irclog2html for #htc-blueangel on 20051009

08:44.46*** join/#htc-blueangel rob_w (n=bob@p213.54.155.212.tisdip.tiscali.de)
08:44.46*** mode/#htc-blueangel [+o rob_w] by ChanServ
09:06.07*** join/#htc-blueangel llange (n=llange@lange.nom.fr)
09:45.49rob_whi llange
12:51.19*** join/#htc-blueangel cp15 (i=xvvldc@p549A2965.dip0.t-ipconnect.de)
12:51.20*** mode/#htc-blueangel [+o cp15] by ChanServ
12:51.40cp15Good morning, everyone
12:51.53cp15Ah, a new guest. Hi llange
12:53.27bhimaApparently the h6300 can use a proprietary multi-plexing protocol.
12:54.03cp15Interesting. Do you have any details?
12:54.21bhimaI know that somebody else implemented it already :)
12:54.51cp15Any source code available?
12:56.23bhimaI don't know if it's been uploaded or not, but it was going to be made available.
12:56.48cp15Ok, if you know more please tell me.
12:57.32cp15Do you know whether GPRS is also working when you make a call?
12:57.44bhimaThe h6300 uses a Wavecom modem; they have their own multi-plexing protocol, which is apparently on by default on the modem
12:58.27cp15Aha, so very different from the BA. But maybe we can use a common userspace interface
12:58.59bhimaYes, I would _definitely_ liike to see that.
12:59.33cp15Does it work with windows ce?
12:59.47bhimaI'm sorry that I keep talking about the stuff I want to do without actually doing anything. I just got back last night from Italy, and am hoping that I can really start to move forward and actually contribute. Right now I feel more like a liason between the h6300 and BA work. :)
12:59.51bhimaIt's for Linux.
13:00.57cp15Ahh... Mistake... I thought the "Yes, I would _definitely_ liike to see that." was the answer to my GPRS question
13:02.18bhimaAhh, no, I would definitely like a common interface.
13:02.48cp15Yes, I know now... But do you know whether GPRS works when you make a call?
13:03.09bhimaYou mean, can you actually transfer GPRS during a call?
13:03.19cp15Yes
13:03.36bhimaoh. No. That is a hardware/protocol limitation of GPRS.
13:03.58cp15Ok, it's fine, I just wanted to know
13:04.28bhimaIt's a number of issues, but cell sites generally have certain channels for voice, and certain for GPRS.
13:04.34bhimaThere are only 8 time slots available.
13:04.50bhimaThey generally hand a particular frequency to GPRS or to voice.
13:05.41bhimaWith only 8 slots available, it would be sort of hard to permit vast differences in data rates once you had voice users on the same frequency.
13:05.48cp15Ok, I thought maybe it is interleaved with voice data
13:06.24bhimaI think that in theory that would be possible; in practice, 8 timeslots makes that less likely to work. Also, GPRS was retro-fitted.
13:06.34bhimaThere are many different vendors of GSM base stations.
13:06.49bhimaSo as I understand it, they wanted to make it easy to retro-fit to existing equipment.
13:06.54cp15So do you know whether the ppp connection is terminated before making/answering a call?
13:07.16cp15Or is it just suspended and the session stays?
13:07.27bhimaIt's fairly easy to unplug a few frequency channels and hook them up to new hardware; it's harder to integrate time sensitive stuff.
13:07.39bhimaI'm pretty confident that PPP stays live.
13:08.00bhimaI know that TCP sessions stay live through calls.
13:08.29cp15Ok, then ppp must stay, since you probably get a different ip in a new session
13:08.50bhimawell, in _theory_ you could get the same IP. In practice, I don't see any reason at all that they'd do that.
13:09.23cp15Which also means keep-alive and timeout features of ppp have to be disabled
13:10.21bhimaIn my experience, timeouts on smartphones are...high....
13:11.00bhimaOn my P800, attempting to start a TCP session without network coverage could take 20 minutes to timeout.
13:11.33bhima(and, of course, there's a bug in the firmware; the e-mail app displays a status dialog. When the phone pops up one of those, the backlight turns on. and stays on..)
13:12.05bhimaThus, in the US, working in a basement, I would have trouble getting a day out of a P800 with 30 minute e-mail checking.
13:12.44cp15It is actively checking for e-mail?
13:13.21bhimaYes. It didn't support IMAP IDLE.
13:13.26bhimaThus, it needed to poll.
13:13.29bhimaI had unlimited GPRS.
13:13.36bhimabut not unlimited SMS. :)
13:13.54cp15Oh, unlimited GPRS sounds nice :-)
13:14.32cp15Do you know how the push email service for backberry works?
13:15.34bhimaThe price has gone up since I left; it's now $6/month for unlimited to your phone.
13:15.59bhimaI'm not sure, but the US blackberry plans all include unlimited data and, I think, unlimited SMS.
13:16.16bhimaI know that T-Mobile US, on the h6315, has a special push e-mail system working via SMS.
13:16.35bhimaThe app intercepts all SMSs, and checks them; if they're for push e-mail it will download the e-mail.
13:17.10cp15But you have to pay one sms for each mail...
13:17.21bhimaNo. Unlimited.
13:17.36cp15Ok, then it's ok
13:17.58bhimaActually, with o2 Germany, Eu3 gets you unlimited e-mail notifications via SMS.
13:22.27llangeHello cp15.
13:22.37cp15Hello llange
13:23.03llangeI do need some advice, however I don't know what would be the best place to ask.
13:24.08cp15Just ask
13:25.02llangeOk. I'm trying to boot linux kernel on HTC Himalaya. I've compiled the recent BlueAngel release with some tweaks, and the kernel boots ok. However, it doesn't exec linuxrc.
13:25.40cp15What are the last messages on the screen?
13:25.57llangeFreeing init memory: 80K
13:25.57llangeLLA after free:
13:25.57llangeLLA : run_init_process : /sbin/init
13:25.57llange[ LLA = llange = my debug messages]
13:26.42llange(it is on a serial console. my command line is : "root=/dev/ram0 ramdisk_size=14336 keepinitrd console=ttyS1,115200 console=tty0"
13:27.18cp15You need to specify init=/linuxrc also
13:27.35cp15Don't ask me why, I don't know
13:27.36llangeyes, I also did.
13:28.15llangeBut no change (only my debug message changing to "run_init_process : /linuxrc"
13:30.04cp15Have you changed the floating point emulation to FPE_FASTFPE? It doesn't work and gives exactly that error condition. Took me a while when I first compiled 2.4.16 to find out
13:30.20cp15I mean 2.6.12 of course
13:31.31llangeI tried both. now it is CONFIG_FPE_NWFPE in my .config.
13:31.49llangewithout extended precision.
13:31.59cp15Have you tried NWFPE and init=/linuxrc?
13:32.46llangeJust right at the moment I am...
13:32.48cp15Maybe you should copy arch/arm/configs/blueangel_defconfig to your .config
13:33.07cp15Getting the right .config is not easy
13:33.54llangeI started with this config. Himalaya is a little different architecture so I had to change a few things.
13:34.27llangeA above all, I have a high (50%) failure rate with HaRet...
13:35.22cp15Oh... I didn't read the Himalaya and thought you have a BA
13:35.56cp15Up to which stage does it get when failing?
13:36.38llangeHaRet ? before filling the thermometer. It gives control back to Windows CE without refreshing the screen.
13:37.05llangeit's v 0.3.2c with serial output.
13:37.12cp15Hmm... Never have seen that
13:37.37cp15Maybe a memory allocation problem?
13:37.38llangeit prints "Testing 123
13:37.38llangePhysical kernel address: a0008000"
13:37.38llangethen... nothing.
13:38.13llange( 5 times since I've told you I'd be trying ...)
13:39.09llange6 =)
13:39.17cp15:-(
13:39.27cp15Have you stopped everything else?
13:39.50llangerebooting (cold) each time. hard once in a while.
13:40.34cp15You have to do a cold reboot? That is strange. It should only be necessary if the eyes went red
13:41.02llangesorry I meant warm reboot.
13:41.12llangeand cold once in a while.
13:42.33llangeI'm expert in screen calibration... is there a way to kill this "application" and to cold boot directly in windows CE without screen calib and introduction ?
13:43.01cp15You can put autorun.exe (get it from the xda-developers.com forum) in a folder named 2577 on the sd card, then haret will start immediately after screen calibration
13:43.24cp15To skip the calibration, you would have to change the pxa rom, which is a bit dangerous
13:44.01cp15Ehmm... After calibration and cut & paste excercise
13:44.09cp15But before customization
13:44.30llangeI'm willing to take the risk. Is there any piece of docs ?
13:44.30llangeI already got rid of the customization. I spared 1 reboot =)
13:45.00cp15look for dumprom and makerom on xda-developers.com
13:45.16cp15But make a rom backup on sd card first
13:45.39llangeI have these, but i don't know what to "remove" from the rom.
13:46.13cp15Ah... I have seen a hint once... I am looking
13:48.38llangefirst boot but CRC error (pb with sdcard I think)... it's not the good day. :-(
13:53.40llangeboot ok. but same pb : Freeing init memory: 80K
13:53.40llangeLLA after free:
13:53.40llangeLLA : run_init_process : /linuxrc
13:54.13cp15:-(
13:54.25cp15What linuxrc have you used?
13:54.45immoloon my himalaya I had to run haret from either ram or the built in storage
13:54.52immoloelse it failed
13:55.00cp15http://wiki.xda-developers.com/index.php?pagename=CompleteTools%2Fwelcome.exe might help you speeding up the boot
13:57.13llangegreat ! I will try this.  Concerning the himalaya, I had success with a previous kernel (2.6.3-hh2-xda0) with login prompt, etc.. from the SD Card. However I'll try to copy HaRet somewhere else.
13:57.35immolooh if you boot into linux its fine
13:57.52immoloharet just crashed everytime for me if ran from sd
13:58.36immoloanyway cp15 I think I found out why fontconfig-native doesn't build
13:58.52cp15What was the reason?
13:59.09immolooe doesn't like root :P
13:59.32cp15I already wondered why you work as root
13:59.44llangeimmolo : I guess you're not playing with your Himalaya anymore ? =) What was the last kernel you did run on it ?
14:00.25immolowell I was working in that chroot then I just copied everything off in the same term and carried on
14:00.34immolomy thought for being lazy
14:00.39immolollange-  2.6.3
14:00.59llangethx
14:01.41immoloit didn't do much though
14:02.02immoloand I could never get usbnet working on my host either :P
14:02.29immolobut as a pdas with black screens go it wasn't too bad
14:02.59llangesame for me concerning usbnet. That's why I'm trying to follow a more active development kernel (2.6.12), but with limited success. =)
14:03.49immolocp15-  anyway I'll start this again as local user most of the bugs should stop and I shouldn't have to hack bluez-utils anymore
14:04.21cp15llange: I hope you did notice I changed GPIO_NR_HIMALAYA_TS_IRQ_N in himalaya_ts3.c which breaks it for himalaya
14:04.24immoloI'm going to have to workout how to use your version rather then the standard version though
14:05.00cp15I guess you have to just change the source uri and remove the patches, but haven't looked in deeper
14:05.04llangecp15 : when you had your troubles with floating point etc... how did you manage to debug this ?
14:05.20cp15If you find it easier to use a patch, just do it
14:06.10cp15At first I was desperate... Then I compared 2.6.12 .config with 2.6.3 .config and played with the values.
14:06.11llangecp15 : I'm not using touchscreen right now.
14:06.37cp15I didn't find a way to debug this
14:06.53llange:-(
14:07.26cp15Which initrd do you use?
14:08.23llangestarted with 2.6.3 and modified it according to http://wiki.xda-developers.com/index.php?pagename=BlueangelKernel
14:08.57llangea few changes ( some echo "hello" > /dev/ttyS0, S1, ... at the beginning, just after symlinks.)
