When pblk receives a sync, all data up to that point in the write buffer
must be comitted to persistent storage, and as flash memory comes with a
minimal write size there is a significant cost involved both in terms
of time for completing the sync and in terms of write amplification
padded sectors for filling up to the minimal write size.

In order to get a better understanding of the costs involved for syncs,
Add a sysfs attribute to pblk: padded_dist, showing a normalized
distribution of sectors padded. In order to facilitate measurements of
specific workloads during the lifetime of the pblk instance, the
distribution can be reset by writing 0 to the attribute.

Do this by introducing counters for each possible padding:
{0..(minimal write size - 1)} and calculate the normalized distribution
when showing the attribute.
Signed-off-by: NHans Holmberg <hans.holmberg@cnexlabs.com>
Signed-off-by: NJavier González <javier@cnexlabs.com>
Rearranged total_buckets statement in pblk_sysfs_get_padding_dist
Signed-off-by: NMatias Bjørling <mb@lightnvm.io>
Signed-off-by: NJens Axboe <axboe@kernel.dk>

In a SSD, write amplification, WA, is defined as the average
number of page writes per user page write. Write amplification
negatively affects write performance and decreases the lifetime
of the disk, so it's a useful metric to add to sysfs.

In plkb's case, the number of writes per user sector is the sum of:

    (1) number of user writes
    (2) number of sectors written by the garbage collector
    (3) number of sectors padded (i.e. due to syncs)

This patch adds persistent counters for 1-3 and two sysfs attributes
to export these along with WA calculated with five decimals:

    write_amp_mileage: the accumulated write amplification stats
                      for the lifetime of the pblk instance

    write_amp_trip: resetable stats to facilitate delta measurements,
                    values reset at creation and if 0 is written
                    to the attribute.

64-bit counters are used as a 32 bit counter would wrap around
already after about 17 TB worth of user data. It will take a
long long time before the 64 bit sector counters wrap around.

The counters are stored after the bad block bitmap in the first
emeta sector of each written line. There is plenty of space in the
first emeta sector, so we don't need to bump the major version of
the line data format.
Signed-off-by: NHans Holmberg <hans.holmberg@cnexlabs.com>
Signed-off-by: NJavier González <javier@cnexlabs.com>
Signed-off-by: NMatias Bjørling <mb@lightnvm.io>
Signed-off-by: NJens Axboe <axboe@kernel.dk>

This patch has been generated as follows:

for verb in set_unlocked clear_unlocked set clear; do
  replace-in-files queue_flag_${verb} blk_queue_flag_${verb%_unlocked} \
    $(git grep -lw queue_flag_${verb} drivers block/bsg*)
done

Except for protecting all queue flag changes with the queue lock
this patch does not change any functionality.

Cc: Mike Snitzer <snitzer@redhat.com>
Cc: Shaohua Li <shli@fb.com>
Cc: Christoph Hellwig <hch@lst.de>
Cc: Hannes Reinecke <hare@suse.de>
Cc: Ming Lei <ming.lei@redhat.com>
Signed-off-by: NBart Van Assche <bart.vanassche@wdc.com>
Reviewed-by: NMartin K. Petersen <martin.petersen@oracle.com>
Reviewed-by: NJohannes Thumshirn <jthumshirn@suse.de>
Acked-by: NMartin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: NJens Axboe <axboe@kernel.dk>

Add the instance name to the information printed out on target creation.
Signed-off-by: NJavier González <javier@cnexlabs.com>
Signed-off-by: NMatias Bjørling <m@bjorling.me>
Signed-off-by: NJens Axboe <axboe@kernel.dk>

Refactor the way we free the write buffer to ensure that all entries get
freed in case of an error on the init sequence.
Signed-off-by: NJavier González <javier@cnexlabs.com>
Signed-off-by: NMatias Bjørling <m@bjorling.me>
Signed-off-by: NJens Axboe <axboe@kernel.dk>

When creating the write thread, ensure that the kthread has been created
before initializing the timer responsible from kicking it. Otherwise, if
the kthread creation fails or gets killed from used space, we risk
kicking an empty thread structure.

