This bug affects systems with a running X11 DISPLAY, but where only pinentry-ncurses is installed. It wrongly reads the input password: no matter what string is chosen, the password becomes:
tomb [W] Detected DISPLAY, but only pinentry-curses is found.
Following the fix in Tomb 2.8 affected users will need to use the line above as password to open their tomb and should change their key with a new password using 'tomb passwd'.
Tomb uses the cryptsetup LUKS volume header default to type luks1 which has been for long the default in cryptsetup. But starting from cryptsetup v2.1 a new default has been introduced (luks2) and the --type option added to specify the old luks1.
Using Tomb version 2.6 (and future releases) the problem opening tombs using recent GNU/Linux distributions is fixed.
Up until and including Tomb's version 2.5 the PBKDF2 wrapper for keys in Tomb has a bug affecting passwords that contain whitespace. Since the passwords are trimmed at the first whitespace, this makes them weaker, while fortunately the KDF transformation still applies.
This issue is fixed in Tomb version 2.6: all users adopting KDF keys that have passwords containing whitespace should change them, knowing that their "old password" is trimmed until the whitespace.
Users adopting GPG keys or plain (without KDF wrapper) can ignore this bug.
An important part of Tomb's security model is to make it hard for attackers to enter in possession of both key and data storage: once that happens, bruteforcing the password can be relatively easy.
Protection from bruteforcing is provided by the KDF module that can
be optionally compiled in extras/kdf-keys and installed.
If a key is buried in an image and then the image is stolen, the KDF protection does not works because attackers can bruteforce easily using steghide dictionary attacks: once found the password is the same for the steg crypto and the key crypto.
Users should keep in mind these issues when planning their encryption scheme and, when relying on steganography, keep the image always mixed in the same folder with many more images since that will be the multiplier making it slightly harder to bruteforce their password.
In most cases consider that password bruteforce is a feasible attack
vector on keys. If there are doubts about a key being compromised is
a good practice to change it using the setkey command on a secure
machine, possibly while off-line or in single user mode.
When used to forge new keys, Tomb version 2.2 incorrectly added a new line ('\n', 0x0A) character at the end of each key's secret sequence before encoding it with GnuPG. This does not affect Tomb regression and compatibility with other Tomb versions as this final newline is ignored in any case, but third party software may have problems. Those writing a software that supports opening Tomb files should always ignore the final newline when present in the secret material obtained after decoding the key with the password.
Due to distraction tomb version 1.5 displays its version as 1.4. Also version 1.5 did not work when using -k - to pipe keys from stdin, plus left the encrypted keys laying around in RAM (tmpfs). This was a minor vulnerability fixed in 1.5.1.
Due to an error in the creation and decoding of key files, release versions 1.3 and 1.3.1 cannot open older tombs, plus the tombs created with them will not be opened with older and newer versions of Tomb.
This bug was fixed in commit 551a7839f500a9ba4b26cd63774019d91615cb16
Those who have created tombs with older versions can simply upgrade to release 1.4 (and any other following release) to fix this issue and be able to operate their tombs normally.
Those who have used Tomb 1.3 or 1.3.1 to create new tombs should use Tomb version 1.3.1 (available from https://files.dyne.org/tomb) to open them and then migrate the contents into a new tomb created using the latest stable Tomb version.
This bug was due to a typo in the code which appended a GnuPG status string to the content of keys. All users of Tomb 1.3.* should pay particular attention to this issue, however that release series was out as latest for less than a month.