Porting Device Driver to 2.6

Dominik Brodowski linux at dominikbrodowski.net
Sun Apr 24 15:35:20 EDT 2005


On Wed, Apr 20, 2005 at 09:26:56AM -0400, Alex Holeczy wrote:
> If the detach() function is suppose to call the release() function and the 
> release() function is not suppose to free any resources allocated by the 
> config() function for the device.

Possibly you want to consider this patch:


which shows how this timer got removed for 2.6.

> By 'serialize this behavior', do you mean that the detach() function 
> should not be called by the 2.6 kernel if the device is still open?

No, I meant that _release() is called inside the other function, but not in
a different thread (the timer thread).


More information about the linux-pcmcia mailing list