OSDN Git Service

mm: Move change_prot_numa outside CONFIG_ARCH_USES_NUMA_PROT_NONE
authorAneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>
Thu, 5 Dec 2013 18:38:22 +0000 (00:08 +0530)
committerBenjamin Herrenschmidt <benh@kernel.crashing.org>
Mon, 9 Dec 2013 00:40:23 +0000 (11:40 +1100)
change_prot_numa should work even if _PAGE_NUMA != _PAGE_PROTNONE.
On archs like ppc64 that don't use _PAGE_PROTNONE and also have
a separate page table outside linux pagetable, we just need to
make sure that when calling change_prot_numa we flush the
hardware page table entry so that next page access  result in a numa
fault.

We still need to make sure we use the numa faulting logic only
when CONFIG_NUMA_BALANCING is set. This implies the migrate-on-fault
(Lazy migration) via mbind will only work if CONFIG_NUMA_BALANCING
is set.

Signed-off-by: Aneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>
Reviewed-by: Rik van Riel <riel@redhat.com>
Acked-by: Mel Gorman <mgorman@suse.de>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
include/linux/mm.h
mm/mempolicy.c

index 1cedd00..a7b4e31 100644 (file)
@@ -1842,7 +1842,7 @@ static inline pgprot_t vm_get_page_prot(unsigned long vm_flags)
 }
 #endif
 
-#ifdef CONFIG_ARCH_USES_NUMA_PROT_NONE
+#ifdef CONFIG_NUMA_BALANCING
 unsigned long change_prot_numa(struct vm_area_struct *vma,
                        unsigned long start, unsigned long end);
 #endif
index eca4a31..9f73b29 100644 (file)
@@ -613,7 +613,7 @@ static inline int queue_pages_pgd_range(struct vm_area_struct *vma,
        return 0;
 }
 
-#ifdef CONFIG_ARCH_USES_NUMA_PROT_NONE
+#ifdef CONFIG_NUMA_BALANCING
 /*
  * This is used to mark a range of virtual addresses to be inaccessible.
  * These are later cleared by a NUMA hinting fault. Depending on these
@@ -627,7 +627,6 @@ unsigned long change_prot_numa(struct vm_area_struct *vma,
                        unsigned long addr, unsigned long end)
 {
        int nr_updated;
-       BUILD_BUG_ON(_PAGE_NUMA != _PAGE_PROTNONE);
 
        nr_updated = change_protection(vma, addr, end, vma->vm_page_prot, 0, 1);
        if (nr_updated)
@@ -641,7 +640,7 @@ static unsigned long change_prot_numa(struct vm_area_struct *vma,
 {
        return 0;
 }
-#endif /* CONFIG_ARCH_USES_NUMA_PROT_NONE */
+#endif /* CONFIG_NUMA_BALANCING */
 
 /*
  * Walk through page tables and collect pages to be migrated.