Mailing List archive

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[linux-dvb] Re: Typhoon CI (budget-ci)



On Friday 28 May 2004 15:44, Andrew de Quincey wrote:
> On Friday 28 May 2004 13:39, Kenneth Aafløy wrote:
> > On Friday 28 May 2004 13:08, Andrew de Quincey wrote:
> > > Aha, that would probably mean its wired the IRQ line (assuming it has
> > > one - it _could_ be polling mode only) to a different GPIO. I think
> > > you're getting too many IRQs for it to be working quite right judging
> > > by that syslog.

I've moved the main ci gpio to be 3 and made DEBICI* != 0x8000000, also the 
interrupt is on 6. With those changes an eject will generate 2 interrupts, 
while an insert gives 3. If I remove the 0x8.. change to debici* the 
interrupts will bounce back and fourth (regenerating it seems). I still can't 
read anything in this mode (every read returns 0), which leads me to my 
question. What is this polling mode you talk about?

I'm also wondering about the XIRQ_EN mode of the debi, because that also use 
gpio3. saa7146:7.15 mentions putting a block transfer on hold by lowering 
gpio3, then beeing able to resume the transfer by lifting it again. Is that 
the only speciality of this mode? Does it have side effects?

What is strange when in this configuration is that inserting/ejecting the cam 
actually locks possibly the pci bus long enough to pause my livetv mythtv 
session for a fraction of a second. So possibly there is still something 
wrong here too.

> > Also, when the module is out the PSR has DEBI_E set, but when it's
> > inserted DEBI_S also gets set. The SSR has no debi related bits set.
>
> Hmm, it does sound like it is definitely using the DEBI if thats the case.

Yes, but it feels strange. I should have emmediate access to the address space 
of the attached device, right?

> > There must be some power line that does not go up, because no read from
> > all of the debi address space returns anything other than all highs,
> > except if I set the TIEN bit of the DEBI_CONFIG word to high (disable).
> > It will then return some value (usually 0xfc) on the first read, then 0
> > on the rest.
> >
> > I'll fiddle some more with it later today.
>
> Yeah - theres various modes the DEBI bus can be in - maybe its different
> for that card. You'll know you get it right when you're able to access the
> CAM's attribute memory.. maybe you could write a thing to try all the
> sensible combinations of DEBI parameters, and see if you can see any of the
> magic strings (e.g. "DVB_CI_MODULE")...

Noticed that the 7146 sheet specifies that tien must be enabled for intel 
mode, to bad I did not check the status regs on that one. Anyways, I did at 
least a small nested for loop with 0 to <6 for gpio(0-3, i<<8), then 
traveling a masked set of configs, looking for the first byte of the config 
space (0x1d). I masked a constant timeout (1111) and tien=0. Still nothing 
showing up.

I've asked Typhoon for some tech specs about the card/ci, but no reply today. 
Just fired off a mail to KNC-1 also in hopes for some info.

Kenneth


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



Home | Main Index | Thread Index