**** BEGIN LOGGING AT Thu Jun 23 02:59:56 2011 Jun 23 05:48:17 anyone know if intone supports flac? Jun 23 06:02:30 gah, intone crashes scanning my music Jun 23 06:39:58 always did for me Jun 23 06:40:37 didn't when I first scanned it in Jun 23 06:40:57 never got it to work, so I gave up on it Jun 23 06:41:22 esp since FR isn't really a great MP3 player anyway ;-P Jun 23 06:52:09 it was my first MP3 player so yeah. most of the time the OS was broken enough that intone was my only use for my FR Jun 23 08:49:09 morning Jun 23 08:49:36 I have a checkout of shr-unstable here and am trying to create a patch to submit Jun 23 08:50:01 when I run OE om-gta02@shr ~/shr-unstable/openembedded/recipes $ git diff Jun 23 08:50:05 I get the following error: Jun 23 08:50:11 error: object directory .git/objects/replace git object reference with relative path echo ../../../../openembedded/.git/objects/ git checkout master does not Jun 23 08:50:16 what am I doing wrong here? Jun 23 08:50:54 sounds like your repo is broken, try git fsck Jun 23 08:50:59 or check out again Jun 23 08:52:19 pabs3: it's a clean checkout using the Makefile from yesterday Jun 23 08:52:25 I only made one bb -k shr-image Jun 23 08:52:41 then edited one file with vim and ran bb package Jun 23 08:53:05 wow Jun 23 08:53:15 git fsck reports metric fsckton of errors Jun 23 08:53:25 *sigh* I'll never understand git Jun 23 08:53:33 a clean checkout... and it's horribly broken already Jun 23 09:11:23 blindcoder: just commit your changes and then use git format-patch Jun 23 09:12:01 OE om-gta02@shr ~/shr-unstable/openembedded/recipes/fltk $ git commit fltkhackdiet_svn.bb Jun 23 09:12:04 error: object directory .git/objects/replace git object reference with relative path echo ../../../../openembedded/.git/objects/ git checkout master does not Jun 23 09:12:18 commit impossible :/ Jun 23 09:12:50 that line continues with: exist; check .git/objects/info/alternates. Jun 23 09:33:46 blindcoder: see if http://ibot.rikers.org/%23openmoko-cdevel/20110606.html.gz starting at 13:30 Jun 23 09:34:27 can help in your problem Jun 23 09:37:46 jluis: thanks, looks like lxsameer has fixed it, sadly he doesn't say how :( Jun 23 09:37:53 and I know nothing about git Jun 23 09:38:11 but when I use a fresh "make setup-shr-chroot" then git seems to work Jun 23 09:38:25 I'll try to use a fresh setup for patch submission Jun 23 09:39:47 ~seen lxsameer Jun 23 09:39:56 lxsameer <~lxsameer@unaffiliated/lxsameer> was last seen on IRC in channel #openmoko-cdevel, 14d 16h 31m 25s ago, saying: 'mrmoku: Notice: An email has been sent to lxsameer@gnu.org with a token to verify your new email address '. Jun 23 09:44:18 blindcoder: are you using shr-chroot? Jun 23 09:44:23 jluis: yes Jun 23 09:44:40 hello golks Jun 23 09:44:44 jluis: I followed http://shr-project.org/trac/wiki/Building%20SHR Jun 23 09:46:17 what do you have on /OE/shr-unstable/openembedded/.git/objects/info/alternates Jun 23 09:46:30 form the chroot Jun 23 09:46:46 replace git object reference with relative path echo ../../../../openembedded/.git/objects/ git checkout master Jun 23 09:47:43 that's the content of the file Jun 23 09:49:51 I replicated a system with dump/restore from a 16GB uSD card to a new 8GB one. System was 700M. The new card boots weird: the kernel starts, then panics with "not synching: No init found. Try passing init= option to the kernel" Jun 23 09:51:27 edit the line to have "/OE/openembedded/.git/objects" or "../../openembedded/.git/objects" Jun 23 09:52:06 the Power button 10s shut down does not work, and the LED blinks red Jun 23 09:52:16 any clue how I can debug this? Jun 23 09:52:56 mounting the new uSD on another host works fine, and fsck -f shows FS clean Jun 23 09:54:29 so just remove two ../ ? Jun 23 09:58:21 blindcoder: and the "replace git ..." it only need the path ( relative or absolute to the object repository) Jun 23 09:59:37 ah, okay Jun 23 10:00:20 jluis: works now, thanks! Jun 23 10:00:28 but where does that line come from? Jun 23 10:02:19 okay, I have the commit and I have the format-patch output Jun 23 10:02:35 where do I send the patch now? I've been unsubscribed from the mailinglists for a while Jun 23 10:02:50 it comes from the makefile Jun 23 10:03:15 ah Jun 23 10:03:32 should I send the patch to shr-devel@lists...? Jun 23 10:05:34 or to openembedded-devel Jun 23 10:08:30 I'm not sure if openembedded is interested in my small fltkhackdiet program ;) Jun 23 10:09:56 the recipe is in openemdedded so somebody was interested on it Jun 23 10:10:25 s/openemdedded/openembedded/ Jun 23 10:10:25 jluis meant: the recipe is in openembedded so somebody was interested on it Jun 23 10:11:19 michelem: what kernel and bootloader are you using? Jun 23 10:11:56 kernel is 2.6.32 I believe, BL is Qi Jun 23 10:19:05 what distro ? I think that no distro ships now with 2.6.32 Jun 23 10:20:14 also the contents of boot and could help Jun 23 10:21:13 distro is Debian Squeeze Jun 23 10:21:27 well, it's on the same phone Jun 23 10:21:43 the old card boots, then I dumped and restored it on a new card, the new card does not work Jun 23 10:21:55 is there a way to get the kernel messages on the phone? Jun 23 10:25:03 holding down the power button during the start increases the verbosity of kernel messages Jun 23 10:25:38 good to know Jun 23 10:26:21 uhm Jun 23 10:26:39 I'm seaching in the mail the kernel parameter to slow-down the card reads Jun 23 10:27:03 I was just about to ask :) Jun 23 10:28:14 VFS says root FS mounted Jun 23 10:37:55 searching for the parm I found that 2.6.32 was failling on some SD see http://www.mail-archive.com/shr-user@lists.shr-project.org/msg05212.html Jun 23 10:49:51 seems relevant; the only catch, those refer ability to read the partition tables, whereas I get VFS: root mounted, so I expect the partition table to be read at least Jun 23 10:50:15 btw, is that glamo_mci.sd_max_clk=5000000 a kernel parameter? Jun 23 10:59:22 michelem: that's the maximum frequency used by the SD card controller. Jun 23 10:59:48 PaulFertser: yes, but where do I set it? can I set it at runtime with sysctl? Jun 23 10:59:50 I think soo see also the thread at http://lists.openmoko.org/nabble.html#nabble-td6066048 that I tink is more apropiate Jun 23 11:00:15 michelem: you can set it runtime by "echoing" in /sys/modules/glamo_mci/parameters/sd_max_clk Jun 23 11:00:24 michelem: or with kernel command line of course. Jun 23 11:00:27 (on boot) Jun 23 11:00:46 I guess mine can be a clock problem. The part table is read, FS mounted, but init is not read. Bus may fail on larger read()s Jun 23 11:00:55 PaulFertser: k, thanks Jun 23 11:02:06 true I cannot pass custom arguments to the kernel with Qi? Jun 23 11:02:38 michelem: if you boot from uSD you can use /boot/append-GTA02 Jun 23 11:02:45 michelem: check Qi wiki page for details :) Jun 23 11:03:05 michelem: what kernel are you using, is it the usual or OpenWrt's one? Jun 23 11:06:09 thanks, bbs Jun 23 11:56:29 my kernel is: "Linux version 2.6.29-rc3-v24 (radek@rp-skunk) (gcc version 4.3.2 (Debian 4.3.2-1.1) ) #24 Sun May 23 23:32:50 CEST 2010" Jun 23 12:02:32 oh yeah, with the lower frequency the new uSD works Jun 23 12:10:19 any tip for exploring higher clock settings? Jun 23 12:11:37 I guess I can raise the sd_slow_ratio progressively Jun 23 12:12:49 * SpeedEvil tries to remember what the maximum PLL settings are for the 2410. Jun 23 12:12:53 (CPU) Jun 23 12:13:02 michelem: I guess it would be much faster to find if you are booting off NAND and not rebooting between adjustments Jun 23 12:13:17 SpeedEvil: LOL Jun 23 12:14:05 PaulFertser: hi :-) Jun 23 12:21:58 well I can test echoing to sysfs as PaulFertser mentioned before Jun 23 12:33:50 out of curiosity, which WM do you folks use? I'm looking for ideas. Need something minimal, customizable, fast. Jun 23 12:34:37 DocScrutinizer: hi :)) Jun 23 12:35:03 michelem: nah, slow_ratio is only in effect when GPS is powered iirc. Jun 23 12:35:32 michelem: i think you might want to check his setup: http://wiki.openmoko.org/wiki/User:Lindi Jun 23 12:35:43 PaulFertser: may be totally placebo, but setting it from 16 down to 2 seems to make ti faster Jun 23 12:35:46 thanks Jun 23 12:35:50 I'm using e17 but not really liking it on FR. Jun 23 12:38:00 so there's no educated way to test the glamo_mci frequencies, just try random value? Jun 23 12:39:32 michelem: probably bisecting? ;) Jun 23 12:39:45 michelem: use at least 1 for sd_drive. Jun 23 12:41:38 PaulFertser: do I find the meaning of those params anywhere? Jun 23 12:41:57 cat sd_drive is currently 0 Jun 23 12:43:27 michelem: not really, some commit messages specify it. Jun 23 12:43:30 michelem: 0 is not enough. Jun 23 12:43:36 michelem: for some cards. Jun 23 12:45:30 "First is Glamo drive strength on SD IOs including CLK." <— wtf? Jun 23 12:48:31 turn it to 11! Jun 23 12:48:32 michelem: makes perfect sense ;) Jun 23 12:48:38 No, 3 is the maximum. Jun 23 12:48:38 *duck* Jun 23 12:49:13 michelem: it was lowered to not interfere with gps. But 0 is really not enough, 1 should have been set the default (and it still allows gps to work). Jun 23 12:49:35 I do not use GPS, it is OFF for power saving. Jun 23 12:49:54 so what is it practically? Jun 23 12:50:02 a one liner would be fine Jun 23 12:50:11 mmh... isn't SD clock turned off while no accesses are underway? Jun 23 12:50:51 also, sd_max_clk and sd_post_power_clock are the number of CPU cycles per each bus check? Jun 23 12:51:22 with the current settings, the system is terribly slow Jun 23 12:53:06 I'll just assume that 1) sd_max_clk is the relevant setting 2) higher is faster Jun 23 12:58:28 michelem: sd_drive is basically "drive strength" i.e. how strong are the pull-ups/downs driving the data lines. Jun 23 12:59:06 michelem: if they drive them really strong, rough edges on transitions have too much hi-freq harmonics. Jun 23 12:59:47 PaulFertser: can you please Jun 23 12:59:47 # for i in /sys/module/glamo_mci/parameters/* ; do echo "* $i" ; cat $i ; done Jun 23 13:00:28 I set it to 1, and sd_max_clk to 10000000, but speed doesn't seem changed Jun 23 13:01:07 michelem: 0 0 16666666 1000000 8 Jun 23 13:01:12 (typed manually) Jun 23 13:01:22 thanks for your patience. Jun 23 13:01:22 michelem: speed depends on max_clk. Jun 23 13:01:38 michelem: so you should try sd_drive=1 (or =2) with the default clk. Jun 23 13:01:53 Also, keep in mind SD access really is slow on FRs. Jun 23 13:02:18 (but more modern kernels perform considerably better in this regard, especially when the screen is blanked) Jun 23 13:02:31 the base system ran OK with the other card, and with this one too (default settings). But there is one specific operation that kills with time Jun 23 13:03:07 that is one simple SQLite INSERT. SQLite creates a lock file, edits the database, then removes the lockfile. This takes like 3-4 seconds Jun 23 13:03:50 PaulFertser: are you recommending to upgrade my 2.6.29 ? Jun 23 13:04:19 We are shipping 12 of these devices for pilot testing down to Africa in a few weeks. Jun 23 13:05:17 michelem: i'm not sure, sorry. I know the kernel folks were working hard to improve it from 2.6.29 but i keep the known to work version on my personal device because it works good enough for me. Jun 23 13:05:37 k, thanks Jun 23 13:09:56 /boot/uImage-GTA0[123].bin <— is this a search sequence, or Qi looks ONLY for GTA0x where x = current platform ? Jun 23 13:13:21 michelem: only Jun 23 13:13:31 michelem: it can also use uImage.bin iirc. Jun 23 13:13:35 thx Jun 23 14:05:00 new kernel works, but all namings have changed again Jun 23 14:06:12 michelem: there's "om" utility that does great work on unification of kernel interfaces for end-users. Jun 23 14:06:42 PaulFertser: they changed namings of that too, gta02-power:blue became gta02:blue:power :P Jun 23 14:07:04 michelem: and in newer kernels sd_drive is automatic, the more the frequency, the stronger the drive, so you do not have to care about it anymore (i hope). Jun 