[PATCH 1/5] net: ethernet: mtk_eth_soc: account for vlan in rx header length

[PATCH 1/5] net: ethernet: mtk_eth_soc: account for vlan in rx header length

[Felix Fietkau]

The network stack assumes that devices can handle an extra VLAN tag without
increasing the MTU

Reviewed-by: Vladimir Oltean <olteanv@gmail.com>
Signed-off-by: Felix Fietkau <nbd@nbd.name>
---
 drivers/net/ethernet/mediatek/mtk_eth_soc.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/net/ethernet/mediatek/mtk_eth_soc.h b/drivers/net/ethernet/mediatek/mtk_eth_soc.h
index 7ca806b4de10..a553265c10de 100644
--- a/drivers/net/ethernet/mediatek/mtk_eth_soc.h
+++ b/drivers/net/ethernet/mediatek/mtk_eth_soc.h
@@ -34,7 +34,7 @@
 #define MTK_QDMA_RING_SIZE	2048
 #define MTK_DMA_SIZE		512
 #define MTK_MAC_COUNT		2
-#define MTK_RX_ETH_HLEN		(ETH_HLEN + ETH_FCS_LEN)
+#define MTK_RX_ETH_HLEN		(VLAN_ETH_HLEN + ETH_FCS_LEN)
 #define MTK_RX_HLEN		(NET_SKB_PAD + MTK_RX_ETH_HLEN + NET_IP_ALIGN)
 #define MTK_DMA_DUMMY_DESC	0xffffffff
 #define MTK_DEFAULT_MSG_ENABLE	(NETIF_MSG_DRV | \
-- 
2.38.1


_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

[PATCH 2/5] net: ethernet: mtk_eth_soc: compile out netsys v2 code on mt7621

[Felix Fietkau]

Avoid some branches in the hot path on low-end devices with limited CPU power,
and reduce code size

Signed-off-by: Felix Fietkau <nbd@nbd.name>
---
 drivers/net/ethernet/mediatek/mtk_eth_soc.h | 8 +++++++-
 1 file changed, 7 insertions(+), 1 deletion(-)

diff --git a/drivers/net/ethernet/mediatek/mtk_eth_soc.h b/drivers/net/ethernet/mediatek/mtk_eth_soc.h
index a553265c10de..628ed72911bd 100644
--- a/drivers/net/ethernet/mediatek/mtk_eth_soc.h
+++ b/drivers/net/ethernet/mediatek/mtk_eth_soc.h
@@ -907,7 +907,13 @@ enum mkt_eth_capabilities {
 #define MTK_MUX_GMAC12_TO_GEPHY_SGMII   \
 	(MTK_ETH_MUX_GMAC12_TO_GEPHY_SGMII | MTK_MUX)
 
-#define MTK_HAS_CAPS(caps, _x)		(((caps) & (_x)) == (_x))
+#ifdef CONFIG_SOC_MT7621
+#define MTK_CAP_MASK MTK_NETSYS_V2
+#else
+#define MTK_CAP_MASK 0
+#endif
+
+#define MTK_HAS_CAPS(caps, _x)		(((caps) & (_x) & ~(MTK_CAP_MASK)) == (_x))
 
 #define MT7621_CAPS  (MTK_GMAC1_RGMII | MTK_GMAC1_TRGMII | \
 		      MTK_GMAC2_RGMII | MTK_SHARED_INT | \
-- 
2.38.1


_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

[PATCH 3/5] net: ethernet: mtk_eth_soc: work around issue with sending small fragments

[Felix Fietkau]

When frames are sent with very small fragments, the DMA engine appears to
lock up and transmit attempts time out. Fix this by detecting the presence
of small fragments and use skb_gso_segment + skb_linearize to deal with
them

