Date: Fri, 16 Sep 2016 08:55:29 -0700 (PDT) From: Jeffrey Walton <noloader@...il.com> To: oss-security@...ts.openwall.com Subject: Fwd: CVE-2016-7420 and dev-brach 'trap' ready for testing The improved code should be available in Master in the next few days. After about three or four weeks we will release Crypto++ 5.6.5. The decision to release Crypto++ 5.6.5 was driven by Error Reporting services like Apport, CrashReporter and WER; and company's like Apple, Google and Microsoft's cooperation with governments to mine sensitive information. On Friday, September 16, 2016 at 11:51:36 AM UTC-4, Jeffrey Walton wrote: > > Hi Everyone, > > CVE-2016-7420 caused us to cut-in CRYPTOPP_ASSERT a little earlier than > expected. <trap.h> and CRYPTOPP_ASSERT have existed in Master for over a > year. We set up a dev-branch called 'trap' to isolate the cut-in during > testing. > > The cut-over to CRYPTOPP_ASSERT occurred at > https://github.com/weidai11/cryptopp/commit/399a1546de71f41598c15edada28e7f0d616f541 > . It tested OK under modern versions of Clang, CGG, Solaris and Visual > Studio. > > The defining factor of CRYPTOPP_ASSERT is it abandons Posix NDEBUG, which > we used to rely upon to remove asserts. We switched strategies, and now we > enable CRYPTOPP_ASSERT if any the following are defined: CRYPTOPP_DEBUG, > DEBUG, _DEBUG. This strategy side steps bad release/production > configurations due to policy (Debian never defines NDEBUG) and > errors/omissions (users or Autotools or CMake or Eclipse <other build > system> fails to define NDEBUG). > > CRYPTOPP_ASSERT also adds a nice feature: it raises SIGTRAP rather than > SIGABRT. SIGABRT will snap the debugger, if present. And it won't follow > Posix's idiotic footsteps and crash the program with a SIGABRT while a > developer is debugging it. > > The last two, DEBUG and _DEBUG, are set in Visual Studio projects by > Microsoft; and they cause CRYPTOPP_DEBUG to be set automatically. BSD, > Linux, Solaris and Unix user will have to -DCRYPTOPP_DEBUG=1 or uncomment > CRYPTOPP_DEBUG in config.h. > > If all goes well with testing, then we will merge Trap dev-branch into > Master this weekend or early next week. Our test script takes two or three > days to run on IoT gadets like BeableBoards and CubieTrucks, so the > earliest we can merge will be late Saturday or Sunday. > > Jeff > Content of type "text/html" skipped
Powered by blists - more mailing lists
Please check out the Open Source Software Security Wiki, which is counterpart to this mailing list.
Confused about mailing lists and their use? Read about mailing lists on Wikipedia and check out these guidelines on proper formatting of your messages.