OpenCloudOS-Kernel/fs/xfs/xfs_trans.c

1434 lines
39 KiB
C
Raw Normal View History

// SPDX-License-Identifier: GPL-2.0
/*
* Copyright (c) 2000-2003,2005 Silicon Graphics, Inc.
* Copyright (C) 2010 Red Hat, Inc.
* All Rights Reserved.
*/
#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_mount.h"
#include "xfs_extent_busy.h"
#include "xfs_quota.h"
#include "xfs_trans.h"
#include "xfs_trans_priv.h"
#include "xfs_log.h"
xfs: AIL needs asynchronous CIL forcing The AIL pushing is stalling on log forces when it comes across pinned items. This is happening on removal workloads where the AIL is dominated by stale items that are removed from AIL when the checkpoint that marks the items stale is committed to the journal. This results is relatively few items in the AIL, but those that are are often pinned as directories items are being removed from are still being logged. As a result, many push cycles through the CIL will first issue a blocking log force to unpin the items. This can take some time to complete, with tracing regularly showing push delays of half a second and sometimes up into the range of several seconds. Sequences like this aren't uncommon: .... 399.829437: xfsaild: last lsn 0x11002dd000 count 101 stuck 101 flushing 0 tout 20 <wanted 20ms, got 270ms delay> 400.099622: xfsaild: target 0x11002f3600, prev 0x11002f3600, last lsn 0x0 400.099623: xfsaild: first lsn 0x11002f3600 400.099679: xfsaild: last lsn 0x1100305000 count 16 stuck 11 flushing 0 tout 50 <wanted 50ms, got 500ms delay> 400.589348: xfsaild: target 0x110032e600, prev 0x11002f3600, last lsn 0x0 400.589349: xfsaild: first lsn 0x1100305000 400.589595: xfsaild: last lsn 0x110032e600 count 156 stuck 101 flushing 30 tout 50 <wanted 50ms, got 460ms delay> 400.950341: xfsaild: target 0x1100353000, prev 0x110032e600, last lsn 0x0 400.950343: xfsaild: first lsn 0x1100317c00 400.950436: xfsaild: last lsn 0x110033d200 count 105 stuck 101 flushing 0 tout 20 <wanted 20ms, got 200ms delay> 401.142333: xfsaild: target 0x1100361600, prev 0x1100353000, last lsn 0x0 401.142334: xfsaild: first lsn 0x110032e600 401.142535: xfsaild: last lsn 0x1100353000 count 122 stuck 101 flushing 8 tout 10 <wanted 10ms, got 10ms delay> 401.154323: xfsaild: target 0x1100361600, prev 0x1100361600, last lsn 0x1100353000 401.154328: xfsaild: first lsn 0x1100353000 401.154389: xfsaild: last lsn 0x1100353000 count 101 stuck 101 flushing 0 tout 20 <wanted 20ms, got 300ms delay> 401.451525: xfsaild: target 0x1100361600, prev 0x1100361600, last lsn 0x0 401.451526: xfsaild: first lsn 0x1100353000 401.451804: xfsaild: last lsn 0x1100377200 count 170 stuck 22 flushing 122 tout 50 <wanted 50ms, got 500ms delay> 401.933581: xfsaild: target 0x1100361600, prev 0x1100361600, last lsn 0x0 .... In each of these cases, every AIL pass saw 101 log items stuck on the AIL (pinned) with very few other items being found. Each pass, a log force was issued, and delay between last/first is the sleep time + the sync log force time. Some of these 101 items pinned the tail of the log. The tail of the log does slowly creep forward (first lsn), but the problem is that the log is actually out of reservation space because it's been running so many transactions that stale items that never reach the AIL but consume log space. Hence we have a largely empty AIL, with long term pins on items that pin the tail of the log that don't get pushed frequently enough to keep log space available. The problem is the hundreds of milliseconds that we block in the log force pushing the CIL out to disk. The AIL should not be stalled like this - it needs to run and flush items that are at the tail of the log with minimal latency. What we really need to do is trigger a log flush, but then not wait for it at all - we've already done our waiting for stuff to complete when we backed off prior to the log force being issued. Even if we remove the XFS_LOG_SYNC from the xfs_log_force() call, we still do a blocking flush of the CIL and that is what is causing the issue. Hence we need a new interface for the CIL to trigger an immediate background push of the CIL to get it moving faster but not to wait on that to occur. While the CIL is pushing, the AIL can also be pushing. We already have an internal interface to do this - xlog_cil_push_now() - but we need a wrapper for it to be used externally. xlog_cil_force_seq() can easily be extended to do what we need as it already implements the synchronous CIL push via xlog_cil_push_now(). Add the necessary flags and "push current sequence" semantics to xlog_cil_force_seq() and convert the AIL pushing to use it. One of the complexities here is that the CIL push does not guarantee that the commit record for the CIL checkpoint is written to disk. The current log force ensures this by submitting the current ACTIVE iclog that the commit record was written to. We need the CIL to actually write this commit record to disk for an async push to ensure that the checkpoint actually makes it to disk and unpins the pinned items in the checkpoint on completion. Hence we need to pass down to the CIL push that we are doing an async flush so that it can switch out the commit_iclog if necessary to get written to disk when the commit iclog is finally released. Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Darrick J. Wong <djwong@kernel.org> Reviewed-by: Allison Henderson <allison.henderson@oracle.com> Signed-off-by: Darrick J. Wong <djwong@kernel.org>
2021-08-11 09:00:44 +08:00
#include "xfs_log_priv.h"
xfs: Improve scalability of busy extent tracking When we free a metadata extent, we record it in the per-AG busy extent array so that it is not re-used before the freeing transaction hits the disk. This array is fixed size, so when it overflows we make further allocation transactions synchronous because we cannot track more freed extents until those transactions hit the disk and are completed. Under heavy mixed allocation and freeing workloads with large log buffers, we can overflow this array quite easily. Further, the array is sparsely populated, which means that inserts need to search for a free slot, and array searches often have to search many more slots that are actually used to check all the busy extents. Quite inefficient, really. To enable this aspect of extent freeing to scale better, we need a structure that can grow dynamically. While in other areas of XFS we have used radix trees, the extents being freed are at random locations on disk so are better suited to being indexed by an rbtree. So, use a per-AG rbtree indexed by block number to track busy extents. This incures a memory allocation when marking an extent busy, but should not occur too often in low memory situations. This should scale to an arbitrary number of extents so should not be a limitation for features such as in-memory aggregation of transactions. However, there are still situations where we can't avoid allocating busy extents (such as allocation from the AGFL). To minimise the overhead of such occurences, we need to avoid doing a synchronous log force while holding the AGF locked to ensure that the previous transactions are safely on disk before we use the extent. We can do this by marking the transaction doing the allocation as synchronous rather issuing a log force. Because of the locking involved and the ordering of transactions, the synchronous transaction provides the same guarantees as a synchronous log force because it ensures that all the prior transactions are already on disk when the synchronous transaction hits the disk. i.e. it preserves the free->allocate order of the extent correctly in recovery. By doing this, we avoid holding the AGF locked while log writes are in progress, hence reducing the length of time the lock is held and therefore we increase the rate at which we can allocate and free from the allocation group, thereby increasing overall throughput. The only problem with this approach is that when a metadata buffer is marked stale (e.g. a directory block is removed), then buffer remains pinned and locked until the log goes to disk. The issue here is that if that stale buffer is reallocated in a subsequent transaction, the attempt to lock that buffer in the transaction will hang waiting the log to go to disk to unlock and unpin the buffer. Hence if someone tries to lock a pinned, stale, locked buffer we need to push on the log to get it unlocked ASAP. Effectively we are trading off a guaranteed log force for a much less common trigger for log force to occur. Ideally we should not reallocate busy extents. That is a much more complex fix to the problem as it involves direct intervention in the allocation btree searches in many places. This is left to a future set of modifications. Finally, now that we track busy extents in allocated memory, we don't need the descriptors in the transaction structure to point to them. We can replace the complex busy chunk infrastructure with a simple linked list of busy extents. This allows us to remove a large chunk of code, making the overall change a net reduction in code size. Signed-off-by: Dave Chinner <david@fromorbit.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Alex Elder <aelder@sgi.com>
2010-05-21 10:07:08 +08:00
#include "xfs_trace.h"
#include "xfs_error.h"
xfs: defer agfl block frees when dfops is available The AGFL fixup code executes before every block allocation/free and rectifies the AGFL based on the current, dynamic allocation requirements of the fs. The AGFL must hold a minimum number of blocks to satisfy a worst case split of the free space btrees caused by the impending allocation operation. The AGFL is also updated to maintain the implicit requirement for a minimum number of free slots to satisfy a worst case join of the free space btrees. Since the AGFL caches individual blocks, AGFL reduction typically involves multiple, single block frees. We've had reports of transaction overrun problems during certain workloads that boil down to AGFL reduction freeing multiple blocks and consuming more space in the log than was reserved for the transaction. Since the objective of freeing AGFL blocks is to ensure free AGFL free slots are available for the upcoming allocation, one way to address this problem is to release surplus blocks from the AGFL immediately but defer the free of those blocks (similar to how file-mapped blocks are unmapped from the file in one transaction and freed via a deferred operation) until the transaction is rolled. This turns AGFL reduction into an operation with predictable log reservation consumption. Add the capability to defer AGFL block frees when a deferred ops list is available to the AGFL fixup code. Add a dfops pointer to the transaction to carry dfops through various contexts to the allocator context. Deferring AGFL frees is conditional behavior based on whether the transaction pointer is populated. The long term objective is to reuse the transaction pointer to clean up all unrelated callchains that pass dfops on the stack along with a transaction and in doing so, consistently defer AGFL blocks from the allocator. A bit of customization is required to handle deferred completion processing because AGFL blocks are accounted against a per-ag reservation pool and AGFL blocks are not inserted into the extent busy list when freed (they are inserted when used and released back to the AGFL). Reuse the majority of the existing deferred extent free infrastructure and customize it appropriately to handle AGFL blocks. Note that this patch only adds infrastructure. It does not change behavior because no callers have been updated to pass ->t_agfl_dfops into the allocation code. Signed-off-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>
2018-05-08 08:38:47 +08:00
#include "xfs_defer.h"
#include "xfs_inode.h"
#include "xfs_dquot_item.h"
#include "xfs_dquot.h"
#include "xfs_icache.h"
struct kmem_cache *xfs_trans_cache;
#if defined(CONFIG_TRACEPOINTS)
static void
xfs_trans_trace_reservations(
struct xfs_mount *mp)
{
struct xfs_trans_res *res;
struct xfs_trans_res *end_res;
int i;
res = (struct xfs_trans_res *)M_RES(mp);
end_res = (struct xfs_trans_res *)(M_RES(mp) + 1);
for (i = 0; res < end_res; i++, res++)
trace_xfs_trans_resv_calc(mp, i, res);
}
#else
# define xfs_trans_trace_reservations(mp)
#endif
/*
* Initialize the precomputed transaction reservation values
* in the mount structure.
*/
void
xfs_trans_init(
struct xfs_mount *mp)
{
xfs_trans_resv_calc(mp, M_RES(mp));
xfs_trans_trace_reservations(mp);
}
/*
* Free the transaction structure. If there is more clean up
* to do when the structure is freed, add it here.
*/
STATIC void
xfs_trans_free(
xfs: Improve scalability of busy extent tracking When we free a metadata extent, we record it in the per-AG busy extent array so that it is not re-used before the freeing transaction hits the disk. This array is fixed size, so when it overflows we make further allocation transactions synchronous because we cannot track more freed extents until those transactions hit the disk and are completed. Under heavy mixed allocation and freeing workloads with large log buffers, we can overflow this array quite easily. Further, the array is sparsely populated, which means that inserts need to search for a free slot, and array searches often have to search many more slots that are actually used to check all the busy extents. Quite inefficient, really. To enable this aspect of extent freeing to scale better, we need a structure that can grow dynamically. While in other areas of XFS we have used radix trees, the extents being freed are at random locations on disk so are better suited to being indexed by an rbtree. So, use a per-AG rbtree indexed by block number to track busy extents. This incures a memory allocation when marking an extent busy, but should not occur too often in low memory situations. This should scale to an arbitrary number of extents so should not be a limitation for features such as in-memory aggregation of transactions. However, there are still situations where we can't avoid allocating busy extents (such as allocation from the AGFL). To minimise the overhead of such occurences, we need to avoid doing a synchronous log force while holding the AGF locked to ensure that the previous transactions are safely on disk before we use the extent. We can do this by marking the transaction doing the allocation as synchronous rather issuing a log force. Because of the locking involved and the ordering of transactions, the synchronous transaction provides the same guarantees as a synchronous log force because it ensures that all the prior transactions are already on disk when the synchronous transaction hits the disk. i.e. it preserves the free->allocate order of the extent correctly in recovery. By doing this, we avoid holding the AGF locked while log writes are in progress, hence reducing the length of time the lock is held and therefore we increase the rate at which we can allocate and free from the allocation group, thereby increasing overall throughput. The only problem with this approach is that when a metadata buffer is marked stale (e.g. a directory block is removed), then buffer remains pinned and locked until the log goes to disk. The issue here is that if that stale buffer is reallocated in a subsequent transaction, the attempt to lock that buffer in the transaction will hang waiting the log to go to disk to unlock and unpin the buffer. Hence if someone tries to lock a pinned, stale, locked buffer we need to push on the log to get it unlocked ASAP. Effectively we are trading off a guaranteed log force for a much less common trigger for log force to occur. Ideally we should not reallocate busy extents. That is a much more complex fix to the problem as it involves direct intervention in the allocation btree searches in many places. This is left to a future set of modifications. Finally, now that we track busy extents in allocated memory, we don't need the descriptors in the transaction structure to point to them. We can replace the complex busy chunk infrastructure with a simple linked list of busy extents. This allows us to remove a large chunk of code, making the overall change a net reduction in code size. Signed-off-by: Dave Chinner <david@fromorbit.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Alex Elder <aelder@sgi.com>
2010-05-21 10:07:08 +08:00
struct xfs_trans *tp)
{
xfs_extent_busy_sort(&tp->t_busy);
xfs_extent_busy_clear(tp->t_mountp, &tp->t_busy, false);
xfs: Improve scalability of busy extent tracking When we free a metadata extent, we record it in the per-AG busy extent array so that it is not re-used before the freeing transaction hits the disk. This array is fixed size, so when it overflows we make further allocation transactions synchronous because we cannot track more freed extents until those transactions hit the disk and are completed. Under heavy mixed allocation and freeing workloads with large log buffers, we can overflow this array quite easily. Further, the array is sparsely populated, which means that inserts need to search for a free slot, and array searches often have to search many more slots that are actually used to check all the busy extents. Quite inefficient, really. To enable this aspect of extent freeing to scale better, we need a structure that can grow dynamically. While in other areas of XFS we have used radix trees, the extents being freed are at random locations on disk so are better suited to being indexed by an rbtree. So, use a per-AG rbtree indexed by block number to track busy extents. This incures a memory allocation when marking an extent busy, but should not occur too often in low memory situations. This should scale to an arbitrary number of extents so should not be a limitation for features such as in-memory aggregation of transactions. However, there are still situations where we can't avoid allocating busy extents (such as allocation from the AGFL). To minimise the overhead of such occurences, we need to avoid doing a synchronous log force while holding the AGF locked to ensure that the previous transactions are safely on disk before we use the extent. We can do this by marking the transaction doing the allocation as synchronous rather issuing a log force. Because of the locking involved and the ordering of transactions, the synchronous transaction provides the same guarantees as a synchronous log force because it ensures that all the prior transactions are already on disk when the synchronous transaction hits the disk. i.e. it preserves the free->allocate order of the extent correctly in recovery. By doing this, we avoid holding the AGF locked while log writes are in progress, hence reducing the length of time the lock is held and therefore we increase the rate at which we can allocate and free from the allocation group, thereby increasing overall throughput. The only problem with this approach is that when a metadata buffer is marked stale (e.g. a directory block is removed), then buffer remains pinned and locked until the log goes to disk. The issue here is that if that stale buffer is reallocated in a subsequent transaction, the attempt to lock that buffer in the transaction will hang waiting the log to go to disk to unlock and unpin the buffer. Hence if someone tries to lock a pinned, stale, locked buffer we need to push on the log to get it unlocked ASAP. Effectively we are trading off a guaranteed log force for a much less common trigger for log force to occur. Ideally we should not reallocate busy extents. That is a much more complex fix to the problem as it involves direct intervention in the allocation btree searches in many places. This is left to a future set of modifications. Finally, now that we track busy extents in allocated memory, we don't need the descriptors in the transaction structure to point to them. We can replace the complex busy chunk infrastructure with a simple linked list of busy extents. This allows us to remove a large chunk of code, making the overall change a net reduction in code size. Signed-off-by: Dave Chinner <david@fromorbit.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Alex Elder <aelder@sgi.com>
2010-05-21 10:07:08 +08:00
trace_xfs_trans_free(tp, _RET_IP_);
xfs_trans_clear_context(tp);
if (!(tp->t_flags & XFS_TRANS_NO_WRITECOUNT))
sb_end_intwrite(tp->t_mountp->m_super);
xfs_trans_free_dqinfo(tp);
kmem_cache_free(xfs_trans_cache, tp);
}
/*
* This is called to create a new transaction which will share the
* permanent log reservation of the given transaction. The remaining
* unused block and rt extent reservations are also inherited. This
* implies that the original transaction is no longer allowed to allocate
* blocks. Locks and log items, however, are no inherited. They must
* be added to the new transaction explicitly.
*/
xfs: defer agfl block frees when dfops is available The AGFL fixup code executes before every block allocation/free and rectifies the AGFL based on the current, dynamic allocation requirements of the fs. The AGFL must hold a minimum number of blocks to satisfy a worst case split of the free space btrees caused by the impending allocation operation. The AGFL is also updated to maintain the implicit requirement for a minimum number of free slots to satisfy a worst case join of the free space btrees. Since the AGFL caches individual blocks, AGFL reduction typically involves multiple, single block frees. We've had reports of transaction overrun problems during certain workloads that boil down to AGFL reduction freeing multiple blocks and consuming more space in the log than was reserved for the transaction. Since the objective of freeing AGFL blocks is to ensure free AGFL free slots are available for the upcoming allocation, one way to address this problem is to release surplus blocks from the AGFL immediately but defer the free of those blocks (similar to how file-mapped blocks are unmapped from the file in one transaction and freed via a deferred operation) until the transaction is rolled. This turns AGFL reduction into an operation with predictable log reservation consumption. Add the capability to defer AGFL block frees when a deferred ops list is available to the AGFL fixup code. Add a dfops pointer to the transaction to carry dfops through various contexts to the allocator context. Deferring AGFL frees is conditional behavior based on whether the transaction pointer is populated. The long term objective is to reuse the transaction pointer to clean up all unrelated callchains that pass dfops on the stack along with a transaction and in doing so, consistently defer AGFL blocks from the allocator. A bit of customization is required to handle deferred completion processing because AGFL blocks are accounted against a per-ag reservation pool and AGFL blocks are not inserted into the extent busy list when freed (they are inserted when used and released back to the AGFL). Reuse the majority of the existing deferred extent free infrastructure and customize it appropriately to handle AGFL blocks. Note that this patch only adds infrastructure. It does not change behavior because no callers have been updated to pass ->t_agfl_dfops into the allocation code. Signed-off-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>
2018-05-08 08:38:47 +08:00
STATIC struct xfs_trans *
xfs_trans_dup(
xfs: defer agfl block frees when dfops is available The AGFL fixup code executes before every block allocation/free and rectifies the AGFL based on the current, dynamic allocation requirements of the fs. The AGFL must hold a minimum number of blocks to satisfy a worst case split of the free space btrees caused by the impending allocation operation. The AGFL is also updated to maintain the implicit requirement for a minimum number of free slots to satisfy a worst case join of the free space btrees. Since the AGFL caches individual blocks, AGFL reduction typically involves multiple, single block frees. We've had reports of transaction overrun problems during certain workloads that boil down to AGFL reduction freeing multiple blocks and consuming more space in the log than was reserved for the transaction. Since the objective of freeing AGFL blocks is to ensure free AGFL free slots are available for the upcoming allocation, one way to address this problem is to release surplus blocks from the AGFL immediately but defer the free of those blocks (similar to how file-mapped blocks are unmapped from the file in one transaction and freed via a deferred operation) until the transaction is rolled. This turns AGFL reduction into an operation with predictable log reservation consumption. Add the capability to defer AGFL block frees when a deferred ops list is available to the AGFL fixup code. Add a dfops pointer to the transaction to carry dfops through various contexts to the allocator context. Deferring AGFL frees is conditional behavior based on whether the transaction pointer is populated. The long term objective is to reuse the transaction pointer to clean up all unrelated callchains that pass dfops on the stack along with a transaction and in doing so, consistently defer AGFL blocks from the allocator. A bit of customization is required to handle deferred completion processing because AGFL blocks are accounted against a per-ag reservation pool and AGFL blocks are not inserted into the extent busy list when freed (they are inserted when used and released back to the AGFL). Reuse the majority of the existing deferred extent free infrastructure and customize it appropriately to handle AGFL blocks. Note that this patch only adds infrastructure. It does not change behavior because no callers have been updated to pass ->t_agfl_dfops into the allocation code. Signed-off-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>
2018-05-08 08:38:47 +08:00
struct xfs_trans *tp)
{
xfs: defer agfl block frees when dfops is available The AGFL fixup code executes before every block allocation/free and rectifies the AGFL based on the current, dynamic allocation requirements of the fs. The AGFL must hold a minimum number of blocks to satisfy a worst case split of the free space btrees caused by the impending allocation operation. The AGFL is also updated to maintain the implicit requirement for a minimum number of free slots to satisfy a worst case join of the free space btrees. Since the AGFL caches individual blocks, AGFL reduction typically involves multiple, single block frees. We've had reports of transaction overrun problems during certain workloads that boil down to AGFL reduction freeing multiple blocks and consuming more space in the log than was reserved for the transaction. Since the objective of freeing AGFL blocks is to ensure free AGFL free slots are available for the upcoming allocation, one way to address this problem is to release surplus blocks from the AGFL immediately but defer the free of those blocks (similar to how file-mapped blocks are unmapped from the file in one transaction and freed via a deferred operation) until the transaction is rolled. This turns AGFL reduction into an operation with predictable log reservation consumption. Add the capability to defer AGFL block frees when a deferred ops list is available to the AGFL fixup code. Add a dfops pointer to the transaction to carry dfops through various contexts to the allocator context. Deferring AGFL frees is conditional behavior based on whether the transaction pointer is populated. The long term objective is to reuse the transaction pointer to clean up all unrelated callchains that pass dfops on the stack along with a transaction and in doing so, consistently defer AGFL blocks from the allocator. A bit of customization is required to handle deferred completion processing because AGFL blocks are accounted against a per-ag reservation pool and AGFL blocks are not inserted into the extent busy list when freed (they are inserted when used and released back to the AGFL). Reuse the majority of the existing deferred extent free infrastructure and customize it appropriately to handle AGFL blocks. Note that this patch only adds infrastructure. It does not change behavior because no callers have been updated to pass ->t_agfl_dfops into the allocation code. Signed-off-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>
2018-05-08 08:38:47 +08:00
struct xfs_trans *ntp;
trace_xfs_trans_dup(tp, _RET_IP_);
ntp = kmem_cache_zalloc(xfs_trans_cache, GFP_KERNEL | __GFP_NOFAIL);
/*
* Initialize the new transaction structure.
