^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1) // SPDX-License-Identifier: GPL-2.0
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3) * Copyright (c) 2000-2006 Silicon Graphics, Inc.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4) * All Rights Reserved.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 6) #include "xfs.h"
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 7) #include "xfs_fs.h"
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 8) #include "xfs_shared.h"
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 9) #include "xfs_format.h"
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 10) #include "xfs_log_format.h"
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 11) #include "xfs_trans_resv.h"
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 12) #include "xfs_bit.h"
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 13) #include "xfs_sb.h"
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 14) #include "xfs_mount.h"
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 15) #include "xfs_defer.h"
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 16) #include "xfs_inode.h"
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 17) #include "xfs_trans.h"
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 18) #include "xfs_log.h"
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 19) #include "xfs_log_priv.h"
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 20) #include "xfs_log_recover.h"
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 21) #include "xfs_trans_priv.h"
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 22) #include "xfs_alloc.h"
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 23) #include "xfs_ialloc.h"
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 24) #include "xfs_trace.h"
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 25) #include "xfs_icache.h"
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 26) #include "xfs_error.h"
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 27) #include "xfs_buf_item.h"
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 28)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 29) #define BLK_AVG(blk1, blk2) ((blk1+blk2) >> 1)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 30)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 31) STATIC int
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 32) xlog_find_zeroed(
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 33) struct xlog *,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 34) xfs_daddr_t *);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 35) STATIC int
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 36) xlog_clear_stale_blocks(
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 37) struct xlog *,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 38) xfs_lsn_t);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 39) #if defined(DEBUG)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 40) STATIC void
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 41) xlog_recover_check_summary(
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 42) struct xlog *);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 43) #else
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 44) #define xlog_recover_check_summary(log)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 45) #endif
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 46) STATIC int
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 47) xlog_do_recovery_pass(
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 48) struct xlog *, xfs_daddr_t, xfs_daddr_t, int, xfs_daddr_t *);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 49)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 50) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 51) * Sector aligned buffer routines for buffer create/read/write/access
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 52) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 53)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 54) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 55) * Verify the log-relative block number and length in basic blocks are valid for
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 56) * an operation involving the given XFS log buffer. Returns true if the fields
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 57) * are valid, false otherwise.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 58) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 59) static inline bool
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 60) xlog_verify_bno(
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 61) struct xlog *log,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 62) xfs_daddr_t blk_no,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 63) int bbcount)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 64) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 65) if (blk_no < 0 || blk_no >= log->l_logBBsize)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 66) return false;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 67) if (bbcount <= 0 || (blk_no + bbcount) > log->l_logBBsize)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 68) return false;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 69) return true;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 70) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 71)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 72) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 73) * Allocate a buffer to hold log data. The buffer needs to be able to map to
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 74) * a range of nbblks basic blocks at any valid offset within the log.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 75) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 76) static char *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 77) xlog_alloc_buffer(
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 78) struct xlog *log,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 79) int nbblks)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 80) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 81) int align_mask = xfs_buftarg_dma_alignment(log->l_targ);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 82)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 83) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 84) * Pass log block 0 since we don't have an addr yet, buffer will be
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 85) * verified on read.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 86) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 87) if (XFS_IS_CORRUPT(log->l_mp, !xlog_verify_bno(log, 0, nbblks))) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 88) xfs_warn(log->l_mp, "Invalid block length (0x%x) for buffer",
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 89) nbblks);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 90) return NULL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 91) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 92)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 93) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 94) * We do log I/O in units of log sectors (a power-of-2 multiple of the
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 95) * basic block size), so we round up the requested size to accommodate
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 96) * the basic blocks required for complete log sectors.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 97) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 98) * In addition, the buffer may be used for a non-sector-aligned block
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 99) * offset, in which case an I/O of the requested size could extend
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 100) * beyond the end of the buffer. If the requested size is only 1 basic
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 101) * block it will never straddle a sector boundary, so this won't be an
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 102) * issue. Nor will this be a problem if the log I/O is done in basic
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 103) * blocks (sector size 1). But otherwise we extend the buffer by one
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 104) * extra log sector to ensure there's space to accommodate this
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 105) * possibility.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 106) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 107) if (nbblks > 1 && log->l_sectBBsize > 1)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 108) nbblks += log->l_sectBBsize;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 109) nbblks = round_up(nbblks, log->l_sectBBsize);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 110) return kmem_alloc_io(BBTOB(nbblks), align_mask, KM_MAYFAIL | KM_ZERO);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 111) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 112)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 113) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 114) * Return the address of the start of the given block number's data
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 115) * in a log buffer. The buffer covers a log sector-aligned region.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 116) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 117) static inline unsigned int
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 118) xlog_align(
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 119) struct xlog *log,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 120) xfs_daddr_t blk_no)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 121) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 122) return BBTOB(blk_no & ((xfs_daddr_t)log->l_sectBBsize - 1));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 123) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 124)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 125) static int
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 126) xlog_do_io(
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 127) struct xlog *log,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 128) xfs_daddr_t blk_no,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 129) unsigned int nbblks,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 130) char *data,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 131) unsigned int op)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 132) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 133) int error;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 134)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 135) if (XFS_IS_CORRUPT(log->l_mp, !xlog_verify_bno(log, blk_no, nbblks))) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 136) xfs_warn(log->l_mp,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 137) "Invalid log block/length (0x%llx, 0x%x) for buffer",
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 138) blk_no, nbblks);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 139) return -EFSCORRUPTED;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 140) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 141)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 142) blk_no = round_down(blk_no, log->l_sectBBsize);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 143) nbblks = round_up(nbblks, log->l_sectBBsize);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 144) ASSERT(nbblks > 0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 145)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 146) error = xfs_rw_bdev(log->l_targ->bt_bdev, log->l_logBBstart + blk_no,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 147) BBTOB(nbblks), data, op);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 148) if (error && !XFS_FORCED_SHUTDOWN(log->l_mp)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 149) xfs_alert(log->l_mp,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 150) "log recovery %s I/O error at daddr 0x%llx len %d error %d",
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 151) op == REQ_OP_WRITE ? "write" : "read",
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 152) blk_no, nbblks, error);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 153) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 154) return error;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 155) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 156)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 157) STATIC int
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 158) xlog_bread_noalign(
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 159) struct xlog *log,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 160) xfs_daddr_t blk_no,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 161) int nbblks,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 162) char *data)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 163) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 164) return xlog_do_io(log, blk_no, nbblks, data, REQ_OP_READ);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 165) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 166)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 167) STATIC int
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 168) xlog_bread(
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 169) struct xlog *log,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 170) xfs_daddr_t blk_no,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 171) int nbblks,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 172) char *data,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 173) char **offset)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 174) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 175) int error;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 176)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 177) error = xlog_do_io(log, blk_no, nbblks, data, REQ_OP_READ);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 178) if (!error)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 179) *offset = data + xlog_align(log, blk_no);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 180) return error;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 181) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 182)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 183) STATIC int
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 184) xlog_bwrite(
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 185) struct xlog *log,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 186) xfs_daddr_t blk_no,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 187) int nbblks,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 188) char *data)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 189) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 190) return xlog_do_io(log, blk_no, nbblks, data, REQ_OP_WRITE);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 191) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 192)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 193) #ifdef DEBUG
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 194) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 195) * dump debug superblock and log record information
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 196) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 197) STATIC void
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 198) xlog_header_check_dump(
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 199) xfs_mount_t *mp,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 200) xlog_rec_header_t *head)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 201) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 202) xfs_debug(mp, "%s: SB : uuid = %pU, fmt = %d",
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 203) __func__, &mp->m_sb.sb_uuid, XLOG_FMT);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 204) xfs_debug(mp, " log : uuid = %pU, fmt = %d",
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 205) &head->h_fs_uuid, be32_to_cpu(head->h_fmt));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 206) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 207) #else
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 208) #define xlog_header_check_dump(mp, head)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 209) #endif
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 210)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 211) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 212) * check log record header for recovery
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 213) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 214) STATIC int
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 215) xlog_header_check_recover(
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 216) xfs_mount_t *mp,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 217) xlog_rec_header_t *head)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 218) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 219) ASSERT(head->h_magicno == cpu_to_be32(XLOG_HEADER_MAGIC_NUM));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 220)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 221) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 222) * IRIX doesn't write the h_fmt field and leaves it zeroed
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 223) * (XLOG_FMT_UNKNOWN). This stops us from trying to recover
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 224) * a dirty log created in IRIX.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 225) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 226) if (XFS_IS_CORRUPT(mp, head->h_fmt != cpu_to_be32(XLOG_FMT))) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 227) xfs_warn(mp,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 228) "dirty log written in incompatible format - can't recover");
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 229) xlog_header_check_dump(mp, head);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 230) return -EFSCORRUPTED;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 231) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 232) if (XFS_IS_CORRUPT(mp, !uuid_equal(&mp->m_sb.sb_uuid,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 233) &head->h_fs_uuid))) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 234) xfs_warn(mp,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 235) "dirty log entry has mismatched uuid - can't recover");
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 236) xlog_header_check_dump(mp, head);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 237) return -EFSCORRUPTED;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 238) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 239) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 240) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 241)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 242) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 243) * read the head block of the log and check the header
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 244) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 245) STATIC int
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 246) xlog_header_check_mount(
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 247) xfs_mount_t *mp,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 248) xlog_rec_header_t *head)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 249) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 250) ASSERT(head->h_magicno == cpu_to_be32(XLOG_HEADER_MAGIC_NUM));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 251)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 252) if (uuid_is_null(&head->h_fs_uuid)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 253) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 254) * IRIX doesn't write the h_fs_uuid or h_fmt fields. If
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 255) * h_fs_uuid is null, we assume this log was last mounted
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 256) * by IRIX and continue.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 257) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 258) xfs_warn(mp, "null uuid in log - IRIX style log");
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 259) } else if (XFS_IS_CORRUPT(mp, !uuid_equal(&mp->m_sb.sb_uuid,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 260) &head->h_fs_uuid))) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 261) xfs_warn(mp, "log has mismatched uuid - can't recover");
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 262) xlog_header_check_dump(mp, head);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 263) return -EFSCORRUPTED;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 264) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 265) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 266) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 267)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 268) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 269) * This routine finds (to an approximation) the first block in the physical
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 270) * log which contains the given cycle. It uses a binary search algorithm.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 271) * Note that the algorithm can not be perfect because the disk will not
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 272) * necessarily be perfect.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 273) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 274) STATIC int
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 275) xlog_find_cycle_start(
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 276) struct xlog *log,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 277) char *buffer,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 278) xfs_daddr_t first_blk,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 279) xfs_daddr_t *last_blk,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 280) uint cycle)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 281) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 282) char *offset;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 283) xfs_daddr_t mid_blk;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 284) xfs_daddr_t end_blk;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 285) uint mid_cycle;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 286) int error;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 287)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 288) end_blk = *last_blk;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 289) mid_blk = BLK_AVG(first_blk, end_blk);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 290) while (mid_blk != first_blk && mid_blk != end_blk) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 291) error = xlog_bread(log, mid_blk, 1, buffer, &offset);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 292) if (error)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 293) return error;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 294) mid_cycle = xlog_get_cycle(offset);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 295) if (mid_cycle == cycle)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 296) end_blk = mid_blk; /* last_half_cycle == mid_cycle */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 297) else
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 298) first_blk = mid_blk; /* first_half_cycle == mid_cycle */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 299) mid_blk = BLK_AVG(first_blk, end_blk);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 300) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 301) ASSERT((mid_blk == first_blk && mid_blk+1 == end_blk) ||
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 302) (mid_blk == end_blk && mid_blk-1 == first_blk));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 303)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 304) *last_blk = end_blk;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 305)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 306) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 307) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 308)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 309) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 310) * Check that a range of blocks does not contain stop_on_cycle_no.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 311) * Fill in *new_blk with the block offset where such a block is
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 312) * found, or with -1 (an invalid block number) if there is no such
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 313) * block in the range. The scan needs to occur from front to back
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 314) * and the pointer into the region must be updated since a later
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 315) * routine will need to perform another test.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 316) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 317) STATIC int
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 318) xlog_find_verify_cycle(
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 319) struct xlog *log,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 320) xfs_daddr_t start_blk,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 321) int nbblks,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 322) uint stop_on_cycle_no,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 323) xfs_daddr_t *new_blk)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 324) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 325) xfs_daddr_t i, j;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 326) uint cycle;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 327) char *buffer;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 328) xfs_daddr_t bufblks;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 329) char *buf = NULL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 330) int error = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 331)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 332) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 333) * Greedily allocate a buffer big enough to handle the full
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 334) * range of basic blocks we'll be examining. If that fails,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 335) * try a smaller size. We need to be able to read at least
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 336) * a log sector, or we're out of luck.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 337) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 338) bufblks = 1 << ffs(nbblks);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 339) while (bufblks > log->l_logBBsize)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 340) bufblks >>= 1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 341) while (!(buffer = xlog_alloc_buffer(log, bufblks))) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 342) bufblks >>= 1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 343) if (bufblks < log->l_sectBBsize)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 344) return -ENOMEM;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 345) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 346)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 347) for (i = start_blk; i < start_blk + nbblks; i += bufblks) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 348) int bcount;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 349)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 350) bcount = min(bufblks, (start_blk + nbblks - i));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 351)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 352) error = xlog_bread(log, i, bcount, buffer, &buf);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 353) if (error)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 354) goto out;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 355)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 356) for (j = 0; j < bcount; j++) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 357) cycle = xlog_get_cycle(buf);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 358) if (cycle == stop_on_cycle_no) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 359) *new_blk = i+j;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 360) goto out;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 361) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 362)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 363) buf += BBSIZE;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 364) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 365) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 366)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 367) *new_blk = -1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 368)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 369) out:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 370) kmem_free(buffer);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 371) return error;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 372) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 373)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 374) static inline int
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 375) xlog_logrec_hblks(struct xlog *log, struct xlog_rec_header *rh)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 376) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 377) if (xfs_sb_version_haslogv2(&log->l_mp->m_sb)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 378) int h_size = be32_to_cpu(rh->h_size);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 379)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 380) if ((be32_to_cpu(rh->h_version) & XLOG_VERSION_2) &&
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 381) h_size > XLOG_HEADER_CYCLE_SIZE)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 382) return DIV_ROUND_UP(h_size, XLOG_HEADER_CYCLE_SIZE);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 383) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 384) return 1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 385) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 386)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 387) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 388) * Potentially backup over partial log record write.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 389) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 390) * In the typical case, last_blk is the number of the block directly after
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 391) * a good log record. Therefore, we subtract one to get the block number
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 392) * of the last block in the given buffer. extra_bblks contains the number
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 393) * of blocks we would have read on a previous read. This happens when the
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 394) * last log record is split over the end of the physical log.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 395) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 396) * extra_bblks is the number of blocks potentially verified on a previous
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 397) * call to this routine.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 398) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 399) STATIC int
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 400) xlog_find_verify_log_record(
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 401) struct xlog *log,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 402) xfs_daddr_t start_blk,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 403) xfs_daddr_t *last_blk,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 404) int extra_bblks)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 405) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 406) xfs_daddr_t i;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 407) char *buffer;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 408) char *offset = NULL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 409) xlog_rec_header_t *head = NULL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 410) int error = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 411) int smallmem = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 412) int num_blks = *last_blk - start_blk;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 413) int xhdrs;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 414)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 415) ASSERT(start_blk != 0 || *last_blk != start_blk);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 416)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 417) buffer = xlog_alloc_buffer(log, num_blks);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 418) if (!buffer) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 419) buffer = xlog_alloc_buffer(log, 1);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 420) if (!buffer)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 421) return -ENOMEM;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 422) smallmem = 1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 423) } else {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 424) error = xlog_bread(log, start_blk, num_blks, buffer, &offset);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 425) if (error)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 426) goto out;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 427) offset += ((num_blks - 1) << BBSHIFT);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 428) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 429)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 430) for (i = (*last_blk) - 1; i >= 0; i--) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 431) if (i < start_blk) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 432) /* valid log record not found */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 433) xfs_warn(log->l_mp,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 434) "Log inconsistent (didn't find previous header)");
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 435) ASSERT(0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 436) error = -EFSCORRUPTED;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 437) goto out;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 438) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 439)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 440) if (smallmem) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 441) error = xlog_bread(log, i, 1, buffer, &offset);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 442) if (error)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 443) goto out;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 444) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 445)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 446) head = (xlog_rec_header_t *)offset;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 447)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 448) if (head->h_magicno == cpu_to_be32(XLOG_HEADER_MAGIC_NUM))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 449) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 450)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 451) if (!smallmem)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 452) offset -= BBSIZE;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 453) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 454)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 455) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 456) * We hit the beginning of the physical log & still no header. Return
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 457) * to caller. If caller can handle a return of -1, then this routine
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 458) * will be called again for the end of the physical log.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 459) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 460) if (i == -1) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 461) error = 1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 462) goto out;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 463) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 464)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 465) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 466) * We have the final block of the good log (the first block
