*/
#include "xfs.h"
#include "xfs_fs.h"
+#include "xfs_shared.h"
#include "xfs_format.h"
+#include "xfs_log_format.h"
+#include "xfs_trans_resv.h"
#include "xfs_bit.h"
-#include "xfs_log.h"
#include "xfs_inum.h"
-#include "xfs_trans.h"
#include "xfs_sb.h"
#include "xfs_ag.h"
#include "xfs_mount.h"
+#include "xfs_da_format.h"
#include "xfs_error.h"
#include "xfs_bmap_btree.h"
#include "xfs_alloc_btree.h"
#include "xfs_btree.h"
#include "xfs_dinode.h"
#include "xfs_inode.h"
+#include "xfs_trans.h"
#include "xfs_inode_item.h"
#include "xfs_alloc.h"
#include "xfs_ialloc.h"
+#include "xfs_log.h"
#include "xfs_log_priv.h"
-#include "xfs_buf_item.h"
#include "xfs_log_recover.h"
#include "xfs_extfree_item.h"
#include "xfs_trans_priv.h"
#include "xfs_cksum.h"
#include "xfs_trace.h"
#include "xfs_icache.h"
-#include "xfs_icreate_item.h"
/* Need all the magic numbers and buffer ops structures from these headers */
-#include "xfs_symlink.h"
#include "xfs_da_btree.h"
-#include "xfs_dir2_format.h"
#include "xfs_dir2.h"
-#include "xfs_attr_leaf.h"
-#include "xfs_attr_remote.h"
#define BLK_AVG(blk1, blk2) ((blk1+blk2) >> 1)
xfs_mount_t *mp,
xlog_rec_header_t *head)
{
- xfs_debug(mp, "%s: SB : uuid = %pU, fmt = %d\n",
+ xfs_debug(mp, "%s: SB : uuid = %pU, fmt = %d",
__func__, &mp->m_sb.sb_uuid, XLOG_FMT);
- xfs_debug(mp, " log : uuid = %pU, fmt = %d\n",
+ xfs_debug(mp, " log : uuid = %pU, fmt = %d",
&head->h_fs_uuid, be32_to_cpu(head->h_fmt));
}
#else
"bad number of regions (%d) in inode log format",
in_f->ilf_size);
ASSERT(0);
+ kmem_free(ptr);
return XFS_ERROR(EIO);
}
* magic number. If we don't recognise the magic number in the buffer, then
* return a LSN of -1 so that the caller knows it was an unrecognised block and
* so can recover the buffer.
+ *
+ * Note: we cannot rely solely on magic number matches to determine that the
+ * buffer has a valid LSN - we also need to verify that it belongs to this
+ * filesystem, so we need to extract the object's LSN and compare it to that
+ * which we read from the superblock. If the UUIDs don't match, then we've got a
+ * stale metadata block from an old filesystem instance that we need to recover
+ * over the top of.
*/
static xfs_lsn_t
xlog_recover_get_buf_lsn(
__uint16_t magic16;
__uint16_t magicda;
void *blk = bp->b_addr;
+ uuid_t *uuid;
+ xfs_lsn_t lsn = -1;
/* v4 filesystems always recover immediately */
if (!xfs_sb_version_hascrc(&mp->m_sb))
case XFS_ABTB_MAGIC:
case XFS_ABTC_MAGIC:
case XFS_IBT_CRC_MAGIC:
- case XFS_IBT_MAGIC:
- return be64_to_cpu(
- ((struct xfs_btree_block *)blk)->bb_u.s.bb_lsn);
+ case XFS_IBT_MAGIC: {
+ struct xfs_btree_block *btb = blk;
+
+ lsn = be64_to_cpu(btb->bb_u.s.bb_lsn);
+ uuid = &btb->bb_u.s.bb_uuid;
+ break;
+ }
case XFS_BMAP_CRC_MAGIC:
- case XFS_BMAP_MAGIC:
- return be64_to_cpu(
- ((struct xfs_btree_block *)blk)->bb_u.l.bb_lsn);
+ case XFS_BMAP_MAGIC: {
+ struct xfs_btree_block *btb = blk;
+
+ lsn = be64_to_cpu(btb->bb_u.l.bb_lsn);
+ uuid = &btb->bb_u.l.bb_uuid;
+ break;
+ }
case XFS_AGF_MAGIC:
- return be64_to_cpu(((struct xfs_agf *)blk)->agf_lsn);
+ lsn = be64_to_cpu(((struct xfs_agf *)blk)->agf_lsn);
+ uuid = &((struct xfs_agf *)blk)->agf_uuid;
+ break;
case XFS_AGFL_MAGIC:
- return be64_to_cpu(((struct xfs_agfl *)blk)->agfl_lsn);
+ lsn = be64_to_cpu(((struct xfs_agfl *)blk)->agfl_lsn);
+ uuid = &((struct xfs_agfl *)blk)->agfl_uuid;
+ break;
case XFS_AGI_MAGIC:
- return be64_to_cpu(((struct xfs_agi *)blk)->agi_lsn);
+ lsn = be64_to_cpu(((struct xfs_agi *)blk)->agi_lsn);
+ uuid = &((struct xfs_agi *)blk)->agi_uuid;
+ break;
case XFS_SYMLINK_MAGIC:
- return be64_to_cpu(((struct xfs_dsymlink_hdr *)blk)->sl_lsn);
+ lsn = be64_to_cpu(((struct xfs_dsymlink_hdr *)blk)->sl_lsn);
+ uuid = &((struct xfs_dsymlink_hdr *)blk)->sl_uuid;
+ break;
case XFS_DIR3_BLOCK_MAGIC:
case XFS_DIR3_DATA_MAGIC:
case XFS_DIR3_FREE_MAGIC:
- return be64_to_cpu(((struct xfs_dir3_blk_hdr *)blk)->lsn);
+ lsn = be64_to_cpu(((struct xfs_dir3_blk_hdr *)blk)->lsn);
+ uuid = &((struct xfs_dir3_blk_hdr *)blk)->uuid;
+ break;
case XFS_ATTR3_RMT_MAGIC:
- return be64_to_cpu(((struct xfs_attr3_rmt_hdr *)blk)->rm_lsn);
+ lsn = be64_to_cpu(((struct xfs_attr3_rmt_hdr *)blk)->rm_lsn);
+ uuid = &((struct xfs_attr3_rmt_hdr *)blk)->rm_uuid;
+ break;
case XFS_SB_MAGIC:
- return be64_to_cpu(((struct xfs_sb *)blk)->sb_lsn);
+ lsn = be64_to_cpu(((struct xfs_dsb *)blk)->sb_lsn);
+ uuid = &((struct xfs_dsb *)blk)->sb_uuid;
+ break;
default:
break;
}
+ if (lsn != (xfs_lsn_t)-1) {
+ if (!uuid_equal(&mp->m_sb.sb_uuid, uuid))
+ goto recover_immediately;
+ return lsn;
+ }
+
magicda = be16_to_cpu(((struct xfs_da_blkinfo *)blk)->magic);
switch (magicda) {
case XFS_DIR3_LEAF1_MAGIC:
case XFS_DIR3_LEAFN_MAGIC:
case XFS_DA3_NODE_MAGIC:
- return be64_to_cpu(((struct xfs_da3_blkinfo *)blk)->lsn);
+ lsn = be64_to_cpu(((struct xfs_da3_blkinfo *)blk)->lsn);
+ uuid = &((struct xfs_da3_blkinfo *)blk)->uuid;
+ break;
default:
break;
}
+ if (lsn != (xfs_lsn_t)-1) {
+ if (!uuid_equal(&mp->m_sb.sb_uuid, uuid))
+ goto recover_immediately;
+ return lsn;
+ }
+
/*
* We do individual object checks on dquot and inode buffers as they
* have their own individual LSN records. Also, we could have a stale
item->ri_buf[i].i_len, __func__);
goto next;
}
- error = xfs_qm_dqcheck(mp, item->ri_buf[i].i_addr,
+ error = xfs_dqcheck(mp, item->ri_buf[i].i_addr,
-1, 0, XFS_QMOPT_DOWARN,
"dquot_buf_recover");
if (error)
xlog_recover_validate_buf_type(mp, bp, buf_f);
}
-/*
- * Do some primitive error checking on ondisk dquot data structures.
- */
-int
-xfs_qm_dqcheck(
- struct xfs_mount *mp,
- xfs_disk_dquot_t *ddq,
- xfs_dqid_t id,
- uint type, /* used only when IO_dorepair is true */
- uint flags,
- char *str)
-{
- xfs_dqblk_t *d = (xfs_dqblk_t *)ddq;
- int errs = 0;
-
- /*
- * We can encounter an uninitialized dquot buffer for 2 reasons:
- * 1. If we crash while deleting the quotainode(s), and those blks got
- * used for user data. This is because we take the path of regular
- * file deletion; however, the size field of quotainodes is never
- * updated, so all the tricks that we play in itruncate_finish
- * don't quite matter.
- *
- * 2. We don't play the quota buffers when there's a quotaoff logitem.
- * But the allocation will be replayed so we'll end up with an
- * uninitialized quota block.
- *
- * This is all fine; things are still consistent, and we haven't lost
- * any quota information. Just don't complain about bad dquot blks.
- */
- if (ddq->d_magic != cpu_to_be16(XFS_DQUOT_MAGIC)) {
- if (flags & XFS_QMOPT_DOWARN)
- xfs_alert(mp,
- "%s : XFS dquot ID 0x%x, magic 0x%x != 0x%x",
- str, id, be16_to_cpu(ddq->d_magic), XFS_DQUOT_MAGIC);
- errs++;
- }
- if (ddq->d_version != XFS_DQUOT_VERSION) {
- if (flags & XFS_QMOPT_DOWARN)
- xfs_alert(mp,
- "%s : XFS dquot ID 0x%x, version 0x%x != 0x%x",
- str, id, ddq->d_version, XFS_DQUOT_VERSION);
- errs++;
- }
-
- if (ddq->d_flags != XFS_DQ_USER &&
- ddq->d_flags != XFS_DQ_PROJ &&
- ddq->d_flags != XFS_DQ_GROUP) {
- if (flags & XFS_QMOPT_DOWARN)
- xfs_alert(mp,
- "%s : XFS dquot ID 0x%x, unknown flags 0x%x",
- str, id, ddq->d_flags);
- errs++;
- }
-
- if (id != -1 && id != be32_to_cpu(ddq->d_id)) {
- if (flags & XFS_QMOPT_DOWARN)
- xfs_alert(mp,
- "%s : ondisk-dquot 0x%p, ID mismatch: "
- "0x%x expected, found id 0x%x",
- str, ddq, id, be32_to_cpu(ddq->d_id));
- errs++;
- }
-
- if (!errs && ddq->d_id) {
- if (ddq->d_blk_softlimit &&
- be64_to_cpu(ddq->d_bcount) >
- be64_to_cpu(ddq->d_blk_softlimit)) {
- if (!ddq->d_btimer) {
- if (flags & XFS_QMOPT_DOWARN)
- xfs_alert(mp,
- "%s : Dquot ID 0x%x (0x%p) BLK TIMER NOT STARTED",
- str, (int)be32_to_cpu(ddq->d_id), ddq);
- errs++;
- }
- }
- if (ddq->d_ino_softlimit &&
- be64_to_cpu(ddq->d_icount) >
- be64_to_cpu(ddq->d_ino_softlimit)) {
- if (!ddq->d_itimer) {
- if (flags & XFS_QMOPT_DOWARN)
- xfs_alert(mp,
- "%s : Dquot ID 0x%x (0x%p) INODE TIMER NOT STARTED",
- str, (int)be32_to_cpu(ddq->d_id), ddq);
- errs++;
- }
- }
- if (ddq->d_rtb_softlimit &&
- be64_to_cpu(ddq->d_rtbcount) >
- be64_to_cpu(ddq->d_rtb_softlimit)) {
- if (!ddq->d_rtbtimer) {
- if (flags & XFS_QMOPT_DOWARN)
- xfs_alert(mp,
- "%s : Dquot ID 0x%x (0x%p) RTBLK TIMER NOT STARTED",
- str, (int)be32_to_cpu(ddq->d_id), ddq);
- errs++;
- }
- }
- }
-
- if (!errs || !(flags & XFS_QMOPT_DQREPAIR))
- return errs;
-
- if (flags & XFS_QMOPT_DOWARN)
- xfs_notice(mp, "Re-initializing dquot ID 0x%x", id);
-
- /*
- * Typically, a repair is only requested by quotacheck.
- */
- ASSERT(id != -1);
- ASSERT(flags & XFS_QMOPT_DQREPAIR);
- memset(d, 0, sizeof(xfs_dqblk_t));
-
- d->dd_diskdq.d_magic = cpu_to_be16(XFS_DQUOT_MAGIC);
- d->dd_diskdq.d_version = XFS_DQUOT_VERSION;
- d->dd_diskdq.d_flags = type;
- d->dd_diskdq.d_id = cpu_to_be32(id);
-
- if (xfs_sb_version_hascrc(&mp->m_sb)) {
- uuid_copy(&d->dd_uuid, &mp->m_sb.sb_uuid);
- xfs_update_cksum((char *)d, sizeof(struct xfs_dqblk),
- XFS_DQUOT_CRC_OFF);
- }
-
- return errs;
-}
-
/*
* Perform a dquot buffer recovery.
* Simple algorithm: if we have found a QUOTAOFF log item of the same type
return error;
}
+/*
+ * Inode fork owner changes
+ *
+ * If we have been told that we have to reparent the inode fork, it's because an
+ * extent swap operation on a CRC enabled filesystem has been done and we are
+ * replaying it. We need to walk the BMBT of the appropriate fork and change the
+ * owners of it.
+ *
+ * The complexity here is that we don't have an inode context to work with, so
+ * after we've replayed the inode we need to instantiate one. This is where the
+ * fun begins.
+ *
+ * We are in the middle of log recovery, so we can't run transactions. That
+ * means we cannot use cache coherent inode instantiation via xfs_iget(), as
+ * that will result in the corresponding iput() running the inode through
+ * xfs_inactive(). If we've just replayed an inode core that changes the link
+ * count to zero (i.e. it's been unlinked), then xfs_inactive() will run
+ * transactions (bad!).
+ *
+ * So, to avoid this, we instantiate an inode directly from the inode core we've
+ * just recovered. We have the buffer still locked, and all we really need to
+ * instantiate is the inode core and the forks being modified. We can do this
+ * manually, then run the inode btree owner change, and then tear down the
+ * xfs_inode without having to run any transactions at all.
+ *
+ * Also, because we don't have a transaction context available here but need to
+ * gather all the buffers we modify for writeback so we pass the buffer_list
+ * instead for the operation to use.
+ */
+
+STATIC int
+xfs_recover_inode_owner_change(
+ struct xfs_mount *mp,
+ struct xfs_dinode *dip,
+ struct xfs_inode_log_format *in_f,
+ struct list_head *buffer_list)
+{
+ struct xfs_inode *ip;
+ int error;
+
+ ASSERT(in_f->ilf_fields & (XFS_ILOG_DOWNER|XFS_ILOG_AOWNER));
+
+ ip = xfs_inode_alloc(mp, in_f->ilf_ino);
+ if (!ip)
+ return ENOMEM;
+
+ /* instantiate the inode */
+ xfs_dinode_from_disk(&ip->i_d, dip);
+ ASSERT(ip->i_d.di_version >= 3);
+
+ error = xfs_iformat_fork(ip, dip);
+ if (error)
+ goto out_free_ip;
+
+
+ if (in_f->ilf_fields & XFS_ILOG_DOWNER) {
+ ASSERT(in_f->ilf_fields & XFS_ILOG_DBROOT);
+ error = xfs_bmbt_change_owner(NULL, ip, XFS_DATA_FORK,
+ ip->i_ino, buffer_list);
+ if (error)
+ goto out_free_ip;
+ }
+
+ if (in_f->ilf_fields & XFS_ILOG_AOWNER) {
+ ASSERT(in_f->ilf_fields & XFS_ILOG_ABROOT);
+ error = xfs_bmbt_change_owner(NULL, ip, XFS_ATTR_FORK,
+ ip->i_ino, buffer_list);
+ if (error)
+ goto out_free_ip;
+ }
+
+out_free_ip:
+ xfs_inode_free(ip);
+ return error;
+}
+
STATIC int
xlog_recover_inode_pass2(
struct xlog *log,
error = bp->b_error;
if (error) {
xfs_buf_ioerror_alert(bp, "xlog_recover_do..(read#2)");
- xfs_buf_relse(bp);
- goto error;
+ goto out_release;
}
ASSERT(in_f->ilf_fields & XFS_ILOG_CORE);
dip = (xfs_dinode_t *)xfs_buf_offset(bp, in_f->ilf_boffset);
* like an inode!
*/
if (unlikely(dip->di_magic != cpu_to_be16(XFS_DINODE_MAGIC))) {
- xfs_buf_relse(bp);
xfs_alert(mp,
"%s: Bad inode magic number, dip = 0x%p, dino bp = 0x%p, ino = %Ld",
__func__, dip, bp, in_f->ilf_ino);
XFS_ERROR_REPORT("xlog_recover_inode_pass2(1)",
XFS_ERRLEVEL_LOW, mp);
error = EFSCORRUPTED;
- goto error;
+ goto out_release;
}
dicp = item->ri_buf[1].i_addr;
if (unlikely(dicp->di_magic != XFS_DINODE_MAGIC)) {
- xfs_buf_relse(bp);
xfs_alert(mp,
"%s: Bad inode log record, rec ptr 0x%p, ino %Ld",
__func__, item, in_f->ilf_ino);
XFS_ERROR_REPORT("xlog_recover_inode_pass2(2)",
XFS_ERRLEVEL_LOW, mp);
error = EFSCORRUPTED;
- goto error;
+ goto out_release;
}
/*
* If the inode has an LSN in it, recover the inode only if it's less
- * than the lsn of the transaction we are replaying.
+ * than the lsn of the transaction we are replaying. Note: we still
+ * need to replay an owner change even though the inode is more recent
+ * than the transaction as there is no guarantee that all the btree
+ * blocks are more recent than this transaction, too.
*/
if (dip->di_version >= 3) {
xfs_lsn_t lsn = be64_to_cpu(dip->di_lsn);
if (lsn && lsn != -1 && XFS_LSN_CMP(lsn, current_lsn) >= 0) {
trace_xfs_log_recover_inode_skip(log, in_f);
error = 0;
- goto out_release;
+ goto out_owner_change;
}
}
dicp->di_flushiter < (DI_MAX_FLUSH >> 1)) {
/* do nothing */
} else {
- xfs_buf_relse(bp);
trace_xfs_log_recover_inode_skip(log, in_f);
error = 0;
- goto error;
+ goto out_release;
}
}
(dicp->di_format != XFS_DINODE_FMT_BTREE)) {
XFS_CORRUPTION_ERROR("xlog_recover_inode_pass2(3)",
XFS_ERRLEVEL_LOW, mp, dicp);
- xfs_buf_relse(bp);
xfs_alert(mp,
"%s: Bad regular inode log record, rec ptr 0x%p, "
"ino ptr = 0x%p, ino bp = 0x%p, ino %Ld",
__func__, item, dip, bp, in_f->ilf_ino);
error = EFSCORRUPTED;
- goto error;
+ goto out_release;
}
} else if (unlikely(S_ISDIR(dicp->di_mode))) {
if ((dicp->di_format != XFS_DINODE_FMT_EXTENTS) &&
(dicp->di_format != XFS_DINODE_FMT_LOCAL)) {
XFS_CORRUPTION_ERROR("xlog_recover_inode_pass2(4)",
XFS_ERRLEVEL_LOW, mp, dicp);
- xfs_buf_relse(bp);
xfs_alert(mp,
"%s: Bad dir inode log record, rec ptr 0x%p, "
"ino ptr = 0x%p, ino bp = 0x%p, ino %Ld",
__func__, item, dip, bp, in_f->ilf_ino);
error = EFSCORRUPTED;
- goto error;
+ goto out_release;
}
}
if (unlikely(dicp->di_nextents + dicp->di_anextents > dicp->di_nblocks)){
XFS_CORRUPTION_ERROR("xlog_recover_inode_pass2(5)",
XFS_ERRLEVEL_LOW, mp, dicp);
- xfs_buf_relse(bp);
xfs_alert(mp,
"%s: Bad inode log record, rec ptr 0x%p, dino ptr 0x%p, "
"dino bp 0x%p, ino %Ld, total extents = %d, nblocks = %Ld",
dicp->di_nextents + dicp->di_anextents,
dicp->di_nblocks);
error = EFSCORRUPTED;
- goto error;
+ goto out_release;
}
if (unlikely(dicp->di_forkoff > mp->m_sb.sb_inodesize)) {
XFS_CORRUPTION_ERROR("xlog_recover_inode_pass2(6)",
XFS_ERRLEVEL_LOW, mp, dicp);
- xfs_buf_relse(bp);
xfs_alert(mp,
"%s: Bad inode log record, rec ptr 0x%p, dino ptr 0x%p, "
"dino bp 0x%p, ino %Ld, forkoff 0x%x", __func__,
item, dip, bp, in_f->ilf_ino, dicp->di_forkoff);
error = EFSCORRUPTED;
- goto error;
+ goto out_release;
}
isize = xfs_icdinode_size(dicp->di_version);
if (unlikely(item->ri_buf[1].i_len > isize)) {
XFS_CORRUPTION_ERROR("xlog_recover_inode_pass2(7)",
XFS_ERRLEVEL_LOW, mp, dicp);
- xfs_buf_relse(bp);
xfs_alert(mp,
"%s: Bad inode log record length %d, rec ptr 0x%p",
__func__, item->ri_buf[1].i_len, item);
error = EFSCORRUPTED;
- goto error;
+ goto out_release;
}
/* The core is in in-core format */
}
if (in_f->ilf_size == 2)
- goto write_inode_buffer;
+ goto out_owner_change;
len = item->ri_buf[2].i_len;
src = item->ri_buf[2].i_addr;
ASSERT(in_f->ilf_size <= 4);
default:
xfs_warn(log->l_mp, "%s: Invalid flag", __func__);
ASSERT(0);
- xfs_buf_relse(bp);
error = EIO;
- goto error;
+ goto out_release;
}
}
-write_inode_buffer:
+out_owner_change:
+ if (in_f->ilf_fields & (XFS_ILOG_DOWNER|XFS_ILOG_AOWNER))
+ error = xfs_recover_inode_owner_change(mp, dip, in_f,
+ buffer_list);
/* re-generate the checksum. */
xfs_dinode_calc_crc(log->l_mp, dip);
*/
dq_f = item->ri_buf[0].i_addr;
ASSERT(dq_f);
- error = xfs_qm_dqcheck(mp, recddq, dq_f->qlf_id, 0, XFS_QMOPT_DOWARN,
+ error = xfs_dqcheck(mp, recddq, dq_f->qlf_id, 0, XFS_QMOPT_DOWARN,
"xlog_recover_dquot_pass2 (log copy)");
if (error)
return XFS_ERROR(EIO);
* was among a chunk of dquots created earlier, and we did some
* minimal initialization then.
*/
- error = xfs_qm_dqcheck(mp, ddq, dq_f->qlf_id, 0, XFS_QMOPT_DOWARN,
+ error = xfs_dqcheck(mp, ddq, dq_f->qlf_id, 0, XFS_QMOPT_DOWARN,
"xlog_recover_dquot_pass2");
if (error) {
xfs_buf_relse(bp);
if (crc != rhead->h_crc) {
if (rhead->h_crc || xfs_sb_version_hascrc(&log->l_mp->m_sb)) {
xfs_alert(log->l_mp,
- "log record CRC mismatch: found 0x%x, expected 0x%x.\n",
+ "log record CRC mismatch: found 0x%x, expected 0x%x.",
le32_to_cpu(rhead->h_crc),
le32_to_cpu(crc));
xfs_hex_dump(dp, 32);