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 COPYRIGHT HOLDER(S) OR AUTHOR(S) 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#include "nv40.h"
25#include "regs.h"
26
27#include <core/client.h>
28#include <core/handle.h>
29#include <subdev/fb.h>
30#include <subdev/timer.h>
31#include <engine/fifo.h>
32
33struct nv40_gr_priv {
34	struct nvkm_gr base;
35	u32 size;
36};
37
38struct nv40_gr_chan {
39	struct nvkm_gr_chan base;
40};
41
42static u64
43nv40_gr_units(struct nvkm_gr *gr)
44{
45	struct nv40_gr_priv *priv = (void *)gr;
46
47	return nv_rd32(priv, 0x1540);
48}
49
50/*******************************************************************************
51 * Graphics object classes
52 ******************************************************************************/
53
54static int
55nv40_gr_object_ctor(struct nvkm_object *parent, struct nvkm_object *engine,
56		    struct nvkm_oclass *oclass, void *data, u32 size,
57		    struct nvkm_object **pobject)
58{
59	struct nvkm_gpuobj *obj;
60	int ret;
61
62	ret = nvkm_gpuobj_create(parent, engine, oclass, 0, parent,
63				 20, 16, 0, &obj);
64	*pobject = nv_object(obj);
65	if (ret)
66		return ret;
67
68	nv_wo32(obj, 0x00, nv_mclass(obj));
69	nv_wo32(obj, 0x04, 0x00000000);
70	nv_wo32(obj, 0x08, 0x00000000);
71#ifdef __BIG_ENDIAN
72	nv_mo32(obj, 0x08, 0x01000000, 0x01000000);
73#endif
74	nv_wo32(obj, 0x0c, 0x00000000);
75	nv_wo32(obj, 0x10, 0x00000000);
76	return 0;
77}
78
79static struct nvkm_ofuncs
80nv40_gr_ofuncs = {
81	.ctor = nv40_gr_object_ctor,
82	.dtor = _nvkm_gpuobj_dtor,
83	.init = _nvkm_gpuobj_init,
84	.fini = _nvkm_gpuobj_fini,
85	.rd32 = _nvkm_gpuobj_rd32,
86	.wr32 = _nvkm_gpuobj_wr32,
87};
88
89static struct nvkm_oclass
90nv40_gr_sclass[] = {
91	{ 0x0012, &nv40_gr_ofuncs, NULL }, /* beta1 */
92	{ 0x0019, &nv40_gr_ofuncs, NULL }, /* clip */
93	{ 0x0030, &nv40_gr_ofuncs, NULL }, /* null */
94	{ 0x0039, &nv40_gr_ofuncs, NULL }, /* m2mf */
95	{ 0x0043, &nv40_gr_ofuncs, NULL }, /* rop */
96	{ 0x0044, &nv40_gr_ofuncs, NULL }, /* patt */
97	{ 0x004a, &nv40_gr_ofuncs, NULL }, /* gdi */
98	{ 0x0062, &nv40_gr_ofuncs, NULL }, /* surf2d */
99	{ 0x0072, &nv40_gr_ofuncs, NULL }, /* beta4 */
100	{ 0x0089, &nv40_gr_ofuncs, NULL }, /* sifm */
101	{ 0x008a, &nv40_gr_ofuncs, NULL }, /* ifc */
102	{ 0x009f, &nv40_gr_ofuncs, NULL }, /* imageblit */
103	{ 0x3062, &nv40_gr_ofuncs, NULL }, /* surf2d (nv40) */
104	{ 0x3089, &nv40_gr_ofuncs, NULL }, /* sifm (nv40) */
105	{ 0x309e, &nv40_gr_ofuncs, NULL }, /* swzsurf (nv40) */
106	{ 0x4097, &nv40_gr_ofuncs, NULL }, /* curie */
107	{},
108};
109
110static struct nvkm_oclass
111nv44_gr_sclass[] = {
112	{ 0x0012, &nv40_gr_ofuncs, NULL }, /* beta1 */
113	{ 0x0019, &nv40_gr_ofuncs, NULL }, /* clip */
114	{ 0x0030, &nv40_gr_ofuncs, NULL }, /* null */
115	{ 0x0039, &nv40_gr_ofuncs, NULL }, /* m2mf */
116	{ 0x0043, &nv40_gr_ofuncs, NULL }, /* rop */
117	{ 0x0044, &nv40_gr_ofuncs, NULL }, /* patt */
118	{ 0x004a, &nv40_gr_ofuncs, NULL }, /* gdi */
119	{ 0x0062, &nv40_gr_ofuncs, NULL }, /* surf2d */
