- Install cargo-msrv using stable toolchain. by @wks in #1215
- Update CI macos image by @qinsoon in #1216
- Performance history canary by @wks in #1209
- Change the default testing branch for Julia tests by @qinsoon in #1221
- Fix auto merge branches by @qinsoon in #1230
- Document the policy about performance testing environment and epochs by @qinsoon in #1206
- Fix a race between forwarding bits and VO bits. by @wks in #1214
- Remove space for nogc link by @Pat-Lafon in #1217
- Make env_logger an optional dependency by @wks in #1226
- Use modern syntax for optional dependencies by @wks in #1229
- Fix nightly build and add
inline
attributes to{un,}likely
by @k-sareen in #1228
- @Pat-Lafon made their first contribution in #1217
Full Changelog: https://github.com/mmtk/mmtk-core/compare/v0.28.0...v0.29.0
- Require ObjectReference to point inside object by @wks in #1195
- Return if a GC ran or not for
handle_user_collection_request
by @k-sareen in #1205
- Update ci-perf-kit to 0.7.6 by @qinsoon in #1199
- Update ci-perf-kit to plot epoch by @qinsoon in #1208
- Skip iterating through unallocated memory for
MonotonePageResource
by @k-sareen in #1189 - Fix find_last_non_zero_bit, and align metadata address before converting to data address. by @qinsoon in #1188
- Panic message and outdated comments by @wks in #1191
- Filter the result for find_prev_non_zero_value_fast to make sure it is in the search range by @qinsoon in #1192
- Assert OBJECT_REF_OFFSET_LOWER_BOUND invariant by @wks in #1194
- Minor fixes by @wks in #1196
- Add a feature immix_stress_copying by @qinsoon in #1202
- Update migration guide for #1205 by @qinsoon in #1211
Full Changelog: https://github.com/mmtk/mmtk-core/compare/v0.27.0...v0.28.0
- Clear pin bits by @udesou in #1166
- Set the unlog bits for objects in the bootimage by @k-sareen in #1168
- Fix bug in line mark when
MARK_LINE_AT_SCAN_TIME=false
by @k-sareen in #1171
- Require object reference to be aligned by @qinsoon in #1159
- Internal pointer support by @qinsoon in #1165
- Heap traversal by @wks in #1174
- Parallelize and workaround CI style checks by @wks in #1184
- Extensible eBPF timeline attributes by @wks in #1162
- Only map with executable permissions when using code space by @k-sareen in #1176
- Fix style check for Rust 1.80 by @qinsoon in #1178
- Extra assertions for mutators and epilogues by @wks in #1182
- Refactor iterate_meta_bits by @wks in #1181
Full Changelog: https://github.com/mmtk/mmtk-core/compare/v0.26.0...v0.27.0
- Clear side forwarding bits properly by @wks in #1138
- Fix mark bit clearing in PrepareChunkMap by @wks in #1148
- Let MarkSweepSpace use BlockPageResource by @wks in #1150
- Add SweepChunk to native MarkSweepSpace by @wks in #1158
- Rename edge to slot by @wks in #1134
- Fix broken link and stale descriptions in doc by @wks in #1139
- API migration guide. by @wks in #1133
- Add back
DummyVM
as a part of the porting guide. Minor changes to MMTk initialization in the porting guide. by @qinsoon in #1142 - Fix broken link to mmtk.h by @wks in #1149
- Fix clippy warnings for Rust 1.79 by @wks in #1151
- Display number of slots in timeline visualization by @wks in #1154
Full Changelog: https://github.com/mmtk/mmtk-core/compare/v0.25.0...v0.26.0
- Allow roots to be pinned for StickyImmix nursery collections by @k-sareen in #1108
- Check
FULL_NURSERY_GC
for StickyImmix as well by @k-sareen in #1124
- Remove data race for marksweep block lists by @qinsoon in #1112
- Remove coordinator and support forking by @wks in #1067
- Allow nursery size to be proportional to the heap size by @qinsoon in #1087
- Remove NULL ObjectReference by @wks in #1064
- Fix write barrier parameter type by @wks in #1130
- Allow querying if the current GC may move objects. by @wks in #1128
- Use host toolchain and explicit target in CI script by @wks in #1109
- Fix ignores in merge check by @qinsoon in #1117
- Fix the openjdk test name in merge check by @qinsoon in #1119
- Bump dependencies after version 0.24 by @wks in #1105
- Don't use
PROT_EXEC
withmmap
on macos by @eileencodes in #1110 - Rename associated types in GCWorkContext by @wks in #1116
- Add missing pinning metadata to mock_vm by @wks in #1121
- Use to_address for SFT access by @wks in #1122
- Ensure reference objects are either traced or cleared by @wks in #1126
- Dump process map on
mmap
failure and fix typos by @k-sareen in #1127 - Fix compiler errors and warnings for Rust 1.78.0 by @wks in #1132
- Remove fl_map and refactor FreeListPageResoure by @wks in #953
- @eileencodes made their first contribution in #1110
Full Changelog: https://github.com/mmtk/mmtk-core/compare/v0.24.0...v0.25.0
- Check and fix broken links in docs by @qinsoon in #1085
