Subj : Re: atomic operations api added to AppCore...
To   : comp.programming.threads
From : SenderX
Date : Sun Feb 27 2005 01:12 am

> Actually, there could be ABA, but it wouldn't be a problem: if a node
> gets enqued and dequed between the two times I look at it, it will
> always have been valid anyway, as the associated memory is never freed..

However, If aba does hit I believe it could corrupt the integrity of the 
data-structure itself:



ANCHOR.A->B->C // initial freelist state


thread a
---------
1. lf = ANCHOR.front;
2. ln = lf->next;
3. cas( &ANCHOR.front, lf, ln )


thread b
--------
1. pops A
2. pushes D
3. pushes A



Think if the execution sequence goes something like this:

a1: ( A->B->C ) lf = A
a2: ( A->B->C ) ln = B
b1: ( B->C )  this puts a2 in jeopardy
b2: ( D->B->C )
b3: ( A->D->B->C )  this puts a3 in jeopardy
a3: ( B->C ) corrupted... oh sh^%it!


a3 corrupts the list because its cas was successful, when it should have 
clearly failed. I haven't looked at your code in detail so I don't know if 
you can suffer from this race-condition.

.