[mythtv-users] Firewire modules do not load

[Jarod Wilson]

On Aug 21, 2009, at 3:08 AM, Axel Thimm wrote:

>> But if you really think you need the old drivers, you
>> will have to either build them yourself, or find someone who ships a
>> kernel module package containing them. Not sure if atrpms does any  
>> more
>> or not.
>
> It doesn't - I thought all firewire issues were a thing of the
> past. But I get more and more request to resurrect the old
> modules. What do you think, should there be legacy modules again?

I'd have to say no, but then, the new stack Just Works for me, and  
apparently doesn't for others... Note that I think its gotta be  
something cable box dependent, because I've hooked up at *least* six  
different controllers to my cable box using the new stack, and they  
all perform just fine. (Controllers in question would be a Ricoh  
R5C832, a JMicron JMB381 rev b, a Via VT6307, an Agere FW323, a Texas  
Instruments TSB43AB23 and whatever it is on the SoundBlaster Audigy.  
Can't recall if I tried an NEC controller or not. Yes, I have a LOT of  
FireWire cards here...).

>> Sidebar: why is it that I keep getting firewire cable boxes that Just
>> Work, rather than one with issues that would require me to actually
>> dig into why we still have issues w/some cable boxes... :)
>
> That's the Inverse Murphy Law and anyone would like to trade places
> with you (because everywhere else the normal Murphy Law applies :)

Heh, indeed.

> Actually I hadn't had any firmware issues since F10 with the (very
> few, read one) firmware capture devices I have.

You have firmware capture devices? ;)

Since Jay Fenlason finished off the ip-over-firewire driver a little  
while back, the only remaining deficiencies I'm aware of are a lockup  
when using nForce2 onboard controllers (extremely rare device) and ALi  
M5271 controller (also rare, but still available as a PCI card). Some  
TI cards have issues with dual-buffer isochronous receive mode, but  
they've all been quirked to just fall back to packet-per-buffer mode  
in the kernel for some time, and the new Agere FW643 has recently been  
discovered to be deficient when using more than one isochronous  
context at once -- first dual-buffer stream works fine, second falls  
down, but its been worked around too (context > 1 uses packet-per- 
buffer, and this is probably only relevant for folks w/more than one  
firewire cable box). Oh, there's also a segfault using dvgrab with  
certain DV cameras that nobody has been able to figure out, since its  
apparently quite rare.

-- 
Jarod Wilson
jarod at wilsonet.com