Re: [RESEND PATCH] base/memory: pass the base_section in add_memory_block

[Michal Hocko]

On Wed 14-06-17 14:05:58, Wei Yang wrote:
> Hi, Michael
> 
> I copied your reply here:
> 
> >[Sorry for a late response]
> >
> >On Wed 07-06-17 16:52:12, Wei Yang wrote:
> >> The second parameter of init_memory_block() is used to calculate the
> >> start_section_nr of this block, which means any section in the same block
> >> would get the same start_section_nr.
> >
> >Could you be more specific what is the problem here?
> >
> 
> There is no problem in this code. I just find a unnecessary calculation and
> remove it in this patch.

This code needs a larger rething rather than here and there small
changes I believe.

> >> This patch passes the base_section to init_memory_block(), so that to
> >> reduce a local variable and a check in every loop.
> >
> >But then you are not handling a memblock which starts with a !present
> >section. The code is quite hairy but I do not see why your change is any
> 
> I don't see the situation you pointed here.
> 
> In add_memory_block(), section_nr is used to record the first section which is
> present. And this variable is used to calculate the section which is passed to
> init_memory_block().
> 
> In init_memory_block(), the section got from add_memory_block(), is used to
> calculate scn_nr, but finally transformed to "start_section_nr". That means in
> init_memory_block(), we just need the "start_section_nr" of a memory_block. We
> don't care about who is the first present section.

You are right. The code is confusing as hell!

That being said, I am not opposing the patch but I would much rather
appreciate a consistent cleanup in the whole memblock vs. sections area.
That would be a larger project but the end result is really worth it.
-- 
Michal Hocko
SUSE Labs