Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
MIME-Version: 1.0
References: <1586703004-13674-1-git-send-email-mkshah@codeaurora.org>
 <1586703004-13674-3-git-send-email-mkshah@codeaurora.org> <158708153779.132238.7823999141480422982@swboyd.mtv.corp.google.com>
In-Reply-To: <158708153779.132238.7823999141480422982@swboyd.mtv.corp.google.com>
From:   Doug Anderson <dianders@chromium.org>
Date:   Fri, 17 Apr 2020 14:16:53 -0700
Message-ID: <CAD=FV=WOwMrTNS7FK0jybJyEOvT-rHarA0m5U6B+pKoeA7RCHQ@mail.gmail.com>
Subject: Re: [PATCH v17 2/6] soc: qcom: rpmh: Update dirty flag only when data changes
To:     Stephen Boyd <swboyd@chromium.org>
Cc:     Maulik Shah <mkshah@codeaurora.org>,
        Bjorn Andersson <bjorn.andersson@linaro.org>,
        Evan Green <evgreen@chromium.org>,
        LKML <linux-kernel@vger.kernel.org>,
        linux-arm-msm <linux-arm-msm@vger.kernel.org>,
        Andy Gross <agross@kernel.org>,
        Matthias Kaehlcke <mka@chromium.org>,
        Rajendra Nayak <rnayak@codeaurora.org>,
        Lina Iyer <ilina@codeaurora.org>, lsrao@codeaurora.org
Content-Type: text/plain; charset="UTF-8"
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On Thu, Apr 16, 2020 at 4:59 PM Stephen Boyd <swboyd@chromium.org> wrote:
>
> Quoting Maulik Shah (2020-04-12 07:50:00)
> > diff --git a/drivers/soc/qcom/rpmh.c b/drivers/soc/qcom/rpmh.c
> > index eb0ded0..03630ae 100644
> > --- a/drivers/soc/qcom/rpmh.c
> > +++ b/drivers/soc/qcom/rpmh.c
> > @@ -133,26 +134,27 @@ static struct cache_req *cache_rpm_request(struct rpmh_ctrlr *ctrlr,
> >
> >         req->addr = cmd->addr;
> >         req->sleep_val = req->wake_val = UINT_MAX;
> > -       INIT_LIST_HEAD(&req->list);
> >         list_add_tail(&req->list, &ctrlr->cache);
> >
> >  existing:
> > +       old_sleep_val = req->sleep_val;
> > +       old_wake_val = req->wake_val;
> > +
> >         switch (state) {
> >         case RPMH_ACTIVE_ONLY_STATE:
> > -               if (req->sleep_val != UINT_MAX)
> > -                       req->wake_val = cmd->data;
> > -               break;
> >         case RPMH_WAKE_ONLY_STATE:
> >                 req->wake_val = cmd->data;
> >                 break;
> >         case RPMH_SLEEP_STATE:
> >                 req->sleep_val = cmd->data;
> >                 break;
> > -       default:
> > -               break;
> >         }
> >
> > -       ctrlr->dirty = true;
> > +       ctrlr->dirty = (req->sleep_val != old_sleep_val ||
> > +                       req->wake_val != old_wake_val) &&
> > +                       req->sleep_val != UINT_MAX &&
> > +                       req->wake_val != UINT_MAX;
>
> Can this change ctrl->dirty from true to false? I'm worried that we need
> to make this a saturating assignment instead of an assignment.
>
>         ctrl->dirty = ctrl->dirty || (req->sleep_val != .. );

This seems like a serious problem with the current code and feels like
we need a fix sooner rather than later.  I'm sorry I missed it in
review (and in fact, I probably suggested the exact code that's here
so it's even more my fault).  :(

I posted:

https://lore.kernel.org/r/20200417141531.1.Ia4b74158497213eabad7c3d474c50bfccb3f342e@changeid

-Doug