drm: xlnx: zynqmp_dpsub: also call drm_helper_hpd_irq_event

Commit Message

Steffen Dirkwinkel Sept. 23, 2024, 7:48 a.m. UTC

From: Steffen Dirkwinkel <[e-mail]>

With hpd going through the bridge as of commit eb2d64bfcc17
("drm: xlnx: zynqmp_dpsub: Report HPD through the bridge")
we don't get hotplug events in userspace on zynqmp hardware.
Also sending hotplug events with drm_helper_hpd_irq_event works.

Fixes: eb2d64bfcc17 ("drm: xlnx: zynqmp_dpsub: Report HPD through the bridge")
Signed-off-by: Steffen Dirkwinkel <[e-mail]>
---
 drivers/gpu/drm/xlnx/zynqmp_dp.c | 4 ++++
 1 file changed, 4 insertions(+)

Comments

Laurent Pinchart Sept. 24, 2024, 6:43 p.m. UTC | #1

Hi Steffen,

Thank you for the patch.

On Mon, Sep 23, 2024 at 09:48:03AM +0200, [e-mail] wrote:
> From: Steffen Dirkwinkel <[e-mail]>
> 
> With hpd going through the bridge as of commit eb2d64bfcc17
> ("drm: xlnx: zynqmp_dpsub: Report HPD through the bridge")
> we don't get hotplug events in userspace on zynqmp hardware.
> Also sending hotplug events with drm_helper_hpd_irq_event works.

Why does the driver need to call both drm_helper_hpd_irq_event() and
drm_bridge_hpd_notify() ? The latter should end up calling
drm_kms_helper_connector_hotplug_event(), which is the same function
that drm_helper_hpd_irq_event() calls.

> Fixes: eb2d64bfcc17 ("drm: xlnx: zynqmp_dpsub: Report HPD through the bridge")
> Signed-off-by: Steffen Dirkwinkel <[e-mail]>
> ---
>  drivers/gpu/drm/xlnx/zynqmp_dp.c | 4 ++++
>  1 file changed, 4 insertions(+)
> 
> diff --git a/drivers/gpu/drm/xlnx/zynqmp_dp.c b/drivers/gpu/drm/xlnx/zynqmp_dp.c
> index 1846c4971fd8..cb823540a412 100644
> --- a/drivers/gpu/drm/xlnx/zynqmp_dp.c
> +++ b/drivers/gpu/drm/xlnx/zynqmp_dp.c
> @@ -17,6 +17,7 @@
>  #include <drm/drm_fourcc.h>
>  #include <drm/drm_modes.h>
>  #include <drm/drm_of.h>
> +#include <drm/drm_probe_helper.h>
>  
>  #include <linux/clk.h>
>  #include <linux/delay.h>
> @@ -1614,6 +1615,9 @@ static void zynqmp_dp_hpd_work_func(struct work_struct *work)
>  					    hpd_work.work);
>  	enum drm_connector_status status;
>  
> +	if (dp->bridge.dev)
> +		drm_helper_hpd_irq_event(dp->bridge.dev);
> +
>  	status = zynqmp_dp_bridge_detect(&dp->bridge);
>  	drm_bridge_hpd_notify(&dp->bridge, status);
>  }

Steffen Dirkwinkel Sept. 25, 2024, 7:54 a.m. UTC | #2

Hi Laurent,

On Tue, 2024-09-24 at 21:43 +0300, Laurent Pinchart wrote:
> Hi Steffen,
> 
> Thank you for the patch.
> 
> On Mon, Sep 23, 2024 at 09:48:03AM +0200, [e-mail] wrote:
> > From: Steffen Dirkwinkel <[e-mail]>
> > 
> > With hpd going through the bridge as of commit eb2d64bfcc17
> > ("drm: xlnx: zynqmp_dpsub: Report HPD through the bridge")
> > we don't get hotplug events in userspace on zynqmp hardware.
> > Also sending hotplug events with drm_helper_hpd_irq_event works.
> 
> Why does the driver need to call both drm_helper_hpd_irq_event() and
> drm_bridge_hpd_notify() ? The latter should end up calling
> drm_kms_helper_connector_hotplug_event(), which is the same function
> that drm_helper_hpd_irq_event() calls.

