aboutsummaryrefslogblamecommitdiffstats
path: root/arch/um/drivers/ubd_kern.c
blob: 9a56ff94308dff7a06a5b0fd5cff8ed3b14249ad (plain) (tree)
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
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
1253
1254
1255
1256
1257
1258
1259
1260
1261
1262
1263
1264
1265
1266
1267
1268
1269
1270
1271
1272
1273
1274
1275
1276
1277
1278
1279
1280
1281
1282
1283
1284
1285
1286
1287
1288
1289
1290
1291
1292
1293
1294
1295
1296
1297
1298
1299
1300
1301
1302
1303
1304
1305
1306
1307
1308
1309
1310
1311
1312
1313
1314
1315
1316
1317
1318
1319
1320
1321
1322
1323
1324
1325
1326
1327
1328
1329
1330
1331
1332
1333
1334
1335
1336
1337
1338
1339
1340
1341
1342
1343
1344
1345
1346
1347
1348
1349
1350
1351
1352
1353
1354
1355
1356
1357
1358
1359
1360
1361
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
1422
1423
1424
1425
1426
1427
1428
1429
1430
1431
1432
1433
1434
1435
1436
1437
1438
1439
1440
1441
1442
1443
1444
1445
1446
1447
1448
1449
1450
1451
1452
1453
1454
1455
1456
1457
1458
1459
1460
1461
1462
1463
1464
1465
1466
1467
1468
1469
1470
1471
1472
1473
1474
1475
1476
1477
1478
1479
1480
1481
1482
1483
1484
1485
1486
1487
1488
1489
1490
1491
1492
1493
1494
1495
1496
1497
1498
1499
1500
1501
1502
1503
1504
1505
1506
1507
1508
1509
1510
1511
1512
1513
1514
1515
1516
1517
1518
1519
1520
1521
1522
1523
1524
1525
1526
1527
1528
1529
1530
1531
1532
1533
1534
1535
1536
1537
1538
1539
1540
1541
1542
1543
1544
1545
1546
1547
1548
1549
1550
1551
1552
1553
1554
1555
1556
1557
1558
1559
1560
1561
1562
1563
1564
1565
1566
1567
1568
1569
1570
1571
1572
1573
1574
1575
1576
1577
1578
1579
1580
1581
1582
1583
1584
1585
1586
1587
1588
1589
1590
1591
1592
1593
1594
1595
1596
1597
1598
1599
1600
1601
1602
1603
1604
1605
1606
1607
1608
1609
1610
1611
1612
1613
1614





























































































































































                                                                             
                                                


































































































































































































































































































































































































































































































































































































































































































































































































                                                                                                




                                                                              

                                                     
           




































































































































































                                                                                
                                           


















































































































































                                                                                
                                                   









































































































































































































































































































































































                                                                                
                        



























































                                                                               
/* 
 * Copyright (C) 2000 Jeff Dike (jdike@karaya.com)
 * Licensed under the GPL
 */

/* 2001-09-28...2002-04-17
 * Partition stuff by James_McMechan@hotmail.com
 * old style ubd by setting UBD_SHIFT to 0
 * 2002-09-27...2002-10-18 massive tinkering for 2.5
 * partitions have changed in 2.5
 * 2003-01-29 more tinkering for 2.5.59-1
 * This should now address the sysfs problems and has
 * the symlink for devfs to allow for booting with
 * the common /dev/ubd/discX/... names rather than
 * only /dev/ubdN/discN this version also has lots of
 * clean ups preparing for ubd-many.
 * James McMechan
 */

#define MAJOR_NR UBD_MAJOR
#define UBD_SHIFT 4

#include "linux/config.h"
#include "linux/module.h"
#include "linux/blkdev.h"
#include "linux/hdreg.h"
#include "linux/init.h"
#include "linux/devfs_fs_kernel.h"
#include "linux/cdrom.h"
#include "linux/proc_fs.h"
#include "linux/ctype.h"
#include "linux/capability.h"
#include "linux/mm.h"
#include "linux/vmalloc.h"
#include "linux/blkpg.h"
#include "linux/genhd.h"
#include "linux/spinlock.h"
#include "asm/segment.h"
#include "asm/uaccess.h"
#include "asm/irq.h"
#include "asm/types.h"
#include "asm/tlbflush.h"
#include "user_util.h"
#include "mem_user.h"
#include "kern_util.h"
#include "kern.h"
#include "mconsole_kern.h"
#include "init.h"
#include "irq_user.h"
#include "irq_kern.h"
#include "ubd_user.h"
#include "2_5compat.h"
#include "os.h"
#include "mem.h"
#include "mem_kern.h"
#include "cow.h"

enum ubd_req { UBD_READ, UBD_WRITE, UBD_MMAP };

struct io_thread_req {
	enum ubd_req op;
	int fds[2];
	unsigned long offsets[2];
	unsigned long long offset;
	unsigned long length;
	char *buffer;
	int sectorsize;
	unsigned long sector_mask;
	unsigned long long cow_offset;
	unsigned long bitmap_words[2];
	int map_fd;
	unsigned long long map_offset;
	int error;
};

extern int open_ubd_file(char *file, struct openflags *openflags,
			 char **backing_file_out, int *bitmap_offset_out,
			 unsigned long *bitmap_len_out, int *data_offset_out,
			 int *create_cow_out);
extern int create_cow_file(char *cow_file, char *backing_file,
			   struct openflags flags, int sectorsize,
			   int alignment, int *bitmap_offset_out,
			   unsigned long *bitmap_len_out,
			   int *data_offset_out);
extern int read_cow_bitmap(int fd, void *buf, int offset, int len);
extern void do_io(struct io_thread_req *req);

static inline int ubd_test_bit(__u64 bit, unsigned char *data)
{
	__u64 n;
	int bits, off;

	bits = sizeof(data[0]) * 8;
	n = bit / bits;
	off = bit % bits;
	return((data[n] & (1 << off)) != 0);
}

static inline void ubd_set_bit(__u64 bit, unsigned char *data)
{
	__u64 n;
	int bits, off;

	bits = sizeof(data[0]) * 8;
	n = bit / bits;
	off = bit % bits;
	data[n] |= (1 << off);
}
/*End stuff from ubd_user.h*/

#define DRIVER_NAME "uml-blkdev"

static DEFINE_SPINLOCK(ubd_io_lock);
static DEFINE_SPINLOCK(ubd_lock);

static void (*do_ubd)(void);

static int ubd_open(struct inode * inode, struct file * filp);
static int ubd_release(struct inode * inode, struct file * file);
static int ubd_ioctl(struct inode * inode, struct file * file,
		     unsigned int cmd, unsigned long arg);

