**** BEGIN LOGGING AT Wed Sep 07 02:59:56 2005 Sep 07 03:07:13 zaurus-a: so whats the problem? Sep 07 03:20:39 03koen 07org.oe.dev * r04029fed... 10/classes/package_ipk.bbclass: classes/package_ipk.bbclass: put HOMEPAGE into the Homepage: control field Sep 07 03:22:31 brb Sep 07 03:23:38 re Sep 07 03:31:53 03koen 07org.oe.dev * r8517a140... 10/packages/abiword/ (abiword-2.3.5/cdump-hack.patch abiword_2.3.5.bb): abiword: add 2.3.5 Sep 07 04:10:52 anyone got bvdd source code? Sep 07 04:16:08 time to commit few more strange things Sep 07 04:20:49 cd .. Sep 07 04:20:50 cd .. Sep 07 04:21:08 * XorA is an idiot Sep 07 04:21:13 XorA: no such file or directory Sep 07 04:33:53 03hrw 07org.oe.dev * r90a67b27... 10/packages/ (4 files in 2 dirs): Sep 07 04:33:53 reversed (R)DEPENDS in opie-freetype and opie-ttf-support: Sep 07 04:33:53 - opie-ttf-support does not need opie-freetype to run as it only create fontdir Sep 07 04:33:53 - opie-freetype need opie-ttf-support because it create fontdir for it Sep 07 04:33:57 03hrw 07org.oe.dev * r715f4d8c... 10/conf/machine/collie.conf: Sep 07 04:33:57 collie.conf: removed few not needed kernel modules: Sep 07 04:33:57 - registers as it has unknown license Sep 07 04:33:58 - devinfo as /proc/deviceinfo is not needed anymore for SDL Sep 07 04:34:00 - gpio as it is not needed Sep 07 04:34:02 03hrw 07org.oe.dev * rdf55dfa1... 10/conf/machine/handheld-common.conf: Sep 07 04:34:04 handhelds-common.conf: removed uinput kernel module Sep 07 04:34:06 - currently only kbdd use it so I add it there Sep 07 04:42:33 hrw|work: use RRECOMMEND for uinput in kbdd Sep 07 04:42:57 it already has it Sep 07 04:43:21 don't forget to push the change Sep 07 04:43:40 and I *really* don't like that it wasn't discussed on the mailinglist Sep 07 04:44:01 touching *core* files should *always* be discussed on oe@ or bugzilla Sep 07 04:44:44 ok - will next time sorry Sep 07 04:45:42 koen: kbdd which is in repo has it already - no change from me Sep 07 04:46:07 otherwise it will look like 'team opie' is butchering core files without discussion Sep 07 04:46:19 *cough* alwin *cough* Sep 07 04:47:47 hmm.. ok. you got right - I'll revert it Sep 07 04:48:32 monotone q: does disapprove fail when a merge with that cset has happened? Sep 07 04:49:06 dont know Sep 07 04:49:13 koen: find it out ;) Sep 07 04:49:45 I think a merge causes the 'cannot invert' message Sep 07 04:49:58 * koen wants empty merges in monotone Sep 07 04:51:13 zecke: any news about the freenx .bbs? Sep 07 04:52:26 changeset disaproved, pushed Sep 07 04:52:31 sorry for that Sep 07 04:52:49 koen: forgot to ask :} Sep 07 04:53:08 but I think it is unlikely Sep 07 04:53:09 03hrw 07org.oe.dev * r95ce8a8d... 10/conf/machine/handheld-common.conf: disapproval of revision df55dfa120c7e309713e729dcd552f9e207de952 Sep 07 04:53:32 koen: we need to do it ourselves Sep 07 04:57:31 03koen 07org.oe.dev * rd009f64e... 10/packages/abiword/abiword_2.3.5.bb: abiword 2.3.5: add libgnomeprint dependency Sep 07 05:00:36 * koen pushes some abandonware Sep 07 05:01:11 * hrw|work cleans repo Sep 07 05:02:19 03koen 07org.oe.dev * r2faf6c31... 10/packages/nonworking/freenx/ (16 files): packages/nonworking/freenx/: add freenx suite to nonworking Sep 07 05:03:52 zecke: I've pushed my try at adding freenx Sep 07 05:04:59 hrw|work: I've forwarded you the mail from the viewmtn author Sep 07 05:06:46 koen: I saw that - thx Sep 07 05:08:05 koen: can you do full (from scratch) build for 16M device and other with target 'opie-image meta-opie-all' with my recipes? Sep 07 05:08:31 koen: testing it on my machines will take 2-3 days from scratch Sep 07 05:09:04 as soon as this build is fnished Sep 07 05:09:09 ok Sep 07 05:09:30 I will add recipes into bug Sep 07 05:09:44 koen: I hope to have more time soon ;) Sep 07 05:14:17 koen: http://bugs.openembedded.org/show_bug.cgi?id=307 Sep 07 05:14:28 morning folks Sep 07 05:14:32 mickeyl: hey Sep 07 05:14:35 hi mickeyl Sep 07 05:14:45 hey mickeyl Sep 07 05:14:46 mickeyl: bug #307 needs testing Sep 07 05:15:00 03koen 07org.oe.dev * rddbf7177... 