linux-kernel - Re: [PATCH] selftests/bpf: simplify cgroup_hierarchical

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives
Hash Suite: Windows password security audit tool. GUI, reports in PDF.
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <CAJD7tkZg2jzDDR6vn5=-TS93Tm3P-YEQ+06KDsjg=Mzkt5LqsA@mail.gmail.com>
Date:   Mon, 29 Aug 2022 18:06:41 -0700
From:   Yosry Ahmed <yosryahmed@...gle.com>
To:     Hao Luo <haoluo@...gle.com>
Cc:     Andrii Nakryiko <andrii@...nel.org>,
        Alexei Starovoitov <ast@...nel.org>,
        Daniel Borkmann <daniel@...earbox.net>,
        Martin KaFai Lau <martin.lau@...ux.dev>,
        Mykola Lysenko <mykolal@...com>, Song Liu <song@...nel.org>,
        Yonghong Song <yhs@...com>,
        John Fastabend <john.fastabend@...il.com>,
        KP Singh <kpsingh@...nel.org>,
        Stanislav Fomichev <sdf@...gle.com>,
        Jiri Olsa <jolsa@...nel.org>, bpf <bpf@...r.kernel.org>,
        Linux Kernel Mailing List <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH] selftests/bpf: simplify cgroup_hierarchical_stats selftest

On Mon, Aug 29, 2022 at 3:15 PM Yosry Ahmed <yosryahmed@...gle.com> wrote:
>
> Hi Hao,
>
> Thanks for taking a look!
>
> On Mon, Aug 29, 2022 at 1:08 PM Hao Luo <haoluo@...gle.com> wrote:
> >
> > On Fri, Aug 26, 2022 at 4:06 PM Yosry Ahmed <yosryahmed@...gle.com> wrote:
> > >
> > > The cgroup_hierarchical_stats selftest is complicated. It has to be,
> > > because it tests an entire workflow of recording, aggregating, and
> > > dumping cgroup stats. However, some of the complexity is unnecessary.
> > > The test now enables the memory controller in a cgroup hierarchy, invokes
> > > reclaim, measure reclaim time, THEN uses that reclaim time to test the
> > > stats collection and aggregation. We don't need to use such a
> > > complicated stat, as the context in which the stat is collected is
> > > orthogonal.
> > >
> > > Simplify the test by using a simple stat instead of reclaim time, the
> > > total number of times a process has ever entered a cgroup. This makes
> > > the test simpler and removes the dependency on the memory controller and
> > > the memory reclaim interface.
> > >
> > > Signed-off-by: Yosry Ahmed <yosryahmed@...gle.com>
> > > ---
> >
> > Yosry, please tag the patch with the repo it should be applied on:
> > bpf-next or bpf.
> >
>
> Will do for v2.
>
> > >
> > > When the test failed on Alexei's setup because the memory controller was
> > > not enabled I realized this is an unnecessary dependency for the test,
> > > which inspired this patch :) I am not sure if this prompt a Fixes tag as
> > > the test wasn't broken.
> > >
> > > ---
> > >  .../prog_tests/cgroup_hierarchical_stats.c    | 157 ++++++---------
> > >  .../bpf/progs/cgroup_hierarchical_stats.c     | 181 ++++++------------
> > >  2 files changed, 118 insertions(+), 220 deletions(-)
> > >
> > [...]
> > > diff --git a/tools/testing/selftests/bpf/progs/cgroup_hierarchical_stats.c b/tools/testing/selftests/bpf/progs/cgroup_hierarchical_stats.c
> > > index 8ab4253a1592..c74362854948 100644
> > > --- a/tools/testing/selftests/bpf/progs/cgroup_hierarchical_stats.c
> > > +++ b/tools/testing/selftests/bpf/progs/cgroup_hierarchical_stats.c
> > > @@ -1,7 +1,5 @@
> > >  // SPDX-License-Identifier: GPL-2.0-only
> > >  /*
> > > - * Functions to manage eBPF programs attached to cgroup subsystems
> > > - *
> >
> > Please also add comments here explaining what the programs in this file do.
> >
>
> Will do.
>
> > >   * Copyright 2022 Google LLC.
> > >   */
> > [...]
> > >
> > > -SEC("tp_btf/mm_vmscan_memcg_reclaim_begin")
> > > -int BPF_PROG(vmscan_start, int order, gfp_t gfp_flags)
> > > +SEC("fentry/cgroup_attach_task")
> >
> > Can we select an attachpoint that is more stable? It seems
> > 'cgroup_attach_task' is an internal helper function in cgroup, and its
> > signature can change. I'd prefer using those commonly used tracepoints
> > and EXPORT'ed functions. IMHO their interfaces are more stable.
> >
>
> Will try to find a more stable attach point. Thanks!

Hey Hao,

I couldn't find any suitable stable attach points under kernel/cgroup.
Most tracepoints are created using TRACE_CGROUP_PATH which only
invokes the tracepoint if the trace event is enabled, which I assume
is not something we can rely on. Otherwise, there is only
trace_cgroup_setup_root() and trace_cgroup_destroy_root() which are
irrelevant here. A lot of EXPORT'ed functions are not called in the
kernel, or cannot be invoked from userspace (the test) in a
straightforward way. Even if they did, future changes to such code
paths can also change in the future, so I don't think there is really
a way to guarantee that future changes don't break the test.

Let me know what you think.

>
> > > +int BPF_PROG(counter, struct cgroup *dst_cgrp, struct task_struct *leader,
> > > +            bool threadgroup)
> > >  {
> > > -       struct task_struct *task = bpf_get_current_task_btf();
> > > -       __u64 *start_time_ptr;
> > > -
> > > -       start_time_ptr = bpf_task_storage_get(&vmscan_start_time, task, 0,
> > > -                                             BPF_LOCAL_STORAGE_GET_F_CREATE);
> > > -       if (start_time_ptr)
> > > -               *start_time_ptr = bpf_ktime_get_ns();
> > > -       return 0;
> > > -}
> > [...]
> > >  }
> > > --
> > > 2.37.2.672.g94769d06f0-goog
> > >