Re: ext2fs kernel profiling: most of the time in thread management

[Samuel Thibault]

Svante Signell, le Thu 29 Dec 2011 11:05:20 +0100, a écrit :
> On Thu, 2011-12-29 at 01:25 +0100, Samuel Thibault wrote:
> > I've done some stat-based kernel profiling (I'll have to post how to do
> > it on the wiki sometime) on ext2fs doing nothing more than
> > 
> > while true; do rm -f blop ; \cp -f blip blop ; done
> 
> > It might be biaised by the clock not being able to tick everywhere in
> > the kernel (though I guess e.g. most of the IPC machinery is running at
> > ipl0?), but I believe it's still a bit revealing: I had already noticed
> > that ext2fs spends most of its time in the kernel (like 90%), and it
> > here seems we're spending a lot of time just managing the ext2fs thread
> > sleeps (no, there aren't many threads in that test, just two dozen).
> 
> It's hard to relate these numbers without comparison: What are the
> corresponding numbers for Linux/kFreeBSD?

I have no idea, but what is sure is that Linux and kFreeBSD do not
have such thread management in such case, since the data is already in
the cache, and the read/write calls of cp are thus mere memcpys, while
GNU/Hurd will go through page faulting, paging, etc.

Samuel