fix(deps): replace yanked chacha20 with SSE2 backend fix - #2729
Conversation
|
Important Review skippedReview was skipped due to path filters ⛔ Files ignored due to path filters (1)
CodeRabbit blocks several paths by default. You can override this behavior by explicitly including those paths in the path filters. For example, including ⚙️ Run configurationConfiguration used: Path: .coderabbit.yaml Review profile: CHILL Plan: Pro Plus Run ID: You can disable this status message by setting the Use the checkbox below for a quick retry:
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
Assay-Runner Lane CheckPASS: delegated runner proof accepted for this PR. Matched attested delegated proof pack: https://github.com/Rul1an/assay/actions/runs/33429837971 Expected delegated gate: Record proof in the PR body or a PR comment using: Changed-path classification:
Contract: |
|
| Branch | codex/2728-chacha20-fixed |
| Testbed | ubuntu-latest-store-code |
Click to view all benchmark results
| Benchmark | Latency | Benchmark Result milliseconds (ms) (Result Δ%) | Upper Boundary milliseconds (ms) (Limit %) |
|---|---|---|---|
| swc/120xlarge | 📈 view plot 🚷 view threshold | 9.23 ms(+15.91%)Baseline: 7.97 ms | 12.56 ms (73.50%) |
| swc/500x400b | 📈 view plot 🚷 view threshold | 30.38 ms(+15.53%)Baseline: 26.30 ms | 41.75 ms (72.76%) |
|
| Branch | codex/2728-chacha20-fixed |
| Testbed | ubuntu-latest |
Click to view all benchmark results
| Benchmark | Latency | Benchmark Result milliseconds (ms) (Result Δ%) | Upper Boundary milliseconds (ms) (Limit %) |
|---|---|---|---|
| sr/wc | 📈 view plot 🚷 view threshold | 54.75 ms(-51.34%)Baseline: 112.51 ms | 168.77 ms (32.44%) |
Exact-head delegated proof acceptedPR #2729 remains at
The earlier missing-proof failure is superseded by the accepted current commit status; it is not rerun or relabeled. Central CI still running when checked. Independent exact-head review remains absent, so no ready, auto-merge, merge or release claim. |
{
"schema": "assay.review-record.v0",
"head_sha": "bd8b7697db87bf1fb187a841bd489cc688666dfc",
"review_completed": true,
"verdict": "READY",
"findings": [],
"no_findings": true,
"builder": {
"agent": "codex",
"instance": "codex/2728-chacha20-fixed"
},
"reviewer": {
"agent": "ruley",
"instance": "Grok Bot Ruley, independent review delivered 2026-08-31 21:33:04-21:33:05 Europe/Amsterdam",
"github_login": "Rul1an"
},
"independence": {
"did_not_build": true,
"did_not_author_governing_spec": true,
"declaration": "Ruley states he did not write issue #2728, the lock delta, or its governing plan. Codex authored the commit and brief. Writer worktree untouched."
},
"relay": {
"posted_by": "Codex via Rul1an GitHub account",
"source": "Ruley's actual Grok Bot Desktop response, not Codex self-review",
"reviewer_posted_to_github": false
},
"scope": {
"base_sha": "f1e2fb644688a7768f12479b227ddde5c9328860",
"files": ["Cargo.lock"],
"full_diff_reviewed": true
},
"verification": [
"Start/end wait-gate holds on the exact candidate and base; no drift.",
"Full delta is Cargo.lock only, +2/-2: chacha20 0.10.0 to 0.10.2 and checksum. Dependency list unchanged; one chacha20 stanza.",
"Registry reports 0.10.0 and 0.10.1 yanked, 0.10.2 not yanked. Independently downloaded 38197-byte crate SHA-256 matches lock: 65c35e4b699c7e15ccbe7ee35c005e4fc0a278d22238a2857e6ce2dadeda1b06.",
"Independently inspected RustCrypto/stream-ciphers #579, #580 and #583. Published 0.10.2 changelog names #580; SSE2 backend no longer uses _mm_extract_epi32 and saves counter with _mm_storeu_si128.",
"Traced HEAD lock route chacha20 <- rand 0.10.0 <- object_store 0.14.1 <- assay-evidence <- assay-mcp-server, also adapter-api/core. rand 0.8.7 and 0.9.2 do not use chacha20.",
"Retained generated-package lock excerpt carries the same 0.10.2 checksum. Ruley inspected the artifact/logs; he did not rerun packaging or builds.",
"Delegated producer 33429837971 succeeded on candidate head. Consumer 33430136909 succeeded; its workflow_run head is trusted main f1e2fb64, while the candidate commit status is lane-check/proof accepted: gates=all. Those are distinct provenance roles.",
"Central CI 33429801222 now succeeds on candidate head. Historical lane-check misses do not supersede the current accepted commit status.",
"Tag v5.5.1 still locks chacha20 0.10.0. The published release is not fixed by this candidate."
