On Thu, Jan 20, 2022 at 11:02:27AM +0800, Menglong Dong wrote: > Hello! > > On Thu, Jan 20, 2022 at 6:03 AM Alexei Starovoitov > <alexei.starovoitov@xxxxxxxxx> wrote: > > > [...] > > > > Looks like > > __sk_buff->remote_port > > bpf_sock_ops->remote_port > > sk_msg_md->remote_port > > are doing the right thing, > > but bpf_sock->dst_port is not correct? > > > > I think it's better to fix it, > > but probably need to consolidate it with > > convert_ctx_accesses() that deals with narrow access. > > I suspect reading u8 from three flavors of 'remote_port' > > won't be correct. > > What's the meaning of 'narrow access'? Do you mean to > make 'remote_port' u16? Or 'remote_port' should be made > accessible with u8? In fact, '*((u16 *)&skops->remote_port + 1)' > won't work, as it only is accessible with u32. u8 access to remote_port won't pass the verifier, but u8 access to dst_port will. Though it will return incorrect data. See how convert_ctx_accesses() handles narrow loads. I think we need to generalize it for different endian fields.
