| Message ID | 20220329113523.19793-1-fw@strlen.de (mailing list archive) |
|---|---|
| Headers | show |
| Series | mptcp: inet diag listen dump support | expand |
On Tue, 29 Mar 2022, Florian Westphal wrote: > Changes in v3: fix intendation, fix uninitialised var (Mat) > extend mptcp diag tests Thanks for adding the tests. Series looks good to me: Reviewed-by: Mat Martineau <mathew.j.martineau@linux.intel.com> > > Changes in v2: drop patch 4 and prefer msk->first (Paolo Abeni) > > Done in a not-very-elegant way: > Iterate over tcp listen hash, then pick out the tcp sockets > with mptcp-ctx structure attached, then take the conn->sk for dumping. > > First patch is preparation/cleanup. > Second patch gets rid of locking to avoid a lockdep splat. > If the socket lock is really needed (I don't see where) I can workaround > this by dropping locks temporarily when iterating the listen hash table, > but its a bit more awkward. > > Sample output: > ss -Mil > State Recv-Q Send-Q Local Address:Port Peer Address:Port > LISTEN 0 20 127.0.0.1:12000 0.0.0.0:* > subflows_max:2 > > Florian Westphal (4): > mptcp: diag: switch to context structure > mptcp: remove locking in mptcp_diag_fill_info > mptcp: listen diag dump support > selftests/mptcp: add diag listen tests > > net/mptcp/mptcp_diag.c | 105 +++++++++++++++++++++- > net/mptcp/sockopt.c | 6 -- > tools/testing/selftests/net/mptcp/diag.sh | 38 ++++++++ > 3 files changed, 140 insertions(+), 9 deletions(-) > > -- > 2.35.1 > > > -- Mat Martineau Intel
Hi Florian, Mat, On 29/03/2022 13:35, Florian Westphal wrote: > Changes in v3: fix intendation, fix uninitialised var (Mat) > extend mptcp diag tests > > Changes in v2: drop patch 4 and prefer msk->first (Paolo Abeni) > > Done in a not-very-elegant way: > Iterate over tcp listen hash, then pick out the tcp sockets > with mptcp-ctx structure attached, then take the conn->sk for dumping. > > First patch is preparation/cleanup. > Second patch gets rid of locking to avoid a lockdep splat. > If the socket lock is really needed (I don't see where) I can workaround > this by dropping locks temporarily when iterating the listen hash table, > but its a bit more awkward. > > Sample output: > ss -Mil > State Recv-Q Send-Q Local Address:Port Peer Address:Port > LISTEN 0 20 127.0.0.1:12000 0.0.0.0:* > subflows_max:2 Thank you for these patches and reviews! Now in our tree (feat. for net-next) with Mat's RvB tag: New patches for t/upstream: - 0432b1c61654: mptcp: diag: switch to context structure - ba9e26b62530: mptcp: remove locking in mptcp_diag_fill_info - b433728093dd: mptcp: listen diag dump support - a9cc2c58bf79: selftests/mptcp: add diag listen tests - Results: ccc95f7c14d9..3b82e3d7eed2 (export) Builds and tests are now in progress: https://cirrus-ci.com/github/multipath-tcp/mptcp_net-next/export/20220330T150002 https://github.com/multipath-tcp/mptcp_net-next/actions/workflows/build-validation.yml?query=branch:export Cheers, Matt
Changes in v3: fix intendation, fix uninitialised var (Mat) extend mptcp diag tests Changes in v2: drop patch 4 and prefer msk->first (Paolo Abeni) Done in a not-very-elegant way: Iterate over tcp listen hash, then pick out the tcp sockets with mptcp-ctx structure attached, then take the conn->sk for dumping. First patch is preparation/cleanup. Second patch gets rid of locking to avoid a lockdep splat. If the socket lock is really needed (I don't see where) I can workaround this by dropping locks temporarily when iterating the listen hash table, but its a bit more awkward. Sample output: ss -Mil State Recv-Q Send-Q Local Address:Port Peer Address:Port LISTEN 0 20 127.0.0.1:12000 0.0.0.0:* subflows_max:2 Florian Westphal (4): mptcp: diag: switch to context structure mptcp: remove locking in mptcp_diag_fill_info mptcp: listen diag dump support selftests/mptcp: add diag listen tests net/mptcp/mptcp_diag.c | 105 +++++++++++++++++++++- net/mptcp/sockopt.c | 6 -- tools/testing/selftests/net/mptcp/diag.sh | 38 ++++++++ 3 files changed, 140 insertions(+), 9 deletions(-)