File: posixio.c

package info (click to toggle)
netcdf 1%3A4.4.1.1-2
  • links: PTS, VCS
  • area: main
  • in suites: stretch
  • size: 96,828 kB
  • ctags: 15,369
  • sloc: ansic: 163,650; sh: 9,294; yacc: 2,457; makefile: 1,208; lex: 1,161; xml: 173; f90: 7; fortran: 6; awk: 2
file content (1909 lines) | stat: -rw-r--r-- 44,982 bytes parent folder | download
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
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
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066
1067
1068
1069
1070
1071
1072
1073
1074
1075
1076
1077
1078
1079
1080
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103
1104
1105
1106
1107
1108
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
1119
1120
1121
1122
1123
1124
1125
1126
1127
1128
1129
1130
1131
1132
1133
1134
1135
1136
1137
1138
1139
1140
1141
1142
1143
1144
1145
1146
1147
1148
1149
1150
1151
1152
1153
1154
1155
1156
1157
1158
1159
1160
1161
1162
1163
1164
1165
1166
1167
1168
1169
1170
1171
1172
1173
1174
1175
1176
1177
1178
1179
1180
1181
1182
1183
1184
1185
1186
1187
1188
1189
1190
1191
1192
1193
1194
1195
1196
1197
1198
1199
1200
1201
1202
1203
1204
1205
1206
1207
1208
1209
1210
1211
1212
1213
1214
1215
1216
1217
1218
1219
1220
1221
1222
1223
1224
1225
1226
1227
1228
1229
1230
1231
1232
1233
1234
1235
1236
1237
1238
1239
1240
1241
1242
1243
1244
1245
1246
1247
1248
1249
1250
1251
1252
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
1615
1616
1617
1618
1619
1620
1621
1622
1623
1624
1625
1626
1627
1628
1629
1630
1631
1632
1633
1634
1635
1636
1637
1638
1639
1640
1641
1642
1643
1644
1645
1646
1647
1648
1649
1650
1651
1652
1653
1654
1655
1656
1657
1658
1659
1660
1661
1662
1663
1664
1665
1666
1667
1668
1669
1670
1671
1672
1673
1674
1675
1676
1677
1678
1679
1680
1681
1682
1683
1684
1685
1686
1687
1688
1689
1690
1691
1692
1693
1694
1695
1696
1697
1698
1699
1700
1701
1702
1703
1704
1705
1706
1707
1708
1709
1710
1711
1712
1713
1714
1715
1716
1717
1718
1719
1720
1721
1722
1723
1724
1725
1726
1727
1728
1729
1730
1731
1732
1733
1734
1735
1736
1737
1738
1739
1740
1741
1742
1743
1744
1745
1746
1747
1748
1749
1750
1751
1752
1753
1754
1755
1756
1757
1758
1759
1760
1761
1762
1763
1764
1765
1766
1767
1768
1769
1770
1771
1772
1773
1774
1775
1776
1777
1778
1779
1780
1781
1782
1783
1784
1785
1786
1787
1788
1789
1790
1791
1792
1793
1794
1795
1796
1797
1798
1799
1800
1801
1802
1803
1804
1805
1806
1807
1808
1809
1810
1811
1812
1813
1814
1815
1816
1817
1818
1819
1820
1821
1822
1823
1824
1825
1826
1827
1828
1829
1830
1831
1832
1833
1834
1835
1836
1837
1838
1839
1840
1841
1842
1843
1844
1845
1846
1847
1848
1849
1850
1851
1852
1853
1854
1855
1856
1857
1858
1859
1860
1861
1862
1863
1864
1865
1866
1867
1868
1869
1870
1871
1872
1873
1874
1875
1876
1877
1878
1879
1880
1881
1882
1883
1884
1885
1886
1887
1888
1889
1890
1891
1892
1893
1894
1895
1896
1897
1898
1899
1900
1901
1902
1903
1904
1905
1906
1907
1908
1909
/*
 *	Copyright 1996, University Corporation for Atmospheric Research
 *	See netcdf/COPYRIGHT file for copying and redistribution conditions.
 */
/* $Id: posixio.c,v 1.89 2010/05/22 21:59:08 dmh Exp $ */

/* For MinGW Build */


#include <config.h>
#include <stdio.h>

/* Windows platforms, including MinGW, Cygwin, Visual Studio */
#if defined(_WIN32) || defined(_WIN64)
#include <windows.h>
#include <winbase.h>
#include <io.h>
#else
#include <unistd.h>
#endif

#include <assert.h>
#include <stdlib.h>
#include <errno.h>

#ifndef NC_NOERR
#define NC_NOERR 0
#endif

#include <sys/types.h>
#include <sys/stat.h>

#include <fcntl.h>
#include <string.h>

#ifndef HAVE_SSIZE_T
typedef int ssize_t;
#endif

#ifndef SEEK_SET
#define SEEK_SET 0
#define SEEK_CUR 1
#define SEEK_END 2
#endif

#include "ncio.h"
#include "fbits.h"
#include "rnd.h"

/* #define INSTRUMENT 1 */
#if INSTRUMENT /* debugging */
#undef NDEBUG
#include <stdio.h>
#include "instr.h"
#endif

#undef MIN  /* system may define MIN somewhere and complain */
#define MIN(mm,nn) (((mm) < (nn)) ? (mm) : (nn))

#if !defined(NDEBUG) && !defined(X_INT_MAX)
#define  X_INT_MAX 2147483647
#endif

#if 0 /* !defined(NDEBUG) && !defined(X_ALIGN) */
#define  X_ALIGN 4
#else
#undef X_ALIGN
#endif

/* These are needed on mingw to get a dll to compile. They really
 * should be provided in sys/stats.h, but what the heck. Let's not be
 * too picky! */
#ifndef S_IRGRP
#define S_IRGRP   0000040
#endif
#ifndef S_IROTH
#define S_IROTH   0000004
#endif
#ifndef S_IWGRP
#define S_IWGRP   0000020
#endif
#ifndef S_IWOTH
#define S_IWOTH   0000002
#endif

/*Forward*/
static int ncio_px_filesize(ncio *nciop, off_t *filesizep);
static int ncio_px_pad_length(ncio *nciop, off_t length);
static int ncio_px_close(ncio *nciop, int doUnlink);
static int ncio_spx_close(ncio *nciop, int doUnlink);


/*
 * Define the following for debugging.
 */
/* #define ALWAYS_NC_SHARE 1 */

/* Begin OS */

#ifndef POSIXIO_DEFAULT_PAGESIZE
#define POSIXIO_DEFAULT_PAGESIZE 4096
#endif

/*! Cross-platform file length.
 *
 * Some versions of Visual Studio are throwing errno 132
 * when fstat is used on large files.  This function is
 * an attempt to get around that.
 *
 * @par fd File Descriptor.
 * @return -1 on error, length of file (in bytes) otherwise.
 */
static size_t nc_get_filelen(const int fd) {

  size_t flen;

#ifdef HAVE_FILE_LENGTH_I64
  __int64 file_len = 0;
  if ((file_len = _filelengthi64(fd)) < 0) {
    return file_len;
  }
  flen = (size_t)file_len;

#else
  int res = 0;
  struct stat sb;
  if((res = fstat(fd,&sb)) <0)
    return res;

  flen = sb.st_size;
#endif

  return flen;

