16:00:15 <pwhalen> #startmeeting Fedora ARM and AArch64 Status Meeting
16:00:15 <zodbot> Meeting started Tue Nov  8 16:00:15 2016 UTC.  The chair is pwhalen. Information about MeetBot at http://wiki.debian.org/MeetBot.
16:00:15 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
16:00:15 <zodbot> The meeting name has been set to 'fedora_arm_and_aarch64_status_meeting'
16:00:15 <pwhalen> #chair pwhalen pbrobinson dgilmore hrw dmarlin yselkowitz jonmasters ahs3 msalter
16:00:15 <zodbot> Current chairs: ahs3 dgilmore dmarlin hrw jonmasters msalter pbrobinson pwhalen yselkowitz
16:00:23 * pbrobinson o/
16:01:02 <pwhalen> morning folks, who's here today?
16:04:23 <pwhalen> #topic 1) ==== Userspace Status  ====
16:05:08 <pwhalen> hows everything looking? anyone aware of any issues?
16:05:22 <pbrobinson> nothing major here, from a rawhide perspective Power64 was merged into primary last week, there was a bit of fall out from that (not as clean as the aarch64 merge) but mostly settled out now
16:05:31 * jonmasters in
16:05:39 * pbrobinson waves at jcm
16:05:49 <jonmasters> morning/afternoon pbrobinson
16:06:09 <pbrobinson> from a F-25 perspective I'm not aware of any major userspace issues at the moment
16:06:18 <pbrobinson> on either aarch64 or ARMv7
16:06:36 <pbrobinson> I'm not sure if the firefox problem was got to the bottom of
16:06:43 <pwhalen> #info No current major issues on either AArch64 or ARMv7.
16:06:52 <pwhalen> no, i dont think there has been any change in the FF bug
16:07:33 <pwhalen> or the other browsers I filed bugs on. I'll check if theres been newer versions, update the bug.
16:07:39 <pwhalen> s
16:08:32 <pwhalen> #topic 2) ==== Kernel Status ====
16:09:18 <pwhalen> #info Latest Rawhide - kernel-4.9.0-0.rc4.git0.1.fc26
16:09:28 <pwhalen> #link http://koji.fedoraproject.org/koji/buildinfo?buildID=816006
16:10:09 <pbrobinson> so neither rric or cov appear to be about but the memory regions patch landed and the 4.8.6-300.fc25 will likely be the GA kernel
16:10:18 <pbrobinson> and that's in today's compose
16:10:43 <pbrobinson> there was a few other ARMv7 fixes there too
16:11:04 <pbrobinson> for various AllWinner and Pandaboard
16:11:20 <pbrobinson> overall I think 4.8.6 is looking reasonable for aarch64/ARMv7
16:11:53 <pbrobinson> anyone aware of any issues?
16:13:45 <pwhalen> #info Latest F25 kernel-4.8.6-301.fc25. Includes fixes for Allwinner and Panda for ARMv7, patch for aarch64 memory regions.
16:13:57 <pwhalen> #link http://koji.fedoraproject.org/koji/buildinfo?buildID=814667
16:15:37 <pwhalen> #info Please test and report any issues to the list or #fedora-arm.
16:15:45 <pwhalen> #topic 3) ==== Bootloader Status ====
16:17:12 <pwhalen> #info Latest uboot-tools-2016.11-0.3.rc3
16:17:16 <pbrobinson> nothing really of note here
16:17:19 <pwhalen> #link http://koji.fedoraproject.org/koji/buildinfo?buildID=814111
16:17:32 <pwhalen> no issues so far.
16:17:52 <pwhalen> #topic 4) ==== F25 Final ====
16:18:19 <pwhalen> #info Fedora 25 Schedule: 2016-11-15 - Final Release
16:18:55 <pbrobinson> so basically if there's anything that needs to be fixed it's tell me now or forever hold your peace (ie it becomes a 0day at best)
16:19:13 <cov> pbrobinson: which memory region patche?
16:19:28 <pbrobinson> cov: the one rric posted around the NUMA stuff
16:19:48 <pbrobinson> things I care about are all based on installer functionality
16:19:54 <cov> okay qdf doesn't have numa yet
16:21:07 <pwhalen> #info F25 Final bugs - https://qa.fedoraproject.org/blockerbugs/milestone/25/final/buglist
16:22:27 <pwhalen> anyone have concerns not mention for f25 final?
