diff --git a/Documentation/networking/netdev-features.txt b/Documentation/networking/netdev-features.txt
index 7413eb05223b..c77f9d57eb91 100644
--- a/Documentation/networking/netdev-features.txt
+++ b/Documentation/networking/netdev-features.txt
@@ -163,3 +163,12 @@ This requests that the NIC receive all possible frames, including errored
 frames (such as bad FCS, etc).  This can be helpful when sniffing a link with
 bad packets on it.  Some NICs may receive more packets if also put into normal
 PROMISC mode.
+
+*  rx-gro-hw
+
+This requests that the NIC enables Hardware GRO (generic receive offload).
+Hardware GRO is basically the exact reverse of TSO, and is generally
+stricter than Hardware LRO.  A packet stream merged by Hardware GRO must
+be re-segmentable by GSO or TSO back to the exact original packet stream.
+Hardware GRO is dependent on RXCSUM since every packet successfully merged
+by hardware must also have the checksum verified by hardware.
diff --git a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.c b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.c
index 4c739d5355d2..01b7f2fc249c 100644
--- a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.c
+++ b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.c
@@ -2482,8 +2482,7 @@ static void bnx2x_bz_fp(struct bnx2x *bp, int index)
 	 */
 	if (bp->dev->features & NETIF_F_LRO)
 		fp->mode = TPA_MODE_LRO;
-	else if (bp->dev->features & NETIF_F_GRO &&
-		 bnx2x_mtu_allows_gro(bp->dev->mtu))
+	else if (bp->dev->features & NETIF_F_GRO_HW)
 		fp->mode = TPA_MODE_GRO;
 	else
 		fp->mode = TPA_MODE_DISABLED;
@@ -4874,6 +4873,9 @@ int bnx2x_change_mtu(struct net_device *dev, int new_mtu)
 	 */
 	dev->mtu = new_mtu;
 
+	if (!bnx2x_mtu_allows_gro(new_mtu))
+		dev->features &= ~NETIF_F_GRO_HW;
+
 	if (IS_PF(bp) && SHMEM2_HAS(bp, curr_cfg))
 		SHMEM2_WR(bp, curr_cfg, CURR_CFG_MET_OS);
 
@@ -4903,10 +4905,13 @@ netdev_features_t bnx2x_fix_features(struct net_device *dev,
 	}
 
 	/* TPA requires Rx CSUM offloading */
-	if (!(features & NETIF_F_RXCSUM)) {
+	if (!(features & NETIF_F_RXCSUM))
+		features &= ~NETIF_F_LRO;
+
+	if (!(features & NETIF_F_GRO) || !bnx2x_mtu_allows_gro(dev->mtu))
+		features &= ~NETIF_F_GRO_HW;
+	if (features & NETIF_F_GRO_HW)
 		features &= ~NETIF_F_LRO;
-		features &= ~NETIF_F_GRO;
-	}
 
 	return features;
 }
@@ -4933,13 +4938,8 @@ int bnx2x_set_features(struct net_device *dev, netdev_features_t features)
 		}
 	}
 
-	/* if GRO is changed while LRO is enabled, don't force a reload */
-	if ((changes & NETIF_F_GRO) && (features & NETIF_F_LRO))
-		changes &= ~NETIF_F_GRO;
-
-	/* if GRO is changed while HW TPA is off, don't force a reload */
-	if ((changes & NETIF_F_GRO) && bp->disable_tpa)
-		changes &= ~NETIF_F_GRO;
+	/* Don't care about GRO changes */
+	changes &= ~NETIF_F_GRO;
 
 	if (changes)
 		bnx2x_reload = true;
diff --git a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_main.c b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_main.c
index 91e2a7560b48..4d0654813de1 100644
--- a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_main.c
+++ b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_main.c
@@ -12400,8 +12400,8 @@ static int bnx2x_init_bp(struct bnx2x *bp)
 
 	/* Set TPA flags */
 	if (bp->disable_tpa) {
-		bp->dev->hw_features &= ~NETIF_F_LRO;
-		bp->dev->features &= ~NETIF_F_LRO;
+		bp->dev->hw_features &= ~(NETIF_F_LRO | NETIF_F_GRO_HW);
+		bp->dev->features &= ~(NETIF_F_LRO | NETIF_F_GRO_HW);
 	}
 
 	if (CHIP_IS_E1(bp))
@@ -13273,7 +13273,7 @@ static int bnx2x_init_dev(struct bnx2x *bp, struct pci_dev *pdev,
 
 	dev->hw_features = NETIF_F_SG | NETIF_F_IP_CSUM | NETIF_F_IPV6_CSUM |
 		NETIF_F_TSO | NETIF_F_TSO_ECN | NETIF_F_TSO6 |
-		NETIF_F_RXCSUM | NETIF_F_LRO | NETIF_F_GRO |
+		NETIF_F_RXCSUM | NETIF_F_LRO | NETIF_F_GRO | NETIF_F_GRO_HW |
 		NETIF_F_RXHASH | NETIF_F_HW_VLAN_CTAG_TX;
 	if (!chip_is_e1x) {
 		dev->hw_features |= NETIF_F_GSO_GRE | NETIF_F_GSO_GRE_CSUM |
@@ -13309,6 +13309,8 @@ static int bnx2x_init_dev(struct bnx2x *bp, struct pci_dev *pdev,
 
 	dev->features |= dev->hw_features | NETIF_F_HW_VLAN_CTAG_RX;
 	dev->features |= NETIF_F_HIGHDMA;
+	if (dev->features & NETIF_F_LRO)
+		dev->features &= ~NETIF_F_GRO_HW;
 
 	/* Add Loopback capability to the device */
 	dev->hw_features |= NETIF_F_LOOPBACK;
diff --git a/drivers/net/ethernet/broadcom/bnxt/bnxt.c b/drivers/net/ethernet/broadcom/bnxt/bnxt.c
index 1d865ae201db..9efbdc6f1fcb 100644
--- a/drivers/net/ethernet/broadcom/bnxt/bnxt.c
+++ b/drivers/net/ethernet/broadcom/bnxt/bnxt.c
@@ -2755,7 +2755,7 @@ void bnxt_set_tpa_flags(struct bnxt *bp)
 		return;
 	if (bp->dev->features & NETIF_F_LRO)
 		bp->flags |= BNXT_FLAG_LRO;
-	if (bp->dev->features & NETIF_F_GRO)
+	else if (bp->dev->features & NETIF_F_GRO_HW)
 		bp->flags |= BNXT_FLAG_GRO;
 }
 
@@ -2843,10 +2843,10 @@ int bnxt_set_rx_skb_mode(struct bnxt *bp, bool page_mode)
 			min_t(u16, bp->max_mtu, BNXT_MAX_PAGE_MODE_MTU);
 		bp->flags &= ~BNXT_FLAG_AGG_RINGS;
 		bp->flags |= BNXT_FLAG_NO_AGG_RINGS | BNXT_FLAG_RX_PAGE_MODE;
-		bp->dev->hw_features &= ~NETIF_F_LRO;
-		bp->dev->features &= ~NETIF_F_LRO;
 		bp->rx_dir = DMA_BIDIRECTIONAL;
 		bp->rx_skb_func = bnxt_rx_page_skb;
+		/* Disable LRO or GRO_HW */
+		netdev_update_features(bp->dev);
 	} else {
 		bp->dev->max_mtu = bp->max_mtu;
 		bp->flags &= ~BNXT_FLAG_RX_PAGE_MODE;
@@ -6788,6 +6788,15 @@ static netdev_features_t bnxt_fix_features(struct net_device *dev,
 	if ((features & NETIF_F_NTUPLE) && !bnxt_rfs_capable(bp))
 		features &= ~NETIF_F_NTUPLE;
 
+	if (bp->flags & BNXT_FLAG_NO_AGG_RINGS)
+		features &= ~(NETIF_F_LRO | NETIF_F_GRO_HW);
+
+	if (!(features & NETIF_F_GRO))
+		features &= ~NETIF_F_GRO_HW;
+
+	if (features & NETIF_F_GRO_HW)
+		features &= ~NETIF_F_LRO;
+
 	/* Both CTAG and STAG VLAN accelaration on the RX side have to be
 	 * turned on or off together.
 	 */
@@ -6821,9 +6830,9 @@ static int bnxt_set_features(struct net_device *dev, netdev_features_t features)
 	bool update_tpa = false;
 
 	flags &= ~BNXT_FLAG_ALL_CONFIG_FEATS;
-	if ((features & NETIF_F_GRO) && !BNXT_CHIP_TYPE_NITRO_A0(bp))
+	if (features & NETIF_F_GRO_HW)
 		flags |= BNXT_FLAG_GRO;
-	if (features & NETIF_F_LRO)
+	else if (features & NETIF_F_LRO)
 		flags |= BNXT_FLAG_LRO;
 
 	if (bp->flags & BNXT_FLAG_NO_AGG_RINGS)
@@ -7924,8 +7933,8 @@ static int bnxt_get_dflt_rings(struct bnxt *bp, int *max_rx, int *max_tx,
 		if (rc)
 			return rc;
 		bp->flags |= BNXT_FLAG_NO_AGG_RINGS;
-		bp->dev->hw_features &= ~NETIF_F_LRO;
-		bp->dev->features &= ~NETIF_F_LRO;
+		bp->dev->hw_features &= ~(NETIF_F_LRO | NETIF_F_GRO_HW);
+		bp->dev->features &= ~(NETIF_F_LRO | NETIF_F_GRO_HW);
 		bnxt_set_ring_params(bp);
 	}
 
