On Mon, 13 May 2002, Rik van Riel wrote: > Hi, > > the following patch implements iowait statistics in a simple way: This follows some work I was doing back in 2.4.10 or so WRT just measuring the delay caused by waiting for page IO. Attached is a patch against 2.4.19-pre8-ac3 which is the offspring of preempt-kernel and Rik's waitio patch. All I can claim is that it still runs and boots, seems to have working preempt, and returns reasonable numbers for iowait on a uni processor. I intend to test more, but I know that this is a popular ac version, and hopefully this will be useful. Until I try this on a real machine and see what tuning of things like hdparm and elvtune do, test rmap again -aa, etc, this is an exercise in getting it to work. I'd be interested in feedback, and I hope Rik will continue his development. -- bill davidsen CTO, TMR Associates, Inc Doing interesting things with little computers since 1979.