[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: D-I and busybox 1.9.1 test results

On Saturday 08 March 2008, Bastian Blank wrote:
> Hu? Neither config.deb nor config.udeb are empty in the repo nor the
> source package.

Hmm. Weird.
If I use cat on the files, I get the content. But if I use less, I only get:
$ less config.deb

*** Contents:

So that's why I thought they were empty. As I said, weird :-)

> > 3) the installer thinks it is running on serial console instead of a
> > virtual terminal
> >
> > This seems to be the most serious/structural issue. The problem is in
> > /lib/d-i/detect-console (called by /lib/d-i-startup/S35term). This does
> > a 'readlink /proc/self/fd/0) which used to return /dev/tty1, but now
> > returns /dev/console, which makes detect-console conclude we are on
> > serial console.
> The other possibility is to remove this code and use own console
> selection for the installer.


> > I did not find any config option to solve this (which does not mean it
> > does not exist.

That is enabled. The description for that is:
  │ If this option is enabled a command starting with hyphen (-)
  │ is run in its own session (setsid(2)) and possibly with a
  │ controlling tty (TIOCSCTTY).  This is not the traditional init
  │ behavior, but is often what you want in an embedded system where
  │ the console is only accessed during development or for maintenance.
  │ NB: using cttyhack applet may work better.

So, maybe it would work too if we prefixed the commands with a hyphen.
In both cases we have a change that will need extensive testing on multiple 
platforms. And looking at that last line, it seems to me that cttyhack 
might well do the right thing in more cases.

> > Some highlights from the diff between the configs for the udeb:
> > New options. Do we want to save some space by disabling this?
> Needs testing.

Yeah. These settings do seem to work and do give useful info. There's also 
an option for verbose usage info, but that would be too much.
Disabling this should not gain very much, but I guess it could be worth 

Attachment: signature.asc
Description: This is a digitally signed message part.

Reply to: