Signed-off-by: NChris Mason <chris.mason@oracle.com>

Dead roots are trees left over after a crash, and they were either in the
process of being removed or were waiting to be removed when the box crashed.
Before, a search of the entire tree of root pointers was done on mount
looking for dead roots.  Now, the search is done the first time we load
a root.

This makes mount faster when there are a large number of snapshots, and it
enables the block accounting code to properly update the block counts on
the latest root as old versions of the root are reaped after a crash.
Signed-off-by: NChris Mason <chris.mason@oracle.com>

XFS updates the ondisk inode size only after the data I/O has finished,
so it needs a hook when the writepage end_bio handler has finished.

Might not be worth applying as-is as the per-page callback is very
ineffcient.  What XFS really wants is a callback when writeout of a
whole extent has completed.  This delayed i_size updates scheme might
be worthwile for btrfs aswell, btw.
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Signed-off-by: NChris Mason <chris.mason@oracle.com>

The writepage_io is not mandatory, e.g. my port of xfs to the extent_map
code does not have one for now.  So handle a NULL pointer gracefully
here.
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Signed-off-by: NChris Mason <chris.mason@oracle.com>

generic_bmap is completely trivial, while the extent to bh mapping in
btrfs is rather complex.  So provide a extent_bmap instead that takes
a get_extent callback and can be used by filesystem using the extent_map
code.
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Signed-off-by: NChris Mason <chris.mason@oracle.com>

The bio completion handlers can be run in any context, e.g. when using
the old ide driver they run in hardirq context with irqs disabled so
lockdep rightfully warns about using write_lock_irq useage in these
handlers.

This patch switches clear_extent_bit and set_extent_bit to
write_lock_irqsave to fix this problem.
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Signed-off-by: NChris Mason <chris.mason@oracle.com>

Signed-off-by: NChristoph Hellwig <hch@lst.de>
Signed-off-by: NChris Mason <chris.mason@oracle.com>

Yan Zheng noticed that set_extent_bit was exiting too early when there
was a hole in the map.  The fix is to reorder the tests to check for the
hole first.
Signed-off-by: NChris Mason <chris.mason@oracle.com>

File data checksums are only done during writepage, so we have to make sure
all pages are written when the snapshot is taken.  This also adds some
locking so that new writes don't race in and add new dirty pages.
Signed-off-by: NChris Mason <chris.mason@oracle.com>

defrag.
Signed-off-by: NChris Mason <chris.mason@oracle.com>

Signed-off-by: NChris Mason <chris.mason@oracle.com>

Signed-off-by: NChris Mason <chris.mason@oracle.com>

Signed-off-by: NChris Mason <chris.mason@oracle.com>

Signed-off-by: NChris Mason <chris.mason@oracle.com>

Signed-off-by: NChris Mason <chris.mason@oracle.com>

Modified form of original patch from Christoph Hellwig to make btrfs
mount into the default subvolume by default.

mount /dev/somedevice:subvolumename to get other subvolumes or
mount /dev/somedevice:. to get the root
Signed-off-by: NChris Mason <chris.mason@oracle.com>

Signed-off-by: NChris Mason <chris.mason@oracle.com>

Fixup a few buffer_head release errors, and fix an off by one in
balance_node_right.
Signed-off-by: NChris Mason <chris.mason@oracle.com>

Signed-off-by: NChris Mason <chris.mason@oracle.com>

Signed-off-by: NChris Mason <chris.mason@oracle.com>

instead of buffer heads.
Signed-off-by: NChris Mason <chris.mason@oracle.com>

Signed-off-by: NChris Mason <chris.mason@oracle.com>

Signed-off-by: NChris Mason <chris.mason@oracle.com>

Send the correct slot down to reada_for_search
Signed-off-by: NChris Mason <chris.mason@oracle.com>

Signed-off-by: NChris Mason <chris.mason@oracle.com>

Signed-off-by: NChris Mason <chris.mason@oracle.com>

This allows the tree walking code to defrag only the newly allocated
buffers, it seems to be a good balance between perfect defragging and the
performance hit of repeatedly reallocating blocks.
Signed-off-by: NChris Mason <chris.mason@oracle.com>

Signed-off-by: NChris Mason <chris.mason@oracle.com>

Signed-off-by: NChris Mason <chris.mason@oracle.com>

Signed-off-by: NChris Mason <chris.mason@oracle.com>

Signed-off-by: NChris Mason <chris.mason@oracle.com>

Signed-off-by: NChris Mason <chris.mason@oracle.com>

This adds two types of btree defrag, a run time form that tries to
defrag recently allocated blocks in the btree when they are still in ram,
and an ioctl that forces defrag of all btree blocks.

File data blocks are not defragged yet, but this can make a huge difference
in sequential btree reads.
Signed-off-by: NChris Mason <chris.mason@oracle.com>

Signed-off-by: NChris Mason <chris.mason@oracle.com>

Before, snapshot deletion was a single atomic unit.  This caused considerable
lock contention and required an unbounded amount of space.  Now,
the drop_progress field in the root item is used to indicate how far along
snapshot deletion is, and to resume where it left off.
Signed-off-by: NChris Mason <chris.mason@oracle.com>

Signed-off-by: NChris Mason <chris.mason@oracle.com>

Signed-off-by: NChris Mason <chris.mason@oracle.com>

Signed-off-by: NChris Mason <chris.mason@oracle.com>

Signed-off-by: NChris Mason <chris.mason@oracle.com>

Almost none of the files including module.h need to do so,
remove them.

Include sched.h in extent-tree.c to silence a warning about cond_resched()
being undeclared.
Signed-off-by: NZach Brown <zach.brown@oracle.com>
Signed-off-by: NChris Mason <chris.mason@oracle.com>