
In a memory channel cluster, rebooting a node without performing a hardware
reset an crash other members with a RM_AUDIT_ACK_BLOCK panic.

   0 boot
   1 panic
   2 rm_crash_node_mask
   3 rm_panic
   4 rm_audit_ack_block
   5 rm_write_sync
   6 rm_get_errcnt_lock
   7 rm_lock_global_error
   8 rm_eh_init_shared_data_req
   9 rm_prail_int
  10 rm_int
  11 Mchan_isr
  12 _XentInt


This fixes issues associated with the initialization of the Memory Channel

A debug kernel [built non-optimized] can panic with a kernel memory fault:

   0 panic
   1 trap
   2 _XentMM
   3 rm_notif_request
   4 rm_lrail_int_ctx
   5 rm_int
   6 Mchan_isr
   7 intr_dispatch_no_post
   8 _XentInt


An error in event log indexing may result in the appearance of superfluous
"rm_event, index too big" messages on the system console.

This patch fixes an initializtion problem in the Memory Channel driver.  A
kernel memory fault during boot is the symptom associated with this issue.

 This fixes a problem in the rm driver. Symptoms:
'panic: "simple_lock time limit exceeded" in rm_get_mgmt_spinlock()'
 



