Re: [PATCH] blk queue io tracing support

[Jens Axboe <axboe@suse.de>]

On Wed, Aug 24 2005, Nathan Scott wrote:
> On Wed, Aug 24, 2005 at 09:08:10AM +0200, Jens Axboe wrote:
> > ...
> > This isn't msec precision, it's usec. sched_clock() is in ns! I already
> > decided that msec is too coarse, but usec _should_ be enough.
> 
> Right you are (I was thinking m-for-micro, not m-for-milli in my head ;)
> - but still, there doesn't seem to be any reason for that divide-by-1000
> and reducing the precision in the kernel rather than in userspace, does
> there?  Doing it the other way means you wont ever have to worry about
> whether it is/isn't sufficient precision for all possible block devices,
> and the precision the tool displays will just be a userspace decision.

I was just worried about wrapping of ->time, but I did make it a 64-bit
unit. So I guess we could go to nsec granularity, there should be plenty
[*] of space. I'll change that too, I'll post an update version later.

* I probably should not make such a statement :-)

-- 
Jens Axboe