10/packages/nonworking/monotone-viz/monotone-viz_0.11.bb: packages/nonworking/monotone-viz/monotone-viz_0.11.bb: add monotone-viz to nonworking till we have a ocaml crosscompiler Sep 07 05:15:48 hrw|work: *nod* Sep 07 05:18:50 depmod: unknown symbols in prism2_usb: Sep 07 05:18:53 depmod: msleep Sep 07 05:18:54 *sigh* Sep 07 05:19:00 msleep debuted in kernel 2.4.29 Sep 07 05:19:07 ~kill 2.4.18 Sep 07 05:19:10 * ibot shoots a magneto-ionized proton gun at 2.4.18 Sep 07 05:19:19 tried to backport: Sep 07 05:19:25 warning: Sep 07 05:19:26 HZ undeclared Sep 07 05:19:31 MAX_JIFFY_OFFSET undeclared Sep 07 05:19:36 TASK_UNINTERRUPTIBLE undeclared Sep 07 05:19:44 shcedule_timeout undeclared Sep 07 05:19:46 grrrrrrr Sep 07 05:20:17 mickeyl: 2.4.xx-lineo is so beauty... I dont understand why you dont like it... ;P Sep 07 05:20:31 i know. everything more current is fluff Sep 07 05:20:50 lineo 2.4 is best in class Embedded Linux Kernel Sep 07 05:29:07 but who is best in 2.6? Sep 07 05:30:46 hrw|work: montavista is best in class AFAIK Sep 07 05:31:01 mickeyl: poor guy Sep 07 05:34:09 * mickeyl s/msleep/mdelay/ Sep 07 05:35:20 Progress - cpufreq is now loaded as a module (compiled in still - I couldn't get it to compile as a module - if anyone has any ideas I'd be happy to hear them) Sep 07 05:35:35 However it doesn't work - there's no /sys entry for it Sep 07 05:36:36 hah *strike* Sep 07 05:36:44 damn module loads Sep 07 05:40:14 ~shoot prism2_usb just because Sep 07 05:40:16 * ibot shoots prism2_usb just because in the head with a spitwad! Sep 07 05:41:23 libetpan 0.38 support gnutls so opie-mail can stop depend on openssl Sep 07 05:42:28 hmm, I echo'ed 1 into the sys entry for the debug parameter for cpufreq, but it's had no effect - I imagine cpufreq's already been tried so it will have no effect. I also tried suspending & resuming, no messages either. Adding printks now. Sep 07 05:42:32 lardman|work: no idea on my site... i'm a kernel illiterate Sep 07 05:42:32 iirc a kernel module needs to register every sysfs node with the kernel Sep 07 05:42:32 you might want to add printk's before and after the respective calls Sep 07 05:42:32 + checking for return values Sep 07 05:42:50 03mickeyl 07org.oe.dev * rc1c5f457... 10/packages/wlan-ng/ (2 files in 2 dirs): wlan-ng-modules 0.2.2: make it work on stoneage kernels Sep 07 05:43:16 yep, that's the next step - I made a typo in one of the patches which is why I was banging my head against a brick wall last night - the module was never included in the kernel Sep 07 05:44:17 hrw|work: that's cool. Sep 07 05:44:54 ~praise wlan-ng-modules for finally, yeah _finally_, indeed i mean FINALLY supporting a fair amount of the Wireless Extension Standard Sep 07 05:44:56 All hail wlan-ng-modules for finally, yeah _finally_, indeed i mean FINALLY supporting a fair amount of the Wireless Extension Standard! Sep 07 05:45:15 wow.. Sep 07 05:45:30 even iwlist wlan0 scan Sep 07 05:45:35 can't believe it :) Sep 07 05:45:36 mickeyl: so finally it suxx less? Sep 07 05:45:40 ah ok, i believe it Sep 07 05:45:43 read the Changelog Sep 07 05:45:52 80% of changes submitted by