[linux-dvb] Re: budget-ci CI interface testing

[Andrew de Quincey <adq_dvb@lidskialf.net>]

On Monday 19 April 2004 22:57, Bernhard Schrenk wrote:
> Hi,
>
> I have continued my tests and the CI on my Nova card is now working with
> the latest CVS version driver, the CAM is decoding the channel and I can
> show it with mplayer. There are some problems to solve, but its mainly
> working. Thank you Andrew for your good work.

Oh wow, some success at last!

> The problems I detected so far:
>
> 1) When there is no CAM in the slot, no TS data is received. I think the
> DD1 port has to be switched, when no CAM is present.

I'll have to check that. I'd assumed that the CI interface would pass the data 
through if a CAM is not present (it's _supposed_ to; EN50221 says so). Maybe 
thats incorrect for this hardware though. Or else I'm setting the passthrough 
bit wrongly.

> 2) Detection of CAM insertion does not work. It detects the CAM at driver
> load only.

Weird, it works for me. And with the latest CVS, it no longer reads back the 
config option and checks it, which I thought had fixed the issue for you. In 
that case, can you send a debug dmesg against the latest CVS please?

> 3) There are possibilities to crash the kernel or put it in an endless
> loop, when the driver is unloaded, but not all applications, using it, are
> closed. 

Ah I see, a module usage count issue.... I thought the DVB core stuff should 
take care of that. Sounds like theres a bug in it. Certainly I know for a 
fact that I can unload the tda1004x driver when an application still has the 
frontend open (Again, I thought the usage counter is handled by the 
dvb_core). I apologise for not reporting this before; It had slipped my mind.

> 4) When I send a command to decode one video and two audio(Stereo 
> and AC3) streams, no TS data is received anymore (neither free, nor
> crypted), till the CAM is reseted. This was the problem, why I wrote in my
> last message that the driver does not work correctly, but I think this has
> nothing to do with the Nova-CI driver, it could be a fault of my
> Cryptoworks-CAM. 
> 5) I have also problems with loosing the TS data, like in 
> 4), without decoding 2 audio streams, but I can not say exactly when it
> happens. Could also be a fault of my CAM or some parts of the Nova have to
> be reinitialized after changing decoding parameters.

If you're getting any data at all, then theres not anything else that can be 
wrong with the CI interface as far as I know. Once data is flowing, any data 
interruptions must be due to the CAM module itself.

Currently I only reset the CAM module on initialisation, or when the CI_RESET 
IOCTL is issued. I specifically dont do it when the ca device is opened. 
Perhaps this is wrong. Anyone have any idea?


-- 
Info:
To unsubscribe send a mail to ecartis@linuxtv.org with "unsubscribe linux-dvb" as subject.