Messages in this thread | | | From | "Rafael J. Wysocki" <> | Subject | Re: [PATCH 2/2] cpuidle: fix sysfs output for power_usage | Date | Mon, 17 Dec 2012 13:09:48 +0100 |
| |
On Monday, December 17, 2012 10:46:07 AM Sivaram Nair wrote: > On Mon, Dec 17, 2012 at 08:56:45AM +0100, Rafael J. Wysocki wrote: > > On Monday, December 17, 2012 09:38:15 AM Sivaram Nair wrote: > > > On Sat, Dec 15, 2012 at 01:03:02AM +0100, Rafael J. Wysocki wrote: > > > > On Friday, December 14, 2012 03:17:37 PM Sivaram Nair wrote: > > > > > cpuidle_state->power_usage is signed; so change the corresponding sysfs > > > > > ops to output signed value instead of unsigned. > > > > > > > > What's actually wrong with printing it as an unsigned int? > > > > > > power_usage could have negative values (for example cpuidle/driver.c > > > inits this value to -1, -2 etc. when drv->power_specified is not set) and > > > these shows up badly in the sysfs output. > > > > Does "badly" mean "as big positive numbers"? > > Yes (sorry for not being clearer). > > > > > Should we actually print them at all in those case? Perhaps it'll be better to > > make the file appear empty then? > > May be, but why is power_usage signed in the first place?
Please read the comment in driver.c:set_power_states() for the answer. :-)
> I also noticed > Daniel Lezcano's patches that reduces the scope of this variable. So, > perhaps we can just ignore this change.
OK
Rafael
-- I speak only for myself. Rafael J. Wysocki, Intel Open Source Technology Center.
| |