}


/*
 * What is the system pagesize?
 */
static size_t
pagesize(void)
{
  size_t pgsz;
#if defined(_WIN32) || defined(_WIN64)
  SYSTEM_INFO info;
#endif
/* Hmm, aren't standards great? */
#if defined(_SC_PAGE_SIZE) && !defined(_SC_PAGESIZE)
#define _SC_PAGESIZE _SC_PAGE_SIZE
#endif

  /* For MinGW Builds */
#if defined(_WIN32) || defined(_WIN64)
  GetSystemInfo(&info);
  pgsz = (size_t)info.dwPageSize;
#elif defined(_SC_PAGESIZE)
  pgsz = (size_t)sysconf(_SC_PAGESIZE);
#elif defined(HAVE_GETPAGESIZE)
  pgsz = (size_t) getpagesize();
#endif
  if(pgsz > 0)
    return (size_t) pgsz;
   return (size_t)POSIXIO_DEFAULT_PAGESIZE;
}

/*
 * What is the preferred I/O block size?
 */
static size_t
blksize(int fd)
{
#if defined(HAVE_ST_BLKSIZE)
	struct stat sb;
	if (fstat(fd, &sb) > -1)
	{
		if(sb.st_blksize >= 8192)
			return (size_t) sb.st_blksize;
		return 8192;
	}
	/* else, silent in the face of error */
#endif
	return (size_t) 2 * pagesize();
}


/*
 * Sortof like ftruncate, except won't make the
 * file shorter.
 */
static int
fgrow(const int fd, const off_t len)
{
	struct stat sb;
	if (fstat(fd, &sb) < 0)
		return errno;
	if (len < sb.st_size)
		return NC_NOERR;
	{
	    const long dumb = 0;
	    /* we don't use ftruncate() due to problem with FAT32 file systems */
	    /* cache current position */
	    const off_t pos = lseek(fd, 0, SEEK_CUR);
	    if(pos < 0)
		return errno;
	    if (lseek(fd, len-sizeof(dumb), SEEK_SET) < 0)
		return errno;
	    if(write(fd, &dumb, sizeof(dumb)) < 0)
		return errno;
	    if (lseek(fd, pos, SEEK_SET) < 0)
		return errno;
	}
	return NC_NOERR;
}


/*
 * Sortof like ftruncate, except won't make the file shorter.  Differs
 * from fgrow by only writing one byte at designated seek position, if
 * needed.
 */
static int
fgrow2(const int fd, const off_t len)
{


  /* There is a problem with fstat on Windows based systems
     which manifests (so far) when Config RELEASE is built.
     Use _filelengthi64 isntead.

     See https://github.com/Unidata/netcdf-c/issues/188

  */


  size_t file_len = nc_get_filelen(fd);
  if(file_len < 0) return errno;
  if(len <= file_len)
    return NC_NOERR;
  {
    const char dumb = 0;
	    /* we don't use ftruncate() due to problem with FAT32 file systems */
	    /* cache current position */
	    const off_t pos = lseek(fd, 0, SEEK_CUR);
	    if(pos < 0)
		return errno;
	    if (lseek(fd, len-1, SEEK_SET) < 0)
		return errno;
	    if(write(fd, &dumb, sizeof(dumb)) < 0)
		return errno;
	    if (lseek(fd, pos, SEEK_SET) < 0)
		return errno;
	}
	return NC_NOERR;
}
/* End OS */
/* Begin px */

/* The px_ functions are for posix systems, when NC_SHARE is not in
   effect. */

/* Write out a "page" of data to the file. The size of the page
   (i.e. the extent) varies.

   nciop - pointer to the file metadata.
   offset - where in the file should this page be written.
   extent - how many bytes should be written.
   vp - pointer to the data to write.
   posp - pointer to current position in file, updated after write.
*/
static int
px_pgout(ncio *const nciop,
	off_t const offset,  const size_t extent,
	void *const vp, off_t *posp)
{
    ssize_t partial;
    size_t nextent;
    char *nvp;
#ifdef X_ALIGN
	assert(offset % X_ALIGN == 0);
#endif

	assert(*posp == OFF_NONE || *posp == lseek(nciop->fd, 0, SEEK_CUR));

	if(*posp != offset)
	{
		if(lseek(nciop->fd, offset, SEEK_SET) != offset)
		{
			return errno;
		}
		*posp = offset;
	}
	/* Old write, didn't handle partial writes correctly */
	/* if(write(nciop->fd, vp, extent) != (ssize_t) extent) */
	/* { */
	/* 	return errno; */
	/* } */
	nextent = extent;
        nvp = vp;
	while((partial = write(nciop->fd, nvp, nextent)) != -1) {
	    if(partial == nextent)
		break;
	    nvp += partial;
	    nextent -= partial;
	}
	if(partial == -1)
	    return errno;
	*posp += extent;

	return NC_NOERR;
}

/*! Read in a page of data.

  @param[in] nciop  A pointer to the ncio struct for this file.
  @param[in] offset The byte offset in file where read starts.
  @param[in] extent The size of the page that will be read.
  @param[in] vp     A pointer to where the data will end up.

  @param[in,out] nreadp Returned number of bytes actually read (may be less than extent).
  @param[in,out] posp The pointer to current position in file, updated after read.
  @return Return 0 on success, otherwise an error code.
*/
static int
px_pgin(ncio *const nciop,
	off_t const offset, const size_t extent,
	void *const vp, size_t *nreadp, off_t *posp)
{
	int status;
	ssize_t nread;
    size_t read_count = 0;
    ssize_t bytes_xfered = 0;
    void *p = vp;
#ifdef X_ALIGN
	assert(offset % X_ALIGN == 0);
	assert(extent % X_ALIGN == 0);
#endif
    /* *posp == OFF_NONE (-1) on first call. This
       is problematic because lseek also returns -1
       on error. Use errno instead. */
    if(*posp != OFF_NONE && *posp != lseek(nciop->fd, 0, SEEK_CUR)) {
      if(errno) {
        status = errno;
        printf("Error %d: %s\n",errno,strerror(errno));
        return status;
      }
    }

	if(*posp != offset)
	{
		if(lseek(nciop->fd, offset, SEEK_SET) != offset)
		{
			status = errno;
			return status;
		}
		*posp = offset;
	}

	errno = 0;
    /* Handle the case where the read is interrupted
       by a signal (see NCF-337,
       http://pubs.opengroup.org/onlinepubs/009695399/functions/read.html)

       When this happens, nread will (should) be the bytes read, and
       errno will be set to EINTR.  On older systems nread might be -1.
       If this is the case, there's not a whole lot we can do about it
       as we can't compute any offsets, so we will attempt to read again.
       This *feels* like it could lead to an infinite loop, but it shouldn't
       unless the read is being constantly interrupted by a signal, and is
       on an older system which returns -1 instead of bytexs read.

       The case where it's a short read is already handled by the function
       (according to the comment below, at least). */
    do {
      nread = read(nciop->fd,vp,extent);
    } while (nread == -1 && errno == EINTR);


    if(nread != (ssize_t)extent) {
      status = errno;
      if( nread == -1 || (status != EINTR && status != NC_NOERR))
        return status;
      /* else it's okay we read less than asked for */
      (void) memset((char *)vp + nread, 0, (ssize_t)extent - nread);
    }

    *nreadp = nread;
	*posp += nread;

	return NC_NOERR;
}

/* This struct is for POSIX systems, with NC_SHARE not in effect. If
   NC_SHARE is used, see ncio_spx.

   blksz - block size for reads and writes to file.
   pos - current read/write position in file.
   bf_offset - file offset corresponding to start of memory buffer
   bf_extent - number of bytes in I/O request
   bf_cnt - number of bytes available in buffer
   bf_base - pointer to beginning of buffer.
   bf_rflags - buffer region flags (defined in ncio.h) tell the lock
   status, read/write permissions, and modification status of regions
   of data in the buffer.
   bf_refcount - buffer reference count.
   slave - used in moves.
*/
typedef struct ncio_px {
	size_t blksz;
	off_t pos;
	/* buffer */
	off_t	bf_offset;
	size_t	bf_extent;
	size_t	bf_cnt;
	void	*bf_base;
	int	bf_rflags;
	int	bf_refcount;
	/* chain for double buffering in px_move */
	struct ncio_px *slave;
} ncio_px;


/*ARGSUSED*/
/* This function indicates the file region starting at offset may be
   released.

