Re: [rvm-research] soft handshake

Hi Albert,

There is nothing simultaneous intended by soft handshakes. We (Fil Pizlo, Daniel Frampton, and I) designed them mostly for use in the biased locking implementation, where a thread can rebias a lock without stopping the world.  It is possible that two or more threads may pass through the softHandshake at about the same time but there are no guarantees of any synchronicity in their processing a soft handshake.

Hope that helps.

Tony


> On 22 Aug 2016, at 11:54 AM, Albert Netymk <[email protected]> wrote:
> 
> I was reading the implementation of
> [softHandshake](https://github.com/JikesRVM/JikesRVM/blob/master/rvm/src/org/jikesrvm/scheduler/RVMThread.java#L3766),
> and my impression is that this method tells all `softHandshake`
> threads to process a yield point on their own pace, assuming all of
> them are running java code (`isInJava() == true`).
> 
> Is it possible that those threads reaches the yield point at the same
> point, and produces a STW? If so, then it would have the same effect
> as hard handshake, right? In other words, it's possible that more than
> one (mutator) thread is stopped even using soft handshake.
> 
> 
> -- 
> Best Regards
> 
> Yours faithfully
> Albert Netymk
> 
> ------------------------------------------------------------------------------
> _______________________________________________
> Jikesrvm-researchers mailing list
> [email protected]
> https://lists.sourceforge.net/lists/listinfo/jikesrvm-researchers


------------------------------------------------------------------------------