eqos: fix kernel panic by adding napi_synchronize

Issue:	Kernel panic if NAPI schduled in ether_close() path.

Fix:	Disable NAPI before freeing dma resources as well as
	call napi_synchronize() before napi_disable() for NAPI
	wait

Bug 3006448
Bug 200650229

Change-Id: I833c150c8bf2834f7d4cb125ca1328203d5df622
Signed-off-by: rakesh goyal <rgoyal@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/c/linux-nvidia/+/2404294
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>

1 files changed, 2 insertions, 0 deletions

diff --git a/drivers/net/ethernet/nvidia/eqos/drv.c b/drivers/net/ethernet/nvidia/eqos/drv.c
index 010726699..f2e396031 100644
--- a/drivers/net/ethernet/nvidia/eqos/drv.c
+++ b/drivers/net/ethernet/nvidia/eqos/drv.c
@@ -166,7 +166,9 @@ static void eqos_all_ch_napi_disable(struct eqos_prv_data *pdata)
         pr_debug("-->eqos_napi_disable\n");
 
         for (qinx = 0; qinx < EQOS_RX_QUEUE_CNT; qinx++) {
+                napi_synchronize(&pdata->rx_queue[qinx].napi);
                 napi_disable(&pdata->rx_queue[qinx].napi);
+                napi_synchronize(&pdata->tx_queue[qinx].napi);
                 napi_disable(&pdata->tx_queue[qinx].napi);
         }