On Fri, Aug 22, 2008 at 01:32:12PM -0400, Alan Stern wrote:
> On Fri, 22 Aug 2008 ajay.gupta@xxxxxx wrote:
>
> > @@ -1785,9 +1784,11 @@ static int musb_urb_enqueue(
> > * REVISIT consider a dedicated qh kmem_cache, so it's harder
> > * for bugs in other kernel code to break this driver...
> > */
>
> This doesn't look right. After calling spin_lock_irqsave you are in an
> atomic context, so you can't pass mem_flags to kzalloc.
>
> Either move the spin_lock_irqsave after the call to kzalloc or else
> change mem_flags to GFP_ATOMIC.
first one, please.
should be like:
qh = kzalloc(sizeof *qh, mem_flags);
if (!qh) {
+ spin_lock_irqsave(&musb->lock, flags);
usb_hcd_unlink_urb_from_ep(hcd, urb);
+ spin_unlock_irqrestore(&musb->lock, flags);
return -ENOMEM;
}
Similarly at the end of enqueue(),
if (ret != 0) {
+ spin_lock_irqsave(&musb->lock, flags);
usb_hcd_unlink_urb_from_ep(hcd, urb);
+ spin_unlock_irqrestore(&musb->lock, flags);
kfree(qh);
}
So hold the musb->lock at the two places in enqueue path and no
other change. If this is fine then I will resubmit with this modification.
--Ajay
--
balbi--
To unsubscribe from this list: send the line "unsubscribe linux-omap" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
