On 17.12.21 18:16, Nadav Amit wrote:
quoted
On Dec 17, 2021, at 3:30 AM, David Hildenbrand [off-list ref] wrote:
Let's return early for hugetlb, which really only relies on the compound
mapcount so far and does not support PageDoubleMap() yet. Use the chance
to cleanup the file-THP case to make it easier to grasp. While at it, use
head_compound_mapcount().
This is a preparation for further changes.
It would be useful to add “no functional change intended” or something.
Absolutely, same applies to other "simplification" patches.
quoted
Reviewed-by: Peter Xu <peterx@redhat.com>
Signed-off-by: David Hildenbrand <redacted>
---
mm/util.c | 11 +++++++----
1 file changed, 7 insertions(+), 4 deletions(-)
diff --git a/mm/util.c b/mm/util.c
index 741ba32a43ac..3239e75c148d 100644
--- a/mm/util.c
+++ b/mm/util.c
@@ -732,15 +732,18 @@ int __page_mapcount(struct page *page)
{
int ret;
- ret = atomic_read(&page->_mapcount) + 1;
+ if (PageHuge(page))
+ return compound_mapcount(page);
Before you return, perhaps you can add an assertion like:
VM_BUG_ON(PageDoubleMap(page));
This would be make the code clearer and would ease debugging in the
future (if support for double-map is expanded).
I'd probably have to add this to a couple of places -- and I assume
anybody working on that has to grep the kernel for use of PageDoubleMap
already.
Thanks!
--
Thanks,
David / dhildenb