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#include "nir/nir_to_tgsi.h"
30
31#include "nv_object.xml.h"
32#include "nv30/nv30-40_3d.xml.h"
33#include "nv30/nv30_context.h"
34#include "nv30/nvfx_shader.h"
35#include "nv30/nv30_state.h"
36
37static void
38nv30_vertprog_destroy(struct nv30_vertprog *vp)
39{
40   util_dynarray_fini(&vp->branch_relocs);
41   nouveau_heap_free(&vp->exec);
42   FREE(vp->insns);
43   vp->insns = NULL;
44   vp->nr_insns = 0;
45
46   util_dynarray_fini(&vp->const_relocs);
47   nouveau_heap_free(&vp->data);
48   FREE(vp->consts);
49   vp->consts = NULL;
50   vp->nr_consts = 0;
51
52   vp->translated = false;
53}
54
55void
56nv30_vertprog_validate(struct nv30_context *nv30)
57{
58   struct nouveau_pushbuf *push = nv30->base.pushbuf;
59   struct nouveau_object *eng3d = nv30->screen->eng3d;
60   struct nv30_vertprog *vp = nv30->vertprog.program;
61   struct nv30_fragprog *fp = nv30->fragprog.program;
62   bool upload_code = false;
63   bool upload_data = false;
64   unsigned i;
65
66   if (nv30->dirty & NV30_NEW_FRAGPROG) {
67      if (memcmp(vp->texcoord, fp->texcoord, sizeof(vp->texcoord))) {
68         if (vp->translated)
69            nv30_vertprog_destroy(vp);
70         memcpy(vp->texcoord, fp->texcoord, sizeof(vp->texcoord));
71      }
72   }
73
74   if (nv30->rast && nv30->rast->pipe.clip_plane_enable != vp->enabled_ucps) {
75      vp->enabled_ucps = nv30->rast->pipe.clip_plane_enable;
76      if (vp->translated)
77         nv30_vertprog_destroy(vp);
78   }
79
80   if (!vp->translated) {
81      vp->translated = _nvfx_vertprog_translate(eng3d->oclass, vp);
82      if (!vp->translated) {
83         nv30->draw_flags |= NV30_NEW_VERTPROG;
84         return;
85      }
86      nv30->dirty |= NV30_NEW_VERTPROG;
87   }
88
89   if (!vp->exec) {
90      struct nouveau_heap *heap = nv30->screen->vp_exec_heap;
91      struct nv30_shader_reloc *reloc = vp->branch_relocs.data;
92      unsigned nr_reloc = vp->branch_relocs.size / sizeof(*reloc);
93      uint32_t *inst, target;
94
95      if (nouveau_heap_alloc(heap, vp->nr_insns, &vp->exec, &vp->exec)) {
96         while (heap->next && heap->size < vp->nr_insns) {
97            struct nouveau_heap **evict = heap->next->priv;
98            nouveau_heap_free(evict);
99         }
100
101         if (nouveau_heap_alloc(heap, vp->nr_insns, &vp->exec, &vp->exec)) {
102            nv30->draw_flags |= NV30_NEW_VERTPROG;
103            return;
104         }
105      }
106
107      if (eng3d->oclass < NV40_3D_CLASS) {
108         while (nr_reloc--) {
109            inst     = vp->insns[reloc->location].data;
110            target   = vp->exec->start + reloc->target;
111
112            inst[2] &= ~0x000007fc;
113            inst[2] |= target << 2;
114            reloc++;
115         }
116      } else {
117         while (nr_reloc--) {
118            inst     = vp->insns[reloc->location].data;
119            target   = vp->exec->start + reloc->target;
120
121            inst[2] &= ~0x0000003f;
122            inst[2] |= target >> 3;
123            inst[3] &= ~0xe0000000;
124            inst[3] |= target << 29;
125            reloc++;
126         }
127      }
128
129      upload_code = true;
130   }
131
132   if (vp->nr_consts && !vp->data) {
133      struct nouveau_heap *heap = nv30->screen->vp_data_heap;
134      struct nv30_shader_reloc *reloc = vp->const_relocs.data;
135      unsigned nr_reloc = vp->const_relocs.size / sizeof(*reloc);
136      uint32_t *inst, target;
137
138      if (nouveau_heap_alloc(heap, vp->nr_consts, vp, &vp->data)) {
139         while (heap->next && heap->size < vp->nr_consts) {
140            struct nv30_vertprog *evp = heap->next->priv;
141            nouveau_heap_free(&evp->data);
142         }
143
144         if (nouveau_heap_alloc(heap, vp->nr_consts, vp, &vp->data)) {
145            nv30->draw_flags |= NV30_NEW_VERTPROG;
146            return;
