From: Marc Horowitz <marc@mi...> - 2005-12-26 17:46:09
I recently purchased a Samsung A900 phone for use with Sprint. I'm
experimenting with the obex push support, and I believe something in
the openobex stack is triggering a firmware bug in my phone. I'm
hoping someone here can tell me what the linux stack is doing
differently, because changing the linux end is likely to be a lot
easier than fixing the phone.
My linux distro is Debian. I'm running the kernel version 2.6.12-1-k7
and the bluez drivers which come with it. The latest packaged version
of bluez-utils is 2.19, so I'm running that. Everything is built
against openobex 1.0.0.
If I push a number from my old phone (Sony-Ericsson T608), or from my
palmtop (Tungsten T3), everything works fine, over and over again.
I have a bunch of vcard files I'd like to load onto the phone. I'm
attempting to use kbtobexclient and btobex, built this evening from
the bluez sf.net CVS repository. The phone and the computer are
If I run this:
btobex push 00:12:47:7E:79:2F 0.vcf
0.vcf looks like this:
TEL;PREF;CELL:+1 617 555 1234
The phone prompts me for a pin, then bluez-pin prompts me, then the
vcard is sent. The phone asks me if I want to accept the push, I say
yes, and the record gets added to the phone book. First, the pin
exchange is unexpected, and after I receive the pb entry, the device
pairing has disappeared on the phone. Second, once the push is
complete, the phone cannot receive any more pushes, from the linux
box, my old phone, or my palmtop. I need to power cycle the phone to
get out of this state. kbtobexclient behaves the same way. Since
both tools are based on openobex, I believe that the problem, whatever
it is, lies there.
I would say that nothing the linux box can do should cause the phone
to behave like this, so there's obviously a bug on the phone's side.
However, the difference in behavior also leads me to wonder if there
might be a bug in the linux stack, too.
Can anybody shed some light on this problem? What might I do to get
this to work better?