summaryrefslogtreecommitdiff
path: root/gc/mmtk
AgeCommit message (Collapse)Author
4 days[ruby/mmtk] Use rb_gc_obj_needs_cleanup_pPeter Zhu
https://github.com/ruby/mmtk/commit/031785b41c
8 days[ruby/mmtk] Fix compiler warnings in mmtk.cPeter Zhu
gc/mmtk/mmtk.c:505:1: warning: function 'rb_mmtk_gc_thread_panic_handler' could be declared with attribute 'noreturn' [-Wmissing-noreturn] 505 | { | ^ gc/mmtk/mmtk.c:987:27: warning: variable length array folded to constant array as an extension [-Wgnu-folding-constant] 987 | char obj_info_buf[info_size]; | ^~~~~~~~~ gc/mmtk/mmtk.c:990:34: warning: variable length array folded to constant array as an extension [-Wgnu-folding-constant] 990 | char parent_obj_info_buf[info_size]; | ^~~~~~~~~ https://github.com/ruby/mmtk/commit/8ce21cdf10
2026-05-23[ruby/mmtk] Switch logging from debug to info levelPeter Zhu
https://github.com/ruby/mmtk/commit/9f730cc709
2026-05-23[ruby/mmtk] Add cpu heap mode based on Tavakolisomeh et al. (MPLR '23)Ufuk Kayserilioglu
Adds MMTK_HEAP_MODE=cpu, a dynamic heap-sizing policy that grows or shrinks the heap after each GC cycle to keep measured GC CPU overhead near a configurable target. The control law follows Tavakolisomeh et al., 'Heap Size Adjustment with CPU Control', MPLR '23: a sigmoid of the (averaged) GC CPU overhead error in (-inf, +inf) maps to a heap-size adjustment factor in (0.5, 1.5). Implementation lives alongside the existing 'ruby' delegated trigger in gc/mmtk/src/heap/. T_GC is wall-clock GC duration; T_APP is process CPU time delta read via clock_gettime(CLOCK_PROCESS_CPUTIME_ID), which correctly credits multi-threaded mutator parallelism. Nursery-only generational GCs are skipped so the trigger only re-sizes at full collections. Configuration: MMTK_GC_CPU_TARGET target GC CPU overhead, percent. Default 5. MMTK_GC_CPU_WINDOW number of recent cycles averaged. Default 3. The default differs from the paper's recommended 15. The paper targets ZGC, a concurrent generational collector; MMTk-Ruby currently ships stop-the-world Immix, where every percent of GC CPU also blocks the mutator. An empirical sweep of MMTK_GC_CPU_TARGET across ruby-bench (railsbench, lobsters, psych-load, liquid-render, lee) found 5-6 to be Pareto-optimal vs the existing 'ruby' heap mode: about 6 percent geomean throughput improvement at essentially equal peak RSS. Targets >=10 trade large amounts of throughput for modest RSS savings on this collector. bin/smoke-test, bin/ruby-mmtk-mode, bin/compare-heap-modes, and doc/testing-cpu-heap-mode.md are included so reviewers and future contributors can reproduce the sweep against ruby/ruby-bench. https://github.com/ruby/mmtk/commit/1f223f5ad5 Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
2026-05-20[ruby/mmtk] Fix compiler warnings for noreturnPeter Zhu
gc/mmtk/mmtk.c:480:1: warning: function 'rb_mmtk_gc_thread_bug' could be declared with attribute 'noreturn' [-Wmissing-noreturn] 480 | { | ^ gc/mmtk/mmtk.c:510:1: warning: function 'rb_mmtk_mutator_thread_panic_handler' could be declared with attribute 'noreturn' [-Wmissing-noreturn] 510 | { | ^ https://github.com/ruby/mmtk/commit/a57746d575
2026-05-17[ruby/mmtk] Move rb_gc_obj_suffix to gc.hPeter Zhu
https://github.com/ruby/mmtk/commit/40053354bd
2026-05-16[ruby/mmtk] Respect alignment in rb_mmtk_alloc_fast_pathPeter Zhu
https://github.com/ruby/mmtk/commit/9772a1d30a
2026-05-14[ruby/mmtk] Remove dead ractor_check_mode fieldPeter Zhu
https://github.com/ruby/mmtk/commit/a46b68fe5b
2026-05-14[ruby/mmtk] Remove call to rb_ractor_setup_belongingPeter Zhu
Don't need to call it after https://github.com/ruby/ruby/pull/16961. https://github.com/ruby/mmtk/commit/ae5638ac00
2026-05-09[ruby/mmtk] Introduce event_hook_mutex for RUBY_INTERNAL_EVENT_FREEOBJPeter Zhu
We can stop using rb_gc_worker_thread_set_vm_context because rb_gc_event_hook now uses rb_gc_get_ec since https://github.com/ruby/ruby/pull/16880 https://github.com/ruby/mmtk/commit/04bcf8456a
