Skip to contentSkip to conversation

net: fix Tx VLAN flag for offload emulation

PATCH Applied Chas Williams 2 participants 5 messages Mar 25, 2019 – Apr 4, 2019

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 net: fix Tx VLAN flag for offload emulation

From: Bill Hong <[email protected]>

A PMD might use rte_vlan_insert to implement Tx VLAN offload. Typically the PMD will insert the VLAN header in the transmit path and then attempt to send the packets. If this fails, the packets are returned to the application which may attempt to send these packets again. If the PKT_TX_VLAN flag is not cleared, the transmit path may attempt to insert the VLAN header again.

Fixes 47aa48b969f8 ("net: fix stripped VLAN flag for offload emulation");
---
 lib/librte_net/rte_ether.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
lib/librte_net/rte_ether.h +1 -1
index c2c5e249f..e0d831113 100644
@@ -408,7 +408,7 @@ static inline int rte_vlan_insert(struct rte_mbuf **m)
 	vh = (struct vlan_hdr *) (nh + 1);
 	vh->vlan_tci = rte_cpu_to_be_16((*m)->vlan_tci);
 
-	(*m)->ol_flags &= ~PKT_RX_VLAN_STRIPPED;
+	(*m)->ol_flags &= ~(PKT_RX_VLAN_STRIPPED | PKT_TX_VLAN);
 
 	return 0;
 }
Signature · 1 line
2.17.2
#1 lore raw
Quoted 23 lines · Chas Williams

From: Bill Hong <[email protected]>

A PMD might use rte_vlan_insert to implement Tx VLAN offload. Typically the PMD will insert the VLAN header in the transmit path and then attempt to send the packets. If this fails, the packets are returned to the application which may attempt to send these packets again. If the PKT_TX_VLAN flag is not cleared, the transmit path may attempt to insert the VLAN header again.

Fixes 47aa48b969f8 ("net: fix stripped VLAN flag for offload emulation");
---
 lib/librte_net/rte_ether.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
lib/librte_net/rte_ether.h +0 -0
index c2c5e249f..e0d831113 100644

lib/librte_net/rte_ether.h static inline int rte_vlan_insert(struct rte_mbuf **m) · line 411

Quoted 5 lines
lib/librte_net/rte_ether.h +1 -1
@@ -408,7 +408,7 @@ static inline int rte_vlan_insert(struct rte_mbuf **m)
 	vh = (struct vlan_hdr *) (nh + 1);
 	vh->vlan_tci = rte_cpu_to_be_16((*m)->vlan_tci);
 
-	(*m)->ol_flags &= ~PKT_RX_VLAN_STRIPPED;
+	(*m)->ol_flags &= ~(PKT_RX_VLAN_STRIPPED | PKT_TX_VLAN);

Hi Chas,

Looks reasonable, AFAIK, PKT_TX_VLAN flag means 'mbuf->vlan_tci' has valid value and a request from application to driver to insert VLAN information. After successful insertion flag can be removed.

But in mbuf header, flags documented as:
PKT_TX_VLAN: TX packet is a 802.1q VLAN packet.
PKT_TX_QINQ: TX packet with double VLAN inserted.

Hi Oliver,

Is above comments correct, or I am missing something J

Also here cleaning 'PKT_RX_VLAN_STRIPPED' looks like to say 'vlan' information is not stripped but in the packet data, but 'RX' in this context can be confusing, should we have a more generic 'PKT_VLAN_STRIPPED' ?

#2 lore raw
Quoted 32 lines · Ferruh Yigit
Quoted 23 lines · Chas Williams

From: Bill Hong <[email protected]>

A PMD might use rte_vlan_insert to implement Tx VLAN offload. Typically the PMD will insert the VLAN header in the transmit path and then attempt to send the packets. If this fails, the packets are returned to the application which may attempt to send these packets again. If the PKT_TX_VLAN flag is not cleared, the transmit path may attempt to insert the VLAN header again.

Fixes 47aa48b969f8 ("net: fix stripped VLAN flag for offload emulation");
---
  lib/librte_net/rte_ether.h | 2 +-
  1 file changed, 1 insertion(+), 1 deletion(-)
lib/librte_net/rte_ether.h +0 -0
index c2c5e249f..e0d831113 100644

lib/librte_net/rte_ether.h static inline int rte_vlan_insert(struct rte_mbuf **m) · line 411

Quoted 5 lines
lib/librte_net/rte_ether.h +1 -1
@@ -408,7 +408,7 @@ static inline int rte_vlan_insert(struct rte_mbuf **m)
  	vh = (struct vlan_hdr *) (nh + 1);
  	vh->vlan_tci = rte_cpu_to_be_16((*m)->vlan_tci);
 
-	(*m)->ol_flags &= ~PKT_RX_VLAN_STRIPPED;
+	(*m)->ol_flags &= ~(PKT_RX_VLAN_STRIPPED | PKT_TX_VLAN);

Hi Chas,

Quoted 7 lines

Looks reasonable, AFAIK, PKT_TX_VLAN flag means 'mbuf->vlan_tci' has valid value and a request from application to driver to insert VLAN information. After successful insertion flag can be removed.

