invalidate_inode_pages2_range may return -EBUSY occasionally
which results Oops. This patch fixes the issue by moving
invalidate_inode_pages2_range into a loop and keeping calling
it until the return value is not -EBUSY.
The EBUSY return is temporary, and can happen when the btrfs release page
function is unable to release a page because the EXTENT_LOCK
bit is set.
Signed-off-by: Yan Zheng <zheng.yan@oracle.com>
Signed-off-by: Chris Mason <chris.mason@oracle.com>
find_zlib_workspace returns an ERR_PTR value in an error case instead of NULL.
A simplified version of the semantic match that finds this problem is as
follows: (http://coccinelle.lip6.fr/)
// <smpl>
@match exists@
expression x, E;
statement S1, S2;
@@
x = find_zlib_workspace(...)
... when != x = E
(
* if (x == NULL || ...) S1 else S2
|
* if (x == NULL && ...) S1 else S2
)
// </smpl>
Signed-off-by: Julia Lawall <julia@diku.dk>
Signed-off-by: Chris Mason <chris.mason@oracle.com>
This takes care of the following entry from Dan's list:
fs/btrfs/inode.c +4788 btrfs_rename(36) warning: variable derefenced before check 'old_inode'
Reported-by: Dan Carpenter <error27@gmail.com>
Cc: Jonathan Corbet <corbet@lwn.net>
Cc: Eugene Teo <eteo@redhat.com>
Cc: Julia Lawall <julia@diku.dk>
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
Signed-off-by: Chris Mason <chris.mason@oracle.com>
The new credentials code broke load_flat_shared_library() as it now uses
an uninitialized cred pointer.
Reported-by: Bernd Schmidt <bernds_cb1@t-online.de>
Tested-by: Bernd Schmidt <bernds_cb1@t-online.de>
Cc: Mike Frysinger <vapier@gentoo.org>
Cc: David Howells <dhowells@redhat.com>
Cc: <stable@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
I suspect that mnt_want_write_file() may have wrong assumption. I think
mnt_want_write_file() is assuming it increments ->mnt_writers if
(file->f_mode & FMODE_WRITE). But, if it's special_file(), it is false?
Signed-off-by: OGAWA Hirofumi <hirofumi@mail.parknet.co.jp>
Acked-by: Dave Hansen <dave@linux.vnet.ibm.com>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: Nick Piggin <nickpiggin@yahoo.com.au>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
The FIEMAP_IOC_FIEMAP mapping ioctl was missing a 32-bit compat handler,
which means that 32-bit suerspace on 64-bit kernels cannot use this ioctl
command.
The structure is nicely aligned, padded, and sized, so it is just this
simple.
Tested w/ 32-bit ioctl tester (from Josef) on a 64-bit kernel on ext4.
Signed-off-by: Eric Sandeen <sandeen@redhat.com>
Cc: <linux-ext4@vger.kernel.org>
Cc: Mark Lord <lkml@rtr.ca>
Cc: Arnd Bergmann <arnd@arndb.de>
Cc: Josef Bacik <josef@redhat.com>
Cc: Jan Kara <jack@suse.cz>
Cc: <stable@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
When freeing an inode that lost race getting added to the inode cache we
must not call into ->destroy_inode, because that would delete the inode
that won the race from the inode cache radix tree.
This patch uses splits a new xfs_inode_free helper out of xfs_ireclaim
and uses that plus __destroy_inode to make sure we really only free
the memory allocted for the inode that lost the race, and not mess with
the inode cache state.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Eric Sandeen <sandeen@sandeen.net>
Reported-by: Alex Samad <alex@samad.com.au>
Reported-by: Andrew Randrianasulu <randrik@mail.ru>
Reported-by: Stephane <sharnois@max-t.com>
Reported-by: Tommy <tommy@news-service.com>
Reported-by: Miah Gregory <mace@darksilence.net>
Reported-by: Gabriel Barazer <gabriel@oxeva.fr>
Reported-by: Leandro Lucarella <llucax@gmail.com>
Reported-by: Daniel Burr <dburr@fami.com.au>
Reported-by: Nickolay <newmail@spaces.ru>
Reported-by: Michael Guntsche <mike@it-loops.com>
Reported-by: Dan Carley <dan.carley+linuxkern-bugs@gmail.com>
Reported-by: Michael Ole Olsen <gnu@gmx.net>
Reported-by: Michael Weissenbacher <mw@dermichi.com>
Reported-by: Martin Spott <Martin.Spott@mgras.net>
Reported-by: Christian Kujau <lists@nerdbynature.de>
Tested-by: Michael Guntsche <mike@it-loops.com>
Tested-by: Dan Carley <dan.carley+linuxkern-bugs@gmail.com>
Tested-by: Christian Kujau <lists@nerdbynature.de>
When we want to tear down an inode that lost the add to the cache race
in XFS we must not call into ->destroy_inode because that would delete
the inode that won the race from the inode cache radix tree.
