On Monday 13 December 2004 4:16 am, Flavio Leitner wrote:
> On Wed, Dec 08, 2004 at 05:17:26PM -0800, David Brownell wrote:
> > On Wednesday 08 December 2004 5:16 am, Flavio Leitner wrote:
> > > Is there anything else that I can help you?
> >
> > Find out whether that hardware is for some reason
> > broken, so that it issues INTR_RD when it shouldn't.
> > That includes verifying that the interrupt counts
> > in /proc/interrupts are absurd for an idle USB host.
>
> Something about 3 or 4 interrupts per second
> for an idle USB host.
That counts as "absurd". Is it a shared IRQ?
> > Also, verify that the messages stop if you get
> > rid of CONFIG_PM.
>
> The problem appears on 2.6.10-rc1 patch and disabling
> CONFIG_PM no message is printed.
And do the IRQs continue at that point? You had the
Serverworks box, right? Does the attached patch change
anything for you?
- Dave
Don't try autosuspend if the hardware isn't known to handle it.
For example, the amd756 erratum 4 workaround needs this logic.
--- 1.49/drivers/usb/host/ohci-hub.c Sat Dec 11 06:22:07 2004
+++ edited/drivers/usb/host/ohci-hub.c Mon Dec 13 08:26:11 2004
@@ -305,7 +305,7 @@
{
struct ohci_hcd *ohci = hcd_to_ohci (hcd);
int ports, i, changed = 0, length = 1;
- int can_suspend = 1;
+ int can_suspend = hcd->can_wakeup;
unsigned long flags;
spin_lock_irqsave (&ohci->lock, flags);
