Re: [RFC][PATCH 2/5] xen:x86: introduce a new hypercall to get RMRR mappings

["Chen, Tiejun" <tiejun.chen@intel.com>]

On 2014/8/15 0:51, Jan Beulich wrote:
>>>> On 14.08.14 at 03:07, <tiejun.chen@intel.com> wrote:
>> On 2014/8/14 2:21, Tian, Kevin wrote:
>>>> From: Chen, Tiejun
>>>> Sent: Tuesday, August 12, 2014 5:40 PM
>>>>
>>>> On 2014/8/12 20:19, Jan Beulich wrote:
>>>>>>>> On 12.08.14 at 12:55, <tiejun.chen@intel.com> wrote:
>>>>>> On 2014/8/8 23:45, Jan Beulich wrote:
>>>>>>>>>> On 07.08.14 at 13:02, <tiejun.chen@intel.com> wrote:
>>>>>>>> +/*
>>>>>>>> + * Returns the RMRR memory map as it was when the domain
>>>>>>>> + * was started.
>>>>>>>> + */
>>>>>>>> +#define XENMEM_RMRR_memory_map           26
>>>>>>>> +typedef struct e820map rmrr_e820_t;
>>>>>>>> +DEFINE_XEN_GUEST_HANDLE(rmrr_e820_t);
>>>>>>>
>>>>>>> Again just as a general remark: What in the world does the "e820"
>>>>>>> in here mean?
>>>>>>
>>>>>> I will redefine a struct to represent this to avoid any confusion.
>>>>>
>>>>> And just to avoid another needless round: The term RMRR shouldn't
>>>>> be in the hypercall public interface definitions either. This needs to
>>>>> be properly abstracted.
>>>>>
>>>>
>>>> Without such a term RMRR I can't figure out what definition should be
>>>> better as you expect, I guess you already have a better case so please
>>>> share it to avoid further discussion.
>>>>
>>>
>>> XENMEM_reserved_memory_map
>>>
>>
>> Okay, but I prefer to XENMEM_DEVICE_reserved_memory_map or
>> XENMEM_PLATFORM_reserved_memory_map since RMRR seems to be dedicated to
>> device or platform, right?
>>
>> Anyway, I'm fine as well once Jan have no any objection to this.
>
> I'd be fine with Kevin's suggestion (as I can't see any other
> reserved memory maps to appear), but I'd also be fine with

I double check this again and looks we already have the following 
structure definition that may be easy to confuse with 
XENMEM_reserved_memory_map,

typedef struct xen_memory_reservation xen_memory_reservation_t;
DEFINE_XEN_GUEST_HANDLE(xen_memory_reservation_t);

> XENMEM_reserved_device_memory_map or some such to

So I'd like to adopt this.

Thanks
Tiejun

> suit your concerns.
>
> Jan
>
>