| gandhijee_ | xranby, no errors, the deb file just has the wrong things in it. | 00:20 |
|---|---|---|
| pnphi | excese me ! ! | 01:10 |
| === lool- is now known as lool | ||
| ppisati | anyone from arm's IS? | 10:24 |
| dmart | ppisati: no, but what's your question? | 10:38 |
| ppisati | dmart: someone told me you had panda's with same die id (and thus same mac address) and you couldn't change it at boot time | 10:41 |
| dmart | can you remember who you were talking to? I haven't experienced that. | 10:42 |
| ppisati | dmart: Tobin told me that and the bug is real (i've a fix for that) | 10:42 |
| ppisati | fix sent | 10:47 |
| === ericm|ubuntu is now known as ericm-dead | ||
| === Quintasan_ is now known as Quintasan | ||
| === jeremiah_ is now known as jeremiah | ||
| GrueMaster | ppisati: Thanks. And it was Canonical IS, not Arm. Lamont specifically. | 15:52 |
| ppisati | GrueMaster: ah ok | 16:31 |
| GrueMaster | ppisati: So, did it make it into the SRU kernel that is coming? If so, I'll test it as soon as I can. | 16:53 |
| ppisati | GrueMaster: nope, not pulled yet | 16:54 |
| GrueMaster | Ok. | 16:54 |
| Epsilonorion_ | What is the default runlevel of Ubuntu 12.04? I checked in rc-sysinit.conf and it says level 2, however my system seems to be running in level 1 (no level 2 scripts started at boot). Is there a command line way to check and make sure of the current configuration? | 18:11 |
| smplman | can someone help me merge the nvidia tegra kernel changes into my source? I have ubuntu 11.10 booted on my Xoom but wifi, bluetooth, and graphics acceleration aren't working. | 18:20 |
| xranby_ac100 | smplman: whick kernel version are you using? | 18:21 |
| xranby_ac100 | which | 18:21 |
| smplman | xranby_ac100: 2.6 | 18:21 |
| xranby_ac100 | 2.6.*? | 18:22 |
| smplman | 2.6.36 i believe | 18:22 |
| xranby_ac100 | smplman: have you tested to use the latest ubuntu kernel sourcetree? | 18:22 |
| smplman | my device is at home atm | 18:22 |
| smplman | no, i can try to update and rebuild when i get home | 18:22 |
| xranby_ac100 | hmm i have to check if the xoom was based on the nvidia-harmony dev board | 18:23 |
| xranby_ac100 | if so then most work applied to the ac100 kernel should be usable for your setup | 18:24 |
| smplman | ventana | 18:24 |
| smplman | i found the nvidia source http://nv-tegra.nvidia.com/gitweb/?p=linux-2.6.git;a=summary | 18:24 |
| smplman | here is the source that i used to build my kernel https://github.com/LIV2/LIV2-Xoom-GNU | 18:24 |
| xranby_ac100 | ok hmm then your xoom are similar to the trimslice | 18:25 |
| xranby_ac100 | you can try the trimslice forum | 18:25 |
| smplman | yea i saw a lot of docs about video acceleration, but only the binary comes with the trimslice | 18:25 |
| smplman | no source | 18:25 |
| xranby_ac100 | there is no source | 18:25 |
| xranby_ac100 | video acceleration on the tegra2 currently only exist as a binary driver from nvidia | 18:26 |
| xranby_ac100 | the same situation apply for all other arm gpu cores currently | 18:26 |
| smplman | ahh i see. Im more interested with wifi atm | 18:26 |
| xranby_ac100 | but you can get the video acceleration for ventana here: http://developer.nvidia.com/content/linux-tegra-l4t-beta-release | 18:27 |
| smplman | xranby_ac100: will give it a shot when i get home from work. WIll be back later | 18:29 |
| smplman | thanks for the help | 18:29 |
| xranby_ac100 | smplman: the ac100 team have managed to get these drivers working on the 3.0 kernels and later | 18:29 |
| xranby_ac100 | the trimslice still use the android kernel tree | 18:29 |
| smplman | so a kernel upgrade may be in order | 18:29 |
| xranby_ac100 | you can try the #ac100 for tegra2 kernel guidance | 18:30 |
| smplman | xranby_ac100: muchos gracias | 18:31 |
| xranby_ac100 | youre welcome | 18:31 |
