On Mon, Aug 19, 2019 at 11:46:37AM +0000, Vlastimil Babka wrote:
>
> On 8/16/19 4:04 PM, Kirill A. Shutemov wrote:
> > On Fri, Aug 16, 2019 at 12:13:59PM +0200, Vlastimil Babka wrote:
> >> Currently, page owner info is only recorded for the first page of a high-order
> >> allocation, and copied to tail pages in the event of a split page. With the
> >> plan to keep previous owner info after freeing the page, it would be benefical
> >> to record page owner for each subpage upon allocation. This increases the
> >> overhead for high orders, but that should be acceptable for a debugging option.
> >>
> >> The order stored for each subpage is the order of the whole allocation. This
> >> makes it possible to calculate the "head" pfn and to recognize "tail" pages
> >> (quoted because not all high-order allocations are compound pages with true
> >> head and tail pages). When reading the page_owner debugfs file, keep skipping
> >> the "tail" pages so that stats gathered by existing scripts don't get inflated.
> >>
> >> Signed-off-by: Vlastimil Babka <vbabka@xxxxxxx>
> >
> > Hm. That's all reasonable, but I have a question: do you see how page
> > owner thing works for THP now?
> >
> > I don't see anything in split_huge_page() path (do not confuse it with
> > split_page() path) that would copy the information to tail pages. Do you?
>
> You're right, it's missing. This patch fixes that and can be added e.g.
> at the end of the series.
I would rather put it the first. Possbily with stable@.
> ----8<----
> From 56ac1b41559eecf52a2d453c49ce66dbbb227c64 Mon Sep 17 00:00:00 2001
> From: Vlastimil Babka <vbabka@xxxxxxx>
> Date: Mon, 19 Aug 2019 13:38:29 +0200
> Subject: [PATCH] mm, page_owner: handle THP splits correctly
>
> THP splitting path is missing the split_page_owner() call that split_page()
> has. As a result, split THP pages are wrongly reported in the page_owner file
> as order-9 pages. Furthermore when the former head page is freed, the remaining
> former tail pages are not listed in the page_owner file at all. This patch
> fixes that by adding the split_page_owner() call into __split_huge_page().
>
> Reported-by: Kirill A. Shutemov <kirill@xxxxxxxxxxxxx>
> Signed-off-by: Vlastimil Babka <vbabka@xxxxxxx>
> ---
> mm/huge_memory.c | 3 +++
> 1 file changed, 3 insertions(+)
>
> diff --git a/mm/huge_memory.c b/mm/huge_memory.c
> index 738065f765ab..d727a0401484 100644
> --- a/mm/huge_memory.c
> +++ b/mm/huge_memory.c
> @@ -32,6 +32,7 @@
> #include <linux/shmem_fs.h>
> #include <linux/oom.h>
> #include <linux/numa.h>
> +#include <linux/page_owner.h>
>
> #include <asm/tlb.h>
> #include <asm/pgalloc.h>
> @@ -2533,6 +2534,8 @@ static void __split_huge_page(struct page *page, struct list_head *list,
>
> remap_page(head);
>
> + split_page_owner(head, HPAGE_PMD_ORDER);
> +
I think it has to be before remap_page(). This way nobody would be able to
mess with the page until it has valid page_owner.
> for (i = 0; i < HPAGE_PMD_NR; i++) {
> struct page *subpage = head + i;
> if (subpage == page)
> --
> 2.22.0
>
--
Kirill A. Shutemov
