blob: 71e5975728cc5dce84851ca6bb5c6c862565c708 [file] [log] [blame] [view]
Dave Rodgmana0e8db02021-06-29 18:05:38 +01001# Migrating from Mbed TLS 2.x to Mbed TLS 3.0
Manuel Pégourié-Gonnard89d4ab02021-04-23 11:54:27 +02002
3This guide details the steps required to migrate from Mbed TLS version 2.x to
4Mbed TLS version 3.0 or greater. Unlike normal releases, Mbed TLS 3.0 breaks
5compatibility with previous versions, so users (and alt implementors) might
6need to change their own code in order to make it work with Mbed TLS 3.0.
7
8Here's the list of breaking changes; each entry should help you answer these
9two questions: (1) am I affected? (2) if yes, what's my migration path?
10
Dave Rodgman1aea4042021-06-29 13:27:15 +010011The changes are detailed below, and include:
Dave Rodgman759c0102021-06-29 15:55:08 +010012
Dave Rodgman949c21b2021-06-29 18:05:04 +010013- Removal of many insecure or obsolete features
14- Tidying up of configuration options (including removing some less useful options).
Dave Rodgman70789732021-06-30 09:18:55 +010015- Changing function signatures, e.g. adding return codes, adding extra parameters, or making some arguments const.
Werner Lewis745fcde2022-06-23 12:19:27 +010016- Removal of functions, macros, and types previously marked as deprecated.
17
18
19## Referring to Mbed TLS 2.x documentation
20
21For features previously marked as deprecated, Mbed TLS 2.x documentation may
22explain how to upgrade, and should be referred to when migrating code. Where a
23migration path is not provided in prior documentation, changes made and the
24upgrade steps required will be explained in this guide.
25
26For some deprecated functions, 2.x documentation will suggest using a variant
27suffixed with `_ret`. In Mbed TLS 3.x, this change may not be required, as most
28of these variants have been renamed without the suffix. The section
29[Rename mbedtls_*_ret...](#rename-mbedtls__ret-cryptography-functions-whose-deprecated-variants-have-been-removed)
30has further detail on which functions this applies to.
Dave Rodgmand8a10172021-06-29 21:45:24 +010031
Dave Rodgmand8a10172021-06-29 21:45:24 +010032
Dave Rodgman4ea56432021-06-30 14:16:22 +010033## General changes
Dave Rodgmand8a10172021-06-29 21:45:24 +010034
Dave Rodgmana0e8db02021-06-29 18:05:38 +010035### Introduce a level of indirection and versioning in the config files
Manuel Pégourié-Gonnard89d4ab02021-04-23 11:54:27 +020036
Dave Rodgmane45e6402021-06-29 13:21:55 +010037`config.h` was split into `build_info.h` and `mbedtls_config.h`.
Manuel Pégourié-Gonnard89d4ab02021-04-23 11:54:27 +020038
Dave Rodgmane45e6402021-06-29 13:21:55 +010039* In code, use `#include <mbedtls/build_info.h>`. Don't include `mbedtls/config.h` and don't refer to `MBEDTLS_CONFIG_FILE`.
40* In build tools, edit `mbedtls_config.h`, or edit `MBEDTLS_CONFIG_FILE` as before.
41* If you had a tool that parsed the library version from `include/mbedtls/version.h`, this has moved to `include/mbedtls/build_info.h`. From C code, both headers now define the `MBEDTLS_VERSION_xxx` macros.
42
43Also, if you have a custom configuration file:
44
45* Don't include `check_config.h` or `config_psa.h` anymore.
46* Don't define `MBEDTLS_CONFIG_H` anymore.
47
48A config file version symbol, `MBEDTLS_CONFIG_VERSION` was introduced.
49Defining it to a particular value will ensure that Mbed TLS interprets
50the config file in a way that's compatible with the config file format
51used by the Mbed TLS release whose `MBEDTLS_VERSION_NUMBER` has the same
52value.
53The only value supported by Mbed TLS 3.0.0 is `0x03000000`.
54
Dave Rodgman70180532021-06-30 18:40:24 +010055### Most structure fields are now private
56
57Direct access to fields of structures (`struct` types) declared in public headers is no longer supported. In Mbed TLS 3, the layout of structures is not considered part of the stable API, and minor versions (3.1, 3.2, etc.) may add, remove, rename, reorder or change the type of structure fields.
58
59There is a small number of exceptions where some fields are guaranteed to remain stable throughout the lifetime of Mbed TLS 3.x. These fields are explicitly documented as public. Please note that even if all the fields of a structure are public, future versions may add new fields. Also, as before, some public fields should be considered read-only, since modifying them may make the structure inconsistent; check the documentation in each case.
60
61Attempting to access a private field directly will result in a compilation error.
62
63If you were accessing structure fields directly, and these fields are not documented as public, you need to change your code. If an accessor (getter/setter) function exists, use that. Direct accessor functions are usually called `mbedtls_<MODULE>_{get,set}_<FIELD>` or `mbedtls_<MODULE>_<STRUCTURE>_{get,set}_<FIELD>`. Accessor functions that change the format may use different verbs, for example `read`/`write` for functions that import/export data from/to a text or byte string.
64
Dave Rodgman017a1992022-03-31 14:07:01 +010065If no accessor function exists, please open an [enhancement request against Mbed TLS](https://github.com/Mbed-TLS/mbedtls/issues/new?template=feature_request.md) and describe your use case. The Mbed TLS development team is aware that some useful accessor functions are missing in the 3.0 release, and we expect to add them to the first minor release(s) (3.1, etc.).
Dave Rodgman70180532021-06-30 18:40:24 +010066
67As a last resort, you can access the field `foo` of a structure `bar` by writing `bar.MBEDTLS_PRIVATE(foo)`. Note that you do so at your own risk, since such code is likely to break in a future minor version of Mbed TLS.
68
Dave Rodgmana0e8db02021-06-29 18:05:38 +010069### Move part of timing module out of the library
Dave Rodgman8cccbe12021-06-29 13:15:50 +010070
71The change affects users who use any of the following functions:
72`mbedtls_timing_self_test()`, `mbedtls_hardclock_poll()`,
73`mbedtls_timing_hardclock()` and `mbedtls_set_alarm()`.
74
75If you were relying on these functions, you'll now need to change to using your
76platform's corresponding functions directly.
Dave Rodgman759c0102021-06-29 15:55:08 +010077
Dave Rodgmana0e8db02021-06-29 18:05:38 +010078### Deprecated net.h file was removed
Dave Rodgman8cccbe12021-06-29 13:15:50 +010079
80The file `include/mbedtls/net.h` was removed because its only function was to
81include `mbedtls/net_sockets.h` which now should be included directly.
Dave Rodgman759c0102021-06-29 15:55:08 +010082
Dave Rodgmand267ec32021-06-29 21:31:58 +010083### Remove `MBEDTLS_CHECK_PARAMS` option
Dave Rodgman8cccbe12021-06-29 13:15:50 +010084
85This change does not affect users who use the default configuration; it only
86affects users who enabled that option.
87
88The option `MBEDTLS_CHECK_PARAMS` (disabled by default) enabled certain kinds
89of “parameter validation”. It covered two kinds of validations:
90
91- In some functions that require a valid pointer, “parameter validation” checks
92that the pointer is non-null. With the feature disabled, a null pointer is not
93treated differently from any other invalid pointer, and typically leads to a
94runtime crash. 90% of the uses of the feature are of this kind.
95- In some functions that take an enum-like argument, “parameter validation”
96checks that the value is a valid one. With the feature disabled, an invalid
97value causes a silent default to one of the valid values.
98
99The default reaction to a failed check was to call a function
100`mbedtls_param_failed()` which the application had to provide. If this function
101returned, its caller returned an error `MBEDTLS_ERR_xxx_BAD_INPUT_DATA`.
