: Add per-connection option to set max TSO frame size
Subsystems (4)
PATCH 1/1 : Add per-connection option to set max TSO frame size
Update: Respun for net-2.6.26 tree.
Update: changed max_gso_frame_size and sk_gso_max_size from signed to unsigned - thanks Stephen!
This patch adds the ability for device drivers to control the size of the TSO frames being sent to them, per TCP connection. By setting the netdevice's max_gso_frame_size value, the socket layer will set the GSO frame size based on that value. This will propogate into the TCP layer, and send TSO's of that size to the hardware.
This can be desirable to help tune the bursty nature of TSO on a per-adapter basis, where one may have 1 GbE and 10 GbE devices coexisting in a system, one running multiqueue and the other not, etc.
This can also be desirable for devices that cannot support full 64 KB TSO's, but still want to benefit from some level of segmentation offloading.
---
include/linux/netdevice.h | 6 ++++++ include/net/sock.h | 2 ++ net/core/dev.c | 1 + net/core/sock.c | 6 ++++-- net/ipv4/tcp_output.c | 4 ++-- 5 files changed, 15 insertions(+), 4 deletions(-)
include/linux/netdevice.h +6 -0
@@ -620,6 +620,7 @@ struct net_device/* Partially transmitted GSO packet. */ struct sk_buff *gso_skb; +u16 max_gso_frame_size;/* * One part is mostly used on xmit path (device)
@@ -1475,6 +1476,11 @@ static inline int netif_needs_gso(struct net_device *dev, struct sk_buff *skb) unlikely(skb->ip_summed != CHECKSUM_PARTIAL)); } +static inline void netif_set_max_gso_size (struct net_device *dev,u16 size) +{ + dev->max_gso_frame_size = size; +} +/* On bonding slaves other than the currently active slave, suppress * duplicates except for 802.3ad ETH_P_SLOW, alb non-mcast/bcast, and * ARP on active-backup slaves with arp_validate enabled.
include/net/sock.h +2 -0
@@ -151,6 +151,7 @@ struct sock_common {* @sk_no_check: %SO_NO_CHECK setting, wether or not checkup packets * @sk_route_caps: route capabilities (e.g. %NETIF_F_TSO) * @sk_gso_type: GSO type (e.g. %SKB_GSO_TCPV4) +* @sk_gso_max_size: Maximum GSO segment size to build * @sk_lingertime: %SO_LINGER l_linger setting * @sk_backlog: always used with the per-socket spinlock held * @sk_callback_lock: used with the callbacks in the end of this struct
@@ -236,6 +237,7 @@ struct sock {gfp_t sk_allocation;int sk_route_caps;int sk_gso_type; +__u16 sk_gso_max_size;int sk_rcvlowat;unsigned long sk_flags;unsigned long sk_lingertime;
net/core/dev.c +1 -0
@@ -4021,6 +4021,7 @@ struct net_device *alloc_netdev_mq(int sizeof_priv, const char *name, } dev->egress_subqueue_count = queue_count; + dev->max_gso_frame_size =65536 ; dev->get_stats = internal_stats;netpoll_netdev_init (dev);
net/core/sock.c +4 -2
@@ -1076,10 +1076,12 @@ void sk_setup_caps(struct sock *sk, struct dst_entry *dst)if (sk->sk_route_caps &NETIF_F_GSO ) sk->sk_route_caps |=NETIF_F_GSO_SOFTWARE ;if (sk_can_gso (sk)) { -if (dst->header_len) +if (dst->header_len) { sk->sk_route_caps &= ~NETIF_F_GSO_MASK ; -else + }else { sk->sk_route_caps |=NETIF_F_SG |NETIF_F_HW_CSUM ; + sk->sk_gso_max_size = dst->dev->max_gso_frame_size; + } } }EXPORT_SYMBOL_GPL (sk_setup_caps);
net/ipv4/tcp_output.c +2 -2
@@ -998,7 +998,7 @@ unsigned int tcp_current_mss(struct sock *sk, int large_allowed) xmit_size_goal = mss_now;if (doing_tso) { - xmit_size_goal = (65535 - + xmit_size_goal = ((sk->sk_gso_max_size -1 ) -inet_csk (sk)->icsk_af_ops->net_header_len -inet_csk (sk)->icsk_ext_hdr_len - tp->tcp_header_len);
@@ -1274,7 +1274,7 @@ static int tcp_tso_should_defer(struct sock *sk, struct sk_buff *skb) limit =min (send_win, cong_win);/* If a full-sized TSO skb can be sent, do it. */ -if (limit >=65536 ) +if (limit >= sk->sk_gso_max_size) goto send_now;if (sysctl_tcp_tso_win_divisor) {
On 27-02-2008 18:26, PJ Waskiewicz wrote:
...
