**** BEGIN LOGGING AT Sun Sep 30 02:59:56 2007 Sep 30 02:59:58 btw, any ATT SIM users getting a CME 30 error on attempts to register? that is what i get on my unicel SIM Sep 30 03:00:15 Next, we just need to get it working, right? Sep 30 03:00:21 It's worrisome because it's in response to the initialization of the GSM modem (CFUN=1). Apparently it can't start that to even attempt to register. I'd check for SIM problems, sagacis Sep 30 03:00:59 Sorry, it works usually, and ALWAYS in another phone, or using qtopia. I only have this problem in moko-land Sep 30 03:01:06 sagacis: Yes, your gsmd is working correctly. It seems that your GSM modem, however, is not. Sep 30 03:01:24 CME 30 is a "no network service" error Sep 30 03:01:28 If I cfun AFTER boot (i.e. manually), it works OK. Sep 30 03:01:42 Ah, that's good info. So there's something wrong with the initialization sequence, probably related to something about your SIM I would expect? Sep 30 03:01:58 Maybe. Sep 30 03:02:12 It was starting and registering all the time 2-3 weeks ago. Sep 30 03:02:19 I had the gsmd stop/start in the matchbox session. Sep 30 03:02:19 rwhitby was mentioning that he too had to restart gsmd after boot to get it to register. Sep 30 03:02:27 oy. Sep 30 03:02:41 If you put the restart in the matchbox session, does that fix it? Sep 30 03:03:07 * mwester wonders if cesarb knows anything about this sort of problem... Sep 30 03:03:39 * technil guesses that cesarb does :) Sep 30 03:03:41 I wonder if we're just not giving the modem enough time, and its not ready for the power up. Sep 30 03:03:47 mwester: the session restart no longer works, although if I restartX after booting (thereby restarting gsmd), it seems to work OK. Sep 30 03:04:01 restarting X shouldn't restart gsmd. Sep 30 03:04:26 If it does, that would probably be a bug, I think. Sep 30 03:04:35 mwester: Unless I put gsmd restart in /etc/matchbox/session Sep 30 03:04:51 Ah, ok Sep 30 03:05:16 heh Sep 30 03:05:21 I wish we knew what EXT ERROR: 5 really meant. Anyone have a manual for the TI GSM modem? Sep 30 03:05:41 OK, after restarting gsmd and restarting X, the modem is alive and registered. Sep 30 03:05:58 Are the EXT errors chip specific? Sep 30 03:07:47 I seem to recall reading that they are specific ("extended") codes, and that we had to have sent the GSM modem a flag to enable it to respond with those non-standard codes. But that might not apply here, I can't find any info that really looks definitive. Sep 30 03:09:38 All right. I'm putting gsmd stop/start back into my matchbox/session Sep 30 03:09:54 sagacis: what carrier is that SIM for? Sep 30 03:10:05 ATT Sep 30 03:10:21 the old one, probably. Pre-cingular Sep 30 03:10:21 When they were Cingular? Sep 30 03:10:29 Pre-cingular! Wow, that's old. Sep 30 03:10:30 I can read your mind Sep 30 03:10:34 I e-bayed it. Sep 30 03:10:36 hehe! Sep 30 03:10:45 From somebody's back room. Sep 30 03:10:59 "Found under the cabinet of a closed store when cleaning up" most likely. Sep 30 03:11:08 Did you have any trouble getting it activated? Sep 30 03:11:11 Nope. Sep 30 03:11:20 Just read them the number, they said "you're all set" Sep 30 03:12:58 Looking at the timestamps in the log, the modem is idle for 8 seconds (plenty long enough) before the CFUN arrives, but it resonds immediately with the unexpected code. Sep 30 03:13:08 mwester: I have a theory that the gsmd doesn't re-register after a crash/reboot correctly, and it won't connect until the provider times out the connection. Sep 30 03:13:11 Is that possible? Sep 30 03:13:33 Wow. 8 seconds wasted in the boot.... Sep 30 03:14:09 It's not supposed to be possible, because the modem is actually powered off when the CPU resets, and not powered back up until /etc/init.d/gsmd turns it on. Sep 30 03:14:23 I have no idea what the state on the other end (network) is, however... Sep 30 03:14:49 my cellID keeps switching. Sep 30 03:15:04 I'm just sitting at my desk. The phone's not moving. Sep 30 03:15:08 (the wasted time is probably X starting up - the dialer powers up the modem.) Sep 30 03:15:37 I put the gsmd stop/start right before the dialer. Is it not running when I call stop, then? Sep 30 03:15:56 I've seen that with