From: surya pratap singh raghuvanshi Date: 2015-03-23T01:22:07-07:00 Subject: [ruby-core:68607] Re: Looking for MRI projects for Ruby Google Summer of Code 2015 --089e0158c31a9be4ab0511f05c25 Content-Type: text/plain; charset=UTF-8 what problem will it create if a fibre holding a mutex continue or resume in another thread ? On Sun, Mar 22, 2015 at 12:59 PM, Tony Arcieri wrote: > On Tue, Mar 17, 2015 at 1:59 PM, SASADA Koichi wrote: > >> Fiber captures also machine stack. How to resume it on the other >> threads? It is very interesting. > > > Hi Koichi, > > I put this project on the Ideas List. It is indeed quite difficult. It's a > project I'm potentially interested in mentoring, but if we chose to attempt > it I would really love your input on it. > > There are many potential states where resuming a fiber across threads > simply will not work, for example if the current fiber is holding a mutex. > So the project would start with determining what states a fiber could be in > where cross-thread resumption is allowed. > > It may simply be infeasible without major changes to MRI. I would consider > it more of a research project than anything, and would probably give it a > low chance of success. > > -- > Tony Arcieri > --089e0158c31a9be4ab0511f05c25 Content-Type: text/html; charset=UTF-8 Content-Transfer-Encoding: quoted-printable
what problem will it create if a fibre holding a mutex con= tinue or resume in another thread ?
=C2=A0

On Sun, Mar 22, 2015 at 12:59 PM, = Tony Arcieri <bascule@gmail.com> wrote:
On Tue, Mar 17, 2015 at 1:59 PM, SASADA Koichi = <ko1@atdot.net>= ; wrote:
Fiber captures also machi= ne stack. How to resume it on the other
threads? It is very interesting.

Hi Koichi,=

I put this project on the Ideas List. It is indee= d quite difficult. It's a project I'm potentially interested in men= toring, but if we chose to attempt it I would really love your input on it.=

There are many potential states where resuming a = fiber across threads simply will not work, for example if the current fiber= is holding a mutex. So the project would start with determining what state= s a fiber could be in where cross-thread resumption is allowed.
<= br>
It may simply be infeasible without major changes to MRI. I w= ould consider it more of a research project than anything, and would probab= ly give it a low chance of success.

--
Tony Arcieri

--089e0158c31a9be4ab0511f05c25--