[PATCH 01/26] drm/dp_mst: Move link address dumping into a function

[PATCH 01/26] drm/dp_mst: Move link address dumping into a function

[Lyude Paul]

Since we're about to be calling this from multiple places. Also it makes
things easier to read!

Cc: Juston Li <juston.li@intel.com>
Cc: Imre Deak <imre.deak@intel.com>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Cc: Harry Wentland <hwentlan@amd.com>
Signed-off-by: Lyude Paul <lyude@redhat.com>
---
 drivers/gpu/drm/drm_dp_mst_topology.c | 35 ++++++++++++++++++---------
 1 file changed, 23 insertions(+), 12 deletions(-)

diff --git a/drivers/gpu/drm/drm_dp_mst_topology.c b/drivers/gpu/drm/drm_dp_mst_topology.c
index 0984b9a34d55..998081b9b205 100644
--- a/drivers/gpu/drm/drm_dp_mst_topology.c
+++ b/drivers/gpu/drm/drm_dp_mst_topology.c
@@ -2013,6 +2013,28 @@ static void drm_dp_queue_down_tx(struct drm_dp_mst_topology_mgr *mgr,
 	mutex_unlock(&mgr->qlock);
 }
 
+static void
+drm_dp_dump_link_address(struct drm_dp_link_address_ack_reply *reply)
+{
+	struct drm_dp_link_addr_reply_port *port_reply;
+	int i;
+
+	for (i = 0; i < reply->nports; i++) {
+		port_reply = &reply->ports[i];
+		DRM_DEBUG_KMS("port %d: input %d, pdt: %d, pn: %d, dpcd_rev: %02x, mcs: %d, ddps: %d, ldps %d, sdp %d/%d\n",
+			      i,
+			      port_reply->input_port,
+			      port_reply->peer_device_type,
+			      port_reply->port_number,
+			      port_reply->dpcd_revision,
+			      port_reply->mcs,
+			      port_reply->ddps,
+			      port_reply->legacy_device_plug_status,
+			      port_reply->num_sdp_streams,
+			      port_reply->num_sdp_stream_sinks);
+	}
+}
+
 static void drm_dp_send_link_address(struct drm_dp_mst_topology_mgr *mgr,
 				     struct drm_dp_mst_branch *mstb)
 {
@@ -2038,18 +2060,7 @@ static void drm_dp_send_link_address(struct drm_dp_mst_topology_mgr *mgr,
 			DRM_DEBUG_KMS("link address nak received\n");
 		} else {
 			DRM_DEBUG_KMS("link address reply: %d\n", txmsg->reply.u.link_addr.nports);
-			for (i = 0; i < txmsg->reply.u.link_addr.nports; i++) {
-				DRM_DEBUG_KMS("port %d: input %d, pdt: %d, pn: %d, dpcd_rev: %02x, mcs: %d, ddps: %d, ldps %d, sdp %d/%d\n", i,
-				       txmsg->reply.u.link_addr.ports[i].input_port,
-				       txmsg->reply.u.link_addr.ports[i].peer_device_type,
-				       txmsg->reply.u.link_addr.ports[i].port_number,
-				       txmsg->reply.u.link_addr.ports[i].dpcd_revision,
-				       txmsg->reply.u.link_addr.ports[i].mcs,
-				       txmsg->reply.u.link_addr.ports[i].ddps,
-				       txmsg->reply.u.link_addr.ports[i].legacy_device_plug_status,
-				       txmsg->reply.u.link_addr.ports[i].num_sdp_streams,
-				       txmsg->reply.u.link_addr.ports[i].num_sdp_stream_sinks);
-			}
+			drm_dp_dump_link_address(&txmsg->reply.u.link_addr);
 
 			drm_dp_check_mstb_guid(mstb, txmsg->reply.u.link_addr.guid);
 
-- 
2.21.0

Re: [PATCH 01/26] drm/dp_mst: Move link address dumping into a function

[Daniel Vetter]

On Wed, Jul 17, 2019 at 09:42:24PM -0400, Lyude Paul wrote:
> Since we're about to be calling this from multiple places. Also it makes
> things easier to read!
> 
> Cc: Juston Li <juston.li@intel.com>
> Cc: Imre Deak <imre.deak@intel.com>
> Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
> Cc: Harry Wentland <hwentlan@amd.com>
> Signed-off-by: Lyude Paul <lyude@redhat.com>

Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>

> ---
>  drivers/gpu/drm/drm_dp_mst_topology.c | 35 ++++++++++++++++++---------
>  1 file changed, 23 insertions(+), 12 deletions(-)
> 
> diff --git a/drivers/gpu/drm/drm_dp_mst_topology.c b/drivers/gpu/drm/drm_dp_mst_topology.c
> index 0984b9a34d55..998081b9b205 100644
> --- a/drivers/gpu/drm/drm_dp_mst_topology.c
> +++ b/drivers/gpu/drm/drm_dp_mst_topology.c
> @@ -2013,6 +2013,28 @@ static void drm_dp_queue_down_tx(struct drm_dp_mst_topology_mgr *mgr,
>  	mutex_unlock(&mgr->qlock);
>  }
>  
> +static void
> +drm_dp_dump_link_address(struct drm_dp_link_address_ack_reply *reply)
> +{
> +	struct drm_dp_link_addr_reply_port *port_reply;
> +	int i;
> +
> +	for (i = 0; i < reply->nports; i++) {
> +		port_reply = &reply->ports[i];
> +		DRM_DEBUG_KMS("port %d: input %d, pdt: %d, pn: %d, dpcd_rev: %02x, mcs: %d, ddps: %d, ldps %d, sdp %d/%d\n",
> +			      i,
> +			      port_reply->input_port,
> +			      port_reply->peer_device_type,
> +			      port_reply->port_number,
> +			      port_reply->dpcd_revision,
> +			      port_reply->mcs,
> +			      port_reply->ddps,
> +			      port_reply->legacy_device_plug_status,
> +			      port_reply->num_sdp_streams,
> +			      port_reply->num_sdp_stream_sinks);
> +	}
> +}
> +
>  static void drm_dp_send_link_address(struct drm_dp_mst_topology_mgr *mgr,
>  				     struct drm_dp_mst_branch *mstb)
>  {
> @@ -2038,18 +2060,7 @@ static void drm_dp_send_link_address(struct drm_dp_mst_topology_mgr *mgr,
>  			DRM_DEBUG_KMS("link address nak received\n");
>  		} else {
>  			DRM_DEBUG_KMS("link address reply: %d\n", txmsg->reply.u.link_addr.nports);
> -			for (i = 0; i < txmsg->reply.u.link_addr.nports; i++) {
> -				DRM_DEBUG_KMS("port %d: input %d, pdt: %d, pn: %d, dpcd_rev: %02x, mcs: %d, ddps: %d, ldps %d, sdp %d/%d\n", i,
> -				       txmsg->reply.u.link_addr.ports[i].input_port,
> -				       txmsg->reply.u.link_addr.ports[i].peer_device_type,
> -				       txmsg->reply.u.link_addr.ports[i].port_number,
> -				       txmsg->reply.u.link_addr.ports[i].dpcd_revision,
> -				       txmsg->reply.u.link_addr.ports[i].mcs,
> -				       txmsg->reply.u.link_addr.ports[i].ddps,
> -				       txmsg->reply.u.link_addr.ports[i].legacy_device_plug_status,
> -				       txmsg->reply.u.link_addr.ports[i].num_sdp_streams,
> -				       txmsg->reply.u.link_addr.ports[i].num_sdp_stream_sinks);
> -			}
> +			drm_dp_dump_link_address(&txmsg->reply.u.link_addr);
>  
>  			drm_dp_check_mstb_guid(mstb, txmsg->reply.u.link_addr.guid);
>  
> -- 
> 2.21.0
> 

-- 
Daniel Vetter
Software Engineer, Intel Corporation
http://blog.ffwll.ch

Re: [PATCH 01/26] drm/dp_mst: Move link address dumping into a function

[Lyude Paul]

*sigh* finally have some time to go through these reviews

jfyi: I realized after looking over this patch that it's not actually needed -
I had been planning on using drm_dp_dump_link_address() for other things, but
ended up deciding to make the final plan to use something that dumps into a
format that's identical to the one we're using for dumping DOWN requests. IMHO
though, this patch does make things look nicer so I'll probably keep it.

Assuming I can still count your r-b as valid with a change to the commit
description?

On Thu, 2019-08-08 at 21:53 +0200, Daniel Vetter wrote:
> On Wed, Jul 17, 2019 at 09:42:24PM -0400, Lyude Paul wrote:
> > Since we're about to be calling this from multiple places. Also it makes
> > things easier to read!
> > 
> > Cc: Juston Li <juston.li@intel.com>
> > Cc: Imre Deak <imre.deak@intel.com>
> > Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
> > Cc: Harry Wentland <hwentlan@amd.com>
> > Signed-off-by: Lyude Paul <lyude@redhat.com>
> 
> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
> 
> > ---
> >  drivers/gpu/drm/drm_dp_mst_topology.c | 35 ++++++++++++++++++---------
> >  1 file changed, 23 insertions(+), 12 deletions(-)
> > 
> > diff --git a/drivers/gpu/drm/drm_dp_mst_topology.c
> > b/drivers/gpu/drm/drm_dp_mst_topology.c
> > index 0984b9a34d55..998081b9b205 100644
> > --- a/drivers/gpu/drm/drm_dp_mst_topology.c
> > +++ b/drivers/gpu/drm/drm_dp_mst_topology.c
> > @@ -2013,6 +2013,28 @@ static void drm_dp_queue_down_tx(struct
> > drm_dp_mst_topology_mgr *mgr,
> >  	mutex_unlock(&mgr->qlock);
> >  }
> >  
> > +static void
> > +drm_dp_dump_link_address(struct drm_dp_link_address_ack_reply *reply)
> > +{
> > +	struct drm_dp_link_addr_reply_port *port_reply;
> > +	int i;
> > +
> > +	for (i = 0; i < reply->nports; i++) {
> > +		port_reply = &reply->ports[i];
> > +		DRM_DEBUG_KMS("port %d: input %d, pdt: %d, pn: %d, dpcd_rev:
> > %02x, mcs: %d, ddps: %d, ldps %d, sdp %d/%d\n",
> > +			      i,
> > +			      port_reply->input_port,
> > +			      port_reply->peer_device_type,
> > +			      port_reply->port_number,
> > +			      port_reply->dpcd_revision,
> > +			      port_reply->mcs,
> > +			      port_reply->ddps,
> > +			      port_reply->legacy_device_plug_status,
> > +			      port_reply->num_sdp_streams,
> > +			      port_reply->num_sdp_stream_sinks);
> > +	}
> > +}
> > +
> >  static void drm_dp_send_link_address(struct drm_dp_mst_topology_mgr *mgr,
> >  				     struct drm_dp_mst_branch *mstb)
> >  {
> > @@ -2038,18 +2060,7 @@ static void drm_dp_send_link_address(struct
> > drm_dp_mst_topology_mgr *mgr,
> >  			DRM_DEBUG_KMS("link address nak received\n");
> >  		} else {
> >  			DRM_DEBUG_KMS("link address reply: %d\n", txmsg-
> > >reply.u.link_addr.nports);
> > -			for (i = 0; i < txmsg->reply.u.link_addr.nports; i++)
> > {
> > -				DRM_DEBUG_KMS("port %d: input %d, pdt: %d, pn:
> > %d, dpcd_rev: %02x, mcs: %d, ddps: %d, ldps %d, sdp %d/%d\n", i,
> > -				       txmsg-
> > >reply.u.link_addr.ports[i].input_port,
> > -				       txmsg-
> > >reply.u.link_addr.ports[i].peer_device_type,
> > -				       txmsg-
> > >reply.u.link_addr.ports[i].port_number,
> > -				       txmsg-
> > >reply.u.link_addr.ports[i].dpcd_revision,
> > -				       txmsg->reply.u.link_addr.ports[i].mcs,
> > -				       txmsg->reply.u.link_addr.ports[i].ddps,
> > -				       txmsg-
> > >reply.u.link_addr.ports[i].legacy_device_plug_status,
> > -				       txmsg-
> > >reply.u.link_addr.ports[i].num_sdp_streams,
> > -				       txmsg-
> > >reply.u.link_addr.ports[i].num_sdp_stream_sinks);
> > -			}
> > +			drm_dp_dump_link_address(&txmsg->reply.u.link_addr);
> >  
> >  			drm_dp_check_mstb_guid(mstb, txmsg-
> > >reply.u.link_addr.guid);
> >  
> > -- 
> > 2.21.0
> > 
-- 
Cheers,
	Lyude Paul

Re: [PATCH 01/26] drm/dp_mst: Move link address dumping into a function

[Daniel Vetter]

On Mon, Aug 26, 2019 at 05:51:26PM -0400, Lyude Paul wrote:
> *sigh* finally have some time to go through these reviews

Hey it took me longer to start even reviewing this, and not even through
:-( than it took you to reply here. So no worries!

> jfyi: I realized after looking over this patch that it's not actually needed -
> I had been planning on using drm_dp_dump_link_address() for other things, but
> ended up deciding to make the final plan to use something that dumps into a
> format that's identical to the one we're using for dumping DOWN requests. IMHO
> though, this patch does make things look nicer so I'll probably keep it.
> 
> Assuming I can still count your r-b as valid with a change to the commit
> description?

Sure.

Cheers, Daniel

> 
> On Thu, 2019-08-08 at 21:53 +0200, Daniel Vetter wrote:
> > On Wed, Jul 17, 2019 at 09:42:24PM -0400, Lyude Paul wrote:
> > > Since we're about to be calling this from multiple places. Also it makes
> > > things easier to read!
> > > 
> > > Cc: Juston Li <juston.li@intel.com>
> > > Cc: Imre Deak <imre.deak@intel.com>
> > > Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
> > > Cc: Harry Wentland <hwentlan@amd.com>
> > > Signed-off-by: Lyude Paul <lyude@redhat.com>
> > 
> > Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
> > 
> > > ---
> > >  drivers/gpu/drm/drm_dp_mst_topology.c | 35 ++++++++++++++++++---------
> > >  1 file changed, 23 insertions(+), 12 deletions(-)
> > > 
> > > diff --git a/drivers/gpu/drm/drm_dp_mst_topology.c
> > > b/drivers/gpu/drm/drm_dp_mst_topology.c
> > > index 0984b9a34d55..998081b9b205 100644
> > > --- a/drivers/gpu/drm/drm_dp_mst_topology.c
> > > +++ b/drivers/gpu/drm/drm_dp_mst_topology.c
> > > @@ -2013,6 +2013,28 @@ static void drm_dp_queue_down_tx(struct
> > > drm_dp_mst_topology_mgr *mgr,
> > >  	mutex_unlock(&mgr->qlock);
> > >  }
> > >  
> > > +static void
> > > +drm_dp_dump_link_address(struct drm_dp_link_address_ack_reply *reply)
> > > +{
> > > +	struct drm_dp_link_addr_reply_port *port_reply;
> > > +	int i;
> > > +
> > > +	for (i = 0; i < reply->nports; i++) {
> > > +		port_reply = &reply->ports[i];
> > > +		DRM_DEBUG_KMS("port %d: input %d, pdt: %d, pn: %d, dpcd_rev:
> > > %02x, mcs: %d, ddps: %d, ldps %d, sdp %d/%d\n",
> > > +			      i,
> > > +			      port_reply->input_port,
> > > +			      port_reply->peer_device_type,
> > > +			      port_reply->port_number,
> > > +			      port_reply->dpcd_revision,
> > > +			      port_reply->mcs,
> > > +			      port_reply->ddps,
> > > +			      port_reply->legacy_device_plug_status,
> > > +			      port_reply->num_sdp_streams,
> > > +			      port_reply->num_sdp_stream_sinks);
> > > +	}
> > > +}
> > > +
> > >  static void drm_dp_send_link_address(struct drm_dp_mst_topology_mgr *mgr,
> > >  				     struct drm_dp_mst_branch *mstb)
> > >  {
> > > @@ -2038,18 +2060,7 @@ static void drm_dp_send_link_address(struct
> > > drm_dp_mst_topology_mgr *mgr,
> > >  			DRM_DEBUG_KMS("link address nak received\n");
> > >  		} else {
> > >  			DRM_DEBUG_KMS("link address reply: %d\n", txmsg-
> > > >reply.u.link_addr.nports);
> > > -			for (i = 0; i < txmsg->reply.u.link_addr.nports; i++)
> > > {
> > > -				DRM_DEBUG_KMS("port %d: input %d, pdt: %d, pn:
> > > %d, dpcd_rev: %02x, mcs: %d, ddps: %d, ldps %d, sdp %d/%d\n", i,
> > > -				       txmsg-
> > > >reply.u.link_addr.ports[i].input_port,
> > > -				       txmsg-
> > > >reply.u.link_addr.ports[i].peer_device_type,
> > > -				       txmsg-
> > > >reply.u.link_addr.ports[i].port_number,
> > > -				       txmsg-
> > > >reply.u.link_addr.ports[i].dpcd_revision,
> > > -				       txmsg->reply.u.link_addr.ports[i].mcs,
> > > -				       txmsg->reply.u.link_addr.ports[i].ddps,
> > > -				       txmsg-
> > > >reply.u.link_addr.ports[i].legacy_device_plug_status,
> > > -				       txmsg-
> > > >reply.u.link_addr.ports[i].num_sdp_streams,
> > > -				       txmsg-
> > > >reply.u.link_addr.ports[i].num_sdp_stream_sinks);
> > > -			}
> > > +			drm_dp_dump_link_address(&txmsg->reply.u.link_addr);
> > >  
> > >  			drm_dp_check_mstb_guid(mstb, txmsg-
> > > >reply.u.link_addr.guid);
> > >  
> > > -- 
> > > 2.21.0
> > > 
> -- 
> Cheers,
> 	Lyude Paul
> 

-- 
Daniel Vetter
Software Engineer, Intel Corporation
http://blog.ffwll.ch

[PATCH 02/26] drm/dp_mst: Destroy mstbs from destroy_connector_work

[Lyude Paul]

Currently we remove MST branch devices from the in-memory topology
immediately when their topology refcount reaches 0. This works just fine
at the moment, but since we're about to add suspend/resume reprobing for
MST topologies we're going to need to be able to travel through the
topology and drop topology refs on branch devices while holding
mgr->mutex. Since we currently can't do this due to the circular locking
dependencies that would incur, move all of the actual work for
drm_dp_destroy_mst_branch_device() into drm_dp_destroy_connector_work()
so we can drop topology refs on MSTBs in any locking context.

Cc: Juston Li <juston.li@intel.com>
Cc: Imre Deak <imre.deak@intel.com>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Cc: Harry Wentland <hwentlan@amd.com>
Signed-off-by: Lyude Paul <lyude@redhat.com>
---
 drivers/gpu/drm/drm_dp_mst_topology.c | 121 +++++++++++++++++---------
 include/drm/drm_dp_mst_helper.h       |  10 +++
 2 files changed, 90 insertions(+), 41 deletions(-)

diff --git a/drivers/gpu/drm/drm_dp_mst_topology.c b/drivers/gpu/drm/drm_dp_mst_topology.c
index 998081b9b205..d7c3d9233834 100644
--- a/drivers/gpu/drm/drm_dp_mst_topology.c
+++ b/drivers/gpu/drm/drm_dp_mst_topology.c
@@ -1108,34 +1108,17 @@ static void drm_dp_destroy_mst_branch_device(struct kref *kref)
 	struct drm_dp_mst_branch *mstb =
 		container_of(kref, struct drm_dp_mst_branch, topology_kref);
 	struct drm_dp_mst_topology_mgr *mgr = mstb->mgr;
-	struct drm_dp_mst_port *port, *tmp;
-	bool wake_tx = false;
-
-	mutex_lock(&mgr->lock);
-	list_for_each_entry_safe(port, tmp, &mstb->ports, next) {
-		list_del(&port->next);
-		drm_dp_mst_topology_put_port(port);
-	}
-	mutex_unlock(&mgr->lock);
-
-	/* drop any tx slots msg */
-	mutex_lock(&mstb->mgr->qlock);
-	if (mstb->tx_slots[0]) {
-		mstb->tx_slots[0]->state = DRM_DP_SIDEBAND_TX_TIMEOUT;
-		mstb->tx_slots[0] = NULL;
-		wake_tx = true;
-	}
-	if (mstb->tx_slots[1]) {
-		mstb->tx_slots[1]->state = DRM_DP_SIDEBAND_TX_TIMEOUT;
-		mstb->tx_slots[1] = NULL;
-		wake_tx = true;
-	}
-	mutex_unlock(&mstb->mgr->qlock);
 
-	if (wake_tx)
-		wake_up_all(&mstb->mgr->tx_waitq);
+	INIT_LIST_HEAD(&mstb->destroy_next);
 
-	drm_dp_mst_put_mstb_malloc(mstb);
+	/*
+	 * This can get called under mgr->mutex, so we need to perform the
+	 * actual destruction of the mstb in another worker
+	 */
+	mutex_lock(&mgr->destroy_connector_lock);
+	list_add(&mstb->destroy_next, &mgr->destroy_branch_device_list);
+	mutex_unlock(&mgr->destroy_connector_lock);
+	schedule_work(&mgr->destroy_connector_work);
 }
 
 /**
@@ -3618,10 +3601,56 @@ static void drm_dp_tx_work(struct work_struct *work)
 	mutex_unlock(&mgr->qlock);
 }
 
+static inline void
+drm_dp_finish_destroy_port(struct drm_dp_mst_port *port)
+{
+	INIT_LIST_HEAD(&port->next);
+
+	port->mgr->cbs->destroy_connector(port->mgr, port->connector);
+
+	drm_dp_port_teardown_pdt(port, port->pdt);
+	port->pdt = DP_PEER_DEVICE_NONE;
+
+	drm_dp_mst_put_port_malloc(port);
+}
+
+static inline void
+drm_dp_finish_destroy_mst_branch_device(struct drm_dp_mst_branch *mstb)
+{
+	struct drm_dp_mst_topology_mgr *mgr = mstb->mgr;
+	struct drm_dp_mst_port *port, *tmp;
+	bool wake_tx = false;
+
+	mutex_lock(&mgr->lock);
+	list_for_each_entry_safe(port, tmp, &mstb->ports, next) {
+		list_del(&port->next);
+		drm_dp_mst_topology_put_port(port);
+	}
+	mutex_unlock(&mgr->lock);
+
+	/* drop any tx slots msg */
+	mutex_lock(&mstb->mgr->qlock);
+	if (mstb->tx_slots[0]) {
+		mstb->tx_slots[0]->state = DRM_DP_SIDEBAND_TX_TIMEOUT;
+		mstb->tx_slots[0] = NULL;
+		wake_tx = true;
+	}
+	if (mstb->tx_slots[1]) {
+		mstb->tx_slots[1]->state = DRM_DP_SIDEBAND_TX_TIMEOUT;
+		mstb->tx_slots[1] = NULL;
+		wake_tx = true;
+	}
+	mutex_unlock(&mstb->mgr->qlock);
+
+	if (wake_tx)
+		wake_up_all(&mstb->mgr->tx_waitq);
+
+	drm_dp_mst_put_mstb_malloc(mstb);
+}
+
 static void drm_dp_destroy_connector_work(struct work_struct *work)
 {
 	struct drm_dp_mst_topology_mgr *mgr = container_of(work, struct drm_dp_mst_topology_mgr, destroy_connector_work);
-	struct drm_dp_mst_port *port;
 	bool send_hotplug = false;
 	/*
 	 * Not a regular list traverse as we have to drop the destroy
@@ -3629,24 +3658,33 @@ static void drm_dp_destroy_connector_work(struct work_struct *work)
 	 * ordering between this lock and the config mutex.
 	 */
 	for (;;) {
+		struct drm_dp_mst_branch *mstb = NULL;
+		struct drm_dp_mst_port *port = NULL;
+
+		/* Destroy any MSTBs first, and then their ports second */
 		mutex_lock(&mgr->destroy_connector_lock);
-		port = list_first_entry_or_null(&mgr->destroy_connector_list, struct drm_dp_mst_port, next);
-		if (!port) {
-			mutex_unlock(&mgr->destroy_connector_lock);
-			break;
+		mstb = list_first_entry_or_null(&mgr->destroy_branch_device_list,
+						struct drm_dp_mst_branch,
+						destroy_next);
+		if (mstb) {
+			list_del(&mstb->destroy_next);
+		} else {
+			port = list_first_entry_or_null(&mgr->destroy_connector_list,
+							struct drm_dp_mst_port,
+							next);
+			if (port)
+				list_del(&port->next);
 		}
-		list_del(&port->next);
 		mutex_unlock(&mgr->destroy_connector_lock);
 
-		INIT_LIST_HEAD(&port->next);
-
-		mgr->cbs->destroy_connector(mgr, port->connector);
-
-		drm_dp_port_teardown_pdt(port, port->pdt);
-		port->pdt = DP_PEER_DEVICE_NONE;
-
-		drm_dp_mst_put_port_malloc(port);
-		send_hotplug = true;
+		if (mstb) {
+			drm_dp_finish_destroy_mst_branch_device(mstb);
+		} else if (port) {
+			drm_dp_finish_destroy_port(port);
+			send_hotplug = true;
+		} else {
+			break;
+		}
 	}
 	if (send_hotplug)
 		drm_kms_helper_hotplug_event(mgr->dev);
@@ -3840,6 +3878,7 @@ int drm_dp_mst_topology_mgr_init(struct drm_dp_mst_topology_mgr *mgr,
 	mutex_init(&mgr->destroy_connector_lock);
 	INIT_LIST_HEAD(&mgr->tx_msg_downq);
 	INIT_LIST_HEAD(&mgr->destroy_connector_list);
+	INIT_LIST_HEAD(&mgr->destroy_branch_device_list);
 	INIT_WORK(&mgr->work, drm_dp_mst_link_probe_work);
 	INIT_WORK(&mgr->tx_work, drm_dp_tx_work);
 	INIT_WORK(&mgr->destroy_connector_work, drm_dp_destroy_connector_work);
diff --git a/include/drm/drm_dp_mst_helper.h b/include/drm/drm_dp_mst_helper.h
index 8c97a5f92c47..c01f3ea72756 100644
--- a/include/drm/drm_dp_mst_helper.h
+++ b/include/drm/drm_dp_mst_helper.h
@@ -143,6 +143,12 @@ struct drm_dp_mst_branch {
 	 */
 	struct kref malloc_kref;
 
+	/**
+	 * @destroy_next: linked-list entry used by
+	 * drm_dp_destroy_connector_work()
+	 */
+	struct list_head destroy_next;
+
 	u8 rad[8];
 	u8 lct;
 	int num_ports;
@@ -578,6 +584,10 @@ struct drm_dp_mst_topology_mgr {
 	 * @destroy_connector_list: List of to be destroyed connectors.
 	 */
 	struct list_head destroy_connector_list;
+	/**
+	 * @destroy_branch_device_list: List of to be destroyed branch devices
+	 */
+	struct list_head destroy_branch_device_list;
 	/**
 	 * @destroy_connector_lock: Protects @connector_list.
 	 */
-- 
2.21.0

Re: [PATCH 02/26] drm/dp_mst: Destroy mstbs from destroy_connector_work

[Daniel Vetter]

On Wed, Jul 17, 2019 at 09:42:25PM -0400, Lyude Paul wrote:
> Currently we remove MST branch devices from the in-memory topology
> immediately when their topology refcount reaches 0. This works just fine
> at the moment, but since we're about to add suspend/resume reprobing for
> MST topologies we're going to need to be able to travel through the
> topology and drop topology refs on branch devices while holding
> mgr->mutex. Since we currently can't do this due to the circular locking
> dependencies that would incur, move all of the actual work for
> drm_dp_destroy_mst_branch_device() into drm_dp_destroy_connector_work()
> so we can drop topology refs on MSTBs in any locking context.

Would be good to point at where exactly the problem is here, maybe also
mentioned the exact future patch that causes the problem. I did go look
around a bit, but didn't spot it.

> 
> Cc: Juston Li <juston.li@intel.com>
> Cc: Imre Deak <imre.deak@intel.com>
> Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
> Cc: Harry Wentland <hwentlan@amd.com>
> Signed-off-by: Lyude Paul <lyude@redhat.com>
> ---
>  drivers/gpu/drm/drm_dp_mst_topology.c | 121 +++++++++++++++++---------
>  include/drm/drm_dp_mst_helper.h       |  10 +++
>  2 files changed, 90 insertions(+), 41 deletions(-)
> 
> diff --git a/drivers/gpu/drm/drm_dp_mst_topology.c b/drivers/gpu/drm/drm_dp_mst_topology.c
> index 998081b9b205..d7c3d9233834 100644
> --- a/drivers/gpu/drm/drm_dp_mst_topology.c
> +++ b/drivers/gpu/drm/drm_dp_mst_topology.c
> @@ -1108,34 +1108,17 @@ static void drm_dp_destroy_mst_branch_device(struct kref *kref)
>  	struct drm_dp_mst_branch *mstb =
>  		container_of(kref, struct drm_dp_mst_branch, topology_kref);
>  	struct drm_dp_mst_topology_mgr *mgr = mstb->mgr;
> -	struct drm_dp_mst_port *port, *tmp;
> -	bool wake_tx = false;
> -
> -	mutex_lock(&mgr->lock);
> -	list_for_each_entry_safe(port, tmp, &mstb->ports, next) {
> -		list_del(&port->next);
> -		drm_dp_mst_topology_put_port(port);
> -	}
> -	mutex_unlock(&mgr->lock);
> -
> -	/* drop any tx slots msg */
> -	mutex_lock(&mstb->mgr->qlock);
> -	if (mstb->tx_slots[0]) {
> -		mstb->tx_slots[0]->state = DRM_DP_SIDEBAND_TX_TIMEOUT;
> -		mstb->tx_slots[0] = NULL;
> -		wake_tx = true;
> -	}
> -	if (mstb->tx_slots[1]) {
> -		mstb->tx_slots[1]->state = DRM_DP_SIDEBAND_TX_TIMEOUT;
> -		mstb->tx_slots[1] = NULL;
> -		wake_tx = true;
> -	}
> -	mutex_unlock(&mstb->mgr->qlock);
>  
> -	if (wake_tx)
> -		wake_up_all(&mstb->mgr->tx_waitq);
> +	INIT_LIST_HEAD(&mstb->destroy_next);
>  
> -	drm_dp_mst_put_mstb_malloc(mstb);
> +	/*
> +	 * This can get called under mgr->mutex, so we need to perform the
> +	 * actual destruction of the mstb in another worker
> +	 */
> +	mutex_lock(&mgr->destroy_connector_lock);
> +	list_add(&mstb->destroy_next, &mgr->destroy_branch_device_list);
> +	mutex_unlock(&mgr->destroy_connector_lock);
> +	schedule_work(&mgr->destroy_connector_work);
>  }
>  
>  /**
> @@ -3618,10 +3601,56 @@ static void drm_dp_tx_work(struct work_struct *work)
>  	mutex_unlock(&mgr->qlock);
>  }
>  
> +static inline void
> +drm_dp_finish_destroy_port(struct drm_dp_mst_port *port)

Bikeshed: I'd call this _delayed_destroy, I think that makes a bit clearer
why there's 2 stages to destroying stuff.

> +{
> +	INIT_LIST_HEAD(&port->next);
> +
> +	port->mgr->cbs->destroy_connector(port->mgr, port->connector);
> +
> +	drm_dp_port_teardown_pdt(port, port->pdt);
> +	port->pdt = DP_PEER_DEVICE_NONE;
> +
> +	drm_dp_mst_put_port_malloc(port);
> +}
> +
> +static inline void
> +drm_dp_finish_destroy_mst_branch_device(struct drm_dp_mst_branch *mstb)
> +{
> +	struct drm_dp_mst_topology_mgr *mgr = mstb->mgr;
> +	struct drm_dp_mst_port *port, *tmp;
> +	bool wake_tx = false;
> +
> +	mutex_lock(&mgr->lock);
> +	list_for_each_entry_safe(port, tmp, &mstb->ports, next) {
> +		list_del(&port->next);
> +		drm_dp_mst_topology_put_port(port);
> +	}
> +	mutex_unlock(&mgr->lock);
> +
> +	/* drop any tx slots msg */
> +	mutex_lock(&mstb->mgr->qlock);
> +	if (mstb->tx_slots[0]) {
> +		mstb->tx_slots[0]->state = DRM_DP_SIDEBAND_TX_TIMEOUT;
> +		mstb->tx_slots[0] = NULL;
> +		wake_tx = true;
> +	}
> +	if (mstb->tx_slots[1]) {
> +		mstb->tx_slots[1]->state = DRM_DP_SIDEBAND_TX_TIMEOUT;
> +		mstb->tx_slots[1] = NULL;
> +		wake_tx = true;
> +	}
> +	mutex_unlock(&mstb->mgr->qlock);
> +
> +	if (wake_tx)
> +		wake_up_all(&mstb->mgr->tx_waitq);
> +
> +	drm_dp_mst_put_mstb_malloc(mstb);
> +}
> +
>  static void drm_dp_destroy_connector_work(struct work_struct *work)
>  {
>  	struct drm_dp_mst_topology_mgr *mgr = container_of(work, struct drm_dp_mst_topology_mgr, destroy_connector_work);
> -	struct drm_dp_mst_port *port;
>  	bool send_hotplug = false;
>  	/*
>  	 * Not a regular list traverse as we have to drop the destroy
> @@ -3629,24 +3658,33 @@ static void drm_dp_destroy_connector_work(struct work_struct *work)
>  	 * ordering between this lock and the config mutex.
>  	 */
>  	for (;;) {
> +		struct drm_dp_mst_branch *mstb = NULL;
> +		struct drm_dp_mst_port *port = NULL;
> +
> +		/* Destroy any MSTBs first, and then their ports second */
>  		mutex_lock(&mgr->destroy_connector_lock);
> -		port = list_first_entry_or_null(&mgr->destroy_connector_list, struct drm_dp_mst_port, next);
> -		if (!port) {
> -			mutex_unlock(&mgr->destroy_connector_lock);
> -			break;
> +		mstb = list_first_entry_or_null(&mgr->destroy_branch_device_list,
> +						struct drm_dp_mst_branch,
> +						destroy_next);
> +		if (mstb) {
> +			list_del(&mstb->destroy_next);
> +		} else {
> +			port = list_first_entry_or_null(&mgr->destroy_connector_list,
> +							struct drm_dp_mst_port,
> +							next);
> +			if (port)
> +				list_del(&port->next);
>  		}

Control flow looks rather awkward here. I'd do either two loops, or if you
want to have just one, rename it to ->delayed_destroy_list and have a
->delayed_destry_cb next to it?

Cheers, Daniel


> -		list_del(&port->next);
>  		mutex_unlock(&mgr->destroy_connector_lock);
>  
> -		INIT_LIST_HEAD(&port->next);
> -
> -		mgr->cbs->destroy_connector(mgr, port->connector);
> -
> -		drm_dp_port_teardown_pdt(port, port->pdt);
> -		port->pdt = DP_PEER_DEVICE_NONE;
> -
> -		drm_dp_mst_put_port_malloc(port);
> -		send_hotplug = true;
> +		if (mstb) {
> +			drm_dp_finish_destroy_mst_branch_device(mstb);
> +		} else if (port) {
> +			drm_dp_finish_destroy_port(port);
> +			send_hotplug = true;
> +		} else {
> +			break;
> +		}
>  	}
>  	if (send_hotplug)
>  		drm_kms_helper_hotplug_event(mgr->dev);
> @@ -3840,6 +3878,7 @@ int drm_dp_mst_topology_mgr_init(struct drm_dp_mst_topology_mgr *mgr,
>  	mutex_init(&mgr->destroy_connector_lock);
>  	INIT_LIST_HEAD(&mgr->tx_msg_downq);
>  	INIT_LIST_HEAD(&mgr->destroy_connector_list);
> +	INIT_LIST_HEAD(&mgr->destroy_branch_device_list);
>  	INIT_WORK(&mgr->work, drm_dp_mst_link_probe_work);
>  	INIT_WORK(&mgr->tx_work, drm_dp_tx_work);
>  	INIT_WORK(&mgr->destroy_connector_work, drm_dp_destroy_connector_work);
> diff --git a/include/drm/drm_dp_mst_helper.h b/include/drm/drm_dp_mst_helper.h
> index 8c97a5f92c47..c01f3ea72756 100644
> --- a/include/drm/drm_dp_mst_helper.h
> +++ b/include/drm/drm_dp_mst_helper.h
> @@ -143,6 +143,12 @@ struct drm_dp_mst_branch {
>  	 */
>  	struct kref malloc_kref;
>  
> +	/**
> +	 * @destroy_next: linked-list entry used by
> +	 * drm_dp_destroy_connector_work()
> +	 */
> +	struct list_head destroy_next;
> +
>  	u8 rad[8];
>  	u8 lct;
>  	int num_ports;
> @@ -578,6 +584,10 @@ struct drm_dp_mst_topology_mgr {
>  	 * @destroy_connector_list: List of to be destroyed connectors.
>  	 */
>  	struct list_head destroy_connector_list;
> +	/**
> +	 * @destroy_branch_device_list: List of to be destroyed branch devices
> +	 */
> +	struct list_head destroy_branch_device_list;
>  	/**
>  	 * @destroy_connector_lock: Protects @connector_list.
>  	 */
> -- 
> 2.21.0
> 

-- 
Daniel Vetter
Software Engineer, Intel Corporation
http://blog.ffwll.ch

Re: [PATCH 02/26] drm/dp_mst: Destroy mstbs from destroy_connector_work

[Lyude Paul]

On Tue, 2019-08-13 at 15:00 +0200, Daniel Vetter wrote:
> On Wed, Jul 17, 2019 at 09:42:25PM -0400, Lyude Paul wrote:
> > Currently we remove MST branch devices from the in-memory topology
> > immediately when their topology refcount reaches 0. This works just fine
> > at the moment, but since we're about to add suspend/resume reprobing for
> > MST topologies we're going to need to be able to travel through the
> > topology and drop topology refs on branch devices while holding
> > mgr->mutex. Since we currently can't do this due to the circular locking
> > dependencies that would incur, move all of the actual work for
> > drm_dp_destroy_mst_branch_device() into drm_dp_destroy_connector_work()
> > so we can drop topology refs on MSTBs in any locking context.
> 
> Would be good to point at where exactly the problem is here, maybe also
> mentioned the exact future patch that causes the problem. I did go look
> around a bit, but didn't spot it.

Ah, I didn't do a great job of explaining whoops! Basically, the issue is that
when reprobing the state during suspend/resume, there's of course going to be
a chance that any MST ports in the topology could have had their PDT changed
while we were suspended. This in turn means that while we're iterating through
each mstb's ports in drm_dp_send_link_address() we want to be able to remove
MSTBs from while under &mgr->lock, something we can't do without handling MSTB
destruction in another thread. I'll mention this in the next version of this
patch

> 
> > Cc: Juston Li <juston.li@intel.com>
> > Cc: Imre Deak <imre.deak@intel.com>
> > Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
> > Cc: Harry Wentland <hwentlan@amd.com>
> > Signed-off-by: Lyude Paul <lyude@redhat.com>
> > ---
> >  drivers/gpu/drm/drm_dp_mst_topology.c | 121 +++++++++++++++++---------
> >  include/drm/drm_dp_mst_helper.h       |  10 +++
> >  2 files changed, 90 insertions(+), 41 deletions(-)
> > 
> > diff --git a/drivers/gpu/drm/drm_dp_mst_topology.c
> > b/drivers/gpu/drm/drm_dp_mst_topology.c
> > index 998081b9b205..d7c3d9233834 100644
> > --- a/drivers/gpu/drm/drm_dp_mst_topology.c
> > +++ b/drivers/gpu/drm/drm_dp_mst_topology.c
> > @@ -1108,34 +1108,17 @@ static void
> > drm_dp_destroy_mst_branch_device(struct kref *kref)
> >  	struct drm_dp_mst_branch *mstb =
> >  		container_of(kref, struct drm_dp_mst_branch, topology_kref);
> >  	struct drm_dp_mst_topology_mgr *mgr = mstb->mgr;
> > -	struct drm_dp_mst_port *port, *tmp;
> > -	bool wake_tx = false;
> > -
> > -	mutex_lock(&mgr->lock);
> > -	list_for_each_entry_safe(port, tmp, &mstb->ports, next) {
> > -		list_del(&port->next);
> > -		drm_dp_mst_topology_put_port(port);
> > -	}
> > -	mutex_unlock(&mgr->lock);
> > -
> > -	/* drop any tx slots msg */
> > -	mutex_lock(&mstb->mgr->qlock);
> > -	if (mstb->tx_slots[0]) {
> > -		mstb->tx_slots[0]->state = DRM_DP_SIDEBAND_TX_TIMEOUT;
> > -		mstb->tx_slots[0] = NULL;
> > -		wake_tx = true;
> > -	}
> > -	if (mstb->tx_slots[1]) {
> > -		mstb->tx_slots[1]->state = DRM_DP_SIDEBAND_TX_TIMEOUT;
> > -		mstb->tx_slots[1] = NULL;
> > -		wake_tx = true;
> > -	}
> > -	mutex_unlock(&mstb->mgr->qlock);
> >  
> > -	if (wake_tx)
> > -		wake_up_all(&mstb->mgr->tx_waitq);
> > +	INIT_LIST_HEAD(&mstb->destroy_next);
> >  
> > -	drm_dp_mst_put_mstb_malloc(mstb);
> > +	/*
> > +	 * This can get called under mgr->mutex, so we need to perform the
> > +	 * actual destruction of the mstb in another worker
> > +	 */
> > +	mutex_lock(&mgr->destroy_connector_lock);
> > +	list_add(&mstb->destroy_next, &mgr->destroy_branch_device_list);
> > +	mutex_unlock(&mgr->destroy_connector_lock);
> > +	schedule_work(&mgr->destroy_connector_work);
> >  }
> >  
> >  /**
> > @@ -3618,10 +3601,56 @@ static void drm_dp_tx_work(struct work_struct
> > *work)
> >  	mutex_unlock(&mgr->qlock);
> >  }
> >  
> > +static inline void
> > +drm_dp_finish_destroy_port(struct drm_dp_mst_port *port)
> 
> Bikeshed: I'd call this _delayed_destroy, I think that makes a bit clearer
> why there's 2 stages to destroying stuff.
> 
> > +{
> > +	INIT_LIST_HEAD(&port->next);
> > +
> > +	port->mgr->cbs->destroy_connector(port->mgr, port->connector);
> > +
> > +	drm_dp_port_teardown_pdt(port, port->pdt);
> > +	port->pdt = DP_PEER_DEVICE_NONE;
> > +
> > +	drm_dp_mst_put_port_malloc(port);
> > +}
> > +
> > +static inline void
> > +drm_dp_finish_destroy_mst_branch_device(struct drm_dp_mst_branch *mstb)
> > +{
> > +	struct drm_dp_mst_topology_mgr *mgr = mstb->mgr;
> > +	struct drm_dp_mst_port *port, *tmp;
> > +	bool wake_tx = false;
> > +
> > +	mutex_lock(&mgr->lock);
> > +	list_for_each_entry_safe(port, tmp, &mstb->ports, next) {
> > +		list_del(&port->next);
> > +		drm_dp_mst_topology_put_port(port);
> > +	}
> > +	mutex_unlock(&mgr->lock);
> > +
> > +	/* drop any tx slots msg */
> > +	mutex_lock(&mstb->mgr->qlock);
> > +	if (mstb->tx_slots[0]) {
> > +		mstb->tx_slots[0]->state = DRM_DP_SIDEBAND_TX_TIMEOUT;
> > +		mstb->tx_slots[0] = NULL;
> > +		wake_tx = true;
> > +	}
> > +	if (mstb->tx_slots[1]) {
> > +		mstb->tx_slots[1]->state = DRM_DP_SIDEBAND_TX_TIMEOUT;
> > +		mstb->tx_slots[1] = NULL;
> > +		wake_tx = true;
> > +	}
> > +	mutex_unlock(&mstb->mgr->qlock);
> > +
> > +	if (wake_tx)
> > +		wake_up_all(&mstb->mgr->tx_waitq);
> > +
> > +	drm_dp_mst_put_mstb_malloc(mstb);
> > +}
> > +
> >  static void drm_dp_destroy_connector_work(struct work_struct *work)
> >  {
> >  	struct drm_dp_mst_topology_mgr *mgr = container_of(work, struct
> > drm_dp_mst_topology_mgr, destroy_connector_work);
> > -	struct drm_dp_mst_port *port;
> >  	bool send_hotplug = false;
> >  	/*
> >  	 * Not a regular list traverse as we have to drop the destroy
> > @@ -3629,24 +3658,33 @@ static void drm_dp_destroy_connector_work(struct
> > work_struct *work)
> >  	 * ordering between this lock and the config mutex.
> >  	 */
> >  	for (;;) {
> > +		struct drm_dp_mst_branch *mstb = NULL;
> > +		struct drm_dp_mst_port *port = NULL;
> > +
> > +		/* Destroy any MSTBs first, and then their ports second */
> >  		mutex_lock(&mgr->destroy_connector_lock);
> > -		port = list_first_entry_or_null(&mgr->destroy_connector_list,
> > struct drm_dp_mst_port, next);
> > -		if (!port) {
> > -			mutex_unlock(&mgr->destroy_connector_lock);
> > -			break;
> > +		mstb = list_first_entry_or_null(&mgr-
> > >destroy_branch_device_list,
> > +						struct drm_dp_mst_branch,
> > +						destroy_next);
> > +		if (mstb) {
> > +			list_del(&mstb->destroy_next);
> > +		} else {
> > +			port = list_first_entry_or_null(&mgr-
> > >destroy_connector_list,
> > +							struct
> > drm_dp_mst_port,
> > +							next);
> > +			if (port)
> > +				list_del(&port->next);
> >  		}
> 
> Control flow looks rather awkward here. I'd do either two loops, or if you
> want to have just one, rename it to ->delayed_destroy_list and have a
> ->delayed_destry_cb next to it?
> 
> Cheers, Daniel
> 
> 
> > -		list_del(&port->next);
> >  		mutex_unlock(&mgr->destroy_connector_lock);
> >  
> > -		INIT_LIST_HEAD(&port->next);
> > -
> > -		mgr->cbs->destroy_connector(mgr, port->connector);
> > -
> > -		drm_dp_port_teardown_pdt(port, port->pdt);
> > -		port->pdt = DP_PEER_DEVICE_NONE;
> > -
> > -		drm_dp_mst_put_port_malloc(port);
> > -		send_hotplug = true;
> > +		if (mstb) {
> > +			drm_dp_finish_destroy_mst_branch_device(mstb);
> > +		} else if (port) {
> > +			drm_dp_finish_destroy_port(port);
> > +			send_hotplug = true;
> > +		} else {
> > +			break;
> > +		}
> >  	}
> >  	if (send_hotplug)
> >  		drm_kms_helper_hotplug_event(mgr->dev);
> > @@ -3840,6 +3878,7 @@ int drm_dp_mst_topology_mgr_init(struct
> > drm_dp_mst_topology_mgr *mgr,
> >  	mutex_init(&mgr->destroy_connector_lock);
> >  	INIT_LIST_HEAD(&mgr->tx_msg_downq);
> >  	INIT_LIST_HEAD(&mgr->destroy_connector_list);
> > +	INIT_LIST_HEAD(&mgr->destroy_branch_device_list);
> >  	INIT_WORK(&mgr->work, drm_dp_mst_link_probe_work);
> >  	INIT_WORK(&mgr->tx_work, drm_dp_tx_work);
> >  	INIT_WORK(&mgr->destroy_connector_work,
> > drm_dp_destroy_connector_work);
> > diff --git a/include/drm/drm_dp_mst_helper.h
> > b/include/drm/drm_dp_mst_helper.h
> > index 8c97a5f92c47..c01f3ea72756 100644
> > --- a/include/drm/drm_dp_mst_helper.h
> > +++ b/include/drm/drm_dp_mst_helper.h
> > @@ -143,6 +143,12 @@ struct drm_dp_mst_branch {
> >  	 */
> >  	struct kref malloc_kref;
> >  
> > +	/**
> > +	 * @destroy_next: linked-list entry used by
> > +	 * drm_dp_destroy_connector_work()
> > +	 */
> > +	struct list_head destroy_next;
> > +
> >  	u8 rad[8];
> >  	u8 lct;
> >  	int num_ports;
> > @@ -578,6 +584,10 @@ struct drm_dp_mst_topology_mgr {
> >  	 * @destroy_connector_list: List of to be destroyed connectors.
> >  	 */
> >  	struct list_head destroy_connector_list;
> > +	/**
> > +	 * @destroy_branch_device_list: List of to be destroyed branch devices
> > +	 */
> > +	struct list_head destroy_branch_device_list;
> >  	/**
> >  	 * @destroy_connector_lock: Protects @connector_list.
> >  	 */
> > -- 
> > 2.21.0
> > 
-- 
Cheers,
	Lyude Paul

[PATCH 03/26] drm/dp_mst: Move test_calc_pbn_mode() into an actual selftest

[Lyude Paul]

Yes, apparently we've been testing this for every single driver load for
quite a long time now. At least that means our PBN calculation is solid!

Anyway, introduce self tests for MST and move this into there.

Cc: Juston Li <juston.li@intel.com>
Cc: Imre Deak <imre.deak@intel.com>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Cc: Harry Wentland <hwentlan@amd.com>
Signed-off-by: Lyude Paul <lyude@redhat.com>
---
 drivers/gpu/drm/drm_dp_mst_topology.c         | 27 --------------
 drivers/gpu/drm/selftests/Makefile            |  2 +-
 .../gpu/drm/selftests/drm_modeset_selftests.h |  1 +
 .../drm/selftests/test-drm_dp_mst_helper.c    | 36 +++++++++++++++++++
 .../drm/selftests/test-drm_modeset_common.h   |  1 +
 5 files changed, 39 insertions(+), 28 deletions(-)
 create mode 100644 drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c

diff --git a/drivers/gpu/drm/drm_dp_mst_topology.c b/drivers/gpu/drm/drm_dp_mst_topology.c
index d7c3d9233834..9e382117896d 100644
--- a/drivers/gpu/drm/drm_dp_mst_topology.c
+++ b/drivers/gpu/drm/drm_dp_mst_topology.c
@@ -45,7 +45,6 @@
  */
 static bool dump_dp_payload_table(struct drm_dp_mst_topology_mgr *mgr,
 				  char *buf);
-static int test_calc_pbn_mode(void);
 
 static void drm_dp_mst_topology_put_port(struct drm_dp_mst_port *port);
 
@@ -3439,30 +3438,6 @@ int drm_dp_calc_pbn_mode(int clock, int bpp)
 }
 EXPORT_SYMBOL(drm_dp_calc_pbn_mode);
 
-static int test_calc_pbn_mode(void)
-{
-	int ret;
-	ret = drm_dp_calc_pbn_mode(154000, 30);
-	if (ret != 689) {
-		DRM_ERROR("PBN calculation test failed - clock %d, bpp %d, expected PBN %d, actual PBN %d.\n",
-				154000, 30, 689, ret);
-		return -EINVAL;
-	}
-	ret = drm_dp_calc_pbn_mode(234000, 30);
-	if (ret != 1047) {
-		DRM_ERROR("PBN calculation test failed - clock %d, bpp %d, expected PBN %d, actual PBN %d.\n",
-				234000, 30, 1047, ret);
-		return -EINVAL;
-	}
-	ret = drm_dp_calc_pbn_mode(297000, 24);
-	if (ret != 1063) {
-		DRM_ERROR("PBN calculation test failed - clock %d, bpp %d, expected PBN %d, actual PBN %d.\n",
-				297000, 24, 1063, ret);
-		return -EINVAL;
-	}
-	return 0;
-}
-
 /* we want to kick the TX after we've ack the up/down IRQs. */
 static void drm_dp_mst_kick_tx(struct drm_dp_mst_topology_mgr *mgr)
 {
@@ -3898,8 +3873,6 @@ int drm_dp_mst_topology_mgr_init(struct drm_dp_mst_topology_mgr *mgr,
 	if (!mgr->proposed_vcpis)
 		return -ENOMEM;
 	set_bit(0, &mgr->payload_mask);
-	if (test_calc_pbn_mode() < 0)
-		DRM_ERROR("MST PBN self-test failed\n");
 
 	mst_state = kzalloc(sizeof(*mst_state), GFP_KERNEL);
 	if (mst_state == NULL)
diff --git a/drivers/gpu/drm/selftests/Makefile b/drivers/gpu/drm/selftests/Makefile
index aae88f8a016c..d2137342b371 100644
--- a/drivers/gpu/drm/selftests/Makefile
+++ b/drivers/gpu/drm/selftests/Makefile
@@ -1,6 +1,6 @@
 # SPDX-License-Identifier: GPL-2.0-only
 test-drm_modeset-y := test-drm_modeset_common.o test-drm_plane_helper.o \
                       test-drm_format.o test-drm_framebuffer.o \
-		      test-drm_damage_helper.o
+		      test-drm_damage_helper.o test-drm_dp_mst_helper.o
 
 obj-$(CONFIG_DRM_DEBUG_SELFTEST) += test-drm_mm.o test-drm_modeset.o test-drm_cmdline_parser.o
diff --git a/drivers/gpu/drm/selftests/drm_modeset_selftests.h b/drivers/gpu/drm/selftests/drm_modeset_selftests.h
index 464753746013..dec3ee3ec96f 100644
--- a/drivers/gpu/drm/selftests/drm_modeset_selftests.h
+++ b/drivers/gpu/drm/selftests/drm_modeset_selftests.h
@@ -32,3 +32,4 @@ selftest(damage_iter_damage_one_intersect, igt_damage_iter_damage_one_intersect)
 selftest(damage_iter_damage_one_outside, igt_damage_iter_damage_one_outside)
 selftest(damage_iter_damage_src_moved, igt_damage_iter_damage_src_moved)
 selftest(damage_iter_damage_not_visible, igt_damage_iter_damage_not_visible)
+selftest(dp_mst_calc_pbn_mode, igt_dp_mst_calc_pbn_mode)
diff --git a/drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c b/drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c
new file mode 100644
index 000000000000..51b2486ec917
--- /dev/null
+++ b/drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c
@@ -0,0 +1,36 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Test cases for for the DRM DP MST helpers
+ */
+
+#define pr_fmt(fmt) "drm_dp_mst_helper: " fmt
+
+#include <drm/drm_dp_mst_helper.h>
+#include <drm/drm_print.h>
+
+#include "test-drm_modeset_common.h"
+
+int igt_dp_mst_calc_pbn_mode(void *ignored)
+{
+	int pbn, i;
+	const struct {
+		int rate;
+		int bpp;
+		int expected;
+	} test_params[] = {
+		{ 154000, 30, 689 },
+		{ 234000, 30, 1047 },
+		{ 297000, 24, 1063 },
+	};
+
+	for (i = 0; i < ARRAY_SIZE(test_params); i++) {
+		pbn = drm_dp_calc_pbn_mode(test_params[i].rate,
+					   test_params[i].bpp);
+		FAIL(pbn != test_params[i].expected,
+		     "Expected PBN %d for clock %d bpp %d, got %d\n",
+		     test_params[i].expected, test_params[i].rate,
+		     test_params[i].bpp, pbn);
+	}
+
+	return 0;
+}
diff --git a/drivers/gpu/drm/selftests/test-drm_modeset_common.h b/drivers/gpu/drm/selftests/test-drm_modeset_common.h
index 8c76f09c12d1..590bda35a683 100644
--- a/drivers/gpu/drm/selftests/test-drm_modeset_common.h
+++ b/drivers/gpu/drm/selftests/test-drm_modeset_common.h
@@ -39,5 +39,6 @@ int igt_damage_iter_damage_one_intersect(void *ignored);
 int igt_damage_iter_damage_one_outside(void *ignored);
 int igt_damage_iter_damage_src_moved(void *ignored);
 int igt_damage_iter_damage_not_visible(void *ignored);
+int igt_dp_mst_calc_pbn_mode(void *ignored);
 
 #endif
-- 
2.21.0

Re: [PATCH 03/26] drm/dp_mst: Move test_calc_pbn_mode() into an actual selftest

[Daniel Vetter]

On Wed, Jul 17, 2019 at 09:42:26PM -0400, Lyude Paul wrote:
> Yes, apparently we've been testing this for every single driver load for
> quite a long time now. At least that means our PBN calculation is solid!
> 
> Anyway, introduce self tests for MST and move this into there.
> 
> Cc: Juston Li <juston.li@intel.com>
> Cc: Imre Deak <imre.deak@intel.com>
> Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
> Cc: Harry Wentland <hwentlan@amd.com>
> Signed-off-by: Lyude Paul <lyude@redhat.com>

More official unit tests, yay!

Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>


> ---
>  drivers/gpu/drm/drm_dp_mst_topology.c         | 27 --------------
>  drivers/gpu/drm/selftests/Makefile            |  2 +-
>  .../gpu/drm/selftests/drm_modeset_selftests.h |  1 +
>  .../drm/selftests/test-drm_dp_mst_helper.c    | 36 +++++++++++++++++++
>  .../drm/selftests/test-drm_modeset_common.h   |  1 +
>  5 files changed, 39 insertions(+), 28 deletions(-)
>  create mode 100644 drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c
> 
> diff --git a/drivers/gpu/drm/drm_dp_mst_topology.c b/drivers/gpu/drm/drm_dp_mst_topology.c
> index d7c3d9233834..9e382117896d 100644
> --- a/drivers/gpu/drm/drm_dp_mst_topology.c
> +++ b/drivers/gpu/drm/drm_dp_mst_topology.c
> @@ -45,7 +45,6 @@
>   */
>  static bool dump_dp_payload_table(struct drm_dp_mst_topology_mgr *mgr,
>  				  char *buf);
> -static int test_calc_pbn_mode(void);
>  
>  static void drm_dp_mst_topology_put_port(struct drm_dp_mst_port *port);
>  
> @@ -3439,30 +3438,6 @@ int drm_dp_calc_pbn_mode(int clock, int bpp)
>  }
>  EXPORT_SYMBOL(drm_dp_calc_pbn_mode);
>  
> -static int test_calc_pbn_mode(void)
> -{
> -	int ret;
> -	ret = drm_dp_calc_pbn_mode(154000, 30);
> -	if (ret != 689) {
> -		DRM_ERROR("PBN calculation test failed - clock %d, bpp %d, expected PBN %d, actual PBN %d.\n",
> -				154000, 30, 689, ret);
> -		return -EINVAL;
> -	}
> -	ret = drm_dp_calc_pbn_mode(234000, 30);
> -	if (ret != 1047) {
> -		DRM_ERROR("PBN calculation test failed - clock %d, bpp %d, expected PBN %d, actual PBN %d.\n",
> -				234000, 30, 1047, ret);
> -		return -EINVAL;
> -	}
> -	ret = drm_dp_calc_pbn_mode(297000, 24);
> -	if (ret != 1063) {
> -		DRM_ERROR("PBN calculation test failed - clock %d, bpp %d, expected PBN %d, actual PBN %d.\n",
> -				297000, 24, 1063, ret);
> -		return -EINVAL;
> -	}
> -	return 0;
> -}
> -
>  /* we want to kick the TX after we've ack the up/down IRQs. */
>  static void drm_dp_mst_kick_tx(struct drm_dp_mst_topology_mgr *mgr)
>  {
> @@ -3898,8 +3873,6 @@ int drm_dp_mst_topology_mgr_init(struct drm_dp_mst_topology_mgr *mgr,
>  	if (!mgr->proposed_vcpis)
>  		return -ENOMEM;
>  	set_bit(0, &mgr->payload_mask);
> -	if (test_calc_pbn_mode() < 0)
> -		DRM_ERROR("MST PBN self-test failed\n");
>  
>  	mst_state = kzalloc(sizeof(*mst_state), GFP_KERNEL);
>  	if (mst_state == NULL)
> diff --git a/drivers/gpu/drm/selftests/Makefile b/drivers/gpu/drm/selftests/Makefile
> index aae88f8a016c..d2137342b371 100644
> --- a/drivers/gpu/drm/selftests/Makefile
> +++ b/drivers/gpu/drm/selftests/Makefile
> @@ -1,6 +1,6 @@
>  # SPDX-License-Identifier: GPL-2.0-only
>  test-drm_modeset-y := test-drm_modeset_common.o test-drm_plane_helper.o \
>                        test-drm_format.o test-drm_framebuffer.o \
> -		      test-drm_damage_helper.o
> +		      test-drm_damage_helper.o test-drm_dp_mst_helper.o
>  
>  obj-$(CONFIG_DRM_DEBUG_SELFTEST) += test-drm_mm.o test-drm_modeset.o test-drm_cmdline_parser.o
> diff --git a/drivers/gpu/drm/selftests/drm_modeset_selftests.h b/drivers/gpu/drm/selftests/drm_modeset_selftests.h
> index 464753746013..dec3ee3ec96f 100644
> --- a/drivers/gpu/drm/selftests/drm_modeset_selftests.h
> +++ b/drivers/gpu/drm/selftests/drm_modeset_selftests.h
> @@ -32,3 +32,4 @@ selftest(damage_iter_damage_one_intersect, igt_damage_iter_damage_one_intersect)
>  selftest(damage_iter_damage_one_outside, igt_damage_iter_damage_one_outside)
>  selftest(damage_iter_damage_src_moved, igt_damage_iter_damage_src_moved)
>  selftest(damage_iter_damage_not_visible, igt_damage_iter_damage_not_visible)
> +selftest(dp_mst_calc_pbn_mode, igt_dp_mst_calc_pbn_mode)
> diff --git a/drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c b/drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c
> new file mode 100644
> index 000000000000..51b2486ec917
> --- /dev/null
> +++ b/drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c
> @@ -0,0 +1,36 @@
> +// SPDX-License-Identifier: GPL-2.0-only
> +/*
> + * Test cases for for the DRM DP MST helpers
> + */
> +
> +#define pr_fmt(fmt) "drm_dp_mst_helper: " fmt
> +
> +#include <drm/drm_dp_mst_helper.h>
> +#include <drm/drm_print.h>
> +
> +#include "test-drm_modeset_common.h"
> +
> +int igt_dp_mst_calc_pbn_mode(void *ignored)
> +{
> +	int pbn, i;
> +	const struct {
> +		int rate;
> +		int bpp;
> +		int expected;
> +	} test_params[] = {
> +		{ 154000, 30, 689 },
> +		{ 234000, 30, 1047 },
> +		{ 297000, 24, 1063 },
> +	};
> +
> +	for (i = 0; i < ARRAY_SIZE(test_params); i++) {
> +		pbn = drm_dp_calc_pbn_mode(test_params[i].rate,
> +					   test_params[i].bpp);
> +		FAIL(pbn != test_params[i].expected,
> +		     "Expected PBN %d for clock %d bpp %d, got %d\n",
> +		     test_params[i].expected, test_params[i].rate,
> +		     test_params[i].bpp, pbn);
> +	}
> +
> +	return 0;
> +}
> diff --git a/drivers/gpu/drm/selftests/test-drm_modeset_common.h b/drivers/gpu/drm/selftests/test-drm_modeset_common.h
> index 8c76f09c12d1..590bda35a683 100644
> --- a/drivers/gpu/drm/selftests/test-drm_modeset_common.h
> +++ b/drivers/gpu/drm/selftests/test-drm_modeset_common.h
> @@ -39,5 +39,6 @@ int igt_damage_iter_damage_one_intersect(void *ignored);
>  int igt_damage_iter_damage_one_outside(void *ignored);
>  int igt_damage_iter_damage_src_moved(void *ignored);
>  int igt_damage_iter_damage_not_visible(void *ignored);
> +int igt_dp_mst_calc_pbn_mode(void *ignored);
>  
>  #endif
> -- 
> 2.21.0
> 

-- 
Daniel Vetter
Software Engineer, Intel Corporation
http://blog.ffwll.ch

[PATCH 04/26] drm/print: Add drm_err_printer()

[Lyude Paul]

A simple convienence function that returns a drm_printer which prints
using pr_err()

Cc: Juston Li <juston.li@intel.com>
Cc: Imre Deak <imre.deak@intel.com>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Cc: Harry Wentland <hwentlan@amd.com>
Signed-off-by: Lyude Paul <lyude@redhat.com>
---
 drivers/gpu/drm/drm_print.c |  6 ++++++
 include/drm/drm_print.h     | 17 +++++++++++++++++
 2 files changed, 23 insertions(+)

diff --git a/drivers/gpu/drm/drm_print.c b/drivers/gpu/drm/drm_print.c
index a17c8a14dba4..6112be358769 100644
--- a/drivers/gpu/drm/drm_print.c
+++ b/drivers/gpu/drm/drm_print.c
@@ -147,6 +147,12 @@ void __drm_printfn_debug(struct drm_printer *p, struct va_format *vaf)
 }
 EXPORT_SYMBOL(__drm_printfn_debug);
 
+void __drm_printfn_err(struct drm_printer *p, struct va_format *vaf)
+{
+	pr_err("%s %pV", p->prefix, vaf);
+}
+EXPORT_SYMBOL(__drm_printfn_err);
+
 /**
  * drm_puts - print a const string to a &drm_printer stream
  * @p: the &drm printer
diff --git a/include/drm/drm_print.h b/include/drm/drm_print.h
index a5d6f2f3e430..112165d3195d 100644
--- a/include/drm/drm_print.h
+++ b/include/drm/drm_print.h
@@ -83,6 +83,7 @@ void __drm_printfn_seq_file(struct drm_printer *p, struct va_format *vaf);
 void __drm_puts_seq_file(struct drm_printer *p, const char *str);
 void __drm_printfn_info(struct drm_printer *p, struct va_format *vaf);
 void __drm_printfn_debug(struct drm_printer *p, struct va_format *vaf);
+void __drm_printfn_err(struct drm_printer *p, struct va_format *vaf);
 
 __printf(2, 3)
 void drm_printf(struct drm_printer *p, const char *f, ...);
@@ -227,6 +228,22 @@ static inline struct drm_printer drm_debug_printer(const char *prefix)
 	return p;
 }
 
+/**
+ * drm_err_printer - construct a &drm_printer that outputs to pr_err()
+ * @prefix: debug output prefix
+ *
+ * RETURNS:
+ * The &drm_printer object
+ */
+static inline struct drm_printer drm_err_printer(const char *prefix)
+{
+	struct drm_printer p = {
+		.printfn = __drm_printfn_err,
+		.prefix = prefix
+	};
+	return p;
+}
+
 /*
  * The following categories are defined:
  *
-- 
2.21.0

Re: [PATCH 04/26] drm/print: Add drm_err_printer()

[Daniel Vetter]

On Wed, Jul 17, 2019 at 09:42:27PM -0400, Lyude Paul wrote:
> A simple convienence function that returns a drm_printer which prints
> using pr_err()
> 
> Cc: Juston Li <juston.li@intel.com>
> Cc: Imre Deak <imre.deak@intel.com>
> Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
> Cc: Harry Wentland <hwentlan@amd.com>
> Signed-off-by: Lyude Paul <lyude@redhat.com>
> ---
>  drivers/gpu/drm/drm_print.c |  6 ++++++
>  include/drm/drm_print.h     | 17 +++++++++++++++++
>  2 files changed, 23 insertions(+)
> 
> diff --git a/drivers/gpu/drm/drm_print.c b/drivers/gpu/drm/drm_print.c
> index a17c8a14dba4..6112be358769 100644
> --- a/drivers/gpu/drm/drm_print.c
> +++ b/drivers/gpu/drm/drm_print.c
> @@ -147,6 +147,12 @@ void __drm_printfn_debug(struct drm_printer *p, struct va_format *vaf)
>  }
>  EXPORT_SYMBOL(__drm_printfn_debug);
>  
> +void __drm_printfn_err(struct drm_printer *p, struct va_format *vaf)
> +{
> +	pr_err("%s %pV", p->prefix, vaf);

DRM printing is a huge bikeshad (or tire fire?). We can't call DRM_ERROR,
but for consistency mabye emulate the layout?

