misuse of compare_testvector

[fperrad]

compare_testvector has two different definitions see https://github.com/libtom/libtomcrypt/blob/develop/src/headers/tomcrypt_misc.h#L102-L109
When the macro definition is used, the comparison compare_testvector() != 0 is wrong.
When the function definition is used, the comparison is useless.
So, this commit removes the comparison.

[sjaeckel]

very true

[sjaeckel]

I just thought would it make sense to add this patch instead?

diff --git a/src/headers/tomcrypt_misc.h b/src/headers/tomcrypt_misc.h
index 91f87c5..76f4f6b 100644
--- a/src/headers/tomcrypt_misc.h
+++ b/src/headers/tomcrypt_misc.h
@@ -105,7 +105,7 @@ void print_hex(const char* what, const void* v, const unsigned long l);
 int compare_testvector(const void* is, const unsigned long is_len, const void* should, const unsigned long should_len, const
 #else
 #define compare_testvector(is, is_len, should, should_len, what, which) \
-   (((is_len) != (should_len)) || (XMEMCMP((is), (should), (is_len)) != 0))
+   ((((is_len) != (should_len)) || (XMEMCMP((is), (should), (is_len)) != 0)) ? 1 : 0)
 #endif
 
 /* $Source$ */

[karel-m]

The macro compare_testvector should be fixed as proposed by @sjaeckel anyway.

Using if (compare_testvector(..)) vs. if(compare_testvector(..) != 0) is just a matter of taste. Its nature is similar to XMEMCMP which we use in both ways if (XMEMCMP(..)) as well as if (XMEMCMP(..) != 0). I think we can leave out != 0 in these cases.