On Thursday 14 June 2007, Jason Curl wrote: > Mike Frysinger wrote: > > On Wednesday 13 June 2007, souf wrote: > >> Le mercredi 13 juin 2007 à 17:13 -0400, Mike Frysinger a écrit : > >>> plus, i think we're talking microseconds difference here, not seconds > >> > >> in an real time system that counts much > > > > if we were talking costs that'd have to be paid over and over, you might > > have an argument, but we're not ... we're talking about a 1 time cost > > only which is paid at startup > > I'm not trying to start a flame war here, but we try here everything we > can to speed up start up times for embedded systems. You're right, 1ms > is insignificant at start when compared to loading and starting the > kernel. Often in Automotive there are requirements of the order of 100ms > to get basic subsystems up and running, so please don't think this is > not an important topic, it should always be in the "back of our minds".
i'm aware of automotive issues, but i have yet to see one where timing requirements were that tight at boot up and could wait until userspace ... that said, i'm really not convinced Souf's implementation of scanning /dev really is faster (or correct) compared to what's in svn now ... there's a lot of messing around with memory in there -mike
signature.asc
Description: This is a digitally signed message part.
_______________________________________________ busybox mailing list [email protected] http://busybox.net/cgi-bin/mailman/listinfo/busybox
