Ryan Zezeski writes:

>
> I have another stuck build and it looks like the futex implementation
> may be the culprit. I will run more tests to verify futex is always
> invovled when stuck. Here is my latest report:
>
> https://gist.github.com/rzezeski/8a967f9cfdd3976869eb#comment-1479375

I dug into this more today and have a few updates on my gist, here is
the latest:

https://gist.github.com/rzezeski/8a967f9cfdd3976869eb#comment-1480012

At this point there are two key pieces of information:

1. The bug-18026 test always locks up on mono `4.0.1.44` and `4.0.2.4`
   but not on master. I'm going to try the 4.2 branch on request of the
   mono creator.

2. When things do lock up they involve the futex syscall. There was a
   recent Linux bug with futex that was often hit on Haswell CPUs (my
   tests have been run on Haswell and Sandybridge) and the symptoms are
   _much_ like what I've seen. Of course, lx has no linux kernel so it
   doesn't directly apply but maybe something similar is happening in
   the lx code?

   https://groups.google.com/forum/#!topic/mechanical-sympathy/QbmpZxp6C64

-Z


-------------------------------------------
smartos-discuss
Archives: https://www.listbox.com/member/archive/184463/=now
RSS Feed: https://www.listbox.com/member/archive/rss/184463/25769125-55cfbc00
Modify Your Subscription: 
https://www.listbox.com/member/?member_id=25769125&id_secret=25769125-7688e9fb
Powered by Listbox: http://www.listbox.com

Reply via email to