**** BEGIN LOGGING AT Mon Jan 28 02:59:57 2008 Jan 28 03:11:11 mwester: I'd hope the proper AT commands would shut it off? like +CFUN=0 Jan 28 03:13:00 Kero: AT@POFF is the command, IIRC. A function to issue that command was added to gsmd, but the last time I checked, gsmd does not automatically issue that function on exit. Jan 28 03:13:55 This issue is one of the reasons why your phone still kills the battery if you shut it down -- the gsm modem sucks the battery flat even though you've powered off the device. Jan 28 03:17:40 gsmd does not. Jan 28 03:18:52 if I 'kill it gently' (TERM), my little framework that started gsmd sees nothing of the sort. Jan 28 03:23:01 I've been doing battle with Openmoko on this issue; I insist that we need to run a shutdown script to power off the gsm modem when the phone shuts down, the Openmoko developers refuse to implement *any* sort of solution. :( Jan 28 03:26:04 given the fragility os gsmd in general... that does not surprise me :( Jan 28 03:27:03 if I get libgsmd-tools to invoke but one CME error 3 when the gsm is connected to an operator, that's usually the last response I get. Jan 28 03:27:47 hey guys Jan 28 03:28:07 err ... i mean hello my moko monkeys Jan 28 03:28:30 has anyone checked out kde4 release video on google video? Jan 28 03:28:59 their new plasma technology is really neat, i could totally see running all of kde on something like the moko devices Jan 28 03:29:15 providing a fast enough cpu and enough flash storage i guess Jan 28 03:29:19 yeah, I even got plasma running on the moko :) Jan 28 03:29:23 but the ui could be totally customized to fit Jan 28 03:29:29 Mek: REALLY? Jan 28 03:29:36 cool! Jan 28 03:29:43 holycow: yeah.. it is a bit slow, but it works quite well :) Jan 28 03:30:22 and it requires large parts of kde4/qt4 of course Jan 28 03:30:28 yes indeed Jan 28 03:30:37 my thought was that of course one wouldn't run all of kde Jan 28 03:31:10 but you would have specialized applications developed on the framework for the device Jan 28 03:31:35 theoretically tho, one could striip down a lot of kde apps for the presentation format Jan 28 03:32:45 Mek: what are your thoughts on the potential for the kde4 framework running on moko devices? Jan 28 03:34:20 might be missing gsm stuff Jan 28 03:34:24 I think it should be possible to make it work... but I haven't really thouht about much details yet... Jan 28 03:34:45 I'd hopde the gsm stuff is/will be implemented in a gui-independent way... Jan 28 03:35:08 ljp: i would imagine probably the hardware abstraction might be missing support for more than that ... but i'm guessing Jan 28 03:35:38 Mek: if any of the kde4 stuff is an indication they will probably abstract that into some sort of daemon/api type of thing Jan 28 03:35:49 and probably indeed some missing bits should be implemented (starting with a widget style to even look a bit acceptable) Jan 28 03:36:06 Mek: yeah indeed Jan 28 03:36:13 might also need more ram, depending on what parts of kde you wanted to run Jan 28 03:36:30 i'm really starting like oxygen so far, its so non offensive compared to the 3.5 iconography Jan 28 03:36:50 I'm not sure that is really true, kde4 should be able to use not significantly more ram than the current UI I'd think Jan 28 03:37:19 i'm not sure i'm one to have an opinion on this, but i think different applications require different interface ideas Jan 28 03:37:47 playing media the ipod has that text based thing down pretty well it seems Jan 28 03:38:15 while say selecting an application the iphone kind of has an interesting idea Jan 28 03:38:53 for web browsing i picture a fully scaled up browser (if one was available theoretically) with an overlaid ui that only comes up with a certain action Jan 28 03:39:03 pft. fourth command gsmd gives is +CMEE. What is that? spec I have here does not list it Jan 28 03:41:24 however the neat thing about some of the iphone/ipod or even the playstation 3 interface ideas for such devices are interesting in that they can be reverse ported to the desktop Jan 28 03:41:40 their minimal approach is perfect for media center types of applications Jan 28 03:41:59 so you could get a re-use type of usecase where a media center shares the interface guidelines for a moko style device Jan 28 03:42:12 whereas the regulard desktop ideas are left for serious work at the desk Jan 28 03:42:50 nm. search functionality of abiword *really sucks* Jan 28 03:50:48 CMEE is extended error reporting Jan 28 03:58:00 even with explanation, I have no clue what +CCWA=1,1 does. Jan 28 03:58:35 "call waiting supplementary service", 3GPP TS 22.083 [5] Jan 28 04:09:30 Mek: thanks for the heads up on running plasma on moko already, thats really cool. Jan 28 04:12:30 Kero: call forwarding and call barring I think Jan 28 04:14:31 but what is configured about it? Both '1's mean "enable". Jan 28 04:15:21 and when would a call be forwarded? in other words, which doc do I need to figure this out? Jan 28 04:15:43 and while at it, what AT commands are used for text messaging? not in the doc I have... Jan 28 04:19:00 * Kero will dig for better docs tomorrow. done enough today ;) Jan 28 04:20:20 http://doc.trolltech.com/qtopia4.2/modememulator-supplementaryservices.html Jan 28 04:49:35 i dont know who had the idea to replace the OSKB with a num pad, but good job. i feel like im texting the console Jan 28 05:00:06 anyone have some code on how to access the mic? or an app that does it besides the dialer? Jan 28 05:00:39 also, anyone have source for a third party openmoko gui app? Jan 28 05:35:52 bkruse: http://sagacis.doesntexist.com/projects/openmoko-dashboard_0.10_src.tar.gz there is a GUI app source Jan 28 05:36:05 KrisAbsinthe: perfect...thank you :] Jan 28 05:36:20 I need to figure out how to make it run all the time or whatever right? so it can monitor for iax calls Jan 28 05:37:25 yup but unless an apop is closed by pressing the power button it is running in the bg Jan 28 05:38:40 so i would say that a simple "if call_receive =true then bring_to_front" or somthing like that psudo code lol Jan 28 05:39:34 well gotta sleep, stupid work and class Jan 28 05:41:36 oh, perfect Jan 28 05:41:46 lol, ya, i will try that Jan 28 05:41:56 I am not sure how to "bring to front" but I will mess with it Jan 28 06:28:12 Yay for ebay idiots. one pound ninety-eight for a bluetooth headset. Including postage. Jan 28 06:28:40 Chinglish description, combined with poor picture. Jan 28 06:32:58 2 pounds? pretty cheap. sounds fishy :) Jan 28 06:33:23 I have it in my hand. Jan 28 06:33:47 and it works? Jan 28 06:33:51 40mm*15mm*15mm and weighing almost nothing. Jan 28 06:34:03 If I had a 'normal' phone... Jan 28 06:34:48 * SpeedEvil reads 'manually using bluetooth Jan 28 06:36:36 should prepare for my second workday this month Jan 28 06:36:47 been sick all the other days :P Jan 28 07:08:51 I have a stupid python question. Jan 28 07:08:58 import dbus Jan 28 07:08:58 ImportError: No module named dbus Jan 28 07:09:10 the bluetooth python headset pairing script Jan 28 07:09:16 what do I do? Jan 28 07:23:03 http://trolltech.com/company/newsroom/announcements/press.2008-01-28.4605718236 Jan 28 07:23:46 http://www.nokia.com/A4136001?newsid=1185531 Jan 28 07:28:17 morning Jan 28 07:28:30 ljp: so when lpotter@trolltech.nokia.com? Jan 28 07:30:27 never. :) Jan 28 07:52:02 -_- Jan 28 08:17:02 good morning Jan 28 08:20:14 your clocks are fast ljp :) Jan 28 08:20:27 * ScaredyCat needs coffee Jan 28 09:36:41 coffee! Jan 28 09:37:26 !tea Jan 28 09:37:28 !macha Jan 28 09:37:37 Nokia now own Trolltech, http://www.engadget.com/2008/01/28/nokia-acquires-trolltech-the-biggest-little-company-youve-ne/1#c10091734 Jan 28 09:37:48 saw that Jan 28 09:37:56 makes you wonder Jan 28 09:38:06 wtf Jan 28 09:38:20 what does that mean for qtopia, qt, kde etc? Jan 28 09:38:25 why would Nokia buy Trolltech when they have gone down the GTK road with Maemo Jan 28 09:38:27 their linux phone must be near completeion Jan 28 09:38:36 really ... makes sense - they were using debian based stuff Jan 28 09:38:49 --> ah nar i think they want to expand their linux Jan 28 09:39:07 but they already have a GTK toolkit Jan 28 09:39:22 it's two completly different departments Jan 28 09:39:29 gtk its very good for phone interfaces Jan 28 09:39:29 isnt Jan 28 09:39:33 I think its maybe a licensing issue Jan 28 09:39:52 hasn't gone through yet Jan 28 09:40:03 maybe they just bought them to vapourise qtopia ;) Jan 28 09:40:08 it has not been ok'd by regulatory commission Jan 28 09:40:08 nar owning trolltech is a good move Jan 28 09:40:22 -hasn't gone through yet no ? Jan 28 09:40:33 Trolltech puts QT under GPL 3 and a week later Nokia buys them, hmm Jan 28 09:40:42 - :) Jan 28 09:40:43 one thing can be said it qtopia is a kick ass phone front end Jan 28 09:40:58 hmmm Jan 28 09:41:03 well i think personally nokia hates the symbian os Jan 28 09:41:04 i can agree with that :) Jan 28 09:41:21 - i have seen the n95 not be able to connect to uni network --> internet (wifi) Jan 28 09:41:28 yeah, I read the Nokia blogs and there is no love for Symbian Jan 28 09:42:06 nokia has been respectful of the gpl so far and their linux products are really cool and this is just part of a trend Jan 28 09:42:21 one thing for sure its says Linux is mature on mobiles as far as Nokia is concerned Jan 28 09:42:22 - samsung are already in the linux phone business Jan 28 09:42:22 nokia spent the whole of GUADEC abusing all other linux phone projects, so its pretty obvious they one in the pipes Jan 28 09:42:34 guadec ? Jan 28 09:42:41 NOTE: Task failed: Fetch failed: Fetch command export