netdev - Re: [PATCH net-next ct-offload v3 00/15] Introduce connection tracking offload

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:   Wed, 11 Mar 2020 21:01:26 -0300
From:   Marcelo Ricardo Leitner <marcelo.leitner@...il.com>
To:     Paul Blakey <paulb@...lanox.com>
Cc:     Saeed Mahameed <saeedm@...lanox.com>,
        Oz Shlomo <ozsh@...lanox.com>,
        Vlad Buslov <vladbu@...lanox.com>,
        David Miller <davem@...emloft.net>,
        "netdev@...r.kernel.org" <netdev@...r.kernel.org>,
        Jiri Pirko <jiri@...lanox.com>, Roi Dayan <roid@...lanox.com>
Subject: Re: [PATCH net-next ct-offload v3 00/15] Introduce connection
 tracking offload

On Wed, Mar 11, 2020 at 07:44:15PM -0300, Marcelo Ricardo Leitner wrote:
> On Thu, Mar 12, 2020 at 12:27:37AM +0200, Paul Blakey wrote:
...
> > if not try skipping flow_action_hw_stats_types_check() calls in mlx5 driver.
> 
> Ok. This one was my main suspect now after some extra printks. I could
> confirm it parse_tc_fdb_actions is returning the error, but not sure
> why yet.

Copied the extack in flow_action_mixed_hw_stats_types_check() onto a
printk and logged the if parameters:

@@ -284,6 +284,8 @@ flow_action_mixed_hw_stats_types_check(const struct flow_action *action,

        flow_action_for_each(i, action_entry, action) {
                if (i && action_entry->hw_stats_type != last_hw_stats_type) {
+                       printk("Mixing HW stats types for actions is not supported, %d %d %d\n",
+                              i, action_entry->hw_stats_type, last_hw_stats_type);
                        NL_SET_ERR_MSG_MOD(extack, "Mixing HW stats types for actions is not supported");

[  188.554636] Mixing HW stats types for actions is not supported, 2 0 3

with iproute-next from today loaded with
'iproute2/net-next v2] tc: m_action: introduce support for hw stats
type', I get a dump like:

# ./tc -s filter show dev vxlan_sys_4789 ingress
filter protocol ip pref 2 flower chain 0
filter protocol ip pref 2 flower chain 0 handle 0x1
  dst_mac 6a:66:2d:48:92:c2
  src_mac 00:00:00:00:0e:b7
  eth_type ipv4
  ip_proto udp
  ip_ttl 64
  src_port 100
  enc_dst_ip 1.1.1.1
  enc_src_ip 1.1.1.2
  enc_key_id 0
  enc_dst_port 4789
  enc_tos 0
  ip_flags nofrag
  not_in_hw
        action order 1: tunnel_key  unset pipe
         index 4 ref 1 bind 1 installed 2432 sec used 0 sec
        Action statistics:
        Sent 223744 bytes 4864 pkt (dropped 0, overlimits 0 requeues 0)
        backlog 0b 0p requeues 0
        no_percpu

        action order 2:  pedit action pipe keys 2
         index 4 ref 1 bind 1 installed 2432 sec used 0 sec
         key #0  at ipv4+8: val 3f000000 mask 00ffffff
         key #1  at udp+0: val 13890000 mask 0000ffff
        Action statistics:
        Sent 223744 bytes 4864 pkt (dropped 0, overlimits 0 requeues 0)
        backlog 0b 0p requeues 0

        action order 3: csum (iph, udp) action pipe
        index 4 ref 1 bind 1 installed 2432 sec used 0 sec
        Action statistics:
        Sent 223744 bytes 4864 pkt (dropped 0, overlimits 0 requeues 0)
        backlog 0b 0p requeues 0
        no_percpu

        action order 4: mirred (Egress Redirect to device eth0) stolen
        index 1 ref 1 bind 1 installed 2432 sec used 0 sec
        Action statistics:
        Sent 223744 bytes 4864 pkt (dropped 0, overlimits 0 requeues 0)
        backlog 0b 0p requeues 0
        cookie 9c7d6b3aa84a32498181d98da0af80d2
        no_percpu

Which is quite interesting as it is not listing any hwstats fields at all.
Considering that due to tcf_action_hw_stats_type_get() and
hw_stats_type initialization in tcf_action_init_1(), the default is
for it to be 3 if not specified (which is then not dumped over netlink
by the kernel).

I'm wondering how it was 0 for i==0,1 and is 3 for i==2.