[linux-dvb] Re: stability problems + closing and re-opening the frontend

["Valsecchi Patrick" <patrick.valsecchi@nagra.com>]

Hi,

In my kind of application (monitoring) I don't really care if I loose some packets. Would it be possible to reset the firmware when that happens? That's what I wanted to try to do by closing the firmware and re-opening it.

Maybe when can trigger that reset when the "waiting for command idle" problem appears. Then send back the tune command. That way, we recover the card without manual operations and without the user level to notice it (maybe a slower tune, though). Or, if you guys think it's too radical, maybe add a reset ioctl. That way, the user level can solve the problem by itself.

I'll try to find a "budget" card. Seems like those are a bit more stable, isn't it?

Thanks.

> -----Original Message-----
> From: linux-dvb-bounce@linuxtv.org
> [mailto:linux-dvb-bounce@linuxtv.org]On Behalf Of Guido Fiala
> Sent: Saturday, April 03, 2004 7:39 PM
> To: linux-dvb@linuxtv.org
> Subject: [linux-dvb] Re: stability problems + closing and 
> re-opening the
> frontend
> 
> 
> On Friday 02 April 2004 18:11, you wrote:
> > Valsecchi Patrick wrote:
> > > All,
> > >
> > > Here is the test code. It's butt ugly, but does the job 
> and it's short.
> > > You need a satellite tuner and need to edit the dvbBug.c 
> file to put two
> > > frequencies you have on your dish. Last run, it crashed 
> the ARM in 7516
> > > iterations. So it may take some time.
> > >
> > > Is it possible it's a firmware bug? If yes, who is 
> developping it? Do we
> > > have the source code?
> > >
> > > Thanks
> >
> > Yes it's possibly a firmware bug introduced after 0.9.4 because that
> > release doesn't show the behavior or at least not within 
> 1000 channel
> > changes or signal-less operation.
> >
> > Guido Fiala, Andreas Share and others have tried working on 
> this just a
> > couple of weeks ago see
> > 
http://www.linuxtv.org/mailinglists/linux-dvb/2004/03-2004/msg00521.html
>
> The firmware has to be closed because of licensing issues and NDA for
> cryto stuff. Johannes Stezenbach at convergence amongst a few others is
> a firmware wizard and said in the thread: "IIRC Ralph said that he tried
> to debug this in the firmware, and
> the ARM just hangs inside a RTSL call. There isn't much we can do then."

That's right - and as was to read at the vdr-ml i traced it down to the 
tuner-feed of the av7110. If the ves1893 (the frontend-chip) does not feed 
any data or interrupts in the ARM, the ARM is perfectly stable, i currently 
reached an uptime of 6 days without ARM-crash while in continuous replay, 
heavy OSD-load, changes in Playmode, jumps etc etc ... 

However, this is only a solution to systems without a dish, as the chip is 
required to decode the Sat-signal.

I had hoped this would help the driver developers to fix this in the firmware, 
but maybe there is simply no way to circumvent this.

Guido


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



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