Pavel Roskin Sep 07 05:45:56 Mr. Orinoco/Spectrum Sep 07 05:45:59 ;) Sep 07 05:47:00 Depends: liblockfile (>= 1.06), libgnutls11 (>= 1.0.23), libgnutls-openssl11 (>= 1.0.23), Sep 07 05:48:32 that's nice Sep 07 05:48:34 really Sep 07 05:48:34 nice Sep 07 05:48:42 die OpenSSL die Sep 07 05:49:02 2005-03-12 - libetpan-0.36cvs4 - hoa Sep 07 05:49:05 support for GNUTLS. Sep 07 05:49:06 Thanks from Rajko Albrecht. Sep 07 05:49:10 FYI: 3.5.4/Opie|tosa looks good enough to not worry about a release. now lets take a look at GPE Sep 07 05:49:32 hrw|work: thanks 'from' alwin? Sep 07 05:49:46 mickeyl: my tosa will be handed over on the 11th Sep 07 05:49:51 zecke: good to hear Sep 07 05:49:51 mickeyl: tosa 2.4 has apm problem which was fixed on collie? Sep 07 05:49:58 mickeyl: I quoted changelog Sep 07 05:50:14 hrw|work: tosa apm problems should be fixed by your patch Sep 07 05:50:18 (same as poodle) Sep 07 05:50:21 ok Sep 07 05:50:26 can you test it? Sep 07 05:50:27 thanks again, btw. Sep 07 05:50:31 i will Sep 07 05:52:42 morning pb_ Sep 07 05:52:48 pb_: hey Sep 07 05:53:50 pb_: hey do you think you will make gcc 4.0.2 bbfiles any time soon Sep 07 06:07:21 hail mickeyl Sep 07 06:07:22 hail zecke Sep 07 06:07:43 zecke: I don't have any immediate plans to do that. Sep 07 06:13:46 pb_: what is stopping you from doing that? Sep 07 06:16:49 03koen 07org.oe.dev * rdac5598f... 10/packages/abiword/abiword_2.3.5.bb: Abiword 2.3.5: add libgnmoeprintui dependency Sep 07 06:23:10 zecke: just sloth Sep 07 06:24:00 ah you give shelter to the sloth-bear as well= Sep 07 06:24:02 ? Sep 07 06:26:13 yeah, it 0wns me. Sep 07 06:26:59 I've been finding it a bit difficult to work up any enthusiasm for oe lately. Sep 07 06:29:12 Good good, I have cpufreq compiling as a group of modules, time to add more printks then test Sep 07 06:32:12 03hrw 07org.oe.dev * r444d1a1c... 10/packages/libetpan/ (libetpan-0.38/etpan_suffix.patch libetpan_0.38.bb): Sep 07 06:32:12 added libetpan 0.38 - based on patch supplied by Frederic Devernay Sep 07 06:32:12 - this version can be build with GNU TLS so I switched to it Sep 07 06:32:12 - another package does not need OpenSSL Sep 07 06:32:12 - closes bug #304 Sep 07 06:36:03 I'll try to build gnutls 1.0.25 (latest 1.0.xx) Sep 07 06:36:27 * koen discovers building from scratch yields interesting results Sep 07 06:36:55 as usual Sep 07 06:37:14 pb_: understandable Sep 07 06:40:37 03koen 07org.oe.dev * ra42933a8... 10/packages/abiword/abiword_2.3.5.bb: abiword 2.3.5: add libgsf dependency Sep 07 06:41:54 * mickeyl can't follow Sep 07 06:43:28 mickey|sun: wise decision ;) Sep 07 06:47:57 03hrw 07org.oe.dev * r120e5d12... 10/packages/gnutls/ (3 files in 2 dirs): added gnutls 1.0.25 Sep 07 06:54:51 hrw|work: Did you see my ipkg woes? Sep 07 06:55:29 yep Sep 07 06:55:54 and depends didn't work either Sep 07 06:56:58 I tested only ipkg-native Sep 07 06:57:34 now installing mc on c7x0 with ipkg 0.99.153 - will look 1st time on device Sep 07 06:57:51 does not work here too Sep 07 06:58:05 mickey|sun: http://www.trolltech.com/campaign/devdays/agenda/europe.html Sep 07 06:58:15 mickey|sun: specially the embedded track is laughable... Sep 07 06:58:24 mickey|sun: Zack Rusing and the future of embedded graphics... Sep 07 06:59:35 zecke: who is he? Sep 07 06:59:48 koen: building from scratch killed my build machine :-( well looks like you are catching the abiword dependancy problems anyway Sep 07 06:59:54 hrw|work: some kde guy hired by TT Sep 07 07:00:04 hrw|work: but he never finished any of the tasks he started ;) Sep 07 07:00:08 aha Sep 07 07:00:17 zecke: that's