1 The dm-integrity target emulates a block device that has additional
2 per-sector tags that can be used for storing integrity information.
4 A general problem with storing integrity tags with every sector is that
5 writing the sector and the integrity tag must be atomic - i.e. in case of
6 crash, either both sector and integrity tag or none of them is written.
8 To guarantee write atomicity, the dm-integrity target uses journal, it
9 writes sector data and integrity tags into a journal, commits the journal
10 and then copies the data and integrity tags to their respective location.
12 The dm-integrity target can be used with the dm-crypt target - in this
13 situation the dm-crypt target creates the integrity data and passes them
14 to the dm-integrity target via bio_integrity_payload attached to the bio.
15 In this mode, the dm-crypt and dm-integrity targets provide authenticated
16 disk encryption - if the attacker modifies the encrypted device, an I/O
17 error is returned instead of random data.
19 The dm-integrity target can also be used as a standalone target, in this
20 mode it calculates and verifies the integrity tag internally. In this
21 mode, the dm-integrity target can be used to detect silent data
22 corruption on the disk or in the I/O path.
24 There's an alternate mode of operation where dm-integrity uses bitmap
25 instead of a journal. If a bit in the bitmap is 1, the corresponding
26 region's data and integrity tags are not synchronized - if the machine
27 crashes, the unsynchronized regions will be recalculated. The bitmap mode
28 is faster than the journal mode, because we don't have to write the data
29 twice, but it is also less reliable, because if data corruption happens
30 when the machine crashes, it may not be detected.
32 When loading the target for the first time, the kernel driver will format
33 the device. But it will only format the device if the superblock contains
34 zeroes. If the superblock is neither valid nor zeroed, the dm-integrity
35 target can't be loaded.
37 To use the target for the first time:
38 1. overwrite the superblock with zeroes
39 2. load the dm-integrity target with one-sector size, the kernel driver
40 will format the device
41 3. unload the dm-integrity target
42 4. read the "provided_data_sectors" value from the superblock
43 5. load the dm-integrity target with the the target size
44 "provided_data_sectors"
45 6. if you want to use dm-integrity with dm-crypt, load the dm-crypt target
46 with the size "provided_data_sectors"
51 1. the underlying block device
53 2. the number of reserved sector at the beginning of the device - the
54 dm-integrity won't read of write these sectors
56 3. the size of the integrity tag (if "-" is used, the size is taken from
57 the internal-hash algorithm)
60 D - direct writes (without journal) - in this mode, journaling is
61 not used and data sectors and integrity tags are written
62 separately. In case of crash, it is possible that the data
63 and integrity tag doesn't match.
64 J - journaled writes - data and integrity tags are written to the
65 journal and atomicity is guaranteed. In case of crash,
66 either both data and tag or none of them are written. The
67 journaled mode degrades write throughput twice because the
68 data have to be written twice.
69 B - bitmap mode - data and metadata are written without any
70 synchronization, the driver maintains a bitmap of dirty
71 regions where data and metadata don't match. This mode can
72 only be used with internal hash.
73 R - recovery mode - in this mode, journal is not replayed,
74 checksums are not checked and writes to the device are not
75 allowed. This mode is useful for data recovery if the
76 device cannot be activated in any of the other standard
79 5. the number of additional arguments
83 journal_sectors:number
84 The size of journal, this argument is used only if formatting the
85 device. If the device is already formatted, the value from the
88 interleave_sectors:number
89 The number of interleaved sectors. This values is rounded down to
90 a power of two. If the device is already formatted, the value from
91 the superblock is used.
94 Don't interleave the data and metadata on on device. Use a
95 separate device for metadata.
98 The number of sectors in one buffer. The value is rounded down to
101 The tag area is accessed using buffers, the buffer size is
102 configurable. The large buffer size means that the I/O size will
103 be larger, but there could be less I/Os issued.
105 journal_watermark:number
106 The journal watermark in percents. When the size of the journal
107 exceeds this watermark, the thread that flushes the journal will
111 Commit time in milliseconds. When this time passes, the journal is
112 written. The journal is also written immediatelly if the FLUSH
115 internal_hash:algorithm(:key) (the key is optional)
116 Use internal hash or crc.
117 When this argument is used, the dm-integrity target won't accept
118 integrity tags from the upper target, but it will automatically
119 generate and verify the integrity tags.
121 You can use a crc algorithm (such as crc32), then integrity target
122 will protect the data against accidental corruption.
123 You can also use a hmac algorithm (for example
124 "hmac(sha256):0123456789abcdef"), in this mode it will provide
125 cryptographic authentication of the data without encryption.
127 When this argument is not used, the integrity tags are accepted
128 from an upper layer target, such as dm-crypt. The upper layer
129 target should check the validity of the integrity tags.
132 Recalculate the integrity tags automatically. It is only valid
133 when using internal hash.
135 journal_crypt:algorithm(:key) (the key is optional)
136 Encrypt the journal using given algorithm to make sure that the
137 attacker can't read the journal. You can use a block cipher here
138 (such as "cbc(aes)") or a stream cipher (for example "chacha20",
139 "salsa20", "ctr(aes)" or "ecb(arc4)").
141 The journal contains history of last writes to the block device,
142 an attacker reading the journal could see the last sector nubmers
143 that were written. From the sector numbers, the attacker can infer
144 the size of files that were written. To protect against this
145 situation, you can encrypt the journal.
147 journal_mac:algorithm(:key) (the key is optional)
148 Protect sector numbers in the journal from accidental or malicious
149 modification. To protect against accidental modification, use a
150 crc algorithm, to protect against malicious modification, use a
151 hmac algorithm with a key.
153 This option is not needed when using internal-hash because in this
154 mode, the integrity of journal entries is checked when replaying
155 the journal. Thus, modified sector number would be detected at
159 The size of a data block in bytes. The larger the block size the
160 less overhead there is for per-block integrity metadata.
161 Supported values are 512, 1024, 2048 and 4096 bytes. If not
162 specified the default block size is 512 bytes.
164 sectors_per_bit:number
165 In the bitmap mode, this parameter specifies the number of
166 512-byte sectors that corresponds to one bitmap bit.
168 bitmap_flush_interval:number
169 The bitmap flush interval in milliseconds. The metadata buffers
170 are synchronized when this interval expires.
173 The journal mode (D/J), buffer_sectors, journal_watermark, commit_time can
174 be changed when reloading the target (load an inactive table and swap the
175 tables with suspend and resume). The other arguments should not be changed
176 when reloading the target because the layout of disk data depend on them
177 and the reloaded target would be non-functional.
180 The layout of the formatted block device:
181 * reserved sectors (they are not used by this target, they can be used for
182 storing LUKS metadata or for other purpose), the size of the reserved
183 area is specified in the target arguments
185 * magic string - identifies that the device was formatted
187 * log2(interleave sectors)
189 * the number of journal sections
190 * provided data sectors - the number of sectors that this target
191 provides (i.e. the size of the device minus the size of all
192 metadata and padding). The user of this target should not send
193 bios that access data beyond the "provided data sectors" limit.
195 SB_FLAG_HAVE_JOURNAL_MAC - a flag is set if journal_mac is used
196 SB_FLAG_RECALCULATING - recalculating is in progress
197 SB_FLAG_DIRTY_BITMAP - journal area contains the bitmap of dirty
199 * log2(sectors per block)
200 * a position where recalculating finished
202 The journal is divided into sections, each section contains:
203 * metadata area (4kiB), it contains journal entries
204 every journal entry contains:
205 * logical sector (specifies where the data and tag should
207 * last 8 bytes of data
208 * integrity tag (the size is specified in the superblock)
209 every metadata sector ends with
210 * mac (8-bytes), all the macs in 8 metadata sectors form a
211 64-byte value. It is used to store hmac of sector
212 numbers in the journal section, to protect against a
213 possibility that the attacker tampers with sector
214 numbers in the journal.
216 * data area (the size is variable; it depends on how many journal
217 entries fit into the metadata area)
218 every sector in the data area contains:
219 * data (504 bytes of data, the last 8 bytes are stored in
222 To test if the whole journal section was written correctly, every
223 512-byte sector of the journal ends with 8-byte commit id. If the
224 commit id matches on all sectors in a journal section, then it is
225 assumed that the section was written correctly. If the commit id
226 doesn't match, the section was written partially and it should not
228 * one or more runs of interleaved tags and data. Each run contains:
229 * tag area - it contains integrity tags. There is one tag for each
230 sector in the data area
231 * data area - it contains data sectors. The number of data sectors
232 in one run must be a power of two. log2 of this value is stored