runtime/debug: SIGSEGV in dumpgoroutine

[bcmills]

From linux-ppc64le-buildlet in https://build.golang.org/log/e905b629a5a27c512d13fcbeffe416796327bba3:

fatal error: unexpected signal during runtime execution
[signal SIGSEGV: segmentation violation code=0x1 addr=0x0 pc=0x18cbc]

runtime stack:
runtime.throw(0x15c4a9, 0x2a)
	/workdir/go/src/runtime/panic.go:1045 +0x5c
runtime.sigpanic()
	/workdir/go/src/runtime/signal_unix.go:573 +0x474
runtime.dumpgoroutine(0xc000001680)
	/workdir/go/src/runtime/heapdump.go:374 +0x23c
runtime.dumpgs()
	/workdir/go/src/runtime/heapdump.go:403 +0xa8
runtime.mdump()
	/workdir/go/src/runtime/heapdump.go:649 +0xc0
runtime.writeheapdump_m(0x7)
	/workdir/go/src/runtime/heapdump.go:672 +0x6c
runtime.runtime_debug_WriteHeapDump.func1()
	/workdir/go/src/runtime/heapdump.go:24 +0x2c
runtime.systemstack(0x0)
	/workdir/go/src/runtime/asm_ppc64x.s:269 +0x94
runtime.mstart()
	/workdir/go/src/runtime/proc.go:1069

goroutine 21 [waiting]:
runtime.systemstack_switch()
	/workdir/go/src/runtime/asm_ppc64x.s:216 +0x10 fp=0xc0001c4e50 sp=0xc0001c4e30 pc=0x6de00
runtime/debug.WriteHeapDump(0x7)
	/workdir/go/src/runtime/heapdump.go:23 +0x60 fp=0xc0001c4e90 sp=0xc0001c4e50 pc=0x17dd0
runtime/debug_test.TestWriteHeapDumpNonempty(0xc0000aa100)
	/workdir/go/src/runtime/debug/heapdump_test.go:25 +0x124 fp=0xc0001c4f60 sp=0xc0001c4e90 pc=0x103b54
testing.tRunner(0xc0000aa100, 0x15e690)
	/workdir/go/src/testing/testing.go:916 +0xc0 fp=0xc0001c4fb0 sp=0xc0001c4f60 pc=0xca1a0
runtime.goexit()
	/workdir/go/src/runtime/asm_ppc64x.s:884 +0x4 fp=0xc0001c4fb0 sp=0xc0001c4fb0 pc=0x703c4
created by testing.(*T).Run
	/workdir/go/src/testing/testing.go:967 +0x2ec

goroutine 1 [chan receive]:
testing.(*T).Run(0xc0000aa100, 0x158713, 0x19, 0x15e690, 0x301)
	/workdir/go/src/testing/testing.go:968 +0x308
testing.runTests.func1(0xc00008c000)
	/workdir/go/src/testing/testing.go:1209 +0x78
testing.tRunner(0xc00008c000, 0xc000054d88)
	/workdir/go/src/testing/testing.go:916 +0xc0
testing.runTests(0xc00000c060, 0x2455e0, 0x7, 0x7, 0x0)
	/workdir/go/src/testing/testing.go:1207 +0x27c
testing.(*M).Run(0xc00000a080, 0x0)
	/workdir/go/src/testing/testing.go:1124 +0x150
main.main()
	_testmain.go:58 +0x130

goroutine 9 [runnable]:
runtime.Goexit()
	/workdir/go/src/runtime/panic.go:608 +0x16c
testing.(*common).SkipNow(0xc00008c200)
	/workdir/go/src/testing/testing.go:749 +0x3c
testing.(*common).Skipf(0xc00008c200, 0x15def2, 0x4a, 0xc000078020, 0x1, 0x1)
	/workdir/go/src/testing/testing.go:735 +0x7c
internal/testenv.SkipFlaky(0x182500, 0xc00008c200, 0x4e6c)
	/workdir/go/src/internal/testenv/testenv.go:241 +0xe4
runtime/debug_test.TestSetGCPercent(0xc00008c200)
	/workdir/go/src/runtime/debug/garbage_test.go:114 +0x5c
testing.tRunner(0xc00008c200, 0x15e670)
	/workdir/go/src/testing/testing.go:916 +0xc0
created by testing.(*T).Run
	/workdir/go/src/testing/testing.go:967 +0x2ec
FAIL	runtime/debug	0.052s

CC @cuonglm @aclements @cherrymui

[cuonglm]

Seems a flaky for me, another fail https://build.golang.org/log/ac9672a53d4ece77dcb8acca1cf4a468c719cceb

[bcmills]

Here's one on freebsd-arm-paulzhol: https://build.golang.org/log/47a67881527b85bd923b75d9aa910a7f62598790

[aclements]

This is on the line

		dumpint(uint64(uintptr(unsafe.Pointer(d.fn.fn))))

where d is a defer.

@danscales, didn't open-coded defers make it possible for there to be a defer object on the chain with _defer.fn set to nil?

Probably we should just emit nil in the heap dump in that case. To the extent that the heap dump is meaningful at all at this point, I think the defer records are only used to find roots the way the GC does, and the GC just ignores d.fn if it's nil.

[gopherbot]

Change https://golang.org/cl/203697 mentions this issue: runtime: fix dumpgoroutine() to deal with open-coded defers

[laboger]

The Goexit failure is discussed here #34575

…

On Sat, Oct 26, 2019 at 3:05 AM Cuong Manh Le ***@***.***> wrote: Seems a flaky for me, another fail https://build.golang.org/log/ac9672a53d4ece77dcb8acca1cf4a468c719cceb — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#35172?email_source=notifications&email_token=ACH7BDFBHYMN4RAQ44MD6DDQQP24BA5CNFSM4JFLKWLKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOECKCSSY#issuecomment-546580811>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ACH7BDCMRBKB7VJLG2U6SATQQP24BANCNFSM4JFLKWLA> .