This patch provides the __destroy_inode helper needed to fix this,
the actual fix will be in th next patch. As XFS was the only reason
destroy_inode was exported we shift the export to the new __destroy_inode.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Eric Sandeen <sandeen@sandeen.net>
Currently inode_init_always calls into ->destroy_inode if the additional
initialization fails. That's not only counter-intuitive because
inode_init_always did not allocate the inode structure, but in case of
XFS it's actively harmful as ->destroy_inode might delete the inode from
a radix-tree that has never been added. This in turn might end up
deleting the inode for the same inum that has been instanciated by
another process and cause lots of cause subtile problems.
Also in the case of re-initializing a reclaimable inode in XFS it would
free an inode we still want to keep alive.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Eric Sandeen <sandeen@sandeen.net>
Check whether index is within bounds before testing the element.
Acked-by: Jeff Layton <jlayton@redhat.com>
Signed-off-by: Roel Kluin <roel.kluin@gmail.com>
Signed-off-by: Steve French <sfrench@us.ibm.com>
Since forceuid is the default, we now need to show when it's disabled.
Signed-off-by: Jeff Layton <jlayton@redhat.com>
Signed-off-by: Steve French <sfrench@us.ibm.com>
This adds a missing unlock of nilfs->ns_writer_mutex in
nilfs_mdt_write_page() function.
Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
This patch fixes the regression reported here:
http://bugzilla.kernel.org/show_bug.cgi?id=13861
commit 4ae1507f6d changed the default
behavior when the uid= or gid= option was specified for a mount. The
existing behavior was to always clobber the ownership information
provided by the server when these options were specified. The above
commit changed this behavior so that these options simply provided
defaults when the server did not provide this information (unless
"forceuid" or "forcegid" were specified)
This patch reverts this change so that the default behavior is restored.
It also adds "noforceuid" and "noforcegid" options to make it so that
ownership information from the server is preserved, even when the mount
has uid= or gid= options specified.
It also adds a couple of printk notices that pop up when forceuid or
forcegid options are specified without a uid= or gid= option.
Reported-by: Tom Chiverton <bugzilla.kernel.org@falkensweb.com>
Reviewed-by: Shirish Pargaonkar <shirishp@us.ibm.com>
Signed-off-by: Jeff Layton <jlayton@redhat.com>
Signed-off-by: Steve French <sfrench@us.ibm.com>
Andrea Gelmini gave me a report that a kernel oops hit on a nilfs
filesystem with a 1KB block size when doing rsync.
This turned out to be caused by an inconsistency of dirty state
between a page and its buffers storing b-tree node blocks.
If the page had multiple buffers split over multiple logs, and if the
logs were written at a time, a dirty flag remained in the page even
every dirty flag in the buffers was cleared.
This will fix the failure by dropping the dirty flag properly for
pages with the discrete multiple b-tree nodes.
Reported-by: Andrea Gelmini <andrea.gelmini@gmail.com>
Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
Tested-by: Andrea Gelmini <andrea.gelmini@gmail.com>
Cc: stable@kernel.org
The async caching thread can end up looping forever if a given
search puts it at the last key in a leaf. It will end up calling
btrfs_next_leaf and then checking if it needs to politely drop
the read semaphore.
Most of the time this looping isn't noticed because it is able to
make progress the next time around. But, during log replay,
we wait on the async caching thread to finish, and the async thread
is waiting on the commit, and no progress is really made.
The fix used here is to copy the key out of the next leaf,
that way our search lands there properly.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
Yan Zheng hit a problem where we tried to remove some free space but failed
because we couldn't find the free space entry. This is because the free space
was held within a bitmap that had a starting offset well before the actual
offset of the free space, and there were free space extents that were in the
same range as that offset, so tree_search_offset returned with NULL because we
couldn't find a free space extent that had that offset. This is fixed by
making sure that if we fail to find the entry, we re-search again with
bitmap_only set to 1 and do an offset_to_bitmap so we can get the appropriate
bitmap. A similar problem happens in btrfs_alloc_from_bitmap for the
clustering code, but that is not as bad since we will just go and redo our
cluster allocation.
Also this adds some debugging checks to make sure that the free space we are
trying to remove from the bitmap is in fact there. This can probably go away
after a while, but since this code is only used by the tree-logging stuff it
would be nice to run with it for a while to make sure there are no problems.
Signed-off-by: Josef Bacik <jbacik@redhat.com>
Signed-off-by: Chris Mason <chris.mason@oracle.com>
VM calculation for nr_to_write seems off. Bump it way
up, this gets simple streaming writes zippy again.
To be reviewed again after Jens' writeback changes.
Signed-off-by: Christoph Hellwig <hch@infradead.org>
Signed-off-by: Eric Sandeen <sandeen@sandeen.net>
Cc: Chris Mason <chris.mason@oracle.com>
Reviewed-by: Felix Blyakher <felixb@sgi.com>
Signed-off-by: Felix Blyakher <felixb@sgi.com>
commit 6321e3ed2a caused
the full bmv_count's worth of getbmapx structures to get
allocated; telling it to do MAXEXTNUM was a bit insane,
resulting in ENOMEM every time.
Chop it down to something reasonable, the number of slots
in the caller's input buffer. If this is too large the
caller may get ENOMEM but the reason should not be a
mystery, and they can try again with something smaller.
We add 1 to the value because in the normal getbmap
world, bmv_count includes the header and xfs_getbmap does:
nex = bmv->bmv_count - 1;
if (nex <= 0)
return XFS_ERROR(EINVAL);
Signed-off-by: Eric Sandeen <sandeen@sandeen.net>
Reviewed-by: Olaf Weber <olaf@sgi.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Felix Blyakher <felixb@sgi.com>
Commit d01730d74d didn't completely fix
the problem since we still take dqio_mutex and i_mutex in the wrong
order. Move taking of i_mutex further down (luckily it's needed only
for updating inode flags) below where dqio_mutex is taken.
Tested-by: Valdis Kletnieks <valdis.kletnieks@vt.edu>
Signed-off-by: Jan Kara <jack@suse.cz>
VAT inode is located in the last block recorded block of the medium. When the
drive errorneously reports number of recorded blocks, we failed to load the VAT
inode and thus mount the medium. This patch makes kernel try to read VAT inode
from the last block of the device if it is different from the last recorded
block.
Signed-off-by: Jan Kara <jack@suse.cz>
The semaphore used by the async caching threads can prevent a
transaction commit, which can make the FS appear to stall. This
releases the semaphore more often when a transaction commit is
in progress.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
The async block group caching code uses the commit_root pointer
to get a stable version of the extent allocation tree for scanning.
This copy of the tree root isn't going to change and it significantly
reduces the complexity of the scanning code.
During a commit, we have a loop where we update the extent allocation
tree root. We need to loop because updating the root pointer in
the tree of tree roots may allocate blocks which may change the
extent allocation tree.
Right now the commit_root pointer is changed inside this loop. It
is more correct to change the commit_root pointer only after all the
looping is done.
Signed-off-by: Yan Zheng <zheng.yan@oracle.com>
Signed-off-by: Chris Mason <chris.mason@oracle.com>
When a file is deleted from a gfs2 filesystem on one node, a dcache
entry for it may still exist on other nodes in the cluster. If this
happens, gfs2 will be unable to free this file on disk. Because of this,
it's possible to have a gfs2 filesystem with no files on it and no free
space. With this patch, when a node receives a callback notifying it
that the file is being deleted on another node, it schedules a new
workqueue thread to remove the file's dcache entry.
Signed-off-by: Benjamin Marzinski <bmarzins@redhat.com>
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
Since both linked and unlinked inodes are counted by rgd->rd_dinodes, It
makes no sense to count them with the used data blocks (first check that
I changed), it makes sense to count them with the linked inodes (second
check), and it makes no sense to care if there are more unlinked inodes
than linked ones. This fixes these errors.
Signed-off-by: Benjamin Marzinski <bmarzins@redhat.com>
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
GFS2 was placing far too many glocks on the reclaim list that were not good
candidates for freeing up from cache. These locks would sit there and
repeatedly get scanned to see if they could be reclaimed, wasting a lot
of time when there was memory pressure. This fix does more checks on the
locks to see if they are actually likely to be removable from cache.
Signed-off-by: Benjamin Marzinski <bmarzins@redhat.com>
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
When searching for unlinked, but still allocated inodes during block
allocation, avoid the block relating to the inode that is doing the
allocation. This fixes a hang caused when an unlinked, but still
open, inode tries to allocate some more blocks and lands up
finding itself during the search for deallocatable inodes.
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
It is possible for gfs2_shrink_glock_memory() to check a glock for
demotion
that's in the process of being freed by gfs2_glock_put(). In this case,
gfs2_shrink_glock_memory() will acquire a new reference to this glock,
and
then try to free the glock itself when it drops the refernce. To solve
this, gfs2_shrink_glock_memory() just needs to check if the glock is in
the process of being freed, and if so skip it without ever unlocking the
lru_lock.
Signed-off-by: Benjamin Marzinski <bmarzins@redhat.com>
Acked-by: Bob Peterson <rpeterso@redhat.com>
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
GFS2 wasn't syncing its statfs info on grows. This causes a problem
when you grow the filesystem on multiple nodes. GFS2 would calculate
the new space based on the resource groups (which are always current),
and then assume that the filesystem had grown the from the existing
statfs size. If you grew the filesystem on two different nodes in a
short time, the second node wouldn't see the statfs size change from the
first node, and would assume that it was grown by a larger amount than
it was. When all these changes were synced out, the total fileystem
size would be incorrect (the first grow would be counted twice).
This patch syncs makes GFS2 read in the statfs changes from disk before
a grow, and write them out after the grow, while the master statfs inode
is locked.
Signed-off-by: Benjamin Marzinski <bmarzins@redhat.com>
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
This patch removes some of the special cases that the shrinker
was trying to deal with. As a result we leave fewer items on
the list and none at all which cannot be demoted. This makes
the list scanning more efficient and solves some issues seen
with large numbers of inodes.
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
This file makes use of various macros defined in files like asm/current.h
or asm-generic/resource.h. All these files can be included via sched.h.
The building of the !MMU ARM kernel (with additional patches) fails
without this change.
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
Acked-by: David Howells <dhowells@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Create bdgrab(). This function copies an existing reference to a
block_device. It is safe to call from any context.
Hibernation code wishes to copy a reference to the active swap device.
Right now it calls bdget() under a spinlock, but this is wrong because
bdget() can sleep. It doesn't need a full bdget() because we already
hold a reference to active swap devices (and the spinlock protects
against swapoff).
Fixes http://bugzilla.kernel.org/show_bug.cgi?id=13827
Signed-off-by: Alan Jenkins <alan-jenkins@tuffmail.co.uk>
Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
The parse_tag_3_packet function does not check if the tag 3 packet contains a
encrypted key size larger than ECRYPTFS_MAX_ENCRYPTED_KEY_BYTES.
Signed-off-by: Ramon de Carvalho Valle <ramon@risesecurity.org>
[tyhicks@linux.vnet.ibm.com: Added printk newline and changed goto to out_free]
Signed-off-by: Tyler Hicks <tyhicks@linux.vnet.ibm.com>
Cc: stable@kernel.org (2.6.27 and 30)
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Tag 11 packets are stored in the metadata section of an eCryptfs file to
store the key signature(s) used to encrypt the file encryption key.
After extracting the packet length field to determine the key signature
length, a check is not performed to see if the length would exceed the
key signature buffer size that was passed into parse_tag_11_packet().
Thanks to Ramon de Carvalho Valle for finding this bug using fsfuzzer.
Signed-off-by: Tyler Hicks <tyhicks@linux.vnet.ibm.com>
Cc: stable@kernel.org (2.6.27 and 30)
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Update directory hardlink count when moving kobjects to a new parent.
Fixes the following problem which occurs when several devices are
moved to the same parent and then unregistered:
> ls -laF /sys/devices/css0/defunct/
> total 0
> drwxr-xr-x 4294967295 root root 0 2009-07-14 17:02 ./
> drwxr-xr-x 114 root root 0 2009-07-14 17:02 ../
> drwxr-xr-x 2 root root 0 2009-07-14 17:01 power/
> -rw-r--r-- 1 root root 4096 2009-07-14 17:01 uevent
Signed-off-by: Peter Oberparleiter <oberpar@linux.vnet.ibm.com>
Cc: stable <stable@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
- don't stop the caching thread until btrfs_commit_super return.
- if caching is interrupted by umount, set last to (u64)-1.
otherwise the un-scanned range of block group will be considered
as free extent.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
If the referral is malformed or the hostname can't be resolved, then
the current code generates an oops. Fix it to handle these errors
gracefully.
Reported-by: Sandro Mathys <sm@sandro-mathys.ch>
Acked-by: Igor Mammedov <niallain@gmail.com>
CC: Stable <stable@kernel.org>
Signed-off-by: Jeff Layton <jlayton@redhat.com>
Signed-off-by: Steve French <sfrench@us.ibm.com>
We are racy with async block caching and unpinning extents. This patch makes
things much less complicated by only unpinning the extent if the block group is
cached. We check the block_group->cached var under the block_group->lock spin
lock. If it is set to BTRFS_CACHE_FINISHED then we update the pinned counters,
and unpin the extent and add the free space back. If it is not set to this, we
start the caching of the block group so the next time we unpin extents we can
unpin the extent. This keeps us from racing with the async caching threads,
lets us kill the fs wide async thread counter, and keeps us from having to set
DELALLOC bits for every extent we hit if there are caching kthreads going.
One thing that needed to be changed was btrfs_free_super_mirror_extents. Now
instead of just looking for LOCKED extents, we also look for DIRTY extents,
since we could have left some extents pinned in the previous transaction that
will never get freed now that we are unmounting, which would cause us to leak
memory. So btrfs_free_super_mirror_extents has been changed to
btrfs_free_pinned_extents, and it will clear the extents locked for the super
mirror, and any remaining pinned extents that may be present. Thank you,
Signed-off-by: Josef Bacik <jbacik@redhat.com>
Signed-off-by: Chris Mason <chris.mason@oracle.com>
dir has already been tested. It seems that this test should be on the
recently returned value inode.
A simplified version of the semantic match that finds this problem is as
follows: (http://www.emn.fr/x-info/coccinelle/)
Signed-off-by: Julia Lawall <julia@diku.dk>
Signed-off-by: Chris Mason <chris.mason@oracle.com>
Allocating new block group is easy when the disk has plenty of space.
But things get difficult as the disk fills up, especially if
the FS has been run through btrfs-vol -b. The balance operation
is likely to make the total bytes available on the device greater
than the largest extent we'll actually be able to allocate.
But the device extent allocation code incorrectly assumes that a device
with 5G free will be able to allocate a 5G extent. It isn't normally a
problem because device extents don't get freed unless btrfs-vol -b
is run.
This fixes the device extent allocator to remember the largest free
extent it can find, and then uses that value as a fallback.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
Btrfs allocates individual extents from block groups, and each
block group has a specific type. It may hold metadata, data
mirrored or striped etc.
When we balance space (btrfs-vol -b) or remove a drive (btrfs-vol -r)
we free block groups. Once a block group is freed, the space it was
using on the device may be available for use by new block groups.
btrfs_remove_block_group was clearing the flag that said
'our devices are full, don't even try to allocate new block groups',
but it was only clearing that flag for a specific type of block group.
This commit clears the full flag for all of the types of block groups,
making it much more likely that we'll be able to balance space when
the drive is close to full.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
The commit_transaction call to wait_ordered_extents when snap_pending
passes nocow_only=1 to process only NOCOW or PREALLOC extents. This isn't
correct for the 'flushoncommit' mode, as it skips extents we just started
IO on in start_delalloc_inodes.
So, in the flushoncommit case, wait on all ordered extents. Otherwise,
only pass the nocow_only flag to wait_ordered_extents if snap_pending.
Signed-off-by: Sage Weil <sage@newdream.net>
Signed-off-by: Chris Mason <chris.mason@oracle.com>
btrfs_split_leaf and btrfs_del_items can end up in a loop
where one is constantly spliting a given leaf and the other
is constantly merging it back with the adjacent nodes.
There is a better fix for this, but in the interest of something
small, this patch just changes btrfs_del_items back to balancing less
often.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
Check objectid of item before checking the item type, otherwise we may return
zero for a key that is actually too low.
