Re: [RFC PATCH net-next v1 4/4] net: page_pool: use netmem_t instead of struct page in API

On 2023/12/22 5:22, Mina Almasry wrote:

On Thu, Dec 21, 2023 at 3:32 AM Yunsheng Lin [off-list ref] wrote:

quoted

On 2023/12/20 11:01, Mina Almasry wrote:

...

quoted

quoted

quoted

quoted

Perhaps we should aim to not export netmem_to_page(),
prevent modules from accessing it directly.

+1.

I looked into this, but it turns out it's a slightly bigger change
that needs some refactoring to make it work. There are few places
where I believe I need to add netmem_to_page() that are exposed to the
drivers via inline helpers, these are:

- skb_frag_page(), which returns NULL if the netmem is not a page, but
needs to do a netmem_to_page() to return the page otherwise.

Is it possible to introduce something like skb_frag_netmem() for
netmem? so that we can keep most existing users of skb_frag_page()
unchanged and avoid adding additional checking overhead for existing
users.

In my experience most current skb_frag_page() users need specifically
the struct page*. Example is illegal_highdma() which
PageHighMem(skb_frag_page())

For illegal_highdma() case, is it possible to use something like
skb_readabe_frag() checking to avoid calling skb_frag_page() for netmem?

But RFC v5 adds skb_frag_netmem() for callsites that want a netmem and
don't care about specifically a page:

https://patchwork.kernel.org/project/netdevbpf/patch/20231218024024.3516870-10-almasrymina@google.com/

quoted

quoted

- The helpers inside skb_add_rx_frag(), which needs to do a
netmem_to_page() to set skb->pfmemalloc.

Similar as above, perhaps introduce something like skb_add_rx_netmem_frag()?

Yes, v3 of this series adds skb_add_rx_frag_netmem():

https://patchwork.kernel.org/project/netdevbpf/patch/20231220214505.2303297-4-almasrymina@google.com/