aboutsummaryrefslogtreecommitdiff
path: root/shared-core/via_dma.c
blob: 1f5d3b942d03953fe9bcf2aa9d06fdd8b7afdd42 (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
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
/* via_dma.c -- DMA support for the VIA Unichrome/Pro
 * 
 * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
 * All Rights Reserved.
 *
 * Copyright 2004 Digeo, Inc., Palo Alto, CA, U.S.A.
 * All Rights Reserved.
 * 
 * Copyright 2004 The Unichrome project.
 * 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, sub license,
 * 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 NON-INFRINGEMENT. IN NO EVENT SHALL
 * THE COPYRIGHT HOLDERS, AUTHORS 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.
 *
 * Authors: Various
 */

#include "drmP.h"
#include "drm.h"
#include "via_drm.h"
#include "via_drv.h"
#include "via_3d_reg.h"

#define CMDBUF_ALIGNMENT_SIZE   (0x100)
#define CMDBUF_ALIGNMENT_MASK   (0xff)

/* defines for VIA 3D registers */
#define VIA_REG_STATUS          0x400
#define VIA_REG_TRANSET         0x43C
#define VIA_REG_TRANSPACE       0x440

/* VIA_REG_STATUS(0x400): Engine Status */
#define VIA_CMD_RGTR_BUSY       0x00000080	/* Command Regulator is busy */
#define VIA_2D_ENG_BUSY         0x00000001	/* 2D Engine is busy */
#define VIA_3D_ENG_BUSY         0x00000002	/* 3D Engine is busy */
#define VIA_VR_QUEUE_BUSY       0x00020000	/* Virtual Queue is busy */

#define SetReg2DAGP(nReg, nData) {				\
	*((uint32_t *)(vb)) = ((nReg) >> 2) | HALCYON_HEADER1;	\
	*((uint32_t *)(vb) + 1) = (nData);			\
	vb = ((uint32_t *)vb) + 2;				\
	dev_priv->dma_low +=8;					\
}

#define via_flush_write_combine() DRM_MEMORYBARRIER()

#define VIA_OUT_RING_QW(w1,w2)			\
	*vb++ = (w1);				\
	*vb++ = (w2);				\
	dev_priv->dma_low += 8; 

static void via_cmdbuf_start(drm_via_private_t * dev_priv);
static void via_cmdbuf_pause(drm_via_private_t * dev_priv);
static void via_cmdbuf_reset(drm_via_private_t * dev_priv);
static void via_cmdbuf_rewind(drm_via_private_t * dev_priv);
static int via_wait_idle(drm_via_private_t * dev_priv);
/*
 * Free space in command buffer.
 */

static uint32_t
via_cmdbuf_space(drm_via_private_t *dev_priv)
{
	uint32_t agp_base = dev_priv->dma_offset + 
		(uint32_t) dev_priv->agpAddr;
	uint32_t hw_addr = *(dev_priv->hw_addr_ptr) - agp_base;
	
	return ((hw_addr <= dev_priv->dma_low) ? 
		(dev_priv->dma_high + hw_addr - dev_priv->dma_low) : 
		(hw_addr - dev_priv->dma_low));
}

/*
 * How much does the command regulator lag behind?
 */

static uint32_t
via_cmdbuf_lag(drm_via_private_t *dev_priv)
{
	uint32_t agp_base = dev_priv->dma_offset + 
		(uint32_t) dev_priv->agpAddr;
	uint32_t hw_addr = *(dev_priv->hw_addr_ptr) - agp_base;
	
	return ((hw_addr <= dev_priv->dma_low) ? 
		(dev_priv->dma_low - hw_addr) : 
		(dev_priv->dma_wrap + dev_priv->dma_low - hw_addr));
}

/*
 * Check that the given size fits in the buffer, otherwise wait.
 */

static inline int
via_cmdbuf_wait(drm_via_private_t * dev_priv, unsigned int size)
{
	uint32_t agp_base = dev_priv->dma_offset + (uint32_t) dev_priv->agpAddr;
	uint32_t cur_addr, hw_addr, next_addr;
	volatile uint32_t *hw_addr_ptr;
	uint32_t count;
	hw_addr_ptr = dev_priv->hw_addr_ptr;
	cur_addr = dev_priv->dma_low;
	next_addr = cur_addr + size + 512*1024;
	count = 1000000;
	do {
	        hw_addr = *hw_addr_ptr - agp_base;
		if (count-- == 0) {
			DRM_ERROR("via_cmdbuf_wait timed out hw %x cur_addr %x next_addr %x\n",
				  hw_addr, cur_addr, next_addr);
			return -1;
		}
	} while ((cur_addr < hw_addr) && (next_addr >= hw_addr));
	return 0;
}


/*
 * Checks whether buffer head has reach the end. Rewind the ring buffer
 * when necessary.
 *
 * Returns virtual pointer to ring buffer.
 */

static inline uint32_t *via_check_dma(drm_via_private_t * dev_priv,
				      unsigned int size)
{
	if ((dev_priv->dma_low + size + 4*CMDBUF_ALIGNMENT_SIZE) > dev_priv->dma_high) {
		via_cmdbuf_rewind(dev_priv);
	}
	if (via_cmdbuf_wait(dev_priv, size) != 0) {
		return NULL;
	}

	return (uint32_t *) (dev_priv->dma_ptr + dev_priv->dma_low);
}

int via_dma_cleanup(drm_device_t * dev)
{
	if (dev->dev_private) {
		drm_via_private_t *dev_priv =
			(drm_via_private_t *) dev->dev_private;

		if (dev_priv->ring.virtual_start) {
			via_cmdbuf_reset(dev_priv);

			drm_core_ioremapfree(&dev_priv->ring.map, dev);
			dev_priv->ring.virtual_start = NULL;
		}

	}

	return 0;
}

static int via_initialize(drm_device_t * dev,
			  drm_via_private_t * dev_priv,
			  drm_via_dma_init_t * init)
{
	if (!dev_priv || !dev_priv->mmio) {
		DRM_ERROR("via_dma_init called before via_map_init\n");
		return DRM_ERR(EFAULT);
	}

	if (dev_priv->ring.virtual_start != NULL) {
		DRM_ERROR("%s called again without calling cleanup\n",
			  __FUNCTION__);
		return DRM_ERR(EFAULT);
	}

	dev_priv->ring.map.offset = dev->agp->base + init->offset;
	dev_priv->ring.map.size = init->size;
	dev_priv->ring.map.type = 0;
	dev_priv->ring.map.flags = 0;
	dev_priv->ring.map.mtrr = 0;

	drm_core_ioremap(&dev_priv->ring.map, dev);

	if (dev_priv->ring.map.handle == NULL) {
		via_dma_cleanup(dev);
		DRM_ERROR("can not ioremap virtual address for"
			  " ring buffer\n");
		return DRM_ERR(ENOMEM);
	}

	dev_priv->ring.virtual_start = dev_priv->ring.map.handle;

	dev_priv->dma_ptr = dev_priv->ring.virtual_start;
	dev_priv->dma_low = 0;
	dev_priv->dma_high = init->size;
	dev_priv->dma_wrap = init->size;
	dev_priv->dma_offset = init->offset;
	dev_priv->last_pause_ptr = NULL;
	dev_priv->hw_addr_ptr = dev_priv->mmio->handle + init->reg_pause_addr;

	via_cmdbuf_start(dev_priv);

	return 0;
}

int via_dma_init(DRM_IOCTL_ARGS)
{
	DRM_DEVICE;
	drm_via_private_t *dev_priv = (drm_via_private_t *) dev->dev_private;
	drm_via_dma_init_t init;
	int retcode = 0;

	DRM_COPY_FROM_USER_IOCTL(init, (drm_via_dma_init_t *) data,
				 sizeof(init));

	switch (init.func) {
	case VIA_INIT_DMA:
		if (!capable(CAP_SYS_ADMIN))
			retcode = DRM_ERR(EPERM);
		else
			retcode = via_initialize(dev, dev_priv, &init);
		break;
	case VIA_CLEANUP_DMA:
		if (!capable(CAP_SYS_ADMIN))
			retcode = DRM_ERR(EPERM);
		else
			retcode = via_dma_cleanup(dev);
		break;
        case VIA_DMA_INITIALIZED:
		retcode = (dev_priv->ring.virtual_start != NULL) ? 
			0: DRM_ERR( EFAULT );
	        break;
	default:
		retcode = DRM_ERR(EINVAL);
		break;
	}

	return retcode;
}

static int via_dispatch_cmdbuffer(drm_device_t * dev, drm_via_cmdbuffer_t * cmd)
{
	drm_via_private_t *dev_priv;
	uint32_t *vb;
	int ret;

	dev_priv = (drm_via_private_t *) dev->dev_private;

	if (dev_priv->ring.virtual_start == NULL) {
		DRM_ERROR("%s called without initializing AGP ring buffer.\n",
			  __FUNCTION__);
		return DRM_ERR(EFAULT);
	}

	if (cmd->size > VIA_PCI_BUF_SIZE) {
		return DRM_ERR(ENOMEM);
	} 


	if (DRM_COPY_FROM_USER(dev_priv->pci_buf, cmd->buf, cmd->size))
		return DRM_ERR(EFAULT);

	/*
	 * Running this function on AGP memory is dead slow. Therefore
	 * we run it on a temporary cacheable system memory buffer and
	 * copy it to AGP memory when ready.
	 */

	
	if ((ret = via_verify_command_stream((uint32_t *)dev_priv->pci_buf, cmd->size, dev, 1))) {
		return ret;
	}
       	
	vb = via_check_dma(dev_priv, cmd->size);
	if (vb == NULL) {
		return DRM_ERR(EAGAIN);
	}

	memcpy(vb, dev_priv->pci_buf, cmd->size);
	
	dev_priv->dma_low += cmd->size;
	via_cmdbuf_pause(dev_priv);

	return 0;
}

int via_driver_dma_quiescent(drm_device_t * dev)
{
	drm_via_private_t *dev_priv = dev->dev_private;

	if (!via_wait_idle(dev_priv)) {
		return DRM_ERR(EBUSY);
	}
	return 0;
}

int via_flush_ioctl(DRM_IOCTL_ARGS)
{
	DRM_DEVICE;

	LOCK_TEST_WITH_RETURN( dev, filp );

	return via_driver_dma_quiescent(dev);
}

int via_cmdbuffer(DRM_IOCTL_ARGS)
{
	DRM_DEVICE;
	drm_via_cmdbuffer_t cmdbuf;
	int ret;

	LOCK_TEST_WITH_RETURN( dev, filp );

	DRM_COPY_FROM_USER_IOCTL(cmdbuf, (drm_via_cmdbuffer_t *) data,
				 sizeof(cmdbuf));

	DRM_DEBUG("via cmdbuffer, buf %p size %lu\n", cmdbuf.buf, cmdbuf.size);

	ret = via_dispatch_cmdbuffer(dev, &cmdbuf);
	if (ret) {
		return ret;
	}

	return 0;
}

static int via_parse_pci_cmdbuffer(drm_device_t * dev, const char *buf,
				   unsigned int size)
{
	drm_via_private_t *dev_priv = dev->dev_private;
	const uint32_t *regbuf = (const uint32_t *) buf;
	const uint32_t *regend = regbuf + (size >> 2);
	const uint32_t *next_fire; 
	int fire_count = 0;
	int ret;
	int check_2d_cmd = 1;



	if ((ret = via_verify_command_stream(regbuf, size, dev, 0)))
		return ret;

	next_fire = dev_priv->fire_offsets[fire_count];
	while (regbuf != regend) {	
		if ( *regbuf == HALCYON_HEADER2 ) {
		  
			regbuf++;
			check_2d_cmd = ( *regbuf != HALCYON_SUB_ADDR0 );
			VIA_WRITE(HC_REG_TRANS_SET + HC_REG_BASE, *regbuf++);
			
