On 12/13/23 3:09 AM, Hou Tao wrote:
Hi,
On 12/13/2023 6:31 AM, Yonghong Song wrote:
For percpu data structure allocation with bpf_global_percpu_ma,
the maximum data size is 4K. But for a system with large
number of cpus, bigger data size (e.g., 2K, 4K) might consume
a lot of memory. For example, the percpu memory consumption
with unit size 2K and 1024 cpus will be 2K * 1K * 1k = 2GB
memory.
We should discourage such usage. Let us limit the maximum data
size to be 512 for bpf_global_percpu_ma allocation.
Signed-off-by: Yonghong Song <yonghong.song@xxxxxxxxx>
---
kernel/bpf/verifier.c | 7 +++++++
1 file changed, 7 insertions(+)
diff --git a/kernel/bpf/verifier.c b/kernel/bpf/verifier.c
index 0c55fe4451e1..e5cb6b7526b6 100644
--- a/kernel/bpf/verifier.c
+++ b/kernel/bpf/verifier.c
@@ -43,6 +43,8 @@ static const struct bpf_verifier_ops * const bpf_verifier_ops[] = {
};
struct bpf_mem_alloc bpf_global_percpu_ma;
+#define LLIST_NODE_SZ sizeof(struct llist_node)
+#define BPF_GLOBAL_PERCPU_MA_MAX_SIZE (512 - LLIST_NODE_SZ)
It seems for per-cpu allocation the extra subtraction is not needed, we
could use all allocated space in per-cpu pointer. Maybe we could update
bpf_mem_alloc() firstly to use size instead of size + sizeof(void *) to
select cache.
Good point. If this works, it can also ensure if users try to allocate
512 bytes. It will go to 512-byte bucket instead of 1024-byte buck.
Will investigate.
/* bpf_check() is a static code analyzer that walks eBPF program
* instruction by instruction and updates register/stack state.
@@ -12091,6 +12093,11 @@ static int check_kfunc_call(struct bpf_verifier_env *env, struct bpf_insn *insn,
}
if (meta.func_id == special_kfunc_list[KF_bpf_percpu_obj_new_impl]) {
+ if (ret_t->size > BPF_GLOBAL_PERCPU_MA_MAX_SIZE) {
+ verbose(env, "bpf_percpu_obj_new type size (%d) is greater than %lu\n",
+ ret_t->size, BPF_GLOBAL_PERCPU_MA_MAX_SIZE);
+ return -EINVAL;
+ }
mutex_lock(&bpf_percpu_ma_lock);
err = bpf_mem_alloc_percpu_unit_init(&bpf_global_percpu_ma, ret_t->size);
mutex_unlock(&bpf_percpu_ma_lock);
