Implement merge method for the snapshot origin to improve read
performance.

Without merge method, dm asks the upper layers to submit smallest possible
bios --- one page. Submitting such small bios impacts performance negatively
when reading or writing the origin device.

Without this patch, CPU consumption when reading the origin on lvm on md-raid0
was 6 to 12%, with this patch, it drops to 1 to 4%.

Note: in my testing, it actually degraded performance in some settings, I
traced it to Maxtor disks having problems with > 512-sector requests.
Reducing the number of sectors to /sys/block/sd*/queue/max_sectors_kb to
256 fixed the read performance. I think we don't have to care about weird
disks that actually degrade performance because of large requests being
sent to them.
Signed-off-by: NMikulas Patocka <mpatocka@redhat.com>
Signed-off-by: NAlasdair G Kergon <agk@redhat.com>

Change bio-based mapped devices no longer to have a fully initialized
request_queue (request_fn, elevator, etc).  This means bio-based DM
devices no longer register elevator sysfs attributes ('iosched/' tree
or 'scheduler' other than "none").

In contrast, a request-based DM device will continue to have a full
request_queue and will register elevator sysfs attributes.  Therefore
a user can determine a DM device's type by checking if elevator sysfs
attributes exist.

First allocate a minimalist request_queue structure for a DM device
(needed for both bio and request-based DM).

Initialization of a full request_queue is deferred until it is known
that the DM device is request-based, at the end of the table load
sequence.

Factor DM device's request_queue initialization:
- common to both request-based and bio-based into dm_init_md_queue().
- specific to request-based into dm_init_request_based_queue().

The md->type_lock mutex is used to protect md->queue, in addition to
md->type, during table_load().

A DM device's first table_load will establish the immutable md->type.
But md->queue initialization, based on md->type, may fail at that time
(because blk_init_allocated_queue cannot allocate memory).  Therefore
any subsequent table_load must (re)try dm_setup_md_queue independently of
establishing md->type.
Signed-off-by: NMike Snitzer <snitzer@redhat.com>
Acked-by: NKiyoshi Ueda <k-ueda@ct.jp.nec.com>
Signed-off-by: NAlasdair G Kergon <agk@redhat.com>

Determine whether a mapped device is bio-based or request-based when
loading its first (inactive) table and don't allow that to be changed
later.

This patch performs different device initialisation in each of the two
cases.  (We don't think it's necessary to add code to support changing
between the two types.)

Allowed md->type transitions:
  DM_TYPE_NONE to DM_TYPE_BIO_BASED
  DM_TYPE_NONE to DM_TYPE_REQUEST_BASED

We now prevent table_load from replacing the inactive table with a
conflicting type of table even after an explicit table_clear.

Introduce 'type_lock' into the struct mapped_device to protect md->type
and to prepare for the next patch that will change the queue
initialization and allocate memory while md->type_lock is held.
Signed-off-by: NMike Snitzer <snitzer@redhat.com>
Acked-by: NKiyoshi Ueda <k-ueda@ct.jp.nec.com>
Signed-off-by: NAlasdair G Kergon <agk@redhat.com>

 drivers/md/dm-ioctl.c    |   15 +++++++++++++++
 drivers/md/dm.c          |   37 ++++++++++++++++++++++++++++++-------
 drivers/md/dm.h          |    5 +++++
 include/linux/dm-ioctl.h |    4 ++--
 4 files changed, 52 insertions(+), 9 deletions(-)

When processing barriers, skip the second flush if processing the bio
failed with -EOPNOTSUPP.  This can happen with discard+barrier requests.
If the device doesn't support discard, there would be two useless
SYNCHRONIZE CACHE commands.  The first dm_flush cannot be so easily
optimized out, so we leave it there.

Previously, -EOPNOTSUPP could be received in dec_pending only with empty
barriers and we ignored that error, assuming the device not supporting
cache flushes has cache always consistent.  With the addition of discard
barriers, this -EOPNOTSUPP can also be generated by discards and we
must record it in md->barrier_error for process_barrier.
Signed-off-by: NMikulas Patocka <mpatocka@redhat.com>
Signed-off-by: NAlasdair G Kergon <agk@redhat.com>

This patch fixes hard-coded value for the size of a chunk that includes
disk header for persistent snapshot. It should be changed to existing
macro NUM_SNAPSHOT_HDR_CHUNKS instead of using hard-coded value 1.
Signed-off-by: NTomohiro Kusumi <kusumi.tomohiro@jp.fujitsu.com>
Signed-off-by: NMikulas Patocka <mpatocka@redhat.com>
Signed-off-by: NAlasdair G Kergon <agk@redhat.com>

Use kstrdup when the goal of an allocation is copy a string into the
allocated region.

