[OpenSSL-3.0LTS] TNF local patch - respect -Wcast-qual

[Takehiko NOZAKI]

TNF don’t like cast (const type *) to (type *) directly, and they often enables -Wcast-qual with -Werror.

they defines __UNCONST macro in sys/cdefs.h:

const type *a;
type *b;
a = __UNCONST(a);

but in this patch, they dont use it for portability(and maybe its implementation is broken future 128bit pointer era), so use intptr_t like:

const type *a;
type *b;
a = (type *)(intptr_t)a;

‌

[Takehiko NOZAKI]

in terms of C specification, cast const *type to type * is completely legal, but sometimes const qualifier is used for memory protection, see following examples taken from GCC’s manual:

  /* p is char ** value.  */
  const char **q = (const char **) p;
  /* Assignment of readonly string to const char * is OK.  */
  *q = "string";
  /* Now char** pointer points to read-only memory.  */
  **p = 'b';

so -Wcast-qual may help to detect assignment of read-only location error.

‌

but TNF, they blindly uses __UNCONST macro and never checks these memories can really be writable.

‌

so they work hard, but they got nothing, it is only same result -Wno-cast-qual how idiot!(same as F*BSD, they do as same by __DECONST macro)

‌

O*BSD keeps avoid to use -Wcast-qual, it seems that they believe -Wcast-qual is harmful, i think.

[Takehiko NOZAKI]

and OpenSSL’s coder is completely suck:

I don't understand C anything at all, and I wrote const qualifier in the mood…

this is why so many OpenSSL fork exists such as LibreSSL/BoringSSL.

[Takehiko NOZAKI]

it seems that TNF abandoned to pass -Wcast-qual error, they enables -Wno-error=cast-qual already. http://cvsweb.netbsd.org/bsdweb.cgi/src/crypto/Makefile.openssl?only_with_tag=MAIN#rev1.14