Re: [Pkg-shadow-devel] Bug#476519: shell-mode garbles input

[Stephane Chazelas]

On Sun, May 04, 2008 at 10:14:26AM +0800, jidanni@jidanni.org wrote:
[...]
> SC> Have you tried changing nobody's shell?
> I don't want to.

Then, can you try:

su - nobody -c ksh
su - nobody -c pdksh
su - nobody -c zsh

Also, what about:

perl -e '$<=$>=$(=$)=65534; exec sh'
perl -e '$<=$>=$(=$)=65534; system sh'

> CR> Does su do anything with process groups?
> That is an advanced question not for little me.

Not as far as the strace output can tell.

> Anyway, the only other human to have reproduced any of this mess is
> http://bugs.debian.org/476519#52
> Nobody :-) has responded to the points of that message #52.

The poster seems to say that what makes a difference is what
*root* does before calling su.

That would be /etc/bash_completion? Not sure how what
bash_completion does could have an impact on su.

You've already ruled out the environment by running "env -i",
you could check the open files with lsof -p "$$", you've
already checked stty -a. You could try running ps to see if it
left some processes running behind.

I see "mesg n" in Sven's post.

That command is meant to change the tty permissions to allow
writes to the owner only, in that case root, but I can't see how
that could affect nobody's sh.

For sh to get a SIGTTIN upon the "read", it would have not to be
in the foreground process group of the terminal.

bash does put itself in the foreground process group by a:
ioctl(255, TIOCSPGRP, getpid())
255 being a dup(dup(2)).

And it reads from 0.

So one way to get that behavior could be if stderr is
redirected to a different terminal as stdin... seems like a bit
far fetched.

running out of ideas here...

-- 
Stéphane