We apply a smoothing to the scale changes in order to keep sawtoothy
behavior from occurring. However our window for checking if we've
missed our target can sometimes be lower than the smoothing interval
(500ms), especially on faster drives like ssd's. In order to deal with
this keep track of the running tally of the previous intervals that we
threw away because we had already done a scale event recently.
This is needed for the ssd case as these low latency drives will have
bursts of latency, and if it happens to be ok for the window that
directly follows the opening of the scale window we could unthrottle
when previous windows we were missing our target.
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
We use an average latency approach for determining if we're missing our
latency target. This works well for rotational storage where we have
generally consistent latencies, but for ssd's and other low latency
devices you have more of a spikey behavior, which means we often won't
throttle misbehaving groups because a lot of IO completes at drastically
faster times than our latency target. Instead keep track of how many
IO's miss our target and how many IO's are done in our time window. If
the p(90) latency is above our target then we know we need to throttle.
With this change in place we are seeing the same throttling behavior
with our testcase on ssd's as we see with rotational drives.
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
There is logic to keep cgroups that haven't done a lot of IO in the most
recent scale window from being punished for over-active higher priority
groups. However for things like ssd's where the windows are pretty
short we'll end up with small numbers of samples, so 5% of samples will
come out to 0 if there aren't enough. Make the floor 1 sample to keep
us from improperly bailing out of scaling down.
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Hitting the case where blk_queue_depth() returned 1 uncovered the fact
that iolatency doesn't actually handle this case properly, it simply
doesn't scale down anybody. For this case we should go straight into
applying the time delay, which we weren't doing. Since we already limit
the floor at 1 request this if statement is not needed, and this allows
us to set our depth to 1 which allows us to apply the delay if needed.
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
We were using blk_queue_depth() assuming that it would return
nr_requests, but we hit a case in production on drives that had to have
NCQ turned off in order for them to not shit the bed which resulted in a
qd of 1, even though the nr_requests was much larger. iolatency really
only cares about requests we are allowed to queue up, as any io that
get's onto the request list is going to be serviced soonish, so we want
to be throttling before the bio gets onto the request list. To make
iolatency work as expected, simply use q->nr_requests instead of
blk_queue_depth() as that is what we actually care about.
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
NSEC_PER_SEC has type long, so 5 * NSEC_PER_SEC is calculated as a long.
However, 5 seconds is 5,000,000,000 nanoseconds, which overflows a
32-bit long. Make sure all of the targets are calculated as 64-bit
values.
Fixes: 6e25cb01ea ("kyber: implement improved heuristics")
Reported-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Omar Sandoval <osandov@fb.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Use new-style DEVICE_ATTR_RO/DEVICE_ATTR_RW to create the sysfs attributes
and register the disk with default sysfs attribute groups.
Signed-off-by: Hannes Reinecke <hare@suse.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Acked-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Bart Van Assche <bart.vanassche@wdc.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Register default sysfs groups during device_add_disk() to avoid a
race condition with udev during startup.
Signed-off-by: Hannes Reinecke <hare@suse.com>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Nitin Gupta <ngupta@vflare.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Bart Van Assche <bart.vanassche@wdc.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Register default sysfs groups during device_add_disk() to avoid a
race condition with udev during startup.
Signed-off-by: Hannes Reinecke <hare@suse.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Acked-by: Ed L. Cachin <ed.cashin@acm.org>
Reviewed-by: Bart Van Assche <bart.vanassche@wdc.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
We should be registering the ns_id attribute as default sysfs
attribute groups, otherwise we have a race condition between
the uevent and the attributes appearing in sysfs.
Suggested-by: Bart van Assche <bvanassche@acm.org>
Reviewed-by: Keith Busch <keith.busch@intel.com>
Signed-off-by: Hannes Reinecke <hare@suse.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Update device_add_disk() to take an 'groups' argument so that
individual drivers can register a device with additional sysfs
attributes.
This avoids race condition the driver would otherwise have if these
groups were to be created with sysfs_add_groups().
Signed-off-by: Martin Wilck <martin.wilck@suse.com>
Signed-off-by: Hannes Reinecke <hare@suse.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Bart Van Assche <bvanassche@acm.org>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
When debugging Kyber, it's really useful to know what latencies we've
been having, how the domain depths have been adjusted, and if we've
actually been throttling. Add three tracepoints, kyber_latency,
kyber_adjust, and kyber_throttled, to record that.
Signed-off-by: Omar Sandoval <osandov@fb.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Kyber's current heuristics have a few flaws:
- It's based on the mean latency, but p99 latency tends to be more
meaningful to anyone who cares about latency. The mean can also be
skewed by rare outliers that the scheduler can't do anything about.
- The statistics calculations are purely time-based with a short window.
This works for steady, high load, but is more sensitive to outliers
with bursty workloads.
- It only considers the latency once an I/O has been submitted to the
device, but the user cares about the time spent in the kernel, as
well.
These are shortcomings of the generic blk-stat code which doesn't quite
fit the ideal use case for Kyber. So, this replaces the statistics with
a histogram used to calculate percentiles of total latency and I/O
latency, which we then use to adjust depths in a slightly more
intelligent manner:
- Sync and async writes are now the same domain.
- Discards are a separate domain.
- Domain queue depths are scaled by the ratio of the p99 total latency
to the target latency (e.g., if the p99 latency is double the target
latency, we will double the queue depth; if the p99 latency is half of
the target latency, we can halve the queue depth).
- We use the I/O latency to determine whether we should scale queue
depths down: we will only scale down if any domain's I/O latency
exceeds the target latency, which is an indicator of congestion in the
device.
These new heuristics are just as scalable as the heuristics they
replace.
Signed-off-by: Omar Sandoval <osandov@fb.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
The domain token sbitmaps are currently initialized to the device queue
depth or 256, whichever is larger, and immediately resized to the
maximum depth for that domain (256, 128, or 64 for read, write, and
other, respectively). The sbitmap is never resized larger than that, so
it's unnecessary to allocate a bitmap larger than the maximum depth.
Let's just allocate it to the maximum depth to begin with. This will use
marginally less memory, and more importantly, give us a more appropriate
number of bits per sbitmap word.
Signed-off-by: Omar Sandoval <osandov@fb.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Kyber will need this in a future change if it is built as a module.
Signed-off-by: Omar Sandoval <osandov@fb.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Commit 4bc6339a58 ("block: move blk_stat_add() to
__blk_mq_end_request()") consolidated some calls using ktime_get() so
we'd only need to call it once. Kyber's ->completed_request() hook also
calls ktime_get(), so let's move it to the same place, too.
Signed-off-by: Omar Sandoval <osandov@fb.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Now that the blk-mq core processes power management requests
(marked with RQF_PREEMPT) in other states than RPM_ACTIVE, enable
runtime power management for blk-mq.
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Reviewed-by: Ming Lei <ming.lei@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Cc: Jianchao Wang <jianchao.w.wang@oracle.com>
Cc: Hannes Reinecke <hare@suse.com>
Cc: Johannes Thumshirn <jthumshirn@suse.de>
Cc: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Instead of allowing requests that are not power management requests
to enter the queue in runtime suspended status (RPM_SUSPENDED), make
the blk_get_request() caller block. This change fixes a starvation
issue: it is now guaranteed that power management requests will be
executed no matter how many blk_get_request() callers are waiting.
For blk-mq, instead of maintaining the q->nr_pending counter, rely
on q->q_usage_counter. Call pm_runtime_mark_last_busy() every time a
request finishes instead of only if the queue depth drops to zero.
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Reviewed-by: Ming Lei <ming.lei@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Cc: Jianchao Wang <jianchao.w.wang@oracle.com>
Cc: Hannes Reinecke <hare@suse.com>
Cc: Johannes Thumshirn <jthumshirn@suse.de>
Cc: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
A later patch will call blk_freeze_queue_start() followed by
blk_mq_unfreeze_queue() without waiting for q_usage_counter to drop
to zero. Make sure that this doesn't cause a kernel warning to appear
by switching from percpu_ref_reinit() to percpu_ref_resurrect(). The
former namely requires that the refcount it operates on is zero.
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Reviewed-by: Ming Lei <ming.lei@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Cc: Jianchao Wang <jianchao.w.wang@oracle.com>
Cc: Hannes Reinecke <hare@suse.com>
Cc: Johannes Thumshirn <jthumshirn@suse.de>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
This function will be used in a later patch to switch the struct
request_queue q_usage_counter from killed back to live. In contrast
to percpu_ref_reinit(), this new function does not require that the
refcount is zero.
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Acked-by: Tejun Heo <tj@kernel.org>
Reviewed-by: Ming Lei <ming.lei@redhat.com>
Cc: Christoph Hellwig <hch@lst.de>
Cc: Jianchao Wang <jianchao.w.wang@oracle.com>
Cc: Hannes Reinecke <hare@suse.com>
Cc: Johannes Thumshirn <jthumshirn@suse.de>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Instead of scheduling runtime resume of a request queue after a
request has been queued, schedule asynchronous resume during request
allocation. The new pm_request_resume() calls occur after
blk_queue_enter() has increased the q_usage_counter request queue
member. This change is needed for a later patch that will make request
allocation block while the queue status is not RPM_ACTIVE.
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Reviewed-by: Ming Lei <ming.lei@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Cc: Jianchao Wang <jianchao.w.wang@oracle.com>
Cc: Hannes Reinecke <hare@suse.com>
Cc: Johannes Thumshirn <jthumshirn@suse.de>
Cc: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Move the pm_request_resume() and pm_runtime_mark_last_busy() calls into
two new functions and thereby separate legacy block layer code from code
that works for both the legacy block layer and blk-mq. A later patch will
add calls to the new functions in the blk-mq code.
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Reviewed-by: Ming Lei <ming.lei@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Cc: Martin K. Petersen <martin.petersen@oracle.com>
Cc: Jianchao Wang <jianchao.w.wang@oracle.com>
Cc: Hannes Reinecke <hare@suse.com>
Cc: Johannes Thumshirn <jthumshirn@suse.de>
Cc: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
The RQF_PREEMPT flag is used for three purposes:
- In the SCSI core, for making sure that power management requests
are executed even if a device is in the "quiesced" state.
- For domain validation by SCSI drivers that use the parallel port.
- In the IDE driver, for IDE preempt requests.
Rename "preempt-only" into "pm-only" because the primary purpose of
this mode is power management. Since the power management core may
but does not have to resume a runtime suspended device before
performing system-wide suspend and since a later patch will set
"pm-only" mode as long as a block device is runtime suspended, make
it possible to set "pm-only" mode from more than one context. Since
with this change scsi_device_quiesce() is no longer idempotent, make
that function return early if it is called for a quiesced queue.
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Acked-by: Martin K. Petersen <martin.petersen@oracle.com>
Reviewed-by: Hannes Reinecke <hare@suse.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Ming Lei <ming.lei@redhat.com>
Cc: Jianchao Wang <jianchao.w.wang@oracle.com>
Cc: Johannes Thumshirn <jthumshirn@suse.de>
Cc: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Move the code for runtime power management from blk-core.c into the
new source file blk-pm.c. Move the corresponding declarations from
<linux/blkdev.h> into <linux/blk-pm.h>. For CONFIG_PM=n, leave out
the declarations of the functions that are not used in that mode.
This patch not only reduces the number of #ifdefs in the block layer
core code but also reduces the size of header file <linux/blkdev.h>
and hence should help to reduce the build time of the Linux kernel
if CONFIG_PM is not defined.
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Reviewed-by: Ming Lei <ming.lei@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Cc: Jianchao Wang <jianchao.w.wang@oracle.com>
Cc: Hannes Reinecke <hare@suse.com>
Cc: Johannes Thumshirn <jthumshirn@suse.de>
Cc: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Nothing Xen specific in these headers, which get included from a lot
of code in the kernel. So prune the includes and move them to the
Xen-specific files that actually use them instead.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Take the Xen check into the core code instead of delegating it to
the architectures.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Having multiple externs in arch headers is not a good way to provide
a common interface.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Now that we don't need an override for BIOVEC_PHYS_MERGEABLE there is
no need to drag this header in.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
We only use it in biovec_phys_mergeable and a m68k paravirt driver,
so just opencode it there. Also remove the pointless unsigned long cast
for the offset in the opencoded instances.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Geert Uytterhoeven <geert@linux-m68k.org>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
These two checks should always be performed together, so merge them into
a single helper.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
The actual recaculation of segments in __blk_recalc_rq_segments will
do this check, so there is no point in forcing it if we know it won't
succeed.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Turn the macro into an inline, move it to blk.h and simplify the
arch hooks a bit.
Also rename the function to biovec_phys_mergeable as there is no need
to shout.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Keep it close to the actual users instead of exposing the function to all
drivers.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Make it easier to understand the purpose of the functions that iterate
over requests by documenting their purpose. Fix several minor spelling
and grammer mistakes in comments in these functions.
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Cc: Christoph Hellwig <hch@lst.de>
Cc: Ming Lei <ming.lei@redhat.com>
Cc: Jianchao Wang <jianchao.w.wang@oracle.com>
Cc: Hannes Reinecke <hare@suse.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
blkg reference counting now uses percpu_ref rather than atomic_t. Let's
make this consistent with css_tryget. This renames blkg_try_get to
blkg_tryget and now returns a bool rather than the blkg or NULL.
Signed-off-by: Dennis Zhou <dennisszhou@gmail.com>
Reviewed-by: Josef Bacik <josef@toxicpanda.com>
Acked-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Now that every bio is associated with a blkg, this puts the use of
blkg_get, blkg_try_get, and blkg_put on the hot path. This switches over
the refcnt in blkg to use percpu_ref.
Signed-off-by: Dennis Zhou <dennisszhou@gmail.com>
Acked-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
blk_get_rl is responsible for identifying which request_list a request
should be allocated to. Try get logic was added earlier, but
semantically the logic was not changed.
This patch makes better use of the bio already having a reference to the
blkg in the hot path. The cold path uses a better fallback of
blkg_lookup_create rather than just blkg_lookup and then falling back to
the q->root_rl. If lookup_create fails with anything but -ENODEV, it
falls back to q->root_rl.
A clarifying comment is added to explain why q->root_rl is used rather
than the root blkg's rl.
Signed-off-by: Dennis Zhou <dennisszhou@gmail.com>
Acked-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
The previous patch in this series removed carrying around a pointer to
the css in blkg. However, the blkg association logic still relied on
taking a reference on the css to ensure we wouldn't fail in getting a
reference for the blkg.
Here the implicit dependency on the css is removed. The association
continues to rely on the tryget logic walking up the blkg tree. This
streamlines the three ways that association can happen: normal, swap,
and writeback.
Acked-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Dennis Zhou <dennisszhou@gmail.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Prior patches ensured that all bios are now associated with some blkg.
This now makes bio->bi_css unnecessary as blkg maintains a reference to
the blkcg already.
This patch removes the field bi_css and transfers corresponding uses to
access via bi_blkg.
Signed-off-by: Dennis Zhou <dennisszhou@gmail.com>
Reviewed-by: Josef Bacik <josef@toxicpanda.com>
Acked-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
One of the goals of this series is to remove a separate reference to
the css of the bio. This can and should be accessed via bio_blkcg. In
this patch, the wbc_init_bio call is changed such that it must be called
after a queue has been associated with the bio.
Signed-off-by: Dennis Zhou <dennisszhou@gmail.com>
Reviewed-by: Josef Bacik <josef@toxicpanda.com>
Acked-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
A prior patch in this series added blkg association to bios issued by
cgroups. There are two other paths that we want to attribute work back
to the appropriate cgroup: swap and writeback. Here we modify the way
swap tags bios to include the blkg. Writeback will be tackle in the next
patch.
Signed-off-by: Dennis Zhou <dennisszhou@gmail.com>
Reviewed-by: Josef Bacik <josef@toxicpanda.com>
Acked-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
bio_issue_init among other things initializes the timestamp for an IO.
Rather than have this logic handled by policies, this consolidates it to
be on the init paths (normal, clone, bounce clone).
Signed-off-by: Dennis Zhou <dennisszhou@gmail.com>
Acked-by: Tejun Heo <tj@kernel.org>
Reviewed-by: Liu Bo <bo.liu@linux.alibaba.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Previously, blkg's were only assigned as needed by blk-iolatency and
blk-throttle. bio->css was also always being associated while blkg was
being looked up and then thrown away in blkcg_bio_issue_check.
This patch begins the cleanup of bio->css and bio->bi_blkg by always
associating a blkg in blkcg_bio_issue_check. This tries to create the
blkg, but if it is not possible, falls back to using the root_blkg of
the request_queue. Therefore, a bio will always be associated with a
blkg. The duplicate association logic is removed from blk-throttle and
blk-iolatency.
Signed-off-by: Dennis Zhou <dennisszhou@gmail.com>
Acked-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
There are several scenarios where blkg_lookup_create can fail. Examples
include the blkcg dying, request_queue is dying, or simply being OOM. At
the end of the day, most handle this by simply falling back to the
q->root_blkg and calling it a day.
This patch implements the notion of closest blkg. During
blkg_lookup_create, if it fails to create, return the closest blkg
found or the q->root_blkg. blkg_try_get_closest is introduced and used
during association so a bio is always attached to a blkg.
Acked-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Dennis Zhou <dennisszhou@gmail.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
To know when to create a blkg, the general pattern is to do a
blkg_lookup and if that fails, lock and then do a lookup again and if
that fails finally create. It doesn't make much sense for everyone who
wants to do creation to write this themselves.
This changes blkg_lookup_create to do locking and implement this
pattern. The old blkg_lookup_create is renamed to __blkg_lookup_create.
If a call site wants to do its own error handling or already owns the
queue lock, they can use __blkg_lookup_create. This will be used in
upcoming patches.
Signed-off-by: Dennis Zhou <dennisszhou@gmail.com>
Reviewed-by: Josef Bacik <josef@toxicpanda.com>
Acked-by: Tejun Heo <tj@kernel.org>
Reviewed-by: Liu Bo <bo.liu@linux.alibaba.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Josef Bacik