samba.git - Unnamed repository; edit this file 'description' to name the repository.

diff options

author	Jeremy Allison <jra@samba.org>	2015-03-05 12:48:47 -0800
committer	Jeremy Allison <jra@samba.org>	2015-03-08 18:24:06 +0100
commit	3289a5d84f73bf044e5767a6c47a3f7bf8357c08 (patch)
tree	b231566d9d0082f792952f696a3af69ad9351935 /python/samba/sd_utils.py
parent	e53f6e9ec2eb05ded39a0caec407bb3bb23039cf (diff)

lib: talloc: Fix bug when calling a destructor.

If the destructor itself calls talloc_set_destructor() and returns -1, the new destructor set is overwritten by talloc. Dectect that and leave the new destructor in place. Signed-off-by: Jeremy Allison <jra@samba.org> Reviewed-by: Ira Cooper <ira@samba.org>

Diffstat (limited to 'python/samba/sd_utils.py')

0 files changed, 0 insertions, 0 deletions

n67'>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 1910 1911 1912 1913 1914 1915 1916 1917 1918 1919 1920 1921 1922 1923 1924 1925 1926 1927 1928 1929 1930 1931 1932 1933 1934 1935 1936 1937 1938 1939 1940 1941 1942 1943 1944 1945 1946 1947 1948 1949 1950 1951 1952 1953 1954 1955 1956 1957 1958 1959 1960 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2036 2037 2038 2039 2040

/*
   Copyright (c) 2013 Red Hat, Inc. <http://www.redhat.com>
   This file is part of GlusterFS.

   This file is licensed to you under your choice of the GNU Lesser
   General Public License, version 3 or any later version (LGPLv3 or
   later), or the GNU General Public License, version 2 (GPLv2), in all
   cases as published by the Free Software Foundation.
*/

#include "xlator.h"
#include "defaults.h"
#include "logging.h"
#include "iobuf.h"
#include "syscall.h"

#include "changelog-helpers.h"
#include "changelog-encoders.h"
#include "changelog-mem-types.h"
#include "changelog-messages.h"

#include "changelog-encoders.h"
#include "changelog-rpc-common.h"
#include <pthread.h>

static void
changelog_cleanup_free_mutex (void *arg_mutex)
{
    pthread_mutex_t *p_mutex = (pthread_mutex_t*) arg_mutex;

    if (p_mutex)
            pthread_mutex_unlock(p_mutex);
}

int
changelog_thread_cleanup (xlator_t *this, pthread_t thr_id)
{
        int   ret    = 0;
        void *retval = NULL;

        /* send a cancel request to the thread */
        ret = pthread_cancel (thr_id);
        if (ret != 0) {
                gf_msg (this->name, GF_LOG_ERROR, errno,
                        CHANGELOG_MSG_PTHREAD_CANCEL_FAILED,
                        "could not cancel thread");
                goto out;
        }

        ret = pthread_join (thr_id, &retval);
        if ((ret != 0) || (retval != PTHREAD_CANCELED)) {
                gf_msg (this->name, GF_LOG_ERROR, errno,
                        CHANGELOG_MSG_PTHREAD_CANCEL_FAILED,
                        "cancel request not adhered as expected");
        }

 out:
        return ret;
}

void *
changelog_get_usable_buffer (changelog_local_t *local)
{
        changelog_log_data_t *cld = NULL;

        if (!local)
                return NULL;

        cld = &local->cld;
        if (!cld->cld_iobuf)
                return NULL;

        return cld->cld_iobuf->ptr;
}

static int
changelog_selector_index (unsigned int selector)
{
        return (ffs (selector) - 1);
}

int
changelog_ev_selected (xlator_t *this,
                       changelog_ev_selector_t *selection,
                       unsigned int selector)
{
        int idx = 0;

        idx = changelog_selector_index (selector);
        gf_msg_debug (this->name, 0,
                      "selector ref count for %d (idx: %d): %d",
                      selector, idx, selection->ref[idx]);
        /* this can be lockless */
        return (idx < CHANGELOG_EV_SELECTION_RANGE
                 && (selection->ref[idx] > 0));
}

void
changelog_select_event (xlator_t *this,
                        changelog_ev_selector_t *selection,
                        unsigned int selector)
{
        int idx = 0;

        LOCK (&selection->reflock);
        {
                while (selector) {
                        idx = changelog_selector_index (selector);
                        if (idx < CHANGELOG_EV_SELECTION_RANGE) {
                                selection->ref[idx]++;
                                gf_msg_debug (this->name, 0,
                                              "selecting event %d", idx);
                        }
                        selector &= ~(1 << idx);
                }
        }
        UNLOCK (&selection->reflock);
}

void
changelog_deselect_event (xlator_t *this,
                          changelog_ev_selector_t *selection,
                          unsigned int selector)
{
        int idx = 0;

        LOCK (&selection->reflock);
        {
                while (selector) {
                        idx = changelog_selector_index (selector);
                        if (idx < CHANGELOG_EV_SELECTION_RANGE) {
                                selection->ref[idx]--;
                                gf_msg_debug (this->name, 0,
                                              "de-selecting event %d", idx);
                        }
                        selector &= ~(1 << idx);
                }
        }
        UNLOCK (&selection->reflock);
}

int
changelog_init_event_selection (xlator_t *this,
                                changelog_ev_selector_t *selection)
{
        int ret = 0;
        int j = CHANGELOG_EV_SELECTION_RANGE;

        ret = LOCK_INIT (&selection->reflock);
        if (ret != 0)
                return -1;

        LOCK (&selection->reflock);
        {
                while (j--) {
                        selection->ref[j] = 0;
                }
        }
        UNLOCK (&selection->reflock);

        return 0;
}

int
changelog_cleanup_event_selection (xlator_t *this,
                                   changelog_ev_selector_t *selection)
{
        int j = CHANGELOG_EV_SELECTION_RANGE;

        LOCK (&selection->reflock);
        {
                while (j--) {
                        if (selection->ref[j] > 0)
                                gf_msg (this->name, GF_LOG_WARNING, 0,
                                        CHANGELOG_MSG_CLEANUP_ON_ACTIVE_REF,
                                        "changelog event selection cleaning up "
                                        " on active references");
                }
        }
        UNLOCK (&selection->reflock);

        return LOCK_DESTROY (&selection->reflock);
}

static void
changelog_perform_dispatch (xlator_t *this,
                            changelog_priv_t *priv, void *mem, size_t size)
{
        char *buf    = NULL;
        void *opaque = NULL;

        buf = rbuf_reserve_write_area (priv->rbuf, size, &opaque);
        if (!buf) {
                gf_msg_callingfn (this->name,
                                  GF_LOG_WARNING, 0,
                                  CHANGELOG_MSG_DISPATCH_EVENT_FAILED,
                                  "failed to dispatch event");
                return;
        }

        memcpy (buf, mem, size);
        rbuf_write_complete (opaque);
}

void
changelog_dispatch_event (xlator_t *this,
                          changelog_priv_t *priv, changelog_event_t *ev)
{
        changelog_ev_selector_t *selection = NULL;

        selection = &priv->ev_selection;
        if (changelog_ev_selected (this, selection, ev->ev_type)) {
                changelog_perform_dispatch (this, priv, ev, CHANGELOG_EV_SIZE);
        }
}

void
changelog_set_usable_record_and_length (changelog_local_t *local,
                                        size_t len, int xr)
{
        changelog_log_data_t *cld = NULL;

        cld = &local->cld;

        cld->cld_ptr_len = len;
        cld->cld_xtra_records = xr;
}

void
changelog_local_cleanup (xlator_t *xl, changelog_local_t *local)
{
        int                   i   = 0;
        changelog_opt_t      *co  = NULL;
        changelog_log_data_t *cld = NULL;

        if (!local)
                return;

        cld = &local->cld;

        /* cleanup dynamic allocation for extra records */
        if (cld->cld_xtra_records) {
                co = (changelog_opt_t *) cld->cld_ptr;
                for (; i < cld->cld_xtra_records; i++, co++)
                        if (co->co_free)
                                co->co_free (co);
        }