120	{ 0x0072, &nv40_gr_ofuncs, NULL }, /* beta4 */
121	{ 0x0089, &nv40_gr_ofuncs, NULL }, /* sifm */
122	{ 0x008a, &nv40_gr_ofuncs, NULL }, /* ifc */
123	{ 0x009f, &nv40_gr_ofuncs, NULL }, /* imageblit */
124	{ 0x3062, &nv40_gr_ofuncs, NULL }, /* surf2d (nv40) */
125	{ 0x3089, &nv40_gr_ofuncs, NULL }, /* sifm (nv40) */
126	{ 0x309e, &nv40_gr_ofuncs, NULL }, /* swzsurf (nv40) */
127	{ 0x4497, &nv40_gr_ofuncs, NULL }, /* curie */
128	{},
129};
130
131/*******************************************************************************
132 * PGRAPH context
133 ******************************************************************************/
134
135static int
136nv40_gr_context_ctor(struct nvkm_object *parent, struct nvkm_object *engine,
137		     struct nvkm_oclass *oclass, void *data, u32 size,
138		     struct nvkm_object **pobject)
139{
140	struct nv40_gr_priv *priv = (void *)engine;
141	struct nv40_gr_chan *chan;
142	int ret;
143
144	ret = nvkm_gr_context_create(parent, engine, oclass, NULL, priv->size,
145				     16, NVOBJ_FLAG_ZERO_ALLOC, &chan);
146	*pobject = nv_object(chan);
147	if (ret)
148		return ret;
149
150	nv40_grctx_fill(nv_device(priv), nv_gpuobj(chan));
151	nv_wo32(chan, 0x00000, nv_gpuobj(chan)->addr >> 4);
152	return 0;
153}
154
155static int
156nv40_gr_context_fini(struct nvkm_object *object, bool suspend)
157{
158	struct nv40_gr_priv *priv = (void *)object->engine;
159	struct nv40_gr_chan *chan = (void *)object;
160	u32 inst = 0x01000000 | nv_gpuobj(chan)->addr >> 4;
161	int ret = 0;
162
163	nv_mask(priv, 0x400720, 0x00000001, 0x00000000);
164
165	if (nv_rd32(priv, 0x40032c) == inst) {
166		if (suspend) {
167			nv_wr32(priv, 0x400720, 0x00000000);
168			nv_wr32(priv, 0x400784, inst);
169			nv_mask(priv, 0x400310, 0x00000020, 0x00000020);
170			nv_mask(priv, 0x400304, 0x00000001, 0x00000001);
171			if (!nv_wait(priv, 0x400300, 0x00000001, 0x00000000)) {
172				u32 insn = nv_rd32(priv, 0x400308);
173				nv_warn(priv, "ctxprog timeout 0x%08x\n", insn);
174				ret = -EBUSY;
175			}
176		}
177
178		nv_mask(priv, 0x40032c, 0x01000000, 0x00000000);
179	}
180
181	if (nv_rd32(priv, 0x400330) == inst)
182		nv_mask(priv, 0x400330, 0x01000000, 0x00000000);
183
184	nv_mask(priv, 0x400720, 0x00000001, 0x00000001);
185	return ret;
186}
187
188static struct nvkm_oclass
189nv40_gr_cclass = {
190	.handle = NV_ENGCTX(GR, 0x40),
191	.ofuncs = &(struct nvkm_ofuncs) {
192		.ctor = nv40_gr_context_ctor,
193		.dtor = _nvkm_gr_context_dtor,
194		.init = _nvkm_gr_context_init,
195		.fini = nv40_gr_context_fini,
196		.rd32 = _nvkm_gr_context_rd32,
197		.wr32 = _nvkm_gr_context_wr32,
198	},
199};
200
201/*******************************************************************************
202 * PGRAPH engine/subdev functions
203 ******************************************************************************/
204
205static void
206nv40_gr_tile_prog(struct nvkm_engine *engine, int i)
207{
208	struct nvkm_fb_tile *tile = &nvkm_fb(engine)->tile.region[i];
209	struct nvkm_fifo *pfifo = nvkm_fifo(engine);
210	struct nv40_gr_priv *priv = (void *)engine;
211	unsigned long flags;
212
213	pfifo->pause(pfifo, &flags);
214	nv04_gr_idle(priv);
215
216	switch (nv_device(priv)->chipset) {
217	case 0x40:
218	case 0x41:
219	case 0x42:
220	case 0x43:
221	case 0x45:
222	case 0x4e:
223		nv_wr32(priv, NV20_PGRAPH_TSIZE(i), tile->pitch);
224		nv_wr32(priv, NV20_PGRAPH_TLIMIT(i), tile->limit);
225		nv_wr32(priv, NV20_PGRAPH_TILE(i), tile->addr);
226		nv_wr32(priv, NV40_PGRAPH_TSIZE1(i), tile->pitch);
227		nv_wr32(priv, NV40_PGRAPH_TLIMIT1(i), tile->limit);
228		nv_wr32(priv, NV40_PGRAPH_TILE1(i), tile->addr);
229		switch (nv_device(priv)->chipset) {
230		case 0x40:
231		case 0x45:
232			nv_wr32(priv, NV20_PGRAPH_ZCOMP(i), tile->zcomp);
233			nv_wr32(priv, NV40_PGRAPH_ZCOMP1(i), tile->zcomp);
234			break;
235		case 0x41:
236		case 0x42:
237		case 0x43:
238			nv_wr32(priv, NV41_PGRAPH_ZCOMP0(i), tile->zcomp);
239			nv_wr32(priv, NV41_PGRAPH_ZCOMP1(i), tile->zcomp);
240			break;
241		default:
242			break;
243		}
244		break;
245	case 0x44:
246	case 0x4a:
247		nv_wr32(priv, NV20_PGRAPH_TSIZE(i), tile->pitch);
248		nv_wr32(priv, NV20_PGRAPH_TLIMIT(i), tile->limit);
249		nv_wr32(priv, NV20_PGRAPH_TILE(i), tile->addr);
250		break;
251	case 0x46:
252	case 0x4c:
253	case 0x47:
254	case 0x49:
255	case 0x4b:
256	case 0x63:
257	case 0x67:
258	case 0x68:
259		nv_wr32(priv, NV47_PGRAPH_TSIZE(i), tile->pitch);
260		nv_wr32(priv, NV47_PGRAPH_TLIMIT(i), tile->limit);
261		nv_wr32(priv, NV47_PGRAPH_TILE(i), tile->addr);
262		nv_wr32(priv, NV40_PGRAPH_TSIZE1(i), tile->pitch);
263		nv_wr32(priv, NV40_PGRAPH_TLIMIT1(i), tile->limit);
264		nv_wr32(priv, NV40_PGRAPH_TILE1(i), tile->addr);
265		switch (nv_device(priv)->chipset) {
266		case 0x47:
267		case 0x49:
268		case 0x4b:
269			nv_wr32(priv, NV47_PGRAPH_ZCOMP0(i), tile->zcomp);
270			nv_wr32(priv, NV47_PGRAPH_ZCOMP1(i), tile->zcomp);
271			break;
272		default:
273			break;
274		}
275		break;
276	default:
277		break;
278	}
279
280	pfifo->start(pfifo, &flags);
281}
282
283static void
284nv40_gr_intr(struct nvkm_subdev *subdev)
285{
286	struct nvkm_fifo *pfifo = nvkm_fifo(subdev);
287	struct nvkm_engine *engine = nv_engine(subdev);
288	struct nvkm_object *engctx;
289	struct nvkm_handle *handle = NULL;
290	struct nv40_gr_priv *priv = (void *)subdev;
291	u32 stat = nv_rd32(priv, NV03_PGRAPH_INTR);
292	u32 nsource = nv_rd32(priv, NV03_PGRAPH_NSOURCE);
293	u32 nstatus = nv_rd32(priv, NV03_PGRAPH_NSTATUS);
294	u32 inst = nv_rd32(priv, 0x40032c) & 0x000fffff;
295	u32 addr = nv_rd32(priv, NV04_PGRAPH_TRAPPED_ADDR);
296	u32 subc = (addr & 0x00070000) >> 16;
297	u32 mthd = (addr & 0x00001ffc);
298	u32 data = nv_rd32(priv, NV04_PGRAPH_TRAPPED_DATA);
299	u32 class = nv_rd32(priv, 0x400160 + subc * 4) & 0xffff;
300	u32 show = stat;
301	int chid;
302
303	engctx = nvkm_engctx_get(engine, inst);
304	chid   = pfifo->chid(pfifo, engctx);
305
306	if (stat & NV_PGRAPH_INTR_ERROR) {
307		if (nsource & NV03_PGRAPH_NSOURCE_ILLEGAL_MTHD) {
308			handle = nvkm_handle_get_class(engctx, class);
