It's possible for the driver to send a CMF_SYNC_WQE to nonresponsive
firmware during reset of the adapter. The phba link_state conditional
check is currently a strict == LPFC_LINK_DOWN, which does not cover
initialization states before reaching the LPFC_LINK_UP state.
Update the phba->link_state conditional to < LPFC_LINK_UP so that all
initialization states are covered before allowing sending CMF_SYNC_WQE.
Update taking of the hbalock to be during this link_state check as well.
Signed-off-by: Justin Tee <justin.tee@broadcom.com>
Link: https://lore.kernel.org/r/20240912232447.45607-3-justintee8345@gmail.com
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
atot = atomic_xchg(&phba->cgn_sync_alarm_cnt, 0);
wtot = atomic_xchg(&phba->cgn_sync_warn_cnt, 0);
+ spin_lock_irqsave(&phba->hbalock, iflags);
+
/* ONLY Managed mode will send the CMF_SYNC_WQE to the HBA */
if (phba->cmf_active_mode != LPFC_CFG_MANAGED ||
- phba->link_state == LPFC_LINK_DOWN)
- return 0;
+ phba->link_state < LPFC_LINK_UP) {
+ ret_val = 0;
+ goto out_unlock;
+ }
- spin_lock_irqsave(&phba->hbalock, iflags);
sync_buf = __lpfc_sli_get_iocbq(phba);
if (!sync_buf) {
lpfc_printf_log(phba, KERN_ERR, LOG_CGN_MGMT,