Merge branch '7.8'
[mesa.git] / src / gallium / drivers / cell / spu / spu_vertex_fetch.c
1 /**************************************************************************
2 *
3 * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas.
4 * (C) Copyright IBM Corporation 2008
5 * All Rights Reserved.
6 *
7 * Permission is hereby granted, free of charge, to any person obtaining a
8 * copy of this software and associated documentation files (the
9 * "Software"), to deal in the Software without restriction, including
10 * without limitation the rights to use, copy, modify, merge, publish,
11 * distribute, sub license, and/or sell copies of the Software, and to
12 * permit persons to whom the Software is furnished to do so, subject to
13 * the following conditions:
14 *
15 * The above copyright notice and this permission notice (including the
16 * next paragraph) shall be included in all copies or substantial portions
17 * of the Software.
18 *
19 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
20 * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
21 * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
22 * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
23 * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
24 * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
25 * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
26 *
27 **************************************************************************/
28
29 /*
30 * Authors:
31 * Keith Whitwell <keith@tungstengraphics.com>
32 * Ian Romanick <idr@us.ibm.com>
33 */
34
35 #include "pipe/p_state.h"
36 #include "pipe/p_shader_tokens.h"
37 #include "spu_exec.h"
38 #include "spu_vertex_shader.h"
39 #include "spu_main.h"
40 #include "spu_dcache.h"
41
42 typedef void (*spu_fetch_func)(qword *out, const qword *in,
43 const qword *shuffle_data);
44
45
46 PIPE_ALIGN_VAR(16) static const qword
47 fetch_shuffle_data[5] = {
48 /* Shuffle used by CVT_64_FLOAT
49 */
50 {
51 0x00, 0x01, 0x02, 0x03, 0x10, 0x11, 0x12, 0x13,
52 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80,
53 },
54
55 /* Shuffle used by CVT_8_USCALED and CVT_8_SSCALED
56 */
57 {
58 0x00, 0x80, 0x80, 0x80, 0x01, 0x80, 0x80, 0x80,
59 0x02, 0x80, 0x80, 0x80, 0x03, 0x80, 0x80, 0x80,
60 },
61
62 /* Shuffle used by CVT_16_USCALED and CVT_16_SSCALED
63 */
64 {
65 0x00, 0x01, 0x80, 0x80, 0x02, 0x03, 0x80, 0x80,
66 0x04, 0x05, 0x80, 0x80, 0x06, 0x07, 0x80, 0x80,
67 },
68
69 /* High value shuffle used by trans4x4.
70 */
71 {
72 0x00, 0x01, 0x02, 0x03, 0x10, 0x11, 0x12, 0x13,
73 0x04, 0x05, 0x06, 0x07, 0x14, 0x15, 0x16, 0x17
74 },
75
76 /* Low value shuffle used by trans4x4.
77 */
78 {
79 0x08, 0x09, 0x0A, 0x0B, 0x18, 0x19, 0x1A, 0x1B,
80 0x0C, 0x0D, 0x0E, 0x0F, 0x1C, 0x1D, 0x1E, 0x1F
81 }
82 };
83
84
85 /**
86 * Fetch vertex attributes for 'count' vertices.
87 */
88 static void generic_vertex_fetch(struct spu_vs_context *draw,
89 struct spu_exec_machine *machine,
90 const unsigned *elts,
91 unsigned count)
92 {
93 unsigned nr_attrs = draw->vertex_fetch.nr_attrs;
94 unsigned attr;
95
96 ASSERT(count <= 4);
97
98 #if DRAW_DBG
99 printf("SPU: %s count = %u, nr_attrs = %u\n",
100 __FUNCTION__, count, nr_attrs);
101 #endif
102
103 /* loop over vertex attributes (vertex shader inputs)
104 */
105 for (attr = 0; attr < nr_attrs; attr++) {
106 const unsigned pitch = draw->vertex_fetch.pitch[attr];
107 const uint64_t src = draw->vertex_fetch.src_ptr[attr];
108 const spu_fetch_func fetch = (spu_fetch_func)
109 (draw->vertex_fetch.code + draw->vertex_fetch.code_offset[attr]);
110 unsigned i;
111 unsigned idx;
112 const unsigned bytes_per_entry = draw->vertex_fetch.size[attr];
113 const unsigned quads_per_entry = (bytes_per_entry + 15) / 16;
114 PIPE_ALIGN_VAR(16) qword in[2 * 4];
115
116
117 /* Fetch four attributes for four vertices.
118 */
119 idx = 0;
120 for (i = 0; i < count; i++) {
121 const uint64_t addr = src + (elts[i] * pitch);
122
123 #if DRAW_DBG
124 printf("SPU: fetching = 0x%llx\n", addr);
125 #endif
126
127 spu_dcache_fetch_unaligned(& in[idx], addr, bytes_per_entry);
128 idx += quads_per_entry;
129 }
130
131 /* Be nice and zero out any missing vertices.
132 */
133 (void) memset(& in[idx], 0, (8 - idx) * sizeof(qword));
134
135
136 /* Convert all 4 vertices to vectors of float.
137 */
138 (*fetch)(&machine->Inputs[attr].xyzw[0].q, in, fetch_shuffle_data);
139 }
140 }
141
142
143 void spu_update_vertex_fetch( struct spu_vs_context *draw )
144 {
145 draw->vertex_fetch.fetch_func = generic_vertex_fetch;
146 }