[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Xen-devel] regression from 22242:7831b8e5aae2 (x86 guest pagetable walker: check for invalid bits in pagetable entries)?

At 14:34 +0000 on 23 Feb (1330007656), Jan Beulich wrote:
> >>> Tim Deegan <tim@xxxxxxx> 02/23/12 11:34 AM >>>
> >I've applied it, since it seemed not to break anything and I'll be away
> >from my test boxes for a while.  Let me know of you're still seeing any
> >problems. 
> Thanks - I had hoped we would have reported testing results earlier, but
> this unfortunately is going pretty slowly.
> One thing though - shouldn't further walking be suppressed if at a
> given level any violation is detected (i.e. if rc became non-zero)? For
> sure the hardware aborts a walk at least if reserved bits are found
> set, and I'd suspect it also doesn't bother continuing the walk if access
> rights don't permit the result to be used.

That sounds plausible, but I recall there being some subtleties about
that when we first looked at it and I haven't time to dig up what they
were right now.  I may be misremembering but I don't want to tinker with
it without being sure.  I'll look into it again when I get a minute.


Xen-devel mailing list



Lists.xenproject.org is hosted with RackSpace, monitoring our
servers 24x7x365 and backed by RackSpace's Fanatical Support®.