aboutsummaryrefslogtreecommitdiff
path: root/nx-X11/extras/Mesa/src/mesa/tnl/t_vtx_x86.c
diff options
context:
space:
mode:
Diffstat (limited to 'nx-X11/extras/Mesa/src/mesa/tnl/t_vtx_x86.c')
-rw-r--r--nx-X11/extras/Mesa/src/mesa/tnl/t_vtx_x86.c385
1 files changed, 385 insertions, 0 deletions
diff --git a/nx-X11/extras/Mesa/src/mesa/tnl/t_vtx_x86.c b/nx-X11/extras/Mesa/src/mesa/tnl/t_vtx_x86.c
new file mode 100644
index 000000000..38cdad451
--- /dev/null
+++ b/nx-X11/extras/Mesa/src/mesa/tnl/t_vtx_x86.c
@@ -0,0 +1,385 @@
+/**************************************************************************
+
+Copyright 2004 Tungsten Graphics Inc., Cedar Park, Texas.
+
+All Rights Reserved.
+
+Permission is hereby granted, free of charge, to any person obtaining a
+copy of this software and associated documentation files (the "Software"),
+to deal in the Software without restriction, including without limitation
+on the rights to use, copy, modify, merge, publish, distribute, sub
+license, and/or sell copies of the Software, and to permit persons to whom
+the Software is furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice (including the next
+paragraph) shall be included in all copies or substantial portions of the
+Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
+ATI, TUNGSTEN GRAPHICS AND/OR THEIR SUPPLIERS BE LIABLE FOR ANY CLAIM,
+DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
+OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
+USE OR OTHER DEALINGS IN THE SOFTWARE.
+
+**************************************************************************/
+
+/*
+ * Authors:
+ * Keith Whitwell <keith@tungstengraphics.com>
+ * Daniel Borca <dborca@yahoo.com>
+ */
+
+
+#include "glheader.h"
+#include "context.h"
+#include "macros.h"
+#include "vtxfmt.h"
+#include "dlist.h"
+#include "state.h"
+#include "light.h"
+#include "api_arrayelt.h"
+#include "api_noop.h"
+#include "t_vtx_api.h"
+#include "simple_list.h"
+
+
+#if defined(USE_X86_ASM) && !defined(HAVE_NONSTANDARD_GLAPIENTRY)
+
+#define EXTERN( FUNC ) \
+extern const char FUNC[]; \
+extern const char FUNC##_end[]
+
+EXTERN( _tnl_x86_Attribute1fv );
+EXTERN( _tnl_x86_Attribute2fv );
+EXTERN( _tnl_x86_Attribute3fv );
+EXTERN( _tnl_x86_Attribute4fv );
+EXTERN( _tnl_x86_Vertex1fv );
+EXTERN( _tnl_x86_Vertex2fv );
+EXTERN( _tnl_x86_Vertex3fv );
+EXTERN( _tnl_x86_Vertex4fv );
+
+EXTERN( _tnl_x86_dispatch_attrf1 );
+EXTERN( _tnl_x86_dispatch_attrf2 );
+EXTERN( _tnl_x86_dispatch_attrf3 );
+EXTERN( _tnl_x86_dispatch_attrf4 );
+EXTERN( _tnl_x86_dispatch_attrfv );
+EXTERN( _tnl_x86_dispatch_multitexcoordf1 );
+EXTERN( _tnl_x86_dispatch_multitexcoordf2 );
+EXTERN( _tnl_x86_dispatch_multitexcoordf3 );
+EXTERN( _tnl_x86_dispatch_multitexcoordf4 );
+EXTERN( _tnl_x86_dispatch_multitexcoordfv );
+EXTERN( _tnl_x86_dispatch_vertexattribf1 );
+EXTERN( _tnl_x86_dispatch_vertexattribf2 );
+EXTERN( _tnl_x86_dispatch_vertexattribf3 );
+EXTERN( _tnl_x86_dispatch_vertexattribf4 );
+EXTERN( _tnl_x86_dispatch_vertexattribfv );
+
+EXTERN( _tnl_x86_choose_fv );
+
+
+#define DONT_KNOW_OFFSETS 1
+
+
+#define DFN( FUNC, CACHE, KEY ) \
+ struct _tnl_dynfn *dfn = MALLOC_STRUCT( _tnl_dynfn );\
+ const char *start = FUNC; \
+ const char *end = FUNC##_end; \
+ int offset = 0; \
+ insert_at_head( &CACHE, dfn ); \
+ dfn->key = KEY; \
+ dfn->code = ALIGN_MALLOC( end - start, 16 ); \
+ memcpy (dfn->code, start, end - start)
+
+
+
+#define FIXUP( CODE, KNOWN_OFFSET, CHECKVAL, NEWVAL ) \
+do { \
+ GLint subst = 0x10101010 + CHECKVAL; \
+ \
+ if (DONT_KNOW_OFFSETS) { \
+ while (*(int *)(CODE+offset) != subst) offset++; \
+ *(int *)(CODE+offset) = (int)(NEWVAL); \
+ if (0) fprintf(stderr, "%s/%d: offset %d, new value: 0x%x\n", __FILE__, __LINE__, offset, (int)(NEWVAL)); \
+ offset += 4; \
+ } \
+ else { \
+ int *icode = (int *)(CODE+KNOWN_OFFSET); \
+ assert (*icode == subst); \
+ *icode = (int)NEWVAL; \
+ } \
+} while (0)
+
+
+
+#define FIXUPREL( CODE, KNOWN_OFFSET, CHECKVAL, NEWVAL )\
+do { \
+ GLint subst = 0x10101010 + CHECKVAL; \
+ \
+ if (DONT_KNOW_OFFSETS) { \
+ while (*(int *)(CODE+offset) != subst) offset++; \
+ *(int *)(CODE+offset) = (int)(NEWVAL) - ((int)(CODE)+offset) - 4; \
+ if (0) fprintf(stderr, "%s/%d: offset %d, new value: 0x%x\n", __FILE__, __LINE__, offset, (int)(NEWVAL) - ((int)(CODE)+offset) - 4); \
+ offset += 4; \
+ } \
+ else { \
+ int *icode = (int *)(CODE+KNOWN_OFFSET); \
+ assert (*icode == subst); \
+ *icode = (int)(NEWVAL) - (int)(icode) - 4; \
+ } \
+} while (0)
+
+
+
+
+/* Build specialized versions of the immediate calls on the fly for
+ * the current state. Generic x86 versions.
+ */
+
+static struct _tnl_dynfn *makeX86Vertex1fv( GLcontext *ctx, int vertex_size )
+{
+ TNLcontext *tnl = TNL_CONTEXT(ctx);
+ DFN ( _tnl_x86_Vertex1fv, tnl->vtx.cache.Vertex[1-1], vertex_size );
+
+ FIXUP(dfn->code, 0, 0, (int)&tnl->vtx.vbptr);
+ FIXUP(dfn->code, 0, 1, vertex_size - 1);
+ FIXUP(dfn->code, 0, 2, (int)&tnl->vtx.vertex[1]);
+ FIXUP(dfn->code, 0, 0, (int)&tnl->vtx.vbptr);
+ FIXUP(dfn->code, 0, 3, (int)&tnl->vtx.counter);
+ FIXUP(dfn->code, 0, 3, (int)&tnl->vtx.counter);
+ FIXUP(dfn->code, 0, 4, (int)ctx);
+ FIXUPREL(dfn->code, 0, 5, (int)&_tnl_wrap_filled_vertex);
+
+ return dfn;
+}
+
+static struct _tnl_dynfn *makeX86Vertex2fv( GLcontext *ctx, int vertex_size )
