15:00:37 <pwhalen> #startmeeting Fedora ARM and AArch64 Status Meeting
15:00:38 <zodbot> Meeting started Tue Dec 15 15:00:37 2020 UTC.
15:00:38 <zodbot> This meeting is logged and archived in a public location.
15:00:38 <zodbot> The chair is pwhalen. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:00:38 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
15:00:38 <zodbot> The meeting name has been set to 'fedora_arm_and_aarch64_status_meeting'
15:00:38 <pwhalen> #chair pwhalen pbrobinson jlinton coremodule
15:00:38 <pwhalen> #topic roll call
15:00:38 <zodbot> Current chairs: coremodule jlinton pbrobinson pwhalen
15:00:46 * pbrobinson is here
15:01:31 * jlinton waves
15:03:39 <pwhalen> howdy gents
15:03:48 <pwhalen> #topic 1) ==== Userspace Status  ====
15:04:00 <pwhalen> Any new user space issues?
15:04:05 <pbrobinson> We should have firefox back with the next build
15:04:10 <pwhalen> Looks like firefox is trying..
15:04:33 <pwhalen> been a few days of failed builds, but i see a bunch going now
15:05:28 <pbrobinson> yes, but I don't think those issues were aarch64 specific
15:05:44 <pwhalen> right, the issues looked to cover all the arches
15:06:01 <pwhalen> #info No new issues reported.
15:06:10 <pwhalen> #topic 2) ==== Kernel Status ====
15:06:44 <pwhalen> I dont think we got an RC7 build last week..
15:07:09 <pbrobinson> there's issues with the kernel on aarch64 and gcc11 which landed in rawhide in the last few days
15:08:00 <pbrobinson> jforbes just sent a status to the kernel list on that, he'll do 5.10 builds for f33 from the stabilisation branch and there's a 5.10 test week on the week of Jan 4th
15:08:34 <jforbes> Yup, gcc 11 fails specifically on aarch64
15:08:43 <pwhalen> ok, looks like aarch64 and x86_64 failed today so hopefully we see a new attempt soon. I had no issues with RC6
15:09:03 <jforbes> There will be no further attempt with 5.10 for rawhide
15:09:24 <pwhalen> ok, is there a bz for aarch64?
15:09:42 <jforbes> X86 fail was a known koji issue, it builds fine, aarch64 has not built since last week (when gcc 11 was put in)
15:10:06 <pbrobinson> jlinton: are you aware of any issues around aarch64 kernel and gcc 11?
15:10:13 <jforbes> Also, how did the compiler get a major revision right before everyone went on vacation, without verifying it builds key things like kernel?
15:10:21 <jlinton> I haven't been building it with gcc11, my build machines are f33.
15:10:41 <jforbes> https://bugzilla.redhat.com/show_bug.cgi?id=1907945 is the bug, and yes, I verified the same srpm for 5.10 does build in koji against f33
15:11:15 <pwhalen> thanks jforbes
15:11:31 <jforbes> I also pinged Jakub on this yesterday, he got back to me, but is on PTO and wants me to do some debugging on my end, which I will try to get to when I grab access to an aarch64 host capable of building the kernel
15:11:51 <jlinton> Although, i've seen that error with gcc10. I was having that problem with the koji builders
15:12:17 <jforbes> But I am also doing half days, so it might be later this week.  Seems getting 5.10 into stabilization is more of a priority at this point
15:12:28 <pbrobinson> pwhalen: do you have a system you can give jforbes access to?
15:12:29 <pwhalen> #info Issues building the latest 5.10 kernel with GCC11 - https://bugzilla.redhat.com/show_bug.cgi?id=1907945
15:13:00 <pwhalen> Should be easy to get a host from beaker, sure
15:13:17 <jforbes> I thought we had a Fedora process for getting host access?
15:13:55 <pwhalen> there is, just looking for that
15:14:22 <jforbes> https://koji.fedoraproject.org/koji/taskinfo?taskID=57446012 is the scratch build I did yesterday for aarch64 on f33, if anyone needs it
15:14:33 <pwhalen> there are these machines - https://fedoraproject.org/wiki/Test_Machine_Resources_For_Package_Maintainers
15:14:53 <pbrobinson> jforbes: I'm not sure if those hosts are back up post DC move, and there's likely faster ones
15:15:08 <pbrobinson> but pwhalen would know of that status than me TBH
15:15:46 <jforbes> I will certainly take faster.  Only real host I have around here with the disk space to build a kernel is an 8GB pi 4, but I am guessing it would take ages to build
15:15:56 <pwhalen> you might be right, I can't recall if those are back. jforbes I'll get you a faster machine and pm the details
15:16:09 <jforbes> Thanks pwhalen
15:16:09 <pwhalen> yea, dont do that :)
15:16:42 <pwhalen> #info Please test and report any issues to the list or #fedora-arm.
15:16:57 <pwhalen> ok, thanks again jforbes.
15:17:04 <pwhalen> #topic 3) ==== Bootloader Status ====
15:19:09 <pwhalen> I don't see anything has changed here, javier submitted a patch for U-Boot (efi boot issue - BZ#1904242)
15:19:16 <pbrobinson> so I worked out the issue with the u-boot bits, it was due to the patches for the rpi-400..... :-/
15:19:20 <pwhalen> Anything else?
15:19:25 <pwhalen> of course :)
15:19:44 <pbrobinson> reported that upstream, the maintainer has reproduced so we should have a fix for the next rev of that
15:20:01 <pbrobinson> I would expect a new u-boot RC today(ish)
15:20:17 <pwhalen> ok, thanks
15:20:31 <pwhalen> #topic 4) ==== Fedora 34 ====
15:20:41 <pwhalen> #info OpenQA Testing - https://openqa.fedoraproject.org/tests/overview?distri=fedora&version=Rawhide&build=Fedora-Rawhide-20201215.n.0&groupid=5
15:21:08 <pwhalen> not looking too bad
15:21:22 <pwhalen> and largely the same as x86_64.
15:21:35 <defolos> on the topic of openQA: I have ordered some hardware and will try to setup bare metal tests with whatever hardware I can get my hands on
15:21:48 <defolos> cannot promise I'll get it done over the christmas break
15:22:15 <pwhalen> the yellow bits are mostly from AVC's
15:22:38 <pwhalen> defolos: very interested, keep us posted
15:22:49 <defolos> I will
15:23:08 <defolos> I have a Pi 4 8GB and a pinebook pro atm
15:24:11 <defolos> not sure if these will even launch too well?
15:24:16 <pbrobinson> I'm going to be working on the PBP on Friday
15:24:24 <jforbes> BTW, I just got the Pi 4 8GB this past weekend, and was pleasantly surprised at just how easy it was to get working with F33
15:24:32 <defolos> 🎉
15:24:42 <defolos> that's great news!
15:24:51 <pbrobinson> jforbes: the 8Gb one has a bunch of quirks around display and USB, but for headless it's mostly OK
15:25:47 <jforbes> pbrobinson: Right now it is running Fedora server, though connected to HDMI until I move it into its intended purpose. Only USB I used was a keyboard though
15:26:37 <pwhalen> #topic 5)  == Open Floor ==
15:27:00 <jforbes> Still waiting for a host to become available at a reasonable price that I can use for kernel builds/debugging/testing
15:27:30 <pwhalen> I believe this will be the last meeting of 2020, I think most people are off for the rest of the year.
15:27:52 <defolos> jforbes: the rockpro64 has quite the punch and is not super expensive
15:29:11 <pwhalen> #info Fedora ARM meetings to resume in 2021 - next meeting scheduled for January 6th, 2021.
15:29:30 <jforbes> defolos: Oh, I am willing to pay more than that, but was looking for something I with at least 16GB of memory that I can plug a decent SSD into and will build a kernel rpm in 2 hours or less
15:29:57 <jlinton> honeycomb
15:29:59 <defolos> jforbes: honeycomb lx2
15:30:10 <defolos> but it's going to set you back about 700 bucks
15:30:35 <defolos> I sorta-kinda-know a guy at arm who loves his for doing all kinds of tinkering with it
15:30:37 <jlinton> oh, actually i don't think it will do the .rpm in under 2 hours.
15:30:54 <jlinton> it can do a kernel, but the rpm takes a lot longer.... tx2?
15:31:14 <jlinton> the fedora configed kernel builds in something like 4 mins on the tx2.
15:31:38 <jlinton> the rpm is more like 30 mins.
15:32:12 <jforbes> Wow, that actually matches my current x86_64 build system here
15:32:21 <jlinton> 256 threads.
15:32:24 <jlinton> :)
15:32:35 <defolos> 🤯
15:33:26 <pwhalen> Thanks for joining all, see you in 2021.
15:33:30 <pwhalen> #endmeeting