Undefined behavior or compiler bug?

Undefined behavior or compiler bug?

[Tibor Billes]

Hi,

I'm using the -fsanitize=undefined flag and got a runtime error after a
compiler upgrade from 7.2 to 9.3. I was happy at first that the upgraded
compiler caught an error that I can fix before it hits me in production.

However after some investigation, I cannot figure out what's wrong in the
code. That's why I seek help on this mailing list, can someone explain
the bug to me? Or is it a bug in the compiler that I should report?

The simplified version of the test code on Compiler Explorer:
https://godbolt.org/z/7rAxJj

It shows that different compiler versions behave differently.

There is a comment on line 16 that explicitly defaults the Derived
constructor. If you switch the comment with line 17, both compiler verions
run fine without producing the runtime error although the defaulted
constructor should be exactly the same the user defined one:
https://godbolt.org/z/UShm-u

Can someone please explain what's wrong?

Thanks in advance,
Tibor Billes

Re: Undefined behavior or compiler bug?

[Jonathan Wakely]

On Mon, 15 Jun 2020 at 18:34, Tibor Billes <tbilles@gmx.com> wrote:
>
> Hi,
>
> I'm using the -fsanitize=undefined flag and got a runtime error after a
> compiler upgrade from 7.2 to 9.3. I was happy at first that the upgraded
> compiler caught an error that I can fix before it hits me in production.
>
> However after some investigation, I cannot figure out what's wrong in the
> code. That's why I seek help on this mailing list, can someone explain
> the bug to me? Or is it a bug in the compiler that I should report?
>
> The simplified version of the test code on Compiler Explorer:
> https://godbolt.org/z/7rAxJj
>
> It shows that different compiler versions behave differently.
>
> There is a comment on line 16 that explicitly defaults the Derived
> constructor. If you switch the comment with line 17, both compiler verions
> run fine without producing the runtime error although the defaulted
> constructor should be exactly the same the user defined one:
> https://godbolt.org/z/UShm-u
>
> Can someone please explain what's wrong?

Looks like a bug, please do report it to bugzilla, thanks.

Re: Undefined behavior or compiler bug?

[Tibor Billes]

On Mon, 15 Jun 2020, Jonathan Wakely wrote:

> On Mon, 15 Jun 2020 at 18:34, Tibor Billes <tbilles@gmx.com> wrote:
> >
> > Hi,
> >
> > I'm using the -fsanitize=undefined flag and got a runtime error after a
> > compiler upgrade from 7.2 to 9.3. I was happy at first that the upgraded
> > compiler caught an error that I can fix before it hits me in production.
> >
> > However after some investigation, I cannot figure out what's wrong in the
> > code. That's why I seek help on this mailing list, can someone explain
> > the bug to me? Or is it a bug in the compiler that I should report?
> >
> > The simplified version of the test code on Compiler Explorer:
> > https://godbolt.org/z/7rAxJj
> >
> > It shows that different compiler versions behave differently.
> >
> > There is a comment on line 16 that explicitly defaults the Derived
> > constructor. If you switch the comment with line 17, both compiler verions
> > run fine without producing the runtime error although the defaulted
> > constructor should be exactly the same the user defined one:
> > https://godbolt.org/z/UShm-u
> >
> > Can someone please explain what's wrong?
>
> Looks like a bug, please do report it to bugzilla, thanks.
>

Ok, here it is: https://gcc.gnu.org/bugzilla/show_bug.cgi?id=95693 .
Thanks for your quick response!

Tibor