Yes, that's pretty much how I had planned to do it.  I'll try and do it
very soon.  (I added stuff to execute snippets of code in the usb
thread just for this purpose.)

    -- Lennart


Jason R Thorpe wrote:

> On Thu, Apr 12, 2001 at 03:45:41AM +0200, Lennart Augustsson wrote:
>
>  > Well, the reason I didn't make the switch is that it doesn't really
>  > work with the aue driver.  It needs to be converted not to use the
>  > ugly hack that allows USB transfers in interrupt context.
>  > Which I've not done yet, it's a little messy because of the mii callbacks
>  > that happen in interrupt context.
>
> It should be easy to deal with this, actually.
>
> The only interrupt-driven MII accesses that happen is the mii_tick()
> stuff that is controlled completely by the driver.  This is so e.g. you
> could call mii_tick() from an MII-change interrupt, if your hardware
> has that -- or you could use the periodic-task-in-thread-context stuff
> that the USB code has to do the one-second MII tick operation.  Then all
> the stuff would happen from thread context, and there would be no trouble.
>
> --
>         -- Jason R. Thorpe <thorpej@zembu.com>