fix(runtime): recover poisoned activated-tool lock reads#7755
Open
77382104 wants to merge 1 commit into
Open
Conversation
github-actions
Bot
added
agent
77382104
force-pushed
the
fix/runtime-activated-tool-lock-poison-recovery
branch
from
f047957 to
352cb5f
Compare
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Summary
master(all contributions)ActivatedToolSetlock panics in per-iteration tool-spec assembly with explicit poison recovery for read-only access so a prior lock poison no longer crashes the turn path.ActivatedToolSetlock panic in deferred tool resolution during tool execution with the same read-recovery behavior so activated deferred tools remain usable after poison.ActivatedToolSetlocks are recovered and existing activated-tool behavior still works.ActivatedToolSetaccess and leaving the rest of the turn/tool execution flow unchanged.ActivatedToolSetis populated, does not replace the underlying mutex type, and does not add any new config, CLI, or persistence behavior.zeroclaw-runtime, specifically turn-time tool spec assembly and activated-tool lookup during tool execution. Other runtime subsystems should be unaffected.type: bug,risk: medium,size: SValidation Evidence (required)
Local validation is the signal CI cannot replace. Run the full battery and paste literal output (tails, failures, warnings — not "all passed").
cargo fmt --all -- --check cargo clippy --all-targets -- -D warnings cargo testcargo fmt --all -- --checkcargo clippy --all-targets -- -D warningsdevprofile [unoptimized + debuginfo] target(s) in 1.12scargo testOutput
ActivatedToolSetaccess paths recover from a poisoned mutex instead of panicking.Security & Privacy Impact (required)
Yes/No for each. Answer any
Yeswith a 1–2 sentence explanation.No)No)No)No)Yes, describe the risk and mitigation:Compatibility (required)
Yes)No)NoorYesto either: exact upgrade steps for existing users:Rollback (required for
risk: mediumandrisk: high)git revert <merge_commit_sha>Noneactivated-tool lock poisoned while assembling iteration tool specs; recovering guard for readoractivated-tool lock poisoned while resolving tool; recovering guard for read. A regression would likely show deferred MCP tools missing from per-turn tool specs or activated deferred tools failing to resolve during execution.Supersede Attribution (required only when
Supersedes #is used)#<pr> by @<author>, one per line):Co-authored-bytrailers added in commit messages for incorporated contributors? (No)No, why (inspiration-only, no direct code/design carry-over):Labels live in the GitHub label UI, not in the body. Maintainers and reviewers with label permissions set
risk:*,size:*, and any missing manual labels via the sidebar. The PR path labeler only owns path/scope labels from.github/labeler.yml. Contributors without label permission can note obvious label mismatches in a comment.Do not add bot/AI attribution footers such as
Co-authored-by: Claude ...or
Created with Claude Codeto the PR body or commit-message tail. Humanco-author trailers are appropriate only for incorporated contributor work under
the supersede-attribution section and privacy contract.
Privacy contract (
docs/book/src/contributing/privacy.md) is a merge gate. Never commit real identities, secrets, personal emails, or PII in diff, tests, fixtures, or docs.