Either way: Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>


> +}
> +EXPORT_SYMBOL(__drm_printfn_err);
> +
>  /**
>   * drm_puts - print a const string to a &drm_printer stream
>   * @p: the &drm printer
> diff --git a/include/drm/drm_print.h b/include/drm/drm_print.h
> index a5d6f2f3e430..112165d3195d 100644
> --- a/include/drm/drm_print.h
> +++ b/include/drm/drm_print.h
> @@ -83,6 +83,7 @@ void __drm_printfn_seq_file(struct drm_printer *p, struct va_format *vaf);
>  void __drm_puts_seq_file(struct drm_printer *p, const char *str);
>  void __drm_printfn_info(struct drm_printer *p, struct va_format *vaf);
>  void __drm_printfn_debug(struct drm_printer *p, struct va_format *vaf);
> +void __drm_printfn_err(struct drm_printer *p, struct va_format *vaf);
>  
>  __printf(2, 3)
>  void drm_printf(struct drm_printer *p, const char *f, ...);
> @@ -227,6 +228,22 @@ static inline struct drm_printer drm_debug_printer(const char *prefix)
>  	return p;
>  }
>  
> +/**
> + * drm_err_printer - construct a &drm_printer that outputs to pr_err()
> + * @prefix: debug output prefix
> + *
> + * RETURNS:
> + * The &drm_printer object
> + */
> +static inline struct drm_printer drm_err_printer(const char *prefix)
> +{
> +	struct drm_printer p = {
> +		.printfn = __drm_printfn_err,
> +		.prefix = prefix
> +	};
> +	return p;
> +}
> +
>  /*
>   * The following categories are defined:
>   *
> -- 
> 2.21.0
> 

-- 
Daniel Vetter
Software Engineer, Intel Corporation
http://blog.ffwll.ch

[PATCH 05/26] drm/dp_mst: Add sideband down request tracing + selftests

[Lyude Paul]

Unfortunately the DP MST helpers do not have much in the way of
debugging utilities. So, let's add some!

This adds basic debugging output for down sideband requests that we send
from the driver, so that we can actually discern what's happening when
sideband requests timeout. Note that with this commit, we'll be dumping
out sideband requests under both of the following conditions:

- When the user has enabled DRM_UT_DP output, of course.
- When the user has enabled DRM_UT_KMS or DRM_UT_DP, and a sideband
  request has failed in some way. This will allow for developers to get
  a basic idea of what's actually happening with failed modesets on MST,
  without needing to have DRM_UT_DP explicitly enabled.

Since there wasn't really a good way of testing that any of this worked,
I ended up writing simple selftests that lightly test sideband message
encoding and decoding as well. Enjoy!

Cc: Juston Li <juston.li@intel.com>
Cc: Imre Deak <imre.deak@intel.com>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Cc: Harry Wentland <hwentlan@amd.com>
Signed-off-by: Lyude Paul <lyude@redhat.com>
---
 drivers/gpu/drm/drm_dp_mst_topology.c         | 308 +++++++++++++++++-
 .../gpu/drm/drm_dp_mst_topology_internal.h    |  24 ++
 .../gpu/drm/selftests/drm_modeset_selftests.h |   1 +
 .../drm/selftests/test-drm_dp_mst_helper.c    | 212 ++++++++++++
 .../drm/selftests/test-drm_modeset_common.h   |   1 +
 include/drm/drm_dp_mst_helper.h               |   2 +-
 6 files changed, 543 insertions(+), 5 deletions(-)
 create mode 100644 drivers/gpu/drm/drm_dp_mst_topology_internal.h

diff --git a/drivers/gpu/drm/drm_dp_mst_topology.c b/drivers/gpu/drm/drm_dp_mst_topology.c
index 9e382117896d..defc5e09fb9a 100644
--- a/drivers/gpu/drm/drm_dp_mst_topology.c
+++ b/drivers/gpu/drm/drm_dp_mst_topology.c
@@ -36,6 +36,8 @@
 #include <drm/drm_print.h>
 #include <drm/drm_probe_helper.h>
 
+#include "drm_dp_mst_topology_internal.h"
+
 /**
  * DOC: dp mst helper
  *
@@ -68,6 +70,8 @@ static int drm_dp_mst_register_i2c_bus(struct drm_dp_aux *aux);
 static void drm_dp_mst_unregister_i2c_bus(struct drm_dp_aux *aux);
 static void drm_dp_mst_kick_tx(struct drm_dp_mst_topology_mgr *mgr);
 
+#define DBG_PREFIX "[dp_mst]"
+
 #define DP_STR(x) [DP_ ## x] = #x
 
 static const char *drm_dp_mst_req_type_str(u8 req_type)
@@ -124,6 +128,43 @@ static const char *drm_dp_mst_nak_reason_str(u8 nak_reason)
 }
 
 #undef DP_STR
+#define DP_STR(x) [DRM_DP_SIDEBAND_TX_ ## x] = #x
+
+static const char *drm_dp_mst_sideband_tx_state_str(int state)
+{
+	static const char * const sideband_reason_str[] = {
+		DP_STR(QUEUED),
+		DP_STR(START_SEND),
+		DP_STR(SENT),
+		DP_STR(RX),
+		DP_STR(TIMEOUT),
+	};
+
+	if (state >= ARRAY_SIZE(sideband_reason_str) ||
+	    !sideband_reason_str[state])
+		return "unknown";
+
+	return sideband_reason_str[state];
+}
+
+static int
+drm_dp_mst_rad_to_str(const u8 rad[8], u8 lct, char *out, size_t len)
+{
+	int i;
+	u8 unpacked_rad[16];
+
+	for (i = 0; i < lct; i++) {
+		if (i % 2)
+			unpacked_rad[i] = rad[i / 2] >> 4;
+		else
+			unpacked_rad[i] = rad[i / 2] & BIT_MASK(4);
+	}
+
+	/* TODO: Eventually add something to printk so we can format the rad
+	 * like this: 1.2.3
+	 */
+	return snprintf(out, len, "%*phC", lct, unpacked_rad);
+}
 
 /* sideband msg handling */
 static u8 drm_dp_msg_header_crc4(const uint8_t *data, size_t num_nibbles)
@@ -256,8 +297,9 @@ static bool drm_dp_decode_sideband_msg_hdr(struct drm_dp_sideband_msg_hdr *hdr,
 	return true;
 }
 
-static void drm_dp_encode_sideband_req(struct drm_dp_sideband_msg_req_body *req,
-				       struct drm_dp_sideband_msg_tx *raw)
+void
+drm_dp_encode_sideband_req(const struct drm_dp_sideband_msg_req_body *req,
+			   struct drm_dp_sideband_msg_tx *raw)
 {
 	int idx = 0;
 	int i;
@@ -362,6 +404,249 @@ static void drm_dp_encode_sideband_req(struct drm_dp_sideband_msg_req_body *req,
 	}
 	raw->cur_len = idx;
 }
+EXPORT_SYMBOL_FOR_TESTS_ONLY(drm_dp_encode_sideband_req);
+
+/* Decode a sideband request we've encoded, mainly used for debugging */
+int
+drm_dp_decode_sideband_req(const struct drm_dp_sideband_msg_tx *raw,
+			   struct drm_dp_sideband_msg_req_body *req)
+{
+	const u8 *buf = raw->msg;
+	int i, idx = 0;
+
+	req->req_type = buf[idx++] & 0x7f;
+	switch (req->req_type) {
+	case DP_ENUM_PATH_RESOURCES:
+	case DP_POWER_DOWN_PHY:
+	case DP_POWER_UP_PHY:
+		req->u.port_num.port_number = (buf[idx] >> 4) & 0xf;
+		break;
+	case DP_ALLOCATE_PAYLOAD:
+		{
+			struct drm_dp_allocate_payload *a =
+				&req->u.allocate_payload;
+
+			a->number_sdp_streams = buf[idx] & 0xf;
+			a->port_number = (buf[idx] >> 4) & 0xf;
+
+			a->vcpi = buf[++idx] & 0x7f;
+
+			a->pbn = buf[++idx] << 8;
+			a->pbn |= buf[++idx];
+
+			idx++;
+			for (i = 0; i < a->number_sdp_streams; i++) {
+				a->sdp_stream_sink[i] =
+					(buf[idx + (i / 2)] >> ((i % 2) ? 0 : 4)) & 0xf;
+			}
+		}
+		break;
+	case DP_QUERY_PAYLOAD:
+		req->u.query_payload.port_number = (buf[idx] >> 4) & 0xf;
+		req->u.query_payload.vcpi = buf[++idx] & 0x7f;
+		break;
+	case DP_REMOTE_DPCD_READ:
+		{
+			struct drm_dp_remote_dpcd_read *r = &req->u.dpcd_read;
+
+			r->port_number = (buf[idx] >> 4) & 0xf;
+
+			r->dpcd_address = (buf[idx] << 16) & 0xf0000;
+			r->dpcd_address |= (buf[++idx] << 8) & 0xff00;
+			r->dpcd_address |= buf[++idx] & 0xff;
+
+			r->num_bytes = buf[++idx];
+		}
+		break;
+	case DP_REMOTE_DPCD_WRITE:
+		{
+			struct drm_dp_remote_dpcd_write *w =
+				&req->u.dpcd_write;
+
+			w->port_number = (buf[idx] >> 4) & 0xf;
+
+			w->dpcd_address = (buf[idx] << 16) & 0xf0000;
+			w->dpcd_address |= (buf[++idx] << 8) & 0xff00;
+			w->dpcd_address |= buf[++idx] & 0xff;
+
+			w->num_bytes = buf[++idx];
+
+			w->bytes = kmemdup(&buf[++idx], w->num_bytes,
+					   GFP_KERNEL);
+			if (!w->bytes)
+				return -ENOMEM;
+		}
+		break;
+	case DP_REMOTE_I2C_READ:
+		{
+			struct drm_dp_remote_i2c_read *r = &req->u.i2c_read;
+			struct drm_dp_remote_i2c_read_tx *tx;
+			bool failed = false;
+
+			r->num_transactions = buf[idx] & 0x3;
+			r->port_number = (buf[idx] >> 4) & 0xf;
+			for (i = 0; i < r->num_transactions; i++) {
+				tx = &r->transactions[i];
+
+				tx->i2c_dev_id = buf[++idx] & 0x7f;
+				tx->num_bytes = buf[++idx];
+				tx->bytes = kmemdup(&buf[++idx],
+						    tx->num_bytes,
+						    GFP_KERNEL);
+				if (!tx->bytes) {
+					failed = true;
+					break;
+				}
+				idx += tx->num_bytes;
+				tx->no_stop_bit = (buf[idx] >> 5) & 0x1;
+				tx->i2c_transaction_delay = buf[idx] & 0xf;
+			}
+
+			if (failed) {
+				for (i = 0; i < r->num_transactions; i++)
+					kfree(tx->bytes);
+				return -ENOMEM;
+			}
+
+			r->read_i2c_device_id = buf[++idx] & 0x7f;
+			r->num_bytes_read = buf[++idx];
+		}
+		break;
+	case DP_REMOTE_I2C_WRITE:
+		{
+			struct drm_dp_remote_i2c_write *w = &req->u.i2c_write;
+
+			w->port_number = (buf[idx] >> 4) & 0xf;
+			w->write_i2c_device_id = buf[++idx] & 0x7f;
+			w->num_bytes = buf[++idx];
+			w->bytes = kmemdup(&buf[++idx], w->num_bytes,
+					   GFP_KERNEL);
+			if (!w->bytes)
+				return -ENOMEM;
+		}
+		break;
+	}
+
+	return 0;
+}
+EXPORT_SYMBOL_FOR_TESTS_ONLY(drm_dp_decode_sideband_req);
+
+void
+drm_dp_dump_sideband_msg_req_body(const struct drm_dp_sideband_msg_req_body *req,
+				  int indent, struct drm_printer *printer)
+{
+	int i;
+
+#define P(f, ...) drm_printf_indent(printer, indent, f, ##__VA_ARGS__)
+	if (req->req_type == DP_LINK_ADDRESS) {
+		/* No contents to print */
+		P("type=%s\n", drm_dp_mst_req_type_str(req->req_type));
+		return;
+	}
+
+	P("type=%s contents:\n", drm_dp_mst_req_type_str(req->req_type));
+	indent++;
+
+	switch (req->req_type) {
+	case DP_ENUM_PATH_RESOURCES:
+	case DP_POWER_DOWN_PHY:
+	case DP_POWER_UP_PHY:
+		P("port=%d\n", req->u.port_num.port_number);
+		break;
+	case DP_ALLOCATE_PAYLOAD:
+		P("port=%d vcpi=%d pbn=%d sdp_streams=%d %*ph\n",
+		  req->u.allocate_payload.port_number,
+		  req->u.allocate_payload.vcpi, req->u.allocate_payload.pbn,
+		  req->u.allocate_payload.number_sdp_streams,
+		  req->u.allocate_payload.number_sdp_streams,
+		  req->u.allocate_payload.sdp_stream_sink);
+		break;
+	case DP_QUERY_PAYLOAD:
+		P("port=%d vcpi=%d\n",
+		  req->u.query_payload.port_number,
+		  req->u.query_payload.vcpi);
+		break;
+	case DP_REMOTE_DPCD_READ:
+		P("port=%d dpcd_addr=%05x len=%d\n",
+		  req->u.dpcd_read.port_number, req->u.dpcd_read.dpcd_address,
+		  req->u.dpcd_read.num_bytes);
+		break;
+	case DP_REMOTE_DPCD_WRITE:
+		P("port=%d addr=%05x len=%d: %*ph\n",
+		  req->u.dpcd_write.port_number,
+		  req->u.dpcd_write.dpcd_address,
+		  req->u.dpcd_write.num_bytes, req->u.dpcd_write.num_bytes,
+		  req->u.dpcd_write.bytes);
+		break;
+	case DP_REMOTE_I2C_READ:
+		P("port=%d num_tx=%d id=%d size=%d:\n",
+		  req->u.i2c_read.port_number,
+		  req->u.i2c_read.num_transactions,
+		  req->u.i2c_read.read_i2c_device_id,
+		  req->u.i2c_read.num_bytes_read);
+
+		indent++;
+		for (i = 0; i < req->u.i2c_read.num_transactions; i++) {
+			const struct drm_dp_remote_i2c_read_tx *rtx =
+				&req->u.i2c_read.transactions[i];
+
+			P("%d: id=%03d size=%03d no_stop_bit=%d tx_delay=%03d: %*ph\n",
+			  i, rtx->i2c_dev_id, rtx->num_bytes,
+			  rtx->no_stop_bit, rtx->i2c_transaction_delay,
+			  rtx->num_bytes, rtx->bytes);
+		}
+		break;
+	case DP_REMOTE_I2C_WRITE:
+		P("port=%d id=%d size=%d: %*ph\n",
+		  req->u.i2c_write.port_number,
+		  req->u.i2c_write.write_i2c_device_id,
+		  req->u.i2c_write.num_bytes, req->u.i2c_write.num_bytes,
+		  req->u.i2c_write.bytes);
+		break;
+	default:
+		P("???\n");
+		break;
+	}
+#undef P
+}
+EXPORT_SYMBOL_FOR_TESTS_ONLY(drm_dp_dump_sideband_msg_req_body);
+
+static inline void
+drm_dp_mst_dump_sideband_msg_tx(struct drm_printer *p,
+				const struct drm_dp_sideband_msg_tx *txmsg)
+{
+	struct drm_dp_sideband_msg_req_body req;
+	char buf[64];
+	int ret;
+	int i;
+
+	drm_dp_mst_rad_to_str(txmsg->dst->rad, txmsg->dst->lct, buf,
+			      sizeof(buf));
+	drm_printf(p, "txmsg cur_offset=%x cur_len=%x seqno=%x state=%s path_msg=%d dst=%s\n",
+		   txmsg->cur_offset, txmsg->cur_len, txmsg->seqno,
+		   drm_dp_mst_sideband_tx_state_str(txmsg->state),
+		   txmsg->path_msg, buf);
+
+	ret = drm_dp_decode_sideband_req(txmsg, &req);
+	if (ret) {
+		drm_printf(p, "<failed to decode sideband req: %d>\n", ret);
+		return;
+	}
+	drm_dp_dump_sideband_msg_req_body(&req, 1, p);
+
+	switch (req.req_type) {
+	case DP_REMOTE_DPCD_WRITE:
+		kfree(req.u.dpcd_write.bytes);
+		break;
+	case DP_REMOTE_I2C_READ:
+		for (i = 0; i < req.u.i2c_read.num_transactions; i++)
+			kfree(req.u.i2c_read.transactions[i].bytes);
+		break;
+	case DP_REMOTE_I2C_WRITE:
+		kfree(req.u.i2c_write.bytes);
+		break;
+	}
+}
 
 static void drm_dp_crc_sideband_chunk_req(u8 *msg, u8 len)
 {
@@ -893,6 +1178,11 @@ static int drm_dp_mst_wait_tx_reply(struct drm_dp_mst_branch *mstb,
 		}
 	}
 out:
+	if (unlikely(ret == -EIO && drm_debug & (DRM_UT_DP | DRM_UT_KMS))) {
+		struct drm_printer p = drm_debug_printer(DBG_PREFIX);
+
+		drm_dp_mst_dump_sideband_msg_tx(&p, txmsg);
+	}
 	mutex_unlock(&mgr->qlock);
 
 	return ret;
@@ -1927,8 +2217,11 @@ static int process_single_tx_qlock(struct drm_dp_mst_topology_mgr *mgr,
 	idx += tosend + 1;
 
 	ret = drm_dp_send_sideband_msg(mgr, up, chunk, idx);
-	if (ret) {
-		DRM_DEBUG_KMS("sideband msg failed to send\n");
+	if (ret && unlikely(drm_debug & (DRM_UT_DP | DRM_UT_KMS))) {
+		struct drm_printer p = drm_debug_printer(DBG_PREFIX);
+
+		drm_printf(&p, "sideband msg failed to send\n");
+		drm_dp_mst_dump_sideband_msg_tx(&p, txmsg);
 		return ret;
 	}
 
@@ -1990,6 +2283,13 @@ static void drm_dp_queue_down_tx(struct drm_dp_mst_topology_mgr *mgr,
 {
 	mutex_lock(&mgr->qlock);
 	list_add_tail(&txmsg->next, &mgr->tx_msg_downq);
+
+	if (unlikely(drm_debug & DRM_UT_DP)) {
+		struct drm_printer p = drm_debug_printer(DBG_PREFIX);
+
+		drm_dp_mst_dump_sideband_msg_tx(&p, txmsg);
+	}
+
 	if (list_is_singular(&mgr->tx_msg_downq))
 		process_single_down_tx_qlock(mgr);
 	mutex_unlock(&mgr->qlock);
diff --git a/drivers/gpu/drm/drm_dp_mst_topology_internal.h b/drivers/gpu/drm/drm_dp_mst_topology_internal.h
new file mode 100644
index 000000000000..eeda9a61c657
--- /dev/null
+++ b/drivers/gpu/drm/drm_dp_mst_topology_internal.h
@@ -0,0 +1,24 @@
+/* SPDX-License-Identifier: GPL-2.0-only
+ *
+ * Declarations for DP MST related functions which are only used in selftests
+ *
+ * Copyright © 2018 Red Hat
+ * Authors:
+ *     Lyude Paul <lyude@redhat.com>
+ */
+
+#ifndef _DRM_DP_MST_HELPER_INTERNAL_H_
+#define _DRM_DP_MST_HELPER_INTERNAL_H_
+
+#include <drm/drm_dp_mst_helper.h>
+
+void
+drm_dp_encode_sideband_req(const struct drm_dp_sideband_msg_req_body *req,
+			   struct drm_dp_sideband_msg_tx *raw);
+int drm_dp_decode_sideband_req(const struct drm_dp_sideband_msg_tx *raw,
+			       struct drm_dp_sideband_msg_req_body *req);
+void
+drm_dp_dump_sideband_msg_req_body(const struct drm_dp_sideband_msg_req_body *req,
+				  int indent, struct drm_printer *printer);
+
+#endif /* !_DRM_DP_MST_HELPER_INTERNAL_H_ */
diff --git a/drivers/gpu/drm/selftests/drm_modeset_selftests.h b/drivers/gpu/drm/selftests/drm_modeset_selftests.h
index dec3ee3ec96f..1898de0b4a4d 100644
--- a/drivers/gpu/drm/selftests/drm_modeset_selftests.h
+++ b/drivers/gpu/drm/selftests/drm_modeset_selftests.h
@@ -33,3 +33,4 @@ selftest(damage_iter_damage_one_outside, igt_damage_iter_damage_one_outside)
 selftest(damage_iter_damage_src_moved, igt_damage_iter_damage_src_moved)
 selftest(damage_iter_damage_not_visible, igt_damage_iter_damage_not_visible)
 selftest(dp_mst_calc_pbn_mode, igt_dp_mst_calc_pbn_mode)
+selftest(dp_mst_sideband_msg_req_decode, igt_dp_mst_sideband_msg_req_decode)
diff --git a/drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c b/drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c
index 51b2486ec917..ceca89babd65 100644
--- a/drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c
+++ b/drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c
@@ -8,6 +8,7 @@
 #include <drm/drm_dp_mst_helper.h>
 #include <drm/drm_print.h>
 
+#include "../drm_dp_mst_topology_internal.h"
 #include "test-drm_modeset_common.h"
 
 int igt_dp_mst_calc_pbn_mode(void *ignored)
@@ -34,3 +35,214 @@ int igt_dp_mst_calc_pbn_mode(void *ignored)
 
 	return 0;
 }
+
+static bool
+sideband_msg_req_equal(const struct drm_dp_sideband_msg_req_body *in,
+		       const struct drm_dp_sideband_msg_req_body *out)
+{
+	const struct drm_dp_remote_i2c_read_tx *txin, *txout;
+	int i;
+
+	if (in->req_type != out->req_type)
+		return false;
+
+	switch (in->req_type) {
+	case DP_ENUM_PATH_RESOURCES:
+	case DP_POWER_UP_PHY:
+	case DP_POWER_DOWN_PHY:
+	case DP_ALLOCATE_PAYLOAD:
+	case DP_QUERY_PAYLOAD:
+	case DP_REMOTE_DPCD_READ:
+		return memcmp(in, out, sizeof(*in)) == 0;
+
+	case DP_REMOTE_I2C_READ:
+#define IN in->u.i2c_read
+#define OUT out->u.i2c_read
+		if (IN.num_bytes_read != OUT.num_bytes_read ||
+		    IN.num_transactions != OUT.num_transactions ||
+		    IN.port_number != OUT.port_number ||
+		    IN.read_i2c_device_id != OUT.read_i2c_device_id)
+			return false;
+
+		for (i = 0; i < IN.num_transactions; i++) {
+			txin = &IN.transactions[i];
+			txout = &OUT.transactions[i];
+
+			if (txin->i2c_dev_id != txout->i2c_dev_id ||
+			    txin->no_stop_bit != txout->no_stop_bit ||
+			    txin->num_bytes != txout->num_bytes ||
+			    txin->i2c_transaction_delay !=
+			    txout->i2c_transaction_delay)
+				return false;
+
+			if (memcmp(txin->bytes, txout->bytes,
+				   txin->num_bytes) != 0)
+				return false;
+		}
+		break;
+#undef IN
+#undef OUT
+
+	case DP_REMOTE_DPCD_WRITE:
+#define IN in->u.dpcd_write
+#define OUT out->u.dpcd_write
+		if (IN.dpcd_address != OUT.dpcd_address ||
+		    IN.num_bytes != OUT.num_bytes ||
+		    IN.port_number != OUT.port_number)
+			return false;
+
+		return memcmp(IN.bytes, OUT.bytes, IN.num_bytes) == 0;
+#undef IN
+#undef OUT
+
+	case DP_REMOTE_I2C_WRITE:
+#define IN in->u.i2c_write
+#define OUT out->u.i2c_write
+		if (IN.port_number != OUT.port_number ||
+		    IN.write_i2c_device_id != OUT.write_i2c_device_id ||
+		    IN.num_bytes != OUT.num_bytes)
+			return false;
+
+		return memcmp(IN.bytes, OUT.bytes, IN.num_bytes) == 0;
+#undef IN
+#undef OUT
+	}
+
+	return true;
+}
+
+static int
+__sideband_msg_req_encode_decode(int line,
+				 struct drm_dp_sideband_msg_req_body *in,
+				 struct drm_dp_sideband_msg_req_body *out)
+{
+	struct drm_printer p = drm_err_printer(pr_fmt());
+	struct drm_dp_sideband_msg_tx txmsg;
+	int i, ret;
+	bool eq;
+
+	drm_dp_encode_sideband_req(in, &txmsg);
+	ret = drm_dp_decode_sideband_req(&txmsg, out);
+	if (ret < 0) {
+		pr_err(pr_fmt("Failed to decode sideband request @ line %d: %d\n"),
+		       line, ret);
+		return ret;
+	}
+
+	eq = sideband_msg_req_equal(in, out);
+	if (!eq) {
+		pr_err(pr_fmt("Encode/decode @ line %d failed, expected:\n"),
+		       line);
+		drm_dp_dump_sideband_msg_req_body(in, 1, &p);
+		pr_err(pr_fmt("Got:\n"));
+		drm_dp_dump_sideband_msg_req_body(out, 1, &p);
+	}
+
+	switch (in->req_type) {
+	case DP_REMOTE_DPCD_WRITE:
+		kfree(out->u.dpcd_write.bytes);
+		break;
+	case DP_REMOTE_I2C_READ:
+		for (i = 0; i < out->u.i2c_read.num_transactions; i++)
+			kfree(out->u.i2c_read.transactions[i].bytes);
+		break;
+	case DP_REMOTE_I2C_WRITE:
+		kfree(out->u.i2c_write.bytes);
+		break;
+	}
+
+	/* Clear everything but the req_type for the input */
+	memset(&in->u, 0, sizeof(in->u));
+
+	/* Clear the output entirely */
+	memset(out, 0, sizeof(*out));
+
+	return eq ? 0 : -EINVAL;
+}
+
+int igt_dp_mst_sideband_msg_req_decode(void *unused)
+{
+	struct drm_dp_sideband_msg_req_body in = { 0 }, out = { 0 };
+	u8 data[] = { 0xff, 0x0, 0xdd };
+	int i;
+
+#define DO_TEST()                                                          \
+	do {                                                               \
+		if (__sideband_msg_req_encode_decode(__LINE__, &in, &out)) \
+			return -EINVAL;                                    \
+	} while (0)
+
+	in.req_type = DP_ENUM_PATH_RESOURCES;
+	in.u.port_num.port_number = 5;
+	DO_TEST();
+
+	in.req_type = DP_POWER_UP_PHY;
+	in.u.port_num.port_number = 5;
+	DO_TEST();
+
+	in.req_type = DP_POWER_DOWN_PHY;
+	in.u.port_num.port_number = 5;
+	DO_TEST();
+
+	in.req_type = DP_ALLOCATE_PAYLOAD;
+	in.u.allocate_payload.number_sdp_streams = 3;
+	for (i = 0; i < in.u.allocate_payload.number_sdp_streams; i++)
+		in.u.allocate_payload.sdp_stream_sink[i] = i + 1;
+	DO_TEST();
+	in.u.allocate_payload.port_number = 0xf;
+	DO_TEST();
+	in.u.allocate_payload.vcpi = 0x7f;
+	DO_TEST();
+	in.u.allocate_payload.pbn = U16_MAX;
+	DO_TEST();
+
+	in.req_type = DP_QUERY_PAYLOAD;
+	in.u.query_payload.port_number = 0xf;
+	DO_TEST();
+	in.u.query_payload.vcpi = 0x7f;
+	DO_TEST();
+
+	in.req_type = DP_REMOTE_DPCD_READ;
+	in.u.dpcd_read.port_number = 0xf;
+	DO_TEST();
+	in.u.dpcd_read.dpcd_address = 0xfedcb;
+	DO_TEST();
+	in.u.dpcd_read.num_bytes = U8_MAX;
+	DO_TEST();
+
+	in.req_type = DP_REMOTE_DPCD_WRITE;
+	in.u.dpcd_write.port_number = 0xf;
+	DO_TEST();
+	in.u.dpcd_write.dpcd_address = 0xfedcb;
+	DO_TEST();
+	in.u.dpcd_write.num_bytes = ARRAY_SIZE(data);
+	in.u.dpcd_write.bytes = data;
+	DO_TEST();
+
+	in.req_type = DP_REMOTE_I2C_READ;
+	in.u.i2c_read.port_number = 0xf;
+	DO_TEST();
+	in.u.i2c_read.read_i2c_device_id = 0x7f;
+	DO_TEST();
+	in.u.i2c_read.num_transactions = 3;
+	in.u.i2c_read.num_bytes_read = ARRAY_SIZE(data) * 3;
+	for (i = 0; i < in.u.i2c_read.num_transactions; i++) {
+		in.u.i2c_read.transactions[i].bytes = data;
+		in.u.i2c_read.transactions[i].num_bytes = ARRAY_SIZE(data);
+		in.u.i2c_read.transactions[i].i2c_dev_id = 0x7f & ~i;
+		in.u.i2c_read.transactions[i].i2c_transaction_delay = 0xf & ~i;
+	}
+	DO_TEST();
+
+	in.req_type = DP_REMOTE_I2C_WRITE;
+	in.u.i2c_write.port_number = 0xf;
+	DO_TEST();
+	in.u.i2c_write.write_i2c_device_id = 0x7f;
+	DO_TEST();
+	in.u.i2c_write.num_bytes = ARRAY_SIZE(data);
+	in.u.i2c_write.bytes = data;
+	DO_TEST();
+
+#undef DO_TEST
+	return 0;
+}
diff --git a/drivers/gpu/drm/selftests/test-drm_modeset_common.h b/drivers/gpu/drm/selftests/test-drm_modeset_common.h
index 590bda35a683..0fcb8bbc6a1b 100644
--- a/drivers/gpu/drm/selftests/test-drm_modeset_common.h
+++ b/drivers/gpu/drm/selftests/test-drm_modeset_common.h
@@ -40,5 +40,6 @@ int igt_damage_iter_damage_one_outside(void *ignored);
 int igt_damage_iter_damage_src_moved(void *ignored);
 int igt_damage_iter_damage_not_visible(void *ignored);
 int igt_dp_mst_calc_pbn_mode(void *ignored);
+int igt_dp_mst_sideband_msg_req_decode(void *ignored);
 
 #endif
diff --git a/include/drm/drm_dp_mst_helper.h b/include/drm/drm_dp_mst_helper.h
index c01f3ea72756..4c8d177e83e5 100644
--- a/include/drm/drm_dp_mst_helper.h
+++ b/include/drm/drm_dp_mst_helper.h
@@ -293,7 +293,7 @@ struct drm_dp_remote_dpcd_write {
 struct drm_dp_remote_i2c_read {
 	u8 num_transactions;
 	u8 port_number;
-	struct {
+	struct drm_dp_remote_i2c_read_tx {
 		u8 i2c_dev_id;
 		u8 num_bytes;
 		u8 *bytes;
-- 
2.21.0

Re: [PATCH 05/26] drm/dp_mst: Add sideband down request tracing + selftests

[Daniel Vetter]

On Wed, Jul 17, 2019 at 09:42:28PM -0400, Lyude Paul wrote:
> Unfortunately the DP MST helpers do not have much in the way of
> debugging utilities. So, let's add some!
> 
> This adds basic debugging output for down sideband requests that we send
> from the driver, so that we can actually discern what's happening when
> sideband requests timeout. Note that with this commit, we'll be dumping
> out sideband requests under both of the following conditions:
> 
> - When the user has enabled DRM_UT_DP output, of course.
> - When the user has enabled DRM_UT_KMS or DRM_UT_DP, and a sideband
>   request has failed in some way. This will allow for developers to get
>   a basic idea of what's actually happening with failed modesets on MST,
>   without needing to have DRM_UT_DP explicitly enabled.
> 
> Since there wasn't really a good way of testing that any of this worked,
> I ended up writing simple selftests that lightly test sideband message
> encoding and decoding as well. Enjoy!
> 
> Cc: Juston Li <juston.li@intel.com>
> Cc: Imre Deak <imre.deak@intel.com>
> Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
> Cc: Harry Wentland <hwentlan@amd.com>
> Signed-off-by: Lyude Paul <lyude@redhat.com>
> ---
>  drivers/gpu/drm/drm_dp_mst_topology.c         | 308 +++++++++++++++++-
>  .../gpu/drm/drm_dp_mst_topology_internal.h    |  24 ++
>  .../gpu/drm/selftests/drm_modeset_selftests.h |   1 +
>  .../drm/selftests/test-drm_dp_mst_helper.c    | 212 ++++++++++++
>  .../drm/selftests/test-drm_modeset_common.h   |   1 +
>  include/drm/drm_dp_mst_helper.h               |   2 +-
>  6 files changed, 543 insertions(+), 5 deletions(-)
>  create mode 100644 drivers/gpu/drm/drm_dp_mst_topology_internal.h
> 
> diff --git a/drivers/gpu/drm/drm_dp_mst_topology.c b/drivers/gpu/drm/drm_dp_mst_topology.c
> index 9e382117896d..defc5e09fb9a 100644
> --- a/drivers/gpu/drm/drm_dp_mst_topology.c
> +++ b/drivers/gpu/drm/drm_dp_mst_topology.c
> @@ -36,6 +36,8 @@
>  #include <drm/drm_print.h>
>  #include <drm/drm_probe_helper.h>
>  
> +#include "drm_dp_mst_topology_internal.h"
> +
>  /**
>   * DOC: dp mst helper
>   *
> @@ -68,6 +70,8 @@ static int drm_dp_mst_register_i2c_bus(struct drm_dp_aux *aux);
>  static void drm_dp_mst_unregister_i2c_bus(struct drm_dp_aux *aux);
>  static void drm_dp_mst_kick_tx(struct drm_dp_mst_topology_mgr *mgr);
>  
> +#define DBG_PREFIX "[dp_mst]"
> +
>  #define DP_STR(x) [DP_ ## x] = #x
>  
>  static const char *drm_dp_mst_req_type_str(u8 req_type)
> @@ -124,6 +128,43 @@ static const char *drm_dp_mst_nak_reason_str(u8 nak_reason)
>  }
>  
>  #undef DP_STR
> +#define DP_STR(x) [DRM_DP_SIDEBAND_TX_ ## x] = #x
> +
> +static const char *drm_dp_mst_sideband_tx_state_str(int state)
> +{
> +	static const char * const sideband_reason_str[] = {
> +		DP_STR(QUEUED),
> +		DP_STR(START_SEND),
> +		DP_STR(SENT),
> +		DP_STR(RX),
> +		DP_STR(TIMEOUT),
> +	};
> +
> +	if (state >= ARRAY_SIZE(sideband_reason_str) ||
> +	    !sideband_reason_str[state])
> +		return "unknown";
> +
> +	return sideband_reason_str[state];
> +}
> +
> +static int
> +drm_dp_mst_rad_to_str(const u8 rad[8], u8 lct, char *out, size_t len)
> +{
> +	int i;
> +	u8 unpacked_rad[16];
> +
> +	for (i = 0; i < lct; i++) {
> +		if (i % 2)
> +			unpacked_rad[i] = rad[i / 2] >> 4;
> +		else
> +			unpacked_rad[i] = rad[i / 2] & BIT_MASK(4);
> +	}
> +
> +	/* TODO: Eventually add something to printk so we can format the rad
> +	 * like this: 1.2.3
> +	 */

	lct *=2;

missing here? And yeah the todo would be sweet, but quite a bit more
typing I guess.

> +	return snprintf(out, len, "%*phC", lct, unpacked_rad);
> +}
>  
>  /* sideband msg handling */
>  static u8 drm_dp_msg_header_crc4(const uint8_t *data, size_t num_nibbles)
> @@ -256,8 +297,9 @@ static bool drm_dp_decode_sideband_msg_hdr(struct drm_dp_sideband_msg_hdr *hdr,
>  	return true;
>  }
>  
> -static void drm_dp_encode_sideband_req(struct drm_dp_sideband_msg_req_body *req,
> -				       struct drm_dp_sideband_msg_tx *raw)
> +void
> +drm_dp_encode_sideband_req(const struct drm_dp_sideband_msg_req_body *req,
> +			   struct drm_dp_sideband_msg_tx *raw)
>  {
>  	int idx = 0;
>  	int i;
> @@ -362,6 +404,249 @@ static void drm_dp_encode_sideband_req(struct drm_dp_sideband_msg_req_body *req,
>  	}
>  	raw->cur_len = idx;
>  }
> +EXPORT_SYMBOL_FOR_TESTS_ONLY(drm_dp_encode_sideband_req);
> +
> +/* Decode a sideband request we've encoded, mainly used for debugging */
> +int
> +drm_dp_decode_sideband_req(const struct drm_dp_sideband_msg_tx *raw,
> +			   struct drm_dp_sideband_msg_req_body *req)
> +{
> +	const u8 *buf = raw->msg;
> +	int i, idx = 0;
> +
> +	req->req_type = buf[idx++] & 0x7f;
> +	switch (req->req_type) {
> +	case DP_ENUM_PATH_RESOURCES:
> +	case DP_POWER_DOWN_PHY:
> +	case DP_POWER_UP_PHY:

OCD warning: the enum isn't ordered the same way here and in
drm_dp_encode_sideband_req().

> +		req->u.port_num.port_number = (buf[idx] >> 4) & 0xf;
> +		break;
> +	case DP_ALLOCATE_PAYLOAD:
> +		{
> +			struct drm_dp_allocate_payload *a =
> +				&req->u.allocate_payload;
> +
> +			a->number_sdp_streams = buf[idx] & 0xf;
> +			a->port_number = (buf[idx] >> 4) & 0xf;
> +
> +			a->vcpi = buf[++idx] & 0x7f;

Complain here if 0x80 is set?

> +
> +			a->pbn = buf[++idx] << 8;
> +			a->pbn |= buf[++idx];
> +
> +			idx++;
> +			for (i = 0; i < a->number_sdp_streams; i++) {
> +				a->sdp_stream_sink[i] =
> +					(buf[idx + (i / 2)] >> ((i % 2) ? 0 : 4)) & 0xf;
> +			}
> +		}
> +		break;
> +	case DP_QUERY_PAYLOAD:
> +		req->u.query_payload.port_number = (buf[idx] >> 4) & 0xf;
> +		req->u.query_payload.vcpi = buf[++idx] & 0x7f;

Same here for the highest bit?

> +		break;
> +	case DP_REMOTE_DPCD_READ:
> +		{
> +			struct drm_dp_remote_dpcd_read *r = &req->u.dpcd_read;
> +
> +			r->port_number = (buf[idx] >> 4) & 0xf;
> +
> +			r->dpcd_address = (buf[idx] << 16) & 0xf0000;
> +			r->dpcd_address |= (buf[++idx] << 8) & 0xff00;
> +			r->dpcd_address |= buf[++idx] & 0xff;
> +
> +			r->num_bytes = buf[++idx];
> +		}
> +		break;
> +	case DP_REMOTE_DPCD_WRITE:
> +		{
> +			struct drm_dp_remote_dpcd_write *w =
> +				&req->u.dpcd_write;
> +
> +			w->port_number = (buf[idx] >> 4) & 0xf;
> +
> +			w->dpcd_address = (buf[idx] << 16) & 0xf0000;
> +			w->dpcd_address |= (buf[++idx] << 8) & 0xff00;
> +			w->dpcd_address |= buf[++idx] & 0xff;
> +
> +			w->num_bytes = buf[++idx];
> +
> +			w->bytes = kmemdup(&buf[++idx], w->num_bytes,
> +					   GFP_KERNEL);

But if we go really strict on validation we'd need to make sure we don't
walk past raw->cur_len ... probably not worth it?

> +			if (!w->bytes)
> +				return -ENOMEM;
> +		}
> +		break;
> +	case DP_REMOTE_I2C_READ:
> +		{
> +			struct drm_dp_remote_i2c_read *r = &req->u.i2c_read;
> +			struct drm_dp_remote_i2c_read_tx *tx;
> +			bool failed = false;
> +
> +			r->num_transactions = buf[idx] & 0x3;
> +			r->port_number = (buf[idx] >> 4) & 0xf;
> +			for (i = 0; i < r->num_transactions; i++) {
> +				tx = &r->transactions[i];
> +
> +				tx->i2c_dev_id = buf[++idx] & 0x7f;
> +				tx->num_bytes = buf[++idx];
> +				tx->bytes = kmemdup(&buf[++idx],
> +						    tx->num_bytes,
> +						    GFP_KERNEL);
> +				if (!tx->bytes) {
> +					failed = true;
> +					break;
> +				}
> +				idx += tx->num_bytes;
> +				tx->no_stop_bit = (buf[idx] >> 5) & 0x1;
> +				tx->i2c_transaction_delay = buf[idx] & 0xf;
> +			}
> +
> +			if (failed) {
> +				for (i = 0; i < r->num_transactions; i++)
> +					kfree(tx->bytes);
> +				return -ENOMEM;
> +			}
> +
> +			r->read_i2c_device_id = buf[++idx] & 0x7f;
> +			r->num_bytes_read = buf[++idx];
> +		}
> +		break;
> +	case DP_REMOTE_I2C_WRITE:
> +		{
> +			struct drm_dp_remote_i2c_write *w = &req->u.i2c_write;
> +
> +			w->port_number = (buf[idx] >> 4) & 0xf;
> +			w->write_i2c_device_id = buf[++idx] & 0x7f;
> +			w->num_bytes = buf[++idx];
> +			w->bytes = kmemdup(&buf[++idx], w->num_bytes,
> +					   GFP_KERNEL);
> +			if (!w->bytes)
> +				return -ENOMEM;
> +		}
> +		break;
> +	}

For safety: Should we validate cur_len here? Or return it, and let the
caller deal with any mismatches.

But I'm kinda leaning towards no safety here, since it's not dealing with
any replies we get from the hw/externally, which might be used for
attacking us. If we later on have the same thing but for
drm_dp_sideband_parse_reply() then a lot more fun.

> +
> +	return 0;
> +}
> +EXPORT_SYMBOL_FOR_TESTS_ONLY(drm_dp_decode_sideband_req);
> +
> +void
> +drm_dp_dump_sideband_msg_req_body(const struct drm_dp_sideband_msg_req_body *req,
> +				  int indent, struct drm_printer *printer)
> +{
> +	int i;
> +
> +#define P(f, ...) drm_printf_indent(printer, indent, f, ##__VA_ARGS__)
> +	if (req->req_type == DP_LINK_ADDRESS) {
> +		/* No contents to print */
> +		P("type=%s\n", drm_dp_mst_req_type_str(req->req_type));
> +		return;
> +	}
> +
> +	P("type=%s contents:\n", drm_dp_mst_req_type_str(req->req_type));
> +	indent++;
> +
> +	switch (req->req_type) {
> +	case DP_ENUM_PATH_RESOURCES:
> +	case DP_POWER_DOWN_PHY:
> +	case DP_POWER_UP_PHY:
> +		P("port=%d\n", req->u.port_num.port_number);
> +		break;
> +	case DP_ALLOCATE_PAYLOAD:
> +		P("port=%d vcpi=%d pbn=%d sdp_streams=%d %*ph\n",
> +		  req->u.allocate_payload.port_number,
> +		  req->u.allocate_payload.vcpi, req->u.allocate_payload.pbn,
> +		  req->u.allocate_payload.number_sdp_streams,
> +		  req->u.allocate_payload.number_sdp_streams,
> +		  req->u.allocate_payload.sdp_stream_sink);
> +		break;
> +	case DP_QUERY_PAYLOAD:
> +		P("port=%d vcpi=%d\n",
> +		  req->u.query_payload.port_number,
> +		  req->u.query_payload.vcpi);
> +		break;
> +	case DP_REMOTE_DPCD_READ:
> +		P("port=%d dpcd_addr=%05x len=%d\n",
> +		  req->u.dpcd_read.port_number, req->u.dpcd_read.dpcd_address,
> +		  req->u.dpcd_read.num_bytes);
> +		break;
> +	case DP_REMOTE_DPCD_WRITE:
> +		P("port=%d addr=%05x len=%d: %*ph\n",
> +		  req->u.dpcd_write.port_number,
> +		  req->u.dpcd_write.dpcd_address,
> +		  req->u.dpcd_write.num_bytes, req->u.dpcd_write.num_bytes,
> +		  req->u.dpcd_write.bytes);
> +		break;
> +	case DP_REMOTE_I2C_READ:
> +		P("port=%d num_tx=%d id=%d size=%d:\n",
> +		  req->u.i2c_read.port_number,
> +		  req->u.i2c_read.num_transactions,
> +		  req->u.i2c_read.read_i2c_device_id,
> +		  req->u.i2c_read.num_bytes_read);
> +
> +		indent++;
> +		for (i = 0; i < req->u.i2c_read.num_transactions; i++) {
> +			const struct drm_dp_remote_i2c_read_tx *rtx =
> +				&req->u.i2c_read.transactions[i];
> +
> +			P("%d: id=%03d size=%03d no_stop_bit=%d tx_delay=%03d: %*ph\n",
> +			  i, rtx->i2c_dev_id, rtx->num_bytes,
> +			  rtx->no_stop_bit, rtx->i2c_transaction_delay,
> +			  rtx->num_bytes, rtx->bytes);
> +		}
> +		break;
> +	case DP_REMOTE_I2C_WRITE:
> +		P("port=%d id=%d size=%d: %*ph\n",
> +		  req->u.i2c_write.port_number,
> +		  req->u.i2c_write.write_i2c_device_id,
> +		  req->u.i2c_write.num_bytes, req->u.i2c_write.num_bytes,
> +		  req->u.i2c_write.bytes);
> +		break;
> +	default:
> +		P("???\n");
> +		break;
> +	}
> +#undef P
> +}
> +EXPORT_SYMBOL_FOR_TESTS_ONLY(drm_dp_dump_sideband_msg_req_body);
> +
> +static inline void
> +drm_dp_mst_dump_sideband_msg_tx(struct drm_printer *p,
> +				const struct drm_dp_sideband_msg_tx *txmsg)
> +{
> +	struct drm_dp_sideband_msg_req_body req;
> +	char buf[64];
> +	int ret;
> +	int i;
> +
> +	drm_dp_mst_rad_to_str(txmsg->dst->rad, txmsg->dst->lct, buf,
> +			      sizeof(buf));
> +	drm_printf(p, "txmsg cur_offset=%x cur_len=%x seqno=%x state=%s path_msg=%d dst=%s\n",
> +		   txmsg->cur_offset, txmsg->cur_len, txmsg->seqno,
> +		   drm_dp_mst_sideband_tx_state_str(txmsg->state),
> +		   txmsg->path_msg, buf);
> +
> +	ret = drm_dp_decode_sideband_req(txmsg, &req);
> +	if (ret) {
> +		drm_printf(p, "<failed to decode sideband req: %d>\n", ret);
> +		return;
> +	}
> +	drm_dp_dump_sideband_msg_req_body(&req, 1, p);
> +
> +	switch (req.req_type) {
> +	case DP_REMOTE_DPCD_WRITE:
> +		kfree(req.u.dpcd_write.bytes);
> +		break;
> +	case DP_REMOTE_I2C_READ:
> +		for (i = 0; i < req.u.i2c_read.num_transactions; i++)
> +			kfree(req.u.i2c_read.transactions[i].bytes);
> +		break;
> +	case DP_REMOTE_I2C_WRITE:
> +		kfree(req.u.i2c_write.bytes);
> +		break;
> +	}
> +}
>  
>  static void drm_dp_crc_sideband_chunk_req(u8 *msg, u8 len)
>  {
> @@ -893,6 +1178,11 @@ static int drm_dp_mst_wait_tx_reply(struct drm_dp_mst_branch *mstb,
>  		}
>  	}
>  out:
> +	if (unlikely(ret == -EIO && drm_debug & (DRM_UT_DP | DRM_UT_KMS))) {

Hm I'd only check for DRM_UT_DP here.

> +		struct drm_printer p = drm_debug_printer(DBG_PREFIX);
> +
> +		drm_dp_mst_dump_sideband_msg_tx(&p, txmsg);
> +	}
>  	mutex_unlock(&mgr->qlock);
>  
>  	return ret;
> @@ -1927,8 +2217,11 @@ static int process_single_tx_qlock(struct drm_dp_mst_topology_mgr *mgr,
>  	idx += tosend + 1;
>  
>  	ret = drm_dp_send_sideband_msg(mgr, up, chunk, idx);
> -	if (ret) {
> -		DRM_DEBUG_KMS("sideband msg failed to send\n");
> +	if (ret && unlikely(drm_debug & (DRM_UT_DP | DRM_UT_KMS))) {

Same.

> +		struct drm_printer p = drm_debug_printer(DBG_PREFIX);
> +
> +		drm_printf(&p, "sideband msg failed to send\n");
> +		drm_dp_mst_dump_sideband_msg_tx(&p, txmsg);
>  		return ret;
>  	}
>  
> @@ -1990,6 +2283,13 @@ static void drm_dp_queue_down_tx(struct drm_dp_mst_topology_mgr *mgr,
>  {
>  	mutex_lock(&mgr->qlock);
>  	list_add_tail(&txmsg->next, &mgr->tx_msg_downq);
> +
> +	if (unlikely(drm_debug & DRM_UT_DP)) {

Like you do here.

> +		struct drm_printer p = drm_debug_printer(DBG_PREFIX);
> +
> +		drm_dp_mst_dump_sideband_msg_tx(&p, txmsg);
> +	}
> +
>  	if (list_is_singular(&mgr->tx_msg_downq))
>  		process_single_down_tx_qlock(mgr);
>  	mutex_unlock(&mgr->qlock);
> diff --git a/drivers/gpu/drm/drm_dp_mst_topology_internal.h b/drivers/gpu/drm/drm_dp_mst_topology_internal.h
> new file mode 100644
> index 000000000000..eeda9a61c657
> --- /dev/null
> +++ b/drivers/gpu/drm/drm_dp_mst_topology_internal.h
> @@ -0,0 +1,24 @@
> +/* SPDX-License-Identifier: GPL-2.0-only
> + *
> + * Declarations for DP MST related functions which are only used in selftests
> + *
> + * Copyright © 2018 Red Hat
> + * Authors:
> + *     Lyude Paul <lyude@redhat.com>
> + */
> +
> +#ifndef _DRM_DP_MST_HELPER_INTERNAL_H_
> +#define _DRM_DP_MST_HELPER_INTERNAL_H_
> +
> +#include <drm/drm_dp_mst_helper.h>
> +
> +void
> +drm_dp_encode_sideband_req(const struct drm_dp_sideband_msg_req_body *req,
> +			   struct drm_dp_sideband_msg_tx *raw);
> +int drm_dp_decode_sideband_req(const struct drm_dp_sideband_msg_tx *raw,
> +			       struct drm_dp_sideband_msg_req_body *req);
> +void
> +drm_dp_dump_sideband_msg_req_body(const struct drm_dp_sideband_msg_req_body *req,
> +				  int indent, struct drm_printer *printer);
> +
> +#endif /* !_DRM_DP_MST_HELPER_INTERNAL_H_ */
> diff --git a/drivers/gpu/drm/selftests/drm_modeset_selftests.h b/drivers/gpu/drm/selftests/drm_modeset_selftests.h
> index dec3ee3ec96f..1898de0b4a4d 100644
> --- a/drivers/gpu/drm/selftests/drm_modeset_selftests.h
> +++ b/drivers/gpu/drm/selftests/drm_modeset_selftests.h
> @@ -33,3 +33,4 @@ selftest(damage_iter_damage_one_outside, igt_damage_iter_damage_one_outside)
>  selftest(damage_iter_damage_src_moved, igt_damage_iter_damage_src_moved)
>  selftest(damage_iter_damage_not_visible, igt_damage_iter_damage_not_visible)
>  selftest(dp_mst_calc_pbn_mode, igt_dp_mst_calc_pbn_mode)
> +selftest(dp_mst_sideband_msg_req_decode, igt_dp_mst_sideband_msg_req_decode)
> diff --git a/drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c b/drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c
> index 51b2486ec917..ceca89babd65 100644
> --- a/drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c
> +++ b/drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c
> @@ -8,6 +8,7 @@
>  #include <drm/drm_dp_mst_helper.h>
>  #include <drm/drm_print.h>
>  
> +#include "../drm_dp_mst_topology_internal.h"
>  #include "test-drm_modeset_common.h"
>  
>  int igt_dp_mst_calc_pbn_mode(void *ignored)
> @@ -34,3 +35,214 @@ int igt_dp_mst_calc_pbn_mode(void *ignored)
>  
>  	return 0;
>  }
> +
> +static bool
> +sideband_msg_req_equal(const struct drm_dp_sideband_msg_req_body *in,
> +		       const struct drm_dp_sideband_msg_req_body *out)
> +{
> +	const struct drm_dp_remote_i2c_read_tx *txin, *txout;
> +	int i;
> +
> +	if (in->req_type != out->req_type)
> +		return false;
> +
> +	switch (in->req_type) {
> +	case DP_ENUM_PATH_RESOURCES:
> +	case DP_POWER_UP_PHY:
> +	case DP_POWER_DOWN_PHY:
> +	case DP_ALLOCATE_PAYLOAD:
> +	case DP_QUERY_PAYLOAD:
> +	case DP_REMOTE_DPCD_READ:
> +		return memcmp(in, out, sizeof(*in)) == 0;

Just memcmp the entire thing for everyrone (we assume kzalloc anyway), and
then only have the additional checks for the few cases we need it? Would
slightly reduce the code.

> +
> +	case DP_REMOTE_I2C_READ:
> +#define IN in->u.i2c_read
> +#define OUT out->u.i2c_read
> +		if (IN.num_bytes_read != OUT.num_bytes_read ||
> +		    IN.num_transactions != OUT.num_transactions ||
> +		    IN.port_number != OUT.port_number ||
> +		    IN.read_i2c_device_id != OUT.read_i2c_device_id)
> +			return false;
> +
> +		for (i = 0; i < IN.num_transactions; i++) {
> +			txin = &IN.transactions[i];
> +			txout = &OUT.transactions[i];
> +
> +			if (txin->i2c_dev_id != txout->i2c_dev_id ||
> +			    txin->no_stop_bit != txout->no_stop_bit ||
> +			    txin->num_bytes != txout->num_bytes ||
> +			    txin->i2c_transaction_delay !=
> +			    txout->i2c_transaction_delay)
> +				return false;
> +
> +			if (memcmp(txin->bytes, txout->bytes,
> +				   txin->num_bytes) != 0)
> +				return false;
> +		}
> +		break;
> +#undef IN
> +#undef OUT
> +
> +	case DP_REMOTE_DPCD_WRITE:
> +#define IN in->u.dpcd_write
> +#define OUT out->u.dpcd_write
> +		if (IN.dpcd_address != OUT.dpcd_address ||
> +		    IN.num_bytes != OUT.num_bytes ||
> +		    IN.port_number != OUT.port_number)
> +			return false;
> +
> +		return memcmp(IN.bytes, OUT.bytes, IN.num_bytes) == 0;
> +#undef IN
> +#undef OUT
> +
> +	case DP_REMOTE_I2C_WRITE:
> +#define IN in->u.i2c_write
> +#define OUT out->u.i2c_write
> +		if (IN.port_number != OUT.port_number ||
> +		    IN.write_i2c_device_id != OUT.write_i2c_device_id ||
> +		    IN.num_bytes != OUT.num_bytes)
> +			return false;
> +
> +		return memcmp(IN.bytes, OUT.bytes, IN.num_bytes) == 0;
> +#undef IN
> +#undef OUT
> +	}
> +
> +	return true;
> +}
> +
> +static int
> +__sideband_msg_req_encode_decode(int line,
> +				 struct drm_dp_sideband_msg_req_body *in,
> +				 struct drm_dp_sideband_msg_req_body *out)
> +{
> +	struct drm_printer p = drm_err_printer(pr_fmt());
> +	struct drm_dp_sideband_msg_tx txmsg;
> +	int i, ret;
> +	bool eq;
> +
> +	drm_dp_encode_sideband_req(in, &txmsg);
> +	ret = drm_dp_decode_sideband_req(&txmsg, out);
> +	if (ret < 0) {
> +		pr_err(pr_fmt("Failed to decode sideband request @ line %d: %d\n"),
> +		       line, ret);
> +		return ret;
> +	}
> +
> +	eq = sideband_msg_req_equal(in, out);
> +	if (!eq) {
> +		pr_err(pr_fmt("Encode/decode @ line %d failed, expected:\n"),
> +		       line);
> +		drm_dp_dump_sideband_msg_req_body(in, 1, &p);
> +		pr_err(pr_fmt("Got:\n"));
> +		drm_dp_dump_sideband_msg_req_body(out, 1, &p);
> +	}
> +
> +	switch (in->req_type) {
> +	case DP_REMOTE_DPCD_WRITE:
> +		kfree(out->u.dpcd_write.bytes);
> +		break;
> +	case DP_REMOTE_I2C_READ:
> +		for (i = 0; i < out->u.i2c_read.num_transactions; i++)
> +			kfree(out->u.i2c_read.transactions[i].bytes);
> +		break;
> +	case DP_REMOTE_I2C_WRITE:
> +		kfree(out->u.i2c_write.bytes);
> +		break;
> +	}
> +
> +	/* Clear everything but the req_type for the input */
> +	memset(&in->u, 0, sizeof(in->u));
> +
> +	/* Clear the output entirely */
> +	memset(out, 0, sizeof(*out));
> +
> +	return eq ? 0 : -EINVAL;
> +}
> +
> +int igt_dp_mst_sideband_msg_req_decode(void *unused)
> +{
> +	struct drm_dp_sideband_msg_req_body in = { 0 }, out = { 0 };
> +	u8 data[] = { 0xff, 0x0, 0xdd };
> +	int i;
> +
> +#define DO_TEST()                                                          \
> +	do {                                                               \
> +		if (__sideband_msg_req_encode_decode(__LINE__, &in, &out)) \
> +			return -EINVAL;                                    \
> +	} while (0)

I had a tiny wtf moment here until I realized this is a macro ... maybe
put the do on the first line and indent everything? Pure bikeshed to help
the blind ...

> +
> +	in.req_type = DP_ENUM_PATH_RESOURCES;
> +	in.u.port_num.port_number = 5;
> +	DO_TEST();
> +
> +	in.req_type = DP_POWER_UP_PHY;
> +	in.u.port_num.port_number = 5;
> +	DO_TEST();
> +
> +	in.req_type = DP_POWER_DOWN_PHY;
> +	in.u.port_num.port_number = 5;
> +	DO_TEST();
> +
> +	in.req_type = DP_ALLOCATE_PAYLOAD;
> +	in.u.allocate_payload.number_sdp_streams = 3;
> +	for (i = 0; i < in.u.allocate_payload.number_sdp_streams; i++)
> +		in.u.allocate_payload.sdp_stream_sink[i] = i + 1;
> +	DO_TEST();
> +	in.u.allocate_payload.port_number = 0xf;
> +	DO_TEST();
> +	in.u.allocate_payload.vcpi = 0x7f;
> +	DO_TEST();
> +	in.u.allocate_payload.pbn = U16_MAX;
> +	DO_TEST();
> +
> +	in.req_type = DP_QUERY_PAYLOAD;
> +	in.u.query_payload.port_number = 0xf;
> +	DO_TEST();
> +	in.u.query_payload.vcpi = 0x7f;
> +	DO_TEST();
> +
> +	in.req_type = DP_REMOTE_DPCD_READ;
> +	in.u.dpcd_read.port_number = 0xf;
> +	DO_TEST();
> +	in.u.dpcd_read.dpcd_address = 0xfedcb;
> +	DO_TEST();
> +	in.u.dpcd_read.num_bytes = U8_MAX;
> +	DO_TEST();
> +
> +	in.req_type = DP_REMOTE_DPCD_WRITE;
> +	in.u.dpcd_write.port_number = 0xf;
> +	DO_TEST();
> +	in.u.dpcd_write.dpcd_address = 0xfedcb;
> +	DO_TEST();
> +	in.u.dpcd_write.num_bytes = ARRAY_SIZE(data);
> +	in.u.dpcd_write.bytes = data;
> +	DO_TEST();
> +
> +	in.req_type = DP_REMOTE_I2C_READ;
> +	in.u.i2c_read.port_number = 0xf;
> +	DO_TEST();
> +	in.u.i2c_read.read_i2c_device_id = 0x7f;
> +	DO_TEST();
> +	in.u.i2c_read.num_transactions = 3;
> +	in.u.i2c_read.num_bytes_read = ARRAY_SIZE(data) * 3;
> +	for (i = 0; i < in.u.i2c_read.num_transactions; i++) {
> +		in.u.i2c_read.transactions[i].bytes = data;
> +		in.u.i2c_read.transactions[i].num_bytes = ARRAY_SIZE(data);
> +		in.u.i2c_read.transactions[i].i2c_dev_id = 0x7f & ~i;
> +		in.u.i2c_read.transactions[i].i2c_transaction_delay = 0xf & ~i;
> +	}
> +	DO_TEST();
> +
> +	in.req_type = DP_REMOTE_I2C_WRITE;
> +	in.u.i2c_write.port_number = 0xf;
> +	DO_TEST();
> +	in.u.i2c_write.write_i2c_device_id = 0x7f;
> +	DO_TEST();
> +	in.u.i2c_write.num_bytes = ARRAY_SIZE(data);
> +	in.u.i2c_write.bytes = data;
> +	DO_TEST();
> +
> +#undef DO_TEST
> +	return 0;
> +}

Extremely nice, more unit tests ftw!