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 467) * of the log record _before_ the head. So we check the uuid.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 468) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 469) if ((error = xlog_header_check_mount(log->l_mp, head)))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 470) goto out;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 471)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 472) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 473) * We may have found a log record header before we expected one.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 474) * last_blk will be the 1st block # with a given cycle #. We may end
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 475) * up reading an entire log record. In this case, we don't want to
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 476) * reset last_blk. Only when last_blk points in the middle of a log
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 477) * record do we update last_blk.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 478) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 479) xhdrs = xlog_logrec_hblks(log, head);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 480)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 481) if (*last_blk - i + extra_bblks !=
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 482) BTOBB(be32_to_cpu(head->h_len)) + xhdrs)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 483) *last_blk = i;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 484)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 485) out:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 486) kmem_free(buffer);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 487) return error;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 488) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 489)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 490) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 491) * Head is defined to be the point of the log where the next log write
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 492) * could go. This means that incomplete LR writes at the end are
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 493) * eliminated when calculating the head. We aren't guaranteed that previous
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 494) * LR have complete transactions. We only know that a cycle number of
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 495) * current cycle number -1 won't be present in the log if we start writing
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 496) * from our current block number.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 497) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 498) * last_blk contains the block number of the first block with a given
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 499) * cycle number.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 500) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 501) * Return: zero if normal, non-zero if error.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 502) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 503) STATIC int
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 504) xlog_find_head(
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 505) struct xlog *log,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 506) xfs_daddr_t *return_head_blk)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 507) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 508) char *buffer;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 509) char *offset;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 510) xfs_daddr_t new_blk, first_blk, start_blk, last_blk, head_blk;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 511) int num_scan_bblks;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 512) uint first_half_cycle, last_half_cycle;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 513) uint stop_on_cycle;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 514) int error, log_bbnum = log->l_logBBsize;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 515)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 516) /* Is the end of the log device zeroed? */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 517) error = xlog_find_zeroed(log, &first_blk);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 518) if (error < 0) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 519) xfs_warn(log->l_mp, "empty log check failed");
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 520) return error;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 521) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 522) if (error == 1) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 523) *return_head_blk = first_blk;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 524)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 525) /* Is the whole lot zeroed? */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 526) if (!first_blk) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 527) /* Linux XFS shouldn't generate totally zeroed logs -
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 528) * mkfs etc write a dummy unmount record to a fresh
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 529) * log so we can store the uuid in there
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 530) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 531) xfs_warn(log->l_mp, "totally zeroed log");
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 532) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 533)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 534) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 535) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 536)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 537) first_blk = 0; /* get cycle # of 1st block */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 538) buffer = xlog_alloc_buffer(log, 1);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 539) if (!buffer)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 540) return -ENOMEM;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 541)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 542) error = xlog_bread(log, 0, 1, buffer, &offset);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 543) if (error)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 544) goto out_free_buffer;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 545)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 546) first_half_cycle = xlog_get_cycle(offset);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 547)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 548) last_blk = head_blk = log_bbnum - 1; /* get cycle # of last block */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 549) error = xlog_bread(log, last_blk, 1, buffer, &offset);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 550) if (error)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 551) goto out_free_buffer;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 552)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 553) last_half_cycle = xlog_get_cycle(offset);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 554) ASSERT(last_half_cycle != 0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 555)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 556) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 557) * If the 1st half cycle number is equal to the last half cycle number,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 558) * then the entire log is stamped with the same cycle number. In this
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 559) * case, head_blk can't be set to zero (which makes sense). The below
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 560) * math doesn't work out properly with head_blk equal to zero. Instead,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 561) * we set it to log_bbnum which is an invalid block number, but this
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 562) * value makes the math correct. If head_blk doesn't changed through
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 563) * all the tests below, *head_blk is set to zero at the very end rather
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 564) * than log_bbnum. In a sense, log_bbnum and zero are the same block
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 565) * in a circular file.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 566) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 567) if (first_half_cycle == last_half_cycle) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 568) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 569) * In this case we believe that the entire log should have
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 570) * cycle number last_half_cycle. We need to scan backwards
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 571) * from the end verifying that there are no holes still
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 572) * containing last_half_cycle - 1. If we find such a hole,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 573) * then the start of that hole will be the new head. The
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 574) * simple case looks like
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 575) * x | x ... | x - 1 | x
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 576) * Another case that fits this picture would be
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 577) * x | x + 1 | x ... | x
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 578) * In this case the head really is somewhere at the end of the
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 579) * log, as one of the latest writes at the beginning was
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 580) * incomplete.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 581) * One more case is
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 582) * x | x + 1 | x ... | x - 1 | x
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 583) * This is really the combination of the above two cases, and
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 584) * the head has to end up at the start of the x-1 hole at the
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 585) * end of the log.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 586) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 587) * In the 256k log case, we will read from the beginning to the
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 588) * end of the log and search for cycle numbers equal to x-1.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 589) * We don't worry about the x+1 blocks that we encounter,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 590) * because we know that they cannot be the head since the log
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 591) * started with x.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 592) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 593) head_blk = log_bbnum;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 594) stop_on_cycle = last_half_cycle - 1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 595) } else {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 596) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 597) * In this case we want to find the first block with cycle
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 598) * number matching last_half_cycle. We expect the log to be
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 599) * some variation on
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 600) * x + 1 ... | x ... | x
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 601) * The first block with cycle number x (last_half_cycle) will
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 602) * be where the new head belongs. First we do a binary search
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 603) * for the first occurrence of last_half_cycle. The binary
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 604) * search may not be totally accurate, so then we scan back
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 605) * from there looking for occurrences of last_half_cycle before
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 606) * us. If that backwards scan wraps around the beginning of
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 607) * the log, then we look for occurrences of last_half_cycle - 1
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 608) * at the end of the log. The cases we're looking for look
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 609) * like
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 610) * v binary search stopped here
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 611) * x + 1 ... | x | x + 1 | x ... | x
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 612) * ^ but we want to locate this spot
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 613) * or
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 614) * <---------> less than scan distance
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 615) * x + 1 ... | x ... | x - 1 | x
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 616) * ^ we want to locate this spot
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 617) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 618) stop_on_cycle = last_half_cycle;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 619) error = xlog_find_cycle_start(log, buffer, first_blk, &head_blk,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 620) last_half_cycle);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 621) if (error)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 622) goto out_free_buffer;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 623) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 624)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 625) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 626) * Now validate the answer. Scan back some number of maximum possible
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 627) * blocks and make sure each one has the expected cycle number. The
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 628) * maximum is determined by the total possible amount of buffering
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 629) * in the in-core log. The following number can be made tighter if
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 630) * we actually look at the block size of the filesystem.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 631) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 632) num_scan_bblks = min_t(int, log_bbnum, XLOG_TOTAL_REC_SHIFT(log));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 633) if (head_blk >= num_scan_bblks) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 634) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 635) * We are guaranteed that the entire check can be performed
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 636) * in one buffer.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 637) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 638) start_blk = head_blk - num_scan_bblks;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 639) if ((error = xlog_find_verify_cycle(log,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 640) start_blk, num_scan_bblks,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 641) stop_on_cycle, &new_blk)))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 642) goto out_free_buffer;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 643) if (new_blk != -1)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 644) head_blk = new_blk;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 645) } else { /* need to read 2 parts of log */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 646) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 647) * We are going to scan backwards in the log in two parts.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 648) * First we scan the physical end of the log. In this part
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 649) * of the log, we are looking for blocks with cycle number
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 650) * last_half_cycle - 1.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 651) * If we find one, then we know that the log starts there, as
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 652) * we've found a hole that didn't get written in going around
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 653) * the end of the physical log. The simple case for this is
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 654) * x + 1 ... | x ... | x - 1 | x
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 655) * <---------> less than scan distance
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 656) * If all of the blocks at the end of the log have cycle number
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 657) * last_half_cycle, then we check the blocks at the start of
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 658) * the log looking for occurrences of last_half_cycle. If we
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 659) * find one, then our current estimate for the location of the
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 660) * first occurrence of last_half_cycle is wrong and we move
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 661) * back to the hole we've found. This case looks like
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 662) * x + 1 ... | x | x + 1 | x ...
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 663) * ^ binary search stopped here
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 664) * Another case we need to handle that only occurs in 256k
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 665) * logs is
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 666) * x + 1 ... | x ... | x+1 | x ...
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 667) * ^ binary search stops here
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 668) * In a 256k log, the scan at the end of the log will see the
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 669) * x + 1 blocks. We need to skip past those since that is
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 670) * certainly not the head of the log. By searching for
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 671) * last_half_cycle-1 we accomplish that.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 672) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 673) ASSERT(head_blk <= INT_MAX &&
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 674) (xfs_daddr_t) num_scan_bblks >= head_blk);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 675) start_blk = log_bbnum - (num_scan_bblks - head_blk);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 676) if ((error = xlog_find_verify_cycle(log, start_blk,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 677) num_scan_bblks - (int)head_blk,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 678) (stop_on_cycle - 1), &new_blk)))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 679) goto out_free_buffer;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 680) if (new_blk != -1) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 681) head_blk = new_blk;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 682) goto validate_head;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 683) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 684)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 685) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 686) * Scan beginning of log now. The last part of the physical
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 687) * log is good. This scan needs to verify that it doesn't find
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 688) * the last_half_cycle.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 689) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 690) start_blk = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 691) ASSERT(head_blk <= INT_MAX);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 692) if ((error = xlog_find_verify_cycle(log,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 693) start_blk, (int)head_blk,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 694) stop_on_cycle, &new_blk)))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 695) goto out_free_buffer;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 696) if (new_blk != -1)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 697) head_blk = new_blk;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 698) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 699)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 700) validate_head:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 701) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 702) * Now we need to make sure head_blk is not pointing to a block in
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 703) * the middle of a log record.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 704) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 705) num_scan_bblks = XLOG_REC_SHIFT(log);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 706) if (head_blk >= num_scan_bblks) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 707) start_blk = head_blk - num_scan_bblks; /* don't read head_blk */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 708)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 709) /* start ptr at last block ptr before head_blk */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 710) error = xlog_find_verify_log_record(log, start_blk, &head_blk, 0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 711) if (error == 1)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 712) error = -EIO;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 713) if (error)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 714) goto out_free_buffer;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 715) } else {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 716) start_blk = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 717) ASSERT(head_blk <= INT_MAX);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 718) error = xlog_find_verify_log_record(log, start_blk, &head_blk, 0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 719) if (error < 0)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 720) goto out_free_buffer;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 721) if (error == 1) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 722) /* We hit the beginning of the log during our search */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 723) start_blk = log_bbnum - (num_scan_bblks - head_blk);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 724) new_blk = log_bbnum;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 725) ASSERT(start_blk <= INT_MAX &&
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 726) (xfs_daddr_t) log_bbnum-start_blk >= 0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 727) ASSERT(head_blk <= INT_MAX);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 728) error = xlog_find_verify_log_record(log, start_blk,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 729) &new_blk, (int)head_blk);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 730) if (error == 1)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 731) error = -EIO;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 732) if (error)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 733) goto out_free_buffer;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 734) if (new_blk != log_bbnum)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 735) head_blk = new_blk;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 736) } else if (error)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 737) goto out_free_buffer;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 738) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 739)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 740) kmem_free(buffer);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 741) if (head_blk == log_bbnum)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 742) *return_head_blk = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 743) else
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 744) *return_head_blk = head_blk;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 745) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 746) * When returning here, we have a good block number. Bad block
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 747) * means that during a previous crash, we didn't have a clean break
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 748) * from cycle number N to cycle number N-1. In this case, we need
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 749) * to find the first block with cycle number N-1.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 750) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 751) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 752)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 753) out_free_buffer:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 754) kmem_free(buffer);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 755) if (error)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 756) xfs_warn(log->l_mp, "failed to find log head");
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 757) return error;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 758) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 759)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 760) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 761) * Seek backwards in the log for log record headers.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 762) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 763) * Given a starting log block, walk backwards until we find the provided number
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 764) * of records or hit the provided tail block. The return value is the number of
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 765) * records encountered or a negative error code. The log block and buffer
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 766) * pointer of the last record seen are returned in rblk and rhead respectively.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 767) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 768) STATIC int
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 769) xlog_rseek_logrec_hdr(
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 770) struct xlog *log,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 771) xfs_daddr_t head_blk,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 772) xfs_daddr_t tail_blk,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 773) int count,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 774) char *buffer,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 775) xfs_daddr_t *rblk,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 776) struct xlog_rec_header **rhead,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 777) bool *wrapped)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 778) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 779) int i;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 780) int error;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 781) int found = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 782) char *offset = NULL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 783) xfs_daddr_t end_blk;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 784)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 785) *wrapped = false;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 786)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 787) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 788) * Walk backwards from the head block until we hit the tail or the first
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 789) * block in the log.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 790) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 791) end_blk = head_blk > tail_blk ? tail_blk : 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 792) for (i = (int) head_blk - 1; i >= end_blk; i--) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 793) error = xlog_bread(log, i, 1, buffer, &offset);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 794) if (error)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 795) goto out_error;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 796)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 797) if (*(__be32 *) offset == cpu_to_be32(XLOG_HEADER_MAGIC_NUM)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 798) *rblk = i;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 799) *rhead = (struct xlog_rec_header *) offset;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 800) if (++found == count)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 801) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 802) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 803) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 804)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 805) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 806) * If we haven't hit the tail block or the log record header count,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 807) * start looking again from the end of the physical log. Note that
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 808) * callers can pass head == tail if the tail is not yet known.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 809) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 810) if (tail_blk >= head_blk && found != count) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 811) for (i = log->l_logBBsize - 1; i >= (int) tail_blk; i--) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 812) error = xlog_bread(log, i, 1, buffer, &offset);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 813) if (error)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 814) goto out_error;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 815)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 816) if (*(__be32 *)offset ==
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 817) cpu_to_be32(XLOG_HEADER_MAGIC_NUM)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 818) *wrapped = true;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 819) *rblk = i;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 820) *rhead = (struct xlog_rec_header *) offset;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 821) if (++found == count)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 822) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 823) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 824) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 825) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 826)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 827) return found;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 828)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 829) out_error:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 830) return error;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 831) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 832)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 833) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 834) * Seek forward in the log for log record headers.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 835) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 836) * Given head and tail blocks, walk forward from the tail block until we find
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 837) * the provided number of records or hit the head block. The return value is the
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 838) * number of records encountered or a negative error code. The log block and
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 839) * buffer pointer of the last record seen are returned in rblk and rhead
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 840) * respectively.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 841) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 842) STATIC int
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 843) xlog_seek_logrec_hdr(
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 844) struct xlog *log,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 845) xfs_daddr_t head_blk,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 846) xfs_daddr_t tail_blk,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 847) int count,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 848) char *buffer,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 849) xfs_daddr_t *rblk,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 850) struct xlog_rec_header **rhead,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 851) bool *wrapped)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 852) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 853) int i;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 854) int error;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 855) int found = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 856) char *offset = NULL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 857) xfs_daddr_t end_blk;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 858)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 859) *wrapped = false;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 860)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 861) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 862) * Walk forward from the tail block until we hit the head or the last
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 863) * block in the log.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 864) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 865) end_blk = head_blk > tail_blk ? head_blk : log->l_logBBsize - 1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 866) for (i = (int) tail_blk; i <= end_blk; i++) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 867) error = xlog_bread(log, i, 1, buffer, &offset);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 868) if (error)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 869) goto out_error;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 870)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 871) if (*(__be32 *) offset == cpu_to_be32(XLOG_HEADER_MAGIC_NUM)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 872) *rblk = i;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 873) *rhead = (struct xlog_rec_header *) offset;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 874) if (++found == count)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 875) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 876) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 877) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 878)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 879) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 880) * If we haven't hit the head block or the log record header count,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 881) * start looking again from the start of the physical log.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 882) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 883) if (tail_blk > head_blk && found != count) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 884) for (i = 0; i < (int) head_blk; i++) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 885) error = xlog_bread(log, i, 1, buffer, &offset);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 886) if (error)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 887) goto out_error;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 888)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 889) if (*(__be32 *)offset ==
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 890) cpu_to_be32(XLOG_HEADER_MAGIC_NUM)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 891) *wrapped = true;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 892) *rblk = i;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 893) *rhead = (struct xlog_rec_header *) offset;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 894) if (++found == count)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 895) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 896) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 897) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 898) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 899)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 900) return found;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 901)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 902) out_error:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 903) return error;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 904) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 905)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 906) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 907) * Calculate distance from head to tail (i.e., unused space in the log).
