**** BEGIN LOGGING AT Wed Apr 30 03:00:02 2008 Apr 30 05:14:34 Hi all. I'm trying to play a video with sound on gta01. Sound seems to only work with mplayer when pulseaudio is stopped. This seems to prevent my phone from playing the ring sound (but I guess the vibrate mode still shows there's an incoming phone call). Does anyone have a solution to play a video with sound but still allow sounds like the phone call ring to work? Apr 30 05:31:43 when did they change the name and why? Apr 30 05:32:18 is the freerunner going to be the "offical" release? Apr 30 05:33:56 ? Apr 30 05:34:06 ah, I see Apr 30 05:34:39 I mean a name change from neo-1973 Apr 30 05:36:07 neo-1973 is a separate piece of hardware Apr 30 05:36:09 different Apr 30 05:36:11 different cpu Apr 30 05:37:00 yeah... I'm reading that Apr 30 05:44:55 different amount of flash Apr 30 05:44:59 a wifi chip added Apr 30 05:45:03 glamo fx chip added Apr 30 05:45:06 A speaker removed Apr 30 05:45:11 yeah Apr 30 05:45:15 it happens to be in the same case Apr 30 05:45:16 And it's got shiny LEDs! Apr 30 05:45:18 and same screen Apr 30 05:45:25 For running-light effects. Apr 30 05:45:28 and same gsm chip Apr 30 05:45:31 same bluetooth Apr 30 05:45:40 it has a new gps module Apr 30 05:45:44 etc. etc. Apr 30 05:45:51 It's black! Apr 30 05:46:47 after I get one it will have apachy and gcc on it :) Apr 30 05:51:44 any good speculation estimate on when the new one will be sold? Apr 30 05:52:19 (the freerunner) Apr 30 05:52:37 this year Apr 30 05:52:42 the best estimate so far :) Apr 30 07:21:20 good morning Apr 30 09:07:18 openmoko: 03thomas * r4410 10/trunk/src/target/opkg/ (libopkg/opkg.c libopkg/opkg.h tests/libopkg_test.c): opkg: implement opkg_list_upgradable_packages function Apr 30 09:57:30 can someone suggest me a fairly stable image. i would need to use my neo for a few days Apr 30 10:25:00 mwester: can you tell me more about http://lists.openmoko.org/pipermail/openmoko-kernel/2008-April/002596.html ? when does that happen? Apr 30 10:26:46 mwester: and would any hardware tests be of use Apr 30 10:30:50 i wrote the muxer and a pygsmd btw, and i would expect to see such a problem early. i encountered just one effect: the initial message of the modem ("AT-Commandinterpreter ready" or similar) is sometimes cut after some chars... Apr 30 10:34:23 is someone able to fix this: "run-parts: /etc/dbus-1/event.d/20hal exited with return code 1" when restarting system dbus? Apr 30 10:43:07 openmoko: 03reddog 07andy * r0b302f8479ac 10/drivers/video/s3c2410fb.c: gta01-fix-resume-redo-par.patch Apr 30 10:43:07 openmoko: 03mwester 07andy * r308ad84b8eb3 10/arch/arm/mach-s3c2410/mach-gta01.c: gta01-fix-jbt-platform-missing-members.patch Apr 30 10:43:08 openmoko: 03mwester 07andy * r76923000356b 10/arch/arm/plat-s3c24xx/neo1973_pm_gps.c: gta01-gps-power-state-resume-preserve.patch Apr 30 10:43:09 openmoko: 03reddog 07andy * r1ac9f711e4be 10/sound/soc/s3c24xx/neo1973_wm8753.c: This-fixes-the-exit-function-so-that-the-module-can.patch Apr 30 10:43:11 openmoko: 03reddog 07andy * r9123dfb5aab1 10/sound/soc/s3c24xx/ (Kconfig neo1973_wm8753.c s3c24xx-i2s.c s3c24xx-pcm.c): This-adds-some-debug-messages-to-the-Neo1937-sound-d.patch Apr 30 10:43:14 openmoko: 03matt 07andy * rd3334b473345 10/drivers/sdio/hcd/s3c24xx/ (s3c24xx_hcd.c s3c24xx_hcd.h): From 2d727270e57c25044d112c6ad9807ddc07d5199a Mon Sep 17 00:00:00 2001 Apr 30 10:51:45 the PIDFILE is wrong, should be PIDFILE=$PIDDIR/pid - can someone change that? Apr 30 10:58:17 emdete: yo Apr 30 10:58:22 how goes your hacking? Apr 30 11:00:33 raster: i'm stuck ;) just doing muxer&gsm stuff. and my gta02 freezes - not shure when. so i lost much time in nothing fancy new. Apr 30 11:01:00 aaah fun Apr 30 11:01:02 isn't it Apr 30 11:01:08 raster: is there any fancy new stuff from your side? (not illume, more driver level)? Apr 30 11:01:36 howdy guys Apr 30 11:02:53 raster: i am not shure why my gta02 freezes but it happens often using grafic&sd together. have you played around with sd when doing grafic? Apr 30 11:05:21 raster: and i would like to do some stuff with scrolling text. i use edje's textblock (which formats the text fine) - how would be the simplies way to scroll in that text if it's higher than the area? how can i determine how high a block gets for a text in a given width? Apr 30 11:06:55 emdete: u mean like x? Apr 30 11:06:55 :) Apr 30 11:07:25 (driver stuff) Apr 30 11:07:44 as for sd+ gfx - not a lot at all really, but i could believe it. sd and gfx share the same bus - hell, the same chip. Apr 30 11:08:08 and as for textblock Apr 30 11:08:19 u can get the formatted text sice from the textblock Apr 30 11:08:30 that gives u the text formatted given withe textblock width and height Apr 30 11:08:40 formatted size may be taller or shorter than the evas textblock object Apr 30 11:08:49 thus u know if the object should be made taller to fit the text Apr 30 11:09:02 and thus u can resize the textblock to be taller - and then juts move it up and down to scroll Apr 30 11:09:03 :) Apr 30 11:09:44 raster: cool, have to check how to access it in py. sounds great. it's anoying to read the first lines of news only all the time ;) Apr 30 11:12:22 aaah Apr 30 11:12:30 rememebr i'm always thinking o the c interface Apr 30 11:12:38 shure, i know. Apr 30 11:12:44 as i wrote that and intended it to do pretty much everything u need.. somehow Apr 30 11:12:46 :) Apr 30 11:13:34 the only thing thats missing is word-break ... not full hyphenation but a hard cut as we discussed some weeks ago... Apr 30 11:14:33 word-break exists Apr 30 11:14:35 whe i wrote a similar thing based on dfb i included picture support. you could put base64ed-pictures into the text, mark it and it rendered it at that place... Apr 30 11:14:40 u put it in the formatting Apr 30 11:15:09 raster: break at word boundaries, shure, but it does not break words longer than a line Apr 30 11:15:32 and: it puts a empty line after a line with a word that was too long Apr 30 11:15:37 oooh Apr 30 11:15:39 that Apr 30 11:15:40 hmm Apr 30 11:15:43 i think it just refuses Apr 30 11:15:47 its a bit stupid there Apr 30 11:15:55 from memory i dont fall back to character-wrap Apr 30 11:16:04 http://scap.linuxtogo.org/files/2ec43d94a5a47f8235d6a9e6439b9b32.png Apr 30 11:16:11 doing wrod-breakign would mean it needs to onow thelanguage and rules on hypenation Apr 30 11:16:14 and no way was i going to try that Apr 30 11:16:15 :) Apr 30 11:16:16 see the line after the url? Apr 30 11:16:43 shure, but it should not include an empty line Apr 30 11:16:49 yeah Apr 30 11:16:53 its just refusing to break it Apr 30 11:16:53 and: hard break would be good anyway Apr 30 11:17:02 so it renders off into lala-land on the right Apr 30 11:17:14 it's lost :( Apr 30 11:17:14 so u'd prefer it fall back to char-wrap then? Apr 30 11:17:27 well if u resized the textblock to be wider it'd be there Apr 30 11:17:31 just cut it at the right boarder Apr 30 11:17:32 if u ask it for its formatted size Apr 30 11:17:36 it'd be wider than the obj u have Apr 30 11:17:45 indicating is just can't get smaller than its formatted width Apr 30 11:17:50 it should do that anyway Apr 30 11:18:15 and pictures - i did plan on doing that in tb Apr 30 11:18:17 eventually Apr 30 11:18:20 not as pictures as such Apr 30 11:18:23 bu "obstacles" Apr 30 11:18:23 resizinf that would need a soldering iron Apr 30 11:18:31 u can slide in arbitrary blocks Apr 30 11:18:45 with either a callback or another evas object handle Apr 30 11:18:56 actually callbaks i think i planned Apr 30 11:18:56 that would be cool, yes Apr 30 11:19:09 callbacks would also be great for hyphenation.... Apr 30 11:19:15 then u can just hook up a few generic callbacks and move/resize an image or whatvere to be in the right spot Apr 30 11:19:53 but back to the 2 problems: the empty line is not intended by me. its made because the word did not fit Apr 30 11:20:28 yeah Apr 30 11:20:40 i think its a bi-product of not resizing to be wide enough Apr 30 11:20:44 and i would sugest to do a hard cut at the right border if a word does not fit. i am interested in the content Apr 30 11:20:48 as i think i intended that to be what u do to fit Apr 30 11:20:58 there is no (non-hw) way of resizing ;) Apr 30 11:21:40 well in this case u'd need a scrollable region Apr 30 11:21:49 thsat can scroll horizontally Apr 30 11:21:52 i don't want to ;) Apr 30 11:21:59 as such url i would actually make a separate field Apr 30 11:22:01 :) Apr 30 11:22:08 but as such Apr 30 11:22:18 the extra line is annoying Apr 30 11:22:26 no, thats just a sample Apr 30 11:22:47 there are ugly long words in german that do the same. Apr 30 11:22:57 sheiße Apr 30 11:23:05 the url is just for reference (more debugging for me) Apr 30 11:23:07 merde Apr 30 11:23:10 o u'd just want char-wrap then Apr 30 11:23:15 yes Apr 30 11:23:22 in case of long words Apr 30 11:23:27 yeah i dont have a fallback if it doesnt fit Apr 30 11:23:33 its kind of an unhandled case Apr 30 11:25:19 ui need to set aside some time to go into the code again and find that case Apr 30 11:27:02 u or i ? :D Apr 30 11:27:06 me Apr 30 11:27:48 that would be really, really good stuff and would help me alot. :D Apr 30 11:48:18 hello :) Apr 30 11:48:50 ccfly: hi Apr 30 11:51:44 hi edistar, ccfly Apr 30 12:26:52 Marcel_M: hast du mal wieder gebuilet in der letzten zeit? Apr 30 12:27:57 ich kann nicht mehr bauen seit einigen tagen, weil ein patch von imlib2 fehlschlaegt Apr 30 12:28:20 emdete: no 20080422 is the latest working one (fgau told me) Apr 30 12:28:38 image, klar, und ipk? Apr 30 12:28:47 /kannst/ du gerade builden? Apr 30 12:29:04 no tryed today Apr 30 12:29:14 warum muss man sich immer nur mit so nem dreck rumschlagen?!? Apr 30 12:29:26 hab ja cron... der tried jeden tag :/ Apr 30 12:30:36 oups, wrong languag | wrong channel... shit, sry Apr 30 12:33:35 openmoko: 03matt_hsu 07andy * r0a4bc69cd721 10/drivers/sdio/hcd/s3c24xx/ (s3c24xx_hcd.c s3c24xx_hcd.h): From cc08b5986dfd8d971ee46ce7045fb7863f99a92a Mon Sep 17 00:00:00 2001 Apr 30 12:33:36 openmoko: 03agreen 07andy * r15467dfc64ac 10/drivers/sdio/hcd/s3c24xx/s3c24xx_hcd.c: clean-sdio-hcd-suspend.patch Apr 30 12:34:29 openmoko: 03matt_hsu 07stable * r463f236121c7 10/drivers/sdio/hcd/s3c24xx/ (s3c24xx_hcd.c s3c24xx_hcd.h): From cc08b5986dfd8d971ee46ce7045fb7863f99a92a Mon Sep 17 00:00:00 2001 Apr 30 12:34:30 openmoko: 03agreen 07stable * r401eb535cb94 10/drivers/sdio/hcd/s3c24xx/s3c24xx_hcd.c: clean-sdio-hcd-suspend.patch Apr 30 12:35:14 openmoko: 03matt_hsu 07andy-tracking * rfdc887813036 10/drivers/sdio/hcd/s3c24xx/ (s3c24xx_hcd.c s3c24xx_hcd.h): From cc08b5986dfd8d971ee46ce7045fb7863f99a92a Mon Sep 17 00:00:00 2001 Apr 30 12:35:14 openmoko: 03agreen 07andy-tracking * r68a96163573e 10/drivers/sdio/hcd/s3c24xx/s3c24xx_hcd.c: clean-sdio-hcd-suspend.patch Apr 30 13:03:23 emdete: what specifically do you wish to know about the flow control problem, and does http://bugzilla.openmoko.org/cgi-bin/bugzilla/show_bug.cgi?id=1376 have more information you might find helpful? Apr 30 13:04:57 Can the modem be run at 19200? Apr 30 13:05:01 emdete: Also, you will