xiaoxiang781216 edited a comment on pull request #5432: URL: https://github.com/apache/incubator-nuttx/pull/5432#issuecomment-1031276071
> > > If this change will be rejected I will probably need to implement something like http://www.makelinux.net/ldd3/chp-7-sect-4.shtml and base POSIX timers and `timerfd` on top of it. > > > > > > Linux timer API even with the latest code still suffer the same issue as wdog since both use the tick as unit. You should add a new set API like hrtimer which use nanosecond unit: https://www.kernel.org/doc/html/latest/timers/hrtimers.html. And implement all other timer related api on top of hrtimer. > > Since the unit of arch(up_alarm_xxx or up_timer_exx) API in tickless mode use nanosecond unit, it's very lucky that we don't need introduce the new arch API to support the high resolution timer. > > The implementation of of high resolution timers will probably need to rely on MCU HW timer and will be an optional feature. > What I mean high resolution is that the software timer(posix timer, timerfd) has the same accuracy as the hardware timer. So the developer can select the right hardware timer to implement arch_timer_xxx or arch_alarm_xxx API. The problem you hit is that wdog accuracy(one tick) is lower than the hardware capability. > Please understand me right. I'm fine if proposed change is rejected by community and I'm just trying to find a way to achieve the best possible operation for timers functionality because believe that it is the right way of doing things. The best way is what Linux has been done. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
