Re: [PATCH v2 net-next 0/2] ip: Batch flushing uncached routes per batched device unregistration.

[Chris Arges <carges@cloudflare.com>]

On 2026-09-29 20:15:06, Kuniyuki Iwashima wrote:
> Chris J Arges reported high RTNL contention during cleanup_net()
> caused by rt_flush_dev() and rt6_uncached_list_flush_dev() iterating
> over the global per-cpu uncached route lists for every netdev in
> dying netns [0].
> 
> This series resolves the issue by batching the uncached route
> cleanup after __rtnl_unlock() in netdev_run_todo().
> 
> [0]: https://lore.kernel.org/netdev/20260917-hash-bucket-route-lists-v3-0-30493a37b6eb@cloudflare.com/
> 
> 
> Changelog:
>   v2:
>     * Drop ->pre_exit_batch()
>     * Do batching after setting NETDEV_UNREGISTERED in netdev_run_todo()
> 
>   v1: https://lore.kernel.org/netdev/20260927202429.2452589-1-kuniyu@google.com/
> 
> 
> Kuniyuki Iwashima (2):
>   ipv4: Batch rt_flush_dev() in netdev_run_todo().
>   ipv6: Batch rt6_uncached_list_flush_dev() in netdev_run_todo().
> 
>  include/net/ip6_route.h |  7 +++++++
>  include/net/route.h     |  6 ++++++
>  net/core/dev.c          |  5 +++++
>  net/ipv4/route.c        | 11 +++++++++--
>  net/ipv6/route.c        | 18 ++++++++++++++----
>  5 files changed, 41 insertions(+), 6 deletions(-)
> 
> -- 
> 2.56.0.rc1.315.gc6ed9934b7-goog
> 
I was able to test and confirm that this patchset reduces RTNL lock contention
with my test case. Please add:
Tested-by: Chris J Arges <carges@cloudflare.com>

Also for "ipv4: Batch rt_flush_dev() in netdev_run_todo()" please include:
Reported-by: Chris J Arges <carges@cloudflare.com>.

Thanks,
--chris