I wrote:

>> I've also run into some problems with serial consoles and the latest
>> scc code.  Doing a shutdown can hang a machine.  The problem is
>> somewhere in the scc code: either in the dialin/dialout support in
>> sccopen(), the spls() around that code, or in my incorporation of
>> Takuya Koumoto's work on the scc driver.  

I forgot to say: that problem seems to be with dialin/dialout support.
On a machine with a serial console which doesn't have carrier wired
up, processes which try and open /dev/console hang. Typically on reboot.

Of course, i could easily have botched applying the dialin/dialout changes.

Anyway, Takyua Koumoto's changes have been in -current for a couple of
days now. it seems to work fine on a 3MIN.


Takyua Koumoto replies:

>Can I assume that there is almost no difference on the
>hardware implementation around the Z8530 between TC-Alphas
>and PMAXs? If so, I think the parameter for wr11 is invalid
>to initialize the channel A(keyboard and mouse). For alpha,
>I have changed it from
>
> 	value = ZSWR11_RXCLK_BAUD | ZSWR11_TXCLK_BAUD |
>		ZSWR11_TRXC_OUT_ENA | ZSWR11_TRXC_BAUD;
>
>to
>
> 	value = ZSWR11_RXCLK_BAUD | ZSWR11_TXCLK_BAUD;
>
>in cold_sccparam(). This is derived from the zs driver of
>other architectures.

Yes, as far as I know, the hardware is the same.  The only difference
I'm aware of is that pmaxes don't seem to handle zs settle time in
hardware. By experiment, the pmaxes really do require delays of about
1.6 usecs between certain register accesses.  But that may be due to
old bugs in the driver.  I have not tried without it for some tiem.

I no longer have a databook with the scc in it, so I dont know what
those bits really mean;).  They look like they are putting the 8530
clock on the TxC and RxC pins.  Does the mouse or keyboard actually
take clock bits from the serial cable? I suppose it's possible.

Do you have a pmax machine with an lk-201 and a DEC puck mouse?
If a serial mouse and keyboard with these changes, I guess it's OK.
It certainly seems to work as-is (or as-was).