On Tue, Mar 08, 2011 at 11:40:33AM -0500, Alan Stern wrote:
> On Mon, 7 Mar 2011, Sarah Sharp wrote:
> > The xHCI driver will need to ensure that HC_STATE_HALT, HC_STATE_RUNNING,
> > and HC_STATE_QUIESCING will be set for both the roothubs.
>
> Fortunately this is no longer necessary, unless xhci-hcd uses
> hcd->state internally.
Alan, I tried to remove all references to HC_STATE_HALT and
HC_STATE_RUNNING in the xHCI driver, since it only sets them and never
tests them.
The problem is, the roothub allocation fails in register_root_hub().
HC_STATE_HALT is defined to be zero, and the USB core never sets
HC_STATE_RUNNING in the roothub allocation function, so if the xHCI
driver never sets HC_STATE_RUNNING itself, the roothub allocation will
fail here because of the second conditional in the if statement:
static int register_root_hub(struct usb_hcd *hcd)
{
...
if (retval == 0) {
spin_lock_irq (&hcd_root_hub_lock);
hcd->rh_registered = 1;
spin_unlock_irq (&hcd_root_hub_lock);
/* Did the HC die before the root hub was registered? */
if (HCD_DEAD(hcd) || hcd->state == HC_STATE_HALT)
usb_hc_died (hcd); /* This time clean up */
}
I really want to move the xHCI driver away from having to set
hcd->state. Can you rework your hcd->state patch to accommodate drivers
that never touch it?
Sarah Sharp
--
To unsubscribe from this list: send the line "unsubscribe linux-usb" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
