Re: + mm-revert-thp-make-madv_hugepage-check-for-mm-def_flags.patch added to -mm tree

[Alex Thorlton <athorlton@sgi.com>]

On Wed, Feb 26, 2014 at 05:57:59PM +0100, Peter Zijlstra wrote:
> On Wed, Feb 26, 2014 at 04:31:44PM +0100, Oleg Nesterov wrote:
> Do we want a comment here, explaining why s390 is special again?

Here's what I've got, with everybody's suggestions spun together:

diff --git a/mm/huge_memory.c b/mm/huge_memory.c
index 82166bf..7f01491 100644
--- a/mm/huge_memory.c
+++ b/mm/huge_memory.c
@@ -1968,8 +1968,6 @@ out:
 int hugepage_madvise(struct vm_area_struct *vma,
                     unsigned long *vm_flags, int advice)
 {
-       struct mm_struct *mm = vma->vm_mm;
-
        switch (advice) {
        case MADV_HUGEPAGE:
                /*
@@ -1977,8 +1975,16 @@ int hugepage_madvise(struct vm_area_struct *vma,
                 */
                if (*vm_flags & (VM_HUGEPAGE | VM_NO_THP))
                        return -EINVAL;
-               if (mm->def_flags & VM_NOHUGEPAGE)
+
+/*
+ * MADV_HUGEPAGE after PRCTL_THP_DISABLE is broken on s390 because
+ * qemu blindly does madvise(MADV_HUGEPAGE) after s390_enable_sie().
+ */
+#ifdef CONFIG_S390
+               if (mm_has_pgste(vma->vm_mm))
                        return -EINVAL;
+#endif
+
                *vm_flags &= ~VM_NOHUGEPAGE;
                *vm_flags |= VM_HUGEPAGE;
                /*

I've compiled and tested and it works ok on my machines (I don't have an
s390 to test on though :).  Is everybody okay with this solution?

BTW, Kirill, I looked at using IS_ENABLED to clean up the ifdef, but it
won't work here, since mm_has_pgste isn't defined unless CONFIG_S390
is defined.  i.e.: This line:

if (IS_ENABLED(CONFIG_S390) && mm_has_pgste(vma->vm_mm))

won't compile.