Merge remote branch 'origin/master' into pipe-video
[mesa.git] / src / gallium / drivers / nvc0 / nvc0_shader_state.c
1 /*
2 * Copyright 2010 Christoph Bumiller
3 *
4 * Permission is hereby granted, free of charge, to any person obtaining a
5 * copy of this software and associated documentation files (the "Software"),
6 * to deal in the Software without restriction, including without limitation
7 * the rights to use, copy, modify, merge, publish, distribute, sublicense,
8 * and/or sell copies of the Software, and to permit persons to whom the
9 * Software is furnished to do so, subject to the following conditions:
10 *
11 * The above copyright notice and this permission notice shall be included in
12 * all copies or substantial portions of the Software.
13 *
14 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
15 * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
16 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
17 * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
18 * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
19 * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
20 * SOFTWARE.
21 */
22
23 #include "pipe/p_context.h"
24 #include "pipe/p_defines.h"
25 #include "pipe/p_state.h"
26 #include "util/u_inlines.h"
27
28 #include "nvc0_context.h"
29
30 static INLINE void
31 nvc0_program_update_context_state(struct nvc0_context *nvc0,
32 struct nvc0_program *prog, int stage)
33 {
34 if (prog->hdr[1])
35 nvc0->state.tls_required |= 1 << stage;
36 else
37 nvc0->state.tls_required &= ~(1 << stage);
38 }
39
40 static boolean
41 nvc0_program_validate(struct nvc0_context *nvc0, struct nvc0_program *prog)
42 {
43 int ret;
44 unsigned size;
45
46 if (prog->translated)
47 return TRUE;
48
49 prog->translated = nvc0_program_translate(prog);
50 if (!prog->translated)
51 return FALSE;
52
53 size = align(prog->code_size + NVC0_SHADER_HEADER_SIZE, 0x100);
54
55 ret = nouveau_resource_alloc(nvc0->screen->text_heap, size, prog,
56 &prog->res);
57 if (ret)
58 return FALSE;
59
60 prog->code_base = prog->res->start;
61
62 nvc0_m2mf_push_linear(&nvc0->base, nvc0->screen->text, prog->code_base,
63 NOUVEAU_BO_VRAM, NVC0_SHADER_HEADER_SIZE, prog->hdr);
64 nvc0_m2mf_push_linear(&nvc0->base, nvc0->screen->text,
65 prog->code_base + NVC0_SHADER_HEADER_SIZE,
66 NOUVEAU_BO_VRAM, prog->code_size, prog->code);
67
68 BEGIN_RING(nvc0->screen->base.channel, RING_3D(MEM_BARRIER), 1);
69 OUT_RING (nvc0->screen->base.channel, 0x1111);
70
71 return TRUE;
72 }
73
74 void
75 nvc0_vertprog_validate(struct nvc0_context *nvc0)
76 {
77 struct nouveau_channel *chan = nvc0->screen->base.channel;
78 struct nvc0_program *vp = nvc0->vertprog;
79
80 if (nvc0->clip.nr > vp->vp.num_ucps) {
81 assert(nvc0->clip.nr <= 6);
82 vp->vp.num_ucps = 6;
83
84 if (vp->translated)
85 nvc0_program_destroy(nvc0, vp);
86 }
87
88 if (!nvc0_program_validate(nvc0, vp))
89 return;
90 nvc0_program_update_context_state(nvc0, vp, 0);
91
92 BEGIN_RING(chan, RING_3D(SP_SELECT(1)), 2);
93 OUT_RING (chan, 0x11);
94 OUT_RING (chan, vp->code_base);
95 BEGIN_RING(chan, RING_3D(SP_GPR_ALLOC(1)), 1);
96 OUT_RING (chan, vp->max_gpr);
97
98 // BEGIN_RING(chan, RING_3D_(0x163c), 1);
99 // OUT_RING (chan, 0);
100 }
101
102 void
103 nvc0_fragprog_validate(struct nvc0_context *nvc0)
104 {
105 struct nouveau_channel *chan = nvc0->screen->base.channel;
106 struct nvc0_program *fp = nvc0->fragprog;
107
108 if (!nvc0_program_validate(nvc0, fp))
109 return;
110 nvc0_program_update_context_state(nvc0, fp, 4);
111
112 BEGIN_RING(chan, RING_3D(SP_SELECT(5)), 2);
113 OUT_RING (chan, 0x51);
114 OUT_RING (chan, fp->code_base);
115 BEGIN_RING(chan, RING_3D(SP_GPR_ALLOC(5)), 1);
116 OUT_RING (chan, fp->max_gpr);
117
118 BEGIN_RING(chan, RING_3D_(0x0360), 2);
119 OUT_RING (chan, 0x20164010);
120 OUT_RING (chan, 0x20);
121 BEGIN_RING(chan, RING_3D_(0x196c), 1);
122 OUT_RING (chan, fp->flags[0]);
123 }
124
125 void
126 nvc0_tctlprog_validate(struct nvc0_context *nvc0)
127 {
128 struct nouveau_channel *chan = nvc0->screen->base.channel;
129 struct nvc0_program *tp = nvc0->tctlprog;
130
131 if (!tp) {
132 BEGIN_RING(chan, RING_3D(SP_SELECT(2)), 1);
133 OUT_RING (chan, 0x20);
134 return;
135 }
136 if (!nvc0_program_validate(nvc0, tp))
137 return;
138 nvc0_program_update_context_state(nvc0, tp, 1);
139
140 BEGIN_RING(chan, RING_3D(SP_SELECT(2)), 2);
141 OUT_RING (chan, 0x21);
142 OUT_RING (chan, tp->code_base);
143 BEGIN_RING(chan, RING_3D(SP_GPR_ALLOC(2)), 1);
144 OUT_RING (chan, tp->max_gpr);
145 }
146
147 void
148 nvc0_tevlprog_validate(struct nvc0_context *nvc0)
149 {
150 struct nouveau_channel *chan = nvc0->screen->base.channel;
151 struct nvc0_program *tp = nvc0->tevlprog;
152
153 if (!tp) {
154 BEGIN_RING(chan, RING_3D(TEP_SELECT), 1);
155 OUT_RING (chan, 0x30);
156 return;
157 }
158 if (!nvc0_program_validate(nvc0, tp))
159 return;
160 nvc0_program_update_context_state(nvc0, tp, 2);
161
162 BEGIN_RING(chan, RING_3D(TEP_SELECT), 1);
163 OUT_RING (chan, 0x31);
164 BEGIN_RING(chan, RING_3D(SP_START_ID(3)), 1);
165 OUT_RING (chan, tp->code_base);
166 BEGIN_RING(chan, RING_3D(SP_GPR_ALLOC(3)), 1);
167 OUT_RING (chan, tp->max_gpr);
168 }
169
170 void
171 nvc0_gmtyprog_validate(struct nvc0_context *nvc0)
172 {
173 struct nouveau_channel *chan = nvc0->screen->base.channel;
174 struct nvc0_program *gp = nvc0->gmtyprog;
175
176 if (!gp) {
177 BEGIN_RING(chan, RING_3D(GP_SELECT), 1);
178 OUT_RING (chan, 0x40);
179 return;
180 }
181 if (!nvc0_program_validate(nvc0, gp))
182 return;
183 nvc0_program_update_context_state(nvc0, gp, 3);
184
185 BEGIN_RING(chan, RING_3D(GP_SELECT), 1);
186 OUT_RING (chan, 0x41);
187 BEGIN_RING(chan, RING_3D(SP_START_ID(4)), 1);
188 OUT_RING (chan, gp->code_base);
189 BEGIN_RING(chan, RING_3D(SP_GPR_ALLOC(4)), 1);
190 OUT_RING (chan, gp->max_gpr);
191 }
192
193 /* It's *is* kind of shader related. We need to inspect the program
194 * to get the output locations right.
195 */
196 void
197 nvc0_tfb_validate(struct nvc0_context *nvc0)
198 {
199 struct nouveau_channel *chan = nvc0->screen->base.channel;
200 struct nvc0_program *vp;
201 struct nvc0_transform_feedback_state *tfb = nvc0->tfb;
202 int b;
203
204 BEGIN_RING(chan, RING_3D(TFB_ENABLE), 1);
205 if (!tfb) {
206 OUT_RING(chan, 0);
207 return;
208 }
209 OUT_RING(chan, 1);
210
211 vp = nvc0->vertprog ? nvc0->vertprog : nvc0->gmtyprog;
212
213 for (b = 0; b < nvc0->num_tfbbufs; ++b) {
214 uint8_t idx, var[128];
215 int i, n;
216 struct nv04_resource *buf = nv04_resource(nvc0->tfbbuf[b]);
217
218 BEGIN_RING(chan, RING_3D(TFB_BUFFER_ENABLE(b)), 5);
219 OUT_RING (chan, 1);
220 OUT_RESRCh(chan, buf, nvc0->tfb_offset[b], NOUVEAU_BO_WR);
221 OUT_RESRCl(chan, buf, nvc0->tfb_offset[b], NOUVEAU_BO_WR);
222 OUT_RING (chan, buf->base.width0 - nvc0->tfb_offset[b]);
223 OUT_RING (chan, 0); /* TFB_PRIMITIVE_ID <- offset ? */
224
225 if (!(nvc0->dirty & NVC0_NEW_TFB))
226 continue;
227
228 BEGIN_RING(chan, RING_3D(TFB_UNK07X0(b)), 3);
229 OUT_RING (chan, 0);
230 OUT_RING (chan, tfb->varying_count[b]);
231 OUT_RING (chan, tfb->stride[b]);
232
233 n = b ? tfb->varying_count[b - 1] : 0;
234 i = 0;
235 for (; i < tfb->varying_count[b]; ++i) {
236 idx = tfb->varying_index[n + i];
237 var[i] = vp->vp.out_pos[idx >> 2] + (idx & 3);
238 }
239 for (; i & 3; ++i)
240 var[i] = 0;
241
242 BEGIN_RING(chan, RING_3D(TFB_VARYING_LOCS(b, 0)), i / 4);
243 OUT_RINGp (chan, var, i / 4);
244 }
245 for (; b < 4; ++b)
246 IMMED_RING(chan, RING_3D(TFB_BUFFER_ENABLE(b)), 0);
247 }