**** BEGIN LOGGING AT Sun Sep 09 02:59:56 2007 Sep 09 03:00:39 oh hooray Sep 09 03:00:40 c++ Sep 09 03:00:49 * raster loves c++... NOT Sep 09 03:00:49 Hey, at least it's not Java. Sep 09 03:01:32 aahha... Sep 09 03:02:43 true Sep 09 03:02:53 come on wiki Sep 09 03:02:56 mail me damnit! Sep 09 03:05:03 * ljp hands raster an herb tea and takes away his jolt cola Sep 09 03:05:39 don't disturb my zen of instantness Sep 09 03:08:41 well, if you believe that all time exists now and we are merely moving through it, everything is instant Sep 09 03:10:12 raster neat - evas/eet/ecore all build under om - just installed them - trying to find a test proggy.. Sep 09 03:12:47 ljp: i believe in instant noodles, meat pies and instant beer. Sep 09 03:12:53 aking: they should build Sep 09 03:13:04 ok Sep 09 03:13:07 added to wiki Sep 09 03:13:20 now others can fix their gettext emacs goop Sep 09 03:13:43 raster: emotion seg faults during build though Sep 09 03:15:01 aking: segv's? Sep 09 03:15:02 wow! Sep 09 03:15:05 during build! Sep 09 03:15:11 what segv's? Sep 09 03:15:22 edje_cc? Sep 09 03:16:28 raster: yup - make[3]: *** [theme.edj] Segmentation fault Sep 09 03:16:39 hmmm Sep 09 03:16:43 using upstream? Sep 09 03:16:46 cvs? Sep 09 03:16:48 or a snapshot? Sep 09 03:17:02 emotion-0.0.1+cvs20070819 Sep 09 03:17:07 i hada problem a while back i fixed in scratchbox Sep 09 03:17:11 no not emotion Sep 09 03:17:12 edje Sep 09 03:17:23 and eet Sep 09 03:17:24 etc. Sep 09 03:17:26 anyuway Sep 09 03:17:31 edje_cc liked to use a fair bit of stack Sep 09 03:17:34 a few hundred kb Sep 09 03:17:38 and qemu hated it Sep 09 03:17:46 edje-0.5.0+cvs20070819 Sep 09 03:17:48 when running arm binaries under aemu emulation Sep 09 03:17:53 and it blew the stack Sep 09 03:17:55 hmm Sep 09 03:18:00 thats not including recent changes Sep 09 03:18:04 ahh Sep 09 03:18:05 it probably has my stack fixes tho Sep 09 03:18:28 strange, just an hour ago, the build completed succesfully. Now it's failing Sep 09 03:18:40 but OE builds a native x86 edje_cc Sep 09 03:18:45 thus that should just work (tm) Sep 09 03:18:55 unless... Sep 09 03:18:58 the native build is broken Sep 09 03:19:06 and doesnt install buffer engines etc. properly Sep 09 03:19:09 yup.. it should.. I'll check Sep 09 03:19:14 amd64 Sep 09 03:19:24 should also be 64bit clean Sep 09 03:20:14 it builds and works fine on 32bit x86 Sep 09 03:20:24 i dont bother with 64bit os installs anymore Sep 09 03:20:27 too much pain Sep 09 03:20:46 (mostly java, flash, w32codecs, etc. too much pain) Sep 09 03:21:57 exquisite will want fb engine support Sep 09 03:22:04 if u want it on your om as a splash screen replacement Sep 09 03:22:16 and that means fb evas engine and ecore fb support Sep 09 03:22:23 unsure of the oe efl packages build that Sep 09 03:26:08 aking: are you talking about openembedded? Sep 09 03:26:48 thomasg: yup Sep 09 03:26:58 same problems here Sep 09 03:27:14 seg fault with edje_cc? Sep 09 03:27:17 do you get some pango errors in your log (or can you build pango at all?)? Sep 09 03:28:00 I assume we're talking about openembedded problems, as I don't have problems under my native gentoo system at all Sep 09 03:28:08 oh hooray Sep 09 03:28:10 glibc now! Sep 09 03:28:12 libnss_compat_pic.a'. Stop. Sep 09 03:28:12 | make[3]: *** Waiting for unfinished jobs.... Sep 09 03:28:12 | make[3]: *** write jobserver: Bad file descriptor. Stop. Sep 09 03:28:12 | make[2]: *** [nis/others] Error 2 Sep 09 03:28:12 | make[2]: Leaving directory `/home/raster/moko/build/tmp/work/armv4t-angstrom-linux-gnueabi/glibc-intermediate-2.5-r7/glibc-2.5' Sep 09 03:28:14 | make[1]: *** [all] Error 2 Sep 09 03:28:17 hehe :) Sep 09 03:29:38 thomasg:yup - native oe build of edje_cc seg faults - lookin ginto it now.. Sep 09 03:30:29 i suspect its a missing engine thing Sep 09 03:30:31 i suspect Sep 09 03:30:49 edje_cc isnt crash hot on checking for that Sep 09 03:30:51 i should fix that Sep 09 03:31:03 heh Sep 09 03:31:19 actually Sep 09 03:31:23 it is good about it Sep 09 03:31:26 it will complain properly Sep 09 03:31:41 fprintf(stderr, "Error. cannot create buffer engine canvas for image load.\n"); Sep 09 03:35:47 raster: here's the output from the seg fault (run manually): http://pastebin.ca/688007 Sep 09 03:35:50 ah, I'm in #openmoko here. was confused as I was assuming I'm in #e :)_ Sep 09 03:37:46 here's my errorlog; http://rafb.net/p/bglmcn13.html Sep 09 03:41:13 aking: doesnt help much Sep 09 03:41:22 died after inlining font info Sep 09 03:41:25 before anythign else Sep 09 03:41:32 i guess it was goign to inline images next Sep 09 03:41:55 ok - I'll try replacing with my other build.. Sep 09 03:42:05 yup Sep 09 03:42:07 images was next Sep 09 03:42:17 that implies buffer engine was going to be involved Sep 09 03:42:33 (edje_cc creates an evas canvas but in-memory only and has evas do the image loaidng and heavy lifting) Sep 09 03:42:40 but edje checks return values Sep 09 03:42:44 from the wrapeprs Sep 09 03:42:48 so its deeper down Sep 09 03:42:52 inside the engine loading code Sep 09 03:42:58 or ecore_evas itself Sep 09 03:43:23 raster: sounds like an ugly one to fix - I'll try a few things here. Sep 09 03:43:37 you will need to basically do this under gdb or valgrind Sep 09 03:43:46 get yourself a shell with all the om environment set up Sep 09 03:43:51 and run the build line by hand under valgrind Sep 09 03:43:55 as its a native exec Sep 09 03:43:58 that should work Sep 09 03:45:52 raster: 'k - if I have time next weekend, I'll try that (I've used gdb many times... ) but I'd rather carry on playing with picogl/neo this weekend Sep 09 03:46:07 ok Sep 09 03:46:08 cool Sep 09 04:02:21 i give up Sep 09 04:02:26 glibc is not screwed Sep 09 04:02:32 got better things to do Sep 09 04:10:30 raster: copying over my gentoo build of edje_cc into /home/moko/build/tmp/staging/x86_64-linux/bin/ works - about to try epdf.. Sep 09 04:26:21 (script) wiki RecentChanges: [[Development_resources/es]] [[Talk:MokoMakefile]] [[User:CesarB/Requirements]] Sep 09 04:27:09 aking: grrr Sep 09 04:27:43 raster: grrr.. right back at you.. so many missing dependencies... trying to build d them.. Sep 09 04:27:59 :( Sep 09 05:58:04 moin Sep 09 06:02:16 damn css html and stuff Sep 09 06:27:23 (script) wiki RecentChanges: [[Neo1973_Battery_Charger]] [[Talk:Neo1973:_GTA01Bv4_versus_GTA02_comparison]] [[Today/2007.2]] [[Neo1973_Power_button/es]] [[OpenMoko_compatible_cellphone_providers/es]] [[Main_Page/es]] Sep 09 06:47:21 (script) openmoko-community: Sean Moss-Pultz Re: speeding up the flow of information Sep 09 06:48:59 what's a good and safe version of u-boot? Sep 09 06:49:33 the 2007.08 snapshot Sep 09 06:49:42 i think i still have the version the phone shipped with so i should probably update Sep 09 06:49:54 dunno if it has the rootfs erase fix in it though Sep 09 06:50:33 there's also some other weird stuff like bootdelay doesn't seem to work Sep 09 06:51:35 i'm i right in expecting that it should give me more time to press aux to get to the bootloader menu? Sep 09 06:52:01 you hold aux in while pressing power Sep 09 06:52:40 "by changing bootdelay you can have more time to take control of u-boot" says http://wiki.openmoko.org/wiki/U-boot#Timeouts Sep 09 06:52:45 bootdelay is only about getting in with a terminal on ttyACM0 if you're not physically at the device Sep 09 06:53:02 oh ok, i'll clarify that writing then Sep 09 06:53:37 is there any way to get to the menu on every boot? Sep 09 06:54:05 dunno Sep 09 06:54:38 ok. thanks for the information! Sep 09 06:56:52 np Sep 09 06:57:59 are you use libglade/ Sep 09 06:58:10 do you use libglade? Sep 09 07:04:41 damn, now i'd want to do some u-boot hacking %-) Sep 09 08:27:24 (script) wiki RecentChanges: [[Bootloader]] [[User:Techiem2]] [[OpenEmbedded]] Sep 09 08:53:19 good morning :) Sep 09 09:00:07 will i be able to deploy multiple cores when using bitbake? Sep 09 09:00:35 i mean, compile something in parallel Sep 09 09:03:41 tuukkah: yes, see PARALLEL_MAKE option for local.conf Sep 09 09:37:21 pH5, is it known to not break openmoko builds? Sep 09 09:42:04 tuukkah i use it, but its not the reason for broken builds (anymore) Sep 09 09:43:16 hmm, are you saying there's no way to get a proper build at the moment? Sep 09 09:43:46 i mean, why the "but" Sep 09 09:43:47 my build just failed on task do_rootfs. Dependency on openmoko-feedreader2 was not satisfied. Sep 09 09:43:54 there is.. but you have to hit the right moment Sep 09 09:44:07 roh :-) Sep 09 09:44:55 roh: we should develop an sms service, when we get the sms we all start our builds :-) Sep 09 09:48:37 isn't version control supposed to help get rid of timing problems - are some changes independent of mtn update? Sep 09 09:51:25 hi, what are the most recent rootfs & kernel image versions I *should* use? Sep 09 09:51:47 * duffyd just about to copy my first rootfs & kernel images onto my newly acquired neo :-) Sep 09 09:52:12 duffyd: According to buidhost, the ones from 3 September are ok Sep 09 09:52:24 joao: thanks Sep 09 09:53:39 joao: mmm, you got a url for that? Sep 09 09:53:49 currently looking at http://buildhost.openmoko.org/OM2007.2/tmp/deploy/glibc/images/fic-gta01/ Sep 09 09:54:46 duffyd: http://buildhost.automated.it/OM2007.2/?M=D and then sort by last modified. I hava a scaredycat one Sep 09 09:56:39 which bitbake targets are interesting? Sep 09 09:57:39 openmoko-devel-image uboot-openmoko openocd-native and dfu-util-native Sep 09 09:58:11 do we have those explained somewhere in the wiki? Sep 09 09:58:45 learning by doing? Sep 09 09:58:54 openmoko-devel-image = kernel & rootfs Sep 09 09:58:57 uboot-openmoko = bootloader Sep 09 09:59:09 openocd-native = on chip debugger /debugboard needed Sep 09 09:59:17 dfu-util-native = device firmware update tool Sep 09 09:59:54 ixs, that much i can guess :-) it would still make sense to me to have them listed somewhere Sep 09 10:02:07 uhm.. it seems the wiki is out of date for manual oe/bb instructions and OM2007.2 Sep 09 10:02:35 but https://wiki.openmoko.org/wiki/MokoMakefile and a walk through the makefile is a lot of explaining Sep 09 10:03:34 roh: as far as I can tell, the mokomakefile is now building 2007.02 so the instructions on the wiki sould work Sep 09 10:04:58 i'm following manual build instructions at http://wiki.openmoko.org/wiki/OpenMoko2007.2 Sep 09 10:07:05 Yeah, OPENMOKO_GNERATION = 20007.2 Sep 09 10:10:27 joao japp.. i meant the manual ones.. Sep 09 10:10:36 japp.. https://wiki.openmoko.org/wiki/OpenMoko2007.2 has the correct ones Sep 09 10:27:24 (script) openmoko-neo1973: Harald Welte Re: Manual for the SC32442 and 1 other mails Sep 09 10:27:24 (script) wiki RecentChanges: [[OpenMoko_Local_Groups:_Trondheim]] [[Neo1973_Battery_Charger]] [[Manually_using_Bluetooth]] [[WiFi_support_in_OpenMoko]] [[GSM]] [[Bootloader]] [[Talk:OpenMoko_Local_Groups:_Norway]] [[Talk:OpenMoko_Local_Groups:_Bucharest]] Sep 09 10:28:54 how to input a number in GB2312 coding but ascii? Sep 09 10:29:57 readon, do you mean iconv -f ascii -t GB2312 Sep 09 10:30:27 seems to be a no-op Sep 09 10:32:40 numbers seems to be ignored Sep 09 10:32:43 by iconv Sep 09 10:34:43 readon, based on wikipedia GB2312 is not actually an encoding but is used as the character set of the EUC-CN and HZ encodings Sep 09 10:39:50 GB2312 is listed in iconv -l Sep 09 10:41:06 perhaps it's a synonym for EUC-CN Sep 09 10:42:33 then again this HZ thing isn't there at all Sep 09 11:00:19 there a known somewhat working semi-recent image combo? Sep 09 11:00:54 not that I know of Sep 09 11:01:46 blindly installed uImage-2.6.22.5-moko11-r2-fic-gta01.bin and OpenMoko-openmoko-devel-image-glibc-ipk-P1-September-Snapshot-20070908-fic-gta01.rootfs.jffs2, gsmd either doesn't work or crashes Neo on startup :] Sep 09 11:02:16 mjr, see bug #788 for a workaround to the crash Sep 09 11:02:49 ack Sep 09 11:12:07 Hello, everyon Sep 09 11:14:42 Hi pavelm! :) Sep 09 11:14:53 Hi, Elrond. Sep 09 11:15:09 SO I tried 2007.2, and it is really better than the older versions. Sep 09 11:15:29 ...but I wonder if I have to work on chroot for gllin now...? Sep 09 11:15:34 ...or did someone already played with that? Sep 09 11:15:35 Except gsmd crashing, I hear. ;) Sep 09 11:16:01 pavelm - chroot for gllin was on my list. But I did not do anything on that. Sep 09 11:16:06 well, I actually have some newer version from scaredy cat... and that one has an older version of gsmd... and that one works. Sep 09 11:16:29 Ahh. Sep 09 11:16:58 ...unfortunately that means no sms support :-( Sep 09 11:17:15 I have decided to wait for the official images to work much better. Sep 09 11:18:24 t68i with broken keyboard is a phone I'm trying to replace. I'm afraid I don't want to wait. Sep 09 11:19:51 My me45 still works nicely. And Bv03 isn't really an option for replacing, you know. :) Sep 09 11:21:39 So now we need a way to modify the apex environment from userspace to fix the gsm powerdown bug Sep 09 11:22:22 rwhitby, where can i read about that? Sep 09 11:23:52 http://bugzilla.openmoko.org/cgi-bin/bugzilla/show_bug.cgi?id=788#c10 Sep 09 11:24:14 anyone got any good uboot menu items I can add while I'm here setenv'ing ? Sep 09 11:24:15 it doesn't mention apex Sep 09 11:24:24 oops - s/apex/uboot/ Sep 09 11:25:11 rwhitby - Increase the bootdelay, so you don't have to always press aux to do anything in u-boot. :) Sep 09 11:25:48 rwhitby - Was the first thing I did, so after a reboot, I just could picocom into u-boot without touching the phone. Sep 09 11:26:55 rwhitby, do you already have "Boot from SD"? Sep 09 11:27:08 tuukkah, righto, I can make calls now. Yay. Sep 09 11:28:23 mjr: Congrats. Does it also ring on incoming call? Sep 09 11:29:06 * rwhitby ipkg upgrades from 0829 to 0908 from buildhost, with "console=ttySAC0,115200" removed from uboot bootargs_base ... Sep 09 11:31:33 no Sep 09 11:42:37 mjr: I did shell script for ringing.. so I was interested if it worked in "mainline" already Sep 09 11:43:44 * rwhitby sees a keyboard icon in the panel first the very first time, instead of a white rectangle ... Sep 09 11:46:20 hmm - GSM_RES missing from /etc/init.d/gsmd Sep 09 11:46:47 I simply can't believe that the person working on gsmd is actually testing anything Sep 09 11:46:53 or /etc/default/gsmd Sep 09 11:47:08 right - /etc/default/gsmd is what I meant Sep 09 11:47:56 rwhitby - I wonder the same about other parts. Sep 09 11:48:02 well in the init script, [ -n $GSM_POW ] etc. doesn't make sense at all. should be [ -n "$GSM_POW" ] Sep 09 11:48:38 ok, these files are in OE, so I can fix them Sep 09 11:48:58 nice Sep 09 11:49:10 lay the fixes on me, baby. Sep 09 11:49:29 what should GSM_RES be set to? Sep 09 11:49:58 my guess would be /sys/bus/platform/devices/gta01-pm-gsm.0/reset Sep 09 11:50:07 yep, that's what I came up with too Sep 09 11:51:05 Hi zecke Sep 09 11:54:13 let's reboot to test that fix before committing it ... Sep 09 11:54:28 rwhitby, what a foreign idea ;-) Sep 09 11:54:54 alien ideas? ;o) Sep 09 11:56:07 Ok, I got gllin to work in chroot. Sep 09 11:56:24 pavelm - Cool! :) Sep 09 11:56:45 pavelm - A chroot in tmpfs? Sep 09 11:57:14 I guess I'll just do a blog posting. Sep 09 11:57:31 No, I just created /chroot. I don't think theres any advantage to tmpfs...? Sep 09 11:59:26 pavelm - There is: Those crappy NVRAM* files wont cripple the flash. :) Sep 09 12:00:05 pavelm - Just replace /chroot by /var/run/gllin-chroot :) Sep 09 12:00:50 * rwhitby has pushed the gsmd init script fixes Sep 09 12:02:00 pavelm - Instead of blogging, put a HOWTO on the wiki and blog the wiki location. :) Sep 09 12:02:03 elrond: :-). Well, I guess the idea is that gllin _needs_ persistent data. Sep 09 12:02:13 elrond: Right. GOod idea. Sep 09 12:02:23 /sys/bus/platform/devices/gta01-pm-gsm.0/reset does nothing on gta01bv04 Sep 09 12:02:46 what about gta01bv3? Sep 09 12:02:50 pavelm - That's the idea. Just it has shown for me, that the NVRAM* makes it worse. It couldn't get any sat with outdated NVRAM*. Sep 09 12:03:08 rwhitby - What's with the bv03? Sep 09 12:03:15 Ahh. Sep 09 12:03:36 elrond: Well, we may want to solve the time problem. No wonder outdated NVRAM* hurts if device's clock are wrong. Sep 09 12:03:42 bv03: Sep 09 12:03:42 root@fon:~$ ls -la /sys/bus/platform/devices/gta01-pm-gsm.0/reset Sep 09 12:03:42 -rw-r--r-- 1 root root 4096 Mar 5 17:13 /sys/bus/platform/devices/gta01-pm-gsm.0/reset Sep 09 12:03:55 on bv3 it can really reset the gsm afaik.. but its buggy on suspend to ram thus we removed q1001 Sep 09 12:04:12 something went very wrong on my system. I don't think I have my / filesystem any more. Sep 09 12:04:20 If I drop out from the chat, I had to reboot. Sep 09 12:04:24 q1001 is the transistor triggering the hard-reset on gsm Sep 09 12:04:57 roh - Can you elaborate a bit on q1001? Sep 09 12:05:18 * pavelm goes on to reboot Sep 09 12:05:43 roh: toggling reset will do no harm, right? Sep 09 12:05:43 roh - BTW: My bv03 survives about a week in OFF (after doing "halt" in the running linux). So this whole q1001 story seems a little more complex. Sep 09 12:05:48 there are 2 gpio lines, power_on and reset going from the app cpu to the gsm. these 2 are made avail via the sysfs entrys. Sep 09 12:06:18 they switch a transistor each which 'presses' the 'button' on the gsm Sep 09 12:06:40 roh - *listening* Sep 09 12:06:59 rwhitby on gta0bv04 nothing happens. the line ends in the void since q1001 is not placed. Sep 09 12:07:31 ok, so my fix is correct then (it's just superfluous but innocuous on bv4) Sep 09 12:07:47 Elrond: the power_on is working as it should. the reset is, as far as i understood only intended for testing and development. thus its not debounced. Sep 09 12:07:55 I'm assuming that future devices will have a positive logic reset ... Sep 09 12:08:47 roh - Why exactly was q1001 removed? Sep 09 12:08:52 Elrond: but since the combination pf pmu and app-cpu we have here results in a very short glitch on the gpios in case of going to deep sleep, it would trigger a reset on the gsm, thus waking up the app cpu Sep 09 12:09:42 and no, the glitch is too massive to be eaten by a cap.. thought of that, calculated and tried it. nope.. q1001 had to go. also thats what ti says. Sep 09 12:09:53 * rwhitby puts the gsmd restart back in /etc/matchbox/session, to see if that fixes (a) the panel icons not appearing and (b) gsm not registering ... Sep 09 12:10:24 also we do not need. it. ive never witnessed a gsm engine crash so far. Sep 09 12:10:28 roh - Ahh. But if the gsm is switched to off (using the power gpio), it doesn't care on the reset "button"? Sep 09 12:10:36 sending 0> and 1 > to power_on triggeres a soft-reset Sep 09 12:11:41 indeed, a gsmd restart in /etc/matchbox/session fixes both (a) and (b). Sep 09 12:11:47 Elrond sending 0 on power_on does not switch the gsm off. for that you have to send at@poff via the serial. thats a limitation of the gsm engine/pmu combo (the gsm has its own pmu) Sep 09 12:12:47 so for a real gsm power off send at@poff and 0> power_on and it should draw no power anymore and really be dead Sep 09 12:12:49 roh - Ahh, so "/etc/init.d/gsm stop" should really send the "at@poff" ? Sep 09 12:12:53 so, my recipe for visible gsm registration success is currently: (1) remove "console=ttySAC0,115200" from uboot bootargs_base, (2) install my fixed gsmd scripts, (3) restart gsmd in /etc/matchbox/session. (2) may well be optional. Sep 09 12:13:12 yes, 2 is optional Sep 09 12:13:13 Elrond i am not shure thats the right place. Sep 09 12:14:01 gsmd restart is perhaps needed because the first time i start it, i get "*** glibc detected *** /usr/sbin/gsmd: free(): invalid pointer: 0x0002e8a8 ***" Sep 09 12:14:02 tuukkah: thx - I just could let the "Can't create directory" errors that I got on a manual gsmd restart persist without being fixed ... Sep 09 12:14:21 s/could/couldn't/ Sep 09 12:14:22 rwhitby meant: tuukkah: thx - I just couldn't let the "Can't create directory" errors that I got on a manual gsmd restart persist without being fixed ... Sep 09 12:14:24 Elrond to be exact, power management should not be a job of initscripts in my opinion Sep 09 12:14:37 rwhitby, yeah, it was really irritating Sep 09 12:14:43 roh - Where should it be done? Sep 09 12:15:02 I wish the people who should be looking at an X log had the same aversion to fatal errors ... Sep 09 12:15:24 Elrond i think that should be done via gsmd, but not by gsmd. meaning gsmd should be able to start and run if the gsm modem is ready or not. Sep 09 12:15:25 * rwhitby was horrified the first time he logged the output of the X server ... Sep 09 12:15:42 rwhitby, roh, should we remove GSM_RES from /etc/default/gsmd for now? Sep 09 12:15:50 it should monitor the state of the modem in case it gets switched on, but now do that by itself. Sep 09 12:16:02 if all it does is delay boot by 3 seconds Sep 09 12:16:14 i'm just not sure what app/lib/daemon should control that. this part is still missing Sep 09 12:16:15 roh - Ahh. Okay. Sep 09 12:16:25 uhmmmmm. Sep 09 12:16:32 tuukkah: well, depends whether it's there in GTA02 or not ... Sep 09 12:16:34 who fucked up the gsmd init script wrt GSM_RES? Sep 09 12:16:43 s/now/not Sep 09 12:16:54 roh - Okay, so the right way to switch the GSM fully off is 1) send 'at@poff' 2) "echo 0 >power_on" correct? Sep 09 12:16:59 ack Sep 09 12:17:02 checking if it is undefined and if true piping something into the undefined file? Sep 09 12:17:33 roh - And powering it up is only "echo 1 >power_on" ? Sep 09 12:17:37 * pavelm is back. "ehci received fatal error" is a bad news if your / fs is on USB disk. Sep 09 12:17:49 ixs: which version are you looking at? Sep 09 12:17:54 Elrond ack Sep 09 12:18:03 rwhitby: yesterdays checkout. just compiled... Sep 09 12:18:09 ah, I just fixed that. Sep 09 12:18:10 that is, just finished compiling. :D Sep 09 12:18:20 rwhitby: kk. thx. upgrading then. Sep 09 12:18:30 (I wasn't the culprit - I just fixed it) Sep 09 12:18:32 Elrond that whole mess is just a corner of the whole 'powermanagement' battlefield. Sep 09 12:19:18 roh - Just for my understanding: What would the following leave the modem in? 1) at@poff 2) echo 0>power_on 3) echo 1>power_on 4) echo 0>power_on Sep 09 12:19:22 roh: is anybody actively discussing this right now? I'm very interested because on the htc device we currently misuse the serial port open/close functions to power on/off the modems. Sep 09 12:19:49 (which is the wrong way, but seemingly is way less fragile) Sep 09 12:19:51 ahrgl.... Sep 09 12:20:03 Hmm - now when I dial, I get a nice message saying "Not connected to network. Current status = Home network registered." WTF? Sep 09 12:20:15 rwhitby: muahhahaa Sep 09 12:20:18 Elrond 1-4 in sequence? i think with a running modem drawing power. Sep 09 12:20:21 I was just writing about the same problem I'm seeing. Sep 09 12:20:28 got you own cell network at home? Sep 09 12:20:37 especially as I'm seeing: Sep 09 12:20:38 ** (matchbox-panel-2:1334): DEBUG: gsm_applet_update_signal_strength: signal strength = 14 Sep 09 12:20:43 *your Sep 09 12:21:12 roh, why would we power off the modem in gsmd init script but not add echo "at@poff" | cu -l /dev/ttySAC0 before that? Sep 09 12:21:15 pH5 good question.. i think we should add switching on/off the power to gsmd but controll it from another app. so yes, well need an abstraction for that. Sep 09 12:21:18 roh - Hmm, so the glitch on the gpios could in theory power up the modem, despite doing (at@poff, echo0) ? Sep 09 12:21:23 So I've got back to registered state, but being able to actually make a call has gone backwards. Sep 09 12:21:48 pH5 i thin open/close on the serial is a bit hard.. when gsmd crashes a modem reset shouldnt be neccessary Sep 09 12:22:06 * rwhitby thinks he's in the groundhog day movie - I was fixing gsm registration last sunday night too ... Sep 09 12:22:07 mhm Sep 09 12:22:22 Elrond yes. the ones on the reset especially, since reset is not debounced, power_on is Sep 09 12:22:32 I think I'm fed up and just put the phone back in it's case and that's it. Sep 09 12:22:48 roh: agreed, though it's a nice incentive to make gsmd crash less often ;) (and currently it assumes an uninitialized modem on startup anyway) Sep 09 12:22:58 Elrond thats why the P1 hardware does not have that transistor anymore. so the reboot-loop cannot happen anymore Sep 09 12:22:59 roh - Ahh. The debouncing avoids the gsm seeing the glitches? Sep 09 12:23:43 Elrond japp. but it happens inside the gsm pmu in hardware. so no software is involved Sep 09 12:24:01 roh - Ahh. Sep 09 12:24:38 So after a (at@poff, echo0) the thing is really off, until we do an echo1 (for long enough). :-) Sep 09 12:24:58 roh: is the transistor still present in the P1 devices that were sent to P0 developers? Sep 09 12:25:08 rwhitby - Yes. Sep 09 12:25:16 rwhitby - The ones from .tw, that is. Sep 09 12:25:24 right - that's the one I've got. Sep 09 12:25:28 pH5 japp. but i think the idea with sysfs-switches is nice.. i would adapt it to other phones and rather add a libgsmd function for controlling it Sep 09 12:25:49 if you wanna rework it... i can tell you where to find it. Sep 09 12:25:52 so my 'reset' entry in /sys *does* do something ;-) Sep 09 12:26:11 rwhitby - Yes, yours and mine. ;o) Sep 09 12:26:29 but thats not for starters *looking serious* ... its in the gsm tincan and so hard on the corner that it takes experience Sep 09 12:26:52 that's my defence then if someone queries my GSM_RES in /etc/default/gsmd - "It's in *my* GTA01!" Sep 09 12:27:06 (script) wiki RecentChanges: [[Hardware:AGPS]] [[Neo1973]] [[OpenMoko_Local_Groups:_Berlin]] Sep 09 12:27:20 roh: I'll do normal smd tran removal, but I'm not going inside the can Sep 09 12:27:43 rwhitby: removal as in pliers? :D Sep 09 12:27:46 (at least not until I get a GTA02 ;-) Sep 09 12:27:46 roh - Okay. That now fully explains, why _my_ Bv03 happily survives a week (or more), while powering it up sometimes in the week. :-) Sep 09 12:28:08 roh - I nerver switched on gsm. ;o) So never needed the correct way to switch it off. Sep 09 12:29:38 hi folks, I'm getting this error. anyone familiar with it? Sep 09 12:29:39 ERROR: log data follows (/home/emre/moko/build/tmp/work/armv4t-angstrom-linux-gnueabi/webkit-gtk-0.0+svnr25457-r25447-r1/temp/log.do_compile.17427) Sep 09 12:29:39 | /home/emre/moko/build/tmp/work/armv4t-angstrom-linux-gnueabi/webkit-gtk-0.0+svnr25457-r25447-r1/temp/run.do_compile.17427: line 223: cd: /home/emre/moko/build/tmp/work/armv4t-angstrom-linux-gnueabi/webkit-gtk-0.0+svnr25457-r25447-r1//JavaScriptCore/pcre: No such file or directory Sep 09 12:29:56 Elrond the whole glitch-thing has something to do with the order we switch of the different voltages on the main pmu in case of suspend to ram Sep 09 12:30:35 Elrond we cannot do that exactly in the order samsung says due to a hardware limitation of the pmu. (and it seems nobody ever knew before we saw the spike) Sep 09 12:31:40 roh - Ahh. :) Sep 09 12:31:45 emre_ cd /home/emre/moko/build && . ../setup_env ; bitbake -cclean webkit-gtk Sep 09 12:32:19 Elrond thats one of the moments you'd like to bash your head against the table.. believe me. Sep 09 12:32:48 roh: you're still seeing that many hardware problems nowadays? Sep 09 12:32:58 roh - I know. I have the hardware developers next door at work. :-) Sep 09 12:33:22 elrond: SO we have an explanation why even powered-off openmokos take a lot of power? Sep 09 12:33:44 ixs were getting better... but there were a lot of 'altlasten' in gta01 till we killed them somehow. Sep 09 12:34:00 pavelm - We knew that for a while. *I* did not know, why _my_ bv03 did not exhibit this. :) Sep 09 12:34:09 roh: ahh? so you're mainly working on gta02 and not gta01? Sep 09 12:34:14 ixs its always a fight against things you cannot change, so often you have to workaround them... Sep 09 12:34:24 roh: yes, true. Sep 09 12:34:40 workarounds piled on workarounds until it is ready for shipping. :D Sep 09 12:34:55 ....and solution is to simply remove a transistor? Sep 09 12:35:04 ...or is sw workaround possible? Sep 09 12:35:06 ixs currently the main ressources are on gta02, because even when we finished some design we see as ready for massproduction it stiĂll will take some time todo so. Sep 09 12:35:20 pavelm: the software workaround is to switch off the amp when not in use IIRC Sep 09 12:35:40 ixs: Ok, that should do the trick. Sep 09 12:35:42 pavelm - For "OFF": There is software workaround possible: Switch off the GSM correctly and fully. Sep 09 12:35:45 roh: ahhkay. you still in .de or did you move to .tw? Sep 09 12:35:54 gllin chroot page is up: http://wiki.openmoko.org/wiki/Gllin Sep 09 12:36:00 (but I'm bad with wiki formatting) Sep 09 12:36:08 mp compared to hand-produced lap samples means to set up the whole enchillada of testing, flashing, packaging, and especially documentation of all that to the bit. Sep 09 12:36:11 Elrond: does the GSM draw that much power now? I thought it was rather sparse compared to amp. Sep 09 12:36:14 elrond: So the problem is with "SLEEP"? Sep 09 12:36:37 ixs i'm still in berlin and will stay here. but i'll jet to taiwan sometimes for a few weeks. Sep 09 12:36:51 ixs - The AMP is the other power drainer. Sep 09 12:36:59 ixs: check this logic out: "if (priv->registered != MOKO_GSMD_CONNECTION_NETREG_HOME Sep 09 12:36:59 || priv->registered != MOKO_GSMD_CONNECTION_NETREG_ROAMING)" Sep 09 12:37:15 and hey.. its not the us, and its nice there. if only the light wasnt that long (or they would get me power and connectivity inflight) Sep 09 12:37:35 rwhitby, did you find that yesterday already or was it someone else? Sep 09 12:37:38 roh: hehe. Did you plan on being at mrmcd? Sep 09 12:37:48 rwhitby: YAY! Sep 09 12:37:52 tuukkah: find what yesterday? Sep 09 12:37:53 ixs uh.. dunno.. when is that? Sep 09 12:37:54 pavelm - For SLEEP: Yes, remove the transistor. Sep 09 12:38:05 roh: 14th to 16th Sep 09 12:38:07 rwhitby, that buggy piece of code. someone mentioned yesterday Sep 09 12:38:13 this month? ugh Sep 09 12:38:18 oh, I just found it then Sep 09 12:38:26 information not flowing :-( Sep 09 12:38:41 roh: jupp. I'll leave from the office on thursday, drive up to darmstadt, do a little neo1973 talk on saturday and just go back on sunday in time for noc-work. Sep 09 12:38:50 roh: Thanks, it works. Sep 09 12:39:01 tuukkah: no, I just didn't check http://bugzilla.openmoko.org/cgi-bin/bugzilla/show_bug.cgi?id=807 Sep 09 12:39:41 rwhitby, still Sep 09 12:40:09 roh - Many thanks for the explanations! :-) I assume, I can put that somewhere in the wiki? Sep 09 12:40:37 ok, anyone here with OM svn write access who can fix moko-dialer.c right now ? Sep 09 12:41:05 cause this one is bloody obvious ... Sep 09 12:41:06 pavelm: just fixed up your page to be readable. Sep 09 12:41:28 Elrond if you find a nice place for it Sep 09 12:41:38 pavelm: do you have the libs needed packaged somewhere? Sep 09 12:42:57 roh - I'll look for one. :-) Sep 09 12:43:17 * rwhitby spies mickey|away in #oe and hounds him there ... Sep 09 12:43:44 thanks, ixs. Sep 09 12:44:14 pavelm: you're welcome. Sep 09 12:44:32 ixs: I took them from openmoko-devel-image-fic-gta01-20070803091138.rootfs.jffs2 , IIRC. Sep 09 12:44:37 argh Sep 09 12:44:38 /files/openmoko-tests/build/tmp/work/armv4t-angstrom-linux-gnueabi/webkit-gtk-0.0+svnr25457-r25447-r1/WebKitBuilds/Debug//lib/libWebKitGdk.so: undefined reference to `WebCore::callOnMainThread(void (*)())' Sep 09 12:44:38 | collect2: ld returned 1 exit status Sep 09 12:44:40 pavelm: thx. Sep 09 12:44:48 where is zecke? Sep 09 12:44:52 ;) Sep 09 12:45:06 mickey|away: ping here too - I've spotted you active in #oe 6 minutes ago - please commit http://bugzilla.openmoko.org/cgi-bin/bugzilla/attachment.cgi?id=281&action=view to enable us to dial again Sep 09 12:46:28 hehe. Sep 09 12:48:05 ~praise mickey|away Sep 09 12:48:06 All hail mickey|away! Sep 09 12:50:04 Okay, as soon as I hear something about "mickey, we love you" or "I want your baby" I'm wondering if mickey joined a boy-band... Sep 09 12:50:31 ixs - *lol* Sep 09 12:51:29 openmoko: 03mickey * r2945 10/trunk/src/target/OM-2007.2/applications/openmoko-dialer2/ (ChangeLog src/moko-dialer.c): Sep 09 12:51:29 openmoko: openmoko-dialer2: fix typo and prevent dialing when netreg has been denied Sep 09 12:51:29 openmoko: patch by cesarb, thanks rwhitby. Sep 09 12:54:38 openmoko: 03mickey * r2946 10/trunk/src/target/OM-2007.2/applications/openmoko-dialer2/ChangeLog: openmoko-dialer2: fix description... Sep 09 12:55:01 hmm, dialing of emergency _should_ be allowed with netreg denied. Sep 09 12:55:12 pavelm: it is now Sep 09 12:55:20 (and the description has been fixed) Sep 09 12:56:01 aha, ok. (Out of curiosity, where did you get list of emergency numbers?) Sep 09 12:56:46 112 is GSM standard. If a network doesn's support that, then it shouldn't be calling itself a GSM network. Sep 09 12:57:43 rwhitby: Ok... IIRC t68i allows you to dial 911 in emergency mode, too, plus some stranger numbers. Sep 09 12:57:58 couldn't we just allow all numbers? Sep 09 12:58:01 that would be phone-specifc Sep 09 12:58:15 tuukkah: this patch does allow all numbers Sep 09 12:58:35 oh, so what's the list about? Sep 09 12:58:37 how about *#06# that is suposed to display IMEI-number? Sep 09 12:58:47 tuukkah: dunno, ask pavelm Sep 09 12:59:58 I have not checked how openmoko currently works. (Where do I get reasonable/latest build?) Sep 09 13:00:02 there's something called "parallel bitbake", anyone know where to get it and how to use it? Sep 09 13:00:17 ...but on t68i, you type 911 instead of PIN number, and it dials emergency. Sep 09 13:00:24 pavelm: see backlog for my sucess recipe Sep 09 13:00:42 ...and you can even type 911 without unlocking keyboard, but I believe that's a bug. Sep 09 13:01:14 pavelm, it's so that you can dial even in panic Sep 09 13:01:28 I flashed an image from 20070829, then ipkg upgraded to latest, then put gsmd restart in /etc/matchbox/session, and that should work. Sep 09 13:01:41 tuukkah in build/conf/local.conf add PARALLEL_MAKE = "-j 4" Sep 09 13:01:41 BB_NUMBER_THREADS = "4" Sep 09 13:01:41 INHERIT += "rm_work" Sep 09 13:02:14 this results in 4 threads of bitbake running down the tree invoking make with -j 4 and deleting the crap again after building and packaging Sep 09 13:02:24 roh, thanks! Sep 09 13:02:30 should this be documented somewhere Sep 09 13:02:37 make sure you only do that on bitbake 1.8.8 or later Sep 09 13:02:49 i think it is.. but where... Sep 09 13:03:20 * rwhitby is back to calls working :-) Sep 09 13:04:04 ok, so one week later I'm exactly back where I was a week ago Sep 09 13:06:06 rwitby: alice in wonderland? :-) Sep 09 13:06:36 * rwhitby tries sending an SMS to the neo Sep 09 13:08:16 PARALLEL_MAKE alone wasn't speeding up much Sep 09 13:10:14 pavelm: it's not a bug - it's a feature of most modern mobiles - i believe it's even specified somewhere Sep 09 13:10:59 * rwhitby sees the AT+CMTI indication of the SMS arriving Sep 09 13:13:10 roh: that transistor on my P1- neo could explain why the gsm registration goes away after some hours Sep 09 13:16:54 rwhitby do you suspend the device in between? Sep 09 13:17:06 if i ctrl-c the bitbake and restart, will it pick up from where it was quicky? Sep 09 13:17:18 I leave it plugged into USB 500mA, and let it do what it does. Sep 09 13:17:48 tuukkah: it should do. unless you do a make update, or you cross GMT midnight. Sep 09 13:17:52 roh: can the GSM module do any sort of power saving? e.g. AT+CFUN=n n >= 2? or is their a special muxer command? Sep 09 13:18:49 It can be turned off Sep 09 13:18:54 send it at@poff Sep 09 13:19:08 at+cfun=0 turns off the modem Sep 09 13:19:16 s/modem/radio Sep 09 13:19:54 I'm unsure if it supports 'do not search if you fall off the network' mode - which can significantly reduce power. Sep 09 13:20:06 search unless Sep 09 13:20:30 And increase it if you're moving of course. Sep 09 13:25:14 damn, no bitbake doesn't cache the bb files anymore. what causes that... Sep 09 13:26:12 tuukkah: this only happens if you change the CACHE variable, or have a TMPDIR which is not absolute and issue bitbake from another directory Sep 09 13:26:34 neither of those should apply Sep 09 13:26:48 i have TMPDIR = "${HOME}/oe/tmp" Sep 09 13:26:51 zecke i don't know exactly.. i assume that turning off all the cellbroadcast events and similar stuff will save a lot of power like on other featurephones Sep 09 13:28:08 perhaps such a TMPDIR doesn't actually work Sep 09 13:29:03 roh: Dunno. The power with at+cfun=0 seems pretty close to the power baseline. To the extent that it usually dominates. Sep 09 13:29:49 I need to do yet more measurements. Sep 09 13:29:49 tuukkah: it should work Sep 09 13:32:37 now with BB_NUMBER_THREADS the cpu usage graphs look much better :-) Sep 09 13:37:40 roh, perhaps the cache is cleaned every time i change local.conf? because now i didn't and the cache was used :-) Sep 09 13:38:28 tuukkah ack. the cache needs to be rebuild if you touch a part of the buildsys Sep 09 13:38:39 tuukkah: not cleaned, but as every file depends on local.conf it will reparse ;) Sep 09 13:38:53 ok, stupid me :-) Sep 09 13:44:47 is this now pretty certain, that the GTA02 will only be ready in December? Sep 09 13:44:58 oh no :( Sep 09 13:45:06 * hypa7ia wants one now Sep 09 13:45:14 my treo is dying :( Sep 09 13:48:38 edistar_: no!?! I want one too :/ Sep 09 13:52:40 meus: check the mailinglist and the wiki Sep 09 13:53:01 ok Sep 09 13:53:12 http://wiki.openmoko.org/wiki/Neo1973 Sep 09 13:53:31 my phone(siemens) is so full of bugs, I wonder why it doesn't commit suicide Sep 09 13:54:09 hmm ok thanks edistar_ Sep 09 13:54:31 edistar_, it would, but the suicide function is buggy Sep 09 13:54:42 and only mono sound then.. Sep 09 13:54:55 cb22_: good one ;) Sep 09 13:55:35 GTA01 aren't available anymore, are they? Sep 09 13:56:07 not sure Sep 09 13:56:16 but i cant see lots of people buying them Sep 09 13:56:24 considering 02 is out soonish Sep 09 13:56:55 cb22_: _ish_ Sep 09 13:58:21 * edistar_ has been waiting for the end-user hardware for a year now :( Sep 09 13:58:43 * cb22_ agress Sep 09 13:58:53 * meus too Sep 09 13:58:54 s/ss/es/ Sep 09 13:59:05 note the _hardware_ bit Sep 09 14:00:00 cb22_: yes, I don't mind a few bugs or missing software etc cause I know it will be there sometime, but I like having something to play with;) Sep 09 14:00:07 same here Sep 09 14:01:12 at least it gives me time to save Sep 09 14:02:55 * cb22_ has been doing odd jobs to raise money Sep 09 14:06:04 cb22_: e.g.? ;) Sep 09 14:06:15 heh Sep 09 14:06:25 doing some graphics for my bro Sep 09 14:06:36 getting paid R1000 = $150 or so Sep 09 14:06:45 that's alright.. Sep 09 14:07:22 I heard that one doesn't need the dev board if you just want to install and update the OS? Sep 09 14:07:38 yeo Sep 09 14:07:40 yep* Sep 09 14:07:46 its only really needed for uboot-dev Sep 09 14:12:09 cb22_: okey Sep 09 14:14:37 The dev board is also very useful for kernel debugging Sep 09 14:17:47 (script) openmoko-community: Sean Moss-Pultz Re: speeding up the flow of information Sep 09 14:19:25 aloril: ? Sep 09 14:19:57 Hm, looks like someone forced the revision of binutils to be the pre-2.18 one, thus unbreaking the build for amd64 Sep 09 14:20:08 * cesarb will have to remove the unbreaking patch from his local overlay then Sep 09 14:20:58 edistar: that scripted line means that there is new mail from Sean at openmoko-community mailing list Sep 09 14:21:14 aloril: I know. I wasn't talking about it. Sep 09 14:21:23 oops Sep 09 14:21:30 * cesarb got confused somewhere :D Sep 09 14:21:53 I think I should have slept more :D Sep 09 14:27:47 (script) wiki RecentChanges: [[OpenMoko_Local_Groups:_Trondheim]] [[OpenMoko2007.2]] [[Gllin]] Sep 09 14:41:25 aloril: thx Sep 09 14:41:59 what is HXD8 stuff? Sep 09 14:42:40 sssh, dont say it too loud Sep 09 14:42:46 other than what source code reveals we don't know (probably not a phone) Sep 09 14:42:47 the black helicopters might start appearing Sep 09 14:43:02 wiki HXD8 page has information based on source code I think Sep 09 14:49:17 ah Sep 09 14:49:31 a future hardware shaped like a heli perhaps? ;) Sep 09 14:52:06 looks interesting, but the resolution is quite small? Sep 09 14:58:23 another question, I compiled the qemu makefile etc. it seems to emulate the hardware? Can't I compile openmoko for x86 easily? then I think it would be much faster Sep 09 14:59:11 edistar you could. but thats something completely different. qemu executes the real arm images and thus is practical for testing Sep 09 14:59:39 in the wiki are some instructions how to use Xoo or xephyr.. but i didnt try yet Sep 09 14:59:58 roh: Xoo would be compiled natively? Sep 09 15:00:11 Xoo is like X11 in X11 afaik Sep 09 15:00:22 roh: Sep 09 15:00:24 roh: ok Sep 09 15:01:17 edistar, another easy thing to do if you just want to see how apps you write will look is to use the GTK theme Sep 09 15:02:12 cb22_: okey Sep 09 15:02:43 Please help me: http://tinyurl.com/2yu2te Sep 09 15:02:49 cb22_: actually I just wanted to try out the interface etc and see if I like it ;) But the emulated version is just way to slow on my PC Sep 09 15:03:55 anyone got xephyr to work? Sep 09 15:05:30 edistar, did you already have a look at http://wiki.openmoko.org/wiki/Xephyr Sep 09 15:05:48 tuukkah: yes Sep 09 15:06:02 I can't find xephyr in my debian repos Sep 09 15:06:04 ;) Sep 09 15:06:21 xserver-xephyr Sep 09 15:06:33 and xoo Sep 09 15:06:58 ok Sep 09 15:07:29 the load of work is going to be compiling openmoko unless you can find prebuild binaries for x86 Sep 09 15:07:52 tuukkah: I have a few days.. Sep 09 15:09:02 compiling openmoko apps only for x86 should be quite fast actually Sep 09 15:09:31 aloril, but where do you find instructions for doing that? Sep 09 15:09:47 just few apps vs whole distro when compiled for arm Sep 09 15:10:24 tuukkah: at wiki somewhere, it has been several months ago (when source became available) when I compiled for PC ;-) Sep 09 15:11:48 (haven't looked, but mokomakefile might also have suitable target) Sep 09 15:12:09 why does mokomakefile need to me in /home/moko/Makefile? Sep 09 15:12:19 It doesn't Sep 09 15:12:41 But you'll have to fix a few things if you choose to move it. Sep 09 15:13:06 mwester: oh... hm Sep 09 15:13:22 if I compile for x86, I can't do the automatic mokomakefile think, can I? Sep 09 15:13:48 You'll have to make manual modification, AFAIK Sep 09 15:14:06 mwester: ok Sep 09 15:14:43 Which rather defeats the whole purpose of the Makefile -- by the time you set it up to build in a directory outside of your home dir, and make it target an x86, well -- you might as well just learn OE and use the native bitbake commands. Sep 09 15:14:58 edistar: mwester: unless something changed recently, you can but mokomakefile wherever you like. Sep 09 15:15:16 s/but/put/ Sep 09 15:15:18 Writchi1 meant: edistar: mwester: unless something changed recently, you can put mokomakefile wherever you like. Sep 09 15:15:35 thx apt Sep 09 15:15:37 ;) Sep 09 15:15:46 I ended up with a couple of directories in my home dir that I had to replace with sym links, perhaps that's been fixed. Sep 09 15:15:46 will the comercial version of neo have camera? Sep 09 15:15:46 There is no camera included. With (battery) powered USB hub, you should be able to attach to (almost) any external camera. See http://wiki.openmoko.org/wiki/Neo1973_Hardware for details on what is included. Also see http://wiki.openmoko.org/wiki/FAQ (also some model next year probably will gave camera module: http://www.channelweb.co.uk/vnunet/news/2192036/first-open-source-mobile-phone ) Sep 09 15:16:16 i put mine in /home/user/om2007.2 for latest and /home/user/openmoko for the earlier builds Sep 09 15:16:17 tuukkah: it could be something normal like: svn checkout && ./autogen.sh && ./configure && make Sep 09 15:17:30 with GTA02 no need for battery powered USB hub, it has power Sep 09 15:17:33 aloril, yeah Sep 09 15:17:57 * mwester finishes reading the log, and with high hopes and great optimism, sets out to sync up and rebuild a new image... :-) Sep 09 15:18:36 how far is the software actually? what works(at least a bit) and what doesn't at all? Sep 09 15:18:44 aloril: that will be nice :) then one can connect a good camera to the phone. geotag the images and send them somewhere :) Sep 09 15:20:03 yeah the statement about gta02's usb power was a nice surprise Sep 09 15:24:40 mjr: where can I read that statement? Sep 09 15:26:17 Maledictus its right.. gta02 will be abĂle to source 100mA at 5V (at the usual penalty of runtime) Sep 09 15:27:06 ok, thats very cool. but how do you know? Sep 09 15:27:34 I think it was laforge who said it on the one of the lists Sep 09 15:27:48 some of the "official" guys anyway Sep 09 15:28:17 yep Sep 09 15:29:07 yay. thank you guys for this cool info :) Sep 09 15:29:25 this phone just gets cooler every day :) Sep 09 15:29:45 anyone tried another bootloader like redboot? Sep 09 15:30:20 Maledictus whatfor? Sep 09 15:30:24 edistar, http://wiki.openmoko.org/wiki/How_to_run_OpenMoko_Apps_on_PC Sep 09 15:30:24 dunno, would anyone want to? Sep 09 15:30:54 roh: just curious. it's open and thats about choice ;) Sep 09 15:31:46 in south africa we have a brand of condoms called choice Sep 09 15:31:55 Maledictus i think you can always try to port it.. i just see no real sense in it Sep 09 15:32:00 6 million of them were just recalled due to failing quality standards :p Sep 09 15:32:12 lol cb22_ Sep 09 15:32:37 ahh, roh = person from sean's speeding up the flow of information email :) Sep 09 15:34:28 cb22 japp. i'm here a few month already, but now we hired michael and sean thought its good to introduce us together Sep 09 15:35:14 cool Sep 09 15:36:00 we will work out some concept this month and then see if you like it. will mean much more regular updates and channels for informations. Sep 09 15:36:45 w00t! :) Sep 09 15:37:24 * cb22 checks his gta02 slush fund: only $100 more needed Sep 09 15:37:34 roh: thanks - the effort is appreciated! Sep 09 15:38:14 roh: cool! Sep 09 15:52:17 * mwester shakes his head sadly, watching his build waste valuable time compiling "oh-puzzles"... Sep 09 15:53:12 mwester most of the time goes by for locales, configure scripts and sourcecode repos Sep 09 15:54:41 hm openocd-native-0.0+svn20070909 needs to get srcrev-ed also Sep 09 15:57:45 roh: I shaved a huge amount of time off the build by disabling locales. I'm not sure what's going to break at run-time because of that, but so far so good. Sep 09 15:59:36 http://www.phoronix.com/scan.php?page=news_item&px=NjA0NQ Sep 09 16:00:29 do locales take a lot of time even if you only build one of them? Sep 09 16:01:50 El_Salvador: funny to see rumors travel so fast Sep 09 16:02:41 Sigh. Sep 09 16:03:29 What's the wiki markup for flashing red, with trumpets, for the 'preliminary estimates on the timeline' phrase. Sep 09 16:03:43 ? Sep 09 16:04:02 aaa? Sep 09 16:04:25 media... Sep 09 16:04:44 SpeedEvil i rather think of some 'this revision of $wikipage is reviewed by openmoko'-page ;) Sep 09 16:05:04 but thats a lot of work Sep 09 16:06:20 roh: that would be the nice way - or even some official timeline :) Sep 09 16:06:44 SpeedEvil even that would only be estimates. Sep 09 16:07:29 Of course - however, you presumably can nail down some dates. at least on the low end. Sep 09 16:07:46 As best-case dates which will probably slip. Sep 09 16:08:22 But you're unlikely to - for example - have the testing phase complete in half the intended time. Sep 09 16:09:48 would be pokering Sep 09 16:12:15 pokering? Sep 09 16:12:31 SpeedEvil poker with dates Sep 09 16:13:35 Ah. Sep 09 16:14:13 I'm going to have to disagree. The maximum date is quite indeterminate. The minimum date much less so. Sep 09 16:14:44 yep. but the press then would always take the minimum date Sep 09 16:14:50 Especially when the minimum date seemingly exceeds published dates. Sep 09 16:27:17 roh: do you know if the powered USB is going to be USB OTG or some aed-hoc non-standard scheme? Sep 09 16:28:22 Writchie, i'd imagine its just normal, powered usb Sep 09 16:28:25 just like on a PC Sep 09 16:28:28 (script) wiki RecentChanges: [[Talk:Neo1973]] [[Advanced_End-user/es]] [[Manually_using_SMS]] [[Manually_using_Bluetooth]] [[Manually_using_GPRS]] [[Manually_using_GSM]] [[Main_Page/es]] [[Media_Content/es]] [[Advanced_End-user]] [[Coding_Guidelines/es]] [[PIM_Storage]] and other changes Sep 09 16:28:34 not OTG Sep 09 16:28:47 Writchie afaik its a powered usb host port on the 'wrong' connector Sep 09 16:29:01 a simple cable hack will solve that Sep 09 16:29:02 is this another connector?? Sep 09 16:29:03 wrong because its mini-b not a Sep 09 16:29:21 b is ok for client. for host its nonstandard. Sep 09 16:29:48 * cb22 imagines ripping off the cables off all his usb devices and replacing them with mini versions Sep 09 16:30:07 the first thing i'm doing is my nice cheap 4gb flash disk Sep 09 16:30:08 * Writchie Writchie imagines smoked laptops Sep 09 16:30:09 roh: couldn't it cause problems with normal "device" mode if the hosts's voltage ends up lower than the neo's? it would cause current going in the wrong direction... Sep 09 16:30:36 s/lower/slightly lowe/ Sep 09 16:30:37 cesarb meant: roh: couldn't it cause problems with normal "device" mode if the hosts's voltage ends up slightly lowe than the neo's? it would cause current going in the wrong direction... Sep 09 16:30:48 urgh Sep 09 16:30:52 slightly lower Sep 09 16:31:01 wow "NOTE: package openmoko-devel-image-1.0: completed" Sep 09 16:32:00 do you think 100ma could power a optical mouse? Sep 09 16:32:03 yes. Sep 09 16:32:07 most optical mice. Sep 09 16:32:12 heh ^-^ nice Sep 09 16:32:32 although it will be more fun hacking batteries into the mouse to power it Sep 09 16:32:32 sb22: the problem will be with USB mass storage devices Sep 09 16:32:52 Writchie, i doubt that, sure, it may not be able to power my 160gb removable laptop hdd Sep 09 16:32:52 many if not most of these take > 100ma Sep 09 16:33:00 but a flash disk should be fine? Sep 09 16:33:04 cb22: why mice? Sep 09 16:33:12 cb22: a keyboard I can understand, but mice? Sep 09 16:33:16 cesarb, because, i go to school. Sep 09 16:33:25 and random things have a certain 'cool factor' about them Sep 09 16:33:31 like having a mouse on a phone Sep 09 16:33:36 bluetooth mouse would be better. Sep 09 16:33:41 true Sep 09 16:33:45 cb22: I have to agree with you that it's random Sep 09 16:33:53 cb22: i don't know where you are getting your information. Virtually all flash diisk requirem more than the 100ma minimum Sep 09 16:34:06 s/disk/sticks/ Sep 09 16:34:17 cesarb, i just want to laugh at all my friends with their phones when i can do things they never even thought possible :) Sep 09 16:34:25 Writchie, i didnt get it from anywhere, i just presumed Sep 09 16:34:31 They tend to use less power in USB1.1 mode. Sep 09 16:34:48 yes - some of them used to work but not virtually all use 2.0 Sep 09 16:34:59 and request the power on enumeration. Sep 09 16:35:16 which is why they can't be plug into usb keyboards with extra USB slots Sep 09 16:35:18 No, I mean they actually use less power when connected to a 1.1 host Sep 09 16:35:27 SpeedEvil: yes Sep 09 16:35:45 # Sep 09 16:35:45 # Power Consumption: Write 36.0mA, Read 26.0mA Sep 09 16:35:58 from a sandisk datasheet Sep 09 16:36:03 another one says >100 though Sep 09 16:36:08 cesarb yes,, there needs to be some kind of gui-lockout to make sure no usb host is connected before switching to host itself Sep 09 16:36:09 WiFi USB adapters also take more than 100ma Sep 09 16:36:48 roh: is there some "Philosophical" objection to USB OTG? Sep 09 16:36:51 Great, now it reliably stops at ATE0V1 instead of at ATZ Sep 09 16:37:35 nothing like reliable failure Sep 09 16:38:02 yep, my mouse is in 100ma mode Sep 09 16:38:40 and my flashdrive is using... Sep 09 16:38:40 * Stephmw catches up on the OTG conversation Sep 09 16:38:55 80ma Sep 09 16:39:01 at least according to windoze Sep 09 16:39:09 * mjr would assume that it's more the practical consideration that the SoC happens to support usb host/gadget and not otg Sep 09 16:39:16 roh: I note in Harald's last mail the S3C2242B being 64MB RAM with integrated SDRAM and flash. Is the GTAv02 going to have a less ram or both integrated and off module RAM. Sep 09 16:39:52 suprisingly enough, my removable hdd apparently draws 96ma Sep 09 16:39:53 mgr: I don't see any problem implemented USB-OTG with the S3C2242B. Sep 09 16:40:16 cb22: where are you getting those numbers from? Sep 09 16:40:23 mjr: from my experience, usb-mass-storage on cellphones comes in two flavours (I'll get to otg next): a) removable card is the only accessible drive, and access is exclusive: either by device *or* by pc Sep 09 16:40:31 windows' usb root hub properties SpeedEvil Sep 09 16:40:32 mjr: with the right USB interface chip Sep 09 16:40:38 mjr: b) device reboots into usb mass storage mode Sep 09 16:40:58 no usb plugged in, then i plug in hdd: power required = 96ma Sep 09 16:41:00 mjr: tbh, both approaches seem to be ways to avoid dealing with sync Sep 09 16:41:03 cb22: sounds unlikely to be direct current readings Sep 09 16:41:19 mjr: otg likely will suffer from some similar stupid constraint Sep 09 16:41:33 Is this a powered USB drive cb? Sep 09 16:41:34 Writchie, yeah sure, but SoC comes with usb 1.1 for "free" Sep 09 16:41:37 SpeedEvil, no Sep 09 16:41:42 its a laptop hdd in a case Sep 09 16:41:49 I give up for today... Why would the modem not answer to "ATZ" (or sometimes the following "ATE0V1")? Sep 09 16:41:52 96mA is basically impossible. Sep 09 16:42:11 All laptop hard drives are at >>250mA Sep 09 16:42:29 beh, dumb windows Sep 09 16:42:42 I measured my Kingston 1GB thing on my keychain as 50mA 'static' in USB2 Sep 09 16:42:46 and some are >> 500ma requiring two USB slots for power Sep 09 16:42:55 And about 30mA on USB1.1 Sep 09 16:43:11 And some 80mA IIRC writing at max speed Sep 09 16:43:14 * cb22 looks around for his multimeter Sep 09 16:43:17 (on USB2) Sep 09 16:43:35 it would probably be a good idea to put this info on the wiki somewhere Sep 09 16:43:38 cesarb: shutdown the neo, remove the battery, wait 30sec, put back the battery and it works Sep 09 16:43:38 SpeedEvil: and what power does it request at enumeration Sep 09 16:43:46 under powered usb devices or something Sep 09 16:44:12 Obri: no, it doesn't Sep 09 16:44:13 hang on Sep 09 16:44:19 I'd guess 100mA Sep 09 16:44:25 Obri: in fact, that's my testcase Sep 09 16:44:32 * SpeedEvil looks for his keys Sep 09 16:44:49 apparently my unknown 1 gig USB stick is happy with 100 mA Sep 09 16:45:01 most usb disks illegally take >0.5a from a single port Sep 09 16:45:08 (according to lsusb) Sep 09 16:45:08 since most implementations don't set the limit at that level... Sep 09 16:45:20 except for macs Sep 09 16:45:35 hm Sep 09 16:45:39 anrp: yeah, my usb stick completely misbehaves on the monitor's builtin hub Sep 09 16:45:44 SpeedEvil: you need a beeper with bluetooth for your keys. think of key_finder-0.1.bb :) Sep 09 16:45:55 anrp: and the bluetooth dongle is worse... lsusb says it's self-powered (!) Sep 09 16:46:24 mjr: how do you get lsusb to give you that info? Sep 09 16:46:33 -v Sep 09 16:46:42 * cesarb has once tought of adding a "quirks" list to the linux kernel which would tell the kernel the correct values, so it could warn you are misplugging the thing Sep 09 16:46:53 Writchie no.. atleat nothing i know of. Sep 09 16:46:56 s/tought/thought/ Sep 09 16:47:04 mjr: cheers Sep 09 16:47:32 roh: thanks. Sep 09 16:47:44 It says max_power 200mA Sep 09 16:47:56 BTW: Does anybody know, if g_serial linux console will be possible once? Sep 09 16:48:22 Writchie i just read myself a bit into otg... i think the point was that its nearly not implemented yet out there and we'd rather not change anything on the usb since it works.. you know how that is when there is a lot to do and you have something which is already done and working Sep 09 16:48:24 dcordes_: I've investigated producing one. It's complex. Sep 09 16:49:02 hm Sep 09 16:49:05 Elrond you can switch of g_ether and load g_serial.. of course.. but its less usefuĂll than ip with multiple connections Sep 09 16:49:12 ppp Sep 09 16:49:15 :) Sep 09 16:49:22 hrr Sep 09 16:49:35 cb22: if you use a usb hub for your external voltage needs there's no problem at all Sep 09 16:49:43 yeah.. have fun.. hack it into gsmd and put it on the drop-ins Sep 09 16:49:43 yep Sep 09 16:49:46 roh: for dual use do you know of any powered dual host/device implementations that are not USB - OTG? Sep 09 16:49:48 s/gsmd/neod Sep 09 16:49:52 roh - The idea is to have the main console on usbtty instead of on a real uart. Sep 09 16:49:58 im pretty sure powered usb hubs are happy with 100ma Sep 09 16:50:13 Writchie currently i know of no massmarket devices ive seen that can do both Sep 09 16:50:39 Writchie shure, there is some hardware which could do usb host unpowered.. but mostly the software does not allow that Sep 09 16:50:41 * SpeedEvil is currently trying to work out how to do powered USB for 1.1, as a user-installable kit. Sep 09 16:50:58 err Sep 09 16:50:59 GTA01 Sep 09 16:51:06 the potential for device damage is one of the primary motivations for OTG. Sep 09 16:51:23 even a partial/poor OTG is better than two devices sourcsing power on the same line. Sep 09 16:51:32 mini-b-mini-b cablĂe, cut positive powerline and some 5V source for the client Sep 09 16:51:49 Writchie ack Sep 09 16:52:11 Isn't the neo having a mini-ab socket? Sep 09 16:52:20 The trivial way round that is to only ever enable 5V when there is not 5V on the input Sep 09 16:52:39 roh: i also note change to large block NAND - this conflicts with prior info but I think it is inevitable. Sep 09 16:53:19 Though there are of course annoying edge cases. Sep 09 16:53:19 Elrond ive never seen a mini-ab.. have a pic? i think its mini-b Sep 09 16:53:25 that zaurus CXX00 devices have powered USB OTG Sep 09 16:53:31 so they managed to work it Sep 09 16:53:56 roh - OTG specs mini-ab, I think. let me look around. Sep 09 16:54:12 roh: how is powered USB done on GTA02 - is it basically just echo 1 >/sys/usb-power? with no special logic? Sep 09 16:54:12 yea you can use small things like keyboards and mice with the zaursus Sep 09 16:54:19 without extra power Sep 09 16:54:38 OTG is 5 pin mini AB IIRC Sep 09 16:54:56 for devices with higher currency needs I have my http://img118.imageshack.us/my.php?image=z1wx9.jpg Can't wait to try it with gta02 Sep 09 16:55:12 Writchie: all mini USB are 5 pin Sep 09 16:55:21 There is a 4 pin variant Sep 09 16:55:24 Writchie: but only mini-AB actually has a use for the ID pin Sep 09 16:55:29 ditto Sep 09 16:55:58 SpeedEvil: never heard of 4-pin mini-USB Sep 09 16:56:22 cesarb: was used on early portables e.g. some mp3 players Sep 09 16:56:24 HUh. "Mini AB Buchsen sind aus der Spezifikation entfernt worden." (wikipedia), translated: "mini ab were removed from the spec"... strange. Sep 09 16:56:36 cb22: you planning to use usb devices with battery? Sep 09 16:58:59 cesarb: I may have assumed that an existing connector I had was microUSB - but it seems proprietory Sep 09 16:59:37 while doing make setup it wants to connect to monotone.openembedded.org, but it seems that it can't, it's taking half an hour now.. Sep 09 16:59:37 SpeedEvil there is a step-up converter for the 5V with up to 100mA which can be switched on... i think there is no driver for that yet Sep 09 17:00:45 xSmurf : yes Sep 09 17:01:01 roh - Notes on mini-ab, but no pic: http://en.wikipedia.org/wiki/USB_On-The-Go#Plug Sep 09 17:01:53 Elrond heh.. were there about 10 minutes ago. Sep 09 17:02:07 ive seen a pic of mini-a .. but not ab yet Sep 09 17:02:49 this otg spec is loony big Sep 09 17:03:01 no wonder laforge ignored it for now Sep 09 17:03:25 seems like proper otg support is rather something for a soc-project Sep 09 17:03:26 roh: the important part is determining the device roles Sep 09 17:03:56 roh: it has things like role switch for instance (two devices with mini-AB sockets and a mini-A to mini-B cable, they can chose which is host and which is device); no wonder it's big Sep 09 17:04:06 the additional hardware is basically easy.. but someone has to design and test it. and then implement the software side Sep 09 17:04:41 Writchie: without the 'soft' role switch, it's a simple matter of reading the ID pin to see if you are in the A or B end of the cable Sep 09 17:04:52 roh - Found it: http://wiki.openmoko.org/wiki/Neo1973_Hardware#USB_Device (USb Device -> last point "mini ab". With link to pic) Sep 09 17:05:05 Writchie: if you are at the mini-A end, you are the host; if you are at the mini-B end, you are the device Sep 09 17:05:19 Writchie: the role switch just allows you to change it to the opposite (and back) later Sep 09 17:05:51 cesarb: basically yes, this is what i mean by partial implementation Sep 09 17:07:31 Writchie: that partial implementation should be very easy to do... IIRC, just one pull-up on the ID pin and some way of reading it back to know if you are the host or the device, and that's it Sep 09 17:07:39 Elrond hmm.. thats wrong.. the device has only mini-b atm Sep 09 17:08:11 NXP ISP1109 does it all along with serial port and audio Sep 09 17:09:20 ISP1109 << $1 Sep 09 17:10:12 cesarb: there are many cases where routing of a couple of wires is severaly impacting functionality of the device. Sep 09 17:10:51 project would wise to harness some of the hardware expertise "open source" for the truly non-proprietary parts. Sep 09 17:11:15 there is no fear that anyone will copy FIC's PM design ;) Sep 09 17:11:32 Writchie: well, I'm more of a software guy, so I tend to forget that routing wires isn't as easy in hardware ;-) Sep 09 17:11:44 prolly only the interchip writing of the TI chipset is truly proprietary Sep 09 17:11:58 s/writing/wiring/ Sep 09 17:11:59 Writchie: and the hammerhead one Sep 09 17:11:59 Writchie meant: prolly only the interchip wiring of the TI chipset is truly proprietary Sep 09 17:12:13 those parts could just be black boxed Sep 09 17:12:38 and an in system PLD Sep 09 17:12:46 we can dream maybe for GTA03 Sep 09 17:13:34 cesarb: routing can be easy if its designed for flexibility. Sep 09 17:13:42 this is not normaly the case for handset hardware Sep 09 17:14:45 one unique aspect of openmoko is that features can be added a year later to the existing base of devices Sep 09 17:15:19 roh - the mini-ab thing was there from the start, I'm quite sure. It's not a "someone faked it up". Sep 09 17:15:29 roh - there := "in the wiki" Sep 09 17:20:22 Elrond: Hehe.. ":=" ;) Sep 09 17:25:45 Elrond hm.. just fixed it Sep 09 17:32:31 Writchie the isp could not help.. its only serial/spi Sep 09 17:32:40 Writchie and is 5x5mm.. which is big Sep 09 17:34:51 re Sep 09 17:35:29 roh: isp1109 multiplexes/level shifts USB, serial, spi, mic, speaker, all on same connector and includes USB OTG logic. Sep 09 17:36:17 yes.. and thats nice. but does take a lot of pcb space and not give us usb2. besides that it would only be avail via spi Sep 09 17:37:08 also the count of external components to the soc is limited by capacitance of the parts and routes on the pcb Sep 09 17:38:16 5 mm is not that big. USB will need different SOC. Gives you serial without needed to bring up a USB device stack. Sep 09 17:38:22 but this is all academic. Sep 09 17:39:33 roh isp1009 would have eliminated need for a debug board Sep 09 17:39:55 i guess not really - you need the jtag Sep 09 17:40:51 then again, i think you do a serial boot Sep 09 17:41:13 s/do/could do/ Sep 09 17:41:14 Writchie meant: then again, i think you could do a serial boot Sep 09 17:43:17 Writchie thats not much of a gain... in the end a real serial is what you want. Sep 09 17:44:59 roh: but rs232 to ttl is widely available for a few bucks Sep 09 17:45:49 in any case USB OTG is the only right way to do both host and device on same connector - IMHO Sep 09 17:48:50 Writchie we'll see. but propably thats something which will not come with gta02 Sep 09 17:49:04 you mean GTA03? Sep 09 17:49:25 or gta02v4? Sep 09 17:49:32 naming afterwards and designs are not decided yet. Sep 09 17:49:39 gta02. Sep 09 17:50:03 No reason you couldn't setup uboot to use a USB serial thingy attached to the host port. Sep 09 17:50:05 gta02 will be powered at wish, but not fully otg compliant. Sep 09 17:50:29 what are you calling the "September 20" freeze Sep 09 17:50:47 SpeedEvil yes and no. uboot currently only knows the udc client mode controller Sep 09 17:50:56 indeed. At the moment. Sep 09 17:51:02 SMOP Sep 09 17:51:22 SpeedEvil: i was thinking pre-u-boot. SoC can boot from serial or NAND depending on pin configuration IIRC Sep 09 17:51:26 SpeedEvil either way. usb client acm is quite fine i think. seen much worse Sep 09 17:51:54 I was meaning if for some reason it was desired to totally replace the debug board Sep 09 17:51:59 roh: if you assume linux's only Sep 09 17:52:09 Writchie i meant uboot Sep 09 18:02:38 what is HXD8? Sep 09 18:04:17 A different device that details have not been revealed of Sep 09 18:05:44 i hope this one has a design review Sep 09 18:07:29 http://wiki.openmoko.org/wiki/HXD8 Sep 09 18:27:50 (script) wiki RecentChanges: [[Phase_1_Software_Testing]] [[OpenMoko_Local_Groups:_Norway]] [[OpenMoko_Local_Groups:_Trondheim]] [[Neo1973_Hardware]] [[Neo1973]] Sep 09 18:29:29 counter Sep 09 18:29:29 http://londerings.novalis.org/wlog/index.php?title=Neo1973_counter Got it, get yours from openmoko.com (and read topic) (1807;252) Sep 09 18:30:59 Does anyone have an idea of what would cause this: gsmd sends "ATZ", modem doesn't answer. I kill -9 gsmd and start it again, it sends the same "ATZ", modem answers, so the modem isn't dead. Sep 09 18:31:18 I'm completely stuck on that... It's the only thing preventing gsmd from working successfully on boot. Sep 09 18:32:03 even a good question is whether we have a bug number for that Sep 09 18:33:23 cesarb: Is there any possibility of a modem lockup coming out of reset? aren't the schematics on the wiki somewhere....I'll try and look at them. Sep 09 18:33:35 tuukkah: I'm investigating it as part of bug 766, but it's probably a separate bug Sep 09 18:33:59 moko-bunny: yes, but I tried putting high sleep times (like 15 seconds) just before gsmd start Sep 09 18:34:13 moko-bunny: and the only schematics you'll find in the wiki are for the debug board Sep 09 18:34:55 cesrab: dang....Harald seems like too good of an engineer to have overlooked that, but I know many ICs have strange powerup requirements Sep 09 18:35:38 moko-bunny: the modem is powering up correctly, since I can kill gsmd and restart it (by hand, not via the init script, so no power changes), and it will answer correctly Sep 09 18:35:53 you should see the fun stuff we did to guarantee a workaround on a freescale 5485 for a power up issue. :) Sep 09 18:36:00 moko-bunny: either it doesn't answer correctly soon after being woken up, or there's some wierd handshaking issue Sep 09 18:36:23 * cesarb is a software guy, so I sometimes don't think of crazy things like ground bounce Sep 09 18:36:52 cesarb: hmm...cool, I'm pretty neophyt-ish with the inner workings, I didn't know whether restarting gsmd did any power cycling...all it seems to do on my hardware is lock it up :) Sep 09 18:37:19 well...not completely true...but if I try and use init.d it locks up Sep 09 18:37:30 moko-bunny: the init script does power-cycling (/etc/init.d/gsmd), which is a pain with the recent lockup-prone kernel Sep 09 18:37:39 Writchie - hxd8 is a washing machine. ;o) Sep 09 18:37:44 moko-bunny: gsmd itself doesn't (and in fact assumes the modem is on) Sep 09 18:38:13 cesarb: ahhh...thus explains a few things. I'm onboard with what you now. Sep 09 18:38:26 s/what// Sep 09 18:38:26 moko-bunny meant: cesarb: ahhh...thus explains a few things. I'm onboard with you now. Sep 09 18:39:10 tuukkah: it's bugs 707 and 705 Sep 09 18:41:00 cesarb: hmmm, well unfortunately I'm thinking that a good look at the full gsm chipset datasheets would be needed. IC documentation seems to have gotten generally worse in recent years...but usually stuff like that is in there. Sep 09 18:42:16 moko-bunny: perhaps. I'd also try to hook something (osciloscope? data logger?) to the serial lines (which are brought out to the debug connector, so it should be easy), but as I said, I'm not a hardware guy, so the best thing I have is a puny digital multimeter Sep 09 18:43:27 cesarb, true Sep 09 18:44:10 great, this time I managed to get the even more bizarre variant (after adding a huge sleep 60 before gsmd start): it answers ATZ with OK... and doesn't answer the following ATE0V1 Sep 09 18:44:11 cesarb: well, I'd be happy to loan you my scope, but it's an old analog piece of junk.....plus I probably live nowhere near you. Sep 09 18:44:59 to make things even more interesting, when it works it tends to echo back the ATZ and ATE0V1... when it doesn't but answer to the ATZ, there's no echo back Sep 09 18:50:33 cesarb: I wonder if the gsm chipset has auto baud detection and is making a wrong selection sometimes on power up? Sep 09 18:51:28 moko-bunny: could be... except that the serial console uses the same baud rate Sep 09 18:52:26 moko-bunny: except if the switch from serial console to gsmd glitches it... Sep 09 18:52:40 * cesarb will try several boots with the serial console disabled to see if things get better Sep 09 18:53:09 What bothers me about this: Why doesn't the power button work? Sep 09 18:53:33 cesarb: cool. How many ATZ's in a row can you send it...it's a bad hack, but trying until you get the ok or maybe 200 tries could be interesting Sep 09 18:53:44 cesarb, have you ever seen problems with uboot console? because i have... Sep 09 18:54:06 i should test more now that i have the latest uboot version Sep 09 18:54:11 tuukkah - You mean the usbtty console? Sep 09 18:54:19 Elrond, yes Sep 09 18:54:27 is it in any way related? Sep 09 18:54:36 tuukkah - You have the u-boot with this whole gsm pass through stuff? Sep 09 18:54:47 tuukkah: it always works fine for me (after I select USB on the uboot menu) Sep 09 18:55:02 (because all previous uboots have massive usbtty problems) Sep 09 18:56:05 * technil notes that minicom has a horrendous time, sometimes it executes a nand erase twice Sep 09 18:56:26 with problems i mean, the echo gets out of sync by one char, printenv output skips characters sometimes Sep 09 18:56:48 i have the same problems with the shipped uboot Sep 09 18:57:03 technil, same as me? Sep 09 18:57:15 tuukkah - That's normal for old uboots. Sep 09 18:57:30 * moko-bunny thinks dialer needs work -> "Not connected to network. Current status = Home network registered" Sep 09 18:57:37 tuukkah: ah, yeah, if for instance I try to paste a long line, it misbehaves, if that's what you are saying... Sep 09 18:57:41 tuukkah - And none of the uboots on buildhost have the fix. Sep 09 18:57:46 moko-bunny: you are late, I fixed that one yesterday Sep 09 18:58:04 moko-bunny: rwhitby sent my patch to mickey and it's already on SVN Sep 09 18:58:10 cesarb - Old news. laforge fixed that a week ago. ;) Sep 09 18:58:20 cesarb: hehehe...doh, well today's build dies for me on....guess what...everyone all at once "webkit" Sep 09 18:58:22 cesarb - At least he blogged on it. I did not try. Sep 09 18:58:33 Elrond: well, I'm using the august snapshot 2007.2 uboot... which is OLD Sep 09 18:58:45 Elrond, oh ok, am i safe to flash an uboot i just compiled? Sep 09 18:59:04 moko-bunny: webkit was dying yesterday night, today I managed to build it... Sep 09 18:59:12 moko-bunny: but that was earlier this morning Sep 09 18:59:12 cesarb - if you have a debug board AND know how to use use it: Maybe. Otherwise: NO Sep 09 18:59:16 trying again :) Sep 09 18:59:43 Elrond, i don't have the board :-( Sep 09 19:00:16 tuukkah - Then wait for a blessed uboot. Sep 09 19:01:24 s/cesarb/tuukkah/ <-- in my second pre-previous statement Sep 09 19:02:02 but this bug has been fixed so there's no open bug? Sep 09 19:06:15 tuukkah - http://bugzilla.openmoko.org/cgi-bin/bugzilla/show_bug.cgi?id=170 Sep 09 19:07:21 ok great Sep 09 19:25:56 cesarb: I think there is a possibility that there is some impact to the serial port, or the serial driver, when it is switched to/from the GSM modem; this might explain both the strange gsm modem behavior as well as the serial console behavior, both problem occurring immediately after the port is switched. Sep 09 19:26:31 mwester: well, it looks like it only happens in the first boot after the battery is replaced, so I'm more inclined to think it's a modem-slow-to-really-wake-up issue Sep 09 19:26:44 has anyone managed to send more then one word in an sms using libgsmd-tool? - or even multiline sms? - are there escape codes for space and/or newline? Sep 09 19:26:54 mwester: if I reboot (or even turn off, keeping the battery) it reacts fine Sep 09 19:27:10 mwester: of course, THEN half of the time the GUI status bar dies for no reason at all :P Sep 09 19:32:06 Ok. I think I'm going to see if I can hack in a reset of the serial driver after the port is switched, and see what happens. Sep 09 19:33:08 mwester: get the debug patch from bug 766, else you will not know when gsmd sends things to the port Sep 09 19:33:57 mwester: (the patch for fixing bug 766 would also be useful, since it avoids gsmd getting stuck on some situations) Sep 09 19:34:35 hi folkd Sep 09 19:34:37 folks Sep 09 19:34:52 so I have 20070908 rootfs with 20070909 kernel fic-gta01 2.6.22.5-moko11+svnr2937-r2 and it does not ask for PIN... Sep 09 19:36:45 Interesting, so CTS and RTS _aren't_ sent out to the debug board... Sep 09 19:36:55 Is the kernel serial console using XON/XOFF? Sep 09 19:37:51 hmm, according tohttp://lists.openmoko.org/pipermail/neo1973-hardware/2007-September/000235.html Sep 09 19:38:01 * technil thinks his builds lost that over a week ago hhf423.. but since it never actually worked.. Sep 09 19:38:14 http://lists.openmoko.org/pipermail/neo1973-hardware/2007-September/000235.html the SC32442 has only 64mb of ram? Sep 09 19:40:39 cb22 nope. there is an external chip connected Sep 09 19:40:47 ah, great Sep 09 19:40:55 lost what a month ago? Sep 09 19:41:26 hhf424 : dialer asking for a sim pin Sep 09 19:41:49 and only a week or so.. Sep 09 19:41:53 not a month Sep 09 19:42:49 well, see phase 1 software testing, it was ok before the new build from today Sep 09 19:43:50 yes, indeed my builds have some unexpected problems :) Sep 09 19:45:05 yours? Sep 09 19:45:17 cesarb: I've never encountered a serial console that uses any form of flow controll. Sep 09 19:47:16 hhf424 : yes, via mokomakefile on 9/01/07 and 9/05/07 Sep 09 19:53:36 technil: are you published Sep 09 19:54:06 publishing to builhost.openmoko? **** ENDING LOGGING AT Sun Sep 09 19:54:28 2007 **** BEGIN LOGGING AT Sun Sep 09 19:56:12 2007 Sep 09 20:01:18 doh Sep 09 20:09:50 Hey all Sep 09 20:25:52 ok, looks like I've managed to reliably make gsmd work (with the GUI) after all resets (but not on initial power up after putting the battery back in), without any powering off/resetting of the modem on /etc/init.d/gsmd or sleep tricks Sep 09 20:26:31 unfortunately, I cannot find out how to make it work on the initial powering up of the phone Sep 09 20:26:45 and the GUI status bar still tends to crash a lot Sep 09 20:26:58 cesarb: putting gsmd stop / start into matchbox/session works for me Sep 09 20:27:19 (script) wiki RecentChanges: [[Phase_1_Software_Testing]] [[Hardware:Neo1973:Alternate_Cases:Camera]] [[OpenMoko_Local_Groups:_Norway]] Sep 09 20:27:37 hhf423: but that's a kudge. I'm doing it _without_ any stop/start or poweroff/poweron dance Sep 09 20:27:38 cesarb: and the gui status bar does no longer crash on me. all with the rootfs from yesterday, kernel from today Sep 09 20:27:54 cesarb: ok, hats off :-) Sep 09 20:29:07 We need something which works every time... How can you call it a phone if gsmd only works sometimes? Sep 09 20:29:27 cesarb: my words exactly Sep 09 20:29:46 my status is: Dialer refuses to make calls, gives "Not connected to network. Current status = Home network registered" and refuses to make a call. Sep 09 20:29:57 hhf423: I fixed that one yesterday Sep 09 20:30:07 so the phone IS connected yet dialer says its not and won't dial Sep 09 20:30:25 hhf423: after gsmd started working well enough that I could get to the dialer without having to restart the GUI Sep 09 20:30:28 cesarb: then I need a newer build Sep 09 20:30:40 hhf423: no, the dialer was telling you it _was_ connected Sep 09 20:31:08 yes, dialer tells me I am connected and then refuses to dial because I am not connected Sep 09 20:31:12 hhf423: the test was something like "if (status != CONNECTED || status != ROAMING) { error_message(); }" Sep 09 20:31:25 hhf423: try to make a truth table for that if statement ;-) Sep 09 20:31:28 Dialer says: "Not connected to network. Current status = Home network registered" Sep 09 20:31:49 and then refuses to dial. that it. I can only be called :-( Sep 09 20:32:10 !ombug 807 Sep 09 20:32:25 * cesarb wonders where the bot went Sep 09 20:33:02 hhf423: anyway, I told the channel, rwhitby took the patch, grabbed mickeyl who was passing by and told him to commit the patch Sep 09 20:33:29 hhf423: so builds from today shouldn't have that bug Sep 09 20:33:39 cesarb: so I should be fine when a new image arrives on buildhost? Sep 09 20:33:59 It's a sad reflection of the state of gsmd that such a bug on the dialer went unnoticed for several days... Sep 09 20:34:07 or I should be fine with the existing image from today? Sep 09 20:35:18 hhf423: no idea... you could ask someone to build the openmoko-dialer2 package and send it to you Sep 09 20:35:41 hhf423: due to a couple of NATs, it's hard for me to send it to you (unless you have IPv6) Sep 09 20:36:02 cesarb: don't worry, I'll wait for the next build Sep 09 20:36:06 Any of you guys know anything about TurboSim? Sep 09 20:36:17 Is it possible that thing would help increase compatibility for those whose SIMs don't seem to work? Sep 09 20:39:05 Hm, is the "Ret" key on the new and improved terminal emulator supposed to act like Enter, or just sit there and do nothing when clicked? Sep 09 20:40:11 cesarb: enter Sep 09 20:40:25 cesarb: but you need to do ipkg install vte before Sep 09 20:40:58 (quick question: is power management any better? what's battery life like these days? worth using openmoko as a portable PIM?) Sep 09 20:41:19 no, no, no Sep 09 20:41:22 paulproteus: I believe PM is still in its infancy Sep 09 20:41:28 paulproteus: there is power management? Sep 09 20:41:33 there is little PIM stuff working Sep 09 20:41:45 Well, the calendar display even from 2007.1 wasn't so bad. Sep 09 20:41:59 SpeedEvil: ah, but there's _feedreader_, which is on task-openmoko-pim! Sep 09 20:44:20 is there anyone here that works with GPSd ? Sep 09 20:44:29 or on navigation in general Sep 09 20:45:01 hhf423: works... how come it isn't on the terminal's RDEPENDS? Sep 09 20:45:33 cesarb: my question exactly. for that dependency is known a while now Sep 09 20:45:48 cesarb: will you file a bug? Sep 09 20:45:54 ok, where should bugs in openmoko-terminal2 be filed for maximum effect? OM or OE? Sep 09 20:46:06 OM I would guess Sep 09 20:46:06 (since it's a bug on the .bb file...) Sep 09 20:46:13 (which is in OE) Sep 09 20:46:31 cesarb om i think Sep 09 20:46:39 roh: thx Sep 09 20:47:15 CathodioN : sort of, why Sep 09 20:49:11 oooh, a new screenshot on scap with sms tool Sep 09 20:51:06 hhf423 feel free, to use it: http://wiki.openmoko.org/wiki/SMSTool :-) Sep 09 20:53:36 kristian-m: ah, cool. will probably try it tomorrow Sep 09 20:54:00 kristian-m: need to make free for phyton again :-) Sep 09 20:54:42 hhf423 its a pity that python with gtk is 9 MB :-/ Sep 09 20:55:05 I put python on the card, using ipkg install -d card Sep 09 20:55:14 kristian-m: yes, lot of space on such a small phone :-) Sep 09 20:56:36 There's about 40MB of packages I have installed on the card Sep 09 20:57:29 gcb77 and then you add it to your $PATH? - hmm nice feature Sep 09 20:58:14 kristian-m: I used to do that, but rwhitby pointed me to ipkg-utils. Using ipkg-link it creates symlinks to your root for all the files in the package Sep 09 20:58:46 gcb77: how does this exactly work? Sep 09 20:59:11 gcb77: so you not only define 'card' in your ipkg.conf? Sep 09 21:00:17 I am getting: ipkg: Cannot create symlink from ./usr/bin/python-config to 'python2.5-config': Operation not permitted Sep 09 21:00:44 gcb77 could you create a wikipage for that? - just an example how to get it work - or do we already have that? Sep 09 21:05:10 hhf423: you need to re-format the card with ext2 Sep 09 21:05:18 aaah Sep 09 21:05:20 by default its vfat Sep 09 21:05:50 Hello, are there any planned changes to payment methods or shipping process? Sep 09 21:06:39 it looks like there are some people around here who would like to order neos in one batch, but finding credit card capable of paying $3000-$4000 is quite difficult:) Sep 09 21:07:11 gcb77: like mkfs -t ext2 -c /media/card ? Sep 09 21:07:46 oh, no mkfs Sep 09 21:07:48 hhf423: umount /media/card first Sep 09 21:08:02 hhf423: then mke2fs /dev/mmcblk0 Sep 09 21:08:10 hhf423: wait, no Sep 09 21:08:17 /dev/mmcblk0p1 Sep 09 21:08:18 got to partition it first Sep 09 21:08:36 why that? Sep 09 21:09:00 hmm, didn't even think about that... I just deleted and re-created the partition Sep 09 21:09:02 mmp other payment options are planned for the gta02 sale Sep 09 21:09:26 The partition is maked as VFAT now, you can probably just change the type Sep 09 21:09:36 roh: any ideas what they would be like? so we can prepare in the meanwhile :) Sep 09 21:09:43 then run mke2fs /dev/mmcblk0p1 Sep 09 21:09:54 (and not panic and look for someone with _big_ credit card :) Sep 09 21:10:48 roh: iirc, some time ago also shipping changes were announced, is there any (even tiny:) possibility that shipments will be done from Europe? Sep 09 21:11:31 mmp we will try to get distributors in every continent/bigger country Sep 09 21:12:19 roh: hmm, that's veery good news, thanks a lot:) Sep 09 21:13:12 I hope it will be possible to use the other payment methods ASAP :) Sep 09 21:13:46 * mmp would like little $450 present under his Christmas tree, but tries to be realistic about that Sep 09 21:15:23 mmp: where is 'here'? my cc can do 3000 USD :-) Sep 09 21:15:55 hhf423: slovakia Sep 09 21:16:13 hmm, not that far from you in fact:) Sep 09 21:16:26 mmp: thats not .cz, no? Sep 09 21:16:40 no, sorry Sep 09 21:16:43 hhf423: no, just country next to it:) Sep 09 21:16:45 its not, of course Sep 09 21:16:53 np:) Sep 09 21:17:39 the problem is of course pushing the money around Sep 09 21:17:40 BTW: Does the current 7.2 image have tools to send/receive files by BT? Sep 09 21:18:13 hhf423: but let's get surprised what will the other payment methods be... it is possible that the final amount will be >$3000... Sep 09 21:18:27 hhf423: hmm, right... . Sep 09 21:19:30 mmp: and the level of trust everyone would be willing to give. I wonder what shipping from .de to you would be? Sep 09 21:20:04 hhf423: :) yes. Well, it would then need to be re-shipped to Slovakia... Sep 09 21:21:26 another problem is that last time I've been doing international money transfer, the payment itself and the charges were roughly 50:50... Sep 09 21:21:35 now "Boot from SD" results in "undefined instruction" and a cpu reset Sep 09 21:21:49 hhf423: but that wasn't big amount after all, so with more money it could be more efficient Sep 09 21:22:18 to "slovakische republik" a DHL package with less than 5kg is ... Sep 09 21:22:39 mmp: would need to use paypal or something Sep 09 21:22:52 mmp: you're not eu yet, no? Sep 09 21:23:09 hhf423: we already are, at least to some extent:) Sep 09 21:23:34 so from this PoV it should be fine... Sep 09 21:23:53 mmp: so no tax stuff? Sep 09 21:24:17 hhf423: ahh, taxes, I forgot:) Sep 09 21:24:19 mmp: maybe direct bank to bank money transfer is then not so expensive? Sep 09 21:24:43 hhf423: friend just told me that the money transfer should cost no more than ~EUR 15,- Sep 09 21:24:46 ah, ok, so its 24 EURs Sep 09 21:24:58 for the shipping, maybe less Sep 09 21:25:04 or EUR 20,- :) Sep 09 21:25:40 mmp: we would need to test that somehow Sep 09 21:26:06 hhf423: but anyway, let's wait for what FIC announces... . if the distributors will be in both germany and czech republic, it would be more effective to just order separate. Sep 09 21:26:16 mmp: sure Sep 09 21:26:35 (I guess there's no chance to open distribution in Slovakia:) Sep 09 21:27:35 mmp: maybe if you find a phone / computer shop who has already experience in this stuff they would not mind Sep 09 21:27:50 hhf423: hmm, nice idea... Sep 09 21:28:35 I'll try to find out more info this week... Sep 09 21:29:31 hhf423: How can I contact you if I find something? I'll be idling here probably until GTA02 will be released (and probably after that, too:) Sep 09 21:29:50 mmp: but I am not sure what you will be really saving. a little bit of shipping maybe and a little bit of UPS carges or so Sep 09 21:30:01 mmp: in czech republic will be distribution of neo's? Sep 09 21:30:13 cegy_: no idea:) Sep 09 21:30:18 is anyone else getting:openmoko/build/tmp/staging/arm-angstrom-linux-gnueabi/lib/libWebKitGdk.so: undefined reference to `WebCore::callOnMainThread(void (*)())' ? Sep 09 21:30:30 mmp: I'll be here. Or you can leave a message in my talk page in the wiki Sep 09 21:30:38 cegy_: just slightly higher chance that it will be there than in Slovakia:) Sep 09 21:31:03 mmp: it would be great :) Sep 09 21:31:30 cegy_: well, many things would be, but I'm not affiliated with OpenMoko/FIC anyhow, those are just my rough guesses Sep 09 21:31:32 :) Sep 09 21:31:42 mmp: I'll