When running the following fio job

[torrent]
filename=torrent-test
rw=randwrite
size=4g
filesize=4g
bs=4k
ioengine=sync

you would see long stalls where no work was being done.  That is because we were
doing all this extra work to read in the file extent outside of the transaction,
however in the random io case this ends up hurting us because the file extents
are not there to begin with.  So axe this logic, since we end up reading in the
file extent when we go to update it anyway.  This took the fio job from 11 mb/s
with several ~10 second stalls to 24 mb/s to a couple of 1-2 second stalls.
Signed-off-by: NJosef Bacik <josef@redhat.com>
Signed-off-by: NChris Mason <chris.mason@oracle.com>

This patch revert's commit

6c090a11

Since it introduces this problem where we can run orphan cleanup on a
volume that can have orphan entries re-added.  Instead of my original
fix, Yan Zheng pointed out that we can just revert my original fix and
then run the orphan cleanup in open_ctree after we look up the fs_root.
I have tested this with all the tests that gave me problems and this
patch fixes both problems.  Thanks,
Signed-off-by: NJosef Bacik <josef@redhat.com>
Signed-off-by: NChris Mason <chris.mason@oracle.com>

commit f2bc9dd07e3424c4ec5f3949961fe053d47bc825
Author: Aneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>
Date:   Wed Jan 20 12:57:53 2010 +0530

    Btrfs: Use correct values when updating inode i_size on fallocate

    Even though we allocate more, we should be updating inode i_size
    as per the arguments passed
Signed-off-by: NAneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>
Signed-off-by: NChris Mason <chris.mason@oracle.com>

The default btrfs mount -o compress mode will quickly back off
compressing a file if it notices that compression does not reduce the
size of the data being written.  This can save considerable CPU because
all future writes to the file go through uncompressed.

But some files are both very large and have mixed data stored in
them.  In that case, we want to add the ability to always try
compressing data before writing it.

This commit adds mount -o compress-force.  A later commit will add
a new inode flag that does the same thing.
Signed-off-by: NChris Mason <chris.mason@oracle.com>

Currently orphan cleanup only ever gets triggered if we cross subvolumes during
a lookup, which means that if we just mount a plain jane fs that has orphans in
it, they will never get cleaned up.  This results in panic's like these

http://www.kerneloops.org/oops.php?number=1109085

where adding an orphan entry results in -EEXIST being returned and we panic.  In
order to fix this, we check to see on lookup if our root has had the orphan
cleanup done, and if not go ahead and do it.  This is easily reproduceable by
running this testcase

#include <sys/types.h>
#include <sys/stat.h>
#include <fcntl.h>
#include <string.h>
#include <unistd.h>
#include <stdio.h>

int main(int argc, char **argv)
{
	char data[4096];
	char newdata[4096];
	int fd1, fd2;

	memset(data, 'a', 4096);
	memset(newdata, 'b', 4096);

	while (1) {
		int i;

		fd1 = creat("file1", 0666);
		if (fd1 < 0)
			break;

		for (i = 0; i < 512; i++)
			write(fd1, data, 4096);

		fsync(fd1);
		close(fd1);

		fd2 = creat("file2", 0666);
		if (fd2 < 0)
			break;

		ftruncate(fd2, 4096 * 512);

		for (i = 0; i < 512; i++)
			write(fd2, newdata, 4096);
		close(fd2);

		i = rename("file2", "file1");
		unlink("file1");
	}

	return 0;
}

and then pulling the power on the box, and then trying to run that test again
when the box comes back up.  I've tested this locally and it fixes the problem.
Thanks to Tomas Carnecky for helping me track this down initially.
Signed-off-by: NJosef Bacik <josef@redhat.com>
Signed-off-by: NChris Mason <chris.mason@oracle.com>

parent 49313cdac7b34c9f7ecbb1780cfc648b1c082cd7 (v2.6.32-1-g49313cd)
commit ff48c08e1c05c67e8348ab6f8a24de8034e0e34d
Author: Jan Engelhardt <jengelh@medozas.de>
Date:   Wed Dec 9 22:57:36 2009 +0100

Btrfs: fix missing last-entry in readdir(3)

When one does a 32-bit readdir(3), the last entry of a directory is
missing. This is however not due to passing a large value to filldir,
but it seems to have to do with glibc doing telldir or something
quirky. In any case, this patch fixes it in practice.
Signed-off-by: NJan Engelhardt <jengelh@medozas.de>
Signed-off-by: NChris Mason <chris.mason@oracle.com>

The recent patch to make fallocate enospc friendly would send
down a NULL trans handle to the allocator.  This moves the
transaction start to properly fix things.
Signed-off-by: NChris Mason <chris.mason@oracle.com>

