| Age | Commit message (Collapse) | Author |
|---|
|
Pull to receive the __arena argument conversion:
67f1f4a48c24 ("sched_ext: Pass kernel arena pointers to ops_cid callbacks")
a8dc810968af ("sched_ext: Convert sub-cap kfuncs to __arena cmask arguments")
a05c5b5cb5cf ("sched_ext: Convert scx_bpf_cid_override() to __arena array arguments")
along with the bpf-next branch carrying the __arena argument support they
depend on.
Conflict in kernel/sched/ext/ext.c between:
c384ab8a0b13 ("sched_ext: Move the config-off sub-cap kfunc stubs into sub.c")
and:
a8dc810968af ("sched_ext: Convert sub-cap kfuncs to __arena cmask arguments")
which updated the stubs in their old ext.c location. Resolved by keeping
ext.c without the stubs and applying the prototype conversion to the
relocated stubs in sub.c.
Signed-off-by: Tejun Heo <tj@kernel.org>
|
|
With sub-schedulers, tasks of different schedulers routinely share rqs and
SMT siblings, but scx_prio_less() consults ops.core_sched_before() only when
both tasks belong to the same scheduler. Every pair spanning two schedulers
falls back to the default ordering, so no scheduler can express ordering
across a scheduler boundary, including a root over its sub-schedulers'
tasks.
Order a pair spanning schedulers by the nearest common ancestor that
implements ops.core_sched_before(): both tasks are in its subtree, making
this the one op where a scheduler is called on tasks it delegated to its
sub-schedulers and may not be scheduling anymore. Same-scheduler pairs keep
using the owning scheduler's op so a parent never orders inside a subtree it
delegated. The op is skipped when the deciding scheduler is bypassing on
either task's CPU.
Update scx_qmap to fall back to the kernel's default ordering when handed a
delegated task it has no task_ctx for.
Signed-off-by: Tejun Heo <tj@kernel.org>
|
|
scx_prio_less() implements prio_less() semantics - %true means that @a is
the lower priority and should run after @b. ops.core_sched_before() is
documented to return %true when @a should run before @b. scx_prio_less()
returns the op's value as-is, inverting the documented semantics at runtime.
Call the op with the arguments swapped.
scx_qmap followed the wiring instead of the documentation and returned %true
for the younger task, so the two inversions canceled out and it behaved as
intended. Flip its comparison to match. scx_qmap is likely the only current
user in or out of the kernel tree. Any scheduler written the same way needs
the same flip, while schedulers following the documentation are fixed by
this change.
Fixes: 7b0888b7cc19 ("sched_ext: Implement core-sched support")
Cc: stable@vger.kernel.org # v6.12+
Signed-off-by: Tejun Heo <tj@kernel.org>
|
|
sched_class->balance() is gone from sched_ext and what balance_one() does is
run dispatch to produce something pickable. Update the balance-era names to
dispatch terms:
- balance_one() -> dispatch_one()
- SCX_RQ_IN_BALANCE -> SCX_RQ_IN_DISPATCH
No BPF scheduler reads the flag. The enum autogen headers gain the new name
with the old entry retained like other removed enumerators, zero-filling at
load time. No functional changes.
Signed-off-by: Tejun Heo <tj@kernel.org>
|
|
fcg_dispatch() true-ups the current cgroup's cvtime when its slice
expires or its DSQ runs empty while the slice is still active:
__sync_fetch_and_add(&cgc->cvtime_delta,
(cpuc->cur_at + cgrp_slice_ns - now) *
FCG_HWEIGHT_ONE / (cgc->hweight ?: 1));
The true-up should be actual minus charged: on CNS_EXPIRE, the overrun
(now - cur_at - cgrp_slice_ns) should be added; on CNS_EMPTY, the
unused portion of the slice should be subtracted. The expression above
has the sign inverted, and in the CNS_EXPIRE case now is already past
cur_at + cgrp_slice_ns, so the u64 subtraction wraps. The
multiplication preserves the two's complement encoding but the
unsigned division by hweight destroys it, adding roughly 2^64/hweight
per expiry instead of a small correction.
Under saturation the hweight budget clamp in cgrp_cap_budget() masks
most of the garbage, so the weight distribution barely moves, but the
accounting is broken all the same. Compute the delta as a signed value
and use fetch_and_add()/fetch_and_sub() so that the dividends stay
positive, as BPF division is unsigned.
Instrumented the true-up and ran a saturated three-leaf cgroup tree on
a 4-CPU VM: without the fix, each expiry added ~5e15 (2^64/hweight
territory) to cvtime_delta; with it, the corrections are back to
slice scale, with the overrun added and the unused portion subtracted.
Fixes: a4103eacc2ab ("sched_ext: Add a cgroup scheduler which uses flattened hierarchy")
Suggested-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Tao Cui <cuitao@kylinos.cn>
Signed-off-by: Tejun Heo <tj@kernel.org>
|
|
libbpf strips the last ___flavor suffix when resolving kfunc externs, so
the bare ___v2 declaration resolves to scx_bpf_dsq_move_to_local, whose
BTF proto lacks @enq_flags. The extern never matches, bpf_ksym_exists()
returns false on every kernel that has the ___v2 kfunc, and the macro
falls back to ___v1, silently dropping @enq_flags.
Add the trailing ___compat suffix used by the other versioned externs in
this file (scx_bpf_dsq_insert___v2, scx_bpf_reenqueue_local___v2).
Any caller passing non-zero @enq_flags through the compat macro silently
loses them.
Fixes: 860683763ebf ("sched_ext: Add enq_flags to scx_bpf_dsq_move_to_local()")
Cc: stable@vger.kernel.org # v7.1+
Assisted-by: Z.ai:glm-5.2
Signed-off-by: fangqiurong <fangqiurong@kylinos.cn>
Signed-off-by: Tejun Heo <tj@kernel.org>
|
|
scx_bpf_dsq_reenq() is registered with KF_IMPLICIT_ARGS and its kernel BTF
prototype omits the trailing bpf_prog_aux argument. The ___compat extern
declares the argument explicitly, so libbpf never matches the prototype and
the weak extern silently stays unresolved on every kernel. The wrapper
always takes the old fallback path, which disables generic reenq users like
scx_qmap's lowpri mechanism and fails non-local reenq with "kernel too old"
even on kernels that have the kfunc.
Drop the explicit aux argument. Also correct the stale v6.20 reference, the
kfunc was added in v7.1.
Fixes: 9c34c5074d1b ("sched_ext: Introduce scx_bpf_dsq_reenq() for remote local DSQ reenqueue")
Cc: stable@vger.kernel.org # v7.1+
Signed-off-by: Tejun Heo <tj@kernel.org>
|
|
fcg_cgroup_set_weight() updates cgc->weight and the parent's
child_weight_sum but doesn't bump hweight_gen, so the hweights cached by
cgrp_refresh_hweight() stay stale until some task activation bumps the
generation. For cgroups whose tasks never go through a 0->n runnable
transition (e.g. persistently busy ones), a cpu.weight change never
propagates to scheduling at all.
Bump hweight_gen on weight changes so the next refresh recomputes with
the new weight.
Verified on a flatcg VM: a live cpu.weight 100->800 change on a busy
cgroup leaves HWT update at 0 and the distribution unchanged; with it,
hweight_gen increments and the refresh recomputes.
Signed-off-by: Tao Cui <cuitao@kylinos.cn>
Signed-off-by: Tejun Heo <tj@kernel.org>
|
|
scx_bpf_cid_override() predates the cid-form arena transition and takes its
arrays as verifier-checked mem+size buffers, forcing scx_qmap to keep the
cpu_to_cid and shard_start arrays in writable bss while the rest of its
state lives in the arena. Unify on arena arguments before cid-form
schedulers start seeing real use.
BPF now translates between BPF and kernel arena addresses for __arena
arguments. Take the arrays as __arena arguments, with the counts passed in
entries. The counts now size the snapshot copies and are bounds-checked
before them.
scx_qmap moves the arrays into struct qmap_arena. As the arena is mmapped at
load, the loader populates them between load and attach instead of before
load.
The arena argument address translation is currently implemented only on
x86-64. Schedulers calling this kfunc load only there for now.
Signed-off-by: Tejun Heo <tj@kernel.org>
|
|
The sub-cap kfuncs take their cmask arguments as __ign pointers. The values
cross the kfunc boundary as unchecked scalars and scx_cmask_ref_init()
rebases them into the arena by hand.
BPF now translates between BPF and kernel arena addresses for __arena
arguments. Tag the cmask arguments __arena so the kfuncs receive kernel
addresses and scx_cmask_ref_init() loses the hand-rolled conversion. The
optional denied_out keeps its NULL not-provided signal via
__arena__nullable. The mandatory masks use plain __arena.
scx_qmap's call sites drop the (void *)(long) casts since the BPF-side
declarations type the cmask arguments __arena and take arena pointers
directly.
The arena argument address translation is currently implemented only on
x86-64. Schedulers calling these kfuncs load only there for now.
Signed-off-by: Tejun Heo <tj@kernel.org>
|
|
|
|
Pull to receive:
c10b216a072f ("sched/core: Handle pick_task() releasing the rq lock")
f3629c63a4af ("sched/core: Make core-sched flips wait for in-flight selections")
ffaab58d2175 ("sched_ext: Replace SCX_RQ_BAL_KEEP with a dispatch verdict return")
3dd52416e44a ("sched_ext: Fix this_rq() assumptions in dispatch kfuncs")
f2da9587118d ("sched_ext: Count rq lock releases in rq->scx.lock_drop_seq")
d954004205c1 ("sched_ext: Fix rq->core_pick corruption under core scheduling")
for the pending core scheduling follow-ups and to resolve the conflicts
with the code reorganization and cap gate work on for-7.3.
ffaab58d2175 converts scx_dispatch_sched() to a dispatch verdict return
which for-7.3 moved from ext.c into inlines.h. Resolved by applying the
conversion to the relocated copy and combining balance_one()'s verdict
returns with the scx_task_can_stay_on_cpu() gate from the cap work.
ffaab58d2175 and 3dd52416e44a update scx_bpf_sub_dispatch() which
for-7.3 moved into sub.c. Resolved by applying the scx_locked_rq()
switch and the verdict test to the sub.c copy.
f2da9587118d instruments the open-coded lock releases in
consume_remote_task() which for-7.3 folded into switch_rq_lock().
Resolved by keeping the accounting in switch_rq_lock() which covers all
its callers.
d954004205c1 widens the put_prev_task_scx() WARN suppression to all
core-sched rqs on the same condition that for-7.3 gated with
scx_task_can_stay_on_cpu(). Resolved by combining both.
Signed-off-by: Tejun Heo <tj@kernel.org>
|
|
SCX_RQ_BAL_KEEP tells the pick to keep running the previous task, a leftover
from when balancing and picking were separate operations. An rq-level flag
only works while dispatches and picks pair up one to one, which core
scheduling breaks: selections interleave through dispatch's lock drops and a
pick can consume a stale flag, keeping a task that has since been dequeued.
Fixing core scheduling support requires the decision to travel with the
dispatch that made it. Make scx_dispatch_sched() and balance_one() return an
explicit verdict instead and drop the flag's plumbing from the tools autogen
enum headers.
Also factor the pick-side invocation, its follow-up queueing and the
post-dispatch checks out of do_pick_task_scx() into dispatch_pick(). No
functional changes intended.
v2: Drop the SCX_RQ_BAL_KEEP plumbing from the tools autogen enum headers
as well (Andrea).
Fixes: 4c95380701f5 ("sched/ext: Fold balance_scx() into pick_task_scx()")
Cc: stable@vger.kernel.org # v6.19+
Signed-off-by: Tejun Heo <tj@kernel.org>
|
|
git://git.kernel.org/pub/scm/linux/kernel/git/bpf/bpf-next into for-7.3-arena-args
Pull bpf-next d114bb989367 ("Merge branch
'add-arena-argument-support-to-kfuncs-and-struct_ops'") to make the __arena
and __arena__nullable kfunc and struct_ops argument suffixes available. The
suffixed arguments will be used to convert sched_ext kfuncs and struct_ops
callbacks that currently pass arena pointers as scalars and rebase them by
hand.
|
|
A sched holds only the cids its parent granted and nothing guarantees that
they cover its tasks' affinities. A task that can run on none of them has
nowhere to go and qmap stalls out: it force-inserts the task onto its first
allowed cid, but the kernel bounces the insert back and the task parks in
SHARED_DSQ, which is drained only on self cids it can't run on.
