OHCI fix for Belkin BusPort card on PowerMac

OHCI fix for Belkin BusPort card on PowerMac

[iweiny]

Hello and thanks to all who have responded to my pleas for help.

I thought I would thank you and at the same time bring up an interesting
behavior I found to get my USB-DC265 Kodak camera to work.

I am running:
PowerMac 8500 (604 processor)
USB BusPort from Belkin (Model: F5U005-MAC) PCI USB card
LinuxPPC dist
2.2.12 kernel (self compiled)
2.3.29 USB stack from Paul Mackerras's rsync tree
opendis-0.0.3-pre4 from http://digitalux.netpedia.net

After many long nights of reading print statements in kernel logs.  I found
that the OHCI driver was IN FACT TALKING to the camera!

However, when requesting more data then the camera had to give the returned
status was "9" and not "0".  The OHCI driver interpreted this as an error and
subsequently the user level program would think no data had been returned.

After verifing the data to be correct I applied a very simple correction of:

/* Thinking this is what my controler returns when it is done? */
#define USB_ST_EOF           9

and used that to signal the ohci.c and dc2xx.c drivers that the "End of File"
had been reached.

My final question is:  Could this status be the same status as

#define USB_ST_DATAUNDERRUN   -9

or What does "data under run" mean?

It is obvious to me that my card has some unique status codes and I may run
into this in the future.  It is for this reason I will not include patches for
any of this unless others are like me with this type of card; trying to get 
bulk messages to work.

Once again Thank you all
Ira Weiny


** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

Re: [linux-usb] OHCI fix for Belkin BusPort card on PowerMac

[Gregory P. Smith]

[-- Attachment #1: Type: text/plain, Size: 1820 bytes --]

On Mon, Dec 06, 1999 at 02:23:12PM -0800, iweiny@pacbell.net wrote:
> 
> I am running:
> PowerMac 8500 (604 processor)
> USB BusPort from Belkin (Model: F5U005-MAC) PCI USB card
> 2.2.12 kernel (self compiled)
> 2.3.29 USB stack from Paul Mackerras's rsync tree
> opendis-0.0.3-pre4 from http://digitalux.netpedia.net
...
> However, when requesting more data then the camera had to give the returned
> status was "9" and not "0".  The OHCI driver interpreted this as an error and
> subsequently the user level program would think no data had been returned.
> 
> After verifing the data to be correct I applied a very simple correction of:
> 
> /* Thinking this is what my controler returns when it is done? */
> #define USB_ST_EOF           9
> 
> and used that to signal the ohci.c and dc2xx.c drivers that the "End of File"
> had been reached.
> 
> My final question is:  Could this status be the same status as
> 
> #define USB_ST_DATAUNDERRUN   -9
> 
> or What does "data under run" mean?

Data underrun means the controller received less data that it asked
for in a transfer.  This is not really an error in many situations;
try modifying the dc2xx driver to accept it.

> It is obvious to me that my card has some unique status codes and I may run
> into this in the future.  It is for this reason I will not include patches for
> any of this unless others are like me with this type of card; trying to get 
> bulk messages to work.

It's not your card, its all controllers and a general "all slightly
different" state of the HCDs at the moment.  We're working on fixing
this with a move to a new cleaner device <-> usb stack/hcd interface.

-- 
Gregory P. Smith   pgp: http://suitenine.com/greg/pgp/
                   C379 1F92 3703 52C9 87C4  BE58 6CDA DB87 105D 9163

[-- Attachment #2: Type: application/pgp-signature, Size: 240 bytes --]

Re: [linux-usb] OHCI fix for Belkin BusPort card on PowerMac

[David Brownell]

"Gregory P. Smith" wrote:
> 
> On Mon, Dec 06, 1999 at 02:23:12PM -0800, iweiny@pacbell.net wrote:
> >
> > I am running:
> > PowerMac 8500 (604 processor)
> > USB BusPort from Belkin (Model: F5U005-MAC) PCI USB card
> > 2.2.12 kernel (self compiled)
> > 2.3.29 USB stack from Paul Mackerras's rsync tree
> > opendis-0.0.3-pre4 from http://digitalux.netpedia.net

(and a DC-265 camera)

> ...
> > However, when requesting more data then the camera had to give the returned
> > status was "9" and not "0".  The OHCI driver interpreted this as an error and
> > subsequently the user level program would think no data had been returned.
> ...
> 
> Data underrun means the controller received less data that it asked
> for in a transfer.  This is not really an error in many situations;
> try modifying the dc2xx driver to accept it.

I'd not be sure it should be OK in this situation, though.  Or that
the "opendis-0.0.3-pre4" release is really free of dependencies on
serial I/O models, where such cases are the norm.  (By the way, the
page above lists 0.0.2 -- dated August, with no USB support -- as the
most current version!)


> It's not your card, its all controllers and a general "all slightly
> different" state of the HCDs at the moment.  We're working on fixing
> this with a move to a new cleaner device <-> usb stack/hcd interface.

That'll be very good to have.


As a quick sanity check, though, is USB working nicely on that
PowerMac for anyone else?  I've not heard many success reports,
pro or con, on Mac USB support.

- Dave

** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/