Merge branch 'support-symmetric-xor-rss-hash'

Ahmed Zaki says:

====================
Support symmetric-xor RSS hash

Patches 1 and 2 modify the get/set_rxh ethtool API to take a pointer to
struct of parameters instead of individual params. This will allow future
changes to the uAPI-shared struct ethtool_rxfh without changing the
drivers' API.

Patch 3 adds the support at the Kernel level, allowing the user to set a
symmetric-xor RSS hash for a netdevice via:

    # ethtool -X eth0 hfunc toeplitz symmetric-xor

and clears the flag via:

    # ethtool -X eth0 hfunc toeplitz

The "symmetric-xor" is set in a new "input_xfrm" field in struct
ethtool_rxfh. Support for the new "symmetric-xor" flag will be later sent
to the "ethtool" user-space tool.

Patch 4 fixes a long standing bug with the ice hash function register
values. The bug has been benign for now since only (asymmetric) Toeplitz
hash (Zero) has been used.

Patches 5 and 6 lay some groundwork refactoring. While the first is
mainly cosmetic, the second is needed since there is no more room in the
previous 64-bit RSS profile ID for the symmetric attribute introduced in
the next patch.

Finally, patches 7 and 8 add the symmetric-xor support for the ice
(E800 PFs) and the iAVF drivers.
====================

Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Jakub Kicinski <[email protected]>

Author: kuba-moo

Documentation/netlink/specs/ethtool.yaml

@@ -908,6 +908,9 @@ attribute-sets:
       -
         name: hkey
         type: binary
+      -
+        name: input_xfrm
+        type: u32
   -
     name: plca
     attributes:
@@ -1598,6 +1601,7 @@ operations:
             - hfunc
             - indir
             - hkey
+            - input_xfrm
       dump: *rss-get-op
     -
       name: plca-get-cfg

Documentation/networking/ethtool-netlink.rst

@@ -1774,12 +1774,16 @@ Kernel response contents:
   ``ETHTOOL_A_RSS_HFUNC``              u32     RSS hash func
   ``ETHTOOL_A_RSS_INDIR``              binary  Indir table bytes
   ``ETHTOOL_A_RSS_HKEY``               binary  Hash key bytes
+  ``ETHTOOL_A_RSS_INPUT_XFRM``         u32     RSS input data transformation
 =====================================  ======  ==========================
 
 ETHTOOL_A_RSS_HFUNC attribute is bitmap indicating the hash function
 being used. Current supported options are toeplitz, xor or crc32.
-ETHTOOL_A_RSS_INDIR attribute returns RSS indrection table where each byte
+ETHTOOL_A_RSS_INDIR attribute returns RSS indirection table where each byte
 indicates queue number.
+ETHTOOL_A_RSS_INPUT_XFRM attribute is a bitmap indicating the type of
+transformation applied to the input protocol fields before given to the RSS
+hfunc. Current supported option is symmetric-xor.
 
 PLCA_GET_CFG
 ============

Documentation/networking/scaling.rst

@@ -44,6 +44,21 @@ by masking out the low order seven bits of the computed hash for the
 packet (usually a Toeplitz hash), taking this number as a key into the
 indirection table and reading the corresponding value.
 
+Some NICs support symmetric RSS hashing where, if the IP (source address,
+destination address) and TCP/UDP (source port, destination port) tuples
+are swapped, the computed hash is the same. This is beneficial in some
+applications that monitor TCP/IP flows (IDS, firewalls, ...etc) and need
+both directions of the flow to land on the same Rx queue (and CPU). The
+"Symmetric-XOR" is a type of RSS algorithms that achieves this hash
+symmetry by XORing the input source and destination fields of the IP
+and/or L4 protocols. This, however, results in reduced input entropy and
+could potentially be exploited. Specifically, the algorithm XORs the input
+as follows::
+
+    # (SRC_IP ^ DST_IP, SRC_IP ^ DST_IP, SRC_PORT ^ DST_PORT, SRC_PORT ^ DST_PORT)
+
+The result is then fed to the underlying RSS algorithm.
+
 Some advanced NICs allow steering packets to queues based on
 programmable filters. For example, webserver bound TCP port 80 packets
 can be directed to their own receive queue. Such “n-tuple” filters can

drivers/net/ethernet/amazon/ena/ena_ethtool.c

@@ -802,15 +802,15 @@ static int ena_indirection_table_get(struct ena_adapter *adapter, u32 *indir)
 	return rc;
 }
 
-static int ena_get_rxfh(struct net_device *netdev, u32 *indir, u8 *key,
-			u8 *hfunc)
+static int ena_get_rxfh(struct net_device *netdev,
+			struct ethtool_rxfh_param *rxfh)
 {
 	struct ena_adapter *adapter = netdev_priv(netdev);
 	enum ena_admin_hash_functions ena_func;
 	u8 func;
 	int rc;
 
-	rc = ena_indirection_table_get(adapter, indir);
+	rc = ena_indirection_table_get(adapter, rxfh->indir);
 	if (rc)
 		return rc;
 
@@ -825,7 +825,7 @@ static int ena_get_rxfh(struct net_device *netdev, u32 *indir, u8 *key,
 		return rc;
 	}
 
-	rc = ena_com_get_hash_key(adapter->ena_dev, key);
+	rc = ena_com_get_hash_key(adapter->ena_dev, rxfh->key);
 	if (rc)
 		return rc;
 
@@ -842,27 +842,27 @@ static int ena_get_rxfh(struct net_device *netdev, u32 *indir, u8 *key,
 		return -EOPNOTSUPP;
 	}
 
-	if (hfunc)
-		*hfunc = func;
+	rxfh->hfunc = func;
 
 	return 0;
 }
 
-static int ena_set_rxfh(struct net_device *netdev, const u32 *indir,
-			const u8 *key, const u8 hfunc)
+static int ena_set_rxfh(struct net_device *netdev,
+			struct ethtool_rxfh_param *rxfh,
+			struct netlink_ext_ack *extack)
 {
 	struct ena_adapter *adapter = netdev_priv(netdev);
 	struct ena_com_dev *ena_dev = adapter->ena_dev;
 	enum ena_admin_hash_functions func = 0;
 	int rc;
 
-	if (indir) {
-		rc = ena_indirection_table_set(adapter, indir);
+	if (rxfh->indir) {
+		rc = ena_indirection_table_set(adapter, rxfh->indir);
 		if (rc)
 			return rc;
 	}
 
-	switch (hfunc) {
+	switch (rxfh->hfunc) {
 	case ETH_RSS_HASH_NO_CHANGE:
 		func = ena_com_get_current_hash_function(ena_dev);
 		break;
@@ -874,12 +874,12 @@ static int ena_set_rxfh(struct net_device *netdev, const u32 *indir,
 		break;
 	default:
 		netif_err(adapter, drv, netdev, "Unsupported hfunc %d\n",
-			  hfunc);
+			  rxfh->hfunc);
 		return -EOPNOTSUPP;
 	}
 
-	if (key || func) {
-		rc = ena_com_fill_hash_function(ena_dev, func, key,
+	if (rxfh->key || func) {
+		rc = ena_com_fill_hash_function(ena_dev, func, rxfh->key,
 						ENA_HASH_KEY_SIZE,
 						0xFFFFFFFF);
 		if (unlikely(rc)) {

drivers/net/ethernet/amd/xgbe/xgbe-ethtool.c

@@ -527,47 +527,48 @@ static u32 xgbe_get_rxfh_indir_size(struct net_device *netdev)
 	return ARRAY_SIZE(pdata->rss_table);
 }
 
-static int xgbe_get_rxfh(struct net_device *netdev, u32 *indir, u8 *key,
-			 u8 *hfunc)
+static int xgbe_get_rxfh(struct net_device *netdev,
+			 struct ethtool_rxfh_param *rxfh)
 {
 	struct xgbe_prv_data *pdata = netdev_priv(netdev);
 	unsigned int i;
 
