SIP on Kindle

[knc1]

Originally Posted by jmseight

Hi knc1,

I have only c and C++ programming experience from 15+ years ago and am trying to learn bash and java programming - from getting the kindle.

Thanks,
James

Sounds like ideal background for a "project head" -

The people who do the actually coding often hate with a passion the chore of running the project. Even the almost fully automated projects supported here at Mobile Read.

So I would suggest you give my suggestion a second thought.

It reads as if you have at least one person already that you might talk into joining your project.

Putting the work up as a "project", either here or at one of the other "free" project hosting sites, has other advantages.

For instance:
pjsua is licensed GPL:
http://www.pjsip.org/pjsua.htm
http://www.pjsip.org/licensing.htm

Which means that once changes or additions are made to it and the resultant binary is distributed, the source must remain available for three years.
Putting the changed source into a publicly hosted project somewhere is a good step towards handling that requirement.

And most project hosts (like Mobile Read) provide additional support services, like a Wiki to keep information about the application.
That way, the information does not get "lost" in an old forum thread.

Plus consider the "bragging rights"
You could still win the first spot on the list of "Kindle Projects" here at Mobile Read.
No one has yet scored a spot on that list.

[knc1]

Originally Posted by jmseight

Hi toehio,

Would you please post the binary for this. I would like to try this out.

Thanks,
James

According to the blog toehio linked to, that binary is compiled against a different version of the C library than what is on the Kindle-3-keyboard.

It (the sources) will probably have to be re-compiled against the correct C library version.

A good "first goal" for your new mobileread dev. project.

[Matan]

A binary compiled according to toehio's instructions, but working with K3 version of glibc is here.

You need to change --capture-dev to 0 in the configuration file. In my tests I could hear the other side, but the other side could not hear me.

[jmseight]

Hi,

I have 2x k3g, one with Debian and one without. I ran alsa-pjsua on Debian chroot and Matan's pjusa on non-Debian. I used toehio config options to optimize, but on the non-Debian version, I had to comment out
--capture-dev 1
because it does not work on the non-Debian version - will cause pjsua to terminate.

I used the k3g's to call each other - because I could not find a good Win7 SIP client to work and the DSi SIP client mic does not work.

On the Debian k3g, I can receive and transmit audio. On the non-Debian k3g, I can only receive audio and the mic does not work.

I think this confirms the behavior of toehio and Matan.

Is the problem the capture-dev in the non-Debian version?

Thanks,
James

[Matan]

Please try with --capture-dev 0 in the non-debian case, and see if it changes anything.

If it does not help, please compare the content of the directories /dev/snd/ in and out of chroot, and compare the content of the files in the directory /proc/asound, to see if there is any difference in and out of chroot.

[jmseight]

Hi Matan,

dev0 gives the same result on the non-Debian.

I checked the /proc/asound directory. But I think the Debian script actually mount the /proc into /mnt/debian/proc and /dev into /mnt/debian/dev. SO they are identical.

I think dev1 is correct, but somehow in the no-Debian version, it cannot open dev1 and causes a crash.

I tried dev 0 and dev 2 and both don't crash but no mic. Please see what happens when I try to use dev 1:

19:46:59.150 pjsua_media.c Opening sound device PCM@16000/1/20ms
19:46:59.152 pjsua_media.c Opening sound device PCM@44100/1/20ms
19:46:59.153 pjsua_media.c Opening sound device PCM@48000/1/20ms
19:46:59.154 pjsua_media.c Opening sound device PCM@32000/1/20ms
19:46:59.155 pjsua_media.c Opening sound device PCM@16000/1/20ms
19:46:59.155 pjsua_media.c Opening sound device PCM@8000/1/20ms
19:46:59.157 pjsua_media.c Unable to open sound device: Invalid number of channels [status=469997]
19:46:59.161 pjsua_core.c Shutting down...
19:46:59.161 pjsua_pres.c Shutting down presence..
19:46:59.161 pjsua_media.c Shutting down media..
19:46:59.549 pa_dev.c PortAudio sound library shutting down..

