**** BEGIN LOGGING AT Tue Oct 28 02:59:57 2008 Oct 28 04:34:45 *yawn* Oct 28 04:36:09 mmmm Oct 28 05:58:15 Hey Oct 28 05:58:21 Ainulindale, already awake? Oct 28 06:06:28 moin mickeyl Oct 28 06:10:51 *yawn* Oct 28 06:10:52 morning Oct 28 06:15:36 morning mickeyl Oct 28 06:15:41 alphaone, already here? Oct 28 06:17:17 mickeyl, would you like to fix a little bug in RetrieveMessage or should I wait for alphaone? Oct 28 06:18:30 which bug? Oct 28 06:18:34 alphaone sleeps, i gues Oct 28 06:18:34 s Oct 28 06:19:08 (sleep) I bet Oct 28 06:20:07 mickeyl, RetrieveMessagebook() returns a timestamp in the message properties, RetrieveMessage() does not. Try it with cli-frameworkd you don't believe me ;) Oct 28 06:22:04 have a patch already? :) Oct 28 06:22:51 no, but I think it should be trivial, one moment Oct 28 06:24:31 mickeyl: have you looked at android rild, I think it is apache license, or could there be other problems? Oct 28 06:25:56 no idea. i have not looked at Android, it's completely uninteresting to me Oct 28 06:26:14 sorry, don't mean ui stuff, modem interface daemon Oct 28 06:26:29 ah Oct 28 06:26:33 no, didn't look at that Oct 28 06:26:36 could take a look Oct 28 06:26:43 but then again, our modem interface is done Oct 28 06:26:48 so i don't know what we should gain Oct 28 06:27:10 didn't know if you were still looking a c or vala or staying python Oct 28 06:27:14 (and no, please lets not jump into the next Python vs. C war) Oct 28 06:27:43 i'm not actively looking for an ogsmd replacement atm., no. Oct 28 06:27:44 plus calypso has quirks Android shurely won't cope with out of the box Oct 28 06:28:02 the plan is to move some frameworkd services to Vala, yeah Oct 28 06:28:05 but surely not the ogsmd Oct 28 06:28:09 which would gain us nothing Oct 28 06:28:16 since the modem is slow Oct 28 06:28:25 but instead would make the code very complicated Oct 28 06:28:53 (calypso quirks) yes, that too Oct 28 06:29:02 but i'd hope they have a modem abstraction Oct 28 06:29:02 calypso is the firmware name? we are looking at omap device and finding very simlilar modem on the dsp core Oct 28 06:29:14 calypso is the chip from Texas Instruments Oct 28 06:30:54 omap, dsp core, modem??? Oct 28 06:31:16 doesn't go together Oct 28 06:31:47 omap850, two cores with modem on one core, modem is essentially same calypso chip, as it's understood Oct 28 06:32:34 hmm :-/ Oct 28 06:33:20 mickeyl, I don't see an obvious bug. Maybe decodeSMS has some problems with parsing the date, but I don't know. Oct 28 06:33:28 there is some code from e28 phone gpl source, that is for omap730 share memory interface, found by ali1234 on linwizard Oct 28 06:33:50 sorry, thank you Oct 28 06:36:37 quickdev: ok, lets wait for daniel then Oct 28 06:36:46 yeah, ok Oct 28 06:37:49 freesmartphone.org: 03mickey 07framework * re8ad25f2e36e 10/framework/subsystems/ogsmd/modems/abstract/mediator.py: ogsmd: 'dir' is a builtin in Python, don't shadow it Oct 28 06:38:13 tmzt: which firmware and tools are provided by TI for this alegedly calypso-compatible function block? Oct 28 06:39:26 tmzt: or which docs & specs at least? Oct 28 06:39:57 to us?, none Oct 28 06:40:25 python has weird vfs built in, doesn't it? Oct 28 06:41:38 Now would be a good time if we could get a firmware update for the #1024 TI calypso no deep sleep issue Oct 28 06:42:34 we have a new firmware version in OM internal tests Oct 28 06:42:45 tmzt: virtual file system in Python? not that i know of Oct 28 06:43:26 we're also about to publish FLUID for Neo to flash out in the field Oct 28 06:43:44 DocScrutinizer, testing towards #1024 issue? Oct 28 06:43:45 finally ! yeah :-))) Oct 28 06:44:07 !mong others Oct 28 06:44:16 great! Oct 28 06:44:21 wow, I found it rather weird to have to send back a device for flashing when wm just keeps(kept) the radio fw in nand Oct 28 06:45:09 yeah, OM got a new wizard for calypso fw - Dieter Oct 28 06:45:17 who is working on gtk?? frontend for fso? Oct 28 06:45:43 tmzt, SHR is working for a frontend for fso, but we're using EFL, not GTK, it's better :) Oct 28 06:46:08 ok, it's just not zhone (python), so I wasn't sure if it was efl or gtk Oct 28 06:46:31 how hard would it be to adapt to different screen sizes/no touchscreen? Oct 28 06:46:34 DocScrutinizer2, would be fantastic if dieter is able to fix it :) Oct 28 06:46:39 (when it's ready) Oct 28 06:47:13 tmzt, should be no problem I think Oct 28 06:47:16 Werner, Dieter and humble me are focused on it Oct 28 06:47:30 tmzt, maybe you want a different theme file...but that's trivial Oct 28 06:48:05 the status bars/tray icons/gsm status, etc. are in seperate app now? Oct 28 06:48:17 but illume can stack them in the right order? Oct 28 06:48:37 the status things are illume modules Oct 28 06:48:42 what order? Oct 28 06:48:51 DocScrutinizer2, are you optimistic? Oct 28 06:48:59 I mean vertically, taking the whole width of the screen Oct 28 06:49:04 horizontally Oct 28 06:49:20 there are vectors to tackle it Oct 28 06:50:02 though we're missing important parts of TI sources Oct 28 06:50:23 DocScrutinizer2: Should I wait for FLUID or patch the software to make it not enter deep sleep? Oct 28 06:50:37 