15:00:42 <coremodule> #startmeeting Fedora ARM Working Group Meeting
15:00:42 <zodbot> Meeting started Tue May  9 15:00:42 2023 UTC.
15:00:42 <zodbot> This meeting is logged and archived in a public location.
15:00:42 <zodbot> The chair is coremodule. Information about MeetBot at https://fedoraproject.org/wiki/Zodbot#Meeting_Functions.
15:00:42 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
15:00:42 <zodbot> The meeting name has been set to 'fedora_arm_working_group_meeting'
15:00:42 <coremodule> #chair pwhalen pbrobinson jlinton coremodule
15:00:42 <zodbot> Current chairs: coremodule jlinton pbrobinson pwhalen
15:00:42 <coremodule> #topic roll call
15:01:03 <coremodule> hello all, who is around today for an arm meeting?
15:01:28 * pwhalen is here
15:01:34 * jlinton waves
15:01:35 <pwhalen> Good morning coremodule
15:02:30 <coremodule> good morning!
15:02:46 <coremodule> shall we? I don't have anything other than the usual for today.
15:02:55 <coremodule> #topic 1) ==== Stable Releases ====
15:03:16 <jlinton> The F38 release images won't install on ampere.
15:03:38 <jlinton> Looks like a systemd bug, with tmpfs? but I've not patched it to see if the problem goes away.
15:05:26 <jlinton> So this is largely a heads up since I've not opened a defect until I understand why that machine at the moment seems to be the only one with the problem (although its multiple ampere altra models, and not just me reproducing it)
15:05:52 <pwhalen> :( I had some issues trying to install as well, but not gone back to see why
15:06:13 <jlinton> There is an upstream systemd bug with the same signature https://github.com/systemd/systemd/issues/27436 but doesn't make sense why it only affects that machine.
15:06:16 <coremodule> #info F38 release images won't install on Ampere hardware, specifically the Ampere Altra.
15:07:40 <jlinton> Also, while looking at the logs there are a ton of selinux issues booting the install ISOs
15:08:09 <jlinton> And to be clear none of them seem to be causing the ampere problem.
15:08:29 <pwhalen> RIght, but not a good look.
15:10:50 <pwhalen> Let's try to get a bug filed, coremodule do you use the enterprise hardware for testing?
15:11:29 <coremodule> no, I don't, just desktop stuff that I have physically. jlinton, can you dig deeper and file a bug?
15:11:49 <coremodule> are there ampere systems on beaker?
15:11:50 <jlinton> Yah, thats the plan for the next day or so.
15:12:24 <coremodule> thanks jlinton
15:12:25 <jlinton> I think there are, mines across the "pond" so its not fun to work with. But I have HW.
15:12:26 <pwhalen> coremodule: yes, we have a bunch of different ampere hw, looks like quite a few new boxes
15:13:23 <jlinton> Also, before I forget, kdump is broken in f37 and f38 now too because of the ZBOOT stuff.
15:13:50 <jlinton> kexec tools patches are on the ML
15:15:34 <coremodule> thanks for the update jlinton
15:15:56 <coremodule> anything else for stable?
15:16:28 * pwhalen has nothign
15:16:33 <coremodule> #topic 2) ==== Kernel Status ====
15:16:33 <coremodule> #info kernel-6.4.0-0.rc1.16.fc39
15:16:33 <coremodule> #link https://koji.fedoraproject.org/koji/buildinfo?buildID=2197254
15:16:44 <coremodule> 6.4 kernel has dropped
15:17:18 <coremodule> there is a 6.3 kernel test week ongoing right now
15:17:30 <coremodule> #info Kernel 6.3 Test Week is ongoing this week!
15:17:35 <coremodule> #link https://fedoraproject.org/wiki/Test_Day:2023-05-07_Kernel_6.3_Test_Week
15:17:52 <pwhalen> thanks, was looking for that too
15:18:17 <pwhalen> I will upgrade my hw here, post some results
15:19:18 <coremodule> thanks pwhalen
15:19:36 <coremodule> #topic 3) ==== Bootloader Status ====
15:19:37 <coremodule> #info uboot-tools-2023.04-1.fc38
15:19:37 <coremodule> #link https://koji.fedoraproject.org/koji/buildinfo?buildID=2182390
15:19:48 <coremodule> no changes to uboot, nothing of note here from me
15:21:47 <coremodule> #topic 4) ==== Fedora 39 status ===
15:21:47 <coremodule> #info OpenQA Testing
15:21:47 <coremodule> #info https://openqa.fedoraproject.org/tests/overview?distri=fedora&version=Rawhide&build=Fedora-Rawhide-20230509.n.0&groupid=5
15:22:26 <coremodule> I personally haven't spent any time looking at f39 since we launched f38. but a quick glance at openQA doesn't appear horrible.
15:23:02 <pwhalen> looks ok for where we are, agreed
15:23:14 <pwhalen> no huge fires :D
15:24:47 <coremodule> so far, so good!
15:24:49 <coremodule> #topic 5) ==== Open Floor ====
15:24:51 <jforbes> Ug, sorry. had other meetings. Thanks for mentioning test week. I expect 6.3.2 tomorrowish.
15:25:10 <coremodule> no worries jforbes!
15:25:11 <jlinton> retrace? Is it working for aarch64?
15:25:42 <jlinton> I was poking around on it yesterday and while its possible to see armv7 issues, aarch64 is missing from the dropdown (KEBKAC?)
15:25:59 <jlinton> and I pretty consistently remember seeing retrace rejections on aarch64 machines.
15:26:28 <jlinton> I think I have a 5 year old bug open about it somewhere too.
15:27:02 <jlinton> although given I can't find it at the moment maybe its time for another one.
15:27:37 <jforbes> I don't see it listed as an arch even if going outside of my regular kernel page
15:28:17 <jforbes> One more thing, relevant to upstream discussion, is anyone still using ThunderX?
15:28:40 <jlinton> I have a couple thunderX2's but no on the original.
15:28:47 <coremodule> I don't have one jforbes
15:29:00 <jlinton> There were some in beaker, did they get decomissioned?
15:29:31 <jforbes> There is a possible change to FORCE_MAX_ORDER for 4K pages, and the only thing that we know of that needs the current value is the original ThunderX
15:30:13 <pwhalen> we have some of both in beaker
15:31:33 <pwhalen> jforbes:  is this in 6.3?
15:33:50 <jforbes> pwhalen: it is 6.4
15:35:09 <pwhalen> OK thanks, I'll try 6.4 rc1 (if I can provision this host)
15:35:14 <jforbes> Upstream is trying to hide FORCE_MAX_ORDER behind if EXPERT for arm64 and drop the value to 10. Our current value is 13, though with other changes it is really 12.  Supposedly a driver for ThunderX is the only thing whcih is a problem
15:35:21 <jforbes> Oh, we are carrying a work around right now
15:36:15 <jforbes> Was thinking about dropping the work around and just going with the default to see if it breaks anything. Ard seems to think ThunderX is all, and "it is really more of a waste of electricity at this point"
15:37:36 <pwhalen> heh, we'll hear about it if it's an issue for someone.
15:37:55 <pwhalen> happy to test if you do
15:38:09 <jforbes> Right, it is early in the rc cycle.
15:40:14 <pwhalen> agreed.
15:41:55 <coremodule> thanks everyone. I'll close the meeting in 5...
15:41:57 <coremodule> 4...
15:41:58 <coremodule> 3...
15:42:00 <coremodule> 2...
15:42:02 <coremodule> 1...
15:42:05 <coremodule> #endmeeting