xref: /arkcompiler/runtime_core/static_core/runtime/options.yaml

# Copyright (c) 2021-2025 Huawei Device Co., Ltd.
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.

module:
  name: runtime
  namespace: ark

options:
- name: boot-panda-files
  type: arg_list_t
  default:
      - $ORIGIN/../pandastdlib/arkstdlib.abc
  description: Boot panda files separated by colon
  delimiter: ":"

- name: panda-files
  type: arg_list_t
  default: []
  description: Panda files separated by colon which is not within boot-panda-files
  delimiter: ":"

- name: boot-intrinsic-spaces
  type: arg_list_t
  default:
    - core
  possible_values:
    - core
  description: Load specified intrinsic spaces. This option is for fine tuning. Use load-runtimes instead
  delimiter: ":"

- name: boot-class-spaces
  type: arg_list_t
  default:
    - core
  possible_values:
    - core
  description: Load specified class spaces. This option is for fine tuning. Use load-runtimes instead
  delimiter: ":"

- name: runtime-type
  type: std::string
  default: core
  possible_values:
    - core
  description: Runtime type. This option is for fine tuning. Use load-runtimes instead

- name: load-runtimes
  type: arg_list_t
  default:
    - core
  possible_values:
    - core
  description: Load specified class and intrinsic spaces and define runtime type
  delimiter: ":"

- name: runtime-compressed-strings-enabled
  type: bool
  default: true
  description: Enable/disable compressed strings

- name: explicit-concurrent-gc-enabled
  type: bool
  default: true
  description: Enable/disable explicit GC run in concurrent

- name: run-gc-in-place
  type: bool
  default: false
  description: Enable/disable running GC only in place

- name: full-gc-bombing-frequency
  type: uint32_t
  default: 0
  description: Switch on running Full GC collection instead of the one in the trigger queue in the GC thread. Value means frequency - each N collection in the GC thread will be a full collection. Can't be use with run-gc-in-place

- name: run-gc-every-safepoint
  type: bool
  default: false
  description: Run GC on every safepoint. GC will be run only if we use gc-trigger-type=debug and run-gc-in-place=true

- name: compiler-enable-jit
  type: bool
  default: true
  description: Enables/disables JIT compiler

- name: compiler-hotness-threshold
  type: uint32_t
  default: 3000
  description: Threshold for "hotness" counter of the method after that it will be compiled

- name: compiler-profiling-threshold
  type: uint32_t
  default: 1500
  description: Threshold to start method profiling.

- name: compiler-enable-osr
  type: bool
  default: true
  description: Enables/disables OSR compilation

- name: debugger-library-path
  type: std::string
  default: ""
  description: Path to debugger library

- name: debugger-enable
  type: bool
  default: false
  description: Is debugger enabled on runtime create

- name: sampling-profiler-create
  type: bool
  default: false
  description: Is the sampling profiler created and available during runtime execution

- name: sampling-profiler-startup-run
  type: bool
  default: false
  description: Is the sampling profiler runned on runtime startup

- name: sampling-profiler-interval
  type: uint32_t
  default: 500
  description: Sampling profiler interval in microseconds

- name: sampling-profiler-output-file
  type: std::string
  default: ""
  description: Name of file to collect trace in .aspt format

- name: debugger-port
  type: uint32_t
  default: 19015
  description: Port for serving debugger clients via socket

- name: debugger-break-on-start
  type: bool
  default: false
  description: Whether to break threads on creation or not

- name: log-components
  type: arg_list_t
  default:
    - all
  possible_values:
    - all
    - alloc
    - mm-obj-events
    - classlinker
    - common
    - core
    - gc
    - gc_trigger
    - reference_processor
    - interpreter
    - compiler
    - pandafile
    - memorypool
    - runtime
    - trace
    - debugger
    - interop
    - verifier
    - compilation_queue
    - aot
    - events
    - scheduler
    - coroutines
  description: '[DEPRECATED] Option was moved to libpandabase/options.yaml, use base_options::Options instead of runtime options.'
  delimiter: ":"
  deprecated: true

