Skip to contentSkip to conversation

net: spacemit: k1-emac: fix jumbo frame support

PATCH Applied Tomas Hlavacek 5 participants 6 messages Jan 30 – Feb 20

Subsystems (5)
lore
Share & tools
lore
#0 lore raw

PATCH v3 1/1 net: spacemit: k1-emac: fix jumbo frame support

The driver never programs the MAC frame size and jabber registers, causing the hardware to reject frames larger than the default 1518 bytes even when larger DMA buffers are allocated.

Program MAC_MAXIMUM_FRAME_SIZE, MAC_TRANSMIT_JABBER_SIZE, and MAC_RECEIVE_JABBER_SIZE based on the configured MTU. Also fix the maximum buffer size from 4096 to 4095, since the descriptor buffer size field is only 12 bits. Account for double VLAN tags in frame size calculations.

Fixes bfec6d7f2001 ("net: spacemit: Add K1 Ethernet MAC")

---
v3:
- Set all three frame/jabber registers, fix 12-bit buffer size field overflow, use actual frame size with VLAN headroom consistently.

v2: https://lore.kernel.org/netdev/[email protected]/
- Added Fixes tag and Cc stable.

v1: https://lore.kernel.org/netdev/[email protected]/

---
 drivers/net/ethernet/spacemit/k1_emac.c | 21 +++++++++++++++------
 1 file changed, 15 insertions(+), 6 deletions(-)
drivers/net/ethernet/spacemit/k1_emac.c +15 -6
index 220eb5ce7583..cd6879d7434c 100644
@@ -12,6 +12,7 @@
 #include <linux/dma-mapping.h>
 #include <linux/etherdevice.h>
 #include <linux/ethtool.h>
+#include <linux/if_vlan.h>
 #include <linux/interrupt.h>
 #include <linux/io.h>
 #include <linux/iopoll.h>
@@ -38,7 +39,7 @@
 
 #define EMAC_DEFAULT_BUFSIZE		1536
 #define EMAC_RX_BUF_2K			2048
-#define EMAC_RX_BUF_4K			4096
+#define EMAC_RX_BUF_MAX			FIELD_MAX(RX_DESC_1_BUFFER_SIZE_1_MASK)
 
 /* Tuning parameters from SpacemiT */
 #define EMAC_TX_FRAMES			64
@@ -202,8 +203,7 @@ static void emac_init_hw(struct emac_priv *priv)
 {
 	/* Destination address for 802.3x Ethernet flow control */
 	u8 fc_dest_addr[ETH_ALEN] = { 0x01, 0x80, 0xc2, 0x00, 0x00, 0x01 };
-
-	u32 rxirq = 0, dma = 0;
+	u32 rxirq = 0, dma = 0, frame_sz;
 
 	regmap_set_bits(priv->regmap_apmu,
 			priv->regmap_apmu_offset + APMU_EMAC_CTRL_REG,
@@ -228,6 +228,15 @@ static void emac_init_hw(struct emac_priv *priv)
 		DEFAULT_TX_THRESHOLD);
 	emac_wr(priv, MAC_RECEIVE_PACKET_START_THRESHOLD, DEFAULT_RX_THRESHOLD);
 
+	/* Set maximum frame size and jabber size based on configured MTU,
+	 * accounting for Ethernet header, double VLAN tags, and FCS.
+	 */
+	frame_sz = priv->ndev->mtu + ETH_HLEN + 2 * VLAN_HLEN + ETH_FCS_LEN;
+
+	emac_wr(priv, MAC_MAXIMUM_FRAME_SIZE, frame_sz);
+	emac_wr(priv, MAC_TRANSMIT_JABBER_SIZE, frame_sz);
+	emac_wr(priv, MAC_RECEIVE_JABBER_SIZE, frame_sz);
+
 	/* Configure flow control (enabled in emac_adjust_link() later) */
 	emac_set_mac_addr_reg(priv, fc_dest_addr, MAC_FC_SOURCE_ADDRESS_HIGH);
 	emac_wr(priv, MAC_FC_PAUSE_HIGH_THRESHOLD, DEFAULT_FC_FIFO_HIGH);
@@ -924,14 +933,14 @@ static int emac_change_mtu(struct net_device *ndev, int mtu)
 		return -EBUSY;
 	}
 
