=== allquixotic_ is now known as allquixotic [03:14] I'm annihilating SD cards, any suggestions for something more durable? (using a beaglebone) [03:16] trelane, what type of warning do you get before they go? [03:22] scientes, none that I've been able to capture before the failure, but lots of error reading, drive seek, etc from the kernel afterwards [03:22] eek, i should make some backups [03:22] [ 17.989135] end_request: I/O error, dev mmcblk0, sector 984986 [03:22] [ 20.791503] mmcblk0: error -110 transferring data, sector 984987, nr 29, cmd response 0x900, card status 0x0 [03:23] starting to think that SD is the achilles heel [03:23] well eSATA is obviously the way to go [03:23] usb is both slow, and doesn't support power spinning down [03:23] scientes: how do we interface with eSATA??? [03:24] I don't think the beaglebone has esata [03:24] but some arm devices do [03:24] scientes, not many I've seen, yet, but they are coming [03:24] such as? [03:25] they have been around for quite a while [03:25] there was even a sheevaplug that had esata years ago [03:25] * scientes doesn't have esata [03:25] scientes, IIRC though that wasn't native esata but a USB adapter [03:25] which is what I'm considering now [03:25] and the 2nd revision of the original sheeaplug can be retrofitted, but i have 1st revision [03:26] yes usb hard drive works [03:26] its just hot cause it never spinns down [03:26] the other option is ntp [03:26] solid state drives??? [03:26] putting solid state drive on usb is a sin [03:27] true... but will it work? [03:27] scientes, I don't need the performance, I need the reliability [03:27] heathkid, of course, it's just slower [03:27] and stop killing DS cards? [03:27] SD [03:27] hard drives are more reliable than ssd in my experience [03:27] as long as you arn't dropping them [03:28] * heathkid thinks maybe he wants to drop them... [03:28] * trelane mumbles something about the cloud and goes to find an umbrella [03:28] * heathkid ponders... maybe only a few hundred G's though... [03:28] heathkid, screw it, we've got a pallet and an incline drop table, and duct tape [03:29] stick the hard drive between several of those and pull the red handle [03:29] or we could just skip that and use... the bat [03:29] I can do a 600G 1/2 sine pulse drop [03:29] I just want something that works and doesn't wear out after a few days [03:29] that should reliably kill just about any hard drive [03:30] not SSD [03:30] depends on construction [03:30] they should construct them better then [03:30] I'd settle for 100G's [03:31] *any* set-top box will survive 70G without even a glitch in the signal... [03:31] regardless I need something that can handle high read/write cycles better than SD [03:31] honestly trelane hard drives are the best way to get reliable space [03:31] with ssd you never have enough space [03:31] very trure [03:31] ture [03:31] bah [03:31] how about "true" [03:32] scientes, actually my problem isn't space, I barely use _ANY_ but I write to the same file once a minute (rrdtool) [03:32] and in about a week, it shits the bed [03:32] trelane, why dont you just do it in memory then? [03:32] thats what almost all embeded machines do [03:32] I've thought about a ram disk, my concern there is preserving data [03:32] hard drives and ram are the only store that can handle that type of stuff [03:33] not to be overly dramatic but if data's lost, the ramifications for this system is people starve [03:33] trelane, you could save it every few hours [03:33] well then use a hard drive [03:33] or multiple hard drives [03:33] that's kind of what I'm thinking [03:33] cloud :/ [03:34] ssds can do it too, as long as there is plenty of free space, cause they just move the writes around [03:34] but hard drives really are the best at writing [03:34] and if you are moving every minute you don't need power saving [03:34] right, and I'm using ~400KB with the database [03:34] solar :) [03:35] ohhhh [03:35] then hard drive might not be best [03:35] system will be deployed to the third world, attached to, err, miscellanious agricultural systems whihc produce food [03:35] you really should be using networked ram disk then [03:36] yeah ram disk is sounding more and more like the solution [03:36] starting to look at my pandaboard es too instead [03:36] if you have a battery and solar ram isn't that bad [03:37] if you were really fancy, you would store it to a known place in ram [03:37] and then when it started up it could actuall use the last runs [03:37] so it could crash (use watchdog device) and recover [03:37] without ever hitting a disk [03:37] that's an interesting idea, since it should never lose power [03:38] you could even back it three times in ram [03:38] or rather, with parity bits [03:39] * scientes is not sure the kernel has anything that would allow implamenting that feature [03:39] ---well-known memory locations that is === awafaa_ is now known as awafaa === phh_ is now known as phh === sangwook is now known as dusaram === zyga is now known as zyga-afk [10:36] so I've got ubuntu running on my beagleboard... how do I speed up the gfx? [10:47] marvin24, I see you have a new branch. Please let me know when you want the ubuntu package updated and which branch should it be rebased on [10:47] janimo: this is for android [10:47] so just ignore it [10:47] ah, so you are testing andorid as well? [10:48] not me, but others seem to be interesed [10:48] but I hate to do the double work [10:48] GNU/Linux should use l4t-r15-ac100 (which is a bit behind of the android branch, thanks nivida!) [10:52] marvin24, did you get some confirmation from nvidia that they indeed are behind and need to port fixes to the L4T branch too? [10:52] anyway should I upgrade to your latest l4t branch now? [10:53] janimo: the branch is taged as "rc" so I guess they plan an update [10:53] janimo: yes, you can do a new package if you like [10:53] unfortunately, the console problem is still present [10:53] marvin24, did you test with the binary driver ? does it run any better ? [10:54] ogra_: what do you mean by "better"? [10:54] console problem meaning you need to pass console=tty1 on the command line? That is ok as long as not a regression from existing 3.1 [10:54] it doesn't crash [10:54] marvin24, "at all" [10:54] nice I see you fixes the logo moving problem on boot [10:54] janimo: yes [10:54] ah, thats an improvement [10:54] well, I didn't tested it much [10:54] that is what users are for ;-) [10:55] well, if you get a mousepointer thats more than i got with what we have in the archive [10:55] marvin24, regarding the zram/zcache stuff [10:55] that seems dropped from the latest branch [10:55] i agree, the rest is for users :) [10:55] janimo: mmh? [10:55] you man zcache updates? [10:55] http://paste.ubuntu.com/1119216/ [10:55] janimo, the code or the option ? [10:56] this is what I get as config delta [10:56] maybe I need to resync with defconfig, I know I did that last time [10:56] bah, crap [10:58] janimo: it should be autoenabled [10:58] at least I didn't removed it [10:58] * janimo checks. [10:58] paz00_defconfig has it [10:58] maybe the disappearance of xvmalloc pulled out those that depended on it [11:00] NCommander, where do i find a wiki doc for booting highbank images in qemu ? i'm searching my butt off but seem to not be able to find anything [11:00] ogra_: Won't work anyway. [11:00] oh? [11:01] janimo: paz00_defconfig is created by "make savedefconfig", so the option should be still needed [11:01] i thought it was the new hotness [11:01] marvin24, ok, resyncing with defconfig now [11:01] janimo: thanks [11:02] ogra_: https://launchpad.net/bugs/1030600 needs fixing first. [11:02] Ubuntu bug 1030600 in linux "please build/install highbank dtb file" [Medium,In progress] [11:02] egh, ok [11:03] ah I see why last time we got failurie to boot from VFAT, merging defconfig has the NLS bits modular [11:04] heh [11:04] * ogra_ reads https://wiki.ubuntu.com/ARM/Server/Install [11:04] "Booting into the Installed System [11:04] Your system should now boot into the installed system. " [11:05] thats a very useful paragraph [11:10] ogra_, that sounds