wifi on a laptop; simple fix was to change a setting. I wonder if we have the same control of the GSM modem and switching cell sites. Sep 30 03:15:57 Then, when the dialer comes up, it's already started Sep 30 03:16:41 I seem to get to a lower sig strength, so it switches to a different cell.... Sep 30 03:16:58 but doesn't like that one as much, so when the first's strength comes up, it switches back Sep 30 03:16:59 The way its supposed to work is that daemon starts early, and runs all the time. Clients like the dialer can come and go, and control the connection. But the gsmd process should never exit, and never fail. Sep 30 03:17:14 Right. Sep 30 03:17:48 You said the dialer powers up the modem. gsmd start doesn't, then? Sep 30 03:18:04 Which makes me think that the actual powerup/down of the GSM modem hardware (not the CFUN command) should also be a function performed by gsmd itself. Sep 30 03:18:19 Right, that's what I was thinking. Sep 30 03:18:28 Oh, there are two "power up" things. Sep 30 03:18:45 Oh. CFUN powers up the amplifier, but the modem is already one? Sep 30 03:18:46 on? Sep 30 03:19:37 There's a GPIO line on the CPU that can apply/remove power to the GSM modem chip itself, and there's the CFUN command that can control the number of services (if I read correctly) that the GSM modem is providing. Sep 30 03:20:11 CFUN=0 puts it in a low-power state (almost-off), CFUN=1 resets the GSM protocol stack in the modem, and enables all functions. Sep 30 03:20:38 The other magic CFUN incantations wouild presumably be ways we could use to reduce power and extend battery life. Sep 30 03:20:43 OK. gsmd does none of that, though? (I'm poring through the code now) Sep 30 03:21:06 gsmd does none of that right now. Sep 30 03:22:12 It'll be tricky, I think -- there's a lot of power management that happens in the kernel, and in other processes. They may need ways to signal gsmd to do stuff. Sep 30 03:22:28 Or maybe neod needs to be a client of gsmd as well, and neod can manage the power. Sep 30 03:23:04 neod is probably the right place - the modem has to work when powered off. Sep 30 03:23:05 cfun=2 is disable transmit, cfun=3 is disable receive, cfun=4 is disable both, dfun=0 is trasceiver off Sep 30 03:23:09 I dunno. Sep 30 03:23:38 when the system is powered off. Sep 30 03:23:43 But I think we need to erect a huge SEP over the power management section of the GSM modem, and figure out why it won't respond properly to a simple CFUN=1 Sep 30 03:23:49 that means intimate setup with the power. Sep 30 03:24:10 mwester: You have a lot of blind eyes. ;-) Sep 30 03:24:14 I found it to be unstable until I gave it a couple of seconds Sep 30 03:24:22 before sending it at+cfun Sep 30 03:24:26 Blindness is necessary for debugging. :) Sep 30 03:24:48 Otherwise one would be overwhelmed by gsmd and the state of OpenMoko! :D Sep 30 03:25:06 One step at a time Sep 30 03:25:28 SpeedEvil: 8 seconds between last command the CFUN=1 in sagacis's log... Sep 30 03:26:16 I mean delay before sending cfun - after powering up the modem Sep 30 03:26:40 http://cl1p.net/gsm.log/ Sep 30 03:27:01 Where does the delay need to be inserted? I can add that... Sep 30 03:27:06 this was just with my hacky awk gsmd Sep 30 03:27:17 As I understand it. Sep 30 03:27:24 When turned on. Sep 30 03:27:41 The modem is off until it sees a positive going edge on power_on Sep 30 03:28:02 then give it several seconds before bothering it with commands. Sep 30 03:28:06 There's two seconds total from power up to first "AT" command being issued. Sep 30 03:28:08 That's not logged Sep 30 03:28:17 Right, not logged. Sep 30 03:29:19 sagacis: Edit the /etc/init.d/gsmd script, and change the sleep time after the "echo '1'" to something bigger - 5 or so... let's see what it does :) Sep 30 03:29:20 Also, we should recognize +EXT ERROR as a valid ERROR response and not continue. Sep 30 03:29:30 OK. Give me a few. Sep 30 03:30:23 What's GSM_RES? Sep 30 03:30:51 rebooting Sep 30 03:32:17 GSM_RES is reset. It's disconnected on the BV4 versions. Sep 30 03:32:36 Ah. Gotcha Sep 30 03:32:46 mwester: the GSM modem doesn't actually turn off until you send it an AT@POFF Sep 30 03:33:14 mwester: so, even if the CPU