Set SCX_ENQ_RESCUE on these inserts so the kernel diverts such tasks to its
rescue path instead of bouncing them. The force-insert covers scheds with
and without children and fires on re-enqueues, and the SHARED_DSQ scan on
every dispatch rescues tasks stranded there - the enqueue-time check misses
a task whose cids were lost while it was already queued. The wrong-cid fault
injection carries the flag too and doubles as a deterministic rescue-traffic
generator.
-B and -q set the root-only rescue bandwidth and quantum ops knobs. -B 0
maps to SCX_RESCUE_DISABLE and turns rescue off kernel-side. Rescue inserts
are counted and reported in the hier stats line.
Signed-off-by: Tejun Heo <tj@kernel.org>
Reviewed-by: Andrea Righi <arighi@nvidia.com>
|
|
pick_direct_dispatch_cid() sent a pinned task straight to its only cpu
without an idle check. An insert onto a time-shared cid is forced IMMED,
which may not wait on a busy cpu - the kernel bounces the task back to
ops.enqueue(), the shortcut re-inserts it, and the loop runs into the
reenqueue repeat limit.
Run pinned tasks through the same idle test as everyone else and queue them
when the cpu is busy. always_enq_immed forces IMMED on every local insert
and already skipped the shortcut for the same reason. The check was the last
reader of the always_enq_immed rodata mirror, so drop the variable. The -I
option still sets SCX_OPS_ALWAYS_ENQ_IMMED.
v2: Drop the now-unread always_enq_immed rodata mirror. (sashiko AI)
Signed-off-by: Tejun Heo <tj@kernel.org>
Reviewed-by: Andrea Righi <arighi@nvidia.com>
|
|
Mechanical regeneration for the enumerators added by the slice write rework
and the rescue machinery. BPF schedulers reference SCX_ENQ_RESCUE, the rest
are enum_defs presence markers.
Signed-off-by: Tejun Heo <tj@kernel.org>
Reviewed-by: Andrea Righi <arighi@nvidia.com>
|
|
A local DSQ insert lacking the needed caps is diverted to the reject DSQ and
bounced back through ops.enqueue() so the scheduler can re-decide. That
recovery assumes the scheduler has somewhere legal to send the task. When it
doesn't, e.g. when the task's affinity is restricted to cids delegated away,
the task starves until the stall watchdog ejects the scheduler. An exiting
task is worse - it skips ops.enqueue() and the rejection becomes a
self-requeuing cycle that burns the CPU until the watchdog fires.
Add SCX_ENQ_RESCUE, a fallback modifier on local DSQ inserts. When the
insert would be rejected for missing caps, the kernel takes over and runs
the task on the target CPU without consulting the owning scheduler. The
kernel sets the flag itself when enqueueing an exiting task.
Rescue is a last-resort forward-progress backstop with a persistent
disadvantage, not a way around cap enforcement. A per-CPU token bucket
accrues rescue_bandwidth_ppt (default 2%) of CPU time and rescues run one at
a time in arrival order. Each is granted a slice of the rescue_quantum_us
(default 5ms) quantum divided across the waiters, waits at the tail of the
local DSQ claiming no priority, and rejoins its scheduler as a fresh arrival
once the slice is served.
The schedulers keep their normal control over an admitted rescuee and may
preempt or reslice it. Service is measured on CPU time actually received, so
neither shortens the rescue. Prolonged denial escalates - the remaining
slice turns into protected execution (SCX_TASK_PROTECTED) and the rescuee
preempts the current task. Escalation is paced by the same bucket, and
delivered service converges on the configured bandwidth no matter how
aggressively the schedulers dispatch.
Both knobs are root-only and SCX_RESCUE_DISABLE turns rescue off, making
SCX_ENQ_RESCUE inserts reject as usual.
v2: - Add SCX_OPS_OPEN() fix-ups for the new ops fields so cpu-form
schedulers setting them still load on older kernels. (Andrea)
Signed-off-by: Tejun Heo <tj@kernel.org>
Reviewed-by: Andrea Righi <arighi@nvidia.com>
|
|
Regenerate enum_defs.autogen.h from the current vmlinux.h to pick up the SCX
enum changes accumulated since the last regeneration, including the
SCX_REENQ_LOCAL_MAX_REPEAT to SCX_REENQ_MAX_REPEAT rename.
Reported-by: Andrea Righi <arighi@nvidia.com>
Link: https://lore.kernel.org/all/amZsEbZJdDgjstPF@gpd4/
Signed-off-by: Tejun Heo <tj@kernel.org>
|
|
scx_bpf_cidperf_set() reaches cpufreq with no cap check, so any cid-form
sub-sched can steer the frequency of any cid in its view, including ones it
holds nothing on.
Gate it behind a new SCX_CAP_PERF rather than SCX_CAP_BASE: hardware control
is a separate axis from queue access - a parent may well delegate scheduling
on a cid without handing over its frequency. PERF neither implies nor is
implied by the other caps. The check runs under the target rq's lock, which
ecaps updates are also folded under, so it is authoritative - a write can
never land after a revoke has taken effect. Denials are counted in
SCX_EV_SUB_CIDPERF_DENIED.
The operation is synchronous and the outcome is reported to the caller:
scx_bpf_cidperf_set() now returns 0 or -errno, -EACCES on denial. The
cid-form interface is still under initial development, so the signature is
changed in place without versioning.
scx_qmap grants PERF alongside its existing cid grants so the cpuperf demo
keeps working in sub-scheds.
Signed-off-by: Tejun Heo <tj@kernel.org>
Reviewed-by: Andrea Righi <arighi@nvidia.com>
|
|
The tools restart when the kernel exits the scheduler with
SCX_ECODE_ACT_RESTART. The restart decision doesn't consult exit_req, so an
exit request arriving while the restart condition persists is ignored and
the tool reloads in a tight loop. Test exit_req before restarting.
scx_userland needs more: its main loop never watches the kernel-side exit
and exit_req doubles as the stats printer's stop signal, set by the teardown
and reset on each restart. Add the missing UEI_EXITED() test and give the
printer its own stop flag so that exit_req only means an exit request and
stays latched like in the other tools.
Signed-off-by: Tejun Heo <tj@kernel.org>
Reviewed-by: Andrea Righi <arighi@nvidia.com>
|
|
ops.cpu_acquire/release() are deprecated in favor of tracking CPU
preemption from a sched_switch tracepoint, see
commit a3f5d4822253 ("sched_ext: Allow scx_bpf_reenqueue_local() to
be called from anywhere"). Loading scx_pair currently emits a
deprecation warning.
Replace the pair_cpu_acquire/release() callbacks with a
tp_btf/sched_switch program that edge-detects the same transitions the
core used to deliver: a release when a running SCX task loses its CPU
to a higher-priority class, and an acquire when the CPU switches back
to an SCX task or idle while marked preempted.
Tasks are classified by effective priority (p->prio) rather than by
policy: rt_mutex_setprio() boosts a PI beneficiary into the rt/dl
classes while leaving its policy untouched, so a policy test would
both miss the release when a boosted task takes the CPU and fire a
spurious acquire when a boosted task replaces a real rt task.
A switch from idle straight to a higher-priority task is deliberately
not treated as a release. The CPU was not running an SCX task, so
there is nothing to drain, and kicking SCX_KICK_PREEMPT |
SCX_KICK_WAIT on every rt wakeup would make the pair CPU wait out rt
bursts it was never coupled to. The old callbacks behaved the same
way, firing ops.cpu_release() only from switch_class() when an SCX
task was put for a higher class.
The tracepoint runs on every context switch in the system, so the
common no-transition case is filtered before taking the pair-shared
lock. This is safe because a CPU's own preempted_mask bit is only ever
written by this tracepoint running on that CPU.
sched_setscheduler() on a running task changes class in place without
a context switch, so such transitions are only observed at the task's
next switch. The old callbacks had the same blind spot in
switch_class(), and try_dispatch() already bounds the resulting wait.
Verified in virtme-ng with the script below. The scheduler must load
without the deprecation warning, stay enabled through the rt churn and
the idle soak (the watchdog would otherwise abort it with "runnable
task stall"), keep its preemption counter advancing, and unregister
cleanly at the end. A PI rt-mutex churn that repeatedly boosts
SCX tasks into the rt class was exercised separately:
#!/bin/bash
# vng --verbose --cpus 8 -m 4G --user root -- ./verify.sh
# FIFO harness: survives even if all SCHED_NORMAL tasks stall
[ "${RT:-0}" = 1 ] || exec chrt -f 5 env RT=1 "$0"
chrt -o 0 ./tools/sched_ext/build/bin/scx_pair &
PAIR=$!
sleep 3
for round in $(seq 10); do
pids=""
for i in 0 1 2 3; do # SCHED_FIFO churn
chrt -f 10 bash -c \
'e=$((SECONDS+1)); while [ $SECONDS -lt $e ]; do :; done' &
pids="$pids $!"
done
for i in 0 1; do # SCHED_NORMAL load under scx
chrt -o 0 bash -c \
'n=0; while [ $n -lt 200000 ]; do n=$((n+1)); done' &
pids="$pids $!"
done
wait $pids # explicit pids, not the scx_pair job
done
sleep 300 # idle soak
kill -INT $PAIR # expect clean unregister in dmesg
Signed-off-by: Cheng-Yang Chou <yphbchou0911@gmail.com>
Reviewed-by: Andrea Righi <arighi@nvidia.com>
Signed-off-by: Tejun Heo <tj@kernel.org>
|
|
Two unrelated things go by "cgroup" in the cid form. Sub-schedulers attach
to cgroups, and the cgroup_*() ops deliver cpu controller events. While the
ops names suggest cgroup2 hierarchy, they actually operate on the cpu
controller.
Rename them to cpuctl_* in struct sched_ext_ops_cid, which has no users
outside scx_qmap yet. The cpu form is deployed ABI and keeps the old names.
The layout is unchanged and the kernel keeps calling through the cpu-form
union view.
Signed-off-by: Tejun Heo <tj@kernel.org>
Reviewed-by: Andrea Righi <arighi@nvidia.com>
|
|
SCX_OPS_OPEN() clears compat-gated ops fields which the running kernel
lacks. The clears dereference cpu-form member names and compile for
cid-form skeletons only because both ops structs currently name their
cgroup ops identically, which an upcoming rename will end. No load-time
fix-up can apply to a cid-form scheduler anyway as the cid form postdates
every compat-gated op.
Factor the skeleton open path out of SCX_OPS_OPEN() and add
SCX_OPS_CID_OPEN() which uses only that shared part. Switch scx_qmap, the
only cid-form scheduler, over.
Signed-off-by: Tejun Heo <tj@kernel.org>
Reviewed-by: Andrea Righi <arighi@nvidia.com>
|
|
Add -J init-fail which makes ops.init_task() fail with -ENOMEM for tasks
whose comm starts with "qmfail", and -J cgrp-init-fail which does the same
in ops.cgroup_init() for cgroups named "qmfail*".
The former exercises the migration veto path: the cgroup.procs write must
fail with the injected errno while the destination sched stays up and the
task stays put. The latter exercises the ownership-return failure path: a
parent failing to re-init a returned cgroup leaves it unowned, and moves and
set_* ops against it must be skipped instead of dereferencing the missing
owner. Matching on "qmfail" names keeps the injecting scheduler's own enable
unaffected.
Signed-off-by: Tejun Heo <tj@kernel.org>
Reviewed-by: Andrea Righi <arighi@nvidia.com>
|
|
With the set_* ops delivered to the parent's sched, a parent qmap instance
now receives cgroup_set_weight for its child subs' attach points. Update
the matching sub_sched_ctx weight and redistribute() in-kernel, and drop
the userspace feed_weights() polling. This exercises the knob routing end
to end.
The self weight is fixed at 100: a cgroup's weight is its parent's knob and
not the scheduler's own business. This drops the self-weight polling and the
repartition PROG_RUN poke with it.
sub_attach seeds the slot with the cgroup's current weight, read through
bpf_cgroup_from_id(), so a weight set before the sub attaches is picked up.
A write racing the attach can still be lost until the next value-changing
cpu.weight write. Acceptable for a demo.
