Can you test it again with this patch on top?

diff --git a/drivers/scsi/BusLogic.c b/drivers/scsi/BusLogic.c
index d057abfcdd5c..9e67f2ee25ee 100644
--- a/drivers/scsi/BusLogic.c
+++ b/drivers/scsi/BusLogic.c

@@ -2554,8 +2554,14 @@ static void blogic_scan_inbox(struct

blogic_adapter *adapter)
         enum blogic_cmplt_code comp_code;

         while ((comp_code = next_inbox->comp_code) != BLOGIC_INBOX_FREE) {
-               struct blogic_ccb *ccb = blogic_inbox_to_ccb(adapter,
adapter->next_inbox);
-               if (comp_code != BLOGIC_CMD_NOTFOUND) {
+               struct blogic_ccb *ccb = blogic_inbox_to_ccb(adapter,
next_inbox);
+               if (!ccb) {
+                       /*
+                        * This should never happen, unless the CCB list is
+                        * corrupted in memory.
+                        */
+                       blogic_warn("Could not find CCB for dma
address 0x%x\n", adapter, next_inbox->ccb);
+               } else if (comp_code != BLOGIC_CMD_NOTFOUND) {
                         if (ccb->status == BLOGIC_CCB_ACTIVE ||
                                         ccb->status == BLOGIC_CCB_RESET) {