        CHANGELOG_IOBUF_UNREF (cld->cld_iobuf);

        if (local->inode)
                inode_unref (local->inode);

        mem_put (local);
}

int
changelog_write (int fd, char *buffer, size_t len)
{
        ssize_t size = 0;
        size_t written = 0;

        while (written < len) {
                size = sys_write (fd,
                              buffer + written, len - written);
                if (size <= 0)
                        break;

                written += size;
        }

        return (written != len);
}

int
htime_update (xlator_t *this,
              changelog_priv_t *priv, unsigned long ts,
              char * buffer)
{
        char changelog_path[PATH_MAX+1]   = {0,};
        int len                           = -1;
        char x_value[25]                  = {0,};
        /* time stamp(10) + : (1) + rolltime (12 ) + buffer (2) */
        int ret                           = 0;

        if (priv->htime_fd ==-1) {
                gf_msg (this->name, GF_LOG_ERROR, 0,
                        CHANGELOG_MSG_HTIME_ERROR,
                        "Htime fd not available for updation");
                ret = -1;
                goto out;
        }
        strncpy (changelog_path, buffer, PATH_MAX);
        len = strlen (changelog_path);
        changelog_path[len] = '\0'; /* redundant */

        if (changelog_write (priv->htime_fd, (void*) changelog_path, len+1 ) < 0) {
                gf_msg (this->name, GF_LOG_ERROR, 0,
                        CHANGELOG_MSG_HTIME_ERROR,
                        "Htime file content write failed");
                ret =-1;
                goto out;
        }

        snprintf (x_value, sizeof x_value, "%lu:%d",
                 ts, priv->rollover_count);

        if (sys_fsetxattr (priv->htime_fd, HTIME_KEY, x_value,
                           strlen (x_value), XATTR_REPLACE)) {
                gf_smsg (this->name, GF_LOG_ERROR, errno,
                         CHANGELOG_MSG_HTIME_ERROR,
                         "Htime xattr updation failed with XATTR_REPLACE",
                         "changelog=%s", changelog_path,
                         NULL);

                if (sys_fsetxattr (priv->htime_fd, HTIME_KEY, x_value,
                                   strlen (x_value), 0)) {
                        gf_smsg (this->name, GF_LOG_ERROR, errno,
                                 CHANGELOG_MSG_HTIME_ERROR,
                                 "Htime xattr updation failed",
                                 "changelog=%s", changelog_path,
                                 NULL);
                        ret = -1;
                        goto out;
                }
        }

        priv->rollover_count +=1;

out:
        return ret;
}

/*
 * Description: Check if the changelog to rollover is empty or not.
 * It is assumed that fd passed is already verified.
 *
 * Returns:
 * 1 : If found empty, changed path from "CHANGELOG.<TS>" to "changelog.<TS>"
 * 0 : If NOT empty, proceed usual.
 */
int
cl_is_empty (xlator_t *this, int fd)
{
        int             ret             = -1;
        size_t          elen            = 0;
        int             encoding        = -1;
        char            buffer[1024]    = {0,};
        struct stat     stbuf           = {0,};
        int             major_version   = -1;
        int             minor_version   = -1;

        ret = sys_fstat (fd, &stbuf);
        if (ret) {
                gf_msg (this->name, GF_LOG_ERROR, errno,
                        CHANGELOG_MSG_FSTAT_OP_FAILED,
                        "Could not stat (CHANGELOG)");
                goto out;
        }

        ret = sys_lseek (fd, 0, SEEK_SET);
        if (ret == -1) {
                gf_msg (this->name, GF_LOG_ERROR, errno,
                        CHANGELOG_MSG_LSEEK_OP_FAILED,
                        "Could not lseek (CHANGELOG)");
                goto out;
        }

        CHANGELOG_GET_HEADER_INFO (fd, buffer, 1024, encoding,
                                   major_version, minor_version, elen);

        if (elen == stbuf.st_size) {
                ret = 1;
        } else {
                ret = 0;
        }

out:
        return ret;
}

/*
 * Description: Updates "CHANGELOG" to "changelog" for writing changelog path
 * to htime file.
 *
 * Returns:
 * 0  : Success
 * -1 : Error
 */
int
update_path (xlator_t *this, char *cl_path)
{
        char low_cl[]     = "changelog";
        char up_cl[]      = "CHANGELOG";
        char *found       = NULL;
        int  ret          = -1;

        found = strstr(cl_path, up_cl);

        if (found == NULL) {
                gf_msg (this->name, GF_LOG_ERROR, errno,
                        CHANGELOG_MSG_LSEEK_OP_FAILED,
                        "Could not find CHANGELOG in changelog path");
                goto out;
        } else {
                strncpy(found, low_cl, strlen(low_cl));
        }

        ret = 0;
out:
        return ret;
}

static int
changelog_rollover_changelog (xlator_t *this,
                              changelog_priv_t *priv, unsigned long ts)
{
        int   ret             = -1;
        int   notify          = 0;
        int   cl_empty_flag   = 0;
        char  ofile[PATH_MAX] = {0,};
        char  nfile[PATH_MAX] = {0,};
        changelog_event_t ev  = {0,};

        if (priv->changelog_fd != -1) {
                ret = sys_fsync (priv->changelog_fd);
                if (ret < 0) {
                        gf_msg (this->name, GF_LOG_ERROR, errno,
                                CHANGELOG_MSG_FSYNC_OP_FAILED,
                                "fsync failed");
                }
                ret = cl_is_empty (this, priv->changelog_fd);
                if (ret == 1) {
                        cl_empty_flag = 1;
                } else if (ret == -1) {
                        /* Log error but proceed as usual */
                        gf_msg (this->name, GF_LOG_WARNING, 0,
                                CHANGELOG_MSG_DETECT_EMPTY_CHANGELOG_FAILED,
                                "Error detecting empty changelog");
                }
                sys_close (priv->changelog_fd);
                priv->changelog_fd = -1;
        }

        (void) snprintf (ofile, PATH_MAX,
                         "%s/"CHANGELOG_FILE_NAME, priv->changelog_dir);
        (void) snprintf (nfile, PATH_MAX,
                         "%s/"CHANGELOG_FILE_NAME".%lu",
                         priv->changelog_dir, ts);

        if (cl_empty_flag == 1) {
                ret = sys_unlink (ofile);
                if (ret) {
                        gf_smsg (this->name, GF_LOG_ERROR, errno,
                                 CHANGELOG_MSG_UNLINK_OP_FAILED,
                                 "error unlinking empty changelog",
                                 "path=%s", ofile,
                                 NULL);
                        ret = 0;  /* Error in unlinking empty changelog should
                                     not break further changelog operation, so
                                     reset return value to 0*/
                }
        } else {
                ret = sys_rename (ofile, nfile);

                if (ret && (errno == ENOENT)) {
                        ret = 0;
                        goto out;
                }
                if (ret) {
                        gf_smsg (this->name, GF_LOG_ERROR, errno,
                                 CHANGELOG_MSG_RENAME_ERROR,
                                 "error renaming",
                                 "from=%s", ofile,
                                 "to=%s", nfile,
                                 NULL);
                }
        }

        if (!ret && (cl_empty_flag == 0)) {
                        notify = 1;
        }

        if (!ret) {
                if (cl_empty_flag) {
                        update_path (this, nfile);
                }
                ret = htime_update (this, priv, ts, nfile);
                if (ret == -1) {
                        gf_msg (this->name, GF_LOG_ERROR,
                                0, CHANGELOG_MSG_HTIME_ERROR,
                                "could not update htime file");
                        goto out;
                }
        }

        if (notify) {
                ev.ev_type = CHANGELOG_OP_TYPE_JOURNAL;
                memcpy (ev.u.journal.path, nfile, strlen (nfile) + 1);
                changelog_dispatch_event (this, priv, &ev);
        }
 out:
        /* If this is explicit rollover initiated by snapshot,
         * wakeup reconfigure thread waiting for changelog to
         * rollover. This should happen even in failure cases as
         * well otherwise snapshot will timeout and fail. Hence
         * moved under out.
         */
        if (priv->explicit_rollover) {
                priv->explicit_rollover = _gf_false;

