diff options
| author |   David S. Miller <davem@davemloft.net> | 2020-04-27 12:43:30 -0700 |
|---|---|---|
| committer | David S. Miller <davem@davemloft.net> | 2020-04-27 12:43:30 -0700 |
| commit | 19f1f467bc5e9874ade015374d8f11a14abb2b3a (patch) | |
| tree | 06e7302af0363a6fad38f8033b7d131fda44dbf8 /net/ipv4/sysctl_net_ipv4.c | |
| parent | Merge branch 'r8169-improve-chip-config-handling' (diff) | |
| parent | selftests: forwarding: tc_actions.sh: add matchall mirror test (diff) | |
| download | wireguard-linux-19f1f467bc5e9874ade015374d8f11a14abb2b3a.tar.xz wireguard-linux-19f1f467bc5e9874ade015374d8f11a14abb2b3a.zip | |
Merge branch 'mlxsw-Rework-matchall-offloading-plumbing'
Ido Schimmel says:
====================
mlxsw: Rework matchall offloading plumbing
Jiri says:
Currently the matchall and flower are handled by registering separate
callbacks in mlxsw. That leads to faulty indication "in_hw_count 2" in
filter show command for every inserted flower filter. That happens
because matchall callback just blindly returns 0 for it and it is
wrongly accounted for as "the offloader".
I inspected different ways to fix this problem. The only clean solution
is to rework handling of matchall in mlxsw a bit. The driver newely
registers one callback for bound block which is called for both matchall
and flower filter insertions.
On the way, iron out the matchall code a bit, push it into a separate
file etc.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
Diffstat (limited to 'net/ipv4/sysctl_net_ipv4.c')
0 files changed, 0 insertions, 0 deletions