Am 20.12.2014 um 15:51 schrieb Russell King - ARM Linux: > On Sat, Dec 20, 2014 at 03:11:54PM +0100, Frank Schäfer wrote: >> Hi Russel, > I guess you won't mind if I mis-spell your name too... Wow... it seems to be very easy to offend you... Sorry, that was definitely not my intention. I did not do this on purpose. It was just a simple mistake and I will try to avoid it in the future. > >> I'd prefer to keep the button initialization related stuff together in >> em28xx_init_buttons() and do the cancel_delayed_work_sync() only if we >> have buttons (dev->num_button_polling_addresses). >> That's how we already do it with the IR work struct (see >> em28xx_ir_suspend()). > Provided all places that touch buttons_query_work are properly updated > that's fine, but to me that is fragile and asking for trouble. It's far > better to ensure that everything is properly initialised so you don't > have to remember to conditionalise every single reference to a work > struct. > > In any case, delayed work struct initialisation is cheap - it doesn't > involve any additional memory, it only initialises the various members > of the struct (and the lockdep information for the static key) so there > really is no argument against always initialising delayed works or normal > works, timers, etc to avoid these kinds of bugs. Fair enough ! Regards, Frank Schäfer -- 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
