Hello, I wrote:

>    Hm, I guess I'm seeing a real issue

    Nevermind, it's just ide_read_sff_dma_status() used for hwif->tp_ops-> 
read_sff_dma_status() method in the driver where it clearly doesn't make 
sense.  But it seems it never gets called in these cases anyway (only adds a 
confusion :-).

> with this patch...

    Not with this patch actually, with the one that introduced 'struct tp_ops'.

>> Thanks,
>> Bart

MBR, Sergei