* Re: [PATCH net-next] ipx: move peII functions
From: David Miller @ 2012-07-19 17:49 UTC (permalink / raw)
To: shemminger; +Cc: acme, netdev
In-Reply-To: <20120718120948.67611bdd@s6510.linuxnetplumber.net>
From: Stephen Hemminger <shemminger@vyatta.com>
Date: Wed, 18 Jul 2012 12:09:48 -0700
> The Ethernet II wrapper is only used by IPX protocol, may have once
> been used by Appletalk but not currently. Therefore it makes sense to
> move it to the IPX dust bin and drop the exports.
>
> Build tested only.
>
> Signed-off-by: Stephen Hemminger <shemminger@vyatta.com>
Applied.
^ permalink raw reply
* Re: [PATCH V2] ipv6: fix incorrect route 'expires' value passed to userspace
From: David Miller @ 2012-07-19 17:49 UTC (permalink / raw)
To: lw; +Cc: netdev, shemminger
In-Reply-To: <50076AD3.1060604@cn.fujitsu.com>
From: Li Wei <lw@cn.fujitsu.com>
Date: Thu, 19 Jul 2012 10:02:59 +0800
>
> When userspace use RTM_GETROUTE to dump route table, with an already
> expired route entry, we always got an 'expires' value(2147157)
> calculated base on INT_MAX.
>
> The reason of this problem is in the following satement:
> rt->dst.expires - jiffies < INT_MAX
> gcc promoted the type of both sides of '<' to unsigned long, thus
> a small negative value would be considered greater than INT_MAX.
>
> This patch fix this by use the same trick as time_after macro to
> avoid the 'unsigned long' type promotion and deal with jiffies
> wrapping.
>
> Also we should do some fix in rtnl_put_cacheinfo() which use
> jiffies_to_clock_t(which take an unsigned long as parameter) to
> convert jiffies to clock_t to handle the negative expires.
>
> Signed-off-by: Li Wei <lw@cn.fujitsu.com>
Your patch is corrupted by your email client and therefore will
not apply cleanly.
I think this isn't the first time your patch submissions have
had this problem, and if so then you should do the necessary
work to prevent problem with more certainty in the future as
such this makes a lot of extra work for other people.
^ permalink raw reply
* Re: [PATCH] Crash in tun
From: David Miller @ 2012-07-19 17:47 UTC (permalink / raw)
To: maxk; +Cc: mikulas, eric.dumazet, netdev
In-Reply-To: <50084761.1030602@qualcomm.com>
From: Max Krasnyansky <maxk@qualcomm.com>
Date: Thu, 19 Jul 2012 10:44:01 -0700
> btw I don't remember now who added the socket business to tun_struct and why.
Is GIT really so broken on your computer that you can't find the
answer to this question in like 5 seconds as I just did?
commit 33dccbb050bbe35b88ca8cf1228dcf3e4d4b3554
Author: Herbert Xu <herbert@gondor.apana.org.au>
Date: Thu Feb 5 21:25:32 2009 -0800
tun: Limit amount of queued packets per device
Unlike a normal socket path, the tuntap device send path does
not have any accounting. This means that the user-space sender
may be able to pin down arbitrary amounts of kernel memory by
continuing to send data to an end-point that is congested.
Even when this isn't an issue because of limited queueing at
most end points, this can also be a problem because its only
response to congestion is packet loss. That is, when those
local queues at the end-point fills up, the tuntap device will
start wasting system time because it will continue to send
data there which simply gets dropped straight away.
Of course one could argue that everybody should do congestion
control end-to-end, unfortunately there are people in this world
still hooked on UDP, and they don't appear to be going away
anywhere fast. In fact, we've always helped them by performing
accounting in our UDP code, the sole purpose of which is to
provide congestion feedback other than through packet loss.
This patch attempts to apply the same bandaid to the tuntap device.
It creates a pseudo-socket object which is used to account our
packets just as a normal socket does for UDP. Of course things
are a little complex because we're actually reinjecting traffic
back into the stack rather than out of the stack.
The stack complexities however should have been resolved by preceding
patches. So this one can simply start using skb_set_owner_w.
For now the accounting is essentially disabled by default for
backwards compatibility. In particular, we set the cap to INT_MAX.
This is so that existing applications don't get confused by the
sudden arrival EAGAIN errors.
In future we may wish (or be forced to) do this by default.
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: David S. Miller <davem@davemloft.net>
diff --git a/drivers/net/tun.c b/drivers/net/tun.c
index 15d6763..0476549 100644
--- a/drivers/net/tun.c
+++ b/drivers/net/tun.c
@@ -64,6 +64,7 @@
#include <net/net_namespace.h>
#include <net/netns/generic.h>
#include <net/rtnetlink.h>
+#include <net/sock.h>
#include <asm/system.h>
#include <asm/uaccess.h>
@@ -95,6 +96,8 @@ struct tun_file {
wait_queue_head_t read_wait;
};
+struct tun_sock;
+
struct tun_struct {
struct tun_file *tfile;
unsigned int flags;
@@ -107,12 +110,24 @@ struct tun_struct {
struct fasync_struct *fasync;
struct tap_filter txflt;
+ struct sock *sk;
+ struct socket socket;
#ifdef TUN_DEBUG
int debug;
#endif
};
+struct tun_sock {
+ struct sock sk;
+ struct tun_struct *tun;
+};
+
+static inline struct tun_sock *tun_sk(struct sock *sk)
+{
+ return container_of(sk, struct tun_sock, sk);
+}
+
static int tun_attach(struct tun_struct *tun, struct file *file)
{
struct tun_file *tfile = file->private_data;
@@ -461,7 +476,8 @@ static unsigned int tun_chr_poll(struct file *file, poll_table * wait)
{
struct tun_file *tfile = file->private_data;
struct tun_struct *tun = __tun_get(tfile);
- unsigned int mask = POLLOUT | POLLWRNORM;
+ struct sock *sk = tun->sk;
+ unsigned int mask = 0;
if (!tun)
return POLLERR;
@@ -473,6 +489,11 @@ static unsigned int tun_chr_poll(struct file *file, poll_table * wait)
if (!skb_queue_empty(&tun->readq))
mask |= POLLIN | POLLRDNORM;
+ if (sock_writeable(sk) ||
+ (!test_and_set_bit(SOCK_ASYNC_NOSPACE, &sk->sk_socket->flags) &&
+ sock_writeable(sk)))
+ mask |= POLLOUT | POLLWRNORM;
+
if (tun->dev->reg_state != NETREG_REGISTERED)
mask = POLLERR;
@@ -482,66 +503,35 @@ static unsigned int tun_chr_poll(struct file *file, poll_table * wait)
/* prepad is the amount to reserve at front. len is length after that.
* linear is a hint as to how much to copy (usually headers). */
-static struct sk_buff *tun_alloc_skb(size_t prepad, size_t len, size_t linear,
- gfp_t gfp)
+static inline struct sk_buff *tun_alloc_skb(struct tun_struct *tun,
+ size_t prepad, size_t len,
+ size_t linear, int noblock)
{
+ struct sock *sk = tun->sk;
struct sk_buff *skb;
- unsigned int i;
-
- skb = alloc_skb(prepad + len, gfp|__GFP_NOWARN);
- if (skb) {
- skb_reserve(skb, prepad);
- skb_put(skb, len);
- return skb;
- }
+ int err;
/* Under a page? Don't bother with paged skb. */
if (prepad + len < PAGE_SIZE)
- return NULL;
+ linear = len;
- /* Start with a normal skb, and add pages. */
- skb = alloc_skb(prepad + linear, gfp);
+ skb = sock_alloc_send_pskb(sk, prepad + linear, len - linear, noblock,
+ &err);
if (!skb)
- return NULL;
+ return ERR_PTR(err);
skb_reserve(skb, prepad);
skb_put(skb, linear);
-
- len -= linear;
-
- for (i = 0; i < MAX_SKB_FRAGS; i++) {
- skb_frag_t *f = &skb_shinfo(skb)->frags[i];
-
- f->page = alloc_page(gfp|__GFP_ZERO);
- if (!f->page)
- break;
-
- f->page_offset = 0;
- f->size = PAGE_SIZE;
-
- skb->data_len += PAGE_SIZE;
- skb->len += PAGE_SIZE;
- skb->truesize += PAGE_SIZE;
- skb_shinfo(skb)->nr_frags++;
-
- if (len < PAGE_SIZE) {
- len = 0;
- break;
- }
- len -= PAGE_SIZE;
- }
-
- /* Too large, or alloc fail? */
- if (unlikely(len)) {
- kfree_skb(skb);
- skb = NULL;
- }
+ skb->data_len = len - linear;
+ skb->len += len - linear;
return skb;
}
/* Get packet from user space buffer */
-static __inline__ ssize_t tun_get_user(struct tun_struct *tun, struct iovec *iv, size_t count)
+static __inline__ ssize_t tun_get_user(struct tun_struct *tun,
+ struct iovec *iv, size_t count,
+ int noblock)
{
struct tun_pi pi = { 0, cpu_to_be16(ETH_P_IP) };
struct sk_buff *skb;
@@ -573,9 +563,11 @@ static __inline__ ssize_t tun_get_user(struct tun_struct *tun, struct iovec *iv,
return -EINVAL;
}
- if (!(skb = tun_alloc_skb(align, len, gso.hdr_len, GFP_KERNEL))) {
- tun->dev->stats.rx_dropped++;
- return -ENOMEM;
+ skb = tun_alloc_skb(tun, align, len, gso.hdr_len, noblock);
+ if (IS_ERR(skb)) {
+ if (PTR_ERR(skb) != -EAGAIN)
+ tun->dev->stats.rx_dropped++;
+ return PTR_ERR(skb);
}
if (skb_copy_datagram_from_iovec(skb, 0, iv, len)) {
@@ -661,7 +653,8 @@ static __inline__ ssize_t tun_get_user(struct tun_struct *tun, struct iovec *iv,
static ssize_t tun_chr_aio_write(struct kiocb *iocb, const struct iovec *iv,
unsigned long count, loff_t pos)
{
- struct tun_struct *tun = tun_get(iocb->ki_filp);
+ struct file *file = iocb->ki_filp;
+ struct tun_struct *tun = file->private_data;
ssize_t result;
if (!tun)
@@ -669,7 +662,8 @@ static ssize_t tun_chr_aio_write(struct kiocb *iocb, const struct iovec *iv,
DBG(KERN_INFO "%s: tun_chr_write %ld\n", tun->dev->name, count);
- result = tun_get_user(tun, (struct iovec *) iv, iov_length(iv, count));
+ result = tun_get_user(tun, (struct iovec *)iv, iov_length(iv, count),
+ file->f_flags & O_NONBLOCK);
tun_put(tun);
return result;
@@ -828,11 +822,40 @@ static struct rtnl_link_ops tun_link_ops __read_mostly = {
.validate = tun_validate,
};
+static void tun_sock_write_space(struct sock *sk)
+{
+ struct tun_struct *tun;
+
+ if (!sock_writeable(sk))
+ return;
+
+ if (sk->sk_sleep && waitqueue_active(sk->sk_sleep))
+ wake_up_interruptible_sync(sk->sk_sleep);
+
+ if (!test_and_clear_bit(SOCK_ASYNC_NOSPACE, &sk->sk_socket->flags))
+ return;
+
+ tun = container_of(sk, struct tun_sock, sk)->tun;
+ kill_fasync(&tun->fasync, SIGIO, POLL_OUT);
+}
+
+static void tun_sock_destruct(struct sock *sk)
+{
+ dev_put(container_of(sk, struct tun_sock, sk)->tun->dev);
+}
+
+static struct proto tun_proto = {
+ .name = "tun",
+ .owner = THIS_MODULE,
+ .obj_size = sizeof(struct tun_sock),
+};
static int tun_set_iff(struct net *net, struct file *file, struct ifreq *ifr)
{
+ struct sock *sk;
struct tun_struct *tun;
struct net_device *dev;
+ struct tun_file *tfile = file->private_data;
int err;
dev = __dev_get_by_name(net, ifr->ifr_name);
@@ -885,14 +908,31 @@ static int tun_set_iff(struct net *net, struct file *file, struct ifreq *ifr)
tun->flags = flags;
tun->txflt.count = 0;
+ err = -ENOMEM;
+ sk = sk_alloc(net, AF_UNSPEC, GFP_KERNEL, &tun_proto);
+ if (!sk)
+ goto err_free_dev;
+
+ /* This ref count is for tun->sk. */
+ dev_hold(dev);
+ sock_init_data(&tun->socket, sk);
+ sk->sk_write_space = tun_sock_write_space;
+ sk->sk_destruct = tun_sock_destruct;
+ sk->sk_sndbuf = INT_MAX;
+ sk->sk_sleep = &tfile->read_wait;
+
+ tun->sk = sk;
+ container_of(sk, struct tun_sock, sk)->tun = tun;
+
tun_net_init(dev);
if (strchr(dev->name, '%')) {
err = dev_alloc_name(dev, dev->name);
if (err < 0)
- goto err_free_dev;
+ goto err_free_sk;
}
+ err = -EINVAL;
err = register_netdevice(tun->dev);
if (err < 0)
goto err_free_dev;
@@ -928,6 +968,8 @@ static int tun_set_iff(struct net *net, struct file *file, struct ifreq *ifr)
strcpy(ifr->ifr_name, tun->dev->name);
return 0;
+ err_free_sk:
+ sock_put(sk);
err_free_dev:
free_netdev(dev);
failed:
@@ -1012,6 +1054,7 @@ static int tun_chr_ioctl(struct inode *inode, struct file *file,
struct tun_struct *tun;
void __user* argp = (void __user*)arg;
struct ifreq ifr;
+ int sndbuf;
int ret;
if (cmd == TUNSETIFF || _IOC_TYPE(cmd) == 0x89)
@@ -1151,6 +1194,22 @@ static int tun_chr_ioctl(struct inode *inode, struct file *file,
ret = dev_set_mac_address(tun->dev, &ifr.ifr_hwaddr);
rtnl_unlock();
break;
+
+ case TUNGETSNDBUF:
+ sndbuf = tun->sk->sk_sndbuf;
+ if (copy_to_user(argp, &sndbuf, sizeof(sndbuf)))
+ ret = -EFAULT;
+ break;
+
+ case TUNSETSNDBUF:
+ if (copy_from_user(&sndbuf, argp, sizeof(sndbuf))) {
+ ret = -EFAULT;
+ break;
+ }
+
+ tun->sk->sk_sndbuf = sndbuf;
+ break;
+
default:
ret = -EINVAL;
break;
@@ -1218,8 +1277,10 @@ static int tun_chr_close(struct inode *inode, struct file *file)
__tun_detach(tun);
/* If desireable, unregister the netdevice. */
- if (!(tun->flags & TUN_PERSIST))
+ if (!(tun->flags & TUN_PERSIST)) {
+ sock_put(tun->sk);
unregister_netdevice(tun->dev);
+ }
rtnl_unlock();
}
diff --git a/fs/compat_ioctl.c b/fs/compat_ioctl.c
index c8f8d59..c03c10d 100644
--- a/fs/compat_ioctl.c
+++ b/fs/compat_ioctl.c
@@ -1988,6 +1988,8 @@ COMPATIBLE_IOCTL(TUNSETGROUP)
COMPATIBLE_IOCTL(TUNGETFEATURES)
COMPATIBLE_IOCTL(TUNSETOFFLOAD)
COMPATIBLE_IOCTL(TUNSETTXFILTER)
+COMPATIBLE_IOCTL(TUNGETSNDBUF)
+COMPATIBLE_IOCTL(TUNSETSNDBUF)
/* Big V */
COMPATIBLE_IOCTL(VT_SETMODE)
COMPATIBLE_IOCTL(VT_GETMODE)
diff --git a/include/linux/if_tun.h b/include/linux/if_tun.h
index 8529f57..049d6c9 100644
--- a/include/linux/if_tun.h
+++ b/include/linux/if_tun.h
@@ -46,6 +46,8 @@
#define TUNSETOFFLOAD _IOW('T', 208, unsigned int)
#define TUNSETTXFILTER _IOW('T', 209, unsigned int)
#define TUNGETIFF _IOR('T', 210, unsigned int)
+#define TUNGETSNDBUF _IOR('T', 211, int)
+#define TUNSETSNDBUF _IOW('T', 212, int)
/* TUNSETIFF ifr flags */
#define IFF_TUN 0x0001
^ permalink raw reply related
* Re: [PATCH] Crash in tun
From: Max Krasnyansky @ 2012-07-19 17:44 UTC (permalink / raw)
To: Mikulas Patocka; +Cc: Eric Dumazet, netdev, davem
In-Reply-To: <alpine.DEB.2.00.1207191746170.7550@artax.karlin.mff.cuni.cz>
On 07/19/2012 09:13 AM, Mikulas Patocka wrote:
>
>
> On Thu, 19 Jul 2012, Eric Dumazet wrote:
>
>> Hi Mikulas
>>
>> A fix for this problem is : http://patchwork.ozlabs.org/patch/170440/
>
> If you call tun_free_netdev beacuse of a jump to an error label
> err_free_sk, your patch still calls it with NULL file, causing a memory
> corruption and a possible crash.
>
> Your patch doesn't fix sockets_in_use underflow.
>
> Maybe we can commit this patch --- it introduces a new flag
> SOCK_EXTERNALLY_ALLOCATED to work around both problems. (it looks quite
> nicer than my previous patch with file = (void *)1).
I definitely like this second version better. Less hacky an all.
btw I don't remember now who added the socket business to tun_struct and why.
It seems to be messy in general. Originally version, back when I was still paying attention to it,
didn't have sockets at all. Only char and net devices. It was much cleaner.
Max
> ---
>
> tun: fix a crash bug and a memory leak
>
> This patch fixes a crash
> tun_chr_close -> netdev_run_todo -> tun_free_netdev -> sk_release_kernel ->
> sock_release -> iput(SOCK_INODE(sock))
> introduced by commit 1ab5ecb90cb6a3df1476e052f76a6e8f6511cb3d
>
> The problem is that this socket is embedded in struct tun_struct, it has
> no inode, iput is called on invalid inode, which modifies invalid memory
> and optionally causes a crash.
>
> sock_release also decrements sockets_in_use, this causes a bug that
> "sockets: used" field in /proc/*/net/sockstat keeps on decreasing when
> creating and closing tun devices.
>
> This patch introduces a flag SOCK_EXTERNALLY_ALLOCATED that instructs
> sock_release to not free the inode and not decrement sockets_in_use,
> fixing both memory corruption and sockets_in_use underflow.
>
> It should be backported to 3.3 an 3.4 stabke.
>
> Signed-off-by: Mikulas Patocka <mikulas@artax.karlin.mff.cuni.cz>
> Cc: stable@kernel.org
>
> ---
> drivers/net/tun.c | 3 +++
> include/linux/net.h | 1 +
> net/socket.c | 3 +++
> 3 files changed, 7 insertions(+)
>
> Index: linux-3.4.5-fast/drivers/net/tun.c
> ===================================================================
> --- linux-3.4.5-fast.orig/drivers/net/tun.c 2012-07-19 17:55:16.000000000 +0200
> +++ linux-3.4.5-fast/drivers/net/tun.c 2012-07-19 17:58:30.000000000 +0200
> @@ -358,6 +358,8 @@ static void tun_free_netdev(struct net_d
> {
> struct tun_struct *tun = netdev_priv(dev);
>
> + BUG_ON(!test_bit(SOCK_EXTERNALLY_ALLOCATED, &tun->socket.flags));
> +
> sk_release_kernel(tun->socket.sk);
> }
>
> @@ -1115,6 +1117,7 @@ static int tun_set_iff(struct net *net,
> tun->flags = flags;
> tun->txflt.count = 0;
> tun->vnet_hdr_sz = sizeof(struct virtio_net_hdr);
> + set_bit(SOCK_EXTERNALLY_ALLOCATED, &tun->socket.flags);
>
> err = -ENOMEM;
> sk = sk_alloc(&init_net, AF_UNSPEC, GFP_KERNEL, &tun_proto);
> Index: linux-3.4.5-fast/include/linux/net.h
> ===================================================================
> --- linux-3.4.5-fast.orig/include/linux/net.h 2012-07-19 17:54:31.000000000 +0200
> +++ linux-3.4.5-fast/include/linux/net.h 2012-07-19 17:55:03.000000000 +0200
> @@ -72,6 +72,7 @@ struct net;
> #define SOCK_NOSPACE 2
> #define SOCK_PASSCRED 3
> #define SOCK_PASSSEC 4
> +#define SOCK_EXTERNALLY_ALLOCATED 5
>
> #ifndef ARCH_HAS_SOCKET_TYPES
> /**
> Index: linux-3.4.5-fast/net/socket.c
> ===================================================================
> --- linux-3.4.5-fast.orig/net/socket.c 2012-07-19 17:56:55.000000000 +0200
> +++ linux-3.4.5-fast/net/socket.c 2012-07-19 17:57:50.000000000 +0200
> @@ -522,6 +522,9 @@ void sock_release(struct socket *sock)
> if (rcu_dereference_protected(sock->wq, 1)->fasync_list)
> printk(KERN_ERR "sock_release: fasync list not empty!\n");
>
> + if (test_bit(SOCK_EXTERNALLY_ALLOCATED, &sock->flags))
> + return;
> +
> percpu_sub(sockets_in_use, 1);
> if (!sock->file) {
> iput(SOCK_INODE(sock));
>
^ permalink raw reply
* [PATCH] net: Fix warnings in dst_ops.h
From: David Miller @ 2012-07-19 17:43 UTC (permalink / raw)
To: netdev
include/net/dst_ops.h:28:20: warning: ‘struct sock’ declared inside parameter list
Signed-off-by: David S. Miller <davem@davemloft.net>
---
include/net/dst_ops.h | 1 +
1 file changed, 1 insertion(+)
diff --git a/include/net/dst_ops.h b/include/net/dst_ops.h
index d079fc6..2f26dfb 100644
--- a/include/net/dst_ops.h
+++ b/include/net/dst_ops.h
@@ -8,6 +8,7 @@ struct dst_entry;
struct kmem_cachep;
struct net_device;
struct sk_buff;
+struct sock;
struct dst_ops {
unsigned short family;
--
1.7.10.4
^ permalink raw reply related
* Re: [PATCH v3 2/7] net-tcp: Fast Open client - cookie cache
From: Eric Dumazet @ 2012-07-19 17:41 UTC (permalink / raw)
To: Yuchung Cheng; +Cc: davem, hkchu, edumazet, ncardwell, sivasankar, netdev
In-Reply-To: <1342716191-19196-3-git-send-email-ycheng@google.com>
On Thu, 2012-07-19 at 09:43 -0700, Yuchung Cheng wrote:
> With help from Eric Dumazet, add Fast Open metrics in tcp metrics cache.
> The basic ones are MSS and the cookies. Later patch will cache more to
> handle unfriendly middleboxes.
>
> Signed-off-by: Yuchung Cheng <ycheng@google.com>
> ---
> include/net/tcp.h | 4 +++
> net/ipv4/tcp_metrics.c | 51 ++++++++++++++++++++++++++++++++++++++++++++++++
> 2 files changed, 55 insertions(+), 0 deletions(-)
Acked-by: Eric Dumazet <edumazet@google.com>
^ permalink raw reply
* Re: [PATCH net-next v2] ipv4: tcp: remove per net tcp_sock
From: David Miller @ 2012-07-19 17:36 UTC (permalink / raw)
To: eric.dumazet; +Cc: netdev, therbert, wsommerfeld
In-Reply-To: <1342719243.2626.4571.camel@edumazet-glaptop>
From: Eric Dumazet <eric.dumazet@gmail.com>
Date: Thu, 19 Jul 2012 19:34:03 +0200
> v2 : move unicast_sock out of ip_send_unicast_reply() body
> init sk_refcnt to 1, in case some driver get/put a reference on
> socket.
The compiler seems much happier with this, applied, thanks Eric :-)
^ permalink raw reply
* [PATCH net-next v2] ipv4: tcp: remove per net tcp_sock
From: Eric Dumazet @ 2012-07-19 17:34 UTC (permalink / raw)
To: David Miller; +Cc: netdev, Tom Herbert, Bill Sommerfeld
From: Eric Dumazet <edumazet@google.com>
tcp_v4_send_reset() and tcp_v4_send_ack() use a single socket
per network namespace.
This leads to bad behavior on multiqueue NICS, because many cpus
contend for the socket lock and once socket lock is acquired, extra
false sharing on various socket fields slow down the operations.
To better resist to attacks, we use a percpu socket. Each cpu can
run without contention, using appropriate memory (local node)
Additional features :
1) We also mirror the queue_mapping of the incoming skb, so that
answers use the same queue if possible.
2) Setting SOCK_USE_WRITE_QUEUE socket flag speedup sock_wfree()
3) We now limit the number of in-flight RST/ACK [1] packets
per cpu, instead of per namespace, and we honor the sysctl_wmem_default
limit dynamically. (Prior to this patch, sysctl_wmem_default value was
copied at boot time, so any further change would not affect tcp_sock
limit)
[1] These packets are only generated when no socket was matched for
the incoming packet.
Reported-by: Bill Sommerfeld <wsommerfeld@google.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Tom Herbert <therbert@google.com>
---
v2 : move unicast_sock out of ip_send_unicast_reply() body
init sk_refcnt to 1, in case some driver get/put a reference on
socket.
include/net/ip.h | 2 -
include/net/netns/ipv4.h | 1
net/ipv4/ip_output.c | 50 +++++++++++++++++++++++--------------
net/ipv4/tcp_ipv4.c | 8 ++---
4 files changed, 36 insertions(+), 25 deletions(-)
diff --git a/include/net/ip.h b/include/net/ip.h
index ec5cfde..bd5e444 100644
--- a/include/net/ip.h
+++ b/include/net/ip.h
@@ -158,7 +158,7 @@ static inline __u8 ip_reply_arg_flowi_flags(const struct ip_reply_arg *arg)
return (arg->flags & IP_REPLY_ARG_NOSRCCHECK) ? FLOWI_FLAG_ANYSRC : 0;
}
-void ip_send_unicast_reply(struct sock *sk, struct sk_buff *skb, __be32 daddr,
+void ip_send_unicast_reply(struct net *net, struct sk_buff *skb, __be32 daddr,
__be32 saddr, const struct ip_reply_arg *arg,
unsigned int len);
diff --git a/include/net/netns/ipv4.h b/include/net/netns/ipv4.h
index 2e089a9..d909c7f 100644
--- a/include/net/netns/ipv4.h
+++ b/include/net/netns/ipv4.h
@@ -38,7 +38,6 @@ struct netns_ipv4 {
struct sock *fibnl;
struct sock **icmp_sk;
- struct sock *tcp_sock;
struct inet_peer_base *peers;
struct tcpm_hash_bucket *tcp_metrics_hash;
unsigned int tcp_metrics_hash_mask;
diff --git a/net/ipv4/ip_output.c b/net/ipv4/ip_output.c
index cc52679..c528f84 100644
--- a/net/ipv4/ip_output.c
+++ b/net/ipv4/ip_output.c
@@ -1463,20 +1463,33 @@ static int ip_reply_glue_bits(void *dptr, char *to, int offset,
/*
* Generic function to send a packet as reply to another packet.
- * Used to send TCP resets so far.
+ * Used to send some TCP resets/acks so far.
*
- * Should run single threaded per socket because it uses the sock
- * structure to pass arguments.
+ * Use a fake percpu inet socket to avoid false sharing and contention.
*/
-void ip_send_unicast_reply(struct sock *sk, struct sk_buff *skb, __be32 daddr,
+static DEFINE_PER_CPU(struct inet_sock, unicast_sock) = {
+ .sk = {
+ .__sk_common = {
+ .skc_refcnt = ATOMIC_INIT(1),
+ },
+ .sk_wmem_alloc = ATOMIC_INIT(1),
+ .sk_allocation = GFP_ATOMIC,
+ .sk_flags = (1UL << SOCK_USE_WRITE_QUEUE),
+ },
+ .pmtudisc = IP_PMTUDISC_WANT,
+};
+
+void ip_send_unicast_reply(struct net *net, struct sk_buff *skb, __be32 daddr,
__be32 saddr, const struct ip_reply_arg *arg,
unsigned int len)
{
- struct inet_sock *inet = inet_sk(sk);
struct ip_options_data replyopts;
struct ipcm_cookie ipc;
struct flowi4 fl4;
struct rtable *rt = skb_rtable(skb);
+ struct sk_buff *nskb;
+ struct sock *sk;
+ struct inet_sock *inet;
if (ip_options_echo(&replyopts.opt.opt, skb))
return;
@@ -1494,38 +1507,39 @@ void ip_send_unicast_reply(struct sock *sk, struct sk_buff *skb, __be32 daddr,
flowi4_init_output(&fl4, arg->bound_dev_if, 0,
RT_TOS(arg->tos),
- RT_SCOPE_UNIVERSE, sk->sk_protocol,
+ RT_SCOPE_UNIVERSE, ip_hdr(skb)->protocol,
ip_reply_arg_flowi_flags(arg),
daddr, saddr,
tcp_hdr(skb)->source, tcp_hdr(skb)->dest);
security_skb_classify_flow(skb, flowi4_to_flowi(&fl4));
- rt = ip_route_output_key(sock_net(sk), &fl4);
+ rt = ip_route_output_key(net, &fl4);
if (IS_ERR(rt))
return;
- /* And let IP do all the hard work.
+ inet = &get_cpu_var(unicast_sock);
- This chunk is not reenterable, hence spinlock.
- Note that it uses the fact, that this function is called
- with locally disabled BH and that sk cannot be already spinlocked.
- */
- bh_lock_sock(sk);
inet->tos = arg->tos;
+ sk = &inet->sk;
sk->sk_priority = skb->priority;
sk->sk_protocol = ip_hdr(skb)->protocol;
sk->sk_bound_dev_if = arg->bound_dev_if;
+ sock_net_set(sk, net);
+ __skb_queue_head_init(&sk->sk_write_queue);
+ sk->sk_sndbuf = sysctl_wmem_default;
ip_append_data(sk, &fl4, ip_reply_glue_bits, arg->iov->iov_base, len, 0,
&ipc, &rt, MSG_DONTWAIT);
- if ((skb = skb_peek(&sk->sk_write_queue)) != NULL) {
+ nskb = skb_peek(&sk->sk_write_queue);
+ if (nskb) {
if (arg->csumoffset >= 0)
- *((__sum16 *)skb_transport_header(skb) +
- arg->csumoffset) = csum_fold(csum_add(skb->csum,
+ *((__sum16 *)skb_transport_header(nskb) +
+ arg->csumoffset) = csum_fold(csum_add(nskb->csum,
arg->csum));
- skb->ip_summed = CHECKSUM_NONE;
+ nskb->ip_summed = CHECKSUM_NONE;
+ skb_set_queue_mapping(nskb, skb_get_queue_mapping(skb));
ip_push_pending_frames(sk, &fl4);
}
- bh_unlock_sock(sk);
+ put_cpu_var(unicast_sock);
ip_rt_put(rt);
}
diff --git a/net/ipv4/tcp_ipv4.c b/net/ipv4/tcp_ipv4.c
index d9caf5c..d7d2fa5 100644
--- a/net/ipv4/tcp_ipv4.c
+++ b/net/ipv4/tcp_ipv4.c
@@ -688,7 +688,7 @@ static void tcp_v4_send_reset(struct sock *sk, struct sk_buff *skb)
net = dev_net(skb_dst(skb)->dev);
arg.tos = ip_hdr(skb)->tos;
- ip_send_unicast_reply(net->ipv4.tcp_sock, skb, ip_hdr(skb)->saddr,
+ ip_send_unicast_reply(net, skb, ip_hdr(skb)->saddr,
ip_hdr(skb)->daddr, &arg, arg.iov[0].iov_len);
TCP_INC_STATS_BH(net, TCP_MIB_OUTSEGS);
@@ -771,7 +771,7 @@ static void tcp_v4_send_ack(struct sk_buff *skb, u32 seq, u32 ack,
if (oif)
arg.bound_dev_if = oif;
arg.tos = tos;
- ip_send_unicast_reply(net->ipv4.tcp_sock, skb, ip_hdr(skb)->saddr,
+ ip_send_unicast_reply(net, skb, ip_hdr(skb)->saddr,
ip_hdr(skb)->daddr, &arg, arg.iov[0].iov_len);
TCP_INC_STATS_BH(net, TCP_MIB_OUTSEGS);
@@ -2624,13 +2624,11 @@ EXPORT_SYMBOL(tcp_prot);
static int __net_init tcp_sk_init(struct net *net)
{
- return inet_ctl_sock_create(&net->ipv4.tcp_sock,
- PF_INET, SOCK_RAW, IPPROTO_TCP, net);
+ return 0;
}
static void __net_exit tcp_sk_exit(struct net *net)
{
- inet_ctl_sock_destroy(net->ipv4.tcp_sock);
}
static void __net_exit tcp_sk_exit_batch(struct list_head *net_exit_list)
^ permalink raw reply related
* Re: [PATCH v3] ipv4: use seqlock for nh_exceptions
From: David Miller @ 2012-07-19 17:30 UTC (permalink / raw)
To: ja; +Cc: netdev
In-Reply-To: <1342642535-2545-1-git-send-email-ja@ssi.bg>
From: Julian Anastasov <ja@ssi.bg>
Date: Wed, 18 Jul 2012 23:15:35 +0300
> Use global seqlock for the nh_exceptions. Call
> fnhe_oldest with the right hash chain. Correct the diff
> value for dst_set_expires.
