Hello Haoqiang, > I think it should be easy since reporting the additional numbers > doesn't require any change to the core logic of the scheduler.
In what way do you suggest that I modify your scheduler to report user/system time separately? > It's charged to the current class. Yes, it might be a cause of > unfairness if interrupts consume significant amount of time. But > isn't it a tradtional thing that Linux does? You are correct in noting that it is the conventional Unix approach, which is precisely what I don't want. Unless it is significantly difficult or expensive to do so, I'd prefer to charge this time to a predefined system class or simply the default /rcfs/taskclass class. Marc > -----Original Message----- > From: Haoqiang Zheng [mailto:[EMAIL PROTECTED] > Sent: Friday, January 28, 2005 9:43 PM > To: Marc E. Fiuczynski > Cc: ckrm-tech > Subject: Re: [ckrm-tech] cpu scheduler feature request > > > On Fri, 28 Jan 2005 13:36:59 -0500, Marc E. Fiuczynski > <[EMAIL PROTECTED]> wrote: > > Looking at a class' stats file, it appears that the CPU > scheduler accounts > > for all time spent on behalf of a class. Would it be difficult > to modify > > the CPU scheduler to maintain stats on how much of that time is in user > > space vs. kernel space? > > I think it should be easy since reporting the additional numbers > doesn't require any change to the core logic of the scheduler. > > > > > Also, when an interrupt is handled by the system, is that time > charged to a > > slice or to the system class? > It's charged to the current class. Yes, it might be a cause of > unfairness if interrupts consume significant amount of time. But > isn't it a tradtional thing that Linux does? > > Haoqiang ------------------------------------------------------- This SF.Net email is sponsored by: IntelliVIEW -- Interactive Reporting Tool for open source databases. Create drag-&-drop reports. Save time by over 75%! Publish reports on the web. Export to DOC, XLS, RTF, etc. Download a FREE copy at http://www.intelliview.com/go/osdn_nl _______________________________________________ ckrm-tech mailing list https://lists.sourceforge.net/lists/listinfo/ckrm-tech
