| All other DEBUG options are set via 
| 
| options <foo>DEBUG

That's not true. in the same dir, RAY_DEBUG isn't like that.
In dev/ic, AIC_DEBUG, HMEDEBUG, PCIDEBUG, WDCDEBUG, are all
on by default. I could go on.

| It'll bail out with a redefinition error. That's 
| 
| 1. Non obvious as hell

The *DEBUG interface isn't documented, and I don't think we
have to stick to it. It might be nice to, but that would
be a good reason to wrap the #define with a #ifndef, not
a good reason to do as you have done.

| 2. Completely opposite of how every other standard debug #define gets set

I think there's more variety than you're aware of.

| This should be setup the same as any other debug arguments. If you want to
| have debug turned on for something, set it via an option in the conf file.
| Otherwise the extra code shouldn't even be generated.

Again, this is debatable. If the extra code doesn't incur a performacne
penalty, then there's no compelling reason why it must not be present,
especially when it can be very useful to have in all kernels.

| Past that I agree with being able to set the check var via ddb.

Parse failure.

| But this all goes back to DEBUG not getting set in GENERIC builds
| which this violates the premise of being able to control.

No. "DEBUG" doesn't mean "*DEBUG".

In summary, I don't see a compelling reason that you're offering why
PCMCIADEBUG must be off by default, and you have just removed a valuable
debugging feature from the operating system, and I object to that.

Response?

--jhawk