On Mon, Jul 24, 2023, at 21:15, Alexei Starovoitov wrote: > On Mon, Jul 24, 2023 at 11:30 AM Arnd Bergmann <arnd@xxxxxxxxxx> wrote: >> On Mon, Jul 24, 2023, at 20:13, Arnd Bergmann wrote: >> >> I have a minimized test case at https://godbolt.org/z/hK4ev17fv >> that shows the problem happening with all versions of gcc >> (4.1 through 14.0) if I force the dec_active() function to be >> inline and force inc_active() to be non-inline. > > That's a bit of cheating, but I see your point now. > How about we do: > diff --git a/kernel/bpf/memalloc.c b/kernel/bpf/memalloc.c > index 51d6389e5152..3fa0944cb975 100644 > --- a/kernel/bpf/memalloc.c > +++ b/kernel/bpf/memalloc.c > @@ -183,11 +183,11 @@ static void inc_active(struct bpf_mem_cache *c, > unsigned long *flags) > WARN_ON_ONCE(local_inc_return(&c->active) != 1); > } > > -static void dec_active(struct bpf_mem_cache *c, unsigned long flags) > +static void dec_active(struct bpf_mem_cache *c, unsigned long *flags) > { > local_dec(&c->active); > if (IS_ENABLED(CONFIG_PREEMPT_RT)) > - local_irq_restore(flags); > + local_irq_restore(*flags); > } Sure, that's fine. Between this and the two suggestions I had (__always_inline or passing the flags from inc_active as a return code), I don't have a strong preference, so pick whichever you like. Arnd