23 14:07:28 yes, I confirm sd_drive et al disappeared Jun 23 14:20:15 also, APM disappeared ffs Jun 23 14:30:00 APM always sucked you know :) Jun 23 14:31:01 is there any other way to get the battery status portably? Jun 23 14:31:55 michelem: modern kernel interface is /sys/class/power_supply/* Jun 23 14:32:34 michelem: there used to be an idea that "hald" should be able to provide that too (on n8x0 and n900 there's a special plugin for it because they're insane enough to not use a kernel driver). Jun 23 14:32:56 michelem: but currently hald is deprecated, so powerkit (?) is considered to be a replacement. Jun 23 14:33:04 I'd stick with /sys/class/power_supply/* Jun 23 14:33:47 DocScrutinizer: any luck? Jun 23 14:34:02 nah, sorry not really Jun 23 14:38:56 excellent, we're nearly there. Last thing is, how to set the moko to power-supply the USB bus when in host mode Jun 23 14:40:10 I used to "echo 1 > /sys/devices/platform/s3c2440-i2c/i2c-adapter/i2c-0/0-0073/neo1973-pm-host.0/hostmode" but that's also vanished Jun 23 14:40:21 and om usb mode host does not power-supply Jun 23 14:40:22 michelem: use "om" Jun 23 14:40:32 michelem: yes, it has another command for that. Jun 23 14:41:19 right, found. Jun 23 14:41:52 I'm a bit reluctant to use "om" since we are considering testing alternative hardware Jun 23 14:42:14 the FR so far was sufficient, but it's really choked Jun 23 14:42:50 # om usb charger-mode power-usb Jun 23 14:42:50 setting usb charger mode: No such file or directory Jun 23 14:43:01 michelem: om version too old? Jun 23 14:44:05 version 0.12, dist-upgrade doesn't claim new ones. Any new? Jun 23 14:45:31 michelem: http://anonscm.debian.org/gitweb/?p=pkg-fso/omhacks.git;a=blob;f=debian/changelog;h=0580d168c6993786ab3b3ebe3111eed43d1595fd;hb=HEAD Jun 23 14:46:32 michelem: so it should be somewhere there in "unstable" or "experimental". Jun 23 15:40:46 PaulFertser: hostmode round 2 ;-P Jun 23 15:40:55 N9(50) Jun 23 15:41:04 just started Jun 23 17:22:49 DocScrutinizer: where at? Jun 23 17:31:02 on #meego and in ##mhd Jun 23 17:31:31 new "member" dm8tbr Jun 23 17:31:57 allegedly done much hostmode and kernel, and was interested in bringing it to N9 Jun 23 17:32:29 PaulFertser: maybe you want to join us again, just to lurk Jun 23 19:25:00 some of you already obtained N9(50) devices? Jun 23 19:26:00 Isn't it will be released only on Semptemer? Jun 23 19:26:39 Hopefully MeeGo will be ported to Openmoko phones... Jun 23 19:29:41 if somebody will do it - yes, but I don't excpect that somebody will Jun 23 19:33:07 Meego looks really nice but it won;t be as open as Openmoko I assume... Jun 23 19:33:14 *won't Jun 23 19:34:43 MeeGo as software - very open, go and download anything Jun 23 19:34:49 It will probably work better as an actual phone though since Nokia phones at least work... Jun 23 19:35:21 The software is open but the implementation probably won't be (e.g. non-free drivers). Jun 23 19:36:05 (My Openmoko works as well although my previous Nokia was more stable) Jun 23 19:40:48 Perhaps this talk is more suitable for #meego and #meego-bar... Jun 23 20:16:47 meego needs some special gfx Jun 23 20:16:59 that's how intel defines meego Jun 23 20:17:15 no way that will run on OM FR Jun 23 20:18:21 Will it run on GTA04? Jun 23 20:28:39 aloniv, I don't think that anything that can be run without porting on anything Jun 23 20:29:01 I think its possible to run meego on any device, but it need some work Jun 23 20:29:35 meego (touchframework, etc) itself just works on top of qt, which works on FR and many other devices Jun 23 20:29:42 I thought Openmoko's hardware doesn't satisfy minimum requirements... Jun 23 20:30:38 minimum requirements == somebody lazy to do some profiling Jun 23 20:31:26 of course a lot of animations should be disabled for some devices Jun 23 20:37:05 I meant that it needs to be recompiled for ARMv4 Jun 23 20:39:34 recompilation itself it is a smallest problem I think **** ENDING LOGGING AT Fri Jun 24 02:59:57 2011