#include "math/m_translate.h"
#include "tnl/tnl.h"
#include "tnl/t_context.h"
-#include "tnl/t_imm_exec.h"
#include "tnl/t_pipeline.h"
+#include "tnl/t_vtx_api.h" /* for _tnl_FlushVertices */
#include "radeon_context.h"
#include "radeon_ioctl.h"
copy_pv_func copy_pv;
GLboolean (*check_tex_sizes)( GLcontext *ctx );
GLuint vertex_size;
- GLuint vertex_stride_shift;
GLuint vertex_format;
} setup_tab[RADEON_MAX_SETUP];
#define GET_TEXSOURCE(n) n
#define GET_VERTEX_FORMAT() RADEON_CONTEXT(ctx)->swtcl.vertex_format
#define GET_VERTEX_STORE() RADEON_CONTEXT(ctx)->swtcl.verts
-#define GET_VERTEX_STRIDE_SHIFT() RADEON_CONTEXT(ctx)->swtcl.vertex_stride_shift
-#define GET_UBYTE_COLOR_STORE() &RADEON_CONTEXT(ctx)->UbyteColor
-#define GET_UBYTE_SPEC_COLOR_STORE() &RADEON_CONTEXT(ctx)->UbyteSecondaryColor
+#define GET_VERTEX_SIZE() RADEON_CONTEXT(ctx)->swtcl.vertex_size * sizeof(GLuint)
#define HAVE_HW_VIEWPORT 1
/* Tiny vertices don't seem to work atm - haven't looked into why.
#define CHECK_HW_DIVIDE (!(ctx->_TriangleCaps & (DD_TRI_LIGHT_TWOSIDE| \
DD_TRI_UNFILLED)))
-#define IMPORT_QUALIFIER
-#define IMPORT_FLOAT_COLORS radeon_import_float_colors
-#define IMPORT_FLOAT_SPEC_COLORS radeon_import_float_spec_colors
-
#define INTERP_VERTEX setup_tab[RADEON_CONTEXT(ctx)->swtcl.SetupIndex].interp
#define COPY_PV_VERTEX setup_tab[RADEON_CONTEXT(ctx)->swtcl.SetupIndex].copy_pv
RADEON_NEWPRIM(rmesa);
rmesa->swtcl.vertex_format = setup_tab[ind].vertex_format;
rmesa->swtcl.vertex_size = setup_tab[ind].vertex_size;
- rmesa->swtcl.vertex_stride_shift = setup_tab[ind].vertex_stride_shift;
}
if (!(ctx->_TriangleCaps & (DD_TRI_LIGHT_TWOSIDE|DD_TRI_UNFILLED))) {
GLuint newinputs )
{
radeonContextPtr rmesa = RADEON_CONTEXT( ctx );
- GLubyte *v = ((GLubyte *)rmesa->swtcl.verts +
- (start << rmesa->swtcl.vertex_stride_shift));
- GLuint stride = 1 << rmesa->swtcl.vertex_stride_shift;
+ GLuint stride = rmesa->swtcl.vertex_size * sizeof(int);
+ GLubyte *v = ((GLubyte *)rmesa->swtcl.verts + (start * stride));
newinputs |= rmesa->swtcl.SetupNewInputs;
rmesa->swtcl.SetupNewInputs = 0;
RADEON_NEWPRIM(rmesa);
rmesa->swtcl.vertex_format = setup_tab[ind].vertex_format;
rmesa->swtcl.vertex_size = setup_tab[ind].vertex_size;
- rmesa->swtcl.vertex_stride_shift = setup_tab[ind].vertex_stride_shift;
}
{
-void radeon_emit_contiguous_verts( GLcontext *ctx, GLuint start, GLuint count )
+static void *radeon_emit_contiguous_verts( GLcontext *ctx,
+ GLuint start,
+ GLuint count,
+ void *dest)
{
radeonContextPtr rmesa = RADEON_CONTEXT(ctx);
- GLuint vertex_size = rmesa->swtcl.vertex_size * 4;
- CARD32 *dest = radeonAllocDmaLowVerts( rmesa, count-start, vertex_size );
- setup_tab[rmesa->swtcl.SetupIndex].emit( ctx, start, count, dest,
- vertex_size );
+ GLuint stride = rmesa->swtcl.vertex_size * 4;
+ setup_tab[rmesa->swtcl.SetupIndex].emit( ctx, start, count, dest, stride );
+ return (void *)((char *)dest + stride * (count - start));
}
}
-static void VERT_FALLBACK( GLcontext *ctx,
- GLuint start,
