On Sat, Mar 14, 2026 at 8:40 PM Jakub Kicinski <[email protected]> wrote: > > On Thu, 12 Mar 2026 10:54:06 +0800 xietangxin wrote: > > Fixes: f2fc6a54585a ("[NETNS][IPV6] route6 - move ip6_dst_ops inside the > > network namespace") > > Cc: [email protected] > > Signed-off-by: xietangxin <[email protected]> > > The Fixes tag should be: > > Fixes: 0287587884b1 ("net: better IFF_XMIT_DST_RELEASE support")
I disagree What was the situation before this patch ? I think virtio_net has been able to hold skbs way before IFF_XMIT_DST_RELEASE has been invented. Some archeology : commit 93f154b594fe47e4a7e5358b309add449a046cd3 Author: Eric Dumazet <[email protected]> Date: Mon May 18 22:19:19 2009 -0700 net: release dst entry in dev_hard_start_xmit() But really at that time struct dst_ops was not per netns The bug came when each netns got a copy of "stuct dst_ops" Not sure if 'fixing' virtio_net is enough. We really need to check all other drivers that might hold skb with dst for more than an RCU grace period. Or... not count dst anymore. What is the point anyway ?
