[next,S63,6/6] i40e/i40evf: Change the way we limit the maximum frame size for Rx

Message ID 1489511727-10959-6-git-send-email-bimmy.pujari@intel.com
State Accepted
Delegated to: Jeff Kirsher
Headers show

Commit Message

Bimmy Pujari March 14, 2017, 5:15 p.m.
From: Alexander Duyck <alexander.h.duyck@intel.com>

This patch changes the way we handle the maximum frame size for the Rx
path.  Previously we were rounding up to 2K for a 1500 MTU and then brining
the max frame size down to MTU plus a fixed amount.  With this patch
applied what we now do is limit the maximum frame to 1.5K minus the value
for NET_IP_ALIGN for standard MTU, and for any MTU greater than 1500 we
allow up to the maximum frame size.  This makes the behavior more
consistent with the other drivers such as igb which had similar logic.  In
addition it reduces the test matrix for MTU since we only have two max
frame sizes that are handled for Rx now.

Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Change-ID: I23a9d3c857e7df04b0ef28c64df63e659c013f3f
---
Testing Hints:
        Verify MTU handling for MTU = 1500.  You should be able to receive
        frames up to 1536 in size on a standard x86 system as NET_IP_ALIGN
        is 0.

        When jumbo frames are enabled you should be able to receive anything
        up to the maximum size supported by hardware which is 9.5K or 9726B
        in size.

        When the "legacy-rx" private flag is set you should be able to
        receive anything up to 9.5K in size.

        The behavior for MTU mismatches will change.  A failing case is any
        in which we trigger a kernel panic.  Receiving a frame larger than
        what was configured for the MTU is not a failure. 

 drivers/net/ethernet/intel/i40e/i40e_main.c        | 26 ++++++++++++----------
 drivers/net/ethernet/intel/i40e/i40e_txrx.h        |  4 +---
 drivers/net/ethernet/intel/i40evf/i40e_txrx.h      |  4 +---
 drivers/net/ethernet/intel/i40evf/i40evf.h         |  4 ----
 drivers/net/ethernet/intel/i40evf/i40evf_main.c    | 16 ++++++++++++-
 .../net/ethernet/intel/i40evf/i40evf_virtchnl.c    | 14 ++++++++----
 6 files changed, 41 insertions(+), 27 deletions(-)

Comments

Bowers, AndrewX March 21, 2017, 4:46 p.m. | #1
> -----Original Message-----
> From: Intel-wired-lan [mailto:intel-wired-lan-bounces@lists.osuosl.org] On
> Behalf Of Bimmy Pujari
> Sent: Tuesday, March 14, 2017 10:15 AM
> To: intel-wired-lan@lists.osuosl.org
> Subject: [Intel-wired-lan] [next PATCH S63 6/6] i40e/i40evf: Change the way
> we limit the maximum frame size for Rx
> 
> From: Alexander Duyck <alexander.h.duyck@intel.com>
> 
> This patch changes the way we handle the maximum frame size for the Rx
> path.  Previously we were rounding up to 2K for a 1500 MTU and then brining
> the max frame size down to MTU plus a fixed amount.  With this patch
> applied what we now do is limit the maximum frame to 1.5K minus the value
> for NET_IP_ALIGN for standard MTU, and for any MTU greater than 1500 we
> allow up to the maximum frame size.  This makes the behavior more
> consistent with the other drivers such as igb which had similar logic.  In
> addition it reduces the test matrix for MTU since we only have two max
> frame sizes that are handled for Rx now.
> 
> Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
> Change-ID: I23a9d3c857e7df04b0ef28c64df63e659c013f3f
> ---
> Testing Hints:
>         Verify MTU handling for MTU = 1500.  You should be able to receive
>         frames up to 1536 in size on a standard x86 system as NET_IP_ALIGN
>         is 0.
> 
>         When jumbo frames are enabled you should be able to receive anything
>         up to the maximum size supported by hardware which is 9.5K or 9726B
>         in size.
> 
>         When the "legacy-rx" private flag is set you should be able to
>         receive anything up to 9.5K in size.
> 
>         The behavior for MTU mismatches will change.  A failing case is any
>         in which we trigger a kernel panic.  Receiving a frame larger than
>         what was configured for the MTU is not a failure.
> 
>  drivers/net/ethernet/intel/i40e/i40e_main.c        | 26 ++++++++++++---------
> -
>  drivers/net/ethernet/intel/i40e/i40e_txrx.h        |  4 +---
>  drivers/net/ethernet/intel/i40evf/i40e_txrx.h      |  4 +---
>  drivers/net/ethernet/intel/i40evf/i40evf.h         |  4 ----
>  drivers/net/ethernet/intel/i40evf/i40evf_main.c    | 16 ++++++++++++-
>  .../net/ethernet/intel/i40evf/i40evf_virtchnl.c    | 14 ++++++++----
>  6 files changed, 41 insertions(+), 27 deletions(-)

