sched_ext is about to be merged upstream. There are some compatibility
breaking changes and we're making the current sched_ext/for-6.11
1edab907b57d ("sched_ext/scx_qmap: Pick idle CPU for direct dispatch on
!wakeup enqueues") the baseline.
Tag everything except scx_mitosis as 1.0.0. As scx_mitosis is still in early
development and is currently temporarily disabled, only the patchlevel is
bumped.
The stats map in scx_rusty is a BPF_MAP_TYPE_PERCPU_ARRAY, with its size
determined by num_possible_cpus(). Initializing it with nr_cpu_ids() can
result in errors such as:
Error: Failed to zero stat
Caused by:
number of values 6 != number of cpus 8
Fix by using num_possible_cpus() to initialize it.
Fixes: 263e02f6 ("rusty: Use nr_cpu_ids instead of nr_cpus_possible")
Signed-off-by: Andrea Righi <andrea.righi@canonical.com>
With commit 5d20f89a ("scheds-rust: build rust schedulers in sequence"),
schedulers are now built serially one after the other to prevent meson
and cargo from forking NxN parallel tasks.
However, this change has made building a single scheduler much more
cumbersome, due to the chain of dependencies.
For example, building scx_rusty using the specific meson target would
still result in all schedulers being built, because they all depend on
each other.
To address this issue, introduce the new meson build option
`serialize=true|false` (default is false).
This option allows to disable the schedulers' build chain, restoring the
old behavior.
With this option enabled, it is now possible to build just a single
scheduler, parallelizing the cargo build properly, without triggering
the build of the others. Example:
$ meson setup build -Dbuildtype=release -Dserialize=false
$ meson compile -C build scx_rusty
Signed-off-by: Andrea Righi <andrea.righi@canonical.com>
When someone is testing schedulers, we often have to ask what version
the scheduler is running as. Now that we can access the build ID from
rust schedulers, let's update scx_rusty to print the build ID when rusty
first starts running.
This results in output such as the following:
```
[void@maniforge scx]$ rusty
19:04:26 [INFO] Running scx_rusty (build ID: 0.8.1-g2043d2537f37c8d75753bb65eb75bca965067564 x86_64-unknown-linux-gnu/debug)
19:04:26 [INFO] NUMA[00] mask= 0b11111111111111111111111111111111
19:04:26 [INFO] DOM[00] mask= 0b00000000111111110000000011111111
19:04:26 [INFO] DOM[01] mask= 0b11111111000000001111111100000000
19:04:26 [INFO] Rusty scheduler started!
```
Signed-off-by: David Vernet <void@manifault.com>
This change adds a new module to the scx_utils crate that provides a
log recorder for metrics-rs. The log recorder will log all metrics to
the console at a configurable interval in an easy to read format. Each
metric type will be displayed in a separate section. Indentation will
be used to show the hierarchy of the metrics. This results in a more
verbose output, but it is easier to read and understand.
scx_rusty was updated to use the log recorder and all explicit metric
logging was removed.
Counters will show the total count and the rate of change per second.
Counters with an additional label, like `type` in
`dispatched_tasks_total` in rusty, will show the count, rate, and
percentage of the total count.
Counters:
dispatched_tasks_total: 65559 [1344.8/s]
prev_idle: 44963 (68.6%) [966.5/s]
wsync_prev_idle: 15696 (23.9%) [317.3/s]
direct_dispatch: 2833 (4.3%) [35.3/s]
dsq: 1804 (2.8%) [21.3/s]
wsync: 262 (0.4%) [4.3/s]
direct_greedy: 1 (0.0%) [0.0/s]
pinned: 0 (0.0%) [0.0/s]
greedy_idle: 0 (0.0%) [0.0/s]
greedy_xnuma: 0 (0.0%) [0.0/s]
direct_greedy_far: 0 (0.0%) [0.0/s]
greedy_local: 0 (0.0%) [0.0/s]
dl_clamped_total: 1290 [20.3/s]
dl_preset_total: 514 [1.0/s]
kick_greedy_total: 6 [0.3/s]
lb_data_errors_total: 0 [0.0/s]
load_balance_total: 0 [0.0/s]
repatriate_total: 0 [0.0/s]
task_errors_total: 0 [0.0/s]
Gauges will show the last set value:
Gauges:
slice_length_us: 20000.00
Histograms will show the average, min, and max. The histogram will be
reset after each log interval to avoid memory leaks, since the data
structure that holds the samples is unbounded.
