[3.19.y-ckt stable] Patch "drm/dp/mst: close deadlock in connector destruction." has been added to staging queue

[Kamal Mostafa]

This is a note to let you know that I have just added a patch titled

    drm/dp/mst: close deadlock in connector destruction.

to the linux-3.19.y-queue branch of the 3.19.y-ckt extended stable tree 
which can be found at:

    http://kernel.ubuntu.com/git/ubuntu/linux.git/log/?h=linux-3.19.y-queue

This patch is scheduled to be released in version 3.19.y-ckt4.

If you, or anyone else, feels it should not be added to this tree, please 
reply to this email.

For more information about the 3.19.y-ckt tree, see
https://wiki.ubuntu.com/Kernel/Dev/ExtendedStable

Thanks.
-Kamal

------

>From cb52b6829451c893c1ad018ad6dafbf2730b5b78 Mon Sep 17 00:00:00 2001
From: Dave Airlie <airlied at redhat.com>
Date: Mon, 15 Jun 2015 10:34:28 +1000
Subject: drm/dp/mst: close deadlock in connector destruction.

commit 6b8eeca65b18ae77e175cc2b6571731f0ee413bf upstream.

I've only seen this once, and I failed to capture the
lockdep backtrace, but I did some investigations.

If we are calling into the MST layer from EDID probing,
we have the mode_config mutex held, if during that EDID
probing, the MST hub goes away, then we can get a deadlock
where the connector destruction function in the driver
tries to retake the mode config mutex.

This offloads connector destruction to a workqueue,
and avoid the subsequenct lock ordering issue.

Acked-by: Daniel Vetter <daniel.vetter at ffwll.ch>
Signed-off-by: Dave Airlie <airlied at redhat.com>
Signed-off-by: Kamal Mostafa <kamal at canonical.com>
---
 drivers/gpu/drm/drm_dp_mst_topology.c | 40 +++++++++++++++++++++++++++++++++--
 include/drm/drm_crtc.h                |  2 ++
 include/drm/drm_dp_mst_helper.h       |  4 ++++
 3 files changed, 44 insertions(+), 2 deletions(-)

diff --git a/drivers/gpu/drm/drm_dp_mst_topology.c b/drivers/gpu/drm/drm_dp_mst_topology.c
index 4d7a4d9..30308ab 100644
--- a/drivers/gpu/drm/drm_dp_mst_topology.c
+++ b/drivers/gpu/drm/drm_dp_mst_topology.c
@@ -863,8 +863,16 @@ static void drm_dp_destroy_port(struct kref *kref)
 		port->vcpi.num_slots = 0;

 		kfree(port->cached_edid);
-		if (port->connector)
-			(*port->mgr->cbs->destroy_connector)(mgr, port->connector);
+
+		/* we can't destroy the connector here, as
+		   we might be holding the mode_config.mutex
+		   from an EDID retrieval */
+		if (port->connector) {
+			mutex_lock(&mgr->destroy_connector_lock);
+			list_add(&port->connector->destroy_list, &mgr->destroy_connector_list);
+			mutex_unlock(&mgr->destroy_connector_lock);
+			schedule_work(&mgr->destroy_connector_work);
+		}
 		drm_dp_port_teardown_pdt(port, port->pdt);

 		if (!port->input && port->vcpi.vcpi > 0)
@@ -2631,6 +2639,30 @@ static void drm_dp_tx_work(struct work_struct *work)
 	mutex_unlock(&mgr->qlock);
 }

+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_connector *connector;
+
+	/*
+	 * Not a regular list traverse as we have to drop the destroy
+	 * connector lock before destroying the connector, to avoid AB->BA
+	 * ordering between this lock and the config mutex.
+	 */
+	for (;;) {
+		mutex_lock(&mgr->destroy_connector_lock);
+		connector = list_first_entry_or_null(&mgr->destroy_connector_list, struct drm_connector, destroy_list);
+		if (!connector) {
+			mutex_unlock(&mgr->destroy_connector_lock);
+			break;
+		}
+		list_del(&connector->destroy_list);
+		mutex_unlock(&mgr->destroy_connector_lock);
+
+		mgr->cbs->destroy_connector(mgr, connector);
+	}
+}
+
 /**
  * drm_dp_mst_topology_mgr_init - initialise a topology manager
  * @mgr: manager struct to initialise
@@ -2650,10 +2682,13 @@ int drm_dp_mst_topology_mgr_init(struct drm_dp_mst_topology_mgr *mgr,
 	mutex_init(&mgr->lock);
 	mutex_init(&mgr->qlock);
 	mutex_init(&mgr->payload_lock);
+	mutex_init(&mgr->destroy_connector_lock);
 	INIT_LIST_HEAD(&mgr->tx_msg_upq);
 	INIT_LIST_HEAD(&mgr->tx_msg_downq);
+	INIT_LIST_HEAD(&mgr->destroy_connector_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_waitqueue_head(&mgr->tx_waitq);
 	mgr->dev = dev;
 	mgr->aux = aux;
@@ -2678,6 +2713,7 @@ EXPORT_SYMBOL(drm_dp_mst_topology_mgr_init);
  */
 void drm_dp_mst_topology_mgr_destroy(struct drm_dp_mst_topology_mgr *mgr)
 {
+	flush_work(&mgr->destroy_connector_work);
 	mutex_lock(&mgr->payload_lock);
 	kfree(mgr->payloads);
 	mgr->payloads = NULL;
diff --git a/include/drm/drm_crtc.h b/include/drm/drm_crtc.h
index b863298..3dab77c 100644
--- a/include/drm/drm_crtc.h
+++ b/include/drm/drm_crtc.h
@@ -689,6 +689,8 @@ struct drm_connector {
 	uint8_t num_h_tile, num_v_tile;
 	uint8_t tile_h_loc, tile_v_loc;
 	uint16_t tile_h_size, tile_v_size;
+
+	struct list_head destroy_list;
 };

 /**
diff --git a/include/drm/drm_dp_mst_helper.h b/include/drm/drm_dp_mst_helper.h
index 00c1da9..180f8a3 100644
--- a/include/drm/drm_dp_mst_helper.h
+++ b/include/drm/drm_dp_mst_helper.h
@@ -463,6 +463,10 @@ struct drm_dp_mst_topology_mgr {
 	struct work_struct work;

 	struct work_struct tx_work;
+
+	struct list_head destroy_connector_list;
+	struct mutex destroy_connector_lock;
+	struct work_struct destroy_connector_work;
 };

 int drm_dp_mst_topology_mgr_init(struct drm_dp_mst_topology_mgr *mgr, struct device *dev, struct drm_dp_aux *aux, int max_dpcd_transaction_bytes, int max_payloads, int conn_base_id);
--
1.9.1