147         }
148      }
149
150      if (eng3d->oclass < NV40_3D_CLASS) {
151         while (nr_reloc--) {
152            inst     = vp->insns[reloc->location].data;
153            target   = vp->data->start + reloc->target;
154
155            inst[1] &= ~0x0007fc000;
156            inst[1] |= (target & 0x1ff) << 14;
157            reloc++;
158         }
159      } else {
160         while (nr_reloc--) {
161            inst     = vp->insns[reloc->location].data;
162            target   = vp->data->start + reloc->target;
163
164            inst[1] &= ~0x0001ff000;
165            inst[1] |= (target & 0x1ff) << 12;
166            reloc++;
167         }
168      }
169
170      upload_code = true;
171      upload_data = true;
172   }
173
174   if (vp->nr_consts) {
175      struct nv04_resource *res = nv04_resource(nv30->vertprog.constbuf);
176
177      for (i = 0; i < vp->nr_consts; i++) {
178         struct nv30_vertprog_data *data = &vp->consts[i];
179
180         if (data->index < 0) {
181            if (!upload_data)
182               continue;
183         } else {
184            float *constbuf = (float *)res->data;
185            if (!upload_data &&
186                !memcmp(data->value, &constbuf[data->index * 4], 16))
187               continue;
188            memcpy(data->value, &constbuf[data->index * 4], 16);
189         }
190
191         BEGIN_NV04(push, NV30_3D(VP_UPLOAD_CONST_ID), 5);
192         PUSH_DATA (push, vp->data->start + i);
193         PUSH_DATAp(push, data->value, 4);
194      }
195   }
196
197   if (upload_code) {
198      BEGIN_NV04(push, NV30_3D(VP_UPLOAD_FROM_ID), 1);
199      PUSH_DATA (push, vp->exec->start);
200      for (i = 0; i < vp->nr_insns; i++) {
201         BEGIN_NV04(push, NV30_3D(VP_UPLOAD_INST(0)), 4);
202         PUSH_DATAp(push, vp->insns[i].data, 4);
203      }
204   }
205
206   if (nv30->dirty & (NV30_NEW_VERTPROG | NV30_NEW_FRAGPROG)) {
207      BEGIN_NV04(push, NV30_3D(VP_START_FROM_ID), 1);
208      PUSH_DATA (push, vp->exec->start);
209      if (eng3d->oclass < NV40_3D_CLASS) {
210         BEGIN_NV04(push, NV30_3D(ENGINE), 1);
211         PUSH_DATA (push, 0x00000013); /* vp instead of ff, somehow */
212      } else {
213         BEGIN_NV04(push, NV40_3D(VP_ATTRIB_EN), 2);
214         PUSH_DATA (push, vp->ir);
215         PUSH_DATA (push, vp->or | fp->vp_or);
216         BEGIN_NV04(push, NV30_3D(ENGINE), 1);
217         PUSH_DATA (push, 0x00000011);
218      }
219   }
220}
221
222static void *
223nv30_vp_state_create(struct pipe_context *pipe,
224                     const struct pipe_shader_state *cso)
225{
226   struct nv30_vertprog *vp = CALLOC_STRUCT(nv30_vertprog);
227   if (!vp)
228      return NULL;
229
230   if (cso->type == PIPE_SHADER_IR_NIR) {
231      vp->pipe.tokens = nir_to_tgsi(cso->ir.nir, pipe->screen);
232   } else {
233      assert(cso->type == PIPE_SHADER_IR_TGSI);
234      /* we need to keep a local copy of the tokens */
235      vp->pipe.tokens = tgsi_dup_tokens(cso->tokens);
236   }
237
238   tgsi_scan_shader(vp->pipe.tokens, &vp->info);
239   return vp;
240}
241
242static void
243nv30_vp_state_delete(struct pipe_context *pipe, void *hwcso)
244{
245   struct nv30_vertprog *vp = hwcso;
246
247   if (vp->translated)
248      nv30_vertprog_destroy(vp);
249
250   if (vp->draw)
251      draw_delete_vertex_shader(nv30_context(pipe)->draw, vp->draw);
252
253   FREE((void *)vp->pipe.tokens);
254   FREE(vp);
255}
256
257static void
258nv30_vp_state_bind(struct pipe_context *pipe, void *hwcso)
259{
260   struct nv30_context *nv30 = nv30_context(pipe);
261
262   nv30->vertprog.program = hwcso;
263   nv30->dirty |= NV30_NEW_VERTPROG;
264}
265
266void
267nv30_vertprog_init(struct pipe_context *pipe)
268{
269   pipe->create_vs_state = nv30_vp_state_create;
270   pipe->bind_vs_state = nv30_vp_state_bind;
271   pipe->delete_vs_state = nv30_vp_state_delete;
272}
273