Re: [PATCH v8 1/4] mm: hugetlb_vmemmap: introduce CONFIG_HUGETLB_PAGE_HAS_OPTIMIZE_VMEMMAP

[Muchun Song <songmuchun@bytedance.com>]

On Wed, Apr 20, 2022 at 04:30:02PM -0700, Mike Kravetz wrote:
> On 4/20/22 10:11, Masahiro Yamada wrote:
> > On Wed, Apr 13, 2022 at 11:48 PM Muchun Song <songmuchun@bytedance.com> wrote:
> >>
> >> If the size of "struct page" is not the power of two but with the feature
> >> of minimizing overhead of struct page associated with each HugeTLB is
> >> enabled, then the vmemmap pages of HugeTLB will be corrupted after
> >> remapping (panic is about to happen in theory).  But this only exists when
> >> !CONFIG_MEMCG && !CONFIG_SLUB on x86_64.  However, it is not a conventional
> >> configuration nowadays.  So it is not a real word issue, just the result
> >> of a code review.  But we have to prevent anyone from configuring that
> >> combined configurations.  In order to avoid many checks like "is_power_of_2
> >> (sizeof(struct page))" through mm/hugetlb_vmemmap.c.  Introduce a new macro
> 
> Sorry for jumping in so late.  I am far from expert in Kconfig so did not pay
> much attention to all those discussions.
> 
> Why not just add one (or a few) simple runtime checks for struct page not a
> power of two before enabling hugetlb vmemmap optimization in the code?  Sure,
> it would be ideal to never build/include the vmemmap optimization code if
> this can be detected at config time.  However, it seems this is a very rare
> combination and the checks for it at config time are very complex.

Right. Iterated and explored through 8 versions, I realized checking
it at config time is very complex.
 
> Would we really need many checks throughout the code as you mention above?
> Or, do we only need to check or two before enabling
> hugetlb_optimize_vmemmap_key?

Yep, now there is only one place where needs to check that size.
I think I should go back to v1, it is simpler.

Thanks Mike.