Re: [Tinycc-devel] Buggy left shift on x86_64

[grischka]

Thomas Preud'homme wrote:
> Thanks for the input. The issue seems to be in tccpp.c parse_number function. 
> The code seems to assume that UL is unsigned int. The code I'm refering to is:
>
>                 if (lcount == 2) {
>                     if (tok == TOK_CINT)
>                         tok = TOK_CLLONG;
>                     else if (tok == TOK_CUINT)
>                         tok = TOK_CULLONG;
>                 }
>
> We could add an ifdef statement to only do the lcount == 2 test if it's a 32 
> bit system but it seems really dirty. Or maybe targets could define the 
> programming model (LP64 & Co) and this could be use here.
>
> Grishka, what do you think is the best approach?

I think best approach is to do both in sequential time order:

1) Add the required dirtiness in order to make it do the right thing.
2) Take benefit from exposed cleanup potential regarding #ifdefs dealing
   with the target 'long' model

As for example tccgen.c:3050:

#if !defined TCC_TARGET_X86_64 || defined TCC_TARGET_PE
        t = (t & ~VT_BTYPE) | VT_INT;
#else
        t = (t & ~VT_BTYPE) | VT_LLONG;
#endif

Also with the 'long double' model, for example tccpe.c:1966:

#ifdef TCC_TARGET_PE
                tok = TOK_CDOUBLE;
                tokc.d = strtod(token_buf, NULL);
#else
                tok = TOK_CLDOUBLE;
                tokc.ld = strtold(token_buf, NULL);
#endif

Have Fun ;)

--- grischka