#define MAX_DEV (8)

/* Changed in early boot */
static int ubd_do_mmap = 0;
#define UBD_MMAP_BLOCK_SIZE PAGE_SIZE

static struct block_device_operations ubd_blops = {
        .owner		= THIS_MODULE,
        .open		= ubd_open,
        .release	= ubd_release,
        .ioctl		= ubd_ioctl,
};

/* Protected by the queue_lock */
static request_queue_t *ubd_queue;

/* Protected by ubd_lock */
static int fake_major = MAJOR_NR;

static struct gendisk *ubd_gendisk[MAX_DEV];
static struct gendisk *fake_gendisk[MAX_DEV];
 
#ifdef CONFIG_BLK_DEV_UBD_SYNC
#define OPEN_FLAGS ((struct openflags) { .r = 1, .w = 1, .s = 1, .c = 0, \
					 .cl = 1 })
#else
#define OPEN_FLAGS ((struct openflags) { .r = 1, .w = 1, .s = 0, .c = 0, \
					 .cl = 1 })
#endif

/* Not protected - changed only in ubd_setup_common and then only to
 * to enable O_SYNC.
 */
static struct openflags global_openflags = OPEN_FLAGS;

struct cow {
	/* This is the backing file, actually */
	char *file;
	int fd;
	unsigned long *bitmap;
	unsigned long bitmap_len;
	int bitmap_offset;
        int data_offset;
};

struct ubd {
	char *file;
	int count;
	int fd;
	__u64 size;
	struct openflags boot_openflags;
	struct openflags openflags;
	int no_cow;
	struct cow cow;
	struct platform_device pdev;

	int map_writes;
	int map_reads;
	int nomap_writes;
	int nomap_reads;
	int write_maps;
};

#define DEFAULT_COW { \
	.file =			NULL, \
        .fd =			-1, \
        .bitmap =		NULL, \
	.bitmap_offset =	0, \
        .data_offset =		0, \
}

#define DEFAULT_UBD { \
	.file = 		NULL, \
	.count =		0, \
	.fd =			-1, \
	.size =			-1, \
	.boot_openflags =	OPEN_FLAGS, \
	.openflags =		OPEN_FLAGS, \
        .no_cow =               0, \
        .cow =			DEFAULT_COW, \
	.map_writes		= 0, \
	.map_reads		= 0, \
	.nomap_writes		= 0, \
	.nomap_reads		= 0, \
	.write_maps		= 0, \
}

struct ubd ubd_dev[MAX_DEV] = { [ 0 ... MAX_DEV - 1 ] = DEFAULT_UBD };

static int ubd0_init(void)
{
	struct ubd *dev = &ubd_dev[0];

	if(dev->file == NULL)
		dev->file = "root_fs";
	return(0);
}

__initcall(ubd0_init);

/* Only changed by fake_ide_setup which is a setup */
static int fake_ide = 0;
static struct proc_dir_entry *proc_ide_root = NULL;
static struct proc_dir_entry *proc_ide = NULL;

static void make_proc_ide(void)
{
	proc_ide_root = proc_mkdir("ide", NULL);
	proc_ide = proc_mkdir("ide0", proc_ide_root);
}

static int proc_ide_read_media(char *page, char **start, off_t off, int count,
			       int *eof, void *data)
{
	int len;

	strcpy(page, "disk\n");
	len = strlen("disk\n");
	len -= off;
	if (len < count){
		*eof = 1;
		if (len <= 0) return 0;
	}
	else len = count;
	*start = page + off;
	return len;
}

static void make_ide_entries(char *dev_name)
{
	struct proc_dir_entry *dir, *ent;
	char name[64];

	if(proc_ide_root == NULL) make_proc_ide();

	dir = proc_mkdir(dev_name, proc_ide);
	if(!dir) return;

	ent = create_proc_entry("media", S_IFREG|S_IRUGO, dir);
	if(!ent) return;
	ent->nlink = 1;
	ent->data = NULL;
	ent->read_proc = proc_ide_read_media;
	ent->write_proc = NULL;
	sprintf(name,"ide0/%s", dev_name);
	proc_symlink(dev_name, proc_ide_root, name);
}

static int fake_ide_setup(char *str)
{
	fake_ide = 1;
	return(1);
}

__setup("fake_ide", fake_ide_setup);

__uml_help(fake_ide_setup,
"fake_ide\n"
"    Create ide0 entries that map onto ubd devices.\n\n"
);

static int parse_unit(char **ptr)
{
	char *str = *ptr, *end;
	int n = -1;

	if(isdigit(*str)) {
		n = simple_strtoul(str, &end, 0);
		if(end == str)
			return(-1);
		*ptr = end;
	}
	else if (('a' <= *str) && (*str <= 'h')) {
		n = *str - 'a';
		str++;
		*ptr = str;
	}
	return(n);
}

static int ubd_setup_common(char *str, int *index_out)
{
	struct ubd *dev;
	struct openflags flags = global_openflags;
	char *backing_file;
	int n, err, i;

	if(index_out) *index_out = -1;
	n = *str;
	if(n == '='){
		char *end;
		int major;

		str++;
		if(!strcmp(str, "mmap")){
			CHOOSE_MODE(printk("mmap not supported by the ubd "
					   "driver in tt mode\n"),
				    ubd_do_mmap = 1);
			return(0);
		}

		if(!strcmp(str, "sync")){
			global_openflags = of_sync(global_openflags);
			return(0);
		}
		major = simple_strtoul(str, &end, 0);
		if((*end != '\0') || (end == str)){
			printk(KERN_ERR 
			       "ubd_setup : didn't parse major number\n");
			return(1);
		}

		err = 1;
 		spin_lock(&ubd_lock);
 		if(fake_major != MAJOR_NR){
 			printk(KERN_ERR "Can't assign a fake major twice\n");
 			goto out1;
 		}
 
 		fake_major = major;

		printk(KERN_INFO "Setting extra ubd major number to %d\n",
		       major);
 		err = 0;
 	out1:
 		spin_unlock(&ubd_lock);
		return(err);
	}

	n = parse_unit(&str);
	if(n < 0){
		printk(KERN_ERR "ubd_setup : couldn't parse unit number "
		       "'%s'\n", str);
		return(1);
	}
	if(n >= MAX_DEV){
		printk(KERN_ERR "ubd_setup : index %d out of range "
		       "(%d devices, from 0 to %d)\n", n, MAX_DEV, MAX_DEV - 1);
		return(1);
	}

	err = 1;
	spin_lock(&ubd_lock);

	dev = &ubd_dev[n];
	if(dev->file != NULL){
		printk(KERN_ERR "ubd_setup : device already configured\n");
		goto out;
	}

