[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <ff77b4e7-49b5-61d8-12f1-b5927eb671a6@fb.com>
Date: Sat, 8 Jan 2022 10:53:49 -0800
From: Yonghong Song <yhs@...com>
To: "Yichun Zhang (agentzh)" <yichun@...nresty.com>
CC: Alexei Starovoitov <ast@...nel.org>,
Daniel Borkmann <daniel@...earbox.net>,
Andrii Nakryiko <andrii@...nel.org>,
Martin KaFai Lau <kafai@...com>,
Song Liu <songliubraving@...com>,
John Fastabend <john.fastabend@...il.com>,
KP Singh <kpsingh@...nel.org>,
Nathan Chancellor <nathan@...nel.org>,
Nick Desaulniers <ndesaulniers@...gle.com>,
<netdev@...r.kernel.org>, <bpf@...r.kernel.org>,
<linux-kernel@...r.kernel.org>, <llvm@...ts.linux.dev>
Subject: Re: [PATCH] bpf: btf: Fix a var size check in validator
On 1/7/22 6:22 PM, Yichun Zhang (agentzh) wrote:
> The btf validator should croak when the variable size is larger than
> its type size, not less. The LLVM optimizer may use smaller sizes for
> the C type.
>
> We ran into this issue with real-world BPF programs emitted by the
> latest version of Clang/LLVM.
Could you give an example to show we have variable size is less than
type_size?
>
> Fixes: 1dc92851849cc ("bpf: kernel side support for BTF Var and DataSec")
> Signed-off-by: Yichun Zhang (agentzh) <yichun@...nresty.com>
> ---
> kernel/bpf/btf.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/kernel/bpf/btf.c b/kernel/bpf/btf.c
> index 9bdb03767db5..2a6967b13ce1 100644
> --- a/kernel/bpf/btf.c
> +++ b/kernel/bpf/btf.c
> @@ -3696,7 +3696,7 @@ static int btf_datasec_resolve(struct btf_verifier_env *env,
> return -EINVAL;
> }
>
> - if (vsi->size < type_size) {
> + if (vsi->size > type_size) {
> btf_verifier_log_vsi(env, v->t, vsi, "Invalid size");
> return -EINVAL;
> }
The fix is incorrect. We do have cases where variable size greater than
type_size. For example,
$ cat t.c
struct t {
int g;
char a[];
} v = {1, {'a', 'b', 'c', 0}};
The type (struct t) size is 4, but the variable size is 8.
Powered by blists - more mailing lists