summaryrefslogtreecommitdiffstats
path: root/drivers/gpu/drm/nouveau/nouveau_chan.c
blob: fd3dbd59d73e18113a6ef357e658570c113155be (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
/*
 * Copyright 2012 Red Hat Inc.
 *
 * Permission is hereby granted, free of charge, to any person obtaining a
 * copy of this software and associated documentation files (the "Software"),
 * to deal in the Software without restriction, including without limitation
 * the rights to use, copy, modify, merge, publish, distribute, sublicense,
 * and/or sell copies of the Software, and to permit persons to whom the
 * Software is furnished to do so, subject to the following conditions:
 *
 * The above copyright notice and this permission notice shall be included in
 * all copies or substantial portions of the Software.
 *
 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
 * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
 * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
 * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
 * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
 * OTHER DEALINGS IN THE SOFTWARE.
 *
 * Authors: Ben Skeggs
 */

#include <nvif/os.h>
#include <nvif/class.h>

/*XXX*/
#include <core/client.h>

#include "nouveau_drm.h"
#include "nouveau_dma.h"
#include "nouveau_bo.h"
#include "nouveau_chan.h"
#include "nouveau_fence.h"
#include "nouveau_abi16.h"

MODULE_PARM_DESC(vram_pushbuf, "Create DMA push buffers in VRAM");
int nouveau_vram_pushbuf;
module_param_named(vram_pushbuf, nouveau_vram_pushbuf, int, 0400);

int
nouveau_channel_idle(struct nouveau_channel *chan)
{
	struct nouveau_cli *cli = (void *)nvif_client(chan->object);
	struct nouveau_fence *fence = NULL;
	int ret;

	ret = nouveau_fence_new(chan, false, &fence);
	if (!ret) {
		ret = nouveau_fence_wait(fence, false, false);
		nouveau_fence_unref(&fence);
	}

	if (ret)
		NV_PRINTK(error, cli, "failed to idle channel 0x%08x [%s]\n",
			  chan->object->handle, nvkm_client(&cli->base)->name);
	return ret;
}

void
nouveau_channel_del(struct nouveau_channel **pchan)
{
	struct nouveau_channel *chan = *pchan;
	if (chan) {
		if (chan->fence) {
			nouveau_channel_idle(chan);
			nouveau_fence(chan->drm)->context_del(chan);
		}
		nvif_object_fini(&chan->nvsw);
		nvif_object_fini(&chan->gart);
		nvif_object_fini(&chan->vram);
		nvif_object_ref(NULL, &chan->object);
		nvif_object_fini(&chan->push.ctxdma);
		nouveau_bo_vma_del(chan->push.buffer, &chan->push.vma);
		nouveau_bo_unmap(chan->push.buffer);
		if (chan->push.buffer && chan->push.buffer->pin_refcnt)
			nouveau_bo_unpin(chan->push.buffer);
		nouveau_bo_ref(NULL, &chan->push.buffer);
		nvif_device_ref(NULL, &chan->device);
		kfree(chan);
	}
	*pchan = NULL;
}

static int
nouveau_channel_prep(struct nouveau_drm *drm, struct nvif_device *device,
		     u32 handle, u32 size, struct nouveau_channel **pchan)
{
	struct nouveau_cli *cli = (void *)nvif_client(&device->base);
	struct nouveau_vmmgr *vmm = nvkm_vmmgr(device);
	struct nv_dma_v0 args = {};
	struct nouveau_channel *chan;
	u32 target;
	int ret;

	chan = *pchan = kzalloc(sizeof(*chan), GFP_KERNEL);
	if (!chan)
		return -ENOMEM;

	nvif_device_ref(device, &chan->device);
	chan->drm = drm;

