Re: [PATCH mm-unstable v9 14/31] s390: Convert various pgalloc functions to use ptdescs

On Thu, Oct 12, 2023 at 12:25 AM Heiko Carstens <[email protected]> wrote:
>
> On Mon, Aug 07, 2023 at 04:04:56PM -0700, Vishal Moola (Oracle) wrote:
> > As part of the conversions to replace pgtable constructor/destructors with
> > ptdesc equivalents, convert various page table functions to use ptdescs.
> >
> > Some of the functions use the *get*page*() helper functions. Convert
> > these to use pagetable_alloc() and ptdesc_address() instead to help
> > standardize page tables further.
> >
> > Acked-by: Mike Rapoport (IBM) <[email protected]>
> > Signed-off-by: Vishal Moola (Oracle) <[email protected]>
> > ---
> >  arch/s390/include/asm/pgalloc.h |   4 +-
> >  arch/s390/include/asm/tlb.h     |   4 +-
> >  arch/s390/mm/pgalloc.c          | 128 ++++++++++++++++----------------
> >  3 files changed, 69 insertions(+), 67 deletions(-)
> ...
> > diff --git a/arch/s390/mm/pgalloc.c b/arch/s390/mm/pgalloc.c
> > index d7374add7820..07fc660a24aa 100644
> > --- a/arch/s390/mm/pgalloc.c
> > +++ b/arch/s390/mm/pgalloc.c
> ...
> > @@ -488,16 +486,20 @@ static void base_pgt_free(unsigned long *table)
> >  static unsigned long *base_crst_alloc(unsigned long val)
> >  {
> >       unsigned long *table;
> > +     struct ptdesc *ptdesc;
> >
> > -     table = (unsigned long *)__get_free_pages(GFP_KERNEL, CRST_ALLOC_ORDER);
> > -     if (table)
> > -             crst_table_init(table, val);
> > +     ptdesc = pagetable_alloc(GFP_KERNEL & ~__GFP_HIGHMEM, CRST_ALLOC_ORDER);
>
> I guess I must miss something, but what is the reason to mask out
> __GFP_HIGHMEM here? It is not part of GFP_KERNEL, nor does s390 support
> HIGHMEM.

You're not missing anything.

This was replacing __get_free_pages() which also doesn't support HIGHMEM,
so I had that in to ensure a non-HIGHMEM allocation in case a
passed-in gfp_flags
had it set. In hindsight since we're just passing in the GFP flags
directly here, we don't
actually need to mask out GFP_HIGHMEM.