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>
32 * Ian Romanick <idr@us.ibm.com>
35 #include <spu_mfcio.h>
37 #include "pipe/p_state.h"
38 #include "pipe/p_shader_tokens.h"
39 #include "util/u_math.h"
40 #include "draw/draw_private.h"
41 #include "draw/draw_context.h"
42 #include "cell/common.h"
43 #include "spu_vertex_shader.h"
48 #define MAX_VERTEX_SIZE ((2 + PIPE_MAX_SHADER_OUTPUTS) * 4 * sizeof(float))
51 #define CLIP_RIGHT_BIT 0x01
52 #define CLIP_LEFT_BIT 0x02
53 #define CLIP_TOP_BIT 0x04
54 #define CLIP_BOTTOM_BIT 0x08
55 #define CLIP_FAR_BIT 0x10
56 #define CLIP_NEAR_BIT 0x20
60 dot4(const float *a, const float *b)
68 static INLINE unsigned
69 compute_clipmask(const float *clip, /*const*/ float plane[][4], unsigned nr)
74 /* Do the hardwired planes first:
76 if (-clip[0] + clip[3] < 0) mask |= CLIP_RIGHT_BIT;
77 if ( clip[0] + clip[3] < 0) mask |= CLIP_LEFT_BIT;
78 if (-clip[1] + clip[3] < 0) mask |= CLIP_TOP_BIT;
79 if ( clip[1] + clip[3] < 0) mask |= CLIP_BOTTOM_BIT;
80 if (-clip[2] + clip[3] < 0) mask |= CLIP_FAR_BIT;
81 if ( clip[2] + clip[3] < 0) mask |= CLIP_NEAR_BIT;
83 /* Followed by any remaining ones:
85 for (i = 6; i < nr; i++) {
86 if (dot4(clip, plane[i]) < 0)
95 * Transform vertices with the current vertex program/shader
96 * Up to four vertices can be shaded at a time.
97 * \param vbuffer the input vertex data
98 * \param elts indexes of four input vertices
99 * \param count number of vertices to shade [1..4]
100 * \param vOut array of pointers to four output vertices
103 run_vertex_program(struct spu_vs_context *draw,
104 unsigned elts[4], unsigned count,
105 const uint64_t *vOut)
107 struct spu_exec_machine *machine = &draw->machine;
110 PIPE_ALIGN_VAR(16) struct spu_exec_vector inputs[PIPE_MAX_ATTRIBS];
111 PIPE_ALIGN_VAR(16) struct spu_exec_vector outputs[PIPE_MAX_ATTRIBS];
112 const float *scale = draw->viewport.scale;
113 const float *trans = draw->viewport.translate;
117 machine->Processor = TGSI_PROCESSOR_VERTEX;
119 ASSERT_ALIGN16(draw->constants);
120 machine->Consts = (float (*)[4]) draw->constants;
122 machine->Inputs = inputs;
123 machine->Outputs = outputs;
125 spu_vertex_fetch( draw, machine, elts, count );
128 spu_exec_machine_run( machine );
131 /* store machine results */
132 for (j = 0; j < count; j++) {
136 unsigned char buffer[sizeof(struct vertex_header)
138 struct vertex_header *const tmpOut =
139 (struct vertex_header *) buffer;
140 const unsigned vert_size = ROUNDUP16(sizeof(struct vertex_header)
142 * draw->num_vs_outputs));
144 mfc_get(tmpOut, vOut[j], vert_size, TAG_VERTEX_BUFFER, 0, 0);
145 wait_on_mask(1 << TAG_VERTEX_BUFFER);
148 /* Handle attr[0] (position) specially:
150 * XXX: Computing the clipmask should be done in the vertex
151 * program as a set of DP4 instructions appended to the
152 * user-provided code.
154 x = tmpOut->clip[0] = machine->Outputs[0].xyzw[0].f[j];
155 y = tmpOut->clip[1] = machine->Outputs[0].xyzw[1].f[j];
156 z = tmpOut->clip[2] = machine->Outputs[0].xyzw[2].f[j];
157 w = tmpOut->clip[3] = machine->Outputs[0].xyzw[3].f[j];
159 tmpOut->clipmask = compute_clipmask(tmpOut->clip, draw->plane,
161 tmpOut->edgeflag = 1;
169 /* Viewport mapping */
170 tmpOut->data[0][0] = x * scale[0] + trans[0];
171 tmpOut->data[0][1] = y * scale[1] + trans[1];
172 tmpOut->data[0][2] = z * scale[2] + trans[2];
173 tmpOut->data[0][3] = w;
175 /* Remaining attributes are packed into sequential post-transform
176 * vertex attrib slots.
178 for (slot = 1; slot < draw->num_vs_outputs; slot++) {
179 tmpOut->data[slot][0] = machine->Outputs[slot].xyzw[0].f[j];
180 tmpOut->data[slot][1] = machine->Outputs[slot].xyzw[1].f[j];
181 tmpOut->data[slot][2] = machine->Outputs[slot].xyzw[2].f[j];
182 tmpOut->data[slot][3] = machine->Outputs[slot].xyzw[3].f[j];
185 mfc_put(tmpOut, vOut[j], vert_size, TAG_VERTEX_BUFFER, 0, 0);
186 } /* loop over vertices */
190 PIPE_ALIGN_VAR(16) unsigned char
191 immediates[(sizeof(float) * 4 * TGSI_EXEC_NUM_IMMEDIATES) + 32];
195 spu_bind_vertex_shader(struct spu_vs_context *draw,
196 struct cell_shader_info *vs)
198 const unsigned immediate_addr = vs->immediates;
199 const unsigned immediate_size =
200 ROUNDUP16((sizeof(float) * 4 * vs->num_immediates)
201 + (immediate_addr & 0x0f));
204 mfc_get(immediates, immediate_addr & ~0x0f, immediate_size,
205 TAG_VERTEX_BUFFER, 0, 0);
207 draw->machine.Instructions = (struct tgsi_full_instruction *)
209 draw->machine.NumInstructions = vs->num_instructions;
211 draw->machine.Declarations = (struct tgsi_full_declaration *)
213 draw->machine.NumDeclarations = vs->num_declarations;
215 draw->num_vs_outputs = vs->num_outputs;
217 /* specify the shader to interpret/execute */
218 spu_exec_machine_init(&draw->machine,
223 wait_on_mask(1 << TAG_VERTEX_BUFFER);
225 (void) memcpy(& draw->machine.Imms, &immediates[immediate_addr & 0x0f],
226 sizeof(float) * 4 * vs->num_immediates);
231 spu_execute_vertex_shader(struct spu_vs_context *draw,
232 const struct cell_command_vs *vs)
236 (void) memcpy(draw->plane, vs->plane, sizeof(float) * 4 * vs->nr_planes);
237 draw->nr_planes = vs->nr_planes;
238 draw->vertex_fetch.nr_attrs = vs->nr_attrs;
240 for (i = 0; i < vs->num_elts; i += 4) {
241 const unsigned batch_size = MIN2(vs->num_elts - i, 4);
243 run_vertex_program(draw, & vs->elts[i], batch_size, &vs->vOut[i]);