SCSI: scsi_remove_target: fix softlockup regression on hot remove

commit bc3f02a795d3b4faa99d37390174be2a75d091bd upstream.

John reports:
 BUG: soft lockup - CPU#2 stuck for 23s! [kworker/u:8:2202]
 [..]
 Call Trace:
  [<ffffffff8141782a>] scsi_remove_target+0xda/0x1f0
  [<ffffffff81421de5>] sas_rphy_remove+0x55/0x60
  [<ffffffff81421e01>] sas_rphy_delete+0x11/0x20
  [<ffffffff81421e35>] sas_port_delete+0x25/0x160
  [<ffffffff814549a3>] mptsas_del_end_device+0x183/0x270

...introduced by commit 3b661a9 "[SCSI] fix hot unplug vs async scan race".

Don't restart lookup of more stargets in the multi-target case, just
arrange to traverse the list once, on the assumption that new targets
are always added at the end.  There is no guarantee that the target will
change state in scsi_target_reap() so we can end up spinning if we
restart.

Acked-by: Jack Wang <jack_wang@usish.com>
LKML-Reference: <CAEhu1-6wq1YsNiscGMwP4ud0Q+MrViRzv=kcWCQSBNc8c68N5Q@mail.gmail.com>
Reported-by: John Drescher <drescherjm@gmail.com>
Tested-by: John Drescher <drescherjm@gmail.com>
Signed-off-by: Dan Williams <djbw@fb.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

diff --git a/drivers/scsi/scsi_sysfs.c b/drivers/scsi/scsi_sysfs.c
index de21547e7c43f4f4382ae33c713faeefd9dc2f6e..51d823f5465a2154b20d3b416e1d698f92f8121f 100644 (file)
--- a/drivers/scsi/scsi_sysfs.c
+++ b/drivers/scsi/scsi_sysfs.c
@@ -988,33 +988,31 @@ static void __scsi_remove_target(struct scsi_target *starget)
 void scsi_remove_target(struct device *dev)
 {
        struct Scsi_Host *shost = dev_to_shost(dev->parent);
-       struct scsi_target *starget, *found;
+       struct scsi_target *starget, *last = NULL;
        unsigned long flags;
 
- restart:
-       found = NULL;
+       /* remove targets being careful to lookup next entry before
+        * deleting the last
+        */
        spin_lock_irqsave(shost->host_lock, flags);
        list_for_each_entry(starget, &shost->__targets, siblings) {
                if (starget->state == STARGET_DEL)
                        continue;
                if (starget->dev.parent == dev || &starget->dev == dev) {
-                       found = starget;
-                       found->reap_ref++;
-                       break;
+                       /* assuming new targets arrive at the end */
+                       starget->reap_ref++;
+                       spin_unlock_irqrestore(shost->host_lock, flags);
+                       if (last)
+                               scsi_target_reap(last);
+                       last = starget;
+                       __scsi_remove_target(starget);
+                       spin_lock_irqsave(shost->host_lock, flags);
                }
        }
        spin_unlock_irqrestore(shost->host_lock, flags);
 
-       if (found) {
-               __scsi_remove_target(found);
-               scsi_target_reap(found);
-               /* in the case where @dev has multiple starget children,
-                * continue removing.
-                *
-                * FIXME: does such a case exist?
-                */
-               goto restart;
-       }
+       if (last)
+               scsi_target_reap(last);
 }
 EXPORT_SYMBOL(scsi_remove_target);