Quoted 9 lines
include/linux/netdevice.h +1
-0@@ -620,6 +620,7 @@ struct net_device/* Partially transmitted GSO packet. */ struct sk_buff *gso_skb; +u16 max_gso_frame_size;
...
Quoted 10 lines
include/linux/netdevice.h +2
-0+static inline void netif_set_max_gso_size(struct net_device *dev, u16 size) + __u16 sk_gso_max_size; int sk_rcvlowat;unsigned long sk_flags;unsigned long sk_lingertime;net/core/dev.c +0
-0
net/core/dev.c struct net_device *alloc_netdev_mq(int sizeof_priv, const char *name, · line 4024
Quoted 4 lines
net/core/dev.c +1
-0@@ -4021,6 +4021,7 @@ struct net_device *alloc_netdev_mq(int sizeof_priv, const char *name, } dev->egress_subqueue_count = queue_count; + dev->max_gso_frame_size =65536 ;
Does it fit into u16?
Isn't it "nicer" with some constant (#define MAX_GSO_FRAME_SIZE ...)?
Why not more consistent names, e.g.:dev->gso_max_size;sk_gso_max_size;netif_set_gso_max_size()
(GSO_MAX_SIZE)
ordev->max_gso_frame_size;sk_max_gso_frame_size;netif_set_max_gso_frame_size()
(MAX_GSO_FRAME_SIZE)
etc.?
Regards,
Jarek P.
Quoted 11 lines · Jarek Poplawski
On 27-02-2008 18:26, PJ Waskiewicz wrote:
...include/linux/netdevice.h struct net_device · line 623
Quoted 9 lines
include/linux/netdevice.h +1
-0@@ -620,6 +620,7 @@ struct net_device/* Partially transmitted GSO packet. */ struct sk_buff *gso_skb; +u16 max_gso_frame_size;
...And since it looks like not very often used - maybe it could be placed a bit lower?
Jarek P.
RE: [PATCH] [NET 2.6.26]: Add per-connection option to set max TSOframe size
Quoted 4 lines
...And since it looks like not very often used - maybe it could be placed a bit lower?
Jarek P.
Yes, I can move it. I just grouped it with the other GSO/TSO elements in the struct. I'll take a closer look and see if I can place it elsewhere for better efficiency. I'll send change along with the other changes from your other feedback.
Thanks,
-PJ Waskiewicz
RE: [PATCH] [NET 2.6.26]: Add per-connection option to set max TSOframe size
In reply to Jarek PoplawskiQuoted 4 lines
net/core/dev.c struct net_device *alloc_netdev_mq(int sizeof_priv, const char *name, · line 4024
Quoted 2 lines
net/core/dev.c +1
-0dev->egress_subqueue_count = queue_count; + dev->max_gso_frame_size =65536 ;Does it fit into u16?
No it doesn't. I originally had this as an int, and borked it when switching it to unsigned. I'll fix this asap.
Quoted 1 line
Isn't it "nicer" with some constant (#define
MAX_GSO_FRAME_SIZE...)?
Yes it is. I'll add this.
Quoted 11 lines
Why not more consistent names, e.g.:
dev->gso_max_size;sk_gso_max_size;netif_set_gso_max_size()
(GSO_MAX_SIZE)
ordev->max_gso_frame_size;sk_max_gso_frame_size;netif_set_max_gso_frame_size()
(MAX_GSO_FRAME_SIZE)
etc.?
Very good point. I'll change this and resend today. Thanks a bunch for the feedback!
Cheers,
-PJ Waskiewicz