	if (index_out)
		*index_out = n;

	for (i = 0; i < 4; i++) {
		switch (*str) {
		case 'r':
			flags.w = 0;
			break;
		case 's':
			flags.s = 1;
			break;
		case 'd':
			dev->no_cow = 1;
			break;
		case '=':
			str++;
			goto break_loop;
		default:
			printk(KERN_ERR "ubd_setup : Expected '=' or flag letter (r,s or d)\n");
			goto out;
		}
		str++;
	}

        if (*str == '=')
		printk(KERN_ERR "ubd_setup : Too many flags specified\n");
        else
		printk(KERN_ERR "ubd_setup : Expected '='\n");
	goto out;

break_loop:
	err = 0;
	backing_file = strchr(str, ',');

	if (!backing_file) {
		backing_file = strchr(str, ':');
	}

	if(backing_file){
		if(dev->no_cow)
			printk(KERN_ERR "Can't specify both 'd' and a "
			       "cow file\n");
		else {
			*backing_file = '\0';
			backing_file++;
		}
	}
	dev->file = str;
	dev->cow.file = backing_file;
	dev->boot_openflags = flags;
out:
	spin_unlock(&ubd_lock);
	return(err);
}

static int ubd_setup(char *str)
{
	ubd_setup_common(str, NULL);
	return(1);
}

__setup("ubd", ubd_setup);
__uml_help(ubd_setup,
"ubd<n><flags>=<filename>[(:|,)<filename2>]\n"
"    This is used to associate a device with a file in the underlying\n"
"    filesystem. When specifying two filenames, the first one is the\n"
"    COW name and the second is the backing file name. As separator you can\n"
"    use either a ':' or a ',': the first one allows writing things like;\n"
"	ubd0=~/Uml/root_cow:~/Uml/root_backing_file\n"
"    while with a ',' the shell would not expand the 2nd '~'.\n"
"    When using only one filename, UML will detect whether to thread it like\n"
"    a COW file or a backing file. To override this detection, add the 'd'\n"
"    flag:\n"
"	ubd0d=BackingFile\n"
"    Usually, there is a filesystem in the file, but \n"
"    that's not required. Swap devices containing swap files can be\n"
"    specified like this. Also, a file which doesn't contain a\n"
"    filesystem can have its contents read in the virtual \n"
"    machine by running 'dd' on the device. <n> must be in the range\n"
"    0 to 7. Appending an 'r' to the number will cause that device\n"
"    to be mounted read-only. For example ubd1r=./ext_fs. Appending\n"
"    an 's' will cause data to be written to disk on the host immediately.\n\n"
);

static int udb_setup(char *str)
{
	printk("udb%s specified on command line is almost certainly a ubd -> "
	       "udb TYPO\n", str);
	return(1);
}

__setup("udb", udb_setup);
__uml_help(udb_setup,
"udb\n"
"    This option is here solely to catch ubd -> udb typos, which can be\n\n"
"    to impossible to catch visually unless you specifically look for\n\n"
"    them.  The only result of any option starting with 'udb' is an error\n\n"
"    in the boot output.\n\n"
);

static int fakehd_set = 0;
static int fakehd(char *str)
{
	printk(KERN_INFO "fakehd : Changing ubd name to \"hd\".\n");
	fakehd_set = 1;
	return 1;
}

__setup("fakehd", fakehd);
__uml_help(fakehd,
"fakehd\n"
"    Change the ubd device name to \"hd\".\n\n"
);

static void do_ubd_request(request_queue_t * q);

/* Only changed by ubd_init, which is an initcall. */
int thread_fd = -1;

/* Changed by ubd_handler, which is serialized because interrupts only
 * happen on CPU 0.
 */
int intr_count = 0;

/* call ubd_finish if you need to serialize */
static void __ubd_finish(struct request *req, int error)
{
	int nsect;

	if(error){
		end_request(req, 0);
		return;
	}
	nsect = req->current_nr_sectors;
	req->sector += nsect;
	req->buffer += nsect << 9;
	req->errors = 0;
	req->nr_sectors -= nsect;
	req->current_nr_sectors = 0;
	end_request(req, 1);
}

static inline void ubd_finish(struct request *req, int error)
{
 	spin_lock(&ubd_io_lock);
	__ubd_finish(req, error);
	spin_unlock(&ubd_io_lock);
}

/* Called without ubd_io_lock held */
static void ubd_handler(void)
{
	struct io_thread_req req;
	struct request *rq = elv_next_request(ubd_queue);
	int n, err;

	do_ubd = NULL;
	intr_count++;
	n = os_read_file(thread_fd, &req, sizeof(req));
	if(n != sizeof(req)){
		printk(KERN_ERR "Pid %d - spurious interrupt in ubd_handler, "
		       "err = %d\n", os_getpid(), -n);
		spin_lock(&ubd_io_lock);
		end_request(rq, 0);
		spin_unlock(&ubd_io_lock);
		return;
	}
        
	if((req.op != UBD_MMAP) &&
	   ((req.offset != ((__u64) (rq->sector)) << 9) ||
	    (req.length != (rq->current_nr_sectors) << 9)))
		panic("I/O op mismatch");
	
	if(req.map_fd != -1){
		err = physmem_subst_mapping(req.buffer, req.map_fd,
					    req.map_offset, 1);
		if(err)
			printk("ubd_handler - physmem_subst_mapping failed, "
			       "err = %d\n", -err);
	}

	ubd_finish(rq, req.error);
	reactivate_fd(thread_fd, UBD_IRQ);	
	do_ubd_request(ubd_queue);
}

static irqreturn_t ubd_intr(int irq, void *dev, struct pt_regs *unused)
{
	ubd_handler();
	return(IRQ_HANDLED);
}

