Commit 50995582 authored by Darrick J. Wong's avatar Darrick J. Wong

xfs: log recovery should replay deferred ops in order

As part of testing log recovery with dm_log_writes, Amir Goldstein
discovered an error in the deferred ops recovery that lead to corruption
of the filesystem metadata if a reflink+rmap filesystem happened to shut
down midway through a CoW remap:

"This is what happens [after failed log recovery]:

"Phase 1 - find and verify superblock...
"Phase 2 - using internal log
"        - zero log...
"        - scan filesystem freespace and inode maps...
"        - found root inode chunk
"Phase 3 - for each AG...
"        - scan (but don't clear) agi unlinked lists...
"        - process known inodes and perform inode discovery...
"        - agno = 0
"data fork in regular inode 134 claims CoW block 376
"correcting nextents for inode 134
"bad data fork in inode 134
"would have cleared inode 134"

Hou Tao dissected the log contents of exactly such a crash:

"According to the implementation of xfs_defer_finish(), these ops should
be completed in the following sequence:

"Have been done:
"(1) CUI: Oper (160)
"(2) BUI: Oper (161)
"(3) CUD: Oper (194), for CUI Oper (160)
"(4) RUI A: Oper (197), free rmap [0x155, 2, -9]

"Should be done:
"(5) BUD: for BUI Oper (161)
"(6) RUI B: add rmap [0x155, 2, 137]
"(7) RUD: for RUI A
"(8) RUD: for RUI B

"Actually be done by xlog_recover_process_intents()
"(5) BUD: for BUI Oper (161)
"(6) RUI B: add rmap [0x155, 2, 137]
"(7) RUD: for RUI B
"(8) RUD: for RUI A

"So the rmap entry [0x155, 2, -9] for COW should be freed firstly,
then a new rmap entry [0x155, 2, 137] will be added. However, as we can see
from the log record in post_mount.log (generated after umount) and the trace
print, the new rmap entry [0x155, 2, 137] are added firstly, then the rmap
entry [0x155, 2, -9] are freed."

