/srv/irclogs.ubuntu.com/2010/10/11/#ubuntu-arm.txt

=== pcacjr_ is now known as pcacjr
=== lilstevie|ZNC is now known as lilstevie
DanaGhttp://www.slashgear.com/pandaboard-offers-ti-cortex-a9-omap4-to-imaginative-devs-04105681/02:25
DanaGSpiffy.02:25
DanaGSo the Pandaboard is really close.02:25
DanaGThough, it looks like form-factor may be just different enough to make it incompatible with Beagle expansion boards.02:28
persiahttp://pandaboard.org/ is already taking developer applications...02:28
DanaGIs the "retail availability" date still NDA?02:29
persiaNo idea :)02:30
persiaAs a result, I believe the answer is probably "Yes", because otherwise I would expect to have heard something.02:30
DanaGThat's a good answer.02:37
=== asac_ is now known as asac
=== JaMa|Zzzz is now known as JaMa
=== amitk-afk is now known as amitk
=== hrw|gone is now known as hrw
hrwmoin08:30
=== vstehle_ is now known as vstehle-laptop
ndecogra: hey!10:22
ndecso release is up, finally!10:22
lag:D10:28
lagAre you please with it ndec?10:29
armin76celebreate!10:29
ndeclag: well extremely!10:29
lagndec: I'm pleased you're pleased :)10:29
ndeclag: it's been a really cool project to work with!10:29
lagndec: So what's next?10:29
lagFor sure10:29
ndeclag: natty ;-)10:29
lagI've enjoyed it10:29
lagWhoooooooo :)10:29
ndeclag: well, next step is to release our public PPA10:29
lagFor sure10:30
lagAre we going to be supporting any more boards?10:31
ndecnot that I know of10:32
armin76zumbi_: celebrate!10:32
armin76then you can support another distro :D10:32
lagarmin76: :-o10:33
lagThere's only need for one distro :)10:33
armin76yeah, and 640k ought to be enough for anyone ;)10:34
ndecarmin76: ;-)10:35
lag:)10:36
=== lilstevie is now known as lilstevie|ZNC
=== ogra changed the topic of #ubuntu-arm to: Ubuntu ARM Discussion & Development | https://wiki.ubuntu.com/ARM | Want to Submit a Bug? https://bugs.launchpad.net/ubuntu/+filebug | Build a rootfs from scratch: https://wiki.ubuntu.com/ARM/RootfsFromScratch | wanna cross build ? see http://idlethread.blogspot.com/2010/09/cross-compilation-redux.html | Maverick is out ! http://cdimage.ubuntu.com/ubuntu-netbook/ports/releases/maverick/
ogra:)10:59
persiaWhy point at cdimage rather than releases?11:00
ograpersia, because thats where our release lives ?11:01
ograshow me the url on r.u.c and i'll point to it11:01
persiaYou're right.11:02
persiaThat's a regression.11:02
persiaWe were on releases.ubuntu.com for karmic and lucid11:03
* ogra doesnt see it as a regression, butu yeah11:04
persiaWell, it indicates we weren't considered part of the release.  Maybe just a social thing.11:04
* hrw agrees with persia11:05
hrwarmel looks like 3rdparty in ubunt11:05
hrwu11:05
persiaAnyway, something to target to be fixed in Natty.11:05
persiahrw, It's most certainly nothing like 3rd party.11:05
hrwbut totally outside of official archive11:05
persiaNo.  Just not blessed with release.11:06
hrwpersia: but also ports.ubuntu.com for last few years instead of merging to archive.ubuntu.com11:06
* hrw hugs apt-cacher-ng11:07
persiaThat's mostly blocked by some technical limitations of our mirror software, sadly.11:07
ograhmm, https://wiki.ubuntu.com/MaverickMeerkat/ReleaseManifest agrees with you11:07
ograi dont think there is code in place to put them on releases.u.c :(11:07
persiaYep.  Go complain to skaet.11:07
persiaThere is code to move from ports/daily{,-live} but maybe not daily-preinstalled.11:08
ograright11:08
ograthough i really wouldnt mind keeping them on cdimage11:09
ograsimply because it's super confusing having them in the same page with all the other images11:10
ogra(and their install instructions)11:10
persiaogra, See, the way to solve that is to make them have the *same* installation instructions :)11:12
ograthats not possible11:12
* persia grumbles at git for being painfully slow and bloated (why download 1GB of stuff for 80MB of data?)11:12
persiaYes it is.  Just requires more work, and more time.11:12
ograthen the install isntructions get confusing11:12
ograyou just move the problem around11:13
persiaI believe the only remaining obstacles are 1) getting everyone to accept a common first-stage bootloader (e.g. UEFI), 2) porting grub2, and 3) ensuring there is *some* flash on every board to hold a vendor-installed first-stage bootloader.11:13
persiaThis is much better than we were two years ago.  I predict we'll be done in only 5 or 6 more.11:14
ograyou are making weird assumptions about HW manufacturers11:14
persiaWhy?  For x86/powerpc the vendors all meet those requirements already.  It's kinda expected for consumer-level general-purpose devices.11:14
ograthere is no flash on any of the publically available boards we support except the beagle C4 which is underpowered for these images11:15
persiaSo?11:15
persiaThat's deficient HW.11:15
ograthere is neither grub but a properly usable u-boot11:15
persiaRight, hence point 2) above.11:15
ograwhich i disagree with11:15
persiawhy?11:15
ograas well as with the other points11:16
ograwe have a properly unified bootloader with u-boot11:16
ograits just missing framebuffer support11:16
persiaDo you not think it makes sense to have first-stage bootloaders managed by folks to do boards (for board-bring-up), and second-stage bootloaders managed by folks that do OSs (for OS bring up)?11:16
persiaEveryone else is doing it.11:16
ograi do ... but thats what we have effectively with x-loader and u-boot11:17
persiaWhy be different ?  Multiple code paths are harder to maintain.11:17
ograwe dont have to maintain them11:17
persiaIf you like x-loader, fine, but x-loader/grub2 is more cleanly aligned.11:17
ograno, its not11:17
