summaryrefslogtreecommitdiff
path: root/linux/drm_memory.h
blob: 98ea89d8500ac529cb4214cb41e9794a7f64da70 (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
/** 
 * \file drm_memory.h 
 * Memory management wrappers for DRM
 *
 * \author Rickard E. (Rik) Faith <faith@valinux.com>
 * \author Gareth Hughes <gareth@valinux.com>
 */

/* 
 * Created: Thu Feb  4 14:00:34 1999 by faith@valinux.com
 *
 * Copyright 1999 Precision Insight, Inc., Cedar Park, Texas.
 * Copyright 2000 VA Linux Systems, Inc., Sunnyvale, California.
 * 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
 * VA LINUX SYSTEMS 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.
 */

#define __NO_VERSION__
#include <linux/config.h>
#include <linux/highmem.h>
#include "drmP.h"

/**
 * Cut down version of drm_memory_debug.h, which used to be called
 * drm_memory.h.  If you want the debug functionality, change 0 to 1
 * below.
 */
#define DEBUG_MEMORY 0

/* Need the 4-argument version of vmap().  */
#if __REALLY_HAVE_AGP && defined(VMAP_4_ARGS)

#include <linux/vmalloc.h>

#ifdef HAVE_PAGE_AGP
#include <asm/agp.h>
#else
# ifdef __powerpc__
#  define PAGE_AGP	__pgprot(_PAGE_KERNEL | _PAGE_NO_CACHE)
# else
#  define PAGE_AGP	PAGE_KERNEL
# endif
#endif

#if LINUX_VERSION_CODE < KERNEL_VERSION(2,5,0)
# define pte_offset_kernel(dir, address)	pte_offset(dir, address)
# define pte_pfn(pte)				(pte_page(pte) - mem_map)
# define pfn_to_page(pfn)			(mem_map + (pfn))
#endif

/*
 * Find the drm_map that covers the range [offset, offset+size).
 */
static inline drm_map_t *
drm_lookup_map (unsigned long offset, unsigned long size, drm_device_t *dev)
{
	struct list_head *list;
	drm_map_list_t *r_list;
	drm_map_t *map;

	list_for_each(list, &dev->maplist->head) {
		r_list = (drm_map_list_t *) list;
		map = r_list->map;
		if (!map)
			continue;
		if (map->offset <= offset && (offset + size) <= (map->offset + map->size))
			return map;
	}
	return NULL;
}

static inline void *
agp_remap (unsigned long offset, unsigned long size, drm_device_t *dev)
{
	unsigned long *phys_addr_map, i, num_pages = PAGE_ALIGN(size) / PAGE_SIZE;
	struct drm_agp_mem *agpmem;
	struct page **page_map;
	void *addr;

	size = PAGE_ALIGN(size);

#ifdef __alpha__
	offset -= dev->hose->mem_space->start;
#endif

	for (agpmem = dev->agp->memory; agpmem; agpmem = agpmem->next)
		if (agpmem->bound <= offset
		    && (agpmem->bound + (agpmem->pages << PAGE_SHIFT)) >= (offset + size))
			break;
	if (!agpmem)
		return NULL;

	/*
	 * OK, we're mapping AGP space on a chipset/platform on which memory accesses by
	 * the CPU do not get remapped by the GART.  We fix this by using the kernel's
	 * page-table instead (that's probably faster anyhow...).
	 */
	/* note: use vmalloc() because num_pages could be large... */
	page_map = vmalloc(num_pages * sizeof(struct page *));
	if (!page_map)
		return NULL;

	phys_addr_map = agpmem->memory->memory + (offset - agpmem->bound) / PAGE_SIZE;
	for (i = 0; i < num_pages; ++i)
		page_map[i] = pfn_to_page(phys_addr_map[i] >> PAGE_SHIFT);
	addr = vmap(page_map, num_pages, VM_IOREMAP, PAGE_AGP);
	vfree(page_map);

	return addr;
}

static inline unsigned long
drm_follow_page (void *vaddr)
{
	pgd_t *pgd = pgd_offset_k((unsigned long) vaddr);
	pmd_t *pmd = pmd_offset(pgd, (unsigned long) vaddr);
	pte_t *ptep = pte_offset_kernel(pmd, (unsigned long) vaddr);
	return pte_pfn(*ptep) << PAGE_SHIFT;
}

#endif /* __REALLY_HAVE_AGP && defined(VMAP_4_ARGS) */

static inline void *drm_ioremap(unsigned long offset, unsigned long size, drm_device_t *dev)
{
#if __REALLY_HAVE_AGP && defined(VMAP_4_ARGS)
	if (dev->agp && dev->agp->cant_use_aperture) {
		drm_map_t *map = drm_lookup_map(offset, size, dev);

		if (map && map->type == _DRM_AGP)
			return agp_remap(offset, size, dev);
	}
#endif

	return ioremap(offset, size);
}

static inline void *drm_ioremap_nocache(unsigned long offset, unsigned long size,
					drm_device_t *dev)
{
#if __REALLY_HAVE_AGP && defined(VMAP_4_ARGS)
	if (dev->agp && dev->agp->cant_use_aperture) {
		drm_map_t *map = drm_lookup_map(offset, size, dev);

		if (map && map->type == _DRM_AGP)
			return agp_remap(offset, size, dev);
	}
#endif

	return ioremap_nocache(offset, size);
}

static inline void drm_ioremapfree(void *pt, unsigned long size, drm_device_t *dev)
{
#if __REALLY_HAVE_AGP && defined(VMAP_4_ARGS)
	/*
	 * This is a bit ugly.  It would be much cleaner if the DRM API would use separate
	 * routines for handling mappings in the AGP space.  Hopefully this can be done in
	 * a future revision of the interface...
	 */
	if (dev->agp && dev->agp->cant_use_aperture
	    && ((unsigned long) pt >= VMALLOC_START && (unsigned long) pt < VMALLOC_END))
	{
		unsigned long offset;
		drm_map_t *map;

		offset = drm_follow_page(pt) | ((unsigned long) pt & ~PAGE_MASK);
		map = drm_lookup_map(offset, size, dev);
		if (map && map->type == _DRM_AGP) {
			vunmap(pt);
			return;
		}
	}
#endif

	iounmap(pt);
}

#if DEBUG_MEMORY
#include "drm_memory_debug.h"
#else

/** No-op. */
void DRM(mem_init)(void)
{
}

/**
 * Called when "/proc/dri/%dev%/mem" is read.
 * 
 * \param buf output buffer.
 * \param start start of output data.
 * \param offset requested start offset.
 * \param len requested number of bytes.
 * \param eof whether there is no more data to return.
 * \param data private data.
 * \return number of written bytes.
 *
 * No-op. 
 */
int DRM(mem_info)(char *buf, char **start, off_t offset,
		  int len, int *eof, void *data)
{
	return 0;
}

/** Wrapper around kmalloc() */
void *DRM(alloc)(size_t size, int area)
{
	return kmalloc(size, GFP_KERNEL);
}

/** Wrapper around kmalloc() */
void *DRM(calloc)(size_t size, size_t nmemb, int area)
{
	void *addr;

	addr = kmalloc(size * nmemb, GFP_KERNEL);
	if (addr != NULL)
		memset((void *)addr, 0, size * nmemb);

	return addr;
}

/** Wrapper around kmalloc() and kfree() */
void *DRM(realloc)(void *oldpt, size_t oldsize, size_t size, int area)
{
	void *pt;

	if (!(pt = kmalloc(size, GFP_KERNEL))) return NULL;
	if (oldpt && oldsize) {
		memcpy(pt, oldpt, oldsize);
		kfree(oldpt);
	}
	return pt;
}

/** Wrapper around kfree() */
void DRM(free)(void *pt, size_t size, int area)
{
	kfree(pt);
}

/**
 * Allocate pages.
 *
 * \param order size order.
 * \param area memory area. (Not used.)
 * \return page address on success, or zero on failure.
 *
 * Allocate and reserve free pages.
 */
unsigned long DRM(alloc_pages)(int order, int area)
{
	unsigned long address;
	unsigned long bytes	  = PAGE_SIZE << order;
	unsigned long addr;
	unsigned int  sz;

	address = __get_free_pages(GFP_KERNEL, order);
	if (!address) 
		return 0;

				/* Zero */
	memset((void *)address, 0, bytes);

				/* Reserve */
	for (addr = address, sz = bytes;
	     sz > 0;
	     addr += PAGE_SIZE, sz -= PAGE_SIZE) {
		SetPageReserved(virt_to_page(addr));
	}

	return address;
}

/**
 * Free pages.
 * 
 * \param address address of the pages to free.
 * \param order size order.
 * \param area memory area. (Not used.)
 *
 * Unreserve and free pages allocated by alloc_pages().
 */
void DRM(free_pages)(unsigned long address, int order, int area)
{
	unsigned long bytes = PAGE_SIZE << order;
	unsigned long addr;
	unsigned int  sz;

	if (!address) 
		return;

	/* Unreserve */
	for (addr = address, sz = bytes;
	     sz > 0;
	     addr += PAGE_SIZE, sz -= PAGE_SIZE) {
		ClearPageReserved(virt_to_page(addr));
	}

	free_pages(address, order);
}

/** Wrapper around drm_ioremap() */
void *DRM(ioremap)(unsigned long offset, unsigned long size, drm_device_t *dev)
{
	return drm_ioremap(offset, size, dev);
}

/** Wrapper around drm_ioremap_nocache() */
void *DRM(ioremap_nocache)(unsigned long offset, unsigned long size, drm_device_t *dev)
{
	return drm_ioremap_nocache(offset, size, dev);
}

/** Wrapper around drm_iounmap() */
void DRM(ioremapfree)(void *pt, unsigned long size, drm_device_t *dev)
{
	drm_ioremapfree(pt, size, dev);
}

#if __REALLY_HAVE_AGP
/** Wrapper around agp_allocate_memory() */
DRM_AGP_MEM *DRM(alloc_agp)(int pages, u32 type)
{
	return DRM(agp_allocate_memory)(pages, type);
}

/** Wrapper around agp_free_memory() */
int DRM(free_agp)(DRM_AGP_MEM *handle, int pages)
{
	return DRM(agp_free_memory)(handle) ? 0 : -EINVAL;
}

/** Wrapper around agp_bind_memory() */
int DRM(bind_agp)(DRM_AGP_MEM *handle, unsigned int start)
{
	return DRM(agp_bind_memory)(handle, start);
}

