svn commit: r231814 - in head/sys: kern sys

[Marcel Moolenaar]

On Feb 17, 2012, at 1:09 AM, Attilio Rao wrote:
>> 
>> The message buffer does not have to be a chunk of memory that
>> we circularly scribble to. It can be a per-cpu linked list of
>> messages even.
> 
> Do you think we could inherit much of the code from KTR subsystem?
> We could use KTR as a scheleton (and possibly improving it) for making
> a general circular-buffer and then create a new KPI, thus add
> implementations for KTR, msgbuf, etc.

Aaah... I like your thinking. See also my response to pjd at .
If printf becomes a wrapper around a log infrastructure, then
ktr could potentially be a log of some "high" verbosity or
even some specific facility (or however we can augment messages
to make filtering and distribution easy and flexible).

I like that. Fewer ways to do fundamentally the same thing is
good. Not to mention that if our printf/log function has the
strength to handle KTR functionality, we also open the door
for getting those KTR messages pushed out of the kernel and
into some analysis or monitoring tool... Watch the events as
they happen...

-- 
Marcel Moolenaar
marcel at xcllnt.net