- name: compiler-nullcheck
  type: uint64_t
  default: 1
  possible_values:
    - 0
    - 1
  description: Dfx option for compiler-nullcheck

- name: reference-dump
  type: uint64_t
  default: 1
  possible_values:
    - 0
    - 1
  description: Dfx option for reference-dump

- name: signal-catcher
  type: uint64_t
  default: 1
  possible_values:
    - 0
    - 1
  description: Dfx option for signal-catcher

- name: signal-handler
  type: uint64_t
  default: 1
  possible_values:
    - 0
    - 1
  description: Dfx option for signal-handler

- name: sigquit-flag
  type: uint64_t
  default: 1
  possible_values:
    - 0
    - 1
  description: Dfx option for sigquit

- name: sigusr1-flag
  type: uint64_t
  default: 1
  possible_values:
    - 0
    - 1
  description: Dfx option for sigusr1

- name: sigusr2-flag
  type: uint64_t
  default: 1
  possible_values:
    - 0
    - 1
  description: Dfx option for sigusr2

- name: mobile-log-flag
  type: uint64_t
  default: 1
  possible_values:
    - 0
    - 1
  description: Dfx option for mobile-log

- name: dfx-log
  type: uint64_t
  default: 0
  possible_values:
    - 0
    - 1
  description: Dfx option for dfx log

- name: gc-type
  type: std::string
  default: g1-gc
  possible_values:
    - epsilon
    - epsilon-g1
    - stw
    - gen-gc
    - g1-gc
    - cmc-gc
  description: Type of used GC

- name: ic-enable
  type: bool
  default: true
  description: switch of inline cache

- name: gc-trigger-type
  type: std::string
  default: heap-trigger
  possible_values:
    - heap-trigger-test
    - heap-trigger
    - adaptive-heap-trigger
    - trigger-heap-occupancy
    - no-gc-for-start-up
    - debug
    - debug-never
    - pause-time-goal-trigger
  description: Type of used GC trigger

- name: log-detailed-gc-info-enabled
  type: bool
  default: false
  description: If print INFO log to get more detailed information in GC.

- name: log-detailed-gc-compaction-info-enabled
  type: bool
  default: false
  description: If print INFO log to get more detailed compaction/promotion information per region in GC.

- name: gc-trigger-percent-threshold
  type: uint32_t
  default: 15
  description: Percent of heap for threshold increasing

- name: gc-trigger-adaptive-multiplier
  type: uint32_t
  default: 3
  description: Adjust jump from too frequent gc. New jump calculate as gc-trigger-adaptive-multiplier * max-max-extra-heap-size

- name: max-trigger-percent
  type: uint32_t
  default: 45
  description: Percent of maximum heap size is used as threshold for trigger-heap-occupancy

- name: skip-startup-gc-count
  type: uint32_t
  default: 10
  description: No stw gc count for no-gc-for-start-up

- name: gc-use-nth-alloc-trigger
  type: bool
  default: false
  description: Allow to trigger GC on n-th allocation

- name: init-young-space-size
  type: uint64_t
  default: 4194304
  description: Initial young space size gen-gc

- name: young-space-size
  type: uint64_t
  default: 4194304
  description: Young space size of gen-gc

- name: young-shared-space-size
  type: uint64_t
  default: 524288
  description: How much space in young-gen are shared (this space is not used for tlabs)

- name: min-extra-heap-size
  type: uint64_t
  default: 5242880
  description: Minimum extra heap size for trigger gc

- name: max-extra-heap-size
  type: uint64_t
  default: 8388608
  description: Maximum extra heap size for trigger gc

- name: gc-debug-trigger-start
  type: uint64_t
  default: 0
  description: First iteration to start returning true in debug trigger

- name: gc-dump-heap
  type: bool
  default: false
  description: Dump heap before and after GC

- name: gc-workers-count
  type: uint32_t
  # default value depends on a specific device model
  default_target_specific:
    HL1CMSM: 2  # mate 60 pro
    ohos: 2 # devboard
    HN1NOAHM: 2 # mate 40 pro
    default: 2
  description: Set number of additional GC helper threads. GC uses these threads to speed up collection doing work in parallel.

