On Tue, Feb 11, 2025 at 3:41 PM Martin KaFai Lau <martin.lau@xxxxxxxxx> wrote:
>
> On 2/8/25 2:32 AM, Jason Xing wrote:
> > Use __bpf_kfunc feature to allow bpf prog dynamically and selectively
>
> s/Use/Add/
>
> Remove "dynamically". A kfunc can only be called dynamically at runtime.
>
> Like:
>
> "Add the bpf_sock_ops_enable_tx_tstamp kfunc to allow BPF programs to
> selectively enable TX timestamping on a skb during tcp_sendmsg..."
Will adjust it.
>
> > to sample/track the skb. For example, the bpf prog will limit tracking
> > X numbers of packets and then will stop there instead of tracing
> > all the sendmsgs of matched flow all along.
> > > Signed-off-by: Jason Xing <kerneljasonxing@xxxxxxxxx>
> > ---
> > kernel/bpf/btf.c | 1 +
> > net/core/filter.c | 27 ++++++++++++++++++++++++++-
> > 2 files changed, 27 insertions(+), 1 deletion(-)
> >
> > diff --git a/kernel/bpf/btf.c b/kernel/bpf/btf.c
> > index 8396ce1d0fba..a65e2eeffb88 100644
> > --- a/kernel/bpf/btf.c
> > +++ b/kernel/bpf/btf.c
> > @@ -8535,6 +8535,7 @@ static int bpf_prog_type_to_kfunc_hook(enum bpf_prog_type prog_type)
> > case BPF_PROG_TYPE_CGROUP_SOCK_ADDR:
> > case BPF_PROG_TYPE_CGROUP_SOCKOPT:
> > case BPF_PROG_TYPE_CGROUP_SYSCTL:
> > + case BPF_PROG_TYPE_SOCK_OPS:
> > return BTF_KFUNC_HOOK_CGROUP;
> > case BPF_PROG_TYPE_SCHED_ACT:
> > return BTF_KFUNC_HOOK_SCHED_ACT;
> > diff --git a/net/core/filter.c b/net/core/filter.c
> > index 7f56d0bbeb00..db20a947e757 100644
> > --- a/net/core/filter.c
> > +++ b/net/core/filter.c
> > @@ -12102,6 +12102,21 @@ __bpf_kfunc int bpf_sk_assign_tcp_reqsk(struct __sk_buff *s, struct sock *sk,
> > #endif
> > }
> >
> > +__bpf_kfunc int bpf_sock_ops_enable_tx_tstamp(struct bpf_sock_ops_kern *skops)
>
> I am ok to always enable txstamp_ack here. Please still add a second "u64 flags"
> argument such that future disable/enable is still possible.
Ok. Will do it.
>
> > +{
> > + struct sk_buff *skb;
> > +
> > + if (skops->op != BPF_SOCK_OPS_TS_SND_CB)
> > + return -EOPNOTSUPP;> +
> > + skb = skops->skb;
> > + TCP_SKB_CB(skb)->txstamp_ack = 2;
>
> Willem (thanks!) has already mentioned there is a bug.
>
> This also brought up that a test is missing: the bpf timestamping and user
> space's SO_TIMESTAMPING can work without interfering others. The current test
> only has SK_BPF_CB_TX_TIMESTAMPING on. A test is needed when both
> SK_BPF_CB_TX_TIMESTAMPING and the user space's SO_TIMESTAMPING are on. The
> expectation is both of them will work together.
Yeah, I did miss this particular test. Let me figure out how to test
it in a proper way.
Thanks,
Jason
>
> > + skb_shinfo(skb)->tx_flags |= SKBTX_BPF;
> > + skb_shinfo(skb)->tskey = TCP_SKB_CB(skb)->seq + skb->len - 1;
> > +
> > + return 0;
> > +}
> > +
> > __bpf_kfunc_end_defs();
> >
> > int bpf_dynptr_from_skb_rdonly(struct __sk_buff *skb, u64 flags,
> > @@ -12135,6 +12150,10 @@ BTF_KFUNCS_START(bpf_kfunc_check_set_tcp_reqsk)
> > BTF_ID_FLAGS(func, bpf_sk_assign_tcp_reqsk, KF_TRUSTED_ARGS)
> > BTF_KFUNCS_END(bpf_kfunc_check_set_tcp_reqsk)
> >
> > +BTF_KFUNCS_START(bpf_kfunc_check_set_sock_ops)
> > +BTF_ID_FLAGS(func, bpf_sock_ops_enable_tx_tstamp, KF_TRUSTED_ARGS)
> > +BTF_KFUNCS_END(bpf_kfunc_check_set_sock_ops)
> > +
> > static const struct btf_kfunc_id_set bpf_kfunc_set_skb = {
> > .owner = THIS_MODULE,
> > .set = &bpf_kfunc_check_set_skb,
> > @@ -12155,6 +12174,11 @@ static const struct btf_kfunc_id_set bpf_kfunc_set_tcp_reqsk = {
> > .set = &bpf_kfunc_check_set_tcp_reqsk,
> > };
> >
> > +static const struct btf_kfunc_id_set bpf_kfunc_set_sock_ops = {
> > + .owner = THIS_MODULE,
> > + .set = &bpf_kfunc_check_set_sock_ops,
> > +};
> > +
> > static int __init bpf_kfunc_init(void)
> > {
> > int ret;
> > @@ -12173,7 +12197,8 @@ static int __init bpf_kfunc_init(void)
> > ret = ret ?: register_btf_kfunc_id_set(BPF_PROG_TYPE_XDP, &bpf_kfunc_set_xdp);
> > ret = ret ?: register_btf_kfunc_id_set(BPF_PROG_TYPE_CGROUP_SOCK_ADDR,
> > &bpf_kfunc_set_sock_addr);
> > - return ret ?: register_btf_kfunc_id_set(BPF_PROG_TYPE_SCHED_CLS, &bpf_kfunc_set_tcp_reqsk);
> > + ret = ret ?: register_btf_kfunc_id_set(BPF_PROG_TYPE_SCHED_CLS, &bpf_kfunc_set_tcp_reqsk);
> > + return ret ?: register_btf_kfunc_id_set(BPF_PROG_TYPE_SOCK_OPS, &bpf_kfunc_set_sock_ops);
> > }
> > late_initcall(bpf_kfunc_init);
> >
>
