From: Coco Li <lixiaoyan@google.com>
Date: Wed, 5 Jan 2022 02:48:38 -0800
Subject: gro: add ability to control gro max packet size
Patch-mainline: v5.17-rc1
Git-commit: eac1b93c14d645ef147b049ace0d5230df755548
References: jsc#PED-1565
Eric Dumazet suggested to allow users to modify max GRO packet size.
We have seen GRO being disabled by users of appliances (such as
wifi access points) because of claimed bufferbloat issues,
or some work arounds in sch_cake, to split GRO/GSO packets.
Instead of disabling GRO completely, one can chose to limit
the maximum packet size of GRO packets, depending on their
latency constraints.
This patch adds a per device gro_max_size attribute
that can be changed with ip link command.
ip link set dev eth0 gro_max_size 16000
Suggested-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: Coco Li <lixiaoyan@google.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Acked-by: Thomas Bogendoerfer <tbogendoerfer@suse.de>
---
include/linux/netdevice.h | 11 +++++++++++
include/uapi/linux/if_link.h | 1 +
net/core/dev.c | 1 +
net/core/gro.c | 6 +++++-
net/core/rtnetlink.c | 22 ++++++++++++++++++++++
tools/include/uapi/linux/if_link.h | 1 +
6 files changed, 41 insertions(+), 1 deletion(-)
--- a/include/linux/netdevice.h
+++ b/include/linux/netdevice.h
@@ -1933,6 +1933,8 @@ enum netdev_ml_priv_type {
* dev->addr_list_lock.
* @unlink_list: As netif_addr_lock() can be called recursively,
* keep a list of interfaces to be deleted.
+ * @gro_max_size: Maximum size of aggregated packet in generic
+ * receive offload (GRO)
*
* @dev_addr_shadow: Copy of @dev_addr to catch direct writes.
* @linkwatch_dev_tracker: refcount tracker used by linkwatch.
@@ -2118,6 +2120,8 @@ struct net_device {
struct bpf_prog __rcu *xdp_prog;
unsigned long gro_flush_timeout;
int napi_defer_hard_irqs;
+#define GRO_MAX_SIZE 65536
+ unsigned int gro_max_size;
rx_handler_func_t __rcu *rx_handler;
void __rcu *rx_handler_data;
@@ -4791,6 +4795,13 @@ static inline void netif_set_gso_max_seg
WRITE_ONCE(dev->gso_max_segs, segs);
}
+static inline void netif_set_gro_max_size(struct net_device *dev,
+ unsigned int size)
+{
+ /* This pairs with the READ_ONCE() in skb_gro_receive() */
+ WRITE_ONCE(dev->gro_max_size, size);
+}
+
static inline void skb_gso_error_unwind(struct sk_buff *skb, __be16 protocol,
int pulled_hlen, u16 mac_offset,
int mac_len)
--- a/include/uapi/linux/if_link.h
+++ b/include/uapi/linux/if_link.h
@@ -347,6 +347,7 @@ enum {
*/
IFLA_PARENT_DEV_NAME,
IFLA_PARENT_DEV_BUS_NAME,
+ IFLA_GRO_MAX_SIZE,
__IFLA_MAX
};
--- a/net/core/dev.c
+++ b/net/core/dev.c
@@ -10121,6 +10121,7 @@ struct net_device *alloc_netdev_mqs(int
dev->gso_max_size = GSO_MAX_SIZE;
dev->gso_max_segs = GSO_MAX_SEGS;
+ dev->gro_max_size = GRO_MAX_SIZE;
dev->upper_level = 1;
dev->lower_level = 1;
#ifdef CONFIG_LOCKDEP
--- a/net/core/gro.c
+++ b/net/core/gro.c
@@ -132,10 +132,14 @@ int skb_gro_receive(struct sk_buff *p, s
unsigned int headlen = skb_headlen(skb);
unsigned int len = skb_gro_len(skb);
unsigned int delta_truesize;
+ unsigned int gro_max_size;
unsigned int new_truesize;
struct sk_buff *lp;
- if (unlikely(p->len + len >= 65536 || NAPI_GRO_CB(skb)->flush))
+ /* pairs with WRITE_ONCE() in netif_set_gro_max_size() */
+ gro_max_size = READ_ONCE(p->dev->gro_max_size);
+
+ if (unlikely(p->len + len >= gro_max_size || NAPI_GRO_CB(skb)->flush))
return -E2BIG;
lp = NAPI_GRO_CB(p)->last;
--- a/net/core/rtnetlink.c
+++ b/net/core/rtnetlink.c
@@ -1037,6 +1037,7 @@ static noinline size_t if_nlmsg_size(con
+ nla_total_size(4) /* IFLA_NUM_RX_QUEUES */
+ nla_total_size(4) /* IFLA_GSO_MAX_SEGS */
+ nla_total_size(4) /* IFLA_GSO_MAX_SIZE */
+ + nla_total_size(4) /* IFLA_GRO_MAX_SIZE */
+ nla_total_size(1) /* IFLA_OPERSTATE */
+ nla_total_size(1) /* IFLA_LINKMODE */
+ nla_total_size(4) /* IFLA_CARRIER_CHANGES */
@@ -1739,6 +1740,7 @@ static int rtnl_fill_ifinfo(struct sk_bu
nla_put_u32(skb, IFLA_NUM_TX_QUEUES, dev->num_tx_queues) ||
nla_put_u32(skb, IFLA_GSO_MAX_SEGS, dev->gso_max_segs) ||
nla_put_u32(skb, IFLA_GSO_MAX_SIZE, dev->gso_max_size) ||
+ nla_put_u32(skb, IFLA_GRO_MAX_SIZE, dev->gro_max_size) ||
#ifdef CONFIG_RPS
nla_put_u32(skb, IFLA_NUM_RX_QUEUES, dev->num_rx_queues) ||
#endif
@@ -1891,6 +1893,7 @@ static const struct nla_policy ifla_poli
[IFLA_PROTO_DOWN_REASON] = { .type = NLA_NESTED },
[IFLA_NEW_IFINDEX] = NLA_POLICY_MIN(NLA_S32, 1),
[IFLA_PARENT_DEV_NAME] = { .type = NLA_NUL_STRING },
+ [IFLA_GRO_MAX_SIZE] = { .type = NLA_U32 },
};
static const struct nla_policy ifla_info_policy[IFLA_INFO_MAX+1] = {
@@ -2303,6 +2306,14 @@ static int validate_linkmsg(struct net_d
}
}
+ if (tb[IFLA_GRO_MAX_SIZE]) {
+ u32 gro_max_size = nla_get_u32(tb[IFLA_GRO_MAX_SIZE]);
+
+ if (gro_max_size > GRO_MAX_SIZE) {
+ NL_SET_ERR_MSG(extack, "too big gro_max_size");
+ return -EINVAL;
+ }
+ }
return 0;
}
@@ -2777,6 +2788,15 @@ static int do_setlink(const struct sk_bu
}
}
+ if (tb[IFLA_GRO_MAX_SIZE]) {
+ u32 gro_max_size = nla_get_u32(tb[IFLA_GRO_MAX_SIZE]);
+
+ if (dev->gro_max_size ^ gro_max_size) {
+ netif_set_gro_max_size(dev, gro_max_size);
+ status |= DO_SETLINK_MODIFIED;
+ }
+ }
+
if (tb[IFLA_OPERSTATE])
set_operstate(dev, nla_get_u8(tb[IFLA_OPERSTATE]));
@@ -3227,6 +3247,8 @@ struct net_device *rtnl_create_link(stru
netif_set_gso_max_size(dev, nla_get_u32(tb[IFLA_GSO_MAX_SIZE]));
if (tb[IFLA_GSO_MAX_SEGS])
netif_set_gso_max_segs(dev, nla_get_u32(tb[IFLA_GSO_MAX_SEGS]));
+ if (tb[IFLA_GRO_MAX_SIZE])
+ netif_set_gro_max_size(dev, nla_get_u32(tb[IFLA_GRO_MAX_SIZE]));
return dev;
}
--- a/tools/include/uapi/linux/if_link.h
+++ b/tools/include/uapi/linux/if_link.h
@@ -347,6 +347,7 @@ enum {
*/
IFLA_PARENT_DEV_NAME,
IFLA_PARENT_DEV_BUS_NAME,
+ IFLA_GRO_MAX_SIZE,
__IFLA_MAX
};