Re: [PATCH 09/15] mm/hmm/devmem: device memory hotplug using ZONE_DEVICE v6

From: Jerome Glisse
Date: Mon Jul 31 2017 - 13:21:33 EST


On Fri, Jul 28, 2017 at 01:10:03PM +0200, Michal Hocko wrote:
> I haven't seen a newer version posted but the same comment applies on
> your hmm-v25-4.9 git version from
> git://people.freedesktop.org/~glisse/linux
>
> On Wed 28-06-17 14:00:41, Jérôme Glisse wrote:
> > This introduce a simple struct and associated helpers for device driver
> > to use when hotpluging un-addressable device memory as ZONE_DEVICE. It
> > will find a unuse physical address range and trigger memory hotplug for
> > it which allocates and initialize struct page for the device memory.
>
> Please document the hotplug semantic some more please (who is in charge,
> what is the lifetime, userspace API to add/remove this memory if any
> etc...).
>
> I can see you call add_pages. Please document why arch_add_memory (like
> devm_memremap_pages) is not used. You also never seem to online the
> range which is in line with nvdim usage and it is OK. But then I fail to
> understand why you need

I added documentation in function and in commit message:
https://cgit.freedesktop.org/~glisse/linux/commit/?h=hmm-next&id=33e236a64da84423c83db401fc62ea13877111f2

Not much to say i am affraid as everything is under control of the device
driver (when hotplug/hotremove happens, memory management, userspace API,
...).

>
> [...]
> > + mem_hotplug_begin();
> > + ret = add_pages(nid, align_start >> PAGE_SHIFT,
> > + align_size >> PAGE_SHIFT, false);
> > + if (ret) {
> > + mem_hotplug_done();
> > + goto error_add_memory;
> > + }
> > + move_pfn_range_to_zone(&NODE_DATA(nid)->node_zones[ZONE_DEVICE],
> > + align_start >> PAGE_SHIFT,
> > + align_size >> PAGE_SHIFT);
> > + mem_hotplug_done();
> > +
> > + for (pfn = devmem->pfn_first; pfn < devmem->pfn_last; pfn++) {
> > + struct page *page = pfn_to_page(pfn);
> > +
> > + /*
> > + * ZONE_DEVICE pages union ->lru with a ->pgmap back
> > + * pointer. It is a bug if a ZONE_DEVICE page is ever
> > + * freed or placed on a driver-private list. Therefore,
> > + * seed the storage with LIST_POISON* values.
> > + */
> > + list_del(&page->lru);
>
> this? The page is not on any list yet - it hasn't been added to the page
> allocator.

Like comments says it was to init page->lru.next|prev with poison values
it is not important so i remove it.

Jérôme