I don't know why we need drm_helper_hpd_irq_event.
I'll try to trace what happens on hotplug.

> 
> > Fixes: eb2d64bfcc17 ("drm: xlnx: zynqmp_dpsub: Report HPD through
> > the bridge")
> > Signed-off-by: Steffen Dirkwinkel <[e-mail]>
> > ---
> >  drivers/gpu/drm/xlnx/zynqmp_dp.c | 4 ++++
> >  1 file changed, 4 insertions(+)
> > 
> > diff --git a/drivers/gpu/drm/xlnx/zynqmp_dp.c
> > b/drivers/gpu/drm/xlnx/zynqmp_dp.c
> > index 1846c4971fd8..cb823540a412 100644
> > --- a/drivers/gpu/drm/xlnx/zynqmp_dp.c
> > +++ b/drivers/gpu/drm/xlnx/zynqmp_dp.c
> > @@ -17,6 +17,7 @@
> >  #include <drm/drm_fourcc.h>
> >  #include <drm/drm_modes.h>
> >  #include <drm/drm_of.h>
> > +#include <drm/drm_probe_helper.h>
> >  
> >  #include <linux/clk.h>
> >  #include <linux/delay.h>
> > @@ -1614,6 +1615,9 @@ static void zynqmp_dp_hpd_work_func(struct
> > work_struct *work)
> >  					    hpd_work.work);
> >  	enum drm_connector_status status;
> >  
> > +	if (dp->bridge.dev)
> > +		drm_helper_hpd_irq_event(dp->bridge.dev);
> > +
> >  	status = zynqmp_dp_bridge_detect(&dp->bridge);
> >  	drm_bridge_hpd_notify(&dp->bridge, status);
> >  }
>

Laurent Pinchart Sept. 25, 2024, 4:36 p.m. UTC | #3

Hi Steffen,

On Wed, Sep 25, 2024 at 09:54:18AM +0200, Steffen Dirkwinkel wrote:
> On Tue, 2024-09-24 at 21:43 +0300, Laurent Pinchart wrote:
> > On Mon, Sep 23, 2024 at 09:48:03AM +0200, [e-mail] wrote:
> > > From: Steffen Dirkwinkel <[e-mail]>
> > > 
> > > With hpd going through the bridge as of commit eb2d64bfcc17
> > > ("drm: xlnx: zynqmp_dpsub: Report HPD through the bridge")
> > > we don't get hotplug events in userspace on zynqmp hardware.
> > > Also sending hotplug events with drm_helper_hpd_irq_event works.
> > 
> > Why does the driver need to call both drm_helper_hpd_irq_event() and
> > drm_bridge_hpd_notify() ? The latter should end up calling
> > drm_kms_helper_connector_hotplug_event(), which is the same function
> > that drm_helper_hpd_irq_event() calls.
> 
> I don't know why we need drm_helper_hpd_irq_event.
> I'll try to trace what happens on hotplug.

Thank you. Let's try to find the best solution based on your findings.

> > > Fixes: eb2d64bfcc17 ("drm: xlnx: zynqmp_dpsub: Report HPD through
> > > the bridge")
> > > Signed-off-by: Steffen Dirkwinkel <[e-mail]>
> > > ---
> > >  drivers/gpu/drm/xlnx/zynqmp_dp.c | 4 ++++
> > >  1 file changed, 4 insertions(+)
> > > 
> > > diff --git a/drivers/gpu/drm/xlnx/zynqmp_dp.c
> > > b/drivers/gpu/drm/xlnx/zynqmp_dp.c
> > > index 1846c4971fd8..cb823540a412 100644
> > > --- a/drivers/gpu/drm/xlnx/zynqmp_dp.c
> > > +++ b/drivers/gpu/drm/xlnx/zynqmp_dp.c
> > > @@ -17,6 +17,7 @@
> > >  #include <drm/drm_fourcc.h>
> > >  #include <drm/drm_modes.h>
> > >  #include <drm/drm_of.h>
> > > +#include <drm/drm_probe_helper.h>
> > >  
> > >  #include <linux/clk.h>
> > >  #include <linux/delay.h>
> > > @@ -1614,6 +1615,9 @@ static void zynqmp_dp_hpd_work_func(struct
> > > work_struct *work)
> > >  					    hpd_work.work);
> > >  	enum drm_connector_status status;
> > >  
> > > +	if (dp->bridge.dev)
> > > +		drm_helper_hpd_irq_event(dp->bridge.dev);
> > > +
> > >  	status = zynqmp_dp_bridge_detect(&dp->bridge);
> > >  	drm_bridge_hpd_notify(&dp->bridge, status);
> > >  }

Steffen Dirkwinkel Oct. 9, 2024, 2:28 p.m. UTC | #4

Hi Laurent,


On Wed, 2024-09-25 at 19:36 +0300, Laurent Pinchart wrote:
> Hi Steffen,
> 
> On Wed, Sep 25, 2024 at 09:54:18AM +0200, Steffen Dirkwinkel wrote:
> > On Tue, 2024-09-24 at 21:43 +0300, Laurent Pinchart wrote:
> > > On Mon, Sep 23, 2024 at 09:48:03AM +0200, [e-mail] wrote:
> > > > From: Steffen Dirkwinkel <[e-mail]>
> > > > 
> > > > With hpd going through the bridge as of commit eb2d64bfcc17
> > > > ("drm: xlnx: zynqmp_dpsub: Report HPD through the bridge")
> > > > we don't get hotplug events in userspace on zynqmp hardware.
> > > > Also sending hotplug events with drm_helper_hpd_irq_event
> > > > works.
> > > 
> > > Why does the driver need to call both drm_helper_hpd_irq_event()
> > > and
> > > drm_bridge_hpd_notify() ? The latter should end up calling
> > > drm_kms_helper_connector_hotplug_event(), which is the same
> > > function
> > > that drm_helper_hpd_irq_event() calls.
> > 
> > I don't know why we need drm_helper_hpd_irq_event.
> > I'll try to trace what happens on hotplug.
> 
> Thank you. Let's try to find the best solution based on your
> findings.

There's just nothing registering for hpd with
"drm_bridge_connector_enable_hpd" or "drm_bridge_hpd_enable". I'm not
sure what the correct way to implement this is. In
"drivers/gpu/drm/bridge/ti-tfp410.c" the driver registers for the
callback and calls "drm_helper_hpd_irq_event" in the callback. I guess
we could also do that, but then we might as well call
drm_helper_hpd_irq_event directly? Some other drivers just call both
like I did here. (drivers/gpu/drm/mediatek/mtk_hdmi.c for example)
For "drivers/gpu/drm/msm/hdmi/hdmi_bridge.c" I also can't find the hpd
enable call and it just calls drm_bridge_hpd_notify.

> 
> > > > Fixes: eb2d64bfcc17 ("drm: xlnx: zynqmp_dpsub: Report HPD
> > > > through
> > > > the bridge")
> > > > Signed-off-by: Steffen Dirkwinkel <[e-mail]>
> > > > ---
> > > >  drivers/gpu/drm/xlnx/zynqmp_dp.c | 4 ++++
> > > >  1 file changed, 4 insertions(+)
> > > > 
> > > > diff --git a/drivers/gpu/drm/xlnx/zynqmp_dp.c
> > > > b/drivers/gpu/drm/xlnx/zynqmp_dp.c
> > > > index 1846c4971fd8..cb823540a412 100644
> > > > --- a/drivers/gpu/drm/xlnx/zynqmp_dp.c
> > > > +++ b/drivers/gpu/drm/xlnx/zynqmp_dp.c
> > > > @@ -17,6 +17,7 @@
> > > >  #include <drm/drm_fourcc.h>
> > > >  #include <drm/drm_modes.h>
> > > >  #include <drm/drm_of.h>
> > > > +#include <drm/drm_probe_helper.h>
> > > >  
> > > >  #include <linux/clk.h>
> > > >  #include <linux/delay.h>
> > > > @@ -1614,6 +1615,9 @@ static void
> > > > zynqmp_dp_hpd_work_func(struct
> > > > work_struct *work)
> > > >  					    hpd_work.work);
> > > >  	enum drm_connector_status status;
> > > >  
> > > > +	if (dp->bridge.dev)
> > > > +		drm_helper_hpd_irq_event(dp->bridge.dev);
> > > > +
> > > >  	status = zynqmp_dp_bridge_detect(&dp->bridge);
> > > >  	drm_bridge_hpd_notify(&dp->bridge, status);
> > > >  }
>

Dmitry Baryshkov Oct. 19, 2024, 3:11 p.m. UTC | #5

On Wed, Oct 09, 2024 at 04:28:26PM +0200, Steffen Dirkwinkel wrote:
> Hi Laurent,
> 
> 
> On Wed, 2024-09-25 at 19:36 +0300, Laurent Pinchart wrote:
> > Hi Steffen,
> > 
> > On Wed, Sep 25, 2024 at 09:54:18AM +0200, Steffen Dirkwinkel wrote:
> > > On Tue, 2024-09-24 at 21:43 +0300, Laurent Pinchart wrote:
> > > > On Mon, Sep 23, 2024 at 09:48:03AM +0200, [e-mail] wrote:
> > > > > From: Steffen Dirkwinkel <[e-mail]>
> > > > > 
> > > > > With hpd going through the bridge as of commit eb2d64bfcc17
> > > > > ("drm: xlnx: zynqmp_dpsub: Report HPD through the bridge")
> > > > > we don't get hotplug events in userspace on zynqmp hardware.
> > > > > Also sending hotplug events with drm_helper_hpd_irq_event
> > > > > works.
> > > > 
> > > > Why does the driver need to call both drm_helper_hpd_irq_event()
> > > > and
> > > > drm_bridge_hpd_notify() ? The latter should end up calling
> > > > drm_kms_helper_connector_hotplug_event(), which is the same
> > > > function
> > > > that drm_helper_hpd_irq_event() calls.
> > > 
> > > I don't know why we need drm_helper_hpd_irq_event.
> > > I'll try to trace what happens on hotplug.
> > 
> > Thank you. Let's try to find the best solution based on your
> > findings.
> 
> There's just nothing registering for hpd with
> "drm_bridge_connector_enable_hpd" or "drm_bridge_hpd_enable". I'm not
> sure what the correct way to implement this is. In
> "drivers/gpu/drm/bridge/ti-tfp410.c" the driver registers for the
> callback and calls "drm_helper_hpd_irq_event" in the callback. I guess
> we could also do that, but then we might as well call
> drm_helper_hpd_irq_event directly? Some other drivers just call both
> like I did here. (drivers/gpu/drm/mediatek/mtk_hdmi.c for example)
> For "drivers/gpu/drm/msm/hdmi/hdmi_bridge.c" I also can't find the hpd
> enable call and it just calls drm_bridge_hpd_notify.

The drm_bridge_connector handles enabling it for you when the driver
calls drm_kms_helper_poll_init() / drm_kms_helper_poll_enable(). It
seems zynqmp_kms calls drm_kms_helper_poll_init() too early, before
creating DP chain, so the HPD doesn't get enabled.

> 
> > 
> > > > > Fixes: eb2d64bfcc17 ("drm: xlnx: zynqmp_dpsub: Report HPD
> > > > > through
> > > > > the bridge")
> > > > > Signed-off-by: Steffen Dirkwinkel <[e-mail]>
> > > > > ---
> > > > >  drivers/gpu/drm/xlnx/zynqmp_dp.c | 4 ++++
> > > > >  1 file changed, 4 insertions(+)
> > > > > 
> > > > > diff --git a/drivers/gpu/drm/xlnx/zynqmp_dp.c
> > > > > b/drivers/gpu/drm/xlnx/zynqmp_dp.c
> > > > > index 1846c4971fd8..cb823540a412 100644
> > > > > --- a/drivers/gpu/drm/xlnx/zynqmp_dp.c
> > > > > +++ b/drivers/gpu/drm/xlnx/zynqmp_dp.c
> > > > > @@ -17,6 +17,7 @@
> > > > >  #include <drm/drm_fourcc.h>
> > > > >  #include <drm/drm_modes.h>
> > > > >  #include <drm/drm_of.h>
> > > > > +#include <drm/drm_probe_helper.h>
> > > > >  
> > > > >  #include <linux/clk.h>
> > > > >  #include <linux/delay.h>
> > > > > @@ -1614,6 +1615,9 @@ static void
> > > > > zynqmp_dp_hpd_work_func(struct
> > > > > work_struct *work)
> > > > >  					    hpd_work.work);
> > > > >  	enum drm_connector_status status;
> > > > >  
> > > > > +	if (dp->bridge.dev)
> > > > > +		drm_helper_hpd_irq_event(dp->bridge.dev);
> > > > > +
> > > > >  	status = zynqmp_dp_bridge_detect(&dp->bridge);
> > > > >  	drm_bridge_hpd_notify(&dp->bridge, status);
> > > > >  }
> > 
>

Steffen Dirkwinkel Oct. 21, 2024, 1:19 p.m. UTC | #6

On Sat, 2024-10-19 at 18:11 +0300, Dmitry Baryshkov wrote:
> On Wed, Oct 09, 2024 at 04:28:26PM +0200, Steffen Dirkwinkel wrote:
> > Hi Laurent,
> > 
> > 
> > On Wed, 2024-09-25 at 19:36 +0300, Laurent Pinchart wrote:
> > > Hi Steffen,
> > > 
> > > On Wed, Sep 25, 2024 at 09:54:18AM +0200, Steffen Dirkwinkel
> > > wrote:
> > > > On Tue, 2024-09-24 at 21:43 +0300, Laurent Pinchart wrote:
> > > > > On Mon, Sep 23, 2024 at 09:48:03AM +0200,
> > > > > [e-mail] wrote:
> > > > > > From: Steffen Dirkwinkel <[e-mail]>
> > > > > > 
> > > > > > With hpd going through the bridge as of commit eb2d64bfcc17
> > > > > > ("drm: xlnx: zynqmp_dpsub: Report HPD through the bridge")
> > > > > > we don't get hotplug events in userspace on zynqmp
> > > > > > hardware.
> > > > > > Also sending hotplug events with drm_helper_hpd_irq_event
> > > > > > works.
> > > > > 
> > > > > Why does the driver need to call both
> > > > > drm_helper_hpd_irq_event()
> > > > > and
> > > > > drm_bridge_hpd_notify() ? The latter should end up calling
> > > > > drm_kms_helper_connector_hotplug_event(), which is the same
> > > > > function
> > > > > that drm_helper_hpd_irq_event() calls.
> > > > 
> > > > I don't know why we need drm_helper_hpd_irq_event.
> > > > I'll try to trace what happens on hotplug.
> > > 
> > > Thank you. Let's try to find the best solution based on your
> > > findings.
> > 
> > There's just nothing registering for hpd with
> > "drm_bridge_connector_enable_hpd" or "drm_bridge_hpd_enable". I'm
> > not
> > sure what the correct way to implement this is. In
> > "drivers/gpu/drm/bridge/ti-tfp410.c" the driver registers for the
> > callback and calls "drm_helper_hpd_irq_event" in the callback. I
> > guess
> > we could also do that, but then we might as well call
> > drm_helper_hpd_irq_event directly? Some other drivers just call
> > both
> > like I did here. (drivers/gpu/drm/mediatek/mtk_hdmi.c for example)
> > For "drivers/gpu/drm/msm/hdmi/hdmi_bridge.c" I also can't find the
> > hpd
> > enable call and it just calls drm_bridge_hpd_notify.
> 
> The drm_bridge_connector handles enabling it for you when the driver
> calls drm_kms_helper_poll_init() / drm_kms_helper_poll_enable(). It
> seems zynqmp_kms calls drm_kms_helper_poll_init() too early, before
> creating DP chain, so the HPD doesn't get enabled.
Yes, that's it. Thank you

