pctv452e

I've been using a Technotrend TT 3600 USB DVB-S2 receiver for a couple of years, which has (largely) been working fine under the S2-liplianin pctv452e driver. I've been aware of a lot of documented problems with running this receiver under the 3.x kernel, so I've stuck with the 2.6 series kernels.

Unfortunately I've now had to upgrade to the 3.2.0 kernel for other unrelated reasons, and it seems that the device is more or less unusable under this kernel. With the stock 3.2.0 kernel, the driver produces numerous I2C errors and is quite unreliable. The I2C errors seem to be produced exclusively as a result of stb_6100_read_reg() reading register F, and notably all calls to stb6100_read_regs() seem to succeed, so I've replaced the stb_6100_read_reg() function with a call to stb6100_read_regs(), so it reads all the registers and then returns the requested one, rather than reading just the requested register. This seems to make the I2C errors disappear.

However, the card is still very unreliable - after about 5 minutes of receiving a channel (using MythTV), it breaks. No errors logged in dmesg, but MythTV logs:

DevRdB(/dev/dvb/adapter0/frontend0) Error: Poll giving up DVBSH(/dev/dvb/adapter0/frontend0) Error: Device error detected DVBRec(7:/dev/dvb/adapter0/frontend0) Error: Stream handler died unexpectedly.

Can anyone give me any pointers that might help? I've searched and searched and all I can see if people saying that it won't work since the DVB-S2 code was integrated into the kernel tree, but I've not seen anyone try to figure out _why_ it won't work.

Thanks.

 - Steve

I'm on the same boat.
I have 3 DVB-S2 cards, one of them is pctv452e, and none of them works reliable. Yes, it's very frustrating that card which claims linux support has this support broken. The problem is that community which uses DVB cards is much smaller then those using other equipment. So there are no test of every device in each kernel release.
I'm reading this list for some time and often see patches which refactor some drivers, but the only check is compilation check.
"totally rewritten DVB-USB-framework" will not help to have more drivers too. Makers of dvb cards often don't provide linux driver, or provide only once in form of patches. You will not apply such patch into new DVB stack. Makers must constantly refactor their drivers to suit new kernels, and there are only a few which do this. I don't know why they simply don't join this list and push drivers into kernel and then provide support.

I don't know what's future of DVB stack, but I see that changes in DVB stack causes meany problems with drivers. Often driver is written once and later unsupported - changes make it non-functional and there is nobody who can fix it. Exactly like with pctv452e driver. Marx

Ps. Steve, could you please give me full version of kernel which works with pctv452e?

Your claims about my DVB-USB-framework work is quite bullshit. I am not going to convert all drivers to the new framework just doing compile tests. I am heavily aware it is too risky thus I implemented it as a alternative, converting drivers one by one when you have hardware in hands to test. If such big changes were made to the existing code surely there is breakages.

And it fixes some existing DVB-USB bugs, actually all I was aware. Also it removes all the hacks, and also bugs coming those hacks, from many drivers. Most notable complex AF9015 and AF9035. See AF9015 before change and after change - code size is reduced something like 25% and binary size too.

As the new DVB-USB fixes many bugs I ask you to test it. I converted pctv452e driver for you:
http://git.linuxtv.org/anttip/media_tree.git/shortlog/refs/heads/pctv452e

Only PCTV device supported currently, not TechnoTrend at that time.

regards
Antti

http://palosaari.fi/

I'm not bashing you, I see you are doing very good job, and not only you but many people here. I don't like mess and refactoring is good thing. The problem is that often makers of hardware doesn't support their linux drivers and changes in DVB core doesn't help it. Like with pctv452e - device was working, i've bought it as supported and now it works only partially.
I don't think there is easy solution to that problem. Card maker should support drivers but in reality we should be happy if he do any working driver for linux.

And thank you for modified driver, I will try it at home and of course will write how it is working. Marx

Hi guys,
I was able to got that device, just because of you, and I cannot reproduce that error. I ran it many hours against signal from modulator yesterday and it didn't hang. I used szap && mplayer. Also I ran w_scan. Only problem I saw was these I2C errors printed.

So you make tests, likely using szap, and say how I can make it hang?

regards
Antti

http://palosaari.fi/