Improve documentation of memory_cleanse()

So far, the documentation of memory_cleanse() is a verbatim copy of
the commit message in BoringSSL, where this code was originally
written. However, our code evolved since then, and the commit message
is not particularly helpful in the code but is rather of historical
interested in BoringSSL only.

This commit improves improves the comments around memory_cleanse()
and gives a better rationale for the method that we use. This commit
touches only comments.

src/support/cleanse.cpp

@@ -11,33 +11,25 @@
 #include <Windows.h> // For SecureZeroMemory.
 #endif
 
-/* Compilers have a bad habit of removing "superfluous" memset calls that
- * are trying to zero memory. For example, when memset()ing a buffer and
- * then free()ing it, the compiler might decide that the memset is
- * unobservable and thus can be removed.
- *
- * Previously we used OpenSSL which tried to stop this by a) implementing
- * memset in assembly on x86 and b) putting the function in its own file
- * for other platforms.
- *
- * This change removes those tricks in favour of using asm directives to
- * scare the compiler away. As best as our compiler folks can tell, this is
- * sufficient and will continue to be so.
- *
- * Adam Langley <agl@google.com>
- * Commit: ad1907fe73334d6c696c8539646c21b11178f20f
- * BoringSSL (LICENSE: ISC)
- */
 void memory_cleanse(void *ptr, size_t len)
 {
 #if defined(_MSC_VER)
+    /* SecureZeroMemory is guaranteed not to be optimized out by MSVC. */
     SecureZeroMemory(ptr, len);
 #else
     std::memset(ptr, 0, len);
 
-    /* As best as we can tell, this is sufficient to break any optimisations that
-       might try to eliminate "superfluous" memsets. If there's an easy way to
-       detect memset_s, it would be better to use that. */
+    /* Memory barrier that scares the compiler away from optimizing out the memset.
+     *
+     * Quoting Adam Langley <agl@google.com> in commit ad1907fe73334d6c696c8539646c21b11178f20f
+     * in BoringSSL (ISC License):
+     *    As best as we can tell, this is sufficient to break any optimisations that
+     *    might try to eliminate "superfluous" memsets.
+     * This method is used in memzero_explicit() the Linux kernel, too. Its advantage is that it
+     * is pretty efficient because the compiler can still implement the memset() efficiently,
+     * just not remove it entirely. See "Dead Store Elimination (Still) Considered Harmful" by
+     * Yang et al. (USENIX Security 2017) for more background.
+     */
     __asm__ __volatile__("" : : "r"(ptr) : "memory");
 #endif
 }

src/support/cleanse.h

@@ -8,7 +8,8 @@
 
 #include <stdlib.h>
 
-// Attempt to overwrite data in the specified memory span.
+/** Secure overwrite a buffer (possibly containing secret data) with zero-bytes. The write
+ * operation will not be optimized out by the compiler. */
 void memory_cleanse(void *ptr, size_t len);
 
 #endif // BITCOIN_SUPPORT_CLEANSE_H