mlxsw: pci: Avoid flow control for EMAD packets

Locally generated packets ingress the device through its CPU port. When
the CPU port is congested and there are not enough credits in its
headroom buffer, packets can be dropped.

While this might be acceptable for data packets that traverse the
network, configuration packets exchanged between the host and the device
(EMADs) should not be subjected to this flow control.

The "sdq_lp" bit in the SDQ (Send Descriptor Queue) context allows the
host to instruct the device to treat packets sent on this queue as
"local processing" and always process them, regardless of the state of
the CPU port's headroom.

Add the definition of this bit and set it for the dedicated SDQ reserved
for the transmission of EMAD packets. This makes the "local processing"
bit in the WQE (Work Queue Element) redundant, so clear it.
Signed-off-by: Danielle Ratson <danieller@nvidia.com>
Signed-off-by: Ido Schimmel <idosch@nvidia.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

drivers/net/ethernet/mellanox/mlxsw/cmd.h

@@ -935,6 +935,18 @@ static inline int mlxsw_cmd_sw2hw_rdq(struct mlxsw_core *mlxsw_core,
  */
 MLXSW_ITEM32(cmd_mbox, sw2hw_dq, cq, 0x00, 24, 8);
 
+enum mlxsw_cmd_mbox_sw2hw_dq_sdq_lp {
+	MLXSW_CMD_MBOX_SW2HW_DQ_SDQ_LP_WQE,
+	MLXSW_CMD_MBOX_SW2HW_DQ_SDQ_LP_IGNORE_WQE,
+};
+
+/* cmd_mbox_sw2hw_dq_sdq_lp
+ * SDQ local Processing
+ * 0: local processing by wqe.lp
+ * 1: local processing (ignoring wqe.lp)
+ */
+MLXSW_ITEM32(cmd_mbox, sw2hw_dq, sdq_lp, 0x00, 23, 1);
+
 /* cmd_mbox_sw2hw_dq_sdq_tclass
  * SDQ: CPU Egress TClass
  * RDQ: Reserved

drivers/net/ethernet/mellanox/mlxsw/pci.c

@@ -285,6 +285,7 @@ static int mlxsw_pci_sdq_init(struct mlxsw_pci *mlxsw_pci, char *mbox,
 			      struct mlxsw_pci_queue *q)
 {
 	int tclass;
+	int lp;
 	int i;
 	int err;
 
@@ -292,9 +293,12 @@ static int mlxsw_pci_sdq_init(struct mlxsw_pci *mlxsw_pci, char *mbox,
 	q->consumer_counter = 0;
 	tclass = q->num == MLXSW_PCI_SDQ_EMAD_INDEX ? MLXSW_PCI_SDQ_EMAD_TC :
 						      MLXSW_PCI_SDQ_CTL_TC;
+	lp = q->num == MLXSW_PCI_SDQ_EMAD_INDEX ? MLXSW_CMD_MBOX_SW2HW_DQ_SDQ_LP_IGNORE_WQE :
+						  MLXSW_CMD_MBOX_SW2HW_DQ_SDQ_LP_WQE;
 
 	/* Set CQ of same number of this SDQ. */
 	mlxsw_cmd_mbox_sw2hw_dq_cq_set(mbox, q->num);
+	mlxsw_cmd_mbox_sw2hw_dq_sdq_lp_set(mbox, lp);
 	mlxsw_cmd_mbox_sw2hw_dq_sdq_tclass_set(mbox, tclass);
 	mlxsw_cmd_mbox_sw2hw_dq_log2_dq_sz_set(mbox, 3); /* 8 pages */
 	for (i = 0; i < MLXSW_PCI_AQ_PAGES; i++) {
@@ -1678,7 +1682,7 @@ static int mlxsw_pci_skb_transmit(void *bus_priv, struct sk_buff *skb,
 
 	wqe = elem_info->elem;
 	mlxsw_pci_wqe_c_set(wqe, 1); /* always report completion */
-	mlxsw_pci_wqe_lp_set(wqe, !!tx_info->is_emad);
+	mlxsw_pci_wqe_lp_set(wqe, 0);
 	mlxsw_pci_wqe_type_set(wqe, MLXSW_PCI_WQE_TYPE_ETHERNET);
 
 	err = mlxsw_pci_wqe_frag_map(mlxsw_pci, wqe, 0, skb->data,