[Gta04-owner] Kernel 3.5 v49-1 for GTA04 has DTMF bug again

Neil Jerram neil at ossau.homelinux.net
Wed Sep 19 10:57:52 CEST 2012

On Wednesday, 19 September 2012 09:20:01, Radek Polak wrote:
> On Tuesday, September 18, 2012 10:37:41 PM Neil Jerram wrote:
> > If so, that's expected because I haven't yet provided an updated fix for
> > that; my first attempt caused an SMS problem and so had to be reverted.
> > (Also note that it's a userspace issue, not kernel.)
> > 
> > I've been running an updated fix for a few days now, and it seems good,
> > but I was being cautious.  It's attached below now, and it would be good
> > if others can try and confirm that it doesn't reintroduce any SMS or
> > modem state problems.  (And also fixes the DTMF problem!)
> Hi Neil,
> i have modified it the patch a bit (modified version attached). It works for me 
> as good as your original one, but does not need QAtChat modifications. What do 
> you think about it? I would like this version better...
> Regards
> Radek

Hi Radek,

That - using send() - was how I first implemented this.  You may remember that I sent half an email by mistake, and then said I had more work to do.

The theoretical problem with send is that if there are other AT commands already executing, or in the command queue, send will not wait for those commands to complete first.  For example, in a call, suppose there's a CLCC query in progress when send is called for a DTMF tone?

For that reason, I thought that using send wouldn't always be reliable, and that we needed a solution which (1) waited for other commands to complete first, and (2) did not wait for a response to the VTS command - i.e. the noResponse solution.

What do you think?  Although it is much simpler, I'm worried that the send solution will occasionally mess up the modem state.


More information about the Gta04-owner mailing list