On Fri, Aug 10, 2018 at 12:31:51PM CDT, Zev Weiss wrote:>On Fri, Aug 10, 2018 at 12:16:20PM CDT, Damien Miller wrote:
>>On Fri, 10 Aug 2018, Zev Weiss wrote:
>>
>>>Also x86_64 here (Core i7-4771, for what it's worth). I too
have 4.17
>>>installed, I just haven't yet gotten around to actually running
it, so 4.14 is
>>>what's presently live on the system. If you want I think you
should be able
>>>to replicate that aspect via 'xbps-install linux4.14' (and
then of course
>>>booting said kernel).
>>
>>I booted to linux-4.14 and test_kex still passes. Have you twiddled any
>>sysctls related to ASLR or security?
>>
>>-d
>
>The only entry in my /etc/sysctl.conf is kernel.yama.ptrace_scope = 0;
>kernel.randomize_va_space is at its default value of 2.
>
>I'm a bit busy at the moment, but I'll hopefully be able to dig in
>with a debugger and investigate in more detail sometime in the next
>couple days.
>
...and it turns out it was just a stale build artifact --
regress/unittests/kex/test_kex.o didn't get recompiled after an update
to kex.h that changed the layout of struct kex. (Since from a cursory
glance it doesn't appear that there's currently any effort being made to
track header dependencies for things in the regress/ directory, I'm
assuming this wouldn't be considered a bug worth fixing.)
After a make clean all tests pass -- apologies for the noise.
Zev