With the nits somehow figured out: Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>

> diff --git a/drivers/gpu/drm/selftests/test-drm_modeset_common.h b/drivers/gpu/drm/selftests/test-drm_modeset_common.h
> index 590bda35a683..0fcb8bbc6a1b 100644
> --- a/drivers/gpu/drm/selftests/test-drm_modeset_common.h
> +++ b/drivers/gpu/drm/selftests/test-drm_modeset_common.h
> @@ -40,5 +40,6 @@ int igt_damage_iter_damage_one_outside(void *ignored);
>  int igt_damage_iter_damage_src_moved(void *ignored);
>  int igt_damage_iter_damage_not_visible(void *ignored);
>  int igt_dp_mst_calc_pbn_mode(void *ignored);
> +int igt_dp_mst_sideband_msg_req_decode(void *ignored);
>  
>  #endif
> diff --git a/include/drm/drm_dp_mst_helper.h b/include/drm/drm_dp_mst_helper.h
> index c01f3ea72756..4c8d177e83e5 100644
> --- a/include/drm/drm_dp_mst_helper.h
> +++ b/include/drm/drm_dp_mst_helper.h
> @@ -293,7 +293,7 @@ struct drm_dp_remote_dpcd_write {
>  struct drm_dp_remote_i2c_read {
>  	u8 num_transactions;
>  	u8 port_number;
> -	struct {
> +	struct drm_dp_remote_i2c_read_tx {
>  		u8 i2c_dev_id;
>  		u8 num_bytes;
>  		u8 *bytes;
> -- 
> 2.21.0
> 

-- 
Daniel Vetter
Software Engineer, Intel Corporation
http://blog.ffwll.ch

Re: [PATCH 05/26] drm/dp_mst: Add sideband down request tracing + selftests

[Lyude Paul]

On Tue, 2019-08-13 at 16:50 +0200, Daniel Vetter wrote:
> On Wed, Jul 17, 2019 at 09:42:28PM -0400, Lyude Paul wrote:
> > Unfortunately the DP MST helpers do not have much in the way of
> > debugging utilities. So, let's add some!
> > 
> > This adds basic debugging output for down sideband requests that we send
> > from the driver, so that we can actually discern what's happening when
> > sideband requests timeout. Note that with this commit, we'll be dumping
> > out sideband requests under both of the following conditions:
> > 
> > - When the user has enabled DRM_UT_DP output, of course.
> > - When the user has enabled DRM_UT_KMS or DRM_UT_DP, and a sideband
> >   request has failed in some way. This will allow for developers to get
> >   a basic idea of what's actually happening with failed modesets on MST,
> >   without needing to have DRM_UT_DP explicitly enabled.
> > 
> > Since there wasn't really a good way of testing that any of this worked,
> > I ended up writing simple selftests that lightly test sideband message
> > encoding and decoding as well. Enjoy!
> > 
> > Cc: Juston Li <juston.li@intel.com>
> > Cc: Imre Deak <imre.deak@intel.com>
> > Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
> > Cc: Harry Wentland <hwentlan@amd.com>
> > Signed-off-by: Lyude Paul <lyude@redhat.com>
> > ---
> >  drivers/gpu/drm/drm_dp_mst_topology.c         | 308 +++++++++++++++++-
> >  .../gpu/drm/drm_dp_mst_topology_internal.h    |  24 ++
> >  .../gpu/drm/selftests/drm_modeset_selftests.h |   1 +
> >  .../drm/selftests/test-drm_dp_mst_helper.c    | 212 ++++++++++++
> >  .../drm/selftests/test-drm_modeset_common.h   |   1 +
> >  include/drm/drm_dp_mst_helper.h               |   2 +-
> >  6 files changed, 543 insertions(+), 5 deletions(-)
> >  create mode 100644 drivers/gpu/drm/drm_dp_mst_topology_internal.h
> > 
> > diff --git a/drivers/gpu/drm/drm_dp_mst_topology.c
> > b/drivers/gpu/drm/drm_dp_mst_topology.c
> > index 9e382117896d..defc5e09fb9a 100644
> > --- a/drivers/gpu/drm/drm_dp_mst_topology.c
> > +++ b/drivers/gpu/drm/drm_dp_mst_topology.c
> > @@ -36,6 +36,8 @@
> >  #include <drm/drm_print.h>
> >  #include <drm/drm_probe_helper.h>
> >  
> > +#include "drm_dp_mst_topology_internal.h"
> > +
> >  /**
> >   * DOC: dp mst helper
> >   *
> > @@ -68,6 +70,8 @@ static int drm_dp_mst_register_i2c_bus(struct drm_dp_aux
> > *aux);
> >  static void drm_dp_mst_unregister_i2c_bus(struct drm_dp_aux *aux);
> >  static void drm_dp_mst_kick_tx(struct drm_dp_mst_topology_mgr *mgr);
> >  
> > +#define DBG_PREFIX "[dp_mst]"
> > +
> >  #define DP_STR(x) [DP_ ## x] = #x
> >  
> >  static const char *drm_dp_mst_req_type_str(u8 req_type)
> > @@ -124,6 +128,43 @@ static const char *drm_dp_mst_nak_reason_str(u8
> > nak_reason)
> >  }
> >  
> >  #undef DP_STR
> > +#define DP_STR(x) [DRM_DP_SIDEBAND_TX_ ## x] = #x
> > +
> > +static const char *drm_dp_mst_sideband_tx_state_str(int state)
> > +{
> > +	static const char * const sideband_reason_str[] = {
> > +		DP_STR(QUEUED),
> > +		DP_STR(START_SEND),
> > +		DP_STR(SENT),
> > +		DP_STR(RX),
> > +		DP_STR(TIMEOUT),
> > +	};
> > +
> > +	if (state >= ARRAY_SIZE(sideband_reason_str) ||
> > +	    !sideband_reason_str[state])
> > +		return "unknown";
> > +
> > +	return sideband_reason_str[state];
> > +}
> > +
> > +static int
> > +drm_dp_mst_rad_to_str(const u8 rad[8], u8 lct, char *out, size_t len)
> > +{
> > +	int i;
> > +	u8 unpacked_rad[16];
> > +
> > +	for (i = 0; i < lct; i++) {
> > +		if (i % 2)
> > +			unpacked_rad[i] = rad[i / 2] >> 4;
> > +		else
> > +			unpacked_rad[i] = rad[i / 2] & BIT_MASK(4);
> > +	}
> > +
> > +	/* TODO: Eventually add something to printk so we can format the rad
> > +	 * like this: 1.2.3
> > +	 */
> 
> 	lct *=2;
> 
> missing here? And yeah the todo would be sweet, but quite a bit more
> typing I guess.

Nope-although I see why one might assume that. lct == Link Count Total, e.g.
how many nodes we have to hop until we reach our destination. Each segment of
the RAD is packed as 4 bits, but the LCT is just stored as a plain value.
> 
> > +	return snprintf(out, len, "%*phC", lct, unpacked_rad);
> > +}
> >  
> >  /* sideband msg handling */
> >  static u8 drm_dp_msg_header_crc4(const uint8_t *data, size_t num_nibbles)
> > @@ -256,8 +297,9 @@ static bool drm_dp_decode_sideband_msg_hdr(struct
> > drm_dp_sideband_msg_hdr *hdr,
> >  	return true;
> >  }
> >  
> > -static void drm_dp_encode_sideband_req(struct
> > drm_dp_sideband_msg_req_body *req,
> > -				       struct drm_dp_sideband_msg_tx *raw)
> > +void
> > +drm_dp_encode_sideband_req(const struct drm_dp_sideband_msg_req_body
> > *req,
> > +			   struct drm_dp_sideband_msg_tx *raw)
> >  {
> >  	int idx = 0;
> >  	int i;
> > @@ -362,6 +404,249 @@ static void drm_dp_encode_sideband_req(struct
> > drm_dp_sideband_msg_req_body *req,
> >  	}
> >  	raw->cur_len = idx;
> >  }
> > +EXPORT_SYMBOL_FOR_TESTS_ONLY(drm_dp_encode_sideband_req);
> > +
> > +/* Decode a sideband request we've encoded, mainly used for debugging */
> > +int
> > +drm_dp_decode_sideband_req(const struct drm_dp_sideband_msg_tx *raw,
> > +			   struct drm_dp_sideband_msg_req_body *req)
> > +{
> > +	const u8 *buf = raw->msg;
> > +	int i, idx = 0;
> > +
> > +	req->req_type = buf[idx++] & 0x7f;
> > +	switch (req->req_type) {
> > +	case DP_ENUM_PATH_RESOURCES:
> > +	case DP_POWER_DOWN_PHY:
> > +	case DP_POWER_UP_PHY:
> 
> OCD warning: the enum isn't ordered the same way here and in
> drm_dp_encode_sideband_req().
> 
> > +		req->u.port_num.port_number = (buf[idx] >> 4) & 0xf;
> > +		break;
> > +	case DP_ALLOCATE_PAYLOAD:
> > +		{
> > +			struct drm_dp_allocate_payload *a =
> > +				&req->u.allocate_payload;
> > +
> > +			a->number_sdp_streams = buf[idx] & 0xf;
> > +			a->port_number = (buf[idx] >> 4) & 0xf;
> > +
> > +			a->vcpi = buf[++idx] & 0x7f;
> 
> Complain here if 0x80 is set?
> 
> > +
> > +			a->pbn = buf[++idx] << 8;
> > +			a->pbn |= buf[++idx];
> > +
> > +			idx++;
> > +			for (i = 0; i < a->number_sdp_streams; i++) {
> > +				a->sdp_stream_sink[i] =
> > +					(buf[idx + (i / 2)] >> ((i % 2) ? 0 :
> > 4)) & 0xf;
> > +			}
> > +		}
> > +		break;
> > +	case DP_QUERY_PAYLOAD:
> > +		req->u.query_payload.port_number = (buf[idx] >> 4) & 0xf;
> > +		req->u.query_payload.vcpi = buf[++idx] & 0x7f;
> 
> Same here for the highest bit?
> 
> > +		break;
> > +	case DP_REMOTE_DPCD_READ:
> > +		{
> > +			struct drm_dp_remote_dpcd_read *r = &req->u.dpcd_read;
> > +
> > +			r->port_number = (buf[idx] >> 4) & 0xf;
> > +
> > +			r->dpcd_address = (buf[idx] << 16) & 0xf0000;
> > +			r->dpcd_address |= (buf[++idx] << 8) & 0xff00;
> > +			r->dpcd_address |= buf[++idx] & 0xff;
> > +
> > +			r->num_bytes = buf[++idx];
> > +		}
> > +		break;
> > +	case DP_REMOTE_DPCD_WRITE:
> > +		{
> > +			struct drm_dp_remote_dpcd_write *w =
> > +				&req->u.dpcd_write;
> > +
> > +			w->port_number = (buf[idx] >> 4) & 0xf;
> > +
> > +			w->dpcd_address = (buf[idx] << 16) & 0xf0000;
> > +			w->dpcd_address |= (buf[++idx] << 8) & 0xff00;
> > +			w->dpcd_address |= buf[++idx] & 0xff;
> > +
> > +			w->num_bytes = buf[++idx];
> > +
> > +			w->bytes = kmemdup(&buf[++idx], w->num_bytes,
> > +					   GFP_KERNEL);
> 
> But if we go really strict on validation we'd need to make sure we don't
> walk past raw->cur_len ... probably not worth it?
> 
> > +			if (!w->bytes)
> > +				return -ENOMEM;
> > +		}
> > +		break;
> > +	case DP_REMOTE_I2C_READ:
> > +		{
> > +			struct drm_dp_remote_i2c_read *r = &req->u.i2c_read;
> > +			struct drm_dp_remote_i2c_read_tx *tx;
> > +			bool failed = false;
> > +
> > +			r->num_transactions = buf[idx] & 0x3;
> > +			r->port_number = (buf[idx] >> 4) & 0xf;
> > +			for (i = 0; i < r->num_transactions; i++) {
> > +				tx = &r->transactions[i];
> > +
> > +				tx->i2c_dev_id = buf[++idx] & 0x7f;
> > +				tx->num_bytes = buf[++idx];
> > +				tx->bytes = kmemdup(&buf[++idx],
> > +						    tx->num_bytes,
> > +						    GFP_KERNEL);
> > +				if (!tx->bytes) {
> > +					failed = true;
> > +					break;
> > +				}
> > +				idx += tx->num_bytes;
> > +				tx->no_stop_bit = (buf[idx] >> 5) & 0x1;
> > +				tx->i2c_transaction_delay = buf[idx] & 0xf;
> > +			}
> > +
> > +			if (failed) {
> > +				for (i = 0; i < r->num_transactions; i++)
> > +					kfree(tx->bytes);
> > +				return -ENOMEM;
> > +			}
> > +
> > +			r->read_i2c_device_id = buf[++idx] & 0x7f;
> > +			r->num_bytes_read = buf[++idx];
> > +		}
> > +		break;
> > +	case DP_REMOTE_I2C_WRITE:
> > +		{
> > +			struct drm_dp_remote_i2c_write *w = &req->u.i2c_write;
> > +
> > +			w->port_number = (buf[idx] >> 4) & 0xf;
> > +			w->write_i2c_device_id = buf[++idx] & 0x7f;
> > +			w->num_bytes = buf[++idx];
> > +			w->bytes = kmemdup(&buf[++idx], w->num_bytes,
> > +					   GFP_KERNEL);
> > +			if (!w->bytes)
> > +				return -ENOMEM;
> > +		}
> > +		break;
> > +	}
> 
> For safety: Should we validate cur_len here? Or return it, and let the
> caller deal with any mismatches.
> 
> But I'm kinda leaning towards no safety here, since it's not dealing with
> any replies we get from the hw/externally, which might be used for
> attacking us. If we later on have the same thing but for
> drm_dp_sideband_parse_reply() then a lot more fun.

Yeah-this was my thinking as well. Additionally, I should probably note the
next batch of patches I'm going to try getting in afterwards actually adds
selftests for our length checking :)

(fun fact: so far, those selftests seem to indicate we're actually doing quite
well at verifying length. Mostly.)

> 
> > +
> > +	return 0;
> > +}
> > +EXPORT_SYMBOL_FOR_TESTS_ONLY(drm_dp_decode_sideband_req);
> > +
> > +void
> > +drm_dp_dump_sideband_msg_req_body(const struct
> > drm_dp_sideband_msg_req_body *req,
> > +				  int indent, struct drm_printer *printer)
> > +{
> > +	int i;
> > +
> > +#define P(f, ...) drm_printf_indent(printer, indent, f, ##__VA_ARGS__)
> > +	if (req->req_type == DP_LINK_ADDRESS) {
> > +		/* No contents to print */
> > +		P("type=%s\n", drm_dp_mst_req_type_str(req->req_type));
> > +		return;
> > +	}
> > +
> > +	P("type=%s contents:\n", drm_dp_mst_req_type_str(req->req_type));
> > +	indent++;
> > +
> > +	switch (req->req_type) {
> > +	case DP_ENUM_PATH_RESOURCES:
> > +	case DP_POWER_DOWN_PHY:
> > +	case DP_POWER_UP_PHY:
> > +		P("port=%d\n", req->u.port_num.port_number);
> > +		break;
> > +	case DP_ALLOCATE_PAYLOAD:
> > +		P("port=%d vcpi=%d pbn=%d sdp_streams=%d %*ph\n",
> > +		  req->u.allocate_payload.port_number,
> > +		  req->u.allocate_payload.vcpi, req->u.allocate_payload.pbn,
> > +		  req->u.allocate_payload.number_sdp_streams,
> > +		  req->u.allocate_payload.number_sdp_streams,
> > +		  req->u.allocate_payload.sdp_stream_sink);
> > +		break;
> > +	case DP_QUERY_PAYLOAD:
> > +		P("port=%d vcpi=%d\n",
> > +		  req->u.query_payload.port_number,
> > +		  req->u.query_payload.vcpi);
> > +		break;
> > +	case DP_REMOTE_DPCD_READ:
> > +		P("port=%d dpcd_addr=%05x len=%d\n",
> > +		  req->u.dpcd_read.port_number, req->u.dpcd_read.dpcd_address,
> > +		  req->u.dpcd_read.num_bytes);
> > +		break;
> > +	case DP_REMOTE_DPCD_WRITE:
> > +		P("port=%d addr=%05x len=%d: %*ph\n",
> > +		  req->u.dpcd_write.port_number,
> > +		  req->u.dpcd_write.dpcd_address,
> > +		  req->u.dpcd_write.num_bytes, req->u.dpcd_write.num_bytes,
> > +		  req->u.dpcd_write.bytes);
> > +		break;
> > +	case DP_REMOTE_I2C_READ:
> > +		P("port=%d num_tx=%d id=%d size=%d:\n",
> > +		  req->u.i2c_read.port_number,
> > +		  req->u.i2c_read.num_transactions,
> > +		  req->u.i2c_read.read_i2c_device_id,
> > +		  req->u.i2c_read.num_bytes_read);
> > +
> > +		indent++;
> > +		for (i = 0; i < req->u.i2c_read.num_transactions; i++) {
> > +			const struct drm_dp_remote_i2c_read_tx *rtx =
> > +				&req->u.i2c_read.transactions[i];
> > +
> > +			P("%d: id=%03d size=%03d no_stop_bit=%d tx_delay=%03d:
> > %*ph\n",
> > +			  i, rtx->i2c_dev_id, rtx->num_bytes,
> > +			  rtx->no_stop_bit, rtx->i2c_transaction_delay,
> > +			  rtx->num_bytes, rtx->bytes);
> > +		}
> > +		break;
> > +	case DP_REMOTE_I2C_WRITE:
> > +		P("port=%d id=%d size=%d: %*ph\n",
> > +		  req->u.i2c_write.port_number,
> > +		  req->u.i2c_write.write_i2c_device_id,
> > +		  req->u.i2c_write.num_bytes, req->u.i2c_write.num_bytes,
> > +		  req->u.i2c_write.bytes);
> > +		break;
> > +	default:
> > +		P("???\n");
> > +		break;
> > +	}
> > +#undef P
> > +}
> > +EXPORT_SYMBOL_FOR_TESTS_ONLY(drm_dp_dump_sideband_msg_req_body);
> > +
> > +static inline void
> > +drm_dp_mst_dump_sideband_msg_tx(struct drm_printer *p,
> > +				const struct drm_dp_sideband_msg_tx *txmsg)
> > +{
> > +	struct drm_dp_sideband_msg_req_body req;
> > +	char buf[64];
> > +	int ret;
> > +	int i;
> > +
> > +	drm_dp_mst_rad_to_str(txmsg->dst->rad, txmsg->dst->lct, buf,
> > +			      sizeof(buf));
> > +	drm_printf(p, "txmsg cur_offset=%x cur_len=%x seqno=%x state=%s
> > path_msg=%d dst=%s\n",
> > +		   txmsg->cur_offset, txmsg->cur_len, txmsg->seqno,
> > +		   drm_dp_mst_sideband_tx_state_str(txmsg->state),
> > +		   txmsg->path_msg, buf);
> > +
> > +	ret = drm_dp_decode_sideband_req(txmsg, &req);
> > +	if (ret) {
> > +		drm_printf(p, "<failed to decode sideband req: %d>\n", ret);
> > +		return;
> > +	}
> > +	drm_dp_dump_sideband_msg_req_body(&req, 1, p);
> > +
> > +	switch (req.req_type) {
> > +	case DP_REMOTE_DPCD_WRITE:
> > +		kfree(req.u.dpcd_write.bytes);
> > +		break;
> > +	case DP_REMOTE_I2C_READ:
> > +		for (i = 0; i < req.u.i2c_read.num_transactions; i++)
> > +			kfree(req.u.i2c_read.transactions[i].bytes);
> > +		break;
> > +	case DP_REMOTE_I2C_WRITE:
> > +		kfree(req.u.i2c_write.bytes);
> > +		break;
> > +	}
> > +}
> >  
> >  static void drm_dp_crc_sideband_chunk_req(u8 *msg, u8 len)
> >  {
> > @@ -893,6 +1178,11 @@ static int drm_dp_mst_wait_tx_reply(struct
> > drm_dp_mst_branch *mstb,
> >  		}
> >  	}
> >  out:
> > +	if (unlikely(ret == -EIO && drm_debug & (DRM_UT_DP | DRM_UT_KMS))) {
> 
> Hm I'd only check for DRM_UT_DP here.
> 
> > +		struct drm_printer p = drm_debug_printer(DBG_PREFIX);
> > +
> > +		drm_dp_mst_dump_sideband_msg_tx(&p, txmsg);
> > +	}
> >  	mutex_unlock(&mgr->qlock);
> >  
> >  	return ret;
> > @@ -1927,8 +2217,11 @@ static int process_single_tx_qlock(struct
> > drm_dp_mst_topology_mgr *mgr,
> >  	idx += tosend + 1;
> >  
> >  	ret = drm_dp_send_sideband_msg(mgr, up, chunk, idx);
> > -	if (ret) {
> > -		DRM_DEBUG_KMS("sideband msg failed to send\n");
> > +	if (ret && unlikely(drm_debug & (DRM_UT_DP | DRM_UT_KMS))) {
> 
> Same.
> 
> > +		struct drm_printer p = drm_debug_printer(DBG_PREFIX);
> > +
> > +		drm_printf(&p, "sideband msg failed to send\n");
> > +		drm_dp_mst_dump_sideband_msg_tx(&p, txmsg);
> >  		return ret;
> >  	}
> >  
> > @@ -1990,6 +2283,13 @@ static void drm_dp_queue_down_tx(struct
> > drm_dp_mst_topology_mgr *mgr,
> >  {
> >  	mutex_lock(&mgr->qlock);
> >  	list_add_tail(&txmsg->next, &mgr->tx_msg_downq);
> > +
> > +	if (unlikely(drm_debug & DRM_UT_DP)) {
> 
> Like you do here.
> 
> > +		struct drm_printer p = drm_debug_printer(DBG_PREFIX);
> > +
> > +		drm_dp_mst_dump_sideband_msg_tx(&p, txmsg);
> > +	}
> > +
> >  	if (list_is_singular(&mgr->tx_msg_downq))
> >  		process_single_down_tx_qlock(mgr);
> >  	mutex_unlock(&mgr->qlock);
> > diff --git a/drivers/gpu/drm/drm_dp_mst_topology_internal.h
> > b/drivers/gpu/drm/drm_dp_mst_topology_internal.h
> > new file mode 100644
> > index 000000000000..eeda9a61c657
> > --- /dev/null
> > +++ b/drivers/gpu/drm/drm_dp_mst_topology_internal.h
> > @@ -0,0 +1,24 @@
> > +/* SPDX-License-Identifier: GPL-2.0-only
> > + *
> > + * Declarations for DP MST related functions which are only used in
> > selftests
> > + *
> > + * Copyright © 2018 Red Hat
> > + * Authors:
> > + *     Lyude Paul <lyude@redhat.com>
> > + */
> > +
> > +#ifndef _DRM_DP_MST_HELPER_INTERNAL_H_
> > +#define _DRM_DP_MST_HELPER_INTERNAL_H_
> > +
> > +#include <drm/drm_dp_mst_helper.h>
> > +
> > +void
> > +drm_dp_encode_sideband_req(const struct drm_dp_sideband_msg_req_body
> > *req,
> > +			   struct drm_dp_sideband_msg_tx *raw);
> > +int drm_dp_decode_sideband_req(const struct drm_dp_sideband_msg_tx *raw,
> > +			       struct drm_dp_sideband_msg_req_body *req);
> > +void
> > +drm_dp_dump_sideband_msg_req_body(const struct
> > drm_dp_sideband_msg_req_body *req,
> > +				  int indent, struct drm_printer *printer);
> > +
> > +#endif /* !_DRM_DP_MST_HELPER_INTERNAL_H_ */
> > diff --git a/drivers/gpu/drm/selftests/drm_modeset_selftests.h
> > b/drivers/gpu/drm/selftests/drm_modeset_selftests.h
> > index dec3ee3ec96f..1898de0b4a4d 100644
> > --- a/drivers/gpu/drm/selftests/drm_modeset_selftests.h
> > +++ b/drivers/gpu/drm/selftests/drm_modeset_selftests.h
> > @@ -33,3 +33,4 @@ selftest(damage_iter_damage_one_outside,
> > igt_damage_iter_damage_one_outside)
> >  selftest(damage_iter_damage_src_moved, igt_damage_iter_damage_src_moved)
> >  selftest(damage_iter_damage_not_visible,
> > igt_damage_iter_damage_not_visible)
> >  selftest(dp_mst_calc_pbn_mode, igt_dp_mst_calc_pbn_mode)
> > +selftest(dp_mst_sideband_msg_req_decode,
> > igt_dp_mst_sideband_msg_req_decode)
> > diff --git a/drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c
> > b/drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c
> > index 51b2486ec917..ceca89babd65 100644
> > --- a/drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c
> > +++ b/drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c
> > @@ -8,6 +8,7 @@
> >  #include <drm/drm_dp_mst_helper.h>
> >  #include <drm/drm_print.h>
> >  
> > +#include "../drm_dp_mst_topology_internal.h"
> >  #include "test-drm_modeset_common.h"
> >  
> >  int igt_dp_mst_calc_pbn_mode(void *ignored)
> > @@ -34,3 +35,214 @@ int igt_dp_mst_calc_pbn_mode(void *ignored)
> >  
> >  	return 0;
> >  }
> > +
> > +static bool
> > +sideband_msg_req_equal(const struct drm_dp_sideband_msg_req_body *in,
> > +		       const struct drm_dp_sideband_msg_req_body *out)
> > +{
> > +	const struct drm_dp_remote_i2c_read_tx *txin, *txout;
> > +	int i;
> > +
> > +	if (in->req_type != out->req_type)
> > +		return false;
> > +
> > +	switch (in->req_type) {
> > +	case DP_ENUM_PATH_RESOURCES:
> > +	case DP_POWER_UP_PHY:
> > +	case DP_POWER_DOWN_PHY:
> > +	case DP_ALLOCATE_PAYLOAD:
> > +	case DP_QUERY_PAYLOAD:
> > +	case DP_REMOTE_DPCD_READ:
> > +		return memcmp(in, out, sizeof(*in)) == 0;
> 
> Just memcmp the entire thing for everyrone (we assume kzalloc anyway), and
> then only have the additional checks for the few cases we need it? Would
> slightly reduce the code.
> 
> > +
> > +	case DP_REMOTE_I2C_READ:
> > +#define IN in->u.i2c_read
> > +#define OUT out->u.i2c_read
> > +		if (IN.num_bytes_read != OUT.num_bytes_read ||
> > +		    IN.num_transactions != OUT.num_transactions ||
> > +		    IN.port_number != OUT.port_number ||
> > +		    IN.read_i2c_device_id != OUT.read_i2c_device_id)
> > +			return false;
> > +
> > +		for (i = 0; i < IN.num_transactions; i++) {
> > +			txin = &IN.transactions[i];
> > +			txout = &OUT.transactions[i];
> > +
> > +			if (txin->i2c_dev_id != txout->i2c_dev_id ||
> > +			    txin->no_stop_bit != txout->no_stop_bit ||
> > +			    txin->num_bytes != txout->num_bytes ||
> > +			    txin->i2c_transaction_delay !=
> > +			    txout->i2c_transaction_delay)
> > +				return false;
> > +
> > +			if (memcmp(txin->bytes, txout->bytes,
> > +				   txin->num_bytes) != 0)
> > +				return false;
> > +		}
> > +		break;
> > +#undef IN
> > +#undef OUT
> > +
> > +	case DP_REMOTE_DPCD_WRITE:
> > +#define IN in->u.dpcd_write
> > +#define OUT out->u.dpcd_write
> > +		if (IN.dpcd_address != OUT.dpcd_address ||
> > +		    IN.num_bytes != OUT.num_bytes ||
> > +		    IN.port_number != OUT.port_number)
> > +			return false;
> > +
> > +		return memcmp(IN.bytes, OUT.bytes, IN.num_bytes) == 0;
> > +#undef IN
> > +#undef OUT
> > +
> > +	case DP_REMOTE_I2C_WRITE:
> > +#define IN in->u.i2c_write
> > +#define OUT out->u.i2c_write
> > +		if (IN.port_number != OUT.port_number ||
> > +		    IN.write_i2c_device_id != OUT.write_i2c_device_id ||
> > +		    IN.num_bytes != OUT.num_bytes)
> > +			return false;
> > +
> > +		return memcmp(IN.bytes, OUT.bytes, IN.num_bytes) == 0;
> > +#undef IN
> > +#undef OUT
> > +	}
> > +
> > +	return true;
> > +}
> > +
> > +static int
> > +__sideband_msg_req_encode_decode(int line,
> > +				 struct drm_dp_sideband_msg_req_body *in,
> > +				 struct drm_dp_sideband_msg_req_body *out)
> > +{
> > +	struct drm_printer p = drm_err_printer(pr_fmt());
> > +	struct drm_dp_sideband_msg_tx txmsg;
> > +	int i, ret;
> > +	bool eq;
> > +
> > +	drm_dp_encode_sideband_req(in, &txmsg);
> > +	ret = drm_dp_decode_sideband_req(&txmsg, out);
> > +	if (ret < 0) {
> > +		pr_err(pr_fmt("Failed to decode sideband request @ line %d:
> > %d\n"),
> > +		       line, ret);
> > +		return ret;
> > +	}
> > +
> > +	eq = sideband_msg_req_equal(in, out);
> > +	if (!eq) {
> > +		pr_err(pr_fmt("Encode/decode @ line %d failed, expected:\n"),
> > +		       line);
> > +		drm_dp_dump_sideband_msg_req_body(in, 1, &p);
> > +		pr_err(pr_fmt("Got:\n"));
> > +		drm_dp_dump_sideband_msg_req_body(out, 1, &p);
> > +	}
> > +
> > +	switch (in->req_type) {
> > +	case DP_REMOTE_DPCD_WRITE:
> > +		kfree(out->u.dpcd_write.bytes);
> > +		break;
> > +	case DP_REMOTE_I2C_READ:
> > +		for (i = 0; i < out->u.i2c_read.num_transactions; i++)
> > +			kfree(out->u.i2c_read.transactions[i].bytes);
> > +		break;
> > +	case DP_REMOTE_I2C_WRITE:
> > +		kfree(out->u.i2c_write.bytes);
> > +		break;
> > +	}
> > +
> > +	/* Clear everything but the req_type for the input */
> > +	memset(&in->u, 0, sizeof(in->u));
> > +
> > +	/* Clear the output entirely */
> > +	memset(out, 0, sizeof(*out));
> > +
> > +	return eq ? 0 : -EINVAL;
> > +}
> > +
> > +int igt_dp_mst_sideband_msg_req_decode(void *unused)
> > +{
> > +	struct drm_dp_sideband_msg_req_body in = { 0 }, out = { 0 };
> > +	u8 data[] = { 0xff, 0x0, 0xdd };
> > +	int i;
> > +
> > +#define
> > DO_TEST()                                                          \
> > +	do {                                                               \
> > +		if (__sideband_msg_req_encode_decode(__LINE__, &in, &out)) \
> > +			return -EINVAL;                                    \
> > +	} while (0)
> 
> I had a tiny wtf moment here until I realized this is a macro ... maybe
> put the do on the first line and indent everything? Pure bikeshed to help
> the blind ...

Yeah this was kind of silly, in the other round of selftests I've got a
cleaner way of doing this - just define DO_TEST() like so:

/* note: I realized, we also don't need to specify an out (just store it on
the stack) */

static inline bool sideband_msg_req_encode_decode(in) {
	...
}

static void actual_test(...) {
	...
	#define DO_TEST() FAIL_ON(!sideband_msg_req_encode_decode(in))
	...
}

Will do the same thing for these selftests

> 
> > +
> > +	in.req_type = DP_ENUM_PATH_RESOURCES;
> > +	in.u.port_num.port_number = 5;
> > +	DO_TEST();
> > +
> > +	in.req_type = DP_POWER_UP_PHY;
> > +	in.u.port_num.port_number = 5;
> > +	DO_TEST();
> > +
> > +	in.req_type = DP_POWER_DOWN_PHY;
> > +	in.u.port_num.port_number = 5;
> > +	DO_TEST();
> > +
> > +	in.req_type = DP_ALLOCATE_PAYLOAD;
> > +	in.u.allocate_payload.number_sdp_streams = 3;
> > +	for (i = 0; i < in.u.allocate_payload.number_sdp_streams; i++)
> > +		in.u.allocate_payload.sdp_stream_sink[i] = i + 1;
> > +	DO_TEST();
> > +	in.u.allocate_payload.port_number = 0xf;
> > +	DO_TEST();
> > +	in.u.allocate_payload.vcpi = 0x7f;
> > +	DO_TEST();
> > +	in.u.allocate_payload.pbn = U16_MAX;
> > +	DO_TEST();
> > +
> > +	in.req_type = DP_QUERY_PAYLOAD;
> > +	in.u.query_payload.port_number = 0xf;
> > +	DO_TEST();
> > +	in.u.query_payload.vcpi = 0x7f;
> > +	DO_TEST();
> > +
> > +	in.req_type = DP_REMOTE_DPCD_READ;
> > +	in.u.dpcd_read.port_number = 0xf;
> > +	DO_TEST();
> > +	in.u.dpcd_read.dpcd_address = 0xfedcb;
> > +	DO_TEST();
> > +	in.u.dpcd_read.num_bytes = U8_MAX;
> > +	DO_TEST();
> > +
> > +	in.req_type = DP_REMOTE_DPCD_WRITE;
> > +	in.u.dpcd_write.port_number = 0xf;
> > +	DO_TEST();
> > +	in.u.dpcd_write.dpcd_address = 0xfedcb;
> > +	DO_TEST();
> > +	in.u.dpcd_write.num_bytes = ARRAY_SIZE(data);
> > +	in.u.dpcd_write.bytes = data;
> > +	DO_TEST();
> > +
> > +	in.req_type = DP_REMOTE_I2C_READ;
> > +	in.u.i2c_read.port_number = 0xf;
> > +	DO_TEST();
> > +	in.u.i2c_read.read_i2c_device_id = 0x7f;
> > +	DO_TEST();
> > +	in.u.i2c_read.num_transactions = 3;
> > +	in.u.i2c_read.num_bytes_read = ARRAY_SIZE(data) * 3;
> > +	for (i = 0; i < in.u.i2c_read.num_transactions; i++) {
> > +		in.u.i2c_read.transactions[i].bytes = data;
> > +		in.u.i2c_read.transactions[i].num_bytes = ARRAY_SIZE(data);
> > +		in.u.i2c_read.transactions[i].i2c_dev_id = 0x7f & ~i;
> > +		in.u.i2c_read.transactions[i].i2c_transaction_delay = 0xf &
> > ~i;
> > +	}
> > +	DO_TEST();
> > +
> > +	in.req_type = DP_REMOTE_I2C_WRITE;
> > +	in.u.i2c_write.port_number = 0xf;
> > +	DO_TEST();
> > +	in.u.i2c_write.write_i2c_device_id = 0x7f;
> > +	DO_TEST();
> > +	in.u.i2c_write.num_bytes = ARRAY_SIZE(data);
> > +	in.u.i2c_write.bytes = data;
> > +	DO_TEST();
> > +
> > +#undef DO_TEST
> > +	return 0;
> > +}
> 
> Extremely nice, more unit tests ftw!
> 
> With the nits somehow figured out: Reviewed-by: Daniel Vetter <
> daniel.vetter@ffwll.ch>
> 
> > diff --git a/drivers/gpu/drm/selftests/test-drm_modeset_common.h
> > b/drivers/gpu/drm/selftests/test-drm_modeset_common.h
> > index 590bda35a683..0fcb8bbc6a1b 100644
> > --- a/drivers/gpu/drm/selftests/test-drm_modeset_common.h
> > +++ b/drivers/gpu/drm/selftests/test-drm_modeset_common.h
> > @@ -40,5 +40,6 @@ int igt_damage_iter_damage_one_outside(void *ignored);
> >  int igt_damage_iter_damage_src_moved(void *ignored);
> >  int igt_damage_iter_damage_not_visible(void *ignored);
> >  int igt_dp_mst_calc_pbn_mode(void *ignored);
> > +int igt_dp_mst_sideband_msg_req_decode(void *ignored);
> >  
> >  #endif
> > diff --git a/include/drm/drm_dp_mst_helper.h
> > b/include/drm/drm_dp_mst_helper.h
> > index c01f3ea72756..4c8d177e83e5 100644
> > --- a/include/drm/drm_dp_mst_helper.h
> > +++ b/include/drm/drm_dp_mst_helper.h
> > @@ -293,7 +293,7 @@ struct drm_dp_remote_dpcd_write {
> >  struct drm_dp_remote_i2c_read {
> >  	u8 num_transactions;
> >  	u8 port_number;
> > -	struct {
> > +	struct drm_dp_remote_i2c_read_tx {
> >  		u8 i2c_dev_id;
> >  		u8 num_bytes;
> >  		u8 *bytes;
> > -- 
> > 2.21.0
> > 
-- 
Cheers,
	Lyude Paul

Re: [PATCH 05/26] drm/dp_mst: Add sideband down request tracing + selftests

[Lyude Paul]

On Tue, 2019-08-13 at 16:50 +0200, Daniel Vetter wrote:
> On Wed, Jul 17, 2019 at 09:42:28PM -0400, Lyude Paul wrote:
> > Unfortunately the DP MST helpers do not have much in the way of
> > debugging utilities. So, let's add some!
> > 
> > This adds basic debugging output for down sideband requests that we send
> > from the driver, so that we can actually discern what's happening when
> > sideband requests timeout. Note that with this commit, we'll be dumping
> > out sideband requests under both of the following conditions:
> > 
> > - When the user has enabled DRM_UT_DP output, of course.
> > - When the user has enabled DRM_UT_KMS or DRM_UT_DP, and a sideband
> >   request has failed in some way. This will allow for developers to get
> >   a basic idea of what's actually happening with failed modesets on MST,
> >   without needing to have DRM_UT_DP explicitly enabled.
> > 
> > Since there wasn't really a good way of testing that any of this worked,
> > I ended up writing simple selftests that lightly test sideband message
> > encoding and decoding as well. Enjoy!
> > 
> > Cc: Juston Li <juston.li@intel.com>
> > Cc: Imre Deak <imre.deak@intel.com>
> > Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
> > Cc: Harry Wentland <hwentlan@amd.com>
> > Signed-off-by: Lyude Paul <lyude@redhat.com>
> > ---
> >  drivers/gpu/drm/drm_dp_mst_topology.c         | 308 +++++++++++++++++-
> >  .../gpu/drm/drm_dp_mst_topology_internal.h    |  24 ++
> >  .../gpu/drm/selftests/drm_modeset_selftests.h |   1 +
> >  .../drm/selftests/test-drm_dp_mst_helper.c    | 212 ++++++++++++
> >  .../drm/selftests/test-drm_modeset_common.h   |   1 +
> >  include/drm/drm_dp_mst_helper.h               |   2 +-
> >  6 files changed, 543 insertions(+), 5 deletions(-)
> >  create mode 100644 drivers/gpu/drm/drm_dp_mst_topology_internal.h
> > 
> > diff --git a/drivers/gpu/drm/drm_dp_mst_topology.c
> > b/drivers/gpu/drm/drm_dp_mst_topology.c
> > index 9e382117896d..defc5e09fb9a 100644
> > --- a/drivers/gpu/drm/drm_dp_mst_topology.c
> > +++ b/drivers/gpu/drm/drm_dp_mst_topology.c
> > @@ -36,6 +36,8 @@
> >  #include <drm/drm_print.h>
> >  #include <drm/drm_probe_helper.h>
> >  
> > +#include "drm_dp_mst_topology_internal.h"
> > +
> >  /**
> >   * DOC: dp mst helper
> >   *
> > @@ -68,6 +70,8 @@ static int drm_dp_mst_register_i2c_bus(struct drm_dp_aux
> > *aux);
> >  static void drm_dp_mst_unregister_i2c_bus(struct drm_dp_aux *aux);
> >  static void drm_dp_mst_kick_tx(struct drm_dp_mst_topology_mgr *mgr);
> >  
> > +#define DBG_PREFIX "[dp_mst]"
> > +
> >  #define DP_STR(x) [DP_ ## x] = #x
> >  
> >  static const char *drm_dp_mst_req_type_str(u8 req_type)
> > @@ -124,6 +128,43 @@ static const char *drm_dp_mst_nak_reason_str(u8
> > nak_reason)
> >  }
> >  
> >  #undef DP_STR
> > +#define DP_STR(x) [DRM_DP_SIDEBAND_TX_ ## x] = #x
> > +
> > +static const char *drm_dp_mst_sideband_tx_state_str(int state)
> > +{
> > +	static const char * const sideband_reason_str[] = {
> > +		DP_STR(QUEUED),
> > +		DP_STR(START_SEND),
> > +		DP_STR(SENT),
> > +		DP_STR(RX),
> > +		DP_STR(TIMEOUT),
> > +	};
> > +
> > +	if (state >= ARRAY_SIZE(sideband_reason_str) ||
> > +	    !sideband_reason_str[state])
> > +		return "unknown";
> > +
> > +	return sideband_reason_str[state];
> > +}
> > +
> > +static int
> > +drm_dp_mst_rad_to_str(const u8 rad[8], u8 lct, char *out, size_t len)
> > +{
> > +	int i;
> > +	u8 unpacked_rad[16];
> > +
> > +	for (i = 0; i < lct; i++) {
> > +		if (i % 2)
> > +			unpacked_rad[i] = rad[i / 2] >> 4;
> > +		else
> > +			unpacked_rad[i] = rad[i / 2] & BIT_MASK(4);
> > +	}
> > +
> > +	/* TODO: Eventually add something to printk so we can format the rad
> > +	 * like this: 1.2.3
> > +	 */
> 
> 	lct *=2;
> 
> missing here? And yeah the todo would be sweet, but quite a bit more
> typing I guess.
> 
> > +	return snprintf(out, len, "%*phC", lct, unpacked_rad);
> > +}
> >  
> >  /* sideband msg handling */
> >  static u8 drm_dp_msg_header_crc4(const uint8_t *data, size_t num_nibbles)
> > @@ -256,8 +297,9 @@ static bool drm_dp_decode_sideband_msg_hdr(struct
> > drm_dp_sideband_msg_hdr *hdr,
> >  	return true;
> >  }
> >  
> > -static void drm_dp_encode_sideband_req(struct
> > drm_dp_sideband_msg_req_body *req,
> > -				       struct drm_dp_sideband_msg_tx *raw)
> > +void
> > +drm_dp_encode_sideband_req(const struct drm_dp_sideband_msg_req_body
> > *req,
> > +			   struct drm_dp_sideband_msg_tx *raw)
> >  {
> >  	int idx = 0;
> >  	int i;
> > @@ -362,6 +404,249 @@ static void drm_dp_encode_sideband_req(struct
> > drm_dp_sideband_msg_req_body *req,
> >  	}
> >  	raw->cur_len = idx;
> >  }
> > +EXPORT_SYMBOL_FOR_TESTS_ONLY(drm_dp_encode_sideband_req);
> > +
> > +/* Decode a sideband request we've encoded, mainly used for debugging */
> > +int
> > +drm_dp_decode_sideband_req(const struct drm_dp_sideband_msg_tx *raw,
> > +			   struct drm_dp_sideband_msg_req_body *req)
> > +{
> > +	const u8 *buf = raw->msg;
> > +	int i, idx = 0;
> > +
> > +	req->req_type = buf[idx++] & 0x7f;
> > +	switch (req->req_type) {
> > +	case DP_ENUM_PATH_RESOURCES:
> > +	case DP_POWER_DOWN_PHY:
> > +	case DP_POWER_UP_PHY:
> 
> OCD warning: the enum isn't ordered the same way here and in
> drm_dp_encode_sideband_req().
> 
> > +		req->u.port_num.port_number = (buf[idx] >> 4) & 0xf;
> > +		break;
> > +	case DP_ALLOCATE_PAYLOAD:
> > +		{
> > +			struct drm_dp_allocate_payload *a =
> > +				&req->u.allocate_payload;
> > +
> > +			a->number_sdp_streams = buf[idx] & 0xf;
> > +			a->port_number = (buf[idx] >> 4) & 0xf;
> > +
> > +			a->vcpi = buf[++idx] & 0x7f;
> 
> Complain here if 0x80 is set?
> 
> > +
> > +			a->pbn = buf[++idx] << 8;
> > +			a->pbn |= buf[++idx];
> > +
> > +			idx++;
> > +			for (i = 0; i < a->number_sdp_streams; i++) {
> > +				a->sdp_stream_sink[i] =
> > +					(buf[idx + (i / 2)] >> ((i % 2) ? 0 :
> > 4)) & 0xf;
> > +			}
> > +		}
> > +		break;
> > +	case DP_QUERY_PAYLOAD:
> > +		req->u.query_payload.port_number = (buf[idx] >> 4) & 0xf;
> > +		req->u.query_payload.vcpi = buf[++idx] & 0x7f;
> 
> Same here for the highest bit?
> 
> > +		break;
> > +	case DP_REMOTE_DPCD_READ:
> > +		{
> > +			struct drm_dp_remote_dpcd_read *r = &req->u.dpcd_read;
> > +
> > +			r->port_number = (buf[idx] >> 4) & 0xf;
> > +
> > +			r->dpcd_address = (buf[idx] << 16) & 0xf0000;
> > +			r->dpcd_address |= (buf[++idx] << 8) & 0xff00;
> > +			r->dpcd_address |= buf[++idx] & 0xff;
> > +
> > +			r->num_bytes = buf[++idx];
> > +		}
> > +		break;
> > +	case DP_REMOTE_DPCD_WRITE:
> > +		{
> > +			struct drm_dp_remote_dpcd_write *w =
> > +				&req->u.dpcd_write;
> > +
> > +			w->port_number = (buf[idx] >> 4) & 0xf;
> > +
> > +			w->dpcd_address = (buf[idx] << 16) & 0xf0000;
> > +			w->dpcd_address |= (buf[++idx] << 8) & 0xff00;
> > +			w->dpcd_address |= buf[++idx] & 0xff;
> > +
> > +			w->num_bytes = buf[++idx];
> > +
> > +			w->bytes = kmemdup(&buf[++idx], w->num_bytes,
> > +					   GFP_KERNEL);
> 
> But if we go really strict on validation we'd need to make sure we don't
> walk past raw->cur_len ... probably not worth it?
> 
> > +			if (!w->bytes)
> > +				return -ENOMEM;
> > +		}
> > +		break;
> > +	case DP_REMOTE_I2C_READ:
> > +		{
> > +			struct drm_dp_remote_i2c_read *r = &req->u.i2c_read;
> > +			struct drm_dp_remote_i2c_read_tx *tx;
> > +			bool failed = false;
> > +
> > +			r->num_transactions = buf[idx] & 0x3;
> > +			r->port_number = (buf[idx] >> 4) & 0xf;
> > +			for (i = 0; i < r->num_transactions; i++) {
> > +				tx = &r->transactions[i];
> > +
> > +				tx->i2c_dev_id = buf[++idx] & 0x7f;
> > +				tx->num_bytes = buf[++idx];
> > +				tx->bytes = kmemdup(&buf[++idx],
> > +						    tx->num_bytes,
> > +						    GFP_KERNEL);
> > +				if (!tx->bytes) {
> > +					failed = true;
> > +					break;
> > +				}
> > +				idx += tx->num_bytes;
> > +				tx->no_stop_bit = (buf[idx] >> 5) & 0x1;
> > +				tx->i2c_transaction_delay = buf[idx] & 0xf;
> > +			}
> > +
> > +			if (failed) {
> > +				for (i = 0; i < r->num_transactions; i++)
> > +					kfree(tx->bytes);
> > +				return -ENOMEM;
> > +			}
> > +
> > +			r->read_i2c_device_id = buf[++idx] & 0x7f;
> > +			r->num_bytes_read = buf[++idx];
> > +		}
> > +		break;
> > +	case DP_REMOTE_I2C_WRITE:
> > +		{
> > +			struct drm_dp_remote_i2c_write *w = &req->u.i2c_write;
> > +
> > +			w->port_number = (buf[idx] >> 4) & 0xf;
> > +			w->write_i2c_device_id = buf[++idx] & 0x7f;
> > +			w->num_bytes = buf[++idx];
> > +			w->bytes = kmemdup(&buf[++idx], w->num_bytes,
> > +					   GFP_KERNEL);
> > +			if (!w->bytes)
> > +				return -ENOMEM;
> > +		}
> > +		break;
> > +	}
> 
> For safety: Should we validate cur_len here? Or return it, and let the
> caller deal with any mismatches.
> 
> But I'm kinda leaning towards no safety here, since it's not dealing with
> any replies we get from the hw/externally, which might be used for
> attacking us. If we later on have the same thing but for
> drm_dp_sideband_parse_reply() then a lot more fun.
> 
> > +
> > +	return 0;
> > +}
> > +EXPORT_SYMBOL_FOR_TESTS_ONLY(drm_dp_decode_sideband_req);
> > +
> > +void
> > +drm_dp_dump_sideband_msg_req_body(const struct
> > drm_dp_sideband_msg_req_body *req,
> > +				  int indent, struct drm_printer *printer)
> > +{
> > +	int i;
> > +
> > +#define P(f, ...) drm_printf_indent(printer, indent, f, ##__VA_ARGS__)
> > +	if (req->req_type == DP_LINK_ADDRESS) {
> > +		/* No contents to print */
> > +		P("type=%s\n", drm_dp_mst_req_type_str(req->req_type));
> > +		return;
> > +	}
> > +
> > +	P("type=%s contents:\n", drm_dp_mst_req_type_str(req->req_type));
> > +	indent++;
> > +
> > +	switch (req->req_type) {
> > +	case DP_ENUM_PATH_RESOURCES:
> > +	case DP_POWER_DOWN_PHY:
> > +	case DP_POWER_UP_PHY:
> > +		P("port=%d\n", req->u.port_num.port_number);
> > +		break;
> > +	case DP_ALLOCATE_PAYLOAD:
> > +		P("port=%d vcpi=%d pbn=%d sdp_streams=%d %*ph\n",
> > +		  req->u.allocate_payload.port_number,
> > +		  req->u.allocate_payload.vcpi, req->u.allocate_payload.pbn,
> > +		  req->u.allocate_payload.number_sdp_streams,
> > +		  req->u.allocate_payload.number_sdp_streams,
> > +		  req->u.allocate_payload.sdp_stream_sink);
> > +		break;
> > +	case DP_QUERY_PAYLOAD:
> > +		P("port=%d vcpi=%d\n",
> > +		  req->u.query_payload.port_number,
> > +		  req->u.query_payload.vcpi);
> > +		break;
> > +	case DP_REMOTE_DPCD_READ:
> > +		P("port=%d dpcd_addr=%05x len=%d\n",
> > +		  req->u.dpcd_read.port_number, req->u.dpcd_read.dpcd_address,
> > +		  req->u.dpcd_read.num_bytes);
> > +		break;
> > +	case DP_REMOTE_DPCD_WRITE:
> > +		P("port=%d addr=%05x len=%d: %*ph\n",
> > +		  req->u.dpcd_write.port_number,
> > +		  req->u.dpcd_write.dpcd_address,
> > +		  req->u.dpcd_write.num_bytes, req->u.dpcd_write.num_bytes,
> > +		  req->u.dpcd_write.bytes);
> > +		break;
> > +	case DP_REMOTE_I2C_READ:
> > +		P("port=%d num_tx=%d id=%d size=%d:\n",
> > +		  req->u.i2c_read.port_number,
> > +		  req->u.i2c_read.num_transactions,
> > +		  req->u.i2c_read.read_i2c_device_id,
> > +		  req->u.i2c_read.num_bytes_read);
> > +
> > +		indent++;
> > +		for (i = 0; i < req->u.i2c_read.num_transactions; i++) {
> > +			const struct drm_dp_remote_i2c_read_tx *rtx =
> > +				&req->u.i2c_read.transactions[i];
> > +
> > +			P("%d: id=%03d size=%03d no_stop_bit=%d tx_delay=%03d:
> > %*ph\n",
> > +			  i, rtx->i2c_dev_id, rtx->num_bytes,
> > +			  rtx->no_stop_bit, rtx->i2c_transaction_delay,
> > +			  rtx->num_bytes, rtx->bytes);
> > +		}
> > +		break;
> > +	case DP_REMOTE_I2C_WRITE:
> > +		P("port=%d id=%d size=%d: %*ph\n",
> > +		  req->u.i2c_write.port_number,
> > +		  req->u.i2c_write.write_i2c_device_id,
> > +		  req->u.i2c_write.num_bytes, req->u.i2c_write.num_bytes,
> > +		  req->u.i2c_write.bytes);
> > +		break;
> > +	default:
> > +		P("???\n");
> > +		break;
> > +	}
> > +#undef P
> > +}
> > +EXPORT_SYMBOL_FOR_TESTS_ONLY(drm_dp_dump_sideband_msg_req_body);
> > +
> > +static inline void
> > +drm_dp_mst_dump_sideband_msg_tx(struct drm_printer *p,
> > +				const struct drm_dp_sideband_msg_tx *txmsg)
> > +{
> > +	struct drm_dp_sideband_msg_req_body req;
> > +	char buf[64];
> > +	int ret;
> > +	int i;
> > +
> > +	drm_dp_mst_rad_to_str(txmsg->dst->rad, txmsg->dst->lct, buf,
> > +			      sizeof(buf));
> > +	drm_printf(p, "txmsg cur_offset=%x cur_len=%x seqno=%x state=%s
> > path_msg=%d dst=%s\n",
> > +		   txmsg->cur_offset, txmsg->cur_len, txmsg->seqno,
> > +		   drm_dp_mst_sideband_tx_state_str(txmsg->state),
> > +		   txmsg->path_msg, buf);
> > +
> > +	ret = drm_dp_decode_sideband_req(txmsg, &req);
> > +	if (ret) {
> > +		drm_printf(p, "<failed to decode sideband req: %d>\n", ret);
> > +		return;
> > +	}
> > +	drm_dp_dump_sideband_msg_req_body(&req, 1, p);
> > +
> > +	switch (req.req_type) {
> > +	case DP_REMOTE_DPCD_WRITE:
> > +		kfree(req.u.dpcd_write.bytes);
> > +		break;
> > +	case DP_REMOTE_I2C_READ:
> > +		for (i = 0; i < req.u.i2c_read.num_transactions; i++)
> > +			kfree(req.u.i2c_read.transactions[i].bytes);
> > +		break;
> > +	case DP_REMOTE_I2C_WRITE:
> > +		kfree(req.u.i2c_write.bytes);
> > +		break;
> > +	}
> > +}
> >  
> >  static void drm_dp_crc_sideband_chunk_req(u8 *msg, u8 len)
> >  {
> > @@ -893,6 +1178,11 @@ static int drm_dp_mst_wait_tx_reply(struct
> > drm_dp_mst_branch *mstb,
> >  		}
> >  	}
> >  out:
> > +	if (unlikely(ret == -EIO && drm_debug & (DRM_UT_DP | DRM_UT_KMS))) {
> 
> Hm I'd only check for DRM_UT_DP here.
> 
> > +		struct drm_printer p = drm_debug_printer(DBG_PREFIX);
> > +
> > +		drm_dp_mst_dump_sideband_msg_tx(&p, txmsg);
> > +	}
> >  	mutex_unlock(&mgr->qlock);
> >  
> >  	return ret;
> > @@ -1927,8 +2217,11 @@ static int process_single_tx_qlock(struct
> > drm_dp_mst_topology_mgr *mgr,
> >  	idx += tosend + 1;
> >  
> >  	ret = drm_dp_send_sideband_msg(mgr, up, chunk, idx);
> > -	if (ret) {
> > -		DRM_DEBUG_KMS("sideband msg failed to send\n");
> > +	if (ret && unlikely(drm_debug & (DRM_UT_DP | DRM_UT_KMS))) {
> 
> Same.
> 
> > +		struct drm_printer p = drm_debug_printer(DBG_PREFIX);
> > +
> > +		drm_printf(&p, "sideband msg failed to send\n");
> > +		drm_dp_mst_dump_sideband_msg_tx(&p, txmsg);
> >  		return ret;
> >  	}
> >  
> > @@ -1990,6 +2283,13 @@ static void drm_dp_queue_down_tx(struct
> > drm_dp_mst_topology_mgr *mgr,
> >  {
> >  	mutex_lock(&mgr->qlock);
> >  	list_add_tail(&txmsg->next, &mgr->tx_msg_downq);
> > +
> > +	if (unlikely(drm_debug & DRM_UT_DP)) {
> 
> Like you do here.
> 
> > +		struct drm_printer p = drm_debug_printer(DBG_PREFIX);
> > +
> > +		drm_dp_mst_dump_sideband_msg_tx(&p, txmsg);
> > +	}
> > +
> >  	if (list_is_singular(&mgr->tx_msg_downq))
> >  		process_single_down_tx_qlock(mgr);
> >  	mutex_unlock(&mgr->qlock);
> > diff --git a/drivers/gpu/drm/drm_dp_mst_topology_internal.h
> > b/drivers/gpu/drm/drm_dp_mst_topology_internal.h
> > new file mode 100644
> > index 000000000000..eeda9a61c657
> > --- /dev/null
> > +++ b/drivers/gpu/drm/drm_dp_mst_topology_internal.h
> > @@ -0,0 +1,24 @@
> > +/* SPDX-License-Identifier: GPL-2.0-only
> > + *
> > + * Declarations for DP MST related functions which are only used in
> > selftests
> > + *
> > + * Copyright © 2018 Red Hat
> > + * Authors:
> > + *     Lyude Paul <lyude@redhat.com>
> > + */
> > +
> > +#ifndef _DRM_DP_MST_HELPER_INTERNAL_H_
> > +#define _DRM_DP_MST_HELPER_INTERNAL_H_
> > +
> > +#include <drm/drm_dp_mst_helper.h>
> > +
> > +void
> > +drm_dp_encode_sideband_req(const struct drm_dp_sideband_msg_req_body
> > *req,
> > +			   struct drm_dp_sideband_msg_tx *raw);
> > +int drm_dp_decode_sideband_req(const struct drm_dp_sideband_msg_tx *raw,
> > +			       struct drm_dp_sideband_msg_req_body *req);
> > +void
> > +drm_dp_dump_sideband_msg_req_body(const struct
> > drm_dp_sideband_msg_req_body *req,
> > +				  int indent, struct drm_printer *printer);
> > +
> > +#endif /* !_DRM_DP_MST_HELPER_INTERNAL_H_ */
> > diff --git a/drivers/gpu/drm/selftests/drm_modeset_selftests.h
> > b/drivers/gpu/drm/selftests/drm_modeset_selftests.h
> > index dec3ee3ec96f..1898de0b4a4d 100644
> > --- a/drivers/gpu/drm/selftests/drm_modeset_selftests.h
> > +++ b/drivers/gpu/drm/selftests/drm_modeset_selftests.h
> > @@ -33,3 +33,4 @@ selftest(damage_iter_damage_one_outside,
> > igt_damage_iter_damage_one_outside)
> >  selftest(damage_iter_damage_src_moved, igt_damage_iter_damage_src_moved)
> >  selftest(damage_iter_damage_not_visible,
> > igt_damage_iter_damage_not_visible)
> >  selftest(dp_mst_calc_pbn_mode, igt_dp_mst_calc_pbn_mode)
> > +selftest(dp_mst_sideband_msg_req_decode,
> > igt_dp_mst_sideband_msg_req_decode)
> > diff --git a/drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c
> > b/drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c
> > index 51b2486ec917..ceca89babd65 100644
> > --- a/drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c
> > +++ b/drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c
> > @@ -8,6 +8,7 @@
> >  #include <drm/drm_dp_mst_helper.h>
> >  #include <drm/drm_print.h>
> >  
> > +#include "../drm_dp_mst_topology_internal.h"
> >  #include "test-drm_modeset_common.h"
> >  
> >  int igt_dp_mst_calc_pbn_mode(void *ignored)
> > @@ -34,3 +35,214 @@ int igt_dp_mst_calc_pbn_mode(void *ignored)
> >  
> >  	return 0;
> >  }
> > +
> > +static bool
> > +sideband_msg_req_equal(const struct drm_dp_sideband_msg_req_body *in,
> > +		       const struct drm_dp_sideband_msg_req_body *out)
> > +{
> > +	const struct drm_dp_remote_i2c_read_tx *txin, *txout;
> > +	int i;
> > +
> > +	if (in->req_type != out->req_type)
> > +		return false;
> > +
> > +	switch (in->req_type) {
> > +	case DP_ENUM_PATH_RESOURCES:
> > +	case DP_POWER_UP_PHY:
> > +	case DP_POWER_DOWN_PHY:
> > +	case DP_ALLOCATE_PAYLOAD:
> > +	case DP_QUERY_PAYLOAD:
> > +	case DP_REMOTE_DPCD_READ:
> > +		return memcmp(in, out, sizeof(*in)) == 0;
> 
> Just memcmp the entire thing for everyrone (we assume kzalloc anyway), and
> then only have the additional checks for the few cases we need it? Would
> slightly reduce the code.

