**** BEGIN LOGGING AT Thu Sep 11 02:59:57 2008 Sep 11 03:34:07 hi all.. i'm trying to build a kernel module (and a userspace application) using Mokomakefile. I have followed the procedure in Mokomakefile (quilt new, add, edit, refresh). The build succeeds and produces 3 ipks (package, dbg and dev) for each package. The problem is that all the packages are empty. There was no error during make. What could be the problem. Is there a way to have a complete debug information (some switch perhaps) so I can Sep 11 03:34:07 see the details of the make? Sep 11 03:36:32 What could be the problem. Is there a way to have a complete debug information (some switch perhaps) so I can Sep 11 03:36:35 Your sentence ends there. Sep 11 03:36:45 Ah, never mind. Sep 11 03:36:50 ... see the details of the make? Sep 11 03:36:52 :) Sep 11 03:37:08 (my chat client split the msg) Sep 11 03:37:28 You created a bitbake recipe for your module? Sep 11 03:37:45 yeah. as described in the Mokomakefile wiki page Sep 11 03:38:22 but it included only the source, the inherit autotools and description version etc. My packages both require only a ./configure, make, make install.. so autotools should be fine Sep 11 03:38:47 You might wanna read http://bec-systems.com/oe/html/recipes_packages.html then Sep 11 03:39:58 hnn.. alright. reading now Sep 11 03:42:42 Also, using bitbake in debug mode may help you resolve this issue more easily Sep 11 03:43:55 1) as the link says, the default should work to create the base package (which it doesn't) and 2) i'm using mokomakefile. can you please tell me how to pass a switch or something through the makefile to bitbake for debug mode? Sep 11 03:45:00 source setup-env; cd build; bitbake -f -c compile name_of_your_bitbake_recipe Sep 11 03:45:17 That's the basic invokation for bitbake using mokomakefile Sep 11 03:46:09 You can then add -v or -D to the bitbake command to make it more verbose Sep 11 03:46:40 hnn. alright. i'll try that Sep 11 05:27:54 my moko build environment is of latest version(2008.08), when im trying to build a sample helloworld package using bitbake method it is throwing error like " sqlite3.OperationalError: attempt to write a readonly database " . can anyone help me in solving this error?? Sep 11 05:32:01 file permissions? Sep 11 05:32:17 does mount show the filesystem rw? Sep 11 05:43:24 tmzt_, if i tried with root then it is saying that " Do not use Bitbake as root ". Sep 11 05:45:52 maddy: Sounds like your install of sqlite3 is borked. Maybe re-installing helps Sep 11 05:51:22 abraxa_: initally it failed two times but third time it build successfully Sep 11 06:10:12 im trying to build trousers using the binary openmoko toolchain ... I get this error with "om-conf trousers": configure: error: openssl is currently the only supported crypto library for trousers. Please install openssl from http://www.openssl.org or the -devel package from your distro Sep 11 06:10:30 FATAL: oe_runconf failed Sep 11 06:13:37 but I already have gnutls-openssl.a in my toolchain lib dir Sep 11 06:21:22 Yes. But gnutls != openssl; check your configure script to see if it requires the openssl lib or if can use he gnutls version. Sep 11 06:24:34 it requires openssl... Sep 11 06:25:07 so, what DO i need then? i have libssl-dev installed on the build system Sep 11 06:25:39 (and i have libssl.a in the toolchain lib too) Sep 11 06:27:37 It can be a bit of a challenge at this point, as there are so very many things that could be wrong. Sep 11 06:27:53 :O ... that was a generic response :) Sep 11 06:28:16 You need to make sure exactly what it is that trousers needs -- abcedf.a and yweiyui.so.4.3.0 and fjdsklfj.h and fildfkjkfjslk.h, etc. Sep 11 06:28:44 You need to make sure that all of those, whatever those really are, are installed in the toolchain int he correct places. Sep 11 06:28:58 hnn. alright Sep 11 06:29:00 Then you need to make sure that the confgure script for trousers is written correctly. Sep 11 06:29:16 ( the configure script is alright ) Sep 11 06:29:27 it compiles fine natively Sep 11 06:29:27 You've used that configure script to cross-compile before? Sep 11 06:30:02 Many configure scripts do not handle cross dev environments properly; it's rather a common issue. Sep 11 06:31:05 no. that could be a problem. it compiles fine with native builds... that's whta I meant Sep 11 06:31:17 i'm checking into your suggestions. Sep 11 06:32:04 There's no "trousers" (or pants, or shirts even) in the OE repository, so that package has not been built by OE or that toolchain AFAIK. Sep 11 06:32:29 :) its not in OE. Sep 11 06:32:53 Frankly, the easiest way to deal with it is to set up an OE environment. Sep 11 06:33:13 The toolchain is a snapshot of a build of part of that environment Sep 11 06:33:14 no pants? Sep 11 06:33:17 blasphemy! Sep 11 06:33:17 i tried this with openmoko's makefile too Sep 11 06:33:47 for some reason, got many dependency errors Sep 11 06:34:17 If you need to add a package, you can just state it's dependencies, and you can be sure that they are correctly staged -- it's much easier than worrying if the problem is related to the toolchain not having the correct libs, or such. Sep 11 06:34:44 hnn. i guess you're right about that Sep 11 06:34:45 Well, if you got dependecy errors in OE, it isn't likely to be better with the toolchain that OE build. :D Sep 11 06:35:36 raster: no socks either. Sep 11 06:35:42 BOO! Sep 11 06:35:46 need socks Sep 11 06:36:00 * mwester checks to see if it has boots. Sep 11 06:36:25 Well, it has bootsplash. Sep 11 06:40:01 Trousers: "The open-source TCG Software Stack". The web page goes on to throw out several hundred other acronyms, but nowhere at all does it describe what trousers does, who might use it, or even define any acronyms (except that the bottom of the page offers the clue that the "TCG" in TCG Software Stack might be "Trusted Computing Group'. Sep 11 06:40:07 Good Grief. Sep 11 06:40:51 I hate it when developers decide that "if you don't know the acronyms, you're clearly too stupid to use our software". Sep 11 06:41:49 i hate it when developers support half arsed system overhead adding technologies ... Sep 11 06:41:52 trousers computing group Sep 11 06:41:57 computing in your pants Sep 11 06:42:25 That's a freerunner in my pocket! Sep 11 06:43:59 mwester: i think i may have a solution to my problem. tell me one thing. I need bitbake to call configure script with a "--with-gui = none"... how can I put that in my bitbake file (or in some other way, tell this to bitbake) ? Sep 11 06:47:35 Check the existing .bb files for examples. Search for something like: EXTRA_OECONF = "--with-gui=none" Sep 11 06:48:04 (and by the way, TSS is a trusted software stack... it helps applications communicate with the low level device drivers for the trusted compting chip