The delayed reference queue maintains pending operations that need to
be done to the extent allocation tree. These are processed by
finding records in the tree that are not currently being processed one at
a time.
This is slow because it uses lots of time searching through the rbtree
and because it creates lock contention on the extent allocation tree
when lots of different procs are running delayed refs at the same time.
This commit changes things to grab a cluster of refs for processing,
using a cursor into the rbtree as the starting point of the next search.
This way we walk smoothly through the rbtree.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
The extent allocation tree maintains a reference count and full
back reference information for every extent allocated in the
filesystem. For subvolume and snapshot trees, every time
a block goes through COW, the new copy of the block adds a reference
on every block it points to.
If a btree node points to 150 leaves, then the COW code needs to go
and add backrefs on 150 different extents, which might be spread all
over the extent allocation tree.
These updates currently happen during btrfs_cow_block, and most COWs
happen during btrfs_search_slot. btrfs_search_slot has locks held
on both the parent and the node we are COWing, and so we really want
to avoid IO during the COW if we can.
This commit adds an rbtree of pending reference count updates and extent
allocations. The tree is ordered by byte number of the extent and byte number
of the parent for the back reference. The tree allows us to:
1) Modify back references in something close to disk order, reducing seeks
2) Significantly reduce the number of modifications made as block pointers
are balanced around
3) Do all of the extent insertion and back reference modifications outside
of the performance critical btrfs_search_slot code.
#3 has the added benefit of greatly reducing the btrfs stack footprint.
The extent allocation tree modifications are done without the deep
(and somewhat recursive) call chains used in the past.
These delayed back reference updates must be done before the transaction
commits, and so the rbtree is tied to the transaction. Throttling is
implemented to help keep the queue of backrefs at a reasonable size.
Since there was a similar mechanism in place for the extent tree
extents, that is removed and replaced by the delayed reference tree.
Yan Zheng <yan.zheng@oracle.com> helped review and fixup this code.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
btrfs_tree_locked was being used to make sure a given extent_buffer was
properly locked in a few places. But, it wasn't correct for UP compiled
kernels.
This switches it to using assert_spin_locked instead, and renames it to
btrfs_assert_tree_locked to better reflect how it was really being used.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
Btrfs is currently using spin_lock_nested with a nested value based
on the tree depth of the block. But, this doesn't quite work because
the max tree depth is bigger than what spin_lock_nested can deal with,
and because locks are sometimes taken before the level field is filled in.
The solution here is to use lockdep_set_class_and_name instead, and to
set the class before unlocking the pages when the block is read from the
disk and just after init of a freshly allocated tree block.
btrfs_clear_path_blocking is also changed to take the locks in the proper
order, and it also makes sure all the locks currently held are properly
set to blocking before it tries to retake the spinlocks. Otherwise, lockdep
gets upset about bad lock orderin.
The lockdep magic cam from Peter Zijlstra <peterz@infradead.org>
Signed-off-by: Chris Mason <chris.mason@oracle.com>
Most of the btrfs metadata operations can be protected by a spinlock,
but some operations still need to schedule.
So far, btrfs has been using a mutex along with a trylock loop,
most of the time it is able to avoid going for the full mutex, so
the trylock loop is a big performance gain.
This commit is step one for getting rid of the blocking locks entirely.
btrfs_tree_lock takes a spinlock, and the code explicitly switches
to a blocking lock when it starts an operation that can schedule.
We'll be able get rid of the blocking locks in smaller pieces over time.
Tracing allows us to find the most common cause of blocking, so we
can start with the hot spots first.
The basic idea is:
btrfs_tree_lock() returns with the spin lock held
btrfs_set_lock_blocking() sets the EXTENT_BUFFER_BLOCKING bit in
the extent buffer flags, and then drops the spin lock. The buffer is
still considered locked by all of the btrfs code.
If btrfs_tree_lock gets the spinlock but finds the blocking bit set, it drops
the spin lock and waits on a wait queue for the blocking bit to go away.
Much of the code that needs to set the blocking bit finishes without actually
blocking a good percentage of the time. So, an adaptive spin is still
used against the blocking bit to avoid very high context switch rates.
btrfs_clear_lock_blocking() clears the blocking bit and returns
with the spinlock held again.
btrfs_tree_unlock() can be called on either blocking or spinning locks,
it does the right thing based on the blocking bit.
ctree.c has a helper function to set/clear all the locked buffers in a
path as blocking.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
Before metadata is written to disk, it is updated to reflect that writeout
has begun. Once this update is done, the block must be cow'd before it
can be modified again.
This update was originally synchronized by using a per-fs spinlock. Today
the buffers for the metadata blocks are locked before writeout begins,
and everyone that tests the flag has the buffer locked as well.
So, the per-fs spinlock (called hash_lock for no good reason) is no
longer required.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
Tracing shows the delay between when an async thread goes to sleep
and when more work is added is often very short. This commit adds
a little bit of delay and extra checking to the code right before
we schedule out.
It allows more work to be added to the worker
without requiring notifications from other procs.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
To improve performance, btrfs_sync_log merges tree log sync
requests. But it wrongly merges sync requests for different
tree logs. If multiple tree logs are synced at the same time,
only one of them actually gets synced.
This patch has following changes to fix the bug:
Move most tree log related fields in btrfs_fs_info to
btrfs_root. This allows merging sync requests separately
for each tree log.