>
> v2: after suggestions from Eric Dumazet:
> * get rid of spin lock fnhe_lock, rearrange update_or_create_fnhe
> * continue daddr search in rt_bind_exception
>
> v3:
> * remove the daddr check before seqlock in rt_bind_exception
> * restart lookup in rt_bind_exception on detected seqlock change,
> as suggested by David Miller
>
> Signed-off-by: Julian Anastasov <ja@ssi.bg>
Applied, thanks a lot Julian.
^ permalink raw reply
* Re: [PATCH net-next] ipv4: tcp: remove per net tcp_sock
From: Eric Dumazet @ 2012-07-19 17:22 UTC (permalink / raw)
To: David Miller; +Cc: netdev, therbert, wsommerfeld
In-Reply-To: <20120719.101225.2227860476502999476.davem@davemloft.net>
On Thu, 2012-07-19 at 10:12 -0700, David Miller wrote:
> Just give this thing a unique name and define it right before the function.
Yep, will do that.
I also will init sk_refcnt to one, just in case a driver wants to get a
reference on socket (and free it later).
^ permalink raw reply
* Re: [PATCH] sfc: initialize dynamic sysfs attributes for lockdep
From: Ben Hutchings @ 2012-07-19 17:19 UTC (permalink / raw)
To: Michal Schmidt; +Cc: netdev, Solarflare linux maintainers
In-Reply-To: <1342717485-24034-1-git-send-email-mschmidt@redhat.com>
On Thu, 2012-07-19 at 19:04 +0200, Michal Schmidt wrote:
> Dynamically allocated sysfs attributes must be initialized using
> sysfs_attr_init(), otherwise lockdep complains:
> BUG: key <address> not in .data!
>
> Signed-off-by: Michal Schmidt <mschmidt@redhat.com>
Acked-by: Ben Hutchings <bhutchings@solarflare.com>
> ---
> drivers/net/ethernet/sfc/mcdi_mon.c | 1 +
> 1 file changed, 1 insertion(+)
>
> diff --git a/drivers/net/ethernet/sfc/mcdi_mon.c b/drivers/net/ethernet/sfc/mcdi_mon.c
> index fb7f65b..1d552f0 100644
> --- a/drivers/net/ethernet/sfc/mcdi_mon.c
> +++ b/drivers/net/ethernet/sfc/mcdi_mon.c
> @@ -222,6 +222,7 @@ efx_mcdi_mon_add_attr(struct efx_nic *efx, const char *name,
> attr->index = index;
> attr->type = type;
> attr->limit_value = limit_value;
> + sysfs_attr_init(&attr->dev_attr.attr);
> attr->dev_attr.attr.name = attr->name;
> attr->dev_attr.attr.mode = S_IRUGO;
> attr->dev_attr.show = reader;
--
Ben Hutchings, Staff Engineer, Solarflare
Not speaking for my employer; that's the marketing department's job.
They asked us to note that Solarflare product names are trademarked.
^ permalink raw reply
* Re: [PATCH net-next] ipv4: tcp: remove per net tcp_sock
From: David Miller @ 2012-07-19 17:12 UTC (permalink / raw)
To: eric.dumazet; +Cc: netdev, therbert, wsommerfeld
In-Reply-To: <20120719.101155.1970854797296520147.davem@davemloft.net>
From: David Miller <davem@davemloft.net>
Date: Thu, 19 Jul 2012 10:11:55 -0700 (PDT)
> From: Eric Dumazet <eric.dumazet@gmail.com>
> Date: Thu, 19 Jul 2012 19:07:47 +0200
>
>> On Thu, 2012-07-19 at 08:45 -0700, David Miller wrote:
>>> From: David Miller <davem@davemloft.net>
>>> Date: Thu, 19 Jul 2012 08:35:44 -0700 (PDT)
>>>
>>> > Looks great, applied, thanks Eric.
>>>
>>> I take that back, it doesn't build:
>>>
>>> net/ipv4/ip_output.c: In function ‘ip_send_unicast_reply’:
>>> net/ipv4/ip_output.c:1481:1: error: section attribute cannot be specified for local variables
>>> net/ipv4/ip_output.c:1481:1: error: section attribute cannot be specified for local variables
>>> net/ipv4/ip_output.c:1481:1: error: declaration of ‘__pcpu_unique_unicast_sock’ with no linkage follows extern declaration
>>> net/ipv4/ip_output.c:1481:1: note: previous declaration of ‘__pcpu_unique_unicast_sock’ was here
>>> net/ipv4/ip_output.c:1481:9: error: section attribute cannot be specified for local variables
>>> net/ipv4/ip_output.c:1481:9: error: weak declaration of ‘unicast_sock’ must be public
>>
>> Strange, it builds on my machines, and I got nice performance boost.
>>
>> Apparently your arch doesnt handle the
>>
>> static DEFINE_PER_CPU(struct inet_sock, unicast_sock)
>>
>> in the function body ?
>
> It's x86-64, standard Fedora 17 install.
Just give this thing a unique name and define it right before the function.
^ permalink raw reply
* Re: [PATCH net-next] ipv4: tcp: remove per net tcp_sock
From: David Miller @ 2012-07-19 17:11 UTC (permalink / raw)
To: eric.dumazet; +Cc: netdev, therbert, wsommerfeld
In-Reply-To: <1342717667.2626.4494.camel@edumazet-glaptop>
From: Eric Dumazet <eric.dumazet@gmail.com>
Date: Thu, 19 Jul 2012 19:07:47 +0200
> On Thu, 2012-07-19 at 08:45 -0700, David Miller wrote:
>> From: David Miller <davem@davemloft.net>
>> Date: Thu, 19 Jul 2012 08:35:44 -0700 (PDT)
>>
>> > Looks great, applied, thanks Eric.
>>
>> I take that back, it doesn't build:
>>
>> net/ipv4/ip_output.c: In function ‘ip_send_unicast_reply’:
>> net/ipv4/ip_output.c:1481:1: error: section attribute cannot be specified for local variables
>> net/ipv4/ip_output.c:1481:1: error: section attribute cannot be specified for local variables
>> net/ipv4/ip_output.c:1481:1: error: declaration of ‘__pcpu_unique_unicast_sock’ with no linkage follows extern declaration
>> net/ipv4/ip_output.c:1481:1: note: previous declaration of ‘__pcpu_unique_unicast_sock’ was here
>> net/ipv4/ip_output.c:1481:9: error: section attribute cannot be specified for local variables
>> net/ipv4/ip_output.c:1481:9: error: weak declaration of ‘unicast_sock’ must be public
>
> Strange, it builds on my machines, and I got nice performance boost.
>
> Apparently your arch doesnt handle the
>
> static DEFINE_PER_CPU(struct inet_sock, unicast_sock)
>
> in the function body ?
It's x86-64, standard Fedora 17 install.
^ permalink raw reply
* Re: [PATCH net/for-next V1 1/1] IB/ipoib: break linkage to neighbouring system
From: David Miller @ 2012-07-19 17:08 UTC (permalink / raw)
To: ogerlitz-VPRAkNaXOzVWk0Htik3J/w
Cc: cl-vYTEC60ixJUAvxtiuMwx3w, shlomop-VPRAkNaXOzVWk0Htik3J/w,
roland-DgEjT+Ai2ygdnm+yROfE0A, linux-rdma-u79uwXL29TY76Z2rM5mHXA,
erezsh-VPRAkNaXOzVWk0Htik3J/w, netdev-u79uwXL29TY76Z2rM5mHXA
In-Reply-To: <500833D9.8000001-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org>
From: Or Gerlitz <ogerlitz-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org>
Date: Thu, 19 Jul 2012 19:20:41 +0300
> On 7/19/2012 6:24 PM, Christoph Lameter wrote:
>> On Thu, 19 Jul 2012, Shlomo Pongartz wrote:
>>
>>> The garbage collection and stale times follow the default ipv4/6
>>> neigh.default.gc_yyy
>>> sysctl values, for example
>>>
>>> net.ipv4.neigh.default.gc_interval = 30
>>> net.ipv4.neigh.default.gc_stale_time = 60
>>>
>>> If given access to these values from IPoIB, we will be happy
>>> to integrate them into that logic
>>
>> It looks like the values are hardcoded right now.
>
> Two points here,
>
> 1s, they are indeed hard-coded since there's no define/enum
> that holds their default values (or maybe we should add one now?), see
> this code snippest from net/ipv4/arp.c
These numbers come from the IPV6 Neighbour Discovery RFCs. IPV4
replicates the Neighbour Unreachability Detection schemes of IPV6 in
pretty much it's entirety, and therefore takes on the same timeout et
al. parameters.
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply
* Re: [PATCH net-next] ipv4: tcp: remove per net tcp_sock
From: Eric Dumazet @ 2012-07-19 17:08 UTC (permalink / raw)
To: David Miller; +Cc: netdev, therbert, wsommerfeld
In-Reply-To: <20120719.083544.1223522161508413373.davem@davemloft.net>
On Thu, 2012-07-19 at 08:35 -0700, David Miller wrote:
> > @@ -2624,13 +2624,11 @@ EXPORT_SYMBOL(tcp_prot);
> >
> > static int __net_init tcp_sk_init(struct net *net)
> > {
> > - return inet_ctl_sock_create(&net->ipv4.tcp_sock,
> > - PF_INET, SOCK_RAW, IPPROTO_TCP, net);
> > + return 0;
> > }
> >
> > static void __net_exit tcp_sk_exit(struct net *net)
> > {
> > - inet_ctl_sock_destroy(net->ipv4.tcp_sock);
> > }
> >
> > static void __net_exit tcp_sk_exit_batch(struct list_head *net_exit_list)
>
> If these no longer really do anything, just send me a patch to kill
> them off entirely.
>
We cant remove them, because of tcp_sk_exit_batch()
^ permalink raw reply
* Re: [PATCH net-next] ipv4: tcp: remove per net tcp_sock
From: Eric Dumazet @ 2012-07-19 17:07 UTC (permalink / raw)
To: David Miller; +Cc: netdev, therbert, wsommerfeld
In-Reply-To: <20120719.084503.136853790544876708.davem@davemloft.net>
On Thu, 2012-07-19 at 08:45 -0700, David Miller wrote:
> From: David Miller <davem@davemloft.net>
> Date: Thu, 19 Jul 2012 08:35:44 -0700 (PDT)
>
> > Looks great, applied, thanks Eric.
>
> I take that back, it doesn't build:
>
> net/ipv4/ip_output.c: In function ‘ip_send_unicast_reply’:
> net/ipv4/ip_output.c:1481:1: error: section attribute cannot be specified for local variables
> net/ipv4/ip_output.c:1481:1: error: section attribute cannot be specified for local variables
> net/ipv4/ip_output.c:1481:1: error: declaration of ‘__pcpu_unique_unicast_sock’ with no linkage follows extern declaration
> net/ipv4/ip_output.c:1481:1: note: previous declaration of ‘__pcpu_unique_unicast_sock’ was here
> net/ipv4/ip_output.c:1481:9: error: section attribute cannot be specified for local variables
> net/ipv4/ip_output.c:1481:9: error: weak declaration of ‘unicast_sock’ must be public
Strange, it builds on my machines, and I got nice performance boost.
