piixide: lost interrupts (Intel Atom Mini-ITX)

[Teemu Rinta-aho <teemu%rinta-aho.org@localhost>]

Hi all,

I have been struggling a while after finding out that piixide
sometimes throws "lost interrupt... retrying... soft error (corrected)"
while reading or writing to a hard drive connected to the SATA port
of an Intel Atom Mini-ITX motherboard. This seems to happen only
when there is a big load on the disks, e.g. when building a NetBSD release.
I have two 1 TB disks configured to RAID-1 with raidframe. I have
been using a similar setup for years but the problems started when
I updated the motherboard from VIA to Intel.

I am currently in the investigation phase. I thought replacing
the disks would help, but they were not the source of the problems.
I am now running a build with a current kernel instead of 5.0.
If the problems continue I will plug the disks into another machine.
But that really doesn't solve the problem with this Intel Atom
platform.

So, if you have any insights or nice ideas to try out to maybe fix
a possible software bug in NetBSD, or a way to prove that it is
a hardware problem, please let me know.

And, if someone happens to know why is this:

piixide0 at pci0 dev 31 function 2: Intel 82801GB/GR Serial ATA/Raid 
Controller (ICH7) (rev. 0x01)
piixide0: bus-master DMA support present
piixide0: primary channel configured to compatibility mode
piixide0: primary channel interrupting at ioapic0 pin 14

even when I have configured "native" mode in the BIOS?

And, where are the SATA 150/300 speeds? Not that I need those but anyway,
there's nothing that looks like SATA here:

wd0: drive supports PIO mode 4, DMA mode 2, Ultra-DMA mode 7
wd0(piixide0:0:0): using PIO mode 4, Ultra-DMA mode 6 (Ultra/133) (using DMA)

BR,
Teemu Rinta-aho