I'll send a patch for it.


> 
> > 
> > > 
> > > > > > Fixes: eb2d64bfcc17 ("drm: xlnx: zynqmp_dpsub: Report HPD
> > > > > > through
> > > > > > the bridge")
> > > > > > Signed-off-by: Steffen Dirkwinkel
> > > > > > <[e-mail]>
> > > > > > ---
> > > > > >  drivers/gpu/drm/xlnx/zynqmp_dp.c | 4 ++++
> > > > > >  1 file changed, 4 insertions(+)
> > > > > > 
> > > > > > diff --git a/drivers/gpu/drm/xlnx/zynqmp_dp.c
> > > > > > b/drivers/gpu/drm/xlnx/zynqmp_dp.c
> > > > > > index 1846c4971fd8..cb823540a412 100644
> > > > > > --- a/drivers/gpu/drm/xlnx/zynqmp_dp.c
> > > > > > +++ b/drivers/gpu/drm/xlnx/zynqmp_dp.c
> > > > > > @@ -17,6 +17,7 @@
> > > > > >  #include <drm/drm_fourcc.h>
> > > > > >  #include <drm/drm_modes.h>
> > > > > >  #include <drm/drm_of.h>
> > > > > > +#include <drm/drm_probe_helper.h>
> > > > > >  
> > > > > >  #include <linux/clk.h>
> > > > > >  #include <linux/delay.h>
> > > > > > @@ -1614,6 +1615,9 @@ static void
> > > > > > zynqmp_dp_hpd_work_func(struct
> > > > > > work_struct *work)
> > > > > >  					   
> > > > > > hpd_work.work);
> > > > > >  	enum drm_connector_status status;
> > > > > >  
> > > > > > +	if (dp->bridge.dev)
> > > > > > +		drm_helper_hpd_irq_event(dp->bridge.dev);
> > > > > > +
> > > > > >  	status = zynqmp_dp_bridge_detect(&dp->bridge);
> > > > > >  	drm_bridge_hpd_notify(&dp->bridge, status);
> > > > > >  }
> > > 
> > 
>

Patch

diff --git a/drivers/gpu/drm/xlnx/zynqmp_dp.c b/drivers/gpu/drm/xlnx/zynqmp_dp.c
index 1846c4971fd8..cb823540a412 100644
--- a/drivers/gpu/drm/xlnx/zynqmp_dp.c
+++ b/drivers/gpu/drm/xlnx/zynqmp_dp.c
@@ -17,6 +17,7 @@ 
 #include <drm/drm_fourcc.h>
 #include <drm/drm_modes.h>
 #include <drm/drm_of.h>
+#include <drm/drm_probe_helper.h>
 
 #include <linux/clk.h>
 #include <linux/delay.h>
@@ -1614,6 +1615,9 @@  static void zynqmp_dp_hpd_work_func(struct work_struct *work)
 					    hpd_work.work);
 	enum drm_connector_status status;
 
+	if (dp->bridge.dev)
+		drm_helper_hpd_irq_event(dp->bridge.dev);
+
 	status = zynqmp_dp_bridge_detect(&dp->bridge);
 	drm_bridge_hpd_notify(&dp->bridge, status);
 }