From: Stephen Hemminger <stephen@networkplumber.org>
To: Antony Antony <antony.antony@secunet.com>
Cc: Steffen Klassert <steffen.klassert@secunet.com>,
David Ahern <dsahern@gmail.com>,
Herbert Xu <herbert@gondor.apana.org.au>,
<netdev@vger.kernel.org>, Eyal Birger <eyal.birger@gmail.com>,
Matt Ellison <matt@arroyo.io>
Subject: Re: Regression in add xfrm interface
Date: Tue, 8 Mar 2022 23:08:20 -0800 [thread overview]
Message-ID: <20220308230820.27145272@hermes.local> (raw)
In-Reply-To: <YidTpIty2fVKTBzM@moon.secunet.de>
On Tue, 8 Mar 2022 14:51:30 +0100
Antony Antony <antony.antony@secunet.com> wrote:
> Hi Stephen,
>
> As Steffen explained bellow if_id = 0 is likely to cause problems in the long
> term. Should we revert the commit because it broke userspace tools?
>
> I notice the Debian bug is in a iproute2 testsuite, also it is in Debian testing! How about fixing test case than reverting the kernel commit?
>
> Another option is revert the commit in current kernel development cycle.
> And send the same fix to ipsec-next without "Fixes" tag.
> Would that be acceptable for Debian testsuite usecase?
>
> regards,
> -antony
>
> On Tue, Mar 08, 2022 at 08:50:13 +0100, Steffen Klassert wrote:
> > On Mon, Mar 07, 2022 at 12:11:23PM -0800, Stephen Hemminger wrote:
> > > There appears to be a regression between 5.10 (Debian 11) and 5.16 (Debian testing)
> > > kernel in handling of ip link xfrm create. This shows up in the iproute2 testsuite
> > > which now fails. This is kernel (not iproute2) regression.
> > >
> > >
> > > Running ip/link/add_type_xfrm.t [iproute2-this/5.16.0-1-amd64]: FAILED
> > >
> > >
> > > Good log:
> > > ::::::::::::::
> > > link/add_type_xfrm.t.iproute2-this.out
> > > ::::::::::::::
> > > [Testing Add XFRM Interface, With IF-ID]
> > > tests/ip/link/add_type_xfrm.t: Add dev-ktyXSm xfrm interface succeeded
> > > tests/ip/link/add_type_xfrm.t: Show dev-ktyXSm xfrm interface succeeded with output:
> > > 2: dev-ktyXSm@lo: <NOARP,M-DOWN> mtu 1500 qdisc noop state DOWN mode DEFAULT group default qlen 1000
> > > link/none promiscuity 0 minmtu 68 maxmtu 65535
> > > xfrm if_id 0xf addrgenmode eui64 numtxqueues 1 numrxqueues 1 gso_max_size 65536 gso_max_segs 65535
> > > test on: "dev-ktyXSm" [SUCCESS]
> > > test on: "if_id 0xf" [SUCCESS]
> > > tests/ip/link/add_type_xfrm.t: Del dev-ktyXSm xfrm interface succeeded
> > > [Testing Add XFRM Interface, No IF-ID]
> > > tests/ip/link/add_type_xfrm.t: Add dev-tkUDaA xfrm interface succeeded
> > > tests/ip/link/add_type_xfrm.t: Show dev-tkUDaA xfrm interface succeeded with output:
> > > 3: dev-tkUDaA@lo: <NOARP,M-DOWN> mtu 1500 qdisc noop state DOWN mode DEFAULT group default qlen 1000
> > > link/none promiscuity 0 minmtu 68 maxmtu 65535
> > > xfrm if_id 0 addrgenmode eui64 numtxqueues 1 numrxqueues 1 gso_max_size 65536 gso_max_segs 65535
> > > test on: "dev-tkUDaA" [SUCCESS]
> > > test on: "if_id 0xf" [SUCCESS]
> > > tests/ip/link/add_type_xfrm.t: Del dev-tkUDaA xfrm interface succeeded
> > >
> > > Failed log:
> > >
> > > [Testing Add XFRM Interface, With IF-ID]
> > > tests/ip/link/add_type_xfrm.t: Add dev-pxNsUc xfrm interface succeeded
> > > tests/ip/link/add_type_xfrm.t: Show dev-pxNsUc xfrm interface succeeded with output:
> > > 2: dev-pxNsUc@lo: <NOARP,M-DOWN> mtu 1500 qdisc noop state DOWN mode DEFAULT group default qlen 1000
> > > link/none promiscuity 0 minmtu 68 maxmtu 65535
> > > xfrm if_id 0xf addrgenmode eui64 numtxqueues 1 numrxqueues 1 gso_max_size 65536 gso_max_segs 65535
> > > test on: "dev-pxNsUc" [SUCCESS]
> > > test on: "if_id 0xf" [SUCCESS]
> > > tests/ip/link/add_type_xfrm.t: Del dev-pxNsUc xfrm interface succeeded
> > > [Testing Add XFRM Interface, No IF-ID]
> >
> > No IF-ID is an invalid configuration, the interface does not work
> > with IF-IF 0. Such an interface will blackhole all packets routed
> > to it. That is because policies and states with no IF-ID are meant
> > for a setup without xfrm interfaces, they will not match the interface.
> >
> > Unfortunately we did not catch this invalid configuration from the
> > beginning and userspace seems to use (or do some tests tests with)
> > xfrm interfaces with IF-ID 0. In that case, I fear we eventually
> > have to revert the cange that catches the invalid configuration.
> >
The test is here please update as appropriate:
https://git.kernel.org/pub/scm/network/iproute2/iproute2.git/tree/testsuite/tests/ip/link/add_type_xfrm.t
And was added by:
commit 286446c1e8c7f5f6eca4959015aa9e482b7adb11
Author: Matt Ellison <matt@arroyo.io>
Date: Thu Apr 4 10:08:45 2019 -0400
ip: support for xfrm interfaces
Interfaces take a 'if_id' which is an interface id which can be set on
an xfrm policy as its interface lookup key (XFRMA_IF_ID).
Signed-off-by: Matt Ellison <matt@arroyo.io>
Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
prev parent reply other threads:[~2022-03-09 7:08 UTC|newest]
Thread overview: 5+ messages / expand[flat|nested] mbox.gz Atom feed top
2022-03-07 20:11 Regression in add xfrm interface Stephen Hemminger
2022-03-08 7:50 ` Steffen Klassert
2022-03-08 13:51 ` Antony Antony
2022-03-08 16:45 ` Stephen Hemminger
2022-03-09 7:08 ` Stephen Hemminger [this message]
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20220308230820.27145272@hermes.local \
--to=stephen@networkplumber.org \
--cc=antony.antony@secunet.com \
--cc=dsahern@gmail.com \
--cc=eyal.birger@gmail.com \
--cc=herbert@gondor.apana.org.au \
--cc=matt@arroyo.io \
--cc=netdev@vger.kernel.org \
--cc=steffen.klassert@secunet.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).