Note that neither nanosleep nor sleep are guarenteed any kind of accuracy in
actually running AFAIK. Ie, the efect of chrony, especially once it has
settled down, on sleep or nanosleep are completely lost in the noise of the
procedures themselves.

William G. Unruh __| Canadian Institute for|____ Tel: +1(604)822-3273
Physics&Astronomy _|___ Advanced Research _|____ Fax: +1(604)822-5324
UBC, Vancouver,BC _|_ Program in Cosmology |____
Canada V6T 1Z1 ____|____ and Gravity ______|_

On Sun, 3 Sep 2017, xxhdx1985126 wrote:


By the way, I guess the subsequent "nanosleep" will also be influenced just like 
"sleep", right?

At 2017-09-02 14:18:52, "Bill Unruh" <> wrote:
>Yes it would. If the clock were slowed by 500PPM (which is a large slew rate)
>a 1 second sleep would take all of .5 milli seconds longer. Now chrony can
>slow the clock more than that but only under severe circumstances and never by
>more than 10%.
>You can tell chrony to jump the clock if its error is too large. (that is an
>infinite slew rate), but that will not happen. In normal operation the slew
>rate may change by a few parts per million-- ie, it is not something to worry
>William G. Unruh __| Canadian Institute for|____ Tel: +1(604)822-3273
>Physics&Astronomy _|___ Advanced Research _|____ Fax: +1(604)822-5324
>UBC, Vancouver,BC _|_ Program in Cosmology |____
>Canada V6T 1Z1 ____|____ and Gravity ______|_
>On Sat, 2 Sep 2017, xxhdx1985126 wrote:
>> Hi, everyone
>> >> I'm a newbie to chrony. I saw these words in "chronyc manual page": "any error in the system clock
>> is corrected by slightly speeding up or slowing down the system clock until 
the error has been
>> removed, and then returning to the system clock’s normal speed". I wonder what 
does the "speed up
>> the system clock" mean? Would it influence the execution of glibc APIs like 
"sleep"? I mean if
>> chrony decides to speed up the system clock, would sebsequent "sleep" 
function calls take less time
>> to return? And is there a max error bound in time for chrony?
>> >> Thanks very much:-) >> >> >>   >> >> >> >>   >> >> >>


Reply via email to