fm10k: fix SM mailbox full condition

Current condition will always incorrectly report a full SM mailbox if an
IES API application is not running. Due to this, the
"fm10k_service_task" will be infinitely queued into the driver's
workqueue. This, in turn, will cause a "kworker" thread to report 100%
CPU utilization and might cause "soft lockup" events or system crashes.

To fix this issue, a new condition is added to determine if the SM
mailbox is in the correct state of FM10K_STATE_OPEN before proceeding.
In other words, an instance of the IES API must be running. If there is,
the remainder of the flow stays the same which is to determine if the SM
mailbox capacity has been exceeded or not and take appropriate action.

Signed-off-by: Ngai-Mint Kwan <ngai-mint.kwan@intel.com>
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

diff --git a/drivers/net/ethernet/intel/fm10k/fm10k_iov.c b/drivers/net/ethernet/intel/fm10k/fm10k_iov.c
index e707d717012faa997a127687ce45d54b27b9e3eb..74160c2095ee4f9ccc02745403e31106beed4cde 100644 (file)
--- a/drivers/net/ethernet/intel/fm10k/fm10k_iov.c
+++ b/drivers/net/ethernet/intel/fm10k/fm10k_iov.c
@@ -244,7 +244,8 @@ process_mbx:
                }
 
                /* guarantee we have free space in the SM mailbox */
-               if (!hw->mbx.ops.tx_ready(&hw->mbx, FM10K_VFMBX_MSG_MTU)) {
+               if (hw->mbx.state == FM10K_STATE_OPEN &&
+                   !hw->mbx.ops.tx_ready(&hw->mbx, FM10K_VFMBX_MSG_MTU)) {
                        /* keep track of how many times this occurs */
                        interface->hw_sm_mbx_full++;