Skip to contentSkip to conversation

bpf: Fix memlock accounting for sock_hash

PATCH In review Andrey Ignatov 2 participants 2 messages Jun 12, 2020

Subsystems (4)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 bpf: Fix memlock accounting for sock_hash

Add missed bpf_map_charge_init() in sock_hash_alloc() and correspondingly bpf_map_charge_finish() on ENOMEM.

It was found accidentally while working on unrelated selftest that checks "map->memory.pages > 0" is true for all map types.

Before:

	# bpftool m l
	...
	3692: sockhash  name m_sockhash  flags 0x0
		key 4B  value 4B  max_entries 8  memlock 0B

After:

	# bpftool m l
	...
	84: sockmap  name m_sockmap  flags 0x0
		key 4B  value 4B  max_entries 8  memlock 4096B
Fixes 604326b41a6f ("bpf, sockmap: convert to generic sk_msg interface")
---
 net/core/sock_map.c | 4 ++++
 1 file changed, 4 insertions(+)
net/core/sock_map.c +4 -0
index 17a40a947546..3a2dc6af1d78 100644
@@ -989,11 +989,15 @@ static struct bpf_map *sock_hash_alloc(union bpf_attr *attr)
 		err = -EINVAL;
 		goto free_htab;
 	}
+	err = bpf_map_charge_init(&htab->map.memory, cost);
+	if (err)
+		goto free_htab;
 
 	htab->buckets = bpf_map_area_alloc(htab->buckets_num *
 					   sizeof(struct bpf_htab_bucket),
 					   htab->map.numa_node);
 	if (!htab->buckets) {
+		bpf_map_charge_finish(&htab->map.memory);
 		err = -ENOMEM;
 		goto free_htab;
 	}
Signature · 1 line
2.24.1
#1 lore raw
Quoted 13 lines · Andrey Ignatov

Add missed bpf_map_charge_init() in sock_hash_alloc() and correspondingly bpf_map_charge_finish() on ENOMEM.

It was found accidentally while working on unrelated selftest that checks "map->memory.pages > 0" is true for all map types.

Before:

        # bpftool m l
        ...
        3692: sockhash  name m_sockhash  flags 0x0
                key 4B  value 4B  max_entries 8  memlock 0B
Quoted 8 lines

After:

        # bpftool m l
        ...
        84: sockmap  name m_sockmap  flags 0x0
                key 4B  value 4B  max_entries 8  memlock 4096B
Fixes 604326b41a6f ("bpf, sockmap: convert to generic sk_msg interface")

That's a nasty one. Good catch.
Applied.