persiaPlus, I think it's easier to use the existing ARM port of UEFI than to port x-loader to everything else.11:18
ograsimply because everyone working with these boards has u-boot experience11:18
ograthe only missing bit is framebuffer support so the general enduser could get a menu11:18
persiaSo?  Training is easy.11:18
* suihkulokki thinks u-boot has way more momentum than uefi11:18
ogra++11:19
ograor grub porting (which hasnt even started)11:19
ograand making the boards more expensive by adding NAND doesnt look like a good solution either11:19
persiasuihkulokki, Could be: I've seen ports to about the same number of architectures for u-boot and UEFI.  I don't really care which is selected, but would prefer there to be one-true-first-stage-bootloader11:20
hrwI think that arm boards should go for one simple setup: vendorbootloader + osbootloader. thats done on x86(-64) and works on arm so far with targets which we support11:20
hrwforget about grub/arm11:20
persiahrw, Precisely.11:20
persiaHuh?  Which osbootloader then?  Why not use the same as everyone else?11:20
hrwpersia: we do have it with omap34 - xloader + uboot11:20
hrwu8500/linaro also goes that way: something-from-stericsson + uboot11:21
hrwatmel at91 (armv5): at91bootstrap + uboot11:21
hrwat9200rm: simple-in-cpu-flash-bootloader + uboot11:22
persiableh!  Standardising on uboot would be fine.  Standardising on it as a *second-stage* bootloader just seems like madness, when it can do first-stage just fine, and it can't do useful device-selection.11:23
hrwpersia: *if* uboot can be 1st stage then let it be. on *rest* let it be 2nd11:23
hrwpersia: atmel at91bootstrap fits inside of cpu flash. you cant get uboot there11:24
persiauboot needs a fair bit of work from what I've seen to allow user selection of boot devices, etc.11:24
persiahrw, That's a HW bug :)11:24
ograu-boot is to big for fitting in the ram you have available at that stage of boot11:25
ograyou *need* a smaller first stage11:25
ograindeed it could be derived from the u-boot code11:25
ogra(as x-loader already is)11:26
persiaAnd has been in several cases.11:26
persiaSo sure, maybe I don't actually care that much about standardisation for hwbootloader.11:26
ograi agree that this should be fixed11:26
ograand i know linaro is working on that at least for omap11:26
persiaI still want the *same* osbootloader for every architecture, and think having a single shared hwbootloader would lead to economies of scale.11:26
hrwpersia: so port uboot for basic x8611:27
ograyou wont get that without the vendors agreeing11:27
hrwpersia: most of PC are fine with legacy bios + grub. you cant get grub outside of x86 anyway11:27
persiahrw, It's been done for a while.11:27
hrwand you cant get PC users with uboot11:27
persiaHrm?  grub2/powerpc is reputed to work.11:28
hrwo.. did not know11:28
persiaAnd u-boot claims to support PPC, ARM, AVR32, x86, M68k, ...11:28
* hrw moved to coreboot on one machine at home11:28
hrwpersia: sure. very few x86 chipsets probably11:28
persiaOh, I'm sure.  probably mostly VIA ones, for that matter.11:29
hrwand they do not boot windows anymore?11:29
persiaI still maintain there is scope for economies of scale to have one true hwbootloader, and one true osbootloader.11:29
persiaI wouldn't expect it for the lower-end VIA embedded x86 cores.11:29
hrwpersia: 5 years and EFI will be x86 desktop standard maybe11:30
persiaMost of them can't run current Ubuntu (insufficient instruction support)11:30
* hrw hugs alix.1c with its 2.098s to get init booted from power on11:30
ograpersia, oh, btw, i fail to see how that bootloader discussion solves the initial problem of having different methods to write differnt install media11:35
persiaogra, Identical OSbootloader means one can use identical procedures cross-arch.11:35
ograi still need to write to different media with different tools11:35
persiaAs a result the means by which I make bootable SD for armel is the *same* as for bootable SD for i386.11:35
persiaWhich means the same instructions, etc.11:36
ograso you run brasero to create an SD card ?11:36
persiaNo.  I use usb-creator.11:36
persia(or at least I did last time I tried that for amd64)11:36
hrwpersia: does any of your pc boots stright from SD?11:36
ograright, on armel you cant even start it :P11:37
persiahrw, At least two of them can.  I thikn a third, but haven't tested it.11:37
ograits not installable due to its dep on syslinux11:37
hrwpersia: and by default you need to press some key during reboot to invoke boot menu and select "this SD please" in it11:37
persiaogra, Right.  See, that's a bug in usb-creator: should use grub2: see discussion with superm1 and Bugabundo in -devel whilst you were sleeping.11:37
ograi wasnt sleeping :)11:38
ograand i followed it roughly while my ac100 kernels were building11:38
persiahrw, Actually, no, on one of them it tries SD by default (as I discovered when forgetting to remove the SD when trying to reboot, and being confused why it failed)11:38
persiaogra, Ah, excellent!  Still, like I said, we're years away.  Even if everyone in the world attempted to start achieving it right now, I don't think it's *possible* to have clean alignment in less than 3 years because of chip layout planning timeframes.11:41
ograyup11:42
ograand given that work on devicetree is already going on in kernel and u-boot i expect that to arrive earlier as a unified arm solution11:43
hrwpersia: my desktop can boot from sata/pata/usb/cf/sd/sm/ms/xd but for non-sata I need to go to bootmenu, for pata even to bios11:43
hrwogra: 11.10 not earlier11:43
persiahrw, That's just a side-effect of your hwbootloader.  If there was the alignment and consensus on a single hwbootloader like I'm advocating, we'd be having a comparable experience.11:43
