Re: [Xen-devel] [V8 2/4] x86/xsaves: enable xsaves/xrstors/xsavec in xen

[Shuai Ruan <shuai.ruan@xxxxxxxxxxxxxxx>]

On Tue, Oct 27, 2015 at 02:13:03AM -0600, Jan Beulich wrote:
> >> > +
> >> > +    ((struct xsave_struct *)dest)->xsave_hdr.xstate_bv = xstate_bv;
> >> > +    ((struct xsave_struct *)dest)->xsave_hdr.xcomp_bv =  0;
> >> 
> >> Wouldn't it be more logical to also memcpy() the header? Which
> >> would do away with at least one of these ugly casts, would
> >> allow folding with the memcpy() done right before, _and_ would
> >> eliminate an (apparent or real I'm not sure without looking in
> >> more detail) information leak (the remainder of the header).
> >> 
> > For machine with no-xsaves support, xcomp_bv should be zero or it will cause
> > GP fault. So we can not just memcpy the header when perform save.
> 
> See how I limited my reply by saying "at least one", which I added
> after realizing that fact? IOW I'd expect you to use memcpy() and
> then only zero that one field. Or you'd memset() the whole header
> to zero, and then only store xstate_bv. As said, otherwise it looks
> like you're adding an information leak.
> 
> Jan
Ok And sorry for misunderstanding your meaning.
> 
> 
> _______________________________________________
> Xen-devel mailing list
> Xen-devel@xxxxxxxxxxxxx
> http://lists.xen.org/xen-devel
> 

_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
http://lists.xen.org/xen-devel