On Fri, Nov 01, 2024 at 12:37:45AM -0700, Hugh Dickins wrote: > On Fri, 1 Nov 2024, Shivank Garg wrote: > > > Instead of accessing vma->vm_policy directly, use vma_policy() like > > other places for consistency. > > > > Signed-off-by: Shivank Garg <shivankg@xxxxxxx> > > NAK. The vma_policy(vma) wrapper exists to avoid #ifdef CONFIG_NUMAs > all over the place; there is no point to it inside CONFIG_NUMA source. > > Hugh > In that case we should do the opposite of this patch inside mempolicy.c Only 1 place to update ~Gregory diff --git a/mm/mempolicy.c b/mm/mempolicy.c index b858e22b259d..beffdc17aad5 100644 --- a/mm/mempolicy.c +++ b/mm/mempolicy.c @@ -1572,7 +1572,7 @@ SYSCALL_DEFINE4(set_mempolicy_home_node, unsigned long, start, unsigned long, le * or MPOL_PREFERRED_MANY we return error. We don't reset * the home node for vmas we already updated before. */ - old = vma_policy(vma); + old = vma->vm_policy; if (!old) { prev = vma; continue;