nv40_fifo.c 10.2 KB
Newer Older
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
/*
 * Copyright (C) 2007 Ben Skeggs.
 * All Rights Reserved.
 *
 * 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 (including the
 * next paragraph) 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 OWNER(S) AND/OR ITS SUPPLIERS 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.
 *
 */

#include "drmP.h"
#include "nouveau_drv.h"
#include "nouveau_drm.h"

#define NV40_RAMFC(c) (dev_priv->ramfc_offset + ((c) * NV40_RAMFC__SIZE))
#define NV40_RAMFC__SIZE 128

int
nv40_fifo_create_context(struct nouveau_channel *chan)
{
	struct drm_device *dev = chan->dev;
	struct drm_nouveau_private *dev_priv = dev->dev_private;
	uint32_t fc = NV40_RAMFC(chan->id);
40
	unsigned long flags;
41 42 43 44 45 46 47 48
	int ret;

	ret = nouveau_gpuobj_new_fake(dev, NV40_RAMFC(chan->id), ~0,
				      NV40_RAMFC__SIZE, NVOBJ_FLAG_ZERO_ALLOC |
				      NVOBJ_FLAG_ZERO_FREE, NULL, &chan->ramfc);
	if (ret)
		return ret;

49 50
	spin_lock_irqsave(&dev_priv->context_switch_lock, flags);

51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68
	dev_priv->engine.instmem.prepare_access(dev, true);
	nv_wi32(dev, fc +  0, chan->pushbuf_base);
	nv_wi32(dev, fc +  4, chan->pushbuf_base);
	nv_wi32(dev, fc + 12, chan->pushbuf->instance >> 4);
	nv_wi32(dev, fc + 24, NV_PFIFO_CACHE1_DMA_FETCH_TRIG_128_BYTES |
			      NV_PFIFO_CACHE1_DMA_FETCH_SIZE_128_BYTES |
			      NV_PFIFO_CACHE1_DMA_FETCH_MAX_REQS_8 |
#ifdef __BIG_ENDIAN
			      NV_PFIFO_CACHE1_BIG_ENDIAN |
#endif
			      0x30000000 /* no idea.. */);
	nv_wi32(dev, fc + 56, chan->ramin_grctx->instance >> 4);
	nv_wi32(dev, fc + 60, 0x0001FFFF);
	dev_priv->engine.instmem.finish_access(dev);

	/* enable the fifo dma operation */
	nv_wr32(dev, NV04_PFIFO_MODE,
		nv_rd32(dev, NV04_PFIFO_MODE) | (1 << chan->id));
69 70

	spin_unlock_irqrestore(&dev_priv->context_switch_lock, flags);
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
	return 0;
}

void
nv40_fifo_destroy_context(struct nouveau_channel *chan)
{
	struct drm_device *dev = chan->dev;

	nv_wr32(dev, NV04_PFIFO_MODE,
		nv_rd32(dev, NV04_PFIFO_MODE) & ~(1 << chan->id));

	if (chan->ramfc)
		nouveau_gpuobj_ref_del(dev, &chan->ramfc);
}

static void
nv40_fifo_do_load_context(struct drm_device *dev, int chid)
{
	struct drm_nouveau_private *dev_priv = dev->dev_private;
	uint32_t fc = NV40_RAMFC(chid), tmp, tmp2;

	dev_priv->engine.instmem.prepare_access(dev, false);

	nv_wr32(dev, NV04_PFIFO_CACHE1_DMA_PUT, nv_ri32(dev, fc + 0));
	nv_wr32(dev, NV04_PFIFO_CACHE1_DMA_GET, nv_ri32(dev, fc + 4));
	nv_wr32(dev, NV10_PFIFO_CACHE1_REF_CNT, nv_ri32(dev, fc + 8));
	nv_wr32(dev, NV04_PFIFO_CACHE1_DMA_INSTANCE, nv_ri32(dev, fc + 12));
	nv_wr32(dev, NV04_PFIFO_CACHE1_DMA_DCOUNT, nv_ri32(dev, fc + 16));
	nv_wr32(dev, NV04_PFIFO_CACHE1_DMA_STATE, nv_ri32(dev, fc + 20));