102
103This feature was only used in some classic (non-PSA) cryptography modules. It was
104not used in X.509, TLS or in PSA crypto, and it was not implemented in all
105classic crypto modules.
106
107This feature has been removed. The library no longer checks for NULL pointers;
108checks for enum-like arguments will be kept or re-introduced on a case-by-case
109basis, but their presence will no longer be dependent on a compile-time option.
110
111Validation of enum-like values is somewhat useful, but not extremely important,
112because the parameters concerned are usually constants in applications.
113
114For more information see issue #4313.
Dave Rodgman759c0102021-06-29 15:55:08 +0100115
Dave Rodgman30dc6032021-06-29 22:20:58 +0100116### Remove the `MBEDTLS_TEST_NULL_ENTROPY` configuration option
Dave Rodgman8cccbe12021-06-29 13:15:50 +0100117
Dave Rodgman30dc6032021-06-29 22:20:58 +0100118This does not affect users who use the default `mbedtls_config.h`, as this option was
119already off by default.
120
121If you were using the `MBEDTLS_TEST_NULL_ENTROPY` option and your platform
122doesn't have any entropy source, you should use `MBEDTLS_ENTROPY_NV_SEED`
123and make sure your device is provisioned with a strong random seed.
124Alternatively, for testing purposes only, you can create and register a fake
125entropy function.
126
Dave Rodgman30dc6032021-06-29 22:20:58 +0100127### Remove the HAVEGE module
128
129This doesn't affect people using the default configuration as it was already
130disabled by default.
131
132This only affects users who called the HAVEGE modules directly (not
133recommended), or users who used it through the entropy module but had it as the
134only source of entropy. If you're in that case, please declare OS or hardware
135RNG interfaces with `mbedtls_entropy_add_source()` and/or use an entropy seed
136file created securely during device provisioning. See
137<https://tls.mbed.org/kb/how-to/add-entropy-sources-to-entropy-pool> for more
138information.
139
Dave Rodgmana3758202021-06-30 14:17:03 +0100140### Remove helpers for the transition from Mbed TLS 1.3 to Mbed TLS 2.0
141
142This only affects people who've been using Mbed TLS since before version 2.0
143and still relied on `compat-1.3.h` in their code.
144
145Please use the new names directly in your code; `scripts/rename.pl` (from any
146of the 2.x releases — no longer included in 3.0) might help you do that.
147
Dave Rodgman30dc6032021-06-29 22:20:58 +0100148
149## Low-level crypto
150
Dave Rodgman92170cc2021-06-30 14:23:27 +0100151Please also refer to the section [High-level crypto](#high-level-crypto) for
Dave Rodgman26ad6c72021-06-30 17:14:01 +0100152changes that could sit in either category.
Dave Rodgman92170cc2021-06-30 14:23:27 +0100153
Dave Rodgman30dc6032021-06-29 22:20:58 +0100154### Deprecated functions were removed from bignum
155
156The function `mbedtls_mpi_is_prime()` was removed. Please use
157`mbedtls_mpi_is_prime_ext()` instead which additionally allows specifying the
158number of Miller-Rabin rounds.
159
160### Deprecated functions were removed from DRBGs
161
Dave Rodgman2b034572021-06-30 18:59:49 +0100162The functions `mbedtls_ctr_drbg_update_ret()` and `mbedtls_hmac_drbg_update_ret()`
163were renamed to replace the corresponding functions without `_ret` appended. Please call
164the name without `_ret` appended and check the return value.
Dave Rodgman30dc6032021-06-29 22:20:58 +0100165
166### Deprecated hex-encoded primes were removed from DHM
167
168The macros `MBEDTLS_DHM_RFC5114_MODP_2048_P`, `MBEDTLS_DHM_RFC5114_MODP_2048_G`,
169`MBEDTLS_DHM_RFC3526_MODP_2048_P`, `MBEDTLS_DHM_RFC3526_MODP_2048_G`,
170`MBEDTLS_DHM_RFC3526_MODP_3072_P`, `MBEDTLS_DHM_RFC3526_MODP_3072_G`,
171`MBEDTLS_DHM_RFC3526_MODP_4096_P `and `MBEDTLS_DHM_RFC3526_MODP_4096_G` were
172removed. The primes from RFC 5114 are deprecated because their derivation is not
173documented and therefore their usage constitutes a security risk; they are fully
174removed from the library. Please use parameters from RFC3526 (still in the
175library, only in binary form) or RFC 7919 (also available in the library) or
176other trusted sources instead.
177
Dave Rodgman2d05e0f2021-06-30 18:42:34 +0100178### Deprecated functions were removed from hashing modules
179
180Modules: MD5, SHA1, SHA256, SHA512, MD.
181
182- The functions `mbedtls_xxx_starts_ret()`, `mbedtls_xxx_update_ret()`,
183 `mbedtls_xxx_finish_ret()` and `mbedtls_xxx_ret()` were renamed to replace
184 the corresponding functions without `_ret` appended. Please call the name without `_ret` appended and check the return value.
185- The function `mbedtls_md_init_ctx()` was removed; please use
186 `mbedtls_md_setup()` instead.
187- The functions `mbedtls_xxx_process()` were removed. You normally don't need
188 to call that from application code. However if you do (or if you want to
189 provide your own version of that function), please use
190 `mbedtls_internal_xxx_process()` instead, and check the return value.
191
Dave Rodgmana3758202021-06-30 14:17:03 +0100192### Change `MBEDTLS_ECP_FIXED_POINT_OPTIM` behavior
193
194The option `MBEDTLS_ECP_FIXED_POINT_OPTIM` now increases code size and it does
195not increase peak RAM usage anymore.
196
197If you are limited by code size, you can define `MBEDTLS_ECP_FIXED_POINT_OPTIM`
198to `0` in your config file. The impact depends on the number and size of
199enabled curves. For example, for P-256 the difference is 1KB; see the documentation
200of this option for details.
201
202### Separated `MBEDTLS_SHA224_C` and `MBEDTLS_SHA256_C`
203
Dave Rodgmana0148312021-06-30 19:08:51 +0100204This does not affect users who use the default `mbedtls_config.h`. `MBEDTLS_SHA256_C`
205was enabled by default. Now both `MBEDTLS_SHA256_C` and `MBEDTLS_SHA224_C` are
Dave Rodgmana3758202021-06-30 14:17:03 +0100206enabled.
207
Dave Rodgmana0148312021-06-30 19:08:51 +0100208If you were using custom config file with `MBEDTLS_SHA256_C` enabled, then
Dave Rodgmana3758202021-06-30 14:17:03 +0100209you will need to add `#define MBEDTLS_SHA224_C` option to your config.
Dave Rodgmana0148312021-06-30 19:08:51 +0100210Current version of the library does not support enabling `MBEDTLS_SHA256_C`
211without `MBEDTLS_SHA224_C`.
Dave Rodgmana3758202021-06-30 14:17:03 +0100212
213### Replaced `MBEDTLS_SHA512_NO_SHA384` with `MBEDTLS_SHA384_C`
214
215This does not affect users who use the default `mbedtls_config.h`.
Dave Rodgmana0148312021-06-30 19:08:51 +0100216`MBEDTLS_SHA512_NO_SHA384` was disabled by default, now `MBEDTLS_SHA384_C` is
Dave Rodgmana3758202021-06-30 14:17:03 +0100217enabled by default.
218
Dave Rodgmana0148312021-06-30 19:08:51 +0100219If you were using a config file with both `MBEDTLS_SHA512_C` and
220MBEDTLS_SHA512_NO_SHA384, then just remove the `MBEDTLS_SHA512_NO_SHA384`.