309			if (handle && !nv_call(handle->object, mthd, data))
310				show &= ~NV_PGRAPH_INTR_ERROR;
311			nvkm_handle_put(handle);
312		}
313
314		if (nsource & NV03_PGRAPH_NSOURCE_DMA_VTX_PROTECTION) {
315			nv_mask(priv, 0x402000, 0, 0);
316		}
317	}
318
319	nv_wr32(priv, NV03_PGRAPH_INTR, stat);
320	nv_wr32(priv, NV04_PGRAPH_FIFO, 0x00000001);
321
322	if (show) {
323		nv_error(priv, "%s", "");
324		nvkm_bitfield_print(nv10_gr_intr_name, show);
325		pr_cont(" nsource:");
326		nvkm_bitfield_print(nv04_gr_nsource, nsource);
327		pr_cont(" nstatus:");
328		nvkm_bitfield_print(nv10_gr_nstatus, nstatus);
329		pr_cont("\n");
330		nv_error(priv,
331			 "ch %d [0x%08x %s] subc %d class 0x%04x mthd 0x%04x data 0x%08x\n",
332			 chid, inst << 4, nvkm_client_name(engctx), subc,
333			 class, mthd, data);
334	}
335
336	nvkm_engctx_put(engctx);
337}
338
339static int
340nv40_gr_ctor(struct nvkm_object *parent, struct nvkm_object *engine,
341	     struct nvkm_oclass *oclass, void *data, u32 size,
342	     struct nvkm_object **pobject)
343{
344	struct nv40_gr_priv *priv;
345	int ret;
346
347	ret = nvkm_gr_create(parent, engine, oclass, true, &priv);
348	*pobject = nv_object(priv);
349	if (ret)
350		return ret;
351
352	nv_subdev(priv)->unit = 0x00001000;
353	nv_subdev(priv)->intr = nv40_gr_intr;
354	nv_engine(priv)->cclass = &nv40_gr_cclass;
355	if (nv44_gr_class(priv))
356		nv_engine(priv)->sclass = nv44_gr_sclass;
357	else
358		nv_engine(priv)->sclass = nv40_gr_sclass;
359	nv_engine(priv)->tile_prog = nv40_gr_tile_prog;
360
361	priv->base.units = nv40_gr_units;
362	return 0;
363}
364
365static int
366nv40_gr_init(struct nvkm_object *object)
367{
368	struct nvkm_engine *engine = nv_engine(object);
369	struct nvkm_fb *pfb = nvkm_fb(object);
370	struct nv40_gr_priv *priv = (void *)engine;
371	int ret, i, j;
372	u32 vramsz;
373
374	ret = nvkm_gr_init(&priv->base);
375	if (ret)
376		return ret;
377
378	/* generate and upload context program */
379	ret = nv40_grctx_init(nv_device(priv), &priv->size);
380	if (ret)
381		return ret;
382
383	/* No context present currently */
384	nv_wr32(priv, NV40_PGRAPH_CTXCTL_CUR, 0x00000000);
385
386	nv_wr32(priv, NV03_PGRAPH_INTR   , 0xFFFFFFFF);
387	nv_wr32(priv, NV40_PGRAPH_INTR_EN, 0xFFFFFFFF);
388
389	nv_wr32(priv, NV04_PGRAPH_DEBUG_0, 0xFFFFFFFF);
390	nv_wr32(priv, NV04_PGRAPH_DEBUG_0, 0x00000000);
391	nv_wr32(priv, NV04_PGRAPH_DEBUG_1, 0x401287c0);
392	nv_wr32(priv, NV04_PGRAPH_DEBUG_3, 0xe0de8055);
393	nv_wr32(priv, NV10_PGRAPH_DEBUG_4, 0x00008000);
394	nv_wr32(priv, NV04_PGRAPH_LIMIT_VIOL_PIX, 0x00be3c5f);
395
396	nv_wr32(priv, NV10_PGRAPH_CTX_CONTROL, 0x10010100);
397	nv_wr32(priv, NV10_PGRAPH_STATE      , 0xFFFFFFFF);
398
399	j = nv_rd32(priv, 0x1540) & 0xff;
400	if (j) {
401		for (i = 0; !(j & 1); j >>= 1, i++)
402			;
403		nv_wr32(priv, 0x405000, i);
404	}
405
406	if (nv_device(priv)->chipset == 0x40) {
407		nv_wr32(priv, 0x4009b0, 0x83280fff);
408		nv_wr32(priv, 0x4009b4, 0x000000a0);
409	} else {
410		nv_wr32(priv, 0x400820, 0x83280eff);
411		nv_wr32(priv, 0x400824, 0x000000a0);
412	}
413
414	switch (nv_device(priv)->chipset) {
415	case 0x40:
416	case 0x45:
417		nv_wr32(priv, 0x4009b8, 0x0078e366);
418		nv_wr32(priv, 0x4009bc, 0x0000014c);
419		break;
420	case 0x41:
421	case 0x42: /* pciid also 0x00Cx */
422	/* case 0x0120: XXX (pciid) */
423		nv_wr32(priv, 0x400828, 0x007596ff);
424		nv_wr32(priv, 0x40082c, 0x00000108);
425		break;
426	case 0x43:
427		nv_wr32(priv, 0x400828, 0x0072cb77);
428		nv_wr32(priv, 0x40082c, 0x00000108);
429		break;
430	case 0x44:
431	case 0x46: /* G72 */
432	case 0x4a:
433	case 0x4c: /* G7x-based C51 */
434	case 0x4e:
435		nv_wr32(priv, 0x400860, 0);
436		nv_wr32(priv, 0x400864, 0);
437		break;
438	case 0x47: /* G70 */
439	case 0x49: /* G71 */
440	case 0x4b: /* G73 */
441		nv_wr32(priv, 0x400828, 0x07830610);
442		nv_wr32(priv, 0x40082c, 0x0000016A);
443		break;
444	default:
445		break;
446	}
447
448	nv_wr32(priv, 0x400b38, 0x2ffff800);
449	nv_wr32(priv, 0x400b3c, 0x00006000);
450
451	/* Tiling related stuff. */
452	switch (nv_device(priv)->chipset) {
453	case 0x44:
454	case 0x4a:
455		nv_wr32(priv, 0x400bc4, 0x1003d888);
456		nv_wr32(priv, 0x400bbc, 0xb7a7b500);
457		break;
458	case 0x46:
459		nv_wr32(priv, 0x400bc4, 0x0000e024);
460		nv_wr32(priv, 0x400bbc, 0xb7a7b520);
461		break;
462	case 0x4c:
463	case 0x4e:
464	case 0x67:
465		nv_wr32(priv, 0x400bc4, 0x1003d888);
466		nv_wr32(priv, 0x400bbc, 0xb7a7b540);
467		break;
468	default:
469		break;
470	}
471
472	/* Turn all the tiling regions off. */
473	for (i = 0; i < pfb->tile.regions; i++)
474		engine->tile_prog(engine, i);
475
476	/* begin RAM config */
477	vramsz = nv_device_resource_len(nv_device(priv), 0) - 1;
478	switch (nv_device(priv)->chipset) {
479	case 0x40:
480		nv_wr32(priv, 0x4009A4, nv_rd32(priv, 0x100200));
481		nv_wr32(priv, 0x4009A8, nv_rd32(priv, 0x100204));
482		nv_wr32(priv, 0x4069A4, nv_rd32(priv, 0x100200));
483		nv_wr32(priv, 0x4069A8, nv_rd32(priv, 0x100204));
484		nv_wr32(priv, 0x400820, 0);
485		nv_wr32(priv, 0x400824, 0);
486		nv_wr32(priv, 0x400864, vramsz);
487		nv_wr32(priv, 0x400868, vramsz);
488		break;
489	default:
490		switch (nv_device(priv)->chipset) {
491		case 0x41:
492		case 0x42:
493		case 0x43:
494		case 0x45:
495		case 0x4e:
496		case 0x44:
497		case 0x4a:
498			nv_wr32(priv, 0x4009F0, nv_rd32(priv, 0x100200));
499			nv_wr32(priv, 0x4009F4, nv_rd32(priv, 0x100204));
500			break;
501		default:
502			nv_wr32(priv, 0x400DF0, nv_rd32(priv, 0x100200));
503			nv_wr32(priv, 0x400DF4, nv_rd32(priv, 0x100204));
504			break;
505		}
506		nv_wr32(priv, 0x4069F0, nv_rd32(priv, 0x100200));
507		nv_wr32(priv, 0x4069F4, nv_rd32(priv, 0x100204));
508		nv_wr32(priv, 0x400840, 0);
509		nv_wr32(priv, 0x400844, 0);
510		nv_wr32(priv, 0x4008A0, vramsz);
511		nv_wr32(priv, 0x4008A4, vramsz);
512		break;
513	}
514
515	return 0;
516}
517
518struct nvkm_oclass
519nv40_gr_oclass = {
520	.handle = NV_ENGINE(GR, 0x40),
521	.ofuncs = &(struct nvkm_ofuncs) {
522		.ctor = nv40_gr_ctor,
523		.dtor = _nvkm_gr_dtor,
524		.init = nv40_gr_init,
525		.fini = _nvkm_gr_fini,
526	},
527};
528