probably find the kernel at http://moko.mwester.net/ to be better able to detect this problem when it occurs; enable the "nspy" debug, suspend/resume a few times, waking with a call or SMS, and see if you don't run into an overrun. Apr 30 13:05:06 mwester: i assume to see lost chars in muxer and/or commands in my daemon set. i don't (as far as i know). so question is what are the special circumstances you saw these? Apr 30 13:05:48 SpeedEvil: There is no documented way; cesarb investigated this because the 115200 baud limited the CPU speeds possible (because of the clock used by the UART). Apr 30 13:06:10 ah - right - I knew that. Apr 30 13:06:30 However, it might be possible to firmware configure it with other speeds. Apr 30 13:06:52 emdete: The ideal case for finding this problem is to modify the kernel to hold the GSM flow-controlled during suspend, release flow-control back to the UART in either the serial driver or the neo1973_pm_gsm module. Apr 30 13:07:22 When the unit wakes because of GSM traffic waiting, as soon as flow control is released, the GSM will send the entire buffer. Apr 30 13:07:34 Doesn't it need to have flow control not on in order that the GSM module can send data to the SoC? Apr 30 13:07:41 It will not honor the flow control during the send of the message; it seems to honor flow control only between messages. Apr 30 13:08:56 SpeedEvil: I mean to say that during suspend/resume, one needs to transform the flow-control pin to a GPIO, and hold the GSM flow-controlled, this will make it wake the CPU with the interrupt line. Apr 30 13:09:28 ah - htere is a seperate interrupt line? Apr 30 13:09:30 Then when the serial port is configured after resume, one needs to switch the GPIO back to the UART's flow-control hardware in order to let the UART manage the flow of dta. Apr 30 13:09:40 SpeedEvil: Yes. Apr 30 13:09:46 Diddn't know that Apr 30 13:09:57 It is rather odd. Apr 30 13:11:26 emdete: Do you communicate with the GSM using AT commands, or does it use some other protocol for your gsmd? Apr 30 13:11:30 mwester: what does "the message"? one package when in muxer mode, one line (until cr/lf) in command mode? Apr 30 13:11:57 mwester: i use my muxer which hasgsm07.10 binary protocol Apr 30 13:12:21 max size of one paket is 100 bytes for the ti chip as far as i remember Apr 30 13:12:45 I assume 07.10 doesn't have something as nice as 'reset last packet' - which'd nicely sidestep the problem? Apr 30 13:12:49 resend Apr 30 13:12:59 That may be significant; there may be different behavior in the gsm07.10 binary mode. I have only tested in AT command mode, where the message is arbitrary length and terminates with Apr 30 13:13:23 emdete: you're testing in suspend mode wakeup? Apr 30 13:14:24 One reason this may have gone undetected so far is that it seems that the messages from the GSM are very short; many of them are 16 chars long and are just CSQ (signal quality) measures. They fit in the FIFO. Apr 30 13:17:34 I was very hopeful that the kernel and patches I put up on my site would fix the problem, and the latest change was to instruct the GSM to not send unsolicited notices on registration changes or signal quality -- this worked, but the result was that the only messages that would wake it were the longer SMS and incoming call messages... and the problem came back. Apr 30 13:18:57 mwester: I think I had an issue on my journey to the uk when in suspend and an sms came it... Apr 30 13:19:03 What does an incoming call look like? Apr 30 13:19:05 screen when white and that was it Apr 30 13:22:15 but it worked pretty well (~9.5 hours before the whole thing stopped responding) Apr 30 13:22:50 woke up half a dozen times for 'no reason' Apr 30 13:23:32 scared the shit out of me a couple of times (travelling at night -/me thinks white flashes == speed camera) Apr 30 13:24:54 SpeedEvil: no. i dont user linux-side-suspend Apr 30 13:26:25 emdete: ah - that's the entire problem. Apr 30 13:26:34 SpeedEvil: yes, understand that now. Apr 30 13:26:41 emdete: in normal use, it's just fine. OK. Apr 30 13:27:23 Well, we don't *know* that it's ok in normal use... Apr 30 13:27:38 Just that suspend/resume sets up the conditions to make the problem expose itself. Apr 30 13:27:50 mwester: I pushed a gigabyte over GPRS over a couple of days. Apr 30 13:27:52 I have observed it twice outside of suspend/resume. Apr 30 13:28:07 mwester: with no wierd hands. Apr 30 13:28:35 Does GPRS use a binary protocol as well? Apr 30 13:29:51 * mwester thinks GPRS does not use AT commands, and that perhaps this problem is specific to AT commands and responses being buffered differently? Apr 30 13:30:36 gprs is ppp Apr 30 13:30:51 * mwester does a double-take, and re-reads SpeedEvil's comment... Apr 30 13:31:05 i use gprs/ppp over the muxer Apr 30 13:31:06 A gigabyte??? Gack! That must have cost more than the phone!~ Apr 30 13:32:10 mwester: Not for people with flatrates Apr 30 13:32:26 Ok, evidence is building to support that this is limited to AT command processing; an observation I've made when analyzing the captured logs was that when an overrun occurs, it *never* drops just a few characters; it always drops all the characters to the end of the message. Apr 30 13:33:17 mwester: free actually Apr 30 13:33:28 mwester: It was free for a couple of days Apr 30 13:33:41 10 days actually Apr 30 13:33:47 Nice promotion! Apr 30 13:34:08 Also - in the t-mobile pay-as-you-go plan I'm on - data is never more than a pound a day. Apr 30 13:34:18 Or I can buy a web pass for 5 days for 2.50 Apr 30 13:34:44 * cesarb sees his konversation icon blink in the corner of the screen Apr 30 13:35:27 Yes, we invoked your name :p (re: baud rates for the GSM, and the fact that you have already investigated this and found that we cannot change it) Apr 30 13:43:05 openmoko: 03andy 07stable * r05f86ffc8bc8 10/defconfig-gta01: From: Mike Westerhof Apr 30 13:45:31 openmoko: 03andy 07stable * rf7d2111513c7 10/ (defconfig-2.6.24-maxmodules defconfig-gta01): From: Mike Westerhof Apr 30 13:46:14 ScaredyCat: you have the power of SHOUT :-) Apr 30 13:46:45 hehe! Apr 30 13:47:01 :) Apr 30 13:50:24 openmoko: 03mwester 07stable * r32a457c16160 10/ (defconfig-2.6.24-maxmodules defconfig-gta01): Stop GTA01 MMC Debugging messages Apr 30 13:53:35 mwester: i have one similar issue with the gsm-modem. sometime when i directly (w/o muxer) connect to the /dev/ttySAC0 i don't get the complete greeting line, its cut in the middle. may that be related to your flow-control issue (remember, no suspend involved) Apr 30 13:55:45 CIA is lagging Apr 30 13:59:10 Cut at any point in particular? Apr 30 14:01:07 SpeedEvil: you know which mesage i mean? this "AT-Command Interpreter ready" or similar Apr 30 14:01:28 - Hostname: docs.openmoko.org Apr 30 14:01:32 SpeedEvil: i don't remember exactly... have to check Apr 30 14:01:33 * ScaredyCat grins Apr 30 14:02:22 emdete: If it is this problem, you'll see the first 15 characters of the text, with the 16th (and final) character being the last character of the message, which is always a newline character. Apr 30 14:02:33 yeah - that Apr 30 14:28:08 mwester: strange. i can't reproduce it right now, but 15chars could be possible, but i don't use any suspend... Apr 30 14:30:06 Ok. But note that it is not dependent upon suspend, suspend/resume is merely the scenario that sets up for this (lots of data to send from GSM combined with longer IRQ latencies