hrwogra: for Linaro 11.05 DeviceTree will not be ready yet11:44
ograhrw, i dont exepct d-t in less than a year11:44
ograbuut thats still quicker than persias estimated 3 years for grub11:45
persia3 years isn't for grub.  It's to fix on-chip ROM codes to not be braindead about what to use to boot hwbootloaders.11:46
persiaAnd 3 years is a lower-bound estimate, really.11:46
persia(as in, if someone decided to start fixing this today, it would take 3 years for the *first* SoC that wasn't broken to be released)11:47
persiaAssuming I don't have kernel issues, are there issues expected with do-release-upgrade from jaunty->karmic->lucid->maverick for armel?11:59
dmartwin go #linaro12:05
dmartwin go #linaro12:05
lagI can feel a new Bootloader coming ... Supports all architectures, all bios', all OS': persia-boot12:13
persialag, That is a frightening concept.  The moreso since I tend to write most stuff in make, which is probably not an ideal language for bootloader development.12:18
lagpersia: :)12:19
lagpersia: I can teach you C - it's a small language12:19
persiathanks, but I've been taught C a few times: I just haven't used it professionally in something like 15 years, amd my hobby C has all been small patches.12:23
=== zyga is now known as zyga-lunch
* ogra sighs ...12:33
ogra3800 mails done, 1400 mails to go12:34
persiareleases are fun :)12:35
lagogra: Are you back home?12:36
ogralag, yes12:36
ograwading through my mail since two days12:36
lagI spoke with ndec this morning - he sounded happy with the release :)12:37
ograon the road i only read what was directly affecting the current work12:37
ogralag, yes, he told me already12:37
ogralag, we still need to solve the sound issue and i need to get the tablet going12:37
ograthats the twoi last bits on my TODO for maverick12:37
lagSounds == userspace (last I heard)12:38
ograoh, and bug 657732 is an intresting one ... i could reproduce it on the tegra12:38
ubot2Launchpad bug 657732 in xaos (Ubuntu) "xaos has artefacts on first frame with -threads (affects: 1) (heat: 8)" [Undecided,New] https://launchpad.net/bugs/65773212:38
lagsound*12:38
ogralag, not clear yet12:38
ogralag, there seem to be initialization bits missing but its not clear where12:39
ograto me it looks like the driver needs the init after the mixers were set12:39
ograwhich would be a clear driver issue12:39
lagQuite12:39
ograbut it could as well be a race in alsa userspace12:40
ograthats why its so hard to pin down12:40
lagI believe Mathieu has been liaising with various kernel and TI people to have this issue solved12:40
ograrama from TI (the omap4 ASoC driver dev) was working directly with lrg on the issue but there was no outcome yet12:41
lagWell if they can't fix it, then we have an issue12:41
ograthey still try i belive12:42
laglrg wrote most of ASoC12:42
ograi know12:42
lagHe would be 'the one' AFAIC12:42
ograand rama most of the omap4 driver ... and he worked on the HW too12:42
lagThen we have the best people 'on it'12:42
ograno12:43
lag?12:43
ograbecause nobody seems to have deep insight in the ubuntu specifics of alsa12:43
lagWhat if we lent diwic to them?12:43
ograas i understand it alsactl restore is supposed to do the same as alsactl init for example12:43
ograbut that either doesnt happen or happens in the wrong order for the device12:44
ogralag, how much does he know about the userspace ?12:44
lagHe is Ubuntu's Audio guy12:44
ograthe prob with alsa is that the line between kernel and userspace is really blurry in ubuntu12:44
lagI think that is the case with ALSA generally12:45
ograthere are many bits expected by the driver that imho should be in userspace12:45
ograbut alsa generally doesnt depend on i.e. pulse to set all mixers :)12:45
ograat least in the rest of the world12:45
=== zyga-lunch is now known as zyga
lagdiwic is afk, but I'll hassle him later12:45
dcordeshi12:45
ogralag, ok, i'll try to get rama on IRC (dont know his nick, i need to find out if TX gets up)12:46
lagWhere is lrg?12:46
lagTX?12:47
lrglag: UK12:47
lagBANG!12:47
ograno idea, i know he uses to live in the UK12:47
lagHello Liam12:47
lrgogra: my panda is very ill12:47
ograheh12:47
lrghey12:47
ogralrg, still ? :(12:47
ogrash*t12:47
lrgogra: hw issue :(12:47
lrgso I'm using SDP now12:47
lrgHey Lag12:47
ograshould be fine as well12:47
lagWhere in the UK are you?12:48
ogralrg, any idea about the issue ?12:48
lrglag: just north of Edinburgh12:48
lagJust over the bridge?12:48
lrgogra: afaict the USB hub is a bit broken12:48
lag:)12:48
lrglag: yes12:48
ogralrg, i meant the sound issue with the SDP4430.conf :)12:48
lagThat's a nice area12:50
lrgogra: ah, I'm just starting to look now that I have 10.10 running on the SDP, although ubiquity kept crashing for me so I had to do some manual hacks to setup a user account etc12:50
lagDo you know Queen's Ferry?12:50
lagDo you know Queensferry?12:50
XorA|gonelag: that would either North or South, there is no such town named on its own12:50
ogralrg, kept crashing ? how ?12:50
lrglag: Yes, very close :)12:50
=== XorA|gone is now known as XorA
lagThat's a scary little place if you're not a 'local' :)12:51
XorAlag: Im going to assume from your IP your in Bristol or South Gloucestershire :-)12:51
lagBristol12:52
lrgogra: it would crash after 30secs in the help pages12:52
ograhrm12:52
ograwith the final release ?12:52
lagWhere are you XorA?12:52
lrgno, a few days ago12:52
ograi know it works for TI Nice who have tested it12:52
ograah12:52
XorAlag: Edinburgh, but I used to work in Aztech West many years ago12:52
lag@ST?12:52
XorAlag: yes12:52
lagI have a few friends there12:53
XorA1999-200012:53
lagThey didn't work there then :)12:53