called the trusted platform module.. in case you're interested) Sep 11 06:48:06 Note that in a bb file, you must have spaces around the = sign (it's not shell-script-style syntax Sep 11 06:48:08 ) Sep 11 06:48:21 hnn. Sep 11 06:48:28 ok. i'll find it Sep 11 06:48:51 The openssh bb file should have a pretty complete set of overrides for things like that, IIRC Sep 11 06:50:11 mwester, I already have a package openssh in my build environment. can't I just put it as a dependency for trousers? Sep 11 06:50:30 * openssl Sep 11 06:50:41 And I am interested in what software does; always curious -- just astonished when a group of folks has obviously gone through a lot of trouble to write some software, but erect a web site that seems that it must have been deliberately designed to be a barrier-to-entry! Sep 11 06:50:58 Yes, you can just list it as a build-time dependency Sep 11 06:51:12 DEPENDS = "openssl" (or whatever) Sep 11 06:51:34 bitbake will then build and stage that package before it runs configure on yours. Sep 11 06:53:01 ok. that should work :) Sep 11 06:54:30 gotta run -- good luck! Sep 11 06:54:36 thanks man Sep 11 06:56:25 morning Sep 11 07:04:28 who is the shr "master" responsible? Sep 11 07:04:34 ;) Sep 11 07:04:40 i'd like to ask some questions Sep 11 07:05:15 do you think that it's interesting, as a community distro, to get more devices supported? Sep 11 07:07:32 i have created a deb package abcd.i386, and im trying to cross-compile it on Arm using qemubuilder. but when im trying to compile the package using qemubuilder im getting segmentation fault. Sep 11 07:19:35 and while we're at it, can someone tell me how I can tell bitbake that my package requires gtk-2.0. I added gtk as dependency but it says, " ERROR: Required build target 'trousers' has no buildable providers. Missing or unbuildable dependency chain was: ['trousers', 'gtk']" (i also tried gtk2.. didn't work) Sep 11 07:25:22 there's no depend in the trousers .bb for trousers, right? Sep 11 07:25:58 yes. there is none Sep 11 07:26:24 of course trousers can't depend on trousers Sep 11 07:26:39 :) Sep 11 07:26:53 i have DEPENDS = "openssl gtk" Sep 11 07:27:12 what does gtk-2.0 provide, gtk or gtk-2.0 or something else? Sep 11 07:28:09 i'm not sure. i do know that my package trousers requires gtk-2.0 Sep 11 07:28:19 is there a way to find out? Sep 11 07:28:25 I mean what is the package name in the gtk*.bb file Sep 11 07:29:47 that's the problem. I can't find a .bb file for gtk Sep 11 07:29:58 there are many but they're coming from other packages Sep 11 07:32:11 i'm trying DEPENDS = "gtk+" Sep 11 07:32:31 yes Sep 11 07:32:37 should be gtk+ Sep 11 07:58:38 i have the sources to this package on my harddisk. I've changed the contents of the source tarball but the build environment is still reading from some cache. where is this cache stored? i need to flush it Sep 11 11:20:25 morning Sep 11 11:45:37 hi mickey Sep 11 11:45:44 what ICE690 means? Sep 11 12:05:21 methril|work: ICE is a a fast gtrain in germany. The number is just the train number. Sep 11 12:05:39 methril|work: mickey like to let people know he is on a train. Sep 11 12:05:41 thanks stefan Sep 11 12:05:47 hehehe Sep 11 12:06:04 Explains his on and off hoping. ;) Sep 11 12:09:33 http://tinderbox.openembedded.net/packages/linux-openmoko/ <- lots of failures Any idea why ? Sep 11 12:31:27 <|FiLiP|> do_fetch aparently failed Sep 11 12:48:51 OdyX: bitbake git fetcher is broken Sep 11 15:18:48 Hi everybody Sep 11 15:19:33 is there somebody who can tell me how to install the linux-headers package in the ubuntu-fso distribution Sep 11 15:19:35 ? Sep 11 15:21:17 ubuntu-fso distribution? Sep 11 15:21:27 You mean debian distro? Sep 11 15:22:26 yeah sorry, of course Sep 11 15:23:04 i meant the debian distro Sep 11 15:26:23 Hrm, did you try dpkg -l *linux*header* ? Sep 11 15:26:29 to see the package name Sep 11 15:27:02 I get the following message Sep 11 15:27:02 No packages found matching linux-headers-2.6.24-etchnhalf.1-all_2.6.24-6~etchnhalf.4_arm.deb. Sep 11 15:27:24 try dpkg -l \*linux*header* Sep 11 15:27:56 ok this gives me 5 packets Sep 11 15:28:18 linux-headers-2.6-all Sep 11 15:28:23 linux-headers-2.6.24-all Sep 11 15:28:34 linux-headers-2.6.24-etchnhall Sep 11 15:28:37 linux-headers-2.6-etchnhall Sep 11 15:28:42 linux-kernel-headers Sep 11 15:28:50 is one of this the right one? Sep 11 15:29:01 I think the last one is, but I'm just guessing Sep 11 15:29:14 ok i'll give it a try Sep 11 15:29:47 thanks much for the trick with "dpkg -l" didn't know this ;-) Sep 11 15:30:12 You're welcome. I wouldn't have remembered it off the top of my head, but I was doing it last night on my ubuntu box :-) Sep 11 15:30:37 I always have to look it up, which is weird, because it's not complicated... I just don't do it very often. Sep 11 15:30:54 ;-) Sep 11 15:31:08 ok i installed it but it said to me that the package is already installed Sep 11 15:31:19 however the directory /lib/modules/2.6.24/build still does not exists Sep 11 15:31:28 should i try to install the others? Sep 11 15:31:38 Yeah, I would try linux-headers-2.6.24-all Sep 11 15:31:54 In fact, the output from dpkg -l indicates which ones are already installed, I think Sep 11 15:33:16 i don't think the headers are installed, since the module i try to compile, complains about missing header files Sep 11 15:33:30 furhter more installation of linux-headers-2.6.24-all gives Sep 11 15:33:31 E: Couldn't find package linux-headers-2.6.24-all Sep 11 15:33:39 strange Sep 11 15:33:41 Yeah, I was just suggesting that you can see which ones you don't need to try, because they're already there Sep 11 15:33:47 yeah Sep 11 15:34:29 Since i found no other way, i installed synaptic and search for linux-headers Sep 11 15:34:42 but the only headers found were for a 2.6.26 kernel Sep 11 15:34:54 this would be the other solution: upgrading the kernel Sep 11 15:35:01 but is this feasible? Sep 11 15:35:10 Ack, I would be scared of that Sep 11 15:35:26 Did you try linux-headers-2.6-all or linux-headers-2.6-etchnhall? Sep 11 15:35:38 yes for the first one Sep 11 15:36:12 and yes for the second one. Sep 11 15:36:15 :-/ Sep 11 15:36:17 Same error: package not found Sep 11 15:36:29 maybe there is something wrong with the sources.list Sep 11 15:36:31 ruppena: Try asking in openmoko-debian Sep 11 15:36:45 sudo apt-get update might help, too Sep 11 15:36:51 already done Sep 11 15:37:27 oh didn't know they have i chan, on the site they indicated this chan Sep 11 15:37:28 thx Sep 11 15:37:35 yw Sep 11 17:15:24 freesmartphone.org: 03sudharsh 07openmoko-gsoc2008 * rd9e485cd7f30 10/odeviced/ (src/plugins/power/power.vala vapi/odeviced.vapi): Sep 11 17:15:24 freesmartphone.org: sync power API with frameworkd Sep 11 17:15:24 freesmartphone.org: builds with vala-svn now Sep 