Re: [PATCH] http.c: clear the 'finished' member once we are done with it

[Carlo Arenas <carenas@gmail.com>]

On Sat, May 7, 2022 at 11:42 AM Junio C Hamano <gitster@pobox.com> wrote:
>
> Carlo Marcelo Arenas Belón <carenas@gmail.com> writes:
>
> > On Fri, May 06, 2022 at 02:17:01PM -0700, Junio C Hamano wrote:
> >> diff --git a/http.c b/http.c
> >> index 229da4d148..85437b1980 100644
> >> --- a/http.c
> >> +++ b/http.c
> >> @@ -1367,6 +1367,9 @@ void run_active_slot(struct active_request_slot *slot)
> >>                      select(max_fd+1, &readfds, &writefds, &excfds, &select_timeout);
> >>              }
> >>      }
> >> +
> >> +    if (slot->finished == &finished)
> >> +            slot->finished = NULL;
> >
> > I am not completely sure yet (since I looked at it long ago and got
> > sidetracked) but I think this might be optimized out (at least by gcc12)
> > since it is technically UB, which is why it never "fixed" the warning.
>
> UB meaning "undefined behaviour"?  Which part is?  Taking the
> address of an on-stack variable "finished"?
> Comparing it with a
> pointer that may or may not have been assigned/overwritten elsewhere
> in a structure?

it is not very intuitive, but using a pointer to a variable that is
out of scope is UB, and in this case the value of slot->finished might
point to an address that is not in our own stack (because it came from
a different thread), hence undefined

Carlo