1/*
2 * Copyright 2012 Red Hat Inc.
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 OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR
18 * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
19 * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
20 * OTHER DEALINGS IN THE SOFTWARE.
21 *
22 * Authors: Ben Skeggs
23 *
24 */
25
26#include "draw/draw_context.h"
27#include "util/u_dynarray.h"
28#include "tgsi/tgsi_parse.h"
29
30#include "nv_object.xml.h"
31#include "nv30/nv30-40_3d.xml.h"
32#include "nv30/nv30_context.h"
33#include "nv30/nvfx_shader.h"
34#include "nv30/nv30_state.h"
35
36static void
37nv30_vertprog_destroy(struct nv30_vertprog *vp)
38{
39   util_dynarray_fini(&vp->branch_relocs);
40   nouveau_heap_free(&vp->exec);
41   FREE(vp->insns);
42   vp->insns = NULL;
43   vp->nr_insns = 0;
44
45   util_dynarray_fini(&vp->const_relocs);
46   nouveau_heap_free(&vp->data);
47   FREE(vp->consts);
48   vp->consts = NULL;
49   vp->nr_consts = 0;
50
51   vp->translated = false;
52}
53
54void
55nv30_vertprog_validate(struct nv30_context *nv30)
56{
57   struct nouveau_pushbuf *push = nv30->base.pushbuf;
58   struct nouveau_object *eng3d = nv30->screen->eng3d;
59   struct nv30_vertprog *vp = nv30->vertprog.program;
60   struct nv30_fragprog *fp = nv30->fragprog.program;
61   bool upload_code = false;
62   bool upload_data = false;
63   unsigned i;
64
65   if (nv30->dirty & NV30_NEW_FRAGPROG) {
66      if (memcmp(vp->texcoord, fp->texcoord, sizeof(vp->texcoord))) {
67         if (vp->translated)
68            nv30_vertprog_destroy(vp);
69         memcpy(vp->texcoord, fp->texcoord, sizeof(vp->texcoord));
70      }
71   }
72
73   if (nv30->rast && nv30->rast->pipe.clip_plane_enable != vp->enabled_ucps) {
74      vp->enabled_ucps = nv30->rast->pipe.clip_plane_enable;
75      if (vp->translated)
76         nv30_vertprog_destroy(vp);
77   }
78
79   if (!vp->translated) {
80      vp->translated = _nvfx_vertprog_translate(eng3d->oclass, vp);
81      if (!vp->translated) {
82         nv30->draw_flags |= NV30_NEW_VERTPROG;
83         return;
84      }
85      nv30->dirty |= NV30_NEW_VERTPROG;
86   }
87
88   if (!vp->exec) {
89      struct nouveau_heap *heap = nv30->screen->vp_exec_heap;
90      struct nv30_shader_reloc *reloc = vp->branch_relocs.data;
91      unsigned nr_reloc = vp->branch_relocs.size / sizeof(*reloc);
92      uint32_t *inst, target;
93
94      if (nouveau_heap_alloc(heap, vp->nr_insns, &vp->exec, &vp->exec)) {
95         while (heap->next && heap->size < vp->nr_insns) {
96            struct nouveau_heap **evict = heap->next->priv;
97            nouveau_heap_free(evict);
98         }
99
100         if (nouveau_heap_alloc(heap, vp->nr_insns, &vp->exec, &vp->exec)) {
101            nv30->draw_flags |= NV30_NEW_VERTPROG;
102            return;
103         }
104      }
105
106      if (eng3d->oclass < NV40_3D_CLASS) {
107         while (nr_reloc--) {
108            inst     = vp->insns[reloc->location].data;
109            target   = vp->exec->start + reloc->target;
110
111            inst[2] &= ~0x000007fc;
112            inst[2] |= target << 2;
113            reloc++;
114         }
115      } else {
116         while (nr_reloc--) {
117            inst     = vp->insns[reloc->location].data;
118            target   = vp->exec->start + reloc->target;
119
120            inst[2] &= ~0x0000003f;
121            inst[2] |= target >> 3;
122            inst[3] &= ~0xe0000000;
123            inst[3] |= target << 29;
124            reloc++;
125         }
126      }
127
128      upload_code = true;
129   }
130
131   if (vp->nr_consts && !vp->data) {
132      struct nouveau_heap *heap = nv30->screen->vp_data_heap;
133      struct nv30_shader_reloc *reloc = vp->const_relocs.data;
134      unsigned nr_reloc = vp->const_relocs.size / sizeof(*reloc);
135      uint32_t *inst, target;
136
137      if (nouveau_heap_alloc(heap, vp->nr_consts, vp, &vp->data)) {
138         while (heap->next && heap->size < vp->nr_consts) {
139            struct nv30_vertprog *evp = heap->next->priv;
140            nouveau_heap_free(&evp->data);
141         }
142
143         if (nouveau_heap_alloc(heap, vp->nr_consts, vp, &vp->data)) {
144            nv30->draw_flags |= NV30_NEW_VERTPROG;
145            return;
146         }
147      }
148
149      if (eng3d->oclass < NV40_3D_CLASS) {
150         while (nr_reloc--) {
151            inst     = vp->insns[reloc->location].data;
152            target   = vp->data->start + reloc->target;
153
154            inst[1] &= ~0x0007fc000;
155            inst[1] |= (target & 0x1ff) << 14;
156            reloc++;
157         }
158      } else {
159         while (nr_reloc--) {
160            inst     = vp->insns[reloc->location].data;
161            target   = vp->data->start + reloc->target;
162
163            inst[1] &= ~0x0001ff000;
164            inst[1] |= (target & 0x1ff) << 12;
165            reloc++;
166         }
167      }
168
169      upload_code = true;
170      upload_data = true;
171   }
172
173   if (vp->nr_consts) {
174      struct nv04_resource *res = nv04_resource(nv30->vertprog.constbuf);
175
176      for (i = 0; i < vp->nr_consts; i++) {
177         struct nv30_vertprog_data *data = &vp->consts[i];
178
179         if (data->index < 0) {
180            if (!upload_data)
181               continue;
182         } else {
183            float *constbuf = (float *)res->data;
184            if (!upload_data &&
185                !memcmp(data->value, &constbuf[data->index * 4], 16))
186               continue;
187            memcpy(data->value, &constbuf[data->index * 4], 16);
188         }
189
190         BEGIN_NV04(push, NV30_3D(VP_UPLOAD_CONST_ID), 5);
191         PUSH_DATA (push, vp->data->start + i);
192         PUSH_DATAp(push, data->value, 4);
193      }
194   }
195
196   if (upload_code) {
197      BEGIN_NV04(push, NV30_3D(VP_UPLOAD_FROM_ID), 1);
198      PUSH_DATA (push, vp->exec->start);
199      for (i = 0; i < vp->nr_insns; i++) {
200         BEGIN_NV04(push, NV30_3D(VP_UPLOAD_INST(0)), 4);
201         PUSH_DATAp(push, vp->insns[i].data, 4);
202      }
203   }
204
205   if (nv30->dirty & (NV30_NEW_VERTPROG | NV30_NEW_FRAGPROG)) {
206      BEGIN_NV04(push, NV30_3D(VP_START_FROM_ID), 1);
207      PUSH_DATA (push, vp->exec->start);
208      if (eng3d->oclass < NV40_3D_CLASS) {
209         BEGIN_NV04(push, NV30_3D(ENGINE), 1);
210         PUSH_DATA (push, 0x00000013); /* vp instead of ff, somehow */
211      } else {
212         BEGIN_NV04(push, NV40_3D(VP_ATTRIB_EN), 2);
213         PUSH_DATA (push, vp->ir);
214         PUSH_DATA (push, vp->or | fp->vp_or);
215         BEGIN_NV04(push, NV30_3D(ENGINE), 1);
216         PUSH_DATA (push, 0x00000011);
217      }
218   }
219}
220
221static void *
222nv30_vp_state_create(struct pipe_context *pipe,
223                     const struct pipe_shader_state *cso)
224{
225   struct nv30_vertprog *vp = CALLOC_STRUCT(nv30_vertprog);
226   if (!vp)
227      return NULL;
228
229   vp->pipe.tokens = tgsi_dup_tokens(cso->tokens);
230   tgsi_scan_shader(vp->pipe.tokens, &vp->info);
231   return vp;
232}
233
234static void
235nv30_vp_state_delete(struct pipe_context *pipe, void *hwcso)
236{
237   struct nv30_vertprog *vp = hwcso;
238
239   if (vp->translated)
240      nv30_vertprog_destroy(vp);
241
242   if (vp->draw)
243      draw_delete_vertex_shader(nv30_context(pipe)->draw, vp->draw);
244
245   FREE((void *)vp->pipe.tokens);
246   FREE(vp);
247}
248
249static void
250nv30_vp_state_bind(struct pipe_context *pipe, void *hwcso)
251{
252   struct nv30_context *nv30 = nv30_context(pipe);
253
254   nv30->vertprog.program = hwcso;
255   nv30->dirty |= NV30_NEW_VERTPROG;
256}
257
258void
259nv30_vertprog_init(struct pipe_context *pipe)
260{
261   pipe->create_vs_state = nv30_vp_state_create;
262   pipe->bind_vs_state = nv30_vp_state_bind;
263   pipe->delete_vs_state = nv30_vp_state_delete;
264}
265