probably order one or two of gta2 exta in any case. there will always be someone who would like to have one if its already in europe :-) Sep 09 21:32:21 hhf423: I guess that even if you order 10 times more, you will be able to get rid of them in just no time:) Sep 09 21:33:02 mmp: heh, but then again I already have a day job :-) don't want to add repackaging neo's to the list :-) Sep 09 21:33:07 looks like everyone here lurks for GTA01 or GTA02 :) Sep 09 21:33:18 hhf423: :-))) of course, I'm just joking Sep 09 21:33:25 sure Sep 09 21:33:55 * hhf423 installs python to the mmc Sep 09 21:34:20 * mmp looks on his ipaq with still-buggy SD controller :-( Sep 09 21:36:49 heh.. my pizza arrived Sep 09 21:37:57 hey -- does someone have the url handy for "getting started", once I have my neo1973 (which I do :) Sep 09 21:38:08 gcb77: if I do ipkg -d card install python, should the symlinks be added automatically? Sep 09 21:38:19 http://wiki.openmoko.org/wiki/Getting_Started_with_your_Neo1973 <-- answering my own question Sep 09 21:38:39 vallor: http://wiki.openmoko.org/wiki/Getting_Started_with_your_Neo1973 Sep 09 21:38:46 ah, there you go :-) Sep 09 21:38:51 maybe also http://wiki.openmoko.org/wiki/SH1 Sep 09 21:39:19 vallor: also have a look at phase 1 software testing for results and workarounds with existing images Sep 09 21:40:03 * vallor is off to get a sim from t-mobile... Sep 09 21:44:54 * hhf423 waves good night Sep 09 21:45:11 grrr Sep 09 21:45:32 why can't the whole image be buildable for more than a day ? Sep 09 21:46:08 how can I build a hello world package if I can't get make openmoko-devel-image to finish ? Sep 09 21:46:40 bitbake commands manually ? Sep 09 21:52:21 mwester: MokoMakefile has a specific target to change between neo and x86 builds Sep 09 21:54:27 rwhitby: the wiki says I should report a problem if a package doesn't get built... is a non-working webkit-gtk a known bug, or should I re-check what I'm doing ? Sep 09 21:55:10 polz clean the package and rebuild Sep 09 21:55:22 roh: tried it, didn't work. Sep 09 21:55:57 i buiĂlt it just some hours ago... have a log? Sep 09 21:56:22 i noticed i needed to restart the build many times to get some packages through with Sep 09 21:56:35 /home/polz/openmoko/build/tmp/staging/arm-angstrom-linux-gnueabi/lib/libWebKitGdk.so: undefined reference to `WebCore::callOnMainThread(void (*)())' Sep 09 21:56:36 | collect2: ld returned 1 exit status Sep 09 21:56:47 pleeeease, not a full rebuild... Sep 09 21:57:02 polz thats fixed in svn Sep 09 21:57:03 the last one took me from friday 'till this morning Sep 09 21:57:19 polz only the webkit package.. Sep 09 21:57:54 does mokomakefile fetch the latest svn if the last build didn't finish ? Sep 09 21:57:54 polz cd build && . ../setup_env ; bitbake -cclean webkit-svn Sep 09 21:58:02 polz bitbake does that. Sep 09 21:58:09 which gets called from mokomakefile Sep 09 21:58:32 wait a sec... Sep 09 21:58:42 webkit-svn or webkit-gtk ? Sep 09 21:59:28 webkit-gtk Sep 09 21:59:41 webkit-gtk.. sorry Sep 09 22:00:02 OK, just checking Sep 09 22:00:26 while we're on this topic, can you force-rebuild an openmoko-devel-image? Sep 09 22:00:36 Any chance we can get ipkg-utils included in the official repos? Sep 09 22:00:39 so ... clean seems to have worked Sep 09 22:00:54 i see it as an option in moko makefile Sep 09 22:02:00 so make update should report It's At revision 2946. , right ? Sep 09 22:02:09 it is one.. i just cannot remember the syntax right now Sep 09 22:02:39 gcb77: I'm adding it to task-openmoko-feed.bb right now Sep 09 22:03:03 polz just building it pulls the latest rev Sep 09 22:03:19 polz make update isnt even neccessary. atm it should even build!!1! ;) Sep 09 22:03:39 NOTE: Executing runqueue Sep 09 22:03:39 NOTE: Running task 3259 of 3321 (ID: 1468, /home/polz/openmoko/openembedded/packages/webkit/webkit-gtk_svn.bb, do_fetch) Sep 09 22:03:51 let's wait and see what happens Sep 09 22:03:57 how can i move my build to a new directory? Using mokomakefile update, monotone still looks for the database in the old place :( Sep 09 22:05:30 riot: you will need to remove the checkouts, and run make setup again Sep 09 22:05:32 riot : i have a feeling you might have to change some values in setup-env Sep 09 22:05:42 ok nm Sep 09 22:05:50 it's the _MTN/options file which is recording the db location Sep 09 22:06:02 and you will want to rebuild from scratch as well if you move things Sep 09 22:07:34 NOTE: package webkit-gtk-0.0+svnr25459-r25447-r1: task do_compile: started Sep 09 22:07:41 gcb77: pushed the addition - it should appear in the official feed the next time mickey|away triggers the build on buildhost. synergy and mplayer too. Sep 09 22:07:48 is the version correct here ? Sep 09 22:08:46 looks like we got back to buildability overnight. Sep 09 22:09:25 riot maybe there is a link, pointing to an old location (for org.openembedded...) Sep 09 22:09:46 rwhitby: so will i have to rebuild completely?! oh noes.. Sep 09 22:10:12 rwhitby: how do i remove them? Sep 09 22:10:24 riot: remove what? Sep 09 22:10:41 rwhitby: the checkouts.. (which?) Sep 09 22:11:04 make clobber-openembedded Sep 09 22:11:44 geez. this sucks. Hope i'll never have to move the environment again :( Sep 09 22:12:45 and now, to dust off the computer a bit.. then sleep Sep 09 22:13:05 You sleep on your computer? Sep 09 22:13:07 :) Sep 09 22:13:51 I added a big section to http://wiki.openmoko.org/wiki/Package_management on how to install packages to the memory card. It would be helpful to have someone review/correct it as needed. Sep 09 22:14:36 rwhitby: thank you for adding ipkg-utils, I've been using ipkg-link and it works great Sep 09 22:15:27 package webkit-gtk-0.0+svnr25459-r25447: failed Sep 09 22:15:44 again the same problem Sep 09 22:16:12 polz log? Sep 09 22:16:14 does mokomakefile really fetch the latest svn, even if the last build failed ? Sep 09 22:16:32 mokomakefile fetches nothing there Sep 09 22:16:32 roh: the error is the same as it was before. Sep 09 22:17:07 polz -rw-r--r-- 1 roh roh 5442 2007-09-09 15:56 webkit-gdklauncher_0.0+svnr25459-r25447-r1_armv4t.ipk Sep 09 22:17:26 built today.. so it seems possible ... lets find out why... Sep 09 22:17:53 polz just building it pulls the latest rev Sep 09 22:18:09 whafuck ? Sep 09 22:18:22 so it's probably in some dependancy Sep 09 22:19:37 but where ? I mean... it still fails to find `WebCore::callOnMainThread(void (*)())' when linking Sep 09 22:19:55 thats why i'm wondering Sep 09 22:20:54 here's what I find weird: Sep 09 22:21:07 i'll rebuilt that and see if it breaks here Sep 09 22:21:14 polz@mojastwar:~/openmoko$ make clean-package-webkit-gtk Sep 09 22:21:14 ( . ./setup-env && cd build && bitbake -c clean webkit-gtk ) Sep 09 22:21:17 ... Sep 09 22:21:20 gcb77 thanks :-) Sep 09 22:21:36 OE Build Configuration: ... Sep 09 22:21:50 DISTRO_VERSION = "P1-September-Snapshot-20070909" Sep 09 22:22:37 why is the version 20070909 ? Or is that OK ? Sep 09 22:23:26 http://build.webkit.org/ also says it builds Sep 09 22:24:20 this is really frustrating. I feel like an idiot. Sep 09 22:25:02 http://build.webkit.org/changes/2881 Sep 09 22:25:07 roh: when you rebuild it, which distro version does it report ? Sep 09 22:25:21 DISTRO_VERSION = "P1-September-Snapshot-20070909" Sep 09 22:28:00 (script) wiki RecentChanges: [[Advanced_End-user]] [[Package_management]] [[User:Hhf423]] [[OpenMoko_Local_Groups:_Trondheim]] [[OpenMoko_Local_Groups:_Norway]] [[SMSTool]] [[Phase_1_Software_Testing]] Sep 09 22:28:25 woo hoo, all SIM'ed up and ready to "t-mobilize"! coffee, here I come! Sep 09 22:29:22 roh: the function seems to be right there, in the file Sep 09 22:33:53 and again, it fails. Sep 09 22:33:57 ARGH! Sep 09 22:35:02 polz: "undefined function" is the error? Sep 09 22:35:49 technil: yeah, and the function is right there, _and_ the bloody package seems to build on roh's machine Sep 09 22:37:49 NOTE: package webkit-gtk-0.0+svnr25459-r25447: completed Sep 09 22:38:08 japp-- builds for me.. recently make update updated Sep 09 22:39:46 so it looks like Snapshot 20070829 with Linux fic-gta01 2.6.22.5-moko11 is the more functional snapshot, for having the thing boot and run (for userland software development) ? Sep 09 22:40:21 http://wiki.openmoko.org/wiki/Phase_1_Software_Testing#20070908_rootfs_with_20070909_kernel_fic-gta01_2.6.22.5-moko11.2Bsvnr2937-r2 Sep 09 22:42:58 NOTE: Unpacking /home/polz/openmoko/sources/WebCore_svn.webkit.org_.repository.webkit.trunk._25459_.tar.gz to /home/polz/openmoko/build/tmp/work/armv4t-angstrom-linux-gnueabi/webkit-gtk-0.0+svnr25459-r25447-r1/ Sep 09 22:42:59 NOTE: Unpacking /home/polz/openmoko/sources/WebKit_svn.webkit.org_.repository.webkit.trunk._25447_.tar.gz to /home/polz/openmoko/build/tmp/work/armv4t-angstrom-linux-gnueabi/webkit-gtk-0.0+svnr25459-r25447-r1/ Sep 09 22:43:06 this seems fishy to me Sep 09 22:49:12 nope.. thats fine. Sep 09 22:49:38 its multiple svn dirs which get pulled and of webkit and webcore the svn rev is used for the packagename Sep 09 22:50:53 what else could I be missing ? Sep 09 22:51:45 btw, I've now tried removing sources/WebCore* and am waiting for the build to finish ... again Sep 09 22:59:51 I wonder if Linus has one of these Sep 09 23:01:25 I doubt it. Judging by the interviews, he's pragmatic enough to wait for a working version. Sep 09 23:06:04 yep, still not working Sep 09 23:06:39 I'll try redirecting all the output to a file and reading at least a part of the crap it spews out while it's failing Sep 09 23:10:12 roh: could this be a problem: Sep 09 23:10:14 NOTE: preferred version 2.5 of glibc not available (for item virtual/arm-angstrom-linux-gnueabi-libc-for-gcc) ? Sep 09 23:14:15 polz: it has said that since the first time I've built it Sep 09 23:14:36 polz: it's just background noise Sep 09 23:17:55 polz: I've been getting the same build error for almost a day Sep 09 23:18:47 woo hoo, I'm booted! Sep 09 23:19:04 hmm... there seem to be some files left after make clean-package-webkit-gdk Sep 09 23:19:19 rm build/tmp/staging/arm-angstrom-linux-gnueabi/lib/libWebKitGdk.* Sep 09 23:19:25 let's see if this helps Sep 09 23:31:38 that seems to have been it. Sep 09 23:32:05 rwhitby, roh, cesarb, aking: rm build/tmp/staging/arm-angstrom-linux-gnueabi/lib/libWebKitGdk.* was the solution. Sep 09 23:32:32 root@fic-gta01:/etc/init.d$ ./gsmd start # this put the thing out into the weeds Sep 09 23:32:34 I'm guessing those files should be removed by make clean-package-webkit-gdk, but aren't Sep 09 23:33:23 vallor: that's a known bug in the kernel. The workaround is to edit /etc/init.d/gsmd and comment the lines where the gsm chip gets powered off Sep 09 23:33:36 polz: ah, thank you! :) Sep 09 23:33:52 polz: great! testing it now.. Sep 09 23:34:37 vallor: another workaround, fyi, is to disable serial console in the kernel command line Sep 09 23:34:59 vallor: it's bug 788 Sep 09 23:35:12 ah, I'll check the bugzilla then Sep 09 23:39:27 polz: it worked! Thanks - I was about to rm -rf build/tmp/* Sep 09 23:40:54 so I've been hearing that some SIM's aren't working? Sep 09 23:42:04 * technil would like info too Sep 09 23:42:18 if I try to dial out, it says that it's still registering with the network Sep 09 23:42:20 on that particular matter Sep 09 23:42:56 at least I'm ssh'ed in Sep 09 23:43:03 and I've got usb networking running Sep 09 23:43:14 (first time I'd set up masquerading ;P ) Sep 09 23:43:58 vallor trying libgsmd-tool is a way to start Sep 09 23:44:22 ah, thank you Sep 09 23:44:55 vallor: 'libgsmd-tool -m shell' should put you into a shell connected to gsmd Sep 09 23:45:03 h for help Sep 09 23:45:38 ah, gsmd exited when I tried to dial (restarting with the init script...) Sep 09 23:47:11 aking: w00t ! :) Sep 09 23:47:51 technil: I get this: Sep 09 23:47:55 L Sep 09 23:47:55 # List operators Sep 09 23:47:57 vallor: register first Sep 09 23:48:02 k Sep 09 23:48:04 vallor: press r Sep 09 23:48:11 okay Sep 09 23:48:11 then wait for it to ask you for a pin Sep 09 23:48:25 okay, waiting Sep 09 23:48:30 r Sep 09 23:48:30 # Register Sep 09 23:48:48 I see a typo ;P Sep 09 23:48:56 U Unregister from netowrk Sep 09 23:48:57 in the help Sep 09 23:49:56 polz: yes, the bitbake clean command does not modify the staging area. Sep 09 23:51:34 hmm -- if the SIM isn't inserted correctly, will it just sit there like it's doing? Sep 09 23:51:45 is #openmoko-devel invite only? Sep 09 23:52:07 vallor: try L, to see if you're picking up any networks Sep 09 23:52:20 k Sep 09 23:52:32 L Sep 09 23:52:32 # List operators Sep 09 23:52:35 and then it sits Sep 09 23:53:27 Q Sep 09 23:53:27 Signal strength Sep 09 23:53:27 Signal undetectable Sep 09 23:53:27 Error rate undetectable Sep 09 23:53:40 I don't think it is talking to the modem -- could that be it? Sep 09 23:54:27 vallor: check the SIM Sep 09 23:54:37 k Sep 09 23:55:21 did you do the O before the r? Sep 09 23:55:53 (O powers up the radio part, o powers it off) Sep 09 23:56:07 also check the /tmp/gsm.log for clues Sep 09 23:56:38 and unless you have the patch for bug 766, wait a bit after the O and before the r, else gsmd can get confused Sep 09 23:56:52 okay! Sep 09 23:56:56 I didn't do the O part Sep 09 23:57:08 the sim is definitely in there, it's t-mobile Sep 09 23:57:23 * polz is going to sleep for a while Sep 09 23:57:35 thank you polz!! Sep 09 23:57:40 vallor: np. Sep 09 23:57:41 and thank yoiu cesarb! :) Sep 10 00:00:38 bootup takes a while Sep 10 00:03:24 hmm, it seems to have stopped with the wm startup, thermometer full -- and plugging in usb didn't get me my usb0 ip interface Sep 10 00:06:53 just rebooted w/factory resteet Sep 10 00:06:56 reset Sep 10 00:08:40 vallor: you just got the bug 788 crash, probably Sep 10 00:08:55 vallor: happens all the time, and is very annoying Sep 10 00:09:29 vallor: you can either comment out the powering off of the modem on /etc/init.d/gsmd, or go to uboot and remove the serial console from the kernel command line Sep 10 00:10:14 okay -- if worse comes to worse, I can just dfu the thing all over, right? Sep 10 00:13:32 vallor: the root fs image and kernel, yes you can "re-dfu" them. to change the env variable that passes options to the kernel, you just need to be able to get to a uboot console Sep 10 00:14:10 okay, I'm at the uboot console -- not sure