Re: [External] Re: [PATCH v23 6/9] mm: hugetlb: alloc the vmemmap pages associated with each HugeTLB page

On Mon, May 10, 2021 at 6:45 PM Oscar Salvador [off-list ref] wrote:

On Mon, May 10, 2021 at 11:00:24AM +0800, Muchun Song wrote:

quoted

--- a/mm/hugetlb.c
+++ b/mm/hugetlb.c

@@ -1376,6 +1376,39 @@ static void remove_hugetlb_page(struct hstate *h, struct page *page,
      h->nr_huge_pages_node[nid]--;
 }

+static void add_hugetlb_page(struct hstate *h, struct page *page,
+                          bool adjust_surplus)
+{
+     int zeroed;
+     int nid = page_to_nid(page);
+
+     VM_BUG_ON_PAGE(!HPageVmemmapOptimized(page), page);
+
+     lockdep_assert_held(&hugetlb_lock);
+
+     INIT_LIST_HEAD(&page->lru);
+     h->nr_huge_pages++;
+     h->nr_huge_pages_node[nid]++;
+
+     if (adjust_surplus) {
+             h->surplus_huge_pages++;
+             h->surplus_huge_pages_node[nid]++;
+     }
+
+     set_compound_page_dtor(page, HUGETLB_PAGE_DTOR);
+     set_page_private(page, 0);

I think this has already been discused, so sorry about this.

The only reason to need the set_page_private() is because of the dissolving
function right? add_hugetlb_page() can only get reached via free_huge_page(),
or dissolve_free_huge_page, and while the former clears the flags, the latter
it does not.

Not exactly right. free_huge_page() does not clear HPageTemporary
if the page is temporarily allocated when freeing. You are right that
dissolve_free_huge_page() does not clear HPageFreed.

I think this function would benefit from some renaming. add_hugetlb_page() gives
me no hint of what is this about, although I can figure it out reading the code.

Because I think it is the reverse operation of remove_hugetlb_page,
I named it add_hugetlb_page. Do you have any suggestions for
renaming?

Thanks.

With that: Reviewed-by: Oscar Salvador [off-list ref]




--
Oscar Salvador
SUSE L3