> Our -current based on a sup from december 28 gives us a lot of "FIFO 
> empty" and write errors to the disk. This is similar to what we were 
> experiencing on our ELC:s earlier on.

I just found out that the esp/dma code doesn't expect the possibilty
that the device might like to chop the transfer in pieces, sending
SAVE POINTER messages in between multiple DATA OUT phases.

It just spits out the "FIFO empty" msg, no data is actually lost.

-pk