On Wed, Mar 2, 2016 at 11:44 AM, Antti Kantee <[email protected]> wrote:

>
> Consuming 100% CPU probably shouldn't happen, though of course one should
> also check by running the same program on the host.  Busylooping usually
> occurs because we have stubbed out some blocking call (e.g. sigsuspend()).
>
> Probably, it could be this case. Because, the Ruby package Example is a
simple http server, which is listening at a specified port and waiting
for connection. I don't know how listening is implemented in Ruby,
but sigsuspend() is quite possible


> I can't give a general recipe for tracking the problem down let alone
> fixing it, apart from "poke around".  Some sort of thread accounting system
> (think ps/top) is planned for the future.  The reason one doesn't exist
> already is that it's not fully straightforward to implement using the
> existing mechanisms.
>
> I will try to use nice and cpulimit for qemu-system process, might be, it
will help


> By "existing sample", do you mean the Example in the Ruby package README?
>
> Yes, Example in the Ruby package


> I cc'd the Ruby package maintainer to improve the chances of getting his
> attention.
>
> Thank you, Antii

Reply via email to