On 16.01.2023 13:01, Wei Chen wrote:
>> -----Original Message-----
>> From: Jan Beulich <jbeul...@suse.com>
>> Sent: 2023年1月16日 19:15
>>
>> On 16.01.2023 10:20, Wei Chen wrote:
>>> On 2023/1/12 16:08, Jan Beulich wrote:
>>>> On 12.01.2023 07:22, Wei Chen wrote:
>>>>>> -----Original Message-----
>>>>>> From: Jan Beulich <jbeul...@suse.com>
>>>>>> Sent: 2023年1月11日 0:38
>>>>>>
>>>>>> On 10.01.2023 09:49, Wei Chen wrote:
>>>>>>> --- a/xen/arch/arm/include/asm/numa.h
>>>>>>> +++ b/xen/arch/arm/include/asm/numa.h
>>>>>>> @@ -22,6 +22,12 @@ typedef u8 nodeid_t;
>>>>>>> */
>>>>>>> #define NR_NODE_MEMBLKS NR_MEM_BANKS
>>>>>>>
>>>>>>> +enum dt_numa_status {
>>>>>>> + DT_NUMA_INIT,
>>>>>>
>>>>>> I don't see any use of this. I also think the name isn't good, as
>> INIT
>>>>>> can be taken for "initializer" as well as "initialized". Suggesting
>> an
>>>>>> alternative would require knowing what the future plans with this are;
>>>>>> right now ...
>>>>>>
>>>>>
>>>>> static enum dt_numa_status __read_mostly device_tree_numa;
>>>>
>>>> There's no DT_NUMA_INIT here. You _imply_ it having a value of zero.
>>>>
>>>
>>> How about I assign device_tree_numa explicitly like:
>>> ... __read_mostly device_tree_numa = DT_NUMA_UNINIT;
>>
>> Well, yes, this is what I was asking for when mentioning the lack of use
>> of the enumerator. Irrespective of that I remain unhappy with the name,
>> though.
>>
>
> How about DT_NUMA_DEF or do you have some suggestions for the name?
Yeah, "DEFAULT" is probably the least bad one.
Jan
