Re: [PATCH 4/9] mm, memory_hotplug: get rid of is_zone_device_section

[Michal Hocko]

On Mon 10-04-17 12:20:02, Jerome Glisse wrote:
> On Mon, Apr 10, 2017 at 01:03:46PM +0200, Michal Hocko wrote:
[...]
> > diff --git a/mm/memory_hotplug.c b/mm/memory_hotplug.c
> > index 342332f29364..1570b3eea493 100644
> > --- a/mm/memory_hotplug.c
> > +++ b/mm/memory_hotplug.c
> > @@ -493,7 +493,7 @@ static int __meminit __add_zone(struct zone *zone, unsigned long phys_start_pfn)
> >  }
> >  
> >  static int __meminit __add_section(int nid, struct zone *zone,
> > -					unsigned long phys_start_pfn)
> > +					unsigned long phys_start_pfn, bool want_memblock)
> >  {
> >  	int ret;
> >  
> > @@ -510,7 +510,10 @@ static int __meminit __add_section(int nid, struct zone *zone,
> >  	if (ret < 0)
> >  		return ret;
> >  
> > -	return register_new_memory(nid, __pfn_to_section(phys_start_pfn));
> > +	if (want_memblock)
> > +		ret = register_new_memory(nid, __pfn_to_section(phys_start_pfn));
> > +
> > +	return ret;
> >  }
> 
> The above is wrong for ZONE_DEVICE sparse_add_one_section() will return a
> positive value (on success) thus ret > 0 and other function in the hotplug
> path will interpret positive value as an error.
> 
> I suggest something like:
> 	if (!want_memblock)
> 		return 0;
> 
> 	return register_new_memory(nid, __pfn_to_section(phys_start_pfn));
> }

You are right! I will fold the following. Thanks!
---