fix(cache): preserve Cache Components ownership during probing

[james-elicx]

Stacked on #3094.

What this fixes

This follow-up aligns staged cacheability probes with Next.js Cache Components ownership semantics:

• a public use cache boundary owns an inner fetch(..., { cache: "no-store" }), so that fetch does not demote the enclosing route
• use cache: private marks the route dynamic and suspends before private user code or cache-key construction
• the outer prerender owns that suspension, so user try/catch and Suspense fallbacks cannot turn it into a static certification
• use cache: private inside public use cache or unstable_cache() fails synchronously, matching Next.js, and cannot persist private output in the outer shared cache
• a genuine application 5xx remains a probe failure rather than being hidden by earlier dynamic usage
• the ordinary App Router request path avoids probe-only ownership allocation

The unsafe legacy-payload upgrade and bounded-fill coordinator have been split into #3099 and #3100. They are not part of this diff.

Next.js parity

The workerd E2E ports the relevant ownership and invalid-nesting behavior from:

• https://github.com/vercel/next.js/blob/canary/test/e2e/app-dir/use-cache/use-cache.test.ts
• https://github.com/vercel/next.js/blob/canary/test/e2e/app-dir/cache-components-errors/use-cache-private.util.ts
• https://github.com/vercel/next.js/blob/canary/packages/next/src/server/use-cache/use-cache-wrapper.ts

[pkg-pr-new]

Open in StackBlitz

npm i https://pkg.pr.new/@vinext/cloudflare@3095

npm i https://pkg.pr.new/create-vinext-app@3095

npm i https://pkg.pr.new/@vinext/types@3095

npm i https://pkg.pr.new/vinext@3095

commit: ad26419

[github-actions]

Performance benchmarks

Compared ad26419 against base 9a08dfc using alternating same-runner rounds. Next.js was unchanged and skipped.

0 improved · 0 regressed · 6 within ±1.5%

Scenario	Framework	Baseline	Current	Change
Client bundle size (gzip)	vinext	142.3 KB	142.3 KB	⚫ +0.0%
Client entry size (gzip)	vinext	129.6 KB	129.6 KB	⚫ +0.0%
Dev server cold start	vinext	3.08 s	3.13 s	⚫ +1.4%
Production build time	vinext	3.43 s	3.42 s	⚫ -0.2%
RSC entry closure size (gzip)	vinext	117.5 KB	117.7 KB	⚫ +0.2%
Server bundle size (gzip)	vinext	199.2 KB	199.5 KB	⚫ +0.1%

View detailed results and traces

_{🟢 improvement · 🔴 regression · ⚫ change below 1.5% · paired base/head}

[github-actions]

Example	Preview	Production	Original
app-router-cloudflare	preview	production
pages-router-cloudflare	preview	production
app-router-playground	preview	production	original
realworld-api-rest	preview	production
nextra-docs-template	preview	production
benchmarks	preview	production
hackernews	preview	production	original
workers-cache	preview	production
web	preview	production

[james-elicx]

/bigbonk review for issues

[james-elicx]

/bigbonk review for issues

[james-elicx]

/bigbonk review for issues

[ask-bonk]

@james-elicx Bonk workflow was cancelled.

View workflow run · To retry, trigger Bonk again.

[james-elicx]

/bigbonk review for issues

[james-elicx]

/bigbonk review for issues

Hard time limit: 20 minutes. Return all findings available before the deadline; do not wait indefinitely for additional analysis.

[james-elicx]

/bigbonk review for issues

Hard time limit: 20 minutes. Review exact head 6291704 and return all findings available before the deadline; do not wait indefinitely for additional analysis.

[james-elicx]

/bigbonk review for issues

Hard time limit: 20 minutes. Review exact head 7f4001a and return all findings available before the deadline; do not wait indefinitely.

[james-elicx]

/bigbonk review for issues

Hard time limit: 20 minutes. Review exact head 7f4001a and return every finding available before the deadline. If the full review cannot finish, post the partial verdict instead of waiting or cancelling without a result.