Skip to contentSkip to conversation

xfs: improve transaction rate scalability

PATCH In review Dave Chinner 4 participants 12 messages May 19, 2020 – May 20, 2020 XFS filesystem

lore
Share & tools
lore
#0 lore raw

v3 cover xfs: improve transaction rate scalability

Hi folks,

This is the third version of the patchset to improve the transaction rate on higher CPU count machines. The previous versions can be found here:

https://lore.kernel.org/linux-xfs/[email protected]/
https://lore.kernel.org/linux-xfs/[email protected]/

Changes for v3 are:
- completely reorganise the struct xfs_mount, not just a subset of the variables. This seems to improve performance even more than the original version. (Christoph requested this.)
- remove the m_active_trans counter rather than convert it to percpu as Darrick suggested.

Results are just as good or better than previous versions, passes fstests without regressions on both low PCU count and high CPU count test VMs.

Cheers,

Dave.

#1 lore raw

PATCH v3 1/2 xfs: separate read-only variables in struct xfs_mount

From: Dave Chinner <[email protected]>

Seeing massive cpu usage from xfs_agino_range() on one machine; instruction level profiles look similar to another machine running the same workload, only one machien is consuming 10x as much CPU as the other and going much slower. The only real difference between the two machines is core count per socket. Both are running identical 16p/16GB virtual machine configurations

Machine A:

  25.83%  [k] xfs_agino_range
  12.68%  [k] __xfs_dir3_data_check
   6.95%  [k] xfs_verify_ino
   6.78%  [k] xfs_dir2_data_entry_tag_p
   3.56%  [k] xfs_buf_find
   2.31%  [k] xfs_verify_dir_ino
   2.02%  [k] xfs_dabuf_map.constprop.0
   1.65%  [k] xfs_ag_block_count

And takes around 13 minutes to remove 50 million inodes.

Machine B:

  13.90%  [k] __pv_queued_spin_lock_slowpath
   3.76%  [k] do_raw_spin_lock
   2.83%  [k] xfs_dir3_leaf_check_int
   2.75%  [k] xfs_agino_range
   2.51%  [k] __raw_callee_save___pv_queued_spin_unlock
   2.18%  [k] __xfs_dir3_data_check
   2.02%  [k] xfs_log_commit_cil

And takes around 5m30s to remove 50 million inodes.

Suspect is cacheline contention on m_sectbb_log which is used in one of the macros in xfs_agino_range. This is a read-only variable but shares a cacheline with m_active_trans which is a global atomic that gets bounced all around the machine.

The workload is trying to run hundreds of thousands of transactions per second and hence cacheline contention will be occuring on this atomic counter. Hence xfs_agino_range() is likely just be an innocent bystander as the cache coherency protocol fights over the cacheline between CPU cores and sockets.

On machine A, this rearrangement of the struct xfs_mount results in the profile changing to:

   9.77%  [kernel]  [k] xfs_agino_range
   6.27%  [kernel]  [k] __xfs_dir3_data_check
   5.31%  [kernel]  [k] __pv_queued_spin_lock_slowpath
   4.54%  [kernel]  [k] xfs_buf_find
   3.79%  [kernel]  [k] do_raw_spin_lock
   3.39%  [kernel]  [k] xfs_verify_ino
   2.73%  [kernel]  [k] __raw_callee_save___pv_queued_spin_unlock

Vastly less CPU usage in xfs_agino_range(), but still 3x the amount of machine B and still runs substantially slower than it should.

Current rm -rf of 50 million files:

vanilla patched
machine A 13m20s 6m42s
machine B 5m30s 5m02s

It's an improvement, hence indicating that separation and further optimisation of read-only global filesystem data is worthwhile, but it clearly isn't the underlying issue causing this specific performance degradation.

---
 fs/xfs/xfs_mount.h | 148 +++++++++++++++++++++++++--------------------
 1 file changed, 82 insertions(+), 66 deletions(-)
fs/xfs/xfs_mount.h +82 -66
index 4835581f3eb00..c1f92c1847bb2 100644
@@ -55,61 +55,25 @@ struct xfs_error_cfg {
 	long		retry_timeout;	/* in jiffies, -1 = infinite */
 };
 
+/*
+ * The struct xfsmount layout is optimised to separate read-mostly variables
+ * from variables that are frequently modified. We put the read-mostly variables
+ * first, then place all the other variables at the end.
+ *
+ * Typically, read-mostly variables are those that are set at mount time and
+ * never changed again, or only change rarely as a result of things like sysfs
+ * knobs being tweaked.
+ */
 typedef struct xfs_mount {
+	struct xfs_sb		m_sb;		/* copy of fs superblock */
 	struct super_block	*m_super;
-
-	/*
-	 * Bitsets of per-fs metadata that have been checked and/or are sick.
-	 * Callers must hold m_sb_lock to access these two fields.
-	 */
-	uint8_t			m_fs_checked;
-	uint8_t			m_fs_sick;
-	/*
-	 * Bitsets of rt metadata that have been checked and/or are sick.
-	 * Callers must hold m_sb_lock to access this field.
-	 */
-	uint8_t			m_rt_checked;
-	uint8_t			m_rt_sick;
-
 	struct xfs_ail		*m_ail;		/* fs active log item list */
-
-	struct xfs_sb		m_sb;		/* copy of fs superblock */
-	spinlock_t		m_sb_lock;	/* sb counter lock */
-	struct percpu_counter	m_icount;	/* allocated inodes counter */
-	struct percpu_counter	m_ifree;	/* free inodes counter */
-	struct percpu_counter	m_fdblocks;	/* free block counter */
-	/*
-	 * Count of data device blocks reserved for delayed allocations,
-	 * including indlen blocks.  Does not include allocated CoW staging
-	 * extents or anything related to the rt device.
-	 */
-	struct percpu_counter	m_delalloc_blks;
-
 	struct xfs_buf		*m_sb_bp;	/* buffer for superblock */
 	char			*m_rtname;	/* realtime device name */
 	char			*m_logname;	/* external log device name */
-	int			m_bsize;	/* fs logical block size */
-	xfs_agnumber_t		m_agfrotor;	/* last ag where space found */
-	xfs_agnumber_t		m_agirotor;	/* last ag dir inode alloced */
-	spinlock_t		m_agirotor_lock;/* .. and lock protecting it */
-	xfs_agnumber_t		m_maxagi;	/* highest inode alloc group */
-	uint			m_allocsize_log;/* min write size log bytes */
-	uint			m_allocsize_blocks; /* min write size blocks */
 	struct xfs_da_geometry	*m_dir_geo;	/* directory block geometry */
 	struct xfs_da_geometry	*m_attr_geo;	/* attribute block geometry */
 	struct xlog		*m_log;		/* log specific stuff */
-	struct xfs_ino_geometry	m_ino_geo;	/* inode geometry */
-	int			m_logbufs;	/* number of log buffers */
-	int			m_logbsize;	/* size of each log buffer */
-	uint			m_rsumlevels;	/* rt summary levels */
-	uint			m_rsumsize;	/* size of rt summary, bytes */
-	/*
-	 * Optional cache of rt summary level per bitmap block with the
-	 * invariant that m_rsum_cache[bbno] <= the minimum i for which
-	 * rsum[i][bbno] != 0. Reads and writes are serialized by the rsumip
-	 * inode lock.
-	 */
-	uint8_t			*m_rsum_cache;
 	struct xfs_inode	*m_rbmip;	/* pointer to bitmap inode */
 	struct xfs_inode	*m_rsumip;	/* pointer to summary inode */
 	struct xfs_inode	*m_rootip;	/* pointer to root directory */
@@ -117,9 +81,26 @@ typedef struct xfs_mount {
 	xfs_buftarg_t		*m_ddev_targp;	/* saves taking the address */
 	xfs_buftarg_t		*m_logdev_targp;/* ptr to log device */
 	xfs_buftarg_t		*m_rtdev_targp;	/* ptr to rt device */
+	/*
+	 * Optional cache of rt summary level per bitmap block with the
+	 * invariant that m_rsum_cache[bbno] <= the minimum i for which
+	 * rsum[i][bbno] != 0. Reads and writes are serialized by the rsumip
+	 * inode lock.
+	 */
+	uint8_t			*m_rsum_cache;
+	struct xfs_mru_cache	*m_filestream;  /* per-mount filestream data */
+	struct workqueue_struct *m_buf_workqueue;
+	struct workqueue_struct	*m_unwritten_workqueue;
+	struct workqueue_struct	*m_cil_workqueue;
+	struct workqueue_struct	*m_reclaim_workqueue;
+	struct workqueue_struct *m_eofblocks_workqueue;
+	struct workqueue_struct	*m_sync_workqueue;
+
+	int			m_bsize;	/* fs logical block size */
 	uint8_t			m_blkbit_log;	/* blocklog + NBBY */
 	uint8_t			m_blkbb_log;	/* blocklog - BBSHIFT */
 	uint8_t			m_agno_log;	/* log #ag's */
+	uint8_t			m_sectbb_log;	/* sectlog - BBSHIFT */
 	uint			m_blockmask;	/* sb_blocksize-1 */
 	uint			m_blockwsize;	/* sb_blocksize in words */
 	uint			m_blockwmask;	/* blockwsize-1 */
@@ -138,47 +119,83 @@ typedef struct xfs_mount {
 	xfs_extlen_t		m_ag_prealloc_blocks; /* reserved ag blocks */
 	uint			m_alloc_set_aside; /* space we can't use */
 	uint			m_ag_max_usable; /* max space per AG */
-	struct radix_tree_root	m_perag_tree;	/* per-ag accounting info */
-	spinlock_t		m_perag_lock;	/* lock for m_perag_tree */
-	struct mutex		m_growlock;	/* growfs mutex */
+	int			m_dalign;	/* stripe unit */
+	int			m_swidth;	/* stripe width */
+	xfs_agnumber_t		m_maxagi;	/* highest inode alloc group */
+	uint			m_allocsize_log;/* min write size log bytes */
+	uint			m_allocsize_blocks; /* min write size blocks */
+	int			m_logbufs;	/* number of log buffers */
+	int			m_logbsize;	/* size of each log buffer */
+	uint			m_rsumlevels;	/* rt summary levels */
+	uint			m_rsumsize;	/* size of rt summary, bytes */
 	int			m_fixedfsid[2];	/* unchanged for life of FS */
-	uint64_t		m_flags;	/* global mount flags */
-	bool			m_finobt_nores; /* no per-AG finobt resv. */
 	uint			m_qflags;	/* quota status flags */
+	uint64_t		m_flags;	/* global mount flags */
+	int64_t			m_low_space[XFS_LOWSP_MAX];
+	struct xfs_ino_geometry	m_ino_geo;	/* inode geometry */
 	struct xfs_trans_resv	m_resv;		/* precomputed res values */
+						/* low free space thresholds */
+	bool			m_always_cow;
+	bool			m_fail_unmount;
+	bool			m_finobt_nores; /* no per-AG finobt resv. */
+	bool			m_update_sb;	/* sb needs update in mount */
+
+	/*
+	 * Bitsets of per-fs metadata that have been checked and/or are sick.
+	 * Callers must hold m_sb_lock to access these two fields.
+	 */
+	uint8_t			m_fs_checked;
+	uint8_t			m_fs_sick;
+	/*
+	 * Bitsets of rt metadata that have been checked and/or are sick.
+	 * Callers must hold m_sb_lock to access this field.
+	 */
+	uint8_t			m_rt_checked;
+	uint8_t			m_rt_sick;
+
+	/*
+	 * End of read-mostly variables. Frequently written variables and locks
+	 * should be placed below this comment from now on. The first variable
+	 * here is marked as cacheline aligned so they it is separated from
+	 * the read-mostly variables.
+	 */
+
+	spinlock_t ____cacheline_aligned m_sb_lock; /* sb counter lock */
+	struct percpu_counter	m_icount;	/* allocated inodes counter */
+	struct percpu_counter	m_ifree;	/* free inodes counter */
+	struct percpu_counter	m_fdblocks;	/* free block counter */
+	/*
+	 * Count of data device blocks reserved for delayed allocations,
+	 * including indlen blocks.  Does not include allocated CoW staging
+	 * extents or anything related to the rt device.
+	 */
+	struct percpu_counter	m_delalloc_blks;
+
+	struct radix_tree_root	m_perag_tree;	/* per-ag accounting info */
+	spinlock_t		m_perag_lock;	/* lock for m_perag_tree */
 	uint64_t		m_resblks;	/* total reserved blocks */
 	uint64_t		m_resblks_avail;/* available reserved blocks */
 	uint64_t		m_resblks_save;	/* reserved blks @ remount,ro */
-	int			m_dalign;	/* stripe unit */
-	int			m_swidth;	/* stripe width */
-	uint8_t			m_sectbb_log;	/* sectlog - BBSHIFT */
 	atomic_t		m_active_trans;	/* number trans frozen */
-	struct xfs_mru_cache	*m_filestream;  /* per-mount filestream data */
 	struct delayed_work	m_reclaim_work;	/* background inode reclaim */
 	struct delayed_work	m_eofblocks_work; /* background eof blocks
 						     trimming */
 	struct delayed_work	m_cowblocks_work; /* background cow blocks
 						     trimming */
-	bool			m_update_sb;	/* sb needs update in mount */
-	int64_t			m_low_space[XFS_LOWSP_MAX];
-						/* low free space thresholds */
 	struct xfs_kobj		m_kobj;
 	struct xfs_kobj		m_error_kobj;
 	struct xfs_kobj		m_error_meta_kobj;
 	struct xfs_error_cfg	m_error_cfg[XFS_ERR_CLASS_MAX][XFS_ERR_ERRNO_MAX];
 	struct xstats		m_stats;	/* per-fs stats */
+	xfs_agnumber_t		m_agfrotor;	/* last ag where space found */
+	xfs_agnumber_t		m_agirotor;	/* last ag dir inode alloced */
+	spinlock_t		m_agirotor_lock;/* .. and lock protecting it */
 
 	/*
 	 * Workqueue item so that we can coalesce multiple inode flush attempts
 	 * into a single flush.
 	 */
 	struct work_struct	m_flush_inodes_work;
-	struct workqueue_struct *m_buf_workqueue;
-	struct workqueue_struct	*m_unwritten_workqueue;
-	struct workqueue_struct	*m_cil_workqueue;
-	struct workqueue_struct	*m_reclaim_workqueue;
-	struct workqueue_struct *m_eofblocks_workqueue;
-	struct workqueue_struct	*m_sync_workqueue;
 
 	/*
 	 * Generation of the filesysyem layout.  This is incremented by each
@@ -190,9 +207,8 @@ typedef struct xfs_mount {
 	 * to various other kinds of pain inflicted on the pNFS server.
 	 */
 	uint32_t		m_generation;
+	struct mutex		m_growlock;	/* growfs mutex */
 
-	bool			m_always_cow;
-	bool			m_fail_unmount;
 #ifdef DEBUG
 	/*
 	 * Frequency with which errors are injected.  Replaces xfs_etest; the
Signature · 1 line
2.26.2.761.g0e0b3e54be
#3 lore raw

Re: [PATCH 1/2] xfs: separate read-only variables in struct xfs_mount

Quoted 4 lines · Christoph Hellwig

Shouldn't m_errortag and m_errortag_kobj also go into the read-mostly section?

Otherwise looks good:

kobjs are reference counted and full of random stuff like lists, sysfs references, etc, so they aren't obviously read-mostly variables. I left all the kobjs in the write section for this reason. The errortag stuff is also debug only code so I didn't so I didn't bother touching it.

Cheers,

Dave.

Signature · 2 lines
Dave Chinner
[email protected]
#4 lore raw

Re: [PATCH 1/2] xfs: separate read-only variables in struct xfs_mount

In reply to Dave Chinner
Quoted 5 lines · Dave Chinner

From: Dave Chinner<[email protected]>

Seeing massive cpu usage from xfs_agino_range() on one machine; instruction level profiles look similar to another machine running the same workload, only one machien is consuming 10x as much CPU as

's/machien/machine/', can be done at the time of applying patch.

Quoted 29 lines

the other and going much slower. The only real difference between the two machines is core count per socket. Both are running identical 16p/16GB virtual machine configurations

Machine A:

   25.83%  [k] xfs_agino_range
   12.68%  [k] __xfs_dir3_data_check
    6.95%  [k] xfs_verify_ino
    6.78%  [k] xfs_dir2_data_entry_tag_p
    3.56%  [k] xfs_buf_find
    2.31%  [k] xfs_verify_dir_ino
    2.02%  [k] xfs_dabuf_map.constprop.0
    1.65%  [k] xfs_ag_block_count

And takes around 13 minutes to remove 50 million inodes.

Machine B:

   13.90%  [k] __pv_queued_spin_lock_slowpath
    3.76%  [k] do_raw_spin_lock
    2.83%  [k] xfs_dir3_leaf_check_int
    2.75%  [k] xfs_agino_range
    2.51%  [k] __raw_callee_save___pv_queued_spin_unlock
    2.18%  [k] __xfs_dir3_data_check
    2.02%  [k] xfs_log_commit_cil

And takes around 5m30s to remove 50 million inodes.

Quoted 7 lines

Suspect is cacheline contention on m_sectbb_log which is used in one of the macros in xfs_agino_range. This is a read-only variable but shares a cacheline with m_active_trans which is a global atomic that gets bounced all around the machine.

The workload is trying to run hundreds of thousands of transactions per second and hence cacheline contention will be occuring on this

's/occuring/occurring/', can be done at the time of applying patch.

Quoted 31 lines

atomic counter. Hence xfs_agino_range() is likely just be an innocent bystander as the cache coherency protocol fights over the cacheline between CPU cores and sockets.

On machine A, this rearrangement of the struct xfs_mount results in the profile changing to:

    9.77%  [kernel]  [k] xfs_agino_range
    6.27%  [kernel]  [k] __xfs_dir3_data_check
    5.31%  [kernel]  [k] __pv_queued_spin_lock_slowpath
    4.54%  [kernel]  [k] xfs_buf_find
    3.79%  [kernel]  [k] do_raw_spin_lock
    3.39%  [kernel]  [k] xfs_verify_ino
    2.73%  [kernel]  [k] __raw_callee_save___pv_queued_spin_unlock

Vastly less CPU usage in xfs_agino_range(), but still 3x the amount of machine B and still runs substantially slower than it should.

Current rm -rf of 50 million files:

vanilla patched
machine A 13m20s 6m42s
machine B 5m30s 5m02s

It's an improvement, hence indicating that separation and further optimisation of read-only global filesystem data is worthwhile, but it clearly isn't the underlying issue causing this specific performance degradation.

---

Reviewed-by on v3 1/2
#5 lore raw

Re: [PATCH 1/2] xfs: separate read-only variables in struct xfs_mount

In reply to Dave Chinner
Quoted 65 lines · Dave Chinner

From: Dave Chinner <[email protected]>

Seeing massive cpu usage from xfs_agino_range() on one machine; instruction level profiles look similar to another machine running the same workload, only one machien is consuming 10x as much CPU as the other and going much slower. The only real difference between the two machines is core count per socket. Both are running identical 16p/16GB virtual machine configurations

Machine A:

  25.83%  [k] xfs_agino_range
  12.68%  [k] __xfs_dir3_data_check
   6.95%  [k] xfs_verify_ino
   6.78%  [k] xfs_dir2_data_entry_tag_p
   3.56%  [k] xfs_buf_find
   2.31%  [k] xfs_verify_dir_ino
   2.02%  [k] xfs_dabuf_map.constprop.0
   1.65%  [k] xfs_ag_block_count

And takes around 13 minutes to remove 50 million inodes.

Machine B:

  13.90%  [k] __pv_queued_spin_lock_slowpath
   3.76%  [k] do_raw_spin_lock
   2.83%  [k] xfs_dir3_leaf_check_int
   2.75%  [k] xfs_agino_range
   2.51%  [k] __raw_callee_save___pv_queued_spin_unlock
   2.18%  [k] __xfs_dir3_data_check
   2.02%  [k] xfs_log_commit_cil

And takes around 5m30s to remove 50 million inodes.

Suspect is cacheline contention on m_sectbb_log which is used in one of the macros in xfs_agino_range. This is a read-only variable but shares a cacheline with m_active_trans which is a global atomic that gets bounced all around the machine.

The workload is trying to run hundreds of thousands of transactions per second and hence cacheline contention will be occuring on this atomic counter. Hence xfs_agino_range() is likely just be an innocent bystander as the cache coherency protocol fights over the cacheline between CPU cores and sockets.

On machine A, this rearrangement of the struct xfs_mount results in the profile changing to:

   9.77%  [kernel]  [k] xfs_agino_range
   6.27%  [kernel]  [k] __xfs_dir3_data_check
   5.31%  [kernel]  [k] __pv_queued_spin_lock_slowpath
   4.54%  [kernel]  [k] xfs_buf_find
   3.79%  [kernel]  [k] do_raw_spin_lock
   3.39%  [kernel]  [k] xfs_verify_ino
   2.73%  [kernel]  [k] __raw_callee_save___pv_queued_spin_unlock

Vastly less CPU usage in xfs_agino_range(), but still 3x the amount of machine B and still runs substantially slower than it should.

Current rm -rf of 50 million files:

vanilla patched
machine A 13m20s 6m42s
machine B 5m30s 5m02s

Quoted 6 lines

It's an improvement, hence indicating that separation and further optimisation of read-only global filesystem data is worthwhile, but it clearly isn't the underlying issue causing this specific performance degradation.

Looks ok,

--D

Quoted 223 lines
---
 fs/xfs/xfs_mount.h | 148 +++++++++++++++++++++++++--------------------
 1 file changed, 82 insertions(+), 66 deletions(-)
fs/xfs/xfs_mount.h +82 -66
index 4835581f3eb00..c1f92c1847bb2 100644
@@ -55,61 +55,25 @@ struct xfs_error_cfg {
 	long		retry_timeout;	/* in jiffies, -1 = infinite */
 };
 
+/*
+ * The struct xfsmount layout is optimised to separate read-mostly variables
+ * from variables that are frequently modified. We put the read-mostly variables
+ * first, then place all the other variables at the end.
+ *
+ * Typically, read-mostly variables are those that are set at mount time and
+ * never changed again, or only change rarely as a result of things like sysfs
+ * knobs being tweaked.
+ */
 typedef struct xfs_mount {
+	struct xfs_sb		m_sb;		/* copy of fs superblock */
 	struct super_block	*m_super;
-
-	/*
-	 * Bitsets of per-fs metadata that have been checked and/or are sick.
-	 * Callers must hold m_sb_lock to access these two fields.
-	 */
-	uint8_t			m_fs_checked;
-	uint8_t			m_fs_sick;
-	/*
-	 * Bitsets of rt metadata that have been checked and/or are sick.
-	 * Callers must hold m_sb_lock to access this field.
-	 */
-	uint8_t			m_rt_checked;
-	uint8_t			m_rt_sick;
-
 	struct xfs_ail		*m_ail;		/* fs active log item list */
-
-	struct xfs_sb		m_sb;		/* copy of fs superblock */
-	spinlock_t		m_sb_lock;	/* sb counter lock */
-	struct percpu_counter	m_icount;	/* allocated inodes counter */
-	struct percpu_counter	m_ifree;	/* free inodes counter */
-	struct percpu_counter	m_fdblocks;	/* free block counter */
-	/*
-	 * Count of data device blocks reserved for delayed allocations,
-	 * including indlen blocks.  Does not include allocated CoW staging
-	 * extents or anything related to the rt device.
-	 */
-	struct percpu_counter	m_delalloc_blks;
-
 	struct xfs_buf		*m_sb_bp;	/* buffer for superblock */
 	char			*m_rtname;	/* realtime device name */
 	char			*m_logname;	/* external log device name */
-	int			m_bsize;	/* fs logical block size */
-	xfs_agnumber_t		m_agfrotor;	/* last ag where space found */
-	xfs_agnumber_t		m_agirotor;	/* last ag dir inode alloced */
-	spinlock_t		m_agirotor_lock;/* .. and lock protecting it */
-	xfs_agnumber_t		m_maxagi;	/* highest inode alloc group */
-	uint			m_allocsize_log;/* min write size log bytes */
-	uint			m_allocsize_blocks; /* min write size blocks */
 	struct xfs_da_geometry	*m_dir_geo;	/* directory block geometry */
 	struct xfs_da_geometry	*m_attr_geo;	/* attribute block geometry */
 	struct xlog		*m_log;		/* log specific stuff */
-	struct xfs_ino_geometry	m_ino_geo;	/* inode geometry */
-	int			m_logbufs;	/* number of log buffers */
-	int			m_logbsize;	/* size of each log buffer */
-	uint			m_rsumlevels;	/* rt summary levels */
-	uint			m_rsumsize;	/* size of rt summary, bytes */
-	/*
-	 * Optional cache of rt summary level per bitmap block with the
-	 * invariant that m_rsum_cache[bbno] <= the minimum i for which
-	 * rsum[i][bbno] != 0. Reads and writes are serialized by the rsumip
-	 * inode lock.
-	 */
-	uint8_t			*m_rsum_cache;
 	struct xfs_inode	*m_rbmip;	/* pointer to bitmap inode */
 	struct xfs_inode	*m_rsumip;	/* pointer to summary inode */
 	struct xfs_inode	*m_rootip;	/* pointer to root directory */
@@ -117,9 +81,26 @@ typedef struct xfs_mount {
 	xfs_buftarg_t		*m_ddev_targp;	/* saves taking the address */
 	xfs_buftarg_t		*m_logdev_targp;/* ptr to log device */
 	xfs_buftarg_t		*m_rtdev_targp;	/* ptr to rt device */
+	/*
+	 * Optional cache of rt summary level per bitmap block with the
+	 * invariant that m_rsum_cache[bbno] <= the minimum i for which
+	 * rsum[i][bbno] != 0. Reads and writes are serialized by the rsumip
+	 * inode lock.
+	 */
+	uint8_t			*m_rsum_cache;
+	struct xfs_mru_cache	*m_filestream;  /* per-mount filestream data */
+	struct workqueue_struct *m_buf_workqueue;
+	struct workqueue_struct	*m_unwritten_workqueue;
+	struct workqueue_struct	*m_cil_workqueue;
+	struct workqueue_struct	*m_reclaim_workqueue;
+	struct workqueue_struct *m_eofblocks_workqueue;
+	struct workqueue_struct	*m_sync_workqueue;
+
+	int			m_bsize;	/* fs logical block size */
 	uint8_t			m_blkbit_log;	/* blocklog + NBBY */
 	uint8_t			m_blkbb_log;	/* blocklog - BBSHIFT */
 	uint8_t			m_agno_log;	/* log #ag's */
+	uint8_t			m_sectbb_log;	/* sectlog - BBSHIFT */
 	uint			m_blockmask;	/* sb_blocksize-1 */
 	uint			m_blockwsize;	/* sb_blocksize in words */
 	uint			m_blockwmask;	/* blockwsize-1 */
@@ -138,47 +119,83 @@ typedef struct xfs_mount {
 	xfs_extlen_t		m_ag_prealloc_blocks; /* reserved ag blocks */
 	uint			m_alloc_set_aside; /* space we can't use */
 	uint			m_ag_max_usable; /* max space per AG */
-	struct radix_tree_root	m_perag_tree;	/* per-ag accounting info */
-	spinlock_t		m_perag_lock;	/* lock for m_perag_tree */
-	struct mutex		m_growlock;	/* growfs mutex */
+	int			m_dalign;	/* stripe unit */
+	int			m_swidth;	/* stripe width */
+	xfs_agnumber_t		m_maxagi;	/* highest inode alloc group */
+	uint			m_allocsize_log;/* min write size log bytes */
+	uint			m_allocsize_blocks; /* min write size blocks */
+	int			m_logbufs;	/* number of log buffers */
+	int			m_logbsize;	/* size of each log buffer */
+	uint			m_rsumlevels;	/* rt summary levels */
+	uint			m_rsumsize;	/* size of rt summary, bytes */
 	int			m_fixedfsid[2];	/* unchanged for life of FS */
-	uint64_t		m_flags;	/* global mount flags */
-	bool			m_finobt_nores; /* no per-AG finobt resv. */
 	uint			m_qflags;	/* quota status flags */
+	uint64_t		m_flags;	/* global mount flags */
+	int64_t			m_low_space[XFS_LOWSP_MAX];
+	struct xfs_ino_geometry	m_ino_geo;	/* inode geometry */
 	struct xfs_trans_resv	m_resv;		/* precomputed res values */
+						/* low free space thresholds */
+	bool			m_always_cow;
+	bool			m_fail_unmount;
+	bool			m_finobt_nores; /* no per-AG finobt resv. */
+	bool			m_update_sb;	/* sb needs update in mount */
+
+	/*
+	 * Bitsets of per-fs metadata that have been checked and/or are sick.
+	 * Callers must hold m_sb_lock to access these two fields.
+	 */
+	uint8_t			m_fs_checked;
+	uint8_t			m_fs_sick;
+	/*
+	 * Bitsets of rt metadata that have been checked and/or are sick.
+	 * Callers must hold m_sb_lock to access this field.
+	 */
+	uint8_t			m_rt_checked;
+	uint8_t			m_rt_sick;
+
+	/*
+	 * End of read-mostly variables. Frequently written variables and locks
+	 * should be placed below this comment from now on. The first variable
+	 * here is marked as cacheline aligned so they it is separated from
+	 * the read-mostly variables.
+	 */
+
+	spinlock_t ____cacheline_aligned m_sb_lock; /* sb counter lock */
+	struct percpu_counter	m_icount;	/* allocated inodes counter */
+	struct percpu_counter	m_ifree;	/* free inodes counter */
+	struct percpu_counter	m_fdblocks;	/* free block counter */
+	/*
+	 * Count of data device blocks reserved for delayed allocations,
+	 * including indlen blocks.  Does not include allocated CoW staging
+	 * extents or anything related to the rt device.
+	 */
+	struct percpu_counter	m_delalloc_blks;
+
+	struct radix_tree_root	m_perag_tree;	/* per-ag accounting info */
+	spinlock_t		m_perag_lock;	/* lock for m_perag_tree */
 	uint64_t		m_resblks;	/* total reserved blocks */
 	uint64_t		m_resblks_avail;/* available reserved blocks */
 	uint64_t		m_resblks_save;	/* reserved blks @ remount,ro */
-	int			m_dalign;	/* stripe unit */
-	int			m_swidth;	/* stripe width */
-	uint8_t			m_sectbb_log;	/* sectlog - BBSHIFT */
 	atomic_t		m_active_trans;	/* number trans frozen */
-	struct xfs_mru_cache	*m_filestream;  /* per-mount filestream data */
 	struct delayed_work	m_reclaim_work;	/* background inode reclaim */
 	struct delayed_work	m_eofblocks_work; /* background eof blocks
 						     trimming */
 	struct delayed_work	m_cowblocks_work; /* background cow blocks
 						     trimming */
-	bool			m_update_sb;	/* sb needs update in mount */
-	int64_t			m_low_space[XFS_LOWSP_MAX];
-						/* low free space thresholds */
 	struct xfs_kobj		m_kobj;
 	struct xfs_kobj		m_error_kobj;
 	struct xfs_kobj		m_error_meta_kobj;
 	struct xfs_error_cfg	m_error_cfg[XFS_ERR_CLASS_MAX][XFS_ERR_ERRNO_MAX];
 	struct xstats		m_stats;	/* per-fs stats */
+	xfs_agnumber_t		m_agfrotor;	/* last ag where space found */
+	xfs_agnumber_t		m_agirotor;	/* last ag dir inode alloced */
+	spinlock_t		m_agirotor_lock;/* .. and lock protecting it */
 
 	/*
 	 * Workqueue item so that we can coalesce multiple inode flush attempts
 	 * into a single flush.
 	 */
 	struct work_struct	m_flush_inodes_work;
-	struct workqueue_struct *m_buf_workqueue;
-	struct workqueue_struct	*m_unwritten_workqueue;
-	struct workqueue_struct	*m_cil_workqueue;
-	struct workqueue_struct	*m_reclaim_workqueue;
-	struct workqueue_struct *m_eofblocks_workqueue;
-	struct workqueue_struct	*m_sync_workqueue;
 
 	/*
 	 * Generation of the filesysyem layout.  This is incremented by each
@@ -190,9 +207,8 @@ typedef struct xfs_mount {
 	 * to various other kinds of pain inflicted on the pNFS server.
 	 */
 	uint32_t		m_generation;
+	struct mutex		m_growlock;	/* growfs mutex */
 
-	bool			m_always_cow;
-	bool			m_fail_unmount;
 #ifdef DEBUG
 	/*
 	 * Frequency with which errors are injected.  Replaces xfs_etest; the
Signature · 1 line
2.26.2.761.g0e0b3e54be
#6 lore raw

PATCH v3 2/2 xfs: remove the m_active_trans counter

In reply to Dave Chinner

From: Dave Chinner <[email protected]>

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.

---
 fs/xfs/xfs_mount.h |  1 -
 fs/xfs/xfs_super.c | 17 +++++------------
 fs/xfs/xfs_trans.c | 27 +++++++++++----------------
 3 files changed, 16 insertions(+), 29 deletions(-)
fs/xfs/xfs_mount.h +0 -1
index c1f92c1847bb2..3725d25ad97e8 100644
@@ -176,7 +176,6 @@ typedef struct xfs_mount {
 	uint64_t		m_resblks;	/* total reserved blocks */
 	uint64_t		m_resblks_avail;/* available reserved blocks */
 	uint64_t		m_resblks_save;	/* reserved blks @ remount,ro */
-	atomic_t		m_active_trans;	/* number trans frozen */
 	struct delayed_work	m_reclaim_work;	/* background inode reclaim */
 	struct delayed_work	m_eofblocks_work; /* background eof blocks
 						     trimming */
fs/xfs/xfs_super.c +5 -12
index aae469f73efeb..fa58cb07c8fdf 100644
@@ -874,8 +874,10 @@ xfs_restore_resvblks(struct xfs_mount *mp)
  * there is no log replay required to write the inodes to disk - this is the
  * primary difference between a sync and a quiesce.
  *
- * Note: xfs_log_quiesce() stops background log work - the callers must ensure
- * it is started again when appropriate.
+ * We cancel log work early here to ensure all transactions the log worker may
+ * run have finished before we clean up and log the superblock and write an
+ * unmount record. The unfreeze process is responsible for restarting the log
+ * worker correctly.
  */
 void
 xfs_quiesce_attr(
@@ -883,9 +885,7 @@ xfs_quiesce_attr(
 {
 	int	error = 0;
 
-	/* wait for all modifications to complete */
-	while (atomic_read(&mp->m_active_trans) > 0)
-		delay(100);
+	cancel_delayed_work_sync(&mp->m_log->l_work);
 
 	/* force the log to unpin objects from the now complete transactions */
 	xfs_log_force(mp, XFS_LOG_SYNC);
@@ -899,12 +899,6 @@ xfs_quiesce_attr(
 	if (error)
 		xfs_warn(mp, "xfs_attr_quiesce: failed to log sb changes. "
 				"Frozen image may not be consistent.");
-	/*
-	 * Just warn here till VFS can correctly support
-	 * read-only remount without racing.
-	 */
-	WARN_ON(atomic_read(&mp->m_active_trans) != 0);
-
 	xfs_log_quiesce(mp);
 }
 
@@ -1793,7 +1787,6 @@ static int xfs_init_fs_context(
 	INIT_RADIX_TREE(&mp->m_perag_tree, GFP_ATOMIC);
 	spin_lock_init(&mp->m_perag_lock);
 	mutex_init(&mp->m_growlock);
-	atomic_set(&mp->m_active_trans, 0);
 	INIT_WORK(&mp->m_flush_inodes_work, xfs_flush_inodes_worker);
 	INIT_DELAYED_WORK(&mp->m_reclaim_work, xfs_reclaim_worker);
 	INIT_DELAYED_WORK(&mp->m_eofblocks_work, xfs_eofblocks_worker);
fs/xfs/xfs_trans.c +11 -16
index b055a5ab53465..217937d743dbb 100644
@@ -68,7 +68,6 @@ xfs_trans_free(
 	xfs_extent_busy_clear(tp->t_mountp, &tp->t_busy, false);
 
 	trace_xfs_trans_free(tp, _RET_IP_);
-	atomic_dec(&tp->t_mountp->m_active_trans);
 	if (!(tp->t_flags & XFS_TRANS_NO_WRITECOUNT))
 		sb_end_intwrite(tp->t_mountp->m_super);
 	xfs_trans_free_dqinfo(tp);
@@ -125,8 +124,6 @@ xfs_trans_dup(
 	xfs_defer_move(ntp, tp);
 
 	xfs_trans_dup_dqinfo(tp, ntp);
-
-	atomic_inc(&tp->t_mountp->m_active_trans);
 	return ntp;
 }
 
@@ -275,7 +272,6 @@ xfs_trans_alloc(
 	 */
 	WARN_ON(resp->tr_logres > 0 &&
 		mp->m_super->s_writers.frozen == SB_FREEZE_COMPLETE);
-	atomic_inc(&mp->m_active_trans);
 
 	tp->t_magic = XFS_TRANS_HEADER_MAGIC;
 	tp->t_flags = flags;
@@ -299,20 +295,19 @@ xfs_trans_alloc(
 
 /*
  * Create an empty transaction with no reservation.  This is a defensive
- * 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.
+ * 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.
  *
- * Note the zero-length reservation; this transaction MUST be cancelled
- * without any dirty data.
+ * Note the zero-length reservation; this transaction MUST be cancelled without
+ * any dirty data.
  *
- * Callers should obtain freeze protection to avoid two conflicts with fs
- * freezing: (1) having active transactions trip the m_active_trans ASSERTs;
- * and (2) grabbing buffers at the same time that freeze is trying to drain
- * the buffer LRU list.
+ * 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(
Signature · 1 line
2.26.2.761.g0e0b3e54be
#7 lore raw

Re: [PATCH 2/2] xfs: remove the m_active_trans counter

Quoted 74 lines · Dave Chinner

From: Dave Chinner <[email protected]>

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.

---
 fs/xfs/xfs_mount.h |  1 -
 fs/xfs/xfs_super.c | 17 +++++------------
 fs/xfs/xfs_trans.c | 27 +++++++++++----------------
 3 files changed, 16 insertions(+), 29 deletions(-)
fs/xfs/xfs_mount.h +0 -1
index c1f92c1847bb2..3725d25ad97e8 100644
@@ -176,7 +176,6 @@ typedef struct xfs_mount {
 	uint64_t		m_resblks;	/* total reserved blocks */
 	uint64_t		m_resblks_avail;/* available reserved blocks */
 	uint64_t		m_resblks_save;	/* reserved blks @ remount,ro */
-	atomic_t		m_active_trans;	/* number trans frozen */
 	struct delayed_work	m_reclaim_work;	/* background inode reclaim */
 	struct delayed_work	m_eofblocks_work; /* background eof blocks
 						     trimming */
fs/xfs/xfs_super.c +4 -0
index aae469f73efeb..fa58cb07c8fdf 100644
@@ -874,8 +874,10 @@ xfs_restore_resvblks(struct xfs_mount *mp)
  * there is no log replay required to write the inodes to disk - this is the
  * primary difference between a sync and a quiesce.
  *
+ * We cancel log work early here to ensure all transactions the log worker may
+ * run have finished before we clean up and log the superblock and write an
+ * unmount record. The unfreeze process is responsible for restarting the log
+ * worker correctly.
  */
 void
 xfs_quiesce_attr(

fs/xfs/xfs_super.c xfs_quiesce_attr( · line 888

Quoted 7 lines
fs/xfs/xfs_super.c +1 -3
@@ -883,9 +885,7 @@ xfs_quiesce_attr(
 {
 	int	error = 0;
 
-	/* wait for all modifications to complete */
-	while (atomic_read(&mp->m_active_trans) > 0)
-		delay(100);
+	cancel_delayed_work_sync(&mp->m_log->l_work);

Shouldn't the cancel_delayed_work_sync for l_work in xfs_log_quiesce be removed now given that we've already cancelled it here?

#8 lore raw

Re: [PATCH 2/2] xfs: remove the m_active_trans counter

Quoted 78 lines · Christoph Hellwig
Quoted 77 lines · Dave Chinner

From: Dave Chinner <[email protected]>

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.

---
 fs/xfs/xfs_mount.h |  1 -
 fs/xfs/xfs_super.c | 17 +++++------------
 fs/xfs/xfs_trans.c | 27 +++++++++++----------------
 3 files changed, 16 insertions(+), 29 deletions(-)
fs/xfs/xfs_mount.h +0 -1
index c1f92c1847bb2..3725d25ad97e8 100644
@@ -176,7 +176,6 @@ typedef struct xfs_mount {
 	uint64_t		m_resblks;	/* total reserved blocks */
 	uint64_t		m_resblks_avail;/* available reserved blocks */
 	uint64_t		m_resblks_save;	/* reserved blks @ remount,ro */
-	atomic_t		m_active_trans;	/* number trans frozen */
 	struct delayed_work	m_reclaim_work;	/* background inode reclaim */
 	struct delayed_work	m_eofblocks_work; /* background eof blocks
 						     trimming */
fs/xfs/xfs_super.c +4 -0
index aae469f73efeb..fa58cb07c8fdf 100644
@@ -874,8 +874,10 @@ xfs_restore_resvblks(struct xfs_mount *mp)
  * there is no log replay required to write the inodes to disk - this is the
  * primary difference between a sync and a quiesce.
  *
+ * We cancel log work early here to ensure all transactions the log worker may
+ * run have finished before we clean up and log the superblock and write an
+ * unmount record. The unfreeze process is responsible for restarting the log
+ * worker correctly.
  */
 void
 xfs_quiesce_attr(
@@ -883,9 +885,7 @@ xfs_quiesce_attr(
 {
 	int	error = 0;
 
Quoted 7 lines

fs/xfs/xfs_super.c xfs_quiesce_attr( · line 888

Quoted 4 lines
fs/xfs/xfs_super.c +1 -3
-	/* wait for all modifications to complete */
-	while (atomic_read(&mp->m_active_trans) > 0)
-		delay(100);
+	cancel_delayed_work_sync(&mp->m_log->l_work);

Shouldn't the cancel_delayed_work_sync for l_work in xfs_log_quiesce be removed now given that we've already cancelled it here?

No, because every other caller of xfs_log_quiesce() requires the work to be cancelled before the log is quiesced.

Cheers,

Dave.

Signature · 2 lines
Dave Chinner
[email protected]
#9 lore raw

Re: [PATCH 2/2] xfs: remove the m_active_trans counter

Quoted 12 lines · Dave Chinner
Quoted 11 lines

fs/xfs/xfs_super.c xfs_quiesce_attr( · line 888

Quoted 8 lines
fs/xfs/xfs_super.c +1 -3
@@ -883,9 +885,7 @@ xfs_quiesce_attr(
 {
 	int	error = 0;
 
-	/* wait for all modifications to complete */
-	while (atomic_read(&mp->m_active_trans) > 0)
-		delay(100);
+	cancel_delayed_work_sync(&mp->m_log->l_work);

Shouldn't the cancel_delayed_work_sync for l_work in xfs_log_quiesce be removed now given that we've already cancelled it here?

Quoted 2 lines

No, because every other caller of xfs_log_quiesce() requires the work to be cancelled before the log is quiesced.

The only other caller is xfs_log_unmount, which could grow it. But I guess an extra cancel isn't harmful in the end.

Reviewed-by on v3 2/2
#10 lore raw

Re: [PATCH 2/2] xfs: remove the m_active_trans counter

In reply to Dave Chinner
Quoted 35 lines · Dave Chinner

From: Dave Chinner <[email protected]>

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

Quoted 5 lines

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.

/me kinda wonders why removing the counter entirely has so little effect on machine B, but seeing as I've been pondering killing this counter myself for years,

--D

Quoted 131 lines
---
 fs/xfs/xfs_mount.h |  1 -
 fs/xfs/xfs_super.c | 17 +++++------------
 fs/xfs/xfs_trans.c | 27 +++++++++++----------------
 3 files changed, 16 insertions(+), 29 deletions(-)
fs/xfs/xfs_mount.h +0 -1
index c1f92c1847bb2..3725d25ad97e8 100644
@@ -176,7 +176,6 @@ typedef struct xfs_mount {
 	uint64_t		m_resblks;	/* total reserved blocks */
 	uint64_t		m_resblks_avail;/* available reserved blocks */
 	uint64_t		m_resblks_save;	/* reserved blks @ remount,ro */
-	atomic_t		m_active_trans;	/* number trans frozen */
 	struct delayed_work	m_reclaim_work;	/* background inode reclaim */
 	struct delayed_work	m_eofblocks_work; /* background eof blocks
 						     trimming */
fs/xfs/xfs_super.c +5 -12
index aae469f73efeb..fa58cb07c8fdf 100644
@@ -874,8 +874,10 @@ xfs_restore_resvblks(struct xfs_mount *mp)
  * there is no log replay required to write the inodes to disk - this is the
  * primary difference between a sync and a quiesce.
  *
- * Note: xfs_log_quiesce() stops background log work - the callers must ensure
- * it is started again when appropriate.
+ * We cancel log work early here to ensure all transactions the log worker may
+ * run have finished before we clean up and log the superblock and write an
+ * unmount record. The unfreeze process is responsible for restarting the log
+ * worker correctly.
  */
 void
 xfs_quiesce_attr(
@@ -883,9 +885,7 @@ xfs_quiesce_attr(
 {
 	int	error = 0;
 
-	/* wait for all modifications to complete */
-	while (atomic_read(&mp->m_active_trans) > 0)
-		delay(100);
+	cancel_delayed_work_sync(&mp->m_log->l_work);
 
 	/* force the log to unpin objects from the now complete transactions */
 	xfs_log_force(mp, XFS_LOG_SYNC);
@@ -899,12 +899,6 @@ xfs_quiesce_attr(
 	if (error)
 		xfs_warn(mp, "xfs_attr_quiesce: failed to log sb changes. "
 				"Frozen image may not be consistent.");
-	/*
-	 * Just warn here till VFS can correctly support
-	 * read-only remount without racing.
-	 */
-	WARN_ON(atomic_read(&mp->m_active_trans) != 0);
-
 	xfs_log_quiesce(mp);
 }
 
@@ -1793,7 +1787,6 @@ static int xfs_init_fs_context(
 	INIT_RADIX_TREE(&mp->m_perag_tree, GFP_ATOMIC);
 	spin_lock_init(&mp->m_perag_lock);
 	mutex_init(&mp->m_growlock);
-	atomic_set(&mp->m_active_trans, 0);
 	INIT_WORK(&mp->m_flush_inodes_work, xfs_flush_inodes_worker);
 	INIT_DELAYED_WORK(&mp->m_reclaim_work, xfs_reclaim_worker);
 	INIT_DELAYED_WORK(&mp->m_eofblocks_work, xfs_eofblocks_worker);
fs/xfs/xfs_trans.c +11 -16
index b055a5ab53465..217937d743dbb 100644
@@ -68,7 +68,6 @@ xfs_trans_free(
 	xfs_extent_busy_clear(tp->t_mountp, &tp->t_busy, false);
 
 	trace_xfs_trans_free(tp, _RET_IP_);
-	atomic_dec(&tp->t_mountp->m_active_trans);
 	if (!(tp->t_flags & XFS_TRANS_NO_WRITECOUNT))
 		sb_end_intwrite(tp->t_mountp->m_super);
 	xfs_trans_free_dqinfo(tp);
@@ -125,8 +124,6 @@ xfs_trans_dup(
 	xfs_defer_move(ntp, tp);
 
 	xfs_trans_dup_dqinfo(tp, ntp);
-
-	atomic_inc(&tp->t_mountp->m_active_trans);
 	return ntp;
 }
 
@@ -275,7 +272,6 @@ xfs_trans_alloc(
 	 */
 	WARN_ON(resp->tr_logres > 0 &&
 		mp->m_super->s_writers.frozen == SB_FREEZE_COMPLETE);
-	atomic_inc(&mp->m_active_trans);
 
 	tp->t_magic = XFS_TRANS_HEADER_MAGIC;
 	tp->t_flags = flags;
@@ -299,20 +295,19 @@ xfs_trans_alloc(
 
 /*
  * Create an empty transaction with no reservation.  This is a defensive
- * 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.
+ * 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.
  *
- * Note the zero-length reservation; this transaction MUST be cancelled
- * without any dirty data.
+ * Note the zero-length reservation; this transaction MUST be cancelled without
+ * any dirty data.
  *
- * Callers should obtain freeze protection to avoid two conflicts with fs
- * freezing: (1) having active transactions trip the m_active_trans ASSERTs;
- * and (2) grabbing buffers at the same time that freeze is trying to drain
- * the buffer LRU list.
+ * 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(
Signature · 1 line
2.26.2.761.g0e0b3e54be
#11 lore raw

Re: [PATCH 2/2] xfs: remove the m_active_trans counter

Quoted 40 lines · Darrick J. Wong
Quoted 39 lines · Dave Chinner

From: Dave Chinner <[email protected]>

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.

Quoted 5 lines
Quoted 1 line

/me kinda wonders why removing the counter entirely has so little effect on machine B, but seeing as I've been pondering killing this counter myself for years,

Because the transaction rate on machine B isn't high enough to that the cacheline bouncing becomes a limiting factor.

Don't forget that the impact of cacheline bouncing is exponential - there is a very small window where it goes from "none at all" to "all the machine", and these two machines sit on either side of that threshold. i.e. the older machine is too slow to hit that threshold, the newer machine hits it easily.

Cheers,

Dave.

Signature · 2 lines
Dave Chinner
[email protected]