how to get it to the kernel options though Sep 10 00:14:46 this is uboot 1.2.0 Sep 10 00:14:54 I didn't flash uboot, should I? Sep 10 00:16:35 vallor: no need to worry, we're not doing anything permanent Sep 10 00:16:47 vallor: you just have to printenv, and change one of the variables Sep 10 00:16:58 vallor: since we will not saveenv, it's not permanent Sep 10 00:17:19 vallor: you have to change (with setenv) the bootargs_default (IIRC) to remove the console=ttySAC0,115200 part Sep 10 00:17:33 oh, I should connect to the serial-over-usb device to get at it? Sep 10 00:17:37 vallor: then you chose the "boot" menu option as usual Sep 10 00:17:39 vallor: yep Sep 10 00:17:43 ah, k Sep 10 00:18:09 we need a userspace u-boot environment modifier ... Sep 10 00:18:16 /dev/ttyACM0 i believe it shows up as Sep 10 00:18:16 vallor: if you don't saveenv, the changes disappear when you reboot, so if you get it wrong you only need to reboot... Sep 10 00:18:24 k Sep 10 00:18:32 rwhitby: we already have one, in perl, in the devirginator directory of the svn repo... Sep 10 00:19:04 I think u-boot actually has one - it's just not packaged Sep 10 00:19:43 vallor: make sure the device is unplugged from the usb cable when you boot into u-boot and select serial over USB Sep 10 00:20:09 if you want a tested to be working-uboot... Sep 10 00:20:25 http://people.openmoko.org/roh/u-boot-gta01bv4-1.2.0+git20070909+svnr2943-r0.bin Sep 10 00:20:26 okay Sep 10 00:20:35 atleast it has workin acm and boots kernels Sep 10 00:21:02 hmm, k Sep 10 00:21:20 here's what I ended up with: Sep 10 00:21:22 minicom: cannot open /dev/ttyACM0: No such file or directory Sep 10 00:21:30 now I'm booting w/out usb connected Sep 10 00:22:06 okay, set to usb -- now plug it in? Sep 10 00:22:13 yup Sep 10 00:22:23 then try minicom Sep 10 00:23:35 dag nabbit, now the new ttyACM0 isn't showing up Sep 10 00:23:43 # dmesg Sep 10 00:23:43 usb 7-2.1.2: device descriptor read/64, error -110 Sep 10 00:23:50 it's okay, I'll just re-dfu it Sep 10 00:25:17 the kernel? Sep 10 00:26:37 that usb message came from dmesg on my workstation] Sep 10 00:27:35 oh I've seen this before Sep 10 00:27:41 (script) wiki RecentChanges: [[Main_Page]] [[User:Jon_Phillips]] [[Advanced_End-user/es]] Sep 10 00:30:35 okay: Sep 10 00:30:37 minicom: cannot open /dev/ttyACM0: No such file or directory Sep 10 00:30:45 but /dev/ttyACM0 is definitely there Sep 10 00:31:05 at least I can do the dfu Sep 10 00:31:58 I think I do want to change uboot -- I'm sick if hitting [aux] to keep the thing from shutting off ;P Sep 10 00:35:03 got a console via the ttyACM0 !!!! Sep 10 00:36:34 vallor take the one i posted. at least i know it works for acm Sep 10 00:38:11 setenv works... Sep 10 00:38:22 okay, do I just say "boot" now? Sep 10 00:38:37 setenv bootargs_base rootftype=jffs2 rot=/dev/mtdblock4 consol=tty0 logevel=8 Sep 10 00:38:51 root, not rot Sep 10 00:38:57 also console Sep 10 00:39:03 uh Sep 10 00:39:09 dang Sep 10 00:39:17 I did a cut-paste, didn't type it in Sep 10 00:39:18 thats why you should use a new uboot. the old one is eating chars on the serials/usb-acm Sep 10 00:39:19 one sec Sep 10 00:39:31 okay, I'll do that then Sep 10 00:45:46 okay, u-boot is on it, starting Sep 10 00:45:50 ...fingers crossed... Sep 10 00:46:17 should be built today Sep 10 00:46:38 yay, I'm on the console again Sep 10 00:48:25 booting... Sep 10 00:48:31 roh: you can always type it by hand, one character at a time... Sep 10 00:48:34 ScaredyCat: you there? Sep 10 00:48:38 trying out your proxy dns Sep 10 00:48:42 roh: (which is what I do) Sep 10 00:48:44 got a small patch for you Sep 10 00:49:12 ScaredyCat: still can't get a ping outside of my host box though using your script...tracking down Sep 10 00:49:13 bootargs_base=rootfstype=jffs2 root=/dev/mtdblock4 console=tty0 loglevel=8 Sep 10 00:49:16 is what it said Sep 10 00:49:39 cesarb that bug is fixed. with current uboot pasting should be no problem Sep 10 00:49:41 it got past the thermometer -- and it's playing the startup sound Sep 10 00:50:19 roh: I will wait until the next "official" uboot snapshot, since a broken uboot would be a pain (even with the debug board nearby) Sep 10 00:50:35 * cesarb wonders why his neo doesn't play any startup sound anymore Sep 10 00:50:39 cesarb i flashed that one myself moments ago. Sep 10 00:50:52 cesarb else i wouln't put it on people. Sep 10 00:51:11 roh: that's the most basic of the possible levels of testing ;-) Sep 10 00:51:31 okay, I'm back in libgsm-tool -- just did "O" Sep 10 00:52:12 L Sep 10 00:52:12 # List operators Sep 10 00:52:22 dang, that's a pain Sep 10 00:52:44 maybe I won't be dialing out today ;P Sep 10 00:54:56 vallor: L only works after r for some reason Sep 10 00:55:05 vallor: and only if you actually registered to the network Sep 10 00:56:41 k Sep 10 00:56:45 I did "r" first Sep 10 00:56:52 I might need to do something on tm's web site Sep 10 00:58:00 rejon: which proxy dns are you referring to? Sep 10 00:58:20 ''Your T-Mobile number was not found in our system. Please be sure you entered the number correctly. If you are a new customer, it can take up to 72 hours before you can register. Please try again later.'' Sep 10 00:58:35 k, no phone fun for me today! hehe Sep 10 00:58:52 vallor: how did that appear on /tmp/gsm.log? Sep 10 00:58:58 I'll look\ Sep 10 00:59:04 vallor: or was it a voice message? Sep 10 00:59:05 rwhitby: dnrd Sep 10 00:59:05 and his gta01 script Sep 10 01:00:34 that was on the web site Sep 10 01:00:42 rwhitby: arggggg....must be missing some setting Sep 10 01:00:43 when I gave it my phone number Sep 10 01:01:09 http://ponzo.net/gsm.log Sep 10 01:01:18 there's the log (before gsmd croaked) Sep 10 01:04:37 vallor: that's a common "stuck gsmd" log Sep 10 01:05:08 I just checked -- the radio wasn't on (using the proc file to check for it being on) Sep 10 01:05:18 I just turned it on with "echo", now trying again with "r" Sep 10 01:05:20 vallor: I've been trying to make gsmd get stuck less often... so far, it works for me on every reboot (but not on the initial boot) Sep 10 01:05:44 vallor: you have to do the 'O' even after the echo, because the echo enables only the AT interpreter part Sep 10 01:05:53 oic Sep 10 01:06:43 oo Sep 10 01:06:45 !!! Sep 10 01:06:55 EVENT: Signal Quality: 14 Sep 10 01:06:55 EVENT: Signal Quality: 99 Sep 10 01:06:55 EVENT: Netreg registered (home network) LocationAreaCode = 0x00CF CellID = 0x36ED Sep 10 01:06:55 EVENT: Signal Quality: 12 Sep 10 01:07:00 stuff is happening!!! Sep 10 01:07:05 hehe Sep 10 01:07:11 vallor: that's the normal output after registering to the network Sep 10 01:07:38 but "L" doesn't show anything Sep 10 01:07:45 oh here it goes again: Sep 10 01:07:50 vallor: L takes some time (a few minutes) to react Sep 10 01:07:57 ah, okay Sep 10 01:08:00 EVENT: Netreg registered (home network) LocationAreaCode = 0x00CF CellID = 0x36F7 Sep 10 01:08:00 vallor: since it has to scan all frequencies Sep 10 01:08:25 this is very exciting, I must say... Sep 10 01:09:02 vallor: I know. It also was for me when I first managed to make it answer (and later dial) Sep 10 01:09:15 :) vallor: new toy? Sep 10 01:09:36 (of course, it should swearword work directly on first try from the GUI, with no command line hacking needed!) Sep 10 01:09:41 absolutely -- if I get it to dial out, I'm going to call my biz partner and tell him his iphone sux ;P Sep 10 01:10:09 ERROR reding from gsm_fd Sep 10 01:10:18 shucks Sep 10 01:10:19 :( Sep 10 01:10:22 gsmd quit Sep 10 01:11:44 if it does it again, I'll strace it Sep 10 01:12:23 # List operators Sep 10 01:12:24 31026 T-Mobile, T-Mobile for short, is our current operator Sep 10 01:12:24 310380 AT&T Wireless, AT&T for short, is available Sep 10 01:13:00 EVENT: Incoming SMS stored at location 1 Sep 10 01:13:55 but the libgsm-tool is segfaulting when I try to read the sms Sep 10 01:15:01 ah Sep 10 01:15:09 "sr=1" is the magical incantation Sep 10 01:15:23 "Welcome to T-Mobile! To change your plan, pay your bill or check your minutes go to my.t-mobile.com. Your mobile number is 707xxxxxxx" Sep 10 01:15:31 guys is a neo actually necessary to help? Sep 10 01:15:59 i'd like helping squashing bugs and stuff.... but i don't have a neo... /yet) Sep 10 01:16:22 r3pek: there is actually an emulator for qemu Sep 10 01:16:25 r3pek: you can also use the qemu Sep 10 01:16:38 that's what i'm trying to setup Sep 10 01:16:46 but it's refusing to build on amd64 :( Sep 10 01:16:56 r3pek: I'm on amd64 Sep 10 01:17:14 i suppose you're not runing gentoo are you? ;) Sep 10 01:18:02 r3pek: no, ubuntu Sep 10 01:18:21 r3pek: the gentoo guys are the only ones who always have trouble building it Sep 10 01:18:53 r3pek: I don't think it being amd64 has anything to do with the problem... Sep 10 01:19:10 i'm having relocation problems doring compilation Sep 10 01:19:12 *during Sep 10 01:19:18 can't understand why.... :/ Sep 10 01:19:29 did you built it with gcc-4? or 3? Sep 10 01:19:42 or just apt-get?! :X Sep 10 01:20:12 r3pek: qemu automatically detects your gcc-3 compiler (except on gentoo), and uses it (you must, however, use gcc-4 for everything else, or you'll get other different funny problems) Sep 10 01:20:17 Is direct.openmoko 's webmaster around? Sep 10 01:20:34 I keep getting the following error when trying to order: "Can't use an undefined value as an ARRAY reference at /usr/local/share/perl/5.8.8/Net/UPS.pm line 240." Sep 10 01:21:06 r3pek: the trick (on every system but gentoo) is to install both gcc-4 and gcc-3 (usually gcc-3.4), using gcc-4 by default and letting qemu automatically use gcc-3.x Sep 10 01:21:13 jeremymc: What are you trying to do? Sep 10 01:21:32 cesarb: i'll try to force it.... brb Sep 10 01:21:33 jeremymc you have to enter the state as 2 letter uppercase abbreviation if youre in the us or canada Sep 10 01:21:35 jeremymc: that means you didn't read the warning saying that on a few countries you must always use 2-digit abbreviations for your state Sep 10 01:21:51 * cesarb wonders if the warning is still there Sep 10 01:22:19 cesarb i think its a bit too hard to find in all the informations the wiki holds Sep 10 01:22:38 roh: the warning wasn't on the order page itself? Sep 10 01:23:10 * cesarb doesn't recall if he saw the warning on the order page or just in one of the several FAQs on the wiki) Sep 10 01:23:13 s/)// Sep 10 01:23:41 OK. Given that provinces in Canada use 3 letters as the short form. Sep 10 01:25:06 Thanks. There is no mention of the 2 letters on the order page. Sep 10 01:26:01 btw, I clicked on the wireless antenna in matchbox, and matchbox-panel-2 died Sep 10 01:26:18 vallor: clicking on the panel usually makes it die ;-) Sep 10 01:26:24 vallor: when it doesn't die by itself Sep 10 01:26:24 Maybe add some kind of error when it is expecting 2 letters... Sep 10 01:26:27 heh, k Sep 10 01:28:40 I just dialed out with libgsmd-tool 's "D" command -- worked, but there was no sound on the neo Sep 10 01:29:04 cedric: how's that? it compiled! :D Sep 10 01:29:28 vallor: you need to dial with the GUI dialer Sep 10 01:29:42 k, here goes Sep 10 01:29:43 vallor: since gsmd by itself doesn't setup the sound mixer Sep 10 01:30:05 vallor: you can also setup the mixer by hand... look it up on the wiki for the correct alsa state file name Sep 10 01:30:31 http://wiki.openmoko.org/wiki/Neo1973_Audio_Subsystem Sep 10 01:30:38 ah, thanks Sep 10 01:30:49 * cesarb the first time dialed by hand but "dummy dialed" with the dialer (which wasn't talking to gsmd since I had started gsmd by hand) to make the sound paths be set up Sep 10 01:32:55 k, the dialer says "currently waiting for network registration" (but the gsm applet shows three bars) Sep 10 01:32:59 vallor: root@fic-gta01:~$ ls /etc |grep .state Sep 10 01:33:34 gsmhandset.state ? Sep 10 01:34:30 builtin mic and speaker Sep 10 01:41:52 ah laddie, that worked! Sep 10 01:43:47 okay, I can call out and take a call (but I'm doing it with libgsmd-tool) Sep 10 01:44:16 * technil wishes he could do the same Sep 10 01:44:44 is it not working for you, technil ? Sep 10 01:52:40 thank you very, very much for your help, technil Sep 10 01:53:21 and all you guys -- thank you over and over again :) Sep 10 02:06:33 vallor: if you restart gsmd in /etc/matchbox/session, you should be able to call after boot without going to the command line (assuming a recent build) Sep 10 02:07:14 hmm, okay, will try Sep 10 02:12:58 hmm, seems to be stuck on reboot Sep 10 02:13:22 (script) planet: Daniel Willmann: Chaos Camp buildup https://www.totalueberwachung.de/blog/articles/2007/08/03/chaos-camp-buildup Daniel Willmann: GUADEC recap https://www.totalueberwachung.de/blog/articles/2007/08/02/guadec-recap Daniel Willmann: See you at GUADEC https://www.totalueberwachung.de/blog/articles/2007/07/15/see-you-at-guadec Daniel Willmann: Can I has a Neo? https://www.totalueberwachung.de/blog/articles/2007/07/06/can-i-has-a-neo Sep 10 02:18:56 vallor: might be bug 788, especially if you didn't do saveenv after changing the bootargs .. IIRC passing the kernel option is lost over reboot Sep 10 02:19:31 I was just going back through and doing saveenv Sep 10 02:19:45 :( quite a pain Sep 10 02:23:35 k, it's booting Sep 10 02:23:41 I have thermometer :) Sep 10 02:23:48 yay Sep 10 02:27:17 okay, I'm signed in with ssh Sep 10 02:27:42 (script) wiki RecentChanges: [[Main_Page/es]] [[External_Feeds]] [[OpenMoko/it]] [[OpenMoko]] [[Booting_from_NFS]] [[My_Account]] [[Commercial_models]] [[Vertical_applications]] [[Business_Development]] Sep 10 02:33:52 oh man, the kernel has nfs support! Sep 10 02:34:10 that's awesome, I can use that for strace output w/out burning ram :) Sep 10 02:34:30 shoo it off over the network... Sep 10 02:37:27 what is /media/net used for, is that a good place to put the nfs mount point? Sep 10 02:37:48 rwhitby or ScaredyCat: did you reply to me? Sep 10 02:38:24 rejon: what was the question? Sep 10 02:39:43 oh, was talking about using dnrd to proxy to moko Sep 10 02:39:48 vallor: not listed in /etc/fstab ... Sep 10 02:40:02 technil: yes, and I see no sign of an automounter Sep 10 02:44:24 * vallor is looking for the magic ipkg name to add joe, his beloved editor since 1992 Sep 10 02:54:04 aloril: your planet script is outputting old entries... **** ENDING LOGGING AT Mon Sep 10 02:59:57 2007