runtime, testing: support Goexit, SkipNow, FailNow by jakebailey · Pull Request #5493 · tinygo-org/tinygo

jakebailey · 2026-07-04T00:22:02Z

In #5438 I expanded out panic/recover behavior to cover a lot of (all?) cases.

I forgot about Goexit.

This PR adds support for it, including the work to make it exit pthreads.

Copilot

Pull request overview

Adds end-to-end support for runtime.Goexit in TinyGo, including integrating it with the testing package’s fatal/skip paths and ensuring thread-based schedulers can terminate goroutines cleanly.

Changes:

Implement Goexit as a special unwind/termination path in the panic/defer machinery and schedulers.
Update testing/benchmark runners to execute test bodies in goroutines so FailNow/SkipNow can terminate via Goexit without killing the parent runner.
Add/extend test coverage and golden outputs for Goexit + fatal/skip behaviors.

Reviewed changes

Copilot reviewed 22 out of 22 changed files in this pull request and generated 1 comment.

Show a summary per file

File	Description
testdata/testing.txt	Updates expected output for new fatal subtest behavior.
testdata/testing.go	Adds `t.Fatal`/`t.SkipNow` subtests with arch/OS gating.
testdata/testing-wasm.txt	New expected output variant for wasm / windows-386 gating.
testdata/recover.txt	Extends expected output for Goexit + recovered deferred panics.
testdata/recover.go	Adds Goexit scenarios covering recovered/nested/loop defers.
testdata/goexit.go	New standalone Goexit scenario program for crash/deadlock assertions.
src/testing/testing.go	Implements `FailNow`/`SkipNow` via `runtime.Goexit`; runs subtests in goroutine.
src/testing/benchmark.go	Runs benchmarks in goroutine to support Goexit semantics.
src/runtime/scheduler_threads.go	Adds scheduler `goexit()` path for thread-backed tasks.
src/runtime/scheduler_none.go	Adds `goexit()` implementation for no-scheduler builds.
src/runtime/scheduler_cores.go	Adds `goexit()` for multicore scheduler builds.
src/runtime/scheduler_cooperative.go	Adds `goexit()` for tasks/asyncify cooperative scheduler.
src/runtime/runtime_windows.go	Defines `abort()` on Windows instead of relying on an external symbol.
src/runtime/panic.go	Tracks pending-Goexit across recovered deferred panics; calls `goexit()` on Goexit completion.
src/internal/task/task_threads.go	Adds task-level `Exit()` and refactors deadlock tracking for thread scheduler.
src/internal/task/task_threads.c	Adds `tinygo_task_exit()` using `pthread_exit`.
src/internal/task/task_stack.go	Hooks task lifecycle accounting into stack-based schedulers.
src/internal/task/task_exit.go	Introduces task lifecycle/deadlock logic for tasks/asyncify/cores schedulers.
src/internal/task/task_asyncify.go	Hooks task lifecycle accounting into asyncify scheduler start.
main_test.go	Adds Goexit crash/deadlock regression test; adjusts expected-output routing for testing.go on wasm/win386.
GNUmakefile	Adds `crypto/ecdh` to Linux test packages; updates skip rationale comment.
compiler/testdata/defer-cortex-m-qemu.ll	Updates deferFrame layout expectations for new Goexit flag field.

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

jakebailey · 2026-07-04T18:09:10Z

+import "sync/atomic"
+
+var (
+	mainTask           *Task
+	liveTasks          uint32
+	mainExitedByGoexit uint32
+)
+
+func addLiveTask(t *Task) {
+	if mainTask == nil {
+		mainTask = t
+	}
+	atomic.AddUint32(&liveTasks, 1)
+}
+
+// Exit exits the current task because runtime.Goexit was called.
+func Exit() {
+	exit(true)
+}
+
+func exit(goexit bool) {
+	t := Current()
+	remaining := atomic.AddUint32(&liveTasks, ^uint32(0))
+	if t == mainTask {


I don't think so? the main task is the main task, so it should only happen once, first.

Copilot

Pull request overview

Copilot reviewed 26 out of 26 changed files in this pull request and generated no new comments.

jakebailey · 2026-07-04T21:03:18Z

+	done := make(chan struct{})
+	go func() {
+		defer close(done)
+		tRunner(&sub, f)
+	}()
+	<-done


The fact that this needs to be done is why I had to add more memory to the encoding/xml tests.

Now that tests run on a new goroutine, their stacks are way smaller. On my mac (where I was debugging this), the main thread has something like 8MB, but the new goroutine only has like 512KB or something. The xml tests have a big test that seems to take somewhere between 4MB and 8MB based on what worked in CI.

Copilot

Pull request overview

Copilot reviewed 25 out of 25 changed files in this pull request and generated 1 comment.

Copilot

Pull request overview

Copilot reviewed 25 out of 25 changed files in this pull request and generated 2 comments.

Keep a pending Goexit separate from the active panic state so recovered deferred panics do not cancel the original Goexit unwind. Goexit should also ignore -panic=trap, since it is not a panic. Wire testing.FailNow and SkipNow through runtime.Goexit, and run tests and benchmarks in goroutines. This lets Goexit terminate the user function without skipping test bookkeeping. Add Goexit/recover coverage and enable crypto/ecdh in the Linux stdlib test list.

The threads scheduler handled runtime.Goexit using the generic deadlock path, which parked the current pthread forever. Add a scheduler-specific goexit hook so it can remove the current task and exit the pthread while ordinary blocking deadlocks still block. Track main goroutine Goexit so the last remaining goroutine reports the expected deadlock instead of letting the process exit successfully. Expand the crash coverage with the Goexit cases covered by Big Go's runtime tests.

Top-level tests now run in goroutines, which puts stdlib tests on TinyGo's default host goroutine stack. encoding/xml's depth-limit test needs a larger stack on Linux and Darwin, so run just that package with a larger stack while leaving the rest of the host stdlib tests on the default stack.

Treat panicState as a bitmask so a recovered panic during Goexit can clear the panic bit while preserving the pending Goexit bit. This removes the separate deferFrame Goexit field and restores the previous defer frame size.

jakebailey · 2026-07-06T04:22:31Z

+movl %ebx, 8(%ebx)
+movl %esi, 12(%ebx)
+movl %edi, 16(%ebx)
+movl %ebp, 20(%ebx)
 1:`
-		constraints = "={eax},{ebx},~{ebx},~{ecx},~{edx},~{esi},~{edi},~{ebp},~{xmm0},~{xmm1},~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{fpsr},~{fpcr},~{flags},~{dirflag},~{memory}"
+		constraints = "={eax},{ebx},~{ecx},~{edx},~{xmm0},~{xmm1},~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{fpsr},~{fpcr},~{flags},~{dirflag},~{memory}"


I'll be honest and say I don't have a great understanding of this; sort of like how I had to deal with clobbering problems in my previous PR, the crash is also clobbering, but seemingly so much so that I can't make LLVM do the right thing, and instead need to manually save everything.

jakebailey force-pushed the goexit branch 6 times, most recently from 05448f8 to 493bcb5 Compare July 4, 2026 07:08

dkegel-fastly requested a review from Copilot July 4, 2026 15:45

Copilot started reviewing on behalf of dkegel-fastly July 4, 2026 15:45 View session

Copilot AI reviewed Jul 4, 2026

View reviewed changes

jakebailey force-pushed the goexit branch 5 times, most recently from 1918a97 to d073e86 Compare July 4, 2026 20:55

jakebailey requested a review from Copilot July 4, 2026 20:56

Copilot started reviewing on behalf of jakebailey July 4, 2026 20:56 View session

Copilot AI reviewed Jul 4, 2026

View reviewed changes

jakebailey commented Jul 4, 2026

View reviewed changes

Comment thread src/runtime/panic.go Outdated

jakebailey force-pushed the goexit branch 2 times, most recently from e631e70 to 429be4b Compare July 5, 2026 05:07

jakebailey marked this pull request as ready for review July 5, 2026 15:22

jakebailey requested a review from Copilot July 5, 2026 15:22

Copilot started reviewing on behalf of jakebailey July 5, 2026 15:22 View session

Copilot AI reviewed Jul 5, 2026

View reviewed changes

Comment thread src/runtime/panic.go

jakebailey force-pushed the goexit branch from 429be4b to 8891058 Compare July 5, 2026 17:29

jakebailey requested a review from Copilot July 6, 2026 02:58

Copilot started reviewing on behalf of jakebailey July 6, 2026 02:58 View session

Copilot AI reviewed Jul 6, 2026

View reviewed changes

Comment thread compiler/defer.go

Comment thread src/internal/task/task_threads.go Outdated

jakebailey force-pushed the goexit branch from 8891058 to a570ea0 Compare July 6, 2026 03:20

jakebailey force-pushed the goexit branch 2 times, most recently from 39ed9eb to ee3f918 Compare July 6, 2026 04:16

jakebailey added 5 commits July 5, 2026 21:17

GNUMakefile: change skip reason note

a60fcad

runtime: encode pending Goexit in panic state

e8883dc

Treat panicState as a bitmask so a recovered panic during Goexit can clear the panic bit while preserving the pending Goexit bit. This removes the separate deferFrame Goexit field and restores the previous defer frame size.

jakebailey force-pushed the goexit branch from ee3f918 to e8883dc Compare July 6, 2026 04:17

jakebailey commented Jul 6, 2026

View reviewed changes

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

Uh oh!

runtime, testing: support Goexit, SkipNow, FailNow#5493

runtime, testing: support Goexit, SkipNow, FailNow#5493
jakebailey wants to merge 5 commits into
tinygo-org:devfrom
jakebailey:goexit

jakebailey commented Jul 4, 2026 •

edited

Loading

Uh oh!

Copilot AI left a comment

Uh oh!

jakebailey Jul 4, 2026

Uh oh!

Copilot AI left a comment

Uh oh!

jakebailey Jul 4, 2026

Uh oh!

Uh oh!

Copilot AI left a comment

Uh oh!

Uh oh!

Copilot AI left a comment

Uh oh!

Uh oh!

Uh oh!

jakebailey Jul 6, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

Uh oh!

Uh oh!

Conversation

jakebailey commented Jul 4, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Reviewed changes

Uh oh!

jakebailey Jul 4, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Uh oh!

jakebailey Jul 4, 2026

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Uh oh!

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Uh oh!

Uh oh!

Uh oh!

jakebailey Jul 6, 2026

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

jakebailey commented Jul 4, 2026 •

edited

Loading