On Tue, Jul 11, 2023 at 9:32 AM Jakub Kicinski <kuba@xxxxxxxxxx> wrote: > > On Tue, 11 Jul 2023 10:20:58 -0600 David Ahern wrote: > > On 7/11/23 10:00 AM, Jakub Kicinski wrote: > > >> RDMA works with the AMD and Intel intree drivers using DMABUF without > > >> requiring struct pages using the DRM hacky scatterlist approach. > > > I see, thanks. We need pages primarily for refcounting. Avoiding all > > > the infamous problems with memory pins. Oh well. > > > > io_uring for example already manages the page pinning. An skb flag was > > added for ZC Tx API to avoid refcounting in the core networking layer. > > Right, we can refcount in similar fashion. Still tracking explicitly > when buffers are handed over to the NIC. > > > Any reason not to allow an alternative representation for skb frags than > > struct page? > > I don't think there's a hard technical reason. We can make it work. I also think we can switch the representation for skb frags to something else. However - please do correct me if I'm wrong - I don't think that is sufficient for device memory TCP. My understanding is that we also need to modify any NIC drivers that want to use device memory TCP to understand a new memory type, and the page pool as well if that's involved. I think in particular modifying the memory type in all the NIC drivers that want to do device memory TCP is difficult. Do you think this is feasible? -- Thanks, Mina
