Commits on Source (33)
-
Daniel Kahn Gillmor authoredac5a4d1f
-
Daniel Kahn Gillmor authored815038b4
-
On systems with restricted PAM security, it may not possible to use su.
4c8864ea -
Daniel Kahn Gillmor authored
It should be more portable to rely on the $PATH to env than to invoke it explicitly as /usr/bin/env.
31eff2a5 -
In a couple of cases, monkeysphere commands running as run create a temporary directory in TMPDIR (provided by environment) and then change the ownership/permissions on that directory for monkeysphere user to use that directory. This works in a normal setup but fails when libpam-tmpdir is installed. This PAM module causes the tmp directory to be /tmp/user/0/ so that it is harder to for users to access each other temporary files. This improves security but causes problem for above situation as the parent directory of the directory to be shared is not allowed access by other users. To fix this, explicitly set the TMPDIR to a known location that can be used to share files across users. /var/lib/monkeysphere/authentication/tmp is a directory that is already being setup and used for such purposes. Reuse it instead of created a new one. Apply the fix conservatively only in cases needed. Closes: #656750. Signed-off-by: Sunil Mohan Adapa <sunil@medhas.org>
82711069 -
Daniel Kahn Gillmor authored
In some very unusual cases, where the system monkeysphere directory has whitespace in the paths, or where some filenames start with a "-", the shell, mktemp, or various components of coreutils could misinterpret the commands given. This cleans up some of those mistakes, many of which have been in the monkeysphere codebase for ages. Signed-off-by: Daniel Kahn Gillmor <dkg@fifthhorseman.net>
b6b3c529 -
Daniel Kahn Gillmor authored
There is no good reason to pass -q to diff in the test suite. If there's a difference, the test suite will fail, and we want to know what the difference is.
48f5a134 -
Daniel Kahn Gillmor authoredc48ffcd6
-
Daniel Kahn Gillmor authored
GnuPG 2.1.11 introduced gpg --export-ssh-key. By relying on it explicitly, we can reduce our dependency on keytrans, no longer needing it for getting SSH-style public key information out of OpenPGP certificates. This increases our versioned dependency on GnuPG, and introduces a requirement to have OpenSSH's ssh-keygen around for generating ssh fingerprints. We aren't yet handling secret transformations, though.
3e2b94b3 -
Daniel Kahn Gillmor authored
Since we're using a modern version of GnuPG, we can remove some of the fancier versioned checks. If we're willing to depend on gpg >= 2.1.17, we can probably move even more, since that's when the --quick-* interface really stabilized.
c4c2e888 -
Daniel Kahn Gillmor authored
agent-transfer used to only work for RSA keys. With this changeset, it grows support for ed25519 keys, so "monkeysphere subkey-to-ssh-agent" should be able to work for ed25519 authentication-capable subkeys as well.
af4bfcd3 -
Daniel Kahn Gillmor authored2a30561c
-
Daniel Kahn Gillmor authored32f36299
-
Daniel Kahn Gillmor authored
Without this, sometimes the build process is interrupted with a prompt: gzip: replaced/man/man1/agent-transfer.1.gz already exists; do you wish to overwrite (y or n)?
2b298cad -
Daniel Kahn Gillmor authored
There are only a few places where monkeysphere is actively creating keys. By default now, use 3072-bit RSA (and be explicit that it's RSA, in preparation for being able to move to a different algorithm)
c2ad1254 -
Daniel Kahn Gillmor authored85f5513d
-
Daniel Kahn Gillmor authored
This new test will only work for GnuPG version 2.1.17 and later, when --quick-add-key was added. It's intended to pave the way for a future version of monkeysphere that has a tighter GnuPG dependency and can drop the dependency on perl entirely.
63f9eb31 -
Daniel Kahn Gillmor authored8a887e38
-
Daniel Kahn Gillmor authored6ed85ae6
-
Daniel Kahn Gillmor authored84e75d72
-
Daniel Kahn Gillmor authored
We were using 1024-bit RSA keys in tests to avoid blocking on systems with limited entropy. As we move into the getrandom() era, this is a situation that we should expect the operating system to handle correctly, and our tools should be tested with sensible limits and sensible key sizes.
4cf70afc -
Daniel Kahn Gillmor authored
This should resolve https://bugs.debian.org/906755
1453f99b -
Daniel Kahn Gillmor authored5f2e56e6
-
Daniel Kahn Gillmor authored
See the argument here: https://bugs.debian.org/897366 I discussed this on IRC with anarcat, and did some code archaeology, and could not find any reason for the excessive paranoia. If the chown/chmod/mv sequence fails, then we should report an error and remove the temporary incoming file, not the pre-existing one.
bc443de6 -
Daniel Kahn Gillmor authored
Since OpenSSH 6.0, the sshd server has supported sshd -T to dump the config without needing to find the file. Encourage the use of a modern version of sshd, and improve the diagnostics as a result.
b1608b91 -
Daniel Kahn Gillmor authored16361f60
-
Daniel Kahn Gillmor authored56ce9502
-
Daniel Kahn Gillmor authored43629c46
-
Daniel Kahn Gillmor authored27a86a05
-
Daniel Kahn Gillmor authoreda8f4ac40
-
Daniel Kahn Gillmor authored
See https://bugs.debian.org/920038 for more information about why this additional debugging information might be useful.
318134aa -
Daniel Kahn Gillmor authored
Since monkeysphere-authentication typically relies on runuser, and ISC cron typically sets only PATH=/usr/bin:/bin So we need to include the expected superuser's $PATH when crontab entry.
d558bf0a -
Daniel Kahn Gillmor authored
Since util-linux 2.31, runuser has has support (sometimes undocumented) for defense against terminal hijacking via TIOCSTI or other techniques, by passing a --pty argument. That option is [no longer experimental](https://github.com/karelzak/util-linux/issues/760 ), so monkeysphere-authentication should use it if it is available. This defends against a compromised monkeysphere user being able to attack the superuser when invoked under a controlling terminal. Signed-off-by: Daniel Kahn Gillmor <dkg@fifthhorseman.net>
97ade311