1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
|
Cryptsetup 2.3.5 Release Notes
==============================
Stable bug-fix release with minor extensions.
All users of cryptsetup 2.x and later should upgrade to this version.
Changes since version 2.3.4
~~~~~~~~~~~~~~~~~~~~~~~~~~~
* Fix partial reads of passphrase from an interactive terminal.
Some stable kernels (5.3.11) started to return buffer from a terminal
in parts of maximal size 64 bytes.
This breaks the reading of passphrases longer than 64 characters
entered through an interactive terminal. The change is already fixed
in later kernel releases, but tools now support such partial read from
terminal properly.
* Fix maximal length of password entered through a terminal.
Now the maximal interactive passphrase length is exactly
512 characters (not 511).
* integritysetup: support new dm-integrity HMAC recalculation options.
In older kernels (since version 4.19), an attacker can force
an automatic recalculation of integrity tags by modifying
the dm-integrity superblock.
This is a problem with a keyed algorithms (HMAC), where it expects
nobody can trigger such recalculation without the key.
(Automatic recalculation will start after the next activation.)
Note that dm-integrity in standalone mode was *not* supposed
to provide cryptographic data integrity protection.
Despite that, we try to keep the system secure if keyed algorithms
are used.
Thank Daniel Glöckner for the original report of this problem.
Authenticated encryption that provides data integrity protection (in
combination with dm-crypt and LUKS2) is not affected by this problem.
The fix in the kernel for this problem contains two parts.
Firstly, the dm-integrity kernel module disables integrity
recalculation if keyed algorithms (HMAC) are used.
This change is included in long-term stable kernels.
Secondly, since the kernel version 5.11, dm-integrity introduces
modified protection where a journal-integrity algorithm guards
superblock; also, journal sections are protected. An attacker cannot
copy sectors from one journal section to another, and the superblock
also contains salt to prevent header replacement from another device.
If you want to protect data with HMAC, you should always also use HMAC
for --journal-integrity. Keys can be independent.
If HMAC is used for data but not for the journal, the recalculation
option is disabled.
If you need to use (insecure) backward compatibility implementation,
two new integritysetup options are introduced:
- Use --integrity-legacy-recalc (instead of --integrity-recalc)
to allow recalculation on legacy devices.
- Use --integrity-legacy-hmac in format action to force old insecure
HMAC format.
Libcryptsetup API also introduces flags
CRYPT_COMPAT_LEGACY_INTEGRITY_HMAC and
CRYPT_COMPAT_LEGACY_INTEGRITY_RECALC
to set these through crypt_set_compatibility() call.
* integritysetup: display of recalculating sector in dump command.
* veritysetup: fix verity FEC if stored in the same image with hashes.
Optional FEC (Forward Error Correction) data should cover the whole
data area, hashes (Merkle tree), and optionally additional metadata
(located after hash area).
Unfortunately, if FEC data is stored in the same file as hash,
the calculation wrongly used the whole file size, thus overlaps with
the FEC area itself. This produced unusable and too large FEC data.
There is no problem if the FEC image is a separate image.
The problem is now fixed, introducing FEC blocks calculation as:
- If the hash device is in a separate image, metadata covers the
whole rest of the image after the hash area. (Unchanged behavior.)
- If hash and FEC device is in the image, metadata ends on the FEC
area offset.
Note: there is also a fix for FEC in the dm-verity kernel (on the way
to stable kernels) that fixes error correction with larger RS roots.
* veritysetup: run FEC repair check even if root hash fails.
Note: The userspace FEC verify command reports are only informational
for now. Code does not check verity hash after FEC recovery in
userspace. The Reed-Solomon decoder can then report the possibility
that it fixed data even if parity is too damaged.
This will be fixed in the next major release.
* veritysetup: do not process hash image if hash area is empty.
Sometimes the device is so small that there is only a root hash
needed, and the hash area is not used.
Also, the size of the hash image is not increased for hash block
alignment in this case.
* veritysetup: store verity hash algorithm in superblock in lowercase.
Otherwise, the kernel could refuse the activation of the device.
* bitlk: fix a crash if the device disappears during BitLocker scan.
* bitlk: show a better error when trying to open an NTFS device.
Both BitLocker version 1 and NTFS have the same signature.
If a user opens an NTFS device without BitLocker, it now correctly
informs that it is not a BITLK device.
* bitlk: add support for startup key protected VMKs.
The startup key can be provided in --key-file option for open command.
* Fix LUKS1 repair code (regression since version 1.7.x).
We cannot trust possibly broken keyslots metadata in repair, so the
code recalculates them instead.
This makes the repair code working again when the master boot record
signature overwrites the LUKS header.
* Fix luksKeyChange for LUKS2 with assigned tokens.
The token references are now correctly assigned to the new keyslot
number.
* Fix cryptsetup resize using LUKS2 tokens.
Code needlessly asked for passphrase even though volume key was
already unlocked via LUKS2 token.
* Print a visible error if device resize is not supported.
* Add error message when suspending wrong non-LUKS device.
* Fix default XTS mode key size in reencryption.
The same luksFormat logic (double key size because XTS uses two keys)
is applied in the reencryption code.
* Rephrase missing locking directory warning and move it to debug level.
The system should later provide a safe transition to tempdir
configuration, so creating locking directory inside libcryptsetup
call is safe.
* Many fixes for the use of cipher_null (empty debug cipher).
Support for this empty cipher was intended as a debug feature and for
measuring performance overhead. Unfortunately, many systems started to
use it as an "empty shell" for LUKS (to enable encryption later).
This use is very dangerous and it creates a false sense of security.
Anyway, to not break such systems, we try to support these
configurations.
Using cipher_null in any production system is strongly discouraged!
Fixes include:
- allow LUKS resume for a device with cipher_null.
- do not upload key in keyring when data cipher is null.
- switch to default cipher when reencrypting cipher_null device.
- replace possible bogus cipher_null keyslots before reencryption.
- fix broken detection of null cipher in LUKS2.
cipher_null is no longer possible to be used in keyslot encryption
in LUKS2, it can be used only for data for debugging purposes.
* Fixes for libpasswdqc 2.0.x (optional passphrase quality check).
* Fixes for problems discovered by various tools for code analysis.
Fixes include a rework of libpopt command line option string leaks.
* Various fixes to man pages.
|
