Re: Current int & float overflow checking is slow.

On Mon, Oct 30, 2017 at 11:24 PM, Andres Freund <andres(at)anarazel(dot)de> wrote:
>> > Perhaps it should rather be pg_add_s32_overflow, or a similar
>> > naming scheme?
>>
>> Not sure what the s is supposed to be? Signed?
>
> Yes, signed. So we could add a u32 or something complementing the
> functions already in the patch. Even though overflow checks are a heck
> of a lot easier to write for unsigned ints, the intrinsics are still
> faster. I don't have any sort of strong feelings on the naming.

Right, I guess including the s is probably a good idea then.

>> I suggest that if we think we don't need -fwrapv any more, we ought to
>> remove it. Otherwise, we won't find out if we're wrong.
>
> I agree that we should do so at some point not too far away in the
> future. Not the least because we don't specify this kind of C dialect in
> a lot of other compilers. Additionally the flag causes some slowdown
> (because e.g. for loop variables are optimized less). But I'm fairly
> certain it needs a bit more care that I've invested as of now - should
> probably at least compile with -Wstrict-overflow=some-higher-level, and
> with ubsan. I'm fairly certain there's more bogus overflow checks
> around...

Makes sense.

--
Robert Haas
EnterpriseDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company