winlibs / libmcrypt Goto Github PK

Is it hard to implement this algorithm for a BMS?
It gives the error below.
Compilation failed due to following error(s).
9 | #include <mcrypt.h>
| ^~~~~~~~~~
compilation terminated.
Best

It seems that mcrypt has its own wrapper for calloc() and calls mlock() with each returned pointer. On regular OSs these extra mlock() calls can make the encryption or decryption operations take up to twice as long. On CentOS then this usage of mlock() can make encryption or decryption operations take 1000s of times longer... seems to be a pathological case, see [1]. The test program at [1] also shows that most malloc() returned pointers cannot be used by mlock(). So should the mlock() used in xmemory.c be deleted?

[1] "mlock() on CentOS is 100 to 520 times slower than the next slowest OS tested. Why?" http://bugs.centos.org/view.php?id=8954