Re: [PATCH 3/3] virtio-net: remove VIRTIO_NET_HDR_F_RSC_INFO compat handling

[Cornelia Huck]

On Tue, 28 Apr 2020 16:58:44 +0800
Jason Wang <address@hidden> wrote:

> On 2020/4/28 下午4:34, Cornelia Huck wrote:
> > On Tue, 28 Apr 2020 16:19:15 +0800
> > Jason Wang <address@hidden> wrote:
> >  
> >> On 2020/4/27 下午6:24, Cornelia Huck wrote:  
> >>> VIRTIO_NET_HDR_F_RSC_INFO is available in the headers now.
> >>>
> >>> Signed-off-by: Cornelia Huck <address@hidden>
> >>> ---
> >>>    hw/net/virtio-net.c | 8 --------
> >>>    1 file changed, 8 deletions(-)
> >>>
> >>> diff --git a/hw/net/virtio-net.c b/hw/net/virtio-net.c
> >>> index e85d902588b3..7449570c7123 100644
> >>> --- a/hw/net/virtio-net.c
> >>> +++ b/hw/net/virtio-net.c
> >>> @@ -77,14 +77,6 @@
> >>>       tso/gso/gro 'off'. */
> >>>    #define VIRTIO_NET_RSC_DEFAULT_INTERVAL 300000
> >>>    
> >>> -/* temporary until standard header include it */
> >>> -#if !defined(VIRTIO_NET_HDR_F_RSC_INFO)
> >>> -
> >>> -#define VIRTIO_NET_HDR_F_RSC_INFO  4 /* rsc_ext data in csum_ fields */
> >>> -#define VIRTIO_NET_F_RSC_EXT       61
> >>> -
> >>> -#endif
> >>> -
> >>>    static inline __virtio16 *virtio_net_rsc_ext_num_packets(
> >>>        struct virtio_net_hdr *hdr)
> >>>    {  
> >>
> >> I think we should not keep the those tricky num_packets/dup_acks.  
> > No real opinion here, patch 3 is only a cleanup.
> >
> > The important one is patch 1, because without it I cannot do a headers
> > update.  
> 
> 
> Yes, at least we should dereference segments/dup_acks instead of 
> csum_start/csum_offsets since the header has been synced.

So what about:

- I merge patch 1 and the header sync now (because I have a bunch of
  patches that depend on it...)
- We change virtio-net to handle that properly on top (probably best
  done by someone familiar with the code base ;)