/* Only changed by ubd_init, which is an initcall. */
static int io_pid = -1;

void kill_io_thread(void)
{
	if(io_pid != -1) 
		os_kill_process(io_pid, 1);
}

__uml_exitcall(kill_io_thread);

static int ubd_file_size(struct ubd *dev, __u64 *size_out)
{
	char *file;

	file = dev->cow.file ? dev->cow.file : dev->file;
	return(os_file_size(file, size_out));
}

static void ubd_close(struct ubd *dev)
{
	if(ubd_do_mmap)
		physmem_forget_descriptor(dev->fd);
	os_close_file(dev->fd);
	if(dev->cow.file == NULL)
		return;

	if(ubd_do_mmap)
		physmem_forget_descriptor(dev->cow.fd);
	os_close_file(dev->cow.fd);
	vfree(dev->cow.bitmap);
	dev->cow.bitmap = NULL;
}

static int ubd_open_dev(struct ubd *dev)
{
	struct openflags flags;
	char **back_ptr;
	int err, create_cow, *create_ptr;

	dev->openflags = dev->boot_openflags;
	create_cow = 0;
	create_ptr = (dev->cow.file != NULL) ? &create_cow : NULL;
	back_ptr = dev->no_cow ? NULL : &dev->cow.file;
	dev->fd = open_ubd_file(dev->file, &dev->openflags, back_ptr,
				&dev->cow.bitmap_offset, &dev->cow.bitmap_len, 
				&dev->cow.data_offset, create_ptr);

	if((dev->fd == -ENOENT) && create_cow){
		dev->fd = create_cow_file(dev->file, dev->cow.file, 
					  dev->openflags, 1 << 9, PAGE_SIZE,
					  &dev->cow.bitmap_offset, 
					  &dev->cow.bitmap_len,
					  &dev->cow.data_offset);
		if(dev->fd >= 0){
			printk(KERN_INFO "Creating \"%s\" as COW file for "
			       "\"%s\"\n", dev->file, dev->cow.file);
		}
	}

	if(dev->fd < 0){
		printk("Failed to open '%s', errno = %d\n", dev->file,
		       -dev->fd);
		return(dev->fd);
	}

	if(dev->cow.file != NULL){
		err = -ENOMEM;
		dev->cow.bitmap = (void *) vmalloc(dev->cow.bitmap_len);
		if(dev->cow.bitmap == NULL){
			printk(KERN_ERR "Failed to vmalloc COW bitmap\n");
			goto error;
		}
		flush_tlb_kernel_vm();

		err = read_cow_bitmap(dev->fd, dev->cow.bitmap, 
				      dev->cow.bitmap_offset, 
				      dev->cow.bitmap_len);
		if(err < 0)
			goto error;

		flags = dev->openflags;
		flags.w = 0;
		err = open_ubd_file(dev->cow.file, &flags, NULL, NULL, NULL, 
				    NULL, NULL);
		if(err < 0) goto error;
		dev->cow.fd = err;
	}
	return(0);
 error:
	os_close_file(dev->fd);
	return(err);
}

static int ubd_new_disk(int major, u64 size, int unit,
			struct gendisk **disk_out)
			
{
	struct gendisk *disk;
	char from[sizeof("ubd/nnnnn\0")], to[sizeof("discnnnnn/disc\0")];
	int err;

	disk = alloc_disk(1 << UBD_SHIFT);
	if(disk == NULL)
		return(-ENOMEM);

	disk->major = major;
	disk->first_minor = unit << UBD_SHIFT;
	disk->fops = &ubd_blops;
	set_capacity(disk, size / 512);
	if(major == MAJOR_NR){
		sprintf(disk->disk_name, "ubd%c", 'a' + unit);
		sprintf(disk->devfs_name, "ubd/disc%d", unit);
		sprintf(from, "ubd/%d", unit);
		sprintf(to, "disc%d/disc", unit);
		err = devfs_mk_symlink(from, to);
		if(err)
			printk("ubd_new_disk failed to make link from %s to "
			       "%s, error = %d\n", from, to, err);
	}
	else {
		sprintf(disk->disk_name, "ubd_fake%d", unit);
		sprintf(disk->devfs_name, "ubd_fake/disc%d", unit);
	}

	/* sysfs register (not for ide fake devices) */
	if (major == MAJOR_NR) {
		ubd_dev[unit].pdev.id   = unit;
		ubd_dev[unit].pdev.name = DRIVER_NAME;
		platform_device_register(&ubd_dev[unit].pdev);
		disk->driverfs_dev = &ubd_dev[unit].pdev.dev;
	}

	disk->private_data = &ubd_dev[unit];
	disk->queue = ubd_queue;
	add_disk(disk);

	*disk_out = disk;
	return 0;
}

#define ROUND_BLOCK(n) ((n + ((1 << 9) - 1)) & (-1 << 9))

static int ubd_add(int n)
{
	struct ubd *dev = &ubd_dev[n];
	int err;

	if(dev->file == NULL)
		return(-ENODEV);

	if (ubd_open_dev(dev))
		return(-ENODEV);

	err = ubd_file_size(dev, &dev->size);
	if(err < 0)
		return(err);

	dev->size = ROUND_BLOCK(dev->size);

	err = ubd_new_disk(MAJOR_NR, dev->size, n, &ubd_gendisk[n]);
	if(err) 
		return(err);
 
	if(fake_major != MAJOR_NR)
		ubd_new_disk(fake_major, dev->size, n, 
			     &fake_gendisk[n]);

	/* perhaps this should also be under the "if (fake_major)" above */
	/* using the fake_disk->disk_name and also the fakehd_set name */
	if (fake_ide)
		make_ide_entries(ubd_gendisk[n]->disk_name);

	ubd_close(dev);
	return 0;
}

static int ubd_config(char *str)
{
	int n, err;

	str = uml_strdup(str);
	if(str == NULL){
		printk(KERN_ERR "ubd_config failed to strdup string\n");
		return(1);
	}
	err = ubd_setup_common(str, &n);
	if(err){
		kfree(str);
		return(-1);
	}
	if(n == -1) return(0);

 	spin_lock(&ubd_lock);
	err = ubd_add(n);
	if(err)
		ubd_dev[n].file = NULL;
 	spin_unlock(&ubd_lock);

	return(err);
}

static int ubd_get_config(char *name, char *str, int size, char **error_out)
{
	struct ubd *dev;
	int n, len = 0;

	n = parse_unit(&name);
	if((n >= MAX_DEV) || (n < 0)){
		*error_out = "ubd_get_config : device number out of range";
		return(-1);
	}

	dev = &ubd_dev[n];
	spin_lock(&ubd_lock);

	if(dev->file == NULL){
		CONFIG_CHUNK(str, size, len, "", 1);
		goto out;
	}

	CONFIG_CHUNK(str, size, len, dev->file, 0);

	if(dev->cow.file != NULL){
		CONFIG_CHUNK(str, size, len, ",", 0);
		CONFIG_CHUNK(str, size, len, dev->cow.file, 1);
	}
	else CONFIG_CHUNK(str, size, len, "", 1);

 out:
	spin_unlock(&ubd_lock);
	return(len);
}

static int ubd_remove(char *str)
{
	struct ubd *dev;
	int n, err = -ENODEV;

	n = parse_unit(&str);

	if((n < 0) || (n >= MAX_DEV))
		return(err);

	dev = &ubd_dev[n];
	if(dev->count > 0)
		return(-EBUSY);	/* you cannot remove a open disk */

	err = 0;
 	spin_lock(&ubd_lock);

	if(ubd_gendisk[n] == NULL)
		goto out;

	del_gendisk(ubd_gendisk[n]);
	put_disk(ubd_gendisk[n]);
	ubd_gendisk[n] = NULL;

	if(fake_gendisk[n] != NULL){
		del_gendisk(fake_gendisk[n]);
		put_disk(fake_gendisk[n]);
		fake_gendisk[n] = NULL;
	}

	platform_device_unregister(&dev->pdev);
	*dev = ((struct ubd) DEFAULT_UBD);
	err = 0;
 out:
 	spin_unlock(&ubd_lock);
	return(err);
}

static struct mc_device ubd_mc = {
	.name		= "ubd",
	.config		= ubd_config,
 	.get_config	= ubd_get_config,
	.remove		= ubd_remove,
};

static int ubd_mc_init(void)
{
	mconsole_register_dev(&ubd_mc);
	return 0;
}

__initcall(ubd_mc_init);

static struct device_driver ubd_driver = {
	.name  = DRIVER_NAME,
	.bus   = &platform_bus_type,
};

int ubd_init(void)
{
        int i;

	devfs_mk_dir("ubd");
	if (register_blkdev(MAJOR_NR, "ubd"))
		return -1;

	ubd_queue = blk_init_queue(do_ubd_request, &ubd_io_lock);
	if (!ubd_queue) {
		unregister_blkdev(MAJOR_NR, "ubd");
		return -1;
	}
		
	if (fake_major != MAJOR_NR) {
		char name[sizeof("ubd_nnn\0")];

		snprintf(name, sizeof(name), "ubd_%d", fake_major);
		devfs_mk_dir(name);
		if (register_blkdev(fake_major, "ubd"))
			return -1;
	}
	driver_register(&ubd_driver);
	for (i = 0; i < MAX_DEV; i++) 
		ubd_add(i);
	return 0;
}

late_initcall(ubd_init);

int ubd_driver_init(void){
	unsigned long stack;
	int err;

	/* Set by CONFIG_BLK_DEV_UBD_SYNC or ubd=sync.*/
	if(global_openflags.s){
		printk(KERN_INFO "ubd: Synchronous mode\n");
		/* Letting ubd=sync be like using ubd#s= instead of ubd#= is
		 * enough. So use anyway the io thread. */
	}
	stack = alloc_stack(0, 0);
	io_pid = start_io_thread(stack + PAGE_SIZE - sizeof(void *), 
				 &thread_fd);
	if(io_pid < 0){
		printk(KERN_ERR 
		       "ubd : Failed to start I/O thread (errno = %d) - "
		       "falling back to synchronous I/O\n", -io_pid);
		io_pid = -1;
		return(0);
	}
	err = um_request_irq(UBD_IRQ, thread_fd, IRQ_READ, ubd_intr, 
			     SA_INTERRUPT, "ubd", ubd_dev);
	if(err != 0)
		printk(KERN_ERR "um_request_irq failed - errno = %d\n", -err);
	return(err);
}

device_initcall(ubd_driver_init);

static int ubd_open(struct inode *inode, struct file *filp)
{
	struct gendisk *disk = inode->i_bdev->bd_disk;
	struct ubd *dev = disk->private_data;
	int err = 0;

	if(dev->count == 0){
		err = ubd_open_dev(dev);
		if(err){
			printk(KERN_ERR "%s: Can't open \"%s\": errno = %d\n",
			       disk->disk_name, dev->file, -err);
			goto out;
		}
	}
	dev->count++;
	set_disk_ro(disk, !dev->openflags.w);

	/* This should no more be needed. And it didn't work anyway to exclude
	 * read-write remounting of filesystems.*/
	/*if((filp->f_mode & FMODE_WRITE) && !dev->openflags.w){
	        if(--dev->count == 0) ubd_close(dev);
	        err = -EROFS;
	}*/
 out:
	return(err);
}

static int ubd_release(struct inode * inode, struct file * file)
{
	struct gendisk *disk = inode->i_bdev->bd_disk;
	struct ubd *dev = disk->private_data;

	if(--dev->count == 0)
		ubd_close(dev);
	return(0);
}

static void cowify_bitmap(__u64 io_offset, int length, unsigned long *cow_mask,
			  __u64 *cow_offset, unsigned long *bitmap,
			  __u64 bitmap_offset, unsigned long *bitmap_words,
			  __u64 bitmap_len)
{
	__u64 sector = io_offset >> 9;
	int i, update_bitmap = 0;

	for(i = 0; i < length >> 9; i++){
		if(cow_mask != NULL)
			ubd_set_bit(i, (unsigned char *) cow_mask);
		if(ubd_test_bit(sector + i, (unsigned char *) bitmap))
			continue;

		update_bitmap = 1;
		ubd_set_bit(sector + i, (unsigned char *) bitmap);
	}

	if(!update_bitmap)
		return;

	*cow_offset = sector / (sizeof(unsigned long) * 8);

	/* This takes care of the case where we're exactly at the end of the
	 * device, and *cow_offset + 1 is off the end.  So, just back it up
	 * by one word.  Thanks to Lynn Kerby for the fix and James McMechan
	 * for the original diagnosis.
	 */
	if(*cow_offset == ((bitmap_len + sizeof(unsigned long) - 1) /
			   sizeof(unsigned long) - 1))
		(*cow_offset)--;

	bitmap_words[0] = bitmap[*cow_offset];
	bitmap_words[1] = bitmap[*cow_offset + 1];

	*cow_offset *= sizeof(unsigned long);
	*cow_offset += bitmap_offset;
}

static void cowify_req(struct io_thread_req *req, unsigned long *bitmap,
		       __u64 bitmap_offset, __u64 bitmap_len)
{
	__u64 sector = req->offset >> 9;
	int i;

	if(req->length > (sizeof(req->sector_mask) * 8) << 9)
		panic("Operation too long");

	if(req->op == UBD_READ) {
		for(i = 0; i < req->length >> 9; i++){
			if(ubd_test_bit(sector + i, (unsigned char *) bitmap))
				ubd_set_bit(i, (unsigned char *) 
					    &req->sector_mask);
                }
	}
	else cowify_bitmap(req->offset, req->length, &req->sector_mask,
			   &req->cow_offset, bitmap, bitmap_offset,
			   req->bitmap_words, bitmap_len);
}

static int mmap_fd(struct request *req, struct ubd *dev, __u64 offset)
{
	__u64 sector;
	unsigned char *bitmap;
	int bit, i;

	/* mmap must have been requested on the command line */
	if(!ubd_do_mmap)
		return(-1);

