On Tue, Apr 22, 2025 at 1:04 AM Feng Yang <[email protected]> wrote: > > Subject: Re: [PATCH bpf-next] bpf: Remove bpf_get_smp_processor_id_proto > > On Mon, 21 Apr 2025 18:53:07 -0700 Alexei Starovoitov > <[email protected]> wrote: > > > On Thu, Apr 17, 2025 at 8:41 PM Feng Yang <[email protected]> wrote: > > > > > > From: Feng Yang <[email protected]> > > > > > > All BPF programs either disable CPU preemption or CPU migration, > > > so the bpf_get_smp_processor_id_proto can be safely removed, > > > and the bpf_get_raw_smp_processor_id_proto in bpf_base_func_proto works > > > perfectly. > > > > > > Suggested-by: Andrii Nakryiko <[email protected]> > > > Signed-off-by: Feng Yang <[email protected]> > > > --- > > > include/linux/bpf.h | 1 - > > > kernel/bpf/core.c | 1 - > > > kernel/bpf/helpers.c | 12 ------------ > > > kernel/trace/bpf_trace.c | 2 -- > > > net/core/filter.c | 6 ------ > > > 5 files changed, 22 deletions(-) > > > > > > diff --git a/include/linux/bpf.h b/include/linux/bpf.h > > > index 3f0cc89c0622..36e525141556 100644 > > > --- a/include/linux/bpf.h > > > +++ b/include/linux/bpf.h > > > @@ -3316,7 +3316,6 @@ extern const struct bpf_func_proto > > > bpf_map_peek_elem_proto; > > > extern const struct bpf_func_proto bpf_map_lookup_percpu_elem_proto; > > > > > > extern const struct bpf_func_proto bpf_get_prandom_u32_proto; > > > -extern const struct bpf_func_proto bpf_get_smp_processor_id_proto; > > > extern const struct bpf_func_proto bpf_get_numa_node_id_proto; > > > extern const struct bpf_func_proto bpf_tail_call_proto; > > > extern const struct bpf_func_proto bpf_ktime_get_ns_proto; > > > diff --git a/kernel/bpf/core.c b/kernel/bpf/core.c > > > index ba6b6118cf50..1ad41a16b86e 100644 > > > --- a/kernel/bpf/core.c > > > +++ b/kernel/bpf/core.c > > > @@ -2943,7 +2943,6 @@ const struct bpf_func_proto bpf_spin_unlock_proto > > > __weak; > > > const struct bpf_func_proto bpf_jiffies64_proto __weak; > > > > > > const struct bpf_func_proto bpf_get_prandom_u32_proto __weak; > > > -const struct bpf_func_proto bpf_get_smp_processor_id_proto __weak; > > > const struct bpf_func_proto bpf_get_numa_node_id_proto __weak; > > > const struct bpf_func_proto bpf_ktime_get_ns_proto __weak; > > > const struct bpf_func_proto bpf_ktime_get_boot_ns_proto __weak; > > > diff --git a/kernel/bpf/helpers.c b/kernel/bpf/helpers.c > > > index e3a2662f4e33..2d2bfb2911f8 100644 > > > --- a/kernel/bpf/helpers.c > > > +++ b/kernel/bpf/helpers.c > > > @@ -149,18 +149,6 @@ const struct bpf_func_proto > > > bpf_get_prandom_u32_proto = { > > > .ret_type = RET_INTEGER, > > > }; > > > > > > -BPF_CALL_0(bpf_get_smp_processor_id) > > > -{ > > > - return smp_processor_id(); > > > -} > > > - > > > -const struct bpf_func_proto bpf_get_smp_processor_id_proto = { > > > - .func = bpf_get_smp_processor_id, > > > - .gpl_only = false, > > > - .ret_type = RET_INTEGER, > > > - .allow_fastcall = true, > > > -}; > > > - > > > > bpf_get_raw_smp_processor_id_proto doesn't have > > allow_fastcall = true > > > > so this breaks tests. > > > > Instead of removing BPF_CALL_0(bpf_get_smp_processor_id) > > we should probably remove BPF_CALL_0(bpf_get_raw_cpu_id) > > and adjust SKF_AD_OFF + SKF_AD_CPU case. > > I don't recall why raw_ version was used back in 2014. > > > > The following two seem to explain the reason: > https://lore.kernel.org/all/7103e2085afa29c006cd5b94a6e4a2ac83efc30d.1467106475.git.dan...@iogearbox.net/ > https://lore.kernel.org/all/02fa71ebe1c560cad489967aa29c653b48932596.1474586162.git.dan...@iogearbox.net/ >
Ahh. socket filters run in RCU CS. They don't disable preemption or migration. Then let's keep things as-is. We still want debugging provided by smp_processor_id(). If we switch everything to raw_ may miss things. Like this example with socket filters.
