15:00:07 #startmeeting Fedora ARM and AArch64 Status Meeting 15:00:07 Meeting started Tue Sep 20 15:00:07 2016 UTC. The chair is pwhalen. Information about MeetBot at http://wiki.debian.org/MeetBot. 15:00:07 Useful Commands: #action #agreed #halp #info #idea #link #topic. 15:00:07 The meeting name has been set to 'fedora_arm_and_aarch64_status_meeting' 15:00:07 #chair pwhalen pbrobinson dgilmore hrw dmarlin yselkowitz jonmasters ahs3 msalter 15:00:07 Current chairs: ahs3 dgilmore dmarlin hrw jonmasters msalter pbrobinson pwhalen yselkowitz 15:00:18 morning folks, whos here today? 15:00:52 Good morning 15:01:29 * pbrobinson o/ 15:02:37 thanks for coming, lets get started.. 15:02:39 #topic 1) ==== Userspace Status ==== 15:03:14 so update pushed for libjpeg-turbo issue on aarch64, please test (should fix the VNC install issue) 15:03:32 * rric o/ 15:03:34 ill test that today 15:03:39 https://bodhi.fedoraproject.org/updates/FEDORA-2016-4d0198119a 15:03:58 #info libjpeg-turbo fix needs testing https://bodhi.fedoraproject.org/updates/FEDORA-2016-4d0198119a 15:05:06 any build issues that need looking at? 15:05:08 Is there a nightly installer? This one is empty: http://dl.fedoraproject.org/pub/fedora-secondary/development/25/Server/aarch64/iso/ 15:05:10 there's been a few other fixes to minor bits here and there but I think it looks reasonable 15:05:27 rric: failing at the moment, should hopefully be fixed tomorrow 15:05:43 it's due to a selinux-policy issue, across most arches 15:06:08 when f25 updates go out today it should resolve the issue (hopefully) for tomorrow 15:06:20 hopefully 15:06:25 the kernel of this is too old: https://ftp-stud.hs-esslingen.de/pub/fedora-secondary/releases/test/25_Alpha/Server/aarch64/iso/ 15:07:00 rric: there's been nightlies since then with much newer kernels 15:07:16 rric, heres some older nightlies - https://arm.koji.fedoraproject.org/compose/branched/ 15:07:31 #info please test selinux-policy update https://bodhi.fedoraproject.org/updates/FEDORA-2016-5f88bebc7c 15:07:48 pbrobinson, ok, thanks for the link 15:08:26 pbrobinson, do you use the installer's image with pxe boot? 15:08:35 rric: yes 15:08:44 rric: with grub 15:08:58 so the grub from the installer's image? 15:09:23 and with serial console? 15:09:27 rric: can you ask in #fedora-arm as PXE install how to is off topic for the meeting 15:09:31 ok 15:09:46 rric: and you can do it there anytime :) 15:10:00 anyone else have any userspace issues? 15:10:17 #topic 2) ==== Kernel Status ==== 15:11:01 we face an upstream bug in page_alloc 15:11:09 #info kernel-4.8.0-0.rc7.git0.1.fc25 available for testing - http://koji.fedoraproject.org/koji/buildinfo?buildID=802366 15:11:21 rric: upstream thread? 15:11:27 1sec 15:11:35 rric: kernel? userspace? 15:11:41 kernel 15:12:31 #info RC7 looks better, armhfp tested on Arndale, Banana Pi, Wandboard. 15:12:40 https://marc.info/?l=linux-arm-kernel&m=147436995129596&w=2 15:13:08 Mark triggered it with hackbench 15:13:20 my kernel stops while loading initrd 15:13:21 rric: is that purely for ThunderX or a generic aarch64 problem? 15:13:26 #info RC7 AArch64 build - http://koji.fedoraproject.org/koji/buildinfo?buildID=802368 15:13:28 generic 15:13:33 also seen on Seattle 15:14:02 kernel BUG at mm/page_alloc.c:1844! 15:16:18 rric: so we're waiting on upstream resolution? 15:16:41 pbrobinson: yes, I am still root causing it, the numa patches are related but didn't fix it yet 15:16:46 rric: has that been reproduced with a fedora kernel? 15:17:05 yes, esp. with it's config 15:17:15 ack. thx. 15:18:08 it was also triggered with a non-numa kernel 15:19:34 any other kernel issues to discuss? 15:20:37 none, other than please test, we're 1 week from Beta freeze so if we want it in Beta it basically needs to land in the next couple of days 15:21:37 #info Please test the latest kernels on your device! We're 1 week from Beta freeze so if we want it in Beta it needs to land in the next couple of days. 15:21:53 I think I have a power gov config change, but I'm verifying it 15:22:28 #topic 3) ==== Bootloader Status ==== 15:22:35 jlinton_: OK, I thought we used the same as generic, hence all other arches 15:23:08 so u-boot 2016.09 GA should go stable with today's updates push (in progress) and land in tomorrow's images so please test 15:24:48 #info U-boot 2016.09 GA should go stable with today's updates push (in progress) and land in tomorrow's images. Please test your device! 