On Mon, 20 Jun 2022 14:51:18 +0100 Matthew Wilcox <willy@xxxxxxxxxxxxx> wrote: > On Mon, Jun 20, 2022 at 03:11:31PM +0800, Muchun Song wrote: > > Thanks for your report. It is fixed in thread [1]. > > > > https://lore.kernel.org/all/20220619133851.68184-3-songmuchun@xxxxxxxxxxxxx/ [1] > > No, it's a different problem. I suggest dropping/reverting > > https://lore.kernel.org/linux-mm/20220617175020.717127-20-willy@xxxxxxxxxxxxx/ Dropping that creates awkwardness. How about just uninlining it? --- a/include/linux/mm.h~mm-convert-destroy_compound_page-to-destroy_large_folio-fix +++ a/include/linux/mm.h @@ -923,13 +923,7 @@ static inline void set_compound_page_dto page[1].compound_dtor = compound_dtor; } -static inline void destroy_large_folio(struct folio *folio) -{ - enum compound_dtor_id dtor = folio_page(folio, 1)->compound_dtor; - - VM_BUG_ON_FOLIO(dtor >= NR_COMPOUND_DTORS, folio); - compound_page_dtors[dtor](&folio->page); -} +void destroy_large_folio(struct folio *folio); static inline int head_compound_pincount(struct page *head) { --- a/mm/page_alloc.c~mm-convert-destroy_compound_page-to-destroy_large_folio-fix +++ a/mm/page_alloc.c @@ -822,6 +822,14 @@ void prep_compound_page(struct page *pag prep_compound_head(page, order); } +void destroy_large_folio(struct folio *folio) +{ + enum compound_dtor_id dtor = folio_page(folio, 1)->compound_dtor; + + VM_BUG_ON_FOLIO(dtor >= NR_COMPOUND_DTORS, folio); + compound_page_dtors[dtor](&folio->page); +} + #ifdef CONFIG_DEBUG_PAGEALLOC unsigned int _debug_guardpage_minorder; _
