Re: [PATCH v4 15/17] x86/VPMU: NMI-based VPMU support

[Boris Ostrovsky <boris.ostrovsky@oracle.com>]

On 02/04/2014 11:41 AM, Jan Beulich wrote:
>>>> On 04.02.14 at 17:31, Boris Ostrovsky <boris.ostrovsky@oracle.com> wrote:
>> On 02/04/2014 06:48 AM, Jan Beulich wrote:
>>> And then I'm still missing a reasonable level of analysis that the
>>> previously non-NMI-only interrupt handler is now safe to use in NMI
>>> context.
>> How about this?
> Looks okay, except ...
>
>> With send_guest_vcpu_virq() and hvm_get_segment_register() for PV(H) and
>> vlapic accesses for HVM moved to sofint, the only routines/macros that
>> vpmu_do_interrupt() calls in NMI mode are:
>> * memcpy()
>> * querying domain type (is_XX_domain())
>> * guest_cpu_user_regs()
>> * XLAT_cpu_user_regs()
>> * raise_softirq()
>> * vcpu_vpmu()
>> * vpmu_ops->arch_vpmu_save()
>> * vpmu_ops->do_interrupt() (in the future for PVH support)
>>
>> The latter two can only access PMU MSRs.
> ... that this additionally needs to exclude things like
> {rd,wr}msr_safe() (i.e. stuff raising exceptions that normally
> get recovered from).


I'll add that. And probably a comment in vendor-specific code to remind 
people that these routines need to be NMI-safe.


-boris