This patch adds a few optnames for bpf_setsockopt:
SO_REUSEADDR, IPV6_AUTOFLOWLABEL, TCP_MAXSEG, TCP_NODELAY,
and TCP_THIN_LINEAR_TIMEOUTS.
Thanks to the previous patches of this set, all additions can reuse
the sk_setsockopt(), do_ipv6_setsockopt(), and do_tcp_setsockopt().
The only change here is to allow them in bpf_setsockopt.
The bpf prog has been able to read all members of a sk by
using PTR_TO_BTF_ID of a sk. The optname additions here can also be
read by the same approach. Meaning there is a way to read
the values back.
These optnames can also be added to bpf_getsockopt() later with
another patch set that makes the bpf_getsockopt() to reuse
the sock_getsockopt(), tcp_getsockopt(), and ip[v6]_getsockopt().
Thus, this patch does not add more duplicated code to
bpf_getsockopt() now.
Reviewed-by: Stanislav Fomichev <sdf@google.com>
Signed-off-by: Martin KaFai Lau <kafai@fb.com>
Link: https://lore.kernel.org/r/20220817061841.4181642-1-kafai@fb.com
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
char *optval, int optlen)
{
switch (optname) {
+ case SO_REUSEADDR:
case SO_SNDBUF:
case SO_RCVBUF:
case SO_KEEPALIVE:
return -EINVAL;
switch (optname) {
+ case TCP_NODELAY:
+ case TCP_MAXSEG:
case TCP_KEEPIDLE:
case TCP_KEEPINTVL:
case TCP_KEEPCNT:
case TCP_SYNCNT:
case TCP_WINDOW_CLAMP:
+ case TCP_THIN_LINEAR_TIMEOUTS:
case TCP_USER_TIMEOUT:
case TCP_NOTSENT_LOWAT:
case TCP_SAVE_SYN:
switch (optname) {
case IPV6_TCLASS:
+ case IPV6_AUTOFLOWLABEL:
if (optlen != sizeof(int))
return -EINVAL;
break;