Skip to contentSkip to conversation

drm/dp_mst: Don't report ports connected if nothing is attached to them

PATCH In review Imre Deak 3 participants 10 messages Feb 1, 2021 – Feb 4, 2021

Subsystems (3)
lore
Share & tools
lore

Imre Deak

author
#0 lore raw

PATCH 1/4 drm/dp_mst: Don't report ports connected if nothing is attached to them

Reporting a port as connected if nothing is attached to them leads to any i2c transactions on this port trying to use an uninitialized i2c adapter, fix this.

Let's account for this case even if branch devices have no good reason to report a port as unplugged with their peer device type set to 'none'.

Fixes db1a07956968 ("drm/dp_mst: Handle SST-only branch device case")

References: https://gitlab.freedesktop.org/drm/intel/-/issues/2987
References: https://gitlab.freedesktop.org/drm/intel/-/issues/1963

---
 drivers/gpu/drm/drm_dp_mst_topology.c | 1 +
 1 file changed, 1 insertion(+)
drivers/gpu/drm/drm_dp_mst_topology.c +1 -0
index e82b596d646c..deb7995f42fa 100644
@@ -4224,6 +4224,7 @@ drm_dp_mst_detect_port(struct drm_connector *connector,
 
 	switch (port->pdt) {
 	case DP_PEER_DEVICE_NONE:
+		break;
 	case DP_PEER_DEVICE_MST_BRANCHING:
 		if (!port->mcs)
 			ret = connector_status_connected;
Signature · 6 lines
2.25.1

_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel
author
#1 lore raw

PATCH 2/4 drm/dp_mst: Don't cache EDIDs for physical ports

Caching EDIDs for physical ports prevents updating the EDID if a port gets reconnected via a Connection Status Notification message, fix this.

Fixes db1a07956968 ("drm/dp_mst: Handle SST-only branch device case")
Cc Wayne Lin <[email protected]>
Cc Lyude Paul <[email protected]>
---
 drivers/gpu/drm/drm_dp_mst_topology.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)
drivers/gpu/drm/drm_dp_mst_topology.c +2 -1
index deb7995f42fa..309afe61afdd 100644
@@ -2302,7 +2302,8 @@ drm_dp_mst_port_add_connector(struct drm_dp_mst_branch *mstb,
 	}
 
 	if (port->pdt != DP_PEER_DEVICE_NONE &&
-	    drm_dp_mst_is_end_device(port->pdt, port->mcs)) {
+	    drm_dp_mst_is_end_device(port->pdt, port->mcs) &&
+	    port->port_num >= DP_MST_LOGICAL_PORT_0) {
 		port->cached_edid = drm_get_edid(port->connector,
 						 &port->aux.ddc);
 		drm_connector_set_tile_property(port->connector);
Signature · 6 lines
2.25.1

_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel
#2 lore raw

RE: [PATCH 2/4] drm/dp_mst: Don't cache EDIDs for physical ports

[AMD Public Use]

Quoted 23 lines

-----Original Message-----
From: Imre Deak <[email protected]>
Sent: Monday, February 1, 2021 8:02 PM
To: [email protected]
Cc: Lin, Wayne <[email protected]>; Lyude Paul <[email protected]>
Subject: [PATCH 2/4] drm/dp_mst: Don't cache EDIDs for physical ports

Caching EDIDs for physical ports prevents updating the EDID if a port gets reconnected via a Connection Status Notification message, fix this.

Fixes db1a07956968 ("drm/dp_mst: Handle SST-only branch device case")
Cc Wayne Lin <[email protected]>
Cc Lyude Paul <[email protected]>
---
 drivers/gpu/drm/drm_dp_mst_topology.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)
drivers/gpu/drm/drm_dp_mst_topology.c +0 -0
index deb7995f42fa..309afe61afdd 100644

drivers/gpu/drm/drm_dp_mst_topology.c drm_dp_mst_port_add_connector(struct drm_dp_mst_branch *mstb, · line 2306

Quoted 6 lines
drivers/gpu/drm/drm_dp_mst_topology.c +2 -1
@@ -2302,7 +2302,8 @@ drm_dp_mst_port_add_connector(struct drm_dp_mst_branch *mstb,
 }
 
 if (port->pdt != DP_PEER_DEVICE_NONE &&
-    drm_dp_mst_is_end_device(port->pdt, port->mcs)) {
+    drm_dp_mst_is_end_device(port->pdt, port->mcs) &&
+    port->port_num >= DP_MST_LOGICAL_PORT_0) {

Hi Imre Deak,

Thanks for the patch!
Just curious that you mean we don't want to fetch the EDID of the sst monitor like below case?

