Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
MIME-Version: 1.0
References: <20230525180209.19497-1-james.morse@arm.com> <20230525180209.19497-13-james.morse@arm.com>
In-Reply-To: <20230525180209.19497-13-james.morse@arm.com>
From:   Peter Newman <peternewman@google.com>
Date:   Tue, 6 Jun 2023 10:49:37 +0200
Message-ID: <CALPaoCjLjngabG32m4X8sSwK-bbZ28oL6BVKSn_dFnDwPzkDkg@mail.gmail.com>
Subject: Re: [PATCH v4 12/24] x86/resctrl: Make resctrl_arch_rmid_read() retry
 when it is interrupted
To:     James Morse <james.morse@arm.com>
Cc:     x86@kernel.org, linux-kernel@vger.kernel.org,
        Fenghua Yu <fenghua.yu@intel.com>,
        Reinette Chatre <reinette.chatre@intel.com>,
        Thomas Gleixner <tglx@linutronix.de>,
        Ingo Molnar <mingo@redhat.com>, Borislav Petkov <bp@alien8.de>,
        H Peter Anvin <hpa@zytor.com>,
        Babu Moger <Babu.Moger@amd.com>,
        shameerali.kolothum.thodi@huawei.com,
        D Scott Phillips OS <scott@os.amperecomputing.com>,
        carl@os.amperecomputing.com, lcherian@marvell.com,
        bobo.shaobowang@huawei.com, tan.shaopeng@fujitsu.com,
        xingxin.hx@openanolis.org, baolin.wang@linux.alibaba.com,
        Jamie Iles <quic_jiles@quicinc.com>,
        Xin Hao <xhao@linux.alibaba.com>, dfustini@baylibre.com
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
Precedence: bulk

Hi James,

On Thu, May 25, 2023 at 8:03=E2=80=AFPM James Morse <james.morse@arm.com> w=
rote:
> +interrupted:
> +       am =3D get_arch_mbm_state(hw_dom, rmid, eventid);
> +       if (am)
> +               start_msr_val =3D atomic64_read(&am->prev_msr);
> +
>         ret =3D __rmid_read(rmid, eventid, &msr_val);
>         if (ret)
>                 return ret;
>
>         am =3D get_arch_mbm_state(hw_dom, rmid, eventid);
>         if (am) {
> -               am->chunks +=3D mbm_overflow_count(am->prev_msr, msr_val,
> -                                                hw_res->mbm_width);
> -               chunks =3D get_corrected_mbm_count(rmid, am->chunks);
> -               am->prev_msr =3D msr_val;
> +               old_msr_val =3D atomic64_cmpxchg(&am->prev_msr, start_msr=
_val,
> +                                              msr_val);
> +               if (old_msr_val !=3D start_msr_val)
> +                       goto interrupted;
> +
> +               chunks =3D mbm_overflow_count(start_msr_val, msr_val,
> +                                           hw_res->mbm_width);
> +               atomic64_add(chunks, &am->chunks);
> +
> +               chunks =3D get_corrected_mbm_count(rmid,
> +                                                atomic64_read(&am->chunk=
s));
>         } else {
>                 chunks =3D msr_val;
>         }

It looks like if __rmid_read() is interrupted by an occupancy counter
read between writing QM_EVTSEL and reading QM_CTR, it will not perform
any update to am->prev_msr, and the interrupted read will return the
same counter value as in the interrupting read.

Maybe there's something you can create to check that's updated unconditiona=
lly?

-Peter