Supermicro X7DBR-8+ hang at boot

[John Baldwin]

On Wednesday 24 October 2007 09:49:06 am Guy Helmer wrote:
> John Baldwin wrote:
> > On Tuesday 23 January 2007 01:17:57 pm Guy Helmer wrote:
> >   
> >> Jack Vogel wrote:
> >>     
> >>> On 1/23/07, Guy Helmer <ghelmer at palisadesys.com> wrote:
> >>>       
> >>>> Using FreeBSD 6.2, I'm having trouble with the Supermicro X7DBR-8+
> >>>> motherboard (dual Xeon 5130 CPUs on the Blackford chipset -
> >>>> 
http://www.supermicro.com/products/motherboard/Xeon1333/5000P/X7DBR-8+.cfm) 
> >>>>
> >>>> hanging after printing the "Waiting 5 seconds for SCSI devices to
> >>>> settle" message.  The hang doesn't always happen - sometimes we have to
> >>>> go through several reboot cycles for it to happen - but sometimes it
> >>>> happens with every reboot.  For those who would suggest that this
> >>>> happens because I'm using Seagate drives, it happens even if we totally
> >>>> remove the SCSI drive (but leave the aic7902 SCSI interfaces enabled)
> >>>> and boot from a SATA disk.  Using FreeBSD 6.1, the Intel gigabit
> >>>> ethernet NICs aren't found but the hang doesn't occur.
> >>>>         
> >>> ...
> >>> If that isnt it, I would suggest installing using ACPI disabled or 
> >>> SAFE if
> >>> needed, and then tweak the kernel after.
> >>>       
> >> hint.apic.0.disabled=1 helped - it hasn't hung yet in several boot 
> >> cycles.  New dmesg is attached below in case it helps anyone see a 
> >> better fix than disabling the APICs.
> >>     
> >
> > So you got an interrupt storm on IRQ 18 when ahd0 tried to probe and ahd0 
got
> > interrupt timeouts.  This indicates that ahd0 really lives on IRQ 18, not 
IRQ
> > 30.  Your BIOS is likely busted since ACPI hardcodes these sort of IRQs.
> >
> > You can override the BIOS by doing:
> >
> > set hw.pci5.2.INTA.irq=18
> >
> > in the loader (or adding a line to loader.conf) and seeing if that fixes 
the
> > boot with APIC enabled.
> >
> >   
> I'm trying to resolve what looks like a similar problem with an IBM 
> Blade Server unit.  I'm reviewing my previous emails on this subject 
> with the verbose boot messages to try to learn what lead you to 
> determine the correct interrupt would be 18, but I can't seem to figure 
> out what data leads to this conclusion.  Any hints?

He got an interrupt storm on IRQ 18 while the ahd0 device on IRQ 30 was timing 
out.

-- 
John Baldwin