Just realized also - this wouldn't work: remember that I2C_READ, DPCD_WRITE,
and I2C_WRITE contain pointers which of course will be different even if the
contents are the same. Unfortunately I don't think there's really a good
solution for this, unless we want to just remove the pointers and store things
within the down request struct itself

> 
> > +
> > +	case DP_REMOTE_I2C_READ:
> > +#define IN in->u.i2c_read
> > +#define OUT out->u.i2c_read
> > +		if (IN.num_bytes_read != OUT.num_bytes_read ||
> > +		    IN.num_transactions != OUT.num_transactions ||
> > +		    IN.port_number != OUT.port_number ||
> > +		    IN.read_i2c_device_id != OUT.read_i2c_device_id)
> > +			return false;
> > +
> > +		for (i = 0; i < IN.num_transactions; i++) {
> > +			txin = &IN.transactions[i];
> > +			txout = &OUT.transactions[i];
> > +
> > +			if (txin->i2c_dev_id != txout->i2c_dev_id ||
> > +			    txin->no_stop_bit != txout->no_stop_bit ||
> > +			    txin->num_bytes != txout->num_bytes ||
> > +			    txin->i2c_transaction_delay !=
> > +			    txout->i2c_transaction_delay)
> > +				return false;
> > +
> > +			if (memcmp(txin->bytes, txout->bytes,
> > +				   txin->num_bytes) != 0)
> > +				return false;
> > +		}
> > +		break;
> > +#undef IN
> > +#undef OUT
> > +
> > +	case DP_REMOTE_DPCD_WRITE:
> > +#define IN in->u.dpcd_write
> > +#define OUT out->u.dpcd_write
> > +		if (IN.dpcd_address != OUT.dpcd_address ||
> > +		    IN.num_bytes != OUT.num_bytes ||
> > +		    IN.port_number != OUT.port_number)
> > +			return false;
> > +
> > +		return memcmp(IN.bytes, OUT.bytes, IN.num_bytes) == 0;
> > +#undef IN
> > +#undef OUT
> > +
> > +	case DP_REMOTE_I2C_WRITE:
> > +#define IN in->u.i2c_write
> > +#define OUT out->u.i2c_write
> > +		if (IN.port_number != OUT.port_number ||
> > +		    IN.write_i2c_device_id != OUT.write_i2c_device_id ||
> > +		    IN.num_bytes != OUT.num_bytes)
> > +			return false;
> > +
> > +		return memcmp(IN.bytes, OUT.bytes, IN.num_bytes) == 0;
> > +#undef IN
> > +#undef OUT
> > +	}
> > +
> > +	return true;
> > +}
> > +
> > +static int
> > +__sideband_msg_req_encode_decode(int line,
> > +				 struct drm_dp_sideband_msg_req_body *in,
> > +				 struct drm_dp_sideband_msg_req_body *out)
> > +{
> > +	struct drm_printer p = drm_err_printer(pr_fmt());
> > +	struct drm_dp_sideband_msg_tx txmsg;
> > +	int i, ret;
> > +	bool eq;
> > +
> > +	drm_dp_encode_sideband_req(in, &txmsg);
> > +	ret = drm_dp_decode_sideband_req(&txmsg, out);
> > +	if (ret < 0) {
> > +		pr_err(pr_fmt("Failed to decode sideband request @ line %d:
> > %d\n"),
> > +		       line, ret);
> > +		return ret;
> > +	}
> > +
> > +	eq = sideband_msg_req_equal(in, out);
> > +	if (!eq) {
> > +		pr_err(pr_fmt("Encode/decode @ line %d failed, expected:\n"),
> > +		       line);
> > +		drm_dp_dump_sideband_msg_req_body(in, 1, &p);
> > +		pr_err(pr_fmt("Got:\n"));
> > +		drm_dp_dump_sideband_msg_req_body(out, 1, &p);
> > +	}
> > +
> > +	switch (in->req_type) {
> > +	case DP_REMOTE_DPCD_WRITE:
> > +		kfree(out->u.dpcd_write.bytes);
> > +		break;
> > +	case DP_REMOTE_I2C_READ:
> > +		for (i = 0; i < out->u.i2c_read.num_transactions; i++)
> > +			kfree(out->u.i2c_read.transactions[i].bytes);
> > +		break;
> > +	case DP_REMOTE_I2C_WRITE:
> > +		kfree(out->u.i2c_write.bytes);
> > +		break;
> > +	}
> > +
> > +	/* Clear everything but the req_type for the input */
> > +	memset(&in->u, 0, sizeof(in->u));
> > +
> > +	/* Clear the output entirely */
> > +	memset(out, 0, sizeof(*out));
> > +
> > +	return eq ? 0 : -EINVAL;
> > +}
> > +
> > +int igt_dp_mst_sideband_msg_req_decode(void *unused)
> > +{
> > +	struct drm_dp_sideband_msg_req_body in = { 0 }, out = { 0 };
> > +	u8 data[] = { 0xff, 0x0, 0xdd };
> > +	int i;
> > +
> > +#define
> > DO_TEST()                                                          \
> > +	do {                                                               \
> > +		if (__sideband_msg_req_encode_decode(__LINE__, &in, &out)) \
> > +			return -EINVAL;                                    \
> > +	} while (0)
> 
> I had a tiny wtf moment here until I realized this is a macro ... maybe
> put the do on the first line and indent everything? Pure bikeshed to help
> the blind ...
> 
> > +
> > +	in.req_type = DP_ENUM_PATH_RESOURCES;
> > +	in.u.port_num.port_number = 5;
> > +	DO_TEST();
> > +
> > +	in.req_type = DP_POWER_UP_PHY;
> > +	in.u.port_num.port_number = 5;
> > +	DO_TEST();
> > +
> > +	in.req_type = DP_POWER_DOWN_PHY;
> > +	in.u.port_num.port_number = 5;
> > +	DO_TEST();
> > +
> > +	in.req_type = DP_ALLOCATE_PAYLOAD;
> > +	in.u.allocate_payload.number_sdp_streams = 3;
> > +	for (i = 0; i < in.u.allocate_payload.number_sdp_streams; i++)
> > +		in.u.allocate_payload.sdp_stream_sink[i] = i + 1;
> > +	DO_TEST();
> > +	in.u.allocate_payload.port_number = 0xf;
> > +	DO_TEST();
> > +	in.u.allocate_payload.vcpi = 0x7f;
> > +	DO_TEST();
> > +	in.u.allocate_payload.pbn = U16_MAX;
> > +	DO_TEST();
> > +
> > +	in.req_type = DP_QUERY_PAYLOAD;
> > +	in.u.query_payload.port_number = 0xf;
> > +	DO_TEST();
> > +	in.u.query_payload.vcpi = 0x7f;
> > +	DO_TEST();
> > +
> > +	in.req_type = DP_REMOTE_DPCD_READ;
> > +	in.u.dpcd_read.port_number = 0xf;
> > +	DO_TEST();
> > +	in.u.dpcd_read.dpcd_address = 0xfedcb;
> > +	DO_TEST();
> > +	in.u.dpcd_read.num_bytes = U8_MAX;
> > +	DO_TEST();
> > +
> > +	in.req_type = DP_REMOTE_DPCD_WRITE;
> > +	in.u.dpcd_write.port_number = 0xf;
> > +	DO_TEST();
> > +	in.u.dpcd_write.dpcd_address = 0xfedcb;
> > +	DO_TEST();
> > +	in.u.dpcd_write.num_bytes = ARRAY_SIZE(data);
> > +	in.u.dpcd_write.bytes = data;
> > +	DO_TEST();
> > +
> > +	in.req_type = DP_REMOTE_I2C_READ;
> > +	in.u.i2c_read.port_number = 0xf;
> > +	DO_TEST();
> > +	in.u.i2c_read.read_i2c_device_id = 0x7f;
> > +	DO_TEST();
> > +	in.u.i2c_read.num_transactions = 3;
> > +	in.u.i2c_read.num_bytes_read = ARRAY_SIZE(data) * 3;
> > +	for (i = 0; i < in.u.i2c_read.num_transactions; i++) {
> > +		in.u.i2c_read.transactions[i].bytes = data;
> > +		in.u.i2c_read.transactions[i].num_bytes = ARRAY_SIZE(data);
> > +		in.u.i2c_read.transactions[i].i2c_dev_id = 0x7f & ~i;
> > +		in.u.i2c_read.transactions[i].i2c_transaction_delay = 0xf &
> > ~i;
> > +	}
> > +	DO_TEST();
> > +
> > +	in.req_type = DP_REMOTE_I2C_WRITE;
> > +	in.u.i2c_write.port_number = 0xf;
> > +	DO_TEST();
> > +	in.u.i2c_write.write_i2c_device_id = 0x7f;
> > +	DO_TEST();
> > +	in.u.i2c_write.num_bytes = ARRAY_SIZE(data);
> > +	in.u.i2c_write.bytes = data;
> > +	DO_TEST();
> > +
> > +#undef DO_TEST
> > +	return 0;
> > +}
> 
> Extremely nice, more unit tests ftw!
> 
> With the nits somehow figured out: Reviewed-by: Daniel Vetter <
> daniel.vetter@ffwll.ch>
> 
> > diff --git a/drivers/gpu/drm/selftests/test-drm_modeset_common.h
> > b/drivers/gpu/drm/selftests/test-drm_modeset_common.h
> > index 590bda35a683..0fcb8bbc6a1b 100644
> > --- a/drivers/gpu/drm/selftests/test-drm_modeset_common.h
> > +++ b/drivers/gpu/drm/selftests/test-drm_modeset_common.h
> > @@ -40,5 +40,6 @@ int igt_damage_iter_damage_one_outside(void *ignored);
> >  int igt_damage_iter_damage_src_moved(void *ignored);
> >  int igt_damage_iter_damage_not_visible(void *ignored);
> >  int igt_dp_mst_calc_pbn_mode(void *ignored);
> > +int igt_dp_mst_sideband_msg_req_decode(void *ignored);
> >  
> >  #endif
> > diff --git a/include/drm/drm_dp_mst_helper.h
> > b/include/drm/drm_dp_mst_helper.h
> > index c01f3ea72756..4c8d177e83e5 100644
> > --- a/include/drm/drm_dp_mst_helper.h
> > +++ b/include/drm/drm_dp_mst_helper.h
> > @@ -293,7 +293,7 @@ struct drm_dp_remote_dpcd_write {
> >  struct drm_dp_remote_i2c_read {
> >  	u8 num_transactions;
> >  	u8 port_number;
> > -	struct {
> > +	struct drm_dp_remote_i2c_read_tx {
> >  		u8 i2c_dev_id;
> >  		u8 num_bytes;
> >  		u8 *bytes;
> > -- 
> > 2.21.0
> > 
-- 
Cheers,
	Lyude Paul

Re: [PATCH 05/26] drm/dp_mst: Add sideband down request tracing + selftests

[Daniel Vetter]

On Tue, Aug 27, 2019 at 12:49:17PM -0400, Lyude Paul wrote:
> On Tue, 2019-08-13 at 16:50 +0200, Daniel Vetter wrote:
> > On Wed, Jul 17, 2019 at 09:42:28PM -0400, Lyude Paul wrote:
> > > Unfortunately the DP MST helpers do not have much in the way of
> > > debugging utilities. So, let's add some!
> > > 
> > > This adds basic debugging output for down sideband requests that we send
> > > from the driver, so that we can actually discern what's happening when
> > > sideband requests timeout. Note that with this commit, we'll be dumping
> > > out sideband requests under both of the following conditions:
> > > 
> > > - When the user has enabled DRM_UT_DP output, of course.
> > > - When the user has enabled DRM_UT_KMS or DRM_UT_DP, and a sideband
> > >   request has failed in some way. This will allow for developers to get
> > >   a basic idea of what's actually happening with failed modesets on MST,
> > >   without needing to have DRM_UT_DP explicitly enabled.
> > > 
> > > Since there wasn't really a good way of testing that any of this worked,
> > > I ended up writing simple selftests that lightly test sideband message
> > > encoding and decoding as well. Enjoy!
> > > 
> > > Cc: Juston Li <juston.li@intel.com>
> > > Cc: Imre Deak <imre.deak@intel.com>
> > > Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
> > > Cc: Harry Wentland <hwentlan@amd.com>
> > > Signed-off-by: Lyude Paul <lyude@redhat.com>
> > > ---
> > >  drivers/gpu/drm/drm_dp_mst_topology.c         | 308 +++++++++++++++++-
> > >  .../gpu/drm/drm_dp_mst_topology_internal.h    |  24 ++
> > >  .../gpu/drm/selftests/drm_modeset_selftests.h |   1 +
> > >  .../drm/selftests/test-drm_dp_mst_helper.c    | 212 ++++++++++++
> > >  .../drm/selftests/test-drm_modeset_common.h   |   1 +
> > >  include/drm/drm_dp_mst_helper.h               |   2 +-
> > >  6 files changed, 543 insertions(+), 5 deletions(-)
> > >  create mode 100644 drivers/gpu/drm/drm_dp_mst_topology_internal.h
> > > 
> > > diff --git a/drivers/gpu/drm/drm_dp_mst_topology.c
> > > b/drivers/gpu/drm/drm_dp_mst_topology.c
> > > index 9e382117896d..defc5e09fb9a 100644
> > > --- a/drivers/gpu/drm/drm_dp_mst_topology.c
> > > +++ b/drivers/gpu/drm/drm_dp_mst_topology.c
> > > @@ -36,6 +36,8 @@
> > >  #include <drm/drm_print.h>
> > >  #include <drm/drm_probe_helper.h>
> > >  
> > > +#include "drm_dp_mst_topology_internal.h"
> > > +
> > >  /**
> > >   * DOC: dp mst helper
> > >   *
> > > @@ -68,6 +70,8 @@ static int drm_dp_mst_register_i2c_bus(struct drm_dp_aux
> > > *aux);
> > >  static void drm_dp_mst_unregister_i2c_bus(struct drm_dp_aux *aux);
> > >  static void drm_dp_mst_kick_tx(struct drm_dp_mst_topology_mgr *mgr);
> > >  
> > > +#define DBG_PREFIX "[dp_mst]"
> > > +
> > >  #define DP_STR(x) [DP_ ## x] = #x
> > >  
> > >  static const char *drm_dp_mst_req_type_str(u8 req_type)
> > > @@ -124,6 +128,43 @@ static const char *drm_dp_mst_nak_reason_str(u8
> > > nak_reason)
> > >  }
> > >  
> > >  #undef DP_STR
> > > +#define DP_STR(x) [DRM_DP_SIDEBAND_TX_ ## x] = #x
> > > +
> > > +static const char *drm_dp_mst_sideband_tx_state_str(int state)
> > > +{
> > > +	static const char * const sideband_reason_str[] = {
> > > +		DP_STR(QUEUED),
> > > +		DP_STR(START_SEND),
> > > +		DP_STR(SENT),
> > > +		DP_STR(RX),
> > > +		DP_STR(TIMEOUT),
> > > +	};
> > > +
> > > +	if (state >= ARRAY_SIZE(sideband_reason_str) ||
> > > +	    !sideband_reason_str[state])
> > > +		return "unknown";
> > > +
> > > +	return sideband_reason_str[state];
> > > +}
> > > +
> > > +static int
> > > +drm_dp_mst_rad_to_str(const u8 rad[8], u8 lct, char *out, size_t len)
> > > +{
> > > +	int i;
> > > +	u8 unpacked_rad[16];
> > > +
> > > +	for (i = 0; i < lct; i++) {
> > > +		if (i % 2)
> > > +			unpacked_rad[i] = rad[i / 2] >> 4;
> > > +		else
> > > +			unpacked_rad[i] = rad[i / 2] & BIT_MASK(4);
> > > +	}
> > > +
> > > +	/* TODO: Eventually add something to printk so we can format the rad
> > > +	 * like this: 1.2.3
> > > +	 */
> > 
> > 	lct *=2;
> > 
> > missing here? And yeah the todo would be sweet, but quite a bit more
> > typing I guess.
> > 
> > > +	return snprintf(out, len, "%*phC", lct, unpacked_rad);
> > > +}
> > >  
> > >  /* sideband msg handling */
> > >  static u8 drm_dp_msg_header_crc4(const uint8_t *data, size_t num_nibbles)
> > > @@ -256,8 +297,9 @@ static bool drm_dp_decode_sideband_msg_hdr(struct
> > > drm_dp_sideband_msg_hdr *hdr,
> > >  	return true;
> > >  }
> > >  
> > > -static void drm_dp_encode_sideband_req(struct
> > > drm_dp_sideband_msg_req_body *req,
> > > -				       struct drm_dp_sideband_msg_tx *raw)
> > > +void
> > > +drm_dp_encode_sideband_req(const struct drm_dp_sideband_msg_req_body
> > > *req,
> > > +			   struct drm_dp_sideband_msg_tx *raw)
> > >  {
> > >  	int idx = 0;
> > >  	int i;
> > > @@ -362,6 +404,249 @@ static void drm_dp_encode_sideband_req(struct
> > > drm_dp_sideband_msg_req_body *req,
> > >  	}
> > >  	raw->cur_len = idx;
> > >  }
> > > +EXPORT_SYMBOL_FOR_TESTS_ONLY(drm_dp_encode_sideband_req);
> > > +
> > > +/* Decode a sideband request we've encoded, mainly used for debugging */
> > > +int
> > > +drm_dp_decode_sideband_req(const struct drm_dp_sideband_msg_tx *raw,
> > > +			   struct drm_dp_sideband_msg_req_body *req)
> > > +{
> > > +	const u8 *buf = raw->msg;
> > > +	int i, idx = 0;
> > > +
> > > +	req->req_type = buf[idx++] & 0x7f;
> > > +	switch (req->req_type) {
> > > +	case DP_ENUM_PATH_RESOURCES:
> > > +	case DP_POWER_DOWN_PHY:
> > > +	case DP_POWER_UP_PHY:
> > 
> > OCD warning: the enum isn't ordered the same way here and in
> > drm_dp_encode_sideband_req().
> > 
> > > +		req->u.port_num.port_number = (buf[idx] >> 4) & 0xf;
> > > +		break;
> > > +	case DP_ALLOCATE_PAYLOAD:
> > > +		{
> > > +			struct drm_dp_allocate_payload *a =
> > > +				&req->u.allocate_payload;
> > > +
> > > +			a->number_sdp_streams = buf[idx] & 0xf;
> > > +			a->port_number = (buf[idx] >> 4) & 0xf;
> > > +
> > > +			a->vcpi = buf[++idx] & 0x7f;
> > 
> > Complain here if 0x80 is set?
> > 
> > > +
> > > +			a->pbn = buf[++idx] << 8;
> > > +			a->pbn |= buf[++idx];
> > > +
> > > +			idx++;
> > > +			for (i = 0; i < a->number_sdp_streams; i++) {
> > > +				a->sdp_stream_sink[i] =
> > > +					(buf[idx + (i / 2)] >> ((i % 2) ? 0 :
> > > 4)) & 0xf;
> > > +			}
> > > +		}
> > > +		break;
> > > +	case DP_QUERY_PAYLOAD:
> > > +		req->u.query_payload.port_number = (buf[idx] >> 4) & 0xf;
> > > +		req->u.query_payload.vcpi = buf[++idx] & 0x7f;
> > 
> > Same here for the highest bit?
> > 
> > > +		break;
> > > +	case DP_REMOTE_DPCD_READ:
> > > +		{
> > > +			struct drm_dp_remote_dpcd_read *r = &req->u.dpcd_read;
> > > +
> > > +			r->port_number = (buf[idx] >> 4) & 0xf;
> > > +
> > > +			r->dpcd_address = (buf[idx] << 16) & 0xf0000;
> > > +			r->dpcd_address |= (buf[++idx] << 8) & 0xff00;
> > > +			r->dpcd_address |= buf[++idx] & 0xff;
> > > +
> > > +			r->num_bytes = buf[++idx];
> > > +		}
> > > +		break;
> > > +	case DP_REMOTE_DPCD_WRITE:
> > > +		{
> > > +			struct drm_dp_remote_dpcd_write *w =
> > > +				&req->u.dpcd_write;
> > > +
> > > +			w->port_number = (buf[idx] >> 4) & 0xf;
> > > +
> > > +			w->dpcd_address = (buf[idx] << 16) & 0xf0000;
> > > +			w->dpcd_address |= (buf[++idx] << 8) & 0xff00;
> > > +			w->dpcd_address |= buf[++idx] & 0xff;
> > > +
> > > +			w->num_bytes = buf[++idx];
> > > +
> > > +			w->bytes = kmemdup(&buf[++idx], w->num_bytes,
> > > +					   GFP_KERNEL);
> > 
> > But if we go really strict on validation we'd need to make sure we don't
> > walk past raw->cur_len ... probably not worth it?
> > 
> > > +			if (!w->bytes)
> > > +				return -ENOMEM;
> > > +		}
> > > +		break;
> > > +	case DP_REMOTE_I2C_READ:
> > > +		{
> > > +			struct drm_dp_remote_i2c_read *r = &req->u.i2c_read;
> > > +			struct drm_dp_remote_i2c_read_tx *tx;
> > > +			bool failed = false;
> > > +
> > > +			r->num_transactions = buf[idx] & 0x3;
> > > +			r->port_number = (buf[idx] >> 4) & 0xf;
> > > +			for (i = 0; i < r->num_transactions; i++) {
> > > +				tx = &r->transactions[i];
> > > +
> > > +				tx->i2c_dev_id = buf[++idx] & 0x7f;
> > > +				tx->num_bytes = buf[++idx];
> > > +				tx->bytes = kmemdup(&buf[++idx],
> > > +						    tx->num_bytes,
> > > +						    GFP_KERNEL);
> > > +				if (!tx->bytes) {
> > > +					failed = true;
> > > +					break;
> > > +				}
> > > +				idx += tx->num_bytes;
> > > +				tx->no_stop_bit = (buf[idx] >> 5) & 0x1;
> > > +				tx->i2c_transaction_delay = buf[idx] & 0xf;
> > > +			}
> > > +
> > > +			if (failed) {
> > > +				for (i = 0; i < r->num_transactions; i++)
> > > +					kfree(tx->bytes);
> > > +				return -ENOMEM;
> > > +			}
> > > +
> > > +			r->read_i2c_device_id = buf[++idx] & 0x7f;
> > > +			r->num_bytes_read = buf[++idx];
> > > +		}
> > > +		break;
> > > +	case DP_REMOTE_I2C_WRITE:
> > > +		{
> > > +			struct drm_dp_remote_i2c_write *w = &req->u.i2c_write;
> > > +
> > > +			w->port_number = (buf[idx] >> 4) & 0xf;
> > > +			w->write_i2c_device_id = buf[++idx] & 0x7f;
> > > +			w->num_bytes = buf[++idx];
> > > +			w->bytes = kmemdup(&buf[++idx], w->num_bytes,
> > > +					   GFP_KERNEL);
> > > +			if (!w->bytes)
> > > +				return -ENOMEM;
> > > +		}
> > > +		break;
> > > +	}
> > 
> > For safety: Should we validate cur_len here? Or return it, and let the
> > caller deal with any mismatches.
> > 
> > But I'm kinda leaning towards no safety here, since it's not dealing with
> > any replies we get from the hw/externally, which might be used for
> > attacking us. If we later on have the same thing but for
> > drm_dp_sideband_parse_reply() then a lot more fun.
> > 
> > > +
> > > +	return 0;
> > > +}
> > > +EXPORT_SYMBOL_FOR_TESTS_ONLY(drm_dp_decode_sideband_req);
> > > +
> > > +void
> > > +drm_dp_dump_sideband_msg_req_body(const struct
> > > drm_dp_sideband_msg_req_body *req,
> > > +				  int indent, struct drm_printer *printer)
> > > +{
> > > +	int i;
> > > +
> > > +#define P(f, ...) drm_printf_indent(printer, indent, f, ##__VA_ARGS__)
> > > +	if (req->req_type == DP_LINK_ADDRESS) {
> > > +		/* No contents to print */
> > > +		P("type=%s\n", drm_dp_mst_req_type_str(req->req_type));
> > > +		return;
> > > +	}
> > > +
> > > +	P("type=%s contents:\n", drm_dp_mst_req_type_str(req->req_type));
> > > +	indent++;
> > > +
> > > +	switch (req->req_type) {
> > > +	case DP_ENUM_PATH_RESOURCES:
> > > +	case DP_POWER_DOWN_PHY:
> > > +	case DP_POWER_UP_PHY:
> > > +		P("port=%d\n", req->u.port_num.port_number);
> > > +		break;
> > > +	case DP_ALLOCATE_PAYLOAD:
> > > +		P("port=%d vcpi=%d pbn=%d sdp_streams=%d %*ph\n",
> > > +		  req->u.allocate_payload.port_number,
> > > +		  req->u.allocate_payload.vcpi, req->u.allocate_payload.pbn,
> > > +		  req->u.allocate_payload.number_sdp_streams,
> > > +		  req->u.allocate_payload.number_sdp_streams,
> > > +		  req->u.allocate_payload.sdp_stream_sink);
> > > +		break;
> > > +	case DP_QUERY_PAYLOAD:
> > > +		P("port=%d vcpi=%d\n",
> > > +		  req->u.query_payload.port_number,
> > > +		  req->u.query_payload.vcpi);
> > > +		break;
> > > +	case DP_REMOTE_DPCD_READ:
> > > +		P("port=%d dpcd_addr=%05x len=%d\n",
> > > +		  req->u.dpcd_read.port_number, req->u.dpcd_read.dpcd_address,
> > > +		  req->u.dpcd_read.num_bytes);
> > > +		break;
> > > +	case DP_REMOTE_DPCD_WRITE:
> > > +		P("port=%d addr=%05x len=%d: %*ph\n",
> > > +		  req->u.dpcd_write.port_number,
> > > +		  req->u.dpcd_write.dpcd_address,
> > > +		  req->u.dpcd_write.num_bytes, req->u.dpcd_write.num_bytes,
> > > +		  req->u.dpcd_write.bytes);
> > > +		break;
> > > +	case DP_REMOTE_I2C_READ:
> > > +		P("port=%d num_tx=%d id=%d size=%d:\n",
> > > +		  req->u.i2c_read.port_number,
> > > +		  req->u.i2c_read.num_transactions,
> > > +		  req->u.i2c_read.read_i2c_device_id,
> > > +		  req->u.i2c_read.num_bytes_read);
> > > +
> > > +		indent++;
> > > +		for (i = 0; i < req->u.i2c_read.num_transactions; i++) {
> > > +			const struct drm_dp_remote_i2c_read_tx *rtx =
> > > +				&req->u.i2c_read.transactions[i];
> > > +
> > > +			P("%d: id=%03d size=%03d no_stop_bit=%d tx_delay=%03d:
> > > %*ph\n",
> > > +			  i, rtx->i2c_dev_id, rtx->num_bytes,
> > > +			  rtx->no_stop_bit, rtx->i2c_transaction_delay,
> > > +			  rtx->num_bytes, rtx->bytes);
> > > +		}
> > > +		break;
> > > +	case DP_REMOTE_I2C_WRITE:
> > > +		P("port=%d id=%d size=%d: %*ph\n",
> > > +		  req->u.i2c_write.port_number,
> > > +		  req->u.i2c_write.write_i2c_device_id,
> > > +		  req->u.i2c_write.num_bytes, req->u.i2c_write.num_bytes,
> > > +		  req->u.i2c_write.bytes);
> > > +		break;
> > > +	default:
> > > +		P("???\n");
> > > +		break;
> > > +	}
> > > +#undef P
> > > +}
> > > +EXPORT_SYMBOL_FOR_TESTS_ONLY(drm_dp_dump_sideband_msg_req_body);
> > > +
> > > +static inline void
> > > +drm_dp_mst_dump_sideband_msg_tx(struct drm_printer *p,
> > > +				const struct drm_dp_sideband_msg_tx *txmsg)
> > > +{
> > > +	struct drm_dp_sideband_msg_req_body req;
> > > +	char buf[64];
> > > +	int ret;
> > > +	int i;
> > > +
> > > +	drm_dp_mst_rad_to_str(txmsg->dst->rad, txmsg->dst->lct, buf,
> > > +			      sizeof(buf));
> > > +	drm_printf(p, "txmsg cur_offset=%x cur_len=%x seqno=%x state=%s
> > > path_msg=%d dst=%s\n",
> > > +		   txmsg->cur_offset, txmsg->cur_len, txmsg->seqno,
> > > +		   drm_dp_mst_sideband_tx_state_str(txmsg->state),
> > > +		   txmsg->path_msg, buf);
> > > +
> > > +	ret = drm_dp_decode_sideband_req(txmsg, &req);
> > > +	if (ret) {
> > > +		drm_printf(p, "<failed to decode sideband req: %d>\n", ret);
> > > +		return;
> > > +	}
> > > +	drm_dp_dump_sideband_msg_req_body(&req, 1, p);
> > > +
> > > +	switch (req.req_type) {
> > > +	case DP_REMOTE_DPCD_WRITE:
> > > +		kfree(req.u.dpcd_write.bytes);
> > > +		break;
> > > +	case DP_REMOTE_I2C_READ:
> > > +		for (i = 0; i < req.u.i2c_read.num_transactions; i++)
> > > +			kfree(req.u.i2c_read.transactions[i].bytes);
> > > +		break;
> > > +	case DP_REMOTE_I2C_WRITE:
> > > +		kfree(req.u.i2c_write.bytes);
> > > +		break;
> > > +	}
> > > +}
> > >  
> > >  static void drm_dp_crc_sideband_chunk_req(u8 *msg, u8 len)
> > >  {
> > > @@ -893,6 +1178,11 @@ static int drm_dp_mst_wait_tx_reply(struct
> > > drm_dp_mst_branch *mstb,
> > >  		}
> > >  	}
> > >  out:
> > > +	if (unlikely(ret == -EIO && drm_debug & (DRM_UT_DP | DRM_UT_KMS))) {
> > 
> > Hm I'd only check for DRM_UT_DP here.
> > 
> > > +		struct drm_printer p = drm_debug_printer(DBG_PREFIX);
> > > +
> > > +		drm_dp_mst_dump_sideband_msg_tx(&p, txmsg);
> > > +	}
> > >  	mutex_unlock(&mgr->qlock);
> > >  
> > >  	return ret;
> > > @@ -1927,8 +2217,11 @@ static int process_single_tx_qlock(struct
> > > drm_dp_mst_topology_mgr *mgr,
> > >  	idx += tosend + 1;
> > >  
> > >  	ret = drm_dp_send_sideband_msg(mgr, up, chunk, idx);
> > > -	if (ret) {
> > > -		DRM_DEBUG_KMS("sideband msg failed to send\n");
> > > +	if (ret && unlikely(drm_debug & (DRM_UT_DP | DRM_UT_KMS))) {
> > 
> > Same.
> > 
> > > +		struct drm_printer p = drm_debug_printer(DBG_PREFIX);
> > > +
> > > +		drm_printf(&p, "sideband msg failed to send\n");
> > > +		drm_dp_mst_dump_sideband_msg_tx(&p, txmsg);
> > >  		return ret;
> > >  	}
> > >  
> > > @@ -1990,6 +2283,13 @@ static void drm_dp_queue_down_tx(struct
> > > drm_dp_mst_topology_mgr *mgr,
> > >  {
> > >  	mutex_lock(&mgr->qlock);
> > >  	list_add_tail(&txmsg->next, &mgr->tx_msg_downq);
> > > +
> > > +	if (unlikely(drm_debug & DRM_UT_DP)) {
> > 
> > Like you do here.
> > 
> > > +		struct drm_printer p = drm_debug_printer(DBG_PREFIX);
> > > +
> > > +		drm_dp_mst_dump_sideband_msg_tx(&p, txmsg);
> > > +	}
> > > +
> > >  	if (list_is_singular(&mgr->tx_msg_downq))
> > >  		process_single_down_tx_qlock(mgr);
> > >  	mutex_unlock(&mgr->qlock);
> > > diff --git a/drivers/gpu/drm/drm_dp_mst_topology_internal.h
> > > b/drivers/gpu/drm/drm_dp_mst_topology_internal.h
> > > new file mode 100644
> > > index 000000000000..eeda9a61c657
> > > --- /dev/null
> > > +++ b/drivers/gpu/drm/drm_dp_mst_topology_internal.h
> > > @@ -0,0 +1,24 @@
> > > +/* SPDX-License-Identifier: GPL-2.0-only
> > > + *
> > > + * Declarations for DP MST related functions which are only used in
> > > selftests
> > > + *
> > > + * Copyright © 2018 Red Hat
> > > + * Authors:
> > > + *     Lyude Paul <lyude@redhat.com>
> > > + */
> > > +
> > > +#ifndef _DRM_DP_MST_HELPER_INTERNAL_H_
> > > +#define _DRM_DP_MST_HELPER_INTERNAL_H_
> > > +
> > > +#include <drm/drm_dp_mst_helper.h>
> > > +
> > > +void
> > > +drm_dp_encode_sideband_req(const struct drm_dp_sideband_msg_req_body
> > > *req,
> > > +			   struct drm_dp_sideband_msg_tx *raw);
> > > +int drm_dp_decode_sideband_req(const struct drm_dp_sideband_msg_tx *raw,
> > > +			       struct drm_dp_sideband_msg_req_body *req);
> > > +void
> > > +drm_dp_dump_sideband_msg_req_body(const struct
> > > drm_dp_sideband_msg_req_body *req,
> > > +				  int indent, struct drm_printer *printer);
> > > +
> > > +#endif /* !_DRM_DP_MST_HELPER_INTERNAL_H_ */
> > > diff --git a/drivers/gpu/drm/selftests/drm_modeset_selftests.h
> > > b/drivers/gpu/drm/selftests/drm_modeset_selftests.h
> > > index dec3ee3ec96f..1898de0b4a4d 100644
> > > --- a/drivers/gpu/drm/selftests/drm_modeset_selftests.h
> > > +++ b/drivers/gpu/drm/selftests/drm_modeset_selftests.h
> > > @@ -33,3 +33,4 @@ selftest(damage_iter_damage_one_outside,
> > > igt_damage_iter_damage_one_outside)
> > >  selftest(damage_iter_damage_src_moved, igt_damage_iter_damage_src_moved)
> > >  selftest(damage_iter_damage_not_visible,
> > > igt_damage_iter_damage_not_visible)
> > >  selftest(dp_mst_calc_pbn_mode, igt_dp_mst_calc_pbn_mode)
> > > +selftest(dp_mst_sideband_msg_req_decode,
> > > igt_dp_mst_sideband_msg_req_decode)
> > > diff --git a/drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c
> > > b/drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c
> > > index 51b2486ec917..ceca89babd65 100644
> > > --- a/drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c
> > > +++ b/drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c
> > > @@ -8,6 +8,7 @@
> > >  #include <drm/drm_dp_mst_helper.h>
> > >  #include <drm/drm_print.h>
> > >  
> > > +#include "../drm_dp_mst_topology_internal.h"
> > >  #include "test-drm_modeset_common.h"
> > >  
> > >  int igt_dp_mst_calc_pbn_mode(void *ignored)
> > > @@ -34,3 +35,214 @@ int igt_dp_mst_calc_pbn_mode(void *ignored)
> > >  
> > >  	return 0;
> > >  }
> > > +
> > > +static bool
> > > +sideband_msg_req_equal(const struct drm_dp_sideband_msg_req_body *in,
> > > +		       const struct drm_dp_sideband_msg_req_body *out)
> > > +{
> > > +	const struct drm_dp_remote_i2c_read_tx *txin, *txout;
> > > +	int i;
> > > +
> > > +	if (in->req_type != out->req_type)
> > > +		return false;
> > > +
> > > +	switch (in->req_type) {
> > > +	case DP_ENUM_PATH_RESOURCES:
> > > +	case DP_POWER_UP_PHY:
> > > +	case DP_POWER_DOWN_PHY:
> > > +	case DP_ALLOCATE_PAYLOAD:
> > > +	case DP_QUERY_PAYLOAD:
> > > +	case DP_REMOTE_DPCD_READ:
> > > +		return memcmp(in, out, sizeof(*in)) == 0;
> > 
> > Just memcmp the entire thing for everyrone (we assume kzalloc anyway), and
> > then only have the additional checks for the few cases we need it? Would
> > slightly reduce the code.
> 
> Just realized also - this wouldn't work: remember that I2C_READ, DPCD_WRITE,
> and I2C_WRITE contain pointers which of course will be different even if the
> contents are the same. Unfortunately I don't think there's really a good
> solution for this, unless we want to just remove the pointers and store things
> within the down request struct itself

Uh ... disappoint :-/

Then at least put that into a comment, and maybe make this the default:
handler?
> 
> > 
> > > +
> > > +	case DP_REMOTE_I2C_READ:
> > > +#define IN in->u.i2c_read
> > > +#define OUT out->u.i2c_read
> > > +		if (IN.num_bytes_read != OUT.num_bytes_read ||
> > > +		    IN.num_transactions != OUT.num_transactions ||
> > > +		    IN.port_number != OUT.port_number ||
> > > +		    IN.read_i2c_device_id != OUT.read_i2c_device_id)
> > > +			return false;
> > > +
> > > +		for (i = 0; i < IN.num_transactions; i++) {
> > > +			txin = &IN.transactions[i];
> > > +			txout = &OUT.transactions[i];
> > > +
> > > +			if (txin->i2c_dev_id != txout->i2c_dev_id ||
> > > +			    txin->no_stop_bit != txout->no_stop_bit ||
> > > +			    txin->num_bytes != txout->num_bytes ||
> > > +			    txin->i2c_transaction_delay !=
> > > +			    txout->i2c_transaction_delay)
> > > +				return false;
> > > +
> > > +			if (memcmp(txin->bytes, txout->bytes,
> > > +				   txin->num_bytes) != 0)
> > > +				return false;
> > > +		}
> > > +		break;
> > > +#undef IN
> > > +#undef OUT
> > > +
> > > +	case DP_REMOTE_DPCD_WRITE:
> > > +#define IN in->u.dpcd_write
> > > +#define OUT out->u.dpcd_write
> > > +		if (IN.dpcd_address != OUT.dpcd_address ||
> > > +		    IN.num_bytes != OUT.num_bytes ||
> > > +		    IN.port_number != OUT.port_number)
> > > +			return false;
> > > +
> > > +		return memcmp(IN.bytes, OUT.bytes, IN.num_bytes) == 0;
> > > +#undef IN
> > > +#undef OUT
> > > +
> > > +	case DP_REMOTE_I2C_WRITE:
> > > +#define IN in->u.i2c_write
> > > +#define OUT out->u.i2c_write
> > > +		if (IN.port_number != OUT.port_number ||
> > > +		    IN.write_i2c_device_id != OUT.write_i2c_device_id ||
> > > +		    IN.num_bytes != OUT.num_bytes)
> > > +			return false;
> > > +
> > > +		return memcmp(IN.bytes, OUT.bytes, IN.num_bytes) == 0;
> > > +#undef IN
> > > +#undef OUT
> > > +	}
> > > +
> > > +	return true;
> > > +}
> > > +
> > > +static int
> > > +__sideband_msg_req_encode_decode(int line,
> > > +				 struct drm_dp_sideband_msg_req_body *in,
> > > +				 struct drm_dp_sideband_msg_req_body *out)
> > > +{
> > > +	struct drm_printer p = drm_err_printer(pr_fmt());
> > > +	struct drm_dp_sideband_msg_tx txmsg;
> > > +	int i, ret;
> > > +	bool eq;
> > > +
> > > +	drm_dp_encode_sideband_req(in, &txmsg);
> > > +	ret = drm_dp_decode_sideband_req(&txmsg, out);
> > > +	if (ret < 0) {
> > > +		pr_err(pr_fmt("Failed to decode sideband request @ line %d:
> > > %d\n"),
> > > +		       line, ret);
> > > +		return ret;
> > > +	}
> > > +
> > > +	eq = sideband_msg_req_equal(in, out);
> > > +	if (!eq) {
> > > +		pr_err(pr_fmt("Encode/decode @ line %d failed, expected:\n"),
> > > +		       line);
> > > +		drm_dp_dump_sideband_msg_req_body(in, 1, &p);
> > > +		pr_err(pr_fmt("Got:\n"));
> > > +		drm_dp_dump_sideband_msg_req_body(out, 1, &p);
> > > +	}
> > > +
> > > +	switch (in->req_type) {
> > > +	case DP_REMOTE_DPCD_WRITE:
> > > +		kfree(out->u.dpcd_write.bytes);
> > > +		break;
> > > +	case DP_REMOTE_I2C_READ:
> > > +		for (i = 0; i < out->u.i2c_read.num_transactions; i++)
> > > +			kfree(out->u.i2c_read.transactions[i].bytes);
> > > +		break;
> > > +	case DP_REMOTE_I2C_WRITE:
> > > +		kfree(out->u.i2c_write.bytes);
> > > +		break;
> > > +	}
> > > +
> > > +	/* Clear everything but the req_type for the input */
> > > +	memset(&in->u, 0, sizeof(in->u));
> > > +
> > > +	/* Clear the output entirely */
> > > +	memset(out, 0, sizeof(*out));
> > > +
> > > +	return eq ? 0 : -EINVAL;
> > > +}
> > > +
> > > +int igt_dp_mst_sideband_msg_req_decode(void *unused)
> > > +{
> > > +	struct drm_dp_sideband_msg_req_body in = { 0 }, out = { 0 };
> > > +	u8 data[] = { 0xff, 0x0, 0xdd };
> > > +	int i;
> > > +
> > > +#define
> > > DO_TEST()                                                          \
> > > +	do {                                                               \
> > > +		if (__sideband_msg_req_encode_decode(__LINE__, &in, &out)) \
> > > +			return -EINVAL;                                    \
> > > +	} while (0)
> > 
> > I had a tiny wtf moment here until I realized this is a macro ... maybe
> > put the do on the first line and indent everything? Pure bikeshed to help
> > the blind ...
> > 
> > > +
> > > +	in.req_type = DP_ENUM_PATH_RESOURCES;
> > > +	in.u.port_num.port_number = 5;
> > > +	DO_TEST();
> > > +
> > > +	in.req_type = DP_POWER_UP_PHY;
> > > +	in.u.port_num.port_number = 5;
> > > +	DO_TEST();
> > > +
> > > +	in.req_type = DP_POWER_DOWN_PHY;
> > > +	in.u.port_num.port_number = 5;
> > > +	DO_TEST();
> > > +
> > > +	in.req_type = DP_ALLOCATE_PAYLOAD;
> > > +	in.u.allocate_payload.number_sdp_streams = 3;
> > > +	for (i = 0; i < in.u.allocate_payload.number_sdp_streams; i++)
> > > +		in.u.allocate_payload.sdp_stream_sink[i] = i + 1;
> > > +	DO_TEST();
> > > +	in.u.allocate_payload.port_number = 0xf;
> > > +	DO_TEST();
> > > +	in.u.allocate_payload.vcpi = 0x7f;
> > > +	DO_TEST();
> > > +	in.u.allocate_payload.pbn = U16_MAX;
> > > +	DO_TEST();
> > > +
> > > +	in.req_type = DP_QUERY_PAYLOAD;
> > > +	in.u.query_payload.port_number = 0xf;
> > > +	DO_TEST();
> > > +	in.u.query_payload.vcpi = 0x7f;
> > > +	DO_TEST();
> > > +
> > > +	in.req_type = DP_REMOTE_DPCD_READ;
> > > +	in.u.dpcd_read.port_number = 0xf;
> > > +	DO_TEST();
> > > +	in.u.dpcd_read.dpcd_address = 0xfedcb;
> > > +	DO_TEST();
> > > +	in.u.dpcd_read.num_bytes = U8_MAX;
> > > +	DO_TEST();
> > > +
> > > +	in.req_type = DP_REMOTE_DPCD_WRITE;
> > > +	in.u.dpcd_write.port_number = 0xf;
> > > +	DO_TEST();
> > > +	in.u.dpcd_write.dpcd_address = 0xfedcb;
> > > +	DO_TEST();
> > > +	in.u.dpcd_write.num_bytes = ARRAY_SIZE(data);
> > > +	in.u.dpcd_write.bytes = data;
> > > +	DO_TEST();
> > > +
> > > +	in.req_type = DP_REMOTE_I2C_READ;
> > > +	in.u.i2c_read.port_number = 0xf;
> > > +	DO_TEST();
> > > +	in.u.i2c_read.read_i2c_device_id = 0x7f;
> > > +	DO_TEST();
> > > +	in.u.i2c_read.num_transactions = 3;
> > > +	in.u.i2c_read.num_bytes_read = ARRAY_SIZE(data) * 3;
> > > +	for (i = 0; i < in.u.i2c_read.num_transactions; i++) {
> > > +		in.u.i2c_read.transactions[i].bytes = data;
> > > +		in.u.i2c_read.transactions[i].num_bytes = ARRAY_SIZE(data);
> > > +		in.u.i2c_read.transactions[i].i2c_dev_id = 0x7f & ~i;
> > > +		in.u.i2c_read.transactions[i].i2c_transaction_delay = 0xf &
> > > ~i;
> > > +	}
> > > +	DO_TEST();
> > > +
> > > +	in.req_type = DP_REMOTE_I2C_WRITE;
> > > +	in.u.i2c_write.port_number = 0xf;
> > > +	DO_TEST();
> > > +	in.u.i2c_write.write_i2c_device_id = 0x7f;
> > > +	DO_TEST();
> > > +	in.u.i2c_write.num_bytes = ARRAY_SIZE(data);
> > > +	in.u.i2c_write.bytes = data;
> > > +	DO_TEST();
> > > +
> > > +#undef DO_TEST
> > > +	return 0;
> > > +}
> > 
> > Extremely nice, more unit tests ftw!
> > 
> > With the nits somehow figured out: Reviewed-by: Daniel Vetter <
> > daniel.vetter@ffwll.ch>
> > 
> > > diff --git a/drivers/gpu/drm/selftests/test-drm_modeset_common.h
> > > b/drivers/gpu/drm/selftests/test-drm_modeset_common.h
> > > index 590bda35a683..0fcb8bbc6a1b 100644
> > > --- a/drivers/gpu/drm/selftests/test-drm_modeset_common.h
> > > +++ b/drivers/gpu/drm/selftests/test-drm_modeset_common.h
> > > @@ -40,5 +40,6 @@ int igt_damage_iter_damage_one_outside(void *ignored);
> > >  int igt_damage_iter_damage_src_moved(void *ignored);
> > >  int igt_damage_iter_damage_not_visible(void *ignored);
> > >  int igt_dp_mst_calc_pbn_mode(void *ignored);
> > > +int igt_dp_mst_sideband_msg_req_decode(void *ignored);
> > >  
> > >  #endif
> > > diff --git a/include/drm/drm_dp_mst_helper.h
> > > b/include/drm/drm_dp_mst_helper.h
> > > index c01f3ea72756..4c8d177e83e5 100644
> > > --- a/include/drm/drm_dp_mst_helper.h
> > > +++ b/include/drm/drm_dp_mst_helper.h
> > > @@ -293,7 +293,7 @@ struct drm_dp_remote_dpcd_write {
> > >  struct drm_dp_remote_i2c_read {
> > >  	u8 num_transactions;
> > >  	u8 port_number;
> > > -	struct {
> > > +	struct drm_dp_remote_i2c_read_tx {
> > >  		u8 i2c_dev_id;
> > >  		u8 num_bytes;
> > >  		u8 *bytes;
> > > -- 
> > > 2.21.0
> > > 
> -- 
> Cheers,
> 	Lyude Paul
> 

-- 
Daniel Vetter
Software Engineer, Intel Corporation
http://blog.ffwll.ch

Re: [PATCH 05/26] drm/dp_mst: Add sideband down request tracing + selftests

[Lyude Paul]

Two things below

On Tue, 2019-08-27 at 19:15 +0200, Daniel Vetter wrote:
> On Tue, Aug 27, 2019 at 12:49:17PM -0400, Lyude Paul wrote:
> > On Tue, 2019-08-13 at 16:50 +0200, Daniel Vetter wrote:
> > > On Wed, Jul 17, 2019 at 09:42:28PM -0400, Lyude Paul wrote:
> > > > Unfortunately the DP MST helpers do not have much in the way of
> > > > debugging utilities. So, let's add some!
> > > > 
> > > > This adds basic debugging output for down sideband requests that we
> > > > send
> > > > from the driver, so that we can actually discern what's happening when
> > > > sideband requests timeout. Note that with this commit, we'll be
> > > > dumping
> > > > out sideband requests under both of the following conditions:
> > > > 
> > > > - When the user has enabled DRM_UT_DP output, of course.
> > > > - When the user has enabled DRM_UT_KMS or DRM_UT_DP, and a sideband
> > > >   request has failed in some way. This will allow for developers to
> > > > get
> > > >   a basic idea of what's actually happening with failed modesets on
> > > > MST,
> > > >   without needing to have DRM_UT_DP explicitly enabled.
> > > > 
> > > > Since there wasn't really a good way of testing that any of this
> > > > worked,
> > > > I ended up writing simple selftests that lightly test sideband message
> > > > encoding and decoding as well. Enjoy!
> > > > 
> > > > Cc: Juston Li <juston.li@intel.com>
> > > > Cc: Imre Deak <imre.deak@intel.com>
> > > > Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
> > > > Cc: Harry Wentland <hwentlan@amd.com>
> > > > Signed-off-by: Lyude Paul <lyude@redhat.com>
> > > > ---
> > > >  drivers/gpu/drm/drm_dp_mst_topology.c         | 308
> > > > +++++++++++++++++-
> > > >  .../gpu/drm/drm_dp_mst_topology_internal.h    |  24 ++
> > > >  .../gpu/drm/selftests/drm_modeset_selftests.h |   1 +
> > > >  .../drm/selftests/test-drm_dp_mst_helper.c    | 212 ++++++++++++
> > > >  .../drm/selftests/test-drm_modeset_common.h   |   1 +
> > > >  include/drm/drm_dp_mst_helper.h               |   2 +-
> > > >  6 files changed, 543 insertions(+), 5 deletions(-)
> > > >  create mode 100644 drivers/gpu/drm/drm_dp_mst_topology_internal.h
> > > > 
> > > > diff --git a/drivers/gpu/drm/drm_dp_mst_topology.c
> > > > b/drivers/gpu/drm/drm_dp_mst_topology.c
> > > > index 9e382117896d..defc5e09fb9a 100644
> > > > --- a/drivers/gpu/drm/drm_dp_mst_topology.c
> > > > +++ b/drivers/gpu/drm/drm_dp_mst_topology.c
> > > > @@ -36,6 +36,8 @@
> > > >  #include <drm/drm_print.h>
> > > >  #include <drm/drm_probe_helper.h>
> > > >  
> > > > +#include "drm_dp_mst_topology_internal.h"
> > > > +
> > > >  /**
> > > >   * DOC: dp mst helper
> > > >   *
> > > > @@ -68,6 +70,8 @@ static int drm_dp_mst_register_i2c_bus(struct
> > > > drm_dp_aux
> > > > *aux);
> > > >  static void drm_dp_mst_unregister_i2c_bus(struct drm_dp_aux *aux);
> > > >  static void drm_dp_mst_kick_tx(struct drm_dp_mst_topology_mgr *mgr);
> > > >  
> > > > +#define DBG_PREFIX "[dp_mst]"
> > > > +
> > > >  #define DP_STR(x) [DP_ ## x] = #x
> > > >  
> > > >  static const char *drm_dp_mst_req_type_str(u8 req_type)
> > > > @@ -124,6 +128,43 @@ static const char *drm_dp_mst_nak_reason_str(u8
> > > > nak_reason)
> > > >  }
> > > >  
> > > >  #undef DP_STR
> > > > +#define DP_STR(x) [DRM_DP_SIDEBAND_TX_ ## x] = #x
> > > > +
> > > > +static const char *drm_dp_mst_sideband_tx_state_str(int state)
> > > > +{
> > > > +	static const char * const sideband_reason_str[] = {
> > > > +		DP_STR(QUEUED),
> > > > +		DP_STR(START_SEND),
> > > > +		DP_STR(SENT),
> > > > +		DP_STR(RX),
> > > > +		DP_STR(TIMEOUT),
> > > > +	};
> > > > +
> > > > +	if (state >= ARRAY_SIZE(sideband_reason_str) ||
> > > > +	    !sideband_reason_str[state])
> > > > +		return "unknown";
> > > > +
> > > > +	return sideband_reason_str[state];
> > > > +}
> > > > +
> > > > +static int
> > > > +drm_dp_mst_rad_to_str(const u8 rad[8], u8 lct, char *out, size_t len)
> > > > +{
> > > > +	int i;
> > > > +	u8 unpacked_rad[16];
> > > > +
> > > > +	for (i = 0; i < lct; i++) {
> > > > +		if (i % 2)
> > > > +			unpacked_rad[i] = rad[i / 2] >> 4;
> > > > +		else
> > > > +			unpacked_rad[i] = rad[i / 2] & BIT_MASK(4);
> > > > +	}
> > > > +
> > > > +	/* TODO: Eventually add something to printk so we can format
> > > > the rad
> > > > +	 * like this: 1.2.3
> > > > +	 */
> > > 
> > > 	lct *=2;
> > > 
> > > missing here? And yeah the todo would be sweet, but quite a bit more
> > > typing I guess.
> > > 
> > > > +	return snprintf(out, len, "%*phC", lct, unpacked_rad);
> > > > +}
> > > >  
> > > >  /* sideband msg handling */
> > > >  static u8 drm_dp_msg_header_crc4(const uint8_t *data, size_t
> > > > num_nibbles)
> > > > @@ -256,8 +297,9 @@ static bool drm_dp_decode_sideband_msg_hdr(struct
> > > > drm_dp_sideband_msg_hdr *hdr,
> > > >  	return true;
> > > >  }
> > > >  
> > > > -static void drm_dp_encode_sideband_req(struct
> > > > drm_dp_sideband_msg_req_body *req,
> > > > -				       struct drm_dp_sideband_msg_tx
> > > > *raw)
> > > > +void
> > > > +drm_dp_encode_sideband_req(const struct drm_dp_sideband_msg_req_body
> > > > *req,
> > > > +			   struct drm_dp_sideband_msg_tx *raw)
> > > >  {
> > > >  	int idx = 0;
> > > >  	int i;
> > > > @@ -362,6 +404,249 @@ static void drm_dp_encode_sideband_req(struct
> > > > drm_dp_sideband_msg_req_body *req,
> > > >  	}
> > > >  	raw->cur_len = idx;
> > > >  }
> > > > +EXPORT_SYMBOL_FOR_TESTS_ONLY(drm_dp_encode_sideband_req);
> > > > +
> > > > +/* Decode a sideband request we've encoded, mainly used for debugging
> > > > */
> > > > +int
> > > > +drm_dp_decode_sideband_req(const struct drm_dp_sideband_msg_tx *raw,
> > > > +			   struct drm_dp_sideband_msg_req_body *req)
> > > > +{
> > > > +	const u8 *buf = raw->msg;
> > > > +	int i, idx = 0;
> > > > +
> > > > +	req->req_type = buf[idx++] & 0x7f;
> > > > +	switch (req->req_type) {
> > > > +	case DP_ENUM_PATH_RESOURCES:
> > > > +	case DP_POWER_DOWN_PHY:
> > > > +	case DP_POWER_UP_PHY:
> > > 
> > > OCD warning: the enum isn't ordered the same way here and in
> > > drm_dp_encode_sideband_req().
> > > 
> > > > +		req->u.port_num.port_number = (buf[idx] >> 4) & 0xf;
> > > > +		break;
> > > > +	case DP_ALLOCATE_PAYLOAD:
> > > > +		{
> > > > +			struct drm_dp_allocate_payload *a =
> > > > +				&req->u.allocate_payload;
> > > > +
> > > > +			a->number_sdp_streams = buf[idx] & 0xf;
> > > > +			a->port_number = (buf[idx] >> 4) & 0xf;
> > > > +
> > > > +			a->vcpi = buf[++idx] & 0x7f;
> > > 
> > > Complain here if 0x80 is set?
> > > 
> > > > +
> > > > +			a->pbn = buf[++idx] << 8;
> > > > +			a->pbn |= buf[++idx];
> > > > +
> > > > +			idx++;
> > > > +			for (i = 0; i < a->number_sdp_streams; i++) {
> > > > +				a->sdp_stream_sink[i] =
> > > > +					(buf[idx + (i / 2)] >> ((i %
> > > > 2) ? 0 :
> > > > 4)) & 0xf;
> > > > +			}
> > > > +		}
> > > > +		break;
> > > > +	case DP_QUERY_PAYLOAD:
> > > > +		req->u.query_payload.port_number = (buf[idx] >> 4) &
> > > > 0xf;
> > > > +		req->u.query_payload.vcpi = buf[++idx] & 0x7f;
> > > 
> > > Same here for the highest bit?
> > > 
> > > > +		break;
> > > > +	case DP_REMOTE_DPCD_READ:
> > > > +		{
> > > > +			struct drm_dp_remote_dpcd_read *r = &req-
> > > > >u.dpcd_read;
> > > > +
> > > > +			r->port_number = (buf[idx] >> 4) & 0xf;
> > > > +
> > > > +			r->dpcd_address = (buf[idx] << 16) & 0xf0000;
> > > > +			r->dpcd_address |= (buf[++idx] << 8) & 0xff00;
> > > > +			r->dpcd_address |= buf[++idx] & 0xff;
> > > > +
> > > > +			r->num_bytes = buf[++idx];
> > > > +		}
> > > > +		break;
> > > > +	case DP_REMOTE_DPCD_WRITE:
> > > > +		{
> > > > +			struct drm_dp_remote_dpcd_write *w =
> > > > +				&req->u.dpcd_write;
> > > > +
> > > > +			w->port_number = (buf[idx] >> 4) & 0xf;
> > > > +
> > > > +			w->dpcd_address = (buf[idx] << 16) & 0xf0000;
> > > > +			w->dpcd_address |= (buf[++idx] << 8) & 0xff00;
> > > > +			w->dpcd_address |= buf[++idx] & 0xff;
> > > > +
> > > > +			w->num_bytes = buf[++idx];
> > > > +
> > > > +			w->bytes = kmemdup(&buf[++idx], w->num_bytes,
> > > > +					   GFP_KERNEL);
> > > 
> > > But if we go really strict on validation we'd need to make sure we don't
> > > walk past raw->cur_len ... probably not worth it?
> > > 
> > > > +			if (!w->bytes)
> > > > +				return -ENOMEM;
> > > > +		}
> > > > +		break;
> > > > +	case DP_REMOTE_I2C_READ:
> > > > +		{
> > > > +			struct drm_dp_remote_i2c_read *r = &req-
> > > > >u.i2c_read;
> > > > +			struct drm_dp_remote_i2c_read_tx *tx;
> > > > +			bool failed = false;
> > > > +
> > > > +			r->num_transactions = buf[idx] & 0x3;
> > > > +			r->port_number = (buf[idx] >> 4) & 0xf;
> > > > +			for (i = 0; i < r->num_transactions; i++) {
> > > > +				tx = &r->transactions[i];
> > > > +
> > > > +				tx->i2c_dev_id = buf[++idx] & 0x7f;
> > > > +				tx->num_bytes = buf[++idx];
> > > > +				tx->bytes = kmemdup(&buf[++idx],
> > > > +						    tx->num_bytes,
> > > > +						    GFP_KERNEL);
> > > > +				if (!tx->bytes) {
> > > > +					failed = true;
> > > > +					break;
> > > > +				}
> > > > +				idx += tx->num_bytes;
> > > > +				tx->no_stop_bit = (buf[idx] >> 5) &
> > > > 0x1;
> > > > +				tx->i2c_transaction_delay = buf[idx] &
> > > > 0xf;
> > > > +			}
> > > > +
> > > > +			if (failed) {
> > > > +				for (i = 0; i < r->num_transactions;
> > > > i++)
> > > > +					kfree(tx->bytes);
> > > > +				return -ENOMEM;
> > > > +			}
> > > > +
> > > > +			r->read_i2c_device_id = buf[++idx] & 0x7f;
> > > > +			r->num_bytes_read = buf[++idx];
> > > > +		}
> > > > +		break;
> > > > +	case DP_REMOTE_I2C_WRITE:
> > > > +		{
> > > > +			struct drm_dp_remote_i2c_write *w = &req-
> > > > >u.i2c_write;
> > > > +
> > > > +			w->port_number = (buf[idx] >> 4) & 0xf;
> > > > +			w->write_i2c_device_id = buf[++idx] & 0x7f;
> > > > +			w->num_bytes = buf[++idx];
> > > > +			w->bytes = kmemdup(&buf[++idx], w->num_bytes,
> > > > +					   GFP_KERNEL);
> > > > +			if (!w->bytes)
> > > > +				return -ENOMEM;
> > > > +		}
> > > > +		break;
> > > > +	}
> > > 
> > > For safety: Should we validate cur_len here? Or return it, and let the
> > > caller deal with any mismatches.
> > > 
> > > But I'm kinda leaning towards no safety here, since it's not dealing
> > > with
> > > any replies we get from the hw/externally, which might be used for
> > > attacking us. If we later on have the same thing but for
> > > drm_dp_sideband_parse_reply() then a lot more fun.
> > > 
> > > > +
> > > > +	return 0;
> > > > +}
> > > > +EXPORT_SYMBOL_FOR_TESTS_ONLY(drm_dp_decode_sideband_req);
> > > > +
> > > > +void
> > > > +drm_dp_dump_sideband_msg_req_body(const struct
> > > > drm_dp_sideband_msg_req_body *req,
> > > > +				  int indent, struct drm_printer
> > > > *printer)
> > > > +{
> > > > +	int i;
> > > > +
> > > > +#define P(f, ...) drm_printf_indent(printer, indent, f,
> > > > ##__VA_ARGS__)
> > > > +	if (req->req_type == DP_LINK_ADDRESS) {
> > > > +		/* No contents to print */
> > > > +		P("type=%s\n", drm_dp_mst_req_type_str(req-
> > > > >req_type));
> > > > +		return;
> > > > +	}
> > > > +
> > > > +	P("type=%s contents:\n", drm_dp_mst_req_type_str(req-
> > > > >req_type));
> > > > +	indent++;
> > > > +
> > > > +	switch (req->req_type) {
> > > > +	case DP_ENUM_PATH_RESOURCES:
> > > > +	case DP_POWER_DOWN_PHY:
> > > > +	case DP_POWER_UP_PHY:
> > > > +		P("port=%d\n", req->u.port_num.port_number);
> > > > +		break;
> > > > +	case DP_ALLOCATE_PAYLOAD:
> > > > +		P("port=%d vcpi=%d pbn=%d sdp_streams=%d %*ph\n",
> > > > +		  req->u.allocate_payload.port_number,
> > > > +		  req->u.allocate_payload.vcpi, req-
> > > > >u.allocate_payload.pbn,
> > > > +		  req->u.allocate_payload.number_sdp_streams,
> > > > +		  req->u.allocate_payload.number_sdp_streams,
> > > > +		  req->u.allocate_payload.sdp_stream_sink);
> > > > +		break;
> > > > +	case DP_QUERY_PAYLOAD:
> > > > +		P("port=%d vcpi=%d\n",
> > > > +		  req->u.query_payload.port_number,
> > > > +		  req->u.query_payload.vcpi);
> > > > +		break;
> > > > +	case DP_REMOTE_DPCD_READ:
> > > > +		P("port=%d dpcd_addr=%05x len=%d\n",
> > > > +		  req->u.dpcd_read.port_number, req-
> > > > >u.dpcd_read.dpcd_address,
> > > > +		  req->u.dpcd_read.num_bytes);
> > > > +		break;
> > > > +	case DP_REMOTE_DPCD_WRITE:
> > > > +		P("port=%d addr=%05x len=%d: %*ph\n",
> > > > +		  req->u.dpcd_write.port_number,
> > > > +		  req->u.dpcd_write.dpcd_address,
> > > > +		  req->u.dpcd_write.num_bytes, req-
> > > > >u.dpcd_write.num_bytes,
> > > > +		  req->u.dpcd_write.bytes);
> > > > +		break;
> > > > +	case DP_REMOTE_I2C_READ:
> > > > +		P("port=%d num_tx=%d id=%d size=%d:\n",
> > > > +		  req->u.i2c_read.port_number,
> > > > +		  req->u.i2c_read.num_transactions,
> > > > +		  req->u.i2c_read.read_i2c_device_id,
> > > > +		  req->u.i2c_read.num_bytes_read);
> > > > +
> > > > +		indent++;
> > > > +		for (i = 0; i < req->u.i2c_read.num_transactions; i++)
> > > > {
> > > > +			const struct drm_dp_remote_i2c_read_tx *rtx =
> > > > +				&req->u.i2c_read.transactions[i];
> > > > +
> > > > +			P("%d: id=%03d size=%03d no_stop_bit=%d
> > > > tx_delay=%03d:
> > > > %*ph\n",
> > > > +			  i, rtx->i2c_dev_id, rtx->num_bytes,
> > > > +			  rtx->no_stop_bit, rtx-
> > > > >i2c_transaction_delay,
> > > > +			  rtx->num_bytes, rtx->bytes);
> > > > +		}
> > > > +		break;
> > > > +	case DP_REMOTE_I2C_WRITE:
> > > > +		P("port=%d id=%d size=%d: %*ph\n",
> > > > +		  req->u.i2c_write.port_number,
> > > > +		  req->u.i2c_write.write_i2c_device_id,
> > > > +		  req->u.i2c_write.num_bytes, req-
> > > > >u.i2c_write.num_bytes,
> > > > +		  req->u.i2c_write.bytes);
> > > > +		break;
> > > > +	default:
> > > > +		P("???\n");
> > > > +		break;
> > > > +	}
> > > > +#undef P
> > > > +}
> > > > +EXPORT_SYMBOL_FOR_TESTS_ONLY(drm_dp_dump_sideband_msg_req_body);
> > > > +
> > > > +static inline void
> > > > +drm_dp_mst_dump_sideband_msg_tx(struct drm_printer *p,
> > > > +				const struct drm_dp_sideband_msg_tx
> > > > *txmsg)
> > > > +{
> > > > +	struct drm_dp_sideband_msg_req_body req;
> > > > +	char buf[64];
> > > > +	int ret;
> > > > +	int i;
> > > > +
> > > > +	drm_dp_mst_rad_to_str(txmsg->dst->rad, txmsg->dst->lct, buf,
> > > > +			      sizeof(buf));
> > > > +	drm_printf(p, "txmsg cur_offset=%x cur_len=%x seqno=%x
> > > > state=%s
> > > > path_msg=%d dst=%s\n",
> > > > +		   txmsg->cur_offset, txmsg->cur_len, txmsg->seqno,
> > > > +		   drm_dp_mst_sideband_tx_state_str(txmsg->state),
> > > > +		   txmsg->path_msg, buf);
> > > > +
> > > > +	ret = drm_dp_decode_sideband_req(txmsg, &req);
> > > > +	if (ret) {
> > > > +		drm_printf(p, "<failed to decode sideband req: %d>\n",
> > > > ret);
> > > > +		return;
> > > > +	}
> > > > +	drm_dp_dump_sideband_msg_req_body(&req, 1, p);
> > > > +
> > > > +	switch (req.req_type) {
> > > > +	case DP_REMOTE_DPCD_WRITE:
> > > > +		kfree(req.u.dpcd_write.bytes);
> > > > +		break;
> > > > +	case DP_REMOTE_I2C_READ:
> > > > +		for (i = 0; i < req.u.i2c_read.num_transactions; i++)
> > > > +			kfree(req.u.i2c_read.transactions[i].bytes);
> > > > +		break;
> > > > +	case DP_REMOTE_I2C_WRITE:
> > > > +		kfree(req.u.i2c_write.bytes);
> > > > +		break;
> > > > +	}
> > > > +}
> > > >  
> > > >  static void drm_dp_crc_sideband_chunk_req(u8 *msg, u8 len)
> > > >  {
> > > > @@ -893,6 +1178,11 @@ static int drm_dp_mst_wait_tx_reply(struct
> > > > drm_dp_mst_branch *mstb,
> > > >  		}
> > > >  	}
> > > >  out:
> > > > +	if (unlikely(ret == -EIO && drm_debug & (DRM_UT_DP |
> > > > DRM_UT_KMS))) {
> > > 
> > > Hm I'd only check for DRM_UT_DP here.

Was going to go ahead with this, but realized it might not be a great idea.
One of the reasons I went with DRM_UT_DP | DRM_UT_KMS here is because
generally, I'd assume that (this is the case for me at least) DRM_UT_DP serves
as a way to trace what's going on with DP, and DRM_UT_KMS is for non-atomic
specific modesetting debug info. That being said, if we fail sending a message
like PAYLOAD_ALLOCATE then the modeset itself is going to not work correctly
and thus - the contents of failure would still be relevant to someone working
on modesetting issues even if they're not interested in the entirity of what's
going on between the sink and the source.

That, and we were already printing when messages timed out under DRM_UT_KMS.
Granted though, that printing was added well before I introduced DRM_UT_DP.
However, I suppose I could just move that back to using DRM_DEBUG_KMS() and
dump the sideband message seperately.

Maybe I'm overthinking this a bit, thoughts?