*/
ntp->t_magic = XFS_TRANS_HEADER_MAGIC;
ntp->t_mountp = tp->t_mountp;
INIT_LIST_HEAD(&ntp->t_items);
xfs: Improve scalability of busy extent tracking When we free a metadata extent, we record it in the per-AG busy extent array so that it is not re-used before the freeing transaction hits the disk. This array is fixed size, so when it overflows we make further allocation transactions synchronous because we cannot track more freed extents until those transactions hit the disk and are completed. Under heavy mixed allocation and freeing workloads with large log buffers, we can overflow this array quite easily. Further, the array is sparsely populated, which means that inserts need to search for a free slot, and array searches often have to search many more slots that are actually used to check all the busy extents. Quite inefficient, really. To enable this aspect of extent freeing to scale better, we need a structure that can grow dynamically. While in other areas of XFS we have used radix trees, the extents being freed are at random locations on disk so are better suited to being indexed by an rbtree. So, use a per-AG rbtree indexed by block number to track busy extents. This incures a memory allocation when marking an extent busy, but should not occur too often in low memory situations. This should scale to an arbitrary number of extents so should not be a limitation for features such as in-memory aggregation of transactions. However, there are still situations where we can't avoid allocating busy extents (such as allocation from the AGFL). To minimise the overhead of such occurences, we need to avoid doing a synchronous log force while holding the AGF locked to ensure that the previous transactions are safely on disk before we use the extent. We can do this by marking the transaction doing the allocation as synchronous rather issuing a log force. Because of the locking involved and the ordering of transactions, the synchronous transaction provides the same guarantees as a synchronous log force because it ensures that all the prior transactions are already on disk when the synchronous transaction hits the disk. i.e. it preserves the free->allocate order of the extent correctly in recovery. By doing this, we avoid holding the AGF locked while log writes are in progress, hence reducing the length of time the lock is held and therefore we increase the rate at which we can allocate and free from the allocation group, thereby increasing overall throughput. The only problem with this approach is that when a metadata buffer is marked stale (e.g. a directory block is removed), then buffer remains pinned and locked until the log goes to disk. The issue here is that if that stale buffer is reallocated in a subsequent transaction, the attempt to lock that buffer in the transaction will hang waiting the log to go to disk to unlock and unpin the buffer. Hence if someone tries to lock a pinned, stale, locked buffer we need to push on the log to get it unlocked ASAP. Effectively we are trading off a guaranteed log force for a much less common trigger for log force to occur. Ideally we should not reallocate busy extents. That is a much more complex fix to the problem as it involves direct intervention in the allocation btree searches in many places. This is left to a future set of modifications. Finally, now that we track busy extents in allocated memory, we don't need the descriptors in the transaction structure to point to them. We can replace the complex busy chunk infrastructure with a simple linked list of busy extents. This allows us to remove a large chunk of code, making the overall change a net reduction in code size. Signed-off-by: Dave Chinner <david@fromorbit.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Alex Elder <aelder@sgi.com>
2010-05-21 10:07:08 +08:00
INIT_LIST_HEAD(&ntp->t_busy);
INIT_LIST_HEAD(&ntp->t_dfops);
ntp->t_firstblock = NULLFSBLOCK;
ASSERT(tp->t_flags & XFS_TRANS_PERM_LOG_RES);
ASSERT(tp->t_ticket != NULL);
ntp->t_flags = XFS_TRANS_PERM_LOG_RES |
(tp->t_flags & XFS_TRANS_RESERVE) |
xfs: preserve rmapbt swapext block reservation from freed blocks The rmapbt extent swap algorithm remaps individual extents between the source inode and the target to trigger reverse mapping metadata updates. If either inode straddles a format or other bmap allocation boundary, the individual unmap and map cycles can trigger repeated bmap block allocations and frees as the extent count bounces back and forth across the boundary. While net block usage is bound across the swap operation, this behavior can prematurely exhaust the transaction block reservation because it continuously drains as the transaction rolls. Each allocation accounts against the reservation and each free returns to global free space on transaction roll. The previous workaround to this problem attempted to detect this boundary condition and provide surplus block reservation to acommodate it. This is insufficient because more remaps can occur than implied by the extent counts; if start offset boundaries are not aligned between the two inodes, for example. To address this problem more generically and dynamically, add a transaction accounting mode that returns freed blocks to the transaction reservation instead of the superblock counters on transaction roll and use it when the rmapbt based algorithm is active. This allows the chain of remap transactions to preserve the block reservation based own its own frees and prevent premature exhaustion regardless of the remap pattern. Note that this is only safe for superblocks with lazy sb accounting, but the latter is required for v5 supers and the rmap feature depends on v5. Fixes: b3fed434822d0 ("xfs: account format bouncing into rmapbt swapext tx reservation") Root-caused-by: Darrick J. Wong <darrick.wong@oracle.com> Signed-off-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>
2020-06-30 05:44:36 +08:00
(tp->t_flags & XFS_TRANS_NO_WRITECOUNT) |
(tp->t_flags & XFS_TRANS_RES_FDBLKS);
/* We gave our writer reference to the new transaction */
tp->t_flags |= XFS_TRANS_NO_WRITECOUNT;
ntp->t_ticket = xfs_log_ticket_get(tp->t_ticket);
2018-03-10 06:01:58 +08:00
ASSERT(tp->t_blk_res >= tp->t_blk_res_used);
ntp->t_blk_res = tp->t_blk_res - tp->t_blk_res_used;
tp->t_blk_res = tp->t_blk_res_used;
2018-03-10 06:01:58 +08:00
ntp->t_rtx_res = tp->t_rtx_res - tp->t_rtx_res_used;
tp->t_rtx_res = tp->t_rtx_res_used;
xfs_trans_switch_context(tp, ntp);
xfs: support embedded dfops in transaction The dfops structure used by multi-transaction operations is typically stored on the stack and carried around by the associated transaction. The lifecycle of dfops does not quite match that of the transaction, but they are tightly related in that the former depends on the latter. The relationship of these objects is tight enough that we can avoid the cumbersome boilerplate code required in most cases to manage them separately by just embedding an xfs_defer_ops in the transaction itself. This means that a transaction allocation returns with an initialized dfops, a transaction commit finishes pending deferred items before the tx commit, a transaction cancel cancels the dfops before the transaction and a transaction dup operation transfers the current dfops state to the new transaction. The dup operation is slightly complicated by the fact that we can no longer just copy a dfops pointer from the old transaction to the new transaction. This is solved through a dfops move helper that transfers the pending items and other dfops state across the transactions. This also requires that transaction rolling code always refer to the transaction for the current dfops reference. Finally, to facilitate incremental conversion to the internal dfops and continue to support the current external dfops mode of operation, create the new ->t_dfops_internal field with a layer of indirection. On allocation, ->t_dfops points to the internal dfops. This state is overridden by callers who re-init a local dfops on the transaction. Once ->t_dfops is overridden, the external dfops reference is maintained as the transaction rolls. This patch adds the fundamental ability to support an internal dfops. All codepaths that perform deferred processing continue to override the internal dfops until they are converted over in subsequent patches. Signed-off-by: Brian Foster <bfoster@redhat.com> Reviewed-by: Bill O'Donnell <billodo@redhat.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Darrick J. Wong <darrick.wong@oracle.com> Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
2018-07-25 04:43:11 +08:00
/* move deferred ops over to the new tp */
xfs_defer_move(ntp, tp);
xfs_trans_dup_dqinfo(tp, ntp);
return ntp;
}
/*
* This is called to reserve free disk blocks and log space for the
* given transaction. This must be done before allocating any resources
* within the transaction.
*
* This will return ENOSPC if there are not enough blocks available.
* It will sleep waiting for available log space.
* The only valid value for the flags parameter is XFS_RES_LOG_PERM, which
* is used by long running transactions. If any one of the reservations
* fails then they will all be backed out.
*
* This does not do quota reservations. That typically is done by the
* caller afterwards.
*/
static int
xfs_trans_reserve(
struct xfs_trans *tp,
struct xfs_trans_res *resp,
uint blocks,
uint rtextents)
{
struct xfs_mount *mp = tp->t_mountp;
int error = 0;
bool rsvd = (tp->t_flags & XFS_TRANS_RESERVE) != 0;
/*
* Attempt to reserve the needed disk blocks by decrementing
* the number needed from the number available. This will
* fail if the count would go below zero.
*/
if (blocks > 0) {
error = xfs_mod_fdblocks(mp, -((int64_t)blocks), rsvd);
if (error != 0)
return -ENOSPC;
tp->t_blk_res += blocks;
}
/*
* Reserve the log space needed for this transaction.
*/
if (resp->tr_logres > 0) {
bool permanent = false;
ASSERT(tp->t_log_res == 0 ||
tp->t_log_res == resp->tr_logres);
ASSERT(tp->t_log_count == 0 ||
tp->t_log_count == resp->tr_logcount);
if (resp->tr_logflags & XFS_TRANS_PERM_LOG_RES) {
tp->t_flags |= XFS_TRANS_PERM_LOG_RES;
permanent = true;
} else {
ASSERT(tp->t_ticket == NULL);
ASSERT(!(tp->t_flags & XFS_TRANS_PERM_LOG_RES));
}
if (tp->t_ticket != NULL) {
ASSERT(resp->tr_logflags & XFS_TRANS_PERM_LOG_RES);
error = xfs_log_regrant(mp, tp->t_ticket);
} else {
error = xfs_log_reserve(mp, resp->tr_logres,
resp->tr_logcount,
&tp->t_ticket, permanent);
}
if (error)
goto undo_blocks;
tp->t_log_res = resp->tr_logres;
tp->t_log_count = resp->tr_logcount;
}
/*
* Attempt to reserve the needed realtime extents by decrementing
* the number needed from the number available. This will
* fail if the count would go below zero.
*/
if (rtextents > 0) {
error = xfs_mod_frextents(mp, -((int64_t)rtextents));
if (error) {
error = -ENOSPC;
goto undo_log;
}
tp->t_rtx_res += rtextents;
}
return 0;
/*
* Error cases jump to one of these labels to undo any
* reservations which have already been performed.
*/
undo_log:
if (resp->tr_logres > 0) {
xfs_log_ticket_ungrant(mp->m_log, tp->t_ticket);
tp->t_ticket = NULL;
tp->t_log_res = 0;
tp->t_flags &= ~XFS_TRANS_PERM_LOG_RES;
}
undo_blocks:
if (blocks > 0) {
xfs_mod_fdblocks(mp, (int64_t)blocks, rsvd);
tp->t_blk_res = 0;
}
return error;
}
int
xfs_trans_alloc(
struct xfs_mount *mp,
struct xfs_trans_res *resp,
uint blocks,
uint rtextents,
uint flags,
struct xfs_trans **tpp)
{
struct xfs_trans *tp;
xfs: don't nest transactions when scanning for eofblocks Brian Foster reported a lockdep warning on xfs/167: ============================================ WARNING: possible recursive locking detected 5.11.0-rc4 #35 Tainted: G W I -------------------------------------------- fsstress/17733 is trying to acquire lock: ffff8e0fd1d90650 (sb_internal){++++}-{0:0}, at: xfs_free_eofblocks+0x104/0x1d0 [xfs] but task is already holding lock: ffff8e0fd1d90650 (sb_internal){++++}-{0:0}, at: xfs_trans_alloc_inode+0x5f/0x160 [xfs] stack backtrace: CPU: 38 PID: 17733 Comm: fsstress Tainted: G W I 5.11.0-rc4 #35 Hardware name: Dell Inc. PowerEdge R740/01KPX8, BIOS 1.6.11 11/20/2018 Call Trace: dump_stack+0x8b/0xb0 __lock_acquire.cold+0x159/0x2ab lock_acquire+0x116/0x370 xfs_trans_alloc+0x1ad/0x310 [xfs] xfs_free_eofblocks+0x104/0x1d0 [xfs] xfs_blockgc_scan_inode+0x24/0x60 [xfs] xfs_inode_walk_ag+0x202/0x4b0 [xfs] xfs_inode_walk+0x66/0xc0 [xfs] xfs_trans_alloc+0x160/0x310 [xfs] xfs_trans_alloc_inode+0x5f/0x160 [xfs] xfs_alloc_file_space+0x105/0x300 [xfs] xfs_file_fallocate+0x270/0x460 [xfs] vfs_fallocate+0x14d/0x3d0 __x64_sys_fallocate+0x3e/0x70 do_syscall_64+0x33/0x40 entry_SYSCALL_64_after_hwframe+0x44/0xa9 The cause of this is the new code that spurs a scan to garbage collect speculative preallocations if we fail to reserve enough blocks while allocating a transaction. While the warning itself is a fairly benign lockdep complaint, it does expose a potential livelock if the rwsem behavior ever changes with regards to nesting read locks when someone's waiting for a write lock. Fix this by freeing the transaction and jumping back to xfs_trans_alloc like this patch in the V4 submission[1]. [1] https://lore.kernel.org/linux-xfs/161142798066.2171939.9311024588681972086.stgit@magnolia/ Fixes: a1a7d05a0576 ("xfs: flush speculative space allocations when we run out of space") Reported-by: Brian Foster <bfoster@redhat.com> Signed-off-by: Darrick J. Wong <djwong@kernel.org> Reviewed-by: Brian Foster <bfoster@redhat.com> Reviewed-by: Allison Henderson <allison.henderson@oracle.com> Reviewed-by: Christoph Hellwig <hch@lst.de>
2021-02-20 01:18:06 +08:00
bool want_retry = true;
int error;
/*
* Allocate the handle before we do our freeze accounting and setting up
* GFP_NOFS allocation context so that we avoid lockdep false positives
* by doing GFP_KERNEL allocations inside sb_start_intwrite().
*/
xfs: don't nest transactions when scanning for eofblocks Brian Foster reported a lockdep warning on xfs/167: ============================================ WARNING: possible recursive locking detected 5.11.0-rc4 #35 Tainted: G W I -------------------------------------------- fsstress/17733 is trying to acquire lock: ffff8e0fd1d90650 (sb_internal){++++}-{0:0}, at: xfs_free_eofblocks+0x104/0x1d0 [xfs] but task is already holding lock: ffff8e0fd1d90650 (sb_internal){++++}-{0:0}, at: xfs_trans_alloc_inode+0x5f/0x160 [xfs] stack backtrace: CPU: 38 PID: 17733 Comm: fsstress Tainted: G W I 5.11.0-rc4 #35 Hardware name: Dell Inc. PowerEdge R740/01KPX8, BIOS 1.6.11 11/20/2018 Call Trace: dump_stack+0x8b/0xb0 __lock_acquire.cold+0x159/0x2ab lock_acquire+0x116/0x370 xfs_trans_alloc+0x1ad/0x310 [xfs] xfs_free_eofblocks+0x104/0x1d0 [xfs] xfs_blockgc_scan_inode+0x24/0x60 [xfs] xfs_inode_walk_ag+0x202/0x4b0 [xfs] xfs_inode_walk+0x66/0xc0 [xfs] xfs_trans_alloc+0x160/0x310 [xfs] xfs_trans_alloc_inode+0x5f/0x160 [xfs] xfs_alloc_file_space+0x105/0x300 [xfs] xfs_file_fallocate+0x270/0x460 [xfs] vfs_fallocate+0x14d/0x3d0 __x64_sys_fallocate+0x3e/0x70 do_syscall_64+0x33/0x40 entry_SYSCALL_64_after_hwframe+0x44/0xa9 The cause of this is the new code that spurs a scan to garbage collect speculative preallocations if we fail to reserve enough blocks while allocating a transaction. While the warning itself is a fairly benign lockdep complaint, it does expose a potential livelock if the rwsem behavior ever changes with regards to nesting read locks when someone's waiting for a write lock. Fix this by freeing the transaction and jumping back to xfs_trans_alloc like this patch in the V4 submission[1]. [1] https://lore.kernel.org/linux-xfs/161142798066.2171939.9311024588681972086.stgit@magnolia/ Fixes: a1a7d05a0576 ("xfs: flush speculative space allocations when we run out of space") Reported-by: Brian Foster <bfoster@redhat.com> Signed-off-by: Darrick J. Wong <djwong@kernel.org> Reviewed-by: Brian Foster <bfoster@redhat.com> Reviewed-by: Allison Henderson <allison.henderson@oracle.com> Reviewed-by: Christoph Hellwig <hch@lst.de>
2021-02-20 01:18:06 +08:00
retry:
tp = kmem_cache_zalloc(xfs_trans_cache, GFP_KERNEL | __GFP_NOFAIL);
if (!(flags & XFS_TRANS_NO_WRITECOUNT))
sb_start_intwrite(mp->m_super);
xfs_trans_set_context(tp);
/*
* Zero-reservation ("empty") transactions can't modify anything, so
* they're allowed to run while we're frozen.
*/
WARN_ON(resp->tr_logres > 0 &&
mp->m_super->s_writers.frozen == SB_FREEZE_COMPLETE);
xfs: preserve rmapbt swapext block reservation from freed blocks The rmapbt extent swap algorithm remaps individual extents between the source inode and the target to trigger reverse mapping metadata updates. If either inode straddles a format or other bmap allocation boundary, the individual unmap and map cycles can trigger repeated bmap block allocations and frees as the extent count bounces back and forth across the boundary. While net block usage is bound across the swap operation, this behavior can prematurely exhaust the transaction block reservation because it continuously drains as the transaction rolls. Each allocation accounts against the reservation and each free returns to global free space on transaction roll. The previous workaround to this problem attempted to detect this boundary condition and provide surplus block reservation to acommodate it. This is insufficient because more remaps can occur than implied by the extent counts; if start offset boundaries are not aligned between the two inodes, for example. To address this problem more generically and dynamically, add a transaction accounting mode that returns freed blocks to the transaction reservation instead of the superblock counters on transaction roll and use it when the rmapbt based algorithm is active. This allows the chain of remap transactions to preserve the block reservation based own its own frees and prevent premature exhaustion regardless of the remap pattern. Note that this is only safe for superblocks with lazy sb accounting, but the latter is required for v5 supers and the rmap feature depends on v5. Fixes: b3fed434822d0 ("xfs: account format bouncing into rmapbt swapext tx reservation") Root-caused-by: Darrick J. Wong <darrick.wong@oracle.com> Signed-off-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>
2020-06-30 05:44:36 +08:00
ASSERT(!(flags & XFS_TRANS_RES_FDBLKS) ||
xfs_has_lazysbcount(mp));
tp->t_magic = XFS_TRANS_HEADER_MAGIC;
tp->t_flags = flags;
tp->t_mountp = mp;
INIT_LIST_HEAD(&tp->t_items);
INIT_LIST_HEAD(&tp->t_busy);
INIT_LIST_HEAD(&tp->t_dfops);
tp->t_firstblock = NULLFSBLOCK;
error = xfs_trans_reserve(tp, resp, blocks, rtextents);
xfs: don't nest transactions when scanning for eofblocks Brian Foster reported a lockdep warning on xfs/167: ============================================ WARNING: possible recursive locking detected 5.11.0-rc4 #35 Tainted: G W I -------------------------------------------- fsstress/17733 is trying to acquire lock: ffff8e0fd1d90650 (sb_internal){++++}-{0:0}, at: xfs_free_eofblocks+0x104/0x1d0 [xfs] but task is already holding lock: ffff8e0fd1d90650 (sb_internal){++++}-{0:0}, at: xfs_trans_alloc_inode+0x5f/0x160 [xfs] stack backtrace: CPU: 38 PID: 17733 Comm: fsstress Tainted: G W I 5.11.0-rc4 #35 Hardware name: Dell Inc. PowerEdge R740/01KPX8, BIOS 1.6.11 11/20/2018 Call Trace: dump_stack+0x8b/0xb0 __lock_acquire.cold+0x159/0x2ab lock_acquire+0x116/0x370 xfs_trans_alloc+0x1ad/0x310 [xfs] xfs_free_eofblocks+0x104/0x1d0 [xfs] xfs_blockgc_scan_inode+0x24/0x60 [xfs] xfs_inode_walk_ag+0x202/0x4b0 [xfs] xfs_inode_walk+0x66/0xc0 [xfs] xfs_trans_alloc+0x160/0x310 [xfs] xfs_trans_alloc_inode+0x5f/0x160 [xfs] xfs_alloc_file_space+0x105/0x300 [xfs] xfs_file_fallocate+0x270/0x460 [xfs] vfs_fallocate+0x14d/0x3d0 __x64_sys_fallocate+0x3e/0x70 do_syscall_64+0x33/0x40 entry_SYSCALL_64_after_hwframe+0x44/0xa9 The cause of this is the new code that spurs a scan to garbage collect speculative preallocations if we fail to reserve enough blocks while allocating a transaction. While the warning itself is a fairly benign lockdep complaint, it does expose a potential livelock if the rwsem behavior ever changes with regards to nesting read locks when someone's waiting for a write lock. Fix this by freeing the transaction and jumping back to xfs_trans_alloc like this patch in the V4 submission[1]. [1] https://lore.kernel.org/linux-xfs/161142798066.2171939.9311024588681972086.stgit@magnolia/ Fixes: a1a7d05a0576 ("xfs: flush speculative space allocations when we run out of space") Reported-by: Brian Foster <bfoster@redhat.com> Signed-off-by: Darrick J. Wong <djwong@kernel.org> Reviewed-by: Brian Foster <bfoster@redhat.com> Reviewed-by: Allison Henderson <allison.henderson@oracle.com> Reviewed-by: Christoph Hellwig <hch@lst.de>
2021-02-20 01:18:06 +08:00
if (error == -ENOSPC && want_retry) {
xfs_trans_cancel(tp);
/*
* We weren't able to reserve enough space for the transaction.
* Flush the other speculative space allocations to free space.
* Do not perform a synchronous scan because callers can hold
* other locks.
*/
xfs_blockgc_flush_all(mp);
xfs: don't nest transactions when scanning for eofblocks Brian Foster reported a lockdep warning on xfs/167: ============================================ WARNING: possible recursive locking detected 5.11.0-rc4 #35 Tainted: G W I -------------------------------------------- fsstress/17733 is trying to acquire lock: ffff8e0fd1d90650 (sb_internal){++++}-{0:0}, at: xfs_free_eofblocks+0x104/0x1d0 [xfs] but task is already holding lock: ffff8e0fd1d90650 (sb_internal){++++}-{0:0}, at: xfs_trans_alloc_inode+0x5f/0x160 [xfs] stack backtrace: CPU: 38 PID: 17733 Comm: fsstress Tainted: G W I 5.11.0-rc4 #35 Hardware name: Dell Inc. PowerEdge R740/01KPX8, BIOS 1.6.11 11/20/2018 Call Trace: dump_stack+0x8b/0xb0 __lock_acquire.cold+0x159/0x2ab lock_acquire+0x116/0x370 xfs_trans_alloc+0x1ad/0x310 [xfs] xfs_free_eofblocks+0x104/0x1d0 [xfs] xfs_blockgc_scan_inode+0x24/0x60 [xfs] xfs_inode_walk_ag+0x202/0x4b0 [xfs] xfs_inode_walk+0x66/0xc0 [xfs] xfs_trans_alloc+0x160/0x310 [xfs] xfs_trans_alloc_inode+0x5f/0x160 [xfs] xfs_alloc_file_space+0x105/0x300 [xfs] xfs_file_fallocate+0x270/0x460 [xfs] vfs_fallocate+0x14d/0x3d0 __x64_sys_fallocate+0x3e/0x70 do_syscall_64+0x33/0x40 entry_SYSCALL_64_after_hwframe+0x44/0xa9 The cause of this is the new code that spurs a scan to garbage collect speculative preallocations if we fail to reserve enough blocks while allocating a transaction. While the warning itself is a fairly benign lockdep complaint, it does expose a potential livelock if the rwsem behavior ever changes with regards to nesting read locks when someone's waiting for a write lock. Fix this by freeing the transaction and jumping back to xfs_trans_alloc like this patch in the V4 submission[1]. [1] https://lore.kernel.org/linux-xfs/161142798066.2171939.9311024588681972086.stgit@magnolia/ Fixes: a1a7d05a0576 ("xfs: flush speculative space allocations when we run out of space") Reported-by: Brian Foster <bfoster@redhat.com> Signed-off-by: Darrick J. Wong <djwong@kernel.org> Reviewed-by: Brian Foster <bfoster@redhat.com> Reviewed-by: Allison Henderson <allison.henderson@oracle.com> Reviewed-by: Christoph Hellwig <hch@lst.de>
2021-02-20 01:18:06 +08:00
want_retry = false;
goto retry;
}
if (error) {
xfs_trans_cancel(tp);
return error;
}
trace_xfs_trans_alloc(tp, _RET_IP_);
*tpp = tp;
return 0;
}
/*
* Create an empty transaction with no reservation. This is a defensive
xfs: remove the m_active_trans counter It's a global atomic counter, and we are hitting it at a rate of half a million transactions a second, so it's bouncing the counter cacheline all over the place on large machines. We don't actually need it anymore - it used to be required because the VFS freeze code could not track/prevent filesystem transactions that were running, but that problem no longer exists. Hence to remove the counter, we simply have to ensure that nothing calls xfs_sync_sb() while we are trying to quiesce the filesytem. That only happens if the log worker is still running when we call xfs_quiesce_attr(). The log worker is cancelled at the end of xfs_quiesce_attr() by calling xfs_log_quiesce(), so just call it early here and then we can remove the counter altogether. Concurrent create, 50 million inodes, identical 16p/16GB virtual machines on different physical hosts. Machine A has twice the CPU cores per socket of machine B: unpatched patched machine A: 3m16s 2m00s machine B: 4m04s 4m05s Create rates: unpatched patched machine A: 282k+/-31k 468k+/-21k machine B: 231k+/-8k 233k+/-11k Concurrent rm of same 50 million inodes: unpatched patched machine A: 6m42s 2m33s machine B: 4m47s 4m47s The transaction rate on the fast machine went from just under 300k/sec to 700k/sec, which indicates just how much of a bottleneck this atomic counter was. Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Darrick J. Wong <darrick.wong@oracle.com> Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
2020-05-21 04:17:11 +08:00
* mechanism for routines that query metadata without actually modifying them --
* if the metadata being queried is somehow cross-linked (think a btree block
* pointer that points higher in the tree), we risk deadlock. However, blocks
* grabbed as part of a transaction can be re-grabbed. The verifiers will
* notice the corrupt block and the operation will fail back to userspace
* without deadlocking.
*
xfs: remove the m_active_trans counter It's a global atomic counter, and we are hitting it at a rate of half a million transactions a second, so it's bouncing the counter cacheline all over the place on large machines. We don't actually need it anymore - it used to be required because the VFS freeze code could not track/prevent filesystem transactions that were running, but that problem no longer exists. Hence to remove the counter, we simply have to ensure that nothing calls xfs_sync_sb() while we are trying to quiesce the filesytem. That only happens if the log worker is still running when we call xfs_quiesce_attr(). The log worker is cancelled at the end of xfs_quiesce_attr() by calling xfs_log_quiesce(), so just call it early here and then we can remove the counter altogether. Concurrent create, 50 million inodes, identical 16p/16GB virtual machines on different physical hosts. Machine A has twice the CPU cores per socket of machine B: unpatched patched machine A: 3m16s 2m00s machine B: 4m04s 4m05s Create rates: unpatched patched machine A: 282k+/-31k 468k+/-21k machine B: 231k+/-8k 233k+/-11k Concurrent rm of same 50 million inodes: unpatched patched machine A: 6m42s 2m33s machine B: 4m47s 4m47s The transaction rate on the fast machine went from just under 300k/sec to 700k/sec, which indicates just how much of a bottleneck this atomic counter was. Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Darrick J. Wong <darrick.wong@oracle.com> Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
2020-05-21 04:17:11 +08:00
* Note the zero-length reservation; this transaction MUST be cancelled without
* any dirty data.
xfs: prohibit fs freezing when using empty transactions I noticed that fsfreeze can take a very long time to freeze an XFS if there happens to be a GETFSMAP caller running in the background. I also happened to notice the following in dmesg: ------------[ cut here ]------------ WARNING: CPU: 2 PID: 43492 at fs/xfs/xfs_super.c:853 xfs_quiesce_attr+0x83/0x90 [xfs] Modules linked in: xfs libcrc32c ip6t_REJECT nf_reject_ipv6 ipt_REJECT nf_reject_ipv4 ip_set_hash_ip ip_set_hash_net xt_tcpudp xt_set ip_set_hash_mac ip_set nfnetlink ip6table_filter ip6_tables bfq iptable_filter sch_fq_codel ip_tables x_tables nfsv4 af_packet [last unloaded: xfs] CPU: 2 PID: 43492 Comm: xfs_io Not tainted 5.6.0-rc4-djw #rc4 Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), BIOS 1.10.2-1ubuntu1 04/01/2014 RIP: 0010:xfs_quiesce_attr+0x83/0x90 [xfs] Code: 7c 07 00 00 85 c0 75 22 48 89 df 5b e9 96 c1 00 00 48 c7 c6 b0 2d 38 a0 48 89 df e8 57 64 ff ff 8b 83 7c 07 00 00 85 c0 74 de <0f> 0b 48 89 df 5b e9 72 c1 00 00 66 90 0f 1f 44 00 00 41 55 41 54 RSP: 0018:ffffc900030f3e28 EFLAGS: 00010202 RAX: 0000000000000001 RBX: ffff88802ac54000 RCX: 0000000000000000 RDX: 0000000000000000 RSI: ffffffff81e4a6f0 RDI: 00000000ffffffff RBP: ffff88807859f070 R08: 0000000000000001 R09: 0000000000000000 R10: 0000000000000000 R11: 0000000000000010 R12: 0000000000000000 R13: ffff88807859f388 R14: ffff88807859f4b8 R15: ffff88807859f5e8 FS: 00007fad1c6c0fc0(0000) GS:ffff88807e000000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 CR2: 00007f0c7d237000 CR3: 0000000077f01003 CR4: 00000000001606a0 Call Trace: xfs_fs_freeze+0x25/0x40 [xfs] freeze_super+0xc8/0x180 do_vfs_ioctl+0x70b/0x750 ? __fget_files+0x135/0x210 ksys_ioctl+0x3a/0xb0 __x64_sys_ioctl+0x16/0x20 do_syscall_64+0x50/0x1a0 entry_SYSCALL_64_after_hwframe+0x49/0xbe These two things appear to be related. The assertion trips when another thread initiates a fsmap request (which uses an empty transaction) after the freezer waited for m_active_trans to hit zero but before the the freezer executes the WARN_ON just prior to calling xfs_log_quiesce. The lengthy delays in freezing happen because the freezer calls xfs_wait_buftarg to clean out the buffer lru list. Meanwhile, the GETFSMAP caller is continuing to grab and release buffers, which means that it can take a very long time for the buffer lru list to empty out. We fix both of these races by calling sb_start_write to obtain freeze protection while using empty transactions for GETFSMAP and for metadata scrubbing. The other two users occur during mount, during which time we cannot fs freeze. Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com> Reviewed-by: Dave Chinner <dchinner@redhat.com>
2020-03-25 14:03:24 +08:00
*
xfs: remove the m_active_trans counter It's a global atomic counter, and we are hitting it at a rate of half a million transactions a second, so it's bouncing the counter cacheline all over the place on large machines. We don't actually need it anymore - it used to be required because the VFS freeze code could not track/prevent filesystem transactions that were running, but that problem no longer exists. Hence to remove the counter, we simply have to ensure that nothing calls xfs_sync_sb() while we are trying to quiesce the filesytem. That only happens if the log worker is still running when we call xfs_quiesce_attr(). The log worker is cancelled at the end of xfs_quiesce_attr() by calling xfs_log_quiesce(), so just call it early here and then we can remove the counter altogether. Concurrent create, 50 million inodes, identical 16p/16GB virtual machines on different physical hosts. Machine A has twice the CPU cores per socket of machine B: unpatched patched machine A: 3m16s 2m00s machine B: 4m04s 4m05s Create rates: unpatched patched machine A: 282k+/-31k 468k+/-21k machine B: 231k+/-8k 233k+/-11k Concurrent rm of same 50 million inodes: unpatched patched machine A: 6m42s 2m33s machine B: 4m47s 4m47s The transaction rate on the fast machine went from just under 300k/sec to 700k/sec, which indicates just how much of a bottleneck this atomic counter was. Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Darrick J. Wong <darrick.wong@oracle.com> Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
2020-05-21 04:17:11 +08:00
* Callers should obtain freeze protection to avoid a conflict with fs freezing
* where we can be grabbing buffers at the same time that freeze is trying to
* drain the buffer LRU list.
*/
int
xfs_trans_alloc_empty(
struct xfs_mount *mp,
struct xfs_trans **tpp)
{
struct xfs_trans_res resv = {0};
return xfs_trans_alloc(mp, &resv, 0, 0, XFS_TRANS_NO_WRITECOUNT, tpp);
}
/*
* Record the indicated change to the given field for application
* to the file system's superblock when the transaction commits.
* For now, just store the change in the transaction structure.
*
* Mark the transaction structure to indicate that the superblock
* needs to be updated before committing.
[XFS] Lazy Superblock Counters When we have a couple of hundred transactions on the fly at once, they all typically modify the on disk superblock in some way. create/unclink/mkdir/rmdir modify inode counts, allocation/freeing modify free block counts. When these counts are modified in a transaction, they must eventually lock the superblock buffer and apply the mods. The buffer then remains locked until the transaction is committed into the incore log buffer. The result of this is that with enough transactions on the fly the incore superblock buffer becomes a bottleneck. The result of contention on the incore superblock buffer is that transaction rates fall - the more pressure that is put on the superblock buffer, the slower things go. The key to removing the contention is to not require the superblock fields in question to be locked. We do that by not marking the superblock dirty in the transaction. IOWs, we modify the incore superblock but do not modify the cached superblock buffer. In short, we do not log superblock modifications to critical fields in the superblock on every transaction. In fact we only do it just before we write the superblock to disk every sync period or just before unmount. This creates an interesting problem - if we don't log or write out the fields in every transaction, then how do the values get recovered after a crash? the answer is simple - we keep enough duplicate, logged information in other structures that we can reconstruct the correct count after log recovery has been performed. It is the AGF and AGI structures that contain the duplicate information; after recovery, we walk every AGI and AGF and sum their individual counters to get the correct value, and we do a transaction into the log to correct them. An optimisation of this is that if we have a clean unmount record, we know the value in the superblock is correct, so we can avoid the summation walk under normal conditions and so mount/recovery times do not change under normal operation. One wrinkle that was discovered during development was that the blocks used in the freespace btrees are never accounted for in the AGF counters. This was once a valid optimisation to make; when the filesystem is full, the free space btrees are empty and consume no space. Hence when it matters, the "accounting" is correct. But that means the when we do the AGF summations, we would not have a correct count and xfs_check would complain. Hence a new counter was added to track the number of blocks used by the free space btrees. This is an *on-disk format change*. As a result of this, lazy superblock counters are a mkfs option and at the moment on linux there is no way to convert an old filesystem. This is possible - xfs_db can be used to twiddle the right bits and then xfs_repair will do the format conversion for you. Similarly, you can convert backwards as well. At some point we'll add functionality to xfs_admin to do the bit twiddling easily.... SGI-PV: 964999 SGI-Modid: xfs-linux-melb:xfs-kern:28652a Signed-off-by: David Chinner <dgc@sgi.com> Signed-off-by: Christoph Hellwig <hch@infradead.org> Signed-off-by: Tim Shimmin <tes@sgi.com>
2007-05-24 13:26:31 +08:00
*
* Because we may not be keeping track of allocated/free inodes and
* used filesystem blocks in the superblock, we do not mark the
* superblock dirty in this transaction if we modify these fields.
* We still need to update the transaction deltas so that they get
* applied to the incore superblock, but we don't want them to
* cause the superblock to get locked and logged if these are the
* only fields in the superblock that the transaction modifies.
*/
void
xfs_trans_mod_sb(
xfs_trans_t *tp,
uint field,
int64_t delta)
{
[XFS] Lazy Superblock Counters When we have a couple of hundred transactions on the fly at once, they all typically modify the on disk superblock in some way. create/unclink/mkdir/rmdir modify inode counts, allocation/freeing modify free block counts. When these counts are modified in a transaction, they must eventually lock the superblock buffer and apply the mods. The buffer then remains locked until the transaction is committed into the incore log buffer. The result of this is that with enough transactions on the fly the incore superblock buffer becomes a bottleneck. The result of contention on the incore superblock buffer is that transaction rates fall - the more pressure that is put on the superblock buffer, the slower things go. The key to removing the contention is to not require the superblock fields in question to be locked. We do that by not marking the superblock dirty in the transaction. IOWs, we modify the incore superblock but do not modify the cached superblock buffer. In short, we do not log superblock modifications to critical fields in the superblock on every transaction. In fact we only do it just before we write the superblock to disk every sync period or just before unmount. This creates an interesting problem - if we don't log or write out the fields in every transaction, then how do the values get recovered after a crash? the answer is simple - we keep enough duplicate, logged information in other structures that we can reconstruct the correct count after log recovery has been performed. It is the AGF and AGI structures that contain the duplicate information; after recovery, we walk every AGI and AGF and sum their individual counters to get the correct value, and we do a transaction into the log to correct them. An optimisation of this is that if we have a clean unmount record, we know the value in the superblock is correct, so we can avoid the summation walk under normal conditions and so mount/recovery times do not change under normal operation. One wrinkle that was discovered during development was that the blocks used in the freespace btrees are never accounted for in the AGF counters. This was once a valid optimisation to make; when the filesystem is full, the free space btrees are empty and consume no space. Hence when it matters, the "accounting" is correct. But that means the when we do the AGF summations, we would not have a correct count and xfs_check would complain. Hence a new counter was added to track the number of blocks used by the free space btrees. This is an *on-disk format change*. As a result of this, lazy superblock counters are a mkfs option and at the moment on linux there is no way to convert an old filesystem. This is possible - xfs_db can be used to twiddle the right bits and then xfs_repair will do the format conversion for you. Similarly, you can convert backwards as well. At some point we'll add functionality to xfs_admin to do the bit twiddling easily.... SGI-PV: 964999 SGI-Modid: xfs-linux-melb:xfs-kern:28652a Signed-off-by: David Chinner <dgc@sgi.com> Signed-off-by: Christoph Hellwig <hch@infradead.org> Signed-off-by: Tim Shimmin <tes@sgi.com>
2007-05-24 13:26:31 +08:00
uint32_t flags = (XFS_TRANS_DIRTY|XFS_TRANS_SB_DIRTY);
xfs_mount_t *mp = tp->t_mountp;
switch (field) {
case XFS_TRANS_SB_ICOUNT:
tp->t_icount_delta += delta;
if (xfs_has_lazysbcount(mp))
[XFS] Lazy Superblock Counters When we have a couple of hundred transactions on the fly at once, they all typically modify the on disk superblock in some way. create/unclink/mkdir/rmdir modify inode counts, allocation/freeing modify free block counts. When these counts are modified in a transaction, they must eventually lock the superblock buffer and apply the mods. The buffer then remains locked until the transaction is committed into the incore log buffer. The result of this is that with enough transactions on the fly the incore superblock buffer becomes a bottleneck. The result of contention on the incore superblock buffer is that transaction rates fall - the more pressure that is put on the superblock buffer, the slower things go. The key to removing the contention is to not require the superblock fields in question to be locked. We do that by not marking the superblock dirty in the transaction. IOWs, we modify the incore superblock but do not modify the cached superblock buffer. In short, we do not log superblock modifications to critical fields in the superblock on every transaction. In fact we only do it just before we write the superblock to disk every sync period or just before unmount. This creates an interesting problem - if we don't log or write out the fields in every transaction, then how do the values get recovered after a crash? the answer is simple - we keep enough duplicate, logged information in other structures that we can reconstruct the correct count after log recovery has been performed. It is the AGF and AGI structures that contain the duplicate information; after recovery, we walk every AGI and AGF and sum their individual counters to get the correct value, and we do a transaction into the log to correct them. An optimisation of this is that if we have a clean unmount record, we know the value in the superblock is correct, so we can avoid the summation walk under normal conditions and so mount/recovery times do not change under normal operation. One wrinkle that was discovered during development was that the blocks used in the freespace btrees are never accounted for in the AGF counters. This was once a valid optimisation to make; when the filesystem is full, the free space btrees are empty and consume no space. Hence when it matters, the "accounting" is correct. But that means the when we do the AGF summations, we would not have a correct count and xfs_check would complain. Hence a new counter was added to track the number of blocks used by the free space btrees. This is an *on-disk format change*. As a result of this, lazy superblock counters are a mkfs option and at the moment on linux there is no way to convert an old filesystem. This is possible - xfs_db can be used to twiddle the right bits and then xfs_repair will do the format conversion for you. Similarly, you can convert backwards as well. At some point we'll add functionality to xfs_admin to do the bit twiddling easily.... SGI-PV: 964999 SGI-Modid: xfs-linux-melb:xfs-kern:28652a Signed-off-by: David Chinner <dgc@sgi.com> Signed-off-by: Christoph Hellwig <hch@infradead.org> Signed-off-by: Tim Shimmin <tes@sgi.com>
2007-05-24 13:26:31 +08:00
flags &= ~XFS_TRANS_SB_DIRTY;
break;
case XFS_TRANS_SB_IFREE:
tp->t_ifree_delta += delta;
if (xfs_has_lazysbcount(mp))
[XFS] Lazy Superblock Counters When we have a couple of hundred transactions on the fly at once, they all typically modify the on disk superblock in some way. create/unclink/mkdir/rmdir modify inode counts, allocation/freeing modify free block counts. When these counts are modified in a transaction, they must eventually lock the superblock buffer and apply the mods. The buffer then remains locked until the transaction is committed into the incore log buffer. The result of this is that with enough transactions on the fly the incore superblock buffer becomes a bottleneck. The result of contention on the incore superblock buffer is that transaction rates fall - the more pressure that is put on the superblock buffer, the slower things go. The key to removing the contention is to not require the superblock fields in question to be locked. We do that by not marking the superblock dirty in the transaction. IOWs, we modify the incore superblock but do not modify the cached superblock buffer. In short, we do not log superblock modifications to critical fields in the superblock on every transaction. In fact we only do it just before we write the superblock to disk every sync period or just before unmount. This creates an interesting problem - if we don't log or write out the fields in every transaction, then how do the values get recovered after a crash? the answer is simple - we keep enough duplicate, logged information in other structures that we can reconstruct the correct count after log recovery has been performed. It is the AGF and AGI structures that contain the duplicate information; after recovery, we walk every AGI and AGF and sum their individual counters to get the correct value, and we do a transaction into the log to correct them. An optimisation of this is that if we have a clean unmount record, we know the value in the superblock is correct, so we can avoid the summation walk under normal conditions and so mount/recovery times do not change under normal operation. One wrinkle that was discovered during development was that the blocks used in the freespace btrees are never accounted for in the AGF counters. This was once a valid optimisation to make; when the filesystem is full, the free space btrees are empty and consume no space. Hence when it matters, the "accounting" is correct. But that means the when we do the AGF summations, we would not have a correct count and xfs_check would complain. Hence a new counter was added to track the number of blocks used by the free space btrees. This is an *on-disk format change*. As a result of this, lazy superblock counters are a mkfs option and at the moment on linux there is no way to convert an old filesystem. This is possible - xfs_db can be used to twiddle the right bits and then xfs_repair will do the format conversion for you. Similarly, you can convert backwards as well. At some point we'll add functionality to xfs_admin to do the bit twiddling easily.... SGI-PV: 964999 SGI-Modid: xfs-linux-melb:xfs-kern:28652a Signed-off-by: David Chinner <dgc@sgi.com> Signed-off-by: Christoph Hellwig <hch@infradead.org> Signed-off-by: Tim Shimmin <tes@sgi.com>
2007-05-24 13:26:31 +08:00
flags &= ~XFS_TRANS_SB_DIRTY;
break;
case XFS_TRANS_SB_FDBLOCKS:
/*
2018-03-10 06:01:58 +08:00
* Track the number of blocks allocated in the transaction.
* Make sure it does not exceed the number reserved. If so,
* shutdown as this can lead to accounting inconsistency.
*/
if (delta < 0) {
tp->t_blk_res_used += (uint)-delta;
2018-03-10 06:01:58 +08:00
if (tp->t_blk_res_used > tp->t_blk_res)
xfs_force_shutdown(mp, SHUTDOWN_CORRUPT_INCORE);
xfs: preserve rmapbt swapext block reservation from freed blocks The rmapbt extent swap algorithm remaps individual extents between the source inode and the target to trigger reverse mapping metadata updates. If either inode straddles a format or other bmap allocation boundary, the individual unmap and map cycles can trigger repeated bmap block allocations and frees as the extent count bounces back and forth across the boundary. While net block usage is bound across the swap operation, this behavior can prematurely exhaust the transaction block reservation because it continuously drains as the transaction rolls. Each allocation accounts against the reservation and each free returns to global free space on transaction roll. The previous workaround to this problem attempted to detect this boundary condition and provide surplus block reservation to acommodate it. This is insufficient because more remaps can occur than implied by the extent counts; if start offset boundaries are not aligned between the two inodes, for example. To address this problem more generically and dynamically, add a transaction accounting mode that returns freed blocks to the transaction reservation instead of the superblock counters on transaction roll and use it when the rmapbt based algorithm is active. This allows the chain of remap transactions to preserve the block reservation based own its own frees and prevent premature exhaustion regardless of the remap pattern. Note that this is only safe for superblocks with lazy sb accounting, but the latter is required for v5 supers and the rmap feature depends on v5. Fixes: b3fed434822d0 ("xfs: account format bouncing into rmapbt swapext tx reservation") Root-caused-by: Darrick J. Wong <darrick.wong@oracle.com> Signed-off-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>
2020-06-30 05:44:36 +08:00
} else if (delta > 0 && (tp->t_flags & XFS_TRANS_RES_FDBLKS)) {
int64_t blkres_delta;
/*
* Return freed blocks directly to the reservation
* instead of the global pool, being careful not to
* overflow the trans counter. This is used to preserve
* reservation across chains of transaction rolls that
* repeatedly free and allocate blocks.
*/
blkres_delta = min_t(int64_t, delta,
UINT_MAX - tp->t_blk_res);
tp->t_blk_res += blkres_delta;
delta -= blkres_delta;
}
tp->t_fdblocks_delta += delta;
if (xfs_has_lazysbcount(mp))
[XFS] Lazy Superblock Counters When we have a couple of hundred transactions on the fly at once, they all typically modify the on disk superblock in some way. create/unclink/mkdir/rmdir modify inode counts, allocation/freeing modify free block counts. When these counts are modified in a transaction, they must eventually lock the superblock buffer and apply the mods. The buffer then remains locked until the transaction is committed into the incore log buffer. The result of this is that with enough transactions on the fly the incore superblock buffer becomes a bottleneck. The result of contention on the incore superblock buffer is that transaction rates fall - the more pressure that is put on the superblock buffer, the slower things go. The key to removing the contention is to not require the superblock fields in question to be locked. We do that by not marking the superblock dirty in the transaction. IOWs, we modify the incore superblock but do not modify the cached superblock buffer. In short, we do not log superblock modifications to critical fields in the superblock on every transaction. In fact we only do it just before we write the superblock to disk every sync period or just before unmount. This creates an interesting problem - if we don't log or write out the fields in every transaction, then how do the values get recovered after a crash? the answer is simple - we keep enough duplicate, logged information in other structures that we can reconstruct the correct count after log recovery has been performed. It is the AGF and AGI structures that contain the duplicate information; after recovery, we walk every AGI and AGF and sum their individual counters to get the correct value, and we do a transaction into the log to correct them. An optimisation of this is that if we have a clean unmount record, we know the value in the superblock is correct, so we can avoid the summation walk under normal conditions and so mount/recovery times do not change under normal operation. One wrinkle that was discovered during development was that the blocks used in the freespace btrees are never accounted for in the AGF counters. This was once a valid optimisation to make; when the filesystem is full, the free space btrees are empty and consume no space. Hence when it matters, the "accounting" is correct. But that means the when we do the AGF summations, we would not have a correct count and xfs_check would complain. Hence a new counter was added to track the number of blocks used by the free space btrees. This is an *on-disk format change*. As a result of this, lazy superblock counters are a mkfs option and at the moment on linux there is no way to convert an old filesystem. This is possible - xfs_db can be used to twiddle the right bits and then xfs_repair will do the format conversion for you. Similarly, you can convert backwards as well. At some point we'll add functionality to xfs_admin to do the bit twiddling easily.... SGI-PV: 964999 SGI-Modid: xfs-linux-melb:xfs-kern:28652a Signed-off-by: David Chinner <dgc@sgi.com> Signed-off-by: Christoph Hellwig <hch@infradead.org> Signed-off-by: Tim Shimmin <tes@sgi.com>
2007-05-24 13:26:31 +08:00
flags &= ~XFS_TRANS_SB_DIRTY;
break;
case XFS_TRANS_SB_RES_FDBLOCKS:
/*
* The allocation has already been applied to the
* in-core superblock's counter. This should only
* be applied to the on-disk superblock.
*/
tp->t_res_fdblocks_delta += delta;
if (xfs_has_lazysbcount(mp))
[XFS] Lazy Superblock Counters When we have a couple of hundred transactions on the fly at once, they all typically modify the on disk superblock in some way. create/unclink/mkdir/rmdir modify inode counts, allocation/freeing modify free block counts. When these counts are modified in a transaction, they must eventually lock the superblock buffer and apply the mods. The buffer then remains locked until the transaction is committed into the incore log buffer. The result of this is that with enough transactions on the fly the incore superblock buffer becomes a bottleneck. The result of contention on the incore superblock buffer is that transaction rates fall - the more pressure that is put on the superblock buffer, the slower things go. The key to removing the contention is to not require the superblock fields in question to be locked. We do that by not marking the superblock dirty in the transaction. IOWs, we modify the incore superblock but do not modify the cached superblock buffer. In short, we do not log superblock modifications to critical fields in the superblock on every transaction. In fact we only do it just before we write the superblock to disk every sync period or just before unmount. This creates an interesting problem - if we don't log or write out the fields in every transaction, then how do the values get recovered after a crash? the answer is simple - we keep enough duplicate, logged information in other structures that we can reconstruct the correct count after log recovery has been performed. It is the AGF and AGI structures that contain the duplicate information; after recovery, we walk every AGI and AGF and sum their individual counters to get the correct value, and we do a transaction into the log to correct them. An optimisation of this is that if we have a clean unmount record, we know the value in the superblock is correct, so we can avoid the summation walk under normal conditions and so mount/recovery times do not change under normal operation. One wrinkle that was discovered during development was that the blocks used in the freespace btrees are never accounted for in the AGF counters. This was once a valid optimisation to make; when the filesystem is full, the free space btrees are empty and consume no space. Hence when it matters, the "accounting" is correct. But that means the when we do the AGF summations, we would not have a correct count and xfs_check would complain. Hence a new counter was added to track the number of blocks used by the free space btrees. This is an *on-disk format change*. As a result of this, lazy superblock counters are a mkfs option and at the moment on linux there is no way to convert an old filesystem. This is possible - xfs_db can be used to twiddle the right bits and then xfs_repair will do the format conversion for you. Similarly, you can convert backwards as well. At some point we'll add functionality to xfs_admin to do the bit twiddling easily.... SGI-PV: 964999 SGI-Modid: xfs-linux-melb:xfs-kern:28652a Signed-off-by: David Chinner <dgc@sgi.com> Signed-off-by: Christoph Hellwig <hch@infradead.org> Signed-off-by: Tim Shimmin <tes@sgi.com>
2007-05-24 13:26:31 +08:00
flags &= ~XFS_TRANS_SB_DIRTY;
break;
case XFS_TRANS_SB_FREXTENTS:
/*
* Track the number of blocks allocated in the
* transaction. Make sure it does not exceed the
* number reserved.
*/
if (delta < 0) {
tp->t_rtx_res_used += (uint)-delta;
ASSERT(tp->t_rtx_res_used <= tp->t_rtx_res);
}
tp->t_frextents_delta += delta;
break;
case XFS_TRANS_SB_RES_FREXTENTS:
/*
* The allocation has already been applied to the
* in-core superblock's counter. This should only
* be applied to the on-disk superblock.
*/
ASSERT(delta < 0);
tp->t_res_frextents_delta += delta;
break;
case XFS_TRANS_SB_DBLOCKS:
tp->t_dblocks_delta += delta;
break;
case XFS_TRANS_SB_AGCOUNT:
ASSERT(delta > 0);
tp->t_agcount_delta += delta;
break;
case XFS_TRANS_SB_IMAXPCT:
tp->t_imaxpct_delta += delta;
break;
case XFS_TRANS_SB_REXTSIZE:
tp->t_rextsize_delta += delta;
break;
case XFS_TRANS_SB_RBMBLOCKS:
tp->t_rbmblocks_delta += delta;
break;
case XFS_TRANS_SB_RBLOCKS:
tp->t_rblocks_delta += delta;
break;
case XFS_TRANS_SB_REXTENTS:
tp->t_rextents_delta += delta;
break;
case XFS_TRANS_SB_REXTSLOG:
tp->t_rextslog_delta += delta;
break;
default:
ASSERT(0);
return;
}
tp->t_flags |= flags;
}
/*
* xfs_trans_apply_sb_deltas() is called from the commit code
* to bring the superblock buffer into the current transaction
* and modify it as requested by earlier calls to xfs_trans_mod_sb().
*
* For now we just look at each field allowed to change and change
* it if necessary.
*/
STATIC void
xfs_trans_apply_sb_deltas(
xfs_trans_t *tp)
{
struct xfs_dsb *sbp;
struct xfs_buf *bp;
int whole = 0;
bp = xfs_trans_getsb(tp);
sbp = bp->b_addr;
[XFS] Lazy Superblock Counters When we have a couple of hundred transactions on the fly at once, they all typically modify the on disk superblock in some way. create/unclink/mkdir/rmdir modify inode counts, allocation/freeing modify free block counts. When these counts are modified in a transaction, they must eventually lock the superblock buffer and apply the mods. The buffer then remains locked until the transaction is committed into the incore log buffer. The result of this is that with enough transactions on the fly the incore superblock buffer becomes a bottleneck. The result of contention on the incore superblock buffer is that transaction rates fall - the more pressure that is put on the superblock buffer, the slower things go. The key to removing the contention is to not require the superblock fields in question to be locked. We do that by not marking the superblock dirty in the transaction. IOWs, we modify the incore superblock but do not modify the cached superblock buffer. In short, we do not log superblock modifications to critical fields in the superblock on every transaction. In fact we only do it just before we write the superblock to disk every sync period or just before unmount. This creates an interesting problem - if we don't log or write out the fields in every transaction, then how do the values get recovered after a crash? the answer is simple - we keep enough duplicate, logged information in other structures that we can reconstruct the correct count after log recovery has been performed. It is the AGF and AGI structures that contain the duplicate information; after recovery, we walk every AGI and AGF and sum their individual counters to get the correct value, and we do a transaction into the log to correct them. An optimisation of this is that if we have a clean unmount record, we know the value in the superblock is correct, so we can avoid the summation walk under normal conditions and so mount/recovery times do not change under normal operation. One wrinkle that was discovered during development was that the blocks used in the freespace btrees are never accounted for in the AGF counters. This was once a valid optimisation to make; when the filesystem is full, the free space btrees are empty and consume no space. Hence when it matters, the "accounting" is correct. But that means the when we do the AGF summations, we would not have a correct count and xfs_check would complain. Hence a new counter was added to track the number of blocks used by the free space btrees. This is an *on-disk format change*. As a result of this, lazy superblock counters are a mkfs option and at the moment on linux there is no way to convert an old filesystem. This is possible - xfs_db can be used to twiddle the right bits and then xfs_repair will do the format conversion for you. Similarly, you can convert backwards as well. At some point we'll add functionality to xfs_admin to do the bit twiddling easily.... SGI-PV: 964999 SGI-Modid: xfs-linux-melb:xfs-kern:28652a Signed-off-by: David Chinner <dgc@sgi.com> Signed-off-by: Christoph Hellwig <hch@infradead.org> Signed-off-by: Tim Shimmin <tes@sgi.com>
2007-05-24 13:26:31 +08:00
/*
* Only update the superblock counters if we are logging them
*/
if (!xfs_has_lazysbcount((tp->t_mountp))) {
if (tp->t_icount_delta)
be64_add_cpu(&sbp->sb_icount, tp->t_icount_delta);
if (tp->t_ifree_delta)
be64_add_cpu(&sbp->sb_ifree, tp->t_ifree_delta);
if (tp->t_fdblocks_delta)
be64_add_cpu(&sbp->sb_fdblocks, tp->t_fdblocks_delta);
if (tp->t_res_fdblocks_delta)
be64_add_cpu(&sbp->sb_fdblocks, tp->t_res_fdblocks_delta);
}
/*
* Updating frextents requires careful handling because it does not
* behave like the lazysb counters because we cannot rely on log
* recovery in older kenels to recompute the value from the rtbitmap.
* This means that the ondisk frextents must be consistent with the
* rtbitmap.
*
* Therefore, log the frextents change to the ondisk superblock and
* update the incore superblock so that future calls to xfs_log_sb
* write the correct value ondisk.
*
* Don't touch m_frextents because it includes incore reservations,
* and those are handled by the unreserve function.
*/
if (tp->t_frextents_delta || tp->t_res_frextents_delta) {
struct xfs_mount *mp = tp->t_mountp;
int64_t rtxdelta;
rtxdelta = tp->t_frextents_delta + tp->t_res_frextents_delta;
spin_lock(&mp->m_sb_lock);
be64_add_cpu(&sbp->sb_frextents, rtxdelta);
mp->m_sb.sb_frextents += rtxdelta;
spin_unlock(&mp->m_sb_lock);
}
if (tp->t_dblocks_delta) {
be64_add_cpu(&sbp->sb_dblocks, tp->t_dblocks_delta);
whole = 1;
}
if (tp->t_agcount_delta) {
be32_add_cpu(&sbp->sb_agcount, tp->t_agcount_delta);
whole = 1;
}
if (tp->t_imaxpct_delta) {
sbp->sb_imax_pct += tp->t_imaxpct_delta;
whole = 1;
}
if (tp->t_rextsize_delta) {
be32_add_cpu(&sbp->sb_rextsize, tp->t_rextsize_delta);
whole = 1;
}
if (tp->t_rbmblocks_delta) {
be32_add_cpu(&sbp->sb_rbmblocks, tp->t_rbmblocks_delta);
whole = 1;
}
if (tp->t_rblocks_delta) {
be64_add_cpu(&sbp->sb_rblocks, tp->t_rblocks_delta);
whole = 1;
}
if (tp->t_rextents_delta) {
be64_add_cpu(&sbp->sb_rextents, tp->t_rextents_delta);
whole = 1;
}
if (tp->t_rextslog_delta) {
sbp->sb_rextslog += tp->t_rextslog_delta;
whole = 1;
}
xfs_trans_buf_set_type(tp, bp, XFS_BLFT_SB_BUF);
if (whole)
/*
* Log the whole thing, the fields are noncontiguous.
*/
xfs_trans_log_buf(tp, bp, 0, sizeof(struct xfs_dsb) - 1);
else
/*
* Since all the modifiable fields are contiguous, we
* can get away with this.
*/
xfs_trans_log_buf(tp, bp, offsetof(struct xfs_dsb, sb_icount),
offsetof(struct xfs_dsb, sb_frextents) +
sizeof(sbp->sb_frextents) - 1);
}
/*
* xfs_trans_unreserve_and_mod_sb() is called to release unused reservations and
* apply superblock counter changes to the in-core superblock. The
* t_res_fdblocks_delta and t_res_frextents_delta fields are explicitly NOT
* applied to the in-core superblock. The idea is that that has already been
* done.
*
* If we are not logging superblock counters, then the inode allocated/free and
* used block counts are not updated in the on disk superblock. In this case,
* XFS_TRANS_SB_DIRTY will not be set when the transaction is updated but we
* still need to update the incore superblock with the changes.
xfs: reduce free inode accounting overhead Shaokun Zhang reported that XFS was using substantial CPU time in percpu_count_sum() when running a single threaded benchmark on a high CPU count (128p) machine from xfs_mod_ifree(). The issue is that the filesystem is empty when the benchmark runs, so inode allocation is running with a very low inode free count. With the percpu counter batching, this means comparisons when the counter is less that 128 * 256 = 32768 use the slow path of adding up all the counters across the CPUs, and this is expensive on high CPU count machines. The summing in xfs_mod_ifree() is only used to fire an assert if an underrun occurs. The error is ignored by the higher level code. Hence this is really just debug code and we don't need to run it on production kernels, nor do we need such debug checks to return error values just to trigger an assert. Finally, xfs_mod_icount/xfs_mod_ifree are only called from xfs_trans_unreserve_and_mod_sb(), so get rid of them and just directly call the percpu_counter_add/percpu_counter_compare functions. The compare functions are now run only on debug builds as they are internal to ASSERT() checks and so only compiled in when ASSERTs are active (CONFIG_XFS_DEBUG=y or CONFIG_XFS_WARN=y). Reported-by: Shaokun Zhang <zhangshaokun@hisilicon.com> Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Darrick J. Wong <darrick.wong@oracle.com> Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
2020-05-21 04:17:11 +08:00
*
* Deltas for the inode count are +/-64, hence we use a large batch size of 128
* so we don't need to take the counter lock on every update.
*/
xfs: reduce free inode accounting overhead Shaokun Zhang reported that XFS was using substantial CPU time in percpu_count_sum() when running a single threaded benchmark on a high CPU count (128p) machine from xfs_mod_ifree(). The issue is that the filesystem is empty when the benchmark runs, so inode allocation is running with a very low inode free count. With the percpu counter batching, this means comparisons when the counter is less that 128 * 256 = 32768 use the slow path of adding up all the counters across the CPUs, and this is expensive on high CPU count machines. The summing in xfs_mod_ifree() is only used to fire an assert if an underrun occurs. The error is ignored by the higher level code. Hence this is really just debug code and we don't need to run it on production kernels, nor do we need such debug checks to return error values just to trigger an assert. Finally, xfs_mod_icount/xfs_mod_ifree are only called from xfs_trans_unreserve_and_mod_sb(), so get rid of them and just directly call the percpu_counter_add/percpu_counter_compare functions. The compare functions are now run only on debug builds as they are internal to ASSERT() checks and so only compiled in when ASSERTs are active (CONFIG_XFS_DEBUG=y or CONFIG_XFS_WARN=y). Reported-by: Shaokun Zhang <zhangshaokun@hisilicon.com> Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Darrick J. Wong <darrick.wong@oracle.com> Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
2020-05-21 04:17:11 +08:00
#define XFS_ICOUNT_BATCH 128
xfs: Introduce delayed logging core code The delayed logging code only changes in-memory structures and as such can be enabled and disabled with a mount option. Add the mount option and emit a warning that this is an experimental feature that should not be used in production yet. We also need infrastructure to track committed items that have not yet been written to the log. This is what the Committed Item List (CIL) is for. The log item also needs to be extended to track the current log vector, the associated memory buffer and it's location in the Commit Item List. Extend the log item and log vector structures to enable this tracking. To maintain the current log format for transactions with delayed logging, we need to introduce a checkpoint transaction and a context for tracking each checkpoint from initiation to transaction completion. This includes adding a log ticket for tracking space log required/used by the context checkpoint. To track all the changes we need an io vector array per log item, rather than a single array for the entire transaction. Using the new log vector structure for this requires two passes - the first to allocate the log vector structures and chain them together, and the second to fill them out. This log vector chain can then be passed to the CIL for formatting, pinning and insertion into the CIL. Formatting of the log vector chain is relatively simple - it's just a loop over the iovecs on each log vector, but it is made slightly more complex because we re-write the iovec after the copy to point back at the memory buffer we just copied into. This code also needs to pin log items. If the log item is not already tracked in this checkpoint context, then it needs to be pinned. Otherwise it is already pinned and we don't need to pin it again. The only other complexity is calculating the amount of new log space the formatting has consumed. This needs to be accounted to the transaction in progress, and the accounting is made more complex becase we need also to steal space from it for log metadata in the checkpoint transaction. Calculate all this at insert time and update all the tickets, counters, etc correctly. Once we've formatted all the log items in the transaction, attach the busy extents to the checkpoint context so the busy extents live until checkpoint completion and can be processed at that point in time. Transactions can then be freed at this point in time. Now we need to issue checkpoints - we are tracking the amount of log space used by the items in the CIL, so we can trigger background checkpoints when the space usage gets to a certain threshold. Otherwise, checkpoints need ot be triggered when a log synchronisation point is reached - a log force event. Because the log write code already handles chained log vectors, writing the transaction is trivial, too. Construct a transaction header, add it to the head of the chain and write it into the log, then issue a commit record write. Then we can release the checkpoint log ticket and attach the context to the log buffer so it can be called during Io completion to complete the checkpoint. We also need to allow for synchronising multiple in-flight checkpoints. This is needed for two things - the first is to ensure that checkpoint commit records appear in the log in the correct sequence order (so they are replayed in the correct order). The second is so that xfs_log_force_lsn() operates correctly and only flushes and/or waits for the specific sequence it was provided with. To do this we need a wait variable and a list tracking the checkpoint commits in progress. We can walk this list and wait for the checkpoints to change state or complete easily, an this provides the necessary synchronisation for correct operation in both cases. Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Alex Elder <aelder@sgi.com>
2010-05-21 12:37:18 +08:00
void
xfs_trans_unreserve_and_mod_sb(
struct xfs_trans *tp)
{
struct xfs_mount *mp = tp->t_mountp;
bool rsvd = (tp->t_flags & XFS_TRANS_RESERVE) != 0;
int64_t blkdelta = 0;
int64_t rtxdelta = 0;
int64_t idelta = 0;
int64_t ifreedelta = 0;
int error;
/* calculate deltas */
if (tp->t_blk_res > 0)
blkdelta = tp->t_blk_res;
if ((tp->t_fdblocks_delta != 0) &&
(xfs_has_lazysbcount(mp) ||
(tp->t_flags & XFS_TRANS_SB_DIRTY)))
blkdelta += tp->t_fdblocks_delta;
if (tp->t_rtx_res > 0)
rtxdelta = tp->t_rtx_res;
if ((tp->t_frextents_delta != 0) &&
(tp->t_flags & XFS_TRANS_SB_DIRTY))
rtxdelta += tp->t_frextents_delta;
if (xfs_has_lazysbcount(mp) ||
(tp->t_flags & XFS_TRANS_SB_DIRTY)) {
idelta = tp->t_icount_delta;
ifreedelta = tp->t_ifree_delta;
}
/* apply the per-cpu counters */
if (blkdelta) {
error = xfs_mod_fdblocks(mp, blkdelta, rsvd);
ASSERT(!error);
}
if (idelta)
xfs: reduce free inode accounting overhead Shaokun Zhang reported that XFS was using substantial CPU time in percpu_count_sum() when running a single threaded benchmark on a high CPU count (128p) machine from xfs_mod_ifree(). The issue is that the filesystem is empty when the benchmark runs, so inode allocation is running with a very low inode free count. With the percpu counter batching, this means comparisons when the counter is less that 128 * 256 = 32768 use the slow path of adding up all the counters across the CPUs, and this is expensive on high CPU count machines. The summing in xfs_mod_ifree() is only used to fire an assert if an underrun occurs. The error is ignored by the higher level code. Hence this is really just debug code and we don't need to run it on production kernels, nor do we need such debug checks to return error values just to trigger an assert. Finally, xfs_mod_icount/xfs_mod_ifree are only called from xfs_trans_unreserve_and_mod_sb(), so get rid of them and just directly call the percpu_counter_add/percpu_counter_compare functions. The compare functions are now run only on debug builds as they are internal to ASSERT() checks and so only compiled in when ASSERTs are active (CONFIG_XFS_DEBUG=y or CONFIG_XFS_WARN=y). Reported-by: Shaokun Zhang <zhangshaokun@hisilicon.com> Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Darrick J. Wong <darrick.wong@oracle.com> Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
2020-05-21 04:17:11 +08:00
percpu_counter_add_batch(&mp->m_icount, idelta,
XFS_ICOUNT_BATCH);
if (ifreedelta)
xfs: reduce free inode accounting overhead Shaokun Zhang reported that XFS was using substantial CPU time in percpu_count_sum() when running a single threaded benchmark on a high CPU count (128p) machine from xfs_mod_ifree(). The issue is that the filesystem is empty when the benchmark runs, so inode allocation is running with a very low inode free count. With the percpu counter batching, this means comparisons when the counter is less that 128 * 256 = 32768 use the slow path of adding up all the counters across the CPUs, and this is expensive on high CPU count machines. The summing in xfs_mod_ifree() is only used to fire an assert if an underrun occurs. The error is ignored by the higher level code. Hence this is really just debug code and we don't need to run it on production kernels, nor do we need such debug checks to return error values just to trigger an assert. Finally, xfs_mod_icount/xfs_mod_ifree are only called from xfs_trans_unreserve_and_mod_sb(), so get rid of them and just directly call the percpu_counter_add/percpu_counter_compare functions. The compare functions are now run only on debug builds as they are internal to ASSERT() checks and so only compiled in when ASSERTs are active (CONFIG_XFS_DEBUG=y or CONFIG_XFS_WARN=y). Reported-by: Shaokun Zhang <zhangshaokun@hisilicon.com> Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Darrick J. Wong <darrick.wong@oracle.com> Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
2020-05-21 04:17:11 +08:00
percpu_counter_add(&mp->m_ifree, ifreedelta);
if (rtxdelta) {
error = xfs_mod_frextents(mp, rtxdelta);
ASSERT(!error);
}
if (!(tp->t_flags & XFS_TRANS_SB_DIRTY))
return;
/* apply remaining deltas */
spin_lock(&mp->m_sb_lock);
mp->m_sb.sb_fdblocks += tp->t_fdblocks_delta + tp->t_res_fdblocks_delta;
mp->m_sb.sb_icount += idelta;
mp->m_sb.sb_ifree += ifreedelta;
/*
* Do not touch sb_frextents here because we are dealing with incore
* reservation. sb_frextents is not part of the lazy sb counters so it
* must be consistent with the ondisk rtbitmap and must never include
* incore reservations.
*/
mp->m_sb.sb_dblocks += tp->t_dblocks_delta;
mp->m_sb.sb_agcount += tp->t_agcount_delta;
mp->m_sb.sb_imax_pct += tp->t_imaxpct_delta;
mp->m_sb.sb_rextsize += tp->t_rextsize_delta;
mp->m_sb.sb_rbmblocks += tp->t_rbmblocks_delta;
mp->m_sb.sb_rblocks += tp->t_rblocks_delta;
mp->m_sb.sb_rextents += tp->t_rextents_delta;
mp->m_sb.sb_rextslog += tp->t_rextslog_delta;
spin_unlock(&mp->m_sb_lock);
/*
* Debug checks outside of the spinlock so they don't lock up the
* machine if they fail.
*/
ASSERT(mp->m_sb.sb_imax_pct >= 0);
ASSERT(mp->m_sb.sb_rextslog >= 0);
return;
}
/* Add the given log item to the transaction's list of log items. */
void
xfs_trans_add_item(
struct xfs_trans *tp,
struct xfs_log_item *lip)
{
ASSERT(lip->li_log == tp->t_mountp->m_log);
ASSERT(lip->li_ailp == tp->t_mountp->m_ail);
ASSERT(list_empty(&lip->li_trans));
ASSERT(!test_bit(XFS_LI_DIRTY, &lip->li_flags));
list_add_tail(&lip->li_trans, &tp->t_items);
trace_xfs_trans_add_item(tp, _RET_IP_);
}
/*
* Unlink the log item from the transaction. the log item is no longer
* considered dirty in this transaction, as the linked transaction has
* finished, either by abort or commit completion.
*/
void
xfs_trans_del_item(
struct xfs_log_item *lip)
{
clear_bit(XFS_LI_DIRTY, &lip->li_flags);
list_del_init(&lip->li_trans);
}
/* Detach and unlock all of the items in a transaction */
static void
xfs_trans_free_items(
struct xfs_trans *tp,
bool abort)
{
struct xfs_log_item *lip, *next;
trace_xfs_trans_free_items(tp, _RET_IP_);
list_for_each_entry_safe(lip, next, &tp->t_items, li_trans) {
xfs_trans_del_item(lip);
if (abort)
set_bit(XFS_LI_ABORTED, &lip->li_flags);
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
if (lip->li_ops->iop_release)
lip->li_ops->iop_release(lip);
}
}
xfs: bulk AIL insertion during transaction commit When inserting items into the AIL from the transaction committed callbacks, we take the AIL lock for every single item that is to be inserted. For a CIL checkpoint commit, this can be tens of thousands of individual inserts, yet almost all of the items will be inserted at the same point in the AIL because they have the same index. To reduce the overhead and contention on the AIL lock for such operations, introduce a "bulk insert" operation which allows a list of log items with the same LSN to be inserted in a single operation via a list splice. To do this, we need to pre-sort the log items being committed into a temporary list for insertion. The complexity is that not every log item will end up with the same LSN, and not every item is actually inserted into the AIL. Items that don't match the commit LSN will be inserted and unpinned as per the current one-at-a-time method (relatively rare), while items that are not to be inserted will be unpinned and freed immediately. Items that are to be inserted at the given commit lsn are placed in a temporary array and inserted into the AIL in bulk each time the array fills up. As a result of this, we trade off AIL hold time for a significant reduction in traffic. lock_stat output shows that the worst case hold time is unchanged, but contention from AIL inserts drops by an order of magnitude and the number of lock traversal decreases significantly. Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Christoph Hellwig <hch@lst.de>
2010-12-20 09:02:19 +08:00
static inline void
xfs_log_item_batch_insert(
struct xfs_ail *ailp,
struct xfs_ail_cursor *cur,
xfs: bulk AIL insertion during transaction commit When inserting items into the AIL from the transaction committed callbacks, we take the AIL lock for every single item that is to be inserted. For a CIL checkpoint commit, this can be tens of thousands of individual inserts, yet almost all of the items will be inserted at the same point in the AIL because they have the same index. To reduce the overhead and contention on the AIL lock for such operations, introduce a "bulk insert" operation which allows a list of log items with the same LSN to be inserted in a single operation via a list splice. To do this, we need to pre-sort the log items being committed into a temporary list for insertion. The complexity is that not every log item will end up with the same LSN, and not every item is actually inserted into the AIL. Items that don't match the commit LSN will be inserted and unpinned as per the current one-at-a-time method (relatively rare), while items that are not to be inserted will be unpinned and freed immediately. Items that are to be inserted at the given commit lsn are placed in a temporary array and inserted into the AIL in bulk each time the array fills up. As a result of this, we trade off AIL hold time for a significant reduction in traffic. lock_stat output shows that the worst case hold time is unchanged, but contention from AIL inserts drops by an order of magnitude and the number of lock traversal decreases significantly. Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Christoph Hellwig <hch@lst.de>
2010-12-20 09:02:19 +08:00
struct xfs_log_item **log_items,
int nr_items,
xfs_lsn_t commit_lsn)
{
int i;
spin_lock(&ailp->ail_lock);
/* xfs_trans_ail_update_bulk drops ailp->ail_lock */
xfs_trans_ail_update_bulk(ailp, cur, log_items, nr_items, commit_lsn);
xfs: bulk AIL insertion during transaction commit When inserting items into the AIL from the transaction committed callbacks, we take the AIL lock for every single item that is to be inserted. For a CIL checkpoint commit, this can be tens of thousands of individual inserts, yet almost all of the items will be inserted at the same point in the AIL because they have the same index. To reduce the overhead and contention on the AIL lock for such operations, introduce a "bulk insert" operation which allows a list of log items with the same LSN to be inserted in a single operation via a list splice. To do this, we need to pre-sort the log items being committed into a temporary list for insertion. The complexity is that not every log item will end up with the same LSN, and not every item is actually inserted into the AIL. Items that don't match the commit LSN will be inserted and unpinned as per the current one-at-a-time method (relatively rare), while items that are not to be inserted will be unpinned and freed immediately. Items that are to be inserted at the given commit lsn are placed in a temporary array and inserted into the AIL in bulk each time the array fills up. As a result of this, we trade off AIL hold time for a significant reduction in traffic. lock_stat output shows that the worst case hold time is unchanged, but contention from AIL inserts drops by an order of magnitude and the number of lock traversal decreases significantly. Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Christoph Hellwig <hch@lst.de>
2010-12-20 09:02:19 +08:00
for (i = 0; i < nr_items; i++) {
struct xfs_log_item *lip = log_items[i];
if (lip->li_ops->iop_unpin)
lip->li_ops->iop_unpin(lip, 0);
}
xfs: bulk AIL insertion during transaction commit When inserting items into the AIL from the transaction committed callbacks, we take the AIL lock for every single item that is to be inserted. For a CIL checkpoint commit, this can be tens of thousands of individual inserts, yet almost all of the items will be inserted at the same point in the AIL because they have the same index. To reduce the overhead and contention on the AIL lock for such operations, introduce a "bulk insert" operation which allows a list of log items with the same LSN to be inserted in a single operation via a list splice. To do this, we need to pre-sort the log items being committed into a temporary list for insertion. The complexity is that not every log item will end up with the same LSN, and not every item is actually inserted into the AIL. Items that don't match the commit LSN will be inserted and unpinned as per the current one-at-a-time method (relatively rare), while items that are not to be inserted will be unpinned and freed immediately. Items that are to be inserted at the given commit lsn are placed in a temporary array and inserted into the AIL in bulk each time the array fills up. As a result of this, we trade off AIL hold time for a significant reduction in traffic. lock_stat output shows that the worst case hold time is unchanged, but contention from AIL inserts drops by an order of magnitude and the number of lock traversal decreases significantly. Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Christoph Hellwig <hch@lst.de>
2010-12-20 09:02:19 +08:00
}
/*
* Bulk operation version of xfs_trans_committed that takes a log vector of
* items to insert into the AIL. This uses bulk AIL insertion techniques to
* minimise lock traffic.
xfs: fix efi item leak on forced shutdown After test 139, kmemleak shows: unreferenced object 0xffff880078b405d8 (size 400): comm "xfs_io", pid 4904, jiffies 4294909383 (age 1186.728s) hex dump (first 32 bytes): 60 c1 17 79 00 88 ff ff 60 c1 17 79 00 88 ff ff `..y....`..y.... 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ................ backtrace: [<ffffffff81afb04d>] kmemleak_alloc+0x2d/0x60 [<ffffffff8115c6cf>] kmem_cache_alloc+0x13f/0x2b0 [<ffffffff814aaa97>] kmem_zone_alloc+0x77/0xf0 [<ffffffff814aab2e>] kmem_zone_zalloc+0x1e/0x50 [<ffffffff8147cd6b>] xfs_efi_init+0x4b/0xb0 [<ffffffff814a4ee8>] xfs_trans_get_efi+0x58/0x90 [<ffffffff81455fab>] xfs_bmap_finish+0x8b/0x1d0 [<ffffffff814851b4>] xfs_itruncate_finish+0x2c4/0x5d0 [<ffffffff814a970f>] xfs_setattr+0x8df/0xa70 [<ffffffff814b5c7b>] xfs_vn_setattr+0x1b/0x20 [<ffffffff8117dc00>] notify_change+0x170/0x2e0 [<ffffffff81163bf6>] do_truncate+0x66/0xa0 [<ffffffff81163d0b>] sys_ftruncate+0xdb/0xe0 [<ffffffff8103a002>] system_call_fastpath+0x16/0x1b [<ffffffffffffffff>] 0xffffffffffffffff The cause of the leak is that the "remove" parameter of IOP_UNPIN() is never set when a CIL push is aborted. This means that the EFI item is never freed if it was in the push being cancelled. The problem is specific to delayed logging, but has uncovered a couple of problems with the handling of IOP_UNPIN(remove). Firstly, we cannot safely call xfs_trans_del_item() from IOP_UNPIN() in the CIL commit failure path or the iclog write failure path because for delayed loging we have no transaction context. Hence we must only call xfs_trans_del_item() if the log item being unpinned has an active log item descriptor. Secondly, xfs_trans_uncommit() does not handle log item descriptor freeing during the traversal of log items on a transaction. It can reference a freed log item descriptor when unpinning an EFI item. Hence it needs to use a safe list traversal method to allow items to be removed from the transaction during IOP_UNPIN(). Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Alex Elder <aelder@sgi.com>
2011-01-27 09:13:35 +08:00
*
* If we are called with the aborted flag set, it is because a log write during
* a CIL checkpoint commit has failed. In this case, all the items in the
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
* checkpoint have already gone through iop_committed and iop_committing, which
xfs: fix efi item leak on forced shutdown After test 139, kmemleak shows: unreferenced object 0xffff880078b405d8 (size 400): comm "xfs_io", pid 4904, jiffies 4294909383 (age 1186.728s) hex dump (first 32 bytes): 60 c1 17 79 00 88 ff ff 60 c1 17 79 00 88 ff ff `..y....`..y.... 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ................ backtrace: [<ffffffff81afb04d>] kmemleak_alloc+0x2d/0x60 [<ffffffff8115c6cf>] kmem_cache_alloc+0x13f/0x2b0 [<ffffffff814aaa97>] kmem_zone_alloc+0x77/0xf0 [<ffffffff814aab2e>] kmem_zone_zalloc+0x1e/0x50 [<ffffffff8147cd6b>] xfs_efi_init+0x4b/0xb0 [<ffffffff814a4ee8>] xfs_trans_get_efi+0x58/0x90 [<ffffffff81455fab>] xfs_bmap_finish+0x8b/0x1d0 [<ffffffff814851b4>] xfs_itruncate_finish+0x2c4/0x5d0 [<ffffffff814a970f>] xfs_setattr+0x8df/0xa70 [<ffffffff814b5c7b>] xfs_vn_setattr+0x1b/0x20 [<ffffffff8117dc00>] notify_change+0x170/0x2e0 [<ffffffff81163bf6>] do_truncate+0x66/0xa0 [<ffffffff81163d0b>] sys_ftruncate+0xdb/0xe0 [<ffffffff8103a002>] system_call_fastpath+0x16/0x1b [<ffffffffffffffff>] 0xffffffffffffffff The cause of the leak is that the "remove" parameter of IOP_UNPIN() is never set when a CIL push is aborted. This means that the EFI item is never freed if it was in the push being cancelled. The problem is specific to delayed logging, but has uncovered a couple of problems with the handling of IOP_UNPIN(remove). Firstly, we cannot safely call xfs_trans_del_item() from IOP_UNPIN() in the CIL commit failure path or the iclog write failure path because for delayed loging we have no transaction context. Hence we must only call xfs_trans_del_item() if the log item being unpinned has an active log item descriptor. Secondly, xfs_trans_uncommit() does not handle log item descriptor freeing during the traversal of log items on a transaction. It can reference a freed log item descriptor when unpinning an EFI item. Hence it needs to use a safe list traversal method to allow items to be removed from the transaction during IOP_UNPIN(). Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Alex Elder <aelder@sgi.com>
2011-01-27 09:13:35 +08:00
* means that checkpoint commit abort handling is treated exactly the same
* as an iclog write error even though we haven't started any IO yet. Hence in
* this case all we need to do is iop_committed processing, followed by an
* iop_unpin(aborted) call.
*
* The AIL cursor is used to optimise the insert process. If commit_lsn is not
* at the end of the AIL, the insert cursor avoids the need to walk
* the AIL to find the insertion point on every xfs_log_item_batch_insert()
* call. This saves a lot of needless list walking and is a net win, even
* though it slightly increases that amount of AIL lock traffic to set it up
* and tear it down.
xfs: bulk AIL insertion during transaction commit When inserting items into the AIL from the transaction committed callbacks, we take the AIL lock for every single item that is to be inserted. For a CIL checkpoint commit, this can be tens of thousands of individual inserts, yet almost all of the items will be inserted at the same point in the AIL because they have the same index. To reduce the overhead and contention on the AIL lock for such operations, introduce a "bulk insert" operation which allows a list of log items with the same LSN to be inserted in a single operation via a list splice. To do this, we need to pre-sort the log items being committed into a temporary list for insertion. The complexity is that not every log item will end up with the same LSN, and not every item is actually inserted into the AIL. Items that don't match the commit LSN will be inserted and unpinned as per the current one-at-a-time method (relatively rare), while items that are not to be inserted will be unpinned and freed immediately. Items that are to be inserted at the given commit lsn are placed in a temporary array and inserted into the AIL in bulk each time the array fills up. As a result of this, we trade off AIL hold time for a significant reduction in traffic. lock_stat output shows that the worst case hold time is unchanged, but contention from AIL inserts drops by an order of magnitude and the number of lock traversal decreases significantly. Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Christoph Hellwig <hch@lst.de>
2010-12-20 09:02:19 +08:00
*/
void
xfs_trans_committed_bulk(
struct xfs_ail *ailp,
struct list_head *lv_chain,
xfs: bulk AIL insertion during transaction commit When inserting items into the AIL from the transaction committed callbacks, we take the AIL lock for every single item that is to be inserted. For a CIL checkpoint commit, this can be tens of thousands of individual inserts, yet almost all of the items will be inserted at the same point in the AIL because they have the same index. To reduce the overhead and contention on the AIL lock for such operations, introduce a "bulk insert" operation which allows a list of log items with the same LSN to be inserted in a single operation via a list splice. To do this, we need to pre-sort the log items being committed into a temporary list for insertion. The complexity is that not every log item will end up with the same LSN, and not every item is actually inserted into the AIL. Items that don't match the commit LSN will be inserted and unpinned as per the current one-at-a-time method (relatively rare), while items that are not to be inserted will be unpinned and freed immediately. Items that are to be inserted at the given commit lsn are placed in a temporary array and inserted into the AIL in bulk each time the array fills up. As a result of this, we trade off AIL hold time for a significant reduction in traffic. lock_stat output shows that the worst case hold time is unchanged, but contention from AIL inserts drops by an order of magnitude and the number of lock traversal decreases significantly. Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Christoph Hellwig <hch@lst.de>
2010-12-20 09:02:19 +08:00
xfs_lsn_t commit_lsn,
bool aborted)
xfs: bulk AIL insertion during transaction commit When inserting items into the AIL from the transaction committed callbacks, we take the AIL lock for every single item that is to be inserted. For a CIL checkpoint commit, this can be tens of thousands of individual inserts, yet almost all of the items will be inserted at the same point in the AIL because they have the same index. To reduce the overhead and contention on the AIL lock for such operations, introduce a "bulk insert" operation which allows a list of log items with the same LSN to be inserted in a single operation via a list splice. To do this, we need to pre-sort the log items being committed into a temporary list for insertion. The complexity is that not every log item will end up with the same LSN, and not every item is actually inserted into the AIL. Items that don't match the commit LSN will be inserted and unpinned as per the current one-at-a-time method (relatively rare), while items that are not to be inserted will be unpinned and freed immediately. Items that are to be inserted at the given commit lsn are placed in a temporary array and inserted into the AIL in bulk each time the array fills up. As a result of this, we trade off AIL hold time for a significant reduction in traffic. lock_stat output shows that the worst case hold time is unchanged, but contention from AIL inserts drops by an order of magnitude and the number of lock traversal decreases significantly. Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Christoph Hellwig <hch@lst.de>
2010-12-20 09:02:19 +08:00
{
#define LOG_ITEM_BATCH_SIZE 32
struct xfs_log_item *log_items[LOG_ITEM_BATCH_SIZE];
struct xfs_log_vec *lv;
struct xfs_ail_cursor cur;
xfs: bulk AIL insertion during transaction commit When inserting items into the AIL from the transaction committed callbacks, we take the AIL lock for every single item that is to be inserted. For a CIL checkpoint commit, this can be tens of thousands of individual inserts, yet almost all of the items will be inserted at the same point in the AIL because they have the same index. To reduce the overhead and contention on the AIL lock for such operations, introduce a "bulk insert" operation which allows a list of log items with the same LSN to be inserted in a single operation via a list splice. To do this, we need to pre-sort the log items being committed into a temporary list for insertion. The complexity is that not every log item will end up with the same LSN, and not every item is actually inserted into the AIL. Items that don't match the commit LSN will be inserted and unpinned as per the current one-at-a-time method (relatively rare), while items that are not to be inserted will be unpinned and freed immediately. Items that are to be inserted at the given commit lsn are placed in a temporary array and inserted into the AIL in bulk each time the array fills up. As a result of this, we trade off AIL hold time for a significant reduction in traffic. lock_stat output shows that the worst case hold time is unchanged, but contention from AIL inserts drops by an order of magnitude and the number of lock traversal decreases significantly. Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Christoph Hellwig <hch@lst.de>
2010-12-20 09:02:19 +08:00
int i = 0;
spin_lock(&ailp->ail_lock);
xfs_trans_ail_cursor_last(ailp, &cur, commit_lsn);
spin_unlock(&ailp->ail_lock);
xfs: bulk AIL insertion during transaction commit When inserting items into the AIL from the transaction committed callbacks, we take the AIL lock for every single item that is to be inserted. For a CIL checkpoint commit, this can be tens of thousands of individual inserts, yet almost all of the items will be inserted at the same point in the AIL because they have the same index. To reduce the overhead and contention on the AIL lock for such operations, introduce a "bulk insert" operation which allows a list of log items with the same LSN to be inserted in a single operation via a list splice. To do this, we need to pre-sort the log items being committed into a temporary list for insertion. The complexity is that not every log item will end up with the same LSN, and not every item is actually inserted into the AIL. Items that don't match the commit LSN will be inserted and unpinned as per the current one-at-a-time method (relatively rare), while items that are not to be inserted will be unpinned and freed immediately. Items that are to be inserted at the given commit lsn are placed in a temporary array and inserted into the AIL in bulk each time the array fills up. As a result of this, we trade off AIL hold time for a significant reduction in traffic. lock_stat output shows that the worst case hold time is unchanged, but contention from AIL inserts drops by an order of magnitude and the number of lock traversal decreases significantly. Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Christoph Hellwig <hch@lst.de>
2010-12-20 09:02:19 +08:00
/* unpin all the log items */
list_for_each_entry(lv, lv_chain, lv_list) {
xfs: bulk AIL insertion during transaction commit When inserting items into the AIL from the transaction committed callbacks, we take the AIL lock for every single item that is to be inserted. For a CIL checkpoint commit, this can be tens of thousands of individual inserts, yet almost all of the items will be inserted at the same point in the AIL because they have the same index. To reduce the overhead and contention on the AIL lock for such operations, introduce a "bulk insert" operation which allows a list of log items with the same LSN to be inserted in a single operation via a list splice. To do this, we need to pre-sort the log items being committed into a temporary list for insertion. The complexity is that not every log item will end up with the same LSN, and not every item is actually inserted into the AIL. Items that don't match the commit LSN will be inserted and unpinned as per the current one-at-a-time method (relatively rare), while items that are not to be inserted will be unpinned and freed immediately. Items that are to be inserted at the given commit lsn are placed in a temporary array and inserted into the AIL in bulk each time the array fills up. As a result of this, we trade off AIL hold time for a significant reduction in traffic. lock_stat output shows that the worst case hold time is unchanged, but contention from AIL inserts drops by an order of magnitude and the number of lock traversal decreases significantly. Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Christoph Hellwig <hch@lst.de>
2010-12-20 09:02:19 +08:00
struct xfs_log_item *lip = lv->lv_item;
xfs_lsn_t item_lsn;
if (aborted)
set_bit(XFS_LI_ABORTED, &lip->li_flags);
if (lip->li_ops->flags & XFS_ITEM_RELEASE_WHEN_COMMITTED) {
lip->li_ops->iop_release(lip);
continue;
}
if (lip->li_ops->iop_committed)
item_lsn = lip->li_ops->iop_committed(lip, commit_lsn);
else
item_lsn = commit_lsn;
xfs: bulk AIL insertion during transaction commit When inserting items into the AIL from the transaction committed callbacks, we take the AIL lock for every single item that is to be inserted. For a CIL checkpoint commit, this can be tens of thousands of individual inserts, yet almost all of the items will be inserted at the same point in the AIL because they have the same index. To reduce the overhead and contention on the AIL lock for such operations, introduce a "bulk insert" operation which allows a list of log items with the same LSN to be inserted in a single operation via a list splice. To do this, we need to pre-sort the log items being committed into a temporary list for insertion. The complexity is that not every log item will end up with the same LSN, and not every item is actually inserted into the AIL. Items that don't match the commit LSN will be inserted and unpinned as per the current one-at-a-time method (relatively rare), while items that are not to be inserted will be unpinned and freed immediately. Items that are to be inserted at the given commit lsn are placed in a temporary array and inserted into the AIL in bulk each time the array fills up. As a result of this, we trade off AIL hold time for a significant reduction in traffic. lock_stat output shows that the worst case hold time is unchanged, but contention from AIL inserts drops by an order of magnitude and the number of lock traversal decreases significantly. Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Christoph Hellwig <hch@lst.de>
2010-12-20 09:02:19 +08:00
xfs: unpin stale inodes directly in IOP_COMMITTED When inodes are marked stale in a transaction, they are treated specially when the inode log item is being inserted into the AIL. It tries to avoid moving the log item forward in the AIL due to a race condition with the writing the underlying buffer back to disk. The was "fixed" in commit de25c18 ("xfs: avoid moving stale inodes in the AIL"). To avoid moving the item forward, we return a LSN smaller than the commit_lsn of the completing transaction, thereby trying to trick the commit code into not moving the inode forward at all. I'm not sure this ever worked as intended - it assumes the inode is already in the AIL, but I don't think the returned LSN would have been small enough to prevent moving the inode. It appears that the reason it worked is that the lower LSN of the inodes meant they were inserted into the AIL and flushed before the inode buffer (which was moved to the commit_lsn of the transaction). The big problem is that with delayed logging, the returning of the different LSN means insertion takes the slow, non-bulk path. Worse yet is that insertion is to a position -before- the commit_lsn so it is doing a AIL traversal on every insertion, and has to walk over all the items that have already been inserted into the AIL. It's expensive. To compound the matter further, with delayed logging inodes are likely to go from clean to stale in a single checkpoint, which means they aren't even in the AIL at all when we come across them at AIL insertion time. Hence these were all getting inserted into the AIL when they simply do not need to be as inodes marked XFS_ISTALE are never written back. Transactional/recovery integrity is maintained in this case by the other items in the unlink transaction that were modified (e.g. the AGI btree blocks) and committed in the same checkpoint. So to fix this, simply unpin the stale inodes directly in xfs_inode_item_committed() and return -1 to indicate that the AIL insertion code does not need to do any further processing of these inodes. Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Alex Elder <aelder@sgi.com>
2011-07-04 13:27:36 +08:00
/* item_lsn of -1 means the item needs no further processing */
xfs: bulk AIL insertion during transaction commit When inserting items into the AIL from the transaction committed callbacks, we take the AIL lock for every single item that is to be inserted. For a CIL checkpoint commit, this can be tens of thousands of individual inserts, yet almost all of the items will be inserted at the same point in the AIL because they have the same index. To reduce the overhead and contention on the AIL lock for such operations, introduce a "bulk insert" operation which allows a list of log items with the same LSN to be inserted in a single operation via a list splice. To do this, we need to pre-sort the log items being committed into a temporary list for insertion. The complexity is that not every log item will end up with the same LSN, and not every item is actually inserted into the AIL. Items that don't match the commit LSN will be inserted and unpinned as per the current one-at-a-time method (relatively rare), while items that are not to be inserted will be unpinned and freed immediately. Items that are to be inserted at the given commit lsn are placed in a temporary array and inserted into the AIL in bulk each time the array fills up. As a result of this, we trade off AIL hold time for a significant reduction in traffic. lock_stat output shows that the worst case hold time is unchanged, but contention from AIL inserts drops by an order of magnitude and the number of lock traversal decreases significantly. Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Christoph Hellwig <hch@lst.de>
2010-12-20 09:02:19 +08:00
if (XFS_LSN_CMP(item_lsn, (xfs_lsn_t)-1) == 0)
continue;
xfs: fix efi item leak on forced shutdown After test 139, kmemleak shows: unreferenced object 0xffff880078b405d8 (size 400): comm "xfs_io", pid 4904, jiffies 4294909383 (age 1186.728s) hex dump (first 32 bytes): 60 c1 17 79 00 88 ff ff 60 c1 17 79 00 88 ff ff `..y....`..y.... 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ................ backtrace: [<ffffffff81afb04d>] kmemleak_alloc+0x2d/0x60 [<ffffffff8115c6cf>] kmem_cache_alloc+0x13f/0x2b0 [<ffffffff814aaa97>] kmem_zone_alloc+0x77/0xf0 [<ffffffff814aab2e>] kmem_zone_zalloc+0x1e/0x50 [<ffffffff8147cd6b>] xfs_efi_init+0x4b/0xb0 [<ffffffff814a4ee8>] xfs_trans_get_efi+0x58/0x90 [<ffffffff81455fab>] xfs_bmap_finish+0x8b/0x1d0 [<ffffffff814851b4>] xfs_itruncate_finish+0x2c4/0x5d0 [<ffffffff814a970f>] xfs_setattr+0x8df/0xa70 [<ffffffff814b5c7b>] xfs_vn_setattr+0x1b/0x20 [<ffffffff8117dc00>] notify_change+0x170/0x2e0 [<ffffffff81163bf6>] do_truncate+0x66/0xa0 [<ffffffff81163d0b>] sys_ftruncate+0xdb/0xe0 [<ffffffff8103a002>] system_call_fastpath+0x16/0x1b [<ffffffffffffffff>] 0xffffffffffffffff The cause of the leak is that the "remove" parameter of IOP_UNPIN() is never set when a CIL push is aborted. This means that the EFI item is never freed if it was in the push being cancelled. The problem is specific to delayed logging, but has uncovered a couple of problems with the handling of IOP_UNPIN(remove). Firstly, we cannot safely call xfs_trans_del_item() from IOP_UNPIN() in the CIL commit failure path or the iclog write failure path because for delayed loging we have no transaction context. Hence we must only call xfs_trans_del_item() if the log item being unpinned has an active log item descriptor. Secondly, xfs_trans_uncommit() does not handle log item descriptor freeing during the traversal of log items on a transaction. It can reference a freed log item descriptor when unpinning an EFI item. Hence it needs to use a safe list traversal method to allow items to be removed from the transaction during IOP_UNPIN(). Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Alex Elder <aelder@sgi.com>
2011-01-27 09:13:35 +08:00
/*
* if we are aborting the operation, no point in inserting the
* object into the AIL as we are in a shutdown situation.
