Subject: Re: ohci overuns
To: Chris Gilbert <chris@paradox.demon.co.uk>
From: Lennart Augustsson <lennart@mail.augustsson.net>
List: port-arm32
Date: 05/18/2001 11:56:27
I could make a quirk for that chip to ignore scheduling overrun.
Could anyone test it with USB devices if I do?

    -- Lennart


Chris Gilbert wrote:

> There's nothing connected to the USB (other than the internal hub).  However
> I suspect interrupt latency maybe a factor?  It certainly happens more when X
> is being used.  Certainly looks to be the chips that's a common factor here
> (both on arm32 and i386)
>
> Cheers,
> Chris
>
> On Friday 18 May 2001  9:57 am, Lennart Augustsson wrote:
> > The scheduling overrun interrupt can only occur when the OHCI controller
> > determines that it cannot complete its periodic tasks within the  1 ms time
> > frame.  While this can technically happen, you would have to have heavy
> > isochronous (e.g. audio) traffic for it to occur.
> > If there are no USB devices connected it cannot happen.  With no audio
> > running I cannot see how it could happen.
> > So if it happens, there's a bug somewhere.  In the chip or in the driver.
> > Having written the driver, I'll blame the chip. :-)
> >
> >     -- Lennart
> >
> > Chris Gilbert wrote:
> > > On Thursday 17 May 2001 10:52 am, Thilo Manske wrote:
> > > > On Wed, May 16 2001 at 23:06:33 +0100, Chris Gilbert wrote:
> > > > > on my arm32 CATS box I've been getting:
> > > > > ohci0: scheduling overrun
> > > > >
> > > > > And I was wondering if anyone else was seeing it.
> > > >
> > > > Yes, I used to get this on i386.
> > > >
> > > > But they look more like this:
> > > > "ohci0: n scheduling overruns"
> > > >
> > > > with n=1..6
> > > >
> > > > It's the onboard USB interface of on Asus P5A, ALi Chipset.
> > > > I don't have USB in my kernel anymore, so I can't quote a better dmesg
> > > > snippet than this:
> > > >
> > > > vendor 0x10b9 product 0x5237 (USB serial bus, interface 0x10, revision
> > > > 0x03) at pci0 dev 2 function 0 not configured
> > >
> > > The CATS runs:
> > > ohci0 at pci0 dev 20 function 0: Acer Labs M5237 USB Host Controller
> > > (rev.0x03)
> > > So perhaps it's a bug/quirk of ALi USB and/or our driver.
> > >
> > > Cheers,
> > > Chris