PATH=/home/mohamed/moko1/build/tmp/staging/i686-linux/bin/arm-angstrom-linux-gnueabi:/home/mohamed/moko1/build/tmp/staging/i686-linux/bin:/home/mohamed/moko1/build/tmp/staging/i686-linux/bin:/home/mohamed/moko1/build/tmp/cross/bin:/home/mohamed/moko1/build/tmp/staging/i686-linux/bin:/home/mohamed/moko1/bitbake/bin:/usr/local/sbin:/usr/local/bin Jan 28 09:42:43 gnome dev conference Jan 28 09:42:43 Hi all Jan 28 09:42:44 I'm using ubuntu 7.10 and I'm trying to build openmoko but the following error rises: Jan 28 09:42:44 probably nokia is commited to Linux and are gonna ditch Symbian at some point and don't want to put all their eggs in the same Linux basket so will support qt and gtk+ up to some point Jan 28 09:42:46 fatal: read error (Connection timed out) Jan 28 09:42:47 fatal: Fetch failure: git://www.denx.de/git/u-boot.git Jan 28 09:42:49 :/usr/sbin:/usr/bin:/sbin:/bin:/usr/games; git fetch git://www.denx.de/git/u-boot.git failed with signal 128, output: Jan 28 09:42:50 NOTE: package u-boot-openmoko-1.3.1+git0ec595243dc99edcd248bbcfbfd5a1dc860bde89+svn3817-r1: task do_fetch: failed Jan 28 09:42:52 ERROR: TaskFailed event exception, aborting Jan 28 09:42:53 NOTE: package u-boot-openmoko-1.3.1+git0ec595243dc99edcd248bbcfbfd5a1dc860bde89+svn3817: failed Jan 28 09:42:55 ERROR: Build of /home/mohamed/moko1/openembedded/packages/u-boot/u-boot-openmoko_1.3.1+git+svn.bb do_fetch failed Jan 28 09:42:58 ERROR: Task 28 (/home/mohamed/moko1/openembedded/packages/u-boot/u-boot-openmoko_1.3.1+git+svn.bb, do_fetch) failed Jan 28 09:43:01 NOTE: Tasks Summary: Attempted 1774 tasks of which 1774 didn't need to be rerun and 1 failed. Jan 28 09:43:03 ERROR: '/home/mohamed/moko1/openembedded/packages/u-boot/u-boot-openmoko_1.3.1+git+svn.bb' failed Jan 28 09:43:05 make: *** [openmoko-devel-image] Error 1 Jan 28 09:43:05 yickes Jan 28 09:43:12 maemo == good for pda, useless for phones Jan 28 09:43:17 mohamed: pastebin /o\ Jan 28 09:43:20 ? really xora Jan 28 09:43:24 i agree Jan 28 09:43:25 XorA: explains why they puchased trolltech, talk down GTK/Gnome then buy Trolltech Jan 28 09:43:27 !pastebin Jan 28 09:43:35 dbmoodb: I assume you never used it Jan 28 09:43:45 what an openmoko or qt on a phone ? Jan 28 09:44:15 Mononoke: what is pastebin /o\ ?? Jan 28 09:44:25 Linux_Galore: remember that Qt is cross platform, not just linux Jan 28 09:44:28 Openmoko will have to have some QT support if they want Skype Jan 28 09:44:57 i know qt powers tons of stuff Jan 28 09:44:59 Qt is easy Jan 28 09:45:03 mohamed: random page( http://pastebin.com/) where you paste your text and post only link on irc Jan 28 09:45:14 Om already has Qt support Jan 28 09:45:17 ljp: yeah but even Qtopia only use there tookits on Linux Mobile devices as far as I can see Jan 28 09:45:34 ok i have the answer Jan 28 09:45:47 Linux_Galore: doesnt mean it has to stay only on linux Jan 28 09:45:57 then again skype runs on just about anything and that QT Jan 28 09:46:28 if nokia wants to build internet applications that work on pc and nokia things, then what better way than come along and extend qt, so we might be expecting a Nokia qt voip app some time soon Jan 28 09:46:36 surely though what will win over on a smart phone is what developers find the easiest to use....not which is technically better Jan 28 09:46:38 that makes the most sense to me - and that would only be one app they could make Jan 28 09:46:40 ljp: ? maybe Nokia have been working on their own OS and want a UI toolkit ? Jan 28 09:47:43 dbmoodb: there are already a few QT VoIP apps Jan 28 09:48:03 yeah but anyone controlled by nokia ? Jan 28 09:48:39 dbmoodb: makes the relationship with Philips and Trolltech interesting Jan 28 09:48:59 ie Philips use QTopia on lots of their wireless stuff Jan 28 09:49:00 well this doesn't go through till later and i expect nokia to maintain trolltech Jan 28 09:49:03 ljp: why does the use of -verbose in qtopia build do different stuff than without, makes debugging hard :-( Jan 28 09:49:14 tis not like they can take the gpl v3 off it Jan 28 09:50:05 XorA: only difference should be a more verbsose output Jan 28 09:50:16 dbmoodb: they can fork it Jan 28 09:50:24 yes i know Jan 28 09:50:32 but they won't Jan 28 09:51:09 dbmoodb: we are talking big megga corp, being nice isnt what they are paid for Jan 28 09:51:43 ljp: here its actually ending with a different error :-( Jan 28 09:52:13 -verbose should be the actual error Jan 28 09:52:20 ljp: Im beginning to think some variable set in OE, is upsetting the configure script in qtopia in some subtle way Jan 28 09:52:24 not sure why that happens Jan 28 09:52:59 XorA: have you looked at the .bb in the qtopia sources? Jan 28 09:53:05 ljp: yes Jan 28 09:53:23 good morning all Jan 28 09:53:41 see if Nokia throw a spanner in the QTopia port for OM Jan 28 09:53:44 linux galore stop, can i go and download maemo source code ? Jan 28 09:54:13 dbmoodb: no, large parts are held under patent so you can only get binaries Jan 28 09:54:23 try again Jan 28 09:54:29 i can get all the gpled parts Jan 28 09:54:35 dbmoodb: the whole dsp is a binary Jan 28 09:54:52 http://maemo.org/development/sources/ Jan 28 09:55:12 dbmoodb: yep thats the GPL "parts" but not the whole lot Jan 28 09:55:28 the parts are greater than the whole Jan 28 09:55:58 Linux_Galore: my stupidity is enough of a spanner, nokia dont need to throw one Jan 28 09:56:02 The documentation made available for downloading on or through this site by Nokia is primarily licensed under the GNU Free Documentation License. --> Jan 28 09:56:13 they seem to be very respectful and open Jan 28 09:56:30 yes, nokia could do something evil, but surely they would have already. Jan 28 09:56:44 dbmoodb: yes but the hardware-> UI stuff, allot of it isnt open Jan 28 09:57:01 .... dude most hardware isn't open Jan 28 09:57:07 dbmoodb: heres a simple test, find out how to play ogg file on a N700/800/810 Jan 28 09:57:14 Mononoke: pastebin doesn't seem to be responding?? Jan 28 09:57:36 dbmoodb: they have locked out allot of the functionality Jan 28 09:57:57 dbmoodb: you cant play a ogg file on a N700/800810 Jan 28 09:58:04 wonder why ? Jan 28 09:58:26 sure but all i'm saying they are not an evil company as we have seen yet. linux_galore this is really not the place to say if a company is going to do something or not - slug would be slightly better Jan 28 09:58:30 "slightly" Jan 28 09:59:20 dbmoodb: I cant even write code to play an ogg file, wow not to open now are we Jan 28 10:00:09 Mononoke: the error is just that a Not Found error and then the connection times out: Jan 28 10:00:11 NOTE: Running task 1775 of 5344 (ID: 28, /home/mohamed/moko1/openembedded/packages/u-boot/u-boot-openmoko_1.3.1+git+svn.bb, do_fetch) Jan 28 10:00:13 NOTE: package u-boot-openmoko-1.3.1+git0ec595243dc99edcd248bbcfbfd5a1dc860bde89+svn3817: started Jan 28 10:00:15 NOTE: package u-boot-openmoko-1.3.1+git0ec595243dc99edcd248bbcfbfd5a1dc860bde89+svn3817-r1: task do_fetch: started Jan 28 10:00:17 NOTE: fetch http://downloads.openmoko.org/sources/git_www.denx.de.git.u-boot.git_0ec595243dc99edcd248bbcfbfd5a1dc860bde89.tar.gz Jan 28 10:00:18 --11:58:32-- http://downloads.openmoko.org/sources/git_www.denx.de.git.u-boot.git_0ec595243dc99edcd248bbcfbfd5a1dc860bde89.tar.gz Jan 28 10:00:20 => `/home/mohamed/moko1/sources/git_www.denx.de.git.u-boot.git_0ec595243dc99edcd248bbcfbfd5a1dc860bde89.tar.gz' Jan 28 10:00:21 Resolving downloads.openmoko.org... 88.198.93.219 Jan 28 10:00:23 Connecting to downloads.openmoko.org|88.198.93.219|:80... connected. Jan 28 10:00:24 HTTP request sent, awaiting response... 404 Not Found Jan 28 10:00:25 11:58:33 ERROR 404: Not Found. Jan 28 10:00:32 mohamed: dont spam' Jan 28 10:01:03 mohamed: use a pastbin service then post the url as part of the question Jan 28 10:01:47 Linux_Galore: I'm sorry I'm not spamming I was just trying to send my problem as pastbin didn't worked Jan 28 10:02:07 !pastbin Jan 28 10:02:49 http://paste.ubuntu-nl.org/ Jan 28 10:02:52 pastebin? Jan 28 10:03:11 past text files onto the web Jan 28 10:03:22 rather then spam the channel Jan 28 10:04:30 ljp: it was PEBCAK problem with -verbose Jan 28 10:06:56 mohamed: normally you would get kicked out after the first paste Jan 28 10:07:41 balrog-kun: ok sorry once more, my question was that I have a problem building openmoko (http://paste.ubuntu-nl.org/53842/) the previous url shows the task which fails Jan 28 10:09:36 XorA: thats easy to fix.. just stand up Jan 28 10:09:44 * XorA stands Jan 28 10:10:40 * balrog-kun robs the chair Jan 28 10:11:41 (One thing about the nokia discussion, then I'll drop it as off-topic; to me, the N8xx platform has seemed to be about going towards more and more proprietary solutions, both with third party apps and Nokia's in-house ones, even if much of the underlying framework itself is open. So we'll just see if qtopia stays both developed and free. Luckily I don't mind particularly much.) Jan 28 10:13:00 * * OM Bug 1199 has been created by gcasse(AT)oralux.org Jan 28 10:13:01 * * gps power status Jan 28 10:13:02 * * http://bugzilla.openmoko.org/cgi-bin/bugzilla/show_bug.cgi?id=1199 Jan 28 10:15:01 * * OM Bug 1200 has been created by gcasse(AT)oralux.org Jan 28 10:15:01 * * bluetooth power status Jan 28 10:15:02 * * http://bugzilla.openmoko.org/cgi-bin/bugzilla/show_bug.cgi?id=1200 Jan 28 10:16:01 any help with that http://paste.ubuntu-nl.org/53842/ (the connection then times out) Jan 28 10:16:37 mohamed: got a firewall between you and net? Jan 28 10:17:39 I'm not sure but I'm using ubuntu 7.10 and I've downloaded security updates, which might contain firewall Jan 28 10:17:59 XorA: should I disable it if any is working? Jan 28 10:18:00 * * OM Bug 1201 has been created by gcasse(AT)oralux.org Jan 28 10:18:01 * * gsm power status Jan 28 10:18:02 * * http://bugzilla.openmoko.org/cgi-bin/bugzilla/show_bug.cgi?id=1201 Jan 28 10:18:20 well, it's 404 so there is a working connection Jan 28 10:18:43 balrog-kun: yes, but he said after tha a connection times out, which I assume is git and svn fethcing not working Jan 28 10:19:00 as git and svn should kick in if tarball not found Jan 28 10:19:07 XorA: oh, right Jan 28 10:19:23 ljp: NOTE: package qtopia-phone-4.3.0-r1: task do_compile: started Jan 28 10:21:56 smlsfb: the gsm connector is acessible when you open the back (as if you were going to remove the battery), it's the hole near one of the screw holes. See the wiki for details. Jan 28 10:22:56 XorA: so what should I do to avoid this? Jan 28 10:23:18 mohamed: no idea Jan 28 10:24:20 you could try skipping u-boot i guess Jan 28 10:24:51 balrog-kun: how could I skip it? Jan 28 10:27:01 cesarb, cool, thanks :) Jan 28 10:27:49 mohamed: i don't know exactly sorry Jan 28 10:27:53 Kero: you can get the manual for the Enfora Enabler IIE (which uses very similar proprietary commands) and the 3gpp specifications for the official commands (07.07 has the base ones; the SMS ones should be in a different spec, you just have to find the correct number) Jan 28 10:27:59 don't have mokomakefile Jan 28 10:28:24 balrog-kun: I have, but what command should I type? Jan 28 10:31:58 well, i don't know since i don't have mokomakefile Jan 28 10:35:48 If you want to skip uboot, you also have to skip compiling the kernel, since it needs a small tool from uboot Jan 28 10:36:09 And shouldn't the fetch be able to get uboot directly from the git repository if the http download fails? Jan 28 10:38:31 cesarb: no it didn't get uboot directly Jan 28 10:41:13 ljp: Im confused is this supposed to be build qvfb for arm it x86? Jan 28 10:41:19 s/it/or/ Jan 28 10:41:19 XorA meant: ljp: Im confused is this supposed to be build qvfb for arm or x86? Jan 28 10:52:20 x86 i believe (if qvfb is enabled in configure) Jan 28 10:52:52 or rather, the host arch Jan 28 10:57:21 ljp: any clues? /home/dp/zaurus/build-neo/tmp/work/fic-gta01-angstrom-linux-gnueabi/qtopia-phone-4.3.0-r1/qtopia-opensource-4.3.0/qtopiacore/qt/src/corelib/codecs/qtextcodecplugin.h:67: Error: Undefined interface Jan 28 11:15:16 Hi folks, I'm trying to understand whether http://moblin.org/repos/projects/moko-finger-scroll.git/ relates in anyway to openmoko Jan 28 12:18:14 hi all. the mixer settings on the gta01 are really complicated. When I use "gsmhandset.state", I hear my own voice through the speaker, which makes the audio quality on a call pretty poor. Which setting should I adjust? Jan 28 12:18:23 I guess I'm after some kind of "microphone mute" setting Jan 28 12:18:34 but I've no idea which of the million options to start looking at :) Jan 28 12:25:28 Hello Everybody Jan 28 12:25:35 tonyg, http://wiki.openmoko.org/wiki/User:KrisAbsinthe Jan 28 12:25:55 KrisAbsinthe, has a lot of alsa/mixer stuff on their wiki page Jan 28 12:26:02 thanks, I'll take a look Jan 28 12:26:18 I just experimented with alsamixer and found the "Bypass Playback Volume" seems to be the setting I wanted Jan 28 12:26:29 I've set it to zero during a call and things seemed to be better Jan 28 12:26:37 cool Jan 28 12:26:53 good luck with it. i know it's been kind of a pain for some people Jan 28 12:34:31 yep, that seems to have fixed it. I'll stick a patch on my userpage Jan 28 12:35:38 awesome :) Jan 28 12:37:44 tonyg, how is the phone functioning for you? Jan 28 12:37:51 can you make/receive calls? Jan 28 12:37:54 http://wiki.openmoko.org/wiki/User:TonyGarnockJones Jan 28 12:37:55 yep Jan 28 12:38:05 cool Jan 28 12:38:08 I am, however, not using very much of the openmoko userland Jan 28 12:38:12 using it on a regular basis at all? Jan 28 12:38:16 trying to. Jan 28 12:38:22 the main problem now is power management. Jan 28 12:38:24 ahh, fair enough Jan 28 12:38:29 yeah, so i hear Jan 28 12:38:31 I get about 4-5 hours on a full charge Jan 28 12:38:42 means I have to carry around the usb connector in my pocket :-) Jan 28 12:38:55 ... fortunately, professional programmer, and lots of machines at home Jan 28 12:39:05 haha, that helps Jan 28 12:39:09 One gripe is that the USB charger I bought for my ipod isn't detected by the GTA01 Jan 28 12:39:21 so I have to have a whole PC turned on to charge the bloody thing Jan 28 12:39:36 ugh, that sounds like a poor design Jan 28 12:39:45 actually it's not Jan 28 12:39:48 well, the rationale is good Jan 28 12:39:49 it's proper design Jan 28 12:39:56 i have a charger for my palm tx, i wonder if it would work Jan 28 12:39:59 as in... honor standards Jan 28 12:40:03 mickeyl, agreed Jan 28 12:40:05 and... don't trash external hardware Jan 28 12:40:08 ahh, only charges if it trusts it? Jan 28 12:40:19 mickeyl, is there a way to force 500mA mode? Jan 28 12:40:19 yep. don't suck 500mA from unknown devices Jan 28 12:40:41 tonyg: ya. echo "fast" > Jan 28 12:40:45 that sounds reasonable Jan 28 12:40:45 **nice** Jan 28 12:40:46 thanks Jan 28 12:40:53 please look in the wiki for the correct devicenode, don't have it offhand Jan 28 12:40:53 brb Jan 28 12:40:56 ends with chgmode Jan 28 12:40:58 iirc Jan 28 12:41:12 nitpick: proc file, not device node :] Jan 28 12:41:22 well Jan 28 12:41:23 sysfile Jan 28 12:41:26 but you're right Jan 28 12:41:30 oops Jan 28 12:41:30 it's not a device node Jan 28 12:41:37 * mjr has to learn to pick his nits more carefully Jan 28 12:41:48 * mickeyl not fully awaken yet Jan 28 12:42:04 Annyway. I think I saw a GUI frontend to forcing fast charging somewhere too. Jan 28 12:42:14 probably SettingsUI Jan 28 12:42:34 hi all Jan 28 12:46:52 http://www.getmiro.com/ looks interesting.... Jan 28 12:47:12 * ScaredyCat just snarfed all the openmoko vids off youtube Jan 28 12:47:47 ScaredyCat, interesting indeed at first glance Jan 28 12:48:00 hey, it's FOSS. .. no complainging ;) Jan 28 12:48:22 and cross platorm too Jan 28 12:48:34 all I need no w is a new keyboard Jan 28 13:02:06 "boo" for firewall reboots :( Jan 28 13:11:58 boo Jan 28 13:14:10 b Jan 28 13:21:24 what would I echo to chgmode to turn off 500mA? "trickle" didn't have an effect Jan 28 13:21:31 (I'm currently plugged in to a linux box) Jan 28 13:21:40 (my apple usb charger is at home) Jan 28 13:26:58 why do you want to turn it off? Jan 28 13:28:06 Indeed I'd also like to be able to turn it off, sometimes when working with a laptop one might want the neo not to slurp at the battery Jan 28 13:28:13 ah Jan 28 13:28:35 haven't felt great enough need yet to do that myself, but it could be a nice small project... Jan 28 13:30:14 SpeedEvil, mostly just to test forcing it back on again :) Jan 28 13:30:33 hehe Jan 28 13:33:48 eh tonyg there is a good mobile meet in london next week if u interested....i am gonna go armed with Neo Jan 28 13:38:37 mjr: I have a hardware solution for that. Jan 28 13:38:50 mjr: 10cm USB cable, with the red wire cut. Jan 28 13:41:17 openmoko: 03werner * r3967 10/trunk/src/target/u-boot/ (patches/build-kludge.patch patches/series scripts/build): (log message trimmed) Jan 28 13:41:17 openmoko: Tracking upstream, with a small build error. Jan 28 13:41:17 openmoko: scripts/build: update to GIT_HEAD 98b742489c09780be6a832eeaa4e5eff824792bb, Jan 28 13:41:17 openmoko: SVN_REV 3967. Jan 28 13:41:17 openmoko: 98b742489c09780be6a832eeaa4e5eff824792bb did not build because Jan 28 13:41:21 openmoko: drivers/spi/mpc8xxx_spi.c referenced asm/mpc8xxx_spi.h, which doesn't exist on Jan 28 13:41:23 openmoko: non-MPC8xxx builds. Instead of fixing the build system to compile only drivers Jan 28 13:44:12 In this mode, /sys/devices/platform/s3c2410-i2c/i2c-adapter/i2c-0/0-0008/chgmode reports 'pre' Jan 28 13:53:09 good morning Jan 28 13:55:37 mickeyl: Who might be interested in a patch for another GSM modem support in gsmd? Jan 28 13:57:28 ScaredyCat, (echo "pre" > chgmode; cat chgmode) still produces "fast_cccv" for me. Never mind, I'll try it out when I get home Jan 28 13:57:39 zedstar, sounds interesting. what's the story? Jan 28 13:57:46 tonyg: in pratice Jan 28 13:58:02 If the battery is charged, it won't draw any more than normal operating current Jan 28 13:58:39 florian: gmsd-devel@lists.openmoko.org Jan 28 13:58:53 florian: or the bugtracker Jan 28 13:59:13 tonyg: it is mobile monday london...some industry meetup thing every month.....this month is about mobile OS's and we have various speakers from the different OS's Jan 28 14:00:00 tongy: http://mobilemonday.org.uk/ Jan 28 14:00:15 mickeyl: ok Jan 28 14:12:32 mickeyl: would you happen to know if the gsm modem in the neo can do other speeds than 115200? Jan 28 14:13:17 (WRT cpufreq and serial driver constraints on operating speeds) Jan 28 14:13:38 SpeedEvil: no idea, sorry Jan 28 14:13:51 interesting question though Jan 28 14:13:57 try the mailing list Jan 28 14:15:43 openmoko: 03thomas * r3968 10/trunk/src/target/opkg/ (opkg_cmd.c opkg_download.c): opkg: improve download error reporting Jan 28 14:17:24 <`pwgen`> hello Jan 28 14:18:23 <`pwgen`> does some know what i can do if gsmd died ( instead of rebooting ) ? Jan 28 14:19:17 yeah Jan 28 14:23:20 Hi Jan 28 14:23:57 The top information bar on the phone is often closing unexpextedly... any way to sart it back without rebooting ? Jan 28 14:24:19 <`pwgen`> the problem is , gsmd does eventnofication via socket,if it dies all communication to phonekit ist lost Jan 28 14:25:04 <`pwgen`> jmichel: xserver + dbus restart .... Jan 28 14:27:40 `pwgen` are these commands because I can't find them on the phone's terminal ? Jan 28 14:30:22 <`pwgen`> make a script with following: /etc/init.d/xserver-nodm stop;/etc/init.d/dbus-1 stop;.../dbus-1 start; ...xserver-nodm start ; Jan 28 14:31:04 <`pwgen`> an run it with : nohup /home/root/myscript.sh