runtime/pprof: TestLabelSystemstack failure on linux-arm

[mknyszek]

#!watchflakes
post <- pkg == "runtime/pprof" && test == "TestLabelSystemstack"

2024-04-03 15:35 go1.21-linux-arm release-branch.go1.21@d8392e69 runtime/pprof.TestLabelSystemstack [ABORT] (log)

[mknyszek]

In triage, this looks like a timer failed to fire. A time.Sleep of 100 milliseconds never woke up. I wonder if this is fixed in the new timer changes?

[mknyszek]

Given that we rewrote the timer implementation on tip, this only happened once, and only on Go 1.21, let's just wait for more failures.

[prattmic]

Note that there are 2 different 100ms sleeps, both of which failed to wake.

goroutine 26699 [sleep, 9 minutes]:
runtime.gopark(0x1e38c4, 0x184a030, 0x13, 0x13, 0x1)
	/home/swarming/.swarming/w/ir/x/w/goroot/src/runtime/proc.go:398 +0x104 fp=0x158ff8c sp=0x158ff78 pc=0x53078
time.Sleep(0x5f5e100)
	/home/swarming/.swarming/w/ir/x/w/goroot/src/runtime/time.go:195 +0x16c fp=0x158ffa8 sp=0x158ff8c pc=0x8886c
runtime/pprof.profileWriter({0x211d3c, 0x1a67a58})
	/home/swarming/.swarming/w/ir/x/w/goroot/src/runtime/pprof/pprof.go:809 +0x58 fp=0x158ffe0 sp=0x158ffa8 pc=0x16e07c
runtime/pprof.StartCPUProfile.func2()
	/home/swarming/.swarming/w/ir/x/w/goroot/src/runtime/pprof/pprof.go:794 +0x30 fp=0x158ffec sp=0x158ffe0 pc=0x16dfa4
runtime.goexit()
	/home/swarming/.swarming/w/ir/x/w/goroot/src/runtime/asm_arm.s:859 +0x4 fp=0x158ffec sp=0x158ffec pc=0x8a760
created by runtime/pprof.StartCPUProfile in goroutine 26698
	/home/swarming/.swarming/w/ir/x/w/goroot/src/runtime/pprof/pprof.go:794 +0x13c

goroutine 26698 [sleep, 9 minutes]:
runtime.gopark(0x1e38c4, 0x184a060, 0x13, 0x13, 0x1)
	/home/swarming/.swarming/w/ir/x/w/goroot/src/runtime/proc.go:398 +0x104 fp=0x148dd0c sp=0x148dcf8 pc=0x53078
time.Sleep(0x5f5e100)
	/home/swarming/.swarming/w/ir/x/w/goroot/src/runtime/time.go:195 +0x16c fp=0x148dd28 sp=0x148dd0c pc=0x8886c
runtime/pprof.parallelLabelHog({0x21249c, 0x1a67a70}, 0x5f5e100, 0x64)
	/home/swarming/.swarming/w/ir/x/w/goroot/src/runtime/pprof/pprof_test.go:1992 +0xfc fp=0x148dd44 sp=0x148dd28 pc=0x18d670
runtime/pprof.TestLabelSystemstack.func1.1({0x21249c, 0x1a67a70})
	/home/swarming/.swarming/w/ir/x/w/goroot/src/runtime/pprof/pprof_test.go:1885 +0x3c fp=0x148dd5c sp=0x148dd44 pc=0x18ce90
runtime/pprof.Do({0x21245c, 0x318ba0}, {{0x15f7d20, 0x1, 0x1}}, 0x148ddd0)
	/home/swarming/.swarming/w/ir/x/w/goroot/src/runtime/pprof/runtime.go:51 +0xbc fp=0x148dd98 sp=0x148dd5c pc=0x17a860
runtime/pprof.TestLabelSystemstack.func1(0x5f5e100)
	/home/swarming/.swarming/w/ir/x/w/goroot/src/runtime/pprof/pprof_test.go:1884 +0x1d0 fp=0x148ddfc sp=0x148dd98 pc=0x18ce18
runtime/pprof.testCPUProfile(0x17be000, 0x148df74, 0x148def0)
	/home/swarming/.swarming/w/ir/x/w/goroot/src/runtime/pprof/pprof_test.go:484 +0x268 fp=0x148de9c sp=0x148ddfc pc=0x181868
runtime/pprof.TestLabelSystemstack(0x17be000)
	/home/swarming/.swarming/w/ir/x/w/goroot/src/runtime/pprof/pprof_test.go:1883 +0xd4 fp=0x148df94 sp=0x148de9c pc=0x18c178
testing.tRunner(0x17be000, 0x1e34f0)
	/home/swarming/.swarming/w/ir/x/w/goroot/src/testing/testing.go:1595 +0x124 fp=0x148dfe0 sp=0x148df94 pc=0xff968
testing.(*T).Run.func1()
	/home/swarming/.swarming/w/ir/x/w/goroot/src/testing/testing.go:1648 +0x30 fp=0x148dfec sp=0x148dfe0 pc=0x10095c
runtime.goexit()
	/home/swarming/.swarming/w/ir/x/w/goroot/src/runtime/asm_arm.s:859 +0x4 fp=0x148dfec sp=0x148dfec pc=0x8a760
created by testing.(*T).Run in goroutine 1
	/home/swarming/.swarming/w/ir/x/w/goroot/src/testing/testing.go:1648 +0x3f0

[gopherbot]

Found new dashboard test flakes for:

#!watchflakes
post <- pkg == "runtime/pprof" && test == "TestLabelSystemstack"

2024-04-03 15:35 go1.21-linux-arm release-branch.go1.21@d8392e69 runtime/pprof.TestLabelSystemstack [ABORT] (log)

=== RUN   TestLabelSystemstack
SIGQUIT: quit
PC=0x8b78c m=0 sigcode=0

goroutine 0 [idle]:
runtime.futex()
	/home/swarming/.swarming/w/ir/x/w/goroot/src/runtime/sys_linux_arm.s:416 +0x1c fp=0xffab15a8 sp=0xffab15a8 pc=0x8b78c
runtime.futexsleep(0x3080d0, 0x0, 0xffffffffffffffff)
	/home/swarming/.swarming/w/ir/x/w/goroot/src/runtime/os_linux.go:69 +0x60 fp=0xffab15d0 sp=0xffab15a8 pc=0x4b670
runtime.notesleep(0x3080d0)
...
r9      0x308750
r10     0x307b48
fp      0x6
ip      0x4
sp      0xffab15a8
lr      0x4b670
pc      0x8b78c
cpsr    0xa0800010
fault   0x0
*** Test killed with quit: ran too long (10m0s).

— watchflakes