14:10.42cp15The changes mentioned on BlueangelKernel are BA-specific and shouldn't be necessary for the Himalaya
14:11.08llangeexcept for the serial ports that changed since 2.6.3
14:12.09llangebtw it seems that BA and HI are not using the same serial port for debugging. Mine is TTYS1 ( BTUART) on the craddle.
14:14.13llangebtw in command line : is the "console=tty0" mandatory ?
14:18.57llangeWhen i keep this "console=tty0", after booting, I don't have any echo on the terminal.
14:18.58llangeWhen I omit this, I do have an echo (when sending characters they are echoed back). Does it sound familiar ?
14:23.20cp15console=tty0 brings the console output on the device display
14:23.42cp15BA uses STUART for the serial connection
14:24.26llangeok so I must get rid console=tty0 as I I'm not targetting the LCD right now (trying to work on the serial connection).
14:24.59cp15You have to specify console=ttyS1 for serial console
14:25.24cp15Ah... Have you enabled LL_DEBUG? Maybe linuxrc is called, but you don't get its output?
14:25.36llangeI had to modify a few file ( debug-macro.S, uncompress.h ) to account for the serial difference. I already have console=ttyS1.
14:26.04llangeI disabled LL_DEBUG recently but I'm going to re-enable it (I was seeing double =) )
14:26.15cp15No, don't do it
14:26.59cp15DEBUG_LL is to debug the boot process and gives earlier printk output, but it makes trouble when using it as console, too
14:27.15llangeAh
14:28.06cp15Can you give me a diff between your .config and the BA config?
14:28.18llangeyes. on the IRC ?
14:28.21cp15Yes
14:30.32llange3,4c3,4
14:30.32llange< # Linux kernel version: 2.6.12-hh2-xda0-ba0
14:30.32llange> # Linux kernel version: 2.6.12-hh2
14:30.32llange110,114c110,113
14:30.32llange< # CONFIG_MACH_HIMALAYA is not set
14:30.47bhimacp15: I don't think that was a good idea. :)
14:31.38*** join/#htc-blueangel llange (n=llange@lange.nom.fr)
14:31.42cp15:-( Sorry... But I already pasted longer texts. Is this a operator privilege?
14:31.55cp15Can you do dcc?
14:32.26llangedon't think so. I'm trying to put it on a URL.
14:34.46immoloexcess flood is a client set limit
14:35.04llangecp15 : try this ? http://himalaya.lange.nom.fr/ba-hi.diff
14:35.28cp15The client decides that the user behaves badly and disconnects him?
14:36.07llangeIt's not really a full featured IRC client, i'm using Gaim.
14:36.14immoloyeah, In mIRC it was anyway
14:36.56immoloand if you remember some of the outputs of bluez-utils I posted they were way over 4 lines :P
14:37.52immoloI think it was designed for when you copy and paste the wrong thing and get yourself banned
14:39.48llangeAnd here is what is neede to compile 2.6.12 for Himalaya : http://himalaya.lange.nom.fr/K2-6-12-hh2-xda0-ba0-himalaya.diff
14:42.40cp15Found nothing unusual
14:43.03llangethanks =)
14:44.46llangeThe bootlog : http://himalaya.lange.nom.fr/K2-6-12-hh2-bootscreen.txt
14:58.12cp15Looks also good
14:58.45cp15Maybe you can try something out: Use init=/lib/ld-2.3.90.so
14:59.10cp15This won't boot your system, but you should get an errormessage from ld and a kernel panic
14:59.50cp15And since ld is statically linked, we can rule out shared library problems
15:03.24llangeI'll try.
15:08.29llangeless often, but sometimes HaRet freezes when the 'spherical bottom' of the thermometer is half-filled.
15:10.27llangecp15 your idea seems interesting : I have a kernel panic :
15:10.27llangeLLA : run_init_process : /lib/ld-2.3.2.so
15:10.27llangeUsage: ld.so [OPTION]... EXECUTABLE-FILE [ARGS-FOR-PROGRAM...]Kernel panic - not syncing: Attempted to kill init!
15:10.51cp15Ok, the program is running
15:11.53llangedoes this also confirm that the default console ( /dev/console ) is well configured ?
15:11.57cp15Yes
15:12.04llangeok
15:12.20cp15Try init=/bin/sh next
15:13.58cp15And happy cut & pasting ;-)
15:15.28llangeJust before trying ld.so, I've realized that I wasn't using the initrd from 2.6.3 but the BA 2.6.12 (not the same libc it seems.)
15:20.57cp15Yes, it has a newer libc to be compatible with the h3900 image. It should't make a problem, but maybe you try the Himalaya initrd if init=/bin/sh doesn't work?
15:23.06llangedoesn't work bin sh (himalaya initrd)
15:24.05cp15Errors or just a hang?
15:25.14llangeno errors, nothing. But there is echo on the TTY
15:27.56cp15Ok, one last try: Try init=/sbin/insmod
15:28.43cp15If it works, the problem is most likely in busybox and I would suggest recompiling it with a lot of debug output
15:29.47cp15If it doesn't work, the problem is in the shared library setup, and I suggest compiling sash (also with lots of debug output) and running it via init=
15:37.22cp15You can also extract sash from http://ftp.de.debian.org/debian/pool/main/s/sash/sash_3.7-7_arm.deb
15:38.01cp15if you don't want to recompile (but you won't get any useful information if it doesn't work)
15:42.42llangewill try. strange because that initrd works for a lot of people isn't it ?
15:42.58llange(and did work for 2.6.3)
15:43.05cp15Yes, don't have an idea where the error might be
15:44.54llangeinsmod : nothing (no output).
15:45.24cp15Ok, shared library...
15:46.51llangewhich means... ? =)
15:47.40cp15don't know
15:48.04llange=)
15:48.35cp15Are you sure you used NWFPE?
15:49.08llangeright now, I just checked it, and 'm doing a 'make clean; make'
15:49.46llangeI also tried a kernel with no floating point emulation ; which changed nothing. no runtime error.