Signed-off-by: Felix Fietkau <nbd@nbd.name>
---
 drivers/net/ethernet/mediatek/mtk_eth_soc.c | 36 +++++++++++++++++++--
 1 file changed, 34 insertions(+), 2 deletions(-)

diff --git a/drivers/net/ethernet/mediatek/mtk_eth_soc.c b/drivers/net/ethernet/mediatek/mtk_eth_soc.c
index 4c9972a94451..e63d2c034ca3 100644
--- a/drivers/net/ethernet/mediatek/mtk_eth_soc.c
+++ b/drivers/net/ethernet/mediatek/mtk_eth_soc.c
@@ -1441,12 +1441,28 @@ static void mtk_wake_queue(struct mtk_eth *eth)
 	}
 }
 
+static bool mtk_skb_has_small_frag(struct sk_buff *skb)
+{
+	int min_size = 16;
+	int i;
+
+	if (skb_headlen(skb) < min_size)
+		return true;
+
+	for (i = 0; i < skb_shinfo(skb)->nr_frags; i++)
+		if (skb_frag_size(&skb_shinfo(skb)->frags[i]) < min_size)
+			return true;
+
+	return false;
+}
+
 static netdev_tx_t mtk_start_xmit(struct sk_buff *skb, struct net_device *dev)
 {
 	struct mtk_mac *mac = netdev_priv(dev);
 	struct mtk_eth *eth = mac->hw;
 	struct mtk_tx_ring *ring = &eth->tx_ring;
 	struct net_device_stats *stats = &dev->stats;
+	struct sk_buff *segs, *next;
 	bool gso = false;
 	int tx_num;
 
@@ -1468,6 +1484,17 @@ static netdev_tx_t mtk_start_xmit(struct sk_buff *skb, struct net_device *dev)
 		return NETDEV_TX_BUSY;
 	}
 
+	if (skb_is_gso(skb) && mtk_skb_has_small_frag(skb)) {
+		segs = skb_gso_segment(skb, dev->features & ~NETIF_F_ALL_TSO);
+		if (IS_ERR(segs))
+			goto drop;
+
+		if (segs) {
+			consume_skb(skb);
+			skb = segs;
+		}
+	}
+
 	/* TSO: fill MSS info in tcp checksum field */
 	if (skb_is_gso(skb)) {
 		if (skb_cow_head(skb, 0)) {
@@ -1483,8 +1510,13 @@ static netdev_tx_t mtk_start_xmit(struct sk_buff *skb, struct net_device *dev)
 		}
 	}
 
-	if (mtk_tx_map(skb, dev, tx_num, ring, gso) < 0)
-		goto drop;
+	skb_list_walk_safe(skb, skb, next) {
+		if ((mtk_skb_has_small_frag(skb) && skb_linearize(skb)) ||
+		    mtk_tx_map(skb, dev, tx_num, ring, gso) < 0) {
+			stats->tx_dropped++;
+			dev_kfree_skb_any(skb);
+		}
+	}
 
 	if (unlikely(atomic_read(&ring->free_count) <= ring->thresh))
 		netif_tx_stop_all_queues(dev);
-- 
2.38.1


_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

Re: [PATCH 3/5] net: ethernet: mtk_eth_soc: work around issue with sending small fragments

[Alexander Lobakin]

From: Felix Fietkau <nbd@nbd.name>
Date: Wed, 23 Nov 2022 10:57:52 +0100

> When frames are sent with very small fragments, the DMA engine appears to
> lock up and transmit attempts time out. Fix this by detecting the presence
> of small fragments and use skb_gso_segment + skb_linearize to deal with
> them

Nit: all of your commit messages don't have a trailing dot (.), not
sure if it's important, but my eye is missing it definitely :D

skb_gso_segment() and skb_linearize() are slow as hell. I think you
can do it differently. I guess only the first (head) and the last
frag can be so small, right?

So, if a frag from shinfo->frags is less than 16, get a new frag of
the minimum acceptable size via netdev_alloc_frag(), copy the data
to it and pad the rest with zeroes. Then increase skb->len and
skb->data_len, skb_frag_unref() the current, "invalid" frag and
replace the pointer to the new frag. I didn't miss anything I
believe... Zero padding the tail is usual thing for NICs. skb frag
substitution is less common, but should be legit.

If skb_headlen() is less than 16, try doing pskb_may_pull() +
__skb_pull() at first. The argument would be `16 - headlen`. If
pskb_may_pull() returns false, then yeah, you have no choice other
than segmenting and linearizing ._.

> 
> Signed-off-by: Felix Fietkau <nbd@nbd.name>
> ---
>  drivers/net/ethernet/mediatek/mtk_eth_soc.c | 36 +++++++++++++++++++--
>  1 file changed, 34 insertions(+), 2 deletions(-)

[...]

>  	if (unlikely(atomic_read(&ring->free_count) <= ring->thresh))
>  		netif_tx_stop_all_queues(dev);
> -- 
> 2.38.1

Thanks,
Olek

_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

Re: [PATCH 3/5] net: ethernet: mtk_eth_soc: work around issue with sending small fragments

[Felix Fietkau]

On 24.11.22 18:54, Alexander Lobakin wrote:
> From: Felix Fietkau <nbd@nbd.name>
> Date: Wed, 23 Nov 2022 10:57:52 +0100
> 
>> When frames are sent with very small fragments, the DMA engine appears to
>> lock up and transmit attempts time out. Fix this by detecting the presence
>> of small fragments and use skb_gso_segment + skb_linearize to deal with
>> them
> 
> Nit: all of your commit messages don't have a trailing dot (.), not
> sure if it's important, but my eye is missing it definitely :D
> 
> skb_gso_segment() and skb_linearize() are slow as hell. I think you
> can do it differently. I guess only the first (head) and the last
> frag can be so small, right?
> 
> So, if a frag from shinfo->frags is less than 16, get a new frag of
> the minimum acceptable size via netdev_alloc_frag(), copy the data
> to it and pad the rest with zeroes. Then increase skb->len and
> skb->data_len, skb_frag_unref() the current, "invalid" frag and
> replace the pointer to the new frag. I didn't miss anything I
> believe... Zero padding the tail is usual thing for NICs. skb frag
> substitution is less common, but should be legit.
> 
> If skb_headlen() is less than 16, try doing pskb_may_pull() +
> __skb_pull() at first. The argument would be `16 - headlen`. If
> pskb_may_pull() returns false, then yeah, you have no choice other
> than segmenting and linearizing ._.
I looked into this some more and spoke with people at MTK. It appears 
that in principle, the DMA engine is able to process very small 
fragments. However, when it is being flooded with them, a FIFO can 
overflow, which causes the hang that I was observing.
I think your suggestion likely would not fix the issue completely.
A MTK engineer also confirmed that my approach is the correct one for 
handling this.
I will send v2 with an updated description.

Thanks,

- Felix


_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

[PATCH 4/5] net: ethernet: mtk_eth_soc: fix flow_offload related refcount bug

[Felix Fietkau]

Since we call flow_block_cb_decref on FLOW_BLOCK_UNBIND, we need to call
flow_block_cb_incref unconditionally, even for a newly allocated cb.
Fixes a use-after-free bug

Fixes: 502e84e2382d ("net: ethernet: mtk_eth_soc: add flow offloading support")
Signed-off-by: Felix Fietkau <nbd@nbd.name>
---
 .../net/ethernet/mediatek/mtk_ppe_offload.c   | 21 ++++++++++++-------
 1 file changed, 13 insertions(+), 8 deletions(-)

diff --git a/drivers/net/ethernet/mediatek/mtk_ppe_offload.c b/drivers/net/ethernet/mediatek/mtk_ppe_offload.c
index 81afd5ee3fbf..412d215b9571 100644
--- a/drivers/net/ethernet/mediatek/mtk_ppe_offload.c
+++ b/drivers/net/ethernet/mediatek/mtk_ppe_offload.c
@@ -554,6 +554,7 @@ mtk_eth_setup_tc_block(struct net_device *dev, struct flow_block_offload *f)
 	struct mtk_eth *eth = mac->hw;
 	static LIST_HEAD(block_cb_list);
 	struct flow_block_cb *block_cb;
+	bool register_block = false;
 	flow_setup_cb_t *cb;
 
