[linux-dvb] Re: Autoscanning, was [PATCH] mt352 TPS_GIVEN shiftswere wrong

[Holger Waechtler <holger@qanu.de>]

Wolfram Joost wrote:
> Hi,
>
>
> > btw: does anybody of you has an idea how we might make use of the
> > automatic channel scan functionality of the MT352 under Linux, how we
> > might need to extend the LinuxDVB API?
>
> I would add two ioctls:
>
> - FE_START_SCAN with the lowest and highest frequency to scan as paramters.
> - FE_CONTINUE_SCAN
>
> After the FE_START_SCAN the kernel-thread monitors the fe-status and send an 
> event if the fe locks or an event when the fe finished scanning. Monitoring 
> than stopps.
>
> The application then can get more information by requesting some sections and 
> of cause it should get the channel-parameters with a FE_GET_FRONTEND ioctl. 
> Then, the application continues scanning by issueing the FE_CONTINUE_SCAN 
> ioctl.
sounds good so far.

> The kernel-thread starts monitoring the fe-status again.
in well-designed interrupt-driven STBs we don't need any frontend thread 
at all.

> The FE_GET_STATUS ioctl should set a new flag to indicate that the 
> scanning-process has finished.
end of scan can get determined by comparing the end frequency with the 
current frequency.

I would propose a 'scan in progress' flag instead (it maps well to the 
internal '352 state bit), this allows us to distinguish between 
weak/instable transponders (lock lost between scan interrupt and 
FE_GET_STATUS), scan-in-progress (no lock because no signal) and end of 
scan.

:) Are you volunteering to implement this feature?

Holger