that would result in the UART not being serviced within the required .5 - 1 ms range) Apr 30 14:34:43 mwester: but that means that we have that problem even without suspend. if the system gets under high load we loose data - right? Apr 30 14:35:05 emdete: the kernel should be buffering it Apr 30 14:35:23 SpeedEvil: does it under higher load scenarios? Apr 30 14:35:24 emdete: it should be happening in interrupts - it's not user code loading it Apr 30 14:35:55 than my cut-greeting is not related? Apr 30 14:36:23 emdete: however, something like a higher priority interrupt that holds off for an unusually long time might Apr 30 14:36:46 over 1.6ms or so Apr 30 14:36:49 what has higher prio? do you have example? Apr 30 14:36:51 emdete: Yes, that is why I am so concerned about this problem -- I do not believe it to be limited to suspend/resume any longer. Apr 30 14:37:12 emdete: I don't recall the interrupt priority wiring on the SoC Apr 30 14:37:23 *everything* has higher priority than UART 0. Apr 30 14:37:24 i saw the cuttet greeting under no-load-at-all Apr 30 14:37:36 maybe sd access..? Apr 30 14:37:47 interrupt priority is software controlled Apr 30 14:38:15 it's that murphy thing again Apr 30 14:38:16 The SoC docs say that the hardware is involved, and UART 0 is on the lowest-priority bit of the hardware. Apr 30 14:38:30 mwester: however if busy-waiting is not helping then i don't think any kind of interrupt set up will work Apr 30 14:38:31 I've not read that bit Apr 30 14:38:53 even FIQ has an incomparable latency compared to busy-wait Apr 30 14:38:55 balrog-kun: busy-waiting *does* fix it -- but one cannot busy-wait forever. Apr 30 14:38:57 If something is masking interrupts that might be the probolem. Apr 30 14:39:06 might it? Apr 30 14:39:09 mwester: uh okay Apr 30 14:39:22 that's something positive :) Apr 30 14:39:33 But I agree that FIQ is the only solution that will resolve this in all possible use cases. Apr 30 14:39:54 * SpeedEvil suffers from acronym decay. Apr 30 14:40:11 FIQ approx equal to NMI on the SoC Apr 30 14:40:17 (more acronyms!) Apr 30 14:40:17 ah Apr 30 14:40:24 dear god ! Apr 30 14:40:25 i don't know, if it's a bug in the modem then FIQ might still be too slow Apr 30 14:41:36 If you wind down the FIFO, you've got over a millisecond Apr 30 14:41:48 i'll have a look tonight at the modem commands again to see if there's a way to do 9600baud Apr 30 14:42:55 one thing i'm sure is that when in multiplexing mode 115200 is the only possible speed Apr 30 14:43:04 it's part of the standard Apr 30 14:44:13 balrog-kun: you mean gsm0710 muxing? Apr 30 14:44:29 emdete: yes Apr 30 14:44:54 I saw stuff in my skimming over the 0710 muxing that implied error checking in packets. Apr 30 14:45:00 but you can have different speed than 115200... Apr 30 14:45:08 is there a 'rerequest packet' ? Apr 30 14:45:09 emdete: the UART's speed is fixed to 115200 iirc, and the individual DLCIs can control their speeds Apr 30 14:45:32 or i may remember wrong Apr 30 14:46:08 i never tested other speed but it should be possible Apr 30 14:46:29 you cannot change speed in muxer mode and the speed of the channels is virtual anyway Apr 30 14:46:57 right Apr 30 14:47:33 and ending muxer mode and restarting will loose all states of channels... Apr 30 14:51:08 there's this ATW command to save settings but gta modem doesn't seem to support it Apr 30 14:51:21 :D Apr 30 14:51:30 but that's not by channel Apr 30 14:51:53 i would prefere freq-change instead of suspend anyway... Apr 30 14:52:22 my neo lasts 5h with everything on - a little more would be sufficient Apr 30 14:53:07 well if you turn off the backlight for the screen switch off GPS Apr 30 14:53:16 it would be neough