Description of problem:

I installed two 64bit adaptec 29160 scsi hbas into an intel tiger4
system. Each hba is attached to an external disk array, containing
14 disks, 7 on each controller (split backplain).
                                                                     
                                            
When I boot the system with the disk tray turned on (disks spinning),
the kernel blows stack somewhere in acpi and reboots after about 30
seconds.
                                                                     
                                            
If the disk tray is powered off, the kernel boots fine I turn the tray
one, load the aic7xxx driver, and all is well the option rom scanning
for the adaptec cards is turned off in the ia64 bios so the adaptec
bios does not even get a chance to load and potentiall install a wacky
interrupt handler.
                                                                     
                                            
<theory>
The kernel blowing stack happens even with an initrd that does not
contain the aic7xxx driver so I am thinking it has something to do
with the intel acpi finding the hbas, making some data structures
for them (with info about the disks attached most likely)
and then when our acpi parser trys to parse the modified acpi tables,
we recurse and the kernel stack goes boom.
</theory>
                                                                     
                                            
Keep in mind that the aic7xxx driver is not even in the picture here,
and that we blow stack so early in the boot process that we are still
parsing the acpi tables.


Version-Release number of selected component (if applicable):

Kernel version 2.4.21-20.EL

How reproducible:

Every time

Steps to Reproduce:
1. power system and disk tray off
2. power disk tray on
3. power system on
4. wait for kernel to go boom during the acpi parsing part of kernel
initialization