Need to check kthread_should_stop after schedule_timeout() before calling
schedule(). This causes threads to sleep with potentially no one to wake them
up causing mount(2) to hang in btrfs_stop_workers waiting for threads to stop.
Signed-off-by: Amit Gud <gud@ksu.edu>
Signed-off-by: Chris Mason <chris.mason@oracle.com>
The 'flushoncommit' mount option forces any data dirtied by a write in a
prior transaction to commit as part of the current commit. This makes
the committed state a fully consistent view of the file system from the
application's perspective (i.e., it includes all completed file system
operations). This was previously the behavior only when a snapshot is
created.
This is used by Ceph to ensure that completed writes make it to the
platter along with the metadata operations they are bound to (by
BTRFS_IOC_TRANS_{START,END}).
Signed-off-by: Sage Weil <sage@newdream.net>
Signed-off-by: Chris Mason <chris.mason@oracle.com>
Add a 'notreelog' mount option to disable the tree log (used by fsync,
O_SYNC writes). This is much slower, but the tree logging produces
inconsistent views into the FS for ceph.
Signed-off-by: Sage Weil <sage@newdream.net>
Signed-off-by: Chris Mason <chris.mason@oracle.com>
btrfs options can change at times other than mount, yet /proc/mounts shows the
options string used when the fs was mounted (an example would be when btrfs
determines that barriers aren't useful and turns them off.) This patch
instead outputs the actual options in use by btrfs.
Signed-off-by: Eric Paris <eparis@redhat.com>
Signed-off-by: Chris Mason <chris.mason@oracle.com>
Because btrfs is copy-on-write, we end up picking new locations for
blocks very often. This makes it fairly difficult to maintain perfect
read patterns over time, but we can at least do some optimizations
for writes.
This is done today by remembering the last place we allocated and
trying to find a free space hole big enough to hold more than just one
allocation. The end result is that we tend to write sequentially to
the drive.
This happens all the time for metadata and it happens for data
when mounted -o ssd. But, the way we record it is fairly racey
and it tends to fragment the free space over time because we are trying
to allocate fairly large areas at once.
This commit gets rid of the races by adding a free space cluster object
with dedicated locking to make sure that only one process at a time
is out replacing the cluster.
The free space fragmentation is somewhat solved by allowing a cluster
to be comprised of smaller free space extents. This part definitely
adds some CPU time to the cluster allocations, but it allows the allocator
to consume the small holes left behind by cow.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
btrfs_next_leaf was using blocking locks when it could have been using
faster spinning ones instead. This adds a few extra checks around
the pieces that block and switches over to spinning locks.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
This patch removes the pinned_mutex. The extent io map has an internal tree
lock that protects the tree itself, and since we only copy the extent io map
when we are committing the transaction we don't need it there. We also don't
need it when caching the block group since searching through the tree is also
protected by the internal map spin lock.
Signed-off-by: Josef Bacik <jbacik@redhat.com>
This patch removes the block group alloc mutex used to protect the free space
tree for allocations and replaces it with a spin lock which is used only to
protect the free space rb tree. This means we only take the lock when we are
directly manipulating the tree, which makes us a touch faster with
multi-threaded workloads.
This patch also gets rid of btrfs_find_free_space and replaces it with
btrfs_find_space_for_alloc, which takes the number of bytes you want to
allocate, and empty_size, which is used to indicate how much free space should
be at the end of the allocation.
It will return an offset for the allocator to use. If we don't end up using it
we _must_ call btrfs_add_free_space to put it back. This is the tradeoff to
kill the alloc_mutex, since we need to make sure nobody else comes along and
takes our space.
Signed-off-by: Josef Bacik <jbacik@redhat.com>
I've replaced the strange looping constructs with a list_for_each_entry on
space_info->block_groups. If we have a hint we just jump into the loop with
the block group and start looking for space. If we don't find anything we
start at the beginning and start looking. We never come out of the loop with a
ref on the block_group _unless_ we found space to use, then we drop it after we
set the trans block_group.
Signed-off-by: Josef Bacik <jbacik@redhat.com>
This patch cleans up the free space cache code a bit. It better documents the
idiosyncrasies of tree_search_offset and makes the code make a bit more sense.
I took out the info allocation at the start of __btrfs_add_free_space and put it
where it makes more sense. This was left over cruft from when alloc_mutex
existed. Also all of the re-searches we do to make sure we inserted properly.
Signed-off-by: Josef Bacik <jbacik@redhat.com>
Btrfs pages being written get set to writeback, and then may go through
a number of steps before they hit the block layer. This includes compression,
checksumming and async bio submission.
The end result is that someone who writes a page and then does
wait_on_page_writeback is likely to unplug the queue before the bio they
cared about got there.
We could fix this by marking bios sync, or by doing more frequent unplugs,
but this commit just changes the async bio submission code to unplug
after it has processed all the bios for a device. The async bio submission
does a fair job of collection bios, so this shouldn't be a huge problem
for reducing merging at the elevator.
For streaming O_DIRECT writes on a 5 drive array, it boosts performance
from 386MB/s to 460MB/s.
Thanks to Hisashi Hifumi for helping with this work.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
Btrfs uses async helper threads to submit write bios so the checksumming
helper threads don't block on the disk.
The submit bio threads may process bios for more than one block device,
so when they find one device congested they try to move on to other
devices instead of blocking in get_request_wait for one device.
This does a pretty good job of keeping multiple devices busy, but the
congested flag has a number of problems. A congested device may still
give you a request, and other procs that aren't backing off the congested
device may starve you out.
This commit uses the io_context stored in current to decide if our process
has been made a batching process by the block layer. If so, it keeps
sending IO down for at least one batch. This helps make sure we do
a good amount of work each time we visit a bdev, and avoids large IO
stalls in multi-device workloads.
It's also very ugly. A better solution is in the works with Jens Axboe.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
Impact: cleanup
Rather than overriding MODULE_PARAM_PREFIX, build via acpi.o so
KBUILD_MODNAME is set to "acpi".
This is the logical way to do it, even though acpi cannot be a module
due to these config options being bool. Those parts of ACPI which can
be modular are not built into the acpi "module".
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Signed-off-by: Len Brown <len.brown@intel.com>
The device_type "soc" is still required for MPC85xx boards.
Signed-off-by: Wolfgang Grandegger <wg@grandegger.com>
Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
Set queue ordered mode. It doesn't really matter what we set here
because we don't ever put any requests on the queue. But we need to set
something other than QUEUE_ORDERED_NONE so that __generic_make_request
passes barrier requests to us.
Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
Move wait queue declaration and unplug to dm_wait_for_completion.
The purpose is to minimize duplicate code in the further patches.
The patch reorders functions a little bit. It doesn't change any
functionality. For proper non-deadlock operation, add_wait_queue must
happen before set_current_state(interruptible) and before the test for
!atomic_read(&md->pending).
Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
Merge pushback and deferred lists into one list - use deferred list
for both deferred and pushed-back bios.
This will be needed for proper support of barrier bios: it is impossible to
support ordering correctly with two lists because the requests on both lists
will be mixed up.
Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
Allow uninterruptible wait for pending IOs.
Add argument "interruptible" to dm_wait_for_completion that specifies
either interruptible or uninterruptible waiting.
Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
Merge __flush_deferred_io() into the only caller, dm_wq_work().
There's no need to have a function that has only one caller.
Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
Move the bio_io_error() calls directly into __split_and_process_bio().
This avoids some code duplication in later patches.
Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
Rename __split_bio() to __split_and_process_bio() because it not only splits
the bio to serveral parts, but also submits them to target drivers.
Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
Remove struct dm_wq_req and move "work" directly into struct mapped_device.
In the revised implementation, the thread will do just one type of work
(processing the queue).
Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
Remove the context field from struct dm_wq_req because we will no longer
need it.
Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
Remove "type" field from struct dm_wq_req because we no longer need it
to have more than one value.
Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
Introduce a function that adds a bio to the head of the list for
use by the patch that will support barriers.
Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
The persistent exception store destructor does not properly
account for all conditions in which it can be called. If it
is called after 'ctr' but before 'read_metadata' (e.g. if
something else in 'snapshot_ctr' fails) then it will attempt
to free areas of memory that haven't been allocated yet.
Signed-off-by: Jonathan Brassow <jbrassow@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
Let the exception store types print out their status through
the new API, rather than having the snapshot code do it.
Adjust the buffer position to allow for the preceding DMEMIT in the
arguments to type->status().
Signed-off-by: Jonathan Brassow <jbrassow@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
First step of having the exception stores parse their own arguments -
generalizing the interface.
Signed-off-by: Jonathan Brassow <jbrassow@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
Use DMEMIT in place of snprintf. This makes it easier later when
other modules are helping to populate our status output.
Signed-off-by: Jonathan Brassow <jbrassow@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
Move some of the last bits from dm-snap.h into dm-snap.c where they
belong and remove dm-snap.h.
Signed-off-by: Jonathan Brassow <jbrassow@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
Move useful functions out of dm-snap.h and stop using dm-snap.h.
Signed-off-by: Jonathan Brassow <jbrassow@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
Move COW device from snapshot to exception store.
Signed-off-by: Jonathan Brassow <jbrassow@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
Move chunk fields from snapshot to exception store.
Signed-off-by: Jonathan Brassow <jbrassow@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
Move target pointer from snapshot to exception store.
Signed-off-by: Jonathan Brassow <jbrassow@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
The logging API needs an extra function to make cluster mirroring
possible. This new function allows us to check whether a mirror
region is being recovered on another machine in the cluster. This
helps us prevent simultaneous recovery I/O and process I/O to the
same locations on disk.
Cluster-aware log modules will implement this function. Single
machine log modules will not. So, there is no performance
penalty for single machine mirrors.
Signed-off-by: Jonathan Brassow <jbrassow@redhat.com>
Acked-by: Heinz Mauelshagen <heinzm@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
Remove the 'dm_dirty_log_internal' structure. The resulting cleanup
eliminates extra memory allocations. Therefore exposing the internal
list_head to the external 'dm_dirty_log_type' structure is a worthwhile
compromise.
Signed-off-by: Mike Snitzer <snitzer@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
Avoid private module usage accounting by removing 'use' from
dm_dirty_log_internal. The standard module reference counting is
sufficient.
Signed-off-by: Mike Snitzer <snitzer@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
Use kzfree() instead of memset() + kfree().
Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Reviewed-by: Pekka Enberg <penberg@cs.helsinki.fi>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
The tt_internal is really just a list_head to manage registered target_type
in a double linked list,
Here embed the list_head into target_type directly,
1. to avoid kmalloc/kfree;
2. then tt_internal is really unneeded;
Cc: stable@kernel.org
Signed-off-by: Cheng Renquan <crquan@gmail.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
Reviewed-by: Alasdair G Kergon <agk@redhat.com>
upgrade_mode() sets bdev to NULL temporarily, and does not have any
locking to exclude anything from seeing that NULL.
In dm_table_any_congested() bdev_get_queue() can dereference that NULL and
cause a reported oops.
Fix this by not changing that field during the mode upgrade.
Cc: stable@kernel.org
Cc: Neil Brown <neilb@suse.de>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
The tt_internal's 'use' field is superfluous: the module's refcount can do
the work properly. An acceptable side-effect is that this increases the
reference counts reported by 'lsmod'.
Remove the superfluous test when removing a target module.
[Crash possible without this on SMP - agk]
Cc: stable@kernel.org
Signed-off-by: Cheng Renquan <crquan@gmail.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
Reviewed-by: Alasdair G Kergon <agk@redhat.com>
Reviewed-by: Jonathan Brassow <jbrassow@redhat.com>
We need to check if the exception was completed after dropping the lock.
After regaining the lock, __find_pending_exception checks if the exception
was already placed into &s->pending hash.
But we don't check if the exception was already completed and placed into
&s->complete hash. If the process waiting in alloc_pending_exception was
delayed at this point because of a scheduling latency and the exception
was meanwhile completed, we'd miss that and allocate another pending
exception for already completed chunk.
It would lead to a situation where two records for the same chunk exist
and potential data corruption because multiple snapshot I/Os to the
affected chunk could be redirected to different locations in the
snapshot.
Cc: stable@kernel.org
Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
It is uncommon and bug-prone to drop a lock in a function that is called with
the lock held, so this is moved to the caller.
Cc: stable@kernel.org
Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
Move looking-up of a pending exception from __find_pending_exception to another
function.
Cc: stable@kernel.org
Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
If someone sends signal to a process performing synchronous dm-io call,
the kernel may crash.
The function sync_io attempts to exit with -EINTR if it has pending signal,
however the structure "io" is allocated on stack, so already submitted io
requests end up touching unallocated stack space and corrupting kernel memory.
sync_io sets its state to TASK_UNINTERRUPTIBLE, so the signal can't break out
of io_schedule() --- however, if the signal was pending before sync_io entered
while (1) loop, the corruption of kernel memory will happen.
There is no way to cancel in-progress IOs, so the best solution is to ignore
signals at this point.
Cc: stable@kernel.org
Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
With my previous patch to save bi_io_vec, the size of dm_raid1_read_record
is significantly increased (the vector list takes 3072 bytes on 32-bit machines
and 4096 bytes on 64-bit machines).
The structure dm_raid1_read_record used to be allocated with kmalloc,
but kmalloc aligns the size on the next power-of-two so an object
slightly greater than 4096 will allocate 8192 bytes of memory and half of
that memory will be wasted.
This patch turns kmalloc into a slab cache which doesn't have this
padding so it will reduce the memory consumed.
Cc: stable@kernel.org
Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
