From: Eric Dumazet <edumazet@google.com>
Date: Sat, 1 Mar 2025 19:13:15 +0000 (+0000)
Subject: bpf: no longer acquire map_idr_lock in bpf_map_inc_not_zero()
X-Git-Tag: ceph-for-6.16-rc1~367^2~34
X-Git-Url: http://git.apps.os.sepia.ceph.com/?a=commitdiff_plain;h=f8ac5a4e1a97354c9a076e9a0c289032618a4f56;p=ceph-client.git

bpf: no longer acquire map_idr_lock in bpf_map_inc_not_zero()

bpf_sk_storage_clone() is the only caller of bpf_map_inc_not_zero()
and is holding rcu_read_lock().

map_idr_lock does not add any protection, just remove the cost
for passive TCP flows.

Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Kui-Feng Lee <kuifeng@meta.com>
Cc: Martin KaFai Lau <martin.lau@kernel.org>
Acked-by: Stanislav Fomichev <sdf@fomichev.me>
Link: https://lore.kernel.org/r/20250301191315.1532629-1-edumazet@google.com
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
---

diff --git a/kernel/bpf/syscall.c b/kernel/bpf/syscall.c
index d799fe8f568e1..1340f6779811c 100644
--- a/kernel/bpf/syscall.c
+++ b/kernel/bpf/syscall.c
@@ -1593,11 +1593,8 @@ struct bpf_map *__bpf_map_inc_not_zero(struct bpf_map *map, bool uref)
 
 struct bpf_map *bpf_map_inc_not_zero(struct bpf_map *map)
 {
-	spin_lock_bh(&map_idr_lock);
-	map = __bpf_map_inc_not_zero(map, false);
-	spin_unlock_bh(&map_idr_lock);
-
-	return map;
+	lockdep_assert(rcu_read_lock_held());
+	return __bpf_map_inc_not_zero(map, false);
 }
 EXPORT_SYMBOL_GPL(bpf_map_inc_not_zero);