Don't insert root item into the log root tree immediately
after log tree is allocated. Root item for log tree is
inserted when log tree get synced for the first time. This
allows syncing the log root tree without first syncing all
log trees.
At tree-log sync, btrfs_sync_log first sync the log tree;
then updates corresponding root item in the log root tree;
sync the log root tree; then update the super block.
Signed-off-by: Yan Zheng <zheng.yan@oracle.com>
Merge list_for_each* and list_entry to list_for_each_entry*
Signed-off-by: Qinghuang Feng <qhfeng.kernel@gmail.com>
Signed-off-by: Chris Mason <chris.mason@oracle.com>
kthread_run() returns the kthread or ERR_PTR(-ENOMEM), not NULL.
Signed-off-by: Qinghuang Feng <qhfeng.kernel@gmail.com>
Signed-off-by: Chris Mason <chris.mason@oracle.com>
The data in fs_info->super_for_commit are zeros before the
first transaction commit. If tree log sync and system crash
both occur before the first transaction commit, super block
will get corrupted.
This fixes it by properly filling in the super_for_commit field at
open time.
Signed-off-by: Yan Zheng <zheng.yan@oracle.com>
bio_end_io for reads without checksumming on and btree writes were
happening without using async thread pools. This means the extent_io.c
code had to use spin_lock_irq and friends on the rb tree locks for
extent state.
There were some irq safe vs unsafe lock inversions between the delallock
lock and the extent state locks. This patch gets rid of them by moving
all end_io code into the thread pools.
To avoid contention and deadlocks between the data end_io processing and the
metadata end_io processing yet another thread pool is added to finish
off metadata writes.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
This patch makes seed device possible to be shared by
multiple mounted file systems. The sharing is achieved
by cloning seed device's btrfs_fs_devices structure.
Thanks you,
Signed-off-by: Yan Zheng <zheng.yan@oracle.com>
This patch implements superblock duplication. Superblocks
are stored at offset 16K, 64M and 256G on every devices.
Spaces used by superblocks are preserved by the allocator,
which uses a reverse mapping function to find the logical
addresses that correspond to superblocks. Thank you,
Signed-off-by: Yan Zheng <zheng.yan@oracle.com>
Btrfs stores checksums for each data block. Until now, they have
been stored in the subvolume trees, indexed by the inode that is
referencing the data block. This means that when we read the inode,
we've probably read in at least some checksums as well.
But, this has a few problems:
* The checksums are indexed by logical offset in the file. When
compression is on, this means we have to do the expensive checksumming
on the uncompressed data. It would be faster if we could checksum
the compressed data instead.
* If we implement encryption, we'll be checksumming the plain text and
storing that on disk. This is significantly less secure.
* For either compression or encryption, we have to get the plain text
back before we can verify the checksum as correct. This makes the raid
layer balancing and extent moving much more expensive.
* It makes the front end caching code more complex, as we have touch
the subvolume and inodes as we cache extents.
* There is potentitally one copy of the checksum in each subvolume
referencing an extent.
The solution used here is to store the extent checksums in a dedicated
tree. This allows us to index the checksums by phyiscal extent
start and length. It means:
* The checksum is against the data stored on disk, after any compression
or encryption is done.
* The checksum is stored in a central location, and can be verified without
following back references, or reading inodes.
This makes compression significantly faster by reducing the amount of
data that needs to be checksummed. It will also allow much faster
raid management code in general.
The checksums are indexed by a key with a fixed objectid (a magic value
in ctree.h) and offset set to the starting byte of the extent. This
allows us to copy the checksum items into the fsync log tree directly (or
any other tree), without having to invent a second format for them.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
This patch gives us the space we will need in order to have different csum
algorithims at some point in the future. We save the csum algorithim type
in the superblock, and use those instead of define's.
Signed-off-by: Josef Bacik <jbacik@redhat.com>
This needs to be applied on top of my previous patches, but is needed for more
than just my new stuff. We're going to the wrong label when we have an error,
we try to stop the workers, but they are started below all of this code. This
fixes it so we go to the right error label and not panic when we fail one of
these cases.
Signed-off-by: Josef Bacik <jbacik@redhat.com>
This adds the necessary disk format for handling compatibility flags
in the future to handle disk format changes. We have a compat_flags,
compat_ro_flags and incompat_flags set for the super block. Compat
flags will be to hold the features that are compatible with older
versions of btrfs, compat_ro flags have features that are compatible
with older versions of btrfs if the fs is mounted read only, and
incompat_flags has features that are incompatible with older versions
of btrfs. This also axes the compat_flags field for the inode and
just makes the flags field a 64bit field, and changes the root item
flags field to 64bit.
Signed-off-by: Josef Bacik <jbacik@redhat.com>
Shut up various sparse warnings about symbols that should be either
static or have their declarations in scope.
Signed-off-by: Christoph Hellwig <hch@lst.de>
The log replay produces dirty roots. These dirty roots
should be dropped immediately if the fs is mounted as
ro. Otherwise they can be added to the dirty root list
again when remounting the fs as rw. Thank you,
Signed-off-by: Yan Zheng <zheng.yan@oracle.com>
The btrfs git kernel trees is used to build a standalone tree for
compiling against older kernels. This commit makes the standalone tree
work with 2.6.27
Signed-off-by: Chris Mason <chris.mason@oracle.com>
fsync log replay can change the filesystem, so it cannot be delayed until
mount -o rw,remount, and it can't be forgotten entirely. So, this patch
changes btrfs to do with reiserfs, ext3 and xfs do, which is to do the
log replay even when mounted readonly.
