Skip to contentSkip to conversation

thunderbolt: A couple of fixes for v6.2-rc

PATCH In review Mika Westerberg 2 participants 6 messages Jan 4, 2023 – Jan 9, 2023

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

cover thunderbolt: A couple of fixes for v6.2-rc

Hi all,

This series includes a couple of fixes for issues found/reported.

Mika Westerberg (3):
thunderbolt: Do not call PM runtime functions in tb_retimer_scan()
thunderbolt: Use correct function to calculate maximum USB3 link rate
thunderbolt: Disable XDomain lane 1 only in software connection manager

 drivers/thunderbolt/retimer.c | 17 +++--------------
 drivers/thunderbolt/tb.c      | 20 +++++++++++++++-----
 drivers/thunderbolt/tunnel.c  |  2 +-
 drivers/thunderbolt/xdomain.c | 17 ++++++++++++-----
 4 files changed, 31 insertions(+), 25 deletions(-)
Signature · 1 line
2.35.1
#1 lore raw

PATCH 1/3 thunderbolt: Do not call PM runtime functions in tb_retimer_scan()

We cannot call PM runtime functions in tb_retimer_scan() because it will also be called when retimers are scanned from userspace (happens when there is no device connected on ChromeOS for instance) and at the same USB4 port runtime resume hook. This leads to hang because neither can proceed.

Fix this by runtime resuming USB4 ports in tb_scan_port() instead. This makes sure the ports are runtime PM active when retimers are added under it while avoiding the reported hang as well.

Fixes 1e56c88adecc ("thunderbolt: Runtime resume USB4 port when retimers are scanned")
---
 drivers/thunderbolt/retimer.c | 17 +++--------------
 drivers/thunderbolt/tb.c      | 20 +++++++++++++++-----
 2 files changed, 18 insertions(+), 19 deletions(-)
drivers/thunderbolt/retimer.c +3 -14
index 6ebe7a2886ec..56008eb91e2e 100644
@@ -427,13 +427,6 @@ int tb_retimer_scan(struct tb_port *port, bool add)
 {
 	u32 status[TB_MAX_RETIMER_INDEX + 1] = {};
 	int ret, i, last_idx = 0;
-	struct usb4_port *usb4;
-
-	usb4 = port->usb4;
-	if (!usb4)
-		return 0;
-
-	pm_runtime_get_sync(&usb4->dev);
 
 	/*
 	 * Send broadcast RT to make sure retimer indices facing this
@@ -441,7 +434,7 @@ int tb_retimer_scan(struct tb_port *port, bool add)
 	 */
 	ret = usb4_port_enumerate_retimers(port);
 	if (ret)
-		goto out;
+		return ret;
 
 	/*
 	 * Enable sideband channel for each retimer. We can do this
@@ -471,11 +464,11 @@ int tb_retimer_scan(struct tb_port *port, bool add)
 			break;
 	}
 
-	ret = 0;
 	if (!last_idx)
-		goto out;
+		return 0;
 
 	/* Add on-board retimers if they do not exist already */
+	ret = 0;
 	for (i = 1; i <= last_idx; i++) {
 		struct tb_retimer *rt;
 
@@ -489,10 +482,6 @@ int tb_retimer_scan(struct tb_port *port, bool add)
 		}
 	}
 
-out:
-	pm_runtime_mark_last_busy(&usb4->dev);
-	pm_runtime_put_autosuspend(&usb4->dev);
-
 	return ret;
 }
 
drivers/thunderbolt/tb.c +15 -5
index 462845804427..3f1ab30c4fb1 100644
@@ -628,11 +628,15 @@ static void tb_scan_port(struct tb_port *port)
 			 * Downstream switch is reachable through two ports.
 			 * Only scan on the primary port (link_nr == 0).
 			 */
+
+	if (port->usb4)
+		pm_runtime_get_sync(&port->usb4->dev);
+
 	if (tb_wait_for_port(port, false) <= 0)
-		return;
+		goto out_rpm_put;
 	if (port->remote) {
 		tb_port_dbg(port, "port already has a remote\n");
-		return;
+		goto out_rpm_put;
 	}
 
 	tb_retimer_scan(port, true);
@@ -647,12 +651,12 @@ static void tb_scan_port(struct tb_port *port)
 		 */
 		if (PTR_ERR(sw) == -EIO || PTR_ERR(sw) == -EADDRNOTAVAIL)
 			tb_scan_xdomain(port);
-		return;
+		goto out_rpm_put;
 	}
 
 	if (tb_switch_configure(sw)) {
 		tb_switch_put(sw);
-		return;
+		goto out_rpm_put;
 	}
 
 	/*
@@ -681,7 +685,7 @@ static void tb_scan_port(struct tb_port *port)
 
 	if (tb_switch_add(sw)) {
 		tb_switch_put(sw);
-		return;
+		goto out_rpm_put;
 	}
 
 	/* Link the switches using both links if available */
@@ -733,6 +737,12 @@ static void tb_scan_port(struct tb_port *port)
 
 	tb_add_dp_resources(sw);
 	tb_scan_switch(sw);
+
+out_rpm_put:
+	if (port->usb4) {
+		pm_runtime_mark_last_busy(&port->usb4->dev);
+		pm_runtime_put_autosuspend(&port->usb4->dev);
+	}
 }
 
 static void tb_deactivate_and_free_tunnel(struct tb_tunnel *tunnel)
Signature · 1 line
2.35.1
#2 lore raw

PATCH 2/3 thunderbolt: Use correct function to calculate maximum USB3 link rate

In reply to Mika Westerberg

We need to take minimum of both sides of the USB3 link into consideration, not just the downstream port. Fix this by calling tb_usb3_max_link_rate() instead.

Fixes 0bd680cd900c ("thunderbolt: Add USB3 bandwidth management")
---
 drivers/thunderbolt/tunnel.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/thunderbolt/tunnel.c +1 -1
index 2c3cf7fc3357..1fc3c29b24f8 100644
@@ -1275,7 +1275,7 @@ static void tb_usb3_reclaim_available_bandwidth(struct tb_tunnel *tunnel,
 		return;
 	} else if (!ret) {
 		/* Use maximum link rate if the link valid is not set */
-		ret = usb4_usb3_port_max_link_rate(tunnel->src_port);
+		ret = tb_usb3_max_link_rate(tunnel->dst_port, tunnel->src_port);
 		if (ret < 0) {
 			tb_tunnel_warn(tunnel, "failed to read maximum link rate\n");
 			return;
Signature · 1 line
2.35.1
#3 lore raw

PATCH 3/3 thunderbolt: Disable XDomain lane 1 only in software connection manager

In reply to Mika Westerberg

When firmware connection manager is in use we should not touch the lane adapter (well or any) configuration space so do this only when we know that the software connection manager is active.

Fixes 8e1de7042596 ("thunderbolt: Add support for XDomain lane bonding")
---
 drivers/thunderbolt/xdomain.c | 17 ++++++++++++-----
 1 file changed, 12 insertions(+), 5 deletions(-)
drivers/thunderbolt/xdomain.c +12 -5
index cfa83486c9da..3c51e47dd86b 100644
@@ -1419,12 +1419,19 @@ static int tb_xdomain_get_properties(struct tb_xdomain *xd)
 	 * registered, we notify the userspace that it has changed.
 	 */
 	if (!update) {
-		struct tb_port *port;
+		/*
+		 * Now disable lane 1 if bonding was not enabled. Do
+		 * this only if bonding was possible at the beginning
+		 * (that is we are the connection manager and there are
+		 * two lanes).
+		 */
+		if (xd->bonding_possible) {
+			struct tb_port *port;
 
-		/* Now disable lane 1 if bonding was not enabled */
-		port = tb_port_at(xd->route, tb_xdomain_parent(xd));
-		if (!port->bonded)
-			tb_port_disable(port->dual_link_port);
+			port = tb_port_at(xd->route, tb_xdomain_parent(xd));
+			if (!port->bonded)
+				tb_port_disable(port->dual_link_port);
+		}
 
 		if (device_add(&xd->dev)) {
 			dev_err(&xd->dev, "failed to add XDomain device\n");
Signature · 1 line
2.35.1
maintainer Acked-by on the v1 series
#4 lore raw
In reply to Mika Westerberg
Quoted 10 lines · Mika Westerberg

Hi all,

This series includes a couple of fixes for issues found/reported.

Mika Westerberg (3):
thunderbolt: Do not call PM runtime functions in tb_retimer_scan()
thunderbolt: Use correct function to calculate maximum USB3 link rate
thunderbolt: Disable XDomain lane 1 only in software connection manager

Quoted 9 lines
 drivers/thunderbolt/retimer.c | 17 +++--------------
 drivers/thunderbolt/tb.c      | 20 +++++++++++++++-----
 drivers/thunderbolt/tunnel.c  |  2 +-
 drivers/thunderbolt/xdomain.c | 17 ++++++++++++-----
 4 files changed, 31 insertions(+), 25 deletions(-)

--
2.35.1

#5 lore raw
Quoted 18 lines · Yehezkel Bernat
Quoted 16 lines · Mika Westerberg

Hi all,

This series includes a couple of fixes for issues found/reported.

Mika Westerberg (3):
thunderbolt: Do not call PM runtime functions in tb_retimer_scan()
thunderbolt: Use correct function to calculate maximum USB3 link rate
thunderbolt: Disable XDomain lane 1 only in software connection manager

 drivers/thunderbolt/retimer.c | 17 +++--------------
 drivers/thunderbolt/tb.c      | 20 +++++++++++++++-----
 drivers/thunderbolt/tunnel.c  |  2 +-
 drivers/thunderbolt/xdomain.c | 17 ++++++++++++-----
 4 files changed, 31 insertions(+), 25 deletions(-)
Quoted 5 lines
Quoted 3 lines

--
2.35.1

Thanks!

All applied to fixes.