- Nov 07, 2016
-
-
David Benjamin authored
RFC 3447, section 8.2.2, steps 3 and 4 states that verifiers must encode the DigestInfo struct and then compare the result against the public key operation result. This implies that one and only one encoding is legal. OpenSSL instead parses with crypto/asn1, then checks that the encoding round-trips, and allows some variations for the parameter. Sufficient laxness in this area can allow signature forgeries, as described in https://www.imperialviolet.org/2014/09/26/pkcs1.html Although there aren't known attacks against OpenSSL's current scheme, this change makes OpenSSL implement the algorithm as specified. This avoids the uncertainty and, more importantly, helps grow a healthy ecosystem. Laxness beyond the spec, particularly in implementations which enjoy wide use, risks harm to the ecosystem for all. A signature producer which only tests against OpenSSL may not notice bugs and accidentally become widely deployed. Thus implementations have a responsibility to honor the specification as tightly as is practical. In some cases, the damage is permanent and the spec deviation and security risk becomes a tax all implementors must forever pay, but not here. Both BoringSSL and Go successfully implemented and deployed RSASSA-PKCS1-v1_5 as specified since their respective beginnings, so this change should be compatible enough to pin down in future OpenSSL releases. See also https://tools.ietf.org/html/draft-thomson-postel-was-wrong-00 As a bonus, by not having to deal with sign/verify differences, this version is also somewhat clearer. It also more consistently enforces digest lengths in the verify_recover codepath. The NID_md5_sha1 codepath wasn't quite doing this right. Reviewed-by:
Kurt Roeckx <kurt@roeckx.be> Reviewed-by:
Rich Salz <rsalz@openssl.org> GH: #1474 (cherry picked from commit 608a0264)
-
Matt Caswell authored
This partially reverts commit c636c1c4 . It also tweaks the documentation and comments in this area. On the client side the documented interface for SSL_CTX_set_verify()/SSL_set_verify() is that setting the flag SSL_VERIFY_PEER causes verfication of the server certificate to take place. Previously what was implemented was that if *any* flag was set then verification would take place. The above commit improved the semantics to be as per the documented interface. However, we have had a report of at least one application where an application was incorrectly using the interface and used *only* SSL_VERIFY_FAIL_IF_NO_PEER_CERT on the client side. In OpenSSL prior to the above commit this still caused verification of the server certificate to take place. After this commit the application silently failed to verify the server certificate. Ideally SSL_CTX_set_verify()/SSL_set_verify() could be modified to indicate if invalid flags were being used. However these are void functions! The simplest short term solution is to revert to the previous behaviour which at least means we "fail closed" rather than "fail open". Thanks to Cory Benfield for reporting this issue. Reviewed-by:
Richard Levitte <levitte@openssl.org> (cherry picked from commit c8e2f98c)
-
Matt Caswell authored
We read it later in grow_init_buf(). If CCS is the first thing received in a flight, then it will use the init_msg from the last flight we received. If the init_buf has been grown in the meantime then it will point to some arbitrary other memory location. This is likely to result in grow_init_buf() attempting to grow to some excessively large amount which is likely to fail. In practice this should never happen because the only time we receive a CCS as the first thing in a flight is in an abbreviated handshake. None of the preceding messages from the server flight would be large enough to trigger this. Reviewed-by:
-
- Nov 06, 2016
-
-
Richard Levitte authored
Reviewed-by:
Andy Polyakov <appro@openssl.org> (Merged from https://github.com/openssl/openssl/pull/1772) (cherry picked from commit 475592e2)
-
Richard Levitte authored
If zlib-dynamic was given but not --with-zlib-lib, LIBZ was defined to the empty string. Instead, give it the default "ZLIB1". Reviewed-by:
-
Richard Levitte authored
VMS only unloads shared libraries at process rundown, so tell the OpenSSL code so by pretending we linked with -znodelete. Reviewed-by:
-
- Nov 04, 2016
-
-
Richard Levitte authored
Since the local symbol table is looked up before the global symbol table, 'arch' assigned in the local symbol table of the DCL where MMS is called would be seen before the 'arch' defined in descrip.mms. Assigning it to the local symbol table in descrip.mms removes that issue. Reviewed-by:
-
Rich Salz authored
Reviewed-by:
-
Richard Levitte authored
Clang on Linux seems to catch things that we might miss otherwise. Also, throw in 'no-deprecated' to make sure we test that as well. Reviewed-by:
Matt Caswell <matt@openssl.org> (Merged from https://github.com/openssl/openssl/pull/1839) (cherry picked from commit 7b195438)
-
- Nov 03, 2016
-
-
Richard Levitte authored
test/shlibloadtest.c assumes all Unix style platforms use .so as shared library extension. This is not the case for Mac OS X, which uses .dylib. Instead of this, have the test recipe find out the extension from configuration data. Reviewed-by:
-
Rich Salz authored
Reviewed-by:
-
Richard Levitte authored
Pre 1.1.0, 'make test' would set the environment variable OPENSSL_DEBUG_MEMORY to "on". This got lost when translating the old build files to the new templates. This changes reintroduces that variable. Reviewed-by:
-
Richard Levitte authored
- Make sure to initialise SHLIB variables - Make sure to make local variables static Reviewed-by:
-
Richard Levitte authored
gcc is kinder, it silently passes quite a few flags to ld, while clang is stricter and wants them prefixed with -Wl, Reviewed-by:
-
Richard Levitte authored
Reviewed-by:
-
Kurt Roeckx authored
Reviewed-by:
Tim Hudson <tjh@openssl.org> Reviewed-by:
-
Mike Aizatsky authored
Reviewed-by:
-
- Nov 02, 2016
-
-
Matt Caswell authored
Rather than leaking a reference, just call GetModuleHandleEx and pin the module on Windows. Reviewed-by:
-
Matt Caswell authored
Instead of deliberately leaking a reference to ourselves, use nodelete which does this more neatly. Only for Linux at the moment. Reviewed-by:
-
Matt Caswell authored
This should demonstrate that the atexit() handling is working properly (or at least not crashing) on process exit. Reviewed-by:
-
Matt Caswell authored
Because we use atexit() to cleanup after ourselves, this will cause a problem if we have been dynamically loaded and then unloaded again: the atexit() handler may no longer be there. Most modern atexit() implementations can handle this, however there are still difficulties if libssl gets unloaded before libcrypto, because of the atexit() callback that libcrypto makes to libssl. The most robust solution seems to be to ensure that libcrypto and libssl never unload. This is done by simply deliberately leaking a dlopen() reference to them. Reviewed-by:
-
Matt Caswell authored
This works the same way as DSO_pathbyaddr() but instead returns a ptr to the DSO that contains the provided symbol. Reviewed-by:
-
Matt Caswell authored
Commit 3d8b2ec4 removed various unused functions. However now we need to use one of them! This commit resurrects DSO_pathbyaddr(). We're not going to resurrect the Windows version though because what we need to achieve can be done a different way on Windows. Reviewed-by:
-
Matt Caswell authored
Reviewed-by:
-
Matt Caswell authored
We should fail if we receive an unrecognised record type Reviewed-by:
-
Matt Caswell authored
TLS1.0 and TLS1.1 say you SHOULD ignore unrecognised record types, but TLS 1.2 says you MUST send an unexpected message alert. We swap to the TLS 1.2 behaviour for all protocol versions to prevent issues where no progress is being made and the peer continually sends unrecognised record types, using up resources processing them. Issue reported by 郭志攀 Reviewed-by:
-
Richard Levitte authored
Reviewed-by:
-
Sergey Bronnikov authored
CLA: trivial Reviewed-by:
Kurt Roeckx <kurt@openssl.org> Reviewed-by:
-
Matt Caswell authored
This test checks that read_ahead works correctly when dealing with large records. Reviewed-by:
-
Matt Caswell authored
The function ssl3_read_n() takes a parameter |clearold| which, if set, causes any old data in the read buffer to be forgotten, and any unread data to be moved to the start of the buffer. This is supposed to happen when we first read the record header. However, the data move was only taking place if there was not already sufficient data in the buffer to satisfy the request. If read_ahead is set then the record header could be in the buffer already from when we read the preceding record. So with read_ahead we can get into a situation where even though |clearold| is set, the data does not get moved to the start of the read buffer when we read the record header. This means there is insufficient room in the read buffer to consume the rest of the record body, resulting in an internal error. This commit moves the |clearold| processing to earlier in ssl3_read_n() to ensure that it always takes place. Reviewed-by:
-
Richard Levitte authored
Forks will have to define their own Reviewed-by:
-
- Nov 01, 2016
-
-
Benjamin Kaduk authored
Reviewed-by:
-
- Oct 31, 2016
-
-
Richard Levitte authored
Reviewed-by:
-
Richard Levitte authored
Fixes #1781 Reviewed-by:
-
Richard Levitte authored
VC-noCE-common and VC-WIN64-common were missing this line: template => 1, Fixes GH#1809 Reviewed-by:
-
- Oct 28, 2016
-
-
Matt Caswell authored
Replace the various length checks in the extension code with a macro to simplify the logic. Reviewed-by: -
Matt Caswell authored
The previous commit inspired a review of all the length checks for the extension adding code. This adds more robust checks and adds checks where some were missing previously. The real solution for this is to use WPACKET which is currently in master - but that cannot be applied to release branches. Reviewed-by: -
Matt Caswell authored
The status request extension did not correctly check its length, meaning that writing the extension could go 2 bytes beyond the buffer size. In practice this makes little difference because, due to logic in buffer.c the buffer is actually over allocated by approximately 5k! Issue reported by Guido Vranken. Reviewed-by: -
Matt Caswell authored
Providing a spkac file with no default section causes a double free. Thanks to Brian Carpenter for reporting this issue. Reviewed-by:
-
Matt Caswell authored
A BIO_read() 0 return indicates that a failure occurred that may be retryable. An SSL_read() 0 return indicates a non-retryable failure. Check that if BIO_read() returns 0, SSL_read() returns <0. Same for SSL_write(). The asyncio test filter BIO already returns 0 on a retryable failure so we build on that. Reviewed-by:
-
