Re: [PATCH] hvm: Allow HVM guests to make console_io hypercall.

[Daniel De Graaf <dgdegra@tycho.nsa.gov>]

On 09/30/2013 04:04 AM, Jan Beulich wrote:
>>>> On 27.09.13 at 23:17, Konrad Rzeszutek Wilk <konrad.wilk@oracle.com> wrote:
>> The console_io hypercall is provided for PV guests and for HVM
>> guests it is done via the 0xe9 port. However the PV hypercall
>> is more efficient as it takes a string rather than one character
>> per write.
>
> While the change itself is fine, there are two questions arising
> from me looking at do_console_io() in this context:
>
> 1) While I can see why VERBOSE ought to control access to
> CONSOLEIO_write, I would think CONSOLEIO_read ought to be
> constrained to the control domain in any case.

I agree on read. Do we still need the restriction on write with
non-VERBOSE builds now that domains' output is restricted to the
guest debug loglevel? I recall this being discussed briefly when
the output changes were being made, but I believe the decision
was only that such a change belonged in a different patch.

> 2) xsm_console_io(), like quite a few other stubs in xsm/dummy.h,
> ignores its "d" parameter and uses current->domain instead. Is
> that really the right way (and if so, why are there other stubs
> that do honor their inputs)?
>
> Jan

No, this is incorrect - it should be using the parameter, same as
any of the other stubs that have a parameter that is always set
to current->domain.

-- 
Daniel De Graaf
National Security Agency