- GLuint count,
- GLuint flags )
-{
- TNLcontext *tnl = TNL_CONTEXT(ctx);
- tnl->Driver.Render.PrimitiveNotify( ctx, flags & PRIM_MODE_MASK );
- tnl->Driver.Render.BuildVertices( ctx, start, count, ~0 );
- tnl->Driver.Render.PrimTabVerts[flags&PRIM_MODE_MASK]( ctx, start, count, flags );
- RADEON_CONTEXT(ctx)->swtcl.SetupNewInputs = VERT_BIT_CLIP;
-}
-
-static void ELT_FALLBACK( GLcontext *ctx,
- GLuint start,
- GLuint count,
- GLuint flags )
-{
- TNLcontext *tnl = TNL_CONTEXT(ctx);
- tnl->Driver.Render.PrimitiveNotify( ctx, flags & PRIM_MODE_MASK );
- tnl->Driver.Render.BuildVertices( ctx, start, count, ~0 );
- tnl->Driver.Render.PrimTabElts[flags&PRIM_MODE_MASK]( ctx, start, count, flags );
- RADEON_CONTEXT(ctx)->swtcl.SetupNewInputs = VERT_BIT_CLIP;
-}
#define LOCAL_VARS radeonContextPtr rmesa = RADEON_CONTEXT(ctx)
-#define ELTS_VARS GLushort *dest
+#define ELTS_VARS( buf ) GLushort *dest = buf
#define INIT( prim ) radeonDmaPrimitive( rmesa, prim )
#define ELT_INIT(prim) radeonEltPrimitive( rmesa, prim )
-#define NEW_PRIMITIVE() RADEON_NEWPRIM( rmesa )
-#define NEW_BUFFER() radeonRefillCurrentDmaRegion( rmesa )
+#define FLUSH() RADEON_NEWPRIM( rmesa )
#define GET_CURRENT_VB_MAX_VERTS() \
(((int)rmesa->dma.current.end - (int)rmesa->dma.current.ptr) / (rmesa->swtcl.vertex_size*4))
#define GET_SUBSEQUENT_VB_MAX_VERTS() \
((RADEON_CMD_BUF_SZ - 1024) / 2)
+static void *radeon_alloc_elts( radeonContextPtr rmesa, int nr )
+{
+ if (rmesa->dma.flush == radeonFlushElts &&
+ rmesa->store.cmd_used + nr*2 < RADEON_CMD_BUF_SZ) {
-/* How do you extend an existing primitive?
- */
-#define ALLOC_ELTS(nr) \
-do { \
- if (rmesa->dma.flush == radeonFlushElts && \
- rmesa->store.cmd_used + nr*2 < RADEON_CMD_BUF_SZ) { \
- \
- dest = (GLushort *)(rmesa->store.cmd_buf + \
- rmesa->store.cmd_used); \
- rmesa->store.cmd_used += nr*2; \
- } \
- else { \
- if (rmesa->dma.flush) { \
- rmesa->dma.flush( rmesa ); \
- } \
- \
- radeonEmitVertexAOS( rmesa, \
- rmesa->swtcl.vertex_size, \
- (rmesa->radeonScreen->gart_buffer_offset + \
- rmesa->swtcl.indexed_verts.buf->buf->idx * \
- RADEON_BUFFER_SIZE + \
- rmesa->swtcl.indexed_verts.start)); \
- \
- dest = radeonAllocEltsOpenEnded( rmesa, \
- rmesa->swtcl.vertex_format, \
- rmesa->swtcl.hw_primitive, \
- nr ); \
- } \
-} while (0)
+ rmesa->store.cmd_used += nr*2;
-#define ALLOC_ELTS_NEW_PRIMITIVE(nr) ALLOC_ELTS( nr )
+ return (void *)(rmesa->store.cmd_buf + rmesa->store.cmd_used);
+ }
+ else {
+ if (rmesa->dma.flush) {
+ rmesa->dma.flush( rmesa );
+ }
+
+ radeonEmitVertexAOS( rmesa,
+ rmesa->swtcl.vertex_size,
+ (rmesa->radeonScreen->gart_buffer_offset +
+ rmesa->swtcl.indexed_verts.buf->buf->idx *
+ RADEON_BUFFER_SIZE +
+ rmesa->swtcl.indexed_verts.start));
+
+ return (void *) radeonAllocEltsOpenEnded( rmesa,
+ rmesa->swtcl.vertex_format,
+ rmesa->swtcl.hw_primitive,
+ nr );
+ }
+}
+
+#define ALLOC_ELTS(nr) radeon_alloc_elts(rmesa, nr)
#ifdef MESA_BIG_ENDIAN
/* We could do without (most of) this ugliness if dest was always 32 bit word aligned... */