When reconstructing the internal log state from the log items found on
disk, it's required that deferred ops replay in exactly the same order
that they would have had the filesystem not gone down.  However,
replaying unfinished deferred ops can create /more/ deferred ops.  These
new deferred ops are finished in the wrong order.  This causes fs
corruption and replay crashes, so let's create a single defer_ops to
handle the subsequent ops created during replay, then use one single
transaction at the end of log recovery to ensure that everything is
replayed in the same order as they're supposed to be.
Reported-by: default avatarAmir Goldstein <amir73il@gmail.com>
Analyzed-by: default avatarHou Tao <houtao1@huawei.com>
Reviewed-by: default avatarChristoph Hellwig <hch@lst.de>
Tested-by: default avatarAmir Goldstein <amir73il@gmail.com>
Signed-off-by: default avatarDarrick J. Wong <darrick.wong@oracle.com>
parent 98c4f78d
...@@ -389,7 +389,8 @@ xfs_bud_init( ...@@ -389,7 +389,8 @@ xfs_bud_init(
int int
xfs_bui_recover( xfs_bui_recover(
struct xfs_mount *mp, struct xfs_mount *mp,
struct xfs_bui_log_item *buip) struct xfs_bui_log_item *buip,
struct xfs_defer_ops *dfops)
{ {
int error = 0; int error = 0;
unsigned int bui_type; unsigned int bui_type;
...@@ -404,9 +405,7 @@ xfs_bui_recover( ...@@ -404,9 +405,7 @@ xfs_bui_recover(
xfs_exntst_t state; xfs_exntst_t state;
struct xfs_trans *tp; struct xfs_trans *tp;
struct xfs_inode *ip = NULL; struct xfs_inode *ip = NULL;
struct xfs_defer_ops dfops;
struct xfs_bmbt_irec irec; struct xfs_bmbt_irec irec;
xfs_fsblock_t firstfsb;
ASSERT(!test_bit(XFS_BUI_RECOVERED, &buip->bui_flags)); ASSERT(!test_bit(XFS_BUI_RECOVERED, &buip->bui_flags));
...@@ -464,7 +463,6 @@ xfs_bui_recover( ...@@ -464,7 +463,6 @@ xfs_bui_recover(
if (VFS_I(ip)->i_nlink == 0) if (VFS_I(ip)->i_nlink == 0)
xfs_iflags_set(ip, XFS_IRECOVERY); xfs_iflags_set(ip, XFS_IRECOVERY);
xfs_defer_init(&dfops, &firstfsb);
/* Process deferred bmap item. */ /* Process deferred bmap item. */
state = (bmap->me_flags & XFS_BMAP_EXTENT_UNWRITTEN) ? state = (bmap->me_flags & XFS_BMAP_EXTENT_UNWRITTEN) ?
...@@ -479,16 +477,16 @@ xfs_bui_recover( ...@@ -479,16 +477,16 @@ xfs_bui_recover(
break; break;
default: default:
error = -EFSCORRUPTED; error = -EFSCORRUPTED;
goto err_dfops; goto err_inode;
} }
xfs_trans_ijoin(tp, ip, 0); xfs_trans_ijoin(tp, ip, 0);
count = bmap->me_len; count = bmap->me_len;
error = xfs_trans_log_finish_bmap_update(tp, budp, &dfops, type, error = xfs_trans_log_finish_bmap_update(tp, budp, dfops, type,
ip, whichfork, bmap->me_startoff, ip, whichfork, bmap->me_startoff,
bmap->me_startblock, &count, state); bmap->me_startblock, &count, state);
if (error) if (error)
goto err_dfops; goto err_inode;
if (count > 0) { if (count > 0) {
ASSERT(type == XFS_BMAP_UNMAP); ASSERT(type == XFS_BMAP_UNMAP);
...@@ -496,16 +494,11 @@ xfs_bui_recover( ...@@ -496,16 +494,11 @@ xfs_bui_recover(
irec.br_blockcount = count; irec.br_blockcount = count;
irec.br_startoff = bmap->me_startoff; irec.br_startoff = bmap->me_startoff;
irec.br_state = state; irec.br_state = state;
error = xfs_bmap_unmap_extent(tp->t_mountp, &dfops, ip, &irec); error = xfs_bmap_unmap_extent(tp->t_mountp, dfops, ip, &irec);
if (error) if (error)
goto err_dfops; goto err_inode;
} }
/* Finish transaction, free inodes. */
error = xfs_defer_finish(&tp, &dfops);
if (error)
goto err_dfops;
set_bit(XFS_BUI_RECOVERED, &buip->bui_flags); set_bit(XFS_BUI_RECOVERED, &buip->bui_flags);
error = xfs_trans_commit(tp); error = xfs_trans_commit(tp);
xfs_iunlock(ip, XFS_ILOCK_EXCL); xfs_iunlock(ip, XFS_ILOCK_EXCL);
...@@ -513,8 +506,6 @@ xfs_bui_recover( ...@@ -513,8 +506,6 @@ xfs_bui_recover(
return error; return error;
err_dfops:
xfs_defer_cancel(&dfops);
err_inode: err_inode:
xfs_trans_cancel(tp); xfs_trans_cancel(tp);
if (ip) { if (ip) {
......
...@@ -93,6 +93,7 @@ struct xfs_bud_log_item *xfs_bud_init(struct xfs_mount *, ...@@ -93,6 +93,7 @@ struct xfs_bud_log_item *xfs_bud_init(struct xfs_mount *,
struct xfs_bui_log_item *); struct xfs_bui_log_item *);
void xfs_bui_item_free(struct xfs_bui_log_item *); void xfs_bui_item_free(struct xfs_bui_log_item *);
void xfs_bui_release(struct xfs_bui_log_item *); void xfs_bui_release(struct xfs_bui_log_item *);
int xfs_bui_recover(struct xfs_mount *mp, struct xfs_bui_log_item *buip); int xfs_bui_recover(struct xfs_mount *mp, struct xfs_bui_log_item *buip,
struct xfs_defer_ops *dfops);
#endif /* __XFS_BMAP_ITEM_H__ */ #endif /* __XFS_BMAP_ITEM_H__ */
...@@ -24,6 +24,7 @@ ...@@ -24,6 +24,7 @@
#include "xfs_bit.h" #include "xfs_bit.h"
#include "xfs_sb.h" #include "xfs_sb.h"
#include "xfs_mount.h" #include "xfs_mount.h"
#include "xfs_defer.h"
#include "xfs_da_format.h" #include "xfs_da_format.h"
#include "xfs_da_btree.h" #include "xfs_da_btree.h"
#include "xfs_inode.h" #include "xfs_inode.h"
...@@ -4716,7 +4717,8 @@ STATIC int ...@@ -4716,7 +4717,8 @@ STATIC int
xlog_recover_process_cui( xlog_recover_process_cui(
struct xfs_mount *mp, struct xfs_mount *mp,
struct xfs_ail *ailp, struct xfs_ail *ailp,
struct xfs_log_item *lip) struct xfs_log_item *lip,
struct xfs_defer_ops *dfops)
{ {
struct xfs_cui_log_item *cuip; struct xfs_cui_log_item *cuip;
int error; int error;
...@@ -4729,7 +4731,7 @@ xlog_recover_process_cui( ...@@ -4729,7 +4731,7 @@ xlog_recover_process_cui(
return 0; return 0;
spin_unlock(&ailp->xa_lock); spin_unlock(&ailp->xa_lock);
error = xfs_cui_recover(mp, cuip); error = xfs_cui_recover(mp, cuip, dfops);
spin_lock(&ailp->xa_lock); spin_lock(&ailp->xa_lock);
return error; return error;
...@@ -4756,7 +4758,8 @@ STATIC int ...@@ -4756,7 +4758,8 @@ STATIC int
xlog_recover_process_bui( xlog_recover_process_bui(
struct xfs_mount *mp, struct xfs_mount *mp,
struct xfs_ail *ailp, struct xfs_ail *ailp,
struct xfs_log_item *lip) struct xfs_log_item *lip,
struct xfs_defer_ops *dfops)
{ {
struct xfs_bui_log_item *buip; struct xfs_bui_log_item *buip;
int error; int error;
...@@ -4769,7 +4772,7 @@ xlog_recover_process_bui( ...@@ -4769,7 +4772,7 @@ xlog_recover_process_bui(
return 0; return 0;
spin_unlock(&ailp->xa_lock); spin_unlock(&ailp->xa_lock);
error = xfs_bui_recover(mp, buip); error = xfs_bui_recover(mp, buip, dfops);
spin_lock(&ailp->xa_lock); spin_lock(&ailp->xa_lock);
return error; return error;
...@@ -4805,6 +4808,46 @@ static inline bool xlog_item_is_intent(struct xfs_log_item *lip) ...@@ -4805,6 +4808,46 @@ static inline bool xlog_item_is_intent(struct xfs_log_item *lip)
} }
} }
/* Take all the collected deferred ops and finish them in order. */
static int
xlog_finish_defer_ops(
struct xfs_mount *mp,
struct xfs_defer_ops *dfops)
{
struct xfs_trans *tp;
int64_t freeblks;
uint resblks;
int error;
/*
* We're finishing the defer_ops that accumulated as a result of
* recovering unfinished intent items during log recovery. We
* reserve an itruncate transaction because it is the largest
* permanent transaction type. Since we're the only user of the fs
* right now, take 93% (15/16) of the available free blocks. Use
* weird math to avoid a 64-bit division.
*/
freeblks = percpu_counter_sum(&mp->m_fdblocks);
if (freeblks <= 0)
return -ENOSPC;
resblks = min_t(int64_t, UINT_MAX, freeblks);
resblks = (resblks * 15) >> 4;
error = xfs_trans_alloc(mp, &M_RES(mp)->tr_itruncate, resblks,
0, XFS_TRANS_RESERVE, &tp);
if (error)
return error;
error = xfs_defer_finish(&tp, dfops);
if (error)
goto out_cancel;
return xfs_trans_commit(tp);
out_cancel:
xfs_trans_cancel(tp);
return error;
}
/* /*
* When this is called, all of the log intent items which did not have * When this is called, all of the log intent items which did not have
* corresponding log done items should be in the AIL. What we do now * corresponding log done items should be in the AIL. What we do now
...@@ -4825,10 +4868,12 @@ STATIC int ...@@ -4825,10 +4868,12 @@ STATIC int
xlog_recover_process_intents( xlog_recover_process_intents(
struct xlog *log) struct xlog *log)
{ {
struct xfs_log_item *lip; struct xfs_defer_ops dfops;
int error = 0;
struct xfs_ail_cursor cur; struct xfs_ail_cursor cur;
struct xfs_log_item *lip;
struct xfs_ail *ailp; struct xfs_ail *ailp;
xfs_fsblock_t firstfsb;
int error = 0;
#if defined(DEBUG) || defined(XFS_WARN) #if defined(DEBUG) || defined(XFS_WARN)
xfs_lsn_t last_lsn; xfs_lsn_t last_lsn;
#endif #endif
...@@ -4839,6 +4884,7 @@ xlog_recover_process_intents( ...@@ -4839,6 +4884,7 @@ xlog_recover_process_intents(
#if defined(DEBUG) || defined(XFS_WARN) #if defined(DEBUG) || defined(XFS_WARN)
last_lsn = xlog_assign_lsn(log->l_curr_cycle, log->l_curr_block); last_lsn = xlog_assign_lsn(log->l_curr_cycle, log->l_curr_block);
#endif #endif
xfs_defer_init(&dfops, &firstfsb);
while (lip != NULL) { while (lip != NULL) {
/* /*
* We're done when we see something other than an intent. * We're done when we see something other than an intent.
...@@ -4859,6 +4905,12 @@ xlog_recover_process_intents( ...@@ -4859,6 +4905,12 @@ xlog_recover_process_intents(
*/ */
ASSERT(XFS_LSN_CMP(last_lsn, lip->li_lsn) >= 0); ASSERT(XFS_LSN_CMP(last_lsn, lip->li_lsn) >= 0);
/*
* NOTE: If your intent processing routine can create more
* deferred ops, you /must/ attach them to the dfops in this
* routine or else those subsequent intents will get
* replayed in the wrong order!
*/
switch (lip->li_type) { switch (lip->li_type) {
case XFS_LI_EFI: case XFS_LI_EFI:
error = xlog_recover_process_efi(log->l_mp, ailp, lip); error = xlog_recover_process_efi(log->l_mp, ailp, lip);
...@@ -4867,10 +4919,12 @@ xlog_recover_process_intents( ...@@ -4867,10 +4919,12 @@ xlog_recover_process_intents(
error = xlog_recover_process_rui(log->l_mp, ailp, lip); error = xlog_recover_process_rui(log->l_mp, ailp, lip);
break; break;
case XFS_LI_CUI: case XFS_LI_CUI:
error = xlog_recover_process_cui(log->l_mp, ailp, lip); error = xlog_recover_process_cui(log->l_mp, ailp, lip,
&dfops);
break; break;
case XFS_LI_BUI: case XFS_LI_BUI:
error = xlog_recover_process_bui(log->l_mp, ailp, lip); error = xlog_recover_process_bui(log->l_mp, ailp, lip,
&dfops);
break; break;
} }
if (error) if (error)
...@@ -4880,6 +4934,11 @@ xlog_recover_process_intents( ...@@ -4880,6 +4934,11 @@ xlog_recover_process_intents(
out: out:
xfs_trans_ail_cursor_done(&cur); xfs_trans_ail_cursor_done(&cur);
spin_unlock(&ailp->xa_lock); spin_unlock(&ailp->xa_lock);
if (error)
xfs_defer_cancel(&dfops);
else
error = xlog_finish_defer_ops(log->l_mp, &dfops);
return error; return error;
} }
......
...@@ -393,7 +393,8 @@ xfs_cud_init( ...@@ -393,7 +393,8 @@ xfs_cud_init(
int int
xfs_cui_recover( xfs_cui_recover(
struct xfs_mount *mp, struct xfs_mount *mp,
struct xfs_cui_log_item *cuip) struct xfs_cui_log_item *cuip,
struct xfs_defer_ops *dfops)
{ {
int i; int i;
int error = 0; int error = 0;
...@@ -405,11 +406,9 @@ xfs_cui_recover( ...@@ -405,11 +406,9 @@ xfs_cui_recover(
struct xfs_trans *tp; struct xfs_trans *tp;
struct xfs_btree_cur *rcur = NULL; struct xfs_btree_cur *rcur = NULL;
enum xfs_refcount_intent_type type; enum xfs_refcount_intent_type type;
xfs_fsblock_t firstfsb;
xfs_fsblock_t new_fsb; xfs_fsblock_t new_fsb;
xfs_extlen_t new_len; xfs_extlen_t new_len;