                pthread_mutex_lock (&priv->bn.bnotify_mutex);
                {
                        if (ret) {
                                priv->bn.bnotify_error = _gf_true;
                                gf_msg (this->name, GF_LOG_ERROR, 0,
                                        CHANGELOG_MSG_EXPLICIT_ROLLOVER_FAILED,
                                        "Fail snapshot because of "
                                        "previous errors");
                        } else {
                                gf_smsg (this->name, GF_LOG_INFO, 0,
                                         CHANGELOG_MSG_BNOTIFY_INFO, "Explicit "
                                         "rollover changelog signaling "
                                         "bnotify",
                                         "changelog=%s", nfile,
                                         NULL);
                        }
                        priv->bn.bnotify = _gf_false;
                        pthread_cond_signal (&priv->bn.bnotify_cond);
                }
                pthread_mutex_unlock (&priv->bn.bnotify_mutex);
        }
        return ret;
}

int
filter_cur_par_dirs (const struct dirent *entry)
{
        if (entry == NULL)
                return 0;

        if ((strcmp(entry->d_name, ".") == 0) ||
            (strcmp(entry->d_name, "..") == 0))
                return 0;
        else
                return 1;
}

/*
 * find_current_htime:
 *       It finds the latest htime file and sets the HTIME_CURRENT
 *       xattr.
 *       RETURN VALUE:
 *           -1 : Error
 *           ret: Number of directory entries;
 */

int
find_current_htime (int ht_dir_fd, const char *ht_dir_path, char *ht_file_bname)
{
        struct dirent       **namelist = NULL;
        int                   ret      = 0;
        int                   cnt      = 0;
        int                   i        = 0;
        xlator_t             *this     = NULL;

        this = THIS;
        GF_ASSERT (this);
        GF_ASSERT (ht_dir_path);

        cnt = scandir (ht_dir_path, &namelist, filter_cur_par_dirs, alphasort);
        if (cnt < 0) {
                gf_msg (this->name, GF_LOG_ERROR, errno,
                        CHANGELOG_MSG_SCAN_DIR_FAILED,
                        "scandir failed");
        } else if (cnt > 0) {
                strncpy (ht_file_bname, namelist[cnt - 1]->d_name, NAME_MAX);
                ht_file_bname[NAME_MAX - 1] = 0;

                if (sys_fsetxattr (ht_dir_fd, HTIME_CURRENT, ht_file_bname,
                    strlen (ht_file_bname), 0)) {
                        gf_msg (this->name, GF_LOG_ERROR, errno,
                                CHANGELOG_MSG_FSETXATTR_FAILED,
                                "fsetxattr failed: HTIME_CURRENT");
                        ret = -1;
                        goto out;
                }

                if (sys_fsync (ht_dir_fd) < 0) {
                        gf_msg (this->name, GF_LOG_ERROR, errno,
                                CHANGELOG_MSG_FSYNC_OP_FAILED,
                                "fsync failed");
                        ret = -1;
                        goto out;
                }
        }

 out:
        for (i = 0; i < cnt; i++)
                free (namelist[i]);
        free (namelist);

        if (ret)
                cnt = ret;

        return cnt;
}

/* Returns 0 on successful open of htime file
 * returns -1 on failure or error
 */
int
htime_open (xlator_t *this,
              changelog_priv_t *priv, unsigned long ts)
{
        int ht_file_fd                  = -1;
        int ht_dir_fd                   = -1;
        int ret                         = 0;
        int cnt                         = 0;
        char ht_dir_path[PATH_MAX]      = {0,};
        char ht_file_path[PATH_MAX]     = {0,};
        char ht_file_bname[NAME_MAX]    = {0,};
        char x_value[NAME_MAX]          = {0,};
        int flags                       = 0;
        unsigned long min_ts            = 0;
        unsigned long max_ts            = 0;
        unsigned long total             = 0;
        unsigned long total1            = 0;
        ssize_t size                    = 0;
        struct stat stat_buf            = {0,};
        unsigned long record_len        = 0;

        CHANGELOG_FILL_HTIME_DIR(priv->changelog_dir, ht_dir_path);

        /* Open htime directory to get HTIME_CURRENT */
        ht_dir_fd = open (ht_dir_path, O_RDONLY);
        if (ht_dir_fd == -1) {
                gf_smsg (this->name, GF_LOG_ERROR, errno,
                         CHANGELOG_MSG_OPEN_FAILED, "open failed",
                         "path=%s", ht_dir_path,
                         NULL);
                ret = -1;
                goto out;
        }

        size = sys_fgetxattr (ht_dir_fd, HTIME_CURRENT, ht_file_bname,
                             sizeof (ht_file_bname));
        if (size < 0) {
                gf_msg (this->name, GF_LOG_ERROR, errno,
                        CHANGELOG_MSG_FGETXATTR_FAILED, "Error extracting"
                        " HTIME_CURRENT.");

                /* If upgrade scenario, find the latest HTIME.TSTAMP file
                 * and use the same. If error, create a new HTIME.TSTAMP
                 * file.
                 */
                cnt = find_current_htime (ht_dir_fd, ht_dir_path,
                                           ht_file_bname);
                if (cnt <= 0) {
                        gf_msg (this->name, GF_LOG_INFO, errno,
                                CHANGELOG_MSG_HTIME_INFO,
                                "HTIME_CURRENT not found. Changelog enabled"
                                " before init");
                        return htime_create (this, priv, ts);
                }

                gf_msg (this->name, GF_LOG_ERROR, errno,
                        CHANGELOG_MSG_HTIME_ERROR, "Error extracting"
                        " HTIME_CURRENT.");
        }

        gf_smsg (this->name, GF_LOG_INFO, 0, CHANGELOG_MSG_HTIME_INFO,
                 "HTIME_CURRENT",
                 "path=%s", ht_file_bname, NULL);
        (void) snprintf (ht_file_path, PATH_MAX, "%s/%s",
                         ht_dir_path, ht_file_bname);

        /* Open in append mode as existing htime file is used */
        flags |= (O_RDWR | O_SYNC | O_APPEND);
        ht_file_fd = open (ht_file_path, flags,
                        S_IRUSR | S_IWUSR | S_IRGRP | S_IROTH);
        if (ht_file_fd < 0) {
                gf_smsg (this->name, GF_LOG_ERROR, errno,
                         CHANGELOG_MSG_OPEN_FAILED,
                         "unable to open htime file",
                         "path=%s", ht_file_path,
                         NULL);
                ret = -1;
                goto out;
        }

        /* save this htime_fd in priv->htime_fd */
        priv->htime_fd = ht_file_fd;

        ret = sys_fstat (ht_file_fd, &stat_buf);
        if (ret < 0) {
                gf_smsg (this->name, GF_LOG_ERROR, errno,
                         CHANGELOG_MSG_HTIME_ERROR,
                         "unable to stat htime file",
                         "path=%s", ht_file_path,
                         NULL);
                ret = -1;
                goto out;
        }

        /* Initialize rollover-number in priv to current number */
        size = sys_fgetxattr (ht_file_fd, HTIME_KEY, x_value, sizeof (x_value));
        if (size < 0) {
                gf_smsg (this->name, GF_LOG_ERROR, errno,
                         CHANGELOG_MSG_FGETXATTR_FAILED, "error extracting max"
                         " timstamp from htime file",
                         "path=%s", ht_file_path,
                         NULL);
                ret = -1;
                goto out;
        }

        sscanf (x_value, "%lu:%lu", &max_ts, &total);

