> On Tue, Oct 17, 2000 at 10:06:32AM -0400, Joseph Sarkes wrote:
> > I had already looked at that, com0 16550a working fifo

> The IDE driver doesn't block serial interrupts (IPL_SERIAL is abobe
> IPL_HIGTH on i386). But the hardware may: the super I/O chip may be
> at lower priority than IDE DMA, so serial PIO transfers may be blocked
> by a IDE DMA operation.

I'm not sure whether it is a priority thing here, or some ide drive
interaction. I originally had an ide drive as master and a cdrom
drive as slave on the primary ide controller, and two ide drives on
the secondary controller, which were the active drives, although
heavy use of any drive seemed to cause the silo overflows. 

I split the two used drives off to the primary port of the
promise udma/100 ide controller on the motherboard, and put the
first ide drive as master on the primary ide controller, and the
cd rom as slave on the secondary ide controller. Now I can't seem
to get any more silo overflows. Hopefully it will stay that way.

> 
> --
> Manuel Bouyer <bouyer@antioche.eu.org>
> --
> 
> 


-- 
Joseph Sarkes		jsarkes@tiac.net