Re: [External] Re: [PATCH 2/5] hugetlb: convert page_huge_active() to HP_Migratable flag

[Muchun Song]

On Sat, Jan 16, 2021 at 12:26 PM Matthew Wilcox <willy@xxxxxxxxxxxxx> wrote:
>
> On Fri, Jan 15, 2021 at 04:31:02PM -0800, Mike Kravetz wrote:
> > +++ b/fs/hugetlbfs/inode.c
> > @@ -735,7 +735,7 @@ static long hugetlbfs_fallocate(struct file *file, int mode, loff_t offset,
> >
> >               mutex_unlock(&hugetlb_fault_mutex_table[hash]);
> >
> > -             set_page_huge_active(page);
> > +             hugetlb_set_page_flag(page, HP_Migratable);
>
> I had understood the request to be more like ...
>
>                 SetHPageMigratable(page);
>
> > +++ b/include/linux/hugetlb.h
> > @@ -480,9 +480,13 @@ unsigned long hugetlb_get_unmapped_area(struct file *file, unsigned long addr,
> >   * HP_Restore_Reserve - Set when a hugetlb page consumes a reservation at
> >   *   allocation time.  Cleared when page is fully instantiated.  Free
> >   *   routine checks flag to restore a reservation on error paths.
> > + * HP_Migratable - Set after a newly allocated page is added to the page
> > + *   cache and/or page tables.  Indicates the page is a candidate for
> > + *   migration.
> >   */
> >  enum hugetlb_page_flags {
> >       HP_Restore_Reserve = 0,
> > +     HP_Migratable,
> >  };
>
> and name these HPG_restore_reserve and HPG_migratable
>
> and generate the calls to hugetlb_set_page_flag etc from macros, eg:
>
> #define TESTHPAGEFLAG(uname, lname)                                     \
> static __always_inline bool HPage##uname(struct page *page)             \
> { return test_bit(HPG_##lname, &page->private); }
> ...
> #define HPAGEFLAG(uname, lname)                                         \
>         TESTHPAGEFLAG(uname, lname)                                     \
>         SETHPAGEFLAG(uname, lname)                                      \
>         CLEARHPAGEFLAG(uname, lname)
>
> HPAGEFLAG(RestoreReserve, restore_reserve)
> HPAGEFLAG(Migratable, migratable)
>
> just to mirror page-flags.h more closely.

I prefer this suggestion. I also made the same suggestion in the
previous RFC version.