    Src->MST device ->SST monitor

I thought we still need to get the EDID even the monitor is connected to the physical output port of mst device. Maybe what we should fix here is why the EDID is not get updated once reconnected via CSN message?

Thanks!

Quoted 5 lines
drivers/gpu/drm/drm_dp_mst_topology.c +0 -1
 port->cached_edid = drm_get_edid(port->connector,
  &port->aux.ddc);
 drm_connector_set_tile_property(port->connector);
--

2.25.1

Regards,
Wayne Lin
_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel

author
#3 lore raw

Re: [PATCH 2/4] drm/dp_mst: Don't cache EDIDs for physical ports

Quoted 33 lines · Wayne

[AMD Public Use]

Quoted 23 lines

-----Original Message-----
From: Imre Deak <[email protected]>
Sent: Monday, February 1, 2021 8:02 PM
To: [email protected]
Cc: Lin, Wayne <[email protected]>; Lyude Paul <[email protected]>
Subject: [PATCH 2/4] drm/dp_mst: Don't cache EDIDs for physical ports

Caching EDIDs for physical ports prevents updating the EDID if a port gets reconnected via a Connection Status Notification message, fix this.

Fixes db1a07956968 ("drm/dp_mst: Handle SST-only branch device case")
Cc Wayne Lin <[email protected]>
Cc Lyude Paul <[email protected]>
---
 drivers/gpu/drm/drm_dp_mst_topology.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)
drivers/gpu/drm/drm_dp_mst_topology.c +0 -0
index deb7995f42fa..309afe61afdd 100644

drivers/gpu/drm/drm_dp_mst_topology.c drm_dp_mst_port_add_connector(struct drm_dp_mst_branch *mstb, · line 2306

Quoted 6 lines
drivers/gpu/drm/drm_dp_mst_topology.c +2 -1
@@ -2302,7 +2302,8 @@ drm_dp_mst_port_add_connector(struct drm_dp_mst_branch *mstb,
 }
 
 if (port->pdt != DP_PEER_DEVICE_NONE &&
-    drm_dp_mst_is_end_device(port->pdt, port->mcs)) {
+    drm_dp_mst_is_end_device(port->pdt, port->mcs) &&
+    port->port_num >= DP_MST_LOGICAL_PORT_0) {

Hi Imre Deak,

Quoted 5 lines

Thanks for the patch!
Just curious that you mean we don't want to fetch the EDID of the sst monitor like below case?

    Src->MST device ->SST monitor

The intention of the mst cached_edid logic is to cache the EDID for logical ports where the EDID cannot change anyway. The EDID on physical ports is fetched during connector probing just as for any other connector.

Quoted 2 lines

I thought we still need to get the EDID even the monitor is connected to the physical output port of mst device.

For sinks attached to phyisical ports we get the EDID whenever probing the corresponding connector.

Quoted 2 lines

Maybe what we should fix here is why the EDID is not get updated once reconnected via CSN message?

This patch fixes the problem that we stopped updating the EDID for physical connectors. After this change it will get updated when probing such connectors.

Quoted 8 lines

Thanks!

Quoted 5 lines
drivers/gpu/drm/drm_dp_mst_topology.c +0 -1
 port->cached_edid = drm_get_edid(port->connector,
  &port->aux.ddc);
 drm_connector_set_tile_property(port->connector);
--

2.25.1

Regards,
Wayne Lin

_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel

#4 lore raw

RE: [PATCH 2/4] drm/dp_mst: Don't cache EDIDs for physical ports

[AMD Official Use Only - Internal Distribution Only]

Quoted 56 lines

-----Original Message-----
From: Imre Deak <[email protected]>
Sent: Tuesday, February 2, 2021 7:22 PM
To: Lin, Wayne <[email protected]>
Cc: [email protected]; Lyude Paul <[email protected]>
Subject: Re: [PATCH 2/4] drm/dp_mst: Don't cache EDIDs for physical ports

Quoted 36 lines · Wayne

[AMD Public Use]

Quoted 26 lines

-----Original Message-----
From: Imre Deak <[email protected]>
Sent: Monday, February 1, 2021 8:02 PM
To: [email protected]
Cc: Lin, Wayne <[email protected]>; Lyude Paul <[email protected]>
Subject: [PATCH 2/4] drm/dp_mst: Don't cache EDIDs for physical ports

Caching EDIDs for physical ports prevents updating the EDID if a port gets reconnected via a Connection Status Notification message, fix this.

