mm: sigbus instead of abusing oom
authorHugh Dickins <>
Tue, 15 Dec 2009 01:59:04 +0000 (17:59 -0800)
committerLinus Torvalds <>
Tue, 15 Dec 2009 16:53:17 +0000 (08:53 -0800)
When do_nonlinear_fault() realizes that the page table must have been
corrupted for it to have been called, it does print_bad_pte() and returns
...  VM_FAULT_OOM, which is hard to understand.

It made some sense when I did it for 2.6.15, when do_page_fault() just
killed the current process; but nowadays it lets the OOM killer decide who
to kill - so page table corruption in one process would be liable to kill

Change it to return VM_FAULT_SIGBUS instead: that doesn't guarantee that
the process will be killed, but is good enough for such a rare
abnormality, accompanied as it is by the "BUG: Bad page map" message.

And recent HWPOISON work has copied that code into do_swap_page(), when it
finds an impossible swap entry: fix that to VM_FAULT_SIGBUS too.

Signed-off-by: Hugh Dickins <>
Cc: Izik Eidus <>
Cc: Andrea Arcangeli <>
Cc: Nick Piggin <>
Reviewed-by: KOSAKI Motohiro <>
Cc: Rik van Riel <>
Cc: Lee Schermerhorn <>
Cc: Andi Kleen <>
Reviewed-by: KAMEZAWA Hiroyuki <>
Reviewed-by: Wu Fengguang <>
Reviewed-by: Minchan Kim <>
Signed-off-by: Andrew Morton <>
Signed-off-by: Linus Torvalds <>

index 543c446..1c9dc46 100644 (file)
@@ -2527,7 +2527,7 @@ static int do_swap_page(struct mm_struct *mm, struct vm_area_struct *vma,
                        ret = VM_FAULT_HWPOISON;
                } else {
                        print_bad_pte(vma, address, orig_pte, NULL);
-                       ret = VM_FAULT_OOM;
+                       ret = VM_FAULT_SIGBUS;
                goto out;
@@ -2923,7 +2923,7 @@ static int do_nonlinear_fault(struct mm_struct *mm, struct vm_area_struct *vma,
                 * Page table corrupted: show pte and kill process.
                print_bad_pte(vma, address, orig_pte, NULL);
-               return VM_FAULT_OOM;
+               return VM_FAULT_SIGBUS;
        pgoff = pte_to_pgoff(orig_pte);