15:50.26llangeshouldn't the kernel 'oops' or panic in case there were a pb with floating point ?
15:50.44cp15Probably
15:54.47cp15Ah... The "NetWinder Floating Point Emulator" message is missing in your boot messages
15:57.19llangeyes you're right. And it was there in 2.6.3
15:57.27llange"<4>NetWinder Floating Point Emulator V0.97 (double precision)"
15:57.45cp15It is also there on my 2.6.12 kernel, I have just rebooted
15:58.13llangeOk I'll check this. thanks for the tip and your time. I'll came back to report success (or failure =) )
15:58.37cp15Ok, then I will continue with the phone application :-)
15:59.11llangegreat news : what is this application ? I do hope being able to call from linux one day =)
15:59.50cp15gomunicator
16:00.00cp15See the xanadux cvs for the newest version
16:00.15cp15I can already make and answer calls on my BA
16:01.47llangegreat ! I hope to try this soon =) (when I have LCD, touchscreen, ... =) )
16:02.04cp15Shouldn't be that difficult
16:04.32llangeOn the road to success : message in bootlog + insmod panicing !! (You've been of great help !)
16:05.35llangewhat is the state of the BA ? is the device autonomous ? touchscreen etc.. ?
16:07.12cp15Great. If you want to join the team get a sourceforge account and write to w4xy@users.sourceforge.net
16:07.55cp15The device is autonomous, the essential things are working, but battery lifetime is way too short (< 24 hours)
16:08.27cp15More details at http://wiki.xda-developers.com/index.php?pagename=BlueangelProgress
16:10.03llangethx
16:14.37*** join/#htc-blueangel rob_w (n=bob@p213.54.155.67.tisdip.tiscali.de)
16:14.37*** mode/#htc-blueangel [+o rob_w] by ChanServ
16:15.08cp15bhima: still there?
16:15.24llangefamiliar login: => thanks a lot cp15 !!!
16:17.46bhimacp15: yup
16:18.27cp15I am planning a BA developers meeting for a weekend. Do you want to come?
16:19.42bhimacp15: Hrm. Yes, but I'm not sure right now when I am actually free. Where would it be?
16:20.30cp15In my flat (at least if it aren't too much people), near Ingolstadt
16:20.33bhimaNovember ~22 is when I move back to the US. My parents are visiting for awhile starting Wednesday. I need to find out there plans.
16:21.18cp15Ok, take a look at http://ba-devel.it-schaller.de/ and choose the dates when you can come
16:21.29bhimaWhere is Ingolstadt?
16:21.43cp15Between munich and nuremberg
16:25.52bhimaOk. On Wednesday, I'll talk to my parents about their plans and will find out when I am likely free.
16:26.37cp15Do you come again to germany after the 22nd?
16:27.23bhimaProbably in May or so for two or three weeks...
16:28.33cp15Thats not too much. But I have been told the internet even works outside of europe ;-)
16:29.23bhimaIt does. It was actually invented outside of Europe. :P
20:27.14*** join/#htc-blueangel lamikr (n=chatzill@aragorn.kortex.jyu.fi)
20:27.46lamikrHi!
20:28.10cp15Hi lamikr
20:29.23cp15How may we help you?
20:29.31lamikrBhima and koen have told me that you are making pretty good progress with gsm and blueangel. I and couple of my friends in "h6300-port" are trying to do same for ipaq h6300... I came here to ask some help.
20:30.32cp15Ah... bhima already told me that he met somebody who made progress on the h6300 gsm
20:30.43cp15What is your current status?
20:30.54lamikrYes, we are able to communicate over /dev/ttyS1
20:31.56lamikrWe are able to send AT commands for making the phone to call, etc... But we are not yet able to control the speaker. The sound driver is however getting better, so there is hope...
20:32.24cp15Do you know which sound chip is used in the device?
20:33.32cp15On the BA GSM sound is pretty easy. The microphone is connected to the sound chip and gsm in parallel, and the gsm speaker output is connected to line in on the sound chip
20:33.54lamikrI tried tsc2101, we have now OSS based driver that was able to play, but everett who has made the drivers has now started to convert it to alsa.
20:35.45lamikrCan you use AT commands like "AT+SPEAKER=0", "AT+SPEAKER=1", etc... for selecting whether you want to use headphone, or normal speakers? According to our wavecom gsm module specs, we can do that. We just need to get the driver to init the speaker and mic.
20:35.51cp15Alsa is definetly better here. I am currently working on a phone app, and I think I will convert it to alsa soon
20:36.11lamikrDo you mean gomunicator?
20:36.15cp15Yes
20:36.31lamikrKoen mentioned it.
20:37.05cp15On the BA speaker/headphone is controlled by the sound chip. Have yet to find out how it is done with the microphone.
20:37.17cp15I have currently no idea how to switch the audio path to bluetooth
20:37.57cp15gomunicator is still a bit alpha, but it is getting useable. I haven't found a ready to use phone application for linux
20:38.03lamikrSo, you would like to use headless bt stuff?
20:38.34cp15Sooner or later, sure.
20:38.39lamikrWell me either. I tried to use libgsm today for sending SMS messages but it made the serial line to hung.
20:39.34cp15Haven't look yet at libgsm. But maybe it would be a good idea to base gomunicator on it.
20:39.54cp15Do you know whether it works synchronous or asynchronous?
20:40.21lamikrI actually downloaded today the prebuild gomunicator from the handhelds page and tried it. I created symlink from /dev/gsm0 to /dev/ttyS1 but I think gomunicator exited because it could not communicate via alsa. So I will now try to build it myself.
20:40.51cp15And you have to patch something to get it to work, I think
20:41.07cp15The BA requires a prefix and a postfix for each command
20:41.57cp15Take a look at sendCommand and remove the writen_ni(..,prefix,..) and write_ni(..,postfix,..)
20:41.59lamikrYes, actually I came here to ask how do you configure it for the crosscompiler. Are you build against the libs OE puts under tmp/staging/arm-linux dir?
20:42.09cp15Yes
20:42.29lamikrGood, can you give me your configure command?
20:42.54cp15I have a small environment file
20:43.00cp15export OE=/home/martin/blueangel/oe
20:43.00cp15export BUILD=$OE/build
20:43.00cp15PATH=$BUILD/tmp/cross/bin:$BUILD/tmp/staging/i686-linux/bin:$PATH
20:43.01cp15export ACLOCAL_USR_SHARE_ACLOCAL=$BUILD/tmp/staging/arm-linux/share/aclocal
20:43.01cp15export PKG_CONFIG_PATH=$BUILD/tmp/staging/arm-linux/share/pkgconfig
20:43.01cp15alias oe_conf='./configure CC=arm-linux-gcc CPPFLAGS="-Wall -I/home/martin/blueangel/oe/build/tmp/staging/arm-linux/include" LDFLAGS="-L/home/martin/blueangel/oe/build/tmp/staging/arm-linux/lib -Wl,-rpath-link,/home/martin/blueangel/oe/build/tmp/staging/arm-linux/lib" --host=arm-elf-linux'
20:43.07cp15alias oe_autom="aclocal -I $ACLOCAL_USR_SHARE_ACLOCAL && autoheader && automake --add-missing --copy && autoconf"
20:43.28cp15I think you have to adapt the pathes a bit
20:43.45lamikrYes, I will try thanks.
20:44.34lamikrBtw, one of as called collin has made socket based multiplexer for GSM. Would you be interested in using that?
20:45.15cp15For that I have to explain the situation on the BA a bit first
20:45.39cp15We have one uart for the GSM module, and a dual ported ram asic
20:46.11lamikrOk, I am listening... Or hold, I ask collin also to join here first!
20:46.16cp15All AT commands and replies go over the uart and GPRS/CSD data goes over the dpram
20:47.04*** join/#htc-blueangel collin_sleep (n=collin@elbarto.betaversion.net)
20:47.06collin_sleephi
20:47.15cp15Hi collin
20:47.33lamikrcp15 can you repeat your last sentence?
20:47.39cp15I have heard you are working on a GSM multiplexer
20:47.52cp15Maybe we can use a common interface to userspace
20:47.53collin_it is working allready ...
20:48.10cp15On the Blueangel there are two interfaces to the GSM module
20:48.13collin_is there something like this?
20:48.22collin_h63xx also has this
20:48.30cp15One is over an uart for AT commands and replies
20:48.33collin_but it's not needed
20:48.54cp15The other one is a dual ported ram asic which is connected to main cpu and gsm cpu
20:49.00collin_ah
20:49.04cp15It is used for GPRS and CSD data transfer
20:49.08collin_I heard about this
20:49.13collin_ok
20:49.21cp15I am currently writing an uart "emulation" for this
20:49.34cp15So setting up a csd call works like this:
20:49.39cp15open uart
20:49.45cp15send "ATDT<number>"
20:49.50cp15wait for "CONNECT"
20:49.54cp15open dpram
20:50.00cp15send/receive data
20:50.03cp15close dpram
20:50.09collin_ok
20:50.12cp15set ATH0 to uart
20:50.38cp15Is the h6300 using something similar?
20:50.39collin_so does your wavecom modem support multiplexing?
20:50.41lamikrI think our situation is quite similar for the data connections.
20:50.42collin_nop
20:50.55collin_on the h63xx we have one uart
20:51.11lamikrWe have two sockets: One for AT commands and another one for data.
20:51.30collin_lamikr let me explain
20:51.38lamikrOk :-)
20:51.56collin_so we have on uart with a multiplexing protocol running onit
20:52.11collin_one uart
20:52.38collin_so mplexd plays the protocol and provides one socket for each stream on the uart
20:53.05collin_cp15: is this clear? or do you need more details?
20:53.23cp15So you can have several at command streams?
20:53.32collin_nop
20:53.44collin_one command stream and one data stream (for GRPS)
20:54.12collin_the data stream is opened when a GPRS connection is created
20:54.33cp15Ah, ok, so it is similar to the Blueangel, but the transfer on the Blueangel is not multiplexed
20:54.46collin_ah do you have a wavecom modem?
20:55.04collin_try sending at+wmux?
20:55.04cp15Don't know. How could I recognize it?
20:55.12cp15Just a moment
20:55.28collin_that shows you the mulitplex mode
20:55.44collin_lamikr: waht was the command to get the manufacturer again?
20:55.49cp15ERROR
20:56.05lamikrLet me check...
20:56.19collin_try at+wmux=0
20:56.32cp15Same
20:56.42collin_ok so no mulitplexing for you :(
20:56.53cp15There is a at+cmux command, but I haven't found out how to really use it
20:57.04lamikrAT+CGMI gives manufacturer info.
20:57.05collin_ah
20:57.22cp15HTC
20:57.23collin_try at+cmux=1
20:57.47collin_cgmi returns htc?
20:57.49cp15AT-Command Interpreter ready
20:57.52cp15Yes
20:57.54lamikrI think mplexd makes it looks like the same than BA has. One channel for AT commands and another one for data. So if BA could have another two socket implementation, we could have similar looking interface in both of the devices for the gomunicator. BA's data socket should just communicate to the dpram. Or what do you collin think?
20:58.19collin_sure
20:58.25collin_thats is it
20:58.47collin_so we just need to make gommunicator socket aware
20:59.20lamikrWell, thats what at least we need to do... But what was your plan cp15?
20:59.54collin_I really need to look at the code of gomm..
21:00.08cp15Well, the internal mechanisms are a bit different, but I think we should expose a common interface
21:00.30cp15If I understand you correctly, you are doing multiplexing/demultiplexing in userspace via a daemon?
21:00.37collin_yes
21:00.37lamikrYes, that would really be good...
21:00.49lamikrcp15: We have two sockets running in /var/run
21:01.09collin_cp15: this is not worth putting into the kernel
21:01.18cp15Hmm... I think you need to use a pty. pppd probably doesn't work with a socket
21:01.31collin_cp15 ... it should
21:01.42collin_I'll check that
21:02.02collin_thats a minor detail :)
21:02.24cp15:-) You can still implement a pppd plugin
21:02.29collin_sure
21:03.23lamikrcollin: Another detail is the name of the interfaces. /var/run/wavecom_mplex_cmd and /var/run/wavecom_mplex_data needs more general names like "universal_ipag_master_of_the_gsm_data_channel" :)
21:03.35collin_hehe
21:04.05collin_lamikr: you can decide over naming if you want
21:04.13collin_:)
21:04.26lamikrWell, that was my proposal :-)
21:04.34collin_so does gomm... need libgsm?
21:04.42lamikrNo
21:04.51collin_cp15: does voice calls work on the BA?
21:04.57cp15Yes
21:05.03cp15I am already using it
21:05.12collin_nice
21:05.29cp15But there are some problems with the power management. Battery lasts < 24 hours
21:05.31collin_so I just need an sound driver?
21:05.35lamikrDoes BA handle powersaving and cpu freq scaling ok?
21:05.37collin_cp15: hehe
21:06.04lamikrok, it seems not...
21:06.36cp15Well, the GSM modems are quite different, it seems. On the BA it was easy: Power on the Amplifier, Power on mic bias, set the mixer volume, and it worked
21:06.59cp15Well, at least after I found out that the semicolon in the ATD command is not optional
21:07.29lamikrcp15: It may be as easy also for the h6300, but I do not have alsa driver that can power mic and speaker on.
21:08.02collin_hehe
21:08.05lamikrThe OSS driver started to work on friday and I have been away the weekend, so I have not tested yet.
21:08.24collin_lamikr: we now have that 14-rc1 issue
21:08.24cp15The mic and speaker power on the BA has nothing to do with the sound chip. They are just two GPIO bits
21:09.18cp15Maybe here we can find a common mechanism
21:09.25lamikrBT was that simple in h6300.
21:09.36cp15I have two mixer controls related to gsm
21:09.46collin_lamikr we have these two GPIOs for speaker and mic
21:10.13cp15One is named "Phone" and controlls the speaker volume. If you are setting volume to 0 and not playing dsp audio, the speaker gets turned off
21:11.18lamikrcp15: So does the driver turn the speaker off in that kind of situation?
21:11.35cp15The other is named "Mic Power" and is a switch. If you are disabling mic power and nod recording dsp audio, mic gets turned off
21:12.27cp15The driver is uda1380.c, and it is a bit self-contained. I have implemented a callback so the lowlevel driver gets informed when power needs to be enabled/disabled
21:13.17collin_ah
21:15.03lamikrOk, so when you turn the volume to 0, that is informed to lowel level driver. Does the driver know whether there is dsp audio playing going on or does it need to be informed by alsa libs?
21:15.14cp15Do you have a wiki/webpage or something where you document your status?
21:15.24collin_yes
21:15.26lamikrYes,
21:15.30cp15Actually on the BA it is a bit more complicated
21:15.32lamikrhttp://handhelds.org/moin/moin.cgi/HpIpaqH6315
21:16.04lamikrThere is our multiplexer also for download.
21:16.27cp15Ok, I will take a look
21:17.15lamikrhttp://www.wavecom.com/media/files/support/Advanced_Software_Solutions/Open_AT/Open_AT_30/Doc-OAT-V3.zip contains AT command manual. 400 pages of data from how to control wavecom with AT commands...
21:17.32cp15As for the audio, it works on the BA like this: There is the UDA1380 chip which is mixer, A/D and D/A converter, headphone and microphone amplifier
21:17.42cp15D/A is connected to PXA SSP port
21:18.00cp15A/D is connected to PXA I2S port (so it won't get to dull :-)
21:18.46cp15The lowlevel driver controls SSP and I2S ports and tells the UDA1380 driver only which samplerate is to be used
21:19.30cp15Rest is done by itself, so it knows when it is playing and recording (becaus it has to setup SSP/I2S) but not if the user changes the mixer setting (which is done by uda1380.c directly)
21:19.44collin_mmh
21:20.40cp15Does reading the sim phonebook work on the wavecom modem with standard gsm commands?
21:21.08collin_I guess so
21:21.21collin_the commands are in the spec asfaras I know
21:21.24lamikryes, it should. I have not tested this because I do not have any contacts saved to sim card. But manual mentions commands for those.
21:21.37lamikrThere are also commands for selecting the ring sound, etc...
21:21.45cp15On the BA it doesn't work. There is a at+crsm command where you can access the sim cars filesystem
21:22.03cp15Is the ring signal procduced by the gsm modem?
21:22.19collin_no I dont think so
21:22.56collin_I guess this is the wavecom stuff thats just there because other modules support it
21:23.30collin_maybe it can do stuff like that if we open the speaker on the device
21:23.40cp15Ah, I have seen you have also the texas bt chip. Which version is it?
21:25.05collin_lamikr: so maybe we also need to enable these special gpios
21:25.21collin_everett: told me that the mic/speaker gpio is not used by the sound card
21:25.29lamikrcp15: We have brf6100.
21:26.14cp15Same as here. You probably want to checkout bluez-utils from xanadux cvs. It allows speeds higher than 115200 baud.
21:26.20lamikrcollin: What are the gpios for mic and speaker? I could try to set them on in the gsm_driver I had in the bt_branch.
21:26.47collin_lamikr: i2c gpio 10 11
21:27.15lamikrcp15: Do you have a some kind of script that you load with the hciattach command? hx4700 has something like that for their brf6150.
21:27.19collin_lamikr : i'll try this in my gsm driver
21:27.29lamikrcollin: Ok!
21:27.58cp15I have found 3 scripts in the windows filesystem, but I haven't yet tried them
21:28.31cp15They are named Init_Rom_V316.bts  Soft_Gemini_Disable_for_P4.bts  Soft_Gemini_Enable_for_P4.bts
21:28.51cp15No idea what soft gemini might be
21:30.17lamikrWe have also some in our windows dir but I have not tried them either. Todd Plumer (the hx4700 bt driver author) send Script query on behalf of us to TI, but he has not received yet any response.
