Merge #818: Add static assertion that uint32_t is unsigned int or wider

c0041b5cfc Add static assertion that uint32_t is unsigned int or wider (Tim Ruffing)

Pull request description:

  Solves one item in #792 .

ACKs for top commit:
  sipa:
    utACK c0041b5cfc
  elichai:
    ACK c0041b5cfc

Tree-SHA512: 9f700e89be39e15983260da94642593d16b9c437171e10377837ac73731ca7ba5dd7e328b3d93d0a24d143fb9e73abd11c578f6b58e2f94c82b783e977173b0c

src/assumptions.h

@@ -7,6 +7,8 @@
 #ifndef SECP256K1_ASSUMPTIONS_H
 #define SECP256K1_ASSUMPTIONS_H
 
+#include <limits.h>
+
 #include "util.h"
 
 /* This library, like most software, relies on a number of compiler implementation defined (but not undefined)
@@ -19,7 +21,11 @@ struct secp256k1_assumption_checker {
        allowed. */
     int dummy_array[(
         /* Bytes are 8 bits. */
-        CHAR_BIT == 8 &&
+        (CHAR_BIT == 8) &&
+
+        /* No integer promotion for uint32_t. This ensures that we can multiply uintXX_t values where XX >= 32
+           without signed overflow, which would be undefined behaviour. */
+        (UINT_MAX <= UINT32_MAX) &&
 
         /* Conversions from unsigned to signed outside of the bounds of the signed type are
            implementation-defined. Verify that they function as reinterpreting the lower