- name: manage-gc-threads-affinity
  type: bool
  default: true
  description: Manage GC Threads affinity on cores

- name: use-weak-cpu-for-gc-concurrent
  type: bool
  default: false
  description: Set GC Thread on weak cores during concurrent phases

- name: g1-region-garbage-rate-threshold
  type: uint32_t
  default: 52
  description: Specify garbage rate threshold of a tenured region to be included into a mixed collection.

- name: g1-number-of-tenured-regions-at-mixed-collection
  type: uint32_t
  default: 20
  description: Specify a max number of tenured regions which can be collected at mixed collection in G1GC.

- name: g1-track-freed-objects
  type: std::string
  default: default
  possible_values:
  - default
  - 'true'
  - 'false'
  description: Specify whether G1GC tracks object removing (i.e. update objects count in memstats and log removed objects). Default means true on Release and false otherwise. If the value is true G1GC doesn't delete classes in concurrent phase because it needs to know object size.

- name: gc-marking-stack-new-tasks-frequency
  type: uint64_t
  default: 10000000
  description: Limit the creation rate of tasks during marking in nanoseconds. if average task creation during marking is less than this value - it increases the stack size limit to create tasks less frequently.

- name: gc-root-marking-stack-max-size
  type: uint32_t
  default: 186
  description: The max stack size for root marking in main thread, if it exceeded, we will send a new task to workers. 0 means unlimited.

- name: gc-workers-marking-stack-max-size
  type: uint32_t
  default: 200
  description: The max stack size for marking stage proceed in workers threads, if it exceeded, we will send a new task to other workers. 0 means unlimited.

- name: gc-parallel-marking-enabled
  type: bool
  default: true
  description: Enable parallel marking in GC if it is supported (now it is G1 and STW). If we don't have gc workers, this options will be ignored.

- name: gc-parallel-compacting-enabled
  type: bool
  default: true
  description: Enable parallel compacting in GC if it is supported (now it is G1). If we don't have gc workers, this options will be ignored.

- name: gc-parallel-ref-updating-enabled
  type: bool
  default: true
  description: Enable parallel references updating in GC if it is supported (now it is G1). If we don't have gc workers, this options will be ignored.

- name: g1-single-pass-compaction-enabled
  type: bool
  default: true
  description: Enable single pass compaction mode for G1 GC if it is supported by VM implementation

- name: reference-processor-enable
  type: bool
  default: true
  description: Enables/disables ReferenceProcessor

- name: log-level
  type: std::string
  default: error
  possible_values:
  - debug
  - info
  - warning
  - error
  - fatal
  description: '[DEPRECATED] Option was moved to libpandabase/options.yaml, use base_options::Options instead of runtime options.'
  deprecated: true

- name: log-stream
  type: std::string
  default: std
  possible_values:
  - std
  - file
  - fast-file
  - dummy
  description: '[DEPRECATED] Option was moved to libpandabase/options.yaml, use base_options::Options instead of runtime options.'
  deprecated: true

- name: log-file
  type: std::string
  default: ""
  description: '[DEPRECATED] Option was moved to libpandabase/options.yaml, use base_options::Options instead of runtime options.'
  deprecated: true

- name: gc-enable-tracing
  type: bool
  default: false
  description: enables/disables tracing gc

- name: g1-enable-concurrent-update-remset
  type: bool
  default: true
  description: enables/disables concurrent update remset thread

- name: g1-min-concurrent-cards-to-process
  type: uint32_t
  default: 2
  description: Minimum number of cards to process from queue in update-remset-thread. Higher number consumes less CPU, but can cause higher pause.