Apparently your arch doesnt handle the
static DEFINE_PER_CPU(struct inet_sock, unicast_sock)
in the function body ?
^ permalink raw reply
* [PATCH] sfc: initialize dynamic sysfs attributes for lockdep
From: Michal Schmidt @ 2012-07-19 17:04 UTC (permalink / raw)
To: netdev; +Cc: Ben Hutchings, Solarflare linux maintainers
Dynamically allocated sysfs attributes must be initialized using
sysfs_attr_init(), otherwise lockdep complains:
BUG: key <address> not in .data!
Signed-off-by: Michal Schmidt <mschmidt@redhat.com>
---
drivers/net/ethernet/sfc/mcdi_mon.c | 1 +
1 file changed, 1 insertion(+)
diff --git a/drivers/net/ethernet/sfc/mcdi_mon.c b/drivers/net/ethernet/sfc/mcdi_mon.c
index fb7f65b..1d552f0 100644
--- a/drivers/net/ethernet/sfc/mcdi_mon.c
+++ b/drivers/net/ethernet/sfc/mcdi_mon.c
@@ -222,6 +222,7 @@ efx_mcdi_mon_add_attr(struct efx_nic *efx, const char *name,
attr->index = index;
attr->type = type;
attr->limit_value = limit_value;
+ sysfs_attr_init(&attr->dev_attr.attr);
attr->dev_attr.attr.name = attr->name;
attr->dev_attr.attr.mode = S_IRUGO;
attr->dev_attr.show = reader;
--
1.7.10.4
^ permalink raw reply related
* Re: [PATCH net-next] asix: AX88172A driver depends on phylib
From: Fengguang Wu @ 2012-07-19 17:01 UTC (permalink / raw)
To: Christian Riesch; +Cc: netdev, kernel-janitors
In-Reply-To: <1342699339-13871-1-git-send-email-christian.riesch@omicron.at>
On Thu, Jul 19, 2012 at 02:02:19PM +0200, Christian Riesch wrote:
> Since commit 16626b0cc3d5afe250850f96759b241f8a403b52 the asix
> driver depends on the phylib. Select phylib when the asix driver is
> selected.
Tested-by: Fengguang Wu <fengguang.wu@intel.com>
^ permalink raw reply
* [PATCH] bridge: update documentation references
From: Stephen Hemminger @ 2012-07-19 17:01 UTC (permalink / raw)
To: David Miller; +Cc: netdev
Update the references to bridge utilities and web pages
to current locations
Signed-off-by: Stephen Hemminger <shemminger@vyatta.com>
--- a/Documentation/networking/bridge.txt 2012-07-16 12:08:02.743695241 -0700
+++ b/Documentation/networking/bridge.txt 2012-07-19 09:43:47.955910072 -0700
@@ -1,7 +1,14 @@
In order to use the Ethernet bridging functionality, you'll need the
-userspace tools. These programs and documentation are available
-at http://www.linuxfoundation.org/en/Net:Bridge. The download page is
-http://prdownloads.sourceforge.net/bridge.
+userspace tools.
+
+Documentation for Linux bridging is on:
+ http://www.linuxfoundation.org/collaborate/workgroups/networking/bridge
+
+The bridge-utilities are maintained at:
+ git://git.kernel.org/pub/scm/linux/kernel/git/shemminger/bridge-utils.git
+
+Additionally, the iproute2 utilities can be used to configure
+bridge devices.
If you still have questions, don't hesitate to post to the mailing list
(more info https://lists.linux-foundation.org/mailman/listinfo/bridge).
^ permalink raw reply
* Re: [PATCH v2] sctp: Implement quick failover draft from tsvwg
From: Joe Perches @ 2012-07-19 16:54 UTC (permalink / raw)
To: Neil Horman
Cc: netdev, Vlad Yasevich, Sridhar Samudrala, David S. Miller,
linux-sctp
In-Reply-To: <20120719104513.GB2070@hmsreliant.think-freely.org>
On Thu, 2012-07-19 at 06:45 -0400, Neil Horman wrote:
> On Wed, Jul 18, 2012 at 01:30:58PM -0700, Joe Perches wrote:
> > On Wed, 2012-07-18 at 14:01 -0400, Neil Horman wrote:
> > > I've seen several attempts recently made to do quick failover of sctp transports
> > > by reducing various retransmit timers and counters. While its possible to
> > > implement a faster failover on multihomed sctp associations, its not
> > > particularly robust, in that it can lead to unneeded retransmits, as well as
> > > false connection failures due to intermittent latency on a network.
[]
> > > @@ -878,12 +896,15 @@ void sctp_assoc_control_transport(struct sctp_association *asoc,
> > []
> > > + if (ulp_notify) {
> > > + memset(&addr, 0, sizeof(struct sockaddr_storage));
> > > + memcpy(&addr, &transport->ipaddr,
> > > + transport->af_specific->sockaddr_len);
> >
> > Perhaps it's better to do the memcpy then the memset of the
> > space left instead.
> >
> > memcpy(&addr, &transport->ipaddr, transport->af_specific->sockaddr_len);
> > memset((char *)&addr) + transport->af_specific->sockaddr_len, 0,
> > sizeof(struct sockaddr_storage) - transport->af_specific->sockaddr_len);
> >
> hmm, not sure about that. It works either way for me, but I've not changed that
> code, just the condition under which it was executed. I'd rather save cleanups
> like that for a separate patch if you don't mind.
Not a bit.
It's almost certain reversing the order is slower for v4
addresses anyway. It might be slower for v6 too given
the arithmetic.
cheers, Joe
^ permalink raw reply
* [PATCH v3] sctp: Implement quick failover draft from tsvwg
From: Neil Horman @ 2012-07-19 16:51 UTC (permalink / raw)
To: netdev
Cc: Neil Horman, Vlad Yasevich, Sridhar Samudrala, David S. Miller,
linux-sctp, joe
In-Reply-To: <1342203998-24037-1-git-send-email-nhorman@tuxdriver.com>
I've seen several attempts recently made to do quick failover of sctp transports
by reducing various retransmit timers and counters. While its possible to
implement a faster failover on multihomed sctp associations, its not
particularly robust, in that it can lead to unneeded retransmits, as well as
false connection failures due to intermittent latency on a network.
Instead, lets implement the new ietf quick failover draft found here:
http://tools.ietf.org/html/draft-nishida-tsvwg-sctp-failover-05
This will let the sctp stack identify transports that have had a small number of
errors, and avoid using them quickly until their reliability can be
re-established. I've tested this out on two virt guests connected via multiple
isolated virt networks and believe its in compliance with the above draft and
works well.
Signed-off-by: Neil Horman <nhorman@tuxdriver.com>
CC: Vlad Yasevich <vyasevich@gmail.com>
CC: Sridhar Samudrala <sri@us.ibm.com>
CC: "David S. Miller" <davem@davemloft.net>
CC: linux-sctp@vger.kernel.org
CC: joe@perches.com
---
Change notes:
V2)
- Added socket option API from section 6.1 of the specification, as per
request from Vlad. Adding this socket option allows us to alter both the path
maximum retransmit value and the path partial failure threshold for each
transport and the association as a whole.
- Added a per transport pf_retrans value, and initialized it from the
association value. This makes each transport independently configurable as per
the socket option above, and prevents changes in the sysctl from bleeding into
an already created association.
V3)
- Cleaned up some line spacing (Joe Perches)
- Fixed some socket option user data sanitization (Vlad Yasevich)
---
Documentation/networking/ip-sysctl.txt | 14 +++++
include/net/sctp/constants.h | 1 +
include/net/sctp/structs.h | 11 +++-
include/net/sctp/user.h | 11 ++++
net/sctp/associola.c | 37 ++++++++++--
net/sctp/outqueue.c | 6 +-
net/sctp/sm_sideeffect.c | 33 +++++++++-
net/sctp/socket.c | 100 ++++++++++++++++++++++++++++++++
net/sctp/sysctl.c | 9 +++
net/sctp/transport.c | 4 +-
10 files changed, 211 insertions(+), 15 deletions(-)
diff --git a/Documentation/networking/ip-sysctl.txt b/Documentation/networking/ip-sysctl.txt
index 47b6c79..c636f9c 100644
--- a/Documentation/networking/ip-sysctl.txt
+++ b/Documentation/networking/ip-sysctl.txt
@@ -1408,6 +1408,20 @@ path_max_retrans - INTEGER
Default: 5
+pf_retrans - INTEGER
+ The number of retransmissions that will be attempted on a given path
+ before traffic is redirected to an alternate transport (should one
+ exist). Note this is distinct from path_max_retrans, as a path that
+ passes the pf_retrans threshold can still be used. Its only
+ deprioritized when a transmission path is selected by the stack. This
+ setting is primarily used to enable fast failover mechanisms without
+ having to reduce path_max_retrans to a very low value. See:
+ http://www.ietf.org/id/draft-nishida-tsvwg-sctp-failover-05.txt
+ for details. Note also that a value of pf_retrans > path_max_retrans
+ disables this feature
+
+ Default: 0
+
rto_initial - INTEGER
The initial round trip timeout value in milliseconds that will be used
in calculating round trip times. This is the initial time interval
diff --git a/include/net/sctp/constants.h b/include/net/sctp/constants.h
index 942b864..d053d2e 100644
--- a/include/net/sctp/constants.h
+++ b/include/net/sctp/constants.h
@@ -334,6 +334,7 @@ typedef enum {
typedef enum {
SCTP_TRANSPORT_UP,
SCTP_TRANSPORT_DOWN,
+ SCTP_TRANSPORT_PF,
} sctp_transport_cmd_t;
/* These are the address scopes defined mainly for IPv4 addresses
diff --git a/include/net/sctp/structs.h b/include/net/sctp/structs.h
index e4652fe..f70726c 100644
--- a/include/net/sctp/structs.h
+++ b/include/net/sctp/structs.h
@@ -160,6 +160,7 @@ extern struct sctp_globals {
int max_retrans_association;
int max_retrans_path;
int max_retrans_init;
+ int pf_retrans;
/*
* Policy for preforming sctp/socket accounting
@@ -258,6 +259,7 @@ extern struct sctp_globals {
#define sctp_sndbuf_policy (sctp_globals.sndbuf_policy)
#define sctp_rcvbuf_policy (sctp_globals.rcvbuf_policy)
#define sctp_max_retrans_path (sctp_globals.max_retrans_path)
+#define sctp_pf_retrans (sctp_globals.pf_retrans)
#define sctp_max_retrans_init (sctp_globals.max_retrans_init)
#define sctp_sack_timeout (sctp_globals.sack_timeout)
#define sctp_hb_interval (sctp_globals.hb_interval)
@@ -987,10 +989,15 @@ struct sctp_transport {
/* This is the max_retrans value for the transport and will
* be initialized from the assocs value. This can be changed
- * using SCTP_SET_PEER_ADDR_PARAMS socket option.
+ * using the SCTP_SET_PEER_ADDR_PARAMS socket option.