/** Wrapper around agp_unbind_memory() */
int DRM(unbind_agp)(DRM_AGP_MEM *handle)
{
	return DRM(agp_unbind_memory)(handle);
}
#endif /* agp */
#endif /* debug_memory */
a id='n1362' href='#n1362'>1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 1419 1420 1421
/* i810_dma.c -- DMA support for the i810 -*- linux-c -*-
 * Created: Mon Dec 13 01:50:01 1999 by jhartmann@precisioninsight.com
 *
 * Copyright 1999 Precision Insight, Inc., Cedar Park, Texas.
 * Copyright 2000 VA Linux Systems, Inc., Sunnyvale, California.
 * 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
 * PRECISION INSIGHT 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: Rickard E. (Rik) Faith <faith@valinux.com>
 *	    Jeff Hartmann <jhartmann@valinux.com>
 *          Keith Whitwell <keith@tungstengraphics.com>
 *
 */

#include "i810.h"
#include "drmP.h"
#include "drm.h"
#include "i810_drm.h"
#include "i810_drv.h"
#include <linux/interrupt.h>	/* For task queue support */
#include <linux/delay.h>
#include <linux/pagemap.h>

#ifdef DO_MUNMAP_4_ARGS
#define DO_MUNMAP(m, a, l)	do_munmap(m, a, l, 1)
#else
#define DO_MUNMAP(m, a, l)	do_munmap(m, a, l)
#endif

#define I810_BUF_FREE		2
#define I810_BUF_CLIENT		1
#define I810_BUF_HARDWARE      	0

#define I810_BUF_UNMAPPED 0
#define I810_BUF_MAPPED   1

#if LINUX_VERSION_CODE <= KERNEL_VERSION(2,4,2)
#define down_write down
#define up_write up
#endif

static inline void i810_print_status_page(drm_device_t *dev)
{
   	drm_device_dma_t *dma = dev->dma;
      	drm_i810_private_t *dev_priv = dev->dev_private;
	u32 *temp = dev_priv->hw_status_page;
   	int i;

   	DRM_DEBUG(  "hw_status: Interrupt Status : %x\n", temp[0]);
   	DRM_DEBUG(  "hw_status: LpRing Head ptr : %x\n", temp[1]);
   	DRM_DEBUG(  "hw_status: IRing Head ptr : %x\n", temp[2]);
      	DRM_DEBUG(  "hw_status: Reserved : %x\n", temp[3]);
	DRM_DEBUG(  "hw_status: Last Render: %x\n", temp[4]);
   	DRM_DEBUG(  "hw_status: Driver Counter : %d\n", temp[5]);
   	for(i = 6; i < dma->buf_count + 6; i++) {
	   	DRM_DEBUG( "buffer status idx : %d used: %d\n", i - 6, temp[i]);
	}
}

static drm_buf_t *i810_freelist_get(drm_device_t *dev)
{
   	drm_device_dma_t *dma = dev->dma;
	int		 i;
   	int 		 used;

	/* Linear search might not be the best solution */

   	for (i = 0; i < dma->buf_count; i++) {
	   	drm_buf_t *buf = dma->buflist[ i ];
	   	drm_i810_buf_priv_t *buf_priv = buf->dev_private;
		/* In use is already a pointer */
	   	used = cmpxchg(buf_priv->in_use, I810_BUF_FREE,
			       I810_BUF_CLIENT);
		if (used == I810_BUF_FREE) {
			return buf;
		}
	}
   	return NULL;
}

/* This should only be called if the buffer is not sent to the hardware
 * yet, the hardware updates in use for us once its on the ring buffer.
 */

static int i810_freelist_put(drm_device_t *dev, drm_buf_t *buf)
{
   	drm_i810_buf_priv_t *buf_priv = buf->dev_private;
   	int used;

   	/* In use is already a pointer */
   	used = cmpxchg(buf_priv->in_use, I810_BUF_CLIENT, I810_BUF_FREE);
	if (used != I810_BUF_CLIENT) {
	   	DRM_ERROR("Freeing buffer thats not in use : %d\n", buf->idx);
	   	return -EINVAL;
	}

   	return 0;
}

static struct file_operations i810_buffer_fops = {
	.open	 = DRM(open),
	.flush	 = DRM(flush),
	.release = DRM(release),
	.ioctl	 = DRM(ioctl),
	.mmap	 = i810_mmap_buffers,
	.fasync  = DRM(fasync),
};

int i810_mmap_buffers(struct file *filp, struct vm_area_struct *vma)
{
	drm_file_t	    *priv	  = filp->private_data;
	drm_device_t	    *dev;
	drm_i810_private_t  *dev_priv;
	drm_buf_t           *buf;
	drm_i810_buf_priv_t *buf_priv;

	lock_kernel();
	dev	 = priv->dev;
	dev_priv = dev->dev_private;
	buf      = dev_priv->mmap_buffer;
	buf_priv = buf->dev_private;

	vma->vm_flags |= (VM_IO | VM_DONTCOPY);
	vma->vm_file = filp;

   	buf_priv->currently_mapped = I810_BUF_MAPPED;
	unlock_kernel();

	if (remap_page_range(DRM_RPR_ARG(vma) vma->vm_start,
			     VM_OFFSET(vma),
			     vma->vm_end - vma->vm_start,
			     vma->vm_page_prot)) return -EAGAIN;
	return 0;
}

static int i810_map_buffer(drm_buf_t *buf, struct file *filp)
{
	drm_file_t	  *priv	  = filp->private_data;
	drm_device_t	  *dev	  = priv->dev;
	drm_i810_buf_priv_t *buf_priv = buf->dev_private;
      	drm_i810_private_t *dev_priv = dev->dev_private;
   	struct file_operations *old_fops;
	int retcode = 0;

