Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.37 as permitted sender) client-ip=23.128.96.37;
Date:   Tue, 12 Sep 2023 10:13:37 -0700
From:   Alexei Starovoitov <alexei.starovoitov@gmail.com>
To:     Chuyi Zhou <zhouchuyi@bytedance.com>
Cc:     bpf@vger.kernel.org, ast@kernel.org, daniel@iogearbox.net,
        andrii@kernel.org, martin.lau@kernel.org, tj@kernel.org,
        linux-kernel@vger.kernel.org
Subject: Re: [PATCH bpf-next v2 2/6] bpf: Introduce css_task open-coded
 iterator kfuncs
Message-ID: <20230912171337.px445hxd76uxxnu6@iphone-544e90d47a76.dhcp.thefacebook.com>
References: <20230912070149.969939-1-zhouchuyi@bytedance.com>
 <20230912070149.969939-3-zhouchuyi@bytedance.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20230912070149.969939-3-zhouchuyi@bytedance.com>
Precedence: bulk

On Tue, Sep 12, 2023 at 03:01:45PM +0800, Chuyi Zhou wrote:
> This patch adds kfuncs bpf_iter_css_task_{new,next,destroy} which allow
> creation and manipulation of struct bpf_iter_css_task in open-coded
> iterator style. These kfuncs actually wrapps css_task_iter_{start,next,
> end}. BPF programs can use these kfuncs through bpf_for_each macro for
> iteration of all tasks under a css.
> 
> css_task_iter_*() would try to get the global spin-lock *css_set_lock*, so
> the bpf side has to be careful in where it allows to use this iter.
> Currently we only allow it in bpf_lsm and bpf iter-s.
> 
> Signed-off-by: Chuyi Zhou <zhouchuyi@bytedance.com>
> ---
>  include/uapi/linux/bpf.h       |  4 +++
>  kernel/bpf/helpers.c           |  3 +++
>  kernel/bpf/task_iter.c         | 48 ++++++++++++++++++++++++++++++++++
>  kernel/bpf/verifier.c          | 23 ++++++++++++++++
>  tools/include/uapi/linux/bpf.h |  4 +++
>  tools/lib/bpf/bpf_helpers.h    |  7 +++++
>  6 files changed, 89 insertions(+)
> 
> diff --git a/include/uapi/linux/bpf.h b/include/uapi/linux/bpf.h
> index 73b155e52204..de02c0971428 100644
> --- a/include/uapi/linux/bpf.h
> +++ b/include/uapi/linux/bpf.h
> @@ -7318,4 +7318,8 @@ struct bpf_iter_num {
>  	__u64 __opaque[1];
>  } __attribute__((aligned(8)));
>  
> +struct bpf_iter_css_task {
> +	__u64 __opaque[1];
> +} __attribute__((aligned(8)));
> +
>  #endif /* _UAPI__LINUX_BPF_H__ */
> diff --git a/kernel/bpf/helpers.c b/kernel/bpf/helpers.c
> index b0a9834f1051..d6a16becfbb9 100644
> --- a/kernel/bpf/helpers.c
> +++ b/kernel/bpf/helpers.c
> @@ -2504,6 +2504,9 @@ BTF_ID_FLAGS(func, bpf_dynptr_slice_rdwr, KF_RET_NULL)
>  BTF_ID_FLAGS(func, bpf_iter_num_new, KF_ITER_NEW)
>  BTF_ID_FLAGS(func, bpf_iter_num_next, KF_ITER_NEXT | KF_RET_NULL)
>  BTF_ID_FLAGS(func, bpf_iter_num_destroy, KF_ITER_DESTROY)
> +BTF_ID_FLAGS(func, bpf_iter_css_task_new, KF_ITER_NEW)

Looking at selftest:
struct cgroup_subsys_state *css = &cgrp->self;

realized that we're missing KF_TRUSTED_ARGS here.

> +BTF_ID_FLAGS(func, bpf_iter_css_task_next, KF_ITER_NEXT | KF_RET_NULL)
> +BTF_ID_FLAGS(func, bpf_iter_css_task_destroy, KF_ITER_DESTROY)
>  BTF_ID_FLAGS(func, bpf_dynptr_adjust)
>  BTF_ID_FLAGS(func, bpf_dynptr_is_null)
>  BTF_ID_FLAGS(func, bpf_dynptr_is_rdonly)
> diff --git a/kernel/bpf/task_iter.c b/kernel/bpf/task_iter.c
> index 7473068ed313..d8539cc05ffd 100644
> --- a/kernel/bpf/task_iter.c
> +++ b/kernel/bpf/task_iter.c
> @@ -803,6 +803,54 @@ const struct bpf_func_proto bpf_find_vma_proto = {
>  	.arg5_type	= ARG_ANYTHING,
>  };
>  
> +struct bpf_iter_css_task_kern {
> +	struct css_task_iter *css_it;
> +} __attribute__((aligned(8)));
> +
> +__bpf_kfunc int bpf_iter_css_task_new(struct bpf_iter_css_task *it,
> +		struct cgroup_subsys_state *css, unsigned int flags)

the verifier does a type check, but it's not strong enough.
We need KF_TRUSTED_ARGS to make sure the pointer is valid.
The BTF_TYPE_SAFE_RCU(struct cgroup) {
probably doesn't need to change, since '&cgrp->self' is not a pointer deref.
The verifier should understand that cgroup_subsys_state is also PTR_TRUSTED
just like 'cgrp' pointer.

Also please add negative tests in patch 6.
Like doing bpf_rcu_read_unlock() in the middle and check that the verifier
catches such mistake.