I rebased Christian Parpart's patch to deny hard link across
subvolumes. Original patch modifies also btrfs_rename, but
I excluded it because we can move across subvolumes now and
it make no problem.
-----------------

Hard link across subvolumes should not allowed in Btrfs.
btrfs_link checks root of 'to' directory is same as root
of 'from' file. If not same, btrfs_link returns -EPERM.
Signed-off-by: NTARUISI Hiroaki <taruishi.hiroak@jp.fujitsu.com>
Signed-off-by: NChris Mason <chris.mason@oracle.com>

iput() can trigger new transactions if we are dropping the
final reference, so calling it in btrfs_commit_transaction
may end up deadlock. This patch adds delayed iput to avoid
the issue.
Signed-off-by: NYan Zheng <zheng.yan@oracle.com>
Signed-off-by: NChris Mason <chris.mason@oracle.com>

Pass transaction handle down to security and ACL initialization
functions, so we can avoid starting nested transactions
Signed-off-by: NYan Zheng <zheng.yan@oracle.com>
Signed-off-by: NChris Mason <chris.mason@oracle.com>

truncating and deleting regular files are unbound operations,
so it's not good to do them in a single transaction. This
patch makes btrfs_truncate and btrfs_delete_inode start a
new transaction after all items in a tree leaf are deleted.
Signed-off-by: NYan Zheng <zheng.yan@oracle.com>
Signed-off-by: NChris Mason <chris.mason@oracle.com>

fallocate(2) may allocate large number of file extents, so it's not
good to do it in a single transaction. This patch make fallocate(2)
start a new transaction for each file extents it allocates.
Signed-off-by: NYan Zheng <zheng.yan@oracle.com>
Signed-off-by: NChris Mason <chris.mason@oracle.com>

We do log replay in a single transaction, so it's not good to do unbound
operations. This patch cleans up orphan inodes cleanup after replaying
the log. It also avoids doing other unbound operations such as truncating
a file during replaying log. These unbound operations are postponed to
the orphan inode cleanup stage.
Signed-off-by: NYan Zheng <zheng.yan@oracle.com>
Signed-off-by: NChris Mason <chris.mason@oracle.com>

There are some cases file extents are inserted without involving
ordered struct. In these cases, we update disk_i_size directly,
without checking pending ordered extent and DELALLOC bit. This
patch extends btrfs_ordered_update_i_size() to handle these cases.
Signed-off-by: NYan Zheng <zheng.yan@oracle.com>
Signed-off-by: NChris Mason <chris.mason@oracle.com>

Rewrite btrfs_drop_extents by using btrfs_duplicate_item, so we can
avoid calling lock_extent within transaction.
Signed-off-by: NYan Zheng <zheng.yan@oracle.com>
Signed-off-by: NChris Mason <chris.mason@oracle.com>

There is a problem where iget5_locked will look for an inode, not find it, and
then subsequently try to allocate it.  Another CPU will have raced in and
allocated the inode instead, so when iget5_locked gets the inode spin lock again
and does a search, it finds the new inode.  So it goes ahead and calls
destroy_inode on the inode it just allocated.  The problem is we don't set
BTRFS_I(inode)->root until the new inode is completely initialized.  This patch
makes us set root to NULL when alloc'ing a new inode, so when we get to
btrfs_destroy_inode and we see that root is NULL we can just free up the memory
and continue on.  This fixes the panic

http://www.kerneloops.org/submitresult.php?number=812690

Thanks,
Signed-off-by: NJosef Bacik <josef@redhat.com>
Signed-off-by: NChris Mason <chris.mason@oracle.com>

Currently compressed IO does not deal with not having its entire extent able to
be allocated.  So if we have enough free space to allocate for the extent, but
its not contiguous, it will fail spectacularly.  This patch fixes this by
falling back on uncompressed IO which lets us spread the delalloc extent across
multiple extents.  I tested this by making us randomly think the reservation had
failed to make it fallback on the uncompressed io way and it seemed to work
fine.  Thanks,
Signed-off-by: NJosef Bacik <josef@redhat.com>
Signed-off-by: NChris Mason <chris.mason@oracle.com>

We weren't reserving metadata space for rename, rmdir and unlink, which could
cause problems.
Signed-off-by: NJosef Bacik <josef@redhat.com>
Signed-off-by: NChris Mason <chris.mason@oracle.com>

Sometimes our start allocation hint when we cow a file can be either
EXTENT_HOLE or some other such place holder, which is not optimal.  So if we
find that our em->block_start is one of these special values, check to see
where the first block of the inode is stored, and use that as a hint.  If that
block is also a special value, just fallback on a hint of 0 and let the
allocator figure out a good place to put the data.
Signed-off-by: NJosef Bacik <josef@redhat.com>
Signed-off-by: NChris Mason <chris.mason@oracle.com>

There's a problem where we don't do any space reservation for truncates, which
can cause you to OOPs because you will be allowed to go off in the weeds a bit
since we don't account for the delalloc bytes that are created as a result of
the truncate.
Signed-off-by: NJosef Bacik <jbacik@redhat.com>
Signed-off-by: NChris Mason <chris.mason@oracle.com>

rpm has a habit of running fdatasync when the file hasn't
changed.  We already detect if a file hasn't been changed
in the current transaction but it might have been sent to
the tree-log in this transaction and not changed since
the last call to fsync.

In this case, we want to avoid a tree log sync, which includes
a number of synchronous writes and barriers.  This commit
extends the existing tracking of the last transaction to change
a file to also track the last sub-transaction.

The end result is that rpm -ivh and -Uvh are roughly twice as fast,
and on par with ext3.
Signed-off-by: NChris Mason <chris.mason@oracle.com>

The extent_type variable was exposed uninit via a goto.  It should be
impossible to trigger because it is protected by a check on another
variable, but this makes sure.
Signed-off-by: NChris Mason <chris.mason@oracle.com>

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

The use of btrfs_dentry_delete is removing dentries from the
dcache when deleting subvolumne. btrfs_dentry_delete ignores
negative dentries. This is incorrect since if we don't remove
the negative dentry, its parent dentry can't be removed.
Signed-off-by: NYan Zheng <zheng.yan@oracle.com>
Signed-off-by: NChris Mason <chris.mason@oracle.com>

This patch fixes an issue with the delalloc metadata space reservation
code.  The problem is we used to free the reservation as soon as we
allocated the delalloc region.  The problem with this is if we are not
inserting an inline extent, we don't actually insert the extent item until
after the ordered extent is written out.  This patch does 3 things,

1) It moves the reservation clearing stuff into the ordered code, so when
we remove the ordered extent we remove the reservation.
2) It adds a EXTENT_DO_ACCOUNTING flag that gets passed when we clear
delalloc bits in the cases where we want to clear the metadata reservation
when we clear the delalloc extent, in the case that we do an inline extent
or we invalidate the page.
3) It adds another waitqueue to the space info so that when we start a fs
wide delalloc flush, anybody else who also hits that area will simply wait
for the flush to finish and then try to make their allocation.

This has been tested thoroughly to make sure we did not regress on
performance.
Signed-off-by: NJosef Bacik <jbacik@redhat.com>
Signed-off-by: NChris Mason <chris.mason@oracle.com>

When compression is on, the cow_file_range code is farmed off to
worker threads.  This allows us to do significant CPU work in parallel
on SMP machines.

But it is a delicate balance around when we clear flags and how.  In
the past we cleared the delalloc flag immediately, which was safe
because the pages stayed locked.

But this is causing problems with the newest ENOSPC code, and with the
recent extent state cleanups we can now clear the delalloc bit at the
same time the uncompressed code does.
Signed-off-by: NChris Mason <chris.mason@oracle.com>

extent_clear_unlock_delalloc has a growing set of ugly parameters
that is very difficult to read and maintain.

This switches to a flag field and well named flag defines.
Signed-off-by: NChris Mason <chris.mason@oracle.com>

[akpm@linux-foundation.org: fix KVM]
Signed-off-by: NAlexey Dobriyan <adobriyan@gmail.com>
Acked-by: NMike Frysinger <vapier@gentoo.org>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

There is a problem where page_mkwrite can be called on a dirtied page that
already has a delalloc range associated with it.  The fix is to clear any
delalloc bits for the range we are dirtying so the space accounting gets
handled properly.  This is the same thing we do in the normal write case, so we
are consistent across the board.  With this patch we no longer leak reserved
space.
Signed-off-by: NJosef Bacik <jbacik@redhat.com>
Signed-off-by: NChris Mason <chris.mason@oracle.com>

At the start of a transaction we do a btrfs_reserve_metadata_space() and
specify how many items we plan on modifying.  Then once we've done our
modifications and such, just call btrfs_unreserve_metadata_space() for
the same number of items we reserved.

For keeping track of metadata needed for data I've had to add an extent_io op
for when we merge extents.  This lets us track space properly when we are doing
sequential writes, so we don't end up reserving way more metadata space than
what we need.

The only place where the metadata space accounting is not done is in the
relocation code.  This is because Yan is going to be reworking that code in the
near future, so running btrfs-vol -b could still possibly result in a ENOSPC
related panic.  This patch also turns off the metadata_ratio stuff in order to
allow users to more efficiently use their disk space.

