Add the attempted DMA address to the 'DMA beyond end of ISA' message so that

we can see if it's a small distance beyond the end, or way out. This may
give some clues as to whether it is being caused by something coalescing
the transfers in spite of the bounce buffers, or simply because of buffer
corruption.  (The BT driver seems to occasionally get hit by from this too,
except that it does not trap the transfer, and the system panics later
with vm_bounce_page_free.)  This "event" usually happens to me during a
savecore (on the rare occasion that a kernel coredump is actually taken
after a crash - the lack of kernel core dumps is another problem...).

sys/i386/isa/aha1542.c

@@ -12,7 +12,7 @@
  * on the understanding that TFS is not responsible for the correct
  * functioning of this software in any circumstances.
  *
- *      $Id: aha1542.c,v 1.52 1995/12/07 12:45:53 davidg Exp $
+ *      $Id: aha1542.c,v 1.53 1995/12/15 00:11:26 bde Exp $
  */
 
 /*
@@ -1565,7 +1565,8 @@ aha_scsi_cmd(xs)
 					if (thisphys > 0xFFFFFF)
 					{
 						printf("aha%d: DMA beyond"
-							" end Of ISA\n", unit);
+							" end Of ISA: 0x%x\n",
+							unit, thisphys);
 						xs->error = XS_DRIVER_STUFFUP;
 						aha_free_ccb(unit, ccb, flags);
 						return (HAD_ERROR);