cvs commit: src/etc remote

[Bruce Evans]

On Mon, 21 Jul 2008, John Baldwin wrote:

> On Monday 21 July 2008 05:22:15 pm Marcel Moolenaar wrote:
>>
>> On Jul 21, 2008, at 8:07 AM, John Baldwin wrote:
>>
>>> On Saturday 19 July 2008 03:08:22 pm Marcel Moolenaar wrote:
>>>> marcel      2008-07-19 19:08:22 UTC
>>>>
>>>>  FreeBSD src repository
>>>>
>>>>  Modified files:
>>>>    etc                  remote
>>>>  Log:
>>>>  SVN rev 180615 on 2008-07-19 19:08:22Z by marcel
>>>>
>>>>  With uart(4) default, change /dev/cuad# to /dev/cuau# so that
>>>>  out-of-the-box FreeBSD is consistent.  Add uart[0-7] as a
>>>>  fingerfriendly shortcut alongside sio[0-7] and com[1-8].
>>>
>>> Perhaps the sioX aliases should either be removed or still
>>> reference /dev/cuadX?  Having them reference the uart(4) devices
>>> might be
>>> confusing.

rc.d/serial was (not quite) similarly broken.  In rc.d/serial, there
is supposed to be a set of config line for each driver.  This is
unfortunately necessary for the same reasons that separate config lines
for ethernet drivers are necessary (it is too hard to map multiple
devices/drivers to the same device namespace, and we don't want that
anyway (anyone here want eth*?).  Now there are no lines for sio
devices, but there are still lines for unusual devices (just a couple
that were maintained enought to have lines there).

>> I was mostly thinking in terms of backward compatibility.
>> sioX will map to a working port now. I can certainly remove
>> sioX from the aliases, but that may break existing setups.
>> I don't know if adding sioX aliases for /dev/cuadX is useful.
>> I think sioX should either not exist or be an useful alias.

I think that uart should not exist, but then I'm biased.

>>
>> Thoughts?
>
> I would just remove them.  I'm not sure how often they are really used since
> they are specific to a driver name as opposed to COM1/COM2 which some boxes
> have as labels on the case for the physical ports.

This is mostly backwards.  COM* was the IBM PC's (or IBM's?) name for
serial ports.  It wasn't really a 386 name, but, 386BSD used it for
the i386 serial driver.  rgrimes hated it more than me, so he changed
the name of the driver from com to sio and the device name from com*
to ttyd*.  I forget if the d was traditional.  More likely the a in
cuaa* was traditional.  "uart" is more specialized than "sio", but not
much more (just the "a" for asynchronous in it).  "sio" is still too
generic (most serial devices are now ethernet or USB), but it is better
than an industry acronym like UART, and much better than "tty".

Of course, the correct name is CIA.  Everyone knows what that is ;-).
(It is Communications Interface Adaptor.  I changed the industry acronym
of ACIA to get a more interesting name.  Initial development of sio's
technology was on a device named an ACIA by its manufacturer (Motorola
6850).)

Bruce