On 6 February 2016 at 04:37, Ben Pfaff <b...@ovn.org> wrote: > On Fri, Jan 29, 2016 at 12:34:12PM -0800, Joe Stringer wrote: >> On 29 January 2016 at 11:37, Joe Stringer <j...@ovn.org> wrote: >> > Depending on the kernel version in use, the nf_conntrack module may >> > register hooks for each namespace and execute conntrack prior to passing >> > packets to OVS (or not). In cases where this happens, the previous flow >> > table in this test would trust the local stack's connection tracking and >> > use it rather than sending packets through conntrack itself (and >> > therefore handling IP defragmentation/fragmentation). >> > >> > This patch revealed two fatal datapath bugs, fixed by these commits: >> > 86c2eb45fd82 datapath: Fix panic sending IP frags over tunnels. >> > XXXXXXXXXXXX datapath: inet: frag: Always orphan skbs inside ip_defrag(). >> > >> > Signed-off-by: Joe Stringer <j...@ovn.org> >> >> I should have made this more clear and perhaps included this testsuite >> with the latter XXXXX patch above. Note that without that patch, this >> test will crash your kernel. > > That worries me. > > Acked-by: Ben Pfaff <b...@ovn.org>
Typically I think that the proper etiquette is to send out the fix and the test updates in the same series, but I had already sent the fix. I updated the commit message with the fix's git commit ID and pushed it. _______________________________________________ dev mailing list dev@openvswitch.org http://openvswitch.org/mailman/listinfo/dev