	if (buf_priv->currently_mapped == I810_BUF_MAPPED) 
		return -EINVAL;

	down_write( &current->mm->mmap_sem );
	old_fops = filp->f_op;
	filp->f_op = &i810_buffer_fops;
	dev_priv->mmap_buffer = buf;
	buf_priv->virtual = (void *)do_mmap(filp, 0, buf->total,
					    PROT_READ|PROT_WRITE,
					    MAP_SHARED,
					    buf->bus_address);
	dev_priv->mmap_buffer = NULL;
	filp->f_op = old_fops;
	if ((unsigned long)buf_priv->virtual > -1024UL) {
		/* Real error */
		DRM_ERROR("mmap error\n");
		retcode = (signed int)buf_priv->virtual;
		buf_priv->virtual = NULL;
	}
	up_write( &current->mm->mmap_sem );

	return retcode;
}

static int i810_unmap_buffer(drm_buf_t *buf)
{
	drm_i810_buf_priv_t *buf_priv = buf->dev_private;
	int retcode = 0;

	if (buf_priv->currently_mapped != I810_BUF_MAPPED)
		return -EINVAL;

	down_write(&current->mm->mmap_sem);
	retcode = DO_MUNMAP(current->mm,
			    (unsigned long)buf_priv->virtual,
			    (size_t) buf->total);
	up_write(&current->mm->mmap_sem);

   	buf_priv->currently_mapped = I810_BUF_UNMAPPED;
   	buf_priv->virtual = NULL;

	return retcode;
}

static int i810_dma_get_buffer(drm_device_t *dev, drm_i810_dma_t *d,
			       struct file *filp)
{
	drm_buf_t	  *buf;
	drm_i810_buf_priv_t *buf_priv;
	int retcode = 0;

	buf = i810_freelist_get(dev);
	if (!buf) {
		retcode = -ENOMEM;
	   	DRM_DEBUG("retcode=%d\n", retcode);
		return retcode;
	}

	retcode = i810_map_buffer(buf, filp);
	if (retcode) {
		i810_freelist_put(dev, buf);
	   	DRM_ERROR("mapbuf failed, retcode %d\n", retcode);
		return retcode;
	}
	buf->filp = filp;
	buf_priv = buf->dev_private;
	d->granted = 1;
   	d->request_idx = buf->idx;
   	d->request_size = buf->total;
   	d->virtual = buf_priv->virtual;

	return retcode;
}

int i810_dma_cleanup(drm_device_t *dev)
{
	drm_device_dma_t *dma = dev->dma;

	/* Make sure interrupts are disabled here because the uninstall ioctl
	 * may not have been called from userspace and after dev_private
	 * is freed, it's too late.
	 */
	if ( drm_core_check_feature(dev, DRIVER_HAVE_IRQ) && dev->irq_enabled )
		DRM(irq_uninstall)(dev);

	if (dev->dev_private) {
		int i;
	   	drm_i810_private_t *dev_priv =
	     		(drm_i810_private_t *) dev->dev_private;

		if (dev_priv->ring.virtual_start) {
		   	DRM(ioremapfree)((void *) dev_priv->ring.virtual_start,
					 dev_priv->ring.Size, dev);
		}
	   	if (dev_priv->hw_status_page) {
		   	pci_free_consistent(dev->pdev, PAGE_SIZE,
					    dev_priv->hw_status_page,
					    dev_priv->dma_status_page);
		   	/* Need to rewrite hardware status page */
		   	I810_WRITE(0x02080, 0x1ffff000);
		}
	   	DRM(free)(dev->dev_private, sizeof(drm_i810_private_t),
			 DRM_MEM_DRIVER);
	   	dev->dev_private = NULL;

		for (i = 0; i < dma->buf_count; i++) {
			drm_buf_t *buf = dma->buflist[ i ];
			drm_i810_buf_priv_t *buf_priv = buf->dev_private;
			if ( buf_priv->kernel_virtual && buf->total )
				DRM(ioremapfree)(buf_priv->kernel_virtual, buf->total, dev);
		}
	}
   	return 0;
}

static int i810_wait_ring(drm_device_t *dev, int n)
{
   	drm_i810_private_t *dev_priv = dev->dev_private;
   	drm_i810_ring_buffer_t *ring = &(dev_priv->ring);
   	int iters = 0;
   	unsigned long end;
	unsigned int last_head = I810_READ(LP_RING + RING_HEAD) & HEAD_ADDR;

	end = jiffies + (HZ*3);
   	while (ring->space < n) {
	   	ring->head = I810_READ(LP_RING + RING_HEAD) & HEAD_ADDR;
	   	ring->space = ring->head - (ring->tail+8);
		if (ring->space < 0) ring->space += ring->Size;
	   
		if (ring->head != last_head) {
			end = jiffies + (HZ*3);
			last_head = ring->head;
		}
	  
	   	iters++;
		if (time_before(end, jiffies)) {
		   	DRM_ERROR("space: %d wanted %d\n", ring->space, n);
		   	DRM_ERROR("lockup\n");
		   	goto out_wait_ring;
		}
		udelay(1);
	}

out_wait_ring:
   	return iters;
}

static void i810_kernel_lost_context(drm_device_t *dev)
{
      	drm_i810_private_t *dev_priv = dev->dev_private;
   	drm_i810_ring_buffer_t *ring = &(dev_priv->ring);