2026-05-08[ruby/mmtk] Remove unnecessary null check.Matt Valentine-House
the only caller of this unconditionally constructs a binding options object now, So actually this is dead code https://github.com/ruby/mmtk/commit/d832004e89
2026-05-08[ruby/mmtk] Introduce support for ractor_belonging.Matt Valentine-House
This is a debug mode in Ruby where an extra word is used after each object to store the address of the Ractor that owns the object, used for debug purposes only. While we're working on Ractors, we also need to be able to test with MMTk enabled, so we should introduce support for this to the MMTk binding as well. As implemented we'll default the binding options to have everything disabled and hardcoded to 0, as was always the case, but if RACTOR_CHECK_MODE is enabled, we'll build and pass a valid RubyBinding object to MMTk. https://github.com/ruby/mmtk/commit/83cb291313
2026-05-04Use EC saved in GC for root markingPeter Zhu
Since EC is thread-local, we previously used rb_gc_worker_thread_set_vm_context in MMTk worker threads to temporarily set the EC. However, this was inelegant and also occasionally caused crashes when marking threads/fibers for the current EC since it will mark the current machine stack twice (once during root marking and once for the fiber). However, since the machine stack is actively being used, the contents may be different when marking the fiber. Since all objects on the machine stack are pinned, this may cause an unpinned object to be pinned, which is not allowed in Immix. The following crash can be observed: Object 0x200fffbc7d8 is trying to pin 0x200ffc80188 0: mmtk_ruby::handle_gc_thread_panic 1: mmtk_ruby::set_panic_hook::{{closure}} 2: <alloc::boxed::Box<dyn for<'a, 'b> core::ops::function::Fn<(&'a std::panic::PanicHookInfo<'b>,), Output = ()> + core::marker::Sync + core::marker::Send> as core::ops::function::Fn<(&std::panic::PanicHookInfo,)>>::call at /rustc/59807616e1fa2540724bfbac14d7976d7e4a3860/library/alloc/src/boxed.rs:2254:9 3: std::panicking::panic_with_hook at /rustc/59807616e1fa2540724bfbac14d7976d7e4a3860/library/std/src/panicking.rs:833:13 4: std::panicking::panic_handler::{closure#0} at /rustc/59807616e1fa2540724bfbac14d7976d7e4a3860/library/std/src/panicking.rs:698:13 5: std::sys::backtrace::__rust_end_short_backtrace::<std::panicking::panic_handler::{closure#0}, !> at /rustc/59807616e1fa2540724bfbac14d7976d7e4a3860/library/std/src/sys/backtrace.rs:182:18 6: __rustc::rust_begin_unwind at /rustc/59807616e1fa2540724bfbac14d7976d7e4a3860/library/std/src/panicking.rs:689:5 7: core::panicking::panic_fmt at /rustc/59807616e1fa2540724bfbac14d7976d7e4a3860/library/core/src/panicking.rs:80:14 8: <mmtk_ruby::scanning::VMScanning as mmtk::vm::scanning::Scanning<mmtk_ruby::Ruby>>::scan_object_and_trace_edges::{{closure}} 9: mmtk_ruby::abi::ObjectClosure::c_function_registered 10: rb_mmtk_call_object_closure at gc/mmtk/mmtk.c:976:19 11: rb_gc_impl_mark_and_pin at gc/mmtk/mmtk.c:1008:5 12: rb_gc_impl_mark_and_pin at gc/mmtk/mmtk.c:1004:1 13: gc_mark_maybe_internal at gc.c:2908:5 14: gc_mark_maybe_internal at gc.c:2906:1 15: gc_mark_maybe_each_location at gc.c:2939:5 16: gc_mark_maybe_each_location at gc.c:2937:1 17: each_location at gc.c:2924:9 18: each_location_ptr at gc.c:2933:5 19: each_location_ptr at gc.c:2930:1 20: rb_gc_mark_machine_context at gc.c:3200:5 21: rb_execution_context_mark at vm.c:3768:9 22: cont_mark at cont.c:1155:5 23: fiber_mark at cont.c:1284:5 24: rb_mmtk_call_gc_mark_children at gc/mmtk/mmtk.c:318:5 25: <mmtk_ruby::scanning::VMScanning as mmtk::vm::scanning::Scanning<mmtk_ruby::Ruby>>::scan_object_and_trace_edges::{{closure}}
2026-04-09Introduce RVALUE_SIZE GC constantMatt Valentine-House
Add GC::INTERNAL_CONSTANTS[:RVALUE_SIZE] to store the usable size (excluding debug overhead) of the smallest pool that can hold a standard RVALUE.
