Skip to content

L
linux
Commit ba493b28 authored by Alexander Duyck's avatar Alexander Duyck Committed by Tim Gardner
Browse files
Options

i40e/i40evf: Do not write to descriptor unless we complete 

BugLink: http://bugs.launchpad.net/bugs/1547674

This patch defers writing to the Tx descriptor bits until we know we have
successfully completed a given operation.  So for example we defer updating
the tunnelling portion of the context descriptor until we have fully
identified the type.

The advantage to this approach is that we can assemble values as we go
instead of having to try and kludge everything together all at once.  As a
result we can significantly clean up the tunneling configuration for
instance as we can just do a pointer walk and do the math for the distance
between each set of points.
Signed-off-by: default avatarAlexander Duyck <aduyck@mirantis.com>
Tested-by: default avatarAndrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: default avatarJeff Kirsher <jeffrey.t.kirsher@intel.com>
(cherry picked from net-next commit 475b4205)
Signed-off-by: default avatarTim Gardner <tim.gardner@canonical.com>
parent f978582a
Hide whitespace changes
Inline Side-by-side
Showing with 84 additions and 76 deletions
drivers/net/ethernet/intel/i40e/i40e_txrx.c
View file @ ba493b28
...@@ -2404,24 +2404,26 @@ static void i40e_tx_enable_csum(struct sk_buff *skb, u32 *tx_flags, ...@@ -2404,24 +2404,26 @@ static void i40e_tx_enable_csum(struct sk_buff *skb, u32 *tx_flags,
unsigned char *hdr; unsigned char *hdr;
} l4; } l4;
unsigned char *exthdr; unsigned char *exthdr;
u32 l4_tunnel = 0; u32 offset, cmd = 0, tunnel = 0;
__be16 frag_off; __be16 frag_off;
u8 l4_proto = 0; u8 l4_proto = 0;
ip.hdr = skb_network_header(skb); ip.hdr = skb_network_header(skb);
l4.hdr = skb_transport_header(skb); l4.hdr = skb_transport_header(skb);
/* compute outer L2 header size */
offset = ((ip.hdr - skb->data) / 2) << I40E_TX_DESC_LENGTH_MACLEN_SHIFT;
if (skb->encapsulation) { if (skb->encapsulation) {
/* define outer network header type */ /* define outer network header type */
if (*tx_flags & I40E_TX_FLAGS_IPV4) { if (*tx_flags & I40E_TX_FLAGS_IPV4) {
if (*tx_flags & I40E_TX_FLAGS_TSO) tunnel |= (*tx_flags & I40E_TX_FLAGS_TSO) ?
*cd_tunneling |= I40E_TX_CTX_EXT_IP_IPV4; I40E_TX_CTX_EXT_IP_IPV4 :
else I40E_TX_CTX_EXT_IP_IPV4_NO_CSUM;
*cd_tunneling |=
I40E_TX_CTX_EXT_IP_IPV4_NO_CSUM;
l4_proto = ip.v4->protocol; l4_proto = ip.v4->protocol;
} else if (*tx_flags & I40E_TX_FLAGS_IPV6) { } else if (*tx_flags & I40E_TX_FLAGS_IPV6) {
*cd_tunneling |= I40E_TX_CTX_EXT_IP_IPV6; tunnel |= I40E_TX_CTX_EXT_IP_IPV6;
exthdr = ip.hdr + sizeof(*ip.v6); exthdr = ip.hdr + sizeof(*ip.v6);
l4_proto = ip.v6->nexthdr; l4_proto = ip.v6->nexthdr;