Fixes db1a07956968 ("drm/dp_mst: Handle SST-only branch device

case")

Cc Wayne Lin <[email protected]>
Cc Lyude Paul <[email protected]>
---
 drivers/gpu/drm/drm_dp_mst_topology.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)
drivers/gpu/drm/drm_dp_mst_topology.c +0 -0
index deb7995f42fa..309afe61afdd 100644

drivers/gpu/drm/drm_dp_mst_topology.c drm_dp_mst_port_add_connector(struct · line 2306

Quoted 6 lines
drivers/gpu/drm/drm_dp_mst_topology.c +2 -1
@@ -2302,7 +2302,8 @@ drm_dp_mst_port_add_connector(struct
 drm_dp_mst_branch *mstb,  }
 
 if (port->pdt != DP_PEER_DEVICE_NONE &&
-    drm_dp_mst_is_end_device(port->pdt, port->mcs)) {
+    drm_dp_mst_is_end_device(port->pdt, port->mcs) &&
+    port->port_num >= DP_MST_LOGICAL_PORT_0) {

Hi Imre Deak,

Quoted 5 lines

Thanks for the patch!
Just curious that you mean we don't want to fetch the EDID of the sst monitor like below case?

    Src->MST device ->SST monitor

The intention of the mst cached_edid logic is to cache the EDID for logical ports where the EDID cannot change anyway. The EDID on physical ports is fetched during connector probing just as for any other connector.

Quoted 2 lines

I thought we still need to get the EDID even the monitor is connected to the physical output port of mst device.

Quoted 8 lines

For sinks attached to phyisical ports we get the EDID whenever probing the corresponding connector.

Quoted 2 lines

Maybe what we should fix here is why the EDID is not get updated once reconnected via CSN message?

This patch fixes the problem that we stopped updating the EDID for physical connectors. After this change it will get updated when probing such connectors.

Appreciate for the explanation.
Thanks!

Quoted 8 lines
Quoted 8 lines

Thanks!

Quoted 5 lines
drivers/gpu/drm/drm_dp_mst_topology.c +0 -1
 port->cached_edid = drm_get_edid(port->connector,
  &port->aux.ddc);
 drm_connector_set_tile_property(port->connector);
--

2.25.1

Regards,
Wayne Lin

Regards,
Wayne Lin
_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel

author
#5 lore raw

Re: [PATCH 2/4] drm/dp_mst: Don't cache EDIDs for physical ports

In reply to Imre Deak
Quoted 7 lines · Imre Deak

Caching EDIDs for physical ports prevents updating the EDID if a port gets reconnected via a Connection Status Notification message, fix this.

Fixes db1a07956968 ("drm/dp_mst: Handle SST-only branch device case")
Cc Wayne Lin <[email protected]>
Cc Lyude Paul <[email protected]>

Pushed patches 2-4 to drm-misc-next, thanks for the review.

Quoted 25 lines
---
 drivers/gpu/drm/drm_dp_mst_topology.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)
drivers/gpu/drm/drm_dp_mst_topology.c +2 -1
index deb7995f42fa..309afe61afdd 100644
@@ -2302,7 +2302,8 @@ drm_dp_mst_port_add_connector(struct drm_dp_mst_branch *mstb,
 	}
 
 	if (port->pdt != DP_PEER_DEVICE_NONE &&
-	    drm_dp_mst_is_end_device(port->pdt, port->mcs)) {
+	    drm_dp_mst_is_end_device(port->pdt, port->mcs) &&
+	    port->port_num >= DP_MST_LOGICAL_PORT_0) {
 		port->cached_edid = drm_get_edid(port->connector,
 						 &port->aux.ddc);
 		drm_connector_set_tile_property(port->connector);
Signature · 6 lines
2.25.1

_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel

_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel

author
#6 lore raw

PATCH 3/4 drm/dp_mst: Remove redundant tile property setting

In reply to Imre Deak

drm_get_edid() already updates the tile property since
commit 2de3a078497b ("drm/dp: Set the connector's TILE property even for DP SST connectors") so no need to update it after calling this function.

