port-amd64/51934: ATF tests hang at kernel/t_ptrace_wait:signal4 under qemu

[gson%gson.org@localhost (Andreas Gustafsson)]

>Number:         51934
>Category:       port-amd64
>Synopsis:       ATF tests hang at kernel/t_ptrace_wait:signal4 under qemu
>Confidential:   no
>Severity:       serious
>Priority:       medium
>Responsible:    port-amd64-maintainer
>State:          open
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Wed Feb 01 15:30:00 +0000 2017
>Originator:     Andreas Gustafsson
>Release:        NetBSD-current, source date >= 2017.01.26.22.32.48
>Organization:

>Environment:
System: NetBSD guava.gson.org 7.0 NetBSD 7.0 (GENERIC) #0: Thu Mar 3 22:52:02 EET 2016 gson%guido.araneus.fi@localhost:/bracket/prod/7.0gson1/amd64/obj/sys/arch/amd64/compile/GENERIC amd64
Architecture: x86_64
Machine: amd64
>Description:

Since kamil's commit of src/tests/kernel/t_ptrace_wait.c 1.64 at CVS
source date 2017.01.26.22.32.48, the ATF test runs on the TNF testbed
have been timing out.  They appear to hang while running the signal4
test case of the kernel/t_ptrace_wait test.  The log from the most
recent failure is here:

  http://releng.netbsd.org/b5reports/amd64/build/2017.01.31.09.17.53/test.log

This does not happen on real hardware, so it may well be a qemu bug
rather than a NetBSD one, but I figure this should have a PR anyway so
we have a PR number to refer to if the test gets marked as an
"expected failure".

I'm wondering if this might be related to kern/49603, since it also
affects traced processes on amd64 under qemu.

>How-To-Repeat:

Run the ATF tests on -current/amd64 under qemu, or inspect the testbed
logs.

>Fix: