https://github.com/torvalds/linux
Revision 9230a0b65b47fe6856c4468ec0175c4987e5bede authored by Dave Chinner on 20 November 2018, 06:50:08 UTC, committed by Darrick J. Wong on 21 November 2018, 18:10:53 UTC
Long saga. There have been days spent following this through dead end
after dead end in multi-GB event traces. This morning, after writing
a trace-cmd wrapper that enabled me to be more selective about XFS
trace points, I discovered that I could get just enough essential
tracepoints enabled that there was a 50:50 chance the fsx config
would fail at ~115k ops. If it didn't fail at op 115547, I stopped
fsx at op 115548 anyway.

That gave me two traces - one where the problem manifested, and one
where it didn't. After refining the traces to have the necessary
information, I found that in the failing case there was a real
extent in the COW fork compared to an unwritten extent in the
working case.

Walking back through the two traces to the point where the CWO fork
extents actually diverged, I found that the bad case had an extra
unwritten extent in it. This is likely because the bug it led me to
had triggered multiple times in those 115k ops, leaving stray
COW extents around. What I saw was a COW delalloc conversion to an
unwritten extent (as they should always be through
xfs_iomap_write_allocate()) resulted in a /written extent/:

xfs_writepage:        dev 259:0 ino 0x83 pgoff 0x17000 size 0x79a00 offset 0 length 0
xfs_iext_remove:      dev 259:0 ino 0x83 state RC|LF|RF|COW cur 0xffff888247b899c0/2 offset 32 block 152 count 20 flag 1 caller xfs_bmap_add_extent_delay_real
xfs_bmap_pre_update:  dev 259:0 ino 0x83 state RC|LF|RF|COW cur 0xffff888247b899c0/1 offset 1 block 4503599627239429 count 31 flag 0 caller xfs_bmap_add_extent_delay_real
xfs_bmap_post_update: dev 259:0 ino 0x83 state RC|LF|RF|COW cur 0xffff888247b899c0/1 offset 1 block 121 count 51 flag 0 caller xfs_bmap_add_ex

Basically, Cow fork before:

	0 1            32          52
	+H+DDDDDDDDDDDD+UUUUUUUUUUU+
	   PREV		RIGHT

COW delalloc conversion allocates:

	  1	       32
	  +uuuuuuuuuuuu+
	  NEW

And the result according to the xfs_bmap_post_update trace was:

	0 1            32          52
	+H+wwwwwwwwwwwwwwwwwwwwwwww+
	   PREV

Which is clearly wrong - it should be a merged unwritten extent,
not an unwritten extent.

That lead me to look at the LEFT_FILLING|RIGHT_FILLING|RIGHT_CONTIG
case in xfs_bmap_add_extent_delay_real(), and sure enough, there's
the bug.

It takes the old delalloc extent (PREV) and adds the length of the
RIGHT extent to it, takes the start block from NEW, removes the
RIGHT extent and then updates PREV with the new extent.

What it fails to do is update PREV.br_state. For delalloc, this is
always XFS_EXT_NORM, while in this case we are converting the
delayed allocation to unwritten, so it needs to be updated to
XFS_EXT_UNWRITTEN. This LF|RF|RC case does not do this, and so
the resultant extent is always written.

And that's the bug I've been chasing for a week - a bmap btree bug,
not a reflink/dedupe/copy_file_range bug, but a BMBT bug introduced
with the recent in core extent tree scalability enhancements.

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>
1 parent 2c30717
Raw File
Tip revision: 9230a0b65b47fe6856c4468ec0175c4987e5bede authored by Dave Chinner on 20 November 2018, 06:50:08 UTC
xfs: delalloc -> unwritten COW fork allocation can go wrong
Tip revision: 9230a0b
blk-stat.h
/* SPDX-License-Identifier: GPL-2.0 */
#ifndef BLK_STAT_H
#define BLK_STAT_H

#include <linux/kernel.h>
#include <linux/blkdev.h>
#include <linux/ktime.h>
#include <linux/rcupdate.h>
#include <linux/timer.h>

/**
 * struct blk_stat_callback - Block statistics callback.
 *
 * A &struct blk_stat_callback is associated with a &struct request_queue. While
 * @timer is active, that queue's request completion latencies are sorted into
 * buckets by @bucket_fn and added to a per-cpu buffer, @cpu_stat. When the
 * timer fires, @cpu_stat is flushed to @stat and @timer_fn is invoked.
 */
struct blk_stat_callback {
	/*
	 * @list: RCU list of callbacks for a &struct request_queue.
	 */
	struct list_head list;

	/**
	 * @timer: Timer for the next callback invocation.
	 */
	struct timer_list timer;

	/**
	 * @cpu_stat: Per-cpu statistics buckets.
	 */
	struct blk_rq_stat __percpu *cpu_stat;

	/**
	 * @bucket_fn: Given a request, returns which statistics bucket it
	 * should be accounted under. Return -1 for no bucket for this
	 * request.
	 */
	int (*bucket_fn)(const struct request *);

	/**
	 * @buckets: Number of statistics buckets.
	 */
	unsigned int buckets;

	/**
	 * @stat: Array of statistics buckets.
	 */
	struct blk_rq_stat *stat;

	/**
	 * @fn: Callback function.
	 */
	void (*timer_fn)(struct blk_stat_callback *);

	/**
	 * @data: Private pointer for the user.
	 */
	void *data;

	struct rcu_head rcu;
};

struct blk_queue_stats *blk_alloc_queue_stats(void);
void blk_free_queue_stats(struct blk_queue_stats *);

void blk_stat_add(struct request *rq, u64 now);

/* record time/size info in request but not add a callback */
void blk_stat_enable_accounting(struct request_queue *q);

/**
 * blk_stat_alloc_callback() - Allocate a block statistics callback.
 * @timer_fn: Timer callback function.
 * @bucket_fn: Bucket callback function.
 * @buckets: Number of statistics buckets.
 * @data: Value for the @data field of the &struct blk_stat_callback.
 *
 * See &struct blk_stat_callback for details on the callback functions.
 *
 * Return: &struct blk_stat_callback on success or NULL on ENOMEM.
 */
struct blk_stat_callback *
blk_stat_alloc_callback(void (*timer_fn)(struct blk_stat_callback *),
			int (*bucket_fn)(const struct request *),
			unsigned int buckets, void *data);

/**
 * blk_stat_add_callback() - Add a block statistics callback to be run on a
 * request queue.
 * @q: The request queue.
 * @cb: The callback.
 *
 * Note that a single &struct blk_stat_callback can only be added to a single
 * &struct request_queue.
 */
void blk_stat_add_callback(struct request_queue *q,
			   struct blk_stat_callback *cb);

/**
 * blk_stat_remove_callback() - Remove a block statistics callback from a
 * request queue.
 * @q: The request queue.
 * @cb: The callback.
 *
 * When this returns, the callback is not running on any CPUs and will not be
 * called again unless readded.
 */
void blk_stat_remove_callback(struct request_queue *q,
			      struct blk_stat_callback *cb);

/**
 * blk_stat_free_callback() - Free a block statistics callback.
 * @cb: The callback.
 *
 * @cb may be NULL, in which case this does nothing. If it is not NULL, @cb must
 * not be associated with a request queue. I.e., if it was previously added with
 * blk_stat_add_callback(), it must also have been removed since then with
 * blk_stat_remove_callback().
 */
void blk_stat_free_callback(struct blk_stat_callback *cb);

/**
 * blk_stat_is_active() - Check if a block statistics callback is currently
 * gathering statistics.
 * @cb: The callback.
 */
static inline bool blk_stat_is_active(struct blk_stat_callback *cb)
{
	return timer_pending(&cb->timer);
}

/**
 * blk_stat_activate_nsecs() - Gather block statistics during a time window in
 * nanoseconds.
 * @cb: The callback.
 * @nsecs: Number of nanoseconds to gather statistics for.
 *
 * The timer callback will be called when the window expires.
 */
static inline void blk_stat_activate_nsecs(struct blk_stat_callback *cb,
					   u64 nsecs)
{
	mod_timer(&cb->timer, jiffies + nsecs_to_jiffies(nsecs));
}

/**
 * blk_stat_activate_msecs() - Gather block statistics during a time window in
 * milliseconds.
 * @cb: The callback.
 * @msecs: Number of milliseconds to gather statistics for.
 *
 * The timer callback will be called when the window expires.
 */
static inline void blk_stat_activate_msecs(struct blk_stat_callback *cb,
					   unsigned int msecs)
{
	mod_timer(&cb->timer, jiffies + msecs_to_jiffies(msecs));
}

void blk_rq_stat_add(struct blk_rq_stat *, u64);
void blk_rq_stat_sum(struct blk_rq_stat *, struct blk_rq_stat *);
void blk_rq_stat_init(struct blk_rq_stat *);

#endif
back to top