuf_set_prim(vbuf, PIPE_PRIM_POINTS);
stage->point( stage, prim );
}
-static void
-vbuf_flush_indices( struct vbuf_stage *vbuf )
-{
- if(!vbuf->nr_indices)
- return;
-
- assert((uint) (vbuf->vertex_ptr - vbuf->vertices) ==
- vbuf->nr_vertices * vbuf->vertex_size / sizeof(unsigned));
-
- vbuf->render->draw(vbuf->render, vbuf->indices, vbuf->nr_indices);
-
- vbuf->nr_indices = 0;
-}
-
/**
* Flush existing vertex buffer and allocate a new one.
- *
- * XXX: We separate flush-on-index-full and flush-on-vb-full, but may
- * raise issues uploading vertices if the hardware wants to flush when
- * we flush.
*/
static void
vbuf_flush_vertices( struct vbuf_stage *vbuf )
{
- if(vbuf->vertices) {
- vbuf_flush_indices(vbuf);
-
+ if(vbuf->vertices) {
+
+ vbuf->render->unmap_vertices( vbuf->render, 0, vbuf->nr_vertices - 1 );
+
+ if (vbuf->nr_indices)
+ {
+ vbuf->render->draw(vbuf->render,
+ vbuf->indices,
+ vbuf->nr_indices );
+
+ vbuf->nr_indices = 0;
+ }
+
/* Reset temporary vertices ids */
if(vbuf->nr_vertices)
draw_reset_vertex_ids( vbuf->stage.draw );
/* Free the vertex buffer */
- vbuf->render->release_vertices(vbuf->render,
- vbuf->vertices,
- vbuf->vertex_size,
- vbuf->nr_vertices);
+ vbuf->render->release_vertices( vbuf->render );
+
vbuf->max_vertices = vbuf->nr_vertices = 0;
vbuf->vertex_ptr = vbuf->vertices = NULL;
-
}
}
@@ -402,9 +385,12 @@ vbuf_alloc_vertices( struct vbuf_stage *vbuf )
* and it will flush itself if necessary to do so. If this does
* fail, we are basically without usable hardware.
*/
- vbuf->vertices = (uint *) vbuf->render->allocate_vertices(vbuf->render,
- (ushort) vbuf->vertex_size,
- (ushort) vbuf->max_vertices);
+ vbuf->render->allocate_vertices(vbuf->render,
+ (ushort) vbuf->vertex_size,
+ (ushort) vbuf->max_vertices);
+
+ vbuf->vertices = (uint *) vbuf->render->map_vertices( vbuf->render );
+
vbuf->vertex_ptr = vbuf->vertices;
}
@@ -415,14 +401,11 @@ vbuf_flush( struct draw_stage *stage, unsigned flags )
{
struct vbuf_stage *vbuf = vbuf_stage( stage );
- vbuf_flush_indices( vbuf );
+ vbuf_flush_vertices( vbuf );
stage->point = vbuf_first_point;
stage->line = vbuf_first_line;
stage->tri = vbuf_first_tri;
-
- if (flags & DRAW_FLUSH_BACKEND)
- vbuf_flush_vertices( vbuf );
}
diff --git a/src/gallium/auxiliary/draw/draw_pt.c b/src/gallium/auxiliary/draw/draw_pt.c
index 4e5ffa09305..9ea0cbe5990 100644
--- a/src/gallium/auxiliary/draw/draw_pt.c
+++ b/src/gallium/auxiliary/draw/draw_pt.c
@@ -87,7 +87,7 @@ draw_pt_arrays(struct draw_context *draw,
opt |= PT_CLIPTEST;
}