        /* 22 = 1(/) + 20(CHANGELOG.TIMESTAMP) + 1(\x00) */
        record_len = strlen(priv->changelog_dir) + 22;
        total1 = stat_buf.st_size/record_len;
        if (total != total1) {
                gf_smsg (this->name, GF_LOG_INFO, 0,
                         CHANGELOG_MSG_TOTAL_LOG_INFO,
                         "Mismatch of changelog count. "
                         "INIT CASE",
                         "xattr_total=%lu", total,
                         "size_total=%lu", total1,
                         NULL);
        }

        gf_smsg (this->name, GF_LOG_INFO, 0,
                 CHANGELOG_MSG_TOTAL_LOG_INFO,
                 "INIT CASE",
                 "min=%lu", min_ts,
                 "max=%lu", max_ts,
                 "total_changelogs=%lu", total,
                 NULL);

        if (total < total1)
                priv->rollover_count = total1 + 1;
        else
                priv->rollover_count = total + 1;

out:
        if (ht_dir_fd != -1)
                sys_close (ht_dir_fd);
        return ret;
}

/* Returns 0 on successful creation of htime file
 * returns -1 on failure or error
 */
int
htime_create (xlator_t *this,
              changelog_priv_t *priv, unsigned long ts)
{
        int ht_file_fd                      = -1;
        int ht_dir_fd                       = -1;
        int ret                             = 0;
        char ht_dir_path[PATH_MAX]          = {0,};
        char ht_file_path[PATH_MAX]         = {0,};
        char ht_file_bname[NAME_MAX + 1]    = {0,};
        int flags                           = 0;

        gf_smsg (this->name, GF_LOG_INFO, 0,
                 CHANGELOG_MSG_HTIME_INFO, "Changelog enable: Creating new "
                 "HTIME file",
                 "name=%lu", ts,
                 NULL);

        CHANGELOG_FILL_HTIME_DIR(priv->changelog_dir, ht_dir_path);

        /* get the htime file name in ht_file_path */
        (void) snprintf (ht_file_path,PATH_MAX,"%s/%s.%lu",ht_dir_path,
                        HTIME_FILE_NAME, ts);

        flags |= (O_CREAT | O_RDWR | O_SYNC);
        ht_file_fd = open (ht_file_path, flags,
                        S_IRUSR | S_IWUSR | S_IRGRP | S_IROTH);
        if (ht_file_fd < 0) {
                gf_smsg (this->name, GF_LOG_ERROR, errno,
                         CHANGELOG_MSG_OPEN_FAILED,
                         "unable to create htime file",
                         "path=%s", ht_file_path,
                         NULL);
                ret = -1;
                goto out;
        }

        if (sys_fsetxattr (ht_file_fd, HTIME_KEY, HTIME_INITIAL_VALUE,
                       sizeof (HTIME_INITIAL_VALUE)-1,  0)) {
                gf_msg (this->name, GF_LOG_ERROR, errno,
                        CHANGELOG_MSG_FSETXATTR_FAILED,
                        "Htime xattr initialization failed");
                ret = -1;
                goto out;
        }

        ret = sys_fsync (ht_file_fd);
        if (ret < 0) {
                gf_msg (this->name, GF_LOG_ERROR, errno,
                        CHANGELOG_MSG_FSYNC_OP_FAILED,
                        "fsync failed");
                goto out;
        }

        /* Set xattr HTIME_CURRENT on htime directory to htime filename */
        ht_dir_fd = open (ht_dir_path, O_RDONLY);
        if (ht_dir_fd == -1) {
                gf_smsg (this->name, GF_LOG_ERROR, errno,
                         CHANGELOG_MSG_OPEN_FAILED, "open failed",
                         "path=%s", ht_dir_path,
                         NULL);
                ret = -1;
                goto out;
        }

        (void) snprintf (ht_file_bname, sizeof (ht_file_bname), "%s.%lu",
                         HTIME_FILE_NAME, ts);
        if (sys_fsetxattr (ht_dir_fd, HTIME_CURRENT, ht_file_bname,
            strlen (ht_file_bname), 0)) {
                gf_msg (this->name, GF_LOG_ERROR, errno,
                        CHANGELOG_MSG_FSETXATTR_FAILED, "fsetxattr failed:"
                        " HTIME_CURRENT");
                ret = -1;
                goto out;
        }

        ret = sys_fsync (ht_dir_fd);
        if (ret < 0) {
                gf_msg (this->name, GF_LOG_ERROR, errno,
                        CHANGELOG_MSG_FSYNC_OP_FAILED,
                        "fsync failed");
                goto out;
        }

        /* save this htime_fd in priv->htime_fd */
        priv->htime_fd = ht_file_fd;
        /* initialize rollover-number in priv to 1 */
        priv->rollover_count = 1;

out:
        if (ht_dir_fd != -1)
                sys_close (ht_dir_fd);
        return ret;
}

/* Description:
 *      Opens the snap changelog to log call path fops in it.
 *      This changelos name is "CHANGELOG.SNAP", stored in
 *      path ".glusterfs/changelogs/csnap".
 * Returns:
 *       0  : On success.
 *      -1  : On failure.
 */
int
changelog_snap_open (xlator_t *this,
                         changelog_priv_t *priv)
{
        int fd                        = -1;
        int ret                       = 0;
        int flags                     = 0;
        char buffer[1024]             = {0,};
        char c_snap_path[PATH_MAX]    = {0,};
        char csnap_dir_path[PATH_MAX] = {0,};

        CHANGELOG_FILL_CSNAP_DIR(priv->changelog_dir, csnap_dir_path);

        (void) snprintf (c_snap_path, PATH_MAX,
                        "%s/"CSNAP_FILE_NAME,
                        csnap_dir_path);

        flags |= (O_CREAT | O_RDWR | O_TRUNC);

        fd = open (c_snap_path, flags,
                        S_IRUSR | S_IWUSR | S_IRGRP | S_IROTH);
        if (fd < 0) {
                gf_smsg (this->name, GF_LOG_ERROR, errno,
                         CHANGELOG_MSG_OPEN_FAILED, "unable to open file",
                         "path=%s", c_snap_path,
                         NULL);
                ret = -1;
                goto out;
        }
        priv->c_snap_fd = fd;

        (void) snprintf (buffer, 1024, CHANGELOG_HEADER,
                        CHANGELOG_VERSION_MAJOR,
                        CHANGELOG_VERSION_MINOR,
                        priv->ce->encoder);
        ret = changelog_snap_write_change (priv, buffer, strlen (buffer));
        if (ret < 0) {
                sys_close (priv->c_snap_fd);
                priv->c_snap_fd = -1;
                goto out;
        }

out:
        return ret;
}

/*
 * Description:
 *      Starts logging fop details in CSNAP journal.
 * Returns:
 *       0 : On success.
 *      -1 : On Failure.
 */
int
changelog_snap_logging_start (xlator_t *this,
                                  changelog_priv_t *priv)
{
        int ret = 0;

        ret = changelog_snap_open (this, priv);
        gf_msg (this->name, GF_LOG_INFO, 0,
                        CHANGELOG_MSG_SNAP_INFO,
                        "Now starting to log in call path");

        return ret;
}

/*
 * Description:
 *      Stops logging fop details in CSNAP journal.
 * Returns:
 *       0 : On success.
 *      -1 : On Failure.
 */
int
changelog_snap_logging_stop (xlator_t *this,
                changelog_priv_t *priv)
{
        int ret         = 0;

        sys_close (priv->c_snap_fd);
        priv->c_snap_fd = -1;

        gf_msg (this->name, GF_LOG_INFO, 0,
                CHANGELOG_MSG_SNAP_INFO,
                "Stopped to log in call path");

        return ret;
}

int
changelog_open_journal (xlator_t *this,
                        changelog_priv_t *priv)
{
        int fd                        = 0;
        int ret                       = -1;
        int flags                     = 0;
        char buffer[1024]             = {0,};
        char changelog_path[PATH_MAX] = {0,};