lrgogra, lag: going for lunch now - will be back shortly12:53
ograk12:53
XorAworked on the disastrous project for ST GFX card12:53
lagSure12:53
lagI believe they've had a couple12:54
lagogra: It looks like diwic specialises in userspace :)13:12
ograawesome13:13
lagI will speak to the chain and see if we can free him up13:13
lagHi diwic13:16
lagThanks for joining13:16
diwicnp13:16
lagogra: ping13:16
ograyes, i'm here :)13:16
lagdiwic: We have some rather taxing issues with regards to audio on the Panda13:17
lagWe have two kernel audio experts on the case13:17
lagBut we are lacking in userspace knowledge13:17
diwicokay13:18
lagCan we set up some kind of brain storming session when lrg is back and rama is available13:18
lagogra: diwic: ?13:19
lagWould you be up for that?13:19
diwicsure, at what time approx?13:19
lagThat would depend13:20
lagogra: Where in the world is rama?13:20
ograTX13:20
lagSo it would be afternoon13:20
ograyeah13:20
laglrg is at lunch currently13:20
diwicMy work day is up in ~ 3 hours13:20
ograand i need to get him to IRC first :)13:20
lagSure13:20
lagIf we can't do it today, I'll set a meeting up and we'll do it later in the week (perhaps tomorrow)13:21
lagogra: Do you have his contact details?13:22
ograno13:22
lagOkay, we'll hunt those down13:22
lagThen speak to lrg13:22
ograright13:22
lagdiwic: I'll let you know more as soon as I do13:22
lagdiwic: Thanks for your time13:22
ograhe knows them for sure13:22
lagWe'll get them one way or other13:22
lagrobclark will probably know them too13:23
diwiclag, okay. For today I can't be online when my work day is up, but tuesday or wednesday might work better, if time zones are a problem13:23
lagdiwic: No problem - we'll work something out13:23
cjjnjusthello, when porting linux to arm, I add printascii("bara bara\n"); in start_kernel. It print other string, and then trap into abort. What wrong?13:26
lagcjjnjust: When you say it prints another string, what do you mean?13:33
cjjnjusti mean it print the other strings. the string is define in other place.13:35
lagYou mean it prints out other strings successfully?13:35
lagIt's just yours that it does not?13:35
cjjnjustlag, the string is in other function, it seems like address error13:36
lagThat's probably what it is then13:36
lagCheck in: arch/arm/kernel/head.S13:37
cjjnjustyes, I have check it again and again...13:37
cjjnjustonce  call a function with args, it will trap into abort mode.13:38
cjjnjustbut it can run to start_kernel, when printk("%s",linux_banner); it go to  abort mode.13:39
lagcjjnjust: Which kernel are you using? And which Arm chip are you trying to use?13:43
cjjnjustlag,2.6.35 s3c241013:44
lagThe s3c2410 is already supported13:46
lagarch/arm/configs/s3c2410_defconfig13:48
cjjnjustI know that .13:48
lagThen why port it?13:48
bercolag: ogra: can you please invite me to your audio brainstorming too?13:49
lagberco: Of course13:49
cjjnjustI just want to add some driver on it .13:49
cjjnjustuse the default config can't work.13:50
lagYou can use it as a base13:50
lagThen edit it13:51
lagAll of the bring-up and debugging (including earlyprintk) code as already been implemented13:51
lagAll of the addressing setup code too13:52
cjjnjustIn fact, I just do what you say.13:53
* ogra_ac dances ...13:54
ogra_acgot an ubuntuized kernel for the ac10013:54
lagcjjnjust: Failing that, take a look at this: http://stackoverflow.com/questions/1359919/arm-data-abort-error-exception-debugging13:54
lagWill enable you to dump the stack and see where it's going wrong13:55
lagogra_ac: Congrats ... booting?13:55
ogra_aclag, sure, its the toshiba source but with ubuntu config13:55
ogra_ac(still 2.6.29 indeed)13:56
ogra_acbut all the android crap is gone, i have swp and all modules i can imagine13:56
lagSo long as you're enjoying yourself :)13:56
ogra_aci|ll test later what mavericks udev thinks about it :)13:56
ogra_acs/swp/swap13:57
lagI knew what you meant13:57
ogra_aclag, well, i have a netbook on which i can do ubuntu arm development, what more would a human need :)13:57
cjjnjustthanks13:57
ogra_aclag, compiling the omap4 kernel package takes just 2h here13:58
ogra_ac(on external USB disk)13:58
lagcjjnjust: No problem - best of luck13:58
lagogra: Here being?13:59
ogra_acon the ac10013:59
lagIt only takes me 15mins :)13:59
ogra_acthats a third of what the buildds take13:59
ogra_acand 30min less than on the panda14:00
ogra_acquite impressive imho14:00
armin76wow, hows that? tegra and panda should be same speed, no?14:04
ogra_acsimilar14:05
armin76but 30mins is a lot14:05
rsalvetiogra_ac: you should be fine with udev: http://paste.ubuntu.com/510877/14:17
rsalvetiminimum is 2.6.2714:17
rsalvetiogra_ac: with usb on omap4 it takes less than 2 hours14:17
rsalvetisomething around 1:50 minutes14:18
ogra_acwith the new mem speed ?14:18
rsalvetibut this I got when testing with 1gb14:18
robclarkogra_ac: on panda with usb disk, I built kernel (just running 'make uImage') in ~35 min..14:19
rsalvetiogra_ac: I have the timing with new mem speed, just need to boot my board14:19
ogra_acrobclark, ubuntu package ...14:19
robclarkahh..  I didn't try that14:19
rsalvetirobclark: so, how is it going with usb disk?14:19
ogra_acrobclark, i build zImage for the tegra in about the same time14:19
robclarkbut on MMC card it was ~1hr.. and with USB disk ~35min (for make uImage)14:20
* rsalveti is setting up all the boards again14:20
ograbug 65728115:00
ubot2Launchpad bug 657281 in ubuntu "Kubuntu Maverick on Omap3 & Omap4: screen goes black and never comes back (affects: 1) (heat: 8)" [Undecided,Incomplete] https://launchpad.net/bugs/65728115:00
diwiclag, any news on the brainstorming meeting? I'll have to go in an hour.15:11
lagNo sorry15:11
laglrg: Are you around?15:11
lagdiwic: I can't see it happening today15:11
lagI'll try and sort something out for another day15:12
lagXorA: Are you in the same place as lrg?15:12
diwiclrg, isn't that Liam? I think he knows ALSA userspace better than I :-)15:13
XorAlag: he is the wrong side of the river15:13
lagAh, do you both work from home?15:13
ogradiwic, but not the ubuntu specifics about it i guess15:14
ograi think our setup has a lot of specialities15:14
lrglag: I'm here, diwic lets do this tomorrow as I'm seeing some funny things with our mixers that could cause config to fail15:15
ogra\o/15:15
ograsomeone with a clue15:15
ograYAY !15:15
diwiclrg, aren't you the one designing the use case manager?15:15
lrgI need to get the mixer thing sorted out and then I should have a better idea whats going on with the config15:15
lrgdiwic: yes, ucm is in perex ucm branch - he's made some changes and I need him to push to master.....15:16
diwiclrg, so then you probably know alsa userspace as well as I, but I don't mind participating in a brainstorming, if not lrg's mixer fixes is what we need.15:17
laglrg: Do you have Rama's contact details?15:18
laglrg: Is it worth pulling him in?15:18
lrglag: yes, rsrimushnam@ti.com15:20
lagOkay, I'll send out an email15:20
lrglag: Rama is in Dallas so he's probably around atm15:20
lagHow does 15:00 UTC grab everyone?15:20
lagTomorrow15:20
lrgand he does have a working pandaboard15:20
lagOr Wednesday15:20
lrglag: most times are fine for me, Rama has more meetings though15:21
lrgbest find a time suitable for Rama andI'll join15:21
lagI'll email him and try to get him on here15:21
diwicworks for me, but 14:00 UTC would be even better.15:21
Baybalis this day a some kind of a holiday in North America?15:22
lagdiwic: The Arm meeting is at 14:00 UTC15:23
lagBaybal: You mean today?15:23
ograBaybal, yes15:23
ograits the day for that guy who got lost trying to find india15:23
diwiclag, so 15:00 UTC tomorrow it is then?15:24
lagBaybal: It's Columbus day15:24
lagdiwic: Provided Rama can make it, yes15:24
laglrg: What's your email address?15:25
lagberco: Likewise15:25
lrglag: lrg@slimlogic.co.uk15:25
lagShall we have it in here, or in it's own channel?15:27
lrglag: own channel may be best15:29
lagNo problem15:29
prpplaguelrg: greetings earthing15:31
bercolag: ok for me at 15.00 UTC15:38
bercolag: d-bercovitz@ti.com15:39
armin76i'm invited too? *g*15:40
lagarmin76: If you want to be15:41
armin76j/k15:44
lagprpplague: David, what are you doing tomorrow @ 15:00 UTC?15:53
diwicsee you in 24 hours then15:54
ojnarmin76/marvin24: it's using a nonstandard load address. I'll see why nvidia decided to change that this week, but for now, if you change it manually in the sources you can boot it through fastboot15:56
ojnOr, you can flash it instead of fastboot as well.15:56
marvin24ojn: you mean from 0xe08000 to 0x108000?15:58
marvin24last is reported by nvflash15:58
marvin24first is TEXT_BASE in config.mk15:58
ojnyeah15:59
lagndec: sebjan: Do you want in?15:59
ogra_acojn, what are you gusy doing about nvrm_daemon ?16:01
ogra_ac*guys16:01
* marvin24 hope it's being killed16:03
sebjanlag: yes, please add me. I'll join if I can16:03
ogra_acmarvin24, ++16:04
ogra_acbut you wont get any power management or sound without it16:04
marvin24ogra_ac: not if you have the right documentation16:06
marvin24I wish there would be *any* public docu about tegra16:07
marvin24but even the supplemental chip docu is hidden16:07
ogra_acyeah16:08
ogra_acevil16:08
ogra_aci wish TI would have a netbook like the ac100 already16:08
marvin24TI is not better, search on the TI website for TPS658600A ...16:09
marvin24controller on the ac10016:09
marvin24only features, no docu (for free)16:09
ogra_achmm, i thought i saw wiring docs at some vendor16:09
ogra_acplatforms like beagle or panda are surely well documented publically16:10
armin76ogra_ac: touchbook?16:10
ogra_acarmin76, shudder ?16:10
ogra_acarmin76, did you ever hold one in your hands ?16:11
armin76nope16:11
ogra_acand did you ever hold an ac100 in your hands to compare16:11
armin76nope either16:12
marvin24armin76: care to make another try with uboot on harmony with the modified loader address?16:14
armin76marvin24: i'll try latter today16:14
marvin24that would be awesome!16:14
prpplaguelag: not sure why?16:15
lagWe're having a Panda Audio Brainstorming session if you want in16:16
egost_hi16:17
egost_i'm playing with pre build ubuntu  on blaze. could someone tell me the usename and passwd please!16:17
hrwogra_ac: ac100 on photos looks like polished product. touchbook... looks like beagleboard on steroids16:17
ogra_acegost_, i guess yuo need to ask in an internal channel about that image ... but i would suspect something like ubuntu/ubuntu16:18
ogra_achrw, yeah, and one is wobbly all over and falls on its face if you remove your hands from the kbd, the other weights nothing and has a super stable case16:19
egost_i tried ubuntu/ubunto  doesnt work16:19
egost_ubuntu/ubuntu  i mean16:19
rsalvetivstehle: ^16:19
ogra_acegost_, where did you get that image ? internal server ?16:19
egost_http://cdimage.ubuntu.com/ubuntu-netbook/ports/daily-preinstalled/current/maverick-preinstalled-netbook-armel+omap4.img.gz16:20
rsalvetiif not you can install a new one by following https://wiki.ubuntu.com/ARM/OMAPMaverickInstall16:20
ogra_aci would suspect there is a readme file stored nexdt to it16:20
ogra_acegost_, oh, that wont work out of the box16:20
ogra_acegost_, see rsalveti's link16:20
rsalvetiegost_: by getting a preinstalled image you need to change mlo, u-boot and unitrd16:20
egost_i did it16:21
rsalvetifollowing the link should give you a working image16:21
egost_i use custom uImage and MLO16:21