   This is for POSIX, without NC_SHARE.  If called with RGN_MODIFIED
   flag, sets the modified flag in pxp->bf_rflags and decrements the
   reference count.

   pxp - pointer to posix non-share ncio_px struct.

   offset - file offset for beginning of to region to be
   released.

   rflags - only RGN_MODIFIED is relevant to this function, others ignored
*/
static int
px_rel(ncio_px *const pxp, off_t offset, int rflags)
{
	assert(pxp->bf_offset <= offset
		 && offset < pxp->bf_offset + (off_t) pxp->bf_extent);
	assert(pIf(fIsSet(rflags, RGN_MODIFIED),
		fIsSet(pxp->bf_rflags, RGN_WRITE)));

	if(fIsSet(rflags, RGN_MODIFIED))
	{
		fSet(pxp->bf_rflags, RGN_MODIFIED);
	}
	pxp->bf_refcount--;

	return NC_NOERR;
}

/* This function indicates the file region starting at offset may be
   released.  Each read or write to the file is bracketed by a call to
   the "get" region function and a call to the "rel" region function.
   If you only read from the memory region, release it with a flag of
   0, if you modify the region, release it with a flag of
   RGN_MODIFIED.

   For POSIX system, without NC_SHARE, this becomes the rel function
   pointed to by the ncio rel function pointer. It mearly checks for
   file write permission, then calls px_rel to do everything.

   nciop - pointer to ncio struct.
   offset - num bytes from beginning of buffer to region to be
   released.
   rflags - only RGN_MODIFIED is relevant to this function, others ignored
*/
static int
ncio_px_rel(ncio *const nciop, off_t offset, int rflags)
{
	ncio_px *const pxp = (ncio_px *)nciop->pvt;

	if(fIsSet(rflags, RGN_MODIFIED) && !fIsSet(nciop->ioflags, NC_WRITE))
		return EPERM; /* attempt to write readonly file */

	return px_rel(pxp, offset, rflags);
}

/* POSIX get. This will "make a region available." Since we're using
   buffered IO, this means that if needed, we'll fetch a new page from
   the file, otherwise, just return a pointer to what's in memory
   already.

   nciop - pointer to ncio struct, containing file info.
   pxp - pointer to ncio_px struct, which contains special metadate
   for posix files without NC_SHARE.
   offset - start byte of region to get.
   extent - how many bytes to read.
   rflags - One of the RGN_* flags defined in ncio.h.
   vpp - pointer to pointer that will receive data.

   NOTES:

   * For blkoffset round offset down to the nearest pxp->blksz. This
   provides the offset (in bytes) to the beginning of the block that
   holds the current offset.

   * diff tells how far into the current block we are.

   * For blkextent round up to the number of bytes at the beginning of
   the next block, after the one that holds our current position, plus
   whatever extra (i.e. the extent) that we are about to grab.

   * The blkextent can't be more than twice the pxp->blksz. That's
   because the pxp->blksize is the sizehint, and in ncio_px_init2 the
   buffer (pointed to by pxp->bf-base) is allocated with 2 *
   *sizehintp. This is checked (unneccesarily) more than once in
   asserts.

   * If this is called on a newly opened file, pxp->bf_offset will be
   OFF_NONE and we'll jump to label pgin to immediately read in a
   page.
*/
static int
px_get(ncio *const nciop, ncio_px *const pxp,
		off_t offset, size_t extent,
		int rflags,
		void **const vpp)
{
	int status = NC_NOERR;

	const off_t blkoffset = _RNDDOWN(offset, (off_t)pxp->blksz);
	off_t diff = (size_t)(offset - blkoffset);
	off_t blkextent = _RNDUP(diff + extent, pxp->blksz);

	assert(extent != 0);
	assert(extent < X_INT_MAX); /* sanity check */
	assert(offset >= 0); /* sanity check */

	if(2 * pxp->blksz < blkextent)
		return E2BIG; /* TODO: temporary kludge */
	if(pxp->bf_offset == OFF_NONE)
	{
		/* Uninitialized */
		if(pxp->bf_base == NULL)
		{
			assert(pxp->bf_extent == 0);
			assert(blkextent <= 2 * pxp->blksz);
			pxp->bf_base = malloc(2 * pxp->blksz);
			if(pxp->bf_base == NULL)
				return ENOMEM;
		}
		goto pgin;
	}
	/* else */
	assert(blkextent <= 2 * pxp->blksz);

	if(blkoffset == pxp->bf_offset)
	{
		/* hit */
 		if(blkextent > pxp->bf_extent)
		{
			/* page in upper */
			void *const middle =
			 	(void *)((char *)pxp->bf_base + pxp->blksz);
			assert(pxp->bf_extent == pxp->blksz);
			status = px_pgin(nciop,
				 pxp->bf_offset + (off_t)pxp->blksz,
				 pxp->blksz,
				 middle,
				 &pxp->bf_cnt,
				 &pxp->pos);
			if(status != NC_NOERR)
				return status;
			pxp->bf_extent = 2 * pxp->blksz;
			pxp->bf_cnt += pxp->blksz;
		}
		goto done;
	}
	/* else */

	if(pxp->bf_extent > pxp->blksz
		 && blkoffset == pxp->bf_offset + (off_t)pxp->blksz)
	{
		/* hit in upper half */
		if(blkextent == pxp->blksz)
		{
			/* all in upper half, no fault needed */
			diff += pxp->blksz;
			goto done;
		}
		/* else */
		if(pxp->bf_cnt > pxp->blksz)
		{
			/* data in upper half */
			void *const middle =
				(void *)((char *)pxp->bf_base + pxp->blksz);
			assert(pxp->bf_extent == 2 * pxp->blksz);
			if(fIsSet(pxp->bf_rflags, RGN_MODIFIED))
			{
				/* page out lower half */
				assert(pxp->bf_refcount <= 0);
				status = px_pgout(nciop,
					pxp->bf_offset,
					pxp->blksz,
					pxp->bf_base,
					&pxp->pos);
				if(status != NC_NOERR)
					return status;
			}
			pxp->bf_cnt -= pxp->blksz;
			/* copy upper half into lower half */
			(void) memcpy(pxp->bf_base, middle, pxp->bf_cnt);
		}
		else		/* added to fix nofill bug */
		{
			assert(pxp->bf_extent == 2 * pxp->blksz);
			/* still have to page out lower half, if modified */
			if(fIsSet(pxp->bf_rflags, RGN_MODIFIED))
			{
				assert(pxp->bf_refcount <= 0);
				status = px_pgout(nciop,
					pxp->bf_offset,
					pxp->blksz,
					pxp->bf_base,
					&pxp->pos);
				if(status != NC_NOERR)
					return status;
			}
		}
		pxp->bf_offset = blkoffset;
		/* pxp->bf_extent = pxp->blksz; */