Histograms:
cpu_busy_pct: avg=1.66 min=1.16 max=2.16
load_avg node=0: avg=0.31 min=0.23 max=0.39
load_avg node=0 dom=0: avg=0.31 min=0.23 max=0.39
processing_duration_us: avg=297.50 min=296.00 max=299.00
Signed-off-by: Jose Fernandez <josef@netflix.com>
Origin assignment of the variable ridx is equivalent to comparing
between "ridx" and "wids - MAX_PIDS". Using u64 max library helper
function to perform the comparison and provide better readability.
Signed-off-by: I Hsin Cheng <richard120310@gmail.com>
Check whether the BalanceState of pull_dom.load inside function
try_find_move_task is actually the variant NeedsPull. It'll perform task
migration in abit more conservative manner when the system is under high
loading situation.
Experiments are performed when the system is compiling linux kernel and
undergoing a large amount of I/O operation at the same time using fio.
The result showns that before the modification, there're 12,6617 times
of task migrations system wide. After the modification, there're 11,5419
times of task migrations system wide.
Signed-off-by: I Hsin Cheng <richard120310@gmail.com>
In scx_rusty, we're currently using topo.nr_cpus_possible() to determine
how many possible CPU IDs we could have on the system. scx_rusty already
accounts for offlined CPUs, so to properly support systems whose
disabled CPUs may be in the middle of the range of possible CPU IDs,
let's instead use topo.nr_cpu_ids().
Signed-off-by: David Vernet <void@manifault.com>
We need a layer of indirection between the stats collection and their
output destinations. Currently, stats are only printed to stdout. Our
goal is to integrate with various telemetry systems such as Prometheus,
StatsD, and custom metric backends like those used by Meta and Netflix.
Importantly, adding a new backend should not require changes to the
existing stats code.
This patch introduces the `metrics` [1] crate, which provides a
framework for defining metrics and publishing them to different
backends.
The initial implementation includes the `dispatched_tasks_count`
metric, tagged with `type`. This metric increments every time a task is
dispatched, emitting the raw count instead of a percentage. A monotonic
counter is the most suitable metric type for this use case, as
percentages can be calculated at query time if needed. Existing logged
metrics continue to print percentages and remain unchanged.
A new flag, `--enable-prometheus`, has been added. When enabled, it
starts a Prometheus endpoint on port 9000 (default is false). This
endpoint allows metrics to be charted in Prometheus or Grafana
dashboards.
Future changes will migrate additional stats to this framework and add
support for other backends.
[1] https://metrics.rs/
Signed-off-by: Jose Fernandez <josef@netflix.com>
Modify the execution sequence before lookup operation for new_domc. If
new_dom_id == NO_DOM_FOUND, lookup operation for new_domc is definitely
going to fail so we don't have to wait until we found that new_domc is
NULL, clearing of cpumask and return operation should be done directly
in that case.
Plus we should avoid using try_lookup_dom_ctx outside the context of
lookup_dom_ctx, as it can keep the interface's consistency.
Signed-off-by: I Hsin Cheng <richard120310@gmail.com>
The rusty dispatch logic is a bit unnecessarily convoluted. Let's clean it up
so that we're just comparing dom ids rather than iterating over arrays nested
inside of pcpu context.
Signed-off-by: David Vernet <void@manifault.com>
Right now, the SCX_WAKE_SYNC logic in rusty is very primitive. We only check to
see if the waker CPU's runqueue is empty, and then migrate the wakee there if
so. We'll want to expand this to be more thorough, such as:
- Checking to see if prev_cpu and waker_cpu share the same LLC when determining
where to migrate
- Check for whether SCX_WAKE_SYNC migration helps load imbalance between cores
- ...
Right now all of that code is just a big blob in the middle of
rusty_select_cpu(). Let's pull it into its own function to improve readability,
and also add some logic to stay on prev_cpu if it shares an LLC with the waker.
