On Sat, 1 Jan 2005 18:25:42 -0500 (EST)
"Greg A. Woods" <woods@weird.com> wrote:

> I would blame the hardware for incorrectly sensing a BREAK _long_
> before I would ever even dream of blaming the driver for acting on
> one.

These systems run unattended all the time without consoles connected, and this one has been used with 1.6 NetBSD alpha for the last two years or so. In that time I have started it many times without a console cable.

Between running 1.6 and 2.0 I installed a new secondary disk (sd1 in this configuration) and installed 2.0 on the disk which originally had 1.6. These were the only changes I made.

One thing which I will try is to install 2.0 on my other dec3000 system. This should resolve any hardware/software issue.

> If there's nothing connected to the port then it would be well within
> the realm of possibility for the UART to sense what appears to be a
> BREAK signal from the noise on the line.

I suppose so. But it ran fine with 1.6, and it should be possible to start this system without a console connected. Please note that the break on startup is consistent. It always happens at the same point.

I would still like to have a look at the relevant source code. I haven't dowloaded it yet so my ability to grep is limited ;). Can anybody suggest a starting point? I will have a look at the cvsweb interface on the web site.

Regards,
-- 
Michael Smith