and TI support is EOL Oct 28 06:51:05 TeLLuS, atm, I'm using AT%SLEEP=2 as a startup command in frameworkd, because I always suffered from the bug Oct 28 06:51:08 do both I'd suggest Oct 28 06:51:21 DocScrutinizer2, EOL? End of life? End of line? ;) Oct 28 06:51:55 quickdev: sounds like i should make it an option Oct 28 06:52:07 mickeyl, yes, would be great Oct 28 06:52:12 simply Evaporated Outof Live Oct 28 06:52:14 calypso_recamping_handling = conservative Oct 28 06:52:25 calypso_recamping_handling = optimistic Oct 28 06:52:29 or something like that Oct 28 06:52:32 * mickeyl notes to TODO Oct 28 06:52:36 yeah Oct 28 06:52:56 mickeyl: :-) Oct 28 06:53:12 quickdev: How do I do that with om2008.9? Oct 28 06:53:38 TeLLuS, don't know, use a FSO based distribution ;) Oct 28 06:54:41 quickdev: I'm trying to make qtopia-x11.. and put it in there. Oct 28 06:56:19 you'd like to port the qtopia stuff to use frameworkd? are you interesting in developing new applications that make use of frameworkd and EFL? Oct 28 06:59:27 quickdev: I just want to be able to use the phone.. and just insert the command somwere. Maybe could use the mux for that also. Oct 28 07:00:14 TeLLuS, I'd only know how to do that on SHR (FSO based) Oct 28 07:01:45 mickeyl: U see 1024? Oct 28 07:02:02 can reproduce it? Oct 28 07:02:03 everywhere Oct 28 07:02:08 well Oct 28 07:02:15 reproducing sounds a bit too "active" for #1024 Oct 28 07:02:20 you just let it sit there Oct 28 07:02:21 and it starts Oct 28 07:02:24 like to test moko1p please Oct 28 07:02:27 ? Oct 28 07:02:28 and wait 30 seconds, hehe Oct 28 07:02:49 moko10 fw Oct 28 07:02:50 I'd like to test it, too, if it's not too difficult to flash Oct 28 07:03:01 DocScrutinizer2: sure Oct 28 07:03:26 although it's too early to see it Oct 28 07:03:31 I'll send binary to U Oct 28 07:03:36 i can test @ rush hour :) Oct 28 07:03:39 ok, cool Oct 28 07:04:03 I have no idea (yet) how to flah Oct 28 07:04:13 guess U know Oct 28 07:04:27 ah, yeah Oct 28 07:04:35 time to hack, ehehe Oct 28 07:05:20 quickdev: we're testing OMinternally for new. Please wait another few days Oct 28 07:05:35 now Oct 28 07:05:46 yeah, no problem :) Oct 28 07:05:58 but inform us about your results! ;) Oct 28 07:06:29 sure, what do U think we're working for ;-) Oct 28 07:09:08 hmm, where's the SHR recipes again? Oct 28 07:10:13 mickeyl, want to build shr? Oct 28 07:10:51 later Oct 28 07:10:56 first, the recipes need to go upstream Oct 28 07:11:01 then you guys should get OE access Oct 28 07:11:03 etc. Oct 28 07:11:18 mickeyl, recipes are in svn://daria.forty-two.fr/shr/trunk/ Oct 28 07:11:21 thanks Oct 28 07:11:34 but we're going to move to git Oct 28 07:11:58 mickeyl: U got mail: moko10-beta1 Oct 28 07:12:17 oki, will report Oct 28 07:12:24 thanks :-) Oct 28 07:13:49 mickeyl: you're usaing ARM-FLUID or LV-RS232 via hsjack to flash? Oct 28 07:13:56 arm-fluid Oct 28 07:14:00 never flashed any other way Oct 28 07:14:02 kewl! Oct 28 07:14:44 how long does it take to setup a new firmware with appropriate tools? Oct 28 07:14:47 could U publish ARM-FLUID binary and a howto? We got permission from TI to publish Oct 28 07:14:57 hmm Oct 28 07:15:34 not sure about the sources, but for sure we may distribute binary for Neo Oct 28 07:16:10 Wolfgang wants to speed up this Oct 28 07:16:14 heh Oct 28 07:16:25 then perhaps i should not do it Oct 28 07:16:37 lol, why? :-) Oct 28 07:16:51 just a bunch of more important tasks Oct 28 07:17:15 anyways, first i need to test it Oct 28 07:17:23 k, so I'll look for someone else who knows how to use it Oct 28 07:17:28 before we have something interesting and tested to give to the users, they don't need the tool anyways Oct 28 07:17:48 no reason to rush, really Oct 28 07:17:50 we waited for so long Oct 28 07:17:53 or rather... them Oct 28 07:18:00 we should distribute the tool and MOKO6 at least, for GTA01 users Oct 28 07:18:20 hmm Oct 28 07:18:23 i don't trust moko6 Oct 28 07:18:35 no experience with it Oct 28 07:18:42 lol, I dont trust calypso FW at all Oct 28 07:18:59 i'd rather release 9-beta1, which has at least been tested in some 01s Oct 28 07:19:26 I don't see the test reports on moko9-beta1 anywhere Oct 28 07:19:39 heh, same as the other non-existing test reports :D Oct 28 07:19:41 also we dn't have the sources for moko9 anymore :-( Oct 28 07:19:49 why does it matter? Oct 28 07:20:11 i know 9b1 works, i tested this quite a while. i can't say the same for 6 Oct 28 07:20:20 nobody knows *what* exactly *is* moko9 Oct 28 07:20:23 but it's not my call anyways Oct 28 07:21:19 so *you* _tested_ moko9? Well so let's bublish it for the devices U tested it for, as soon as possible. If U think it's worth being published Oct 28 07:21:35 *Publish Oct 28 07:21:39 yo Oct 28 07:22:29 btw., no firmware arrived yet Oct 28 07:22:32 mickey@openmoko.org? Oct 28 07:22:34 toss me some pointers to pipermail_HowTo plus files and I'll place it on my people/ Oct 28 07:22:52 hmm Oct 28 07:22:56 you really want to rush it Oct 28 07:22:59 but ok Oct 28 07:23:08 Fwd: Moko10 firmware image Oct 28 07:23:12 Von: Oct 28 07:23:13 Joerg Reisenweber Oct 28 07:23:13 An: Oct 28 07:23:13 Michael Lauer , Wolfgang Spraul