[PATCH net-next v4] net: ixp4xx_eth: Support changing the MTU

[PATCH net-next v4] net: ixp4xx_eth: Support changing the MTU

[Linus Walleij]

As we don't specify the MTU in the driver, the framework
will fall back to 1500 bytes and this doesn't work very
well when we try to attach a DSA switch:

  eth1: mtu greater than device maximum
  ixp4xx_eth c800a000.ethernet eth1: error -22 setting
  MTU to 1504 to include DSA overhead

After locating an out-of-tree patch in OpenWrt I found
suitable code to set the MTU on the interface and ported
it and updated it. Now the MTU gets set properly.

Reviewed-by: Jacob Keller <jacob.e.keller@intel.com>
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
---
Changes in v4:
- Rebase on the merged patch "net: ixp4xx_eth: Specify min/max MTU"
  so it can be easily applied.
- Link to v3: https://lore.kernel.org/r/20230928-ixp4xx-eth-mtu-v3-1-cb18eaa0edb9@linaro.org

Changes in v3:
- Fix some coding style and initialization style.
- Collect Jacob's review tag.
- Link to v2: https://lore.kernel.org/r/20230925-ixp4xx-eth-mtu-v2-1-393caab75cb0@linaro.org

Changes in v2:
- Don't just set min/max MTU: implement the interface for actually
  changing it as well.
- Link to v1: https://lore.kernel.org/r/20230923-ixp4xx-eth-mtu-v1-1-9e88b908e1b2@linaro.org
---
 drivers/net/ethernet/xscale/ixp4xx_eth.c | 67 +++++++++++++++++++++++++++++---
 1 file changed, 61 insertions(+), 6 deletions(-)

diff --git a/drivers/net/ethernet/xscale/ixp4xx_eth.c b/drivers/net/ethernet/xscale/ixp4xx_eth.c
index 8f40287c8d58..910094ab553c 100644
--- a/drivers/net/ethernet/xscale/ixp4xx_eth.c
+++ b/drivers/net/ethernet/xscale/ixp4xx_eth.c
@@ -64,7 +64,15 @@
 
 #define POOL_ALLOC_SIZE		(sizeof(struct desc) * (RX_DESCS + TX_DESCS))
 #define REGS_SIZE		0x1000
-#define MAX_MRU			1536 /* 0x600 */
+
+/* MRU is said to be 14320 in a code dump, the SW manual says that
+ * MRU/MTU is 16320 and includes VLAN and ethernet headers.
+ * See "IXP400 Software Programmer's Guide" section 10.3.2, page 161.
+ *
+ * FIXME: we have chosen the safe default (14320) but if you can test
+ * jumboframes, experiment with 16320 and see what happens!
+ */
+#define MAX_MRU			(14320 - VLAN_ETH_HLEN)
 #define RX_BUFF_SIZE		ALIGN((NET_IP_ALIGN) + MAX_MRU, 4)
 
 #define NAPI_WEIGHT		16
@@ -1183,6 +1191,54 @@ static void destroy_queues(struct port *port)
 	}
 }
 
+static int ixp4xx_do_change_mtu(struct net_device *dev, int new_mtu)
+{
+	struct port *port = netdev_priv(dev);
+	struct npe *npe = port->npe;
+	int framesize, chunks;
+	struct msg msg = {};
+
+	/* adjust for ethernet headers */
+	framesize = new_mtu + VLAN_ETH_HLEN;
+	/* max rx/tx 64 byte chunks */
+	chunks = DIV_ROUND_UP(framesize, 64);
+
+	msg.cmd = NPE_SETMAXFRAMELENGTHS;
+	msg.eth_id = port->id;
+
+	/* Firmware wants to know buffer size in 64 byte chunks */
+	msg.byte2 = chunks << 8;
+	msg.byte3 = chunks << 8;
+
+	msg.byte4 = msg.byte6 = framesize >> 8;
+	msg.byte5 = msg.byte7 = framesize & 0xff;
+
+	if (npe_send_recv_message(npe, &msg, "ETH_SET_MAX_FRAME_LENGTH"))
+		return -EIO;
+	netdev_dbg(dev, "set MTU on NPE %s to %d bytes\n",
+		   npe_name(npe), new_mtu);
+
+	return 0;
+}
+
+static int ixp4xx_eth_change_mtu(struct net_device *dev, int new_mtu)
+{
+	int ret;
+
+	/* MTU can only be changed when the interface is up. We also
+	 * set the MTU from dev->mtu when opening the device.
+	 */
+	if (dev->flags & IFF_UP) {
+		ret = ixp4xx_do_change_mtu(dev, new_mtu);
+		if (ret < 0)
+			return ret;
+	}
+
+	dev->mtu = new_mtu;
+
+	return 0;
+}
+
 static int eth_open(struct net_device *dev)
 {
 	struct port *port = netdev_priv(dev);
@@ -1233,6 +1289,8 @@ static int eth_open(struct net_device *dev)
 	if (npe_send_recv_message(port->npe, &msg, "ETH_SET_FIREWALL_MODE"))
 		return -EIO;
 
+	ixp4xx_do_change_mtu(dev, dev->mtu);
+
 	if ((err = request_queues(port)) != 0)
 		return err;
 
@@ -1375,6 +1433,7 @@ static int eth_close(struct net_device *dev)
 static const struct net_device_ops ixp4xx_netdev_ops = {
 	.ndo_open = eth_open,
 	.ndo_stop = eth_close,
+	.ndo_change_mtu = ixp4xx_eth_change_mtu,
 	.ndo_start_xmit = eth_xmit,
 	.ndo_set_rx_mode = eth_set_mcast_list,
 	.ndo_eth_ioctl = eth_ioctl,
@@ -1489,12 +1548,8 @@ static int ixp4xx_eth_probe(struct platform_device *pdev)
 	ndev->dev.dma_mask = dev->dma_mask;
 	ndev->dev.coherent_dma_mask = dev->coherent_dma_mask;
 
-	/* Maximum frame size is 16320 bytes and includes VLAN and
-	 * ethernet headers. See "IXP400 Software Programmer's Guide"
-	 * section 10.3.2, page 161.
-	 */
 	ndev->min_mtu = ETH_MIN_MTU;
-	ndev->max_mtu = 16320 - VLAN_ETH_HLEN;
+	ndev->max_mtu = MAX_MRU;
 
 	netif_napi_add_weight(ndev, &port->napi, eth_poll, NAPI_WEIGHT);
 

---
base-commit: f22d2ec3d4e72d1f24153533ba0ce8fa926cdd37
change-id: 20230923-ixp4xx-eth-mtu-c041d7efe932

Best regards,
-- 
Linus Walleij <linus.walleij@linaro.org>

Re: [PATCH net-next v4] net: ixp4xx_eth: Support changing the MTU

[patchwork-bot+netdevbpf]

Hello:

This patch was applied to netdev/net-next.git (main)
by Jakub Kicinski <kuba@kernel.org>:

On Thu, 05 Oct 2023 00:43:53 +0200 you wrote:
> As we don't specify the MTU in the driver, the framework
> will fall back to 1500 bytes and this doesn't work very
> well when we try to attach a DSA switch:
> 
>   eth1: mtu greater than device maximum
>   ixp4xx_eth c800a000.ethernet eth1: error -22 setting
>   MTU to 1504 to include DSA overhead
> 
> [...]

Here is the summary with links:
  - [net-next,v4] net: ixp4xx_eth: Support changing the MTU
    https://git.kernel.org/netdev/net-next/c/4f08c2570239

You are awesome, thank you!
-- 
Deet-doot-dot, I am a bot.
https://korg.docs.kernel.org/patchwork/pwbot.html