Re: Droid 4 modem support in kernel & ofono

[Tony Lindgren <tony@atomide.com>]

* Pavel Machek <pavel@ucw.cz> [190207 22:18]:
> Hi!
> 
> > > Ok, so I got calls and smses somehow working in kernel&ofono ... which
> > > is really all I need.
> > 
> > Nice :)
> > 
> > I think the SIM card reading and writing should be doable
> > using dlci10 /dev/motmdm10 for AT+CRSM calls..
> 
> Might be. Sorry, this is outside of my area of interest, because LTE
> would not be on usable frequencies, anyway.

Oh I did not mean LTE, I meant for general SIM access for
ofono for pin codes etc. Anyways, if anybody needs the sim,
it should be available at /dev/motmdm10.

> > > I pushed the tree to git@github.com:pavelmachek/ofono.git , branch
> > > d4... But I had to do some "rather interesting" hacks. D4 modem
> > > expects packets and current kernel drivers rely on write() boundaries
> > > and flush(). .. which is a bit of problem for in ofonod, as it expects
> > > to work with bytestream with no explicit packet boundaries.
> > > 
> > > However D4 still uses normal AT commands, so... it would be good to be
> > > able to use AT parsing framework in ofono.
> > > 
> > > I believe easiest solution would be to automatically do the packet
> > > splitting in kernel, it should be as easy as splitting on \r and
> > > ^Z. (Currently packets are only generated when \r or ^Z is seen on
> > > write boundary, but that does not work well for ofono).
> > 
> > OK yeah it's worth trying. I hit that issue too with the
> > flush needed for droid4-sms-tools scripts. And the traffic
> > we're seeing is minimal and AFAIK there's no network port
> > for ts27010. And SMS messages are PDU encoded anyways.
> > 
> > Hmm should we do it for \r\n and \r? Otherwise the \n
> > will be left out of the packet :)
> 
> I guess splitting on \n makes sense, yes. 
> 
> Ofono normally uses just \r. Translating it to \r\n in kernel (ttys
> already do that) would be super nice, but I think this is easy enough
> to handle in ofono.

Do you have some simple test case with printf?

This test case works just fine already:

$ printf "AT+CFUN=1\rAT+SCRN=0\r" > /dev/motmdm1

But maybe the modem behaves in a different way for different
dlci.. So a minimal test case would be good to have :)

Regards,

Tony