[harivansh-afk]

TLDR

Stacked on #5409 (wb-2). Only Gen2 has cards, because minor collections trace Gen0 and Gen1 together. mark_card_for_ptr already ignored every other container. The barrier still classified both sides fully: up to four chunk scans, then a fifth in mark_card_for_ptr to find the card. Most containers are young, and a young container is the worst case: every scan runs to the end and misses.

How

• write_barrier: look up the container's Gen2 location first. If it's in Gen2 and the value isn't, mark the card at that location. A young container costs one Gen2 scan.
• conservative_write_barrier calls mark_card_for_ptr directly.
• Small helpers: locate_in_gen2, mark_card_at.
• The same cards are marked as before: container in Gen2 and value in CompileTime, Gen0 or Gen1. wb-1's matrix pins this.

Results (wb-2 + wb-3 vs #5264 at 19f2ae2328, 10 interleaved rounds)

workload	Spark native	Spark bytecode	x86 native	x86 bytecode
richards-50	1.57×	1.29×	1.13×	1.19×
array-build-sum-100k	1.32×	1.09×	1.08×	1.09×
grid-alloc-1000x100	1.30×	1.09×	1.11×	1.13×
merge-sort-20k	1.07×	1.05×	1.08×	1.33×
allocation-high-retention	1.05×	1.04×	1.02×	1.14×
k-nucleotide-25k	1.03×	1.06×	1.02×*	1.03×
cycles-into-live	1.02×	1.02×	1.01×	1.03×
live-walk / json-parser-15	1.00×	1.00–1.02×	1.01×* / 1.00×	1.00–1.03×

* 20-trial rerun; the first 10-trial run read 0.97× / 0.96× with a ~25% trial spread.

• Spark: aarch64, Cortex-X925, pinned core. x86: Freestyle bench-x86-v1.
• x86 gains less because an atomic there is one inline lock instruction, not an outlined call.
• Profile, native richards on Spark: the lock atomics went from ~31% to ≤0.2%, and generation_of no longer appears.
• A constant-time Gen2 lookup (wb-4) is parked: what's left of the scan doesn't show in the profile while Gen2 is a few chunks.

Measurements: instrumentation/measurements/2026-10-25-write-barrier-ab-{spark,x86,x86-rerun}/.

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated
developer-docs	Ready	Preview	Oct 25, 2026 9:43pm UTC

[coderabbitai]

Important

Draft PR not reviewed

Draft PRs are not automatically reviewed by default.

• Trigger a manual review

To automatically review draft PRs, update your CodeRabbit configuration:

reviews:
  auto_review:
    drafts: true

• Autofix · Keep fixing CodeRabbit findings and required CI, and resolving merge conflicts

Warning

Billing warning: we have not been able to collect payment for this subscription for more than 72 hours. Please update the payment method or pay any pending invoices in Billing to avoid service interruption.

_{Comment @coderabbitai help to get the list of available commands.}

[github-actions]

⏭️ Performance benchmarks were skipped

Perf benchmarks (CodSpeed) are opt-in on pull requests — they no longer run on every push. They always run automatically after merge to canary/main.

To run them on this PR, do any of the following, then push a commit (or re-run CI):

• Add RUN_CODSPEED=1 to the PR description, or
• Include run-perf or /perf in the PR title or any commit message.

[github-actions]

Binary size checks failed

❌ 2 violations · ✅ 1 passed

⚠️ Please fix the size gate issues or acknowledge them by updating baselines.

	Artifact	Platform	File	Gzip	Gated on	Baseline	Delta	Status
✅	baml-cli	Linux	🔒 64.1 MB	27.3 MB	file	73.1 MB	-9.0 MB (-12.3%)	OK
❌	packed-program	Linux	🔒 41.4 MB	16.9 MB	file	29.2 MB	+12.2 MB (+41.9%)	FAIL
❌	bridge_wasm	WASM	27.4 MB	🔒 8.0 MB	gzip	5.7 MB	+2.4 MB (+41.8%)	FAIL

🔒 = the size this artifact is GATED on (ceiling + delta). Binaries gate on file size (installed binary); WASM gates on gzip (download size). The other size is shown for information only.

Details & how to fix

Violations:

• packed-program (Linux) file_bytes: 41.4 MB exceeds limit of 30.1 MB (exceeded by +11.3 MB, policy: max_file_bytes)
• packed-program (Linux) file_delta_pct: +41.9% exceeds limit of 3.0% (exceeded by +38.9pp, policy: max_delta_pct)
• bridge_wasm (WASM) gzip_bytes: 8.0 MB exceeds limit of 5.9 MB (exceeded by +2.2 MB, policy: max_gzip_bytes)
• bridge_wasm (WASM) gzip_delta_pct: +41.8% exceeds limit of 3.0% (exceeded by +38.8pp, policy: max_delta_pct)

Add/update baselines:

.ci/size-gate/wasm32-unknown-unknown.toml:

[artifacts.bridge_wasm]
file_bytes = 27365333
gzip_bytes = 8029812

.ci/size-gate/x86_64-unknown-linux-gnu.toml:

[artifacts.packed-program]
file_bytes = 41368224
gzip_bytes = 16867680

---

Generated by cargo size-gate · workflow run

[harivansh-afk]

Closing: superseded by the shared-runtime consolidation tracked from #5566. Branch kept for reference.