🚀 go-pugleaf

In article <1992Dec28.152533.25758@chpc.org> ralf@chpc.org (Ralph Valentino) writes:
>I just upgraded my pc's serial card to 16550A's and I've been having
>problems with Mach (mk77 ux37).  Mouse access from them seem fine, but
>if I send any character to a serial port via getty, kermit, or even
>cat, the system locks up tight - so tight it won't even enter the
>debugger.
>
>If I disable the fifo right after I test if it's there, (in
>mk/kernel/i386at/com.c) then everything works fine, but of course this
>defeats the whole purpose of the upgrade.
>
>Has anyone run into a similar problem or have any suggestions as to
>how I can track this down?
>
>-Ralph
>
>ps: I tested the fifo capabilities with OS/2 and 2.6MSD; it seems to
>    work fine.
>
>===============
>Ralph Valentino   (ralf@chpc.org)  (ralf@wpi.wpi.edu)
>Hardware Engineer,  Worcester  Polytechnic  Institute
>Center for High Performance Computing, Marlborough MA

The 16550 turns on 2 unused bits in one of the status registers to let the user know
that it is an enhanced 16450/8250.  Windows 3.0 would do horrible things with this
because it used the data in the status register to directly index into a jump table.
Obviously, using whatever address is in location 10xx is not the same as using whatever
is in location 00xx.  Your symptoms sound very similar.

This is the 1st time I've "tried" to post something in news, so I apologize if I've
screwed it up...

	Joe B.