Oct 28 07:23:13 Datum: Oct 28 07:23:13 2008-10-28 08:09 Oct 28 07:23:18 hmm, k Oct 28 07:23:20 underway then Oct 28 07:23:46 https://people.internal.openmoko.org/stefan/fluid_0.0+svn20070817-r2_armv4t_eabi.ipk Oct 28 07:23:47 well is some 2.5MB Oct 28 07:24:00 this you need to flash Oct 28 07:24:17 and the explanation was sent to some of the lists by werner Oct 28 07:24:19 need to dig that out Oct 28 07:24:22 been a while Oct 28 07:24:29 (werner) got this one Oct 28 07:24:54 ah, good Oct 28 07:24:59 which devices U tested moko9 not to brick them? ;-) Oct 28 07:25:00 on which list was taht? Oct 28 07:25:09 mompl Oct 28 07:25:11 01Bv4 Oct 28 07:25:16 02v5 Oct 28 07:25:17 02v6 Oct 28 07:25:39 http://lists.openmoko.org/pipermail/openmoko-devel/2008-April/002605.html Oct 28 07:26:17 thanks Oct 28 07:26:19 (mail arrived) Oct 28 07:26:27 could you prepare a 5iner mail to me, so I got it non-volatile? Oct 28 07:26:34 5-liner Oct 28 07:27:02 sure Oct 28 07:27:05 thanks Oct 28 07:27:09 np Oct 28 07:27:12 actually Oct 28 07:27:37 i might as well do the wiki page Oct 28 07:27:52 hat's what I thought ;-) Oct 28 07:27:55 mmh Oct 28 07:28:08 note to self: don't be available so early on irc Oct 28 07:28:11 *shrug* Oct 28 07:28:27 I'll place the ipkg and moko9 bin to my people/joerg Oct 28 07:28:35 ok Oct 28 07:28:36 lol Oct 28 07:28:40 i link to these Oct 28 07:28:45 k Oct 28 07:28:59 you will also need the gpio ipkg Oct 28 07:29:05 i'll build one Oct 28 07:30:30 well, I'll see whether I can set perms to g+w for the drawer about mokoFW, so U can copy to this place Oct 28 07:30:48 ok Oct 28 07:30:56 i can also mail it to you Oct 28 07:30:57 it's small Oct 28 07:31:20 ok, go ahaed Oct 28 07:31:21 216K Oct 28 07:31:28 np Oct 28 07:31:28 done Oct 28 07:31:31 mickeyl, one simple question: Can opkg differ between frameworkd-gitr-833 and frameworkd-gitr-838 ? If yes, how does it differentiate git revisions? Oct 28 07:32:01 quickdev: IIRC it uses lexical comparison Oct 28 07:32:12 the detailed algorithm is in opkg :) Oct 28 07:32:25 to differenciate git versions, you need to add a prefix Oct 28 07:32:34 we ported some Om stuff back to OE Oct 28 07:32:42 mickeyl, so 838 is considered newer as 833, damn..the opposite is true Oct 28 07:32:42 that puts the revision number into the prefix Oct 28 07:32:43 mickeyl: already awake? Oct 28 07:32:50 stefan_schmidt: unfortunately, ya Oct 28 07:32:54 mickeyl: heh Oct 28 07:33:04 mickeyl: Our trac does not like git any more Oct 28 07:33:04 Warning: Can't synchronize with the repository (Unsupported version control system "git": Can't find an appropriate component, maybe the corresponding plugin was not enabled? ). Look in the Trac log for more information. Oct 28 07:33:20 argh Oct 28 07:34:21 let me try to restart Oct 28 07:34:31 ah no Oct 28 07:34:34 trac is no longer an own daemon Oct 28 07:34:35 hmm Oct 28 07:35:45 sftp://joerg@sita.openmoko.org/home/joerg/public_html/calypso_moko_FW/s3c24xx-gpio_1.0+svnr4130-r2.1_armv4t.ipk Oct 28 07:36:15 mickeyl: aka people.om/joerg/calypso_moko_FW/s3c24xx-gpio_1.0+svnr4130-r2.1_armv4t.ipk Oct 28 07:36:24 k Oct 28 07:41:11 mickeyl, shr guys had problem with frameworkd version yesterday. Of course we can't add a prefix, because frameworkd is maintained by you. You ported some Om stuff back to OE? How could we make use of that? Oct 28 07:41:21 *problems Oct 28 07:42:32 there's some new option for that Oct 28 07:42:50 BB_GIT_CLONE_FOR_SRCREV = "1" Oct 28 07:42:58 use that and rebuild everything Oct 28 07:43:06 then you should get nice increasing revisions Oct 28 07:43:08 mickeyl, we need to add that to local.conf, right? Oct 28 07:43:12 ya Oct 28 07:43:34 will all package names change? Oct 28 07:43:54 mickeyl: ETA for MOKO9-beta1 2:30min Oct 28 07:46:35 quickdev: no name will change Oct 28 07:46:47 but all revisions for packages built from git Oct 28 07:47:56 mickeyl, so I only have to rebuild git packages, right? Oct 28 07:48:05 I'll see Oct 28 07:48:12 have to go now Oct 28 07:48:16 thanks for that valuable info Oct 28 07:48:28 and good luck with the new FW :) Oct 28 07:48:30 cya Oct 28 07:49:39 cu Oct 28 07:50:03 mickeyl: Tried to build fso-image the since yesterday? gst-plugins-bad_0.10.6.b failed Oct 28 07:50:26 | No package 'gstreamer-plugins-base-0.10' found Oct 28 07:50:26 | configure: error: no gstreamer-plugins-base-0.10 >= 0.10.17 (GStreamer Base Plug-ins Library) found Oct 28 07:51:12 Still we have gst-plugins-base_0.10.17.bb in the repo Oct 28 07:51:16 * stefan_schmidt is puzzled Oct 28 07:51:42 ya, zecke broke it Oct 28 07:51:48 no time to check yet Oct 28 07:52:01 ah Oct 28 07:52:01 something wrt. removing libmad Oct 28 07:52:10 interestingly, it builds fine for me here Oct 28 07:52:13 but breaks on the buildserver Oct 28 07:52:17 ok Oct 28 07:52:39 hmm it does not build for me on my workstation either Oct 28 08:02:14 ok, trac should be fixed again Oct 28 08:02:37 mickeyl: Great. What was the problem? Oct 28 08:03:05 after an aptitude upgrade the gitplugin refused to work Oct 28 08:03:18 had to update that (manually, since we have no package) and change some permissions Oct 28 08:03:40 ah Oct 28 08:03:53 The joy of