[dpdk-dev] [PATCH v2 06/30] i40e/base: check for stopped admin queue

Helin Zhang helin.zhang at intel.com
Thu Feb 18 07:29:56 CET 2016


It's possible that while waiting for the spinlock, another
entity (that owns the spinlock) has shut down the admin queue.
If it then attempts to use the queue, it will panic.
It adds a check for this condition on the receive side. This
matches an existing check on the send queue side.

Signed-off-by: Helin Zhang <helin.zhang at intel.com>
---
 drivers/net/i40e/base/i40e_adminq.c | 8 ++++++++
 1 file changed, 8 insertions(+)

diff --git a/drivers/net/i40e/base/i40e_adminq.c b/drivers/net/i40e/base/i40e_adminq.c
index 998582c..e1a162e 100644
--- a/drivers/net/i40e/base/i40e_adminq.c
+++ b/drivers/net/i40e/base/i40e_adminq.c
@@ -1035,6 +1035,13 @@ enum i40e_status_code i40e_clean_arq_element(struct i40e_hw *hw,
 	/* take the lock before we start messing with the ring */
 	i40e_acquire_spinlock(&hw->aq.arq_spinlock);
 
+	if (hw->aq.arq.count == 0) {
+		i40e_debug(hw, I40E_DEBUG_AQ_MESSAGE,
+			   "AQRX: Admin queue not initialized.\n");
+		ret_code = I40E_ERR_QUEUE_EMPTY;
+		goto clean_arq_element_err;
+	}
+
 	/* set next_to_use to head */
 #ifdef PF_DRIVER
 #ifdef INTEGRATED_VF
@@ -1113,6 +1120,7 @@ clean_arq_element_out:
 	/* Set pending if needed, unlock and return */
 	if (pending != NULL)
 		*pending = (ntc > ntu ? hw->aq.arq.count : 0) + (ntu - ntc);
+clean_arq_element_err:
 	i40e_release_spinlock(&hw->aq.arq_spinlock);
 
 #ifdef PF_DRIVER
-- 
2.5.0



More information about the dev mailing list