> > > 
> > > > +		struct drm_printer p = drm_debug_printer(DBG_PREFIX);
> > > > +
> > > > +		drm_dp_mst_dump_sideband_msg_tx(&p, txmsg);
> > > > +	}
> > > >  	mutex_unlock(&mgr->qlock);
> > > >  
> > > >  	return ret;
> > > > @@ -1927,8 +2217,11 @@ static int process_single_tx_qlock(struct
> > > > drm_dp_mst_topology_mgr *mgr,
> > > >  	idx += tosend + 1;
> > > >  
> > > >  	ret = drm_dp_send_sideband_msg(mgr, up, chunk, idx);
> > > > -	if (ret) {
> > > > -		DRM_DEBUG_KMS("sideband msg failed to send\n");
> > > > +	if (ret && unlikely(drm_debug & (DRM_UT_DP | DRM_UT_KMS))) {
> > > 
> > > Same.
> > > 
> > > > +		struct drm_printer p = drm_debug_printer(DBG_PREFIX);
> > > > +
> > > > +		drm_printf(&p, "sideband msg failed to send\n");
> > > > +		drm_dp_mst_dump_sideband_msg_tx(&p, txmsg);
> > > >  		return ret;
> > > >  	}
> > > >  
> > > > @@ -1990,6 +2283,13 @@ static void drm_dp_queue_down_tx(struct
> > > > drm_dp_mst_topology_mgr *mgr,
> > > >  {
> > > >  	mutex_lock(&mgr->qlock);
> > > >  	list_add_tail(&txmsg->next, &mgr->tx_msg_downq);
> > > > +
> > > > +	if (unlikely(drm_debug & DRM_UT_DP)) {
> > > 
> > > Like you do here.
> > > 
> > > > +		struct drm_printer p = drm_debug_printer(DBG_PREFIX);
> > > > +
> > > > +		drm_dp_mst_dump_sideband_msg_tx(&p, txmsg);
> > > > +	}
> > > > +
> > > >  	if (list_is_singular(&mgr->tx_msg_downq))
> > > >  		process_single_down_tx_qlock(mgr);
> > > >  	mutex_unlock(&mgr->qlock);
> > > > diff --git a/drivers/gpu/drm/drm_dp_mst_topology_internal.h
> > > > b/drivers/gpu/drm/drm_dp_mst_topology_internal.h
> > > > new file mode 100644
> > > > index 000000000000..eeda9a61c657
> > > > --- /dev/null
> > > > +++ b/drivers/gpu/drm/drm_dp_mst_topology_internal.h
> > > > @@ -0,0 +1,24 @@
> > > > +/* SPDX-License-Identifier: GPL-2.0-only
> > > > + *
> > > > + * Declarations for DP MST related functions which are only used in
> > > > selftests
> > > > + *
> > > > + * Copyright © 2018 Red Hat
> > > > + * Authors:
> > > > + *     Lyude Paul <lyude@redhat.com>
> > > > + */
> > > > +
> > > > +#ifndef _DRM_DP_MST_HELPER_INTERNAL_H_
> > > > +#define _DRM_DP_MST_HELPER_INTERNAL_H_
> > > > +
> > > > +#include <drm/drm_dp_mst_helper.h>
> > > > +
> > > > +void
> > > > +drm_dp_encode_sideband_req(const struct drm_dp_sideband_msg_req_body
> > > > *req,
> > > > +			   struct drm_dp_sideband_msg_tx *raw);
> > > > +int drm_dp_decode_sideband_req(const struct drm_dp_sideband_msg_tx
> > > > *raw,
> > > > +			       struct drm_dp_sideband_msg_req_body
> > > > *req);
> > > > +void
> > > > +drm_dp_dump_sideband_msg_req_body(const struct
> > > > drm_dp_sideband_msg_req_body *req,
> > > > +				  int indent, struct drm_printer
> > > > *printer);
> > > > +
> > > > +#endif /* !_DRM_DP_MST_HELPER_INTERNAL_H_ */
> > > > diff --git a/drivers/gpu/drm/selftests/drm_modeset_selftests.h
> > > > b/drivers/gpu/drm/selftests/drm_modeset_selftests.h
> > > > index dec3ee3ec96f..1898de0b4a4d 100644
> > > > --- a/drivers/gpu/drm/selftests/drm_modeset_selftests.h
> > > > +++ b/drivers/gpu/drm/selftests/drm_modeset_selftests.h
> > > > @@ -33,3 +33,4 @@ selftest(damage_iter_damage_one_outside,
> > > > igt_damage_iter_damage_one_outside)
> > > >  selftest(damage_iter_damage_src_moved,
> > > > igt_damage_iter_damage_src_moved)
> > > >  selftest(damage_iter_damage_not_visible,
> > > > igt_damage_iter_damage_not_visible)
> > > >  selftest(dp_mst_calc_pbn_mode, igt_dp_mst_calc_pbn_mode)
> > > > +selftest(dp_mst_sideband_msg_req_decode,
> > > > igt_dp_mst_sideband_msg_req_decode)
> > > > diff --git a/drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c
> > > > b/drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c
> > > > index 51b2486ec917..ceca89babd65 100644
> > > > --- a/drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c
> > > > +++ b/drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c
> > > > @@ -8,6 +8,7 @@
> > > >  #include <drm/drm_dp_mst_helper.h>
> > > >  #include <drm/drm_print.h>
> > > >  
> > > > +#include "../drm_dp_mst_topology_internal.h"
> > > >  #include "test-drm_modeset_common.h"
> > > >  
> > > >  int igt_dp_mst_calc_pbn_mode(void *ignored)
> > > > @@ -34,3 +35,214 @@ int igt_dp_mst_calc_pbn_mode(void *ignored)
> > > >  
> > > >  	return 0;
> > > >  }
> > > > +
> > > > +static bool
> > > > +sideband_msg_req_equal(const struct drm_dp_sideband_msg_req_body *in,
> > > > +		       const struct drm_dp_sideband_msg_req_body *out)
> > > > +{
> > > > +	const struct drm_dp_remote_i2c_read_tx *txin, *txout;
> > > > +	int i;
> > > > +
> > > > +	if (in->req_type != out->req_type)
> > > > +		return false;
> > > > +
> > > > +	switch (in->req_type) {
> > > > +	case DP_ENUM_PATH_RESOURCES:
> > > > +	case DP_POWER_UP_PHY:
> > > > +	case DP_POWER_DOWN_PHY:
> > > > +	case DP_ALLOCATE_PAYLOAD:
> > > > +	case DP_QUERY_PAYLOAD:
> > > > +	case DP_REMOTE_DPCD_READ:
> > > > +		return memcmp(in, out, sizeof(*in)) == 0;
> > > 
> > > Just memcmp the entire thing for everyrone (we assume kzalloc anyway),
> > > and
> > > then only have the additional checks for the few cases we need it? Would
> > > slightly reduce the code.
> > 
> > Just realized also - this wouldn't work: remember that I2C_READ,
> > DPCD_WRITE,
> > and I2C_WRITE contain pointers which of course will be different even if
> > the
> > contents are the same. Unfortunately I don't think there's really a good
> > solution for this, unless we want to just remove the pointers and store
> > things
> > within the down request struct itself
> 
> Uh ... disappoint :-/
> 
> Then at least put that into a comment, and maybe make this the default:
> handler?
> > > > +
> > > > +	case DP_REMOTE_I2C_READ:
> > > > +#define IN in->u.i2c_read
> > > > +#define OUT out->u.i2c_read
> > > > +		if (IN.num_bytes_read != OUT.num_bytes_read ||
> > > > +		    IN.num_transactions != OUT.num_transactions ||
> > > > +		    IN.port_number != OUT.port_number ||
> > > > +		    IN.read_i2c_device_id != OUT.read_i2c_device_id)
> > > > +			return false;
> > > > +
> > > > +		for (i = 0; i < IN.num_transactions; i++) {
> > > > +			txin = &IN.transactions[i];
> > > > +			txout = &OUT.transactions[i];
> > > > +
> > > > +			if (txin->i2c_dev_id != txout->i2c_dev_id ||
> > > > +			    txin->no_stop_bit != txout->no_stop_bit ||
> > > > +			    txin->num_bytes != txout->num_bytes ||
> > > > +			    txin->i2c_transaction_delay !=
> > > > +			    txout->i2c_transaction_delay)
> > > > +				return false;
> > > > +
> > > > +			if (memcmp(txin->bytes, txout->bytes,
> > > > +				   txin->num_bytes) != 0)
> > > > +				return false;
> > > > +		}
> > > > +		break;
> > > > +#undef IN
> > > > +#undef OUT
> > > > +
> > > > +	case DP_REMOTE_DPCD_WRITE:
> > > > +#define IN in->u.dpcd_write
> > > > +#define OUT out->u.dpcd_write
> > > > +		if (IN.dpcd_address != OUT.dpcd_address ||
> > > > +		    IN.num_bytes != OUT.num_bytes ||
> > > > +		    IN.port_number != OUT.port_number)
> > > > +			return false;
> > > > +
> > > > +		return memcmp(IN.bytes, OUT.bytes, IN.num_bytes) == 0;
> > > > +#undef IN
> > > > +#undef OUT
> > > > +
> > > > +	case DP_REMOTE_I2C_WRITE:
> > > > +#define IN in->u.i2c_write
> > > > +#define OUT out->u.i2c_write
> > > > +		if (IN.port_number != OUT.port_number ||
> > > > +		    IN.write_i2c_device_id != OUT.write_i2c_device_id
> > > > ||
> > > > +		    IN.num_bytes != OUT.num_bytes)
> > > > +			return false;
> > > > +
> > > > +		return memcmp(IN.bytes, OUT.bytes, IN.num_bytes) == 0;
> > > > +#undef IN
> > > > +#undef OUT
> > > > +	}
> > > > +
> > > > +	return true;
> > > > +}
> > > > +
> > > > +static int
> > > > +__sideband_msg_req_encode_decode(int line,
> > > > +				 struct drm_dp_sideband_msg_req_body
> > > > *in,
> > > > +				 struct drm_dp_sideband_msg_req_body
> > > > *out)
> > > > +{
> > > > +	struct drm_printer p = drm_err_printer(pr_fmt());
> > > > +	struct drm_dp_sideband_msg_tx txmsg;
> > > > +	int i, ret;
> > > > +	bool eq;
> > > > +
> > > > +	drm_dp_encode_sideband_req(in, &txmsg);
> > > > +	ret = drm_dp_decode_sideband_req(&txmsg, out);
> > > > +	if (ret < 0) {
> > > > +		pr_err(pr_fmt("Failed to decode sideband request @
> > > > line %d:
> > > > %d\n"),
> > > > +		       line, ret);
> > > > +		return ret;
> > > > +	}
> > > > +
> > > > +	eq = sideband_msg_req_equal(in, out);
> > > > +	if (!eq) {
> > > > +		pr_err(pr_fmt("Encode/decode @ line %d failed,
> > > > expected:\n"),
> > > > +		       line);
> > > > +		drm_dp_dump_sideband_msg_req_body(in, 1, &p);
> > > > +		pr_err(pr_fmt("Got:\n"));
> > > > +		drm_dp_dump_sideband_msg_req_body(out, 1, &p);
> > > > +	}
> > > > +
> > > > +	switch (in->req_type) {
> > > > +	case DP_REMOTE_DPCD_WRITE:
> > > > +		kfree(out->u.dpcd_write.bytes);
> > > > +		break;
> > > > +	case DP_REMOTE_I2C_READ:
> > > > +		for (i = 0; i < out->u.i2c_read.num_transactions; i++)
> > > > +			kfree(out->u.i2c_read.transactions[i].bytes);
> > > > +		break;
> > > > +	case DP_REMOTE_I2C_WRITE:
> > > > +		kfree(out->u.i2c_write.bytes);
> > > > +		break;
> > > > +	}
> > > > +
> > > > +	/* Clear everything but the req_type for the input */
> > > > +	memset(&in->u, 0, sizeof(in->u));
> > > > +
> > > > +	/* Clear the output entirely */
> > > > +	memset(out, 0, sizeof(*out));
> > > > +
> > > > +	return eq ? 0 : -EINVAL;
> > > > +}
> > > > +
> > > > +int igt_dp_mst_sideband_msg_req_decode(void *unused)
> > > > +{
> > > > +	struct drm_dp_sideband_msg_req_body in = { 0 }, out = { 0 };
> > > > +	u8 data[] = { 0xff, 0x0, 0xdd };
> > > > +	int i;
> > > > +
> > > > +#define
> > > > DO_TEST()                                                          \
> > > > +	do
> > > > {                                                               \
> > > > +		if (__sideband_msg_req_encode_decode(__LINE__, &in,
> > > > &out)) \
> > > > +			return
> > > > -EINVAL;                                    \
> > > > +	} while (0)
> > > 
> > > I had a tiny wtf moment here until I realized this is a macro ... maybe
> > > put the do on the first line and indent everything? Pure bikeshed to
> > > help
> > > the blind ...
> > > 
> > > > +
> > > > +	in.req_type = DP_ENUM_PATH_RESOURCES;
> > > > +	in.u.port_num.port_number = 5;
> > > > +	DO_TEST();
> > > > +
> > > > +	in.req_type = DP_POWER_UP_PHY;
> > > > +	in.u.port_num.port_number = 5;
> > > > +	DO_TEST();
> > > > +
> > > > +	in.req_type = DP_POWER_DOWN_PHY;
> > > > +	in.u.port_num.port_number = 5;
> > > > +	DO_TEST();
> > > > +
> > > > +	in.req_type = DP_ALLOCATE_PAYLOAD;
> > > > +	in.u.allocate_payload.number_sdp_streams = 3;
> > > > +	for (i = 0; i < in.u.allocate_payload.number_sdp_streams; i++)
> > > > +		in.u.allocate_payload.sdp_stream_sink[i] = i + 1;
> > > > +	DO_TEST();
> > > > +	in.u.allocate_payload.port_number = 0xf;
> > > > +	DO_TEST();
> > > > +	in.u.allocate_payload.vcpi = 0x7f;
> > > > +	DO_TEST();
> > > > +	in.u.allocate_payload.pbn = U16_MAX;
> > > > +	DO_TEST();
> > > > +
> > > > +	in.req_type = DP_QUERY_PAYLOAD;
> > > > +	in.u.query_payload.port_number = 0xf;
> > > > +	DO_TEST();
> > > > +	in.u.query_payload.vcpi = 0x7f;
> > > > +	DO_TEST();
> > > > +
> > > > +	in.req_type = DP_REMOTE_DPCD_READ;
> > > > +	in.u.dpcd_read.port_number = 0xf;
> > > > +	DO_TEST();
> > > > +	in.u.dpcd_read.dpcd_address = 0xfedcb;
> > > > +	DO_TEST();
> > > > +	in.u.dpcd_read.num_bytes = U8_MAX;
> > > > +	DO_TEST();
> > > > +
> > > > +	in.req_type = DP_REMOTE_DPCD_WRITE;
> > > > +	in.u.dpcd_write.port_number = 0xf;
> > > > +	DO_TEST();
> > > > +	in.u.dpcd_write.dpcd_address = 0xfedcb;
> > > > +	DO_TEST();
> > > > +	in.u.dpcd_write.num_bytes = ARRAY_SIZE(data);
> > > > +	in.u.dpcd_write.bytes = data;
> > > > +	DO_TEST();
> > > > +
> > > > +	in.req_type = DP_REMOTE_I2C_READ;
> > > > +	in.u.i2c_read.port_number = 0xf;
> > > > +	DO_TEST();
> > > > +	in.u.i2c_read.read_i2c_device_id = 0x7f;
> > > > +	DO_TEST();
> > > > +	in.u.i2c_read.num_transactions = 3;
> > > > +	in.u.i2c_read.num_bytes_read = ARRAY_SIZE(data) * 3;
> > > > +	for (i = 0; i < in.u.i2c_read.num_transactions; i++) {
> > > > +		in.u.i2c_read.transactions[i].bytes = data;
> > > > +		in.u.i2c_read.transactions[i].num_bytes =
> > > > ARRAY_SIZE(data);
> > > > +		in.u.i2c_read.transactions[i].i2c_dev_id = 0x7f & ~i;
> > > > +		in.u.i2c_read.transactions[i].i2c_transaction_delay =
> > > > 0xf &
> > > > ~i;
> > > > +	}
> > > > +	DO_TEST();
> > > > +
> > > > +	in.req_type = DP_REMOTE_I2C_WRITE;
> > > > +	in.u.i2c_write.port_number = 0xf;
> > > > +	DO_TEST();
> > > > +	in.u.i2c_write.write_i2c_device_id = 0x7f;
> > > > +	DO_TEST();
> > > > +	in.u.i2c_write.num_bytes = ARRAY_SIZE(data);
> > > > +	in.u.i2c_write.bytes = data;
> > > > +	DO_TEST();
> > > > +
> > > > +#undef DO_TEST
> > > > +	return 0;
> > > > +}
> > > 
> > > Extremely nice, more unit tests ftw!
> > > 
> > > With the nits somehow figured out: Reviewed-by: Daniel Vetter <
> > > daniel.vetter@ffwll.ch>
> > > 
> > > > diff --git a/drivers/gpu/drm/selftests/test-drm_modeset_common.h
> > > > b/drivers/gpu/drm/selftests/test-drm_modeset_common.h
> > > > index 590bda35a683..0fcb8bbc6a1b 100644
> > > > --- a/drivers/gpu/drm/selftests/test-drm_modeset_common.h
> > > > +++ b/drivers/gpu/drm/selftests/test-drm_modeset_common.h
> > > > @@ -40,5 +40,6 @@ int igt_damage_iter_damage_one_outside(void
> > > > *ignored);
> > > >  int igt_damage_iter_damage_src_moved(void *ignored);
> > > >  int igt_damage_iter_damage_not_visible(void *ignored);
> > > >  int igt_dp_mst_calc_pbn_mode(void *ignored);
> > > > +int igt_dp_mst_sideband_msg_req_decode(void *ignored);
> > > >  
> > > >  #endif
> > > > diff --git a/include/drm/drm_dp_mst_helper.h
> > > > b/include/drm/drm_dp_mst_helper.h
> > > > index c01f3ea72756..4c8d177e83e5 100644
> > > > --- a/include/drm/drm_dp_mst_helper.h
> > > > +++ b/include/drm/drm_dp_mst_helper.h
> > > > @@ -293,7 +293,7 @@ struct drm_dp_remote_dpcd_write {
> > > >  struct drm_dp_remote_i2c_read {
> > > >  	u8 num_transactions;
> > > >  	u8 port_number;
> > > > -	struct {
> > > > +	struct drm_dp_remote_i2c_read_tx {
> > > >  		u8 i2c_dev_id;
> > > >  		u8 num_bytes;
> > > >  		u8 *bytes;
> > > > -- 
> > > > 2.21.0
> > > > 
> > -- 
> > Cheers,
> > 	Lyude Paul
> > 
-- 
Cheers,
	Lyude Paul

Re: [PATCH 05/26] drm/dp_mst: Add sideband down request tracing + selftests

[Daniel Vetter]

On Fri, Aug 30, 2019 at 08:31:06PM -0400, Lyude Paul wrote:
> Two things below
> 
> On Tue, 2019-08-27 at 19:15 +0200, Daniel Vetter wrote:
> > On Tue, Aug 27, 2019 at 12:49:17PM -0400, Lyude Paul wrote:
> > > On Tue, 2019-08-13 at 16:50 +0200, Daniel Vetter wrote:
> > > > On Wed, Jul 17, 2019 at 09:42:28PM -0400, Lyude Paul wrote:
> > > > > Unfortunately the DP MST helpers do not have much in the way of
> > > > > debugging utilities. So, let's add some!
> > > > > 
> > > > > This adds basic debugging output for down sideband requests that we
> > > > > send
> > > > > from the driver, so that we can actually discern what's happening when
> > > > > sideband requests timeout. Note that with this commit, we'll be
> > > > > dumping
> > > > > out sideband requests under both of the following conditions:
> > > > > 
> > > > > - When the user has enabled DRM_UT_DP output, of course.
> > > > > - When the user has enabled DRM_UT_KMS or DRM_UT_DP, and a sideband
> > > > >   request has failed in some way. This will allow for developers to
> > > > > get
> > > > >   a basic idea of what's actually happening with failed modesets on
> > > > > MST,
> > > > >   without needing to have DRM_UT_DP explicitly enabled.
> > > > > 
> > > > > Since there wasn't really a good way of testing that any of this
> > > > > worked,
> > > > > I ended up writing simple selftests that lightly test sideband message
> > > > > encoding and decoding as well. Enjoy!
> > > > > 
> > > > > Cc: Juston Li <juston.li@intel.com>
> > > > > Cc: Imre Deak <imre.deak@intel.com>
> > > > > Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
> > > > > Cc: Harry Wentland <hwentlan@amd.com>
> > > > > Signed-off-by: Lyude Paul <lyude@redhat.com>
> > > > > ---
> > > > >  drivers/gpu/drm/drm_dp_mst_topology.c         | 308
> > > > > +++++++++++++++++-
> > > > >  .../gpu/drm/drm_dp_mst_topology_internal.h    |  24 ++
> > > > >  .../gpu/drm/selftests/drm_modeset_selftests.h |   1 +
> > > > >  .../drm/selftests/test-drm_dp_mst_helper.c    | 212 ++++++++++++
> > > > >  .../drm/selftests/test-drm_modeset_common.h   |   1 +
> > > > >  include/drm/drm_dp_mst_helper.h               |   2 +-
> > > > >  6 files changed, 543 insertions(+), 5 deletions(-)
> > > > >  create mode 100644 drivers/gpu/drm/drm_dp_mst_topology_internal.h
> > > > > 
> > > > > diff --git a/drivers/gpu/drm/drm_dp_mst_topology.c
> > > > > b/drivers/gpu/drm/drm_dp_mst_topology.c
> > > > > index 9e382117896d..defc5e09fb9a 100644
> > > > > --- a/drivers/gpu/drm/drm_dp_mst_topology.c
> > > > > +++ b/drivers/gpu/drm/drm_dp_mst_topology.c
> > > > > @@ -36,6 +36,8 @@
> > > > >  #include <drm/drm_print.h>
> > > > >  #include <drm/drm_probe_helper.h>
> > > > >  
> > > > > +#include "drm_dp_mst_topology_internal.h"
> > > > > +
> > > > >  /**
> > > > >   * DOC: dp mst helper
> > > > >   *
> > > > > @@ -68,6 +70,8 @@ static int drm_dp_mst_register_i2c_bus(struct
> > > > > drm_dp_aux
> > > > > *aux);
> > > > >  static void drm_dp_mst_unregister_i2c_bus(struct drm_dp_aux *aux);
> > > > >  static void drm_dp_mst_kick_tx(struct drm_dp_mst_topology_mgr *mgr);
> > > > >  
> > > > > +#define DBG_PREFIX "[dp_mst]"
> > > > > +
> > > > >  #define DP_STR(x) [DP_ ## x] = #x
> > > > >  
> > > > >  static const char *drm_dp_mst_req_type_str(u8 req_type)
> > > > > @@ -124,6 +128,43 @@ static const char *drm_dp_mst_nak_reason_str(u8
> > > > > nak_reason)
> > > > >  }
> > > > >  
> > > > >  #undef DP_STR
> > > > > +#define DP_STR(x) [DRM_DP_SIDEBAND_TX_ ## x] = #x
> > > > > +
> > > > > +static const char *drm_dp_mst_sideband_tx_state_str(int state)
> > > > > +{
> > > > > +	static const char * const sideband_reason_str[] = {
> > > > > +		DP_STR(QUEUED),
> > > > > +		DP_STR(START_SEND),
> > > > > +		DP_STR(SENT),
> > > > > +		DP_STR(RX),
> > > > > +		DP_STR(TIMEOUT),
> > > > > +	};
> > > > > +
> > > > > +	if (state >= ARRAY_SIZE(sideband_reason_str) ||
> > > > > +	    !sideband_reason_str[state])
> > > > > +		return "unknown";
> > > > > +
> > > > > +	return sideband_reason_str[state];
> > > > > +}
> > > > > +
> > > > > +static int
> > > > > +drm_dp_mst_rad_to_str(const u8 rad[8], u8 lct, char *out, size_t len)
> > > > > +{
> > > > > +	int i;
> > > > > +	u8 unpacked_rad[16];
> > > > > +
> > > > > +	for (i = 0; i < lct; i++) {
> > > > > +		if (i % 2)
> > > > > +			unpacked_rad[i] = rad[i / 2] >> 4;
> > > > > +		else
> > > > > +			unpacked_rad[i] = rad[i / 2] & BIT_MASK(4);
> > > > > +	}
> > > > > +
> > > > > +	/* TODO: Eventually add something to printk so we can format
> > > > > the rad
> > > > > +	 * like this: 1.2.3
> > > > > +	 */
> > > > 
> > > > 	lct *=2;
> > > > 
> > > > missing here? And yeah the todo would be sweet, but quite a bit more
> > > > typing I guess.
> > > > 
> > > > > +	return snprintf(out, len, "%*phC", lct, unpacked_rad);
> > > > > +}
> > > > >  
> > > > >  /* sideband msg handling */
> > > > >  static u8 drm_dp_msg_header_crc4(const uint8_t *data, size_t
> > > > > num_nibbles)
> > > > > @@ -256,8 +297,9 @@ static bool drm_dp_decode_sideband_msg_hdr(struct
> > > > > drm_dp_sideband_msg_hdr *hdr,
> > > > >  	return true;
> > > > >  }
> > > > >  
> > > > > -static void drm_dp_encode_sideband_req(struct
> > > > > drm_dp_sideband_msg_req_body *req,
> > > > > -				       struct drm_dp_sideband_msg_tx
> > > > > *raw)
> > > > > +void
> > > > > +drm_dp_encode_sideband_req(const struct drm_dp_sideband_msg_req_body
> > > > > *req,
> > > > > +			   struct drm_dp_sideband_msg_tx *raw)
> > > > >  {
> > > > >  	int idx = 0;
> > > > >  	int i;
> > > > > @@ -362,6 +404,249 @@ static void drm_dp_encode_sideband_req(struct
> > > > > drm_dp_sideband_msg_req_body *req,
> > > > >  	}
> > > > >  	raw->cur_len = idx;
> > > > >  }
> > > > > +EXPORT_SYMBOL_FOR_TESTS_ONLY(drm_dp_encode_sideband_req);
> > > > > +
> > > > > +/* Decode a sideband request we've encoded, mainly used for debugging
> > > > > */
> > > > > +int
> > > > > +drm_dp_decode_sideband_req(const struct drm_dp_sideband_msg_tx *raw,
> > > > > +			   struct drm_dp_sideband_msg_req_body *req)
> > > > > +{
> > > > > +	const u8 *buf = raw->msg;
> > > > > +	int i, idx = 0;
> > > > > +
> > > > > +	req->req_type = buf[idx++] & 0x7f;
> > > > > +	switch (req->req_type) {
> > > > > +	case DP_ENUM_PATH_RESOURCES:
> > > > > +	case DP_POWER_DOWN_PHY:
> > > > > +	case DP_POWER_UP_PHY:
> > > > 
> > > > OCD warning: the enum isn't ordered the same way here and in
> > > > drm_dp_encode_sideband_req().
> > > > 
> > > > > +		req->u.port_num.port_number = (buf[idx] >> 4) & 0xf;
> > > > > +		break;
> > > > > +	case DP_ALLOCATE_PAYLOAD:
> > > > > +		{
> > > > > +			struct drm_dp_allocate_payload *a =
> > > > > +				&req->u.allocate_payload;
> > > > > +
> > > > > +			a->number_sdp_streams = buf[idx] & 0xf;
> > > > > +			a->port_number = (buf[idx] >> 4) & 0xf;
> > > > > +
> > > > > +			a->vcpi = buf[++idx] & 0x7f;
> > > > 
> > > > Complain here if 0x80 is set?
> > > > 
> > > > > +
> > > > > +			a->pbn = buf[++idx] << 8;
> > > > > +			a->pbn |= buf[++idx];
> > > > > +
> > > > > +			idx++;
> > > > > +			for (i = 0; i < a->number_sdp_streams; i++) {
> > > > > +				a->sdp_stream_sink[i] =
> > > > > +					(buf[idx + (i / 2)] >> ((i %
> > > > > 2) ? 0 :
> > > > > 4)) & 0xf;
> > > > > +			}
> > > > > +		}
> > > > > +		break;
> > > > > +	case DP_QUERY_PAYLOAD:
> > > > > +		req->u.query_payload.port_number = (buf[idx] >> 4) &
> > > > > 0xf;
> > > > > +		req->u.query_payload.vcpi = buf[++idx] & 0x7f;
> > > > 
> > > > Same here for the highest bit?
> > > > 
> > > > > +		break;
> > > > > +	case DP_REMOTE_DPCD_READ:
> > > > > +		{
> > > > > +			struct drm_dp_remote_dpcd_read *r = &req-
> > > > > >u.dpcd_read;
> > > > > +
> > > > > +			r->port_number = (buf[idx] >> 4) & 0xf;
> > > > > +
> > > > > +			r->dpcd_address = (buf[idx] << 16) & 0xf0000;
> > > > > +			r->dpcd_address |= (buf[++idx] << 8) & 0xff00;
> > > > > +			r->dpcd_address |= buf[++idx] & 0xff;
> > > > > +
> > > > > +			r->num_bytes = buf[++idx];
> > > > > +		}
> > > > > +		break;
> > > > > +	case DP_REMOTE_DPCD_WRITE:
> > > > > +		{
> > > > > +			struct drm_dp_remote_dpcd_write *w =
> > > > > +				&req->u.dpcd_write;
> > > > > +
> > > > > +			w->port_number = (buf[idx] >> 4) & 0xf;
> > > > > +
> > > > > +			w->dpcd_address = (buf[idx] << 16) & 0xf0000;
> > > > > +			w->dpcd_address |= (buf[++idx] << 8) & 0xff00;
> > > > > +			w->dpcd_address |= buf[++idx] & 0xff;
> > > > > +
> > > > > +			w->num_bytes = buf[++idx];
> > > > > +
> > > > > +			w->bytes = kmemdup(&buf[++idx], w->num_bytes,
> > > > > +					   GFP_KERNEL);
> > > > 
> > > > But if we go really strict on validation we'd need to make sure we don't
> > > > walk past raw->cur_len ... probably not worth it?
> > > > 
> > > > > +			if (!w->bytes)
> > > > > +				return -ENOMEM;
> > > > > +		}
> > > > > +		break;
> > > > > +	case DP_REMOTE_I2C_READ:
> > > > > +		{
> > > > > +			struct drm_dp_remote_i2c_read *r = &req-
> > > > > >u.i2c_read;
> > > > > +			struct drm_dp_remote_i2c_read_tx *tx;
> > > > > +			bool failed = false;
> > > > > +
> > > > > +			r->num_transactions = buf[idx] & 0x3;
> > > > > +			r->port_number = (buf[idx] >> 4) & 0xf;
> > > > > +			for (i = 0; i < r->num_transactions; i++) {
> > > > > +				tx = &r->transactions[i];
> > > > > +
> > > > > +				tx->i2c_dev_id = buf[++idx] & 0x7f;
> > > > > +				tx->num_bytes = buf[++idx];
> > > > > +				tx->bytes = kmemdup(&buf[++idx],
> > > > > +						    tx->num_bytes,
> > > > > +						    GFP_KERNEL);
> > > > > +				if (!tx->bytes) {
> > > > > +					failed = true;
> > > > > +					break;
> > > > > +				}
> > > > > +				idx += tx->num_bytes;
> > > > > +				tx->no_stop_bit = (buf[idx] >> 5) &
> > > > > 0x1;
> > > > > +				tx->i2c_transaction_delay = buf[idx] &
> > > > > 0xf;
> > > > > +			}
> > > > > +
> > > > > +			if (failed) {
> > > > > +				for (i = 0; i < r->num_transactions;
> > > > > i++)
> > > > > +					kfree(tx->bytes);
> > > > > +				return -ENOMEM;
> > > > > +			}
> > > > > +
> > > > > +			r->read_i2c_device_id = buf[++idx] & 0x7f;
> > > > > +			r->num_bytes_read = buf[++idx];
> > > > > +		}
> > > > > +		break;
> > > > > +	case DP_REMOTE_I2C_WRITE:
> > > > > +		{
> > > > > +			struct drm_dp_remote_i2c_write *w = &req-
> > > > > >u.i2c_write;
> > > > > +
> > > > > +			w->port_number = (buf[idx] >> 4) & 0xf;
> > > > > +			w->write_i2c_device_id = buf[++idx] & 0x7f;
> > > > > +			w->num_bytes = buf[++idx];
> > > > > +			w->bytes = kmemdup(&buf[++idx], w->num_bytes,
> > > > > +					   GFP_KERNEL);
> > > > > +			if (!w->bytes)
> > > > > +				return -ENOMEM;
> > > > > +		}
> > > > > +		break;
> > > > > +	}
> > > > 
> > > > For safety: Should we validate cur_len here? Or return it, and let the
> > > > caller deal with any mismatches.
> > > > 
> > > > But I'm kinda leaning towards no safety here, since it's not dealing
> > > > with
> > > > any replies we get from the hw/externally, which might be used for
> > > > attacking us. If we later on have the same thing but for
> > > > drm_dp_sideband_parse_reply() then a lot more fun.
> > > > 
> > > > > +
> > > > > +	return 0;
> > > > > +}
> > > > > +EXPORT_SYMBOL_FOR_TESTS_ONLY(drm_dp_decode_sideband_req);
> > > > > +
> > > > > +void
> > > > > +drm_dp_dump_sideband_msg_req_body(const struct
> > > > > drm_dp_sideband_msg_req_body *req,
> > > > > +				  int indent, struct drm_printer
> > > > > *printer)
> > > > > +{
> > > > > +	int i;
> > > > > +
> > > > > +#define P(f, ...) drm_printf_indent(printer, indent, f,
> > > > > ##__VA_ARGS__)
> > > > > +	if (req->req_type == DP_LINK_ADDRESS) {
> > > > > +		/* No contents to print */
> > > > > +		P("type=%s\n", drm_dp_mst_req_type_str(req-
> > > > > >req_type));
> > > > > +		return;
> > > > > +	}
> > > > > +
> > > > > +	P("type=%s contents:\n", drm_dp_mst_req_type_str(req-
> > > > > >req_type));
> > > > > +	indent++;
> > > > > +
> > > > > +	switch (req->req_type) {
> > > > > +	case DP_ENUM_PATH_RESOURCES:
> > > > > +	case DP_POWER_DOWN_PHY:
> > > > > +	case DP_POWER_UP_PHY:
> > > > > +		P("port=%d\n", req->u.port_num.port_number);
> > > > > +		break;
> > > > > +	case DP_ALLOCATE_PAYLOAD:
> > > > > +		P("port=%d vcpi=%d pbn=%d sdp_streams=%d %*ph\n",
> > > > > +		  req->u.allocate_payload.port_number,
> > > > > +		  req->u.allocate_payload.vcpi, req-
> > > > > >u.allocate_payload.pbn,
> > > > > +		  req->u.allocate_payload.number_sdp_streams,
> > > > > +		  req->u.allocate_payload.number_sdp_streams,
> > > > > +		  req->u.allocate_payload.sdp_stream_sink);
> > > > > +		break;
> > > > > +	case DP_QUERY_PAYLOAD:
> > > > > +		P("port=%d vcpi=%d\n",
> > > > > +		  req->u.query_payload.port_number,
> > > > > +		  req->u.query_payload.vcpi);
> > > > > +		break;
> > > > > +	case DP_REMOTE_DPCD_READ:
> > > > > +		P("port=%d dpcd_addr=%05x len=%d\n",
> > > > > +		  req->u.dpcd_read.port_number, req-
> > > > > >u.dpcd_read.dpcd_address,
> > > > > +		  req->u.dpcd_read.num_bytes);
> > > > > +		break;
> > > > > +	case DP_REMOTE_DPCD_WRITE:
> > > > > +		P("port=%d addr=%05x len=%d: %*ph\n",
> > > > > +		  req->u.dpcd_write.port_number,
> > > > > +		  req->u.dpcd_write.dpcd_address,
> > > > > +		  req->u.dpcd_write.num_bytes, req-
> > > > > >u.dpcd_write.num_bytes,
> > > > > +		  req->u.dpcd_write.bytes);
> > > > > +		break;
> > > > > +	case DP_REMOTE_I2C_READ:
> > > > > +		P("port=%d num_tx=%d id=%d size=%d:\n",
> > > > > +		  req->u.i2c_read.port_number,
> > > > > +		  req->u.i2c_read.num_transactions,
> > > > > +		  req->u.i2c_read.read_i2c_device_id,
> > > > > +		  req->u.i2c_read.num_bytes_read);
> > > > > +
> > > > > +		indent++;
> > > > > +		for (i = 0; i < req->u.i2c_read.num_transactions; i++)
> > > > > {
> > > > > +			const struct drm_dp_remote_i2c_read_tx *rtx =
> > > > > +				&req->u.i2c_read.transactions[i];
> > > > > +
> > > > > +			P("%d: id=%03d size=%03d no_stop_bit=%d
> > > > > tx_delay=%03d:
> > > > > %*ph\n",
> > > > > +			  i, rtx->i2c_dev_id, rtx->num_bytes,
> > > > > +			  rtx->no_stop_bit, rtx-
> > > > > >i2c_transaction_delay,
> > > > > +			  rtx->num_bytes, rtx->bytes);
> > > > > +		}
> > > > > +		break;
> > > > > +	case DP_REMOTE_I2C_WRITE:
> > > > > +		P("port=%d id=%d size=%d: %*ph\n",
> > > > > +		  req->u.i2c_write.port_number,
> > > > > +		  req->u.i2c_write.write_i2c_device_id,
> > > > > +		  req->u.i2c_write.num_bytes, req-
> > > > > >u.i2c_write.num_bytes,
> > > > > +		  req->u.i2c_write.bytes);
> > > > > +		break;
> > > > > +	default:
> > > > > +		P("???\n");
> > > > > +		break;
> > > > > +	}
> > > > > +#undef P
> > > > > +}
> > > > > +EXPORT_SYMBOL_FOR_TESTS_ONLY(drm_dp_dump_sideband_msg_req_body);
> > > > > +
> > > > > +static inline void
> > > > > +drm_dp_mst_dump_sideband_msg_tx(struct drm_printer *p,
> > > > > +				const struct drm_dp_sideband_msg_tx
> > > > > *txmsg)
> > > > > +{
> > > > > +	struct drm_dp_sideband_msg_req_body req;
> > > > > +	char buf[64];
> > > > > +	int ret;
> > > > > +	int i;
> > > > > +
> > > > > +	drm_dp_mst_rad_to_str(txmsg->dst->rad, txmsg->dst->lct, buf,
> > > > > +			      sizeof(buf));
> > > > > +	drm_printf(p, "txmsg cur_offset=%x cur_len=%x seqno=%x
> > > > > state=%s
> > > > > path_msg=%d dst=%s\n",
> > > > > +		   txmsg->cur_offset, txmsg->cur_len, txmsg->seqno,
> > > > > +		   drm_dp_mst_sideband_tx_state_str(txmsg->state),
> > > > > +		   txmsg->path_msg, buf);
> > > > > +
> > > > > +	ret = drm_dp_decode_sideband_req(txmsg, &req);
> > > > > +	if (ret) {
> > > > > +		drm_printf(p, "<failed to decode sideband req: %d>\n",
> > > > > ret);
> > > > > +		return;
> > > > > +	}
> > > > > +	drm_dp_dump_sideband_msg_req_body(&req, 1, p);
> > > > > +
> > > > > +	switch (req.req_type) {
> > > > > +	case DP_REMOTE_DPCD_WRITE:
> > > > > +		kfree(req.u.dpcd_write.bytes);
> > > > > +		break;
> > > > > +	case DP_REMOTE_I2C_READ:
> > > > > +		for (i = 0; i < req.u.i2c_read.num_transactions; i++)
> > > > > +			kfree(req.u.i2c_read.transactions[i].bytes);
> > > > > +		break;
> > > > > +	case DP_REMOTE_I2C_WRITE:
> > > > > +		kfree(req.u.i2c_write.bytes);
> > > > > +		break;
> > > > > +	}
> > > > > +}
> > > > >  
> > > > >  static void drm_dp_crc_sideband_chunk_req(u8 *msg, u8 len)
> > > > >  {
> > > > > @@ -893,6 +1178,11 @@ static int drm_dp_mst_wait_tx_reply(struct
> > > > > drm_dp_mst_branch *mstb,
> > > > >  		}
> > > > >  	}
> > > > >  out:
> > > > > +	if (unlikely(ret == -EIO && drm_debug & (DRM_UT_DP |
> > > > > DRM_UT_KMS))) {
> > > > 
> > > > Hm I'd only check for DRM_UT_DP here.
> 
> Was going to go ahead with this, but realized it might not be a great idea.
> One of the reasons I went with DRM_UT_DP | DRM_UT_KMS here is because
> generally, I'd assume that (this is the case for me at least) DRM_UT_DP serves
> as a way to trace what's going on with DP, and DRM_UT_KMS is for non-atomic
> specific modesetting debug info. That being said, if we fail sending a message
> like PAYLOAD_ALLOCATE then the modeset itself is going to not work correctly
> and thus - the contents of failure would still be relevant to someone working
> on modesetting issues even if they're not interested in the entirity of what's
> going on between the sink and the source.
> 
> That, and we were already printing when messages timed out under DRM_UT_KMS.
> Granted though, that printing was added well before I introduced DRM_UT_DP.
> However, I suppose I could just move that back to using DRM_DEBUG_KMS() and
> dump the sideband message seperately.
> 
> Maybe I'm overthinking this a bit, thoughts?

Yes. Or at least, the drm debugging stuff we have is a mess already, let's
try to not make it even more fun. I think consistently using DP for this
stuff would make a lot more sense (and for general debugging I guess
everyone just gets to add the DP bit to their drm.debug option).
-Daniel

> 
> > > > 
> > > > > +		struct drm_printer p = drm_debug_printer(DBG_PREFIX);
> > > > > +
> > > > > +		drm_dp_mst_dump_sideband_msg_tx(&p, txmsg);
> > > > > +	}
> > > > >  	mutex_unlock(&mgr->qlock);
> > > > >  
> > > > >  	return ret;
> > > > > @@ -1927,8 +2217,11 @@ static int process_single_tx_qlock(struct
> > > > > drm_dp_mst_topology_mgr *mgr,
> > > > >  	idx += tosend + 1;
> > > > >  
> > > > >  	ret = drm_dp_send_sideband_msg(mgr, up, chunk, idx);
> > > > > -	if (ret) {
> > > > > -		DRM_DEBUG_KMS("sideband msg failed to send\n");
> > > > > +	if (ret && unlikely(drm_debug & (DRM_UT_DP | DRM_UT_KMS))) {
> > > > 
> > > > Same.
> > > > 
> > > > > +		struct drm_printer p = drm_debug_printer(DBG_PREFIX);
> > > > > +
> > > > > +		drm_printf(&p, "sideband msg failed to send\n");
> > > > > +		drm_dp_mst_dump_sideband_msg_tx(&p, txmsg);
> > > > >  		return ret;
> > > > >  	}
> > > > >  
> > > > > @@ -1990,6 +2283,13 @@ static void drm_dp_queue_down_tx(struct
> > > > > drm_dp_mst_topology_mgr *mgr,
> > > > >  {
> > > > >  	mutex_lock(&mgr->qlock);
> > > > >  	list_add_tail(&txmsg->next, &mgr->tx_msg_downq);
> > > > > +
> > > > > +	if (unlikely(drm_debug & DRM_UT_DP)) {
> > > > 
> > > > Like you do here.
> > > > 
> > > > > +		struct drm_printer p = drm_debug_printer(DBG_PREFIX);
> > > > > +
> > > > > +		drm_dp_mst_dump_sideband_msg_tx(&p, txmsg);
> > > > > +	}
> > > > > +
> > > > >  	if (list_is_singular(&mgr->tx_msg_downq))
> > > > >  		process_single_down_tx_qlock(mgr);
> > > > >  	mutex_unlock(&mgr->qlock);
> > > > > diff --git a/drivers/gpu/drm/drm_dp_mst_topology_internal.h
> > > > > b/drivers/gpu/drm/drm_dp_mst_topology_internal.h
> > > > > new file mode 100644
> > > > > index 000000000000..eeda9a61c657
> > > > > --- /dev/null
> > > > > +++ b/drivers/gpu/drm/drm_dp_mst_topology_internal.h
> > > > > @@ -0,0 +1,24 @@
> > > > > +/* SPDX-License-Identifier: GPL-2.0-only
> > > > > + *
> > > > > + * Declarations for DP MST related functions which are only used in
> > > > > selftests
> > > > > + *
> > > > > + * Copyright © 2018 Red Hat
> > > > > + * Authors:
> > > > > + *     Lyude Paul <lyude@redhat.com>
> > > > > + */
> > > > > +
> > > > > +#ifndef _DRM_DP_MST_HELPER_INTERNAL_H_
> > > > > +#define _DRM_DP_MST_HELPER_INTERNAL_H_
> > > > > +
> > > > > +#include <drm/drm_dp_mst_helper.h>
> > > > > +
> > > > > +void
> > > > > +drm_dp_encode_sideband_req(const struct drm_dp_sideband_msg_req_body
> > > > > *req,
> > > > > +			   struct drm_dp_sideband_msg_tx *raw);
> > > > > +int drm_dp_decode_sideband_req(const struct drm_dp_sideband_msg_tx
> > > > > *raw,
> > > > > +			       struct drm_dp_sideband_msg_req_body
> > > > > *req);
> > > > > +void
> > > > > +drm_dp_dump_sideband_msg_req_body(const struct
> > > > > drm_dp_sideband_msg_req_body *req,
> > > > > +				  int indent, struct drm_printer
> > > > > *printer);
> > > > > +
> > > > > +#endif /* !_DRM_DP_MST_HELPER_INTERNAL_H_ */
> > > > > diff --git a/drivers/gpu/drm/selftests/drm_modeset_selftests.h
> > > > > b/drivers/gpu/drm/selftests/drm_modeset_selftests.h
> > > > > index dec3ee3ec96f..1898de0b4a4d 100644
> > > > > --- a/drivers/gpu/drm/selftests/drm_modeset_selftests.h
> > > > > +++ b/drivers/gpu/drm/selftests/drm_modeset_selftests.h
> > > > > @@ -33,3 +33,4 @@ selftest(damage_iter_damage_one_outside,
> > > > > igt_damage_iter_damage_one_outside)
> > > > >  selftest(damage_iter_damage_src_moved,
> > > > > igt_damage_iter_damage_src_moved)
> > > > >  selftest(damage_iter_damage_not_visible,
> > > > > igt_damage_iter_damage_not_visible)
> > > > >  selftest(dp_mst_calc_pbn_mode, igt_dp_mst_calc_pbn_mode)
> > > > > +selftest(dp_mst_sideband_msg_req_decode,
> > > > > igt_dp_mst_sideband_msg_req_decode)
> > > > > diff --git a/drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c
> > > > > b/drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c
> > > > > index 51b2486ec917..ceca89babd65 100644
> > > > > --- a/drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c
> > > > > +++ b/drivers/gpu/drm/selftests/test-drm_dp_mst_helper.c
> > > > > @@ -8,6 +8,7 @@
> > > > >  #include <drm/drm_dp_mst_helper.h>
> > > > >  #include <drm/drm_print.h>
> > > > >  
> > > > > +#include "../drm_dp_mst_topology_internal.h"
> > > > >  #include "test-drm_modeset_common.h"
> > > > >  
> > > > >  int igt_dp_mst_calc_pbn_mode(void *ignored)
> > > > > @@ -34,3 +35,214 @@ int igt_dp_mst_calc_pbn_mode(void *ignored)
> > > > >  
> > > > >  	return 0;
> > > > >  }
> > > > > +
> > > > > +static bool
> > > > > +sideband_msg_req_equal(const struct drm_dp_sideband_msg_req_body *in,
> > > > > +		       const struct drm_dp_sideband_msg_req_body *out)
> > > > > +{
> > > > > +	const struct drm_dp_remote_i2c_read_tx *txin, *txout;
> > > > > +	int i;
> > > > > +
> > > > > +	if (in->req_type != out->req_type)
> > > > > +		return false;
> > > > > +
> > > > > +	switch (in->req_type) {
> > > > > +	case DP_ENUM_PATH_RESOURCES:
> > > > > +	case DP_POWER_UP_PHY:
> > > > > +	case DP_POWER_DOWN_PHY:
> > > > > +	case DP_ALLOCATE_PAYLOAD:
> > > > > +	case DP_QUERY_PAYLOAD:
> > > > > +	case DP_REMOTE_DPCD_READ:
> > > > > +		return memcmp(in, out, sizeof(*in)) == 0;
> > > > 
> > > > Just memcmp the entire thing for everyrone (we assume kzalloc anyway),
> > > > and
> > > > then only have the additional checks for the few cases we need it? Would
> > > > slightly reduce the code.
> > > 
> > > Just realized also - this wouldn't work: remember that I2C_READ,
> > > DPCD_WRITE,
> > > and I2C_WRITE contain pointers which of course will be different even if
> > > the
> > > contents are the same. Unfortunately I don't think there's really a good
> > > solution for this, unless we want to just remove the pointers and store
> > > things
> > > within the down request struct itself
> > 
> > Uh ... disappoint :-/
> > 
> > Then at least put that into a comment, and maybe make this the default:
> > handler?
> > > > > +
> > > > > +	case DP_REMOTE_I2C_READ:
> > > > > +#define IN in->u.i2c_read
> > > > > +#define OUT out->u.i2c_read
> > > > > +		if (IN.num_bytes_read != OUT.num_bytes_read ||
> > > > > +		    IN.num_transactions != OUT.num_transactions ||
> > > > > +		    IN.port_number != OUT.port_number ||
> > > > > +		    IN.read_i2c_device_id != OUT.read_i2c_device_id)
> > > > > +			return false;
> > > > > +
> > > > > +		for (i = 0; i < IN.num_transactions; i++) {
> > > > > +			txin = &IN.transactions[i];
> > > > > +			txout = &OUT.transactions[i];
> > > > > +
> > > > > +			if (txin->i2c_dev_id != txout->i2c_dev_id ||
> > > > > +			    txin->no_stop_bit != txout->no_stop_bit ||
> > > > > +			    txin->num_bytes != txout->num_bytes ||
> > > > > +			    txin->i2c_transaction_delay !=
> > > > > +			    txout->i2c_transaction_delay)
> > > > > +				return false;
> > > > > +
> > > > > +			if (memcmp(txin->bytes, txout->bytes,
> > > > > +				   txin->num_bytes) != 0)
> > > > > +				return false;
> > > > > +		}
> > > > > +		break;
> > > > > +#undef IN
> > > > > +#undef OUT
> > > > > +
> > > > > +	case DP_REMOTE_DPCD_WRITE:
> > > > > +#define IN in->u.dpcd_write
> > > > > +#define OUT out->u.dpcd_write
> > > > > +		if (IN.dpcd_address != OUT.dpcd_address ||
> > > > > +		    IN.num_bytes != OUT.num_bytes ||
> > > > > +		    IN.port_number != OUT.port_number)
> > > > > +			return false;
> > > > > +
> > > > > +		return memcmp(IN.bytes, OUT.bytes, IN.num_bytes) == 0;
> > > > > +#undef IN
> > > > > +#undef OUT
> > > > > +
> > > > > +	case DP_REMOTE_I2C_WRITE:
> > > > > +#define IN in->u.i2c_write
> > > > > +#define OUT out->u.i2c_write
> > > > > +		if (IN.port_number != OUT.port_number ||
> > > > > +		    IN.write_i2c_device_id != OUT.write_i2c_device_id
> > > > > ||
> > > > > +		    IN.num_bytes != OUT.num_bytes)
> > > > > +			return false;
> > > > > +
> > > > > +		return memcmp(IN.bytes, OUT.bytes, IN.num_bytes) == 0;
> > > > > +#undef IN
> > > > > +#undef OUT
> > > > > +	}
> > > > > +
> > > > > +	return true;
> > > > > +}
> > > > > +
> > > > > +static int
> > > > > +__sideband_msg_req_encode_decode(int line,
> > > > > +				 struct drm_dp_sideband_msg_req_body
> > > > > *in,
> > > > > +				 struct drm_dp_sideband_msg_req_body
> > > > > *out)
> > > > > +{
> > > > > +	struct drm_printer p = drm_err_printer(pr_fmt());
> > > > > +	struct drm_dp_sideband_msg_tx txmsg;
> > > > > +	int i, ret;
> > > > > +	bool eq;
> > > > > +
> > > > > +	drm_dp_encode_sideband_req(in, &txmsg);
> > > > > +	ret = drm_dp_decode_sideband_req(&txmsg, out);
> > > > > +	if (ret < 0) {
> > > > > +		pr_err(pr_fmt("Failed to decode sideband request @
> > > > > line %d:
> > > > > %d\n"),
> > > > > +		       line, ret);
> > > > > +		return ret;
> > > > > +	}
> > > > > +
> > > > > +	eq = sideband_msg_req_equal(in, out);
> > > > > +	if (!eq) {
> > > > > +		pr_err(pr_fmt("Encode/decode @ line %d failed,
> > > > > expected:\n"),
> > > > > +		       line);
> > > > > +		drm_dp_dump_sideband_msg_req_body(in, 1, &p);
> > > > > +		pr_err(pr_fmt("Got:\n"));
> > > > > +		drm_dp_dump_sideband_msg_req_body(out, 1, &p);
> > > > > +	}
> > > > > +
> > > > > +	switch (in->req_type) {
> > > > > +	case DP_REMOTE_DPCD_WRITE:
> > > > > +		kfree(out->u.dpcd_write.bytes);
> > > > > +		break;
> > > > > +	case DP_REMOTE_I2C_READ:
> > > > > +		for (i = 0; i < out->u.i2c_read.num_transactions; i++)
> > > > > +			kfree(out->u.i2c_read.transactions[i].bytes);
> > > > > +		break;
> > > > > +	case DP_REMOTE_I2C_WRITE:
> > > > > +		kfree(out->u.i2c_write.bytes);
> > > > > +		break;
> > > > > +	}
> > > > > +
> > > > > +	/* Clear everything but the req_type for the input */
> > > > > +	memset(&in->u, 0, sizeof(in->u));
> > > > > +
> > > > > +	/* Clear the output entirely */
> > > > > +	memset(out, 0, sizeof(*out));
> > > > > +
> > > > > +	return eq ? 0 : -EINVAL;
> > > > > +}
> > > > > +
> > > > > +int igt_dp_mst_sideband_msg_req_decode(void *unused)
> > > > > +{
> > > > > +	struct drm_dp_sideband_msg_req_body in = { 0 }, out = { 0 };
> > > > > +	u8 data[] = { 0xff, 0x0, 0xdd };
> > > > > +	int i;
> > > > > +
> > > > > +#define
> > > > > DO_TEST()                                                          \
> > > > > +	do
> > > > > {                                                               \
> > > > > +		if (__sideband_msg_req_encode_decode(__LINE__, &in,
> > > > > &out)) \
> > > > > +			return
> > > > > -EINVAL;                                    \
> > > > > +	} while (0)
> > > > 
> > > > I had a tiny wtf moment here until I realized this is a macro ... maybe
> > > > put the do on the first line and indent everything? Pure bikeshed to
> > > > help
> > > > the blind ...
> > > > 
> > > > > +
> > > > > +	in.req_type = DP_ENUM_PATH_RESOURCES;
> > > > > +	in.u.port_num.port_number = 5;
> > > > > +	DO_TEST();
> > > > > +
> > > > > +	in.req_type = DP_POWER_UP_PHY;
> > > > > +	in.u.port_num.port_number = 5;
> > > > > +	DO_TEST();
> > > > > +
> > > > > +	in.req_type = DP_POWER_DOWN_PHY;
> > > > > +	in.u.port_num.port_number = 5;
> > > > > +	DO_TEST();
> > > > > +
> > > > > +	in.req_type = DP_ALLOCATE_PAYLOAD;
> > > > > +	in.u.allocate_payload.number_sdp_streams = 3;
> > > > > +	for (i = 0; i < in.u.allocate_payload.number_sdp_streams; i++)
> > > > > +		in.u.allocate_payload.sdp_stream_sink[i] = i + 1;
> > > > > +	DO_TEST();
> > > > > +	in.u.allocate_payload.port_number = 0xf;
> > > > > +	DO_TEST();
> > > > > +	in.u.allocate_payload.vcpi = 0x7f;
> > > > > +	DO_TEST();
> > > > > +	in.u.allocate_payload.pbn = U16_MAX;
> > > > > +	DO_TEST();
> > > > > +
> > > > > +	in.req_type = DP_QUERY_PAYLOAD;
> > > > > +	in.u.query_payload.port_number = 0xf;
> > > > > +	DO_TEST();
> > > > > +	in.u.query_payload.vcpi = 0x7f;
> > > > > +	DO_TEST();
> > > > > +
> > > > > +	in.req_type = DP_REMOTE_DPCD_READ;
> > > > > +	in.u.dpcd_read.port_number = 0xf;
> > > > > +	DO_TEST();
> > > > > +	in.u.dpcd_read.dpcd_address = 0xfedcb;
> > > > > +	DO_TEST();
> > > > > +	in.u.dpcd_read.num_bytes = U8_MAX;
> > > > > +	DO_TEST();
> > > > > +
> > > > > +	in.req_type = DP_REMOTE_DPCD_WRITE;
> > > > > +	in.u.dpcd_write.port_number = 0xf;
> > > > > +	DO_TEST();
> > > > > +	in.u.dpcd_write.dpcd_address = 0xfedcb;
> > > > > +	DO_TEST();
> > > > > +	in.u.dpcd_write.num_bytes = ARRAY_SIZE(data);
> > > > > +	in.u.dpcd_write.bytes = data;
> > > > > +	DO_TEST();
> > > > > +
> > > > > +	in.req_type = DP_REMOTE_I2C_READ;
> > > > > +	in.u.i2c_read.port_number = 0xf;
> > > > > +	DO_TEST();
> > > > > +	in.u.i2c_read.read_i2c_device_id = 0x7f;
> > > > > +	DO_TEST();
> > > > > +	in.u.i2c_read.num_transactions = 3;
> > > > > +	in.u.i2c_read.num_bytes_read = ARRAY_SIZE(data) * 3;
> > > > > +	for (i = 0; i < in.u.i2c_read.num_transactions; i++) {
> > > > > +		in.u.i2c_read.transactions[i].bytes = data;
> > > > > +		in.u.i2c_read.transactions[i].num_bytes =
> > > > > ARRAY_SIZE(data);
> > > > > +		in.u.i2c_read.transactions[i].i2c_dev_id = 0x7f & ~i;
> > > > > +		in.u.i2c_read.transactions[i].i2c_transaction_delay =
> > > > > 0xf &
> > > > > ~i;
> > > > > +	}
> > > > > +	DO_TEST();
> > > > > +
> > > > > +	in.req_type = DP_REMOTE_I2C_WRITE;
> > > > > +	in.u.i2c_write.port_number = 0xf;
> > > > > +	DO_TEST();
> > > > > +	in.u.i2c_write.write_i2c_device_id = 0x7f;
> > > > > +	DO_TEST();
> > > > > +	in.u.i2c_write.num_bytes = ARRAY_SIZE(data);
> > > > > +	in.u.i2c_write.bytes = data;
> > > > > +	DO_TEST();
> > > > > +
> > > > > +#undef DO_TEST
> > > > > +	return 0;
> > > > > +}
> > > > 
> > > > Extremely nice, more unit tests ftw!
> > > > 
> > > > With the nits somehow figured out: Reviewed-by: Daniel Vetter <
> > > > daniel.vetter@ffwll.ch>
> > > > 
> > > > > diff --git a/drivers/gpu/drm/selftests/test-drm_modeset_common.h
> > > > > b/drivers/gpu/drm/selftests/test-drm_modeset_common.h
> > > > > index 590bda35a683..0fcb8bbc6a1b 100644
> > > > > --- a/drivers/gpu/drm/selftests/test-drm_modeset_common.h
> > > > > +++ b/drivers/gpu/drm/selftests/test-drm_modeset_common.h
> > > > > @@ -40,5 +40,6 @@ int igt_damage_iter_damage_one_outside(void
> > > > > *ignored);
> > > > >  int igt_damage_iter_damage_src_moved(void *ignored);
> > > > >  int igt_damage_iter_damage_not_visible(void *ignored);
> > > > >  int igt_dp_mst_calc_pbn_mode(void *ignored);
> > > > > +int igt_dp_mst_sideband_msg_req_decode(void *ignored);
> > > > >  
> > > > >  #endif
> > > > > diff --git a/include/drm/drm_dp_mst_helper.h
> > > > > b/include/drm/drm_dp_mst_helper.h
> > > > > index c01f3ea72756..4c8d177e83e5 100644
> > > > > --- a/include/drm/drm_dp_mst_helper.h
> > > > > +++ b/include/drm/drm_dp_mst_helper.h
> > > > > @@ -293,7 +293,7 @@ struct drm_dp_remote_dpcd_write {
> > > > >  struct drm_dp_remote_i2c_read {
> > > > >  	u8 num_transactions;
> > > > >  	u8 port_number;
> > > > > -	struct {
> > > > > +	struct drm_dp_remote_i2c_read_tx {
> > > > >  		u8 i2c_dev_id;
> > > > >  		u8 num_bytes;
> > > > >  		u8 *bytes;
> > > > > -- 
> > > > > 2.21.0
> > > > > 
> > > -- 
> > > Cheers,
> > > 	Lyude Paul
> > > 
> -- 
> Cheers,
> 	Lyude Paul
> 

-- 
Daniel Vetter
Software Engineer, Intel Corporation
http://blog.ffwll.ch

[PATCH 06/26] drm/dp_mst: Move PDT teardown for ports into destroy_connector_work

[Lyude Paul]

This will allow us to add some locking for port PDTs, which can't be
done from drm_dp_destroy_port() since we don't know what locks the
caller might be holding. Also, this gets rid of a good bit of unneeded
code.

Cc: Juston Li <juston.li@intel.com>
Cc: Imre Deak <imre.deak@intel.com>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Cc: Harry Wentland <hwentlan@amd.com>
Signed-off-by: Lyude Paul <lyude@redhat.com>
---
 drivers/gpu/drm/drm_dp_mst_topology.c | 42 +++++++++++----------------
 1 file changed, 17 insertions(+), 25 deletions(-)

diff --git a/drivers/gpu/drm/drm_dp_mst_topology.c b/drivers/gpu/drm/drm_dp_mst_topology.c
index defc5e09fb9a..0295e007c836 100644
--- a/drivers/gpu/drm/drm_dp_mst_topology.c
+++ b/drivers/gpu/drm/drm_dp_mst_topology.c
@@ -1509,31 +1509,22 @@ static void drm_dp_destroy_port(struct kref *kref)
 		container_of(kref, struct drm_dp_mst_port, topology_kref);
 	struct drm_dp_mst_topology_mgr *mgr = port->mgr;
 
-	if (!port->input) {
-		kfree(port->cached_edid);
-
-		/*
-		 * The only time we don't have a connector
-		 * on an output port is if the connector init
-		 * fails.
-		 */
-		if (port->connector) {
-			/* we can't destroy the connector here, as
-			 * we might be holding the mode_config.mutex
-			 * from an EDID retrieval */
-
-			mutex_lock(&mgr->destroy_connector_lock);
-			list_add(&port->next, &mgr->destroy_connector_list);
-			mutex_unlock(&mgr->destroy_connector_lock);
-			schedule_work(&mgr->destroy_connector_work);
-			return;
-		}
-		/* no need to clean up vcpi
-		 * as if we have no connector we never setup a vcpi */
-		drm_dp_port_teardown_pdt(port, port->pdt);
-		port->pdt = DP_PEER_DEVICE_NONE;
+	/* There's nothing that needs locking to destroy an input port yet */
+	if (port->input) {
+		drm_dp_mst_put_port_malloc(port);
+		return;
 	}
-	drm_dp_mst_put_port_malloc(port);
+
+	kfree(port->cached_edid);
+
+	/*
+	 * we can't destroy the connector here, as we might be holding the
+	 * mode_config.mutex from an EDID retrieval
+	 */
+	mutex_lock(&mgr->destroy_connector_lock);
+	list_add(&port->next, &mgr->destroy_connector_list);
+	mutex_unlock(&mgr->destroy_connector_lock);
+	schedule_work(&mgr->destroy_connector_work);
 }
 
 /**
@@ -3881,7 +3872,8 @@ drm_dp_finish_destroy_port(struct drm_dp_mst_port *port)
 {
 	INIT_LIST_HEAD(&port->next);
 
-	port->mgr->cbs->destroy_connector(port->mgr, port->connector);
+	if (port->connector)
+		port->mgr->cbs->destroy_connector(port->mgr, port->connector);
 
 	drm_dp_port_teardown_pdt(port, port->pdt);
 	port->pdt = DP_PEER_DEVICE_NONE;
-- 
2.21.0

Re: [PATCH 06/26] drm/dp_mst: Move PDT teardown for ports into destroy_connector_work

[Daniel Vetter]

On Wed, Jul 17, 2019 at 09:42:29PM -0400, Lyude Paul wrote:
> This will allow us to add some locking for port PDTs, which can't be
> done from drm_dp_destroy_port() since we don't know what locks the
> caller might be holding. Also, this gets rid of a good bit of unneeded
> code.
> 
> Cc: Juston Li <juston.li@intel.com>
> Cc: Imre Deak <imre.deak@intel.com>
> Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
> Cc: Harry Wentland <hwentlan@amd.com>
> Signed-off-by: Lyude Paul <lyude@redhat.com>
> ---
>  drivers/gpu/drm/drm_dp_mst_topology.c | 42 +++++++++++----------------
>  1 file changed, 17 insertions(+), 25 deletions(-)
> 
> diff --git a/drivers/gpu/drm/drm_dp_mst_topology.c b/drivers/gpu/drm/drm_dp_mst_topology.c
> index defc5e09fb9a..0295e007c836 100644
> --- a/drivers/gpu/drm/drm_dp_mst_topology.c
> +++ b/drivers/gpu/drm/drm_dp_mst_topology.c
> @@ -1509,31 +1509,22 @@ static void drm_dp_destroy_port(struct kref *kref)
>  		container_of(kref, struct drm_dp_mst_port, topology_kref);
>  	struct drm_dp_mst_topology_mgr *mgr = port->mgr;
>  
> -	if (!port->input) {
> -		kfree(port->cached_edid);
> -
> -		/*
> -		 * The only time we don't have a connector
> -		 * on an output port is if the connector init
> -		 * fails.
> -		 */
> -		if (port->connector) {
> -			/* we can't destroy the connector here, as
> -			 * we might be holding the mode_config.mutex
> -			 * from an EDID retrieval */
> -
> -			mutex_lock(&mgr->destroy_connector_lock);
> -			list_add(&port->next, &mgr->destroy_connector_list);
> -			mutex_unlock(&mgr->destroy_connector_lock);
> -			schedule_work(&mgr->destroy_connector_work);
> -			return;
> -		}
> -		/* no need to clean up vcpi
> -		 * as if we have no connector we never setup a vcpi */
> -		drm_dp_port_teardown_pdt(port, port->pdt);
> -		port->pdt = DP_PEER_DEVICE_NONE;
> +	/* There's nothing that needs locking to destroy an input port yet */
> +	if (port->input) {
> +		drm_dp_mst_put_port_malloc(port);
> +		return;
>  	}
> -	drm_dp_mst_put_port_malloc(port);
> +
> +	kfree(port->cached_edid);
> +
> +	/*
> +	 * we can't destroy the connector here, as we might be holding the
> +	 * mode_config.mutex from an EDID retrieval
> +	 */
> +	mutex_lock(&mgr->destroy_connector_lock);
> +	list_add(&port->next, &mgr->destroy_connector_list);
> +	mutex_unlock(&mgr->destroy_connector_lock);
> +	schedule_work(&mgr->destroy_connector_work);

So if I'm not completely blind this just flattens the above code flow (by
inverting the if (port->input)).