...@@ -2430,33 +2432,38 @@ static void i40e_tx_enable_csum(struct sk_buff *skb, u32 *tx_flags, ...@@ -2430,33 +2432,38 @@ static void i40e_tx_enable_csum(struct sk_buff *skb, u32 *tx_flags,
&l4_proto, &frag_off); &l4_proto, &frag_off);
} }
/* compute outer L3 header size */
tunnel |= ((l4.hdr - ip.hdr) / 4) <<
I40E_TXD_CTX_QW0_EXT_IPLEN_SHIFT;
/* switch IP header pointer from outer to inner header */
ip.hdr = skb_inner_network_header(skb);
/* define outer transport */ /* define outer transport */
switch (l4_proto) { switch (l4_proto) {
case IPPROTO_UDP: case IPPROTO_UDP:
l4_tunnel = I40E_TXD_CTX_UDP_TUNNELING; tunnel |= I40E_TXD_CTX_UDP_TUNNELING;
*tx_flags |= I40E_TX_FLAGS_UDP_TUNNEL; *tx_flags |= I40E_TX_FLAGS_UDP_TUNNEL;
break; break;
case IPPROTO_GRE: case IPPROTO_GRE:
l4_tunnel = I40E_TXD_CTX_GRE_TUNNELING; tunnel |= I40E_TXD_CTX_GRE_TUNNELING;
*tx_flags |= I40E_TX_FLAGS_UDP_TUNNEL; *tx_flags |= I40E_TX_FLAGS_UDP_TUNNEL;
break; break;
default: default:
return; return;
} }
/* compute tunnel header size */
tunnel |= ((ip.hdr - l4.hdr) / 2) <<
I40E_TXD_CTX_QW0_NATLEN_SHIFT;
/* record tunnel offload values */
*cd_tunneling |= tunnel;
/* switch L4 header pointer from outer to inner */ /* switch L4 header pointer from outer to inner */
ip.hdr = skb_inner_network_header(skb);
l4.hdr = skb_inner_transport_header(skb); l4.hdr = skb_inner_transport_header(skb);
l4_proto = 0; l4_proto = 0;
/* Now set the ctx descriptor fields */
*cd_tunneling |= (skb_network_header_len(skb) >> 2) <<
I40E_TXD_CTX_QW0_EXT_IPLEN_SHIFT |
l4_tunnel |
((skb_inner_network_offset(skb) -
skb_transport_offset(skb)) >> 1) <<
I40E_TXD_CTX_QW0_NATLEN_SHIFT;
/* reset type as we transition from outer to inner headers */ /* reset type as we transition from outer to inner headers */
*tx_flags &= ~(I40E_TX_FLAGS_IPV4 | I40E_TX_FLAGS_IPV6); *tx_flags &= ~(I40E_TX_FLAGS_IPV4 | I40E_TX_FLAGS_IPV6);
if (ip.v4->version == 4) if (ip.v4->version == 4)
...@@ -2471,13 +2478,11 @@ static void i40e_tx_enable_csum(struct sk_buff *skb, u32 *tx_flags, ...@@ -2471,13 +2478,11 @@ static void i40e_tx_enable_csum(struct sk_buff *skb, u32 *tx_flags,
/* the stack computes the IP header already, the only time we /* the stack computes the IP header already, the only time we
* need the hardware to recompute it is in the case of TSO. * need the hardware to recompute it is in the case of TSO.
*/ */
if (*tx_flags & I40E_TX_FLAGS_TSO) { cmd |= (*tx_flags & I40E_TX_FLAGS_TSO) ?
*td_cmd |= I40E_TX_DESC_CMD_IIPT_IPV4_CSUM; I40E_TX_DESC_CMD_IIPT_IPV4_CSUM :
} else { I40E_TX_DESC_CMD_IIPT_IPV4;
*td_cmd |= I40E_TX_DESC_CMD_IIPT_IPV4;
}
} else if (*tx_flags & I40E_TX_FLAGS_IPV6) { } else if (*tx_flags & I40E_TX_FLAGS_IPV6) {
*td_cmd |= I40E_TX_DESC_CMD_IIPT_IPV6; cmd |= I40E_TX_DESC_CMD_IIPT_IPV6;
exthdr = ip.hdr + sizeof(*ip.v6); exthdr = ip.hdr + sizeof(*ip.v6);
l4_proto = ip.v6->nexthdr; l4_proto = ip.v6->nexthdr;
...@@ -2486,35 +2491,34 @@ static void i40e_tx_enable_csum(struct sk_buff *skb, u32 *tx_flags, ...@@ -2486,35 +2491,34 @@ static void i40e_tx_enable_csum(struct sk_buff *skb, u32 *tx_flags,
&l4_proto, &frag_off); &l4_proto, &frag_off);
} }
/* Now set the td_offset for IP header length */ /* compute inner L3 header size */
*td_offset = ((l4.hdr - ip.hdr) / 4) << I40E_TX_DESC_LENGTH_IPLEN_SHIFT; offset |= ((l4.hdr - ip.hdr) / 4) << I40E_TX_DESC_LENGTH_IPLEN_SHIFT;
/* words in MACLEN + dwords in IPLEN + dwords in L4Len */
*td_offset |= (skb_network_offset(skb) >> 1) <<
I40E_TX_DESC_LENGTH_MACLEN_SHIFT;
/* Enable L4 checksum offloads */ /* Enable L4 checksum offloads */
switch (l4_proto) { switch (l4_proto) {
case IPPROTO_TCP: case IPPROTO_TCP:
/* enable checksum offloads */ /* enable checksum offloads */
*td_cmd |= I40E_TX_DESC_CMD_L4T_EOFT_TCP; cmd |= I40E_TX_DESC_CMD_L4T_EOFT_TCP;
*td_offset |= l4.tcp->doff << offset |= l4.tcp->doff << I40E_TX_DESC_LENGTH_L4_FC_LEN_SHIFT;
I40E_TX_DESC_LENGTH_L4_FC_LEN_SHIFT;
break; break;
case IPPROTO_SCTP: case IPPROTO_SCTP:
/* enable SCTP checksum offload */ /* enable SCTP checksum offload */
*td_cmd |= I40E_TX_DESC_CMD_L4T_EOFT_SCTP; cmd |= I40E_TX_DESC_CMD_L4T_EOFT_SCTP;
*td_offset |= (sizeof(struct sctphdr) >> 2) << offset |= (sizeof(struct sctphdr) >> 2) <<
I40E_TX_DESC_LENGTH_L4_FC_LEN_SHIFT; I40E_TX_DESC_LENGTH_L4_FC_LEN_SHIFT;
break; break;
case IPPROTO_UDP: case IPPROTO_UDP:
/* enable UDP checksum offload */ /* enable UDP checksum offload */
*td_cmd |= I40E_TX_DESC_CMD_L4T_EOFT_UDP; cmd |= I40E_TX_DESC_CMD_L4T_EOFT_UDP;
*td_offset |= (sizeof(struct udphdr) >> 2) << offset |= (sizeof(struct udphdr) >> 2) <<
I40E_TX_DESC_LENGTH_L4_FC_LEN_SHIFT; I40E_TX_DESC_LENGTH_L4_FC_LEN_SHIFT;
break; break;
default: default:
break; break;
} }
*td_cmd |= cmd;
*td_offset |= offset;
} }
/** /**
......
drivers/net/ethernet/intel/i40evf/i40e_txrx.c
View file @ ba493b28
...@@ -1621,24 +1621,26 @@ static void i40e_tx_enable_csum(struct sk_buff *skb, u32 *tx_flags, ...@@ -1621,24 +1621,26 @@ static void i40e_tx_enable_csum(struct sk_buff *skb, u32 *tx_flags,
unsigned char *hdr; unsigned char *hdr;
} l4; } l4;
unsigned char *exthdr; unsigned char *exthdr;
u32 l4_tunnel = 0; u32 offset, cmd = 0, tunnel = 0;
__be16 frag_off; __be16 frag_off;
u8 l4_proto = 0; u8 l4_proto = 0;
ip.hdr = skb_network_header(skb); ip.hdr = skb_network_header(skb);
l4.hdr = skb_transport_header(skb); l4.hdr = skb_transport_header(skb);
/* compute outer L2 header size */
offset = ((ip.hdr - skb->data) / 2) << I40E_TX_DESC_LENGTH_MACLEN_SHIFT;
if (skb->encapsulation) { if (skb->encapsulation) {
/* define outer network header type */ /* define outer network header type */
if (*tx_flags & I40E_TX_FLAGS_IPV4) { if (*tx_flags & I40E_TX_FLAGS_IPV4) {
if (*tx_flags & I40E_TX_FLAGS_TSO) tunnel |= (*tx_flags & I40E_TX_FLAGS_TSO) ?