2026-04-09Introduce power-of-two size poolsMatt Valentine-House
Replace the RVALUE_SLOT_SIZE-multiplier based pool sizes with explicit power-of-two (and near-power-of-two) slot sizes. On 64-bit this gives 12 heaps (32, 40, 64, 80, 96, 128, 160, 256, 512, 640, 768, 1024) instead of 5, providing finer granularity and less internal fragmentation. On 32-bit the layout is 5 heaps (32, 64, 128, 256, 512).
2026-04-01[ruby/mmtk] Add moving_gc_count to GC.statPeter Zhu
Outputs the number of GC cycles that are moving. https://github.com/ruby/mmtk/commit/fef8f04186
2026-03-09Look up slot sizes for allocations in a tableMatt Valentine-House
Also remove BASE_SLOT_SIZE.
2026-03-04[ruby/mmtk] Cargo formatMatt Valentine-House
https://github.com/ruby/mmtk/commit/7889da7c0e
2026-03-04[ruby/mmtk] Distribute batch candidates across parallel bucketsMatt Valentine-House
Instead of sending all 128 buffered objects to one bucket, round-robin distribute them across all worker buckets so parallel obj_free work stays balanced. https://github.com/ruby/mmtk/commit/e1f926cd21
2026-03-04[ruby/mmtk] Flush obj_free buffers before shutdown finalizersMatt Valentine-House
shutdown_call_finalizer reads candidates from the Rust-side WeakProcessor, but the main ractor's C-side buffer may not have been flushed yet (ractor_cache_free runs later). Flush all remaining buffers before reading candidates. https://github.com/ruby/mmtk/commit/7e01232134
2026-03-04[ruby/mmtk] Fix Cargo format issuesMatt Valentine-House
https://github.com/ruby/mmtk/commit/26ec9f7f89
2026-03-04[ruby/mmtk] Buffer obj_free candidates.Matt Valentine-House
Previously, every object allocation in rb_gc_impl_new_obj made a per-object FFI call into Rust (mmtk_add_obj_free_candidate), which acquired a mutex on one of the WeakProcessor's candidate vecs, pushed a single element, and released the mutex. That's an FFI crossing + mutex lock/unlock on every single allocation. Now, each MMTk_ractor_cache has two local buffers (parallel-freeable and non-parallel-freeable, 128 entries each). On allocation, we just store the pointer into the local buffer. When a buffer fills up, we flush the entire batch in one FFI call using mmtk_add_obj_free_candidates, which does a single mutex acquisition and extend_from_slice for the whole batch. We picked 128 as our buffer size at random. We should probably investigate further what an optimum size for this is https://github.com/ruby/mmtk/commit/23c4a9a676
2026-02-12[ruby/mmtk] Implement rb_gc_impl_during_gc_pMatt Valentine-House
https://github.com/ruby/mmtk/commit/039fa648fe
2026-02-05Bump git2 in /gc/mmtk in the cargo group across 1 directorydependabot[bot]
Bumps the cargo group with 1 update in the /gc/mmtk directory: [git2](https://github.com/rust-lang/git2-rs). Updates `git2` from 0.20.2 to 0.20.4 - [Changelog](https://github.com/rust-lang/git2-rs/blob/git2-0.20.4/CHANGELOG.md) - [Commits](https://github.com/rust-lang/git2-rs/compare/git2-0.20.2...git2-0.20.4) --- updated-dependencies: - dependency-name: git2 dependency-version: 0.20.4 dependency-type: indirect dependency-group: cargo ... Signed-off-by: dependabot[bot] <support@github.com>