- Allow binding to implement GC trigger by @qinsoon in #1083
- Resolve issues for unit tests on 32 bits by @qinsoon in #1095
- Use SFTDenseChunkMap on 64bits when vm_space is enabled by @qinsoon in #1094
- Fix warnings with the stable Rust toolchain by @wks in #1099
- Use proper logging for the treadmill by @wks in #1101
Full Changelog: https://github.com/mmtk/mmtk-core/compare/v0.23.0...v0.24.0
- Ask from binding if GC is disabled by @udesou in #1075
- Stop using the deprecated set-output in our CI configs by @qinsoon in #1059
- Rename CI tests to minimal/extended tests by @qinsoon in #1073
- Properly get PR number for merge group by @qinsoon in #1077
- MMTk Enhancement Proposal by @wks in #1056
- Timeline visualization tool by @wks in #1022
- Add a document for project principles by @qinsoon in #1069
- Fix un-cleared VO bits for contiguous monotone PR by @wks in #1071
- Use live memory to estimate collection pages in mem balancer by @qinsoon in #1050
- Fix issues in sanity GC by @qinsoon in #1079
Full Changelog: https://github.com/mmtk/mmtk-core/compare/v0.22.1...v0.23.0
- Revert a mistake in v0.22 that prevents cargo publish by @qinsoon in #1065
Full Changelog: https://github.com/mmtk/mmtk-core/compare/v0.22.0...v0.22.1
- Flush mutator buffers in
destroy_mutator
by @k-sareen in #1045 - NULL and movement check in process_edge by @wks in #1032
- Mention revoking TLABs of all mutators after a GC by @k-sareen in #1018
- Add missing docs for the vm module by @qinsoon in #1021
- Add missing docs for some util modules by @qinsoon in #1024
- Add missing docs for the rest of the util module by @qinsoon in #1026
- Add missing docs for the rest of the code base (merge after #1026) by @qinsoon in #1028
- Update some stale comments by @k-sareen in #1036
- Use new running scripts in ci-perf-kit by @qinsoon in #1005
- Upload perf build and log as artifacts (merge after #1005) by @qinsoon in #1011
- Fix multiple issues in recent CI changes by @qinsoon in #1014
- Fix CI for OpenJDK nogc config by @qinsoon in #1019
- Use ci-perf-kit 0.7.4: ignore runs with unmatched key/value in the logs. by @qinsoon in #1025
- Fix Ruby binding test repo checkout by @wks in #1048
- Post-release dependency version bump for v0.21.0 by @wks in #1013
- Introduce MockVM by @qinsoon in #1049
Full Changelog: https://github.com/mmtk/mmtk-core/compare/v0.21.0...v0.22.0
- Move some states from BasePlan to GlobalState by @qinsoon in #949
- Extract set_collection_kind and set_gc_status by @qinsoon in #957
- Expose
is_emergency_collection
to VM bindings by @wks in #997
- Clear all side forwarding bits by @wks in #974
- Stop using the name "ForwardingWord" by @wks in #976
- Implement get_gc_trigger() for LockFreeImmortalSpace by @qinsoon in #1003
- Expose alloc_slow. Add a section in user guide about allocation optimization by @qinsoon in #967
- Use BumpPointer::default() by @qinsoon in #993
- Fix mention of
to_object_reference()
in comment by @k-sareen in #998
- Fix rustdoc warnings in Rust 1.73 by @qinsoon in #978
- Add a section about debugging tips in user guide by @qinsoon in #987
- Update doc comments by @wks in #1000
- Add document for release by @qinsoon in #1007
- Move constants and add comments for stress copying by @wks in #1008
- Update bibtex of the tracing paper to include the DOI by @caizixian in #1009
- Fix performance baseline workflow by @qinsoon in #971
- Allow auto merge binding PRs by @qinsoon in #988
- Apply a delay before we enable auto merge on a PR by @qinsoon in #992
- Use
cargo generate-lockfile
to update JikesRVM's Cargo.lock by @qinsoon in #996
- Add option to exclude kernel events from perf results by @k-sareen in #995
Full Changelog: https://github.com/mmtk/mmtk-core/compare/v0.20.0...v0.21.0
- Refactor derive macros and add HasSpaces trait by @wks in #934
- Make MarkCompact LOS support 2nd transitive closure by @wenyuzhao in #944
- Disabling PrepareMutator from PlanConstraints by @wks in #964
- Add ExternalPageResource and allow discontiguous VM space by @qinsoon in #864
- Discontiguous mark compact space support by @wenyuzhao in #939
- Discontiguous PageProtect GC support by @wenyuzhao in #946
- Fix vo-bit reset for discontiguous space by @wenyuzhao in #948
- Boot-time configurable heap constants by @wenyuzhao in #899
- This PR enables transitively pinning (TP) objects from particular roots for Immix/StickyImmix by @udesou in #897
- Let VM control when or if to read env var options by @wks in #955
- Update doc comment of Scanning::process_weak_refs by @wks in #919
- Binding test for Ruby by @wks in #916
- Fix api-check CI by @wks in #923
- Fix default value for RUBY_BINDING_REPO by @qinsoon in #926
- Add a ready-to-merge check by @qinsoon in #910
- Run ready to merge check for PRs by @qinsoon in #928
