Potential uninitialized memory access in mesalink_BIO_new_mem_buf

[shinmao]

Hi, I found that the function mesalink_BIO_new_mem_buf might incorrectly assume buf_ptr points to any type that has non-zero length (all the example cases also show the assumption); however, it ignores the case that buf_ptr could also point to struct type.

mesalink/src/libcrypto/bio.rs

Lines 720 to 733 in ef12dd6

	pub extern "C" fn mesalink_BIO_new_mem_buf<'a>(
	buf_ptr: *mut c_void,
	len: c_int,
	) -> *mut MESALINK_BIO<'a> {
	if buf_ptr.is_null() {
	return ptr::null_mut();
	}
	let buflen = if len < 0 {
	unsafe { libc::strlen(buf_ptr as *const c_char) }
	} else {
	len as usize
	};
	let buf_ptr = buf_ptr as *mut u8;
	let buf = unsafe { slice::from_raw_parts_mut(buf_ptr, buflen) };

Even in C, the struct could also contain padding bytes, which means the slice created at line 733 might point to uninitialized padding bytes. We consider that the function could add pre-condition check to make sure buf_ptr is kind of plain old data.