#endif
#define EMIT_TWO_ELTS(offset, x, y) *(GLuint *)(dest+offset) = ((y)<<16)|(x);
#define INCR_ELTS( nr ) dest += nr
+#define ELTPTR dest
#define RELEASE_ELT_VERTS() \
radeonReleaseDmaRegion( rmesa, &rmesa->swtcl.indexed_verts, __FUNCTION__ )
-#define EMIT_VERTS( ctx, j, nr ) \
- radeon_emit_contiguous_verts(ctx, j, (j)+(nr))
#define EMIT_INDEXED_VERTS( ctx, start, count ) \
radeon_emit_indexed_verts( ctx, start, count )
+#define ALLOC_VERTS( nr ) \
+ radeonAllocDmaLowVerts( rmesa, nr, rmesa->swtcl.vertex_size * 4 )
+#define EMIT_VERTS( ctx, j, nr, buf ) \
+ radeon_emit_contiguous_verts(ctx, j, (j)+(nr), buf)
+
#define TAG(x) radeon_dma_##x
#include "tnl_dd/t_dd_dmatmp.h"
static GLboolean radeon_run_render( GLcontext *ctx,
- struct gl_pipeline_stage *stage )
+ struct tnl_pipeline_stage *stage )
{
radeonContextPtr rmesa = RADEON_CONTEXT(ctx);
TNLcontext *tnl = TNL_CONTEXT(ctx);
struct vertex_buffer *VB = &tnl->vb;
- GLuint i, length, flags = 0;
render_func *tab = TAG(render_tab_verts);
+ GLuint i;
if (rmesa->swtcl.indexed_verts.buf && (!VB->Elts || stage->changed_inputs))
RELEASE_ELT_VERTS();
- if (VB->ClipOrMask || /* No clipping */
- rmesa->swtcl.RenderIndex != 0 || /* No per-vertex manipulations */
- ctx->Line.StippleFlag) /* GH: THIS IS A HACK!!! */
+ if (rmesa->swtcl.RenderIndex != 0 ||
+ !radeon_dma_validate_render( ctx, VB ))
return GL_TRUE;
tnl->Driver.Render.Start( ctx );
if (VB->Elts) {
tab = TAG(render_tab_elts);
- if (!rmesa->swtcl.indexed_verts.buf)
- if (!TAG(emit_elt_verts)(ctx, 0, VB->Count))
- return GL_TRUE; /* too many vertices */
+ if (!rmesa->swtcl.indexed_verts.buf) {
+ if (VB->Count > GET_SUBSEQUENT_VB_MAX_VERTS())
+ return GL_TRUE;
+ EMIT_INDEXED_VERTS(ctx, 0, VB->Count);
+ }
}
- for (i = 0 ; !(flags & PRIM_LAST) ; i += length)
+ for (i = 0 ; i < VB->PrimitiveCount ; i++)
{
- flags = VB->Primitive[i];
- length = VB->PrimitiveLength[i];
+ GLuint prim = VB->Primitive[i].mode;
+ GLuint start = VB->Primitive[i].start;
+ GLuint length = VB->Primitive[i].count;
+
+ if (!length)
+ continue;
if (RADEON_DEBUG & DEBUG_PRIMS)
fprintf(stderr, "radeon_render.c: prim %s %d..%d\n",
- _mesa_lookup_enum_by_nr(flags & PRIM_MODE_MASK),
- i, i+length);
+ _mesa_lookup_enum_by_nr(prim & PRIM_MODE_MASK),
+ start, start+length);
if (length)
- tab[flags & PRIM_MODE_MASK]( ctx, i, i + length, flags );
+ tab[prim & PRIM_MODE_MASK]( ctx, start, start + length, prim );
}
tnl->Driver.Render.Finish( ctx );
static void radeon_check_render( GLcontext *ctx,
- struct gl_pipeline_stage *stage )
+ struct tnl_pipeline_stage *stage )
{
- GLuint inputs = VERT_BIT_POS | VERT_BIT_CLIP | VERT_BIT_COLOR0;
-
- if (ctx->RenderMode == GL_RENDER) {
- if (ctx->_TriangleCaps & DD_SEPARATE_SPECULAR)
- inputs |= VERT_BIT_COLOR1;
-
- if (ctx->Texture.Unit[0]._ReallyEnabled)
- inputs |= VERT_BIT_TEX0;
-
- if (ctx->Texture.Unit[1]._ReallyEnabled)
- inputs |= VERT_BIT_TEX1;
-
- if (ctx->Fog.Enabled)
- inputs |= VERT_BIT_FOG;
- }
-