	/* No idea what 0x2058 is.. */
	tmp   = nv_ri32(dev, fc + 24);
	tmp2  = nv_rd32(dev, 0x2058) & 0xFFF;
	tmp2 |= (tmp & 0x30000000);
	nv_wr32(dev, 0x2058, tmp2);
	tmp  &= ~0x30000000;
	nv_wr32(dev, NV04_PFIFO_CACHE1_DMA_FETCH, tmp);

	nv_wr32(dev, NV04_PFIFO_CACHE1_ENGINE, nv_ri32(dev, fc + 28));
	nv_wr32(dev, NV04_PFIFO_CACHE1_PULL1, nv_ri32(dev, fc + 32));
	nv_wr32(dev, NV10_PFIFO_CACHE1_ACQUIRE_VALUE, nv_ri32(dev, fc + 36));
	tmp = nv_ri32(dev, fc + 40);
	nv_wr32(dev, NV10_PFIFO_CACHE1_ACQUIRE_TIMESTAMP, tmp);
	nv_wr32(dev, NV10_PFIFO_CACHE1_ACQUIRE_TIMEOUT, nv_ri32(dev, fc + 44));
	nv_wr32(dev, NV10_PFIFO_CACHE1_SEMAPHORE, nv_ri32(dev, fc + 48));
	nv_wr32(dev, NV10_PFIFO_CACHE1_DMA_SUBROUTINE, nv_ri32(dev, fc + 52));
	nv_wr32(dev, NV40_PFIFO_GRCTX_INSTANCE, nv_ri32(dev, fc + 56));

	/* Don't clobber the TIMEOUT_ENABLED flag when restoring from RAMFC */
	tmp  = nv_rd32(dev, NV04_PFIFO_DMA_TIMESLICE) & ~0x1FFFF;
	tmp |= nv_ri32(dev, fc + 60) & 0x1FFFF;
	nv_wr32(dev, NV04_PFIFO_DMA_TIMESLICE, tmp);

	nv_wr32(dev, 0x32e4, nv_ri32(dev, fc + 64));
	/* NVIDIA does this next line twice... */
	nv_wr32(dev, 0x32e8, nv_ri32(dev, fc + 68));
	nv_wr32(dev, 0x2088, nv_ri32(dev, fc + 76));
	nv_wr32(dev, 0x3300, nv_ri32(dev, fc + 80));

	dev_priv->engine.instmem.finish_access(dev);

	nv_wr32(dev, NV03_PFIFO_CACHE1_GET, 0);
	nv_wr32(dev, NV03_PFIFO_CACHE1_PUT, 0);
}

int
nv40_fifo_load_context(struct nouveau_channel *chan)
{
	struct drm_device *dev = chan->dev;
	uint32_t tmp;

	nv40_fifo_do_load_context(dev, chan->id);

	/* Set channel active, and in DMA mode */
	nv_wr32(dev, NV03_PFIFO_CACHE1_PUSH1,
		     NV40_PFIFO_CACHE1_PUSH1_DMA | chan->id);
	nv_wr32(dev, NV04_PFIFO_CACHE1_DMA_PUSH, 1);

	/* Reset DMA_CTL_AT_INFO to INVALID */
	tmp = nv_rd32(dev, NV04_PFIFO_CACHE1_DMA_CTL) & ~(1 << 31);
	nv_wr32(dev, NV04_PFIFO_CACHE1_DMA_CTL, tmp);

	return 0;
}

int
nv40_fifo_unload_context(struct drm_device *dev)
{
	struct drm_nouveau_private *dev_priv = dev->dev_private;
	struct nouveau_fifo_engine *pfifo = &dev_priv->engine.fifo;
	uint32_t fc, tmp;
	int chid;

	chid = pfifo->channel_id(dev);
	if (chid < 0 || chid >= dev_priv->engine.fifo.channels)
		return 0;
	fc = NV40_RAMFC(chid);

	dev_priv->engine.instmem.prepare_access(dev, true);
	nv_wi32(dev, fc + 0, nv_rd32(dev, NV04_PFIFO_CACHE1_DMA_PUT));
	nv_wi32(dev, fc + 4, nv_rd32(dev, NV04_PFIFO_CACHE1_DMA_GET));
	nv_wi32(dev, fc + 8, nv_rd32(dev, NV10_PFIFO_CACHE1_REF_CNT));
	nv_wi32(dev, fc + 12, nv_rd32(dev, NV04_PFIFO_CACHE1_DMA_INSTANCE));
	nv_wi32(dev, fc + 16, nv_rd32(dev, NV04_PFIFO_CACHE1_DMA_DCOUNT));
	nv_wi32(dev, fc + 20, nv_rd32(dev, NV04_PFIFO_CACHE1_DMA_STATE));
	tmp  = nv_rd32(dev, NV04_PFIFO_CACHE1_DMA_FETCH);
	tmp |= nv_rd32(dev, 0x2058) & 0x30000000;
	nv_wi32(dev, fc + 24, tmp);
	nv_wi32(dev, fc + 28, nv_rd32(dev, NV04_PFIFO_CACHE1_ENGINE));
	nv_wi32(dev, fc + 32, nv_rd32(dev, NV04_PFIFO_CACHE1_PULL1));
	nv_wi32(dev, fc + 36, nv_rd32(dev, NV10_PFIFO_CACHE1_ACQUIRE_VALUE));
	tmp = nv_rd32(dev, NV10_PFIFO_CACHE1_ACQUIRE_TIMESTAMP);
	nv_wi32(dev, fc + 40, tmp);
	nv_wi32(dev, fc + 44, nv_rd32(dev, NV10_PFIFO_CACHE1_ACQUIRE_TIMEOUT));
	nv_wi32(dev, fc + 48, nv_rd32(dev, NV10_PFIFO_CACHE1_SEMAPHORE));
	/* NVIDIA read 0x3228 first, then write DMA_GET here.. maybe something
	 * more involved depending on the value of 0x3228?
	 */
	nv_wi32(dev, fc + 52, nv_rd32(dev, NV04_PFIFO_CACHE1_DMA_GET));
	nv_wi32(dev, fc + 56, nv_rd32(dev, NV40_PFIFO_GRCTX_INSTANCE));
	nv_wi32(dev, fc + 60, nv_rd32(dev, NV04_PFIFO_DMA_TIMESLICE) & 0x1ffff);
	/* No idea what the below is for exactly, ripped from a mmio-trace */
	nv_wi32(dev, fc + 64, nv_rd32(dev, NV40_PFIFO_UNK32E4));
	/* NVIDIA do this next line twice.. bug? */
	nv_wi32(dev, fc + 68, nv_rd32(dev, 0x32e8));
	nv_wi32(dev, fc + 76, nv_rd32(dev, 0x2088));
	nv_wi32(dev, fc + 80, nv_rd32(dev, 0x3300));
#if 0 /* no real idea which is PUT/GET in UNK_48.. */
	tmp  = nv_rd32(dev, NV04_PFIFO_CACHE1_GET);
	tmp |= (nv_rd32(dev, NV04_PFIFO_CACHE1_PUT) << 16);
	nv_wi32(dev, fc + 72, tmp);
#endif
	dev_priv->engine.instmem.finish_access(dev);

	nv40_fifo_do_load_context(dev, pfifo->channels - 1);
	nv_wr32(dev, NV03_PFIFO_CACHE1_PUSH1,
		     NV40_PFIFO_CACHE1_PUSH1_DMA | (pfifo->channels - 1));
	return 0;
}

static void
nv40_fifo_init_reset(struct drm_device *dev)
{
	int i;

	nv_wr32(dev, NV03_PMC_ENABLE,
		nv_rd32(dev, NV03_PMC_ENABLE) & ~NV_PMC_ENABLE_PFIFO);
	nv_wr32(dev, NV03_PMC_ENABLE,
		nv_rd32(dev, NV03_PMC_ENABLE) |  NV_PMC_ENABLE_PFIFO);

