Re: [PATCH v4 2/3] arm: export platform_op XENPF_settime64

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

On 16/11/15 08:08, Ian Campbell wrote:
> On Thu, 2015-11-12 at 17:46 +0000, Stefano Stabellini wrote:
>> Call update_domain_wallclock_time at domain initialization.
>> Set time_offset_seconds to the number of seconds between physical boot
>> and domain initialization: it is going to be used to get/set the
>> wallclock time.
>> Add time_offset_seconds to system_time when before calling do_settime,
>> so that system_time actually accounts for all the time in nsec between
>> machine boot and when the wallclock was set.
>>
>> Expose xsm_platform_op to ARM.
>>
>> Signed-off-by: Stefano Stabellini <stefano.stabellini@eu.citrix.com>
>
> Acked-by: Ian Campbell <ian.campbell@citrix.com>
>
> An aside:[...]
> @@ -1332,6 +1332,75 @@ static int flask_deassign_dtdevice(struct domain
>> *d, const char *dtpath)
>>   }
>>   #endif /* HAS_PASSTHROUGH && HAS_DEVICE_TREE */
>>
>> +static int flask_platform_op(uint32_t op)
>> +{
>> +    switch ( op )
>> +    {
>> +#ifdef CONFIG_X86
>> +    /* These operations have their own XSM hooks */
>> +    case XENPF_cpu_online:
>> +    case XENPF_cpu_offline:
>> +    case XENPF_cpu_hotadd:
>> +    case XENPF_mem_hotadd:
>> +        return 0;
>
> Should this not then be an error (e.g. fail closed)?

During the invocation of these operations, two XSM hooks are called: this
one (from above the switch) and the individual hook (inside the switch).
This hook needs to allow access so that the more detailed hook is called.

> Also, although only implemented today for x86 they don't seem inherently
> any more x86 specific than many of the other things below, so maybe the
> ifdef could be ditched?

The #ifdef is there mostly as a failsafe reminder to ensure that the
implementation for other architectures actually calls the same XSM hooks
that the x86 version does.


-- 
Daniel De Graaf
National Security Agency