On Fri, May 2, 2008 at 10:45 AM, Mark Lord <liml@xxxxxx> wrote: > Grant Grundler wrote: ... > > I reccomend setting per_loop to 10 or 20 at least. 5us is not very long. > > It takes about 1us to do an MMIO read. Each additional MMIO read will > exacerbate > > the problem of DMA not finishing since MMIO reads interfere with DMA > streams. > > > > Also can you add a comment why 15ms is right amount of time to wait? > > > .. > > I would, if I knew. That's just a random number. > My own timings with a handful of drives here indicated > that it takes a few hundred microseconds worst case, > so I just picked a bigger number to start with. > > No rationale == no comment in the original. Ok. Often there is a rationale and the original author just didn't include it. > But if you have a good theory for what that number ought to be, > then go for it -- just patch it and stick an explanation in. I'd need to study the EDMA engine docs more carefully than I have to date. > > BTW, I have no idea when EDMA stops on it's own or if we could > > safely get EDMA to stop "prematurely" (e.g. disable PCI Bus Master). > > > .. > > I don't know what you're talking about there. I don't either. :) Just trying to explain I was making up the example. thanks, grant > > ??? > -- To unsubscribe from this list: send the line "unsubscribe linux-ide" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html