ATA and APIC IRQ conflict
pisymbol at gmail.com
Mon May 5 21:37:00 UTC 2008
On Mon, May 5, 2008 at 2:52 PM, David Naylor <naylor.b.david at gmail.com> wrote:
> On Sunday 04 May 2008 01:17:16 you wrote:
> > On Sat, May 3, 2008 at 9:25 AM, David Naylor <naylor.b.david at gmail.com>
> > > I am trying to install FreeBSD on a gigabyte board (based on nForce 650i
> > > chipset (with MCP51 controller)) however no SATA hard drives are
> > > detected. However, if APIC is disabled then I have access to the hard
> > > drives (and only one CPU). It has been suggested that it is an IRQ
> > > conflict (or something else related to APIC).
> > I'll bet its interrupt routing more than IRQ conflicts (which these
> > days on a modern system really doesn't make sense). More than likely
> > there is some funkiness going on with the BIOS's ACPI tables (I would
> > start with the MADT and verify its entries look kosher).
> Considering that the system works with APIC disabled I think you are right. I
> am learning on the fly here, sorry.
You said the system works but later on you say you can't see second
My guess is the system still doesn't work right! :D!
> There does not appear to be any BIOS options except for the suspend state,
> currently set to S1? (And a few power-on options, all disabled)
Typically when you disable APIC your in classical 8259 mode which
should work. I've seen power management screw up a box royally on a
soft reset too but again, not sure that is your problem just yet.
> > Though it could be a power management problem and have nothing to do
> > with interrupts (though with the SATA hard drives not being detected I
> > would guess its interrupts).
> Both linux and windows vista run on this system without a problem... Except
> when I reboot from FreeBSD then windows BSOD's and linux keeps resetting ata
> (without ever actually booting). After having run FreeBSD I need to do a
> system power down (not just a reboot) then the problem disappears.
Seems like after a soft reset, ACPI firmware is left in an
inconsistent state (that might be because whatever version of BSD you
are running to bring it up completely properly).
> Also even with APIC disabled FreeBSD fails to detect my second CD ROM drive
> (both are IDE) and it fails to read the first one.
Still not good, sounds like an interrupt routing issue again. I'm
assuming the drivers load but nothing is detected (that would rule out
any PCI bus hierarchy nastiness).
> I doubt it is a BIOS specific problem since this is the second board (first
> one was from Asus) that has this problem. Just to be safe I flashed the BIOS
> to the latest one, no luck :-(
I'm kinda losing it. What version of FreeBSD are you using 7.0
RELEASE or top of tree (built from -CURRENT)? I should have asked
this to begin with (and you should post it in future postings). My
guess is you are trying a STABLE release (6.3 or 7.0) etc.
If you can dump the AML code (even from Linux) that would be very very
helpful. Here is a utility that works from DOS:
dmesg during an ACPI boot also is needed (again turn on debugging).
Most of the time ACPI problems (from my limited experience) is really
due to "interesting" ACPI table code that can throw off the OS.
Interrupt routing issues tend to be MADT related but I have no idea
what is your specific problem so far.
More information about the freebsd-questions