IRC logs for #baserock for Wednesday, 2016-01-06

*** edcragg has quit IRC01:06
*** bwh_ is now known as bwh02:33
*** bwh has quit IRC02:36
*** gtristan has quit IRC02:38
*** bwh has joined #baserock02:41
*** gtristan has joined #baserock02:43
*** bwh has quit IRC02:44
*** gtristan has quit IRC05:06
*** gtristan has joined #baserock06:32
*** bruce_ has joined #baserock08:30
*** paulw has joined #baserock08:40
*** paulw has quit IRC08:40
*** bashrc has joined #baserock09:08
*** ssam2 has joined #baserock09:19
*** ChanServ sets mode: +v ssam209:19
*** jonathanmaw has joined #baserock09:36
*** edcragg has joined #baserock10:13
*** Lachlan1975 has joined #baserock10:15
rdalei'm trying to use ybd to build a minimal system: ../ybd/ systems/minimal-system-x86_64-generic.morph x86_6410:17
rdalebut i'm getting an error 'Too many mounts (maximum of 50)' - does anyone know what that means?10:17
pedroalvarezhm... wasn't that because linux-user-chroot?10:18
pedroalvarezIIRC the workaround was to un-install  linux-user-chroot so that sandboxlib (used by ybd) uses chroot10:18
rdalehmm, would that stop morph from working?10:19
pedroalvarezoh, so you are running ybd in a baserock image?10:19
pedroalvarezthen it might not be the linux-user-chroot problem10:19
rdaleno i'm running ybd in my normal laptop environment, so it should be ok to remove linux-user-chroot then10:21
pedroalvarezjust a note: that problem in linux-user-chroot was fixed, but is not included on distros10:21
pedroalvarezwell, looks like fedora has the fix in10:24
paulsherwoodi didn't get round to fixingit10:24
pedroalvarezI don't understand why that version works fine for morph then10:26
rdalei've removed linux-user-chroot anyway and it is getting past the 50 mounts error, and failing to compile stage1-binutils instead10:26
paulsherwoodrdale: what's the error?10:28
paulsherwoodrdale: are you root?10:31
pedroalvarezlooks like ubuntu also includes the fixed linux-user-chroot version10:31
paulsherwoodsadly, you need to be10:31
paulsherwoodthis is an ugly error :/10:32
paulsherwoodrdale: i've raised on this. patches gratefully accepted :)10:34
paulsherwoodpedroalvarez: iirc ybd does more mounts, because it builds recursively and creates staging as it goes10:36
pedroalvarezthen.. by reading I take that ybd can't use linux-user-chroot :/10:37
paulsherwoodi've not tried it with the 'fixed' l-u-c... it should work10:41
paulsherwood(but then i've never seen the device or resource busy error, either, so this may be distro-specific)10:43
gtristanI originally passed a build with a locally built l-u-c with the arbitrary max mounts set to 102410:43
gtristanwhen I filed that bug, I had max-mounts at 1024, and still failed :-/10:44
gtristanit may be because I also had instances: 2 ?10:44
pedroalvarezI continue thinking that if all these facts are true, ybd shouldn't attempt to use linux-user-chroot10:45
paulsherwoodyou're probably right10:47
pedroalvarezradiofree: I'm wondering if you know anything about possible issues of upgrading weston to 1.9 for genivi baseline.10:53
pedroalvarezi take the Jetson TK1 patches are still needed10:54
radiofreepedroalvarez: if you upgrade weston then the version of ivi-extension used in the GDP won't work10:56
radiofreenot sure if we care about that though, since GDP is yocto?10:56
paulsherwoodradiofree: only the yocto GDP is yocto?10:57
paulsherwoodthere is a baserock GDP too, iirc?10:57
radiofreethat's more like a proof of concept10:57
radiofreei thought the only "official" gdp is the yocto one?10:57
radiofreeat least, no one seems to be doing any work on the baserock gdp10:57
paulsherwooda proof of concept for a demo system?10:57
radiofreepedroalvarez: yes the patches are needed,  there's already a branch for it though
radiofreethat will work on all supported weston platforms by the way, though you might prefer to just use the 1.9.0 tag there10:58
pedroalvarezradiofree: oh, that's nice.I'll stop making those patches10:58
radiofreepaulsherwood: i haven't been following it, but when I left it it was still a personal branch of definitions since the whole thing is such a hacky mess that I wouldn't want to merge it10:59