The semantic patch that makes this change is as follows:
(http://coccinelle.lip6.fr/)

// <smpl>
@@
expression from,to;
expression flag,E1,E2;
statement S;
@@

-  to = kmalloc(strlen(from) + 1,flag);
+  to = kstrdup(from, flag);
   ... when != \(from = E1 \| to = E1 \)
   if (to==NULL || ...) S
   ... when != \(from = E2 \| to = E2 \)
-  strcpy(to, from);
// </smpl>
Signed-off-by: NJulia Lawall <julia@diku.dk>
Signed-off-by: NAlasdair G Kergon <agk@redhat.com>

The dm control device does not implement read/write, so it has no use for
seeking.  Using no_llseek prevents falling back to default_llseek, which
requires the BKL.
Signed-off-by: NArnd Bergmann <arnd@arndb.de>
Signed-off-by: NFrederic Weisbecker <fweisbec@gmail.com>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NAlasdair G Kergon <agk@redhat.com>

This patch separates the device deletion code from dm_put()
to make sure the deletion happens in the process context.

By this patch, device deletion always occurs in an ioctl (process)
context and dm_put() can be called in interrupt context.
As a result, the request-based dm's bad dm_put() usage pointed out
by Mikulas below disappears.
    http://marc.info/?l=dm-devel&m=126699981019735&w=2

Without this patch, I confirmed there is a case to crash the system:
    dm_put() => dm_table_destroy() => vfree() => BUG_ON(in_interrupt())

Some more backgrounds and details:
In request-based dm, a device opener can remove a mapped_device
while the last request is still completing, because bios in the last
request complete first and then the device opener can close and remove
the mapped_device before the last request completes:
  CPU0                                          CPU1
  =================================================================
  <<INTERRUPT>>
  blk_end_request_all(clone_rq)
    blk_update_request(clone_rq)
      bio_endio(clone_bio) == end_clone_bio
        blk_update_request(orig_rq)
          bio_endio(orig_bio)
                                                <<I/O completed>>
                                                dm_blk_close()
                                                dev_remove()
                                                  dm_put(md)
                                                    <<Free md>>
   blk_finish_request(clone_rq)
     ....
     dm_end_request(clone_rq)
       free_rq_clone(clone_rq)
       blk_end_request_all(orig_rq)
       rq_completed(md)

So request-based dm used dm_get()/dm_put() to hold md for each I/O
until its request completion handling is fully done.
However, the final dm_put() can call the device deletion code which
must not be run in interrupt context and may cause kernel panic.

To solve the problem, this patch moves the device deletion code,
dm_destroy(), to predetermined places that is actually deleting
the mapped_device in ioctl (process) context, and changes dm_put()
just to decrement the reference count of the mapped_device.
By this change, dm_put() can be used in any context and the symmetric
model below is introduced:
    dm_create():  create a mapped_device
    dm_destroy(): destroy a mapped_device
    dm_get():     increment the reference count of a mapped_device
    dm_put():     decrement the reference count of a mapped_device

dm_destroy() waits for all references of the mapped_device to disappear,
then deletes the mapped_device.

dm_destroy() uses active waiting with msleep(1), since deleting
the mapped_device isn't performance-critical task.
And since at this point, nobody opens the mapped_device and no new
reference will be taken, the pending counts are just for racing
completing activity and will eventually decrease to zero.

For the unlikely case of the forced module unload, dm_destroy_immediate(),
which doesn't wait and forcibly deletes the mapped_device, is also
introduced and used in dm_hash_remove_all().  Otherwise, "rmmod -f"
may be stuck and never return.
And now, because the mapped_device is deleted at this point, subsequent
accesses to the mapped_device may cause NULL pointer references.

Cc: stable@kernel.org
Signed-off-by: NKiyoshi Ueda <k-ueda@ct.jp.nec.com>
Signed-off-by: NJun'ichi Nomura <j-nomura@ce.jp.nec.com>
Signed-off-by: NAlasdair G Kergon <agk@redhat.com>

This patch changes dm_hash_remove_all() to release _hash_lock when
removing a device.  After removing the device, dm_hash_remove_all()
takes _hash_lock and searches the hash from scratch again.

This patch is a preparation for the next patch, which changes device
deletion code to wait for md reference to be 0.  Without this patch,
the wait in the next patch may cause AB-BA deadlock:
  CPU0                                CPU1
  -----------------------------------------------------------------------
  dm_hash_remove_all()
    down_write(_hash_lock)
                                      table_status()
                                        md = find_device()
                                               dm_get(md)
                                                 <increment md->holders>
                                        dm_get_live_or_inactive_table()
                                          dm_get_inactive_table()
                                            down_write(_hash_lock)
    <in the md deletion code>
      <wait for md->holders to be 0>
Signed-off-by: NKiyoshi Ueda <k-ueda@ct.jp.nec.com>
Signed-off-by: NJun'ichi Nomura <j-nomura@ce.jp.nec.com>
Cc: stable@kernel.org
Signed-off-by: NAlasdair G Kergon <agk@redhat.com>

This patch prevents access to mapped_device which is being deleted.

Currently, even after a mapped_device has been removed from the hash,
it could be accessed through idr_find() using minor number.
That could cause a race and NULL pointer reference below:
  CPU0                          CPU1
  ------------------------------------------------------------------
  dev_remove(param)
    down_write(_hash_lock)
    dm_lock_for_deletion(md)
      spin_lock(_minor_lock)
      set_bit(DMF_DELETING)
      spin_unlock(_minor_lock)
    __hash_remove(hc)
    up_write(_hash_lock)
                                dev_status(param)
                                  md = find_device(param)
                                         down_read(_hash_lock)
                                         __find_device_hash_cell(param)
                                           dm_get_md(param->dev)
                                             md = dm_find_md(dev)
                                                    spin_lock(_minor_lock)
                                                    md = idr_find(MINOR(dev))
                                                    spin_unlock(_minor_lock)
    dm_put(md)
      free_dev(md)
                                             dm_get(md)
                                         up_read(_hash_lock)
                                  __dev_status(md, param)
                                  dm_put(md)

This patch fixes such problems.
Signed-off-by: NKiyoshi Ueda <k-ueda@ct.jp.nec.com>
Signed-off-by: NJun'ichi Nomura <j-nomura@ce.jp.nec.com>
Cc: stable@kernel.org
Signed-off-by: NAlasdair G Kergon <agk@redhat.com>

All the dm ioctls that generate uevents set the DM_UEVENT_GENERATED flag so
that userspace knows whether or not to wait for a uevent to be processed
before continuing,

The dm rename ioctl sets this flag but was not structured to return it
to userspace.  This patch restructures the rename ioctl processing to
behave like the other ioctls that return data and so fix this.
Signed-off-by: NPeter Rajnoha <prajnoha@redhat.com>
Signed-off-by: NAlasdair G Kergon <agk@redhat.com>

__dev_status() cannot fail so make it void and simplify callers.
Signed-off-by: NAlasdair G Kergon <agk@redhat.com>

Remove useless __dev_status call while processing an ioctl that sets up
device geometry and target message.  The data is not returned to
userspace so there is no point collecting it and in the case of
target_message it is collected before processing the message so if it
did return it might be stale.
Signed-off-by: NPeter Rajnoha <prajnoha@redhat.com>
Signed-off-by: NAlasdair G Kergon <agk@redhat.com>

Validate chunk size against both origin and snapshot sector size

Don't allow chunk size smaller than either origin or snapshot logical
sector size. Reading or writing data not aligned to sector size is not
allowed and causes immediate errors.

This requires us to open the origin before initialising the
exception store and to export dm_snap_origin.

Cc: stable@kernel.org
Signed-off-by: NMikulas Patocka <mpatocka@redhat.com>
Reviewed-by: NMike Snitzer <snitzer@redhat.com>
Signed-off-by: NAlasdair G Kergon <agk@redhat.com>

Iterate both origin and snapshot devices

iterate_devices method should call the callback for all the devices where
the bio may be remapped. Thus, snapshot_iterate_devices should call the callback
for both snapshot and origin underlying devices because it remaps some bios
to the snapshot and some to the origin.

snapshot_iterate_devices called the callback only for the origin device.
This led to badly calculated device limits if snapshot and origin were placed
on different types of disks.

Cc: stable@kernel.org
Signed-off-by: NMikulas Patocka <mpatocka@redhat.com>
Reviewed-by: NMike Snitzer <snitzer@redhat.com>
Signed-off-by: NAlasdair G Kergon <agk@redhat.com>

multipath_ctr() forgets to return an error after detecting
missing path parameters.  Fix this.
Signed-off-by: NPatrick LoPresti <lopresti@gmail.com>
Cc: stable@kernel.org
Signed-off-by: NAlasdair G Kergon <agk@redhat.com>

There is only one error exit from do_md_stop, so make that more
explicit and discard the 'err' variable.
Also drop the 'revalidate' variable by moving the unlock calls around.
Signed-off-by: NNeilBrown <neilb@suse.de>

Move the deletion of sysfs attributes from reconfig_mutex to
open_mutex didn't really help as a process can try to take
open_mutex while holding reconfig_mutex, so the same deadlock can
happen, just requiring one more process to be involved in the chain.

I looks like I cannot easily use locking to wait for the sysfs
deletion to complete, so don't.

The only things that we cannot do while the deletions are still
pending is other things which can change the sysfs namespace: run,
takeover, stop.  Each of these can fail with -EBUSY.
So set a flag while doing a sysfs deletion, and fail run, takeover,
stop if that flag is set.

This is suitable for 2.6.35.x

Cc: stable@kernel.org
Signed-off-by: NNeilBrown <neilb@suse.de>

Commit b821eaa5 "md: remove ->changed and related code" moved
revalidate_disk() under open_mutex, and lockdep noticed.

[ INFO: possible circular locking dependency detected ]
2.6.32-mdadm-locking #1
-------------------------------------------------------
mdadm/3640 is trying to acquire lock:
 (&bdev->bd_mutex){+.+.+.}, at: [<ffffffff811acecb>] revalidate_disk+0x5b/0x90

but task is already holding lock:
 (&mddev->open_mutex){+.+...}, at: [<ffffffffa055e07a>] do_md_stop+0x4a/0x4d0 [md_mod]

which lock already depends on the new lock.

It is suitable for 2.6.35.x

Cc: <stable@kernel.org>
Reported-by: NPrzemyslaw Czarnowski <przemyslaw.hawrylewicz.czarnowski@intel.com>
Signed-off-by: NDan Williams <dan.j.williams@intel.com>
Signed-off-by: NNeilBrown <neilb@suse.de>

The open and release block_device_operations are currently
called with the BKL held. In order to change that, we must
first make sure that all drivers that currently rely
on this have no regressions.

This blindly pushes the BKL into all .open and .release
operations for all block drivers to prepare for the
next step. The drivers can subsequently replace the BKL
with their own locks or remove it completely when it can
be shown that it is not needed.

The functions blkdev_get and blkdev_put are the only
remaining users of the big kernel lock in the block
layer, besides a few uses in the ioctl code, none
of which need to serialize with blkdev_{get,put}.

Most of these two functions is also under the protection
of bdev->bd_mutex, including the actual calls to
->open and ->release, and the common code does not
access any global data structures that need the BKL.
Signed-off-by: NArnd Bergmann <arnd@arndb.de>
Acked-by: NChristoph Hellwig <hch@infradead.org>
Signed-off-by: NJens Axboe <jaxboe@fusionio.com>

This removes q->prepare_flush_fn completely (changes the
blk_queue_ordered API).
Signed-off-by: NFUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp>
Reviewed-by: NChristoph Hellwig <hch@lst.de>
Signed-off-by: NJens Axboe <jaxboe@fusionio.com>

use REQ_FLUSH flag instead.
Signed-off-by: NFUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp>
Cc: Alasdair G Kergon <agk@redhat.com>
Reviewed-by: NChristoph Hellwig <hch@lst.de>
Signed-off-by: NJens Axboe <jaxboe@fusionio.com>

Remove the current bio flags and reuse the request flags for the bio, too.
This allows to more easily trace the type of I/O from the filesystem
down to the block driver.  There were two flags in the bio that were
missing in the requests:  BIO_RW_UNPLUG and BIO_RW_AHEAD.  Also I've
renamed two request flags that had a superflous RW in them.

Note that the flags are in bio.h despite having the REQ_ name - as
blkdev.h includes bio.h that is the only way to go for now.
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Signed-off-by: NJens Axboe <jaxboe@fusionio.com>

Remove all the trivial wrappers for the cmd_type and cmd_flags fields in
struct requests.  This allows much easier grepping for different request
types instead of unwinding through macros.
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Signed-off-by: NJens Axboe <jaxboe@fusionio.com>

If the 'bio_split' path in raid10-read is used while
resync/recovery is happening it is possible to deadlock.
Fix this be elevating ->nr_waiting for the duration of both
parts of the split request.

This fixes a bug that has been present since 2.6.22
but has only started manifesting recently for unknown reasons.
It is suitable for and -stable since then.
Reported-by: NJustin Bronder <jsbronder@gentoo.org>
Tested-by: NJustin Bronder <jsbronder@gentoo.org>
Signed-off-by: NNeilBrown <neilb@suse.de>
Cc: stable@kernel.org

dm makes this distinction between ->ctr and ->resume, so we need to
too.

Also get the new bitmap_load to clear out the bitmap first, as this is
most consistent with the dm suspend/resume approach
Signed-off-by: NNeilBrown <neilb@suse.de>

This allows md/raid5 to fully work as a dm target.

Normally md uses a 'filemap' which contains a list of pages of bits
each of which may be written separately.
dm-log uses and all-or-nothing approach to writing the log, so
when using a dm-log, ->filemap is NULL and the flags normally stored
in filemap_attr are stored in ->logattrs instead.
Signed-off-by: NNeilBrown <neilb@suse.de>

A bitmap is stored as one page per 2048 bits.
If none of the bits are set, the page is not allocated.

When bitmap_get_counter finds that a page isn't allocate,
it just reports that one bit work of space isn't flagged,
rather than reporting that 2048 bits worth of space are
unflagged.
This can cause searches for flagged bits (e.g. bitmap_close_sync)
to do more work than is really necessary.

So change bitmap_get_counter (when creating) to report a number of
blocks that more accurately reports the range of the device for which
no counter currently exists.
Signed-off-by: NNeilBrown <neilb@suse.de>

1/ use md_unplug in bitmap.c as we will soon be using bitmaps under
  arrays with no queue attached.

2/ Don't bother plugging the queue when we set a bit in the bitmap.
   The reason for this was to encourage as many bits as possible to
   get set before we unplug and write stuff out.
   However every personality already plugs the queue after
   bitmap_startwrite either directly (raid1/raid10) or be setting
   STRIPE_BIT_DELAY which causes the queue to be plugged later
   (raid5).
Signed-off-by: NNeilBrown <neilb@suse.de>

For dm-raid45 we will want to use bitmaps in dm-targets which don't
have entries in sysfs, so cope with the mddev not living in sysfs.
Signed-off-by: NNeilBrown <neilb@suse.de>

Fixes some whitespace problems
Fixed some checkpatch.pl complaints.
Replaced kmalloc ... memset(0), with kzalloc
Fixed an unlikely memory leak on an error path.
Reformatted a number of 'if/else' sets, sometimes
replacing goto with an else clause.
Removed some old comments and commented-out code.
Signed-off-by: NNeilBrown <neilb@suse.de>

Also remove remaining accesses to ->queue and ->gendisk when ->queue
is NULL (As it is in a DM target).
Signed-off-by: NNeilBrown <neilb@suse.de>

If an array doesn't have a 'queue' then md_do_sync cannot
unplug it.
In that case it will have a 'plugger', so make that available
to the mddev, and use it to unplug the array if needed.
Signed-off-by: NNeilBrown <neilb@suse.de>

md/raid5 uses the plugging infrastructure provided by the block layer
and 'struct request_queue'.  However when we plug raid5 under dm there
is no request queue so we cannot use that.

So create a similar infrastructure that is much lighter weight and use
it for raid5.
Signed-off-by: NNeilBrown <neilb@suse.de>

the dm module will need this for dm-raid45.

Also only access ->queue->backing_dev_info->congested_fn
if ->queue actually exists.  It won't in a dm target.
Signed-off-by: NNeilBrown <neilb@suse.de>

dm-raid456 does not provide a 'queue' for raid5 to use,
so we must make raid5 stop depending on the queue.

First: read_ahead
dm handles read-ahead adjustment fully in userspace, so
simply don't do any readahead adjustments if there is
no queue.

Also re-arrange code slightly so all the accesses to ->queue are
together.

Finally, move the blk_queue_merge_bvec function into the 'if' as
the ->split_io setting in dm-raid456 has the same effect.
Signed-off-by: NNeilBrown <neilb@suse.de>

dm uses scheduled work to raise events to user-space.
So allow md device to have work_structs and schedule them on an error.
Signed-off-by: NNeilBrown <neilb@suse.de>

export entry points for starting and stopping md arrays.
This will be used by a module to make md/raid5 work under
dm.
Also stop calling md_stop_writes from md_stop, as that won't
work well with dm - it will want to call the two separately.
Signed-off-by: NNeilBrown <neilb@suse.de>

This functionality will be needed separately in a subsequent patch, so
split it into it's own exported function.
Signed-off-by: NNeilBrown <neilb@suse.de>

When MD_CHANGE_CLEAN is set we might block in md_write_start.
So we should only set it when fairly sure that something will clear
it.

There are two places where it is set so as to encourage a metadata
update to record the progress of resync/recovery.  This should only
be done if the internal metadata update mechanisms are in use, which
can be tested by by inspecting '->persistent'.
Signed-off-by: NNeilBrown <neilb@suse.de>