cl: preserve recover-visible panic source lines

[cpunion]

Summary

• retain same-package synchronous call paths whose panic PCs can be observed by a recovering defer
• emit statement PC anchors at existing implicit panic sites without broadening panic generation
• enable 5 unique GOROOT compatibility cases, removing 15 platform/version xfail entries

GOROOT cases

• fixedbugs/bug347.go
• fixedbugs/bug348.go
• fixedbugs/issue27201.go
• fixedbugs/issue29504.go
• fixedbugs/issue4562.go

Scope

Depends on #2034 for the recover-time panic stack snapshot. The new work is kept in four layered commits: regression tests, implementation, tracking-rule documentation, and xfail cleanup.

This intentionally excludes the nil-check cases handled by #2256 and unrelated GC/finalizer timing failures.

Validation

• go test ./cl -run "TestRuntimeCallerFuncSetKeepsRecoverObservableCallees|TestCompileRuntimeCallerPanicPCLineMetadata|TestRuntimeCallerPackageDetection|TestRuntimeCallerAnalysisEdgeCases" -count=1
• go test -vet=off ./test/go -run "^TestRuntimeStatementLineInfo$" -count=1
• macOS/arm64, Go 1.26.5: all 5 GOROOT cases passed with empty xfail/not-applicable files
• Linux/arm64, Go 1.26.5: all 5 GOROOT cases passed with empty xfail/not-applicable files
• Linux/amd64, Go 1.26.5: PR primary build, both test shards, release artifact test, and coverage passed

[fennoai]

Review summary

Careful, well-documented rework of LLGo's recover/panic model: a panicNode linked list on g, per-invocation recover-frame binding, transparent-wrapper aliasing, volatile recover slots, and a conservative mayRecover propagation across cl/, ssa/, and the runtime. The PanicTypeAssert signature change is applied uniformly (verified against the _testrt/_testgo CHECK lines), and nil-handling in the new SSA helpers is solid and test-covered (ssa/recover_metadata_test.go). SIGBUS wiring (SIGBUS != 0 guard + per-platform constants, Linux 0x7, Darwin 0xa) is correct.

One correctness divergence and a few compile-time performance notes are inline. No memory-safety defects were found in the moveToDefer/Recover/abortPanics unlink-and-free ordering.

Non-blocking confirmation (pre-existing, not changed here): Panic still stores the panic value in a non-GC c.Malloc'd node (arg any, z_rt.go:~204), same as before this PR. Since deferred functions run arbitrary allocating user code between the longjmp and Recover, the panic value's GC reachability while a panic is in flight is worth confirming — but this property predates the PR and is out of its scope.

[github-actions]

LLGo baseline benchmarks

58e3ace843f6 | workflow run | long-term charts

Program measurements

Platform	Workload	File size	vs base	Build	vs base	Run	vs base
Linux	cprintf	18656 B	+0.0%	310.592 ms	+0.6% (worse)	1.363 ms	-0.1% (better)
Linux	fmtprintf	4561936 B	+142.4% (worse)	3.498 s	+13.1% (worse)	9.780 ms	+83.2% (worse)
Linux	println	68512 B	+0.0%	311.598 ms	+1.1% (worse)	1.761 ms	-0.7% (better)
macOS	cprintf	84672 B	+0.0%	320.115 ms	+4.1% (worse)	2.495 ms	-26.5% (better)
macOS	fmtprintf	2990064 B	+58.3% (worse)	2.507 s	+10.3% (worse)	13.919 ms	+13.9% (worse)
macOS	println	121216 B	+0.0%	294.650 ms	+4.7% (worse)	3.097 ms	+2.9% (worse)

Core language and compiler benchmarks

Platform	Benchmark	ns/op	vs base
Linux	BenchmarkLookupPCRandom	12.260 ns/op	+0.0%
Linux	BenchmarkMergeCompilerFlags	145.600 ns/op	+0.4% (worse)
Linux	BenchmarkMergeLinkerFlags	98.050 ns/op	+3.7% (worse)
Linux	BenchmarkChannelBuffered	36.390 ns/op	-0.4% (better)
Linux	BenchmarkChannelHandoff	25477 ns/op	+5.7% (worse)
Linux	BenchmarkDefer	48.370 ns/op	-7.2% (better)
Linux	BenchmarkDirectCall	1.757 ns/op	-0.1% (better)
Linux	BenchmarkGlobalRead	1.757 ns/op	-0.1% (better)
Linux	BenchmarkGlobalWrite	2.805 ns/op	-0.0% (better)
Linux	BenchmarkGoroutine	29778 ns/op	-1.1% (better)
Linux	BenchmarkInterfaceCall	8.443 ns/op	-1.3% (better)
Linux	BenchmarkRuntimeGetG	2.116 ns/op	-14.0% (better)
macOS	BenchmarkLookupPCRandom	10.670 ns/op	-1.1% (better)
macOS	BenchmarkMergeCompilerFlags	108.800 ns/op	-31.8% (better)
macOS	BenchmarkMergeLinkerFlags	59.760 ns/op	-25.9% (better)
macOS	BenchmarkChannelBuffered	20.420 ns/op	-4.4% (better)
macOS	BenchmarkChannelHandoff	6089 ns/op	-4.8% (better)
macOS	BenchmarkDefer	24.580 ns/op	-12.6% (better)
macOS	BenchmarkDirectCall	0.946 ns/op	-6.6% (better)
macOS	BenchmarkGlobalRead	1.016 ns/op	-2.4% (better)
macOS	BenchmarkGlobalWrite	0.945 ns/op	-8.8% (better)
macOS	BenchmarkGoroutine	24367 ns/op	-19.9% (better)
macOS	BenchmarkInterfaceCall	4.089 ns/op	-9.7% (better)
macOS	BenchmarkRuntimeGetG	1.887 ns/op	-9.8% (better)

Compared with 6bc422655d08 measured in the same runner job.

[codecov]

Codecov Report

❌ Patch coverage is 95.55556% with 4 lines in your changes missing coverage. Please review.

Files with missing lines	Patch %	Lines
cl/instr.go	94.87%	2 Missing and 2 partials ⚠️

📢 Thoughts on this report? Let us know!