- stage->inputs = inputs;
+ stage->inputs = TNL_CONTEXT(ctx)->render_inputs;
}
-static void dtr( struct gl_pipeline_stage *stage )
+static void dtr( struct tnl_pipeline_stage *stage )
{
(void)stage;
}
-const struct gl_pipeline_stage _radeon_render_stage =
+const struct tnl_pipeline_stage _radeon_render_stage =
{
"radeon render",
(_DD_NEW_SEPARATE_SPECULAR |
static GLboolean run_texrect_stage( GLcontext *ctx,
- struct gl_pipeline_stage *stage )
+ struct tnl_pipeline_stage *stage )
{
struct texrect_stage_data *store = TEXRECT_STAGE_DATA(stage);
radeonContextPtr rmesa = RADEON_CONTEXT(ctx);
if (stage->changed_inputs & VERT_BIT_TEX(i)) {
struct gl_texture_object *texObj = ctx->Texture.Unit[i].CurrentRect;
- struct gl_texture_image *texImage = texObj->Image[texObj->BaseLevel];
+ struct gl_texture_image *texImage = texObj->Image[0][texObj->BaseLevel];
const GLfloat iw = 1.0/texImage->Width;
const GLfloat ih = 1.0/texImage->Height;
GLfloat *in = (GLfloat *)VB->TexCoordPtr[i]->data;
/* Called the first time stage->run() is invoked.
*/
static GLboolean alloc_texrect_data( GLcontext *ctx,
- struct gl_pipeline_stage *stage )
+ struct tnl_pipeline_stage *stage )
{
struct vertex_buffer *VB = &TNL_CONTEXT(ctx)->vb;
struct texrect_stage_data *store;
static void check_texrect( GLcontext *ctx,
- struct gl_pipeline_stage *stage )
+ struct tnl_pipeline_stage *stage )
{
GLuint flags = 0;
}
-static void free_texrect_data( struct gl_pipeline_stage *stage )
+static void free_texrect_data( struct tnl_pipeline_stage *stage )
{
struct texrect_stage_data *store = TEXRECT_STAGE_DATA(stage);
GLuint i;
}
-const struct gl_pipeline_stage _radeon_texrect_stage =
+const struct tnl_pipeline_stage _radeon_texrect_stage =
{
"radeon texrect stage", /* name */
_NEW_TEXTURE, /* check_state */
***********************************************************************/
#undef LOCAL_VARS
+#undef ALLOC_VERTS
#define CTX_ARG radeonContextPtr rmesa
#define CTX_ARG2 rmesa
#define GET_VERTEX_DWORDS() rmesa->swtcl.vertex_size
#undef LOCAL_VARS
#define LOCAL_VARS \
radeonContextPtr rmesa = RADEON_CONTEXT(ctx); \
- const GLuint shift = rmesa->swtcl.vertex_stride_shift; \
const char *radeonverts = (char *)rmesa->swtcl.verts;
-#define VERT(x) (radeonVertex *)(radeonverts + (x << shift))
+#define VERT(x) (radeonVertex *)(radeonverts + (x * vertsize * sizeof(int)))
#define VERTEX radeonVertex
#undef TAG
#define TAG(x) radeon_##x
#define VERT_Y(_v) _v->v.y
#define VERT_Z(_v) _v->v.z
#define AREA_IS_CCW( a ) (a < 0)
-#define GET_VERTEX(e) (rmesa->swtcl.verts + (e<<rmesa->swtcl.vertex_stride_shift))
+#define GET_VERTEX(e) (rmesa->swtcl.verts + (e * rmesa->swtcl.vertex_size * sizeof(int)))
+
+#define VERT_SET_RGBA( v, c ) \
+do { \
+ radeon_color_t *color = (radeon_color_t *)&((v)->ui[coloroffset]); \
+ UNCLAMPED_FLOAT_TO_UBYTE(color->red, (c)[0]); \
+ UNCLAMPED_FLOAT_TO_UBYTE(color->green, (c)[1]); \
+ UNCLAMPED_FLOAT_TO_UBYTE(color->blue, (c)[2]); \
+ UNCLAMPED_FLOAT_TO_UBYTE(color->alpha, (c)[3]); \