- Remove cast ref to mut everywhere by @playXE in #893
- Benchmark Rust code by @qinsoon in #933
- Fix broken links in the tutorial by @caizixian in #936
- Update doc to add a section for LTO by @qinsoon in #937
- Add CARGO_INCREMENTAL=0 to work around clippy 1.72 bug by @qinsoon in #938
- Fix issues for cargo fmt in 1.72 by @qinsoon in #940
- Use atomic operations for SFT map and remove unsafe code by @qinsoon in #931
- Fix outdated Rust version in README by @qinsoon in #942
- Fix length of Map64::descriptor_map by @wks in #956
Full Changelog: https://github.com/mmtk/mmtk-core/compare/v0.19.0...v0.20.0
- Remove a warning in sticky immix trace_object_nursery by @qinsoon in #815
- Change default plan to GenImmix by @qinsoon in #819
- Remove redundant clear_nursery() by @tianleq in #799
- Introduce VMSpace, and allow VMSpace to be set lazily by @qinsoon in #802
- Fix an issue that the aligned VM space may not match the original location by @qinsoon in #809
- Remove some uses of mem::transmute in marksweep block by @qinsoon in #826
- Fix
is_live
for ImmixSpace by @wks in #842 - Sweep abandoned blocks in eager sweeping by @qinsoon in #830
- Fix VO bits for Immix by @wks in #849
- Fix unaligned access by @wks in #887
- Rename "alloc bit" to "valid-object bit" (VO bit), the second attempt. by @wks in #791
- Add MarkState. Use MarkState for ImmortalSpace. by @qinsoon in #796
- Change info logging to debug in ImmortalSpace by @qinsoon in #801
- Reset nursery_index in finalizable processor if we remove objects from candidates by @qinsoon in #807
- Allow bulk set side metadata by @qinsoon in #808
- Some fixes for sanity GC by @qinsoon in #806
- Use extreme assertion for metadata mapped assert by @wks in #812
- Install the missing deps for i686 CI tests by @qinsoon in #816
- Use min nursery as mem balancer's extra memory by @qinsoon in #820
- Sort dependencies in alphabetical order by @k-sareen in #822
- Use sysinfo instead of sys-info-rs by @k-sareen in #827
- Let sysinfo only load memory-related info by @wks in #836
- Add extreme assertion for barrier slow path calls by @ArberSephirotheca in #833
- Refactor: Use
Atomic<Address>
where appropriate by @ClSlaid in #843 - Update porting guide by @k-sareen in #857
- Fix typo in doc comment by @k-sareen in #859
- Replace debug_assert in side_after with assert by @qinsoon in #873
- Work around stack overflow in array_from_fn by @wks in #876
- Let ObjectReference implement Ord by @wks in #870
- Add USDT tracepoints for key GC activities by @caizixian in #883
- Fix UB in SFTMap implementations by @playXE in #879
- Run CI build/unit test with latest stable Rust toolchain by @qinsoon in #885
- Document MSRV policy by @wks in #881
- Fix performance regression test scripts by @qinsoon in #892
- Run V8 binding tests on GitHub hosted runner by @caizixian in #900
- Add tracing tools and documentation by @caizixian in #898
- Run benchmarks for more plans on OpenJDK by @qinsoon in #901
- Apply style check on auxiliary crates (macros and dummyvm) by @caizixian in #913
- Call Collection::out_of_memory if the allocation size is larger than max heap size by @qinsoon in #896
- Add a unit test for comma-separated bulk option parsing by @caizixian in #911
- Merge tutorial and porting guide into user guide by @qinsoon in #907
- Fix broken links in README and cargo doc warnings by @caizixian in #912
- Add object() in MemorySlice by @qinsoon in #798
- Remove Collection::COORDINATOR_ONLY_STW by @ArberSephirotheca in #814
- Refactor: Change
ActivePlan::mutators()
's return type by @ArberSephirotheca in #817 - Replace alloc-related
offset
type tousize
insteadisize
by @fepicture in #838 - Rename ambiguous
scan_thread_root{,s}
functions by @k-sareen in #846 - Update comments on bind_mutator by @qinsoon in #854
- Counting VM-allocated pages into heap size. by @wks in #866
- Expose Allocators type to public API by @playXE in #880
- Collect live bytes during GC by @qinsoon in #768
- Tidy up mutator scan API by @qinsoon in #875
- Implement transparent hugepage support by @caizixian in #905
- Implement AllocatorInfo by @playXE in #889
- Add comma as an alternative options string separator by @wenyuzhao in #909
- @ArberSephirotheca made their first contribution in #814
- @fepicture made their first contribution in #838
- @ClSlaid made their first contribution in #843
- @playXE made their first contribution in #880
Full Changelog: https://github.com/mmtk/mmtk-core/compare/v0.18.0...v0.19.0
- Add a new plan,
StickyImmix
. This is an variant of immix using a sticky mark bit. This plan allows generational behaviors without using a compulsory copying nursery. - Side log bits in generational plans are now set using relaxed store for the entire byte as an optimization.