15:25:21 #topic 4) ==== F25 Beta ==== 15:25:31 so we freeze one week today 15:25:48 I think over all we're looking pretty reasonable over both ARMv7 and aarch64 15:25:52 #info 2016-09-27 Beta Freeze 15:26:11 #info 2016-10-11 Beta Release 15:26:13 I have one feature I plan on landing later this week, more on that one next week. 15:26:22 #info blocker list is at https://qa.fedoraproject.org/blockerbugs/milestone/25/beta/buglist 15:26:52 pbrobinson: what features are you planning to land? 15:26:52 armhfp, we have two Beta blockers on initial-setup. if you're testing, make sure to remove the root password so you can log in. 15:26:56 so probably the biggest issue ATM is the initial-setup bugs. The maintainer is currently working on them now. so hopefully fix RSN 15:27:18 cov: rel-eng related change for ARMv7 so not affecting aarch64 15:27:27 okay 15:28:18 any other concerns for Beta? 15:28:25 PCI MSI support (ACPI IORT) patches are queued for 4.9 if you're willing to take a backport 15:28:26 firefox 15:28:28 https://wiki.codeaurora.org/xwiki/bin/Linux+Kernel/WebHome#HACPIIORT2CGICITS2CPCIEMSI 15:28:47 cov: they're a bit large for my liking TBH 15:29:14 impact is that PCI cards either fall back to legacy interrupts or don't work at all 15:29:14 agree, firefox would be nice to get fixed. i checked some other browsers and all have issues (midori, otter browser) 15:29:26 jlinton_: did you get a change to test v49 at all? new build just recently (also might be affected by the libjpeg-turbo bug? ) 15:29:44 good point pbrobinson , will test as well 15:29:47 I retested 48 15:30:03 cov: * some PCI cards correct? Also ACPI isn't default 15:31:15 pbrobinson: some drivers can fall back to legacy interrupts. other drivers can't and just give up 15:31:23 jlinton_: so 49 is on it's way to f25 testing, will be there in a few hours, so will be stable before Beta freeze 15:31:25 I'm not sure what you mean by ACPI isn't default 15:31:49 Ok, I will test that version, I wasn't sure if it was going to be in F25 15:31:50 cov: you have to explicitly put ACPI=force on the command line, else it defaults to DT 15:31:56 no 15:32:04 that's not how it works 15:32:26 at least not how it works on the systems I have access to 15:32:29 cov: well on mustang/seattle it's the way it works, I don't have anything else to test against 15:32:34 If there isn't a DT it uses ACPI.. 15:32:58 So, a few of the server machines don't have DT 15:33:04 isn't a *stub* DT? there's always at least a stub DT to satisfy the boot protocol 15:33:18 Yah, stub 15:34:20 on the systems I have access to such as QDF2432, there isn't a full DT, just the stub pointing to ACPI 15:36:36 cov: well I don't see anything against v5 of the patch (last on the linux-pci list) saying it's been pulled and I have no particular urge (or really the time) to pull a 1200+ line patchset 15:37:21 pbrobinson: it's in Marc Zyngier's ARM Platforms tree on the irq/irqchip-4.9 branch 15:37:26 also in -next 15:37:50 pbrobinson, jlinton_ no change with .49 and new libjpeg-turbo. will update the bz 15:38:11 http://git.kernel.org/cgit/linux/kernel/git/next/linux-next.git/log/?qt=author&q=nowicki 15:38:35 pwhalen: That was fast, took my machine maybe an hour to do a bt with the debug info's 15:40:52 jlinton_, bt in progress 15:41:59 cov; I'm really not convinced of the feature vs my time/sanity means we can't wait for 4.9.... you'll need to _REALLY_ convince me on this 15:43:56 #topic 5) == Open Floor == 15:44:08 anything else for today? 15:44:21 pbrobinson: is there some way to make it take less time for you, such as I submit the change to the mailing list after testing it across various hardware? 15:44:54 cov: one mailing list vs the other doesn't really save me time 15:45:11 so what takes time? testing? 15:45:52 cov: I basically have zero time free this week to deal (already negative) with anything kernel as I have a 100 things else to deal with and I'm travelling the next two weeks. 15:46:10 cov: Linaro connect next week, Linux Con Europe the following one 15:47:48 definitely busy. is there some chance another maintainer might be able to help out? 15:49:38 cov: it's not just that, I'm yet to be convinced the complete value over waiting for 4.9 to land 15:50:57 will the fedora 25 kernel rebase onto 4.9? 15:51:16 perhaps you gents can continue in #fedora-arm, lets wrap this up 15:51:28 anything else for today? 15:51:54 #endmeeting