ssa: panic with runtime type assertion errors by cpunion · Pull Request #1892 · xgo-dev/llgo

cpunion · 2026-05-22T04:29:53Z

Summary

lower failed non-comma-ok type assertions through a runtime TypeAssertionError helper instead of panic(string)
add stable test/go coverage for recovered runtime.Error values on interface-to-interface and interface-to-concrete assertions
remove fixed fixedbugs/issue16130.go GOROOT xfails; leave issue32187 comparison xfails untouched

Testing

go test ./test/go -run 'TestInterfaceAssert|TestNilInterfaceSameTypeAssertPanics|TestInterfaceCompareUncomparable' -count=1\n- go test ./test/go -count=1\n- (cd runtime && go test ./internal/runtime -run 'TestDoesNotExist' -count=1)\n- LLGO_ROOT="$PWD" go run -tags=dev ./cmd/llgo test -run 'TestInterfaceAssert|TestNilInterfaceSameTypeAssertPanics|TestInterfaceCompareUncomparable' ./test/go\n- go test ./ssa -count=1\n- go test ./test/goroot -run TestGoRootRunCases/fixedbugs/issue16130.go -count=1 -args -goroot "$(go env GOROOT)" -dirs fixedbugs -case '^fixedbugs/issue16130\.go$' -xfail /tmp/llgo-no-xfail.yaml\n- go test ./test/goroot -run TestGoRootRunCases/fixedbugs/issue16130.go -count=1 -args -goroot "$(go env GOROOT)" -dirs fixedbugs -case '^fixedbugs/issue16130\.go$'\n\n## Notes\nGOROOT CI is slow/disabled, so this PR does not depend on automatic full GOROOT runs. The targeted local GOROOT check used the current machine GOROOT (go env GOROOT = /opt/homebrew/Cellar/go@1.24/1.24.11/libexec). Normal PR CI should still run from the fork branch.\n\nA non-targeted local LLGO_ROOT="$PWD" go run -tags=dev ./cmd/llgo test ./test/go run reached test execution but failed on darwin temp-dir cleanup with operation not permitted in existing temp-dir cleanup paths; the focused interface assertion llgo test above passed.

codecov-commenter · 2026-05-22T17:28:01Z

Codecov Report

✅ All modified and coverable lines are covered by tests.

📢 Thoughts on this report? Let us know!

cpunion · 2026-05-24T03:07:12Z

Update for latest interface/nointerface coverage (head c9e354c):

Scope:

interface/nointerface only; no chan/nil/print/recover/GC/liveness/finalizer/goroutine changes.
//go:nointerface methods are recorded during AST directive scan and omitted from runtime ABI method tables, while direct method calls/method expressions remain available.

GOROOT xfails removed after targeted local verification:

typeparam/mdempsky/15.go (//go:nointerface, generics, promoted methods)
fixedbugs/issue47928.go (promoted //go:nointerface method)
typeparam/mdempsky/16.go (interface type assertion panic text; already fixed by this PR's earlier assertion-error work)

Coverage:

Added test/go/nointerface_test.go plus llgo/non-llgo expectation files. Host Go's default toolchain does not filter nointerface without the fieldtrack experiment, while llgo should filter under the llgo tag.

Local tests:

go test ./test/go -count=1
go test ./ssa -count=1
go test ./cl -count=1
go test ./test/goroot -run TestGoRootRunCases -count=1 -args -goroot "$(go env GOROOT)" -dirs typeparam/mdempsky,fixedbugs -case '^(typeparam/mdempsky/(15|16)|fixedbugs/issue47928)\.go$' -directive-mode runlike -run-timeout 120s

Not covered here:

Remaining non-nointerface xfails such as typeparam/nested.go and unrelated chan/nil/print/recover/GC cases are intentionally left for their owning domains.