 		assert(blkextent == 2 * pxp->blksz);
		{
			/* page in upper */
			void *const middle =
			 	(void *)((char *)pxp->bf_base + pxp->blksz);
			status = px_pgin(nciop,
				 pxp->bf_offset + (off_t)pxp->blksz,
				 pxp->blksz,
				 middle,
				 &pxp->bf_cnt,
				 &pxp->pos);
			if(status != NC_NOERR)
				return status;
			pxp->bf_extent = 2 * pxp->blksz;
			pxp->bf_cnt += pxp->blksz;
		}
		goto done;
	}
	/* else */

	if(blkoffset == pxp->bf_offset - (off_t)pxp->blksz)
	{
		/* wants the page below */
		void *const middle =
			(void *)((char *)pxp->bf_base + pxp->blksz);
		size_t upper_cnt = 0;
		if(pxp->bf_cnt > pxp->blksz)
		{
			/* data in upper half */
			assert(pxp->bf_extent == 2 * pxp->blksz);
			if(fIsSet(pxp->bf_rflags, RGN_MODIFIED))
			{
				/* page out upper half */
				assert(pxp->bf_refcount <= 0);
				status = px_pgout(nciop,
					pxp->bf_offset + (off_t)pxp->blksz,
					pxp->bf_cnt - pxp->blksz,
					middle,
					&pxp->pos);
				if(status != NC_NOERR)
					return status;
			}
			pxp->bf_cnt = pxp->blksz;
			pxp->bf_extent = pxp->blksz;
		}
		if(pxp->bf_cnt > 0)
		{
			/* copy lower half into upper half */
			(void) memcpy(middle, pxp->bf_base, pxp->blksz);
			upper_cnt = pxp->bf_cnt;
		}
		/* read page below into lower half */
		status = px_pgin(nciop,
			 blkoffset,
			 pxp->blksz,
			 pxp->bf_base,
			 &pxp->bf_cnt,
			 &pxp->pos);
		if(status != NC_NOERR)
			return status;
		pxp->bf_offset = blkoffset;
		if(upper_cnt != 0)
		{
			pxp->bf_extent = 2 * pxp->blksz;
			pxp->bf_cnt = pxp->blksz + upper_cnt;
		}
		else
		{
			pxp->bf_extent = pxp->blksz;
		}
		goto done;
	}
	/* else */

	/* no overlap */
	if(fIsSet(pxp->bf_rflags, RGN_MODIFIED))
	{
		assert(pxp->bf_refcount <= 0);
		status = px_pgout(nciop,
			pxp->bf_offset,
			pxp->bf_cnt,
			pxp->bf_base,
			&pxp->pos);
		if(status != NC_NOERR)
			return status;
		pxp->bf_rflags = 0;
	}

pgin:
	status = px_pgin(nciop,
		 blkoffset,
		 blkextent,
		 pxp->bf_base,
		 &pxp->bf_cnt,
		 &pxp->pos);
	if(status != NC_NOERR)
		return status;
	 pxp->bf_offset = blkoffset;
	 pxp->bf_extent = blkextent;

done:
	extent += diff;
	if(pxp->bf_cnt < extent)
		pxp->bf_cnt = extent;
	assert(pxp->bf_cnt <= pxp->bf_extent);

	pxp->bf_rflags |= rflags;
	pxp->bf_refcount++;

#ifndef __CHAR_UNSIGNED__
    *vpp = (void *)((char *)pxp->bf_base + diff);
#else
    *vpp = (void *)((signed char*)pxp->bf_base + diff);
#endif
	return NC_NOERR;
}

/* Request that the region (offset, extent) be made available through
   *vpp.

   This function converts a file region specified by an offset and
   extent to a memory pointer. The region may be locked until the
   corresponding call to rel().

   For POSIX systems, without NC_SHARE. This function gets a page of
   size extent?

   This is a wrapper for the function px_get, which does all the heavy
   lifting.

   nciop - pointer to ncio struct for this file.
   offset - offset (from beginning of file?) to the data we want to
   read.
   extent - the number of bytes to read from the file.
   rflags - One of the RGN_* flags defined in ncio.h.
   vpp - handle to point at data when it's been read.
*/
static int
ncio_px_get(ncio *const nciop,
		off_t offset, size_t extent,
		int rflags,
		void **const vpp)
{
	ncio_px *const pxp = (ncio_px *)nciop->pvt;

	if(fIsSet(rflags, RGN_WRITE) && !fIsSet(nciop->ioflags, NC_WRITE))
		return EPERM; /* attempt to write readonly file */

	/* reclaim space used in move */
	if(pxp->slave != NULL)
	{
		if(pxp->slave->bf_base != NULL)
		{
			free(pxp->slave->bf_base);
			pxp->slave->bf_base = NULL;
			pxp->slave->bf_extent = 0;
			pxp->slave->bf_offset = OFF_NONE;
		}
		free(pxp->slave);
		pxp->slave = NULL;
	}
	return px_get(nciop, pxp, offset, extent, rflags, vpp);
}


/* ARGSUSED */
static int
px_double_buffer(ncio *const nciop, off_t to, off_t from,
			size_t nbytes, int rflags)
{
	ncio_px *const pxp = (ncio_px *)nciop->pvt;
	int status = NC_NOERR;
	void *src;
	void *dest;

#if INSTRUMENT
fprintf(stderr, "\tdouble_buffr %ld %ld %ld\n",
		 (long)to, (long)from, (long)nbytes);
#endif
	status = px_get(nciop, pxp, to, nbytes, RGN_WRITE,
			&dest);
	if(status != NC_NOERR)
		return status;

	if(pxp->slave == NULL)
	{
		pxp->slave = (ncio_px *) malloc(sizeof(ncio_px));
		if(pxp->slave == NULL)
			return ENOMEM;

		pxp->slave->blksz = pxp->blksz;
		/* pos done below */
		pxp->slave->bf_offset = pxp->bf_offset;
		pxp->slave->bf_extent = pxp->bf_extent;
		pxp->slave->bf_cnt = pxp->bf_cnt;
		pxp->slave->bf_base = malloc(2 * pxp->blksz);
		if(pxp->slave->bf_base == NULL)
			return ENOMEM;
		(void) memcpy(pxp->slave->bf_base, pxp->bf_base,
			 pxp->bf_extent);
		pxp->slave->bf_rflags = 0;
		pxp->slave->bf_refcount = 0;
		pxp->slave->slave = NULL;
	}

	pxp->slave->pos = pxp->pos;
	status = px_get(nciop, pxp->slave, from, nbytes, 0,
			&src);
	if(status != NC_NOERR)
		return status;
	if(pxp->pos != pxp->slave->pos)
	{
		/* position changed, sync */
		pxp->pos = pxp->slave->pos;
	}

	(void) memcpy(dest, src, nbytes);

	(void)px_rel(pxp->slave, from, 0);
	(void)px_rel(pxp, to, RGN_MODIFIED);

	return status;
}

/* Like memmove(), safely move possibly overlapping data.

   Copy one region to another without making anything available to
   higher layers. May be just implemented in terms of get() and rel(),
   or may be tricky to be efficient. Only used in by nc_enddef()
   after redefinition.

   nciop - pointer to ncio struct with file info.
   to - src for move?
   from - dest for move?
   nbytes - number of bytes to move.
   rflags - One of the RGN_* flags defined in ncio.h. The only
   reasonable flag value is RGN_NOLOCK.
*/
static int
ncio_px_move(ncio *const nciop, off_t to, off_t from,
			size_t nbytes, int rflags)
{
	ncio_px *const pxp = (ncio_px *)nciop->pvt;
	int status = NC_NOERR;
	off_t lower;
	off_t upper;
	char *base;
	size_t diff;
	size_t extent;

	if(to == from)
		return NC_NOERR; /* NOOP */

	if(fIsSet(rflags, RGN_WRITE) && !fIsSet(nciop->ioflags, NC_WRITE))
		return EPERM; /* attempt to write readonly file */

	rflags &= RGN_NOLOCK; /* filter unwanted flags */

	if(to > from)
	{
		/* growing */
		lower = from;
		upper = to;
	}
	else
	{
		/* shrinking */
		lower = to;
		upper = from;
	}
	diff = (size_t)(upper - lower);
	extent = diff + nbytes;

#if INSTRUMENT
fprintf(stderr, "ncio_px_move %ld %ld %ld %ld %ld\n",
		 (long)to, (long)from, (long)nbytes, (long)lower, (long)extent);
#endif
	if(extent > pxp->blksz)
	{
		size_t remaining = nbytes;

if(to > from)
{
		off_t frm = from + nbytes;
		off_t toh = to + nbytes;
		for(;;)
		{
			size_t loopextent = MIN(remaining, pxp->blksz);
			frm -= loopextent;
			toh -= loopextent;

			status = px_double_buffer(nciop, toh, frm,
				 	loopextent, rflags) ;
			if(status != NC_NOERR)
				return status;
			remaining -= loopextent;

			if(remaining == 0)
				break; /* normal loop exit */
		}
}
else
{
		for(;;)
		{
			size_t loopextent = MIN(remaining, pxp->blksz);

			status = px_double_buffer(nciop, to, from,
				 	loopextent, rflags) ;
			if(status != NC_NOERR)
				return status;
			remaining -= loopextent;

			if(remaining == 0)
				break; /* normal loop exit */
			to += loopextent;
			from += loopextent;
		}
}
		return NC_NOERR;
	}

#if INSTRUMENT
fprintf(stderr, "\tncio_px_move small\n");
#endif
	status = px_get(nciop, pxp, lower, extent, RGN_WRITE|rflags,
			(void **)&base);

	if(status != NC_NOERR)
		return status;

	if(to > from)
		(void) memmove(base + diff, base, nbytes);
	else
		(void) memmove(base, base + diff, nbytes);

	(void) px_rel(pxp, lower, RGN_MODIFIED);

	return status;
}


/* Flush any buffers to disk. May be a no-op on if I/O is unbuffered.
   This function is used when NC_SHARE is NOT used.
*/
static int
ncio_px_sync(ncio *const nciop)
{
	ncio_px *const pxp = (ncio_px *)nciop->pvt;
	int status = NC_NOERR;
	if(fIsSet(pxp->bf_rflags, RGN_MODIFIED))
	{
		assert(pxp->bf_refcount <= 0);
		status = px_pgout(nciop, pxp->bf_offset,
			pxp->bf_cnt,
			pxp->bf_base, &pxp->pos);
		if(status != NC_NOERR)
			return status;
		pxp->bf_rflags = 0;
	}
	else if (!fIsSet(pxp->bf_rflags, RGN_WRITE))
	{
	    /*
	     * The dataset is readonly.  Invalidate the buffers so
	     * that the next ncio_px_get() will actually read data.
	     */
	    pxp->bf_offset = OFF_NONE;
	    pxp->bf_cnt = 0;
	}
	return status;
}

/* Internal function called at close to
   free up anything hanging off pvt.
*/
static void
ncio_px_freepvt(void *const pvt)
{
	ncio_px *const pxp = (ncio_px *)pvt;
	if(pxp == NULL)
		return;

	if(pxp->slave != NULL)
	{
		if(pxp->slave->bf_base != NULL)
		{
			free(pxp->slave->bf_base);
			pxp->slave->bf_base = NULL;
			pxp->slave->bf_extent = 0;
			pxp->slave->bf_offset = OFF_NONE;
		}
		free(pxp->slave);
		pxp->slave = NULL;
	}

	if(pxp->bf_base != NULL)
	{
		free(pxp->bf_base);
		pxp->bf_base = NULL;
		pxp->bf_extent = 0;
		pxp->bf_offset = OFF_NONE;
	}
}


/* This is the second half of the ncio initialization. This is called
   after the file has actually been opened.

   The most important thing that happens is the allocation of a block
   of memory at pxp->bf_base. This is going to be twice the size of
   the chunksizehint (rounded up to the nearest sizeof(double)) passed
   in from nc__create or nc__open. The rounded chunksizehint (passed
   in here in sizehintp) is going to be stored as pxp->blksize.

   According to our "contract" we are not allowed to ask for an extent
   larger than this chunksize/sizehint/blksize from the ncio get
   function.

   nciop - pointer to the ncio struct
   sizehintp - pointer to a size hint that will be rounded up and
   passed back to the caller.
   isNew - true if this is being called from ncio_create for a new
   file.
*/
static int
ncio_px_init2(ncio *const nciop, size_t *sizehintp, int isNew)
{
	ncio_px *const pxp = (ncio_px *)nciop->pvt;
	const size_t bufsz = 2 * *sizehintp;

	assert(nciop->fd >= 0);

	pxp->blksz = *sizehintp;

	assert(pxp->bf_base == NULL);

	/* this is separate allocation because it may grow */
	pxp->bf_base = malloc(bufsz);
	if(pxp->bf_base == NULL)
		return ENOMEM;
	/* else */
	pxp->bf_cnt = 0;
	if(isNew)
	{
		/* save a read */
		pxp->pos = 0;
		pxp->bf_offset = 0;
		pxp->bf_extent = bufsz;
		(void) memset(pxp->bf_base, 0, pxp->bf_extent);
	}
	return NC_NOERR;
}


/* This is the first of a two-part initialization of the ncio struct.
   Here the rel, get, move, sync, and free function pointers are set
   to their POSIX non-NC_SHARE functions (ncio_px_*).

   The ncio_px struct is also partially initialized.
*/
static void
ncio_px_init(ncio *const nciop)
{
	ncio_px *const pxp = (ncio_px *)nciop->pvt;

	*((ncio_relfunc **)&nciop->rel) = ncio_px_rel; /* cast away const */
	*((ncio_getfunc **)&nciop->get) = ncio_px_get; /* cast away const */
	*((ncio_movefunc **)&nciop->move) = ncio_px_move; /* cast away const */
	*((ncio_syncfunc **)&nciop->sync) = ncio_px_sync; /* cast away const */
	*((ncio_filesizefunc **)&nciop->filesize) = ncio_px_filesize; /* cast away const */
	*((ncio_pad_lengthfunc **)&nciop->pad_length) = ncio_px_pad_length; /* cast away const */
	*((ncio_closefunc **)&nciop->close) = ncio_px_close; /* cast away const */

	pxp->blksz = 0;
	pxp->pos = -1;
	pxp->bf_offset = OFF_NONE;
	pxp->bf_extent = 0;
	pxp->bf_rflags = 0;
	pxp->bf_refcount = 0;
	pxp->bf_base = NULL;
	pxp->slave = NULL;

}

/* Begin spx */

/* This is the struct that gets hung of ncio->pvt(?) when the NC_SHARE
   flag is used.
*/
typedef struct ncio_spx {
	off_t pos;
	/* buffer */
	off_t	bf_offset;
	size_t	bf_extent;
	size_t	bf_cnt;
	void	*bf_base;
} ncio_spx;


/*ARGSUSED*/
/* This function releases the region specified by offset.