- Add constants
STRESS_DEFRAG
andDEFRAG_EVERY_BLOCK
to allow immix to stress defrag copying for debugging.
- Add a new write barrier method
object_probable_write
. The method can be called before the fields of an object may get updated without a normal write barrier. - Add a feature
immix_non_moving
to replace the currentimmix_no_defrag
feature. This is only intended for debugging uses to rule out issues from copying.
- Refactor
Mmapper
andVMMap
. Their implementation is now chosen dynamically, rather than statically based on the pointer size of the architecture. This allows us to use a more appropriate implementation to support compressed pointers. - Revert changes in d341506 about forwarding bits, and bulk zero side forward bits for defrag source blocks.
- Fix a bug in the scheduler where a worker may spuriously wake up and attempt to open new buckets while the coordinator is executing a coordinator work packet, which would result in an assertion failure.
- Fix a bug in the mem balancer where subtraction may silently overflow and cause unexpected statistics to be used for computing new heap sizes.
- Fix a bug where
ScanObjectsWork
does not properly set the worker when it internally usesProcessEdgesWork
. - Fix a bug where nodes pushed during root scanning were not cached for sanity GC.
- Fix a bug where
harness_begin
does not force a full heap GC for generational plans. - Add info-level logging for each GC to show the memory usage before and after the GC, and elapsed time for the GC.
- Fix a bug where
SemiSpace::get_available_pages
returned unused pages as 'available pages'. It should return half of the unused pages as 'available'. - Fix a bug where generational plans may make full-heap GC decision before all the releasing work is done. Now plans have a
Plan::end_of_gc()
method that is executed after all the GC work is done. - Fix a bug where generational plans tended to trigger full heap GCs when heap is full. Now we properly use Appel-style nursery and allow the nursery to extend to the heap size.
- Add a feature
immix_zero_on_release
as a debug feature for Immix to eagerly zero any reclaimed memory. - Fix a bug where the alloc bits for dead objects were still set after the objects were reclaimed. Now Immix eagerly clears the alloc bits once the memory is reclaimed.
- Fix a bug where Immix uses forwarding bits and forwarding pointers but did not declare it.
- Fix a bug about the maximum object size in
FreeListAllocator
. The allowed maximum object size depends on both the block size and the largest bin size.
- Add a 'sentinel' mechanism to execute a specified 'sentinel' work packet when a bucket is drained, and prevent the GC from entering the next stage depending
on the sentinel packet. This makes it possible to expand the transitive closure multiple times. It replaces
GCWorkScheduler::closure_end()
.
- Add a new set of APIs to support binding-specific weak reference processing:
- Add
Scanning::process_weak_refs()
andScanning::forward_weak_refs()
. They both supply anObjectTracerContext
argument for retaining and updating weak references.Scanning::process_weak_refs()
allows a boolean return value to indicate if the method should be called again by MMTk after finishing transitive closures for the weak references, which can be used to implement ephemeron. - Add
Collection::post_forwarding()
which is called by MMTk after all weak reference related work is done. A binding can use this call for any post processing for weak references, such as enqueue cleared weak references to a queue (Java). - These replace old methods for the same purpose, like
Collection::process_weak_refs()
,Collection::vm_release()
, andmemory_manager::on_closure_end
.
- Add
- Upgrade the Rust toolchain we use to 1.66.1 and upgrade MSRV to 1.61.0.
- Improve the boot time significantly for MMTk:
- Space descriptor map is created as an zeroed vector.
- Mmapper coalesces the mmap requests for adjancent chunks, and reduces our
mmap
system calls at boot time.
- Add
GCTrigger
to allow different heuristics to trigger a GC:- Implement
FixedHeapSizeTrigger
that triggers the GC when the maximum heap size is reached (the current approach). - Implement
MemBalancerTrigger
(https://dl.acm.org/doi/pdf/10.1145/3563323) as the dynamic heap resize heuristic.
- Implement
- Refactor
SFTMap
so it is dynamically created now. - Fix a bug where
Address::store
dropped the value after store. - Fix a bug about an incorrect pointer cast in unsafe code in
CommonFreeListPageResource
. - Remove all inline directives from our code base. We rely on Rust compiler and PGO for inlining decisions.
- Remove
SynchronizedCounter
.
- Refactor
MarkSweep
to work with both our native mark sweep policy and the malloc mark sweep policy backed by malloc libraries.
- Add
FreeListAllocator
which is implemented as aMiMalloc
allocator. - Fix a bug in
ImmixAllocator
that alignment is properly taken into consideration when deciding whether to do overflow allocation.
- Add
MarkSweepSpace
:- It uses our native MiMalloc implementation in MMTk core.
- When the feature
malloc_mark_sweep
is enabled, it uses the selected malloc library to back up its allocation.
- Malloc mark sweep now accounts for memory based on page usage, and each malloc library may use a different page size.
- The immix space now uses the newly added
BlockPageResource
. - Changes to Space Function Table (SFT) to improve our boot time and reduce the memory footprint:
- Rename the current SFT map to
SFTSparseChunkMap
, and only use it for 32 bits architectures. - Add
SFTSpaceMap
and by default use it for 64 bits architectures. - Add
SFTDenseChunkMap
and use it when we have no control of the virtual address range for a MMTk space on 64 bits architectures.
- Rename the current SFT map to
- Add an option
thread_affinity
to set processor affinity for MMTk GC threads. - Add
AllocationSemantics::NonMoving
to allocate objects that are known to be non-moving at allocation time. - Add
ReferenceGlue::is_referent_cleared
to allow some bindings to use a special value rather than a normal null reference for a cleared referent. - Add
pin
,unpin
, andis_pinned
for object pinning. Note that some spaces do not support object pinning, and using these methods may cause panic if the space does not support object pinning. - Refactor
ObjectReference
:- MMTk core now pervasively uses
ObjectModel::ref_to_address
to get an address from an object reference for setting per-object side metadata. - Add
ObjectModel::address_to_ref
that does the opposite ofref_to_address
: getting an object reference from an address that is returned byref_to_address
. - Add
ObjectModel::ref_to_header
for the binding to tell us the base header address from an object reference. - Rename
ObjectModel::object_start_ref
toObjectModel::ref_to_object_start
(to be consistent with other methods). - Remove
ObjectModel::OBJECT_REF_OFFSET_BEYOND_CELL
, as we no longer use the raw address of an object reference. - Add
ObjectModel::UNIFIED_OBJECT_REFERENCE_ADDRESS
. If a binding uses the same address forObjectReference
,ref_to_address
andref_to_object_start
, they should set this totrue
. MMTk can utilize this information for optimization. - Add
ObjectModel::OBJECT_REF_OFFSET_LOWER_BOUND
to specify the minimam value of the possible offsets between an allocation result and object reference's raw address.
- MMTk core now pervasively uses
destroy_mutator()
no longer requires a boxed mutator as its argument. Instead, a mutable reference to the mutator is required. It is made clear that the binding should manage the lifetime of the boxed mutator from abind_mutator()
call.- Remove
VMBinding::LOG_MIN_ALIGNMENT
andVMBinding::MAX_ALIGNMENT_SHIFT
(so we only keepVMBinding::MIN_ALIGNMENT
andVMBinding::MAX_ALIGNMENT
).
- Add a lock-free
BlockPageResource
that can be used for policies that always allocate memory at the granularity of a fixed sized block. This page resource facilitates block allocation and reclamation, and uses lock-free operations where possible. - Fix a race condition in
FreeListPageResource
when multiple threads release pages. - Fix a bug in
fetch_and/or
in our metadata implementation. - Fix a bug in side metadata bulk zeroing that may zero unrelated bits if the zeroed region cannot be mapped to whole metadata bytes.
- Remove unused
meta_data_pages_per_region
in page resource implementations. - Remove the use of
MaybeUninit::uninit().assume_init()
inFreeListPageResource
which has undefined behaviors and causes the illegal instruction error with newer Rust toolchains. - Remove the trait bound
From<Address>
andInto<Address>
forRegion
, as we cannot guarantee safe conversion between those two types. - Extract
Chunk
andChunkMap
from the immix policy, and make it available for all the policies.
- Generational plans now support bounded nursery size and fixed nursery size.
- Immix can be used in a non-moving variant (with no defragmentation) to facilitate early porting stages
where a non-moving GC is expected by the VM. Enable the feature
immix_no_defrag
to use the variant. Note that this variant performs poorly, compared to normal immix.
- Add
mod build_info
for bindings to get information about the current build. - Add
trait Edge
. A binding can implement its own edge type if they need more sophisiticated edges than a simple address slot, e.g. to support compressed pointers, base pointers with offsets, or tagged pointers. - Add APIs for implementing write barriers. MMTk provides subusming barriers
object_reference_write()
and pre/post write barriersobject_reference_write_pre/post()
. - Add APIs for implementing barriers for memory copying such as
array_copy
in Java. MMTk providesmemory_region_copy()
(subsuming) andmemory_region_copy_pre/post()
. - The
ignore_system_g_c
option is renamed toignore_system_gc
to be consistent with our naming convention. - The
max/min_nursery
option is replaced bynursery
. Bindings can usenursery=Fixed:<size>
orBounded:<size>
to config the nursery size. - Metadata access methods now requires a type parameter for the metadata value.
- Metadata compare-exchange methods now return a
Result
rather than a boolean, which is more consistent with Rust atomic types. - Metadata now supports
fetch_and
,fetch_or
andfetch_update
. - Header metadata access methods in
ObjectModel
now have default implementations.
- Remove all stdout printing in MMTk.
- Fix a bug that
CopySpace
should not try zeroing alloc bit if there is no allocation in the space. - Fix a few issues in documentation.
