Re: [PATCH] Some clean up of the time code.

[george anzinger <george@mvista.com>]

Andrew Morton wrote:
> george anzinger <george@mvista.com> wrote:
> 
>>-void do_settimeofday(struct timeval *tv)
>> +int do_settimeofday(struct timespec *tv)
>>  {
>> +	if ((unsigned long)tv->tv_nsec > NSEC_PER_SEC)
>> +		return -EINVAL;
>> +
> 
> 
> Should that be ">="?
> 
> Is there any reasonable way to avoid breaking existing
> do_settimeofday() implementations? That's just more grief all round.

Hm. Giving this more thought, the main reason for the change was to 
move to the timespec from the timeval, i.e. nanoseconds instead of 
microseconds.  The error check was put in because the function was 
already being changed.  The reason to move to the timespec is to 
complete the change made to xtime and to more correctly align with the 
POSIX clock_settime, both of which use timespec.

I suspect Linus would prefer this direction...

Comments?
> 
> 

-- 
George Anzinger   george@mvista.com
High-res-timers:  http://sourceforge.net/projects/high-res-timers/
Preemption patch: http://www.kernel.org/pub/linux/kernel/people/rml