@@ -8108,7 +8117,11 @@ static int bnxt_init_one(struct pci_dev *pdev, const struct pci_device_id *ent)
 	dev->vlan_features = dev->hw_features | NETIF_F_HIGHDMA;
 	dev->hw_features |= NETIF_F_HW_VLAN_CTAG_RX | NETIF_F_HW_VLAN_CTAG_TX |
 			    NETIF_F_HW_VLAN_STAG_RX | NETIF_F_HW_VLAN_STAG_TX;
+	if (!BNXT_CHIP_TYPE_NITRO_A0(bp))
+		dev->hw_features |= NETIF_F_GRO_HW;
 	dev->features |= dev->hw_features | NETIF_F_HIGHDMA;
+	if (dev->features & NETIF_F_GRO_HW)
+		dev->features &= ~NETIF_F_LRO;
 	dev->priv_flags |= IFF_UNICAST_FLT;
 
 #ifdef CONFIG_BNXT_SRIOV
diff --git a/drivers/net/ethernet/qlogic/qede/qede.h b/drivers/net/ethernet/qlogic/qede/qede.h
index a3a70ade411f..8a336517baac 100644
--- a/drivers/net/ethernet/qlogic/qede/qede.h
+++ b/drivers/net/ethernet/qlogic/qede/qede.h
@@ -494,6 +494,8 @@ int qede_vlan_rx_kill_vid(struct net_device *dev, __be16 proto, u16 vid);
 void qede_vlan_mark_nonconfigured(struct qede_dev *edev);
 int qede_configure_vlan_filters(struct qede_dev *edev);
 
+netdev_features_t qede_fix_features(struct net_device *dev,
+				    netdev_features_t features);
 int qede_set_features(struct net_device *dev, netdev_features_t features);
 void qede_set_rx_mode(struct net_device *ndev);
 void qede_config_rx_mode(struct net_device *ndev);
diff --git a/drivers/net/ethernet/qlogic/qede/qede_ethtool.c b/drivers/net/ethernet/qlogic/qede/qede_ethtool.c
index dae741270022..4ca3847fffd4 100644
--- a/drivers/net/ethernet/qlogic/qede/qede_ethtool.c
+++ b/drivers/net/ethernet/qlogic/qede/qede_ethtool.c
@@ -940,6 +940,9 @@ int qede_change_mtu(struct net_device *ndev, int new_mtu)
 	DP_VERBOSE(edev, (NETIF_MSG_IFUP | NETIF_MSG_IFDOWN),
 		   "Configuring MTU size of %d\n", new_mtu);
 
+	if (new_mtu > PAGE_SIZE)
+		ndev->features &= ~NETIF_F_GRO_HW;
+
 	/* Set the mtu field and re-start the interface if needed */
 	args.u.mtu = new_mtu;
 	args.func = &qede_update_mtu;
diff --git a/drivers/net/ethernet/qlogic/qede/qede_filter.c b/drivers/net/ethernet/qlogic/qede/qede_filter.c
index c1a0708a7d7c..77aa826227e5 100644
--- a/drivers/net/ethernet/qlogic/qede/qede_filter.c
+++ b/drivers/net/ethernet/qlogic/qede/qede_filter.c
@@ -895,19 +895,26 @@ static void qede_set_features_reload(struct qede_dev *edev,
 	edev->ndev->features = args->u.features;
 }
 
+netdev_features_t qede_fix_features(struct net_device *dev,
+				    netdev_features_t features)
+{
+	struct qede_dev *edev = netdev_priv(dev);
+
+	if (edev->xdp_prog || edev->ndev->mtu > PAGE_SIZE ||
+	    !(features & NETIF_F_GRO))
+		features &= ~NETIF_F_GRO_HW;
+
+	return features;
+}
+
 int qede_set_features(struct net_device *dev, netdev_features_t features)
 {
 	struct qede_dev *edev = netdev_priv(dev);
 	netdev_features_t changes = features ^ dev->features;
 	bool need_reload = false;
 
-	/* No action needed if hardware GRO is disabled during driver load */
-	if (changes & NETIF_F_GRO) {
-		if (dev->features & NETIF_F_GRO)
-			need_reload = !edev->gro_disable;
-		else
-			need_reload = edev->gro_disable;
-	}
+	if (changes & NETIF_F_GRO_HW)
+		need_reload = true;
 
 	if (need_reload) {
 		struct qede_reload_args args;
diff --git a/drivers/net/ethernet/qlogic/qede/qede_main.c b/drivers/net/ethernet/qlogic/qede/qede_main.c
index 57332b3e5e64..90d79ae2a48f 100644
--- a/drivers/net/ethernet/qlogic/qede/qede_main.c
+++ b/drivers/net/ethernet/qlogic/qede/qede_main.c
@@ -545,6 +545,7 @@ static const struct net_device_ops qede_netdev_ops = {
 #endif
 	.ndo_vlan_rx_add_vid = qede_vlan_rx_add_vid,
 	.ndo_vlan_rx_kill_vid = qede_vlan_rx_kill_vid,
+	.ndo_fix_features = qede_fix_features,
 	.ndo_set_features = qede_set_features,
 	.ndo_get_stats64 = qede_get_stats64,
 #ifdef CONFIG_QED_SRIOV
@@ -572,6 +573,7 @@ static const struct net_device_ops qede_netdev_vf_ops = {
 	.ndo_change_mtu = qede_change_mtu,
 	.ndo_vlan_rx_add_vid = qede_vlan_rx_add_vid,
 	.ndo_vlan_rx_kill_vid = qede_vlan_rx_kill_vid,
+	.ndo_fix_features = qede_fix_features,
 	.ndo_set_features = qede_set_features,
 	.ndo_get_stats64 = qede_get_stats64,
 	.ndo_udp_tunnel_add = qede_udp_tunnel_add,
@@ -589,6 +591,7 @@ static const struct net_device_ops qede_netdev_vf_xdp_ops = {
 	.ndo_change_mtu = qede_change_mtu,
 	.ndo_vlan_rx_add_vid = qede_vlan_rx_add_vid,
 	.ndo_vlan_rx_kill_vid = qede_vlan_rx_kill_vid,
+	.ndo_fix_features = qede_fix_features,
 	.ndo_set_features = qede_set_features,
 	.ndo_get_stats64 = qede_get_stats64,
 	.ndo_udp_tunnel_add = qede_udp_tunnel_add,
@@ -676,7 +679,7 @@ static void qede_init_ndev(struct qede_dev *edev)
 	ndev->priv_flags |= IFF_UNICAST_FLT;
 
 	/* user-changeble features */
-	hw_features = NETIF_F_GRO | NETIF_F_SG |
+	hw_features = NETIF_F_GRO | NETIF_F_GRO_HW | NETIF_F_SG |
 		      NETIF_F_IP_CSUM | NETIF_F_IPV6_CSUM |
 		      NETIF_F_TSO | NETIF_F_TSO6;
 
@@ -1228,18 +1231,9 @@ static int qede_alloc_sge_mem(struct qede_dev *edev, struct qede_rx_queue *rxq)
 	dma_addr_t mapping;
 	int i;
 
-	/* Don't perform FW aggregations in case of XDP */
-	if (edev->xdp_prog)
-		edev->gro_disable = 1;
-
 	if (edev->gro_disable)
 		return 0;
 
-	if (edev->ndev->mtu > PAGE_SIZE) {
-		edev->gro_disable = 1;
-		return 0;
-	}
-
 	for (i = 0; i < ETH_TPA_MAX_AGGS_NUM; i++) {
 		struct qede_agg_info *tpa_info = &rxq->tpa_info[i];
 		struct sw_rx_data *replace_buf = &tpa_info->buffer;
@@ -1269,6 +1263,7 @@ static int qede_alloc_sge_mem(struct qede_dev *edev, struct qede_rx_queue *rxq)
 err:
 	qede_free_sge_mem(edev, rxq);
 	edev->gro_disable = 1;
+	edev->ndev->features &= ~NETIF_F_GRO_HW;
 	return -ENOMEM;
 }
 
@@ -1511,7 +1506,7 @@ static void qede_init_fp(struct qede_dev *edev)
 			 edev->ndev->name, queue_id);
 	}
 
-	edev->gro_disable = !(edev->ndev->features & NETIF_F_GRO);
+	edev->gro_disable = !(edev->ndev->features & NETIF_F_GRO_HW);
 }
 
 static int qede_set_real_num_queues(struct qede_dev *edev)
diff --git a/include/linux/netdev_features.h b/include/linux/netdev_features.h
index b1b0ca7ccb2b..db84c516bcfb 100644
--- a/include/linux/netdev_features.h
+++ b/include/linux/netdev_features.h
@@ -78,6 +78,8 @@ enum {
 	NETIF_F_HW_ESP_TX_CSUM_BIT,	/* ESP with TX checksum offload */
 	NETIF_F_RX_UDP_TUNNEL_PORT_BIT, /* Offload of RX port for UDP tunnels */
 
+	NETIF_F_GRO_HW_BIT,		/* Hardware Generic receive offload */
+
 	/*
 	 * Add your fresh new feature above and remember to update
 	 * netdev_features_strings[] in net/core/ethtool.c and maybe
@@ -97,6 +99,7 @@ enum {
 #define NETIF_F_FRAGLIST	__NETIF_F(FRAGLIST)
 #define NETIF_F_FSO		__NETIF_F(FSO)
 #define NETIF_F_GRO		__NETIF_F(GRO)
+#define NETIF_F_GRO_HW		__NETIF_F(GRO_HW)
 #define NETIF_F_GSO		__NETIF_F(GSO)
 #define NETIF_F_GSO_ROBUST	__NETIF_F(GSO_ROBUST)
 #define NETIF_F_HIGHDMA		__NETIF_F(HIGHDMA)
diff --git a/net/core/dev.c b/net/core/dev.c
index b0eee49a2489..c7db39926769 100644
--- a/net/core/dev.c
+++ b/net/core/dev.c
@@ -1542,6 +1542,23 @@ void dev_disable_lro(struct net_device *dev)
 }
 EXPORT_SYMBOL(dev_disable_lro);
 
+/**
+ *	dev_disable_gro_hw - disable HW Generic Receive Offload on a device
+ *	@dev: device
+ *
+ *	Disable HW Generic Receive Offload (GRO_HW) on a net device.  Must be
+ *	called under RTNL.  This is needed if Generic XDP is installed on
+ *	the device.
+ */
+static void dev_disable_gro_hw(struct net_device *dev)
+{
+	dev->wanted_features &= ~NETIF_F_GRO_HW;
+	netdev_update_features(dev);
+
+	if (unlikely(dev->features & NETIF_F_GRO_HW))
+		netdev_WARN(dev, "failed to disable GRO_HW!\n");
+}
+
 static int call_netdevice_notifier(struct notifier_block *nb, unsigned long val,
 				   struct net_device *dev)
 {
@@ -4564,6 +4581,7 @@ static int generic_xdp_install(struct net_device *dev, struct netdev_bpf *xdp)
 		} else if (new && !old) {
 			static_key_slow_inc(&generic_xdp_needed);
 			dev_disable_lro(dev);
+			dev_disable_gro_hw(dev);
 		}
 		break;
 
@@ -7424,6 +7442,18 @@ static netdev_features_t netdev_fix_features(struct net_device *dev,
 		features &= ~dev->gso_partial_features;
 	}
 
+	if (!(features & NETIF_F_RXCSUM)) {
+		/* NETIF_F_GRO_HW implies doing RXCSUM since every packet
+		 * successfully merged by hardware must also have the
+		 * checksum verified by hardware.  If the user does not
+		 * want to enable RXCSUM, logically, we should disable GRO_HW.
+		 */
+		if (features & NETIF_F_GRO_HW) {
+			netdev_dbg(dev, "Dropping NETIF_F_GRO_HW since no RXCSUM feature.\n");
+			features &= ~NETIF_F_GRO_HW;
+		}
+	}
+
 	return features;
 }
 
diff --git a/net/core/ethtool.c b/net/core/ethtool.c
index f8fcf450a36e..50a79203043b 100644
--- a/net/core/ethtool.c
+++ b/net/core/ethtool.c
@@ -73,6 +73,7 @@ static const char netdev_features_strings[NETDEV_FEATURE_COUNT][ETH_GSTRING_LEN]
 	[NETIF_F_LLTX_BIT] =             "tx-lockless",
 	[NETIF_F_NETNS_LOCAL_BIT] =      "netns-local",
 	[NETIF_F_GRO_BIT] =              "rx-gro",
+	[NETIF_F_GRO_HW_BIT] =           "rx-gro-hw",
 	[NETIF_F_LRO_BIT] =              "rx-lro",
 
 	[NETIF_F_TSO_BIT] =              "tx-tcp-segmentation",