From: Janosch Frank <frankja@xxxxxxxxxxxxx> commit 528a9539348a0234375dfaa1ca5dbbb2f8f8e8d2 upstream. If the pmd is soft dirty we must mark the pte as soft dirty (and not dirty). This fixes some cases for guest migration with huge page backings. Cc: <stable@xxxxxxxxxxxxxxx> # 4.8 Fixes: bc29b7ac1d9f ("s390/mm: clean up pte/pmd encoding") Reviewed-by: Christian Borntraeger <borntraeger@xxxxxxxxxx> Reviewed-by: Gerald Schaefer <gerald.schaefer@xxxxxxxxxx> Signed-off-by: Janosch Frank <frankja@xxxxxxxxxxxxx> Signed-off-by: Heiko Carstens <hca@xxxxxxxxxxxxx> Signed-off-by: Greg Kroah-Hartman <gregkh@xxxxxxxxxxxxxxxxxxx> --- arch/s390/mm/hugetlbpage.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) --- a/arch/s390/mm/hugetlbpage.c +++ b/arch/s390/mm/hugetlbpage.c @@ -117,7 +117,7 @@ static inline pte_t __rste_to_pte(unsign _PAGE_YOUNG); #ifdef CONFIG_MEM_SOFT_DIRTY pte_val(pte) |= move_set_bit(rste, _SEGMENT_ENTRY_SOFT_DIRTY, - _PAGE_DIRTY); + _PAGE_SOFT_DIRTY); #endif pte_val(pte) |= move_set_bit(rste, _SEGMENT_ENTRY_NOEXEC, _PAGE_NOEXEC);