Date: Wed, 29 Apr 2020 11:12:46 -0700 From: Eric Joyner <erj@freebsd.org> To: Aleksandr Fedorov <wigneddoom@yandex.ru> Cc: "src-committers@freebsd.org" <src-committers@freebsd.org>, "svn-src-all@freebsd.org" <svn-src-all@freebsd.org>, "svn-src-head@freebsd.org" <svn-src-head@freebsd.org> Subject: Re: svn commit: r360398 - head/sys/net Message-ID: <CAKdFRZiRHCd1VapmJRt0kye%2BDJsfmXcjZD-1n0zAUvpqXcb8-Q@mail.gmail.com> In-Reply-To: <122841588148797@mail.yandex.ru> References: <202004272202.03RM2iR2012757@repo.freebsd.org> <122841588148797@mail.yandex.ru>
next in thread | previous in thread | raw e-mail | index | archive | help
Hi Aleksandr, Part of this patch was to introduce a driver-dependent iflib function that would allow drivers to tell iflib whether or not to flap the interface when a vlan is registered or unregistered. I think this could be fixed for em(4) by having it implement that function to either always return false (so iflib doesn't do an init/stop), or maybe have it only do a reset when vlanhwfilter is enabled; I'm going to take a look at doing that. - Eric On Wed, Apr 29, 2020 at 1:27 AM Aleksandr Fedorov <wigneddoom@yandex.ru> wrote: > Sorry for the late response, but I think that this changes makes Intel > cards with iflib drivers fully unusable. > We already have a regression which leads to an interface flapping > when adding vlan with enabled vlanhwfilter. But at least there is a > workaround - disabling hwfilter. > After this change, the interface will always flap. > > https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=241785 > https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=240818 > https://lists.freebsd.org/pipermail/freebsd-net/2018-November/052184.html > >
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?CAKdFRZiRHCd1VapmJRt0kye%2BDJsfmXcjZD-1n0zAUvpqXcb8-Q>