2026-01-29[ruby/mmtk] Add weak_references_count to GC.statPeter Zhu
https://github.com/ruby/mmtk/commit/86fa2fd4af
2026-01-21[ruby/mmtk] Pass whether GC is moving to rb_mmtk_update_global_tablesPeter Zhu
https://github.com/ruby/mmtk/commit/002faa8f92
2026-01-20[ruby/mmtk] Restore MMTK_DEBUG around zeroing in obj_freePeter Zhu
https://github.com/ruby/mmtk/commit/ae69d69768
2026-01-18[ruby/mmtk] Fix assertion in rb_gc_impl_ractor_cache_free for RUBY_FREE_AT_EXITPeter Zhu
https://github.com/ruby/mmtk/commit/e2a89d5a22
2026-01-18[ruby/mmtk] Check RUBY_FL_FINALIZE on new location of object if movedPeter Zhu
https://github.com/ruby/mmtk/commit/a31deee342
2026-01-18[ruby/mmtk] Replace RUBY_ASSERT with MMTK_ASSERTPeter Zhu
https://github.com/ruby/mmtk/commit/e92baf15ef
2026-01-13[ruby/mmtk] Implement object movement support for finalizer tablePeter Zhu
https://github.com/ruby/mmtk/commit/e5e2c1c347
2026-01-12[ruby/mmtk] Clear slot when object is freedPeter Zhu
https://github.com/ruby/mmtk/commit/87290e45b2
2026-01-09[ruby/mmtk] Assert that objects are not T_NONE in the write barrierPeter Zhu
https://github.com/ruby/mmtk/commit/59d27203e2
2026-01-09[ruby/mmtk] Add MMTK_ASSERTPeter Zhu
https://github.com/ruby/mmtk/commit/e34d5cf32f
2026-01-07[ruby/mmtk] Check that a and b are valid objects in write barrierPeter Zhu
https://github.com/ruby/mmtk/commit/350625ebb3
2026-01-07[ruby/mmtk] Set MMTK_DEBUG in C compiler for debug buildsPeter Zhu
https://github.com/ruby/mmtk/commit/12a3904b04
2026-01-05[ruby/mmtk] Assert pinning object is not movedPeter Zhu
https://github.com/ruby/mmtk/commit/8813e76bf8
2026-01-04[ruby/mmtk] Process obj_free candidates in parallelPeter Zhu
Redos commit 544770d which seems to have accidentally been undone in b27d935.
2026-01-03Also output GC thread backtrace in rb_mmtk_gc_thread_bugPeter Zhu
2026-01-03[ruby/mmtk] Use rb_mmtk_gc_thread_bug for rb_mmtk_call_object_closurePeter Zhu
https://github.com/ruby/mmtk/commit/308936296a
2026-01-03[ruby/mmtk] Propagate crash of GC thread to mutator threadPeter Zhu
This allows the mutator thread to dump its backtrace when a GC thread crashes. https://github.com/ruby/mmtk/commit/40ff9ffee7
2026-01-02[ruby/mmtk] Check for T_NONE during markingPeter Zhu
https://github.com/ruby/mmtk/commit/c3e338bb25
2026-01-02[ruby/mmtk] Assert target is not pinned during normal tracingPeter Zhu
https://github.com/ruby/mmtk/commit/58210c88ed
2026-01-01[ruby/mmtk] Format imports to be each on a new linePeter Zhu
https://github.com/ruby/mmtk/commit/42adba630e
2025-12-31Use `is_obj_encoding` instead of `is_data_encoding`Nobuyoshi Nakada
The argument to `is_data_encoding` is assumed to be `T_DATA`.
