>This morning (yes, in bed ;)) I was thinking about the two options for
>caching SBE-pages, and I got rather confused.
Tststs.. ;)
> For each code page, we have
>an "original" and a "scanned" code page.. (in addition to another page
>which has SBE info in it). It's the "scanned" page that is mapped at the
>corresponding physical
>address, right ?
Scanned page is mapped via entry into I TLB. Guest can access original page
as data. Keep in mind our differentiation into code and data - guest has
access to scanned code only for execution, not for read or write. RW access
is to original code so that the guest handles the "right" code (from its
point of view). Hope I understood and explained it correctly.
But the point about I'm pondering is: how to detect the point of time a
rescan is necessary? Test the D-bit after every instruction is too costly,
but larger time windows may be too large!??
jens
P.S. I'm planning to write some doc about our SBE next days. Have to scan
Kevins virtualization paper and mailing list archiv for that - or has somebody
a summary of all ideas? Kevin?