crashes, it's still on Sep 30 03:33:22 This time, the cfun returned "ERROR" I'll clip it. Sep 30 03:33:43 cesarb: That's interesting... and bad. Sep 30 03:33:52 http://cl1p.net/gsm.log2/ Sep 30 03:33:58 A crashed phone will kill the battery quickly. Sep 30 03:34:50 gsmd is rather optimistic: After telling me "Radio power-up failed", it still tries to register on the network. :-) Sep 30 03:34:56 sagacis: Add another few seconds... maybe it'll just respond "ER" or even just "E" :) Sep 30 03:35:09 If I wait 30 seconds, will it invert? Sep 30 03:35:12 mwester: not only that, the audio amp has exactly the same behaviour (won't turn off unless you set "Amp Mode" to "Off", or if you have a patch which is not in the kernel yet AFAIK it will turn off on a clean shutdown) Sep 30 03:35:30 Hmm... Sep 30 03:35:48 Is the patch in bugzilla? Sep 30 03:36:14 I've been popping the battery. That should work. Sep 30 03:36:26 Although my back clamshell is getting looser. Sep 30 03:36:30 * mwester has a growing collection of kernel patches that all seem to be growing moldy in bugzilla... Sep 30 03:36:53 sagacis: the back of mine is actually seldom attached :) Sep 30 03:37:37 That would be safer, but I carry it to work most days. Sep 30 03:37:51 Of course, I don't use it much.... I carry a backup phone, too. Sep 30 03:38:05 Duct tape will hold that back on, should it get too loose. :p Sep 30 03:38:22 * mwester is a huge fan of duct tape. Sep 30 03:38:35 and WD-40 Sep 30 03:38:40 Yup. Sep 30 03:38:48 I could go the iPhone route and super-glue it. Sep 30 03:39:08 If it moves, and it ain't supposed to, use the duct tape. If it doesn't move, but it's supposed to, use WD-40. Who needs toolboxes? Sep 30 03:39:41 mwester: yeah, it's in bugzilla Sep 30 03:40:29 this is odd: Stopping GSM daemon: stopped /usr/sbin/gsmd (pid 1117). -- sh: /usr/bin/cpp: not found --gsmd. (the dashes represent \n) Sep 30 03:40:31 Ok. I have a note to go search for that. Sep 30 03:42:08 good luck bugchasers Sep 30 03:43:05 mwester: 781 Sep 30 03:46:02 Interesting set of comments. And a significant power "suck" when the device is supposed to be off. Sep 30 03:46:19 Hm, I'm so far failing to find more information on google for "EXT ERROR"... only about "CME ERROR" Sep 30 03:46:36 Yeah, google washed out for me too. Sep 30 03:46:48 mwester: disabling both the amp and the modem is enough to make it consume so little when powered off SpeedEvil couldn't measure with his rig Sep 30 03:47:43 mwester: the enfora enabler IIE docs (which tend to be quite close to what neo's modem uses, so close the "engineering mode" commands are the same) also doesn't seem to say anything about it Sep 30 03:49:16 Also, the "EXT: I" on the second log is a mistery too (it's visibly some extended information, and the real error is "ERROR") Sep 30 03:51:49 cesarb: So if we fix those two items alone, many of the ugly battery recovery episodes will just go away... Sep 30 03:52:00 mwester: google for 'ml_parse "ext: I"' returned one interesting result... a gsm log from laforge accidentally in the middle of a huge patch in 2006 Sep 30 03:52:05 ...and I won't have to leave my neo on the umbilical cord overnight all the time... Sep 30 03:52:43 mwester: I leave it both night and day (two different chords) Sep 30 03:55:18 http://www.socialdailynews.com/diverse/not-animated.php <-- stupid brain Sep 30 03:55:43 cesarb: do you know if anyone has done any work on the AT@POFF thing for gsmd? Sep 30 03:56:16 mwester: I think SpeedEvil played with echoing it in the initscript :D Sep 30 03:56:43 bedboi: Mine too http://www.pizdaus.com/pics/6I5muh2Keb4v.jpg Sep 30 03:56:46 no - just echo - no script :) Sep 30 03:56:47 That's kinda what I was thinking of doing as well... Sep 30 03:57:15 for crashes, the best way would be to have a watchdog, which would allow uboot to take over and reboot (linux can then get back to managing everything) Sep 30 03:58:02 How often do we see bad crashes? Usually, I've only seen process crashes. Sep 30 03:58:17 That's a good idea. And even letting uboot take control for a few moments on poweroff to make sure it's all *really* powered off would be reasonable, I think. Sep 30 03:58:56 sagacis: That's why I put that kernel