| Date | Wed, 13 Nov 2019 23:30:58 +0100 (CET) | From | Thomas Gleixner <> | Subject | Re: [PATCH 22/23] [RFC] y2038: itimer: use ktime_t internally |
| |
On Fri, 8 Nov 2019, Arnd Bergmann wrote: > Avoid the intermediate step going from timeval to timespec64 to > ktime_t and back by using ktime_t throughout the code. > > I was going back and forth between the two implementations. > This patch is optional: if we want it, it could be folded into > the patch converting to itimerspec64. > > On an arm32 build, this version actually produces 10% larger > code than the timespec64 version, while on x86-64 it's the > same as before, and the number of source lines stays the > same as well.
Right. For 32bit without a native 64/32 division this is going to be more text and I'm not really convinvced that this buys us anything.
Thanks,
tglx
|