Четверг, 27 февраля 2014, 10:24 +01:00 от Linus Walleij <linus.walleij@xxxxxxxxxx>: > On Mon, Feb 24, 2014 at 4:34 PM, Roger Quadros <rogerq@xxxxxx> wrote: > > On 02/24/2014 04:25 PM, Linus Walleij wrote: > >> On Sat, Feb 15, 2014 at 2:34 PM, Alexander Shiyan <shc_work@xxxxxxx> wrote: > >> > >>> - status = twl_i2c_write_u8(TWL4030_MODULE_LED, cached_leden, > >>> - TWL4030_LED_LEDEN_REG); > >>> + > >>> + twl_i2c_write_u8(TWL4030_MODULE_LED, cached_leden, > >>> + TWL4030_LED_LEDEN_REG); > >> > >> Isn't the right fix to actually *check* this status instead? > >> > >> TI dudes? > > > > Yes we should check for error. But the only action we can take is maybe print an error message > > as all the users of this function return void. e.g. twl_set(). > > > > It seems the set() hook of struct gpio_chip also doesn't expect any return value. Wondering if that should change. > > I think in this case simply printing an error and bailing out is > just fine, people will be able to debug from there. WARN_ON_ONCE() is enough? v2? --- ��.n��������+%������w��{.n�����{�� b���ܨ}���Ơz�j:+v�����w����ޙ��&�)ߡ�a����z�ޗ���ݢj��w�f