summaryrefslogtreecommitdiffstats
path: root/fs/xfs/xfs_rmap_item.c
diff options
context:
space:
mode:
authorDarrick J. Wong <darrick.wong@oracle.com>2016-08-03 12:04:45 +1000
committerDave Chinner <david@fromorbit.com>2016-08-03 12:04:45 +1000
commit5880f2d78ff17c6ee7c7f6d4071bfd13090c264c (patch)
tree6f90df23bcb16a03e1e1bb90bccf2336467c0cba /fs/xfs/xfs_rmap_item.c
parentabf09233817b5ea1241db0c187136d3b4738d218 (diff)
downloadlinux-5880f2d78ff17c6ee7c7f6d4071bfd13090c264c.tar.bz2
xfs: create rmap update intent log items
Create rmap update intent/done log items to record redo information in the log. Because we need to roll transactions between updating the bmbt mapping and updating the reverse mapping, we also have to track the status of the metadata updates that will be recorded in the post-roll transactions, just in case we crash before committing the final transaction. This mechanism enables log recovery to finish what was already started. Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com> Reviewed-by: Brian Foster <bfoster@redhat.com> Signed-off-by: Dave Chinner <david@fromorbit.com>
Diffstat (limited to 'fs/xfs/xfs_rmap_item.c')
-rw-r--r--fs/xfs/xfs_rmap_item.c459
1 files changed, 459 insertions, 0 deletions
diff --git a/fs/xfs/xfs_rmap_item.c b/fs/xfs/xfs_rmap_item.c
new file mode 100644
index 000000000000..5398b8478f02
--- /dev/null
+++ b/fs/xfs/xfs_rmap_item.c
@@ -0,0 +1,459 @@
+/*
+ * Copyright (C) 2016 Oracle. All Rights Reserved.
+ *
+ * Author: Darrick J. Wong <darrick.wong@oracle.com>
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License
+ * as published by the Free Software Foundation; either version 2
+ * of the License, or (at your option) any later version.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301, USA.
+ */
+#include "xfs.h"
+#include "xfs_fs.h"
+#include "xfs_format.h"
+#include "xfs_log_format.h"
+#include "xfs_trans_resv.h"
+#include "xfs_mount.h"
+#include "xfs_trans.h"
+#include "xfs_trans_priv.h"
+#include "xfs_buf_item.h"
+#include "xfs_rmap_item.h"
+#include "xfs_log.h"
+
+
+kmem_zone_t *xfs_rui_zone;
+kmem_zone_t *xfs_rud_zone;
+
+static inline struct xfs_rui_log_item *RUI_ITEM(struct xfs_log_item *lip)
+{
+ return container_of(lip, struct xfs_rui_log_item, rui_item);
+}
+
+void
+xfs_rui_item_free(
+ struct xfs_rui_log_item *ruip)
+{
+ if (ruip->rui_format.rui_nextents > XFS_RUI_MAX_FAST_EXTENTS)
+ kmem_free(ruip);
+ else
+ kmem_zone_free(xfs_rui_zone, ruip);
+}
+
+/*
+ * This returns the number of iovecs needed to log the given rui item.
+ * We only need 1 iovec for an rui item. It just logs the rui_log_format
+ * structure.
+ */
+static inline int
+xfs_rui_item_sizeof(
+ struct xfs_rui_log_item *ruip)
+{
+ return sizeof(struct xfs_rui_log_format) +
+ (ruip->rui_format.rui_nextents - 1) *
+ sizeof(struct xfs_map_extent);
+}
+
+STATIC void
+xfs_rui_item_size(
+ struct xfs_log_item *lip,
+ int *nvecs,
+ int *nbytes)
+{
+ *nvecs += 1;
+ *nbytes += xfs_rui_item_sizeof(RUI_ITEM(lip));
+}
+
+/*
+ * This is called to fill in the vector of log iovecs for the
+ * given rui log item. We use only 1 iovec, and we point that
+ * at the rui_log_format structure embedded in the rui item.
+ * It is at this point that we assert that all of the extent
+ * slots in the rui item have been filled.
+ */
+STATIC void
+xfs_rui_item_format(
+ struct xfs_log_item *lip,
+ struct xfs_log_vec *lv)
+{
+ struct xfs_rui_log_item *ruip = RUI_ITEM(lip);
+ struct xfs_log_iovec *vecp = NULL;
+
+ ASSERT(atomic_read(&ruip->rui_next_extent) ==
+ ruip->rui_format.rui_nextents);
+
+ ruip->rui_format.rui_type = XFS_LI_RUI;
+ ruip->rui_format.rui_size = 1;
+
+ xlog_copy_iovec(lv, &vecp, XLOG_REG_TYPE_RUI_FORMAT, &ruip->rui_format,
+ xfs_rui_item_sizeof(ruip));
+}
+
+/*
+ * Pinning has no meaning for an rui item, so just return.
+ */
+STATIC void
+xfs_rui_item_pin(
+ struct xfs_log_item *lip)
+{
+}
+
+/*
+ * The unpin operation is the last place an RUI is manipulated in the log. It is
+ * either inserted in the AIL or aborted in the event of a log I/O error. In
+ * either case, the RUI transaction has been successfully committed to make it
+ * this far. Therefore, we expect whoever committed the RUI to either construct
+ * and commit the RUD or drop the RUD's reference in the event of error. Simply
+ * drop the log's RUI reference now that the log is done with it.
+ */
+STATIC void
+xfs_rui_item_unpin(
+ struct xfs_log_item *lip,
+ int remove)
+{
+ struct xfs_rui_log_item *ruip = RUI_ITEM(lip);
+
+ xfs_rui_release(ruip);
+}
+
+/*
+ * RUI items have no locking or pushing. However, since RUIs are pulled from
+ * the AIL when their corresponding RUDs are committed to disk, their situation
+ * is very similar to being pinned. Return XFS_ITEM_PINNED so that the caller
+ * will eventually flush the log. This should help in getting the RUI out of
+ * the AIL.
+ */
+STATIC uint
+xfs_rui_item_push(
+ struct xfs_log_item *lip,
+ struct list_head *buffer_list)
+{
+ return XFS_ITEM_PINNED;
+}
+
+/*
+ * The RUI has been either committed or aborted if the transaction has been
+ * cancelled. If the transaction was cancelled, an RUD isn't going to be
+ * constructed and thus we free the RUI here directly.
+ */
+STATIC void
+xfs_rui_item_unlock(
+ struct xfs_log_item *lip)
+{
+ if (lip->li_flags & XFS_LI_ABORTED)
+ xfs_rui_item_free(RUI_ITEM(lip));
+}
+
+/*
+ * The RUI is logged only once and cannot be moved in the log, so simply return
+ * the lsn at which it's been logged.
+ */
+STATIC xfs_lsn_t
+xfs_rui_item_committed(
+ struct xfs_log_item *lip,
+ xfs_lsn_t lsn)
+{
+ return lsn;
+}
+
+/*
+ * The RUI dependency tracking op doesn't do squat. It can't because
+ * it doesn't know where the free extent is coming from. The dependency
+ * tracking has to be handled by the "enclosing" metadata object. For
+ * example, for inodes, the inode is locked throughout the extent freeing
+ * so the dependency should be recorded there.
+ */
+STATIC void
+xfs_rui_item_committing(
+ struct xfs_log_item *lip,
+ xfs_lsn_t lsn)
+{
+}
+
+/*
+ * This is the ops vector shared by all rui log items.
+ */
+static const struct xfs_item_ops xfs_rui_item_ops = {
+ .iop_size = xfs_rui_item_size,
+ .iop_format = xfs_rui_item_format,
+ .iop_pin = xfs_rui_item_pin,
+ .iop_unpin = xfs_rui_item_unpin,
+ .iop_unlock = xfs_rui_item_unlock,
+ .iop_committed = xfs_rui_item_committed,
+ .iop_push = xfs_rui_item_push,
+ .iop_committing = xfs_rui_item_committing,
+};
+
+/*
+ * Allocate and initialize an rui item with the given number of extents.
+ */
+struct xfs_rui_log_item *
+xfs_rui_init(
+ struct xfs_mount *mp,
+ uint nextents)
+
+{
+ struct xfs_rui_log_item *ruip;
+ uint size;
+
+ ASSERT(nextents > 0);
+ if (nextents > XFS_RUI_MAX_FAST_EXTENTS) {
+ size = (uint)(sizeof(struct xfs_rui_log_item) +
+ ((nextents - 1) * sizeof(struct xfs_map_extent)));
+ ruip = kmem_zalloc(size, KM_SLEEP);
+ } else {
+ ruip = kmem_zone_zalloc(xfs_rui_zone, KM_SLEEP);
+ }
+
+ xfs_log_item_init(mp, &ruip->rui_item, XFS_LI_RUI, &xfs_rui_item_ops);
+ ruip->rui_format.rui_nextents = nextents;
+ ruip->rui_format.rui_id = (uintptr_t)(void *)ruip;
+ atomic_set(&ruip->rui_next_extent, 0);
+ atomic_set(&ruip->rui_refcount, 2);
+
+ return ruip;
+}
+
+/*
+ * Copy an RUI format buffer from the given buf, and into the destination
+ * RUI format structure. The RUI/RUD items were designed not to need any
+ * special alignment handling.
+ */
+int
+xfs_rui_copy_format(
+ struct xfs_log_iovec *buf,
+ struct xfs_rui_log_format *dst_rui_fmt)
+{
+ struct xfs_rui_log_format *src_rui_fmt;
+ uint len;
+
+ src_rui_fmt = buf->i_addr;
+ len = sizeof(struct xfs_rui_log_format) +
+ (src_rui_fmt->rui_nextents - 1) *
+ sizeof(struct xfs_map_extent);
+
+ if (buf->i_len != len)
+ return -EFSCORRUPTED;
+
+ memcpy((char *)dst_rui_fmt, (char *)src_rui_fmt, len);
+ return 0;
+}
+
+/*
+ * Freeing the RUI requires that we remove it from the AIL if it has already
+ * been placed there. However, the RUI may not yet have been placed in the AIL
+ * when called by xfs_rui_release() from RUD processing due to the ordering of
+ * committed vs unpin operations in bulk insert operations. Hence the reference
+ * count to ensure only the last caller frees the RUI.
+ */
+void
+xfs_rui_release(
+ struct xfs_rui_log_item *ruip)
+{
+ if (atomic_dec_and_test(&ruip->rui_refcount)) {
+ xfs_trans_ail_remove(&ruip->rui_item, SHUTDOWN_LOG_IO_ERROR);
+ xfs_rui_item_free(ruip);
+ }
+}
+
+static inline struct xfs_rud_log_item *RUD_ITEM(struct xfs_log_item *lip)
+{
+ return container_of(lip, struct xfs_rud_log_item, rud_item);
+}
+
+STATIC void
+xfs_rud_item_free(struct xfs_rud_log_item *rudp)
+{
+ if (rudp->rud_format.rud_nextents > XFS_RUD_MAX_FAST_EXTENTS)
+ kmem_free(rudp);
+ else
+ kmem_zone_free(xfs_rud_zone, rudp);
+}
+
+/*
+ * This returns the number of iovecs needed to log the given rud item.
+ * We only need 1 iovec for an rud item. It just logs the rud_log_format
+ * structure.
+ */
+static inline int
+xfs_rud_item_sizeof(
+ struct xfs_rud_log_item *rudp)
+{
+ return sizeof(struct xfs_rud_log_format) +
+ (rudp->rud_format.rud_nextents - 1) *
+ sizeof(struct xfs_map_extent);
+}
+
+STATIC void
+xfs_rud_item_size(
+ struct xfs_log_item *lip,
+ int *nvecs,
+ int *nbytes)
+{
+ *nvecs += 1;
+ *nbytes += xfs_rud_item_sizeof(RUD_ITEM(lip));
+}
+
+/*
+ * This is called to fill in the vector of log iovecs for the
+ * given rud log item. We use only 1 iovec, and we point that
+ * at the rud_log_format structure embedded in the rud item.
+ * It is at this point that we assert that all of the extent
+ * slots in the rud item have been filled.
+ */
+STATIC void
+xfs_rud_item_format(
+ struct xfs_log_item *lip,
+ struct xfs_log_vec *lv)
+{
+ struct xfs_rud_log_item *rudp = RUD_ITEM(lip);
+ struct xfs_log_iovec *vecp = NULL;
+
+ ASSERT(rudp->rud_next_extent == rudp->rud_format.rud_nextents);
+
+ rudp->rud_format.rud_type = XFS_LI_RUD;
+ rudp->rud_format.rud_size = 1;
+
+ xlog_copy_iovec(lv, &vecp, XLOG_REG_TYPE_RUD_FORMAT, &rudp->rud_format,
+ xfs_rud_item_sizeof(rudp));
+}
+
+/*
+ * Pinning has no meaning for an rud item, so just return.
+ */
+STATIC void
+xfs_rud_item_pin(
+ struct xfs_log_item *lip)
+{
+}
+
+/*
+ * Since pinning has no meaning for an rud item, unpinning does
+ * not either.
+ */
+STATIC void
+xfs_rud_item_unpin(
+ struct xfs_log_item *lip,
+ int remove)
+{
+}
+
+/*
+ * There isn't much you can do to push on an rud item. It is simply stuck
+ * waiting for the log to be flushed to disk.
+ */
+STATIC uint
+xfs_rud_item_push(
+ struct xfs_log_item *lip,
+ struct list_head *buffer_list)
+{
+ return XFS_ITEM_PINNED;
+}
+
+/*
+ * The RUD is either committed or aborted if the transaction is cancelled. If
+ * the transaction is cancelled, drop our reference to the RUI and free the
+ * RUD.
+ */
+STATIC void
+xfs_rud_item_unlock(
+ struct xfs_log_item *lip)
+{
+ struct xfs_rud_log_item *rudp = RUD_ITEM(lip);
+
+ if (lip->li_flags & XFS_LI_ABORTED) {
+ xfs_rui_release(rudp->rud_ruip);
+ xfs_rud_item_free(rudp);
+ }
+}
+
+/*
+ * When the rud item is committed to disk, all we need to do is delete our
+ * reference to our partner rui item and then free ourselves. Since we're
+ * freeing ourselves we must return -1 to keep the transaction code from
+ * further referencing this item.
+ */
+STATIC xfs_lsn_t
+xfs_rud_item_committed(
+ struct xfs_log_item *lip,
+ xfs_lsn_t lsn)
+{
+ struct xfs_rud_log_item *rudp = RUD_ITEM(lip);
+
+ /*
+ * Drop the RUI reference regardless of whether the RUD has been
+ * aborted. Once the RUD transaction is constructed, it is the sole
+ * responsibility of the RUD to release the RUI (even if the RUI is
+ * aborted due to log I/O error).
+ */
+ xfs_rui_release(rudp->rud_ruip);
+ xfs_rud_item_free(rudp);
+
+ return (xfs_lsn_t)-1;
+}
+
+/*
+ * The RUD dependency tracking op doesn't do squat. It can't because
+ * it doesn't know where the free extent is coming from. The dependency
+ * tracking has to be handled by the "enclosing" metadata object. For
+ * example, for inodes, the inode is locked throughout the extent freeing
+ * so the dependency should be recorded there.
+ */
+STATIC void
+xfs_rud_item_committing(
+ struct xfs_log_item *lip,
+ xfs_lsn_t lsn)
+{
+}
+
+/*
+ * This is the ops vector shared by all rud log items.
+ */
+static const struct xfs_item_ops xfs_rud_item_ops = {
+ .iop_size = xfs_rud_item_size,
+ .iop_format = xfs_rud_item_format,
+ .iop_pin = xfs_rud_item_pin,
+ .iop_unpin = xfs_rud_item_unpin,
+ .iop_unlock = xfs_rud_item_unlock,
+ .iop_committed = xfs_rud_item_committed,
+ .iop_push = xfs_rud_item_push,
+ .iop_committing = xfs_rud_item_committing,
+};
+
+/*
+ * Allocate and initialize an rud item with the given number of extents.
+ */
+struct xfs_rud_log_item *
+xfs_rud_init(
+ struct xfs_mount *mp,
+ struct xfs_rui_log_item *ruip,
+ uint nextents)
+
+{
+ struct xfs_rud_log_item *rudp;
+ uint size;
+
+ ASSERT(nextents > 0);
+ if (nextents > XFS_RUD_MAX_FAST_EXTENTS) {
+ size = (uint)(sizeof(struct xfs_rud_log_item) +
+ ((nextents - 1) * sizeof(struct xfs_map_extent)));
+ rudp = kmem_zalloc(size, KM_SLEEP);
+ } else {
+ rudp = kmem_zone_zalloc(xfs_rud_zone, KM_SLEEP);
+ }
+
+ xfs_log_item_init(mp, &rudp->rud_item, XFS_LI_RUD, &xfs_rud_item_ops);
+ rudp->rud_ruip = ruip;
+ rudp->rud_format.rud_nextents = nextents;
+ rudp->rud_format.rud_rui_id = ruip->rui_format.rui_id;
+
+ return rudp;
+}