mm: fix missing wake-up event for FSDAX pages

FSDAX page refcounts are 1-based, rather than 0-based: if refcount is
1, then the page is freed.  The FSDAX pages can be pinned through GUP,
then they will be unpinned via unpin_user_page() using a folio variant
to put the page, however, folio variants did not consider this special
case, the result will be to miss a wakeup event (like the user of
__fuse_dax_break_layouts()).  This results in a task being permanently
stuck in TASK_INTERRUPTIBLE state.

Since FSDAX pages are only possibly obtained by GUP users, so fix GUP
instead of folio_put() to lower overhead.

Link: https://lkml.kernel.org/r/[email protected]
Fixes: d8ddc09 ("mm/gup: Add gup_put_folio()")
Signed-off-by: Muchun Song <[email protected]>
Suggested-by: Matthew Wilcox <[email protected]>
Cc: Jason Gunthorpe <[email protected]>
Cc: John Hubbard <[email protected]>
Cc: William Kucharski <[email protected]>
Cc: Dan Williams <[email protected]>
Cc: Jan Kara <[email protected]>
Cc: <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>

include/linux/mm.h

@@ -1130,23 +1130,27 @@ static inline bool is_zone_movable_page(const struct page *page)
 #if defined(CONFIG_ZONE_DEVICE) && defined(CONFIG_FS_DAX)
 DECLARE_STATIC_KEY_FALSE(devmap_managed_key);
 
-bool __put_devmap_managed_page(struct page *page);
-static inline bool put_devmap_managed_page(struct page *page)
+bool __put_devmap_managed_page_refs(struct page *page, int refs);
+static inline bool put_devmap_managed_page_refs(struct page *page, int refs)
 {
 	if (!static_branch_unlikely(&devmap_managed_key))
 		return false;
 	if (!is_zone_device_page(page))
 		return false;
-	return __put_devmap_managed_page(page);
+	return __put_devmap_managed_page_refs(page, refs);
 }
-
 #else /* CONFIG_ZONE_DEVICE && CONFIG_FS_DAX */
-static inline bool put_devmap_managed_page(struct page *page)
+static inline bool put_devmap_managed_page_refs(struct page *page, int refs)
 {
 	return false;
 }
 #endif /* CONFIG_ZONE_DEVICE && CONFIG_FS_DAX */
 
+static inline bool put_devmap_managed_page(struct page *page)
+{
+	return put_devmap_managed_page_refs(page, 1);
+}
+
 /* 127: arbitrary random number, small enough to assemble well */
 #define folio_ref_zero_or_close_to_overflow(folio) \
 	((unsigned int) folio_ref_count(folio) + 127u <= 127u)

mm/gup.c

@@ -87,7 +87,8 @@ static inline struct folio *try_get_folio(struct page *page, int refs)
 	 * belongs to this folio.
 	 */
 	if (unlikely(page_folio(page) != folio)) {
-		folio_put_refs(folio, refs);
+		if (!put_devmap_managed_page_refs(&folio->page, refs))
+			folio_put_refs(folio, refs);
 		goto retry;
 	}
 
@@ -176,7 +177,8 @@ static void gup_put_folio(struct folio *folio, int refs, unsigned int flags)
 			refs *= GUP_PIN_COUNTING_BIAS;
 	}
 
-	folio_put_refs(folio, refs);
+	if (!put_devmap_managed_page_refs(&folio->page, refs))
+		folio_put_refs(folio, refs);
 }
 
 /**

mm/memremap.c

@@ -499,7 +499,7 @@ void free_zone_device_page(struct page *page)
 }
 
 #ifdef CONFIG_FS_DAX
-bool __put_devmap_managed_page(struct page *page)
+bool __put_devmap_managed_page_refs(struct page *page, int refs)
 {
 	if (page->pgmap->type != MEMORY_DEVICE_FS_DAX)
 		return false;
@@ -509,9 +509,9 @@ bool __put_devmap_managed_page(struct page *page)
 	 * refcount is 1, then the page is free and the refcount is
 	 * stable because nobody holds a reference on the page.
 	 */
-	if (page_ref_dec_return(page) == 1)
+	if (page_ref_sub_return(page, refs) == 1)
 		wake_up_var(&page->_refcount);
 	return true;
 }
-EXPORT_SYMBOL(__put_devmap_managed_page);
+EXPORT_SYMBOL(__put_devmap_managed_page_refs);
 #endif /* CONFIG_FS_DAX */