Re: [PATCH] usb: core: don't try to reset_device() a port that gotjust disconnected

[Alan Stern]

On Tue, 30 Jul 2013, Julius Werner wrote:

> The USB hub driver's event handler contains a check to catch SuperSpeed
> devices that transitioned into the SS.Inactive state and tries to fix
> them with a reset. It decides whether to do a plain hub port reset or
> call the usb_reset_device() method based on whether there was a device
> attached to the port.
> 
> However, there are device/hub combinations (found with a JetFlash
> Transcend mass storage stick (8564:1000) on the root hub of an Intel
> LynxPoint PCH) which can transition to the SS.Inactive state on
> disconnect (and stay there long enough for the host to notice). In this
> case, above-mentioned reset check will call usb_reset_device() on the
> now disconnected device. This leads to a futile attempt of asking a
> bunch of floating wires to disable device-initiated U1 and U2, each
> stalling khubd for a 5 second timeout before finally accepting the
> device's fate amongst a flurry of error messages.
 
Wait a moment.  Why does each of these attempts lead to a 5-second
timeout?  Why don't they fail immediately?

> This patch makes the choice of reset dependent on the port status that
> has just been read from the hub, as opposed to any previous in-kernel
> state, thus working around this problem.
> 
> Signed-off-by: Julius Werner <jwerner@xxxxxxxxxxxx>
> ---
>  drivers/usb/core/hub.c | 4 ++--
>  1 file changed, 2 insertions(+), 2 deletions(-)
> 
> diff --git a/drivers/usb/core/hub.c b/drivers/usb/core/hub.c
> index 4a8a1d6..ace0c52 100644
> --- a/drivers/usb/core/hub.c
> +++ b/drivers/usb/core/hub.c
> @@ -4798,7 +4798,7 @@ static void hub_events(void)
>  					hub->ports[i - 1]->child;
>  
>  				dev_dbg(hub_dev, "warm reset port %d\n", i);
> -				if (!udev) {
> +				if (!(portstatus & USB_PORT_STAT_CONNECTION)) {
>  					status = hub_port_reset(hub, i,
>  							NULL, HUB_BH_RESET_TIME,
>  							true);
> @@ -4808,8 +4808,8 @@ static void hub_events(void)
>  					usb_lock_device(udev);
>  					status = usb_reset_device(udev);
>  					usb_unlock_device(udev);

What will happen here if udev is NULL?  Maybe you should change the 
test to (!udev || !(portstatus & ...)).

> +					connect_change = 0;
>  				}
> -				connect_change = 0;
>  			}

Alan Stern

--
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/