On Thu, 11 Jul 2024 20:30:07 -0700 (PDT) Hugh Dickins <hughd@xxxxxxxxxx> wrote:
> In looking through a "git log -p --reverse v6.10-rc6..HEAD"
> of mm-everything, I just noticed a small anomaly in mm-stable:
> ac90c56bbd73 "mm/ksm: refactor out try_to_merge_with_zero_page()"
> modifies mm/hugetlb.c! undoing a oneliner which had been folded-merged into
> 003af997c8a9 "hugetlb: force allocating surplus hugepages on mempolicy allowed nodes"
> which comes just before it.
>
Huh thanks. This cycle has been pretty bad with new hotfixes coming in
and messing up mm-stable.
--- a/mm/hugetlb.c~mm-hugetlbc-undo-errant-change
+++ a/mm/hugetlb.c
@@ -2634,6 +2634,7 @@ static int gather_surplus_pages(struct h
retry:
spin_unlock_irq(&hugetlb_lock);
for (i = 0; i < needed; i++) {
+ folio = NULL;
for_each_node_mask(node, cpuset_current_mems_allowed) {
if (!mbind_nodemask || node_isset(node, *mbind_nodemask)) {
folio = alloc_surplus_hugetlb_folio(h, htlb_alloc_mask(h),
_
