On Fri, Jun 8, 2012 at 6:45 PM, Xi Wang <xi.wang@xxxxxxxxx> wrote:
> On Jun 8, 2012, at 6:47 AM, Xi Wang wrote:
>> - oldvalue = cast_pseudo(ep, oldvalue, src->ctype, expr->ctype);
>> - opcode = opcode_sign(op_trans[expr->op - SPECIAL_BASE], src->ctype);
>> - dst = add_binary_op(ep, src->ctype, opcode, oldvalue, value);
>> - value = cast_pseudo(ep, dst, expr->ctype, src->ctype);
>> + opcode = opcode_sign(op_trans[expr->op - SPECIAL_BASE], expr->ctype);
>> + value = cast_pseudo(ep, value, src->ctype, expr->ctype);
>> + value = add_binary_op(ep, expr->ctype, opcode, oldvalue, value);
>
> For sparse-llvm, is it better to get rid of the cast_pseudo()
> at all?
>
> For example, what to emit for p += 1:
>
> 1) pointer p + integer 1,
>
> 2) pointer + pointer [cast from integer], or
>
> 3) (pointer)(integer [cast from pointer] + integer)?
Dunno.
This:
void *foo(void *p)
{
p += 1;
return p;
}
is converted as follows by clang:
define i8* @foo(i8* %p) nounwind uwtable readnone {
%1 = getelementptr i8* %p, i64 1
ret i8* %1
}
[ You can try it out yourself here: http://llvm.org/demo/index.cgi ]
So it seems that LLVM really wants to use the "getelementptr" thingy
for these. I have no idea what makes the most sense from sparse IR
point of view.
Jeff, Linus, Chris, hmm?
Pekka
--
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
