When compiling under x32, __uint64_t-variables don't want to shift by 32 bits, what's wrong?

Question

linux

The variable is not shifted to values ​​< 32, but when the shift is stronger, the value is reset to zero. In fact, the variable behaves like a 32-bit variable, but I declare it as unsigned long long int.
What is the reason for this behavior?

Answer 1

Here the problem is very tricky. both arguments must be 64bit.
For example:
__int64 a = 1ull << 33; // will work,
__int64 b = 1 << 33; // zero out

Answer 2

In x32, regardless of the type of the arguments, byte shift always returns a type under the size of the register. So if the value has gone beyond x32, then that's it, although an assembler insert can help.