mm/gup: Move page table entry dereference into helper function
authorKirill A. Shutemov <[email protected]>
Thu, 16 Mar 2017 15:26:51 +0000 (18:26 +0300)
committerIngo Molnar <[email protected]>
Sat, 18 Mar 2017 08:48:02 +0000 (09:48 +0100)
This is a preparation patch for the transition of x86 to the generic GUP_fast()
implementation.

On x86 PAE, page table entry is larger than sizeof(long) and we would
need to provide a helper that can read the entry atomically.

Signed-off-by: Kirill A. Shutemov <[email protected]>
Cc: Andrew Morton <[email protected]>
Cc: Aneesh Kumar K . V <[email protected]>
Cc: Borislav Petkov <[email protected]>
Cc: Catalin Marinas <[email protected]>
Cc: Dann Frazier <[email protected]>
Cc: Dave Hansen <[email protected]>
Cc: H. Peter Anvin <[email protected]>
Cc: Linus Torvalds <[email protected]>
Cc: Peter Zijlstra <[email protected]>
Cc: Rik van Riel <[email protected]>
Cc: Steve Capper <[email protected]>
Cc: Thomas Gleixner <[email protected]>
Cc: [email protected]
Cc: [email protected]
Link: http://lkml.kernel.org/r/[email protected]
Signed-off-by: Ingo Molnar <[email protected]>
mm/gup.c

index a62a778ce4ece3bb128031066d2844462af0f38c..e83db38deb17df3bbf0fa3058da540ec4f081a81 100644 (file)
--- a/mm/gup.c
+++ b/mm/gup.c
@@ -1189,6 +1189,17 @@ struct page *get_dump_page(unsigned long addr)
  */
 #ifdef CONFIG_HAVE_GENERIC_RCU_GUP
 
+#ifndef gup_get_pte
+/*
+ * We assume that the PTE can be read atomically. If this is not the case for
+ * your architecture, please provide the helper.
+ */
+static inline pte_t gup_get_pte(pte_t *ptep)
+{
+       return READ_ONCE(*ptep);
+}
+#endif
+
 #ifdef __HAVE_ARCH_PTE_SPECIAL
 static int gup_pte_range(pmd_t pmd, unsigned long addr, unsigned long end,
                         int write, struct page **pages, int *nr)
@@ -1198,14 +1209,7 @@ static int gup_pte_range(pmd_t pmd, unsigned long addr, unsigned long end,
 
        ptem = ptep = pte_offset_map(&pmd, addr);
        do {
-               /*
-                * In the line below we are assuming that the pte can be read
-                * atomically. If this is not the case for your architecture,
-                * please wrap this in a helper function!
-                *
-                * for an example see gup_get_pte in arch/x86/mm/gup.c
-                */
-               pte_t pte = READ_ONCE(*ptep);
+               pte_t pte = gup_get_pte(ptep);
                struct page *head, *page;
 
                /*