Subj : Re: experimental smr-based 100% lock-free eventcount schema for windows... To : comp.programming.threads From : SenderX Date : Wed Feb 02 2005 12:38 am "SenderX" wrote in message news:Uq2dnR0UnJmv953fRVn-hg@comcast.com... > Here we go... This solves the signaling race I found. I think there might be > another racer that could leak waitsets... Humm, I need to study this issue > further... There's another bug, the hazard pointer needs to be set to null in the wait function if the eventcount compare failed. Humm, instead of scribbling this schema down on usenet, I think I'll get some working example code going so others can test it. I might as well add the code to the windows section of my libraries paper. AppCore uses native kernel objects sometimes on the slow paths, so this experiential win32 waitset is sort of pointless to my current library, but I could add it under an optimizations section. I decided to call the evil paper beast the "AppCore System's Journal"... Currently it contains 56+ pages of text and source code ( C and i686 asm ), in tiny little print using two-columns... Its a bitch to write! Might as well be a technical book on how to abstract native sync primitives from multi-operating systems away into common api, and how to marry a lock-free layer directly on top of the native abstraction. :O .