xplat: examine _WIN64, should it be _M_X64?

[jianchun]

e.g. CallInfo.h, is following "unused" patch applicable to xplat x64?

        CallInfo(CallFlags flags, ushort count)
            : Flags(flags)
            , Count(count)
#ifdef _WIN64
            , unused(0)
#endif
        {
        }

        // Assumes big-endian layout
        // If the size of the count is changed, change should happen at following places also
        //  - scriptdirect.idl
        //  - LowererMDArch::LoadInputParamCount
        //
        unsigned  Count : 24;
        CallFlags Flags : 8;
#ifdef _WIN64
        unsigned unused : 32;
#endif

[obastemur]

How about _M_X64_OR_ARM64 instead of _M_X64 ? Especially for the sample you've shared.

[jianchun]

Yes, that may be better. Or maybe a new symbol BIT64.

[LouisLaf]

I don't like _M_X64 of _M_X64_OR_ARM64 as these are specific CPUs is the defines are just about 64 targets. BIT64 is better but not that descriptive. How about TARGET64 or HOST64?

[jianchun]

I don't like _M_X64 of _M_X64_OR_ARM64 as these are specific CPUs is the defines are just about 64 targets. BIT64 is better but not that descriptive. How about TARGET64 or HOST64?

+1 TARGET64

[jianchun]

We now have TARGET_64. But lots of code (including new code) still uses _WIN64. Needs to search for all and evaluate which should change to TARGET_64.