netdev - Re: [PATCH net] openvswitch: Fix egress tunnel info.

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]

Message-ID: <CALnjE+qLPeJN1h1qFnYuiipbJH06Gtx_7vJGzjocVoH40fNUcQ@mail.gmail.com>
Date:	Tue, 6 Oct 2015 12:11:05 -0700
From:	Pravin Shelar <pshelar@...ira.com>
To:	Jiri Benc <jbenc@...hat.com>
Cc:	netdev <netdev@...r.kernel.org>
Subject: Re: [PATCH net] openvswitch: Fix egress tunnel info.

On Tue, Oct 6, 2015 at 11:55 AM, Jiri Benc <jbenc@...hat.com> wrote:
> On Tue, 6 Oct 2015 11:26:31 -0700, Pravin Shelar wrote:
>> I do not see need to drop and reallocate dst in this operation. I just
>> need to set source IP address and source and dst port in the metadata
>> dst already set in skb.
>
> If I'm looking at the code correctly, metadata_dst is stored in the
> action and each skb gets only a reference to it. Modifying it would
> modify the shared metadata_dst (see execute_set_action).
>
right, I was thinking about updating just the source IPv4 address
which should be same for given tunnel metadata. But that can get
complicated for updates to ipv6 address.

>> This fill_metadata function is not called for every packet so
>> ndo_start_xmit() still needs to do route lookup.
>
> Yes. I meant that in those cases where the fill_metadata function was
> called, we may skip the lookup. Just an optimization and not an
> important one. I'm not even sure it can currently happen as the skb is
> cloned for each action.
>
The function is called for sampling which is rare operation. So I do
not want to complicate code to optimize such case.

>> egress_tun_info in dp_upcall_info is required to check for failure. It
>> would be only set on successful fill_metadata operation.
>
> Or you can just set dst to NULL on failure.
>

In that case we would need to extra logic to update error stats
correctly device xmit operation. For example geneve_xmit() updates
different error stats for route lookup.
--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html