Re: [PATCH v1 1/2] mm/page_alloc.c: Don't set pages PageReserved() when offlining

[David Hildenbrand]

On 21.10.19 17:47, Michal Hocko wrote:
> On Mon 21-10-19 17:39:36, David Hildenbrand wrote:
> > On 21.10.19 16:43, Michal Hocko wrote:
> [...]
> > > We still set PageReserved before onlining pages and that one should be
> > > good to go as well (memmap_init_zone).
> > > Thanks!
> >
> > memmap_init_zone() is called when onlining memory. There, set all pages to
> > reserved right now (on context == MEMMAP_HOTPLUG). We clear PG_reserved when
> > onlining a page to the buddy (e.g., generic_online_page). If we would online
> > a memory block with holes, we would want to keep all such pages
> > (!pfn_valid()) set to reserved. Also, there might be other side effects.
>
> Isn't it sufficient to have those pages in a poisoned state? They are
> not onlined so their state is basically undefined anyway. I do not see
> how PageReserved makes this any better.

It is what people have been using for a long time. Memory hole -> 
PG_reserved. The memmap is valid, but people want to tell "this here is 
crap, don't look at it".

> Also is the hole inside a hotplugable memory something we really have to
> care about. Has anybody actually seen a platform to require that?

That's what I was asking. I can see "support" for this was added 
basically right from the beginning. I'd say we rip that out and 
cleanup/simplify. I am not aware of a platform that requires this. 
Especially, memory holes on DIMMs (detected during boot) seem like an 
unlikely thing.


--

Thanks,

David / dhildenb