Commit Graph

5202 Commits

Author SHA1 Message Date
Gilles Peskine
18ac716021 RSA: wipe more stack buffers
MGF mask and PSS salt are not highly sensitive, but wipe them anyway
for good hygiene.
2017-05-16 10:22:37 +01:00
Gilles Peskine
4a7f6a0ddb RSA: wipe stack buffers
The RSA private key functions rsa_rsaes_pkcs1_v15_decrypt and
rsa_rsaes_oaep_decrypt put sensitive data (decryption results) on the
stack. Wipe it before returning.

Thanks to Laurent Simon for reporting this issue.
2017-05-16 10:22:37 +01:00
Simon Butcher
77da95357f Add fix for #667 to ChangeLog 2017-05-16 10:22:37 +01:00
Simon Butcher
4775e83337 Fix ChangeLog
Fixes the ChangeLog following the merge of the ECP HW acceleration interfaces
in aab9efb.
2017-05-16 10:22:37 +01:00
Janos Follath
6e876988de Fix bug in threading sample implementation #667 2017-05-16 10:22:37 +01:00
Janos Follath
45182a0065 Add Changelog entry for RSA exponent blinding 2017-05-16 10:22:37 +01:00
Janos Follath
f9203b4139 Add exponent blinding to RSA with CRT
The sliding window exponentiation algorithm is vulnerable to
side-channel attacks. As a countermeasure we add exponent blinding in
order to prevent combining the results of different measurements.

This commit handles the case when the Chinese Remainder Theorem is used
to accelerate the computation.
2017-05-16 10:22:37 +01:00
Janos Follath
e81102e476 Add exponent blinding to RSA without CRT
The sliding window exponentiation algorithm is vulnerable to
side-channel attacks. As a countermeasure we add exponent blinding in
order to prevent combining the results of fifferent measurements.

This commits handles the case when the Chinese Remainder Theorem is NOT
used to accelerate computations.
2017-05-16 10:22:37 +01:00
Janos Follath
7a8a090f55 Add Changelog entry for CA list suppression 2017-05-16 10:22:37 +01:00
Janos Follath
89baba2375 Add tests for the CA suppression option 2017-05-16 10:22:37 +01:00
Janos Follath
4817e27d4d Add the CA list suppression option to ssl_server2
Adding the CA suppression list option to the 'ssl_server2' sample
program is a prerequisite for adding tests for this feature to the
integration test suite (ssl-opt.sh).
2017-05-16 10:22:37 +01:00
Janos Follath
088ce43ffe Implement optional CA list suppression in Certificate Request
According to RFC5246 the server can indicate the known Certificate
Authorities or can constrain the aurhorisation space by sending a
certificate list. This part of the message is optional and if omitted,
the client may send any certificate in the response.

The previous behaviour of mbed TLS was to always send the name of all the
CAs that are configured as root CAs. In certain cases this might cause
usability and privacy issues for example:
- If the list of the CA names is longer than the peers input buffer then
  the handshake will fail
- If the configured CAs belong to third parties, this message gives away
  information on the relations to these third parties

Therefore we introduce an option to suppress the CA list in the
Certificate Request message.

Providing this feature as a runtime option comes with a little cost in
code size and advantages in maintenance and flexibility.
2017-05-16 10:22:37 +01:00
Andres Amaya Garcia
75fdf631fd Remove obsolete macros from compat-1.3.h 2017-05-16 10:22:37 +01:00
Hanno Becker
cf7ae7e7d3 Improve documentation 2017-05-15 11:50:13 +01:00
Hanno Becker
0d0cd4b30e Split long lines 2017-05-15 11:50:13 +01:00
Hanno Becker
7e5437a972 Remember suitable hash function for any signature algorithm.
This commit changes `ssl_parse_signature_algorithms_ext` to remember
one suitable ( := supported by client and by our config ) hash
algorithm per signature algorithm.

It also modifies the ciphersuite checking function
`ssl_ciphersuite_match` to refuse a suite if there
is no suitable hash algorithm.

