On Fri, Jun 7, 2024 at 1:37 PM ran xiaokai <ranxiaokai627@xxxxxxx> wrote:
>
> > > From: Ran Xiaokai <ran.xiaokai@xxxxxxxxxx>
> > >
> > > When I did a large folios split test, a WARNING
> > > "[ 5059.122759][ T166] Cannot split file folio to non-0 order"
> > > was triggered. But the test cases are only for anonmous folios.
> > > while mapping_large_folio_support() is only reasonable for page
> > > cache folios.
> > >
> > > In split_huge_page_to_list_to_order(), the folio passed to
> > > mapping_large_folio_support() maybe anonmous folio. The
> > > folio_test_anon() check is missing. So the split of the anonmous THP
> > > is failed. This is also the same for shmem_mapping(). We'd better add
> > > a check for both. But the shmem_mapping() in __split_huge_page() is
> > > not involved, as for anonmous folios, the end parameter is set to -1, so
> > > (head[i].index >= end) is always false. shmem_mapping() is not called.
> > >
> > > Also add a VM_WARN_ON_ONCE() in mapping_large_folio_support()
> > > for anon mapping, So we can detect the wrong use more easily.
> > >
> > > THP folios maybe exist in the pagecache even the file system doesn't
> > > support large folio, it is because when CONFIG_TRANSPARENT_HUGEPAGE
> > > is enabled, khugepaged will try to collapse read-only file-backed pages
> > > to THP. But the mapping does not actually support multi order
> > > large folios properly.
> > >
> > > Using /sys/kernel/debug/split_huge_pages to verify this, with this
> > > patch, large anon THP is successfully split and the warning is ceased.
> > >
> > > Signed-off-by: Ran Xiaokai <ran.xiaokai@xxxxxxxxxx>
> > > ---
> > > include/linux/pagemap.h | 4 ++++
> > > mm/huge_memory.c | 27 ++++++++++++++++-----------
> > > 2 files changed, 20 insertions(+), 11 deletions(-)
> > >
> > > diff --git a/include/linux/pagemap.h b/include/linux/pagemap.h
> > > index ee633712bba0..59f1df0cde5a 100644
> > > --- a/include/linux/pagemap.h
> > > +++ b/include/linux/pagemap.h
> > > @@ -381,6 +381,10 @@ static inline void mapping_set_large_folios(struct address_space *mapping)
> > > */
> > > static inline bool mapping_large_folio_support(struct address_space *mapping)
> > > {
> > > + /* AS_LARGE_FOLIO_SUPPORT is only reasonable for pagecache folios */
> > > + VM_WARN_ONCE((unsigned long)mapping & PAGE_MAPPING_ANON,
> > > + "Anonymous mapping always supports large folio");
> > > +
> > > return IS_ENABLED(CONFIG_TRANSPARENT_HUGEPAGE) &&
> > > test_bit(AS_LARGE_FOLIO_SUPPORT, &mapping->flags);
> > > }
> > > diff --git a/mm/huge_memory.c b/mm/huge_memory.c
> > > index 317de2afd371..62d57270b08e 100644
> > > --- a/mm/huge_memory.c
> > > +++ b/mm/huge_memory.c
> > > @@ -3009,30 +3009,35 @@ int split_huge_page_to_list_to_order(struct page *page, struct list_head *list,
> > > if (new_order >= folio_order(folio))
> > > return -EINVAL;
> > >
> > > - /* Cannot split anonymous THP to order-1 */
> > > - if (new_order == 1 && folio_test_anon(folio)) {
> > > - VM_WARN_ONCE(1, "Cannot split to order-1 folio");
> > > - return -EINVAL;
> > > - }
> > > -
> > > - if (new_order) {
> > > - /* Only swapping a whole PMD-mapped folio is supported */
> > > - if (folio_test_swapcache(folio))
> > > + if (folio_test_anon(folio)) {
> > > + /* Cannot split anonymous THP to order-1 */
> >
> > This is simply what the code is indicating. Shouldn't we phrase
> > it differently to explain "why" but not "how"? for example, anon
> > order-1 mTHP is not supported?
>
> Hi, Barry,
> Good comments, thanks.
> Is "order-1 is not a anonymouns mTHP suitable order." better?
could pick up some words from include/linux/huge_mm.h, particularly
those words regarding "a limitation of the THP implementation".
/*
* Mask of all large folio orders supported for anonymous THP; all orders up to
* and including PMD_ORDER, except order-0 (which is not "huge") and order-1
* (which is a limitation of the THP implementation).
*/
#define THP_ORDERS_ALL_ANON ((BIT(PMD_ORDER + 1) - 1) & ~(BIT(0) | BIT(1)))
perhaps, you can even do
if (order > 0 && !(bit(order) & THP_ORDERS_ALL_ANON))
return -EINVAL;
This is self-commented. Either way is fine.
>
> > Otherwise, it looks good to me.
> >
> > Reviewed-by: Barry Song <baohua@xxxxxxxxxx>
>
Thanks
Barry
