Re: [PATCH v2 2/3] usb: xhci: This reworks ff8cbf250b448aac35589f6075082c3fcad8a8fe

[Alan Stern]

On Tue, 4 Nov 2014, Lu Baolu wrote:

> xhci: clear root port wake on bits if controller isn't wake-up capable
> 
> When system is being suspended, if host device is not wakeup capable,
> xhci_suspend() needs to clear all root port wake on bits.  Otherwise,
> some platforms may generate spurious wakeup, even if PCI PME# is dis-
> abled.
> 
> Signed-off-by: Lu Baolu <baolu.lu@xxxxxxxxxxxxxxx>
> ---
>  drivers/usb/host/xhci.c | 42 ++++++++++++++++++++++++++++++++++++++++++
>  1 file changed, 42 insertions(+)
> 
> diff --git a/drivers/usb/host/xhci.c b/drivers/usb/host/xhci.c
> index 2a5d45b..cd57aae 100644
> --- a/drivers/usb/host/xhci.c
> +++ b/drivers/usb/host/xhci.c
> @@ -35,6 +35,8 @@
>  #define DRIVER_AUTHOR "Sarah Sharp"
>  #define DRIVER_DESC "'eXtensible' Host Controller (xHC) Driver"
>  
> +#define	PORT_WAKE_BITS	(PORT_WKOC_E | PORT_WKDISC_E | PORT_WKCONN_E)
> +
>  /* Some 0.95 hardware can't handle the chain bit on a Link TRB being cleared */
>  static int link_quirk;
>  module_param(link_quirk, int, S_IRUGO | S_IWUSR);
> @@ -851,6 +853,42 @@ static void xhci_clear_command_ring(struct xhci_hcd *xhci)
>  	xhci_set_cmd_ring_deq(xhci);
>  }
>  
> +static void xhci_disable_port_wake_on_bits(struct xhci_hcd *xhci)
> +{
> +	int port_index;
> +	__le32 __iomem **port_array;
> +	unsigned long flags;
> +	u32 t1, t2;
> +
> +	spin_lock_irqsave(&xhci->lock, flags);
> +
> +	/* disble usb3 ports Wake bits*/
> +	port_index = xhci->num_usb3_ports;
> +	port_array = xhci->usb3_ports;
> +	while (port_index--) {
> +		t1 = readl(port_array[port_index]);
> +		t2 = xhci_port_state_to_neutral(t1);
> +		t2 &= ~PORT_WAKE_BITS;
> +		t1 = xhci_port_state_to_neutral(t1);
> +		if (t1 != t2)
> +			writel(t2, port_array[port_index]);
> +	}
> +
> +	/* disble usb2 ports Wake bits*/
> +	port_index = xhci->num_usb2_ports;
> +	port_array = xhci->usb2_ports;
> +	while (port_index--) {
> +		t1 = readl(port_array[port_index]);
> +		t2 = xhci_port_state_to_neutral(t1);
> +		t2 &= ~PORT_WAKE_BITS;
> +		t1 = xhci_port_state_to_neutral(t1);
> +		if (t1 != t2)
> +			writel(t2, port_array[port_index]);
> +	}
> +
> +	spin_unlock_irqrestore(&xhci->lock, flags);
> +}
> +
>  /*
>   * Stop HC (not bus-specific)
>   *
> @@ -868,6 +906,10 @@ int xhci_suspend(struct xhci_hcd *xhci)
>  			xhci->shared_hcd->state != HC_STATE_SUSPENDED)
>  		return -EINVAL;
>  
> +	/* Clear root port wake on bits if not wakeup capable. */
> +	if (!device_may_wakeup(hcd->self.controller))
> +		xhci_disable_port_wake_on_bits(xhci);
> +
>  	/* Don't poll the roothubs on bus suspend. */
>  	xhci_dbg(xhci, "%s: stopping port polling.\n", __func__);
>  	clear_bit(HCD_FLAG_POLL_RH, &hcd->flags);

This is better but still wrong.  Remember, this same code gets called
for system suspend _and_ for runtime suspend.  During runtime suspend,
wakeup is always supposed to be turned on, even if device_may_wakeup()
is false.  That's because device_may_wakeup() refers only to system
suspend.  What you need to test is the do_wakeup flag, which should be
passed into xhci_suspend() by xhci_pci_suspend() and
xhci_plat_suspend().

Another problem is in the patch description and the comments.  If 
device_may_wakeup() returns false, it doesn't mean the controller isn't 
wakeup-capable -- it means the controller isn't _allowed_ to wake up 
the system.  Those are two different things.

Finally, the code in xhci_disable_port_wake_on_bits() looks a little 
peculiar -- I'm not sure about all those calls to 
xhci_port_state_to_neutral().  But I'm not an expert on that; Mathias 
will have to advise on it.

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/