- Add patches for building on riscv64/aarch64 and s390
- Update to version 1.0.6:
* version: v1.0.6 Cargo.lock updates
* version: v1.0.6
* scx_mitosis: add RCU-like synchronization
* scx_bpfland: drop lowlatency mode and the priority DSQ
* scx_lavd: deprioritize a long runtime by prioritizing frequencies further
* scx_lavd: avoid self-IPI at preemption
* scx_bpfland: strict domain affinity
* scx_bpfland: consider all tasks as regular if priority DSQ is congested
* scx_bpfland: do not overly prioritize WAKE_SYNC tasks
* scx_bpfland: do not exclude exiting tasks
* scx_bpfland: consistent vruntime update
* scx_rusty: Fix filtering logic error
* scx_rusty: Restore push domain tasks when no task found
* scx_layered: remove ->open from layer struct
* scx_layered: point costc to global struct when initializing budgets
* scx_layered: Refactor cost naming
* scx_lavd: drop all kernel lock tracing
* scx_lavd: improve readabilty
* scx: sync default flags with CachyOS Kernel Manager
* scx_lavd: fix a variable name to kill a warning
* scx_lavd: fix CI error for missing kernel symbols
* scx_layered Fix trace format
* scx_layered: Add additional drain to fallback DSQs
* scx_layered: Fix dump output format
* scx_lavd: create DSQs on their associated NUMA nodes
* scx_lavd: optimize consume_starving_task()
* scx_lavd: fix warnings
* scx_bpfland: get rid of the global dynamic nvcsw threshold
* scx_bpfland: get rid of MAX_LATENCY_WEIGHT
* scx_bpfland: evaluate nvcsw without using kernel metrics
* scx_layered: Add layer CPU cost to dump
* scx_layered: Add layer name to bpf
* scx_lavd: Fix typos
* scx_lavd: Correct the type of taskc within lavd_dispatch()
* scx_layered: Refactor dispatch
* scx_lavd: boost task's latency criticality when pinned to a cpu
* scx_lavd: move reset_lock_futex_boost() to ops.running()
* scx_lavd: calculate task's latency criticality in the direct dispatch path
* scx_layered: Add cost accounting
* scx_lavd: kick CPU explicitly at the ops.enqueue() path
* scx_lavd: yield for preemption only when a task is ineligible
* layered/timers: support verifying on older kernels and fix logic
* make timerlib work the best it can with tooling
* fix lsp to work after multiarch support
* scx_layered: Fix declarations in timer
* scx_layered: Add monitor
* scx_lavd: fix uninitialized memory access comp_preemption_info()
* add retries to kernel clone step
* scx_loader: introduce configuration
* Revert "fix ci errors due to __str update in kfunc signature"
* Downgrade version to 1.2.0
* Set minimal meson version to silent new warnings
* scx_lavd: proactively kick a CPU at the ops.enqueue() path
* scx_lavd: change the greedy penalty function
* scx_lavd: change how to use the context information for latency criticality
* ci: enable SCHED_MC in the virtme-ng kernel config
* ci: enable verbose mode when testing schedulers
* Add script for generating per-arch vmlinux.h files
* fix ci errors due to __str update in kfunc signature
* scx_layered: Add timer helpers
* scx_lavd: remove the time restriction in preemption
* scx_lavd: incorporate task's weight to latency criticality
* scx_lavd: respect a chosen cpu even if it is not idle
* scx_lavd: add ops.cpu_release()
* scx_lavd: fix/work around a verifier error
* layered: bpf: add layer kind to layer
* scx_lavd: refill time slice once for a lock holder
* scx_lavd: direct dispatch when there is an idle CPU
* scx_lavd: give more penalty to long-running tasks
* scx_layered: do not attempt ctx lookup on tasks exited before running on scx
* Use per-arch vmlinux.h v2
* scx_rustland: Adjust task's vruntime budget based on latency weight
* scx_rusty: Cleanup cpumask casting
* scx_lavd: remove unnecessary load tracking
* scx_lavd: add missing reset_lock_futex_boost()
* scx_layered: Read CPU topology for building CpuPool
* scx_bpfland: tune default settings
* scx_bpfland: rework lowlatency mode to adjust tasks priority
* remove apt fast from ci setup
* scx_rustland_core: update documentation about the new API
* layered: make disable_topology arg require equals
* layered: add RandomTopo layer growth algorithm
* Revert "Use per-arch vmlinux.h"
* Revert "Add vmlinux.h for multiple arch"
* scx_rlfifo: operate in a more work-conserving way
* scx_rustland: clarify EDF scheduling
* scx_rustland: smooth vruntime update
* scx_rustland: use built-in nvcsw metrics
* scx_rustland_core: bump up version to 2.2.2
* scx_rustland_core: get rid of the SCX_ENQ_WAKEUP logic
* scx_rustland_core: kick an idle CPU after global dispatch
* scx_rustland_core: keep CPUs alive with pending tasks
* scx_layered: Refactor refresh cpumasks
* scx_rustland_core: restart scheduler on hotplug events
* scx_rustland_core: allow user-space scheduler to run indefinitely
* scx_rustland_core: pass nvcsw, slice and dsq_vtime to user-space
* scx_rustland_core: update CPU idle selection logic
* scx_mitosis: Handle pinned tasks
* scx_layered: Refactor layer iteration
* scx_mitosis: handle enqueue() on !wakeup
* Update developer guide with Perfetto info
* scx_layered: Add LLC integration test
* scripts: Convert sched ftrace helper scripts to python
* scx_layered: Add topology integration test
* scx_mitosis: Fix static assertion of scx_bpf_task_cgroup failing __weak check
* Add vmlinux.h for multiple arch
* Use per-arch vmlinux.h
* scripts: Add ftrace helper scripts
* scx_lavd: fix int overflow in calculating avg_lat_cri
* scx_lavd: remove unnecessary downscaling in deadline calculation
* scripts: Add vtime distribution script
* scx_lavd: do not inspect scx_lavd process itself
* scx_bpfland: get rid of per-CPU DSQs
* scx_bpfland: clarify error code returned by pick_idle_cpu()
* scx_bpfland: always refill task timeslice in ops.dispatch()
* scx_bpfland: pick a random idle CPU when prev_cpu is not valid
* scx_layered: Cleanup non topology path
* layered: make default value for disable_topology dynamic
* layered: move configuration into library component
* scx_rustland_core: use handle_mm_fault kprobe
* scx_lavd: fix incorrect task comparison for preemption
* make lsp work nice on util include
* scx_layered: fix lsp etc on util.bpf.c
* scx_lavd: do not preempt while holding a lock
* scx_lavd: boost latency criticality when a task holds a lock
* scx_lavd: tracing all blocking locks and futexes
* [layered] Implement reverse weight DSQ algorithm
* scx_layered: Change default DSQ iter algo
* scx_layered: Cleanup topology preempt path
* scx_bpfland: fix cpumask initialization error
* scx_layered: Refactor topo preemption
* setup matrix job to run key paths of layered through verifier/stress test
* scx_layered: Cleanup debug messages
* scx_layered: Fix verifier errors
* mitosis: apply autoformatting
* layered: attempt to work steal from own llc before others
* mitosis: fix build
* scx_layered: Update docs for layer_preempt_weight_disable
* scx_layered: Rename load_adj statistic
* layered: split dispatch into no_topo version
* scx_layered: Update CI to show stats
* enable configuring layer iteration when no topo
* [rusty] Fix load stats when host is under-utilized
* update stress test settings to constants used in test_scheds
* Revert "Merge pull request #746 from likewhatevs/layered-delay"
* scx_layered: Make stress-ng non exclusive in example
* layered: cleanup warnings in bpf compilation
* scx_layered: Add verbose output on stress tests
* scx_layered: Add verbose output on CI logs
* scx_layered: Add stress-ng example layer
* scx_bpfland: prevent per-CPU DSQ stall with per-CPU kthreads
* scx_layered: Add doc comment to CpuPool
* formatting
* formatting
* formatting
* flatten and simplify dispatch
* scx_rusty: Fix BPF crash during CPU hotplug
* docs: Add Gentoo installation instructions
* scx_layered: Update idle topology selection order
* scx_layered: Improve perf on non topo aware paths
* scx_layered: Add idle smt layer configuration
* scx_layered: Use idle smt mask for idle selection
* scx_layered: Add big cpumask
* Drop journald from libalpn hooks
* scx_bpfland: rework lowlatency mode
* scx_loader: Add SwitchScheduler methods to DBUS interface
* Replace PID with Task Pointer in Rusty
* scx_layered: Add better debug output of iter algo
* scx_layered: Cleanup layer growth weight limits
* scx_layered: Cleanup layer infeasible weight calc
* scx_layered: Cleanup from code review
* scx_layered: Add default dsq iter algo
* scx_layered: Add DSQ iteration algo
* scx_layered: Refactor stats variable names
* scx_layered: Use dcycle for infeasible weights
* scx_layered: Cleanup unused metrics
* scx_layered: Cargo fmt
* scx_layered: Add weighted layer DSQ iteration
* scx_layered: Add flags for growth and preemption
* scx_layered: Remove LoadLedger from stats
* scx_layered: Refactor load calculation
* scx_layered: Add weights to userspace layer config
* scx_layered: Add stats for adjusted layer weights
* scx_layered: Add layer weights
* scx_layered: Add Reverse layer growth algo
* scx_layered: Break up layer_core_order function
* scx_layered: Move layer core growth logic to separate module
* scx_lavd: split main.bpf.c into multiple files
* scx_layered: Add per layer time slices to stats
* scx_rusty: fix single dom short-circuit
* have bpf builds use separate cache
* add link to generated docs in developer guide
* update urls to point to bpf-next
* remove lint step from bpf workflow
* enable periodic build/test against bpf-next
* version: Cargo.lock update
* scx_lavd: support CPU hotplug correctly
* scx_lavd: fix incorrect preemtability test
OBS-URL: https://build.opensuse.org/request/show/1227554
OBS-URL: https://build.opensuse.org/package/show/Base:System/scx?expand=0&rev=2