*cd_tunneling |= I40E_TX_CTX_EXT_IP_IPV4; I40E_TX_CTX_EXT_IP_IPV4 :
else I40E_TX_CTX_EXT_IP_IPV4_NO_CSUM;
*cd_tunneling |=
I40E_TX_CTX_EXT_IP_IPV4_NO_CSUM;
l4_proto = ip.v4->protocol; l4_proto = ip.v4->protocol;
} else if (*tx_flags & I40E_TX_FLAGS_IPV6) { } else if (*tx_flags & I40E_TX_FLAGS_IPV6) {
*cd_tunneling |= I40E_TX_CTX_EXT_IP_IPV6; tunnel |= I40E_TX_CTX_EXT_IP_IPV6;
exthdr = ip.hdr + sizeof(*ip.v6); exthdr = ip.hdr + sizeof(*ip.v6);
l4_proto = ip.v6->nexthdr; l4_proto = ip.v6->nexthdr;
...@@ -1647,33 +1649,38 @@ static void i40e_tx_enable_csum(struct sk_buff *skb, u32 *tx_flags, ...@@ -1647,33 +1649,38 @@ static void i40e_tx_enable_csum(struct sk_buff *skb, u32 *tx_flags,
&l4_proto, &frag_off); &l4_proto, &frag_off);
} }
/* compute outer L3 header size */
tunnel |= ((l4.hdr - ip.hdr) / 4) <<
I40E_TXD_CTX_QW0_EXT_IPLEN_SHIFT;
/* switch IP header pointer from outer to inner header */
ip.hdr = skb_inner_network_header(skb);
/* define outer transport */ /* define outer transport */
switch (l4_proto) { switch (l4_proto) {
case IPPROTO_UDP: case IPPROTO_UDP:
l4_tunnel = I40E_TXD_CTX_UDP_TUNNELING; tunnel |= I40E_TXD_CTX_UDP_TUNNELING;
*tx_flags |= I40E_TX_FLAGS_VXLAN_TUNNEL; *tx_flags |= I40E_TX_FLAGS_VXLAN_TUNNEL;
break; break;
case IPPROTO_GRE: case IPPROTO_GRE:
l4_tunnel = I40E_TXD_CTX_GRE_TUNNELING; tunnel |= I40E_TXD_CTX_GRE_TUNNELING;
*tx_flags |= I40E_TX_FLAGS_VXLAN_TUNNEL; *tx_flags |= I40E_TX_FLAGS_VXLAN_TUNNEL;
break; break;
default: default:
return; return;
} }
/* compute tunnel header size */
tunnel |= ((ip.hdr - l4.hdr) / 2) <<
I40E_TXD_CTX_QW0_NATLEN_SHIFT;
/* record tunnel offload values */
*cd_tunneling |= tunnel;
/* switch L4 header pointer from outer to inner */ /* switch L4 header pointer from outer to inner */
ip.hdr = skb_inner_network_header(skb);
l4.hdr = skb_inner_transport_header(skb); l4.hdr = skb_inner_transport_header(skb);
l4_proto = 0; l4_proto = 0;
/* Now set the ctx descriptor fields */
*cd_tunneling |= (skb_network_header_len(skb) >> 2) <<
I40E_TXD_CTX_QW0_EXT_IPLEN_SHIFT |
l4_tunnel |
((skb_inner_network_offset(skb) -
skb_transport_offset(skb)) >> 1) <<
I40E_TXD_CTX_QW0_NATLEN_SHIFT;
/* reset type as we transition from outer to inner headers */ /* reset type as we transition from outer to inner headers */
*tx_flags &= ~(I40E_TX_FLAGS_IPV4 | I40E_TX_FLAGS_IPV6); *tx_flags &= ~(I40E_TX_FLAGS_IPV4 | I40E_TX_FLAGS_IPV6);
if (ip.v4->version == 4) if (ip.v4->version == 4)
...@@ -1688,13 +1695,11 @@ static void i40e_tx_enable_csum(struct sk_buff *skb, u32 *tx_flags, ...@@ -1688,13 +1695,11 @@ static void i40e_tx_enable_csum(struct sk_buff *skb, u32 *tx_flags,
/* the stack computes the IP header already, the only time we /* the stack computes the IP header already, the only time we
* need the hardware to recompute it is in the case of TSO. * need the hardware to recompute it is in the case of TSO.
*/ */
if (*tx_flags & I40E_TX_FLAGS_TSO) { cmd |= (*tx_flags & I40E_TX_FLAGS_TSO) ?
*td_cmd |= I40E_TX_DESC_CMD_IIPT_IPV4_CSUM; I40E_TX_DESC_CMD_IIPT_IPV4_CSUM :
} else { I40E_TX_DESC_CMD_IIPT_IPV4;
*td_cmd |= I40E_TX_DESC_CMD_IIPT_IPV4;
}
} else if (*tx_flags & I40E_TX_FLAGS_IPV6) { } else if (*tx_flags & I40E_TX_FLAGS_IPV6) {
*td_cmd |= I40E_TX_DESC_CMD_IIPT_IPV6; cmd |= I40E_TX_DESC_CMD_IIPT_IPV6;
exthdr = ip.hdr + sizeof(*ip.v6); exthdr = ip.hdr + sizeof(*ip.v6);
l4_proto = ip.v6->nexthdr; l4_proto = ip.v6->nexthdr;
...@@ -1703,35 +1708,34 @@ static void i40e_tx_enable_csum(struct sk_buff *skb, u32 *tx_flags, ...@@ -1703,35 +1708,34 @@ static void i40e_tx_enable_csum(struct sk_buff *skb, u32 *tx_flags,
&l4_proto, &frag_off); &l4_proto, &frag_off);
} }
/* Now set the td_offset for IP header length */ /* compute inner L3 header size */
*td_offset = ((l4.hdr - ip.hdr) / 4) << I40E_TX_DESC_LENGTH_IPLEN_SHIFT; offset |= ((l4.hdr - ip.hdr) / 4) << I40E_TX_DESC_LENGTH_IPLEN_SHIFT;
/* words in MACLEN + dwords in IPLEN + dwords in L4Len */
*td_offset |= (skb_network_offset(skb) >> 1) <<
I40E_TX_DESC_LENGTH_MACLEN_SHIFT;
/* Enable L4 checksum offloads */ /* Enable L4 checksum offloads */
switch (l4_proto) { switch (l4_proto) {
case IPPROTO_TCP: case IPPROTO_TCP:
/* enable checksum offloads */ /* enable checksum offloads */
*td_cmd |= I40E_TX_DESC_CMD_L4T_EOFT_TCP; cmd |= I40E_TX_DESC_CMD_L4T_EOFT_TCP;
*td_offset |= l4.tcp->doff << offset |= l4.tcp->doff << I40E_TX_DESC_LENGTH_L4_FC_LEN_SHIFT;
I40E_TX_DESC_LENGTH_L4_FC_LEN_SHIFT;
break; break;
case IPPROTO_SCTP: case IPPROTO_SCTP:
/* enable SCTP checksum offload */ /* enable SCTP checksum offload */
*td_cmd |= I40E_TX_DESC_CMD_L4T_EOFT_SCTP; cmd |= I40E_TX_DESC_CMD_L4T_EOFT_SCTP;
*td_offset |= (sizeof(struct sctphdr) >> 2) << offset |= (sizeof(struct sctphdr) >> 2) <<
I40E_TX_DESC_LENGTH_L4_FC_LEN_SHIFT; I40E_TX_DESC_LENGTH_L4_FC_LEN_SHIFT;
break; break;
case IPPROTO_UDP: case IPPROTO_UDP:
/* enable UDP checksum offload */ /* enable UDP checksum offload */
*td_cmd |= I40E_TX_DESC_CMD_L4T_EOFT_UDP; cmd |= I40E_TX_DESC_CMD_L4T_EOFT_UDP;
*td_offset |= (sizeof(struct udphdr) >> 2) << offset |= (sizeof(struct udphdr) >> 2) <<
I40E_TX_DESC_LENGTH_L4_FC_LEN_SHIFT; I40E_TX_DESC_LENGTH_L4_FC_LEN_SHIFT;
break; break;
default: default:
break; break;
} }
*td_cmd |= cmd;
*td_offset |= offset;
} }
/** /**
......
Markdown is supported
0% or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment
GitLab Nexedi Edition | About GitLab | About Nexedi | 沪ICP备2021021310号-2 | 沪ICP备2021021310号-7