What does a 64-bit system and 4GB RAM have to do with using 64bit timestamps?

32bit systems can use 64bit values without issue. In fact, even an 8bit system can handle 256bit values or even longer ones without issue.

The bittiness of a CPU and its address space have nothing to do with the length of usable data unless you end up with data longer than the RAM volume (and even then there's swap).

32-bit systems could only handle 4 GB of RAM

I don’t understand why people always say that. Pentium Pro could handle 64 GB even though it was a 32 bit CPU. It had a 36 bit address bus. Later models are the same.

The only issue would be critical issues that run on 32-bit systems and must be fixed before that date.

So, many banks and government agencies which still run on mainframes...

No, those can have 64 bit value as well, this is only a problem for applications which haven't switched to using them.

It's a problem with the internal represensation of a C/C++ type alias called time_t, mostly. That's the thing that holds the number of elapsed seconds since midnight on Jan. 1, 1970, which is the most common low-level representation of date and time on computers. In theory, time_t could point to a 32-bit type even on a 64-bit system, but I don't think anyone's actually dumb enough to do that. It affects more than C/C++ code because most programming languages end up calling C libraries once you go down enough levels.

In other words, there's no way you can tell whether a given application is affected or not unless you're aware of the code details, regardless of the bitness of the program and the processor it's running on.

It only depends whether the app and its OS/kernel interface use a 32-bit value to store the time information.

32-bit architecture or OS has nothing to do with this bug, for example 16-bit architectures must've used 32-bit time, too (otherwise they'd be able to only count up to 32-65 seconds).

500 Internal Brain Error