Fwd: Fix AES and incorrect argument to _freea() under Microsoft compilers (PR 306, Issue 302)

[Jeffrey Walton]

FYI...

---------- Forwarded message ----------
From: Jeffrey Walton <nolo...@gmail.com>
Date: Sat, Sep 24, 2016 at 6:55 PM
Subject: Fix AES and incorrect argument to _freea() under Microsoft
compilers (PR 306, Issue 302)
To: <redacted>

Hi Everyone,

John analyzed the issue further and discovered some more facets for
the problem. Additional analysis is at (I was not aware Microsoft had
an _ALLOCA_S_THRESHOLD define):

* https://github.com/weidai11/cryptopp/pull/301#issuecomment-249390026

We merged his Pull Request at:

* https://github.com/weidai11/cryptopp/commit/31e776d4e7aaaa9e6171cf34f7eb0dc5e0a57610

As soon as I finish with DLIES and ECIES I will turn towards AES. I
think we can move the byte buffer to a class member and re-use it
across calls. Using a SecByteBlock for the temporary storage should
not violate the per-object safety guaranteed by the library.

Jeff