11 17:16:37 mickey|zecke, is this zecke or mickeyl? Sep 11 17:17:24 * mickey|zecke at zecke's place Sep 11 17:18:05 gprs actually is not too bad Sep 11 17:18:19 (on the neo) Sep 11 17:19:32 mickey|zecke, ah, ok... happy hacking.. i will be giving a talk on FSO at a Foss-subconference.. i shall send the slides to the smartphones-ml for review.. and I am stuck with some pkgconfig while trying to build my code for release :(.. Sep 11 17:20:02 (talk) ah, cool Sep 11 17:20:03 (pkgconfig) bummer, Sep 11 17:20:44 i get "checking for GLIB... configure: error: The pkg-config script could not be found or is too old." in my package that DEPENDS on glib-2.0.. But build/tmp/staging/x86_64-linux/usr/bin/pkg-config --list-all shows that the glib-2.0 is present.. additionally i did a make rebuild-package-glib-2.0 Sep 11 17:21:03 (x-paste from #oe) Sep 11 17:21:40 mickey|zecke, are you irc'ing on gprs? Sep 11 17:22:03 yep Sep 11 17:22:05 on my neo Sep 11 17:22:39 using FSO Sep 11 17:22:41 ms3 :) Sep 11 17:23:18 mickey|zecke, i did flash ms3, but didnt get a chance to play with it :/, i was hoping to release immediately after ms3 announcement Sep 11 17:23:58 redownload Sep 11 17:24:05 we had a showstopper Sep 11 17:24:16 and just updated Sep 11 17:24:17 i'm sending the announcement now Sep 11 17:24:55 mickey|zecke, oh, i downloaded it yesterday evening... when it was still UPLOAD_IN_PROGRESS Sep 11 17:26:12 suspend doesn't last more than 30 seconds Sep 11 17:26:17 Can you help me read some dbus error ? http://pastebin.com/d1a9553cf ? thx ... Sep 11 17:26:19 and resume by itself Sep 11 17:26:21 on MS3 Sep 11 17:27:14 «soc-audio: resume work completed» Sep 11 17:27:14 hmm, sleeps pretty long here Sep 11 17:27:27 debug logs appreciated Sep 11 17:33:28 mickeyAtZecke, http://pastebin.com/m467c90f9 Sep 11 17:34:53 and yes i 've got a microSD card plugged Sep 11 18:16:33 Hi. Is a "WiFi connectivitiy" (or GUI) planned in any future milestone of FSO (or it is out of the scope of FSO) ? Sep 11 18:22:46 OdyX, iwconfig IS the tool which allow wi-fi connection Sep 11 18:23:28 interface is up and is called eth0 of mine Sep 11 18:36:51 hi everybody Sep 11 18:38:02 i am trying to port frozen bubble and make controls use accelerometers... Sep 11 18:38:20 1st step a python wrapper script Sep 11 18:38:28 2nd step native c Sep 11 18:38:46 admiral0: Did you look at how Duke Nukem 3D is doing it? Sep 11 18:38:47 my only problem is setting the toolchain Sep 11 18:39:08 no not yet Sep 11 18:39:20 admiral0: unless you have probems w/ the python wrapper script, why switch to C? :-) Sep 11 18:39:21 (but i love dukenukem 4 freerunner) Sep 11 18:39:34 speed Sep 11 18:40:02 i think that even the frameworkd is a bad idea (the python thing) Sep 11 18:40:45 python is powerful, but not very good choice for an embedded device Sep 11 18:41:00 admiral0: I think you should see if you have a perf problem (and where it is) before just rewriting Sep 11 18:41:11 embedded devices now have 200 mhz + cpus Sep 11 18:41:40 i know... but i am an efficiency maniac :D Sep 11 18:41:51 cool :-) Sep 11 18:42:08 we are going very OT from my issue Sep 11 18:42:11 :P Sep 11 18:42:24 I am an efficiency maniac, but for *my* time, not my computer's time Sep 11 18:42:26 how do i use the toolchain... Sep 11 18:42:27 heh Sep 11 18:42:44 Can't you just dl it and compile? Sep 11 18:42:52 no Sep 11 18:43:05 you can also think about compiling on the neo, in which case all those cross-compile issues go away Sep 11 18:43:10 there is a thing called dependency hell Sep 11 18:43:14 admiral0: ugh Sep 11 18:43:27 chroot, baby Sep 11 18:43:36 i compiled on the FR... Sep 11 18:43:52 So whadaya need the toolchain for? Sep 11 18:43:52 ages Sep 11 18:43:55 milleniums Sep 11 18:45:02 now i think my only option is http://wiki.openmoko.org/wiki/Host-based_development_with_Xoo_and_Xephyr Sep 11 18:45:51 Ainulindale: you never answer my love letters any more Sep 11 18:46:50 wow, love letters. A new way to get in contact :D Sep 11 18:47:08 or not, in this case Sep 11 18:47:08 admiral0: it's actually quite old ;) Sep 11 18:47:15 lambda: OK. So I should just build or port an appropriate gui ;) Sep 11 18:47:17 admiral0: the toolchain is a good choice for "Hello World", but if you have a development effort that requires supporting packages and libraries that are not already part of the tool chain, then you should just set up the Master Makefile build environement and use that. Sep 11 18:47:59 mwester: you mean mokoMakefile? Sep 11 18:48:24 Yep Sep 11 18:48:26 i'm not a openmoko programming god... Sep 11 18:48:32 You will be soon. Sep 11 18:48:38 mwester: i tried it once Sep 11 18:48:55 Pretty cool, huh? Sep 11 18:49:01 yes Sep 11 18:49:25 but i rm -rf'd it because i was angry at it Sep 11 18:49:50 I rm -rf it regularly, just because I don't trust incremental anythings. Sep 11 18:49:58 after 5h of compiling failed the compilation of qt Sep 11 18:50:16 that's why programmers suicide Sep 11 18:50:22 lol Sep 11 18:52:27 I'm building feeds as well as multiple images from the OE environment; it's not unusual for me to encounter a failure 24 hours or later into a build. Sep 11 18:52:56 Although if i devote both cores to building 2007.2, it usually completes in only 8 hours or so. Sep 11 18:53:15 O_o Sep 11 18:53:35 (qt is a pain, because that part cannot be built with any parallelization, so only one core can be in use) :( Sep 11 18:54:10 what do you think about Host based development? Sep 11 18:54:32 Not practical for anything beyond "Hello World". Sep 11 18:55:08 it's about the libs, isn't it? Sep 11 18:55:18 Insufficient RAM, poor I/O speed to the SD card, and even with 8GB, not enough space for an entire distro build. Sep 11 18:56:15 Although it may be a very practical approach to things like kernel modules -- they tend to be relatively self-contained, and only depend on their own sources and the kernel headers. Sep 11 18:57:17 I do intend to have gcc on my freerunner at some point -- there's something just too "over the top" about having a full C/C++ compiler on one's cell phone.... Sep 11 18:57:24 :-D Sep 11 18:57:59 no i meant that one: http://wiki.openmoko.org/wiki/Host-based_development_with_Xoo_and_Xephyr Sep 11 18:58:04 (Although python is probably a better on-device dev environment than C/C++) Sep 11 18:58:18 although i know that compiling on the FR is slow as hell Sep 11 18:58:40 mwester: maybe you can answer my question: I had some issues linking qt-phone-x11 because some X11 libs were missing on the host. Why would that package link against host libs? Sep 11 18:58:52 bug Sep 11 18:59:14 mwester: Hmm, I wonder about building a big ramdisk on your PC and sharing it as a drive (and swapspace) on the neo Sep 11 18:59:15 it should never reference any host libs or header files Sep 11 18:59:24 mwester: well that's what I thought, but I wasn't shure. Sep 11 18:59:26 I'd be interested to see what that did to compile time Sep 11 18:59:45 wurp2|working: somebody has an iscsi initiator already running on the freerunner over wifi. Sep 11 18:59:45 wurp2|working: not very much Sep 11 18:59:54 it has only 400MHz Sep 11 19:00:09 admiral0: My first computer had 4khz Sep 11 19:00:23 mwester, did you already compile a kernel module on the fr? Sep 11 19:00:35 Memory is a much bigger constraint than cpu speed, almost always Sep 11 19:00:36 wurp2|working: and you were compiling large apps on that machine? Sep 11 19:00:38 ;) Sep 11 19:00:49 ruppena|home: No, I use a cross environment -- that's just a reasonable thing to do once it stabilizes a bit more. Sep 11 19:00:52 wurp2|working: in to that end it's disk IO. Sep 11 19:01:06 Kensan: Well, I compiled large apps 3 months ago on my 750 mhz desktop Sep 11 19:01:18 750? Sep 11 19:01:30 P-III as well ;) :D Sep 11 19:01:38 motherboard had no headers for the cpu frequency jumpers :-O Sep 11 19:01:38 mwester, how do you compile a kernel module with cross environment? Since you need the right headers... Sep 11 19:01:42 wurp2|working: you cannot compare a desktop machine to an embedded device simply by comparing the cpu freq Sep 11 19:01:45 wurp2|working: my first computer a Pentium II 350Mhz Sep 11 19:01:55 Kensan: That was part of my poinot Sep 11 19:02:00 *point Sep 11 19:02:00 (if my memories are correct) Sep 11 19:02:14 I know that arm cpu is much slower than pentium of same freq Sep 11 19:02:36 different arch Sep 11 19:02:45 but I still stand by my assertion: for almost any app other than graphics monsters, memory is *much* more important than cpu speed Sep 11 19:02:51 Kensan is right; I/O is a very significant part -- my main buidl machine is a 2.4GHz tower; it builds far faster than my 2.66GHz laptop -- both with OE, running Fedora 9 Sep 11 19:03:32 mwester:of course Sep 11 19:03:40 I/O is vital Sep 11 19:03:51 ruppena|home: If it's already in the kernel tree, just edit defconfig. If not, I steal the bitbake recipe from one of the other out-of-tree kernel modules and hack it until it works. No science,just hacking mainly. Sep 11 19:04:36 mwester, ok thx a lot, that will help me Sep 11 19:04:54 mwester: I saw you did various patches. What issues did you fix and did they make their way into the moko repos? Sep 11 19:05:49 All of the above said, it still must be observed that most systems spend the majority of their life in the idle loop -- in the end, does it make any real difference if we use a slower machine to do the job, as long as the build tools have enough memory to run at all? Sep 11 19:07:03 Kensan: the important bits are all in the stable kernel; there are a number of debug tools that are not, and probably never will be. There are also some "tweaks" that are not yet in; they wait for an image that runs well enough for people to begin to pay attention to some of the optimizations. Sep 11 19:07:53 Right now making it work is probably a priority, so I'll not bother Andy and the rest of the Om team with little optimizations and such. :) Sep 11 19:08:19 mwester: So suspend/resume is settled? Sep 11 19:08:22 mwester: what are the main kernel issues currently? Sep 11 19:09:00 mwester: sorry for asking all this, but I was not able to keep up with all the various MLs... Sep 11 19:11:12 i'm going back to frozen bubble Sep 11 19:11:57 (compiling for FR, not playing unfortunately) Sep 11 19:17:19 wurp2|working: No, suspend/resume is pending Andy's major re-org work in the newer kernel. Sep 11 19:17:34 great Sep 11 19:18:22 Kensan: The major issues vary depending on who you speak with -- for example, I don't care very much about wifi but I suspect some others may feel that the wifi driver should be a priority. Sep 11 19:19:08 Kensan: My list includes the White Screen Of Death problem, suspend/resume issues in general (but audio suspend/resume in particular). Sep 11 19:20:25 Kensan: And then we also have yet to implement a major bit of functionality, which is a light-weight periodic resume -- we need something that can wake every few minutes, resume up a few drivers -- just enough to check on the battery and the charger, etc -- and then suspend again if all is well. Sep 11 19:20:55 Kensan: Until we do that last bit, we'll continue to have the "battery-drained-so-flat-that-it-cannot-even-boot-to-start-charging" problem. Sep 11 19:21:30 Oh -- and the GTA01 UART overrun problem is still out there, and needs to be fixed -- but that's a massive effort. Sep 11 19:22:01 mwester: Personally, it still boggles my mind how a cell phone can rely on software to start charging. I'd think that's something to be done in hardware. Sep 11 19:22:08 mwester: but when the FR is drained so flat, they can use the uboot menu to start up, no? Sep 11 19:22:48 abraxa_: You are so right on that. We need that tiny micro that Om management just can't seem to accept. :( Sep 11 19:22:48 mwester: UART overrun? Do you have any pointers on that issue? Sep 11 19:23:26 Kensan: It's only on the GTA01; been discussed over and over on the mailing lists earlier this year. There's no one place that captures it all, I'm afraid. Sep 11 19:24:12 wurp2|working: I think there are still some open issues involving older u-boot versions where things can actually get into a state where the battery cannot be recovered. I'm not completely clear on that, though. Sep 11 19:24:13 mwester: ok. On the kernel ML? I have a GTA01 lying around which I would still like to get the most out of... Sep 11 19:25:12 Kensan: Ah, ok. Yes, on the kernel ML. Basically, when the interrupt latencies are high (such as during and immediately after resume), the serial interrupt is not serviced quickly enough and the FIFO on the UART will overflow, and it will drop characters from the GSM. Sep 11 19:25:31 The problem seems to be that the GSM does not honor hardware flow-control quickly enough. Sep 11 19:25:47 But there has been complete and total silence on that issue from Om. Sep 11 19:26:06 They will not comment on that bug (it's in the bug tracking tool). Sep 11 19:26:42 mwester: ah ic... Sep 11 19:27:05 The workaround is to hold the GSM flowcontrolled from suspend beyond completion of the resume -- and release it once things have "quieted down". Sep 11 19:27:22 Then the interrupts get serviced quickly enough and all is well. Sep 11 19:28:26 At least, as long as you do not use GPRS; there has been one report of a failure with GPRS that sounds to me like the GSM overflowed the buffer. That's why I think we still need to fix it in the kernel with a horrible workaround, but it's proven difficult to get working. Sep 11 19:29:46 mwester: hm... Sep 11 19:31:36 mwester: well I will try to read up on it on the ML. Sep 11 19:31:42 mwester: thanks for the info! Sep 11 19:41:14 mwester: I don't think we'll need that "periodic resume" feature to prevent draining the battery. The PMU can generate an interrupt when Vbat falls below a threshold, so the resume handler should be able to detect that and take the appropriate actions. Sep 11 19:42:14 mmontour: Not on the GTA01. And I seem to recall that there was some nuance of the feature you note, or some other use case, that would also require that ability on the GTA02. Sep 11 19:42:41 But implementing the battery-low feature would be a good start in any case. Sep 11 19:43:00 Agreed about GTA01 - I was only considering the FR. Sep 11 19:43:26 mwester: So it's prohibitively amp-hour expensive to wake up, estimate minimum sleep time until battery will be low, & set a timer to wake then? Sep 11 19:43:44 One use case for the periodic-resume on FR was to flash an LED to indicate a new SMS / missed call / etc. Sep 11 19:44:57 I wouldn't say so; I'd say that would be less-than-ideal -- it would certainly consume more power than necessary, but worse than that it would wake up the display, and restart all other user-space apps. It would be annoying to have one's pocket light up like a beacon every 5 minutes in the theater, for example. Sep 11 19:47:20 mwester: Sure, but I wouldn't expect it to be every 5 minutes Sep 11 19:48:19 I would think the desired behavior would be to estimate the shortest time when the battery power will fall below resume + 10 minutes of use, and wake then. If less than 30 minutes of use, shut down. If not, re-estimate & suspend again Sep 11 19:48:36 With the numbers of course being configuraable Sep 11 19:49:09 I don't know how to implement it, but I think we need a mechanism where the phone only resumes far enough to deal with the specific wake-reason and then suspends again. E.g. USB-insertion negotiates the correct 100/500/1000mA current limit only, without touching audio/gsm/backlight/etc. Sep 11 19:53:13 mmontour: Sure, but I'm more concerned with the good solution we could have now than the great solution we can have in 4 months Sep 11 20:01:47 good evening Sep 11 20:04:31 hi Sep 11 20:21:24 mwester: so the gsm cannot send an interrupt that could wake the app processor on event? Sep 11 20:21:48 Kensan: Yes, it does do that. Sep 11 20:23:26 mwester, Kensan, are you talking about future hardware or our current one? (just joined) Sep 11 20:23:47 quickdev: I was refering to the current gsm chip. Sep 11 20:24:17 mwester: ah I missread mmontour's message, never mind. Sep 11 20:42:35 AndreasD_, available? :) Sep 11 21:00:22 i'm trying to compile a kernel so i can test a few things related to my WLAN problems. i cloned the OM Git tree and checked out 'stable' and copied defconfig-gta02 to .config, but when i run "make vmlinuz" i get errors like ':warning: trying to assign nonexistent symbol HARDIRQS_SW_RESEND' and i'm asked for more options. did i miss a step? Sep 11 21:01:03 i'm no stranger to compiling kernels, but i guess the default config should work with the unmodified code... Sep 11 21:04:40 gismo: hi Sep 11 21:05:03 gismo: I’m preparing the milestone3 package of frameworkd now. Have you tried that out already? Sep 11 21:10:58 nomeata: I tried the kernel and posted on smartphone-kernel (it works, but I don't like it) Sep 11 21:11:08 nomeata: for the rest I'm still preparing the Italian DebConf Sep 11 21:11:20 ok Sep 11 21:11:22 nomeata: and I'll be busy until Monday, at least Sep 11 21:11:44 sounds good, I’ll be offline from saturday afternoon, so you can take up the bug reports then :-) Sep 11 21:12:12 nomeata: sure, no problem at all ;-) Sep 11 21:12:12 mwester: about the UART issue: it seems that a solution acceptable for all parties involved in the discussion is rewriting some parts of the s3c24xx serial driver. Sep 11 21:12:40 nomeata: BTW, I've some ideas/comments/suggestions I'll probably post next week... Sep 11 21:12:53 nomeata: FWIW I'm also flashing official m3! Sep 11 21:13:05 so am I at this very moment :-9 Sep 11 21:13:20 (I need to get a copy of the scenario files that we have added to the frameworkd package) Sep 11 21:13:52 nomeata: you don't need to flash it, you can greb them from the .tar.bz2 I think Sep 11 21:14:05 I know, but I want it flashed anyways Sep 11 21:14:18 nomeata: /me too :-D Sep 11 21:14:39 It doesn’t hurt to keep the various modules in /etc/modules, even though they are now linked statically, right? Sep 11 21:15:03 nomeata: not at all, the worst is an "echo" error Sep 11 21:15:37 nomeata: and if we list something in /etc/modules we know what's needed for what Sep 11 21:16:13 nomeata: I'm for releasing a new version of our install.sh script, it gets quite a lot of improvements Sep 11 21:16:32 gismo: yes, after I used it to install milestone3 I’ll do that Sep 11 21:16:46 nomeata: oh, good! Sep 11 21:17:01 Hi, can someone help with window system settings? I got an englightement taskbar, wanna get a rid of it. and I lost my icons on the top (battery, gsm status) Sep 11 21:17:27 2008.8 Sep 11 21:17:39 oh, and qpe takes 70-80% of CPU. Sep 11 21:17:54 (for some minutes after boot) Sep 11 21:17:57 euh, the moving ying-yang symbol when starting milestone3 is ugly! Sep 11 21:20:31 nomeata: hopefully, I'm not yet there... Sep 11 21:32:48 nomeata: does the SMS appcliation work for you? Sep 11 21:34:06 gismo: haven’t tried it yet Sep 11 21:34:15 I don’t have my SIM card in the FreeRunner ATM Sep 11 21:35:56 nomeata: ah, OK, it doesn't work here, let me reboot and see Sep 11 21:38:55 gismo: nomeata: hi, what's new on debian? ;) Sep 11 21:39:16 * nomeata is packaging milestone3 right now Sep 11 21:39:37 super nomeata! Sep 11 21:40:00 nomeata: nice Sep 11 21:40:08 i'm ready to test and report bugs :) Sep 11 21:40:17 nomeata: does it have any testsuite btw? Sep 11 21:40:36 lindi-: I'm preparing a talk for the Italian DebConf Sep 11 21:40:37 lindi-: testsuite, as in automatic tests? not that I know of Sep 11 21:40:49 lindi-: and then I'll be back, as usual :-D Sep 11 21:41:29 nomeata: ok Sep 11 21:42:34 i have so many things on my todo-list that it's hard to decide which one to work on Sep 11 21:43:15 lindi-: don't tell me that... Sep 11 21:47:22 lindi-: learn about "procrastination" => don't do any of your To-Do list and watch the consequences ;) Sep 11 21:53:28 http://wiki.debian.org/Teams/DebianFSO lists things like "Automatically detect a USB client device at the USB port, and switch to host mode." which are not debian-only in anyway. i have a long list of similar issues, should i add them to the wiki? ;) Sep 11 21:54:16 lindi-: feel free Sep 11 21:54:36 lindi-: there's also Enrico's list -> http://lists.linuxtogo.org/pipermail/smartphones-userland/2008-September/000012.html Sep 11 21:54:43 gismo: my vision is that we could have a really large public TODO