[linux-dvb] Updated Mantis VP-2033 remote control patch for Manu's jusst.de Mantis branch
rscheidegger_lists at hispeed.ch
Mon May 19 19:47:22 CEST 2008
On 19.05.2008 18:52, Pauli Borodulin wrote:
>> On 5/17/08, Pauli Borodulin <pauli at borodulin.fi> wrote:
>>> Since there has been some direct requests for this via email, I'm
>>> posting a updated version of Kristian Slavov's original remote control
>>> patch for Manu's jusst.de Mantis branch. The new version is
>>> functionally the same as the one I posted in March.
> Markus Rechberger wrote:
>> +int mantis_rc_exit(struct mantis_pci *mantis)
>> + mmwrite(mmread(MANTIS_INT_MASK) & (~MANTIS_INT_IRQ1),
>> + cancel_delayed_work(&mantis->ir.rc_query_work);
>> + input_unregister_device(mantis->ir.rc_dev);
>> + dprintk(verbose, MANTIS_DEBUG, 1, "RC unregistered");
>> + return 0;
>> this might be dangerous when unloading the driver because the callback
>> function might still be running after cancel_delayed_work.
>> I ran into that problem a while ago and it could lock up the whole
>> input system.
> Thanks Markus! I created a revised version of the patch with call to
> flush_scheduled_work() after calling cancel_delayed_work. I also changed
> RC polling a bit so that it would work alike on different kernel
> configurations (thanks go to Kristian Slavov for pointing this out).
> The patch is against
No offense, but I like my patch much better :-) . I fail to see why
polling has to be done - just for half-working (at best on some cards,
not at all if the native repeat rate is too low) "improved" auto-repeat.
I was under the impression that using cancel_rearming_delayed_work
instead of cancel_delayed_work (as I did in my patch) would make it
unnecessary to call flush_scheduled_work (but I just followed some other
drivers and could be easily wrong).
As for the IR codes being in common ir code, I didn't care but the
dvb-usb driver also uses its own tables - though I see this driver
probably has reasons to do so.
More information about the linux-dvb