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 908) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 909) static inline int
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 910) xlog_tail_distance(
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 911) struct xlog *log,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 912) xfs_daddr_t head_blk,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 913) xfs_daddr_t tail_blk)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 914) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 915) if (head_blk < tail_blk)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 916) return tail_blk - head_blk;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 917)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 918) return tail_blk + (log->l_logBBsize - head_blk);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 919) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 920)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 921) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 922) * Verify the log tail. This is particularly important when torn or incomplete
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 923) * writes have been detected near the front of the log and the head has been
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 924) * walked back accordingly.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 925) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 926) * We also have to handle the case where the tail was pinned and the head
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 927) * blocked behind the tail right before a crash. If the tail had been pushed
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 928) * immediately prior to the crash and the subsequent checkpoint was only
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 929) * partially written, it's possible it overwrote the last referenced tail in the
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 930) * log with garbage. This is not a coherency problem because the tail must have
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 931) * been pushed before it can be overwritten, but appears as log corruption to
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 932) * recovery because we have no way to know the tail was updated if the
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 933) * subsequent checkpoint didn't write successfully.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 934) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 935) * Therefore, CRC check the log from tail to head. If a failure occurs and the
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 936) * offending record is within max iclog bufs from the head, walk the tail
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 937) * forward and retry until a valid tail is found or corruption is detected out
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 938) * of the range of a possible overwrite.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 939) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 940) STATIC int
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 941) xlog_verify_tail(
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 942) struct xlog *log,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 943) xfs_daddr_t head_blk,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 944) xfs_daddr_t *tail_blk,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 945) int hsize)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 946) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 947) struct xlog_rec_header *thead;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 948) char *buffer;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 949) xfs_daddr_t first_bad;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 950) int error = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 951) bool wrapped;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 952) xfs_daddr_t tmp_tail;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 953) xfs_daddr_t orig_tail = *tail_blk;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 954)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 955) buffer = xlog_alloc_buffer(log, 1);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 956) if (!buffer)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 957) return -ENOMEM;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 958)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 959) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 960) * Make sure the tail points to a record (returns positive count on
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 961) * success).
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 962) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 963) error = xlog_seek_logrec_hdr(log, head_blk, *tail_blk, 1, buffer,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 964) &tmp_tail, &thead, &wrapped);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 965) if (error < 0)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 966) goto out;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 967) if (*tail_blk != tmp_tail)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 968) *tail_blk = tmp_tail;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 969)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 970) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 971) * Run a CRC check from the tail to the head. We can't just check
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 972) * MAX_ICLOGS records past the tail because the tail may point to stale
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 973) * blocks cleared during the search for the head/tail. These blocks are
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 974) * overwritten with zero-length records and thus record count is not a
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 975) * reliable indicator of the iclog state before a crash.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 976) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 977) first_bad = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 978) error = xlog_do_recovery_pass(log, head_blk, *tail_blk,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 979) XLOG_RECOVER_CRCPASS, &first_bad);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 980) while ((error == -EFSBADCRC || error == -EFSCORRUPTED) && first_bad) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 981) int tail_distance;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 982)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 983) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 984) * Is corruption within range of the head? If so, retry from
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 985) * the next record. Otherwise return an error.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 986) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 987) tail_distance = xlog_tail_distance(log, head_blk, first_bad);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 988) if (tail_distance > BTOBB(XLOG_MAX_ICLOGS * hsize))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 989) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 990)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 991) /* skip to the next record; returns positive count on success */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 992) error = xlog_seek_logrec_hdr(log, head_blk, first_bad, 2,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 993) buffer, &tmp_tail, &thead, &wrapped);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 994) if (error < 0)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 995) goto out;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 996)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 997) *tail_blk = tmp_tail;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 998) first_bad = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 999) error = xlog_do_recovery_pass(log, head_blk, *tail_blk,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1000) XLOG_RECOVER_CRCPASS, &first_bad);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1001) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1002)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1003) if (!error && *tail_blk != orig_tail)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1004) xfs_warn(log->l_mp,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1005) "Tail block (0x%llx) overwrite detected. Updated to 0x%llx",
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1006) orig_tail, *tail_blk);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1007) out:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1008) kmem_free(buffer);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1009) return error;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1010) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1011)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1012) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1013) * Detect and trim torn writes from the head of the log.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1014) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1015) * Storage without sector atomicity guarantees can result in torn writes in the
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1016) * log in the event of a crash. Our only means to detect this scenario is via
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1017) * CRC verification. While we can't always be certain that CRC verification
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1018) * failure is due to a torn write vs. an unrelated corruption, we do know that
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1019) * only a certain number (XLOG_MAX_ICLOGS) of log records can be written out at
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1020) * one time. Therefore, CRC verify up to XLOG_MAX_ICLOGS records at the head of
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1021) * the log and treat failures in this range as torn writes as a matter of
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1022) * policy. In the event of CRC failure, the head is walked back to the last good
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1023) * record in the log and the tail is updated from that record and verified.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1024) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1025) STATIC int
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1026) xlog_verify_head(
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1027) struct xlog *log,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1028) xfs_daddr_t *head_blk, /* in/out: unverified head */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1029) xfs_daddr_t *tail_blk, /* out: tail block */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1030) char *buffer,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1031) xfs_daddr_t *rhead_blk, /* start blk of last record */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1032) struct xlog_rec_header **rhead, /* ptr to last record */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1033) bool *wrapped) /* last rec. wraps phys. log */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1034) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1035) struct xlog_rec_header *tmp_rhead;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1036) char *tmp_buffer;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1037) xfs_daddr_t first_bad;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1038) xfs_daddr_t tmp_rhead_blk;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1039) int found;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1040) int error;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1041) bool tmp_wrapped;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1042)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1043) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1044) * Check the head of the log for torn writes. Search backwards from the
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1045) * head until we hit the tail or the maximum number of log record I/Os
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1046) * that could have been in flight at one time. Use a temporary buffer so
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1047) * we don't trash the rhead/buffer pointers from the caller.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1048) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1049) tmp_buffer = xlog_alloc_buffer(log, 1);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1050) if (!tmp_buffer)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1051) return -ENOMEM;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1052) error = xlog_rseek_logrec_hdr(log, *head_blk, *tail_blk,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1053) XLOG_MAX_ICLOGS, tmp_buffer,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1054) &tmp_rhead_blk, &tmp_rhead, &tmp_wrapped);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1055) kmem_free(tmp_buffer);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1056) if (error < 0)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1057) return error;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1058)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1059) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1060) * Now run a CRC verification pass over the records starting at the
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1061) * block found above to the current head. If a CRC failure occurs, the
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1062) * log block of the first bad record is saved in first_bad.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1063) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1064) error = xlog_do_recovery_pass(log, *head_blk, tmp_rhead_blk,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1065) XLOG_RECOVER_CRCPASS, &first_bad);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1066) if ((error == -EFSBADCRC || error == -EFSCORRUPTED) && first_bad) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1067) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1068) * We've hit a potential torn write. Reset the error and warn
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1069) * about it.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1070) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1071) error = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1072) xfs_warn(log->l_mp,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1073) "Torn write (CRC failure) detected at log block 0x%llx. Truncating head block from 0x%llx.",
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1074) first_bad, *head_blk);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1075)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1076) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1077) * Get the header block and buffer pointer for the last good
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1078) * record before the bad record.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1079) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1080) * Note that xlog_find_tail() clears the blocks at the new head
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1081) * (i.e., the records with invalid CRC) if the cycle number
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1082) * matches the current cycle.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1083) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1084) found = xlog_rseek_logrec_hdr(log, first_bad, *tail_blk, 1,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1085) buffer, rhead_blk, rhead, wrapped);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1086) if (found < 0)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1087) return found;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1088) if (found == 0) /* XXX: right thing to do here? */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1089) return -EIO;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1090)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1091) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1092) * Reset the head block to the starting block of the first bad
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1093) * log record and set the tail block based on the last good
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1094) * record.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1095) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1096) * Bail out if the updated head/tail match as this indicates
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1097) * possible corruption outside of the acceptable
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1098) * (XLOG_MAX_ICLOGS) range. This is a job for xfs_repair...
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1099) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1100) *head_blk = first_bad;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1101) *tail_blk = BLOCK_LSN(be64_to_cpu((*rhead)->h_tail_lsn));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1102) if (*head_blk == *tail_blk) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1103) ASSERT(0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1104) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1105) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1106) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1107) if (error)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1108) return error;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1109)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1110) return xlog_verify_tail(log, *head_blk, tail_blk,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1111) be32_to_cpu((*rhead)->h_size));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1112) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1113)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1114) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1115) * We need to make sure we handle log wrapping properly, so we can't use the
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1116) * calculated logbno directly. Make sure it wraps to the correct bno inside the
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1117) * log.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1118) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1119) * The log is limited to 32 bit sizes, so we use the appropriate modulus
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1120) * operation here and cast it back to a 64 bit daddr on return.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1121) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1122) static inline xfs_daddr_t
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1123) xlog_wrap_logbno(
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1124) struct xlog *log,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1125) xfs_daddr_t bno)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1126) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1127) int mod;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1128)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1129) div_s64_rem(bno, log->l_logBBsize, &mod);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1130) return mod;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1131) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1132)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1133) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1134) * Check whether the head of the log points to an unmount record. In other
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1135) * words, determine whether the log is clean. If so, update the in-core state
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1136) * appropriately.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1137) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1138) static int
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1139) xlog_check_unmount_rec(
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1140) struct xlog *log,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1141) xfs_daddr_t *head_blk,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1142) xfs_daddr_t *tail_blk,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1143) struct xlog_rec_header *rhead,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1144) xfs_daddr_t rhead_blk,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1145) char *buffer,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1146) bool *clean)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1147) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1148) struct xlog_op_header *op_head;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1149) xfs_daddr_t umount_data_blk;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1150) xfs_daddr_t after_umount_blk;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1151) int hblks;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1152) int error;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1153) char *offset;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1154)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1155) *clean = false;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1156)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1157) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1158) * Look for unmount record. If we find it, then we know there was a
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1159) * clean unmount. Since 'i' could be the last block in the physical
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1160) * log, we convert to a log block before comparing to the head_blk.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1161) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1162) * Save the current tail lsn to use to pass to xlog_clear_stale_blocks()
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1163) * below. We won't want to clear the unmount record if there is one, so
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1164) * we pass the lsn of the unmount record rather than the block after it.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1165) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1166) hblks = xlog_logrec_hblks(log, rhead);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1167) after_umount_blk = xlog_wrap_logbno(log,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1168) rhead_blk + hblks + BTOBB(be32_to_cpu(rhead->h_len)));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1169)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1170) if (*head_blk == after_umount_blk &&
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1171) be32_to_cpu(rhead->h_num_logops) == 1) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1172) umount_data_blk = xlog_wrap_logbno(log, rhead_blk + hblks);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1173) error = xlog_bread(log, umount_data_blk, 1, buffer, &offset);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1174) if (error)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1175) return error;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1176)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1177) op_head = (struct xlog_op_header *)offset;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1178) if (op_head->oh_flags & XLOG_UNMOUNT_TRANS) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1179) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1180) * Set tail and last sync so that newly written log
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1181) * records will point recovery to after the current
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1182) * unmount record.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1183) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1184) xlog_assign_atomic_lsn(&log->l_tail_lsn,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1185) log->l_curr_cycle, after_umount_blk);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1186) xlog_assign_atomic_lsn(&log->l_last_sync_lsn,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1187) log->l_curr_cycle, after_umount_blk);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1188) *tail_blk = after_umount_blk;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1189)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1190) *clean = true;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1191) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1192) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1193)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1194) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1195) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1196)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1197) static void
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1198) xlog_set_state(
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1199) struct xlog *log,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1200) xfs_daddr_t head_blk,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1201) struct xlog_rec_header *rhead,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1202) xfs_daddr_t rhead_blk,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1203) bool bump_cycle)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1204) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1205) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1206) * Reset log values according to the state of the log when we
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1207) * crashed. In the case where head_blk == 0, we bump curr_cycle
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1208) * one because the next write starts a new cycle rather than
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1209) * continuing the cycle of the last good log record. At this
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1210) * point we have guaranteed that all partial log records have been
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1211) * accounted for. Therefore, we know that the last good log record
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1212) * written was complete and ended exactly on the end boundary
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1213) * of the physical log.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1214) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1215) log->l_prev_block = rhead_blk;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1216) log->l_curr_block = (int)head_blk;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1217) log->l_curr_cycle = be32_to_cpu(rhead->h_cycle);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1218) if (bump_cycle)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1219) log->l_curr_cycle++;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1220) atomic64_set(&log->l_tail_lsn, be64_to_cpu(rhead->h_tail_lsn));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1221) atomic64_set(&log->l_last_sync_lsn, be64_to_cpu(rhead->h_lsn));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1222) xlog_assign_grant_head(&log->l_reserve_head.grant, log->l_curr_cycle,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1223) BBTOB(log->l_curr_block));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1224) xlog_assign_grant_head(&log->l_write_head.grant, log->l_curr_cycle,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1225) BBTOB(log->l_curr_block));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1226) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1227)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1228) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1229) * Find the sync block number or the tail of the log.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1230) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1231) * This will be the block number of the last record to have its
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1232) * associated buffers synced to disk. Every log record header has
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1233) * a sync lsn embedded in it. LSNs hold block numbers, so it is easy
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1234) * to get a sync block number. The only concern is to figure out which
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1235) * log record header to believe.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1236) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1237) * The following algorithm uses the log record header with the largest
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1238) * lsn. The entire log record does not need to be valid. We only care
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1239) * that the header is valid.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1240) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1241) * We could speed up search by using current head_blk buffer, but it is not
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1242) * available.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1243) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1244) STATIC int
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1245) xlog_find_tail(
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1246) struct xlog *log,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1247) xfs_daddr_t *head_blk,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1248) xfs_daddr_t *tail_blk)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1249) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1250) xlog_rec_header_t *rhead;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1251) char *offset = NULL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1252) char *buffer;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1253) int error;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1254) xfs_daddr_t rhead_blk;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1255) xfs_lsn_t tail_lsn;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1256) bool wrapped = false;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1257) bool clean = false;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1258)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1259) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1260) * Find previous log record
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1261) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1262) if ((error = xlog_find_head(log, head_blk)))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1263) return error;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1264) ASSERT(*head_blk < INT_MAX);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1265)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1266) buffer = xlog_alloc_buffer(log, 1);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1267) if (!buffer)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1268) return -ENOMEM;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1269) if (*head_blk == 0) { /* special case */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1270) error = xlog_bread(log, 0, 1, buffer, &offset);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1271) if (error)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1272) goto done;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1273)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1274) if (xlog_get_cycle(offset) == 0) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1275) *tail_blk = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1276) /* leave all other log inited values alone */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1277) goto done;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1278) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1279) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1280)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1281) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1282) * Search backwards through the log looking for the log record header
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1283) * block. This wraps all the way back around to the head so something is
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1284) * seriously wrong if we can't find it.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1285) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1286) error = xlog_rseek_logrec_hdr(log, *head_blk, *head_blk, 1, buffer,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1287) &rhead_blk, &rhead, &wrapped);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1288) if (error < 0)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1289) goto done;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1290) if (!error) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1291) xfs_warn(log->l_mp, "%s: couldn't find sync record", __func__);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1292) error = -EFSCORRUPTED;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1293) goto done;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1294) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1295) *tail_blk = BLOCK_LSN(be64_to_cpu(rhead->h_tail_lsn));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1296)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1297) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1298) * Set the log state based on the current head record.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1299) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1300) xlog_set_state(log, *head_blk, rhead, rhead_blk, wrapped);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1301) tail_lsn = atomic64_read(&log->l_tail_lsn);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1302)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1303) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1304) * Look for an unmount record at the head of the log. This sets the log
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1305) * state to determine whether recovery is necessary.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1306) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1307) error = xlog_check_unmount_rec(log, head_blk, tail_blk, rhead,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1308) rhead_blk, buffer, &clean);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1309) if (error)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1310) goto done;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1311)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1312) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1313) * Verify the log head if the log is not clean (e.g., we have anything
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1314) * but an unmount record at the head). This uses CRC verification to
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1315) * detect and trim torn writes. If discovered, CRC failures are
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1316) * considered torn writes and the log head is trimmed accordingly.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1317) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1318) * Note that we can only run CRC verification when the log is dirty
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1319) * because there's no guarantee that the log data behind an unmount
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1320) * record is compatible with the current architecture.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1321) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1322) if (!clean) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1323) xfs_daddr_t orig_head = *head_blk;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1324)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1325) error = xlog_verify_head(log, head_blk, tail_blk, buffer,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1326) &rhead_blk, &rhead, &wrapped);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1327) if (error)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1328) goto done;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1329)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1330) /* update in-core state again if the head changed */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1331) if (*head_blk != orig_head) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1332) xlog_set_state(log, *head_blk, rhead, rhead_blk,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1333) wrapped);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1334) tail_lsn = atomic64_read(&log->l_tail_lsn);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1335) error = xlog_check_unmount_rec(log, head_blk, tail_blk,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1336) rhead, rhead_blk, buffer,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1337) &clean);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1338) if (error)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1339) goto done;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1340) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1341) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1342)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1343) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1344) * Note that the unmount was clean. If the unmount was not clean, we
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1345) * need to know this to rebuild the superblock counters from the perag
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1346) * headers if we have a filesystem using non-persistent counters.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1347) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1348) if (clean)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1349) log->l_mp->m_flags |= XFS_MOUNT_WAS_CLEAN;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1350)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1351) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1352) * Make sure that there are no blocks in front of the head
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1353) * with the same cycle number as the head. This can happen
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1354) * because we allow multiple outstanding log writes concurrently,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1355) * and the later writes might make it out before earlier ones.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1356) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1357) * We use the lsn from before modifying it so that we'll never
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1358) * overwrite the unmount record after a clean unmount.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1359) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1360) * Do this only if we are going to recover the filesystem
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1361) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1362) * NOTE: This used to say "if (!readonly)"
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1363) * However on Linux, we can & do recover a read-only filesystem.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1364) * We only skip recovery if NORECOVERY is specified on mount,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1365) * in which case we would not be here.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1366) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1367) * But... if the -device- itself is readonly, just skip this.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1368) * We can't recover this device anyway, so it won't matter.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1369) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1370) if (!xfs_readonly_buftarg(log->l_targ))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1371) error = xlog_clear_stale_blocks(log, tail_lsn);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1372)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1373) done:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1374) kmem_free(buffer);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1375)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1376) if (error)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1377) xfs_warn(log->l_mp, "failed to locate log tail");
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1378) return error;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1379) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1380)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1381) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1382) * Is the log zeroed at all?