-	frame_len = mtu + ETH_HLEN + ETH_FCS_LEN;
+	frame_len = mtu + ETH_HLEN + 2 * VLAN_HLEN + ETH_FCS_LEN;
 
 	if (frame_len <= EMAC_DEFAULT_BUFSIZE)
 		priv->dma_buf_sz = EMAC_DEFAULT_BUFSIZE;
 	else if (frame_len <= EMAC_RX_BUF_2K)
 		priv->dma_buf_sz = EMAC_RX_BUF_2K;
 	else
-		priv->dma_buf_sz = EMAC_RX_BUF_4K;
+		priv->dma_buf_sz = EMAC_RX_BUF_MAX;
 
 	ndev->mtu = mtu;
 
@@ -2005,7 +2014,7 @@ static int emac_probe(struct platform_device *pdev)
 	ndev->hw_features = NETIF_F_SG;
 	ndev->features |= ndev->hw_features;
 
-	ndev->max_mtu = EMAC_RX_BUF_4K - (ETH_HLEN + ETH_FCS_LEN);
+	ndev->max_mtu = EMAC_RX_BUF_MAX - (ETH_HLEN + 2 * VLAN_HLEN + ETH_FCS_LEN);
 	ndev->pcpu_stat_type = NETDEV_PCPU_STAT_DSTATS;
 
 	priv = netdev_priv(ndev);
Signature · 1 line
2.52.0
#1 lore raw
This patch was applied to netdev/net.git (main) by Jakub Kicinski <[email protected]>:

Hello:

This patch was applied to netdev/net.git (main)
by Jakub Kicinski <[email protected]>:

Quoted 11 lines · On Fri, 30 Jan 2026 11:23:01 +0100 you wrote:

The driver never programs the MAC frame size and jabber registers, causing the hardware to reject frames larger than the default 1518 bytes even when larger DMA buffers are allocated.

Program MAC_MAXIMUM_FRAME_SIZE, MAC_TRANSMIT_JABBER_SIZE, and MAC_RECEIVE_JABBER_SIZE based on the configured MTU. Also fix the maximum buffer size from 4096 to 4095, since the descriptor buffer size field is only 12 bits. Account for double VLAN tags in frame size calculations.

[...]

Here is the summary with links:
- [net,v3] net: spacemit: k1-emac: fix jumbo frame support

    https://git.kernel.org/netdev/net/c/3125fc170169

You are awesome, thank you!

Signature · 2 lines
Deet-doot-dot, I am a bot.
https://korg.docs.kernel.org/patchwork/pwbot.html
#2 lore raw

Re: [PATCH net v3] net: spacemit: k1-emac: fix jumbo frame support: manual merge

In reply to Tomas Hlavacek

Hello,

+cc: Mark for Linux Next

Quoted 9 lines · Tomas Hlavacek

The driver never programs the MAC frame size and jabber registers, causing the hardware to reject frames larger than the default 1518 bytes even when larger DMA buffers are allocated.

Program MAC_MAXIMUM_FRAME_SIZE, MAC_TRANSMIT_JABBER_SIZE, and MAC_RECEIVE_JABBER_SIZE based on the configured MTU. Also fix the maximum buffer size from 4096 to 4095, since the descriptor buffer size field is only 12 bits. Account for double VLAN tags in frame size calculations.

FYI, we got a small conflict when merging 'net' in 'net-next' in the MPTCP tree due to this patch applied in 'net':

3125fc170169 ("net: spacemit: k1-emac: fix jumbo frame support")

and this one from 'net-next':

f66086798f91 ("net: spacemit: Remove broken flow control support")

----- Generic Message -----
The best is to avoid conflicts between 'net' and 'net-next' trees but if they cannot be avoided when preparing patches, a note about how to fix them is much appreciated.

The conflict has been resolved on our side [1] and the resolution we suggest is attached to this email. Please report any issues linked to this conflict resolution as it might be used by others. If you worked on the mentioned patches, don't hesitate to ACK this conflict resolution.
---------------------------

Regarding this conflict, both commits modified independent code from the same context: code related to the flow control has been removed, and the one to frame/jabber size has been added.

Rerere cache is available in [2].