*/
if (aborted) {
ASSERT(xlog_is_shutdown(ailp->ail_log));
if (lip->li_ops->iop_unpin)
lip->li_ops->iop_unpin(lip, 1);
xfs: fix efi item leak on forced shutdown After test 139, kmemleak shows: unreferenced object 0xffff880078b405d8 (size 400): comm "xfs_io", pid 4904, jiffies 4294909383 (age 1186.728s) hex dump (first 32 bytes): 60 c1 17 79 00 88 ff ff 60 c1 17 79 00 88 ff ff `..y....`..y.... 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ................ backtrace: [<ffffffff81afb04d>] kmemleak_alloc+0x2d/0x60 [<ffffffff8115c6cf>] kmem_cache_alloc+0x13f/0x2b0 [<ffffffff814aaa97>] kmem_zone_alloc+0x77/0xf0 [<ffffffff814aab2e>] kmem_zone_zalloc+0x1e/0x50 [<ffffffff8147cd6b>] xfs_efi_init+0x4b/0xb0 [<ffffffff814a4ee8>] xfs_trans_get_efi+0x58/0x90 [<ffffffff81455fab>] xfs_bmap_finish+0x8b/0x1d0 [<ffffffff814851b4>] xfs_itruncate_finish+0x2c4/0x5d0 [<ffffffff814a970f>] xfs_setattr+0x8df/0xa70 [<ffffffff814b5c7b>] xfs_vn_setattr+0x1b/0x20 [<ffffffff8117dc00>] notify_change+0x170/0x2e0 [<ffffffff81163bf6>] do_truncate+0x66/0xa0 [<ffffffff81163d0b>] sys_ftruncate+0xdb/0xe0 [<ffffffff8103a002>] system_call_fastpath+0x16/0x1b [<ffffffffffffffff>] 0xffffffffffffffff The cause of the leak is that the "remove" parameter of IOP_UNPIN() is never set when a CIL push is aborted. This means that the EFI item is never freed if it was in the push being cancelled. The problem is specific to delayed logging, but has uncovered a couple of problems with the handling of IOP_UNPIN(remove). Firstly, we cannot safely call xfs_trans_del_item() from IOP_UNPIN() in the CIL commit failure path or the iclog write failure path because for delayed loging we have no transaction context. Hence we must only call xfs_trans_del_item() if the log item being unpinned has an active log item descriptor. Secondly, xfs_trans_uncommit() does not handle log item descriptor freeing during the traversal of log items on a transaction. It can reference a freed log item descriptor when unpinning an EFI item. Hence it needs to use a safe list traversal method to allow items to be removed from the transaction during IOP_UNPIN(). Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Alex Elder <aelder@sgi.com>
2011-01-27 09:13:35 +08:00
continue;
}
xfs: bulk AIL insertion during transaction commit When inserting items into the AIL from the transaction committed callbacks, we take the AIL lock for every single item that is to be inserted. For a CIL checkpoint commit, this can be tens of thousands of individual inserts, yet almost all of the items will be inserted at the same point in the AIL because they have the same index. To reduce the overhead and contention on the AIL lock for such operations, introduce a "bulk insert" operation which allows a list of log items with the same LSN to be inserted in a single operation via a list splice. To do this, we need to pre-sort the log items being committed into a temporary list for insertion. The complexity is that not every log item will end up with the same LSN, and not every item is actually inserted into the AIL. Items that don't match the commit LSN will be inserted and unpinned as per the current one-at-a-time method (relatively rare), while items that are not to be inserted will be unpinned and freed immediately. Items that are to be inserted at the given commit lsn are placed in a temporary array and inserted into the AIL in bulk each time the array fills up. As a result of this, we trade off AIL hold time for a significant reduction in traffic. lock_stat output shows that the worst case hold time is unchanged, but contention from AIL inserts drops by an order of magnitude and the number of lock traversal decreases significantly. Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Christoph Hellwig <hch@lst.de>
2010-12-20 09:02:19 +08:00
if (item_lsn != commit_lsn) {
/*
* Not a bulk update option due to unusual item_lsn.
* Push into AIL immediately, rechecking the lsn once
* we have the ail lock. Then unpin the item. This does
* not affect the AIL cursor the bulk insert path is
* using.
xfs: bulk AIL insertion during transaction commit When inserting items into the AIL from the transaction committed callbacks, we take the AIL lock for every single item that is to be inserted. For a CIL checkpoint commit, this can be tens of thousands of individual inserts, yet almost all of the items will be inserted at the same point in the AIL because they have the same index. To reduce the overhead and contention on the AIL lock for such operations, introduce a "bulk insert" operation which allows a list of log items with the same LSN to be inserted in a single operation via a list splice. To do this, we need to pre-sort the log items being committed into a temporary list for insertion. The complexity is that not every log item will end up with the same LSN, and not every item is actually inserted into the AIL. Items that don't match the commit LSN will be inserted and unpinned as per the current one-at-a-time method (relatively rare), while items that are not to be inserted will be unpinned and freed immediately. Items that are to be inserted at the given commit lsn are placed in a temporary array and inserted into the AIL in bulk each time the array fills up. As a result of this, we trade off AIL hold time for a significant reduction in traffic. lock_stat output shows that the worst case hold time is unchanged, but contention from AIL inserts drops by an order of magnitude and the number of lock traversal decreases significantly. Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Christoph Hellwig <hch@lst.de>
2010-12-20 09:02:19 +08:00
*/
spin_lock(&ailp->ail_lock);
xfs: bulk AIL insertion during transaction commit When inserting items into the AIL from the transaction committed callbacks, we take the AIL lock for every single item that is to be inserted. For a CIL checkpoint commit, this can be tens of thousands of individual inserts, yet almost all of the items will be inserted at the same point in the AIL because they have the same index. To reduce the overhead and contention on the AIL lock for such operations, introduce a "bulk insert" operation which allows a list of log items with the same LSN to be inserted in a single operation via a list splice. To do this, we need to pre-sort the log items being committed into a temporary list for insertion. The complexity is that not every log item will end up with the same LSN, and not every item is actually inserted into the AIL. Items that don't match the commit LSN will be inserted and unpinned as per the current one-at-a-time method (relatively rare), while items that are not to be inserted will be unpinned and freed immediately. Items that are to be inserted at the given commit lsn are placed in a temporary array and inserted into the AIL in bulk each time the array fills up. As a result of this, we trade off AIL hold time for a significant reduction in traffic. lock_stat output shows that the worst case hold time is unchanged, but contention from AIL inserts drops by an order of magnitude and the number of lock traversal decreases significantly. Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Christoph Hellwig <hch@lst.de>
2010-12-20 09:02:19 +08:00
if (XFS_LSN_CMP(item_lsn, lip->li_lsn) > 0)
xfs_trans_ail_update(ailp, lip, item_lsn);
else
spin_unlock(&ailp->ail_lock);
if (lip->li_ops->iop_unpin)
lip->li_ops->iop_unpin(lip, 0);
xfs: bulk AIL insertion during transaction commit When inserting items into the AIL from the transaction committed callbacks, we take the AIL lock for every single item that is to be inserted. For a CIL checkpoint commit, this can be tens of thousands of individual inserts, yet almost all of the items will be inserted at the same point in the AIL because they have the same index. To reduce the overhead and contention on the AIL lock for such operations, introduce a "bulk insert" operation which allows a list of log items with the same LSN to be inserted in a single operation via a list splice. To do this, we need to pre-sort the log items being committed into a temporary list for insertion. The complexity is that not every log item will end up with the same LSN, and not every item is actually inserted into the AIL. Items that don't match the commit LSN will be inserted and unpinned as per the current one-at-a-time method (relatively rare), while items that are not to be inserted will be unpinned and freed immediately. Items that are to be inserted at the given commit lsn are placed in a temporary array and inserted into the AIL in bulk each time the array fills up. As a result of this, we trade off AIL hold time for a significant reduction in traffic. lock_stat output shows that the worst case hold time is unchanged, but contention from AIL inserts drops by an order of magnitude and the number of lock traversal decreases significantly. Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Christoph Hellwig <hch@lst.de>
2010-12-20 09:02:19 +08:00
continue;
}
/* Item is a candidate for bulk AIL insert. */
log_items[i++] = lv->lv_item;
if (i >= LOG_ITEM_BATCH_SIZE) {
xfs_log_item_batch_insert(ailp, &cur, log_items,
xfs: bulk AIL insertion during transaction commit When inserting items into the AIL from the transaction committed callbacks, we take the AIL lock for every single item that is to be inserted. For a CIL checkpoint commit, this can be tens of thousands of individual inserts, yet almost all of the items will be inserted at the same point in the AIL because they have the same index. To reduce the overhead and contention on the AIL lock for such operations, introduce a "bulk insert" operation which allows a list of log items with the same LSN to be inserted in a single operation via a list splice. To do this, we need to pre-sort the log items being committed into a temporary list for insertion. The complexity is that not every log item will end up with the same LSN, and not every item is actually inserted into the AIL. Items that don't match the commit LSN will be inserted and unpinned as per the current one-at-a-time method (relatively rare), while items that are not to be inserted will be unpinned and freed immediately. Items that are to be inserted at the given commit lsn are placed in a temporary array and inserted into the AIL in bulk each time the array fills up. As a result of this, we trade off AIL hold time for a significant reduction in traffic. lock_stat output shows that the worst case hold time is unchanged, but contention from AIL inserts drops by an order of magnitude and the number of lock traversal decreases significantly. Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Christoph Hellwig <hch@lst.de>
2010-12-20 09:02:19 +08:00
LOG_ITEM_BATCH_SIZE, commit_lsn);
i = 0;
}
}
/* make sure we insert the remainder! */
if (i)
xfs_log_item_batch_insert(ailp, &cur, log_items, i, commit_lsn);
spin_lock(&ailp->ail_lock);
xfs_trans_ail_cursor_done(&cur);
spin_unlock(&ailp->ail_lock);
xfs: bulk AIL insertion during transaction commit When inserting items into the AIL from the transaction committed callbacks, we take the AIL lock for every single item that is to be inserted. For a CIL checkpoint commit, this can be tens of thousands of individual inserts, yet almost all of the items will be inserted at the same point in the AIL because they have the same index. To reduce the overhead and contention on the AIL lock for such operations, introduce a "bulk insert" operation which allows a list of log items with the same LSN to be inserted in a single operation via a list splice. To do this, we need to pre-sort the log items being committed into a temporary list for insertion. The complexity is that not every log item will end up with the same LSN, and not every item is actually inserted into the AIL. Items that don't match the commit LSN will be inserted and unpinned as per the current one-at-a-time method (relatively rare), while items that are not to be inserted will be unpinned and freed immediately. Items that are to be inserted at the given commit lsn are placed in a temporary array and inserted into the AIL in bulk each time the array fills up. As a result of this, we trade off AIL hold time for a significant reduction in traffic. lock_stat output shows that the worst case hold time is unchanged, but contention from AIL inserts drops by an order of magnitude and the number of lock traversal decreases significantly. Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Christoph Hellwig <hch@lst.de>
2010-12-20 09:02:19 +08:00
}
xfs: add log item precommit operation For inodes that are dirty, we have an attached cluster buffer that we want to use to track the dirty inode through the AIL. Unfortunately, locking the cluster buffer and adding it to the transaction when the inode is first logged in a transaction leads to buffer lock ordering inversions. The specific problem is ordering against the AGI buffer. When modifying unlinked lists, the buffer lock order is AGI -> inode cluster buffer as the AGI buffer lock serialises all access to the unlinked lists. Unfortunately, functionality like xfs_droplink() logs the inode before calling xfs_iunlink(), as do various directory manipulation functions. The inode can be logged way down in the stack as far as the bmapi routines and hence, without a major rewrite of lots of APIs there's no way we can avoid the inode being logged by something until after the AGI has been logged. As we are going to be using ordered buffers for inode AIL tracking, there isn't a need to actually lock that buffer against modification as all the modifications are captured by logging the inode item itself. Hence we don't actually need to join the cluster buffer into the transaction until just before it is committed. This means we do not perturb any of the existing buffer lock orders in transactions, and the inode cluster buffer is always locked last in a transaction that doesn't otherwise touch inode cluster buffers. We do this by introducing a precommit log item method. This commit just introduces the mechanism; the inode item implementation is in followup commits. The precommit items need to be sorted into consistent order as we may be locking multiple items here. Hence if we have two dirty inodes in cluster buffers A and B, and some other transaction has two separate dirty inodes in the same cluster buffers, locking them in different orders opens us up to ABBA deadlocks. Hence we sort the items on the transaction based on the presence of a sort log item method. Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Darrick J. Wong <djwong@kernel.org> Reviewed-by: Christoph Hellwig <hch@lst.de>
2022-07-14 09:47:26 +08:00
/*
* Sort transaction items prior to running precommit operations. This will
* attempt to order the items such that they will always be locked in the same
* order. Items that have no sort function are moved to the end of the list
* and so are locked last.
*
* This may need refinement as different types of objects add sort functions.
*
* Function is more complex than it needs to be because we are comparing 64 bit
* values and the function only returns 32 bit values.
*/
static int
xfs_trans_precommit_sort(
void *unused_arg,
const struct list_head *a,
const struct list_head *b)
{
struct xfs_log_item *lia = container_of(a,
struct xfs_log_item, li_trans);
struct xfs_log_item *lib = container_of(b,
struct xfs_log_item, li_trans);
int64_t diff;
/*
* If both items are non-sortable, leave them alone. If only one is
* sortable, move the non-sortable item towards the end of the list.
*/
if (!lia->li_ops->iop_sort && !lib->li_ops->iop_sort)
return 0;
if (!lia->li_ops->iop_sort)
return 1;
if (!lib->li_ops->iop_sort)
return -1;
diff = lia->li_ops->iop_sort(lia) - lib->li_ops->iop_sort(lib);
if (diff < 0)
return -1;
if (diff > 0)
return 1;
return 0;
}
/*
* Run transaction precommit functions.
*
* If there is an error in any of the callouts, then stop immediately and
* trigger a shutdown to abort the transaction. There is no recovery possible
* from errors at this point as the transaction is dirty....
*/
static int
xfs_trans_run_precommits(
struct xfs_trans *tp)
{
struct xfs_mount *mp = tp->t_mountp;
struct xfs_log_item *lip, *n;
int error = 0;
/*
* Sort the item list to avoid ABBA deadlocks with other transactions
* running precommit operations that lock multiple shared items such as
* inode cluster buffers.
*/
list_sort(NULL, &tp->t_items, xfs_trans_precommit_sort);
/*
* Precommit operations can remove the log item from the transaction
* if the log item exists purely to delay modifications until they
* can be ordered against other operations. Hence we have to use
* list_for_each_entry_safe() here.
*/
list_for_each_entry_safe(lip, n, &tp->t_items, li_trans) {
if (!test_bit(XFS_LI_DIRTY, &lip->li_flags))
continue;
if (lip->li_ops->iop_precommit) {
error = lip->li_ops->iop_precommit(tp, lip);
if (error)
break;
}
}
if (error)
xfs_force_shutdown(mp, SHUTDOWN_CORRUPT_INCORE);
return error;
}
/*
* Commit the given transaction to the log.
*
* XFS disk error handling mechanism is not based on a typical
* transaction abort mechanism. Logically after the filesystem
* gets marked 'SHUTDOWN', we can't let any new transactions
* be durable - ie. committed to disk - because some metadata might
* be inconsistent. In such cases, this returns an error, and the
* caller may assume that all locked objects joined to the transaction
* have already been unlocked as if the commit had succeeded.
* Do not reference the transaction structure after this call.
*/
static int
__xfs_trans_commit(
struct xfs_trans *tp,
bool regrant)
{
struct xfs_mount *mp = tp->t_mountp;
xfs: xfs_trans_commit() path must check for log shutdown If a shut races with xfs_trans_commit() and we have shut down the filesystem but not the log, we will still cancel the transaction. This can result in aborting dirty log items instead of committing and pinning them whilst the log is still running. Hence we can end up with dirty, unlogged metadata that isn't in the AIL in memory that can be flushed to disk via writeback clustering. This was discovered from a g/388 trace where an inode log item was having IO completed on it and it wasn't in the AIL, hence tripping asserts xfs_ail_check(). Inode cluster writeback started long after the filesystem shutdown started, and long after the transaction containing the dirty inode was aborted and the log item marked XFS_LI_ABORTED. The inode was seen as dirty and unpinned, so it was flushed. IO completion tried to remove the inode from the AIL, at which point stuff went bad: XFS (pmem1): Log I/O Error (0x6) detected at xfs_fs_goingdown+0xa3/0xf0 (fs/xfs/xfs_fsops.c:500). Shutting down filesystem. XFS: Assertion failed: in_ail, file: fs/xfs/xfs_trans_ail.c, line: 67 XFS (pmem1): Please unmount the filesystem and rectify the problem(s) Workqueue: xfs-buf/pmem1 xfs_buf_ioend_work RIP: 0010:assfail+0x27/0x2d Call Trace: <TASK> xfs_ail_check+0xa8/0x180 xfs_ail_delete_one+0x3b/0xf0 xfs_buf_inode_iodone+0x329/0x3f0 xfs_buf_ioend+0x1f8/0x530 xfs_buf_ioend_work+0x15/0x20 process_one_work+0x1ac/0x390 worker_thread+0x56/0x3c0 kthread+0xf6/0x120 ret_from_fork+0x1f/0x30 </TASK> xfs_trans_commit() needs to check log state for shutdown, not mount state. It cannot abort dirty log items while the log is still running as dirty items must remained pinned in memory until they are either committed to the journal or the log has shut down and they can be safely tossed away. Hence if the log has not shut down, the xfs_trans_commit() path must allow completed transactions to commit to the CIL and pin the dirty items even if a mount shutdown has started. Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Darrick J. Wong <djwong@kernel.org> Signed-off-by: Darrick J. Wong <djwong@kernel.org>
2022-03-30 09:22:01 +08:00
struct xlog *log = mp->m_log;
xfs: xfs_log_force_lsn isn't passed a LSN In doing an investigation into AIL push stalls, I was looking at the log force code to see if an async CIL push could be done instead. This lead me to xfs_log_force_lsn() and looking at how it works. xfs_log_force_lsn() is only called from inode synchronisation contexts such as fsync(), and it takes the ip->i_itemp->ili_last_lsn value as the LSN to sync the log to. This gets passed to xlog_cil_force_lsn() via xfs_log_force_lsn() to flush the CIL to the journal, and then used by xfs_log_force_lsn() to flush the iclogs to the journal. The problem is that ip->i_itemp->ili_last_lsn does not store a log sequence number. What it stores is passed to it from the ->iop_committing method, which is called by xfs_log_commit_cil(). The value this passes to the iop_committing method is the CIL context sequence number that the item was committed to. As it turns out, xlog_cil_force_lsn() converts the sequence to an actual commit LSN for the related context and returns that to xfs_log_force_lsn(). xfs_log_force_lsn() overwrites it's "lsn" variable that contained a sequence with an actual LSN and then uses that to sync the iclogs. This caused me some confusion for a while, even though I originally wrote all this code a decade ago. ->iop_committing is only used by a couple of log item types, and only inode items use the sequence number it is passed. Let's clean up the API, CIL structures and inode log item to call it a sequence number, and make it clear that the high level code is using CIL sequence numbers and not on-disk LSNs for integrity synchronisation purposes. Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Brian Foster <bfoster@redhat.com> Reviewed-by: Darrick J. Wong <djwong@kernel.org> Reviewed-by: Allison Henderson <allison.henderson@oracle.com> Signed-off-by: Darrick J. Wong <djwong@kernel.org>
2021-06-18 23:21:52 +08:00
xfs_csn_t commit_seq = 0;
int error = 0;
int sync = tp->t_flags & XFS_TRANS_SYNC;
trace_xfs_trans_commit(tp, _RET_IP_);
xfs: add log item precommit operation For inodes that are dirty, we have an attached cluster buffer that we want to use to track the dirty inode through the AIL. Unfortunately, locking the cluster buffer and adding it to the transaction when the inode is first logged in a transaction leads to buffer lock ordering inversions. The specific problem is ordering against the AGI buffer. When modifying unlinked lists, the buffer lock order is AGI -> inode cluster buffer as the AGI buffer lock serialises all access to the unlinked lists. Unfortunately, functionality like xfs_droplink() logs the inode before calling xfs_iunlink(), as do various directory manipulation functions. The inode can be logged way down in the stack as far as the bmapi routines and hence, without a major rewrite of lots of APIs there's no way we can avoid the inode being logged by something until after the AGI has been logged. As we are going to be using ordered buffers for inode AIL tracking, there isn't a need to actually lock that buffer against modification as all the modifications are captured by logging the inode item itself. Hence we don't actually need to join the cluster buffer into the transaction until just before it is committed. This means we do not perturb any of the existing buffer lock orders in transactions, and the inode cluster buffer is always locked last in a transaction that doesn't otherwise touch inode cluster buffers. We do this by introducing a precommit log item method. This commit just introduces the mechanism; the inode item implementation is in followup commits. The precommit items need to be sorted into consistent order as we may be locking multiple items here. Hence if we have two dirty inodes in cluster buffers A and B, and some other transaction has two separate dirty inodes in the same cluster buffers, locking them in different orders opens us up to ABBA deadlocks. Hence we sort the items on the transaction based on the presence of a sort log item method. Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Darrick J. Wong <djwong@kernel.org> Reviewed-by: Christoph Hellwig <hch@lst.de>
2022-07-14 09:47:26 +08:00
error = xfs_trans_run_precommits(tp);
if (error) {
if (tp->t_flags & XFS_TRANS_PERM_LOG_RES)
xfs_defer_cancel(tp);
goto out_unreserve;
}
/*
* Finish deferred items on final commit. Only permanent transactions
* should ever have deferred ops.
*/
WARN_ON_ONCE(!list_empty(&tp->t_dfops) &&
!(tp->t_flags & XFS_TRANS_PERM_LOG_RES));
if (!regrant && (tp->t_flags & XFS_TRANS_PERM_LOG_RES)) {
error = xfs_defer_finish_noroll(&tp);
if (error)
xfs: support embedded dfops in transaction The dfops structure used by multi-transaction operations is typically stored on the stack and carried around by the associated transaction. The lifecycle of dfops does not quite match that of the transaction, but they are tightly related in that the former depends on the latter. The relationship of these objects is tight enough that we can avoid the cumbersome boilerplate code required in most cases to manage them separately by just embedding an xfs_defer_ops in the transaction itself. This means that a transaction allocation returns with an initialized dfops, a transaction commit finishes pending deferred items before the tx commit, a transaction cancel cancels the dfops before the transaction and a transaction dup operation transfers the current dfops state to the new transaction. The dup operation is slightly complicated by the fact that we can no longer just copy a dfops pointer from the old transaction to the new transaction. This is solved through a dfops move helper that transfers the pending items and other dfops state across the transactions. This also requires that transaction rolling code always refer to the transaction for the current dfops reference. Finally, to facilitate incremental conversion to the internal dfops and continue to support the current external dfops mode of operation, create the new ->t_dfops_internal field with a layer of indirection. On allocation, ->t_dfops points to the internal dfops. This state is overridden by callers who re-init a local dfops on the transaction. Once ->t_dfops is overridden, the external dfops reference is maintained as the transaction rolls. This patch adds the fundamental ability to support an internal dfops. All codepaths that perform deferred processing continue to override the internal dfops until they are converted over in subsequent patches. Signed-off-by: Brian Foster <bfoster@redhat.com> Reviewed-by: Bill O'Donnell <billodo@redhat.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Darrick J. Wong <darrick.wong@oracle.com> Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
2018-07-25 04:43:11 +08:00
goto out_unreserve;
}
/*
* If there is nothing to be logged by the transaction,
* then unlock all of the items associated with the
* transaction and free the transaction structure.
* Also make sure to return any reserved blocks to
* the free pool.
*/
if (!(tp->t_flags & XFS_TRANS_DIRTY))
goto out_unreserve;
xfs: xfs_trans_commit() path must check for log shutdown If a shut races with xfs_trans_commit() and we have shut down the filesystem but not the log, we will still cancel the transaction. This can result in aborting dirty log items instead of committing and pinning them whilst the log is still running. Hence we can end up with dirty, unlogged metadata that isn't in the AIL in memory that can be flushed to disk via writeback clustering. This was discovered from a g/388 trace where an inode log item was having IO completed on it and it wasn't in the AIL, hence tripping asserts xfs_ail_check(). Inode cluster writeback started long after the filesystem shutdown started, and long after the transaction containing the dirty inode was aborted and the log item marked XFS_LI_ABORTED. The inode was seen as dirty and unpinned, so it was flushed. IO completion tried to remove the inode from the AIL, at which point stuff went bad: XFS (pmem1): Log I/O Error (0x6) detected at xfs_fs_goingdown+0xa3/0xf0 (fs/xfs/xfs_fsops.c:500). Shutting down filesystem. XFS: Assertion failed: in_ail, file: fs/xfs/xfs_trans_ail.c, line: 67 XFS (pmem1): Please unmount the filesystem and rectify the problem(s) Workqueue: xfs-buf/pmem1 xfs_buf_ioend_work RIP: 0010:assfail+0x27/0x2d Call Trace: <TASK> xfs_ail_check+0xa8/0x180 xfs_ail_delete_one+0x3b/0xf0 xfs_buf_inode_iodone+0x329/0x3f0 xfs_buf_ioend+0x1f8/0x530 xfs_buf_ioend_work+0x15/0x20 process_one_work+0x1ac/0x390 worker_thread+0x56/0x3c0 kthread+0xf6/0x120 ret_from_fork+0x1f/0x30 </TASK> xfs_trans_commit() needs to check log state for shutdown, not mount state. It cannot abort dirty log items while the log is still running as dirty items must remained pinned in memory until they are either committed to the journal or the log has shut down and they can be safely tossed away. Hence if the log has not shut down, the xfs_trans_commit() path must allow completed transactions to commit to the CIL and pin the dirty items even if a mount shutdown has started. Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Darrick J. Wong <djwong@kernel.org> Signed-off-by: Darrick J. Wong <djwong@kernel.org>
2022-03-30 09:22:01 +08:00
/*
* We must check against log shutdown here because we cannot abort log
* items and leave them dirty, inconsistent and unpinned in memory while
* the log is active. This leaves them open to being written back to
* disk, and that will lead to on-disk corruption.
*/
if (xlog_is_shutdown(log)) {
error = -EIO;
goto out_unreserve;
}
ASSERT(tp->t_ticket != NULL);
/*
* If we need to update the superblock, then do it now.
*/
if (tp->t_flags & XFS_TRANS_SB_DIRTY)
xfs_trans_apply_sb_deltas(tp);
xfs_trans_apply_dquot_deltas(tp);
xfs: xfs_trans_commit() path must check for log shutdown If a shut races with xfs_trans_commit() and we have shut down the filesystem but not the log, we will still cancel the transaction. This can result in aborting dirty log items instead of committing and pinning them whilst the log is still running. Hence we can end up with dirty, unlogged metadata that isn't in the AIL in memory that can be flushed to disk via writeback clustering. This was discovered from a g/388 trace where an inode log item was having IO completed on it and it wasn't in the AIL, hence tripping asserts xfs_ail_check(). Inode cluster writeback started long after the filesystem shutdown started, and long after the transaction containing the dirty inode was aborted and the log item marked XFS_LI_ABORTED. The inode was seen as dirty and unpinned, so it was flushed. IO completion tried to remove the inode from the AIL, at which point stuff went bad: XFS (pmem1): Log I/O Error (0x6) detected at xfs_fs_goingdown+0xa3/0xf0 (fs/xfs/xfs_fsops.c:500). Shutting down filesystem. XFS: Assertion failed: in_ail, file: fs/xfs/xfs_trans_ail.c, line: 67 XFS (pmem1): Please unmount the filesystem and rectify the problem(s) Workqueue: xfs-buf/pmem1 xfs_buf_ioend_work RIP: 0010:assfail+0x27/0x2d Call Trace: <TASK> xfs_ail_check+0xa8/0x180 xfs_ail_delete_one+0x3b/0xf0 xfs_buf_inode_iodone+0x329/0x3f0 xfs_buf_ioend+0x1f8/0x530 xfs_buf_ioend_work+0x15/0x20 process_one_work+0x1ac/0x390 worker_thread+0x56/0x3c0 kthread+0xf6/0x120 ret_from_fork+0x1f/0x30 </TASK> xfs_trans_commit() needs to check log state for shutdown, not mount state. It cannot abort dirty log items while the log is still running as dirty items must remained pinned in memory until they are either committed to the journal or the log has shut down and they can be safely tossed away. Hence if the log has not shut down, the xfs_trans_commit() path must allow completed transactions to commit to the CIL and pin the dirty items even if a mount shutdown has started. Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Darrick J. Wong <djwong@kernel.org> Signed-off-by: Darrick J. Wong <djwong@kernel.org>
2022-03-30 09:22:01 +08:00
xlog_cil_commit(log, tp, &commit_seq, regrant);
xfs_trans_free(tp);
/*
* If the transaction needs to be synchronous, then force the
* log out now and wait for it.
*/
if (sync) {
xfs: xfs_log_force_lsn isn't passed a LSN In doing an investigation into AIL push stalls, I was looking at the log force code to see if an async CIL push could be done instead. This lead me to xfs_log_force_lsn() and looking at how it works. xfs_log_force_lsn() is only called from inode synchronisation contexts such as fsync(), and it takes the ip->i_itemp->ili_last_lsn value as the LSN to sync the log to. This gets passed to xlog_cil_force_lsn() via xfs_log_force_lsn() to flush the CIL to the journal, and then used by xfs_log_force_lsn() to flush the iclogs to the journal. The problem is that ip->i_itemp->ili_last_lsn does not store a log sequence number. What it stores is passed to it from the ->iop_committing method, which is called by xfs_log_commit_cil(). The value this passes to the iop_committing method is the CIL context sequence number that the item was committed to. As it turns out, xlog_cil_force_lsn() converts the sequence to an actual commit LSN for the related context and returns that to xfs_log_force_lsn(). xfs_log_force_lsn() overwrites it's "lsn" variable that contained a sequence with an actual LSN and then uses that to sync the iclogs. This caused me some confusion for a while, even though I originally wrote all this code a decade ago. ->iop_committing is only used by a couple of log item types, and only inode items use the sequence number it is passed. Let's clean up the API, CIL structures and inode log item to call it a sequence number, and make it clear that the high level code is using CIL sequence numbers and not on-disk LSNs for integrity synchronisation purposes. Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Brian Foster <bfoster@redhat.com> Reviewed-by: Darrick J. Wong <djwong@kernel.org> Reviewed-by: Allison Henderson <allison.henderson@oracle.com> Signed-off-by: Darrick J. Wong <djwong@kernel.org>
2021-06-18 23:21:52 +08:00
error = xfs_log_force_seq(mp, commit_seq, XFS_LOG_SYNC, NULL);
XFS_STATS_INC(mp, xs_trans_sync);
} else {
XFS_STATS_INC(mp, xs_trans_async);
}
return error;
out_unreserve:
xfs_trans_unreserve_and_mod_sb(tp);
/*
* It is indeed possible for the transaction to be not dirty but
* the dqinfo portion to be. All that means is that we have some
* (non-persistent) quota reservations that need to be unreserved.
*/
xfs_trans_unreserve_and_mod_dquots(tp);
if (tp->t_ticket) {
xfs: xfs_trans_commit() path must check for log shutdown If a shut races with xfs_trans_commit() and we have shut down the filesystem but not the log, we will still cancel the transaction. This can result in aborting dirty log items instead of committing and pinning them whilst the log is still running. Hence we can end up with dirty, unlogged metadata that isn't in the AIL in memory that can be flushed to disk via writeback clustering. This was discovered from a g/388 trace where an inode log item was having IO completed on it and it wasn't in the AIL, hence tripping asserts xfs_ail_check(). Inode cluster writeback started long after the filesystem shutdown started, and long after the transaction containing the dirty inode was aborted and the log item marked XFS_LI_ABORTED. The inode was seen as dirty and unpinned, so it was flushed. IO completion tried to remove the inode from the AIL, at which point stuff went bad: XFS (pmem1): Log I/O Error (0x6) detected at xfs_fs_goingdown+0xa3/0xf0 (fs/xfs/xfs_fsops.c:500). Shutting down filesystem. XFS: Assertion failed: in_ail, file: fs/xfs/xfs_trans_ail.c, line: 67 XFS (pmem1): Please unmount the filesystem and rectify the problem(s) Workqueue: xfs-buf/pmem1 xfs_buf_ioend_work RIP: 0010:assfail+0x27/0x2d Call Trace: <TASK> xfs_ail_check+0xa8/0x180 xfs_ail_delete_one+0x3b/0xf0 xfs_buf_inode_iodone+0x329/0x3f0 xfs_buf_ioend+0x1f8/0x530 xfs_buf_ioend_work+0x15/0x20 process_one_work+0x1ac/0x390 worker_thread+0x56/0x3c0 kthread+0xf6/0x120 ret_from_fork+0x1f/0x30 </TASK> xfs_trans_commit() needs to check log state for shutdown, not mount state. It cannot abort dirty log items while the log is still running as dirty items must remained pinned in memory until they are either committed to the journal or the log has shut down and they can be safely tossed away. Hence if the log has not shut down, the xfs_trans_commit() path must allow completed transactions to commit to the CIL and pin the dirty items even if a mount shutdown has started. Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Darrick J. Wong <djwong@kernel.org> Signed-off-by: Darrick J. Wong <djwong@kernel.org>
2022-03-30 09:22:01 +08:00
if (regrant && !xlog_is_shutdown(log))
xfs_log_ticket_regrant(log, tp->t_ticket);
else
xfs: xfs_trans_commit() path must check for log shutdown If a shut races with xfs_trans_commit() and we have shut down the filesystem but not the log, we will still cancel the transaction. This can result in aborting dirty log items instead of committing and pinning them whilst the log is still running. Hence we can end up with dirty, unlogged metadata that isn't in the AIL in memory that can be flushed to disk via writeback clustering. This was discovered from a g/388 trace where an inode log item was having IO completed on it and it wasn't in the AIL, hence tripping asserts xfs_ail_check(). Inode cluster writeback started long after the filesystem shutdown started, and long after the transaction containing the dirty inode was aborted and the log item marked XFS_LI_ABORTED. The inode was seen as dirty and unpinned, so it was flushed. IO completion tried to remove the inode from the AIL, at which point stuff went bad: XFS (pmem1): Log I/O Error (0x6) detected at xfs_fs_goingdown+0xa3/0xf0 (fs/xfs/xfs_fsops.c:500). Shutting down filesystem. XFS: Assertion failed: in_ail, file: fs/xfs/xfs_trans_ail.c, line: 67 XFS (pmem1): Please unmount the filesystem and rectify the problem(s) Workqueue: xfs-buf/pmem1 xfs_buf_ioend_work RIP: 0010:assfail+0x27/0x2d Call Trace: <TASK> xfs_ail_check+0xa8/0x180 xfs_ail_delete_one+0x3b/0xf0 xfs_buf_inode_iodone+0x329/0x3f0 xfs_buf_ioend+0x1f8/0x530 xfs_buf_ioend_work+0x15/0x20 process_one_work+0x1ac/0x390 worker_thread+0x56/0x3c0 kthread+0xf6/0x120 ret_from_fork+0x1f/0x30 </TASK> xfs_trans_commit() needs to check log state for shutdown, not mount state. It cannot abort dirty log items while the log is still running as dirty items must remained pinned in memory until they are either committed to the journal or the log has shut down and they can be safely tossed away. Hence if the log has not shut down, the xfs_trans_commit() path must allow completed transactions to commit to the CIL and pin the dirty items even if a mount shutdown has started. Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Darrick J. Wong <djwong@kernel.org> Signed-off-by: Darrick J. Wong <djwong@kernel.org>
2022-03-30 09:22:01 +08:00
xfs_log_ticket_ungrant(log, tp->t_ticket);
tp->t_ticket = NULL;
}
xfs_trans_free_items(tp, !!error);
xfs_trans_free(tp);
XFS_STATS_INC(mp, xs_trans_empty);
return error;
}
int
xfs_trans_commit(
struct xfs_trans *tp)
{
return __xfs_trans_commit(tp, false);
}
/*
xfs: xfs_trans_commit() path must check for log shutdown If a shut races with xfs_trans_commit() and we have shut down the filesystem but not the log, we will still cancel the transaction. This can result in aborting dirty log items instead of committing and pinning them whilst the log is still running. Hence we can end up with dirty, unlogged metadata that isn't in the AIL in memory that can be flushed to disk via writeback clustering. This was discovered from a g/388 trace where an inode log item was having IO completed on it and it wasn't in the AIL, hence tripping asserts xfs_ail_check(). Inode cluster writeback started long after the filesystem shutdown started, and long after the transaction containing the dirty inode was aborted and the log item marked XFS_LI_ABORTED. The inode was seen as dirty and unpinned, so it was flushed. IO completion tried to remove the inode from the AIL, at which point stuff went bad: XFS (pmem1): Log I/O Error (0x6) detected at xfs_fs_goingdown+0xa3/0xf0 (fs/xfs/xfs_fsops.c:500). Shutting down filesystem. XFS: Assertion failed: in_ail, file: fs/xfs/xfs_trans_ail.c, line: 67 XFS (pmem1): Please unmount the filesystem and rectify the problem(s) Workqueue: xfs-buf/pmem1 xfs_buf_ioend_work RIP: 0010:assfail+0x27/0x2d Call Trace: <TASK> xfs_ail_check+0xa8/0x180 xfs_ail_delete_one+0x3b/0xf0 xfs_buf_inode_iodone+0x329/0x3f0 xfs_buf_ioend+0x1f8/0x530 xfs_buf_ioend_work+0x15/0x20 process_one_work+0x1ac/0x390 worker_thread+0x56/0x3c0 kthread+0xf6/0x120 ret_from_fork+0x1f/0x30 </TASK> xfs_trans_commit() needs to check log state for shutdown, not mount state. It cannot abort dirty log items while the log is still running as dirty items must remained pinned in memory until they are either committed to the journal or the log has shut down and they can be safely tossed away. Hence if the log has not shut down, the xfs_trans_commit() path must allow completed transactions to commit to the CIL and pin the dirty items even if a mount shutdown has started. Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Darrick J. Wong <djwong@kernel.org> Signed-off-by: Darrick J. Wong <djwong@kernel.org>
2022-03-30 09:22:01 +08:00
* Unlock all of the transaction's items and free the transaction. If the
* transaction is dirty, we must shut down the filesystem because there is no
* way to restore them to their previous state.
*
xfs: xfs_trans_commit() path must check for log shutdown If a shut races with xfs_trans_commit() and we have shut down the filesystem but not the log, we will still cancel the transaction. This can result in aborting dirty log items instead of committing and pinning them whilst the log is still running. Hence we can end up with dirty, unlogged metadata that isn't in the AIL in memory that can be flushed to disk via writeback clustering. This was discovered from a g/388 trace where an inode log item was having IO completed on it and it wasn't in the AIL, hence tripping asserts xfs_ail_check(). Inode cluster writeback started long after the filesystem shutdown started, and long after the transaction containing the dirty inode was aborted and the log item marked XFS_LI_ABORTED. The inode was seen as dirty and unpinned, so it was flushed. IO completion tried to remove the inode from the AIL, at which point stuff went bad: XFS (pmem1): Log I/O Error (0x6) detected at xfs_fs_goingdown+0xa3/0xf0 (fs/xfs/xfs_fsops.c:500). Shutting down filesystem. XFS: Assertion failed: in_ail, file: fs/xfs/xfs_trans_ail.c, line: 67 XFS (pmem1): Please unmount the filesystem and rectify the problem(s) Workqueue: xfs-buf/pmem1 xfs_buf_ioend_work RIP: 0010:assfail+0x27/0x2d Call Trace: <TASK> xfs_ail_check+0xa8/0x180 xfs_ail_delete_one+0x3b/0xf0 xfs_buf_inode_iodone+0x329/0x3f0 xfs_buf_ioend+0x1f8/0x530 xfs_buf_ioend_work+0x15/0x20 process_one_work+0x1ac/0x390 worker_thread+0x56/0x3c0 kthread+0xf6/0x120 ret_from_fork+0x1f/0x30 </TASK> xfs_trans_commit() needs to check log state for shutdown, not mount state. It cannot abort dirty log items while the log is still running as dirty items must remained pinned in memory until they are either committed to the journal or the log has shut down and they can be safely tossed away. Hence if the log has not shut down, the xfs_trans_commit() path must allow completed transactions to commit to the CIL and pin the dirty items even if a mount shutdown has started. Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Darrick J. Wong <djwong@kernel.org> Signed-off-by: Darrick J. Wong <djwong@kernel.org>
2022-03-30 09:22:01 +08:00
* If the transaction has made a log reservation, make sure to release it as
* well.
*
* This is a high level function (equivalent to xfs_trans_commit()) and so can
* be called after the transaction has effectively been aborted due to the mount
* being shut down. However, if the mount has not been shut down and the
* transaction is dirty we will shut the mount down and, in doing so, that
* guarantees that the log is shut down, too. Hence we don't need to be as
* careful with shutdown state and dirty items here as we need to be in
* xfs_trans_commit().
*/
void
xfs_trans_cancel(
struct xfs_trans *tp)
{
struct xfs_mount *mp = tp->t_mountp;
xfs: xfs_trans_commit() path must check for log shutdown If a shut races with xfs_trans_commit() and we have shut down the filesystem but not the log, we will still cancel the transaction. This can result in aborting dirty log items instead of committing and pinning them whilst the log is still running. Hence we can end up with dirty, unlogged metadata that isn't in the AIL in memory that can be flushed to disk via writeback clustering. This was discovered from a g/388 trace where an inode log item was having IO completed on it and it wasn't in the AIL, hence tripping asserts xfs_ail_check(). Inode cluster writeback started long after the filesystem shutdown started, and long after the transaction containing the dirty inode was aborted and the log item marked XFS_LI_ABORTED. The inode was seen as dirty and unpinned, so it was flushed. IO completion tried to remove the inode from the AIL, at which point stuff went bad: XFS (pmem1): Log I/O Error (0x6) detected at xfs_fs_goingdown+0xa3/0xf0 (fs/xfs/xfs_fsops.c:500). Shutting down filesystem. XFS: Assertion failed: in_ail, file: fs/xfs/xfs_trans_ail.c, line: 67 XFS (pmem1): Please unmount the filesystem and rectify the problem(s) Workqueue: xfs-buf/pmem1 xfs_buf_ioend_work RIP: 0010:assfail+0x27/0x2d Call Trace: <TASK> xfs_ail_check+0xa8/0x180 xfs_ail_delete_one+0x3b/0xf0 xfs_buf_inode_iodone+0x329/0x3f0 xfs_buf_ioend+0x1f8/0x530 xfs_buf_ioend_work+0x15/0x20 process_one_work+0x1ac/0x390 worker_thread+0x56/0x3c0 kthread+0xf6/0x120 ret_from_fork+0x1f/0x30 </TASK> xfs_trans_commit() needs to check log state for shutdown, not mount state. It cannot abort dirty log items while the log is still running as dirty items must remained pinned in memory until they are either committed to the journal or the log has shut down and they can be safely tossed away. Hence if the log has not shut down, the xfs_trans_commit() path must allow completed transactions to commit to the CIL and pin the dirty items even if a mount shutdown has started. Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Darrick J. Wong <djwong@kernel.org> Signed-off-by: Darrick J. Wong <djwong@kernel.org>
2022-03-30 09:22:01 +08:00
struct xlog *log = mp->m_log;
bool dirty = (tp->t_flags & XFS_TRANS_DIRTY);
trace_xfs_trans_cancel(tp, _RET_IP_);
/*
* It's never valid to cancel a transaction with deferred ops attached,
* because the transaction is effectively dirty. Complain about this
* loudly before freeing the in-memory defer items.
*/
if (!list_empty(&tp->t_dfops)) {
ASSERT(xfs_is_shutdown(mp) || list_empty(&tp->t_dfops));
ASSERT(tp->t_flags & XFS_TRANS_PERM_LOG_RES);
dirty = true;
xfs_defer_cancel(tp);
}
xfs: support embedded dfops in transaction The dfops structure used by multi-transaction operations is typically stored on the stack and carried around by the associated transaction. The lifecycle of dfops does not quite match that of the transaction, but they are tightly related in that the former depends on the latter. The relationship of these objects is tight enough that we can avoid the cumbersome boilerplate code required in most cases to manage them separately by just embedding an xfs_defer_ops in the transaction itself. This means that a transaction allocation returns with an initialized dfops, a transaction commit finishes pending deferred items before the tx commit, a transaction cancel cancels the dfops before the transaction and a transaction dup operation transfers the current dfops state to the new transaction. The dup operation is slightly complicated by the fact that we can no longer just copy a dfops pointer from the old transaction to the new transaction. This is solved through a dfops move helper that transfers the pending items and other dfops state across the transactions. This also requires that transaction rolling code always refer to the transaction for the current dfops reference. Finally, to facilitate incremental conversion to the internal dfops and continue to support the current external dfops mode of operation, create the new ->t_dfops_internal field with a layer of indirection. On allocation, ->t_dfops points to the internal dfops. This state is overridden by callers who re-init a local dfops on the transaction. Once ->t_dfops is overridden, the external dfops reference is maintained as the transaction rolls. This patch adds the fundamental ability to support an internal dfops. All codepaths that perform deferred processing continue to override the internal dfops until they are converted over in subsequent patches. Signed-off-by: Brian Foster <bfoster@redhat.com> Reviewed-by: Bill O'Donnell <billodo@redhat.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Darrick J. Wong <darrick.wong@oracle.com> Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
2018-07-25 04:43:11 +08:00
/*
xfs: xfs_trans_commit() path must check for log shutdown If a shut races with xfs_trans_commit() and we have shut down the filesystem but not the log, we will still cancel the transaction. This can result in aborting dirty log items instead of committing and pinning them whilst the log is still running. Hence we can end up with dirty, unlogged metadata that isn't in the AIL in memory that can be flushed to disk via writeback clustering. This was discovered from a g/388 trace where an inode log item was having IO completed on it and it wasn't in the AIL, hence tripping asserts xfs_ail_check(). Inode cluster writeback started long after the filesystem shutdown started, and long after the transaction containing the dirty inode was aborted and the log item marked XFS_LI_ABORTED. The inode was seen as dirty and unpinned, so it was flushed. IO completion tried to remove the inode from the AIL, at which point stuff went bad: XFS (pmem1): Log I/O Error (0x6) detected at xfs_fs_goingdown+0xa3/0xf0 (fs/xfs/xfs_fsops.c:500). Shutting down filesystem. XFS: Assertion failed: in_ail, file: fs/xfs/xfs_trans_ail.c, line: 67 XFS (pmem1): Please unmount the filesystem and rectify the problem(s) Workqueue: xfs-buf/pmem1 xfs_buf_ioend_work RIP: 0010:assfail+0x27/0x2d Call Trace: <TASK> xfs_ail_check+0xa8/0x180 xfs_ail_delete_one+0x3b/0xf0 xfs_buf_inode_iodone+0x329/0x3f0 xfs_buf_ioend+0x1f8/0x530 xfs_buf_ioend_work+0x15/0x20 process_one_work+0x1ac/0x390 worker_thread+0x56/0x3c0 kthread+0xf6/0x120 ret_from_fork+0x1f/0x30 </TASK> xfs_trans_commit() needs to check log state for shutdown, not mount state. It cannot abort dirty log items while the log is still running as dirty items must remained pinned in memory until they are either committed to the journal or the log has shut down and they can be safely tossed away. Hence if the log has not shut down, the xfs_trans_commit() path must allow completed transactions to commit to the CIL and pin the dirty items even if a mount shutdown has started. Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Darrick J. Wong <djwong@kernel.org> Signed-off-by: Darrick J. Wong <djwong@kernel.org>
2022-03-30 09:22:01 +08:00
* See if the caller is relying on us to shut down the filesystem. We
* only want an error report if there isn't already a shutdown in
* progress, so we only need to check against the mount shutdown state
* here.
*/
if (dirty && !xfs_is_shutdown(mp)) {
XFS_ERROR_REPORT("xfs_trans_cancel", XFS_ERRLEVEL_LOW, mp);
xfs_force_shutdown(mp, SHUTDOWN_CORRUPT_INCORE);
}
#ifdef DEBUG
xfs: xfs_trans_commit() path must check for log shutdown If a shut races with xfs_trans_commit() and we have shut down the filesystem but not the log, we will still cancel the transaction. This can result in aborting dirty log items instead of committing and pinning them whilst the log is still running. Hence we can end up with dirty, unlogged metadata that isn't in the AIL in memory that can be flushed to disk via writeback clustering. This was discovered from a g/388 trace where an inode log item was having IO completed on it and it wasn't in the AIL, hence tripping asserts xfs_ail_check(). Inode cluster writeback started long after the filesystem shutdown started, and long after the transaction containing the dirty inode was aborted and the log item marked XFS_LI_ABORTED. The inode was seen as dirty and unpinned, so it was flushed. IO completion tried to remove the inode from the AIL, at which point stuff went bad: XFS (pmem1): Log I/O Error (0x6) detected at xfs_fs_goingdown+0xa3/0xf0 (fs/xfs/xfs_fsops.c:500). Shutting down filesystem. XFS: Assertion failed: in_ail, file: fs/xfs/xfs_trans_ail.c, line: 67 XFS (pmem1): Please unmount the filesystem and rectify the problem(s) Workqueue: xfs-buf/pmem1 xfs_buf_ioend_work RIP: 0010:assfail+0x27/0x2d Call Trace: <TASK> xfs_ail_check+0xa8/0x180 xfs_ail_delete_one+0x3b/0xf0 xfs_buf_inode_iodone+0x329/0x3f0 xfs_buf_ioend+0x1f8/0x530 xfs_buf_ioend_work+0x15/0x20 process_one_work+0x1ac/0x390 worker_thread+0x56/0x3c0 kthread+0xf6/0x120 ret_from_fork+0x1f/0x30 </TASK> xfs_trans_commit() needs to check log state for shutdown, not mount state. It cannot abort dirty log items while the log is still running as dirty items must remained pinned in memory until they are either committed to the journal or the log has shut down and they can be safely tossed away. Hence if the log has not shut down, the xfs_trans_commit() path must allow completed transactions to commit to the CIL and pin the dirty items even if a mount shutdown has started. Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Darrick J. Wong <djwong@kernel.org> Signed-off-by: Darrick J. Wong <djwong@kernel.org>
2022-03-30 09:22:01 +08:00
/* Log items need to be consistent until the log is shut down. */
if (!dirty && !xlog_is_shutdown(log)) {
struct xfs_log_item *lip;
list_for_each_entry(lip, &tp->t_items, li_trans)
ASSERT(!xlog_item_is_intent_done(lip));
}
#endif
xfs_trans_unreserve_and_mod_sb(tp);
xfs_trans_unreserve_and_mod_dquots(tp);
if (tp->t_ticket) {
xfs: xfs_trans_commit() path must check for log shutdown If a shut races with xfs_trans_commit() and we have shut down the filesystem but not the log, we will still cancel the transaction. This can result in aborting dirty log items instead of committing and pinning them whilst the log is still running. Hence we can end up with dirty, unlogged metadata that isn't in the AIL in memory that can be flushed to disk via writeback clustering. This was discovered from a g/388 trace where an inode log item was having IO completed on it and it wasn't in the AIL, hence tripping asserts xfs_ail_check(). Inode cluster writeback started long after the filesystem shutdown started, and long after the transaction containing the dirty inode was aborted and the log item marked XFS_LI_ABORTED. The inode was seen as dirty and unpinned, so it was flushed. IO completion tried to remove the inode from the AIL, at which point stuff went bad: XFS (pmem1): Log I/O Error (0x6) detected at xfs_fs_goingdown+0xa3/0xf0 (fs/xfs/xfs_fsops.c:500). Shutting down filesystem. XFS: Assertion failed: in_ail, file: fs/xfs/xfs_trans_ail.c, line: 67 XFS (pmem1): Please unmount the filesystem and rectify the problem(s) Workqueue: xfs-buf/pmem1 xfs_buf_ioend_work RIP: 0010:assfail+0x27/0x2d Call Trace: <TASK> xfs_ail_check+0xa8/0x180 xfs_ail_delete_one+0x3b/0xf0 xfs_buf_inode_iodone+0x329/0x3f0 xfs_buf_ioend+0x1f8/0x530 xfs_buf_ioend_work+0x15/0x20 process_one_work+0x1ac/0x390 worker_thread+0x56/0x3c0 kthread+0xf6/0x120 ret_from_fork+0x1f/0x30 </TASK> xfs_trans_commit() needs to check log state for shutdown, not mount state. It cannot abort dirty log items while the log is still running as dirty items must remained pinned in memory until they are either committed to the journal or the log has shut down and they can be safely tossed away. Hence if the log has not shut down, the xfs_trans_commit() path must allow completed transactions to commit to the CIL and pin the dirty items even if a mount shutdown has started. Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Darrick J. Wong <djwong@kernel.org> Signed-off-by: Darrick J. Wong <djwong@kernel.org>
2022-03-30 09:22:01 +08:00
xfs_log_ticket_ungrant(log, tp->t_ticket);
tp->t_ticket = NULL;
}
xfs_trans_free_items(tp, dirty);
xfs_trans_free(tp);
}
/*
* Roll from one trans in the sequence of PERMANENT transactions to
* the next: permanent transactions are only flushed out when
* committed with xfs_trans_commit(), but we still want as soon
* as possible to let chunks of it go to the log. So we commit the
* chunk we've been working on and get a new transaction to continue.
*/
int
xfs_trans_roll(
struct xfs_trans **tpp)
{
struct xfs_trans *trans = *tpp;
struct xfs_trans_res tres;
int error;
trace_xfs_trans_roll(trans, _RET_IP_);
/*
* Copy the critical parameters from one trans to the next.
*/
tres.tr_logres = trans->t_log_res;
tres.tr_logcount = trans->t_log_count;
*tpp = xfs_trans_dup(trans);
/*
* Commit the current transaction.
* If this commit failed, then it'd just unlock those items that
* are not marked ihold. That also means that a filesystem shutdown
* is in progress. The caller takes the responsibility to cancel
* the duplicate transaction that gets returned.
*/
error = __xfs_trans_commit(trans, true);
if (error)
return error;
/*
* Reserve space in the log for the next transaction.
* This also pushes items in the "AIL", the list of logged items,
* out to disk if they are taking up space at the tail of the log
* that we want to use. This requires that either nothing be locked
* across this call, or that anything that is locked be logged in
* the prior and the next transactions.
*/
tres.tr_logflags = XFS_TRANS_PERM_LOG_RES;
return xfs_trans_reserve(*tpp, &tres, 0, 0);
}
/*
* Allocate an transaction, lock and join the inode to it, and reserve quota.
*
* The caller must ensure that the on-disk dquots attached to this inode have
* already been allocated and initialized. The caller is responsible for
* releasing ILOCK_EXCL if a new transaction is returned.
*/
int
xfs_trans_alloc_inode(
struct xfs_inode *ip,
struct xfs_trans_res *resv,
unsigned int dblocks,
unsigned int rblocks,
bool force,
struct xfs_trans **tpp)
{
struct xfs_trans *tp;
struct xfs_mount *mp = ip->i_mount;
bool retried = false;
int error;
retry:
error = xfs_trans_alloc(mp, resv, dblocks,
rblocks / mp->m_sb.sb_rextsize,
force ? XFS_TRANS_RESERVE : 0, &tp);
if (error)
return error;
xfs_ilock(ip, XFS_ILOCK_EXCL);
xfs_trans_ijoin(tp, ip, 0);
error = xfs_qm_dqattach_locked(ip, false);
if (error) {
/* Caller should have allocated the dquots! */
ASSERT(error != -ENOENT);
goto out_cancel;
}
error = xfs_trans_reserve_quota_nblks(tp, ip, dblocks, rblocks, force);
if ((error == -EDQUOT || error == -ENOSPC) && !retried) {
xfs_trans_cancel(tp);
xfs_iunlock(ip, XFS_ILOCK_EXCL);
xfs_blockgc_free_quota(ip, 0);
retried = true;
goto retry;
}
if (error)
goto out_cancel;
*tpp = tp;
return 0;
out_cancel:
xfs_trans_cancel(tp);
xfs_iunlock(ip, XFS_ILOCK_EXCL);
return error;
}
/*
* Allocate an transaction in preparation for inode creation by reserving quota
* against the given dquots. Callers are not required to hold any inode locks.
*/
int
xfs_trans_alloc_icreate(
struct xfs_mount *mp,
struct xfs_trans_res *resv,
struct xfs_dquot *udqp,
struct xfs_dquot *gdqp,
struct xfs_dquot *pdqp,
unsigned int dblocks,
struct xfs_trans **tpp)
{
struct xfs_trans *tp;
bool retried = false;
int error;
retry:
error = xfs_trans_alloc(mp, resv, dblocks, 0, 0, &tp);
if (error)
return error;
error = xfs_trans_reserve_quota_icreate(tp, udqp, gdqp, pdqp, dblocks);
if ((error == -EDQUOT || error == -ENOSPC) && !retried) {
xfs_trans_cancel(tp);
xfs_blockgc_free_dquots(mp, udqp, gdqp, pdqp, 0);
retried = true;
goto retry;
}
if (error) {
xfs_trans_cancel(tp);
return error;
}
*tpp = tp;
return 0;
}
/*
* Allocate an transaction, lock and join the inode to it, and reserve quota
* in preparation for inode attribute changes that include uid, gid, or prid
* changes.
*
* The caller must ensure that the on-disk dquots attached to this inode have
* already been allocated and initialized. The ILOCK will be dropped when the
* transaction is committed or cancelled.
*/
int
xfs_trans_alloc_ichange(
struct xfs_inode *ip,
struct xfs_dquot *new_udqp,
struct xfs_dquot *new_gdqp,
struct xfs_dquot *new_pdqp,
bool force,
struct xfs_trans **tpp)
{
struct xfs_trans *tp;
struct xfs_mount *mp = ip->i_mount;
struct xfs_dquot *udqp;
struct xfs_dquot *gdqp;
struct xfs_dquot *pdqp;
bool retried = false;
int error;
retry:
error = xfs_trans_alloc(mp, &M_RES(mp)->tr_ichange, 0, 0, 0, &tp);
if (error)
return error;
xfs_ilock(ip, XFS_ILOCK_EXCL);
xfs_trans_ijoin(tp, ip, XFS_ILOCK_EXCL);
error = xfs_qm_dqattach_locked(ip, false);
if (error) {
/* Caller should have allocated the dquots! */
ASSERT(error != -ENOENT);
goto out_cancel;
}
/*
* For each quota type, skip quota reservations if the inode's dquots
* now match the ones that came from the caller, or the caller didn't
* pass one in. The inode's dquots can change if we drop the ILOCK to
* perform a blockgc scan, so we must preserve the caller's arguments.
*/
udqp = (new_udqp != ip->i_udquot) ? new_udqp : NULL;
gdqp = (new_gdqp != ip->i_gdquot) ? new_gdqp : NULL;
pdqp = (new_pdqp != ip->i_pdquot) ? new_pdqp : NULL;
if (udqp || gdqp || pdqp) {
unsigned int qflags = XFS_QMOPT_RES_REGBLKS;
if (force)
qflags |= XFS_QMOPT_FORCE_RES;
/*
* Reserve enough quota to handle blocks on disk and reserved
* for a delayed allocation. We'll actually transfer the
* delalloc reservation between dquots at chown time, even
* though that part is only semi-transactional.
*/
error = xfs_trans_reserve_quota_bydquots(tp, mp, udqp, gdqp,
pdqp, ip->i_nblocks + ip->i_delayed_blks,
1, qflags);
if ((error == -EDQUOT || error == -ENOSPC) && !retried) {
xfs_trans_cancel(tp);
xfs_blockgc_free_dquots(mp, udqp, gdqp, pdqp, 0);
retried = true;
goto retry;
}
if (error)
goto out_cancel;
}
*tpp = tp;
return 0;
out_cancel:
xfs_trans_cancel(tp);
return error;
}
/*
* Allocate an transaction, lock and join the directory and child inodes to it,
* and reserve quota for a directory update. If there isn't sufficient space,
* @dblocks will be set to zero for a reservationless directory update and
* @nospace_error will be set to a negative errno describing the space
* constraint we hit.
*
* The caller must ensure that the on-disk dquots attached to this inode have
* already been allocated and initialized. The ILOCKs will be dropped when the
* transaction is committed or cancelled.
*/
int
xfs_trans_alloc_dir(
struct xfs_inode *dp,
struct xfs_trans_res *resv,
struct xfs_inode *ip,
unsigned int *dblocks,
struct xfs_trans **tpp,
int *nospace_error)
{
struct xfs_trans *tp;
struct xfs_mount *mp = ip->i_mount;
unsigned int resblks;
bool retried = false;
int error;
retry:
*nospace_error = 0;
resblks = *dblocks;
error = xfs_trans_alloc(mp, resv, resblks, 0, 0, &tp);
if (error == -ENOSPC) {
*nospace_error = error;
resblks = 0;
error = xfs_trans_alloc(mp, resv, resblks, 0, 0, &tp);
}
if (error)
return error;
xfs_lock_two_inodes(dp, XFS_ILOCK_EXCL, ip, XFS_ILOCK_EXCL);
xfs_trans_ijoin(tp, dp, XFS_ILOCK_EXCL);
xfs_trans_ijoin(tp, ip, XFS_ILOCK_EXCL);
error = xfs_qm_dqattach_locked(dp, false);
if (error) {
/* Caller should have allocated the dquots! */
ASSERT(error != -ENOENT);
goto out_cancel;
}
error = xfs_qm_dqattach_locked(ip, false);
if (error) {
/* Caller should have allocated the dquots! */
ASSERT(error != -ENOENT);
goto out_cancel;
}
if (resblks == 0)
goto done;
error = xfs_trans_reserve_quota_nblks(tp, dp, resblks, 0, false);
if (error == -EDQUOT || error == -ENOSPC) {
if (!retried) {
xfs_trans_cancel(tp);
xfs_blockgc_free_quota(dp, 0);
retried = true;
goto retry;
}
*nospace_error = error;
resblks = 0;
error = 0;
}
if (error)
goto out_cancel;
done:
*tpp = tp;
*dblocks = resblks;
return 0;
out_cancel:
xfs_trans_cancel(tp);
return error;
}