Re: [PATCH v5 2/2] allow hardware domain != dom0

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

On 04/17/2014 03:24 AM, Jan Beulich wrote:
>>>> On 16.04.14 at 21:13, <dgdegra@tycho.nsa.gov> wrote:
>> On 04/16/2014 03:06 PM, Andrew Cooper wrote:
>>> On 16/04/14 19:56, Daniel De Graaf wrote:
>>>>    static unsigned int __read_mostly extra_dom0_irqs = 256;
>>>>    static unsigned int __read_mostly extra_domU_irqs = 32;
>>>>    static void __init parse_extra_guest_irqs(const char *s)
>>>> @@ -192,7 +242,7 @@ custom_param("extra_guest_irqs", parse_extra_guest_irqs);
>>>>    struct domain *domain_create(
>>>>        domid_t domid, unsigned int domcr_flags, uint32_t ssidref)
>>>>    {
>>>> -    struct domain *d, **pd;
>>>> +    struct domain *d, **pd, *old_hwdom = NULL;
>>>>        enum { INIT_xsm = 1u<<0, INIT_watchdog = 1u<<1, INIT_rangeset = 1u<<2,
>>>>               INIT_evtchn = 1u<<3, INIT_gnttab = 1u<<4, INIT_arch = 1u<<5 };
>>>>        int err, init_status = 0;
>>>> @@ -237,10 +287,12 @@ struct domain *domain_create(
>>>>        else if ( domcr_flags & DOMCRF_pvh )
>>>>            d->guest_type = guest_type_pvh;
>>>>
>>>> -    if ( domid == 0 )
>>>> +    if ( domid == 0 || domid == hardware_domid )
>>>>        {
>>>> +        BUG_ON(domid >= DOMID_FIRST_RESERVED);
>>>
>>> Domid is a signed type.
>>>
>>> You also need ensure it is not negative, as assign_integer_param() from
>>> the command line parsing writes all values as unsigned.
>>
>> While this is true, the domain ID has already been validated by the caller
>> of domain_create and so there is no need to check for domid < 0 here.  If
>> someone assigns an out-of-range domain ID to the hardware_domid field, the
>> system will act the same as if any other unused domain ID is specified: a
>> technically working but realistically unusable system.
>
> The thing is that you check the wrong entity: domid is always valid
> (as it is being picked by the caller) - you really want to BUG_ON() or
> panic() upon seeing hardware_domid >= DOMID_FIRST_RESERVED (and
> indeed I think in this case panic() would be the better option, as it
> gives a descriptive message for bad user input rather than a crash
> the reason for which one needs to look up in sources).
>
> As that's the only (non-cosmetic) change request, I'd be fine doing
> that change while committing, unless I hear to the contrary.
>
> Jan

That change is fine with me.


-- 
Daniel De Graaf
National Security Agency