        (void) snprintf (changelog_path, PATH_MAX,
                         "%s/"CHANGELOG_FILE_NAME,
                         priv->changelog_dir);

        flags |= (O_CREAT | O_RDWR);
        if (priv->fsync_interval == 0)
                flags |= O_SYNC;

        fd = open (changelog_path, flags,
                   S_IRUSR | S_IWUSR | S_IRGRP | S_IROTH);
        if (fd < 0) {
                gf_smsg (this->name, GF_LOG_ERROR, errno,
                         CHANGELOG_MSG_OPEN_FAILED,
                         "unable to open/create changelog file."
                         " change-logging will be"
                         " inactive",
                         "path=%s", changelog_path,
                         NULL);
                goto out;
        }

        priv->changelog_fd = fd;

        (void) snprintf (buffer, 1024, CHANGELOG_HEADER,
                         CHANGELOG_VERSION_MAJOR,
                         CHANGELOG_VERSION_MINOR,
                         priv->ce->encoder);
        ret = changelog_write_change (priv, buffer, strlen (buffer));
        if (ret) {
                sys_close (priv->changelog_fd);
                priv->changelog_fd = -1;
                goto out;
        }

        ret = 0;

 out:
        return ret;
}

int
changelog_start_next_change (xlator_t *this,
                             changelog_priv_t *priv,
                             unsigned long ts, gf_boolean_t finale)
{
        int ret = -1;

        ret = changelog_rollover_changelog (this, priv, ts);

        if (!ret && !finale)
                ret = changelog_open_journal (this, priv);

        return ret;
}

/**
 * return the length of entry
 */
size_t
changelog_entry_length ()
{
        return sizeof (changelog_log_data_t);
}

int
changelog_fill_rollover_data (changelog_log_data_t *cld, gf_boolean_t is_last)
{
        struct timeval tv = {0,};

        cld->cld_type = CHANGELOG_TYPE_ROLLOVER;

        if (gettimeofday (&tv, NULL))
                return -1;

        cld->cld_roll_time = (unsigned long) tv.tv_sec;
        cld->cld_finale = is_last;
        return 0;
}

int
changelog_snap_write_change (changelog_priv_t *priv, char *buffer, size_t len)
{
        return changelog_write (priv->c_snap_fd, buffer, len);
}

int
changelog_write_change (changelog_priv_t *priv, char *buffer, size_t len)
{
        return changelog_write (priv->changelog_fd, buffer, len);
}

/*
 * Descriptions:
 *      Writes fop details in ascii format to CSNAP.
 * Issues:
 *      Not Encoding agnostic.
 * Returns:
 *      0 : On Success.
 *     -1 : On Failure.
 */
int
changelog_snap_handle_ascii_change (xlator_t *this,
                                    changelog_log_data_t *cld)
{
        size_t            off      = 0;
        size_t            gfid_len = 0;
        char             *gfid_str = NULL;
        char             *buffer   = NULL;
        changelog_priv_t *priv     = NULL;
        int               ret      = 0;

        if (this == NULL) {
                ret = -1;
                goto out;
        }

        priv = this->private;

        if (priv == NULL) {
                ret = -1;
                goto out;
        }

        gfid_str = uuid_utoa (cld->cld_gfid);
        gfid_len = strlen (gfid_str);

        /*  extra bytes for decorations */
        buffer = alloca (gfid_len + cld->cld_ptr_len + 10);
        CHANGELOG_STORE_ASCII (priv, buffer,
                        off, gfid_str, gfid_len, cld);

        CHANGELOG_FILL_BUFFER (buffer, off, "\0", 1);

        ret = changelog_snap_write_change (priv, buffer, off);

        if (ret < 0) {
                gf_msg (this->name, GF_LOG_ERROR, 0,
                        CHANGELOG_MSG_WRITE_FAILED,
                        "error writing csnap to disk");
        }
        gf_msg (this->name, GF_LOG_INFO, 0, CHANGELOG_MSG_SNAP_INFO,
                        "Successfully wrote to csnap");
        ret = 0;
out:
        return ret;
}

int
changelog_handle_change (xlator_t *this,
                         changelog_priv_t *priv, changelog_log_data_t *cld)
{
        int ret = 0;

        if (CHANGELOG_TYPE_IS_ROLLOVER (cld->cld_type)) {
                changelog_encode_change (priv);
                ret = changelog_start_next_change (this, priv,
                                                   cld->cld_roll_time,
                                                   cld->cld_finale);
                if (ret)
                        gf_msg (this->name, GF_LOG_ERROR, 0,
                                CHANGELOG_MSG_GET_TIME_OP_FAILED,
                                "Problem rolling over changelog(s)");
                goto out;
        }

        /**
         * case when there is reconfigure done (disabling changelog) and there
         * are still fops that have updates in prgress.
         */
        if (priv->changelog_fd == -1)
                return 0;

        if (CHANGELOG_TYPE_IS_FSYNC (cld->cld_type)) {
                ret = sys_fsync (priv->changelog_fd);
                if (ret < 0) {
                        gf_msg (this->name, GF_LOG_ERROR, errno,
                                CHANGELOG_MSG_FSYNC_OP_FAILED,
                                "fsync failed");
                }
                goto out;
        }

        ret = priv->ce->encode (this, cld);
        if (ret) {
                gf_msg (this->name, GF_LOG_ERROR, 0,
                        CHANGELOG_MSG_WRITE_FAILED,
                        "error writing changelog to disk");
        }

 out:
        return ret;
}

changelog_local_t *
changelog_local_init (xlator_t *this, inode_t *inode,
                      uuid_t gfid, int xtra_records,
                      gf_boolean_t update_flag)
{
        changelog_local_t *local = NULL;
        struct iobuf      *iobuf = NULL;

        /**
         * We relax the presence of inode if @update_flag is true.
         * The caller (implmentation of the fop) needs to be careful to
         * not blindly use local->inode.
         */
        if (!update_flag && !inode) {
                gf_msg_callingfn (this->name, GF_LOG_WARNING, 0,
                                  CHANGELOG_MSG_INODE_NOT_FOUND,
                                  "inode needed for version checking !!!");
                goto out;
        }

        if (xtra_records) {
                iobuf = iobuf_get2 (this->ctx->iobuf_pool,
                                    xtra_records * CHANGELOG_OPT_RECORD_LEN);
                if (!iobuf)
                        goto out;
        }

        local = mem_get0 (this->local_pool);
        if (!local) {
                CHANGELOG_IOBUF_UNREF (iobuf);
                goto out;
        }

        local->update_no_check = update_flag;

        gf_uuid_copy (local->cld.cld_gfid, gfid);

        local->cld.cld_iobuf = iobuf;
        local->cld.cld_xtra_records = 0; /* set by the caller */

        if (inode)
                local->inode = inode_ref (inode);

 out:
        return local;
}

int
changelog_forget (xlator_t *this, inode_t *inode)
{
        uint64_t ctx_addr = 0;
        changelog_inode_ctx_t *ctx = NULL;

        inode_ctx_del (inode, this, &ctx_addr);
        if (!ctx_addr)
                return 0;

        ctx = (changelog_inode_ctx_t *) (long) ctx_addr;
        GF_FREE (ctx);

        return 0;
}

int
changelog_inject_single_event (xlator_t *this,
                               changelog_priv_t *priv,
                               changelog_log_data_t *cld)
{
        return priv->cd.dispatchfn (this, priv, priv->cd.cd_data, cld, NULL);
}

/* Wait till all the black fops are drained */
void
changelog_drain_black_fops (xlator_t *this, changelog_priv_t *priv)
{
        int ret = 0;