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1383) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1384) * The last binary search should be changed to perform an X block read
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1385) * once X becomes small enough. You can then search linearly through
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1386) * the X blocks. This will cut down on the number of reads we need to do.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1387) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1388) * If the log is partially zeroed, this routine will pass back the blkno
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1389) * of the first block with cycle number 0. It won't have a complete LR
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1390) * preceding it.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1391) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1392) * Return:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1393) * 0 => the log is completely written to
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1394) * 1 => use *blk_no as the first block of the log
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1395) * <0 => error has occurred
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1396) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1397) STATIC int
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1398) xlog_find_zeroed(
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1399) struct xlog *log,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1400) xfs_daddr_t *blk_no)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1401) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1402) char *buffer;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1403) char *offset;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1404) uint first_cycle, last_cycle;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1405) xfs_daddr_t new_blk, last_blk, start_blk;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1406) xfs_daddr_t num_scan_bblks;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1407) int error, log_bbnum = log->l_logBBsize;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1408)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1409) *blk_no = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1410)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1411) /* check totally zeroed log */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1412) buffer = xlog_alloc_buffer(log, 1);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1413) if (!buffer)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1414) return -ENOMEM;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1415) error = xlog_bread(log, 0, 1, buffer, &offset);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1416) if (error)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1417) goto out_free_buffer;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1418)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1419) first_cycle = xlog_get_cycle(offset);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1420) if (first_cycle == 0) { /* completely zeroed log */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1421) *blk_no = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1422) kmem_free(buffer);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1423) return 1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1424) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1425)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1426) /* check partially zeroed log */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1427) error = xlog_bread(log, log_bbnum-1, 1, buffer, &offset);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1428) if (error)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1429) goto out_free_buffer;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1430)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1431) last_cycle = xlog_get_cycle(offset);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1432) if (last_cycle != 0) { /* log completely written to */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1433) kmem_free(buffer);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1434) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1435) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1436)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1437) /* we have a partially zeroed log */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1438) last_blk = log_bbnum-1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1439) error = xlog_find_cycle_start(log, buffer, 0, &last_blk, 0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1440) if (error)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1441) goto out_free_buffer;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1442)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1443) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1444) * Validate the answer. Because there is no way to guarantee that
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1445) * the entire log is made up of log records which are the same size,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1446) * we scan over the defined maximum blocks. At this point, the maximum
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1447) * is not chosen to mean anything special. XXXmiken
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1448) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1449) num_scan_bblks = XLOG_TOTAL_REC_SHIFT(log);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1450) ASSERT(num_scan_bblks <= INT_MAX);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1451)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1452) if (last_blk < num_scan_bblks)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1453) num_scan_bblks = last_blk;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1454) start_blk = last_blk - num_scan_bblks;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1455)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1456) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1457) * We search for any instances of cycle number 0 that occur before
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1458) * our current estimate of the head. What we're trying to detect is
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1459) * 1 ... | 0 | 1 | 0...
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1460) * ^ binary search ends here
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1461) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1462) if ((error = xlog_find_verify_cycle(log, start_blk,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1463) (int)num_scan_bblks, 0, &new_blk)))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1464) goto out_free_buffer;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1465) if (new_blk != -1)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1466) last_blk = new_blk;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1467)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1468) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1469) * Potentially backup over partial log record write. We don't need
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1470) * to search the end of the log because we know it is zero.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1471) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1472) error = xlog_find_verify_log_record(log, start_blk, &last_blk, 0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1473) if (error == 1)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1474) error = -EIO;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1475) if (error)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1476) goto out_free_buffer;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1477)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1478) *blk_no = last_blk;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1479) out_free_buffer:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1480) kmem_free(buffer);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1481) if (error)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1482) return error;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1483) return 1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1484) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1485)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1486) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1487) * These are simple subroutines used by xlog_clear_stale_blocks() below
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1488) * to initialize a buffer full of empty log record headers and write
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1489) * them into the log.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1490) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1491) STATIC void
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1492) xlog_add_record(
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1493) struct xlog *log,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1494) char *buf,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1495) int cycle,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1496) int block,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1497) int tail_cycle,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1498) int tail_block)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1499) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1500) xlog_rec_header_t *recp = (xlog_rec_header_t *)buf;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1501)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1502) memset(buf, 0, BBSIZE);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1503) recp->h_magicno = cpu_to_be32(XLOG_HEADER_MAGIC_NUM);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1504) recp->h_cycle = cpu_to_be32(cycle);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1505) recp->h_version = cpu_to_be32(
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1506) xfs_sb_version_haslogv2(&log->l_mp->m_sb) ? 2 : 1);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1507) recp->h_lsn = cpu_to_be64(xlog_assign_lsn(cycle, block));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1508) recp->h_tail_lsn = cpu_to_be64(xlog_assign_lsn(tail_cycle, tail_block));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1509) recp->h_fmt = cpu_to_be32(XLOG_FMT);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1510) memcpy(&recp->h_fs_uuid, &log->l_mp->m_sb.sb_uuid, sizeof(uuid_t));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1511) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1512)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1513) STATIC int
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1514) xlog_write_log_records(
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1515) struct xlog *log,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1516) int cycle,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1517) int start_block,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1518) int blocks,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1519) int tail_cycle,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1520) int tail_block)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1521) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1522) char *offset;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1523) char *buffer;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1524) int balign, ealign;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1525) int sectbb = log->l_sectBBsize;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1526) int end_block = start_block + blocks;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1527) int bufblks;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1528) int error = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1529) int i, j = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1530)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1531) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1532) * Greedily allocate a buffer big enough to handle the full
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1533) * range of basic blocks to be written. If that fails, try
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1534) * a smaller size. We need to be able to write at least a
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1535) * log sector, or we're out of luck.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1536) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1537) bufblks = 1 << ffs(blocks);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1538) while (bufblks > log->l_logBBsize)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1539) bufblks >>= 1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1540) while (!(buffer = xlog_alloc_buffer(log, bufblks))) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1541) bufblks >>= 1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1542) if (bufblks < sectbb)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1543) return -ENOMEM;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1544) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1545)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1546) /* We may need to do a read at the start to fill in part of
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1547) * the buffer in the starting sector not covered by the first
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1548) * write below.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1549) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1550) balign = round_down(start_block, sectbb);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1551) if (balign != start_block) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1552) error = xlog_bread_noalign(log, start_block, 1, buffer);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1553) if (error)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1554) goto out_free_buffer;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1555)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1556) j = start_block - balign;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1557) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1558)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1559) for (i = start_block; i < end_block; i += bufblks) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1560) int bcount, endcount;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1561)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1562) bcount = min(bufblks, end_block - start_block);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1563) endcount = bcount - j;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1564)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1565) /* We may need to do a read at the end to fill in part of
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1566) * the buffer in the final sector not covered by the write.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1567) * If this is the same sector as the above read, skip it.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1568) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1569) ealign = round_down(end_block, sectbb);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1570) if (j == 0 && (start_block + endcount > ealign)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1571) error = xlog_bread_noalign(log, ealign, sectbb,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1572) buffer + BBTOB(ealign - start_block));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1573) if (error)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1574) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1575)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1576) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1577)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1578) offset = buffer + xlog_align(log, start_block);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1579) for (; j < endcount; j++) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1580) xlog_add_record(log, offset, cycle, i+j,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1581) tail_cycle, tail_block);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1582) offset += BBSIZE;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1583) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1584) error = xlog_bwrite(log, start_block, endcount, buffer);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1585) if (error)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1586) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1587) start_block += endcount;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1588) j = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1589) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1590)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1591) out_free_buffer:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1592) kmem_free(buffer);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1593) return error;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1594) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1595)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1596) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1597) * This routine is called to blow away any incomplete log writes out
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1598) * in front of the log head. We do this so that we won't become confused
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1599) * if we come up, write only a little bit more, and then crash again.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1600) * If we leave the partial log records out there, this situation could
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1601) * cause us to think those partial writes are valid blocks since they
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1602) * have the current cycle number. We get rid of them by overwriting them
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1603) * with empty log records with the old cycle number rather than the
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1604) * current one.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1605) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1606) * The tail lsn is passed in rather than taken from
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1607) * the log so that we will not write over the unmount record after a
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1608) * clean unmount in a 512 block log. Doing so would leave the log without
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1609) * any valid log records in it until a new one was written. If we crashed
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1610) * during that time we would not be able to recover.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1611) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1612) STATIC int
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1613) xlog_clear_stale_blocks(
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1614) struct xlog *log,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1615) xfs_lsn_t tail_lsn)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1616) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1617) int tail_cycle, head_cycle;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1618) int tail_block, head_block;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1619) int tail_distance, max_distance;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1620) int distance;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1621) int error;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1622)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1623) tail_cycle = CYCLE_LSN(tail_lsn);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1624) tail_block = BLOCK_LSN(tail_lsn);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1625) head_cycle = log->l_curr_cycle;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1626) head_block = log->l_curr_block;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1627)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1628) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1629) * Figure out the distance between the new head of the log
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1630) * and the tail. We want to write over any blocks beyond the
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1631) * head that we may have written just before the crash, but
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1632) * we don't want to overwrite the tail of the log.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1633) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1634) if (head_cycle == tail_cycle) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1635) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1636) * The tail is behind the head in the physical log,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1637) * so the distance from the head to the tail is the
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1638) * distance from the head to the end of the log plus
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1639) * the distance from the beginning of the log to the
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1640) * tail.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1641) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1642) if (XFS_IS_CORRUPT(log->l_mp,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1643) head_block < tail_block ||
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1644) head_block >= log->l_logBBsize))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1645) return -EFSCORRUPTED;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1646) tail_distance = tail_block + (log->l_logBBsize - head_block);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1647) } else {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1648) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1649) * The head is behind the tail in the physical log,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1650) * so the distance from the head to the tail is just
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1651) * the tail block minus the head block.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1652) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1653) if (XFS_IS_CORRUPT(log->l_mp,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1654) head_block >= tail_block ||
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1655) head_cycle != tail_cycle + 1))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1656) return -EFSCORRUPTED;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1657) tail_distance = tail_block - head_block;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1658) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1659)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1660) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1661) * If the head is right up against the tail, we can't clear
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1662) * anything.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1663) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1664) if (tail_distance <= 0) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1665) ASSERT(tail_distance == 0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1666) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1667) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1668)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1669) max_distance = XLOG_TOTAL_REC_SHIFT(log);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1670) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1671) * Take the smaller of the maximum amount of outstanding I/O
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1672) * we could have and the distance to the tail to clear out.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1673) * We take the smaller so that we don't overwrite the tail and
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1674) * we don't waste all day writing from the head to the tail
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1675) * for no reason.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1676) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1677) max_distance = min(max_distance, tail_distance);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1678)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1679) if ((head_block + max_distance) <= log->l_logBBsize) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1680) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1681) * We can stomp all the blocks we need to without
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1682) * wrapping around the end of the log. Just do it
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1683) * in a single write. Use the cycle number of the
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1684) * current cycle minus one so that the log will look like:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1685) * n ... | n - 1 ...
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1686) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1687) error = xlog_write_log_records(log, (head_cycle - 1),
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1688) head_block, max_distance, tail_cycle,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1689) tail_block);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1690) if (error)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1691) return error;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1692) } else {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1693) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1694) * We need to wrap around the end of the physical log in
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1695) * order to clear all the blocks. Do it in two separate
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1696) * I/Os. The first write should be from the head to the
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1697) * end of the physical log, and it should use the current
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1698) * cycle number minus one just like above.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1699) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1700) distance = log->l_logBBsize - head_block;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1701) error = xlog_write_log_records(log, (head_cycle - 1),
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1702) head_block, distance, tail_cycle,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1703) tail_block);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1704)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1705) if (error)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1706) return error;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1707)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1708) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1709) * Now write the blocks at the start of the physical log.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1710) * This writes the remainder of the blocks we want to clear.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1711) * It uses the current cycle number since we're now on the
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1712) * same cycle as the head so that we get:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1713) * n ... n ... | n - 1 ...
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1714) * ^^^^^ blocks we're writing
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1715) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1716) distance = max_distance - (log->l_logBBsize - head_block);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1717) error = xlog_write_log_records(log, head_cycle, 0, distance,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1718) tail_cycle, tail_block);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1719) if (error)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1720) return error;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1721) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1722)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1723) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1724) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1725)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1726) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1727) * Release the recovered intent item in the AIL that matches the given intent
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1728) * type and intent id.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1729) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1730) void
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1731) xlog_recover_release_intent(
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1732) struct xlog *log,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1733) unsigned short intent_type,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1734) uint64_t intent_id)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1735) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1736) struct xfs_ail_cursor cur;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1737) struct xfs_log_item *lip;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1738) struct xfs_ail *ailp = log->l_ailp;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1739)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1740) spin_lock(&ailp->ail_lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1741) for (lip = xfs_trans_ail_cursor_first(ailp, &cur, 0); lip != NULL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1742) lip = xfs_trans_ail_cursor_next(ailp, &cur)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1743) if (lip->li_type != intent_type)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1744) continue;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1745) if (!lip->li_ops->iop_match(lip, intent_id))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1746) continue;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1747)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1748) spin_unlock(&ailp->ail_lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1749) lip->li_ops->iop_release(lip);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1750) spin_lock(&ailp->ail_lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1751) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1752) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1753)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1754) xfs_trans_ail_cursor_done(&cur);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1755) spin_unlock(&ailp->ail_lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1756) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1757)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1758) /******************************************************************************
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1759) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1760) * Log recover routines
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1761) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1762) ******************************************************************************
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1763) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1764) static const struct xlog_recover_item_ops *xlog_recover_item_ops[] = {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1765) &xlog_buf_item_ops,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1766) &xlog_inode_item_ops,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1767) &xlog_dquot_item_ops,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1768) &xlog_quotaoff_item_ops,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1769) &xlog_icreate_item_ops,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1770) &xlog_efi_item_ops,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1771) &xlog_efd_item_ops,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1772) &xlog_rui_item_ops,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1773) &xlog_rud_item_ops,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1774) &xlog_cui_item_ops,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1775) &xlog_cud_item_ops,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1776) &xlog_bui_item_ops,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1777) &xlog_bud_item_ops,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1778) };
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1779)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1780) static const struct xlog_recover_item_ops *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1781) xlog_find_item_ops(
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1782) struct xlog_recover_item *item)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1783) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1784) unsigned int i;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1785)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1786) for (i = 0; i < ARRAY_SIZE(xlog_recover_item_ops); i++)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1787) if (ITEM_TYPE(item) == xlog_recover_item_ops[i]->item_type)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1788) return xlog_recover_item_ops[i];
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1789)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1790) return NULL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1791) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1792)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1793) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1794) * Sort the log items in the transaction.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1795) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1796) * The ordering constraints are defined by the inode allocation and unlink
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1797) * behaviour. The rules are:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1798) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1799) * 1. Every item is only logged once in a given transaction. Hence it
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1800) * represents the last logged state of the item. Hence ordering is
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1801) * dependent on the order in which operations need to be performed so
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1802) * required initial conditions are always met.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1803) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1804) * 2. Cancelled buffers are recorded in pass 1 in a separate table and
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1805) * there's nothing to replay from them so we can simply cull them
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1806) * from the transaction. However, we can't do that until after we've
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1807) * replayed all the other items because they may be dependent on the
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1808) * cancelled buffer and replaying the cancelled buffer can remove it
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1809) * form the cancelled buffer table. Hence they have tobe done last.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1810) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1811) * 3. Inode allocation buffers must be replayed before inode items that
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1812) * read the buffer and replay changes into it. For filesystems using the
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1813) * ICREATE transactions, this means XFS_LI_ICREATE objects need to get
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1814) * treated the same as inode allocation buffers as they create and
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1815) * initialise the buffers directly.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1816) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1817) * 4. Inode unlink buffers must be replayed after inode items are replayed.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1818) * This ensures that inodes are completely flushed to the inode buffer
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1819) * in a "free" state before we remove the unlinked inode list pointer.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1820) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1821) * Hence the ordering needs to be inode allocation buffers first, inode items
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1822) * second, inode unlink buffers third and cancelled buffers last.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1823) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1824) * But there's a problem with that - we can't tell an inode allocation buffer
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1825) * apart from a regular buffer, so we can't separate them. We can, however,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1826) * tell an inode unlink buffer from the others, and so we can separate them out
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1827) * from all the other buffers and move them to last.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1828) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1829) * Hence, 4 lists, in order from head to tail:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1830) * - buffer_list for all buffers except cancelled/inode unlink buffers
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1831) * - item_list for all non-buffer items
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1832) * - inode_buffer_list for inode unlink buffers
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1833) * - cancel_list for the cancelled buffers
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1834) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1835) * Note that we add objects to the tail of the lists so that first-to-last
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1836) * ordering is preserved within the lists. Adding objects to the head of the
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1837) * list means when we traverse from the head we walk them in last-to-first
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1838) * order. For cancelled buffers and inode unlink buffers this doesn't matter,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1839) * but for all other items there may be specific ordering that we need to
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1840) * preserve.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1841) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1842) STATIC int
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1843) xlog_recover_reorder_trans(
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1844) struct xlog *log,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1845) struct xlog_recover *trans,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1846) int pass)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1847) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1848) struct xlog_recover_item *item, *n;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1849) int error = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1850) LIST_HEAD(sort_list);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1851) LIST_HEAD(cancel_list);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1852) LIST_HEAD(buffer_list);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1853) LIST_HEAD(inode_buffer_list);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1854) LIST_HEAD(item_list);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1855)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1856) list_splice_init(&trans->r_itemq, &sort_list);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1857) list_for_each_entry_safe(item, n, &sort_list, ri_list) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1858) enum xlog_recover_reorder fate = XLOG_REORDER_ITEM_LIST;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1859)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1860) item->ri_ops = xlog_find_item_ops(item);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1861) if (!item->ri_ops) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1862) xfs_warn(log->l_mp,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1863) "%s: unrecognized type of log operation (%d)",
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1864) __func__, ITEM_TYPE(item));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1865) ASSERT(0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1866) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1867) * return the remaining items back to the transaction
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1868) * item list so they can be freed in caller.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1869) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1870) if (!list_empty(&sort_list))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1871) list_splice_init(&sort_list, &trans->r_itemq);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1872) error = -EFSCORRUPTED;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1873) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1874) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1875)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1876) if (item->ri_ops->reorder)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1877) fate = item->ri_ops->reorder(item);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1878)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1879) switch (fate) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1880) case XLOG_REORDER_BUFFER_LIST:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1881) list_move_tail(&item->ri_list, &buffer_list);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1882) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1883) case XLOG_REORDER_CANCEL_LIST:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1884) trace_xfs_log_recover_item_reorder_head(log,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1885) trans, item, pass);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1886) list_move(&item->ri_list, &cancel_list);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1887) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1888) case XLOG_REORDER_INODE_BUFFER_LIST:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1889) list_move(&item->ri_list, &inode_buffer_list);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1890) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1891) case XLOG_REORDER_ITEM_LIST:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1892) trace_xfs_log_recover_item_reorder_tail(log,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1893) trans, item, pass);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1894) list_move_tail(&item->ri_list, &item_list);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1895) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1896) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1897) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1898)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1899) ASSERT(list_empty(&sort_list));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1900) if (!list_empty(&buffer_list))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1901) list_splice(&buffer_list, &trans->r_itemq);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1902) if (!list_empty(&item_list))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1903) list_splice_tail(&item_list, &trans->r_itemq);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1904) if (!list_empty(&inode_buffer_list))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1905) list_splice_tail(&inode_buffer_list, &trans->r_itemq);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1906) if (!list_empty(&cancel_list))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1907) list_splice_tail(&cancel_list, &trans->r_itemq);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1908) return error;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1909) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1910)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1911) void
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1912) xlog_buf_readahead(
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1913) struct xlog *log,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1914) xfs_daddr_t blkno,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1915) uint len,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1916) const struct xfs_buf_ops *ops)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1917) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1918) if (!xlog_is_buffer_cancelled(log, blkno, len))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1919) xfs_buf_readahead(log->l_mp->m_ddev_targp, blkno, len, ops);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1920) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1921)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1922) STATIC int
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1923) xlog_recover_items_pass2(
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1924) struct xlog *log,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1925) struct xlog_recover *trans,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1926) struct list_head *buffer_list,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1927) struct list_head *item_list)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1928) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1929) struct xlog_recover_item *item;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1930) int error = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1931)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1932) list_for_each_entry(item, item_list, ri_list) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1933) trace_xfs_log_recover_item_recover(log, trans, item,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1934) XLOG_RECOVER_PASS2);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1935)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1936) if (item->ri_ops->commit_pass2)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1937) error = item->ri_ops->commit_pass2(log, buffer_list,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1938) item, trans->r_lsn);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1939) if (error)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1940) return error;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1941) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1942)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1943) return error;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1944) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1945)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1946) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1947) * Perform the transaction.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1948) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1949) * If the transaction modifies a buffer or inode, do it now. Otherwise,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1950) * EFIs and EFDs get queued up by adding entries into the AIL for them.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1951) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1952) STATIC int
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1953) xlog_recover_commit_trans(
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1954) struct xlog *log,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1955) struct xlog_recover *trans,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1956) int pass,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1957) struct list_head *buffer_list)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1958) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1959) int error = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1960) int items_queued = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1961) struct xlog_recover_item *item;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1962) struct xlog_recover_item *next;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1963) LIST_HEAD (ra_list);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1964) LIST_HEAD (done_list);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1965)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1966) #define XLOG_RECOVER_COMMIT_QUEUE_MAX 100
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1967)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1968) hlist_del_init(&trans->r_list);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1969)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1970) error = xlog_recover_reorder_trans(log, trans, pass);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1971) if (error)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1972) return error;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1973)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1974) list_for_each_entry_safe(item, next, &trans->r_itemq, ri_list) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1975) trace_xfs_log_recover_item_recover(log, trans, item, pass);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1976)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1977) switch (pass) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1978) case XLOG_RECOVER_PASS1:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1979) if (item->ri_ops->commit_pass1)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1980) error = item->ri_ops->commit_pass1(log, item);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1981) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1982) case XLOG_RECOVER_PASS2:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1983) if (item->ri_ops->ra_pass2)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1984) item->ri_ops->ra_pass2(log, item);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1985) list_move_tail(&item->ri_list, &ra_list);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1986) items_queued++;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1987) if (items_queued >= XLOG_RECOVER_COMMIT_QUEUE_MAX) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1988) error = xlog_recover_items_pass2(log, trans,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1989) buffer_list, &ra_list);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1990) list_splice_tail_init(&ra_list, &done_list);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1991) items_queued = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1992) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1993)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1994) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1995) default:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1996) ASSERT(0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1997) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1998)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1999) if (error)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2000) goto out;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2001) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2002)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2003) out:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2004) if (!list_empty(&ra_list)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2005) if (!error)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2006) error = xlog_recover_items_pass2(log, trans,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2007) buffer_list, &ra_list);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2008) list_splice_tail_init(&ra_list, &done_list);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2009) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2010)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2011) if (!list_empty(&done_list))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2012) list_splice_init(&done_list, &trans->r_itemq);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2013)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2014) return error;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2015) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2016)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2017) STATIC void
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2018) xlog_recover_add_item(
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2019) struct list_head *head)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2020) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2021) struct xlog_recover_item *item;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2022)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2023) item = kmem_zalloc(sizeof(struct xlog_recover_item), 0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2024) INIT_LIST_HEAD(&item->ri_list);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2025) list_add_tail(&item->ri_list, head);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2026) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2027)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2028) STATIC int
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2029) xlog_recover_add_to_cont_trans(
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2030) struct xlog *log,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2031) struct xlog_recover *trans,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2032) char *dp,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2033) int len)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2034) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2035) struct xlog_recover_item *item;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2036) char *ptr, *old_ptr;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2037) int old_len;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2038)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2039) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2040) * If the transaction is empty, the header was split across this and the
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2041) * previous record. Copy the rest of the header.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2042) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2043) if (list_empty(&trans->r_itemq)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2044) ASSERT(len <= sizeof(struct xfs_trans_header));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2045) if (len > sizeof(struct xfs_trans_header)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2046) xfs_warn(log->l_mp, "%s: bad header length", __func__);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2047) return -EFSCORRUPTED;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2048) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2049)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2050) xlog_recover_add_item(&trans->r_itemq);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2051) ptr = (char *)&trans->r_theader +
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2052) sizeof(struct xfs_trans_header) - len;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2053) memcpy(ptr, dp, len);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2054) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2055) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2056)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2057) /* take the tail entry */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2058) item = list_entry(trans->r_itemq.prev, struct xlog_recover_item,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2059) ri_list);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2060)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2061) old_ptr = item->ri_buf[item->ri_cnt-1].i_addr;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2062) old_len = item->ri_buf[item->ri_cnt-1].i_len;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2063)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2064) ptr = krealloc(old_ptr, len + old_len, GFP_KERNEL | __GFP_NOFAIL);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2065) memcpy(&ptr[old_len], dp, len);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2066) item->ri_buf[item->ri_cnt-1].i_len += len;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2067) item->ri_buf[item->ri_cnt-1].i_addr = ptr;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2068) trace_xfs_log_recover_item_add_cont(log, trans, item, 0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2069) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2070) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2071)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2072) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2073) * The next region to add is the start of a new region. It could be
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2074) * a whole region or it could be the first part of a new region. Because
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2075) * of this, the assumption here is that the type and size fields of all
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2076) * format structures fit into the first 32 bits of the structure.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2077) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2078) * This works because all regions must be 32 bit aligned. Therefore, we
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2079) * either have both fields or we have neither field. In the case we have
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2080) * neither field, the data part of the region is zero length. We only have
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2081) * a log_op_header and can throw away the header since a new one will appear
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2082) * later. If we have at least 4 bytes, then we can determine how many regions
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2083) * will appear in the current log item.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2084) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2085) STATIC int
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2086) xlog_recover_add_to_trans(
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2087) struct xlog *log,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2088) struct xlog_recover *trans,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2089) char *dp,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2090) int len)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2091) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2092) struct xfs_inode_log_format *in_f; /* any will do */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2093) struct xlog_recover_item *item;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2094) char *ptr;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2095)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2096) if (!len)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2097) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2098) if (list_empty(&trans->r_itemq)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2099) /* we need to catch log corruptions here */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2100) if (*(uint *)dp != XFS_TRANS_HEADER_MAGIC) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2101) xfs_warn(log->l_mp, "%s: bad header magic number",
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2102) __func__);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2103) ASSERT(0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2104) return -EFSCORRUPTED;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2105) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2106)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2107) if (len > sizeof(struct xfs_trans_header)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2108) xfs_warn(log->l_mp, "%s: bad header length", __func__);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2109) ASSERT(0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2110) return -EFSCORRUPTED;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2111) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2112)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2113) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2114) * The transaction header can be arbitrarily split across op
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2115) * records. If we don't have the whole thing here, copy what we
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2116) * do have and handle the rest in the next record.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2117) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2118) if (len == sizeof(struct xfs_trans_header))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2119) xlog_recover_add_item(&trans->r_itemq);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2120) memcpy(&trans->r_theader, dp, len);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2121) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2122) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2123)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2124) ptr = kmem_alloc(len, 0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2125) memcpy(ptr, dp, len);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2126) in_f = (struct xfs_inode_log_format *)ptr;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2127)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2128) /* take the tail entry */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2129) item = list_entry(trans->r_itemq.prev, struct xlog_recover_item,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2130) ri_list);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2131) if (item->ri_total != 0 &&
