Re: [MiNT] EHCI host controller driver

[David Gálvez <dgalvez75@gmail.com>]

2014-04-30 18:22 GMT+02:00 Vincent Rivière <vincent.riviere@freesbee.fr>:
> On 30/04/2014 16:25, David Gálvez wrote:
>>
>> Yes support for USB 2.0 host controllers that follows the EHCI
>> standard. Until now there were only support for the isp116x (used by
>> EtherNat and NetUSBee) and sl811 (used by the Unicorn) host
>> controllers.
>
>
> Thanks for that information.
> Maybe your new code will support the FireBee soon?
> Or maybe ARAnyM and libusb?
>
>

This driver in theory supports any ATARI with PCI-BIOS including the
FireBee, but for the FireBee it doesn't add anything new that has been
already offered by FireTOS own drivers, (well not true anymore, now
MiNT has the USBtoSerial driver). Besides FireTOS XHDI implementation
doesn't support XHNewCookie( ) function so MiNT's own mass storage
driver can't be installed :-/

For ARanyM there is a special separate driver, which abstract all this
low level stuff, it supports what libusb supports, so Aranym+MiNT have
USB 2.0 since time ago.

>> When you said that 68000 is affected by this, it's compiling for 68000
>> CPU or the entire family (68020 to 68060 also).
>
>
> __attribute__ ((packed)) causes bad code for the old 68000 only.
> Since newer processors can access words and longs on odd addresses, gcc
> produces optimal move.w and move.l code as usual, in any case.
>

Ok, thanks for the explanation. Although I don't expect 68000 machines
with PCI, I'll remove the __attribute__ ((packed)) just in case we
compile the drivers for 68000.