> On Sun, 2010-05-09 at 21:31 -0700, Yinghai wrote:
>>> This is clumsy. Doesn't work with NUMA either. Also,
>>
>> it works NUMA. for x86 we managed to call that one time from
>> 2.6.34-rc1. (still called early_res_to_bootmem().
>>
http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commitdiff;h=1842f9...
>>
>>> reserve_bootmem_generic() only exists on x86 afaik.
>>
>> previous patch add weak version
>> [PATCH 06/22] bootmem, x86: Add weak version of
>> reserve_bootmem_generic
>
> It might work on x86, it won't work on sparc as far as I understand.
> In
> any case, I don't think that should be in lmb.c, that's boilerplate
> crap
> that eventually will go away.
>
> One thing you don't seem to get is that several archs already have
> some
> code to move from lmb to bootmem.
>
> The right approach when making something "generic" like that is not to
> make up another one deep into x86 gunkyness and make -that- generic
> ignoring the existing implementations.
>
> Instead, make sure that the existing implementations are in sync, and
> then move -that- to a generic place, eventually adding (in a separate
> patch) the changes needed by x86 if any.
>
