15:00:06 #startmeeting Fedora ARM and AArch64 Status Meeting 15:00:06 Meeting started Tue May 3 15:00:06 2016 UTC. The chair is pwhalen. Information about MeetBot at http://wiki.debian.org/MeetBot. 15:00:06 Useful Commands: #action #agreed #halp #info #idea #link #topic. 15:00:06 The meeting name has been set to 'fedora_arm_and_aarch64_status_meeting' 15:00:06 #chair pwhalen bconoboy pbrobinson dgilmore hrw jsmith dmarlin yselkowitz 15:00:06 Current chairs: bconoboy dgilmore dmarlin hrw jsmith pbrobinson pwhalen yselkowitz 15:00:16 morning folks, whos here today? 15:00:39 O/ 15:01:25 * pbrobinson is sort of here, just closing out another meeting 15:01:27 morning/evening herw 15:01:42 Afternoon ;-) 15:01:49 doh :) 15:01:55 * pbrobinson now here 15:02:23 nice, lets get started 15:02:26 #topic 1) ==== Package Status & Issues ==== 15:02:45 .hello linuxmodder 15:02:46 linuxmodder: linuxmodder 'Corey W Sheldon' 15:02:53 Pwhalen: we have sunny spring. Not 'just one feet of snow' winter like Canada has. 15:03:13 firefox stack is fixed, I literally have very little to report. 15:03:24 libreoffice is probably our biggest now 15:03:34 herw, we are finally getting some nice weather.. 15:04:01 #info firefox stack is fixed. 15:04:22 #info libreoffice currently biggest blocker. 15:04:41 good news. ok, anyone else have any other packages to mention? 15:04:47 there will be some corner case packages but there's nothing major, likely a few enhancements between now and GA 15:05:02 * jonmasters in 15:05:30 good morning jonmasters 15:05:36 morning Paul! 15:05:38 #topic 2) ==== Kernel Status ==== 15:06:06 morning also pbrobinson - already seen him on the twitters ;) 15:06:13 ive admittedly been focused on f24, has anyone done any testing with the latest 4.6? 15:06:25 Works fine on mustang. 15:06:30 I think 4.5 is looking good for beta. 15:06:45 jonmasters: hey 15:06:47 4.5.2 boots on pine64+ 15:06:58 doing some testing of 4.6 for things like PCIe 15:07:12 Jonmasters: can you check fedora on cello? 15:07:23 i saw you crazy guys working on the pine64 this past weekend.. great news 15:07:27 hoping that at least the basic ability to enumerate PCI (but possibly without quirks - so Moonshot and Mustang will need a small patch) 15:07:28 * pwhalen needs one 15:07:35 ...in 4.7. Not 4.6. 15:08:07 Pwhalen: needs some blobs, some integration. 15:08:11 On Cello, the issue was that the PCIe 100MHz clock wasn't being enabled. Waiting on whether we can work around with firmware but it might be better to wait for new boards (couple weeks) 15:08:19 jonmasters: OK, so the Cello pci-e issue is not a HW blocking and it's off to prod? 15:08:29 ouch, ok.. 15:08:31 jonmasters: ah! Oops! 15:08:45 pbrobinson: waiting to find out what can be done in fw - current boards have a hw hack we've been applying 15:09:53 Hope to grab cello before July. Would be great dev platform for work out of home. 15:09:56 OK, so people can please test 4.6 and report issues. I've seen one on BBone but not had time to test wider 15:10:01 without PCIe clock, couple of devices show as there but can't be used hence I give them some credit that it wasn't so easy to QE before these first boards left the line - they rushed them out to get us to look at them 15:10:02 #info kernel-4.5.2-302.fc24 looks good for beta on all hardware (aarch64 & armhfp). 15:10:32 #info Please test 4.6 and report issues. 15:10:50 pbrobinson, what happened on the bbone? black? 15:11:09 on 4.6, I'm lighting various fires on getting upstream all of the remaining core enablement pieces. Have a document I am updating with upstream status for Linaro folks 15:11:11 pwhalen: some issue with crypto stack, I need to look gurther 15:11:14 further 15:11:37 (will share that document with folks who need it) 15:11:48 Jonmasters: the red/yellow/green one? 15:11:52 overall this week I'll be moving all my devices to 4.6 kernel 15:11:53 herw: yup 15:11:57 ok, i hit some memory issues during update on the bb white. been looking at that since i know others have the black 15:12:14 jonmasters, excellent 15:12:42 so anything else kernel related, jonmasters I'll ping you OOB on some bits 15:12:44 ok, any other kernel news? 15:12:46 pwhalen: I thank pbrobinson for reminding me how unacceptable it is for ARM to not have upstreamed this stuff a million years ago 15:12:52 pbrobinson: ok 15:13:15 tracking some core enablement for v8 in 4.6 but don't need to go into it here 15:13:34 oh, I expect NUMA to "just work" in 4.6 if anyone did have a NUMA v8 box 15:13:53 Thunderx? 15:13:57 jonmasters: the patches aren't upstream in 4.6, maybe you mean 4.7? 15:14:18 yeah - NUMA doesn't "work" but the system will boot and work sufficiently without it, which is a milestone 15:14:29 sorry, I should have said "just boot" not "just work" 15:15:19 Ok, u-boot? 15:15:24 thunderx NUMA works fine with posted patches (not yet upstream) -- re-uses x86 code so does seem to "just work"... 15:15:32 ahs3: agree 15:16:10 #topic 3) ==== Bootloader Status ==== 15:16:43 seattle and others just pretend to be a single NUMA node (been using that for daily driver) 15:16:55 * jonmasters on the hook to investigate status of U-Boot support for 32-bit VMs on 64-bit (will sync with pbrobinson once I have an update) 15:17:07 Pine64 needs binary blobs and piece of opensource code to get u-boot working. Good news: Fedora u-boot works (after blobbing). 