[Bug 1584] New: 2.6.0-test10 hangs in aix79xxx after iocell_workaround message

[Bug 1584] New: 2.6.0-test10 hangs in aix79xxx after iocell_workaround message

[Martin J. Bligh]

http://bugme.osdl.org/show_bug.cgi?id=1584

           Summary: 2.6.0-test10 hangs in aix79xxx after iocell_workaround
                    message
    Kernel Version: 2.6.0-test10 SMP
            Status: NEW
          Severity: blocking
             Owner: andmike@us.ibm.com
         Submitter: berkley@cs.wustl.edu


Distribution:
Hardware Environment: P4-Xeon-2.4GHz 2GB Adaptec 39320/AIC7902
Software Environment: Redhat 9.0
Problem Description:
Booting 2.6.0-test10 fails in AIC79XX (spinlock problem?) after executing
iocell_workaround in aic79xx_core.c (last message with ahd_debug set to 15)
since there were no changes to aic* since -test9 (which works fine) I'm assuming
the problem is in scsi.c which has many changes, most involving adding spinlocks.
The aic79xx module is loaded in the kernel (=y, not as a module =m)
at compile time.

Steps to reproduce:

boot -test10 kernel with either a 39320 U320 scsi controller or the motherboard
equivalent the AIC7902. With no debug on, the kernel hangs hard (no
<ctrl-alt-del>) just after showing the IDE partitions.
Setting the module parameter ahd_debug=15 to enable lots of messages shows
printf("%s: Setting up iocell workaround\n", ahd_name(ahd)); at 5383 in
aic79xx.c as the last known entry in syslog.