Re: [PATCH 0/3] selftests/net: A couple of typos fixes in key-management test

[Jakub Kicinski <kuba@kernel.org>]

On Wed, 24 Jan 2024 17:46:10 +0000 Dmitry Safonov wrote:
> >> Thanks!
> >>
> >> I'll send a patch for it in version 2 (as I anyway need to address
> >> Simon's feedback).  
> > 
> > Hi Dmitry!
> > 
> > I put TCP_AO and VETH in the config and the tests seem to fail with  
> 
> Thanks for wiring it up and for https://netdev.bots.linux.dev/status.html!
> 
> > selftests: net/tcp_ao: rst_ipv4
> > not ok 1 # error 834[lib/kconfig.c:143] Failed to initialize kconfig 2: No such file or directory
> > # Planned tests != run tests (0 != 1)
> > # Totals: pass:0 fail:0 xfail:0 xpass:0 skip:0 error:1  
> 
> Hehe, yeah I wanted to detect kernels with !CONFIG_TCP_AO, to SKIP the
> test, rather than FAIL it, which this lib/kconfig.c does.
> But from a glance, I think it's failing in your run because there are
> checks with and without TCP_AO, but I didn't think of checking for
> the hashing algorithms support.
> 
> I think what happens is has_tcp_ao():
> : strcpy(tmp.alg_name, "hmac(sha1)");
> ...
> : if (setsockopt(sk, IPPROTO_TCP, TCP_AO_ADD_KEY, &tmp, sizeof(tmp)) < 0)
> 
> Could you check that what I suppose is failing, is actually failing?
> [dima@Mindolluin linux-master]$ grep -e '\<CONFIG_CRYPTO_SHA1\>' -e
> '\<CONFIG_CRYPTO_HMAC\>' .config
> CONFIG_CRYPTO_HMAC=y
> CONFIG_CRYPTO_SHA1=y

FWIW the config used is uploaded with the results. If you click on 
the remote it should take you to a location like this:

https://netdev-2.bots.linux.dev/vmksft-tcp-ao/results/435369/

and there should be a config file in there. 

> If that's the case, I'll  add the detection for hashing algorithms to
> lib/kconfig.c (together with a patch for
> tools/testing/selftests/net/config).
> And also heads up for key-management.c - that tries a bunch of hashing
> algorithms to check that the work and that the key rotation between
> different algorithms works:
> 
> : const char *test_algos[] = {
> : 	"cmac(aes128)",
> : 	"hmac(sha1)", "hmac(sha512)", "hmac(sha384)", "hmac(sha256)",
> : 	"hmac(sha224)", "hmac(sha3-512)",
> : 	/* only if !CONFIG_FIPS */
> : #define TEST_NON_FIPS_ALGOS	2
> : 	"hmac(rmd160)", "hmac(md5)"
> : };

I was stuck in a meeting and I started playing around with the options 
for TCP-AO :) I added these options now:

CONFIG_CRYPTO_HMAC=y
CONFIG_CRYPTO_SHA1=y
CONFIG_CRYPTO_RMD160=y
CONFIG_IPV6=y
CONFIG_TCP_AO=y
CONFIG_TCP_MD5SIG=y
CONFIG_VETH=m

And it looks much better! There are still some failures:

https://netdev.bots.linux.dev/contest.html?branch=net-next-2024-01-24--18-00&executor=vmksft-tcp-ao

I added VRF so that should hopefully take care of the MD5 skips
on the next run. But the failures of the rst-ip* tests don't look 
like an obvious config problem.