patch together -- really bad crashes were rather frequent for me for a while. :) Now I only crash it once or twice a day. Sep 30 03:59:15 sagacis: could you try things with the "debugging patch" attached to bug 766 (or perhaps with all three)? that debugging patch will make things a lot less muddy... Sep 30 03:59:56 cesarb: do those patches apply to the updated gmsd? Last time I tried, the code was very different in some of those places.. Sep 30 04:00:04 sagacis: both debugging patches shouldn't change anything on gsmd's behaviour (other than being more verbose); the middle patch actually fixes a bug Sep 30 04:00:08 mwester: dunno, didn't look Sep 30 04:00:31 The commit comments imply that they fixed the bug you identified as well. Sep 30 04:00:47 mwester: I recall seeing changes in a completely different place for a very related bug (with user<->gsmd connections, while mine is for modem<->gsmd) Sep 30 04:01:06 Ok, perhaps that was it. Sep 30 04:01:23 But in any case, your patches wouldn't apply when I tried. :( Sep 30 04:04:50 mwester: all three patches applied to me to latest svn Sep 30 04:05:07 mwester: but that's from my local copy of the patches, I'll check the copies at bugzilla Sep 30 04:05:11 OK, patch apply help. Sep 30 04:05:22 The patches are not from the root. They are from gsmd/src or somewhere Sep 30 04:06:54 yep, all three patches are identical to what's on bugzilla Sep 30 04:07:10 sagacis: they are from gsmd's root Sep 30 04:07:41 sagacis: very easy to apply on a local overlay, just add to the SRC_URI ;-) Sep 30 04:07:51 cesarb: but what's built in OE is SVN + a set of OE patches. Sep 30 04:07:58 Did you check the OE patches? Sep 30 04:08:09 mwester: hm, forgot that :P Sep 30 04:08:12 * cesarb goes try Sep 30 04:08:27 the only oe patch I have is a makefile patch Sep 30 04:09:53 10 patches in mine. Sep 30 04:10:52 /u/moko/build/tmp/work/fic-gta01-angstrom-linux-gnueabi/libgsmd-0.1+svnr2957-r31/gsm Sep 30 04:11:58 Yup, me too. Sep 30 04:14:56 mwester: great, the very first patch of these shuffles the code to a completely different place :( Sep 30 04:18:06 mwester: the new code doesn't even make sense in places Sep 30 04:18:25 for instance, in atcmd_submit there's a line "llist_empty(&g->pending_atcmds);" Sep 30 04:18:45 isn't llist_empty a function which returns whether the list is empty, and has NO side effects? Sep 30 04:19:48 In the original code that call was in the conditional of an if, which was part of the select() FD_WRITE logic Sep 30 04:23:42 "statement has no effect" would be the diagnosis from a good static code analyzer. Sep 30 04:25:08 Well, the patch won't work for sure, according to the code I have Sep 30 04:28:08 * mwester just finished building a kernel with the audio power-off patch in it. Sep 30 04:28:14 mwester: statements without effect are often caused by C++-like abstraction levels, so even good static analyzers cannot warn of all of them Sep 30 04:28:38 mwester: and the bug I reported on bug 766 seems to have been fixed (but I cannot tell which patch fixed if) Sep 30 04:28:46 s/ if/ it/ Sep 30 04:28:46 cesarb meant: mwester: and the bug I reported on bug 766 seems to have been fixed (but I cannot tell which patch fixed it) Sep 30 04:28:57 SpeedEvil: any tricks or advice for powering off the modem in the /etc/init.d/gsm script? Sep 30 04:29:02 echo -e 'O\nr' | libgsmd-tool -m shell doesn't make gsmd get completely confused anymore Sep 30 04:31:03 however, all these patches seem to add yet another buffering layer, which means that even if I moved the debug print to where the code now is (it's on atcmd_wake_queue instead of atcmd_select_cb now), it wouldn't be enough Sep 30 04:31:12 how about 'echo "AT@POFF" | libgsmd-tool -m atmcd' or something? Sep 30 04:31:25 the other debugging patch probably isn't needed anymore, the new layer won't lose the CR I hope Sep 30 04:32:09 hm, no, it can still lose the cr if the new buffer is full on a unlucky place Sep 30 04:32:35 sagacis: it'd only confuse gsmd, since the modem is going away... either gsmd itself should do it, or you should do it after killing gsmd Sep 30 04:32:53 Can you pipe to the device directly? Sep 30 04:32:57 sagacis: of course, you might not