Re: [PATCH v1 0/7] Add HPD support for Rockchip Analogix DP

[Damon Ding]

Hi Heiko,

On 7/31/2026 5:23 PM, Heiko Stübner wrote:
> Hi Damon,
>
> Am Freitag, 31. Juli 2026, 05:08:45 Mitteleuropäische Sommerzeit schrieb Damon Ding:
> > On 7/31/2026 2:49 AM, Heiko Stübner wrote:
> > > Am Donnerstag, 30. Juli 2026, 05:27:37 Mitteleuropäische Sommerzeit schrieb Damon Ding:
> > > > This series improves the HPD (Hotplug Detect) interrupt handling in
> > > > the Analogix DP driver to enable reliable native HPD pin detection on
> > > > Rockchip platforms, and introduces platform-specific HPD detection
> > > > schemes with fine-grained interrupt control.
> > >
> > > [...]
> > >
> > > > Tested on RK3576 with both native HPD pin and GPIO HPD configurations.
> > > >
> > > > Native HPD pin mode:
> > > >
> > > >     &edp {
> > > >         status = "okay";
> > > >         pinctrl-names = "default";
> > > >         pinctrl-0 = <&edp_txm0_pins>;
> > > >     };
> > > >
> > > > GPIO HPD mode:
> > > >
> > > >     &edp {
> > > >         status = "okay";
> > > >         pinctrl-names = "default";
> > > >         pinctrl-0 = <&edp0_hpd>;
> > > >         hpd-gpios = <&gpio4 RK_PC1 GPIO_ACTIVE_HIGH>;
> > > >     };
> > > >
> > > >     &pinctrl {
> > > >         edp {
> > > >             edp0_hpd: edp0-hpd {
> > > >                 rockchip,pins = <4 RK_PC1 0 &pcfg_pull_none>;
> > > >             };
> > > >         };
> > > >     };
> > >
> > > Yay and thanks a lot for working on that.
> > >
> > > If I roll my display-port carrier board back to the analogix-dp being
> > > the endpoint, this works exactly as expected :-) .
> > >
> > > I can plug in the display at some point after the boot and have it
> > > detected and setup correctly. Same with unplugging it while
> > > the system is running. The connection status is correctly reported
> > > in sysfs.
> > >
> > >
> > > Though there is a caveat. When the eDP is not the last part, this does not
> > > work anymore. The detect function always returns "connected" if there is
> > > a next bridge available.
> > >
> > > My board [0] has a physical DP connector on it, so models it as
> > > dp-connector, but in that case the Analogix driver always reports being
> > > connected.
> > >
> > >
> > > [0] https://lore.kernel.org/linux-rockchip/20251009225050.88192-3-heiko@xxxxxxxxx/
> > > ... same effect with hpd-gpios in either the edp node or dp-connector node
> >
> > My previous tests were carried out on the modified RK3576 IOTEST board
> > DTS based on RK3576 EVB1, which includes dp-connector by default. I also
> > did extra verification without dp-connector, and HPD detection works
> > correctly in both cases.
> >
> > I've taken a look at the Tiger board DT configuration. The pinctrl
> > settings for the eDP HPD GPIO may need adjustment:
> >
> > &pinctrl {
> > 	edp0 {
> > 		edp0_hpd_l: edp0-hpd-l-pin {
> > 			rockchip,pins = <4 RK_PB5 RK_FUNC_GPIO &pcfg_pull_down>;
> > 		};
> > 	};
> > };
> >
> > It is recommended to configure GPIO4B5 as pull-down or pll-none. Pull-up
> > configuration may prevent GPIO interrupts from being triggered properly,
> > resulting in a permanent "connected" status.
>
> At least on that DP-carrier the pin direction is different.
> I.e. it is active-low, meaning 0 for plugged in.
>
> In the schematics I do see a hw pull up too - so pull-none in the pinconfig
> would also be possible, but pull-down is definitly wrong on _this_ board.
>
> So we're talking about the same code, I put the dts up on [0] .
>
> I'm currently running with a printk diff [1] to showcase the issue.
>
> So on boot, with the display unplugged I get:
> [   14.757911] ---> drm_bridge_connector_detect: detect
> [   14.757914] ---> analogix_dp_bridge_detect a
> [   14.757918] ---> drm_bridge_connector_detect: detect
> [   14.773779] ---> analogix_dp_bridge_detect a
> [   14.838658] rockchip-dp fdec0000.edp: failed to read dpcd caps: -110
> [   14.838662] [drm:analogix_dp_bridge_atomic_enable [analogix_dp]] *ERROR* dp commit error, ret = -110
> [   14.838670] rockchip-dp fdec0000.edp: failed to set bridge, retry: 0
> [   14.884791] rockchip-dp fdec0000.edp: failed to read dpcd caps: -110
> [   14.884793] [drm:analogix_dp_bridge_atomic_enable [analogix_dp]] *ERROR* dp commit error, ret = -110
> [   14.884796] rockchip-dp fdec0000.edp: failed to set bridge, retry: 1
> [   14.930925] rockchip-dp fdec0000.edp: failed to read dpcd caps: -110
> [   14.930927] [drm:analogix_dp_bridge_atomic_enable [analogix_dp]] *ERROR* dp commit error, ret = -110
> [   14.930930] rockchip-dp fdec0000.edp: failed to set bridge, retry: 2
> [   14.977032] rockchip-dp fdec0000.edp: failed to read dpcd caps: -110
> [   14.977034] [drm:analogix_dp_bridge_atomic_enable [analogix_dp]] *ERROR* dp commit error, ret = -110
> [   14.977037] rockchip-dp fdec0000.edp: failed to set bridge, retry: 3
> [   15.023150] rockchip-dp fdec0000.edp: failed to read dpcd caps: -110
> [   15.023152] [drm:analogix_dp_bridge_atomic_enable [analogix_dp]] *ERROR* dp commit error, ret = -110
> [   15.023155] rockchip-dp fdec0000.edp: failed to set bridge, retry: 4
> [   15.023168] rockchip-dp fdec0000.edp: too many times retry set bridge, give it up
> [   15.052485] Console: switching to colour frame buffer device 80x30
> [   15.210457] rockchip-drm display-subsystem: [drm] fb0: rockchipdrmfb frame buffer device
> [   15.223444] ---> drm_bridge_connector_detect: detect
> [   15.231974] ---> analogix_dp_bridge_detect a
> [   15.239587] ---> drm_bridge_connector_detect: detect
> [   15.245341] ---> analogix_dp_bridge_detect a
>
> So analogix_dp_bridge_detect() is called and always returns connected.
>
> Plugging in the the display, I get:
> [  290.592096] ---> display_connector_detect gpio value is 1   {display-connector returning connected}
> [  290.598648] ---> drm_bridge_connector_detect: detect
> [  290.604637] ---> analogix_dp_bridge_detect a
> [  290.609826] ---> drm_bridge_connector_detect: detect
> [  290.615784] ---> analogix_dp_bridge_detect a
>
> Unplugging the display, I get:
> [  302.081444] ---> display_connector_detect gpio value is 0   {display-connector returning disconnected}
> [  302.087996] ---> drm_bridge_connector_detect: detect
> [  302.093945] ---> analogix_dp_bridge_detect a
> [  302.099131] ---> drm_bridge_connector_detect: detect
> [  302.105090] ---> analogix_dp_bridge_detect a
> [  302.144834] ---> drm_bridge_connector_detect: detect
> [  302.150784] ---> analogix_dp_bridge_detect a
> [  302.155933] ---> drm_bridge_connector_detect: detect
> [  302.161859] ---> analogix_dp_bridge_detect a
>
>
> So it looks like on the gpio-side the detection is correct and/but
> analogix_dp_bridge_detect() is always returning a connected state
> ("b" output is never reached).
>
> I'm still working on understanding how the whole detection logic works
> in drm ;-) .

I think I've spotted the environmental difference between our setups.
Your DTS places the hpd-gpios property under the dp-connector node,
which is a scenario I haven't covered yet. I will reproduce and test 
this case locally.

As a quick experiment, could you try removing hpd-gpios from 
dp-connector and move it under the edp node instead? As expected, GPIO 
HPD should work properly with that configuration.

Best regards,
Damon

> [0] https://github.com/mmind/linux-rockchip/blob/dev/v7.3/rk3588-tiger-dp-carrier-v5/arch/arm64/boot/dts/rockchip/rk3588-tiger-displayport-carrier.dts
> [1]
> diff --git a/drivers/gpu/drm/bridge/analogix/analogix_dp_core.c b/drivers/gpu/drm/bridge/analogix/analogix_dp_core.c
> index 7a89548a7a885..0accdf62637ae 100644
> --- a/drivers/gpu/drm/bridge/analogix/analogix_dp_core.c
> +++ b/drivers/gpu/drm/bridge/analogix/analogix_dp_core.c
> @@ -915,8 +915,10 @@ analogix_dp_bridge_detect(struct drm_bridge *bridge, struct drm_connector *conne
>          struct analogix_dp_device *dp = to_dp(bridge);
>          enum drm_connector_status status = connector_status_disconnected;
>   
> +printk("---> %s a\n", __func__);
>          if (dp->plat_data->next_bridge)
>                  return connector_status_connected;
> +printk("---> %s b\n", __func__);
>   
>          if (!analogix_dp_detect_hpd(dp))
>                  status = connector_status_connected;
> diff --git a/drivers/gpu/drm/bridge/display-connector.c b/drivers/gpu/drm/bridge/display-connector.c
> index 16c0631adeb18..b0c07694b82e3 100644
> --- a/drivers/gpu/drm/bridge/display-connector.c
> +++ b/drivers/gpu/drm/bridge/display-connector.c
> @@ -45,6 +45,7 @@ static enum drm_connector_status display_connector_detect(struct drm_bridge *bri
>          struct display_connector *conn = to_display_connector(bridge);
>   
>          if (conn->hpd_gpio) {
> +printk("---> %s gpio value is %d\n", __func__, gpiod_get_value_cansleep(conn->hpd_gpio));
>                  if (gpiod_get_value_cansleep(conn->hpd_gpio))
>                          return connector_status_connected;
>                  else
> @@ -84,6 +85,7 @@ static enum drm_connector_status display_connector_detect(struct drm_bridge *bri
>   static enum drm_connector_status
>   display_connector_bridge_detect(struct drm_bridge *bridge, struct drm_connector *connector)
>   {
> +printk("---> %s\n", __func__);
>          return display_connector_detect(bridge);
>   }
>   
> diff --git a/drivers/gpu/drm/display/drm_bridge_connector.c b/drivers/gpu/drm/display/drm_bridge_connector.c
> index 649969fca1413..8816d837c9a77 100644
> --- a/drivers/gpu/drm/display/drm_bridge_connector.c
> +++ b/drivers/gpu/drm/display/drm_bridge_connector.c
> @@ -218,6 +218,7 @@ drm_bridge_connector_detect(struct drm_connector *connector, bool force)
>          enum drm_connector_status status;
>   
>          if (detect) {
> +printk("---> %s: detect\n", __func__);
>                  status = detect->funcs->detect(detect, connector);
>   
>                  if (hdmi)