| Epsilonorion_ | Does using update.rc somehow change the runlevel? | 18:37 |
| Epsilonorion_ | Anyone have any info they can share on how to setup a startup script for a user within ubuntu server? I setup a autologin for a user. I then created a startup script that I setup with update-rc.d. The script seems to run, but not within the user account. The user account also does not seem to auto login anymore. Finally, I seem to have issues killing the autoscript when desired. I can go through ps aux, but that is the only | 19:31 |
| Epsilonorion_ | way. | 19:31 |
| GrueMaster | Epsilonorion_: Have the script change uuid during startup. A simple way would to have the init script call sudo <user> <script>. | 19:33 |
| infinity | Epsilonorion_: For system-provided (ie: packaged) stuff running as another user, you want "su" in your init scrpit. | 19:35 |
| infinity | Epsilonorion_: For user-provided, @reboot cronjobs are the way to run things at boot as users. | 19:35 |
| Epsilonorion_ | is it suggested to run as a user. I don't specifically need to. I can just run it as a background service if that is best? | 19:36 |
| Epsilonorion_ | I had just assumed that using sudo would ask for a password, and I don't want that | 19:37 |
| infinity | Epsilonorion_: su, not sudo. Init scripts run as root anyway, no need for fancy wrappers. | 19:37 |
| smplman | xranby_ac100: my wifi/bluetooth chipset is the bcm4329 | 19:38 |
| Epsilonorion_ | infinity: so then on the line where I call the function to run, I use "su username function". How should I handle the shutdown? Just use ps aux, or create a stop procedure within the script? | 19:39 |
| infinity | Epsilonorion_: If stopping actually matters (ie: if just killing the process won't do), then yes, you need a stop function. | 19:40 |
| infinity | Epsilonorion_: And your start would look something like: su -c "/path/to/binary -option -option" - username | 19:41 |
| infinity | Epsilonorion_: A third option is learning how to use start-stop-daemon (check the manpage), which lets you feed --user and --group. | 19:42 |
| Epsilonorion_ | infinity: killing the process will do, however I want an easy way for a user to kill it instead of ps aux and looking for the script. Thanks for the correct su line. Is there a correct way to stop the program in this case? | 19:42 |
| victorp | ping | 19:42 |
| infinity | victorp: Should the channel pong? ;) | 19:42 |
| victorp | infinity, not even sure how that got there | 19:43 |
| victorp | ooops | 19:43 |
| victorp | :) | 19:43 |
| infinity | Epsilonorion_: If you want it easily killable (and there isn't a /path/to/binary -shutdown option), then you want to start it with start-stop-daemon. | 19:43 |
| GrueMaster | Epsilonorion_: Your init script can just call kill <PID> if it isn't doing anything important. | 19:43 |
| infinity | Epsilonorion_: start-stop-daemon can be asked to write a pidfile. | 19:43 |
| infinity | Epsilonorion_: Than your stop target can kill that same pid. | 19:43 |
| infinity | GrueMaster: Just randomly scanning for PIDs is disastrous, if people may be running more than one copy of the same binary. | 19:44 |
| infinity | (user copies, chroots, etc) | 19:44 |
| GrueMaster | infinity: I didn't imply random scanning. I was actually going to suggest storing the PID in /var/run/<app>. | 19:44 |
| * GrueMaster is in multiple realities this morning. | 19:45 | |
| infinity | GrueMaster: Sure, which, if his app doesn't do it, is best done with start-stop-daemon, so I guess we're on a similar page. ;) | 19:45 |
| GrueMaster | yes. I was actually going to look for an example init script that I wrote many moons ago that does this. | 19:46 |
| Epsilonorion_ | I follow what you guys are saying, and have seen examples of how to store the PID to some extent, however I am new to creating a script in this manner. Is there a good resource/example of how to do this | 19:46 |
| GrueMaster | http://tldp.org/LDP/abs/html/index.html | 19:46 |
| GrueMaster | That's what I use for bash scripting. | 19:46 |
| infinity | Epsilonorion_: grep start-stop-daemon /etc/init.d/* | 19:46 |
| infinity | Epsilonorion_: The ssh init script looks like a good one to steal from. | 19:47 |
| infinity | Epsilonorion_: Ooo, kerneloops is even better, since it has the --chuid bit you want too. ;) | 19:47 |
| infinity | kerneloops:start-stop-daemon --start --quiet --oknodo --chuid kernoops:adm --pidfile $pidfile --exec $exec | 19:47 |
| infinity | kerneloops:start-stop-daemon --stop --quiet --oknodo --pidfile $pidfile | 19:47 |
| infinity | ^-- Both start and stop invocations. | 19:47 |
| infinity | "$exec" in that case would be "/path/to/binary --with --options" | 19:48 |
| Epsilonorion_ | I found both the site and kerneloops | 19:48 |
| infinity | And $pidfile should be /run/mydaemon.pid | 19:48 |
| infinity | And --chuid is self-evident, I assume, just change it to user:group that you wanted. | 19:48 |
| Epsilonorion_ | so doing this method, should I kill the autologin procedure. It would seem that following this method is closer to creating a service for all users. | 19:53 |
| GrueMaster | Epsilonorion_: Unless you are doing multi-user and each user will need a sparate instance of this app, I would make it an init daemon. | 19:56 |
| GrueMaster | Far less overhead. | 19:57 |
| Epsilonorion_ | I just keep wondering if I am making this too complicated. I started out simply wanting to create a script that would allow a single program to startup when the user auto logged in. This way I could easily type ps -all, find the user (or the program) and end it. | 19:57 |
| infinity | Epsilonorion_: You want an init script if it should be running even if the user isn't logged in. | 19:58 |
| GrueMaster | If you want something to run when a user logs in, just add it to the ~/.bashrc (or /etc/bash.bashrc to be global). | 19:58 |
| infinity | Epsilonorion_: If the goal is to have it run only when the user is logged in, you want it in their session. | 19:58 |
| infinity | Epsilonorion_: As Gruemaster says, .bashrc for a CLI login, or for GUI logins, there's session preferences/editing in the various GUI control panels. | 19:59 |
| steev | i have a question regarding the ubuntu udev package. recent versions of udev seem to have a kernel version check for > 2.6.32, but since I'm using an older kernel with the portions that udev expects backported, i like to remove that check from the udev script, however I can't seem to find where it resides on Ubuntu, anyone know where it is? | 19:59 |
| Epsilonorion_ | I technically do not need it to run when the user logs in. The system is a stand alone system. Users log in to debug code or get logs, however the application should run as soon as it is fully powered up. I have only been familiar with creating startup scripts for when users log in | 19:59 |
| infinity | steev: rgrep 2.6.32 through the udev sources? | 20:00 |
| Epsilonorion_ | I don't mind letting it be a daemon that runs without a user, as long as when a user logs in they can easily end it | 20:00 |
| infinity | Epsilonorion_: Right, if you want it running regardless of people being logged in, the init script you were writing is the right way to go. | 20:00 |
| GrueMaster | With it being an init script, a user can end it with "sudo service <appname> stop". | 20:01 |
| steev | infinity: well it normally resides in /etc/init.d/udev but that doesn't exactly exist in ubuntu. | 20:01 |
| Epsilonorion_ | that was what I thought on both accounts. Then I will stick with the init script. I assume that I do not need to worry abou tying it into a specific user in this case then | 20:02 |
| GrueMaster | Epsilonorion_: No, but for security it is recommended that the process have a non-root user that it runs as (i.e. mysql, apache, etc). | 20:03 |
| scientes | will ubuntu support this? http://www.wholesaletabletspcs.com/wholesale-Ainol-Novo7-Elf-7-Inch-Android-4-0-Tablet-PC-Capatitive-Multitouch-screen-Allwinner-many-core-A10-1GB-DDR3-camera-3G-WIFI-2160P-decoding-HDMI.html | 20:03 |
| GrueMaster | Not required, but recommended. | 20:03 |
| scientes | I want a cheap tablet with dual-core | 20:03 |
| scientes | and HDMI+3G+Wifi+nice: GPS | 20:03 |
| infinity | steev: We have /etc/init/udev.conf, but I see no kernel checks, there. | 20:03 |
| Epsilonorion_ | understood. Thanks GrueMaster, infinity. Hopefully I won't break anything | 20:03 |
| infinity | steev: Are you sure it's not in the C source? | 20:04 |
| scientes | also, anybody having success with multi-seat? | 20:04 |
| steev | infinity: i'm 100% positive it's not in the C source | 20:04 |
| GrueMaster | steev: It could also be in the initrd stuff. | 20:04 |
| infinity | Nope, I see no version checks in the udev initramfs hooks either. | 20:05 |
| infinity | steev: I honestly see no checks anywhere. Are you sure something's checking, or just assuming that it must? | 20:05 |
| steev | infinity: it does on debian, and gentoo, i just wanted to pre-empt it on ubuntu (doing a clean install) | 20:06 |
| infinity | Oh, wait. | 20:06 |
| infinity | MINKVER="2.6.24" | 20:06 |
| steev | where do you find that? | 20:06 |
| infinity | Oddly, it only checks when building the initrd. | 20:06 |
| infinity | Actually, I suppose that makes some sense. | 20:06 |
| infinity | /usr/share/initramfs-tools/hooks/udev | 20:07 |
| GrueMaster | scientes: Unknown if that will support Ubuntu or not. Is it armv7? I can't find any relevant specs on the cpu. | 20:07 |
| steev | interesting, and good to know | 20:07 |
| infinity | GrueMaster: Looks like the "Allwinner A10" is an A8. | 20:08 |
| infinity | (Cuase that's not confusing) | 20:08 |
| steev | infinity: it is an A8, i have one, it's quite nice | 20:08 |
| infinity | 1.5G A8 and Mali400. | 20:08 |
| steev | one of only 2 tablets i have that don't make me want to throw it out the window | 20:09 |
| GrueMaster | I read "Allwinner A10 is just under the architecture of ARM Cortex-A8". http://androidling.wordpress.com/2011/09/12/allwinner-technology-a10-tablet-solution-2160p/ | 20:09 |
| scientes | guerby, yes it is armv7---cortexA10 to be exact | 20:09 |
| scientes | oh wait A* i guess | 20:09 |
| infinity | scientes: It's an A8, not an "A10". Confusing product name. | 20:09 |
| scientes | yeah agreed | 20:09 |
| infinity | scientes: Anyhow, what did you mean by "will Ubuntu support it"? | 20:09 |
| infinity | scientes: Ubuntu's userspace will run on it. | 20:10 |
| scientes | gotcha | 20:10 |
| infinity | scientes: We don't have a kernel for it, but other kernels would work, if you built your own, or used the Android one. | 20:10 |
| scientes | anyone gotten multiseat to work on these tablets? | 20:10 |
| scientes | "A10 is just under the architecture of ARM Cortex-A8." | 20:10 |
| scientes | damn that is misleading | 20:10 |
| scientes | IE use the HDMI w/ usb keyboard+mouse as a seperate computer | 20:11 |
| scientes | with a differn't user of course | 20:11 |
| === npitre_ is now known as npitre | ||
| GrueMaster | scientes: We're just getting to the point of running alongside Android on tablets. Not sure if multiseat would work in this instance. Likely not, as the systems are really underpowered for something like that. | 20:13 |
| scientes | well I wanted to do a very targeted thing on the tablet, and desktop-like on the hdmi | 20:13 |
| scientes | with a dual-core one | 20:13 |
| GrueMaster | You need a lot of memory to have two running desktop instances that are usable. | 20:13 |
| scientes | and use one of the 1GB ones | 20:14 |
| scientes | but you need multiseat support (two xorgs, seperate input handling) to use both at the same time | 20:14 |
| scientes | like I wanted opencpn (navigation) on the tablet, and desktop like stuff on the hdmi | 20:15 |
| GrueMaster | All I can say is that it is untested. | 20:15 |
| scientes | supposedly systemd is working to bring back multiseat support in linux | 20:15 |
| scientes | but i havn't gotten it to work | 20:15 |
| scientes | the main thing is to be able to run two Xorg servers at the same time | 20:16 |
| scientes | I also want this on x86 | 20:16 |
| scientes | anyone managed to upgrade the kernel on the sharp netwalker? | 20:18 |
| Epsilonorion_ | GrueMaster, infinity: I have finished the init script and tested it. The script seems to be working fine, however, I need a feature I am unure of how to get. When I made a simple script that ran the program, if I typed "service script" I would get the output of the function started by the script, however, I do not get that feature now. Is there a way to do this? | 21:00 |
| GrueMaster | I'm not sure I understand, but I think the init scripts usually trap output to /var/log/syslog. You can also have your script create and append to it's own log. | 21:05 |
| Epsilonorion_ | GrueMaster: Can I do that by adding a pipe at the end of the start-stop-daemon command? If so, is there a way to make the log file more dynamic (datestamped) | 21:07 |
| GrueMaster | I really don't know the full answers to this. I'd recommend googling at this point. There are a ton of documents online for this. Look for init script development. | 21:08 |
| Epsilonorion_ | GrueMaster: And to better explain what I am doing, the program will have multiple debugging printfs that are useful to check for errors when directly connected to the system. I currently do not know how to get these printfs to show up if the daemon starts up on boot up. I killed the --quiet and --background flags and it shows up now though | 21:09 |
| Epsilonorion_ | GrueMaster: Understood thanks again and sorry for taking your time | 21:09 |
| GrueMaster | No problem, I'm glad to help where I can. | 21:09 |
| scientes | https://lwn.net/Articles/482760/ OOOOO, can i do this with vhashify from vserver to consolidate ram and disk usage? | 23:24 |
| scientes | http://www.ubuntu.com/devices/android/features-and-specs | 23:49 |
| scientes | 512MB ram | 23:49 |
| scientes | really? | 23:49 |
| GrueMaster | scientes: Most of the cell phones out there have 512M. Some have 1G. | 23:51 |
| infinity | Certainly most of the dual core ones, which is also a listed "requirement". | 23:52 |
| scientes | well you have 1GB requirement for desktops | 23:54 |
| scientes | seems like you had to bend it | 23:54 |
| scientes | cause its not like you are changing the stack that much | 23:55 |
| GrueMaster | We do? Hrm, better tellthat to the AC100 and beagleXM users. | 23:55 |
| scientes | nono i don't say 512 to too little | 23:55 |
| scientes | its just that you are saying you need 1GB to x86 people | 23:55 |
| GrueMaster | Well, remember that x86 binaries are much larger too. Especially for x86-64. | 23:56 |
| scientes | true true | 23:56 |
| scientes | so is chromium signif faster on arm? | 23:57 |
| GrueMaster | Couldn't tell you. I'm too busy testing stuff to do comparative benchmarking. | 23:58 |
| scientes | just wondering why chomium is default over firefox | 23:58 |
| scientes | BTW, i was trying to debug opencpn in a precise chroot (arm) | 23:59 |
| scientes | and i couldn't get gdb to load the debug symbols | 23:59 |
| GrueMaster | Erm, I have Firefox as default on my daily Panda image. | 23:59 |
| scientes | you have to use target remote cause qemu doesn't implament ptrace | 23:59 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!