Jan 28 14:31:23 <`pwgen`> this should restart your gui Jan 28 14:31:33 `pwgen` Thanks I'll try this right now Jan 28 14:32:21 <`pwgen`> the problem is, the panel will be started from windowmanager and if it dies, ic could not be restartet without the windowmanager Jan 28 14:33:21 <`pwgen`> and it seems that there are probs with dbus-communication when the panel died Jan 28 15:34:44 agh, the momo thing clashes with a friend's leaving drinks, zedstar, but in general I'm up for any London-based OpenMoko-relevant events Jan 28 15:42:51 tonyg: ok cool....should be quite a few events coming up Jan 28 16:12:28 * zilux fights against bitbake Jan 28 16:12:35 advantage bitbake Jan 28 16:17:52 hello Jan 28 16:18:00 lo Jan 28 16:51:48 I'm gonna build a little app that you run in cron, monitors your charge rate, and warns you if you're doing slow charge, w/ a button to switch to fasst Jan 28 16:52:00 I've been doing gui stuff w/ pygtk Jan 28 16:52:30 Is there a scripting interface to build gtk guis that doesn't require new packages installed on neo? Jan 28 16:52:42 I don't want people to have to install pygtk to use my app Jan 28 16:53:20 wurp2|away: i would use pure gtk+ and C Jan 28 16:53:56 Well, I don't have a dev env set up Jan 28 16:54:00 MokoMakefile is giving me fits Jan 28 16:54:13 and I'm not crazy about C for anything but talking to hardware, anyway Jan 28 16:54:46 using dfu-util, but am unable to list the device, have updated the fstab with the part of the filesystem Jan 28 16:55:09 *usbfs Jan 28 16:56:11 wurp2|away: i don't see any other solutions Jan 28 16:56:52 can i seek a help on a problem: Unable to list the device using dfu-util Jan 28 16:56:59 OK, thanks Jan 28 16:57:26 dukeNukem: You don't have a /dev/ttyACM0 after plugging your neo into your linux box via usb cable? Jan 28 16:58:05 bedboi: I thought maybe someone knew of a pre-installed Tk-ish solution on the neo Jan 28 16:58:10 wurp2|away:nope Jan 28 16:58:18 There is a comment about that on the wiki, I think Jan 28 16:58:26 wurp2|away: tk stuff is quite out-dated Jan 28 16:58:29 Indicating that you probably are missing a kernel module Jan 28 16:58:53 bedboi: I actually have never done any, but for what I need, any dirt-simple thing will work Jan 28 16:59:24 wurp2|away: so python is your friend, i think that the only stuff installed is gtk+ Jan 28 16:59:24 I will go with pygtk for now, possibly migrating to C after I get a dev env set up Jan 28 16:59:34 wurp2|away: My current status: Just recieved the device couple of mins ago.. when i booted found a kernel panic Jan 28 16:59:52 dukeNukem: Try searching wiki.openmoko.org for ACM0 Jan 28 17:00:07 dukeNukem: that's what i call "welcome" Jan 28 17:00:12 dukeNukem: Yeah, the device ships with nothing burned to rootfs or kernel Jan 28 17:00:17 at least, nothing useful Jan 28 17:00:53 wurp2|away: Yup that's true, henceofth tried to start from scratch, and found a binary for dfu Jan 28 17:00:59 *henceforth Jan 28 17:01:39 dukeNukem: surely Jan 28 17:01:47 dukeNukem: What distro are you using? Jan 28 17:02:06 fc7 Jan 28 17:02:44 wurp2|away: morever my usb path is slightly different then the one asked to look at , in the wiki Jan 28 17:03:23 wurp2|away: i have /proc/bus/004/001 and morever when i lsusb, i find the device in this path Jan 28 17:06:56 dukeNukem: I would start by searching for neo and fedora core on google Jan 28 17:07:12 Worst case it sounds as if you might need to create /dev/ttyACM0 yourself Jan 28 17:07:16 Or... Jan 28 17:07:34 Download an ubuntu live boot disk and connect that way temporarily, if you just want to make it work soon Jan 28 17:07:39 Nokia bought Trolltech... Hrm. Jan 28 17:07:46 yeah Jan 28 17:07:59 lrj linked to an announcement trolltech sent to devs about it Jan 28 17:08:01 I didn't read it Jan 28 17:08:19 kdean06: Did you find the little pygtk gui script I put on the wiki? Jan 28 17:08:31 I realized after you dropped off that I don't know your email addy Jan 28 17:08:36 Nope, I didn't look for it there. :) Jan 28 17:09:16 Yeah, I dropped a comment in the chat room about it, but I didn't have much hope you'd see it since you weren't on then Jan 28 17:09:55 wurp2: fc7 is incompatible or something..? Jan 28 17:10:24 Nah, I just don't have any experience with it Jan 28 17:10:38 And apparently your install is not working Jan 28 17:10:57 but I know an install of ubuntu 7.10 works out of the box, since I use it Jan 28 17:11:04 wurp2: any other channel i can give it a shot[ur suggestion :) ] Jan 28 17:11:36 I'm sorry; I didn't understand the 'any other channel' part of that... Jan 28 17:12:07 say an other irc portal frens of yours Jan 28 17:12:12 *wurp2: Jan 28 17:12:43 I don't know of any other channel to talk about neo stuff Jan 28 17:12:53 I am no expert; I'm just loud Jan 28 17:13:02 wurp2: well said Jan 28 17:13:18 wurp2: say would it support on a gentoo live Jan 28 17:13:29 I dunno Jan 28 17:13:45 I only know that it works on mine :-) Jan 28 17:13:50 nice Jan 28 17:14:00 wurp2: thx for the all the help :) Jan 28 17:15:25 You're welcome; sorry I didn't have any real help :-/ Jan 28 17:17:00 CM: this. this is pride. http://blog.asystemapart.com/ Jan 28 17:17:41 CM: or rather, this http://blog.asystemapart.com/post/24887656 Jan 28 17:20:01 wurp2, you could try the SDK for building a little C GTK+ app Jan 28 17:20:14 in fact, the SDK may include bits&pieces for working with python/pygtk Jan 28 17:22:51 wurp2, http://wiki.openmoko.org/wiki/Toolchain#Downloading_and_installing Jan 28 17:25:06 MokoMakefile has failed for me like 30 times over 2 weeks Jan 28 17:25:43 And I'm not sure I want to devote the time to figure out how it's incompatible with my system, or the time to keep running it over & over and hoping someone will have fixed it Jan 28 17:25:59 wurp2, hence the prebuilt sdk, complete with all gcc and libraries etc Jan 28 17:26:00 So python is my prefered dev mode right no Jan 28 17:26:06 s/no$/now/ Jan 28 17:26:30 Ah, sorry, I thought you were sending me a link for mokomake again Jan 28 17:26:34 :) Jan 28 17:26:38 I will save that link Jan 28 17:26:50 Maybe I will do the app in C, thanks Jan 28 17:28:02 wurp2, much as I hate to recommend C to anyone :) Jan 28 17:28:30 (I've been doing rudimentary gui programming for openmoko in erlang, myself... using erlang's gtknode, which is just barely usable, and *still* more pleasant than C. Just.) Jan 28 17:29:49 huh? Jan 28 17:30:09 * ScaredyCat sends tonyg to the corner Jan 28 17:32:29 jana_ecal_utils_guess_location -- does that do anything remotely intelligent? Jan 28 17:36:44 * tonyg stands in the corner, happily making calls and sending sms using an erlang-based userland ;-) Jan 28 17:38:52 tonyg: My criterion is the most straightforward way to do gtk w/o requiring new packages Jan 28 17:39:16 wurp2, absolutely C then :) Jan 28 17:39:19 C is near my last choice, but it's marginally better than making people install pygtk Jan 28 17:39:55 I've done plenty of C programming, but my opinion of it is about the same as yours Jan 28 17:39:59 IIUC your app is relatively small and simple Jan 28 17:40:09 very small & simple Jan 28 17:40:38 yeah ... 'here install my python app', 'oh yea, you need a crapload of libs too... sorry' Jan 28 17:41:04 'oh, and they wont all fit, so you need to us your sdcard' Jan 28 17:41:24 ./openmoko-preferences Jan 28 17:41:27 ScaredyCat: I install almost everything to sdcard anyway Jan 28 17:41:53 everything but ipkg-link ;-) Jan 28 17:42:37 mmm... Jan 28 17:42:58 Any ETA on the Neo1973 yet, seems to keep getting delayed. :-( Jan 28 17:43:50 Maybe march if you check the community updates Daemonax Jan 28 17:44:22 if you mean the freerunner Jan 28 17:45:37 spydon: thanks I saw that in the topic, I thought freerunner wasn't the final product. Jan 28 17:46:33 Well it is the final product but the software is not ready yet Jan 28 17:46:55 Ah neat. Jan 28 18:01:18 hi Jan 28 18:31:58 Hello there. I'm a programmer from The Netherlands, and I'm trying to set up an OpenMoko development environment. I have an iMac G5 with Virtual PC installed, in which I installed Ubuntu (so as to have a standard environment). When using the MokoMakefile to make a development image, I get an error: "/usr/bin/env: bad interpreter". Does anyone have any idea what could be wrong? Jan 28 18:33:12 Witch: ouch Jan 28 18:33:18 Witch, i just installed qemu and got a few errors ..... running again solved it Jan 28 18:33:42 I don't think I can run qemu yet... nothing's built! Jan 28 18:35:37 Witch, sorry i should have said i ran the commands here http://wiki.openmoko.org/wiki/Automatic_emulation_in_Ubuntu which got a few errors which were solved by repeating the failed commands Jan 28 18:35:49 ubuntu 7.1 Jan 28 18:57:30 hmm. cat chgmode --> pre, and echo fast > chgmode; cat chgmode --> pre Jan 28 18:57:38 same for echo fast_cccv Jan 28 18:59:37 ah! the trick is "echo -n fast_cccv > chgmode" Jan 28 18:59:38 yay Jan 28 19:00:00 wurp2, your tool will need to not-newline the buffer before writing :) Jan 28 19:00:16 * tonyg_ seems to be logged in twice. doh Jan 28 19:07:43 This somehow bothers me: Jan 28 19:07:45 openmoko@openmoko-devel:~/moko$ which sh Jan 28 19:07:49 /bin/sh Jan 28 19:07:51 tonyg_: Thanks. I actually already have a script for it Jan 28 19:07:51 openmoko@openmoko-devel:~/moko$ ls -al /bin/sh Jan 28 19:07:53 lrwxrwxrwx 1 root root 4 2007-12-01 19:29 /bin/sh -> bash Jan 28 19:07:57 openmoko@openmoko-devel:~/moko$ sh Jan 28 19:07:59 sh-3.2$ exit Jan 28 19:08:00 openmoko@openmoko-devel:~/moko$ bash Jan 28 19:08:02 openmoko@openmoko-devel:~/moko$ Jan 28 19:08:03 Or is it normal that 'bash' reports itself as 'sh' (and is order of magnitude faster in starting up!) when run as sh? Jan 28 19:08:08 