Re: [PATCH RFC bpf-next v2 1/5] netlink: specs: Add XDP RX checksum capability to XDP metadata specs

[Stanislav Fomichev <stfomichev@gmail.com>]

On 02/12, Lorenzo Bianconi wrote:
> > On 02/10, Lorenzo Bianconi wrote:
> > > > On 09/26, Jesper Dangaard Brouer wrote:
> > > > > 
> > > > > 
> > > > > On 26/09/2025 06.20, Stanislav Fomichev wrote:
> > > > > > On 09/25, Lorenzo Bianconi wrote:
> > > > > > > Introduce XDP RX checksum capability to XDP metadata specs. XDP RX
> > > > > > > checksum will be use by devices capable of exposing receive checksum
> > > > > > > result via bpf_xdp_metadata_rx_checksum().
> > > > > > > Moreover, introduce xmo_rx_checksum netdev callback in order allow the
> > > > > > > eBPF program bounded to the device to retrieve the RX checksum result
> > > > > > > computed by the hw NIC.
> > > > > > > 
> > > > > > > Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
> > > > > > > ---
> > > > > > >   Documentation/netlink/specs/netdev.yaml |  5 +++++
> > > > > > >   include/net/xdp.h                       | 14 ++++++++++++++
> > > > > > >   net/core/xdp.c                          | 29 +++++++++++++++++++++++++++++
> > > > > > >   3 files changed, 48 insertions(+)
> > > > > > > 
> > > > > > > diff --git a/Documentation/netlink/specs/netdev.yaml b/Documentation/netlink/specs/netdev.yaml
> > > > > > > index e00d3fa1c152d7165e9485d6d383a2cc9cef7cfd..00699bf4a7fdb67c6b9ee3548098b0c933fd39a4 100644
> > > > > > > --- a/Documentation/netlink/specs/netdev.yaml
> > > > > > > +++ b/Documentation/netlink/specs/netdev.yaml
> > > > > > > @@ -61,6 +61,11 @@ definitions:
> > > > > > >           doc: |
> > > > > > >             Device is capable of exposing receive packet VLAN tag via
> > > > > > >             bpf_xdp_metadata_rx_vlan_tag().
> > > > > > > +      -
> > > > > > > +        name: checksum
> > > > > > > +        doc: |
> > > > > > > +          Device is capable of exposing receive checksum result via
> > > > > > > +          bpf_xdp_metadata_rx_checksum().
> > > > > > >     -
> > > > > > >       type: flags
> > > > > > >       name: xsk-flags
> > > > > > > diff --git a/include/net/xdp.h b/include/net/xdp.h
> > > > > > > index aa742f413c358575396530879af4570dc3fc18de..9ab9ac10ae2074b70618a9d4f32544d8b9a30b63 100644
> > > > > > > --- a/include/net/xdp.h
> > > > > > > +++ b/include/net/xdp.h
> > > > > > > @@ -586,6 +586,10 @@ void xdp_attachment_setup(struct xdp_attachment_info *info,
> > > > > > >   			   NETDEV_XDP_RX_METADATA_VLAN_TAG, \
> > > > > > >   			   bpf_xdp_metadata_rx_vlan_tag, \
> > > > > > >   			   xmo_rx_vlan_tag) \
> > > > > > > +	XDP_METADATA_KFUNC(XDP_METADATA_KFUNC_RX_CHECKSUM, \
> > > > > > > +			   NETDEV_XDP_RX_METADATA_CHECKSUM, \
> > > > > > > +			   bpf_xdp_metadata_rx_checksum, \
> > > > > > > +			   xmo_rx_checksum)
> > > > > > >   enum xdp_rx_metadata {
> > > > > > >   #define XDP_METADATA_KFUNC(name, _, __, ___) name,
> > > > > > > @@ -643,12 +647,22 @@ enum xdp_rss_hash_type {
> > > > > > >   	XDP_RSS_TYPE_L4_IPV6_SCTP_EX = XDP_RSS_TYPE_L4_IPV6_SCTP | XDP_RSS_L3_DYNHDR,
> > > > > > >   };
> > > > > > > +enum xdp_checksum {
> > > > > > > +	XDP_CHECKSUM_NONE		= CHECKSUM_NONE,
> > > > > > > +	XDP_CHECKSUM_UNNECESSARY	= CHECKSUM_UNNECESSARY,
> > > > > > > +	XDP_CHECKSUM_COMPLETE		= CHECKSUM_COMPLETE,
> > > > > > > +	XDP_CHECKSUM_PARTIAL		= CHECKSUM_PARTIAL,
> > > > > > > +};
> > > > > > 
> > > > > > Btw, might be worth mentioning, awhile ago we had settled on a smaller set of
> > > > > > exposed types:
> > > > > > 
> > > > > > https://lore.kernel.org/netdev/20230811161509.19722-13-larysa.zaremba@intel.com/
> > > > > > 
> > > > > > Maybe go through the previous postings and check if the arguments are
> > > > > > still relevant? (or explain why we want more checksum now)
> > > > > 
> > > > > IHMO the linked proposal reduced the types too much.
> > > > 
> > > > IIRC, PARTIAL was removed because it's mostly (or only) a TX feature?
> > > > So no real need to expose it as an rx hint. And I think empty xdp_csum_status
> > > > in that proposal might have indicated NONE?
> > > 
> > > Sorry for the (very) late reply. According to [0] CHECKSUM_PARTIAL can be used
> > > even on Rx side, right?
> > 
> > So is this for virtio (which I don't think you need)? Or something else?
> 
> I forgot to mention before CHECKSUM_PARTIAL is used for the veth use-case
> when the packet is coming from the networking stack.

But what do you do with that partial state in BPF? I doubt you're
calculating/verifying it? Can we treat/export it as NONE/UNNECESSARY for now?