],
"non_claims": [
"Code READY only. At review time draft/review-record gate blocked merging; Wave 0 remained in progress. Codex must reconcile live final gates before landing.",
"ARM/macOS tests and hosted Ubuntu success do not establish that the SSE2 backend executed.",
"Upstream Miri undefined behavior is not a locally reproduced Assay crash.",
"cargo package --no-verify is package inspection, not compilation. Local writer logs were read, not rerun.",
"No edits, builds, auth, public post or merge performed by Ruley. Review binds only this exact SHA."
]
} |
There was a problem hiding this comment.
Copilot wasn't able to review any files in this pull request.
💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.
Summary
Fixes #2728. Updates only the workspace lockfile from yanked
chacha20 0.10.0to0.10.2, with the published registry checksum. No other dependency, feature, version or policy changes.Upstream #579 reported an unavailable SSE4.1 instruction in the SSE2 RNG/legacy backend. #580 fixes the backend; #583 records that 0.10.0 and 0.10.1 were yanked. This is a supported-target dependency repair, not a locally reproduced Assay crash or exploitation claim.
The normal/build dependency path reaches the MCP package through
rand -> object_store -> assay-evidence, not only dev dependencies. The locally generated MCP package lock also contains 0.10.2. Existing published v5.5.1 packages are immutable and remain unchanged by this PR.Verification
Measured at
bd8b7697db87bf1fb187a841bd489cc688666dfc, basef1e2fb644688a7768f12479b227ddde5c9328860, worktree/Users/roelschuurkes/wt-codex-2728-chacha20; Rust/Cargo 1.96.0, macOS arm64, one build job and worktree-local target directory.cargo fmt --all -- --checkandgit diff --check: pass.cargo test --locked -p assay-evidence -p assay-adapter-api -p assay-core -p assay-mcp-server --lib: 1,405 passed, 0 failed. These are library tests, not a full repository suite.cargo test --locked -p assay-mcp-server --tests: 492 passed, 0 failed, 2 ignored across 41 test targets. The two ignored entries aredescendant_spawner_helper_process, which is re-executed as a child by its owning tests. This invocation includes the MCP library tests already counted above; the two totals must not be added as unique tests.cargo clippy --locked --workspace --all-targets --exclude assay-it --exclude assay-ebpf -- -D warnings: pass.cargo tree --locked -p assay-mcp-server --target x86_64-unknown-linux-gnu -e normal,build -i chacha20: fixed version appears on the production dependency path. This is dependency resolution, not an x86 execution test.cargo package -p assay-mcp-server --locked --no-verify: archive created and its embedded Cargo.lock inspected. This command is packaging inspection, not a package compilation test or publication.65c35e4b699c7e15ccbe7ee35c005e4fc0a278d22238a2857e6ce2dadeda1b06.Pending Landing Evidence
Independent exact-head review and required GitHub CI/delegated proof are still required. Cargo.lock requires delegated
gates=all; no proof on another SHA is accepted. macOS arm64 tests do not execute the SSE2 backend. No local Miri reproduction or full x86 behavioral claim is made.A follow-up patch release is needed to deliver this fix through published locked installation artifacts. Do not silently relabel v5.5.1 host evidence as a new-release pass.