Re: [Jikesrvm-researchers] Can two or more threads enter heapExhausted() in VM_Allocator?

[Perry C. <per...@us...>]

If you look at the semi-space collector's heapExhausted method, you'll see
that it calls VM_GCUtil.outOfMemory, gc1, or sysFail.  The body of
outOfMemory  is guarded by an explicit lock outOfMemoryLock.  The call to
sysFail is a fatal condition so it doesn't matter much.  The call to gc1 is
the typical case and it is certainly possible that multiple threads call
it.  However, gc1 merely calls VM_Collector.collect which then calls the
synchronized VM_HandShake.requestAndAwaitCompletion.

Perry


"Kim, Woo Jin" <wj...@ss...>@www-124.southbury.usf.ibm.com on
07/11/2002 02:01:46 AM

Please respond to jik...@ww...

Sent by:    jik...@ww...


To:    "jalapeno researchers"
       <jik...@ww...>
cc:
Subject:    [Jikesrvm-researchers] Can two or more threads enter
       heapExhausted() in VM_Allocator?



Hello,

I'm studying garbage collection. I have a question about synchronization.

I think that there is no synchronization in heapExhausted() method in
VM_Allocator.

If it is true, two or more threads (VM_Processors, exactly) can enter this
method.

Is there any problem in this case?

I can't figure out that no problem occurs in this case.

In addition, I am not sure that this case can occur because of
quasi-preemption feature of RVM.

Which is true? That fact helps me implement what I want corrently.

Thank you. I hope to see your reply.

_______________________________________________
Jikesrvm-researchers mailing list
Jik...@ww...
http://www-124.ibm.com/developerworks/oss/mailman/listinfo/jikesrvm-researchers