   For POSIX system, with NC_SHARE, this becomes the rel function
   pointed to by the ncio rel function pointer. It mearly checks for
   file write permission, then calls px_rel to do everything.

   nciop - pointer to ncio struct.

   offset - beginning of region.

   rflags - One of the RGN_* flags defined in ncio.h. If set to
   RGN_MODIFIED it means that the data in this region were modified,
   and it needs to be written out to the disk immediately (since we
   are not buffering with NC_SHARE on).

*/
static int
ncio_spx_rel(ncio *const nciop, off_t offset, int rflags)
{
	ncio_spx *const pxp = (ncio_spx *)nciop->pvt;
	int status = NC_NOERR;

	assert(pxp->bf_offset <= offset);
	assert(pxp->bf_cnt != 0);
	assert(pxp->bf_cnt <= pxp->bf_extent);
#ifdef X_ALIGN
	assert(offset < pxp->bf_offset + X_ALIGN);
	assert(pxp->bf_cnt % X_ALIGN == 0 );
#endif

	if(fIsSet(rflags, RGN_MODIFIED))
	{
		if(!fIsSet(nciop->ioflags, NC_WRITE))
			return EPERM; /* attempt to write readonly file */

		status = px_pgout(nciop, pxp->bf_offset,
			pxp->bf_cnt,
			pxp->bf_base, &pxp->pos);
		/* if error, invalidate buffer anyway */
	}
	pxp->bf_offset = OFF_NONE;
	pxp->bf_cnt = 0;
	return status;
}


/* Request that the region (offset, extent) be made available through
   *vpp.

   This function converts a file region specified by an offset and
   extent to a memory pointer. The region may be locked until the
   corresponding call to rel().

   For POSIX systems, with NC_SHARE.

   nciop - pointer to ncio struct for this file.
   offset - offset (from beginning of file?) to the data we want to
   read.
   extent - the number of bytes we want.
   rflags - One of the RGN_* flags defined in ncio.h. May be RGN_NOLOCK.
   vpp - handle to point at data when it's been read.
*/
static int
ncio_spx_get(ncio *const nciop,
		off_t offset, size_t extent,
		int rflags,
		void **const vpp)
{
	ncio_spx *const pxp = (ncio_spx *)nciop->pvt;
	int status = NC_NOERR;
#ifdef X_ALIGN
	size_t rem;
#endif

	if(fIsSet(rflags, RGN_WRITE) && !fIsSet(nciop->ioflags, NC_WRITE))
		return EPERM; /* attempt to write readonly file */

	assert(extent != 0);
	assert(extent < X_INT_MAX); /* sanity check */

	assert(pxp->bf_cnt == 0);

#ifdef X_ALIGN
	rem = (size_t)(offset % X_ALIGN);
	if(rem != 0)
	{
		offset -= rem;
		extent += rem;
	}

	{
      const size_t rndup = extent % X_ALIGN;
      if(rndup != 0)
        extent += X_ALIGN - rndup;
	}

	assert(offset % X_ALIGN == 0);
	assert(extent % X_ALIGN == 0);
#endif

	if(pxp->bf_extent < extent)
	{
		if(pxp->bf_base != NULL)
		{
			free(pxp->bf_base);
			pxp->bf_base = NULL;
			pxp->bf_extent = 0;
		}
		assert(pxp->bf_extent == 0);
		pxp->bf_base = malloc(extent+1);
		if(pxp->bf_base == NULL)
			return ENOMEM;
		pxp->bf_extent = extent;
	}

	status = px_pgin(nciop, offset,
		 extent,
		 pxp->bf_base,
		 &pxp->bf_cnt, &pxp->pos);
	if(status != NC_NOERR)
		return status;

	pxp->bf_offset = offset;

	if(pxp->bf_cnt < extent)
		pxp->bf_cnt = extent;

#ifdef X_ALIGN
	*vpp = (char *)pxp->bf_base + rem;
#else
	*vpp = pxp->bf_base;
#endif
	return NC_NOERR;
}


#if 0
/*ARGSUSED*/
static int
strategy(ncio *const nciop, off_t to, off_t offset,
			size_t extent, int rflags)
{
	static ncio_spx pxp[1];
	int status = NC_NOERR;
#ifdef X_ALIGN
	size_t rem;
#endif

	assert(extent != 0);
	assert(extent < X_INT_MAX); /* sanity check */
#if INSTRUMENT
fprintf(stderr, "strategy %ld at %ld to %ld\n",
	 (long)extent, (long)offset, (long)to);
#endif


#ifdef X_ALIGN
	rem = (size_t)(offset % X_ALIGN);
	if(rem != 0)
	{
		offset -= rem;
		extent += rem;
	}

	{
		const size_t rndup = extent % X_ALIGN;
		if(rndup != 0)
			extent += X_ALIGN - rndup;
	}

	assert(offset % X_ALIGN == 0);
	assert(extent % X_ALIGN == 0);
#endif

	if(pxp->bf_extent < extent)
	{
		if(pxp->bf_base != NULL)
		{
			free(pxp->bf_base);
			pxp->bf_base = NULL;
			pxp->bf_extent = 0;
		}
		assert(pxp->bf_extent == 0);
		pxp->bf_base = malloc(extent);
		if(pxp->bf_base == NULL)
			return ENOMEM;
		pxp->bf_extent = extent;
	}

	status = px_pgin(nciop, offset,
		 extent,
		 pxp->bf_base,
		 &pxp->bf_cnt, &pxp->pos);
	if(status != NC_NOERR)
		return status;

	pxp->bf_offset = to; /* TODO: XALIGN */

	if(pxp->bf_cnt < extent)
		pxp->bf_cnt = extent;

	status = px_pgout(nciop, pxp->bf_offset,
		pxp->bf_cnt,
		pxp->bf_base, &pxp->pos);
	/* if error, invalidate buffer anyway */
	pxp->bf_offset = OFF_NONE;
	pxp->bf_cnt = 0;
	return status;
}
#endif

/* Copy one region to another without making anything available to
   higher layers. May be just implemented in terms of get() and rel(),
   or may be tricky to be efficient.  Only used in by nc_enddef()
   after redefinition.

   nciop - pointer to ncio struct for this file.
   to - dest for move?
   from - src for move?
   nbytes - number of bytes to move.
   rflags - One of the RGN_* flags defined in ncio.h.
*/
static int
ncio_spx_move(ncio *const nciop, off_t to, off_t from,
			size_t nbytes, int rflags)
{
	int status = NC_NOERR;
	off_t lower = from;
	off_t upper = to;
	char *base;
	size_t diff;
	size_t extent;

	rflags &= RGN_NOLOCK; /* filter unwanted flags */

	if(to == from)
		return NC_NOERR; /* NOOP */

	if(to > from)
	{
		/* growing */
		lower = from;
		upper = to;
	}
	else
	{
		/* shrinking */
		lower = to;
		upper = from;
	}

	diff = (size_t)(upper - lower);
	extent = diff + nbytes;

	status = ncio_spx_get(nciop, lower, extent, RGN_WRITE|rflags,
			(void **)&base);

	if(status != NC_NOERR)
		return status;

	if(to > from)
		(void) memmove(base + diff, base, nbytes);
	else
		(void) memmove(base, base + diff, nbytes);

