On Fri, 11 Sep 2020 at 01:53, Alexei Starovoitov
<alexei.starovoitov@xxxxxxxxx> wrote:
>
> On Thu, Sep 10, 2020 at 4:03 AM Lorenz Bauer <lmb@xxxxxxxxxxxxxx> wrote:
> >
> > As Alexei points out, struct bpf_sk_lookup_kern has two 4-byte holes.
> > This leads to suboptimal instructions being generated (IPv4, x86):
> >
> > Fix this by moving around sport and dport. pahole confirms there
> > are no more holes:
> >
> > Fixes: e9ddbb7707ff ("bpf: Introduce SK_LOOKUP program type with a dedicated attach point")
> > Suggested-by: Alexei Starovoitov <ast@xxxxxxxxxx>
> > Signed-off-by: Lorenz Bauer <lmb@xxxxxxxxxxxxxx>
>
> Applied to bpf-next.
> I feel it's a bit of a stretch to consider it a fix, but if you really
> insist I can let it go as a fix
> and reapply to a different tree. Just let me know.
Thanks, bpf-next is fine by me.
--
Lorenz Bauer | Systems Engineer
6th Floor, County Hall/The Riverside Building, SE1 7PB, UK
www.cloudflare.com