        /* clean up framework of pthread_mutex is required here as
         * 'reconfigure' terminates the changelog_rollover thread
         * on graph change.
         */
        pthread_cleanup_push (changelog_cleanup_free_mutex,
                                        &priv->dm.drain_black_mutex);
        ret = pthread_mutex_lock (&priv->dm.drain_black_mutex);
        if (ret)
                gf_smsg (this->name, GF_LOG_ERROR, errno,
                         CHANGELOG_MSG_PTHREAD_ERROR, "pthread error",
                         "error=%d", ret,
                         NULL);
        while (priv->dm.black_fop_cnt > 0) {
                gf_msg_debug (this->name, 0,
                              "Condtional wait on black fops: %ld",
                              priv->dm.black_fop_cnt);
                priv->dm.drain_wait_black = _gf_true;
                ret = pthread_cond_wait (&priv->dm.drain_black_cond,
                                         &priv->dm.drain_black_mutex);
                if (ret)
                        gf_smsg (this->name, GF_LOG_ERROR, errno,
                                 CHANGELOG_MSG_PTHREAD_COND_WAIT_FAILED,
                                 "pthread cond wait failed",
                                 "error=%d", ret,
                                 NULL);
        }
        priv->dm.drain_wait_black = _gf_false;
        ret = pthread_mutex_unlock (&priv->dm.drain_black_mutex);
        if (ret)
                gf_smsg (this->name, GF_LOG_ERROR, errno,
                         CHANGELOG_MSG_PTHREAD_ERROR, "pthread error",
                         "error=%d", ret,
                         NULL);
        pthread_cleanup_pop (0);
        gf_msg_debug (this->name, 0,
                      "Woke up: Conditional wait on black fops");
}

/* Wait till all the white  fops are drained */
void
changelog_drain_white_fops (xlator_t *this, changelog_priv_t *priv)
{
        int ret = 0;

        /* clean up framework of pthread_mutex is required here as
         * 'reconfigure' terminates the changelog_rollover thread
         * on graph change.
         */
        pthread_cleanup_push (changelog_cleanup_free_mutex,
                                        &priv->dm.drain_white_mutex);
        ret = pthread_mutex_lock (&priv->dm.drain_white_mutex);
        if (ret)
                gf_smsg (this->name, GF_LOG_ERROR, errno,
                         CHANGELOG_MSG_PTHREAD_ERROR, "pthread error",
                         "error=%d", ret,
                         NULL);
        while (priv->dm.white_fop_cnt > 0) {
                gf_msg_debug (this->name, 0,
                              "Condtional wait on white fops : %ld",
                        priv->dm.white_fop_cnt);
                priv->dm.drain_wait_white = _gf_true;
                ret = pthread_cond_wait (&priv->dm.drain_white_cond,
                                         &priv->dm.drain_white_mutex);
                if (ret)
                        gf_smsg (this->name, GF_LOG_ERROR, errno,
                                 CHANGELOG_MSG_PTHREAD_COND_WAIT_FAILED,
                                 "pthread cond wait failed",
                                 "error=%d", ret,
                                 NULL);
        }
        priv->dm.drain_wait_white = _gf_false;
        ret = pthread_mutex_unlock (&priv->dm.drain_white_mutex);
        if (ret)
                gf_smsg (this->name, GF_LOG_ERROR, errno,
                         CHANGELOG_MSG_PTHREAD_ERROR, "pthread error",
                         "error=%d", ret,
                         NULL);
        pthread_cleanup_pop (0);
        gf_msg_debug (this->name, 0,
                      "Woke up: Conditional wait on white fops");
}

/**
 * TODO: these threads have many thing in common (wake up after
 * a certain time etc..). move them into separate routine.
 */
void *
changelog_rollover (void *data)
{
        int                     ret             = 0;
        xlator_t               *this            = NULL;
        struct timespec         tv              = {0,};
        changelog_log_data_t    cld             = {0,};
        changelog_time_slice_t *slice           = NULL;
        changelog_priv_t       *priv            = data;

        this = priv->cr.this;
        slice = &priv->slice;

        while (1) {
                (void) pthread_testcancel();

                tv.tv_sec  = time (NULL) + priv->rollover_time;
                tv.tv_nsec = 0;
                ret = 0; /* Reset ret to zero */

               /* The race between actual rollover and explicit rollover is
                * handled. If actual rollover is being done and the
                * explicit rollover event comes, the event is not missed.
                * Since explicit rollover sets 'cr.notify' to true, this
                * thread doesn't wait on 'pthread_cond_timedwait'.
                */
                pthread_cleanup_push (changelog_cleanup_free_mutex,
                                      &priv->cr.lock);
                pthread_mutex_lock (&priv->cr.lock);
                {
                        while (ret == 0 && !priv->cr.notify)
                                ret = pthread_cond_timedwait (&priv->cr.cond,
                                                              &priv->cr.lock,
                                                              &tv);
                        if (ret == 0)
                                priv->cr.notify = _gf_false;
                }
                pthread_mutex_unlock (&priv->cr.lock);
                pthread_cleanup_pop (0);

                if (ret == 0) {
                        gf_msg (this->name, GF_LOG_INFO, 0,
                                CHANGELOG_MSG_BARRIER_INFO,
                                "Explicit wakeup on barrier notify");
                        priv->explicit_rollover = _gf_true;
                } else if (ret && ret != ETIMEDOUT) {
                        gf_msg (this->name, GF_LOG_ERROR, errno,
                                CHANGELOG_MSG_SELECT_FAILED,
                                "pthread_cond_timedwait failed");
                        continue;
                } else if (ret && ret == ETIMEDOUT) {
                        gf_msg_debug (this->name, 0, "Wokeup on timeout");
                }

               /* Reading curent_color without lock is fine here
                * as it is only modified here and is next to reading.
                */
                if (priv->current_color == FOP_COLOR_BLACK) {
                        LOCK(&priv->lock);
                                priv->current_color = FOP_COLOR_WHITE;
                        UNLOCK(&priv->lock);
                        gf_msg_debug (this->name, 0, "Black fops"
                                      " to be drained:%ld",
                                      priv->dm.black_fop_cnt);
                        changelog_drain_black_fops (this, priv);
                } else {
                        LOCK(&priv->lock);
                                priv->current_color = FOP_COLOR_BLACK;
                        UNLOCK(&priv->lock);
                        gf_msg_debug (this->name, 0, "White fops"
                                      " to be drained:%ld",
                                      priv->dm.white_fop_cnt);
                        changelog_drain_white_fops (this, priv);
                }

                /* Adding delay of 1 second only during explicit rollover:
                 *
                 * Changelog rollover can happen either due to actual
                 * or the explict rollover during snapshot. Actual
                 * rollover is controlled by tuneable called 'rollover-time'.
                 * The minimum granularity for rollover-time is 1 second.
                 * Explicit rollover is asynchronous in nature and happens
                 * during snapshot.
                 *
                 * Basically, rollover renames the current CHANGELOG file
                 * to CHANGELOG.TIMESTAMP. Let's assume, at time 't1',
                 * actual and explicit rollover raced against  each
                 * other and actual rollover won the race renaming the
                 * CHANGELOG file to CHANGELOG.t1 and opens a new
                 * CHANGELOG file. There is high chance that, an immediate
                 * explicit rollover at time 't1' can happen with in the same
                 * second to rename CHANGELOG file to CHANGELOG.t1 resulting in
                 * purging the earlier CHANGELOG.t1 file created by actual
                 * rollover. So adding a delay of 1 second guarantees unique
                 * CHANGELOG.TIMESTAMP during  explicit rollover.
                 */
                if (priv->explicit_rollover == _gf_true)
                        sleep (1);

                ret = changelog_fill_rollover_data (&cld, _gf_false);
                if (ret) {
                        gf_msg (this->name, GF_LOG_ERROR, 0,
                                CHANGELOG_MSG_GET_TIME_OP_FAILED,
                                "failed to fill rollover data");
                        continue;
                }

                _mask_cancellation ();

                LOCK (&priv->lock);
                {
                        ret = changelog_inject_single_event (this, priv, &cld);
                        if (!ret)
                                SLICE_VERSION_UPDATE (slice);
                }
                UNLOCK (&priv->lock);

                _unmask_cancellation ();
        }

        return NULL;
}

void *
changelog_fsync_thread (void *data)
{
        int                   ret  = 0;
        xlator_t             *this = NULL;
        struct timeval        tv   = {0,};
        changelog_log_data_t  cld  = {0,};
        changelog_priv_t     *priv = data;

        this = priv->cf.this;
        cld.cld_type = CHANGELOG_TYPE_FSYNC;

        while (1) {
                (void) pthread_testcancel();

                tv.tv_sec  = priv->fsync_interval;
                tv.tv_usec = 0;

                ret = select (0, NULL, NULL, NULL, &tv);
                if (ret)
                        continue;

