Re: [Xen-devel] [PATCH 04/19] xen: arm: provide and use a handle_raz_wi helper

[Ian Campbell <ian.campbell@xxxxxxxxxx>]

On Thu, 2015-04-02 at 16:45 +0100, Julien Grall wrote:
> 
> On 02/04/2015 16:31, Ian Campbell wrote:
> > On Thu, 2015-04-02 at 16:14 +0100, Julien Grall wrote:
> >> Hi Ian,
> >>
> >> On 31/03/2015 11:07, Ian Campbell wrote:
> >>> Reduces the use of goto in the trap handlers to none.
> >>>
> >>> Some explcitily 32-bit types become register_t here, but that's OK, on
> >>
> >> s/explcitily/explicitly/
> >>
> >>> 32-bit they are 32-bit already and on 64-bit it is fine/harmless to
> >>> set the larger register, a 32-bit guest won't see the top half in any
> >>> case.
> >>
> >> What about 32-bit userspace on 64-bit kernel? Are we sure that a guest
> >> kernel won't only save the bottom half of the register?
> >
> > That would be fine, since the userspace couldn't see the top half anyway
> > so not saving it doesn't hurt.
> >
> > In any case, the trap here has been talking from 32-bit mode and that is
> > where we will return, so I'm not sure the guest kernel enters the
> > picture, does it?
> 
> It's possible for the kernel to access only a part of the 64 bit 
> registers and preserve the other part with a valid data to use later.
> 
> AFAICT, nothing prevent a guest to use the top half of the registers for 
> his own purpose. It would only need to save/restore the bottom half of a 
> 64 bit registers.

Writing to the bottom half (e.g. w0) of a register implicitly clears the
top half, IIRC, so I think a kernel is unlikely to want to do this, even
if it could (which I'm not quite convinced of).

> 
> By resetting the 64-bit register, we will corrupt the top half of the 
> registers and potentially (if the use case is valid) crash the kernel.
> 
> Although I didn't say that I would write a such guest ;)
> 
> Regards,
> 



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