		} else if ( check_2d_cmd && ((*regbuf & HALCYON_HEADER1MASK) == HALCYON_HEADER1 )) {

			register uint32_t addr = ( (*regbuf++ ) & ~HALCYON_HEADER1MASK) << 2;
			VIA_WRITE( addr, *regbuf++ );

		} else if ( (fire_count < dev_priv->num_fire_offsets) && 
			    (regbuf == next_fire) &&
			    (( *regbuf & HALCYON_FIREMASK ) == HALCYON_FIRECMD) ) {

			next_fire = dev_priv->fire_offsets[++fire_count];

			VIA_WRITE(HC_REG_TRANS_SPACE + HC_REG_BASE, *regbuf++);

			if ( ( regbuf != regend ) && 
			     ((*regbuf & HALCYON_FIREMASK) == HALCYON_FIRECMD))
				regbuf++;
			if (( *regbuf & HALCYON_CMDBMASK ) != HC_ACMD_HCmdB )
				check_2d_cmd = 1;
		} else {

			VIA_WRITE(HC_REG_TRANS_SPACE + HC_REG_BASE , *regbuf++);

		}
	} 
	return 0;

}

static int via_dispatch_pci_cmdbuffer(drm_device_t * dev,
				      drm_via_cmdbuffer_t * cmd)
{
	drm_via_private_t *dev_priv = dev->dev_private;
	int ret;

	if (cmd->size > VIA_PCI_BUF_SIZE) {
		return DRM_ERR(ENOMEM);
	} 
	if (DRM_COPY_FROM_USER(dev_priv->pci_buf, cmd->buf, cmd->size))
		return DRM_ERR(EFAULT);
	ret = via_parse_pci_cmdbuffer(dev, dev_priv->pci_buf, cmd->size);
	return ret;
}

int via_pci_cmdbuffer(DRM_IOCTL_ARGS)
{
	DRM_DEVICE;
	drm_via_cmdbuffer_t cmdbuf;
	int ret;

	LOCK_TEST_WITH_RETURN( dev, filp );

	DRM_COPY_FROM_USER_IOCTL(cmdbuf, (drm_via_cmdbuffer_t *) data,
				 sizeof(cmdbuf));

	DRM_DEBUG("via_pci_cmdbuffer, buf %p size %lu\n", cmdbuf.buf,
		  cmdbuf.size);

	ret = via_dispatch_pci_cmdbuffer(dev, &cmdbuf);
	if (ret) {
		return ret;
	}

	return 0;
}


static inline uint32_t *via_align_buffer(drm_via_private_t * dev_priv,
					 uint32_t * vb, int qw_count)
{
	for (; qw_count > 0; --qw_count) {
		VIA_OUT_RING_QW(HC_DUMMY, HC_DUMMY);
	}
	return vb;
}


/*
 * This function is used internally by ring buffer mangement code.
 *
 * Returns virtual pointer to ring buffer.
 */
static inline uint32_t *via_get_dma(drm_via_private_t * dev_priv)
{
	return (uint32_t *) (dev_priv->dma_ptr + dev_priv->dma_low);
}

/*
 * Hooks a segment of data into the tail of the ring-buffer by
 * modifying the pause address stored in the buffer itself. If
 * the regulator has already paused, restart it.
 */
static int via_hook_segment(drm_via_private_t *dev_priv,
			    uint32_t pause_addr_hi, uint32_t pause_addr_lo,
			    int no_pci_fire)
{
	int paused, count;
	volatile uint32_t *paused_at = dev_priv->last_pause_ptr;

	via_flush_write_combine();
	while(! *(via_get_dma(dev_priv)-1));
	*dev_priv->last_pause_ptr = pause_addr_lo;
	via_flush_write_combine();

	/*
	 * The below statement is inserted to really force the flush.
	 * Not sure it is needed.
	 */

	while(! *dev_priv->last_pause_ptr);
	dev_priv->last_pause_ptr = via_get_dma(dev_priv) - 1;
	while(! *dev_priv->last_pause_ptr);


	paused = 0;
	count = 20; 

	while (!(paused = (VIA_READ(0x41c) & 0x80000000)) && count--);
	if ((count <= 8) && (count >= 0)) {
		uint32_t rgtr, ptr;
		rgtr = *(dev_priv->hw_addr_ptr);
		ptr = ((char *)dev_priv->last_pause_ptr - dev_priv->dma_ptr) + 
			dev_priv->dma_offset + (uint32_t) dev_priv->agpAddr + 4 - 
			CMDBUF_ALIGNMENT_SIZE;
		if (rgtr <= ptr) {
			DRM_ERROR("Command regulator\npaused at count %d, address %x, "
				  "while current pause address is %x.\n"
				  "Please mail this message to "
				  "<unichrome-devel@lists.sourceforge.net>\n",
				  count, rgtr, ptr);
		}
	}
		
	if (paused && !no_pci_fire) {
	        uint32_t rgtr,ptr;
		uint32_t ptr_low;

		count = 1000000;
		while ((VIA_READ(VIA_REG_STATUS) & VIA_CMD_RGTR_BUSY) && count--);
		
		rgtr = *(dev_priv->hw_addr_ptr);
		ptr = ((char *)paused_at - dev_priv->dma_ptr) + 
			dev_priv->dma_offset + (uint32_t) dev_priv->agpAddr + 4;
		

		ptr_low = (ptr > 3*CMDBUF_ALIGNMENT_SIZE) ? 
			ptr - 3*CMDBUF_ALIGNMENT_SIZE : 0;
		if (rgtr <= ptr && rgtr >= ptr_low) {
			VIA_WRITE(VIA_REG_TRANSET, (HC_ParaType_PreCR << 16));
			VIA_WRITE(VIA_REG_TRANSPACE, pause_addr_hi);
			VIA_WRITE(VIA_REG_TRANSPACE, pause_addr_lo);
		} 
	}
	return paused;
}



static int via_wait_idle(drm_via_private_t * dev_priv)
{
	int count = 10000000;
	while (count-- && (VIA_READ(VIA_REG_STATUS) &
			   (VIA_CMD_RGTR_BUSY | VIA_2D_ENG_BUSY |
			    VIA_3D_ENG_BUSY))) ;
	return count;
}

static uint32_t *via_align_cmd(drm_via_private_t * dev_priv, uint32_t cmd_type,
			       uint32_t addr, uint32_t *cmd_addr_hi, 
			       uint32_t *cmd_addr_lo,
			       int skip_wait)
{
	uint32_t agp_base;
	uint32_t cmd_addr, addr_lo, addr_hi;
	uint32_t *vb;
	uint32_t qw_pad_count;

	if (!skip_wait)
		via_cmdbuf_wait(dev_priv, 2*CMDBUF_ALIGNMENT_SIZE);

	vb = via_get_dma(dev_priv);
	VIA_OUT_RING_QW( HC_HEADER2 | ((VIA_REG_TRANSET >> 2) << 12) |
			 (VIA_REG_TRANSPACE >> 2), HC_ParaType_PreCR << 16); 
	agp_base = dev_priv->dma_offset + (uint32_t) dev_priv->agpAddr;
	qw_pad_count = (CMDBUF_ALIGNMENT_SIZE >> 3) -
		((dev_priv->dma_low & CMDBUF_ALIGNMENT_MASK) >> 3);

	
	cmd_addr = (addr) ? addr : 
		agp_base + dev_priv->dma_low - 8 + (qw_pad_count << 3);
	addr_lo = ((HC_SubA_HAGPBpL << 24) | (cmd_type & HC_HAGPBpID_MASK) |
		   (cmd_addr & HC_HAGPBpL_MASK));
	addr_hi = ((HC_SubA_HAGPBpH << 24) | (cmd_addr >> 24));

	vb = via_align_buffer(dev_priv, vb, qw_pad_count - 1);
	VIA_OUT_RING_QW(*cmd_addr_hi = addr_hi, 
			*cmd_addr_lo = addr_lo);
	return vb;
}




static void via_cmdbuf_start(drm_via_private_t * dev_priv)
{
	uint32_t pause_addr_lo, pause_addr_hi;
	uint32_t start_addr, start_addr_lo;
	uint32_t end_addr, end_addr_lo;
	uint32_t command;
	uint32_t agp_base;


	dev_priv->dma_low = 0;

	agp_base = dev_priv->dma_offset + (uint32_t) dev_priv->agpAddr;
	start_addr = agp_base;
	end_addr = agp_base + dev_priv->dma_high;

	start_addr_lo = ((HC_SubA_HAGPBstL << 24) | (start_addr & 0xFFFFFF));
	end_addr_lo = ((HC_SubA_HAGPBendL << 24) | (end_addr & 0xFFFFFF));
	command = ((HC_SubA_HAGPCMNT << 24) | (start_addr >> 24) |
		   ((end_addr & 0xff000000) >> 16));

	dev_priv->last_pause_ptr = 
		via_align_cmd(dev_priv, HC_HAGPBpID_PAUSE, 0, 
			      &pause_addr_hi, & pause_addr_lo, 1) - 1;

	via_flush_write_combine();
	while(! *dev_priv->last_pause_ptr);

	VIA_WRITE(VIA_REG_TRANSET, (HC_ParaType_PreCR << 16));
	VIA_WRITE(VIA_REG_TRANSPACE, command);
	VIA_WRITE(VIA_REG_TRANSPACE, start_addr_lo);
	VIA_WRITE(VIA_REG_TRANSPACE, end_addr_lo);

	VIA_WRITE(VIA_REG_TRANSPACE, pause_addr_hi);
	VIA_WRITE(VIA_REG_TRANSPACE, pause_addr_lo);

	VIA_WRITE(VIA_REG_TRANSPACE, command | HC_HAGPCMNT_MASK);
}

static inline void via_dummy_bitblt(drm_via_private_t * dev_priv)
{
	uint32_t *vb = via_get_dma(dev_priv);
	SetReg2DAGP(0x0C, (0 | (0 << 16)));
	SetReg2DAGP(0x10, 0 | (0 << 16));
	SetReg2DAGP(0x0, 0x1 | 0x2000 | 0xAA000000); 
}


static void via_cmdbuf_jump(drm_via_private_t * dev_priv)
{
	uint32_t agp_base;
	uint32_t pause_addr_lo, pause_addr_hi;
	uint32_t jump_addr_lo, jump_addr_hi;
	volatile uint32_t *last_pause_ptr;
	uint32_t dma_low_save1, dma_low_save2;

	agp_base = dev_priv->dma_offset + (uint32_t) dev_priv->agpAddr;
	via_align_cmd(dev_priv,  HC_HAGPBpID_JUMP, 0, &jump_addr_hi, 
		      &jump_addr_lo, 0);
	
	dev_priv->dma_wrap = dev_priv->dma_low;


	/*
	 * Wrap command buffer to the beginning.
	 */

	dev_priv->dma_low = 0;
	if (via_cmdbuf_wait(dev_priv, CMDBUF_ALIGNMENT_SIZE) != 0) {
		DRM_ERROR("via_cmdbuf_jump failed\n");
	}

	via_dummy_bitblt(dev_priv);
	via_dummy_bitblt(dev_priv); 

	last_pause_ptr = via_align_cmd(dev_priv,  HC_HAGPBpID_PAUSE, 0, &pause_addr_hi, 
				       &pause_addr_lo, 0) -1;
	via_align_cmd(dev_priv,  HC_HAGPBpID_PAUSE, 0, &pause_addr_hi, 
		      &pause_addr_lo, 0);

	*last_pause_ptr = pause_addr_lo;
	dma_low_save1 = dev_priv->dma_low;

