Received-SPF: pass (google.com: domain of linux-kernel+bounces-123540-linux.lists.archive=gmail.com@vger.kernel.org designates 2604:1380:45e3:2400::1 as permitted sender) client-ip=2604:1380:45e3:2400::1;
Sender: Tejun Heo <htejun@gmail.com>
Date: Thu, 28 Mar 2024 10:02:11 -1000
From: Tejun Heo <tj@kernel.org>
To: Alexei Starovoitov <alexei.starovoitov@gmail.com>
Cc: Djalal Harouni <tixxdz@gmail.com>, Zefan Li <lizefan.x@bytedance.com>,
	Johannes Weiner <hannes@cmpxchg.org>,
	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>, Stanislav Fomichev <sdf@google.com>,
	Hao Luo <haoluo@google.com>, Jiri Olsa <jolsa@kernel.org>,
	Mykola Lysenko <mykolal@fb.com>, Shuah Khan <shuah@kernel.org>,
	LKML <linux-kernel@vger.kernel.org>,
	"open list:CONTROL GROUP (CGROUP)" <cgroups@vger.kernel.org>,
	bpf <bpf@vger.kernel.org>,
	"open list:KERNEL SELFTEST FRAMEWORK" <linux-kselftest@vger.kernel.org>
Subject: Re: [RFC PATCH bpf-next 0/3] bpf: freeze a task cgroup from bpf
Message-ID: <ZgXMww9kJiKi4Vmd@slm.duckdns.org>
References: <20240327-ccb56fc7a6e80136db80876c@djalal>
 <20240327225334.58474-1-tixxdz@gmail.com>
 <ZgWnPZtwBYfHEFzf@slm.duckdns.org>
 <CAADnVQK6BUGZFCATD8Ejcfob5sKK-b8HUD_4o8Q6s9FM72L4iQ@mail.gmail.com>
 <ZgWv19ySvoACAll4@slm.duckdns.org>
 <CAADnVQLhWDcX-7XCdo-W=jthU=9iPqODwrE6c9fvU8sfAJ5ARg@mail.gmail.com>
Precedence: bulk
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <CAADnVQLhWDcX-7XCdo-W=jthU=9iPqODwrE6c9fvU8sfAJ5ARg@mail.gmail.com>

Hello,

On Thu, Mar 28, 2024 at 12:46:03PM -0700, Alexei Starovoitov wrote:
> To use kernel_file_open() it would need path, inode, cred.

Yeah, it's more involved and potentially more controversial. That said, just
to push the argument a bit further, at least for path, it'd be nice to have
a helper anyway which can return cgroup path. I don't know whether we'd need
direct inode access. For cred, just share some root cred?

> None of that is available now.
> Allocating all these structures just to wrap a cgroup pointer
> feels like overkill.
> Of course, it would solve the need to introduce other
> cgroup apis that are already available via text based cgroupfs
> read/write. So there are pros and cons in both approaches.
> Maybe the 3rd option would be to expose:
> cgroup_lock() as a special blend of acquire plus lock.

Oh, let's not expose that. That has been a source of problem for some use
cases and we may have to change how cgroups are locked.

> Then there will be no need for bpf_task_freeze_cgroup() with task
> argument. Instead cgroup_freeze() will be such kfunc that
> takes cgroup argument and the verifier will check that
> cgroup was acquired/locked.
> Sort-of what we check to access bpf_rb_root.

There's also cgroup.kill which would be useful for similar use cases. We can
add interface for both but idk. Let's say we have something like the
following (pardon the bad naming):

  bpf_cgroup_knob_write(struct cgroup *cgrp, char *filename, char *buf)

Would that work? I'm not necessarily in love with the idea or against adding
separate helpers but the duplication still bothers me a bit.

Thanks.

-- 
tejun