cpunion · 2026-05-24T04:17:16Z

Update for commit 64ae33c:\n\n- Added reflect/interface type identity coverage for reflect.Type.Method and MethodByName: Method.Func.Interface() now asserts to and calls func(*T) / func(T) using the canonical LLGo closure-backed function type.\n- Fixed reflect.Type interface invokes to request method ABI type initialization, included closure ABI symbols for reflect method FuncOf lookup, and boxed Type.Method Func values as LLGo closure values while preserving public reflect Type identity.\n- recover.go classification: without #1882 it still stops at the existing recover-frame failure (spurious recover). With #1882's recover-frame fix layered locally, the previous func(*T1) vs func(*T1) type identity failure is gone; the next observed failure is missing recover 10, which is recover semantics and intentionally not covered here. No recover.go xfail removed.\n- Local tests: go test ./test/go -count=1; go run ./cmd/llgo test -run TestReflectTypeMethodFuncInterfaceTypeIdentity -count=1 ./test/go; go test ./ssa -count=1; go test ./cl -count=1; go test ./internal/build -count=1; focused goroot recover.go with existing xfail passed.

cpunion · 2026-05-24T06:31:37Z

Coverage update for fixedbugs/issue26094.go (commit 5733be6aa6aa6f2bb3782a00ae49e7bc8c960989):

Classified as the same local type identity / interface assertion root as this draft PR: different scopes can produce the same printed type name, but interface assertion must compare scoped type identity and panic with different scopes.
Added stable test/go coverage in TestInterfaceAssertRejectsSameNameTypesFromDifferentScopes for local-vs-local and local-vs-package same-name type assertions.
Removed only the now-covered fixedbugs/issue26094.go goroot xfail entries for darwin/arm64 and linux/amd64. No nil/chan/print/recover/GC/finalizer/goroutine cases were changed.

Focused local validation:

go test ./test/go -run 'TestInterfaceAssert(RejectsSameNameTypesFromDifferentScopes|To.*Panics)' -count=1\n- go test ./test/goroot -run TestGoRootRunCases/fixedbugs/issue26094.go -count=1 -args -goroot /Users/lijie/sdk/go1.26.0 -dirs fixedbugs -case '^fixedbugs/issue26094\\.go$' -progress 30s\n- go test ./test/go -count=1\n- go test ./ssa/... ./cl/... ./runtime/... -count=1 reported ssa and cl OK; the command itself returned FAIL because ./runtime/... is a separate module path from the repo root.\n- (cd runtime && go test ./... -count=1) passed.

cpunion · 2026-06-06T04:05:32Z

Merged latest xgo-dev/main and resolved the remaining #1892 conflicts.

Conflict/coverage updates:

Kept the mainline reflect Method.Func closure boxing and combined it with this PR's interface assertion closure matching semantics.
Fixed the malformed adjacent xfail entries for typeparam/issue51521.go and typeparam/orderedmap.go while keeping ssa: panic with runtime type assertion errors #1892-owned xfails removed.
Added package-level coverage for //go:nointerface parsing/filtering and reflect-method closure ABI symbol filtering: cl/import_coverage_test.go, ssa/nointerface_test.go, internal/build/main_module_test.go.

Local verification:

go test -timeout 20m ./test/go -run 'Test(InterfaceAssert|NilInterfaceSameTypeAssertPanics|InterfaceCompareUncomparable|NoInterface|ReflectTypeMethodFuncInterfaceTypeIdentity)' -count=1\n- go test -timeout 20m ./ssa ./cl -run 'Test.*(Interface|NoInterface|FuncType|Method|TypeAssert|Import|ABI|Closure)|TestDoesNotExist' -count=1\n- cd runtime && go test -timeout 20m ./internal/runtime ./internal/lib/reflect -count=1\n- go test -timeout 30m ./test/goroot -run TestGoRootRunCases -count=1 -args -goroot /Users/lijie/sdk/go1.26.0 -dirs fixedbugs,typeparam/mdempsky -case '^(fixedbugs/(issue16130|issue26094|issue47928)|typeparam/mdempsky/(15|16))\.go$' -directive-mode runlike -run-timeout 120s\n- go test -timeout 20m ./cl -coverprofile=/tmp/llgo-1892-cl.cover -count=1 (93.8%)\n- go test -timeout 20m ./ssa -coverprofile=/tmp/llgo-1892-ssa.cover -count=1 (87.9%)\n- go test -timeout 20m ./internal/build -coverprofile=/tmp/llgo-1892-build.cover -count=1 (67.4%)\n- LLGO_ROOT="$PWD" go run -tags=dev ./cmd/llgo test -run 'Test(InterfaceAssert|NilInterfaceSameTypeAssertPanics|NoInterface|ReflectTypeMethodFuncInterfaceTypeIdentity)' -count=1 ./test/go\n- git diff --check && git diff --cached --check

…table On Mach-O, pointer slots that name exported functions — every __llgo_stub.* wrapper and any exported Go function — are emitted as chained-fixup BIND nodes, not rebases. The rewriter only decoded rebase nodes, so all stub records (and some entry records) were dropped as unowned and never reached the prebuilt ftab; FuncForPC on function values silently fell back to dladdr (~6µs per fresh pc on darwin). - Parse the LC_DYLD_CHAINED_FIXUPS imports table and resolve bind ordinals to their in-image definitions. - Match canonical owners against the record symbolID with underscore normalization (debug/macho's suffix-shared string table can surface one mangling underscore more or less than the source-level name). - Splice the prebuilt header's base slot back into the fixup chain as a live rebase node: dyld writes the slid text base at load, so the runtime reads a ready runtime PC with no slide arithmetic (non-PIE ELF link-time values already equal runtime addresses). - LLGO_PCLNPOST=0 escape hatch keeps first-use construction. Fresh-pc FuncForPC slow path: darwin 6-8µs -> 1.2-1.7µs, linux 6.8µs -> 0.5µs; first-in-process lookup: darwin ~32µs -> ~14µs, linux ~6.8µs -> ~4µs. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

Pure-compute probes (recursive fib, JSON round-trip, sort.Ints, map churn) with no runtime introspection, so one harness run covers both the introspection extremes and what the funcinfo machinery costs code that never asks for it. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

Go's pclntab pages are touched by its own runtime (traceback, GC) long before user code queries it, so its first FuncForPC never pays page-in. Mirror that: when the prebuilt table is present, init adopts it (zero-copy, sub-µs), touches the pages the lookup path reads (blob, funcinfo records, string offsets, strings), runs one synthetic lookup to warm the code paths, and write-warms the FuncForPC cache pages. First-in-process FuncForPC: darwin ~17µs -> ~2.8µs, linux ~6.6µs -> ~1.0µs. Startup cost is page-count-bound (tens of µs on stdlib-sized tables, invisible next to ~3ms process startup; hello-world medians unchanged). Non-prebuilt binaries stay fully lazy: first-use construction allocates, which has no place in init, and programs that never introspect pay nothing. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

-depths generates deep_<N> scenarios at configurable call depths; -bigsizes generates bigfunc scenarios (funcs x statements) whose large bodies stress statement-level pcline density, mid-function pc symbolization, and ordinary performance of big method bodies. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

This reverts commit 4f2bc0a.

cpunion · 2026-07-02T15:03:43Z

Rebased onto #2016 (codex/pclntab-linkphase-p1, which includes #2012) per the review-order plan: #2012 → #2016 → this line of semantics fixes. Conflicts resolved were additive (context fields, the noinline condition set, and runtime.Panic now calls SavePanicCallerFrames() before the panic-node bookkeeping). Note the PR base is still main, so the diff shows #2012/#2016 commits until those merge.

cpunion · 2026-07-06T01:40:17Z

Superseded by #2031, which re-expresses this PR's surviving value (go:nointerface support, scope-disambiguated assertion messages, MatchesClosure and reflect.Method fixes) on the stage-5 base; the funcinfo/pclntab machinery carried here was superseded by #2012/#2016/#2019.

…upport Re-expresses the surviving value of xgo-dev#1892 on the xgo-dev#2023 base (its other ~10k lines were an earlier draft of the funcinfo/pclntab machinery, since superseded by xgo-dev#2012/xgo-dev#2016/xgo-dev#2019): - //go:nointerface methods no longer satisfy interfaces: the pragma is recorded at import (cl/import.go -> Program.SetNoInterfaceMethod) and filtered out of abi uncommon method tables; GlobalDCE metadata takes the filtered selection list. Retires the typeparam/mdempsky/15 goroot xfails (validated go1.26 darwin, ci directive mode). - Type-assertion failure between same-named types from different scopes appends '(types from different scopes)' (gc wording). - MatchesClosure unwraps closure types on both sides, fixing assertions between closure and plain function types. - reflect.(*rtype).Method builds the method func value with the closure layout allocation (fixes reflect method-value identity). - xfail: retire fixedbugs/issue16130, fixedbugs/issue29735 (validated go1.24 + go1.26 darwin), and reword typeparam/mdempsky/16 - the panic value now implements runtime.Error; the message still lacks the source interface type and prints the command-line-arguments package path.

cpunion mentioned this pull request May 24, 2026

Fix nil deref checks for address expressions #1896

Merged

cpunion mentioned this pull request May 24, 2026

runtime: preserve recover error values #1882

Closed

cpunion marked this pull request as ready for review June 1, 2026 03:17

This was referenced Jun 4, 2026

fix zero-sized deref comparisons #1920

Merged

fix(build): default non-constant untyped shifts #1911

Merged

Fix slice-to-array conversion panics #1899

Merged

cpunion force-pushed the codex/goroot-interface-coverage branch 5 times, most recently from 14ffd15 to f0c4fd2 Compare June 10, 2026 11:30

cpunion force-pushed the codex/goroot-interface-coverage branch from 71068c1 to 2cf0202 Compare June 20, 2026 02:54

cpunion force-pushed the codex/goroot-interface-coverage branch 2 times, most recently from aafb17c to 8412508 Compare June 27, 2026 14:54

cpunion added 11 commits July 2, 2026 03:27

build: add Go-style pclntab findfunc index

2870ba8

ssa: emit DCE-safe function metadata

f365b73

runtime: add line info for stack frames

e81a007

runtime: add statement line caller frames

1e15db2

runtime: compress funcinfo table

407e0dd

test: cover indirect runtime caller paths

a3115cd

cl: narrow runtime caller tracking

c6c128a

runtime: add compact pc-line funcinfo table

4692e15

runtime: optimize FuncForPC metadata lookup

b29906c

runtime: slim FuncForPC cache hot path

734368f

cl: make pc-line labels clone-safe

85b1c13

cpunion and others added 18 commits July 2, 2026 22:01

doc: record P3 findings in pclntab-linkphase design

cd8506d

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

benchmark: keep full LTO from constant-folding the plain fib probe

a3e679c

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

ssa: panic with runtime type assertion errors

61bb547

fix nointerface method tables

3a2c17c

drop fixed mdempsky interface xfail

861d826

reflect: preserve method func type identity

c3c2918

test: cover scoped type assertion identity

1801f66

reflect: fix method func closure values

ac739eb

test: cover nointerface filtering helpers

5b11c1b

test: drop passing goroot xfails

02f99c5

runtime: disambiguate same-name type assertion errors

6de8673

cl: collect detached package linknames

2805be2

Revert "cl: collect detached package linknames"

35b235a

This reverts commit 4f2bc0a.

runtime: detect scoped type assertion names

aacb3cc

cpunion force-pushed the codex/goroot-interface-coverage branch from 8412508 to aacb3cc Compare July 2, 2026 15:03

This was referenced Jul 2, 2026

Proposal: runtime funcinfo metadata and fast FuncForPC lookup #2004

Closed

ssa,runtime: gc-conformant type-assertion panics and go:nointerface support #2031

Merged

cpunion closed this Jul 6, 2026

cpunion mentioned this pull request Jul 8, 2026

runtime,cl,build: Go-style caller info — funcinfo tables, link-phase pcln, FP unwinder, gc-verified conformance (stage-5 backbone) #2024

Merged

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

ssa: panic with runtime type assertion errors#1892

ssa: panic with runtime type assertion errors#1892
cpunion wants to merge 56 commits into
xgo-dev:mainfrom
cpunion:codex/goroot-interface-coverage

cpunion commented May 22, 2026

Uh oh!

codecov-commenter commented May 22, 2026 •

edited by codecov Bot

Loading

Uh oh!

cpunion commented May 24, 2026

Uh oh!

cpunion commented May 24, 2026

Uh oh!

cpunion commented May 24, 2026

Uh oh!

cpunion commented Jun 6, 2026

Uh oh!

cpunion commented Jul 2, 2026

Uh oh!

cpunion commented Jul 6, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

Uh oh!

Conversation

cpunion commented May 22, 2026

Summary

Testing

Uh oh!

codecov-commenter commented May 22, 2026 • edited by codecov Bot Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Codecov Report

Uh oh!

cpunion commented May 24, 2026

Uh oh!

cpunion commented May 24, 2026

Uh oh!

cpunion commented May 24, 2026

Uh oh!

cpunion commented Jun 6, 2026

Uh oh!

cpunion commented Jul 2, 2026

Uh oh!

cpunion commented Jul 6, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

codecov-commenter commented May 22, 2026 •

edited by codecov Bot

Loading