Re: [PATCH 6/7] mm: convert generic code to 5-level paging

[Michal Hocko]

On Wed 08-03-17 18:21:30, Kirill A. Shutemov wrote:
> On Wed, Mar 08, 2017 at 02:57:35PM +0100, Michal Hocko wrote:
> > On Mon 06-03-17 23:45:13, Kirill A. Shutemov wrote:
> > > Convert all non-architecture-specific code to 5-level paging.
> > > 
> > > It's mostly mechanical adding handling one more page table level in
> > > places where we deal with pud_t.
> > > 
> > > Signed-off-by: Kirill A. Shutemov <kirill.shutemov@xxxxxxxxxxxxxxx>
> > 
> > OK, I haven't spotted anything major. I am just scratching my head about
> > the __ARCH_HAS_5LEVEL_HACK leak into kasan_init.c (see below). Why do we
> > need it?  It looks more than ugly but I am not familiar with kasan so
> > maybe this is really necessary.
> 
> Yeah ugly.
> 
> kasan_zero_p4d is only defined if we have real page table level. It's okay
> if the page table level is folded properly -- using pgtable-nop4d.h -- in
> this case pgd_populate() is nop and we don't reference kasan_zero_p4d.
> 
> With 5level-fixup.h, pgd_populate() is not nop, so we would reference
> kasan_zero_p4d and build breaks. We don't need this as p4d_populate()
> would do what we really need in this case.
> 
> We can drop the hack once all architectures that support kasan would be
> converted to pgtable-nop4d.h -- amd64 and x86 at the moment.

But those architectures even do not enable kasan
$ git grep "select *HAVE_ARCH_KASAN"
arch/arm64/Kconfig:     select HAVE_ARCH_KASAN if SPARSEMEM_VMEMMAP && !(ARM64_16K_PAGES && ARM64_VA_BITS_48)
arch/x86/Kconfig:       select HAVE_ARCH_KASAN                  if X86_64 && SPARSEMEM_VMEMMAP

both arm64 and x86 (64b) do compile fine without the ifdef... So I guess
we should be fine without it.
-- 
Michal Hocko
SUSE Labs