On l=F8rdag, mar 15, 2003, at 02:52 Europe/Copenhagen, Henry B. Hotz=20
wrote:

> At 10:53 AM +0100 3/14/03, Daniel Eggert wrote:
>> On torsdag, mar 13, 2003, at 10:49 Europe/Copenhagen, Daniel Eggert=20=

>> wrote:
>>
>>>
>>> On torsdag, mar 13, 2003, at 02:16 Europe/Copenhagen, John Gordon=20
>>> wrote:
>>>> The problem is that sometimes when a lot of data is passing across=20=

>>>> the serial
>>>> link (e.g. scrolling down a page in an editor) the link will hang.=20=

>>>> The system
>>>> seems to be still alive (I can ping it). I will set up ssh support=20=

>>>> on it next
>>>> so I can take a look at what is happening, but thought I'd ask if=20=

>>>> anybody has
>>>> seen anything like this before?
>>>>
>>>
>>> This is a know problem. Try using ssh instead if you can. Hopefully=20=

>>> some day someone will fix this. It's a bug in the driver for the=20
>>> serial port.
>>>
>>
>> It is in the PR-database as #10540.
>>
>> Is anybody working on a fix? Does anyone have a clue, what could be=20=

>> wrong? ..or how to investigate this? I guess the bug is somewhere=20
>> inside src/sys/arch/macppc/dev/zs.c and I wouldn't mind helping...
>>
>> Any comments?
>
> I think the last person to look at this seriously and comment on it=20
> was John Klos.  He came to the conclusion that there was something=20
> whacked in the MI infrastructure w.r.t. how interrupts were traded=20
> among parts of the system.

This sounds like a very good explanation. It would certainly explain=20
why the serial driver hangs during bursts and not otherwise.

/Daniel=