embedded Sep 07 07:00:22 leaving out features Sep 07 07:00:52 zecke: everything zack doesn't finish is bloat of fluff Sep 07 07:01:02 NAiL: oops. I had 0.99.153-r0 on device - upgrading to r1 Sep 07 07:01:13 koen: right, doing the first 10% is good enough, as long as you get publicity and a job... Sep 07 07:03:20 NAiL: works with 0.99.153-r1 Sep 07 07:03:36 will pastebin in a minute Sep 07 07:04:16 off Sep 07 07:04:17 odd Sep 07 07:04:24 153-r1 didn't work at all on mine Sep 07 07:04:28 NAiL: http://pastebin.ca/22296 Sep 07 07:05:57 03koen 07org.oe.dev * r301eec9b... 10/packages/gaim/gaim_2.0.0-cvs-mco1.bb: gaim 2.0: s/PV/PN/ in cvsdate Sep 07 07:09:27 Good evening Sep 07 07:09:41 hi Sep 07 07:15:27 03koen 07org.oe.dev * r5302bee6... 10/conf/distro/preferred-gpe-versions-2.7.inc: conf/distro/preferred-gpe-versions-2.7.inc: update gaim PV Sep 07 07:45:44 good morning kergoth Sep 07 07:45:47 hey Sep 07 07:47:40 kergoth: hey Sep 07 07:47:41 hi kergoth Sep 07 07:48:57 ~emulate peter Sep 07 07:48:59 Chris, everything I say is a lie. Except that. And that. And that. And that. And that. An Sep 07 07:49:33 jbowler-zzz: the alsa patch strsignal is in cvs Sep 07 07:52:07 mithro: still working on btsco? Sep 07 07:52:13 koen: yeah Sep 07 07:52:31 i can't figure out why it can't find the sound device Sep 07 07:53:09 btsco would make a nice addition for VoIP on a pda Sep 07 07:53:16 yes Sep 07 07:53:27 it means you can use a bluetooth headset with your PDA :P Sep 07 07:56:07 i think i have endian or alignment issues Sep 07 07:56:16 or both Sep 07 07:57:47 yeah Sep 07 07:58:08 i can't test it on my zaurus because i don't have a bluetooth card Sep 07 07:59:02 don't zauri have bluetooth builtin? Sep 07 07:59:14 * koen whistles innocently Sep 07 08:00:51 * XorA curses bluetooth Sep 07 08:01:25 XorA: good Sep 07 08:02:07 mithro: I'd lent you mine, but it's a socket revE Sep 07 08:02:13 the one thing I cannot find is a card people say works with linux and is sensible price Sep 07 08:04:11 XorA: the bluemonkey one? Sep 07 08:04:37 koen: dont mind which, in uk there doesnt seem to be anything other than belkin available Sep 07 08:04:51 XorA: or sucky socket revE Sep 07 08:06:00 XorA: http://www.expansys.com/product.asp?code=ARM-BLCF Sep 07 08:06:21 the version *without* the pcmcia adapter is more expensive Sep 07 08:06:40 expansys seems like a pretty sweet place. just ordered a cell phone from them yesterday Sep 07 08:06:58 ou just have to keep an eye on the price Sep 07 08:07:22 some items are ridiciously more expensive at expansys Sep 07 08:07:25 Bernardo: which means? Sep 07 08:07:28 * kergoth nods Sep 07 08:08:28 koen: been waiting months for them to actually stock that though Sep 07 08:08:33 koen: maybe this time :-) Sep 07 08:08:45 anyway time to go to backup storage site Sep 07 08:11:11 mithro: 230kbps max, no suspend/resum support Sep 07 08:11:13 mithro: it is bluetooth 1.0, no sco Sep 07 08:12:07 I've bought from expansys too, their daily specials are (sometimes) cheap Sep 07 08:13:31 * kergoth picked up http://www.expansys-usa.com/product.asp?code=116371 Sep 07 08:16:04 I bought an eBuyer value BT CF card. Sep 07 08:16:12 I've had it working on the Z. Sep 07 08:18:26 They don't seem to list it any more though. Sep 07 08:18:29 Bernardo: ahh okay Sep 07 08:18:45 kergoth: that one looks nice... I am trying to find a affordable sony k750i Sep 07 08:22:48 yah, virtually any cheap cf bluetooth card should be fine. Sep 07 08:22:57 * kergoth kicks svn Sep 07 08:23:21 * pb_ stabs ieee 802.11 Sep 07 08:25:31 cya later Sep 07 08:25:43 later zecke_ Sep 07 08:29:37 bye zecke_ Sep 07 08:41:05 RP: where in /sys/ I find touchscreen? Sep 07 08:42:54 currently /dev/sharp_ts has to be created by hand Sep 07 08:43:55 on 2.6, sane touchscreen interfaces use the input system Sep 07 08:44:08 which input node a touchscreen is, depends on the actual model Sep 07 08:44:12 on our zaurii, it's event1 Sep 07 08:44:24 mickeyl: ah Sep 07 08:44:42 ok. so its got created Sep 07 08:44:49 udev is weird sometimes Sep 07 08:45:19 especially because it isn't correctly started by our script Sep 07 08:45:36 I replaced it with changed debian script Sep 07 08:45:41 ah that's better Sep 07 08:45:43 also replaced some rules Sep 07 08:45:57 are you sure its rules are loaded? Sep 07 08:46:03 mickeyl: there's no way to garantee it's event1 Sep 07 08:46:14 a user might load an extra module at boot Sep 07 08:46:27 or compile a different kernel, etc Sep 07 08:46:28 koen: udev rule can fix it Sep 07 08:46:34 we should really have a script look through the event interfaces and ask them if they're a touchscreen Sep 07 08:46:35 heh Sep 07 08:46:43 kergoth: detect-stylus Sep 07 08:46:44 or yeah, udev could do it, good point Sep 07 08:46:46 koen: uhm... this information wasn't generic Sep 07 08:46:48 * kergoth nods Sep 07 08:46:51 i was talking about our zaurus kernels Sep 07 08:46:57 mickeyl: me too Sep 07 08:46:58 not any other configurations any other people might do Sep 07 08:47:14 on our distributions it's event1 for the zaurus Sep 07 08:47:20 there's nothing you can do about it Sep 07 08:47:25 http://pastebin.ca/22302 is touchscreen on c760 Sep 07 08:47:36 mickeyl: load the key driver later as the ts driver Sep 07 08:47:49 koen: how so? we compile touchscreen and keyboard into the kernel Sep 07 08:48:07 mickeyl: we do, but who know what users will do Sep 07 08:48:25 or the input people upstream Sep 07 08:48:25 if users change it and break it, that isnt something you shoudl be supporting Sep 07 08:48:32 mickeyl: or other devels... XorA boots from /dev/mmcblk0p1, RP tries to boot from /dev/hda1... Sep 07 08:48:50 but yeah, a udev rule to create /dev/input/ts or something based on a sysfs attribute would be less failure prone, since it wouldnt be based on registration order Sep 07 08:48:51 kergoth: hardcode assumptions are bad, no matter what users do Sep 07 08:48:54 kergoth: http://handhelds.org/cgi-bin/cvsweb.cgi/gpe/base/detect-stylus/detect-stylus.c?rev=1.13&content-type=text/x-cvsweb-markup Sep 07 08:48:57 * kergoth nods Sep 07 08:49:03 ok. reboot - will look how it will start Sep 07 08:49:14 remove the last 20 lines if X is against your religion Sep 07 08:49:47 I still don't know what we are talking about Sep 07 08:50:01 are we trying to solve a general problem of finding out where a touchscreen is no matter what kernel version we ahve? Sep 07 08:50:08 or are we talking about 2.6 and zaurus only Sep 07 08:50:11 or what? Sep 07 08:50:22 I think we (you) are talking about 2.6 only, but all platforms. Sep 07 08:51:11 aha. excellent. then where's the problem? :D Sep 07 08:51:15 afaik, there is no good way to detect the touchscreen device under 2.4, other than finding out the machine type and consulting a lookup table. Sep 07 08:51:43 indeed. there's no standard mechanism. too many different device knodes and device paths Sep 07 08:51:53 but, well, the code has already been written to do that for all the current 2.4 platforms, and there are unlikely to be any more, so that isn't a big deal. Sep 07 08:51:54 hmm.. reflashing husky made it boot faster - no jffs2 errors Sep 07 08:52:17 can we assume that all ipaqs,toshibas, and whatever pda's and pads we mainly support under 2.6 use the input system? Sep 07 08:52:50 mickeyl: The order of the keyboard and touchscreen devices under 2.6 is just a coincidence and in theory could change at any moment :-/ Sep 07 08:53:22 mickeyl: It comes back to driver registration order and we shouldn't really depend on it... Sep 07 08:53:22 RP: is that so? then i hope that the userland people will be informed about that Sep 07 08:54:08 mickeyl: There is nothing the kernel can do to prevent that and tools like udev are there to deal with it Sep 07 08:54:43 Having said that, I don't plan to alter the order so it will only change if the kernel init code gets rearranged Sep 07 08:55:16 mickeyl: I did some investigating with direct booting of the microdrive on spitz and those warning unable to open initial console errors Sep 07 08:55:24 then again, when using tslib I still don't think that having to deal without a device node is the proper abstraction. since tslib already knows about everything something like tslib_get_fileno() to add into a select loop and then tslib_get_value() would make more sense than having to deal with the ts device as well Sep 07 08:55:41 s/without/with/ Sep 07 08:55:48 (first occurance) Sep 07 08:56:09 RP: any findings yet? Sep 07 08:56:18 Abrstracting a find me a touchscreen function into tslib sounds good. You'd have to work out how to deal with multiple devices though Sep 07 08:56:35 tslib_get_fileno would be flawed Sep 07 08:56:40 someone want to add lsb-base into OE? Sep 07 08:56:43 mickeyl: The /dev driectory appears to get trashed somehow. All the files in it disappear, hence the warning Sep 07 08:56:50 RP: ah. no /dev/console Sep 07 08:56:59 since tslib is free to add or remove events to the stream, that select() could return when you have no events, or hang when you have an event Sep 07 08:57:03 but yeah, something like that would be nice Sep 07 08:57:07 They're there in hdimage1.tgz and not there when the system boots Sep 07 08:57:35 mickeyl: The other strange thing is it jumps straight into runlevel 5 Sep 07 08:57:54 runlevel S and friends don't get run - hence no keymap amongst other things Sep 07 08:58:43 Having looked at the updater.sh code, I don't know how the dev entries disappear :-/ Sep 07 08:58:51 RP: Its not at the point that udev remounts /dev as /.dev and puts the ramfs over the top, is it? Sep 07 08:58:56 I need to confirm they do straight after a reflash Sep 07 08:59:13 RP: patched the updater.sh yet do drop you into a shell after flash? Sep 07 08:59:22 s/flash/extracting the image/ Sep 07 08:59:40 mickeyl: yes, I think it is safe to assume that all 2.6 devices use the input system Sep 07 08:59:47 pb_: that's good to know. Sep 07 08:59:48 really Sep 07 08:59:56 * gridzero was just trying to figure out the lack of rcS himself.... Sep 07 08:59:57 so, afaik, there shouldn't really be any problem Sep 07 09:00:02 hi Sep 07 09:00:09 my enthusiasm to support 2.4 is lowering in a linear fashion Sep 07 09:00:21 I agree with you that having to tell tslib about the device node is a bit sucky. Sep 07 09:00:21 gridzero: where? on my debian desktop and on Z I dont have /.dev/ or any other that kind of mount Sep 07 09:00:26 gridzero: I wondered about that but mount shows no ramfs mounted at this point. The kernel itself is complaining about the lack of a /dev/console so its long before udev gets started Sep 07 09:00:36 bleh Sep 07 09:00:49 quite Sep 07 09:00:55 absolutely Sep 07 09:01:11 yay my warranty on my ibook is over and I havae nice line on the screen Sep 07 09:01:13 pb_: Is there any need for the h3600 /dev/input/ts0 device under 2.6 now? Sep 07 09:01:25 RP: there has never been any need for that under 2.6 Sep 07 09:02:03 pb_: Who still wants that there then? Sep 07 09:02:09 Nobody, as far as I know. Sep 07 09:02:29 * RP is tempted to suggest its removal Sep 07 09:02:44 afaik, both the tslib and tslibdev drivers are historical relics of no value. Sep 07 09:02:49 mickeyl: Patching updater.sh to drop to a shell it probably the nest step Sep 07 09:02:52 er, tsdev and tslibdev Sep 07 09:03:37 but, who knows, there are a lot of crazy people out there. maybe someone is using them. Sep 07 09:03:50 pb_: It was patched by zap in 2004... Sep 07 09:04:26 ...? Sep 07 09:05:11 pb_: i'm curious about your enthusiasm problem. please don't tell me the SCM is responsible for that... Sep 07 09:07:11 * mickeyl nominates "there are a lot of crazy people out there" as quote of the month Sep 07 09:09:41 looks like rfcomm does not have sysfs support Sep 07 09:11:42 mickeyl: ... most of them wearing military uniforms of one kind or another... Sep 07 09:13:50 hmm Sep 07 09:14:19 the dangerous people are not those with the uniforms Sep 07 09:14:25 we can recognize these Sep 07 09:14:42 I'm trying to modprobe my cpufreq module (which I've got compiled as a stand-alone module), however it gives me the following error: insmod: cannot insert '/lib/...': Unknown symbol in module (-1): No such file or directory Sep 07 09:14:56 ok. /dev/rfcomm0 created automatically Sep 07 09:15:03 Does this mean I'm missing a symbol, or that I'm missing a symbol because it's looking for another module to insert? Sep 07 09:15:23 lardman|work: does dmesg tell you more about that? any missing symbols upon depmod -ae? Sep 07 09:15:39 lardman|work: modinfo cpufreq Sep 07 09:15:57 thanks both, dmesg has tol me Sep 07 09:16:03 s/tol/told Sep 07 09:17:11 I'm not sure it's happy being a stand-alone module: It's after various things, one of which is 'sysfs_create_file' which I presume ought to exist elsewhere Sep 07 09:19:00 has anyone got 'mysql' to build using bitbake? Sep 07 09:19:31 mickeyl: I don't think it's due to any one thing, more a confluence of minor factors. The SCM is certainly part of that, in the sense that making casual patches is rather more painful than it used to be, but I'm sure I could put up with that if I were otherwise sufficiently enthused. Sep 07 09:19:34 I get: 'checking for atomic_add... configure: error: cannot run test program while cross compiling' using X86 as my type Sep 07 09:20:05 cu all Sep 07 09:20:25 dwildes: you probably need to add a site file entry for that. Sep 07 09:20:46 I'm not sure exactly what you mean by "using X86 as my type". What is your TARGET_ARCH? Sep 07 09:21:13 pb_: ok, thanks. Sep 07 09:22:15 pb: I'm just using 'machine = x86' and 'target_os = linux'. that's what I meant by type, sorry for confusion Sep 07 09:22:41 the target_arch isn't set, because it said it would read it from the 'machine' attribute, according to local.conf Sep 07 09:23:19 dwildes: okay. x86.conf sets TARGET_ARCH to "i386", so you need to inspect site/i386-linux. Sep 07 09:24:01 ahhh :) Sep 07 09:24:15 if you look at i586-linux, it has some "mysql" lines in it that i386-linux doesn't. Sep 07 09:24:27 I suspect that, if you copy those into the i386 version, it will start to work. Sep 07 09:24:43 I'll see what I come up with. Thanks a billion pb!! Sep 07 09:25:05 yep - you're right, i386 doesn't have anything for atomic_add Sep 07 09:31:15 koen|away: wait, one question, you told me the sl6000 orange mud patch is in the kdrive bugtracker. have a link? i didn't find it Sep 07 09:31:27 (the bug, not the tracker) Sep 07 09:32:29 03nail 07org.oe.dev * r1776da2d... 10/packages/meta/openslug-packages.bb: openslug-packages: Remove duplicate icecast. It should only be in BROKEN, since it currently is. Sep 07 09:40:41 If a compiled-in module fails to find a symbol (function) does it fail silently? Sep 07 09:40:49 nah Sep 07 09:40:57 you can't link the zImage in that case Sep 07 09:41:04 right Sep 07 09:41:31 I've put loads of printks in cpufreq, and it's loaded (in /sys/modules) but nothing in dmesg Sep 07 09:42:10 (I've gone back to the compiled-in version as I'd no idea how to get rid of the missing symbols errors for the stand-alone version) Sep 07 09:48:01 night all Sep 07 09:49:41 hey guys Sep 07 09:49:54 hey MSpin. ltns. how ya been? Sep 07 09:50:03 howdy MSpin Sep 07 09:50:04 it's been good...busy as hell Sep 07 09:50:38 I finally convinced these people that embedded linux is the way to go Sep 07 09:51:18 and the plan is to use OE Sep 07 09:51:28 woot Sep 07 09:51:30 good news Sep 07 09:52:35 I got this totally sexy PXA255 board on my desk, hehe Sep 07 09:52:41 it's the size of a SO-DIMM Sep 07 09:52:47 nice Sep 07 09:53:24 anyway....I'm trying to get OE to build.. Sep 07 09:53:58 libtool-native-1.5.10-r3 configure is bailing with "configure: error: source directory already configured" Sep 07 09:54:07 that's in the wiki Sep 07 09:54:11 frequentproblems or so Sep 07 09:54:23 * MSpin looks again Sep 07 09:54:26 iirc you should prevent using links to your oe build dir Sep 07 09:55:36 hmm i can't find it either :/ Sep 07 09:55:53 yeah, if there are symlinks in the path, a canonicalization of the path wont match the path, which cauess things to blow up Sep 07 09:55:56 dont recall why, but.. Sep 07 09:55:58 fyi, I've been playing with udev-068, extras/start_udev, and the gentoo rules file, and it's working reasonably well. startup time is much better since (I think) it's not doing hotplug for all ttys, etc. We'd still need to add rules for MMC etc. though, and I'm now seeing some strangeness where usbnet isn't coming online until I remove/reattach the cable Sep 07 09:56:05 d0h...I forgot /home was a symlink to /data/home Sep 07 09:56:19 mreimer: excellent (most of it) Sep 07 09:56:27 thanks! I'll add that to one of the wiki pages Sep 07 09:57:05 mreimer: desktop w/ 2.6.13 distros are talking about 40% faster startup as wel Sep 07 09:57:34 that sounds about right. I think we might even get more when we dump hotplug entirely Sep 07 09:57:42 *nod* Sep 07 09:58:08 has anyone in here gotten a Cxx00 with a 2.4 kernel to connect to a desktop with a 2.6 kernel via USB? I can't get the usb0 device to show up on my desktop whatever I do. Sep 07 09:58:25 it sees a device but no networking Sep 07 09:58:36 (yes, I modprobe'd usbnet) Sep 07 10:00:06 03jbowler 07org.oe.dev * rb532e7ba... 10/packages/uclibc/uclibc-0.9.28/ucslugc/uClibc.config: Sep 07 10:00:06 uclibc: ucslugc config 0.9.28 updated Sep 07 10:00:06 The config now sets the processor option to XScale, correct for ucslugc Sep 07 10:00:06 (which is NSLU2) Sep 07 10:00:10 03jbowler 07org.oe.dev * r5d9c9e1e... 10/packages/binutils/ (3 files in 2 dirs): (log message trimmed) Sep 07 10:00:10 binutils: patch 2.16 for uclibc, enable 2.16 on uclibc builds Sep 07 10:00:10 2.16 now includes the patches which ensure that HOST_CC_ARCH is passed to Sep 07 10:00:11 the build, other than that this commit should only affect binutils for Sep 07 10:00:13 linux-uclibc targets. The patches which enable recognition of Sep 07 10:00:15 linux-uclibc*' as well as 'linux-gnu*' have integrated - in fact there are Sep 07 10:00:17