bpf, sockmap, tcp: sk_prot needs inuse_idx set for proc stats
authorJohn Fastabend <john.fastabend@gmail.com>
Mon, 12 Jul 2021 19:55:46 +0000 (12:55 -0700)
committerGreg Kroah-Hartman <gregkh@linuxfoundation.org>
Wed, 28 Jul 2021 12:35:37 +0000 (14:35 +0200)
commitc260442431b4a9bb17e0ae4b727cc6838bea5f33
tree38a6e22ac0e2a33dd873e2058dfd8a69d2de6ea5
parent715f378f42909c401ec043f5150c4fdf57fb8889
bpf, sockmap, tcp: sk_prot needs inuse_idx set for proc stats

[ Upstream commit 228a4a7ba8e99bb9ef980b62f71e3be33f4aae69 ]

The proc socket stats use sk_prot->inuse_idx value to record inuse sock
stats. We currently do not set this correctly from sockmap side. The
result is reading sock stats '/proc/net/sockstat' gives incorrect values.
The socket counter is incremented correctly, but because we don't set the
counter correctly when we replace sk_prot we may omit the decrement.

To get the correct inuse_idx value move the core_initcall that initializes
the TCP proto handlers to late_initcall. This way it is initialized after
TCP has the chance to assign the inuse_idx value from the register protocol
handler.

Fixes: 604326b41a6fb ("bpf, sockmap: convert to generic sk_msg interface")
Suggested-by: Jakub Sitnicki <jakub@cloudflare.com>
Signed-off-by: John Fastabend <john.fastabend@gmail.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Reviewed-by: Cong Wang <cong.wang@bytedance.com>
Link: https://lore.kernel.org/bpf/20210712195546.423990-3-john.fastabend@gmail.com
Signed-off-by: Sasha Levin <sashal@kernel.org>
net/ipv4/tcp_bpf.c