                _mask_cancellation ();

                ret = changelog_inject_single_event (this, priv, &cld);
                if (ret)
                        gf_msg (this->name, GF_LOG_ERROR, 0,
                                CHANGELOG_MSG_INJECT_FSYNC_FAILED,
                                "failed to inject fsync event");

                _unmask_cancellation ();
        }

        return NULL;
}

/* macros for inode/changelog version checks */

#define INODE_VERSION_UPDATE(priv, inode, iver, slice, type) do {       \
                LOCK (&inode->lock);                                    \
                {                                                       \
                        LOCK (&priv->lock);                             \
                        {                                               \
                                *iver = slice->changelog_version[type]; \
                        }                                               \
                        UNLOCK (&priv->lock);                           \
                }                                                       \
                UNLOCK (&inode->lock);                                  \
        } while (0)

#define INODE_VERSION_EQUALS_SLICE(priv, ver, slice, type, upd) do {    \
                LOCK (&priv->lock);                                     \
                {                                                       \
                        upd = (ver == slice->changelog_version[type])   \
                                ? _gf_false : _gf_true;                 \
                }                                                       \
                UNLOCK (&priv->lock);                                   \
        } while (0)

static int
__changelog_inode_ctx_set (xlator_t *this,
                           inode_t *inode, changelog_inode_ctx_t *ctx)
{
        uint64_t ctx_addr = (uint64_t) ctx;
        return __inode_ctx_set (inode, this, &ctx_addr);
}

/**
 * one shot routine to get the address and the value of a inode version
 * for a particular type.
 */
changelog_inode_ctx_t *
__changelog_inode_ctx_get (xlator_t *this,
                           inode_t *inode, unsigned long **iver,
                           unsigned long *version, changelog_log_type type)
{
        int                    ret      = 0;
        uint64_t               ctx_addr = 0;
        changelog_inode_ctx_t *ctx      = NULL;

        ret = __inode_ctx_get (inode, this, &ctx_addr);
        if (ret < 0)
                ctx_addr = 0;
        if (ctx_addr != 0) {
                ctx = (changelog_inode_ctx_t *) (long)ctx_addr;
                goto out;
        }

        ctx = GF_CALLOC (1, sizeof (*ctx), gf_changelog_mt_inode_ctx_t);
        if (!ctx)
                goto out;

        ret = __changelog_inode_ctx_set (this, inode, ctx);
        if (ret) {
                GF_FREE (ctx);
                ctx = NULL;
        }

 out:
        if (ctx && iver && version) {
                *iver = CHANGELOG_INODE_VERSION_TYPE (ctx, type);
                *version = **iver;
        }

        return ctx;
}

static changelog_inode_ctx_t *
changelog_inode_ctx_get (xlator_t *this,
                         inode_t *inode, unsigned long **iver,
                         unsigned long *version, changelog_log_type type)
{
        changelog_inode_ctx_t *ctx = NULL;

        LOCK (&inode->lock);
        {
                ctx = __changelog_inode_ctx_get (this,
                                                 inode, iver, version, type);
        }
        UNLOCK (&inode->lock);

        return ctx;
}

/**
 * This is the main update routine. Locking has been made granular so as to
 * maximize parallelism of fops - I'll try to explain it below using execution
 * timelines.
 *
 * Basically, the contention is between multiple execution threads of this
 * routine and the roll-over thread. So, instead of having a big lock, we hold
 * granular locks: inode->lock and priv->lock. Now I'll explain what happens
 * when there is an update and a roll-over at just about the same time.
 * NOTE:
 *  - the dispatcher itself synchronizes updates via it's own lock
 *  - the slice version in incremented by the roll-over thread
 *
 * Case 1: When the rollover thread wins before the inode version can be
 * compared with the slice version.
 *
 *          [updater]                 |             [rollover]
 *                                    |
 *                                    |           <SLICE: 1, 1, 1>
 * <changelog_update>                 |
 *   <changelog_inode_ctx_get>        |
 *      <CTX: 1, 1, 1>                |
 *                                    |         <dispatch-rollover-event>
 *                                    |         LOCK (&priv->lock)
 *                                    |            <SLICE_VERSION_UPDATE>
 *                                    |              <SLICE: 2, 2, 2>
 *                                    |         UNLOCK (&priv->lock)
 *                                    |
 * LOCK (&priv->lock)                 |
 *   <INODE_VERSION_EQUALS_SLICE>     |
 *    I: 1 <-> S: 2                   |
 *    update: true                    |
 * UNLOCK (&priv->lock)               |
 *                                    |
 * <if update == true>                |
 *  <dispath-update-event>            |
 *  <INODE_VERSION_UPDATE>            |
 *   LOCK (&inode->lock)              |
 *    LOCK (&priv->lock)              |
 *     <CTX: 2, 1, 1>                 |
 *    UNLOCK (&priv->lock)            |
 *   UNLOCK (&inode->lock)            |
 *
 * Therefore, the change gets recorded in the next change (no lost change). If
 * the slice version was ahead of the inode version (say I:1, S: 2), then
 * anyway the comparison would result in a update (I: 1, S: 3).
 *
 * If the rollover time is too less, then there is another contention when the
 * updater tries to bring up inode version to the slice version (this is also
 * the case when the roll-over thread wakes up during INODE_VERSION_UPDATE.
 *
 *   <CTX: 1, 1, 1>                   |       <SLICE: 2, 2, 2>
 *                                    |
 *                                    |
 * <dispath-update-event>             |
 * <INODE_VERSION_UPDATE>             |
 *  LOCK (&inode->lock)               |
 *   LOCK (&priv->lock)               |
 *    <CTX: 2, 1, 1>                  |
 *   UNLOCK (&priv->lock)             |
 *  UNLOCK (&inode->lock)             |
 *                                    |         <dispatch-rollover-event>
 *                                    |         LOCK (&priv->lock)
 *                                    |            <SLICE_VERSION_UPDATE>
 *                                    |              <SLICE: 3, 3, 3>
 *                                    |         UNLOCK (&priv->lock)
 *
 *
 * Case 2: When the fop thread wins
 *
 *          [updater]                 |             [rollover]
 *                                    |
 *                                    |           <SLICE: 1, 1, 1>
 * <changelog_update>                 |
 *   <changelog_inode_ctx_get>        |
 *      <CTX: 0, 0, 0>                |
 *                                    |
 * LOCK (&priv->lock)                 |
 *   <INODE_VERSION_EQUALS_SLICE>     |
 *    I: 0 <-> S: 1                   |
 *    update: true                    |
 * UNLOCK (&priv->lock)               |
 *                                    |         <dispatch-rollover-event>
 *                                    |         LOCK (&priv->lock)
 *                                    |            <SLICE_VERSION_UPDATE>
 *                                    |              <SLICE: 2, 2, 2>
 *                                    |         UNLOCK (&priv->lock)
 * <if update == true>                |
 *  <dispath-update-event>            |
 *  <INODE_VERSION_UPDATE>            |
 *   LOCK (&inode->lock)              |
 *    LOCK (&priv->lock)              |
 *     <CTX: 2, 0, 0>                 |
 *    UNLOCK (&priv->lock)            |
 *   UNLOCK (&inode->lock)            |
 *
 * Here again, if the inode version was equal to the slice version (I: 1, S: 1)
 * then there is no need to record an update (as the equality of the two version
 * signifies an update was recorded in the current time slice).
 */
void
changelog_update (xlator_t *this, changelog_priv_t *priv,
                  changelog_local_t *local, changelog_log_type type)
{
        int                     ret        = 0;
        unsigned long          *iver       = NULL;
        unsigned long           version    = 0;
        inode_t                *inode      = NULL;
        changelog_time_slice_t *slice      = NULL;
        changelog_inode_ctx_t  *ctx        = NULL;
        changelog_log_data_t   *cld_0      = NULL;
        changelog_log_data_t   *cld_1      = NULL;
        changelog_local_t      *next_local = NULL;
        gf_boolean_t            need_upd   = _gf_true;

        slice = &priv->slice;

        /**
         * for fops that do not require inode version checking
         */
        if (local->update_no_check)
                goto update;

        inode = local->inode;

        ctx = changelog_inode_ctx_get (this,
                                       inode, &iver, &version, type);
        if (!ctx)
                goto update;

        INODE_VERSION_EQUALS_SLICE (priv, version, slice, type, need_upd);

 update:
        if (need_upd) {
                cld_0 = &local->cld;
                cld_0->cld_type = type;

                if ( (next_local = local->prev_entry) != NULL ) {
                        cld_1 = &next_local->cld;
                        cld_1->cld_type = type;
                }

                ret = priv->cd.dispatchfn (this, priv,
                                           priv->cd.cd_data, cld_0, cld_1);

                /**
                 * update after the dispatcher has successfully done
                 * it's job.
                 */
                if (!local->update_no_check && iver && !ret)
                        INODE_VERSION_UPDATE (priv, inode, iver, slice, type);
        }

        return;
}

/* Begin: Geo-rep snapshot dependency changes */

/* changelog_color_fop_and_inc_cnt: Assign color and inc fop cnt.
 *
 * Assigning color and increment of corresponding fop count should happen
 * in a lock (i.e., there should be no window between them). If it does not,
 * we might miss draining those fops which are colored but not yet incremented
 * the count. Let's assume black fops are draining. If the black fop count
 * reaches zero, we say draining is completed but we miss black fops which are
 * not incremented fop count but color is assigned black.
 */

void
changelog_color_fop_and_inc_cnt (xlator_t *this, changelog_priv_t *priv,
                                                 changelog_local_t *local)
{
        if (!priv || !local)
                return;

        LOCK (&priv->lock);
        {
                local->color = priv->current_color;
                changelog_inc_fop_cnt (this, priv, local);
        }
        UNLOCK (&priv->lock);
}

/* Increments the respective fop counter based on the fop color */
void
changelog_inc_fop_cnt (xlator_t *this, changelog_priv_t *priv,
                                       changelog_local_t *local)
{
        int ret = 0;

        if (local) {
                if (local->color == FOP_COLOR_BLACK) {
                        ret = pthread_mutex_lock (&priv->dm.drain_black_mutex);
                        CHANGELOG_PTHREAD_ERROR_HANDLE_0 (ret, out);
                        {
                                priv->dm.black_fop_cnt++;
                        }
                        ret = pthread_mutex_unlock(&priv->dm.drain_black_mutex);
                        CHANGELOG_PTHREAD_ERROR_HANDLE_0 (ret, out);
                } else {
                        ret = pthread_mutex_lock (&priv->dm.drain_white_mutex);
                        CHANGELOG_PTHREAD_ERROR_HANDLE_0 (ret, out);
                        {
                                priv->dm.white_fop_cnt++;
                        }
                        ret = pthread_mutex_unlock(&priv->dm.drain_white_mutex);
                        CHANGELOG_PTHREAD_ERROR_HANDLE_0 (ret, out);
                }
        }
 out:
        return;
}

/* Decrements the respective fop counter based on the fop color */
void
changelog_dec_fop_cnt (xlator_t *this, changelog_priv_t *priv,
                                       changelog_local_t *local)
{
        int ret = 0;

        if (local) {
                if (local->color == FOP_COLOR_BLACK) {
                        ret = pthread_mutex_lock (&priv->dm.drain_black_mutex);
                        CHANGELOG_PTHREAD_ERROR_HANDLE_0 (ret, out);
                        {
                                priv->dm.black_fop_cnt--;
                                if (priv->dm.black_fop_cnt == 0 &&
                                    priv->dm.drain_wait_black == _gf_true) {
                                        ret = pthread_cond_signal (
                                                    &priv->dm.drain_black_cond);
                                        CHANGELOG_PTHREAD_ERROR_HANDLE_0 (ret,
                                                                           out);
                                        gf_msg_debug (this->name, 0,
                                                      "Signalled "
                                                      "draining of black");
                                }
                        }
                        ret = pthread_mutex_unlock(&priv->dm.drain_black_mutex);
                        CHANGELOG_PTHREAD_ERROR_HANDLE_0 (ret, out);
                } else {
                        ret = pthread_mutex_lock (&priv->dm.drain_white_mutex);
                        CHANGELOG_PTHREAD_ERROR_HANDLE_0 (ret, out);
                        {
                                priv->dm.white_fop_cnt--;
                                if (priv->dm.white_fop_cnt == 0 &&
                                    priv->dm.drain_wait_white == _gf_true) {
                                        ret = pthread_cond_signal (
                                                    &priv->dm.drain_white_cond);
                                        CHANGELOG_PTHREAD_ERROR_HANDLE_0 (ret,
                                                                           out);
                                        gf_msg_debug (this->name, 0,
                                                      "Signalled "
                                                      "draining of white");
                                }
                        }
                        ret = pthread_mutex_unlock(&priv->dm.drain_white_mutex);
                        CHANGELOG_PTHREAD_ERROR_HANDLE_0 (ret, out);
                }
        }
 out:
        return;
}

/* Write to a pipe setup between changelog main thread and changelog
 * rollover thread to initiate explicit rollover of changelog journal.
 */
int
changelog_barrier_notify (changelog_priv_t *priv, char *buf)
{
        int ret = 0;

        pthread_mutex_lock (&priv->cr.lock);
        {
                ret = pthread_cond_signal (&priv->cr.cond);
                priv->cr.notify = _gf_true;
        }
        pthread_mutex_unlock (&priv->cr.lock);
        return ret;
}

/* Clean up flags set on barrier notification */
void
changelog_barrier_cleanup (xlator_t *this, changelog_priv_t *priv,
                                                struct list_head *queue)
{
        int ret = 0;

        LOCK (&priv->bflags.lock);
                priv->bflags.barrier_ext = _gf_false;
        UNLOCK (&priv->bflags.lock);

        ret = pthread_mutex_lock (&priv->bn.bnotify_mutex);
        CHANGELOG_PTHREAD_ERROR_HANDLE_0 (ret, out);
        {
                priv->bn.bnotify = _gf_false;
        }
        ret = pthread_mutex_unlock (&priv->bn.bnotify_mutex);
        CHANGELOG_PTHREAD_ERROR_HANDLE_0 (ret, out);

        /* Disable changelog barrier and dequeue fops */
        LOCK (&priv->lock);
        {
                if (priv->barrier_enabled == _gf_true)
                        __chlog_barrier_disable (this, queue);
                else
                        ret = -1;
        }
        UNLOCK (&priv->lock);
        if (ret == 0)
                chlog_barrier_dequeue_all(this, queue);

 out:
        return;
}
/* End: Geo-Rep snapshot dependency changes */

int32_t
changelog_fill_entry_buf (call_frame_t *frame, xlator_t *this,
                          loc_t *loc, changelog_local_t **local)
{
        changelog_opt_t  *co       = NULL;
        size_t            xtra_len = 0;
        char             *dup_path = NULL;
        char             *bname    = NULL;
        inode_t          *parent   = NULL;

        GF_ASSERT (this);

        parent = inode_parent (loc->inode, 0, 0);
        if (!parent) {
                gf_smsg (this->name, GF_LOG_ERROR, errno,
                         CHANGELOG_MSG_INODE_NOT_FOUND,
                         "Parent inode not found",
                         "gfid=%s", uuid_utoa (loc->inode->gfid),
                         NULL);
                goto err;
        }

        CHANGELOG_INIT_NOCHECK (this, *local, loc->inode, loc->inode->gfid, 5);
        if (!(*local)) {
                gf_msg (this->name, GF_LOG_ERROR, 0,
                        CHANGELOG_MSG_LOCAL_INIT_FAILED, "changelog local"
                        " initiatilization failed");
                goto err;
        }

        co = changelog_get_usable_buffer (*local);
        if (!co) {
                gf_msg (this->name, GF_LOG_ERROR, 0,


context:
space:
mode: