bnx2x: MTU for FCoE L2 ring

Always configure an FCoE L2 ring with a mini-jumbo MTU size (2500).
To do that we had to move the rx_buf_size parameter from per
function level to a per ring level.

Signed-off-by: Vladislav Zolotarov <vladz@broadcom.com>
Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

5 files changed, 57 insertions, 21 deletions

diff --git a/drivers/net/bnx2x/bnx2x.h b/drivers/net/bnx2x/bnx2x.h
index ff87ec33d00e..c29b37e5e743 100644
--- a/drivers/net/bnx2x/bnx2x.h
+++ b/drivers/net/bnx2x/bnx2x.h
@@ -341,6 +341,8 @@ struct bnx2x_fastpath {
         /* chip independed shortcut into rx_prods_offset memory */
         u32                     ustorm_rx_prods_offset;
 
+        u32                     rx_buf_size;
+
         dma_addr_t              status_blk_mapping;
 
         struct sw_tx_bd         *tx_buf_ring;
@@ -428,6 +430,10 @@ struct bnx2x_fastpath {
 };
 
 #define bnx2x_fp(bp, nr, var)           (bp->fp[nr].var)
+
+/* Use 2500 as a mini-jumbo MTU for FCoE */
+#define BNX2X_FCOE_MINI_JUMBO_MTU       2500
+
 #ifdef BCM_CNIC
 /* FCoE L2 `fastpath' is right after the eth entries */
 #define FCOE_IDX                        BNX2X_NUM_ETH_QUEUES(bp)
@@ -911,7 +917,6 @@ struct bnx2x {
         int                     tx_ring_size;
 
         u32                     rx_csum;
-        u32                     rx_buf_size;
 /* L2 header size + 2*VLANs (8 bytes) + LLC SNAP (8 bytes) */
 #define ETH_OVREHEAD            (ETH_HLEN + 8 + 8)
 #define ETH_MIN_PACKET_SIZE             60
diff --git a/drivers/net/bnx2x/bnx2x_cmn.c b/drivers/net/bnx2x/bnx2x_cmn.c
index 710ce5d04c53..844afcec79b4 100644
--- a/drivers/net/bnx2x/bnx2x_cmn.c
+++ b/drivers/net/bnx2x/bnx2x_cmn.c
@@ -232,7 +232,7 @@ static void bnx2x_tpa_start(struct bnx2x_fastpath *fp, u16 queue,
         /* move empty skb from pool to prod and map it */
         prod_rx_buf->skb = fp->tpa_pool[queue].skb;
         mapping = dma_map_single(&bp->pdev->dev, fp->tpa_pool[queue].skb->data,
-                                 bp->rx_buf_size, DMA_FROM_DEVICE);
+                                 fp->rx_buf_size, DMA_FROM_DEVICE);
         dma_unmap_addr_set(prod_rx_buf, mapping, mapping);
 
         /* move partial skb from cons to pool (don't unmap yet) */
@@ -333,13 +333,13 @@ static void bnx2x_tpa_stop(struct bnx2x *bp, struct bnx2x_fastpath *fp,
         struct sw_rx_bd *rx_buf = &fp->tpa_pool[queue];
         struct sk_buff *skb = rx_buf->skb;
         /* alloc new skb */
-        struct sk_buff *new_skb = netdev_alloc_skb(bp->dev, bp->rx_buf_size);
+        struct sk_buff *new_skb = netdev_alloc_skb(bp->dev, fp->rx_buf_size);
 
         /* Unmap skb in the pool anyway, as we are going to change
            pool entry status to BNX2X_TPA_STOP even if new skb allocation
            fails. */
         dma_unmap_single(&bp->pdev->dev, dma_unmap_addr(rx_buf, mapping),
-                         bp->rx_buf_size, DMA_FROM_DEVICE);
+                         fp->rx_buf_size, DMA_FROM_DEVICE);
 
         if (likely(new_skb)) {
                 /* fix ip xsum and give it to the stack */
@@ -349,10 +349,10 @@ static void bnx2x_tpa_stop(struct bnx2x *bp, struct bnx2x_fastpath *fp,
                 prefetch(((char *)(skb)) + L1_CACHE_BYTES);
 
 #ifdef BNX2X_STOP_ON_ERROR
-                if (pad + len > bp->rx_buf_size) {
+                if (pad + len > fp->rx_buf_size) {
                         BNX2X_ERR("skb_put is about to fail...  "
                                   "pad %d  len %d  rx_buf_size %d\n",
-                                  pad, len, bp->rx_buf_size);
+                                  pad, len, fp->rx_buf_size);
                         bnx2x_panic();
                         return;
                 }
@@ -582,7 +582,7 @@ int bnx2x_rx_int(struct bnx2x_fastpath *fp, int budget)
                         if (likely(bnx2x_alloc_rx_skb(bp, fp, bd_prod) == 0)) {
                                 dma_unmap_single(&bp->pdev->dev,
                                         dma_unmap_addr(rx_buf, mapping),
-                                                 bp->rx_buf_size,
+                                                 fp->rx_buf_size,
                                                  DMA_FROM_DEVICE);
                                 skb_reserve(skb, pad);
                                 skb_put(skb, len);
@@ -821,19 +821,16 @@ void bnx2x_init_rx_rings(struct bnx2x *bp)
         u16 ring_prod;
         int i, j;
 
-        bp->rx_buf_size = bp->dev->mtu + ETH_OVREHEAD + BNX2X_RX_ALIGN +
-                IP_HEADER_ALIGNMENT_PADDING;
-
-        DP(NETIF_MSG_IFUP,
-           "mtu %d  rx_buf_size %d\n", bp->dev->mtu, bp->rx_buf_size);
-
         for_each_rx_queue(bp, j) {
                 struct bnx2x_fastpath *fp = &bp->fp[j];
 
+                DP(NETIF_MSG_IFUP,
+                   "mtu %d  rx_buf_size %d\n", bp->dev->mtu, fp->rx_buf_size);
+
                 if (!fp->disable_tpa) {
                         for (i = 0; i < max_agg_queues; i++) {
                                 fp->tpa_pool[i].skb =
-                                   netdev_alloc_skb(bp->dev, bp->rx_buf_size);
+                                   netdev_alloc_skb(bp->dev, fp->rx_buf_size);
                                 if (!fp->tpa_pool[i].skb) {
                                         BNX2X_ERR("Failed to allocate TPA "
                                                   "skb pool for queue[%d] - "
@@ -941,7 +938,7 @@ static void bnx2x_free_rx_skbs(struct bnx2x *bp)
 
                         dma_unmap_single(&bp->pdev->dev,
                                          dma_unmap_addr(rx_buf, mapping),
-                                         bp->rx_buf_size, DMA_FROM_DEVICE);
+                                         fp->rx_buf_size, DMA_FROM_DEVICE);
 
                         rx_buf->skb = NULL;
                         dev_kfree_skb(skb);
@@ -1249,6 +1246,31 @@ static inline int bnx2x_set_real_num_queues(struct bnx2x *bp)
         return rc;
 }
 
+static inline void bnx2x_set_rx_buf_size(struct bnx2x *bp)
+{
+        int i;
+
+        for_each_queue(bp, i) {
+                struct bnx2x_fastpath *fp = &bp->fp[i];
+
+                /* Always use a mini-jumbo MTU for the FCoE L2 ring */
+                if (IS_FCOE_IDX(i))
+                        /*
+                         * Although there are no IP frames expected to arrive to
+                         * this ring we still want to add an
+                         * IP_HEADER_ALIGNMENT_PADDING to prevent a buffer
+                         * overrun attack.
+                         */
+                        fp->rx_buf_size =
+                                BNX2X_FCOE_MINI_JUMBO_MTU + ETH_OVREHEAD +
+                                BNX2X_RX_ALIGN + IP_HEADER_ALIGNMENT_PADDING;
+                else
+                        fp->rx_buf_size =
+                                bp->dev->mtu + ETH_OVREHEAD + BNX2X_RX_ALIGN +
+                                IP_HEADER_ALIGNMENT_PADDING;
+        }
+}
+
 /* must be called with rtnl_lock */
 int bnx2x_nic_load(struct bnx2x *bp, int load_mode)
 {
@@ -1272,6 +1294,9 @@ int bnx2x_nic_load(struct bnx2x *bp, int load_mode)
         /* must be called before memory allocation and HW init */
         bnx2x_ilt_set_info(bp);
 
+        /* Set the receive queues buffer size */
+        bnx2x_set_rx_buf_size(bp);
+
         if (bnx2x_alloc_mem(bp))
                 return -ENOMEM;
 
diff --git a/drivers/net/bnx2x/bnx2x_cmn.h b/drivers/net/bnx2x/bnx2x_cmn.h
index 03eb4d68e6bb..f062d5d20fa9 100644
--- a/drivers/net/bnx2x/bnx2x_cmn.h
+++ b/drivers/net/bnx2x/bnx2x_cmn.h
@@ -822,11 +822,11 @@ static inline int bnx2x_alloc_rx_skb(struct bnx2x *bp,
         struct eth_rx_bd *rx_bd = &fp->rx_desc_ring[index];
         dma_addr_t mapping;
 
-        skb = netdev_alloc_skb(bp->dev, bp->rx_buf_size);
+        skb = netdev_alloc_skb(bp->dev, fp->rx_buf_size);
         if (unlikely(skb == NULL))
                 return -ENOMEM;
 
-        mapping = dma_map_single(&bp->pdev->dev, skb->data, bp->rx_buf_size,
+        mapping = dma_map_single(&bp->pdev->dev, skb->data, fp->rx_buf_size,
                                  DMA_FROM_DEVICE);
         if (unlikely(dma_mapping_error(&bp->pdev->dev, mapping))) {
                 dev_kfree_skb(skb);
@@ -892,7 +892,7 @@ static inline void bnx2x_free_tpa_pool(struct bnx2x *bp,
                 if (fp->tpa_state[i] == BNX2X_TPA_START)
                         dma_unmap_single(&bp->pdev->dev,
                                          dma_unmap_addr(rx_buf, mapping),
-                                         bp->rx_buf_size, DMA_FROM_DEVICE);
+                                         fp->rx_buf_size, DMA_FROM_DEVICE);
 
                 dev_kfree_skb(skb);
                 rx_buf->skb = NULL;
diff --git a/drivers/net/bnx2x/bnx2x_ethtool.c b/drivers/net/bnx2x/bnx2x_ethtool.c
index 5b44a8b48509..816fef6d3844 100644
--- a/drivers/net/bnx2x/bnx2x_ethtool.c
+++ b/drivers/net/bnx2x/bnx2x_ethtool.c
@@ -1618,7 +1618,7 @@ static int bnx2x_run_loopback(struct bnx2x *bp, int loopback_mode, u8 link_up)
         /* prepare the loopback packet */
         pkt_size = (((bp->dev->mtu < ETH_MAX_PACKET_SIZE) ?
                      bp->dev->mtu : ETH_MAX_PACKET_SIZE) + ETH_HLEN);
-        skb = netdev_alloc_skb(bp->dev, bp->rx_buf_size);
+        skb = netdev_alloc_skb(bp->dev, fp_rx->rx_buf_size);
         if (!skb) {
                 rc = -ENOMEM;
                 goto test_loopback_exit;
diff --git a/drivers/net/bnx2x/bnx2x_main.c b/drivers/net/bnx2x/bnx2x_main.c
index 5e3f94878153..722450631302 100644
--- a/drivers/net/bnx2x/bnx2x_main.c
+++ b/drivers/net/bnx2x/bnx2x_main.c
@@ -2473,8 +2473,14 @@ static void bnx2x_pf_rx_cl_prep(struct bnx2x *bp,
         rxq_init->sge_map = fp->rx_sge_mapping;
         rxq_init->rcq_map = fp->rx_comp_mapping;
         rxq_init->rcq_np_map = fp->rx_comp_mapping + BCM_PAGE_SIZE;
-        rxq_init->mtu = bp->dev->mtu;
-        rxq_init->buf_sz = bp->rx_buf_size;
+
+        /* Always use mini-jumbo MTU for FCoE L2 ring */
+        if (IS_FCOE_FP(fp))
+                rxq_init->mtu = BNX2X_FCOE_MINI_JUMBO_MTU;
+        else
+                rxq_init->mtu = bp->dev->mtu;
+
+        rxq_init->buf_sz = fp->rx_buf_size;
         rxq_init->cl_qzone_id = fp->cl_qzone_id;
         rxq_init->cl_id = fp->cl_id;
         rxq_init->spcl_id = fp->cl_id;