15:17:07 so firstly I'd like to congratulate herw for the fight he's started on bootloaders on G+ ... it's hilarious! 15:17:30 herw: yep, I'm working on a proper image this week along side with other bits 15:17:31 sounds fun 15:17:32 url? 15:17:47 yselkowitz: check his profile 15:17:56 https://plus.google.com/+MarcinJuszkiewicz/posts/J79qhndV6FY 15:18:00 yselkowitz: or url for what? 15:18:21 in other (more important) news 15:18:45 Pbrobinson: can we build atf part as part of u-boot or extra package? 15:18:53 the grub2 issue is being actively worked on, won't block beta on it but we'll have it fixed in time for GA 15:19:27 herw: I'm looking at it, might need to get legal involved, we'll sort something out one way or another 15:19:33 Ok 15:19:39 herw: that is hilarious - too many of the vendors just don't get that it's better to have a small amount of first stage flash on board or even whole thing to remove silly SD card hacks. They just don't get it. I tried to make that part of the CE spec for 96boards but failed to persuade people :( 15:20:13 May blog about it later in week. 15:20:19 jonmasters: it would be good to get it mandated in 96b CE edition when there's a refresh 15:20:31 the device manufacturers just want to make hacks, not real machines. And yea pbrobinson I am trying :) 15:20:44 brb 15:21:04 jonmasters: I can understand on something like pin64 where it's really cheap, but none of the 96boards are that cheap 15:21:13 so other than grub2 mentioned above 15:21:40 if people could test u-boot 2016.05rc3 that would be fab, likely a new version coming very soon 15:22:07 #info the grub2 issue is being actively worked on, won't block beta on it but we'll have it fixed in time for GA 15:22:20 #info test u-boot 2016.05rc3 that would be fab, likely a new version coming very soon 15:22:49 ok, anything else for bootloaders? 15:23:18 #topic 4) ==== F24 Beta ==== 15:23:30 #info Current Test results (Beta 1.4) 15:23:30 #link https://fedoraproject.org/wiki/Test_Results:Current_Summary 15:23:32 so aarch64 beta is building right now 15:23:53 First run I did missed some bits, like cloud stacks and other fixes 15:24:18 Cloud... Need to test them in Openstack. 15:24:29 herw: please do 15:24:38 herw: let me know if you need any details 15:24:45 First need to get it work 15:25:39 so I'll send an announce out as soon as it's ready 15:25:48 back 15:26:11 pbrobinson, let me know prior and i'll do a page for tracking.. will it be named beta 1.4? 15:26:28 pwhalen: at the moment 1.2 15:26:56 ok, wasnt sure if we were following pa. sounds good 15:27:54 so for armhfp, everything looks reasonable. we have a couple of issues with work arounds.. it would be helpful for folks to test their favourite device and update the hardware support.. 15:28:08 ive covered what i have here. 15:28:15 pwhalen: packages sets we are, the exact version not currently, but we might get there ;-) 15:28:39 pbrobinson, right, ok 15:29:21 I plan to get some tech preview base images out later in the week for devices like Pine64, Jetson TX1, Geekbox etc 15:30:30 pbrobinson, excellent 15:30:43 anything else for beta? 15:31:09 #topic 5) == Open Floor == 15:32:16 so I know we're still on F-24, but is there features (other than pci-e support in ACPI jonmasters!) that people are interested in for F-25 15:32:40 as in working on themselves that might need help/assistance 15:33:00 I hope for u-boot spl for pine64 before f25 release. 15:34:01 pbrobinson: I'm pushing all of the v8 vendors saying I expect F25 to work out of the box or I'm going to mock them 15:34:17 Cool 15:34:18 hrw: you mean one you don't have to cat with TrustZone 15:34:27 Exactly that 15:34:30 jonmasters: I already so but then I'm not as high up as you 15:36:26 note there will be no mass rebuild in F-25 so it should be a calmer cycle (famous last words!) 15:36:52 thank goodness. 15:36:56 anything else for open floor? 15:37:00 Did someone know can other ethernet ports (than over USB one) work on mustang? 15:37:04 none from me 15:37:22 1GbE one I mean 15:37:51 hrww, good question 15:38:13 hrww: what do you mean 15:38:30 Mustang has three rj45 ports. 15:38:32 as in do you mean the 3 on board ports or something else 15:38:46 there are three different interfaces on mustang 15:38:50 I use "over USB one" as it worked since beginning 15:39:24 hrww: the wired ones should all work, one is 10G, but I use one of the wired ones for like ever. Without ACPI I think all of them now work 15:39:39 there's RGMII, SGMII, and XFI 15:40:01 the RGMIIs work fine out of the box, the others depend upon firmware version but work 15:40:12 (RGMII is 1g) 15:40:21 Ok, will add cables. 15:40:42 I also use 10G at home via a Mellanox PCIe card but I know you're using the slot :) 15:41:21 Jcm: send me card and switch? :-D 15:41:28 lol :) 15:41:41 the switches are cheap on eBay, shipping to you will cost more :) 15:42:06 anything else for the meeting? 15:42:16 Nope 15:42:20 if you're bored, see the Media Independent Interface article on wikipedia and look at those three terms I used for the details on how they work etc. 15:42:49 #endmeeting