If there is a replacement device, then recover to it,
reading from any drives - maybe the one being replaced, maybe not.
Signed-off-by: NNeilBrown <neilb@suse.de>

If we need to resync an array which has replacement devices,
we always write any block checked to every replacement.

If the resync was bitmap-based resync we will then complete the
replacement normally.
If it was a full resync, we mark the replacements as fully recovered
when the resync finishes so no further recovery is needed.
Signed-off-by: NNeilBrown <neilb@suse.de>

When writing, we need to submit two writes, one to the original,
and one to the replacements - if there is a replacement.

If the write to the replacement results in a write error we just
fail the device.  We only try to record write errors to the
original.

This only handles writing new data.  Writing for resync/recovery
will come later.
Signed-off-by: NNeilBrown <neilb@suse.de>

Enhance raid10_remove_disk to be able to remove ->replacement
as well as ->rdev
Signed-off-by: NNeilBrown <neilb@suse.de>

When reading (for array reads, not for recovery etc) we read from the
replacement device if it has recovered far enough.
This requires storing the chosen rdev in the 'r10_bio' so we can make
sure to drop the ref on the right device when the read finishes.
Signed-off-by: NNeilBrown <neilb@suse.de>

It makes more sense to return an rdev than just an index as
read_balance() gets a reference to the rdev and so returning
the pointer make this more idiomatic.

This will be needed in a future patch when we might return
a 'replacement' rdev instead of the main rdev.
Signed-off-by: NNeilBrown <neilb@suse.de>

Allow each slot in the RAID10 to have 2 devices, the want_replacement
and the replacement.

Also an r10bio to have 2 bios, and for resync/recovery allocate the
second bio if there are any replacement devices.
Signed-off-by: NNeilBrown <neilb@suse.de>

Now that WantReplacement drives are replaced cleanly, mark a drive
as WantReplacement when we see a write error.  It might get failed soon so
the WantReplacement flag is irrelevant, but if the write error is recorded
in the bad block log, we still want to activate any spare that might
be available.
Reviewed-by: NDan Williams <dan.j.williams@intel.com>
Signed-off-by: NNeilBrown <neilb@suse.de>

When attempting to add a spare to a RAID[456] array, also consider
adding it as a replacement for a want_replacement device.

This requires that common md code attempt hot_add even when the array
is not formally degraded.
Reviewed-by: NDan Williams <dan.j.williams@intel.com>
Signed-off-by: NNeilBrown <neilb@suse.de>

If a Replacement is seen, file it as such.

If we see two replacements (or two normal devices) for the one slot,
abort.
Reviewed-by: NDan Williams <dan.j.williams@intel.com>
Signed-off-by: NNeilBrown <neilb@suse.de>

When recovery completes - as reported by a call to ->spare_active,
we clear In_sync on the original and set it on the replacement.

Then when the original gets removed we move the replacement from
'replacement' to 'rdev'.

This could race with other code that is looking at these pointers,
so we use memory barriers and careful ordering to ensure that
a reader might see one device twice, but never no devices.
Then the readers guard against using both devices, which could
only happen when writing.
Signed-off-by: NNeilBrown <neilb@suse.de>

During recovery we want to write to the replacement but not
the original.  So we have two new flags
 - R5_NeedReplace if this stripe has a replacement that needs to
   be written at some stage
 - R5_WantReplace if NeedReplace, and the data is available, and
   a 'sync' has been requested on this stripe.

We also distinguish between 'sync and replace' which need to read
all other devices, and 'replace' which only needs to read the
devices being replaced.

Note that during resync we always write to any replacement device.
It might not need to be written to, but as we don't read to compare,
we have to write to be sure.
Signed-off-by: NNeilBrown <neilb@suse.de>

When writing, we need to submit two writes, one to the original, and
one to the replacement - if there is a replacement.

If the write to the replacement results in a write error, we just fail
the device.  We only try to record write errors to the original.

When writing for recovery, we shouldn't write to the original.  This
will be addressed in a subsequent patch that generally addresses
recovery.
Reviewed-by: NDan Williams <dan.j.williams@intel.com>
Signed-off-by: NNeilBrown <neilb@suse.de>

Enhance raid5_remove_disk to be able to remove ->replacement
as well as ->rdev.
Reviewed-by: NDan Williams <dan.j.williams@intel.com>
Signed-off-by: NNeilBrown <neilb@suse.de>

If a replacement device is present and has been recovered far enough,
then use it for reading into the stripe cache.

If we get an error we don't try to repair it, we just fail the device.
A replacement device that gives errors does not sound sensible.

This requires removing the setting of R5_ReadError when we get
a read error during a read that bypasses the cache.  It was probably
a bad idea anyway as we don't know that every block in the read
caused an error, and it could cause ReadError to be set for the
replacement device, which is bad.
Signed-off-by: NNeilBrown <neilb@suse.de>

We current initialise some fields of a bio when preparing a
stripe_head, and again just before submitting the request.

Remove the duplication by only setting the fields that lower level
devices don't touch in raid5_build_block, and only set the changeable
fields in ops_run_io.
Reviewed-by: NDan Williams <dan.j.williams@intel.com>
Signed-off-by: NNeilBrown <neilb@suse.de>

Remove some #defines that are no longer used, and replace some
others with an enum.
And remove an unused field.
Reviewed-by: NDan Williams <dan.j.williams@intel.com>
Signed-off-by: NNeilBrown <neilb@suse.de>

Just enhance data structures to record a second device per slot to be
used as a 'replacement' device, replacing the original.
We also have a second bio in each slot in each stripe_head.  This will
only be used when writing to the array - we need to write to both the
original and the replacement at the same time, so will need two bios.

For now, only try using the replacement drive for aligned-reads.
In this case, we prefer the replacement if it has been recovered far
enough, otherwise use the original.

This includes a small enhancement.  Previously we would only do
aligned reads if the target device was fully recovered.  Now we also
do them if it has recovered far enough.
Reviewed-by: NDan Williams <dan.j.williams@intel.com>
Signed-off-by: NNeilBrown <neilb@suse.de>

hot-replace is a feature being added to md which will allow a
device to be replaced without removing it from the array first.

With hot-replace a spare can be activated and recovery can start while
the original device is still in place, thus allowing a transition from
an unreliable device to a reliable device without leaving the array
degraded during the transition.  It can also be use when the original
device is still reliable but it not wanted for some reason.

This will eventually be supported in RAID4/5/6 and RAID10.

This patch adds a super-block flag to distinguish the replacement
device.  If an old kernel sees this flag it will reject the device.

It also adds two per-device flags which are viewable and settable via
sysfs.
   "want_replacement" can be set to request that a device be replaced.
   "replacement" is set to show that this device is replacing another
   device.

The "rd%d" links in /sys/block/mdXx/md only apply to the original
device, not the replacement.  We currently don't make links for the
replacement - there doesn't seem to be a need.
Signed-off-by: NNeilBrown <neilb@suse.de>

Soon an array will be able to have multiple devices with the
same raid_disk number (an original and a replacement).  So removing
a device based on the number won't work.  So pass the actual device
handle instead.
Reviewed-by: NDan Williams <dan.j.williams@intel.com>
Signed-off-by: NNeilBrown <neilb@suse.de>

When setting the slot number on a device in an active array we
currently check that the number is not already in use.
We then call into the personality's hot_add_disk function
which performs the same test and returns the same error.

Thus the common test is not needed.

As we will shortly be changing some personalities to allow duplicates
in some cases (to support hot-replace), the common test will become
inconvenient.

So remove the common test.
Reviewed-by: NDan Williams <dan.j.williams@intel.com>
Signed-off-by: NNeilBrown <neilb@suse.de>

For each active region corresponding to a bit in the bitmap with have
a 14bit counter (and some flags).
This counts
   number of active writes + bit in the on-disk bitmap + delay-needed.

The "delay-needed" is because we always want a delay before clearing a
bit.  So the number here is normally number of active writes plus 2.
If there have been no writes for a while, we drop to 1.
If still no writes we clear the bit and drop to 0.

So for consistency, when setting bit from the on-disk bitmap or by
request from user-space it is best to set the counter to '2' to start
with.

In particular we might also set the NEEDED_MASK flag at this time, and
in all other cases NEEDED_MASK is only set when the counter is 2 or
more.
Signed-off-by: NNeilBrown <neilb@suse.de>

While using etags to find free_pages(), I stumbled across this debug
definition of free_pages() that is to be used while debugging some raid
code in userspace. The __get_free_pages() allocates the correct size,
but the free_pages() does not match. free_pages(), like
__get_free_pages(), takes an order and not a size.
Acked-by: NH. Peter Anvin <hpa@zytor.com>
Signed-off-by: NSteven Rostedt <rostedt@goodmis.org>
Signed-off-by: NNeilBrown <neilb@suse.de>

When an array is being reshaped to change the number of devices,
the two halves can be differently degraded.  e.g. one could be
missing a device and the other not.

So we need to be more careful about calculating the 'degraded'
attribute.

Instead of just inc/dec at appropriate times, perform a full
re-calculation examining both possible cases.  This doesn't happen
often so it not a big cost, and we already have most of the code to
do it.
Signed-off-by: NNeilBrown <neilb@suse.de>

We have a variable 'mddev' in this function, but repeatedly get the
same value by dereferencing bitmap->mddev.
There is room for simplification here...
Signed-off-by: NNeilBrown <neilb@suse.de>

The info is already available in /proc/mdstat and /sys/block in
an accessible form so there is no point in putting a road-block in
the ioctl for information gathering.
Signed-off-by: NNeilBrown <neilb@suse.de>

commit d0a4bb49 introduced a
regression which is annoying but fairly harmless.

When writing to an array that is undergoing recovery (a spare
in being integrated into the array), writing to the array will
set bits in the bitmap, but they will not be cleared when the
write completes.

For bits covering areas that have not been recovered yet this is not a
problem as the recovery will clear the bits.  However bits set in
already-recovered region will stay set and never be cleared.
This doesn't risk data integrity.  The only negatives are:
 - next time there is a crash, more resyncing than necessary will
   be done.
 - the bitmap doesn't look clean, which is confusing.

While an array is recovering we don't want to update the
'events_cleared' setting in the bitmap but we do still want to clear
bits that have very recently been set - providing they were written to
the recovering device.

So split those two needs - which previously both depended on 'success'
and always clear the bit of the write went to all devices.
Signed-off-by: NNeilBrown <neilb@suse.de>

Before performing a recovery we try to remove any spares that
might not be working, then add any that might have become relevant.

Currently we abort on the first spare that cannot be added.
This is a false optimisation.
It is conceivable that - depending on rules in the personality - a
subsequent spare might be accepted.
Also the loop does other things like count the available spares and
reset the 'recovery_offset' value.

If we abort early these might not happen properly.

So remove the early abort.

In particular if you have an array what is undergoing recovery and
which has extra spares, then the recovery may not restart after as
reboot as the could of 'spares' might end up as zero.
Reported-by: NAnssi Hannula <anssi.hannula@iki.fi>
Signed-off-by: NNeilBrown <neilb@suse.de>

While reshaping a degraded array (as when reshaping a RAID0 by first
converting it to a degraded RAID4) we currently get confused about
which devices are in_sync.  In most cases we get it right, but in the
region that is being reshaped we need to treat non-failed devices as
in-sync when we have the data but haven't actually written it out yet.
Reported-by: NAdam Kwolek <adam.kwolek@intel.com>
Signed-off-by: NNeilBrown <neilb@suse.de>

commit d70ed2e4
broke hot-add to a linear array.
After that commit, metadata if not written to devices until they
have been fully integrated into the array as determined by
saved_raid_disk.  That patch arranged to clear that field after
a recovery completed.

However for linear arrays, there is no recovery - the integration is
instantaneous.  So we need to explicitly clear the saved_raid_disk
field.
Signed-off-by: NNeilBrown <neilb@suse.de>

NULL pointer access causes crash in raid5 module.
Signed-off-by: NAdam Kwolek <adam.kwolek@intel.com>
Signed-off-by: NNeilBrown <neilb@suse.de>

Once a device is failed we really want to completely ignore it.
It should go away soon anyway.

In particular the presence of bad blocks on it should not cause us to
block as we won't be trying to write there anyway.

So as soon as we can check if a device is Faulty, do so and pretend
that it is already gone if it is Faulty.
Signed-off-by: NNeilBrown <neilb@suse.de>

When we mark blocks as bad we need them to be acknowledged by the
metadata handler promptly.

For an in-kernel metadata handler that was already being done.  But
for an external metadata handler we need to alert it of the change by
sending a notification through the sysfs file.  This adds that
notification.
Signed-off-by: NNeilBrown <neilb@suse.de>

Once a device is marked Faulty the badblocks - whether acknowledged or
not - become irrelevant.  So they shouldn't cause the device to be
marked as Blocked.

Without this patch, a process might write "-blocked" to clear the
Blocked status, but while that will correctly fail the device, it
won't remove the apparent 'blocked' status.
Signed-off-by: NNeilBrown <neilb@suse.de>

When we are accessing an mddev via sysfs we know that the
mddev cannot disappear because it has an embedded kobj which
is refcounted by sysfs.
And we also take the mddev_lock.
However this is not enough.

The final mddev_put could have been called and the
mddev_delayed_delete is waiting for sysfs to let go so it can destroy
the kobj and mddev.
In this state there are a lot of changes that should not be attempted.

To to guard against this we:
 - initialise mddev->all_mddevs in on last put so the state can be
   easily detected.
 - in md_attr_show and md_attr_store, check ->all_mddevs under
   all_mddevs_lock and mddev_get the mddev if it still appears to
   be active.

This means that if we get to sysfs as the mddev is being deleted we
will get -EBUSY.

rdev_attr_store and rdev_attr_show are similar but already have
sufficient protection.  They check that rdev->mddev still points to
mddev after taking mddev_lock.  As this is cleared  before delayed
removal which can only be requested under the mddev_lock, this
ensure the rdev and mddev are still alive.
Signed-off-by: NNeilBrown <neilb@suse.de>

We like md devices to disappear when they really are not needed.
However it is not possible to tell from the current state whether it
is needed or not.  We can only tell from recent history of changes.

In particular immediately after we create an md device it looks very
similar to immediately after we have finished with it.

So we always preserve a newly created md device until something
significant happens.  This state is stored in 'hold_active'.

The normal case is to keep it until an ioctl happens, as that will
normally either activate it, or explicitly de-activate it.  If it
doesn't then it was probably created by mistake and it is now time to
get rid of it.

We can also modify an array via sysfs (instead of via ioctl) and we
currently treat any change via sysfs like an ioctl as a sign that if
it now isn't more active, it should be destroyed.
However this is not appropriate as changes made via sysfs are more
gradual so we should look for a more definitive change.

So this patch only clears 'hold_active' from UNTIL_IOCTL to clear when
the array_state is changed via sysfs.  Other changes via sysfs
are ignored.
Signed-off-by: NNeilBrown <neilb@suse.de>

Page attributes are set using __set_bit rather than set_bit as
it normally called under a spinlock so the extra atomicity is not
needed.

However there are two places where we might set or clear page
attributes without holding the spinlock.
So add the spinlock in those cases.

This might be the cause of occasional reports that bits a aren't
getting clear properly - theory is that BITMAP_PAGE_PENDING gets lost
when BITMAP_PAGE_NEEDWRITE is set or cleared.  This is an
inconvenience, not a threat to data safety.
Signed-off-by: NNeilBrown <neilb@suse.de>

All updates that occur under STRIPE_ACTIVE should be globally visible
when STRIPE_ACTIVE clears.  test_and_set_bit() implies a barrier, but
clear_bit() does not.

This is suitable for 3.1-stable.
Signed-off-by: NDan Williams <dan.j.williams@intel.com>
Signed-off-by: NNeilBrown <neilb@suse.de>
Cc: stable@kernel.org

When the number of failed devices exceeds the allowed number
we must abort any active parity operations (checks or updates) as they
are no longer meaningful, and can lead to a BUG_ON in
handle_parity_checks6.

This bug was introduce by commit 6c0069c0
in 2.6.29.
Reported-by: NManish Katiyar <mkatiyar@gmail.com>
Tested-by: NManish Katiyar <mkatiyar@gmail.com>
Acked-by: NDan Williams <dan.j.williams@intel.com>
Signed-off-by: NNeilBrown <neilb@suse.de>
Cc: stable@kernel.org

.. with new name.  Because nothing says "really solid kernel release"
like naming it after an extinct animal that just happened to be in the
news lately.