2025-12-31[ruby/mmtk] Split ProcessObjFreeCandidates to parallel and non-parallelPeter Zhu
This makes it easier to visualize in profilers which one is non-parallel. https://github.com/ruby/mmtk/commit/ba68b2ef3b
2025-12-31[ruby/mmtk] Process obj_free candidates in parallelPeter Zhu
This commit allows objects that are safe to be freed in parallel to do so. A decrease in object freeing time can be seen in profiles. The benchmarks don't show much difference. Before: -------------- -------------------- ---------- --------- bench sequential free (ms) stddev (%) RSS (MiB) activerecord 242.3 7.4 84.3 chunky-png 439.1 0.6 75.6 erubi-rails 1221.2 4.2 132.7 hexapdf 1544.8 1.8 429.1 liquid-c 42.7 7.4 48.5 liquid-compile 41.4 8.3 52.2 liquid-render 100.6 3.0 56.8 mail 108.9 2.1 65.1 psych-load 1536.9 0.6 43.4 railsbench 1633.5 2.6 146.2 rubocop 126.5 15.8 142.1 ruby-lsp 129.6 9.7 112.2 sequel 47.9 6.5 44.6 shipit 1152.0 2.7 315.2 -------------- -------------------- ---------- --------- After: -------------- ------------------ ---------- --------- bench parallel free (ms) stddev (%) RSS (MiB) activerecord 235.1 5.5 87.4 chunky-png 440.8 0.8 68.1 erubi-rails 1105.3 0.8 128.0 hexapdf 1578.3 4.1 405.1 liquid-c 42.6 7.1 48.4 liquid-compile 41.5 8.1 52.1 liquid-render 101.2 2.8 53.3 mail 109.7 2.7 64.8 psych-load 1567.7 1.1 44.4 railsbench 1644.9 1.9 150.9 rubocop 125.6 15.4 148.5 ruby-lsp 127.9 5.8 104.6 sequel 48.2 6.1 44.1 shipit 1215.3 4.7 320.5 -------------- ------------------ ---------- --------- https://github.com/ruby/mmtk/commit/4f0b5fd2eb
2025-12-30[ruby/mmtk] Use MMTK_HEAP_COUNT for SIZE_POOL_COUNTPeter Zhu
https://github.com/ruby/mmtk/commit/290a2aec4e
2025-12-30Add RVALUE_OLD_AGE to GC::INTERNAL_CONSTANTS for MMTkPeter Zhu
2025-12-29Implement moving Immix in MMTkPeter Zhu
This commit implements moving Immix in MMTk, which allows objects to move in the GC. The performance of this implementation is not yet amazing. It is very similar to non-moving Immix in many of them and slightly slower in others. The benchmark results is shown below. -------------- ----------------- ---------- --------- bench Moving Immix (ms) stddev (%) RSS (MiB) activerecord 241.9 0.5 86.6 chunky-png 447.8 0.8 74.9 erubi-rails 1183.9 0.8 136.1 hexapdf 1607.9 2.6 402.3 liquid-c 45.4 6.7 44.9 liquid-compile 44.1 9.3 53.0 liquid-render 105.4 4.5 55.9 lobsters 650.1 9.7 418.4 mail 115.4 2.1 64.4 psych-load 1656.8 0.8 43.6 railsbench 1653.5 1.3 149.8 rubocop 127.0 15.6 142.1 ruby-lsp 130.7 10.5 99.4 sequel 52.8 7.2 45.6 shipit 1187.0 3.9 311.0 -------------- ----------------- ---------- --------- -------------- --------------------- ---------- --------- bench Non-moving Immix (ms) stddev (%) RSS (MiB) activerecord 218.9 2.7 86.1 chunky-png 464.6 0.8 66.7 erubi-rails 1119.0 4.3 132.7 hexapdf 1539.8 1.8 425.2 liquid-c 40.6 6.9 45.2 liquid-compile 40.6 8.1 52.9 liquid-render 99.3 2.3 48.3 mail 107.4 5.3 65.4 psych-load 1535.6 1.0 39.5 railsbench 1565.6 1.1 149.6 rubocop 122.5 14.3 146.7 ruby-lsp 128.4 10.7 106.4 sequel 44.1 4.0 45.7 shipit 1154.5 2.7 358.5 -------------- --------------------- ---------- ---------
generated by cgit v1.2.3 (git 2.25.1) at 2026-06-08 07:58:13 +0000