- if (!draw->rasterizer->bypass_vs) {
+ if (!draw->rasterizer->bypass_vs_clip_and_viewport) {
opt |= PT_SHADE;
}
}
diff --git a/src/gallium/auxiliary/draw/draw_pt.h b/src/gallium/auxiliary/draw/draw_pt.h
index c02f229110a..6f3e1e0289b 100644
--- a/src/gallium/auxiliary/draw/draw_pt.h
+++ b/src/gallium/auxiliary/draw/draw_pt.h
@@ -173,9 +173,7 @@ void draw_pt_emit( struct pt_emit *emit,
void draw_pt_emit_linear( struct pt_emit *emit,
const float (*vertex_data)[4],
- unsigned vertex_count,
unsigned stride,
- unsigned start,
unsigned count );
void draw_pt_emit_destroy( struct pt_emit *emit );
@@ -217,7 +215,7 @@ boolean draw_pt_post_vs_run( struct pt_post_vs *pvs,
void draw_pt_post_vs_prepare( struct pt_post_vs *pvs,
boolean bypass_clipping,
- boolean identity_viewport,
+ boolean bypass_viewport,
boolean opengl );
struct pt_post_vs *draw_pt_post_vs_create( struct draw_context *draw );
diff --git a/src/gallium/auxiliary/draw/draw_pt_emit.c b/src/gallium/auxiliary/draw/draw_pt_emit.c
index 232dfdaed2d..064e16c295c 100644
--- a/src/gallium/auxiliary/draw/draw_pt_emit.c
+++ b/src/gallium/auxiliary/draw/draw_pt_emit.c
@@ -165,6 +165,9 @@ void draw_pt_emit( struct pt_emit *emit,
*/
draw_do_flush( draw, DRAW_FLUSH_BACKEND );
+ if (vertex_count == 0)
+ return;
+
if (vertex_count >= UNDEFINED_VERTEX_ID) {
assert(0);
return;
@@ -178,9 +181,11 @@ void draw_pt_emit( struct pt_emit *emit,
return;
}
- hw_verts = render->allocate_vertices(render,
- (ushort)translate->key.output_stride,
- (ushort)vertex_count);
+ render->allocate_vertices(render,
+ (ushort)translate->key.output_stride,
+ (ushort)vertex_count);
+
+ hw_verts = render->map_vertices( render );
if (!hw_verts) {
assert(0);
return;
@@ -201,22 +206,21 @@ void draw_pt_emit( struct pt_emit *emit,
vertex_count,
hw_verts );
+ render->unmap_vertices( render,
+ 0,
+ vertex_count - 1 );
+
render->draw(render,
elts,
count);
- render->release_vertices(render,
- hw_verts,
- translate->key.output_stride,
- vertex_count);
+ render->release_vertices(render);
}
void draw_pt_emit_linear(struct pt_emit *emit,
const float (*vertex_data)[4],
- unsigned vertex_count,
unsigned stride,
- unsigned start,
unsigned count)
{
struct draw_context *draw = emit->draw;
@@ -231,26 +235,23 @@ void draw_pt_emit_linear(struct pt_emit *emit,
*/
draw_do_flush( draw, DRAW_FLUSH_BACKEND );
- if (count >= UNDEFINED_VERTEX_ID) {
- assert(0);
- return;
- }
+ if (count >= UNDEFINED_VERTEX_ID)
+ goto fail;
/* XXX: and work out some way to coordinate the render primitive
* between vbuf.c and here...
