On 10/3/2011 4:26 PM, Yi Zou wrote:
With commit "77a2b73 [SCSI] libfc: fix fc_eh_host_reset", the exch lock is
already
held when fc_exch_abort_locked(), but fc_seq_send() that is called along the
path is gonna grab the same exch lock again, causing dead-lock. Drop the exch
lock before fc_seq_send() in fc_exch_abort_locked() should fix it.
v1:
This is reported by Bhanu with the following kernel log:
https://lists.open-fcoe.org/pipermail/devel/2011-September/011767.html
v2:
No need to hold exch lock for fc_frame_alloc(), so just drop it before the frame
alloc and grab it before returning from fc_exch_abort_locked()
Patch looks good to me.
Thanks,
Bhanu
Reported-by: Bhanu Prakash Gollapudi<bprak...@broadcom.com>
Signed-off-by: Yi Zou<yi....@intel.com>
---
drivers/scsi/libfc/fc_exch.c | 3 +++
1 files changed, 3 insertions(+), 0 deletions(-)
diff --git a/drivers/scsi/libfc/fc_exch.c b/drivers/scsi/libfc/fc_exch.c
index 7c055fd..8325489 100644
--- a/drivers/scsi/libfc/fc_exch.c
+++ b/drivers/scsi/libfc/fc_exch.c
@@ -617,6 +617,7 @@ static int fc_exch_abort_locked(struct fc_exch *ep,
/*
* Send an abort for the sequence that timed out.
*/
+ spin_unlock_bh(&ep->ex_lock);
fp = fc_frame_alloc(ep->lp, 0);
if (fp) {
fc_fill_fc_hdr(fp, FC_RCTL_BA_ABTS, ep->did, ep->sid,
@@ -624,6 +625,8 @@ static int fc_exch_abort_locked(struct fc_exch *ep,
error = fc_seq_send(ep->lp, sp, fp);
} else
error = -ENOBUFS;
+
+ spin_lock_bh(&ep->ex_lock);
return error;
}
_______________________________________________
devel mailing list
devel@open-fcoe.org
https://lists.open-fcoe.org/mailman/listinfo/devel
_______________________________________________
devel mailing list
devel@open-fcoe.org
https://lists.open-fcoe.org/mailman/listinfo/devel