+{
+ TNLcontext *tnl = TNL_CONTEXT(ctx);
+ DFN ( _tnl_x86_Vertex2fv, tnl->vtx.cache.Vertex[2-1], vertex_size );
+
+ FIXUP(dfn->code, 0, 0, (int)&tnl->vtx.vbptr);
+ FIXUP(dfn->code, 0, 1, vertex_size - 2);
+ FIXUP(dfn->code, 0, 2, (int)&tnl->vtx.vertex[2]);
+ FIXUP(dfn->code, 0, 0, (int)&tnl->vtx.vbptr);
+ FIXUP(dfn->code, 0, 3, (int)&tnl->vtx.counter);
+ FIXUP(dfn->code, 0, 3, (int)&tnl->vtx.counter);
+ FIXUP(dfn->code, 0, 4, (int)ctx);
+ FIXUPREL(dfn->code, 0, 5, (int)&_tnl_wrap_filled_vertex);
+
+ return dfn;
+}
+
+static struct _tnl_dynfn *makeX86Vertex3fv( GLcontext *ctx, int vertex_size )
+{
+ TNLcontext *tnl = TNL_CONTEXT(ctx);
+
+ switch (vertex_size) {
+ default: {
+ DFN ( _tnl_x86_Vertex3fv, tnl->vtx.cache.Vertex[3-1], vertex_size );
+
+ FIXUP(dfn->code, 0, 0, (int)&tnl->vtx.vbptr);
+ FIXUP(dfn->code, 0, 1, vertex_size - 3);
+ FIXUP(dfn->code, 0, 2, (int)&tnl->vtx.vertex[3]);
+ FIXUP(dfn->code, 0, 0, (int)&tnl->vtx.vbptr);
+ FIXUP(dfn->code, 0, 3, (int)&tnl->vtx.counter);
+ FIXUP(dfn->code, 0, 3, (int)&tnl->vtx.counter);
+ FIXUP(dfn->code, 0, 4, (int)ctx);
+ FIXUPREL(dfn->code, 0, 5, (int)&_tnl_wrap_filled_vertex);
+ return dfn;
+ }
+ }
+}
+
+static struct _tnl_dynfn *makeX86Vertex4fv( GLcontext *ctx, int vertex_size )
+{
+ TNLcontext *tnl = TNL_CONTEXT(ctx);
+ DFN ( _tnl_x86_Vertex4fv, tnl->vtx.cache.Vertex[4-1], vertex_size );
+
+ FIXUP(dfn->code, 0, 0, (int)&tnl->vtx.vbptr);
+ FIXUP(dfn->code, 0, 1, vertex_size - 4);
+ FIXUP(dfn->code, 0, 2, (int)&tnl->vtx.vertex[4]);
+ FIXUP(dfn->code, 0, 0, (int)&tnl->vtx.vbptr);
+ FIXUP(dfn->code, 0, 3, (int)&tnl->vtx.counter);
+ FIXUP(dfn->code, 0, 3, (int)&tnl->vtx.counter);
+ FIXUP(dfn->code, 0, 4, (int)ctx);
+ FIXUPREL(dfn->code, 0, 5, (int)&_tnl_wrap_filled_vertex);
+
+ return dfn;
+}
+
+
+static struct _tnl_dynfn *makeX86Attribute1fv( GLcontext *ctx, int dest )
+{
+ TNLcontext *tnl = TNL_CONTEXT(ctx);
+ DFN ( _tnl_x86_Attribute1fv, tnl->vtx.cache.Attribute[1-1], dest );
+
+ FIXUP(dfn->code, 0, 0, dest);
+
+ return dfn;
+}
+
+static struct _tnl_dynfn *makeX86Attribute2fv( GLcontext *ctx, int dest )
+{
+ TNLcontext *tnl = TNL_CONTEXT(ctx);
+ DFN ( _tnl_x86_Attribute2fv, tnl->vtx.cache.Attribute[2-1], dest );
+
+ FIXUP(dfn->code, 0, 0, dest);
+ FIXUP(dfn->code, 0, 1, 4+dest);
+
+ return dfn;
+}
+
+static struct _tnl_dynfn *makeX86Attribute3fv( GLcontext *ctx, int dest )
+{
+ TNLcontext *tnl = TNL_CONTEXT(ctx);
+ DFN ( _tnl_x86_Attribute3fv, tnl->vtx.cache.Attribute[3-1], dest );
+
+ FIXUP(dfn->code, 0, 0, dest);
+ FIXUP(dfn->code, 0, 1, 4+dest);
+ FIXUP(dfn->code, 0, 2, 8+dest);
+
+ return dfn;
+}
+
+static struct _tnl_dynfn *makeX86Attribute4fv( GLcontext *ctx, int dest )
+{
+ TNLcontext *tnl = TNL_CONTEXT(ctx);
+ DFN ( _tnl_x86_Attribute4fv, tnl->vtx.cache.Attribute[4-1], dest );
+
+ FIXUP(dfn->code, 0, 0, dest);
+ FIXUP(dfn->code, 0, 1, 4+dest);
+ FIXUP(dfn->code, 0, 2, 8+dest);
+ FIXUP(dfn->code, 0, 3, 12+dest);
+
+ return dfn;
+}
+
+
+void _tnl_InitX86Codegen( struct _tnl_dynfn_generators *gen )
+{
+ gen->Vertex[0] = makeX86Vertex1fv;
+ gen->Vertex[1] = makeX86Vertex2fv;
+ gen->Vertex[2] = makeX86Vertex3fv;
+ gen->Vertex[3] = makeX86Vertex4fv;
+ gen->Attribute[0] = makeX86Attribute1fv;
+ gen->Attribute[1] = makeX86Attribute2fv;
+ gen->Attribute[2] = makeX86Attribute3fv;
+ gen->Attribute[3] = makeX86Attribute4fv;
+}
+
+
+#define MKDISP(FUNC, SIZE, ATTR, WARP) \
+do { \
+ char *code; \
+ const char *start = WARP; \
+ const char *end = WARP##_end; \
+ int offset = 0; \
+ code = ALIGN_MALLOC( end - start, 16 ); \
+ memcpy (code, start, end - start); \
+ FIXUP(code, 0, 0, (int)&(TNL_CONTEXT(ctx)->vtx.tabfv[ATTR][SIZE-1]));\
+ *(void **)&vfmt->FUNC = code; \
+} while (0)
+
+
+/* Install the codegen'ed versions of the 2nd level dispatch
+ * functions. We should keep a list and free them in the end...
+ */
+void _tnl_x86_exec_vtxfmt_init( GLcontext *ctx )
+{
+ GLvertexformat *vfmt = &(TNL_CONTEXT(ctx)->exec_vtxfmt);
+
+ MKDISP(Color3f, 3, _TNL_ATTRIB_COLOR0, _tnl_x86_dispatch_attrf3);
+ MKDISP(Color3fv, 3, _TNL_ATTRIB_COLOR0, _tnl_x86_dispatch_attrfv);
+ MKDISP(Color4f, 4, _TNL_ATTRIB_COLOR0, _tnl_x86_dispatch_attrf4);
+ MKDISP(Color4fv, 4, _TNL_ATTRIB_COLOR0, _tnl_x86_dispatch_attrfv);
+ MKDISP(FogCoordfEXT, 1, _TNL_ATTRIB_FOG, _tnl_x86_dispatch_attrf1);
+ MKDISP(FogCoordfvEXT, 1, _TNL_ATTRIB_FOG, _tnl_x86_dispatch_attrfv);
+ MKDISP(Normal3f, 3, _TNL_ATTRIB_NORMAL, _tnl_x86_dispatch_attrf3);
+ MKDISP(Normal3fv, 3, _TNL_ATTRIB_NORMAL, _tnl_x86_dispatch_attrfv);
+ MKDISP(SecondaryColor3fEXT, 3, _TNL_ATTRIB_COLOR1, _tnl_x86_dispatch_attrf3);
+ MKDISP(SecondaryColor3fvEXT,3, _TNL_ATTRIB_COLOR1, _tnl_x86_dispatch_attrfv);
+ MKDISP(TexCoord1f, 1, _TNL_ATTRIB_TEX0, _tnl_x86_dispatch_attrf1);
+ MKDISP(TexCoord1fv, 1, _TNL_ATTRIB_TEX0, _tnl_x86_dispatch_attrfv);
+ MKDISP(TexCoord2f, 2, _TNL_ATTRIB_TEX0, _tnl_x86_dispatch_attrf2);
+ MKDISP(TexCoord2fv, 2, _TNL_ATTRIB_TEX0, _tnl_x86_dispatch_attrfv);
+ MKDISP(TexCoord3f, 3, _TNL_ATTRIB_TEX0, _tnl_x86_dispatch_attrf3);
+ MKDISP(TexCoord3fv, 3, _TNL_ATTRIB_TEX0, _tnl_x86_dispatch_attrfv);
+ MKDISP(TexCoord4f, 4, _TNL_ATTRIB_TEX0, _tnl_x86_dispatch_attrf4);
+ MKDISP(TexCoord4fv, 4, _TNL_ATTRIB_TEX0, _tnl_x86_dispatch_attrfv);
+ MKDISP(Vertex2f, 2, _TNL_ATTRIB_POS, _tnl_x86_dispatch_attrf2);
+ MKDISP(Vertex2fv, 2, _TNL_ATTRIB_POS, _tnl_x86_dispatch_attrfv);
+ MKDISP(Vertex3f, 3, _TNL_ATTRIB_POS, _tnl_x86_dispatch_attrf3);
+ MKDISP(Vertex3fv, 3, _TNL_ATTRIB_POS, _tnl_x86_dispatch_attrfv);
+ MKDISP(Vertex4f, 4, _TNL_ATTRIB_POS, _tnl_x86_dispatch_attrf4);
+ MKDISP(Vertex4fv, 4, _TNL_ATTRIB_POS, _tnl_x86_dispatch_attrfv);
+
+ MKDISP(MultiTexCoord1fARB, 1, _TNL_ATTRIB_TEX0, _tnl_x86_dispatch_multitexcoordf1);
+ MKDISP(MultiTexCoord1fvARB, 1, _TNL_ATTRIB_TEX0, _tnl_x86_dispatch_multitexcoordfv);
+ MKDISP(MultiTexCoord2fARB, 2, _TNL_ATTRIB_TEX0, _tnl_x86_dispatch_multitexcoordf2);
+ MKDISP(MultiTexCoord2fvARB, 2, _TNL_ATTRIB_TEX0, _tnl_x86_dispatch_multitexcoordfv);
+ MKDISP(MultiTexCoord3fARB, 3, _TNL_ATTRIB_TEX0, _tnl_x86_dispatch_multitexcoordf3);
+ MKDISP(MultiTexCoord3fvARB, 3, _TNL_ATTRIB_TEX0, _tnl_x86_dispatch_multitexcoordfv);
+ MKDISP(MultiTexCoord4fARB, 4, _TNL_ATTRIB_TEX0, _tnl_x86_dispatch_multitexcoordf4);
+ MKDISP(MultiTexCoord4fvARB, 4, _TNL_ATTRIB_TEX0, _tnl_x86_dispatch_multitexcoordfv);
+
+ MKDISP(VertexAttrib1fNV, 1, 0, _tnl_x86_dispatch_vertexattribf1);
+ MKDISP(VertexAttrib1fvNV, 1, 0, _tnl_x86_dispatch_vertexattribfv);
+ MKDISP(VertexAttrib2fNV, 2, 0, _tnl_x86_dispatch_vertexattribf2);
+ MKDISP(VertexAttrib2fvNV, 2, 0, _tnl_x86_dispatch_vertexattribfv);
+ MKDISP(VertexAttrib3fNV, 3, 0, _tnl_x86_dispatch_vertexattribf3);
+ MKDISP(VertexAttrib3fvNV, 3, 0, _tnl_x86_dispatch_vertexattribfv);
+ MKDISP(VertexAttrib4fNV, 4, 0, _tnl_x86_dispatch_vertexattribf4);
+ MKDISP(VertexAttrib4fvNV, 4, 0, _tnl_x86_dispatch_vertexattribfv);
+}
+
+
+/* Install the codegen'ed choosers.
+ * We should keep a list and free them in the end...
+ */
+void _tnl_x86choosers( tnl_attrfv_func (*choose)[4],
+ tnl_attrfv_func (*do_choose)( GLuint attr,
+ GLuint sz ))
+{
+ int attr, size;
+
+ for (attr = 0; attr < _TNL_MAX_ATTR_CODEGEN; attr++) {
+ for (size = 0; size < 4; size++) {
+ char *code;
+ const char *start = _tnl_x86_choose_fv;
+ const char *end = _tnl_x86_choose_fv_end;
+ int offset = 0;
+ code = ALIGN_MALLOC( end - start, 16 );
+ memcpy (code, start, end - start);
+ FIXUP(code, 0, 0, attr);
+ FIXUP(code, 0, 1, size + 1);
+ FIXUPREL(code, 0, 2, do_choose);
+ choose[attr][size] = (tnl_attrfv_func)code;
+ }
+ }
+}
+
+#else
+
+void _tnl_InitX86Codegen( struct _tnl_dynfn_generators *gen )
+{
+ (void) gen;
+}
+
+
+void _tnl_x86_exec_vtxfmt_init( GLcontext *ctx )
+{
+ (void) ctx;
+}
+
+
+void _tnl_x86choosers( tnl_attrfv_func (*choose)[4],
+ tnl_attrfv_func (*do_choose)( GLuint attr,
+ GLuint sz ))
+{
+ (void) choose;
+ (void) do_choose;
+}
+
+#endif