ci: run GOROOT for go-test-compat pull requests

[cpunion]

Summary

Run the existing GOROOT matrix as opt-in pull-request checks when a PR has the existing go-test-compat label.

• adding go-test-compat starts the matrix automatically;
• pushing another commit or reopening the PR reruns the matrix while the label remains present;
• adding an unrelated label does not restart an already labeled PR;
• PRs without go-test-compat skip the GOROOT jobs;
• manual dispatch and the nightly schedule remain unchanged;
• no required-check or repository-ruleset change is included.

The existing pull-request concurrency key cancels an older GOROOT run when a labeled PR receives a newer commit.

Validation

• GOTOOLCHAIN=local go run github.com/rhysd/actionlint/cmd/actionlint@v1.7.7 .github/workflows/goroot.yml
• git diff --check

[fennoai]

GOROOT PR label-gating review

This change adds a pull_request trigger to the GOROOT workflow and gates both the goroot matrix job and the goroot-summary job behind the go-test-compat label. Overall it is well-constructed and secure.

Strengths

• Correct trigger: uses pull_request (not pull_request_target) with permissions: contents: read and no secret usage — fork PRs cannot exfiltrate secrets or write to the repo.
• The gate correctly distinguishes the labeled action (reads github.event.label.name) from synchronize/reopened (reads contains(...labels.*.name, ...)), avoiding the common bug where a just-added label isn't yet visible in the label set.
• Both jobs are gated in sync, so on unlabeled PRs the summary job is skipped rather than failing on missing artifacts / the report_count -ne 16 check.

Notes (non-blocking)

• Inline comments below cover the substantive items (CI cost on synchronize, opened coverage, and gate duplication/drift risk).
• Pre-existing, adjacent to the diff: the matrix comment at .github/workflows/goroot.yml:43-44 is contradictory — it states "Go 1.25 is intentionally omitted because it is not a compatibility target," but 1.25.0 is present in the matrix (line 36) and assigned the compatibility lane (lines 39-40). Not introduced by this PR, but worth correcting while nearby.
• The label acts as the trust gate for running fork code; ensure go-test-compat is only appliable by trusted maintainers (a repo-settings concern, outside this diff).

No blocking issues found.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.

📢 Thoughts on this report? Let us know!

[github-actions]

LLGo baseline benchmarks

688be2abfe7e | workflow run | long-term charts

Program measurements

Platform	Workload	File size	vs base	Text size	vs base	Build	vs base	Run	vs base
Linux	cprintf	19256 B	0 B / +0.0%	387 B	0 B / +0.0%	278.885 ms	+20.6 ms / +8.0% (worse)	975.085 us	+147.6 us / +17.8% (worse)
Linux	cprintf-lto	19088 B	0 B / +0.0%	368 B	0 B / +0.0%	270.121 ms	+9.538 ms / +3.7% (worse)	941.397 us	+129.3 us / +15.9% (worse)
Linux	fmtprintf	1653232 B	+8 B / +0.0004839% (worse)	501083 B	0 B / +0.0%	2.183 s	-240.3 ms / -9.9% (better)	2.192 ms	-267.3 us / -10.9% (better)
Linux	fmtprintf-lto	1525824 B	0 B / +0.0%	459700 B	0 B / +0.0%	6.491 s	-104.6 ms / -1.6% (better)	2.019 ms	-71.55 us / -3.4% (better)
Linux	println	62032 B	0 B / +0.0%	15261 B	0 B / +0.0%	262.665 ms	-14.04 ms / -5.1% (better)	1.139 ms	+63.54 us / +5.9% (worse)
Linux	println-lto	53840 B	0 B / +0.0%	12882 B	0 B / +0.0%	413.661 ms	+37.12 ms / +9.9% (worse)	1.166 ms	+163.7 us / +16.3% (worse)
macOS	cprintf	84480 B	0 B / +0.0%	16493 B	0 B / +0.0%	423.501 ms	-90.59 ms / -17.6% (better)	3.290 ms	-1.986 ms / -37.6% (better)
macOS	cprintf-lto	100704 B	0 B / +0.0%	16473 B	0 B / +0.0%	510.667 ms	+44.13 ms / +9.5% (worse)	3.117 ms	+259.9 us / +9.1% (worse)
macOS	fmtprintf	1498256 B	0 B / +0.0%	882133 B	0 B / +0.0%	2.604 s	-878.4 ms / -25.2% (better)	6.361 ms	-581.4 us / -8.4% (better)
macOS	fmtprintf-lto	1208976 B	0 B / +0.0%	875129 B	0 B / +0.0%	6.540 s	-3.73 s / -36.3% (better)	5.345 ms	-8.878 ms / -62.4% (better)
macOS	println	114832 B	0 B / +0.0%	34849 B	0 B / +0.0%	516.193 ms	-7.418 ms / -1.4% (better)	4.116 ms	-955 us / -18.8% (better)
macOS	println-lto	118656 B	0 B / +0.0%	32489 B	0 B / +0.0%	697.318 ms	-234.1 ms / -25.1% (better)	4.377 ms	-3.064 ms / -41.2% (better)

Core language and compiler benchmarks

Platform	Benchmark	ns/op	vs base
Linux	BenchmarkLookupPCRandom	8.089 ns/op	+0.015 ns/op / +0.2% (worse)
Linux	BenchmarkMergeCompilerFlags	100.200 ns/op	-0.4 ns/op / -0.4% (better)
Linux	BenchmarkMergeLinkerFlags	64.080 ns/op	-0.21 ns/op / -0.3% (better)
Linux	BenchmarkChannelBuffered	40 ns/op	-0.04 ns/op / -0.1% (better)
Linux	BenchmarkChannelHandoff	24436 ns/op	+214 ns/op / +0.9% (worse)
Linux	BenchmarkDefer	35.630 ns/op	-2.56 ns/op / -6.7% (better)
Linux	BenchmarkDirectCall	0.247 ns/op	-0.004 ns/op / -1.6% (better)
Linux	BenchmarkGlobalRead	0.355 ns/op	-0.0025 ns/op / -0.7% (better)
Linux	BenchmarkGlobalWrite	6.958 ns/op	-0.016 ns/op / -0.2% (better)
Linux	BenchmarkGoroutine	36129 ns/op	+3538 ns/op / +10.9% (worse)
Linux	BenchmarkInterfaceCall	4.306 ns/op	+0.02 ns/op / +0.5% (worse)
Linux	BenchmarkRuntimeGetG	0.723 ns/op	-0.0001 ns/op / -0.01384% (better)
macOS	BenchmarkLookupPCRandom	11.620 ns/op	-5.35 ns/op / -31.5% (better)
macOS	BenchmarkMergeCompilerFlags	106.200 ns/op	-160.6 ns/op / -60.2% (better)
macOS	BenchmarkMergeLinkerFlags	70.610 ns/op	-147.1 ns/op / -67.6% (better)
macOS	BenchmarkChannelBuffered	24.530 ns/op	-10.77 ns/op / -30.5% (better)
macOS	BenchmarkChannelHandoff	6971 ns/op	-4204 ns/op / -37.6% (better)
macOS	BenchmarkDefer	34.730 ns/op	-10.05 ns/op / -22.4% (better)
macOS	BenchmarkDirectCall	1.251 ns/op	-0.055 ns/op / -4.2% (better)
macOS	BenchmarkGlobalRead	1.159 ns/op	-0.156 ns/op / -11.9% (better)
macOS	BenchmarkGlobalWrite	1.197 ns/op	-0.135 ns/op / -10.1% (better)
macOS	BenchmarkGoroutine	44324 ns/op	+14519 ns/op / +48.7% (worse)
macOS	BenchmarkInterfaceCall	5.424 ns/op	-1.263 ns/op / -18.9% (better)
macOS	BenchmarkRuntimeGetG	2.226 ns/op	-0.794 ns/op / -26.3% (better)

Compared with 6fe8babd468c measured in the same runner job.