>  }
>  
>  /**
> @@ -3881,7 +3872,8 @@ drm_dp_finish_destroy_port(struct drm_dp_mst_port *port)
>  {
>  	INIT_LIST_HEAD(&port->next);
>  
> -	port->mgr->cbs->destroy_connector(port->mgr, port->connector);
> +	if (port->connector)

And this here I can't connect with the commit message. I'm confused, did
something go wrong with some rebase here, and this patch should have a
different title/summary?
-Daniel

> +		port->mgr->cbs->destroy_connector(port->mgr, port->connector);
>  
>  	drm_dp_port_teardown_pdt(port, port->pdt);
>  	port->pdt = DP_PEER_DEVICE_NONE;
> -- 
> 2.21.0
> 

-- 
Daniel Vetter
Software Engineer, Intel Corporation
http://blog.ffwll.ch

Re: [PATCH 06/26] drm/dp_mst: Move PDT teardown for ports into destroy_connector_work

[Lyude Paul]

On Tue, 2019-08-13 at 16:52 +0200, Daniel Vetter wrote:
> On Wed, Jul 17, 2019 at 09:42:29PM -0400, Lyude Paul wrote:
> > This will allow us to add some locking for port PDTs, which can't be
> > done from drm_dp_destroy_port() since we don't know what locks the
> > caller might be holding. Also, this gets rid of a good bit of unneeded
> > code.
> > 
> > Cc: Juston Li <juston.li@intel.com>
> > Cc: Imre Deak <imre.deak@intel.com>
> > Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
> > Cc: Harry Wentland <hwentlan@amd.com>
> > Signed-off-by: Lyude Paul <lyude@redhat.com>
> > ---
> >  drivers/gpu/drm/drm_dp_mst_topology.c | 42 +++++++++++----------------
> >  1 file changed, 17 insertions(+), 25 deletions(-)
> > 
> > diff --git a/drivers/gpu/drm/drm_dp_mst_topology.c
> > b/drivers/gpu/drm/drm_dp_mst_topology.c
> > index defc5e09fb9a..0295e007c836 100644
> > --- a/drivers/gpu/drm/drm_dp_mst_topology.c
> > +++ b/drivers/gpu/drm/drm_dp_mst_topology.c
> > @@ -1509,31 +1509,22 @@ static void drm_dp_destroy_port(struct kref *kref)
> >  		container_of(kref, struct drm_dp_mst_port, topology_kref);
> >  	struct drm_dp_mst_topology_mgr *mgr = port->mgr;
> >  
> > -	if (!port->input) {
> > -		kfree(port->cached_edid);
> > -
> > -		/*
> > -		 * The only time we don't have a connector
> > -		 * on an output port is if the connector init
> > -		 * fails.
> > -		 */
> > -		if (port->connector) {
> > -			/* we can't destroy the connector here, as
> > -			 * we might be holding the mode_config.mutex
> > -			 * from an EDID retrieval */
> > -
> > -			mutex_lock(&mgr->destroy_connector_lock);
> > -			list_add(&port->next, &mgr->destroy_connector_list);
> > -			mutex_unlock(&mgr->destroy_connector_lock);
> > -			schedule_work(&mgr->destroy_connector_work);
> > -			return;
> > -		}
> > -		/* no need to clean up vcpi
> > -		 * as if we have no connector we never setup a vcpi */
> > -		drm_dp_port_teardown_pdt(port, port->pdt);
> > -		port->pdt = DP_PEER_DEVICE_NONE;
> > +	/* There's nothing that needs locking to destroy an input port yet */
> > +	if (port->input) {
> > +		drm_dp_mst_put_port_malloc(port);
> > +		return;
> >  	}
> > -	drm_dp_mst_put_port_malloc(port);
> > +
> > +	kfree(port->cached_edid);
> > +
> > +	/*
> > +	 * we can't destroy the connector here, as we might be holding the
> > +	 * mode_config.mutex from an EDID retrieval
> > +	 */
> > +	mutex_lock(&mgr->destroy_connector_lock);
> > +	list_add(&port->next, &mgr->destroy_connector_list);
> > +	mutex_unlock(&mgr->destroy_connector_lock);
> > +	schedule_work(&mgr->destroy_connector_work);
> 
> So if I'm not completely blind this just flattens the above code flow (by
> inverting the if (port->input)).

Now I'm really remembering why I refactored this. The control flow on the
previous version of this is pretty misleading. To summarize so it's a bit more
obvious:

if (port->input) {
	drm_dp_mst_put_port_malloc(port);
	return;
} else if (port->connector) {
	add_connector_to_destroy_list();
	return;
	/* ^ now, this is where PDT teardown happens */
} else {
	drm_dp_port_teardown_pdt(port, port->pdt);
}
/* free edid etc etc */

So, I suppose the title of this patch would be more accurate if it was
"drm/dp_mst: Remove PDT teardown in destroy_port() and refactor"
I'll go ahead and change that

> 
> >  }
> >  
> >  /**
> > @@ -3881,7 +3872,8 @@ drm_dp_finish_destroy_port(struct drm_dp_mst_port
> > *port)
> >  {
> >  	INIT_LIST_HEAD(&port->next);
> >  
> > -	port->mgr->cbs->destroy_connector(port->mgr, port->connector);
> > +	if (port->connector)
> 
> And this here I can't connect with the commit message. I'm confused, did
> something go wrong with some rebase here, and this patch should have a
> different title/summary?
> -Daniel

No, this is correct. In the previous drm_dp_destroy_port() function we only
added a port to the delayed destroy work if it had a connector on it. Now that
we add ports unconditionally, we have to check port->connector before trying
to call port->mgr->cbs->destroy_connector() since port->connector is no longer
guaranteed to be != NULL.

> 
> > +		port->mgr->cbs->destroy_connector(port->mgr, port->connector);
> >  
> >  	drm_dp_port_teardown_pdt(port, port->pdt);
> >  	port->pdt = DP_PEER_DEVICE_NONE;
> > -- 
> > 2.21.0
> > 
-- 
Cheers,
	Lyude Paul

[PATCH 07/26] drm/dp_mst: Get rid of list clear in drm_dp_finish_destroy_port()

[Lyude Paul]

This seems to be some leftover detritus from before the port/mstb kref
cleanup and doesn't do anything anymore, so get rid of it.

Cc: Juston Li <juston.li@intel.com>
Cc: Imre Deak <imre.deak@intel.com>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Cc: Harry Wentland <hwentlan@amd.com>
Signed-off-by: Lyude Paul <lyude@redhat.com>
---
 drivers/gpu/drm/drm_dp_mst_topology.c | 2 --
 1 file changed, 2 deletions(-)

diff --git a/drivers/gpu/drm/drm_dp_mst_topology.c b/drivers/gpu/drm/drm_dp_mst_topology.c
index 0295e007c836..ec6865e1af75 100644
--- a/drivers/gpu/drm/drm_dp_mst_topology.c
+++ b/drivers/gpu/drm/drm_dp_mst_topology.c
@@ -3870,8 +3870,6 @@ static void drm_dp_tx_work(struct work_struct *work)
 static inline void
 drm_dp_finish_destroy_port(struct drm_dp_mst_port *port)
 {
-	INIT_LIST_HEAD(&port->next);
-
 	if (port->connector)
 		port->mgr->cbs->destroy_connector(port->mgr, port->connector);
 
-- 
2.21.0

Re: [PATCH 07/26] drm/dp_mst: Get rid of list clear in drm_dp_finish_destroy_port()

[Daniel Vetter]

On Wed, Jul 17, 2019 at 09:42:30PM -0400, Lyude Paul wrote:
> This seems to be some leftover detritus from before the port/mstb kref
> cleanup and doesn't do anything anymore, so get rid of it.
> 
> Cc: Juston Li <juston.li@intel.com>
> Cc: Imre Deak <imre.deak@intel.com>
> Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
> Cc: Harry Wentland <hwentlan@amd.com>
> Signed-off-by: Lyude Paul <lyude@redhat.com>

Maybe move this earlier, before patch 2. Either way:

Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>


> ---
>  drivers/gpu/drm/drm_dp_mst_topology.c | 2 --
>  1 file changed, 2 deletions(-)
> 
> diff --git a/drivers/gpu/drm/drm_dp_mst_topology.c b/drivers/gpu/drm/drm_dp_mst_topology.c
> index 0295e007c836..ec6865e1af75 100644
> --- a/drivers/gpu/drm/drm_dp_mst_topology.c
> +++ b/drivers/gpu/drm/drm_dp_mst_topology.c
> @@ -3870,8 +3870,6 @@ static void drm_dp_tx_work(struct work_struct *work)
>  static inline void
>  drm_dp_finish_destroy_port(struct drm_dp_mst_port *port)
>  {
> -	INIT_LIST_HEAD(&port->next);
> -
>  	if (port->connector)
>  		port->mgr->cbs->destroy_connector(port->mgr, port->connector);
>  
> -- 
> 2.21.0
> 

-- 
Daniel Vetter
Software Engineer, Intel Corporation
http://blog.ffwll.ch

[PATCH 08/26] drm/dp_mst: Refactor drm_dp_send_enum_path_resources

[Lyude Paul]

Use more pointers so we don't have to write out
txmsg->reply.u.path_resources each time. Also, fix line wrapping +
rearrange local variables.

Cc: Juston Li <juston.li@intel.com>
Cc: Imre Deak <imre.deak@intel.com>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Cc: Harry Wentland <hwentlan@amd.com>
Signed-off-by: Lyude Paul <lyude@redhat.com>
---
 drivers/gpu/drm/drm_dp_mst_topology.c | 24 ++++++++++++++++--------
 1 file changed, 16 insertions(+), 8 deletions(-)

diff --git a/drivers/gpu/drm/drm_dp_mst_topology.c b/drivers/gpu/drm/drm_dp_mst_topology.c
index ec6865e1af75..57c9c605ee17 100644
--- a/drivers/gpu/drm/drm_dp_mst_topology.c
+++ b/drivers/gpu/drm/drm_dp_mst_topology.c
@@ -2350,12 +2350,14 @@ static void drm_dp_send_link_address(struct drm_dp_mst_topology_mgr *mgr,
 	kfree(txmsg);
 }
 
-static int drm_dp_send_enum_path_resources(struct drm_dp_mst_topology_mgr *mgr,
-					   struct drm_dp_mst_branch *mstb,
-					   struct drm_dp_mst_port *port)
+static int
+drm_dp_send_enum_path_resources(struct drm_dp_mst_topology_mgr *mgr,
+				struct drm_dp_mst_branch *mstb,
+				struct drm_dp_mst_port *port)
 {
-	int len;
+	struct drm_dp_enum_path_resources_ack_reply *path_res;
 	struct drm_dp_sideband_msg_tx *txmsg;
+	int len;
 	int ret;
 
 	txmsg = kzalloc(sizeof(*txmsg), GFP_KERNEL);
@@ -2369,14 +2371,20 @@ static int drm_dp_send_enum_path_resources(struct drm_dp_mst_topology_mgr *mgr,
 
 	ret = drm_dp_mst_wait_tx_reply(mstb, txmsg);
 	if (ret > 0) {
+		path_res = &txmsg->reply.u.path_resources;
+
 		if (txmsg->reply.reply_type == DP_SIDEBAND_REPLY_NAK) {
 			DRM_DEBUG_KMS("enum path resources nak received\n");
 		} else {
-			if (port->port_num != txmsg->reply.u.path_resources.port_number)
+			if (port->port_num != path_res->port_number)
 				DRM_ERROR("got incorrect port in response\n");
-			DRM_DEBUG_KMS("enum path resources %d: %d %d\n", txmsg->reply.u.path_resources.port_number, txmsg->reply.u.path_resources.full_payload_bw_number,
-			       txmsg->reply.u.path_resources.avail_payload_bw_number);
-			port->available_pbn = txmsg->reply.u.path_resources.avail_payload_bw_number;
+
+			DRM_DEBUG_KMS("enum path resources %d: %d %d\n",
+				      path_res->port_number,
+				      path_res->full_payload_bw_number,
+				      path_res->avail_payload_bw_number);
+			port->available_pbn =
+				path_res->avail_payload_bw_number;
 		}
 	}
 
-- 
2.21.0

Re: [PATCH 08/26] drm/dp_mst: Refactor drm_dp_send_enum_path_resources

[Daniel Vetter]

On Wed, Jul 17, 2019 at 09:42:31PM -0400, Lyude Paul wrote:
> Use more pointers so we don't have to write out
> txmsg->reply.u.path_resources each time. Also, fix line wrapping +
> rearrange local variables.
> 
> Cc: Juston Li <juston.li@intel.com>
> Cc: Imre Deak <imre.deak@intel.com>
> Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
> Cc: Harry Wentland <hwentlan@amd.com>
> Signed-off-by: Lyude Paul <lyude@redhat.com>

Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>

> ---
>  drivers/gpu/drm/drm_dp_mst_topology.c | 24 ++++++++++++++++--------
>  1 file changed, 16 insertions(+), 8 deletions(-)
> 
> diff --git a/drivers/gpu/drm/drm_dp_mst_topology.c b/drivers/gpu/drm/drm_dp_mst_topology.c
> index ec6865e1af75..57c9c605ee17 100644
> --- a/drivers/gpu/drm/drm_dp_mst_topology.c
> +++ b/drivers/gpu/drm/drm_dp_mst_topology.c
> @@ -2350,12 +2350,14 @@ static void drm_dp_send_link_address(struct drm_dp_mst_topology_mgr *mgr,
>  	kfree(txmsg);
>  }
>  
> -static int drm_dp_send_enum_path_resources(struct drm_dp_mst_topology_mgr *mgr,
> -					   struct drm_dp_mst_branch *mstb,
> -					   struct drm_dp_mst_port *port)
> +static int
> +drm_dp_send_enum_path_resources(struct drm_dp_mst_topology_mgr *mgr,
> +				struct drm_dp_mst_branch *mstb,
> +				struct drm_dp_mst_port *port)
>  {
> -	int len;
> +	struct drm_dp_enum_path_resources_ack_reply *path_res;
>  	struct drm_dp_sideband_msg_tx *txmsg;
> +	int len;
>  	int ret;
>  
>  	txmsg = kzalloc(sizeof(*txmsg), GFP_KERNEL);
> @@ -2369,14 +2371,20 @@ static int drm_dp_send_enum_path_resources(struct drm_dp_mst_topology_mgr *mgr,
>  
>  	ret = drm_dp_mst_wait_tx_reply(mstb, txmsg);
>  	if (ret > 0) {
> +		path_res = &txmsg->reply.u.path_resources;
> +
>  		if (txmsg->reply.reply_type == DP_SIDEBAND_REPLY_NAK) {
>  			DRM_DEBUG_KMS("enum path resources nak received\n");
>  		} else {
> -			if (port->port_num != txmsg->reply.u.path_resources.port_number)
> +			if (port->port_num != path_res->port_number)
>  				DRM_ERROR("got incorrect port in response\n");
> -			DRM_DEBUG_KMS("enum path resources %d: %d %d\n", txmsg->reply.u.path_resources.port_number, txmsg->reply.u.path_resources.full_payload_bw_number,
> -			       txmsg->reply.u.path_resources.avail_payload_bw_number);
> -			port->available_pbn = txmsg->reply.u.path_resources.avail_payload_bw_number;
> +
> +			DRM_DEBUG_KMS("enum path resources %d: %d %d\n",
> +				      path_res->port_number,
> +				      path_res->full_payload_bw_number,
> +				      path_res->avail_payload_bw_number);
> +			port->available_pbn =
> +				path_res->avail_payload_bw_number;
>  		}
>  	}
>  
> -- 
> 2.21.0
> 

-- 
Daniel Vetter
Software Engineer, Intel Corporation
http://blog.ffwll.ch

[PATCH 09/26] drm/dp_mst: Remove huge conditional in drm_dp_mst_handle_up_req()

[Lyude Paul]

Which reduces indentation and makes this function more legible.

Cc: Juston Li <juston.li@intel.com>
Cc: Imre Deak <imre.deak@intel.com>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Cc: Harry Wentland <hwentlan@amd.com>
Signed-off-by: Lyude Paul <lyude@redhat.com>
---
 drivers/gpu/drm/drm_dp_mst_topology.c | 90 +++++++++++++--------------
 1 file changed, 45 insertions(+), 45 deletions(-)

diff --git a/drivers/gpu/drm/drm_dp_mst_topology.c b/drivers/gpu/drm/drm_dp_mst_topology.c
index 57c9c605ee17..b867a2e8f779 100644
--- a/drivers/gpu/drm/drm_dp_mst_topology.c
+++ b/drivers/gpu/drm/drm_dp_mst_topology.c
@@ -3126,7 +3126,9 @@ static int drm_dp_mst_handle_down_rep(struct drm_dp_mst_topology_mgr *mgr)
 
 static int drm_dp_mst_handle_up_req(struct drm_dp_mst_topology_mgr *mgr)
 {
-	int ret = 0;
+	struct drm_dp_sideband_msg_req_body msg;
+	struct drm_dp_mst_branch *mstb = NULL;
+	bool seqno;
 
 	if (!drm_dp_get_one_sb_msg(mgr, true)) {
 		memset(&mgr->up_req_recv, 0,
@@ -3134,62 +3136,60 @@ static int drm_dp_mst_handle_up_req(struct drm_dp_mst_topology_mgr *mgr)
 		return 0;
 	}
 
-	if (mgr->up_req_recv.have_eomt) {
-		struct drm_dp_sideband_msg_req_body msg;
-		struct drm_dp_mst_branch *mstb = NULL;
-		bool seqno;
-
-		if (!mgr->up_req_recv.initial_hdr.broadcast) {
-			mstb = drm_dp_get_mst_branch_device(mgr,
-							    mgr->up_req_recv.initial_hdr.lct,
-							    mgr->up_req_recv.initial_hdr.rad);
-			if (!mstb) {
-				DRM_DEBUG_KMS("Got MST reply from unknown device %d\n", mgr->up_req_recv.initial_hdr.lct);
-				memset(&mgr->up_req_recv, 0, sizeof(struct drm_dp_sideband_msg_rx));
-				return 0;
-			}
-		}
+	if (!mgr->up_req_recv.have_eomt)
+		return 0;
 
-		seqno = mgr->up_req_recv.initial_hdr.seqno;
-		drm_dp_sideband_parse_req(&mgr->up_req_recv, &msg);
+	if (!mgr->up_req_recv.initial_hdr.broadcast) {
+		mstb = drm_dp_get_mst_branch_device(mgr,
+						    mgr->up_req_recv.initial_hdr.lct,
+						    mgr->up_req_recv.initial_hdr.rad);
+		if (!mstb) {
+			DRM_DEBUG_KMS("Got MST reply from unknown device %d\n", mgr->up_req_recv.initial_hdr.lct);
+			memset(&mgr->up_req_recv, 0, sizeof(struct drm_dp_sideband_msg_rx));
+			return 0;
+		}
+	}
 
-		if (msg.req_type == DP_CONNECTION_STATUS_NOTIFY) {
-			drm_dp_send_up_ack_reply(mgr, mgr->mst_primary, msg.req_type, seqno, false);
+	seqno = mgr->up_req_recv.initial_hdr.seqno;
+	drm_dp_sideband_parse_req(&mgr->up_req_recv, &msg);
 
-			if (!mstb)
-				mstb = drm_dp_get_mst_branch_device_by_guid(mgr, msg.u.conn_stat.guid);
+	if (msg.req_type == DP_CONNECTION_STATUS_NOTIFY) {
+		drm_dp_send_up_ack_reply(mgr, mgr->mst_primary, msg.req_type, seqno, false);
 
-			if (!mstb) {
-				DRM_DEBUG_KMS("Got MST reply from unknown device %d\n", mgr->up_req_recv.initial_hdr.lct);
-				memset(&mgr->up_req_recv, 0, sizeof(struct drm_dp_sideband_msg_rx));
-				return 0;
-			}
+		if (!mstb)
+			mstb = drm_dp_get_mst_branch_device_by_guid(mgr, msg.u.conn_stat.guid);
 
-			drm_dp_update_port(mstb, &msg.u.conn_stat);
+		if (!mstb) {
+			DRM_DEBUG_KMS("Got MST reply from unknown device %d\n", mgr->up_req_recv.initial_hdr.lct);
+			memset(&mgr->up_req_recv, 0, sizeof(struct drm_dp_sideband_msg_rx));
+			return 0;
+		}
 
-			DRM_DEBUG_KMS("Got CSN: pn: %d ldps:%d ddps: %d mcs: %d ip: %d pdt: %d\n", msg.u.conn_stat.port_number, msg.u.conn_stat.legacy_device_plug_status, msg.u.conn_stat.displayport_device_plug_status, msg.u.conn_stat.message_capability_status, msg.u.conn_stat.input_port, msg.u.conn_stat.peer_device_type);
-			drm_kms_helper_hotplug_event(mgr->dev);
+		drm_dp_update_port(mstb, &msg.u.conn_stat);
 
-		} else if (msg.req_type == DP_RESOURCE_STATUS_NOTIFY) {
-			drm_dp_send_up_ack_reply(mgr, mgr->mst_primary, msg.req_type, seqno, false);
-			if (!mstb)
-				mstb = drm_dp_get_mst_branch_device_by_guid(mgr, msg.u.resource_stat.guid);
+		DRM_DEBUG_KMS("Got CSN: pn: %d ldps:%d ddps: %d mcs: %d ip: %d pdt: %d\n", msg.u.conn_stat.port_number, msg.u.conn_stat.legacy_device_plug_status, msg.u.conn_stat.displayport_device_plug_status, msg.u.conn_stat.message_capability_status, msg.u.conn_stat.input_port, msg.u.conn_stat.peer_device_type);
+		drm_kms_helper_hotplug_event(mgr->dev);
 
-			if (!mstb) {
-				DRM_DEBUG_KMS("Got MST reply from unknown device %d\n", mgr->up_req_recv.initial_hdr.lct);
-				memset(&mgr->up_req_recv, 0, sizeof(struct drm_dp_sideband_msg_rx));
-				return 0;
-			}
+	} else if (msg.req_type == DP_RESOURCE_STATUS_NOTIFY) {
+		drm_dp_send_up_ack_reply(mgr, mgr->mst_primary, msg.req_type, seqno, false);
+		if (!mstb)
+			mstb = drm_dp_get_mst_branch_device_by_guid(mgr, msg.u.resource_stat.guid);
 
-			DRM_DEBUG_KMS("Got RSN: pn: %d avail_pbn %d\n", msg.u.resource_stat.port_number, msg.u.resource_stat.available_pbn);
+		if (!mstb) {
+			DRM_DEBUG_KMS("Got MST reply from unknown device %d\n", mgr->up_req_recv.initial_hdr.lct);
+			memset(&mgr->up_req_recv, 0, sizeof(struct drm_dp_sideband_msg_rx));
+			return 0;
 		}
 
-		if (mstb)
-			drm_dp_mst_topology_put_mstb(mstb);
-
-		memset(&mgr->up_req_recv, 0, sizeof(struct drm_dp_sideband_msg_rx));
+		DRM_DEBUG_KMS("Got RSN: pn: %d avail_pbn %d\n", msg.u.resource_stat.port_number, msg.u.resource_stat.available_pbn);
 	}
-	return ret;
+
+	if (mstb)
+		drm_dp_mst_topology_put_mstb(mstb);
+
+	memset(&mgr->up_req_recv, 0, sizeof(struct drm_dp_sideband_msg_rx));
+
+	return 0;
 }
 
 /**
-- 
2.21.0

Re: [PATCH 09/26] drm/dp_mst: Remove huge conditional in drm_dp_mst_handle_up_req()

[Daniel Vetter]

On Wed, Jul 17, 2019 at 09:42:32PM -0400, Lyude Paul wrote:
> Which reduces indentation and makes this function more legible.

Indeed, nice one.

Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>


> 
> Cc: Juston Li <juston.li@intel.com>
> Cc: Imre Deak <imre.deak@intel.com>
> Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
> Cc: Harry Wentland <hwentlan@amd.com>
> Signed-off-by: Lyude Paul <lyude@redhat.com>
> ---
>  drivers/gpu/drm/drm_dp_mst_topology.c | 90 +++++++++++++--------------
>  1 file changed, 45 insertions(+), 45 deletions(-)
> 
> diff --git a/drivers/gpu/drm/drm_dp_mst_topology.c b/drivers/gpu/drm/drm_dp_mst_topology.c
> index 57c9c605ee17..b867a2e8f779 100644
> --- a/drivers/gpu/drm/drm_dp_mst_topology.c
> +++ b/drivers/gpu/drm/drm_dp_mst_topology.c
> @@ -3126,7 +3126,9 @@ static int drm_dp_mst_handle_down_rep(struct drm_dp_mst_topology_mgr *mgr)
>  
>  static int drm_dp_mst_handle_up_req(struct drm_dp_mst_topology_mgr *mgr)
>  {
> -	int ret = 0;
> +	struct drm_dp_sideband_msg_req_body msg;
> +	struct drm_dp_mst_branch *mstb = NULL;
> +	bool seqno;
>  
>  	if (!drm_dp_get_one_sb_msg(mgr, true)) {
>  		memset(&mgr->up_req_recv, 0,
> @@ -3134,62 +3136,60 @@ static int drm_dp_mst_handle_up_req(struct drm_dp_mst_topology_mgr *mgr)
>  		return 0;
>  	}
>  
> -	if (mgr->up_req_recv.have_eomt) {
> -		struct drm_dp_sideband_msg_req_body msg;
> -		struct drm_dp_mst_branch *mstb = NULL;
> -		bool seqno;
> -
> -		if (!mgr->up_req_recv.initial_hdr.broadcast) {
> -			mstb = drm_dp_get_mst_branch_device(mgr,
> -							    mgr->up_req_recv.initial_hdr.lct,
> -							    mgr->up_req_recv.initial_hdr.rad);
> -			if (!mstb) {
> -				DRM_DEBUG_KMS("Got MST reply from unknown device %d\n", mgr->up_req_recv.initial_hdr.lct);
> -				memset(&mgr->up_req_recv, 0, sizeof(struct drm_dp_sideband_msg_rx));
> -				return 0;
> -			}
> -		}
> +	if (!mgr->up_req_recv.have_eomt)
> +		return 0;
>  
> -		seqno = mgr->up_req_recv.initial_hdr.seqno;
> -		drm_dp_sideband_parse_req(&mgr->up_req_recv, &msg);
> +	if (!mgr->up_req_recv.initial_hdr.broadcast) {
> +		mstb = drm_dp_get_mst_branch_device(mgr,
> +						    mgr->up_req_recv.initial_hdr.lct,
> +						    mgr->up_req_recv.initial_hdr.rad);
> +		if (!mstb) {
> +			DRM_DEBUG_KMS("Got MST reply from unknown device %d\n", mgr->up_req_recv.initial_hdr.lct);
> +			memset(&mgr->up_req_recv, 0, sizeof(struct drm_dp_sideband_msg_rx));
> +			return 0;
> +		}
> +	}
>  
> -		if (msg.req_type == DP_CONNECTION_STATUS_NOTIFY) {
> -			drm_dp_send_up_ack_reply(mgr, mgr->mst_primary, msg.req_type, seqno, false);
> +	seqno = mgr->up_req_recv.initial_hdr.seqno;
> +	drm_dp_sideband_parse_req(&mgr->up_req_recv, &msg);
>  
> -			if (!mstb)
> -				mstb = drm_dp_get_mst_branch_device_by_guid(mgr, msg.u.conn_stat.guid);
> +	if (msg.req_type == DP_CONNECTION_STATUS_NOTIFY) {
> +		drm_dp_send_up_ack_reply(mgr, mgr->mst_primary, msg.req_type, seqno, false);
>  
> -			if (!mstb) {
> -				DRM_DEBUG_KMS("Got MST reply from unknown device %d\n", mgr->up_req_recv.initial_hdr.lct);
> -				memset(&mgr->up_req_recv, 0, sizeof(struct drm_dp_sideband_msg_rx));
> -				return 0;
> -			}
> +		if (!mstb)
> +			mstb = drm_dp_get_mst_branch_device_by_guid(mgr, msg.u.conn_stat.guid);
>  
> -			drm_dp_update_port(mstb, &msg.u.conn_stat);
> +		if (!mstb) {
> +			DRM_DEBUG_KMS("Got MST reply from unknown device %d\n", mgr->up_req_recv.initial_hdr.lct);
> +			memset(&mgr->up_req_recv, 0, sizeof(struct drm_dp_sideband_msg_rx));
> +			return 0;
> +		}
>  
> -			DRM_DEBUG_KMS("Got CSN: pn: %d ldps:%d ddps: %d mcs: %d ip: %d pdt: %d\n", msg.u.conn_stat.port_number, msg.u.conn_stat.legacy_device_plug_status, msg.u.conn_stat.displayport_device_plug_status, msg.u.conn_stat.message_capability_status, msg.u.conn_stat.input_port, msg.u.conn_stat.peer_device_type);
> -			drm_kms_helper_hotplug_event(mgr->dev);
> +		drm_dp_update_port(mstb, &msg.u.conn_stat);
>  
> -		} else if (msg.req_type == DP_RESOURCE_STATUS_NOTIFY) {
> -			drm_dp_send_up_ack_reply(mgr, mgr->mst_primary, msg.req_type, seqno, false);
> -			if (!mstb)
> -				mstb = drm_dp_get_mst_branch_device_by_guid(mgr, msg.u.resource_stat.guid);
> +		DRM_DEBUG_KMS("Got CSN: pn: %d ldps:%d ddps: %d mcs: %d ip: %d pdt: %d\n", msg.u.conn_stat.port_number, msg.u.conn_stat.legacy_device_plug_status, msg.u.conn_stat.displayport_device_plug_status, msg.u.conn_stat.message_capability_status, msg.u.conn_stat.input_port, msg.u.conn_stat.peer_device_type);
> +		drm_kms_helper_hotplug_event(mgr->dev);
>  
> -			if (!mstb) {
> -				DRM_DEBUG_KMS("Got MST reply from unknown device %d\n", mgr->up_req_recv.initial_hdr.lct);
> -				memset(&mgr->up_req_recv, 0, sizeof(struct drm_dp_sideband_msg_rx));
> -				return 0;
> -			}
> +	} else if (msg.req_type == DP_RESOURCE_STATUS_NOTIFY) {
> +		drm_dp_send_up_ack_reply(mgr, mgr->mst_primary, msg.req_type, seqno, false);
> +		if (!mstb)
> +			mstb = drm_dp_get_mst_branch_device_by_guid(mgr, msg.u.resource_stat.guid);
>  
> -			DRM_DEBUG_KMS("Got RSN: pn: %d avail_pbn %d\n", msg.u.resource_stat.port_number, msg.u.resource_stat.available_pbn);
> +		if (!mstb) {
> +			DRM_DEBUG_KMS("Got MST reply from unknown device %d\n", mgr->up_req_recv.initial_hdr.lct);
> +			memset(&mgr->up_req_recv, 0, sizeof(struct drm_dp_sideband_msg_rx));
> +			return 0;
>  		}
>  
> -		if (mstb)
> -			drm_dp_mst_topology_put_mstb(mstb);
> -
> -		memset(&mgr->up_req_recv, 0, sizeof(struct drm_dp_sideband_msg_rx));
> +		DRM_DEBUG_KMS("Got RSN: pn: %d avail_pbn %d\n", msg.u.resource_stat.port_number, msg.u.resource_stat.available_pbn);
>  	}
> -	return ret;
> +
> +	if (mstb)
> +		drm_dp_mst_topology_put_mstb(mstb);
> +
> +	memset(&mgr->up_req_recv, 0, sizeof(struct drm_dp_sideband_msg_rx));
> +
> +	return 0;
>  }
>  
>  /**
> -- 
> 2.21.0
> 

-- 
Daniel Vetter
Software Engineer, Intel Corporation
http://blog.ffwll.ch

[PATCH 10/26] drm/dp_mst: Constify guid in drm_dp_get_mst_branch_by_guid()

[Lyude Paul]

And it's helper, we'll be using this in just a moment.

Cc: Juston Li <juston.li@intel.com>
Cc: Imre Deak <imre.deak@intel.com>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Cc: Harry Wentland <hwentlan@amd.com>
Signed-off-by: Lyude Paul <lyude@redhat.com>
---
 drivers/gpu/drm/drm_dp_mst_topology.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/drivers/gpu/drm/drm_dp_mst_topology.c b/drivers/gpu/drm/drm_dp_mst_topology.c
index b867a2e8f779..0894bc237082 100644
--- a/drivers/gpu/drm/drm_dp_mst_topology.c
+++ b/drivers/gpu/drm/drm_dp_mst_topology.c
@@ -1971,7 +1971,7 @@ static struct drm_dp_mst_branch *drm_dp_get_mst_branch_device(struct drm_dp_mst_
 
 static struct drm_dp_mst_branch *get_mst_branch_device_by_guid_helper(
 	struct drm_dp_mst_branch *mstb,
-	uint8_t *guid)
+	const uint8_t *guid)
 {
 	struct drm_dp_mst_branch *found_mstb;
 	struct drm_dp_mst_port *port;
@@ -1995,7 +1995,7 @@ static struct drm_dp_mst_branch *get_mst_branch_device_by_guid_helper(
 
 static struct drm_dp_mst_branch *
 drm_dp_get_mst_branch_device_by_guid(struct drm_dp_mst_topology_mgr *mgr,
-				     uint8_t *guid)
+				     const uint8_t *guid)
 {
 	struct drm_dp_mst_branch *mstb;
 	int ret;
-- 
2.21.0

[PATCH 11/26] drm/dp_mst: Refactor drm_dp_mst_handle_up_req()

[Lyude Paul]

There's a couple of changes here, so to summarize:

* Remove the big ugly mgr->up_req_recv.have_eomt conditional to save on
  indenting
* Store &mgr->up_req_recv.initial_hdr in a variable so we don't keep
  going over 80 character long lines
* De-duplicate code for calling drm_dp_send_up_ack_reply() and getting
  the MSTB via it's GUID
* Remove all of the duplicate calls to memset() and just use a goto
  instead
* Actually do line wrapping
* Remove the unnecessary if (mstb) check before calling
  drm_dp_mst_topology_put_mstb() - we are guaranteed to always have
  mstb != NULL at that point in the function

Cc: Juston Li <juston.li@intel.com>
Cc: Imre Deak <imre.deak@intel.com>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Cc: Harry Wentland <hwentlan@amd.com>
Signed-off-by: Lyude Paul <lyude@redhat.com>
---
 drivers/gpu/drm/drm_dp_mst_topology.c | 75 ++++++++++++++-------------
 1 file changed, 38 insertions(+), 37 deletions(-)

diff --git a/drivers/gpu/drm/drm_dp_mst_topology.c b/drivers/gpu/drm/drm_dp_mst_topology.c
index 0894bc237082..5dea219bf744 100644
--- a/drivers/gpu/drm/drm_dp_mst_topology.c
+++ b/drivers/gpu/drm/drm_dp_mst_topology.c
@@ -3127,68 +3127,69 @@ static int drm_dp_mst_handle_down_rep(struct drm_dp_mst_topology_mgr *mgr)
 static int drm_dp_mst_handle_up_req(struct drm_dp_mst_topology_mgr *mgr)
 {
 	struct drm_dp_sideband_msg_req_body msg;
+	struct drm_dp_sideband_msg_hdr *hdr = &mgr->up_req_recv.initial_hdr;
 	struct drm_dp_mst_branch *mstb = NULL;
+	const u8 *guid;
 	bool seqno;
 
-	if (!drm_dp_get_one_sb_msg(mgr, true)) {
-		memset(&mgr->up_req_recv, 0,
-		       sizeof(struct drm_dp_sideband_msg_rx));
-		return 0;
-	}
+	if (!drm_dp_get_one_sb_msg(mgr, true))
+		goto out;
 
 	if (!mgr->up_req_recv.have_eomt)
 		return 0;
 
-	if (!mgr->up_req_recv.initial_hdr.broadcast) {
-		mstb = drm_dp_get_mst_branch_device(mgr,
-						    mgr->up_req_recv.initial_hdr.lct,
-						    mgr->up_req_recv.initial_hdr.rad);
+	if (!hdr->broadcast) {
+		mstb = drm_dp_get_mst_branch_device(mgr, hdr->lct, hdr->rad);
 		if (!mstb) {
-			DRM_DEBUG_KMS("Got MST reply from unknown device %d\n", mgr->up_req_recv.initial_hdr.lct);
-			memset(&mgr->up_req_recv, 0, sizeof(struct drm_dp_sideband_msg_rx));
-			return 0;
+			DRM_DEBUG_KMS("Got MST reply from unknown device %d\n",
+				      hdr->lct);
+			goto out;
 		}
 	}
 
-	seqno = mgr->up_req_recv.initial_hdr.seqno;
+	seqno = hdr->seqno;
 	drm_dp_sideband_parse_req(&mgr->up_req_recv, &msg);
 
-	if (msg.req_type == DP_CONNECTION_STATUS_NOTIFY) {
-		drm_dp_send_up_ack_reply(mgr, mgr->mst_primary, msg.req_type, seqno, false);
+	if (msg.req_type == DP_CONNECTION_STATUS_NOTIFY)
+		guid = msg.u.conn_stat.guid;
+	else if (msg.req_type == DP_RESOURCE_STATUS_NOTIFY)
+		guid = msg.u.resource_stat.guid;
+	else
+		goto out;
 
-		if (!mstb)
-			mstb = drm_dp_get_mst_branch_device_by_guid(mgr, msg.u.conn_stat.guid);
+	drm_dp_send_up_ack_reply(mgr, mgr->mst_primary, msg.req_type, seqno,
+				 false);
 
+	if (!mstb) {
+		mstb = drm_dp_get_mst_branch_device_by_guid(mgr, guid);
 		if (!mstb) {
-			DRM_DEBUG_KMS("Got MST reply from unknown device %d\n", mgr->up_req_recv.initial_hdr.lct);
-			memset(&mgr->up_req_recv, 0, sizeof(struct drm_dp_sideband_msg_rx));
-			return 0;
+			DRM_DEBUG_KMS("Got MST reply from unknown device %d\n",
+				      hdr->lct);
+			goto out;
 		}
+	}
 
+	if (msg.req_type == DP_CONNECTION_STATUS_NOTIFY) {
 		drm_dp_update_port(mstb, &msg.u.conn_stat);
 
-		DRM_DEBUG_KMS("Got CSN: pn: %d ldps:%d ddps: %d mcs: %d ip: %d pdt: %d\n", msg.u.conn_stat.port_number, msg.u.conn_stat.legacy_device_plug_status, msg.u.conn_stat.displayport_device_plug_status, msg.u.conn_stat.message_capability_status, msg.u.conn_stat.input_port, msg.u.conn_stat.peer_device_type);
-		drm_kms_helper_hotplug_event(mgr->dev);
+		DRM_DEBUG_KMS("Got CSN: pn: %d ldps:%d ddps: %d mcs: %d ip: %d pdt: %d\n",
+			      msg.u.conn_stat.port_number,
+			      msg.u.conn_stat.legacy_device_plug_status,
+			      msg.u.conn_stat.displayport_device_plug_status,
+			      msg.u.conn_stat.message_capability_status,
+			      msg.u.conn_stat.input_port,
+			      msg.u.conn_stat.peer_device_type);
 
+		drm_kms_helper_hotplug_event(mgr->dev);
 	} else if (msg.req_type == DP_RESOURCE_STATUS_NOTIFY) {
-		drm_dp_send_up_ack_reply(mgr, mgr->mst_primary, msg.req_type, seqno, false);
-		if (!mstb)
-			mstb = drm_dp_get_mst_branch_device_by_guid(mgr, msg.u.resource_stat.guid);
-
-		if (!mstb) {
-			DRM_DEBUG_KMS("Got MST reply from unknown device %d\n", mgr->up_req_recv.initial_hdr.lct);
-			memset(&mgr->up_req_recv, 0, sizeof(struct drm_dp_sideband_msg_rx));
-			return 0;
-		}
-
-		DRM_DEBUG_KMS("Got RSN: pn: %d avail_pbn %d\n", msg.u.resource_stat.port_number, msg.u.resource_stat.available_pbn);
+		DRM_DEBUG_KMS("Got RSN: pn: %d avail_pbn %d\n",
+			      msg.u.resource_stat.port_number,
+			      msg.u.resource_stat.available_pbn);
 	}
 
-	if (mstb)
-		drm_dp_mst_topology_put_mstb(mstb);
-
+	drm_dp_mst_topology_put_mstb(mstb);
+out:
 	memset(&mgr->up_req_recv, 0, sizeof(struct drm_dp_sideband_msg_rx));
-
 	return 0;
 }
 
-- 
2.21.0

[PATCH 12/26] drm/dp_mst: Refactor drm_dp_mst_handle_down_rep()

[Lyude Paul]

* Remove the big ugly have_eomt conditional
* Store &mgr->down_rep_recv.initial_hdr in a var to make line wrapping
  easier
* Remove duplicate memset() calls
* Actually wrap lines

Cc: Juston Li <juston.li@intel.com>
Cc: Imre Deak <imre.deak@intel.com>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Cc: Harry Wentland <hwentlan@amd.com>
Signed-off-by: Lyude Paul <lyude@redhat.com>
---
 drivers/gpu/drm/drm_dp_mst_topology.c | 102 +++++++++++++-------------
 1 file changed, 50 insertions(+), 52 deletions(-)

diff --git a/drivers/gpu/drm/drm_dp_mst_topology.c b/drivers/gpu/drm/drm_dp_mst_topology.c
index 5dea219bf744..9e7fc6a96080 100644
--- a/drivers/gpu/drm/drm_dp_mst_topology.c
+++ b/drivers/gpu/drm/drm_dp_mst_topology.c
@@ -3060,68 +3060,66 @@ static bool drm_dp_get_one_sb_msg(struct drm_dp_mst_topology_mgr *mgr, bool up)
 
 static int drm_dp_mst_handle_down_rep(struct drm_dp_mst_topology_mgr *mgr)
 {
-	int ret = 0;
+	struct drm_dp_sideband_msg_tx *txmsg;
+	struct drm_dp_mst_branch *mstb;
+	struct drm_dp_sideband_msg_hdr *hdr = &mgr->down_rep_recv.initial_hdr;
+	int slot = -1;
+
+	if (!drm_dp_get_one_sb_msg(mgr, false))
+		goto clear_down_rep_recv;
 
-	if (!drm_dp_get_one_sb_msg(mgr, false)) {
-		memset(&mgr->down_rep_recv, 0,
-		       sizeof(struct drm_dp_sideband_msg_rx));
+	if (!mgr->down_rep_recv.have_eomt)
 		return 0;
+
+	mstb = drm_dp_get_mst_branch_device(mgr, hdr->lct, hdr->rad);
+	if (!mstb) {
+		DRM_DEBUG_KMS("Got MST reply from unknown device %d\n",
+			      hdr->lct);
+		goto clear_down_rep_recv;
 	}
 
-	if (mgr->down_rep_recv.have_eomt) {
-		struct drm_dp_sideband_msg_tx *txmsg;
-		struct drm_dp_mst_branch *mstb;
-		int slot = -1;
-		mstb = drm_dp_get_mst_branch_device(mgr,
-						    mgr->down_rep_recv.initial_hdr.lct,
-						    mgr->down_rep_recv.initial_hdr.rad);
+	/* find the message */
+	slot = hdr->seqno;
+	mutex_lock(&mgr->qlock);
+	txmsg = mstb->tx_slots[slot];
+	/* remove from slots */
+	mutex_unlock(&mgr->qlock);
 
-		if (!mstb) {
-			DRM_DEBUG_KMS("Got MST reply from unknown device %d\n", mgr->down_rep_recv.initial_hdr.lct);
-			memset(&mgr->down_rep_recv, 0, sizeof(struct drm_dp_sideband_msg_rx));
-			return 0;
-		}
+	if (!txmsg) {
+		DRM_DEBUG_KMS("Got MST reply with no msg %p %d %d %02x %02x\n",
+			      mstb, hdr->seqno, hdr->lct, hdr->rad[0],
+			      mgr->down_rep_recv.msg[0]);
+		goto no_msg;
+	}
 
-		/* find the message */
-		slot = mgr->down_rep_recv.initial_hdr.seqno;
-		mutex_lock(&mgr->qlock);
-		txmsg = mstb->tx_slots[slot];
-		/* remove from slots */
-		mutex_unlock(&mgr->qlock);
-
-		if (!txmsg) {
-			DRM_DEBUG_KMS("Got MST reply with no msg %p %d %d %02x %02x\n",
-			       mstb,
-			       mgr->down_rep_recv.initial_hdr.seqno,
-			       mgr->down_rep_recv.initial_hdr.lct,
-				      mgr->down_rep_recv.initial_hdr.rad[0],
-				      mgr->down_rep_recv.msg[0]);
-			drm_dp_mst_topology_put_mstb(mstb);
-			memset(&mgr->down_rep_recv, 0, sizeof(struct drm_dp_sideband_msg_rx));
-			return 0;
-		}
+	drm_dp_sideband_parse_reply(&mgr->down_rep_recv, &txmsg->reply);
 
-		drm_dp_sideband_parse_reply(&mgr->down_rep_recv, &txmsg->reply);
+	if (txmsg->reply.reply_type == DP_SIDEBAND_REPLY_NAK)
+		DRM_DEBUG_KMS("Got NAK reply: req 0x%02x (%s), reason 0x%02x (%s), nak data 0x%02x\n",
+			      txmsg->reply.req_type,
+			      drm_dp_mst_req_type_str(txmsg->reply.req_type),
+			      txmsg->reply.u.nak.reason,
+			      drm_dp_mst_nak_reason_str(txmsg->reply.u.nak.reason),
+			      txmsg->reply.u.nak.nak_data);
 
-		if (txmsg->reply.reply_type == DP_SIDEBAND_REPLY_NAK)
-			DRM_DEBUG_KMS("Got NAK reply: req 0x%02x (%s), reason 0x%02x (%s), nak data 0x%02x\n",
-				      txmsg->reply.req_type,
-				      drm_dp_mst_req_type_str(txmsg->reply.req_type),
-				      txmsg->reply.u.nak.reason,
-				      drm_dp_mst_nak_reason_str(txmsg->reply.u.nak.reason),
-				      txmsg->reply.u.nak.nak_data);
-
-		memset(&mgr->down_rep_recv, 0, sizeof(struct drm_dp_sideband_msg_rx));
-		drm_dp_mst_topology_put_mstb(mstb);
+	memset(&mgr->down_rep_recv, 0, sizeof(struct drm_dp_sideband_msg_rx));
+	drm_dp_mst_topology_put_mstb(mstb);
 
-		mutex_lock(&mgr->qlock);
-		txmsg->state = DRM_DP_SIDEBAND_TX_RX;
-		mstb->tx_slots[slot] = NULL;
-		mutex_unlock(&mgr->qlock);
+	mutex_lock(&mgr->qlock);
+	txmsg->state = DRM_DP_SIDEBAND_TX_RX;
+	mstb->tx_slots[slot] = NULL;
+	mutex_unlock(&mgr->qlock);
 
-		wake_up_all(&mgr->tx_waitq);
-	}
-	return ret;
+	wake_up_all(&mgr->tx_waitq);
+
+	return 0;
+
+no_msg:
+	drm_dp_mst_topology_put_mstb(mstb);
+clear_down_rep_recv:
+	memset(&mgr->down_rep_recv, 0, sizeof(struct drm_dp_sideband_msg_rx));
+
+	return 0;
 }
 
 static int drm_dp_mst_handle_up_req(struct drm_dp_mst_topology_mgr *mgr)
-- 
2.21.0

[PATCH 13/26] drm/dp_mst: Destroy topology_mgr mutexes

[Lyude Paul]

Turns out we've been forgetting for a while now to actually destroy any
of the mutexes that we create in drm_dp_mst_topology_mgr. So, let's do
that.

Cc: Juston Li <juston.li@intel.com>
Cc: Imre Deak <imre.deak@intel.com>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Cc: Harry Wentland <hwentlan@amd.com>
Signed-off-by: Lyude Paul <lyude@redhat.com>
---
 drivers/gpu/drm/drm_dp_mst_topology.c | 5 +++++
 1 file changed, 5 insertions(+)

diff --git a/drivers/gpu/drm/drm_dp_mst_topology.c b/drivers/gpu/drm/drm_dp_mst_topology.c
index 9e7fc6a96080..26541af47d91 100644
--- a/drivers/gpu/drm/drm_dp_mst_topology.c
+++ b/drivers/gpu/drm/drm_dp_mst_topology.c
@@ -4205,6 +4205,11 @@ void drm_dp_mst_topology_mgr_destroy(struct drm_dp_mst_topology_mgr *mgr)
 	mgr->aux = NULL;
 	drm_atomic_private_obj_fini(&mgr->base);
 	mgr->funcs = NULL;
+
+	mutex_destroy(&mgr->destroy_connector_lock);
+	mutex_destroy(&mgr->payload_lock);
+	mutex_destroy(&mgr->qlock);
+	mutex_destroy(&mgr->lock);
 }
 EXPORT_SYMBOL(drm_dp_mst_topology_mgr_destroy);
 
-- 
2.21.0

[PATCH 14/26] drm/dp_mst: Cleanup drm_dp_send_link_address() a bit

[Lyude Paul]

Declare local pointer to the drm_dp_link_address_ack_reply struct
instead of constantly dereferencing it through the union in
txmsg->reply. Then, invert the order of conditionals so we don't have to
do the bulk of the work inside them, and can wrap lines even less. Then
finally, rearrange variable declarations a bit.

Cc: Juston Li <juston.li@intel.com>
Cc: Imre Deak <imre.deak@intel.com>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Cc: Harry Wentland <hwentlan@amd.com>
Signed-off-by: Lyude Paul <lyude@redhat.com>
---
 drivers/gpu/drm/drm_dp_mst_topology.c | 42 +++++++++++++++------------
 1 file changed, 23 insertions(+), 19 deletions(-)

diff --git a/drivers/gpu/drm/drm_dp_mst_topology.c b/drivers/gpu/drm/drm_dp_mst_topology.c
index 26541af47d91..13adfb77bf00 100644
--- a/drivers/gpu/drm/drm_dp_mst_topology.c
+++ b/drivers/gpu/drm/drm_dp_mst_topology.c
@@ -2311,9 +2311,9 @@ drm_dp_dump_link_address(struct drm_dp_link_address_ack_reply *reply)
 static void drm_dp_send_link_address(struct drm_dp_mst_topology_mgr *mgr,
 				     struct drm_dp_mst_branch *mstb)
 {
-	int len;
 	struct drm_dp_sideband_msg_tx *txmsg;
-	int ret;
+	struct drm_dp_link_address_ack_reply *reply;
+	int i, len, ret;
 
 	txmsg = kzalloc(sizeof(*txmsg), GFP_KERNEL);
 	if (!txmsg)
@@ -2325,28 +2325,32 @@ static void drm_dp_send_link_address(struct drm_dp_mst_topology_mgr *mgr,
 	mstb->link_address_sent = true;
 	drm_dp_queue_down_tx(mgr, txmsg);
 
+	/* FIXME: Actually do some real error handling here */
 	ret = drm_dp_mst_wait_tx_reply(mstb, txmsg);
-	if (ret > 0) {
-		int i;
+	if (ret <= 0) {
+		DRM_ERROR("Sending link address failed with %d\n", ret);
+		goto out;
+	}
+	if (txmsg->reply.reply_type == DP_SIDEBAND_REPLY_NAK) {
+		DRM_ERROR("link address NAK received\n");
+		ret = -EIO;
+		goto out;
+	}
 
-		if (txmsg->reply.reply_type == DP_SIDEBAND_REPLY_NAK) {
-			DRM_DEBUG_KMS("link address nak received\n");
-		} else {
-			DRM_DEBUG_KMS("link address reply: %d\n", txmsg->reply.u.link_addr.nports);
-			drm_dp_dump_link_address(&txmsg->reply.u.link_addr);
+	reply = &txmsg->reply.u.link_addr;
+	DRM_DEBUG_KMS("link address reply: %d\n", reply->nports);
+	drm_dp_dump_link_address(reply);
 
-			drm_dp_check_mstb_guid(mstb, txmsg->reply.u.link_addr.guid);
+	drm_dp_check_mstb_guid(mstb, reply->guid);
 
-			for (i = 0; i < txmsg->reply.u.link_addr.nports; i++) {
-				drm_dp_add_port(mstb, mgr->dev, &txmsg->reply.u.link_addr.ports[i]);
-			}
-			drm_kms_helper_hotplug_event(mgr->dev);
-		}
-	} else {
-		mstb->link_address_sent = false;
-		DRM_DEBUG_KMS("link address failed %d\n", ret);
-	}
+	for (i = 0; i < reply->nports; i++)
+		drm_dp_add_port(mstb, mgr->dev, &reply->ports[i]);
+
+	drm_kms_helper_hotplug_event(mgr->dev);
 
+out:
+	if (ret <= 0)
+		mstb->link_address_sent = false;
 	kfree(txmsg);
 }
 
-- 
2.21.0

[PATCH 15/26] drm/dp_mst: Refactor pdt setup/teardown, add more locking

[Lyude Paul]

Since we're going to be implementing suspend/resume reprobing very soon,
we need to make sure we are extra careful to ensure that our locking
actually protects the topology state where we expect it to. Turns out
this isn't the case with drm_dp_port_setup_pdt() and
drm_dp_port_teardown_pdt(), both of which change port->mstb without
grabbing &mgr->lock.

Additionally, since most callers of these functions are just using it to
teardown the port's previous PDT and setup a new one we can simplify
things a bit and combine drm_dp_port_setup_pdt() and
drm_dp_port_teardown_pdt() into a single function:
drm_dp_port_set_pdt(). This function also handles actually ensuring that
we grab the correct locks when we need to modify port->mstb.

Cc: Juston Li <juston.li@intel.com>
Cc: Imre Deak <imre.deak@intel.com>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Cc: Harry Wentland <hwentlan@amd.com>
Signed-off-by: Lyude Paul <lyude@redhat.com>
---
 drivers/gpu/drm/drm_dp_mst_topology.c | 181 +++++++++++++++-----------
 include/drm/drm_dp_mst_helper.h       |   6 +-
 2 files changed, 110 insertions(+), 77 deletions(-)

diff --git a/drivers/gpu/drm/drm_dp_mst_topology.c b/drivers/gpu/drm/drm_dp_mst_topology.c
index 13adfb77bf00..68fda2131feb 100644
--- a/drivers/gpu/drm/drm_dp_mst_topology.c
+++ b/drivers/gpu/drm/drm_dp_mst_topology.c
@@ -1485,24 +1485,6 @@ drm_dp_mst_topology_put_mstb(struct drm_dp_mst_branch *mstb)
 	kref_put(&mstb->topology_kref, drm_dp_destroy_mst_branch_device);
 }
 
-static void drm_dp_port_teardown_pdt(struct drm_dp_mst_port *port, int old_pdt)
-{
-	struct drm_dp_mst_branch *mstb;
-
-	switch (old_pdt) {
-	case DP_PEER_DEVICE_DP_LEGACY_CONV:
-	case DP_PEER_DEVICE_SST_SINK:
-		/* remove i2c over sideband */
-		drm_dp_mst_unregister_i2c_bus(&port->aux);
-		break;
-	case DP_PEER_DEVICE_MST_BRANCHING:
-		mstb = port->mstb;
-		port->mstb = NULL;
-		drm_dp_mst_topology_put_mstb(mstb);
-		break;
-	}
-}
-
 static void drm_dp_destroy_port(struct kref *kref)
 {
 	struct drm_dp_mst_port *port =
@@ -1712,38 +1694,79 @@ static u8 drm_dp_calculate_rad(struct drm_dp_mst_port *port,
 	return parent_lct + 1;
 }
 
-/*
- * return sends link address for new mstb
- */
-static bool drm_dp_port_setup_pdt(struct drm_dp_mst_port *port)
+static int drm_dp_port_set_pdt(struct drm_dp_mst_port *port, u8 new_pdt)
 {
-	int ret;
-	u8 rad[6], lct;
-	bool send_link = false;
+	struct drm_dp_mst_topology_mgr *mgr = port->mgr;
+	struct drm_dp_mst_branch *mstb;
+	u8 rad[8], lct;
+	int ret = 0;
+
+	if (port->pdt == new_pdt)
+		return 0;
+
+	/* Teardown the old pdt, if there is one */
+	switch (port->pdt) {
+	case DP_PEER_DEVICE_DP_LEGACY_CONV:
+	case DP_PEER_DEVICE_SST_SINK:
+		/*
+		 * If the new PDT would also have an i2c bus, don't bother
+		 * with reregistering it
+		 */
+		if (new_pdt == DP_PEER_DEVICE_DP_LEGACY_CONV ||
+		    new_pdt == DP_PEER_DEVICE_SST_SINK) {
+			port->pdt = new_pdt;
+			return 0;
+		}
+
+		/* remove i2c over sideband */
+		drm_dp_mst_unregister_i2c_bus(&port->aux);
+		break;
+	case DP_PEER_DEVICE_MST_BRANCHING:
+		mutex_lock(&mgr->lock);
+		drm_dp_mst_topology_put_mstb(port->mstb);
+		port->mstb = NULL;
+		mutex_unlock(&mgr->lock);
+		break;
+	}
+
+	port->pdt = new_pdt;
 	switch (port->pdt) {
 	case DP_PEER_DEVICE_DP_LEGACY_CONV:
 	case DP_PEER_DEVICE_SST_SINK:
 		/* add i2c over sideband */
 		ret = drm_dp_mst_register_i2c_bus(&port->aux);
 		break;
+
 	case DP_PEER_DEVICE_MST_BRANCHING:
 		lct = drm_dp_calculate_rad(port, rad);
+		mstb = drm_dp_add_mst_branch_device(lct, rad);
+		if (!mstb) {
+			ret = -ENOMEM;
+			DRM_ERROR("Failed to create MSTB for port %p", port);
+			goto out;
+		}
 
-		port->mstb = drm_dp_add_mst_branch_device(lct, rad);
-		if (port->mstb) {
-			port->mstb->mgr = port->mgr;
-			port->mstb->port_parent = port;
-			/*
-			 * Make sure this port's memory allocation stays
-			 * around until its child MSTB releases it
-			 */
-			drm_dp_mst_get_port_malloc(port);
+		mutex_lock(&mgr->lock);
+		port->mstb = mstb;
+		mstb->mgr = port->mgr;
+		mstb->port_parent = port;
 
-			send_link = true;
-		}
+		/*
+		 * Make sure this port's memory allocation stays
+		 * around until its child MSTB releases it
+		 */
+		drm_dp_mst_get_port_malloc(port);
+		mutex_unlock(&mgr->lock);
+
+		/* And make sure we send a link address for this */
+		ret = 1;
 		break;
 	}
-	return send_link;
+
+out:
+	if (ret < 0)
+		port->pdt = DP_PEER_DEVICE_NONE;
+	return ret;
 }
 
 static void drm_dp_check_mstb_guid(struct drm_dp_mst_branch *mstb, u8 *guid)
@@ -1793,10 +1816,9 @@ static void drm_dp_add_port(struct drm_dp_mst_branch *mstb,
 			    struct drm_device *dev,
 			    struct drm_dp_link_addr_reply_port *port_msg)
 {
+	struct drm_dp_mst_topology_mgr *mgr = mstb->mgr;
 	struct drm_dp_mst_port *port;
-	bool ret;
 	bool created = false;
-	int old_pdt = 0;
 	int old_ddps = 0;
 
 	port = drm_dp_get_port(mstb, port_msg->port_number);
@@ -1808,7 +1830,7 @@ static void drm_dp_add_port(struct drm_dp_mst_branch *mstb,
 		kref_init(&port->malloc_kref);
 		port->parent = mstb;
 		port->port_num = port_msg->port_number;
-		port->mgr = mstb->mgr;
+		port->mgr = mgr;
 		port->aux.name = "DPMST";
 		port->aux.dev = dev->dev;
 
@@ -1820,11 +1842,9 @@ static void drm_dp_add_port(struct drm_dp_mst_branch *mstb,
 
 		created = true;
 	} else {
-		old_pdt = port->pdt;
 		old_ddps = port->ddps;
 	}
 
-	port->pdt = port_msg->peer_device_type;
 	port->input = port_msg->input_port;
 	port->mcs = port_msg->mcs;
 	port->ddps = port_msg->ddps;
@@ -1836,29 +1856,33 @@ static void drm_dp_add_port(struct drm_dp_mst_branch *mstb,
 	/* manage mstb port lists with mgr lock - take a reference
 	   for this list */
 	if (created) {
-		mutex_lock(&mstb->mgr->lock);
+		mutex_lock(&mgr->lock);
 		drm_dp_mst_topology_get_port(port);
 		list_add(&port->next, &mstb->ports);
-		mutex_unlock(&mstb->mgr->lock);
+		mutex_unlock(&mgr->lock);
 	}
 
 	if (old_ddps != port->ddps) {
 		if (port->ddps) {
 			if (!port->input) {
-				drm_dp_send_enum_path_resources(mstb->mgr,
-								mstb, port);
+				drm_dp_send_enum_path_resources(mgr, mstb,
+								port);
 			}
 		} else {
 			port->available_pbn = 0;
 		}
 	}
 
-	if (old_pdt != port->pdt && !port->input) {
-		drm_dp_port_teardown_pdt(port, old_pdt);
-
-		ret = drm_dp_port_setup_pdt(port);
-		if (ret == true)
-			drm_dp_send_link_address(mstb->mgr, port->mstb);
+	if (!port->input) {
+		int ret = drm_dp_port_set_pdt(port,
+					      port_msg->peer_device_type);
+		if (ret == 1) {
+			drm_dp_send_link_address(mgr, port->mstb);
+		} else if (ret < 0) {
+			DRM_ERROR("Failed to change PDT on port %p: %d\n",
+				  port, ret);
+			goto fail;
+		}
 	}
 
 	if (created && !port->input) {
@@ -1866,18 +1890,11 @@ static void drm_dp_add_port(struct drm_dp_mst_branch *mstb,
 
 		build_mst_prop_path(mstb, port->port_num, proppath,
 				    sizeof(proppath));
-		port->connector = (*mstb->mgr->cbs->add_connector)(mstb->mgr,
-								   port,
-								   proppath);
-		if (!port->connector) {
-			/* remove it from the port list */
-			mutex_lock(&mstb->mgr->lock);
-			list_del(&port->next);
-			mutex_unlock(&mstb->mgr->lock);
-			/* drop port list reference */
-			drm_dp_mst_topology_put_port(port);
-			goto out;
-		}
+		port->connector = (*mgr->cbs->add_connector)(mgr, port,
+							     proppath);
+		if (!port->connector)
+			goto fail;
+
 		if ((port->pdt == DP_PEER_DEVICE_DP_LEGACY_CONV ||
 		     port->pdt == DP_PEER_DEVICE_SST_SINK) &&
 		    port->port_num >= DP_MST_LOGICAL_PORT_0) {
@@ -1885,28 +1902,38 @@ static void drm_dp_add_port(struct drm_dp_mst_branch *mstb,
 							 &port->aux.ddc);
 			drm_connector_set_tile_property(port->connector);
 		}
-		(*mstb->mgr->cbs->register_connector)(port->connector);
+
+		(*mgr->cbs->register_connector)(port->connector);
 	}
 
-out:
 	/* put reference to this port */
 	drm_dp_mst_topology_put_port(port);
+	return;
+
+fail:
+	/* Remove it from the port list */
+	mutex_lock(&mgr->lock);
+	list_del(&port->next);
+	mutex_unlock(&mgr->lock);
+
+	/* Drop the port list reference */
+	drm_dp_mst_topology_put_port(port);
+	/* And now drop our reference */
+	drm_dp_mst_topology_put_port(port);
 }
 
 static void drm_dp_update_port(struct drm_dp_mst_branch *mstb,
 			       struct drm_dp_connection_status_notify *conn_stat)
 {
 	struct drm_dp_mst_port *port;
-	int old_pdt;
 	int old_ddps;
 	bool dowork = false;
+
 	port = drm_dp_get_port(mstb, conn_stat->port_number);
 	if (!port)
 		return;
 
 	old_ddps = port->ddps;
-	old_pdt = port->pdt;
-	port->pdt = conn_stat->peer_device_type;
 	port->mcs = conn_stat->message_capability_status;
 	port->ldps = conn_stat->legacy_device_plug_status;
 	port->ddps = conn_stat->displayport_device_plug_status;
@@ -1918,11 +1945,17 @@ static void drm_dp_update_port(struct drm_dp_mst_branch *mstb,
 			port->available_pbn = 0;
 		}
 	}
-	if (old_pdt != port->pdt && !port->input) {
-		drm_dp_port_teardown_pdt(port, old_pdt);
 
-		if (drm_dp_port_setup_pdt(port))
+	if (!port->input) {
+		int ret = drm_dp_port_set_pdt(port,
+					      conn_stat->peer_device_type);
+		if (ret == 1) {
 			dowork = true;
+		} else if (ret < 0) {
+			DRM_ERROR("Failed to change PDT for port %p: %d\n",
+				  port, ret);
+			dowork = false;
+		}
 	}
 
 	drm_dp_mst_topology_put_port(port);
@@ -3884,9 +3917,7 @@ drm_dp_finish_destroy_port(struct drm_dp_mst_port *port)
 	if (port->connector)
 		port->mgr->cbs->destroy_connector(port->mgr, port->connector);
 
-	drm_dp_port_teardown_pdt(port, port->pdt);
-	port->pdt = DP_PEER_DEVICE_NONE;
-
+	drm_dp_port_set_pdt(port, DP_PEER_DEVICE_NONE);
 	drm_dp_mst_put_port_malloc(port);
 }
 
diff --git a/include/drm/drm_dp_mst_helper.h b/include/drm/drm_dp_mst_helper.h
index 4c8d177e83e5..f704d217c9e0 100644
--- a/include/drm/drm_dp_mst_helper.h
+++ b/include/drm/drm_dp_mst_helper.h
@@ -55,8 +55,10 @@ struct drm_dp_vcpi {
  * @num_sdp_stream_sinks: Number of stream sinks
  * @available_pbn: Available bandwidth for this port.
  * @next: link to next port on this branch device
- * @mstb: branch device attach below this port
- * @aux: i2c aux transport to talk to device connected to this port.
+ * @mstb: branch device on this port, protected by
+ * &drm_dp_mst_topology_mgr.lock
+ * @aux: i2c aux transport to talk to device connected to this port, protected
+ * by &drm_dp_mst_topology_mgr.lock
  * @parent: branch device parent of this port
  * @vcpi: Virtual Channel Payload info for this port.
  * @connector: DRM connector this port is connected to.
-- 
2.21.0

[PATCH 16/26] drm/dp_mst: Rename drm_dp_add_port and drm_dp_update_port

[Lyude Paul]

The names for these functions are rather confusing. drm_dp_add_port()
sounds like a function that would simply create a port and add it to a
topology, and do nothing more. Similarly, drm_dp_update_port() would be
assumed to be the function that should be used to update port
information after initial creation.

While those assumptions are currently correct in how these functions are
used, a quick glance at drm_dp_add_port() reveals that drm_dp_add_port()
can also update the information on a port, and seems explicitly designed
to do so. This can be explained pretty simply by the fact that there's
more situations that would involve updating the port information based
on a link address response as opposed to a connection status
notification than the driver's initial topology probe. Case in point:
reprobing link addresses after suspend/resume.

Since we're about to start using drm_dp_add_port() differently for
suspend/resume reprobing, let's rename both functions to clarify what
they actually do.

Cc: Juston Li <juston.li@intel.com>
Cc: Imre Deak <imre.deak@intel.com>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Cc: Harry Wentland <hwentlan@amd.com>
Signed-off-by: Lyude Paul <lyude@redhat.com>
---
 drivers/gpu/drm/drm_dp_mst_topology.c | 17 ++++++++++-------
 1 file changed, 10 insertions(+), 7 deletions(-)

diff --git a/drivers/gpu/drm/drm_dp_mst_topology.c b/drivers/gpu/drm/drm_dp_mst_topology.c
index 68fda2131feb..35ced8514e18 100644
--- a/drivers/gpu/drm/drm_dp_mst_topology.c
+++ b/drivers/gpu/drm/drm_dp_mst_topology.c
@@ -1812,9 +1812,10 @@ static void build_mst_prop_path(const struct drm_dp_mst_branch *mstb,
 	strlcat(proppath, temp, proppath_size);
 }
 
-static void drm_dp_add_port(struct drm_dp_mst_branch *mstb,
-			    struct drm_device *dev,
-			    struct drm_dp_link_addr_reply_port *port_msg)
+static void
+drm_dp_mst_handle_link_address_port(struct drm_dp_mst_branch *mstb,
+				    struct drm_device *dev,
+				    struct drm_dp_link_addr_reply_port *port_msg)
 {
 	struct drm_dp_mst_topology_mgr *mgr = mstb->mgr;
 	struct drm_dp_mst_port *port;
@@ -1922,8 +1923,9 @@ static void drm_dp_add_port(struct drm_dp_mst_branch *mstb,
 	drm_dp_mst_topology_put_port(port);
 }
 
-static void drm_dp_update_port(struct drm_dp_mst_branch *mstb,
-			       struct drm_dp_connection_status_notify *conn_stat)
+static void
+drm_dp_mst_handle_conn_stat(struct drm_dp_mst_branch *mstb,
+			    struct drm_dp_connection_status_notify *conn_stat)
 {
 	struct drm_dp_mst_port *port;
 	int old_ddps;
@@ -2377,7 +2379,8 @@ static void drm_dp_send_link_address(struct drm_dp_mst_topology_mgr *mgr,
 	drm_dp_check_mstb_guid(mstb, reply->guid);
 
 	for (i = 0; i < reply->nports; i++)
-		drm_dp_add_port(mstb, mgr->dev, &reply->ports[i]);
+		drm_dp_mst_handle_link_address_port(mstb, mgr->dev,
+						    &reply->ports[i]);
 
 	drm_kms_helper_hotplug_event(mgr->dev);
 
@@ -3205,7 +3208,7 @@ static int drm_dp_mst_handle_up_req(struct drm_dp_mst_topology_mgr *mgr)
 	}
 
 	if (msg.req_type == DP_CONNECTION_STATUS_NOTIFY) {
-		drm_dp_update_port(mstb, &msg.u.conn_stat);
+		drm_dp_mst_handle_conn_stat(mstb, &msg.u.conn_stat);
 
 		DRM_DEBUG_KMS("Got CSN: pn: %d ldps:%d ddps: %d mcs: %d ip: %d pdt: %d\n",
 			      msg.u.conn_stat.port_number,
-- 
2.21.0

[PATCH 17/26] drm/dp_mst: Remove lies in {up,down}_rep_recv documentation

[Lyude Paul]

These are most certainly accessed from far more than the mgr work. In
fact, up_req_recv is -only- ever accessed from outside the mgr work.

Cc: Juston Li <juston.li@intel.com>
Cc: Imre Deak <imre.deak@intel.com>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Cc: Harry Wentland <hwentlan@amd.com>
Signed-off-by: Lyude Paul <lyude@redhat.com>
---
 include/drm/drm_dp_mst_helper.h | 8 ++------
 1 file changed, 2 insertions(+), 6 deletions(-)

diff --git a/include/drm/drm_dp_mst_helper.h b/include/drm/drm_dp_mst_helper.h
index f704d217c9e0..8e04a1bd2e9d 100644
--- a/include/drm/drm_dp_mst_helper.h
+++ b/include/drm/drm_dp_mst_helper.h
@@ -489,15 +489,11 @@ struct drm_dp_mst_topology_mgr {
 	int conn_base_id;
 
 	/**
-	 * @down_rep_recv: Message receiver state for down replies. This and
-	 * @up_req_recv are only ever access from the work item, which is
-	 * serialised.
+	 * @down_rep_recv: Message receiver state for down replies.
 	 */
 	struct drm_dp_sideband_msg_rx down_rep_recv;
 	/**
-	 * @up_req_recv: Message receiver state for up requests. This and
-	 * @down_rep_recv are only ever access from the work item, which is
-	 * serialised.
+	 * @up_req_recv: Message receiver state for up requests.
 	 */
 	struct drm_dp_sideband_msg_rx up_req_recv;
 
-- 
2.21.0

[PATCH 18/26] drm/dp_mst: Handle UP requests asynchronously

[Lyude Paul]

Once upon a time, hotplugging devices on MST branches actually worked in
DRM. Now, it only works in amdgpu (likely because of how it's hotplug
handlers are implemented). On both i915 and nouveau, hotplug
notifications from MST branches are noticed - but trying to respond to
them causes messaging timeouts and causes the whole topology state to go
out of sync with reality, usually resulting in the user needing to
replug the entire topology in hopes that it actually fixes things.

The reason for this is because the way we currently handle UP requests
in MST is completely bogus. drm_dp_mst_handle_up_req() is called from
drm_dp_mst_hpd_irq(), which is usually called from the driver's hotplug
handler. Because we handle sending the hotplug event from this function,
we actually cause the driver's hotplug handler (and in turn, all
sideband transactions) to block on
drm_device->mode_config.connection_mutex. This makes it impossible to
send any sideband messages from the driver's connector probing
functions, resulting in the aforementioned sideband message timeout.

There's even more problems with this beyond breaking hotplugging on MST
branch devices. It also makes it almost impossible to protect
drm_dp_mst_port struct members under a lock because we then have to
worry about dealing with all of the lock dependency issues that ensue.

So, let's finally actually fix this issue by handling the processing of
up requests asyncronously. This way we can send sideband messages from
most contexts without having to deal with getting blocked if we hold
connection_mutex. This also fixes MST branch device hotplugging on i915,
finally!

Cc: Juston Li <juston.li@intel.com>
Cc: Imre Deak <imre.deak@intel.com>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Cc: Harry Wentland <hwentlan@amd.com>
Signed-off-by: Lyude Paul <lyude@redhat.com>
---
 drivers/gpu/drm/drm_dp_mst_topology.c | 148 +++++++++++++++++++-------
 include/drm/drm_dp_mst_helper.h       |  16 +++
 2 files changed, 123 insertions(+), 41 deletions(-)

diff --git a/drivers/gpu/drm/drm_dp_mst_topology.c b/drivers/gpu/drm/drm_dp_mst_topology.c
index 35ced8514e18..ee28b372afa4 100644
--- a/drivers/gpu/drm/drm_dp_mst_topology.c
+++ b/drivers/gpu/drm/drm_dp_mst_topology.c
@@ -45,6 +45,12 @@
  * protocol. The helpers contain a topology manager and bandwidth manager.
  * The helpers encapsulate the sending and received of sideband msgs.
  */
+struct drm_dp_pending_up_req {
+	struct drm_dp_sideband_msg_hdr hdr;
+	struct drm_dp_sideband_msg_req_body msg;
+	struct list_head next;
+};
+
 static bool dump_dp_payload_table(struct drm_dp_mst_topology_mgr *mgr,
 				  char *buf);
 
@@ -1178,7 +1184,7 @@ static int drm_dp_mst_wait_tx_reply(struct drm_dp_mst_branch *mstb,
 		}
 	}
 out:
-	if (unlikely(ret == -EIO && drm_debug & (DRM_UT_DP | DRM_UT_KMS))) {
+	if (ret == -EIO && unlikely(drm_debug & (DRM_UT_DP | DRM_UT_KMS))) {
 		struct drm_printer p = drm_debug_printer(DBG_PREFIX);
 
 		drm_dp_mst_dump_sideband_msg_tx(&p, txmsg);
@@ -2990,6 +2996,7 @@ void drm_dp_mst_topology_mgr_suspend(struct drm_dp_mst_topology_mgr *mgr)
 	drm_dp_dpcd_writeb(mgr->aux, DP_MSTM_CTRL,
 			   DP_MST_EN | DP_UPSTREAM_IS_SRC);
 	mutex_unlock(&mgr->lock);
+	flush_work(&mgr->up_req_work);
 	flush_work(&mgr->work);
 	flush_work(&mgr->destroy_connector_work);
 }
@@ -3162,12 +3169,70 @@ static int drm_dp_mst_handle_down_rep(struct drm_dp_mst_topology_mgr *mgr)
 	return 0;
 }
 
+static inline void
+drm_dp_mst_process_up_req(struct drm_dp_mst_topology_mgr *mgr,
+			  struct drm_dp_pending_up_req *up_req)
+{
+	struct drm_dp_mst_branch *mstb = NULL;
+	struct drm_dp_sideband_msg_req_body *msg = &up_req->msg;
+	struct drm_dp_sideband_msg_hdr *hdr = &up_req->hdr;
+
+	if (hdr->broadcast) {
+		const u8 *guid = NULL;
+
+		if (msg->req_type == DP_CONNECTION_STATUS_NOTIFY)
+			guid = msg->u.conn_stat.guid;
+		else if (msg->req_type == DP_RESOURCE_STATUS_NOTIFY)
+			guid = msg->u.resource_stat.guid;
+
+		mstb = drm_dp_get_mst_branch_device_by_guid(mgr, guid);
+	} else {
+		mstb = drm_dp_get_mst_branch_device(mgr, hdr->lct, hdr->rad);
+	}
+
+	if (!mstb) {
+		DRM_DEBUG_KMS("Got MST reply from unknown device %d\n",
+			      hdr->lct);
+		return;
+	}
+
+	/* TODO: Add missing handler for DP_RESOURCE_STATUS_NOTIFY events */
+	if (msg->req_type == DP_CONNECTION_STATUS_NOTIFY) {
+		drm_dp_mst_handle_conn_stat(mstb, &msg->u.conn_stat);
+		drm_kms_helper_hotplug_event(mgr->dev);
+	}
+
+	drm_dp_mst_topology_put_mstb(mstb);
+}
+
+static void drm_dp_mst_up_req_work(struct work_struct *work)
+{
+	struct drm_dp_mst_topology_mgr *mgr =
+		container_of(work, struct drm_dp_mst_topology_mgr,
+			     up_req_work);
+	struct drm_dp_pending_up_req *up_req;
+
+	while (true) {
+		mutex_lock(&mgr->up_req_lock);
+		up_req = list_first_entry_or_null(&mgr->up_req_list,
+						  struct drm_dp_pending_up_req,
+						  next);
+		if (up_req)
+			list_del(&up_req->next);
+		mutex_unlock(&mgr->up_req_lock);
+
+		if (!up_req)
+			break;
+
+		drm_dp_mst_process_up_req(mgr, up_req);
+		kfree(up_req);
+	}
+}
+
 static int drm_dp_mst_handle_up_req(struct drm_dp_mst_topology_mgr *mgr)
 {
-	struct drm_dp_sideband_msg_req_body msg;
 	struct drm_dp_sideband_msg_hdr *hdr = &mgr->up_req_recv.initial_hdr;
-	struct drm_dp_mst_branch *mstb = NULL;
-	const u8 *guid;
+	struct drm_dp_pending_up_req *up_req;
 	bool seqno;
 
 	if (!drm_dp_get_one_sb_msg(mgr, true))
@@ -3176,56 +3241,53 @@ static int drm_dp_mst_handle_up_req(struct drm_dp_mst_topology_mgr *mgr)
 	if (!mgr->up_req_recv.have_eomt)
 		return 0;
 
-	if (!hdr->broadcast) {
-		mstb = drm_dp_get_mst_branch_device(mgr, hdr->lct, hdr->rad);
-		if (!mstb) {
-			DRM_DEBUG_KMS("Got MST reply from unknown device %d\n",
-				      hdr->lct);
-			goto out;
-		}
+	up_req = kzalloc(sizeof(*up_req), GFP_KERNEL);
+	if (!up_req) {
+		DRM_ERROR("Not enough memory to process MST up req\n");
+		return -ENOMEM;
 	}
+	INIT_LIST_HEAD(&up_req->next);
 
 	seqno = hdr->seqno;
-	drm_dp_sideband_parse_req(&mgr->up_req_recv, &msg);
+	drm_dp_sideband_parse_req(&mgr->up_req_recv, &up_req->msg);
 
-	if (msg.req_type == DP_CONNECTION_STATUS_NOTIFY)
-		guid = msg.u.conn_stat.guid;
-	else if (msg.req_type == DP_RESOURCE_STATUS_NOTIFY)
-		guid = msg.u.resource_stat.guid;
-	else
+	if (up_req->msg.req_type != DP_CONNECTION_STATUS_NOTIFY &&
+	    up_req->msg.req_type != DP_RESOURCE_STATUS_NOTIFY) {
+		DRM_DEBUG_KMS("Received unknown up req type, ignoring: %x\n",
+			      up_req->msg.req_type);
+		kfree(up_req);
 		goto out;
-
-	drm_dp_send_up_ack_reply(mgr, mgr->mst_primary, msg.req_type, seqno,
-				 false);
-
-	if (!mstb) {
-		mstb = drm_dp_get_mst_branch_device_by_guid(mgr, guid);
-		if (!mstb) {
-			DRM_DEBUG_KMS("Got MST reply from unknown device %d\n",
-				      hdr->lct);
-			goto out;
-		}
 	}
 
-	if (msg.req_type == DP_CONNECTION_STATUS_NOTIFY) {
-		drm_dp_mst_handle_conn_stat(mstb, &msg.u.conn_stat);
+	drm_dp_send_up_ack_reply(mgr, mgr->mst_primary, up_req->msg.req_type,
+				 seqno, false);
+
+	if (up_req->msg.req_type == DP_CONNECTION_STATUS_NOTIFY) {
+		const struct drm_dp_connection_status_notify *conn_stat =
+			&up_req->msg.u.conn_stat;
 
 		DRM_DEBUG_KMS("Got CSN: pn: %d ldps:%d ddps: %d mcs: %d ip: %d pdt: %d\n",
-			      msg.u.conn_stat.port_number,
-			      msg.u.conn_stat.legacy_device_plug_status,
-			      msg.u.conn_stat.displayport_device_plug_status,
-			      msg.u.conn_stat.message_capability_status,
-			      msg.u.conn_stat.input_port,
-			      msg.u.conn_stat.peer_device_type);
+			      conn_stat->port_number,
+			      conn_stat->legacy_device_plug_status,
+			      conn_stat->displayport_device_plug_status,
+			      conn_stat->message_capability_status,
+			      conn_stat->input_port,
+			      conn_stat->peer_device_type);
+	} else if (up_req->msg.req_type == DP_RESOURCE_STATUS_NOTIFY) {
+		const struct drm_dp_resource_status_notify *res_stat =
+			&up_req->msg.u.resource_stat;
 
-		drm_kms_helper_hotplug_event(mgr->dev);
-	} else if (msg.req_type == DP_RESOURCE_STATUS_NOTIFY) {
 		DRM_DEBUG_KMS("Got RSN: pn: %d avail_pbn %d\n",
-			      msg.u.resource_stat.port_number,
-			      msg.u.resource_stat.available_pbn);
+			      res_stat->port_number,
+			      res_stat->available_pbn);
 	}
 
-	drm_dp_mst_topology_put_mstb(mstb);
+	up_req->hdr = *hdr;
+	mutex_lock(&mgr->up_req_lock);
+	list_add_tail(&up_req->next, &mgr->up_req_list);
+	mutex_unlock(&mgr->up_req_lock);
+	queue_work(system_long_wq, &mgr->up_req_work);
+
 out:
 	memset(&mgr->up_req_recv, 0, sizeof(struct drm_dp_sideband_msg_rx));
 	return 0;
@@ -4186,12 +4248,15 @@ int drm_dp_mst_topology_mgr_init(struct drm_dp_mst_topology_mgr *mgr,
 	mutex_init(&mgr->qlock);
 	mutex_init(&mgr->payload_lock);
 	mutex_init(&mgr->destroy_connector_lock);
+	mutex_init(&mgr->up_req_lock);
 	INIT_LIST_HEAD(&mgr->tx_msg_downq);
 	INIT_LIST_HEAD(&mgr->destroy_connector_list);
 	INIT_LIST_HEAD(&mgr->destroy_branch_device_list);
+	INIT_LIST_HEAD(&mgr->up_req_list);
 	INIT_WORK(&mgr->work, drm_dp_mst_link_probe_work);
 	INIT_WORK(&mgr->tx_work, drm_dp_tx_work);
 	INIT_WORK(&mgr->destroy_connector_work, drm_dp_destroy_connector_work);
+	INIT_WORK(&mgr->up_req_work, drm_dp_mst_up_req_work);
 	init_waitqueue_head(&mgr->tx_waitq);
 	mgr->dev = dev;
 	mgr->aux = aux;
@@ -4248,6 +4313,7 @@ void drm_dp_mst_topology_mgr_destroy(struct drm_dp_mst_topology_mgr *mgr)
 	mutex_destroy(&mgr->payload_lock);
 	mutex_destroy(&mgr->qlock);
 	mutex_destroy(&mgr->lock);
+	mutex_destroy(&mgr->up_req_lock);
 }
 EXPORT_SYMBOL(drm_dp_mst_topology_mgr_destroy);
 
diff --git a/include/drm/drm_dp_mst_helper.h b/include/drm/drm_dp_mst_helper.h
index 8e04a1bd2e9d..ac1a68574a1a 100644
--- a/include/drm/drm_dp_mst_helper.h
+++ b/include/drm/drm_dp_mst_helper.h
@@ -595,6 +595,22 @@ struct drm_dp_mst_topology_mgr {
 	 * avoid locking inversion.
 	 */
 	struct work_struct destroy_connector_work;
+
+	/**
+	 * @up_req_list: List of pending up requests from the topology that
+	 * need to be processed, in chronological order.
+	 */
+	struct list_head up_req_list;
+	/**
+	 * @up_req_lock: Protects @up_req_list
+	 */
+	struct mutex up_req_lock;
+	/**
+	 * @up_req_work: Work item to process up requests received from the
+	 * topology. Needed to avoid blocking hotplug handling and sideband
+	 * transmissions.
+	 */
+	struct work_struct up_req_work;
 };
 
 int drm_dp_mst_topology_mgr_init(struct drm_dp_mst_topology_mgr *mgr,
-- 
2.21.0

[PATCH 19/26] drm/dp_mst: Protect drm_dp_mst_port members with connection_mutex

[Lyude Paul]

Yes-you read that right. Currently there is literally no locking in
place for any of the drm_dp_mst_port struct members that can be modified
in response to a link address response, or a connection status response.
Which literally means if we're unlucky enough to have any sort of
hotplugging event happen before we're finished with reprobing link
addresses, we'll race and the contents of said struct members becomes
undefined. Fun!

So, finally add some simple locking protections to our MST helpers by
protecting any drm_dp_mst_port members which can be changed by link
address responses or connection status notifications under
drm_device->mode_config.connection_mutex.

Cc: Juston Li <juston.li@intel.com>
Cc: Imre Deak <imre.deak@intel.com>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Cc: Harry Wentland <hwentlan@amd.com>
Signed-off-by: Lyude Paul <lyude@redhat.com>
---
 drivers/gpu/drm/drm_dp_mst_topology.c | 144 +++++++++++++++++++-------
 include/drm/drm_dp_mst_helper.h       |  39 +++++--
 2 files changed, 133 insertions(+), 50 deletions(-)

diff --git a/drivers/gpu/drm/drm_dp_mst_topology.c b/drivers/gpu/drm/drm_dp_mst_topology.c
index ee28b372afa4..dffd80f31537 100644
--- a/drivers/gpu/drm/drm_dp_mst_topology.c
+++ b/drivers/gpu/drm/drm_dp_mst_topology.c
@@ -1352,6 +1352,7 @@ static void drm_dp_free_mst_port(struct kref *kref)
 		container_of(kref, struct drm_dp_mst_port, malloc_kref);
 
 	drm_dp_mst_put_mstb_malloc(port->parent);
+	mutex_destroy(&port->lock);
 	kfree(port);
 }
 
@@ -1818,6 +1819,36 @@ static void build_mst_prop_path(const struct drm_dp_mst_branch *mstb,
 	strlcat(proppath, temp, proppath_size);
 }
 
+static void
+drm_dp_mst_port_add_connector(struct drm_dp_mst_branch *mstb,
+			      struct drm_dp_mst_port *port)
+{
+	struct drm_dp_mst_topology_mgr *mgr = port->mgr;
+	char proppath[255];
+	int ret;
+
+	build_mst_prop_path(mstb, port->port_num, proppath, sizeof(proppath));
+	port->connector = mgr->cbs->add_connector(mgr, port, proppath);
+	if (!port->connector) {
+		ret = -ENOMEM;
+		goto error;
+	}
+
+	if ((port->pdt == DP_PEER_DEVICE_DP_LEGACY_CONV ||
+	     port->pdt == DP_PEER_DEVICE_SST_SINK) &&
+	    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);
+	}
+
+	mgr->cbs->register_connector(port->connector);
+	return;
+
+error:
+	DRM_ERROR("Failed to create connector for port %p: %d\n", port, ret);
+}
+
 static void
 drm_dp_mst_handle_link_address_port(struct drm_dp_mst_branch *mstb,
 				    struct drm_device *dev,
@@ -1825,8 +1856,12 @@ drm_dp_mst_handle_link_address_port(struct drm_dp_mst_branch *mstb,
 {
 	struct drm_dp_mst_topology_mgr *mgr = mstb->mgr;
 	struct drm_dp_mst_port *port;
-	bool created = false;
-	int old_ddps = 0;
+	struct drm_dp_mst_branch *child_mstb = NULL;
+	struct drm_connector *connector_to_destroy = NULL;
+	int old_ddps = 0, ret;
+	u8 new_pdt = DP_PEER_DEVICE_NONE;
+	bool created = false, send_link_addr = false,
+	     create_connector = false;
 
 	port = drm_dp_get_port(mstb, port_msg->port_number);
 	if (!port) {
@@ -1835,6 +1870,7 @@ drm_dp_mst_handle_link_address_port(struct drm_dp_mst_branch *mstb,
 			return;
 		kref_init(&port->topology_kref);
 		kref_init(&port->malloc_kref);
+		mutex_init(&port->lock);
 		port->parent = mstb;
 		port->port_num = port_msg->port_number;
 		port->mgr = mgr;
@@ -1848,11 +1884,17 @@ drm_dp_mst_handle_link_address_port(struct drm_dp_mst_branch *mstb,
 		drm_dp_mst_get_mstb_malloc(mstb);
 
 		created = true;
-	} else {
-		old_ddps = port->ddps;
 	}
 
+	mutex_lock(&port->lock);
+	drm_modeset_lock(&dev->mode_config.connection_mutex, NULL);
+
+	if (!created)
+		old_ddps = port->ddps;
+
 	port->input = port_msg->input_port;
+	if (!port->input)
+		new_pdt = port_msg->peer_device_type;
 	port->mcs = port_msg->mcs;
 	port->ddps = port_msg->ddps;
 	port->ldps = port_msg->legacy_device_plug_status;
@@ -1880,44 +1922,58 @@ drm_dp_mst_handle_link_address_port(struct drm_dp_mst_branch *mstb,
 		}
 	}
 
-	if (!port->input) {
-		int ret = drm_dp_port_set_pdt(port,
-					      port_msg->peer_device_type);
-		if (ret == 1) {
-			drm_dp_send_link_address(mgr, port->mstb);
-		} else if (ret < 0) {
-			DRM_ERROR("Failed to change PDT on port %p: %d\n",
-				  port, ret);
-			goto fail;
+	ret = drm_dp_port_set_pdt(port, new_pdt);
+	if (ret == 1) {
+		send_link_addr = true;
+	} else if (ret < 0) {
+		DRM_ERROR("Failed to change PDT on port %p: %d\n",
+			  port, ret);
+		goto fail_unlock;
+	}
+
+	if (send_link_addr) {
+		mutex_lock(&mgr->lock);
+		if (port->mstb) {
+			child_mstb = port->mstb;
+			drm_dp_mst_get_mstb_malloc(child_mstb);
 		}
+		mutex_unlock(&mgr->lock);
 	}
 
-	if (created && !port->input) {
-		char proppath[255];
+	/*
+	 * We unset port->connector before dropping connection_mutex so that
+	 * there's no chance any of the atomic MST helpers can accidentally
+	 * associate a to-be-destroyed connector with a port.
+	 */
+	if (port->connector && port->input) {
+		connector_to_destroy = port->connector;
+		port->connector = NULL;
+	} else if (!port->connector && !port->input) {
+		create_connector = true;
+	}
 
-		build_mst_prop_path(mstb, port->port_num, proppath,
-				    sizeof(proppath));
-		port->connector = (*mgr->cbs->add_connector)(mgr, port,
-							     proppath);
-		if (!port->connector)
-			goto fail;
+	drm_modeset_unlock(&dev->mode_config.connection_mutex);
 
-		if ((port->pdt == DP_PEER_DEVICE_DP_LEGACY_CONV ||
-		     port->pdt == DP_PEER_DEVICE_SST_SINK) &&
-		    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);
-		}
+	if (connector_to_destroy)
+		mgr->cbs->destroy_connector(mgr, connector_to_destroy);
+	else if (create_connector)
+		drm_dp_mst_port_add_connector(mstb, port);
+
+	mutex_unlock(&port->lock);
 
-		(*mgr->cbs->register_connector)(port->connector);
+	if (send_link_addr && child_mstb) {
+		drm_dp_send_link_address(mgr, child_mstb);
+		drm_dp_mst_put_mstb_malloc(child_mstb);
 	}
 
 	/* put reference to this port */
 	drm_dp_mst_topology_put_port(port);
 	return;
 
-fail:
+fail_unlock:
+	drm_modeset_unlock(&dev->mode_config.connection_mutex);
+	mutex_unlock(&port->lock);
+
 	/* Remove it from the port list */
 	mutex_lock(&mgr->lock);
 	list_del(&port->next);
@@ -1933,6 +1989,7 @@ static void
 drm_dp_mst_handle_conn_stat(struct drm_dp_mst_branch *mstb,
 			    struct drm_dp_connection_status_notify *conn_stat)
 {
+	struct drm_device *dev = mstb->mgr->dev;
 	struct drm_dp_mst_port *port;
 	int old_ddps;
 	bool dowork = false;
@@ -1941,6 +1998,8 @@ drm_dp_mst_handle_conn_stat(struct drm_dp_mst_branch *mstb,
 	if (!port)
 		return;
 
+	drm_modeset_lock(&dev->mode_config.connection_mutex, NULL);
+
 	old_ddps = port->ddps;
 	port->mcs = conn_stat->message_capability_status;
 	port->ldps = conn_stat->legacy_device_plug_status;
@@ -1966,6 +2025,7 @@ drm_dp_mst_handle_conn_stat(struct drm_dp_mst_branch *mstb,
 		}
 	}
 
+	drm_modeset_unlock(&dev->mode_config.connection_mutex);
 	drm_dp_mst_topology_put_port(port);
 	if (dowork)
 		queue_work(system_long_wq, &mstb->mgr->work);
@@ -2058,28 +2118,34 @@ drm_dp_get_mst_branch_device_by_guid(struct drm_dp_mst_topology_mgr *mgr,
 static void drm_dp_check_and_send_link_address(struct drm_dp_mst_topology_mgr *mgr,
 					       struct drm_dp_mst_branch *mstb)
 {
+	struct drm_device *dev = mgr->dev;
 	struct drm_dp_mst_port *port;
-	struct drm_dp_mst_branch *mstb_child;
+
 	if (!mstb->link_address_sent)
 		drm_dp_send_link_address(mgr, mstb);
 
 	list_for_each_entry(port, &mstb->ports, next) {
-		if (port->input)
-			continue;
+		struct drm_dp_mst_branch *mstb_child = NULL;
+
+		drm_modeset_lock(&dev->mode_config.connection_mutex, NULL);
 
-		if (!port->ddps)
+		if (port->input || !port->ddps) {
+			drm_modeset_unlock(&dev->mode_config.connection_mutex);
 			continue;
+		}
 
 		if (!port->available_pbn)
 			drm_dp_send_enum_path_resources(mgr, mstb, port);
 
-		if (port->mstb) {
+		if (port->mstb)
 			mstb_child = drm_dp_mst_topology_get_mstb_validated(
 			    mgr, port->mstb);
-			if (mstb_child) {
-				drm_dp_check_and_send_link_address(mgr, mstb_child);
-				drm_dp_mst_topology_put_mstb(mstb_child);
-			}
+
+		drm_modeset_unlock(&dev->mode_config.connection_mutex);
+
+		if (mstb_child) {
+			drm_dp_check_and_send_link_address(mgr, mstb_child);
+			drm_dp_mst_topology_put_mstb(mstb_child);
 		}
 	}
 }
diff --git a/include/drm/drm_dp_mst_helper.h b/include/drm/drm_dp_mst_helper.h
index ac1a68574a1a..aed68d7e6492 100644
--- a/include/drm/drm_dp_mst_helper.h
+++ b/include/drm/drm_dp_mst_helper.h
@@ -45,23 +45,34 @@ struct drm_dp_vcpi {
 /**
  * struct drm_dp_mst_port - MST port
  * @port_num: port number
- * @input: if this port is an input port.
- * @mcs: message capability status - DP 1.2 spec.
- * @ddps: DisplayPort Device Plug Status - DP 1.2
- * @pdt: Peer Device Type
- * @ldps: Legacy Device Plug Status
- * @dpcd_rev: DPCD revision of device on this port
- * @num_sdp_streams: Number of simultaneous streams
- * @num_sdp_stream_sinks: Number of stream sinks
- * @available_pbn: Available bandwidth for this port.
+ * @input: if this port is an input port. Protected by
+ * &drm_device.mode_config.connection_mutex.
+ * @mcs: message capability status - DP 1.2 spec. Protected by
+ * &drm_device.mode_config.connection_mutex.
+ * @ddps: DisplayPort Device Plug Status - DP 1.2. Protected by
+ * &drm_device.mode_config.connection_mutex.
+ * @pdt: Peer Device Type. Protected by
+ * &drm_device.mode_config.connection_mutex.
+ * @ldps: Legacy Device Plug Status. Protected by
+ * &drm_device.mode_config.connection_mutex.
+ * @dpcd_rev: DPCD revision of device on this port. Protected by
+ * &drm_device.mode_config.connection_mutex.
+ * @num_sdp_streams: Number of simultaneous streams. Protected by
+ * &drm_device.mode_config.connection_mutex.
+ * @num_sdp_stream_sinks: Number of stream sinks. Protected by
+ * &drm_device.mode_config.connection_mutex.
+ * @available_pbn: Available bandwidth for this port. Protected by
+ * &drm_device.mode_config.connection_mutex.
  * @next: link to next port on this branch device
  * @mstb: branch device on this port, protected by
  * &drm_dp_mst_topology_mgr.lock
  * @aux: i2c aux transport to talk to device connected to this port, protected
- * by &drm_dp_mst_topology_mgr.lock
+ * by &drm_device.mode_config.connection_mutex.
  * @parent: branch device parent of this port
  * @vcpi: Virtual Channel Payload info for this port.
- * @connector: DRM connector this port is connected to.
+ * @connector: DRM connector this port is connected to. Protected by @lock.
+ * When there is already a connector registered for this port, this is also
+ * protected by &drm_device.mode_config.connection_mutex.
  * @mgr: topology manager this port lives under.
  *
  * This structure represents an MST port endpoint on a device somewhere
@@ -100,6 +111,12 @@ struct drm_dp_mst_port {
 	struct drm_connector *connector;
 	struct drm_dp_mst_topology_mgr *mgr;
 
+	/**
+	 * @lock: Protects @connector. If needed, this lock should be grabbed
+	 * before &drm_device.mode_config.connection_mutex.
+	 */
+	struct mutex lock;
+
 	/**
 	 * @cached_edid: for DP logical ports - make tiling work by ensuring
 	 * that the EDID for all connectors is read immediately.
-- 
2.21.0

[PATCH 20/26] drm/dp_mst: Don't forget to update port->input in drm_dp_mst_handle_conn_stat()

[Lyude Paul]

This probably hasn't caused any problems up until now since it's
probably nearly impossible to encounter this in the wild, however if we
were to receive a connection status notification from the MST hub after
resume while we're in the middle of reprobing the link addresses for a
topology then there's a much larger chance that a port could have
changed from being an output port to input port (or vice versa). If we
forget to update this bit of information, we'll potentially ignore a
valid PDT change on a downstream port because we think it's an input
port.

So, make sure we read the input_port field in connection status
notifications in drm_dp_mst_handle_conn_stat() to prevent this from
happening once we've implemented suspend/resume reprobing.

Cc: Juston Li <juston.li@intel.com>
Cc: Imre Deak <imre.deak@intel.com>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Cc: Harry Wentland <hwentlan@amd.com>
Signed-off-by: Lyude Paul <lyude@redhat.com>
---
 drivers/gpu/drm/drm_dp_mst_topology.c | 51 +++++++++++++++++++--------
 1 file changed, 37 insertions(+), 14 deletions(-)

diff --git a/drivers/gpu/drm/drm_dp_mst_topology.c b/drivers/gpu/drm/drm_dp_mst_topology.c
index dffd80f31537..126db36c9337 100644
--- a/drivers/gpu/drm/drm_dp_mst_topology.c
+++ b/drivers/gpu/drm/drm_dp_mst_topology.c
@@ -1989,18 +1989,23 @@ static void
 drm_dp_mst_handle_conn_stat(struct drm_dp_mst_branch *mstb,
 			    struct drm_dp_connection_status_notify *conn_stat)
 {
-	struct drm_device *dev = mstb->mgr->dev;
+	struct drm_dp_mst_topology_mgr *mgr = mstb->mgr;
+	struct drm_device *dev = mgr->dev;
 	struct drm_dp_mst_port *port;
-	int old_ddps;
-	bool dowork = false;
+	struct drm_connector *connector_to_destroy = NULL;
+	int old_ddps, ret;
+	u8 new_pdt;
+	bool dowork = false, create_connector = false;
 
 	port = drm_dp_get_port(mstb, conn_stat->port_number);
 	if (!port)
 		return;
 
+	mutex_lock(&port->lock);
 	drm_modeset_lock(&dev->mode_config.connection_mutex, NULL);
 
 	old_ddps = port->ddps;
+	port->input = conn_stat->input_port;
 	port->mcs = conn_stat->message_capability_status;
 	port->ldps = conn_stat->legacy_device_plug_status;
 	port->ddps = conn_stat->displayport_device_plug_status;
@@ -2013,23 +2018,41 @@ drm_dp_mst_handle_conn_stat(struct drm_dp_mst_branch *mstb,
 		}
 	}
 
-	if (!port->input) {
-		int ret = drm_dp_port_set_pdt(port,
-					      conn_stat->peer_device_type);
-		if (ret == 1) {
-			dowork = true;
-		} else if (ret < 0) {
-			DRM_ERROR("Failed to change PDT for port %p: %d\n",
-				  port, ret);
-			dowork = false;
-		}
+	new_pdt = port->input ? DP_PEER_DEVICE_NONE : conn_stat->peer_device_type;
+
+	ret = drm_dp_port_set_pdt(port, new_pdt);
+	if (ret == 1) {
+		dowork = true;
+	} else if (ret < 0) {
+		DRM_ERROR("Failed to change PDT for port %p: %d\n",
+			  port, ret);
+		dowork = false;
+	}
+
+	/*
+	 * We unset port->connector before dropping connection_mutex so that
+	 * there's no chance any of the atomic MST helpers can accidentally
+	 * associate a to-be-destroyed connector with a port.
+	 */
+	if (port->connector && port->input) {
+		connector_to_destroy = port->connector;
+		port->connector = NULL;
+	} else if (!port->connector && !port->input) {
+		create_connector = true;
 	}
 
 	drm_modeset_unlock(&dev->mode_config.connection_mutex);
+
+	if (connector_to_destroy)
+		mgr->cbs->destroy_connector(mgr, connector_to_destroy);
+	else if (create_connector)
+		drm_dp_mst_port_add_connector(mstb, port);
+
+	mutex_unlock(&port->lock);
+
 	drm_dp_mst_topology_put_port(port);
 	if (dowork)
 		queue_work(system_long_wq, &mstb->mgr->work);
-
 }
 
 static struct drm_dp_mst_branch *drm_dp_get_mst_branch_device(struct drm_dp_mst_topology_mgr *mgr,
-- 
2.21.0

[PATCH 21/26] drm/nouveau: Don't grab runtime PM refs for HPD IRQs

[Lyude Paul]

In order for suspend/resume reprobing to work, we need to be able to
perform sideband communications during suspend/resume, along with
runtime PM suspend/resume. In order to do so, we also need to make sure
that nouveau doesn't bother grabbing a runtime PM reference to do so,
since otherwise we'll start deadlocking runtime PM again.

Note that we weren't able to do this before, because of the DP MST
helpers processing UP requests from topologies in the same context as
drm_dp_mst_hpd_irq() which would have caused us to open ourselves up to
receiving hotplug events and deadlocking with runtime suspend/resume.
Now that those requests are handled asynchronously, this change should
be completely safe.

Cc: Juston Li <juston.li@intel.com>
Cc: Imre Deak <imre.deak@intel.com>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Cc: Harry Wentland <hwentlan@amd.com>
Signed-off-by: Lyude Paul <lyude@redhat.com>
---
 drivers/gpu/drm/nouveau/nouveau_connector.c | 33 +++++++++++----------
 1 file changed, 17 insertions(+), 16 deletions(-)

diff --git a/drivers/gpu/drm/nouveau/nouveau_connector.c b/drivers/gpu/drm/nouveau/nouveau_connector.c
index 4116ee62adaf..e9e78696a728 100644
--- a/drivers/gpu/drm/nouveau/nouveau_connector.c
+++ b/drivers/gpu/drm/nouveau/nouveau_connector.c
@@ -1129,6 +1129,16 @@ nouveau_connector_hotplug(struct nvif_notify *notify)
 	const char *name = connector->name;
 	struct nouveau_encoder *nv_encoder;
 	int ret;
+	bool plugged = (rep->mask != NVIF_NOTIFY_CONN_V0_UNPLUG);
+
+	if (rep->mask & NVIF_NOTIFY_CONN_V0_IRQ) {
+		NV_DEBUG(drm, "service %s\n", name);
+		drm_dp_cec_irq(&nv_connector->aux);
+		if ((nv_encoder = find_encoder(connector, DCB_OUTPUT_DP)))
+			nv50_mstm_service(nv_encoder->dp.mstm);
+
+		return NVIF_NOTIFY_KEEP;
+	}
 
 	ret = pm_runtime_get(drm->dev->dev);
 	if (ret == 0) {
@@ -1149,25 +1159,16 @@ nouveau_connector_hotplug(struct nvif_notify *notify)
 		return NVIF_NOTIFY_DROP;
 	}
 
-	if (rep->mask & NVIF_NOTIFY_CONN_V0_IRQ) {
-		NV_DEBUG(drm, "service %s\n", name);
-		drm_dp_cec_irq(&nv_connector->aux);
-		if ((nv_encoder = find_encoder(connector, DCB_OUTPUT_DP)))
-			nv50_mstm_service(nv_encoder->dp.mstm);
-	} else {
-		bool plugged = (rep->mask != NVIF_NOTIFY_CONN_V0_UNPLUG);
-
+	if (!plugged)
+		drm_dp_cec_unset_edid(&nv_connector->aux);
+	NV_DEBUG(drm, "%splugged %s\n", plugged ? "" : "un", name);
+	if ((nv_encoder = find_encoder(connector, DCB_OUTPUT_DP))) {
 		if (!plugged)
-			drm_dp_cec_unset_edid(&nv_connector->aux);
-		NV_DEBUG(drm, "%splugged %s\n", plugged ? "" : "un", name);
-		if ((nv_encoder = find_encoder(connector, DCB_OUTPUT_DP))) {
-			if (!plugged)
-				nv50_mstm_remove(nv_encoder->dp.mstm);
-		}
-
-		drm_helper_hpd_irq_event(connector->dev);
+			nv50_mstm_remove(nv_encoder->dp.mstm);
 	}
 
+	drm_helper_hpd_irq_event(connector->dev);
+
 	pm_runtime_mark_last_busy(drm->dev->dev);
 	pm_runtime_put_autosuspend(drm->dev->dev);
 	return NVIF_NOTIFY_KEEP;
-- 
2.21.0

[PATCH 22/26] drm/amdgpu: Iterate through DRM connectors correctly

[Lyude Paul]

Currently, every single piece of code in amdgpu that loops through
connectors does it incorrectly and doesn't use the proper list iteration
helpers, drm_connector_list_iter_begin() and
drm_connector_list_iter_end(). Yeesh.

So, do that.

Cc: Juston Li <juston.li@intel.com>
Cc: Imre Deak <imre.deak@intel.com>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Cc: Harry Wentland <hwentlan@amd.com>
Signed-off-by: Lyude Paul <lyude@redhat.com>
---
 .../gpu/drm/amd/amdgpu/amdgpu_connectors.c    | 13 +++++-
 drivers/gpu/drm/amd/amdgpu/amdgpu_device.c    | 20 +++++++---
 drivers/gpu/drm/amd/amdgpu/amdgpu_display.c   |  5 ++-
 drivers/gpu/drm/amd/amdgpu/amdgpu_encoders.c  | 40 +++++++++++++------
 drivers/gpu/drm/amd/amdgpu/amdgpu_irq.c       |  5 ++-
 drivers/gpu/drm/amd/amdgpu/dce_v10_0.c        | 34 ++++++++++++----
 drivers/gpu/drm/amd/amdgpu/dce_v11_0.c        | 34 ++++++++++++----
 drivers/gpu/drm/amd/amdgpu/dce_v6_0.c         | 40 ++++++++++++++-----
 drivers/gpu/drm/amd/amdgpu/dce_v8_0.c         | 34 ++++++++++++----
 .../gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c | 33 ++++++++-------
 .../drm/amd/display/amdgpu_dm/amdgpu_dm_irq.c | 10 ++++-
 11 files changed, 195 insertions(+), 73 deletions(-)

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_connectors.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_connectors.c
index 73b2ede773d3..2cabaaecf28a 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_connectors.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_connectors.c
@@ -1022,8 +1022,12 @@ amdgpu_connector_dvi_detect(struct drm_connector *connector, bool force)
 			 */
 			if (amdgpu_connector->shared_ddc && (ret == connector_status_connected)) {
 				struct drm_connector *list_connector;
+				struct drm_connector_list_iter iter;
 				struct amdgpu_connector *list_amdgpu_connector;
-				list_for_each_entry(list_connector, &dev->mode_config.connector_list, head) {
+
+				drm_connector_list_iter_begin(dev, &iter);
+				drm_for_each_connector_iter(list_connector,
+							    &iter) {
 					if (connector == list_connector)
 						continue;
 					list_amdgpu_connector = to_amdgpu_connector(list_connector);
@@ -1040,6 +1044,7 @@ amdgpu_connector_dvi_detect(struct drm_connector *connector, bool force)
 						}
 					}
 				}
+				drm_connector_list_iter_end(&iter);
 			}
 		}
 	}
@@ -1501,6 +1506,7 @@ amdgpu_connector_add(struct amdgpu_device *adev,
 {
 	struct drm_device *dev = adev->ddev;
 	struct drm_connector *connector;
+	struct drm_connector_list_iter iter;
 	struct amdgpu_connector *amdgpu_connector;
 	struct amdgpu_connector_atom_dig *amdgpu_dig_connector;
 	struct drm_encoder *encoder;
@@ -1514,10 +1520,12 @@ amdgpu_connector_add(struct amdgpu_device *adev,
 		return;
 
 	/* see if we already added it */
-	list_for_each_entry(connector, &dev->mode_config.connector_list, head) {
+	drm_connector_list_iter_begin(dev, &iter);
+	drm_for_each_connector_iter(connector, &iter) {
 		amdgpu_connector = to_amdgpu_connector(connector);
 		if (amdgpu_connector->connector_id == connector_id) {
 			amdgpu_connector->devices |= supported_device;
+			drm_connector_list_iter_end(&iter);
 			return;
 		}
 		if (amdgpu_connector->ddc_bus && i2c_bus->valid) {
@@ -1532,6 +1540,7 @@ amdgpu_connector_add(struct amdgpu_device *adev,
 			}
 		}
 	}
+	drm_connector_list_iter_end(&iter);
 
 	/* check if it's a dp bridge */
 	list_for_each_entry(encoder, &dev->mode_config.encoder_list, head) {
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
index 7401bc95c15b..f8d433b013e1 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
@@ -2909,6 +2909,7 @@ int amdgpu_device_suspend(struct drm_device *dev, bool suspend, bool fbcon)
 	struct amdgpu_device *adev;
 	struct drm_crtc *crtc;
 	struct drm_connector *connector;
+	struct drm_connector_list_iter iter;
 	int r;
 
 	if (dev == NULL || dev->dev_private == NULL) {
@@ -2931,9 +2932,11 @@ int amdgpu_device_suspend(struct drm_device *dev, bool suspend, bool fbcon)
 	if (!amdgpu_device_has_dc_support(adev)) {
 		/* turn off display hw */
 		drm_modeset_lock_all(dev);
-		list_for_each_entry(connector, &dev->mode_config.connector_list, head) {
-			drm_helper_connector_dpms(connector, DRM_MODE_DPMS_OFF);
-		}
+		drm_connector_list_iter_begin(dev, &iter);
+		drm_for_each_connector_iter(connector, &iter)
+			drm_helper_connector_dpms(connector,
+						  DRM_MODE_DPMS_OFF);
+		drm_connector_list_iter_end(&iter);
 		drm_modeset_unlock_all(dev);
 			/* unpin the front buffers and cursors */
 		list_for_each_entry(crtc, &dev->mode_config.crtc_list, head) {
@@ -3012,6 +3015,7 @@ int amdgpu_device_suspend(struct drm_device *dev, bool suspend, bool fbcon)
 int amdgpu_device_resume(struct drm_device *dev, bool resume, bool fbcon)
 {
 	struct drm_connector *connector;
+	struct drm_connector_list_iter iter;
 	struct amdgpu_device *adev = dev->dev_private;
 	struct drm_crtc *crtc;
 	int r = 0;
@@ -3082,9 +3086,13 @@ int amdgpu_device_resume(struct drm_device *dev, bool resume, bool fbcon)
 
 			/* turn on display hw */
 			drm_modeset_lock_all(dev);
-			list_for_each_entry(connector, &dev->mode_config.connector_list, head) {
-				drm_helper_connector_dpms(connector, DRM_MODE_DPMS_ON);
-			}
+
+			drm_connector_list_iter_begin(dev, &iter);
+			drm_for_each_connector_iter(connector, &iter)
+				drm_helper_connector_dpms(connector,
+							  DRM_MODE_DPMS_ON);
+			drm_connector_list_iter_end(&iter);
+
 			drm_modeset_unlock_all(dev);
 		}
 		amdgpu_fbdev_set_suspend(adev, 0);
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_display.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_display.c
index 535650967b1a..d342471ad3bb 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_display.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_display.c
@@ -369,11 +369,13 @@ void amdgpu_display_print_display_setup(struct drm_device *dev)
 	struct amdgpu_connector *amdgpu_connector;
 	struct drm_encoder *encoder;
 	struct amdgpu_encoder *amdgpu_encoder;
+	struct drm_connector_list_iter iter;
 	uint32_t devices;
 	int i = 0;
 
+	drm_connector_list_iter_begin(dev, &iter);
 	DRM_INFO("AMDGPU Display Connectors\n");
-	list_for_each_entry(connector, &dev->mode_config.connector_list, head) {
+	drm_for_each_connector_iter(connector, &iter) {
 		amdgpu_connector = to_amdgpu_connector(connector);
 		DRM_INFO("Connector %d:\n", i);
 		DRM_INFO("  %s\n", connector->name);
@@ -437,6 +439,7 @@ void amdgpu_display_print_display_setup(struct drm_device *dev)
 		}
 		i++;
 	}
+	drm_connector_list_iter_end(&iter);
 }
 
 /**
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_encoders.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_encoders.c
index 571a6dfb473e..61fcf247a638 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_encoders.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_encoders.c
@@ -37,12 +37,14 @@ amdgpu_link_encoder_connector(struct drm_device *dev)
 {
 	struct amdgpu_device *adev = dev->dev_private;
 	struct drm_connector *connector;
+	struct drm_connector_list_iter iter;
 	struct amdgpu_connector *amdgpu_connector;
 	struct drm_encoder *encoder;
 	struct amdgpu_encoder *amdgpu_encoder;
 
+	drm_connector_list_iter_begin(dev, &iter);
 	/* walk the list and link encoders to connectors */
-	list_for_each_entry(connector, &dev->mode_config.connector_list, head) {
+	drm_for_each_connector_iter(connector, &iter) {
 		amdgpu_connector = to_amdgpu_connector(connector);
 		list_for_each_entry(encoder, &dev->mode_config.encoder_list, head) {
 			amdgpu_encoder = to_amdgpu_encoder(encoder);
@@ -55,6 +57,7 @@ amdgpu_link_encoder_connector(struct drm_device *dev)
 			}
 		}
 	}
+	drm_connector_list_iter_end(&iter);
 }
 
 void amdgpu_encoder_set_active_device(struct drm_encoder *encoder)
@@ -62,8 +65,10 @@ void amdgpu_encoder_set_active_device(struct drm_encoder *encoder)
 	struct drm_device *dev = encoder->dev;
 	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
 	struct drm_connector *connector;
+	struct drm_connector_list_iter iter;
 
-	list_for_each_entry(connector, &dev->mode_config.connector_list, head) {
+	drm_connector_list_iter_begin(dev, &iter);
+	drm_for_each_connector_iter(connector, &iter) {
 		if (connector->encoder == encoder) {
 			struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
 			amdgpu_encoder->active_device = amdgpu_encoder->devices & amdgpu_connector->devices;
@@ -72,6 +77,7 @@ void amdgpu_encoder_set_active_device(struct drm_encoder *encoder)
 				  amdgpu_connector->devices, encoder->encoder_type);
 		}
 	}
+	drm_connector_list_iter_end(&iter);
 }
 
 struct drm_connector *
@@ -79,15 +85,20 @@ amdgpu_get_connector_for_encoder(struct drm_encoder *encoder)
 {
 	struct drm_device *dev = encoder->dev;
 	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
-	struct drm_connector *connector;
+	struct drm_connector *connector, *found = NULL;
+	struct drm_connector_list_iter iter;
 	struct amdgpu_connector *amdgpu_connector;
 
-	list_for_each_entry(connector, &dev->mode_config.connector_list, head) {
+	drm_connector_list_iter_begin(dev, &iter);
+	drm_for_each_connector_iter(connector, &iter) {
 		amdgpu_connector = to_amdgpu_connector(connector);
-		if (amdgpu_encoder->active_device & amdgpu_connector->devices)
-			return connector;
+		if (amdgpu_encoder->active_device & amdgpu_connector->devices) {
+			found = connector;
+			break;
+		}
 	}
-	return NULL;
+	drm_connector_list_iter_end(&iter);
+	return found;
 }
 
 struct drm_connector *
@@ -95,15 +106,20 @@ amdgpu_get_connector_for_encoder_init(struct drm_encoder *encoder)
 {
 	struct drm_device *dev = encoder->dev;
 	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
-	struct drm_connector *connector;
+	struct drm_connector *connector, *found = NULL;
+	struct drm_connector_list_iter iter;
 	struct amdgpu_connector *amdgpu_connector;
 
-	list_for_each_entry(connector, &dev->mode_config.connector_list, head) {
+	drm_connector_list_iter_begin(dev, &iter);
+	drm_for_each_connector_iter(connector, &iter) {
 		amdgpu_connector = to_amdgpu_connector(connector);
-		if (amdgpu_encoder->devices & amdgpu_connector->devices)
-			return connector;
+		if (amdgpu_encoder->devices & amdgpu_connector->devices) {
+			found = connector;
+			break;
+		}
 	}
-	return NULL;
+	drm_connector_list_iter_end(&iter);
+	return found;
 }
 
 struct drm_encoder *amdgpu_get_external_encoder(struct drm_encoder *encoder)
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.c
index 2a3f5ec298db..977e121204e6 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.c
@@ -87,10 +87,13 @@ static void amdgpu_hotplug_work_func(struct work_struct *work)
 	struct drm_device *dev = adev->ddev;
 	struct drm_mode_config *mode_config = &dev->mode_config;
 	struct drm_connector *connector;
+	struct drm_connector_list_iter iter;
 
 	mutex_lock(&mode_config->mutex);
-	list_for_each_entry(connector, &mode_config->connector_list, head)
+	drm_connector_list_iter_begin(dev, &iter);
+	drm_for_each_connector_iter(connector, &iter)
 		amdgpu_connector_hotplug(connector);
+	drm_connector_list_iter_end(&iter);
 	mutex_unlock(&mode_config->mutex);
 	/* Just fire off a uevent and let userspace tell us what to do */
 	drm_helper_hpd_irq_event(dev);
diff --git a/drivers/gpu/drm/amd/amdgpu/dce_v10_0.c b/drivers/gpu/drm/amd/amdgpu/dce_v10_0.c
index 1ffd1963e765..957335d41c56 100644
--- a/drivers/gpu/drm/amd/amdgpu/dce_v10_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/dce_v10_0.c
@@ -326,9 +326,11 @@ static void dce_v10_0_hpd_init(struct amdgpu_device *adev)
 {
 	struct drm_device *dev = adev->ddev;
 	struct drm_connector *connector;
+	struct drm_connector_list_iter iter;
 	u32 tmp;
 
-	list_for_each_entry(connector, &dev->mode_config.connector_list, head) {
+	drm_connector_list_iter_begin(dev, &iter);
+	drm_for_each_connector_iter(connector, &iter) {
 		struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
 
 		if (amdgpu_connector->hpd.hpd >= adev->mode_info.num_hpd)
@@ -364,6 +366,7 @@ static void dce_v10_0_hpd_init(struct amdgpu_device *adev)
 		amdgpu_irq_get(adev, &adev->hpd_irq,
 			       amdgpu_connector->hpd.hpd);
 	}
+	drm_connector_list_iter_end(&iter);
 }
 
 /**
@@ -378,9 +381,11 @@ static void dce_v10_0_hpd_fini(struct amdgpu_device *adev)
 {
 	struct drm_device *dev = adev->ddev;
 	struct drm_connector *connector;
+	struct drm_connector_list_iter iter;
 	u32 tmp;
 
-	list_for_each_entry(connector, &dev->mode_config.connector_list, head) {
+	drm_connector_list_iter_begin(dev, &iter);
+	drm_for_each_connector_iter(connector, &iter) {
 		struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
 
 		if (amdgpu_connector->hpd.hpd >= adev->mode_info.num_hpd)
@@ -393,6 +398,7 @@ static void dce_v10_0_hpd_fini(struct amdgpu_device *adev)
 		amdgpu_irq_put(adev, &adev->hpd_irq,
 			       amdgpu_connector->hpd.hpd);
 	}
+	drm_connector_list_iter_end(&iter);
 }
 
 static u32 dce_v10_0_hpd_get_gpio_reg(struct amdgpu_device *adev)
@@ -1215,10 +1221,12 @@ static void dce_v10_0_afmt_audio_select_pin(struct drm_encoder *encoder)
 static void dce_v10_0_audio_write_latency_fields(struct drm_encoder *encoder,
 						struct drm_display_mode *mode)
 {
-	struct amdgpu_device *adev = encoder->dev->dev_private;
+	struct drm_device *dev = encoder->dev;
+	struct amdgpu_device *adev = dev->dev_private;
 	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
 	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
 	struct drm_connector *connector;
+	struct drm_connector_list_iter iter;
 	struct amdgpu_connector *amdgpu_connector = NULL;
 	u32 tmp;
 	int interlace = 0;
@@ -1226,12 +1234,14 @@ static void dce_v10_0_audio_write_latency_fields(struct drm_encoder *encoder,
 	if (!dig || !dig->afmt || !dig->afmt->pin)
 		return;
 
-	list_for_each_entry(connector, &encoder->dev->mode_config.connector_list, head) {
+	drm_connector_list_iter_begin(dev, &iter);
+	drm_for_each_connector_iter(connector, &iter) {
 		if (connector->encoder == encoder) {
 			amdgpu_connector = to_amdgpu_connector(connector);
 			break;
 		}
 	}
+	drm_connector_list_iter_end(&iter);
 
 	if (!amdgpu_connector) {
 		DRM_ERROR("Couldn't find encoder's connector\n");
@@ -1257,10 +1267,12 @@ static void dce_v10_0_audio_write_latency_fields(struct drm_encoder *encoder,
 
 static void dce_v10_0_audio_write_speaker_allocation(struct drm_encoder *encoder)
 {
-	struct amdgpu_device *adev = encoder->dev->dev_private;
+	struct drm_device *dev = encoder->dev;
+	struct amdgpu_device *adev = dev->dev_private;
 	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
 	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
 	struct drm_connector *connector;
+	struct drm_connector_list_iter iter;
 	struct amdgpu_connector *amdgpu_connector = NULL;
 	u32 tmp;
 	u8 *sadb = NULL;
@@ -1269,12 +1281,14 @@ static void dce_v10_0_audio_write_speaker_allocation(struct drm_encoder *encoder
 	if (!dig || !dig->afmt || !dig->afmt->pin)
 		return;
 
-	list_for_each_entry(connector, &encoder->dev->mode_config.connector_list, head) {
+	drm_connector_list_iter_begin(dev, &iter);
+	drm_for_each_connector_iter(connector, &iter) {
 		if (connector->encoder == encoder) {
 			amdgpu_connector = to_amdgpu_connector(connector);
 			break;
 		}
 	}
+	drm_connector_list_iter_end(&iter);
 
 	if (!amdgpu_connector) {
 		DRM_ERROR("Couldn't find encoder's connector\n");
@@ -1309,10 +1323,12 @@ static void dce_v10_0_audio_write_speaker_allocation(struct drm_encoder *encoder
 
 static void dce_v10_0_audio_write_sad_regs(struct drm_encoder *encoder)
 {
-	struct amdgpu_device *adev = encoder->dev->dev_private;
+	struct drm_device *dev = encoder->dev;
+	struct amdgpu_device *adev = dev->dev_private;
 	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
 	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
 	struct drm_connector *connector;
+	struct drm_connector_list_iter iter;
 	struct amdgpu_connector *amdgpu_connector = NULL;
 	struct cea_sad *sads;
 	int i, sad_count;
@@ -1335,12 +1351,14 @@ static void dce_v10_0_audio_write_sad_regs(struct drm_encoder *encoder)
 	if (!dig || !dig->afmt || !dig->afmt->pin)
 		return;
 
-	list_for_each_entry(connector, &encoder->dev->mode_config.connector_list, head) {
+	drm_connector_list_iter_begin(dev, &iter);
+	drm_for_each_connector_iter(connector, &iter) {
 		if (connector->encoder == encoder) {
 			amdgpu_connector = to_amdgpu_connector(connector);
 			break;
 		}
 	}
+	drm_connector_list_iter_end(&iter);
 
 	if (!amdgpu_connector) {
 		DRM_ERROR("Couldn't find encoder's connector\n");
diff --git a/drivers/gpu/drm/amd/amdgpu/dce_v11_0.c b/drivers/gpu/drm/amd/amdgpu/dce_v11_0.c
index 9e0782b54066..fd7b29f03633 100644
--- a/drivers/gpu/drm/amd/amdgpu/dce_v11_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/dce_v11_0.c
@@ -344,9 +344,11 @@ static void dce_v11_0_hpd_init(struct amdgpu_device *adev)
 {
 	struct drm_device *dev = adev->ddev;
 	struct drm_connector *connector;
+	struct drm_connector_list_iter iter;
 	u32 tmp;
 
-	list_for_each_entry(connector, &dev->mode_config.connector_list, head) {
+	drm_connector_list_iter_begin(dev, &iter);
+	drm_for_each_connector_iter(connector, &iter) {
 		struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
 
 		if (amdgpu_connector->hpd.hpd >= adev->mode_info.num_hpd)
@@ -381,6 +383,7 @@ static void dce_v11_0_hpd_init(struct amdgpu_device *adev)
 		dce_v11_0_hpd_set_polarity(adev, amdgpu_connector->hpd.hpd);
 		amdgpu_irq_get(adev, &adev->hpd_irq, amdgpu_connector->hpd.hpd);
 	}
+	drm_connector_list_iter_end(&iter);
 }
 
 /**
@@ -395,9 +398,11 @@ static void dce_v11_0_hpd_fini(struct amdgpu_device *adev)
 {
 	struct drm_device *dev = adev->ddev;
 	struct drm_connector *connector;
+	struct drm_connector_list_iter iter;
 	u32 tmp;
 
