Re: [PATCH 0/4] g_NCR5380: PDMA fixes and cleanup

[Finn Thain <fthain@telegraphics.com.au>]

On Fri, 23 Jun 2017, Ondrej Zary wrote:

> On Friday 23 June 2017 13:01:53 Finn Thain wrote:
> > On Fri, 23 Jun 2017, I wrote:
> > > Does this patch help? It should be applied on top of this series of 4.
> >
> > Sorry, I sent the wrong diff. Please try this patch instead.
> 
> Thanks, much better now: both HDD and CD-ROM seem to work on DTC and non-DTC 
> chips. I get many of these messages with CD-ROM:
> [  912.397076] generic_NCR5380_pread: No end dma signal (4096/4096)
> [  913.141225] generic_NCR5380_pread: No end dma signal (4096/4096)
> 
> Maybe just remove this error message as in my original patch?
> 

The "data loop" algorithm in the datasheet syas that if End of DMA doesn't 
become asserted after a transfer, this is an error condition.

Your log (above) shows 4096/4096 bytes so I think that End of DMA is 
tested too soon: the datasheet says that End of DMA takes 100 ns to become 
asserted even after all of the relevant signals reach their final state.

I'll re-do the series to account for this.

Thanks for testing.

--