But in mbuf header, flags documented as:
PKT_TX_VLAN: TX packet is a 802.1q VLAN packet.
PKT_TX_QINQ: TX packet with double VLAN inserted.

Note the usage slightly below:

	/**
	 * Bitmask of all supported packet Tx offload features flags,
	 * which can be set for packet.
	 */
	#define PKT_TX_OFFLOAD_MASK (    \
			PKT_TX_OUTER_IPV6 |      \
			PKT_TX_OUTER_IPV4 |      \
			PKT_TX_OUTER_IP_CKSUM |  \
			PKT_TX_VLAN_PKT |        \

So this implies that PKT_TX_VLAN_PKT (which is PKT_TX_VLAN's old name) is an offload feature, not that the actual packet has an 802.1q header (offloaded or otherwise).

The same seems to apply for PKT_TX_QINQ.

Quoted 9 lines

Hi Oliver,

Is above comments correct, or I am missing something J

Also here cleaning 'PKT_RX_VLAN_STRIPPED' looks like to say 'vlan' information is not stripped but in the packet data, but 'RX' in this context can be confusing, should we have a more generic 'PKT_VLAN_STRIPPED' ?

Reviewed-by
#3 lore raw
Quoted 35 lines · Chas Williams
Quoted 32 lines · Ferruh Yigit
Quoted 23 lines · Chas Williams

From: Bill Hong <[email protected]>

A PMD might use rte_vlan_insert to implement Tx VLAN offload. Typically the PMD will insert the VLAN header in the transmit path and then attempt to send the packets. If this fails, the packets are returned to the application which may attempt to send these packets again. If the PKT_TX_VLAN flag is not cleared, the transmit path may attempt to insert the VLAN header again.

Fixes 47aa48b969f8 ("net: fix stripped VLAN flag for offload emulation");
---
  lib/librte_net/rte_ether.h | 2 +-
  1 file changed, 1 insertion(+), 1 deletion(-)
lib/librte_net/rte_ether.h +0 -0
index c2c5e249f..e0d831113 100644

lib/librte_net/rte_ether.h static inline int rte_vlan_insert(struct rte_mbuf **m) · line 411

Quoted 5 lines
lib/librte_net/rte_ether.h +1 -1
@@ -408,7 +408,7 @@ static inline int rte_vlan_insert(struct rte_mbuf **m)
  	vh = (struct vlan_hdr *) (nh + 1);
  	vh->vlan_tci = rte_cpu_to_be_16((*m)->vlan_tci);
 
-	(*m)->ol_flags &= ~PKT_RX_VLAN_STRIPPED;
+	(*m)->ol_flags &= ~(PKT_RX_VLAN_STRIPPED | PKT_TX_VLAN);

Hi Chas,

Quoted 7 lines
Quoted 7 lines

Looks reasonable, AFAIK, PKT_TX_VLAN flag means 'mbuf->vlan_tci' has valid value and a request from application to driver to insert VLAN information. After successful insertion flag can be removed.

But in mbuf header, flags documented as:
PKT_TX_VLAN: TX packet is a 802.1q VLAN packet.
PKT_TX_QINQ: TX packet with double VLAN inserted.

Hi Olivier,

Any comment on them, should they be fixed?

Quoted 13 lines

Note the usage slightly below:

	/**
	 * Bitmask of all supported packet Tx offload features flags,
	 * which can be set for packet.
	 */
	#define PKT_TX_OFFLOAD_MASK (    \
			PKT_TX_OUTER_IPV6 |      \
			PKT_TX_OUTER_IPV4 |      \
			PKT_TX_OUTER_IP_CKSUM |  \
			PKT_TX_VLAN_PKT |        \
Quoted 5 lines

So this implies that PKT_TX_VLAN_PKT (which is PKT_TX_VLAN's old name) is an offload feature, not that the actual packet has an 802.1q header (offloaded or otherwise).

The same seems to apply for PKT_TX_QINQ.

Agree, I just would like to confirm, so lgtm:

Quoted 10 lines
Quoted 10 lines

Hi Oliver,

Is above comments correct, or I am missing something J

Also here cleaning 'PKT_RX_VLAN_STRIPPED' looks like to say 'vlan' information is not stripped but in the packet data, but 'RX' in this context can be confusing, should we have a more generic 'PKT_VLAN_STRIPPED' ?

Applied
#4 lore raw
In reply to Chas Williams
Quoted 9 lines · Chas Williams

From: Bill Hong <[email protected]>

A PMD might use rte_vlan_insert to implement Tx VLAN offload. Typically the PMD will insert the VLAN header in the transmit path and then attempt to send the packets. If this fails, the packets are returned to the application which may attempt to send these packets again. If the PKT_TX_VLAN flag is not cleared, the transmit path may attempt to insert the VLAN header again.

Quoted 5 lines
Fixes 47aa48b969f8 ("net: fix stripped VLAN flag for offload emulation");

Applied to dpdk-next-net/master, thanks.