On Thu, Sep 29, 2022 at 6:37 AM Lorenzo Bianconi <lorenzo.bianconi@xxxxxxxxxx> wrote: > > > On 9/28/22 11:07 AM, Jesper Dangaard Brouer wrote: > > > > > > On 28/09/2022 15.54, Marcus Wichelmann wrote: > > >> > > >> I'm working for a cloud hosting provider and we're working on a new > > >> XDP-based networking stack for our VM-Hosts that uses XDP to > > >> accelerate the connectivity of our qemu/KVM VMs to the outside. > > >> > > > > > > Welcome to the community! Sounds like an excellent use-case and > > > opportunity for speeding up the RX packets from physical NIC into the > > > VM. Good to hear someone (again) having this use-case. I've personally > > > > +1 +1. This is somehow what I wanted to achieve in the past. > > > > > not been focused on this use-case lately, mostly because community > > > members that I was interacting with changed jobs, away from cloud > > > hosting companies. Good to have a user back in this area! > > > > > > > > >> For this, we use XDP_REDIRECT to forward packets between the physical > > >> host NIC and the VM tap devices. The main issue we have now is, that > > >> our VM guests have some virtio NIC offloads enabled: rx/tx > > >> checksumming, TSO/GSO, GRO and Scatter-Gather. > > > > > > Supporting RX-checksumming is part of the plans for XDP-hints, although > > > virtio_net is not part of my initial patchset. > > > > Lorenzo and I both had versions of a patch to propagate rx csum > > validation to the VM on xdp redirect. I do not recall a version after > > this one: > > > > https://lore.kernel.org/netdev/cover.1622222367.git.lorenzo@xxxxxxxxxx/ > > > > and I lost of track of what change is needed for it to go in. > > iirc the blocking point was the missing CHECKSUM_COMPLETE support (we support > just CHECKSUM_UNNECESSARY). We can conver it with XDP hw-hints. How about other metadata like gso_type, gso_segs, csum_start and csum_offset etc? This looks like a must for GSO to work even if multi-buffer is supported. Thanks > > Regards, > Lorenzo > > > > > > > > > XDP-redirect with GRO and Scatter-Gather frames are part of the > > > multi-buff effort (Cc Lorenzo), but currently XDP_REDIRECT with > > > multi-buff is disabled (except for cpumap), because the lack of > > > XDP-feature bits, meaning we cannot determine (in kernel) if receiving > > > net_device supports multi-buff (Cc Kumar). > > > > > >> Currently, these offloads (especially TSO/GSO) are incompatible with > > >> XDP_REDIRECT and result in packets being dropped. Because disabling > > >> these offloads in all our customer VMs is not a good option, we're > > >> searching for ways to support these offloads with XDP. > > >> > > > > > > To David Ahern, didn't the kernel recently loosen up on having to > > > disable these offloads for KVM virtio_net? > > > > not that I am aware. Still need tx offloads disabled. > > > > > > This summarizes what I was looking into back in 2020, along with the > > current state of XDP for VM use case: > > > > https://legacy.netdevconf.info/0x14/pub/slides/24/netdev-0x14-XDP-and-the-cloud.pdf > > > > source code is still on github too. > >
|