Re: [Pharo-dev] threading in Pharo

---

Hi Ben,

On Wed, Mar 26, 2014 at 7:04 AM, Ben Coman <[hidden email]> wrote:

Eliot Miranda wrote:

On Tue, Mar 25, 2014 at 10:21 AM, Eliot Miranda <[hidden email]> wrote:

Hi Igor,

    you have a point but I disagree.  The scheduler is defined in the implementation section of the blue book.  It could be more explicit, but the blue book scheduler is a known and simple system.  In my threading work I've made sure to preserve its semantics (cooperative within priorities, preemptive across priorities, thread switch at activation of non-primitive sends and backward branches).  The only serious bug I know of (that preempting sends a process to the back of its run-queue) was addressed in Cog and the VW VM.  We can and should write up the semantics, but they're not undefined, they're simple and they do work.

 

Oops.  I lie.  There is a serious bug with Semaphore>>critical: and the like.  There's a suspension point after signal and before block evaluation that can result in deadlock if a higher priority process signals the semaphore.  Since the higher priority process is running the lower priority process never makes progress to run the ensure block or cvlear the flag or whatever it needs to do, and hence the system deadlocks.  This requires thought ;-)  So I do know of one outstanding bug.

My grasp of concurrency controls hasn't been tested in 20 years, so naively I would say: If a lower priority process "L" holds** a semaphore when the higher priority process "H" signals the semaphore, temporarily raise the priority of the "L", or temporarily lower the priority of "H".  Is there something simple that I missing?

There are a couple of obvious solutions.  One is to avoid the suspension point somehow, another is adding priority inversion. Adding priority inversion to the scheduler is a significant change which would increase complexity.  You yourself have glossed over detail by saying "temporarily".  I've glossed over a lot with "somehow".  How would these "temporarily"s and "somehow"s be implemented?  Your design sketches are appreciated.

cc'ing vm-dev and squeak dev lists.  this is an important discussion.

cheers -ben

Eliot (phone)

On Mar 25, 2014, at 10:11 AM, Igor Stasenko <[hidden email]> wrote:

On 25 March 2014 17:31, Eliot Miranda <[hidden email]> wrote:

Hi Igor,

On Tue, Mar 25, 2014 at 5:05 AM, Igor Stasenko <[hidden email]> wrote:

On 24 March 2014 22:54, [hidden email] <[hidden email]> wrote:

On Mon, Mar 24, 2014 at 8:23 PM, Alexandre Bergel <[hidden email]> wrote:

>> I am working on a memory model for expandable collection in Pharo. Currently, OrderedCollection, Dictionary and other expandable collections use a internal array to store their data. My new collection library recycle these array instead of letting the garbage collector dispose them. I simply insert the arrays in an ordered collection when an array is not necessary anymore. And I remove one when I need one.
>
> Hm, is that really going to be worth the trouble?

This technique reduces the consumption of about 15% of memory.

>> At the end, #add:  and #remove: are performed on these polls of arrays. I haven’t been able to spot any problem regarding concurrency and I made no effort in preventing them. I have a simple global collection and each call site of "OrderedCollection new” can pick an element of my global collection.
>>
>> I have the impression that I simply need to guard the access to the global poll, which is basically guarding #add:  #remove: and #includes:
>
> One of the AtomicCollections might be the right things for you?

I will have a look at it.

>> What is funny, is that I did not care at all about multi-threading and concurrency, and I have not spotted any problem so far.
>
> There isn’t any ‘multi-threading’ like in Java, you got a much more control version: cooperative on the same priority, preemptive between priorities.
> So, I am not surprised. And well, these operations are likely not to be problematic when they are racy, except when the underling data structure could get into an inconsistent state itself. The overall operations (adding/removing/searing) are racy on the application level anyway.
>
> However, much more interesting would be to know what kind of benefit do you see for such reuse?
> And especially, with Spur around the corner, will it still pay off then? Or is it an application-specific optimization?

I am exploring a new design of the collection library of Pharo. Not all the (academic) ideas will be worth porting into the mainstream of Pharo. But some of them yes.

Thanks for all your help guys! You’re great!

Cheers,
Alexandre

--
_,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:
Alexandre Bergel  http://www.bergel.eu
^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;.

An interesting method I stumbled upon which may help in understanding how these things do work. 

BlockClosure>>valueUnpreemptively

"Evaluate the receiver (block), without the possibility of preemption by higher priority processes. Use this facility VERY sparingly!"

"Think about using Block>>valueUninterruptably first, and think about using Semaphore>>critical: before that, and think about redesigning your application even before that! 

After you've done all that thinking, go right ahead and use it..."

| activeProcess oldPriority result semaphore |

activeProcess := Processor activeProcess.

oldPriority := activeProcess priority.

activeProcess priority: Processor highestPriority.

result := self ensure: [activeProcess priority: oldPriority]. 

I would not recommend you to use this method for anything.

This method heavily relies on how process scheduler works, and in case of any changes, it may break everything.

For the sake of programming, one shall never assume there is a way to "stop the world while i busy doing something".

Really?  Surely any system as interactive as Smalltalk can benefit from a stop-the-rest-of-the-world scheduling facility, and surely packaging it as BlockClosure>>valueUnpreemptively would be a convenient way of doing so.  Surely the right attitude for an implementor of a threading system for Smalltalk would be "Sure, I can implement that, even in a truly concurrent, multi-processor environment".  It may take some doing but it's an important facility to have.  It shouldn't be abused, but when you need it, you need it.

There should be hard guarantees from VM to do it. Right now there's none. That's my point.
Like special primitive(s) for disabling interrupts/scheduling and enabling it back again.

Let us be realistic: the above implementation is based on insider's knowledge how scheduling works, lacking any notion of contract between VM and image.

Right now it just based on implementation detail rather than on guaranteed and well defined semantic.

 

It is no doubt, sometimes you may need such hammer to stop the world.
And it is no doubt (to me) that one should avoid using it unless it is impossible to do otherwise.