12:02:41 <jki> #startmeeting CIP IRC weekly meeting 12:02:41 <collab-meetbot`> Meeting started Thu Aug 17 12:02:41 2023 UTC and is due to finish in 60 minutes. The chair is jki. Information about MeetBot at http://wiki.debian.org/MeetBot. 12:02:41 <collab-meetbot`> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 12:02:41 <collab-meetbot`> The meeting name has been set to 'cip_irc_weekly_meeting' 12:02:49 <jki> #topic AI review 12:02:54 <jki> 1. create kernelci pipeline for buster images (arisut) 12:03:58 <jki> arisut: your isar-cip-core patch is related, I assume 12:04:08 <jki> so, still ongoing, right? 12:04:13 <arisut> i sended a mail to isar-cip for a patch fir adding buster, bookworm and bullseye 12:04:49 <arisut> Any prospect on merging time? Or questions? 12:05:19 <jki> I'm still wondering about the number of jobs we spawn, and if that is needed for every CI run 12:05:50 <jki> specifically as we seem to hit some wall at git.kernel.org around fetching tarballs 12:06:05 <jki> but let's follow up on the mailing list on that, ok? 12:06:59 <arisut> Yeah i see that, we cannot mirror it anymore? 12:07:32 <jki> we had issues with github - or was it gitlab? - in the past as well 12:08:20 <jki> see commit 73f779e067b75a7fbb09bdcf6f8744b41b3c1802 12:08:37 <arisut> Ok 12:08:39 <jki> "gitlab.com turned out to be too unreliable for fetching on-the-fly generated kernel tarballs in CI. Let's hope kernel.org will do better." - well... 12:08:55 <arisut> Other than that looks ok? 12:09:16 <jki> but even with that solved, the number of jobs also affects our AWS bill 12:09:49 <jki> and I would like to manage that according to our actual needs, that's why I was challenging the expansion 12:10:01 <arisut> But still isar-cip images need to be tested with the respective linux-cip kernel 12:10:32 <jki> sure, then may drop some other jobs we no longer need after adding your new ones? 12:10:34 <arisut> As we decided same as Debian is doing 12:11:42 <jki> again, let's resume the ML discussion thread, I can follow up after this meeting 12:11:45 <arisut> Buster with 4.19.y-cip bullseye with 5.10.y-cip and bookworm with 6.1.y-cip 12:11:50 <arisut> Ok 12:11:56 <jki> 2. draft press release about 6.1-cip (jan) 12:12:10 <arisut> Also one question I couldn't understand it, sorry 12:12:13 <jki> I started to write, will hopefully send something to the members list later 12:12:41 <jki> yes? do you have one more question? 12:13:28 <arisut> One question you asked in the ML, i replayed to that that I couldn't understand 12:13:43 <jki> ok, will check that again 12:13:48 <arisut> Thanks 12:14:01 <jki> then let's move on 12:14:09 <jki> #topic Kernel maintenance updates 12:14:36 <jki> [uli] "reviewing 6.1.45" 12:14:36 <masami> This week reported 2 new CVEs and 19 updated CVEs. 12:14:41 <pave1> I did reviews, 6.1.44 and renesas patches. 12:15:09 <masami> to all: thank you for answering to Dinesh-san's email. 12:15:46 <pave1> Speaking about security... 12:16:04 <pave1> I don't believe CVEs are too useful for us. 12:16:25 <pave1> Not everything security related has a CVE, and not nearly all CVEs are security problems that affects us. 12:17:15 <pave1> I wonder if we could push security team to watch CVEs, and talk to us if there's something that we really need to care about? 12:17:35 <pave1> I guess they already have to watc the CVEs for everything non-kernel...? 12:18:14 <jki> to my understanding, the security team is not yet in execution mode, rather in gap analysis and process definition mode 12:18:45 <jki> there will likely be more todos as outcomes of their analysis and assessor discussions 12:18:45 <pave1> May be true. 12:18:46 <masami> I heard from security team creating a script to watch CVEs for packages 12:19:10 <jki> ...and pull from Debian DB? 12:19:17 <jki> all that is done by Debian already 12:19:38 <jki> for their packages, not for our own ones, including the CIP kernel 12:19:44 <masami> yes. it may get CVE information from debian. 12:20:03 <pave1> But should we push the process in that direction? Security team watches for security bugs, if we have a security bug they don't like, they talk to us? 12:20:15 <masami> and cip-kernel-sec 12:20:29 <pave1> And we keep the cip-kernel-sec mostly up-to-date, so they can see our status. 12:21:12 <jki> but we need to look at patches ourselves anyway 12:21:24 <pave1> Yes, we do. 12:21:32 <jki> if some happen to relate to CVEs does not really matter 12:21:45 <pave1> But the patches we see may or may not have CVE annotation. 12:21:51 <jki> right 12:22:08 <jki> that must be clear in the process documentation that the kernel team is not CVE-driven 12:22:20 <pave1> Right. 12:22:20 <jki> CVEs are by-products 12:22:36 <jki> no difference to mainline here 12:23:13 <pave1> Reading CVEs needs special skills. I assume security team has them, and I'd like them to watch the CVE stream for "this needs fixing for CIP" bugs. 12:24:33 <jki> which should generally result in, "yes, we already know" 12:24:57 <jki> concretely: how is the situation beyond 6.1 for recent Downfalls & Co.? 12:25:12 <jki> are there / will there be backports to older kernels? 12:25:27 <pave1> Yes. But that means we don't need to go through CVE feeds and don't need to have processes for that :-). 12:26:47 <jki> ok, getting that - did we ever spot information in CVEs that weren't already in the stable patch stream? 12:27:20 <pave1> 6.1.44 has patches for "Gather Data Sampling" and they were queued for 4.14 and up. 12:28:01 <pave1> 6.1.44 has patches for "Speculative RAS Overflo\ 12:28:09 <pave1> w mitigation" and they are for 5.10 and up. 12:28:49 <pave1> jki: Not really. We get the information from stable, not CVEs. Not that we are watching CVEs too closely. 12:29:10 <pave1> "Gather Data Sampling" is the Intel problem 12:29:23 <pave1> "Speculative RAS Overflow" is the AMD problem, AFAICT. 12:30:00 <jki> well, masami is reporting here every week - are you reading up details of the CVEs as well? 12:31:20 <masami> yes. if nvd or other source have vulneratiblity details I read it. 12:32:11 <pave1> I try to go through his emails, yes. (Still need to go through today's one). If something looks interesting, I sometimes do investigate, but that may happen once in two months or so. 12:32:45 <jki> then, would that time elsewhere be invested with even better efficiency? 12:32:49 <jki> and where? 12:32:58 <jki> asking openly 12:34:30 <pave1> I'm quite happy how it currently works. But I'd hate to have to specify formal rules for CVE investigation. 12:35:16 <jki> well, we have to describe how we work so that others can understand it without having to do it themselves 12:35:47 <jki> doesn't mean that we are bound to only work like that - to my understanding 12:37:29 <jki> i think this is something to discuss, not only with the security team, but they also need to present it to the assessors 12:38:36 <pave1> My preffered solution would be "security team looks at CVEs". (And we make some informal effort to "already know" if something obvious pops up). 12:39:44 <jki> then follow up on Dinesh email on that point so that also other can see and comment 12:40:06 <jki> including kernel team members on leave right now 12:40:26 <jki> ok, anything else on this topic? 12:40:49 <jki> 5 12:40:51 <jki> 4 12:40:52 <jki> 3 12:40:53 <pave1> Yep, will follow up in the email. 12:40:56 <jki> 2 12:40:58 <jki> 1 12:41:01 <jki> #topic Kernel release status 12:41:06 <jki> 4.4 12:41:25 <pave1> uli released 4.4-cip, I follwed up with 4.4-cip-rt. Should be ok. 12:41:32 <jki> 4.19 12:42:08 <pave1> 4.19-cip-rt was released, based on slightly old versions. 12:42:33 <jki> older than latest plain cip? 12:42:42 <jki> nope 12:42:57 <pave1> Not that old, but not completely fresh either. 12:43:21 <jki> yeah, a new release for base 4.19-cip should be due soon 12:43:25 <jki> ok 12:43:27 <jki> 5.10 12:43:46 <pave1> -rt is due in September, we should be ok. 12:44:06 <jki> vanilla should come soon 12:44:10 <jki> 6.1 12:44:47 <jki> looks all recent 12:45:00 <jki> ok... 12:45:03 <jki> #topic Kernel testing 12:45:03 <pave1> I'll need to check, I don't think I have suitable -rt, 6.1-rt is due soon. 12:45:20 <jki> ok, thanks 12:45:28 <jki> anything for testing today? 12:45:59 <arisut> only the isar-cip patch and discussion 12:46:03 <arisut> From me 12:46:09 <arisut> patersonc: ? 12:46:16 <jki> right, we already had that above 12:46:30 <jki> I think he is on leave 12:46:39 <arisut> Ok 12:47:03 <jki> then... moving on? 12:47:07 <jki> 5 12:47:08 <jki> 4 12:47:10 <jki> 3 12:47:11 <sietze> I realized to get some tests sent to our SQUAD staging instance 12:47:11 <jki> 2 12:47:26 <jki> ah, cool! 12:47:47 <sietze> Note sure if anybody is interested in that; this is how it currently looks like: http://squad.ciplatform.org:8000/cip-kernel/linux-cip/build/6.1.38-cip1_093191f30/ 12:48:12 <sietze> We're still testing though and we still need to get https going 12:48:48 <jki> what will be the key benefits in the end when everything works? 12:49:46 <jki> single page summary? those compare features? 12:50:14 <sietze> Yes, better test overview, test results vs time, abilitiy to include other tests than kernel tests 12:50:31 <jki> ok, great 12:51:46 <jki> ok - anything else? 12:51:59 <jki> 3 12:52:01 <jki> 2 12:52:03 <jki> 1 12:52:07 <jki> #topic AOB 12:52:48 <jki> question: do you have options to host your cip releases as tarballs somewhere? 12:53:01 <jki> would that be useful beyond the CI case? 12:54:59 <jki> ok, will try to ask that also on the mailing list as follow up 12:55:13 <jki> other topic: I'm off next week, who can take over? 12:55:43 <pave1> I can take over, I guess. 12:55:52 <jki> pavel: thanks! 12:55:58 <pave1> You are welcome :-). 12:56:03 <jki> anything else for today? 12:56:58 <jki> 3 12:57:00 <jki> 2 12:57:02 <jki> 1 12:57:03 <jki> #endmeeting