out-of-package-system installs :) Oct 28 08:04:10 Let's see how zecke broke my build. Oct 28 08:04:28 It seems it only can be one patch he merged from OM Oct 28 08:04:35 yea Oct 28 08:04:59 f4fcadb061442a394cbe50041eb6acd9d4125eba Oct 28 08:05:37 uh, some openmoko only depends Oct 28 08:08:20 * stefan_schmidt tries a build with that one reverted Oct 28 08:31:50 mickeyl: http://trac.freesmartphone.org/ticket/201 Oct 28 08:32:02 mickeyl: Should we put tzdata into the fso-image? Oct 28 08:33:01 alphaone, mickeyl Oct 28 08:33:09 Bitbake problem here between 833 and 838 Oct 28 08:33:19 It won't regenerate frameworkd-config properly by itself Oct 28 08:51:15 mickeyl, already some info about the firmware? :) Oct 28 08:56:03 quickdev: we got new version 10h ago, we're going to test it internally now. please be patient! Oct 28 09:00:17 quickdev: U won't want to find new FW is melting down *YOUR* device, no? ;-) Oct 28 09:03:49 Ooh you can create meltdowns? Sweet! I knew it was a cool device, but this, wow.. Oct 28 09:10:29 nio_: it work under progress ;-) Oct 28 09:11:34 probably needs some hw-extensions to create a decent core meltdown Oct 28 09:12:00 e.g a nice little nuclear battery Oct 28 09:12:26 Well, it would solve the battery life problem..) Oct 28 09:30:22 *flashing moko10b1* Oct 28 09:33:14 :-) Oct 28 09:34:30 quickdev: hey there Oct 28 09:35:03 mickeyl, alphaone : well, any info on this issue with bitbake and git revisions of frameworkd ? Oct 28 09:35:18 issue? Oct 28 09:35:26 * alphaone must have missed it during the weekend Oct 28 09:35:29 Yep, the package didn't rebuild, we had to clean it manually Oct 28 09:35:35 I just told about it yesterday Oct 28 09:41:53 *booting frameworkd* Oct 28 09:52:53 freesmartphone.org: 03jluebbe 07stabilization/milestone4 * r133829643175 10framework/framework/persist.py: persist: readd logging Oct 28 09:53:12 freesmartphone.org: 03jluebbe 07framework * rb4b1aa607004 10/framework/persist.py: Merge branch 'stabilization/milestone4' Oct 28 10:16:56 Ainulindale, frameworkd thing can be fixed Oct 28 10:17:00 mickey|bbl told me how Oct 28 10:17:02 will tell you later Oct 28 10:17:05 don't bother with it Oct 28 10:17:06 cya Oct 28 10:32:11 mickey|bbl: when you get back, can you recommend a good u-boot for me to upgrade to? Oct 28 10:32:20 * pb__ fed up with having to manually erase nand before flashing Oct 28 10:36:13 http://wiki.openmoko.org/wiki/GSM/Flashing Oct 28 10:36:45 MOKO10-beta1 tested but doesn't fix #1024. Oct 28 10:37:03 * DocScrutinizer will kepp you informed Oct 28 10:38:02 PLEASE NOTE: MOKO9 is worth nothing probably, so the whole thing is mere academic for now Oct 28 10:40:41 Well, the real point is that people can flash gsm fw now :) Oct 28 10:46:14 stefan_schmidt: yes I agree. That is good news indeed. Oct 28 10:47:06 time to reverse-engineer it :) Oct 28 10:49:04 DocScrutinizer: Those buzzfix pictures look rather scary... ;) Oct 28 10:52:30 hey! how can i unmount /dev/mmcblkp01 when it keeps saying: umount: /media/card: device is busy ? Oct 28 10:55:48 sicu: sudo lsof | grep /media/card Oct 28 10:57:18 lindi-: on the FR ... lsof: not found Oct 28 10:57:30 havn't got a cardreader Oct 28 11:00:00 sicu: sudo apt-get install lsof Oct 28 11:00:05 sicu: it's telling you that at least once process is still using something on the card. you have to track down which process.. Oct 28 11:05:00 sicu: from where (directory on FR) are you trying to unmount? Oct 28 11:14:04 lindi- Weiss: thx, was qpe, mediaserv, clock, and quicklaun that held it ... killing qpe worked =D Oct 28 11:14:19 Kensan_: was trying from root Oct 28 11:50:41 also note: Tony is concerned there might have been issues with IMEI being nulled while flashing new firmware. THIS MIGHT BRICK YOUR DEVICE as no network will let you register, and you need to send device in for IMEI re-setting Oct 28 11:52:41 we are about to investigate further on IMEI issue, as well as on #1024, #666 et al. For now NO USE TO UPDATE calypso firmware - strongly deprecated! Oct 28 12:08:12 raster: hey! Oct 28 12:10:16 boo Oct 28 13:01:29 raster: still no disrespectfull words for my patches? ;) Oct 28 13:02:41 Kensan_: not yet Oct 28 13:02:42 :) Oct 28 13:02:49 i have applied some to my local tree herew Oct 28 13:03:08 raster: ok. Oct 28 13:03:19 raster: nothing broke? Oct 28 13:03:38 raster: btw, first boot will take considerably longer sit it will tar up /dev Oct 28 13:03:45 i think bootup may have slowed Oct 28 13:03:50 thanks to prelink Oct 28 13:03:52 i think Oct 28 13:03:54 yes Oct 28 13:03:54 raster: oh Oct 28 13:03:57 that first boot Oct 28 13:03:58 :) Oct 28 13:04:32 maybe tis that Oct 28 13:04:32 :) Oct 28 13:04:33 raster: will see how one can put device-specific dev.tar into oe Oct 28 13:04:33 raster: yeah, give it a second shot Oct 28 13:05:46 raster: plus prelink should be run on firstboot too. Oct 28 13:06:00 raster: need to put that in some script somewhere Oct 28 13:06:27 raster: do you think readahead could help speed up things? Oct 28 13:08:03 it could Oct 28 13:08:08 though given out current io problems Oct 28 13:08:10 not much Oct 28 13:08:17 as reading ahead will peg the cpu Oct 28 13:08:23 so u arent going to get any parallelism Oct 28 13:11:52 raster: hm yeah. So I guess upstart will not help much either... Oct 28 13:11:59 nup Oct 28 13:12:05 we need to either Oct 28 13:12:21 1. reduce compute 2. reduce io requirments (just load less) Oct 28 13:12:44 3. remove any stupid "Waiting" that waits for neiother a solidly booked cpu Oct 28 13:12:46 nore io Oct 28 13:14:09 raster: right. did you get any insider info from your samsung contacts about the nand issue? Oct 28 13:14:37 is this discussion about init system for OM? Oct 28 13:16:42 Sup3rkiddo: speeding up boottime yes. Oct 28 13:17:11 oh ok.....theres OpenRC and ruinit as well :D Oct 28 13:17:41 Sup3rkiddo: the point is they won't help Oct 28 13:17:50 Sup3rkiddo: I believe other init systems will not improve the situation Oct 28 13:17:55 there's only one CPU and reading from the flash uses 100% CPU Oct 28 13:18:21 oh...bummer... Oct 28 13:18:22 Kensan_: some of it we already knew Oct 28 13:18:29 theoretical bw is 6mb/sec from onenand Oct 28 13:18:43 bgut we barely scrape 1/2 of that and peg the cpu at max Oct 28 13:19:08 they seem to think we need to check the nand controller setup code (duh!) Oct 28 13:19:10 oh so its the case of a small pipe and a big tap? :D Oct 28 13:19:13 raster: yeah, we are well below that mark... Oct 28 13:19:13 so nothing we dont already know Oct 28 13:19:39 raster: hm... :( Oct 28 13:19:41 right now theres 2 options Oct 28 13:19:47 1 hyper-optimise the nand reads with cpu Oct 28 13:19:55 so we slice every cycle from the loop we can Oct 28 13:20:11 and make sure we keep it well busy reading as fast as nand can provide data Oct 28 13:20:31 raster: so ubifs could help quite a bit with the write caching etc... Oct 28 13:20:40 (so no 1 lop them go do something else for 200 cycles, then come back unles u literally DO have 200 cycles betwene words being available) Oct 28 13:20:51 write isnt the issue atm Oct 28 13:20:55 it can be a future one Oct 28 13:21:03 raster: yeah true. Oct 28 13:24:02 nikolajsheller ? Oct 28 13:25:09 right now tho.. fixing e's intl code to not be so picky about oeo's "strange" locale setup Oct 28 13:26:00 raster: There's some other stuff too, that can be taken care of before digging deeper into this. Oct 28 13:26:24 Kensan: regarding sysvinit versus others, an init system that didn't involve quite so many shell scripts would probably be an improvement; starting the shell is a fairly expensive operation. Oct 28 13:26:43 obviously you could stick with the core of sysvinit and rewrite the scripty bits, though. no need to jettison the whole thing, necessarily. Oct 28 13:27:51 raster: I think I saw in the bootcharts that OM is using readahead. Oct 28 13:28:11 apropos nand, it's a shame that the samsung cpu doesn't allow RnB to be an interrupt source. background-loading readahead data is a bit of a loss if you need to busy-wait for each sector to become available. Oct 28 13:28:13 pb__: yes definitely. dash was a *minor* improvement. Oct 28 13:28:44 using dash instead of bash that is Oct 28 13:29:12 Kensan: iirc, the last benchmark run I did on gta01 showed about 35 milliseconds to start the busybox default shell (ash, I guess). Oct 28 13:29:21 part of the problem is that fork/exec is just an expensive operation on arm Oct 28 13:30:09 maybe the best thing is to do some detailed nand read timings Oct 28 13:30:22 and clock exactyl how many ns between words being available Oct 28 13:30:26 and read a word Oct 28 13:30:30 then go off and do somehtnig else Oct 28 13:30:46 and make sure u interrupt exactl on or just before new word is available Oct 28 13:31:10 so u avoid clockinb/spinning on reads for more than absolutely necessarly Oct 28 13:31:14 the time between words is more or less negligible, it's the time between sectors that can be significant. Oct 28 13:31:19 pb__: ic, what would be a init-replacement you think would fit that shoe? Oct 28 13:31:21 I think the sector read time is usually specified as something like 10us. Oct 28 13:31:25 pb__: fork+exec is expensive anywhere Oct 28 13:31:30 actually fork isnt the baddie Oct 28 13:31:31 exec is Oct 28 13:31:33 its evil Oct 28 13:31:34 :) Oct 28 13:32:01 pb__: if every system daemon has to write new init scripts it will be a hassle and should be avoided so something "compatible" with sysvinit would be nice... Oct 28 13:33:07 Kensan: yeah. I think what you probably need to do is to optimise the startup for the daemons that are part of the base firmware load, and accept that it will get a bit slower if the user installs non-standard things after the fact. Oct 28 13:33:32 doing a completely customised startup system is obviously not going to be good news for ease of adding extra packages. Oct 28 13:34:24 the way we do that on some of our own embedded products is that we use sysvinit but with the various rcX.d directories basically empty as shipped: all the base stuff is actually run out of /linuxrc or directly from inittab entries. Oct 28 13:34:29 pb__: yes, I am actually hoping that the optimiziations can be applied to OE in general. Oct 28 13:34:38 right Oct 28 13:35:19 pb__: yeah, I cut down most stuff that's unneeded on bootup which helps. Oct 28 13:35:58 pb__: e.g. bluetooth/wlan does not need to be turned on right on bootup. The user should actively turn it on when he needs it. Oct 28 13:36:01 pb__: thats the right way to go Oct 28 13:36:09 Kensan: so, an easy place to start would probably be to replace the script implementation of /etc/init.d/rc with something else. Oct 28 13:36:12 so system base is one blob of c binary for example Oct 28 13:36:14 Kensan: yah, indeed Oct 28 13:36:21 its not changable withotu writign c and haveing source Oct 28 13:36:25 or completely replacing Oct 28 13:36:34 but thats not meant to be modified (much) ever Oct 28 13:36:42 as it should do all the things the system really needs Oct 28 13:36:45 and do it really efficiently Oct 28 13:37:17 * pb__ lunchtime now Oct 28 13:37:35 raster: but the issue with openmoko is, that it *should* work the same for all supported devices... I don't see OM/community maintaining device specific linuxrc-style init stuff. Oct 28 13:37:46 imho of course Oct 28 13:37:47 :) Oct 28 13:39:39 Kensan: in case you're interested, here are the benchmark numbers for shell, fork and exec: http://copper.reciva.com/pb/lmbench.txt Oct 28 13:39:46 those are on gta01 but I don't imagine gta02 will be all that different Oct 28 13:40:58 pb__: thank you. I will take a look at it tonight Oct 28 13:41:54 pb__: wow.. still have a netwinder? Oct 28 14:02:10 mickey|bbl: http://trac.freesmartphone.org/ticket/205 Oct 28 14:31:02 shoragan: fso-image built Oct 28 15:00:14 Hey Oct 28 15:00:18 Ainulindale, here? Oct 28 15:00:24 hey quickdev Oct 28 15:00:35 hey alphaone Oct 28 15:00:45 alphaone, btw...did mickey tell you about the RetrieveMessage bug? Oct 28 15:01:03 quickdev: The events/ring problem? Oct 28 15:01:16 Or the problem when sending with local numbers? Oct 28 15:01:22 none of them ;) Oct 28 15:01:25 Anyway, both should be fixed Oct 28 15:01:26 heh Oct 28 15:01:29 damn :-) Oct 28 15:01:53 When you call RetrieveMessagebook a timestamp is returned, but not when calling RetrieveMessage ;) Oct 28 15:02:06 quickdev: yes Oct 28 15:02:29 Ainulindale, do you want to know how to solve the frameworkd version thing? ;) Oct 28 15:02:40 quickdev: Ah! Oct 28 15:03:05 quickdev: That is the kind of error report that points me where I applies my stupidity :-) Oct 28 15:03:07 Thanks Oct 28 15:03:15 no prob ;) Oct 28 15:03:15 Will fix in a minute Oct 28 15:03:23 mickey|bbl, sorry for disturbing, how's the fw going? :) Oct 28 15:03:28 quickdev: already did Oct 28 15:04:04 Ainulindale, fine, with the bitbake config parameter? Oct 28 15:04:04 quickdev: AFAIK it's still 1024 prone Oct 28 15:04:25 quickdev: bitbake -c clean and bitbake frameworkd with the good MACHINE Oct 28 15:04:25 alphaone, it's a pity ;) Oct 28 15:04:32 quickdev: and the image has been generated Oct 28 15:05:04 I'd like to see this illume module now Oct 28 15:05:08 quickdev: Yeah, but it seems we're making at least some sort of progress in that department Oct 28 15:05:14 But I don't know where nikolajsheller is Oct 28 15:05:26 Ainulindale, fine, but things should build automatically, there's a special BB setting..one moment Oct 28 15:05:41 alphaone, yeah :) Oct 28 15:05:43 quickdev: yes they should but they didn't Oct 28 15:05:48 Hence my quick & dirty hack =) Oct 28 15:05:54 Ainulindale, we can fix that..one minute ;) Oct 28 15:06:15 hi guys Oct 28 15:06:27 hi methril Oct 28 15:06:32 what is the daemon who deals directly with the GSM HW? in FSO? Oct 28 15:06:34 Ainulindale, http://rafb.net/p/E7MsO120.html Oct 28 15:06:45 methril, frameworkd Oct 28 15:07:02 but it directly sends the AT cmds? to the serial line? Oct 28 15:07:21 i wasn't able to configure for other GSM modem Oct 28 15:07:35 methril, afaik through gsm0710muxd Oct 28 15:07:57 thanks Oct 28 15:08:02 time to research ;) Oct 28 15:12:08 Ainulindale, maybe we need to build everything from scratch with that setting Oct 28 15:12:54 We could Oct 28 15:13:52 quickdev: what is the bug for that again Oct 28 15:14:19 alphaone, I don't get you Oct 28 15:14:32 which bug for what? ;) Oct 28 15:15:29 Ainulindale, Optionally use git-rev-list to get a sortable revision. With setting BB_GIT_CLONE_FOR_SRCREV="1" you can get a sensible and global (per repository with only fast forwards) revision. The downsides are you will have to have a repository at parse time which means you will git-clone certain trees you don't even use. This is also the reason why this is optional. This also means you might need to download your git checkouts Oct 28 15:15:29 to get this feature working. Oct 28 15:16:12 quickdev: by the way, which tool would you suggest for this planning thing ? Oct 28 15:16:34 Ainulindale, why not a text file? Oct 28 15:17:16 quickdev: link to the bug for the timestamp-foo Oct 28 15:17:29 alphaone, I haven't reported a bug Oct 28 15:19:49 quickdev: I had some bugs today Oct 28 15:19:56 Call incoming, clicked on accept Oct 28 15:20:00 And it didn't accept and kept ringing Oct 28 15:20:06 So I guess it was a frameworkd problem Oct 28 15:20:15 Ainulindale, frameworkd log? Oct 28 15:20:44 Had to reboot Oct 28 15:21:01 By the way, what is this issue with ld.so.conf and the postinst things ? Oct 28 15:21:49 Ainulindale, don't know Oct 28 15:22:35 Ainulindale, I'm stopping the buildhost and I'll clean the builds Oct 28 15:22:43 Ainulindale: me too, I had problem with ld.so.conf Oct 28 15:23:07 but we are not alone .. Oct 28 15:23:12 http://n2.nabble.com/-fso-testing--Where-has-ld.so.conf-gone--td1375314.html Oct 28 15:23:55 quickdev: Ok, no problem. Oct 28 15:25:40 Well I have to go to the dentist (again), see you in a while Oct 28 15:26:06 ok, see you Oct 28 15:26:08 good luck Oct 28 15:36:08 rwhitby, how do I make the buildhost build from scratch? remove tmp directory? Oct 28 15:40:50 freesmartphone.org: 03daniel 07stabilization/milestone4 * rf112e62e169c 10framework/framework/subsystems/ogsmd/modems/ti_calypso/ (channel.py unsolicited.py): Oct 28 15:40:50 freesmartphone.org: ogsmd: Fix TI calypso recamping logic (#190) Oct 28 15:40:50 freesmartphone.org: The timer is now removed if we disable GSM resource. Oct 28 15:40:50 freesmartphone.org: 03daniel 07stabilization/milestone4 * r1e3c0641e262 10framework/framework/subsystems/ogsmd/modems/abstract/mediator.py: ogsmd: Correctly match on SMS category in RetrieveMessage Oct 28 15:43:13 alphaone, thanks, will integrate it later :) Oct 28 15:45:01 ogsmd: Fix TI calypso recamping logic (#190) <-- nice :) Oct 28 15:48:24 here the weather sux Oct 28 15:48:33 rain rain rain Oct 28 15:48:34 freesmartphone.org: 03daniel 07framework * r46fd8ed54877 10/framework/subsystems/ogsmd/modems/ (3 files in 2 dirs): Oct 28 15:48:34 freesmartphone.org: Merge commit 'origin/stabilization/milestone4' Oct 28 15:48:34 freesmartphone.org: Conflicts: Oct 28 15:48:34 freesmartphone.org: framework/subsystems/ogsmd/modems/abstract/mediator.py Oct 28 15:48:35 freesmartphone.org: (changed dir -> direction since dir is a python builtin) Oct 28 16:42:24 dum de dum Oct 28 19:12:31 whats the command to turn off gsm Oct 28 19:15:34 kd8ikt: find /sys | grep gsm | grep power_on? Oct 28 19:16:32 /sys/devices/platform/neo1973-pm-gsm.0/power_on Oct 28 19:16:42 yay Oct 28 19:16:47 echo 0 >> Oct 28 19:17:01 er just one > Oct 28 19:17:33 freesmartphone.org: 03morphis * r310 10/trunk/python-elementary/ (8 files in 2 dirs): Added more functionality and setup.py Oct 28 19:18:20 oh wait there is a init.d script to turn it off Oct 28 20:07:04 Hey Oct 28 20:39:36 hello Oct 28 20:40:20 hey nikolajsheller :) Oct 28 20:40:38 slow night :) Oct 28 20:42:03 slow night? mh? ;) Oct 28 20:43:52 quickdev: well, on shr the web area is symlinked to tmp, so you will want to move tmp to tmp.old, and change the symlinks in the web area so that the feeds don't suddenly disappear. Once everything is rebuilt, then you change the symlink back and delete tmp.old. Oct 28 20:44:16 europe :) Oct 28 20:44:18 rwhitby, yeah, thought something like that, thanks! Oct 28 20:46:31 I'm still trying to get illume to build Oct 28 20:46:32 I've tried with: svn://svn.projects.openmoko.org/svnroot/illume Oct 28 20:46:32 and: http://svn.enlightenment.org/svn/e/trunk/TMP/oe/ Oct 28 20:48:07 but I'm getting non-obvious problems with missing typedefs and the like. I you know of a guide on how to build from scratch, I would be much obliged Oct 28 20:48:51 nikolajsheller, I like your endurance :) Oct 28 20:49:14 nikolajsheller, did you build e, edje, etc.? Oct 28 20:49:36 quickdev: yes. they seem to be in place Oct 28 20:50:08 I haven't ever built illume, but could you paste errors log on rafb.net/pastebin ? :) Oct 28 20:50:32 I've tried building illume from: svn://svn.projects.openmoko.org/svnroot/illume Oct 28 20:51:47 nikolajsheller, error log :) Oct 28 20:52:10 quickdev: http://rafb.net/p/d4bO2n57.html Oct 28 20:53:20 nikolajsheller, is that the full error log? Oct 28 20:53:35 gsm_get.c:335: ecore_string_init(); Oct 28 20:54:03 no, just the section that fails. 2 secs Oct 28 20:54:41 nikolajsheller, it probably does not find the core lib Oct 28 20:54:52 I need the full log Oct 28 20:55:28 it's the compiler that fails. I seems as though it doesn't get the prototyes from the required header? Oct 28 20:56:25 ah... how do I pipe stderr into a file? make &1> file or such.... Oct 28 20:59:27 make 2>&1 > file.log (afaik) Oct 28 21:00:29 quickdev: illume build fail (whole log): http://rafb.net/p/aR3s9188.html Oct 28 21:01:45 nikolajsheller, /usr/local/lib/libecore.so - does that file exist? Oct 28 21:04:23 /usr/local/lib/libecore.so -> libecore.so.0.9.9 Oct 28 21:04:23 /usr/local/lib/libecore.so.0 -> libecore.so.0.9.9 Oct 28 21:04:23 /usr/local/lib/libecore.so.0.9.9 Oct 28 21:04:46 it softlinks... Oct 28 21:05:43 nikolajsheller, strings libecore.so | grep ecore_string_init - does that output ecore_string_init ? Oct 28 21:06:59 nope.... Oct 28 21:07:17 hmm? Oct 28 21:07:53 nikolajsheller, my grep does....so your library is wrong...maybe a wrong version Oct 28 21:08:17 where did you get it from? ecore.. Oct 28 21:08:31 interesting. I built it from the repository version... Oct 28 21:08:45 that I am not sure of Oct 28 21:09:10 I didn't keep track of where the files came from... Oct 28 21:09:22 try to install it again Oct 28 21:09:39 but I read the error in the log as a compiler error, not a linker error? Oct 28 21:10:13 the compiler makes sure that all functions used can be referenced in the libraries Oct 28 21:10:14 But it seems you nailed the problem :) Oct 28 21:11:20 Thank you so much for your help. I'll see if I can find out why my ecore is a couple of functions short Oct 28 21:12:18 nikolajsheller, run file libecore.so.0.9.9 Oct 28 21:12:58 libecore.so.0.9.9: ELF 64-bit LSB shared object, x86-64, version 1 (SYSV), not stripped Oct 28 21:13:24 nikolajsheller: in bash, its : program &> file Oct 28 21:14:26 nikolajsheller, but strings libecore.so outputs many functions, right? Oct 28 21:14:35 thseiler: thank you Oct 28 21:15:14 quickdev: yes. But nothing for ecore_stri* Oct 28 21:15:48 quickdev: yes. But nothing for ecore_str* yeild a bunch. Oct 28 21:16:30 quickdev: so it seems like I have built a version without string functions... Oct 28 21:20:28 nikolajsheller, http://trac.enlightenment.org/e/changeset/36675/trunk/ecore/src/lib/ecore/Ecore_Data.h Oct 28 21:23:04 nikolajsheller, so your compiler probably does not get the newest headers Oct 28 21:25:04 odd... I checked out from svn within this week the change is 13 days old Oct 28 21:25:57 yeah, that's why your library hasn't got the functions....but the headers were not updated Oct 28 21:26:11 so gcc still thinks there's an ecore_string_init function Oct 28 21:26:46 try to reinstall ecore and check that/where the headers are installed Oct 28 21:27:52 quickdev: very cool. Will do. Thank you very much for your help! Oct 28 21:28:04 no prob Oct 28 21:33:24 rwhitby, can I modify the symlink to tmp? I think I can't.. Oct 28 21:33:55 if you're logged in you can Oct 28 21:34:06 I am, which file is it? Oct 28 21:34:56 whoever changed it back last time knows ... Oct 28 21:35:01 (that wasn't me) Oct 28 21:35:08 probably Ainulindale Oct 28 21:58:04 mickey|bbl, here? got some frameworkd error from my today's log Oct 28 21:58:29 *some errors Oct 28 21:59:42 or alphaone :) but I think it's to late, right? :) Oct 28 22:22:53 quickdev: what did I do ? Oct 28 22:23:07 Ainulindale, what did you do? Oct 28 22:23:10 I don't know Oct 28 22:23:11 The symlink Oct 28 22:23:19 ah, nothing...already solved Oct 28 22:23:40 What was the problem ? Oct 28 22:24:02 what I did: cd shr-testing, mv tmp tmp.old.....and then I adjusted the symlink......so the images are still available while the buildhost is now building a complete new tmp folder Oct 28 22:24:19 Oh ok :-) Oct 28 22:24:40 quickdev: By the way did you see the explanation I put on the wiki ? Oct 28 22:24:54 yeah, that was missing Oct 28 22:25:20 Ainulindale, I discovered another frameworkd failure our image was suffering from Oct 28 22:25:30 Yet another one ? :-) Oct 28 22:25:42 What is it this time? Oct 28 22:25:44 yeah, the incoming message signals breaks frameworkd Oct 28 22:25:57 Nice :-) Oct 28 22:26:04 alphaone: Fix this or I shall destroy you Oct 28 22:26:10 except that, it's quite stable here Oct 28 22:27:44 mickey|bbl, alphaone, http://rafb.net/p/FriqtY77.html, notice that I'm not using HEAD, but a few revision old version Oct 28 22:28:11 Oh I know about that Oct 28 22:28:13 That's oeventsd again Oct 28 22:28:20 charlie is a breaker :-) Oct 28 22:28:31 I don't think it's charlie's fault, we'll see Oct 28 22:28:43 Ainulindale, hopefully we won't have package probs again after the rebuild Oct 28 22:28:52 You mean postinst ? Oct 28 22:28:57 quickdev: by the way Oct 28 22:29:04 Tomorrow I'm going on holidays Oct 28 22:29:10 I'll have internet and probably will be around Oct 28 22:29:14 Ainulindale, I mean opkg update && opkg upgrade, for example frameworkd prob Oct 28 22:29:25 But I might be not that available from monday to wednesday Oct 28 22:29:26 where are you going to? :) Oct 28 22:29:30 To my parents Oct 28 22:29:33 South of France Oct 28 22:29:39 (Carcassonne) Oct 28 22:29:49 ok, have some fine days :) Oct 28 22:29:55 quickdev: To see my mother before her reeducation Oct 28 22:30:26 quickdev: oh and by the way, if you want a bit of laugh Oct 28 22:30:33 If I want to fix everything that's wrong with my teeth Oct 28 22:30:44 It'll cost me approximatively 16 000 euros Oct 28 22:30:58 Isn't it funny ? Oct 28 22:31:02 damn, sell your car! ;) Oct 28 22:31:15 My car is worth well Oct 28 22:31:16 100 euros ? Oct 28 22:31:26 euro :o Oct 28 22:31:31 not euros :o Oct 28 22:31:35 sounds bad Oct 28 22:31:38 Yeah, same thing Oct 28 22:31:55 It's supposed not to be plural but even on my checkbooks the bank puts a plural Oct 28 22:31:58 you can sell your FR :o Oct 28 22:32:09 Go to die ! Oct 28 22:32:27 * Hire2 jump from the window Oct 28 22:32:31 jumps Oct 28 22:34:10 16000 for a teeth? Oct 28 22:34:22 Not a tooth but teeth Oct 28 22:34:25 Several Oct 28 22:34:51 That's what you got when you like to eat good food :-) Oct 28 22:36:01 oh yes sorry :D Oct 28 22:36:17 hoh many teeth you have broken? Oct 28 22:36:21 how Oct 28 22:36:58 Far too much Oct 28 22:37:52 sadly, me too I have to fix two teeth :( Oct 28 22:38:09 i have broken two widsom teeth :\ Oct 28 22:42:51 * Hire2 starts to watch 1x13 of Monk Oct 28 23:25:14 hello people Oct 28 23:25:54 I had a question Oct 29 00:05:35 ring ring rin **** ENDING LOGGING AT Wed Oct 29 02:59:57 2008