- name: g1-hot-cards-processing-frequency
  type: uint32_t
  default: 7
  description: Frequency of proccessing hot cards in update-remset-thread

- name: g1-promotion-region-alive-rate
  type: uint32_t
  default: 75
  description: Minimum percentage of alive bytes in young region to promote it into tenured without moving. 100 value means never.

- name: g1-full-gc-region-fragmentation-rate
  type: uint32_t
  default: 10
  description: Maximum fragmentation percentage for a tenured region with no garbage inside to skip it on full collection. If the region has bigger fragmentation, we will collect it during a full collection in spite of it has no garbage inside. 0 value means we will collect every region.

- name: g1-pause-time-goal
  description: Enable G1 pause time goal
  sub_options:
  - name: max-gc-pause
    type: uint32_t
    default: 10
    description: Maximum sum of G1 GC pauses during gc-pause-interval in milliseconds
  - name: gc-pause-interval
    type: uint32_t
    default: 11
    description: Time interval for max-gc-pause in milliseconds

- name: distributed-profiling
  type: bool
  default: false
  description: Enable/disable collection of information for distributed profiling

- name: aot-file
  type: std::string
  default: ""
  description: path to aot file

- name: aot-files
  type: arg_list_t
  default: []
  description: List of aot files to be loaded
  delimiter: ':'

- name: aot-verify-abs-path
  type: bool
  default: true
  description: Use absolute path to panda files in Aot verification

- name: adaptive-tlab-size
  type: bool
  default: true
  description: Enable/disable adaptive TLAB size according to runtime allocation statistics

- name: init-tlab-size
  type: uint64_t
  default: 4096
  description: Initial TLAB size and maximum size of object to be allocated in TLAB. TLAB size remains constant unless adaptive-tlab-size is enabled. Also, this value is used for YoungAllocMaxSize

- name: max-tlab-size
  type: uint64_t
  default: 16384
  description: Max TLAB size. Used when adaptive-tlab-size is enabled

- name: use-tlab-for-allocations
  type: bool
  default: true
  description: Enable/disable TLAB using for object allocations. Now, it is ignored for all GCs except GenGC. Temporary option for quick switch between modes.

- name: init-heap-size-limit
  type: uint64_t
  default: 536870912
  description: Initial size of pool used for objects (Xms). If it was not set explicitly then it set as heap-size-limit

- name: heap-size-limit
  type: uint32_t
  default: 536870912
  description: Max heap size

- name: internal-memory-size-limit
  type: uint64_t
  default: 2147483648
  description: Max internal memory used by the VM

- name: frames-memory-size-limit
  type: uint64_t
  default: 2147483648
  description: Max internal memory used for frames in the VM

- name: min-heap-free-percentage
  type: uint32_t
  default: 30
  description: Minimum percentage of free space in heap. Must be 0 <= min-heap-free-percentage <= max-heap-free-percentage <= 100

- name: max-heap-free-percentage
  type: uint32_t
  default: 70
  description: Maximum percentage of free space in heap. Must be 0 <= min-heap-free-percentage <= max-heap-free-percentage <= 100

- name: code-cache-size-limit
  type: uint64_t
  default: 33554432
  description: The limit for compiled code size.

- name: compiler-memory-size-limit
  type: uint64_t
  default: 268435456
  description: Max memory used by the compiler

- name: print-memory-statistics
  type: bool
  default: false
  description: Enable/disable printing memory statistics in the end of the program

- name: print-gc-statistics
  type: bool
  default: false
  description: Enable/disable printing gc statistics in the end of the program

- name: no-async-jit
  type: bool
  default: false
  description: Perform compilation in the main thread or in parallel worker

- name: compiler-queue-type
  type: std::string
  default: counter-priority
  possible_values:
    - simple
    - counter-priority
    - aged-counter-priority
  description: Type of compiler queue

