UbuntuIRC / 2012 /07 /11 /#ubuntu-arm.txt
niansa
Initial commit
4aa5fce
=== prp^2 is now known as prpplague
=== txwikinger2 is now known as txwikinger
=== jkridner__ is now known as jkridner
=== wgrant_ is now known as wgrant
=== calculu5 is now known as calculus
=== jeremiah_travel is now known as jeremiah
[08:45] <jeremiah> .c
=== jkridner__ is now known as jkridner
=== zyga is now known as zyga-food
=== zyga-food is now known as zyga
=== prpplague is now known as prp^2
[15:33] <janimo> ogra_, marvin24 did the 3.1 kernel issue have something in common with https://bugs.launchpad.net/ubuntu/+source/upstart/+bug/936667 ?
[15:33] <ubot2> Ubuntu bug 936667 in upstart "Upstart early job logging causes boot failure for systems with no initramfs (error is "No available ptys")" [High,Fix released]
[15:34] <ogra_> janimo, well, we always have an initrd
[15:34] <janimo> any news on that particular nvidia kernel issue?
[15:34] <janimo> ogra_, is that still holding up quantal ac100 images?
[15:35] <ogra_> and that bug is fixed since mid-precise
[15:35] <ogra_> no, what was holding up was my time to fix ac100-tarball-installer and additionally research the oops i get on boot
[15:36] <ogra_> the installer is fixed now
[15:36] <ogra_> tomorrow if i have new images i'll move on with the kernel issue
[15:36] <janimo> ogra_, kernel oops on install or first boot?
[15:36] <janimo> ok
[15:36] <ogra_> sadly ui cant quickly roll images manually atm
[15:36] <ogra_> we only have one builder thats constantly occupied
[15:37] <janimo> is that a regression?
[15:37] <ogra_> janimo, first boot, some initrd issue i think
[15:37] <ogra_> we dropped all non panda HW from the datacenter
[15:37] <janimo> I thought the builder situations was only going to get better
[15:37] <janimo> ah, a good move otherwise
[15:37] <janimo> and not enough new pandas?
[15:37] <ogra_> and until tere is a new mandabox server we only have one machine for livefs builds
[15:38] <ogra_> i think we have a lot of new pandas, but someone needs to set them up
=== zyga_ is now known as zyga
[15:51] <hrw> ogra_: build iamge at home?
[15:51] <ogra_> no, thanks, i have better stuff to do than to set up the ubuntu build infrastructure at home
[16:22] * GrueMaster has the infrastructure, but no loinger has the time or desire.
[16:28] <janimo> the fact our build insfrastructure is so involved as opposed to dead-simple live-build means noone can really take over properly without too much effort
[16:33] <marvin24> janimo, ogra_: I hope the console can be fixed soon
[16:33] <marvin24> otherwise we need to boot with console=tty1 as a workaround
[16:34] <marvin24> janimo: did you tried to compile a 3.5 kernel for tegra already?
[16:37] <janimo> marvin24, no, since Iheard from you it has no graphics :)
[16:39] <marvin24> it has, but a bit buggy
[16:40] <marvin24> maybe still makes no sense to try it
[16:40] <janimo> I may give arm-soc a try - it has more tegra stuff than 3.5
[16:41] <janimo> I just saw too many kernel tree lately that don't even build, so I am not too attracted to yet another one
[16:41] <janimo> :)
[16:42] <marvin24> janimo: yes, me too :-)
[16:47] <janimo> lilstevie, do you know anything about the status of the 3.1 jhinta kernel for tf101? It sort of boots here, but hangs on starting network manager
[16:48] <janimo> the xda forums are horrendous
[16:48] <marvin24> I wonder if tf101 also has the console problem
[16:50] <janimo> marvin24, I had to boot with --no-log
[16:51] <janimo> marvin24, , otherwise not sure how the console problem manifests itself, I did not boot ac100 with 3.1 yet
[16:51] <janimo> it just prints out the kernel logs, then starting init and a few scripts, pausing on NM
[16:51] <marvin24> yes, it oopes shortly after entering userspace boot
[16:52] <marvin24> but looks random
[16:52] <janimo> no oops on screen and no ramconsole built in this kernel
[16:52] <marvin24> is plymouth installed?
[16:53] <janimo> marvin24, yes. Should I make it -x ?
[16:53] <janimo> meanwhile I'll rebuild the kernel with ramconsole on
[16:54] <marvin24> no, it hangs here only with plymouth installed
[16:55] <marvin24> but uninstall is also only a workaround
[16:55] <janimo> marvin24, iirc ogra said making lib/plymouthd or some unexecutable is a workaround
[16:55] <marvin24> yes, or add console=tty1
[16:55] <marvin24> and leave it executable
[16:56] <janimo> marvin24, do you know if tegra2 kernels can support multiple hw from the same build? Ie. ventana, harmony, ac100, tf101 all built in?
[16:56] <janimo> marvin24, thanks, I'll try that too
[16:56] <janimo> will that redirect console output to somewhere invisible?
[16:56] <marvin24> the kernel does, but the bootloader doesn't
[16:56] <marvin24> janimo: yes
[16:57] <marvin24> in mainline, everything will depend on device tree
[16:57] <marvin24> so you can boot the same kernel on all tegras
[16:57] <marvin24> but we ee
[16:57] <marvin24> ... need uboot for this
[16:58] <marvin24> uboot passes device tree to the kernel (same kernel for all)
[16:58] <marvin24> I don't know how other archs are doing it now
[16:58] <janimo> marvin24, in 3.1 there's no device tree yet afaik
[16:58] <marvin24> yep, only in mainline
[16:58] * janimo wonders if a single zImage could be used on both ac100 and tf101
[16:59] <marvin24> sure, add tf101 to mainline
[16:59] <marvin24> or add a device-tree
[17:00] <marvin24> as I said, *all* tegras (tegra2 and tegra3) should be bootable from a single kernel
[17:00] <janimo> so with current 2.6.x/3.0 non-DT, what is uboot specific in the kernel? I see nothing in what I build for in the confis
[17:00] <janimo> configs
[17:00] <janimo> marvin24, yes in mainline, I was wondering pre-DT where we are and will likely still be in 12.10
[17:00] <marvin24> the current bootloader (fastboot) does not support device trees
[17:00] <marvin24> while u-boot does
[17:01] <marvin24> so, mainline + u-boot or fastboot + 3.1
[17:01] <marvin24> (uboot can also boot 3.1 kernels)
[17:01] <marvin24> but hw support is not finished
[17:02] <marvin24> we have to wait - again :-(
[17:02] <marvin24> but regarding DT support, tegra is pretty complete
[17:09] <janimo> marvin24, chmod -x /sbin/plymouthd made it boot into X. will try console= too thanks!
[17:10] <marvin24> janimo: good
[17:10] <janimo> marvin24, do you think it is a nvidia kernel bug?
[17:11] <marvin24> yes, I talked to an engineer from nvidia
[17:11] <marvin24> and he was able to reproduce it on ventana (tegra2 dev board)
[17:11] <marvin24> could also be ubuntu related ...
[17:11] <janimo> do they plan to address it in 3.1?
[17:11] <marvin24> but the kernel shouldn't oops
[17:11] <marvin24> janimo: yes
[17:11] <janimo> possibly, ubuntu userspace is a bit different
[17:12] <marvin24> we tried to track it down (and still try)
[17:13] <marvin24> janimo: I will request a new kernel build when it's fixed ;-)
[17:14] <janimo> marvin24, sure ;)
[17:14] <janimo> marvin24, back to the previous issue - the idea of single image for multiple kernels is _dependent_ on device tree? Since in menuconfig you can check any number of arches to support
[17:15] <marvin24> with "arches", do you mean omap, tegra, ...
[17:15] <marvin24> or boards?
[17:16] <marvin24> you can select different boards, but our bootloader is buggy and does not accept multi-board kernels
[17:17] <marvin24> janimo: no idea what tf101 bootloader does
[17:17] <janimo> marvin24, afacit just takes an android bootimg and starts it :)
[17:18] <janimo> like the ac100 one, not sure about details
[17:18] <janimo> marvin24, ah so the ac100 bootloader would choke on a kernel image that has any other board than paz00 enabled too?
[17:18] <janimo> I meant just tegra2 based ones (paz00 and tf101 in particular)
[17:22] <marvin24> in fact, no one knows what it does
[17:22] <marvin24> s*y
[17:31] <marvin24> I had the impression, that the bootloader reads the machine-id from the kernel and feeds it back to the kernel
[19:12] <LetoThe2nd> prpplague: howdy pardner
[19:12] <prpplague> greetings
[19:13] <LetoThe2nd> prpplague: found time to look at the video i told you?
[19:13] <prpplague> LetoThe2nd: sorry no, yesterday was crazy i had some car troubles from the previous afternoon
[19:13] <prpplague> LetoThe2nd: can you give me the link again? i have it at home
[19:13] <LetoThe2nd> prpplague: ah kay... hope you got things roted out?
[19:14] <prpplague> LetoThe2nd: yea
[19:14] <LetoThe2nd> sorted out, even :/
[19:14] <prpplague> LetoThe2nd: i had the car worked on friday, turns out they put one of the nuts back on without the cotter pin
[19:14] <LetoThe2nd> aw :(
[19:15] <prpplague> LetoThe2nd: damn near got killed by a large truck on monday afternoon after the front passenger tie rod came off while i was cross an highway intersection
[19:15] <LetoThe2nd> prpplague: http://www.youtube.com/watch?v=ufBCoHP-8Yk&list=UUXHs7mzjNipr-xp1Bps2zew&index=2&feature=plcp
[19:16] * prpplague looks
[19:16] <prpplague> LetoThe2nd: you said about 50 seconds in?
[19:16] <LetoThe2nd> prpplague: interesting part is from 0:55 onwards or so.
[19:16] <LetoThe2nd> prpplague: glad nothing bad happened then
[19:16] <prpplague> LetoThe2nd: this video is work safe correct?
[19:16] <LetoThe2nd> prpplague: totally yes.
[19:17] <LetoThe2nd> prpplague: its from an educational fair near me.
[19:18] <prpplague> LetoThe2nd: ahh dandy!
[19:18] <LetoThe2nd> prpplague: and inside that rc car is a panda :)
[19:18] <prpplague> LetoThe2nd: yea i think those guys were in cambridge a while back at ELC
[19:18] <prpplague> LetoThe2nd: with an early prototype of the design
[19:20] <LetoThe2nd> prpplague: will meet those guys sometime in the next two weeks probably, was really surprised there are other panda users in this small town.
[19:22] <prpplague> LetoThe2nd: hehe
[19:22] <LetoThe2nd> prpplague: btw got one or two minutes for private?
[19:22] <LetoThe2nd> (just noticed we're not even in #pandaboard)
[19:22] * prpplague checks the time
[19:22] <prpplague> LetoThe2nd: yea, got about 5-10 mintues
[21:04] <smplman> does anyone know where i can find the status of the SGX drivers on a BB XM running 12.04?
[21:04] <smplman> and audio also
[22:32] <xranby_ac100> smplman: check the linaro website
[22:32] <xranby_ac100> smplman: all the latest linaro releases are based on 12.04 with improved kernels and drivers where available
[22:34] <xranby_ac100> smplman: sorry i think i gave you bad advice.. they only support the pandaboard nowdays