MIME-Version: 1.0
References: <cover.1676067791.git.quic_sudaraja@quicinc.com>
 <CAJuCfpHWQ8NV=iR3BN+pt1c8FynCnRqyyriHb1gLxFgiNVrwjA@mail.gmail.com>
 <e944536f-a04c-5528-601e-d7f505a761e8@quicinc.com> <CAJuCfpGLkkS2yx0d9+2nYtEtxANSH5H3EgCmWZax4N-ieEBG7g@mail.gmail.com>
 <15cd8816-b474-0535-d854-41982d3bbe5c@quicinc.com> <CAJuCfpHihLgHCcsAqMJ_o2u7Ux9B5HFGsV2y_L2_5GXYAGYLnw@mail.gmail.com>
 <82406da2-799e-f0b4-bce0-7d47486030d4@quicinc.com> <CAJuCfpHrhO7_fMwNuu2hdQob=MPjZTW8eaJpNhEhPmDMqz2qTA@mail.gmail.com>
 <Y/ix53x8i/ViuBXf@dhcp22.suse.cz> <CAJuCfpG1c5v6qOotPV6t-o1T1p7EbOLs90eFnoEBjts+gafovA@mail.gmail.com>
 <Y/yxSqhpQh9yR7L4@dhcp22.suse.cz>
In-Reply-To: <Y/yxSqhpQh9yR7L4@dhcp22.suse.cz>
From:   Suren Baghdasaryan <surenb@google.com>
Date:   Mon, 27 Feb 2023 09:49:59 -0800
Message-ID: <CAJuCfpHXwbtjdhH_K9r28-LUYtzC6+NaZ+RrKCtiNBk8PfXhLw@mail.gmail.com>
Subject: Re: [PATCH] psi: reduce min window size to 50ms
To:     Michal Hocko <mhocko@suse.com>
Cc:     Sudarshan Rajagopalan <quic_sudaraja@quicinc.com>,
        David Hildenbrand <david@redhat.com>,
        Johannes Weiner <hannes@cmpxchg.org>,
        Mike Rapoport <rppt@kernel.org>,
        Oscar Salvador <osalvador@suse.de>,
        Anshuman Khandual <anshuman.khandual@arm.com>,
        mark.rutland@arm.com, will@kernel.org,
        virtualization@lists.linux-foundation.org, linux-mm@kvack.org,
        linux-kernel@vger.kernel.org, linux-arm-kernel@lists.infradead.org,
        linux-arm-msm@vger.kernel.org,
        Trilok Soni <quic_tsoni@quicinc.com>,
        Sukadev Bhattiprolu <quic_sukadev@quicinc.com>,
        Srivatsa Vaddagiri <quic_svaddagi@quicinc.com>,
        Patrick Daly <quic_pdaly@quicinc.com>, johunt@akamai.com
Content-Type: text/plain; charset="UTF-8"
Precedence: bulk

On Mon, Feb 27, 2023 at 5:34 AM Michal Hocko <mhocko@suse.com> wrote:
>
> On Fri 24-02-23 13:07:57, Suren Baghdasaryan wrote:
> > On Fri, Feb 24, 2023 at 4:47 AM Michal Hocko <mhocko@suse.com> wrote:
> > >
> > > On Tue 14-02-23 11:34:30, Suren Baghdasaryan wrote:
> > > [...]
> > > > Your suggestion to have this limit configurable sounds like obvious
> > > > solution. I would like to get some opinions from other maintainers.
> > > > Johannes, WDYT? CC'ing Michal to chime in as well since this is mostly
> > > > related to memory stalls.
> > >
> > > I do not think that making this configurable helps much. Many users will
> > > be bound to distribution config and also it would be hard to experiment
> > > with a recompile cycle every time. This seems just too impractical.
> > >
> > > Is there any reason why we shouldn't allow any timeout? Shorter
> > > timeouts could be restricted to a priviledged context to avoid an easy
> > > way to swamp system by too frequent polling.
> >
> > Hmm, ok. Maybe then we just ensure that only privileged users can set
> > triggers and remove the min limit (use a >0 check)?
>
> This could break existing userspace which is not privileged. I would
> just go with CAP_SYS_NICE or similar with small (sub min) timeouts.

Yeah, that's what I meant. /proc/pressure/* files already check for
CAP_SYS_RESOURCE
(https://elixir.bootlin.com/linux/latest/source/kernel/sched/psi.c#L1440)
but per-cgroup pressure files do not have this check. I think the
original patch which added this check
(https://lore.kernel.org/all/20210402025833.27599-1-johunt@akamai.com/)
missed the cgroup ones. This should be easy to add but I wonder if
that was left that way intentionally.

CC'ing the author. Josh, Johannes is that inconsistency between system
pressure files and cgroup-specific ones intentional? Can we change
them all to check for CAP_SYS_RESOURCE?

>
> > > Btw. it seems that there is is only a limit on a single trigger per fd
> > > but no limits per user so it doesn't sound too hard to end up with too
> > > much polling even with a larger timeouts. To me it seems like we need to
> > > contain the polling thread to be bound by the cpu controller.
> >
> > Hmm. We have one "psimon" thread per cgroup (+1 system-level one) and
> > poll_min_period for each thread is chosen as the min() of polling
> > periods between triggers created in that group. So, a bad trigger that
> > causes overly aggressive polling and polling thread being throttled,
> > might affect other triggers in that cgroup.
>
> Yes, and why that would be a problem?

If unprivileged processes are allowed to add new triggers then a
malicious process can add a bad trigger and affect other legit
processes. That sounds like a problem to me.
Thanks,
Suren.

> --
> Michal Hocko
> SUSE Labs