Received-SPF: pass (google.com: domain of linux-kernel+bounces-104815-linux.lists.archive=gmail.com@vger.kernel.org designates 147.75.199.223 as permitted sender) client-ip=147.75.199.223;
Date: Fri, 15 Mar 2024 11:50:49 -0700
In-Reply-To: <20240315182804.216081-1-josef@netflix.com>
Precedence: bulk
Mime-Version: 1.0
References: <20240315182804.216081-1-josef@netflix.com>
Message-ID: <ZfSYiWcE0_au_ZPt@google.com>
Subject: Re: [PATCH bpf-next 1/2] bpf: add btf_task_get_cgroup_id kfunc
From: Stanislav Fomichev <sdf@google.com>
To: Jose Fernandez <josef@netflix.com>
Cc: Alexei Starovoitov <ast@kernel.org>, Daniel Borkmann <daniel@iogearbox.net>, 
	Andrii Nakryiko <andrii@kernel.org>, Martin KaFai Lau <martin.lau@linux.dev>, 
	Eduard Zingerman <eddyz87@gmail.com>, Song Liu <song@kernel.org>, 
	Yonghong Song <yonghong.song@linux.dev>, John Fastabend <john.fastabend@gmail.com>, 
	KP Singh <kpsingh@kernel.org>, Hao Luo <haoluo@google.com>, Jiri Olsa <jolsa@kernel.org>, 
	bpf@vger.kernel.org, linux-kernel@vger.kernel.org, 
	linux-kselftest@vger.kernel.org, Tycho Andersen <tycho@tycho.pizza>
Content-Type: text/plain; charset="utf-8"

On 03/15, Jose Fernandez wrote:
> This patch enhances the BPF helpers by adding a kfunc to retrieve the
> cgroup v2 ID of a specific task, addressing a previous limitation where
> only bpf_task_get_cgroup1 was available for cgroup v1. The new kfunc is
> particularly useful for scenarios where obtaining the cgroup ID of a
> task other than the "current" one is necessary, which the existing
> bpf_get_current_cgroup_id helper cannot accommodate. A specific use case
> at Netflix involved the sched_switch tracepoint, where we had to get
> the cgroup IDs of both the previous and next tasks.
> 
> The bpf_task_get_cgroup_id kfunc returns a task's cgroup ID, correctly
> implementing RCU read locking and unlocking for safe data access, and
> leverages existing cgroup.h helpers to fetch the cgroup and its ID.
> 
> Signed-off-by: Jose Fernandez <josef@netflix.com>
> Reviewed-by: Tycho Andersen <tycho@tycho.pizza>
> ---
>  kernel/bpf/helpers.c | 22 ++++++++++++++++++++++
>  1 file changed, 22 insertions(+)
> 
> diff --git a/kernel/bpf/helpers.c b/kernel/bpf/helpers.c
> index a89587859571..8038b2bd3488 100644
> --- a/kernel/bpf/helpers.c
> +++ b/kernel/bpf/helpers.c
> @@ -2266,6 +2266,27 @@ bpf_task_get_cgroup1(struct task_struct *task, int hierarchy_id)
>  		return NULL;
>  	return cgrp;
>  }
> +
> +/**
> + * bpf_task_get_cgroup_id - Get the cgroup ID of a task.
> + * @task: The target task
> + *
> + * This function returns the ID of the task's default cgroup, primarily
> + * designed for use with cgroup v2. In cgroup v1, the concept of default
> + * cgroup varies by subsystem, and while this function will work with
> + * cgroup v1, it's recommended to use bpf_task_get_cgroup1 instead.
> + */
> +__bpf_kfunc u64 bpf_task_get_cgroup_id(struct task_struct *task)
> +{
> +	struct cgroup *cgrp;
> +	u64 cgrp_id;
> +
> +	rcu_read_lock();
> +	cgrp = task_dfl_cgroup(task);
> +	cgrp_id = cgroup_id(cgrp);
> +	rcu_read_unlock();
> +	return cgrp_id;
> +}
>  #endif /* CONFIG_CGROUPS */
>  
>  /**
> @@ -2573,6 +2594,7 @@ BTF_ID_FLAGS(func, bpf_cgroup_ancestor, KF_ACQUIRE | KF_RCU | KF_RET_NULL)
>  BTF_ID_FLAGS(func, bpf_cgroup_from_id, KF_ACQUIRE | KF_RET_NULL)
>  BTF_ID_FLAGS(func, bpf_task_under_cgroup, KF_RCU)
>  BTF_ID_FLAGS(func, bpf_task_get_cgroup1, KF_ACQUIRE | KF_RCU | KF_RET_NULL)
> +BTF_ID_FLAGS(func, bpf_task_get_cgroup_id, KF_RCU)

Any reason we are not returning 'struct cgroup' pointer? That should
allow using other kfuncs that are all 'struct cgrop' based as well.