Re: [PATCH v4 1/4] driver/base: Optimize memory block registration to reduce boot time

[Donet Tom <donettom@linux.ibm.com>]

On 5/16/25 3:42 PM, David Hildenbrand wrote:
>>> We'd have to be smart about memory blocks that fall into multiple 
>>> regions,
>>> but it should be a corner case and doable.
>>
>> This is a corner case that should be handled regardless of the loop 
>> order.
>> And I don't think it's handled today at all.
>>
>> If we have a block that crosses node boundaries, current 
>> implementation of
>> register_mem_block_under_node_early() will register it under the first
>> node.
>
> At least upstream behavior should be that it would be linked under all 
> nodes. At least that's what I remember :)
>
>>> OTOH, we usually don't expect having a lot of regions, so iterating 
>>> over
>>> them is probably not a big bottleneck? Anyhow, just wanted to raise it.
>>
>> There would be at least a region per node and having
>>
>> for_each_online_node()
>>     for_each_mem_region()
>>
>> makes the loop O(n²) for no good reason.
>
> Yes, that's why I mentioned it. If we have many nodes it might 
> definitely be relevant.


Thanks David and Mike

I will implement this and send the next version.