]> www.infradead.org Git - users/dwmw2/linux.git/commitdiff
bnxt_en: Don't query FW when netif_running() is false.
authorPavan Chebbi <pavan.chebbi@broadcom.com>
Wed, 26 Aug 2020 05:08:32 +0000 (01:08 -0400)
committerGreg Kroah-Hartman <gregkh@linuxfoundation.org>
Wed, 9 Sep 2020 17:12:25 +0000 (19:12 +0200)
[ Upstream commit c1c2d77408022a398a1a7c51cf20488c922629de ]

In rare conditions like two stage OS installation, the
ethtool's get_channels function may be called when the
device is in D3 state, leading to uncorrectable PCI error.
Check netif_running() first before making any query to FW
which involves writing to BAR.

Fixes: db4723b3cd2d ("bnxt_en: Check max_tx_scheduler_inputs value from firmware.")
Signed-off-by: Pavan Chebbi <pavan.chebbi@broadcom.com>
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>
drivers/net/ethernet/broadcom/bnxt/bnxt_ethtool.c

index 1f512e7c3d434b6513783e96436ca1a95516e846..de9b34a255cf1d5da2ab3367d01fcb043d43c580 100644 (file)
@@ -769,7 +769,7 @@ static void bnxt_get_channels(struct net_device *dev,
        int max_tx_sch_inputs;
 
        /* Get the most up-to-date max_tx_sch_inputs. */
-       if (BNXT_NEW_RM(bp))
+       if (netif_running(dev) && BNXT_NEW_RM(bp))
                bnxt_hwrm_func_resc_qcaps(bp, false);
        max_tx_sch_inputs = hw_resc->max_tx_sch_inputs;