Re: New Amiga drivers coming for Linux 4.18
Hi Finn,
On Mon, Apr 23, 2018 at 11:08 AM, Finn Thain <fthain@telegraphics.com.au> wrote:
> On Sun, 22 Apr 2018, Michael Schmitz wrote:
>
>>
>> a bus error is encountered once the processor attempts to transfer
>> additional bytes after the ESP has ended DMA mode.
>>
>
> No, these algorithms continue to transfer after a bus error.
You are correct - if DMA mode hasn't ended and the bus error was
caused by the glue logic because data is not yet available, transfer
resumes after a delay. If DMA mode has ended prematurely, transfer
will eventually time out.
The task queue does not appear to anticipate premature end of DMA so
this code would need modifications to cope with that case.
>
>> Am I missing something else here?
>>
>
> Yeah, a circuit schematic.
>
> Unfortunately I don't have one either, so we have to try to draw
> inferences.
>
> I don't think IRQs are relevant to this discussion but I'm just guessing.
The end of DMA interrupt could be used to signal the task queue
transfer loop to exit with residual,but the old driver didn't make use
of such a feature. Clearly needs a bit more thought should the case
arise.
But a schematic or more details about the hardware would be essential.
Cheers,
Michael
Reply to: