[MLton] x86_64 port status

[Matthew Fluet]

> Here are all the warnings I see, grouped by type.  If anyone knows how
> to solve the problem by changing the code or how to locally turn off
> the warning, please speak up.
>
> --------------------------------------------------------------------------------
>
> basis/System/Date.c:62: warning: format not a string literal, format string not checked

We could disable this warning either by disabling -Wformat=2 or by passing 
-Wno-format for the compilation of Date.c.

> gc/int-inf.c:68: warning: right shift count >= width of type

I get this warning even if I turn off all command line warning flags;
We could eliminate the code, as its only there for configurations where 
the size of an ML object pointer is greater than the size of mp_limb_t, 
which seems unlikely.

> gc/mark-compact.c:18: warning: statement with no effect
> gc/mark-compact.c:30: warning: statement with no effect

I don't get these warnings with on my system.

> gdtoa/gdtoa.h:116: warning: redundant redeclaration of `strtof' in same scope
> /usr/include/stdlib.h:347: warning: previous declaration of `strtof'
> gdtoa/gdtoa.h:117: warning: redundant redeclaration of `strtod' in same scope
> /usr/include/stdlib.h:326: warning: previous declaration of `strtod'

We could patch them out, via the currently idempotent gdtoa-patch.

> gc/cheney-copy.c:13: warning: inlining failed in call to `updateWeaksForCheneyCopy'
> gc/cheney-copy.c:13: warning: inlining failed in call to `updateWeaksForCheneyCopy'

I also don't get these warnings on my system, although I needed to tweak 
the --param options to enable sufficient inlining.  As Henry noted, we 
could disable the warning (by not passing -Winline), but I note that gc.c 
in HEAD doesn't inline everything we request either.  The correct --param 
options also depend on the version of gcc and, presumably, the target 
architecture.