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
In-Reply-To: <07f1e701-f921-b06a-7492-63810a937322@quicinc.com>
References: <1649970223-9522-1-git-send-email-quic_khsieh@quicinc.com>
 <CAE-0n51rp73v6tod98TX3Y_q8TuOppJVdm9Te_9kSNyqyFuoog@mail.gmail.com> <07f1e701-f921-b06a-7492-63810a937322@quicinc.com>
From:   Stephen Boyd <swboyd@chromium.org>
User-Agent: alot/0.10
Date:   Mon, 25 Apr 2022 16:33:18 -0700
Message-ID: <CAE-0n53=PCDWh--tYOrpEFJG1sVh7V_G5+d0dEhenXsbKFyWxA@mail.gmail.com>
Subject: Re: [PATCH] drm/msm/dp: tear down main link at unplug handle immediately
To:     Kuogee Hsieh <quic_khsieh@quicinc.com>, agross@kernel.org,
        airlied@linux.ie, bjorn.andersson@linaro.org, daniel@ffwll.ch,
        dmitry.baryshkov@linaro.org, robdclark@gmail.com, sean@poorly.run,
        vkoul@kernel.org
Cc:     quic_abhinavk@quicinc.com, quic_aravindh@quicinc.com,
        quic_sbillaka@quicinc.com, freedreno@lists.freedesktop.org,
        dri-devel@lists.freedesktop.org, linux-arm-msm@vger.kernel.org,
        linux-kernel@vger.kernel.org
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
Precedence: bulk

Quoting Kuogee Hsieh (2022-04-25 15:29:30)
>
> On 4/20/2022 3:38 PM, Stephen Boyd wrote:
> > Quoting Kuogee Hsieh (2022-04-14 14:03:43)
> >
> >> diff --git a/drivers/gpu/drm/msm/dp/dp_display.c b/drivers/gpu/drm/msm=
/dp/dp_display.c
> >> index 01453db..f5bd8f5 100644
> >> --- a/drivers/gpu/drm/msm/dp/dp_display.c
> >> +++ b/drivers/gpu/drm/msm/dp/dp_display.c
> >> @@ -615,24 +598,21 @@ static int dp_hpd_unplug_handle(struct dp_displa=
y_private *dp, u32 data)
> >>                  if (dp->link->sink_count =3D=3D 0) {
> >>                          dp_display_host_phy_exit(dp);
> >>                  }
> >> +               dp_display_notify_disconnect(&dp->pdev->dev);
> >>                  mutex_unlock(&dp->event_mutex);
> >>                  return 0;
> >> -       }
> >> -
> >> -       if (state =3D=3D ST_DISCONNECT_PENDING) {
> >> +       } else if (state =3D=3D ST_DISCONNECT_PENDING) {
> >>                  mutex_unlock(&dp->event_mutex);
> >>                  return 0;
> >> -       }
> >> -
> >> -       if (state =3D=3D ST_CONNECT_PENDING) {
> >> -               /* wait until CONNECTED */
> >> -               dp_add_event(dp, EV_HPD_UNPLUG_INT, 0, 1); /* delay =
=3D 1 */
> >> +       } else if (state =3D=3D ST_CONNECT_PENDING) {
> > I take it that ST_CONNECT_PENDING is sort of like "userspace hasn't
> > handled the uevent yet and modeset hasn't been called but the link is
> > setup and now we want to tear it down". The state name may want to be
> > changed to something else.
> yes, how about change to=C2=A0 ST_MAINLINK_READY?

Sure.

> >> @@ -1529,8 +1480,11 @@ int msm_dp_display_enable(struct msm_dp *dp, st=
ruct drm_encoder *encoder)
> >>
> >>          mutex_lock(&dp_display->event_mutex);
> >>
> >> -       /* stop sentinel checking */
> >> -       dp_del_event(dp_display, EV_CONNECT_PENDING_TIMEOUT);
> >> +       state =3D dp_display->hpd_state;
> >> +       if (state !=3D ST_DISPLAY_OFF && state !=3D ST_CONNECT_PENDING=
) {
> > Is this to guard against userspace doing an atomic commit when the
> > display isn't connected? Is that even possible?
>
> No, it used to guard follow scenario in timing order,
>
> 1) plugin had been handled and mainlink is ready,
>
> 2)=C2=A0 userspace hasn't handled the uevent yet and modeset hasn't been =
called
>
> 3) unplugged happen, mainlink be teared down
>
> 4) user space start to response to uevent=C2=A0 and try to enable display=
.
> (it too late since mainlink had been teared down)
>

Ok. Thanks for clarifying.