But already once I've forgotten to run it and ran my neo down until it shut off Jan 28 19:08:14 while it was plugged in :-) Jan 28 19:08:36 ouch :) yeah I've done that too. optimistically assumed that the usb-charger was a smartish device Jan 28 19:08:42 next morning --> no phone. Jan 28 19:08:53 Witch: bash probably detects its name and runs differently based on it Jan 28 19:09:15 wurp2: Ok, so that _may_ be normal behaviour, then... Jan 28 19:09:25 tonyg: It seems to only detect usb chargers properly if you can network through it Jan 28 19:09:49 Witch: yeah. That's a total guess on my part, but I've seen other apps do it Jan 28 19:10:48 I still have this error after following every instruction I can find: Jan 28 19:10:50 openmoko@openmoko-devel:~/moko$ bitbake/bin/bitbake Jan 28 19:10:51 -bash: bitbake/bin/bitbake: /usr/bin/env: bad interpreter: Permission denied Jan 28 19:11:38 This seems normal (to me): Jan 28 19:11:39 openmoko@openmoko-devel:~/moko$ ls -al /usr/bin/env Jan 28 19:11:41 -rwxr-xr-x 1 root root 13160 2007-03-05 06:25 /usr/bin/env Jan 28 19:11:42 openmoko@openmoko-devel:~/moko$ file /usr/bin/env Jan 28 19:11:44 /usr/bin/env: ELF 32-bit LSB executable, Intel 80386, version 1 (SYSV), for GNU/Linux 2.6.0, dynamically linked (uses shared libs), stripped Jan 28 19:12:08 Witch, "head -1 bitbake/bin/bitbake" Jan 28 19:12:43 tonyg: Jan 28 19:12:45 Witch: Did you download the files on a windows box? Jan 28 19:12:48 openmoko@openmoko-devel:~/moko$ head -1 bitbake/bin/bitbake Jan 28 19:12:49 #!/usr/bin/env python Jan 28 19:12:53 That sounds like an end of line issue Jan 28 19:12:57 maybe Jan 28 19:13:05 wurp2: Nope, Ubuntu... Jan 28 19:13:05 wurp2, could be! Jan 28 19:13:15 also, Witch, "which python" Jan 28 19:13:40 or even, "head -1 `which python` | xxd | head -5" Jan 28 19:13:47 tonyg: Jan 28 19:13:47 "file `which python`" Jan 28 19:13:48 $ which python Jan 28 19:13:50 /usr/bin/python Jan 28 19:15:06 $ head -1 `which python` | xxd | head -5 Jan 28 19:15:07 0000000: 7f45 4c46 0101 0100 0000 0000 0000 0000 .ELF............ Jan 28 19:15:09 0000010: 0200 0300 0100 0000 9087 0508 3400 0000 ............4... Jan 28 19:15:10 0000020: 7ca3 1100 0000 0000 3400 2000 0700 2800 |.......4. ...(. Jan 28 19:15:12 0000030: 1b00 1a00 0600 0000 3400 0000 3480 0408 ........4...4... Jan 28 19:15:14 0000040: 3480 0408 e000 0000 e000 0000 0500 0000 4............... Jan 28 19:15:15 $ file `which python` Jan 28 19:15:16 /usr/bin/python: symbolic link to `python2.5' Jan 28 19:15:38 :( that seems eminently sensible so far Jan 28 19:15:44 how about plain old "python"? Jan 28 19:16:07 My notes about a gui to help w/ fastcharge: http://pastebin.ca/876282 Jan 28 19:16:09 input welcome Jan 28 19:16:22 $ python Jan 28 19:16:24 Python 2.5.1 (r251:54863, May 2 2007, 16:56:35) Jan 28 19:16:28 [GCC 4.1.2 (Ubuntu 4.1.2-0ubuntu4)] on linux2 Jan 28 19:16:30 Type "help", "copyright", "credits" or "license" for more information. Jan 28 19:16:31 >>> Jan 28 19:16:52 tonyg: Yes, I've only seen 'sensible' things so far! Jan 28 19:17:26 Perhaps there's a bug in Virtual PC that mucks bash? Somehow? Jan 28 19:18:46 ! sounds awfully unlikely Jan 28 19:18:53 have you dotted-in your setup-env script? Jan 28 19:18:54 Perhaps related: Jan 28 19:18:55 $ make qemu Jan 28 19:18:57 [ -e build/qemu ] || \ Jan 28 19:18:58 ( mkdir -p build/qemu ) Jan 28 19:19:00 [ -e build/qemu/Makefile ] || \ Jan 28 19:19:01 ( . ./setup-env && cd build/qemu && \ Jan 28 19:19:03 ${OMDIR}/openmoko/trunk/src/host/qemu-neo1973/configure \ Jan 28 19:19:04 --target-list=arm-softmmu ) Jan 28 19:19:06 /bin/sh: /home/openmoko/moko/openmoko/trunk/src/host/qemu-neo1973/configure: /bin/sh: bad interpreter: Permission denied Jan 28 19:19:07 make: *** [setup-qemu] Error 126 Jan 28 19:19:18 I usually "cd build; . ../setup-env; bitbake somepackage" Jan 28 19:19:47 I would try "/usr/bin/env python" from the command line Jan 28 19:19:59 And try 'vi -b' on one of the failing scripts Jan 28 19:20:08 to check for invisible stuff in the first line Jan 28 19:20:50 itermittent didn't fit Jan 28 19:21:02 sleep time Jan 28 19:21:04 sii ja all Jan 28 19:21:07 wurp2: Jan 28 19:21:09 $ /usr/bin/env python Jan 28 19:21:10 Python 2.5.1 (r251:54863, May 2 2007, 16:56:35) Jan 28 19:21:12 [GCC 4.1.2 (Ubuntu 4.1.2-0ubuntu4)] on linux2 Jan 28 19:21:13 Type "help", "copyright", "credits" or "license" for more information. Jan 28 19:21:15 >>> Jan 28 19:21:16 Starts up without trouble. Jan 28 19:22:06 the error from the makefile is extremely bizarre. Jan 28 19:22:18 tonyg: Jan 28 19:22:20 $ . ../setup-env ; bitbake help Jan 28 19:22:21 -bash: /home/openmoko/moko/bitbake/bin/bitbake: /usr/bin/env: bad interpreter: Permission denied Jan 28 19:22:32 ok. Jan 28 19:22:35 any insight from vi -b? Jan 28 19:22:37 I'm out of clues :) Jan 28 19:22:58 wurp2: which files _should_ i vi -b Jan 28 19:23:22 /home/openmoko/moko/bitbake/bin/bitbake Jan 28 19:24:15 then if it looks OK run it by hand (just to see if it bails right away like it did run the other way) Jan 28 19:24:26 wurp2: first lines look like this: Jan 28 19:24:28 #!/usr/bin/env python Jan 28 19:24:30 # ex:ts=4:sw=4:sts=4:et Jan 28 19:24:31 # -*- tab-width: 4; c-basic-offset: 4; indent-tabs-mode: nil -*- Jan 28 19:24:33 # Jan 28 19:24:36 and if it does, then make a copy and hack stuff out of it until it stops failing Jan 28 19:24:41 then look at what you hacked out Jan 28 19:25:10 Witch: So nothing funny at the end of the #!/usr/bin/env python line? (like ^M or an unreadable character) Jan 28 19:25:35 wurp2: Immediately bails out with: Jan 28 19:25:35 $ bitbake/bin/bitbake Jan 28 19:25:35 -bash: bitbake/bin/bitbake: /usr/bin/env: bad interpreter: Permission denied Jan 28 19:25:39 A file with nothing in it but: Jan 28 19:25:40 #!/usr/bin/env python Jan 28 19:25:43 print "hi" Jan 28 19:26:13 should run, and print hi. The kind of error you're getting from bitbake should only happen if starting it as a python script is failing altogether Jan 28 19:26:30 btw, how are you running it? In a terminal running bash? Jan 28 19:26:48 ~seen writchie Jan 28 19:26:50 btw, I'm getting booted periodically. I may go away unexpectedly :-( Jan 28 19:26:51 writchie was last seen on IRC in channel #openmoko, 41d 34m 36s ago, saying: 'some 4 band 3g phones are called 3 band gsm 900/1800/1900 with UMTS 2100'. Jan 28 19:26:56 hmm Jan 28 19:28:25 wurp2: Damn and blast reboots ;-) Jan 28 19:28:26 Anyway, curiouser and curiouser! Jan 28 19:28:27 $ cat test Jan 28 19:28:29 !/usr/bin/env python Jan 28 19:28:31 print "hi" Jan 28 19:28:32 Jan 28 19:28:33 $ ./test Jan 28 19:28:35 -bash: ./test: Permission denied Jan 28 19:28:36 $ ls -al test Jan 28 19:28:38 -rwxr-xr-x 1 openmoko openmoko 33 2008-01-28 12:55 test Jan 28 19:28:43 chmod +x test Jan 28 19:28:57 It _is_ +x! Jan 28 19:29:04 O Jan 28 19:29:10 That's what makes it curious :-) Jan 28 19:29:31 What about changing python to bash, and print to echo Jan 28 19:29:33 ? Jan 28 19:29:49 wurp: what about me putting a # here in the first place :-( Jan 28 19:30:15 wurp: $ ./test Jan 28 19:30:16 -bash: ./test: /usr/bin/env: bad interpreter: Permission denied Jan 28 19:30:16 And did test come from copying & editing bitbake, or did you create it fresh? Jan 28 19:30:24 K Jan 28 19:30:35 I editted it fresh (and missed the #) Jan 28 19:30:45 Right, I see now Jan 28 19:30:47 :-) Jan 28 19:31:05 But running env from the command line works? Jan 28 19:31:15 How's 'ls -l test' look? Jan 28 19:32:44 yes, gives me all kinds of environment settings (I guess that that its job, isn't it?) Jan 28 19:33:11 $ ls -l test Jan 28 19:33:13 -rwxr-xr-x 1 openmoko openmoko 34 2008-01-28 12:58 test Jan 28 19:33:24 Yeah, if you don't give it a command to run Jan 28 19:33:38 echo $USER? Jan 28 19:34:00 and echo $SHELL Jan 28 19:34:01 openmoko Jan 28 19:34:09 ping mickeyl Jan 28 19:34:17 "/bin/bash" Jan 28 19:34:35 wow, that's impressive :-) Jan 28 19:34:40 pöng Jan 28 19:35:06 My only advice at this point is google "/usr/bin/env: bad interpreter: Permission denied" Jan 28 19:35:09 :-( Jan 28 19:35:11 I wasn't going to try to get a single / into the irc channel somehow :-) Jan 28 19:35:20 / Jan 28 19:35:26 heh Jan 28 19:35:34 just prepend with a space Jan 28 19:35:40 (showoff :-) ) Jan 28 19:36:06 Yes, but if I run it, I'd have to prepend the space somehow... Ok, typing " is more of a burden, but still... Jan 28 19:36:24 ScaredyCat: pong Jan 28 19:36:51 sooy.. Jan 28 19:37:24 mickeyl: what is the preferred method for applicaitons to do stuff with gsm... via dbus? Jan 28 19:38:19 hey, perhaps I've found something! Jan 28 19:38:57 my 'moko' dir is a different file system (I made a new file system to store OM stuff on). Jan 28 19:39:09 I guess I need an 'exec' in my mtab somewhere. Jan 28 19:39:21 ScaredyCat: in the future, yes Jan 28 19:39:32 this is pending Jan 28 19:39:34 a) the dbus API Jan 28 19:39:36 oh... Jan 28 19:39:37 b) the phone server Jan 28 19:39:44 lol Jan 28 19:39:53 ok, so I can't do much atm? Jan 28 19:39:56 both on my agenda for the next 3 months Jan 28 19:39:58 until the api is done? Jan 28 19:40:01 well, you can help me :D Jan 28 19:40:13 you might ot want that ;) Jan 28 19:40:17 not Jan 28 19:40:24 heh Jan 28 19:40:36 well, at least you could participate in the discussions Jan 28 19:40:43 all application programmers are invited Jan 28 19:40:46 since it's the API you want to work with Jan 28 19:40:47 :) Jan 28 19:40:56 see smartphones-standards mailing list Jan 28 19:41:06 linuxtogo.org/cgi-bin/mailman/listinfo Jan 28 19:41:17 ta Jan 28 19:41:35 wurp2, tonyg: I have my moko dir on a different file system. I needed the exec' permission in /etc/mtab to be able to run things from it! Jan 28 19:41:48 doh! Jan 28 19:41:49 :) Jan 28 19:42:12 Whew! Finally I can develop stuff for my next phone! :-) Jan 28 19:42:29 ...although it may take a while! Jan 28 19:43:29 * ScaredyCat waits for email Jan 28 19:49:16 I've modified the wiki so that other people getting this error message may get it earlier than I did :-O Jan 28 19:56:04 freesmartphone.org: 03emdete * r61 10/trunk/software/gsm0710muxd/ (10 files in 3 dirs): added glib/dbus stuff to muxer Jan 28 19:59:39 ~[5~[5~[5~[5~[6~[6~[6~[6~[6~[6~[6~[6~[6~[6~[6~[6~[6~[6~[6~[6~is there anything known about the date when the devices with wifi are expected? Jan 28 19:59:42 that's too long, Spida Jan 28 20:01:28 march is the best guess atm. Jan 28 20:02:46 hey guys Jan 28 20:02:55 finally got my moko back :p Jan 28 20:03:15 anyone on here that can give me some advice on qtopia images? Jan 28 20:03:45 i.e is 271107 the latest stable image (as it says on the wiki) or can I find a more up to date image elsewhere? Jan 28 20:14:57 mickeyl: do things like GetProviderName work ? Jan 28 20:15:11 in how far? Jan 28 20:15:29 like i said previously, there's no phone server code yet Jan 28 20:15:46 i just started working on the architecture of it now that i have the first API draft Jan 28 20:15:52 dbus_g_proxy_call(data.network_proxy,"GetProviderName",&error, G_TYPE_INVALID, G_TYPE_STRING, &name, G_TYPE_INVALID); Jan 28 20:15:57 should htat work? Jan 28 20:16:27 or have I misunderstood dbus calls Jan 28 20:16:40 looks good on first outlook Jan 28 20:17:33 phonekit is using this interface as well Jan 28 20:17:45 so if the object and interface paths are correct, it should work Jan 28 20:18:41 try with python Jan 28 20:18:41 hmm... Jan 28 20:18:48 eeeeeeeeeeewwww Jan 28 20:18:48 you can prototype so much faster Jan 28 20:18:50 directly on the device Jan 28 20:19:02 ok.. food first then dbus... Jan 28 20:19:24 ~bon appetit Jan 28 20:19:25 well, bon appetit is smacznego. Guten Appetit. Eet Smakelijk. God Appetitt. Buon Appetito. Buen apetito Bom Apetite. buen apetito Smaklig måltid!. Hyvää ruokahalua. Bo Proveito Jan 28 20:20:02 apt: sorry, I am on umts and got disconnected Jan 28 20:20:22 trolltech....geez Jan 28 20:20:29 Did that surprise anyone? Jan 28 20:20:40 me Jan 28 20:20:46 what the hell are nokia gonig to do wth trolltech? Jan 28 20:20:47 i did not know that they're that short on cash Jan 28 20:23:20 although the stocks speak a language everyone understands... Jan 28 20:24:12 nokia now has qt, gtk and symbian covered..... Jan 28 20:26:50 second-hand report from their press conference says that they're gonna use Qt for S40/S60(/PC) development Jan 28 20:27:02 Qt or Qtopia? Jan 28 20:27:02 and that they're not planning a linux-based phone Jan 28 20:27:25 and that Maemo will stay on the tablets with the G* foundation Jan 28 20:27:49 hmm Jan 28 20:27:49 my information doesn't say Jan 28 20:27:58 I wonder will HTC ever stop being utter cunts and releae a video driver for the TyTn II Jan 28 20:28:00 doesn't sound that convincing to me Jan 28 20:28:05 too much fragmentation Jan 28 20:28:10 but we'll see Jan 28 20:29:39 ewon: shame on them! I'm selling mine. Jan 28 20:33:30 mickeyl: agreed Jan 28 20:35:23 I dunno, it sounds like something Nokia would do :] Jan 28 20:35:47 but we'll see indeed Jan 28 20:36:08 mjr: I'm really curious about the S40 angle Jan 28 20:36:20 mjr: as I don't see how that would fit under NOS Jan 28 20:36:50 ewon, did you read harald's recent blog about TyTn? (not dir related to vid dr) Jan 28 20:37:18 I've just tried a simple SDL program that blits a coloured rectangle to the display as quickly as possible under X11. I'm seeing 13 fps when I include an SDL_flip() after each frame, and 34 "fps" if I omit the SDL_flip (so nothing is drawn, just the blit to the X-client-side screen buffer happens, presumably) Jan 28 20:37:27 That's not very fast. Jan 28 20:37:46 What are the chances of GTA02's 2D acceleration helping out in cases like these? Jan 28 20:38:47 since the bus to gpu AIUI is slower, you'd probably not want to do the blitting yourself through SDL but rather tell the X server to draw it, which then might use some acceleration :] Jan 28 20:39:10 mjr, not knowing much about SDL, wouldn't it operate that way internally? Jan 28 20:39:25 br3n: no? Jan 28 20:39:29 or do you mean draw the coloured rectangles itself? :) Jan 28 20:39:52 I assumed "blit" == "loop through an area in the framebuffer and set pixels" Jan 28 20:40:04 ewon, rumoured gpl violation turned out to be false... :-) Jan 28 20:40:06 perhaps I was mistaken Jan 28 20:40:07 mjr, no, I'm using SDL_BlitSurface Jan 28 20:40:09 oh well Jan 28 20:40:47 * tonyg hunts for the old code I had around somewhere for /dev/fb* mmap()ed access Jan 28 20:40:57 * tonyg mixes third and first person Jan 28 20:42:58 why are you using SDL? Jan 28 20:45:11 doh. setting 16bit mode doubles fps :-) (of course!) Jan 28 20:45:18 33fps with a flip after each blit Jan 28 20:45:28 mickeyl, because I'm interested in drawing without a widget-set Jan 28 20:45:35 as fast as possible Jan 28 20:45:40 using cairo occasionally Jan 28 20:45:42 hmm Jan 28 20:45:56 I'm much happier now 16bit mode works. Jan 28 20:46:02 i'd use EFL then Jan 28 20:46:04 use EFL with the software-x11 backend. Once xrender accelleration is done for gta02, you can switch to that backend. Jan 28 20:46:49 does it work nicely with cairo? Jan 28 20:46:56 and does it do sprite-style blits? Jan 28 20:47:33 (aside: is the neo's screen a 16-bit display?) Jan 28 20:48:18 heh, actually I'd have to wait to switch to the xrender backend until I had a gta02 as well :) Jan 28 20:49:16 efl is bitmap stuff, cairo is vector stuff, dunno about working together Jan 28 20:49:22 * tonyg nods Jan 28 20:49:38 and yeah 16-bit Jan 28 20:49:39 I'll check it out. Thanks for the tip. I'd glanced at it previously, but the docs and code are not as... visible as SDLs Jan 28 20:52:24 tonyg: So are you using SDL for more than the blit? Jan 28 20:52:57 cworth, not currently. I'm actually using cairo for all my drawing and it's... not fast on the openmoko. Jan 28 20:52:58 tonyg: If not, cairo_paint with a cairo xlib surface should be plenty fast, (unless the neo has very broken X drivers for some reason). Jan 28 20:53:02 Admittedly, I was using 32-bits. Jan 28 20:53:05 I'm about to try 16-bits Jan 28 20:53:09 It may run twice as quickly Jan 28 20:53:13 which would be almost usable. Jan 28 20:53:28 My plan is to cairo onto some bitmaps which I then blit around until I really need to redraw the bitmaps. Jan 28 20:53:37 tonyg: sure Jan 28 20:53:46 (unless I can get suitable performance out of plain ol' cairo) Jan 28 20:54:21 tonyg: Cacheing rendered content is often a very good idea. Jan 28 20:54:26 * tonyg nods Jan 28 20:54:54 (Though once we have nice "3d-like acceleration" then just re-rendering everthing every frame should be fine too---that's how 3D applications typically work after all.) Jan 28 20:55:14 yeah. Jan 28 20:58:35 * tonyg . o O (doh! cairo doesn't do 16 bit) Jan 28 20:59:16 tonyg: It does if you use cairo-xlib. Jan 28 20:59:35 hmm. good point. Jan 28 20:59:58 ... otoh, i'd have to write to Xlib :-/ Jan 28 21:00:05 which is tedious Jan 28 21:00:28 tonyg: Not really. Just a tiny bit of setup and then it's just cairo. Jan 28 21:00:48 XOpenDisplay; XCreatePixmap; cairo_xlib_surface_create and you're on your way Jan 28 21:01:20 tonyg: Can you extract Xlib Drawable idenitifers from your SDL surfaces? Jan 28 21:01:26 I need to take xevents too; and if I'm going to be rendering into offscreen pixmaps... Jan 28 21:01:26 Then it would be even easier. Jan 28 21:01:33 cworth, perhaps so Jan 28 21:01:44 * cworth goes for breakfast Jan 28 21:29:58 Unable to resolve kernel panic issues Jan 28 21:30:16 using Debian distro, not able to detect neo1973 Jan 28 21:30:39 the detection through dfu-util is not achieved Jan 28 21:32:22 do you have the cdc-acm module loaded? Jan 28 21:32:32 this is the USB serial port. Jan 28 21:32:41 virtual serial port Jan 28 21:33:10 when i lsusb i get the following output Jan 28 21:33:22 cdc-acm module loaded? Jan 28 21:33:23 this is the USB serial Jan 28 21:33:31 wacky. Jan 28 21:33:31 oops sorry about that Jan 28 21:33:37 Lsusb shouldn't say that Jan 28 21:33:39 :);2~ Jan 28 21:33:44 Bus 005 Device 001: ID 0000:0000 Jan 28 21:33:45 Bus 004 Device 001: ID 0000:0000 Jan 28 21:33:45 Bus 003 Device 003: ID 1457:5122 Jan 28 21:33:45 Bus 003 Device 001: ID 0000:0000 Jan 28 21:33:45 Bus 001 Device 001: ID 0000:0000 Jan 28 21:33:45 Bus 002 Device 001: ID 0000:0000 Jan 28 21:33:50 SpeedEvil Jan 28 21:33:52 1457 is the neo Jan 28 21:33:55 yup Jan 28 21:34:14 SpeedEvil: but how do i get it detected under dfu-util Jan 28 21:34:24 you know you have to press the aux button when it boots to put it into the right mode? Jan 28 21:34:52 but i am shown a kernel panic Jan 28 21:35:17 right - that's as there is no root filesystem installed Jan 28 21:35:42 SpeedEvil: Ack!! Someone was complaining about the same thing earlier Jan 28 21:35:44 SpeedEvil:yes Jan 28 21:35:53 And I totally forgot to tell them to hold the aux button when they turned it on Jan 28 21:36:19 sudo modprobe ftdi_sio vendor=0x1457 product=0x5122 Jan 28 21:36:22 deepak: So you have the boot menu showing while dfu-util is failing to find the neo? Jan 28 21:36:26 then it will work Jan 28 21:36:45 http://wiki.openmoko.org/wiki/Flashing_openmoko Jan 28 21:36:48 is of yse Jan 28 21:41:25 raster: after modprobe ,tried dfu-util --list... no result Jan 28 21:42:01 SpeedEvil:I tried there Jan 28 21:42:20 your neo should be in uboot mode Jan 28 21:42:22 wurp2|in_and_out: Nope Jan 28 21:42:51 raster: How exactly can i get into uboot mode Jan 28 21:42:52 if it is off or booted into the kernel it won't work Jan 28 21:43:12 And on Debian you need to be root for dfu-util to detect it... Just covering bases. :P Jan 28 21:43:19 it will be in uboot for about 3 seconds between power on and kernel boot Jan 28 21:43:37 deepak: it's explined on the flashing page Jan 28 21:43:44 oki got the uboot menu Jan 28 21:43:47 you need to gold down the aux button Jan 28 21:43:55 right - now try dfu Jan 28 21:44:19 On an unrelated matter. I found the bluetooth pairing script for headsets. Jan 28 21:44:29 however, python lacks dbus, what do I do? Jan 28 21:44:48 thx a lot gyuys Jan 28 21:44:53 *guys Jan 28 21:50:43 SpeedEvil: http://www.python.org/doc/current/ext/simpleExample.html Jan 28 21:51:08 Then publish the result as an ipkg (or opkg) so the rest of us don't have to do that Jan 28 21:51:16 :-/ Jan 28 21:52:14 umm - no. I assume that there must be some moderately easy way - otherwise it wouldn't ahve been put in [[manually using bluetooth]] Jan 28 21:53:31 Ah, I thought you were saying you found a script somewhere else for some other platform Jan 28 21:53:51 I'm presuming 'ipkg list | grep dbus | grep py' doesn't return anything useful? Jan 28 21:54:00 no Jan 28 21:57:26 I'm just trying to get my BT headset working :) Jan 28 21:57:52 SpeedEvil, once i get the basic u-boot menu Jan 28 21:58:07 deepak: you run the dfu... thingy Jan 28 21:58:07 SpeedEvil: I need to select the boot to usb option is it..? Jan 28 21:58:11 no Jan 28 21:58:19 don't select anything Jan 28 21:58:21 SpeedEvil: Yes i did and i detected the device Jan 28 21:58:44 SpeedEvil: Now also have the menu Jan 28 21:58:50 ignore it Jan 28 21:59:01 power off and on if you've flashed Jan 28 21:59:16 nope Jan 28 21:59:32 looking for the imae Jan 28 21:59:34 *image Jan 28 21:59:44 it'll timeout then Jan 28 22:00:08 SpeedEvil:i mean once i download it, i need to flash the rootfs right..? Jan 28 22:00:34 yes Jan 28 22:02:04 SpeedEvil:sorry about that, that was a very silley question though Jan 28 22:19:46 gn folks Jan 28 22:25:57 is the wiki down? Jan 28 22:26:21 seems like it Jan 28 22:26:23 jr1: seems so, planet. is also dead Jan 28 22:28:07 great, another reason for mokomakefile to fail (flashing qemu randomly needs a file from the wiki) Jan 28 22:30:07 it only needs to download it once in the lifetime of the qemu tree Jan 28 22:30:41 freesmartphone.org: 03mickeyl * r62 10/trunk/software/gsm0710muxd/configure.ac: gsm0710muxd: fix PKG_CONFIG check Jan 28 22:31:39 balrog-kun: which doesn't help much the new guys (there should be at least two, if I'm following the chatter right) Jan 28 22:32:23 cesarb: it's likely that there aren't many new users during the wiki downtimes :p Jan 28 22:32:59 balrog-kun: heh true Jan 28 22:33:22 * cesarb suddenly remembers he's logging his cpufreq progress on a wiki page. oh sh-- Jan 28 22:35:30 freesmartphone.org: 03mickeyl * r63 10/trunk/software/gsm0710muxd/configure.ac: gsm0710muxd: checking for dbus-glib-1 is enough Jan 28 22:43:34 wtf, the default rootfs has... strace??? Jan 28 22:43:40 (and I just used it!) Jan 28 22:45:32 Now *that's* a right proper rootfs. Jan 28 22:47:44 cesarb: the devel-image sure does Jan 28 22:47:51 * cesarb managed to make the stats code work (easy, just tell it where the frequencies table is)... and it crashed within vsprintf within the kernel. urgh. Jan 28 22:48:07 raster: is there any other? ;-) Jan 28 22:48:15 hhehehhehe Jan 28 22:48:18 * cesarb guesses openmoko-image probably doesn't even boot Jan 28 22:49:10 possibly Jan 28 22:49:32 on the gta02 it really isnt so important Jan 28 22:49:34 with 256m of flash Jan 28 22:49:49 we can throw some debug goodies in Jan 28 22:50:05 as long as its jffs2 Jan 28 22:50:09 if we move to yaffs Jan 28 22:50:20 we will not compress anything so - we'll begin to bloat up Jan 28 22:59:48 great, it looks like the stats/trans_table sysfs file is bigger than one page, which causes the code to pass a "negative" length to snprintf (PAGE_SIZE - len)... Jan 28 23:00:26 bzzzt Jan 28 23:00:30 bad code. bad Jan 28 23:03:14 hooray Jan 28 23:03:22 we still cant suspend and resume Jan 28 23:03:56 and wootsors - still cant reliably make a call and receiving a call... hahahah Jan 28 23:04:35 * raster cries Jan 28 23:04:47 raster: Overated anyway. Just change the specs to match this. ;) Jan 28 23:04:50 raster: at least the cpufreq code I'm writing is starting to work better, so there is less need for suspend/resume... Jan 28 23:05:32 cesarb: we will need suspend/resume Jan 28 23:05:33 cesarb, yaaaaay Jan 28 23:05:36 * tonyg cheers cesarb on Jan 28 23:05:46 even if we clokc down severely i doubt we will manage more than 1-2days battery life Jan 28 23:05:50 on a new good battery Jan 28 23:06:02 raster, even 1 day is better than the 4-5h I'm seeing Jan 28 23:06:18 tonyg: sure - but really - for it to be a viable phone Jan 28 23:06:22 I have to carry round the usb cable with me, and find a PC every 5 hours :-) Jan 28 23:06:23 we ned to suspend/resume Jan 28 23:06:28 raster, agreed Jan 28 23:06:49 * mwester thinks it will be even better when shutting down the phone also turns off the GSM modem. Jan 28 23:06:51 but I'd be able to use it as my personal phone with a 1day battery life Jan 28 23:06:51 my n800 can manage something just shy of a week if i leave it alone Jan 28 23:07:07 so could my sony t610, on a new battery Jan 28 23:07:12 mwester: that's relatively trivial Jan 28 23:07:30 mwester: it can already do that, you just need the script Jan 28 23:07:44 :( Technically it's trivial. OM refuses to do so Jan 28 23:07:49 mwester, is that the reason that a switched-off phone quite quickly turns unswitchonable? Jan 28 23:07:54 tonyg: Yes. Jan 28 23:08:00 the *only* reason? Jan 28 23:08:08 raster: the problem with suspend is that a program has to do nonstandard things instead of just a selec(...,timeout) Jan 28 23:08:09 ie. no other dev sucking current? Jan 28 23:08:20 s/selec/select/ Jan 28 23:08:20 cesarb meant: raster: the problem with suspend is that a program has to do nonstandard things instead of just a select(...,timeout) Jan 28 23:08:21 tonyg: at this point, I believe that everything *except* the GSM modem has been addressed via kernel patches and the like. Jan 28 23:08:33 :) righto Jan 28 23:08:53 SpeedEvil, so the script for that is basically at@poff followed by echo 0 > power_on, right? Jan 28 23:08:54 cesarb: a well written program with a well written select loop should work fine Jan 28 23:09:20 yes Jan 28 23:09:23 >tonyg Jan 28 23:09:27 neat, thanks SpeedEvil Jan 28 23:09:39 tonyg: I submitted a bug and a shutdown script to fix this problem, OM refuses to incorporate claiming this should be done by gsmd -- which is wrong in the first place, but it's even more wrong since nobody will fix gsmd to actually do it. Jan 28 23:09:42 raster: my evil plan is to combine parts of the suspend code (which has to tell the gsm modem to wakeup and turn off the serial) with cpufreq, thus allowing it to go into absurdly-deep 1MHz mode Jan 28 23:10:00 cesarb: 1mhz might be doable Jan 28 23:10:04 raster: the "sleepy linux" proposal which has been floated around some weeks ago would also be a very good trick Jan 28 23:10:20 i wonder if its feasible and how good that is power-wise compared to suspend Jan 28 23:10:27 i suspect it wont be far off being as good as suspend Jan 28 23:10:29 I also plan into testing if SLOW _really_ can't be combined with IDLE... I'm suspecting it can Jan 28 23:10:45 mwester, istm the kernel should do it? as a last-resort Jan 28 23:10:48 tho suspend is still better imho Jan 28 23:11:01 i do know my own mainloop code handles suspend perfectly well Jan 28 23:11:11 it has code in there for detecting just that Jan 28 23:11:15 and any oher hiccups Jan 28 23:11:17 eg Jan 28 23:11:22 killall -STOP proc Jan 28 23:11:25 for 1 min Jan 28 23:11:28 then -CONT it Jan 28 23:11:29 raster: of course, this is all GTA01-only, since I don't have a table of frequency settings which would be needed for cpufreq on GTA02 Jan 28 23:11:33 it will know what happened Jan 28 23:11:40 and figure things out Jan 28 23:11:51 (without signal handlers needed) Jan 28 23:12:04 the problem with 1mhz is Jan 28 23:12:05 raster: it can figure things out, but it can't easily wake up 1 minute later if it wants to Jan 28 23:12:07 tonyg: I think so, but that's difficult since the shutdown of the GSM moodem is via an "AT@POFF" command sent via the serial port; no other means exists to turn it off (it's wired to the battery directly) Jan 28 23:12:08 the system is still alive Jan 28 23:12:12 raster: say you have an alarm clock app Jan 28 23:12:17 processes that were polling will keeppolling Jan 28 23:12:20 raster: it won't run during suspend, obviously Jan 28 23:12:28 but now their polls will eat up a lot more of the cpu - if not all Jan 28 23:12:37 mwester, righto. hm. so a shutdown script, then, if not gsmd. ick. Jan 28 23:12:48 cesarb: oh - of course Jan 28 23:12:55 you need a way to queue such wakup interrupts Jan 28 23:13:03 raster: meaning you need to write the app specially to use some random neo-only api Jan 28 23:13:08 cesarb: but i believe your code will be useful for GTA02 too with adapatation? Jan 28 23:13:11 tonyg: at this point, I'd settle for just *SOMETHING* doing it. Jan 28 23:13:18 raster: or you need something like the "sleepy linux" thingy Jan 28 23:13:29 cesarb: chances are we need such a generic api for linux anyway Jan 28 23:13:37 i.e. less will be left to do on GTA02 when your code works on GTA01 Jan 28 23:14:14 sleepy linux: http://lwn.net/Articles/263180/ Jan 28 23:14:23 balrog-kun: true Jan 28 23:14:58 cesarb: great Jan 28 23:15:39 balrog-kun: the most you'd need to rewrite would be a small part of the serial driver (which is separate for each cpu