	nv_wr32(dev, 0x003224, 0x000f0078);
	nv_wr32(dev, 0x003210, 0x00000000);
	nv_wr32(dev, 0x003270, 0x00000000);
	nv_wr32(dev, 0x003240, 0x00000000);
	nv_wr32(dev, 0x003244, 0x00000000);
	nv_wr32(dev, 0x003258, 0x00000000);
	nv_wr32(dev, 0x002504, 0x00000000);
	for (i = 0; i < 16; i++)
		nv_wr32(dev, 0x002510 + (i * 4), 0x00000000);
	nv_wr32(dev, 0x00250c, 0x0000ffff);
	nv_wr32(dev, 0x002048, 0x00000000);
	nv_wr32(dev, 0x003228, 0x00000000);
	nv_wr32(dev, 0x0032e8, 0x00000000);
	nv_wr32(dev, 0x002410, 0x00000000);
	nv_wr32(dev, 0x002420, 0x00000000);
	nv_wr32(dev, 0x002058, 0x00000001);
	nv_wr32(dev, 0x00221c, 0x00000000);
	/* something with 0x2084, read/modify/write, no change */
	nv_wr32(dev, 0x002040, 0x000000ff);
	nv_wr32(dev, 0x002500, 0x00000000);
	nv_wr32(dev, 0x003200, 0x00000000);

	nv_wr32(dev, NV04_PFIFO_DMA_TIMESLICE, 0x2101ffff);
}

static void
nv40_fifo_init_ramxx(struct drm_device *dev)
{
	struct drm_nouveau_private *dev_priv = dev->dev_private;

	nv_wr32(dev, NV03_PFIFO_RAMHT, (0x03 << 24) /* search 128 */ |
				       ((dev_priv->ramht_bits - 9) << 16) |
				       (dev_priv->ramht_offset >> 8));
	nv_wr32(dev, NV03_PFIFO_RAMRO, dev_priv->ramro_offset>>8);

	switch (dev_priv->chipset) {
	case 0x47:
	case 0x49:
	case 0x4b:
		nv_wr32(dev, 0x2230, 1);
		break;
	default:
		break;
	}

	switch (dev_priv->chipset) {
	case 0x40:
	case 0x41:
	case 0x42:
	case 0x43:
	case 0x45:
	case 0x47:
	case 0x48:
	case 0x49:
	case 0x4b:
		nv_wr32(dev, NV40_PFIFO_RAMFC, 0x30002);
		break;
	default:
		nv_wr32(dev, 0x2230, 0);
		nv_wr32(dev, NV40_PFIFO_RAMFC,
			((nouveau_mem_fb_amount(dev) - 512 * 1024 +
			  dev_priv->ramfc_offset) >> 16) | (3 << 16));
		break;
	}
}

static void
nv40_fifo_init_intr(struct drm_device *dev)
{
	nv_wr32(dev, 0x002100, 0xffffffff);
	nv_wr32(dev, 0x002140, 0xffffffff);
}

int
nv40_fifo_init(struct drm_device *dev)
{
	struct drm_nouveau_private *dev_priv = dev->dev_private;
	struct nouveau_fifo_engine *pfifo = &dev_priv->engine.fifo;
	int i;

	nv40_fifo_init_reset(dev);
	nv40_fifo_init_ramxx(dev);

	nv40_fifo_do_load_context(dev, pfifo->channels - 1);
	nv_wr32(dev, NV03_PFIFO_CACHE1_PUSH1, pfifo->channels - 1);

	nv40_fifo_init_intr(dev);
	pfifo->enable(dev);
	pfifo->reassign(dev, true);

	for (i = 0; i < dev_priv->engine.fifo.channels; i++) {
		if (dev_priv->fifos[i]) {
			uint32_t mode = nv_rd32(dev, NV04_PFIFO_MODE);
			nv_wr32(dev, NV04_PFIFO_MODE, mode | (1 << i));
		}
	}

	return 0;
}