[PATCH] usb: core: Fix side effect of clear port feature in hub port reset

From: hyunho747.kim
Date: Tue Sep 01 2015 - 03:29:45 EST

After successful port reset by set_port_feature, some devices show
immediate link connection which generates port connect change interrupt.
But, the next step is an unconditional usb_clear_port_feature
and this flow always clears USB_PORT_FEAT_C_CONNECTION bit in port status
though next kick_khubd is reserved by link connection interrupt.
This flow can make an ambiguous state in the next port_evnet operation,
port_status is connected state such as 0x203 but
port_change is zero value caused by the previous clear port feature.
So, port_event can't call hub_port_connect_change and
there is no other way to peform connect procedure.

Signed-off-by: hyunho747.kim <hyunho747.kim@xxxxxxx>
drivers/usb/core/hub.c | 11 +++++++----
1 file changed, 7 insertions(+), 4 deletions(-)

diff --git a/drivers/usb/core/hub.c b/drivers/usb/core/hub.c
index 73dfa19..4508aff 100644
--- a/drivers/usb/core/hub.c
+++ b/drivers/usb/core/hub.c
@@ -2761,18 +2761,21 @@ static int hub_port_reset(struct usb_hub *hub, int port1,

/* Check for disconnect or reset */
if (status == 0 || status == -ENOTCONN || status == -ENODEV) {
- usb_clear_port_feature(hub->hdev, port1,
+ if (status)
+ usb_clear_port_feature(hub->hdev, port1,

if (!hub_is_superspeed(hub->hdev))
goto done;

- usb_clear_port_feature(hub->hdev, port1,
+ if (status) {
+ usb_clear_port_feature(hub->hdev, port1,
- usb_clear_port_feature(hub->hdev, port1,
+ usb_clear_port_feature(hub->hdev, port1,
- usb_clear_port_feature(hub->hdev, port1,
+ usb_clear_port_feature(hub->hdev, port1,
+ }

* If a USB 3.0 device migrates from reset to an error

To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/