	/* The buffer must be page aligned */
	if(((unsigned long) req->buffer % UBD_MMAP_BLOCK_SIZE) != 0)
		return(-1);

	/* The request must be a page long */
	if((req->current_nr_sectors << 9) != PAGE_SIZE)
		return(-1);

	if(dev->cow.file == NULL)
		return(dev->fd);

	sector = offset >> 9;
	bitmap = (unsigned char *) dev->cow.bitmap;
	bit = ubd_test_bit(sector, bitmap);

	for(i = 1; i < req->current_nr_sectors; i++){
		if(ubd_test_bit(sector + i, bitmap) != bit)
			return(-1);
	}

	if(bit || (rq_data_dir(req) == WRITE))
		offset += dev->cow.data_offset;

	/* The data on disk must be page aligned */
	if((offset % UBD_MMAP_BLOCK_SIZE) != 0)
		return(-1);

	return(bit ? dev->fd : dev->cow.fd);
}

static int prepare_mmap_request(struct ubd *dev, int fd, __u64 offset,
				struct request *req,
				struct io_thread_req *io_req)
{
	int err;

	if(rq_data_dir(req) == WRITE){
		/* Writes are almost no-ops since the new data is already in the
		 * host page cache
		 */
		dev->map_writes++;
		if(dev->cow.file != NULL)
			cowify_bitmap(io_req->offset, io_req->length,
				      &io_req->sector_mask, &io_req->cow_offset,
				      dev->cow.bitmap, dev->cow.bitmap_offset,
				      io_req->bitmap_words,
				      dev->cow.bitmap_len);
	}
	else {
		int w;

		if((dev->cow.file != NULL) && (fd == dev->cow.fd))
			w = 0;
		else w = dev->openflags.w;

		if((dev->cow.file != NULL) && (fd == dev->fd))
			offset += dev->cow.data_offset;

		err = physmem_subst_mapping(req->buffer, fd, offset, w);
		if(err){
			printk("physmem_subst_mapping failed, err = %d\n",
			       -err);
			return(1);
		}
		dev->map_reads++;
	}
	io_req->op = UBD_MMAP;
	io_req->buffer = req->buffer;
	return(0);
}

/* Called with ubd_io_lock held */
static int prepare_request(struct request *req, struct io_thread_req *io_req)
{
	struct gendisk *disk = req->rq_disk;
	struct ubd *dev = disk->private_data;
	__u64 offset;
	int len, fd;

	if(req->rq_status == RQ_INACTIVE) return(1);

	/* This should be impossible now */
	if((rq_data_dir(req) == WRITE) && !dev->openflags.w){
		printk("Write attempted on readonly ubd device %s\n", 
		       disk->disk_name);
		end_request(req, 0);
		return(1);
	}

	offset = ((__u64) req->sector) << 9;
	len = req->current_nr_sectors << 9;

	io_req->fds[0] = (dev->cow.file != NULL) ? dev->cow.fd : dev->fd;
	io_req->fds[1] = dev->fd;
	io_req->map_fd = -1;
	io_req->cow_offset = -1;
	io_req->offset = offset;
	io_req->length = len;
	io_req->error = 0;
	io_req->sector_mask = 0;

	fd = mmap_fd(req, dev, io_req->offset);
	if(fd > 0){
		/* If mmapping is otherwise OK, but the first access to the
		 * page is a write, then it's not mapped in yet.  So we have
		 * to write the data to disk first, then we can map the disk
		 * page in and continue normally from there.
		 */
		if((rq_data_dir(req) == WRITE) && !is_remapped(req->buffer)){
			io_req->map_fd = dev->fd;
			io_req->map_offset = io_req->offset +
				dev->cow.data_offset;
			dev->write_maps++;
		}
		else return(prepare_mmap_request(dev, fd, io_req->offset, req,
						 io_req));
	}

	if(rq_data_dir(req) == READ)
		dev->nomap_reads++;
	else dev->nomap_writes++;

	io_req->op = (rq_data_dir(req) == READ) ? UBD_READ : UBD_WRITE;
	io_req->offsets[0] = 0;
	io_req->offsets[1] = dev->cow.data_offset;
	io_req->buffer = req->buffer;
	io_req->sectorsize = 1 << 9;

	if(dev->cow.file != NULL)
		cowify_req(io_req, dev->cow.bitmap, dev->cow.bitmap_offset,
			   dev->cow.bitmap_len);

	return(0);
}

/* Called with ubd_io_lock held */
static void do_ubd_request(request_queue_t *q)
{
	struct io_thread_req io_req;
	struct request *req;
	int err, n;

	if(thread_fd == -1){
		while((req = elv_next_request(q)) != NULL){
			err = prepare_request(req, &io_req);
			if(!err){
				do_io(&io_req);
				__ubd_finish(req, io_req.error);
			}
		}
	}
	else {
		if(do_ubd || (req = elv_next_request(q)) == NULL)
			return;
		err = prepare_request(req, &io_req);
		if(!err){
			do_ubd = ubd_handler;
			n = os_write_file(thread_fd, (char *) &io_req,
					 sizeof(io_req));
			if(n != sizeof(io_req))
				printk("write to io thread failed, "
				       "errno = %d\n", -n);
		}
	}
}

static int ubd_ioctl(struct inode * inode, struct file * file,
		     unsigned int cmd, unsigned long arg)
{
	struct hd_geometry __user *loc = (struct hd_geometry __user *) arg;
	struct ubd *dev = inode->i_bdev->bd_disk->private_data;
	struct hd_driveid ubd_id = {
		.cyls		= 0,
		.heads		= 128,
		.sectors	= 32,
	};

	switch (cmd) {
	        struct hd_geometry g;
		struct cdrom_volctrl volume;
	case HDIO_GETGEO:
		if(!loc) return(-EINVAL);
		g.heads = 128;
		g.sectors = 32;
		g.cylinders = dev->size / (128 * 32 * 512);
		g.start = get_start_sect(inode->i_bdev);
		return(copy_to_user(loc, &g, sizeof(g)) ? -EFAULT : 0);

	case HDIO_GET_IDENTITY:
		ubd_id.cyls = dev->size / (128 * 32 * 512);
		if(copy_to_user((char __user *) arg, (char *) &ubd_id,
				 sizeof(ubd_id)))
			return(-EFAULT);
		return(0);
		
	case CDROMVOLREAD:
		if(copy_from_user(&volume, (char __user *) arg, sizeof(volume)))
			return(-EFAULT);
		volume.channel0 = 255;
		volume.channel1 = 255;
		volume.channel2 = 255;
		volume.channel3 = 255;
		if(copy_to_user((char __user *) arg, &volume, sizeof(volume)))
			return(-EFAULT);
		return(0);
	}
	return(-EINVAL);
}

static int ubd_check_remapped(int fd, unsigned long address, int is_write,
			      __u64 offset)
{
	__u64 bitmap_offset;
	unsigned long new_bitmap[2];
	int i, err, n;

