Re: [PATCH bpf-next v2] bpf: make update_prog_stats always_inline

[Menglong Dong]

On Tue, Jun 24, 2025 at 12:26 AM Alexei Starovoitov
<alexei.starovoitov@xxxxxxxxx> wrote:
>
> On Fri, Jun 20, 2025 at 9:57 PM Menglong Dong <menglong8.dong@xxxxxxxxx> wrote:
> >
> > The function update_prog_stats() will be called in the bpf trampoline.
> > In most cases, it will be optimized by the compiler by making it inline.
> > However, we can't rely on the compiler all the time, and just make it
> > __always_inline to reduce the possible overhead.
> >
> > Signed-off-by: Menglong Dong <dongml2@xxxxxxxxxxxxxxx>
> > ---
> > v2:
> > - split out __update_prog_stats() and make update_prog_stats()
> >   __always_inline, as Alexei's advice
> > ---
> >  kernel/bpf/trampoline.c | 23 ++++++++++++++---------
> >  1 file changed, 14 insertions(+), 9 deletions(-)
> >
> > diff --git a/kernel/bpf/trampoline.c b/kernel/bpf/trampoline.c
> > index c4b1a98ff726..1f92246117eb 100644
> > --- a/kernel/bpf/trampoline.c
> > +++ b/kernel/bpf/trampoline.c
> > @@ -911,18 +911,16 @@ static u64 notrace __bpf_prog_enter_recur(struct bpf_prog *prog, struct bpf_tram
> >         return bpf_prog_start_time();
> >  }
> >
> > -static void notrace update_prog_stats(struct bpf_prog *prog,
> > -                                     u64 start)
> > +static void notrace __update_prog_stats(struct bpf_prog *prog, u64 start)
> >  {
> >         struct bpf_prog_stats *stats;
> >
> > -       if (static_branch_unlikely(&bpf_stats_enabled_key) &&
> > -           /* static_key could be enabled in __bpf_prog_enter*
> > -            * and disabled in __bpf_prog_exit*.
> > -            * And vice versa.
> > -            * Hence check that 'start' is valid.
> > -            */
> > -           start > NO_START_TIME) {
> > +       /* static_key could be enabled in __bpf_prog_enter*
> > +        * and disabled in __bpf_prog_exit*.
> > +        * And vice versa.
> > +        * Hence check that 'start' is valid.
> > +        */
>
>
> Instead of old networking style I reformatted above to normal
> kernel style comment.
>
> > +       if (start > NO_START_TIME) {
>
> and refactored it to <= and removed extra indent in below.
> while applying.

Looks much better, thanks a lot ~