221If you were using a config file with `MBEDTLS_SHA512_C` and without
222`MBEDTLS_SHA512_NO_SHA384` and you need the SHA-384 algorithm, then add
Dave Rodgmana3758202021-06-30 14:17:03 +0100223`#define MBEDTLS_SHA384_C` to your config file.
224
Dave Rodgman68547182021-06-30 18:53:09 +0100225### GCM multipart interface: application changes
226
227The GCM module now supports arbitrary chunked input in the multipart interface.
228This changes the interface for applications using the GCM module directly for multipart operations.
229Applications using one-shot GCM or using GCM via the `mbedtls_cipher_xxx` or `psa_aead_xxx` interfaces do not require any changes.
230
231* `mbedtls_gcm_starts()` now only sets the mode and the nonce (IV). Call the new function `mbedtls_gcm_update_ad()` to pass the associated data.
232* `mbedtls_gcm_update()` now takes an extra parameter to indicate the actual output length. In Mbed TLS 2.x, applications had to pass inputs consisting of whole 16-byte blocks except for the last block (this limitation has been lifted). In this case:
233 * As long as the input remains block-aligned, the output length is exactly the input length, as before.
234 * If the length of the last input is not a multiple of 16, alternative implementations may return the last partial block in the call to `mbedtls_gcm_finish()` instead of returning it in the last call to `mbedtls_gcm_update()`.
235* `mbedtls_gcm_finish()` now takes an extra output buffer for the last partial block. This is needed for alternative implementations that can only process a whole block at a time.
236
237### GCM interface changes: impact for alternative implementations
238
Dave Rodgman24826502021-06-30 19:00:48 +0100239The GCM multipart interface has changed as described in [“GCM multipart interface: application changes”](#gcm-multipart-interface-application-changes). The consequences for an alternative implementation of GCM (`MBEDTLS_GCM_ALT`) are as follows:
Dave Rodgman68547182021-06-30 18:53:09 +0100240
241* `mbedtls_gcm_starts()` now only sets the mode and the nonce (IV). The new function `mbedtls_gcm_update_ad()` receives the associated data. It may be called multiple times.
242* `mbedtls_gcm_update()` now allows arbitrary-length inputs, takes an extra parameter to indicate the actual output length. Alternative implementations may choose between two modes:
243 * Always return the partial output immediately, even if it does not consist of a whole number of blocks.
244 * Buffer the data for the last partial block, to be returned in the next call to `mbedtls_gcm_update()` or `mbedtls_gcm_finish()`.
245* `mbedtls_gcm_finish()` now takes an extra output buffer for the last partial block if needed.
246
Dave Rodgmana3758202021-06-30 14:17:03 +0100247### The configuration option `MBEDTLS_ECP_NO_INTERNAL_RNG` was removed
248
249This doesn't affect users of the default configuration; it only affects people
250who were explicitly setting this option.
251
252This was a trade-off between code size and countermeasures; it is no longer
253relevant as the countermeasure is now always on at no cost in code size.
254
255### SHA-512 and SHA-256 output type change
256
Dave Rodgman8e5020d2021-07-02 12:16:03 +0100257The output parameter of `mbedtls_sha256_finish()`, `mbedtls_sha256()`, `mbedtls_sha512_finish()`, `mbedtls_sha512()` now has a pointer type rather than array type. This makes no difference in terms of C semantics, but removes spurious warnings in some compilers when outputting a SHA-384 hash into a 48-byte buffer or a SHA-224 hash into a 28-byte buffer.
Dave Rodgmana3758202021-06-30 14:17:03 +0100258
259This makes no difference to a vast majority of applications. If your code takes a pointer to one of these functions, you may need to change the type of the pointer.
260
261Alternative implementations of the SHA256 and SHA512 modules must adjust their functions' prototype accordingly.
262
Dave Rodgman30dc6032021-06-29 22:20:58 +0100263### Deprecated error codes for hardware failures were removed
264
Werner Lewis016cec12022-06-23 12:33:35 +0100265- The macros `MBEDTLS_ERR_xxx_FEATURE_UNAVAILABLE` from various crypto modules
Dave Rodgman30dc6032021-06-29 22:20:58 +0100266 were removed; `MBEDTLS_ERR_PLATFORM_FEATURE_UNSUPPORTED` is now used
267 instead.
Werner Lewis016cec12022-06-23 12:33:35 +0100268- The macro `MBEDTLS_ERR_RSA_UNSUPPORTED_OPERATION` was removed;
269 `MBEDTLS_ERR_PLATFORM_FEATURE_UNSUPPORTED` is now used instead.
Dave Rodgman30dc6032021-06-29 22:20:58 +0100270- The macros `MBEDTLS_ERR_xxx_HW_ACCEL_FAILED` from various crypto modules
271 were removed; `MBEDTLS_ERR_PLATFORM_HW_ACCEL_FAILED` is now used instead.
272
Werner Lewis016cec12022-06-23 12:33:35 +0100273### Deprecated error codes for invalid input data were removed
274
275- The macros `MBEDTLS_ERR_xxx_INVALID_KEY_LENGTH` from ARIA and Camellia
276 modules were removed; `MBEDTLS_ERR_xxx_BAD_INPUT_DATA` is now used instead.
277
Dave Rodgman507827e2021-06-30 18:54:35 +0100278### Remove the mode parameter from RSA functions
279
280This affects all users who use the RSA encryption, decryption, sign and
281verify APIs.
282
283The RSA module no longer supports private-key operations with the public key or
284vice versa. As a consequence, RSA operation functions no longer have a mode
285parameter. If you were calling RSA operations with the normal mode (public key
286for verification or encryption, private key for signature or decryption), remove
Werner Lewis016cec12022-06-23 12:33:35 +0100287the `MBEDTLS_RSA_PUBLIC` or `MBEDTLS_RSA_PRIVATE` argument. If you were calling
Dave Rodgman507827e2021-06-30 18:54:35 +0100288RSA operations with the wrong mode, which rarely makes sense from a security
289perspective, this is no longer supported.
290
Dave Rodgman71596682021-06-30 18:56:20 +0100291### Deprecated functions were removed from AES
292
293The functions `mbedtls_aes_encrypt()` and `mbedtls_aes_decrypt()` were
294removed.
295
296If you're simply using the AES module, you should be calling the higher-level
297functions `mbedtls_aes_crypt_xxx()`.
298
299If you're providing an alternative implementation using
300`MBEDTLS_AES_ENCRYPT_ALT` or `MBEDTLS_AES_DECRYPT_ALT`, you should be
301replacing the removed functions with `mbedtls_internal_aes_encrypt()` and
302`mbedtls_internal_aes_decrypt()` respectively.
303
Dave Rodgman8128b692021-06-30 18:56:33 +0100304### Deprecated functions were removed from ECDSA
305
306The functions `mbedtls_ecdsa_write_signature_det()` and
307`mbedtls_ecdsa_sign_det()` were removed. They were superseded by
308`mbedtls_ecdsa_write_signature()` and `mbedtls_ecdsa_sign_det_ext()`
309respectively.
310
Dave Rodgmanb4d15b12021-06-30 18:57:37 +0100311### Rename `mbedtls_*_ret()` cryptography functions whose deprecated variants have been removed
312
313This change affects users who were using the `mbedtls_*_ret()` cryptography
314functions.
315
316Those functions were created based on now-deprecated functions according to a
317requirement that a function needs to return a value. This change brings back the
318original names of those functions. The renamed functions are:
319
320| name before this change | after the change |
321|--------------------------------|----------------------------|
322| `mbedtls_ctr_drbg_update_ret` | `mbedtls_ctr_drbg_update` |
323| `mbedtls_hmac_drbg_update_ret` | `mbedtls_hmac_drbg_update` |
324| `mbedtls_md5_starts_ret` | `mbedtls_md5_starts` |
325| `mbedtls_md5_update_ret` | `mbedtls_md5_update` |
326| `mbedtls_md5_finish_ret` | `mbedtls_md5_finish` |
327| `mbedtls_md5_ret` | `mbedtls_md5` |
328| `mbedtls_ripemd160_starts_ret` | `mbedtls_ripemd160_starts` |
329| `mbedtls_ripemd160_update_ret` | `mbedtls_ripemd160_update` |
330| `mbedtls_ripemd160_finish_ret` | `mbedtls_ripemd160_finish` |
331| `mbedtls_ripemd160_ret` | `mbedtls_ripemd160` |
332| `mbedtls_sha1_starts_ret` | `mbedtls_sha1_starts` |
333| `mbedtls_sha1_update_ret` | `mbedtls_sha1_update` |
334| `mbedtls_sha1_finish_ret` | `mbedtls_sha1_finish` |
335| `mbedtls_sha1_ret` | `mbedtls_sha1` |
336| `mbedtls_sha256_starts_ret` | `mbedtls_sha256_starts` |
337| `mbedtls_sha256_update_ret` | `mbedtls_sha256_update` |
338| `mbedtls_sha256_finish_ret` | `mbedtls_sha256_finish` |
339| `mbedtls_sha256_ret` | `mbedtls_sha256` |
340| `mbedtls_sha512_starts_ret` | `mbedtls_sha512_starts` |
341| `mbedtls_sha512_update_ret` | `mbedtls_sha512_update` |
342| `mbedtls_sha512_finish_ret` | `mbedtls_sha512_finish` |
343| `mbedtls_sha512_ret` | `mbedtls_sha512` |
344
345To migrate to the this change the user can keep the `*_ret` names in their code
346and include the `compat_2.x.h` header file which holds macros with proper
347renaming or to rename those functions in their code according to the list from
348mentioned header file.
349
Dave Rodgmanb0e6bb52021-06-30 20:03:55 +0100350### Remove the RNG parameter from RSA verify functions
351
352RSA verification functions also no longer take random generator arguments (this
353was only needed when using a private key). This affects all applications using
354the RSA verify functions.
355
Dave Rodgman7b743192021-06-30 20:10:10 +0100356### Remove the padding parameters from `mbedtls_rsa_init()`
357
358This affects all users who use the RSA encryption, decryption, sign and
359verify APIs.
360
361The function `mbedtls_rsa_init()` no longer supports selecting the PKCS#1 v2.1
362encoding and its hash. It just selects the PKCS#1 v1.5 encoding by default. If
363you were using the PKCS#1 v2.1 encoding you now need, subsequently to the call
364to `mbedtls_rsa_init()`, to call `mbedtls_rsa_set_padding()` to set it.
365
366To choose the padding type when initializing a context, instead of
367
368```C
369 mbedtls_rsa_init(ctx, padding, hash_id);
370```
371
372use
373
374```C
375 mbedtls_rsa_init(ctx);
376 mbedtls_rsa_set_padding(ctx, padding, hash_id);
377```
378
379To use PKCS#1 v1.5 padding, instead of
380
381```C
382 mbedtls_rsa_init(ctx, MBEDTLS_RSA_PKCS_V15, <ignored>);
383```
384
385just use
386
387```C
388 mbedtls_rsa_init(ctx);
389```
390
Dave Rodgman897a95f2021-06-30 18:50:57 +0100391
392## High-level crypto
393
394Please also refer to the section [Low-level crypto](#low-level-crypto) for
395changes that could sit in either category.
396
Dave Rodgman30dc6032021-06-29 22:20:58 +0100397### Calling `mbedtls_cipher_finish()` is mandatory for all multi-part operations
398
399This only affects people who use the cipher module to perform AEAD operations
400using the multi-part API.
401
402Previously, the documentation didn't state explicitly if it was OK to call
403`mbedtls_cipher_check_tag()` or `mbedtls_cipher_write_tag()` directly after
404the last call to `mbedtls_cipher_update()` — that is, without calling
405`mbedtls_cipher_finish()` in-between. If you code was missing that call,
406please add it and be prepared to get as much as 15 bytes of output.
407
408Currently the output is always 0 bytes, but it may be more when alternative
409implementations of the underlying primitives are in use, or with future
410versions of the library.
411
Dave Rodgman30dc6032021-06-29 22:20:58 +0100412### Remove MD2, MD4, RC4, Blowfish and XTEA algorithms
413
414This change affects users of the MD2, MD4, RC4, Blowfish and XTEA algorithms.
415
416They are already niche or obsolete and most of them are weak or broken. For
417those reasons possible users should consider switching to modern and safe
418alternatives to be found in literature.
419
Dave Rodgman30dc6032021-06-29 22:20:58 +0100420### Deprecated functions were removed from cipher
421
422The functions `mbedtls_cipher_auth_encrypt()` and
423`mbedtls_cipher_auth_decrypt()` were removed. They were superseded by
424`mbedtls_cipher_auth_encrypt_ext()` and `mbedtls_cipher_auth_decrypt_ext()`
425respectively which additionally support key wrapping algorithms such as
426NIST_KW.
427
Dave Rodgman30dc6032021-06-29 22:20:58 +0100428### Extra parameter for the output buffer size
429
430The following functions now take an extra parameter indicating the size of the output buffer:
431
432* `mbedtls_ecdsa_write_signature()`, `mbedtls_ecdsa_write_signature_restartable()`
433* `mbedtls_pk_sign()`, `mbedtls_pk_sign_restartable()`
434
435The requirements for the output buffer have not changed, but passing a buffer that is too small now reliably causes the functions to return an error, rather than overflowing the buffer.
Dave Rodgman8cccbe12021-06-29 13:15:50 +0100436
Dave Rodgmana0e8db02021-06-29 18:05:38 +0100437### Signature functions now require the hash length to match the expected value
Dave Rodgman8cccbe12021-06-29 13:15:50 +0100438
439This affects users of the PK API as well as users of the low-level API in the RSA module. Users of the PSA API or of the ECDSA module are unaffected.
440
441All the functions in the RSA module that accept a `hashlen` parameter used to
442ignore it unless the `md_alg` parameter was `MBEDTLS_MD_NONE`, indicating raw
443data was signed. The `hashlen` parameter is now always the size that is read
444from the `hash` input buffer. This length must be equal to the output size of
445the hash algorithm used when signing a hash. (The requirements when signing
446raw data are unchanged.) This affects the following functions:
447
448* `mbedtls_rsa_pkcs1_sign`, `mbedtls_rsa_pkcs1_verify`
449* `mbedtls_rsa_rsassa_pkcs1_v15_sign`, `mbedtls_rsa_rsassa_pkcs1_v15_verify`
450* `mbedtls_rsa_rsassa_pss_sign`, `mbedtls_rsa_rsassa_pss_verify`
451* `mbedtls_rsa_rsassa_pss_sign_ext`, `mbedtls_rsa_rsassa_pss_verify_ext`
452
453The signature functions in the PK module no longer accept 0 as the `hash_len` parameter. The `hash_len` parameter is now always the size that is read from the `hash` input buffer. This affects the following functions:
454
455* `mbedtls_pk_sign`, `mbedtls_pk_verify`
456* `mbedtls_pk_sign_restartable`, `mbedtls_pk_verify_restartable`
457* `mbedtls_pk_verify_ext`
458
459The migration path is to pass the correct value to those functions.
Dave Rodgman759c0102021-06-29 15:55:08 +0100460
Dave Rodgman30dc6032021-06-29 22:20:58 +0100461### Some function parameters were made const
462
463Various functions in the PK and ASN.1 modules had a `const` qualifier added to
464some of their parameters.
465
466This normally doesn't affect your code, unless you use pointers to reference
467those functions. In this case, you'll need to update the type of your pointers
468in order to match the new signature.
469
Dave Rodgman9637bd32021-06-30 20:07:57 +0100470### The RNG parameter is now mandatory for all functions that accept one
471
472This change affects all users who called a function accepting a `f_rng`
473parameter with `NULL` as the value of this argument; this is no longer
474supported.
475
476The changed functions are: the X.509 CRT and CSR writing functions; the PK and
477RSA sign and decrypt functions; `mbedtls_rsa_private()`; the functions in DHM
478and ECDH that compute the shared secret; the scalar multiplication functions in
479ECP.
480
481You now need to pass a properly seeded, cryptographically secure RNG to all
482functions that accept a `f_rng` parameter. It is of course still possible to
483pass `NULL` as the context pointer `p_rng` if your RNG function doesn't need a
484context.
485
486Alternative implementations of a module (enabled with the `MBEDTLS_module_ALT`
487configuration options) may have their own internal and are free to ignore the
488`f_rng` argument but must allow users to pass one anyway.
489
490### Some functions gained an RNG parameter
491
492This affects users of the following functions: `mbedtls_ecp_check_pub_priv()`,
493`mbedtls_pk_check_pair()`, `mbedtls_pk_parse_key()`, and
494`mbedtls_pk_parse_keyfile()`.
495
496You now need to pass a properly seeded, cryptographically secure RNG when
497calling these functions. It is used for blinding, a countermeasure against
498side-channel attacks.
499
Dave Rodgman30dc6032021-06-29 22:20:58 +0100500
Dave Rodgman30dc6032021-06-29 22:20:58 +0100501## PSA
502
503### Deprecated names for PSA constants and types were removed
504
505Some constants and types that were present in beta versions of the PSA Crypto
506API were removed from version 1.0 of specification. Please switch to the new
507names provided by the 1.0 specification instead.
508
509
Dave Rodgman4ea56432021-06-30 14:16:22 +0100510## Changes that only affect alternative implementations
Dave Rodgman30dc6032021-06-29 22:20:58 +0100511
512### Internal / alt-focused headers were moved to a private location
513
514This shouldn't affect users who took care not to include headers that
515were documented as internal, despite being in the public include directory.
516
517If you're providing alt implementations of ECP or RSA, you'll need to add our
518`library` directory to your include path when building your alt
519implementations, and note that `ecp_internal.h` and `rsa_internal.h` have been
520renamed to `ecp_internal_alt.h` and `rsa_alt_helpers.h` respectively.
521
522If you're a library user and used to rely on having access to a structure or
523function that's now in a private header, please reach out on the mailing list
524and explain your need; we'll consider adding a new API in a future version.
525
526### CCM interface changes: impact for alternative implementations
527
528The CCM interface has changed with the addition of support for
529multi-part operations. Five new API functions have been defined:
Dave Rodgman9d341782021-06-30 18:35:43 +0100530 `mbedtls_ccm_starts()`, `mbedtls_ccm_set_lengths()`,
531 `mbedtls_ccm_update_ad()`, `mbedtls_ccm_update()` and `mbedtls_ccm_finish()`.
Dave Rodgman30dc6032021-06-29 22:20:58 +0100532Alternative implementations of CCM (`MBEDTLS_CCM_ALT`) have now to
533implement those additional five API functions.
534
535
Dave Rodgman30dc6032021-06-29 22:20:58 +0100536## X.509
537
Dave Rodgman30dc6032021-06-29 22:20:58 +0100538### Remove the certs module from the library
539
540This should not affect production use of the library, as the certificates and
541keys included there were never suitable for production use.
542
543However it might affect you if you relied on them for testing purposes. In
544that case, please embed your own test certificates in your test code; now that
545`certs.c` is out of the library there is no longer any stability guaranteed
546and it may change in incompatible ways at any time.
547
548### Change the API to allow adding critical extensions to CSRs
549
550This affects applications that call the `mbedtls_x509write_csr_set_extension`
551function.
552
Dave Rodgman6753a772021-06-30 17:15:28 +0100553The API is changed to include the parameter `critical` which enables marking an
Dave Rodgman30dc6032021-06-29 22:20:58 +0100554extension included in a CSR as critical. To get the previous behavior pass 0.
555
Dave Rodgmana0148312021-06-30 19:08:51 +0100556### Remove the config option `MBEDTLS_X509_ALLOW_UNSUPPORTED_CRITICAL_EXTENSION`
Dave Rodgman30dc6032021-06-29 22:20:58 +0100557
558This change does not affect users of the default configuration; it only affects
559users who enable this option.
560
561The X.509 standard says that implementations must reject critical extensions that
562they don't recognize, and this is what Mbed TLS does by default. This option
563allowed to continue parsing those certificates but didn't provide a convenient
564way to handle those extensions.
565
566The migration path from that option is to use the
567`mbedtls_x509_crt_parse_der_with_ext_cb()` function which is functionally
568equivalent to `mbedtls_x509_crt_parse_der()`, and/or
569`mbedtls_x509_crt_parse_der_nocopy()` but it calls the callback with every
570unsupported certificate extension and additionally the "certificate policies"
571extension if it contains any unsupported certificate policies.
572
573### Remove `MBEDTLS_X509_CHECK_*_KEY_USAGE` options from `mbedtls_config.h`
574
575This change affects users who have chosen the configuration options to disable the
576library's verification of the `keyUsage` and `extendedKeyUsage` fields of x509
577certificates.
578
579The `MBEDTLS_X509_CHECK_KEY_USAGE` and `MBEDTLS_X509_CHECK_EXTENDED_KEY_USAGE`
580configuration options are removed and the X509 code now behaves as if they were
581always enabled. It is consequently not possible anymore to disable at compile
582time the verification of the `keyUsage` and `extendedKeyUsage` fields of X509
583certificates.
584
585The verification of the `keyUsage` and `extendedKeyUsage` fields is important,
586disabling it can cause security issues and it is thus not recommended. If the
587verification is for some reason undesirable, it can still be disabled by means
588of the verification callback function passed to `mbedtls_x509_crt_verify()` (see
589the documentation of this function for more information).
590
591### Remove the `MBEDTLS_X509_ALLOW_EXTENSIONS_NON_V3` option
592
593This change does not affect users who were using the default configuration, as
594this option was already disabled by default. Also, it does not affect users who
595are working with current V3 X.509 certificates.
596
597Extensions were added in V3 of the X.509 specification, so pre-V3 certificates
598containing extensions were never compliant. Mbed TLS now rejects them with a
599parsing error in all configurations, as it did previously in the default
600configuration.
601
602If you are working with the pre-V3 certificates you need to switch to the
603current ones.
604
Dave Rodgmanb1c6b4a2021-06-30 14:17:21 +0100605### Strengthen default algorithm selection for X.509
Dave Rodgman30dc6032021-06-29 22:20:58 +0100606
Dave Rodgman7d2ac882021-06-30 19:02:36 +0100607This is described in the section [Strengthen default algorithm selection for X.509 and TLS](#strengthen-default-algorithm-selection-for-x.509-and-tls).
Dave Rodgman30dc6032021-06-29 22:20:58 +0100608
Dave Rodgmanaa1fba22021-06-30 18:41:24 +0100609### Remove wrapper for libpkcs11-helper
610
611This doesn't affect people using the default configuration as it was already
612disabled by default.
613
614If you used to rely on this module in order to store your private keys
615securely, please have a look at the key management facilities provided by the
616PSA crypto API. If you have a use case that's not covered yet by this API,
617please reach out on the mailing list.
618
Dave Rodgmanc936bbb2021-06-30 14:19:30 +0100619
Dave Rodgman30dc6032021-06-29 22:20:58 +0100620## SSL
621
622### Remove support for TLS 1.0, 1.1 and DTLS 1.0
623
624This change affects users of the TLS 1.0, 1.1 and DTLS 1.0 protocols.
625
626These versions have been deprecated by RFC 8996.
627Keeping them in the library creates opportunities for misconfiguration
628and possibly downgrade attacks. More generally, more code means a larger attack
629surface, even if the code is supposedly not used.
630
631The migration path is to adopt the latest versions of the protocol.
632
633As a consequence of removing TLS 1.0, support for CBC record splitting was
634also removed, as it was a work-around for a weakness in this particular
635version. There is no migration path since the feature is no longer relevant.
636
637As a consequence of currently supporting only one version of (D)TLS (and in the
638future 1.3 which will have a different version negotiation mechanism), support
639for fallback SCSV (RFC 7507) was also removed. There is no migration path as
640it's no longer useful with TLS 1.2 and later.
641
642As a consequence of currently supporting only one version of (D)TLS (and in the
643future 1.3 which will have a different concept of ciphersuites), support for
644configuring ciphersuites separately for each version via
645`mbedtls_ssl_conf_ciphersuites_for_version()` was removed. Use
646`mbedtls_ssl_conf_ciphersuites()` to configure ciphersuites to use with (D)TLS
6471.2; in the future a different API will be added for (D)TLS 1.3.
648
649### Remove support for SSL 3.0
650
651This doesn't affect people using the default configuration as it was already
652disabled by default.
653
654This only affects TLS users who explicitly enabled `MBEDTLS_SSL_PROTO_SSL3`
655and relied on that version in order to communicate with peers that are not up
656to date. If one of your peers is in that case, please try contacting them and
657encouraging them to upgrade their software.
658
659### Remove support for parsing SSLv2 ClientHello
660
661This doesn't affect people using the default configuration as it was already
662disabled by default.
663
664This only affects TLS servers that have clients who send an SSLv2 ClientHello.
665These days clients are very unlikely to do that. If you have a client that
666does, please try contacting them and encouraging them to upgrade their
667software.
668
669### Remove support for truncated HMAC
670
671This affects users of truncated HMAC, that is, users who called
672`mbedtls_ssl_conf_truncated_hmac( ..., MBEDTLS_SSL_TRUNC_HMAC_ENABLED)`,
673regardless of whether the standard version was used or compatibility version
674(`MBEDTLS_SSL_TRUNCATED_HMAC_COMPAT`).
675
676The recommended migration path for people who want minimal overhead is to use a
677CCM-8 ciphersuite.
678
679### Remove support for TLS record-level compression
680
681This doesn't affect people using the default configuration as it was already
682disabled by default.
683
684This only affects TLS users who enabled `MBEDTLS_ZLIB_SUPPORT`. This will not
685cause any failures however if you used to enable TLS record-level compression
686you may find that your bandwidth usage increases without compression. There's
687no general solution to this problem; application protocols might have their
688own compression mechanisms and are in a better position than the TLS stack to
689avoid variants of the CRIME and BREACH attacks.
690
691### Remove support for TLS RC4-based ciphersuites
692
693This does not affect people who used the default `mbedtls_config.h` and the default
694list of ciphersuites, as RC4-based ciphersuites were already not negotiated in
695that case.
696
697Please switch to any of the modern, recommended ciphersuites (based on
698AES-GCM, AES-CCM or ChachaPoly for example) and if your peer doesn't support
699any, encourage them to upgrade their software.
700
701### Remove support for TLS single-DES ciphersuites
702
703This doesn't affect people using the default configuration as it was already
704disabled by default.
705
706Please switch to any of the modern, recommended ciphersuites (based on
707AES-GCM, AES-CCM or ChachaPoly for example) and if your peer doesn't support
708any, encourage them to upgrade their software.
709
710### Remove support for TLS record-level hardware acceleration
711
712This doesn't affect people using the default configuration as it was already
713disabled by default.
714
715This feature had been broken for a while so we doubt anyone still used it.
716However if you did, please reach out on the mailing list and let us know about
717your use case.
718
719### Remove config option `MBEDTLS_SSL_DEFAULT_TICKET_LIFETIME`
720
721This doesn't affect people using the default configuration.
722
723This option has not had any effect for a long time. Please use the `lifetime`
724parameter of `mbedtls_ssl_ticket_setup()` instead.
725
Dave Rodgman30dc6032021-06-29 22:20:58 +0100726### Combine the `MBEDTLS_SSL_CID_PADDING_GRANULARITY` and `MBEDTLS_SSL_TLS1_3_PADDING_GRANULARITY` options
727
728This change affects users who modified the default `mbedtls_config.h` padding granularity
729settings, i.e. enabled at least one of the options.
730
731The `mbedtls_config.h` options `MBEDTLS_SSL_CID_PADDING_GRANULARITY` and
732`MBEDTLS_SSL_TLS1_3_PADDING_GRANULARITY` were combined into one option because
733they used exactly the same padding mechanism and hence their respective padding
734granularities can be used in exactly the same way. This change simplifies the
735code maintenance.
736
737The new single option `MBEDTLS_SSL_CID_TLS1_3_PADDING_GRANULARITY` can be used
738for both DTLS-CID and TLS 1.3.
739
740### TLS now favors faster curves over larger curves
741
742The default preference order for curves in TLS now favors resource usage (performance and memory consumption) over size. The exact order is unspecified and may change, but generally you can expect 256-bit curves to be preferred over larger curves.
743
744If you prefer a different order, call `mbedtls_ssl_conf_curves()` when configuring a TLS connection.
745
746### SSL key export interface change
747
748This affects users of the SSL key export APIs:
Dave Rodgman26c12eb2021-06-30 19:58:00 +0100749```
Dave Rodgman30dc6032021-06-29 22:20:58 +0100750 mbedtls_ssl_conf_export_keys_cb()
751 mbedtls_ssl_conf_export_keys_ext_cb()
752```
753
754Those APIs have been removed and replaced by the new API
755`mbedtls_ssl_set_export_keys_cb()`. This API differs from
756the previous key export API in the following ways:
757
758- It is no longer bound to an SSL configuration, but to an
759 SSL context. This allows users to more easily identify the
760 connection an exported key belongs to.
761- It no longer exports raw keys and IV.
762- A secret type parameter has been added to identify which key
763 is being exported. For TLS 1.2, only the master secret is
764 exported, but upcoming TLS 1.3 support will add other kinds of keys.
765- The callback now specifies a void return type, rather than
766 returning an error code. It is the responsibility of the application
767 to handle failures in the key export callback, for example by
768 shutting down the TLS connection.
769
770For users which do not rely on raw keys and IV, adjusting to the new
771callback type should be straightforward — see the example programs
772`programs/ssl/ssl_client2` and `programs/ssl/ssl_server2` for callbacks
773for NSSKeylog, EAP-TLS and DTLS-SRTP.
774
775Users which require access to the raw keys used to secure application
776traffic may derive those by hand based on the master secret and the
777handshake transcript hashes which can be obtained from the raw data
778on the wire. Such users are also encouraged to reach out to the
779Mbed TLS team on the mailing list, to let the team know about their
780use case.
781
782### Remove MaximumFragmentLength (MFL) query API
783
784This affects users which use the MFL query APIs
785`mbedtls_ssl_get_{input,output}_max_frag_len()` to
786infer upper bounds on the plaintext size of incoming and
787outgoing record.
788
789Users should switch to `mbedtls_ssl_get_max_{in,out}_record_payload()`
790instead, which also provides such upper bounds but takes more factors
791than just the MFL configuration into account.
792
793### Relaxed semantics for PSK configuration
794
795This affects users which call the PSK configuration APIs
796`mbedtlsl_ssl_conf_psk()` and `mbedtls_ssl_conf_psk_opaque()`
797multiple times on the same SSL configuration.
