Re: [PATCH] scsi: lpfc: Fix improper handling of refcount in lpfc_bsg_hba_get_event()

[Justin Tee]

Hi Qiu-ji,

Similar to the other suggested patch, this does not look logically
correct.  if (evt_dat == NULL) evaluates to true, then that means the
list_for_each_entry_safe(evt, evt_next, &phba->ct_ev_waiters, node)
loop did not find an evt lpfc_bsg_event object of interest or that the
phba->ct_ev_waiters list is empty.

Why would this patch want to call lpfc_bsg_event_unref on an evt
object that is not of specified interest indicated by the bsg
event_req object?

Even worse, as mentioned in the other email, this patch could kref_put
on the phba->ct_ev_waiters head which is not a preallocated
lpfc_bsg_event object leading to references on an uninitialized memory
region.

Sorry, but I cannot acknowledge this patch as well.

Regards,
Justin