Skip to content

correction: Bound alignment cache growth - #24

Merged
BenWestgate merged 2 commits into
reviewability-v1from
21-bound-correction-alignment-cache
Sep 24, 2026
Merged

BenWestgate merged 2 commits into
reviewability-v1from
21-bound-correction-alignment-cache

Conversation

@BenWestgate

@BenWestgate BenWestgate commented Sep 22, 2026 •

Copy link
Copy Markdown
Owner

Bound _syndrome_alignment to eight recent entries so caller-controlled HRPs cannot create process-lifetime cache growth. Retains normal cache reuse and adds churn coverage.\n\n\n\nValidation: 862 pytest tests; mypy; Ruff check/format; 57 frozen differential-correction cases; 100 mixed attacker HRP/length entries left currsize=8.

Fixes #21.

Limit attacker-keyed syndrome alignment state to eight recent entries while retaining reuse for normal correction work. Add churn and cache-hit regression coverage.\n\nFixes #21.

@chatgpt-codex-connector chatgpt-codex-connector Bot left a comment

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 55385ca6bd

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

  • Open a pull request for review
  • Mark a draft as ready
  • Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

Comment thread src/codex32/correction.py Outdated
Size the bounded alignment LRU for the full 11-length generic unknown-target window so repeated search phases reuse every alignment without reintroducing unbounded attacker-controlled state. Add a regression proving the full window remains hot.\n\nSecurity: preserves the resource bound from #21 while avoiding cache thrash that can consume the fixed correction deadline.\n\nValidation: test_alignment.py 26 passed; full pytest 863 passed; optimized pytest 863 passed; Ruff, format, mypy, and git diff --check passed.
@BenWestgate BenWestgate added the gate: adversarial review Resolve, merge, or explicitly defer before the next full adversarial review. label Sep 24, 2026
@BenWestgate BenWestgate self-assigned this Sep 24, 2026

@BenWestgate BenWestgate left a comment

Copy link
Copy Markdown
Owner Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

crACK fceaeec

@BenWestgate
BenWestgate merged commit fba20e3 into reviewability-v1 Sep 24, 2026
12 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

gate: adversarial review Resolve, merge, or explicitly defer before the next full adversarial review.

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant