[00:00] <GrueMaster> brandini: Yes. Multiple trees.  http://kernel.ubuntu.com
[00:01] <brandini> phew, why couldn't I find that
[00:07] <brandini> GrueMaster: I'm still looking for the assembly code you guys wrote to initialized the panda :)
[00:09] <GrueMaster> I believe TI keeps it in the mach-omap2 stuff under arch/arm.  Not sure.
[00:09] <GrueMaster> Or plat-omap.
[00:10] <brandini> they release source for that?
[00:13] <GrueMaster> If you clone the ubuntu/ubuntu-oneiric.git tree, there should be a branch in it for ti-omap4 (iirc).
[00:13] <brandini> thank you :)
[00:14] <GrueMaster> It is a separate branch though, not the main trunk.
[00:16] <GrueMaster> Looks like ubuntu-3.0.0-1205.10 (according to git web).
[03:39] <twb> lilstevie: ping
[03:39] <twb> lilstevie: nag re send me known-good u-boot.bin
[08:21] <twb> lilstevie: poke poke
[08:21] <CodeWar> whats a good hardware to test ubuntu-arm on ? preferrably a SMP one
[08:22] <CodeWar> Tegra part of the supported list?
[08:22] <twb> CodeWar: efikamx should be good, they seem to actually be contributing to debian
[08:23] <twb> tegra is more like "if the moon is waxing and you're standing on one leg, it works OK...ish"
[08:24] <infinity> We don't have installers for efika systems right now (and they're not SMP).
[08:24] <infinity> If you want SMP, a Toshiba AC100 netbook  or a TI PandaBoard are your best bets.
[08:24] <twb> infinity: oh, sorry
[08:24] <twb> AC100s are about the same as TF101s, aren't they?  i.e. sucky?
[08:24] <infinity> Mine works great.
[08:25] <twb> OK
[08:25] <infinity> Outperforms the Panda by no small margin, if it had a US keyboard layout, I'd actually use it as my primary netbook.
[08:25] <infinity> But the Uk keyboard makes me want to kick puppies. :P
[08:25] <twb> stupid enter key?
[08:26] <infinity> Stupid everything.  Keyboard layouts are a religious thing. :)
[08:26] <CodeWar> AC101 dual core A9 .. decent enough let me look it up
[08:26] <twb> Well you can remap it if it's just the caps
[08:26] <infinity> CodeWar: It's a Tegra2.
[08:27] <infinity> twb: Yeah, remapping it fails a bit because you end up with a teeny-tiny \| key, due to the enter key eating most of its neighbours.
[08:27] <twb> yeah OK
[08:27] <infinity> Om now now.
[08:27] <infinity> nom nom too.
[08:27] <twb> That's one of my biggest hates on keyboards, that big enter key
[08:27] <CodeWar> Asus Transformer 2 .. is that expected to work :-)
[08:28] <twb> I mean half the time I type ^M anyway
[08:28] <CodeWar> would be best
[08:28] <infinity> CodeWar: There have been some people fiddling with the TF2.  It has no official support, but I know you can make it work with enough effort.
[08:28] <twb> CodeWar: I have a TF101 (Eee Pad Transformer 32G); currently it only works with crappy old 2.6.36
[08:28] <infinity> CodeWar: Panda or AC100 work out of the box, which is appealing if you just want to get to hacking.
[08:28] <twb> When lilstevie comes back from the pub or his girlfriend's or whatever and helps me, I might make some more progress :P
[08:29] <CodeWar> thanks guys .. still trying to wrap my head around these various models ..
[08:29] <diwic> infinity, I was a little surprised when I read the release notes, AC100 and IMX.53 (I think?) was listed, but not the Pandaboard.
[08:30] <infinity> (And most devices that ship with Android can just be abused to boot Ubuntu with an Android kernel, but the user experience there will vary, depending)
[08:30] <twb> That's cheating
[08:30] <twb> it doesn't count
[08:30] <infinity> diwic: AC100 and i.MX53 were listed as new, omap3 and omap4 were already supported.
[08:30] <diwic> infinity, ok, that explains it, thanks
[08:31] <soren> You know what would be awesome? If someone sold a complete get-started-on-hacking-Ubuntu-on-ARM kits at UDS. Pandaboard, power supply, SD card with Ubuntu pre-installed, USB-serial dongle, whatever else one might need.
[08:31] <twb> sheevaplug used to ship with ubuntu pre-installed
[08:32] <infinity> soren: To be fair, that's more or less what you get if you order a Freescale i.MX53.  They ship with an SD with Ubuntu.  Though, I need to find someone at Freescale to talk to about refreshing their image to use a saner kernel.
[08:32] <twb> But IMO the goal should be a pure blend
[08:32] <infinity> It would be nice if TI did the same thing with the Panda package, though.
[08:32] <infinity> Not that it's rocket science to download an image and make it go, but the user experince is fairly shiny when it "just works" without having to read.
[08:34] <soren> infinity: Looking at the i.MX53 now. Glancing at the specs, it looks somewhat less beefy.
[08:34] <infinity> soren: Well, it's a single-core A8, which is less cool than a Panda or AC100, but for people doing a lot of building, the on-baord SATA more than makes up for it.
[08:35] <soren> infinity: Oh, shiny. I didn't notice that.
[08:35] <infinity> soren: Building most things on A8/A9 systems is almost entirely I/O bound, not CPU.
[08:35] <twb> infinity: what, you don't use iSCSI pointing at the SAN for everything ? ;-)
[08:35] <soren> infinity: Ok. Well, let me rephrase then..
[08:35] <soren> You know what would be awesome? If someone sold a complete get-started-on-hacking-Ubuntu-on-ARM kits at UDS. Freescale i.MX53, power supply, SD card with Ubuntu pre-installed, USB-serial dongle, whatever else one might need.
[08:36] <infinity> twb: It doesn't matter how cool my networking tech is, you can't get past the part where Pandas have a 100bit ethernet adapter hanging off a USB 2.0 bus. :P
[08:36] <soren> 100bit? Holy crap.
[08:36] <infinity> Mbit.
[08:36] <infinity> Typing is hard.
[08:36] <soren> Oh. Those.
[08:36] <infinity> It feels like 100bit.
[08:36] <twb> that's the uart :P
[08:37] <infinity> (To be clear, I have no issues with the Panda's architecture, it's a dev board meant to be a giant cell phone, and it works great for what it's meant to do... It's just a lousy desktop or build server due to USB being your limiting factor for any storage)
[08:38] <soren> At least for me, having to buy unknown hardware and bits and pieces just to even get started has put me off for a looong time. Now I've actually bought a pandaboard, but still haven't gotten it to work.
[08:38] <infinity> soren: Flash oneiric image to SD, insert, boot.
[08:38] <soren> Done that.
[08:38] <infinity> soren: It's pretty straightforward these days.
[08:38] <soren> Doesn't work. It just lights up very briefly, then turns off.
[08:38] <soren> No clue why.
[08:38] <infinity> That sounds unpleasant.
[08:38] <infinity> The turning off bit.
[08:38] <infinity> They don't do that.
[08:39] <soren> I RMA'ed the board (not just for this reason), but the new one does the same.
[08:39] <infinity> Oh, actually, it might do that if it fails to find anything interesting on the SD, I don't recall.
[08:39] <soren> What the heck. I'll give it another go.
[08:39] <infinity> Wiggle the card, rewrite it moar bettar, use a different one?
[08:40] <soren> Where' the current "Idiot's guide to Ubuntu on Pandaboard"?
[08:40] <soren> I've tried two different card.
[08:40] <infinity> https://wiki.ubuntu.com/ARM/OMAP
[08:40] <soren> Same micro-SD-regular-SD converter, though.
[08:41] <infinity> Oh, I've had serious issues with micro->regular->Panda, though I'd always assumed it was just my own cards and adapters being shit.
[08:41] <infinity> (Which it probably is)
[08:41] <soren> Maybe I should see if I could find a regular SD card somewhere.
[08:41] <soren> ...or just try a different adapter.
[08:41] <infinity> But yeah.  SD/MMC is about the worst choice for installatoin media ever, but we have no options.
[08:41] <soren> HAdn't thought of that.
[08:42] <twb> infinity: ferrite core
[08:42] <infinity> If I recall, the Panda will just appear to "do notihng" if you light it up and it sees nothing of interest in the SD slot.
[08:42] <soren> infinity: Well, if no SD card is in, it stays on.
[08:42] <soren> When my SD card is in, it shuts off.
[08:42] <infinity> Well, nothing of interest, after looking.
[08:42] <infinity> It's a bit fiddly in those first few miliseconds. :P
[08:42] <infinity> After that, it's great!
[08:44] <soren> I guess it could be the power supply.
[08:45] <soren> It says it goes up to 2.5A.
[08:47] <twb> Or dirty power that is pissing it off because it's a switching-mode PSU and thus sensitive to problems in the wave
[08:47] <twb> We went through three soekris net5501 PSUs here before we gave up
[08:49] <soren> infinity: The for the PAndaboard I want to use the OMAP4 image, right?
[08:50] <infinity> soren: Yup.
[08:51]  * soren downloads
[08:54] <soren> Oops, accidentally grabbed the natty image. Will that work ok?
[08:54] <infinity> It will, but why start out-of-date?
[08:54] <infinity> Especially if you're using SD.  Upgrading is SLOW. :P
[08:56] <soren> I just want to see that damn thing work.
[08:56] <soren> Also, the oneiric image seems to be more than 3 times bigger. Writing out the natty image takes long enough.
[09:03] <soren> Holy crap! It stayed on!
[09:03] <infinity> Magic.
[09:04]  * soren headdesks
[09:04] <soren> It's been that adapter all along, then!
[09:06] <soren> And the one I'm using now is identical. *sigh*
[09:12] <ogra_> lool, poke
[09:17] <ogra_> hmm
[09:17] <ogra_> infinity, do you happen to know where the new flash-kernel lives in debian atm ? did it migrate to unstable or is it still in experimental ?
[09:18] <ogra_> (i would like to file a sync request ahead of time so we have it immediately after opening)
[09:18] <infinity> ogra_: Not sure.  I'd assume either experimental, in some VCS somewhere, or on lool's hard drive.
[09:18] <ogra_> i suspect we will run into a pile of issues
[09:18] <ogra_> so the sooner we get it the better
[09:18] <infinity> ogra_: I'm not positive that lool thinks it's ready for prime-time, but I haven't talked to him about it for a couple of months.
[09:19] <ogra_> i know its ready for the arches we support but might handle things different than we do for these
[09:19] <infinity> ogra_: Given that this is an LTS coming up, I'm not sure I'm inclined to switch.
[09:19] <ogra_> i know its not ready for "old" arches
[09:19] <infinity> ogra_: Dealing with the bugs we know seems saner.
[09:19] <ogra_> which we dont care about
[09:19] <ogra_> and i know its used by default in armhf in debian
[09:20] <ogra_> but thats not in any official repo
[09:20] <infinity> Almost.
[09:20] <infinity> But is it?  They don't have installation media.
[09:20] <ogra_> infinity, well, you had a spec to actually use its database for HW stuff iirc
[09:20] <ogra_> and i would like to get rid of the crap ahead of the LTS
[09:21] <ogra_> else we need to carry the hackish version for 5 years
[09:21] <infinity> http://ports.debian.net/debian/pool-armhf/main/f/flash-kernel/
[09:21] <soren> What does "hf" stand for, btw?
[09:21] <ogra_> right, i dont think we can sync from there
[09:21] <diwic> hard float
[09:21] <ogra_> soren, hard float
[09:21] <soren> Ah.
[09:22] <diwic> ogra_, does the official oneiric image for AC100 have working sound?
[09:22] <infinity> ogra_: Eh.  I don't mind carrying it for 5 years.  Once it's set up, it works.  It's not like it's a maintenance burden.
[09:22] <ogra_> diwic, btw, doi you remember who waas chiming in when we talked about 5.1 recievers ? i owe him a beer :)
[09:22] <infinity> diwic: It does if you use headphones.
[09:22] <ogra_> diwic, only with a kernel update thats not in an SRU yet and with some alsamixer adjustments
[09:23] <ogra_> and it breaks after resume from suspend
[09:23] <ogra_> but all these issues should be fixable and will make it into SRUs
[09:23] <infinity> ogra_: The problem with that spec is that it assumes some debian-cd violence and other things.  Which I'm happy to do, but I feel like it would be a waste of timein a cycle where we should be focussing on stabilising existing software, not introducing new stuff.
[09:23] <diwic> ogra_, sorry, don't remember the 5.1 receiver stuff?
[09:24] <ogra_> infinity, well, for me thats not so much introducing new stuff but getting rid of the horrid hacks
[09:24] <infinity> ogra_: I really do want to implement that spec and get the new hw DB idea working right, but this just feels like the wrong time.  I dunno.  We'll argue about it in Orlando. ;)
[09:24] <ogra_> diwic, yeah, its started with me asking you about spectrum analyzer software
[09:24] <diwic> right
[09:24] <ogra_> he chimed in and recommended me to rather get a denon ...
[09:25] <ogra_> after 3 weeks of bitter ear pain with the new yamaha i returned it on monday ... now i have a denon, no more earpain, waaaay better sound :)
[09:25]  * soren pats his Denon receiver in his desk
[09:26] <ogra_> and i got it reaaaaly cheap because they felt pity for giving me something that produces ear pain :)
[09:26] <ogra_> (like a 500€ discount *g*)
[09:26] <soren> Wow.
[09:27] <ogra_> soren, for the SACD player i bought i got a 1000€ discout because the box was missing ....500 isnt that much ;)
[09:28] <soren> If you can get a €1000 it must have been rather pricey to begin with.
[09:28] <lilstevie> ogra_: is there anything major different between the AC100 image and the omap builds?
[09:28] <soren> I think mine cost €1000 total.
[09:28] <soren> 11 years ago.
[09:29] <ogra_> lilstevie, look at ac100-tarball-installer source, that has all the specialities ... beyond that indeed it has its own kernel and bootloader setup
[09:29] <lilstevie> ok :)
[09:30] <ogra_> soren, yeah, my old system costed me 600, including cup sized speakers etc ... i thought its time for an upgrade and invested 5000
[09:30] <lilstevie> ogra_: was mainly wondering for basing my transformer stuff from one of them
[09:30] <ogra_> lilstevie, feel free, but you might need some adjustments
[09:31] <lilstevie> well I do have a different bootloader setup
[09:31] <ogra_> right, and a different kernel
[09:31] <lilstevie> yeah
[09:54] <lool> ogra_, infinity: New f-k is in experimental and in git
[09:54] <lool> there are more changes to be done, but I guess it's already better than what we have
[09:54] <ogra_> lool, do you think its suitable for an LTS ?
[09:54] <ogra_> yeah, thats what i think too
[09:54] <lool> it doesn't support SD card right now
[09:55] <ogra_> and i dont want to sit on what we have for 5 years
[09:55] <ogra_> argh, seriously ?
[09:55] <lool> well, it doesn't have any OMAP mechanism; you could point it at /boot though
[09:55] <lool> I don't think much OMAP support made it to Debian
[09:56] <ogra_> does it have omap4 ?
[09:56] <ogra_> shouldnt be to hard to derive from there
[09:56] <lool> that included lack of OMAP4
[09:56] <ogra_> bah, k
[10:29] <brandini> morning all
[10:31] <brandini> is com0 at 0x49020000 on the panda?
[10:38] <FunkyPenguin> aloha, does ubuntu build from a single source or do you have a separate one for arm vs x86?
[10:38] <FunkyPenguin> wondering if i look for a package in packages.ubuntu.com does that cover arm?
[10:46] <ogra_> most of the time, yes
[10:46] <ogra_> all packages use the same source, packages.u.c wont tell you if the binary exists though
[10:47] <ogra_> http://qa.ubuntuwire.org/ftbfs/ has a list of all failed builds
[10:48] <FunkyPenguin> ogra_: great, thanks
[10:49] <ogra_> you can also look on launchpad.net/ubuntu/
[11:38] <soren> What's the rationale behind that /var/lib/preinstalled-pool thing in Oneiric?
[11:41] <soren> What's worse is that the very first package I installed from there has a checksum mismatch :(
[11:44] <ogra_> that shouldnt happen, infinity ^^^
[11:44] <soren> The python-setuptools package I have in there is all NULLs.
[11:44] <ogra_> its what we ship as pool on the x86 server isos and should support installs without network
[11:44]  * soren checks the image
[11:45] <ogra_> right, compare your md5
[11:46] <soren> Image checksum matches.
[11:46] <ogra_> hmpf
[11:46] <infinity> Yeah, but does it match once you've written it to SD?
[11:47] <soren> That's what I'm about to find out.
[11:47] <soren> WEll... Sort of.
[11:47] <infinity> I can write an image out in a sec and have a look.
[11:47] <infinity> Where "in a sec" is "sometime this afternoon".
[11:47] <soren> I'll mount the image on my laptop and see if the problem exists there, too. Otherwise, it probably got screwed up when I wrote it to the SD.
[11:49] <soren> Nope, it's fine if I mount it on my laptop.
[11:49] <soren> Weird.
[11:49] <soren> No idea where it got scrwed up, then.
[11:49] <ogra_> hmm, porobably a kernel issue then... mmc driver or filesystem driver issue
[11:50] <soren> Most of the files in there seem fine.
[11:51] <ogra_> on your laptop...
[11:51] <soren> Just 9 of them have this problem.
[11:51] <soren> No no, on the SD card.
[11:51] <ogra_> if inserted in your laptop or in the panda ?
[11:51] <soren> Panda
[11:51] <ogra_> ah
[11:51] <ogra_> hmm
[11:52] <ogra_> bad SD ?
[11:52] <ogra_> i.e. some borked blocks that shouldnt have been written to ?
[11:52] <soren> No idea. It's brandh new. Just unwrapped it a couple of hours ago.
[11:53] <soren> dmesg is silent.
[11:53] <soren> (on this subject, I mean)
[11:53] <soren> Plenty of other stuff in dmesg.
[11:55] <ogra_> hmm
[12:03] <infinity> I could rant about SD/MMC quality again, but I seem to do that often enough.
[12:04] <infinity> While I think it's cool that we provide an SD grow-root installation method (and, indeeed, that's the simplest way for people to test their hardware and get started), I still think the goal should be to encourage people to get their system on reliable external storage ASAP.
[12:05] <infinity> And just use the SD for uBoot.
[12:06] <ogra_> or support boards with nand booting only :P
[12:06] <ogra_> we just need to convince vendors to put more money in :)
[12:07] <infinity> ogra_: I don't mind treating the SD slot as a hardwires flash/firmware area.  At the end of the day, the behaviour is the same.
[12:07] <infinity> ogra_: And that's what we do when we netboot, for instance.
[12:07] <infinity> It's also how my i.MX53 and Panda both run when they're at home.
[12:07] <infinity> s/hardwires/hardwired/
[12:08] <ogra_> infinity, right, thats also what we did on babbage ... though there we had the prob that we had to install to the livefs media while running from it
[12:21] <xranby> wow... i got opengl-es to work on my pandaboard using oneiric  and using the default package sources..  i simply installed every omap4 package and then it turned ON
[12:22] <xranby> i was expecting to get the packages from the ti ppa.. but they where already in main
[12:22] <infinity> *raise brow*
[12:22] <xranby> infinity: want a screenshot?
[12:22] <infinity> There's no way that's hardware accelerated if it involves nothing from TI.
[12:26] <xranby> infinity: http://openjdk.gudinna.com/lwjgl-es/pandaboard-LWJGL.png
[12:26] <xranby> running at 130fps
[12:28] <infinity> xranby: 130fps sounds like software rendering to me.
[12:29] <xranby> infinity: libEGL comes from the SGX omap 4 package
[12:29] <xranby> infinity: this are usning java -> opengl-es bindings
[12:30] <ogra_> infinity, es2gears in SW rendering on the ac100 gets me 20-30 frames
[12:31] <ogra_> i think its about 180 if accelerated
[12:34] <infinity> xranby: I can't find this package you're referring to.
[12:34] <xranby> infinity: let me try generate a list of installed packages
[12:34] <infinity> xranby: dpkg -S /path/to/file
[12:34] <xranby> are the some way to list the packages that got installed  today?
[12:34] <twb> Would that be the "benchmark" that has a --i-acknowledge-this-is-not-a-benchmark option?
[12:39] <xranby> infinity: http://paste.ubuntu.com/707964/    perhaps i got them from the ppa after all
[12:40] <ogra_> twb, i think that was dropped again at some piunt :)
[12:40] <ogra_> *point
[12:40] <infinity> xranby: "apt-cache policy packagename" will tell you where it comes from.
[12:40] <infinity> xranby: But yes, those aren't in the archive.
