141
Keyfiles
TrueCrypt keyfile is a file whose content is combined with a password. The user can use any kind
of file as a TrueCrypt keyfile. The user can also generate a keyfile using the built-in keyfile
generator, which utilizes the TrueCrypt RNG to generate a file with random content (for more
information, see the section
Random Number Generator
).
The maximum size of a keyfile is not limited; however, only its first 1,048,576 bytes (1 MB) are
processed (all remaining bytes are ignored due to performance issues connected with processing
extremely large files). The user can supply one or more keyfiles (the number of keyfiles is not
limited).
Keyfiles can be stored on PKCS-11-compliant [23] security tokens and
smart cards protected by
multiple PIN codes (which can be entered either using a hardware PIN pad or via the TrueCrypt
GUI).
Keyfiles are processed and applied to a password using the following method:
1.
Let
P
be a TrueCrypt volume password supplied by user (may be empty)
2.
Let
KP
be the keyfile pool
3.
Let
kpl
be the
size of the keyfile pool
KP
, in bytes (64, i.e., 512 bits);
kpl
must be a multiple of the output size of a hash function
H
4.
Let
pl
be the length of the password
P
, in bytes (in the current version: 0
≤
pl
≤
64)
5.
if
kpl > pl
, append (
kpl – pl
)
zero bytes to the password
P
(thus
pl = kpl
)
6.
Fill the keyfile pool
KP
with
kpl
zero bytes.
7.
For each keyfile perform the following steps:
a.
Set the position of the keyfile pool cursor to the beginning
of the pool
b.
Initialize the hash function
H
c.
Load all bytes of the keyfile one by one, and for each loaded byte perform the
following steps:
i.
Hash the loaded byte using the hash function
H
without
initializing the hash,
to obtain an intermediate hash (state)
M.
Do not finalize the hash (the state is
retained for next round).
ii.
Divide the state
M
into individual bytes.
For example, if the hash output size is 4 bytes, (
T
0
||
T
1
||
T
2
||
T
3
) =
M
iii.
Write these bytes (obtained in step 7.c.ii) individually to the keyfile pool with
the modulo 2
8
addition operation (not by replacing the old values in the pool)
at the position of the pool cursor.
After a byte is written, the pool cursor
position is advanced by one byte. When the cursor reaches the end of the
pool, its position is set to the beginning of the pool.
8.
Apply the content of the keyfile pool to the password
P
using the following method:
a.
Divide the password
P
into
individual bytes
B
0
...
B
pl
-1
.
Note that if the password was shorter than the keyfile pool, then the password was padded with zero
bytes to the length of the pool in Step 5 (hence, at this point the length of the password is always
greater than or equal to the length of the keyfile pool).
b.
Divide the keyfile pool
KP
into individual bytes
G
0
...
G
kpl
-1
c.
For 0
≤
i
<
kpl
perform:
B
i
=
B
i
⊕
G
i
d.
P
=
B
0
||
B
1
|| ... ||
B
pl
-2
||
B
pl
-1
142
9.
The password
P
(after the keyfile pool content has been applied to it) is now passed to the
header key derivation function PBKDF2 (PKCS #5 v2), which processes it (along with salt
and other data) using a cryptographically secure hash algorithm selected by the user (e.g.,
SHA-512). See
the section
Header Key Derivation, Salt, and Iteration Count
for more
information.
The role of the hash function
H
is merely to perform diffusion [2]. CRC-32 is used as the hash
function
H
. Note that the output of CRC-32 is subsequently processed using a cryptographically
secure hash algorithm: The keyfile pool content (in addition to being hashed using CRC-32) is
applied to the
password, which is then passed to the header key derivation function PBKDF2
(PKCS #5 v2), which processes it (along with salt and other data) using a cryptographically
secure hash algorithm selected by the user (e.g., SHA-512). The resultant values are used to
form the header key and the secondary header key (XTS mode).