*/
- if (!draw->render->set_primitive(draw->render, emit->prim)) {
- assert(0);
- return;
- }
+ if (!draw->render->set_primitive(draw->render, emit->prim))
+ goto fail;
- hw_verts = render->allocate_vertices(render,
- (ushort)translate->key.output_stride,
- (ushort)count);
- if (!hw_verts) {
- assert(0);
- return;
- }
+ if (!render->allocate_vertices(render,
+ (ushort)translate->key.output_stride,
+ (ushort)count))
+ goto fail;
+
+ hw_verts = render->map_vertices( render );
+ if (!hw_verts)
+ goto fail;
translate->set_buffer(translate, 0,
vertex_data, stride);
@@ -261,12 +262,12 @@ void draw_pt_emit_linear(struct pt_emit *emit,
translate->run(translate,
0,
- vertex_count,
+ count,
hw_verts);
if (0) {
unsigned i;
- for (i = 0; i < vertex_count; i++) {
+ for (i = 0; i < count; i++) {
debug_printf("\n\n%s vertex %d:\n", __FUNCTION__, i);
draw_dump_emitted_vertex( emit->vinfo,
(const uint8_t *)hw_verts +
@@ -274,13 +275,17 @@ void draw_pt_emit_linear(struct pt_emit *emit,
}
}
+ render->unmap_vertices( render, 0, count - 1 );
+
+ render->draw_arrays(render, 0, count);
+
+ render->release_vertices(render);
- render->draw_arrays(render, start, count);
+ return;
- render->release_vertices(render,
- hw_verts,
- translate->key.output_stride,
- vertex_count);
+fail:
+ assert(0);
+ return;
}
struct pt_emit *draw_pt_emit_create( struct draw_context *draw )
diff --git a/src/gallium/auxiliary/draw/draw_pt_fetch_emit.c b/src/gallium/auxiliary/draw/draw_pt_fetch_emit.c
index dcb7744b177..6b7d02a19bc 100644
--- a/src/gallium/auxiliary/draw/draw_pt_fetch_emit.c
+++ b/src/gallium/auxiliary/draw/draw_pt_fetch_emit.c
@@ -234,9 +234,11 @@ static void fetch_emit_run( struct draw_pt_middle_end *middle,
return;
}
- hw_verts = draw->render->allocate_vertices( draw->render,
- (ushort)feme->translate->key.output_stride,
- (ushort)fetch_count );
+ draw->render->allocate_vertices( draw->render,
+ (ushort)feme->translate->key.output_stride,
+ (ushort)fetch_count );
+
+ hw_verts = draw->render->map_vertices( draw->render );
if (!hw_verts) {
assert(0);
return;
@@ -259,6 +261,10 @@ static void fetch_emit_run( struct draw_pt_middle_end *middle,
}
}
+ draw->render->unmap_vertices( draw->render,
+ 0,
+ (ushort)(fetch_count - 1) );
+
/* XXX: Draw arrays path to avoid re-emitting index list again and
* again.
*/
@@ -268,10 +274,7 @@ static void fetch_emit_run( struct draw_pt_middle_end *middle,
/* Done -- that was easy, wasn't it:
*/
- draw->render->release_vertices( draw->render,
- hw_verts,
- feme->translate->key.output_stride,
- fetch_count );
+ draw->render->release_vertices( draw->render );
}
@@ -288,18 +291,17 @@ static void fetch_emit_run_linear( struct draw_pt_middle_end *middle,
*/
draw_do_flush( draw, DRAW_FLUSH_BACKEND );
- if (count >= UNDEFINED_VERTEX_ID) {
- assert(0);
- return;
- }
+ if (count >= UNDEFINED_VERTEX_ID)
+ goto fail;
- hw_verts = draw->render->allocate_vertices( draw->render,
- (ushort)feme->translate->key.output_stride,
- (ushort)count );
- if (!hw_verts) {
- assert(0);
- return;
- }
+ if (!draw->render->allocate_vertices( draw->render,
+ (ushort)feme->translate->key.output_stride,
+ (ushort)count ))
+ goto fail;
+
+ hw_verts = draw->render->map_vertices( draw->render );
+ if (!hw_verts)
+ goto fail;
/* Single routine to fetch vertices and emit HW verts.
*/
@@ -317,20 +319,21 @@ static void fetch_emit_run_linear( struct draw_pt_middle_end *middle,
}
}
+ draw->render->unmap_vertices( draw->render, 0, count - 1 );
+
/* XXX: Draw arrays path to avoid re-emitting index list again and
* again.
*/
- draw->render->draw_arrays( draw->render,
- 0, /*start*/
- count );
+ draw->render->draw_arrays( draw->render, 0, count );
/* Done -- that was easy, wasn't it:
*/
- draw->render->release_vertices( draw->render,
- hw_verts,
- feme->translate->key.output_stride,
- count );
+ draw->render->release_vertices( draw->render );
+ return;
+fail:
+ assert(0);
+ return;
}
@@ -351,9 +354,12 @@ static boolean fetch_emit_run_linear_elts( struct draw_pt_middle_end *middle,
if (count >= UNDEFINED_VERTEX_ID)
return FALSE;
- hw_verts = draw->render->allocate_vertices( draw->render,
- (ushort)feme->translate->key.output_stride,
- (ushort)count );
+ if (!draw->render->allocate_vertices( draw->render,
+ (ushort)feme->translate->key.output_stride,
+ (ushort)count ))
+ return FALSE;
+
+ hw_verts = draw->render->map_vertices( draw->render );
if (!hw_verts)
return FALSE;
@@ -364,6 +370,8 @@ static boolean fetch_emit_run_linear_elts( struct draw_pt_middle_end *middle,
count,
hw_verts );
+ draw->render->unmap_vertices( draw->render, 0, (ushort)(count - 1) );
+
/* XXX: Draw arrays path to avoid re-emitting index list again and
* again.
