Re: MSR driver

Stephan Meyer (Stephan.Meyer@pobox.com)
Fri, 14 Nov 1997 21:32:23 +0100 (MET)


On Fri, 14 Nov 1997, Dean Gaudet wrote:
> Out of curiosity, what MSRs other than the pemo MSRs are people interested
> in dinking with? The pemo stuff is a special case that can be dealt with
> another way entirely -- it can become part of task state and switched like
> the FPU is switched.

Yes, just like that.

IMHO, there should be two interfaces to the MSRs, both implemented as
machine-independent to user-processes, regardless of the interface used.

(a) Bare-bones access to physical MSRs of each machine. The usermode
program needs to take care of accessing the right registers.
We cannot leave this out as there are more interesting things to
fiddle with than "just" the performance counters.

(b) Machine dependent access to performance counters, perhaps even
providing a common list of event identifiers. Say, a union of
the events found on different processors.

IMHO there should be no multiplexing support in the kernel as this adds to
kernel bloat and complexity and be implemented just as well in a usermode
application (see my web page for the link to perfmon),

> Dean

------------------------------------------------
Stephan.Meyer@pobox.com
meyerst@informatik.uni-muenchen.de
http://pobox.com/~stephan.meyer/
2A 64 F0 73 02 91 10 FC 18 CC 83 1E E2 2C 7E 79

So what would an angel say
The devil wants to know