[SCSI] libfc: Fix panic in fc_exch_recv
Bhanu Prakash Gollapudi [Sat, 11 Feb 2012 01:18:51 +0000 (17:18 -0800)]
Adding and removing the host into the zone causes this panic.

BUG: unable to handle kernel NULL pointer dereference at 00000000000000a0
IP: [<ffffffffa0491707>] fc_exch_recv+0xc57/0xe70 [libfc]
Call Trace:
[<ffffffffa050e04b>] bnx2fc_l2_rcv_thread+0x37b/0x430 [bnx2fc]
[<ffffffffa050dcd0>] ? bnx2fc_l2_rcv_thread+0x0/0x430 [bnx2fc]
[<ffffffff81090886>] kthread+0x96/0xa0
[<ffffffff8100c14a>] child_rip+0xa/0x20
[<ffffffff810907f0>] ? kthread+0x0/0xa0
[<ffffffff8100c140>] ? child_rip+0x0/0x20

During fc_exch_reset, the active exchanges are aborted and the exch is deleted.
As part of processing ABTS response, due to 'ep' being NULL, any access to ep in
fc_exch_recv_bls() causes this panic. Fixed to access 'ep' only if non-NULL.

Reviewed-by: Neerav Parikh <neerav.parikh@intel.com>
Signed-off-by: Bhanu Prakash Gollapudi <bprakash@broadcom.com>
Signed-off-by: Robert Love <robert.w.love@intel.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

drivers/scsi/libfc/fc_exch.c

index 4d70d96..630291f 100644 (file)
@@ -1642,9 +1642,10 @@ static void fc_exch_recv_bls(struct fc_exch_mgr *mp, struct fc_frame *fp)
                case FC_RCTL_ACK_0:
                        break;
                default:
-                       FC_EXCH_DBG(ep, "BLS rctl %x - %s received",
-                                   fh->fh_r_ctl,
-                                   fc_exch_rctl_name(fh->fh_r_ctl));
+                       if (ep)
+                               FC_EXCH_DBG(ep, "BLS rctl %x - %s received",
+                                           fh->fh_r_ctl,
+                                           fc_exch_rctl_name(fh->fh_r_ctl));
                        break;
                }
                fc_frame_free(fp);