USB test code problem with stm32 eval board

USB test code problem with stm32 eval board

[GaurangT]

Hi,
        I want to test USB support Added for STM32 in eCos. For that i have
added stm32-usb-20090519.epk pacakge and got "ST STM3210E USB EVAL board" in
templates and able to add STM32 USB driver and it compiles fine. 

Now i am testing to test code of give in
package/io/usb/slave/tests/usbtraget.c.it compiles fine but when i run this
code following output on hyper terminal.

STM32 USB : Setting USB device address = 1 

and windows to ask install red hat ecos test software or driver install.

plz how to test stm32 usb driver?

plz help any one?

Thanks,


-- 
View this message in context: http://www.nabble.com/USB-test-code--problem-with-stm32-eval-board-tp23810900p23810900.html
Sent from the Sourceware - ecos-devel mailing list archive at Nabble.com.

Re: USB test code problem with stm32 eval board

[Chris Holgate]

GaurangT wrote:
> Hi,
>         I want to test USB support Added for STM32 in eCos. For that i have
> added stm32-usb-20090519.epk pacakge and got "ST STM3210E USB EVAL board" in
> templates and able to add STM32 USB driver and it compiles fine. 
> 
> Now i am testing to test code of give in
> package/io/usb/slave/tests/usbtraget.c.it compiles fine but when i run this
> code following output on hyper terminal.
> 
> STM32 USB : Setting USB device address = 1 
> 
> and windows to ask install red hat ecos test software or driver install.
> 
> plz how to test stm32 usb driver?

It's in the manual...

http://ecos.sourceware.org/docs-latest/ref/usbs-testing.html

Unfortunately, the one testcase which is provided by default
(bulk-boundaries) behaves a bit oddly on my system - which I'm inclined
to put down to changes in the way Linux usbfs behaves since the original
was written.

When running the bulk-boundaries test, incoming transfers are treated as
_not_ having zero length packet termination and outgoing transfers _do_
insert zero length packet termination.  This behaviour really should be
consistent between in and out transfers.

I've not done much more investigation myself, since I've been using my
proprietary class driver for testing - but I'm sure that a bit of
tweaking to the TCL should get it to behave.  If this were a low level
Linux driver, this would be achievable by:

1) Forcing zero length packet termination on IN endpoints by issuing an
URB with an oversized buffer so that all incoming packets are treated as
short.  Or...

2) Disabling zero length packet termination on OUT endpoints.

Chris.