798
799In Mbed TLS 2.x, users would observe later calls overwriting
800the effect of earlier calls, with the prevailing PSK being
801the one that has been configured last. In Mbed TLS 3.0,
802calling `mbedtls_ssl_conf_[opaque_]psk()` multiple times
803will return an error, leaving the first PSK intact.
804
805To achieve equivalent functionality when migrating to Mbed TLS 3.0,
806users calling `mbedtls_ssl_conf_[opaque_]psk()` multiple times should
807remove all but the last call, so that only one call to _either_
808`mbedtls_ssl_conf_psk()` _or_ `mbedtls_ssl_conf_psk_opaque()`
809remains.
810
811### Remove the configuration to enable weak ciphersuites in SSL / TLS
812
813This does not affect users who use the default `mbedtls_config.h`, as this option was
814already off by default.
815
816If you were using a weak cipher, please switch to any of the modern,
817recommended ciphersuites (based on AES-GCM, AES-CCM or ChachaPoly for example)
818and if your peer doesn't support any, encourage them to upgrade their software.
819
820If you were using a ciphersuite without encryption, you just have to
Dave Rodgman9d341782021-06-30 18:35:43 +0100821enable `MBEDTLS_CIPHER_NULL_CIPHER` now.
Dave Rodgman30dc6032021-06-29 22:20:58 +0100822
823### Remove the `MBEDTLS_SSL_MAX_CONTENT_LEN` configuration option
824
825This affects users who use the `MBEDTLS_SSL_MAX_CONTENT_LEN` option to
826set the maximum length of incoming and outgoing plaintext fragments,
827which can save memory by reducing the size of the TLS I/O buffers.
828
829This option is replaced by the more fine-grained options
830`MBEDTLS_SSL_IN_CONTENT_LEN` and `MBEDTLS_SSL_OUT_CONTENT_LEN` that set
831the maximum incoming and outgoing plaintext fragment lengths, respectively.
832
833### Remove the SSL API `mbedtls_ssl_get_session_pointer()`
834
835This affects two classes of users:
836
8371. Users who manually inspect parts of the current session through
838 direct structure field access.
839
8402. Users of session resumption who query the current session
841 via `mbedtls_ssl_get_session_pointer()` prior to saving or exporting
842 it via `mbedtls_ssl_session_copy()` or `mbedtls_ssl_session_save()`,
843 respectively.
844
845Migration paths:
846
8471. Mbed TLS 3.0 does not offer a migration path for the use case 1: Like many
848 other Mbed TLS structures, the structure of `mbedtls_ssl_session` is no
849 longer part of the public API in Mbed TLS 3.0, and direct structure field
Dave Rodgmana5a3cce2021-06-30 11:06:58 +0100850 access is no longer supported. Please see the [section on private structure fields](#most-structure-fields-are-now-private) for more details.
Dave Rodgman30dc6032021-06-29 22:20:58 +0100851
8522. Users should replace calls to `mbedtls_ssl_get_session_pointer()` by
853 calls to `mbedtls_ssl_get_session()` as demonstrated in the example
854 program `programs/ssl/ssl_client2.c`.
855
856### Remove `MBEDTLS_SSL_DTLS_BADMAC_LIMIT` option
857
858This change does not affect users who used the default `mbedtls_config.h`, as the option
Dave Rodgman9d341782021-06-30 18:35:43 +0100859`MBEDTLS_SSL_DTLS_BADMAC_LIMIT` was already on by default.
Dave Rodgman30dc6032021-06-29 22:20:58 +0100860
861This option was a trade-off between functionality and code size: it allowed
862users who didn't need that feature to avoid paying the cost in code size, by
863disabling it.
864
865This option is no longer present, but its functionality is now always enabled.
866
867### Deprecated functions were removed from SSL
868
869The function `mbedtls_ssl_conf_dh_param()` was removed. Please use
870`mbedtls_ssl_conf_dh_param_bin()` or `mbedtls_ssl_conf_dh_param_ctx()` instead.
871
872The function `mbedtls_ssl_get_max_frag_len()` was removed. Please use
873`mbedtls_ssl_get_max_out_record_payload()` and
874`mbedtls_ssl_get_max_in_record_payload()`
875instead.
876
877### Remove `MBEDTLS_SSL_RECORD_CHECKING` option and enable its action by default
878
Dave Rodgman9d341782021-06-30 18:35:43 +0100879This change does not affect users who use the default `mbedtls_config.h`, as the
880option `MBEDTLS_SSL_RECORD_CHECKING` was already on by default.
Dave Rodgman30dc6032021-06-29 22:20:58 +0100881
882This option was added only to control compilation of one function,
Dave Rodgman9d341782021-06-30 18:35:43 +0100883 `mbedtls_ssl_check_record()`, which is only useful in some specific cases, so it
Dave Rodgman30dc6032021-06-29 22:20:58 +0100884was made optional to allow users who don't need it to save some code space.
Dave Rodgmane4ec8462021-06-30 09:52:40 +0100885However, the same effect can be achieved by using link-time garbage collection.
Dave Rodgman30dc6032021-06-29 22:20:58 +0100886
887Users who changed the default setting of the option need to change the config/
888build system to remove that change.
Dave Rodgman759c0102021-06-29 15:55:08 +0100889
Dave Rodgmana0e8db02021-06-29 18:05:38 +0100890### Session Cache API Change
Dave Rodgman8cccbe12021-06-29 13:15:50 +0100891
892This affects users who use `mbedtls_ssl_conf_session_cache()`
893to configure a custom session cache implementation different
894from the one Mbed TLS implements in `library/ssl_cache.c`.
895
896Those users will need to modify the API of their session cache
897implementation to that of a key-value store with keys being
898session IDs and values being instances of `mbedtls_ssl_session`:
899
Dave Rodgman10963272021-06-30 19:11:22 +0100900```C
Dave Rodgman8cccbe12021-06-29 13:15:50 +0100901typedef int mbedtls_ssl_cache_get_t( void *data,
902 unsigned char const *session_id,
903 size_t session_id_len,
904 mbedtls_ssl_session *session );
905typedef int mbedtls_ssl_cache_set_t( void *data,
906 unsigned char const *session_id,
907 size_t session_id_len,
908 const mbedtls_ssl_session *session );
909```
910
911Since the structure of `mbedtls_ssl_session` is no longer public from 3.0
912onwards, portable session cache implementations must not access fields of
913`mbedtls_ssl_session`. See the corresponding migration guide. Users that
914find themselves unable to migrate their session cache functionality without
Dave Rodgman759c0102021-06-29 15:55:08 +0100915accessing fields of `mbedtls_ssl_session` should describe their use case
Dave Rodgman8cccbe12021-06-29 13:15:50 +0100916on the Mbed TLS mailing list.
Dave Rodgman759c0102021-06-29 15:55:08 +0100917
Dave Rodgmanb491b2b2021-06-30 09:46:07 +0100918### Changes in the SSL error code space
Dave Rodgman8cccbe12021-06-29 13:15:50 +0100919
920This affects users manually checking for the following error codes:
Dave Rodgman759c0102021-06-29 15:55:08 +0100921
Dave Rodgman8cccbe12021-06-29 13:15:50 +0100922- `MBEDTLS_ERR_SSL_CERTIFICATE_REQUIRED`
923- `MBEDTLS_ERR_SSL_INVALID_VERIFY_HASH`
924- `MBEDTLS_ERR_SSL_CERTIFICATE_TOO_LARGE`
925- `MBEDTLS_ERR_SSL_NO_CIPHER_CHOSEN`
926- `MBEDTLS_ERR_SSL_NO_USABLE_CIPHERSUITE`
927- `MBEDTLS_ERR_SSL_BAD_HS_XXX`
928
929Migration paths:
Dave Rodgman8cccbe12021-06-29 13:15:50 +0100930- `MBEDTLS_ERR_SSL_CERTIFICATE_TOO_LARGE` has been removed, and
931 `MBEDTLS_ERR_SSL_BUFFER_TOO_SMALL` is returned instead if the user's own certificate
Dave Rodgman759c0102021-06-29 15:55:08 +0100932 is too large to fit into the output buffers.
933
934 Users should check for `MBEDTLS_ERR_SSL_BUFFER_TOO_SMALL` instead, and potentially
935 compare the size of their own certificate against the configured size of the output buffer to
936 understand if the error is due to an overly large certificate.
Dave Rodgman8cccbe12021-06-29 13:15:50 +0100937
Dave Rodgmanb491b2b2021-06-30 09:46:07 +0100938- `MBEDTLS_ERR_SSL_NO_CIPHER_CHOSEN` and `MBEDTLS_ERR_SSL_NO_USABLE_CIPHERSUITE` have been
939 replaced by `MBEDTLS_ERR_SSL_HANDSHAKE_FAILURE`.
940
Dave Rodgmana54c1682021-06-30 11:11:07 +0100941- All codes of the form `MBEDTLS_ERR_SSL_BAD_HS_XXX` have been replaced by various alternatives, which give more information about the type of error raised.
Dave Rodgmanb491b2b2021-06-30 09:46:07 +0100942
943 Users should check for the newly introduced generic error codes
944
945 * `MBEDTLS_ERR_SSL_DECODE_ERROR`
946 * `MBEDTLS_ERR_SSL_ILLEGAL_PARAMETER`,
947 * `MBEDTLS_ERR_SSL_HANDSHAKE_FAILURE`
948 * `MBEDTLS_ERR_SSL_BAD_PROTOCOL_VERSION`
949 * `MBEDTLS_ERR_SSL_BAD_CERTIFICATE`
950 * `MBEDTLS_ERR_SSL_UNRECOGNIZED_NAME`
951 * `MBEDTLS_ERR_SSL_UNSUPPORTED_EXTENSION`
952 * `MBEDTLS_ERR_SSL_NO_APPLICATION_PROTOCOL`
953
954 and the pre-existing generic error codes
955
956 * `MBEDTLS_ERR_SSL_UNEXPECTED_MESSAGE`
957 * `MBEDTLS_ERR_SSL_INTERNAL_ERROR`
958
959 instead.
Dave Rodgman759c0102021-06-29 15:55:08 +0100960
Dave Rodgmand267ec32021-06-29 21:31:58 +0100961### Modified semantics of `mbedtls_ssl_{get,set}_session()`
Dave Rodgman8cccbe12021-06-29 13:15:50 +0100962
963This affects users who call `mbedtls_ssl_get_session()` or
964`mbedtls_ssl_set_session()` multiple times on the same SSL context
965representing an established TLS 1.2 connection.
966Those users will now observe the second call to fail with
967`MBEDTLS_ERR_SSL_FEATURE_UNAVAILABLE`.
968
969Migration path:
970- Exporting the same TLS 1.2 connection multiple times via
971 `mbedtls_ssl_get_session()` leads to multiple copies of
972 the same session. This use of `mbedtls_ssl_get_session()`
973 is discouraged, and the following should be considered:
974 * If the various session copies are later loaded into
975 fresh SSL contexts via `mbedtls_ssl_set_session()`,
976 export via `mbedtls_ssl_get_session()` only once and
977 load the same session into different contexts via
978 `mbedtls_ssl_set_session()`. Since `mbedtls_ssl_set_session()`
979 makes a copy of the session that's being loaded, this
980 is functionally equivalent.
981 * If the various session copies are later serialized
982 via `mbedtls_ssl_session_save()`, export and serialize
983 the session only once via `mbedtls_ssl_get_session()` and
984 `mbedtls_ssl_session_save()` and make copies of the raw
985 data instead.
986- Calling `mbedtls_ssl_set_session()` multiple times in Mbed TLS 2.x
987 is not useful since subsequent calls overwrite the effect of previous
Dave Rodgman759c0102021-06-29 15:55:08 +0100988 calls. Applications achieve equivalent functional behavior by
Dave Rodgman8cccbe12021-06-29 13:15:50 +0100989 issuing only the very last call to `mbedtls_ssl_set_session()`.
990
Dave Rodgmand267ec32021-06-29 21:31:58 +0100991### Turn `MBEDTLS_SSL_SRV_RESPECT_CLIENT_PREFERENCE` configuration option into a runtime option
Dave Rodgman8cccbe12021-06-29 13:15:50 +0100992
Dave Rodgman9d341782021-06-30 18:35:43 +0100993This change affects users who were enabling `MBEDTLS_SSL_SRV_RESPECT_CLIENT_PREFERENCE`
Dave Rodgman8cccbe12021-06-29 13:15:50 +0100994option in the `mbedtls_config.h`
995
996This option has been removed and a new function with similar functionality has
997been introduced into the SSL API.
998
999This new function `mbedtls_ssl_conf_preference_order()` can be used to
1000change the preferred order of ciphersuites on the server to those used on the client,
1001e.g.: `mbedtls_ssl_conf_preference_order(ssl_config, MBEDTLS_SSL_SRV_CIPHERSUITE_ORDER_CLIENT)`
1002has the same effect as enabling the removed option. The default state is to use
1003the server order of suites.
Dave Rodgmane45e6402021-06-29 13:21:55 +01001004
Dave Rodgmana3758202021-06-30 14:17:03 +01001005### Strengthen default algorithm selection for X.509 and TLS
1006
1007The default X.509 verification profile (`mbedtls_x509_crt_profile_default`) and the default curve and hash selection in TLS have changed. They are now aligned, except that the X.509 profile only lists curves that support signature verification.
1008
1009Hashes and curves weaker than 255 bits (security strength less than 128 bits) are no longer accepted by default. The following hashes have been removed: SHA-1 (formerly only accepted for key exchanges but not for certificate signatures), SHA-224 (weaker hashes were already not accepted). The following curves have been removed: secp192r1, secp224r1, secp192k1, secp224k1.
1010
1011The compile-time options `MBEDTLS_TLS_DEFAULT_ALLOW_SHA1_IN_CERTIFICATES` and `MBEDTLS_TLS_DEFAULT_ALLOW_SHA1_IN_KEY_EXCHANGE` are no longer available.
1012
1013The curve secp256k1 has also been removed from the default X.509 and TLS profiles. [RFC 8422](https://datatracker.ietf.org/doc/html/rfc8422#section-5.1.1) deprecates it in TLS, and it is very rarely used, although it is not known to be weak at the time of writing.
1014
1015If you still need to accept certificates signed with algorithms that have been removed from the default profile, call `mbedtls_x509_crt_verify_with_profile` instead of `mbedtls_x509_crt_verify` and pass a profile that allows the curves and hashes you want. For example, to allow SHA-224:
Dave Rodgman10963272021-06-30 19:11:22 +01001016```C
Dave Rodgmana3758202021-06-30 14:17:03 +01001017mbedtls_x509_crt_profile my_profile = mbedtls_x509_crt_profile_default;
1018my_profile.allowed_mds |= MBEDTLS_X509_ID_FLAG( MBEDTLS_MD_SHA224 );
1019```
1020
1021If you still need to allow hashes and curves in TLS that have been removed from the default configuration, call `mbedtls_ssl_conf_sig_hashes()` and `mbedtls_ssl_conf_curves()` with the desired lists.
Dave Rodgman3f669432021-06-30 18:43:49 +01001022
1023### Remove 3DES ciphersuites
1024
1025This change does not affect users using default settings for 3DES in `mbedtls_config.h`
1026because the 3DES ciphersuites were disabled by that.
1027
10283DES has weaknesses/limitations and there are better alternatives, and more and
1029more standard bodies are recommending against its use in TLS.
1030
1031The migration path here is to chose from the alternatives recommended in the
1032literature, such as AES.