   	ring->head = I810_READ(LP_RING + RING_HEAD) & HEAD_ADDR;
     	ring->tail = I810_READ(LP_RING + RING_TAIL);
     	ring->space = ring->head - (ring->tail+8);
     	if (ring->space < 0) ring->space += ring->Size;
}

static int i810_freelist_init(drm_device_t *dev, drm_i810_private_t *dev_priv)
{
      	drm_device_dma_t *dma = dev->dma;
   	int my_idx = 24;
   	u32 *hw_status = (u32 *)(dev_priv->hw_status_page + my_idx);
   	int i;

	if (dma->buf_count > 1019) {
	   	/* Not enough space in the status page for the freelist */
	   	return -EINVAL;
	}

   	for (i = 0; i < dma->buf_count; i++) {
	   	drm_buf_t *buf = dma->buflist[ i ];
	   	drm_i810_buf_priv_t *buf_priv = buf->dev_private;

	   	buf_priv->in_use = hw_status++;
	   	buf_priv->my_use_idx = my_idx;
	   	my_idx += 4;

	   	*buf_priv->in_use = I810_BUF_FREE;

		buf_priv->kernel_virtual = DRM(ioremap)(buf->bus_address,
							buf->total, dev);
	}
	return 0;
}

static int i810_dma_initialize(drm_device_t *dev,
			       drm_i810_private_t *dev_priv,
			       drm_i810_init_t *init)
{
	struct list_head *list;

   	memset(dev_priv, 0, sizeof(drm_i810_private_t));

	list_for_each(list, &dev->maplist->head) {
		drm_map_list_t *r_list = list_entry(list, drm_map_list_t, head);
		if (r_list->map &&
		    r_list->map->type == _DRM_SHM &&
		    r_list->map->flags & _DRM_CONTAINS_LOCK ) {
			dev_priv->sarea_map = r_list->map;
 			break;
 		}
 	}
	if (!dev_priv->sarea_map) {
		dev->dev_private = (void *)dev_priv;
	   	i810_dma_cleanup(dev);
	   	DRM_ERROR("can not find sarea!\n");
	   	return -EINVAL;
	}
	dev_priv->mmio_map = drm_core_findmap(dev, init->mmio_offset);
	if (!dev_priv->mmio_map) {
		dev->dev_private = (void *)dev_priv;
	   	i810_dma_cleanup(dev);
	   	DRM_ERROR("can not find mmio map!\n");
	   	return -EINVAL;
	}
	dev->agp_buffer_map = drm_core_findmap(dev, init->buffers_offset);
	if (!dev->agp_buffer_map) {
		dev->dev_private = (void *)dev_priv;
	   	i810_dma_cleanup(dev);
	   	DRM_ERROR("can not find dma buffer map!\n");
	   	return -EINVAL;
	}

	dev_priv->sarea_priv = (drm_i810_sarea_t *)
		((u8 *)dev_priv->sarea_map->handle +
		 init->sarea_priv_offset);

   	dev_priv->ring.Start = init->ring_start;
   	dev_priv->ring.End = init->ring_end;
   	dev_priv->ring.Size = init->ring_size;

   	dev_priv->ring.virtual_start = DRM(ioremap)(dev->agp->base +
						    init->ring_start,
						    init->ring_size, dev);

   	if (dev_priv->ring.virtual_start == NULL) {
		dev->dev_private = (void *) dev_priv;
	   	i810_dma_cleanup(dev);
	   	DRM_ERROR("can not ioremap virtual address for"
			  " ring buffer\n");
	   	return -ENOMEM;
	}

   	dev_priv->ring.tail_mask = dev_priv->ring.Size - 1;

	dev_priv->w = init->w;
	dev_priv->h = init->h;
	dev_priv->pitch = init->pitch;
	dev_priv->back_offset = init->back_offset;
	dev_priv->depth_offset = init->depth_offset;
	dev_priv->front_offset = init->front_offset;

	dev_priv->overlay_offset = init->overlay_offset;
	dev_priv->overlay_physical = init->overlay_physical;

	dev_priv->front_di1 = init->front_offset | init->pitch_bits;
	dev_priv->back_di1 = init->back_offset | init->pitch_bits;
	dev_priv->zi1 = init->depth_offset | init->pitch_bits;

   	/* Program Hardware Status Page */
   	dev_priv->hw_status_page =
		pci_alloc_consistent(dev->pdev, PAGE_SIZE,
						&dev_priv->dma_status_page);
   	if (!dev_priv->hw_status_page) {
		dev->dev_private = (void *)dev_priv;
		i810_dma_cleanup(dev);
		DRM_ERROR("Can not allocate hardware status page\n");
		return -ENOMEM;
	}
   	memset(dev_priv->hw_status_page, 0, PAGE_SIZE);
   	DRM_DEBUG("hw status page @ %p\n", dev_priv->hw_status_page);

	I810_WRITE(0x02080, dev_priv->dma_status_page);
   	DRM_DEBUG("Enabled hardware status page\n");

   	/* Now we need to init our freelist */
	if (i810_freelist_init(dev, dev_priv) != 0) {
		dev->dev_private = (void *)dev_priv;
	   	i810_dma_cleanup(dev);
	   	DRM_ERROR("Not enough space in the status page for"
			  " the freelist\n");
	   	return -ENOMEM;
	}
	dev->dev_private = (void *)dev_priv;

   	return 0;
}

/* i810 DRM version 1.1 used a smaller init structure with different
 * ordering of values than is currently used (drm >= 1.2). There is
 * no defined way to detect the XFree version to correct this problem,
 * however by checking using this procedure we can detect the correct
 * thing to do.
 *
 * #1 Read the Smaller init structure from user-space
 * #2 Verify the overlay_physical is a valid physical address, or NULL
 *    If it isn't then we have a v1.1 client. Fix up params.
 *    If it is, then we have a 1.2 client... get the rest of the data.
 */
int i810_dma_init_compat(drm_i810_init_t *init, unsigned long arg)
{

	/* Get v1.1 init data */
	if (copy_from_user(init, (drm_i810_pre12_init_t __user *)arg,
			  sizeof(drm_i810_pre12_init_t))) {
		return -EFAULT;
	}

	if ((!init->overlay_physical) || (init->overlay_physical > 4096)) {

		/* This is a v1.2 client, just get the v1.2 init data */
		DRM_INFO("Using POST v1.2 init.\n");
		if (copy_from_user(init, (drm_i810_init_t __user *)arg,
				   sizeof(drm_i810_init_t))) {
			return -EFAULT;
		}
	} else {

		/* This is a v1.1 client, fix the params */
		DRM_INFO("Using PRE v1.2 init.\n");
	 	init->pitch_bits = init->h;
	 	init->pitch = init->w;
	 	init->h = init->overlay_physical;
	 	init->w = init->overlay_offset;
	 	init->overlay_physical = 0;
	 	init->overlay_offset = 0;
	}

	return 0;
}

int i810_dma_init(struct inode *inode, struct file *filp,
		  unsigned int cmd, unsigned long arg)
{
   	drm_file_t *priv = filp->private_data;
   	drm_device_t *dev = priv->dev;
   	drm_i810_private_t *dev_priv;
   	drm_i810_init_t init;
   	int retcode = 0;

	/* Get only the init func */
	if (copy_from_user(&init, (void __user *)arg, sizeof(drm_i810_init_func_t))) 
		return -EFAULT;

   	switch(init.func) {
	 	case I810_INIT_DMA:
	 	       	/* This case is for backward compatibility. It
			 * handles XFree 4.1.0 and 4.2.0, and has to
			 * do some parameter checking as described below.
			 * It will someday go away.
			 */
			retcode = i810_dma_init_compat(&init, arg);
			if (retcode)
				return retcode;

	   		dev_priv = DRM(alloc)(sizeof(drm_i810_private_t),
					     DRM_MEM_DRIVER);
	   		if (dev_priv == NULL)
				return -ENOMEM;
	   		retcode = i810_dma_initialize(dev, dev_priv, &init);
			break;

		default:
	 	case I810_INIT_DMA_1_4:
			DRM_INFO("Using v1.4 init.\n");
  			if (copy_from_user(&init, (drm_i810_init_t __user *)arg,
					  sizeof(drm_i810_init_t))) {
				return -EFAULT;
			}
	   		dev_priv = DRM(alloc)(sizeof(drm_i810_private_t),
					     DRM_MEM_DRIVER);
			if (dev_priv == NULL) 
				return -ENOMEM;
	   		retcode = i810_dma_initialize(dev, dev_priv, &init);
			break;

	 	case I810_CLEANUP_DMA:
		        DRM_INFO("DMA Cleanup\n");
	   		retcode = i810_dma_cleanup(dev);
              	   	break;
	}

   	return retcode;
}



/* Most efficient way to verify state for the i810 is as it is
 * emitted.  Non-conformant state is silently dropped.
 *
 * Use 'volatile' & local var tmp to force the emitted values to be
 * identical to the verified ones.
 */
static void i810EmitContextVerified( drm_device_t *dev,
				     volatile unsigned int *code )
{
   	drm_i810_private_t *dev_priv = dev->dev_private;
	int i, j = 0;
	unsigned int tmp;
	RING_LOCALS;

	BEGIN_LP_RING( I810_CTX_SETUP_SIZE );

	OUT_RING( GFX_OP_COLOR_FACTOR );
	OUT_RING( code[I810_CTXREG_CF1] );

	OUT_RING( GFX_OP_STIPPLE );
	OUT_RING( code[I810_CTXREG_ST1] );

	for ( i = 4 ; i < I810_CTX_SETUP_SIZE ; i++ ) {
		tmp = code[i];

		if ((tmp & (7<<29)) == (3<<29) &&
		    (tmp & (0x1f<<24)) < (0x1d<<24))
		{
			OUT_RING( tmp );
			j++;
		}
		else printk("constext state dropped!!!\n");
	}

	if (j & 1)
		OUT_RING( 0 );

	ADVANCE_LP_RING();
}

static void i810EmitTexVerified( drm_device_t *dev,
				 volatile unsigned int *code )
{
   	drm_i810_private_t *dev_priv = dev->dev_private;
	int i, j = 0;
	unsigned int tmp;
	RING_LOCALS;

	BEGIN_LP_RING( I810_TEX_SETUP_SIZE );

	OUT_RING( GFX_OP_MAP_INFO );
	OUT_RING( code[I810_TEXREG_MI1] );
	OUT_RING( code[I810_TEXREG_MI2] );
	OUT_RING( code[I810_TEXREG_MI3] );

	for ( i = 4 ; i < I810_TEX_SETUP_SIZE ; i++ ) {
		tmp = code[i];

		if ((tmp & (7<<29)) == (3<<29) &&
		    (tmp & (0x1f<<24)) < (0x1d<<24))
		{
			OUT_RING( tmp );
			j++;
		}
		else printk("texture state dropped!!!\n");
	}

	if (j & 1)
		OUT_RING( 0 );

	ADVANCE_LP_RING();
}


/* Need to do some additional checking when setting the dest buffer.
 */
static void i810EmitDestVerified( drm_device_t *dev,
				  volatile unsigned int *code )
{
   	drm_i810_private_t *dev_priv = dev->dev_private;
	unsigned int tmp;
	RING_LOCALS;

	BEGIN_LP_RING( I810_DEST_SETUP_SIZE + 2 );

	tmp = code[I810_DESTREG_DI1];
	if (tmp == dev_priv->front_di1 || tmp == dev_priv->back_di1) {
		OUT_RING( CMD_OP_DESTBUFFER_INFO );
		OUT_RING( tmp );
	} else
	   DRM_DEBUG("bad di1 %x (allow %x or %x)\n",
		     tmp, dev_priv->front_di1, dev_priv->back_di1);

	/* invarient:
	 */
	OUT_RING( CMD_OP_Z_BUFFER_INFO );
	OUT_RING( dev_priv->zi1 );

	OUT_RING( GFX_OP_DESTBUFFER_VARS );
	OUT_RING( code[I810_DESTREG_DV1] );

	OUT_RING( GFX_OP_DRAWRECT_INFO );
	OUT_RING( code[I810_DESTREG_DR1] );
	OUT_RING( code[I810_DESTREG_DR2] );
	OUT_RING( code[I810_DESTREG_DR3] );
	OUT_RING( code[I810_DESTREG_DR4] );
	OUT_RING( 0 );

	ADVANCE_LP_RING();
}



static void i810EmitState( drm_device_t *dev )
{
   	drm_i810_private_t *dev_priv = dev->dev_private;
      	drm_i810_sarea_t *sarea_priv = dev_priv->sarea_priv;
	unsigned int dirty = sarea_priv->dirty;
	
	DRM_DEBUG("%s %x\n", __FUNCTION__, dirty);

	if (dirty & I810_UPLOAD_BUFFERS) {
		i810EmitDestVerified( dev, sarea_priv->BufferState );
		sarea_priv->dirty &= ~I810_UPLOAD_BUFFERS;
	}

	if (dirty & I810_UPLOAD_CTX) {
		i810EmitContextVerified( dev, sarea_priv->ContextState );
		sarea_priv->dirty &= ~I810_UPLOAD_CTX;
	}

	if (dirty & I810_UPLOAD_TEX0) {
		i810EmitTexVerified( dev, sarea_priv->TexState[0] );
		sarea_priv->dirty &= ~I810_UPLOAD_TEX0;
	}

	if (dirty & I810_UPLOAD_TEX1) {
		i810EmitTexVerified( dev, sarea_priv->TexState[1] );
		sarea_priv->dirty &= ~I810_UPLOAD_TEX1;
	}
}



/* need to verify
 */
static void i810_dma_dispatch_clear( drm_device_t *dev, int flags,
				     unsigned int clear_color,
				     unsigned int clear_zval )
{
   	drm_i810_private_t *dev_priv = dev->dev_private;
      	drm_i810_sarea_t *sarea_priv = dev_priv->sarea_priv;
	int nbox = sarea_priv->nbox;
	drm_clip_rect_t *pbox = sarea_priv->boxes;
	int pitch = dev_priv->pitch;
	int cpp = 2;
	int i;
	RING_LOCALS;
	
	if ( dev_priv->current_page == 1 ) {
	        unsigned int tmp = flags;
	       
		flags &= ~(I810_FRONT | I810_BACK);
		if (tmp & I810_FRONT) flags |= I810_BACK;
		if (tmp & I810_BACK) flags |= I810_FRONT;
	}

  	i810_kernel_lost_context(dev);

      	if (nbox > I810_NR_SAREA_CLIPRECTS)
     		nbox = I810_NR_SAREA_CLIPRECTS;

	for (i = 0 ; i < nbox ; i++, pbox++) {
		unsigned int x = pbox->x1;
		unsigned int y = pbox->y1;
		unsigned int width = (pbox->x2 - x) * cpp;
		unsigned int height = pbox->y2 - y;
		unsigned int start = y * pitch + x * cpp;

		if (pbox->x1 > pbox->x2 ||
		    pbox->y1 > pbox->y2 ||
		    pbox->x2 > dev_priv->w ||
		    pbox->y2 > dev_priv->h)
			continue;

	   	if ( flags & I810_FRONT ) {
			BEGIN_LP_RING( 6 );
			OUT_RING( BR00_BITBLT_CLIENT |
				  BR00_OP_COLOR_BLT | 0x3 );
			OUT_RING( BR13_SOLID_PATTERN | (0xF0 << 16) | pitch );
			OUT_RING( (height << 16) | width );
			OUT_RING( start );
			OUT_RING( clear_color );
			OUT_RING( 0 );
			ADVANCE_LP_RING();
		}

		if ( flags & I810_BACK ) {
			BEGIN_LP_RING( 6 );
			OUT_RING( BR00_BITBLT_CLIENT |
				  BR00_OP_COLOR_BLT | 0x3 );
			OUT_RING( BR13_SOLID_PATTERN | (0xF0 << 16) | pitch );
			OUT_RING( (height << 16) | width );
			OUT_RING( dev_priv->back_offset + start );
			OUT_RING( clear_color );
			OUT_RING( 0 );
			ADVANCE_LP_RING();
		}

		if ( flags & I810_DEPTH ) {
			BEGIN_LP_RING( 6 );
			OUT_RING( BR00_BITBLT_CLIENT |
				  BR00_OP_COLOR_BLT | 0x3 );
			OUT_RING( BR13_SOLID_PATTERN | (0xF0 << 16) | pitch );
			OUT_RING( (height << 16) | width );
			OUT_RING( dev_priv->depth_offset + start );
			OUT_RING( clear_zval );
			OUT_RING( 0 );
			ADVANCE_LP_RING();
		}
	}
}

static void i810_dma_dispatch_swap( drm_device_t *dev )
{
   	drm_i810_private_t *dev_priv = dev->dev_private;
      	drm_i810_sarea_t *sarea_priv = dev_priv->sarea_priv;
	int nbox = sarea_priv->nbox;
	drm_clip_rect_t *pbox = sarea_priv->boxes;
	int pitch = dev_priv->pitch;
	int cpp = 2;
	int i;
	RING_LOCALS;

	DRM_DEBUG("swapbuffers\n");

  	i810_kernel_lost_context(dev);

      	if (nbox > I810_NR_SAREA_CLIPRECTS)
     		nbox = I810_NR_SAREA_CLIPRECTS;

	for (i = 0 ; i < nbox; i++, pbox++)
	{
		unsigned int w = pbox->x2 - pbox->x1;
		unsigned int h = pbox->y2 - pbox->y1;
		unsigned int dst = pbox->x1*cpp + pbox->y1*pitch;
		unsigned int start = dst;

		if (pbox->x1 > pbox->x2 ||
		    pbox->y1 > pbox->y2 ||
		    pbox->x2 > dev_priv->w ||
		    pbox->y2 > dev_priv->h)
			continue;

		BEGIN_LP_RING( 6 );
		OUT_RING( BR00_BITBLT_CLIENT | BR00_OP_SRC_COPY_BLT | 0x4 );
		OUT_RING( pitch | (0xCC << 16));
		OUT_RING( (h << 16) | (w * cpp));
		if (dev_priv->current_page == 0)
		  OUT_RING(dev_priv->front_offset + start);
		else
		  OUT_RING(dev_priv->back_offset + start);
		OUT_RING( pitch );
		if (dev_priv->current_page == 0)
		  OUT_RING(dev_priv->back_offset + start);
		else
		  OUT_RING(dev_priv->front_offset + start);
		ADVANCE_LP_RING();
	}
}


static void i810_dma_dispatch_vertex(drm_device_t *dev,
				     drm_buf_t *buf,
				     int discard,
				     int used)
{
   	drm_i810_private_t *dev_priv = dev->dev_private;
	drm_i810_buf_priv_t *buf_priv = buf->dev_private;
   	drm_i810_sarea_t *sarea_priv = dev_priv->sarea_priv;
   	drm_clip_rect_t *box = sarea_priv->boxes;
   	int nbox = sarea_priv->nbox;
	unsigned long address = (unsigned long)buf->bus_address;
	unsigned long start = address - dev->agp->base;
	int i = 0;
   	RING_LOCALS;

   	i810_kernel_lost_context(dev);

   	if (nbox > I810_NR_SAREA_CLIPRECTS)
		nbox = I810_NR_SAREA_CLIPRECTS;

	if (used > 4*1024)
		used = 0;

	if (sarea_priv->dirty)
	   i810EmitState( dev );

	if (buf_priv->currently_mapped == I810_BUF_MAPPED) {
		unsigned int prim = (sarea_priv->vertex_prim & PR_MASK);

		*(u32 *)buf_priv->kernel_virtual = ((GFX_OP_PRIMITIVE | prim | ((used/4)-2)));

		if (used & 4) {
			*(u32 *)((u32)buf_priv->kernel_virtual + used) = 0;
			used += 4;
		}

		i810_unmap_buffer(buf);
	}

	if (used) {
		do {
			if (i < nbox) {
				BEGIN_LP_RING(4);
				OUT_RING( GFX_OP_SCISSOR | SC_UPDATE_SCISSOR |
					  SC_ENABLE );
				OUT_RING( GFX_OP_SCISSOR_INFO );
				OUT_RING( box[i].x1 | (box[i].y1<<16) );
				OUT_RING( (box[i].x2-1) | ((box[i].y2-1)<<16) );
				ADVANCE_LP_RING();
			}

			BEGIN_LP_RING(4);
			OUT_RING( CMD_OP_BATCH_BUFFER );
			OUT_RING( start | BB1_PROTECTED );
			OUT_RING( start + used - 4 );
			OUT_RING( 0 );
			ADVANCE_LP_RING();

		} while (++i < nbox);
	}

	if (discard) {
		dev_priv->counter++;

		(void) cmpxchg(buf_priv->in_use, I810_BUF_CLIENT,
			       I810_BUF_HARDWARE);

		BEGIN_LP_RING(8);
		OUT_RING( CMD_STORE_DWORD_IDX );
		OUT_RING( 20 );
		OUT_RING( dev_priv->counter );
		OUT_RING( CMD_STORE_DWORD_IDX );
		OUT_RING( buf_priv->my_use_idx );
		OUT_RING( I810_BUF_FREE );
		OUT_RING( CMD_REPORT_HEAD );
		OUT_RING( 0 );
		ADVANCE_LP_RING();
	}