	(void) ncio_spx_rel(nciop, lower, RGN_MODIFIED);

	return status;
}


/*ARGSUSED*/
/* Flush any buffers to disk. May be a no-op on if I/O is unbuffered.
*/
static int
ncio_spx_sync(ncio *const nciop)
{
	/* NOOP */
	return NC_NOERR;
}

static void
ncio_spx_freepvt(void *const pvt)
{
	ncio_spx *const pxp = (ncio_spx *)pvt;
	if(pxp == NULL)
		return;

	if(pxp->bf_base != NULL)
	{
		free(pxp->bf_base);
		pxp->bf_base = NULL;
		pxp->bf_offset = OFF_NONE;
		pxp->bf_extent = 0;
		pxp->bf_cnt = 0;
	}
}


/* This does the second half of the ncio_spx struct initialization for
   POSIX systems, with NC_SHARE on.

   nciop - pointer to ncio struct for this file. File has been opened.
   sizehintp - pointer to a size which will be rounded up to the
   nearest 8-byt boundary and then used as the max size "chunk" (or
   page) to read from the file.
*/
static int
ncio_spx_init2(ncio *const nciop, const size_t *const sizehintp)
{
	ncio_spx *const pxp = (ncio_spx *)nciop->pvt;

	assert(nciop->fd >= 0);

	pxp->bf_extent = *sizehintp;

	assert(pxp->bf_base == NULL);

	/* this is separate allocation because it may grow */
	pxp->bf_base = malloc(pxp->bf_extent);
	if(pxp->bf_base == NULL)
	{
		pxp->bf_extent = 0;
		return ENOMEM;
	}
	/* else */
	return NC_NOERR;
}


/* First half of init for ncio_spx struct, setting the rel, get, move,
   snyc, and free function pointers to the NC_SHARE versions of these
   functions (i.e. the ncio_spx_* functions).
*/
static void
ncio_spx_init(ncio *const nciop)
{
	ncio_spx *const pxp = (ncio_spx *)nciop->pvt;

	*((ncio_relfunc **)&nciop->rel) = ncio_spx_rel; /* cast away const */
	*((ncio_getfunc **)&nciop->get) = ncio_spx_get; /* cast away const */
	*((ncio_movefunc **)&nciop->move) = ncio_spx_move; /* cast away const */
	*((ncio_syncfunc **)&nciop->sync) = ncio_spx_sync; /* cast away const */
	/* shared with _px_ */
	*((ncio_filesizefunc **)&nciop->filesize) = ncio_px_filesize; /* cast away const */
	*((ncio_pad_lengthfunc **)&nciop->pad_length) = ncio_px_pad_length; /* cast away const */
	*((ncio_closefunc **)&nciop->close) = ncio_spx_close; /* cast away const */

	pxp->pos = -1;
	pxp->bf_offset = OFF_NONE;
	pxp->bf_extent = 0;
	pxp->bf_cnt = 0;
	pxp->bf_base = NULL;
}


/* */

/* This will call whatever free function is attached to the free
   function pointer in ncio. It's called from ncio_close, and from
   ncio_open and ncio_create when an error occurs that the file
   metadata must be freed.
*/
static void
ncio_px_free(ncio *nciop)
{
	if(nciop == NULL)
		return;
	if(nciop->pvt != NULL)
		ncio_px_freepvt(nciop->pvt);
	free(nciop);
}

static void
ncio_spx_free(ncio *nciop)
{
	if(nciop == NULL)
		return;
	if(nciop->pvt != NULL)
		ncio_spx_freepvt(nciop->pvt);
	free(nciop);
}


/* Create a new ncio struct to hold info about the file. This will
   create and init the ncio_px or ncio_spx struct (the latter if
   NC_SHARE is used.)
*/
static ncio *
ncio_px_new(const char *path, int ioflags)
{
	size_t sz_ncio = M_RNDUP(sizeof(ncio));
	size_t sz_path = M_RNDUP(strlen(path) +1);
	size_t sz_ncio_pvt;
	ncio *nciop;

#if ALWAYS_NC_SHARE /* DEBUG */
	fSet(ioflags, NC_SHARE);
#endif

	if(fIsSet(ioflags, NC_SHARE))
		sz_ncio_pvt = sizeof(ncio_spx);
	else
		sz_ncio_pvt = sizeof(ncio_px);

	nciop = (ncio *) malloc(sz_ncio + sz_path + sz_ncio_pvt);
	if(nciop == NULL)
		return NULL;

	nciop->ioflags = ioflags;
	*((int *)&nciop->fd) = -1; /* cast away const */

	nciop->path = (char *) ((char *)nciop + sz_ncio);
	(void) strcpy((char *)nciop->path, path); /* cast away const */

				/* cast away const */
	*((void **)&nciop->pvt) = (void *)(nciop->path + sz_path);

	if(fIsSet(ioflags, NC_SHARE))
		ncio_spx_init(nciop);
	else
		ncio_px_init(nciop);

	return nciop;
}


/* Public below this point */
#ifndef NCIO_MINBLOCKSIZE
#define NCIO_MINBLOCKSIZE 256
#endif
#ifndef NCIO_MAXBLOCKSIZE
#define NCIO_MAXBLOCKSIZE 268435456 /* sanity check, about X_SIZE_T_MAX/8 */
#endif

#ifdef S_IRUSR
#define NC_DEFAULT_CREAT_MODE \
        (S_IRUSR|S_IWUSR|S_IRGRP|S_IWGRP|S_IROTH|S_IWOTH) /* 0666 */

#else
#define NC_DEFAULT_CREAT_MODE 0666
#endif

/* Create a file, and the ncio struct to go with it. This function is
   only called from nc__create_mp.

   path - path of file to create.
   ioflags - flags from nc_create
   initialsz - From the netcdf man page: "The argument
   Iinitialsize sets the initial size of the file at creation time."
   igeto -
   igetsz -
   sizehintp - this eventually goes into pxp->blksz and is the size of
   a page of data for buffered reads and writes.
   nciopp - pointer to a pointer that will get location of newly
   created and inited ncio struct.
   igetvpp - pointer to pointer which will get the location of ?
*/
int
posixio_create(const char *path, int ioflags,
	size_t initialsz,
	off_t igeto, size_t igetsz, size_t *sizehintp,
	void* parameters,
	ncio **nciopp, void **const igetvpp)
{
	ncio *nciop;
	int oflags = (O_RDWR|O_CREAT);
	int fd;
	int status;

	if(initialsz < (size_t)igeto + igetsz)
		initialsz = (size_t)igeto + igetsz;

	fSet(ioflags, NC_WRITE);

	if(path == NULL || *path == 0)
		return EINVAL;

	nciop = ncio_px_new(path, ioflags);
	if(nciop == NULL)
		return ENOMEM;

	if(fIsSet(ioflags, NC_NOCLOBBER))
		fSet(oflags, O_EXCL);
	else
		fSet(oflags, O_TRUNC);
#ifdef O_BINARY
	fSet(oflags, O_BINARY);
#endif
#ifdef vms
	fd = open(path, oflags, NC_DEFAULT_CREAT_MODE, "ctx=stm");
#else
	/* Should we mess with the mode based on NC_SHARE ?? */
	fd = open(path, oflags, NC_DEFAULT_CREAT_MODE);
#endif
#if 0
	(void) fprintf(stderr, "ncio_create(): path=\"%s\"\n", path);
	(void) fprintf(stderr, "ncio_create(): oflags=0x%x\n", oflags);
#endif
	if(fd < 0)
	{
		status = errno;
		goto unwind_new;
	}
	*((int *)&nciop->fd) = fd; /* cast away const */

	if(*sizehintp < NCIO_MINBLOCKSIZE)
	{
		/* Use default */
		*sizehintp = blksize(fd);
	}
	else if(*sizehintp >= NCIO_MAXBLOCKSIZE)
	{
		/* Use maximum allowed value */
		*sizehintp = NCIO_MAXBLOCKSIZE;
	}
	else
	{
		*sizehintp = M_RNDUP(*sizehintp);
	}

	if(fIsSet(nciop->ioflags, NC_SHARE))
		status = ncio_spx_init2(nciop, sizehintp);
	else
		status = ncio_px_init2(nciop, sizehintp, 1);

	if(status != NC_NOERR)
		goto unwind_open;

	if(initialsz != 0)
	{
		status = fgrow(fd, (off_t)initialsz);
		if(status != NC_NOERR)
			goto unwind_open;
	}

	if(igetsz != 0)
	{
		status = nciop->get(nciop,
				igeto, igetsz,
                        	RGN_WRITE,
                        	igetvpp);
		if(status != NC_NOERR)
			goto unwind_open;
	}

	*nciopp = nciop;
	return NC_NOERR;

unwind_open:
	(void) close(fd);
	/* ?? unlink */
	/*FALLTHRU*/
unwind_new:
	ncio_close(nciop,!fIsSet(ioflags, NC_NOCLOBBER));
	return status;
}


/* This function opens the data file. It is only called from nc.c,
   from nc__open_mp and nc_delete_mp.

   path - path of data file.

   ioflags - flags passed into nc_open.

   igeto - looks like this function can do an initial page get, and
   igeto is going to be the offset for that. But it appears to be
   unused

   igetsz - the size in bytes of initial page get (a.k.a. extent). Not
   ever used in the library.

   sizehintp - pointer to sizehint parameter from nc__open or
   nc__create. This is used to set pxp->blksz.

   Here's what the man page has to say:

   "The argument referenced by chunksize controls a space versus time
   tradeoff, memory allocated in the netcdf library versus number of
   system calls.

   Because of internal requirements, the value may not be set to
   exactly the value requested. The actual value chosen is returned by reference.

   Using the value NC_SIZEHINT_DEFAULT causes the library to choose a
   default. How the system choses the default depends on the
   system. On many systems, the "preferred I/O block size" is
   available from the stat() system call, struct stat member
   st_blksize. If this is available it is used. Lacking that, twice
   the system pagesize is used. Lacking a call to discover the system
   pagesize, we just set default chunksize to 8192.

   The chunksize is a property of a given open netcdf descriptor ncid,
   it is not a persistent property of the netcdf dataset."

   nciopp - pointer to pointer that will get address of newly created
   and inited ncio struct.

   igetvpp - handle to pass back pointer to data from inital page
   read, if this were ever used, which it isn't.
*/
int
posixio_open(const char *path,
	int ioflags,
	off_t igeto, size_t igetsz, size_t *sizehintp,
        void* parameters,
	ncio **nciopp, void **const igetvpp)
{
	ncio *nciop;
	int oflags = fIsSet(ioflags, NC_WRITE) ? O_RDWR : O_RDONLY;
	int fd = -1;
	int status = 0;

	if(path == NULL || *path == 0)
		return EINVAL;

	nciop = ncio_px_new(path, ioflags);
	if(nciop == NULL)
		return ENOMEM;

#ifdef O_BINARY
	/*#if _MSC_VER*/
	fSet(oflags, O_BINARY);
#endif

#ifdef vms
	fd = open(path, oflags, 0, "ctx=stm");
#else
	fd = open(path, oflags, 0);
#endif
	if(fd < 0)
	{
		status = errno;
		goto unwind_new;
	}
	*((int *)&nciop->fd) = fd; /* cast away const */

	if(*sizehintp < NCIO_MINBLOCKSIZE)
	{
		/* Use default */
		*sizehintp = blksize(fd);
	}
	else if(*sizehintp >= NCIO_MAXBLOCKSIZE)
	{
		/* Use maximum allowed value */
		*sizehintp = NCIO_MAXBLOCKSIZE;
	}
	else
	{
		*sizehintp = M_RNDUP(*sizehintp);
	}

	if(fIsSet(nciop->ioflags, NC_SHARE))
		status = ncio_spx_init2(nciop, sizehintp);
	else
		status = ncio_px_init2(nciop, sizehintp, 0);

	if(status != NC_NOERR)
		goto unwind_open;

	if(igetsz != 0)
	{
		status = nciop->get(nciop,
				igeto, igetsz,
                        	0,
                        	igetvpp);
		if(status != NC_NOERR)
			goto unwind_open;
	}

	*nciopp = nciop;
	return NC_NOERR;

unwind_open:
	(void) close(fd); /* assert fd >= 0 */
	/*FALLTHRU*/
unwind_new:
	ncio_close(nciop,0);
	return status;
}

/*
 * Get file size in bytes.
 */
static int
ncio_px_filesize(ncio *nciop, off_t *filesizep)
{


	/* There is a problem with fstat on Windows based systems
		which manifests (so far) when Config RELEASE is built.
		Use _filelengthi64 isntead. */
#ifdef HAVE_FILE_LENGTH_I64

	__int64 file_len = 0;
	if( (file_len = _filelengthi64(nciop->fd)) < 0) {
		return errno;
	}

	*filesizep = file_len;

#else
    struct stat sb;
    assert(nciop != NULL);
    if (fstat(nciop->fd, &sb) < 0)
	return errno;
    *filesizep = sb.st_size;
#endif
	return NC_NOERR;
}

/*
 * Sync any changes to disk, then truncate or extend file so its size
 * is length.  This is only intended to be called before close, if the
 * file is open for writing and the actual size does not match the
 * calculated size, perhaps as the result of having been previously
 * written in NOFILL mode.
 */
static int
ncio_px_pad_length(ncio *nciop, off_t length)
{

	int status = NC_NOERR;

	if(nciop == NULL)
		return EINVAL;

	if(!fIsSet(nciop->ioflags, NC_WRITE))
	        return EPERM; /* attempt to write readonly file */

	status = nciop->sync(nciop);
	if(status != NC_NOERR)
	        return status;

 	status = fgrow2(nciop->fd, length);
 	if(status != NC_NOERR)
	        return status;
	return NC_NOERR;
}


/* Write out any dirty buffers to disk and
   ensure that next read will get data from disk.

   Sync any changes, then close the open file associated with the ncio
   struct, and free its memory.

   nciop - pointer to ncio to close.

   doUnlink - if true, unlink file
*/
static int
ncio_px_close(ncio *nciop, int doUnlink)
{
	int status = NC_NOERR;
	if(nciop == NULL)
		return EINVAL;
	if(nciop->fd > 0) {
	    status = nciop->sync(nciop);
	    (void) close(nciop->fd);
	}
	if(doUnlink)
		(void) unlink(nciop->path);
	ncio_px_free(nciop);
	return status;
}

static int
ncio_spx_close(ncio *nciop, int doUnlink)
{
	int status = NC_NOERR;
	if(nciop == NULL)
		return EINVAL;
	if(nciop->fd > 0) {
	    status = nciop->sync(nciop);
	    (void) close(nciop->fd);
	}
	if(doUnlink)
		(void) unlink(nciop->path);
	ncio_spx_free(nciop);
	return status;
}