aboutsummaryrefslogtreecommitdiffstats
path: root/src
Commit message (Collapse)AuthorAgeFilesLines
...
* Another FFI struct vs class fixJack Lloyd2017-09-301-1/+1
|
* Retract explicit on OIDJack Lloyd2017-09-301-1/+1
| | | | This conversion is often useful
* FFI objects have to be structs because we expose them to CJack Lloyd2017-09-302-5/+4
|
* Use explicit on more single-argument constructorsJack Lloyd2017-09-3014-20/+23
|
* Avoid bogus declarations in GMACJack Lloyd2017-09-301-25/+3
| | | | | These functions are declared/defined on MessageAuthenticationCode, and call the start_msg virtual.
* More fiddling with SonarJack Lloyd2017-09-301-1/+1
|
* Address some MSVC warningsJack Lloyd2017-09-308-135/+140
|
* Use class instead of struct for objects with member functionsJack Lloyd2017-09-3014-157/+184
| | | | Flagged by Sonar and quite reasonable
* Attempt at fixing Sonar exclusionsJack Lloyd2017-09-301-1/+2
|
* Add annotation so GCC/Clang/MSVC know it is an allocation function.Jack Lloyd2017-09-302-1/+12
|
* Merge GH #1231 Hide secure_allocator allocate in a functionJack Lloyd2017-09-305-22/+55
|\
| * Missing include, noticed by OS XJack Lloyd2017-09-291-0/+1
| |
| * In secure_allocator, hide mlock/new usage in a function in mem_opsJack Lloyd2017-09-295-22/+54
| | | | | | | | | | | | | | Switch to calloc/free instead of new/delete - shouldn't matter since we are only allocate integral types. This change reduces the size of libbotan-2.so by ~300 Kb on my system.
* | Thinko fixJack Lloyd2017-09-301-1/+1
| |
* | Compat typedef PBKDF -> S2KJack Lloyd2017-09-291-1/+12
|/ | | | [ci skip]
* Windows!!!!!!!Jack Lloyd2017-09-291-1/+1
|
* Work around some MSVC 2013 sillynessJack Lloyd2017-09-291-1/+3
|
* Ensure files are closed before trying to remove itJack Lloyd2017-09-291-0/+4
| | | | Windows ...
* Avoid using std::rand in test codeJack Lloyd2017-09-291-2/+4
| | | | It makes static analyzers mad
* Avoid throwing in pool allocator deallocation pathJack Lloyd2017-09-291-8/+1
| | | | | | std::terminate can ruin your day Coverity find
* Check return value of std::remove in testJack Lloyd2017-09-291-2/+5
| | | | Coverity find
* Further Travis matrix fiddlingJack Lloyd2017-09-291-3/+3
| | | | | The fuzzer build is still in the critical path, everything else still finished before it.
* Fiddle with Travis test ordering a bitJack Lloyd2017-09-291-6/+6
| | | | | Having several long tests at the end means we end up not being able to make good use of Travis running 5 builds for us in parallel.
* Fix test error if mceies disabledJack Lloyd2017-09-291-3/+5
| | | | [ci skip]
* Merge GH #1229 Simplify allocator and avoid MSVC perf issueJack Lloyd2017-09-291-37/+5
|\
| * Add back size_type typedefJack Lloyd2017-09-291-0/+1
| | | | | | | | | | | | | | Lacking this seems to cause interesting issues with Apple Clang on 32-bit ARM. It seems like it implicitly defines a size_type that is the same size as size_t, but not actually size_t, so we get an unsigned long vs unsigned int mismatch on the type.
| * Of course MSVC 2013 has to be specialJack Lloyd2017-09-291-0/+7
| |
| * Simplifiy secure_allocatorJack Lloyd2017-09-291-44/+4
| | | | | | | | | | | | | | | | | | According to https://howardhinnant.github.io/allocator_boilerplate.html we don't need most of what was in there in C++11 and later. I think I originally wrote that code referencing a C++03 doc. Specifically avoiding construct, destruct prevents a performance issue in MSVC (GH #1228)
* | Avoid crash in DLIES tests if AEADs disabledJack Lloyd2017-09-291-0/+6
| |
* | Remove sanitizer and mini-static builds from TravisJack Lloyd2017-09-291-6/+0
| | | | | | | | | | | | | | | | | | | | Sanitizer mode isn't testing anything extra vs the fuzzer tests which also use the sanitizers. Similarly mini-static is pretty well covered by the other minimized build and the static build. Will cut about 7-10 minutes from Travis build time.
* | Build test scriptJack Lloyd2017-09-291-3/+3
| | | | | | | | [ci skip]
* | Build script workJack Lloyd2017-09-291-11/+29
| | | | | | | | [ci skip]
* | Unused variableJack Lloyd2017-09-291-1/+1
| | | | | | | | [ci skip]
* | In TLS CBC padding check, only need to verify last 256 bytesJack Lloyd2017-09-292-10/+42
| | | | | | | | | | No reason to scan the entire record since it's assured at most 256 bytes of padding are used. Inspired by GH #1227
* | More futzing with the build em all scriptJack Lloyd2017-09-291-31/+36
|/ | | | [ci skip]
* Add a test of TLS CBC padding verificationJack Lloyd2017-09-294-9/+88
| | | | See also GH #1227
* Include pk_keys.h in pkcs8.h and x509_key.hJack Lloyd2017-09-292-2/+2
| | | | Not strictly required but seems likely to cause problems for applications.
* Fix fuzzer buildJack Lloyd2017-09-291-0/+1
|
* Require HMAC for HKDFJack Lloyd2017-09-281-0/+4
| | | | Not actually strictly required but in practice doesn't make much sense.
* Fixes for MSVCJack Lloyd2017-09-282-2/+2
| | | | BOTAN_UNUSED "uses" the RNG :/
* More build test scriptingJack Lloyd2017-09-281-1/+5
| | | | [ci skip]
* Expand build test to also cover single-module exclusionsJack Lloyd2017-09-281-9/+13
|
* Further header cleanupsJack Lloyd2017-09-2830-23/+67
|
* Add variants of PKCS8::load_key that don't require an RNG argumentJack Lloyd2017-09-282-32/+112
| | | | | It's not used and only there for compat with existing callers, but no reason we can't offer a version that doesn't require it.
* Disable clcache on AppVeyorJack Lloyd2017-09-282-19/+1
| | | | | | AppVeyor's sharding of the build cache across VM providers means the effective cache hit rate is basically zero. So using clcache just slows down the build without much benefit.
* Build the PDFJack Lloyd2017-09-281-19/+31
|
* Rewrite website generator script in PythonJack Lloyd2017-09-284-50/+144
|
* Doxygen commentJack Lloyd2017-09-281-0/+6
| | | | [ci skip]
* Header cleanupsJack Lloyd2017-09-285-5/+3
|
* Merge GH #1224 Address side channel in Montgomery exponentiationJack Lloyd2017-09-286-36/+98
|\
generated by cgit v1.2.3 (git 2.39.1) at 2024-12-05 03:37:49 +0000