Messages in this thread | | | Date | Fri, 18 Nov 2011 20:06:38 +0100 | From | Johan Hovold <> | Subject | Re: [PATCH] Revert "leds: save the delay values after a successful call to blink_set()" |
| |
On Wed, Nov 16, 2011 at 10:21:00PM +0100, Antonio Ospite wrote: > On Mon, 7 Nov 2011 11:36:37 +0100 > Johan Hovold <jhovold@gmail.com> wrote: > > > This reverts commit 6123b0e274503a0d3588e84fbe07c9aa01bfaf5d. > > > > The problem this patch intends to solve has already been fixed by commit > > 7a5caabd090b8f7 (drivers/leds/ledtrig-timer.c: fix broken sysfs delay > > handling). > > > > Signed-off-by: Johan Hovold <jhovold@gmail.com> > > Sorry for not replying earlier, I was on vacation. > > Thanks Johan, I've just verified the timer trigger works fine on 3.1 > indeed even without my changes. But keep reading, please. > > > --- > > > > Hi, > > > > A fix for the problem with the delay parameters not being stored is already in > > 3.1 (and stable). Rather than storing both values at every call to > > led_blink_set only the updated value is stored in led_delay_{on,off}_store in > > ledtrig-timer.c. > > > > I thought that this was better fixed in the led_blink_set() directly, > as delay_on and delay_off are led_cdev fields after all, not > necessarily specific to ledtrig-timer, what about other users of them? > For example I can see led_blink_set() used in: net/mac80211/led.c > and led_trigger_blink(), which uses led_blink_set() in: > drivers/power/power_supply_leds.c > > Maybe I am missing something.
When ledtrig-timer was generalised to led_set_blink() with fallback to software blinking, the blink_delay_{on,off} parameters were moved to led_classdev and now have two, partly overlapping uses: to store the ledtrig-timer delay parameters and to implement generic software blinking.
Note that both uses are internal -- drivers implementing hardware blinking through blink_set() receives both parameters at each call and need not access these fields directly. And apart from the ledtrig-timer sysfs interface there is no other way to read them back either.
In particular, there is currently no need to update them at each call to led_blink_set() as far as I can see.
Thanks, Johan
| |