On Sat, 1 Dec 2007, Benjamin Herrenschmidt wrote:

On Fri, 2007-11-30 at 16:11 -0600, T Ziomek wrote:

quoted

quoted

Possibly, though you aren't supposed to leave EARLY_DEBUG enabled
once you are done debugging :-)

I'm probably not the only person that would turn it on when needed,
think
"well, no harm in leaving it on for the rest of my development, and it
might be handy; just turn it off when we're done".

It's these kind of non-obvious but undocumented things that make a lot
of
OSS code a pain to work with for non-experts [1].  What's the harm in
giving folks a heads-up?

There is no harm, I didn't say I wasn't going to document it, you do
have a point there, I was just mentioning by the way, that leaving
EARLY_DEBUG is generally not a good idea in production.

One of the things that arhc/powerpc provides is the ability for you to
have a single kernel image boot boards with different 4xx processors for
example, or different fsl booke processors. You lose that if you leave
early debug on as it usually contain hard coded addresses for a given
board.

This is typically useful if you have several revisions / versions of
your product, which could use different processor revisions or even
model, and want a single kernel image to support them.

Makes sense (my last PPC work was with arch/ppc, when arch/powerpc was
just getting started, and I'm not using PPCs at the moment).

Maybe the better comment to add for EARLY_DEBUG is to turn it off when
not of immediate concern?

-- 
   /"\  ASCII Ribbon Campaign   |
   \ /                          |   Email to user 'CTZ001'
    X        Against HTML       |             at 'email.mot.com'
   / \     in e-mail & news     |