+} while (0)
-#define VERT_SET_RGBA( v, c ) v->ui[coloroffset] = LE32_TO_CPU(*(GLuint *)c)
#define VERT_COPY_RGBA( v0, v1 ) v0->ui[coloroffset] = v1->ui[coloroffset]
-#define VERT_SAVE_RGBA( idx ) color[idx] = CPU_TO_LE32(v[idx]->ui[coloroffset])
-#define VERT_RESTORE_RGBA( idx ) v[idx]->ui[coloroffset] = LE32_TO_CPU(color[idx])
-
-#define VERT_SET_SPEC( v0, c ) if (havespec) { \
- v0->v.specular.red = (c)[0]; \
- v0->v.specular.green = (c)[1]; \
- v0->v.specular.blue = (c)[2]; }
-#define VERT_COPY_SPEC( v0, v1 ) if (havespec) { \
- v0->v.specular.red = v1->v.specular.red; \
- v0->v.specular.green = v1->v.specular.green; \
- v0->v.specular.blue = v1->v.specular.blue; }
-#define VERT_SAVE_SPEC( idx ) if (havespec) spec[idx] = CPU_TO_LE32(v[idx]->ui[5])
-#define VERT_RESTORE_SPEC( idx ) if (havespec) v[idx]->ui[5] = LE32_TO_CPU(spec[idx])
+
+#define VERT_SET_SPEC( v0, c ) \
+do { \
+ if (havespec) { \
+ UNCLAMPED_FLOAT_TO_UBYTE(v0->v.specular.red, (c)[0]); \
+ UNCLAMPED_FLOAT_TO_UBYTE(v0->v.specular.green, (c)[1]); \
+ UNCLAMPED_FLOAT_TO_UBYTE(v0->v.specular.blue, (c)[2]); \
+ } \
+} while (0)
+#define VERT_COPY_SPEC( v0, v1 ) \
+do { \
+ if (havespec) { \
+ v0->v.specular.red = v1->v.specular.red; \
+ v0->v.specular.green = v1->v.specular.green; \
+ v0->v.specular.blue = v1->v.specular.blue; \
+ } \
+} while (0)
+
+/* These don't need LE32_TO_CPU() as they used to save and restore
+ * colors which are already in the correct format.
+ */
+#define VERT_SAVE_RGBA( idx ) color[idx] = v[idx]->ui[coloroffset]
+#define VERT_RESTORE_RGBA( idx ) v[idx]->ui[coloroffset] = color[idx]
+#define VERT_SAVE_SPEC( idx ) if (havespec) spec[idx] = v[idx]->ui[5]
+#define VERT_RESTORE_SPEC( idx ) if (havespec) v[idx]->ui[5] = spec[idx]
#undef LOCAL_VARS
#undef TAG
/* Render unclipped begin/end objects */
/**********************************************************************/
-#define VERT(x) (radeonVertex *)(radeonverts + (x << shift))
+#define VERT(x) (radeonVertex *)(radeonverts + (x * vertsize * sizeof(int)))
#define RENDER_POINTS( start, count ) \
for ( ; start < count ; start++) \
radeon_point( rmesa, VERT(start) )
#undef LOCAL_VARS
#define LOCAL_VARS \
radeonContextPtr rmesa = RADEON_CONTEXT(ctx); \
- const GLuint shift = rmesa->swtcl.vertex_stride_shift; \
+ const GLuint vertsize = rmesa->swtcl.vertex_size; \
const char *radeonverts = (char *)rmesa->swtcl.verts; \
const GLuint * const elt = TNL_CONTEXT(ctx)->vb.Elts; \
const GLboolean stipple = ctx->Line.StippleFlag; \
void radeonFlushVertices( GLcontext *ctx, GLuint flags )
{
- _tnl_flush_vertices( ctx, flags );
+ _tnl_FlushVertices( ctx, flags );
if (flags & FLUSH_STORED_VERTICES)
RADEON_NEWPRIM( RADEON_CONTEXT( ctx ) );
rmesa->swtcl.verts = 0;
}
- if (rmesa->UbyteSecondaryColor.Ptr) {
- ALIGN_FREE(rmesa->UbyteSecondaryColor.Ptr);
- rmesa->UbyteSecondaryColor.Ptr = 0;
- }
-
- if (rmesa->UbyteColor.Ptr) {
- ALIGN_FREE(rmesa->UbyteColor.Ptr);
- rmesa->UbyteColor.Ptr = 0;
- }
}
projects / mesa.git / blobdiff