OSDN Git Service

scsi: lpfc: Fix nvme remoteport registration race conditions
authorJames Smart <jsmart2021@gmail.com>
Mon, 9 Apr 2018 21:24:30 +0000 (14:24 -0700)
committerMartin K. Petersen <martin.petersen@oracle.com>
Wed, 18 Apr 2018 23:34:05 +0000 (19:34 -0400)
commitb15bd3e6212e747ebd1b37a542898e88ad05bb17
tree430386525c7068cd5408cdf62ac67d2aa6ade0c4
parentb04744ce52a8da883c8b87b66082f9805bb4ca32
scsi: lpfc: Fix nvme remoteport registration race conditions

On tests adding and removing a remote port, calls to nvme_info would
eventually show fewer target ports discovered than were present in the
san. Additionally, the following error messages were seen:

  6031 RemotePort Registration failed err: -116, DID x471301

There is a race condition that exists between the driver and the nvme
transport on remote port unregister vs the confirmed deletion. It's
possible that the driver may rediscover the remote port and reregister
the remote port before a prior unregister delete callback was made (as
it rebinded to the prior remoteport structure). However, the driver was
coded to expect the callback before seeing the remote port again thus a
new registration. The logic results in the driver having an invalid
remoteport pointer set.

Correct by tracking when waiting for the delete callback. In cases where
the ndlp remoteport pointer is updated, it is only cleared when the wait
has not been superceded by a prior registration.

Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: James Smart <james.smart@broadcom.com>
Reviewed-by: Hannes Reinecke <hare@suse.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
drivers/scsi/lpfc/lpfc_nvme.c