Finally, it adds the corresponding entry to the ChangeLog.
2017-05-15 11:50:11 +01:00
Hanno Becker
1aa267cbc3 Introduce macros and functions to characterize certain ciphersuites.
The routine `mbedtls_ssl_write_server_key_exchange` heavily depends on
what kind of cipher suite is active: some don't need a
ServerKeyExchange at all, some need (EC)DH parameters but no server
signature, some require both. Each time we want to restrict a certain
piece of code to some class of ciphersuites, it is guarded by a
lengthy concatentation of configuration checks determining whether at
least one of the relevant cipher suites is enabled in the config; on
the code level, it is guarded by the check whether one of these
cipher suites is the active one.

To ease readability of the code, this commit introduces several helper
macros and helper functions that can be used to determine whether a
certain class of ciphersuites (a) is active in the config, and
(b) contains the currently present ciphersuite.
2017-05-15 11:46:57 +01:00
Andres Amaya Garcia
af610a0baf Fix check-doxy-blocks.pl errors (cmac.c ecjpake.h) 2017-05-12 00:18:04 +01:00
Andres Amaya Garcia
d3f0f5e1c8 Fix check-doxy-blocks.pl to run from root dir
Modify tests/scripts/check-doxy-blocks.pl to ensure that:
  * It can only be run from the mbed TLS root directory.
  * An error code is returned to the environment when a potential error
    in the source code is found.
2017-05-12 00:16:50 +01:00
Janos Follath
5634b8609b Apply feedback to ECP internal interface documentation 2017-05-11 22:42:14 +01:00
Janos Follath
aab9efb4ce Add doxygen documentation to the new ECP interface
Document the functions in the Elliptic Curve Point module hardware
acceleration to guide silicon vendors when implementing the drivers.
2017-05-11 22:42:14 +01:00
Janos Follath
6c8ccd5be4 Fix cleanup label alignment 2017-05-11 22:42:14 +01:00
Janos Follath
7e2406f6f0 Add Changelog entry 2017-05-11 22:42:14 +01:00
Janos Follath
69b2051e76 Update version features with ECP macros 2017-05-11 22:42:14 +01:00
Janos Follath
333d163523 Remove obsolote ECP abstraction macros
Some macros have been removed from the Elliptic Curve Point abstraction
layer, and they shouldn't have remained in the check_config.h
2017-05-11 22:42:14 +01:00
Janos Follath
c44ab97cc9 Apply review feedback on ECP interface
Naming conventions are revised, style issues corrected and minor
optimisation added.
2017-05-11 22:42:14 +01:00
Janos Follath
b8a90fb51c Update ECP hardware abstraction interface
With this commit the Elliptic Curve Point interface is rewised. Two
compile time options has been removed to simplify the interface and
the function names got a new prefix that indicates that these functions
are for internal use and not part of the public interface.
2017-05-11 22:42:14 +01:00
Janos Follath
ee782bcd54 Adjust documentation to new ECP function names. 2017-05-11 22:42:14 +01:00
Janos Follath
430d3376c9 Add thread safety to ECP hardware acceleration
The intended use of the abstraction layer for Elliptic Curve Point
arithmetic is to enable using hardware cryptographic accelerators.
These devices are a shared resource and the driver code rarely provides
thread safety.

This commit adds mutexes to the abstraction layer to protect the device
in a multi-threaded environment.
2017-05-11 22:42:14 +01:00
Janos Follath
552f6b6d17 Add global mutex for asymmetric crypto accelerator
The primary use case behind providing an abstraction layer to enable
alternative Elliptic Curve Point arithmetic implementation, is making
use of cryptographic acceleration hardware if it is present.

