> Am Mittwoch, 9. Juli 2003 14:42 schrieb Rob van Nieuwkerk:
> > >
> > > > > Just as an experiment, could you connect a second device to the bus?
> > > >
> > > > Sure.
> > > > But I'll need to find/make another USB connector: the board I use is
> > > > a small embedded Geode board with only one USB connector, the second one
> > > > is just 4 pins on the board. Will work something out.
> > > >
> > > > But what would be a usable device as the second device ?
> > > > At the moment I can choose between:
> > > >
> > > > - another Philips webcam
> > > > - SanDisk CF reader
> > > > - Nokia DKU-5 mobile phone cable
> > > >
> > > > Another camera will probably suffer from the same problem and the CF-reader
> > > > & cable won't probably generate interrupts by themselves ? I can find/buy
> > > > another USB device if necessary. Will any of the above do ?
> > >
> > > Just a hub would do. I'd just like to see whether another device on the bus
> > > will keep the controller alive.
> > > But it's not important enough to justify buying something for that.
> >
> > Hi Oliver,
> >
> > I can't get the cable for the second USB port (and don't have stuff
> > available now for building my own). But if I understand correctly
> > connecting a hub to the USB port and connecting the camera via the
> > hub would be a useful test (I don't know to much about USB-hubs) ?
> >
> > I don't have a hub around but I can get one, no problem !
>
> Something stops OHCI. In fact I think this code from
> static void hc_interrupt (int irq, void * __ohci, struct pt_regs * r)
> is to blame:
>
> // FIXME: this assumes SOF (1/ms) interrupts don't get lost...
> if (ints & OHCI_INTR_SF) {
> unsigned int frame = le16_to_cpu (ohci->hcca->frame_no) & 1;
> writel (OHCI_INTR_SF, ®s->intrdisable);
> if (ohci->ed_rm_list[!frame] != NULL) {
> dl_del_list (ohci, !frame);
> }
> if (ohci->ed_rm_list[frame] != NULL)
> writel (OHCI_INTR_SF, ®s->intrenable);
> }
>
> Unfortunately I don't have time to debug OHCI on a silicon level this week.
>
> But there's a slim chance that a second device on the bus that'll cause
> activity continuously (hub, mouse) might help.
>
> Regards
> Oliver
Hi Oliver,
FYI:
I arranged a hub and connected the Philips webcam to the hub and
connected the hub to the machine's USB port. And I killed the webcam
by a "cat /dev/hda > /dev/null".
After that I plugged in the USB CF reader in the hub and the OHCI
interrupt count increases then (and also when removing the reader). So
OHCI can still get interrupt when there's a dead camera.
There were kernel messages about the device being plugged in and removed.
Please tell me if you want more/other tests/experiments !
greetings,
Rob van Nieuwkerk
-------------------------------------------------------
This SF.Net email sponsored by: Parasoft
Error proof Web apps, automate testing & more.
Download & eval WebKing and get a free book.
www.parasoft.com/bulletproofapps
_______________________________________________
[EMAIL PROTECTED]
To unsubscribe, use the last form field at:
https://lists.sourceforge.net/lists/listinfo/linux-usb-devel