-	list_for_each_entry(connector, &dev->mode_config.connector_list, head) {
+	drm_connector_list_iter_begin(dev, &iter);
+	drm_for_each_connector_iter(connector, &iter) {
 		struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
 
 		if (amdgpu_connector->hpd.hpd >= adev->mode_info.num_hpd)
@@ -409,6 +414,7 @@ static void dce_v11_0_hpd_fini(struct amdgpu_device *adev)
 
 		amdgpu_irq_put(adev, &adev->hpd_irq, amdgpu_connector->hpd.hpd);
 	}
+	drm_connector_list_iter_end(&iter);
 }
 
 static u32 dce_v11_0_hpd_get_gpio_reg(struct amdgpu_device *adev)
@@ -1241,10 +1247,12 @@ static void dce_v11_0_afmt_audio_select_pin(struct drm_encoder *encoder)
 static void dce_v11_0_audio_write_latency_fields(struct drm_encoder *encoder,
 						struct drm_display_mode *mode)
 {
-	struct amdgpu_device *adev = encoder->dev->dev_private;
+	struct drm_device *dev = encoder->dev;
+	struct amdgpu_device *adev = dev->dev_private;
 	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
 	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
 	struct drm_connector *connector;
+	struct drm_connector_list_iter iter;
 	struct amdgpu_connector *amdgpu_connector = NULL;
 	u32 tmp;
 	int interlace = 0;
@@ -1252,12 +1260,14 @@ static void dce_v11_0_audio_write_latency_fields(struct drm_encoder *encoder,
 	if (!dig || !dig->afmt || !dig->afmt->pin)
 		return;
 
-	list_for_each_entry(connector, &encoder->dev->mode_config.connector_list, head) {
+	drm_connector_list_iter_begin(dev, &iter);
+	drm_for_each_connector_iter(connector, &iter) {
 		if (connector->encoder == encoder) {
 			amdgpu_connector = to_amdgpu_connector(connector);
 			break;
 		}
 	}
+	drm_connector_list_iter_end(&iter);
 
 	if (!amdgpu_connector) {
 		DRM_ERROR("Couldn't find encoder's connector\n");
@@ -1283,10 +1293,12 @@ static void dce_v11_0_audio_write_latency_fields(struct drm_encoder *encoder,
 
 static void dce_v11_0_audio_write_speaker_allocation(struct drm_encoder *encoder)
 {
-	struct amdgpu_device *adev = encoder->dev->dev_private;
+	struct drm_device *dev = encoder->dev;
+	struct amdgpu_device *adev = dev->dev_private;
 	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
 	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
 	struct drm_connector *connector;
+	struct drm_connector_list_iter iter;
 	struct amdgpu_connector *amdgpu_connector = NULL;
 	u32 tmp;
 	u8 *sadb = NULL;
@@ -1295,12 +1307,14 @@ static void dce_v11_0_audio_write_speaker_allocation(struct drm_encoder *encoder
 	if (!dig || !dig->afmt || !dig->afmt->pin)
 		return;
 
-	list_for_each_entry(connector, &encoder->dev->mode_config.connector_list, head) {
+	drm_connector_list_iter_begin(dev, &iter);
+	drm_for_each_connector_iter(connector, &iter) {
 		if (connector->encoder == encoder) {
 			amdgpu_connector = to_amdgpu_connector(connector);
 			break;
 		}
 	}
+	drm_connector_list_iter_end(&iter);
 
 	if (!amdgpu_connector) {
 		DRM_ERROR("Couldn't find encoder's connector\n");
@@ -1335,10 +1349,12 @@ static void dce_v11_0_audio_write_speaker_allocation(struct drm_encoder *encoder
 
 static void dce_v11_0_audio_write_sad_regs(struct drm_encoder *encoder)
 {
-	struct amdgpu_device *adev = encoder->dev->dev_private;
+	struct drm_device *dev = encoder->dev;
+	struct amdgpu_device *adev = dev->dev_private;
 	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
 	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
 	struct drm_connector *connector;
+	struct drm_connector_list_iter iter;
 	struct amdgpu_connector *amdgpu_connector = NULL;
 	struct cea_sad *sads;
 	int i, sad_count;
@@ -1361,12 +1377,14 @@ static void dce_v11_0_audio_write_sad_regs(struct drm_encoder *encoder)
 	if (!dig || !dig->afmt || !dig->afmt->pin)
 		return;
 
-	list_for_each_entry(connector, &encoder->dev->mode_config.connector_list, head) {
+	drm_connector_list_iter_begin(dev, &iter);
+	drm_for_each_connector_iter(connector, &iter) {
 		if (connector->encoder == encoder) {
 			amdgpu_connector = to_amdgpu_connector(connector);
 			break;
 		}
 	}
+	drm_connector_list_iter_end(&iter);
 
 	if (!amdgpu_connector) {
 		DRM_ERROR("Couldn't find encoder's connector\n");
diff --git a/drivers/gpu/drm/amd/amdgpu/dce_v6_0.c b/drivers/gpu/drm/amd/amdgpu/dce_v6_0.c
index 4bf453e07dca..f137ac1cf54e 100644
--- a/drivers/gpu/drm/amd/amdgpu/dce_v6_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/dce_v6_0.c
@@ -277,9 +277,11 @@ static void dce_v6_0_hpd_init(struct amdgpu_device *adev)
 {
 	struct drm_device *dev = adev->ddev;
 	struct drm_connector *connector;
+	struct drm_connector_list_iter iter;
 	u32 tmp;
 
-	list_for_each_entry(connector, &dev->mode_config.connector_list, head) {
+	drm_connector_list_iter_begin(dev, &iter);
+	drm_for_each_connector_iter(connector, &iter) {
 		struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
 
 		if (amdgpu_connector->hpd.hpd >= adev->mode_info.num_hpd)
@@ -305,7 +307,7 @@ static void dce_v6_0_hpd_init(struct amdgpu_device *adev)
 		dce_v6_0_hpd_set_polarity(adev, amdgpu_connector->hpd.hpd);
 		amdgpu_irq_get(adev, &adev->hpd_irq, amdgpu_connector->hpd.hpd);
 	}
-
+	drm_connector_list_iter_end(&iter);
 }
 
 /**
@@ -320,9 +322,11 @@ static void dce_v6_0_hpd_fini(struct amdgpu_device *adev)
 {
 	struct drm_device *dev = adev->ddev;
 	struct drm_connector *connector;
+	struct drm_connector_list_iter iter;
 	u32 tmp;
 
-	list_for_each_entry(connector, &dev->mode_config.connector_list, head) {
+	drm_connector_list_iter_begin(dev, &iter);
+	drm_for_each_connector_iter(connector, &iter) {
 		struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
 
 		if (amdgpu_connector->hpd.hpd >= adev->mode_info.num_hpd)
@@ -334,6 +338,7 @@ static void dce_v6_0_hpd_fini(struct amdgpu_device *adev)
 
 		amdgpu_irq_put(adev, &adev->hpd_irq, amdgpu_connector->hpd.hpd);
 	}
+	drm_connector_list_iter_end(&iter);
 }
 
 static u32 dce_v6_0_hpd_get_gpio_reg(struct amdgpu_device *adev)
@@ -1120,20 +1125,24 @@ static void dce_v6_0_audio_select_pin(struct drm_encoder *encoder)
 static void dce_v6_0_audio_write_latency_fields(struct drm_encoder *encoder,
 						struct drm_display_mode *mode)
 {
-	struct amdgpu_device *adev = encoder->dev->dev_private;
+	struct drm_device *dev = encoder->dev;
+	struct amdgpu_device *adev = dev->dev_private;
 	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
 	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
 	struct drm_connector *connector;
+	struct drm_connector_list_iter iter;
 	struct amdgpu_connector *amdgpu_connector = NULL;
 	int interlace = 0;
 	u32 tmp;
 
-	list_for_each_entry(connector, &encoder->dev->mode_config.connector_list, head) {
+	drm_connector_list_iter_begin(dev, &iter);
+	drm_for_each_connector_iter(connector, &iter) {
 		if (connector->encoder == encoder) {
 			amdgpu_connector = to_amdgpu_connector(connector);
 			break;
 		}
 	}
+	drm_connector_list_iter_end(&iter);
 
 	if (!amdgpu_connector) {
 		DRM_ERROR("Couldn't find encoder's connector\n");
@@ -1160,21 +1169,25 @@ static void dce_v6_0_audio_write_latency_fields(struct drm_encoder *encoder,
 
 static void dce_v6_0_audio_write_speaker_allocation(struct drm_encoder *encoder)
 {
-	struct amdgpu_device *adev = encoder->dev->dev_private;
+	struct drm_device *dev = encoder->dev;
+	struct amdgpu_device *adev = dev->dev_private;
 	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
 	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
 	struct drm_connector *connector;
+	struct drm_connector_list_iter iter;
 	struct amdgpu_connector *amdgpu_connector = NULL;
 	u8 *sadb = NULL;
 	int sad_count;
 	u32 tmp;
 
-	list_for_each_entry(connector, &encoder->dev->mode_config.connector_list, head) {
+	drm_connector_list_iter_begin(dev, &iter);
+	drm_for_each_connector_iter(connector, &iter) {
 		if (connector->encoder == encoder) {
 			amdgpu_connector = to_amdgpu_connector(connector);
 			break;
 		}
 	}
+	drm_connector_list_iter_end(&iter);
 
 	if (!amdgpu_connector) {
 		DRM_ERROR("Couldn't find encoder's connector\n");
@@ -1217,10 +1230,12 @@ static void dce_v6_0_audio_write_speaker_allocation(struct drm_encoder *encoder)
 
 static void dce_v6_0_audio_write_sad_regs(struct drm_encoder *encoder)
 {
-	struct amdgpu_device *adev = encoder->dev->dev_private;
+	struct drm_device *dev = encoder->dev;
+	struct amdgpu_device *adev = dev->dev_private;
 	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
 	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
 	struct drm_connector *connector;
+	struct drm_connector_list_iter iter;
 	struct amdgpu_connector *amdgpu_connector = NULL;
 	struct cea_sad *sads;
 	int i, sad_count;
@@ -1240,12 +1255,14 @@ static void dce_v6_0_audio_write_sad_regs(struct drm_encoder *encoder)
 		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR13, HDMI_AUDIO_CODING_TYPE_WMA_PRO },
 	};
 
-	list_for_each_entry(connector, &encoder->dev->mode_config.connector_list, head) {
+	drm_connector_list_iter_begin(dev, &iter);
+	drm_for_each_connector_iter(connector, &iter) {
 		if (connector->encoder == encoder) {
 			amdgpu_connector = to_amdgpu_connector(connector);
 			break;
 		}
 	}
+	drm_connector_list_iter_end(&iter);
 
 	if (!amdgpu_connector) {
 		DRM_ERROR("Couldn't find encoder's connector\n");
@@ -1628,6 +1645,7 @@ static void dce_v6_0_afmt_setmode(struct drm_encoder *encoder,
 	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
 	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
 	struct drm_connector *connector;
+	struct drm_connector_list_iter iter;
 	struct amdgpu_connector *amdgpu_connector = NULL;
 	int em = amdgpu_atombios_encoder_get_encoder_mode(encoder);
 	int bpc = 8;
@@ -1635,12 +1653,14 @@ static void dce_v6_0_afmt_setmode(struct drm_encoder *encoder,
 	if (!dig || !dig->afmt)
 		return;
 
-	list_for_each_entry(connector, &encoder->dev->mode_config.connector_list, head) {
+	drm_connector_list_iter_begin(dev, &iter);
+	drm_for_each_connector_iter(connector, &iter) {
 		if (connector->encoder == encoder) {
 			amdgpu_connector = to_amdgpu_connector(connector);
 			break;
 		}
 	}
+	drm_connector_list_iter_end(&iter);
 
 	if (!amdgpu_connector) {
 		DRM_ERROR("Couldn't find encoder's connector\n");
diff --git a/drivers/gpu/drm/amd/amdgpu/dce_v8_0.c b/drivers/gpu/drm/amd/amdgpu/dce_v8_0.c
index b23418ca8f6a..c97558e03393 100644
--- a/drivers/gpu/drm/amd/amdgpu/dce_v8_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/dce_v8_0.c
@@ -271,9 +271,11 @@ static void dce_v8_0_hpd_init(struct amdgpu_device *adev)
 {
 	struct drm_device *dev = adev->ddev;
 	struct drm_connector *connector;
+	struct drm_connector_list_iter iter;
 	u32 tmp;
 
-	list_for_each_entry(connector, &dev->mode_config.connector_list, head) {
+	drm_connector_list_iter_begin(dev, &iter);
+	drm_for_each_connector_iter(connector, &iter) {
 		struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
 
 		if (amdgpu_connector->hpd.hpd >= adev->mode_info.num_hpd)
@@ -299,6 +301,7 @@ static void dce_v8_0_hpd_init(struct amdgpu_device *adev)
 		dce_v8_0_hpd_set_polarity(adev, amdgpu_connector->hpd.hpd);
 		amdgpu_irq_get(adev, &adev->hpd_irq, amdgpu_connector->hpd.hpd);
 	}
+	drm_connector_list_iter_end(&iter);
 }
 
 /**
@@ -313,9 +316,11 @@ static void dce_v8_0_hpd_fini(struct amdgpu_device *adev)
 {
 	struct drm_device *dev = adev->ddev;
 	struct drm_connector *connector;
+	struct drm_connector_list_iter iter;
 	u32 tmp;
 
-	list_for_each_entry(connector, &dev->mode_config.connector_list, head) {
+	drm_connector_list_iter_begin(dev, &iter);
+	drm_for_each_connector_iter(connector, &iter) {
 		struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
 
 		if (amdgpu_connector->hpd.hpd >= adev->mode_info.num_hpd)
@@ -327,6 +332,7 @@ static void dce_v8_0_hpd_fini(struct amdgpu_device *adev)
 
 		amdgpu_irq_put(adev, &adev->hpd_irq, amdgpu_connector->hpd.hpd);
 	}
+	drm_connector_list_iter_end(&iter);
 }
 
 static u32 dce_v8_0_hpd_get_gpio_reg(struct amdgpu_device *adev)
@@ -1153,10 +1159,12 @@ static void dce_v8_0_afmt_audio_select_pin(struct drm_encoder *encoder)
 static void dce_v8_0_audio_write_latency_fields(struct drm_encoder *encoder,
 						struct drm_display_mode *mode)
 {
-	struct amdgpu_device *adev = encoder->dev->dev_private;
+	struct drm_device *dev = encoder->dev;
+	struct amdgpu_device *adev = dev->dev_private;
 	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
 	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
 	struct drm_connector *connector;
+	struct drm_connector_list_iter iter;
 	struct amdgpu_connector *amdgpu_connector = NULL;
 	u32 tmp = 0, offset;
 
@@ -1165,12 +1173,14 @@ static void dce_v8_0_audio_write_latency_fields(struct drm_encoder *encoder,
 
 	offset = dig->afmt->pin->offset;
 
-	list_for_each_entry(connector, &encoder->dev->mode_config.connector_list, head) {
+	drm_connector_list_iter_begin(dev, &iter);
+	drm_for_each_connector_iter(connector, &iter) {
 		if (connector->encoder == encoder) {
 			amdgpu_connector = to_amdgpu_connector(connector);
 			break;
 		}
 	}
+	drm_connector_list_iter_end(&iter);
 
 	if (!amdgpu_connector) {
 		DRM_ERROR("Couldn't find encoder's connector\n");
@@ -1210,10 +1220,12 @@ static void dce_v8_0_audio_write_latency_fields(struct drm_encoder *encoder,
 
 static void dce_v8_0_audio_write_speaker_allocation(struct drm_encoder *encoder)
 {
-	struct amdgpu_device *adev = encoder->dev->dev_private;
+	struct drm_device *dev = encoder->dev;
+	struct amdgpu_device *adev = dev->dev_private;
 	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
 	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
 	struct drm_connector *connector;
+	struct drm_connector_list_iter iter;
 	struct amdgpu_connector *amdgpu_connector = NULL;
 	u32 offset, tmp;
 	u8 *sadb = NULL;
@@ -1224,12 +1236,14 @@ static void dce_v8_0_audio_write_speaker_allocation(struct drm_encoder *encoder)
 
 	offset = dig->afmt->pin->offset;
 
-	list_for_each_entry(connector, &encoder->dev->mode_config.connector_list, head) {
+	drm_connector_list_iter_begin(dev, &iter);
+	drm_for_each_connector_iter(connector, &iter) {
 		if (connector->encoder == encoder) {
 			amdgpu_connector = to_amdgpu_connector(connector);
 			break;
 		}
 	}
+	drm_connector_list_iter_end(&iter);
 
 	if (!amdgpu_connector) {
 		DRM_ERROR("Couldn't find encoder's connector\n");
@@ -1259,11 +1273,13 @@ static void dce_v8_0_audio_write_speaker_allocation(struct drm_encoder *encoder)
 
 static void dce_v8_0_audio_write_sad_regs(struct drm_encoder *encoder)
 {
-	struct amdgpu_device *adev = encoder->dev->dev_private;
+	struct drm_device *dev = encoder->dev;
+	struct amdgpu_device *adev = dev->dev_private;
 	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
 	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
 	u32 offset;
 	struct drm_connector *connector;
+	struct drm_connector_list_iter iter;
 	struct amdgpu_connector *amdgpu_connector = NULL;
 	struct cea_sad *sads;
 	int i, sad_count;
@@ -1288,12 +1304,14 @@ static void dce_v8_0_audio_write_sad_regs(struct drm_encoder *encoder)
 
 	offset = dig->afmt->pin->offset;
 
-	list_for_each_entry(connector, &encoder->dev->mode_config.connector_list, head) {
+	drm_connector_list_iter_begin(dev, &iter);
+	drm_for_each_connector_iter(connector, &iter) {
 		if (connector->encoder == encoder) {
 			amdgpu_connector = to_amdgpu_connector(connector);
 			break;
 		}
 	}
+	drm_connector_list_iter_end(&iter);
 
 	if (!amdgpu_connector) {
 		DRM_ERROR("Couldn't find encoder's connector\n");
diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c
index 0242d693f4f6..0ef49db0f08d 100644
--- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c
+++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c
@@ -752,27 +752,29 @@ static int detect_mst_link_for_all_connectors(struct drm_device *dev)
 {
 	struct amdgpu_dm_connector *aconnector;
 	struct drm_connector *connector;
+	struct drm_connector_list_iter iter;
 	int ret = 0;
 
-	drm_modeset_lock(&dev->mode_config.connection_mutex, NULL);
-
-	list_for_each_entry(connector, &dev->mode_config.connector_list, head) {
+	drm_connector_list_iter_begin(dev, &iter);
+	drm_for_each_connector_iter(connector, &iter) {
 		aconnector = to_amdgpu_dm_connector(connector);
 		if (aconnector->dc_link->type == dc_connection_mst_branch &&
 		    aconnector->mst_mgr.aux) {
 			DRM_DEBUG_DRIVER("DM_MST: starting TM on aconnector: %p [id: %d]\n",
-					aconnector, aconnector->base.base.id);
+					 aconnector,
+					 aconnector->base.base.id);
 
 			ret = drm_dp_mst_topology_mgr_set_mst(&aconnector->mst_mgr, true);
 			if (ret < 0) {
 				DRM_ERROR("DM_MST: Failed to start MST\n");
-				((struct dc_link *)aconnector->dc_link)->type = dc_connection_single;
-				return ret;
-				}
+				aconnector->dc_link->type =
+					dc_connection_single;
+				break;
 			}
+		}
 	}
+	drm_connector_list_iter_end(&iter);
 
-	drm_modeset_unlock(&dev->mode_config.connection_mutex);
 	return ret;
 }
 
@@ -810,14 +812,13 @@ static void s3_handle_mst(struct drm_device *dev, bool suspend)
 {
 	struct amdgpu_dm_connector *aconnector;
 	struct drm_connector *connector;
+	struct drm_connector_list_iter iter;
 	struct drm_dp_mst_topology_mgr *mgr;
 	int ret;
 	bool need_hotplug = false;
 
-	drm_modeset_lock(&dev->mode_config.connection_mutex, NULL);
-
-	list_for_each_entry(connector, &dev->mode_config.connector_list,
-			    head) {
+	drm_connector_list_iter_begin(dev, &iter);
+	drm_for_each_connector_iter(connector, &iter) {
 		aconnector = to_amdgpu_dm_connector(connector);
 		if (aconnector->dc_link->type != dc_connection_mst_branch ||
 		    aconnector->mst_port)
@@ -835,8 +836,7 @@ static void s3_handle_mst(struct drm_device *dev, bool suspend)
 			}
 		}
 	}
-
-	drm_modeset_unlock(&dev->mode_config.connection_mutex);
+	drm_connector_list_iter_end(&iter);
 
 	if (need_hotplug)
 		drm_kms_helper_hotplug_event(dev);
@@ -1018,6 +1018,7 @@ static int dm_resume(void *handle)
 	struct amdgpu_display_manager *dm = &adev->dm;
 	struct amdgpu_dm_connector *aconnector;
 	struct drm_connector *connector;
+	struct drm_connector_list_iter iter;
 	struct drm_crtc *crtc;
 	struct drm_crtc_state *new_crtc_state;
 	struct dm_crtc_state *dm_new_crtc_state;
@@ -1050,7 +1051,8 @@ static int dm_resume(void *handle)
 	amdgpu_dm_irq_resume_early(adev);
 
 	/* Do detection*/
-	list_for_each_entry(connector, &ddev->mode_config.connector_list, head) {
+	drm_connector_list_iter_begin(ddev, &iter);
+	drm_for_each_connector_iter(connector, &iter) {
 		aconnector = to_amdgpu_dm_connector(connector);
 
 		/*
@@ -1078,6 +1080,7 @@ static int dm_resume(void *handle)
 		amdgpu_dm_update_connector_after_detect(aconnector);
 		mutex_unlock(&aconnector->hpd_lock);
 	}
+	drm_connector_list_iter_end(&iter);
 
 	/* Force mode set in atomic commit */
 	for_each_new_crtc_in_state(dm->cached_state, crtc, new_crtc_state, i)
diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_irq.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_irq.c
index fa5d503d379c..64445c4cc4c2 100644
--- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_irq.c
+++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_irq.c
@@ -732,8 +732,10 @@ void amdgpu_dm_hpd_init(struct amdgpu_device *adev)
 {
 	struct drm_device *dev = adev->ddev;
 	struct drm_connector *connector;
+	struct drm_connector_list_iter iter;
 
-	list_for_each_entry(connector, &dev->mode_config.connector_list, head) {
+	drm_connector_list_iter_begin(dev, &iter);
+	drm_for_each_connector_iter(connector, &iter) {
 		struct amdgpu_dm_connector *amdgpu_dm_connector =
 				to_amdgpu_dm_connector(connector);
 
@@ -751,6 +753,7 @@ void amdgpu_dm_hpd_init(struct amdgpu_device *adev)
 					true);
 		}
 	}
+	drm_connector_list_iter_end(&iter);
 }
 
 /**
@@ -765,8 +768,10 @@ void amdgpu_dm_hpd_fini(struct amdgpu_device *adev)
 {
 	struct drm_device *dev = adev->ddev;
 	struct drm_connector *connector;
+	struct drm_connector_list_iter iter;
 
-	list_for_each_entry(connector, &dev->mode_config.connector_list, head) {
+	drm_connector_list_iter_begin(dev, &iter);
+	drm_for_each_connector_iter(connector, &iter) {
 		struct amdgpu_dm_connector *amdgpu_dm_connector =
 				to_amdgpu_dm_connector(connector);
 		const struct dc_link *dc_link = amdgpu_dm_connector->dc_link;
@@ -779,4 +784,5 @@ void amdgpu_dm_hpd_fini(struct amdgpu_device *adev)
 					false);
 		}
 	}
+	drm_connector_list_iter_end(&iter);
 }
-- 
2.21.0

[PATCH 23/26] drm/amdgpu/dm: Resume short HPD IRQs before resuming MST topology

[Lyude Paul]

Since we're going to be reprobing the entire topology state on resume
now using sideband transactions, we need to ensure that we actually have
short HPD irqs enabled before calling drm_dp_mst_topology_mgr_resume().
So, do that.

Cc: Juston Li <juston.li@intel.com>
Cc: Imre Deak <imre.deak@intel.com>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Cc: Harry Wentland <hwentlan@amd.com>
Signed-off-by: Lyude Paul <lyude@redhat.com>
---
 drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c
index 0ef49db0f08d..e33e080cf16d 100644
--- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c
+++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c
@@ -1041,15 +1041,15 @@ static int dm_resume(void *handle)
 	/* program HPD filter */
 	dc_resume(dm->dc);
 
-	/* On resume we need to  rewrite the MSTM control bits to enamble MST*/
-	s3_handle_mst(ddev, false);
-
 	/*
 	 * early enable HPD Rx IRQ, should be done before set mode as short
 	 * pulse interrupts are used for MST
 	 */
 	amdgpu_dm_irq_resume_early(adev);
 
+	/* On resume we need to  rewrite the MSTM control bits to enamble MST*/
+	s3_handle_mst(ddev, false);
+
 	/* Do detection*/
 	drm_connector_list_iter_begin(ddev, &iter);
 	drm_for_each_connector_iter(connector, &iter) {
-- 
2.21.0

[PATCH 24/26] drm/dp_mst: Add basic topology reprobing when resuming

[Lyude Paul]

Finally! For a very long time, our MST helpers have had one very
annoying issue: They don't know how to reprobe the topology state when
coming out of suspend. This means that if a user has a machine connected
to an MST topology and decides to suspend their machine, we lose all
topology changes that happened during that period. That can be a big
problem if the machine was connected to a different topology on the same
port before resuming, as we won't bother reprobing any of the ports and
likely cause the user's monitors not to come back up as expected.

So, we start fixing this by teaching our MST helpers how to reprobe the
link addresses of each connected topology when resuming. As it turns
out, the behavior that we want here is identical to the behavior we want
when initially probing a newly connected MST topology, with a couple of
important differences:

- We need to be more careful about handling the potential races between
  events from the MST hub that could change the topology state as we're
  performing the link address reprobe
- We need to be more careful about handling unlikely state changes on
  ports - such as an input port turning into an output port, something
  that would be far more likely to happen in situations like the MST hub
  we're connected to being changed while we're suspend

Both of which have been solved by previous commits. That leaves one
requirement:

- We need to prune any MST ports in our in-memory topology state that
  were present when suspending, but have not appeared in the post-resume
  link address response from their parent branch device

Which we can now handle in this commit by modifying
drm_dp_send_link_address(). We then introduce suspend/resume reprobing
by introducing drm_dp_mst_topology_mgr_invalidate_mstb(), which we call
in drm_dp_mst_topology_mgr_suspend() to traverse the in-memory topology
state to indicate that each mstb needs it's link address resent and PBN
resources reprobed.

On resume, we start back up &mgr->work and have it reprobe the topology
in the same way we would on a hotplug, removing any leftover ports that
no longer appear in the topology state.

Cc: Juston Li <juston.li@intel.com>
Cc: Imre Deak <imre.deak@intel.com>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Cc: Harry Wentland <hwentlan@amd.com>
Signed-off-by: Lyude Paul <lyude@redhat.com>
---
 .../gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c |   2 +-
 drivers/gpu/drm/drm_dp_mst_topology.c         | 138 +++++++++++++-----
 drivers/gpu/drm/i915/display/intel_dp.c       |   3 +-
 drivers/gpu/drm/nouveau/dispnv50/disp.c       |   6 +-
 include/drm/drm_dp_mst_helper.h               |   3 +-
 5 files changed, 112 insertions(+), 40 deletions(-)

diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c
index e33e080cf16d..2ebd4811ff40 100644
--- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c
+++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c
@@ -829,7 +829,7 @@ static void s3_handle_mst(struct drm_device *dev, bool suspend)
 		if (suspend) {
 			drm_dp_mst_topology_mgr_suspend(mgr);
 		} else {
-			ret = drm_dp_mst_topology_mgr_resume(mgr);
+			ret = drm_dp_mst_topology_mgr_resume(mgr, true);
 			if (ret < 0) {
 				drm_dp_mst_topology_mgr_set_mst(mgr, false);
 				need_hotplug = true;
diff --git a/drivers/gpu/drm/drm_dp_mst_topology.c b/drivers/gpu/drm/drm_dp_mst_topology.c
index 126db36c9337..320158970e25 100644
--- a/drivers/gpu/drm/drm_dp_mst_topology.c
+++ b/drivers/gpu/drm/drm_dp_mst_topology.c
@@ -1931,6 +1931,14 @@ drm_dp_mst_handle_link_address_port(struct drm_dp_mst_branch *mstb,
 		goto fail_unlock;
 	}
 
+	/*
+	 * If this port wasn't just created, then we're reprobing because
+	 * we're coming out of suspend. In this case, always resend the link
+	 * address if there's an MSTB on this port
+	 */
+	if (!created && port->pdt == DP_PEER_DEVICE_MST_BRANCHING)
+		send_link_addr = true;
+
 	if (send_link_addr) {
 		mutex_lock(&mgr->lock);
 		if (port->mstb) {
@@ -2443,7 +2451,8 @@ static void drm_dp_send_link_address(struct drm_dp_mst_topology_mgr *mgr,
 {
 	struct drm_dp_sideband_msg_tx *txmsg;
 	struct drm_dp_link_address_ack_reply *reply;
-	int i, len, ret;
+	struct drm_dp_mst_port *port, *tmp;
+	int i, len, ret, port_mask = 0;
 
 	txmsg = kzalloc(sizeof(*txmsg), GFP_KERNEL);
 	if (!txmsg)
@@ -2473,9 +2482,28 @@ static void drm_dp_send_link_address(struct drm_dp_mst_topology_mgr *mgr,
 
 	drm_dp_check_mstb_guid(mstb, reply->guid);
 
-	for (i = 0; i < reply->nports; i++)
+	for (i = 0; i < reply->nports; i++) {
+		port_mask |= BIT(reply->ports[i].port_number);
 		drm_dp_mst_handle_link_address_port(mstb, mgr->dev,
 						    &reply->ports[i]);
+	}
+
+	/* Prune any ports that are currently a part of mstb in our in-memory
+	 * topology, but were not seen in this link address. Usually this
+	 * means that they were removed while the topology was out of sync,
+	 * e.g. during suspend/resume
+	 */
+	mutex_lock(&mgr->lock);
+	list_for_each_entry_safe(port, tmp, &mstb->ports, next) {
+		if (port_mask & BIT(port->port_num))
+			continue;
+
+		DRM_DEBUG_KMS("port %d was not in link address, removing\n",
+			      port->port_num);
+		list_del(&port->next);
+		drm_dp_mst_topology_put_port(port);
+	}
+	mutex_unlock(&mgr->lock);
 
 	drm_kms_helper_hotplug_event(mgr->dev);
 
@@ -3072,6 +3100,23 @@ int drm_dp_mst_topology_mgr_set_mst(struct drm_dp_mst_topology_mgr *mgr, bool ms
 }
 EXPORT_SYMBOL(drm_dp_mst_topology_mgr_set_mst);
 
+static void
+drm_dp_mst_topology_mgr_invalidate_mstb(struct drm_dp_mst_branch *mstb)
+{
+	struct drm_dp_mst_port *port;
+
+	/* The link address will need to be re-sent on resume */
+	mstb->link_address_sent = false;
+
+	list_for_each_entry(port, &mstb->ports, next) {
+		/* The PBN for each port will also need to be re-probed */
+		port->available_pbn = 0;
+
+		if (port->mstb)
+			drm_dp_mst_topology_mgr_invalidate_mstb(port->mstb);
+	}
+}
+
 /**
  * drm_dp_mst_topology_mgr_suspend() - suspend the MST manager
  * @mgr: manager to suspend
@@ -3088,60 +3133,85 @@ void drm_dp_mst_topology_mgr_suspend(struct drm_dp_mst_topology_mgr *mgr)
 	flush_work(&mgr->up_req_work);
 	flush_work(&mgr->work);
 	flush_work(&mgr->destroy_connector_work);
+
+	mutex_lock(&mgr->lock);
+	if (mgr->mst_state && mgr->mst_primary)
+		drm_dp_mst_topology_mgr_invalidate_mstb(mgr->mst_primary);
+	mutex_unlock(&mgr->lock);
 }
 EXPORT_SYMBOL(drm_dp_mst_topology_mgr_suspend);
 
 /**
  * drm_dp_mst_topology_mgr_resume() - resume the MST manager
  * @mgr: manager to resume
+ * @sync: whether or not to perform topology reprobing synchronously
  *
  * This will fetch DPCD and see if the device is still there,
  * if it is, it will rewrite the MSTM control bits, and return.
  *
- * if the device fails this returns -1, and the driver should do
+ * If the device fails this returns -1, and the driver should do
  * a full MST reprobe, in case we were undocked.
+ *
+ * During system resume (where it is assumed that the driver will be calling
+ * drm_atomic_helper_resume()) this function should be called beforehand with
+ * @sync set to true. In contexts like runtime resume where the driver is not
+ * expected to be calling drm_atomic_helper_resume(), this function should be
+ * called with @sync set to false in order to avoid deadlocking.
+ *
+ * Returns: -1 if the MST topology was removed while we were suspended, 0
+ * otherwise.
  */
-int drm_dp_mst_topology_mgr_resume(struct drm_dp_mst_topology_mgr *mgr)
+int drm_dp_mst_topology_mgr_resume(struct drm_dp_mst_topology_mgr *mgr,
+				   bool sync)
 {
-	int ret = 0;
+	int ret;
+	u8 guid[16];
 
 	mutex_lock(&mgr->lock);
+	if (!mgr->mst_primary)
+		goto out_fail;
 
-	if (mgr->mst_primary) {
-		int sret;
-		u8 guid[16];
+	ret = drm_dp_dpcd_read(mgr->aux, DP_DPCD_REV, mgr->dpcd,
+			       DP_RECEIVER_CAP_SIZE);
+	if (ret != DP_RECEIVER_CAP_SIZE) {
+		DRM_DEBUG_KMS("dpcd read failed - undocked during suspend?\n");
+		goto out_fail;
+	}
 
-		sret = drm_dp_dpcd_read(mgr->aux, DP_DPCD_REV, mgr->dpcd, DP_RECEIVER_CAP_SIZE);
-		if (sret != DP_RECEIVER_CAP_SIZE) {
-			DRM_DEBUG_KMS("dpcd read failed - undocked during suspend?\n");
-			ret = -1;
-			goto out_unlock;
-		}
+	ret = drm_dp_dpcd_writeb(mgr->aux, DP_MSTM_CTRL,
+				 DP_MST_EN |
+				 DP_UP_REQ_EN |
+				 DP_UPSTREAM_IS_SRC);
+	if (ret < 0) {
+		DRM_DEBUG_KMS("mst write failed - undocked during suspend?\n");
+		goto out_fail;
+	}
 
-		ret = drm_dp_dpcd_writeb(mgr->aux, DP_MSTM_CTRL,
-					 DP_MST_EN | DP_UP_REQ_EN | DP_UPSTREAM_IS_SRC);
-		if (ret < 0) {
-			DRM_DEBUG_KMS("mst write failed - undocked during suspend?\n");
-			ret = -1;
-			goto out_unlock;
-		}
+	/* Some hubs forget their guids after they resume */
+	ret = drm_dp_dpcd_read(mgr->aux, DP_GUID, guid, 16);
+	if (ret != 16) {
+		DRM_DEBUG_KMS("dpcd read failed - undocked during suspend?\n");
+		goto out_fail;
+	}
+	drm_dp_check_mstb_guid(mgr->mst_primary, guid);
 
-		/* Some hubs forget their guids after they resume */
-		sret = drm_dp_dpcd_read(mgr->aux, DP_GUID, guid, 16);
-		if (sret != 16) {
-			DRM_DEBUG_KMS("dpcd read failed - undocked during suspend?\n");
-			ret = -1;
-			goto out_unlock;
-		}
-		drm_dp_check_mstb_guid(mgr->mst_primary, guid);
+	/* For the final step of resuming the topology, we need to bring the
+	 * state of our in-memory topology back into sync with reality. So,
+	 * restart the probing process as if we're probing a new hub
+	 */
+	queue_work(system_long_wq, &mgr->work);
+	mutex_unlock(&mgr->lock);
 
-		ret = 0;
-	} else
-		ret = -1;
+	if (sync) {
+		DRM_DEBUG_KMS("Waiting for link probe work to finish re-syncing topology...\n");
+		flush_work(&mgr->work);
+	}
 
-out_unlock:
+	return 0;
+
+out_fail:
 	mutex_unlock(&mgr->lock);
-	return ret;
+	return -1;
 }
 EXPORT_SYMBOL(drm_dp_mst_topology_mgr_resume);
 
diff --git a/drivers/gpu/drm/i915/display/intel_dp.c b/drivers/gpu/drm/i915/display/intel_dp.c
index 0eb5d66f87a7..7190ff5c3649 100644
--- a/drivers/gpu/drm/i915/display/intel_dp.c
+++ b/drivers/gpu/drm/i915/display/intel_dp.c
@@ -7366,7 +7366,8 @@ void intel_dp_mst_resume(struct drm_i915_private *dev_priv)
 		if (!intel_dp->can_mst)
 			continue;
 
-		ret = drm_dp_mst_topology_mgr_resume(&intel_dp->mst_mgr);
+		ret = drm_dp_mst_topology_mgr_resume(&intel_dp->mst_mgr,
+						     true);
 		if (ret) {
 			intel_dp->is_mst = false;
 			drm_dp_mst_topology_mgr_set_mst(&intel_dp->mst_mgr,
diff --git a/drivers/gpu/drm/nouveau/dispnv50/disp.c b/drivers/gpu/drm/nouveau/dispnv50/disp.c
index 7ba373f493b2..c44b54eeddce 100644
--- a/drivers/gpu/drm/nouveau/dispnv50/disp.c
+++ b/drivers/gpu/drm/nouveau/dispnv50/disp.c
@@ -1300,14 +1300,14 @@ nv50_mstm_fini(struct nv50_mstm *mstm)
 }
 
 static void
-nv50_mstm_init(struct nv50_mstm *mstm)
+nv50_mstm_init(struct nv50_mstm *mstm, bool runtime)
 {
 	int ret;
 
 	if (!mstm || !mstm->mgr.mst_state)
 		return;
 
-	ret = drm_dp_mst_topology_mgr_resume(&mstm->mgr);
+	ret = drm_dp_mst_topology_mgr_resume(&mstm->mgr, !runtime);
 	if (ret == -1) {
 		drm_dp_mst_topology_mgr_set_mst(&mstm->mgr, false);
 		drm_kms_helper_hotplug_event(mstm->mgr.dev);
@@ -2257,7 +2257,7 @@ nv50_display_init(struct drm_device *dev, bool resume, bool runtime)
 		if (encoder->encoder_type != DRM_MODE_ENCODER_DPMST) {
 			struct nouveau_encoder *nv_encoder =
 				nouveau_encoder(encoder);
-			nv50_mstm_init(nv_encoder->dp.mstm);
+			nv50_mstm_init(nv_encoder->dp.mstm, runtime);
 		}
 	}
 
diff --git a/include/drm/drm_dp_mst_helper.h b/include/drm/drm_dp_mst_helper.h
index aed68d7e6492..eece28525d52 100644
--- a/include/drm/drm_dp_mst_helper.h
+++ b/include/drm/drm_dp_mst_helper.h
@@ -683,7 +683,8 @@ void drm_dp_mst_dump_topology(struct seq_file *m,
 
 void drm_dp_mst_topology_mgr_suspend(struct drm_dp_mst_topology_mgr *mgr);
 int __must_check
-drm_dp_mst_topology_mgr_resume(struct drm_dp_mst_topology_mgr *mgr);
+drm_dp_mst_topology_mgr_resume(struct drm_dp_mst_topology_mgr *mgr,
+			       bool sync);
 struct drm_dp_mst_topology_state *drm_atomic_get_mst_topology_state(struct drm_atomic_state *state,
 								    struct drm_dp_mst_topology_mgr *mgr);
 int __must_check
-- 
2.21.0

[PATCH 25/26] drm/dp_mst: Also print unhashed pointers for malloc/topology references

[Lyude Paul]

Currently we only print mstb/port pointer addresses in our malloc and
topology refcount functions using the hashed-by-default %p, but
unfortunately if you're trying to debug a use-after-free error caused by
a refcounting error then this really isn't terribly useful. On the other
hand though, everything in the rest of the DP MST helpers uses hashed
pointer values as well and probably isn't useful to convert to unhashed.
So, let's just get the best of both worlds and print both the hashed and
unhashed pointer in our malloc/topology refcount debugging output. This
will hopefully make it a lot easier to figure out which port/mstb is
causing KASAN to get upset.

Cc: Juston Li <juston.li@intel.com>
Cc: Imre Deak <imre.deak@intel.com>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Cc: Harry Wentland <hwentlan@amd.com>
Signed-off-by: Lyude Paul <lyude@redhat.com>
---
 drivers/gpu/drm/drm_dp_mst_topology.c | 34 ++++++++++++++++-----------
 1 file changed, 20 insertions(+), 14 deletions(-)

diff --git a/drivers/gpu/drm/drm_dp_mst_topology.c b/drivers/gpu/drm/drm_dp_mst_topology.c
index 320158970e25..fcd3f98cbc34 100644
--- a/drivers/gpu/drm/drm_dp_mst_topology.c
+++ b/drivers/gpu/drm/drm_dp_mst_topology.c
@@ -1325,7 +1325,8 @@ static void
 drm_dp_mst_get_mstb_malloc(struct drm_dp_mst_branch *mstb)
 {
 	kref_get(&mstb->malloc_kref);
-	DRM_DEBUG("mstb %p (%d)\n", mstb, kref_read(&mstb->malloc_kref));
+	DRM_DEBUG("mstb %p/%px (%d)\n",
+		  mstb, mstb, kref_read(&mstb->malloc_kref));
 }
 
 /**
@@ -1342,7 +1343,8 @@ drm_dp_mst_get_mstb_malloc(struct drm_dp_mst_branch *mstb)
 static void
 drm_dp_mst_put_mstb_malloc(struct drm_dp_mst_branch *mstb)
 {
-	DRM_DEBUG("mstb %p (%d)\n", mstb, kref_read(&mstb->malloc_kref) - 1);
+	DRM_DEBUG("mstb %p/%px (%d)\n",
+		  mstb, mstb, kref_read(&mstb->malloc_kref) - 1);
 	kref_put(&mstb->malloc_kref, drm_dp_free_mst_branch_device);
 }
 
@@ -1377,7 +1379,8 @@ void
 drm_dp_mst_get_port_malloc(struct drm_dp_mst_port *port)
 {
 	kref_get(&port->malloc_kref);
-	DRM_DEBUG("port %p (%d)\n", port, kref_read(&port->malloc_kref));
+	DRM_DEBUG("port %p/%px (%d)\n",
+		  port, port, kref_read(&port->malloc_kref));
 }
 EXPORT_SYMBOL(drm_dp_mst_get_port_malloc);
 
@@ -1394,7 +1397,8 @@ EXPORT_SYMBOL(drm_dp_mst_get_port_malloc);
 void
 drm_dp_mst_put_port_malloc(struct drm_dp_mst_port *port)
 {
-	DRM_DEBUG("port %p (%d)\n", port, kref_read(&port->malloc_kref) - 1);
+	DRM_DEBUG("port %p/%px (%d)\n",
+		  port, port, kref_read(&port->malloc_kref) - 1);
 	kref_put(&port->malloc_kref, drm_dp_free_mst_port);
 }
 EXPORT_SYMBOL(drm_dp_mst_put_port_malloc);
@@ -1445,8 +1449,8 @@ drm_dp_mst_topology_try_get_mstb(struct drm_dp_mst_branch *mstb)
 	int ret = kref_get_unless_zero(&mstb->topology_kref);
 
 	if (ret)
-		DRM_DEBUG("mstb %p (%d)\n", mstb,
-			  kref_read(&mstb->topology_kref));
+		DRM_DEBUG("mstb %p/%px (%d)\n",
+			  mstb, mstb, kref_read(&mstb->topology_kref));
 
 	return ret;
 }
@@ -1469,7 +1473,8 @@ static void drm_dp_mst_topology_get_mstb(struct drm_dp_mst_branch *mstb)
 {
 	WARN_ON(kref_read(&mstb->topology_kref) == 0);
 	kref_get(&mstb->topology_kref);
-	DRM_DEBUG("mstb %p (%d)\n", mstb, kref_read(&mstb->topology_kref));
+	DRM_DEBUG("mstb %p/%px (%d)\n",
+		  mstb, mstb, kref_read(&mstb->topology_kref));
 }
 
 /**
@@ -1487,8 +1492,8 @@ static void drm_dp_mst_topology_get_mstb(struct drm_dp_mst_branch *mstb)
 static void
 drm_dp_mst_topology_put_mstb(struct drm_dp_mst_branch *mstb)
 {
-	DRM_DEBUG("mstb %p (%d)\n",
-		  mstb, kref_read(&mstb->topology_kref) - 1);
+	DRM_DEBUG("mstb %p/%px (%d)\n",
+		  mstb, mstb, kref_read(&mstb->topology_kref) - 1);
 	kref_put(&mstb->topology_kref, drm_dp_destroy_mst_branch_device);
 }
 
@@ -1544,8 +1549,8 @@ drm_dp_mst_topology_try_get_port(struct drm_dp_mst_port *port)
 	int ret = kref_get_unless_zero(&port->topology_kref);
 
 	if (ret)
-		DRM_DEBUG("port %p (%d)\n", port,
-			  kref_read(&port->topology_kref));
+		DRM_DEBUG("port %p/%px (%d)\n",
+			  port, port, kref_read(&port->topology_kref));
 
 	return ret;
 }
@@ -1567,7 +1572,8 @@ static void drm_dp_mst_topology_get_port(struct drm_dp_mst_port *port)
 {
 	WARN_ON(kref_read(&port->topology_kref) == 0);
 	kref_get(&port->topology_kref);
-	DRM_DEBUG("port %p (%d)\n", port, kref_read(&port->topology_kref));
+	DRM_DEBUG("port %p/%px (%d)\n",
+		  port, port, kref_read(&port->topology_kref));
 }
 
 /**
@@ -1583,8 +1589,8 @@ static void drm_dp_mst_topology_get_port(struct drm_dp_mst_port *port)
  */
 static void drm_dp_mst_topology_put_port(struct drm_dp_mst_port *port)
 {
-	DRM_DEBUG("port %p (%d)\n",
-		  port, kref_read(&port->topology_kref) - 1);
+	DRM_DEBUG("port %p/%px (%d)\n",
+		  port, port, kref_read(&port->topology_kref) - 1);
 	kref_put(&port->topology_kref, drm_dp_destroy_port);
 }
 
-- 
2.21.0

[PATCH 26/26] drm/dp_mst: Add topology ref history tracking for debugging

[Lyude Paul]

For very subtle mistakes with topology refs, it can be rather difficult
to trace them down with the debugging info that we already have. I had
one such issue recently while trying to implement suspend/resume
reprobing for MST, and ended up coming up with this.

Inspired by Chris Wilson's wakeref tracking for i915, this adds a very
similar feature to the DP MST helpers, which allows for partial tracking
of topology refs for both ports and branch devices. This is a lot less
advanced then wakeref tracking: we merely keep a count of all of the
spots where a topology ref has been grabbed or dropped, then dump out
that history in chronological order when a port or branch device's
topology refcount reaches 0. So far, I've found this incredibly useful
for debugging topology refcount errors.

Since this has the potential to be somewhat slow and loud, we add an
expert kernel config option to enable or disable this feature,
CONFIG_DRM_DEBUG_DP_MST_TOPOLOGY_REFS.

Cc: Juston Li <juston.li@intel.com>
Cc: Imre Deak <imre.deak@intel.com>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Cc: Harry Wentland <hwentlan@amd.com>
Signed-off-by: Lyude Paul <lyude@redhat.com>
---
 drivers/gpu/drm/Kconfig               |  14 ++
 drivers/gpu/drm/drm_dp_mst_topology.c | 230 +++++++++++++++++++++++++-
 include/drm/drm_dp_mst_helper.h       |  45 +++++
 3 files changed, 285 insertions(+), 4 deletions(-)

diff --git a/drivers/gpu/drm/Kconfig b/drivers/gpu/drm/Kconfig
index b9362b4f6353..e1fd27f1fa34 100644
--- a/drivers/gpu/drm/Kconfig
+++ b/drivers/gpu/drm/Kconfig
@@ -89,6 +89,20 @@ config DRM_KMS_FB_HELPER
 	help
 	  FBDEV helpers for KMS drivers.
 
+config DRM_DEBUG_DP_MST_TOPOLOGY_REFS
+        bool "Enable refcount backtrace history in the DP MST helpers"
+        select STACKDEPOT
+        depends on DRM_KMS_HELPER
+        depends on DEBUG_KERNEL
+        depends on EXPERT
+        help
+          Enables debug tracing for topology refs in DRM's DP MST helpers. A
+          history of each topology reference/dereference will be printed to the
+          kernel log once a port or branch device's topology refcount reaches 0.
+
+          This has the potential to use a lot of memory and print some very
+          large kernel messages. If in doubt, say "N".
+
 config DRM_FBDEV_EMULATION
 	bool "Enable legacy fbdev support for your modesetting driver"
 	depends on DRM
diff --git a/drivers/gpu/drm/drm_dp_mst_topology.c b/drivers/gpu/drm/drm_dp_mst_topology.c
index fcd3f98cbc34..443d05d563aa 100644
--- a/drivers/gpu/drm/drm_dp_mst_topology.c
+++ b/drivers/gpu/drm/drm_dp_mst_topology.c
@@ -28,6 +28,13 @@
 #include <linux/sched.h>
 #include <linux/seq_file.h>
 
+#if IS_ENABLED(CONFIG_DRM_DEBUG_DP_MST_TOPOLOGY_REFS)
+#include <linux/stackdepot.h>
+#include <linux/sort.h>
+#include <linux/timekeeping.h>
+#include <linux/math64.h>
+#endif
+
 #include <drm/drm_atomic.h>
 #include <drm/drm_atomic_helper.h>
 #include <drm/drm_dp_mst_helper.h>
@@ -1403,12 +1410,189 @@ drm_dp_mst_put_port_malloc(struct drm_dp_mst_port *port)
 }
 EXPORT_SYMBOL(drm_dp_mst_put_port_malloc);
 
+#if IS_ENABLED(CONFIG_DRM_DEBUG_DP_MST_TOPOLOGY_REFS)
+
+#define STACK_DEPTH 8
+
+static noinline void
+__topology_ref_save(struct drm_dp_mst_topology_mgr *mgr,
+		    struct drm_dp_mst_topology_ref_history *history,
+		    enum drm_dp_mst_topology_ref_type type)
+{
+	struct drm_dp_mst_topology_ref_entry *entry = NULL;
+	depot_stack_handle_t backtrace;
+	ulong stack_entries[STACK_DEPTH];
+	uint n;
+	int i;
+
+	n = stack_trace_save(stack_entries, ARRAY_SIZE(stack_entries), 1);
+	backtrace = stack_depot_save(stack_entries, n, GFP_KERNEL);
+	if (!backtrace)
+		goto fail_alloc;
+
+	/* Try to find an existing entry for this backtrace */
+	for (i = 0; i < history->len; i++) {
+		if (history->entries[i].backtrace == backtrace) {
+			entry = &history->entries[i];
+			break;
+		}
+	}
+
+	/* Otherwise add one */
+	if (!entry) {
+		struct drm_dp_mst_topology_ref_entry *new;
+		int new_len = history->len + 1;
+
+		new = krealloc(history->entries, sizeof(*new) * new_len,
+			       GFP_KERNEL);
+		if (!new)
+			goto fail_alloc;
+
+		entry = &new[history->len];
+		history->len = new_len;
+		history->entries = new;
+
+		entry->backtrace = backtrace;
+		entry->type = type;
+		entry->count = 0;
+	}
+	entry->count++;
+	entry->ts_nsec = ktime_get_ns();
+
+	return;
+fail_alloc:
+	DRM_WARN_ONCE("Failed to allocate memory for topology refcount backtrace\n");
+}
+
+static int
+topology_ref_history_cmp(const void *a, const void *b)
+{
+	const struct drm_dp_mst_topology_ref_entry *entry_a = a, *entry_b = b;
+
+	if (entry_a->ts_nsec > entry_b->ts_nsec)
+		return 1;
+	else if (entry_a->ts_nsec < entry_b->ts_nsec)
+		return -1;
+	else
+		return 0;
+}
+
+static inline const char *
+topology_ref_type_to_str(enum drm_dp_mst_topology_ref_type type)
+{
+	if (type == DRM_DP_MST_TOPOLOGY_REF_GET)
+		return "get";
+	else
+		return "put";
+}
+
+static void
+__dump_topology_ref_history(struct drm_dp_mst_topology_ref_history *history,
+			    void *ptr, const char *type_str)
+{
+	struct drm_printer p = drm_debug_printer(DBG_PREFIX);
+	char *buf = kzalloc(PAGE_SIZE, GFP_KERNEL);
+	int i;
+
+	if (!buf)
+		return;
+
+	if (!history->len)
+		goto out;
+
+	/* First, sort the list so that it goes from oldest to newest
+	 * reference entry
+	 */
+	sort(history->entries, history->len, sizeof(*history->entries),
+	     topology_ref_history_cmp, NULL);
+
+	drm_printf(&p,
+		   "%s (%p/%px) topology count reached 0, dumping history:\n",
+		   type_str, ptr, ptr);
+
+	for (i = 0; i < history->len; i++) {
+		const struct drm_dp_mst_topology_ref_entry *entry =
+			&history->entries[i];
+		ulong *entries;
+		uint nr_entries;
+		u64 ts_nsec = entry->ts_nsec;
+		u64 rem_nsec = do_div(ts_nsec, 1000000000);
+
+		nr_entries = stack_depot_fetch(entry->backtrace, &entries);
+		stack_trace_snprint(buf, PAGE_SIZE, entries, nr_entries, 4);
+
+		drm_printf(&p, "  %d %ss (last at %5llu.%06llu):\n%s",
+			   entry->count,
+			   topology_ref_type_to_str(entry->type),
+			   ts_nsec, rem_nsec / 1000, buf);
+	}
+
+	/* Now free the history, since this is the only time we expose it */
+	kfree(history->entries);
+out:
+	kfree(buf);
+}
+
+static __always_inline void
+drm_dp_mst_dump_mstb_topology_history(struct drm_dp_mst_branch *mstb)
+{
+	__dump_topology_ref_history(&mstb->topology_ref_history, mstb,
+				    "MSTB");
+}
+
+static __always_inline void
+drm_dp_mst_dump_port_topology_history(struct drm_dp_mst_port *port)
+{
+	__dump_topology_ref_history(&port->topology_ref_history, port,
+				    "Port");
+}
+
+static __always_inline void
+save_mstb_topology_ref(struct drm_dp_mst_branch *mstb,
+		       enum drm_dp_mst_topology_ref_type type)
+{
+	__topology_ref_save(mstb->mgr, &mstb->topology_ref_history, type);
+}
+
+static __always_inline void
+save_port_topology_ref(struct drm_dp_mst_port *port,
+		       enum drm_dp_mst_topology_ref_type type)
+{
+	__topology_ref_save(port->mgr, &port->topology_ref_history, type);
+}
+
+static inline void
+topology_ref_history_lock(struct drm_dp_mst_topology_mgr *mgr)
+{
+	mutex_lock(&mgr->topology_ref_history_lock);
+}
+
+static inline void
+topology_ref_history_unlock(struct drm_dp_mst_topology_mgr *mgr)
+{
+	mutex_unlock(&mgr->topology_ref_history_lock);
+}
+#else
+static inline void
+topology_ref_history_lock(struct drm_dp_mst_topology_mgr *mgr) {}
+static inline void
+topology_ref_history_unlock(struct drm_dp_mst_topology_mgr *mgr) {}
+static inline void
+drm_dp_mst_dump_mstb_topology_history(struct drm_dp_mst_branch *mstb) {}
+static inline void
+drm_dp_mst_dump_port_topology_history(struct drm_dp_mst_port *port) {}
+#define save_mstb_topology_ref(mstb, type)
+#define save_port_topology_ref(port, type)
+#endif
+
 static void drm_dp_destroy_mst_branch_device(struct kref *kref)
 {
 	struct drm_dp_mst_branch *mstb =
 		container_of(kref, struct drm_dp_mst_branch, topology_kref);
 	struct drm_dp_mst_topology_mgr *mgr = mstb->mgr;
 
+	drm_dp_mst_dump_mstb_topology_history(mstb);
+
 	INIT_LIST_HEAD(&mstb->destroy_next);
 
 	/*
@@ -1446,11 +1630,18 @@ static void drm_dp_destroy_mst_branch_device(struct kref *kref)
 static int __must_check
 drm_dp_mst_topology_try_get_mstb(struct drm_dp_mst_branch *mstb)
 {
-	int ret = kref_get_unless_zero(&mstb->topology_kref);
+	int ret;
+
+	topology_ref_history_lock(mstb->mgr);
+	ret = kref_get_unless_zero(&mstb->topology_kref);
 
-	if (ret)
+	if (ret) {
 		DRM_DEBUG("mstb %p/%px (%d)\n",
 			  mstb, mstb, kref_read(&mstb->topology_kref));
+		save_mstb_topology_ref(mstb, DRM_DP_MST_TOPOLOGY_REF_GET);
+	}
+
+	topology_ref_history_unlock(mstb->mgr);
 
 	return ret;
 }
@@ -1471,10 +1662,15 @@ drm_dp_mst_topology_try_get_mstb(struct drm_dp_mst_branch *mstb)
  */
 static void drm_dp_mst_topology_get_mstb(struct drm_dp_mst_branch *mstb)
 {
+	topology_ref_history_lock(mstb->mgr);
+
+	save_mstb_topology_ref(mstb, DRM_DP_MST_TOPOLOGY_REF_GET);
 	WARN_ON(kref_read(&mstb->topology_kref) == 0);
 	kref_get(&mstb->topology_kref);
 	DRM_DEBUG("mstb %p/%px (%d)\n",
 		  mstb, mstb, kref_read(&mstb->topology_kref));
+
+	topology_ref_history_unlock(mstb->mgr);
 }
 
 /**
@@ -1492,9 +1688,14 @@ static void drm_dp_mst_topology_get_mstb(struct drm_dp_mst_branch *mstb)
 static void
 drm_dp_mst_topology_put_mstb(struct drm_dp_mst_branch *mstb)
 {
+	topology_ref_history_lock(mstb->mgr);
+
 	DRM_DEBUG("mstb %p/%px (%d)\n",
 		  mstb, mstb, kref_read(&mstb->topology_kref) - 1);
+	save_mstb_topology_ref(mstb, DRM_DP_MST_TOPOLOGY_REF_PUT);
 	kref_put(&mstb->topology_kref, drm_dp_destroy_mst_branch_device);
+
+	topology_ref_history_unlock(mstb->mgr);
 }
 
 static void drm_dp_destroy_port(struct kref *kref)
@@ -1503,6 +1704,8 @@ static void drm_dp_destroy_port(struct kref *kref)
 		container_of(kref, struct drm_dp_mst_port, topology_kref);
 	struct drm_dp_mst_topology_mgr *mgr = port->mgr;
 
+	drm_dp_mst_dump_port_topology_history(port);
+
 	/* There's nothing that needs locking to destroy an input port yet */
 	if (port->input) {
 		drm_dp_mst_put_port_malloc(port);
@@ -1546,12 +1749,18 @@ static void drm_dp_destroy_port(struct kref *kref)
 static int __must_check
 drm_dp_mst_topology_try_get_port(struct drm_dp_mst_port *port)
 {
-	int ret = kref_get_unless_zero(&port->topology_kref);
+	int ret;
 
-	if (ret)
+	topology_ref_history_lock(port->mgr);
+	ret = kref_get_unless_zero(&port->topology_kref);
+
+	if (ret) {
 		DRM_DEBUG("port %p/%px (%d)\n",
 			  port, port, kref_read(&port->topology_kref));
+		save_port_topology_ref(port, DRM_DP_MST_TOPOLOGY_REF_GET);
+	}
 
+	topology_ref_history_unlock(port->mgr);
 	return ret;
 }
 
@@ -1570,10 +1779,15 @@ drm_dp_mst_topology_try_get_port(struct drm_dp_mst_port *port)
  */
 static void drm_dp_mst_topology_get_port(struct drm_dp_mst_port *port)
 {
+	topology_ref_history_lock(port->mgr);
+
 	WARN_ON(kref_read(&port->topology_kref) == 0);
 	kref_get(&port->topology_kref);
 	DRM_DEBUG("port %p/%px (%d)\n",
 		  port, port, kref_read(&port->topology_kref));
+	save_port_topology_ref(port, DRM_DP_MST_TOPOLOGY_REF_GET);
+
+	topology_ref_history_unlock(port->mgr);
 }
 
 /**
@@ -1589,9 +1803,14 @@ static void drm_dp_mst_topology_get_port(struct drm_dp_mst_port *port)
  */
 static void drm_dp_mst_topology_put_port(struct drm_dp_mst_port *port)
 {
+	topology_ref_history_lock(port->mgr);
+
 	DRM_DEBUG("port %p/%px (%d)\n",
 		  port, port, kref_read(&port->topology_kref) - 1);
+	save_port_topology_ref(port, DRM_DP_MST_TOPOLOGY_REF_PUT);
 	kref_put(&port->topology_kref, drm_dp_destroy_port);
+
+	topology_ref_history_unlock(port->mgr);
 }
 
 static struct drm_dp_mst_branch *
@@ -4414,6 +4633,9 @@ int drm_dp_mst_topology_mgr_init(struct drm_dp_mst_topology_mgr *mgr,
 	mutex_init(&mgr->payload_lock);
 	mutex_init(&mgr->destroy_connector_lock);
 	mutex_init(&mgr->up_req_lock);
+#if IS_ENABLED(CONFIG_DRM_DEBUG_DP_MST_TOPOLOGY_REFS)
+	mutex_init(&mgr->topology_ref_history_lock);
+#endif
 	INIT_LIST_HEAD(&mgr->tx_msg_downq);
 	INIT_LIST_HEAD(&mgr->destroy_connector_list);
 	INIT_LIST_HEAD(&mgr->destroy_branch_device_list);
diff --git a/include/drm/drm_dp_mst_helper.h b/include/drm/drm_dp_mst_helper.h
index eece28525d52..1e05ffee226f 100644
--- a/include/drm/drm_dp_mst_helper.h
+++ b/include/drm/drm_dp_mst_helper.h
@@ -26,6 +26,26 @@
 #include <drm/drm_dp_helper.h>
 #include <drm/drm_atomic.h>
 
+#if IS_ENABLED(CONFIG_DRM_DEBUG_DP_MST_TOPOLOGY_REFS)
+#include <linux/stackdepot.h>
+#include <linux/timekeeping.h>
+
+enum drm_dp_mst_topology_ref_type {
+	DRM_DP_MST_TOPOLOGY_REF_GET,
+	DRM_DP_MST_TOPOLOGY_REF_PUT,
+};
+
+struct drm_dp_mst_topology_ref_history {
+	struct drm_dp_mst_topology_ref_entry {
+		enum drm_dp_mst_topology_ref_type type;
+		int count;
+		ktime_t ts_nsec;
+		depot_stack_handle_t backtrace;
+	} *entries;
+	int len;
+};
+#endif /* IS_ENABLED(CONFIG_DRM_DEBUG_DP_MST_TOPOLOGY_REFS) */
+
 struct drm_dp_mst_branch;
 
 /**
@@ -92,6 +112,14 @@ struct drm_dp_mst_port {
 	 */
 	struct kref malloc_kref;
 
+#if IS_ENABLED(CONFIG_DRM_DEBUG_DP_MST_TOPOLOGY_REFS)
+	/**
+	 * @topology_ref_history: A history of each topology
+	 * reference/dereference. See CONFIG_DRM_DEBUG_DP_MST_TOPOLOGY_REFS.
+	 */
+	struct drm_dp_mst_topology_ref_history topology_ref_history;
+#endif
+
 	u8 port_num;
 	bool input;
 	bool mcs;
@@ -162,6 +190,14 @@ struct drm_dp_mst_branch {
 	 */
 	struct kref malloc_kref;
 
+#if IS_ENABLED(CONFIG_DRM_DEBUG_DP_MST_TOPOLOGY_REFS)
+	/**
+	 * @topology_ref_history: A history of each topology
+	 * reference/dereference. See CONFIG_DRM_DEBUG_DP_MST_TOPOLOGY_REFS.
+	 */
+	struct drm_dp_mst_topology_ref_history topology_ref_history;
+#endif
+
 	/**
 	 * @destroy_next: linked-list entry used by
 	 * drm_dp_destroy_connector_work()
@@ -628,6 +664,15 @@ struct drm_dp_mst_topology_mgr {
 	 * transmissions.
 	 */
 	struct work_struct up_req_work;
+
+#if IS_ENABLED(CONFIG_DRM_DEBUG_DP_MST_TOPOLOGY_REFS)
+	/**
+	 * @topology_ref_history_lock: protects
+	 * &drm_dp_mst_port.topology_ref_history and
+	 * &drm_dp_mst_branch.topology_ref_history.
+	 */
+	struct mutex topology_ref_history_lock;
+#endif
 };
 
 int drm_dp_mst_topology_mgr_init(struct drm_dp_mst_topology_mgr *mgr,
-- 
2.21.0