To provide thread safety for the hardware accelerator we need a mutex
to guard it.
2017-05-11 22:42:14 +01:00
Janos Follath
16e63ea2e2 Fix alternative ECP function names
The alternative Elliptic Curve Point arithmetic functions didn't have
an 'mbedtls_' prefix as required by check-names.sh.
2017-05-11 22:42:14 +01:00
Janos Follath
4d9c69dde8 Fix ECP alternative macro names
The compile time macros enabling the initialisation and deinitialisation
in the alternative Elliptic Curve Point arithmetic implementation had
names that did not end with '_ALT' as required by check-names.sh.
2017-05-11 22:42:14 +01:00
Janos Follath
47d28f08a6 Remove alt_internal directory 2017-05-11 22:42:14 +01:00
Janos Follath
372697b6b7 Clean up ECP abstraction layer code
This commit fixes several style issues and fixes in the documentation
of the Elliptic Curve Point arithmetic abstraction layer.
2017-05-11 22:42:14 +01:00
Janos Follath
b069753313 ECP: Add module and function level replacement options. 2017-05-11 22:42:14 +01:00
Andres AG
5c79d25d94 Add PK tests to avoid hashlen overflow for RSA 2017-05-11 21:58:25 +01:00
Andres AG
72849877d0 Fix data loss in unsigned int cast in PK
This patch introduces some additional checks in the PK module for 64-bit
systems only. The problem is that the API functions in the PK
abstraction accept a size_t value for the hashlen, while the RSA module
accepts an unsigned int for the hashlen. Instead of silently casting
size_t to unsigned int, this change checks whether the hashlen overflows
an unsigned int and returns an error.
2017-05-11 21:55:17 +01:00
Manuel Pégourié-Gonnard
c1380de887 RSA PKCS1v1.5 verification: check padding length
The test case was generated by modifying our signature code so that it
produces a 7-byte long padding (which also means garbage at the end, so it is
essential in to check that the error that is detected first is indeed the
padding rather than the final length check).
2017-05-11 13:10:13 +02:00
Gilles Peskine
3df98f50a9 Changelog entry for the bug fixes 2017-05-10 17:47:40 +02:00
Gilles Peskine
92e4426169 SSLv3: when refusing renegotiation, stop processing
Before the code was sending a fatal alert but then processing any
further data from the peer.

Internal reference: IOTSSL-1384
2017-05-10 17:31:13 +02:00
Gilles Peskine
c94f7352fa Ignore failures when sending fatal alerts
In many places in TLS handling, some code detects a fatal error, sends
a fatal alert message, and returns to the caller. If sending the alert
fails, then return the error that triggered the alert, rather than
overriding the return status. This effectively causes alert sending
failures to be ignored. Formerly the code was inconsistently sometimes
doing one, sometimes the other.

In general ignoring the alert is the right thing: what matters to the
caller is the original error. A typical alert failure is that the
connection is already closed.

One case which remains not handled correctly is if the alert remains
in the output buffer (WANT_WRITE). Then it won't be sent, or will be
truncated. We'd need to either delay the application error or record
the write buffering notice; to be done later.
2017-05-10 17:31:02 +02:00
Gilles Peskine
8498cb3687 Cleaned up double variable declaration 2017-05-10 15:39:40 +02:00
Gilles Peskine
064a85ca48 Code portability fix 2017-05-10 10:46:40 +02:00
Hanno Becker
072e806a07 Adapt ChangeLog 2017-05-09 10:47:49 +01:00
Hanno Becker
b2034b7785 Fix potential stack underflow in mpi_read_file.
When provided with an empty line, mpi_read_file causes a numeric
underflow resulting in a stack underflow. This commit fixes this and
adds some documentation to mpi_read_file.
2017-05-09 10:29:06 +01:00
Hanno Becker
4bcb4914c5 Abort modular inversion when modulus is one.
The modular inversion function hangs when provided with the modulus 1. This commit refuses this modulus with a BAD_INPUT error code. It also adds a test for this case.
2017-05-08 14:47:04 +01:00
Gilles Peskine
e7e7650480 More length checks in RSA PKCS1v15 verify
Added one check that I'd missed, and made the style more uniform.
2017-05-04 12:48:39 +02:00
Gilles Peskine
0e17eb05f8 More length checks in RSA PKCS1v15 verify
Tighten ASN.1 parsing of RSA PKCS#1 v1.5 signatures, to avoid a
potential Bleichenbacher-style attack.
2017-05-03 18:56:10 +02:00
Gilles Peskine
36091fe3ef Added changelog entry 2017-05-03 16:55:03 +02:00
Gilles Peskine
1cc8e3472a Send TLS alerts in many more cases
The TLS client and server code was usually closing the connection in
case of a fatal error without sending an alert. This commit adds
alerts in many cases.

Added one test case to detect that we send the alert, where a server
complains that the client's certificate is from an unknown CA (case
tracked internally as IOTSSL-1330).
2017-05-03 16:28:34 +02:00