[PATCH] mm: handle zone device pages in release_pages()

[PATCH] mm: handle zone device pages in release_pages()

[Ralph Campbell]

release_pages() is an optimized, inlined version of __put_pages() except
that zone device struct pages that are not page_is_devmap_managed()
(i.e., memory_type MEMORY_DEVICE_GENERIC and MEMORY_DEVICE_PCI_P2PDMA),
fall through to the code that could return the zone device page to the
page allocator instead of adjusting the pgmap reference count.
Clearly these type of pages are not having the reference count decremented
to zero via release_pages() or page allocation problems would be seen.
Just to be safe, handle the 1 to zero case in release_pages() like
__put_page() does.

Signed-off-by: Ralph Campbell <rcampbell@nvidia.com>
---

I found this by code inspection while working on converting ZONE_DEVICE
struct pages to have zero based reference counts. I don't think there is
an actual problem that this fixes, it's more to future proof new uses of
release_pages().

This is for Andrew Morton's mm tree after the merge window.

 mm/swap.c | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/mm/swap.c b/mm/swap.c
index 0eb057141a04..106f519c45ac 100644
--- a/mm/swap.c
+++ b/mm/swap.c
@@ -907,6 +907,9 @@ void release_pages(struct page **pages, int nr)
 				put_devmap_managed_page(page);
 				continue;
 			}
+			if (put_page_testzero(page))
+				put_dev_pagemap(page->pgmap);
+			continue;
 		}
 
 		if (!put_page_testzero(page))
-- 
2.20.1

Re: [PATCH] mm: handle zone device pages in release_pages()

[Christoph Hellwig]

Looks correct to me:

Reviewed-by: Christoph Hellwig <hch@lst.de>