	/*
	 * Now, set a trap that will pause the regulator if it tries to rerun the old
	 * command buffer. (Which may happen if via_hook_segment detecs a command regulator pause
	 * and reissues the jump command over PCI, while the regulator has already taken the jump
	 * and actually paused at the current buffer end).
	 * There appears to be no other way to detect this condition, since the hw_addr_pointer
	 * does not seem to get updated immediately when a jump occurs.
	 */

	last_pause_ptr = via_align_cmd(dev_priv,  HC_HAGPBpID_PAUSE, 0, &pause_addr_hi, 
				       &pause_addr_lo, 0) -1;
	via_align_cmd(dev_priv,  HC_HAGPBpID_PAUSE, 0, &pause_addr_hi, 
		      &pause_addr_lo, 0);
	*last_pause_ptr = pause_addr_lo;

	dma_low_save2 = dev_priv->dma_low;
	dev_priv->dma_low = dma_low_save1;	
	via_hook_segment( dev_priv, jump_addr_hi, jump_addr_lo, 0);
	dev_priv->dma_low = dma_low_save2;
	via_hook_segment( dev_priv, pause_addr_hi, pause_addr_lo, 0);
}


static void via_cmdbuf_rewind(drm_via_private_t * dev_priv)
{
	via_cmdbuf_jump(dev_priv); 
}

static void via_cmdbuf_flush(drm_via_private_t * dev_priv, uint32_t cmd_type)
{
	uint32_t pause_addr_lo, pause_addr_hi;

	via_align_cmd(dev_priv, cmd_type, 0, &pause_addr_hi, &pause_addr_lo, 0);
	via_hook_segment( dev_priv, pause_addr_hi, pause_addr_lo, 0);
}


static void via_cmdbuf_pause(drm_via_private_t * dev_priv)
{
	via_cmdbuf_flush(dev_priv, HC_HAGPBpID_PAUSE);
}

static void via_cmdbuf_reset(drm_via_private_t * dev_priv)
{
	via_cmdbuf_flush(dev_priv, HC_HAGPBpID_STOP);
	via_wait_idle(dev_priv);
}

/*
 * User interface to the space and lag functions.
 */

int 
via_cmdbuf_size(DRM_IOCTL_ARGS)
{
	DRM_DEVICE;
	drm_via_cmdbuf_size_t d_siz;
	int ret = 0;
	uint32_t tmp_size, count;
	drm_via_private_t *dev_priv;

	DRM_DEBUG("via cmdbuf_size\n");
	LOCK_TEST_WITH_RETURN( dev, filp );

	dev_priv = (drm_via_private_t *) dev->dev_private;

	if (dev_priv->ring.virtual_start == NULL) {
		DRM_ERROR("%s called without initializing AGP ring buffer.\n",
			  __FUNCTION__);
		return DRM_ERR(EFAULT);
	}

	DRM_COPY_FROM_USER_IOCTL(d_siz, (drm_via_cmdbuf_size_t *) data,
				 sizeof(d_siz));


	count = 1000000;
	tmp_size = d_siz.size;
	switch(d_siz.func) {
	case VIA_CMDBUF_SPACE:
		while (((tmp_size = via_cmdbuf_space(dev_priv)) < d_siz.size) && count--) {
			if (!d_siz.wait) {
				break;
			}
		}
		if (!count) {
			DRM_ERROR("VIA_CMDBUF_SPACE timed out.\n");
			ret = DRM_ERR(EAGAIN);
		}
		break;
	case VIA_CMDBUF_LAG:
		while (((tmp_size = via_cmdbuf_lag(dev_priv)) > d_siz.size) && count--) {
			if (!d_siz.wait) {
				break;
			}
		}
		if (!count) {
			DRM_ERROR("VIA_CMDBUF_LAG timed out.\n");
			ret = DRM_ERR(EAGAIN);
		}
		break;
	default:
		ret = DRM_ERR(EFAULT);
	}
	d_siz.size = tmp_size;

	DRM_COPY_TO_USER_IOCTL((drm_via_cmdbuf_size_t *) data, d_siz,
			       sizeof(d_siz));
	return ret;
}