Copy framesize_t (4 bytes) to nextsize (8 bytes) then shift 32 bits to zero out
the most significant 32 bits.
Use of #ifdef __LP64__ suggested by @brynet.
memcpy((char *)&packeta[1] - packeta->dwSpaceLeft, packet, dwSize);
Source/msg.cpp does this calculation, which will be wrong if any padding is
added. To get this to work correctly, pack the structures, as suggested by
@AJenbo.
Also, define TMEGAPKT_SIZE because pNext can be 4 or 8 bytes.
The following build error was observed when compiling with Clang:
In file included from Source/appfat.cpp:1:
In file included from ./Source/diablo.h:5:
In file included from ./Source/../types.h:52:
In file included from ./SourceS/miniwin.h:19:
In file included from /usr/lib/clang/8.0.1/include/x86intrin.h:29:
In file included from /usr/lib/clang/8.0.1/include/immintrin.h:28:
/usr/lib/clang/8.0.1/include/mmintrin.h:81:40: error: cannot initialize a parameter of type
'__attribute__((__vector_size__(2 * sizeof(int)))) int' (vector of 2 'int' values) with an rvalue of type
'__v2si' (aka 'int')
return __builtin_ia32_vec_ext_v2si((__v2si)__m, 0);
^~~~~~~~~~~
Clang version:
$ clang -v
clang version 8.0.1 (tags/RELEASE_801/final)
First I tried to replace the x86intrin.h include with:
static inline unsigned int _rotl(unsigned int value, int shift)
{
return (value << shift) | (value >> (32 - shift));
}
static inline unsigned int _rotr(unsigned int value, int shift)
{
return (value >> shift) | (value << (32 - shift));
}
But that resulted in an error stating that _rotl and _rotr were
predeclared in Clang:
./SourceS/miniwin.h:20:28: error: definition of builtin function '_rotl'
static inline unsigned int _rotl(unsigned int value, int shift)
Thus, we simply declare but don't define these functions when
using Clang.
Anders Jenbo
Nam Nguyen
qndel
Robin Eklind
jimmy-cell
Robin Eklind