**** BEGIN LOGGING AT Thu May 06 02:59:56 2010 May 06 03:39:23 hmm, how do I get the u-boot to user the initramfs flashed by flash-kernel? May 06 08:22:40 lool: that's great -- i wish i could have more time to do this kind of work :) May 06 09:35:01 lool: if your around, can you take a look at approving this: https://blueprints.edge.launchpad.net/ubuntu/+spec/mobile-maverick-arm-improved-subarch-detection/ May 06 09:55:55 rcn-ee, so looks like we have a workaround for the rootstock hang ! May 06 10:04:25 ogra: \o/ May 06 10:12:09 ogra: what do you think of using multistrap instead debootstrap? May 06 10:12:52 what does it gain us over just using the mirror argument to rootstock ? May 06 10:13:31 note that there is no problem to point it to a debian mirror ... it will likely just break on the hardcoded ubuntu stuff after debootstrap atm May 06 10:13:52 which someone will have to fix for debian if a debian compatibility is wanted May 06 10:14:04 ogra: oem or thrid parties repositories to be supported May 06 10:14:16 s,thrid,third May 06 10:14:41 ogra: i'll have a look for debian compat but in terms of multistrap May 06 10:14:55 if you find it appropiate May 06 10:14:58 does it behave like debootstrap otherwise ? May 06 10:15:14 i have no objections if it doesnt break the current behavior May 06 10:15:37 multistrap behaves like debootstrap, but you can also add $your_business_mirror for customized packages May 06 10:15:51 (i.e. it has to work with qemu-arm-static ... and should support all the ubuntu debootstrap scripts we use by default) May 06 10:15:52 ogra: sure, i'll try not to break current May 06 10:16:50 qemu should not be a problem, I'll have a look to debootstrap ubuntu stuff, but do you have a patched debootstrap? May 06 10:17:23 in ubuntu ? just a bit i think May 06 10:17:49 it might be good then to add a switch so one can pick up which on to use (atm) debootstrap or multistrap May 06 10:17:57 hmm 1.0.20ubuntu1 ... the version indicates that everything should be in debian May 06 10:18:42 * ogra reads the description May 06 10:18:47 "A debootstrap replacement with multiple repository support" May 06 10:18:55 the word replacement scares me a bit May 06 10:19:27 well, it is just a matter of having a switch for each one of them (at the moment) May 06 10:19:52 switching back and forth should be no prob May 06 10:19:58 ogra: you might not see it right now, but I only see benefits from using multistrap May 06 10:20:38 it only scares me that i know debootstrap is maintained by the same person in ubuntu and debian so i can 100% rely on it to work for both May 06 10:20:48 without having to fix patch or touch it at all May 06 10:21:08 Also, all the rest of our tools use debootstrap, so it would make something the odd one out. May 06 10:21:08 yes, that's a plus, we maintain multistrap and we are not hostile May 06 10:21:19 but as i said, being able to switch between debootstrap and multistrap shouldnt be any prob at all May 06 10:21:20 we as emdebian May 06 10:21:41 zumbi: If you'd be happy to *also* maintain it in Ubuntu, that makes the decision less awkward :) May 06 10:21:49 ogra: Do you support --debootstrap-mirror already? May 06 10:22:13 persia, i support only the mirror atm May 06 10:22:18 persia: sure, do i have to fins a sponsor for ubuntu or do an nm? May 06 10:22:29 no reason --debootstrap-mirror couldnt be added with a few lines though May 06 10:22:33 (apologies for my spelling) May 06 10:22:44 There are no maintainers in Ubuntu, so everything is kinda NM (but we pretend we're in Uploaders without changing it). May 06 10:22:49 zumbi, we sync from debian :) May 06 10:23:02 If you can get a single source tree that works well for both environments, then we can just sync, which is easiest. May 06 10:23:07 if there is ubuntu support in the debian package we'll automatically get it May 06 10:23:39 ok, that should not be a problem or if it is we should work on it to fix it. May 06 10:23:41 if you look at debootstrap, it just has two ubuntu scripts (and a bunch of links) added over debian May 06 10:24:21 hmm, no i lied, its a few more May 06 10:24:56 9 actually May 06 10:25:07 but using the same since the gutsy release May 06 10:25:17 anyway, i'll have a look sometime, do not expect any soon :) May 06 10:25:27 ok May 06 10:25:37 btw, wookey will be arround at uds May 06 10:25:43 oh, nice May 06 10:25:45 he is our project leader May 06 10:26:00 i met him at a CELF in amsterdam May 06 10:26:12 we were discussing solar