---
 drivers/gpu/drm/drm_dp_mst_topology.c | 4 +---
 1 file changed, 1 insertion(+), 3 deletions(-)
drivers/gpu/drm/drm_dp_mst_topology.c +1 -3
index 309afe61afdd..43a40660136c 100644
@@ -2303,11 +2303,9 @@ drm_dp_mst_port_add_connector(struct drm_dp_mst_branch *mstb,
 
 	if (port->pdt != DP_PEER_DEVICE_NONE &&
 	    drm_dp_mst_is_end_device(port->pdt, port->mcs) &&
-	    port->port_num >= DP_MST_LOGICAL_PORT_0) {
+	    port->port_num >= DP_MST_LOGICAL_PORT_0)
 		port->cached_edid = drm_get_edid(port->connector,
 						 &port->aux.ddc);
-		drm_connector_set_tile_property(port->connector);
-	}
 
 	drm_connector_register(port->connector);
 	return;
Signature · 6 lines
2.25.1

_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel
author
#7 lore raw

PATCH 4/4 drm/dp_mst: Use DP_MST_LOGICAL_PORT_0 instead of magic number

In reply to Imre Deak

Use the macro defined for the first logical port instead of the corresponding magic number.

---
 drivers/gpu/drm/drm_dp_mst_topology.c | 3 +--
 1 file changed, 1 insertion(+), 2 deletions(-)
drivers/gpu/drm/drm_dp_mst_topology.c +1 -2
index 43a40660136c..0a54506c2773 100644
@@ -4232,9 +4232,8 @@ drm_dp_mst_detect_port(struct drm_connector *connector,
 	case DP_PEER_DEVICE_SST_SINK:
 		ret = connector_status_connected;
 		/* for logical ports - cache the EDID */
-		if (port->port_num >= 8 && !port->cached_edid) {
+		if (port->port_num >= DP_MST_LOGICAL_PORT_0 && !port->cached_edid)
 			port->cached_edid = drm_get_edid(connector, &port->aux.ddc);
-		}
 		break;
 	case DP_PEER_DEVICE_DP_LEGACY_CONV:
 		if (port->ldps)
Signature · 6 lines
2.25.1

_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel
Reviewed-by on v1 1/4
#8 lore raw
In reply to Imre Deak

For the whole series:

Quoted 32 lines · Imre Deak

Reporting a port as connected if nothing is attached to them leads to any i2c transactions on this port trying to use an uninitialized i2c adapter, fix this.

Let's account for this case even if branch devices have no good reason to report a port as unplugged with their peer device type set to 'none'.

Fixes db1a07956968 ("drm/dp_mst: Handle SST-only branch device case")

References: https://gitlab.freedesktop.org/drm/intel/-/issues/2987
References: https://gitlab.freedesktop.org/drm/intel/-/issues/1963

---
 drivers/gpu/drm/drm_dp_mst_topology.c | 1 +

 1 file changed, 1 insertion(+)
drivers/gpu/drm/drm_dp_mst_topology.c +0 -0
index e82b596d646c..deb7995f42fa 100644
@@ -4224,6 +4224,7 @@ drm_dp_mst_detect_port(struct drm_connector *connector,


        switch (port->pdt) {
        case DP_PEER_DEVICE_NONE:
+               break;
        case DP_PEER_DEVICE_MST_BRANCHING:
                if (!port->mcs)
                        ret = connector_status_connected;

Signature · 13 lines
Sincerely,
   Lyude Paul (she/her)
   Software Engineer at Red Hat
   
Note: I deal with a lot of emails and have a lot of bugs on my plate. If you've
asked me a question, are waiting for a review/merge on a patch, etc. and I
haven't responded in a while, please feel free to send me another email to check
on my status. I don't bite!

_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel
author
#9 lore raw
In reply to Imre Deak
Quoted 7 lines · Imre Deak

Reporting a port as connected if nothing is attached to them leads to any i2c transactions on this port trying to use an uninitialized i2c adapter, fix this.

Let's account for this case even if branch devices have no good reason to report a port as unplugged with their peer device type set to 'none'.

Quoted 8 lines
Fixes db1a07956968 ("drm/dp_mst: Handle SST-only branch device case")

References: https://gitlab.freedesktop.org/drm/intel/-/issues/2987
References: https://gitlab.freedesktop.org/drm/intel/-/issues/1963

Thanks for the report and review, I pushed this one patch to drm-misc-fixes.

I fixed a typo in the commit message.

Quoted 23 lines
---
 drivers/gpu/drm/drm_dp_mst_topology.c | 1 +
 1 file changed, 1 insertion(+)
drivers/gpu/drm/drm_dp_mst_topology.c +1 -0
index e82b596d646c..deb7995f42fa 100644
@@ -4224,6 +4224,7 @@ drm_dp_mst_detect_port(struct drm_connector *connector,
 
 	switch (port->pdt) {
 	case DP_PEER_DEVICE_NONE:
+		break;
 	case DP_PEER_DEVICE_MST_BRANCHING:
 		if (!port->mcs)
 			ret = connector_status_connected;
Signature · 6 lines
2.25.1

_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx