linux/fs/xfs/xfs_bmap_item.c

711 lines
18 KiB
C
Raw Normal View History

// SPDX-License-Identifier: GPL-2.0+
/*
* Copyright (C) 2016 Oracle. All Rights Reserved.
* Author: Darrick J. Wong <darrick.wong@oracle.com>
*/
#include "xfs.h"
#include "xfs_fs.h"
#include "xfs_format.h"
#include "xfs_log_format.h"
#include "xfs_trans_resv.h"
#include "xfs_bit.h"
#include "xfs_shared.h"
#include "xfs_mount.h"
#include "xfs_defer.h"
#include "xfs_inode.h"
#include "xfs_trans.h"
#include "xfs_trans_priv.h"
#include "xfs_bmap_item.h"
#include "xfs_log.h"
#include "xfs_bmap.h"
#include "xfs_icache.h"
#include "xfs_bmap_btree.h"
#include "xfs_trans_space.h"
#include "xfs_error.h"
#include "xfs_log_priv.h"
#include "xfs_log_recover.h"
struct kmem_cache *xfs_bui_cache;
struct kmem_cache *xfs_bud_cache;
static const struct xfs_item_ops xfs_bui_item_ops;
static inline struct xfs_bui_log_item *BUI_ITEM(struct xfs_log_item *lip)
{
return container_of(lip, struct xfs_bui_log_item, bui_item);
}
STATIC void
xfs_bui_item_free(
struct xfs_bui_log_item *buip)
{
kmem_free(buip->bui_item.li_lv_shadow);
kmem_cache_free(xfs_bui_cache, buip);
}
/*
* Freeing the BUI requires that we remove it from the AIL if it has already
* been placed there. However, the BUI may not yet have been placed in the AIL
* when called by xfs_bui_release() from BUD 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 BUI.
*/
STATIC void
xfs_bui_release(
struct xfs_bui_log_item *buip)
{
ASSERT(atomic_read(&buip->bui_refcount) > 0);
if (!atomic_dec_and_test(&buip->bui_refcount))
return;
xfs_trans_ail_delete(&buip->bui_item, 0);
xfs_bui_item_free(buip);
}
STATIC void
xfs_bui_item_size(
struct xfs_log_item *lip,
int *nvecs,
int *nbytes)
{
struct xfs_bui_log_item *buip = BUI_ITEM(lip);
*nvecs += 1;
*nbytes += xfs_bui_log_format_sizeof(buip->bui_format.bui_nextents);
}
/*
* This is called to fill in the vector of log iovecs for the
* given bui log item. We use only 1 iovec, and we point that
* at the bui_log_format structure embedded in the bui item.
* It is at this point that we assert that all of the extent
* slots in the bui item have been filled.
*/
STATIC void
xfs_bui_item_format(
struct xfs_log_item *lip,
struct xfs_log_vec *lv)
{
struct xfs_bui_log_item *buip = BUI_ITEM(lip);
struct xfs_log_iovec *vecp = NULL;
ASSERT(atomic_read(&buip->bui_next_extent) ==
buip->bui_format.bui_nextents);
buip->bui_format.bui_type = XFS_LI_BUI;
buip->bui_format.bui_size = 1;
xlog_copy_iovec(lv, &vecp, XLOG_REG_TYPE_BUI_FORMAT, &buip->bui_format,
xfs_bui_log_format_sizeof(buip->bui_format.bui_nextents));
}
/*
* The unpin operation is the last place an BUI 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 BUI transaction has been successfully committed to make it
* this far. Therefore, we expect whoever committed the BUI to either construct
* and commit the BUD or drop the BUD's reference in the event of error. Simply
* drop the log's BUI reference now that the log is done with it.
*/
STATIC void
xfs_bui_item_unpin(
struct xfs_log_item *lip,
int remove)
{
struct xfs_bui_log_item *buip = BUI_ITEM(lip);
xfs_bui_release(buip);
}
/*
* The BUI has been either committed or aborted if the transaction has been
* cancelled. If the transaction was cancelled, an BUD isn't going to be
* constructed and thus we free the BUI here directly.
*/
STATIC void
xfs: split iop_unlock The iop_unlock method is called when comitting or cancelling a transaction. In the latter case, the transaction may or may not be aborted. While there is no known problem with the current code in practice, this implementation is limited in that any log item implementation that might want to differentiate between a commit and a cancellation must rely on the aborted state. The aborted bit is only set when the cancelled transaction is dirty, however. This means that there is no way to distinguish between a commit and a clean transaction cancellation. For example, intent log items currently rely on this distinction. The log item is either transferred to the CIL on commit or released on transaction cancel. There is currently no possibility for a clean intent log item in a transaction, but if that state is ever introduced a cancel of such a transaction will immediately result in memory leaks of the associated log item(s). This is an interface deficiency and landmine. To clean this up, replace the iop_unlock method with an iop_release method that is specific to transaction cancel. The existing iop_committing method occurs at the same time as iop_unlock in the commit path and there is no need for two separate callbacks here. Overload the iop_committing method with the current commit time iop_unlock implementations to eliminate the need for the latter and further simplify the interface. Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Brian Foster <bfoster@redhat.com> Reviewed-by: Darrick J. Wong <darrick.wong@oracle.com> Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
2019-06-29 10:27:32 +08:00
xfs_bui_item_release(
struct xfs_log_item *lip)
{
xfs: split iop_unlock The iop_unlock method is called when comitting or cancelling a transaction. In the latter case, the transaction may or may not be aborted. While there is no known problem with the current code in practice, this implementation is limited in that any log item implementation that might want to differentiate between a commit and a cancellation must rely on the aborted state. The aborted bit is only set when the cancelled transaction is dirty, however. This means that there is no way to distinguish between a commit and a clean transaction cancellation. For example, intent log items currently rely on this distinction. The log item is either transferred to the CIL on commit or released on transaction cancel. There is currently no possibility for a clean intent log item in a transaction, but if that state is ever introduced a cancel of such a transaction will immediately result in memory leaks of the associated log item(s). This is an interface deficiency and landmine. To clean this up, replace the iop_unlock method with an iop_release method that is specific to transaction cancel. The existing iop_committing method occurs at the same time as iop_unlock in the commit path and there is no need for two separate callbacks here. Overload the iop_committing method with the current commit time iop_unlock implementations to eliminate the need for the latter and further simplify the interface. Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Brian Foster <bfoster@redhat.com> Reviewed-by: Darrick J. Wong <darrick.wong@oracle.com> Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
2019-06-29 10:27:32 +08:00
xfs_bui_release(BUI_ITEM(lip));
}
/*
* Allocate and initialize an bui item with the given number of extents.
*/
STATIC struct xfs_bui_log_item *
xfs_bui_init(
struct xfs_mount *mp)
{
struct xfs_bui_log_item *buip;
buip = kmem_cache_zalloc(xfs_bui_cache, GFP_KERNEL | __GFP_NOFAIL);
xfs_log_item_init(mp, &buip->bui_item, XFS_LI_BUI, &xfs_bui_item_ops);
buip->bui_format.bui_nextents = XFS_BUI_MAX_FAST_EXTENTS;
buip->bui_format.bui_id = (uintptr_t)(void *)buip;
atomic_set(&buip->bui_next_extent, 0);
atomic_set(&buip->bui_refcount, 2);
return buip;
}
static inline struct xfs_bud_log_item *BUD_ITEM(struct xfs_log_item *lip)
{
return container_of(lip, struct xfs_bud_log_item, bud_item);
}
STATIC void
xfs_bud_item_size(
struct xfs_log_item *lip,
int *nvecs,
int *nbytes)
{
*nvecs += 1;
*nbytes += sizeof(struct xfs_bud_log_format);
}
/*
* This is called to fill in the vector of log iovecs for the
* given bud log item. We use only 1 iovec, and we point that
* at the bud_log_format structure embedded in the bud item.
* It is at this point that we assert that all of the extent
* slots in the bud item have been filled.
*/
STATIC void
xfs_bud_item_format(
struct xfs_log_item *lip,
struct xfs_log_vec *lv)
{
struct xfs_bud_log_item *budp = BUD_ITEM(lip);
struct xfs_log_iovec *vecp = NULL;
budp->bud_format.bud_type = XFS_LI_BUD;
budp->bud_format.bud_size = 1;
xlog_copy_iovec(lv, &vecp, XLOG_REG_TYPE_BUD_FORMAT, &budp->bud_format,
sizeof(struct xfs_bud_log_format));
}
/*
* The BUD is either committed or aborted if the transaction is cancelled. If
* the transaction is cancelled, drop our reference to the BUI and free the
* BUD.
*/
STATIC void
xfs: split iop_unlock The iop_unlock method is called when comitting or cancelling a transaction. In the latter case, the transaction may or may not be aborted. While there is no known problem with the current code in practice, this implementation is limited in that any log item implementation that might want to differentiate between a commit and a cancellation must rely on the aborted state. The aborted bit is only set when the cancelled transaction is dirty, however. This means that there is no way to distinguish between a commit and a clean transaction cancellation. For example, intent log items currently rely on this distinction. The log item is either transferred to the CIL on commit or released on transaction cancel. There is currently no possibility for a clean intent log item in a transaction, but if that state is ever introduced a cancel of such a transaction will immediately result in memory leaks of the associated log item(s). This is an interface deficiency and landmine. To clean this up, replace the iop_unlock method with an iop_release method that is specific to transaction cancel. The existing iop_committing method occurs at the same time as iop_unlock in the commit path and there is no need for two separate callbacks here. Overload the iop_committing method with the current commit time iop_unlock implementations to eliminate the need for the latter and further simplify the interface. Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Brian Foster <bfoster@redhat.com> Reviewed-by: Darrick J. Wong <darrick.wong@oracle.com> Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
2019-06-29 10:27:32 +08:00
xfs_bud_item_release(
struct xfs_log_item *lip)
{
struct xfs_bud_log_item *budp = BUD_ITEM(lip);
xfs: split iop_unlock The iop_unlock method is called when comitting or cancelling a transaction. In the latter case, the transaction may or may not be aborted. While there is no known problem with the current code in practice, this implementation is limited in that any log item implementation that might want to differentiate between a commit and a cancellation must rely on the aborted state. The aborted bit is only set when the cancelled transaction is dirty, however. This means that there is no way to distinguish between a commit and a clean transaction cancellation. For example, intent log items currently rely on this distinction. The log item is either transferred to the CIL on commit or released on transaction cancel. There is currently no possibility for a clean intent log item in a transaction, but if that state is ever introduced a cancel of such a transaction will immediately result in memory leaks of the associated log item(s). This is an interface deficiency and landmine. To clean this up, replace the iop_unlock method with an iop_release method that is specific to transaction cancel. The existing iop_committing method occurs at the same time as iop_unlock in the commit path and there is no need for two separate callbacks here. Overload the iop_committing method with the current commit time iop_unlock implementations to eliminate the need for the latter and further simplify the interface. Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Brian Foster <bfoster@redhat.com> Reviewed-by: Darrick J. Wong <darrick.wong@oracle.com> Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
2019-06-29 10:27:32 +08:00
xfs_bui_release(budp->bud_buip);
kmem_free(budp->bud_item.li_lv_shadow);
kmem_cache_free(xfs_bud_cache, budp);
}
static struct xfs_log_item *
xfs_bud_item_intent(
struct xfs_log_item *lip)
{
return &BUD_ITEM(lip)->bud_buip->bui_item;
}
static const struct xfs_item_ops xfs_bud_item_ops = {
.flags = XFS_ITEM_RELEASE_WHEN_COMMITTED |
XFS_ITEM_INTENT_DONE,
.iop_size = xfs_bud_item_size,
.iop_format = xfs_bud_item_format,
xfs: split iop_unlock The iop_unlock method is called when comitting or cancelling a transaction. In the latter case, the transaction may or may not be aborted. While there is no known problem with the current code in practice, this implementation is limited in that any log item implementation that might want to differentiate between a commit and a cancellation must rely on the aborted state. The aborted bit is only set when the cancelled transaction is dirty, however. This means that there is no way to distinguish between a commit and a clean transaction cancellation. For example, intent log items currently rely on this distinction. The log item is either transferred to the CIL on commit or released on transaction cancel. There is currently no possibility for a clean intent log item in a transaction, but if that state is ever introduced a cancel of such a transaction will immediately result in memory leaks of the associated log item(s). This is an interface deficiency and landmine. To clean this up, replace the iop_unlock method with an iop_release method that is specific to transaction cancel. The existing iop_committing method occurs at the same time as iop_unlock in the commit path and there is no need for two separate callbacks here. Overload the iop_committing method with the current commit time iop_unlock implementations to eliminate the need for the latter and further simplify the interface. Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Brian Foster <bfoster@redhat.com> Reviewed-by: Darrick J. Wong <darrick.wong@oracle.com> Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
2019-06-29 10:27:32 +08:00
.iop_release = xfs_bud_item_release,
.iop_intent = xfs_bud_item_intent,
};
static struct xfs_bud_log_item *
xfs_trans_get_bud(
struct xfs_trans *tp,
struct xfs_bui_log_item *buip)
{
struct xfs_bud_log_item *budp;
budp = kmem_cache_zalloc(xfs_bud_cache, GFP_KERNEL | __GFP_NOFAIL);
xfs_log_item_init(tp->t_mountp, &budp->bud_item, XFS_LI_BUD,
&xfs_bud_item_ops);
budp->bud_buip = buip;
budp->bud_format.bud_bui_id = buip->bui_format.bui_id;
xfs_trans_add_item(tp, &budp->bud_item);
return budp;
}
/*
* Finish an bmap update and log it to the BUD. Note that the
* transaction is marked dirty regardless of whether the bmap update
* succeeds or fails to support the BUI/BUD lifecycle rules.
*/
static int
xfs_trans_log_finish_bmap_update(
struct xfs_trans *tp,
struct xfs_bud_log_item *budp,
enum xfs_bmap_intent_type type,
struct xfs_inode *ip,
int whichfork,
xfs_fileoff_t startoff,
xfs_fsblock_t startblock,
xfs_filblks_t *blockcount,
xfs_exntst_t state)
{
int error;
error = xfs_bmap_finish_one(tp, ip, type, whichfork, startoff,
startblock, blockcount, state);
/*
* Mark the transaction dirty, even on error. This ensures the
* transaction is aborted, which:
*
* 1.) releases the BUI and frees the BUD
* 2.) shuts down the filesystem
*/
tp->t_flags |= XFS_TRANS_DIRTY | XFS_TRANS_HAS_INTENT_DONE;
set_bit(XFS_LI_DIRTY, &budp->bud_item.li_flags);
return error;
}
/* Sort bmap intents by inode. */
static int
xfs_bmap_update_diff_items(
void *priv,
const struct list_head *a,
const struct list_head *b)
{
struct xfs_bmap_intent *ba;
struct xfs_bmap_intent *bb;
ba = container_of(a, struct xfs_bmap_intent, bi_list);
bb = container_of(b, struct xfs_bmap_intent, bi_list);
return ba->bi_owner->i_ino - bb->bi_owner->i_ino;
}
/* Set the map extent flags for this mapping. */
static void
xfs_trans_set_bmap_flags(
struct xfs_map_extent *bmap,
enum xfs_bmap_intent_type type,
int whichfork,
xfs_exntst_t state)
{
bmap->me_flags = 0;
switch (type) {
case XFS_BMAP_MAP:
case XFS_BMAP_UNMAP:
bmap->me_flags = type;
break;
default:
ASSERT(0);
}
if (state == XFS_EXT_UNWRITTEN)
bmap->me_flags |= XFS_BMAP_EXTENT_UNWRITTEN;
if (whichfork == XFS_ATTR_FORK)
bmap->me_flags |= XFS_BMAP_EXTENT_ATTR_FORK;
}
/* Log bmap updates in the intent item. */
STATIC void
xfs_bmap_update_log_item(
struct xfs_trans *tp,
struct xfs_bui_log_item *buip,
struct xfs_bmap_intent *bmap)
{
uint next_extent;
struct xfs_map_extent *map;
tp->t_flags |= XFS_TRANS_DIRTY;
set_bit(XFS_LI_DIRTY, &buip->bui_item.li_flags);
/*
* atomic_inc_return gives us the value after the increment;
* we want to use it as an array index so we need to subtract 1 from
* it.
*/
next_extent = atomic_inc_return(&buip->bui_next_extent) - 1;
ASSERT(next_extent < buip->bui_format.bui_nextents);
map = &buip->bui_format.bui_extents[next_extent];
map->me_owner = bmap->bi_owner->i_ino;
map->me_startblock = bmap->bi_bmap.br_startblock;
map->me_startoff = bmap->bi_bmap.br_startoff;
map->me_len = bmap->bi_bmap.br_blockcount;
xfs_trans_set_bmap_flags(map, bmap->bi_type, bmap->bi_whichfork,
bmap->bi_bmap.br_state);
}
static struct xfs_log_item *
xfs_bmap_update_create_intent(
struct xfs_trans *tp,
struct list_head *items,
unsigned int count,
bool sort)
{
struct xfs_mount *mp = tp->t_mountp;
struct xfs_bui_log_item *buip = xfs_bui_init(mp);
struct xfs_bmap_intent *bmap;
ASSERT(count == XFS_BUI_MAX_FAST_EXTENTS);
xfs_trans_add_item(tp, &buip->bui_item);
if (sort)
list_sort(mp, items, xfs_bmap_update_diff_items);
list_for_each_entry(bmap, items, bi_list)
xfs_bmap_update_log_item(tp, buip, bmap);
return &buip->bui_item;
}
/* Get an BUD so we can process all the deferred rmap updates. */
static struct xfs_log_item *
xfs_bmap_update_create_done(
struct xfs_trans *tp,
struct xfs_log_item *intent,
unsigned int count)
{
return &xfs_trans_get_bud(tp, BUI_ITEM(intent))->bud_item;
}
/* Process a deferred rmap update. */
STATIC int
xfs_bmap_update_finish_item(
struct xfs_trans *tp,
struct xfs_log_item *done,
struct list_head *item,
struct xfs_btree_cur **state)
{
struct xfs_bmap_intent *bmap;
xfs_filblks_t count;
int error;
bmap = container_of(item, struct xfs_bmap_intent, bi_list);
count = bmap->bi_bmap.br_blockcount;
error = xfs_trans_log_finish_bmap_update(tp, BUD_ITEM(done),
bmap->bi_type,
bmap->bi_owner, bmap->bi_whichfork,
bmap->bi_bmap.br_startoff,
bmap->bi_bmap.br_startblock,
&count,
bmap->bi_bmap.br_state);
if (!error && count > 0) {
ASSERT(bmap->bi_type == XFS_BMAP_UNMAP);
bmap->bi_bmap.br_blockcount = count;
return -EAGAIN;
}
kmem_cache_free(xfs_bmap_intent_cache, bmap);
return error;
}
/* Abort all pending BUIs. */
STATIC void
xfs_bmap_update_abort_intent(
struct xfs_log_item *intent)
{
xfs_bui_release(BUI_ITEM(intent));
}
/* Cancel a deferred rmap update. */
STATIC void
xfs_bmap_update_cancel_item(
struct list_head *item)
{
struct xfs_bmap_intent *bmap;
bmap = container_of(item, struct xfs_bmap_intent, bi_list);
kmem_cache_free(xfs_bmap_intent_cache, bmap);
}
const struct xfs_defer_op_type xfs_bmap_update_defer_type = {
.max_items = XFS_BUI_MAX_FAST_EXTENTS,
.create_intent = xfs_bmap_update_create_intent,
.abort_intent = xfs_bmap_update_abort_intent,
.create_done = xfs_bmap_update_create_done,
.finish_item = xfs_bmap_update_finish_item,
.cancel_item = xfs_bmap_update_cancel_item,
};
/* Is this recovered BUI ok? */
static inline bool
xfs_bui_validate(
struct xfs_mount *mp,
struct xfs_bui_log_item *buip)
{
struct xfs_map_extent *bmap;
/* Only one mapping operation per BUI... */
if (buip->bui_format.bui_nextents != XFS_BUI_MAX_FAST_EXTENTS)
return false;
bmap = &buip->bui_format.bui_extents[0];
if (bmap->me_flags & ~XFS_BMAP_EXTENT_FLAGS)
return false;
switch (bmap->me_flags & XFS_BMAP_EXTENT_TYPE_MASK) {
case XFS_BMAP_MAP:
case XFS_BMAP_UNMAP:
break;
default:
return false;
}
if (!xfs_verify_ino(mp, bmap->me_owner))
return false;
if (!xfs_verify_fileext(mp, bmap->me_startoff, bmap->me_len))
return false;
return xfs_verify_fsbext(mp, bmap->me_startblock, bmap->me_len);
}
/*
* Process a bmap update intent item that was recovered from the log.
* We need to update some inode's bmbt.