ProcessEdgesWork
is no longer exposed in theScanning
trait. Instead,RootsWorkFactory
is introduced for the bindings to create more work packets.Collection::stop_all_mutators()
now provides a callbackmutator_visitor
. The implementation is required to callmutator_visitor
for each mutator once it is stopped. This requirement was implicit prior to this change.- Now MMTk creation is done in the builder pattern:
MMTKBuilder
is introduced. Command line argument processing API (process()
andprocess_bulk()
) now takes&MMTKBuilder
as an argument instead of&MMTK
.gc_init()
is renamed tommtk_init()
.mmtk_init()
now takes&MMTKBuilder
as an argument, and returns an MMTk instanceBox<MMTK>
.heap_size
(which used to be an argument forgc_init()
) is now an MMTk option.- All the options now can be set through the command line argument processing API.
- Node enqueuing is supported:
- Add
Scanning::support_edge_enqueuing()
. A binding may returnfalse
if they cannot do edge scanning for certain objects. - For objects that cannot be enqueued as edges,
Scanning::scan_object_and_trace_edges()
will be called.
- Add
- Fixed a bug that may cause deadlock when GC workers are parked and the coordinator is still executing work.
Plan::gc_init()
andSpace::init()
are removed. Initialization is now properly done in the respective constructors.
- Fixed a bug that in GC stress testing, the allocator slowpath may double count the allocated bytes.
- Fixed a bug that in GC stress testing, the Immix allocator may miss the updates to the allocated bytes in some cases.
- Added work stealing mechanisms to the scheduler: a GC worker may steal work packets from other workers.
- Fixed a bug that work buckets may be incorrectly opened when there is still work left in workers' local bucket.
- Added an associate type
Finalizable
toReferenceGlue
, with which, a binding can define their own finalizer type. - Added a set of malloc APIs that allows a binding to do malloc using MMTk.
- Added
vm_trace_object()
toActivePlan
. When tracing an object that is not in any of MMTk spaces, MMTk will call this method and allow bindings to handle the object.
trait TransitiveClosure
is split into two different traits:EdgeVisitor
andObjectQueue
, andTransitiveClosure
is now removed.- Fixed a bug that the work packet statistics were not collected correctly if different work packets used the same display name.
- Fixed a bug that the work packet statistics and the phase statistics use different time units. Now they both use milliseconds.
- Fixed a bug that
acquire_lock
was used to lock a larger scope than what was necessary, which caused bad performance when we have many allocation threads (e.g. more than 24 threads).
- Introduced
trait PlanTraceObject
and procedural macros to derive implementation for it for all the current plans. - Introduced a work packet type
PlanProcessEdges
that usesPlanTraceObject
. All the current plans use this type for tracing objects.
- Introduced
trait PolicyTraceObject
. Added an implementation for each policy.
- Preliminary support for Java-style weak reference is added (set the option
no_reference_types=false
to enable it). Related APIs are slightly changed. - The type parameter
TransitiveClosure
inScanning::scan_object()/scan_objects()
is now replaced withvm::EdgeVisitor
. - Minor changes to
Scanning::scan_object()/scan_objects()
so they are more consistent.
- Fixed a bug in object forwarding: an object can leave the being-forwarded state without actually being forwarded, and this now won't cause a panic.
- Introduced a new work packet type
SFTProcessEdges
. Most plans now useSFTProcessEdges
for tracing objects, and no longer need to implement any plan-specific work packet. Mark compact and immix plans still use their own tracing work packet.
- Fixed a bug that
ImmixCopyContext
did not set the mark bit after copying an object. - Fixed a bug that
MarkCompactSpace
usedObjectReference
andAddress
interchangably. NowMarkCompactSpace
properly deals withObjectReference
.
is_mapped_object()
is superseded byis_in_mmtk_spaces()
. It returns true if the given object reference is in MMTk spaces, but it does not guarantee that the object reference actually points to an object.is_mmtk_object()
is added. It can be used to check if an object reference points to an object (useful for conservative stack canning).is_mmtk_object()
is only availble when theis_mmtk_object
feature is enabled.
- MMTk core now builds with stable Rust toolchains (minimal supported Rust version 1.57.0).
- Fixed a bug that MMTk may not map metadata and SFT for an object reference if the object reference is in a different chunk from the allocated address.
- Added
trait Region
andstruct RegionIterator<R>
to allow convenient iteration through memory regions.
- Removed plan-specific copy contexts. Now each plan needs to provide a configuration for
GCWorkerCopyContext
(similar to how they configMutator
). - Fixed a bug that
needs_log_bit
was always set totrue
for generational plans, no matter their barrier used the log bit or not. - Fixed a bug that we may overflow when calculating
get_available_pages()
.
- Refactored copy context. Now a copying policy provides its copy context.
- Mark sweep and mark compact now uses
ObjectIterator
for linear scan.
- Introduced
GCController
, a counterpart ofGCWorker
, for the controller thread. - Refactored
GCWorker
. NowGCWorker
is seperated into two parts, a thread local partGCWorker
which is owned by GC threads, and a shared partGCWorkerShared
that is shared between GC threads and the scheduler. - Refactored the creation of the scheduler and the workers to remove some unnecessary
Option<T>
andRwLock<T>
.
- Added
process_bulk()
that allows bindings to pass options as a string of key-value pairs. ObjectModel::copy()
now takesCopySemantics
as a parameter.- Renamed
Collection::spawn_worker_thread()
tospawn_gc_thread()
, which is now used to spawn both GC worker and GC controller. Collection::out_of_memory()
now takesAllocationError
as a parameter which hints the binding on how to handle the OOM error.Collection::out_of_memory()
now allows a binding to return from the method in the case of a non-critical OOM. If a binding returns,alloc()
will return a zero address.
- Added
ObjectIterator
that provides linear scanning through a region to iterate objects using the alloc bit. - Added a feature
work_packet_stats
to optionally collect work packet statistics. Note that MMTk used to always collect work packet statistics. - Optimized the access to the SFT map.
- Fixed a few issues with documentation.
- The example header file
mmtk.h
now uses the prefixmmtk_
for all the functions.
- Added a Lisp2-style mark compact plan.
- Added a GCWorkContext type for each plan which specifies the types used for this plan's GC work packet.
- Changed the allocation semantics mapping for each plan. Now each plan has 1-to-1 mapping between allocation semantics and spaces.
- Fixed a few bugs for Immix space when
DEFRAG
is disabled.
- Added an option
precise_stress
(which defaults totrue
). For precise stress test, MMTk will check for stress GC in each allocation (including thread local fastpath allocation). For non-precise stress test, MMTk only checks for stress GC in global allocation. - Refactored the code about counting scanned stacks to make it easier to read.
- Added
schedule_common()
to schedule the common work packets for all the plans. - Added proepr implementation for checking and triggering full heap collection for all the plans.
- Fixed a bug that collection triggers were not properly cleared after a GC.
- Fixed a bug that objects in generational copying's or semispace's tospace were traced.
- Added a parameter
roots: bool
toProcessEdgesWork::new()
to indicate whether the packet contains root edges. - Refactored
ImmixProcessEdges.trace_object()
so it can deal with both defrag GC and fast GC. - Fixed a bug in Immix that recyclable blocks were not defragment source which could cause OOM without attempting to evacuate recyclable blocks.
- Fixed a bug that nursery large objects had their unlogged bits set, and were treated as mature objects.
- Fixed a bug that SFT entries were not set for
MallocSpace
. - Fixed a bug that SFT entries may not be correctly set if the start address is not chunk aligned.
- Supported proper stress test for all the allocators.
- Supported proper alignment and offset for
MallocAllocator
.
- (Breaking change) Renamed
enable_collection()
toinitialize_collection()
. - Added
enable_collection()
anddisable_collection()
. When MMTk collection is disabled, MMTk allows allocation without triggering GCs. - Added
COORDINATOR_ONLY_STW
to theCollection
trait. If this is set, theStopMutators
work can only done by the MMTk coordinator thread. Otherwise, any GC thread may be used to stop mutators.
- Added assertions in
extreme_assertions
to check if side metadata access is within their bounds. - Added
SideMetadataSpec.name
to help debug. - Added a macro in
util::metadata::side_metadata::spec_defs
to help define side metadata specs without explicitly laying out specs and providing offsets for each spec. - Renamed
SideMetadataSpec.log_min_obj_size
toSideMetadataSpec.log_bytes_in_region
to avoid ambiguity. - Fixed some issues and outdated code in the MMTk tutorial.
- Fixed a bug that may cause incorrect perf event values.
- Refactored to extract common generational code from the existing generational copying plan.
- Added the generational immix plan, a two-generation algorithm that uses immix as its mature generation.
- Upgraded the Rust toolchain we use to nightly-2021-09-17 (rustc 1.57.0-nightly).
- Added a new feature
global_alloc_bit
: mmtk-core will set a bit for each allocated object. This will later be used to implement heap iteration and to support tracing internal pointers. - Refactored the scheduler simplify the implementation by removing the abstract
Scheduler
,Context
andWorkerLocal
. - Renamed the incorrect parameter name
primary
tofull_heap
in a fewprepare()
/release()
methods. - Renamed the phases in statistics reports from
mu
(mutator)/gc
toother
/stw
(stop-the-world) so they won't cause confusion in concurrenct GC plans. - Fixed a few misuses of side metadata methods that caused concurrency issues in accessing the unlogged bit.
- Fixed a bug in
MallocSpace
that caused side metadata was not mapped correctly if an object crossed chunk boundary. - Fixed a bug in
MallocSpace
that it may incorrectly consider a chunk's side metadata is mapped. - Fixed a bug in side metadata implementation that may cause side metadata not mapped if the side metadata size is less than a page.
- Fixed regression in
LockFreeImmortalSpace
. - Fixed a few typos in the tutorial.
- Added the Immix plan, an efficient mark-region garbage collector.
- Added a large code space for BasePlan (included by all the plans).
- Added the Immix allocator.
- Added the Immix space, and related data structures.
- Added
get_forwarded_object()
andis_reachable()
to the space function table (SFT). - Improved the Malloc space (marksweep).
- It now supports parallel sweeping.
- It now uses side metadata for page and chunk bits.
- It now supports bulk check for live objects if mark bits are side metadata.
- Fixed a bug that the side forwarding bits were not cleared for CopySpace.
- Refactored the metadata specs in the ObjectModel trait. Now each spec has a specific type (e.g. VMGlobalLogBitSpec), and provides a simpler constructor for binding implementers to create them.
- Added support for VM specific weak reference processing:
- added an API function
on_closure_end()
to add a callback so the binding will be notified when a object transitive closure ends. - added
vm_release()
andprocess_weak_refs()
to theCollection
trait.
- added an API function
- Added a few new plan constraints.
- needs_log_bit: indicates whether a plan will use the global log bit.
- may_trace_duplicate_edges: indicates whether a plan may allow benign races and trace duplicate edges in a GC.
- max_non_los_default_alloc_bytes: indicates the maximum object size (in bytes) that can be allocated with the default allocator for the plan. For objects that are larger than this, they should be allocated with AllocationSemantics.Los.
- Added utilities to measure process-wide perf events.
- Refactored the offset field for SideMetadataSpec with a union type SideMetadataOffset.
- Fixed a concurrency bug that may cause munprotect fail for the PageProtect plan.
- Fixed a few issues with the tutorial and documentation.
- Added a new plan PageProtect, a plan to help debugging. It allocates each object to a separate page, and protects pages when the pages are released.
- Major changes to the ObjectModel trait: now a binding must specify each per-object metadata used by mmtk-core, whether the metadata resides in header bits provided by the VM or side tables provided by mmtk-core. For in-header metadata, a binding can further implement how it can be accessed, in case the bits are not always available to mmtk-core.
- Refactored metadata to provide unified access to per-object metadata (in-header or side).
- Refactored work packet statistics to allow other types of stats other than execution times.
- Added the feature 'perf_counter' and the option 'perf_events' to collect data from hardware performance counters for work packets.
- 'extreme_assertions' now also checks if values stored in side metadata are correct.
- Fixed a bug that GenCopy may report OOM without doing a full heap GC.
- Fixed a bug that existing mmapping of side metadata memory may get overwritten.
- Fixed a bug that FreeListPageResource may return an incorrect new_chunk flag for the first allocation.
- Upgrade the Rust version we use to nightly-2021-05-12 (1.54.0-nightly).
- The type OpaquePointer is now superseded by more specific types (VMThread/VMMutatorThread/VMWorkerThread).
- The methods
ActivePlan::mutator()
andActivePlan::is_mutator()
are no longer unsafe due to the change above. - The method
ActivePlan::worker()
is removed. - Internal modules are no longer public visible. API functions are provided where necessary.
- Added a feature 'extreme_assertions' that will enable expensive assertions to help debugging.
- Refactored to clean up some unsafe code.
- Improved OOM message for failed mmapping.
- Improved page accounting - the memory used by side metadata is also included and reported.
- Improved worker statistics - the total time per work packet type will be reported.
- Fixed a bug that MMTk may overwrite existing memory mapping.
- Fixed a bug that stress GC was not triggered properly in GenCopy.
- Fixed a bug in FragmentedMmapper that may cause a memory region being mmapped twice.
- Changed the dependency of hoard-sys to v0.1.1.
- The dependencies of malloc implementations are optional.
- Changed the dependency of hoard-sys to v0.1 from crates.io.
- Added a marksweep implementation.
- GC plans are now selected at run-time.
- Added MallocAllocator that can be used as a freelist allocator.
- Added a few implementations of malloc/free that can be chosen as build-time features.
- Added MallocSpace.
- Added support for finalization.
- HAS_GC_BYTE in the ObjectModel trait is superseded by a feature 'side_gc_byte`.
- Added a general side metadata implementation.
- Added a framework for collecting analysis data.
- Added a framework that allows triggering analysis or sanity GC at byte-granularity.
- Added a tutorial for GC implementors.
- Added a porting guide for MMTk users (language implementors).
- GC workers now cache work locally.
- Fixed concurrency bugs in stack scanning and pointer forwarding.
- Refactored the
ObjectModel
trait and it is clearer now that MMTk expects a GC byte from the VM. - Removed methods in the API that were marked as deprecated.
- Minor changes to a few methods/traits in API.
- Rewrote the implementation of GC byte and forwarding word due to the API change.
- Calling
gc_init()
will not fail now if the binding has initialized its own logger. - Fixed a few bugs about incorrect entries in SFT map.
- Fixed wrong allocator config for gencopy.
- Fixed a bug that caused MMTk to panic with OOM in stress tests.
- Fixed a bug that caused the first discontiguous space descriptor being considered as empty.
Added the following plans:
- NoGC
- SemiSpace
- Generational Copying GC
Added the following allocators:
- Bump Pointer Allocator
- Large Object Allocator
Added the following space policies:
- Immortal (including variants)
- Large object
- Copy
- Introduced bi-directional API between VM and MMTk
- Implemented a scheduler, GC work packets and related statistics collecting mechanisms.
- Implemented sanity checking.