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2132) item->ri_total == item->ri_cnt) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2133) /* tail item is in use, get a new one */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2134) xlog_recover_add_item(&trans->r_itemq);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2135) item = list_entry(trans->r_itemq.prev,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2136) struct xlog_recover_item, ri_list);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2137) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2138)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2139) if (item->ri_total == 0) { /* first region to be added */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2140) if (in_f->ilf_size == 0 ||
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2141) in_f->ilf_size > XLOG_MAX_REGIONS_IN_ITEM) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2142) xfs_warn(log->l_mp,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2143) "bad number of regions (%d) in inode log format",
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2144) in_f->ilf_size);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2145) ASSERT(0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2146) kmem_free(ptr);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2147) return -EFSCORRUPTED;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2148) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2149)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2150) item->ri_total = in_f->ilf_size;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2151) item->ri_buf =
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2152) kmem_zalloc(item->ri_total * sizeof(xfs_log_iovec_t),
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2153) 0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2154) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2155)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2156) if (item->ri_total <= item->ri_cnt) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2157) xfs_warn(log->l_mp,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2158) "log item region count (%d) overflowed size (%d)",
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2159) item->ri_cnt, item->ri_total);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2160) ASSERT(0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2161) kmem_free(ptr);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2162) return -EFSCORRUPTED;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2163) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2164)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2165) /* Description region is ri_buf[0] */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2166) item->ri_buf[item->ri_cnt].i_addr = ptr;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2167) item->ri_buf[item->ri_cnt].i_len = len;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2168) item->ri_cnt++;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2169) trace_xfs_log_recover_item_add(log, trans, item, 0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2170) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2171) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2172)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2173) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2174) * Free up any resources allocated by the transaction
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2175) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2176) * Remember that EFIs, EFDs, and IUNLINKs are handled later.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2177) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2178) STATIC void
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2179) xlog_recover_free_trans(
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2180) struct xlog_recover *trans)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2181) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2182) struct xlog_recover_item *item, *n;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2183) int i;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2184)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2185) hlist_del_init(&trans->r_list);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2186)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2187) list_for_each_entry_safe(item, n, &trans->r_itemq, ri_list) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2188) /* Free the regions in the item. */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2189) list_del(&item->ri_list);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2190) for (i = 0; i < item->ri_cnt; i++)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2191) kmem_free(item->ri_buf[i].i_addr);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2192) /* Free the item itself */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2193) kmem_free(item->ri_buf);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2194) kmem_free(item);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2195) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2196) /* Free the transaction recover structure */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2197) kmem_free(trans);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2198) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2199)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2200) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2201) * On error or completion, trans is freed.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2202) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2203) STATIC int
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2204) xlog_recovery_process_trans(
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2205) struct xlog *log,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2206) struct xlog_recover *trans,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2207) char *dp,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2208) unsigned int len,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2209) unsigned int flags,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2210) int pass,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2211) struct list_head *buffer_list)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2212) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2213) int error = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2214) bool freeit = false;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2215)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2216) /* mask off ophdr transaction container flags */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2217) flags &= ~XLOG_END_TRANS;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2218) if (flags & XLOG_WAS_CONT_TRANS)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2219) flags &= ~XLOG_CONTINUE_TRANS;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2220)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2221) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2222) * Callees must not free the trans structure. We'll decide if we need to
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2223) * free it or not based on the operation being done and it's result.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2224) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2225) switch (flags) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2226) /* expected flag values */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2227) case 0:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2228) case XLOG_CONTINUE_TRANS:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2229) error = xlog_recover_add_to_trans(log, trans, dp, len);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2230) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2231) case XLOG_WAS_CONT_TRANS:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2232) error = xlog_recover_add_to_cont_trans(log, trans, dp, len);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2233) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2234) case XLOG_COMMIT_TRANS:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2235) error = xlog_recover_commit_trans(log, trans, pass,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2236) buffer_list);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2237) /* success or fail, we are now done with this transaction. */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2238) freeit = true;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2239) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2240)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2241) /* unexpected flag values */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2242) case XLOG_UNMOUNT_TRANS:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2243) /* just skip trans */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2244) xfs_warn(log->l_mp, "%s: Unmount LR", __func__);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2245) freeit = true;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2246) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2247) case XLOG_START_TRANS:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2248) default:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2249) xfs_warn(log->l_mp, "%s: bad flag 0x%x", __func__, flags);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2250) ASSERT(0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2251) error = -EFSCORRUPTED;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2252) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2253) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2254) if (error || freeit)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2255) xlog_recover_free_trans(trans);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2256) return error;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2257) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2258)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2259) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2260) * Lookup the transaction recovery structure associated with the ID in the
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2261) * current ophdr. If the transaction doesn't exist and the start flag is set in
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2262) * the ophdr, then allocate a new transaction for future ID matches to find.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2263) * Either way, return what we found during the lookup - an existing transaction
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2264) * or nothing.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2265) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2266) STATIC struct xlog_recover *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2267) xlog_recover_ophdr_to_trans(
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2268) struct hlist_head rhash[],
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2269) struct xlog_rec_header *rhead,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2270) struct xlog_op_header *ohead)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2271) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2272) struct xlog_recover *trans;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2273) xlog_tid_t tid;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2274) struct hlist_head *rhp;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2275)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2276) tid = be32_to_cpu(ohead->oh_tid);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2277) rhp = &rhash[XLOG_RHASH(tid)];
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2278) hlist_for_each_entry(trans, rhp, r_list) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2279) if (trans->r_log_tid == tid)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2280) return trans;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2281) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2282)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2283) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2284) * skip over non-start transaction headers - we could be
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2285) * processing slack space before the next transaction starts
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2286) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2287) if (!(ohead->oh_flags & XLOG_START_TRANS))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2288) return NULL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2289)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2290) ASSERT(be32_to_cpu(ohead->oh_len) == 0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2291)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2292) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2293) * This is a new transaction so allocate a new recovery container to
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2294) * hold the recovery ops that will follow.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2295) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2296) trans = kmem_zalloc(sizeof(struct xlog_recover), 0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2297) trans->r_log_tid = tid;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2298) trans->r_lsn = be64_to_cpu(rhead->h_lsn);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2299) INIT_LIST_HEAD(&trans->r_itemq);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2300) INIT_HLIST_NODE(&trans->r_list);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2301) hlist_add_head(&trans->r_list, rhp);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2302)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2303) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2304) * Nothing more to do for this ophdr. Items to be added to this new
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2305) * transaction will be in subsequent ophdr containers.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2306) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2307) return NULL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2308) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2309)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2310) STATIC int
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2311) xlog_recover_process_ophdr(
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2312) struct xlog *log,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2313) struct hlist_head rhash[],
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2314) struct xlog_rec_header *rhead,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2315) struct xlog_op_header *ohead,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2316) char *dp,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2317) char *end,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2318) int pass,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2319) struct list_head *buffer_list)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2320) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2321) struct xlog_recover *trans;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2322) unsigned int len;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2323) int error;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2324)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2325) /* Do we understand who wrote this op? */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2326) if (ohead->oh_clientid != XFS_TRANSACTION &&
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2327) ohead->oh_clientid != XFS_LOG) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2328) xfs_warn(log->l_mp, "%s: bad clientid 0x%x",
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2329) __func__, ohead->oh_clientid);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2330) ASSERT(0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2331) return -EFSCORRUPTED;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2332) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2333)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2334) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2335) * Check the ophdr contains all the data it is supposed to contain.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2336) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2337) len = be32_to_cpu(ohead->oh_len);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2338) if (dp + len > end) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2339) xfs_warn(log->l_mp, "%s: bad length 0x%x", __func__, len);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2340) WARN_ON(1);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2341) return -EFSCORRUPTED;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2342) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2343)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2344) trans = xlog_recover_ophdr_to_trans(rhash, rhead, ohead);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2345) if (!trans) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2346) /* nothing to do, so skip over this ophdr */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2347) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2348) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2349)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2350) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2351) * The recovered buffer queue is drained only once we know that all
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2352) * recovery items for the current LSN have been processed. This is
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2353) * required because:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2354) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2355) * - Buffer write submission updates the metadata LSN of the buffer.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2356) * - Log recovery skips items with a metadata LSN >= the current LSN of
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2357) * the recovery item.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2358) * - Separate recovery items against the same metadata buffer can share
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2359) * a current LSN. I.e., consider that the LSN of a recovery item is
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2360) * defined as the starting LSN of the first record in which its
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2361) * transaction appears, that a record can hold multiple transactions,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2362) * and/or that a transaction can span multiple records.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2363) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2364) * In other words, we are allowed to submit a buffer from log recovery
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2365) * once per current LSN. Otherwise, we may incorrectly skip recovery
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2366) * items and cause corruption.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2367) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2368) * We don't know up front whether buffers are updated multiple times per
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2369) * LSN. Therefore, track the current LSN of each commit log record as it
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2370) * is processed and drain the queue when it changes. Use commit records
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2371) * because they are ordered correctly by the logging code.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2372) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2373) if (log->l_recovery_lsn != trans->r_lsn &&
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2374) ohead->oh_flags & XLOG_COMMIT_TRANS) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2375) error = xfs_buf_delwri_submit(buffer_list);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2376) if (error)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2377) return error;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2378) log->l_recovery_lsn = trans->r_lsn;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2379) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2380)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2381) return xlog_recovery_process_trans(log, trans, dp, len,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2382) ohead->oh_flags, pass, buffer_list);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2383) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2384)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2385) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2386) * There are two valid states of the r_state field. 0 indicates that the
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2387) * transaction structure is in a normal state. We have either seen the
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2388) * start of the transaction or the last operation we added was not a partial
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2389) * operation. If the last operation we added to the transaction was a
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2390) * partial operation, we need to mark r_state with XLOG_WAS_CONT_TRANS.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2391) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2392) * NOTE: skip LRs with 0 data length.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2393) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2394) STATIC int
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2395) xlog_recover_process_data(
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2396) struct xlog *log,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2397) struct hlist_head rhash[],
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2398) struct xlog_rec_header *rhead,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2399) char *dp,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2400) int pass,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2401) struct list_head *buffer_list)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2402) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2403) struct xlog_op_header *ohead;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2404) char *end;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2405) int num_logops;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2406) int error;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2407)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2408) end = dp + be32_to_cpu(rhead->h_len);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2409) num_logops = be32_to_cpu(rhead->h_num_logops);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2410)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2411) /* check the log format matches our own - else we can't recover */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2412) if (xlog_header_check_recover(log->l_mp, rhead))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2413) return -EIO;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2414)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2415) trace_xfs_log_recover_record(log, rhead, pass);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2416) while ((dp < end) && num_logops) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2417)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2418) ohead = (struct xlog_op_header *)dp;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2419) dp += sizeof(*ohead);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2420) ASSERT(dp <= end);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2421)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2422) /* errors will abort recovery */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2423) error = xlog_recover_process_ophdr(log, rhash, rhead, ohead,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2424) dp, end, pass, buffer_list);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2425) if (error)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2426) return error;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2427)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2428) dp += be32_to_cpu(ohead->oh_len);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2429) num_logops--;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2430) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2431) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2432) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2433)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2434) /* Take all the collected deferred ops and finish them in order. */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2435) static int
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2436) xlog_finish_defer_ops(
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2437) struct xfs_mount *mp,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2438) struct list_head *capture_list)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2439) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2440) struct xfs_defer_capture *dfc, *next;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2441) struct xfs_trans *tp;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2442) struct xfs_inode *ip;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2443) int error = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2444)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2445) list_for_each_entry_safe(dfc, next, capture_list, dfc_list) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2446) struct xfs_trans_res resv;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2447)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2448) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2449) * Create a new transaction reservation from the captured
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2450) * information. Set logcount to 1 to force the new transaction
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2451) * to regrant every roll so that we can make forward progress
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2452) * in recovery no matter how full the log might be.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2453) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2454) resv.tr_logres = dfc->dfc_logres;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2455) resv.tr_logcount = 1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2456) resv.tr_logflags = XFS_TRANS_PERM_LOG_RES;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2457)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2458) error = xfs_trans_alloc(mp, &resv, dfc->dfc_blkres,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2459) dfc->dfc_rtxres, XFS_TRANS_RESERVE, &tp);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2460) if (error)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2461) return error;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2462)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2463) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2464) * Transfer to this new transaction all the dfops we captured
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2465) * from recovering a single intent item.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2466) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2467) list_del_init(&dfc->dfc_list);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2468) xfs_defer_ops_continue(dfc, tp, &ip);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2469)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2470) error = xfs_trans_commit(tp);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2471) if (ip) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2472) xfs_iunlock(ip, XFS_ILOCK_EXCL);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2473) xfs_irele(ip);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2474) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2475) if (error)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2476) return error;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2477) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2478)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2479) ASSERT(list_empty(capture_list));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2480) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2481) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2482)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2483) /* Release all the captured defer ops and capture structures in this list. */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2484) static void
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2485) xlog_abort_defer_ops(
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2486) struct xfs_mount *mp,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2487) struct list_head *capture_list)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2488) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2489) struct xfs_defer_capture *dfc;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2490) struct xfs_defer_capture *next;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2491)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2492) list_for_each_entry_safe(dfc, next, capture_list, dfc_list) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2493) list_del_init(&dfc->dfc_list);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2494) xfs_defer_ops_release(mp, dfc);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2495) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2496) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2497) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2498) * When this is called, all of the log intent items which did not have
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2499) * corresponding log done items should be in the AIL. What we do now
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2500) * is update the data structures associated with each one.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2501) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2502) * Since we process the log intent items in normal transactions, they
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2503) * will be removed at some point after the commit. This prevents us
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2504) * from just walking down the list processing each one. We'll use a
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2505) * flag in the intent item to skip those that we've already processed
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2506) * and use the AIL iteration mechanism's generation count to try to
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2507) * speed this up at least a bit.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2508) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2509) * When we start, we know that the intents are the only things in the
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2510) * AIL. As we process them, however, other items are added to the
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2511) * AIL.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2512) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2513) STATIC int
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2514) xlog_recover_process_intents(
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2515) struct xlog *log)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2516) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2517) LIST_HEAD(capture_list);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2518) struct xfs_ail_cursor cur;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2519) struct xfs_log_item *lip;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2520) struct xfs_ail *ailp;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2521) int error = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2522) #if defined(DEBUG) || defined(XFS_WARN)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2523) xfs_lsn_t last_lsn;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2524) #endif
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2525)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2526) ailp = log->l_ailp;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2527) spin_lock(&ailp->ail_lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2528) #if defined(DEBUG) || defined(XFS_WARN)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2529) last_lsn = xlog_assign_lsn(log->l_curr_cycle, log->l_curr_block);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2530) #endif
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2531) for (lip = xfs_trans_ail_cursor_first(ailp, &cur, 0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2532) lip != NULL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2533) lip = xfs_trans_ail_cursor_next(ailp, &cur)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2534) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2535) * We're done when we see something other than an intent.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2536) * There should be no intents left in the AIL now.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2537) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2538) if (!xlog_item_is_intent(lip)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2539) #ifdef DEBUG
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2540) for (; lip; lip = xfs_trans_ail_cursor_next(ailp, &cur))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2541) ASSERT(!xlog_item_is_intent(lip));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2542) #endif
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2543) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2544) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2545)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2546) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2547) * We should never see a redo item with a LSN higher than
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2548) * the last transaction we found in the log at the start
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2549) * of recovery.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2550) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2551) ASSERT(XFS_LSN_CMP(last_lsn, lip->li_lsn) >= 0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2552)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2553) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2554) * NOTE: If your intent processing routine can create more
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2555) * deferred ops, you /must/ attach them to the capture list in
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2556) * the recover routine or else those subsequent intents will be
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2557) * replayed in the wrong order!
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2558) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2559) spin_unlock(&ailp->ail_lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2560) error = lip->li_ops->iop_recover(lip, &capture_list);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2561) spin_lock(&ailp->ail_lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2562) if (error)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2563) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2564) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2565)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2566) xfs_trans_ail_cursor_done(&cur);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2567) spin_unlock(&ailp->ail_lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2568) if (error)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2569) goto err;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2570)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2571) error = xlog_finish_defer_ops(log->l_mp, &capture_list);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2572) if (error)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2573) goto err;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2574)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2575) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2576) err:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2577) xlog_abort_defer_ops(log->l_mp, &capture_list);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2578) return error;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2579) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2580)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2581) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2582) * A cancel occurs when the mount has failed and we're bailing out.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2583) * Release all pending log intent items so they don't pin the AIL.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2584) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2585) STATIC void
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2586) xlog_recover_cancel_intents(
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2587) struct xlog *log)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2588) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2589) struct xfs_log_item *lip;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2590) struct xfs_ail_cursor cur;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2591) struct xfs_ail *ailp;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2592)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2593) ailp = log->l_ailp;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2594) spin_lock(&ailp->ail_lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2595) lip = xfs_trans_ail_cursor_first(ailp, &cur, 0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2596) while (lip != NULL) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2597) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2598) * We're done when we see something other than an intent.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2599) * There should be no intents left in the AIL now.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2600) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2601) if (!xlog_item_is_intent(lip)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2602) #ifdef DEBUG
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2603) for (; lip; lip = xfs_trans_ail_cursor_next(ailp, &cur))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2604) ASSERT(!xlog_item_is_intent(lip));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2605) #endif
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2606) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2607) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2608)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2609) spin_unlock(&ailp->ail_lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2610) lip->li_ops->iop_release(lip);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2611) spin_lock(&ailp->ail_lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2612) lip = xfs_trans_ail_cursor_next(ailp, &cur);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2613) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2614)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2615) xfs_trans_ail_cursor_done(&cur);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2616) spin_unlock(&ailp->ail_lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2617) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2618)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2619) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2620) * This routine performs a transaction to null out a bad inode pointer
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2621) * in an agi unlinked inode hash bucket.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2622) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2623) STATIC void
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2624) xlog_recover_clear_agi_bucket(
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2625) xfs_mount_t *mp,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2626) xfs_agnumber_t agno,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2627) int bucket)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2628) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2629) xfs_trans_t *tp;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2630) xfs_agi_t *agi;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2631) xfs_buf_t *agibp;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2632) int offset;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2633) int error;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2634)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2635) error = xfs_trans_alloc(mp, &M_RES(mp)->tr_clearagi, 0, 0, 0, &tp);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2636) if (error)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2637) goto out_error;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2638)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2639) error = xfs_read_agi(mp, tp, agno, &agibp);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2640) if (error)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2641) goto out_abort;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2642)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2643) agi = agibp->b_addr;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2644) agi->agi_unlinked[bucket] = cpu_to_be32(NULLAGINO);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2645) offset = offsetof(xfs_agi_t, agi_unlinked) +
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2646) (sizeof(xfs_agino_t) * bucket);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2647) xfs_trans_log_buf(tp, agibp, offset,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2648) (offset + sizeof(xfs_agino_t) - 1));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2649)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2650) error = xfs_trans_commit(tp);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2651) if (error)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2652) goto out_error;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2653) return;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2654)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2655) out_abort:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2656) xfs_trans_cancel(tp);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2657) out_error:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2658) xfs_warn(mp, "%s: failed to clear agi %d. Continuing.", __func__, agno);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2659) return;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2660) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2661)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2662) STATIC xfs_agino_t
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2663) xlog_recover_process_one_iunlink(
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2664) struct xfs_mount *mp,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2665) xfs_agnumber_t agno,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2666) xfs_agino_t agino,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2667) int bucket)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2668) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2669) struct xfs_buf *ibp;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2670) struct xfs_dinode *dip;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2671) struct xfs_inode *ip;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2672) xfs_ino_t ino;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2673) int error;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2674)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2675) ino = XFS_AGINO_TO_INO(mp, agno, agino);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2676) error = xfs_iget(mp, NULL, ino, 0, 0, &ip);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2677) if (error)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2678) goto fail;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2679)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2680) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2681) * Get the on disk inode to find the next inode in the bucket.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2682) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2683) error = xfs_imap_to_bp(mp, NULL, &ip->i_imap, &dip, &ibp, 0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2684) if (error)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2685) goto fail_iput;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2686)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2687) xfs_iflags_clear(ip, XFS_IRECOVERY);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2688) ASSERT(VFS_I(ip)->i_nlink == 0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2689) ASSERT(VFS_I(ip)->i_mode != 0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2690)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2691) /* setup for the next pass */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2692) agino = be32_to_cpu(dip->di_next_unlinked);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2693) xfs_buf_relse(ibp);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2694)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2695) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2696) * Prevent any DMAPI event from being sent when the reference on
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2697) * the inode is dropped.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2698) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2699) ip->i_d.di_dmevmask = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2700)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2701) xfs_irele(ip);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2702) return agino;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2703)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2704) fail_iput:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2705) xfs_irele(ip);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2706) fail:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2707) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2708) * We can't read in the inode this bucket points to, or this inode
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2709) * is messed up. Just ditch this bucket of inodes. We will lose
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2710) * some inodes and space, but at least we won't hang.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2711) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2712) * Call xlog_recover_clear_agi_bucket() to perform a transaction to
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2713) * clear the inode pointer in the bucket.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2714) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2715) xlog_recover_clear_agi_bucket(mp, agno, bucket);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2716) return NULLAGINO;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2717) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2718)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2719) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2720) * Recover AGI unlinked lists
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2721) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2722) * This is called during recovery to process any inodes which we unlinked but
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2723) * not freed when the system crashed. These inodes will be on the lists in the
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2724) * AGI blocks. What we do here is scan all the AGIs and fully truncate and free
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2725) * any inodes found on the lists. Each inode is removed from the lists when it
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2726) * has been fully truncated and is freed. The freeing of the inode and its
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2727) * removal from the list must be atomic.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2728) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2729) * If everything we touch in the agi processing loop is already in memory, this
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2730) * loop can hold the cpu for a long time. It runs without lock contention,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2731) * memory allocation contention, the need wait for IO, etc, and so will run
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2732) * until we either run out of inodes to process, run low on memory or we run out
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2733) * of log space.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2734) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2735) * This behaviour is bad for latency on single CPU and non-preemptible kernels,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2736) * and can prevent other filesytem work (such as CIL pushes) from running. This
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2737) * can lead to deadlocks if the recovery process runs out of log reservation
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2738) * space. Hence we need to yield the CPU when there is other kernel work
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2739) * scheduled on this CPU to ensure other scheduled work can run without undue
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2740) * latency.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2741) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2742) STATIC void
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2743) xlog_recover_process_iunlinks(
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2744) struct xlog *log)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2745) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2746) xfs_mount_t *mp;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2747) xfs_agnumber_t agno;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2748) xfs_agi_t *agi;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2749) xfs_buf_t *agibp;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2750) xfs_agino_t agino;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2751) int bucket;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2752) int error;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2753)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2754) mp = log->l_mp;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2755)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2756) for (agno = 0; agno < mp->m_sb.sb_agcount; agno++) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2757) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2758) * Find the agi for this ag.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2759) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2760) error = xfs_read_agi(mp, NULL, agno, &agibp);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2761) if (error) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2762) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2763) * AGI is b0rked. Don't process it.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2764) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2765) * We should probably mark the filesystem as corrupt
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2766) * after we've recovered all the ag's we can....
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2767) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2768) continue;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2769) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2770) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2771) * Unlock the buffer so that it can be acquired in the normal
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2772) * course of the transaction to truncate and free each inode.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2773) * Because we are not racing with anyone else here for the AGI
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2774) * buffer, we don't even need to hold it locked to read the
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2775) * initial unlinked bucket entries out of the buffer. We keep
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2776) * buffer reference though, so that it stays pinned in memory
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2777) * while we need the buffer.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2778) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2779) agi = agibp->b_addr;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2780) xfs_buf_unlock(agibp);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2781)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2782) for (bucket = 0; bucket < XFS_AGI_UNLINKED_BUCKETS; bucket++) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2783) agino = be32_to_cpu(agi->agi_unlinked[bucket]);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2784) while (agino != NULLAGINO) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2785) agino = xlog_recover_process_one_iunlink(mp,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2786) agno, agino, bucket);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2787) cond_resched();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2788) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2789) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2790) xfs_buf_rele(agibp);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2791) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2792) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2793)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2794) STATIC void
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2795) xlog_unpack_data(
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2796) struct xlog_rec_header *rhead,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2797) char *dp,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2798) struct xlog *log)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2799) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2800) int i, j, k;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2801)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2802) for (i = 0; i < BTOBB(be32_to_cpu(rhead->h_len)) &&
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2803) i < (XLOG_HEADER_CYCLE_SIZE / BBSIZE); i++) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2804) *(__be32 *)dp = *(__be32 *)&rhead->h_cycle_data[i];
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2805) dp += BBSIZE;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2806) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2807)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2808) if (xfs_sb_version_haslogv2(&log->l_mp->m_sb)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2809) xlog_in_core_2_t *xhdr = (xlog_in_core_2_t *)rhead;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2810) for ( ; i < BTOBB(be32_to_cpu(rhead->h_len)); i++) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2811) j = i / (XLOG_HEADER_CYCLE_SIZE / BBSIZE);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2812) k = i % (XLOG_HEADER_CYCLE_SIZE / BBSIZE);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2813) *(__be32 *)dp = xhdr[j].hic_xheader.xh_cycle_data[k];
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2814) dp += BBSIZE;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2815) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2816) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2817) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2818)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2819) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2820) * CRC check, unpack and process a log record.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2821) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2822) STATIC int
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2823) xlog_recover_process(
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2824) struct xlog *log,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2825) struct hlist_head rhash[],
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2826) struct xlog_rec_header *rhead,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2827) char *dp,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2828) int pass,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2829) struct list_head *buffer_list)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2830) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2831) __le32 old_crc = rhead->h_crc;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2832) __le32 crc;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2833)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2834) crc = xlog_cksum(log, rhead, dp, be32_to_cpu(rhead->h_len));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2835)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2836) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2837) * Nothing else to do if this is a CRC verification pass. Just return
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2838) * if this a record with a non-zero crc. Unfortunately, mkfs always
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2839) * sets old_crc to 0 so we must consider this valid even on v5 supers.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2840) * Otherwise, return EFSBADCRC on failure so the callers up the stack
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2841) * know precisely what failed.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2842) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2843) if (pass == XLOG_RECOVER_CRCPASS) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2844) if (old_crc && crc != old_crc)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2845) return -EFSBADCRC;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2846) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2847) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2848)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2849) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2850) * We're in the normal recovery path. Issue a warning if and only if the
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2851) * CRC in the header is non-zero. This is an advisory warning and the
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2852) * zero CRC check prevents warnings from being emitted when upgrading
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2853) * the kernel from one that does not add CRCs by default.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2854) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2855) if (crc != old_crc) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2856) if (old_crc || xfs_sb_version_hascrc(&log->l_mp->m_sb)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2857) xfs_alert(log->l_mp,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2858) "log record CRC mismatch: found 0x%x, expected 0x%x.",
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2859) le32_to_cpu(old_crc),
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2860) le32_to_cpu(crc));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2861) xfs_hex_dump(dp, 32);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2862) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2863)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2864) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2865) * If the filesystem is CRC enabled, this mismatch becomes a
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2866) * fatal log corruption failure.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2867) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2868) if (xfs_sb_version_hascrc(&log->l_mp->m_sb)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2869) XFS_ERROR_REPORT(__func__, XFS_ERRLEVEL_LOW, log->l_mp);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2870) return -EFSCORRUPTED;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2871) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2872) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2873)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2874) xlog_unpack_data(rhead, dp, log);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2875)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2876) return xlog_recover_process_data(log, rhash, rhead, dp, pass,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2877) buffer_list);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2878) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2879)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2880) STATIC int
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2881) xlog_valid_rec_header(
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2882) struct xlog *log,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2883) struct xlog_rec_header *rhead,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2884) xfs_daddr_t blkno,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2885) int bufsize)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2886) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2887) int hlen;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2888)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2889) if (XFS_IS_CORRUPT(log->l_mp,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2890) rhead->h_magicno != cpu_to_be32(XLOG_HEADER_MAGIC_NUM)))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2891) return -EFSCORRUPTED;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2892) if (XFS_IS_CORRUPT(log->l_mp,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2893) (!rhead->h_version ||
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2894) (be32_to_cpu(rhead->h_version) &
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2895) (~XLOG_VERSION_OKBITS))))) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2896) xfs_warn(log->l_mp, "%s: unrecognised log version (%d).",
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2897) __func__, be32_to_cpu(rhead->h_version));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2898) return -EFSCORRUPTED;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2899) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2900)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2901) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2902) * LR body must have data (or it wouldn't have been written)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2903) * and h_len must not be greater than LR buffer size.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2904) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2905) hlen = be32_to_cpu(rhead->h_len);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2906) if (XFS_IS_CORRUPT(log->l_mp, hlen <= 0 || hlen > bufsize))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2907) return -EFSCORRUPTED;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2908)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2909) if (XFS_IS_CORRUPT(log->l_mp,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2910) blkno > log->l_logBBsize || blkno > INT_MAX))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2911) return -EFSCORRUPTED;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2912) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2913) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2914)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2915) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2916) * Read the log from tail to head and process the log records found.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2917) * Handle the two cases where the tail and head are in the same cycle
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2918) * and where the active portion of the log wraps around the end of
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2919) * the physical log separately. The pass parameter is passed through
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2920) * to the routines called to process the data and is not looked at
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2921) * here.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2922) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2923) STATIC int
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2924) xlog_do_recovery_pass(
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2925) struct xlog *log,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2926) xfs_daddr_t head_blk,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2927) xfs_daddr_t tail_blk,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2928) int pass,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2929) xfs_daddr_t *first_bad) /* out: first bad log rec */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2930) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2931) xlog_rec_header_t *rhead;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2932) xfs_daddr_t blk_no, rblk_no;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2933) xfs_daddr_t rhead_blk;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2934) char *offset;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2935) char *hbp, *dbp;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2936) int error = 0, h_size, h_len;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2937) int error2 = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2938) int bblks, split_bblks;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2939) int hblks, split_hblks, wrapped_hblks;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2940) int i;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2941) struct hlist_head rhash[XLOG_RHASH_SIZE];
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2942) LIST_HEAD (buffer_list);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2943)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2944) ASSERT(head_blk != tail_blk);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2945) blk_no = rhead_blk = tail_blk;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2946)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2947) for (i = 0; i < XLOG_RHASH_SIZE; i++)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2948) INIT_HLIST_HEAD(&rhash[i]);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2949)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2950) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2951) * Read the header of the tail block and get the iclog buffer size from
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2952) * h_size. Use this to tell how many sectors make up the log header.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2953) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2954) if (xfs_sb_version_haslogv2(&log->l_mp->m_sb)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2955) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2956) * When using variable length iclogs, read first sector of
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2957) * iclog header and extract the header size from it. Get a
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2958) * new hbp that is the correct size.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2959) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2960) hbp = xlog_alloc_buffer(log, 1);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2961) if (!hbp)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2962) return -ENOMEM;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2963)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2964) error = xlog_bread(log, tail_blk, 1, hbp, &offset);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2965) if (error)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2966) goto bread_err1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2967)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2968) rhead = (xlog_rec_header_t *)offset;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2969)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2970) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2971) * xfsprogs has a bug where record length is based on lsunit but
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2972) * h_size (iclog size) is hardcoded to 32k. Now that we
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2973) * unconditionally CRC verify the unmount record, this means the
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2974) * log buffer can be too small for the record and cause an
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2975) * overrun.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2976) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2977) * Detect this condition here. Use lsunit for the buffer size as
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2978) * long as this looks like the mkfs case. Otherwise, return an
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2979) * error to avoid a buffer overrun.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2980) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2981) h_size = be32_to_cpu(rhead->h_size);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2982) h_len = be32_to_cpu(rhead->h_len);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2983) if (h_len > h_size && h_len <= log->l_mp->m_logbsize &&
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2984) rhead->h_num_logops == cpu_to_be32(1)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2985) xfs_warn(log->l_mp,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2986) "invalid iclog size (%d bytes), using lsunit (%d bytes)",
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2987) h_size, log->l_mp->m_logbsize);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2988) h_size = log->l_mp->m_logbsize;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2989) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2990)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2991) error = xlog_valid_rec_header(log, rhead, tail_blk, h_size);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2992) if (error)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2993) goto bread_err1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2994)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2995) hblks = xlog_logrec_hblks(log, rhead);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2996) if (hblks != 1) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2997) kmem_free(hbp);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2998) hbp = xlog_alloc_buffer(log, hblks);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2999) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3000) } else {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3001) ASSERT(log->l_sectBBsize == 1);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3002) hblks = 1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3003) hbp = xlog_alloc_buffer(log, 1);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3004) h_size = XLOG_BIG_RECORD_BSIZE;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3005) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3006)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3007) if (!hbp)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3008) return -ENOMEM;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3009) dbp = xlog_alloc_buffer(log, BTOBB(h_size));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3010) if (!dbp) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3011) kmem_free(hbp);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3012) return -ENOMEM;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3013) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3014)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3015) memset(rhash, 0, sizeof(rhash));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3016) if (tail_blk > head_blk) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3017) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3018) * Perform recovery around the end of the physical log.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3019) * When the head is not on the same cycle number as the tail,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3020) * we can't do a sequential recovery.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3021) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3022) while (blk_no < log->l_logBBsize) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3023) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3024) * Check for header wrapping around physical end-of-log
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3025) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3026) offset = hbp;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3027) split_hblks = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3028) wrapped_hblks = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3029) if (blk_no + hblks <= log->l_logBBsize) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3030) /* Read header in one read */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3031) error = xlog_bread(log, blk_no, hblks, hbp,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3032) &offset);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3033) if (error)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3034) goto bread_err2;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3035) } else {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3036) /* This LR is split across physical log end */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3037) if (blk_no != log->l_logBBsize) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3038) /* some data before physical log end */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3039) ASSERT(blk_no <= INT_MAX);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3040) split_hblks = log->l_logBBsize - (int)blk_no;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3041) ASSERT(split_hblks > 0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3042) error = xlog_bread(log, blk_no,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3043) split_hblks, hbp,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3044) &offset);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3045) if (error)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3046) goto bread_err2;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3047) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3048)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3049) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3050) * Note: this black magic still works with
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3051) * large sector sizes (non-512) only because:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3052) * - we increased the buffer size originally
