Ingo Molnar <mingo@elte.hu> writes:I would suggest to drop the tsc.c change. The "small errors" can be really large on some systems and you can also see large backward jumps. I have a proper (but complicated) solution pending in ftp://ftp.firstfloor.org/pub/ak/x86_64/quilt/sched-clock-share BTW with all this CPU time measurement it would be really nice to report it to the user too. It seems a bit bizarre that the scheduler keeps track of ns, but top only knows jiffies with large sampling errors. -Andi -
