Re: CVS commit: src/etc

[Izumi Tsutsui <tsutsui%ceres.dti.ne.jp@localhost>]

gdt%ir.bbn.com@localhost wrote:

> So is there an actual conflict, or a reason not to assign a number?  The
> notion that we shouldn't assign major numbers before 5.0 seems tenuous,
> and moving a MI device to MD files seems much more intrusive, plus it
> leaves mess to be cleaned up later.

The problem is that there is no suggested way how to assign
a new (but traditional one which can't use new MI conf/major)
device major on each port.

For example, several ports have sparse numbering.
Some of them are reserved for future devices,
some might be left without any particular reason,
others might be numbers used by removed devices.
No easy way to determine reasonable one for the new device
on ports I'm not familiar with.

Another example, MI nsmb(4) has odd numbers among all ports.
"98" seems assigned on several ports with some intention,
but not on all ports.

Random assignments will make current situation more messy,
so I chose fixing MAKEDEV files for now, which looked less problematic
because even if it looks intrusive it can be reverted safely.
---
Izumi Tsutsui