	/* allocate memory for dma push buffer */
	target = TTM_PL_FLAG_TT;
	if (nouveau_vram_pushbuf)
		target = TTM_PL_FLAG_VRAM;

	ret = nouveau_bo_new(drm->dev, size, 0, target, 0, 0, NULL, NULL,
			    &chan->push.buffer);
	if (ret == 0) {
		ret = nouveau_bo_pin(chan->push.buffer, target);
		if (ret == 0)
			ret = nouveau_bo_map(chan->push.buffer);
	}

	if (ret) {
		nouveau_channel_del(pchan);
		return ret;
	}

	/* create dma object covering the *entire* memory space that the
	 * pushbuf lives in, this is because the GEM code requires that
	 * we be able to call out to other (indirect) push buffers
	 */
	chan->push.vma.offset = chan->push.buffer->bo.offset;

	if (device->info.family >= NV_DEVICE_INFO_V0_TESLA) {
		ret = nouveau_bo_vma_add(chan->push.buffer, cli->vm,
					&chan->push.vma);
		if (ret) {
			nouveau_channel_del(pchan);
			return ret;
		}

		args.target = NV_DMA_V0_TARGET_VM;
		args.access = NV_DMA_V0_ACCESS_VM;
		args.start = 0;
		args.limit = cli->vm->vmm->limit - 1;
	} else
	if (chan->push.buffer->bo.mem.mem_type == TTM_PL_VRAM) {
		if (device->info.family == NV_DEVICE_INFO_V0_TNT) {
			/* nv04 vram pushbuf hack, retarget to its location in
			 * the framebuffer bar rather than direct vram access..
			 * nfi why this exists, it came from the -nv ddx.
			 */
			args.target = NV_DMA_V0_TARGET_PCI;
			args.access = NV_DMA_V0_ACCESS_RDWR;
			args.start = nv_device_resource_start(nvkm_device(device), 1);
			args.limit = args.start + device->info.ram_user - 1;
		} else {
			args.target = NV_DMA_V0_TARGET_VRAM;
			args.access = NV_DMA_V0_ACCESS_RDWR;
			args.start = 0;
			args.limit = device->info.ram_user - 1;
		}
	} else {
		if (chan->drm->agp.stat == ENABLED) {
			args.target = NV_DMA_V0_TARGET_AGP;
			args.access = NV_DMA_V0_ACCESS_RDWR;
			args.start = chan->drm->agp.base;
			args.limit = chan->drm->agp.base +
				     chan->drm->agp.size - 1;
		} else {
			args.target = NV_DMA_V0_TARGET_VM;
			args.access = NV_DMA_V0_ACCESS_RDWR;
			args.start = 0;
			args.limit = vmm->limit - 1;
		}
	}

	ret = nvif_object_init(nvif_object(device), NULL, NVDRM_PUSH |
			       (handle & 0xffff), NV_DMA_FROM_MEMORY,
			       &args, sizeof(args), &chan->push.ctxdma);
	if (ret) {
		nouveau_channel_del(pchan);
		return ret;
	}

	return 0;
}

static int
nouveau_channel_ind(struct nouveau_drm *drm, struct nvif_device *device,
		    u32 handle, u32 engine, struct nouveau_channel **pchan)
{
	static const u16 oclasses[] = { KEPLER_CHANNEL_GPFIFO_A,
					FERMI_CHANNEL_GPFIFO,
					G82_CHANNEL_GPFIFO,
					NV50_CHANNEL_GPFIFO,
					0 };
	const u16 *oclass = oclasses;
	union {
		struct nv50_channel_gpfifo_v0 nv50;
		struct kepler_channel_gpfifo_a_v0 kepler;
	} args, *retn;
	struct nouveau_channel *chan;
	u32 size;
	int ret;

	/* allocate dma push buffer */
	ret = nouveau_channel_prep(drm, device, handle, 0x12000, &chan);
	*pchan = chan;
	if (ret)
		return ret;