Signed-off-by: David Vernet <void@manifault.com>
It seems that task_set_domain() is nearly at the point where it can
cause the verifier to get confused and think that it's exceeding the
number of available instructions per program. I've seen this a number of
times when making small changes to task_set_domain(), and it's once
again happened @vax-r (I-Hsin Cheng) made a small cleanup change to
rusty in https://github.com/sched-ext/scx/pull/362.
To avoid this, let's just make dom_xfer_task() a separate global program
so that the verifier doens't have to worry about branch pruning, etc
depending on what the caller does. This should hopefully make
task_set_domain() (and its callers) much less brittle.
Signed-off-by: David Vernet <void@manifault.com>
In preparation of upstreaming, let's set the min version requirement at the
released v6.9 kernels. Drop __COMPAT_scx_bpf_switch_call(). The open helper
macros now check the existence of SCX_OPS_SWITCH_PARTIAL and abort if not.
The bpf_ prefix is used for BPF API. Rename bpf_log2() to u32_log2() and
bpf_log2l() to u64_log2(). While at it, relocate them below compiler
directive helpers.
READ_ONCE()/WRITE_ONCE() macros are added in commit 0932fde, we should
be able to utilize the macros to get around the possibility of data
races for domc->min_vruntime.
Signed-off-by: I Hsin Cheng <richard120310@gmail.com>
Make restart handling with user_exit_info simpler and consistently use the
load and report macros consistently across the rust schedulers. This makes
all schedulers automatically handle auto restarts from CPU hotplug events.
Note that this is necessary even for scx_lavd which has CPU hotplug
operations as CPU hotplug operations which took place between skel open and
scheduler init can still trigger restart.
In cpumask_intersects_domain(), we check whether a given cpumask has any
CPUs in common with the specified domain by looking at the const, static
dom_cpumasks map. This map is only really necessary when creating the
domain struct bpf_cpumask objects at scheduler load time. After that, we
can just use the actual struct bpf_cpumask object embedded in the domain
context. Let's use that and cpumask kfuncs instead.
This allows rusty to load with
https://github.com/sched-ext/sched_ext/pull/216.
Signed-off-by: David Vernet <void@manifault.com>
When building with warnings enabled, a few obvious bugs are pointed out:
- We're not correctly calculating waker frequency
- We're not taking the min of avg_run_raw compared to max latency
- We're missing an element from sched_prio_to_weight
Fix these. With these changes, interactivity is seemingly improved. We
go from ~12 sec / turn -> 11 seconds / turn in the Civ 6 AI benchmark
with a 4 x nproc CPU hogging workload in the background. It's clear,
however, that we really need preemption.
Signed-off-by: David Vernet <void@manifault.com>
C SCX_OPS_ATTACH() and rust scx_ops_attach() macros were not calling
.attach() and were only attaching the struct_ops. This meant that all
non-struct_ops BPF programs contained in the skels were never attached which
breaks e.g. scx_layered.
Let's fix it by adding .attach() invocation the the attach macros.
Now that the scx_ops_open!() macro is available, let's use it in scx_rusty to
cover all cases of when hotplug can happen.
Signed-off-by: David Vernet <void@manifault.com>
Now that the kernel exports the SCX_ECODE_ACT_RESTART exit code, we can
remove the custom hotplug logic from scx_rusty, and instead rely on the
built-in logic from the kernel. There's still a corner case that we're not
honoring: when a hotplug event happens on the init path. A future change will
address this as well.
Signed-off-by: David Vernet <void@manifault.com>
scx_rusty's intention is to support hotplug by automatically restarting
whenever a hotplug event is encountered. Now that we're not trying to
consume a bogus DSQ in the rusty_dispatch() on a newly hotplugged CPU,
let's just remove offline tracking. It's really just there as a sanity
check, but it triggers if an offline task is made runnable during a
hotplug event before the ops.hotplug() callback has been invoked.
Signed-off-by: David Vernet <void@manifault.com>
There's currently a slight issue on existing kernels on the hotplug
path wherein we can start to receive scheduling callbacks on a CPU
before that CPU has received hotplug events. For CPUs going online, this
can possibly confuse a scheduler because it may not be expecting
anything to ever happen on that CPU, and therefore may do things that
could cause the scheduler to crash. For example, without this patch in
scx_rusty, we try to consume from a bogus DSQ that doesn't exist, which
causes ext.c to boot out the scheduler.
