>>> On 01.04.19 at 17:35, wrote:
> On 01/04/2019 15:55, Jan Beulich wrote:
>>
>>> Secondly, when a guest executes CPUID, this doesn't
>>> typically result in Xen executing a CPUID instruction in practice.
>> Wait - this is true for DomU, but used to be false for (PV) Dom0,
>> until you've
On 01/04/2019 15:55, Jan Beulich wrote:
>
>> Secondly, when a guest executes CPUID, this doesn't
>> typically result in Xen executing a CPUID instruction in practice.
> Wait - this is true for DomU, but used to be false for (PV) Dom0,
> until you've switched over from pv_cpuid() to guest_cpuid().
>>> On 01.04.19 at 16:01, wrote:
> There are a number of bugs. There are no read/write hooks on the HVM side, so
> guest accesses fall into the "read/write-discard" defaults, which bypass the
> correct faulting behaviour and the Intel special case.
>
> For the PV side, writes are discarded
There are a number of bugs. There are no read/write hooks on the HVM side, so
guest accesses fall into the "read/write-discard" defaults, which bypass the
correct faulting behaviour and the Intel special case.
For the PV side, writes are discarded (again, bypassing proper faulting),
except for a