struct xfs_bmbt_irec irec; struct xfs_bmbt_irec irec;
struct xfs_defer_ops dfops;
bool requeue_only = false; bool requeue_only = false;
ASSERT(!test_bit(XFS_CUI_RECOVERED, &cuip->cui_flags)); ASSERT(!test_bit(XFS_CUI_RECOVERED, &cuip->cui_flags));
...@@ -465,7 +464,6 @@ xfs_cui_recover( ...@@ -465,7 +464,6 @@ xfs_cui_recover(
return error; return error;
cudp = xfs_trans_get_cud(tp, cuip); cudp = xfs_trans_get_cud(tp, cuip);
xfs_defer_init(&dfops, &firstfsb);
for (i = 0; i < cuip->cui_format.cui_nextents; i++) { for (i = 0; i < cuip->cui_format.cui_nextents; i++) {
refc = &cuip->cui_format.cui_extents[i]; refc = &cuip->cui_format.cui_extents[i];
refc_type = refc->pe_flags & XFS_REFCOUNT_EXTENT_TYPE_MASK; refc_type = refc->pe_flags & XFS_REFCOUNT_EXTENT_TYPE_MASK;
...@@ -485,7 +483,7 @@ xfs_cui_recover( ...@@ -485,7 +483,7 @@ xfs_cui_recover(
new_len = refc->pe_len; new_len = refc->pe_len;
} else } else
error = xfs_trans_log_finish_refcount_update(tp, cudp, error = xfs_trans_log_finish_refcount_update(tp, cudp,
&dfops, type, refc->pe_startblock, refc->pe_len, dfops, type, refc->pe_startblock, refc->pe_len,
&new_fsb, &new_len, &rcur); &new_fsb, &new_len, &rcur);
if (error) if (error)
goto abort_error; goto abort_error;
...@@ -497,21 +495,21 @@ xfs_cui_recover( ...@@ -497,21 +495,21 @@ xfs_cui_recover(
switch (type) { switch (type) {
case XFS_REFCOUNT_INCREASE: case XFS_REFCOUNT_INCREASE:
error = xfs_refcount_increase_extent( error = xfs_refcount_increase_extent(
tp->t_mountp, &dfops, &irec); tp->t_mountp, dfops, &irec);
break; break;
case XFS_REFCOUNT_DECREASE: case XFS_REFCOUNT_DECREASE:
error = xfs_refcount_decrease_extent( error = xfs_refcount_decrease_extent(
tp->t_mountp, &dfops, &irec); tp->t_mountp, dfops, &irec);
break; break;
case XFS_REFCOUNT_ALLOC_COW: case XFS_REFCOUNT_ALLOC_COW:
error = xfs_refcount_alloc_cow_extent( error = xfs_refcount_alloc_cow_extent(
tp->t_mountp, &dfops, tp->t_mountp, dfops,
irec.br_startblock, irec.br_startblock,
irec.br_blockcount); irec.br_blockcount);
break; break;
case XFS_REFCOUNT_FREE_COW: case XFS_REFCOUNT_FREE_COW:
error = xfs_refcount_free_cow_extent( error = xfs_refcount_free_cow_extent(
tp->t_mountp, &dfops, tp->t_mountp, dfops,
irec.br_startblock, irec.br_startblock,
irec.br_blockcount); irec.br_blockcount);
break; break;
...@@ -525,17 +523,12 @@ xfs_cui_recover( ...@@ -525,17 +523,12 @@ xfs_cui_recover(
} }
xfs_refcount_finish_one_cleanup(tp, rcur, error); xfs_refcount_finish_one_cleanup(tp, rcur, error);
error = xfs_defer_finish(&tp, &dfops);
if (error)
goto abort_defer;
set_bit(XFS_CUI_RECOVERED, &cuip->cui_flags); set_bit(XFS_CUI_RECOVERED, &cuip->cui_flags);
error = xfs_trans_commit(tp); error = xfs_trans_commit(tp);
return error; return error;
abort_error: abort_error:
xfs_refcount_finish_one_cleanup(tp, rcur, error); xfs_refcount_finish_one_cleanup(tp, rcur, error);
abort_defer:
xfs_defer_cancel(&dfops);
xfs_trans_cancel(tp); xfs_trans_cancel(tp);
return error; return error;
} }
...@@ -96,6 +96,7 @@ struct xfs_cud_log_item *xfs_cud_init(struct xfs_mount *, ...@@ -96,6 +96,7 @@ struct xfs_cud_log_item *xfs_cud_init(struct xfs_mount *,
struct xfs_cui_log_item *); struct xfs_cui_log_item *);
void xfs_cui_item_free(struct xfs_cui_log_item *); void xfs_cui_item_free(struct xfs_cui_log_item *);
void xfs_cui_release(struct xfs_cui_log_item *); void xfs_cui_release(struct xfs_cui_log_item *);
int xfs_cui_recover(struct xfs_mount *mp, struct xfs_cui_log_item *cuip); int xfs_cui_recover(struct xfs_mount *mp, struct xfs_cui_log_item *cuip,
struct xfs_defer_ops *dfops);
#endif /* __XFS_REFCOUNT_ITEM_H__ */ #endif /* __XFS_REFCOUNT_ITEM_H__ */
Markdown is supported
0%
or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment