Re: [PATCH net-next 0/2] fixes for ipsec selftests

On 6/21/2018 9:56 AM, Anders Roxell wrote:

On Thu, 21 Jun 2018 at 02:32, Shannon Nelson [off-list ref] wrote:

quoted

On 6/20/2018 4:18 PM, Anders Roxell wrote:

quoted

On Thu, 21 Jun 2018 at 00:26, Shannon Nelson [off-list ref] wrote:

quoted

On 6/20/2018 12:09 PM, Anders Roxell wrote:

quoted

On Wed, 20 Jun 2018 at 07:42, Shannon Nelson [off-list ref] wrote:

quoted

A couple of bad behaviors in the ipsec selftest were pointed out
by Anders Roxell [off-list ref] and are addressed here.

Shannon Nelson (2):
     selftests: rtnetlink: hide complaint from terminated monitor
     selftests: rtnetlink: use a local IP address for IPsec tests

    tools/testing/selftests/net/rtnetlink.sh | 11 +++++++----
    1 file changed, 7 insertions(+), 4 deletions(-)

--
2.7.4

Hi Shannon,

With this patches applied and my config patch.

I still get this error when I run the ipsec test:

FAIL: can't add fou port 7777, skipping test
RTNETLINK answers: Operation not supported
FAIL: can't add macsec interface, skipping test
RTNETLINK answers: Protocol not supported
RTNETLINK answers: No such process
RTNETLINK answers: No such process
FAIL: ipsec

One of the odd things I noticed about this script is that there really
aren't any diagnosis messages, just PASS or FAIL.  I followed this
custom when I added the ipsec tests, but I think this is something that
should change so we can get some idea of what breaks.

I'm curious about the "RTNETLINK answers" messages and where they might
be coming from, especially "RTNETLINK answers: Protocol not supported".

I added: "set -x" in the beginning of the rtnetlink.sh script.
+ ip x s add proto esp src 10.66.17.140 dst 10.66.17.141 spi 0x07 mode
transport reqid 0x07 replay-window 32 aead 'rfc4106(gcm(aes))'
0x3132333435
363738393031323334353664636261 128 sel src 10.66.17.140/24 dst 10.66.17.141/24
RTNETLINK answers: Protocol not supported

Okay, so ip didn't like this command...

quoted

quoted

What are the XFRM and AES settings in your kernel config - what is the
output from
          egrep -i "xfrm|_aes" .config

CONFIG_XFRM=y
CONFIG_XFRM_ALGO=y
CONFIG_XFRM_USER=y
CONFIG_INET_XFRM_MODE_TUNNEL=y
CONFIG_INET6_XFRM_MODE_TRANSPORT=y
CONFIG_INET6_XFRM_MODE_TUNNEL=y
CONFIG_INET6_XFRM_MODE_BEET=y
CONFIG_CRYPTO_AES=y

And this is probably why - there seem to be a few config variables
missing, including CONFIG_INET_XFRM_MODE_TRANSPORT, which might be why
the ip command fails above.

Here's what I have in my config:
CONFIG_XFRM=y
CONFIG_XFRM_OFFLOAD=y
CONFIG_XFRM_ALGO=m
CONFIG_XFRM_USER=m
# CONFIG_XFRM_SUB_POLICY is not set
# CONFIG_XFRM_MIGRATE is not set
CONFIG_XFRM_STATISTICS=y
CONFIG_XFRM_IPCOMP=m
CONFIG_INET_XFRM_TUNNEL=m
CONFIG_INET_XFRM_MODE_TRANSPORT=m
CONFIG_INET_XFRM_MODE_TUNNEL=m
CONFIG_INET_XFRM_MODE_BEET=m
CONFIG_INET6_XFRM_TUNNEL=m
CONFIG_INET6_XFRM_MODE_TRANSPORT=m
CONFIG_INET6_XFRM_MODE_TUNNEL=m
CONFIG_INET6_XFRM_MODE_BEET=m
CONFIG_INET6_XFRM_MODE_ROUTEOPTIMIZATION=m
CONFIG_SECURITY_NETWORK_XFRM=y
CONFIG_CRYPTO_AES=y
# CONFIG_CRYPTO_AES_TI is not set
CONFIG_CRYPTO_AES_X86_64=m
CONFIG_CRYPTO_AES_NI_INTEL=m
CONFIG_CRYPTO_CAMELLIA_AESNI_AVX_X86_64=m
CONFIG_CRYPTO_CAMELLIA_AESNI_AVX2_X86_64=m
CONFIG_CRYPTO_DEV_PADLOCK_AES=m

Can I talk you into adding CONFIG_INET_XFRM_MODE_TRANSPORT to your
config

Yes you can.

quoted

and trying again?

same issue with CONFIG_INET_XFRM_MODE_TRANSPORT=y

Interesting.  I took only CONFIG_INET_XFRM_MODE_TRANSPORT out of my 
config and was able to see the "Protocol not supported" message.  I'm 
not familiar enough with the crypto algorithm setup, but I suspect 
there's a combination of the other missing CONFIGs that are needed along 
with CONFIG_INET_XFRM_MODE_TRANSPORT.

My knee-jerk reaction voice wants to say this is the test working as 
expected, pointing out to us that the kernel config is not up to what it 
should be.  However, perhaps a better answer is that the test should be 
reworked to just skip the rest if it can't set up the expected test 
environment, as is done in the macsec case.

So the remaining question then is should the test be marked as failed, 
as in the macsec test if it can't set up it's interface, or just skipped?

sln

Cheers,
Anders