care about gsmd being confused at that time... Sep 30 04:33:27 O Sep 30 04:33:32 oops Sep 30 04:33:59 I'll have to build some rig so I can measure the current on power-down, so I can see if these patches are working. Sep 30 04:34:10 great. Now that I've powered off the modem, how do I bring it back? Sep 30 04:34:18 Oh,yeah. Nevermind. I'm getting tired Sep 30 04:35:15 sagacis: just echo 1 to that power thingy on /sys Sep 30 04:35:21 sagacis: the init script is enough Sep 30 04:35:37 cesarb: Thanks. I realized what I was asking when I pressed Sep 30 04:36:22 sagacis: that power thingy on /sys is a lot like the neo's power button: press it to turn the modem on, but it doesn't turn the modem off (pretend for an instant neo's power button doesn't have the hold-five-seconds beahviour) Sep 30 04:36:23 It's gotta be echo 1 - when the state of the power thingy is 0 Sep 30 04:36:39 0-1 transition is the only thing that does anything Sep 30 04:36:50 not so useful that way. Sep 30 04:36:55 does a soft-reset of the modem Sep 30 04:37:20 Which is why we don't have the reset any more? Sep 30 04:37:28 sagacis: actually, it's very useful - you can turn the modem on! how would you use it if you couldn't turn it on? ;-) Sep 30 04:37:43 It's half of what we need. Sep 30 04:37:46 I think reset went away as it's going away in 02 hardware Sep 30 04:38:06 OK, I had to gsmd stop/start twice to get it to come back. Sep 30 04:38:09 If it's on, you just send at@poff to kill it Sep 30 04:38:14 after the @POFF Sep 30 04:38:30 sagacis: probably because the first start/stop echoed 0 to the file ;-) Sep 30 04:40:18 it looks like init.d/gsmd echos both 0 and 1 to the file to trigger it. Sep 30 04:40:22 * cesarb notices the only part of the neo which doesn't seem problematic is the bluetooth chip -- rwhitby has even been doing some "look, no hands!"-style play with it (treo ssh over BT to the neo, to make a ipkg upgrade also over BT to a nslu...) Sep 30 04:40:33 actually, scratch that, BT has the wakeup problems too :D Sep 30 04:40:40 cesarb: I can't get BT working with my headset. Sep 30 04:40:55 The usb port has been pretty good though. Sep 30 04:41:18 I bet the GPS chip rocks! Sep 30 04:41:30 Phhhhtttthhhtttt!!!! Sep 30 04:41:40 sagacis: ...with its OABI-and-good-luck-finding-it driver, I'd guess ;-) Sep 30 04:41:54 Don't anyone get me started again... :) Sep 30 04:41:55 the backlight works well. Sep 30 04:42:05 mwesterI was gonna go there Sep 30 04:42:06 multi-level, even Sep 30 04:42:21 runs the battery down real good Sep 30 04:42:45 SpeedEvil: and the screen, and the touchscreen... but that's all the LCM, isn't it? so it only counts as one piece ;-) Sep 30 04:42:46 y'know, init.d/gsmd start does all the things gsmd stop does. start is really a restart. Sep 30 04:42:48 Hmm... a dimmer for it would be nice. Sep 30 04:43:13 GUISettings has a dimmer Sep 30 04:43:19 so, we have USB and LCM working without any stress... Sep 30 04:43:19 or whatever that tool is called Sep 30 04:43:39 sagacis: yes. It shouldn't be; that's just in there so that lazy people can type "gsmd start" instead of "gsmd restart". Sep 30 04:44:13 restart calls stop-start, which is really stop-stop-start Sep 30 04:44:22 Silly, isn't it? Sep 30 04:44:36 mwester: no, the "restart-like" behaviour was added as part of voodoo debugging to make it work Sep 30 04:44:40 But, it has to do that, because the power switch only works when going from 0-1 Sep 30 04:44:51 Actually, the "sleeps" and the diddling with the power on and reset should be done by the daemon, not the init script. Sep 30 04:45:06 There are some nasty race conditions that are still present because of that. Sep 30 04:45:11 mwester: it was believed for a while that turning off and on the modem (yeah, I know that echoing 0 doesn't turn it off...) would cure all gsmd woes Sep 30 04:45:34 mwester: the sleeps are also "let's add some sleeps so it'll work" Sep 30 04:45:46 mwester: they even added echos to the non-connected reset line! Sep 30 04:45:52 cesarb: Then I should just take that out, as self-appointed foster caretaker of the gsmd script. Sep 30 04:46:22 mwester: keep only the echo 1 to turn it on (which actually should be within gsmd itself...) Sep 30 04:46:40 (and in fact gsmd should only do it after being asked to! think airplanes) Sep 30 04:46:48 Ooh. good point. Sep 30 04:46:55 cesarb: You should keep the 0 if the echo 1 only works from 0 state. Sep 30 04:47:15 sagacis: good point Sep 30 04:47:37 sagacis: if it were within gsmd, it could simply read the file to check if it's on 0, and if not echo 0 and wait Sep 30 04:47:38 Turning on the modem doesn't actually cause it to emit RF - only once you've done at+cfun Sep 30 04:47:53 (script) wiki RecentChanges: [[Gsmd]] Sep 30 04:47:53 right Sep 30 04:48:06 SpeedEvil: hm, good point... it's still very pointless to do so Sep 30 04:48:16 cesarb: the touchscreen is noisy. Sep 30 04:48:17 The kernel driver should do the 1->0->1 transistion as necessary. Sep 30 04:48:39 the touchscreen, or the sound attached to it? Sep 30 04:48:56 Wow, it squeals Sep 30 04:49:27 Say what? Sep 30 04:49:31 mwester: no. that would no longer be a direct mapping to the GPIO pin Sep 30 04:49:43 mwester: which means it would be a completely different sysfs file Sep 30 04:49:58 mwester: much like the power_sequence one Sep 30 04:50:10 mwester: which does a complex dance when asked to power on Sep 30 04:50:20 There's no point in there being a direct mapping, the kernel driver already diddles with the serial port switch and disables/enables the console driver, all from that same code. Sep 30 04:50:27 * cesarb wonders how many hours did harald have to waste to implement that Sep 30 04:50:35 Not enough. Sep 30 04:50:43 It should have been done in the serial driver. Sep 30 04:51:41 I've been toying with moving the GSM power-up code into the serial driver for a while. It simplifies a lot, because there's way too much cross-referencing of each other's data. Sep 30 04:51:45 mwester: I meand the power_sequence thing (for the gps) Sep 30 04:51:55 For the gps or gsm? Sep 30 04:51:55 s/meand/meant Sep 30 04:52:20 mwester: the gps has a power_sequence file instead of a simpler power_on, because its power sequencing requirements are finicky Sep 30 04:52:42 Ah, ok - sorry, thought you meant the GSM stuff. Sep 30 04:53:25 I don't know about the GPS, it doesn't work here. :( Sep 30 05:01:45 Most phones get their time/date from the signal, yes? Sep 30 05:02:22 sagacis: there are some annoying ones that don't Sep 30 05:02:37 besides the neo? Sep 30 05:03:10 sagacis: one of my relatives has an inexpensive one which doesn't. Another one has a normal one which does. And my razr does. Sep 30 05:03:23 sagacis: I've also seen a motorola (nextel probably) which didn't Sep 30 05:03:50 sagacis: the neo probably wouldn't need it, since it has the much more precise GPS signal (however, that means you need a geographical TZ database) Sep 30 05:04:26 (...or a cell provider/LAC<->TZ database, which should be much easier to make precise enough) Sep 30 05:04:28 And a gps driver. I'm looking for the AT command to get that information, or does it come from a header? Sep 30 05:09:37 sagacis: I couldn't find an AT command to get it... however, it might be a cell broadcast Sep 30 05:09:43 * cesarb does some google research Sep 30 05:11:58 I though it was one of the unsolicited commands, actually... Sep 30 05:12:04 I guess that would be a broadcast. Sep 30 05:12:38 mwester: did you mean undocumented? Sep 30 05:12:44 btw, using the gsm tower time isn't always a good idea. it tends to be wrong by an hour for about a week around the time of a summer time change. Sep 30 05:13:20 (I always end up enabling auto-update to get the minutes correct, disabling it, and adjusting the hour manually) Sep 30 05:13:38 bluelightning: No, unsolicited -- that's a term used in libgsm source. Sep 30 05:13:50 oh right Sep 30 05:14:05 * mwester boots his shiny new power-saving kernel... Sep 30 05:14:27 ah, btw, the one phone which doesn't have auto-update for the time... doesn't have a RTC battery too! (remove the battery, and it'll ask you for the time next boot) Sep 30 05:16:17 AT%CTZV=1 is what you want for unsolisited timezone and date info Sep 30 05:16:24 hm, there is an AT+CCLK on the gsm standards... but that's probably only for the modem's RTC Sep 30 05:16:49 ljp: cool, does it work on the neo? Sep 30 05:17:29 ljp Thanks! Sep 30 05:17:40 Yes Sep 30 05:17:50 I just called it on the neo and got an OK Sep 30 05:18:02 Now, I just wait to get the timezone info, I guess Sep 30 05:18:30 * mwester bets ljp would know what "EXT ERROR: 5" might mean in response to CFUN=1 Sep 30 05:18:41 not sure if it works, i dont htink my gsm privider sends that info. my boss's is supposed to Sep 30 05:19:02 i might if I was at my desk Sep 30 05:19:30 It's in the gsm.log, so we're already calling it. Sep 30 05:19:41 there is also a gsm spec for time zone stuff Sep 30 05:27:22 The network name time and timezone information will normally be transferred from the network to the ME: 1)Upon registering on the network. 2)When the MS geographically relocates to a different Local Time Zone. 3)When the network changes its Local Time Zone, e.g. between summer and winter time. 4) When the network changes its identity. 5) At any time during a signalling connection with Mobile Station. Further details of this feature Sep 30 05:27:26 From the spec Sep 30 05:28:16 http://www.3gpp.org/ftp/Specs/html-info/0242.htm is the timezone spec Sep 30 05:31:51 g'night. Sep 30 06:47:53 (script) wiki RecentChanges: [[GSM02.07]] Sep 30 07:01:47 whats the easiest way to recover a USB hardisks partition table, this is the first time I have seen one break Sep 30 07:02:15 there is one partition on the hardisk Sep 30 07:07:45 actually it is quite common Sep 30 07:13:51 Linux_Galore: testdisk and (if it doesn't work) photorec Sep 30 07:17:26 yeah Im installing testdisk now Sep 30 08:47:42 moin Sep 30 08:47:53 (script) wiki RecentChanges: [[Wishlist:BuiltInScriptingLanguage]] Sep 30 08:48:09 my mokomakefile is deny building again Sep 30 08:48:29 qemu stopps with unable to find a register to spill in class `GENERAL_REGS' Sep 30 08:48:45 or more detailed http://rafb.net/p/YINSpO79.html Sep 30 08:48:54 has anyone encounterd this error before? Sep 30 08:57:04 never mind, i found excatly the same error here: http://wiki.openmoko.org/wiki/Talk:OpenMoko_under_QEMU Sep 30 09:38:37 is anyone able to run the 29th build from buildhost.automated.it in qemo? Sep 30 09:38:38 ifchinscouldkill: See http://wiki.openmoko.org/wiki/FAQ (and also SH1, SH1_FAQ, Neo1973, Developer_preview and Neo1973_Hardware#GTA01Bv4 pages) (Hopefully these links answers your question. This autotrigger for new nick and question during first 15 minutes done only once.) Sep 30 10:23:56 k Sep 30 10:42:37 if i wanna reinstall openmoko OS in neo1973, should i buy the advanced kit ,? is the base kit enough? Sep 30 10:42:38 winux: See http://wiki.openmoko.org/wiki/FAQ (and also SH1, SH1_FAQ, Neo1973, Developer_preview and Neo1973_Hardware#GTA01Bv4 pages) (Hopefully these links answers your question. This autotrigger for new nick and question during first 15 minutes done only once.) Sep 30 10:45:30 if you don't know that you need the advanced kit, you don't Sep 30 10:45:53 winux: you don't need the advanced kit Sep 30 10:46:10 and if you do, find someone near you who has it to unbrick the phone if you managed to do that Sep 30 10:46:33 wow how come qtopia is so fast in qemu? Sep 30 10:47:36 * ljp thinks qemu is really slow Sep 30 10:48:13 (script) wiki RecentChanges: [[Buying_Interest_List]] [[FreeBSD]] [[Talk:FreeBSD]] [[Talk:OpenMoko_under_QEMU]] [[IPhone/it]] Sep 30 11:02:27 hi Sep 30 11:08:21 how often does the buildhost.automated.it update the openmoko image? Sep 30 11:47:59 IfChinsCouldKill: I update most days that there have been changes Sep 30 11:48:19 s/update/update it/ Sep 30 11:48:19 ScaredyCat meant: IfChinsCouldKill: I update it most days that there have been changes Sep 30 11:56:32 Ah ok. The 29th build doesn't boot in qemu so I was curious when to expect the next build ;) Sep 30 12:00:01 I'll take a look at it, jsut flashing it to my neo Sep 30 12:09:02 yup - seems dbus changes broke it Sep 30 12:09:10 will build a new one Sep 30 12:09:27 cool thanks Sep 30 12:46:26 gah, no... Evolution SUCKS. Sep 30 12:46:45 wonderful, that apt-get nowadays is able to autoremove dependencies! Sep 30 12:46:47 You creationist? Sep 30 12:47:22 haha, loL! no.. i love mutts. Sep 30 12:47:38 .oO( now it becomes weird ) Sep 30 12:47:54 (script) wiki RecentChanges: [[User:Alessandro]] Sep 30 13:08:21 now with gtk2 bindings available! http://chmeee.dyndns.org/om/ruby/feed Sep 30 13:09:33 as well as the scripts and settings that I used to compile it all. Sep 30 13:27:36 How long does 2007.2 take to boot? :) Sep 30 13:28:50 long :) You should see progress, most of the time, tho. Sep 30 13:29:58 Kero: Haha, I was wondering because I dont have a Moko yet Sep 30 13:30:03 Waiting for the GTA02 Sep 30 13:30:30 How responsive is it in the real Neo1973? Sep 30 13:30:38 Using over qemu and pretty darn slow :P Sep 30 13:32:12 oh, quicker on the real Neo. Sep 30 13:32:38 A lot? Sep 30 13:32:43 It's VERY laggy ;) Sep 30 13:33:30 depends on the HW you run qemu on :) Sep 30 13:33:46 starting applications is slowish Sep 30 13:33:50 1.66ghz Core 2 Duo, 1gb ram, Linux Sep 30 13:33:56 On the NEo? Sep 30 13:34:01 yeah Sep 30 13:34:04 using the moko-gtk-theme over the old pixmap theme helps a lot Sep 30 13:34:14 gtk responds quickly enough. Sep 30 13:34:35 I am looking to port Sonata MPD client to the Moko =) Sep 30 13:34:36 I have the feeling quite a few stylus-clicks disappear (dspite the feedback sound) Sep 30 13:34:51 No idea how yet Sep 30 13:34:54 gta02 will have a faster CPU for you :) Sep 30 13:34:55 But I want to try =D Sep 30 13:34:59 Yup Sep 30 13:35:12 Me and money is sorta making me =( Sep 30 13:35:23 I cross-compiled Ruby, using the OE/OM environment Sep 30 13:35:56 easy enough (as cross compiling goes...) Sep 30 13:36:23 Isn't PyGTK already on the Moko? Sep 30 13:38:58 ~botmail zecke how about http://en.pastebin.ca/raw/720569 and http://en.pastebin.ca/raw/720570 for openmoko-feedreader2, first one is to fix the build on gcc 4.2 Sep 30 13:40:14 nox-Hand: probably. I prefer Ruby. the point in this case was, porting isn't too hard :) Sep 30 14:40:00 good afternoon Sep 30 14:40:09 is this a developers-only channel? Sep 30 14:40:09 joolz: See http://wiki.openmoko.org/wiki/FAQ (and also SH1, SH1_FAQ, Neo1973, Developer_preview and Neo1973_Hardware#GTA01Bv4 pages) (Hopefully these links answers your question. This autotrigger for new nick and question during first 15 minutes done only once.) Sep 30 14:40:20 hehe Sep 30 14:40:22 will do Sep 30 14:40:34 developers and general talk Sep 30 14:40:42 ok Sep 30 14:40:52 i do development, though mainly db work Sep 30 14:41:13 G'morning. Sep 30 14:41:23 i'm certainly no openmoko devver, but i'm very interested in neo / openmoko as a consumer Sep 30 14:44:03 joolz: We're not really to the consumer level yet. Sep 30 14:45:13 * SpeedEvil giggles at the understatement. Sep 30 14:45:34 Shouldn't neod be on the dbus? Sep 30 14:45:59 It's going to be at the very least several months before Openmoko is user ready, for even the most uncritical of users. Sep 30 14:46:42 We of course will answer sane questions. Sep 30 14:47:14 But most user questions - other than on the hardware are simply 'it's not implemented yet, and no there isn't a timescale'. Sep 30 14:47:33 (script) wiki RecentChanges: [[User:Paulzag]] [[OpenMoko_Local_Groups:_Sydney]] Sep 30 14:52:32 oops Sep 30 14:52:53 I made the vibrator pulse... forever. Apparently I can write an infinite loop. :( Sep 30 15:15:59 good morning everyone. Sep 30 15:16:11 DukeOfURL: howdy Sep 30 15:17:15 I don't have /var/lib/dbus/machine-id in my rootfs, and this means dbus-uuidgen wasn't run. Does anyone know about this? Sep 30 15:17:36 dbus-uuidgen isn't in rootfs either Sep 30 15:22:03 however, dbus-uuidgen is in ./build/tmp/staging/i686-linux/bin/dbus-uuidgen on the development system Sep 30 15:24:34 Sorry. I'm new to dbus. I'll take a look at my neo when boot completes. Sep 30 15:25:45 I have machine-id on the neo Sep 30 15:26:29 for some reason the 'make openmoko-devel-image' isn't including dbug-uuidgen Sep 30 15:27:26 That seems like an oversight. Sep 30 15:29:16 i get these during the build: ERROR: QA issue: libdbus-1.la failed sanity test (reference to workdir or installed) Sep 30 15:29:41 I get so many QA errors. I haven't looked for libdbus errors. Sep 30 15:31:23 when I 'make rebuild-package-dbus' I get: NOTE: the following files were installed but not shipped in any package: Sep 30 15:31:33 NOTE: /usr/bin/dbus-daemon Sep 30 15:31:43