This patch makes it so we track how much metadata we need for an inode's
delayed allocation extents by tracking how many extents are currently
waiting for allocation.  It introduces two new callbacks for the
extent_io tree's, merge_extent_hook and split_extent_hook.  These help
us keep track of when we merge delalloc extents together and split them
up.  Reservations are handled prior to any actually dirty'ing occurs,
and then we unreserve after we dirty.

btrfs_unreserve_metadata_for_delalloc() will make the appropriate
unreservations as needed based on the number of reservations we
currently have and the number of extents we currently have.  Doing the
reservation outside of doing any of the actual dirty'ing lets us do
things like filemap_flush() the inode to try and force delalloc to
happen, or as a last resort actually start allocation on all delalloc
inodes in the fs.  This has survived dbench, fs_mark and an fsx torture
test.
Signed-off-by: NJosef Bacik <jbacik@redhat.com>
Signed-off-by: NChris Mason <chris.mason@oracle.com>

A recent change enforces only one access point to each subvolume. The first
directory entry (the one added when the subvolume/snapshot was created) is
treated as valid access point, all other subvolume links are linked to dummy
empty directories. The dummy directories are temporary inodes that only in
memory, so we can not rename file into them.
Signed-off-by: NYan Zheng <zheng.yan@oracle.com>
Signed-off-by: NChris Mason <chris.mason@oracle.com>

For every hardlink in btrfs, there is a corresponding inode back
reference. All inode back references for hardlinks in a given
directory are stored in single b-tree item. The size of b-tree item
is limited by the size of b-tree leaf, so we can only create limited
number of hardlinks to a given file in a directory.

The original code lacks of the check, it oops if the number of
hardlinks goes over the limit. This patch fixes the issue by adding
check to btrfs_link and btrfs_rename.
Signed-off-by: NYan Zheng <zheng.yan@oracle.com>
Signed-off-by: NChris Mason <chris.mason@oracle.com>

Signed-off-by: NAlexey Dobriyan <adobriyan@gmail.com>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Signed-off-by: NAlexey Dobriyan <adobriyan@gmail.com>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

This patch adds snapshot/subvolume destroy ioctl.  A subvolume that isn't being
used and doesn't contains links to other subvolumes can be destroyed.
Signed-off-by: NYan Zheng <zheng.yan@oracle.com>
Signed-off-by: NChris Mason <chris.mason@oracle.com>

btrfs allows subvolumes and snapshots anywhere in the directory tree.
If we snapshot a subvolume that contains a link to other subvolume
called subvolA, subvolA can be accessed through both the original
subvolume and the snapshot. This is similar to creating hard link to
directory, and has the very similar problems.

The aim of this patch is enforcing there is only one access point to
each subvolume. Only the first directory entry (the one added when
the subvolume/snapshot was created) is treated as valid access point.
The first directory entry is distinguished by checking root forward
reference. If the corresponding root forward reference is missing,
we know the entry is not the first one.

This patch also adds snapshot/subvolume rename support, the code
allows rename subvolume link across subvolumes.
Signed-off-by: NYan Zheng <zheng.yan@oracle.com>
Signed-off-by: NChris Mason <chris.mason@oracle.com>

The new back reference format does not allow reusing objectid of
deleted snapshot/subvol. So we use ++highest_objectid to allocate
objectid for new snapshot/subvol.

Now we use ++highest_objectid to allocate objectid for both new inode
and new snapshot/subvolume, so this patch removes 'find hole' code in
btrfs_find_free_objectid.
Signed-off-by: NYan Zheng <zheng.yan@oracle.com>
Signed-off-by: NChris Mason <chris.mason@oracle.com>

The allocator has some nice knobs for sending hints about where
to try and allocate new blocks, but when we're doing file allocations
we're not sending any hint at all.

This commit adds a simple extent map search to see if we can
quickly and easily find a hint for the allocator.
Signed-off-by: NChris Mason <chris.mason@oracle.com>

Cc: chris.mason@oracle.com
Signed-off-by: NAndi Kleen <ak@linux.intel.com>

When btrfs_get_extent is reading inline file items for readpage,
it needs to copy the inline extent into the page.  If the
inline extent doesn't cover all of the page, that means there
is a hole in the file, or that our file is smaller than one
page.

readpage does zeroing for the case where the file is smaller than one
page, but nobody is currently zeroing for the case where there is
a hole after the inline item.

This commit changes btrfs_get_extent to zero fill the page past
the end of the inline item.
Signed-off-by: NChris Mason <chris.mason@oracle.com>