On a readonly device if log replay is required, the mount is aborted.
Getting all of this right had required fixing up some of the error
handling in open_ctree.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
While building large bios in writepages, btrfs may end up waiting
for other page writeback to finish if WB_SYNC_ALL is used.
While it is waiting, the bio it is building has a number of pages with the
writeback bit set and they aren't getting to the disk any time soon. This
lowers the latencies of writeback in general by sending down the bio being
built before waiting for other pages.
The bio submission code tries to limit the total number of async bios in
flight by waiting when we're over a certain number of async bios. But,
the waits are happening while writepages is building bios, and this can easily
lead to stalls and other problems for people calling wait_on_page_writeback.
The current fix is to let the congestion tests take care of waiting.
sync() and others make sure to drain the current async requests to make
sure that everything that was pending when the sync was started really get
to disk. The code would drain pending requests both before and after
submitting a new request.
But, if one of the requests is waiting for page writeback to finish,
the draining waits might block that page writeback. This changes the
draining code to only wait after submitting the bio being processed.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
For larger multi-device filesystems, there was logic to limit the
number of devices unplugged to just the page that was sent to our sync_page
function.
But, the code wasn't always unplugging the right device. Since this was
just an optimization, disable it for now.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
For a directory tree:
/mnt/subvolA/subvolB
btrfsctl -s /mnt/subvolA/subvolB /mnt
Will create a directory loop with subvolA under subvolB. This
commit uses the forward refs for each subvol and snapshot to error out
before creating the loop.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
Each subvolume has its own private inode number space, and so we need
to fill in different device numbers for each subvolume to avoid confusing
applications.
This commit puts a struct super_block into struct btrfs_root so it can
call set_anon_super() and get a different device number generated for
each root.
btrfs_rename is changed to prevent renames across subvols.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
Before, all snapshots and subvolumes lived in a single flat directory. This
was awkward and confusing because the single flat directory was only writable
with the ioctls.
This commit changes the ioctls to create subvols and snapshots at any
point in the directory tree. This requires making separate ioctls for
snapshot and subvol creation instead of a combining them into one.
The subvol ioctl does:
btrfsctl -S subvol_name parent_dir
After the ioctl is done subvol_name lives inside parent_dir.
The snapshot ioctl does:
btrfsctl -s path_for_snapshot root_to_snapshot
path_for_snapshot can be an absolute or relative path. btrfsctl breaks it up
into directory and basename components.
root_to_snapshot can be any file or directory in the FS. The snapshot
is taken of the entire root where that file lives.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
Seed device is a special btrfs with SEEDING super flag
set and can only be mounted in read-only mode. Seed
devices allow people to create new btrfs on top of it.
The new FS contains the same contents as the seed device,
but it can be mounted in read-write mode.
This patch does the following:
1) split code in btrfs_alloc_chunk into two parts. The first part does makes
the newly allocated chunk usable, but does not do any operation that modifies
the chunk tree. The second part does the the chunk tree modifications. This
division is for the bootstrap step of adding storage to the seed device.
2) Update device management code to handle seed device.
The basic idea is: For an FS grown from seed devices, its
seed devices are put into a list. Seed devices are
opened on demand at mounting time. If any seed device is
missing or has been changed, btrfs kernel module will
refuse to mount the FS.
3) make btrfs_find_block_group not return NULL when all
block groups are read-only.
Signed-off-by: Yan Zheng <zheng.yan@oracle.com>
This patch adds mount ro and remount support. The main
changes in patch are: adding btrfs_remount and related
helper function; splitting the transaction related code
out of close_ctree into btrfs_commit_super; updating
allocator to properly handle read only block group.
Signed-off-by: Yan Zheng <zheng.yan@oracle.com>
This fixes latency problems on metadata reads by making sure they
don't go through the async submit queue, and by tuning down the amount
of readahead done during btree searches.
Also, the btrfs bdi congestion function is tuned to ignore the
number of pending async bios and checksums pending. There is additional
code that throttles new async bios now and the congestion function
doesn't need to worry about it anymore.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
When btrfs unplugs, it tries to find the correct device to unplug
via search through the extent_map tree. This avoids unplugging
a device that doesn't need it, but is a waste of time for filesystems
with a small number of devices.
This patch checks the total number of devices before doing the
search.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
When reading compressed extents, try to put pages into the page cache
for any pages covered by the compressed extent that readpages didn't already
preload.
Add an async work queue to handle transformations at delayed allocation processing
time. Right now this is just compression. The workflow is:
1) Find offsets in the file marked for delayed allocation
2) Lock the pages
3) Lock the state bits
4) Call the async delalloc code
The async delalloc code clears the state lock bits and delalloc bits. It is
important this happens before the range goes into the work queue because
otherwise it might deadlock with other work queue items that try to lock
those extent bits.
The file pages are compressed, and if the compression doesn't work the
pages are written back directly.
An ordered work queue is used to make sure the inodes are written in the same
order that pdflush or writepages sent them down.
This changes extent_write_cache_pages to let the writepage function
update the wbc nr_written count.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
Btrfs uses kernel threads to create async work queues for cpu intensive
operations such as checksumming and decompression. These work well,
but they make it difficult to keep IO order intact.
A single writepages call from pdflush or fsync will turn into a number
of bios, and each bio is checksummed in parallel. Once the checksum is
computed, the bio is sent down to the disk, and since we don't control
the order in which the parallel operations happen, they might go down to
the disk in almost any order.
The code deals with this somewhat by having deep work queues for a single
kernel thread, making it very likely that a single thread will process all
the bios for a single inode.
This patch introduces an explicitly ordered work queue. As work structs
are placed into the queue they are put onto the tail of a list. They have
three callbacks:
->func (cpu intensive processing here)
->ordered_func (order sensitive processing here)
->ordered_free (free the work struct, all processing is done)
The work struct has three callbacks. The func callback does the cpu intensive
work, and when it completes the work struct is marked as done.
Every time a work struct completes, the list is checked to see if the head
is marked as done. If so the ordered_func callback is used to do the
order sensitive processing and the ordered_free callback is used to do
any cleanup. Then we loop back and check the head of the list again.
This patch also changes the checksumming code to use the ordered workqueues.
One a 4 drive array, it increases streaming writes from 280MB/s to 350MB/s.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
This patch adds transaction IDs to root tree pointers.
Transaction IDs in tree pointers are compared with the
generation numbers in block headers when reading root
blocks of trees. This can detect some types of IO errors.
Signed-off-by: Yan Zheng <zheng.yan@oracle.com>
This patch removes the giant fs_info->alloc_mutex and replaces it with a bunch
of little locks.
There is now a pinned_mutex, which is used when messing with the pinned_extents
extent io tree, and the extent_ins_mutex which is used with the pending_del and
extent_ins extent io trees.
The locking for the extent tree stuff was inspired by a patch that Yan Zheng
wrote to fix a race condition, I cleaned it up some and changed the locking
around a little bit, but the idea remains the same. Basically instead of
holding the extent_ins_mutex throughout the processing of an extent on the
extent_ins or pending_del trees, we just hold it while we're searching and when
we clear the bits on those trees, and lock the extent for the duration of the
operations on the extent.
Also to keep from getting hung up waiting to lock an extent, I've added a
try_lock_extent so if we cannot lock the extent, move on to the next one in the
tree and we'll come back to that one. I have tested this heavily and it does
not appear to break anything. This has to be applied on top of my
find_free_extent redo patch.
I tested this patch on top of Yan's space reblancing code and it worked fine.
The only thing that has changed since the last version is I pulled out all my
debugging stuff, apparently I forgot to run guilt refresh before I sent the
last patch out. Thank you,
Signed-off-by: Josef Bacik <jbacik@redhat.com>
This patch improves the space balancing code to keep more sharing
of tree blocks. The only case that breaks sharing of tree blocks is
data extents get fragmented during balancing. The main changes in
this patch are:
Add a 'drop sub-tree' function. This solves the problem in old code
that BTRFS_HEADER_FLAG_WRITTEN check breaks sharing of tree block.
Remove relocation mapping tree. Relocation mappings are stored in
struct btrfs_ref_path and updated dynamically during walking up/down
the reference path. This reduces CPU usage and simplifies code.
This patch also fixes a bug. Root items for reloc trees should be
updated in btrfs_free_reloc_root.
Signed-off-by: Yan Zheng <zheng.yan@oracle.com>
This is a large change for adding compression on reading and writing,
both for inline and regular extents. It does some fairly large
surgery to the writeback paths.
Compression is off by default and enabled by mount -o compress. Even
when the -o compress mount option is not used, it is possible to read
compressed extents off the disk.
If compression for a given set of pages fails to make them smaller, the
file is flagged to avoid future compression attempts later.
* While finding delalloc extents, the pages are locked before being sent down
to the delalloc handler. This allows the delalloc handler to do complex things
such as cleaning the pages, marking them writeback and starting IO on their
behalf.
* Inline extents are inserted at delalloc time now. This allows us to compress
the data before inserting the inline extent, and it allows us to insert
an inline extent that spans multiple pages.
* All of the in-memory extent representations (extent_map.c, ordered-data.c etc)
are changed to record both an in-memory size and an on disk size, as well
as a flag for compression.
From a disk format point of view, the extent pointers in the file are changed
to record the on disk size of a given extent and some encoding flags.
Space in the disk format is allocated for compression encoding, as well
as encryption and a generic 'other' field. Neither the encryption or the
'other' field are currently used.
In order to limit the amount of data read for a single random read in the
file, the size of a compressed extent is limited to 128k. This is a
software only limit, the disk format supports u64 sized compressed extents.
In order to limit the ram consumed while processing extents, the uncompressed
size of a compressed extent is limited to 256k. This is a software only limit
and will be subject to tuning later.
Checksumming is still done on compressed extents, and it is done on the
uncompressed version of the data. This way additional encodings can be
layered on without having to figure out which encoding to checksum.
Compression happens at delalloc time, which is basically singled threaded because
it is usually done by a single pdflush thread. This makes it tricky to
spread the compression load across all the cpus on the box. We'll have to
look at parallel pdflush walks of dirty inodes at a later time.
Decompression is hooked into readpages and it does spread across CPUs nicely.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
This improves the comments at the top of many functions. It didn't
dive into the guts of functions because I was trying to
avoid merging problems with the new allocator and back reference work.
extent-tree.c and volumes.c were both skipped, and there is definitely
more work todo in cleaning and commenting the code.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
btrfs-vol -a /dev/xxx will zero the first and last two MB of the device.
The kernel code needs to wait for this IO to finish before it adds
the device.
btrfs metadata IO does not happen through the block device inode. A
separate address space is used, allowing the zero filled buffer heads in
the block device inode to be written to disk after FS metadata starts
going down to the disk via the btrfs metadata inode.
The end result is zero filled metadata blocks after adding new devices
into the filesystem.
The fix is a simple filemap_write_and_wait on the block device inode
before actually inserting it into the pool of available devices.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
This patch updates the space balancing code to utilize the new
backref format. Before, btrfs-vol -b would break any COW links
on data blocks or metadata. This was slow and caused the amount
of space used to explode if a large number of snapshots were present.
The new code can keeps the sharing of all data extents and
most of the tree blocks.
To maintain the sharing of data extents, the space balance code uses
a seperate inode hold data extent pointers, then updates the references
to point to the new location.
To maintain the sharing of tree blocks, the space balance code uses
reloc trees to relocate tree blocks in reference counted roots.
There is one reloc tree for each subvol, and all reloc trees share
same root key objectid. Reloc trees are snapshots of the latest
committed roots of subvols (root->commit_root).
To relocate a tree block referenced by a subvol, there are two steps.
COW the block through subvol's reloc tree, then update block pointer in
the subvol to point to the new block. Since all reloc trees share
same root key objectid, doing special handing for tree blocks
owned by them is easy. Once a tree block has been COWed in one
reloc tree, we can use the resulting new block directly when the
same block is required to COW again through other reloc trees.
In this way, relocated tree blocks are shared between reloc trees,
so they are also shared between subvols.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
Btrfs has a cache of reference counts in leaves, allowing it to
avoid reading tree leaves while deleting snapshots. To reduce
contention with multiple subvolumes, this cache is private to each
subvolume.
This patch adds shared reference cache support. The new space
balancing code plays with multiple subvols at the same time, So
the old per-subvol reference cache is not well suited.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
Btrfs metadata writeback is fairly expensive. Once a tree block is written
it must be cowed before it can be changed again. The btree writepages
code has a threshold based on a count of dirty btree bytes which is
updated as IO is sent out.
This changes btree_writepages to skip the writeout if there are less
than 32MB of dirty bytes from the btrees, improving performance
across many workloads.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
Btrfs had compatibility code for kernels back to 2.6.18. These have
been removed, and will be maintained in a separate backport
git tree from now on.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
This patch makes the back reference system to explicit record the
location of parent node for all types of extents. The location of
parent node is placed into the offset field of backref key. Every
time a tree block is balanced, the back references for the affected
lower level extents are updated.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
Tree blocks were using async bio submission, but the sum was still
being done directly during writepage. This moves the checksumming
into the worker thread.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
1) replace the per fs_info extent_io_tree that tracked free space with two
rb-trees per block group to track free space areas via offset and size. The
reason to do this is because most allocations come with a hint byte where to
start, so we can usually find a chunk of free space at that hint byte to satisfy
the allocation and get good space packing. If we cannot find free space at or
after the given offset we fall back on looking for a chunk of the given size as
close to that given offset as possible. When we fall back on the size search we
also try to find a slot as close to the size we want as possible, to avoid
breaking small chunks off of huge areas if possible.
2) remove the extent_io_tree that tracked the block group cache from fs_info and
replaced it with an rb-tree thats tracks block group cache via offset. also
added a per space_info list that tracks the block group cache for the particular
space so we can lookup related block groups easily.
3) cleaned up the allocation code to make it a little easier to read and a
little less complicated. Basically there are 3 steps, first look from our
provided hint. If we couldn't find from that given hint, start back at our
original search start and look for space from there. If that fails try to
allocate space if we can and start looking again. If not we're screwed and need
to start over again.
4) small fixes. there were some issues in volumes.c where we wouldn't allocate
the rest of the disk. fixed cow_file_range to actually pass the alloc_hint,
which has helped a good bit in making the fs_mark test I run have semi-normal
results as we run out of space. Generally with data allocations we don't track
where we last allocated from, so everytime we did a data allocation we'd search
through every block group that we have looking for free space. Now searching a
block group with no free space isn't terribly time consuming, it was causing a
slight degradation as we got more data block groups. The alloc_hint has fixed
this slight degredation and made things semi-normal.
There is still one nagging problem I'm working on where we will get ENOSPC when
there is definitely plenty of space. This only happens with metadata
allocations, and only when we are almost full. So you generally hit the 85%
mark first, but sometimes you'll hit the BUG before you hit the 85% wall. I'm
still tracking it down, but until then this seems to be pretty stable and make a
significant performance gain.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
This is the same way the transaction code makes sure that all the
other tree blocks are safely on disk. There's an extent_io tree
for each root, and any blocks allocated to the tree logs are
recorded in that tree.
At tree-log sync, the extent_io tree is walked to flush down the
dirty pages and wait for them.
The main benefit is less time spent walking the tree log and skipping
clean pages, and getting sequential IO down to the drive.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
Since tree log blocks get freed every transaction, they never really
need to be written to disk. This skips the step where we update
metadata to record they were allocated.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
* Pin down data blocks to prevent them from being reallocated like so:
trans 1: allocate file extent
trans 2: free file extent
trans 3: free file extent during old snapshot deletion
trans 3: allocate file extent to new file
trans 3: fsync new file
Before the tree logging code, this was legal because the fsync
would commit the transation that did the final data extent free
and the transaction that allocated the extent to the new file
at the same time.
With the tree logging code, the tree log subtransaction can commit
before the transaction that freed the extent. If we crash,
we're left with two different files using the extent.
* Don't wait in start_transaction if log replay is going on. This
avoids deadlocks from iput while we're cleaning up link counts in the
replay code.
* Don't deadlock in replay_one_name by trying to read an inode off
the disk while holding paths for the directory
* Hold the buffer lock while we mark a buffer as written. This
closes a race where someone is changing a buffer while we write it.
They are supposed to mark it dirty again after they change it, but
this violates the cow rules.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
File syncs and directory syncs are optimized by copying their
items into a special (copy-on-write) log tree. There is one log tree per
subvolume and the btrfs super block points to a tree of log tree roots.
After a crash, items are copied out of the log tree and back into the
subvolume. See tree-log.c for all the details.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
The current code waits for the count of async bio submits to get below
a given threshold if it is too high right after adding the latest bio
to the work queue. This isn't optimal because the caller may have
sequential adjacent bios pending they are waiting to send down the pipe.
This changeset requires the caller to wait on the async bio count,
and changes the async checksumming submits to wait for async bios any
time they self throttle.
The end result is much higher sequential throughput.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
Before, the btrfs bdi congestion function was used to test for too many
async bios. This keeps that check to throttle pdflush, but also
adds a check while queuing bios.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
Before this change, btrfs would use a bdi congestion function to make
sure there weren't too many pending async checksum work items.
This change makes the process creating async work items wait instead,
leading to fewer congestion returns from the bdi. This improves
pdflush background_writeout scanning.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
After writing out all the remaining btree blocks in the transaction,
the commit code would use filemap_fdatawait to make sure it was all
on disk. This means it would wait for blocks written by other procs
as well.
The new code walks the list of blocks for this transaction again
and waits only for those required by this transaction.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
* Make walk_down_tree wake up throttled tasks more often
* Make walk_down_tree call cond_resched during long loops
* As the size of the ref cache grows, wait longer in throttle
* Get rid of the reada code in walk_down_tree, the leaves don't get
read anymore, thanks to the ref cache.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
Large streaming reads make for large bios, which means each entry on the
list async work queues represents a large amount of data. IO
congestion throttling on the device was kicking in before the async
worker threads decided a single thread was busy and needed some help.
The end result was that a streaming read would result in a single CPU
running at 100% instead of balancing the work off to other CPUs.
This patch also changes the pre-IO checksum lookup done by reads to
work on a per-bio basis instead of a per-page. This results in many
extra btree lookups on large streaming reads. Doing the checksum lookup
right before bio submit allows us to reuse searches while processing
adjacent offsets.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
The memory reclaiming issue happens when snapshot exists. In that
case, some cache entries may not be used during old snapshot dropping,
so they will remain in the cache until umount.
The patch adds a field to struct btrfs_leaf_ref to record create time. Besides,
the patch makes all dead roots of a given snapshot linked together in order of
create time. After a old snapshot was completely dropped, we check the dead
root list and remove all cache entries created before the oldest dead root in
the list.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
It was incorrectly clearing the up to date flag on the buffer even
when the buffer properly verified.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
A large reference cache is directly related to a lot of work pending
for the cleaner thread. This throttles back new operations based on
the size of the reference cache so the cleaner thread will be able to keep
up.
Overall, this actually makes the FS faster because the cleaner thread will
be more likely to find things in cache.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
This changes the reference cache to make a single cache per root
instead of one cache per transaction, and to key by the byte number
of the disk block instead of the keys inside.
This makes it much less likely to have cache misses if a snapshot
or something has an extra reference on a higher node or a leaf while
the first transaction that added the leaf into the cache is dropping.
Some throttling is added to functions that free blocks heavily so they
wait for old transactions to drop.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
Much of the IO done while dropping snapshots is done looking up
leaves in the filesystem trees to see if they point to any extents and
to drop the references on any extents found.
This creates a cache so that IO isn't required.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
Before setting an extent to delalloc, the code needs to wait for
pending ordered extents.
Also, the relocation code needs to wait for ordered IO before scanning
the block group again. This is because the extents are not removed
until the IO for the new extents is finished
Signed-off-by: Chris Mason <chris.mason@oracle.com>
Checksum items are not inserted into the tree until all of the io from a
given extent is complete. This means one dirty page from an extent may
be written, freed, and then read again before the entire extent is on disk
and the checksum item is inserted.
The checksums themselves are stored in the ordered extent so they can
be inserted in bulk when IO is complete. On read, if a checksum item isn't
found, the ordered extents were being searched for a checksum record.
This all worked most of the time, but the checksum insertion code tries
to reduce the number of tree operations by pre-inserting checksum items
based on i_size and a few other factors. This means the read code might
find a checksum item that hasn't yet really been filled in.
This commit changes things to check the ordered extents first and only
dive into the btree if nothing was found. This removes the need for
extra locking and is more reliable.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
Before, extent buffers were a temporary object, meant to map a number of pages
at once and collect operations on them.
But, a few extra fields have crept in, and they are also the best place to
store a per-tree block lock field as well. This commit puts the extent
buffers into an rbtree, and ensures a single extent buffer for each
tree block.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
btrfs_commit_transaction has to loop waiting for any writers in the
transaction to finish before it can proceed. btrfs_start_transaction
should be polite and not join a transaction that is in the process
of being finished off.
There are a few places that can't wait, basically the ones doing IO that
might be needed to finish the transaction. For them, btrfs_join_transaction
is added.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
Higher layers sometimes call set_page_dirty without asking the filesystem
to help. This causes many problems for the data=ordered and cow code.
This commit detects pages that haven't been properly setup for IO and
kicks off an async helper to deal with them.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
The old data=ordered code would force commit to wait until
all the data extents from the transaction were fully on disk. This
introduced large latencies into the commit and stalled new writers
in the transaction for a long time.
The new code changes the way data allocations and extents work:
* When delayed allocation is filled, data extents are reserved, and
the extent bit EXTENT_ORDERED is set on the entire range of the extent.
A struct btrfs_ordered_extent is allocated an inserted into a per-inode
rbtree to track the pending extents.
* As each page is written EXTENT_ORDERED is cleared on the bytes corresponding
to that page.
* When all of the bytes corresponding to a single struct btrfs_ordered_extent
are written, The previously reserved extent is inserted into the FS
btree and into the extent allocation trees. The checksums for the file
data are also updated.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
The btree defragger wasn't making forward progress because the new key wasn't
being saved by the btrfs_search_forward function.
This also disables the automatic btree defrag, it wasn't scaling well to
huge filesystems. The auto-defrag needs to be done differently.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
This creates one kthread for commits and one kthread for
deleting old snapshots. All the work queues are removed.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
The existing throttle mechanism was often not sufficient to prevent
new writers from coming in and making a given transaction run forever.
This adds an explicit wait at the end of most operations so they will
allow the current transaction to close.
There is no wait inside file_write, inode updates, or cow filling, all which
have different deadlock possibilities.
This is a temporary measure until better asynchronous commit support is
added. This code leads to stalls as it waits for data=ordered
writeback, and it really needs to be fixed.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
One lock per btree block can make for significant congestion if everyone
has to wait for IO at the high levels of the btree. This drops
locks held by a path when doing reads during a tree search.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
Extent alloctions are still protected by a large alloc_mutex.
Objectid allocations are covered by a objectid mutex
Other btree operations are protected by a lock on individual btree nodes
Signed-off-by: Chris Mason <chris.mason@oracle.com>
The allocation trees and the chunk trees are serialized via their own
dedicated mutexes. This means allocation location is still not very
fine grained.
The main FS btree is protected by locks on each block in the btree. Locks
are taken top / down, and as processing finishes on a given level of the
tree, the lock is released after locking the lower level.
The end result of a search is now a path where only the lowest level
is locked. Releasing or freeing the path drops any locks held.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
If a bio submission is after a lock holder waiting for the bio
on the work queue, it is possible to deadlock. Move the bios
into their own pool.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
mount -o thread_pool_size changes the default, which is
min(num_cpus + 2, 8). Larger thread pools would make more sense on
very large disk arrays.
This mount option controls the max size of each thread pool. There
are multiple thread pools, so the total worker count will be larger
than the mount option.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
Btrfs has been using workqueues to spread the checksumming load across
other CPUs in the system. But, workqueues only schedule work on the
same CPU that queued the work, giving them a limited benefit for systems with
higher CPU counts.
This code adds a generic facility to schedule work with pools of kthreads,
and changes the bio submission code to queue bios up. The queueing is
important to make sure large numbers of procs on the system don't
turn streaming workloads into random workloads by sending IO down
concurrently.
The end result of all of this is much higher performance (and CPU usage) when
doing checksumming on large machines. Two worker pools are created,
one for writes and one for endio processing. The two could deadlock if
we tried to service both from a single pool.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
Also adds lots of comments to describe what's going on here.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Chris Mason <chris.mason@oracle.com>
This avoids IO stalls and poorly ordered IO from inline writers mixing in
with the async submission queue
Signed-off-by: Chris Mason <chris.mason@oracle.com>
The async submit workqueue was absorbing too many requests, leading to long
stalls where the async submitters were stalling.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
Devices can change after the scan ioctls are done, and btrfs_open_devices
needs to be able to verify them as they are opened and used by the FS.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
When duplicate copies exist, writes are allowed to fail to one of those
copies. This changeset includes a few changes that allow the FS to
continue even when some IOs fail.
It also adds verification of the parent generation number for btree blocks.
This generation is stored in the pointer to a block, and it ensures
that missed writes to are detected.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
This required a few structural changes to the code that manages bdev pointers:
The VFS super block now gets an anon-bdev instead of a pointer to the
lowest bdev. This allows us to avoid swapping the super block bdev pointer
around at run time.
The code to read in the super block no longer goes through the extent
buffer interface. Things got ugly keeping the mapping constant.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
2.6.18 seems to get caught in an infinite loop when
cancel_rearming_delayed_workqueue is called more than once, so this switches
to cancel_delayed_work, which is arguably more correct.
Also, balance_dirty_pages can run into problems with 2.6.18 based kernels
because it doesn't have the per-bdi dirty limits. This avoids calling
balance_dirty_pages on the btree inode unless there is actually something
to balance, which is a good optimization in general.
Finally there's a compile fix for ordered-data.h
Signed-off-by: Chris Mason <chris.mason@oracle.com>
btrfs_invalidatepage is not allowed to leave pages around on the lru.
Any such pages will trigger an oops later on because the VM will see
page->private and assume it is a buffer head.
This also forces extra flushes of the async work queues before
dropping all the pages on the btree inode during unmount. Left over
items on the work queues are one possible cause of busy state ranges
during truncate_inode_pages.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
The btree inode should only have a single extent_map in the cache,
it doesn't make sense to ever drop it.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
This allows checksumming to happen in parallel among many cpus, and
keeps us from bogging down pdflush with the checksumming code.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
Block headers now store the chunk tree uuid
Chunk items records the device uuid for each stripes
Device extent items record better back refs to the chunk tree
Block groups record better back refs to the chunk tree
The chunk tree format has also changed. The objectid of BTRFS_CHUNK_ITEM_KEY
used to be the logical offset of the chunk. Now it is a chunk tree id,
with the logical offset being stored in the offset field of the key.
This allows a single chunk tree to record multiple logical address spaces,
upping the number of bytes indexed by a chunk tree from 2^64 to
2^128.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
This includes fixing a missing spinlock init call that caused oops on mount
for most kernels other than 2.6.25.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
Fix for a endianess BUG when using btrfs v0.13 with kernels older than 2.6.23
Problem:
Has of v0.13, btrfs-progs is using crc32c.c equivalent to the one found on
linux-2.6.23/lib/libcrc32c.c Since crc32c_le() changed in linux-2.6.23, when
running btrfs v0.13 with older kernels we have a missmatch between the versions
of crc32c_le() from btrfs-progs and libcrc32c in the kernel. This missmatch
causes a bug when using btrfs on big endian machines.
Solution:
btrfs_crc32c() macro that when compiling for kernels older than 2.6.23, does
endianess conversion to parameters and return value of crc32c().
This endianess conversion nullifies the differences in implementation
of crc32c_le().
If kernel 2.6.23 or better, it calls crc32c().
Signed-off-by: Miguel Sousa Filipe <miguel.filipe@gmail.com>
---
Signed-off-by: Chris Mason <chris.mason@oracle.com>
Before it was done by the bio end_io routine, the work queue code is able
to scale much better with faster IO subsystems.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
Before, metadata checksumming was done by the callers of read_tree_block,
which would set EXTENT_CSUM bits in the extent tree to show that a given
range of pages was already checksummed and didn't need to be verified
again.
But, those bits could go away via try_to_releasepage, and the end
result was bogus checksum failures on pages that never left the cache.
The new code validates checksums when the page is read. It is a little
tricky because metadata blocks can span pages and a single read may
end up going via multiple bios.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
Checksums were only verified by btrfs_read_tree_block, which meant the
functions to probe the page cache for blocks were not validating checksums.
Normally this is fine because the buffers will only be in cache if they
have already been validated.
But, there is a window while the buffer is being read from disk where
it could be up to date in the cache but not yet verified. This patch
makes sure all buffers go through checksum verification before they
are used.
This is safer, and it prevents modification of buffers before they go
through the csum code.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
This allows detection of blocks that have already been written in the
running transaction so they can be recowed instead of modified again.
It is step one in trusting the transid field of the block pointers.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
Reduce CPU time searching for free blocks by optimizing find_first_extent_bit
Fix find_free_extent to make better use of the last_alloc hint. Before it
was often finding blocks just before the hint.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
The end_bio routines are changed to take a pointer to the extent state
struct, and the state tree is walked in order to set/clear appropriate
bits as IO completes. This greatly reduces the number of rbtree searches
done by the end_bio handlers, and reduces lock contention.
The extent_io releasepage function is changed to avoid expensive searches
for locked state.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
There is now extent_map for mapping offsets in the file to disk and
extent_io for state tracking, IO submission and extent_bufers.
The new extent_map code shifts from [start,end] pairs to [start,len], and
pushes the locking out into the caller. This allows a few performance
optimizations and is easier to use.
A number of extent_map usage bugs were fixed, mostly with failing
to remove extent_map entries when changing the file.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
This forces file data extents down the disk along with the metadata that
references them. The current implementation is fairly simple, and just
writes out all of the dirty pages in an inode before the commit.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
This is intended to prevent accidentally filling the drive. A determined
user can still make things oops.
It includes some accounting of the current bytes under delayed allocation,
but this will change as things get optimized
Signed-off-by: Chris Mason <chris.mason@oracle.com>
This patch adds a helper function 'update_pinned_extents' to
extent-tree.c. The usage of the helper function is similar to
'update_block_group', the last parameter of the function indicates
pin vs unpin.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
btrfs_btree_balance_dirty is changed to pass the number of pages dirtied
for more accurate dirty throttling. This lets the VM make better decisions
about when to force some writeback.
Signed-off-by: Chris 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: Chris 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: Chris Mason <chris.mason@oracle.com>