 	if (!eth->soc->offload_version)
@@ -568,16 +569,20 @@ mtk_eth_setup_tc_block(struct net_device *dev, struct flow_block_offload *f)
 	switch (f->command) {
 	case FLOW_BLOCK_BIND:
 		block_cb = flow_block_cb_lookup(f->block, cb, dev);
-		if (block_cb) {
-			flow_block_cb_incref(block_cb);
-			return 0;
+		if (!block_cb) {
+			block_cb = flow_block_cb_alloc(cb, dev, dev, NULL);
+			if (IS_ERR(block_cb))
+				return PTR_ERR(block_cb);
+
+			register_block = true;
 		}
-		block_cb = flow_block_cb_alloc(cb, dev, dev, NULL);
-		if (IS_ERR(block_cb))
-			return PTR_ERR(block_cb);
 
-		flow_block_cb_add(block_cb, f);
-		list_add_tail(&block_cb->driver_list, &block_cb_list);
+		flow_block_cb_incref(block_cb);
+
+		if (register_block) {
+			flow_block_cb_add(block_cb, f);
+			list_add_tail(&block_cb->driver_list, &block_cb_list);
+		}
 		return 0;
 	case FLOW_BLOCK_UNBIND:
 		block_cb = flow_block_cb_lookup(f->block, cb, dev);
-- 
2.38.1


_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

[PATCH 5/5] net: ethernet: mtk_eth_soc: drop generic vlan rx offload, only use DSA untagging

[Felix Fietkau]

Through testing I found out that hardware vlan rx offload support seems to
have some hardware issues. At least when using multiple MACs and when receiving
tagged packets on the secondary MAC, the hardware can sometimes start to emit
wrong tags on the first MAC as well.

In order to avoid such issues, drop the feature configuration and use the
offload feature only for DSA hardware untagging on MT7621/MT7622 devices which
only use one MAC.

Signed-off-by: Felix Fietkau <nbd@nbd.name>
---
 drivers/net/ethernet/mediatek/mtk_eth_soc.c | 101 ++++++++------------
 drivers/net/ethernet/mediatek/mtk_eth_soc.h |   1 -
 2 files changed, 39 insertions(+), 63 deletions(-)

diff --git a/drivers/net/ethernet/mediatek/mtk_eth_soc.c b/drivers/net/ethernet/mediatek/mtk_eth_soc.c
index e63d2c034ca3..d5843660f259 100644
--- a/drivers/net/ethernet/mediatek/mtk_eth_soc.c
+++ b/drivers/net/ethernet/mediatek/mtk_eth_soc.c
@@ -2040,29 +2040,16 @@ static int mtk_poll_rx(struct napi_struct *napi, int budget,
 		if (reason == MTK_PPE_CPU_REASON_HIT_UNBIND_RATE_REACHED)
 			mtk_ppe_check_skb(eth->ppe[0], skb, hash);
 
-		if (netdev->features & NETIF_F_HW_VLAN_CTAG_RX) {
-			if (MTK_HAS_CAPS(eth->soc->caps, MTK_NETSYS_V2)) {
-				if (trxd.rxd3 & RX_DMA_VTAG_V2)
-					__vlan_hwaccel_put_tag(skb,
-						htons(RX_DMA_VPID(trxd.rxd4)),
-						RX_DMA_VID(trxd.rxd4));
-			} else if (trxd.rxd2 & RX_DMA_VTAG) {
-				__vlan_hwaccel_put_tag(skb, htons(RX_DMA_VPID(trxd.rxd3)),
-						       RX_DMA_VID(trxd.rxd3));
-			}
-		}
-
 		/* When using VLAN untagging in combination with DSA, the
 		 * hardware treats the MTK special tag as a VLAN and untags it.
 		 */
-		if (skb_vlan_tag_present(skb) && netdev_uses_dsa(netdev)) {
-			unsigned int port = ntohs(skb->vlan_proto) & GENMASK(2, 0);
+		if (!MTK_HAS_CAPS(eth->soc->caps, MTK_NETSYS_V2) &&
+		    (trxd.rxd2 & RX_DMA_VTAG) && netdev_uses_dsa(netdev)) {
+			unsigned int port = RX_DMA_VPID(trxd.rxd3) & GENMASK(2, 0);
 
