Hi Chris, On 17 August 2017 at 08:19, Christopher Li <sparse@xxxxxxxxxxx> wrote: > On Wed, Aug 16, 2017 at 11:34 AM, Luc Van Oostenryck > <luc.vanoostenryck@xxxxxxxxx> wrote: >> Here by 'simple' we really mean 'worth of putting in a register'. >> >> We select all scalar types as well as struct and unions with a >> size not bigger than a long (register). >> >> Global and static variables, variables which address have been >> taken and volatiles variables are never selected. >> >> Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@xxxxxxxxx> >> --- >> symbol.h | 33 +++++++++++++++++++++++++++++++++ >> 1 file changed, 33 insertions(+) >> >> diff --git a/symbol.h b/symbol.h >> index 327449611..3f075c5bc 100644 >> --- a/symbol.h >> +++ b/symbol.h >> @@ -407,6 +407,39 @@ static inline int is_scalar_type(struct symbol *type) >> return 0; >> } >> >> +static inline int is_simple_type(struct symbol *type) > ... >> + case SYM_STRUCT: >> + case SYM_UNION: >> + return type->bit_size <= long_ctype.bit_size; > > I think the function name should be some thing along the > line of "type fits register" or is_registerable_type. > > union and struct are composite types. It is a bit > confusing calling them simple, if some one just looking > at the code without look at the commit message. > Apart from that I faced a problem with this and had to disable STRUCT/UNION from being treated as simple types. The problem was when generating LLVM output for https://github.com/dibyendumajumdar/dmr_c/blob/master/tests/eic/testcbrt.c. I will investigate more and report back my findings. Meanwhile I would be interested to see the results of a test with above. Regards Dibyendu -- To unsubscribe from this list: send the line "unsubscribe linux-sparse" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
