The Inbox: System-cwp.662.mcz

Hi Nicolas,

On Mon, Jan 13, 2014 at 10:20 AM, Nicolas Cellier <[hidden email]> wrote:

2014/1/13 Eliot Miranda <[hidden email]>

Hi David,

On Sun, Jan 12, 2014 at 5:13 PM, David T. Lewis <[hidden email]> wrote:

On Sun, Jan 12, 2014 at 09:53:27PM +0100, Bert Freudenberg wrote:
>
> On 12.01.2014, at 20:42, David T. Lewis <[hidden email]> wrote:
>

> > It is worth noting that allObjectsDo: relies on assumptions about how
> > the objects memory works internally. It requires that #someObject will
> > always answer the object at the lowest address in the object memory, and
> > also that a newly allocated object will always be placed at a higher
> > address location than all other objects. Either of these assumptions is
> > likely to be a problem as new and better object memories and garbage
> > collectors are implemented.
> >
> > Dave
>
> Right (as Eliot's vm-dev post shows).
>
> So IMHO the only sensible semantics of allObjectsDo: is as in "allObjects do:" -
> which might be implemented as a primitive in some VMs soonish. It *should not*
> enumerate objects created after calling the method.
>

On Sun, Jan 12, 2014 at 12:01:00PM -0800, Eliot Miranda wrote:
>
> The bug is in implementing allObjects in terms of someObject and nextObject
> in the first place.  It's cheap and cheerful but horribly error-prone and
> restrictive.  It's cheap because the collection of objects doesn't have to
> be created, and on the original 16-bit Smalltalk machines that was really
> important.  It's horribly restrictive because it assumes much about the
> implementation.
>
> Before closures a sentinel wasn't even needed because enumerating the block
> didn't create a new object (the block context was reused).  So the code had
> to be rewritten just to support closures.
>
> Spur has a generation scavenger operating in a distinct new space and that
> doesn't jive well with a consistent ordering at all.  So far the system is
> limping along by tenuring all objects on someObject and someInstance (so
> that newSpace is either empty, or doesn't contain any instances of a
> specific class) and having nextObject enumerate only objects in oldSpace.
>
> But I think now we can afford a primitive that answers all the objects
> (remember that average object size means that such a collection will be ~
> 10% of the heap, average object size in Squeak V3 is about 10.6 words).  At
> least that's what Spur will do, along with an allInstancesOf: primitive.
>  And then the become example won't cause any problems at all.  Far more
> reliable.  I suppose there are circumstances when enumerating without a
> container is the only feasible approach, but VisualWorks has got along with
> only an allObjects primitive for a long time now.  I suspect we can too.
>

Implementation attached. Works on interpreter VM, not yet tested on Cog but
it should be ok there also. If no objections or better suggestions I will
commit it to VMMaker tomorrow.

InterpreterPrimitives>>primitiveAllObjects
        "Answer an array of all objects that exist when the primitive is called, excluding those
        that may be garbage collected as a side effect of allocating the result array. Multiple
        references to nil in the last slots of the array are an indication that garbage collection
        occured, such that some of the unreferenced objects that existed at the time of calling
        the primitive no longer exist. Sender is responsible for handling multiple references to
        nil in the result array."

Instead of filling the unused slots with nil or 0, I think you should shorten the object so that it contains each object only once, and contains only the objects.  Cog contains some code for shortening.  See [New]ObjectMemory>>shorten:toIndexableSize:. 

HTH

--
best,

Eliot

Oh, I missed this one. I can't check now, could it cleanly address my hugly trick of modifying the header like explained at
http://smallissimo.blogspot.fr/2013/04/still-hacking-largeintegerplugins-in.html ?

It would on SqueakV3.  On Spur it could be more difficult as there's no support for one-double-word free objects, so one can't free a singe 64-bit pair (objects are rounded up to 64-bits in length).  But perhaps you should inflict this on me to force my hand ;-).  The problem is that a single 64-bit word is big enough for an object header but not for an object header plus an indirection in the first slot, which takes 129 bits, when rounded up to a 64-bit boundary.  So such fragments are only reclaimable if an object next to them is freed.  They can't be compacted because there's no forwarding pointer.  So I expect in Spur I'd just have to disable the code and always create the copy.  However, what *can* be shortened, always, is the last object in eden.  So if the object is newly created it can easily be shortened; all that needs ot happen is the allocation pointer, freeStart, is cut-back.

HTH,

Eliot

Oh, I missed this one. I can't check now, could it cleanly address my hugly trick of modifying the header like explained at
http://smallissimo.blogspot.fr/2013/04/still-hacking-largeintegerplugins-in.html ?

It would on SqueakV3.  On Spur it could be more difficult as there's no support for one-double-word free objects, so one can't free a singe 64-bit pair (objects are rounded up to 64-bits in length).  But perhaps you should inflict this on me to force my hand ;-).  The problem is that a single 64-bit word is big enough for an object header but not for an object header plus an indirection in the first slot, which takes 129 bits, when rounded up to a 64-bit boundary.

 So such fragments are only reclaimable if an object next to them is freed.  They can't be compacted because there's no forwarding pointer.  So I expect in Spur I'd just have to disable the code and always create the copy.  However, what *can* be shortened, always, is the last object in eden.  So if the object is newly created it can easily be shortened; all that needs ot happen is the allocation pointer, freeStart, is cut-back.

HTH,

Eliot