Some engines optimize operation by reading ahead in the descriptor chain
such that descriptor2 may start execution before descriptor1 completes.
If descriptor2 depends on the result from descriptor1 then a fence is
required (on descriptor2) to disable this optimization.  The async_tx
api could implicitly identify dependencies via the 'depend_tx'
parameter, but that would constrain cases where the dependency chain
only specifies a completion order rather than a data dependency.  So,
provide an ASYNC_TX_FENCE to explicitly identify data dependencies.
Signed-off-by: NDan Williams <dan.j.williams@intel.com>

Now that the resources to handle stripe_head operations are allocated
percpu it is possible for raid5d to distribute stripe handling over
multiple cores.  This conversion also adds a call to cond_resched() in
the non-multicore case to prevent one core from getting monopolized for
raid operations.

Cc: Arjan van de Ven <arjan@linux.intel.com>
Signed-off-by: NDan Williams <dan.j.williams@intel.com>

These routines have been replaced by there asynchronous counterparts.
Signed-off-by: NYuri Tikhonov <yur@emcraft.com>
Signed-off-by: NIlya Yanok <yanok@emcraft.com>
Signed-off-by: NDan Williams <dan.j.williams@intel.com>

1/ Use STRIPE_OP_BIOFILL to offload completion of read requests to
   raid_run_ops
2/ Implement a handler for sh->reconstruct_state similar to the raid5 case
   (adds handling of Q parity)
3/ Prevent handle_parity_checks6 from running concurrently with 'compute'
   operations
4/ Hook up raid_run_ops
Signed-off-by: NYuri Tikhonov <yur@emcraft.com>
Signed-off-by: NIlya Yanok <yanok@emcraft.com>
Signed-off-by: NDan Williams <dan.j.williams@intel.com>

[ Based on an original patch by Yuri Tikhonov ]

Implement the state machine for handling the RAID-6 parities check and
repair functionality.  Note that the raid6 case does not need to check
for new failures, like raid5, as it will always writeback the correct
disks.  The raid5 case can be updated to check zero_sum_result to avoid
getting confused by new failures rather than retrying the entire check
operation.
Signed-off-by: NYuri Tikhonov <yur@emcraft.com>
Signed-off-by: NIlya Yanok <yanok@emcraft.com>
Signed-off-by: NDan Williams <dan.j.williams@intel.com>

In the synchronous implementation of stripe dirtying we processed a
degraded stripe with one call to handle_stripe_dirtying6().  I.e.
compute the missing blocks from the other drives, then copy in the new
data and reconstruct the parities.

In the asynchronous case we do not perform stripe operations directly.
Instead, operations are scheduled with flags to be later serviced by
raid_run_ops.  So, for the degraded case the final reconstruction step
can only be carried out after all blocks have been brought up to date by
being read, or computed.  Like the raid5 case schedule_reconstruction()
sets STRIPE_OP_RECONSTRUCT to request a parity generation pass and
through operation chaining can handle compute and reconstruct in a
single raid_run_ops pass.

[dan.j.williams@intel.com: fixup handle_stripe_dirtying6 gating]
Signed-off-by: NYuri Tikhonov <yur@emcraft.com>
Signed-off-by: NIlya Yanok <yanok@emcraft.com>
Signed-off-by: NDan Williams <dan.j.williams@intel.com>

Modify handle_stripe_fill6 to work asynchronously by introducing
fetch_block6 as the raid6 analog of fetch_block5 (schedule compute
operations for missing/out-of-sync disks).

[dan.j.williams@intel.com: compute D+Q in one pass]
Signed-off-by: NYuri Tikhonov <yur@emcraft.com>
Signed-off-by: NIlya Yanok <yanok@emcraft.com>
Signed-off-by: NDan Williams <dan.j.williams@intel.com>

Extend schedule_reconstruction5 for reuse by the raid6 path.  Add
support for generating Q and BUG() if a request is made to perform
'prexor'.
Signed-off-by: NYuri Tikhonov <yur@emcraft.com>
Signed-off-by: NIlya Yanok <yanok@emcraft.com>
Signed-off-by: NDan Williams <dan.j.williams@intel.com>

[ Based on an original patch by Yuri Tikhonov ]

The raid_run_ops routine uses the asynchronous offload api and
the stripe_operations member of a stripe_head to carry out xor+pq+copy
operations asynchronously, outside the lock.

The operations performed by RAID-6 are the same as in the RAID-5 case
except for no support of STRIPE_OP_PREXOR operations. All the others
are supported:
STRIPE_OP_BIOFILL
 - copy data into request buffers to satisfy a read request
STRIPE_OP_COMPUTE_BLK
 - generate missing blocks (1 or 2) in the cache from the other blocks
STRIPE_OP_BIODRAIN
 - copy data out of request buffers to satisfy a write request
STRIPE_OP_RECONSTRUCT
 - recalculate parity for new data that has entered the cache
STRIPE_OP_CHECK
 - verify that the parity is correct

The flow is the same as in the RAID-5 case, and reuses some routines, namely:
1/ ops_complete_postxor (renamed to ops_complete_reconstruct)
2/ ops_complete_compute (updated to set up to 2 targets uptodate)
3/ ops_run_check (renamed to ops_run_check_p for xor parity checks)

[neilb@suse.de: fixes to get it to pass mdadm regression suite]
Reviewed-by: NAndre Noll <maan@systemlinux.org>
Signed-off-by: NYuri Tikhonov <yur@emcraft.com>
Signed-off-by: NIlya Yanok <yanok@emcraft.com>
Signed-off-by: NDan Williams <dan.j.williams@intel.com>

ops_complete_compute5 can be reused in the raid6 path if it is updated to
generically handle a second target.
Signed-off-by: NDan Williams <dan.j.williams@intel.com>

Replace the flat zero_sum_result with a collection of flags to contain
the P (xor) zero-sum result, and the soon to be utilized Q (raid6 reed
solomon syndrome) zero-sum result.  Use the SUM_CHECK_ namespace instead
of DMA_ since these flags will be used on non-dma-zero-sum enabled
platforms.
Reviewed-by: NAndre Noll <maan@systemlinux.org>
Acked-by: NMaciej Sosnowski <maciej.sosnowski@intel.com>
Signed-off-by: NDan Williams <dan.j.williams@intel.com>

Use percpu memory rather than stack for storing the buffer lists used in
parity calculations.  Include space for dma address conversions and pass
that to async_tx via the async_submit_ctl.scribble pointer.

[ Impact: move memory pressure from stack to heap ]
Signed-off-by: NDan Williams <dan.j.williams@intel.com>

In preparation for asynchronous handling of raid6 operations move the
spare page to a percpu allocation to allow multiple simultaneous
synchronous raid6 recovery operations.

Make this allocation cpu hotplug aware to maximize allocation
efficiency.
Signed-off-by: NDan Williams <dan.j.williams@intel.com>

Add missing call to safe_put_page from stop() by unifying open coded
raid5_conf_t de-allocation under free_conf().
Signed-off-by: NDan Williams <dan.j.williams@intel.com>

Now that we support changing the chunksize, we calculate
"reshape_sectors" to be the max of number of sectors in old
and new chunk size.
However there is one please where we still use 'chunksize'
rather than 'reshape_sectors'.
This causes a reshape that reduces the size of chunks to freeze.
Signed-off-by: NNeilBrown <neilb@suse.de>

md has functionality to 'quiesce' and array so that all pending
IO completed and no new IO starts.  This is used to achieve a
stable state before making internal changes.

Currently this quiescing applies equally to normal IO, resync
IO, and reshape IO.
However there is a problem with applying it to reshape IO.
Reshape can have multiple 'stripe_heads' that must be active together.
If the quiesce come between allocating the first and the last of
such a collection, then we deadlock, as the last will not be allocated
until the quiesce is lifted, the quiesce will not be lifted until the
first (which has been allocated) gets used, and that first cannot be
used until the last is allocated.

It is not necessary to inhibit reshape IO when a quiesce is
requested.  Those places in the code that require a full quiesce will
ensure the reshape thread is not running at all.

So allow reshape requests to get access to new stripe_heads without
being blocked by a 'quiesce'.

This only affects in-place reshapes (i.e. where the array does not
grow or shrink) and these are only newly supported.  So this patch is
not needed in earlier kernels.
Signed-off-by: NNeilBrown <neilb@suse.de>

mddev->raid_disks can be changed and any time by a request from
user-space.  It is a suggestion as to what number of raid_disks is
desired.

conf->raid_disks can only be changed by the raid5 module with suitable
locks in place.  It is a statement as to the current number of
raid_disks.

There are two places where the latter should be used, but the former
is used.  This can lead to a crash when reshaping an array.

This patch changes to mddev-> to conf->
Signed-off-by: NNeilBrown <neilb@suse.de>

Prepare the api for the arrival of a new parameter, 'scribble'.  This
will allow callers to identify scratchpad memory for dma address or page
address conversions.  As this adds yet another parameter, take this
opportunity to convert the common submission parameters (flags,
dependency, callback, and callback argument) into an object that is
passed by reference.

Also, take this opportunity to fix up the kerneldoc and add notes about
the relevant ASYNC_TX_* flags for each routine.

[ Impact: moves api pass-by-value parameters to a pass-by-reference struct ]
Signed-off-by: NAndre Noll <maan@systemlinux.org>
Acked-by: NMaciej Sosnowski <maciej.sosnowski@intel.com>
Signed-off-by: NDan Williams <dan.j.williams@intel.com>

In support of inter-channel chaining async_tx utilizes an ack flag to
gate whether a dependent operation can be chained to another.  While the
flag is not set the chain can be considered open for appending.  Setting
the ack flag closes the chain and flags the descriptor for garbage
collection.  The ASYNC_TX_DEP_ACK flag essentially means "close the
chain after adding this dependency".  Since each operation can only have
one child the api now implicitly sets the ack flag at dependency
submission time.  This removes an unnecessary management burden from
clients of the api.

[ Impact: clean up and enforce one dependency per operation ]
Reviewed-by: NAndre Noll <maan@systemlinux.org>
Acked-by: NMaciej Sosnowski <maciej.sosnowski@intel.com>
Signed-off-by: NDan Williams <dan.j.williams@intel.com>

A recent patch to raid5.c use min on an int and a sector_t.
This isn't allowed.
So change it to min_t(sector_t,x,y).
Signed-off-by: NNeilBrown <neilb@suse.de>

As sector_t in unsigned, we cannot afford to let 'safepos' etc go
negative.
So replace
   a -= b;
by
   a -= min(b,a);
Signed-off-by: NNeilBrown <neilb@suse.de>

There are circumstances when a user-space process might need to
"oversee" a resync/reshape process.  For example when doing an
in-place reshape of a raid5, it is prudent to take a backup of each
section before reshaping it as this is the only way to provide
safety against an unplanned shutdown (i.e. crash/power failure).

The sync_max sysfs value can be used to stop the resync from
advancing beyond a particular point.
So user-space can:
  suspend IO to the first section and back it up
  set 'sync_max' to the end of the section
  wait for 'sync_completed' to reach that point
  resume IO on the first section and move on to the next section.

However this process requires the kernel and user-space to run in
lock-step which could introduce unnecessary delays.

It would be better if a 'double buffered' approach could be used with
userspace and kernel space working on different sections with the
'next' section always ready when the 'current' section is finished.

One problem with implementing this is that sync_completed is only
guaranteed to be updated when the sync process reaches sync_max.
(it is updated on a time basis at other times, but it is hard to rely
on that).  This defeats some of the double buffering.

With this patch, sync_completed (and reshape_position) get updated as
the current position approaches sync_max, so there is room for
userspace to advance sync_max early without losing updates.

To be precise, sync_completed is updated when the current sync
position reaches half way between the current value of sync_completed
and the value of sync_max.  This will usually be a good time for user
space to update sync_max.

If sync_max does not get updated, the updates to sync_completed
(together with associated metadata updates) will occur at an
exponentially increasing frequency which will get unreasonably fast
(one update every page) immediately before the process hits sync_max
and stops.  So the update rate will be unreasonably fast only for an
insignificant period of time.
Signed-off-by: NNeilBrown <neilb@suse.de>

The sync_completed file reports how much of a resync (or recovery or
reshape) has been completed.
However due to the possibility of out-of-order completion of writes,
it is not certain to be accurate.

We have an internal value - mddev->curr_resync_completed - which is an
accurate value (though it might not always be quite so uptodate).

So:
 - make curr_resync_completed be uptodate a little more often,
   particularly when raid5 reshape updates status in the metadata
 - report curr_resync_completed in the sysfs file
 - allow poll/select to report all updates to md/sync_completed.

This makes sync_completed completed usable by any external metadata
handler that wants to record this status information in its metadata.
Signed-off-by: NNeilBrown <neilb@suse.de>

'zero_sum' does not properly describe the operation of generating parity
and checking that it validates against an existing buffer.  Change the
name of the operation to 'val' (for 'validate').  This is in
anticipation of the p+q case where it is a requirement to identify the
target parity buffers separately from the source buffers, because the
target parity buffers will not have corresponding pq coefficients.
Reviewed-by: NAndre Noll <maan@systemlinux.org>
Acked-by: NMaciej Sosnowski <maciej.sosnowski@intel.com>
Signed-off-by: NDan Williams <dan.j.williams@intel.com>

We currently update the metadata :
 1/ every 3Megabytes
 2/ When the place we will write new-layout data to is recorded in
    the metadata as still containing old-layout data.

Rule one exists to avoid having to re-do too much reshaping in the
face of a crash/restart.  So it should really be time based rather
than size based.  So change it to "every 10 seconds".

Rule two turns out to be too harsh when restriping an array
'in-place', as in that case the metadata much be updates for every
stripe.
For the in-place update, it can only possibly be safe from a crash if
some user-space program data a backup of every e.g. few hundred
stripes before allowing them to be reshaped.  In that case, the
constant metadata update is pointless.
So only update the metadata if the new metadata will report that the
end of the 'old-layout' data is beyond where we are currently
writing 'new-layout' data.
Signed-off-by: NNeilBrown <neilb@suse.de>

... and to be certain the that make_request doesn't wait forever,
add a 'wake_up' when ->reshape_progress has been set to MaxSector
Signed-off-by: NNeilBrown <neilb@suse.de>

This was only needed when the code was experimental.  Most of it
is well tested now, so the option is no longer useful.
Signed-off-by: NNeilBrown <neilb@suse.de>

When we are reshaping an array, it is very important that we read
the data from a particular sector offset before writing new data
at that offset.

In most cases when growing or shrinking an array we read long before
we even consider writing.  But when restriping an array without
changing it size, there is a small possibility that we might have
some data to available write before the read has happened at the same
location.  This would require some stripes to be in cache already.

To guard against this small possibility, we check, before writing,
that the 'old' stripe at the same location is not in the process of
being read.  And we ensure that we mark all 'source' stripes as such
before allowing new 'destination' stripes to proceed.
Signed-off-by: NNeilBrown <neilb@suse.de>

If an array has 3 or more devices, we allow the chunksize or layout
to be changed and when a reshape starts, we use these as the 'new'
values.
Signed-off-by: NNeilBrown <neilb@suse.de>

This ensures that even when old and new stripes are overlapping,
we will try to read all of the old before having to write any
of the new.
Signed-off-by: NNeilBrown <neilb@suse.de>

Add prev_algo to raid5_conf_t along the same lines as prev_chunk
and previous_raid_disks.
Signed-off-by: NNeilBrown <neilb@suse.de>

Add "prev_chunk" to raid5_conf_t, similar to "previous_raid_disks", to
remember what the chunk size was before the reshape that is currently
underway.

This seems like duplication with "chunk_size" and "new_chunk" in
mddev_t, and to some extent it is, but there are differences.
The values in mddev_t are always defined and often the same.
The prev* values are only defined if a reshape is underway.

Also (and more significantly) the raid5_conf_t values will be changed
at the same time (inside an appropriate lock) that the reshape is
started by setting reshape_position.  In contrast, the new_chunk value
is set when the sysfs file is written which could be well before the
reshape starts.
Signed-off-by: NNeilBrown <neilb@suse.de>

During a raid5 reshape, we have some stripes in the cache that are
'before' the reshape (and are still to be processed) and some that are
'after'.  They are currently differentiated by having different
->disks values as the only reshape current supported involves changing
the number of disks.

However we will soon support reshapes that do not change the number
of disks (chunk parity or chunk size).  So make the difference more
explicit with a 'generation' number.
Signed-off-by: NNeilBrown <neilb@suse.de>

When reshaping a raid5 to have fewer devices, we work from the end of
the array to the beginning.
md_do_sync gives addresses to sync_request that go from the beginning
to the end.  So largely ignore them use the internal state variable
"reshape_progress" to keep track of what to do next.

Never allow the size to be reduced below the minimum (4 for raid6,
3 otherwise).

We require that the size of the array has already been reduced before
the array is reshaped to a smaller size.  This is because simply
reducing the size is an easily reversible operation, while the reshape
is immediately destructive and so is not reversible for the blocks at
the ends of the devices.
Thus to reshape an array to have fewer devices, you must first write
an appropriately small size to md/array_size.

When reshape finished, we remove any drives that are no longer
needed and fix up ->degraded.
Signed-off-by: NNeilBrown <neilb@suse.de>

When reducing the number of devices in a raid4/5/6, the reshape
process has to start at the end of the array and work down to the
beginning.  So we need to handle expand_progress and expand_lo
differently.

This patch renames "expand_progress" and "expand_lo" to avoid the
implication that anything is getting bigger (expand->reshape) and
every place they are used, we make sure that they are used the right
way depending on whether delta_disks is positive or negative.
Signed-off-by: NNeilBrown <neilb@suse.de>

Currently raid5 (the only module that supports restriping)
notices that the reshape has finished be sync_request being
given a large value, and handles any cleanup them.

This patch changes it so md_check_recovery calls into an
explicit finish_reshape method as well.

The clean-up from sync_request can do things that need to be
done promptly, typically things local to the raid5_conf_t
structure.

The "finish_reshape" method is called under the mddev_lock
so it can do things involving reconfiguring the device.

This allows us to get rid of md_set_array_sectors_locked, which
would have caused a deadlock if you tried to stop and array
while a reshape was happening.
Signed-off-by: NNeilBrown <neilb@suse.de>

This is the first of four patches which combine to allow md/raid5 to
reduce the number of devices in the array by restriping the data over
a subset of the devices.

If the number of disks in a raid4/5/6 is being reduced, then the
default size must be based on the new number, not the old number
of devices.
In general, it should be based on the smaller of new and old.
Signed-off-by: NNeilBrown <neilb@suse.de>

We now have this value in stripe_head so we don't need to duplicate
it.
Signed-off-by: NNeilBrown <neilb@suse.de>

Move the raid6 data processing routines into a standalone module
(raid6_pq) to prepare them to be called from async_tx wrappers and other
non-md drivers/modules.  This precludes a circular dependency of raid456
needing the async modules for data processing while those modules in
turn depend on raid456 for the base level synchronous raid6 routines.

To support this move:
1/ The exportable definitions in raid6.h move to include/linux/raid/pq.h
2/ The raid6_call, recovery calls, and table symbols are exported
3/ Extra #ifdef __KERNEL__ statements to enable the userspace raid6test to
   compile
Signed-off-by: NDan Williams <dan.j.williams@intel.com>
Signed-off-by: NNeilBrown <neilb@suse.de>

raid4 allows only one failed disk.
Signed-off-by: NAndre Noll <maan@systemlinux.org>
Signed-off-by: NNeilBrown <neilb@suse.de>