When app tries promisc/allmulti setting, fm10k will check if a valid glort
is acquired, if not then exit without doing anything. It's a long journey
for VF to acquire glort info from VF to PF mailbox, PF to switch mailbox.
It could be a long interval that's out of DPDK's control. Thus, app may
fail on promisc/allmulti setting in VF. In fact, we don't need a valid
glort value in VF, so this patch just skips the glort check for VF.
Fixes: df02ba864695 ("fm10k: support promiscuous mode")
Signed-off-by: Wang Xiao W <xiao.w.wang at intel.com>
---
drivers/net/fm10k/fm10k_ethdev.c | 8 ++++----
1 file changed, 4 insertions(+), 4 deletions(-)
diff --git a/drivers/net/fm10k/fm10k_ethdev.c b/drivers/net/fm10k/fm10k_ethdev.c
index c2d377f..b3aefdb 100644
--- a/drivers/net/fm10k/fm10k_ethdev.c
+++ b/drivers/net/fm10k/fm10k_ethdev.c
@@ -947,7 +947,7 @@ fm10k_dev_promiscuous_enable(struct rte_eth_dev *dev)
PMD_INIT_FUNC_TRACE();
/* Return if it didn't acquire valid glort range */
- if (!fm10k_glort_valid(hw))
+ if ((hw->mac.type == fm10k_mac_pf) && !fm10k_glort_valid(hw))
return;
fm10k_mbx_lock(hw);
@@ -969,7 +969,7 @@ fm10k_dev_promiscuous_disable(struct rte_eth_dev *dev)
PMD_INIT_FUNC_TRACE();
/* Return if it didn't acquire valid glort range */
- if (!fm10k_glort_valid(hw))
+ if ((hw->mac.type == fm10k_mac_pf) && !fm10k_glort_valid(hw))
return;
if (dev->data->all_multicast == 1)
@@ -995,7 +995,7 @@ fm10k_dev_allmulticast_enable(struct rte_eth_dev *dev)
PMD_INIT_FUNC_TRACE();
/* Return if it didn't acquire valid glort range */
- if (!fm10k_glort_valid(hw))
+ if ((hw->mac.type == fm10k_mac_pf) && !fm10k_glort_valid(hw))
return;
/* If promiscuous mode is enabled, it doesn't make sense to enable
@@ -1026,7 +1026,7 @@ fm10k_dev_allmulticast_disable(struct rte_eth_dev *dev)
PMD_INIT_FUNC_TRACE();
/* Return if it didn't acquire valid glort range */
- if (!fm10k_glort_valid(hw))
+ if ((hw->mac.type == fm10k_mac_pf) && !fm10k_glort_valid(hw))
return;
if (dev->data->promiscuous) {
--
1.9.3