*/
STATIC int
xfs_bui_item_recover(
struct xfs_log_item *lip,
xfs: proper replay of deferred ops queued during log recovery When we replay unfinished intent items that have been recovered from the log, it's possible that the replay will cause the creation of more deferred work items. As outlined in commit 509955823cc9c ("xfs: log recovery should replay deferred ops in order"), later work items have an implicit ordering dependency on earlier work items. Therefore, recovery must replay the items (both recovered and created) in the same order that they would have been during normal operation. For log recovery, we enforce this ordering by using an empty transaction to collect deferred ops that get created in the process of recovering a log intent item to prevent them from being committed before the rest of the recovered intent items. After we finish committing all the recovered log items, we allocate a transaction with an enormous block reservation, splice our huge list of created deferred ops into that transaction, and commit it, thereby finishing all those ops. This is /really/ hokey -- it's the one place in XFS where we allow nested transactions; the splicing of the defer ops list is is inelegant and has to be done twice per recovery function; and the broken way we handle inode pointers and block reservations cause subtle use-after-free and allocator problems that will be fixed by this patch and the two patches after it. Therefore, replace the hokey empty transaction with a structure designed to capture each chain of deferred ops that are created as part of recovering a single unfinished log intent. Finally, refactor the loop that replays those chains to do so using one transaction per chain. Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com> Reviewed-by: Brian Foster <bfoster@redhat.com> Reviewed-by: Christoph Hellwig <hch@lst.de>
2020-09-26 08:39:37 +08:00
struct list_head *capture_list)
{
struct xfs_bmbt_irec irec;
struct xfs_bui_log_item *buip = BUI_ITEM(lip);
struct xfs_trans *tp;
struct xfs_inode *ip = NULL;
struct xfs_mount *mp = lip->li_log->l_mp;
struct xfs_map_extent *bmap;
struct xfs_bud_log_item *budp;
xfs_filblks_t count;
xfs_exntst_t state;
unsigned int bui_type;
int whichfork;
int iext_delta;
int error = 0;
if (!xfs_bui_validate(mp, buip)) {
XFS_CORRUPTION_ERROR(__func__, XFS_ERRLEVEL_LOW, mp,
&buip->bui_format, sizeof(buip->bui_format));
return -EFSCORRUPTED;
}
bmap = &buip->bui_format.bui_extents[0];
state = (bmap->me_flags & XFS_BMAP_EXTENT_UNWRITTEN) ?
XFS_EXT_UNWRITTEN : XFS_EXT_NORM;
whichfork = (bmap->me_flags & XFS_BMAP_EXTENT_ATTR_FORK) ?
XFS_ATTR_FORK : XFS_DATA_FORK;
bui_type = bmap->me_flags & XFS_BMAP_EXTENT_TYPE_MASK;
error = xlog_recover_iget(mp, bmap->me_owner, &ip);
if (error)
return error;
/* Allocate transaction and do the work. */
error = xfs_trans_alloc(mp, &M_RES(mp)->tr_itruncate,
XFS_EXTENTADD_SPACE_RES(mp, XFS_DATA_FORK), 0, 0, &tp);
if (error)
goto err_rele;
budp = xfs_trans_get_bud(tp, buip);
xfs_ilock(ip, XFS_ILOCK_EXCL);
xfs_trans_ijoin(tp, ip, 0);
if (bui_type == XFS_BMAP_MAP)
iext_delta = XFS_IEXT_ADD_NOSPLIT_CNT;
else
iext_delta = XFS_IEXT_PUNCH_HOLE_CNT;
error = xfs_iext_count_may_overflow(ip, whichfork, iext_delta);
if (error == -EFBIG)
error = xfs_iext_count_upgrade(tp, ip, iext_delta);
if (error)
goto err_cancel;
count = bmap->me_len;
error = xfs_trans_log_finish_bmap_update(tp, budp, bui_type, ip,
whichfork, bmap->me_startoff, bmap->me_startblock,
&count, state);
if (error == -EFSCORRUPTED)
XFS_CORRUPTION_ERROR(__func__, XFS_ERRLEVEL_LOW, mp, bmap,
sizeof(*bmap));
if (error)
goto err_cancel;
if (count > 0) {
ASSERT(bui_type == XFS_BMAP_UNMAP);
irec.br_startblock = bmap->me_startblock;
irec.br_blockcount = count;
irec.br_startoff = bmap->me_startoff;
irec.br_state = state;
xfs_bmap_unmap_extent(tp, ip, &irec);
}
/*
* Commit transaction, which frees the transaction and saves the inode
* for later replay activities.
*/
error = xfs_defer_ops_capture_and_commit(tp, capture_list);
if (error)
goto err_unlock;
xfs_iunlock(ip, XFS_ILOCK_EXCL);
xfs_irele(ip);
return 0;
err_cancel:
xfs_trans_cancel(tp);
err_unlock:
xfs_iunlock(ip, XFS_ILOCK_EXCL);
err_rele:
xfs_irele(ip);
return error;
}
STATIC bool
xfs_bui_item_match(
struct xfs_log_item *lip,
uint64_t intent_id)
{
return BUI_ITEM(lip)->bui_format.bui_id == intent_id;
}
xfs: periodically relog deferred intent items There's a subtle design flaw in the deferred log item code that can lead to pinning the log tail. Taking up the defer ops chain examples from the previous commit, we can get trapped in sequences like this: Caller hands us a transaction t0 with D0-D3 attached. The defer ops chain will look like the following if the transaction rolls succeed: t1: D0(t0), D1(t0), D2(t0), D3(t0) t2: d4(t1), d5(t1), D1(t0), D2(t0), D3(t0) t3: d5(t1), D1(t0), D2(t0), D3(t0) ... t9: d9(t7), D3(t0) t10: D3(t0) t11: d10(t10), d11(t10) t12: d11(t10) In transaction 9, we finish d9 and try to roll to t10 while holding onto an intent item for D3 that we logged in t0. The previous commit changed the order in which we place new defer ops in the defer ops processing chain to reduce the maximum chain length. Now make xfs_defer_finish_noroll capable of relogging the entire chain periodically so that we can always move the log tail forward. Most chains will never get relogged, except for operations that generate very long chains (large extents containing many blocks with different sharing levels) or are on filesystems with small logs and a lot of ongoing metadata updates. Callers are now required to ensure that the transaction reservation is large enough to handle logging done items and new intent items for the maximum possible chain length. Most callers are careful to keep the chain lengths low, so the overhead should be minimal. The decision to relog an intent item is made based on whether the intent was logged in a previous checkpoint, since there's no point in relogging an intent into the same checkpoint. Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com> Reviewed-by: Brian Foster <bfoster@redhat.com>
2020-09-28 07:18:13 +08:00
/* Relog an intent item to push the log tail forward. */
static struct xfs_log_item *
xfs_bui_item_relog(
struct xfs_log_item *intent,
struct xfs_trans *tp)
{
struct xfs_bud_log_item *budp;
struct xfs_bui_log_item *buip;
struct xfs_map_extent *extp;
unsigned int count;
count = BUI_ITEM(intent)->bui_format.bui_nextents;
extp = BUI_ITEM(intent)->bui_format.bui_extents;
tp->t_flags |= XFS_TRANS_DIRTY;
budp = xfs_trans_get_bud(tp, BUI_ITEM(intent));
set_bit(XFS_LI_DIRTY, &budp->bud_item.li_flags);
buip = xfs_bui_init(tp->t_mountp);
memcpy(buip->bui_format.bui_extents, extp, count * sizeof(*extp));
atomic_set(&buip->bui_next_extent, count);
xfs_trans_add_item(tp, &buip->bui_item);
set_bit(XFS_LI_DIRTY, &buip->bui_item.li_flags);
return &buip->bui_item;
}
static const struct xfs_item_ops xfs_bui_item_ops = {
.flags = XFS_ITEM_INTENT,
.iop_size = xfs_bui_item_size,
.iop_format = xfs_bui_item_format,
.iop_unpin = xfs_bui_item_unpin,
.iop_release = xfs_bui_item_release,
.iop_recover = xfs_bui_item_recover,
.iop_match = xfs_bui_item_match,
xfs: periodically relog deferred intent items There's a subtle design flaw in the deferred log item code that can lead to pinning the log tail. Taking up the defer ops chain examples from the previous commit, we can get trapped in sequences like this: Caller hands us a transaction t0 with D0-D3 attached. The defer ops chain will look like the following if the transaction rolls succeed: t1: D0(t0), D1(t0), D2(t0), D3(t0) t2: d4(t1), d5(t1), D1(t0), D2(t0), D3(t0) t3: d5(t1), D1(t0), D2(t0), D3(t0) ... t9: d9(t7), D3(t0) t10: D3(t0) t11: d10(t10), d11(t10) t12: d11(t10) In transaction 9, we finish d9 and try to roll to t10 while holding onto an intent item for D3 that we logged in t0. The previous commit changed the order in which we place new defer ops in the defer ops processing chain to reduce the maximum chain length. Now make xfs_defer_finish_noroll capable of relogging the entire chain periodically so that we can always move the log tail forward. Most chains will never get relogged, except for operations that generate very long chains (large extents containing many blocks with different sharing levels) or are on filesystems with small logs and a lot of ongoing metadata updates. Callers are now required to ensure that the transaction reservation is large enough to handle logging done items and new intent items for the maximum possible chain length. Most callers are careful to keep the chain lengths low, so the overhead should be minimal. The decision to relog an intent item is made based on whether the intent was logged in a previous checkpoint, since there's no point in relogging an intent into the same checkpoint. Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com> Reviewed-by: Brian Foster <bfoster@redhat.com>
2020-09-28 07:18:13 +08:00
.iop_relog = xfs_bui_item_relog,
};
/*
* Copy an BUI format buffer from the given buf, and into the destination
* BUI format structure. The BUI/BUD items were designed not to need any
* special alignment handling.
*/
static int
xfs_bui_copy_format(
struct xfs_log_iovec *buf,
struct xfs_bui_log_format *dst_bui_fmt)
{
struct xfs_bui_log_format *src_bui_fmt;
uint len;
src_bui_fmt = buf->i_addr;
len = xfs_bui_log_format_sizeof(src_bui_fmt->bui_nextents);
if (buf->i_len == len) {
memcpy(dst_bui_fmt, src_bui_fmt, len);
return 0;
}
XFS_ERROR_REPORT(__func__, XFS_ERRLEVEL_LOW, NULL);
return -EFSCORRUPTED;
}
/*
* This routine is called to create an in-core extent bmap update
* item from the bui format structure which was logged on disk.
* It allocates an in-core bui, copies the extents from the format
* structure into it, and adds the bui to the AIL with the given
* LSN.
*/
STATIC int
xlog_recover_bui_commit_pass2(
struct xlog *log,
struct list_head *buffer_list,
struct xlog_recover_item *item,
xfs_lsn_t lsn)
{
int error;
struct xfs_mount *mp = log->l_mp;
struct xfs_bui_log_item *buip;
struct xfs_bui_log_format *bui_formatp;
bui_formatp = item->ri_buf[0].i_addr;
if (bui_formatp->bui_nextents != XFS_BUI_MAX_FAST_EXTENTS) {
XFS_ERROR_REPORT(__func__, XFS_ERRLEVEL_LOW, log->l_mp);
return -EFSCORRUPTED;
}
buip = xfs_bui_init(mp);
error = xfs_bui_copy_format(&item->ri_buf[0], &buip->bui_format);
if (error) {
xfs_bui_item_free(buip);
return error;
}
atomic_set(&buip->bui_next_extent, bui_formatp->bui_nextents);
/*
* Insert the intent into the AIL directly and drop one reference so
* that finishing or canceling the work will drop the other.
*/
xfs_trans_ail_insert(log->l_ailp, &buip->bui_item, lsn);
xfs_bui_release(buip);
return 0;
}
const struct xlog_recover_item_ops xlog_bui_item_ops = {
.item_type = XFS_LI_BUI,
.commit_pass2 = xlog_recover_bui_commit_pass2,
};
/*
* This routine is called when an BUD format structure is found in a committed
* transaction in the log. Its purpose is to cancel the corresponding BUI if it
* was still in the log. To do this it searches the AIL for the BUI with an id
* equal to that in the BUD format structure. If we find it we drop the BUD
* reference, which removes the BUI from the AIL and frees it.
*/
STATIC int
xlog_recover_bud_commit_pass2(
struct xlog *log,
struct list_head *buffer_list,
struct xlog_recover_item *item,
xfs_lsn_t lsn)
{
struct xfs_bud_log_format *bud_formatp;
bud_formatp = item->ri_buf[0].i_addr;
if (item->ri_buf[0].i_len != sizeof(struct xfs_bud_log_format)) {
XFS_ERROR_REPORT(__func__, XFS_ERRLEVEL_LOW, log->l_mp);
return -EFSCORRUPTED;
}
xlog_recover_release_intent(log, XFS_LI_BUI, bud_formatp->bud_bui_id);
return 0;
}
const struct xlog_recover_item_ops xlog_bud_item_ops = {
.item_type = XFS_LI_BUD,
.commit_pass2 = xlog_recover_bud_commit_pass2,
};