*/
@@ -373,10 +381,7 @@ static boolean fetch_emit_run_linear_elts( struct draw_pt_middle_end *middle,
/* Done -- that was easy, wasn't it:
*/
- draw->render->release_vertices( draw->render,
- hw_verts,
- feme->translate->key.output_stride,
- count );
+ draw->render->release_vertices( draw->render );
return TRUE;
}
diff --git a/src/gallium/auxiliary/draw/draw_pt_fetch_shade_emit.c b/src/gallium/auxiliary/draw/draw_pt_fetch_shade_emit.c
index 84ffe3296af..44147aed9be 100644
--- a/src/gallium/auxiliary/draw/draw_pt_fetch_shade_emit.c
+++ b/src/gallium/auxiliary/draw/draw_pt_fetch_shade_emit.c
@@ -101,7 +101,8 @@ static void fse_prepare( struct draw_pt_middle_end *middle,
fse->key.nr_elements = MAX2(fse->key.nr_outputs, /* outputs - translate to hw format */
fse->key.nr_inputs); /* inputs - fetch from api format */
- fse->key.viewport = !draw->identity_viewport;
+ fse->key.viewport = (!draw->rasterizer->bypass_vs_clip_and_viewport &&
+ !draw->identity_viewport);
fse->key.clip = !draw->bypass_clipping;
fse->key.const_vbuffers = 0;
@@ -234,19 +235,17 @@ static void fse_run_linear( struct draw_pt_middle_end *middle,
*/
draw_do_flush( draw, DRAW_FLUSH_BACKEND );
- if (count >= UNDEFINED_VERTEX_ID) {
- assert(0);
- return;
- }
+ if (count >= UNDEFINED_VERTEX_ID)
+ goto fail;
- hw_verts = draw->render->allocate_vertices( draw->render,
- (ushort)fse->key.output_stride,
- (ushort)count );
+ if (!draw->render->allocate_vertices( draw->render,
+ (ushort)fse->key.output_stride,
+ (ushort)count ))
+ goto fail;
- if (!hw_verts) {
- assert(0);
- return;
- }
+ hw_verts = draw->render->map_vertices( draw->render );
+ if (!hw_verts)
+ goto fail;
/* Single routine to fetch vertices, run shader and emit HW verts.
* Clipping is done elsewhere -- either by the API or on hardware,
@@ -256,13 +255,7 @@ static void fse_run_linear( struct draw_pt_middle_end *middle,
start, count,
hw_verts );
- /* Draw arrays path to avoid re-emitting index list again and
- * again.
- */
- draw->render->draw_arrays( draw->render,
- 0,
- count );
-
+
if (0) {
unsigned i;
for (i = 0; i < count; i++) {
@@ -274,12 +267,24 @@ static void fse_run_linear( struct draw_pt_middle_end *middle,
(const uint8_t *)hw_verts + fse->key.output_stride * i );
}
}
+
+ draw->render->unmap_vertices( draw->render, 0, (ushort)(count - 1) );
+ /* Draw arrays path to avoid re-emitting index list again and
+ * again.
+ */
+ draw->render->draw_arrays( draw->render,
+ 0,
+ count );
+
+
+ draw->render->release_vertices( draw->render );
+
+ return;
- draw->render->release_vertices( draw->render,
- hw_verts,
- fse->key.output_stride,
- count );
+fail:
+ assert(0);
+ return;
}
@@ -298,18 +303,17 @@ fse_run(struct draw_pt_middle_end *middle,
*/
draw_do_flush( draw, DRAW_FLUSH_BACKEND );
- if (fetch_count >= UNDEFINED_VERTEX_ID) {
- assert(0);
- return;
- }
+ if (fetch_count >= UNDEFINED_VERTEX_ID)
+ goto fail;
- hw_verts = draw->render->allocate_vertices( draw->render,
- (ushort)fse->key.output_stride,
- (ushort)fetch_count );
- if (!hw_verts) {
- assert(0);
- return;
- }
+ if (!draw->render->allocate_vertices( draw->render,
+ (ushort)fse->key.output_stride,
+ (ushort)fetch_count ))
+ goto fail;
+
+ hw_verts = draw->render->map_vertices( draw->render );
+ if (!hw_verts)
+ goto fail;
/* Single routine to fetch vertices, run shader and emit HW verts.
@@ -319,9 +323,6 @@ fse_run(struct draw_pt_middle_end *middle,
fetch_count,
hw_verts );
- draw->render->draw( draw->render,
- draw_elts,
- draw_count );
if (0) {
unsigned i;
@@ -333,12 +334,19 @@ fse_run(struct draw_pt_middle_end *middle,
}
}
+ draw->render->unmap_vertices( draw->render, 0, (ushort)(fetch_count - 1) );
+
+ draw->render->draw( draw->render,
+ draw_elts,
+ draw_count );
+
- draw->render->release_vertices( draw->render,
- hw_verts,
- fse->key.output_stride,
- fetch_count );
+ draw->render->release_vertices( draw->render );
+ return;
+fail:
+ assert(0);
+ return;
}
@@ -360,13 +368,14 @@ static boolean fse_run_linear_elts( struct draw_pt_middle_end *middle,
if (count >= UNDEFINED_VERTEX_ID)
return FALSE;
- hw_verts = draw->render->allocate_vertices( draw->render,
- (ushort)fse->key.output_stride,
- (ushort)count );
+ if (!draw->render->allocate_vertices( draw->render,
+ (ushort)fse->key.output_stride,
+ (ushort)count ))
+ return FALSE;
- if (!hw_verts) {
+ hw_verts = draw->render->map_vertices( draw->render );
+ if (!hw_verts)
return FALSE;
- }
/* Single routine to fetch vertices, run shader and emit HW verts.
* Clipping is done elsewhere -- either by the API or on hardware,
@@ -382,11 +391,9 @@ static boolean fse_run_linear_elts( struct draw_pt_middle_end *middle,
draw_count );
+ draw->render->unmap_vertices( draw->render, 0, (ushort)(count - 1) );
- draw->render->release_vertices( draw->render,
- hw_verts,
- fse->key.output_stride,
- count );
+ draw->render->release_vertices( draw->render );
return TRUE;
}
diff --git a/src/gallium/auxiliary/draw/draw_pt_fetch_shade_pipeline.c b/src/gallium/auxiliary/draw/draw_pt_fetch_shade_pipeline.c
index ec3b41c320a..11ac90fc563 100644
--- a/src/gallium/auxiliary/draw/draw_pt_fetch_shade_pipeline.c
+++ b/src/gallium/auxiliary/draw/draw_pt_fetch_shade_pipeline.c
@@ -84,7 +84,8 @@ static void fetch_pipeline_prepare( struct draw_pt_middle_end *middle,
*/
draw_pt_post_vs_prepare( fpme->post_vs,
(boolean)draw->bypass_clipping,
- (boolean)draw->identity_viewport,
+ (boolean)(draw->identity_viewport ||
+ draw->rasterizer->bypass_vs_clip_and_viewport),
(boolean)draw->rasterizer->gl_rasterization_rules );
@@ -140,9 +141,9 @@ static void fetch_pipeline_run( struct draw_pt_middle_end *middle,
(char *)pipeline_verts );
/* Run the shader, note that this overwrites the data[] parts of
- * the pipeline verts. If there is no shader, ie a bypass shader,
- * then the inputs == outputs, and are already in the correct
- * place.
+ * the pipeline verts. If there is no shader, eg if
+ * bypass_vs_clip_and_viewport, then the inputs == outputs, and are
+ * already in the correct place.
*/
if (opt & PT_SHADE)
{
@@ -216,9 +217,9 @@ static void fetch_pipeline_linear_run( struct draw_pt_middle_end *middle,
(char *)pipeline_verts );
/* Run the shader, note that this overwrites the data[] parts of
- * the pipeline verts. If there is no shader, ie a bypass shader,
- * then the inputs == outputs, and are already in the correct
- * place.
+ * the pipeline verts. If there is no shader, ie if
+ * bypass_vs_clip_and_viewport, then the inputs == outputs, and are
+ * already in the correct place.
*/
if (opt & PT_SHADE)
{
@@ -251,9 +252,7 @@ static void fetch_pipeline_linear_run( struct draw_pt_middle_end *middle,
else {
draw_pt_emit_linear( fpme->emit,
(const float (*)[4])pipeline_verts->data,
- count,
fpme->vertex_size,
- 0, /*start*/
count );
}
@@ -288,9 +287,9 @@ static boolean fetch_pipeline_linear_run_elts( struct draw_pt_middle_end *middle
(char *)pipeline_verts );
/* Run the shader, note that this overwrites the data[] parts of
- * the pipeline verts. If there is no shader, ie a bypass shader,
- * then the inputs == outputs, and are already in the correct
- * place.
+ * the pipeline verts. If there is no shader, ie if
+ * bypass_vs_clip_and_viewport, then the inputs == outputs, and are
+ * already in the correct place.
*/
if (opt & PT_SHADE)
{
diff --git a/src/gallium/auxiliary/draw/draw_pt_post_vs.c b/src/gallium/auxiliary/draw/draw_pt_post_vs.c
index 96dc706b998..00d7197b132 100644
--- a/src/gallium/auxiliary/draw/draw_pt_post_vs.c
+++ b/src/gallium/auxiliary/draw/draw_pt_post_vs.c
@@ -200,11 +200,11 @@ boolean draw_pt_post_vs_run( struct pt_post_vs *pvs,
void draw_pt_post_vs_prepare( struct pt_post_vs *pvs,
boolean bypass_clipping,
- boolean identity_viewport,
+ boolean bypass_viewport,
boolean opengl )
{
if (bypass_clipping) {
- if (identity_viewport)
+ if (bypass_viewport)
pvs->run = post_vs_none;
else
pvs->run = post_vs_viewport;
diff --git a/src/gallium/auxiliary/draw/draw_vbuf.h b/src/gallium/auxiliary/draw/draw_vbuf.h
index a1c4c14445b..cccd3bf4358 100644
--- a/src/gallium/auxiliary/draw/draw_vbuf.h
+++ b/src/gallium/auxiliary/draw/draw_vbuf.h
@@ -80,9 +80,14 @@ struct vbuf_render {
* Hardware renderers will use ttm memory, others will just malloc
* something.
*/
- void *(*allocate_vertices)( struct vbuf_render *,
- ushort vertex_size,
- ushort nr_vertices );
+ boolean (*allocate_vertices)( struct vbuf_render *,
+ ushort vertex_size,
+ ushort nr_vertices );
+
+ void *(*map_vertices)( struct vbuf_render * );
+ void (*unmap_vertices)( struct vbuf_render *,
+ ushort min_index,
+ ushort max_index );
/**
* Notify the renderer of the current primitive when it changes.
@@ -109,10 +114,7 @@ struct vbuf_render {
/**
* Called when vbuf is done with this set of vertices:
*/
- void (*release_vertices)( struct vbuf_render *,
- void *vertices,
- unsigned vertex_size,
- unsigned vertices_used );
+ void (*release_vertices)( struct vbuf_render * );
void (*destroy)( struct vbuf_render * );
};
diff --git a/src/gallium/auxiliary/draw/draw_vs_aos.c b/src/gallium/auxiliary/draw/draw_vs_aos.c
index 0c693a4a65c..9e37a26c1e2 100644
--- a/src/gallium/auxiliary/draw/draw_vs_aos.c
+++ b/src/gallium/auxiliary/draw/draw_vs_aos.c
@@ -32,7 +32,7 @@
#include "util/u_memory.h"
#include "util/u_math.h"
#include "pipe/p_shader_tokens.h"
-#include "pipe/p_debug.h"
+#include "util/u_debug.h"
#include "tgsi/tgsi_parse.h"
#include "tgsi/tgsi_util.h"
#include "tgsi/tgsi_exec.h"
@@ -143,7 +143,7 @@ static struct x86_reg get_reg_ptr(struct aos_compilation *cp,
return x86_make_disp(aos_get_x86(cp, 1, X86_CONSTANTS), idx * 4 * sizeof(float));
default:
- ERROR(cp, "unknown reg file");
+ AOS_ERROR(cp, "unknown reg file");
return x86_make_reg(0,0);
}
}
@@ -177,7 +177,7 @@ static void spill( struct aos_compilation *cp, unsigned idx )
(cp->xmm[idx].file != TGSI_FILE_INPUT && /* inputs are fetched into xmm & set dirty */
cp->xmm[idx].file != TGSI_FILE_OUTPUT &&
cp->xmm[idx].file != TGSI_FILE_TEMPORARY)) {
- ERROR(cp, "invalid spill");
+ AOS_ERROR(cp, "invalid spill");
return;
}
else {
@@ -284,6 +284,15 @@ void aos_release_xmm_reg( struct aos_compilation *cp,
}
+static void aos_soft_release_xmm( struct aos_compilation *cp,
+ struct x86_reg reg )
+{
+ if (reg.file == file_XMM) {
+ assert(cp->xmm[reg.idx].last_used == cp->insn_counter);
+ cp->xmm[reg.idx].last_used = cp->insn_counter - 1;
+ }
+}
+
/* Mark an xmm reg as holding the current copy of a shader reg.
@@ -534,7 +543,7 @@ static struct x86_reg fetch_src( struct aos_compilation *cp,
switch (swizzle) {
case TGSI_EXTSWIZZLE_ZERO:
case TGSI_EXTSWIZZLE_ONE:
- ERROR(cp, "not supporting full swizzles yet in tgsi_aos_sse2");
+ AOS_ERROR(cp, "not supporting full swizzles yet in tgsi_aos_sse2");
break;
default:
@@ -555,7 +564,7 @@ static struct x86_reg fetch_src( struct aos_compilation *cp,
break;
default:
- ERROR(cp, "unsupported sign-mode");
+ AOS_ERROR(cp, "unsupported sign-mode");
break;
}
}
@@ -584,15 +593,17 @@ static struct x86_reg fetch_src( struct aos_compilation *cp,
sse_mulps(cp->func, dst, tmp);
aos_release_xmm_reg(cp, tmp.idx);
+ aos_soft_release_xmm(cp, imm_swz);
}
else if (negs) {
struct x86_reg imm_negs = aos_get_internal_xmm(cp, IMM_NEGS);
sse_mulps(cp->func, dst, imm_negs);
+ aos_soft_release_xmm(cp, imm_negs);
}
if (abs && abs != 0xf) {
- ERROR(cp, "unsupported partial abs");
+ AOS_ERROR(cp, "unsupported partial abs");
}
else if (abs) {
struct x86_reg neg = aos_get_internal(cp, IMM_NEGS);
@@ -603,8 +614,10 @@ static struct x86_reg fetch_src( struct aos_compilation *cp,
sse_maxps(cp->func, dst, tmp);
aos_release_xmm_reg(cp, tmp.idx);
+ aos_soft_release_xmm(cp, neg);
}
+ aos_soft_release_xmm(cp, arg0);
return dst;
}
@@ -657,7 +670,7 @@ static void x87_fld_src( struct aos_compilation *cp,
break;
default:
- ERROR(cp, "unsupported sign-mode");
+ AOS_ERROR(cp, "unsupported sign-mode");
break;
}
}
@@ -1559,7 +1572,6 @@ static boolean emit_RCP( struct aos_compilation *cp, const struct tgsi_full_inst
*/
static boolean emit_RSQ( struct aos_compilation *cp, const struct tgsi_full_instruction *op )
{
-
if (0) {
struct x86_reg arg0 = fetch_src(cp, &op->FullSrcRegisters[0]);
struct x86_reg r = aos_get_xmm_reg(cp);
@@ -1568,21 +1580,30 @@ static boolean emit_RSQ( struct aos_compilation *cp, const struct tgsi_full_inst
return TRUE;
}
else {
- struct x86_reg arg0 = fetch_src(cp, &op->FullSrcRegisters[0]);
- struct x86_reg r = aos_get_xmm_reg(cp);
+ struct x86_reg arg0 = fetch_src(cp, &op->FullSrcRegisters[0]);
+ struct x86_reg r = aos_get_xmm_reg(cp);
struct x86_reg neg_half = get_reg_ptr( cp, AOS_FILE_INTERNAL, IMM_RSQ );
struct x86_reg one_point_five = x86_make_disp( neg_half, 4 );
struct x86_reg src = get_xmm_writable( cp, arg0 );
-
- sse_rsqrtss( cp->func, r, src ); /* rsqrtss(a) */
- sse_mulss( cp->func, src, neg_half ); /* -.5 * a */
- sse_mulss( cp->func, src, r ); /* -.5 * a * r */
- sse_mulss( cp->func, src, r ); /* -.5 * a * r * r */
- sse_addss( cp->func, src, one_point_five ); /* 1.5 - .5 * a * r * r */
- sse_mulss( cp->func, r, src ); /* r * (1.5 - .5 * a * r * r) */
+ struct x86_reg neg = aos_get_internal(cp, IMM_NEGS);
+ struct x86_reg tmp = aos_get_xmm_reg(cp);
+
+ sse_movaps(cp->func, tmp, src);
+ sse_mulps(cp->func, tmp, neg);
+ sse_maxps(cp->func, tmp, src);
+
+ sse_rsqrtss( cp->func, r, tmp ); /* rsqrtss(a) */
+ sse_mulss( cp->func, tmp, neg_half ); /* -.5 * a */
+ sse_mulss( cp->func, tmp, r ); /* -.5 * a * r */
+ sse_mulss( cp->func, tmp, r ); /* -.5 * a * r * r */
+ sse_addss( cp->func, tmp, one_point_five ); /* 1.5 - .5 * a * r * r */
+ sse_mulss( cp->func, r, tmp ); /* r * (1.5 - .5 * a * r * r) */
store_scalar_dest(cp, &op->FullDstRegisters[0], r);
+
+ aos_release_xmm_reg(cp, tmp.idx);
+
return TRUE;
}
}
diff --git a/src/gallium/auxiliary/draw/draw_vs_aos.h b/src/gallium/auxiliary/draw/draw_vs_aos.h
index 264387517b0..2cf72ddf7b1 100644
--- a/src/gallium/auxiliary/draw/draw_vs_aos.h
+++ b/src/gallium/auxiliary/draw/draw_vs_aos.h
@@ -204,7 +204,7 @@ struct x86_reg aos_get_internal_xmm( struct aos_compilation *cp,
unsigned imm );
-#define ERROR(cp, msg) \
+#define AOS_ERROR(cp, msg) \
do { \
if (0) debug_printf("%s: x86 translation failed: %s\n", __FUNCTION__, msg); \
cp->error = 1; \
diff --git a/src/gallium/auxiliary/draw/draw_vs_aos_io.c b/src/gallium/auxiliary/draw/draw_vs_aos_io.c
index 39f75b50b76..a6eb37d1280 100644
--- a/src/gallium/auxiliary/draw/draw_vs_aos_io.c
+++ b/src/gallium/auxiliary/draw/draw_vs_aos_io.c
@@ -199,7 +199,7 @@ static boolean load_input( struct aos_compilation *cp,
emit_load_R8G8B8A8_UNORM(cp, dataXMM, src);
break;
default:
- ERROR(cp, "unhandled input format");
+ AOS_ERROR(cp, "unhandled input format");
return FALSE;
}
@@ -410,7 +410,7 @@ static boolean emit_output( struct aos_compilation *cp,
}
break;
default:
- ERROR(cp, "unhandled output format");
+ AOS_ERROR(cp, "unhandled output format");
return FALSE;
}
generated by cgit v1.2.3 (git 2.39.1) at 2024-12-11 10:09:26 +0000