version), the table of frequencies, and the hclk/pclk dividers code... that'd be about half of the cpufreq driver, since it currently is hardwired to a single cpu type Jan 28 23:15:49 balrog-kun: the really annoying parts would be all done Jan 28 23:16:23 balrog-kun: and the hardest part would in fact find a good table of PLL settings (just "max speed" and the slow modes aren't nearly enough) Jan 28 23:16:53 cesarb: but the tables can not be easily inferred from the specs? Jan 28 23:17:06 also i suppose they're board-specific, not only cpu-specific? Jan 28 23:17:06 cesarb: so basically suspend to ram allthe time (as much as possible) much like cpufreq automatic/concervative goernors may drop to a lower clock rate Jan 28 23:17:28 mwester: the *correct* way to do it would be to have *parts of gsmd in the kernel* (at least the part which talks to the AT intrepreter), so you could easily know where, when, and how to insert AT@POFF Jan 28 23:17:33 and then use interrupts from input devices for wakup (or rtc) Jan 28 23:17:37 The other part that is required is the new GSM firmware that's been claimed to be coming. Jan 28 23:17:56 cesarb: the kernel doesn't know when you want to power off Jan 28 23:18:01 balrog-kun: the tables are cpu-specific, not board-specific (well, they depend on the chosen crystal too...) Jan 28 23:18:12 cesarb: only userspace really knows that. Jan 28 23:18:24 GSM firmware that uses lower power in suspend Jan 28 23:18:33 SpeedEvil: the kernel knows you want to power off... it has to, since it's the kernel which powers the PMU off! Jan 28 23:19:30 SpeedEvil: in fact, I'm proposing a "neo GSM driver" which would do part of the gsmd job :D Jan 28 23:19:37 * cesarb is perhaps too much of a kernel programmer :D Jan 28 23:19:49 True I suppose. What I meant is that the kernel doesn't make the decision to turn off and remain in a state the modem can wake the kernel, or to turn off and turn the modem off. Jan 28 23:20:05 Some other policy thing needs to do that. Jan 28 23:20:05 cesarb: yeah, i was thinking that it may be a good idea to make gsmd talk to a kind of virtual serial port, which does the suspend/resume transparently to the userspace, on top of the physical pot Jan 28 23:20:26 the kernel would emulate that port Jan 28 23:20:41 SpeedEvil: nah, just have an IOCTL to set the desired power state (full-off or modem on), or even better leave it on on suspend and off on poweroff (the kernel does know the difference) Jan 28 23:20:43 well, once you have MUXing in kernelspace it makes sense that the MUXer controls gsm power completely Jan 28 23:20:53 balrog-kun: exactly Jan 28 23:21:04 ^^ what mickeyl said. Jan 28 23:21:05 mickeyl: true of course. Jan 28 23:21:23 mickeyl: it'd prolly have to be a bit above the muxer, since the command would be muxed itself, just like ICMP with IP... Jan 28 23:21:23 ~seen writchie Jan 28 23:21:27 writchie was last seen on IRC in channel #openmoko, 41d 4h 29m 11s ago, saying: 'some 4 band 3g phones are called 3 band gsm 900/1800/1900 with UMTS 2100'. Jan 28 23:21:35 woah 41d Jan 28 23:21:43 And after looking at the mess that is the serial driver, a pseudo-driver that can be more carefully controlled would be a good idea. Jan 28 23:21:52 mwester: yeah Jan 28 23:21:59 mickeyl: He took a month of for finishing the kernel muxer ;) Jan 28 23:22:04 *cough* Jan 28 23:22:06 hope dies last Jan 28 23:22:18 mickeyl: True for you. :) Jan 28 23:22:20 man Jan 28 23:22:27 mwester: horrible race conditions seem to emerge if you try to jam all the suspend/resume/... into the serial driver, yes. Jan 28 23:22:35 raster: dude Jan 28 23:22:48 if gsmd/libgsm dont radically improve soon i'm going to be forced to replace them with something that works... just out of frustration Jan 28 23:22:49 :) Jan 28 23:23:03 SpeedEvil: Yep - as proven by something as simple as the console driver living in there along with the normal serial driver. Jan 28 23:23:14 well Jan 28 23:23:17 somehow other serial drivers can deal much better with suspend resume (or even with just plain functioning) than the s3c24xx serial Jan 28 23:23:27 * stefan_schmidt suggest raster uses his tuesday dinner break for this task. Jan 28 23:23:31 once the dbus API is done, the phone server will be started Jan 28 23:23:37 and once this works, we can do a new backend Jan 28 23:23:40 mwester: the console driver isn't "simple" Jan 28 23:23:42 getting rid of libgsmd Jan 28 23:23:56 mwester: it has to run in unusual situations (before most drivers are up, during a crash...) Jan 28 23:23:57 balrog-kun: I'm thinking the problem may have to do with the FIFO; there's not a lot of special handling in that driver for the various special modes one might wish to use. Jan 28 23:24:01 i'm sitting out at a nice loounge bar last night Jan 28 23:24:03 by that time all applications use the dbus API and nothing needs to be changed on the upper layers Jan 28 23:24:05 mwester: which makes it do strange and unusual things Jan 28 23:24:15 and it took 3 boots and 8 tries until fdialler finally made a phone call. Jan 28 23:24:23 of course incoming calls didnt work Jan 28 23:24:32 and this is like an image i flashed on last week Jan 28 23:24:40 cesarb: Yep, that's exactly right. So it ends up being a very intrusive driver once the real driver is operational :( Jan 28 23:24:58 ya, gsmd is the weak spot right now Jan 28 23:25:00 but like i said Jan 28 23:25:05 i think that's a sane plan Jan 28 23:25:12 mwester: I'm not even TRYING to LOOK at it to see if I need to hook something cpufreq-related in the console part Jan 28 23:25:40 cesarb: There's nothing there that's aware of the freq or timing. Jan 28 23:25:48 (which is part of the problem) Jan 28 23:26:22 For example, did you notice that there's absolutely NO locking in the serial driver to protect it from other interrupts? Jan 28 23:28:21 mwester: the locking is about 2 layers above Jan 28 23:28:29 mwester: in the generic serial driver Jan 28 23:28:45 So at this point the new fix for the infamous bug #788 is looking to be to just drop the console output on the floor if flow control is turned on; it's just too big a rewrite to do the locking so that the console driver can "barge in", break flowcontrol, transmit, and enable flow control. Jan 28 23:29:57 cesarb: That only works if everything below goes through that layer; which is not the case if we do anything interesting in the console code. Jan 28 23:31:19 mwester: the correct way to do it would be to use the unlocked code during early boot and use the locking after a certain point Jan 28 23:31:46 Yes. I think there are other drivers that seem to do that (there's locking in the bluefin driver, IIRC). Jan 28 23:31:54 * cesarb would like to know why there isn't a generic serial console layer for it to use. Or perhaps there is, I didn't look closely Jan 28 23:33:16 There is such a layer, but the normal tty layer and the console layer converge at the hardware, and the console requires that a character be transmitted by the hardware (not buffered) -- so it uses a separate path to the hardware. Jan 28 23:34:08 So if you call printk, printk will not return until the final character has been placed into the UART TX register. Jan 28 23:35:30 cesarb: A simplification would be to assume that during normal (non-debug) situations, the GSM modem will be active, and therefore there will never be any activity on the console at all. Jan 28 23:38:25 mwester: isn't that simplification isomorphic to "don't use console=ttySAC0 on uboot unless you are debugging"? Jan 28 23:40:01 The kernel code that powers up the GSM modem and switches the serial port mux to the modem also suspends console activity on that serial port; so there will be no activity during that timeframe. Jan 28 23:41:08 Removing "console=ttySAC0" is insurance to avoid the corner cases that can result in lockups during the transition in and out of that console-suspended state. Jan 29 00:20:39 ok goodnight folks. Jan 29 00:33:17 * cesarb looks at s3c2410_udc.c and wonders if the USB clock and cores are enabled even if there isn't anything plugged in Jan 29 00:33:48 Does anyone know which pin can I look at to know if there's power on the USB power line? Jan 29 00:34:18 cesarb: measure between the outer ones Jan 29 00:34:19 cesarb: Test? I mean there are just 5 :) Jan 29 00:34:27 cesarb: it's one of those on the side. Jan 29 00:34:32 cesarb: the neo one? Jan 29 00:34:37 SpeedEvil: I mean from within the SoC... Jan 29 00:34:48 cesarb: ah - don't think it's hooked up Jan 29 00:34:55 cesarb: only through the PMU Jan 29 00:34:57 I want to somehow make the driver know it's got Vusb *before* it enables the USB core Jan 29 00:35:12 and make the cpufreq dance stop as long as it's plugged into something Jan 29 00:35:35 The minor problem is of course, it's quite possible to be connected to a UsB host not supplying any power. Jan 29 00:35:43 but that's probably not the usual case Jan 29 00:36:27 I just found a benchmarks page. Jan 29 00:36:57 Current processors are at about a million times the speed of a pdp/11-03 Jan 29 00:37:21 The neo is probably on the order of a hundred thousand times faster. Jan 29 00:38:47 SpeedEvil: connecting to a powerless host is not a problem... lots of devices fail in that situation too (it seems it's quite common to use Vusb as a signal to actually turn on the pull-up/down in the data line which signals the host it's on) Jan 29 00:39:08 SpeedEvil: unless you really want to use it OTG style Jan 29 00:39:45 cesarb: I know - I was just commenting as I've used it in the 'cut red wire' case - for battery current testing Jan 29 01:50:44 YAY got the Test GTA01 today, GSM 850 works Jan 29 02:15:20 KrisAbsinthe: it does? **** ENDING LOGGING AT Tue Jan 29 02:59:57 2008