Upstream claims its faster if combined with an optimizing compiler and I can confirm that in some tests, so lets see how it works out in practice. Git-Dch: Ignore
@@ -50,3 +50,6 @@
#define APT_8_CLEANER_HEADERS
#define APT_9_CLEANER_HEADERS
#define APT_10_CLEANER_HEADERS
+
+/* unrolling is faster combined with an optimizing compiler */
+#define SHA2_UNROLL_TRANSFORM