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3053) * by 1 sector giving us enough extra space
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3054) * for the second read;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3055) * - the log start is guaranteed to be sector
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3056) * aligned;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3057) * - we read the log end (LR header start)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3058) * _first_, then the log start (LR header end)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3059) * - order is important.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3060) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3061) wrapped_hblks = hblks - split_hblks;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3062) error = xlog_bread_noalign(log, 0,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3063) wrapped_hblks,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3064) offset + BBTOB(split_hblks));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3065) if (error)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3066) goto bread_err2;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3067) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3068) rhead = (xlog_rec_header_t *)offset;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3069) error = xlog_valid_rec_header(log, rhead,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3070) split_hblks ? blk_no : 0, h_size);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3071) if (error)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3072) goto bread_err2;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3073)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3074) bblks = (int)BTOBB(be32_to_cpu(rhead->h_len));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3075) blk_no += hblks;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3076)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3077) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3078) * Read the log record data in multiple reads if it
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3079) * wraps around the end of the log. Note that if the
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3080) * header already wrapped, blk_no could point past the
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3081) * end of the log. The record data is contiguous in
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3082) * that case.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3083) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3084) if (blk_no + bblks <= log->l_logBBsize ||
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3085) blk_no >= log->l_logBBsize) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3086) rblk_no = xlog_wrap_logbno(log, blk_no);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3087) error = xlog_bread(log, rblk_no, bblks, dbp,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3088) &offset);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3089) if (error)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3090) goto bread_err2;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3091) } else {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3092) /* This log record is split across the
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3093) * physical end of log */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3094) offset = dbp;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3095) split_bblks = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3096) if (blk_no != log->l_logBBsize) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3097) /* some data is before the physical
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3098) * end of log */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3099) ASSERT(!wrapped_hblks);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3100) ASSERT(blk_no <= INT_MAX);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3101) split_bblks =
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3102) log->l_logBBsize - (int)blk_no;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3103) ASSERT(split_bblks > 0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3104) error = xlog_bread(log, blk_no,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3105) split_bblks, dbp,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3106) &offset);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3107) if (error)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3108) goto bread_err2;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3109) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3110)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3111) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3112) * Note: this black magic still works with
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3113) * large sector sizes (non-512) only because:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3114) * - we increased the buffer size originally
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3115) * by 1 sector giving us enough extra space
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3116) * for the second read;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3117) * - the log start is guaranteed to be sector
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3118) * aligned;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3119) * - we read the log end (LR header start)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3120) * _first_, then the log start (LR header end)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3121) * - order is important.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3122) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3123) error = xlog_bread_noalign(log, 0,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3124) bblks - split_bblks,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3125) offset + BBTOB(split_bblks));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3126) if (error)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3127) goto bread_err2;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3128) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3129)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3130) error = xlog_recover_process(log, rhash, rhead, offset,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3131) pass, &buffer_list);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3132) if (error)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3133) goto bread_err2;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3134)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3135) blk_no += bblks;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3136) rhead_blk = blk_no;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3137) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3138)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3139) ASSERT(blk_no >= log->l_logBBsize);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3140) blk_no -= log->l_logBBsize;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3141) rhead_blk = blk_no;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3142) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3143)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3144) /* read first part of physical log */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3145) while (blk_no < head_blk) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3146) error = xlog_bread(log, blk_no, hblks, hbp, &offset);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3147) if (error)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3148) goto bread_err2;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3149)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3150) rhead = (xlog_rec_header_t *)offset;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3151) error = xlog_valid_rec_header(log, rhead, blk_no, h_size);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3152) if (error)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3153) goto bread_err2;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3154)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3155) /* blocks in data section */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3156) bblks = (int)BTOBB(be32_to_cpu(rhead->h_len));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3157) error = xlog_bread(log, blk_no+hblks, bblks, dbp,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3158) &offset);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3159) if (error)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3160) goto bread_err2;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3161)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3162) error = xlog_recover_process(log, rhash, rhead, offset, pass,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3163) &buffer_list);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3164) if (error)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3165) goto bread_err2;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3166)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3167) blk_no += bblks + hblks;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3168) rhead_blk = blk_no;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3169) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3170)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3171) bread_err2:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3172) kmem_free(dbp);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3173) bread_err1:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3174) kmem_free(hbp);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3175)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3176) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3177) * Submit buffers that have been added from the last record processed,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3178) * regardless of error status.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3179) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3180) if (!list_empty(&buffer_list))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3181) error2 = xfs_buf_delwri_submit(&buffer_list);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3182)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3183) if (error && first_bad)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3184) *first_bad = rhead_blk;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3185)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3186) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3187) * Transactions are freed at commit time but transactions without commit
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3188) * records on disk are never committed. Free any that may be left in the
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3189) * hash table.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3190) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3191) for (i = 0; i < XLOG_RHASH_SIZE; i++) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3192) struct hlist_node *tmp;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3193) struct xlog_recover *trans;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3194)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3195) hlist_for_each_entry_safe(trans, tmp, &rhash[i], r_list)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3196) xlog_recover_free_trans(trans);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3197) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3198)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3199) return error ? error : error2;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3200) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3201)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3202) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3203) * Do the recovery of the log. We actually do this in two phases.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3204) * The two passes are necessary in order to implement the function
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3205) * of cancelling a record written into the log. The first pass
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3206) * determines those things which have been cancelled, and the
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3207) * second pass replays log items normally except for those which
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3208) * have been cancelled. The handling of the replay and cancellations
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3209) * takes place in the log item type specific routines.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3210) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3211) * The table of items which have cancel records in the log is allocated
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3212) * and freed at this level, since only here do we know when all of
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3213) * the log recovery has been completed.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3214) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3215) STATIC int
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3216) xlog_do_log_recovery(
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3217) struct xlog *log,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3218) xfs_daddr_t head_blk,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3219) xfs_daddr_t tail_blk)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3220) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3221) int error, i;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3222)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3223) ASSERT(head_blk != tail_blk);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3224)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3225) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3226) * First do a pass to find all of the cancelled buf log items.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3227) * Store them in the buf_cancel_table for use in the second pass.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3228) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3229) log->l_buf_cancel_table = kmem_zalloc(XLOG_BC_TABLE_SIZE *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3230) sizeof(struct list_head),
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3231) 0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3232) for (i = 0; i < XLOG_BC_TABLE_SIZE; i++)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3233) INIT_LIST_HEAD(&log->l_buf_cancel_table[i]);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3234)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3235) error = xlog_do_recovery_pass(log, head_blk, tail_blk,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3236) XLOG_RECOVER_PASS1, NULL);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3237) if (error != 0) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3238) kmem_free(log->l_buf_cancel_table);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3239) log->l_buf_cancel_table = NULL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3240) return error;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3241) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3242) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3243) * Then do a second pass to actually recover the items in the log.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3244) * When it is complete free the table of buf cancel items.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3245) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3246) error = xlog_do_recovery_pass(log, head_blk, tail_blk,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3247) XLOG_RECOVER_PASS2, NULL);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3248) #ifdef DEBUG
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3249) if (!error) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3250) int i;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3251)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3252) for (i = 0; i < XLOG_BC_TABLE_SIZE; i++)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3253) ASSERT(list_empty(&log->l_buf_cancel_table[i]));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3254) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3255) #endif /* DEBUG */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3256)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3257) kmem_free(log->l_buf_cancel_table);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3258) log->l_buf_cancel_table = NULL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3259)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3260) return error;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3261) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3262)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3263) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3264) * Do the actual recovery
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3265) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3266) STATIC int
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3267) xlog_do_recover(
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3268) struct xlog *log,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3269) xfs_daddr_t head_blk,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3270) xfs_daddr_t tail_blk)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3271) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3272) struct xfs_mount *mp = log->l_mp;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3273) struct xfs_buf *bp = mp->m_sb_bp;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3274) struct xfs_sb *sbp = &mp->m_sb;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3275) int error;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3276)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3277) trace_xfs_log_recover(log, head_blk, tail_blk);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3278)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3279) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3280) * First replay the images in the log.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3281) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3282) error = xlog_do_log_recovery(log, head_blk, tail_blk);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3283) if (error)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3284) return error;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3285)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3286) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3287) * If IO errors happened during recovery, bail out.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3288) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3289) if (XFS_FORCED_SHUTDOWN(mp))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3290) return -EIO;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3291)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3292) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3293) * We now update the tail_lsn since much of the recovery has completed
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3294) * and there may be space available to use. If there were no extent
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3295) * or iunlinks, we can free up the entire log and set the tail_lsn to
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3296) * be the last_sync_lsn. This was set in xlog_find_tail to be the
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3297) * lsn of the last known good LR on disk. If there are extent frees
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3298) * or iunlinks they will have some entries in the AIL; so we look at
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3299) * the AIL to determine how to set the tail_lsn.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3300) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3301) xlog_assign_tail_lsn(mp);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3302)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3303) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3304) * Now that we've finished replaying all buffer and inode updates,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3305) * re-read the superblock and reverify it.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3306) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3307) xfs_buf_lock(bp);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3308) xfs_buf_hold(bp);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3309) error = _xfs_buf_read(bp, XBF_READ);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3310) if (error) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3311) if (!XFS_FORCED_SHUTDOWN(mp)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3312) xfs_buf_ioerror_alert(bp, __this_address);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3313) ASSERT(0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3314) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3315) xfs_buf_relse(bp);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3316) return error;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3317) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3318)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3319) /* Convert superblock from on-disk format */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3320) xfs_sb_from_disk(sbp, bp->b_addr);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3321) xfs_buf_relse(bp);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3322)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3323) /* re-initialise in-core superblock and geometry structures */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3324) xfs_reinit_percpu_counters(mp);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3325) error = xfs_initialize_perag(mp, sbp->sb_agcount, &mp->m_maxagi);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3326) if (error) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3327) xfs_warn(mp, "Failed post-recovery per-ag init: %d", error);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3328) return error;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3329) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3330) mp->m_alloc_set_aside = xfs_alloc_set_aside(mp);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3331)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3332) xlog_recover_check_summary(log);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3333)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3334) /* Normal transactions can now occur */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3335) log->l_flags &= ~XLOG_ACTIVE_RECOVERY;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3336) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3337) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3338)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3339) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3340) * Perform recovery and re-initialize some log variables in xlog_find_tail.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3341) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3342) * Return error or zero.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3343) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3344) int
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3345) xlog_recover(
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3346) struct xlog *log)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3347) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3348) xfs_daddr_t head_blk, tail_blk;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3349) int error;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3350)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3351) /* find the tail of the log */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3352) error = xlog_find_tail(log, &head_blk, &tail_blk);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3353) if (error)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3354) return error;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3355)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3356) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3357) * The superblock was read before the log was available and thus the LSN
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3358) * could not be verified. Check the superblock LSN against the current
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3359) * LSN now that it's known.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3360) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3361) if (xfs_sb_version_hascrc(&log->l_mp->m_sb) &&
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3362) !xfs_log_check_lsn(log->l_mp, log->l_mp->m_sb.sb_lsn))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3363) return -EINVAL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3364)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3365) if (tail_blk != head_blk) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3366) /* There used to be a comment here:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3367) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3368) * disallow recovery on read-only mounts. note -- mount
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3369) * checks for ENOSPC and turns it into an intelligent
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3370) * error message.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3371) * ...but this is no longer true. Now, unless you specify
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3372) * NORECOVERY (in which case this function would never be
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3373) * called), we just go ahead and recover. We do this all
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3374) * under the vfs layer, so we can get away with it unless
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3375) * the device itself is read-only, in which case we fail.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3376) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3377) if ((error = xfs_dev_is_read_only(log->l_mp, "recovery"))) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3378) return error;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3379) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3380)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3381) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3382) * Version 5 superblock log feature mask validation. We know the
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3383) * log is dirty so check if there are any unknown log features
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3384) * in what we need to recover. If there are unknown features
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3385) * (e.g. unsupported transactions, then simply reject the
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3386) * attempt at recovery before touching anything.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3387) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3388) if (XFS_SB_VERSION_NUM(&log->l_mp->m_sb) == XFS_SB_VERSION_5 &&
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3389) xfs_sb_has_incompat_log_feature(&log->l_mp->m_sb,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3390) XFS_SB_FEAT_INCOMPAT_LOG_UNKNOWN)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3391) xfs_warn(log->l_mp,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3392) "Superblock has unknown incompatible log features (0x%x) enabled.",
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3393) (log->l_mp->m_sb.sb_features_log_incompat &
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3394) XFS_SB_FEAT_INCOMPAT_LOG_UNKNOWN));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3395) xfs_warn(log->l_mp,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3396) "The log can not be fully and/or safely recovered by this kernel.");
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3397) xfs_warn(log->l_mp,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3398) "Please recover the log on a kernel that supports the unknown features.");
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3399) return -EINVAL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3400) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3401)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3402) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3403) * Delay log recovery if the debug hook is set. This is debug
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3404) * instrumention to coordinate simulation of I/O failures with
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3405) * log recovery.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3406) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3407) if (xfs_globals.log_recovery_delay) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3408) xfs_notice(log->l_mp,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3409) "Delaying log recovery for %d seconds.",
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3410) xfs_globals.log_recovery_delay);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3411) msleep(xfs_globals.log_recovery_delay * 1000);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3412) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3413)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3414) xfs_notice(log->l_mp, "Starting recovery (logdev: %s)",
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3415) log->l_mp->m_logname ? log->l_mp->m_logname
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3416) : "internal");
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3417)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3418) error = xlog_do_recover(log, head_blk, tail_blk);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3419) log->l_flags |= XLOG_RECOVERY_NEEDED;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3420) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3421) return error;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3422) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3423)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3424) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3425) * In the first part of recovery we replay inodes and buffers and build
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3426) * up the list of extent free items which need to be processed. Here
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3427) * we process the extent free items and clean up the on disk unlinked
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3428) * inode lists. This is separated from the first part of recovery so
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3429) * that the root and real-time bitmap inodes can be read in from disk in
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3430) * between the two stages. This is necessary so that we can free space
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3431) * in the real-time portion of the file system.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3432) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3433) int
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3434) xlog_recover_finish(
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3435) struct xlog *log)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3436) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3437) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3438) * Now we're ready to do the transactions needed for the
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3439) * rest of recovery. Start with completing all the extent
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3440) * free intent records and then process the unlinked inode
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3441) * lists. At this point, we essentially run in normal mode
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3442) * except that we're still performing recovery actions
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3443) * rather than accepting new requests.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3444) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3445) if (log->l_flags & XLOG_RECOVERY_NEEDED) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3446) int error;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3447) error = xlog_recover_process_intents(log);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3448) if (error) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3449) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3450) * Cancel all the unprocessed intent items now so that
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3451) * we don't leave them pinned in the AIL. This can
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3452) * cause the AIL to livelock on the pinned item if
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3453) * anyone tries to push the AIL (inode reclaim does
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3454) * this) before we get around to xfs_log_mount_cancel.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3455) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3456) xlog_recover_cancel_intents(log);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3457) xfs_alert(log->l_mp, "Failed to recover intents");
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3458) return error;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3459) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3460)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3461) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3462) * Sync the log to get all the intents out of the AIL.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3463) * This isn't absolutely necessary, but it helps in
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3464) * case the unlink transactions would have problems
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3465) * pushing the intents out of the way.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3466) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3467) xfs_log_force(log->l_mp, XFS_LOG_SYNC);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3468)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3469) xlog_recover_process_iunlinks(log);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3470)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3471) xlog_recover_check_summary(log);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3472)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3473) xfs_notice(log->l_mp, "Ending recovery (logdev: %s)",
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3474) log->l_mp->m_logname ? log->l_mp->m_logname
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3475) : "internal");
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3476) log->l_flags &= ~XLOG_RECOVERY_NEEDED;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3477) } else {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3478) xfs_info(log->l_mp, "Ending clean mount");
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3479) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3480) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3481) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3482)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3483) void
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3484) xlog_recover_cancel(
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3485) struct xlog *log)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3486) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3487) if (log->l_flags & XLOG_RECOVERY_NEEDED)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3488) xlog_recover_cancel_intents(log);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3489) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3490)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3491) #if defined(DEBUG)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3492) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3493) * Read all of the agf and agi counters and check that they
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3494) * are consistent with the superblock counters.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3495) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3496) STATIC void
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3497) xlog_recover_check_summary(
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3498) struct xlog *log)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3499) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3500) xfs_mount_t *mp;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3501) xfs_buf_t *agfbp;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3502) xfs_buf_t *agibp;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3503) xfs_agnumber_t agno;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3504) uint64_t freeblks;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3505) uint64_t itotal;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3506) uint64_t ifree;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3507) int error;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3508)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3509) mp = log->l_mp;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3510)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3511) freeblks = 0LL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3512) itotal = 0LL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3513) ifree = 0LL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3514) for (agno = 0; agno < mp->m_sb.sb_agcount; agno++) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3515) error = xfs_read_agf(mp, NULL, agno, 0, &agfbp);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3516) if (error) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3517) xfs_alert(mp, "%s agf read failed agno %d error %d",
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3518) __func__, agno, error);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3519) } else {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3520) struct xfs_agf *agfp = agfbp->b_addr;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3521)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3522) freeblks += be32_to_cpu(agfp->agf_freeblks) +
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3523) be32_to_cpu(agfp->agf_flcount);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3524) xfs_buf_relse(agfbp);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3525) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3526)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3527) error = xfs_read_agi(mp, NULL, agno, &agibp);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3528) if (error) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3529) xfs_alert(mp, "%s agi read failed agno %d error %d",
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3530) __func__, agno, error);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3531) } else {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3532) struct xfs_agi *agi = agibp->b_addr;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3533)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3534) itotal += be32_to_cpu(agi->agi_count);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3535) ifree += be32_to_cpu(agi->agi_freecount);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3536) xfs_buf_relse(agibp);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3537) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3538) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3539) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3540) #endif /* DEBUG */
Orange Pi5 kernel
Deprecated Linux kernel 5.10.110 for OrangePi 5/5B/5+ boards
3 Commits
0 Branches
0 Tags