While at it, add a traced ops.cgroup_move() so tests can observe move
delivery.
Signed-off-by: Tejun Heo <tj@kernel.org>
Reviewed-by: Andrea Righi <arighi@nvidia.com>
|
|
Cross-merge BPF and other fixes after downstream PR.
Signed-off-by: Kumar Kartikeya Dwivedi <memxor@gmail.com>
|
|
The comment above dispatch_highpri() still references
scx_bpf_dispatch[_vtime]_from_dsq(), which was renamed to
scx_bpf_dsq_move[_vtime]() in v6.13 to unload the overloaded
"dispatch" verb. The code below already uses the new names; only the
comment was left behind during the rename.
Fixes: 5cbb302880f5 ("sched_ext: Rename scx_bpf_dispatch[_vtime]_from_dsq*() -> scx_bpf_dsq_move[_vtime]*()")
Signed-off-by: Liang Luo <luoliang@kylinos.cn>
Signed-off-by: Tejun Heo <tj@kernel.org>
|
|
Add a fault-injection mode to the scx_qmap sub-scheduler that deliberately
dispatches one of its own tasks to a cid it does not hold. The kernel cap
check must reject it and re-enqueue with SCX_TASK_REENQ_CAP, so the
nr_inject_attempts counter tracks nr_reenq_cap one to one, exercising the
delivery-time cap enforcement.
Signed-off-by: Tejun Heo <tj@kernel.org>
Reviewed-by: Andrea Righi <arighi@nvidia.com>
|
|
sched_ext sub-scheduling began as dispatch delegation only: a parent could
call into a child cgroup sub-scheduler's ops.dispatch() from its own
dispatch path, but could not delegate cpus to the child for enqueue and the
other paths. sched_ext has since gained cap-based cid delegation, where a
parent grants and revokes a child's per-cid caps. Expand scx_qmap to
demonstrate it.
scx_qmap can now delegate the cids it holds exclusively, split among itself
and its children by cpu.weight. Each gets the floor of its share as
dedicated cids. The leftover from rounding forms a shared pool,
round-robined among them as an ENQ_IMMED time-share.
This shape is deliberate. Exclusive cids exercise the basic grant and revoke
of ownership, and the shared pool exercises time-sharing one cid across
several schedulers. The implemented policy is impractical, but it covers
most of what a practical sub-scheduler would need without overcomplicating
qmap.
Delegation nests. A cid a node receives from its parent only as a
round-robin share stays self-local and is never re-delegated. A node left
with no exclusive cid, e.g. after its cpus went offline, evicts its
children.
v5: Highpri dispatch masked with self_cids, single-read dispatch cgroup_id, feed_weights race comment. (sashiko AI)
v4: Track all idle cids and mask with self_cids at the dispatch pick, dropping the reseed. (sashiko AI)
v3: Dispatch IMMED flags, repartition accounting order, partition-input snapshot. (sashiko AI)
v2: Use __sync_fetch_and_add() for the shared nr_dsps counter. (sashiko AI)
Signed-off-by: Tejun Heo <tj@kernel.org>
Reviewed-by: Andrea Righi <arighi@nvidia.com>
|
|
Add cmask_next_and2_set() and its round-robin wrap, extending
cmask_next_and_set() to a three-mask intersection: the next cid set in all
three masks at or after @start. A caller iterating the intersection of
three cmasks can then scan it in one pass, folding the third mask into the
word-level AND rather than skipping non-members one candidate at a time.
Signed-off-by: Tejun Heo <tj@kernel.org>
Reviewed-by: Andrea Righi <arighi@nvidia.com>
|
|
A cid-form scheduler can grant caps to and revoke them from its child
sub-schedulers but has no way to tear one down. Add scx_bpf_sub_kill() to
evict a direct child with a printf-style reason that reaches the child's
scx_exit_info. No exit code is taken because the child is a separate
scheduler whose exit-code semantics the parent cannot know. The child and
its subtree are disabled through the usual async path under a new exit kind,
SCX_EXIT_PARENT_KILL.
The bstr formatting infrastructure in ext.c is exposed through internal.h
with scx_ prefixes so the kfunc, which lives in sub.c, can format the
reason.
Signed-off-by: Tejun Heo <tj@kernel.org>
Reviewed-by: Andrea Righi <arighi@nvidia.com>
|
|
Caps are per-cid permissions parents delegate to direct children via
scx_bpf_sub_grant() / scx_bpf_sub_revoke(). A child's cap set is always a
subset of its parent's. Sub-scheds check their caps locally, and cross-sched
communication is needed only when the delegation set itself changes.
Caps will be used to implement sub-sched scheduling on the enqueue path.
Picking a cid for a task at a leaf depends on which cids the leaf is allowed
to use, and resolving that programmatically on every enqueue would mean a
cross-sched round-trip call chain, possibly retrying if the request can't be
granted as-is. The dispatch path is different - it runs as top-down
recursion via scx_bpf_sub_dispatch().
Locking is per shard. cid space is split into shards, and each sub-sched has
its own pshard->lock for each shard. Operations are broken up on shard
boundaries. Different shards never contend. Shards are expected to be
topology-aligned and likely to serve as the locality unit when cids are
allocated to schedulers, so per-shard lock granularity scales naturally with
the allocation pattern.
This patch adds the framework with a single dummy cap. Real caps land in
later patches.
The enable path is reordered for pshards. scx_arena_pool_init() moves ahead
of scx_link_sched() so the pshards are allocated before the sched becomes
reachable - scx_alloc_pshards() skips allocation when the arena pool isn't
initialized.
- scx_bpf_sub_grant(): Per-cid all-or-nothing grant to direct child.
- scx_bpf_sub_revoke(): Clear caps on @cmask across @child and its subtree.
- scx_bpf_sub_caps(): Lockless snapshot of caps on a cid range.
/sys/kernel/sched_ext/SCHED/caps shows the caps each scheduler currently
holds.
v4: Move the pshard[] full build/publish and the err_disable scx_error() recording to earlier patches. (sashiko AI)
v3: Build pshard[] fully before publishing it, read it with READ_ONCE. (sashiko AI)
v2: Validate ops before scx_link_sched() publishes the sub. (sashiko AI)
Signed-off-by: Tejun Heo <tj@kernel.org>
Reviewed-by: Andrea Righi <arighi@nvidia.com>
|
|
An overridden cid mapping invalidates the auto-generated shard layout, so
the override call has to provide both. Extend scx_bpf_cid_override() with a
shard_start[] array that lists the first cid of each shard (starting at 0,
strictly increasing, last shard implicitly extends to num_possible_cpus()).
A scheduler that wants only custom shards with the auto-generated cid
mapping can read the current mapping and pass it back unchanged.
Overridden shards can span NUMA nodes, so scx_shard_node[] is rebuilt by
majority count: each shard is assigned to the node that owns the most cpus
in it.
v2: Snapshot the caller's cpu_to_cid/shard_start arrays before validating. (sashiko AI)
Signed-off-by: Tejun Heo <tj@kernel.org>
Reviewed-by: Andrea Righi <arighi@nvidia.com>
|
|
A cid-form scheduler that calls scx_bpf_cid_override() to install a custom
cid layout can only do so from ops.init(). Enable-path setup that depends on
the cid layout thus has to run after ops.init(), and ops.init() itself can't
use anything derived from the final layout, which turned out to be too
restrictive.
Add an ops.init_cids() callback dedicated to finalizing the cid layout. It
runs before the rest of the enable-path setup, so the final layout is in
effect for everything that follows including ops.init(), which now runs
after the arena pool and cmask scratch allocations.
scx_bpf_cid_override() is restricted to ops.init_cids() at load time. It
sits in a kfunc set gated by SCX_KF_ALLOW_INIT_CIDS, a flag set only on the
init_cids op, so the verifier rejects a call from any other context. The
runtime root-only check is dropped as ops.init_cids() only runs during root
enable.
The qmap demo moves its override into a dedicated qmap_init_cids() and,
while at it, introduces an enum for the cid override modes instead of
hard-coded integers.
Signed-off-by: Tejun Heo <tj@kernel.org>
Reviewed-by: Andrea Righi <arighi@nvidia.com>
|
|
In fcg_read_stats(), the memset() that zeroes the output @stats array
sits after the calloc() failure check. When calloc() fails, the
function returns without writing @stats.
The caller in main() declares acc_stats uninitialized, passes it as
the @stats argument, and then reads it unconditionally:
__u64 acc_stats[FCG_NR_STATS];
fcg_read_stats(skel, acc_stats);
stats[i] = acc_stats[i] - last_stats[i]; // reads garbage
Because fcg_read_stats() returns void, the caller cannot detect the
failure. Reading the uninitialized array is undefined behavior, and
the garbage is further copied into last_stats via memcpy(), corrupting
the baseline used by the next interval.
This regression was introduced by commit cabd76bbc036 ("tools/sched_ext:
scx_flatcg: fix potential stack overflow from VLA in fcg_read_stats"),
which replaced the VLA with calloc() and inserted the failure check
before the existing memset().
Move the memset() above the calloc() failure check so @stats is always
zeroed regardless of allocation outcome.
Fixes: cabd76bbc036 ("tools/sched_ext: scx_flatcg: fix potential stack overflow from VLA in fcg_read_stats")
Signed-off-by: Liang Luo <luoliang@kylinos.cn>
Reviewed-by: Andrea Righi <arighi@nvidia.com>
Signed-off-by: Tejun Heo <tj@kernel.org>
|
|
bpf_cpumask_populate() writes to its destination with bitmap_copy(), but
the destination is typed as struct cpumask *. That allows the verifier to
accept borrowed cpumask pointers returned by read-only kfuncs, such as
scx_bpf_get_online_cpumask(), as a writable destination.
Make the destination a struct bpf_cpumask * so populate follows the same
ownership rule as the other mutating cpumask kfuncs. Query kfuncs continue
to accept const struct cpumask * inputs.
Fixes: 950ad93df2fc ("bpf: add kfunc for populating cpumask bits")
Signed-off-by: Nicholas Dudar <main.kalliope@gmail.com>
Acked-by: Tejun Heo <tj@kernel.org>
Reviewed-by: Emil Tsalapatis <emil@etsalapatis.com>
Link: https://lore.kernel.org/bpf/20260709182800.2037938-2-main.kalliope@gmail.com
Assisted-by: Claude:claude-opus-4-8
Signed-off-by: Kumar Kartikeya Dwivedi <memxor@gmail.com>
|
|
scx_bpf_kick_cid() returned an error code, but the value conveys nothing
actionable and no caller consumes it. The kick is asynchronous, so a
successful return only means it was queued. An invalid @cid is already
reported through scx_error() by scx_cid_to_cpu(), and a missing scheduler
leaves nothing to kick.
Make scx_bpf_kick_cid() return void to match scx_bpf_kick_cpu(). The
cid-form kfuncs are not in practical use yet, so the ABI change is safe.
Signed-off-by: Tejun Heo <tj@kernel.org>
Reviewed-by: Andrea Righi <arighi@nvidia.com>
|
|
scx_qmap.c and the shared scx_qmap.h mixed __u64/__u32/__s32 with the bare
typedefs that scx/common.h provides. Convert the remaining __-prefixed
integer types to the bare forms for consistency. The struct fields become
bare u64 (uint64_t), so the stats printfs that fed them to %llu now cast to
unsigned long long. No functional change.
Signed-off-by: Tejun Heo <tj@kernel.org>
Reviewed-by: Andrea Righi <arighi@nvidia.com>
|
|
Pull to receive:
db4e9defd2e8 ("sched_ext: Record an error on errno-only sub-enable failure")
49b3378a750c ("sched_ext: Fix premature ops->priv publication in scx_alloc_and_add_sched()")
e6979d05c6a6 ("tools/sched_ext: scx - Fix cmask_subset(), cmask_equal() and cmask_weight()")
for further sub-sched changes and to resolve the conflicts with the
sub-sched updates on for-7.3.
db4e9defd2e8 adds scx_error() to the sub-enable err_disable sink which
for-7.3 moved from ext.c into sub.c. Resolved by applying the fix to
scx_sub_enable_workfn() in sub.c.
49b3378a750c drops RCU_INIT_POINTER() from an scx_alloc_and_add_sched()
unwind label whose body changed with for-7.3's stall_cpus addition.
Resolved by dropping the line from the updated unwind.
Signed-off-by: Tejun Heo <tj@kernel.org>
|
|
cmask_equal(), cmask_weight() and cmask_subset() bounded their word walks
with CMASK_NR_WORDS(nr_cids), which pads by one word and can't tell the last
word in use without @base. The walks could thus cover a slack word past the
active range, which cmask_reframe() leaves non-zero: a stale bit there gave
cmask_equal() a spurious mismatch, cmask_weight() an inflated count, and
cmask_subset() a spurious violation. cmask_subset() could also read
@b->bits[] one word past its allocation (within the arena's fault-recovered
range, so harmless), and deviated from the kernel scx_cmask_subset() by
failing any @a range that doesn't nest inside @b's even when the overhanging
bits are all clear.
Bound the cmask_equal() and cmask_weight() walks by the words the range
actually spans, with early returns for empty ranges. Rewrite cmask_subset()
to match the kernel semantics: scan @a's overhangs for set bits with
cmask_next_set() and walk the words of the range intersection.
cmask_subset() moves below cmask_next_set(), which it now uses. Padding bits
don't need masking as every cmask helper keeps them clear.
Fixes: a58e6b79b432 ("sched_ext: Add cmask, a base-windowed bitmap over cid space")
Signed-off-by: Tejun Heo <tj@kernel.org>
Reviewed-by: Andrea Righi <arighi@nvidia.com>
|
|
__COMPAT_read_enum() and __COMPAT_struct_has_field() open-code the
vlen lookup via the raw BTF_INFO_VLEN(t->info) UAPI macro. libbpf
exposes btf_vlen() for exactly this purpose; use it in the three
call sites, matching the pattern in kernel/bpf/inode.c and
tools/bpf/bpftool.
btf_vlen() returns __u32 (since commit cacd6729c0923, "libbpf:
Adjust btf_vlen() to return a __u32", which expanded the BTF vlen
field from 16 to 24 bits). Declare the loop counters as __u32 to
match the return type, keeping the comparison as a plain
'__u32 < __u32' and silencing the -Wsign-compare warnings.
No functional change.
Suggested-by: Andrea Righi <arighi@nvidia.com>
Signed-off-by: Liang Luo <luoliang@kylinos.cn>
Reviewed-by: Andrea Righi <arighi@nvidia.com>
Signed-off-by: Tejun Heo <tj@kernel.org>
|
|
Four example schedulers (scx_simple, scx_cpu0, scx_sdt, scx_userland)
declare the variable that holds getopt()'s return value as __u32.
getopt() returns int and uses -1 to mark end-of-options; storing that
sentinel in an unsigned variable turns it into 0xffffffff, and the
subsequent 'opt != -1' test only happens to keep working because both
operands of != are promoted to the same 0xffffffff.
Declare the variable as __s32 instead, matching getopt()'s actual
contract and the style already used in scx_qmap.c (int opt) and
scx_pair/central/flatcg.c (__s32 opt). This also silences the
-Wsign-compare warnings emitted for the affected files.
Signed-off-by: Liang Luo <luoliang@kylinos.cn>
Signed-off-by: Tejun Heo <tj@kernel.org>
|
|
bypass charging
1. cgrp_cap_budget() used __sync_fetch_and_sub(&cgc->cvtime_delta,
cgc->cvtime_delta) to atomically read and clear cvtime_delta. However,
this is not a true atomic read-clear operation: the second argument
(cgc->cvtime_delta) is evaluated as a normal read before the atomic
fetch_and_sub executes. If a concurrent __sync_fetch_and_add() happens
between the read and the sub, the added value gets included in the
returned delta AND remains in cvtime_delta, causing double charging.
Example:
CPU 0 runs cgrp_cap_budget(), CPU 1 runs fcg_stopping().
Assume cvtime_delta = 100 initially.
T1 CPU 0: sub_val = cvtime_delta = 100 cvtime_delta = 100
T2 CPU 1: __sync_fetch_and_add(&cvtime_delta, 10) cvtime_delta = 110
T3 CPU 0: __sync_fetch_and_sub(&cvtime_delta, sub_val) cvtime_delta = 10
returns old=110
delta = 110 (includes the 10 from CPU 1), but cvtime_delta = 10
(the 10 also remains). The 10 is charged twice: once in delta
(applied to cgv_node->cvtime) and once in the residual cvtime_delta
(fetched again next time).
Fix by using __sync_fetch_and_and(&cgc->cvtime_delta, 0).
Disassembly comparison:
(1) delta = __sync_fetch_and_sub(&cgc->cvtime_delta, cgc->cvtime_delta);
228: (79) r7 = *(u64 *)(r9 +40)
229: (87) r7 = -r7
230: (db) r7 = atomic64_fetch_add((u64 *)(r9 +40), r7) //r9 may be changed
(2) delta = __sync_fetch_and_and(&cgc->cvtime_delta, 0);
228: (b7) r8 = 0
229: (db) r8 = atomic64_xchg((u64 *)(r9 +40), r8)
2. The bypass charging path in fcg_stopping() charges raw execution time
to cvtime_delta without scaling by the inverse of the cgroup hweight.
Since cvtime_delta is eventually applied to cgv_node->cvtime which is
in vtime space (weight-scaled), the bypass path should also scale by
FCG_HWEIGHT_ONE / hweight to match the units used by the dispatch path.
Fixes: a4103eacc2ab ("sched_ext: Add a cgroup scheduler which uses flattened hierarchy")
Signed-off-by: Wanwu Li <liwanwu@kylinos.cn>
Reviewed-by: Andrea Righi <arighi@nvidia.com>
Signed-off-by: Tejun Heo <tj@kernel.org>
|
|
scx_bpf_cpu_rq() exposes rq pointers without requiring the rq lock and
has emitted a deprecation warning since commit
5c48d88fe004 ("sched_ext: deprecation warn for scx_bpf_cpu_rq()").
The supported replacements cover the intended uses:
scx_bpf_locked_rq() for locked rq access and
scx_bpf_cpu_curr() for remote curr lookup.
Remove the kfunc, its BTF registrations, the deprecation warning
state, and the BPF-side prototype and compat fallback.
Signed-off-by: Christian Loehle <christian.loehle@arm.com>
Reviewed-by: Andrea Righi <arighi@nvidia.com>
Reviewed-by: Hongyan Xia <hongyan.xia@transsion.com>
Signed-off-by: Tejun Heo <tj@kernel.org>
|
|
The sched_ext sources had grown to ten ext* files directly under
kernel/sched/. Move them into a new kernel/sched/ext/ subdirectory and drop
the now-redundant ext_ prefix. ext.c/h keep their names.
kernel/sched/ext.{c,h} -> kernel/sched/ext/ext.{c,h}
kernel/sched/ext_internal.h -> kernel/sched/ext/internal.h
kernel/sched/ext_types.h -> kernel/sched/ext/types.h
kernel/sched/ext_idle.{c,h} -> kernel/sched/ext/idle.{c,h}
kernel/sched/ext_cid.{c,h} -> kernel/sched/ext/cid.{c,h}
kernel/sched/ext_arena.{c,h} -> kernel/sched/ext/arena.{c,h}
The include paths in build_policy.c and sched.h, the MAINTAINERS glob, and a
few documentation and comment references are updated to match. No code or
symbol changes.
Suggested-by: Linus Torvalds <torvalds@linux-foundation.org>
Reviewed-by: Andrea Righi <arighi@nvidia.com>
Signed-off-by: Tejun Heo <tj@kernel.org>
|
|
git://git.kernel.org/pub/scm/linux/kernel/git/tj/sched_ext
Pull sched_ext updates from Tejun Heo:
"Most of this continues the in-development sub-scheduler support, which
lets a root BPF scheduler delegate to nested sub-schedulers. The
dispatch-path building blocks landed in 7.1. A follow-up patchset in
development will complete enqueue-path support for hierarchical
scheduling. This cycle adds most of that infrastructure:
- Topological CPU IDs (cids): a dense, topology-ordered CPU numbering
where the CPUs of a core, LLC, or NUMA node form contiguous ranges,
so a topology unit becomes a (start, length) slice. Raw CPU numbers
are sparse and don't track topological closeness, which makes them
clumsy for sharding work across sub-schedulers and awkward in BPF.
- cmask: bitmaps windowed over a slice of cid space, so a
sub-scheduler can track, for example, the idle cids of its shard
without a full NR_CPUS cpumask.
- A struct_ops variant that cid-form sub-schedulers register with,
along with the cid-form kfuncs they call.
- BPF arena integration, which sub-scheduler support is built on. The
bpf-next additions let the kernel read and write the BPF
scheduler's arena directly, turning it into a real kernel/BPF
shared-memory channel. Shared state like the per-CPU cmask now
lives there.
- scx_qmap is reworked to exercise the new arena and cid interfaces.
Additionally:
- Exit-dump improvements: dump the faulting CPU first, expose the
exit CPU to BPF and userspace, and normalize the dump header.
- Misc kfuncs and cleanups: a task-ID lookup kfunc, __printf checking
on the error and dump formatters, header reorganization, and
assorted fixes"
* tag 'sched_ext-for-7.2' of git://git.kernel.org/pub/scm/linux/kernel/git/tj/sched_ext: (59 commits)
sched_ext: Add scx_arena_to_kaddr() / scx_kaddr_to_arena()
sched_ext: Make scx_bpf_kick_cid() return s32
sched_ext: Add scx_cmask_test() and scx_cmask_for_each_cid()
tools/sched_ext: Order single-cid cmask helpers as (cid, mask)
sched_ext: Order single-cid cmask helpers as (cid, mask)
selftests/sched_ext: Fix dsq_move_to_local check
sched_ext: Guard BPF arena helper calls to fix 32-bit build
sched_ext: idle: Fix errno loss in scx_idle_init()
sched_ext: Convert ops.set_cmask() to arena-resident cmask
sched_ext: Sub-allocator over kernel-claimed BPF arena pages
sched_ext: Require an arena for cid-form schedulers
sched_ext: Add cmask mask ops
sched_ext: Track bits[] storage size in struct scx_cmask
sched_ext: Rename scx_cmask.nr_bits to nr_cids
tools/sched_ext: scx_qmap: Fix qa arena placement
sched_ext: Mark !CONFIG_EXT_SUB_SCHED dummy stubs static inline
sched_ext: Replace tryget_task_struct() with get_task_struct()
sched_ext: Add scx_task_iter_relock() and use it in scx_root_enable_workfn()
sched_ext: Fix ops_cid layout assert
sched_ext: Use offsetofend on both sides of the ops_cid layout assert
...
|
|
Switch scx_bpf_kick_cid() from void to s32 so future cap enforcement can
surface failures. cid interface is introduced in this cycle and has no
external users, so the ABI change is safe. Subsequent patches will add
-EPERM returns when the calling sub-sched lacks the required cap on the
target cid.
v2: Return scx_cid_to_cpu()'s errno instead of -EINVAL. (Andrea)
Signed-off-by: Tejun Heo <tj@kernel.org>
Reviewed-by: Andrea Righi <arighi@nvidia.com>
|
|
The BPF arena single-cid cmask helpers take the cmask first and the cid
second. Reorder them to (cid, mask) to match the kernel-side helpers and
the test_bit(nr, addr), cpumask_test_cpu(cpu, mask) convention. Range and
iteration helpers keep (mask, start).
Signed-off-by: Tejun Heo <tj@kernel.org>
Reviewed-by: Andrea Righi <arighi@nvidia.com>
|
|
scx_show_state.py still reads scx_aborting and scx_bypass_depth as
global symbols. Those symbols no longer exist after the state was moved
into struct scx_sched, so the drgn script fails when it reaches either
field.
Read aborting and bypass_depth from scx_root instead. This preserves the
script's current root-scheduler view: with sub-scheduler support, the
reported values are for the root scheduler and sub-schedulers are not
enumerated.
Fixes: 5c8d98a1b4de ("sched_ext: Move bypass state into scx_sched")
Fixes: c1743da43cf5 ("sched_ext: Move aborting flag to per-scheduler field")
Signed-off-by: Zicheng Qu <quzicheng@huawei.com>
Reviewed-by: Andrea Righi <arighi@nvidia.com>
Signed-off-by: Tejun Heo <tj@kernel.org>
|
