Re: [PATCH v4 2/2] mm: memory_hotplug: make hugetlb_optimize_vmemmap compatible with memmap_on_memory

[Muchun Song]

On Mon, Jun 20, 2022 at 10:44:40AM +0200, Oscar Salvador wrote:
> On Mon, Jun 20, 2022 at 04:29:11PM +0800, Muchun Song wrote:
> > > > Although it works, I think PageVmemmapSelfHosted() check for the 1st pfn's
> > > > vmemmap page is not always reliable.  Since we reused PG_owner_priv_1
> > > > as PG_vmemmap_self_hosted, the test is noly reliable for vmemmap page's
> > > > vmemmap page.  Other non-vmemmap page can be flagged with PG_owner_priv_1.
> > > > So this check can be false-positive. Maybe the following code snippet is
> > > > the solution.
> > > 
> > > How could that happen for pages used for backing a vmemmap?
> > >
> > 
> > It cannot happen for memmap_on_memory case. Howwver, it can happen for other
> > cases. E.g. the 1st pfn (of boot memory block) whose vmemmap page may be flagged
> > as PG_owner_priv_1 (if PG_swapcache is set). Then, the check is false-positive.
> 
> If this can really happen, which I am not that sure tbh, maybe a way out would be

I need to clarify this only can be happened by using this approach implemented
in this patch.

For a boot memory block, the vemmmap pages are not slef-hosted.  So the 1st pfn (of
this memory block) can be allocated to other users. e.g. an anonymous page with
PG_swapcache set.  In this patch, ALIGN_DOWN(pfn, PHYS_PFN(memory_block_size_bytes()))
will located on this anonymous page, then the check is false-positive.

[                  boot memory block                  ]
[        section        ][...][        section        ]
[                   usable memory                     ]

> to just define a new page-type as we did in previous versions of memmap_on_memory.
> In that way we would not for flags, but for its type.
>

I think we do not need to introduced a new flag, we just make sure the page
passed to PageVmemmapSelfHosted() is a backing page for vmemmap. Then we
cannot incur false-positive.  The feasible solution is walking page tables
to find a vmemmap page's backing page.

Thanks.
 
> But as I said, I am not entirely sure about the potential fallout of what you mention.
> 
> 
> -- 
> Oscar Salvador
> SUSE Labs
>