list and then mark who is working on which entry Sep 11 21:54:51 Kensan: No, I think maybe you are looking at the wrong issues. The fact that OM is very unwilling to consider changes to non-OM drivers that cannot be pushed into the generic kernel is a small part of the problem. The real problem is that the GSM does not respond to hardware flowcontrol fast enough, so we need to re-write very significant parts of the UART driver to use DMA or FIQ interrupts. Sep 11 21:55:03 gismo: since i read from the list that somebody is working on getting zhone run as non-root but right now i can't remember who it was Sep 11 21:55:30 lindi-: nomeata Sep 11 21:55:41 gismo: thanks for that link Sep 11 21:55:49 lindi-: you're welcome Sep 11 21:55:56 a separate wiki page with a really really long TODO list will be nice :) Sep 11 21:56:07 lindi-: I really think that people should coordinate stuff on mailing lists Sep 11 21:56:09 at least it will remove duplication of work if interested parties can add their contact info there Sep 11 21:56:14 lindi-: again, feel free to add a link to a new page Sep 11 21:56:27 someone working with shr ? Sep 11 21:56:40 lindi-: what about a #openmoko-todo where all people involved are idling/working? Sep 11 21:56:47 lindi-: not quite serious.. just a random idea Sep 11 21:56:53 gismo: but the wiki would only document the current state, you'd use mailing lists of course for discussing the issues themselves Sep 11 21:56:56 lindi-: no, I’m not actually working on that, it’s just a todo :-) Sep 11 21:57:13 gismo: with mailing lists its hard to answer the question "who is working on getting zhone work as non-root" Sep 11 21:57:42 lindi-: well, I really hate wikis, so I'm the wrong person for this kind of stuff Sep 11 21:57:51 lindi-: as I'm hate blogs ;-) Sep 11 21:57:57 gismo: i don't like the slowness of firefox either :P Sep 11 21:58:18 gismo: how about a text file that only _I_ can edit? ;)) Sep 11 21:58:22 mwester: I was mostly referring to bug#788 and in particular the discussion on the ml: http://lists.openmoko.org/pipermail/openmoko-kernel/2008-January/000548.html Sep 11 21:58:51 lindi-: [OT] apt-get --purge autoremove iceweasel && apt-get install conkeror Sep 11 21:59:01 mwester: but I have to admit that I have not read "the whole thread", which I just found out goes on and one :) I will have to read up a bit more. Sep 11 21:59:23 OK, it seems I lost the SMS application on FSO-m3 :-( Sep 11 21:59:28 gismo: i have reported several bugs to them already :) Sep 11 22:00:25 Kensan: There are/were numerous GSM-related issues; they all got entangled on the mailing lists into one gigantic thread that went on and on and on. Sep 11 22:00:45 mwester: yes, I am still trying to dig through it. Sep 11 22:01:26 mwester: so one could say that basically the kernel driver needs to workaround issues of the GSM with regards to flow control. Sep 11 22:02:13 Kensan: There are still a bunch of them around. I have come to the conclusion that Harald understood the "big picture" of how it all was to work, but nobody else at Om ever troubled themselves to think about it -- and the end result was the mess you are reading; a vast conglomeration of GMS/UART problems with numerous people all solving symptoms but not the problem itself. Sep 11 22:03:36 I'm also beginning to wonder how many current Linux coders understand RS232, flow control, and such stuff. I think youth is part of the problem ;) Sep 11 22:03:59 mwester: hehe might be part of it :) Sep 11 22:04:29 mwester: RS232 is ancient ;) I have only once used it, for ancient hw :P but I suppose it's still standard for modems etc Sep 11 22:05:14 mwester: playing doom via serial-link was my first real multiplayer experience ;) Sep 11 22:05:39 Yep, it's ancient, and limited, and the assumptions it makes are horrible. Sep 11 22:06:09 mwester: is it really rs232 btw? Sep 11 22:06:46 No, RS232 has voltage considerations -- the GSM/UART interface is RS-232 but at 0v to 3.3v Sep 11 22:07:03 that's what i thought Sep 11 22:07:24 mwester: well I need to read more, I still don't see "the big picture"... not even the small one for that matter ;) Sep 11 22:08:07 Yeah, that's the same problem everyone else had -- everyone saw lots of little problems, and never saw the whole issue. Sep 11 22:08:22 Re bug 788 -- that was resolved. Sep 11 22:08:41 We got a patch that "worked around" the issue. Sep 11 22:08:48 For a while. Sep 11 22:08:53 Then another use case came up. Sep 11 22:09:05 And (surprise!!!) it locked up again. Sep 11 22:09:11 packaging milestone3 for Debian means installer.sh, fso-frameworkd and zhone, right? Or am I missing anything important? Sep 11 22:09:29 http://wiki.debian.org/DebianOnFreeRunnerTodo#preview Sep 11 22:09:30 says Sep 11 22:09:34 cannot concatenate 'str' and 'instance' objects Sep 11 22:09:42 it ate my todo list! ;) Sep 11 22:09:43 The final result is that the original fix, in modified form, ended up in the console code as it was, when all was said and done, the only way to solve the problem. Sep 11 22:09:58 Months of elapsed time, hours and hours of wasted effort Sep 11 22:09:59 mwester: but the issue is restricted to GTA01 since FR does not share UART between console and gsm, is that correct? Sep 11 22:10:10 Only to end up at the same solution I had originally. Sep 11 22:10:56 Kensan: that depends on what GSM issues you specifically refer to. Sep 11 22:11:14 The general ones that apply to the GTA02 as well are resolved. Sep 11 22:11:33 The ones referenced in 788 are GTA01 specific because of the shared UART. Sep 11 22:11:34 nomeata: I think you missed nothing Sep 11 22:11:50 gismo: ok, thx Sep 11 22:11:56 nomeata: you're welcome Sep 11 22:11:59 mwester: ah I see. Sep 11 22:12:01 (gsm0710muxd had no changes) Sep 11 22:12:39 The open issue with UART overruns on the GTA01 are a hardware issue with the GSM, and have nothing to do with the shared console. I theorize that the problem does not appear ont he GTA02 because the SoC on the GTA02 is different, and the UART ont hat device has a 64byte FIFO instead of only 16 bytes. Sep 11 22:13:26 mwester: interesting. Sep 11 22:13:43 :) tiresome at this point... Sep 11 22:14:14 I wish OM would respond to https://docs.openmoko.org/trac/ticket/1376 Sep 11 22:14:15 mwester: you seem to have a good overview over everything Sep 11 22:14:22 mwester: I can imagine. Well I just discovered this lovely thread on the ML so this is all interesting to me at this point :) Sep 11 22:14:38 edistar: no, not really -- only on the GSM/UART comms. :( Sep 11 22:17:20 mwester: well I might be back with some more tiresome questions, but I should read the postings on the ML first. Sep 11 22:19:28 mwester: thanks for the information anyway! good luck with further investigations, you are one of the few community guys really doing a ton of work! (you don't work for OM, do you?) Sep 11 22:20:26 nomeata: have you already started reinstalling? Sep 11 22:20:47 nomeata: I'm pushing the bluetooth modules patch and also a small fix for the help output Sep 11 22:21:54 gismo: no, not yet. I’m testing the packages first directly, before I’ll upload them and test them with the installer Sep 11 22:22:28 nomeata: OK, perfect, thx! Sep 11 22:22:54 nomeata: done Sep 11 22:23:17 mwester: I concurr with edistar. Keep up the awesome work! Sep 11 22:26:37 gismo: ok, at least without a SIM card the new packages seem to work nicely, and nothing complained yet that I didn’t upgrade the binary glue packages. But then, python has the tendency to complain at the very latest possible point of time :-9 Sep 11 22:26:48 So I’ll upload the packages to pkg-fso, and do an installation test Sep 11 22:28:55 nomeata: +1! Sep 11 22:36:12 ok, packages are uploaded. time for an installation test Sep 11 22:38:52 gismo: it's now http://wiki.debian.org/DebianOnFreeRunnerTodo Sep 11 22:39:18 lindi-: better to use http://wiki.debian.org/DebianOnFreeRunner/ToDo, but leave it like this for the moment Sep 11 22:39:51 gismo: ah, good point Sep 11 22:39:52 lindi-: BTW, could you post that link in reply to Enrico's post? Sep 11 22:39:58 gismo: i already did :) Sep 11 22:40:36 lindi-: ah, OK Sep 11 22:40:54 lindi-: you didn’t quite hit the wiki’s syntax for lists Sep 11 22:41:45 nomeata: i'm working on that :) Sep 11 22:43:41 nomeata: better now? Sep 11 22:44:11 gismo: also maybe we should talk about DebianOnNeo instead of freerunner since it works on both neo1973 and freerunner? Sep 11 22:44:56 lindi-: that's another point for the future Sep 11 22:45:52 lindi-: I mean, it's on my ToDo list, too big to be pasted somewhere ;-) Sep 11 22:46:18 * nomeata runs the installation. Sep 11 22:46:56 nomeata: what are we missing the most WRT the FSO image? Illume, right? Sep 11 22:47:20 gismo: right. Sep 11 22:47:22 gismo: haha :) Sep 11 22:49:10 nomeata: I'm upgrading... Sep 11 22:51:05 i thought i could look into packaking uboot, would that make much sense? i figured it should be pretty simple but i wonder why it is not in debian yet Sep 11 22:51:29 lindi-: and then install it from the running system to NAND? Sep 11 22:51:47 nomeata: not sure how safe that is Sep 11 22:51:53 lindi-: it should work, but then also the installer.sh’s uboot stage should work Sep 11 22:52:07 lindi-: so first step is to make sure that works Sep 11 22:52:40 lindi-: I had strange problems with that, but actually, only once. Maybe someone else should just test what I’ve done there, maybe it works after all or needs just small tweaks. Sep 11 22:52:46 nomeata: my main interest is in the fact that i'd like to study u-boot sources and see if it could add battery indicator there easily :) Sep 11 22:54:15 nomeata: of course the package could include the u-boot binary and ask the user to use dfu-util to actually install it? Sep 11 22:54:46 lindi-: it should be like that, but we need to properly generate the uboot binary Sep 11 22:55:41 yes, so i need to figure out what version of uboot is used and what patches have been added to it Sep 11 22:55:49 lindi-: exactly Sep 11 22:56:08 lindi-: http://git.openmoko.org/?p=u-boot.git;a=summary Sep 11 22:58:06 is 2008-04-07 really the latest change? Sep 11 22:58:30 hmm no Sep 11 23:01:47 nomeata: is it normal that when you upload a new version to pkg-fso the old one is deleted? Sep 11 23:02:15 gismo: yes Sep 11 23:02:16 nomeata: FWIW m3 on Debian doesn't have access to contacts nor SMSs (after a reboot) Sep 11 23:02:30 nomeata: about pkg-fso, I think we should keep at least one version before Sep 11 23:02:32 that's a bug i hipe? ;) Sep 11 23:02:46 gismo: but I think this can be changed - ask phil about it Sep 11 23:03:08 nomeata: OK, I'll send a mail tomorrow night or during the w/e Sep 11 23:06:07 gismo: actually since install.sh does not download u-boot we could all be using different u-boot version Sep 11 23:07:51 lindi-: exactly, but if you package it, we can install that specific version Sep 11 23:08:04 Fun: Running install.sh on the neo and apt-get upgrade on the desktop in two xterms, and suddenlty wondering why there is openoffice installed on the phone. Sep 11 23:08:43 nomeata: does the touchscreen still have space for writing? Sep 11 23:08:47 lindi-: it's like the kernel: having a specific version helps the developers fixing bugs Sep 11 23:09:13 lindi-: I think that some packages should be "frozen", especially since the device is very specific Sep 11 23:09:27 gismo: yes Sep 11 23:15:36 lindi-: http://git.openmoko.org/?p=u-boot.git;a=shortlog;h=stable is the current u-boot Sep 11 23:17:37 gismo: upgrade went fine? Sep 11 23:18:13 * nomeata still waits for the installation to finish Sep 11 23:19:45 nomeata: [01:02:15] nomeata: FWIW m3 on Debian doesn't have access to contacts nor SMSs (after a reboot) Sep 11 23:19:56 nomeata: that's why I asked for the old version ;-) Sep 11 23:20:13 nomeata: but it seems the leds work (at least with the USB cable in the PWR one is blue) Sep 11 23:21:31 can someone run "opkg search /bin/frameworkd" ? thx I can't find anymore it Sep 11 23:22:42 we ought to make sure that the ring tones we ship can be run with only one set of gstreamer plugins (probably “good”), to get rid of -bad and -ugly and their dependencies. Sep 11 23:23:07 mmontour: i read the git cheatsheet but i'm not quite sure how i should get the sources for that. 'git clone git://git.openmoko.org/u-boot' fetched me source code but 'git branch stable' says 'error: pathspec 'stable' did not match any.' Sep 11 23:23:13 Hi everyone, Sep 11 23:23:13 (.. when we replace the existing ones with free ones and maybe put them in their own package) Sep 11 23:23:36 just a quick question: where can I put a scriptfile I want to link to in the wiki? Sep 11 23:23:46 mmontour: i mean 'git checkout stable' Sep 11 23:23:49 nomeata: I vote for Enrico's ringtone at http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=495668#33 Sep 11 23:24:23 ah, haven’t seen that Sep 11 23:24:23 hmm, odd, i don't recall getting mail about that ringtone Sep 11 23:24:24 nomeata: and, really, I find quite strange to have a .sid/.mp3 ringtone for a F/LOSS phone Sep 11 23:25:04 lindi-: do you prefer the s-userland link? http://lists.linuxtogo.org/pipermail/smartphones-userland/2008-September/000014.html :-D Sep 11 23:25:06 I heave heard that .sid is much easier on the CPU than ogg Sep 11 23:25:13 lindi-: try "git checkout origin/stable" (may need a "-b" in there too) Sep 11 23:25:21 * mmontour is also a git n00b Sep 11 23:25:24 gismo: but i was submitter of that bug Sep 11 23:25:30 gismo: so i thought i should have got email :) Sep 11 23:25:40 lindi-: no, you need to subscribe, too Sep 11 23:26:22 gismo: hmm, is this special for the pkg-fso packages? i usually get email from bugs i have submitted Sep 11 23:26:27 nomeata: I don't know anything about CPU usage for .sid or .ogg, to me it seems a phylosophy question Sep 11 23:26:42 lindi-: no, it's general: when you receive mails people directly write you Sep 11 23:26:46 mmontour: 'git checkout: to checkout the requested commit you need to specify a name for a new branch which is created and switched to' Sep 11 23:27:02 gismo: ah Sep 11 23:27:03 lindi-: it's how the Debian BTS is done (and there's sometime some discussion to change it) Sep 11 23:27:05 gismo: yes, I’m just quoting their reason Sep 11 23:27:54 mmontour: 'git checkout -b origin/stable' does not print anything and exits with 0 Sep 11 23:29:15 "git checkout origin/stable" should work now Sep 11 23:29:19 nice, installation went right through. time to reboot Sep 11 23:30:09 mmontour: it seems to now, maybe i did a typo Sep 11 23:31:12 I repeat my question: Is there a place an on-and-off developer like me can stuff files to link them in the wiki? A simple "No" will do if there is not Sep 11 23:31:13 I think you need to do the "-b" first to create a local branch, then another checkout without the "-b". But I don't really understand what's going on inside of git when you do it. Sep 11 23:32:35 mmontour: are you sure that is the current uboot? grepping for neo1973 finds only defines but the uboot command line definitely recognizes commands prefixed with neo1973 Sep 11 23:33:13 'grep -r vibrator .' does not find anything from the sources either Sep 11 23:33:32 install.sh pushed to http://pkg-fso.alioth.debian.org/freerunner/ Sep 11 23:34:47 I'm sure that http://git.openmoko.org/?p=u-boot.git;a=shortlog;h=stable is correct, but I'm less certain about the local git commands to make sure that you're building from that tree Sep 11 23:36:13 mmontour: ok, and if i have not switched to the right branch then grep won't find the neo specific uboot commands either? Sep 11 23:36:19 mmontour: you can do `git checkout -b $LOCAL $REMOTE` to just checkout $REMOTE into a newly-created $LOCAL Sep 11 23:36:38 mmontour: since my u-boot has all kinds of funny neo specific commands (just discovered them today) Sep 11 23:37:20 gismo: ok, thanks. Sep 11 23:37:37 * mmontour wonders if anyone sells a git-cheatsheet coffee mug, like that vi one Sep 11 23:39:00 i think i'm going to fall asleep soon, thanks for all your help! Sep 11 23:40:46 gismo: I’ll send a small mail around and then go to bed Sep 11 23:41:06 gismo: anything important to mention therein? Sep 11 23:41:16 (about milestone3 available in Debian) Sep 11 23:42:44 nomeata: I don't think so, it's just that we missed contacts and SMS ;-) Sep 11 23:43:32 it's funny: tomorrow I've a 900km motorbike ride and it's *heavy* raining here in Geneva Sep 11 23:43:33 gismo: but that’s also with FSO, isn’t it? Sep 11 23:43:54 Or did I mess up and should fix that first? Sep 11 23:44:26 nomeata: the SMS with FSO, yes, but contact no (I'm rebooting to be sure) Sep 11 23:48:17 nomeata: OK, Debian is just missing contacts, SMS is not working with upstream FSO Sep 11 23:49:11 gismo: hmm. I wonder where such an regression compared to FSO comes from Sep 11 23:49:48 can someone help me getting debian to work again ? Sep 11 23:50:13 I had it running, then the sd card crashed, not apt-get segfaults Sep 11 23:50:22 now apt-get segfaults... Sep 11 23:50:37 I removed the *bin files, and others, no clue... Sep 11 23:50:42 neolynx: has your / been checked at boot? Sep 11 23:51:05 yes, fsck said it's clean Sep 11 23:51:13 I reinstalled apt Sep 11 23:51:17 installed strace Sep 11 23:51:34 I'm a bit out of ideas :/ Sep 11 23:51:49 nomeata: I don't know and the three patches you left out don't seem to be involved in SMS handling Sep 11 23:52:12 gismo: well, I’ll have a look later, it’s too late now. Sep 11 23:52:22 nomeata: BTW, it's not that I don't see SMSs (I'm sure I've), but the application is not available at all Sep 11 23:52:30 nomeata: no problem, rest well Sep 11 23:52:39 morning Sep 12 00:25:17 wgetting apt twice gives me: Sep 12 00:25:26 cfddae2b5d85019e787387ba4e71ecde apt_0.7.14_armel.deb Sep 12 00:25:26 1d41cc36a379b5c1d82b99bff0b94aa3 apt_0.7.14_armel.deb.1 Sep 12 00:25:51 neolynx: are you sure your mirror is OK? Sep 12 00:26:21 well, the *deb had the same sum in the beginning Sep 12 00:26:44 I suspect my 8GB sd card is broken Sep 12 00:26:59 or the kernel does not handle it right Sep 12 00:27:19 I read something about downclocking, but did not find something concrete Sep 12 00:27:32 neolynx: maybe you're suffering from http://docs.openmoko.org/trac/ticket/1743 Sep 12 00:27:37 neolynx: s/from/of/ Sep 12 00:28:34 well, the thing is, dmesg shows no errors at all Sep 12 00:28:39 everything seems find Sep 12 00:28:40 fine Sep 12 00:28:49 but my files are going corrupt all the time Sep 12 00:29:20 neolynx: I've never experienced it, so I'm not of big help here Sep 12 00:29:52 no problem :) thanks anyway Sep 12 00:30:27 debian without apt is just no fun :-( Sep 12 00:34:04 neolynx: I'd suggest to try with the 512MB included in the box Sep 12 00:34:30 gismo: good idea Sep 12 00:34:53 is there a howto ? or images to download ? Sep 12 00:46:25 ha ! Sep 12 00:46:34 reinstalled libstdc++6_4.3.2-1_armel.deb and apt works again Sep 12 00:46:44 Anyone know what the plan for wlan support in FSO is? Sep 12 00:46:53 make it work? Sep 12 00:46:59 :D Sep 12 00:47:11 I mean more like, using networkmanager, using hand-crafted Sep 12 00:47:16 python, etc etc Sep 12 00:47:56 Ah, ok. I'm not a fan of networkmanager, personally, but I somethnig of that sort would be nice (assuming the driver worked right0 Sep 12 00:48:36 'night all Sep 12 00:51:18 the nice thing about networkmanager, I guess, would be that it already has a dbus interface and python frontends. Sep 12 00:52:42 * torindel thinks we should use something more lightweight ;] Sep 12 00:53:37 I don't think such a thing exists Sep 12 00:53:54 and writing wep/wpa/wpa+/wpa2 handling code from scratch isn't my kind of fun Sep 12 00:53:55 we could make fso speak directly do wpa_supplicant (it have controll socket) Sep 12 00:54:11 oh, that's true Sep 12 00:54:18 ssh to your phone, configure the networks you want, wpa_suplicant auto recognition <- lightweight Sep 12 00:54:20 networkmenager doesnt do wpa directly Sep 12 00:54:41 it uses wpa_supplicant too Sep 12 00:55:43 anyway i think fso should talk to wpa_supplicant Sep 12 00:56:49 http://xkcd.com/416/ Sep 12 01:00:50 mwester: :) Sep 12 01:00:55 n8 Sep 12 01:02:23 odd Sep 12 01:02:32 the control panel for FSO says that suspend is turned on a second after blanking Sep 12 01:02:38 but it's pretty clearly not suspending Sep 12 01:03:09 (for illume in FSO3, I mean) **** ENDING LOGGING AT Fri Sep 12 02:59:57 2008