	/* create channel object */
	do {
		if (oclass[0] >= KEPLER_CHANNEL_GPFIFO_A) {
			args.kepler.version = 0;
			args.kepler.engine  = engine;
			args.kepler.pushbuf = chan->push.ctxdma.handle;
			args.kepler.ilength = 0x02000;
			args.kepler.ioffset = 0x10000 + chan->push.vma.offset;
			size = sizeof(args.kepler);
		} else {
			args.nv50.version = 0;
			args.nv50.pushbuf = chan->push.ctxdma.handle;
			args.nv50.ilength = 0x02000;
			args.nv50.ioffset = 0x10000 + chan->push.vma.offset;
			size = sizeof(args.nv50);
		}

		ret = nvif_object_new(nvif_object(device), handle, *oclass++,
				      &args, size, &chan->object);
		if (ret == 0) {
			retn = chan->object->data;
			if (chan->object->oclass >= KEPLER_CHANNEL_GPFIFO_A)
				chan->chid = retn->kepler.chid;
			else
				chan->chid = retn->nv50.chid;
			return ret;
		}
	} while (*oclass);

	nouveau_channel_del(pchan);
	return ret;
}

static int
nouveau_channel_dma(struct nouveau_drm *drm, struct nvif_device *device,
		    u32 handle, struct nouveau_channel **pchan)
{
	static const u16 oclasses[] = { NV40_CHANNEL_DMA,
					NV17_CHANNEL_DMA,
					NV10_CHANNEL_DMA,
					NV03_CHANNEL_DMA,
					0 };
	const u16 *oclass = oclasses;
	struct nv03_channel_dma_v0 args, *retn;
	struct nouveau_channel *chan;
	int ret;

	/* allocate dma push buffer */
	ret = nouveau_channel_prep(drm, device, handle, 0x10000, &chan);
	*pchan = chan;
	if (ret)
		return ret;

	/* create channel object */
	args.version = 0;
	args.pushbuf = chan->push.ctxdma.handle;
	args.offset = chan->push.vma.offset;

	do {
		ret = nvif_object_new(nvif_object(device), handle, *oclass++,
				      &args, sizeof(args), &chan->object);
		if (ret == 0) {
			retn = chan->object->data;
			chan->chid = retn->chid;
			return ret;
		}
	} while (ret && *oclass);

	nouveau_channel_del(pchan);
	return ret;
}

static int
nouveau_channel_init(struct nouveau_channel *chan, u32 vram, u32 gart)
{
	struct nvif_device *device = chan->device;
	struct nouveau_cli *cli = (void *)nvif_client(&device->base);
	struct nouveau_vmmgr *vmm = nvkm_vmmgr(device);
	struct nouveau_software_chan *swch;
	struct nv_dma_v0 args = {};
	int ret, i;
	bool save;

	nvif_object_map(chan->object);

	/* allocate dma objects to cover all allowed vram, and gart */
	if (device->info.family < NV_DEVICE_INFO_V0_FERMI) {
		if (device->info.family >= NV_DEVICE_INFO_V0_TESLA) {
			args.target = NV_DMA_V0_TARGET_VM;
			args.access = NV_DMA_V0_ACCESS_VM;
			args.start = 0;
			args.limit = cli->vm->vmm->limit - 1;
		} else {
			args.target = NV_DMA_V0_TARGET_VRAM;
			args.access = NV_DMA_V0_ACCESS_RDWR;
			args.start = 0;
			args.limit = device->info.ram_user - 1;
		}

		ret = nvif_object_init(chan->object, NULL, vram,
				       NV_DMA_IN_MEMORY, &args,
				       sizeof(args), &chan->vram);
		if (ret)
			return ret;

		if (device->info.family >= NV_DEVICE_INFO_V0_TESLA) {
			args.target = NV_DMA_V0_TARGET_VM;
			args.access = NV_DMA_V0_ACCESS_VM;
			args.start = 0;
			args.limit = cli->vm->vmm->limit - 1;
		} else
		if (chan->drm->agp.stat == ENABLED) {
			args.target = NV_DMA_V0_TARGET_AGP;
			args.access = NV_DMA_V0_ACCESS_RDWR;
			args.start = chan->drm->agp.base;
			args.limit = chan->drm->agp.base +
				     chan->drm->agp.size - 1;
		} else {
			args.target = NV_DMA_V0_TARGET_VM;
			args.access = NV_DMA_V0_ACCESS_RDWR;
			args.start = 0;
			args.limit = vmm->limit - 1;
		}

		ret = nvif_object_init(chan->object, NULL, gart,
				       NV_DMA_IN_MEMORY, &args,
				       sizeof(args), &chan->gart);
		if (ret)
			return ret;
	}

	/* initialise dma tracking parameters */
	switch (chan->object->oclass & 0x00ff) {
	case 0x006b:
	case 0x006e:
		chan->user_put = 0x40;
		chan->user_get = 0x44;
		chan->dma.max = (0x10000 / 4) - 2;
		break;
	default:
		chan->user_put = 0x40;
		chan->user_get = 0x44;
		chan->user_get_hi = 0x60;
		chan->dma.ib_base =  0x10000 / 4;
		chan->dma.ib_max  = (0x02000 / 8) - 1;
		chan->dma.ib_put  = 0;
		chan->dma.ib_free = chan->dma.ib_max - chan->dma.ib_put;
		chan->dma.max = chan->dma.ib_base;
		break;
	}

	chan->dma.put = 0;
	chan->dma.cur = chan->dma.put;
	chan->dma.free = chan->dma.max - chan->dma.cur;

	ret = RING_SPACE(chan, NOUVEAU_DMA_SKIPS);
	if (ret)
		return ret;

	for (i = 0; i < NOUVEAU_DMA_SKIPS; i++)
		OUT_RING(chan, 0x00000000);

	/* allocate software object class (used for fences on <= nv05) */
	if (device->info.family < NV_DEVICE_INFO_V0_CELSIUS) {
		ret = nvif_object_init(chan->object, NULL, 0x006e, 0x006e,
				       NULL, 0, &chan->nvsw);
		if (ret)
			return ret;

		swch = (void *)nvkm_object(&chan->nvsw)->parent;
		swch->flip = nouveau_flip_complete;
		swch->flip_data = chan;

		ret = RING_SPACE(chan, 2);
		if (ret)
			return ret;

		BEGIN_NV04(chan, NvSubSw, 0x0000, 1);
		OUT_RING  (chan, chan->nvsw.handle);
		FIRE_RING (chan);
	}

	/* initialise synchronisation */
	save = cli->base.super;
	cli->base.super = true; /* hack until fencenv50 fixed */
	ret = nouveau_fence(chan->drm)->context_new(chan);
	cli->base.super = save;
	return ret;
}

int
nouveau_channel_new(struct nouveau_drm *drm, struct nvif_device *device,
		    u32 handle, u32 arg0, u32 arg1,
		    struct nouveau_channel **pchan)
{
	struct nouveau_cli *cli = (void *)nvif_client(&device->base);
	bool super;
	int ret;

	/* hack until fencenv50 is fixed, and agp access relaxed */
	super = cli->base.super;
	cli->base.super = true;

	ret = nouveau_channel_ind(drm, device, handle, arg0, pchan);
	if (ret) {
		NV_PRINTK(debug, cli, "ib channel create, %d\n", ret);
		ret = nouveau_channel_dma(drm, device, handle, pchan);
		if (ret) {
			NV_PRINTK(debug, cli, "dma channel create, %d\n", ret);
			goto done;
		}
	}

	ret = nouveau_channel_init(*pchan, arg0, arg1);
	if (ret) {
		NV_PRINTK(error, cli, "channel failed to initialise, %d\n", ret);
		nouveau_channel_del(pchan);
	}

done:
	cli->base.super = super;
	return ret;
}