runtime: TestVDSO failures

[mknyszek]

#!watchflakes
post <- test == "TestVDSO" && date < "2023-08-08"

2023-08-07 19:31 linux-arm64-longtest go@d367ec6a runtime.TestVDSO (log)
2023-08-07 21:16 darwin-amd64-longtest go@008ab9ad runtime.TestVDSO (log)
2023-08-07 19:31 linux-arm64-longtest go@d367ec6a runtime.TestVDSO (log)

[gopherbot]

Found new dashboard test flakes for:

#!watchflakes
post <- test == "TestVDSO"

2023-08-07 19:31 linux-arm64-longtest go@d367ec6a runtime.TestVDSO (log)

--- FAIL: TestVDSO (0.65s)
    crash_test.go:158: /tmp/workdir/tmp/go-build1177904178/testprog.exe SignalInVDSO: exit status 2
    crash_test.go:161: output:
        runtime: g1: frame.sp=0x400029af2f top=0x400029afd0
        	stack=[0x400029a000-0x400029b000
        fatal error: traceback did not unwind completely

        runtime stack:
        runtime.throw({0x10ccf6?, 0x1?})
        	/tmp/workdir/go/src/runtime/panic.go:859 +0x40 fp=0xfffff3b14370 sp=0xfffff3b14340 pc=0x4a230
        runtime.(*unwinder).finishInternal(0xfffff3b14488?)
        	/tmp/workdir/go/src/runtime/traceback.go:561 +0x138 fp=0xfffff3b143b0 sp=0xfffff3b14370 pc=0x710e8
        runtime.(*unwinder).next(0xfffff3b14488?)
        	/tmp/workdir/go/src/runtime/traceback.go:442 +0x2cc fp=0xfffff3b14440 sp=0xfffff3b143b0 pc=0x70efc
        runtime.copystack(0x40000021a0, 0x1000)
        	/tmp/workdir/go/src/runtime/stack.go:931 +0x30c fp=0xfffff3b14840 sp=0xfffff3b14440 pc=0x65c2c
        runtime.newstack()
        	/tmp/workdir/go/src/runtime/stack.go:1113 +0x384 fp=0xfffff3b149f0 sp=0xfffff3b14840 pc=0x660f4
        runtime.morestack()
        	/tmp/workdir/go/src/runtime/asm_arm64.s:316 +0x70 fp=0xfffff3b149f0 sp=0xfffff3b149f0 pc=0x7d5d0


        wanted:
        success

2023-08-07 21:16 darwin-amd64-longtest go@008ab9ad runtime.TestVDSO (log)

--- FAIL: TestVDSO (0.96s)
    crash_test.go:158: /tmp/buildlet/tmp/go-build2813057027/testprog.exe SignalInVDSO: exit status 2
    crash_test.go:161: output:
        runtime: pcdata is 136 and 4 locals stack map entries for main.signalInVDSO (targetpc=0xa477032)
        fatal error: bad symbol table

        runtime stack:
        runtime.throw({0xa47a3ad?, 0x2?})
        	/tmp/buildlet/go/src/runtime/panic.go:859 +0x5c fp=0x70000c7c0b48 sp=0x70000c7c0b18 pc=0xa3ed91c
        runtime.(*stkframe).getStackMap(0x70000c7c0ce8, 0x1?)
        	/tmp/buildlet/go/src/runtime/stkframe.go:202 +0x5bb fp=0x70000c7c0c20 sp=0x70000c7c0b48 pc=0xa40849b
        runtime.adjustframe(0x70000c7c0ce8, 0x70000c7c0d50)
        	/tmp/buildlet/go/src/runtime/stack.go:678 +0x92 fp=0x70000c7c0cb0 sp=0x70000c7c0c20 pc=0xa4061b2
        runtime.copystack(0xc0000061a0, 0x1000?)
        	/tmp/buildlet/go/src/runtime/stack.go:931 +0x30a fp=0x70000c7c0da8 sp=0x70000c7c0cb0 pc=0xa40698a
        runtime.newstack()
        	/tmp/buildlet/go/src/runtime/stack.go:1112 +0x47f fp=0x70000c7c0f58 sp=0x70000c7c0da8 pc=0xa406f5f
        runtime.morestack()
        	/tmp/buildlet/go/src/runtime/asm_amd64.s:593 +0x8f fp=0x70000c7c0f60 sp=0x70000c7c0f58 pc=0xa41d96f


        wanted:
        success

2023-08-10 21:54 linux-s390x-ibm go@bf2a6d19 runtime.TestVDSO (log)

--- FAIL: TestVDSO (0.00s)
    crash_test.go:158: /data/golang/workdir/tmp/go-build1673624692/testprog.exe SignalInVDSO failed to start: context deadline exceeded

— watchflakes

[bcmills]

Note that the linux-s390x-ibm failure was a timeout (#60413), not corruption like the other two.

[bcmills]

I think this is a duplicate of #62101, and probably fixed by https://go.dev/cl/516856.

[aclements]

I concur. Both the commit range and the failure mode indicate that the non-linux-s390x-ibm failures are a duplicate of #62101, which is fixed.