- name: compiler-task-life-span
  type: uint32_t
  default: 500
  description: A time in milliseconds of task life span in compiler counter priority queue

- name: compiler-queue-max-length
  type: uint32_t
  default: 100
  description: Max length of compiler queue

- name: compiler-epoch-duration
  type: uint32_t
  default: 500
  description: Duration in milliseconds of an epoch in compiler aged counter priority queue

- name: compiler-death-counter-value
  type: uint32_t
  default: 500
  description: Minimum value of aged counter, which will be still considered

- name: limit-standard-alloc
  type: bool
  default: false
  description: Limit standard allocations usage in runtime

- name: internal-allocator-type
  type: std::string
  default: default
  possible_values:
    - panda_allocators
    - malloc
    - default
  description: Default means using malloc for Release, and panda allocators for Debug builds. If we use malloc, we will ignore limit-standard-alloc option.

- name: finalizer-timeout-ms
  type: uint32_t
  default: 10000
  description: Set finalizer daemon watchdog time

- name: properties
  type: arg_list_t
  default:
    - user.locale=en-US
  description: command-line property settings of VM

- name: events-output
  type: std::string
  default: none
  possible_values:
    - none
    - csv
    - memory
    - log
  description: Specify events output

- name: events-file
  type: std::string
  default: events.csv
  description: Path to events output file (if needed)

- name: mem-alloc-dump-exec
  type: bool
  default: false
  description: true if need to dump

- name: mem-alloc-dump-file
  type: std::string
  default: mem_dump.csv
  description: Path to mem dump file (if needed)

- name: startup-time
  type: bool
  default: false
  description: Print the start time of command execution

- name: max-free
  type: uint32_t
  default: 8388608
  description: Trigger native memory recycling watermark, default 8 M

- name: native-gc-trigger-type
  type: std::string
  default: simple-strategy
  possible_values:
    - no-native-gc-trigger
    - simple-strategy
  description: Type of native gc trigger

- name: profiler-enabled
  type: bool
  default: true
  description: whether open the pgo profiler when jit diabled

- name: profilesaver-sleeping-time-ms
  type: uint32_t
  default: 2000
  description: Set wake up period time of daemon thread for profile saver

- name: profilesaver-delta-number-threshold
  type: uint64_t
  default: 0
  description: delta number threshold of newly added method or class for profile saver

- name: profilesaver-enabled
  type: bool
  default: false
  description: whether open the profilesaver functions

- name: start-as-zygote
  type: bool
  default: false
  description: whether current vm is zygote

- name: verification-mode
  type: std::string
  default: disabled
  possible_values:
    - disabled
    - on-the-fly
    - ahead-of-time
  description: Verification mode.

- name: verification-update-cache
  type: bool
  default: true
  description: Whether to update verification result cache

- name: verification-threads
  type: uint32_t
  default: 1
  description: number of verification threads

- name: verification-config-file
  type: std::string
  default: "default"
  description: filename with debug config for verifier

- name: verification-cache-file
  type: std::string
  default: ""
  description: verifier cache filename

- name: enable-fast-heap-verifier
  type: bool
  default: true
  description: Enable fast heap verifier, which is faster than ordinary heap verifier but uses more internal memory

- name: heap-verifier
  type: arg_list_t
  default: []
  possible_values:
    - pre
    - into
    - post
    - before_g1_concurrent
    - fail_on_verification
  description: parameters for heap verification
  delimiter: ":"

- name: max-global-ref-size
  type: uint32_t
  default: 65535
  description: max size of global reference storage

- name: global-reference-size-check-enabled
  type: bool
  default: false
  description: whether open the global reference size check

- name: concurrent-gc-enabled
  type: bool
  default: true
  description: whether concurrent GC is enabled

- name: snapshot-serialize-enabled
  type: bool
  default: false
  description: whether snapshot serialize is enabled

- name: snapshot-deserialize-enabled
  type: bool
  default: true
  description: whether snapshot deserialize is enabled

