[mythtv-users] Sudden 'disappearance' of dvb tuners

[John Pilkington]

On 20/04/13 20:52, HP-mini wrote:
> On Sat, 2013-04-20 at 11:45 +0100, John Pilkington wrote:
>> I just rebooted my Fedora 17/MythTV 0.25.3 main box after upgrades that
>> included kde 4.10.2 and dvb-apps.  Mythbackend couldn't see any tuners.
>> dmesg showed the pci cards ok but no firmware load for the usb device.
>>
>> Mythtvsetup didn't see the cards either, until I tried running femon for
>> each adapter.  pci cards then worked, usb not found.  Unplug/replug usb;
>> firmware loaded and it's probably ok.
>>
>> Someone else posted recently about tuners that worked ok after being
>> used once, but I haven't relocated the post.  This feels similar.
>>
>> This box is non-standard at present because the f17 nVidia packages that
>> I would like aren't at ATrpms, so I'm still running an f16 kernel - and
>> wondering where to go next!
>>
>> John P
> Could just be warm start (no firmware reload) was problem with dvb-apps
> update tho that seems very unlikely. And the f/w almost never changes.

Something changed.  IIRC I rebooted twice - warm reboot - and got the 
same result each time.  I had had adapter recognition problems before - 
Ticket #11058 etc - but after the femon workaround during mythtvsetup 
reboots have been ok, until today.  I think it's working now but will 
see what happens at the next reboot.

>
> You could try building v4l-dvb from source..
> http://linuxtv.org/wiki/index.php/How_to_Obtain,_Build_and_Install_V4L-DVB_Device_Drivers
>
> The "basic" approach is fairly easy..
>
> I recall having to disable some IR remotes to get this to build on
> 2.6.32 kernel..
>

I've been trying that for the SL6 laptop for the past week; the failure 
mode changes but no success yet.   There's a daily build-status report 
too, although it doesn't cover the exact SL6 kernel.  If things don't 
improve in the next few days I'll have another try at editing the 
hardware list; when I tried it earlier in the week, that bombed too.

The obvious f16/f17 problem is the missing ATrpms nvidia kmdl, not v4l, 
which those kernels can probably handle - but now with a new gotcha. 
Yes, a v4l rebuild might address that.

I could probably build the missing nvidia kmdl too, but a switch to 
rpmfusion might offer a longer-term solution.