Though this issue will soon be fixed in ext.c, let's explicitly avoid
dispatching from an onlining CPU in rusty so that we properly support
hotplug on older kernels as well.
Signed-off-by: David Vernet <void@manifault.com>
scx_lavd implemented 32 and 64 bit versions of a base-2 logarithm
function. This is now also used in rusty. To avoid code duplication,
let's pull it into a shared header.
Note that there is technically a functional change here as we remove the
always inline compiler directive. We instead assume that the compiler
will know best whether or not to inline the function.
Signed-off-by: David Vernet <void@manifault.com>
In user space in rusty, the tuner detects system utilization, and uses
it to inform how we do load balancing, our greedy / direct cpumasks,
etc. Something else we could be doing but currently aren't, is using
system utilization to inform how we dispatch tasks. We currently have a
static, unchanging slice length for the runtime of the program, but this
is inefficient for all scenarios.
Giving a task a long slice length does have advantages, such as
decreasing the number of involuntary context switches, decreasing the
overhead of preemption by doing it less frequently, possibly getting
better cache locality due to a task running on a CPU for a longer amount
of time, etc. On the other hand, long slices can be problematic as well.
When a system is highly utilized, a CPU-hogging task running for too
long can harm interactive tasks. When the system is under-utilized,
those interactive tasks can likely find an idle, or under-utilized core
to run on. When the system is over-utilized, however, they're likely to
have to park in a runqueue.
Thus, in order to better accommodate such scenarios, this patch
implements a rudimentary slice scaling mechanism in scx_rusty. Rather
than having one global, static slice length, we instead have a dynamic,
global slice length that can be changed depending on system utilization.
When over-utilized, we go with a longer slice length, and vice versa for
when the system is under-utilized. With Terraria, this results in
roughly a 50% improvement in mean FPS when playing on an AMD Ryzen 9
7950X, while running Spotify, and stress-ng -c $((4 * $(nproc))).
Signed-off-by: David Vernet <void@manifault.com>
scx_rusty doesn't do terribly well with interactive workloads. In order
to improve the situation, this patch adds support for basic deadline
scheduling in rusty. This approach doesn't incorporate eligibility, and
simply uses a crude avg_runtime tracking approach to scaling a task's
deadline.
In a series of follow-on changes, we'll update the scheduler to use more
indicators for interactivity that affect both slice length, and deadline
calculation.
Signed-off-by: David Vernet <void@manifault.com>
Let's remove the extraneous copy pasting and use a lookup helper like we
do for task and pcpu context.
Signed-off-by: David Vernet <void@manifault.com>
A LoadEntity gets the load to transfer between two entities by taking
the minimum of their imbalances and reducing its abs value by
xfer_ratio.
In practice self.imbal(), the push node or domain, always has positive
imbalance and other.imbal(), the pull node or domain, always has
negative imbalance, so other.imbal() is always the minimum even though
the abs value of its imbalance might be greater than the abs value of
self.imbal(). It seems like the intent is to take the minimum of the
two absolute values instead to avoid overbalancing at the puller, so
make both values abs.
Signed-off-by: Daniel Jordan <daniel.m.jordan@oracle.com>
Rusty's load balancer calculates load differently based on average
system CPU utilization in create_domain_hierarchy(). At >= 99.999%
utilization, load is the product of a task's weight and duty cycle;
below that, load is the same as the task's duty cycle.
populate_tasks_by_load(), however, always uses the product when
calculating per-task load so that in the sub-99.999% util case, load is
inflated, typically by a factor of 100 with a normal priority task.
Tasks look too heavy to migrate as a result because a single task would
transfer more load than the domain imbalance allows, leading to
significant imbalance in some cases.
Make populate_tasks_by_load() calculate task load the same way as
domain load, checking lb_apply_weight.
Signed-off-by: Daniel Jordan <daniel.m.jordan@oracle.com>
We're currently cloning cpumasks returned by calls to {Core, Cache,
Node, Topology}::span(). If a caller needs to clone it, they can. Let's
not penalize the callers that just want to query the underlying cpumask.
Signed-off-by: David Vernet <void@manifault.com>
We have a lot of boilerplate code where we create a cpumask, initialize
it, and then bpf_kptr_xchg() it into the map. In an effort to slightly
reduce the amount of boilerplate, let's create a helper that can
alleviate some of it.