Cheers,
Matt

1: https://github.com/multipath-tcp/mptcp_net-next/commit/286ef2be604b
2: https://github.com/multipath-tcp/mptcp-upstream-rr-cache/commit/45e15

Cheers,
Matt

Signature · 1 line
Sponsored by the NGI0 Core fund.
drivers/net/ethernet/spacemit/k1_emac.c +0 -0
diff --cc drivers/net/ethernet/spacemit/k1_emac.c
index 40c9504b7444,b49c4708bf9e..dab0772c5b9d
@@@ -193,7 -201,9 +194,7 @@@ static void emac_reset_hw(struct emac_p
  
  static void emac_init_hw(struct emac_priv *priv)
  {
- 	u32 rxirq = 0, dma = 0;
 -	/* Destination address for 802.3x Ethernet flow control */
 -	u8 fc_dest_addr[ETH_ALEN] = { 0x01, 0x80, 0xc2, 0x00, 0x00, 0x01 };
+ 	u32 rxirq = 0, dma = 0, frame_sz;
  
  	regmap_set_bits(priv->regmap_apmu,
  			priv->regmap_apmu_offset + APMU_EMAC_CTRL_REG,
@@@ -218,6 -228,21 +219,15 @@@
  		DEFAULT_TX_THRESHOLD);
  	emac_wr(priv, MAC_RECEIVE_PACKET_START_THRESHOLD, DEFAULT_RX_THRESHOLD);
  
+ 	/* Set maximum frame size and jabber size based on configured MTU,
+ 	 * accounting for Ethernet header, double VLAN tags, and FCS.
+ 	 */
+ 	frame_sz = priv->ndev->mtu + ETH_HLEN + 2 * VLAN_HLEN + ETH_FCS_LEN;
+ 
+ 	emac_wr(priv, MAC_MAXIMUM_FRAME_SIZE, frame_sz);
+ 	emac_wr(priv, MAC_TRANSMIT_JABBER_SIZE, frame_sz);
+ 	emac_wr(priv, MAC_RECEIVE_JABBER_SIZE, frame_sz);
+ 
 -	/* Configure flow control (enabled in emac_adjust_link() later) */
 -	emac_set_mac_addr_reg(priv, fc_dest_addr, MAC_FC_SOURCE_ADDRESS_HIGH);
 -	emac_wr(priv, MAC_FC_PAUSE_HIGH_THRESHOLD, DEFAULT_FC_FIFO_HIGH);
 -	emac_wr(priv, MAC_FC_HIGH_PAUSE_TIME, DEFAULT_FC_PAUSE_TIME);
 -	emac_wr(priv, MAC_FC_PAUSE_LOW_THRESHOLD, 0);
 -
  	/* RX IRQ mitigation */
  	rxirq = FIELD_PREP(MREGBIT_RECEIVE_IRQ_FRAME_COUNTER_MASK,
  			   EMAC_RX_FRAMES);
#3 lore raw

Re: [PATCH net v3] net: spacemit: k1-emac: fix jumbo frame support: manual merge

Hi Matthieu and Mark,

Quoted 7 lines · Matthieu Baerts

[...]

The conflict has been resolved on our side [1] and the resolution we suggest is attached to this email. Please report any issues linked to this conflict resolution as it might be used by others. If you worked on the mentioned patches, don't hesitate to ACK this conflict resolution.
---------------------------

Your merge is correct. Thanks for handling it.

Vivian "dramforever" Wang

#4 lore raw

Re: [PATCH net v3] net: spacemit: k1-emac: fix jumbo frame support: manual merge

In reply to Matthieu Baerts

Dear Matthieu,

On Tue, Feb 3, 2026 at 10:39 AM Matthieu Baerts <[email protected]> wrote:
[...]

Quoted 4 lines

The conflict has been resolved on our side [1] and the resolution we suggest is attached to this email. Please report any issues linked to this conflict resolution as it might be used by others. If you worked on the mentioned patches, don't hesitate to ACK this conflict resolution.

It looks good to me. Thank you!

Tomas

#5 lore raw
In reply to Tomas Hlavacek
This patch was applied to riscv/linux.git (fixes) by Jakub Kicinski <[email protected]>:

Hello:

This patch was applied to riscv/linux.git (fixes)
by Jakub Kicinski <[email protected]>:

Quoted 11 lines · On Fri, 30 Jan 2026 11:23:01 +0100 you wrote:

The driver never programs the MAC frame size and jabber registers, causing the hardware to reject frames larger than the default 1518 bytes even when larger DMA buffers are allocated.

Program MAC_MAXIMUM_FRAME_SIZE, MAC_TRANSMIT_JABBER_SIZE, and MAC_RECEIVE_JABBER_SIZE based on the configured MTU. Also fix the maximum buffer size from 4096 to 4095, since the descriptor buffer size field is only 12 bits. Account for double VLAN tags in frame size calculations.

[...]

Here is the summary with links:
- [net,v3] net: spacemit: k1-emac: fix jumbo frame support

    https://git.kernel.org/riscv/c/3125fc170169

You are awesome, thank you!

Signature · 2 lines
Deet-doot-dot, I am a bot.
https://korg.docs.kernel.org/patchwork/pwbot.html