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.
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).
Separates platform time abstraction into it's own header from the
general platform abstraction as both depend on different build options.
(MBEDTLS_PLATFORM_C vs MBEDTLS_HAVE_TIME)
The server code parses the client hello extensions even when the
protocol is SSLv3 and this behaviour is non compliant with rfc6101.
Also the server sends extensions in the server hello and omitting
them may prevent interoperability problems.
By looking just at that test, it looks like 2 + dn_size could overflow. In
fact that can't happen as that would mean we've read a CA cert of size is too
big to be represented by a size_t.
However, it's best for code to be more obviously free of overflow without
having to reason about the bigger picture.
* development: (73 commits)
Bump yotta dependencies version
Fix typo in documentation
Corrected misleading fn description in ssl_cache.h
Corrected URL/reference to MPI library
Fix yotta dependencies
Fix minor spelling mistake in programs/pkey/gen_key.c
Bump version to 2.1.2
Fix CVE number in ChangeLog
Add 'inline' workaround where needed
Fix references to non-standard SIZE_T_MAX
Fix yotta version dependencies again
Upgrade yotta dependency versions
Fix compile error in net.c with musl libc
Add missing warning in doc
Remove inline workaround when not useful
Fix macroization of inline in C++
Changed attribution for Guido Vranken
Merge of IOTSSL-476 - Random malloc in pem_read()
Fix for IOTSSL-473 Double free error
Fix potential overflow in CertificateRequest
...
Conflicts:
include/mbedtls/ssl_internal.h
library/ssl_cli.c
This bug becomes noticeable when the extension following the "supported point
formats" extension has a number starting with 0x01, which is the case of the
EC J-PAKE extension, which explains what I noticed the bug now.
This will be immediately backported to the stable branches,
see the corresponding commits for impact analysis.
This is more consistent, as it doesn't make any sense for a user to be able to
set up an EC J-PAKE password with TLS if the corresponding key exchange is
disabled.
Arguably this is what we should de for other key exchanges as well instead of
depending on ECDH_C etc, but this is an independent issue, so let's just do
the right thing with the new key exchange and fix the other ones later. (This
is a marginal issue anyway, since people who disable all ECDH key exchange are
likely to also disable ECDH_C in order to minimize footprint.)
There is only one length byte but for some reason we skipped two, resulting in
reading one byte past the end of the extension. Fortunately, even if that
extension is at the very end of the ClientHello, it can't be at the end of the
buffer since the ClientHello length is at most SSL_MAX_CONTENT_LEN and the
buffer has some more room after that for MAC and so on. So there is no
buffer overread.
Possible consequences are:
- nothing, if the next byte is 0x00, which is a comment first byte for other
extensions, which is why the bug remained unnoticed
- using a point format that was not offered by the peer if next byte is 0x01.
In that case the peer will reject our ServerKeyExchange message and the
handshake will fail.
- thinking that we don't have a common point format even if we do, which will
cause us to immediately abort the handshake.
None of these are a security issue.
The same bug was fixed client-side in fd35af15
The Thread spec says we need those for EC J-PAKE too.
However, we won't be using the information, so we can skip the parsing
functions in an EC J-PAKE only config; keep the writing functions in order to
comply with the spec.
May happen with a faulty configuration (eg no allowed curve but trying to use
ECDHE key exchange), but not trigger able remotely.
(Found with Clang's scan-build.)
While at it, fix the following:
- on server with RSA_PSK, we don't want to set flags (client auth happens via
the PSK, no cert is expected).
- use safer tests (eg == OPTIONAL vs != REQUIRED)
A simple series of sed invocations.
This is the first step, purely internal changes. The conf substructure is not
ready to be shared between contexts yet.
* commit 'ce60fbe':
Fix potential timing difference with RSA PMS
Update Changelog for recent merge
Added more constant-time code and removed biases in the prime number generation routines.
Conflicts:
library/bignum.c
library/ssl_srv.c
Note from future self: actually md_init_ctx will be re-introduced with the
same signature later, and a new function with the additional argument will be
added.
- more freedom for us to change it in the future
- enforces hygiene
- performance impact of making accessors no longer inline should really be
negligible
Outline:
if( condition )
foo = value;
/* stuff that does not change condition */
if( condition )
/* use foo */
else
/* don't use foo */
For some reason, it appears to only kick in with -Os with gcc 4.9.3
* mbedtls-1.3:
Rename website and repository
Move private macro from header to C file
Add some missing 'static' on a few objects
Fix whitespace issues
Minor portability fix in benchmark
* development: (100 commits)
Update Changelog for the mem-measure branch
Fix issues introduced when rebasing
Fix compile error in memory_buffer_alloc_selftest
Code cosmetics
Add curve25519 to ecc-heap.sh
Add curve25519 to the benchmark program
Fix compile issue when buffer_alloc not available
New script ecc-heap.sh
Fix unused variable issue in some configs
Rm usunused member in private struct
Add heap usage for PK in benchmark
Use memory_buffer_alloc() in benchmark if available
Only define mode_func if mode is enabled (CBC etc)
PKCS8 encrypted key depend on PKCS5 or PKCS12
Disable SRV_C for client measurement
Output stack+heap usage with massif
Enable NIST_OPTIM by default for config-suite-b
Refactor memory.sh
Adapt memory.sh to config-suite-b
Adapt mini-client for config-suite-b.h
...
Conflicts:
ChangeLog
include/polarssl/net.h
library/Makefile
library/error.c
library/ssl_tls.c
programs/Makefile
programs/ssl/ssl_client2.c
programs/ssl/ssl_server2.c
tests/Makefile
Some people recommend using bit operations to avoid the compiler producing a
branch on `ret != 0`, but:
- this makes the code less readable,
- here I got a warning from some compilers about unsigned unary minus
- and anyway modern compilers don't produce a branch here, checked on x64 and
arm with various -O values.
* development: (46 commits)
Fix url again
Fix small bug in base64_encode()
Fix depend that was checked but not documented
Fix dependency that was not checked
Minor gitginore fixes
Move some ignore patterns to subdirectories
Ignore CMake/MSVC-related build files.
Re-categorize changelog entry
Fix misattribution
Minor nits with stdout/stderr.
Add cmake compatibility targets
Add script for polarssl symlink creation
Fix more stdio inclusion issues
Add debug info for cert/suite selection
Fix possible portability issue
Fix bug in ssl_get_verify_result()
aescrypt2.c local char array not initial
Update Changelog
Fix mips64 bignum implementation
Fix usage string of ssl_client2
...
Conflicts:
include/polarssl/ssl.h
library/CMakeLists.txt
library/Makefile
programs/Makefile
programs/ssl/ssl_client2.c
programs/ssl/ssl_server2.c
visualc/VS2010/PolarSSL.sln
visualc/VS2010/mbedTLS.vcxproj
visualc/VS6/mbedtls.dsp
visualc/VS6/mbedtls.dsw
* development:
Update copyright
Fix issue in compat.sh
Rename doxyfile
Rename to mbed TLS in tests/
Rename to mbed TLS in examples
Remove old test certificates.
Rename to mbed TLS in the documentation/comments
Change name to mbed TLS in the copyright notice
Conflicts:
doxygen/input/doc_mainpage.h
doxygen/mbedtls.doxyfile
include/polarssl/version.h
tests/compat.sh
* development:
Adapt tests to new defaults/errors.
Fix typos/cosmetics in Changelog
Disable RC4 by default in example programs.
Add ssl_set_arc4_support()
Set min version to TLS 1.0 in programs
Conflicts:
include/polarssl/ssl.h
library/ssl_cli.c
library/ssl_srv.c
tests/compat.sh
* commit '36adc36':
Add support for getrandom()
Use library default for trunc-hmac in ssl_client2
Make truncated hmac a runtime option server-side
Fix portability issue in script
Specific error for suites in common but none good
Prefer SHA-1 certificates for pre-1.2 clients
Some more refactoring/tuning.
Minor refactoring
Conflicts:
include/polarssl/error.h
include/polarssl/ssl.h
library/error.c
* commit 'f6080b8':
Fix warning in reduced configs
Adapt to "negative" switch for renego
Add tests for periodic renegotiation
Make renego period configurable
Auto-renegotiate before sequence number wrapping
Update Changelog for compile-option renegotiation
Switch from an enable to a disable flag
Save 48 bytes if SSLv3 is not defined
Make renegotiation a compile-time option
Add tests for renego security enforcement
Conflicts:
include/polarssl/ssl.h
library/ssl_cli.c
library/ssl_srv.c
library/ssl_tls.c
programs/ssl/ssl_server2.c
tests/ssl-opt.sh
* commit 'd7e2483': (57 commits)
Skip signature_algorithms ext if PSK only
Fix bug in ssl_client2 reconnect option
Cosmetics in ssl_server2
Improve debugging message.
Fix net_usleep for durations greater than 1 second
Use pk_load_file() in X509
Create ticket keys only if enabled
Fix typo in #ifdef
Clarify documentation a bit
Fix comment on resumption
Update comment from draft to RFC
Use more #ifdef's on CLI_C and SRV_C in ssl_tls.c
Add recursion.pl to all.sh
Allow x509_crt_verify_child() in recursion.pl
Set a compile-time limit to X.509 chain length
Fix 3DES -> DES in all.sh (+ time estimates)
Add curves.pl to all.sh
Rework all.sh to use MSan instead of valgrind
Fix depends on individual curves in tests
Add script to test depends on individual curves
...
Conflicts:
CMakeLists.txt
programs/ssl/ssl_client2.c
Rationale: if people want to disable RC4 but otherwise keep the default suite
list, it was cumbersome. Also, since it uses a global array,
ssl_list_ciphersuite() is not a convenient place. So the SSL modules look like
the best place, even if it means temporarily adding one SSL setting.
Reading the documentation of ssl_set_truncated_hmac() may give the impression
I changed the default for clients but I didn't, the old documentation was
wrong.
* etm:
Fix warning in reduced config
Update Changelog for EtM
Keep EtM state across renegotiations
Adjust minimum length for EtM
Don't send back EtM extension if not using CBC
Fix for the RFC erratum
Implement EtM
Preparation for EtM
Implement initial negotiation of EtM
Conflicts:
include/polarssl/check_config.h
* session-hash:
Update Changelog for session-hash
Make session-hash depend on TLS versions
Forbid extended master secret with SSLv3
compat.sh: allow git version of gnutls
compat.sh: make options a bit more robust
Implement extended master secret
Add negotiation of Extended Master Secret
Conflicts:
include/polarssl/check_config.h
programs/ssl/ssl_server2.c
Dummy fixed content for now.
Also, seems to be a race condition in the way the socket is closed and
reopened, leading to a few "random" failures in compat.sh. A fix is planned
for later.
