runtime: TestGoroutineLeakProfile failures on Darwin in race mode

[gopherbot]

#!watchflakes
default <- goos == "darwin" && pkg == "runtime" && test == "TestGoroutineLeakProfile"

Issue created automatically to collect these failures.

Example (log):

=== RUN   TestGoroutineLeakProfile/Grpc862
=== PAUSE TestGoroutineLeakProfile/Grpc862
=== CONT  TestGoroutineLeakProfile/Grpc862
    goroutineleakprofile_test.go:518: /Users/swarming/.swarming/w/ir/x/t/go-build132654619/testgoroutineleakprofile/goker.exe Grpc862 (1.15780725s): ok
    goroutineleakprofile_test.go:518: /Users/swarming/.swarming/w/ir/x/t/go-build132654619/testgoroutineleakprofile/goker.exe Grpc862 (1.176112917s): ok
    goroutineleakprofile_test.go:518: /Users/swarming/.swarming/w/ir/x/t/go-build132654619/testgoroutineleakprofile/goker.exe Grpc862: signal: segmentation fault
    goroutineleakprofile_test.go:520: Test Grpc862 produced no output. Is the goroutine leak profile collected?
--- FAIL: TestGoroutineLeakProfile/Grpc862 (2.34s)

— watchflakes

[gopherbot]

Found new dashboard test flakes for:

#!watchflakes
default <- pkg == "runtime" && test == "TestGoroutineLeakProfile/Grpc862"

2025-11-04 01:16 gotip-darwin-arm64-race go@dadbac0c runtime.TestGoroutineLeakProfile/Grpc862 (log)

=== RUN   TestGoroutineLeakProfile/Grpc862
=== PAUSE TestGoroutineLeakProfile/Grpc862
=== CONT  TestGoroutineLeakProfile/Grpc862
    goroutineleakprofile_test.go:518: /Users/swarming/.swarming/w/ir/x/t/go-build132654619/testgoroutineleakprofile/goker.exe Grpc862 (1.15780725s): ok
    goroutineleakprofile_test.go:518: /Users/swarming/.swarming/w/ir/x/t/go-build132654619/testgoroutineleakprofile/goker.exe Grpc862 (1.176112917s): ok
    goroutineleakprofile_test.go:518: /Users/swarming/.swarming/w/ir/x/t/go-build132654619/testgoroutineleakprofile/goker.exe Grpc862: signal: segmentation fault
    goroutineleakprofile_test.go:520: Test Grpc862 produced no output. Is the goroutine leak profile collected?
--- FAIL: TestGoroutineLeakProfile/Grpc862 (2.34s)

2025-11-27 00:20 gotip-darwin-arm64-race go@c048a9a1 runtime.TestGoroutineLeakProfile/Grpc862 (log)

=== RUN   TestGoroutineLeakProfile/Grpc862
=== PAUSE TestGoroutineLeakProfile/Grpc862
=== CONT  TestGoroutineLeakProfile/Grpc862
    goroutineleakprofile_test.go:518: /Users/swarming/.swarming/w/ir/x/t/go-build512501604/testgoroutineleakprofile/goker.exe Grpc862 (1.049224792s): ok
    goroutineleakprofile_test.go:518: /Users/swarming/.swarming/w/ir/x/t/go-build512501604/testgoroutineleakprofile/goker.exe Grpc862 (1.025833083s): ok
    goroutineleakprofile_test.go:518: /Users/swarming/.swarming/w/ir/x/t/go-build512501604/testgoroutineleakprofile/goker.exe Grpc862: signal: segmentation fault
    goroutineleakprofile_test.go:520: Test Grpc862 produced no output. Is the goroutine leak profile collected?
--- FAIL: TestGoroutineLeakProfile/Grpc862 (2.08s)

— watchflakes

[thepudds]

See also #76617, #76577 from the last couple of days.

(Also there are some older possibly related ones like #75822 and #75729, but on the surface, those seem like they might be different than this more recent round of ~3).

[thepudds]

@VladSaiocUber, FYI.

[mknyszek]

Both this and #76646 seem to do something on darwin-amd64-race that causes a crash. If I were to guess, it's probably detecting a real race. But we don't know for sure.

I'll send a CL to print the output of any failing sub-process here. We should hopefully get more information.

[gopherbot]

Change https://go.dev/cl/726524 mentions this issue: runtime: print output in case of segfault in goroutine leak tests