21:30.52lamikrSo you manage to get 921600 speed even without any scripts?
21:30.56cp15:-) I have also contacted him. He told me the command to set the speed
21:31.10cp15Yes
21:31.20lamikrOh, what it is?
21:31.46collin_cp15: bluetooth audio (SCO) working?
21:32.03cp15unsigned char cmd[16];
21:32.04cp15<PROTECTED>
21:32.04cp15<PROTECTED>
21:32.04cp15<PROTECTED>
21:32.09cp15<PROTECTED>
21:32.13cp15<PROTECTED>
21:32.15cp15<PROTECTED>
21:32.17cp15<PROTECTED>
21:32.19cp15<PROTECTED>
21:32.21cp15<PROTECTED>
21:32.23cp15<PROTECTED>
21:32.25cp15<PROTECTED>
21:32.27cp15<PROTECTED>
21:32.29cp15<PROTECTED>
21:32.31cp15<PROTECTED>
21:32.35cp15<PROTECTED>
21:32.37cp15<PROTECTED>
21:32.39cp15<PROTECTED>
21:32.43cp15<PROTECTED>
21:32.45cp15<PROTECTED>
21:32.47cp15<PROTECTED>
21:32.49cp15<PROTECTED>
21:32.51cp15<PROTECTED>
21:32.53cp15<PROTECTED>
21:32.55cp15<PROTECTED>
21:32.57cp15<PROTECTED>
21:32.59cp15<PROTECTED>
21:33.01cp15<PROTECTED>
21:33.09cp15<PROTECTED>
21:33.11cp15<PROTECTED>
21:33.13cp15<PROTECTED>
21:33.15cp15No, SCO isn't working. I have tried the alsa interface, but only low frequency static noise came out of the headset
21:33.27cp15The high speed patch is in the xanadux cvs
21:34.06cp15I will try again with a linux pc as other sco station which gives more possibilities to debug
21:34.06collin_cp15: do you have this special bluetooth pcm pin?
21:34.27collin_I use my bt headset with my pc
21:34.40cp15Don't know. What is this for a pin?
21:34.51collin_I dont know
21:35.08lamikrcp15: I will put this info up and try later, now I am more interested in from the gsm :-)
21:35.37cp15Interfacing of gsm to bluetooth is done without cpu interaction, the connection stays alive for a few seconds when you boot linux
21:37.02collin_this should be so
21:37.39cp15Well, I have hoped it could be done in software because this would give a lot more possibilities
21:37.54collin_mmh thats true
21:38.20collin_can you only use BT for the phone on the BA?
21:38.31cp15Like recording a phone call, a caller sensitive answering machine
21:38.55collin_mmh
21:39.16cp15BT data transfer works, SCO might work (without success right now)
21:39.39collin_so just the GSM audio streams are only to the speaker/mic and BT
21:39.50collin_not connected to the sound card
21:40.26cp15They are connected to the sound chip, but only on the speaker side
21:40.48collin_that sucks
21:40.53cp15Yeah :-)
21:41.11collin_so you can only record
21:41.24cp15Maybe I will hack the radio rom so I get the audio stream via the data channel
21:41.29collin_hehe
21:42.15lamikrI need go to sleep soon, but can we summarize. Did we have any plan from the common gommunicator interface? Does the socket need to be changed to pty?
21:42.32collin_the pty thing is for later
21:42.36collin_that only ppp/gprs
21:43.09collin_but it would be good to have communicator to be able to read/write aT commands to a socket
21:43.19cp15Well, it woldn't be hard to implement sockets in gomunicator. But many other applications (minicom, pppd, sendfax) will have trouble with it
21:43.33collin_minicom works with sockets
21:43.48cp15Really? Ok, then remove it from the list
21:44.02cp15Good to know
21:44.09collin_I will check if I can make it a pty
21:44.11lamikrYes, but how about BA? What kind of interface you are planning for the data and gomunicator?
21:44.33cp15Currenty I have two uarts. One for AT, one for data
21:44.57lamikrSo you have implemented some kind of virtual uart?
21:45.04cp15Yes
21:45.38cp15Maybe we can solve it this way: /dev/gsm0 is a symlink to whatever device is required to send at commands
21:46.05cp15gomunicator makes a lstat on it, and if it is a socket, it does a connect call
21:46.17collin_that would be nice
21:46.29lamikrYes, that would be ok.
21:47.05lamikrIn which files you are doing the uart connections in gommunicator?
21:47.12cp15if /dev/gsm0data exists (symlink to some device), the data/command are separate channels
21:47.45cp15phoneComms.c, but it is a mess
21:48.56collin_cp15: can you switch on/off the gsm module?
21:48.56cp15Robert Woehrle and I are currently working on cleaning it up
21:49.23cp15Yes, there are two GPIOs. One is for powering on/off the gsm module, and one is the reset line for the gsm cpu
21:49.34lamikrDoes he have any timeframe for committing in the changes?
21:49.54cp15As usual: No
21:50.38cp15I will also take a look at libgsm, maybe it will make phoneComms.c obsolete
21:51.06lamikrOk, well I work with the eclipse and have already took the code from CVS, so it is pretty easy for me to compare changes in the cvs code. (Eclipse has very good visual diff tools for source codes.)
21:52.38lamikrcp15: I tried following command today without success. gsmsendsms -d /dev/ttyS1 -C <service-center-number> -I AT -b 115200 <target-number> "hello"
21:52.57lamikrWith and without "-I AT"
21:53.56cp15Oh no, why it is written in c++?
21:54.56lamikrcp15: yes, it makes linking hard. And I needed to apply one patch from net to make it compile with gcc 3.4. (OE has problems also for building it)
21:55.47*** join/#htc-blueangel _cronos_ (n=cronos@ron34-1-82-224-149-102.fbx.proxad.net)
21:55.47cp15And it is completely synchronous. This will make it hard to use it for gomunicator
21:57.17cp15As for the general infrastructure: Do you guys think gomunicator should initiate a gprs connection and then launch pppd, or should it be done by an external program
21:57.38collin_both would be nice
21:57.41lamikrThe things that may be very usefull from libgsm are the commands that transfers SMS messages to PDU format.
21:57.56collin_at least it must be possible to trigger it fro mthe outside
21:58.39cp15I have already written a little conversion routine for pdu, but it is far from complete
21:58.53collin_just take it from libgsm
21:58.56lamikrWavecom has commands both for sending SMS messages either in text or PDU format, but I believe PDU is more advantaged.
21:59.28cp15Yes, and you don't have parsing problems. One line is always a complete reply
21:59.50lamikrcp15: If you want to build libgsm with gcc 3.4, you need this patch, I made changes by hand, so I did not tested whether it applies cleanly with patch command. http://prometheus.cs.wmich.edu/gentoo/rsync/app-mobilephone/gsmlib/files/gsmlib-1.11-include-gcc34-fix.patch
21:59.57bhimaI definitely want the ability to programmatically establish PPP. I usually want rules for the GPRS link. Usually I want the link to stay up all the time; if it died, something went wrong and I want to restart it.
22:00.53cp15If we start pppd from gomunicator, I would implement a dbus interface, so you can start/stop gprs via a script or from another program
22:01.05cp15But maybe a gsmd would also be a nice idea
22:01.19cp15Thanks for the patch
22:02.37cp15Oh, already midnight
22:03.34bhimaoops. I have to be in the office in the morning :(
22:03.35lamikrcp15: In here it is 1 am, I also need to go to sleep. Thank's for all info, I will try to compile gommunicator tomorrow and take some look to the sources...
22:04.00cp15Patches are always welcome :-)
22:04.07collin_lamikr: I'm trying the gpios in a second
22:04.24lamikrcollin: Ok, I will wait!
22:04.49collin_lamikr: more like 5min
22:05.30lamikrno problem, I will try to talk with evertt in h6300-port for a while about 2.6.14 kernel.
22:06.02cp15collin: Where are you from?
22:06.22collin_germany
22:06.41cp15Ah... Wasn't visible from the domain name :-)
22:06.59collin_it's my box at 1und1
22:07.16cp15Ah, I also have a root server there :-)
22:07.22collin_:)
22:08.24immoloyou guys never seen the reviews of 1and1 then?
22:08.44collin_why?
22:08.45cp15What reviews?
22:09.11immolofrom what I've heard 1and1 have one of the worse customer support records going
22:09.51bhimaThe demo worked, so now I can play. :)
22:09.59cp15Well, can't say much about customer support. I only needed it once in 3 years, because I issued shutdown -h now on the wrong machine :-)
22:10.08immololol
22:10.35collin_cp15: I nerver ever mailed them other to get IPv6 addresses and that worked within 24h
22:10.42immoloI did that on my home server, I had it running for 2 months and pulled up the wrong term to issue a halt to
22:10.57collin_also I can self reboot mt boxes without support :)
22:11.03cp15They turned it on again, but they wanted 5 euros.
22:11.43cp15Apart from that I only noticed one 60 minute network problem in 3 years
22:11.43immolostrange though, when I was running my hosting company the feedback I heard on 1and1 was pretty poor
22:12.04collin_root servers?
22:12.07immoloI got quite a bit of custom from them as well because of it
22:12.35immolonaw I only did httpd etc
22:12.43collin_I guess your customers dont build thier own OS :)
22:12.48immoloI had a couple of game servers
22:13.01immolocollin_-  No, I had to do it for them :P
22:13.28cp15Ok, you can't expect big support for a 1 euro per month service
22:13.32collin_so then its clear why then compalin about 1und1
22:14.28immolowell what I heard was mainly from what was posted on review forums when I was hunting down new clients
22:14.59immolobut seeing they are doing very well for themself they can't be all bad
22:15.37bhimamumble mumble Microsoft mumble
22:15.46collin_LOL
22:15.47immoloheh
22:15.55cp15Ok, probably 1und1 has also it's dark sites. I have read news where they offered dsl customers with too much traffic that they pay 100 euro if the customer quits his contract
22:16.23immolo1and1 do dsl?
22:16.40cp15Yes, in germany at least
22:17.22cp15But I think they are using the t-com (ex german telecom monopolist) backbone
22:17.35collin_yes they do
22:17.45immolowell they will probley use BT in the UK
22:17.48immoloeveryone else does
22:18.16cp15No, arcor doesn't, and I think there are a few others
22:18.41immolomost do as they all need a BT line
22:19.17cp15Sorry.. Thought the answer was from collin
22:19.20bhimaWhat can you get for Eu1/month?
22:19.43immoloI could a mars bar for a euro :P
22:19.47immoloand not much more
22:20.00bhimaFrom 1and1, I mean. :P
22:20.12collin_1min of support :)
22:20.37cp15Oh, they even offer one .de domain with web and mail forwarding for 0.29 euro/month
22:20.59immolothat sucks I can only get .com,org or net
22:21.06immolono co.uk
22:22.01immolobut then again my domains only cost ?1
22:23.27cp15So, I will disconnect here now. I want to complete ringtones for gomunicator and will then go to bed
22:23.46collin_nihght
22:23.51immolocool, did you get ogg support going?
22:24.15cp15No, just raw audio :-) You have to convert your ogg first ;-)
22:24.30immolook then :P
22:24.43cp15But I think I will use the gstreamer framework which allows ogg, mp3, wav and midi is WIP
22:24.50immoloanyway night, I'm going to see if I can find my linux sdcard
22:25.00cp15Night
22:25.04immolomidi isn't that important
22:25.43immolodamn I'm sure I left it at the pub last night
22:27.53immolothank god for that
22:28.11immoloI put it in my hidden pocket

Generated by irclog2html.pl by Jeff Waugh - find it at freshmeat.net! Modified by Tim Riker to work with blootbot logs, split per channel, etc.