>>> "Jan Beulich" <jbeulich@novell.com> 27.02.07 10:45
>>>
>While I see that the example kernel configs have drm generally enabled
>(obviously except for the xenU ones, though I''m a little surprised
ia64''s
>has this on), looking at the sources makes me wonder if this (specifically
>the interaction with AGP) has really been tested. There''s a very
>suspicious use of pte_pfn() in there (this is what actually caught my
>attention the other day when doing something entirely unrelated), two
>__va() instances that appear to take bus addresses as input, and two
>questionable virt_to_phys() ones.
Digging further into this it seems that no x86 capable driver would
currently be affected, and I know too little about the ppc/ia64 ports
to be able to tell whether they could suffer from problems here (there
are AGP drivers for these platforms that set cant_use_aperture, and
that is what - afaict - triggers use of the possibly problematic code
sequences).
Jan
_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xensource.com
http://lists.xensource.com/xen-devel