- name: snapshot-file
  type: std::string
  default: "/system/etc/snapshot"
  description: snapshot file

- name: framework-abc-file
  type: std::string
  default: "strip.native.min.abc"
  description: snapshot file

- name: boot-an-location
  type: std::string
  default: "/data/service/el1/public/for-all-app/framework_ark_cache"
  description: default path to .an files for boot panda files, try to load ARK .an file from this dir if not empty when enable-an

- name: enable-an
  description: Try to load ARK .an file for boot panda files from boot-an-location or base on abc file location
  sub_options:
  - name: force
    type: bool
    default: false
    description: Crash if there is no .an file for both location based on .abc file and boot-an-location

- name: verify-call-stack
  type: bool
  default: false
  description: Verify call stack before using of the StackWalker

- name: safepoint-backtrace
  type: bool
  default: false
  description: Print backtrace each time a thread gets suspended

- name: aot-string-gc-roots-limit
  type: uint64_t
  default: 1000
  description: Maximum amount of AOT string PLT-slots registered as GC roots

- name: resolve-string-aot-threshold
  type: uint32_t
  default: 1
  description: Minimum amount of ResolveStringAot calls before storing string pointer into PLT-slot

- name: icu-data-path
  type: std::string
  default: "default"
  description: Path to generated icu data file

- name: ark-aot
  type: bool
  default: false
  description: check if runtime start by ark_aot

- name: intrinsics-blacklist
  type: arg_list_t
  default: []
  description: List of methods for which intrinsics should not be linked
  delimiter: ","

- name: verify-entrypoints
  type: bool
  default: false
  description: Verify compiler entrypoints

- name: ignore-daemon-memory-leaks-threshold
  type: uint32_t
  default: 0
  description: Ignore daemon memory leaks during runtime destruction with the specified threshold

- name: interpreter-type
  type: std::string
  default: llvm
  possible_values: [cpp, irtoc, llvm]
  description: Interpreter implementation type

- name: abc-layout-profile-location
  type: std::string
  default: "/data/profile/"
  description: Specify the location of abc layout profile

- name: intrusive-test
  type: uint32_t
  default: 0
  description: Numerical identifier of an intrusive test

- name: profile-output
  type: std::string
  default: "profile.bin"
  description: Specify the location the collected profile information

- name: call-profiling-table-size
  type: uint32_t
  default: 1000
  description: Size of call profiling table size

- name: coroutine-stack-size-pages
  type: uint32_t
  default: 64
  description: defines stack size for stackful coroutines (in number of pages)

- name: coroutines-stack-mem-limit
  type: uint64_t
  default: 268435456
  description: defines the total amount of memory that can be used for stackful coroutine stacks allocation (in bytes)

- name: use-coroutine-pool
  type: bool
  default: false
  description: caches coroutine and coroutine context instances in a pool to speedup creation/destruction

- name: workers-type
  type: std::string
  default: taskmanager
  possible_values:
  - threadpool
  - taskmanager
  description: Set type of additional VM workers. VM uses these workers to speed up internal algorithms (JIT, GC, etc.)

- name: taskmanager-workers-count
  type: uint32_t
  # default value depends on a specific device model
  default_target_specific:
    HL1CMSM: 3 # mate 60 pro
    ohos: 2 # devboard
    HN1NOAHM: 4 # mate 40 pro
    default: 4
  description: Number of worker threads for the taskmanager. Option is used only for --worker-type=taskmanager

- name: task-stats-type
  type: std::string
  default: no-task-stats
  possible_values:
  - no-task-stats
  - light-task-stats
  description: Set type of task stats that will collect task manager metrics, if use set `no-task-stats` there will be no stats collection

- name: load-in-boot
  type: bool
  default: false
  description: Load entrypoint panda file in boot-panda-files

- name: use-string-caches
  type: bool
  default: true
  description: Whether to create cache for strings created from integer/floating point numbers.