[12:41] <ogra_> well, we definitely dont ahve ubuntu-omap4-extra in any official archive
[12:41] <ogra_> and you have the sw-center added ppa .list
[12:42] <xranby> infinity: ok i can now confirm that they did come from the ppa
[12:42] <xranby> thanks ti
[12:42] <ogra_> :)
[12:42] <ogra_> send flowers to ndec and his team :)
[12:42] <xranby> ndec: cheers!
[12:42] <ogra_> (or probably better bottles of old wine)
[12:43] <twb> ogra_: well that's bloody stupid
[12:43] <xranby> ndec: i have lwjgl java bindings working on the pandabord using your latest oneiric drivers..
[12:43] <xranby> nice
[12:52] <ndec> yes, gfx libs have been in PPA for a while now. video decoders are coming soon...
[13:02] <xranby> ndec: ok meanwhile i have filed a bug against the LWJGL upstream to add support for all your drivers extensions http://lwjgl.org/forum/index.php/topic,4237.0.html
[13:04] <xranby> ndec: when i install teh libEGL and friens  all .so ends with .so.1   are this intentional?
[13:05] <xranby> the opengl-es userspace applications are looking for the  libEGL.so symlinks  and fails to find the library
[13:06]  * ndec thought we had fixed that.
[13:13] <ndec> xranby: yes, this is true the .so is only in the -dev package (e.g. libegl1-sgx-omap4-dev)
[13:14] <ndec> to me it's more a bug in the applications, ... but we've seen that in the past, and since we cannot change all applications, we need to update our package.
[13:14] <ndec> for now, you just need to install the -dev
[13:14] <xranby> ok thank you for checking
[13:14] <infinity> There's a longstanding history of bainry GL apps looking for .so instead of the actual library. :/
[13:15] <infinity> But we don't ship .so in any library package, for sanity reasons.
[13:15] <infinity> binary*
[13:16] <xranby> hmm i wonder why.. the brainy GL apps simply try to link againt EGL
[13:16] <xranby> 'to get into this state
[13:17] <xranby> -lEGL are passed tot he linker
[13:17] <xranby> and it makes the linker pick to use the .so
[13:17] <infinity> xranby: Err, wait.
[13:18] <infinity> xranby: You need the -dev (and the .so) to link...
[13:18] <infinity> But it should then link to .so.1
[13:18] <xranby> usually you can have EGL mesa installed to build it
[13:19] <xranby> so the app links against the mesa EGL -dev package
[13:19] <xranby> libegl1-mesa-dev - free implementation of the EGL API -- development files
[13:20] <infinity> I'm still not sure what your bug is here...
[13:20] <infinity> You need dev packages installed to compile.
[13:20] <infinity> But not to run.
[13:21] <infinity> If "ldd mybinary" shows that you've linked to an unversioned .so, that's certainly a bug (ie: if you need it at runtime).
[13:23] <ndec> we used to have wrong SONAME in our libs, but this is fixed. objdump -p /usr/lib/libEGL.so | grep SONAME will tell you libEGL.so.1 (and it used to be the .so). the soname is what the app will link against.
[13:23] <ndec> perhaps the mesa-dev package has wrong soname?
[13:23] <infinity> ndec: Almost certainly not. :P
[13:24] <infinity> (If it did, nothing on my system would work)
[13:24] <ndec> xranby: is your app dynamically linked against the .so, or do you have a dlopen to the .so . i think firefox does (did?) that
[13:25] <xranby> let me check
[13:32] <xranby> ldd at points to /usr/lib/libEGL.so   i guess that makes it dynamically link
[13:33]  * xranby are rebooting his board... for some reason the usb mouse/keyboard refused to enumerate
[13:39] <brandini> morning dudes
[15:43] <_Thomas> Does anyone here work with the ubuntu release for the Linaro Origen-board?
[15:43] <_Thomas> (I'm wondering if anyone knows the status of getting HW opengl on that board)
[15:44] <ogra_> _Thomas, as i said in the other channel, ubuntu-arm doesnt support the origen board
[15:46] <_Thomas> ok
[19:07] <stephen__> pandaboard: linaro image vs ti-omap ppa
[19:07] <stephen__> which is better?
[22:59] <brandini> so when I have the load address in uboot, is that looking for KERNEL_BASE_PHYS=0x80300000?