On 06/24/2014 03:41 AM, Nishanth Aravamudan wrote:
> On 21.06.2014 [13:08:59 +1000], Alexey Kardashevskiy wrote:
>> On 06/21/2014 05:10 AM, Nishanth Aravamudan wrote:
>>> On 16.06.2014 [17:53:49 +1000], Alexey Kardashevskiy wrote:
>>>> Current QEMU does not support memoryless NUMA nodes.
>>>> This prepares SPAPR for that.
>>>>
>>>> This moves 2 calls of spapr_populate_memory_node() into
>>>> the existing loop which handles nodes other than than
>>>> the first one.
>>>
>>> <snip>
>>>
>>>> @@ -719,6 +704,12 @@ static int spapr_populate_memory(sPAPREnvironment
>>>> *spapr, void *fdt)
>>>> node_size = ram_size - mem_start;
>>>> }
>>>> }
>>>> + if (!mem_start) {
>>>> + /* ppc_spapr_init() checks for rma_size <= node0_size already
>>>> */
>>>> + spapr_populate_memory_node(fdt, i, 0, spapr->rma_size);
>>>> + mem_start += spapr->rma_size;
>>>> + node_size -= spapr->rma_size;
>>>> + }
>>>
>>> Why is this needed to be separate? The RMA fits in the first node, per
>>> the comment and the prior checks, so can't we just leave the first node
>>> alone?
>>
>> This is the way to tell SLOF what memory it can use. It can use RMA and it
>> will use first available memory node.
>
> Right, but why does the RMA need to be in it's own memory node? Can't it
> just be part of the first present memory node?
As I understand SLOF can try using the whole node.
--
Alexey