like an instruction that cannot be missed === zyga-afk is now known as zyga [12:13] infinity, hmm, the SUBARCH hardcoding seems to really get in our way in debian-cd now [12:13] i wonder if its actually in the environment and we could just drop it [12:15] * ogra_ cant rember why we put it there first place [12:18] We hardcode SUBARCH? [12:19] ogra_: Oh, in post-boot? [12:20] ogra_: Looks like just some unfortunate cargo-culting, that should definitely be fixable. [12:20] (Or in boot, rather, whatever) [12:20] yeah, if i'm in the office i'll drop it and throw out some testbuilds [12:21] * ogra_ doesnt want to do that from unstable hotel network [12:21] worst case we need to drop the symlinks and actually use a script per subarch but i'd rather not [12:22] ogra_: I'm not sure it's in the environment, but that could perhaps be fixed. [12:22] ogra_: Check tools/boot/sarge/boot-m68k, which tries to guess subarch based on the kernel in play. We could do something like that, perhaps. [12:22] ah, yeah [12:23] well, lets first tryif we have the var set, i think lool added it back then because of some discrepancy between live and d-i builds [12:23] ogra_: But breaking the symlinks wouldn't be world-ending either. Just means we have to remember to keep all the derivatives in sync. [12:23] right [12:24] well, i would like to unify it to one script (or a functions file we source at the top of the scripts) actually [12:24] build.sh: export SUBARCH="${FULLARCH#*+}" [12:24] instead of having a per subarch wood made of symlinks [12:24] It *should* be in the environment. [12:24] k [12:25] s/wood/forest/ :) [13:13] ogra_, do you use bluetooth headsets with the ac100? [13:14] janimo, not atm, but yeah, i did a lot in the past [13:14] it can get a little choppy under load with a usb BT dongle [13:14] not sure how well it works with builtin BT [13:14] did you not try the built-in? [13:15] marvin24, your newest branch boots here, does not seem worse than what I had before. Still needs going to a VT and back to X after a screen blank [13:15] i only got a device that has BT as my last one [13:15] all others i always had had no BT [13:16] ok. Could it depend on the headset too? [13:16] I wonder if using BT headset with the ac100 is worth it [13:16] probably, but i doubt it [13:17] i havent gotten mumble to work reliably with it ... gets to choppy due to load ... if yuo just want to play back music it should be fine though [13:17] but i think duplex is hard for it [13:17] heh, I have issues with mumble even on the wired headsets quite often [13:17] heh [13:19] ikepanhc, still awake? :) [13:20] janimo: yes, anything? [13:21] ikepanhc, nope, I just assumed you went to sleep a while ago :) [13:21] janimo: just back to home, lay on the bed with laptop [13:23] janimo: do you remember how to send request to promote to proposed? [13:23] janimo: the build is done and it looks good [13:23] ikepanhc, from PPA? [13:23] janimo: yes [13:23] I used a script and did not need intervention from others, it needs upload permissions though. [13:23] Let's see if your upload rights are enough [13:24] just a moment [13:24] janimo: though the status not changed yet https://bugs.launchpad.net/ubuntu/+source/linux-armadaxp/+bug/1030308 [13:24] Ubuntu bug 1030308 in linux-armadaxp "linux-armadaxp: 3.2.0-1606.9 -proposed tracker" [Medium,In progress] [13:25] ikepanhc, can you bzr branch this? bzr+ssh://bazaar.launchpad.net/+branch/ubuntu-archive-tools/ [13:25] janimo: I have that already [13:25] then copy-proposed-kernel.py precise linux-armadaxp [13:25] that is a script inside that repo [13:25] that is what I used and worked fine [13:25] janimo: thanks, that's what I need [13:26] ikepanhc: Your "prepare package" is still In Progress. [13:26] janimo: Drop the .py, those symlinks will go away eventually. [13:26] infinity: yes, but the build is done [13:26] Anyhow, I can go the copy. I'll need to accept it and fix overrides anyway. [13:26] hmpf, so omap already fell over after dropping SUBARCH= [13:26] infinity, did not look close enough to realize it was a symlink :) [13:26] infinity: https://launchpad.net/~canonical-kernel-team/+archive/ppa/+build/3690595 [13:26] * ogra_ waits for omap4 to get some logs [13:27] marvin24, so you say you have no issues with X not being able to restore from screen blanking unless going via a VT? [13:28] tegra_pwm tegra_pwm.o: pwm_disable called on disabled PWM [13:28] ikepanhc: If you set 'prepare package' to "Fix Released", the bot will take care of the rest, I believe. [13:28] ikepanhc: (The rest, as in setting the next task, not doing the upload) [13:28] marvin24, ^is the only thing in dmesg that may be related [13:29] infinity: I thought bot shall change the status [13:29] janimo: the console blank is "fixed" by installing the nvifia drivers [13:29] ^ never seen this message before [13:30] ikepanhc: It might. You'd have to ask Brad. It certainly doesn't update statuses on the tasks I do. [13:30] marvin24, rebooting with nvidia-tegra installed now. I wanted to try it out without first, to get the basics tested [13:31] infinity: how about give it few hours. and see how it goes? [13:31] ikepanhc: Or, I could just ask Brad. :P [13:31] not necessary, I love to wait and see [13:32] ikepanhc, the bot usually changed the statuses in a few hours, IIRC, probably does not check every 15 minutes [13:43] bah, crap, actually deploying my fix on nusakan would probably have helped [13:46] ogra_, I too see the fuse: module is already loaded error [13:46] even though fuse is not listed by lsmod [13:46] on the ac100 [13:46] do we have it builtin ? [13:46] infinity any idea what can cause such confusion? [13:47] ogra_, no, it is module [13:47] it is in /lib/modules [13:47] and according to config [13:47] I get fuse:module is already loaded in dmesg every time I try to modprobe it [13:47] and Error inserting fuse as modprobe output on the console [13:48] other modules insert fine [13:48] marvin24, with the new kernel and the nvidia drivers launching gnome-terminal kills X and sends me to lightdm [13:48] marvin24, what userland are you testing with usually? [13:49] infinity, btw, i found a very elegant solution to the boot.scr stuff in flash-kernel ... i'll just compoletely drop boot.scr ;) [13:49] ogra_: ? [13:50] ogra_: Oh, in favor of uENV.txt or something? [13:50] infinity, we planned to move to uEnv.tx anyway ... [13:50] yeah ! [13:50] and the baeuty of it is that you can split it into uEnv and preEnv.txt files [13:50] uEnv will carry all generic bits, preEnv will get the cdmline [13:51] so preEnv can be created by f-k-i and contain root= ... (and reside in /etc/flash-kernel for user convenience [13:51] ) [13:51] * ogra_ goes for a smoke [13:51] janimo: unity-2d [13:52] did you got these errors also with the older build? [13:52] marvin24, same here, unity-2d [13:53] marvin24, I need to check. It looks like I did not have the tegra drivers installed in a while [13:53] mmh, I often use the terminal [13:53] janimo: check with e.g. es2_info [13:54] marvin24, that was the plan, that's why I installed the nvidia drivers, to test some gles [13:54] janimo: some other user on #ac100 also reported crashes (but on wheezy) [13:54] but got derailed when it crashed X [13:54] marvin24, you're on 12.04? [13:54] this is quantal [13:54] yes [13:55] I still wait until it becomes "stable" ... [14:09] marvin24, weird, I get an X crash when starting xterm and as soon as I get into a sudo shell [14:09] bizarre [14:10] * janimo wonders if some device nodes related to nvidia drivers suddenly behave differently when an admin user does something to them indirectly [14:10] janimo: also crashes with a user xterm? [14:10] no, just as soon as I sudo in that [14:10] reproducibly [14:11] great [14:11] I'll test this out when I'm back home [14:13] * janimo keeps dreaming of the world where working in the software industry will not involve any hardware at all [14:14] * janimo unfairly blames hw for majority of software bugs too [14:14] the problem here is more non-open-source software [14:14] than hw [14:16] indeed, closed source. We do not even know for sure if nvidia did not put some _broken hardware_ inside those binaries!! [14:21] you may try to shake the driver a bit [14:22] maybe some loose skews will fall out [14:22] *screws [14:22] which you can send back to NV [14:24] the driver may get shaken indeed as part of me throwing the ac100 out the window :) [14:25] some nice first world problems we've got here [15:50] janimo: what Xorg version does quantal use [15:50] ? [15:50] 1.12.3 ? [15:50] precise seems to use 1.11.3 (abi 11) [15:51] if wheezy also used 1.12.3, then I guess this is a driver (abi 12) problem === Quintasan_ is now known as Quintasan [15:52] this is abi 12 indeed [15:53] marvin24, I narrowsed down the crash to running .bashrc from the sudo shell [15:53] within that setting the PS1 prompt using ANSI escape sequences [15:53] wow, even more bizarre [15:53] no idea why that would crash x though [15:53] I do not yet have a minimal sequence that crashes [15:53] if you change the prompt? [15:53] yes, that line in .bashrc causes the crash [15:53] did you disable CONFIG_USERPACE_UTF8_SUPPORT ? [15:53] :P [15:54] ogra_, heh. I actually need to pastebin the config changes for review [15:55] ogra_, marvin24 http://paste.ubuntu.com/1119676/ [15:55] all changes due to sync with defconfig, nothing explicitly changed by me [15:57] does TEGRA_OTG actually work ? [15:57] ogra_, no idea [15:57] * ogra_ thought it didnt [15:57] wow, what disabled USB_MASS_STORAGE in the past ? [15:57] well, "theoreticly it should" (C) [15:57] ogra_, snafus probably [15:57] or is that diff against a local config ? [15:58] is frontswap efficivily enabled? [15:58] * ogra_ hopes not [15:58] I'm still experimenting with otg [15:59] it does something, but not right yet [15:59] (or I fail to do it right) [15:59] well, it shouldnt do any harm to have it on at least [15:59] even if its not fully working [15:59] well, it crashes kernel when used together with uboot [15:59] but that's a different "Baustelle" [16:00] ogra_, can flash-kernel be forced to flash an older kernel [16:01] I downgrade with dpkg -i on ac100 but it did not run flash-kernel [16:01] yes, it takes the desired kernel version as option [16:03] (the zz-flash-kernel script in /etc/kernel.d should have actually done that on your package install [16:03] ) [16:26] <[XeN]> I'm looking for the ubuntu version that runs on the exynos 4212 SoC but can't find it. Can anyone help? [16:37] [XeN]: give the linaro leb a try for origen on your board [16:37] at linaro we have the samsung landing team maintaining the kernel for it [16:37] <[XeN]> what is the linaro leb? [16:39] <[XeN]> I'm talking about the ODROID-X. Ubuntu already runs on the board but I can't find the image. Only OMAP, iMX etc. [16:39] [XeN]: that's why rsalveti suggests Linaro images for Origen [16:40] <[XeN]> what's origen? [16:40] origen is the developer board based on 4410 [16:40] it's not the same one, but I believe the kernel should probably be quite compatible with [16:40] at linaro we have what we call as leb, based on both ubuntu and android [16:41] supporting the boards we have available [16:41] and regarding samsung, origen is the one used [16:41] rsalveti: 4210 is on origen [16:41] <[XeN]> ok. got it [16:42] sorry, 4210 [16:42] <[XeN]> so the odroid is 4412 quad core. should be compatible [16:44] <[XeN]> the thing I'm looking for is the gfx drivers and kernel sources. the odroid people will release the kernel src too. but only for 3.0.15. [16:44] <[XeN]> for the gfx I found an open one but in alpha state [16:45] <[XeN]> I just thought there is a ubuntu image for the platform where I can grab these stuff from. [16:46] [XeN]: There's no Ubuntu image for it, no. [16:46] <[XeN]> would be interesting what the guys at hardkernel use to show the boards capabilities. [17:44] bug 1028905 [17:44] Launchpad bug 1028905 in cdrom-detect "cdrom-detect in quantal omap4 hangs trying to look for install media on an extended partition" [High,Confirmed] https://launchpad.net/bugs/1028905 === shadeslayer is now known as shadeslayer_ === shadeslayer_ is now known as shadeslayer