Tested-by: Andrew Bowers <andrewx.bowers@intel.com>

Patch

diff --git a/drivers/net/ethernet/intel/i40e/i40e_main.c b/drivers/net/ethernet/intel/i40e/i40e_main.c
index 4cfe9491..724a5a4 100644
--- a/drivers/net/ethernet/intel/i40e/i40e_main.c
+++ b/drivers/net/ethernet/intel/i40e/i40e_main.c
@@ -3002,7 +3002,8 @@  static int i40e_configure_rx_ring(struct i40e_ring *ring)
 
 	ring->rx_buf_len = vsi->rx_buf_len;
 
-	rx_ctx.dbuff = ring->rx_buf_len >> I40E_RXQ_CTX_DBUFF_SHIFT;
+	rx_ctx.dbuff = DIV_ROUND_UP(ring->rx_buf_len,
+				    BIT_ULL(I40E_RXQ_CTX_DBUFF_SHIFT));
 
 	rx_ctx.base = (ring->dma / 128);
 	rx_ctx.qlen = ring->count;
@@ -3082,17 +3083,18 @@  static int i40e_vsi_configure_rx(struct i40e_vsi *vsi)
 	int err = 0;
 	u16 i;
 
-	if (vsi->netdev && (vsi->netdev->mtu > ETH_DATA_LEN))
-		vsi->max_frame = vsi->netdev->mtu + ETH_HLEN
-			       + ETH_FCS_LEN + VLAN_HLEN;
-	else
-		vsi->max_frame = I40E_RXBUFFER_2048;
-
-	vsi->rx_buf_len = I40E_RXBUFFER_2048;
-
-	/* round up for the chip's needs */
-	vsi->rx_buf_len = ALIGN(vsi->rx_buf_len,
-				BIT_ULL(I40E_RXQ_CTX_DBUFF_SHIFT));
+	if (!vsi->netdev || (vsi->back->flags & I40E_FLAG_LEGACY_RX)) {
+		vsi->max_frame = I40E_MAX_RXBUFFER;
+		vsi->rx_buf_len = I40E_RXBUFFER_2048;
+#if (PAGE_SIZE < 8192)
+	} else if (vsi->netdev->mtu <= ETH_DATA_LEN) {
+		vsi->max_frame = I40E_RXBUFFER_1536 - NET_IP_ALIGN;
+		vsi->rx_buf_len = I40E_RXBUFFER_1536 - NET_IP_ALIGN;
+#endif
+	} else {
+		vsi->max_frame = I40E_MAX_RXBUFFER;
+		vsi->rx_buf_len = I40E_RXBUFFER_2048;
+	}
 
 	/* set up individual rings */
 	for (i = 0; i < vsi->num_queue_pairs && !err; i++)
diff --git a/drivers/net/ethernet/intel/i40e/i40e_txrx.h b/drivers/net/ethernet/intel/i40e/i40e_txrx.h
index eb73372..d6609de 100644
--- a/drivers/net/ethernet/intel/i40e/i40e_txrx.h
+++ b/drivers/net/ethernet/intel/i40e/i40e_txrx.h
@@ -117,10 +117,8 @@  enum i40e_dyn_idx_t {
 
 /* Supported Rx Buffer Sizes (a multiple of 128) */
 #define I40E_RXBUFFER_256   256
+#define I40E_RXBUFFER_1536  1536  /* 128B aligned standard Ethernet frame */
 #define I40E_RXBUFFER_2048  2048
-#define I40E_RXBUFFER_3072  3072   /* For FCoE MTU of 2158 */
-#define I40E_RXBUFFER_4096  4096
-#define I40E_RXBUFFER_8192  8192
 #define I40E_MAX_RXBUFFER   9728  /* largest size for single descriptor */
 
 /* NOTE: netdev_alloc_skb reserves up to 64 bytes, NET_IP_ALIGN means we
diff --git a/drivers/net/ethernet/intel/i40evf/i40e_txrx.h b/drivers/net/ethernet/intel/i40evf/i40e_txrx.h
index aba40ed..3bb4d73 100644
--- a/drivers/net/ethernet/intel/i40evf/i40e_txrx.h
+++ b/drivers/net/ethernet/intel/i40evf/i40e_txrx.h
@@ -104,10 +104,8 @@  enum i40e_dyn_idx_t {
 
 /* Supported Rx Buffer Sizes (a multiple of 128) */
 #define I40E_RXBUFFER_256   256
+#define I40E_RXBUFFER_1536  1536  /* 128B aligned standard Ethernet frame */
 #define I40E_RXBUFFER_2048  2048
-#define I40E_RXBUFFER_3072  3072   /* For FCoE MTU of 2158 */
-#define I40E_RXBUFFER_4096  4096
-#define I40E_RXBUFFER_8192  8192
 #define I40E_MAX_RXBUFFER   9728  /* largest size for single descriptor */
 
 /* NOTE: netdev_alloc_skb reserves up to 64 bytes, NET_IP_ALIGN means we
diff --git a/drivers/net/ethernet/intel/i40evf/i40evf.h b/drivers/net/ethernet/intel/i40evf/i40evf.h
index e60cbfa..d61ecf6 100644
--- a/drivers/net/ethernet/intel/i40evf/i40evf.h
+++ b/drivers/net/ethernet/intel/i40evf/i40evf.h
@@ -72,10 +72,6 @@  struct i40e_vsi {
 #define I40EVF_MAX_RXD		4096
 #define I40EVF_MIN_RXD		64
 #define I40EVF_REQ_DESCRIPTOR_MULTIPLE	32
-
-/* Supported Rx Buffer Sizes */
-#define I40EVF_RXBUFFER_2048	2048
-#define I40EVF_MAX_RXBUFFER	16384  /* largest size for single descriptor */
 #define I40EVF_MAX_AQ_BUF_SIZE	4096
 #define I40EVF_AQ_LEN		32
 #define I40EVF_AQ_MAX_ERR	20 /* times to try before resetting AQ */
diff --git a/drivers/net/ethernet/intel/i40evf/i40evf_main.c b/drivers/net/ethernet/intel/i40evf/i40evf_main.c
index 6d666bd..fb2811c 100644
--- a/drivers/net/ethernet/intel/i40evf/i40evf_main.c
+++ b/drivers/net/ethernet/intel/i40evf/i40evf_main.c
@@ -686,12 +686,26 @@  static void i40evf_configure_tx(struct i40evf_adapter *adapter)
  **/
 static void i40evf_configure_rx(struct i40evf_adapter *adapter)
 {
+	unsigned int rx_buf_len = I40E_RXBUFFER_2048;
+	struct net_device *netdev = adapter->netdev;
 	struct i40e_hw *hw = &adapter->hw;
 	int i;
 
+	/* Legacy Rx will always default to a 2048 buffer size. */
+#if (PAGE_SIZE < 8192)
+	if (!(adapter->flags & I40EVF_FLAG_LEGACY_RX)) {
+		/* We use a 1536 buffer size for configurations with
+		 * standard Ethernet mtu.  On x86 this gives us enough room
+		 * for shared info and 192 bytes of padding.
+		 */
+		if (netdev->mtu <= ETH_DATA_LEN)
+			rx_buf_len = I40E_RXBUFFER_1536 - NET_IP_ALIGN;
+	}
+#endif
+
 	for (i = 0; i < adapter->num_active_queues; i++) {
 		adapter->rx_rings[i].tail = hw->hw_addr + I40E_QRX_TAIL1(i);
-		adapter->rx_rings[i].rx_buf_len = I40EVF_RXBUFFER_2048;
+		adapter->rx_rings[i].rx_buf_len = rx_buf_len;
 	}
 }
 
diff --git a/drivers/net/ethernet/intel/i40evf/i40evf_virtchnl.c b/drivers/net/ethernet/intel/i40evf/i40evf_virtchnl.c
index 4bc2488..032be8d 100644
--- a/drivers/net/ethernet/intel/i40evf/i40evf_virtchnl.c
+++ b/drivers/net/ethernet/intel/i40evf/i40evf_virtchnl.c
@@ -234,7 +234,7 @@  void i40evf_configure_queues(struct i40evf_adapter *adapter)
 	struct i40e_virtchnl_vsi_queue_config_info *vqci;
 	struct i40e_virtchnl_queue_pair_info *vqpi;
 	int pairs = adapter->num_active_queues;
-	int i, len;
+	int i, len, max_frame = I40E_MAX_RXBUFFER;
 
 	if (adapter->current_op != I40E_VIRTCHNL_OP_UNKNOWN) {
 		/* bail because we already have a command pending */
@@ -249,6 +249,11 @@  void i40evf_configure_queues(struct i40evf_adapter *adapter)
 	if (!vqci)
 		return;
 
+	/* Limit maximum frame size when jumbo frames is not enabled */
+	if (!(adapter->flags & I40EVF_FLAG_LEGACY_RX) &&
+	    (adapter->netdev->mtu <= ETH_DATA_LEN))
+		max_frame = I40E_RXBUFFER_1536 - NET_IP_ALIGN;
+
 	vqci->vsi_id = adapter->vsi_res->vsi_id;
 	vqci->num_queue_pairs = pairs;
 	vqpi = vqci->qpair;
@@ -264,9 +269,10 @@  void i40evf_configure_queues(struct i40evf_adapter *adapter)
 		vqpi->rxq.queue_id = i;
 		vqpi->rxq.ring_len = adapter->rx_rings[i].count;
 		vqpi->rxq.dma_ring_addr = adapter->rx_rings[i].dma;
-		vqpi->rxq.max_pkt_size = adapter->netdev->mtu
-					+ ETH_HLEN + VLAN_HLEN + ETH_FCS_LEN;
-		vqpi->rxq.databuffer_size = adapter->rx_rings[i].rx_buf_len;
+		vqpi->rxq.max_pkt_size = max_frame;
+		vqpi->rxq.databuffer_size =
+			ALIGN(adapter->rx_rings[i].rx_buf_len,
+			      BIT_ULL(I40E_RXQ_CTX_DBUFF_SHIFT));
 		vqpi++;
 	}