egost_it booted16:21
rsalvetithat's the problem16:21
egost_but i cannot login16:21
ogra_acit didnt exec the uInitrd16:21
rsalvetiat least if it doesn't work with the default uinitrd and u-boot, it's not going to open you the installer16:21
rsalvetithen you don't have a user at the system16:22
ogra_acno, you cant loig in if the uInitrd didnt trigger the user setup16:22
egost_ok i understand16:22
rsalvetiegost_: try first by using these files, then after installing it you can change it the way you like it16:22
egost_ok i will try it  ths!16:23
jcrigbyogra_ac, I have a panda board with serial number 750-2151-002(b) and one red wire.  Can you tell me what version that is?16:26
ograa red wire ?16:26
ograhmm16:26
GrueMasterSounds like an ES2.016:26
ogracould be es2.016:26
jcrigbyrework16:26
ndecjcrigby: is it green or black?16:26
jcrigbyndec, black16:26
ndecjcrigby: when did you get it?16:27
GrueMasterBlack:  ES2.0 8L16:27
hrwegost_: take sd card from board, plug to linux desktop, edit /etc/shadow16:27
rsalveti6layers16:27
GrueMasterhrw: Not the issue.16:27
jcrigbyndec, got it from asac at Prague16:27
rsalvetiusually the black with red wire is the es2.0 6 layers16:27
rsalvetithat doesn't boot anymore with our kernel16:27
ograyeah16:28
ndecjcrigby: black is 6-layer board16:28
GrueMasterOops, rsalveti is right.16:28
jcrigbyok, that is what I thought16:28
jcrigbythanks for the verification16:28
jcrigbyJamieBennett, ^^16:29
hrwjcrigby: another device to be bricked to the wall?16:29
rsalvetijcrigby: you can still boot it by reverting patch http://kernel.ubuntu.com/git?p=ubuntu/ubuntu-maverick.git;a=commit;h=b5485d193f8a422901fe7e40154295097012186016:29
jcrigbyhrw, :)16:29
rsalvetiand using the older x-loader16:29
ndecjcrigby: hrw: that can be a decent build machine, though..16:29
rsalvetithat's how I'm using mine16:29
ndecndec: I think beta image used to work on this board16:29
ograndec, until next security fix of the kernel indeed16:30
GrueMasterRC should work as well.16:30
JamieBennettjcrigby: that is sad news :(16:30
GrueMasterBeta was ES1.16:30
ndecogra: so you can use it as a decent build if you disable security updates ;-)16:30
ograright, beta only worked on es116:30
ograndec, heh, yeah16:30
rsalvetirc should be the best one for es2.0 6 layers16:30
* JamieBennett will be pestering ogra and GrueMaster for some Linaro testing then ;)16:30
* GrueMaster thinks JamieBennett can buy his own board when they ship. :P16:31
hrwJamieBennett: soon I will be able to do linaro testing on pandaboard too16:31
hrwGrueMaster: first they have to sale...16:31
JamieBennettGrueMaster, hrw: need it sooner than that16:31
hrwJamieBennett: refresh TI contacts?16:32
rsalvetiJamieBennett: what do you need to test on it?16:33
JamieBennettrsalveti: I need to see if our Linaro images work for a start16:33
ograJamieBennett, as long as they use the ubuntu kernel and bootloader :P16:34
rsalvetiyup16:34
rsalvetiand x-loader16:34
* ogra meant to summarize u-boot and x-loader under bootloader :)16:34
JamieBennettrsalveti: right they do but that doesn't mean our netbook/ALIP/Plasma Handset/Headless images work, just that they have the right ingredients ;)16:35
ograheadless will, for sure16:35
GrueMasterheh16:36
ograall the rest depends on what driver support you need i guess16:36
GrueMasterJamieBennett: Can't you test your images with qemu?  :P16:36
* JamieBennett bemoans the lack of hardware16:36
ogralol16:36
JamieBennettGrueMaster: heh16:37
ojnogra/marvin24: nvrm is slowly getting replaced, but it's not a small project16:38
ojnpower management should work fine without it16:38
ograojn, awesome to hear16:39
ograwell, frequency scaling surely doesnt16:39
hrwJamieBennett: there always will be lack of hardware16:39
ojncore should do. other parts of the chip is harder to deal with, as with all other socs.16:39
ograah16:39
ograwell, i'm kind of stuck with the toshiba kernel16:40
ojnyeah, then you're sol16:40
hrwogra: binary or source?16:40
ograhrw, source16:40
ograbut 2.6.2916:40
hrweclair one?16:41
ojnnvidia legacy sources are _horrible_. wince code ported over to linux, as far as I can tell.16:41
ograyeah, there are a lot of references to windows in the code16:41
ograeven to x8616:41
ograhrw, 2.6.29 one16:41
ograthere are android patches but i disabled them all16:42
GrueMasterSweet,16:42
NCommanderojn: what's nvidia legacy?16:42
ojnNCommander: their old port, not the one re-done by us and android.16:42
ojnus, them and android, I should say.16:42
ograweho cares about android :P16:42
NCommanderojn: you mean tegra or something?16:43
ojnogra: they're doing good work on the base platform support in this case. :-)16:43
ojnNcommander: yeah16:43
ograojn, indeed :)16:43
NCommanderojn: is there a less sucky base port?16:43
ograbut i dont want their modules16:43
ojnNCommander: yes16:43
ograNCommander, forget about it on the ac10016:43
ojnogra: the linux-tegra- branch in their kernel/tegra.git is android-free16:43
* NCommander is a cyanogenmod developer, and we've re-ported several HTC device ports16:43
NCommanderojn: trivial to re-add16:44
NCommanderI have a list of the patches somewhere16:44
ojn(tne android-tegra- branch is the one with the android pieces)16:44
NCommanderBut its mostly openbinder, the rest is just gravy16:44
ojnNCommander: I don't know what you're talking about. I'm _not_ working on android, and I'm not interested in it.16:44
Baybalcan we port upstream onto it?16:44
Baybalgood morning16:44
NCommanderojn: I was just bringing up that the android stuff isn't really that crufty16:44
NCommanderBaybal: onto what? AC100?16:44
Baybalyes16:44
ojnNCommander: yeah, they seem to be active on .35 at the moment16:45
ograBaybal, which upstream ?16:45
NCommanderProbably. Depends how extensive tegra's mods are, but I don't really feel like playing with Russell King when I'm not a kernel dev16:45
Baybalkernel.org16:45
ograBaybal, if nvidia sends their surces upstream, sure16:45
ojnBaybal: kernel.org is not quite there for tegra. Stuff is queued up, but what's already merged is just very basic stuff16:45
ograNCommander, tegra isnt so bad16:45
NCommanderogra: I haven't poked their tree. If they're using proper git, then that helps a lot16:46
ograNCommander, toshiba adds sauce on top that needs a lot of work16:46
ojn*yawn*16:46
ojnOk, have fun guys.16:46
NCommanderogra: hrm. If I ever bother to port AC100 to cyanogenmod (which is doubtful, Ubuntu is more interesting), I'll probably work of NVIDIA's tree and then manually add the Android kernel bits16:46
NCommanderLot less messy.16:46
Baybalbut I hope, Tosh didn't remap registers?16:46
ograNCommander, i thought david got an ac100 for you16:47
NCommanderogra: I won't get it until UDS.16:47
NCommanderogra: and I still have to track a power plug down for it, as davidm didn't bring it16:47
NCommanderogra: I haven't looked at the kernel situation yet, but I'd love ot see the kernel in archive if possible ...16:47
ograNCommander, only the wire16:47
ograNCommander, no way16:48
ograNCommander, you only need the cable from the wall to the power brick16:48
NCommanderogra: I take it stock tegra kernel doesn't boot?16:48
ogracosts $2 or so16:48
NCommanderogra: ah, i got like 10 of those at home, and GrueMaster probably has a box full16:48
ograNCommander, no, there are ODM patches16:48
ograi have the source and have a properly ubuntuized kernel running here16:49
ograbut o wont forward port it16:49
NCommanderogra: ODM?16:49
NCommanderoh16:49
NCommanderd'oh16:49
ojnogra: ah, you were able to replace the kernel after all? nice16:49
ograand that still only gives you basic support16:49
NCommander(and yuck)16:49
ograojn, yeah, got the source on friday16:49
Baybalis this Toshiba oem_ec a register based interface?16:49
ograhuge ugly tarball16:49
NCommanderogra: got a link?16:49
ograhttp://share.grandou.net/ac100/16:50
* NCommander downloads16:50
ograNCommander, i also have a boot.img and kernel modules for it packaged up16:50
ograping me once you get yours16:50
ograeven though i might have a proper maverick image ready by then16:50
NCommanderogra: thanks. I'm going to just look to see how far this tree diverges from cm-kernel tip.16:50
ojnogra: what wifi chipset does it use?16:51
NCommander(probably very, but depending on how (in)sane it is, I might be able to convince someone to do a port)16:51
ograralink 307016:51
ograor at least something that runs with that driver16:51
ojnok16:51
ograrsalveti, btw, thanks for the udev tip, maverick udev runs fine now16:52
rsalvetiogra: cool16:52
ograsadly xorg still needs an xorg.conf and the kbd driver :/16:52
rsalvetiwell, getting better16:52
rsalveti:-)16:52
ograbut that and disabling pulse should be the only hacks left16:52
ograi wish i knew what nvrm_daemon does to enable the codec16:52
NCommanderogra: magic poke? :-)16:53
hrwogra: I think that it will take some time before xorg.conf will finally die. on my desktop I still have one16:53
ograi have proper alsa devices but nothing attached until nvrm:daemon runs16:53
ograhrw, well, i need it for the input devices16:53
hrwogra: I for output ;D16:53
ograthe stuff that was supposed to magically work16:53
ograoutput works out of the box through framebuffer16:54
ograand its not slow16:54
GrueMastersame here on my desktop.  Nvidia dual monitor configuration.16:54
ograwhat bothers me is that we blantly dropped the kbd driver16:55
hrwGrueMaster: ati opensource + two monitors here16:55
ograsince "all keyboards work through evdev now"16:55
ograwhich obviously isnt true16:55
vstehlersalveti: Hi! I wanted to warn you: I will rename the sgx-dkms package on the PPA.16:55
rsalvetivstehle: ok, thanks :-)16:55
vstehlersalveti: It will be called 'pvr-omap4-kernel-dkms'16:55
vstehlersalveti: (Could not find a longer name ;)16:56
rsalvetiyeah, better one16:56
hrwogra: you meant "all 8bit keyboards" I think16:56
ogravstehle, i culd make one up if you ran out of chars16:56
vstehleogra: :)16:56
ograhrw, well, i doubt mine has more than 8bit stuff ... but it isnt recognized at all16:56
ograi need to compile the kbd driver package from debian to make it work16:57
NCommanderogra: so you need to use the tobisha tree and not the tegra tree to get a working AC100 kernel? :-/16:57
ograand need the xorg.conf for all input devices16:57
hrwogra: I have keyboard which has 9bit keys... works on linux console but not in x11 ;(16:57
ograNCommander, xactly16:57
ograNCommander, the diff between both would be intresting16:58
NCommanderogra: I suspect its got to be relatively minimal16:58
ograNCommander, but the toshiba tarball doesnt have git references in it i think16:58
NCommanderyeah16:58
NCommanderI'm looking at that now16:58
ograso diff and patch are your tools here16:58
ograheh16:59
* ogra just tried a minimally modified panda image on the ac10016:59
ograworks fine16:59
GrueMastersweet.16:59
ograxorg, xserver-xorg-input-kbd, diverting the pulse binary and unpacking my modules tgz on it17:00
ograeven my BT dongle works now17:00
rsalveticool17:00
NCommanderogra: looking at nv-tegra.nvidia.com, it looks like everything Tobisha did is in the 2.6.29 branch ...17:02
marvin24_DTogra: do you think it's worth a try to get u-boot running on ac100?17:02
ogramarvin24, sure, why not ... i would love to not need a second machine to upgrade the kernel17:02
ogramvflash is really annoying17:03
* NCommander would agree except we have no user-accessible serial port so it would be a bit tricky17:03