 			if (port < ARRAY_SIZE(eth->dsa_meta) &&
 			    eth->dsa_meta[port])
 				skb_dst_set_noref(skb, &eth->dsa_meta[port]->dst);
-
-			__vlan_hwaccel_clear_tag(skb);
 		}
 
 		skb_record_rx_queue(skb, 0);
@@ -2885,29 +2872,11 @@ static netdev_features_t mtk_fix_features(struct net_device *dev,
 
 static int mtk_set_features(struct net_device *dev, netdev_features_t features)
 {
-	struct mtk_mac *mac = netdev_priv(dev);
-	struct mtk_eth *eth = mac->hw;
 	netdev_features_t diff = dev->features ^ features;
-	int i;
 
 	if ((diff & NETIF_F_LRO) && !(features & NETIF_F_LRO))
 		mtk_hwlro_netdev_disable(dev);
 
-	/* Set RX VLAN offloading */
-	if (!(diff & NETIF_F_HW_VLAN_CTAG_RX))
-		return 0;
-
-	mtk_w32(eth, !!(features & NETIF_F_HW_VLAN_CTAG_RX),
-		MTK_CDMP_EG_CTRL);
-
-	/* sync features with other MAC */
-	for (i = 0; i < MTK_MAC_COUNT; i++) {
-		if (!eth->netdev[i] || eth->netdev[i] == dev)
-			continue;
-		eth->netdev[i]->features &= ~NETIF_F_HW_VLAN_CTAG_RX;
-		eth->netdev[i]->features |= features & NETIF_F_HW_VLAN_CTAG_RX;
-	}
-
 	return 0;
 }
 
@@ -3207,30 +3176,6 @@ static int mtk_open(struct net_device *dev)
 	struct mtk_eth *eth = mac->hw;
 	int i, err;
 
-	if (mtk_uses_dsa(dev) && !eth->prog) {
-		for (i = 0; i < ARRAY_SIZE(eth->dsa_meta); i++) {
-			struct metadata_dst *md_dst = eth->dsa_meta[i];
-
-			if (md_dst)
-				continue;
-
-			md_dst = metadata_dst_alloc(0, METADATA_HW_PORT_MUX,
-						    GFP_KERNEL);
-			if (!md_dst)
-				return -ENOMEM;
-
-			md_dst->u.port_info.port_id = i;
-			eth->dsa_meta[i] = md_dst;
-		}
-	} else {
-		/* Hardware special tag parsing needs to be disabled if at least
-		 * one MAC does not use DSA.
-		 */
-		u32 val = mtk_r32(eth, MTK_CDMP_IG_CTRL);
-		val &= ~MTK_CDMP_STAG_EN;
-		mtk_w32(eth, val, MTK_CDMP_IG_CTRL);
-	}
-
 	err = phylink_of_phy_connect(mac->phylink, mac->of_node, 0);
 	if (err) {
 		netdev_err(dev, "%s: could not attach PHY: %d\n", __func__,
@@ -3267,6 +3212,39 @@ static int mtk_open(struct net_device *dev)
 	phylink_start(mac->phylink);
 	netif_tx_start_all_queues(dev);
 
+	if (MTK_HAS_CAPS(eth->soc->caps, MTK_NETSYS_V2))
+		return 0;
+
+	if (mtk_uses_dsa(dev) && !eth->prog) {
+		for (i = 0; i < ARRAY_SIZE(eth->dsa_meta); i++) {
+			struct metadata_dst *md_dst = eth->dsa_meta[i];
+
+			if (md_dst)
+				continue;
+
+			md_dst = metadata_dst_alloc(0, METADATA_HW_PORT_MUX,
+						    GFP_KERNEL);
+			if (!md_dst)
+				return -ENOMEM;
+
+			md_dst->u.port_info.port_id = i;
+			eth->dsa_meta[i] = md_dst;
+		}
+	} else {
+		/* Hardware special tag parsing needs to be disabled if at least
+		 * one MAC does not use DSA.
+		 */
+		u32 val = mtk_r32(eth, MTK_CDMP_IG_CTRL);
+		val &= ~MTK_CDMP_STAG_EN;
+		mtk_w32(eth, val, MTK_CDMP_IG_CTRL);
+
+		val = mtk_r32(eth, MTK_CDMQ_IG_CTRL);
+		val &= ~MTK_CDMQ_STAG_EN;
+		mtk_w32(eth, val, MTK_CDMQ_IG_CTRL);
+
+		mtk_w32(eth, 0, MTK_CDMP_EG_CTRL);
+	}
+
 	return 0;
 }
 
@@ -3593,10 +3571,9 @@ static int mtk_hw_init(struct mtk_eth *eth)
 	if (!MTK_HAS_CAPS(eth->soc->caps, MTK_NETSYS_V2)) {
 		val = mtk_r32(eth, MTK_CDMP_IG_CTRL);
 		mtk_w32(eth, val | MTK_CDMP_STAG_EN, MTK_CDMP_IG_CTRL);
-	}
 
-	/* Enable RX VLan Offloading */
-	mtk_w32(eth, 1, MTK_CDMP_EG_CTRL);
+		mtk_w32(eth, 1, MTK_CDMP_EG_CTRL);
+	}
 
 	/* set interrupt delays based on current Net DIM sample */
 	mtk_dim_rx(&eth->rx_dim.work);
@@ -4197,7 +4174,7 @@ static int mtk_add_mac(struct mtk_eth *eth, struct device_node *np)
 		eth->netdev[id]->hw_features |= NETIF_F_LRO;
 
 	eth->netdev[id]->vlan_features = eth->soc->hw_features &
-		~(NETIF_F_HW_VLAN_CTAG_TX | NETIF_F_HW_VLAN_CTAG_RX);
+		~NETIF_F_HW_VLAN_CTAG_TX;
 	eth->netdev[id]->features |= eth->soc->hw_features;
 	eth->netdev[id]->ethtool_ops = &mtk_ethtool_ops;
 
diff --git a/drivers/net/ethernet/mediatek/mtk_eth_soc.h b/drivers/net/ethernet/mediatek/mtk_eth_soc.h
index 628ed72911bd..fa1907b62ffd 100644
--- a/drivers/net/ethernet/mediatek/mtk_eth_soc.h
+++ b/drivers/net/ethernet/mediatek/mtk_eth_soc.h
@@ -48,7 +48,6 @@
 #define MTK_HW_FEATURES		(NETIF_F_IP_CSUM | \
 				 NETIF_F_RXCSUM | \
 				 NETIF_F_HW_VLAN_CTAG_TX | \
-				 NETIF_F_HW_VLAN_CTAG_RX | \
 				 NETIF_F_SG | NETIF_F_ALL_TSO | \
 				 NETIF_F_IPV6_CSUM |\
 				 NETIF_F_HW_TC)
-- 
2.38.1


_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

Aw: [PATCH 5/5] net: ethernet: mtk_eth_soc: drop generic vlan rx offload, only use DSA untagging

[Frank Wunderlich]

Hi,

tested this series on 6.1-next and it fixes my vlan-issues on gmac (Bananapi-R3 sfp-wan). maybe it can be
backported including depencies to 6.1 as this becomes the next lts?

this series on 6.1 is not enough to fix vlan-issue...so i miss anything there

but for next

Tested-By: Frank Wunderlich <frank-w@public-files.de>

regards Frank

_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel