netdev - Conntrack offload and ingress

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>] [day] [month] [year] [list]

Message-ID: <e4ba278d-3308-ada8-d4ab-4d3a6c489216@gmail.com>
Date:   Fri, 1 Apr 2022 11:40:52 +0100
From:   Edward Cree <ecree.xilinx@...il.com>
To:     pablo@...filter.org
Cc:     Network Development <netdev@...r.kernel.org>,
        netfilter-devel@...r.kernel.org
Subject: Conntrack offload and ingress_ifindex

Pablo,

When developing a conntrack offload driver, I was quite surprised to find
 that CT entries passed to the driver's flow-block callback (as registered
 with nf_flow_table_offload_add_cb()) include a match on ingress_ifindex,
 with mask 0xffffffff and key 0.  This is especially confusing as AIUI 0 is
 not a valid ifindex.
>From reading the calling code, looking at git logs etc, I can't determine
 the intended semantics of this match; could you clarify what (if anything)
 drivers are expected to do with it?
(Looking at other drivers it appears that e.g. mlx5e simply ignores it, as
 its test for `key & mask` in mlx5_tc_ct_set_tuple_match() will be false.)

-ed