Signed-off-by: Yan Zheng <zheng.yan@oracle.com>
Signed-off-by: Chris Mason <chris.mason@oracle.com>
find_free_dev_extent does not properly handle the case where
the device is not complete free, and there is a free extent
at the beginning of the device.
Signed-off-by: Yan Zheng <zheng.yan@oracle.com>
Signed-off-by: Chris Mason <chris.mason@oracle.com>
comp_keys is duplicating what is done in btrfs_comp_cpu_keys, so just
call it.
Signed-off-by: Diego Calleja <diegocg@gmail.com>
Signed-off-by: Chris Mason <chris.mason@oracle.com>
This patch moves the caching of the block group off to a kthread in order to
allow people to allocate sooner. Instead of blocking up behind the caching
mutex, we instead kick of the caching kthread, and then attempt to make an
allocation. If we cannot, we wait on the block groups caching waitqueue, which
the caching kthread will wake the waiting threads up everytime it finds 2 meg
worth of space, and then again when its finished caching. This is how I tested
the speedup from this
mkfs the disk
mount the disk
fill the disk up with fs_mark
unmount the disk
mount the disk
time touch /mnt/foo
Without my changes this took 11 seconds on my box, with these changes it now
takes 1 second.
Another change thats been put in place is we lock the super mirror's in the
pinned extent map in order to keep us from adding that stuff as free space when
caching the block group. This doesn't really change anything else as far as the
pinned extent map is concerned, since for actual pinned extents we use
EXTENT_DIRTY, but it does mean that when we unmount we have to go in and unlock
those extents to keep from leaking memory.
I've also added a check where when we are reading block groups from disk, if the
amount of space used == the size of the block group, we go ahead and mark the
block group as cached. This drastically reduces the amount of time it takes to
cache the block groups. Using the same test as above, except doing a dd to a
file and then unmounting, it used to take 33 seconds to umount, now it takes 3
seconds.
This version uses the commit_root in the caching kthread, and then keeps track
of how many async caching threads are running at any given time so if one of the
async threads is still running as we cross transactions we can wait until its
finished before handling the pinned extents. Thank you,
Signed-off-by: Josef Bacik <jbacik@redhat.com>
Signed-off-by: Chris Mason <chris.mason@oracle.com>
Currently btrfs has a problem where it can use a ridiculous amount of RAM simply
tracking free space. As free space gets fragmented, we end up with thousands of
entries on an rb-tree per block group, which usually spans 1 gig of area. Since
we currently don't ever flush free space cache back to disk this gets to be a
bit unweildly on large fs's with lots of fragmentation.
This patch solves this problem by using PAGE_SIZE bitmaps for parts of the free
space cache. Initially we calculate a threshold of extent entries we can
handle, which is however many extent entries we can cram into 16k of ram. The
maximum amount of RAM that should ever be used to track 1 gigabyte of diskspace
will be 32k of RAM, which scales much better than we did before.
Once we pass the extent threshold, we start adding bitmaps and using those
instead for tracking the free space. This patch also makes it so that any free
space thats less than 4 * sectorsize we go ahead and put into a bitmap. This is
nice since we try and allocate out of the front of a block group, so if the
front of a block group is heavily fragmented and then has a huge chunk of free
space at the end, we go ahead and add the fragmented areas to bitmaps and use a
normal extent entry to track the big chunk at the back of the block group.
I've also taken the opportunity to revamp how we search for free space.
Previously we indexed free space via an offset indexed rb tree and a bytes
indexed rb tree. I've dropped the bytes indexed rb tree and use only the offset
indexed rb tree. This cuts the number of tree operations we were doing
previously down by half, and gives us a little bit of a better allocation
pattern since we will always start from a specific offset and search forward
from there, instead of searching for the size we need and try and get it as
close as possible to the offset we want.
I've given this a healthy amount of testing pre-new format stuff, as well as
post-new format stuff. I've booted up my fedora box which is installed on btrfs
with this patch and ran with it for a few days without issues. I've not seen
any performance regressions in any of my tests.
Since the last patch Yan Zheng fixed a problem where we could have overlapping
entries, so updating their offset inline would cause problems. Thanks,
Signed-off-by: Josef Bacik <jbacik@redhat.com>
Signed-off-by: Chris Mason <chris.mason@oracle.com>
Yan Zheng