On Tue, Jul 26, 2016 at 3:59 PM, pravin shelar <pshe...@ovn.org> wrote: > On Tue, Jul 26, 2016 at 3:53 PM, Jesse Gross <je...@kernel.org> wrote: >> On Tue, Jul 26, 2016 at 3:24 PM, Pravin B Shelar <pshe...@ovn.org> wrote: >>> diff --git a/datapath/linux/compat/include/net/udp.h >>> b/datapath/linux/compat/include/net/udp.h >>> index fa49fa5..266e70a 100644 >>> --- a/datapath/linux/compat/include/net/udp.h >>> +++ b/datapath/linux/compat/include/net/udp.h >>> @@ -54,7 +54,7 @@ static inline __sum16 udp_v4_check(int len, __be32 saddr, >>> } >>> #endif >>> >>> -#ifndef HAVE_UDP_SET_CSUM >>> +#if LINUX_VERSION_CODE < KERNEL_VERSION(3,18,0) >> >> I'm a little nervous about these version checks being hard to maintain >> - especially since they don't correspond to anything obvious in this >> function upstream. Maybe we could just declare a #define with a name >> that would make it clearer. That might actually be useful in any case >> since I suspect that we will start seeing some backports in >> distributions that will allow us to avoid doing OVS segmentation even >> on older kernels. > > Is it fine if I do it as part of separate patch? This patch is about > fixing the UDP checksum issue. And the requested change is about > general code improvement.
Yes, that's fine. I think we'll want to convert all of the GSO related 3.18 version checks to use this symbol, so that's mostly not related to checksums anyways. Acked-by: Jesse Gross <je...@kernel.org> _______________________________________________ dev mailing list dev@openvswitch.org http://openvswitch.org/mailman/listinfo/dev
