Myrddin Emrys wrote:
> Here's a summary of my current position.

Hmm, all look ok.

> I don't know where else to look. The userland code doesn't use the major
> number at all... it just uses the device filename. What else could be the
> issue? What locations could I insert debugging print statements that would
> help you guys diagnose my issue? I put some printf statements into
> altqopen(), but that only confirmed that it was not being called properly/at
> all.

I have no clue; EBUSY is unlikely even for an inconsistent device
number...

Another place you can look at is the entry point for the special
device open call: spec_open() in sys-altq/miscfs/specfs/spec_vnops.c.

You can insert something like:

       if (maj == 75)
		printf("cdevsw[75].d_open = %p\n", cdevsw[maj].d_open);

to get the address of the open routine for cdev 75.
Then, compare it with the result of "nm /netbsd | grep altqopen".

-Kenjiro