Also, since the kthread creation can be interrupted form user space,
adapt the error path to not report an error when this happens, since it
is intentional that the instance creation is aborted.
Signed-off-by: NJavier González <javier@cnexlabs.com>
Updated source to reflect the new timer_setup API.
Signed-off-by: NMatias Bjørling <m@bjorling.me>
Signed-off-by: NJens Axboe <axboe@kernel.dk>

Allow to set the over-provision percentage on target creation. In case
that the value is not provided, fall back to the default value set by
the target.

In pblk, set the default OP to 11% of the total size of the device
Signed-off-by: NJavier González <javier@cnexlabs.com>
Signed-off-by: NHans Holmberg <hans.holmberg@cnexlabs.com>
Signed-off-by: NMatias Bjørling <m@bjorling.me>
Signed-off-by: NJens Axboe <axboe@kernel.dk>

Until now, pblk's rate-limiter has used a heuristic to reserve space for
GC I/O given that the over-provision area was fixed.

In preparation for allowing to define the over-provision area on target
creation, define a dedicated free_block counter in the rate-limiter to
track the number of blocks being used for user data.
Signed-off-by: NJavier González <javier@cnexlabs.com>
Signed-off-by: NHans Holmberg <hans.holmberg@cnexlabs.com>
Signed-off-by: NMatias Bjørling <m@bjorling.me>
Signed-off-by: NJens Axboe <axboe@kernel.dk>

Prepare for the 2.0 revision by adapting the geometry
structures to coexist with the 1.2 revision.
Signed-off-by: NMatias Bjørling <m@bjorling.me>
Reviewed-by: NJavier González <javier@cnexlabs.com>
Signed-off-by: NMatias Bjørling <m@bjorling.me>
Signed-off-by: NJens Axboe <axboe@kernel.dk>

In preparation for unconditionally passing the struct timer_list pointer to
all timer callbacks, switch to using the new timer_setup() and from_timer()
to pass the timer pointer explicitly.

Cc: Matias Bjorling <mb@lightnvm.io>
Cc: linux-block@vger.kernel.org
Signed-off-by: NKees Cook <keescook@chromium.org>

Lockdep complains about being in atomic context while freeing line
metadata - and rightly so as we take a spinlock and end up calling
vfree that might sleep(in pblk_mfree).

There is no need for holding the line manager free_lock while
freeing line metadata as the pipeline as stopped, so remove the lock.

Fixes: 588726d3 ("lightnvm: pblk: fail gracefully on irrec. error")
Signed-off-by: NHans Holmberg <hans.holmberg@cnexlabs.com>
Signed-off-by: NMatias Bjørling <m@bjorling.me>
Signed-off-by: NJens Axboe <axboe@kernel.dk>

Print the CRC of the logical-to-physical mapping during exit and
after recovering the L2P table to facilitate detection of meta
data corruption/recovery issues.

The CRC printed after recovery should match the CRC printed during
the previous exit - if it doesn't this indicates that either the meta
data written to the disk is corrupt or recovery failed.
Signed-off-by: NHans Holmberg <hans.holmberg@cnexlabs.com>
Signed-off-by: NMatias Bjørling <m@bjorling.me>
Signed-off-by: NJens Axboe <axboe@kernel.dk>

Start GC if needed, directly after init, as we might
need to garbage collect in order to make room for user writes.

Create a helper function that allows to kick GC without exposing the
internals of the GC/rate-limiter interaction.
Signed-off-by: NHans Holmberg <hans.holmberg@cnexlabs.com>
Signed-off-by: NMatias Bjørling <m@bjorling.me>
Signed-off-by: NJens Axboe <axboe@kernel.dk>

When rebuilding the L2P table, any full lines (lines without any
valid sectors) will be identified. If these lines are not freed,
we risk not being able to allocate the first data line.

This patch refactors the part of GC that frees empty lines
into a separate function and adds a call to this after the
L2P table has been rebuilt.
Signed-off-by: NHans Holmberg <hans.holmberg@cnexlabs.com>
Signed-off-by: NMatias Bjørling <m@bjorling.me>
Signed-off-by: NJens Axboe <axboe@kernel.dk>

GC can be kicked after it has been shut down when closing the last
line during exit, resulting in accesses to freed structures.

Make sure that GC is not triggered while it is not operational.
Also make sure that GC won't be re-activated during exit when
running on another processor by using timer_del_sync.
Signed-off-by: NHans Holmberg <hans.holmberg@cnexlabs.com>
Signed-off-by: NMatias Bjørling <m@bjorling.me>
Signed-off-by: NJens Axboe <axboe@kernel.dk>

If pblk_core_init fails lets destroy all global caches.
Signed-off-by: NRakesh Pandit <rakesh@tuxera.com>
Signed-off-by: NMatias Bjørling <m@bjorling.me>
Signed-off-by: NJens Axboe <axboe@kernel.dk>

pblk_line_gc_list seems to had a bug since the introduction of pblk in
getting GC list for a line. In b20ba1bc while redesigning the GC
algorithm, the naming for the GC thresholds was altered, but the
values for high_thrs and mid_thrs were not. The result is that when
moving to the GC lists, the mid threshold is never evaluated.

Fixes: a4bd217b("lightnvm: physical block device (pblk) target")
Signed-off-by: NRakesh Pandit <rakesh@tuxera.com>
Signed-off-by: NMatias Bjørling <m@bjorling.me>
Signed-off-by: NJens Axboe <axboe@kernel.dk>

Make sure that the variable controlling block threshold for allocating
extra metadata sectors in case of a line with bad blocks does not get a
negative value. Otherwise, the line will be marked as corrupted and
wasted.
Signed-off-by: NJavier González <javier@cnexlabs.com>
Signed-off-by: NMatias Bjørling <m@bjorling.me>
Signed-off-by: NJens Axboe <axboe@kernel.dk>

Metadata I/Os are scheduled to minimize their impact on user data I/Os.
When there are enough LUNs instantiated (i.e., enough bandwidth), it is
easy to interleave metadata and data one after the other so that
metadata I/Os are the ones being blocked and not vice-versa.

We do this by calculating the distance between the I/Os in terms of the
LUNs that are not in used, and selecting a free LUN that satisfies a
the simple heuristic that metadata is scheduled behind. The per-LUN
semaphores guarantee consistency. This works fine on >1 LUN
configuration. However, when a single LUN is instantiated, this design
leads to a deadlock, where metadata waits to be scheduled on a free LUN.

This patch implements the 1 LUN case by simply scheduling the metadada
I/O after the data I/O. In the process, we refactor the way a line is
replaced to ensure that metadata writes are submitted after data writes
in order to guarantee block sequentiality. Note that, since there is
only one LUN, both I/Os will block each other by design. However, such
configuration only pursues tight read latencies, not write bandwidth.
Signed-off-by: NJavier González <javier@cnexlabs.com>
Signed-off-by: NMatias Bjørling <m@bjorling.me>
Signed-off-by: NJens Axboe <axboe@kernel.dk>

When a line is recycled during garbage collection, reads can still be
issued to the line. If the line is freed in the middle of this process,
data corruption might occur.

This patch guarantees that lines are not freed in the middle of reads
that target them (lines). Specifically, we use the existing line
reference to decide when a line is eligible for being freed after the
recycle process.
Signed-off-by: NJavier González <javier@cnexlabs.com>
Signed-off-by: NMatias Bjørling <m@bjorling.me>
Signed-off-by: NJens Axboe <axboe@kernel.dk>

pblk holds two sector bitmaps: one to keep track of the mapped sectors
while the line is active and another one to keep track of the invalid
sectors. The latter is kept during the whole live of the line, until it
is recycled. Since we cannot guarantee forward progress for the mempool
in this case, get rid of the mempool and simply allocate memory through
kmalloc.
Reported-by: NJens Axboe <axboe@kernel.dk>
Signed-off-by: NJavier González <javier@cnexlabs.com>
Signed-off-by: NMatias Bjørling <m@bjorling.me>
Signed-off-by: NJens Axboe <axboe@kernel.dk>

Since read and erase paths offer different guarantees for inflight I/Os,
separate the mempools to set the right min_nr for each on creation.
Reported-by: NJens Axboe <axboe@kernel.dk>
Signed-off-by: NJavier González <javier@cnexlabs.com>
Signed-off-by: NMatias Bjørling <m@bjorling.me>
Signed-off-by: NJens Axboe <axboe@kernel.dk>

In pblk, we have a mempool to allocate a generic structure that we
pass along workqueues. This is heavily used in the GC path in order
to have enough inflight reads and fully utilize the GC bandwidth.

However, the current GC path copies data to the host memory and puts it
back into the write buffer. This requires a vmalloc allocation for the
data and a memory copy. Thus, guaranteeing the allocation by using a
mempool for the structure in itself does not give us much. Until we
implement support for vector copy to avoid moving data through the host,
just allocate the workqueue structure using kmalloc.

This allows us to have a much smaller mempool.
Reported-by: NJens Axboe <axboe@kernel.dk>
Signed-off-by: NJavier González <javier@cnexlabs.com>
Signed-off-by: NMatias Bjørling <m@bjorling.me>
Signed-off-by: NJens Axboe <axboe@kernel.dk>

pblk uses an internal page mempool for allocating pages on internal
bios. The main two users of this memory pool are partial reads (reads
with some sectors in cache and some on media) and padded writes, which
need to add dummy pages to an existing bio already containing valid
data (and with a large enough bioset allocated). In both cases, the
maximum number of pages per bio is defined by the maximum number of
physical sectors supported by the underlying device.

This patch fixes a bad mempool allocation, where the min_nr of elements
on the pool was fixed (to 16), which is lower than the maximum number
of sectors supported by NVMe (as of the time for this patch). Instead,
use the maximum number of allowed sectors reported by the device.
Reported-by: NJens Axboe <axboe@kernel.dk>
Signed-off-by: NJavier González <javier@cnexlabs.com>
Signed-off-by: NMatias Bjørling <m@bjorling.me>
Signed-off-by: NJens Axboe <axboe@kernel.dk>

On low LUN configurations, make sure not to send bios that are bigger
than the buffer size.

Fixes: a4bd217b ("lightnvm: physical block device (pblk) target")
Signed-off-by: NJavier González <javier@cnexlabs.com>
Signed-off-by: NMatias Bjørling <m@bjorling.me>
Signed-off-by: NJens Axboe <axboe@kernel.dk>

Initialize the stat counter for garbage collected reads.

Fixes: a4bd217b ("lightnvm: physical block device (pblk) target")
Signed-off-by: NJavier González <javier@cnexlabs.com>
Signed-off-by: NMatias Bjørling <m@bjorling.me>
Signed-off-by: NJens Axboe <axboe@kernel.dk>

This usually happens if we are developing with qemu and ll2pmode has
default value. Improve description.
Signed-off-by: NRakesh Pandit <rakesh@tuxera.com>
Reviewed-by: NJavier González <javier@cnexlabs.com>
Signed-off-by: NMatias Bjørling <m@bjorling.me>
Signed-off-by: NJens Axboe <axboe@kernel.dk>

Use appropriate memory free calls based on allocation type used and
also fix number of times free is called if kmalloc fails.
Signed-off-by: NRakesh Pandit <rakesh@tuxera.com>
Reviewed-by: NJavier González <javier@cnexlabs.com>
Signed-off-by: NMatias Bjørling <m@bjorling.me>
Signed-off-by: NJens Axboe <axboe@kernel.dk>

If target type module e.g. pblk here is unloaded (rmmod) while module
is in use (after creating target) system crashes.  We fix this by
using module API refcnt.
Signed-off-by: NRakesh Pandit <rakesh@tuxera.com>
Signed-off-by: NMatias Bjørling <m@bjorling.me>
Signed-off-by: NJens Axboe <axboe@kernel.dk>

Add a sanity check to the pblk initialization sequence in order to
ensure that enough LUNs have been allocated to store the line metadata.
Signed-off-by: NJavier González <javier@cnexlabs.com>
Signed-off-by: NMatias Bjørling <matias@cnexlabs.com>
Signed-off-by: NJens Axboe <axboe@kernel.dk>

When removing a pblk instance, pad the current line using asynchronous
I/O. This reduces the removal time from ~1 minute in the worst case to a
couple of seconds.
Signed-off-by: NJavier González <javier@cnexlabs.com>
Signed-off-by: NMatias Bjørling <matias@cnexlabs.com>
Signed-off-by: NJens Axboe <axboe@kernel.dk>

Prevent pblk->lines being double freed in case of an error during pblk
initialization.

Fixes: dd2a4343: "lightnvm: pblk: sched. metadata on write thread"
Reported-by: NDan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: NJavier González <javier@cnexlabs.com>
Signed-off-by: NMatias Bjørling <matias@cnexlabs.com>
Signed-off-by: NJens Axboe <axboe@kernel.dk>

Due to user writes being decoupled from media writes because of the need
of an intermediate write buffer, irrecoverable media write errors lead
to pblk stalling; user writes fill up the buffer and end up in an
infinite retry loop.

In order to let user writes fail gracefully, it is necessary for pblk to
keep track of its own internal state and prevent further writes from
being placed into the write buffer.

This patch implements a state machine to keep track of internal errors
and, in case of failure, fail further user writes in an standard way.
Depending on the type of error, pblk will do its best to persist
buffered writes (which are already acknowledged) and close down on a
graceful manner. This way, data might be recovered by re-instantiating
pblk. Such state machine paves out the way for a state-based FTL log.
Signed-off-by: NJavier González <javier@cnexlabs.com>
Signed-off-by: NMatias Bjørling <matias@cnexlabs.com>
Signed-off-by: NJens Axboe <axboe@kernel.dk>

Make constants to define sizes for internal mempools and workqueues. In
this process, adjust the values to be more meaningful given the internal
constrains of the FTL. In order to do this for workqueues, separate the
current auxiliary workqueue into two dedicated workqueues to manage
lines being closed and bad blocks.
Signed-off-by: NJavier González <javier@cnexlabs.com>
Signed-off-by: NMatias Bjørling <matias@cnexlabs.com>
Signed-off-by: NJens Axboe <axboe@kernel.dk>

Decouple bad block discovery from line allocation logic. This allows to
return meaningful error codes in case of bad block discovery failure.
Signed-off-by: NJavier González <javier@cnexlabs.com>
Signed-off-by: NMatias Bjørling <matias@cnexlabs.com>
Signed-off-by: NJens Axboe <axboe@kernel.dk>

smeta size will always be suitable for a kmalloc allocation. Simplify
the code and leave the vmalloc fallback only for emeta, where the pblk
configuration has an impact.
Signed-off-by: NJavier González <javier@cnexlabs.com>
Signed-off-by: NMatias Bjørling <matias@cnexlabs.com>
Signed-off-by: NJens Axboe <axboe@kernel.dk>

After refactoring the metadata path, the backpointer controlling
synced I/Os in a line becomes unnecessary; metadata is scheduled
on the write thread, thus we know when the end of the line is reached
and act on it directly.
Signed-off-by: NJavier González <javier@cnexlabs.com>
Signed-off-by: NMatias Bjørling <matias@cnexlabs.com>
Signed-off-by: NJens Axboe <axboe@kernel.dk>

At the moment, line metadata is persisted on a separate work queue, that
is kicked each time that a line is closed. The assumption when designing
this was that freeing the write thread from creating a new write request
was better than the potential impact of writes colliding on the media
(user I/O and metadata I/O). Experimentation has proven that this
assumption is wrong; collision can cause up to 25% of bandwidth and
introduce long tail latencies on the write thread, which potentially
cause user write threads to spend more time spinning to get a free entry
on the write buffer.

This patch moves the metadata logic to the write thread. When a line is
closed, remaining metadata is written in memory and is placed on a
metadata queue. The write thread then takes the metadata corresponding
to the previous line, creates the write request and schedules it to
minimize collisions on the media. Using this approach, we see that we
can saturate the media's bandwidth, which helps reducing both write
latencies and the spinning time for user writer threads.
Signed-off-by: NJavier González <javier@cnexlabs.com>
Signed-off-by: NMatias Bjørling <matias@cnexlabs.com>
Signed-off-by: NJens Axboe <axboe@kernel.dk>

Read requests allocate some extra memory to store its per I/O context.
Instead of requiring yet another memory pool for other type of requests,
generalize this context allocation (and change naming accordingly).
Signed-off-by: NJavier González <javier@cnexlabs.com>
Signed-off-by: NMatias Bjørling <matias@cnexlabs.com>
Signed-off-by: NJens Axboe <axboe@kernel.dk>

Erase I/Os are scheduled with the following goals in mind: (i) minimize
LUNs collisions with write I/Os, and (ii) even out the price of erasing
on every write, instead of putting all the burden on when garbage
collection runs. This works well on the current design, but is specific
to the default mapping algorithm.

This patch generalizes the erase path so that other mapping algorithms
can select an arbitrary line to be erased instead. It also gets rid of
the erase semaphore since it creates jittering for user writes.
Signed-off-by: NJavier González <javier@cnexlabs.com>
Signed-off-by: NMatias Bjørling <matias@cnexlabs.com>
Signed-off-by: NJens Axboe <axboe@kernel.dk>