summaryrefslogtreecommitdiff
path: root/fs/xfs/xfs_ialloc.c
diff options
context:
space:
mode:
Diffstat (limited to 'fs/xfs/xfs_ialloc.c')
-rw-r--r--fs/xfs/xfs_ialloc.c177
1 files changed, 133 insertions, 44 deletions
diff --git a/fs/xfs/xfs_ialloc.c b/fs/xfs/xfs_ialloc.c
index a815412eab80..7a0c17d7ec09 100644
--- a/fs/xfs/xfs_ialloc.c
+++ b/fs/xfs/xfs_ialloc.c
@@ -36,6 +36,9 @@
#include "xfs_rtalloc.h"
#include "xfs_error.h"
#include "xfs_bmap.h"
+#include "xfs_cksum.h"
+#include "xfs_buf_item.h"
+#include "xfs_icreate_item.h"
/*
@@ -148,12 +151,16 @@ xfs_check_agi_freecount(
#endif
/*
- * Initialise a new set of inodes.
+ * Initialise a new set of inodes. When called without a transaction context
+ * (e.g. from recovery) we initiate a delayed write of the inode buffers rather
+ * than logging them (which in a transaction context puts them into the AIL
+ * for writeback rather than the xfsbufd queue).
*/
-STATIC int
+int
xfs_ialloc_inode_init(
struct xfs_mount *mp,
struct xfs_trans *tp,
+ struct list_head *buffer_list,
xfs_agnumber_t agno,
xfs_agblock_t agbno,
xfs_agblock_t length,
@@ -165,6 +172,7 @@ xfs_ialloc_inode_init(
int version;
int i, j;
xfs_daddr_t d;
+ xfs_ino_t ino = 0;
/*
* Loop over the new block(s), filling in the inodes.
@@ -183,13 +191,41 @@ xfs_ialloc_inode_init(
}
/*
- * Figure out what version number to use in the inodes we create.
- * If the superblock version has caught up to the one that supports
- * the new inode format, then use the new inode version. Otherwise
- * use the old version so that old kernels will continue to be
- * able to use the file system.
+ * Figure out what version number to use in the inodes we create. If
+ * the superblock version has caught up to the one that supports the new
+ * inode format, then use the new inode version. Otherwise use the old
+ * version so that old kernels will continue to be able to use the file
+ * system.
+ *
+ * For v3 inodes, we also need to write the inode number into the inode,
+ * so calculate the first inode number of the chunk here as
+ * XFS_OFFBNO_TO_AGINO() only works within a filesystem block, not
+ * across multiple filesystem blocks (such as a cluster) and so cannot
+ * be used in the cluster buffer loop below.
+ *
+ * Further, because we are writing the inode directly into the buffer
+ * and calculating a CRC on the entire inode, we have ot log the entire
+ * inode so that the entire range the CRC covers is present in the log.
+ * That means for v3 inode we log the entire buffer rather than just the
+ * inode cores.
*/
- if (xfs_sb_version_hasnlink(&mp->m_sb))
+ if (xfs_sb_version_hascrc(&mp->m_sb)) {
+ version = 3;
+ ino = XFS_AGINO_TO_INO(mp, agno,
+ XFS_OFFBNO_TO_AGINO(mp, agbno, 0));
+
+ /*
+ * log the initialisation that is about to take place as an
+ * logical operation. This means the transaction does not
+ * need to log the physical changes to the inode buffers as log
+ * recovery will know what initialisation is actually needed.
+ * Hence we only need to log the buffers as "ordered" buffers so
+ * they track in the AIL as if they were physically logged.
+ */
+ if (tp)
+ xfs_icreate_log(tp, agno, agbno, XFS_IALLOC_INODES(mp),
+ mp->m_sb.sb_inodesize, length, gen);
+ } else if (xfs_sb_version_hasnlink(&mp->m_sb))
version = 2;
else
version = 1;
@@ -204,27 +240,58 @@ xfs_ialloc_inode_init(
XBF_UNMAPPED);
if (!fbuf)
return ENOMEM;
- /*
- * Initialize all inodes in this buffer and then log them.
- *
- * XXX: It would be much better if we had just one transaction
- * to log a whole cluster of inodes instead of all the
- * individual transactions causing a lot of log traffic.
- */
+
+ /* Initialize the inode buffers and log them appropriately. */
fbuf->b_ops = &xfs_inode_buf_ops;
- xfs_buf_zero(fbuf, 0, ninodes << mp->m_sb.sb_inodelog);
+ xfs_buf_zero(fbuf, 0, BBTOB(fbuf->b_length));
for (i = 0; i < ninodes; i++) {
int ioffset = i << mp->m_sb.sb_inodelog;
- uint isize = sizeof(struct xfs_dinode);
+ uint isize = xfs_dinode_size(version);
free = xfs_make_iptr(mp, fbuf, i);
free->di_magic = cpu_to_be16(XFS_DINODE_MAGIC);
free->di_version = version;
free->di_gen = cpu_to_be32(gen);
free->di_next_unlinked = cpu_to_be32(NULLAGINO);
- xfs_trans_log_buf(tp, fbuf, ioffset, ioffset + isize - 1);
+
+ if (version == 3) {
+ free->di_ino = cpu_to_be64(ino);
+ ino++;
+ uuid_copy(&free->di_uuid, &mp->m_sb.sb_uuid);
+ xfs_dinode_calc_crc(mp, free);
+ } else if (tp) {
+ /* just log the inode core */
+ xfs_trans_log_buf(tp, fbuf, ioffset,
+ ioffset + isize - 1);
+ }
+ }
+
+ if (tp) {
+ /*
+ * Mark the buffer as an inode allocation buffer so it
+ * sticks in AIL at the point of this allocation
+ * transaction. This ensures the they are on disk before
+ * the tail of the log can be moved past this
+ * transaction (i.e. by preventing relogging from moving
+ * it forward in the log).
+ */
+ xfs_trans_inode_alloc_buf(tp, fbuf);
+ if (version == 3) {
+ /*
+ * Mark the buffer as ordered so that they are
+ * not physically logged in the transaction but
+ * still tracked in the AIL as part of the
+ * transaction and pin the log appropriately.
+ */
+ xfs_trans_ordered_buf(tp, fbuf);
+ xfs_trans_log_buf(tp, fbuf, 0,
+ BBTOB(fbuf->b_length) - 1);
+ }
+ } else {
+ fbuf->b_flags |= XBF_DONE;
+ xfs_buf_delwri_queue(fbuf, buffer_list);
+ xfs_buf_relse(fbuf);
}
- xfs_trans_inode_alloc_buf(tp, fbuf);
}
return 0;
}
@@ -269,7 +336,7 @@ xfs_ialloc_ag_alloc(
* First try to allocate inodes contiguous with the last-allocated
* chunk of inodes. If the filesystem is striped, this will fill
* an entire stripe unit with inodes.
- */
+ */
agi = XFS_BUF_TO_AGI(agbp);
newino = be32_to_cpu(agi->agi_newino);
agno = be32_to_cpu(agi->agi_seqno);
@@ -279,8 +346,6 @@ xfs_ialloc_ag_alloc(
(args.agbno < be32_to_cpu(agi->agi_length)))) {
args.fsbno = XFS_AGB_TO_FSB(args.mp, agno, args.agbno);
args.type = XFS_ALLOCTYPE_THIS_BNO;
- args.mod = args.total = args.wasdel = args.isfl =
- args.userdata = args.minalignslop = 0;
args.prod = 1;
/*
@@ -333,8 +398,6 @@ xfs_ialloc_ag_alloc(
* Allocate a fixed-size extent of inodes.
*/
args.type = XFS_ALLOCTYPE_NEAR_BNO;
- args.mod = args.total = args.wasdel = args.isfl =
- args.userdata = args.minalignslop = 0;
args.prod = 1;
/*
* Allow space for the inode btree to split.
@@ -372,8 +435,8 @@ xfs_ialloc_ag_alloc(
* rather than a linear progression to prevent the next generation
* number from being easily guessable.
*/
- error = xfs_ialloc_inode_init(args.mp, tp, agno, args.agbno,
- args.len, random32());
+ error = xfs_ialloc_inode_init(args.mp, tp, NULL, agno, args.agbno,
+ args.len, prandom_u32());
if (error)
return error;
@@ -585,8 +648,7 @@ xfs_ialloc_get_rec(
struct xfs_btree_cur *cur,
xfs_agino_t agino,
xfs_inobt_rec_incore_t *rec,
- int *done,
- int left)
+ int *done)
{
int error;
int i;
@@ -694,12 +756,12 @@ xfs_dialloc_ag(
pag->pagl_leftrec != NULLAGINO &&
pag->pagl_rightrec != NULLAGINO) {
error = xfs_ialloc_get_rec(tcur, pag->pagl_leftrec,
- &trec, &doneleft, 1);
+ &trec, &doneleft);
if (error)
goto error1;
error = xfs_ialloc_get_rec(cur, pag->pagl_rightrec,
- &rec, &doneright, 0);
+ &rec, &doneright);
if (error)
goto error1;
} else {
@@ -1457,6 +1519,7 @@ xfs_ialloc_log_agi(
/*
* Log the allocation group inode header buffer.
*/
+ xfs_trans_buf_set_type(tp, bp, XFS_BLFT_AGI_BUF);
xfs_trans_log_buf(tp, bp, first, last);
}
@@ -1474,19 +1537,23 @@ xfs_check_agi_unlinked(
#define xfs_check_agi_unlinked(agi)
#endif
-static void
+static bool
xfs_agi_verify(
struct xfs_buf *bp)
{
struct xfs_mount *mp = bp->b_target->bt_mount;
struct xfs_agi *agi = XFS_BUF_TO_AGI(bp);
- int agi_ok;
+ if (xfs_sb_version_hascrc(&mp->m_sb) &&
+ !uuid_equal(&agi->agi_uuid, &mp->m_sb.sb_uuid))
+ return false;
/*
* Validate the magic number of the agi block.
*/
- agi_ok = agi->agi_magicnum == cpu_to_be32(XFS_AGI_MAGIC) &&
- XFS_AGI_GOOD_VERSION(be32_to_cpu(agi->agi_versionnum));
+ if (agi->agi_magicnum != cpu_to_be32(XFS_AGI_MAGIC))
+ return false;
+ if (!XFS_AGI_GOOD_VERSION(be32_to_cpu(agi->agi_versionnum)))
+ return false;
/*
* during growfs operations, the perag is not fully initialised,
@@ -1494,30 +1561,52 @@ xfs_agi_verify(
* use it by using uncached buffers that don't have the perag attached
* so we can detect and avoid this problem.
*/
- if (bp->b_pag)
- agi_ok = agi_ok && be32_to_cpu(agi->agi_seqno) ==
- bp->b_pag->pag_agno;
+ if (bp->b_pag && be32_to_cpu(agi->agi_seqno) != bp->b_pag->pag_agno)
+ return false;
- if (unlikely(XFS_TEST_ERROR(!agi_ok, mp, XFS_ERRTAG_IALLOC_READ_AGI,
- XFS_RANDOM_IALLOC_READ_AGI))) {
- XFS_CORRUPTION_ERROR(__func__, XFS_ERRLEVEL_LOW, mp, agi);
- xfs_buf_ioerror(bp, EFSCORRUPTED);
- }
xfs_check_agi_unlinked(agi);
+ return true;
}
static void
xfs_agi_read_verify(
struct xfs_buf *bp)
{
- xfs_agi_verify(bp);
+ struct xfs_mount *mp = bp->b_target->bt_mount;
+ int agi_ok = 1;
+
+ if (xfs_sb_version_hascrc(&mp->m_sb))
+ agi_ok = xfs_verify_cksum(bp->b_addr, BBTOB(bp->b_length),
+ offsetof(struct xfs_agi, agi_crc));
+ agi_ok = agi_ok && xfs_agi_verify(bp);
+
+ if (unlikely(XFS_TEST_ERROR(!agi_ok, mp, XFS_ERRTAG_IALLOC_READ_AGI,
+ XFS_RANDOM_IALLOC_READ_AGI))) {
+ XFS_CORRUPTION_ERROR(__func__, XFS_ERRLEVEL_LOW, mp, bp->b_addr);
+ xfs_buf_ioerror(bp, EFSCORRUPTED);
+ }
}
static void
xfs_agi_write_verify(
struct xfs_buf *bp)
{
- xfs_agi_verify(bp);
+ struct xfs_mount *mp = bp->b_target->bt_mount;
+ struct xfs_buf_log_item *bip = bp->b_fspriv;
+
+ if (!xfs_agi_verify(bp)) {
+ XFS_CORRUPTION_ERROR(__func__, XFS_ERRLEVEL_LOW, mp, bp->b_addr);
+ xfs_buf_ioerror(bp, EFSCORRUPTED);
+ return;
+ }
+
+ if (!xfs_sb_version_hascrc(&mp->m_sb))
+ return;
+
+ if (bip)
+ XFS_BUF_TO_AGI(bp)->agi_lsn = cpu_to_be64(bip->bli_item.li_lsn);
+ xfs_update_cksum(bp->b_addr, BBTOB(bp->b_length),
+ offsetof(struct xfs_agi, agi_crc));
}
const struct xfs_buf_ops xfs_agi_buf_ops = {
generated by cgit v1.2.3 (git 2.39.0) at 2025-12-12 04:54:06 +0300