Signed-off-by: David Vernet <void@manifault.com>
There are some random issues in the code, like unused variables, and bad
print formatters. I'm not sure why the compiler isn't consistently
complaining, but let's fix them.
Signed-off-by: David Vernet <void@manifault.com>
In scx_rusty, now that we have a complete view of the host's topology
thanks to the Topology crate, we can update our calls to
scx_bpf_create_dsq() to create the DSQ on the NUMA node of the domain.
It's unclear how much this will end up mattering for performance in the
typical case, but we might as well do the right thing given that host
topolgoy is static, and we have the information.
Signed-off-by: David Vernet <void@manifault.com>
The scx_rusty scheduler does not support hotplug, and expects a static
host topology throughout its runtime. Though the kernel does have
support for detecting hotplug events, we currently don't detect this in
the kernel, nor surface it to user space when it happens. Now that we
have scx_bpf_exit(), we can gracefully exit the kernel in the event of a
hotplug, and communicate to user space that it should restart the
scheduler.
This patch adds that support to scx_rusty. Note that this assumes that
we're running on a recent enough kernel that has scx_bpf_exit(). If it
doesn't, then we instead just error out of the kernel scheduler and exit
the application.
Signed-off-by: David Vernet <void@manifault.com>
In rusty_select_cpu(), if a task is WAKE_SYNC, we'll currently migrate
the task to that CPU if there are any idle cores on the system. As in
[0], this condition is insufficient, as there could be idle cores
elsewhere on the system, but still tasks piled up on a single local DSQ.
Let's add a condition that the local DSQ has to be empty in order to
apply the WAKE_SYNC migration.
Before patch:
[void@maniforge src]$ hackbench
Running in process mode with 10 groups using 40 file descriptors each (== 400 tasks)
Each sender will pass 100 messages of 100 bytes
Time: 0.433
With patch:
[void@maniforge src]$ hackbench
Running in process mode with 10 groups using 40 file descriptors each (== 400 tasks)
Each sender will pass 100 messages of 100 bytes
Time: 0.035
Signed-off-by: David Vernet <void@manifault.com>
There were a few issues, e.g. us still mentioning the infeasible weights
problem, and arguments using underscores despite clap rendering them
with dashes. Let's fix them up.
Signed-off-by: David Vernet <void@manifault.com>
As described in https://bugzilla.kernel.org/show_bug.cgi?id=218109,
https://github.com/sched-ext/scx/issues/147 and
https://github.com/sched-ext/sched_ext/issues/69, AMD chips can
sometimes report fully disabled CPUs as offline, which causes us to
count them when looking at /sys/devices/system/cpu/possible.
Additionally, systems can have holes in their active CPU maps. For
example, a system with CPUs 0, 1, 2, 3 possible, may have only 0 and 2
active. To address this, we need to do a few things:
1. Update topology.rs to be clear that it's returning the number of
_possible_ CPUs in the system. Also update Topology to only record
online CPUs when creating its span and iterating over sysfs when
creating domains. It was previously trying to record when a CPU was
online, but this was actually broken as the topology directory isn't
present in sysfs when the CPU is offline.
2. Schedulers should not be relying on nr_possible_cpus for anything
other than interacting with per-CPU data (e.g. for stats extraction),
or e.g. verifying maximum sizes of statically sized arrays in BPF. It
should _not_ be used for e.g. performing load calculations, etc. With
that said, we'll also need to update schedulers to not rely on the
nr_possible_cpus figure being exported by the topology crate. We do
that for rusty in this patch, but don't fix any of the others other
than updating how they call topology.rs.
3. Account for the fact that LLC IDs may be non-contiguous. For example,
if there is a single core in an LLC, then if we assign LLC IDs to
domains, then the domain IDs won't be contiguous. This doesn't fit
our current model which is used by e.g. infeasible_weights.rs. We'll
update some of the code in rusty to accomodate this, but we'll need
to do more.
4. Update schedulers to properly reset themselves in the event of a
hotplug event. We'll take care of that in a follow-on change.
Signed-off-by: David Vernet <void@manifault.com>
If a CPU is offline, it could cause an LLC to go offline, which could
cause us to have non-contiguous domain IDs. Right now, a few places in
code assume contiguous domain IDs, such as in the infeasible weights
crate. Let's update domain.rs and load_balaance.rs to do the right
thing. We'll fix the others later.
Signed-off-by: David Vernet <void@manifault.com>
We implement functions or(), and(), and xor() for cpumasks, but we
should also implement the bitwise ops for those operations in case
people prefer that syntax.
Signed-off-by: David Vernet <void@manifault.com>
Given the complexity of migrating load between nodes (we're doing four
nested loops), we should add a comment explaining what we're doing. This
commit does that. In addition, we use a VecDeque to store (and then
restore) push nodes and push domains so that we can re-add them to their
respective lists in load-sorted order rather than reverse-load-sorted
order. This allows us to avoid having to do unnecessary right-shifts
every time a push object is re-added to its containing list.
Signed-off-by: David Vernet <void@manifault.com>
Fixing alignment, moving a couple bail! calls around, and adding a
missing break from move_between_nodes() that lets us bail out of a loop
early.
Signed-off-by: David Vernet <void@manifault.com>
As Tejun pointed out in review, the disadvantage of using
push/pull/balanced lists is that if the domains inside the nodes are
balanced, we won't be able to push load between them. I'd originally
done it that way both as an optimization, but also to allow me to
iterate over the lists of pushable and pullable domains mutably. That
was addressed in the prior commit, but the nodes themselves were still
put into 3 buckets.
I think this is generally just a cleaner way of doing things, so let's
just collapse the nodes into a flat list as well. This prevents us from
having to coalesce the lists, std::mem::swap them, etc.
Signed-off-by: David Vernet <void@manifault.com>
Tejun pointed out that a possible issue exists in the current
implementation, wherein if you have two NUMA nodes that are imbalanced,
but their domains are internally balanced, we'll fail to migrate between
them if all nodes are in the balanced_nodes list.
To address this, let's just use a single global list for all types of
domains, and do checking internally for imbalances. The reason it was
done this way in the first place was to allow me to mutably iterate over
both vectors in a nested loop. The way around that is to just use loop
{} and push/pop domains from the list.
We could do the same thing for the NUMA nodes themselves, which are also
in 3 separate lists in the LoadBalancer. We'll do that in a subsequent
commit.
Signed-off-by: David Vernet <void@manifault.com>
In scx_rusty, a CPU that is going to go idle will attempt to steal tasks
from remote domains when its domain has no tasks to run, and a remote
domain has at least greedy_threshold enqueued tasks. This stealing is
temporary, but of course has a cost in that the CPU that's stealing the
task may cause it to suffer from cache misses, or in the case of
multi-node machines, remote NUMA accesses and working sets split across
multiple domains.
Given the higher cost of x NUMA work stealing, let's add a separate flag
that lets users tune the threshold for doing cross NUMA greedy task
stealing.
Signed-off-by: David Vernet <void@manifault.com>
We removed the debug!() output that was previously present in main.rs. Let's
add more debug!() output that helps debug the current LB hierarchy.
Signed-off-by: David Vernet <void@manifault.com>
The scx_rusty load balancer is currently no longer exporting statistics such as
domain load averages, load sums, etc. Now that we're also balancing by NUMA,
we'll need a way to hierarchically illustrate load balancing statistics. This
patch adds support for that.
Signed-off-by: David Vernet <void@manifault.com>
updating stats printing
Signed-off-by: David Vernet <void@manifault.com>
Users may want to toggle whether tasks can be temporarily sent to idle CPUs on
remote NUMA nodes. By default, we want it to be disabled as a task spanning
multiple NUMA nodes will end up having its working set spanning both nodes,
which is probably not desirable. However, in case a workload really wants to
encourage work conservation, let's add a flag that allows them to toggle it.
Signed-off-by: David Vernet <void@manifault.com>
scx_rusty currently pushes tasks to idle cores if the direct greedy threshold
is exceeded, even if the core is on a remote NUMA node. This behavior is
probably not desired in most scenarios. The worst that will happen if a task is
pushed to an idle core in the same node is some L3 cache miss traffic, but for
multiple NUMA nodes, it could cause the task to have its working set span
multiple nodes.
Let's disable direct greedy work stealing across NUMA nodes. A future commit
will add a flag that's disabled by default, and let's users turn this on if
they really want to encourage work conservation.