16:22:32 <pbrobinson> cov: well it was thought that jlinton might have seen similar issues with the Juno
16:22:50 <pbrobinson> cov: so more a test today's compose on your hardware and let me know how it works
16:23:15 <cov> okay, will do
16:24:16 <jlinton> Its possible that if it was a nomap problem with DMA
16:24:39 <jlinton> that I might have been dependent on which firmware I was running, but I didnt' note that in the defect, and I haven't seen the problem since
16:24:55 <pbrobinson> jlinton: yes, so if you can test too and let me know that would be fab
16:25:36 <jlinton> Yes, If/when i see it again, I will try to track it down..
16:27:53 <pwhalen> thanks
16:27:54 <pwhalen> #topic 5) == Open Floor ==
16:28:11 <pwhalen> anything for open floor?
16:28:21 <jlinton> (although Juno isn't NUMA, its possible I may have been running with an ACPI fake numa setup which artifically splits the cores/memory)
16:28:32 <jonmasters> ah, just pulled that NUMA patch down - yeah, it isn't necessarily just going to affect NUMA machines
16:29:00 <jlinton> Yah, its going to affect machines with nomap holes IIRC.
16:29:09 <jonmasters> exactly what I was about to say :)
16:29:21 <jonmasters> so mostly numa but not exclusively
16:31:05 <pbrobinson> yes, correct
16:31:20 <pbrobinson> although there's still no upstream decision on the right solution to the fix
16:32:00 <jonmasters> btw, I spoke in person with bjorn about PCIe
16:32:22 <pbrobinson> jonmasters: which bits?
16:32:47 <jonmasters> he said he had not meant to drop the ball on the ECAM quirks we're carrying in Fedora and hoped to either squeeze them in for 4.9 as a fix (doubtful but I need to catch up) or in 4.10
16:33:49 <pbrobinson> OK
16:34:04 <cov> what quirks are being carried?
16:34:24 <pbrobinson> cov: let me find the patch for you
16:34:41 <pbrobinson> primarily for x-gene and old generation thunder X
16:34:43 <cov> on the list before plumbers Bjorn was happy with the framework but not neccessarily all the users
16:35:54 <pbrobinson> cov: http://pkgs.fedoraproject.org/cgit/rpms/kernel.git/tree/arm64-pcie-quirks.patch?h=f25
16:36:09 <jonmasters> cov: yea, he was going to followup
16:36:14 <cov> he put his pci/ecam-v6 branch back up
16:36:20 <cov> it has patches 1 and 2
16:36:22 <jonmasters> good
16:36:37 <jonmasters> even getting the framework in is a good step forward
16:38:03 <pbrobinson> yes, well even if it's pulled into 4.10 I can then pull the final patch set back to 4.9 to replace the version we currently have
16:38:22 <pbrobinson> 4.8 as far as I'm concerned is done in that regard
16:42:22 <pwhalen> last call for open floor..
16:43:01 * cov would like to experiment with chainloading Grub from LK
16:43:06 <cov> not sure when I'll find time
16:43:49 <cov> but that seems to be a missing piece to connect some of the Qualcomm mobile device upstreaming efforts by folks like Jeremy McNicoll with some future Fedora release
16:43:51 <pbrobinson> cov: what's the usercase?
16:44:00 <cov> Fedora on a Nexus phone
16:44:24 <pbrobinson> cov: well I'd just like a sane bootloader on some of the qcom SBCs
16:44:49 <jlinton> Also, of gnome goes to mozjs31, the mozjs31 mmap patch will need to be applied. I will get that added to the defect
16:44:53 <cov> what makes LK insane?
16:45:10 <cov> the Android boot format isn't all that different than UEFI's system partition style
16:45:32 <pbrobinson> cov: the q-boot stuff is quite weird compared to either u-boot or uEFI
16:45:51 <cov> I'm not familiar with q-boot
16:46:50 <cov> would you rather see uboot or UEFI chainloaded, instead of Grub?
16:47:56 <pbrobinson> cov: I don't particularly care, if we get u-boot or uEFI we can then do standard process
16:48:13 <pbrobinson> IE it's then no different than any other device
16:50:27 <cov> okay thanks
16:51:45 <pbrobinson> anything else from anyone else?
16:52:24 <pwhalen> if anyone has anything else, please drop in #fedora-arm, we can continue there.
16:52:26 <pwhalen> #endmeeting