linux/block/blk-iocost.c

3413 lines
95 KiB
C
Raw Normal View History

blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
/* SPDX-License-Identifier: GPL-2.0
*
* IO cost model based controller.
*
* Copyright (C) 2019 Tejun Heo <tj@kernel.org>
* Copyright (C) 2019 Andy Newell <newella@fb.com>
* Copyright (C) 2019 Facebook
*
* One challenge of controlling IO resources is the lack of trivially
* observable cost metric. This is distinguished from CPU and memory where
* wallclock time and the number of bytes can serve as accurate enough
* approximations.
*
* Bandwidth and iops are the most commonly used metrics for IO devices but
* depending on the type and specifics of the device, different IO patterns
* easily lead to multiple orders of magnitude variations rendering them
* useless for the purpose of IO capacity distribution. While on-device
* time, with a lot of clutches, could serve as a useful approximation for
* non-queued rotational devices, this is no longer viable with modern
* devices, even the rotational ones.
*
* While there is no cost metric we can trivially observe, it isn't a
* complete mystery. For example, on a rotational device, seek cost
* dominates while a contiguous transfer contributes a smaller amount
* proportional to the size. If we can characterize at least the relative
* costs of these different types of IOs, it should be possible to
* implement a reasonable work-conserving proportional IO resource
* distribution.
*
* 1. IO Cost Model
*
* IO cost model estimates the cost of an IO given its basic parameters and
* history (e.g. the end sector of the last IO). The cost is measured in
* device time. If a given IO is estimated to cost 10ms, the device should
* be able to process ~100 of those IOs in a second.
*
* Currently, there's only one builtin cost model - linear. Each IO is
* classified as sequential or random and given a base cost accordingly.
* On top of that, a size cost proportional to the length of the IO is
* added. While simple, this model captures the operational
* characteristics of a wide varienty of devices well enough. Default
* paramters for several different classes of devices are provided and the
* parameters can be configured from userspace via
* /sys/fs/cgroup/io.cost.model.
*
* If needed, tools/cgroup/iocost_coef_gen.py can be used to generate
* device-specific coefficients.
*
* 2. Control Strategy
*
* The device virtual time (vtime) is used as the primary control metric.
* The control strategy is composed of the following three parts.
*
* 2-1. Vtime Distribution
*
* When a cgroup becomes active in terms of IOs, its hierarchical share is
* calculated. Please consider the following hierarchy where the numbers
* inside parentheses denote the configured weights.
*
* root
* / \
* A (w:100) B (w:300)
* / \
* A0 (w:100) A1 (w:100)
*
* If B is idle and only A0 and A1 are actively issuing IOs, as the two are
* of equal weight, each gets 50% share. If then B starts issuing IOs, B
* gets 300/(100+300) or 75% share, and A0 and A1 equally splits the rest,
* 12.5% each. The distribution mechanism only cares about these flattened
* shares. They're called hweights (hierarchical weights) and always add
* upto 1 (WEIGHT_ONE).
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
*
* A given cgroup's vtime runs slower in inverse proportion to its hweight.
* For example, with 12.5% weight, A0's time runs 8 times slower (100/12.5)
* against the device vtime - an IO which takes 10ms on the underlying
* device is considered to take 80ms on A0.
*
* This constitutes the basis of IO capacity distribution. Each cgroup's
* vtime is running at a rate determined by its hweight. A cgroup tracks
* the vtime consumed by past IOs and can issue a new IO iff doing so
* wouldn't outrun the current device vtime. Otherwise, the IO is
* suspended until the vtime has progressed enough to cover it.
*
* 2-2. Vrate Adjustment
*
* It's unrealistic to expect the cost model to be perfect. There are too
* many devices and even on the same device the overall performance
* fluctuates depending on numerous factors such as IO mixture and device
* internal garbage collection. The controller needs to adapt dynamically.
*
* This is achieved by adjusting the overall IO rate according to how busy
* the device is. If the device becomes overloaded, we're sending down too
* many IOs and should generally slow down. If there are waiting issuers
* but the device isn't saturated, we're issuing too few and should
* generally speed up.
*
* To slow down, we lower the vrate - the rate at which the device vtime
* passes compared to the wall clock. For example, if the vtime is running
* at the vrate of 75%, all cgroups added up would only be able to issue
* 750ms worth of IOs per second, and vice-versa for speeding up.
*
* Device business is determined using two criteria - rq wait and
* completion latencies.
*
* When a device gets saturated, the on-device and then the request queues
* fill up and a bio which is ready to be issued has to wait for a request
* to become available. When this delay becomes noticeable, it's a clear
* indication that the device is saturated and we lower the vrate. This
* saturation signal is fairly conservative as it only triggers when both
* hardware and software queues are filled up, and is used as the default
* busy signal.
*
* As devices can have deep queues and be unfair in how the queued commands
* are executed, soley depending on rq wait may not result in satisfactory
* control quality. For a better control quality, completion latency QoS
* parameters can be configured so that the device is considered saturated
* if N'th percentile completion latency rises above the set point.
*
* The completion latency requirements are a function of both the
* underlying device characteristics and the desired IO latency quality of
* service. There is an inherent trade-off - the tighter the latency QoS,
* the higher the bandwidth lossage. Latency QoS is disabled by default
* and can be set through /sys/fs/cgroup/io.cost.qos.
*
* 2-3. Work Conservation
*
* Imagine two cgroups A and B with equal weights. A is issuing a small IO
* periodically while B is sending out enough parallel IOs to saturate the
* device on its own. Let's say A's usage amounts to 100ms worth of IO
* cost per second, i.e., 10% of the device capacity. The naive
* distribution of half and half would lead to 60% utilization of the
* device, a significant reduction in the total amount of work done
* compared to free-for-all competition. This is too high a cost to pay
* for IO control.
*
* To conserve the total amount of work done, we keep track of how much
* each active cgroup is actually using and yield part of its weight if
* there are other cgroups which can make use of it. In the above case,
* A's weight will be lowered so that it hovers above the actual usage and
* B would be able to use the rest.
*
* As we don't want to penalize a cgroup for donating its weight, the
* surplus weight adjustment factors in a margin and has an immediate
* snapback mechanism in case the cgroup needs more IO vtime for itself.
*
* Note that adjusting down surplus weights has the same effects as
* accelerating vtime for other cgroups and work conservation can also be
* implemented by adjusting vrate dynamically. However, squaring who can
* donate and should take back how much requires hweight propagations
* anyway making it easier to implement and understand as a separate
* mechanism.
*
* 3. Monitoring
*
* Instead of debugfs or other clumsy monitoring mechanisms, this
* controller uses a drgn based monitoring script -
* tools/cgroup/iocost_monitor.py. For details on drgn, please see
* https://github.com/osandov/drgn. The ouput looks like the following.
*
* sdb RUN per=300ms cur_per=234.218:v203.695 busy= +1 vrate= 62.12%
* active weight hweight% inflt% dbt delay usages%
* test/a * 50/ 50 33.33/ 33.33 27.65 2 0*041 033:033:033
* test/b * 100/ 100 66.67/ 66.67 17.56 0 0*000 066:079:077
*
* - per : Timer period
* - cur_per : Internal wall and device vtime clock
* - vrate : Device virtual time rate against wall clock
* - weight : Surplus-adjusted and configured weights
* - hweight : Surplus-adjusted and configured hierarchical weights
* - inflt : The percentage of in-flight IO cost at the end of last period
* - del_ms : Deferred issuer delay induction level and duration
* - usages : Usage history
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
*/
#include <linux/kernel.h>
#include <linux/module.h>
#include <linux/timer.h>
#include <linux/time64.h>
#include <linux/parser.h>
#include <linux/sched/signal.h>
#include <linux/blk-cgroup.h>
#include <asm/local.h>
#include <asm/local64.h>
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
#include "blk-rq-qos.h"
#include "blk-stat.h"
#include "blk-wbt.h"
#ifdef CONFIG_TRACEPOINTS
/* copied from TRACE_CGROUP_PATH, see cgroup-internal.h */
#define TRACE_IOCG_PATH_LEN 1024
static DEFINE_SPINLOCK(trace_iocg_path_lock);
static char trace_iocg_path[TRACE_IOCG_PATH_LEN];
#define TRACE_IOCG_PATH(type, iocg, ...) \
do { \
unsigned long flags; \
if (trace_iocost_##type##_enabled()) { \
spin_lock_irqsave(&trace_iocg_path_lock, flags); \
cgroup_path(iocg_to_blkg(iocg)->blkcg->css.cgroup, \
trace_iocg_path, TRACE_IOCG_PATH_LEN); \
trace_iocost_##type(iocg, trace_iocg_path, \
##__VA_ARGS__); \
spin_unlock_irqrestore(&trace_iocg_path_lock, flags); \
} \
} while (0)
#else /* CONFIG_TRACE_POINTS */
#define TRACE_IOCG_PATH(type, iocg, ...) do { } while (0)
#endif /* CONFIG_TRACE_POINTS */
enum {
MILLION = 1000000,
/* timer period is calculated from latency requirements, bound it */
MIN_PERIOD = USEC_PER_MSEC,
MAX_PERIOD = USEC_PER_SEC,
/*
blk-iocost: revamp donation amount determination iocost has various safety nets to combat inuse adjustment calculation inaccuracies. With Andy's method implemented in transfer_surpluses(), inuse adjustment calculations are now accurate and we can make donation amount determinations accurate too. * Stop keeping track of past usage history and using the maximum. Act on the immediate usage information. * Remove donation constraints defined by SURPLUS_* constants. Donate whatever isn't used. * Determine the donation amount so that the iocg will end up with MARGIN_TARGET_PCT budget at the end of the coming period assuming the same usage as the previous period. TARGET is set at 50% of period, which is the previous maximum. This provides smooth convergence for most repetitive IO patterns. * Apply donation logic early at 20% budget. There's no risk in doing so as the calculation is based on the delta between the current budget and the target budget at the end of the coming period. * Remove preemptive iocg activation for zero cost IOs. As donation can reach near zero now, the mere activation doesn't provide any protection anymore. In the unlikely case that this becomes a problem, the right solution is assigning appropriate costs for such IOs. This significantly improves the donation determination logic while also simplifying it. Now all donations are immediate, exact and smooth. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2020-09-01 18:52:49 +00:00
* iocg->vtime is targeted at 50% behind the device vtime, which
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
* serves as its IO credit buffer. Surplus weight adjustment is
* immediately canceled if the vtime margin runs below 10%.
*/
MARGIN_MIN_PCT = 10,
blk-iocost: revamp donation amount determination iocost has various safety nets to combat inuse adjustment calculation inaccuracies. With Andy's method implemented in transfer_surpluses(), inuse adjustment calculations are now accurate and we can make donation amount determinations accurate too. * Stop keeping track of past usage history and using the maximum. Act on the immediate usage information. * Remove donation constraints defined by SURPLUS_* constants. Donate whatever isn't used. * Determine the donation amount so that the iocg will end up with MARGIN_TARGET_PCT budget at the end of the coming period assuming the same usage as the previous period. TARGET is set at 50% of period, which is the previous maximum. This provides smooth convergence for most repetitive IO patterns. * Apply donation logic early at 20% budget. There's no risk in doing so as the calculation is based on the delta between the current budget and the target budget at the end of the coming period. * Remove preemptive iocg activation for zero cost IOs. As donation can reach near zero now, the mere activation doesn't provide any protection anymore. In the unlikely case that this becomes a problem, the right solution is assigning appropriate costs for such IOs. This significantly improves the donation determination logic while also simplifying it. Now all donations are immediate, exact and smooth. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2020-09-01 18:52:49 +00:00
MARGIN_LOW_PCT = 20,
MARGIN_TARGET_PCT = 50,
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
INUSE_ADJ_STEP_PCT = 25,
/* Have some play in timer operations */
TIMER_SLACK_PCT = 1,
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
/* 1/64k is granular enough and can easily be handled w/ u32 */
WEIGHT_ONE = 1 << 16,
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
/*
* As vtime is used to calculate the cost of each IO, it needs to
* be fairly high precision. For example, it should be able to
* represent the cost of a single page worth of discard with
* suffificient accuracy. At the same time, it should be able to
* represent reasonably long enough durations to be useful and
* convenient during operation.
*
* 1s worth of vtime is 2^37. This gives us both sub-nanosecond
* granularity and days of wrap-around time even at extreme vrates.
*/
VTIME_PER_SEC_SHIFT = 37,
VTIME_PER_SEC = 1LLU << VTIME_PER_SEC_SHIFT,
VTIME_PER_USEC = VTIME_PER_SEC / USEC_PER_SEC,
VTIME_PER_NSEC = VTIME_PER_SEC / NSEC_PER_SEC,
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
/* bound vrate adjustments within two orders of magnitude */
VRATE_MIN_PPM = 10000, /* 1% */
VRATE_MAX_PPM = 100000000, /* 10000% */
VRATE_MIN = VTIME_PER_USEC * VRATE_MIN_PPM / MILLION,
VRATE_CLAMP_ADJ_PCT = 4,
/* if IOs end up waiting for requests, issue less */
RQ_WAIT_BUSY_PCT = 5,
/* unbusy hysterisis */
UNBUSY_THR_PCT = 75,
/*
* The effect of delay is indirect and non-linear and a huge amount of
* future debt can accumulate abruptly while unthrottled. Linearly scale
* up delay as debt is going up and then let it decay exponentially.
* This gives us quick ramp ups while delay is accumulating and long
* tails which can help reducing the frequency of debt explosions on
* unthrottle. The parameters are experimentally determined.
*
* The delay mechanism provides adequate protection and behavior in many
* cases. However, this is far from ideal and falls shorts on both
* fronts. The debtors are often throttled too harshly costing a
* significant level of fairness and possibly total work while the
* protection against their impacts on the system can be choppy and
* unreliable.
*
* The shortcoming primarily stems from the fact that, unlike for page
* cache, the kernel doesn't have well-defined back-pressure propagation
* mechanism and policies for anonymous memory. Fully addressing this
* issue will likely require substantial improvements in the area.
*/
MIN_DELAY_THR_PCT = 500,
MAX_DELAY_THR_PCT = 25000,
MIN_DELAY = 250,
MAX_DELAY = 250 * USEC_PER_MSEC,
/* halve debts if avg usage over 100ms is under 50% */
DFGV_USAGE_PCT = 50,
DFGV_PERIOD = 100 * USEC_PER_MSEC,
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
/* don't let cmds which take a very long time pin lagging for too long */
MAX_LAGGING_PERIODS = 10,
/* switch iff the conditions are met for longer than this */
AUTOP_CYCLE_NSEC = 10LLU * NSEC_PER_SEC,
/*
* Count IO size in 4k pages. The 12bit shift helps keeping
* size-proportional components of cost calculation in closer
* numbers of digits to per-IO cost components.
*/
IOC_PAGE_SHIFT = 12,
IOC_PAGE_SIZE = 1 << IOC_PAGE_SHIFT,
IOC_SECT_TO_PAGE_SHIFT = IOC_PAGE_SHIFT - SECTOR_SHIFT,
/* if apart further than 16M, consider randio for linear model */
LCOEF_RANDIO_PAGES = 4096,
};
enum ioc_running {
IOC_IDLE,
IOC_RUNNING,
IOC_STOP,
};
/* io.cost.qos controls including per-dev enable of the whole controller */
enum {
QOS_ENABLE,
QOS_CTRL,
NR_QOS_CTRL_PARAMS,
};
/* io.cost.qos params */
enum {
QOS_RPPM,
QOS_RLAT,
QOS_WPPM,
QOS_WLAT,
QOS_MIN,
QOS_MAX,
NR_QOS_PARAMS,
};
/* io.cost.model controls */
enum {
COST_CTRL,
COST_MODEL,
NR_COST_CTRL_PARAMS,
};
/* builtin linear cost model coefficients */
enum {
I_LCOEF_RBPS,
I_LCOEF_RSEQIOPS,
I_LCOEF_RRANDIOPS,
I_LCOEF_WBPS,
I_LCOEF_WSEQIOPS,
I_LCOEF_WRANDIOPS,
NR_I_LCOEFS,
};
enum {
LCOEF_RPAGE,
LCOEF_RSEQIO,
LCOEF_RRANDIO,
LCOEF_WPAGE,
LCOEF_WSEQIO,
LCOEF_WRANDIO,
NR_LCOEFS,
};
enum {
AUTOP_INVALID,
AUTOP_HDD,
AUTOP_SSD_QD1,
AUTOP_SSD_DFL,
AUTOP_SSD_FAST,
};
struct ioc_gq;
struct ioc_params {
u32 qos[NR_QOS_PARAMS];
u64 i_lcoefs[NR_I_LCOEFS];
u64 lcoefs[NR_LCOEFS];
u32 too_fast_vrate_pct;
u32 too_slow_vrate_pct;
};
struct ioc_margins {
s64 min;
blk-iocost: revamp donation amount determination iocost has various safety nets to combat inuse adjustment calculation inaccuracies. With Andy's method implemented in transfer_surpluses(), inuse adjustment calculations are now accurate and we can make donation amount determinations accurate too. * Stop keeping track of past usage history and using the maximum. Act on the immediate usage information. * Remove donation constraints defined by SURPLUS_* constants. Donate whatever isn't used. * Determine the donation amount so that the iocg will end up with MARGIN_TARGET_PCT budget at the end of the coming period assuming the same usage as the previous period. TARGET is set at 50% of period, which is the previous maximum. This provides smooth convergence for most repetitive IO patterns. * Apply donation logic early at 20% budget. There's no risk in doing so as the calculation is based on the delta between the current budget and the target budget at the end of the coming period. * Remove preemptive iocg activation for zero cost IOs. As donation can reach near zero now, the mere activation doesn't provide any protection anymore. In the unlikely case that this becomes a problem, the right solution is assigning appropriate costs for such IOs. This significantly improves the donation determination logic while also simplifying it. Now all donations are immediate, exact and smooth. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2020-09-01 18:52:49 +00:00
s64 low;
s64 target;
};
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
struct ioc_missed {
local_t nr_met;
local_t nr_missed;
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
u32 last_met;
u32 last_missed;
};
struct ioc_pcpu_stat {
struct ioc_missed missed[2];
local64_t rq_wait_ns;
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
u64 last_rq_wait_ns;
};
/* per device */
struct ioc {
struct rq_qos rqos;
bool enabled;
struct ioc_params params;
struct ioc_margins margins;
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
u32 period_us;
u32 timer_slack_ns;
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
u64 vrate_min;
u64 vrate_max;
spinlock_t lock;
struct timer_list timer;
struct list_head active_iocgs; /* active cgroups */
struct ioc_pcpu_stat __percpu *pcpu_stat;
enum ioc_running running;
atomic64_t vtime_rate;
u64 vtime_base_rate;
s64 vtime_err;
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
seqcount_spinlock_t period_seqcount;
u64 period_at; /* wallclock starttime */
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
u64 period_at_vtime; /* vtime starttime */
atomic64_t cur_period; /* inc'd each period */
int busy_level; /* saturation history */
bool weights_updated;
atomic_t hweight_gen; /* for lazy hweights */
/* debt forgivness */
u64 dfgv_period_at;
u64 dfgv_period_rem;
u64 dfgv_usage_us_sum;
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
u64 autop_too_fast_at;
u64 autop_too_slow_at;
int autop_idx;
bool user_qos_params:1;
bool user_cost_model:1;
};
struct iocg_pcpu_stat {
local64_t abs_vusage;
};
struct iocg_stat {
u64 usage_us;
u64 wait_us;
u64 indebt_us;
u64 indelay_us;
};
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
/* per device-cgroup pair */
struct ioc_gq {
struct blkg_policy_data pd;
struct ioc *ioc;
/*
* A iocg can get its weight from two sources - an explicit
* per-device-cgroup configuration or the default weight of the
* cgroup. `cfg_weight` is the explicit per-device-cgroup
* configuration. `weight` is the effective considering both
* sources.
*
* When an idle cgroup becomes active its `active` goes from 0 to
* `weight`. `inuse` is the surplus adjusted active weight.
* `active` and `inuse` are used to calculate `hweight_active` and
* `hweight_inuse`.
*
* `last_inuse` remembers `inuse` while an iocg is idle to persist
* surplus adjustments.
*
* `inuse` may be adjusted dynamically during period. `saved_*` are used
* to determine and track adjustments.
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
*/
u32 cfg_weight;
u32 weight;
u32 active;
u32 inuse;
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
u32 last_inuse;
s64 saved_margin;
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
sector_t cursor; /* to detect randio */
/*
* `vtime` is this iocg's vtime cursor which progresses as IOs are
* issued. If lagging behind device vtime, the delta represents
* the currently available IO budget. If runnning ahead, the
* overage.
*
* `vtime_done` is the same but progressed on completion rather
* than issue. The delta behind `vtime` represents the cost of
* currently in-flight IOs.
*/
atomic64_t vtime;
atomic64_t done_vtime;
iocost: protect iocg->abs_vdebt with iocg->waitq.lock abs_vdebt is an atomic_64 which tracks how much over budget a given cgroup is and controls the activation of use_delay mechanism. Once a cgroup goes over budget from forced IOs, it has to pay it back with its future budget. The progress guarantee on debt paying comes from the iocg being active - active iocgs are processed by the periodic timer, which ensures that as time passes the debts dissipate and the iocg returns to normal operation. However, both iocg activation and vdebt handling are asynchronous and a sequence like the following may happen. 1. The iocg is in the process of being deactivated by the periodic timer. 2. A bio enters ioc_rqos_throttle(), calls iocg_activate() which returns without anything because it still sees that the iocg is already active. 3. The iocg is deactivated. 4. The bio from #2 is over budget but needs to be forced. It increases abs_vdebt and goes over the threshold and enables use_delay. 5. IO control is enabled for the iocg's subtree and now IOs are attributed to the descendant cgroups and the iocg itself no longer issues IOs. This leaves the iocg with stuck abs_vdebt - it has debt but inactive and no further IOs which can activate it. This can end up unduly punishing all the descendants cgroups. The usual throttling path has the same issue - the iocg must be active while throttled to ensure that future event will wake it up - and solves the problem by synchronizing the throttling path with a spinlock. abs_vdebt handling is another form of overage handling and shares a lot of characteristics including the fact that it isn't in the hottest path. This patch fixes the above and other possible races by strictly synchronizing abs_vdebt and use_delay handling with iocg->waitq.lock. Signed-off-by: Tejun Heo <tj@kernel.org> Reported-by: Vlad Dmitriev <vvd@fb.com> Cc: stable@vger.kernel.org # v5.4+ Fixes: e1518f63f246 ("blk-iocost: Don't let merges push vtime into the future") Signed-off-by: Jens Axboe <axboe@kernel.dk>
2020-05-04 23:27:54 +00:00
u64 abs_vdebt;
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
/* current delay in effect and when it started */
u64 delay;
u64 delay_at;
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
/*
* The period this iocg was last active in. Used for deactivation
* and invalidating `vtime`.
*/
atomic64_t active_period;
struct list_head active_list;
/* see __propagate_weights() and current_hweight() for details */
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
u64 child_active_sum;
u64 child_inuse_sum;
blk-iocost: implement Andy's method for donation weight updates iocost implements work conservation by reducing iocg->inuse and propagating the adjustment upwards proportionally. However, while I knew the target absolute hierarchical proportion - adjusted hweight_inuse, I couldn't figure out how to determine the iocg->inuse adjustment to achieve that and approximated the adjustment by scaling iocg->inuse using the proportion of the needed hweight_inuse changes. When nested, these scalings aren't accurate even when adjusting a single node as the donating node also receives the benefit of the donated portion. When multiple nodes are donating as they often do, they can be wildly wrong. iocost employed various safety nets to combat the inaccuracies. There are ample buffers in determining how much to donate, the adjustments are conservative and gradual. While it can achieve a reasonable level of work conservation in simple scenarios, the inaccuracies can easily add up leading to significant loss of total work. This in turn makes it difficult to closely cap vrate as vrate adjustment is needed to compensate for the loss of work. The combination of inaccurate donation calculations and vrate adjustments can lead to wide fluctuations and clunky overall behaviors. Andy Newell devised a method to calculate the needed ->inuse updates to achieve the target hweight_inuse's. The method is compatible with the proportional inuse adjustment propagation which allows all hot path operations to be local to each iocg. To roughly summarize, Andy's method divides the tree into donating and non-donating parts, calculates global donation rate which is used to determine the target hweight_inuse for each node, and then derives per-level proportions. There's non-trivial amount of math involved. Please refer to the following pdfs for detailed descriptions. https://drive.google.com/file/d/1PsJwxPFtjUnwOY1QJ5AeICCcsL7BM3bo https://drive.google.com/file/d/1vONz1-fzVO7oY5DXXsLjSxEtYYQbOvsE https://drive.google.com/file/d/1WcrltBOSPN0qXVdBgnKm4mdp9FhuEFQN This patch implements Andy's method in transfer_surpluses(). This makes the donation calculations accurate per cycle and enables further improvements in other parts of the donation logic. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2020-09-01 18:52:48 +00:00
u64 child_adjusted_sum;
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
int hweight_gen;
u32 hweight_active;
u32 hweight_inuse;
blk-iocost: implement Andy's method for donation weight updates iocost implements work conservation by reducing iocg->inuse and propagating the adjustment upwards proportionally. However, while I knew the target absolute hierarchical proportion - adjusted hweight_inuse, I couldn't figure out how to determine the iocg->inuse adjustment to achieve that and approximated the adjustment by scaling iocg->inuse using the proportion of the needed hweight_inuse changes. When nested, these scalings aren't accurate even when adjusting a single node as the donating node also receives the benefit of the donated portion. When multiple nodes are donating as they often do, they can be wildly wrong. iocost employed various safety nets to combat the inaccuracies. There are ample buffers in determining how much to donate, the adjustments are conservative and gradual. While it can achieve a reasonable level of work conservation in simple scenarios, the inaccuracies can easily add up leading to significant loss of total work. This in turn makes it difficult to closely cap vrate as vrate adjustment is needed to compensate for the loss of work. The combination of inaccurate donation calculations and vrate adjustments can lead to wide fluctuations and clunky overall behaviors. Andy Newell devised a method to calculate the needed ->inuse updates to achieve the target hweight_inuse's. The method is compatible with the proportional inuse adjustment propagation which allows all hot path operations to be local to each iocg. To roughly summarize, Andy's method divides the tree into donating and non-donating parts, calculates global donation rate which is used to determine the target hweight_inuse for each node, and then derives per-level proportions. There's non-trivial amount of math involved. Please refer to the following pdfs for detailed descriptions. https://drive.google.com/file/d/1PsJwxPFtjUnwOY1QJ5AeICCcsL7BM3bo https://drive.google.com/file/d/1vONz1-fzVO7oY5DXXsLjSxEtYYQbOvsE https://drive.google.com/file/d/1WcrltBOSPN0qXVdBgnKm4mdp9FhuEFQN This patch implements Andy's method in transfer_surpluses(). This makes the donation calculations accurate per cycle and enables further improvements in other parts of the donation logic. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2020-09-01 18:52:48 +00:00
u32 hweight_donating;
u32 hweight_after_donation;
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
struct list_head walk_list;
struct list_head surplus_list;
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
struct wait_queue_head waitq;
struct hrtimer waitq_timer;
/* timestamp at the latest activation */
u64 activated_at;
/* statistics */
struct iocg_pcpu_stat __percpu *pcpu_stat;
struct iocg_stat local_stat;
struct iocg_stat desc_stat;
struct iocg_stat last_stat;
u64 last_stat_abs_vusage;
blk-iocost: revamp donation amount determination iocost has various safety nets to combat inuse adjustment calculation inaccuracies. With Andy's method implemented in transfer_surpluses(), inuse adjustment calculations are now accurate and we can make donation amount determinations accurate too. * Stop keeping track of past usage history and using the maximum. Act on the immediate usage information. * Remove donation constraints defined by SURPLUS_* constants. Donate whatever isn't used. * Determine the donation amount so that the iocg will end up with MARGIN_TARGET_PCT budget at the end of the coming period assuming the same usage as the previous period. TARGET is set at 50% of period, which is the previous maximum. This provides smooth convergence for most repetitive IO patterns. * Apply donation logic early at 20% budget. There's no risk in doing so as the calculation is based on the delta between the current budget and the target budget at the end of the coming period. * Remove preemptive iocg activation for zero cost IOs. As donation can reach near zero now, the mere activation doesn't provide any protection anymore. In the unlikely case that this becomes a problem, the right solution is assigning appropriate costs for such IOs. This significantly improves the donation determination logic while also simplifying it. Now all donations are immediate, exact and smooth. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2020-09-01 18:52:49 +00:00
u64 usage_delta_us;
u64 wait_since;
u64 indebt_since;
u64 indelay_since;
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
/* this iocg's depth in the hierarchy and ancestors including self */
int level;
struct ioc_gq *ancestors[];
};
/* per cgroup */
struct ioc_cgrp {
struct blkcg_policy_data cpd;
unsigned int dfl_weight;
};
struct ioc_now {
u64 now_ns;
u64 now;
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
u64 vnow;
u64 vrate;
};
struct iocg_wait {
struct wait_queue_entry wait;
struct bio *bio;
u64 abs_cost;
bool committed;
};
struct iocg_wake_ctx {
struct ioc_gq *iocg;
u32 hw_inuse;
s64 vbudget;
};
static const struct ioc_params autop[] = {
[AUTOP_HDD] = {
.qos = {
[QOS_RLAT] = 250000, /* 250ms */
[QOS_WLAT] = 250000,
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
[QOS_MIN] = VRATE_MIN_PPM,
[QOS_MAX] = VRATE_MAX_PPM,
},
.i_lcoefs = {
[I_LCOEF_RBPS] = 174019176,
[I_LCOEF_RSEQIOPS] = 41708,
[I_LCOEF_RRANDIOPS] = 370,
[I_LCOEF_WBPS] = 178075866,
[I_LCOEF_WSEQIOPS] = 42705,
[I_LCOEF_WRANDIOPS] = 378,
},
},
[AUTOP_SSD_QD1] = {
.qos = {
[QOS_RLAT] = 25000, /* 25ms */
[QOS_WLAT] = 25000,
[QOS_MIN] = VRATE_MIN_PPM,
[QOS_MAX] = VRATE_MAX_PPM,
},
.i_lcoefs = {
[I_LCOEF_RBPS] = 245855193,
[I_LCOEF_RSEQIOPS] = 61575,
[I_LCOEF_RRANDIOPS] = 6946,
[I_LCOEF_WBPS] = 141365009,
[I_LCOEF_WSEQIOPS] = 33716,
[I_LCOEF_WRANDIOPS] = 26796,
},
},
[AUTOP_SSD_DFL] = {
.qos = {
[QOS_RLAT] = 25000, /* 25ms */
[QOS_WLAT] = 25000,
[QOS_MIN] = VRATE_MIN_PPM,
[QOS_MAX] = VRATE_MAX_PPM,
},
.i_lcoefs = {
[I_LCOEF_RBPS] = 488636629,
[I_LCOEF_RSEQIOPS] = 8932,
[I_LCOEF_RRANDIOPS] = 8518,
[I_LCOEF_WBPS] = 427891549,
[I_LCOEF_WSEQIOPS] = 28755,
[I_LCOEF_WRANDIOPS] = 21940,
},
.too_fast_vrate_pct = 500,
},
[AUTOP_SSD_FAST] = {
.qos = {
[QOS_RLAT] = 5000, /* 5ms */
[QOS_WLAT] = 5000,
[QOS_MIN] = VRATE_MIN_PPM,
[QOS_MAX] = VRATE_MAX_PPM,
},
.i_lcoefs = {
[I_LCOEF_RBPS] = 3102524156LLU,
[I_LCOEF_RSEQIOPS] = 724816,
[I_LCOEF_RRANDIOPS] = 778122,
[I_LCOEF_WBPS] = 1742780862LLU,
[I_LCOEF_WSEQIOPS] = 425702,
[I_LCOEF_WRANDIOPS] = 443193,
},
.too_slow_vrate_pct = 10,
},
};
/*
* vrate adjust percentages indexed by ioc->busy_level. We adjust up on
* vtime credit shortage and down on device saturation.
*/
static u32 vrate_adj_pct[] =
{ 0, 0, 0, 0,
1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2,
4, 4, 4, 4, 4, 4, 4, 4, 8, 8, 8, 8, 8, 8, 8, 8, 16 };
static struct blkcg_policy blkcg_policy_iocost;
/* accessors and helpers */
static struct ioc *rqos_to_ioc(struct rq_qos *rqos)
{
return container_of(rqos, struct ioc, rqos);
}
static struct ioc *q_to_ioc(struct request_queue *q)
{
return rqos_to_ioc(rq_qos_id(q, RQ_QOS_COST));
}
static const char *q_name(struct request_queue *q)
{
if (blk_queue_registered(q))
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
return kobject_name(q->kobj.parent);
else
return "<unknown>";
}
static const char __maybe_unused *ioc_name(struct ioc *ioc)
{
return q_name(ioc->rqos.q);
}
static struct ioc_gq *pd_to_iocg(struct blkg_policy_data *pd)
{
return pd ? container_of(pd, struct ioc_gq, pd) : NULL;
}
static struct ioc_gq *blkg_to_iocg(struct blkcg_gq *blkg)
{
return pd_to_iocg(blkg_to_pd(blkg, &blkcg_policy_iocost));
}
static struct blkcg_gq *iocg_to_blkg(struct ioc_gq *iocg)
{
return pd_to_blkg(&iocg->pd);
}
static struct ioc_cgrp *blkcg_to_iocc(struct blkcg *blkcg)
{
return container_of(blkcg_to_cpd(blkcg, &blkcg_policy_iocost),
struct ioc_cgrp, cpd);
}
/*
* Scale @abs_cost to the inverse of @hw_inuse. The lower the hierarchical
blk-iocost: Account force-charged overage in absolute vtime Currently, when a bio needs to be force-charged and there isn't enough budget, vtime is simply pushed into the future. This means that the cost of the whole bio is scaled using the current hweight and then charged immediately. Until the global vtime advances beyond this future vtime, the cgroup won't be allowed to issue normal IOs. This is incorrect and can lead to, for example, exploding vrate or extended stalls if vrate range is constrained. Consider the following scenario. 1. A cgroup with a very low hweight runs out of budget. 2. A storm of swap-out happens on it. All of them are scaled according to the current low hweight and charged to vtime pushing it to a far future. 3. All other cgroups go idle and now the above cgroup has access to the whole device. However, because vtime is already wound using the past low hweight, what its current hweight is doesn't matter until global vtime catches up to the local vtime. 4. As a result, either vrate gets ramped up extremely or the IOs stall while the underlying device is idle. This is because the hweight the overage is calculated at is different from the hweight that it's being paid at. Fix it by remembering the overage in absoulte vtime and continuously paying with the actual budget according to the current hweight at each period. Note that non-forced bios which wait already remembers the cost in absolute vtime. This brings forced-bio accounting in line. Signed-off-by: Tejun Heo <tj@kernel.org> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-09-04 19:45:52 +00:00
* weight, the more expensive each IO. Must round up.
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
*/
static u64 abs_cost_to_cost(u64 abs_cost, u32 hw_inuse)
{
return DIV64_U64_ROUND_UP(abs_cost * WEIGHT_ONE, hw_inuse);
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
}
blk-iocost: Account force-charged overage in absolute vtime Currently, when a bio needs to be force-charged and there isn't enough budget, vtime is simply pushed into the future. This means that the cost of the whole bio is scaled using the current hweight and then charged immediately. Until the global vtime advances beyond this future vtime, the cgroup won't be allowed to issue normal IOs. This is incorrect and can lead to, for example, exploding vrate or extended stalls if vrate range is constrained. Consider the following scenario. 1. A cgroup with a very low hweight runs out of budget. 2. A storm of swap-out happens on it. All of them are scaled according to the current low hweight and charged to vtime pushing it to a far future. 3. All other cgroups go idle and now the above cgroup has access to the whole device. However, because vtime is already wound using the past low hweight, what its current hweight is doesn't matter until global vtime catches up to the local vtime. 4. As a result, either vrate gets ramped up extremely or the IOs stall while the underlying device is idle. This is because the hweight the overage is calculated at is different from the hweight that it's being paid at. Fix it by remembering the overage in absoulte vtime and continuously paying with the actual budget according to the current hweight at each period. Note that non-forced bios which wait already remembers the cost in absolute vtime. This brings forced-bio accounting in line. Signed-off-by: Tejun Heo <tj@kernel.org> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-09-04 19:45:52 +00:00
/*
* The inverse of abs_cost_to_cost(). Must round up.
*/
static u64 cost_to_abs_cost(u64 cost, u32 hw_inuse)
{
return DIV64_U64_ROUND_UP(cost * hw_inuse, WEIGHT_ONE);
blk-iocost: Account force-charged overage in absolute vtime Currently, when a bio needs to be force-charged and there isn't enough budget, vtime is simply pushed into the future. This means that the cost of the whole bio is scaled using the current hweight and then charged immediately. Until the global vtime advances beyond this future vtime, the cgroup won't be allowed to issue normal IOs. This is incorrect and can lead to, for example, exploding vrate or extended stalls if vrate range is constrained. Consider the following scenario. 1. A cgroup with a very low hweight runs out of budget. 2. A storm of swap-out happens on it. All of them are scaled according to the current low hweight and charged to vtime pushing it to a far future. 3. All other cgroups go idle and now the above cgroup has access to the whole device. However, because vtime is already wound using the past low hweight, what its current hweight is doesn't matter until global vtime catches up to the local vtime. 4. As a result, either vrate gets ramped up extremely or the IOs stall while the underlying device is idle. This is because the hweight the overage is calculated at is different from the hweight that it's being paid at. Fix it by remembering the overage in absoulte vtime and continuously paying with the actual budget according to the current hweight at each period. Note that non-forced bios which wait already remembers the cost in absolute vtime. This brings forced-bio accounting in line. Signed-off-by: Tejun Heo <tj@kernel.org> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-09-04 19:45:52 +00:00
}
static void iocg_commit_bio(struct ioc_gq *iocg, struct bio *bio,
u64 abs_cost, u64 cost)
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
{
struct iocg_pcpu_stat *gcs;
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
bio->bi_iocost_cost = cost;
atomic64_add(cost, &iocg->vtime);
gcs = get_cpu_ptr(iocg->pcpu_stat);
local64_add(abs_cost, &gcs->abs_vusage);
put_cpu_ptr(gcs);
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
}
static void iocg_lock(struct ioc_gq *iocg, bool lock_ioc, unsigned long *flags)
{
if (lock_ioc) {
spin_lock_irqsave(&iocg->ioc->lock, *flags);
spin_lock(&iocg->waitq.lock);
} else {
spin_lock_irqsave(&iocg->waitq.lock, *flags);
}
}
static void iocg_unlock(struct ioc_gq *iocg, bool unlock_ioc, unsigned long *flags)
{
if (unlock_ioc) {
spin_unlock(&iocg->waitq.lock);
spin_unlock_irqrestore(&iocg->ioc->lock, *flags);
} else {
spin_unlock_irqrestore(&iocg->waitq.lock, *flags);
}
}
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
#define CREATE_TRACE_POINTS
#include <trace/events/iocost.h>
static void ioc_refresh_margins(struct ioc *ioc)
{
struct ioc_margins *margins = &ioc->margins;
u32 period_us = ioc->period_us;
u64 vrate = ioc->vtime_base_rate;
margins->min = (period_us * MARGIN_MIN_PCT / 100) * vrate;
blk-iocost: revamp donation amount determination iocost has various safety nets to combat inuse adjustment calculation inaccuracies. With Andy's method implemented in transfer_surpluses(), inuse adjustment calculations are now accurate and we can make donation amount determinations accurate too. * Stop keeping track of past usage history and using the maximum. Act on the immediate usage information. * Remove donation constraints defined by SURPLUS_* constants. Donate whatever isn't used. * Determine the donation amount so that the iocg will end up with MARGIN_TARGET_PCT budget at the end of the coming period assuming the same usage as the previous period. TARGET is set at 50% of period, which is the previous maximum. This provides smooth convergence for most repetitive IO patterns. * Apply donation logic early at 20% budget. There's no risk in doing so as the calculation is based on the delta between the current budget and the target budget at the end of the coming period. * Remove preemptive iocg activation for zero cost IOs. As donation can reach near zero now, the mere activation doesn't provide any protection anymore. In the unlikely case that this becomes a problem, the right solution is assigning appropriate costs for such IOs. This significantly improves the donation determination logic while also simplifying it. Now all donations are immediate, exact and smooth. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2020-09-01 18:52:49 +00:00
margins->low = (period_us * MARGIN_LOW_PCT / 100) * vrate;
margins->target = (period_us * MARGIN_TARGET_PCT / 100) * vrate;
}
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
/* latency Qos params changed, update period_us and all the dependent params */
static void ioc_refresh_period_us(struct ioc *ioc)
{
u32 ppm, lat, multi, period_us;
lockdep_assert_held(&ioc->lock);
/* pick the higher latency target */
if (ioc->params.qos[QOS_RLAT] >= ioc->params.qos[QOS_WLAT]) {
ppm = ioc->params.qos[QOS_RPPM];
lat = ioc->params.qos[QOS_RLAT];
} else {
ppm = ioc->params.qos[QOS_WPPM];
lat = ioc->params.qos[QOS_WLAT];
}
/*
* We want the period to be long enough to contain a healthy number
* of IOs while short enough for granular control. Define it as a
* multiple of the latency target. Ideally, the multiplier should
* be scaled according to the percentile so that it would nominally
* contain a certain number of requests. Let's be simpler and
* scale it linearly so that it's 2x >= pct(90) and 10x at pct(50).
*/
if (ppm)
multi = max_t(u32, (MILLION - ppm) / 50000, 2);
else
multi = 2;
period_us = multi * lat;
period_us = clamp_t(u32, period_us, MIN_PERIOD, MAX_PERIOD);
/* calculate dependent params */
ioc->period_us = period_us;
ioc->timer_slack_ns = div64_u64(
(u64)period_us * NSEC_PER_USEC * TIMER_SLACK_PCT,
100);
ioc_refresh_margins(ioc);
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
}
static int ioc_autop_idx(struct ioc *ioc)
{
int idx = ioc->autop_idx;
const struct ioc_params *p = &autop[idx];
u32 vrate_pct;
u64 now_ns;
/* rotational? */
if (!blk_queue_nonrot(ioc->rqos.q))
return AUTOP_HDD;
/* handle SATA SSDs w/ broken NCQ */
if (blk_queue_depth(ioc->rqos.q) == 1)
return AUTOP_SSD_QD1;
/* use one of the normal ssd sets */
if (idx < AUTOP_SSD_DFL)
return AUTOP_SSD_DFL;
/* if user is overriding anything, maintain what was there */
if (ioc->user_qos_params || ioc->user_cost_model)
return idx;
/* step up/down based on the vrate */
vrate_pct = div64_u64(ioc->vtime_base_rate * 100, VTIME_PER_USEC);
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
now_ns = ktime_get_ns();
if (p->too_fast_vrate_pct && p->too_fast_vrate_pct <= vrate_pct) {
if (!ioc->autop_too_fast_at)
ioc->autop_too_fast_at = now_ns;
if (now_ns - ioc->autop_too_fast_at >= AUTOP_CYCLE_NSEC)
return idx + 1;
} else {
ioc->autop_too_fast_at = 0;
}
if (p->too_slow_vrate_pct && p->too_slow_vrate_pct >= vrate_pct) {
if (!ioc->autop_too_slow_at)
ioc->autop_too_slow_at = now_ns;
if (now_ns - ioc->autop_too_slow_at >= AUTOP_CYCLE_NSEC)
return idx - 1;
} else {
ioc->autop_too_slow_at = 0;
}
return idx;
}
/*
* Take the followings as input
*
* @bps maximum sequential throughput
* @seqiops maximum sequential 4k iops
* @randiops maximum random 4k iops
*
* and calculate the linear model cost coefficients.
*
* *@page per-page cost 1s / (@bps / 4096)
* *@seqio base cost of a seq IO max((1s / @seqiops) - *@page, 0)
* @randiops base cost of a rand IO max((1s / @randiops) - *@page, 0)
*/
static void calc_lcoefs(u64 bps, u64 seqiops, u64 randiops,
u64 *page, u64 *seqio, u64 *randio)
{
u64 v;
*page = *seqio = *randio = 0;
if (bps)
*page = DIV64_U64_ROUND_UP(VTIME_PER_SEC,
DIV_ROUND_UP_ULL(bps, IOC_PAGE_SIZE));
if (seqiops) {
v = DIV64_U64_ROUND_UP(VTIME_PER_SEC, seqiops);
if (v > *page)
*seqio = v - *page;
}
if (randiops) {
v = DIV64_U64_ROUND_UP(VTIME_PER_SEC, randiops);
if (v > *page)
*randio = v - *page;
}
}
static void ioc_refresh_lcoefs(struct ioc *ioc)
{
u64 *u = ioc->params.i_lcoefs;
u64 *c = ioc->params.lcoefs;
calc_lcoefs(u[I_LCOEF_RBPS], u[I_LCOEF_RSEQIOPS], u[I_LCOEF_RRANDIOPS],
&c[LCOEF_RPAGE], &c[LCOEF_RSEQIO], &c[LCOEF_RRANDIO]);
calc_lcoefs(u[I_LCOEF_WBPS], u[I_LCOEF_WSEQIOPS], u[I_LCOEF_WRANDIOPS],
&c[LCOEF_WPAGE], &c[LCOEF_WSEQIO], &c[LCOEF_WRANDIO]);
}
static bool ioc_refresh_params(struct ioc *ioc, bool force)
{
const struct ioc_params *p;
int idx;
lockdep_assert_held(&ioc->lock);
idx = ioc_autop_idx(ioc);
p = &autop[idx];
if (idx == ioc->autop_idx && !force)
return false;
if (idx != ioc->autop_idx)
atomic64_set(&ioc->vtime_rate, VTIME_PER_USEC);
ioc->autop_idx = idx;
ioc->autop_too_fast_at = 0;
ioc->autop_too_slow_at = 0;
if (!ioc->user_qos_params)
memcpy(ioc->params.qos, p->qos, sizeof(p->qos));
if (!ioc->user_cost_model)
memcpy(ioc->params.i_lcoefs, p->i_lcoefs, sizeof(p->i_lcoefs));
ioc_refresh_period_us(ioc);
ioc_refresh_lcoefs(ioc);
ioc->vrate_min = DIV64_U64_ROUND_UP((u64)ioc->params.qos[QOS_MIN] *
VTIME_PER_USEC, MILLION);
ioc->vrate_max = div64_u64((u64)ioc->params.qos[QOS_MAX] *
VTIME_PER_USEC, MILLION);
return true;
}
/*
* When an iocg accumulates too much vtime or gets deactivated, we throw away
* some vtime, which lowers the overall device utilization. As the exact amount
* which is being thrown away is known, we can compensate by accelerating the
* vrate accordingly so that the extra vtime generated in the current period
* matches what got lost.
*/
static void ioc_refresh_vrate(struct ioc *ioc, struct ioc_now *now)
{
s64 pleft = ioc->period_at + ioc->period_us - now->now;
s64 vperiod = ioc->period_us * ioc->vtime_base_rate;
s64 vcomp, vcomp_min, vcomp_max;
lockdep_assert_held(&ioc->lock);
/* we need some time left in this period */
if (pleft <= 0)
goto done;
/*
* Calculate how much vrate should be adjusted to offset the error.
* Limit the amount of adjustment and deduct the adjusted amount from
* the error.
*/
vcomp = -div64_s64(ioc->vtime_err, pleft);
vcomp_min = -(ioc->vtime_base_rate >> 1);
vcomp_max = ioc->vtime_base_rate;
vcomp = clamp(vcomp, vcomp_min, vcomp_max);
ioc->vtime_err += vcomp * pleft;
atomic64_set(&ioc->vtime_rate, ioc->vtime_base_rate + vcomp);
done:
/* bound how much error can accumulate */
ioc->vtime_err = clamp(ioc->vtime_err, -vperiod, vperiod);
}
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
/* take a snapshot of the current [v]time and vrate */
static void ioc_now(struct ioc *ioc, struct ioc_now *now)
{
unsigned seq;
now->now_ns = ktime_get();
now->now = ktime_to_us(now->now_ns);
now->vrate = atomic64_read(&ioc->vtime_rate);
/*
* The current vtime is
*
* vtime at period start + (wallclock time since the start) * vrate
*
* As a consistent snapshot of `period_at_vtime` and `period_at` is
* needed, they're seqcount protected.
*/
do {
seq = read_seqcount_begin(&ioc->period_seqcount);
now->vnow = ioc->period_at_vtime +
(now->now - ioc->period_at) * now->vrate;
} while (read_seqcount_retry(&ioc->period_seqcount, seq));
}
static void ioc_start_period(struct ioc *ioc, struct ioc_now *now)
{
WARN_ON_ONCE(ioc->running != IOC_RUNNING);
write_seqcount_begin(&ioc->period_seqcount);
ioc->period_at = now->now;
ioc->period_at_vtime = now->vnow;
write_seqcount_end(&ioc->period_seqcount);
ioc->timer.expires = jiffies + usecs_to_jiffies(ioc->period_us);
add_timer(&ioc->timer);
}
/*
* Update @iocg's `active` and `inuse` to @active and @inuse, update level
* weight sums and propagate upwards accordingly. If @save, the current margin
* is saved to be used as reference for later inuse in-period adjustments.
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
*/
static void __propagate_weights(struct ioc_gq *iocg, u32 active, u32 inuse,
bool save, struct ioc_now *now)
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
{
struct ioc *ioc = iocg->ioc;
int lvl;
lockdep_assert_held(&ioc->lock);
inuse = clamp_t(u32, inuse, 1, active);
iocg->last_inuse = iocg->inuse;
if (save)
iocg->saved_margin = now->vnow - atomic64_read(&iocg->vtime);
if (active == iocg->active && inuse == iocg->inuse)
return;
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
for (lvl = iocg->level - 1; lvl >= 0; lvl--) {
struct ioc_gq *parent = iocg->ancestors[lvl];
struct ioc_gq *child = iocg->ancestors[lvl + 1];
u32 parent_active = 0, parent_inuse = 0;
/* update the level sums */
parent->child_active_sum += (s32)(active - child->active);
parent->child_inuse_sum += (s32)(inuse - child->inuse);
/* apply the udpates */
child->active = active;
child->inuse = inuse;
/*
* The delta between inuse and active sums indicates that
* that much of weight is being given away. Parent's inuse
* and active should reflect the ratio.
*/
if (parent->child_active_sum) {
parent_active = parent->weight;
parent_inuse = DIV64_U64_ROUND_UP(
parent_active * parent->child_inuse_sum,
parent->child_active_sum);
}
/* do we need to keep walking up? */
if (parent_active == parent->active &&
parent_inuse == parent->inuse)
break;
active = parent_active;
inuse = parent_inuse;
}
ioc->weights_updated = true;
}
static void commit_weights(struct ioc *ioc)
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
{
lockdep_assert_held(&ioc->lock);
if (ioc->weights_updated) {
/* paired with rmb in current_hweight(), see there */
smp_wmb();
atomic_inc(&ioc->hweight_gen);
ioc->weights_updated = false;
}
}
static void propagate_weights(struct ioc_gq *iocg, u32 active, u32 inuse,
bool save, struct ioc_now *now)
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
{
__propagate_weights(iocg, active, inuse, save, now);
commit_weights(iocg->ioc);
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
}
static void current_hweight(struct ioc_gq *iocg, u32 *hw_activep, u32 *hw_inusep)
{
struct ioc *ioc = iocg->ioc;
int lvl;
u32 hwa, hwi;
int ioc_gen;
/* hot path - if uptodate, use cached */
ioc_gen = atomic_read(&ioc->hweight_gen);
if (ioc_gen == iocg->hweight_gen)
goto out;
/*
* Paired with wmb in commit_weights(). If we saw the updated
* hweight_gen, all the weight updates from __propagate_weights() are
* visible too.
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
*
* We can race with weight updates during calculation and get it
* wrong. However, hweight_gen would have changed and a future
* reader will recalculate and we're guaranteed to discard the
* wrong result soon.
*/
smp_rmb();
hwa = hwi = WEIGHT_ONE;
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
for (lvl = 0; lvl <= iocg->level - 1; lvl++) {
struct ioc_gq *parent = iocg->ancestors[lvl];
struct ioc_gq *child = iocg->ancestors[lvl + 1];
u64 active_sum = READ_ONCE(parent->child_active_sum);
u64 inuse_sum = READ_ONCE(parent->child_inuse_sum);
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
u32 active = READ_ONCE(child->active);
u32 inuse = READ_ONCE(child->inuse);
/* we can race with deactivations and either may read as zero */
if (!active_sum || !inuse_sum)
continue;
active_sum = max_t(u64, active, active_sum);
hwa = div64_u64((u64)hwa * active, active_sum);
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
inuse_sum = max_t(u64, inuse, inuse_sum);
hwi = div64_u64((u64)hwi * inuse, inuse_sum);
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
}
iocg->hweight_active = max_t(u32, hwa, 1);
iocg->hweight_inuse = max_t(u32, hwi, 1);
iocg->hweight_gen = ioc_gen;
out:
if (hw_activep)
*hw_activep = iocg->hweight_active;
if (hw_inusep)
*hw_inusep = iocg->hweight_inuse;
}
/*
* Calculate the hweight_inuse @iocg would get with max @inuse assuming all the
* other weights stay unchanged.
*/
static u32 current_hweight_max(struct ioc_gq *iocg)
{
u32 hwm = WEIGHT_ONE;
u32 inuse = iocg->active;
u64 child_inuse_sum;
int lvl;
lockdep_assert_held(&iocg->ioc->lock);
for (lvl = iocg->level - 1; lvl >= 0; lvl--) {
struct ioc_gq *parent = iocg->ancestors[lvl];
struct ioc_gq *child = iocg->ancestors[lvl + 1];
child_inuse_sum = parent->child_inuse_sum + inuse - child->inuse;
hwm = div64_u64((u64)hwm * inuse, child_inuse_sum);
inuse = DIV64_U64_ROUND_UP(parent->active * child_inuse_sum,
parent->child_active_sum);
}
return max_t(u32, hwm, 1);
}
static void weight_updated(struct ioc_gq *iocg, struct ioc_now *now)
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
{
struct ioc *ioc = iocg->ioc;
struct blkcg_gq *blkg = iocg_to_blkg(iocg);
struct ioc_cgrp *iocc = blkcg_to_iocc(blkg->blkcg);
u32 weight;
lockdep_assert_held(&ioc->lock);
weight = iocg->cfg_weight ?: iocc->dfl_weight;
if (weight != iocg->weight && iocg->active)
propagate_weights(iocg, weight, iocg->inuse, true, now);
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
iocg->weight = weight;
}
static bool iocg_activate(struct ioc_gq *iocg, struct ioc_now *now)
{
struct ioc *ioc = iocg->ioc;
u64 last_period, cur_period;
u64 vtime, vtarget;
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
int i;
/*
* If seem to be already active, just update the stamp to tell the
* timer that we're still active. We don't mind occassional races.
*/
if (!list_empty(&iocg->active_list)) {
ioc_now(ioc, now);
cur_period = atomic64_read(&ioc->cur_period);
if (atomic64_read(&iocg->active_period) != cur_period)
atomic64_set(&iocg->active_period, cur_period);
return true;
}
/* racy check on internal node IOs, treat as root level IOs */
if (iocg->child_active_sum)
return false;
spin_lock_irq(&ioc->lock);
ioc_now(ioc, now);
/* update period */
cur_period = atomic64_read(&ioc->cur_period);
last_period = atomic64_read(&iocg->active_period);
atomic64_set(&iocg->active_period, cur_period);
/* already activated or breaking leaf-only constraint? */
if (!list_empty(&iocg->active_list))
goto succeed_unlock;
for (i = iocg->level - 1; i > 0; i--)
if (!list_empty(&iocg->ancestors[i]->active_list))
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
goto fail_unlock;
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
if (iocg->child_active_sum)
goto fail_unlock;
/*
* Always start with the target budget. On deactivation, we throw away
* anything above it.
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
*/
vtarget = now->vnow - ioc->margins.target;
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
vtime = atomic64_read(&iocg->vtime);
atomic64_add(vtarget - vtime, &iocg->vtime);
atomic64_add(vtarget - vtime, &iocg->done_vtime);
vtime = vtarget;
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
/*
* Activate, propagate weight and start period timer if not
* running. Reset hweight_gen to avoid accidental match from
* wrapping.
*/
iocg->hweight_gen = atomic_read(&ioc->hweight_gen) - 1;
list_add(&iocg->active_list, &ioc->active_iocgs);
propagate_weights(iocg, iocg->weight,
iocg->last_inuse ?: iocg->weight, true, now);
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
TRACE_IOCG_PATH(iocg_activate, iocg, now,
last_period, cur_period, vtime);
iocg->activated_at = now->now;
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
if (ioc->running == IOC_IDLE) {
ioc->running = IOC_RUNNING;
ioc->dfgv_period_at = now->now;
ioc->dfgv_period_rem = 0;
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
ioc_start_period(ioc, now);
}
succeed_unlock:
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
spin_unlock_irq(&ioc->lock);
return true;
fail_unlock:
spin_unlock_irq(&ioc->lock);
return false;
}
static bool iocg_kick_delay(struct ioc_gq *iocg, struct ioc_now *now)
{
struct ioc *ioc = iocg->ioc;
struct blkcg_gq *blkg = iocg_to_blkg(iocg);
u64 tdelta, delay, new_delay;
s64 vover, vover_pct;
u32 hwa;
lockdep_assert_held(&iocg->waitq.lock);
/* calculate the current delay in effect - 1/2 every second */
tdelta = now->now - iocg->delay_at;
if (iocg->delay)
delay = iocg->delay >> div64_u64(tdelta, USEC_PER_SEC);
else
delay = 0;
/* calculate the new delay from the debt amount */
current_hweight(iocg, &hwa, NULL);
vover = atomic64_read(&iocg->vtime) +
abs_cost_to_cost(iocg->abs_vdebt, hwa) - now->vnow;
vover_pct = div64_s64(100 * vover,
ioc->period_us * ioc->vtime_base_rate);
if (vover_pct <= MIN_DELAY_THR_PCT)
new_delay = 0;
else if (vover_pct >= MAX_DELAY_THR_PCT)
new_delay = MAX_DELAY;
else
new_delay = MIN_DELAY +
div_u64((MAX_DELAY - MIN_DELAY) *
(vover_pct - MIN_DELAY_THR_PCT),
MAX_DELAY_THR_PCT - MIN_DELAY_THR_PCT);
/* pick the higher one and apply */
if (new_delay > delay) {
iocg->delay = new_delay;
iocg->delay_at = now->now;
delay = new_delay;
}
if (delay >= MIN_DELAY) {
if (!iocg->indelay_since)
iocg->indelay_since = now->now;
blkcg_set_delay(blkg, delay * NSEC_PER_USEC);
return true;
} else {
if (iocg->indelay_since) {
iocg->local_stat.indelay_us += now->now - iocg->indelay_since;
iocg->indelay_since = 0;
}
iocg->delay = 0;
blkcg_clear_delay(blkg);
return false;
}
}
static void iocg_incur_debt(struct ioc_gq *iocg, u64 abs_cost,
struct ioc_now *now)
{
struct iocg_pcpu_stat *gcs;
lockdep_assert_held(&iocg->ioc->lock);
lockdep_assert_held(&iocg->waitq.lock);
WARN_ON_ONCE(list_empty(&iocg->active_list));
/*
* Once in debt, debt handling owns inuse. @iocg stays at the minimum
* inuse donating all of it share to others until its debt is paid off.
*/
if (!iocg->abs_vdebt && abs_cost) {
iocg->indebt_since = now->now;
propagate_weights(iocg, iocg->active, 0, false, now);
}
iocg->abs_vdebt += abs_cost;
gcs = get_cpu_ptr(iocg->pcpu_stat);
local64_add(abs_cost, &gcs->abs_vusage);
put_cpu_ptr(gcs);
}
static void iocg_pay_debt(struct ioc_gq *iocg, u64 abs_vpay,
struct ioc_now *now)
{
lockdep_assert_held(&iocg->ioc->lock);
lockdep_assert_held(&iocg->waitq.lock);
/* make sure that nobody messed with @iocg */
WARN_ON_ONCE(list_empty(&iocg->active_list));
WARN_ON_ONCE(iocg->inuse > 1);
iocg->abs_vdebt -= min(abs_vpay, iocg->abs_vdebt);
/* if debt is paid in full, restore inuse */
if (!iocg->abs_vdebt) {
iocg->local_stat.indebt_us += now->now - iocg->indebt_since;
iocg->indebt_since = 0;
propagate_weights(iocg, iocg->active, iocg->last_inuse,
false, now);
}
}
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
static int iocg_wake_fn(struct wait_queue_entry *wq_entry, unsigned mode,
int flags, void *key)
{
struct iocg_wait *wait = container_of(wq_entry, struct iocg_wait, wait);
struct iocg_wake_ctx *ctx = (struct iocg_wake_ctx *)key;
u64 cost = abs_cost_to_cost(wait->abs_cost, ctx->hw_inuse);
ctx->vbudget -= cost;
if (ctx->vbudget < 0)
return -1;
iocg_commit_bio(ctx->iocg, wait->bio, wait->abs_cost, cost);
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
/*
* autoremove_wake_function() removes the wait entry only when it
* actually changed the task state. We want the wait always
* removed. Remove explicitly and use default_wake_function().
*/
list_del_init(&wq_entry->entry);
wait->committed = true;
default_wake_function(wq_entry, mode, flags, key);
return 0;
}
/*
* Calculate the accumulated budget, pay debt if @pay_debt and wake up waiters
* accordingly. When @pay_debt is %true, the caller must be holding ioc->lock in
* addition to iocg->waitq.lock.
*/
static void iocg_kick_waitq(struct ioc_gq *iocg, bool pay_debt,
struct ioc_now *now)
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
{
struct ioc *ioc = iocg->ioc;
struct iocg_wake_ctx ctx = { .iocg = iocg };
u64 vshortage, expires, oexpires;
blk-iocost: Account force-charged overage in absolute vtime Currently, when a bio needs to be force-charged and there isn't enough budget, vtime is simply pushed into the future. This means that the cost of the whole bio is scaled using the current hweight and then charged immediately. Until the global vtime advances beyond this future vtime, the cgroup won't be allowed to issue normal IOs. This is incorrect and can lead to, for example, exploding vrate or extended stalls if vrate range is constrained. Consider the following scenario. 1. A cgroup with a very low hweight runs out of budget. 2. A storm of swap-out happens on it. All of them are scaled according to the current low hweight and charged to vtime pushing it to a far future. 3. All other cgroups go idle and now the above cgroup has access to the whole device. However, because vtime is already wound using the past low hweight, what its current hweight is doesn't matter until global vtime catches up to the local vtime. 4. As a result, either vrate gets ramped up extremely or the IOs stall while the underlying device is idle. This is because the hweight the overage is calculated at is different from the hweight that it's being paid at. Fix it by remembering the overage in absoulte vtime and continuously paying with the actual budget according to the current hweight at each period. Note that non-forced bios which wait already remembers the cost in absolute vtime. This brings forced-bio accounting in line. Signed-off-by: Tejun Heo <tj@kernel.org> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-09-04 19:45:52 +00:00
s64 vbudget;
u32 hwa;
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
lockdep_assert_held(&iocg->waitq.lock);
current_hweight(iocg, &hwa, NULL);
blk-iocost: Account force-charged overage in absolute vtime Currently, when a bio needs to be force-charged and there isn't enough budget, vtime is simply pushed into the future. This means that the cost of the whole bio is scaled using the current hweight and then charged immediately. Until the global vtime advances beyond this future vtime, the cgroup won't be allowed to issue normal IOs. This is incorrect and can lead to, for example, exploding vrate or extended stalls if vrate range is constrained. Consider the following scenario. 1. A cgroup with a very low hweight runs out of budget. 2. A storm of swap-out happens on it. All of them are scaled according to the current low hweight and charged to vtime pushing it to a far future. 3. All other cgroups go idle and now the above cgroup has access to the whole device. However, because vtime is already wound using the past low hweight, what its current hweight is doesn't matter until global vtime catches up to the local vtime. 4. As a result, either vrate gets ramped up extremely or the IOs stall while the underlying device is idle. This is because the hweight the overage is calculated at is different from the hweight that it's being paid at. Fix it by remembering the overage in absoulte vtime and continuously paying with the actual budget according to the current hweight at each period. Note that non-forced bios which wait already remembers the cost in absolute vtime. This brings forced-bio accounting in line. Signed-off-by: Tejun Heo <tj@kernel.org> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-09-04 19:45:52 +00:00
vbudget = now->vnow - atomic64_read(&iocg->vtime);
/* pay off debt */
if (pay_debt && iocg->abs_vdebt && vbudget > 0) {
u64 abs_vbudget = cost_to_abs_cost(vbudget, hwa);
u64 abs_vpay = min_t(u64, abs_vbudget, iocg->abs_vdebt);
u64 vpay = abs_cost_to_cost(abs_vpay, hwa);
blk-iocost: Account force-charged overage in absolute vtime Currently, when a bio needs to be force-charged and there isn't enough budget, vtime is simply pushed into the future. This means that the cost of the whole bio is scaled using the current hweight and then charged immediately. Until the global vtime advances beyond this future vtime, the cgroup won't be allowed to issue normal IOs. This is incorrect and can lead to, for example, exploding vrate or extended stalls if vrate range is constrained. Consider the following scenario. 1. A cgroup with a very low hweight runs out of budget. 2. A storm of swap-out happens on it. All of them are scaled according to the current low hweight and charged to vtime pushing it to a far future. 3. All other cgroups go idle and now the above cgroup has access to the whole device. However, because vtime is already wound using the past low hweight, what its current hweight is doesn't matter until global vtime catches up to the local vtime. 4. As a result, either vrate gets ramped up extremely or the IOs stall while the underlying device is idle. This is because the hweight the overage is calculated at is different from the hweight that it's being paid at. Fix it by remembering the overage in absoulte vtime and continuously paying with the actual budget according to the current hweight at each period. Note that non-forced bios which wait already remembers the cost in absolute vtime. This brings forced-bio accounting in line. Signed-off-by: Tejun Heo <tj@kernel.org> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-09-04 19:45:52 +00:00
lockdep_assert_held(&ioc->lock);
atomic64_add(vpay, &iocg->vtime);
atomic64_add(vpay, &iocg->done_vtime);
iocg_pay_debt(iocg, abs_vpay, now);
vbudget -= vpay;
}
if (iocg->abs_vdebt || iocg->delay)
iocg_kick_delay(iocg, now);
blk-iocost: Account force-charged overage in absolute vtime Currently, when a bio needs to be force-charged and there isn't enough budget, vtime is simply pushed into the future. This means that the cost of the whole bio is scaled using the current hweight and then charged immediately. Until the global vtime advances beyond this future vtime, the cgroup won't be allowed to issue normal IOs. This is incorrect and can lead to, for example, exploding vrate or extended stalls if vrate range is constrained. Consider the following scenario. 1. A cgroup with a very low hweight runs out of budget. 2. A storm of swap-out happens on it. All of them are scaled according to the current low hweight and charged to vtime pushing it to a far future. 3. All other cgroups go idle and now the above cgroup has access to the whole device. However, because vtime is already wound using the past low hweight, what its current hweight is doesn't matter until global vtime catches up to the local vtime. 4. As a result, either vrate gets ramped up extremely or the IOs stall while the underlying device is idle. This is because the hweight the overage is calculated at is different from the hweight that it's being paid at. Fix it by remembering the overage in absoulte vtime and continuously paying with the actual budget according to the current hweight at each period. Note that non-forced bios which wait already remembers the cost in absolute vtime. This brings forced-bio accounting in line. Signed-off-by: Tejun Heo <tj@kernel.org> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-09-04 19:45:52 +00:00
/*
* Debt can still be outstanding if we haven't paid all yet or the
* caller raced and called without @pay_debt. Shouldn't wake up waiters
* under debt. Make sure @vbudget reflects the outstanding amount and is
* not positive.
*/
if (iocg->abs_vdebt) {
s64 vdebt = abs_cost_to_cost(iocg->abs_vdebt, hwa);
vbudget = min_t(s64, 0, vbudget - vdebt);
}
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
/*
* Wake up the ones which are due and see how much vtime we'll need for
* the next one. As paying off debt restores hw_inuse, it must be read
* after the above debt payment.
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
*/
ctx.vbudget = vbudget;
current_hweight(iocg, NULL, &ctx.hw_inuse);
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
__wake_up_locked_key(&iocg->waitq, TASK_NORMAL, &ctx);
if (!waitqueue_active(&iocg->waitq)) {
if (iocg->wait_since) {
iocg->local_stat.wait_us += now->now - iocg->wait_since;
iocg->wait_since = 0;
}
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
return;
}
if (!iocg->wait_since)
iocg->wait_since = now->now;
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
if (WARN_ON_ONCE(ctx.vbudget >= 0))
return;
/* determine next wakeup, add a timer margin to guarantee chunking */
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
vshortage = -ctx.vbudget;
expires = now->now_ns +
DIV64_U64_ROUND_UP(vshortage, ioc->vtime_base_rate) *
NSEC_PER_USEC;
expires += ioc->timer_slack_ns;
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
/* if already active and close enough, don't bother */
oexpires = ktime_to_ns(hrtimer_get_softexpires(&iocg->waitq_timer));
if (hrtimer_is_queued(&iocg->waitq_timer) &&
abs(oexpires - expires) <= ioc->timer_slack_ns)
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
return;
hrtimer_start_range_ns(&iocg->waitq_timer, ns_to_ktime(expires),
ioc->timer_slack_ns, HRTIMER_MODE_ABS);
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
}
static enum hrtimer_restart iocg_waitq_timer_fn(struct hrtimer *timer)
{
struct ioc_gq *iocg = container_of(timer, struct ioc_gq, waitq_timer);
bool pay_debt = READ_ONCE(iocg->abs_vdebt);
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
struct ioc_now now;
unsigned long flags;
ioc_now(iocg->ioc, &now);
iocg_lock(iocg, pay_debt, &flags);
iocg_kick_waitq(iocg, pay_debt, &now);
iocg_unlock(iocg, pay_debt, &flags);
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
return HRTIMER_NORESTART;
}
static void ioc_lat_stat(struct ioc *ioc, u32 *missed_ppm_ar, u32 *rq_wait_pct_p)
{
u32 nr_met[2] = { };
u32 nr_missed[2] = { };
u64 rq_wait_ns = 0;
int cpu, rw;
for_each_online_cpu(cpu) {
struct ioc_pcpu_stat *stat = per_cpu_ptr(ioc->pcpu_stat, cpu);
u64 this_rq_wait_ns;
for (rw = READ; rw <= WRITE; rw++) {
u32 this_met = local_read(&stat->missed[rw].nr_met);
u32 this_missed = local_read(&stat->missed[rw].nr_missed);
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
nr_met[rw] += this_met - stat->missed[rw].last_met;
nr_missed[rw] += this_missed - stat->missed[rw].last_missed;
stat->missed[rw].last_met = this_met;
stat->missed[rw].last_missed = this_missed;
}
this_rq_wait_ns = local64_read(&stat->rq_wait_ns);
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
rq_wait_ns += this_rq_wait_ns - stat->last_rq_wait_ns;
stat->last_rq_wait_ns = this_rq_wait_ns;
}
for (rw = READ; rw <= WRITE; rw++) {
if (nr_met[rw] + nr_missed[rw])
missed_ppm_ar[rw] =
DIV64_U64_ROUND_UP((u64)nr_missed[rw] * MILLION,
nr_met[rw] + nr_missed[rw]);
else
missed_ppm_ar[rw] = 0;
}
*rq_wait_pct_p = div64_u64(rq_wait_ns * 100,
ioc->period_us * NSEC_PER_USEC);
}
/* was iocg idle this period? */
static bool iocg_is_idle(struct ioc_gq *iocg)
{
struct ioc *ioc = iocg->ioc;
/* did something get issued this period? */
if (atomic64_read(&iocg->active_period) ==
atomic64_read(&ioc->cur_period))
return false;
/* is something in flight? */
if (atomic64_read(&iocg->done_vtime) != atomic64_read(&iocg->vtime))
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
return false;
return true;
}
/*
* Call this function on the target leaf @iocg's to build pre-order traversal
* list of all the ancestors in @inner_walk. The inner nodes are linked through
* ->walk_list and the caller is responsible for dissolving the list after use.
*/
static void iocg_build_inner_walk(struct ioc_gq *iocg,
struct list_head *inner_walk)
{
int lvl;
WARN_ON_ONCE(!list_empty(&iocg->walk_list));
/* find the first ancestor which hasn't been visited yet */
for (lvl = iocg->level - 1; lvl >= 0; lvl--) {
if (!list_empty(&iocg->ancestors[lvl]->walk_list))
break;
}
/* walk down and visit the inner nodes to get pre-order traversal */
while (++lvl <= iocg->level - 1) {
struct ioc_gq *inner = iocg->ancestors[lvl];
/* record traversal order */
list_add_tail(&inner->walk_list, inner_walk);
}
}
/* collect per-cpu counters and propagate the deltas to the parent */
static void iocg_flush_stat_one(struct ioc_gq *iocg, struct ioc_now *now)
{
struct ioc *ioc = iocg->ioc;
struct iocg_stat new_stat;
u64 abs_vusage = 0;
u64 vusage_delta;
int cpu;
lockdep_assert_held(&iocg->ioc->lock);
/* collect per-cpu counters */
for_each_possible_cpu(cpu) {
abs_vusage += local64_read(
per_cpu_ptr(&iocg->pcpu_stat->abs_vusage, cpu));
}
vusage_delta = abs_vusage - iocg->last_stat_abs_vusage;
iocg->last_stat_abs_vusage = abs_vusage;
iocg->usage_delta_us = div64_u64(vusage_delta, ioc->vtime_base_rate);
iocg->local_stat.usage_us += iocg->usage_delta_us;
/* propagate upwards */
new_stat.usage_us =
iocg->local_stat.usage_us + iocg->desc_stat.usage_us;
new_stat.wait_us =
iocg->local_stat.wait_us + iocg->desc_stat.wait_us;
new_stat.indebt_us =
iocg->local_stat.indebt_us + iocg->desc_stat.indebt_us;
new_stat.indelay_us =
iocg->local_stat.indelay_us + iocg->desc_stat.indelay_us;
/* propagate the deltas to the parent */
if (iocg->level > 0) {
struct iocg_stat *parent_stat =
&iocg->ancestors[iocg->level - 1]->desc_stat;
parent_stat->usage_us +=
new_stat.usage_us - iocg->last_stat.usage_us;
parent_stat->wait_us +=
new_stat.wait_us - iocg->last_stat.wait_us;
parent_stat->indebt_us +=
new_stat.indebt_us - iocg->last_stat.indebt_us;
parent_stat->indelay_us +=
new_stat.indelay_us - iocg->last_stat.indelay_us;
}
iocg->last_stat = new_stat;
}
/* get stat counters ready for reading on all active iocgs */
static void iocg_flush_stat(struct list_head *target_iocgs, struct ioc_now *now)
{
LIST_HEAD(inner_walk);
struct ioc_gq *iocg, *tiocg;
/* flush leaves and build inner node walk list */
list_for_each_entry(iocg, target_iocgs, active_list) {
iocg_flush_stat_one(iocg, now);
iocg_build_inner_walk(iocg, &inner_walk);
}
/* keep flushing upwards by walking the inner list backwards */
list_for_each_entry_safe_reverse(iocg, tiocg, &inner_walk, walk_list) {
iocg_flush_stat_one(iocg, now);
list_del_init(&iocg->walk_list);
}
}
/*
* Determine what @iocg's hweight_inuse should be after donating unused
* capacity. @hwm is the upper bound and used to signal no donation. This
* function also throws away @iocg's excess budget.
*/
static u32 hweight_after_donation(struct ioc_gq *iocg, u32 old_hwi, u32 hwm,
u32 usage, struct ioc_now *now)
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
{
struct ioc *ioc = iocg->ioc;
u64 vtime = atomic64_read(&iocg->vtime);
blk-iocost: revamp donation amount determination iocost has various safety nets to combat inuse adjustment calculation inaccuracies. With Andy's method implemented in transfer_surpluses(), inuse adjustment calculations are now accurate and we can make donation amount determinations accurate too. * Stop keeping track of past usage history and using the maximum. Act on the immediate usage information. * Remove donation constraints defined by SURPLUS_* constants. Donate whatever isn't used. * Determine the donation amount so that the iocg will end up with MARGIN_TARGET_PCT budget at the end of the coming period assuming the same usage as the previous period. TARGET is set at 50% of period, which is the previous maximum. This provides smooth convergence for most repetitive IO patterns. * Apply donation logic early at 20% budget. There's no risk in doing so as the calculation is based on the delta between the current budget and the target budget at the end of the coming period. * Remove preemptive iocg activation for zero cost IOs. As donation can reach near zero now, the mere activation doesn't provide any protection anymore. In the unlikely case that this becomes a problem, the right solution is assigning appropriate costs for such IOs. This significantly improves the donation determination logic while also simplifying it. Now all donations are immediate, exact and smooth. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2020-09-01 18:52:49 +00:00
s64 excess, delta, target, new_hwi;
/* debt handling owns inuse for debtors */
if (iocg->abs_vdebt)
return 1;
/* see whether minimum margin requirement is met */
if (waitqueue_active(&iocg->waitq) ||
time_after64(vtime, now->vnow - ioc->margins.min))
return hwm;
/* throw away excess above target */
excess = now->vnow - vtime - ioc->margins.target;
if (excess > 0) {
atomic64_add(excess, &iocg->vtime);
atomic64_add(excess, &iocg->done_vtime);
vtime += excess;
ioc->vtime_err -= div64_u64(excess * old_hwi, WEIGHT_ONE);
}
blk-iocost: revamp donation amount determination iocost has various safety nets to combat inuse adjustment calculation inaccuracies. With Andy's method implemented in transfer_surpluses(), inuse adjustment calculations are now accurate and we can make donation amount determinations accurate too. * Stop keeping track of past usage history and using the maximum. Act on the immediate usage information. * Remove donation constraints defined by SURPLUS_* constants. Donate whatever isn't used. * Determine the donation amount so that the iocg will end up with MARGIN_TARGET_PCT budget at the end of the coming period assuming the same usage as the previous period. TARGET is set at 50% of period, which is the previous maximum. This provides smooth convergence for most repetitive IO patterns. * Apply donation logic early at 20% budget. There's no risk in doing so as the calculation is based on the delta between the current budget and the target budget at the end of the coming period. * Remove preemptive iocg activation for zero cost IOs. As donation can reach near zero now, the mere activation doesn't provide any protection anymore. In the unlikely case that this becomes a problem, the right solution is assigning appropriate costs for such IOs. This significantly improves the donation determination logic while also simplifying it. Now all donations are immediate, exact and smooth. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2020-09-01 18:52:49 +00:00
/*
* Let's say the distance between iocg's and device's vtimes as a
* fraction of period duration is delta. Assuming that the iocg will
* consume the usage determined above, we want to determine new_hwi so
* that delta equals MARGIN_TARGET at the end of the next period.
*
* We need to execute usage worth of IOs while spending the sum of the
* new budget (1 - MARGIN_TARGET) and the leftover from the last period
* (delta):
*
* usage = (1 - MARGIN_TARGET + delta) * new_hwi
*
* Therefore, the new_hwi is:
*
* new_hwi = usage / (1 - MARGIN_TARGET + delta)
*/
delta = div64_s64(WEIGHT_ONE * (now->vnow - vtime),
now->vnow - ioc->period_at_vtime);
target = WEIGHT_ONE * MARGIN_TARGET_PCT / 100;
new_hwi = div64_s64(WEIGHT_ONE * usage, WEIGHT_ONE - target + delta);
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
blk-iocost: revamp donation amount determination iocost has various safety nets to combat inuse adjustment calculation inaccuracies. With Andy's method implemented in transfer_surpluses(), inuse adjustment calculations are now accurate and we can make donation amount determinations accurate too. * Stop keeping track of past usage history and using the maximum. Act on the immediate usage information. * Remove donation constraints defined by SURPLUS_* constants. Donate whatever isn't used. * Determine the donation amount so that the iocg will end up with MARGIN_TARGET_PCT budget at the end of the coming period assuming the same usage as the previous period. TARGET is set at 50% of period, which is the previous maximum. This provides smooth convergence for most repetitive IO patterns. * Apply donation logic early at 20% budget. There's no risk in doing so as the calculation is based on the delta between the current budget and the target budget at the end of the coming period. * Remove preemptive iocg activation for zero cost IOs. As donation can reach near zero now, the mere activation doesn't provide any protection anymore. In the unlikely case that this becomes a problem, the right solution is assigning appropriate costs for such IOs. This significantly improves the donation determination logic while also simplifying it. Now all donations are immediate, exact and smooth. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2020-09-01 18:52:49 +00:00
return clamp_t(s64, new_hwi, 1, hwm);
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
}
blk-iocost: implement Andy's method for donation weight updates iocost implements work conservation by reducing iocg->inuse and propagating the adjustment upwards proportionally. However, while I knew the target absolute hierarchical proportion - adjusted hweight_inuse, I couldn't figure out how to determine the iocg->inuse adjustment to achieve that and approximated the adjustment by scaling iocg->inuse using the proportion of the needed hweight_inuse changes. When nested, these scalings aren't accurate even when adjusting a single node as the donating node also receives the benefit of the donated portion. When multiple nodes are donating as they often do, they can be wildly wrong. iocost employed various safety nets to combat the inaccuracies. There are ample buffers in determining how much to donate, the adjustments are conservative and gradual. While it can achieve a reasonable level of work conservation in simple scenarios, the inaccuracies can easily add up leading to significant loss of total work. This in turn makes it difficult to closely cap vrate as vrate adjustment is needed to compensate for the loss of work. The combination of inaccurate donation calculations and vrate adjustments can lead to wide fluctuations and clunky overall behaviors. Andy Newell devised a method to calculate the needed ->inuse updates to achieve the target hweight_inuse's. The method is compatible with the proportional inuse adjustment propagation which allows all hot path operations to be local to each iocg. To roughly summarize, Andy's method divides the tree into donating and non-donating parts, calculates global donation rate which is used to determine the target hweight_inuse for each node, and then derives per-level proportions. There's non-trivial amount of math involved. Please refer to the following pdfs for detailed descriptions. https://drive.google.com/file/d/1PsJwxPFtjUnwOY1QJ5AeICCcsL7BM3bo https://drive.google.com/file/d/1vONz1-fzVO7oY5DXXsLjSxEtYYQbOvsE https://drive.google.com/file/d/1WcrltBOSPN0qXVdBgnKm4mdp9FhuEFQN This patch implements Andy's method in transfer_surpluses(). This makes the donation calculations accurate per cycle and enables further improvements in other parts of the donation logic. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2020-09-01 18:52:48 +00:00
/*
* For work-conservation, an iocg which isn't using all of its share should
* donate the leftover to other iocgs. There are two ways to achieve this - 1.
* bumping up vrate accordingly 2. lowering the donating iocg's inuse weight.
*
* #1 is mathematically simpler but has the drawback of requiring synchronous
* global hweight_inuse updates when idle iocg's get activated or inuse weights
* change due to donation snapbacks as it has the possibility of grossly
* overshooting what's allowed by the model and vrate.
*
* #2 is inherently safe with local operations. The donating iocg can easily
* snap back to higher weights when needed without worrying about impacts on
* other nodes as the impacts will be inherently correct. This also makes idle
* iocg activations safe. The only effect activations have is decreasing
* hweight_inuse of others, the right solution to which is for those iocgs to
* snap back to higher weights.
*
* So, we go with #2. The challenge is calculating how each donating iocg's
* inuse should be adjusted to achieve the target donation amounts. This is done
* using Andy's method described in the following pdf.
*
* https://drive.google.com/file/d/1PsJwxPFtjUnwOY1QJ5AeICCcsL7BM3bo
*
* Given the weights and target after-donation hweight_inuse values, Andy's
* method determines how the proportional distribution should look like at each
* sibling level to maintain the relative relationship between all non-donating
* pairs. To roughly summarize, it divides the tree into donating and
* non-donating parts, calculates global donation rate which is used to
* determine the target hweight_inuse for each node, and then derives per-level
* proportions.
*
* The following pdf shows that global distribution calculated this way can be
* achieved by scaling inuse weights of donating leaves and propagating the
* adjustments upwards proportionally.
*
* https://drive.google.com/file/d/1vONz1-fzVO7oY5DXXsLjSxEtYYQbOvsE
*
* Combining the above two, we can determine how each leaf iocg's inuse should
* be adjusted to achieve the target donation.
*
* https://drive.google.com/file/d/1WcrltBOSPN0qXVdBgnKm4mdp9FhuEFQN
*
* The inline comments use symbols from the last pdf.
*
* b is the sum of the absolute budgets in the subtree. 1 for the root node.
* f is the sum of the absolute budgets of non-donating nodes in the subtree.
* t is the sum of the absolute budgets of donating nodes in the subtree.
* w is the weight of the node. w = w_f + w_t
* w_f is the non-donating portion of w. w_f = w * f / b
* w_b is the donating portion of w. w_t = w * t / b
* s is the sum of all sibling weights. s = Sum(w) for siblings
* s_f and s_t are the non-donating and donating portions of s.
*
* Subscript p denotes the parent's counterpart and ' the adjusted value - e.g.
* w_pt is the donating portion of the parent's weight and w'_pt the same value
* after adjustments. Subscript r denotes the root node's values.
*/
static void transfer_surpluses(struct list_head *surpluses, struct ioc_now *now)
{
blk-iocost: implement Andy's method for donation weight updates iocost implements work conservation by reducing iocg->inuse and propagating the adjustment upwards proportionally. However, while I knew the target absolute hierarchical proportion - adjusted hweight_inuse, I couldn't figure out how to determine the iocg->inuse adjustment to achieve that and approximated the adjustment by scaling iocg->inuse using the proportion of the needed hweight_inuse changes. When nested, these scalings aren't accurate even when adjusting a single node as the donating node also receives the benefit of the donated portion. When multiple nodes are donating as they often do, they can be wildly wrong. iocost employed various safety nets to combat the inaccuracies. There are ample buffers in determining how much to donate, the adjustments are conservative and gradual. While it can achieve a reasonable level of work conservation in simple scenarios, the inaccuracies can easily add up leading to significant loss of total work. This in turn makes it difficult to closely cap vrate as vrate adjustment is needed to compensate for the loss of work. The combination of inaccurate donation calculations and vrate adjustments can lead to wide fluctuations and clunky overall behaviors. Andy Newell devised a method to calculate the needed ->inuse updates to achieve the target hweight_inuse's. The method is compatible with the proportional inuse adjustment propagation which allows all hot path operations to be local to each iocg. To roughly summarize, Andy's method divides the tree into donating and non-donating parts, calculates global donation rate which is used to determine the target hweight_inuse for each node, and then derives per-level proportions. There's non-trivial amount of math involved. Please refer to the following pdfs for detailed descriptions. https://drive.google.com/file/d/1PsJwxPFtjUnwOY1QJ5AeICCcsL7BM3bo https://drive.google.com/file/d/1vONz1-fzVO7oY5DXXsLjSxEtYYQbOvsE https://drive.google.com/file/d/1WcrltBOSPN0qXVdBgnKm4mdp9FhuEFQN This patch implements Andy's method in transfer_surpluses(). This makes the donation calculations accurate per cycle and enables further improvements in other parts of the donation logic. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2020-09-01 18:52:48 +00:00
LIST_HEAD(over_hwa);
LIST_HEAD(inner_walk);
struct ioc_gq *iocg, *tiocg, *root_iocg;
u32 after_sum, over_sum, over_target, gamma;
blk-iocost: implement Andy's method for donation weight updates iocost implements work conservation by reducing iocg->inuse and propagating the adjustment upwards proportionally. However, while I knew the target absolute hierarchical proportion - adjusted hweight_inuse, I couldn't figure out how to determine the iocg->inuse adjustment to achieve that and approximated the adjustment by scaling iocg->inuse using the proportion of the needed hweight_inuse changes. When nested, these scalings aren't accurate even when adjusting a single node as the donating node also receives the benefit of the donated portion. When multiple nodes are donating as they often do, they can be wildly wrong. iocost employed various safety nets to combat the inaccuracies. There are ample buffers in determining how much to donate, the adjustments are conservative and gradual. While it can achieve a reasonable level of work conservation in simple scenarios, the inaccuracies can easily add up leading to significant loss of total work. This in turn makes it difficult to closely cap vrate as vrate adjustment is needed to compensate for the loss of work. The combination of inaccurate donation calculations and vrate adjustments can lead to wide fluctuations and clunky overall behaviors. Andy Newell devised a method to calculate the needed ->inuse updates to achieve the target hweight_inuse's. The method is compatible with the proportional inuse adjustment propagation which allows all hot path operations to be local to each iocg. To roughly summarize, Andy's method divides the tree into donating and non-donating parts, calculates global donation rate which is used to determine the target hweight_inuse for each node, and then derives per-level proportions. There's non-trivial amount of math involved. Please refer to the following pdfs for detailed descriptions. https://drive.google.com/file/d/1PsJwxPFtjUnwOY1QJ5AeICCcsL7BM3bo https://drive.google.com/file/d/1vONz1-fzVO7oY5DXXsLjSxEtYYQbOvsE https://drive.google.com/file/d/1WcrltBOSPN0qXVdBgnKm4mdp9FhuEFQN This patch implements Andy's method in transfer_surpluses(). This makes the donation calculations accurate per cycle and enables further improvements in other parts of the donation logic. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2020-09-01 18:52:48 +00:00
/*
* It's pretty unlikely but possible for the total sum of
* hweight_after_donation's to be higher than WEIGHT_ONE, which will
* confuse the following calculations. If such condition is detected,
* scale down everyone over its full share equally to keep the sum below
* WEIGHT_ONE.
*/
after_sum = 0;
over_sum = 0;
list_for_each_entry(iocg, surpluses, surplus_list) {
blk-iocost: implement Andy's method for donation weight updates iocost implements work conservation by reducing iocg->inuse and propagating the adjustment upwards proportionally. However, while I knew the target absolute hierarchical proportion - adjusted hweight_inuse, I couldn't figure out how to determine the iocg->inuse adjustment to achieve that and approximated the adjustment by scaling iocg->inuse using the proportion of the needed hweight_inuse changes. When nested, these scalings aren't accurate even when adjusting a single node as the donating node also receives the benefit of the donated portion. When multiple nodes are donating as they often do, they can be wildly wrong. iocost employed various safety nets to combat the inaccuracies. There are ample buffers in determining how much to donate, the adjustments are conservative and gradual. While it can achieve a reasonable level of work conservation in simple scenarios, the inaccuracies can easily add up leading to significant loss of total work. This in turn makes it difficult to closely cap vrate as vrate adjustment is needed to compensate for the loss of work. The combination of inaccurate donation calculations and vrate adjustments can lead to wide fluctuations and clunky overall behaviors. Andy Newell devised a method to calculate the needed ->inuse updates to achieve the target hweight_inuse's. The method is compatible with the proportional inuse adjustment propagation which allows all hot path operations to be local to each iocg. To roughly summarize, Andy's method divides the tree into donating and non-donating parts, calculates global donation rate which is used to determine the target hweight_inuse for each node, and then derives per-level proportions. There's non-trivial amount of math involved. Please refer to the following pdfs for detailed descriptions. https://drive.google.com/file/d/1PsJwxPFtjUnwOY1QJ5AeICCcsL7BM3bo https://drive.google.com/file/d/1vONz1-fzVO7oY5DXXsLjSxEtYYQbOvsE https://drive.google.com/file/d/1WcrltBOSPN0qXVdBgnKm4mdp9FhuEFQN This patch implements Andy's method in transfer_surpluses(). This makes the donation calculations accurate per cycle and enables further improvements in other parts of the donation logic. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2020-09-01 18:52:48 +00:00
u32 hwa;
blk-iocost: implement Andy's method for donation weight updates iocost implements work conservation by reducing iocg->inuse and propagating the adjustment upwards proportionally. However, while I knew the target absolute hierarchical proportion - adjusted hweight_inuse, I couldn't figure out how to determine the iocg->inuse adjustment to achieve that and approximated the adjustment by scaling iocg->inuse using the proportion of the needed hweight_inuse changes. When nested, these scalings aren't accurate even when adjusting a single node as the donating node also receives the benefit of the donated portion. When multiple nodes are donating as they often do, they can be wildly wrong. iocost employed various safety nets to combat the inaccuracies. There are ample buffers in determining how much to donate, the adjustments are conservative and gradual. While it can achieve a reasonable level of work conservation in simple scenarios, the inaccuracies can easily add up leading to significant loss of total work. This in turn makes it difficult to closely cap vrate as vrate adjustment is needed to compensate for the loss of work. The combination of inaccurate donation calculations and vrate adjustments can lead to wide fluctuations and clunky overall behaviors. Andy Newell devised a method to calculate the needed ->inuse updates to achieve the target hweight_inuse's. The method is compatible with the proportional inuse adjustment propagation which allows all hot path operations to be local to each iocg. To roughly summarize, Andy's method divides the tree into donating and non-donating parts, calculates global donation rate which is used to determine the target hweight_inuse for each node, and then derives per-level proportions. There's non-trivial amount of math involved. Please refer to the following pdfs for detailed descriptions. https://drive.google.com/file/d/1PsJwxPFtjUnwOY1QJ5AeICCcsL7BM3bo https://drive.google.com/file/d/1vONz1-fzVO7oY5DXXsLjSxEtYYQbOvsE https://drive.google.com/file/d/1WcrltBOSPN0qXVdBgnKm4mdp9FhuEFQN This patch implements Andy's method in transfer_surpluses(). This makes the donation calculations accurate per cycle and enables further improvements in other parts of the donation logic. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2020-09-01 18:52:48 +00:00
current_hweight(iocg, &hwa, NULL);
after_sum += iocg->hweight_after_donation;
blk-iocost: implement Andy's method for donation weight updates iocost implements work conservation by reducing iocg->inuse and propagating the adjustment upwards proportionally. However, while I knew the target absolute hierarchical proportion - adjusted hweight_inuse, I couldn't figure out how to determine the iocg->inuse adjustment to achieve that and approximated the adjustment by scaling iocg->inuse using the proportion of the needed hweight_inuse changes. When nested, these scalings aren't accurate even when adjusting a single node as the donating node also receives the benefit of the donated portion. When multiple nodes are donating as they often do, they can be wildly wrong. iocost employed various safety nets to combat the inaccuracies. There are ample buffers in determining how much to donate, the adjustments are conservative and gradual. While it can achieve a reasonable level of work conservation in simple scenarios, the inaccuracies can easily add up leading to significant loss of total work. This in turn makes it difficult to closely cap vrate as vrate adjustment is needed to compensate for the loss of work. The combination of inaccurate donation calculations and vrate adjustments can lead to wide fluctuations and clunky overall behaviors. Andy Newell devised a method to calculate the needed ->inuse updates to achieve the target hweight_inuse's. The method is compatible with the proportional inuse adjustment propagation which allows all hot path operations to be local to each iocg. To roughly summarize, Andy's method divides the tree into donating and non-donating parts, calculates global donation rate which is used to determine the target hweight_inuse for each node, and then derives per-level proportions. There's non-trivial amount of math involved. Please refer to the following pdfs for detailed descriptions. https://drive.google.com/file/d/1PsJwxPFtjUnwOY1QJ5AeICCcsL7BM3bo https://drive.google.com/file/d/1vONz1-fzVO7oY5DXXsLjSxEtYYQbOvsE https://drive.google.com/file/d/1WcrltBOSPN0qXVdBgnKm4mdp9FhuEFQN This patch implements Andy's method in transfer_surpluses(). This makes the donation calculations accurate per cycle and enables further improvements in other parts of the donation logic. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2020-09-01 18:52:48 +00:00
if (iocg->hweight_after_donation > hwa) {
over_sum += iocg->hweight_after_donation;
list_add(&iocg->walk_list, &over_hwa);
}
}
blk-iocost: implement Andy's method for donation weight updates iocost implements work conservation by reducing iocg->inuse and propagating the adjustment upwards proportionally. However, while I knew the target absolute hierarchical proportion - adjusted hweight_inuse, I couldn't figure out how to determine the iocg->inuse adjustment to achieve that and approximated the adjustment by scaling iocg->inuse using the proportion of the needed hweight_inuse changes. When nested, these scalings aren't accurate even when adjusting a single node as the donating node also receives the benefit of the donated portion. When multiple nodes are donating as they often do, they can be wildly wrong. iocost employed various safety nets to combat the inaccuracies. There are ample buffers in determining how much to donate, the adjustments are conservative and gradual. While it can achieve a reasonable level of work conservation in simple scenarios, the inaccuracies can easily add up leading to significant loss of total work. This in turn makes it difficult to closely cap vrate as vrate adjustment is needed to compensate for the loss of work. The combination of inaccurate donation calculations and vrate adjustments can lead to wide fluctuations and clunky overall behaviors. Andy Newell devised a method to calculate the needed ->inuse updates to achieve the target hweight_inuse's. The method is compatible with the proportional inuse adjustment propagation which allows all hot path operations to be local to each iocg. To roughly summarize, Andy's method divides the tree into donating and non-donating parts, calculates global donation rate which is used to determine the target hweight_inuse for each node, and then derives per-level proportions. There's non-trivial amount of math involved. Please refer to the following pdfs for detailed descriptions. https://drive.google.com/file/d/1PsJwxPFtjUnwOY1QJ5AeICCcsL7BM3bo https://drive.google.com/file/d/1vONz1-fzVO7oY5DXXsLjSxEtYYQbOvsE https://drive.google.com/file/d/1WcrltBOSPN0qXVdBgnKm4mdp9FhuEFQN This patch implements Andy's method in transfer_surpluses(). This makes the donation calculations accurate per cycle and enables further improvements in other parts of the donation logic. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2020-09-01 18:52:48 +00:00
if (after_sum >= WEIGHT_ONE) {
/*
* The delta should be deducted from the over_sum, calculate
* target over_sum value.
*/
u32 over_delta = after_sum - (WEIGHT_ONE - 1);
WARN_ON_ONCE(over_sum <= over_delta);
over_target = over_sum - over_delta;
} else {
over_target = 0;
}
list_for_each_entry_safe(iocg, tiocg, &over_hwa, walk_list) {
if (over_target)
iocg->hweight_after_donation =
div_u64((u64)iocg->hweight_after_donation *
over_target, over_sum);
list_del_init(&iocg->walk_list);
}
/*
* Build pre-order inner node walk list and prepare for donation
* adjustment calculations.
*/
list_for_each_entry(iocg, surpluses, surplus_list) {
iocg_build_inner_walk(iocg, &inner_walk);
}
root_iocg = list_first_entry(&inner_walk, struct ioc_gq, walk_list);
WARN_ON_ONCE(root_iocg->level > 0);
list_for_each_entry(iocg, &inner_walk, walk_list) {
iocg->child_adjusted_sum = 0;
iocg->hweight_donating = 0;
iocg->hweight_after_donation = 0;
}
/*
* Propagate the donating budget (b_t) and after donation budget (b'_t)
* up the hierarchy.
*/
list_for_each_entry(iocg, surpluses, surplus_list) {
struct ioc_gq *parent = iocg->ancestors[iocg->level - 1];
parent->hweight_donating += iocg->hweight_donating;
parent->hweight_after_donation += iocg->hweight_after_donation;
}
list_for_each_entry_reverse(iocg, &inner_walk, walk_list) {
if (iocg->level > 0) {
struct ioc_gq *parent = iocg->ancestors[iocg->level - 1];
parent->hweight_donating += iocg->hweight_donating;
parent->hweight_after_donation += iocg->hweight_after_donation;
}
}
/*
* Calculate inner hwa's (b) and make sure the donation values are
* within the accepted ranges as we're doing low res calculations with
* roundups.
*/
list_for_each_entry(iocg, &inner_walk, walk_list) {
if (iocg->level) {
struct ioc_gq *parent = iocg->ancestors[iocg->level - 1];
iocg->hweight_active = DIV64_U64_ROUND_UP(
(u64)parent->hweight_active * iocg->active,
parent->child_active_sum);
}
iocg->hweight_donating = min(iocg->hweight_donating,
iocg->hweight_active);
iocg->hweight_after_donation = min(iocg->hweight_after_donation,
iocg->hweight_donating - 1);
if (WARN_ON_ONCE(iocg->hweight_active <= 1 ||
iocg->hweight_donating <= 1 ||
iocg->hweight_after_donation == 0)) {
pr_warn("iocg: invalid donation weights in ");
pr_cont_cgroup_path(iocg_to_blkg(iocg)->blkcg->css.cgroup);
pr_cont(": active=%u donating=%u after=%u\n",
iocg->hweight_active, iocg->hweight_donating,
iocg->hweight_after_donation);
}
}
/*
* Calculate the global donation rate (gamma) - the rate to adjust
* non-donating budgets by.
*
* No need to use 64bit multiplication here as the first operand is
* guaranteed to be smaller than WEIGHT_ONE (1<<16).
*
* We know that there are beneficiary nodes and the sum of the donating
* hweights can't be whole; however, due to the round-ups during hweight
* calculations, root_iocg->hweight_donating might still end up equal to
* or greater than whole. Limit the range when calculating the divider.
blk-iocost: implement Andy's method for donation weight updates iocost implements work conservation by reducing iocg->inuse and propagating the adjustment upwards proportionally. However, while I knew the target absolute hierarchical proportion - adjusted hweight_inuse, I couldn't figure out how to determine the iocg->inuse adjustment to achieve that and approximated the adjustment by scaling iocg->inuse using the proportion of the needed hweight_inuse changes. When nested, these scalings aren't accurate even when adjusting a single node as the donating node also receives the benefit of the donated portion. When multiple nodes are donating as they often do, they can be wildly wrong. iocost employed various safety nets to combat the inaccuracies. There are ample buffers in determining how much to donate, the adjustments are conservative and gradual. While it can achieve a reasonable level of work conservation in simple scenarios, the inaccuracies can easily add up leading to significant loss of total work. This in turn makes it difficult to closely cap vrate as vrate adjustment is needed to compensate for the loss of work. The combination of inaccurate donation calculations and vrate adjustments can lead to wide fluctuations and clunky overall behaviors. Andy Newell devised a method to calculate the needed ->inuse updates to achieve the target hweight_inuse's. The method is compatible with the proportional inuse adjustment propagation which allows all hot path operations to be local to each iocg. To roughly summarize, Andy's method divides the tree into donating and non-donating parts, calculates global donation rate which is used to determine the target hweight_inuse for each node, and then derives per-level proportions. There's non-trivial amount of math involved. Please refer to the following pdfs for detailed descriptions. https://drive.google.com/file/d/1PsJwxPFtjUnwOY1QJ5AeICCcsL7BM3bo https://drive.google.com/file/d/1vONz1-fzVO7oY5DXXsLjSxEtYYQbOvsE https://drive.google.com/file/d/1WcrltBOSPN0qXVdBgnKm4mdp9FhuEFQN This patch implements Andy's method in transfer_surpluses(). This makes the donation calculations accurate per cycle and enables further improvements in other parts of the donation logic. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2020-09-01 18:52:48 +00:00
*
* gamma = (1 - t_r') / (1 - t_r)
*/
gamma = DIV_ROUND_UP(
(WEIGHT_ONE - root_iocg->hweight_after_donation) * WEIGHT_ONE,
WEIGHT_ONE - min_t(u32, root_iocg->hweight_donating, WEIGHT_ONE - 1));
blk-iocost: implement Andy's method for donation weight updates iocost implements work conservation by reducing iocg->inuse and propagating the adjustment upwards proportionally. However, while I knew the target absolute hierarchical proportion - adjusted hweight_inuse, I couldn't figure out how to determine the iocg->inuse adjustment to achieve that and approximated the adjustment by scaling iocg->inuse using the proportion of the needed hweight_inuse changes. When nested, these scalings aren't accurate even when adjusting a single node as the donating node also receives the benefit of the donated portion. When multiple nodes are donating as they often do, they can be wildly wrong. iocost employed various safety nets to combat the inaccuracies. There are ample buffers in determining how much to donate, the adjustments are conservative and gradual. While it can achieve a reasonable level of work conservation in simple scenarios, the inaccuracies can easily add up leading to significant loss of total work. This in turn makes it difficult to closely cap vrate as vrate adjustment is needed to compensate for the loss of work. The combination of inaccurate donation calculations and vrate adjustments can lead to wide fluctuations and clunky overall behaviors. Andy Newell devised a method to calculate the needed ->inuse updates to achieve the target hweight_inuse's. The method is compatible with the proportional inuse adjustment propagation which allows all hot path operations to be local to each iocg. To roughly summarize, Andy's method divides the tree into donating and non-donating parts, calculates global donation rate which is used to determine the target hweight_inuse for each node, and then derives per-level proportions. There's non-trivial amount of math involved. Please refer to the following pdfs for detailed descriptions. https://drive.google.com/file/d/1PsJwxPFtjUnwOY1QJ5AeICCcsL7BM3bo https://drive.google.com/file/d/1vONz1-fzVO7oY5DXXsLjSxEtYYQbOvsE https://drive.google.com/file/d/1WcrltBOSPN0qXVdBgnKm4mdp9FhuEFQN This patch implements Andy's method in transfer_surpluses(). This makes the donation calculations accurate per cycle and enables further improvements in other parts of the donation logic. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2020-09-01 18:52:48 +00:00
/*
* Calculate adjusted hwi, child_adjusted_sum and inuse for the inner
* nodes.
*/
list_for_each_entry(iocg, &inner_walk, walk_list) {
struct ioc_gq *parent;
u32 inuse, wpt, wptp;
u64 st, sf;
if (iocg->level == 0) {
/* adjusted weight sum for 1st level: s' = s * b_pf / b'_pf */
iocg->child_adjusted_sum = DIV64_U64_ROUND_UP(
iocg->child_active_sum * (WEIGHT_ONE - iocg->hweight_donating),
WEIGHT_ONE - iocg->hweight_after_donation);
continue;
}
parent = iocg->ancestors[iocg->level - 1];
/* b' = gamma * b_f + b_t' */
iocg->hweight_inuse = DIV64_U64_ROUND_UP(
(u64)gamma * (iocg->hweight_active - iocg->hweight_donating),
WEIGHT_ONE) + iocg->hweight_after_donation;
/* w' = s' * b' / b'_p */
inuse = DIV64_U64_ROUND_UP(
(u64)parent->child_adjusted_sum * iocg->hweight_inuse,
parent->hweight_inuse);
/* adjusted weight sum for children: s' = s_f + s_t * w'_pt / w_pt */
st = DIV64_U64_ROUND_UP(
iocg->child_active_sum * iocg->hweight_donating,
iocg->hweight_active);
sf = iocg->child_active_sum - st;
wpt = DIV64_U64_ROUND_UP(
(u64)iocg->active * iocg->hweight_donating,
iocg->hweight_active);
wptp = DIV64_U64_ROUND_UP(
(u64)inuse * iocg->hweight_after_donation,
iocg->hweight_inuse);
iocg->child_adjusted_sum = sf + DIV64_U64_ROUND_UP(st * wptp, wpt);
}
/*
* All inner nodes now have ->hweight_inuse and ->child_adjusted_sum and
* we can finally determine leaf adjustments.
*/
list_for_each_entry(iocg, surpluses, surplus_list) {
struct ioc_gq *parent = iocg->ancestors[iocg->level - 1];
u32 inuse;
/*
* In-debt iocgs participated in the donation calculation with
* the minimum target hweight_inuse. Configuring inuse
* accordingly would work fine but debt handling expects
* @iocg->inuse stay at the minimum and we don't wanna
* interfere.
*/
if (iocg->abs_vdebt) {
WARN_ON_ONCE(iocg->inuse > 1);
continue;
}
blk-iocost: implement Andy's method for donation weight updates iocost implements work conservation by reducing iocg->inuse and propagating the adjustment upwards proportionally. However, while I knew the target absolute hierarchical proportion - adjusted hweight_inuse, I couldn't figure out how to determine the iocg->inuse adjustment to achieve that and approximated the adjustment by scaling iocg->inuse using the proportion of the needed hweight_inuse changes. When nested, these scalings aren't accurate even when adjusting a single node as the donating node also receives the benefit of the donated portion. When multiple nodes are donating as they often do, they can be wildly wrong. iocost employed various safety nets to combat the inaccuracies. There are ample buffers in determining how much to donate, the adjustments are conservative and gradual. While it can achieve a reasonable level of work conservation in simple scenarios, the inaccuracies can easily add up leading to significant loss of total work. This in turn makes it difficult to closely cap vrate as vrate adjustment is needed to compensate for the loss of work. The combination of inaccurate donation calculations and vrate adjustments can lead to wide fluctuations and clunky overall behaviors. Andy Newell devised a method to calculate the needed ->inuse updates to achieve the target hweight_inuse's. The method is compatible with the proportional inuse adjustment propagation which allows all hot path operations to be local to each iocg. To roughly summarize, Andy's method divides the tree into donating and non-donating parts, calculates global donation rate which is used to determine the target hweight_inuse for each node, and then derives per-level proportions. There's non-trivial amount of math involved. Please refer to the following pdfs for detailed descriptions. https://drive.google.com/file/d/1PsJwxPFtjUnwOY1QJ5AeICCcsL7BM3bo https://drive.google.com/file/d/1vONz1-fzVO7oY5DXXsLjSxEtYYQbOvsE https://drive.google.com/file/d/1WcrltBOSPN0qXVdBgnKm4mdp9FhuEFQN This patch implements Andy's method in transfer_surpluses(). This makes the donation calculations accurate per cycle and enables further improvements in other parts of the donation logic. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2020-09-01 18:52:48 +00:00
/* w' = s' * b' / b'_p, note that b' == b'_t for donating leaves */
inuse = DIV64_U64_ROUND_UP(
parent->child_adjusted_sum * iocg->hweight_after_donation,
parent->hweight_inuse);
TRACE_IOCG_PATH(inuse_transfer, iocg, now,
iocg->inuse, inuse,
iocg->hweight_inuse,
iocg->hweight_after_donation);
__propagate_weights(iocg, iocg->active, inuse, true, now);
blk-iocost: implement Andy's method for donation weight updates iocost implements work conservation by reducing iocg->inuse and propagating the adjustment upwards proportionally. However, while I knew the target absolute hierarchical proportion - adjusted hweight_inuse, I couldn't figure out how to determine the iocg->inuse adjustment to achieve that and approximated the adjustment by scaling iocg->inuse using the proportion of the needed hweight_inuse changes. When nested, these scalings aren't accurate even when adjusting a single node as the donating node also receives the benefit of the donated portion. When multiple nodes are donating as they often do, they can be wildly wrong. iocost employed various safety nets to combat the inaccuracies. There are ample buffers in determining how much to donate, the adjustments are conservative and gradual. While it can achieve a reasonable level of work conservation in simple scenarios, the inaccuracies can easily add up leading to significant loss of total work. This in turn makes it difficult to closely cap vrate as vrate adjustment is needed to compensate for the loss of work. The combination of inaccurate donation calculations and vrate adjustments can lead to wide fluctuations and clunky overall behaviors. Andy Newell devised a method to calculate the needed ->inuse updates to achieve the target hweight_inuse's. The method is compatible with the proportional inuse adjustment propagation which allows all hot path operations to be local to each iocg. To roughly summarize, Andy's method divides the tree into donating and non-donating parts, calculates global donation rate which is used to determine the target hweight_inuse for each node, and then derives per-level proportions. There's non-trivial amount of math involved. Please refer to the following pdfs for detailed descriptions. https://drive.google.com/file/d/1PsJwxPFtjUnwOY1QJ5AeICCcsL7BM3bo https://drive.google.com/file/d/1vONz1-fzVO7oY5DXXsLjSxEtYYQbOvsE https://drive.google.com/file/d/1WcrltBOSPN0qXVdBgnKm4mdp9FhuEFQN This patch implements Andy's method in transfer_surpluses(). This makes the donation calculations accurate per cycle and enables further improvements in other parts of the donation logic. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2020-09-01 18:52:48 +00:00
}
/* walk list should be dissolved after use */
list_for_each_entry_safe(iocg, tiocg, &inner_walk, walk_list)
list_del_init(&iocg->walk_list);
}
/*
* A low weight iocg can amass a large amount of debt, for example, when
* anonymous memory gets reclaimed aggressively. If the system has a lot of
* memory paired with a slow IO device, the debt can span multiple seconds or
* more. If there are no other subsequent IO issuers, the in-debt iocg may end
* up blocked paying its debt while the IO device is idle.
*
* The following protects against such cases. If the device has been
* sufficiently idle for a while, the debts are halved and delays are
* recalculated.
*/
static void ioc_forgive_debts(struct ioc *ioc, u64 usage_us_sum, int nr_debtors,
struct ioc_now *now)
{
struct ioc_gq *iocg;
u64 dur, usage_pct, nr_cycles;
/* if no debtor, reset the cycle */
if (!nr_debtors) {
ioc->dfgv_period_at = now->now;
ioc->dfgv_period_rem = 0;
ioc->dfgv_usage_us_sum = 0;
return;
}
/*
* Debtors can pass through a lot of writes choking the device and we
* don't want to be forgiving debts while the device is struggling from
* write bursts. If we're missing latency targets, consider the device
* fully utilized.
*/
if (ioc->busy_level > 0)
usage_us_sum = max_t(u64, usage_us_sum, ioc->period_us);
ioc->dfgv_usage_us_sum += usage_us_sum;
if (time_before64(now->now, ioc->dfgv_period_at + DFGV_PERIOD))
return;
/*
* At least DFGV_PERIOD has passed since the last period. Calculate the
* average usage and reset the period counters.
*/
dur = now->now - ioc->dfgv_period_at;
usage_pct = div64_u64(100 * ioc->dfgv_usage_us_sum, dur);
ioc->dfgv_period_at = now->now;
ioc->dfgv_usage_us_sum = 0;
/* if was too busy, reset everything */
if (usage_pct > DFGV_USAGE_PCT) {
ioc->dfgv_period_rem = 0;
return;
}
/*
* Usage is lower than threshold. Let's forgive some debts. Debt
* forgiveness runs off of the usual ioc timer but its period usually
* doesn't match ioc's. Compensate the difference by performing the
* reduction as many times as would fit in the duration since the last
* run and carrying over the left-over duration in @ioc->dfgv_period_rem
* - if ioc period is 75% of DFGV_PERIOD, one out of three consecutive
* reductions is doubled.
*/
nr_cycles = dur + ioc->dfgv_period_rem;
ioc->dfgv_period_rem = do_div(nr_cycles, DFGV_PERIOD);
list_for_each_entry(iocg, &ioc->active_iocgs, active_list) {
u64 __maybe_unused old_debt, __maybe_unused old_delay;
if (!iocg->abs_vdebt && !iocg->delay)
continue;
spin_lock(&iocg->waitq.lock);
old_debt = iocg->abs_vdebt;
old_delay = iocg->delay;
if (iocg->abs_vdebt)
iocg->abs_vdebt = iocg->abs_vdebt >> nr_cycles ?: 1;
if (iocg->delay)
iocg->delay = iocg->delay >> nr_cycles ?: 1;
iocg_kick_waitq(iocg, true, now);
TRACE_IOCG_PATH(iocg_forgive_debt, iocg, now, usage_pct,
old_debt, iocg->abs_vdebt,
old_delay, iocg->delay);
spin_unlock(&iocg->waitq.lock);
}
}
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
static void ioc_timer_fn(struct timer_list *timer)
{
struct ioc *ioc = container_of(timer, struct ioc, timer);
struct ioc_gq *iocg, *tiocg;
struct ioc_now now;
LIST_HEAD(surpluses);
int nr_debtors = 0, nr_shortages = 0, nr_lagging = 0;
u64 usage_us_sum = 0;
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
u32 ppm_rthr = MILLION - ioc->params.qos[QOS_RPPM];
u32 ppm_wthr = MILLION - ioc->params.qos[QOS_WPPM];
u32 missed_ppm[2], rq_wait_pct;
u64 period_vtime;
blk-iocost: revamp donation amount determination iocost has various safety nets to combat inuse adjustment calculation inaccuracies. With Andy's method implemented in transfer_surpluses(), inuse adjustment calculations are now accurate and we can make donation amount determinations accurate too. * Stop keeping track of past usage history and using the maximum. Act on the immediate usage information. * Remove donation constraints defined by SURPLUS_* constants. Donate whatever isn't used. * Determine the donation amount so that the iocg will end up with MARGIN_TARGET_PCT budget at the end of the coming period assuming the same usage as the previous period. TARGET is set at 50% of period, which is the previous maximum. This provides smooth convergence for most repetitive IO patterns. * Apply donation logic early at 20% budget. There's no risk in doing so as the calculation is based on the delta between the current budget and the target budget at the end of the coming period. * Remove preemptive iocg activation for zero cost IOs. As donation can reach near zero now, the mere activation doesn't provide any protection anymore. In the unlikely case that this becomes a problem, the right solution is assigning appropriate costs for such IOs. This significantly improves the donation determination logic while also simplifying it. Now all donations are immediate, exact and smooth. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2020-09-01 18:52:49 +00:00
int prev_busy_level;
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
/* how were the latencies during the period? */
ioc_lat_stat(ioc, missed_ppm, &rq_wait_pct);
/* take care of active iocgs */
spin_lock_irq(&ioc->lock);
ioc_now(ioc, &now);
period_vtime = now.vnow - ioc->period_at_vtime;
if (WARN_ON_ONCE(!period_vtime)) {
spin_unlock_irq(&ioc->lock);
return;
}
/*
* Waiters determine the sleep durations based on the vrate they
* saw at the time of sleep. If vrate has increased, some waiters
* could be sleeping for too long. Wake up tardy waiters which
* should have woken up in the last period and expire idle iocgs.
*/
list_for_each_entry_safe(iocg, tiocg, &ioc->active_iocgs, active_list) {
if (!waitqueue_active(&iocg->waitq) && !iocg->abs_vdebt &&
!iocg->delay && !iocg_is_idle(iocg))
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
continue;
spin_lock(&iocg->waitq.lock);
/* flush wait and indebt stat deltas */
if (iocg->wait_since) {
iocg->local_stat.wait_us += now.now - iocg->wait_since;
iocg->wait_since = now.now;
}
if (iocg->indebt_since) {
iocg->local_stat.indebt_us +=
now.now - iocg->indebt_since;
iocg->indebt_since = now.now;
}
if (iocg->indelay_since) {
iocg->local_stat.indelay_us +=
now.now - iocg->indelay_since;
iocg->indelay_since = now.now;
}
if (waitqueue_active(&iocg->waitq) || iocg->abs_vdebt ||
iocg->delay) {
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
/* might be oversleeping vtime / hweight changes, kick */
iocg_kick_waitq(iocg, true, &now);
if (iocg->abs_vdebt || iocg->delay)
nr_debtors++;
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
} else if (iocg_is_idle(iocg)) {
/* no waiter and idle, deactivate */
u64 vtime = atomic64_read(&iocg->vtime);
s64 excess;
/*
* @iocg has been inactive for a full duration and will
* have a high budget. Account anything above target as
* error and throw away. On reactivation, it'll start
* with the target budget.
*/
excess = now.vnow - vtime - ioc->margins.target;
if (excess > 0) {
u32 old_hwi;
current_hweight(iocg, NULL, &old_hwi);
ioc->vtime_err -= div64_u64(excess * old_hwi,
WEIGHT_ONE);
}
__propagate_weights(iocg, 0, 0, false, &now);
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
list_del_init(&iocg->active_list);
}
spin_unlock(&iocg->waitq.lock);
}
commit_weights(ioc);
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
/*
* Wait and indebt stat are flushed above and the donation calculation
* below needs updated usage stat. Let's bring stat up-to-date.
*/
iocg_flush_stat(&ioc->active_iocgs, &now);
blk-iocost: revamp donation amount determination iocost has various safety nets to combat inuse adjustment calculation inaccuracies. With Andy's method implemented in transfer_surpluses(), inuse adjustment calculations are now accurate and we can make donation amount determinations accurate too. * Stop keeping track of past usage history and using the maximum. Act on the immediate usage information. * Remove donation constraints defined by SURPLUS_* constants. Donate whatever isn't used. * Determine the donation amount so that the iocg will end up with MARGIN_TARGET_PCT budget at the end of the coming period assuming the same usage as the previous period. TARGET is set at 50% of period, which is the previous maximum. This provides smooth convergence for most repetitive IO patterns. * Apply donation logic early at 20% budget. There's no risk in doing so as the calculation is based on the delta between the current budget and the target budget at the end of the coming period. * Remove preemptive iocg activation for zero cost IOs. As donation can reach near zero now, the mere activation doesn't provide any protection anymore. In the unlikely case that this becomes a problem, the right solution is assigning appropriate costs for such IOs. This significantly improves the donation determination logic while also simplifying it. Now all donations are immediate, exact and smooth. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2020-09-01 18:52:49 +00:00
/* calc usage and see whether some weights need to be moved around */
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
list_for_each_entry(iocg, &ioc->active_iocgs, active_list) {
blk-iocost: revamp donation amount determination iocost has various safety nets to combat inuse adjustment calculation inaccuracies. With Andy's method implemented in transfer_surpluses(), inuse adjustment calculations are now accurate and we can make donation amount determinations accurate too. * Stop keeping track of past usage history and using the maximum. Act on the immediate usage information. * Remove donation constraints defined by SURPLUS_* constants. Donate whatever isn't used. * Determine the donation amount so that the iocg will end up with MARGIN_TARGET_PCT budget at the end of the coming period assuming the same usage as the previous period. TARGET is set at 50% of period, which is the previous maximum. This provides smooth convergence for most repetitive IO patterns. * Apply donation logic early at 20% budget. There's no risk in doing so as the calculation is based on the delta between the current budget and the target budget at the end of the coming period. * Remove preemptive iocg activation for zero cost IOs. As donation can reach near zero now, the mere activation doesn't provide any protection anymore. In the unlikely case that this becomes a problem, the right solution is assigning appropriate costs for such IOs. This significantly improves the donation determination logic while also simplifying it. Now all donations are immediate, exact and smooth. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2020-09-01 18:52:49 +00:00
u64 vdone, vtime, usage_us, usage_dur;
u32 usage, hw_active, hw_inuse;
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
/*
* Collect unused and wind vtime closer to vnow to prevent
* iocgs from accumulating a large amount of budget.
*/
vdone = atomic64_read(&iocg->done_vtime);
vtime = atomic64_read(&iocg->vtime);
current_hweight(iocg, &hw_active, &hw_inuse);
/*
* Latency QoS detection doesn't account for IOs which are
* in-flight for longer than a period. Detect them by
* comparing vdone against period start. If lagging behind
* IOs from past periods, don't increase vrate.
*/
if ((ppm_rthr != MILLION || ppm_wthr != MILLION) &&
!atomic_read(&iocg_to_blkg(iocg)->use_delay) &&
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
time_after64(vtime, vdone) &&
time_after64(vtime, now.vnow -
MAX_LAGGING_PERIODS * period_vtime) &&
time_before64(vdone, now.vnow - period_vtime))
nr_lagging++;
/*
blk-iocost: revamp donation amount determination iocost has various safety nets to combat inuse adjustment calculation inaccuracies. With Andy's method implemented in transfer_surpluses(), inuse adjustment calculations are now accurate and we can make donation amount determinations accurate too. * Stop keeping track of past usage history and using the maximum. Act on the immediate usage information. * Remove donation constraints defined by SURPLUS_* constants. Donate whatever isn't used. * Determine the donation amount so that the iocg will end up with MARGIN_TARGET_PCT budget at the end of the coming period assuming the same usage as the previous period. TARGET is set at 50% of period, which is the previous maximum. This provides smooth convergence for most repetitive IO patterns. * Apply donation logic early at 20% budget. There's no risk in doing so as the calculation is based on the delta between the current budget and the target budget at the end of the coming period. * Remove preemptive iocg activation for zero cost IOs. As donation can reach near zero now, the mere activation doesn't provide any protection anymore. In the unlikely case that this becomes a problem, the right solution is assigning appropriate costs for such IOs. This significantly improves the donation determination logic while also simplifying it. Now all donations are immediate, exact and smooth. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2020-09-01 18:52:49 +00:00
* Determine absolute usage factoring in in-flight IOs to avoid
* high-latency completions appearing as idle.
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
*/
usage_us = iocg->usage_delta_us;
usage_us_sum += usage_us;
blk-iocost: revamp donation amount determination iocost has various safety nets to combat inuse adjustment calculation inaccuracies. With Andy's method implemented in transfer_surpluses(), inuse adjustment calculations are now accurate and we can make donation amount determinations accurate too. * Stop keeping track of past usage history and using the maximum. Act on the immediate usage information. * Remove donation constraints defined by SURPLUS_* constants. Donate whatever isn't used. * Determine the donation amount so that the iocg will end up with MARGIN_TARGET_PCT budget at the end of the coming period assuming the same usage as the previous period. TARGET is set at 50% of period, which is the previous maximum. This provides smooth convergence for most repetitive IO patterns. * Apply donation logic early at 20% budget. There's no risk in doing so as the calculation is based on the delta between the current budget and the target budget at the end of the coming period. * Remove preemptive iocg activation for zero cost IOs. As donation can reach near zero now, the mere activation doesn't provide any protection anymore. In the unlikely case that this becomes a problem, the right solution is assigning appropriate costs for such IOs. This significantly improves the donation determination logic while also simplifying it. Now all donations are immediate, exact and smooth. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2020-09-01 18:52:49 +00:00
if (vdone != vtime) {
u64 inflight_us = DIV64_U64_ROUND_UP(
cost_to_abs_cost(vtime - vdone, hw_inuse),
ioc->vtime_base_rate);
usage_us = max(usage_us, inflight_us);
}
blk-iocost: revamp donation amount determination iocost has various safety nets to combat inuse adjustment calculation inaccuracies. With Andy's method implemented in transfer_surpluses(), inuse adjustment calculations are now accurate and we can make donation amount determinations accurate too. * Stop keeping track of past usage history and using the maximum. Act on the immediate usage information. * Remove donation constraints defined by SURPLUS_* constants. Donate whatever isn't used. * Determine the donation amount so that the iocg will end up with MARGIN_TARGET_PCT budget at the end of the coming period assuming the same usage as the previous period. TARGET is set at 50% of period, which is the previous maximum. This provides smooth convergence for most repetitive IO patterns. * Apply donation logic early at 20% budget. There's no risk in doing so as the calculation is based on the delta between the current budget and the target budget at the end of the coming period. * Remove preemptive iocg activation for zero cost IOs. As donation can reach near zero now, the mere activation doesn't provide any protection anymore. In the unlikely case that this becomes a problem, the right solution is assigning appropriate costs for such IOs. This significantly improves the donation determination logic while also simplifying it. Now all donations are immediate, exact and smooth. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2020-09-01 18:52:49 +00:00
/* convert to hweight based usage ratio */
if (time_after64(iocg->activated_at, ioc->period_at))
usage_dur = max_t(u64, now.now - iocg->activated_at, 1);
else
usage_dur = max_t(u64, now.now - ioc->period_at, 1);
blk-iocost: revamp donation amount determination iocost has various safety nets to combat inuse adjustment calculation inaccuracies. With Andy's method implemented in transfer_surpluses(), inuse adjustment calculations are now accurate and we can make donation amount determinations accurate too. * Stop keeping track of past usage history and using the maximum. Act on the immediate usage information. * Remove donation constraints defined by SURPLUS_* constants. Donate whatever isn't used. * Determine the donation amount so that the iocg will end up with MARGIN_TARGET_PCT budget at the end of the coming period assuming the same usage as the previous period. TARGET is set at 50% of period, which is the previous maximum. This provides smooth convergence for most repetitive IO patterns. * Apply donation logic early at 20% budget. There's no risk in doing so as the calculation is based on the delta between the current budget and the target budget at the end of the coming period. * Remove preemptive iocg activation for zero cost IOs. As donation can reach near zero now, the mere activation doesn't provide any protection anymore. In the unlikely case that this becomes a problem, the right solution is assigning appropriate costs for such IOs. This significantly improves the donation determination logic while also simplifying it. Now all donations are immediate, exact and smooth. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2020-09-01 18:52:49 +00:00
usage = clamp_t(u32,
DIV64_U64_ROUND_UP(usage_us * WEIGHT_ONE,
usage_dur),
1, WEIGHT_ONE);
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
/* see whether there's surplus vtime */
WARN_ON_ONCE(!list_empty(&iocg->surplus_list));
if (hw_inuse < hw_active ||
(!waitqueue_active(&iocg->waitq) &&
blk-iocost: revamp donation amount determination iocost has various safety nets to combat inuse adjustment calculation inaccuracies. With Andy's method implemented in transfer_surpluses(), inuse adjustment calculations are now accurate and we can make donation amount determinations accurate too. * Stop keeping track of past usage history and using the maximum. Act on the immediate usage information. * Remove donation constraints defined by SURPLUS_* constants. Donate whatever isn't used. * Determine the donation amount so that the iocg will end up with MARGIN_TARGET_PCT budget at the end of the coming period assuming the same usage as the previous period. TARGET is set at 50% of period, which is the previous maximum. This provides smooth convergence for most repetitive IO patterns. * Apply donation logic early at 20% budget. There's no risk in doing so as the calculation is based on the delta between the current budget and the target budget at the end of the coming period. * Remove preemptive iocg activation for zero cost IOs. As donation can reach near zero now, the mere activation doesn't provide any protection anymore. In the unlikely case that this becomes a problem, the right solution is assigning appropriate costs for such IOs. This significantly improves the donation determination logic while also simplifying it. Now all donations are immediate, exact and smooth. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2020-09-01 18:52:49 +00:00
time_before64(vtime, now.vnow - ioc->margins.low))) {
u32 hwa, old_hwi, hwm, new_hwi;
/*
* Already donating or accumulated enough to start.
* Determine the donation amount.
*/
current_hweight(iocg, &hwa, &old_hwi);
hwm = current_hweight_max(iocg);
new_hwi = hweight_after_donation(iocg, old_hwi, hwm,
usage, &now);
if (new_hwi < hwm) {
blk-iocost: implement Andy's method for donation weight updates iocost implements work conservation by reducing iocg->inuse and propagating the adjustment upwards proportionally. However, while I knew the target absolute hierarchical proportion - adjusted hweight_inuse, I couldn't figure out how to determine the iocg->inuse adjustment to achieve that and approximated the adjustment by scaling iocg->inuse using the proportion of the needed hweight_inuse changes. When nested, these scalings aren't accurate even when adjusting a single node as the donating node also receives the benefit of the donated portion. When multiple nodes are donating as they often do, they can be wildly wrong. iocost employed various safety nets to combat the inaccuracies. There are ample buffers in determining how much to donate, the adjustments are conservative and gradual. While it can achieve a reasonable level of work conservation in simple scenarios, the inaccuracies can easily add up leading to significant loss of total work. This in turn makes it difficult to closely cap vrate as vrate adjustment is needed to compensate for the loss of work. The combination of inaccurate donation calculations and vrate adjustments can lead to wide fluctuations and clunky overall behaviors. Andy Newell devised a method to calculate the needed ->inuse updates to achieve the target hweight_inuse's. The method is compatible with the proportional inuse adjustment propagation which allows all hot path operations to be local to each iocg. To roughly summarize, Andy's method divides the tree into donating and non-donating parts, calculates global donation rate which is used to determine the target hweight_inuse for each node, and then derives per-level proportions. There's non-trivial amount of math involved. Please refer to the following pdfs for detailed descriptions. https://drive.google.com/file/d/1PsJwxPFtjUnwOY1QJ5AeICCcsL7BM3bo https://drive.google.com/file/d/1vONz1-fzVO7oY5DXXsLjSxEtYYQbOvsE https://drive.google.com/file/d/1WcrltBOSPN0qXVdBgnKm4mdp9FhuEFQN This patch implements Andy's method in transfer_surpluses(). This makes the donation calculations accurate per cycle and enables further improvements in other parts of the donation logic. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2020-09-01 18:52:48 +00:00
iocg->hweight_donating = hwa;
iocg->hweight_after_donation = new_hwi;
list_add(&iocg->surplus_list, &surpluses);
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
} else {
TRACE_IOCG_PATH(inuse_shortage, iocg, &now,
iocg->inuse, iocg->active,
iocg->hweight_inuse, new_hwi);
__propagate_weights(iocg, iocg->active,
iocg->active, true, &now);
nr_shortages++;
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
}
} else {
/* genuinely short on vtime */
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
nr_shortages++;
}
}
if (!list_empty(&surpluses) && nr_shortages)
transfer_surpluses(&surpluses, &now);
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
commit_weights(ioc);
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
/* surplus list should be dissolved after use */
list_for_each_entry_safe(iocg, tiocg, &surpluses, surplus_list)
list_del_init(&iocg->surplus_list);
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
/*
* If q is getting clogged or we're missing too much, we're issuing
* too much IO and should lower vtime rate. If we're not missing
* and experiencing shortages but not surpluses, we're too stingy
* and should increase vtime rate.
*/
prev_busy_level = ioc->busy_level;
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
if (rq_wait_pct > RQ_WAIT_BUSY_PCT ||
missed_ppm[READ] > ppm_rthr ||
missed_ppm[WRITE] > ppm_wthr) {
iocost: don't let vrate run wild while there's no saturation signal When the QoS targets are met and nothing is being throttled, there's no way to tell how saturated the underlying device is - it could be almost entirely idle, at the cusp of saturation or anywhere inbetween. Given that there's no information, it's best to keep vrate as-is in this state. Before 7cd806a9a953 ("iocost: improve nr_lagging handling"), this was the case - if the device isn't missing QoS targets and nothing is being throttled, busy_level was reset to zero. While fixing nr_lagging handling, 7cd806a9a953 ("iocost: improve nr_lagging handling") broke this. Now, while the device is hitting QoS targets and nothing is being throttled, vrate keeps getting adjusted according to the existing busy_level. This led to vrate keeping climing till it hits max when there's an IO issuer with limited request concurrency if the vrate started low. vrate starts getting adjusted upwards until the issuer can issue IOs w/o being throttled. From then on, QoS targets keeps getting met and nothing on the system needs throttling and vrate keeps getting increased due to the existing busy_level. This patch makes the following changes to the busy_level logic. * Reset busy_level if nr_shortages is zero to avoid the above scenario. * Make non-zero nr_lagging block lowering nr_level but still clear positive busy_level if there's clear non-saturation signal - QoS targets are met and nr_shortages is non-zero. nr_lagging's role is preventing adjusting vrate upwards while there are long-running commands and it shouldn't keep busy_level positive while there's clear non-saturation signal. * Restructure code for clarity and add comments. Signed-off-by: Tejun Heo <tj@kernel.org> Reported-by: Andy Newell <newella@fb.com> Fixes: 7cd806a9a953 ("iocost: improve nr_lagging handling") Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-10-15 00:18:11 +00:00
/* clearly missing QoS targets, slow down vrate */
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
ioc->busy_level = max(ioc->busy_level, 0);
ioc->busy_level++;
} else if (rq_wait_pct <= RQ_WAIT_BUSY_PCT * UNBUSY_THR_PCT / 100 &&
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
missed_ppm[READ] <= ppm_rthr * UNBUSY_THR_PCT / 100 &&
missed_ppm[WRITE] <= ppm_wthr * UNBUSY_THR_PCT / 100) {
iocost: don't let vrate run wild while there's no saturation signal When the QoS targets are met and nothing is being throttled, there's no way to tell how saturated the underlying device is - it could be almost entirely idle, at the cusp of saturation or anywhere inbetween. Given that there's no information, it's best to keep vrate as-is in this state. Before 7cd806a9a953 ("iocost: improve nr_lagging handling"), this was the case - if the device isn't missing QoS targets and nothing is being throttled, busy_level was reset to zero. While fixing nr_lagging handling, 7cd806a9a953 ("iocost: improve nr_lagging handling") broke this. Now, while the device is hitting QoS targets and nothing is being throttled, vrate keeps getting adjusted according to the existing busy_level. This led to vrate keeping climing till it hits max when there's an IO issuer with limited request concurrency if the vrate started low. vrate starts getting adjusted upwards until the issuer can issue IOs w/o being throttled. From then on, QoS targets keeps getting met and nothing on the system needs throttling and vrate keeps getting increased due to the existing busy_level. This patch makes the following changes to the busy_level logic. * Reset busy_level if nr_shortages is zero to avoid the above scenario. * Make non-zero nr_lagging block lowering nr_level but still clear positive busy_level if there's clear non-saturation signal - QoS targets are met and nr_shortages is non-zero. nr_lagging's role is preventing adjusting vrate upwards while there are long-running commands and it shouldn't keep busy_level positive while there's clear non-saturation signal. * Restructure code for clarity and add comments. Signed-off-by: Tejun Heo <tj@kernel.org> Reported-by: Andy Newell <newella@fb.com> Fixes: 7cd806a9a953 ("iocost: improve nr_lagging handling") Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-10-15 00:18:11 +00:00
/* QoS targets are being met with >25% margin */
if (nr_shortages) {
/*
* We're throttling while the device has spare
* capacity. If vrate was being slowed down, stop.
*/
ioc->busy_level = min(ioc->busy_level, 0);
iocost: don't let vrate run wild while there's no saturation signal When the QoS targets are met and nothing is being throttled, there's no way to tell how saturated the underlying device is - it could be almost entirely idle, at the cusp of saturation or anywhere inbetween. Given that there's no information, it's best to keep vrate as-is in this state. Before 7cd806a9a953 ("iocost: improve nr_lagging handling"), this was the case - if the device isn't missing QoS targets and nothing is being throttled, busy_level was reset to zero. While fixing nr_lagging handling, 7cd806a9a953 ("iocost: improve nr_lagging handling") broke this. Now, while the device is hitting QoS targets and nothing is being throttled, vrate keeps getting adjusted according to the existing busy_level. This led to vrate keeping climing till it hits max when there's an IO issuer with limited request concurrency if the vrate started low. vrate starts getting adjusted upwards until the issuer can issue IOs w/o being throttled. From then on, QoS targets keeps getting met and nothing on the system needs throttling and vrate keeps getting increased due to the existing busy_level. This patch makes the following changes to the busy_level logic. * Reset busy_level if nr_shortages is zero to avoid the above scenario. * Make non-zero nr_lagging block lowering nr_level but still clear positive busy_level if there's clear non-saturation signal - QoS targets are met and nr_shortages is non-zero. nr_lagging's role is preventing adjusting vrate upwards while there are long-running commands and it shouldn't keep busy_level positive while there's clear non-saturation signal. * Restructure code for clarity and add comments. Signed-off-by: Tejun Heo <tj@kernel.org> Reported-by: Andy Newell <newella@fb.com> Fixes: 7cd806a9a953 ("iocost: improve nr_lagging handling") Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-10-15 00:18:11 +00:00
/*
* If there are IOs spanning multiple periods, wait
* them out before pushing the device harder.
iocost: don't let vrate run wild while there's no saturation signal When the QoS targets are met and nothing is being throttled, there's no way to tell how saturated the underlying device is - it could be almost entirely idle, at the cusp of saturation or anywhere inbetween. Given that there's no information, it's best to keep vrate as-is in this state. Before 7cd806a9a953 ("iocost: improve nr_lagging handling"), this was the case - if the device isn't missing QoS targets and nothing is being throttled, busy_level was reset to zero. While fixing nr_lagging handling, 7cd806a9a953 ("iocost: improve nr_lagging handling") broke this. Now, while the device is hitting QoS targets and nothing is being throttled, vrate keeps getting adjusted according to the existing busy_level. This led to vrate keeping climing till it hits max when there's an IO issuer with limited request concurrency if the vrate started low. vrate starts getting adjusted upwards until the issuer can issue IOs w/o being throttled. From then on, QoS targets keeps getting met and nothing on the system needs throttling and vrate keeps getting increased due to the existing busy_level. This patch makes the following changes to the busy_level logic. * Reset busy_level if nr_shortages is zero to avoid the above scenario. * Make non-zero nr_lagging block lowering nr_level but still clear positive busy_level if there's clear non-saturation signal - QoS targets are met and nr_shortages is non-zero. nr_lagging's role is preventing adjusting vrate upwards while there are long-running commands and it shouldn't keep busy_level positive while there's clear non-saturation signal. * Restructure code for clarity and add comments. Signed-off-by: Tejun Heo <tj@kernel.org> Reported-by: Andy Newell <newella@fb.com> Fixes: 7cd806a9a953 ("iocost: improve nr_lagging handling") Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-10-15 00:18:11 +00:00
*/
if (!nr_lagging)
ioc->busy_level--;
iocost: don't let vrate run wild while there's no saturation signal When the QoS targets are met and nothing is being throttled, there's no way to tell how saturated the underlying device is - it could be almost entirely idle, at the cusp of saturation or anywhere inbetween. Given that there's no information, it's best to keep vrate as-is in this state. Before 7cd806a9a953 ("iocost: improve nr_lagging handling"), this was the case - if the device isn't missing QoS targets and nothing is being throttled, busy_level was reset to zero. While fixing nr_lagging handling, 7cd806a9a953 ("iocost: improve nr_lagging handling") broke this. Now, while the device is hitting QoS targets and nothing is being throttled, vrate keeps getting adjusted according to the existing busy_level. This led to vrate keeping climing till it hits max when there's an IO issuer with limited request concurrency if the vrate started low. vrate starts getting adjusted upwards until the issuer can issue IOs w/o being throttled. From then on, QoS targets keeps getting met and nothing on the system needs throttling and vrate keeps getting increased due to the existing busy_level. This patch makes the following changes to the busy_level logic. * Reset busy_level if nr_shortages is zero to avoid the above scenario. * Make non-zero nr_lagging block lowering nr_level but still clear positive busy_level if there's clear non-saturation signal - QoS targets are met and nr_shortages is non-zero. nr_lagging's role is preventing adjusting vrate upwards while there are long-running commands and it shouldn't keep busy_level positive while there's clear non-saturation signal. * Restructure code for clarity and add comments. Signed-off-by: Tejun Heo <tj@kernel.org> Reported-by: Andy Newell <newella@fb.com> Fixes: 7cd806a9a953 ("iocost: improve nr_lagging handling") Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-10-15 00:18:11 +00:00
} else {
/*
* Nobody is being throttled and the users aren't
* issuing enough IOs to saturate the device. We
* simply don't know how close the device is to
* saturation. Coast.
*/
ioc->busy_level = 0;
}
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
} else {
iocost: don't let vrate run wild while there's no saturation signal When the QoS targets are met and nothing is being throttled, there's no way to tell how saturated the underlying device is - it could be almost entirely idle, at the cusp of saturation or anywhere inbetween. Given that there's no information, it's best to keep vrate as-is in this state. Before 7cd806a9a953 ("iocost: improve nr_lagging handling"), this was the case - if the device isn't missing QoS targets and nothing is being throttled, busy_level was reset to zero. While fixing nr_lagging handling, 7cd806a9a953 ("iocost: improve nr_lagging handling") broke this. Now, while the device is hitting QoS targets and nothing is being throttled, vrate keeps getting adjusted according to the existing busy_level. This led to vrate keeping climing till it hits max when there's an IO issuer with limited request concurrency if the vrate started low. vrate starts getting adjusted upwards until the issuer can issue IOs w/o being throttled. From then on, QoS targets keeps getting met and nothing on the system needs throttling and vrate keeps getting increased due to the existing busy_level. This patch makes the following changes to the busy_level logic. * Reset busy_level if nr_shortages is zero to avoid the above scenario. * Make non-zero nr_lagging block lowering nr_level but still clear positive busy_level if there's clear non-saturation signal - QoS targets are met and nr_shortages is non-zero. nr_lagging's role is preventing adjusting vrate upwards while there are long-running commands and it shouldn't keep busy_level positive while there's clear non-saturation signal. * Restructure code for clarity and add comments. Signed-off-by: Tejun Heo <tj@kernel.org> Reported-by: Andy Newell <newella@fb.com> Fixes: 7cd806a9a953 ("iocost: improve nr_lagging handling") Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-10-15 00:18:11 +00:00
/* inside the hysterisis margin, we're good */
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
ioc->busy_level = 0;
}
ioc->busy_level = clamp(ioc->busy_level, -1000, 1000);
if (ioc->busy_level > 0 || (ioc->busy_level < 0 && !nr_lagging)) {
u64 vrate = ioc->vtime_base_rate;
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
u64 vrate_min = ioc->vrate_min, vrate_max = ioc->vrate_max;
/* rq_wait signal is always reliable, ignore user vrate_min */
if (rq_wait_pct > RQ_WAIT_BUSY_PCT)
vrate_min = VRATE_MIN;
/*
* If vrate is out of bounds, apply clamp gradually as the
* bounds can change abruptly. Otherwise, apply busy_level
* based adjustment.
*/
if (vrate < vrate_min) {
vrate = div64_u64(vrate * (100 + VRATE_CLAMP_ADJ_PCT),
100);
vrate = min(vrate, vrate_min);
} else if (vrate > vrate_max) {
vrate = div64_u64(vrate * (100 - VRATE_CLAMP_ADJ_PCT),
100);
vrate = max(vrate, vrate_max);
} else {
int idx = min_t(int, abs(ioc->busy_level),
ARRAY_SIZE(vrate_adj_pct) - 1);
u32 adj_pct = vrate_adj_pct[idx];
if (ioc->busy_level > 0)
adj_pct = 100 - adj_pct;
else
adj_pct = 100 + adj_pct;
vrate = clamp(DIV64_U64_ROUND_UP(vrate * adj_pct, 100),
vrate_min, vrate_max);
}
trace_iocost_ioc_vrate_adj(ioc, vrate, missed_ppm, rq_wait_pct,
nr_lagging, nr_shortages);
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
ioc->vtime_base_rate = vrate;
ioc_refresh_margins(ioc);
} else if (ioc->busy_level != prev_busy_level || nr_lagging) {
trace_iocost_ioc_vrate_adj(ioc, atomic64_read(&ioc->vtime_rate),
missed_ppm, rq_wait_pct, nr_lagging,
nr_shortages);
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
}
ioc_refresh_params(ioc, false);
ioc_forgive_debts(ioc, usage_us_sum, nr_debtors, &now);
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
/*
* This period is done. Move onto the next one. If nothing's
* going on with the device, stop the timer.
*/
atomic64_inc(&ioc->cur_period);
if (ioc->running != IOC_STOP) {
if (!list_empty(&ioc->active_iocgs)) {
ioc_start_period(ioc, &now);
} else {
ioc->busy_level = 0;
ioc->vtime_err = 0;
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
ioc->running = IOC_IDLE;
}
ioc_refresh_vrate(ioc, &now);
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
}
spin_unlock_irq(&ioc->lock);
}
static u64 adjust_inuse_and_calc_cost(struct ioc_gq *iocg, u64 vtime,
u64 abs_cost, struct ioc_now *now)
{
struct ioc *ioc = iocg->ioc;
struct ioc_margins *margins = &ioc->margins;
u32 __maybe_unused old_inuse = iocg->inuse, __maybe_unused old_hwi;
u32 hwi, adj_step;
s64 margin;
u64 cost, new_inuse;
current_hweight(iocg, NULL, &hwi);
old_hwi = hwi;
cost = abs_cost_to_cost(abs_cost, hwi);
margin = now->vnow - vtime - cost;
/* debt handling owns inuse for debtors */
if (iocg->abs_vdebt)
return cost;
/*
* We only increase inuse during period and do so iff the margin has
* deteriorated since the previous adjustment.
*/
if (margin >= iocg->saved_margin || margin >= margins->low ||
iocg->inuse == iocg->active)
return cost;
spin_lock_irq(&ioc->lock);
/* we own inuse only when @iocg is in the normal active state */
if (iocg->abs_vdebt || list_empty(&iocg->active_list)) {
spin_unlock_irq(&ioc->lock);
return cost;
}
/*
* Bump up inuse till @abs_cost fits in the existing budget.
* adj_step must be determined after acquiring ioc->lock - we might
* have raced and lost to another thread for activation and could
* be reading 0 iocg->active before ioc->lock which will lead to
* infinite loop.
*/
new_inuse = iocg->inuse;
adj_step = DIV_ROUND_UP(iocg->active * INUSE_ADJ_STEP_PCT, 100);
do {
new_inuse = new_inuse + adj_step;
propagate_weights(iocg, iocg->active, new_inuse, true, now);
current_hweight(iocg, NULL, &hwi);
cost = abs_cost_to_cost(abs_cost, hwi);
} while (time_after64(vtime + cost, now->vnow) &&
iocg->inuse != iocg->active);
spin_unlock_irq(&ioc->lock);
TRACE_IOCG_PATH(inuse_adjust, iocg, now,
old_inuse, iocg->inuse, old_hwi, hwi);
return cost;
}
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
static void calc_vtime_cost_builtin(struct bio *bio, struct ioc_gq *iocg,
bool is_merge, u64 *costp)
{
struct ioc *ioc = iocg->ioc;
u64 coef_seqio, coef_randio, coef_page;
u64 pages = max_t(u64, bio_sectors(bio) >> IOC_SECT_TO_PAGE_SHIFT, 1);
u64 seek_pages = 0;
u64 cost = 0;
switch (bio_op(bio)) {
case REQ_OP_READ:
coef_seqio = ioc->params.lcoefs[LCOEF_RSEQIO];
coef_randio = ioc->params.lcoefs[LCOEF_RRANDIO];
coef_page = ioc->params.lcoefs[LCOEF_RPAGE];
break;
case REQ_OP_WRITE:
coef_seqio = ioc->params.lcoefs[LCOEF_WSEQIO];
coef_randio = ioc->params.lcoefs[LCOEF_WRANDIO];
coef_page = ioc->params.lcoefs[LCOEF_WPAGE];
break;
default:
goto out;
}
if (iocg->cursor) {
seek_pages = abs(bio->bi_iter.bi_sector - iocg->cursor);
seek_pages >>= IOC_SECT_TO_PAGE_SHIFT;
}
if (!is_merge) {
if (seek_pages > LCOEF_RANDIO_PAGES) {
cost += coef_randio;
} else {
cost += coef_seqio;
}
}
cost += pages * coef_page;
out:
*costp = cost;
}
static u64 calc_vtime_cost(struct bio *bio, struct ioc_gq *iocg, bool is_merge)
{
u64 cost;
calc_vtime_cost_builtin(bio, iocg, is_merge, &cost);
return cost;
}
static void calc_size_vtime_cost_builtin(struct request *rq, struct ioc *ioc,
u64 *costp)
{
unsigned int pages = blk_rq_stats_sectors(rq) >> IOC_SECT_TO_PAGE_SHIFT;
switch (req_op(rq)) {
case REQ_OP_READ:
*costp = pages * ioc->params.lcoefs[LCOEF_RPAGE];
break;
case REQ_OP_WRITE:
*costp = pages * ioc->params.lcoefs[LCOEF_WPAGE];
break;
default:
*costp = 0;
}
}
static u64 calc_size_vtime_cost(struct request *rq, struct ioc *ioc)
{
u64 cost;
calc_size_vtime_cost_builtin(rq, ioc, &cost);
return cost;
}
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
static void ioc_rqos_throttle(struct rq_qos *rqos, struct bio *bio)
{
struct blkcg_gq *blkg = bio->bi_blkg;
struct ioc *ioc = rqos_to_ioc(rqos);
struct ioc_gq *iocg = blkg_to_iocg(blkg);
struct ioc_now now;
struct iocg_wait wait;
u64 abs_cost, cost, vtime;
bool use_debt, ioc_locked;
unsigned long flags;
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
/* bypass IOs if disabled or for root cgroup */
if (!ioc->enabled || !iocg->level)
return;
/* calculate the absolute vtime cost */
abs_cost = calc_vtime_cost(bio, iocg, false);
if (!abs_cost)
return;
blk-iocost: revamp donation amount determination iocost has various safety nets to combat inuse adjustment calculation inaccuracies. With Andy's method implemented in transfer_surpluses(), inuse adjustment calculations are now accurate and we can make donation amount determinations accurate too. * Stop keeping track of past usage history and using the maximum. Act on the immediate usage information. * Remove donation constraints defined by SURPLUS_* constants. Donate whatever isn't used. * Determine the donation amount so that the iocg will end up with MARGIN_TARGET_PCT budget at the end of the coming period assuming the same usage as the previous period. TARGET is set at 50% of period, which is the previous maximum. This provides smooth convergence for most repetitive IO patterns. * Apply donation logic early at 20% budget. There's no risk in doing so as the calculation is based on the delta between the current budget and the target budget at the end of the coming period. * Remove preemptive iocg activation for zero cost IOs. As donation can reach near zero now, the mere activation doesn't provide any protection anymore. In the unlikely case that this becomes a problem, the right solution is assigning appropriate costs for such IOs. This significantly improves the donation determination logic while also simplifying it. Now all donations are immediate, exact and smooth. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2020-09-01 18:52:49 +00:00
if (!iocg_activate(iocg, &now))
return;
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
iocg->cursor = bio_end_sector(bio);
vtime = atomic64_read(&iocg->vtime);
cost = adjust_inuse_and_calc_cost(iocg, vtime, abs_cost, &now);
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
/*
* If no one's waiting and within budget, issue right away. The
* tests are racy but the races aren't systemic - we only miss once
* in a while which is fine.
*/
iocost: protect iocg->abs_vdebt with iocg->waitq.lock abs_vdebt is an atomic_64 which tracks how much over budget a given cgroup is and controls the activation of use_delay mechanism. Once a cgroup goes over budget from forced IOs, it has to pay it back with its future budget. The progress guarantee on debt paying comes from the iocg being active - active iocgs are processed by the periodic timer, which ensures that as time passes the debts dissipate and the iocg returns to normal operation. However, both iocg activation and vdebt handling are asynchronous and a sequence like the following may happen. 1. The iocg is in the process of being deactivated by the periodic timer. 2. A bio enters ioc_rqos_throttle(), calls iocg_activate() which returns without anything because it still sees that the iocg is already active. 3. The iocg is deactivated. 4. The bio from #2 is over budget but needs to be forced. It increases abs_vdebt and goes over the threshold and enables use_delay. 5. IO control is enabled for the iocg's subtree and now IOs are attributed to the descendant cgroups and the iocg itself no longer issues IOs. This leaves the iocg with stuck abs_vdebt - it has debt but inactive and no further IOs which can activate it. This can end up unduly punishing all the descendants cgroups. The usual throttling path has the same issue - the iocg must be active while throttled to ensure that future event will wake it up - and solves the problem by synchronizing the throttling path with a spinlock. abs_vdebt handling is another form of overage handling and shares a lot of characteristics including the fact that it isn't in the hottest path. This patch fixes the above and other possible races by strictly synchronizing abs_vdebt and use_delay handling with iocg->waitq.lock. Signed-off-by: Tejun Heo <tj@kernel.org> Reported-by: Vlad Dmitriev <vvd@fb.com> Cc: stable@vger.kernel.org # v5.4+ Fixes: e1518f63f246 ("blk-iocost: Don't let merges push vtime into the future") Signed-off-by: Jens Axboe <axboe@kernel.dk>
2020-05-04 23:27:54 +00:00
if (!waitqueue_active(&iocg->waitq) && !iocg->abs_vdebt &&
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
time_before_eq64(vtime + cost, now.vnow)) {
iocg_commit_bio(iocg, bio, abs_cost, cost);
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
return;
}
blk-iocost: Account force-charged overage in absolute vtime Currently, when a bio needs to be force-charged and there isn't enough budget, vtime is simply pushed into the future. This means that the cost of the whole bio is scaled using the current hweight and then charged immediately. Until the global vtime advances beyond this future vtime, the cgroup won't be allowed to issue normal IOs. This is incorrect and can lead to, for example, exploding vrate or extended stalls if vrate range is constrained. Consider the following scenario. 1. A cgroup with a very low hweight runs out of budget. 2. A storm of swap-out happens on it. All of them are scaled according to the current low hweight and charged to vtime pushing it to a far future. 3. All other cgroups go idle and now the above cgroup has access to the whole device. However, because vtime is already wound using the past low hweight, what its current hweight is doesn't matter until global vtime catches up to the local vtime. 4. As a result, either vrate gets ramped up extremely or the IOs stall while the underlying device is idle. This is because the hweight the overage is calculated at is different from the hweight that it's being paid at. Fix it by remembering the overage in absoulte vtime and continuously paying with the actual budget according to the current hweight at each period. Note that non-forced bios which wait already remembers the cost in absolute vtime. This brings forced-bio accounting in line. Signed-off-by: Tejun Heo <tj@kernel.org> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-09-04 19:45:52 +00:00
/*
* We're over budget. This can be handled in two ways. IOs which may
* cause priority inversions are punted to @ioc->aux_iocg and charged as
* debt. Otherwise, the issuer is blocked on @iocg->waitq. Debt handling
* requires @ioc->lock, waitq handling @iocg->waitq.lock. Determine
* whether debt handling is needed and acquire locks accordingly.
iocost: protect iocg->abs_vdebt with iocg->waitq.lock abs_vdebt is an atomic_64 which tracks how much over budget a given cgroup is and controls the activation of use_delay mechanism. Once a cgroup goes over budget from forced IOs, it has to pay it back with its future budget. The progress guarantee on debt paying comes from the iocg being active - active iocgs are processed by the periodic timer, which ensures that as time passes the debts dissipate and the iocg returns to normal operation. However, both iocg activation and vdebt handling are asynchronous and a sequence like the following may happen. 1. The iocg is in the process of being deactivated by the periodic timer. 2. A bio enters ioc_rqos_throttle(), calls iocg_activate() which returns without anything because it still sees that the iocg is already active. 3. The iocg is deactivated. 4. The bio from #2 is over budget but needs to be forced. It increases abs_vdebt and goes over the threshold and enables use_delay. 5. IO control is enabled for the iocg's subtree and now IOs are attributed to the descendant cgroups and the iocg itself no longer issues IOs. This leaves the iocg with stuck abs_vdebt - it has debt but inactive and no further IOs which can activate it. This can end up unduly punishing all the descendants cgroups. The usual throttling path has the same issue - the iocg must be active while throttled to ensure that future event will wake it up - and solves the problem by synchronizing the throttling path with a spinlock. abs_vdebt handling is another form of overage handling and shares a lot of characteristics including the fact that it isn't in the hottest path. This patch fixes the above and other possible races by strictly synchronizing abs_vdebt and use_delay handling with iocg->waitq.lock. Signed-off-by: Tejun Heo <tj@kernel.org> Reported-by: Vlad Dmitriev <vvd@fb.com> Cc: stable@vger.kernel.org # v5.4+ Fixes: e1518f63f246 ("blk-iocost: Don't let merges push vtime into the future") Signed-off-by: Jens Axboe <axboe@kernel.dk>
2020-05-04 23:27:54 +00:00
*/
use_debt = bio_issue_as_root_blkg(bio) || fatal_signal_pending(current);
ioc_locked = use_debt || READ_ONCE(iocg->abs_vdebt);
retry_lock:
iocg_lock(iocg, ioc_locked, &flags);
/*
* @iocg must stay activated for debt and waitq handling. Deactivation
* is synchronized against both ioc->lock and waitq.lock and we won't
* get deactivated as long as we're waiting or has debt, so we're good
* if we're activated here. In the unlikely cases that we aren't, just
* issue the IO.
*/
iocost: protect iocg->abs_vdebt with iocg->waitq.lock abs_vdebt is an atomic_64 which tracks how much over budget a given cgroup is and controls the activation of use_delay mechanism. Once a cgroup goes over budget from forced IOs, it has to pay it back with its future budget. The progress guarantee on debt paying comes from the iocg being active - active iocgs are processed by the periodic timer, which ensures that as time passes the debts dissipate and the iocg returns to normal operation. However, both iocg activation and vdebt handling are asynchronous and a sequence like the following may happen. 1. The iocg is in the process of being deactivated by the periodic timer. 2. A bio enters ioc_rqos_throttle(), calls iocg_activate() which returns without anything because it still sees that the iocg is already active. 3. The iocg is deactivated. 4. The bio from #2 is over budget but needs to be forced. It increases abs_vdebt and goes over the threshold and enables use_delay. 5. IO control is enabled for the iocg's subtree and now IOs are attributed to the descendant cgroups and the iocg itself no longer issues IOs. This leaves the iocg with stuck abs_vdebt - it has debt but inactive and no further IOs which can activate it. This can end up unduly punishing all the descendants cgroups. The usual throttling path has the same issue - the iocg must be active while throttled to ensure that future event will wake it up - and solves the problem by synchronizing the throttling path with a spinlock. abs_vdebt handling is another form of overage handling and shares a lot of characteristics including the fact that it isn't in the hottest path. This patch fixes the above and other possible races by strictly synchronizing abs_vdebt and use_delay handling with iocg->waitq.lock. Signed-off-by: Tejun Heo <tj@kernel.org> Reported-by: Vlad Dmitriev <vvd@fb.com> Cc: stable@vger.kernel.org # v5.4+ Fixes: e1518f63f246 ("blk-iocost: Don't let merges push vtime into the future") Signed-off-by: Jens Axboe <axboe@kernel.dk>
2020-05-04 23:27:54 +00:00
if (unlikely(list_empty(&iocg->active_list))) {
iocg_unlock(iocg, ioc_locked, &flags);
iocg_commit_bio(iocg, bio, abs_cost, cost);
iocost: protect iocg->abs_vdebt with iocg->waitq.lock abs_vdebt is an atomic_64 which tracks how much over budget a given cgroup is and controls the activation of use_delay mechanism. Once a cgroup goes over budget from forced IOs, it has to pay it back with its future budget. The progress guarantee on debt paying comes from the iocg being active - active iocgs are processed by the periodic timer, which ensures that as time passes the debts dissipate and the iocg returns to normal operation. However, both iocg activation and vdebt handling are asynchronous and a sequence like the following may happen. 1. The iocg is in the process of being deactivated by the periodic timer. 2. A bio enters ioc_rqos_throttle(), calls iocg_activate() which returns without anything because it still sees that the iocg is already active. 3. The iocg is deactivated. 4. The bio from #2 is over budget but needs to be forced. It increases abs_vdebt and goes over the threshold and enables use_delay. 5. IO control is enabled for the iocg's subtree and now IOs are attributed to the descendant cgroups and the iocg itself no longer issues IOs. This leaves the iocg with stuck abs_vdebt - it has debt but inactive and no further IOs which can activate it. This can end up unduly punishing all the descendants cgroups. The usual throttling path has the same issue - the iocg must be active while throttled to ensure that future event will wake it up - and solves the problem by synchronizing the throttling path with a spinlock. abs_vdebt handling is another form of overage handling and shares a lot of characteristics including the fact that it isn't in the hottest path. This patch fixes the above and other possible races by strictly synchronizing abs_vdebt and use_delay handling with iocg->waitq.lock. Signed-off-by: Tejun Heo <tj@kernel.org> Reported-by: Vlad Dmitriev <vvd@fb.com> Cc: stable@vger.kernel.org # v5.4+ Fixes: e1518f63f246 ("blk-iocost: Don't let merges push vtime into the future") Signed-off-by: Jens Axboe <axboe@kernel.dk>
2020-05-04 23:27:54 +00:00
return;
}
/*
* We're over budget. If @bio has to be issued regardless, remember
* the abs_cost instead of advancing vtime. iocg_kick_waitq() will pay
* off the debt before waking more IOs.
*
blk-iocost: Account force-charged overage in absolute vtime Currently, when a bio needs to be force-charged and there isn't enough budget, vtime is simply pushed into the future. This means that the cost of the whole bio is scaled using the current hweight and then charged immediately. Until the global vtime advances beyond this future vtime, the cgroup won't be allowed to issue normal IOs. This is incorrect and can lead to, for example, exploding vrate or extended stalls if vrate range is constrained. Consider the following scenario. 1. A cgroup with a very low hweight runs out of budget. 2. A storm of swap-out happens on it. All of them are scaled according to the current low hweight and charged to vtime pushing it to a far future. 3. All other cgroups go idle and now the above cgroup has access to the whole device. However, because vtime is already wound using the past low hweight, what its current hweight is doesn't matter until global vtime catches up to the local vtime. 4. As a result, either vrate gets ramped up extremely or the IOs stall while the underlying device is idle. This is because the hweight the overage is calculated at is different from the hweight that it's being paid at. Fix it by remembering the overage in absoulte vtime and continuously paying with the actual budget according to the current hweight at each period. Note that non-forced bios which wait already remembers the cost in absolute vtime. This brings forced-bio accounting in line. Signed-off-by: Tejun Heo <tj@kernel.org> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-09-04 19:45:52 +00:00
* This way, the debt is continuously paid off each period with the
iocost: protect iocg->abs_vdebt with iocg->waitq.lock abs_vdebt is an atomic_64 which tracks how much over budget a given cgroup is and controls the activation of use_delay mechanism. Once a cgroup goes over budget from forced IOs, it has to pay it back with its future budget. The progress guarantee on debt paying comes from the iocg being active - active iocgs are processed by the periodic timer, which ensures that as time passes the debts dissipate and the iocg returns to normal operation. However, both iocg activation and vdebt handling are asynchronous and a sequence like the following may happen. 1. The iocg is in the process of being deactivated by the periodic timer. 2. A bio enters ioc_rqos_throttle(), calls iocg_activate() which returns without anything because it still sees that the iocg is already active. 3. The iocg is deactivated. 4. The bio from #2 is over budget but needs to be forced. It increases abs_vdebt and goes over the threshold and enables use_delay. 5. IO control is enabled for the iocg's subtree and now IOs are attributed to the descendant cgroups and the iocg itself no longer issues IOs. This leaves the iocg with stuck abs_vdebt - it has debt but inactive and no further IOs which can activate it. This can end up unduly punishing all the descendants cgroups. The usual throttling path has the same issue - the iocg must be active while throttled to ensure that future event will wake it up - and solves the problem by synchronizing the throttling path with a spinlock. abs_vdebt handling is another form of overage handling and shares a lot of characteristics including the fact that it isn't in the hottest path. This patch fixes the above and other possible races by strictly synchronizing abs_vdebt and use_delay handling with iocg->waitq.lock. Signed-off-by: Tejun Heo <tj@kernel.org> Reported-by: Vlad Dmitriev <vvd@fb.com> Cc: stable@vger.kernel.org # v5.4+ Fixes: e1518f63f246 ("blk-iocost: Don't let merges push vtime into the future") Signed-off-by: Jens Axboe <axboe@kernel.dk>
2020-05-04 23:27:54 +00:00
* actual budget available to the cgroup. If we just wound vtime, we
* would incorrectly use the current hw_inuse for the entire amount
* which, for example, can lead to the cgroup staying blocked for a
* long time even with substantially raised hw_inuse.
*
* An iocg with vdebt should stay online so that the timer can keep
* deducting its vdebt and [de]activate use_delay mechanism
* accordingly. We don't want to race against the timer trying to
* clear them and leave @iocg inactive w/ dangling use_delay heavily
* penalizing the cgroup and its descendants.
blk-iocost: Account force-charged overage in absolute vtime Currently, when a bio needs to be force-charged and there isn't enough budget, vtime is simply pushed into the future. This means that the cost of the whole bio is scaled using the current hweight and then charged immediately. Until the global vtime advances beyond this future vtime, the cgroup won't be allowed to issue normal IOs. This is incorrect and can lead to, for example, exploding vrate or extended stalls if vrate range is constrained. Consider the following scenario. 1. A cgroup with a very low hweight runs out of budget. 2. A storm of swap-out happens on it. All of them are scaled according to the current low hweight and charged to vtime pushing it to a far future. 3. All other cgroups go idle and now the above cgroup has access to the whole device. However, because vtime is already wound using the past low hweight, what its current hweight is doesn't matter until global vtime catches up to the local vtime. 4. As a result, either vrate gets ramped up extremely or the IOs stall while the underlying device is idle. This is because the hweight the overage is calculated at is different from the hweight that it's being paid at. Fix it by remembering the overage in absoulte vtime and continuously paying with the actual budget according to the current hweight at each period. Note that non-forced bios which wait already remembers the cost in absolute vtime. This brings forced-bio accounting in line. Signed-off-by: Tejun Heo <tj@kernel.org> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-09-04 19:45:52 +00:00
*/
if (use_debt) {
iocg_incur_debt(iocg, abs_cost, &now);
blk-iocost: switch to fixed non-auto-decaying use_delay The use_delay mechanism was introduced by blk-iolatency to hold memory allocators accountable for the reclaim and other shared IOs they cause. The duration of the delay is dynamically balanced between iolatency increasing the value on each target miss and it auto-decaying as time passes and threads get delayed on it. While this works well for iolatency, iocost's control model isn't compatible with it. There is no repeated "violation" events which can be balanced against auto-decaying. iocost instead knows how much a given cgroup is over budget and wants to prevent that cgroup from issuing IOs while over budget. Until now, iocost has been adding the cost of force-issued IOs. However, this doesn't reflect the amount which is already over budget and is simply not enough to counter the auto-decaying allowing anon-memory leaking low priority cgroup to go over its alloted share of IOs. As auto-decaying doesn't make much sense for iocost, this patch introduces a different mode of operation for use_delay - when blkcg_set_delay() are used insted of blkcg_add/use_delay(), the delay duration is not auto-decayed until it is explicitly cleared with blkcg_clear_delay(). iocost is updated to keep the delay duration synchronized to the budget overage amount. With this change, iocost can effectively police cgroups which generate significant amount of force-issued IOs. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Josef Bacik <josef@toxicpanda.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2020-04-13 16:27:55 +00:00
if (iocg_kick_delay(iocg, &now))
blkcg_schedule_throttle(rqos->q,
(bio->bi_opf & REQ_SWAP) == REQ_SWAP);
iocg_unlock(iocg, ioc_locked, &flags);
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
return;
}
/* guarantee that iocgs w/ waiters have maximum inuse */
if (!iocg->abs_vdebt && iocg->inuse != iocg->active) {
if (!ioc_locked) {
iocg_unlock(iocg, false, &flags);
ioc_locked = true;
goto retry_lock;
}
propagate_weights(iocg, iocg->active, iocg->active, true,
&now);
}
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
/*
* Append self to the waitq and schedule the wakeup timer if we're
* the first waiter. The timer duration is calculated based on the
* current vrate. vtime and hweight changes can make it too short
* or too long. Each wait entry records the absolute cost it's
* waiting for to allow re-evaluation using a custom wait entry.
*
* If too short, the timer simply reschedules itself. If too long,
* the period timer will notice and trigger wakeups.
*
* All waiters are on iocg->waitq and the wait states are
* synchronized using waitq.lock.
*/
init_waitqueue_func_entry(&wait.wait, iocg_wake_fn);
wait.wait.private = current;
wait.bio = bio;
wait.abs_cost = abs_cost;
wait.committed = false; /* will be set true by waker */
__add_wait_queue_entry_tail(&iocg->waitq, &wait.wait);
iocg_kick_waitq(iocg, ioc_locked, &now);
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
iocg_unlock(iocg, ioc_locked, &flags);
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
while (true) {
set_current_state(TASK_UNINTERRUPTIBLE);
if (wait.committed)
break;
io_schedule();
}
/* waker already committed us, proceed */
finish_wait(&iocg->waitq, &wait.wait);
}
static void ioc_rqos_merge(struct rq_qos *rqos, struct request *rq,
struct bio *bio)
{
struct ioc_gq *iocg = blkg_to_iocg(bio->bi_blkg);
struct ioc *ioc = iocg->ioc;
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
sector_t bio_end = bio_end_sector(bio);
struct ioc_now now;
u64 vtime, abs_cost, cost;
iocost: protect iocg->abs_vdebt with iocg->waitq.lock abs_vdebt is an atomic_64 which tracks how much over budget a given cgroup is and controls the activation of use_delay mechanism. Once a cgroup goes over budget from forced IOs, it has to pay it back with its future budget. The progress guarantee on debt paying comes from the iocg being active - active iocgs are processed by the periodic timer, which ensures that as time passes the debts dissipate and the iocg returns to normal operation. However, both iocg activation and vdebt handling are asynchronous and a sequence like the following may happen. 1. The iocg is in the process of being deactivated by the periodic timer. 2. A bio enters ioc_rqos_throttle(), calls iocg_activate() which returns without anything because it still sees that the iocg is already active. 3. The iocg is deactivated. 4. The bio from #2 is over budget but needs to be forced. It increases abs_vdebt and goes over the threshold and enables use_delay. 5. IO control is enabled for the iocg's subtree and now IOs are attributed to the descendant cgroups and the iocg itself no longer issues IOs. This leaves the iocg with stuck abs_vdebt - it has debt but inactive and no further IOs which can activate it. This can end up unduly punishing all the descendants cgroups. The usual throttling path has the same issue - the iocg must be active while throttled to ensure that future event will wake it up - and solves the problem by synchronizing the throttling path with a spinlock. abs_vdebt handling is another form of overage handling and shares a lot of characteristics including the fact that it isn't in the hottest path. This patch fixes the above and other possible races by strictly synchronizing abs_vdebt and use_delay handling with iocg->waitq.lock. Signed-off-by: Tejun Heo <tj@kernel.org> Reported-by: Vlad Dmitriev <vvd@fb.com> Cc: stable@vger.kernel.org # v5.4+ Fixes: e1518f63f246 ("blk-iocost: Don't let merges push vtime into the future") Signed-off-by: Jens Axboe <axboe@kernel.dk>
2020-05-04 23:27:54 +00:00
unsigned long flags;
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
/* bypass if disabled or for root cgroup */
if (!ioc->enabled || !iocg->level)
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
return;
abs_cost = calc_vtime_cost(bio, iocg, true);
if (!abs_cost)
return;
ioc_now(ioc, &now);
vtime = atomic64_read(&iocg->vtime);
cost = adjust_inuse_and_calc_cost(iocg, vtime, abs_cost, &now);
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
/* update cursor if backmerging into the request at the cursor */
if (blk_rq_pos(rq) < bio_end &&
blk_rq_pos(rq) + blk_rq_sectors(rq) == iocg->cursor)
iocg->cursor = bio_end;
/*
iocost: protect iocg->abs_vdebt with iocg->waitq.lock abs_vdebt is an atomic_64 which tracks how much over budget a given cgroup is and controls the activation of use_delay mechanism. Once a cgroup goes over budget from forced IOs, it has to pay it back with its future budget. The progress guarantee on debt paying comes from the iocg being active - active iocgs are processed by the periodic timer, which ensures that as time passes the debts dissipate and the iocg returns to normal operation. However, both iocg activation and vdebt handling are asynchronous and a sequence like the following may happen. 1. The iocg is in the process of being deactivated by the periodic timer. 2. A bio enters ioc_rqos_throttle(), calls iocg_activate() which returns without anything because it still sees that the iocg is already active. 3. The iocg is deactivated. 4. The bio from #2 is over budget but needs to be forced. It increases abs_vdebt and goes over the threshold and enables use_delay. 5. IO control is enabled for the iocg's subtree and now IOs are attributed to the descendant cgroups and the iocg itself no longer issues IOs. This leaves the iocg with stuck abs_vdebt - it has debt but inactive and no further IOs which can activate it. This can end up unduly punishing all the descendants cgroups. The usual throttling path has the same issue - the iocg must be active while throttled to ensure that future event will wake it up - and solves the problem by synchronizing the throttling path with a spinlock. abs_vdebt handling is another form of overage handling and shares a lot of characteristics including the fact that it isn't in the hottest path. This patch fixes the above and other possible races by strictly synchronizing abs_vdebt and use_delay handling with iocg->waitq.lock. Signed-off-by: Tejun Heo <tj@kernel.org> Reported-by: Vlad Dmitriev <vvd@fb.com> Cc: stable@vger.kernel.org # v5.4+ Fixes: e1518f63f246 ("blk-iocost: Don't let merges push vtime into the future") Signed-off-by: Jens Axboe <axboe@kernel.dk>
2020-05-04 23:27:54 +00:00
* Charge if there's enough vtime budget and the existing request has
* cost assigned.
*/
if (rq->bio && rq->bio->bi_iocost_cost &&
iocost: protect iocg->abs_vdebt with iocg->waitq.lock abs_vdebt is an atomic_64 which tracks how much over budget a given cgroup is and controls the activation of use_delay mechanism. Once a cgroup goes over budget from forced IOs, it has to pay it back with its future budget. The progress guarantee on debt paying comes from the iocg being active - active iocgs are processed by the periodic timer, which ensures that as time passes the debts dissipate and the iocg returns to normal operation. However, both iocg activation and vdebt handling are asynchronous and a sequence like the following may happen. 1. The iocg is in the process of being deactivated by the periodic timer. 2. A bio enters ioc_rqos_throttle(), calls iocg_activate() which returns without anything because it still sees that the iocg is already active. 3. The iocg is deactivated. 4. The bio from #2 is over budget but needs to be forced. It increases abs_vdebt and goes over the threshold and enables use_delay. 5. IO control is enabled for the iocg's subtree and now IOs are attributed to the descendant cgroups and the iocg itself no longer issues IOs. This leaves the iocg with stuck abs_vdebt - it has debt but inactive and no further IOs which can activate it. This can end up unduly punishing all the descendants cgroups. The usual throttling path has the same issue - the iocg must be active while throttled to ensure that future event will wake it up - and solves the problem by synchronizing the throttling path with a spinlock. abs_vdebt handling is another form of overage handling and shares a lot of characteristics including the fact that it isn't in the hottest path. This patch fixes the above and other possible races by strictly synchronizing abs_vdebt and use_delay handling with iocg->waitq.lock. Signed-off-by: Tejun Heo <tj@kernel.org> Reported-by: Vlad Dmitriev <vvd@fb.com> Cc: stable@vger.kernel.org # v5.4+ Fixes: e1518f63f246 ("blk-iocost: Don't let merges push vtime into the future") Signed-off-by: Jens Axboe <axboe@kernel.dk>
2020-05-04 23:27:54 +00:00
time_before_eq64(atomic64_read(&iocg->vtime) + cost, now.vnow)) {
iocg_commit_bio(iocg, bio, abs_cost, cost);
iocost: protect iocg->abs_vdebt with iocg->waitq.lock abs_vdebt is an atomic_64 which tracks how much over budget a given cgroup is and controls the activation of use_delay mechanism. Once a cgroup goes over budget from forced IOs, it has to pay it back with its future budget. The progress guarantee on debt paying comes from the iocg being active - active iocgs are processed by the periodic timer, which ensures that as time passes the debts dissipate and the iocg returns to normal operation. However, both iocg activation and vdebt handling are asynchronous and a sequence like the following may happen. 1. The iocg is in the process of being deactivated by the periodic timer. 2. A bio enters ioc_rqos_throttle(), calls iocg_activate() which returns without anything because it still sees that the iocg is already active. 3. The iocg is deactivated. 4. The bio from #2 is over budget but needs to be forced. It increases abs_vdebt and goes over the threshold and enables use_delay. 5. IO control is enabled for the iocg's subtree and now IOs are attributed to the descendant cgroups and the iocg itself no longer issues IOs. This leaves the iocg with stuck abs_vdebt - it has debt but inactive and no further IOs which can activate it. This can end up unduly punishing all the descendants cgroups. The usual throttling path has the same issue - the iocg must be active while throttled to ensure that future event will wake it up - and solves the problem by synchronizing the throttling path with a spinlock. abs_vdebt handling is another form of overage handling and shares a lot of characteristics including the fact that it isn't in the hottest path. This patch fixes the above and other possible races by strictly synchronizing abs_vdebt and use_delay handling with iocg->waitq.lock. Signed-off-by: Tejun Heo <tj@kernel.org> Reported-by: Vlad Dmitriev <vvd@fb.com> Cc: stable@vger.kernel.org # v5.4+ Fixes: e1518f63f246 ("blk-iocost: Don't let merges push vtime into the future") Signed-off-by: Jens Axboe <axboe@kernel.dk>
2020-05-04 23:27:54 +00:00
return;
}
/*
* Otherwise, account it as debt if @iocg is online, which it should
* be for the vast majority of cases. See debt handling in
* ioc_rqos_throttle() for details.
*/
spin_lock_irqsave(&ioc->lock, flags);
spin_lock(&iocg->waitq.lock);
iocost: protect iocg->abs_vdebt with iocg->waitq.lock abs_vdebt is an atomic_64 which tracks how much over budget a given cgroup is and controls the activation of use_delay mechanism. Once a cgroup goes over budget from forced IOs, it has to pay it back with its future budget. The progress guarantee on debt paying comes from the iocg being active - active iocgs are processed by the periodic timer, which ensures that as time passes the debts dissipate and the iocg returns to normal operation. However, both iocg activation and vdebt handling are asynchronous and a sequence like the following may happen. 1. The iocg is in the process of being deactivated by the periodic timer. 2. A bio enters ioc_rqos_throttle(), calls iocg_activate() which returns without anything because it still sees that the iocg is already active. 3. The iocg is deactivated. 4. The bio from #2 is over budget but needs to be forced. It increases abs_vdebt and goes over the threshold and enables use_delay. 5. IO control is enabled for the iocg's subtree and now IOs are attributed to the descendant cgroups and the iocg itself no longer issues IOs. This leaves the iocg with stuck abs_vdebt - it has debt but inactive and no further IOs which can activate it. This can end up unduly punishing all the descendants cgroups. The usual throttling path has the same issue - the iocg must be active while throttled to ensure that future event will wake it up - and solves the problem by synchronizing the throttling path with a spinlock. abs_vdebt handling is another form of overage handling and shares a lot of characteristics including the fact that it isn't in the hottest path. This patch fixes the above and other possible races by strictly synchronizing abs_vdebt and use_delay handling with iocg->waitq.lock. Signed-off-by: Tejun Heo <tj@kernel.org> Reported-by: Vlad Dmitriev <vvd@fb.com> Cc: stable@vger.kernel.org # v5.4+ Fixes: e1518f63f246 ("blk-iocost: Don't let merges push vtime into the future") Signed-off-by: Jens Axboe <axboe@kernel.dk>
2020-05-04 23:27:54 +00:00
if (likely(!list_empty(&iocg->active_list))) {
iocg_incur_debt(iocg, abs_cost, &now);
if (iocg_kick_delay(iocg, &now))
blkcg_schedule_throttle(rqos->q,
(bio->bi_opf & REQ_SWAP) == REQ_SWAP);
iocost: protect iocg->abs_vdebt with iocg->waitq.lock abs_vdebt is an atomic_64 which tracks how much over budget a given cgroup is and controls the activation of use_delay mechanism. Once a cgroup goes over budget from forced IOs, it has to pay it back with its future budget. The progress guarantee on debt paying comes from the iocg being active - active iocgs are processed by the periodic timer, which ensures that as time passes the debts dissipate and the iocg returns to normal operation. However, both iocg activation and vdebt handling are asynchronous and a sequence like the following may happen. 1. The iocg is in the process of being deactivated by the periodic timer. 2. A bio enters ioc_rqos_throttle(), calls iocg_activate() which returns without anything because it still sees that the iocg is already active. 3. The iocg is deactivated. 4. The bio from #2 is over budget but needs to be forced. It increases abs_vdebt and goes over the threshold and enables use_delay. 5. IO control is enabled for the iocg's subtree and now IOs are attributed to the descendant cgroups and the iocg itself no longer issues IOs. This leaves the iocg with stuck abs_vdebt - it has debt but inactive and no further IOs which can activate it. This can end up unduly punishing all the descendants cgroups. The usual throttling path has the same issue - the iocg must be active while throttled to ensure that future event will wake it up - and solves the problem by synchronizing the throttling path with a spinlock. abs_vdebt handling is another form of overage handling and shares a lot of characteristics including the fact that it isn't in the hottest path. This patch fixes the above and other possible races by strictly synchronizing abs_vdebt and use_delay handling with iocg->waitq.lock. Signed-off-by: Tejun Heo <tj@kernel.org> Reported-by: Vlad Dmitriev <vvd@fb.com> Cc: stable@vger.kernel.org # v5.4+ Fixes: e1518f63f246 ("blk-iocost: Don't let merges push vtime into the future") Signed-off-by: Jens Axboe <axboe@kernel.dk>
2020-05-04 23:27:54 +00:00
} else {
iocg_commit_bio(iocg, bio, abs_cost, cost);
iocost: protect iocg->abs_vdebt with iocg->waitq.lock abs_vdebt is an atomic_64 which tracks how much over budget a given cgroup is and controls the activation of use_delay mechanism. Once a cgroup goes over budget from forced IOs, it has to pay it back with its future budget. The progress guarantee on debt paying comes from the iocg being active - active iocgs are processed by the periodic timer, which ensures that as time passes the debts dissipate and the iocg returns to normal operation. However, both iocg activation and vdebt handling are asynchronous and a sequence like the following may happen. 1. The iocg is in the process of being deactivated by the periodic timer. 2. A bio enters ioc_rqos_throttle(), calls iocg_activate() which returns without anything because it still sees that the iocg is already active. 3. The iocg is deactivated. 4. The bio from #2 is over budget but needs to be forced. It increases abs_vdebt and goes over the threshold and enables use_delay. 5. IO control is enabled for the iocg's subtree and now IOs are attributed to the descendant cgroups and the iocg itself no longer issues IOs. This leaves the iocg with stuck abs_vdebt - it has debt but inactive and no further IOs which can activate it. This can end up unduly punishing all the descendants cgroups. The usual throttling path has the same issue - the iocg must be active while throttled to ensure that future event will wake it up - and solves the problem by synchronizing the throttling path with a spinlock. abs_vdebt handling is another form of overage handling and shares a lot of characteristics including the fact that it isn't in the hottest path. This patch fixes the above and other possible races by strictly synchronizing abs_vdebt and use_delay handling with iocg->waitq.lock. Signed-off-by: Tejun Heo <tj@kernel.org> Reported-by: Vlad Dmitriev <vvd@fb.com> Cc: stable@vger.kernel.org # v5.4+ Fixes: e1518f63f246 ("blk-iocost: Don't let merges push vtime into the future") Signed-off-by: Jens Axboe <axboe@kernel.dk>
2020-05-04 23:27:54 +00:00
}
spin_unlock(&iocg->waitq.lock);
spin_unlock_irqrestore(&ioc->lock, flags);
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
}
static void ioc_rqos_done_bio(struct rq_qos *rqos, struct bio *bio)
{
struct ioc_gq *iocg = blkg_to_iocg(bio->bi_blkg);
if (iocg && bio->bi_iocost_cost)
atomic64_add(bio->bi_iocost_cost, &iocg->done_vtime);
}
static void ioc_rqos_done(struct rq_qos *rqos, struct request *rq)
{
struct ioc *ioc = rqos_to_ioc(rqos);
struct ioc_pcpu_stat *ccs;
u64 on_q_ns, rq_wait_ns, size_nsec;
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
int pidx, rw;
if (!ioc->enabled || !rq->alloc_time_ns || !rq->start_time_ns)
return;
switch (req_op(rq) & REQ_OP_MASK) {
case REQ_OP_READ:
pidx = QOS_RLAT;
rw = READ;
break;
case REQ_OP_WRITE:
pidx = QOS_WLAT;
rw = WRITE;
break;
default:
return;
}
on_q_ns = ktime_get_ns() - rq->alloc_time_ns;
rq_wait_ns = rq->start_time_ns - rq->alloc_time_ns;
size_nsec = div64_u64(calc_size_vtime_cost(rq, ioc), VTIME_PER_NSEC);
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
ccs = get_cpu_ptr(ioc->pcpu_stat);
if (on_q_ns <= size_nsec ||
on_q_ns - size_nsec <= ioc->params.qos[pidx] * NSEC_PER_USEC)
local_inc(&ccs->missed[rw].nr_met);
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
else
local_inc(&ccs->missed[rw].nr_missed);
local64_add(rq_wait_ns, &ccs->rq_wait_ns);
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
put_cpu_ptr(ccs);
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
}
static void ioc_rqos_queue_depth_changed(struct rq_qos *rqos)
{
struct ioc *ioc = rqos_to_ioc(rqos);
spin_lock_irq(&ioc->lock);
ioc_refresh_params(ioc, false);
spin_unlock_irq(&ioc->lock);
}
static void ioc_rqos_exit(struct rq_qos *rqos)
{
struct ioc *ioc = rqos_to_ioc(rqos);
blkcg_deactivate_policy(rqos->q, &blkcg_policy_iocost);
spin_lock_irq(&ioc->lock);
ioc->running = IOC_STOP;
spin_unlock_irq(&ioc->lock);
del_timer_sync(&ioc->timer);
free_percpu(ioc->pcpu_stat);
kfree(ioc);
}
static struct rq_qos_ops ioc_rqos_ops = {
.throttle = ioc_rqos_throttle,
.merge = ioc_rqos_merge,
.done_bio = ioc_rqos_done_bio,
.done = ioc_rqos_done,
.queue_depth_changed = ioc_rqos_queue_depth_changed,
.exit = ioc_rqos_exit,
};
static int blk_iocost_init(struct request_queue *q)
{
struct ioc *ioc;
struct rq_qos *rqos;
int i, cpu, ret;
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
ioc = kzalloc(sizeof(*ioc), GFP_KERNEL);
if (!ioc)
return -ENOMEM;
ioc->pcpu_stat = alloc_percpu(struct ioc_pcpu_stat);
if (!ioc->pcpu_stat) {
kfree(ioc);
return -ENOMEM;
}
for_each_possible_cpu(cpu) {
struct ioc_pcpu_stat *ccs = per_cpu_ptr(ioc->pcpu_stat, cpu);
for (i = 0; i < ARRAY_SIZE(ccs->missed); i++) {
local_set(&ccs->missed[i].nr_met, 0);
local_set(&ccs->missed[i].nr_missed, 0);
}
local64_set(&ccs->rq_wait_ns, 0);
}
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
rqos = &ioc->rqos;
rqos->id = RQ_QOS_COST;
rqos->ops = &ioc_rqos_ops;
rqos->q = q;
spin_lock_init(&ioc->lock);
timer_setup(&ioc->timer, ioc_timer_fn, 0);
INIT_LIST_HEAD(&ioc->active_iocgs);
ioc->running = IOC_IDLE;
ioc->vtime_base_rate = VTIME_PER_USEC;
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
atomic64_set(&ioc->vtime_rate, VTIME_PER_USEC);
seqcount_spinlock_init(&ioc->period_seqcount, &ioc->lock);
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
ioc->period_at = ktime_to_us(ktime_get());
atomic64_set(&ioc->cur_period, 0);
atomic_set(&ioc->hweight_gen, 0);
spin_lock_irq(&ioc->lock);
ioc->autop_idx = AUTOP_INVALID;
ioc_refresh_params(ioc, true);
spin_unlock_irq(&ioc->lock);
rq_qos_add(q, rqos);
ret = blkcg_activate_policy(q, &blkcg_policy_iocost);
if (ret) {
rq_qos_del(q, rqos);
free_percpu(ioc->pcpu_stat);
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
kfree(ioc);
return ret;
}
return 0;
}
static struct blkcg_policy_data *ioc_cpd_alloc(gfp_t gfp)
{
struct ioc_cgrp *iocc;
iocc = kzalloc(sizeof(struct ioc_cgrp), gfp);
if (!iocc)
return NULL;
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
iocc->dfl_weight = CGROUP_WEIGHT_DFL * WEIGHT_ONE;
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
return &iocc->cpd;
}
static void ioc_cpd_free(struct blkcg_policy_data *cpd)
{
kfree(container_of(cpd, struct ioc_cgrp, cpd));
}
static struct blkg_policy_data *ioc_pd_alloc(gfp_t gfp, struct request_queue *q,
struct blkcg *blkcg)
{
int levels = blkcg->css.cgroup->level + 1;
struct ioc_gq *iocg;
iocg = kzalloc_node(struct_size(iocg, ancestors, levels), gfp, q->node);
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
if (!iocg)
return NULL;
iocg->pcpu_stat = alloc_percpu_gfp(struct iocg_pcpu_stat, gfp);
if (!iocg->pcpu_stat) {
kfree(iocg);
return NULL;
}
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
return &iocg->pd;
}
static void ioc_pd_init(struct blkg_policy_data *pd)
{
struct ioc_gq *iocg = pd_to_iocg(pd);
struct blkcg_gq *blkg = pd_to_blkg(&iocg->pd);
struct ioc *ioc = q_to_ioc(blkg->q);
struct ioc_now now;
struct blkcg_gq *tblkg;
unsigned long flags;
ioc_now(ioc, &now);
iocg->ioc = ioc;
atomic64_set(&iocg->vtime, now.vnow);
atomic64_set(&iocg->done_vtime, now.vnow);
atomic64_set(&iocg->active_period, atomic64_read(&ioc->cur_period));
INIT_LIST_HEAD(&iocg->active_list);
INIT_LIST_HEAD(&iocg->walk_list);
INIT_LIST_HEAD(&iocg->surplus_list);
iocg->hweight_active = WEIGHT_ONE;
iocg->hweight_inuse = WEIGHT_ONE;
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
init_waitqueue_head(&iocg->waitq);
hrtimer_init(&iocg->waitq_timer, CLOCK_MONOTONIC, HRTIMER_MODE_ABS);
iocg->waitq_timer.function = iocg_waitq_timer_fn;
iocg->level = blkg->blkcg->css.cgroup->level;
for (tblkg = blkg; tblkg; tblkg = tblkg->parent) {
struct ioc_gq *tiocg = blkg_to_iocg(tblkg);
iocg->ancestors[tiocg->level] = tiocg;
}
spin_lock_irqsave(&ioc->lock, flags);
weight_updated(iocg, &now);
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
spin_unlock_irqrestore(&ioc->lock, flags);
}
static void ioc_pd_free(struct blkg_policy_data *pd)
{
struct ioc_gq *iocg = pd_to_iocg(pd);
struct ioc *ioc = iocg->ioc;
unsigned long flags;
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
if (ioc) {
spin_lock_irqsave(&ioc->lock, flags);
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
if (!list_empty(&iocg->active_list)) {
struct ioc_now now;
ioc_now(ioc, &now);
propagate_weights(iocg, 0, 0, false, &now);
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
list_del_init(&iocg->active_list);
}
WARN_ON_ONCE(!list_empty(&iocg->walk_list));
WARN_ON_ONCE(!list_empty(&iocg->surplus_list));
spin_unlock_irqrestore(&ioc->lock, flags);
2019-09-10 16:15:25 +00:00
hrtimer_cancel(&iocg->waitq_timer);
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
}
free_percpu(iocg->pcpu_stat);
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
kfree(iocg);
}
static size_t ioc_pd_stat(struct blkg_policy_data *pd, char *buf, size_t size)
{
struct ioc_gq *iocg = pd_to_iocg(pd);
struct ioc *ioc = iocg->ioc;
size_t pos = 0;
if (!ioc->enabled)
return 0;
if (iocg->level == 0) {
unsigned vp10k = DIV64_U64_ROUND_CLOSEST(
ioc->vtime_base_rate * 10000,
VTIME_PER_USEC);
pos += scnprintf(buf + pos, size - pos, " cost.vrate=%u.%02u",
vp10k / 100, vp10k % 100);
}
pos += scnprintf(buf + pos, size - pos, " cost.usage=%llu",
iocg->last_stat.usage_us);
if (blkcg_debug_stats)
pos += scnprintf(buf + pos, size - pos,
" cost.wait=%llu cost.indebt=%llu cost.indelay=%llu",
iocg->last_stat.wait_us,
iocg->last_stat.indebt_us,
iocg->last_stat.indelay_us);
return pos;
}
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
static u64 ioc_weight_prfill(struct seq_file *sf, struct blkg_policy_data *pd,
int off)
{
const char *dname = blkg_dev_name(pd->blkg);
struct ioc_gq *iocg = pd_to_iocg(pd);
if (dname && iocg->cfg_weight)
seq_printf(sf, "%s %u\n", dname, iocg->cfg_weight / WEIGHT_ONE);
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
return 0;
}
static int ioc_weight_show(struct seq_file *sf, void *v)
{
struct blkcg *blkcg = css_to_blkcg(seq_css(sf));
struct ioc_cgrp *iocc = blkcg_to_iocc(blkcg);
seq_printf(sf, "default %u\n", iocc->dfl_weight / WEIGHT_ONE);
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
blkcg_print_blkgs(sf, blkcg, ioc_weight_prfill,
&blkcg_policy_iocost, seq_cft(sf)->private, false);
return 0;
}
static ssize_t ioc_weight_write(struct kernfs_open_file *of, char *buf,
size_t nbytes, loff_t off)
{
struct blkcg *blkcg = css_to_blkcg(of_css(of));
struct ioc_cgrp *iocc = blkcg_to_iocc(blkcg);
struct blkg_conf_ctx ctx;
struct ioc_now now;
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
struct ioc_gq *iocg;
u32 v;
int ret;
if (!strchr(buf, ':')) {
struct blkcg_gq *blkg;
if (!sscanf(buf, "default %u", &v) && !sscanf(buf, "%u", &v))
return -EINVAL;
if (v < CGROUP_WEIGHT_MIN || v > CGROUP_WEIGHT_MAX)
return -EINVAL;
spin_lock(&blkcg->lock);
iocc->dfl_weight = v * WEIGHT_ONE;
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
hlist_for_each_entry(blkg, &blkcg->blkg_list, blkcg_node) {
struct ioc_gq *iocg = blkg_to_iocg(blkg);
if (iocg) {
spin_lock_irq(&iocg->ioc->lock);
ioc_now(iocg->ioc, &now);
weight_updated(iocg, &now);
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
spin_unlock_irq(&iocg->ioc->lock);
}
}
spin_unlock(&blkcg->lock);
return nbytes;
}
ret = blkg_conf_prep(blkcg, &blkcg_policy_iocost, buf, &ctx);
if (ret)
return ret;
iocg = blkg_to_iocg(ctx.blkg);
if (!strncmp(ctx.body, "default", 7)) {
v = 0;
} else {
if (!sscanf(ctx.body, "%u", &v))
goto einval;
if (v < CGROUP_WEIGHT_MIN || v > CGROUP_WEIGHT_MAX)
goto einval;
}
spin_lock(&iocg->ioc->lock);
iocg->cfg_weight = v * WEIGHT_ONE;
ioc_now(iocg->ioc, &now);
weight_updated(iocg, &now);
spin_unlock(&iocg->ioc->lock);
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
blkg_conf_finish(&ctx);
return nbytes;
einval:
blkg_conf_finish(&ctx);
return -EINVAL;
}
static u64 ioc_qos_prfill(struct seq_file *sf, struct blkg_policy_data *pd,
int off)
{
const char *dname = blkg_dev_name(pd->blkg);
struct ioc *ioc = pd_to_iocg(pd)->ioc;
if (!dname)
return 0;
seq_printf(sf, "%s enable=%d ctrl=%s rpct=%u.%02u rlat=%u wpct=%u.%02u wlat=%u min=%u.%02u max=%u.%02u\n",
dname, ioc->enabled, ioc->user_qos_params ? "user" : "auto",
ioc->params.qos[QOS_RPPM] / 10000,
ioc->params.qos[QOS_RPPM] % 10000 / 100,
ioc->params.qos[QOS_RLAT],
ioc->params.qos[QOS_WPPM] / 10000,
ioc->params.qos[QOS_WPPM] % 10000 / 100,
ioc->params.qos[QOS_WLAT],
ioc->params.qos[QOS_MIN] / 10000,
ioc->params.qos[QOS_MIN] % 10000 / 100,
ioc->params.qos[QOS_MAX] / 10000,
ioc->params.qos[QOS_MAX] % 10000 / 100);
return 0;
}
static int ioc_qos_show(struct seq_file *sf, void *v)
{
struct blkcg *blkcg = css_to_blkcg(seq_css(sf));
blkcg_print_blkgs(sf, blkcg, ioc_qos_prfill,
&blkcg_policy_iocost, seq_cft(sf)->private, false);
return 0;
}
static const match_table_t qos_ctrl_tokens = {
{ QOS_ENABLE, "enable=%u" },
{ QOS_CTRL, "ctrl=%s" },
{ NR_QOS_CTRL_PARAMS, NULL },
};
static const match_table_t qos_tokens = {
{ QOS_RPPM, "rpct=%s" },
{ QOS_RLAT, "rlat=%u" },
{ QOS_WPPM, "wpct=%s" },
{ QOS_WLAT, "wlat=%u" },
{ QOS_MIN, "min=%s" },
{ QOS_MAX, "max=%s" },
{ NR_QOS_PARAMS, NULL },
};
static ssize_t ioc_qos_write(struct kernfs_open_file *of, char *input,
size_t nbytes, loff_t off)
{
struct gendisk *disk;
struct ioc *ioc;
u32 qos[NR_QOS_PARAMS];
bool enable, user;
char *p;
int ret;
disk = blkcg_conf_get_disk(&input);
if (IS_ERR(disk))
return PTR_ERR(disk);
ioc = q_to_ioc(disk->queue);
if (!ioc) {
ret = blk_iocost_init(disk->queue);
if (ret)
goto err;
ioc = q_to_ioc(disk->queue);
}
spin_lock_irq(&ioc->lock);
memcpy(qos, ioc->params.qos, sizeof(qos));
enable = ioc->enabled;
user = ioc->user_qos_params;
spin_unlock_irq(&ioc->lock);
while ((p = strsep(&input, " \t\n"))) {
substring_t args[MAX_OPT_ARGS];
char buf[32];
int tok;
s64 v;
if (!*p)
continue;
switch (match_token(p, qos_ctrl_tokens, args)) {
case QOS_ENABLE:
match_u64(&args[0], &v);
enable = v;
continue;
case QOS_CTRL:
match_strlcpy(buf, &args[0], sizeof(buf));
if (!strcmp(buf, "auto"))
user = false;
else if (!strcmp(buf, "user"))
user = true;
else
goto einval;
continue;
}
tok = match_token(p, qos_tokens, args);
switch (tok) {
case QOS_RPPM:
case QOS_WPPM:
if (match_strlcpy(buf, &args[0], sizeof(buf)) >=
sizeof(buf))
goto einval;
if (cgroup_parse_float(buf, 2, &v))
goto einval;
if (v < 0 || v > 10000)
goto einval;
qos[tok] = v * 100;
break;
case QOS_RLAT:
case QOS_WLAT:
if (match_u64(&args[0], &v))
goto einval;
qos[tok] = v;
break;
case QOS_MIN:
case QOS_MAX:
if (match_strlcpy(buf, &args[0], sizeof(buf)) >=
sizeof(buf))
goto einval;
if (cgroup_parse_float(buf, 2, &v))
goto einval;
if (v < 0)
goto einval;
qos[tok] = clamp_t(s64, v * 100,
VRATE_MIN_PPM, VRATE_MAX_PPM);
break;
default:
goto einval;
}
user = true;
}
if (qos[QOS_MIN] > qos[QOS_MAX])
goto einval;
spin_lock_irq(&ioc->lock);
if (enable) {
blk_stat_enable_accounting(ioc->rqos.q);
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
blk_queue_flag_set(QUEUE_FLAG_RQ_ALLOC_TIME, ioc->rqos.q);
ioc->enabled = true;
} else {
blk_queue_flag_clear(QUEUE_FLAG_RQ_ALLOC_TIME, ioc->rqos.q);
ioc->enabled = false;
}
if (user) {
memcpy(ioc->params.qos, qos, sizeof(qos));
ioc->user_qos_params = true;
} else {
ioc->user_qos_params = false;
}
ioc_refresh_params(ioc, true);
spin_unlock_irq(&ioc->lock);
put_disk_and_module(disk);
return nbytes;
einval:
ret = -EINVAL;
err:
put_disk_and_module(disk);
return ret;
}
static u64 ioc_cost_model_prfill(struct seq_file *sf,
struct blkg_policy_data *pd, int off)
{
const char *dname = blkg_dev_name(pd->blkg);
struct ioc *ioc = pd_to_iocg(pd)->ioc;
u64 *u = ioc->params.i_lcoefs;
if (!dname)
return 0;
seq_printf(sf, "%s ctrl=%s model=linear "
"rbps=%llu rseqiops=%llu rrandiops=%llu "
"wbps=%llu wseqiops=%llu wrandiops=%llu\n",
dname, ioc->user_cost_model ? "user" : "auto",
u[I_LCOEF_RBPS], u[I_LCOEF_RSEQIOPS], u[I_LCOEF_RRANDIOPS],
u[I_LCOEF_WBPS], u[I_LCOEF_WSEQIOPS], u[I_LCOEF_WRANDIOPS]);
return 0;
}
static int ioc_cost_model_show(struct seq_file *sf, void *v)
{
struct blkcg *blkcg = css_to_blkcg(seq_css(sf));
blkcg_print_blkgs(sf, blkcg, ioc_cost_model_prfill,
&blkcg_policy_iocost, seq_cft(sf)->private, false);
return 0;
}
static const match_table_t cost_ctrl_tokens = {
{ COST_CTRL, "ctrl=%s" },
{ COST_MODEL, "model=%s" },
{ NR_COST_CTRL_PARAMS, NULL },
};
static const match_table_t i_lcoef_tokens = {
{ I_LCOEF_RBPS, "rbps=%u" },
{ I_LCOEF_RSEQIOPS, "rseqiops=%u" },
{ I_LCOEF_RRANDIOPS, "rrandiops=%u" },
{ I_LCOEF_WBPS, "wbps=%u" },
{ I_LCOEF_WSEQIOPS, "wseqiops=%u" },
{ I_LCOEF_WRANDIOPS, "wrandiops=%u" },
{ NR_I_LCOEFS, NULL },
};
static ssize_t ioc_cost_model_write(struct kernfs_open_file *of, char *input,
size_t nbytes, loff_t off)
{
struct gendisk *disk;
struct ioc *ioc;
u64 u[NR_I_LCOEFS];
bool user;
char *p;
int ret;
disk = blkcg_conf_get_disk(&input);
if (IS_ERR(disk))
return PTR_ERR(disk);
ioc = q_to_ioc(disk->queue);
if (!ioc) {
ret = blk_iocost_init(disk->queue);
if (ret)
goto err;
ioc = q_to_ioc(disk->queue);
}
spin_lock_irq(&ioc->lock);
memcpy(u, ioc->params.i_lcoefs, sizeof(u));
user = ioc->user_cost_model;
spin_unlock_irq(&ioc->lock);
while ((p = strsep(&input, " \t\n"))) {
substring_t args[MAX_OPT_ARGS];
char buf[32];
int tok;
u64 v;
if (!*p)
continue;
switch (match_token(p, cost_ctrl_tokens, args)) {
case COST_CTRL:
match_strlcpy(buf, &args[0], sizeof(buf));
if (!strcmp(buf, "auto"))
user = false;
else if (!strcmp(buf, "user"))
user = true;
else
goto einval;
continue;
case COST_MODEL:
match_strlcpy(buf, &args[0], sizeof(buf));
if (strcmp(buf, "linear"))
goto einval;
continue;
}
tok = match_token(p, i_lcoef_tokens, args);
if (tok == NR_I_LCOEFS)
goto einval;
if (match_u64(&args[0], &v))
goto einval;
u[tok] = v;
user = true;
}
spin_lock_irq(&ioc->lock);
if (user) {
memcpy(ioc->params.i_lcoefs, u, sizeof(u));
ioc->user_cost_model = true;
} else {
ioc->user_cost_model = false;
}
ioc_refresh_params(ioc, true);
spin_unlock_irq(&ioc->lock);
put_disk_and_module(disk);
return nbytes;
einval:
ret = -EINVAL;
err:
put_disk_and_module(disk);
return ret;
}
static struct cftype ioc_files[] = {
{
.name = "weight",
.flags = CFTYPE_NOT_ON_ROOT,
.seq_show = ioc_weight_show,
.write = ioc_weight_write,
},
{
.name = "cost.qos",
.flags = CFTYPE_ONLY_ON_ROOT,
.seq_show = ioc_qos_show,
.write = ioc_qos_write,
},
{
.name = "cost.model",
.flags = CFTYPE_ONLY_ON_ROOT,
.seq_show = ioc_cost_model_show,
.write = ioc_cost_model_write,
},
{}
};
static struct blkcg_policy blkcg_policy_iocost = {
.dfl_cftypes = ioc_files,
.cpd_alloc_fn = ioc_cpd_alloc,
.cpd_free_fn = ioc_cpd_free,
.pd_alloc_fn = ioc_pd_alloc,
.pd_init_fn = ioc_pd_init,
.pd_free_fn = ioc_pd_free,
.pd_stat_fn = ioc_pd_stat,
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
};
static int __init ioc_init(void)
{
return blkcg_policy_register(&blkcg_policy_iocost);
}
static void __exit ioc_exit(void)
{
blkcg_policy_unregister(&blkcg_policy_iocost);
blkcg: implement blk-iocost This patchset implements IO cost model based work-conserving proportional controller. While io.latency provides the capability to comprehensively prioritize and protect IOs depending on the cgroups, its protection is binary - the lowest latency target cgroup which is suffering is protected at the cost of all others. In many use cases including stacking multiple workload containers in a single system, it's necessary to distribute IO capacity with better granularity. One challenge of controlling IO resources is the lack of trivially observable cost metric. The most common metrics - bandwidth and iops - can be off by orders of magnitude depending on the device type and IO pattern. However, the cost isn't a complete mystery. Given several key attributes, we can make fairly reliable predictions on how expensive a given stream of IOs would be, at least compared to other IO patterns. The function which determines the cost of a given IO is the IO cost model for the device. This controller distributes IO capacity based on the costs estimated by such model. The more accurate the cost model the better but the controller adapts based on IO completion latency and as long as the relative costs across differents IO patterns are consistent and sensible, it'll adapt to the actual performance of the device. Currently, the only implemented cost model is a simple linear one with a few sets of default parameters for different classes of device. This covers most common devices reasonably well. All the infrastructure to tune and add different cost models is already in place and a later patch will also allow using bpf progs for cost models. Please see the top comment in blk-iocost.c and documentation for more details. v2: Rebased on top of RQ_ALLOC_TIME changes and folded in Rik's fix for a divide-by-zero bug in current_hweight() triggered by zero inuse_sum. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Andy Newell <newella@fb.com> Cc: Josef Bacik <jbacik@fb.com> Cc: Rik van Riel <riel@surriel.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2019-08-28 22:05:58 +00:00
}
module_init(ioc_init);
module_exit(ioc_exit);