probably Apr 30 14:54:08 no, gps off is no way ;) Apr 30 14:55:31 mmmm Apr 30 14:55:42 * mjr could live with intermittent gps ;) Apr 30 14:55:45 Wednesday April 30 2008 -- 16:55:32 +02:00 Apr 30 14:55:49 one day left? :-P Apr 30 14:55:57 mjr: if agps/supl is there... Apr 30 14:56:16 * SpeedEvil wishes the gps would do as it claimed Apr 30 14:56:29 Or had docs Apr 30 14:56:39 TI or ublox ? Apr 30 14:56:43 SpeedEvil: Which one? Apr 30 14:57:24 there is doc enough about agps with ublox and supl Apr 30 14:57:24 hh Apr 30 14:58:37 The hammerhead - which I spent several thousand hours tring to reverse engineer, Apr 30 14:58:45 s/hours/minutes/ Apr 30 14:58:45 SpeedEvil meant: The hammerhead - which I spent several thousand minutes tring to reverse engineer, Apr 30 15:02:01 s/\-.*$/Our AT command set is magical. Wanna see, wanna see ? Sykes !/ Apr 30 15:02:57 oh damn GPS Apr 30 15:03:08 I'll have to coin a new regex now. Apr 30 15:40:55 hiya Apr 30 15:41:33 anyone got suggestions for a new image to flash? i've mostly just been running the default OM ones on mine lately, but i have a feeling there are others out there pushing images. Apr 30 16:01:03 try one of the development builds or one of the scaredy cat's Apr 30 16:01:10 usually they work Apr 30 16:01:26 scaredy cat has a few nice surprises in it though Apr 30 16:01:54 like candy?!? Apr 30 16:02:13 like cats !!! Apr 30 16:02:20 lots an lots of LOLcats ! Apr 30 16:03:54 I feel like that might slow the image down :) Apr 30 16:05:38 the ScaredyCat builds used to have added nostrils Apr 30 16:06:05 yes you might find yourself being a bit dizzy and you might see a lot of wavy vidid colors such as in variety of drug abuse cases from 80s Apr 30 16:06:09 * mwester shakes his head over that particular wallpaper choice... Apr 30 16:06:15 but that's a price everyone is willing to pay dammit ! Apr 30 16:07:05 I liked those nostrils, they showed character Apr 30 16:08:17 Oh! Is *that* where character is found, then ;) :D Apr 30 16:08:58 * ScaredyCat takes some 'moon' pics for the next image Apr 30 16:13:09 XorA: what was that fudge again, with the not dl'ing form om ? Apr 30 16:17:39 remove CVS_TARBALL_STASH Apr 30 16:19:07 ta Apr 30 17:26:12 FAIL! Apr 30 17:26:43 israel: you're leaving in Israel ? Apr 30 17:26:54 MichaelShiloh: there ? Apr 30 19:06:15 balrog-kun: ping Apr 30 19:06:23 mickeyl: opng Apr 30 19:06:50 hi. would you assume that forwarding ttySAC0 via nc/pty works in MUXer mode? Apr 30 19:07:11 (because it doesn't and now i need to find out why9 Apr 30 19:07:54 mickeyl: humm humm, it should work just like forwarding plaintext AT commands i think Apr 30 19:08:07 mickeyl: have you tried the same setup with plain AT commands? Apr 30 19:08:18 yes Apr 30 19:08:22 the same setup works fine with plain AT Apr 30 19:09:30 emdete might know if anything other than the characters needs to also be forwarded, like BREAKs, XON, XOFF and thing like that Apr 30 19:09:38 these signals are not fowarded Apr 30 19:10:59 oh i wonder if there's something that buffers data until a newline in the nc/pty setup, there shouldn't be Apr 30 19:12:24 hmm Apr 30 19:12:41 i wonder whether we could come up with a testing program Apr 30 19:12:52 other than the full MUXer which may have its own problems on top Apr 30 19:13:54 mickeyl: do you get any error message from the MUXer? Apr 30 19:14:09 balrog-kun: yo, loads of Apr 30 19:14:28 basically something like "can't talk with the modem" Apr 30 19:14:33 :D Apr 30 19:14:51 not very meaningful :) Apr 30 19:14:55 hello mr modem hihihihi Apr 30 19:15:01 not for someone who didn't write the code, ya Apr 30 19:15:16 emdete could probably debug that better, but he's doing something else on the MUXer atm. Apr 30 19:15:29 i think it would be very helpful though Apr 30 19:15:37 the TS07.10 protocol specifies things like timeouts which are by default 10 ms on some messages Apr 30 19:15:54 i wonder if the network latency is too bad Apr 30 19:16:06 oh right, could be Apr 30 19:16:55 if nothing works then forwarding individual DLCs should work but i understand that's not so much fun Apr 30 19:18:00 oh, iirc the initial AT+CMUX already specifies some of the timeout values, maybe that can be increased easily there Apr 30 19:19:05 +CMUX: (1),(0),(1-5),(10-100),(1-255),(0-100),(2-255),(1-255),(1-7) Apr 30 19:19:22 i tried different framesizes, but that had no effect Apr 30 19:19:30 the (10-100),(1-255),(0-100),(2-255),(1-255) ranges are allowed ranges in milliseconds or something Apr 30 19:19:52 mickeyl: how do u setup this nc stuff? alloc a channel and then? Apr 30 19:20:07 emdete: it goes like that: Apr 30 19:20:17 turn on gsm on neo Apr 30 19:20:29 kill all programs using gsm on neo Apr 30 19:20:35 then run netcat with: Apr 30 19:20:45 nc -l -p 5000 < /dev/ttySAC0 > /dev/ttySAC0 Apr 30 19:20:50 setup on neo done now Apr 30 19:20:52 go to PC and Apr 30 19:21:04 ./pty 192.168.0.202 5000 Apr 30 19:21:09 it shows you a pty Apr 30 19:21:20 then run the muxer on your PC with that pty as serial port Apr 30 19:21:27 done Apr 30 19:21:35 and the rest can run on PC Apr 30 19:21:44 _if_ it would work :) Apr 30 19:22:08 ah, muxer host-side... understand. i thought you mean running nc on one channel of the mobile Apr 30 19:22:22 no, the fun is completely running everything but the modem on PC Apr 30 19:22:26 could u use a terminal on the pts? Apr 30 19:22:35 yes, works fine with plain AT commands Apr 30 19:22:42 i can run gsmd fine Apr 30 19:22:47 that is fun Apr 30 19:22:53 interesting setup... could speedup muxer develop alot... Apr 30 19:22:58 yep Apr 30 19:23:28 it would help to develop the complete phone stuff on PC Apr 30 19:23:37 pygsmd, ophoned Apr 30 19:23:41 up to the UI programs Apr 30 19:23:42 but muxer proto is binary... maybe netcat doesnt like it? Apr 30 19:24:04 netcat doesn't care Apr 30 19:24:05 * balrog-kun wondered that too Apr 30 19:24:11 netcat should not care at all Apr 30 19:24:17 but i didn't look at its code Apr 30 19:24:21 agreed, but why does it? Apr 30 19:24:27 it doesn't do any terminal munging Apr 30 19:24:34 but it works fine for gsmd which cares very much about the \r\n\r\n or something on line endings, so it's supposed to like binary Apr 30 19:24:49 Qwell: that will be the problem, perhaps you need to setup the term first? Apr 30 19:25:16 balrog-kun: binary means more that newlines ;) Apr 30 19:25:35 well, yes Apr 30 19:25:47 i've made backups of /dev/mtdblock* through netcat though Apr 30 19:25:53 :D Apr 30 19:26:13 another story for the thread 'i like the neo' on the ml Apr 30 19:27:42 maybe we should have a modified nc that does the baud rate / c_flag / o_flag setup stuff on the tty, like gsmd Apr 30 19:27:49 or like the muxer Apr 30 19:28:02 are there any differences in the ways gsmd and the muxer setup the tty? Apr 30 19:29:14 don't even need a modified nc. use the stty app and nc's awesome GAPING_SECURITY_HOLE define Apr 30 19:29:48 execute some shell script that sets the tty params, and then basically does the same redirecting of stdin/stdout Apr 30 19:30:18 hmm, the flags are not reset after closing /dev/ttySAC0? Apr 30 19:30:35 then running the muxer once on the neo and quitting it should do the same Apr 30 19:30:44 and then nc Apr 30 19:30:51 hmm interesting approach Apr 30 19:30:54 that could work Apr 30 19:32:03 mickeyl: no, muxer switches off the modem Apr 30 19:32:08 AT@POFF... Apr 30 19:32:26 not if i kill it Apr 30 19:32:28