Re: SMP per-CPU statistics under FreeBSD.

On Sun, Apr 23, 2006 at 01:30:18AM -0500, Bill WIlson wrote:
> On Sun, 23 Apr 2006 01:10:00 +0000
> Marco van Tol <[email protected]> wrote:
> 
> > The attached patch makes it work only in non-client/server mode.  It's
> > really cool, I like it a lot. :)
> > 
> > In client/server mode it only shows one CPU.  I'll look into why that is,
> > but I wanted to send you this patch already so you can see whether you like
> > the way I wrote it or not.  It'd be nice if this ended up in your CVS tree
> > at some point when you're satisfied with what I did. ;)
> > 
> > I tested it both on FreeBSD 6.1-RC1 without the pcpu_time patch, which makes
> > it work the way it does now, and on FreeBSD 7-CURRENT with the pcpu_time
> > patch which makes it show the number of CPU's in my system with specific
> > data for each of them.
> 
> As far as being satisfied with it, I have to rely on yours and others input
> since I don't have access to a FreeBSD machine.
> 
> > I used the netbsd.c in sysdeps/ as sample code while writing this so thanks
> > to Bill Wilson who is listed as its author for that.
> 
> Actually, that's confusing.  The copyright notice there is for "GKrellM" for which
> I am overall the author except for several of the machine dependent files in the
> sysdeps directory.  And for those files I should make the copyright notices
> more clear since I am author of linux.c, but Anthony Mallet is author of
> netbsd.c and Hajimu UMEMOTO is author of freebsd.c. 

You're right, sorry.  My fault though, I didn't read it clear enough. :(
It was late, is that valid?

Thanks to both you for the overall code, and Anthony Mallet for the netbsd.c
code which I used as sample code.  And sorry I got it the wrong way around
the first time.

> So I hope that
> Hajimu will have the chance to review these changes.

Hope so too, and hope he likes what he sees. :)

> For client/server mode debugging, run the client gkrellm with the
> "-d 0x1000" flag and look for <cpu_setup> and <cpu> lines in the output.
> It really should be working so long as you are running the gkrellmd
> compiled with your freebsd.c changes.

Ah, thanks.  I'll see what needs doing to make it work.

Thanks,

Marco

-- 
The difference between theory and practice
is a lot bigger in practice then in theory
- Peter van der Linden in `Deep C Secrets'

_______________________________________________
Gkrellm mailing list
[email protected]
http://ninja.linux-phreak.biz/mailman/listinfo/gkrellm