On Thu, Jan 17, 2019 at 10:07 AM Cong Wang <xiyou.wangc...@gmail.com> wrote: > > On Thu, Jan 17, 2019 at 3:43 AM yupeng <yupeng0...@gmail.com> wrote: > > diff --git a/include/uapi/linux/snmp.h b/include/uapi/linux/snmp.h > > index 86dc24a96c90..fb0a17337bf2 100644 > > --- a/include/uapi/linux/snmp.h > > +++ b/include/uapi/linux/snmp.h > > @@ -178,7 +178,6 @@ enum > > LINUX_MIB_ARPFILTER, /* ArpFilter */ > > LINUX_MIB_TIMEWAITED, /* TimeWaited */ > > LINUX_MIB_TIMEWAITRECYCLED, /* TimeWaitRecycled */ > > - LINUX_MIB_TIMEWAITKILLED, /* TimeWaitKilled */ > > LINUX_MIB_PAWSACTIVEREJECTED, /* PAWSActiveRejected */ > > LINUX_MIB_PAWSESTABREJECTED, /* PAWSEstabRejected */ > > LINUX_MIB_DELAYEDACKS, /* DelayedACKs */ > > This is no way to go, it is breaking user-space. > > You should at least keep this definition even if this counter is unused.
>From the snmp.h commit history, I didn't aware that we should keep the user-space compatibility. For example, the commit 713bafea9292 removed LINUX_MIB_TCPFACKREORDER, and the commit 4f693b55c3d2 inserted LINUX_MIB_TCPBACKLOGCOALESCE to the middle of the linux mib definitions enum. I think these commits would break the user-space compatibility too. Did they use any method to avoid breaking the user-space compatibility? Or if I misunderstand anything, please let me know. Thanks a lot.
