fix: container-builtin reference leaks (builtins.c) - #235
Merged
Conversation
Four builtins build a fresh list/dict per element and append it to an output list. The inner fields used list_append_owned/dict_set_owned, but the OUTER append of the finished container used plain list_append (which increfs), so the builtin's own ref on row/pair/r was never released and leaked one container per element: builtin_scan_tokens list_append(out, row) -> _owned builtin_scan_int_tokens list_append(out, row) -> _owned builtin_tokenize_with_names list_append(result,pair) -> _owned builtin_nearest_in_range_all list_append(result,r) -> _owned (scan_tokens was a sibling of the traced leakers, same shape, not exercised by a counted test — fixed for consistency.) Clears test_builtin_indirect + test_coverage_gaps (counted) and the marker-only test_string_math. Harness leak tally 8 -> 6. Release 1941/1941, ASan 1941/1941 (tally 6). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
InauguralPhysicist
added a commit
that referenced
this pull request
Jun 19, 2026
… recovery Patch release on top of 0.16.0. Highlights: - Remote HTTP DoS fixed (Content-Length SEGV; missing _GNU_SOURCE → 64-bit pointer truncation) and the implicit-declaration class hard-errored in the build (#239/#240). - ASan leak tally driven 10 → 4 (spawn-thread floor) across the runtime, HTTP code routes, container builtins, and the store extension (#234/#235/#236/#233). - OSR/JIT coverage for observed-assign loops recovered: #211's conservative scanner bail replaced by the real fix — reset the OP_POP last_imm peephole at forward-jump targets — so those loops OSR-compile again (~34%, #231). - Docs: clarified the implicit-`n` parameter shadows an outer `n` (#241, not a bug). Also repairs CHANGELOG drift: commit 8fe0731 had accidentally deleted the `## [0.16.0] — 2026-06-18` header, folding released 0.16.0 content back into [Unreleased]. The header is restored (with its summary), the windowed-predicate + #212/#213/#214 entries are reattributed to 0.16.0 (their code shipped in the v0.16.0 tag), and everything post-tag is grouped under the new [0.16.1]. VERSION 0.16.0 → 0.16.1. Release 1943/1943; ASan 1943/1943, leak tally 4. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Four builtins build a fresh list/dict per element and append it to an output list. The inner fields used list_append_owned/dict_set_owned, but the OUTER append of the finished container used plain list_append (which increfs), so the builtin's own ref on row/pair/r was never released and leaked one container per element:
builtin_scan_tokens list_append(out, row) -> _owned
builtin_scan_int_tokens list_append(out, row) -> _owned
builtin_tokenize_with_names list_append(result,pair) -> _owned
builtin_nearest_in_range_all list_append(result,r) -> _owned
(scan_tokens was a sibling of the traced leakers, same shape, not exercised by a counted test — fixed for consistency.)
Clears test_builtin_indirect + test_coverage_gaps (counted) and the marker-only test_string_math. Harness leak tally 8 -> 6.
Release 1941/1941, ASan 1941/1941 (tally 6).
What does this PR do?
Checklist
make testpasses locallydocs/BUILTINS.mddocs/STDLIB.md