From: Chengwen Feng <fengcheng...@huawei.com>
The opcode of the link status notification message reported by the
firmware is zero, it will be filtered out because driver treats it as
already processed message. As a result, the PF can't update the link
status in a timely manner.
Because only VF can set opcode to zero when process mailbox message,
so we add a judgment to make sure the PF messages will not filter out.
Fixes: dbbbad23e380 ("net/hns3: fix VF handling LSC event in secondary process")
Cc: sta...@dpdk.org
Signed-off-by: Chengwen Feng <fengcheng...@huawei.com>
Signed-off-by: Min Hu (Connor) <humi...@huawei.com>
---
drivers/net/hns3/hns3_mbx.c | 10 +++++++---
1 file changed, 7 insertions(+), 3 deletions(-)
diff --git a/drivers/net/hns3/hns3_mbx.c b/drivers/net/hns3/hns3_mbx.c
index 3b35c02..ba04ac9 100644
--- a/drivers/net/hns3/hns3_mbx.c
+++ b/drivers/net/hns3/hns3_mbx.c
@@ -438,16 +438,19 @@ hns3_handle_mbx_msg_out_intr(struct hns3_hw *hw)
void
hns3_dev_handle_mbx_msg(struct hns3_hw *hw)
{
+ struct hns3_adapter *hns = HNS3_DEV_HW_TO_ADAPTER(hw);
struct hns3_cmq_ring *crq = &hw->cmq.crq;
struct hns3_mbx_pf_to_vf_cmd *req;
struct hns3_cmd_desc *desc;
+ bool handle_out;
uint8_t opcode;
uint16_t flag;
rte_spinlock_lock(&hw->cmq.crq.lock);
- if (rte_eal_process_type() != RTE_PROC_PRIMARY ||
- !rte_thread_is_intr()) {
+ handle_out = (rte_eal_process_type() != RTE_PROC_PRIMARY ||
+ !rte_thread_is_intr()) && hns->is_vf;
+ if (handle_out) {
/*
* Currently, any threads in the primary and secondary processes
* could send mailbox sync request, so it will need to process
@@ -491,7 +494,8 @@ hns3_dev_handle_mbx_msg(struct hns3_hw *hw)
continue;
}
- if (desc->opcode == 0) {
+ handle_out = hns->is_vf && desc->opcode == 0;
+ if (handle_out) {
/* Message already processed by other thread */
crq->desc[crq->next_to_use].flag = 0;
hns3_mbx_ring_ptr_move_crq(crq);
--
2.7.4
