[linux-dvb] Re: [PATCH] Skystar2 pid filters reorganization

["Augusto Cardoso" <augusto@carhil.net>]

Hi!

There is no need for delay between writing and reading email. B2C2 had a board that was FPGA based and under certain circumstances, it was necessary to wait. With the ASIC based chips, it's no longer needed. It's possible that in some old driver, that was reversed engineered, some of these old code still there.

The only time you really need to use the flexcop hw pid and section filters, and possibly the demux, is for the USB product. This is because the USB bus doesn't have the band width to open the whole transponder.The current driver doesn't support the B2C2 USB product. I know they sold some good quantities in Italy and perhaps Germany and Russia(?). The USB product uses FCIII, but most of the registers are the same. There are some FCIII PCI cards as well. 
B2C2 is coming up with a new DVB-T PCI and USB product, at a low price. They use FCII, FCIIB and FCIII. They don't make FCII any more, but still have some inventory. FCIIB and FCIII are still in production.
The skystar2.c driver could be modify to support the USB product, if there is enough demand for that.

IMHO the main goal should be to have a very stable driver in the first release of the 2.6.0 kernel.

Having the HW filters is important only for multiple cards per system or for the USB. Otherwise, it should be fine with the whole transponder feature. Note that the chip is filtering the NULL packets. This can be turn off as well, but it shouldn't.

I haven't seen a list of current bugs or issues. The only thing that I'm aware is the slow scan. It sometimes fails as well. Increasing the timeout improves the fail rate.

Do you have a list of problems?

Overall, I think the skystar2.c is a very nice driver, specially if you take in consideration that it was done with no support from B2C2. I think that was a remarkable thing. It has room for improvement.

	Augusto

*********** REPLY SEPARATOR  ***********

On 11/27/2003 at 9:31 PM Wolfgang Thiel wrote:

>On Thu, Nov 27, 2003 at 09:13:32AM -0800, Augusto Cardoso wrote:
>> ...
>
>Hi Augusto, many thanks for your interesting email.
>Do you think some delay is needed between writing and reading the
>registers?
>
>It's nice to have a more-or-less working skystar2 driver also for
>2.6B, but, IMHO, we should reduce the pace trying to improve it.
>We all know that it has a lot of bugs, and we should try to locate
>these first.
>
>My proposal:
>Let's start with a debug on a very low level, first:
>Let's start with open_whole_bandwidth() first, and forget about
>hw-filters: that way, we can check the upper layer, and - if it
>isn't the upper layer's fault - we will have to check the irq rou-
>tines.
>Scan will be a good test program for this, as it is requesting quite
>a few pids/filters at the same time.
>
>If this turns out to be fine one day, we can add hw-filters (0-5, and
>the additional 0-31), one day.
>
>I also propose a new debug value: debug=2 should mean the debug
>we already have, the filter stuff dprintk's should be changed to some-
>thing like iprintk, with a debug value of 1.
>
>What do you think?
>
>Wolfgang
>
>
>-- 
>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.