Re: [PATCH net-next] page_pool: Rename frag_users to frag_cnt

[Ilias Apalodimas]

Hi Yunsheng,

On Fri, 15 Dec 2023 at 14:34, Yunsheng Lin <linyunsheng@xxxxxxxxxx> wrote:
>
> On 2023/12/15 19:58, Ilias Apalodimas wrote:
> > Hi Yunsheng,
> >
> > On Fri, 15 Dec 2023 at 13:10, Yunsheng Lin <linyunsheng@xxxxxxxxxx> wrote:
> >>
> >> On 2023/12/15 15:31, Ilias Apalodimas wrote:
> >>> Since [0] got merged, it's clear that 'pp_ref_count' is used to track
> >>> the number of users for each page. On struct_page though we have
> >>> a member called 'frag_users'. Despite of what the name suggests this is
> >>> not the number of users. It instead represents the number of fragments of
> >>> the current page. When we have a single page this is set to one. When we
> >>> split the page this is set to the actual number of frags and later used
> >>> in page_pool_drain_frag() to infer the real number of users.
> >>>
> >>> So let's rename it to something that matches the description above
> >>>
> >>> [0]
> >>> Link: https://lore.kernel.org/netdev/20231212044614.42733-2-liangchen.linux@xxxxxxxxx/
> >>> Signed-off-by: Ilias Apalodimas <ilias.apalodimas@xxxxxxxxxx>
> >>> ---
> >>>  include/net/page_pool.h | 2 +-
> >>>  net/core/page_pool.c    | 8 ++++----
> >>>  2 files changed, 5 insertions(+), 5 deletions(-)
> >>>
> >>> diff --git a/include/net/page_pool.h b/include/net/page_pool.h
> >>> index 813c93499f20..957cd84bb3f4 100644
> >>> --- a/include/net/page_pool.h
> >>> +++ b/include/net/page_pool.h
> >>> @@ -158,7 +158,7 @@ struct page_pool {
> >>>       u32 pages_state_hold_cnt;
> >>>       unsigned int frag_offset;
> >>>       struct page *frag_page;
> >>> -     long frag_users;
> >>> +     long frag_cnt;
> >>
> >> I would rename it to something like refcnt_bais to mirror the pagecnt_bias
> >> in struct page_frag_cache.
> >
> > Sure
> >
> >>
> >>>
> >>>  #ifdef CONFIG_PAGE_POOL_STATS
> >>>       /* these stats are incremented while in softirq context */
> >>> diff --git a/net/core/page_pool.c b/net/core/page_pool.c
> >>> index 9b203d8660e4..19a56a52ac8f 100644
> >>> --- a/net/core/page_pool.c
> >>> +++ b/net/core/page_pool.c
> >>> @@ -659,7 +659,7 @@ EXPORT_SYMBOL(page_pool_put_page_bulk);
> >>>  static struct page *page_pool_drain_frag(struct page_pool *pool,
> >>>                                        struct page *page)
> >>>  {
> >>> -     long drain_count = BIAS_MAX - pool->frag_users;
> >>> +     long drain_count = BIAS_MAX - pool->frag_cnt;
> >>
> >> drain_count = pool->refcnt_bais;
> >
> > I think this is a typo right? This still remains
>
> It would be better to invert logic too, as it is mirroring:
>
> https://elixir.bootlin.com/linux/v6.7-rc5/source/mm/page_alloc.c#L4745

This is still a bit confusing for me since the actual bias is the
number of fragments that you initially split the page. But I am fine
with having a common approach. I'll send the rename again shortly, and
I can send the logic invert a bit later (or feel free to send it,
since it was your idea).

Thanks
/Ilias