I've been working to diagnose an issue with an old game (mid 90s) that does not run properly on modern hardware. The issue arises when using a CPU clocked above 4.2Ghz which causes unsigned integers to roll over. The function they use in their code to call the system's real-time clock DOES return a 64-bit integer but then elsewhere in the code they do brilliant things like this:

Code:
int64 clock();

int old = clock();
..
(unsigned int)clock() - old <= something
There are hundreds of calls to the clock function which use the wrong type so I don't see it being possible to patch these data types. But would it be possible instead to patch the code to divide the clock by some number so that it will fit into a signed 32-bit integer?

I've dealt with modifying code by altering or NOPing instructions but this here looks as though I'll need to add new instructions; something I've not done before. Is this possible?

This is what I have to work with:

Code:
push   ebp
mov   ebp,esp
push   ecx
push   ecx
and   dword ptr [ebp-08h],00000000h
and   dword ptr [ebp-04h],00000000h
push   esi
rdtsc
mov   [ebp-08h],eax
mov   [ebp-04h],edx
mov   ecx,[ebp-08h]
mov   edx,[ebp-04h]
xor   eax,eax
xor   esi,esi
or   eax,ecx
or   edx,esi
sub   eax,[L00988C60]
pop   esi
sbb   edx,[L00988C64]
leave
retn
Thanks.