> >> This is definitely not a very detailed analysis of what's happening, but
> >> I would think that the latter process takes at least as much time as the
> >> former. In other words, I don't see how the altp2m case improves
> >> performance.
> >
> > I don't think it does either.
>
> But then I have misread your statement that "However, this solution
> (while supported) is not particularly ideal as it still creates
> significant performance overhead." - I've read it to mean that the
> significant performance overhead you're talking about applies to the
> emulation case, as opposed to the better altp2m way - which, as it turns
> out, neither of us believes.
>
Right, what I meant there is execution tracing with EPT is not optimal because of the extra overhead of the unrelated instruction fetch violations on the page. The performance gain is achieved with the breakpoint trick in DRAKVUF, not specific to altp2m (only triggers at specific points instead of the whole page). It only applies for excecution tracing, but in my use-case that had been the major goal and thus the main source of overhead. Here I highlight that this lighter monitoring can still be used on multi-vCPU systems safely with altp2m by doing the shadow copy/remapping method. I can try to rework this part a bit to avoid the confusion.
Cheers,
Tamas
_______________________________________________
Publicity mailing list
Publicity@xxxxxxxxxxxxxxxxxxxxhttp://lists.xenproject.org/cgi-bin/mailman/listinfo/publicity