Signed-off-by: David Vernet <void@manifault.com>
Right now, scx_rusty has no notion of domains spanning NUMA nodes, and makes no
distinction when making load balancing decisions, or work stealing. This can
cause problems on multi-NUMA machines, as load balancing and work stealing
across NUMA nodes has significantly different cost from across L3 cache
boundaries.
In order to better support multi-NUMA machines, this commit adds another layer
to the rusty load balancer, which balances across NUMA nodes using a different
cost function from balancing across domains. Load balancing now takes place
over the span of two passes:
1. In the first pass, we fix imbalances across NUMA nodes by moving tasks
between domains across those NUMA node boundaries. We require a load
imbalance of at least 17% in order to move load at this stage. The ratio of
load imbalance we attempt to adjust (50%) and the maximum amount of load
we're allowed to push out of a domain (50%) is still the same as when
balancing between domains inside a NUMA node, but this is easy to tune with
the current setup.
2. Once we've balanced across NUMA nodes, we iterate over all nodes and balance
between the domains within each NUMA node. The cost function here is the
same as what it has been thus far: we require at least a 5% imbalance in
order to trigger load balancing.
There are a few additional changes / improvements to load balancing in this
commit:
1. NUMA nodes and domains are now ordered according to their load by using
SortedVec objects. We were previously using BTreeMap keyed by load, but this
was suboptimal due to the fact that it doesn't allow duplicate entries.
2. We're no longer exporting load balancing statistics as a vector of data such
as load sums, averages, and imbalances. This is instead all encapsulated in
the load balancing hierarchy we setup in lb.load_balance(). These statistics
are not yet exported, but they will be in a subsequent commit.
One of the issues with this commit is that it does introduce some
almost-identical logic that somehow begs to be deduplicated. For example, when
we balance between NUMA nodes, the logic for iterating over push nodes and
pushing to pull nodes is very similar to the logic of iterating over push
domains and pull domains when balancing within a node. It may be that this can
be improved.
The following are some benchmarks run on an Intel Xeon Gold 6138 (2 x 40 core
processor):
kcompile
--------
On Commit a27648c74210 ("afs: Fix setting of mtime when creating a
file/dir/symlink"):
1. make allyesconfig
2. make -j $(nproc) built-in.a
3. make -j clean
4. goto 2
Runtime
-------
o-----------o-----------o----------o
| scx_rusty | CFS | Delta |
---------o-----------o-----------o----------o
Mean | 562.688s | 566.085s | -.6% |
---------o-----------o-----------o----------o
Variance | 0.54387 | 0.72431 | -24.9% |
---------o-----------o-----------o----------o
o-----------o-----------o----------o
| rusty NUMA| rusty ORIG| Delta |
---------o-----------o-----------o----------o
Mean | 562.688s | 563.209s | -.092% |
---------o-----------o-----------o----------o
Variance | 0.54387 | 0.42038 | 29.38% |
---------o-----------o-----------o----------o
scx_rusty with NUMA awareness clearly beats CFS, but only barely beats
scx_rusty without it. This isn't necessarily super surprising given that
this is kcompile, which has very poor front-end CPU locality. Further
experimentation with toggling the cost function for performing
migrations may improve this further.
CPU util
--------
o-----------o-----------o----------o
| scx_rusty | CFS | Delta |
---------o-----------o-----------o----------o
Mean | 7654.25% | 7551.67% | 1.11% |
---------o-----------o-----------o----------o
Variance | 165.35714 | 158.3333 | 4.436% |
---------o-----------o-----------o----------o
o-----------o-----------o----------o
| rusty NUMA| rusty ORIG| Delta |
---------o-----------o-----------o----------o
Mean | 7654.25% | 7641.57% | 0.1659% |
---------o-----------o-----------o----------o
Variance | 165.35714 | 1230.619 | -86.5% |
---------o-----------o-----------o----------o
As expected, CPU util is quite a bit higher with scx_rusty than it is
with CFS. Further experiments that could be interesting are always
enabling direct-greedy stealing between domains within a NUMA node, and
then comparing rusty NUMA and rusty ORIG. rusty NUMA prevents stealing
between NUMA nodes, so this would show whether the locality introduced
by NUMA awareness appropriately offsets the loss of work conservation.