marvin24_DTgood, I made a small patch based on harmony, but no luck up to now17:03
NCommander^- to dothe port17:03
marvin24_DTI don't know how to setup this mux things and there is no docu17:04
ograNCommander, there is an OTG port ... if you can wire it up in u-boot to be serial you are all set17:04
* rsalveti lunch17:04
ograbut even without, having a bootloader that can read from vfat is still better than flashing17:04
NCommanderogra: that requires fun low-level code :-/17:04
NCommanderand a USB stack17:04
ograi'm sure tegra has something like that17:04
ograthere is kernel code to steal from17:05
* NCommander git clone's the tegra kernel17:05
NCommanderI'm *really* curious on how far Tobisha changed things17:05
ograthey added definitely LCD code17:07
NCommanderright, but that's a driver17:08
NCommanderAs long as arm/arch/* hasn't been heavy modified, it should be pretty straightforwad to get a stock 29 kernel going on the ac10017:08
ogralook for the paz00 (or similar) dir17:09
ograthat should have the ODM specific bits17:09
NCommanderogra: Tobisha forked off NVIDIA release 9.12.1017:11
ograright17:12
NCommandernow to just generate a diff and to see how extensive they changed things17:12
hrwplaying with vendor kernels... suxx always17:13
* NCommander loves git clone --reference17:14
NCommanderhrw: well, we'll see how high the suckage factor is17:14
hrwyes17:15
marvin24_DTNCommander: there is a patch at http://attachments.wetpaintserv.us/yl1upugLD3-jVOiRg2dsLQ136098817:15
marvin24_DTagainst eclair-9.12.1217:15
hrwlot of time passed since http://marcin.juszkiewicz.com.pl/2007/06/21/extracting-diffs-from-vendor-kernels/ post... I learn new tricks, more people to ask ;D17:15
hrwanyway - time for me17:15
NCommandermarvin24_DT: http://git.chromium.org/cgi-bin/gitweb.cgi?p=u-boot.git;a=summary17:17
hrwhave a nice rest of day17:17
=== hrw is now known as hrw|gone
marvin24_DTNCommander: yeah, I know17:17
marvin24_DTlp/~marvin/ac100/u-boot has some patches17:18
marvin24_DTproblem is, no console ...17:18
ogramake it hardcoded look for mmcblk1p1 and fatload boot.scr17:19
ograthe you dont need a console17:19
ogra*then17:19
ograits a bit of a blind flight, but if it works you can just edit boot.scr17:19
NCommanderhrw|gone: ugh, this is a nasty pile of hacks17:21
NCommanderer marvin24_DT & ogra17:21
ograheh, news at ten17:21
NCommanderits not super bad ...17:22
NCommanderBit messy, but servicable17:23
NCommanderif tegra had a proper master tree, I'd probably look at porting this onto their HEAD ...17:24
ograNCommander, ojn has a good tree i heard17:24
NCommanderogra: for Tegra or AC100?17:24
ografor tegra17:24
NCommanderhrm17:25
NCommanderI'll have to look at it, but this is actually not bad at all17:25
NCommanderThere's some board specific code that probably needs refactoring, but the vast maority are minor driver tweaks17:25
=== xfaf is now known as zul
marvin24_DTogra: I'm a little confused how to setup an u-boot script, I guess you are more familar with that (and u-boot)18:39
=== XorA is now known as XorA|gone
marvin24_DTogra: I added you to my launchpad (u-boot+kernel) project in case you are interested18:45
armin76marvin24_DT: -TEXT_BASE = 0x00e0800018:52
armin76+TEXT_BASE = 0x0010800018:52
armin76thats what i should do?18:53
devilhornsogra, have a minute ?18:57
marvin24_DTarmin76: yep19:03
marvin24_DTit at least changes something here (usb went way)19:03
armin76oh, damn19:03
armin76i did it wrong yesterday19:03
marvin24_DT?19:04
armin76i put u-boot instead of u-boot.bin :D19:04
armin76it works now19:04
marvin24_DTwith changed address?19:04
marvin24_DTor without?19:04
armin76with19:05
marvin24_DTgreat!19:05
armin76let me test with the original one19:05
armin76nope, doesn't work19:07
marvin24_DTok, thanks for testing!19:09
armin76http://dpaste.com/256412/19:10
armin76np, thanks for telling me about the load addr :)19:10
marvin24_DTthanks should go to ojn19:10
marvin24_DTbut it seems that it does not find the flash?19:11
armin76seems like it19:11
armin76marvin24_DT: but you can't kinda flash uboot, can you?19:28
marvin24_DTarmin76: you could try to flash it as a kernel and see if it boots up19:28
ojnyeah19:29
ojnthat's the easiest way to do it on a current fastboot setup19:29
ojnjust give it as the kernel argument19:29
marvin24_DTojn: why is the flash not detected?19:29
marvin24_DTstill TODO?19:29
armin76i have tried that, didn't work19:31
marvin24_DTmaybe again a problem with the init vector19:32
ojnmarvin24_DT: what flash?19:35
marvin24_DTsee http://dpaste.com/256412/19:35
ojnarmin76: ok. I haven't tried recently, so things might have changed19:35
ojnmarvin24_DT: Ah, emmc. Well, probably because the emmc is not on that controller on your board? which one is that on? harmony?19:36
ojnTHe MMC that is probed is the one next ot the PCI-e slots there, not the bayonet one. :(19:36
armin76bayonet?19:37
marvin24_DTprobing is done in board/tegra2/common - correct?19:37
ojnarmin76: the one that klicks in and out. the one ont he other side doesn't.19:39
armin76oh19:39
ogra_acwohoo !!19:39
ogra_ac3G works19:39
ogra_achorrible lag but enough for IRC19:40
ogra_acasac, is it normal that i have to configure the connection before inserting the SIM ?19:41
* ogra_ac found that very confusing19:41
marvin24_DTah, probing is done in tegra_sdmmc19:44
* marvin24_DT needs a serial console19:44
hrw|goneNCommander: that was over 3 years ago post21:05
hrw|goneah..21:05
* hrw|gone off21:05
=== lilstevie is now known as lilstevie|ZNC
vstehle-homelool: Hello Loïc, are you awake? I am desperately looking for an URL about "croco". Would you have this at hand, please?22:05

Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!