On Fri, 6 Mar 2009, Jon Smirl wrote:
> It can't find the end point. It is asking for EP 81 on interface 0.
You mean 0x81.
> if ((uurb->endpoint & USB_ENDPOINT_DIR_MASK) != 0) {
> is_in = 1;
> ep = ps->dev->ep_in[uurb->endpoint & USB_ENDPOINT_NUMBER_MASK];
> } else {
> is_in = 0;
> ep = ps->dev->ep_out[uurb->endpoint & USB_ENDPOINT_NUMBER_MASK];
> }
> if (!ep)
> return -ENOENT;
>
> Should it be there?
Yes, indeed it should. In your original message you said you were
using 2.6.28. It's entirely possible that a bug in this area has been
fixed in the .stable series -- have you tried 2.6.28.7?
> All of this works ok if ftdi_sio is unloaded. Does having ftdi_sio
> bound effect these arrays?
Not directly, but there could be an indirect effect.
> The device only has one interface.
You mean only one configuration.
> Should the fact the the device was
> already in config #1 been detected before generating the error?
Before generating which error? It _was_ detected before the failure of
the USBDEVFS_SETCONFIGURATION ioctl:
> usb 7-2: usbfs: interface 1 claimed by ftdi_sio while 'jtag' sets
> config #1
Look at proc_setconfig() in devio.c to see what happens when libftdi
does that ioctl.
> An attempt to set config #1 when config #1 was already set should be a
> nop, right?
No, not really. If a driver is already bound to an interface when your
program does the Set-Config then the result should be a simple error
with no side effects. So with ftdi_sio bound to interface 1, the
Set-Config should have done nothing. (You might want to try adding
code at the beginning and end of proc_setconfig() to check whether
ps->dev->ep_in[1] is set. You might also want to print out its value
just before the check in proc_do_submiturb(), just to be extra
certain.)
But... If none of the interfaces are already bound then the Set-Config
request is sent to the device, resetting it, and the kernel resets some
of its internal state (including those arrays of endpoint pointers).
Alan Stern
--
To unsubscribe from this list: send the line "unsubscribe linux-usb" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
