selftests: bpf: Less strict size check in sockopt_sk

Originally, the kernel strictly checked the size of the optval in
getsockopt(TCP_ZEROCOPY_RECEIVE) to be equal to sizeof(struct
tcp_zerocopy_receive). With c8856c0, this was changed to allow
optvals of different sizes.

The bpf code in the sockopt_sk test was still performing the strict size
check. This fix adapts the kernel behavior from c8856c0 in the
selftest, i.e., just check if the required fields are there.

Fixes: 9cacf81 ("bpf: Remove extra lock_sock for TCP_ZEROCOPY_RECEIVE")
Signed-off-by: Felix Maurer <fmaurer@redhat.com>
Reviewed-by: Stanislav Fomichev <sdf@google.com>
Link: https://lore.kernel.org/r/6f569cca2e45473f9a724d54d03fdfb45f29e35f.1643129402.git.fmaurer@redhat.com
Signed-off-by: Alexei Starovoitov <ast@kernel.org>

tools/testing/selftests/bpf/progs/sockopt_sk.c

@@ -72,7 +72,8 @@ int _getsockopt(struct bpf_sockopt *ctx)
 		 * reasons.
 		 */
 
-		if (optval + sizeof(struct tcp_zerocopy_receive) > optval_end)
+		/* Check that optval contains address (__u64) */
+		if (optval + sizeof(__u64) > optval_end)
 			return 0; /* bounds check */
 
 		if (((struct tcp_zerocopy_receive *)optval)->address != 0)