Re: drm/msm/dp: hold one runtime PM reference per plugged state
From: Dmitry Baryshkov
Date: Wed Nov 11 2026 - 12:54:02 EST
On Wed, Oct 07, 2026 at 01:39:00AM -0300, Val Packett wrote:
>
> On 9/29/26 10:06 AM, Joonhoe Kim wrote:
> > The plug handler takes a runtime PM reference every time it runs and
> > sets dp->plugged; the unplug handler drops a single reference when
> > dp->plugged is set. msm_dp_bridge_detect() also takes a reference and
> > marks the sink plugged, and keeps that reference as well -- on every
> > detect while the sink stays connected. On top of that, in the current
> > tree the plug handler runs more than once per plug:
> > drm_bridge_connector_detect() calls .detect() and then .hpd_notify(),
> > and pmic_glink altmode reports an IRQ_HPD as "connected" too. Every
> > extra reference is leaked on unplug.
> >
> > On a Lenovo TB323FU (SM8850, DP over the USB-C port via pmic_glink
> > altmode) the DP controller's usage count reached 28-34 after a few
> > plugs (rpm_usage tracepoint), so the controller never runtime-suspended
> > after the display was unplugged. It in turn kept the MDSS core GDSC on
> > while all displays were off.
> >
> > Make the plugged state own exactly one reference: the plug handler
> > only takes it when the sink was not plugged yet, and detect() returns
> > its own reference when the plugged state already holds one, or drops
> > both when it finds the sink gone.
> >
> > With this the count stays at 1-3 while a display is connected, the
> > controller suspends 3-4 s after unplug, MDSS suspends and its GDSC
> > powers off on DPMS off, and replugging (quickly or after 20 s) still
> > brings the picture back. Only tested on this device.
> >
> > Fixes: 3ea2d1c3d154 ("drm/msm/dp: turn link_ready into plugged")
> > Assisted-by: LLM
> > Signed-off-by: Joonhoe Kim <26rote@xxxxxxxxx>
> > ---
> > Related pending work: "drm: handle IRQ_HPD events correctly" v5
> > (https://lore.kernel.org/all/20260920-hpd-irq-events-v5-0-ff1de0090c72@xxxxxxxxxxxxxxxx/)
> > removes the IRQ_HPD-as-plug path, and MST v7 26/35
> > (https://lore.kernel.org/all/20260929-msm-dp-mst-v7-26-4b81473185e7@xxxxxxxxxxxxxxxx/)
> > the detect-based hpd_notify for DP; the reference that
> > msm_dp_bridge_detect() keeps is leaked either way. MST v7 27/35 touches
> > the same lines as the first hunk; happy to rebase on top of it.
>
> Also kinda related / touching the same places, tho not yet sent to the lists
> AFAIK:
>
> "drm/msm/dp: Mark the link bad when an active sink is replugged"
> https://github.com/radxa/kernel/commit/eb3247f6f9a39dcb15cf4b1df6d4732b1489b410
>
> > drivers/gpu/drm/msm/dp/dp_display.c | 23 +++++++++++++++++++----
> > 1 file changed, 19 insertions(+), 4 deletions(-)
> >
> >
> > base-commit: 6375e61c01e93e35ee7acd336a689ac1fae4b509
> >
> > diff --git a/drivers/gpu/drm/msm/dp/dp_display.c b/drivers/gpu/drm/msm/dp/dp_display.c
> > index 6abe30bdff12..4e47b659c71d 100644
> > --- a/drivers/gpu/drm/msm/dp/dp_display.c
> > +++ b/drivers/gpu/drm/msm/dp/dp_display.c
> > @@ -412,10 +412,17 @@ static int msm_dp_hpd_plug_handle(struct msm_dp_display_private *dp)
> > guard(mutex)(&dp->plugged_lock);
> > - ret = pm_runtime_resume_and_get(&pdev->dev);
> > - if (ret) {
> > - DRM_ERROR("failed to pm_runtime_resume\n");
> > - return ret;
> > + /*
> > + * The plugged state holds one runtime PM reference, dropped by the
> > + * unplug handler. The Type-C side repeats "connected" notifications,
> > + * and detect() may have seen the sink first: take it only once.
> > + */
> > + if (!dp->plugged) {
> > + ret = pm_runtime_resume_and_get(&pdev->dev);
> > + if (ret) {
> > + DRM_ERROR("failed to pm_runtime_resume\n");
> > + return ret;
> > + }
> > }
> > msm_dp_aux_enable_xfers(dp->aux, true);
> > @@ -897,6 +904,7 @@ enum drm_connector_status msm_dp_bridge_detect(struct drm_bridge *bridge,
> > u8 dpcd[DP_RECEIVER_CAP_SIZE];
> > struct drm_dp_desc desc;
> > bool phy_deinit;
> > + bool was_plugged;
> > int ret;
> > dp = to_dp_bridge(bridge)->msm_dp_display;
> > @@ -910,6 +918,7 @@ enum drm_connector_status msm_dp_bridge_detect(struct drm_bridge *bridge,
> > return status;
> > }
> > + was_plugged = priv->plugged;
> > phy_deinit = msm_dp_display_host_phy_init(priv);
> > msm_dp_aux_enable_xfers(priv->aux, true);
> > @@ -960,6 +969,12 @@ enum drm_connector_status msm_dp_bridge_detect(struct drm_bridge *bridge,
> > }
> > pm_runtime_put_sync(&dp->pdev->dev);
> > + /* the sink is gone: drop the plugged state's reference too */
> > + if (was_plugged)
> > + pm_runtime_put_sync(&dp->pdev->dev);
> > + } else if (was_plugged) {
> > + /* the plugged state already holds its reference */
> > + pm_runtime_put_sync(&dp->pdev->dev);
> > }
> > return status;
> This last chunk looks odd, is that not equivalent to just doing
>
> if (was_plugged) pm_runtime_put_sync(&dp->pdev->dev);
>
> on the outer level, outside of the outer if/else?
I'll fix that while applying.
Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@xxxxxxxxxxxxxxxx>
--
With best wishes
Dmitry