On Sun, Jun 27, 2010 at 7:17 PM, Andy Walls <awalls@xxxxxxxxxxxxxxxx> wrote: > > Jarrod, > > Looking at the patches branch from your WIP git tree: > > Is mceusb_init_input_dev() supposed to allocate a struct ir_input_dev? > It looks like ir_register_input() handles that, and it is trashing your > pointer (memory leak). Eep, crap, you're right. Fixed locally (I think), will test it out and ship off the patch probably tomorrow (exhausting weekend of watching futbol and some heavy-duty bbq'ing, need to turn in early... ;). Just double-checked, I actually cribbed that incorrectness from imon.c, so I'll need to fix it there too. D'oh. > Mauro and Jarrod, > > When ir_register_input() fails, it doesn't indicate whether or not it > was able to register the input_dev or not. To me it looks like it can > return with failure with the input_dev either way depending on the case. > This makes proper cleanup of the input_dev in my cx23885_input_init() > function difficult in the failure case, since the input subsystem has > two different deallocators depending on if the device had been > registered or not. Hm. I've done a double-take a few times now, but if input_register_device is successful, and something later in __ir_input_register fails, input_unregister_device *does* get called within __ir_input_register, so all you should have to do is call input_free_device in your init function's error path, no? -- Jarod Wilson jarod@xxxxxxxxxxxx -- To unsubscribe from this list: send the line "unsubscribe linux-media" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