pedroalvarezso yes, you are right. gdp-hmi will need rework to support new versions of things...11:00
paulsherwoodradiofree: many projects seem quite happy to merge a hacky mess, but i appreciate your point :-)11:00
pedroalvarezI wonder then why yocto is upgrading weston on its baseline11:00
radiofreeon the yocto side it doesn't sound like it's getting any better, maybe we can do it for baserock, properly (and just fix the required version of weston/ivi-extension)11:00
radiofreei.e i believe they're still targeting qt 5.4 + loads of gstreamer 1 patches, since that's in the yocto baseline11:01
radiofreerather than using qt 5.5, which has gstreamer 1 support built in11:01
paulsherwoods/yocto/wind river/ pedroalvarez - maybe email them in public to ask?11:01
paulsherwoodgenivi-projects and meta-ivi would be the place, i think11:01
radiofreepedroalvarez: there was some forward porting going on, let me check11:02
radiofreepedroalvarez: yeah best ask, what version are they upgrading to?11:03
radiofreeright, so that's fine, however i believe the GDP is going to target an older baseline?11:05
paulsherwoodthis conversation should be in #automotive :)11:06
radiofreeby all means upgrade to 1.9 for our baseline, see the recent jetson patches11:06
radiofreeyou can now use everything from upstream :) (well.. apart from those 2 weston patches)11:06
radiofreebtw you need to upgrade u-boot for that to work, you can do that from userspace (so you could use 'upgrade', then update u-boot from within the new image) rather than reflashing the whole thing11:08
radiofreei'm working on a script for that (/boot/u-boot/, should i just add that to the u-boot morph file, or create a branch on bsp-support?11:09
pedroalvarezso that whenever you upgrade the system, the script runs and installs the new version?11:11
pedroalvareznot sure where it should go, tbh11:11
pedroalvarez sounds interesting11:11
radiofreeyou'd have to run it manually11:11
radiofreethere's already a /boot/u-boot/ folder on the target, it contains some files you need to extract when you're doing the initial flash11:15
radiofreepartitioning support is in baserock now? is there any example of that (might move the jetson image over to that)11:16
pedroalvarezit is! and I would love to use that for the jetsons too11:18
pedroalvarezradiofree: hasn't been merged yet, but here what was needed to create rpi images:
pedroalvarez(relevant because it uses partitioning)11:20
radiofreenice, will take a look11:28
radiofreepaulsherwood: does ybd support this partitioning stuff?11:56
paulsherwoodradiofree: i don't know... isn't that just scripts in defintions?12:01
radiofreei think so12:02
paulsherwoodshould work, then :)12:03
* paulsherwood crosses fingers12:03
pedroalvarezyeah, I did the "deploying GDP to raspberry pi" demo using ybd12:05
paulsherwoodoh yes...
paulsherwood(it's had a couple of hundred views, btw)12:06
radiofreepaulsherwood: ybd cache, is that for releases?12:09
paulsherwoodradiofree: i've been updating it on an ad-hoc basis12:09
paulsherwood(last time was in december)12:09
radiofreelooks like the latest tag is cached12:10
paulsherwoodi was planning to do a cron job but didn't get round to ir12:10
rdalewhat are .cida file in ybd?12:18
paulsherwoodiirc i was considering having a process where we'd convert from .morph, .bb etc into some generic 'best' representation, and i called it cida... but i didn't get very far with it12:20
rdaleah ok12:20 has a bit of stuff in it irrc12:21
rdaleyes, that's what i'm looking at12:21
*** gtristan has quit IRC13:27
*** Lachlan1975 has quit IRC13:34
*** De|ta has quit IRC14:03
*** De|ta has joined #baserock14:03
*** gtristan has joined #baserock14:07
radiofreepedroalvarez: there were some existing patches to upgrade to weston 1.9 btw, they have been merged14:21
radiofreenot sure if they touched the baseline though14:21
radiofreeif you upgrade to 1.9 make sure you upgrade the ivi-extensions as well, 1.5.0 i think14:21
radiofreeif you're going to be upgrading, could you take a look at these patches as well?
radiofreepedroalvarez: regarding partitioning, anything i put /boot will automatically get moved to the correct partition then?14:26
*** toscalix has joined #baserock14:26
pedroalvarezradiofree: yes14:26
pedroalvarezI sent an email a while ago about possible issues with partitioining etc14:27
*** toscalix has quit IRC14:30
radiofreei'll have to move the u-boot stuff i left in /boot/u-boot to somewhere else then14:31
radiofree/usr/share/jetson/u-boot ? (there's a couple of other config files i want to add)14:32
pedroalvarezthat works too14:32
pedroalvarezmaybe /usr/lib/foo work too14:32
radiofreepaulsherwood: ybd doesn't appear to work with the partitioning stuff14:33
paulsherwoododd... pedroalvarez had it working i thought?14:33
paulsherwoodwhat's the result?14:34
* pedroalvarez nods14:34
radiofreeit just create a normal image file14:34
radiofreehere's my cluster
pedroalvarezradiofree: wait, I think there was a last-minute change in partitioning.. one sec14:35
pedroalvarezbefore it was merged I mean14:36
radiofreei'm using the latest release of baserock14:36
pedroalvarezUSE_PARTITIONING: yes14:36
pedroalvarezthere is another example in definitions.git14:36
radiofreeok, i'll give that ago14:36
*** toscalix has joined #baserock14:38
*** toscalix has quit IRC14:40
radiofreeok, that work14:42
radiofreeextlinux.conf still gets created in the root of the btrfs partition though14:43
radiofreealso extlinux.conf (for u-boot) needs to be in /extlinux/extlinux.conf14:45
radiofree(on the boot partition)14:45
radiofreei'd rather rely on the deploy creating that, so i guess it needs to change to specific a partition (and/or path?) to use?14:46
radiofreei'll double check with a newever version of u-boot, maybe /extlinux.conf loads now14:46
radiofreeit does not...14:48
pedroalvarezyeah, all of this needs a bit of thinking..14:48
*** gtristan has quit IRC14:48
pedroalvarezI tried to put my thoughts on that email, maybe now you have something to add14:49
radiofreepedroalvarez: i could always add a boot.scr to /boot that just calls "sysboot ... /extlinux.conf", however i'd still need to be able to specify the partition extlinux.conf gets written to14:49
*** toscalix has joined #baserock14:50
*** toscalix has quit IRC14:52
*** edcragg has quit IRC15:05
*** toscalix has joined #baserock15:06
*** edcragg has joined #baserock15:07
*** gtristan has joined #baserock15:07
*** toscalix has quit IRC15:08
*** toscalix has joined #baserock15:14
*** toscalix has quit IRC15:14
*** toscalix has joined #baserock15:16
*** toscalix has quit IRC15:16
*** bashrc has quit IRC15:21
*** bashrc has joined #baserock15:22
pedroalvarezrichard_maw: quick comment about your email: you have assumed that all tarball imports have the -tarball suffix15:37
richard_mawyes, purpose was to understand roughly how many of each classification we're using, given how many -tarball imports we have, I assumed any non -tarball tar imports would be a sufficient minority to not matter15:38
pedroalvarezany opinions about doing this?
pedroalvarezI think it will help with errors like this:
richard_mawpedroalvarez: the `if not exists(path)` check is now redundant and can be removed15:46
richard_mawnot sure exactly how you're triggering that otherwise, needs a close race condition15:47
richard_maware you sure that's the source of the errors?15:47
pedroalvarezvarious morph instances trying to create the folder at the same time15:47
pedroalvarezrichard_maw: I'm not sure, but it looked like the place to me15:48
radiofreewould "BOOTLOADER_CONFIG_LOCATION" be acceptable to people?15:49
radiofreedefaults to current 'real_root' (i.e / of the btrfs partition)15:49
richard_mawradiofree: tbh I'd specialise it by bootloader rather than assuming that every bootloader has a base path for its config files15:50
richard_mawcould be baked into the partition table flags etc.15:50
richard_mawpedroalvarez: aye, that's grand then, just remove the redundant check whether the path exists if you're safely handling creation15:51
radiofreerichard_maw: there's already BOOTLOADER_CONFIG_FORMAT15:51
radiofreehowever for the jetson that needs to go in /boot/extlinux/extlinux.conf rather than /extlinux.conf15:51
richard_mawradiofree: aye, but that selects which bootloader to use, rather than configuration for the selected bootloader15:52
radiofreewell, it creates the extlinux.conf file15:52
radiofreefor x86 images that needs to go in / of the btrfs partition, for jetson images that needs to go somewhere else15:53
richard_mawI see where you're coming from now. I think of BOOTLOADER_CONFIG_FORMAT as deciding which bootloader config file to install while you're interpreting it as what format to write the config file in15:54
richard_mawI'd still prefer EXTLINUX_CONFIG_LOCATION over BOOTLOADER_CONFIG_LOCATION since you might have a reason to install multiple bootloader configs, and you could have a bootloader configured by partition metadata, at which point the configuration wouldn't be in a file15:56
radiofreethat's fine with me!15:56
radiofreeseems good to me15:57
pedroalvarezrichard_maw: I'll check these changes for some days in Mason before sending the patch, thanks!15:57
*** jonathanmaw has quit IRC16:28
pedroalvarezupgrading to weston 1.9 for baseline was really smooth16:35
pedroalvareznot tested on a jetson yet, though, but I don't see why it would fail16:36
radiofreeit works for me16:36
pedroalvarezit just worked16:37
paulsherwoodpedroalvarez: should this be mentioned on #automotive too :)?16:41
* pedroalvarez pats himself16:41
radiofreepedroalvarez: did you try with the kernel 4.4 patches?16:46
pedroalvarezdidn't have time yet, was using x8616:48
pedroalvareznow might be the moment16:48
pedroalvarezradiofree: question, re-flashing needed to test the u-boot upgrade, right?16:49
pedroalvarezthat's an obvious yes..16:49
radiofreepedroalvarez: no!16:55
radiofreeyou can use morph upgrade16:55
radiofreehowever you'll have to reflash u-boot, i can tell you how to do that from userspace if you want16:56
pedroalvarezoh, right16:56
pedroalvarezradiofree: please16:56
radiofreewell, actually it requires an app to be installed called "cbootimage" to generate something, i'll e-mail you the generated img though16:58
radiofree(it's pretty small)16:58
* pedroalvarez goes to grab a jetson to test the patches16:58
radiofreei'm going to submit a script for this once the new stuff has been merged16:59
radiofreei'll e-mail you "upgraded-uboot"17:00
radiofree(it's not *just* the u-boot.bin)17:00
radiofreepedroalvarez: biff17:02
radiofreeactually a newever version of u-boot than in the patches, bleeding edge...17:03
pedroalvarezhow should I proceed once I've `morph upgraded` the system?17:05
radiofreecopy that file to your jetson17:06
radiofreerun the commands in that link i sent you17:06
* pedroalvarez should read emails first before asking17:07
pedroalvarezoh no, link is in here, missed that message17:08
pedroalvarezhah, scary script :)17:08
radiofreeit's fine, trust me™17:13
*** bruce_ has quit IRC17:24
*** ssam2 has quit IRC17:24
*** bashrc has quit IRC17:50
pedroalvarezradiofree: that worked :)17:53
pedroalvarezI should test graphics before merging17:57
pedroalvarezradiofree: that doesn't work :(18:17
pedroalvarezweston on genivi baseline with those patches18:17
radiofreeok, will fix that tomorrow18:22
radiofree(forgot to create the gk20a folder)18:25
jjardonpedroalvarez: patch updated18:30
*** edcragg has quit IRC18:34
pedroalvarezWill test first thing tomorrow morning18:50
*** rdale_ct has joined #baserock20:21
*** rdale has quit IRC20:25
*** doffm has quit IRC22:04
*** doffm has joined #baserock22:04
*** tiagogomes_ has quit IRC22:05
*** faybrocklebank has quit IRC22:05
*** tiagogomes_ has joined #baserock22:07
*** faybrocklebank has joined #baserock22:10
*** edcragg has joined #baserock23:42

Generated by 2.14.0 by Marius Gedminas - find it at!