Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::3:3 as permitted sender) client-ip=2620:137:e000::3:3;
Message-ID: <88825625-283a-d521-3e7e-058784455577@bytedance.com>
Date:   Wed, 13 Sep 2023 12:56:55 +0800
MIME-Version: 1.0
User-Agent: Mozilla/5.0 (Macintosh; Intel Mac OS X 10.15; rv:102.0)
 Gecko/20100101 Thunderbird/102.15.0
Subject: Re: [PATCH bpf-next v2 2/6] bpf: Introduce css_task open-coded
 iterator kfuncs
To:     Martin KaFai Lau <martin.lau@linux.dev>
Cc:     ast@kernel.org, daniel@iogearbox.net, andrii@kernel.org,
        martin.lau@kernel.org, tj@kernel.org, linux-kernel@vger.kernel.org,
        bpf@vger.kernel.org
References: <20230912070149.969939-1-zhouchuyi@bytedance.com>
 <20230912070149.969939-3-zhouchuyi@bytedance.com>
 <8b272d63-5dd7-13bd-7691-d061895fdbe1@linux.dev>
From:   Chuyi Zhou <zhouchuyi@bytedance.com>
In-Reply-To: <8b272d63-5dd7-13bd-7691-d061895fdbe1@linux.dev>
Content-Type: text/plain; charset=UTF-8; format=flowed
Content-Transfer-Encoding: 8bit
Precedence: bulk

Hello Martin.

在 2023/9/13 03:57, Martin KaFai Lau 写道:
> On 9/12/23 12:01 AM, Chuyi Zhou wrote:
>> +__bpf_kfunc int bpf_iter_css_task_new(struct bpf_iter_css_task *it,
>> +        struct cgroup_subsys_state *css, unsigned int flags)
>> +{
>> +    struct bpf_iter_css_task_kern *kit = (void *)it;
>> +
>> +    BUILD_BUG_ON(sizeof(struct bpf_iter_css_task_kern) != 
>> sizeof(struct bpf_iter_css_task));
>> +    BUILD_BUG_ON(__alignof__(struct bpf_iter_css_task_kern) !=
>> +                    __alignof__(struct bpf_iter_css_task));
>> +
>> +    switch (flags) {
>> +    case CSS_TASK_ITER_PROCS | CSS_TASK_ITER_THREADED:
>> +    case CSS_TASK_ITER_PROCS:
>> +    case 0:
>> +        break;
>> +    default:
>> +        return -EINVAL;
>> +    }
>> +
>> +    kit->css_it = kzalloc(sizeof(struct css_task_iter), GFP_KERNEL);
>> +    if (!kit->css_it)
>> +        return -ENOMEM;
>> +    css_task_iter_start(css, flags, kit->css_it);
>> +    return 0;
>> +}
>> +
> 
>> +static bool check_css_task_iter_allowlist(struct bpf_verifier_env *env)
>> +{
>> +    enum bpf_prog_type prog_type = resolve_prog_type(env->prog);
>> +
>> +    switch (prog_type) {
>> +    case BPF_PROG_TYPE_LSM:
> 
> This will allow the non-sleepable lsm prog to call 
> bpf_iter_css_task_new. The above kzalloc(GFP_KERNEL) in 
> bpf_iter_css_task_new should trigger a lockdep error in the lsm selftest 
> in patch 6.

Thanks for your test. I missed the dmesg error since I did not set 
CONFIG_LOCKDEP.

I think here we can use kzalloc(GFP_ATOMIC) to eliminate this error if 
we want to use this iter in non-sleepable lsm prog. I just tested, it 
works well.

> 
>> +        return true;
>> +    case BPF_TRACE_ITER:
>> +        return env->prog->aux->sleepable;
>> +    default:
>> +        return false;
>> +    }
>> +}
>