[lkml]   [1997]   [Jan]   [23]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
Messages in this thread
SubjectRe: Q: Serial Console Patch?
On 23 Jan, Scott J. Ellentuch wrote:
> In' own words (And I ">_") :
> >
> > Hi there,
> >
> > I just read the notes about the serial console patch ... and I've a few more
> > questions:
> >
> Well, I have a little I'd like to say on it too :
> I'm currently setting up remote servers at sites that I don't have easy
> access to nor intelligent life forms nearby. To make matters worse, I only
> get a small amount of space and am using a special rack mount computer, power
> cycling device and terminal server. I'm supposed to do alot of administration
> remotely.
> Making kernel changes is the scariest thing in the world for me. I'm
> afraid some day I'll make a bad mistake on a kernel that will prevent the
> system from booting. When this happens, I'll be in BAD shape. So, what I'm
> looking for is the following. Please forgive references to other UN*Xs and
> hardware.
> On IBM A*X systems, when it boots it broadcasts a message "Press <F1> on
> the console". If you press it on any device, the console or any serially
> attached device, the system changes that to the console. If nothing is
> pressed, it remains with serial0.

Lilo's 'serial=' option is not dependent or connected to the serial
console kernel patches in any way. Additionally, lilo performs IO on
both the serial port and the console, so nothing is lost.

Now, at the lilo prompt you can choose kernels or put something like
"console=xxxxx' to change the serial console to an other port or put
"console=" to disable the serial console features.

> Now, a friend of mine came up with another scenerio. Lets say LILO gets a
> "fallback" kernel and parameter name/list (This could be just another entry).
> When LILO runs, it checks to see if the last boot was ok. If so, it continues
> normal operation. If not, it does a "fallback" entry. After the system
> totally boots, we could then mark something to say "boot done". This
> again could save my skin.

This sounds unrealiable. Just get a lilo prompt and choose your kernel.

> What I don't want to see, and what I think the serial console patch is,
> is that the serial port is ALWAYS the console.

True, but as I said, lilo does not and with lilo you can specify a
different console.

> I'd be interested to hear from anyone else, especially ISP minded, that
> thinks this is something worth while. I've CC'd the author of LILO, and

I think that your needs have already been addressed. There are other
problems with serial console though. One is that it doesn't tell the
kernel what speed the console is running at. This makes a 38400 console
a little ugly but possible. The other is the kernel monotor that pops
up when a break is detected. I modify the patch to prevent this because
a port with no cable attached at the time will get spurious breaks
leaving the machine hung.

I use serial console for embedded Linux machines with no monitor and
keyboard, not for remote or ISP perposes.


Tyson D Sawyer <> RWI, Inc. has been supplying leading edge
Senior Systems Engineer mobile robotics technology since 1983
Real World Interface, Inc.

 \ /
  Last update: 2005-03-22 13:38    [W:0.089 / U:0.812 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site