Bug#519314: Regression: no longer adds my SSH keys to the agent

2009-03-19 Thread Josh Triplett
Package: libpam-ssh
Followup-For: Bug #519314

I very much like the idea that pam-ssh has added a mechanism to allow
users to configure additional SSH keys to unlock on login without
requiring system-wide configuration via the keyfiles option.  The
login-keys.d directory seems like a sensible way to do this.

However, this configurability should not come at the expense of
working out-of-the-box without per-user configuration.  Unlike previous
versions, this new version of libpam-ssh no longer adds my keys to the
agent by default.  This represents a regression from previous versions.

I read the debian-devel thread in question, and the only complaint I saw
about using id_* by default related to taking advantage of a key with a
weak or well-known passphrase to log in as the user.  This only seems
like a problem when configuring pam-ssh as a login mechanism, rather
than only a mechanism to run an ssh-agent and unlock keys as many people
do.  Thus, I have a proposal which I think will address these concerns
while still working out-of-the-box:

- Always attempt to unlock all keys matching id_* and add them to the
  ssh-agent.  Alternatively, at least unlock keys matching id_rsa,
  id_dsa, and identity, as SSH does and the previous version of pam-ssh
  did.  Since you can only unlock keys this way if you know their
  passphrase, this shouldn't represent a security problem.
- For authentication purposes, only treat keys linked from login-keys.d
  as sufficient for authentication.  This ensures that people don't
  inadvertently allow authentication via an unexpected key.

Thus, if I have no login-keys.d directory, pam-ssh will not allow me to
use my SSH keys to log in, but it will still unlock any keys that my
normal password unlocks and add them to ssh-agent.

Now, this still potentially represents a regression for people who use
pam-ssh for authentication.  However, it seems more reasonable to allow
a regression there to prevent a potential security problem.  I don't use
pam-ssh for authentication, so I can't speak for the people who do; the
behavior I described above will at least fix the regression for people
who use pam-ssh just to unlock their keys.

- Josh Triplett



-- 
To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org



Bug#519314: Regression: no longer adds my SSH keys to the agent

2009-03-12 Thread Jens Peter Secher
I am truly sorry that you are being bothered by the changes in the
new Debian version of libpam-ssh.

The story is that the new version of libpam-ssh will look for keys in
$HOME/.ssh/login-keys.d/, and only there.  I implemented that
functionality after a discussion on the debian-devel mailing list [1]
and an announcement on pam mailing list (with no response).  Since then,
the new version of libpam-ssh has been available in experimental, and
because I have had no complaints about it so far, I have uploaded it to
unstable.

Personally I think the new setup is better because it relieves the
administrator from doing any special setup for users, while at the same
time enabling individual users to use as many keys as they like (cf
[3]).  But maybe libpam-ssh should look for the same set as key as ssh
does in addition to $HOME/.ssh/login-keys.d/ ?

[1] http://lists.debian.org/debian-devel/2008/12/msg00193.html
[2] https://www.redhat.com/archives/pam-list/2008-December/msg4.html
[3] http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=477272

Cheers,
-- 
Jens Peter Secher.
_DD6A 05B0 174E BFB2 D4D9 B52E 0EE5 978A FE63 E8A1 jpsecher gmail com_.
A. Because it breaks the logical sequence of discussion.
Q. Why is top posting bad?




-- 
To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org



Bug#519314: Regression: no longer adds my SSH keys to the agent

2009-03-11 Thread Josh Triplett
Package: libpam-ssh
Version: 1.92-5
Severity: normal

After upgrading from libpam-ssh 1.91.0-9.3 to libpam-ssh 1.92-5,
libpam-ssh no longer uses my password to add my SSH keys to the
ssh-agent it starts.

The changelog for libpam-ssh suggested that it would now unlock any key
matching id_*, which my keys do.

Another entry in the changelog suggested that libpam-ssh now looks for
keys in $HOME/.ssh/login-keys.d/ , but I certainly hope it doesn't
*require* keys to appear in that directory.  That would represent a
regression from previous versions, which found my keys without that
extra configuration.

- Josh Triplett

-- System Information:
Debian Release: squeeze/sid
  APT prefers unstable
  APT policy: (500, 'unstable'), (1, 'experimental')
Architecture: amd64 (x86_64)

Kernel: Linux 2.6.28-1-amd64 (SMP w/2 CPU cores)
Locale: LANG=en_US.UTF-8, LC_CTYPE=en_US.UTF-8 (charmap=UTF-8)
Shell: /bin/sh linked to /bin/bash

Versions of packages libpam-ssh depends on:
ii  libc6 2.9-4  GNU C Library: Shared libraries
ii  libpam0g  1.0.1-7Pluggable Authentication Modules l
ii  libssl0.9.8   0.9.8g-15  SSL shared libraries

Versions of packages libpam-ssh recommends:
ii  openssh-client [ssh-client]   1:5.1p1-5  secure shell client, an rlogin/rsh

libpam-ssh suggests no packages.

-- no debconf information



-- 
To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org