8.x and Modems

[Steven Nikkel]

On Mon, 28 Mar 2011, John Baldwin wrote:
>>>>>>>>>> I recently upgraded my trusty old 4.x system to 8.1 and the one
> little
>>>>>>> bit
>>>>>>>>>> I can't get working is the internal ISA modem in the system. On 4.x
> it
>>>>>>> was
>>>>>>>>>> detected automatically by the sio driver:
>>>>>>>>>>
>>>>>>>>>> /kernel: sio4: <U.S. Robotics Sportster 33600 FAX/Voice Int>
> at
>>>>>>> port
>>>>>>>>>> 0x3e8-0x3ef irq 5 on isa0
>>>>>>>>>> /kernel: sio4: type 16550A
>>>>>>>>>
>>>>>>
>>>>>> Ok, here's the proper verbose dump: http://pastebin.com/DJ1z0k4D
>>>>>> I've set it back to PnP mode and taken out all the specific hints.
>>>>>
>>>>> Hmm, no helpful bootverbose messages in the pnp.c code it seems.
>>>>>
>>
>> A first glance tells me the only thing that could be happening is a time out
> trying to read resources. So sure enough after a couple hours of compiling and
> reboots that is the issue. The first check shows it reads a few resources then
> times out trying to read one 45 bytes long. I updated the time out
> dramatically in pnp_get_resource_info to verify and sure enough PNP now
> successfully initializes the card. Curiously I looked at pnpinfo and it seems
> to use pretty much the exact same process, except it was working previously.
> (PS: I set the count to 1000 and delay to 100, extreme overkill, but I didn't
> want to wait for several kernel compiles to get it to work)
>
> Can you try just increasing the DELAY()?  If that works I'll commit it.  I
> wonder if DELAY(1) is broken somehow.  (Would be odd if DELAY(2) worked for
> example, but DELAY(1) didn't.)
>
>> Here are some relevant snippets from dmesg with my debugging bits:
> http://pastebin.com/uKRReBc0
>>
>> Now onto the next bit: uart doesn't seem to identify the card properly. This
> was about the same message I was getting when I hard coded the settings on the
> board (no-pnp) and put them in device.hints
>>
>> However, it does seem to produce results now when I try to use it. Different
> than previous usage, but at least appears to be useful.
>
> So is the card working correctly now?  Is the only problem the string name in
> dmesg?
>
> -- 
> John Baldwin
>

I upped to DELAY(2) in pnp_get_resource_info and left everything else as 
what it was originally, modem was detected properly.

Curiously the uart driver now reports a different port type:
uart5: <16550 or compatible> at port 0x3e8-0x3ef irq 5 pnpid USR0011 on 
isa0
uart5: [FILTER]
uart5: fast interrupt

I'm still testing the modem, I'm only using a very limit set of its 
functionality. It doesn't function like it did in 4.x, so I'm having to 
access it differently, but its similar behaviour.

---
Steven Nikkel
steven_nikkel at ertyu.org