[Lldb-commits] [PATCH] D13362: Enhance test runner timeout logic to detect successful completion of spawned process when stdout/stderr are shared to still-existing child processes

[Pavel Labath via lldb-commits]

labath added a comment.

In http://reviews.llvm.org/D13362#258314, @tfiala wrote:

> The OS X one seems to be related to load (not terribly surprising).  I'm seeing timeouts as I crank up the --threads.  But I don't think that's the whole story.   On the Linux side, I've been cranking down the threads and I'm still seeing it.
>
> In light of the more fundamental issue I just filed here:
>  https://llvm.org/bugs/show_bug.cgi?id=25019
>
> I'm going to address that piece first (upstream) before I take this part on here.  That part needs to be straightened out first.  The then I'll come back here.


I think that is a reasonable course of action. Having three threads servicing the same process seems wasteful/hard to maintain and has a very big race potential.


http://reviews.llvm.org/D13362