-	if (indir) {
+	if (rxfh->indir) {
 		for (i = 0; i < ARRAY_SIZE(pdata->rss_table); i++)
-			indir[i] = XGMAC_GET_BITS(pdata->rss_table[i],
-						  MAC_RSSDR, DMCH);
+			rxfh->indir[i] = XGMAC_GET_BITS(pdata->rss_table[i],
+							MAC_RSSDR, DMCH);
 	}
 
-	if (key)
-		memcpy(key, pdata->rss_key, sizeof(pdata->rss_key));
+	if (rxfh->key)
+		memcpy(rxfh->key, pdata->rss_key, sizeof(pdata->rss_key));
 
-	if (hfunc)
-		*hfunc = ETH_RSS_HASH_TOP;
+	rxfh->hfunc = ETH_RSS_HASH_TOP;
 
 	return 0;
 }
 
-static int xgbe_set_rxfh(struct net_device *netdev, const u32 *indir,
-			 const u8 *key, const u8 hfunc)
+static int xgbe_set_rxfh(struct net_device *netdev,
+			 struct ethtool_rxfh_param *rxfh,
+			 struct netlink_ext_ack *extack)
 {
 	struct xgbe_prv_data *pdata = netdev_priv(netdev);
 	struct xgbe_hw_if *hw_if = &pdata->hw_if;
 	unsigned int ret;
 
-	if (hfunc != ETH_RSS_HASH_NO_CHANGE && hfunc != ETH_RSS_HASH_TOP) {
+	if (rxfh->hfunc != ETH_RSS_HASH_NO_CHANGE &&
+	    rxfh->hfunc != ETH_RSS_HASH_TOP) {
 		netdev_err(netdev, "unsupported hash function\n");
 		return -EOPNOTSUPP;
 	}
 
-	if (indir) {
-		ret = hw_if->set_rss_lookup_table(pdata, indir);
+	if (rxfh->indir) {
+		ret = hw_if->set_rss_lookup_table(pdata, rxfh->indir);
 		if (ret)
 			return ret;
 	}
 
-	if (key) {
-		ret = hw_if->set_rss_hash_key(pdata, key);
+	if (rxfh->key) {
+		ret = hw_if->set_rss_hash_key(pdata, rxfh->key);
 		if (ret)
 			return ret;
 	}

drivers/net/ethernet/aquantia/atlantic/aq_ethtool.c

@@ -447,30 +447,30 @@ static u32 aq_ethtool_get_rss_key_size(struct net_device *ndev)
 	return sizeof(cfg->aq_rss.hash_secret_key);
 }
 
-static int aq_ethtool_get_rss(struct net_device *ndev, u32 *indir, u8 *key,
-			      u8 *hfunc)
+static int aq_ethtool_get_rss(struct net_device *ndev,
+			      struct ethtool_rxfh_param *rxfh)
 {
 	struct aq_nic_s *aq_nic = netdev_priv(ndev);
 	struct aq_nic_cfg_s *cfg;
 	unsigned int i = 0U;
 
 	cfg = aq_nic_get_cfg(aq_nic);
 
-	if (hfunc)
-		*hfunc = ETH_RSS_HASH_TOP; /* Toeplitz */
-	if (indir) {
+	rxfh->hfunc = ETH_RSS_HASH_TOP; /* Toeplitz */
+	if (rxfh->indir) {
 		for (i = 0; i < AQ_CFG_RSS_INDIRECTION_TABLE_MAX; i++)
-			indir[i] = cfg->aq_rss.indirection_table[i];
+			rxfh->indir[i] = cfg->aq_rss.indirection_table[i];
 	}
-	if (key)
-		memcpy(key, cfg->aq_rss.hash_secret_key,
+	if (rxfh->key)
+		memcpy(rxfh->key, cfg->aq_rss.hash_secret_key,
 		       sizeof(cfg->aq_rss.hash_secret_key));
 
 	return 0;
 }
 
-static int aq_ethtool_set_rss(struct net_device *netdev, const u32 *indir,
-			      const u8 *key, const u8 hfunc)
+static int aq_ethtool_set_rss(struct net_device *netdev,
+			      struct ethtool_rxfh_param *rxfh,
+			      struct netlink_ext_ack *extack)
 {
 	struct aq_nic_s *aq_nic = netdev_priv(netdev);
 	struct aq_nic_cfg_s *cfg;
@@ -482,16 +482,17 @@ static int aq_ethtool_set_rss(struct net_device *netdev, const u32 *indir,
 	rss_entries = cfg->aq_rss.indirection_table_size;
 
 	/* We do not allow change in unsupported parameters */
-	if (hfunc != ETH_RSS_HASH_NO_CHANGE && hfunc != ETH_RSS_HASH_TOP)
+	if (rxfh->hfunc != ETH_RSS_HASH_NO_CHANGE &&
+	    rxfh->hfunc != ETH_RSS_HASH_TOP)
 		return -EOPNOTSUPP;
 	/* Fill out the redirection table */
-	if (indir)
+	if (rxfh->indir)
 		for (i = 0; i < rss_entries; i++)
-			cfg->aq_rss.indirection_table[i] = indir[i];
+			cfg->aq_rss.indirection_table[i] = rxfh->indir[i];
 
 	/* Fill out the rss hash key */
-	if (key) {
-		memcpy(cfg->aq_rss.hash_secret_key, key,
+	if (rxfh->key) {
+		memcpy(cfg->aq_rss.hash_secret_key, rxfh->key,
 		       sizeof(cfg->aq_rss.hash_secret_key));
 		err = aq_nic->aq_hw_ops->hw_rss_hash_set(aq_nic->aq_hw,
 			&cfg->aq_rss);

drivers/net/ethernet/broadcom/bnx2x/bnx2x_ethtool.c

@@ -3486,16 +3486,15 @@ static u32 bnx2x_get_rxfh_indir_size(struct net_device *dev)
 	return T_ETH_INDIRECTION_TABLE_SIZE;
 }
 
-static int bnx2x_get_rxfh(struct net_device *dev, u32 *indir, u8 *key,
-			  u8 *hfunc)
+static int bnx2x_get_rxfh(struct net_device *dev,
+			  struct ethtool_rxfh_param *rxfh)
 {
 	struct bnx2x *bp = netdev_priv(dev);
 	u8 ind_table[T_ETH_INDIRECTION_TABLE_SIZE] = {0};
 	size_t i;
 
-	if (hfunc)
-		*hfunc = ETH_RSS_HASH_TOP;
-	if (!indir)
+	rxfh->hfunc = ETH_RSS_HASH_TOP;
+	if (!rxfh->indir)
 		return 0;
 
 	/* Get the current configuration of the RSS indirection table */
@@ -3511,25 +3510,27 @@ static int bnx2x_get_rxfh(struct net_device *dev, u32 *indir, u8 *key,
 	 * queue.
 	 */
 	for (i = 0; i < T_ETH_INDIRECTION_TABLE_SIZE; i++)
-		indir[i] = ind_table[i] - bp->fp->cl_id;
+		rxfh->indir[i] = ind_table[i] - bp->fp->cl_id;
 
 	return 0;
 }
 
-static int bnx2x_set_rxfh(struct net_device *dev, const u32 *indir,
-			  const u8 *key, const u8 hfunc)
+static int bnx2x_set_rxfh(struct net_device *dev,
+			  struct ethtool_rxfh_param *rxfh,
+			  struct netlink_ext_ack *extack)
 {
 	struct bnx2x *bp = netdev_priv(dev);
 	size_t i;
 
 	/* We require at least one supported parameter to be changed and no
 	 * change in any of the unsupported parameters
 	 */
-	if (key ||
-	    (hfunc != ETH_RSS_HASH_NO_CHANGE && hfunc != ETH_RSS_HASH_TOP))
+	if (rxfh->key ||
+	    (rxfh->hfunc != ETH_RSS_HASH_NO_CHANGE &&
+	     rxfh->hfunc != ETH_RSS_HASH_TOP))
 		return -EOPNOTSUPP;
 
-	if (!indir)
+	if (!rxfh->indir)
 		return 0;
 
 	for (i = 0; i < T_ETH_INDIRECTION_TABLE_SIZE; i++) {
@@ -3542,7 +3543,7 @@ static int bnx2x_set_rxfh(struct net_device *dev, const u32 *indir,
 		 * align the received table to the Client ID of the leading RSS
 		 * queue
 		 */
-		bp->rss_conf_obj.ind_table[i] = indir[i] + bp->fp->cl_id;
+		bp->rss_conf_obj.ind_table[i] = rxfh->indir[i] + bp->fp->cl_id;
 	}
 
 	if (bp->state == BNX2X_STATE_OPEN)

drivers/net/ethernet/broadcom/bnxt/bnxt_ethtool.c

@@ -1333,49 +1333,49 @@ static u32 bnxt_get_rxfh_key_size(struct net_device *dev)
 	return HW_HASH_KEY_SIZE;
 }
 
-static int bnxt_get_rxfh(struct net_device *dev, u32 *indir, u8 *key,
-			 u8 *hfunc)
+static int bnxt_get_rxfh(struct net_device *dev,
+			 struct ethtool_rxfh_param *rxfh)
 {
 	struct bnxt *bp = netdev_priv(dev);
 	struct bnxt_vnic_info *vnic;
 	u32 i, tbl_size;
 
-	if (hfunc)
-		*hfunc = ETH_RSS_HASH_TOP;
+	rxfh->hfunc = ETH_RSS_HASH_TOP;
 
 	if (!bp->vnic_info)
 		return 0;
 
 	vnic = &bp->vnic_info[0];
-	if (indir && bp->rss_indir_tbl) {
+	if (rxfh->indir && bp->rss_indir_tbl) {
 		tbl_size = bnxt_get_rxfh_indir_size(dev);
 		for (i = 0; i < tbl_size; i++)
-			indir[i] = bp->rss_indir_tbl[i];
+			rxfh->indir[i] = bp->rss_indir_tbl[i];
 	}
 
-	if (key && vnic->rss_hash_key)
-		memcpy(key, vnic->rss_hash_key, HW_HASH_KEY_SIZE);
+	if (rxfh->key && vnic->rss_hash_key)
+		memcpy(rxfh->key, vnic->rss_hash_key, HW_HASH_KEY_SIZE);
 
 	return 0;
 }
 
-static int bnxt_set_rxfh(struct net_device *dev, const u32 *indir,
-			 const u8 *key, const u8 hfunc)
+static int bnxt_set_rxfh(struct net_device *dev,
+			 struct ethtool_rxfh_param *rxfh,
+			 struct netlink_ext_ack *extack)
 {
 	struct bnxt *bp = netdev_priv(dev);
 	int rc = 0;
 
-	if (hfunc && hfunc != ETH_RSS_HASH_TOP)
+	if (rxfh->hfunc && rxfh->hfunc != ETH_RSS_HASH_TOP)
 		return -EOPNOTSUPP;
 
-	if (key)
+	if (rxfh->key)
 		return -EOPNOTSUPP;
 
-	if (indir) {
+	if (rxfh->indir) {
 		u32 i, pad, tbl_size = bnxt_get_rxfh_indir_size(dev);
 
 		for (i = 0; i < tbl_size; i++)
-			bp->rss_indir_tbl[i] = indir[i];
+			bp->rss_indir_tbl[i] = rxfh->indir[i];
 		pad = bp->rss_indir_tbl_entries - tbl_size;
 		if (pad)
 			memset(&bp->rss_indir_tbl[i], 0, pad * sizeof(u16));