Thanks,
James

[hawhill]

You probably have different ALSA libraries (not kernel APIs) in Debian and the Kindle environment. Also, the Debian pjsua might be patched. Does it even run PortAudio? I ask because pjsua seems to run a PortAudio driver.

The Kindle is picky about what sound formats and channel numbers it supports. Did you try to check if it uses 2 channels (pseudo stereo)? Also it might make sense to try rates like 22050 and 11025.

[twobob]

Originally Posted by jmseight

Hi Matan,

19:46:59.155 pjsua_media.c Opening sound device PCM@8000/1/20ms
19:46:59.157 pjsua_media.c Unable to open sound device: Invalid number of channels [status=469997]

Thanks,
James

That's a classic alsa lib fubar message. and nothing to do with the channels.

I can duplicate this exact behaviour in another application I'm compiling on linux host by simply removing the reference to the /usr/share/alsa-lib-xxxx library location.

Whilst this doesn't help fix it it does confirm what was said above... Libs problem probably.

Alsa libs on device are .13 AIUI

[twobob]

The problem lies here: BLAH

EDIT: NOPE IT DOESN'T IT CAN PROBABLY WORK AS IS

'division by zero' in kernel error seems to hang the audio.
Perhaps this is the issue.

I say this because the "CLICK" you hear when initialising your audio is the same click I was hearing when trying to test:

arecord -D 'plughw' -r 8000 -f S16_LE -c 2 | aplay -D 'plughw'

this does not work initially for the reasons outlined below.

perhaps this could be addressed.


I tested pjsua again and once again the kindle side eventually died on:

--end msg--
 13:01:30.740    pjsua_app.c  Call 1 is DISCONNECTED [reason=408 (Request Timeout)]
 13:01:30.740    pjsua_app.c  
  [DISCONNCTD] To: sip:192.168.1.65
    Call time: 00h:00m:00s, 1st res in 32006 ms, conn in 0ms
    SRTP status: Not active Crypto-suite: (null)
 13:01:31.740  pjsua_media.c  Closing sound device after idle for 1 seconds
 13:01:31.740  pjsua_media.c  Closing mx35luigi:  (hw:0,0) sound playback device and mx35luigi:  (hw:0,0) sound capture device

>>>>

I know this can be prevented with flags (the closing the device) but I suspect it speaks of a deeper error.

One more thing all this set DEV=0 as opposed to -1 stuff. from my reading that is basically saying Choose Device 0 as opposed to choosing the first from a list of ANY - containing one item - device 0

I.E. pointless

Hope that helps


The vague lead on a fix:
=======================================

FROM http://elinux.org/RidgeRun_LeopardBoard_SDK_Hints

Solutions to known issues
'division by zero' in kernel error when I try to use ALSA input

This is a known issue that has been discussed on the ASOC mailing list. The problem is that the function davinci_pcm_enqueue_dma performs this operations:

data_type = prtd->params->data_type;
count = period_size / data_type;

The first time, data_type is set to zero, so we get the error Division by zero in kernel. This is the reason why the pipe arecord | aplay works after running arecord or aplay first.

Apply the Media:Asoc-davinci-pcm.patch to the kernel 2.6.29 to resolve this issue.

================================================== ======

[nick-tech]

Has this project lead to any usable output? I would love to Sip with my kindle?

[knc1]

Originally Posted by nick-tech

Has this project lead to any usable output? I would love to Sip with my kindle?

No - lack of time
Are you volunteering?

[nick-tech]

I am, but I don't really know how to code. I'm more into electronics.

But if I can help in any way, let me know.

[knc1]

Originally Posted by nick-tech

I am, but I don't really know how to code. I'm more into electronics.

But if I can help in any way, let me know.

That is nice to know.

But the person with the most complete (non-released) Kindle VoIP code is away from the forum dealing with non-hobby issues.

The recent problems (opportunities for creative coding) presented us by Amazon on the 4Gbyte devices along with firmware version 5.4.5 will probably grab everyone's free time first, before this (or anything else on the "to do" list).