	/* If it's not a write access, we can't do anything about it */
	if(!is_write)
		return(0);

	/* We have a write */
	for(i = 0; i < sizeof(ubd_dev) / sizeof(ubd_dev[0]); i++){
		struct ubd *dev = &ubd_dev[i];

		if((dev->fd != fd) && (dev->cow.fd != fd))
			continue;

		/* It's a write to a ubd device */

		/* This should be impossible now */
		if(!dev->openflags.w){
			/* It's a write access on a read-only device - probably
			 * shouldn't happen.  If the kernel is trying to change
			 * something with no intention of writing it back out,
			 * then this message will clue us in that this needs
			 * fixing
			 */
			printk("Write access to mapped page from readonly ubd "
			       "device %d\n", i);
			return(0);
		}

		/* It's a write to a writeable ubd device - it must be COWed
		 * because, otherwise, the page would have been mapped in
		 * writeable
		 */

		if(!dev->cow.file)
			panic("Write fault on writeable non-COW ubd device %d",
			      i);

		/* It should also be an access to the backing file since the
		 * COW pages should be mapped in read-write
		 */

		if(fd == dev->fd)
			panic("Write fault on a backing page of ubd "
			      "device %d\n", i);

		/* So, we do the write, copying the backing data to the COW
		 * file...
		 */

		err = os_seek_file(dev->fd, offset + dev->cow.data_offset);
		if(err < 0)
			panic("Couldn't seek to %lld in COW file of ubd "
			      "device %d, err = %d",
			      offset + dev->cow.data_offset, i, -err);

		n = os_write_file(dev->fd, (void *) address, PAGE_SIZE);
		if(n != PAGE_SIZE)
			panic("Couldn't copy data to COW file of ubd "
			      "device %d, err = %d", i, -n);

		/* ... updating the COW bitmap... */

		cowify_bitmap(offset, PAGE_SIZE, NULL, &bitmap_offset,
			      dev->cow.bitmap, dev->cow.bitmap_offset,
			      new_bitmap, dev->cow.bitmap_len);

		err = os_seek_file(dev->fd, bitmap_offset);
		if(err < 0)
			panic("Couldn't seek to %lld in COW file of ubd "
			      "device %d, err = %d", bitmap_offset, i, -err);

		n = os_write_file(dev->fd, new_bitmap, sizeof(new_bitmap));
		if(n != sizeof(new_bitmap))
			panic("Couldn't update bitmap  of ubd device %d, "
			      "err = %d", i, -n);

		/* Maybe we can map the COW page in, and maybe we can't.  If
		 * it is a pre-V3 COW file, we can't, since the alignment will
		 * be wrong.  If it is a V3 or later COW file which has been
		 * moved to a system with a larger page size, then maybe we
		 * can't, depending on the exact location of the page.
		 */

		offset += dev->cow.data_offset;

		/* Remove the remapping, putting the original anonymous page
		 * back.  If the COW file can be mapped in, that is done.
		 * Otherwise, the COW page is read in.
		 */

		if(!physmem_remove_mapping((void *) address))
			panic("Address 0x%lx not remapped by ubd device %d",
			      address, i);
		if((offset % UBD_MMAP_BLOCK_SIZE) == 0)
			physmem_subst_mapping((void *) address, dev->fd,
					      offset, 1);
		else {
			err = os_seek_file(dev->fd, offset);
			if(err < 0)
				panic("Couldn't seek to %lld in COW file of "
				      "ubd device %d, err = %d", offset, i,
				      -err);

			n = os_read_file(dev->fd, (void *) address, PAGE_SIZE);
			if(n != PAGE_SIZE)
				panic("Failed to read page from offset %llx of "
				      "COW file of ubd device %d, err = %d",
				      offset, i, -n);
		}

		return(1);
	}

	/* It's not a write on a ubd device */
	return(0);
}

static struct remapper ubd_remapper = {
	.list	= LIST_HEAD_INIT(ubd_remapper.list),
	.proc	= ubd_check_remapped,
};

static int ubd_remapper_setup(void)
{
	if(ubd_do_mmap)
		register_remapper(&ubd_remapper);

	return(0);
}

__initcall(ubd_remapper_setup);

static int same_backing_files(char *from_cmdline, char *from_cow, char *cow)
{
	struct uml_stat buf1, buf2;
	int err;

	if(from_cmdline == NULL) return(1);
	if(!strcmp(from_cmdline, from_cow)) return(1);

	err = os_stat_file(from_cmdline, &buf1);
	if(err < 0){
		printk("Couldn't stat '%s', err = %d\n", from_cmdline, -err);
		return(1);
	}
	err = os_stat_file(from_cow, &buf2);
	if(err < 0){
		printk("Couldn't stat '%s', err = %d\n", from_cow, -err);
		return(1);
	}
	if((buf1.ust_dev == buf2.ust_dev) && (buf1.ust_ino == buf2.ust_ino))
		return(1);

	printk("Backing file mismatch - \"%s\" requested,\n"
	       "\"%s\" specified in COW header of \"%s\"\n",
	       from_cmdline, from_cow, cow);
	return(0);
}

static int backing_file_mismatch(char *file, __u64 size, time_t mtime)
{
	unsigned long modtime;
	long long actual;
	int err;

	err = os_file_modtime(file, &modtime);
	if(err < 0){
		printk("Failed to get modification time of backing file "
		       "\"%s\", err = %d\n", file, -err);
		return(err);
	}

	err = os_file_size(file, &actual);
	if(err < 0){
		printk("Failed to get size of backing file \"%s\", "
		       "err = %d\n", file, -err);
		return(err);
	}

  	if(actual != size){
		/*__u64 can be a long on AMD64 and with %lu GCC complains; so
		 * the typecast.*/
		printk("Size mismatch (%llu vs %llu) of COW header vs backing "
		       "file\n", (unsigned long long) size, actual);
		return(-EINVAL);
	}
	if(modtime != mtime){
		printk("mtime mismatch (%ld vs %ld) of COW header vs backing "
		       "file\n", mtime, modtime);
		return(-EINVAL);
	}
	return(0);
}

int read_cow_bitmap(int fd, void *buf, int offset, int len)
{
	int err;

	err = os_seek_file(fd, offset);
	if(err < 0)
		return(err);

	err = os_read_file(fd, buf, len);
	if(err < 0)
		return(err);

	return(0);
}

int open_ubd_file(char *file, struct openflags *openflags,
		  char **backing_file_out, int *bitmap_offset_out,
		  unsigned long *bitmap_len_out, int *data_offset_out,
		  int *create_cow_out)
{
	time_t mtime;
	unsigned long long size;
	__u32 version, align;
	char *backing_file;
	int fd, err, sectorsize, same, mode = 0644;

	fd = os_open_file(file, *openflags, mode);
	if(fd < 0){
		if((fd == -ENOENT) && (create_cow_out != NULL))
			*create_cow_out = 1;
                if(!openflags->w ||
                   ((fd != -EROFS) && (fd != -EACCES))) return(fd);
		openflags->w = 0;
		fd = os_open_file(file, *openflags, mode);
		if(fd < 0)
			return(fd);
        }

	err = os_lock_file(fd, openflags->w);
	if(err < 0){
		printk("Failed to lock '%s', err = %d\n", file, -err);
		goto out_close;
	}

	if(backing_file_out == NULL) return(fd);

	err = read_cow_header(file_reader, &fd, &version, &backing_file, &mtime,
			      &size, &sectorsize, &align, bitmap_offset_out);
	if(err && (*backing_file_out != NULL)){
		printk("Failed to read COW header from COW file \"%s\", "
		       "errno = %d\n", file, -err);
		goto out_close;
	}
	if(err) return(fd);

	if(backing_file_out == NULL) return(fd);

	same = same_backing_files(*backing_file_out, backing_file, file);

	if(!same && !backing_file_mismatch(*backing_file_out, size, mtime)){
		printk("Switching backing file to '%s'\n", *backing_file_out);
		err = write_cow_header(file, fd, *backing_file_out,
				       sectorsize, align, &size);
		if(err){
			printk("Switch failed, errno = %d\n", -err);
			return(err);
		}
	}
	else {
		*backing_file_out = backing_file;
		err = backing_file_mismatch(*backing_file_out, size, mtime);
		if(err) goto out_close;
	}

	cow_sizes(version, size, sectorsize, align, *bitmap_offset_out,
		  bitmap_len_out, data_offset_out);

        return(fd);
 out_close:
	os_close_file(fd);
	return(err);
}

int create_cow_file(char *cow_file, char *backing_file, struct openflags flags,
		    int sectorsize, int alignment, int *bitmap_offset_out,
		    unsigned long *bitmap_len_out, int *data_offset_out)
{
	int err, fd;

	flags.c = 1;
	fd = open_ubd_file(cow_file, &flags, NULL, NULL, NULL, NULL, NULL);
	if(fd < 0){
		err = fd;
		printk("Open of COW file '%s' failed, errno = %d\n", cow_file,
		       -err);
		goto out;
	}

	err = init_cow_file(fd, cow_file, backing_file, sectorsize, alignment,
			    bitmap_offset_out, bitmap_len_out,
			    data_offset_out);
	if(!err)
		return(fd);
	os_close_file(fd);
 out:
	return(err);
}

static int update_bitmap(struct io_thread_req *req)
{
	int n;

	if(req->cow_offset == -1)
		return(0);

	n = os_seek_file(req->fds[1], req->cow_offset);
	if(n < 0){
		printk("do_io - bitmap lseek failed : err = %d\n", -n);
		return(1);
	}

	n = os_write_file(req->fds[1], &req->bitmap_words,
		          sizeof(req->bitmap_words));
	if(n != sizeof(req->bitmap_words)){
		printk("do_io - bitmap update failed, err = %d fd = %d\n", -n,
		       req->fds[1]);
		return(1);
	}

	return(0);
}

void do_io(struct io_thread_req *req)
{
	char *buf;
	unsigned long len;
	int n, nsectors, start, end, bit;
	int err;
	__u64 off;

	if(req->op == UBD_MMAP){
		/* Touch the page to force the host to do any necessary IO to
		 * get it into memory
		 */
		n = *((volatile int *) req->buffer);
		req->error = update_bitmap(req);
		return;
	}

	nsectors = req->length / req->sectorsize;
	start = 0;
	do {
		bit = ubd_test_bit(start, (unsigned char *) &req->sector_mask);
		end = start;
		while((end < nsectors) &&
		      (ubd_test_bit(end, (unsigned char *)
				    &req->sector_mask) == bit))
			end++;

		off = req->offset + req->offsets[bit] +
			start * req->sectorsize;
		len = (end - start) * req->sectorsize;
		buf = &req->buffer[start * req->sectorsize];

		err = os_seek_file(req->fds[bit], off);
		if(err < 0){
			printk("do_io - lseek failed : err = %d\n", -err);
			req->error = 1;
			return;
		}
		if(req->op == UBD_READ){
			n = 0;
			do {
				buf = &buf[n];
				len -= n;
				n = os_read_file(req->fds[bit], buf, len);
				if (n < 0) {
					printk("do_io - read failed, err = %d "
					       "fd = %d\n", -n, req->fds[bit]);
					req->error = 1;
					return;
				}
			} while((n < len) && (n != 0));
			if (n < len) memset(&buf[n], 0, len - n);
		} else {
			n = os_write_file(req->fds[bit], buf, len);
			if(n != len){
				printk("do_io - write failed err = %d "
				       "fd = %d\n", -n, req->fds[bit]);
				req->error = 1;
				return;
			}
		}

		start = end;
	} while(start < nsectors);

	req->error = update_bitmap(req);
}

/* Changed in start_io_thread, which is serialized by being called only
 * from ubd_init, which is an initcall.
 */
int kernel_fd = -1;

/* Only changed by the io thread */
int io_count = 0;

int io_thread(void *arg)
{
	struct io_thread_req req;
	int n;

	ignore_sigwinch_sig();
	while(1){
		n = os_read_file(kernel_fd, &req, sizeof(req));
		if(n != sizeof(req)){
			if(n < 0)
				printk("io_thread - read failed, fd = %d, "
				       "err = %d\n", kernel_fd, -n);
			else {
				printk("io_thread - short read, fd = %d, "
				       "length = %d\n", kernel_fd, n);
			}
			continue;
		}
		io_count++;
		do_io(&req);
		n = os_write_file(kernel_fd, &req, sizeof(req));
		if(n != sizeof(req))
			printk("io_thread - write failed, fd = %d, err = %d\n",
			       kernel_fd, -n);
	}
}

/*
 * Overrides for Emacs so that we follow Linus's tabbing style.
 * Emacs will notice this stuff at the end of the file and automatically
 * adjust the settings for this buffer only.  This must remain at the end
 * of the file.
 * ---------------------------------------------------------------------------
 * Local variables:
 * c-file-style: "linux"
 * End:
 */