1 /**************************************************************************
3 * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas.
6 * Permission is hereby granted, free of charge, to any person obtaining a
7 * copy of this software and associated documentation files (the
8 * "Software"), to deal in the Software without restriction, including
9 * without limitation the rights to use, copy, modify, merge, publish,
10 * distribute, sub license, and/or sell copies of the Software, and to
11 * permit persons to whom the Software is furnished to do so, subject to
12 * the following conditions:
14 * The above copyright notice and this permission notice (including the
15 * next paragraph) shall be included in all copies or substantial portions
18 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
19 * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
20 * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
21 * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
22 * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
23 * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
24 * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
26 **************************************************************************/
30 * Keith Whitwell <keith@tungstengraphics.com>
34 #include "pipe/p_util.h"
35 #include "draw_private.h"
36 #include "draw_context.h"
37 #include "draw_vertex.h"
39 #include "pipe/tgsi/exec/tgsi_core.h"
41 static INLINE
unsigned
42 compute_clipmask(float cx
, float cy
, float cz
, float cw
)
46 if (-cx
+ cw
< 0) mask
|= CLIP_RIGHT_BIT
;
47 if ( cx
+ cw
< 0) mask
|= CLIP_LEFT_BIT
;
48 if (-cy
+ cw
< 0) mask
|= CLIP_TOP_BIT
;
49 if ( cy
+ cw
< 0) mask
|= CLIP_BOTTOM_BIT
;
50 if (-cz
+ cw
< 0) mask
|= CLIP_FAR_BIT
;
51 if ( cz
+ cw
< 0) mask
|= CLIP_NEAR_BIT
;
59 #if !defined(XSTDCALL)
61 #define XSTDCALL __stdcall
67 #if defined(USE_X86_ASM) || defined(SLANG_X86)
68 typedef void (XSTDCALL
*sse2_function
)(
69 const struct tgsi_exec_vector
*input
,
70 struct tgsi_exec_vector
*output
,
72 struct tgsi_exec_vector
*temporary
);
76 * Transform vertices with the current vertex program/shader
77 * Up to four vertices can be shaded at a time.
78 * \param vbuffer the input vertex data
79 * \param elts indexes of four input vertices
80 * \param count number of vertices to shade [1..4]
81 * \param vOut array of pointers to four output vertices
84 run_vertex_program(struct draw_context
*draw
,
85 unsigned elts
[4], unsigned count
,
86 struct vertex_header
*vOut
[])
88 struct tgsi_exec_machine machine
;
91 ALIGN16_DECL(struct tgsi_exec_vector
, inputs
, PIPE_ATTRIB_MAX
);
92 ALIGN16_DECL(struct tgsi_exec_vector
, outputs
, PIPE_ATTRIB_MAX
);
93 const float *scale
= draw
->viewport
.scale
;
94 const float *trans
= draw
->viewport
.translate
;
97 assert(draw
->vertex_shader
.output_semantic_name
[0]
98 == TGSI_SEMANTIC_POSITION
);
101 memset( &machine
, 0, sizeof( machine
) );
104 /* init machine state */
105 tgsi_exec_machine_init(&machine
,
106 draw
->vertex_shader
.tokens
,
110 /* Consts does not require 16 byte alignment. */
111 machine
.Consts
= (float (*)[4]) draw
->mapped_constants
;
113 machine
.Inputs
= ALIGN16_ASSIGN(inputs
);
114 machine
.Outputs
= ALIGN16_ASSIGN(outputs
);
116 draw_vertex_fetch( draw
, &machine
, elts
, count
);
119 if( draw
->vertex_shader
.executable
!= NULL
) {
120 #if defined(USE_X86_ASM) || defined(SLANG_X86)
121 sse2_function func
= (sse2_function
) draw
->vertex_shader
.executable
;
132 tgsi_exec_machine_run( &machine
);
136 /* store machine results */
137 for (j
= 0; j
< count
; j
++) {
141 /* Handle attr[0] (position) specially: */
142 x
= vOut
[j
]->clip
[0] = machine
.Outputs
[0].xyzw
[0].f
[j
];
143 y
= vOut
[j
]->clip
[1] = machine
.Outputs
[0].xyzw
[1].f
[j
];
144 z
= vOut
[j
]->clip
[2] = machine
.Outputs
[0].xyzw
[2].f
[j
];
145 w
= vOut
[j
]->clip
[3] = machine
.Outputs
[0].xyzw
[3].f
[j
];
147 vOut
[j
]->clipmask
= compute_clipmask(x
, y
, z
, w
) | draw
->user_clipmask
;
148 vOut
[j
]->edgeflag
= 1;
156 /* Viewport mapping */
157 vOut
[j
]->data
[0][0] = x
* scale
[0] + trans
[0];
158 vOut
[j
]->data
[0][1] = y
* scale
[1] + trans
[1];
159 vOut
[j
]->data
[0][2] = z
* scale
[2] + trans
[2];
160 vOut
[j
]->data
[0][3] = w
;
162 /* Remaining attributes are packed into sequential post-transform
163 * vertex attrib slots.
164 * Skip 0 since we just did it above.
165 * Subtract two because of the VERTEX_HEADER, CLIP_POS attribs.
167 for (slot
= 1; slot
< draw
->vertex_info
.num_attribs
- 2; slot
++) {
168 vOut
[j
]->data
[slot
][0] = machine
.Outputs
[slot
].xyzw
[0].f
[j
];
169 vOut
[j
]->data
[slot
][1] = machine
.Outputs
[slot
].xyzw
[1].f
[j
];
170 vOut
[j
]->data
[slot
][2] = machine
.Outputs
[slot
].xyzw
[2].f
[j
];
171 vOut
[j
]->data
[slot
][3] = machine
.Outputs
[slot
].xyzw
[3].f
[j
];
173 printf("output %d: %f %f %f %f\n", slot,
174 vOut[j]->data[slot][0],
175 vOut[j]->data[slot][1],
176 vOut[j]->data[slot][2],
177 vOut[j]->data[slot][3]);
180 } /* loop over vertices */
185 * Called by the draw module when the vertx cache needs to be flushed.
186 * This involves running the vertex shader.
188 void draw_vertex_shader_queue_flush( struct draw_context
*draw
)
192 // fprintf(stderr, " q(%d) ", draw->vs.queue_nr );
194 /* run vertex shader on vertex cache entries, four per invokation */
195 for (i
= 0; i
< draw
->vs
.queue_nr
; i
+= 4) {
196 struct vertex_header
*dests
[4];
200 for (j
= 0; j
< 4; j
++) {
201 elts
[j
] = draw
->vs
.queue
[i
+ j
].elt
;
202 dests
[j
] = draw
->vs
.queue
[i
+ j
].dest
;
205 n
= MIN2(4, draw
->vs
.queue_nr
- i
);
209 run_vertex_program(draw
, elts
, n
, dests
);
212 draw
->vs
.queue_nr
= 0;