thermal systems ;) May 06 10:26:20 since we both own one May 06 10:26:23 yes, hehe, he likes that stuff May 06 10:27:25 well, thanks for the attention May 06 10:31:45 zumbi, btw :) http://people.canonical.com/~ogra/watchthesun/ May 06 10:35:03 is that a demo or your system? cool -- i used to work for a company which did things like that with sensors and stuff :) May 06 10:36:10 uhm.. and it is real time :) May 06 10:36:13 yep :) May 06 10:36:33 well, delayed, it rsyncs the data to the public server every hour May 06 10:36:44 the relatime ubuntu machine sits in my basement May 06 10:39:08 wookey had a similar thing, http://www.wookware.org/talks/hotwaterballoon-Fosdem2009.pdf May 06 10:39:27 he tracks temps with rrdtool graphs May 06 10:40:10 yeah, i havent found the time yet for the matching rrd setup May 06 10:40:29 ist on ym TODO list but my job doesnt leave me much time for it May 06 10:41:33 the site i showed you was actually supposed to only be the start, clicking on a sensor on the page should pop up the mrtg/rrd graph May 06 10:43:21 nice May 06 10:44:13 there is a project (flukso.net) which might be of help May 06 10:44:44 oh, nice ! May 06 10:44:49 they have a monitoring rrdtool based system with git repo, http://github.com/icarus75/flukso May 06 10:45:18 cool May 06 10:45:18 but I have not check out the code to see if it is generic or if it is to much hardware dependent May 06 10:45:34 * ogra would love to bring ubuntu into home aoutomation some day May 06 10:46:52 there was a talk/bof on that topic on later debconf, interesting stuff.. yes May 06 11:11:38 asac: Any input on the improved-subarch detection spec? May 06 11:12:06 asac: I'm going to prepapprove it, we can decide to schedule or not May 06 17:55:36 hey ogra, (qemu apt bug) those 3 times that it did work, did you just stick the "echo 3 /proc/cpu/alignment" in the install script? my two machines still locked up... May 06 18:05:34 rcn-ee: You might want to check the #ubuntu-devel logs: soren reported that the symptoms were very common to some seen for x86 in hardy. May 06 18:06:19 thanks persia, it's almost to the point i should just auto login there too. ;) May 06 18:11:46 It's not a bad idea in general. Most of the core discussions for all architectures happen there, and it's a point of coordination between all the different development teams. May 06 18:59:20 rcn-ee, its "echo 3 >/proc/cpu/alignment" May 06 19:00:41 rcn-ee, needs to go somewhere after the mount stuff in the installer script May 06 19:19:16 persia: ping May 06 19:19:32 * persia mumbles about contentless pings May 06 19:20:03 ha May 06 19:20:57 persia: just wondering about the possibilities of Ubuntu on ARM for Archos 5it? ARM Core A8 May 06 19:21:55 persia: specs are 32GB of storage, an 800MHz ARM Cortex-A8 CPU, 256MB of RAM, 802/11b/g/n WiFi, a USB port, and runs Android 1.6 May 06 19:22:44 * persia is trying to figure out *which* Cortex-A8 May 06 19:22:47 dunno if suitable to hackin ubuntu on to it.. but Id be willing to help out if its usable as a device we may want to target May 06 19:22:55 hmm.... May 06 19:23:34 dont have the unit in fromt of me to see if I can figure it out persia May 06 19:23:45 Yeah, should work fine. May 06 19:24:25 http://cdimage.ubuntu.com/kubuntu-netbook/ports/daily-live/current/lucid-netbook-armel+omap.img would be the image you want. May 06 19:24:49 Might need some hints to the kernel: it *should* boot on any OMAP3, but it's never quite that simple :) May 06 19:25:35 I think there's some special codecs and video drivers, but I don't know the package names offhand. May 06 19:26:25 And, as much as you like Kubuntu, running it with only 256MB of RAM will be extra fun :) May 06 19:26:46 persia: ya well aware of that. im thinking ubuntu for this May 06 19:26:59 Just remove the "k" from the URL :) May 06 19:27:08 ya May 06 19:27:39 persia: so, this thing has an internal flash, and I assume I just write the img to it using the imagewriter ya? May 06 19:28:14 persia: also, do we have an easy way to backup whats there? so it can be restored? or thoughts on this? May 06 19:28:26 No, it's not going to be quite that simple :) May 06 19:28:35 figured as much May 06 19:28:56 First step would be to track down some device-specific forum, and make sure you know how to backup & reflash the flash. May 06 19:29:30 Next would be to try to find a way to boot *selectable* kernels (because you want to make sure you don't flash something that bricks the device) May 06 19:29:51 http://archosfans.com has some useful links. I believe Archos released a developer linux image for the 5it May 06 19:29:56 Then try booting the kernel. When that doesn't work, find/make a kernel that boots, and try to get that into the Ubuntu kernel tree. May 06 19:30:22 Note that folks will be annoyed if you make their devices stop booting along the way :) May 06 19:30:32 heh May 06 19:30:39 (most folks running the omap kernels seem to have BeagleBoards) May 06 19:31:16 This may be something for me to poke around with at UDS in the evenings May 06 19:31:20 *once* you have a working kernel, then you probably want to do an HD-install, following one of tbm's guides on how to install to a NAS. May 06 19:31:58 persia: this unit has a SD card slot and I have a 8 GB card in it May 06 19:32:05 could use that? May 06 19:32:16 If you can find a way to convince the bootloader to boot off SD, sure :) May 06 19:32:22 lol May 06 19:32:28 loads of fun May 06 19:32:40 It's probably running some version of uboot. May 06 19:33:04 But determining which, and how to adjust the config, and how to avoid bricking it in the process is where you get to have fun :) May 06 19:33:32 * rgreening remembers the easy days of dd May 06 19:34:02 dd works, but you have to be *running an OS* to run dd :) May 06 19:34:30 This kind of fiddling isn't arch-specific: there's plenty of powerpc designs that have the same issues. May 06 19:35:11 It's about not having a standard for all devices to use the *same* hardware-bringup interface, so one can link into OS-specific second-stage bootloaders (e.g. grub) May 06 19:35:17 here's a link to someone else hacking on this apparantly: http://fladnag.net/blog/category/archubuntu/ May 06 19:35:30 tru dat May 06 19:36:51 Cool, so the find-way-to-swap-kernel-without-bricking-it and find-kernel-config-that-works steps should be easy. May 06 19:37:29 I think he is using the develop edition firmware from archos May 06 19:38:32 I may have to pop him a msg May 06 19:38:37 ty persia May 06 19:38:55 Let me know how it goes, and if you get it working, please document *how* on the wiki. May 06 19:39:30 I'd like a little more RAM, but if it works, and can have USB Host somehow, I'd personally prefer it to a BeagleBoard. May 06 19:39:40 * persia likes shiny plastic cases May 06 19:39:57 Note that it will suffer from all the standard issues that Ubuntu has at 800x480 :) May 06 19:40:17 You probably want to install lubuntu-desktop or similar on it. May 06 19:40:57 heh. May 06 19:41:14 just watching this. part way thru a ubuntu desktop http://www.youtube.com/watch?v=5GIyeonA5jE May 06 19:42:27 he has to reboot/hard shut as not touch and no keyboard. hah May 06 19:44:03 Looks like Archos as a "mini-dock" that you want to get to have a keyboard. May 06 19:44:33 or a bluetooth kdeboard. I have a bluetooth mouse and it works well with it May 06 19:44:46 yes. I have used a mouse with this May 06 19:44:48 :P May 06 22:00:13 in chicago ... May 06 22:37:48 * cwillu_at_work stalks rcn-ee May 06 22:47:34 Might be getting late in the day for that :) rcn-ee's are best caught a few hours earlier. May 06 22:49:39 ya, I only got my images in a row a few minutes ago May 06 22:49:52 I've found a hardware dependent btrfs failure :) May 06 22:50:00 or a btrfs dependent hardware failure :/ May 06 22:50:23 same card and image works in both beagles and overos if the rootfs is ext3 May 06 22:50:25 nah i'm back.. ;) wow... so it's only btrfs on overo.. May 06 22:50:29 yep May 06 22:50:39 and at least that one specific file May 06 22:51:13 that's just crazy... so what do we do next.. get the btrfs's guys involved/ May 06 22:51:20 I poked them as well, yes May 06 22:52:29 I'm guessing it's something with the "retrying using single block read" honestly, i.e., either btrfs isn't flushing properly, or something May 06 22:52:50 ooo, reply in #btrfs May 06 22:53:05 cwillu_at_work, btw.. just in case... you haven't tried the 2.6.33.3-l1 right? (i think i found another gcc/kernel bug... May 06 22:53:30 nope May 06 22:53:38 you saw the pastes right? May 06 22:54:01 good, lucid's version stops at uncompressing.. (actually just loading as i see it..) grabbing a beer and taking shoes off. ;) May 06 22:57:17 shhhh, he's thinking :p May 06 22:57:59 patiently waits.. (they don't log their channel do they) May 06 22:58:55 uncertain May 06 22:59:35 usually I'm trying to get cmason's attention May 06 22:59:35 i couldn't find anything quick oh, well it'll be interesting to see what he thinks.. May 06 22:59:50 I wouldn't hold your breath yet :) May 06 23:01:14 i was just thinking of another possibilty, little out their..... omap core bug between revisions? May 06 23:02:03 overo: OMAP3525 ES2.1 beagle: OMAP3430/3530 ES3.0 May 06 23:03:14 actuallly let me try btrfs on my old Bx board... May 06 23:03:19 it's ES2... May 06 23:06:20 * rcn-ee quickly rebuilds a new rootfs, as it has the older 2.6.32.11 kernel... May 06 23:13:43 cwillu_at_work, this issue didnt' happen with 2.6.32.11 right? I'm still rebuilding the new image, but this with 2.6.32.11 on my old beagle OMAP3430/3530 ES2.1 http://pastebin.com/5DhUsa27 May 06 23:14:18 nm it would help to have btrfs tools isntalled... May 06 23:14:23 heh May 06 23:14:26 not fatal though May 06 23:14:32 I'd dying long before I get to btrfs-fsck May 06 23:14:46 ah okay... May 06 23:15:54 I'm going to retry with 2.6.32.1-x1.0 (i.e., the random uImage I have lying around :p) May 06 23:16:10 rootstock is almost done for me, just tarring... May 06 23:17:31 booting now May 06 23:18:07 init: procps main process (666) terminated with status 255 May 06 23:18:54 [ 40.506011] device fsid c24abdbd5df3226a-2ff6a72b00d09595 devid 1 transid 6466 /dev/root May 06 23:18:54 [ 40.516143] btrfs: use spread ssd allocation scheme May 06 23:18:54 [ 40.552825] VFS: Mounted root (btrfs filesystem) readonly on device 0:13. May 06 23:18:54 [ 40.559814] Freeing init memory: 184K May 06 23:18:54 init: procps main process (666) terminated with status 255 May 06 23:19:08 and then it dies waiting for rootfs May 06 23:19:44 init prompt comes up to bring up a maintenance shell, but the shell doesn't start May 06 23:21:11 and... there's the csum error May 06 23:21:12 part of the shell scripts are in the uinird now... May 06 23:21:31 but it should still run from the disk, like previously.. May 06 23:21:33 be aware, I'm not using an initrd :p May 06 23:22:01 shouldn't matter.. just less random notuseful errors on bootup May 06 23:22:12 init: procps main process (666) terminated with status 255 May 06 23:22:27 sounds familiar May 06 23:22:47 oh i was quoting you... it's untaring to my mmc that refused to umount... May 06 23:22:51 ah, k May 06 23:23:00 heh May 06 23:23:08 refusing to umount generally means it's still writing everything out :p May 06 23:23:22 umount is the thing that takes 10-15 minutes on my mkcard script :) May 06 23:23:35 exactly what i was seeing... May 06 23:23:50 The easy way to avoid that is to have much less RAM :) May 06 23:23:56 heh May 06 23:24:17 csum error in /sbin/mountall this time May 06 23:24:21 race condition! May 06 23:24:59 or, what happens on 2.6.32 instead of 2.6.33 May 06 23:26:03 persia, not really avoiding anything, it's the same wait time :p May 06 23:26:49 but don't you worry, my zippy2 will be here monday, and then I can make my root images from the ram constrained heaven of a beagle! May 06 23:26:53 man it really want's the btrfs tools... but otherwise no csum error.. http://pastebin.com/Fya99DmH May 06 23:27:16 btw, check your zippy2, some have the zippy1 id, forum post in the beagleboard forum about it... May 06 23:27:23 k May 06 23:28:13 btrfs.fsck both doesn't do anything, and doesn't take the same options as normal fsck May 06 23:28:55 can you hack up /etc/init/mount to not do the fsck? May 06 23:29:22 your mouting ro from the bootargs? May 06 23:29:31 rw shoudl by pass the fsck pass.. May 06 23:29:49 or modify /etc/fstab... ;) May 06 23:29:51 was referring to your paste May 06 23:29:53 umounts... May 06 23:30:46 I mount rw generally, doesn't make any difference when I tried it the other day though May 06 23:31:01 (no difference to getting the overo to boot, that is) May 06 23:31:38 * cwillu_at_work considers ordering a pizza May 06 23:33:37 and i get login... http://pastebin.com/KLUUqKVB so probally not a ES2.1 to ES3 bug... May 06 23:34:54 yep May 06 23:36:26 would btrfs file check show us anything? May 06 23:37:08 just reading a file is sufficient May 06 23:37:38 it would show a csum error if the file contents changed, and a tree csum error if the metadata was changed (and there's automatically duplication of metadata pages) May 06 23:37:55 part of the joke is that it doesn't leave much for fsck to do :) May 06 23:38:18 yeah, cause it takes care of alot of it... May 06 23:43:20 * cwillu_at_work runs badblocks against the mmc reader, and orders a pizza May 07 00:56:19 lool: you around? **** ENDING LOGGING AT Fri May 07 02:59:56 2010