On Thu, 8 Jul 1999, Robert Walsh wrote:
> We're currently profiling the kernel NFS daemon (the entire path from
> network to disk) using SPECsfs and other benchmarking mechanisms, and
> before I start working on a home-grown profiling mechanism I'd like to
> make sure I'm not reinventing the wheel.
You might want to check on the tool i've been working on, which enables
you to see exactly what if happening on your system at all times. It
would enable you to see what time is spent in kernel, what time is
spent in the rest of the system, and why.
Moreover, the mechanisms I use are easily extendable. Therefore, if
you find the information given insufficient, you can make the kernel
generate more information without modifying too many things.
You can find the Linux
Trace Toolkit at the following address :
http://www.info.polymtl.ca/users/karym/www/trace/
The information on how to use it is there and so is some information
on how it works. If you need more info, don't hesitate to contact me
if you need more information.
Of course, it's all under GPL ...
Regards.
==============================================
Karim Yaghmour
[email protected]
Computer Engineer
Ecole Polytechnique de Montreal
==============================================