[Bug 224837] Extremly unaccurate nanosleep() called to sleep for 0.6 seconds actually sleeps up to 0.640 seconds

[bugzilla-noreply at freebsd.org]

https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=224837

Aleksander Derevianko <aeder at list.ru> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
            Product|Base System                 |Documentation
                 CC|                            |aeder at list.ru
            Version|11.0-RELEASE                |Latest
          Component|kern                        |Documentation
           Severity|Affects Many People         |Affects Some People

--- Comment #2 from Aleksander Derevianko <aeder at list.ru> ---
(In reply to Conrad Meyer from comment #1)

Thanks, it works like a charm! Now precision is pretty good.

Sleeping for 600 ms, real sleep is 600.081164
Sleeping for 600 ms, real sleep is 600.062304
Sleeping for 600 ms, real sleep is 600.063774
Sleeping for 600 ms, real sleep is 600.101515
Sleeping for 600 ms, real sleep is 600.088049
Sleeping for 600 ms, real sleep is 600.066452
Sleeping for 600 ms, real sleep is 600.062314
Sleeping for 600 ms, real sleep is 600.084151
Sleeping for 600 ms, real sleep is 600.059051
Sleeping for 600 ms, real sleep is 600.062662
Sleeping for 600 ms, real sleep is 600.064423


Now I can google kern.timecounter.alloweddeviation and find a lot of references
to it, but before that googling for "nanosleep + freebsd" return nothing
useful.

May be, adding it in nanosleep(2) man page?

-- 
You are receiving this mail because:
You are the assignee for the bug.