Major PFs
---------
o-----------o-----------o----------o
| scx_rusty | CFS | Delta |
---------o-----------o-----------o----------o
Mean | 5332 | 3950 | 36.566% |
---------o-----------o-----------o----------o
Variance | 6975.5 | 5986.333 | 16.5237% |
---------o-----------o-----------o----------o
o-----------o-----------o----------o
| rusty NUMA| rusty ORIG| Delta |
---------o-----------o-----------o----------o
Mean | 5332 | 5336.5 | -.084% |
---------o-----------o-----------o----------o
Variance | 6975.5 | 955.5 | 630.03% |
---------o-----------o-----------o----------o
Also as expected, major page faults are far highe higher with scx_rusty
than with CFS. This is expected even with NUMA awareness, given that
scx_rusty is still less sticky than CFS.
Further experiments that could be interesting are tuning the threshold
for which we perform x NUMA migrations to try and keep this value even
lower. The rate of major page faults between rusty NUMA and rusty ORIG
were very close, though rusty NUMA was a bit lower.
Signed-off-by: David Vernet <void@manifault.com>
More cleanup of scx_rusty. Let's move the LoadBalancer out of rusty.rs and into
its own file. It will soon be extended quite a bit to support multi-NUMA and
other multivariate LB cost functions, so it's time to clean things up and split
it out.
Signed-off-by: David Vernet <void@manifault.com>
rusty.rs is growing a bit unwieldy. We're going to want to update its load
balancing logic somewhat significantly to account for multi-NUMA and other cost
functions, so let's start cleaning the code up so that things are more
logically segmented and easier to work with.
To start, we move the Tuner and DomainGroup/Domain objects into their own
modules.
Signed-off-by: David Vernet <void@manifault.com>
This is to potentinally reduce issues with folks
using different versions of libbpf at runtime.
This also:
- makes static linking of libbpf the default
- adds steps in `meson setup` to fetch libbpf and make it
Now that we have a new 'infeasible' crate that abstracts the logic for
implementing the infeasible weights solution. Let's update rusty to use
it.
Signed-off-by: David Vernet <void@manifault.com>
scx_rusty has logic in the scheduler to inspect the host to
automatically build scheduling domains across every L3 cache. This would
be generically useful for many different types of schedulers, so let's
add it to the scx_utils crate so it can be used by others.
Signed-off-by: David Vernet <void@manifault.com>
As described in [0], there is an open problem in load balancing called
the "infeasible weights" problem. Essentially, the problem boils down to
the fact that a task with disproportionately high load can be granted
more CPU time than they can actually consume per their duty cycle.
This patch implements a solution to that problem, wherein we apply the
algorithm described in this paper to adjust all infeasible weights in
the system down to a feasible wight that gives them their full duty
cycle, while allowing the remaining feasible tasks on the system to
share the remaining compute capacity on the machine.
[0]: https://drive.google.com/file/d/1fAoWUlmW-HTp6akuATVpMxpUpvWcGSAv/view?usp=drive_link
Signed-off-by: David Vernet <void@manifault.com>
After updates to reflect the updated init and direct dispatch API, the
schedulers aren't compatible with older kernels. Bump versions and publish
releases.
In the latest kernel, sched_ext API has changed in two areas:
- ops.prep_enable/cancel_enable/enable/disable() replaced with
ops.init_task/enable/disable/exit_task().
- scx_bpf_dispatch() can now be called from ops.select_cpu(). Also,
SCX_ENQ_LOCAL flag is removed. Instead, users can call
scx_bpf_select_cpu_dfl() from ops.select_cpu() and use the @is_idle out
param value to determine whether to dispatch directly.
This commit updates all schedules so that they build.
- Init functions renamed / merged / split.
- ops.select_cpu() is added to several schedulers and local direct
disptching logic is moved there.
This is the minimum update which is need to make the schedulers build and
work. It needs further update to e.g. move vtime udpates to ops.enable().
This because each scheduler has it's own Rust Crate
and it's better if they had a README associated with each one.
https://crates.io/crates/scx_layered
- combine c and kernel-examples as it's confusing to have both
- rename 'rust-user' and 'c-user' to just 'rust' and 'c', which is simpler
- update and fix sync-to-kernel.sh
