com1 incorrectly associated with ttyd1, com2 with ttyd0

[Bruce Evans]

On Wed, 30 Nov 2005, Joe Rhett wrote:

>>> On Thursday 17 November 2005 05:03 pm, Joe Rhett wrote:
>>>> We can't.  Serial A is a 9pin serial port, and Serial B is the rj45 console
>>>> port.  This is how the motherboard is built. We need Serial B to be the
>>>> console.
>>>>
>>>> /boot/device.hints clearly indicates that 3f8 should map to sio1.  Why
>>>> isn't it using these hints?
>
>> On Thu, 17 Nov 2005, John Baldwin wrote:
>>> What if you disable ACPI?  I think the ACPI bus doesn't use the port
>>> information to honor "wiring" requests but just reads the flags.
>>
>> On Thu, 17 Nov 2005, John Baldwin wrote:
>> ACPI me harder :-).
>>
>> If it doesn't match the port info then it probably gets the flags wrong
>> too, by applying the flags intended for the port at 0x3f8 (intended unit 0)
>> to the port that ends up as unit 0 (0x2f8 here).
>
> Pardon me, what is "it"?   The BIOS or the apci module?

ACPI in FreeBSD.  It sees the ACPI configuration and can/should see the
static configuration.

>> Console initialization uses the flags to decide the console port(s),
>> so swapped flags probably give swapped console ports.
>
> I don't know if this is related, but ACPI also seems to find devices which
> are disabled (ie parallel parts, etc) and then complains because it can't
> allocate resources for them.

It's not clear that disabling in the BIOS should disable for all OSes.

> It sounds like ACPI isn't reading the BIOS data correctly at all.  What
> debug options can I enable to get more information?

Don't know.  I avoid ACPI if possible :-).  I suspect that FreeBSD can see
ACPI tables but not all BIOS tables, so any soft disabling in the BIOS gets
lost.

Bruce