*/
__u16 pathmaxrxt;
+ /* This is the partially failed retrans value for the transport
+ * and will be initialized from the assocs value. This can be changed
+ * using the SCTP_PEER_ADDR_THLDS socket option
+ */
+ int pf_retrans;
/* PMTU : The current known path MTU. */
__u32 pathmtu;
@@ -1660,6 +1667,8 @@ struct sctp_association {
*/
int max_retrans;
+ int pf_retrans;
+
/* Maximum number of times the endpoint will retransmit INIT */
__u16 max_init_attempts;
diff --git a/include/net/sctp/user.h b/include/net/sctp/user.h
index 0842ef0..1b02d7a 100644
--- a/include/net/sctp/user.h
+++ b/include/net/sctp/user.h
@@ -93,6 +93,7 @@ typedef __s32 sctp_assoc_t;
#define SCTP_GET_ASSOC_NUMBER 28 /* Read only */
#define SCTP_GET_ASSOC_ID_LIST 29 /* Read only */
#define SCTP_AUTO_ASCONF 30
+#define SCTP_PEER_ADDR_THLDS 31
/* Internal Socket Options. Some of the sctp library functions are
* implemented using these socket options.
@@ -649,6 +650,7 @@ struct sctp_paddrinfo {
*/
enum sctp_spinfo_state {
SCTP_INACTIVE,
+ SCTP_PF,
SCTP_ACTIVE,
SCTP_UNCONFIRMED,
SCTP_UNKNOWN = 0xffff /* Value used for transport state unknown */
@@ -741,4 +743,13 @@ typedef struct {
int sd;
} sctp_peeloff_arg_t;
+/*
+ * Peer Address Thresholds socket option
+ */
+struct sctp_paddrthlds {
+ sctp_assoc_t spt_assoc_id;
+ struct sockaddr_storage spt_address;
+ __u16 spt_pathmaxrxt;
+ __u16 spt_pathpfthld;
+};
#endif /* __net_sctp_user_h__ */
diff --git a/net/sctp/associola.c b/net/sctp/associola.c
index 5bc9ab1..90fe36b 100644
--- a/net/sctp/associola.c
+++ b/net/sctp/associola.c
@@ -124,6 +124,8 @@ static struct sctp_association *sctp_association_init(struct sctp_association *a
* socket values.
*/
asoc->max_retrans = sp->assocparams.sasoc_asocmaxrxt;
+ asoc->pf_retrans = sctp_pf_retrans;
+
asoc->rto_initial = msecs_to_jiffies(sp->rtoinfo.srto_initial);
asoc->rto_max = msecs_to_jiffies(sp->rtoinfo.srto_max);
asoc->rto_min = msecs_to_jiffies(sp->rtoinfo.srto_min);
@@ -685,6 +687,9 @@ struct sctp_transport *sctp_assoc_add_peer(struct sctp_association *asoc,
/* Set the path max_retrans. */
peer->pathmaxrxt = asoc->pathmaxrxt;
+ /* And the partial failure retrnas threshold */
+ peer->pf_retrans = asoc->pf_retrans;
+
/* Initialize the peer's SACK delay timeout based on the
* association configured value.
*/
@@ -840,6 +845,7 @@ void sctp_assoc_control_transport(struct sctp_association *asoc,
struct sctp_ulpevent *event;
struct sockaddr_storage addr;
int spc_state = 0;
+ bool ulp_notify = true;
/* Record the transition on the transport. */
switch (command) {
@@ -853,6 +859,14 @@ void sctp_assoc_control_transport(struct sctp_association *asoc,
spc_state = SCTP_ADDR_CONFIRMED;
else
spc_state = SCTP_ADDR_AVAILABLE;
+ /* Don't inform ULP about transition from PF to
+ * active state and set cwnd to 1, see SCTP
+ * Quick failover draft section 5.1, point 5
+ */
+ if (transport->state == SCTP_PF) {
+ ulp_notify = false;
+ transport->cwnd = 1;
+ }
transport->state = SCTP_ACTIVE;
break;
@@ -871,6 +885,11 @@ void sctp_assoc_control_transport(struct sctp_association *asoc,
spc_state = SCTP_ADDR_UNREACHABLE;
break;
+ case SCTP_TRANSPORT_PF:
+ transport->state = SCTP_PF;
+ ulp_notify = false;
+ break;
+
default:
return;
}
@@ -878,12 +897,15 @@ void sctp_assoc_control_transport(struct sctp_association *asoc,
/* Generate and send a SCTP_PEER_ADDR_CHANGE notification to the
* user.
*/
- memset(&addr, 0, sizeof(struct sockaddr_storage));
- memcpy(&addr, &transport->ipaddr, transport->af_specific->sockaddr_len);
- event = sctp_ulpevent_make_peer_addr_change(asoc, &addr,
- 0, spc_state, error, GFP_ATOMIC);
- if (event)
- sctp_ulpq_tail_event(&asoc->ulpq, event);
+ if (ulp_notify) {
+ memset(&addr, 0, sizeof(struct sockaddr_storage));
+ memcpy(&addr, &transport->ipaddr,
+ transport->af_specific->sockaddr_len);
+ event = sctp_ulpevent_make_peer_addr_change(asoc, &addr,
+ 0, spc_state, error, GFP_ATOMIC);
+ if (event)
+ sctp_ulpq_tail_event(&asoc->ulpq, event);
+ }
/* Select new active and retran paths. */
@@ -899,7 +921,8 @@ void sctp_assoc_control_transport(struct sctp_association *asoc,
transports) {
if ((t->state == SCTP_INACTIVE) ||
- (t->state == SCTP_UNCONFIRMED))
+ (t->state == SCTP_UNCONFIRMED) ||
+ (t->state == SCTP_PF))
continue;
if (!first || t->last_time_heard > first->last_time_heard) {
second = first;
diff --git a/net/sctp/outqueue.c b/net/sctp/outqueue.c
index a0fa19f..e7aa177c 100644
--- a/net/sctp/outqueue.c
+++ b/net/sctp/outqueue.c
@@ -792,7 +792,8 @@ static int sctp_outq_flush(struct sctp_outq *q, int rtx_timeout)
if (!new_transport)
new_transport = asoc->peer.active_path;
} else if ((new_transport->state == SCTP_INACTIVE) ||
- (new_transport->state == SCTP_UNCONFIRMED)) {
+ (new_transport->state == SCTP_UNCONFIRMED) ||
+ (new_transport->state == SCTP_PF)) {
/* If the chunk is Heartbeat or Heartbeat Ack,
* send it to chunk->transport, even if it's
* inactive.
@@ -987,7 +988,8 @@ static int sctp_outq_flush(struct sctp_outq *q, int rtx_timeout)
new_transport = chunk->transport;
if (!new_transport ||
((new_transport->state == SCTP_INACTIVE) ||
- (new_transport->state == SCTP_UNCONFIRMED)))
+ (new_transport->state == SCTP_UNCONFIRMED) ||
+ (new_transport->state == SCTP_PF)))
new_transport = asoc->peer.active_path;
if (new_transport->state == SCTP_UNCONFIRMED)
continue;
diff --git a/net/sctp/sm_sideeffect.c b/net/sctp/sm_sideeffect.c
index c96d1a8..285e26a 100644
--- a/net/sctp/sm_sideeffect.c
+++ b/net/sctp/sm_sideeffect.c
@@ -76,6 +76,8 @@ static int sctp_side_effects(sctp_event_t event_type, sctp_subtype_t subtype,
sctp_cmd_seq_t *commands,
gfp_t gfp);
+static void sctp_cmd_hb_timer_update(sctp_cmd_seq_t *cmds,
+ struct sctp_transport *t);
/********************************************************************
* Helper functions
********************************************************************/
@@ -470,7 +472,8 @@ sctp_timer_event_t *sctp_timer_events[SCTP_NUM_TIMEOUT_TYPES] = {
* notification SHOULD be sent to the upper layer.
*
*/
-static void sctp_do_8_2_transport_strike(struct sctp_association *asoc,
+static void sctp_do_8_2_transport_strike(sctp_cmd_seq_t *commands,
+ struct sctp_association *asoc,
struct sctp_transport *transport,
int is_hb)
{
@@ -495,6 +498,23 @@ static void sctp_do_8_2_transport_strike(struct sctp_association *asoc,
transport->error_count++;
}
+ /* If the transport error count is greater than the pf_retrans
+ * threshold, and less than pathmaxrtx, then mark this transport
+ * as Partially Failed, ee SCTP Quick Failover Draft, secon 5.1,
+ * point 1
+ */
+ if ((transport->state != SCTP_PF) &&
+ (asoc->pf_retrans < transport->pathmaxrxt) &&
+ (transport->error_count > asoc->pf_retrans)) {
+
+ sctp_assoc_control_transport(asoc, transport,
+ SCTP_TRANSPORT_PF,
+ 0);
+
+ /* Update the hb timer to resend a heartbeat every rto */
+ sctp_cmd_hb_timer_update(commands, transport);
+ }
+
if (transport->state != SCTP_INACTIVE &&
(transport->error_count > transport->pathmaxrxt)) {
SCTP_DEBUG_PRINTK_IPADDR("transport_strike:association %p",
@@ -699,6 +719,10 @@ static void sctp_cmd_transport_on(sctp_cmd_seq_t *cmds,
SCTP_HEARTBEAT_SUCCESS);
}
+ if (t->state == SCTP_PF)
+ sctp_assoc_control_transport(asoc, t, SCTP_TRANSPORT_UP,
+ SCTP_HEARTBEAT_SUCCESS);
+
/* The receiver of the HEARTBEAT ACK should also perform an
* RTT measurement for that destination transport address
* using the time value carried in the HEARTBEAT ACK chunk.
@@ -1565,8 +1589,8 @@ static int sctp_cmd_interpreter(sctp_event_t event_type,
case SCTP_CMD_STRIKE:
/* Mark one strike against a transport. */
- sctp_do_8_2_transport_strike(asoc, cmd->obj.transport,
- 0);
+ sctp_do_8_2_transport_strike(commands, asoc,
+ cmd->obj.transport, 0);
break;
case SCTP_CMD_TRANSPORT_IDLE:
@@ -1576,7 +1600,8 @@ static int sctp_cmd_interpreter(sctp_event_t event_type,
case SCTP_CMD_TRANSPORT_HB_SENT:
t = cmd->obj.transport;
- sctp_do_8_2_transport_strike(asoc, t, 1);
+ sctp_do_8_2_transport_strike(commands, asoc,
+ t, 1);
t->hb_sent = 1;
break;
diff --git a/net/sctp/socket.c b/net/sctp/socket.c
index b3b8a8d..fef9bfa 100644
--- a/net/sctp/socket.c
+++ b/net/sctp/socket.c
@@ -3470,6 +3470,56 @@ static int sctp_setsockopt_auto_asconf(struct sock *sk, char __user *optval,
}
+/*
+ * SCTP_PEER_ADDR_THLDS
+ *
+ * This option allows us to alter the partially failed threshold for one or all
+ * transports in an association. See Section 6.1 of:
+ * http://www.ietf.org/id/draft-nishida-tsvwg-sctp-failover-05.txt
+ */
+static int sctp_setsockopt_paddr_thresholds(struct sock *sk,
+ char __user *optval,
+ unsigned int optlen)
+{
+ struct sctp_paddrthlds val;
+ struct sctp_transport *trans;
+ struct sctp_association *asoc;
+
+ if (optlen < sizeof(struct sctp_paddrthlds))
+ return -EINVAL;
+ if (copy_from_user(&val, (struct sctp_paddrthlds __user *)optval,
+ sizeof(struct sctp_paddrthlds)))
+ return -EFAULT;
+
+ /* path_max_retrans shouldn't ever be zero */
+ if (!val.spt_pathmaxrxt)
+ return -EINVAL;
+
+ if (sctp_is_any(sk, (const union sctp_addr *)&val.spt_address)) {
+ asoc = sctp_id2assoc(sk, val.spt_assoc_id);
+ if (!asoc)
+ return -ENOENT;
+ list_for_each_entry(trans, &asoc->peer.transport_addr_list,
+ transports) {
+ trans->pathmaxrxt = val.spt_pathmaxrxt;
+ trans->pf_retrans = val.spt_pathpfthld;
+ }
+
+ asoc->pf_retrans = val.spt_pathpfthld;
+ asoc->pathmaxrxt = val.spt_pathmaxrxt;
+ } else {
+ trans = sctp_addr_id2transport(sk, &val.spt_address,
+ val.spt_assoc_id);
+ if (!trans)
+ return -ENOENT;
+
+ trans->pathmaxrxt = val.spt_pathmaxrxt;
+ trans->pf_retrans = val.spt_pathpfthld;
+ }
+
+ return 0;
+}
+
/* API 6.2 setsockopt(), getsockopt()
*
* Applications use setsockopt() and getsockopt() to set or retrieve
@@ -3619,6 +3669,9 @@ SCTP_STATIC int sctp_setsockopt(struct sock *sk, int level, int optname,
case SCTP_AUTO_ASCONF:
retval = sctp_setsockopt_auto_asconf(sk, optval, optlen);
break;
+ case SCTP_PEER_ADDR_THLDS:
+ retval = sctp_setsockopt_paddr_thresholds(sk, optval, optlen);
+ break;
default:
retval = -ENOPROTOOPT;
break;
@@ -5490,6 +5543,50 @@ static int sctp_getsockopt_assoc_ids(struct sock *sk, int len,
return 0;
}
+/*
+ * SCTP_PEER_ADDR_THLDS
+ *
+ * This option allows us to fetch the partially failed threshold for one or all
+ * transports in an association. See Section 6.1 of:
+ * http://www.ietf.org/id/draft-nishida-tsvwg-sctp-failover-05.txt
+ */
+static int sctp_getsockopt_paddr_thresholds(struct sock *sk,
+ char __user *optval,
+ int optlen)
+{
+ struct sctp_paddrthlds val;
+ struct sctp_transport *trans;
+ struct sctp_association *asoc;
+
+ if (optlen < sizeof(struct sctp_paddrthlds))
+ return -EINVAL;
+ optlen = sizeof(struct sctp_paddrthlds);
+ if (copy_from_user(&val, (struct sctp_paddrthlds __user *)optval, optlen))
+ return -EFAULT;
+
+ if (sctp_is_any(sk, (const union sctp_addr *)&val.spt_address)) {
+ asoc = sctp_id2assoc(sk, val.spt_assoc_id);
+ if (!asoc)
+ return -ENOENT;
+
+ val.spt_pathpfthld = asoc->pf_retrans;
+ val.spt_pathmaxrxt = asoc->pathmaxrxt;
+ } else {
+ trans = sctp_addr_id2transport(sk, &val.spt_address,
+ val.spt_assoc_id);
+ if (!trans)
+ return -ENOENT;
+
+ val.spt_pathmaxrxt = trans->pathmaxrxt;
+ val.spt_pathpfthld = trans->pf_retrans;
+ }
+
+ if (copy_to_user(optval, &val, optlen))
+ return -EFAULT;
+
+ return 0;
+}
+
SCTP_STATIC int sctp_getsockopt(struct sock *sk, int level, int optname,
char __user *optval, int __user *optlen)
{
@@ -5628,6 +5725,9 @@ SCTP_STATIC int sctp_getsockopt(struct sock *sk, int level, int optname,
case SCTP_AUTO_ASCONF:
retval = sctp_getsockopt_auto_asconf(sk, len, optval, optlen);
break;
+ case SCTP_PEER_ADDR_THLDS:
+ retval = sctp_getsockopt_paddr_thresholds(sk, optval, len);
+ break;
default:
retval = -ENOPROTOOPT;
break;
diff --git a/net/sctp/sysctl.c b/net/sctp/sysctl.c
index e5fe639..2b2bfe9 100644
--- a/net/sctp/sysctl.c
+++ b/net/sctp/sysctl.c
@@ -141,6 +141,15 @@ static ctl_table sctp_table[] = {
.extra2 = &int_max
},
{
+ .procname = "pf_retrans",
+ .data = &sctp_pf_retrans,
+ .maxlen = sizeof(int),
+ .mode = 0644,
+ .proc_handler = proc_dointvec_minmax,
+ .extra1 = &zero,
+ .extra2 = &int_max
+ },
+ {
.procname = "max_init_retransmits",
.data = &sctp_max_retrans_init,
.maxlen = sizeof(int),
diff --git a/net/sctp/transport.c b/net/sctp/transport.c
index b026ba0..194d0f3 100644
--- a/net/sctp/transport.c
+++ b/net/sctp/transport.c
@@ -85,6 +85,7 @@ static struct sctp_transport *sctp_transport_init(struct sctp_transport *peer,
/* Initialize the default path max_retrans. */
peer->pathmaxrxt = sctp_max_retrans_path;
+ peer->pf_retrans = sctp_pf_retrans;
INIT_LIST_HEAD(&peer->transmitted);
INIT_LIST_HEAD(&peer->send_ready);
@@ -585,7 +586,8 @@ unsigned long sctp_transport_timeout(struct sctp_transport *t)
{
unsigned long timeout;
timeout = t->rto + sctp_jitter(t->rto);
- if (t->state != SCTP_UNCONFIRMED)
+ if ((t->state != SCTP_UNCONFIRMED) &&
+ (t->state != SCTP_PF))
timeout += t->hbinterval;
timeout += jiffies;
return timeout;
--
1.7.7.6
^ permalink raw reply related
* Re: [PATCH] net, cgroup: Fix boot failure due to iteration of uninitialized list
From: Neil Horman @ 2012-07-19 16:44 UTC (permalink / raw)
To: Srivatsa S. Bhat
Cc: gaofeng, eric.dumazet, davem, linux-kernel, netdev, mark.d.rustad,
john.r.fastabend, lizefan
In-Reply-To: <20120719162532.23505.85946.stgit@srivatsabhat.in.ibm.com>
On Thu, Jul 19, 2012 at 09:57:37PM +0530, Srivatsa S. Bhat wrote:
> After commit ef209f15 (net: cgroup: fix access the unallocated memory in
> netprio cgroup), boot fails with the following NULL pointer dereference:
>
> Initializing cgroup subsys devices
> Initializing cgroup subsys freezer
> Initializing cgroup subsys net_cls
> Initializing cgroup subsys blkio
> Initializing cgroup subsys perf_event
> Initializing cgroup subsys net_prio
> BUG: unable to handle kernel NULL pointer dereference at 0000000000000698
> IP: [<ffffffff8145e8d6>] cgrp_create+0xf6/0x190
> PGD 0
> Oops: 0000 [#1] SMP
> CPU 0
> Modules linked in:
>
> Pid: 0, comm: swapper/0 Not tainted 3.5.0-rc7-mandeep #1 IBM IBM System x -[7870C4Q]-/68Y8033
> RIP: 0010:[<ffffffff8145e8d6>] [<ffffffff8145e8d6>] cgrp_create+0xf6/0x190
> RSP: 0000:ffffffff81a01ea8 EFLAGS: 00010213
> RAX: 0000000000000000 RBX: ffffffffffffff10 RCX: 0000000000000000
> RDX: 0000000000000000 RSI: 0000000000000246 RDI: ffffffff81aa70a0
> RBP: ffffffff81a01ed8 R08: 0000000000000000 R09: 0000000000000000
> R10: ffff8808ff8641c0 R11: 6e697a696c616974 R12: 0000000000000001
> R13: ffff8808ff8641c0 R14: 0000000000000000 R15: 0000000000093970
> FS: 0000000000000000(0000) GS:ffff8808ffc00000(0000) knlGS:0000000000000000
> CS: 0010 DS: 0000 ES: 0000 CR0: 000000008005003b
> CR2: 0000000000000698 CR3: 0000000001a0b000 CR4: 00000000000006b0
> DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
> DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400
> Process swapper/0 (pid: 0, threadinfo ffffffff81a00000, task ffffffff81a13420)
> Stack:
> ffffffff81a01eb8 ffffffff818060ff ffffffff81d75ec8 ffffffff81aa8960
> ffffffff81aa8960 ffffffff81b4c2c0 ffffffff81a01ef8 ffffffff81b1cb78
> 0000000000000018 0000000000000048 ffffffff81a01f18 ffffffff81b1ce13
> Call Trace:
> [<ffffffff81b1cb78>] cgroup_init_subsys+0x83/0x169
> [<ffffffff81b1ce13>] cgroup_init+0x36/0x119
> [<ffffffff81affef7>] start_kernel+0x3ba/0x3ef
> [<ffffffff81aff95b>] ? kernel_init+0x27b/0x27b
> [<ffffffff81aff356>] x86_64_start_reservations+0x131/0x136
> [<ffffffff81aff45e>] x86_64_start_kernel+0x103/0x112
> Code: 01 48 3d f8 e1 ec 81 48 8d 98 10 ff ff ff 75 1b eb 73 0f 1f 00 48 8b 83 f0 00 00 00 48 3d f8 e1 ec 81 48 8d 98 10 ff ff ff 74 5a <48> 8b 83 88 07 00 00 48 85 c0 74 de 44 3b 60 10 76 d8 44 89 e6
> RIP [<ffffffff8145e8d6>] cgrp_create+0xf6/0x190
> RSP <ffffffff81a01ea8>
> CR2: 0000000000000698
> ---[ end trace a7919e7f17c0a725 ]---
> Kernel panic - not syncing: Attempted to kill the idle task!
>
> The code corresponds to:
>
> update_netdev_tables():
> for_each_netdev(&init_net, dev) {
> map = rtnl_dereference(dev->priomap); <---- HERE
>
>
> The list head is initialized in netdev_init(), which is called much
> later than cgrp_create(). So the problem is that we are calling
> update_netdev_tables() way too early (in cgrp_create()), which will
> end up traversing the not-yet-circular linked list. So at some point,
> the dev pointer will become NULL and hence dev->priomap becomes an
> invalid access.
>
> To fix this, just remove the update_netdev_tables() function entirely,
> since it appears that write_update_netdev_table() will handle things
> just fine.
>
> Signed-off-by: Srivatsa S. Bhat <srivatsa.bhat@linux.vnet.ibm.com>
> ---
>
> Requesting a thorough review of this patch, since I am not sure whether
> removing update_netdev_tables() is perfectly OK and whether that is the
> right thing to do.
>
We could do this I suppose, but this has already been fixed by
734b65417b24d6eea3e3d7457e1f11493890ee1d
>
^ permalink raw reply
* [PATCH v3 7/7] net-tcp: Fast Open client - cookie-less mode
From: Yuchung Cheng @ 2012-07-19 16:43 UTC (permalink / raw)
To: davem, hkchu, edumazet, ncardwell; +Cc: sivasankar, netdev, Yuchung Cheng
In-Reply-To: <1342716191-19196-1-git-send-email-ycheng@google.com>
In trusted networks, e.g., intranet, data-center, the client does not
need to use Fast Open cookie to mitigate DoS attacks. In cookie-less
mode, sendmsg() with MSG_FASTOPEN flag will send SYN-data regardless
of cookie availability.
Signed-off-by: Yuchung Cheng <ycheng@google.com>
Acked-by: Eric Dumazet <edumazet@google.com>
---
Documentation/networking/ip-sysctl.txt | 2 ++
include/linux/tcp.h | 1 +
include/net/tcp.h | 1 +
net/ipv4/tcp_input.c | 8 ++++++--
net/ipv4/tcp_output.c | 6 +++++-
5 files changed, 15 insertions(+), 3 deletions(-)
diff --git a/Documentation/networking/ip-sysctl.txt b/Documentation/networking/ip-sysctl.txt
index 03964e0..5f3ef7f 100644
--- a/Documentation/networking/ip-sysctl.txt
+++ b/Documentation/networking/ip-sysctl.txt
@@ -476,6 +476,8 @@ tcp_fastopen - INTEGER
The values (bitmap) are:
1: Enables sending data in the opening SYN on the client
+ 5: Enables sending data in the opening SYN on the client regardless
+ of cookie availability.
Default: 0
diff --git a/include/linux/tcp.h b/include/linux/tcp.h
index 1edf96a..9febfb6 100644
--- a/include/linux/tcp.h
+++ b/include/linux/tcp.h
@@ -387,6 +387,7 @@ struct tcp_sock {
u8 repair_queue;
u8 do_early_retrans:1,/* Enable RFC5827 early-retransmit */
early_retrans_delayed:1, /* Delayed ER timer installed */
+ syn_data:1, /* SYN includes data */
syn_fastopen:1; /* SYN includes Fast Open option */
/* RTT measurement */
diff --git a/include/net/tcp.h b/include/net/tcp.h
index e07878d..bc7c134 100644
--- a/include/net/tcp.h
+++ b/include/net/tcp.h
@@ -214,6 +214,7 @@ extern void tcp_time_wait(struct sock *sk, int state, int timeo);
/* Bit Flags for sysctl_tcp_fastopen */
#define TFO_CLIENT_ENABLE 1
+#define TFO_CLIENT_NO_COOKIE 4 /* Data in SYN w/o cookie option */
extern struct inet_timewait_death_row tcp_death_row;
diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c
index c49a4fc..e67d685 100644
--- a/net/ipv4/tcp_input.c
+++ b/net/ipv4/tcp_input.c
@@ -5650,7 +5650,7 @@ static bool tcp_rcv_fastopen_synack(struct sock *sk, struct sk_buff *synack,
struct tcp_fastopen_cookie *cookie)
{
struct tcp_sock *tp = tcp_sk(sk);
- struct sk_buff *data = tcp_write_queue_head(sk);
+ struct sk_buff *data = tp->syn_data ? tcp_write_queue_head(sk) : NULL;
u16 mss = tp->rx_opt.mss_clamp;
bool syn_drop;
@@ -5665,6 +5665,9 @@ static bool tcp_rcv_fastopen_synack(struct sock *sk, struct sk_buff *synack,
mss = opt.mss_clamp;
}
+ if (!tp->syn_fastopen) /* Ignore an unsolicited cookie */
+ cookie->len = -1;
+
/* The SYN-ACK neither has cookie nor acknowledges the data. Presumably
* the remote receives only the retransmitted (regular) SYNs: either
* the original SYN-data or the corresponding SYN-ACK is lost.
@@ -5816,7 +5819,8 @@ static int tcp_rcv_synsent_state_process(struct sock *sk, struct sk_buff *skb,
tcp_finish_connect(sk, skb);
- if (tp->syn_fastopen && tcp_rcv_fastopen_synack(sk, skb, &foc))
+ if ((tp->syn_fastopen || tp->syn_data) &&
+ tcp_rcv_fastopen_synack(sk, skb, &foc))
return -1;
if (sk->sk_write_pending ||
diff --git a/net/ipv4/tcp_output.c b/net/ipv4/tcp_output.c
index c5cfd5e..27a32ac 100644
--- a/net/ipv4/tcp_output.c
+++ b/net/ipv4/tcp_output.c
@@ -2864,6 +2864,7 @@ static int tcp_send_syn_data(struct sock *sk, struct sk_buff *syn)
struct sk_buff *syn_data = NULL, *data;
unsigned long last_syn_loss = 0;
+ tp->rx_opt.mss_clamp = tp->advmss; /* If MSS is not cached */
tcp_fastopen_cache_get(sk, &tp->rx_opt.mss_clamp, &fo->cookie,
&syn_loss, &last_syn_loss);
/* Recurring FO SYN losses: revert to regular handshake temporarily */
@@ -2873,7 +2874,9 @@ static int tcp_send_syn_data(struct sock *sk, struct sk_buff *syn)
goto fallback;
}
- if (fo->cookie.len <= 0)
+ if (sysctl_tcp_fastopen & TFO_CLIENT_NO_COOKIE)
+ fo->cookie.len = -1;
+ else if (fo->cookie.len <= 0)
goto fallback;
/* MSS for SYN-data is based on cached MSS and bounded by PMTU and
@@ -2916,6 +2919,7 @@ static int tcp_send_syn_data(struct sock *sk, struct sk_buff *syn)
fo->copied = data->len;
if (tcp_transmit_skb(sk, syn_data, 0, sk->sk_allocation) == 0) {
+ tp->syn_data = (fo->copied > 0);
NET_INC_STATS(sock_net(sk), LINUX_MIB_TCPFASTOPENACTIVE);
goto done;
}
--
1.7.7.3
^ permalink raw reply related
* [PATCH v3 0/7] TCP Fast Open client
From: Yuchung Cheng @ 2012-07-19 16:43 UTC (permalink / raw)
To: davem, hkchu, edumazet, ncardwell; +Cc: sivasankar, netdev, Yuchung Cheng
ChangeLog since v2:
- Added seqlock to update Fast Open metrics
- Move TCP magic code in inet_wait_for_connect() to inet_stream_connect()
- Move up MSG_FASTOPEN macro for better header formatting
ChangeLog since v1:
- Reduce tons of code by storing Fast Open stats in the TCP metrics :)
- Clarify the purpose of using an experimental option in patch 1/7
This patch series implement the client functionality of TCP Fast Open.
TCP Fast Open (TFO) allows data to be carried in the SYN and SYN-ACK
packets and consumed by the receiving end during the initial connection
handshake, thus providing a saving of up to one full round trip time (RTT)
compared to standard TCP requiring a three-way handshake (3WHS) to
complete before data can be exchanged.
The protocol change is detailed in the IETF internet draft at
http://www.ietf.org/id/draft-ietf-tcpm-fastopen-00.txt . The research
paper (http://conferences.sigcomm.org/co-next/2011/papers/1569470463.pdf)
studied the performance impact of HTTP using Fast Open, based on this
Linux implementation and the Chrome browser.
To use Fast Open, the client application (active SYN sender) must
replace connect() socket call with sendmsg() or sendto() with the new
MSG_FASTOPEN flag. If the server supports Fast Open the data exchange
starts at TCP handshake. Otherwise the connection will automatically
fall back to conventional TCP.
Yuchung Cheng (7):
net-tcp: Fast Open base
net-tcp: Fast Open client - cookie cache
net-tcp: Fast Open client - sending SYN-data
net-tcp: Fast Open client - receiving SYN-ACK
net-tcp: Fast Open client - sendmsg(MSG_FASTOPEN)
net-tcp: Fast Open client - detecting SYN-data drops
net-tcp: Fast Open client - cookie-less mode
Documentation/networking/ip-sysctl.txt | 13 +++
include/linux/snmp.h | 1 +
include/linux/socket.h | 1 +
include/linux/tcp.h | 17 ++++-
include/net/inet_common.h | 6 +-
include/net/tcp.h | 28 ++++++-
net/ipv4/Makefile | 2 +-
net/ipv4/af_inet.c | 29 +++++--
net/ipv4/proc.c | 1 +
net/ipv4/syncookies.c | 2 +-
net/ipv4/sysctl_net_ipv4.c | 7 ++
net/ipv4/tcp.c | 61 ++++++++++++-
net/ipv4/tcp_fastopen.c | 11 +++
net/ipv4/tcp_input.c | 76 ++++++++++++++--
net/ipv4/tcp_ipv4.c | 5 +-
net/ipv4/tcp_metrics.c | 61 +++++++++++++
net/ipv4/tcp_minisocks.c | 4 +-
net/ipv4/tcp_output.c | 153 +++++++++++++++++++++++++++++---
net/ipv6/syncookies.c | 2 +-
net/ipv6/tcp_ipv6.c | 2 +-
20 files changed, 438 insertions(+), 44 deletions(-)
create mode 100644 net/ipv4/tcp_fastopen.c
--
1.7.7.3
^ permalink raw reply
* [PATCH v3 6/7] net-tcp: Fast Open client - detecting SYN-data drops
From: Yuchung Cheng @ 2012-07-19 16:43 UTC (permalink / raw)
To: davem, hkchu, edumazet, ncardwell; +Cc: sivasankar, netdev, Yuchung Cheng
In-Reply-To: <1342716191-19196-1-git-send-email-ycheng@google.com>
On paths with firewalls dropping SYN with data or experimental TCP options,
Fast Open connections will have experience SYN timeout and bad performance.
The solution is to track such incidents in the cookie cache and disables
Fast Open temporarily.
Since only the original SYN includes data and/or Fast Open option, the
SYN-ACK has some tell-tale sign (tcp_rcv_fastopen_synack()) to detect
such drops. If a path has recurring Fast Open SYN drops, Fast Open is
disabled for 2^(recurring_losses) minutes starting from four minutes up to
roughly one and half day. sendmsg with MSG_FASTOPEN flag will succeed but
it behaves as connect() then write().
Signed-off-by: Yuchung Cheng <ycheng@google.com>
---
include/net/tcp.h | 6 ++++--
net/ipv4/tcp_input.c | 10 +++++++++-
net/ipv4/tcp_metrics.c | 16 +++++++++++++---
net/ipv4/tcp_output.c | 13 +++++++++++--
4 files changed, 37 insertions(+), 8 deletions(-)
diff --git a/include/net/tcp.h b/include/net/tcp.h
index c025810..e07878d 100644
--- a/include/net/tcp.h
+++ b/include/net/tcp.h
@@ -409,9 +409,11 @@ extern bool tcp_peer_is_proven(struct request_sock *req, struct dst_entry *dst,
extern bool tcp_remember_stamp(struct sock *sk);
extern bool tcp_tw_remember_stamp(struct inet_timewait_sock *tw);
extern void tcp_fastopen_cache_get(struct sock *sk, u16 *mss,
- struct tcp_fastopen_cookie *cookie);
+ struct tcp_fastopen_cookie *cookie,
+ int *syn_loss, unsigned long *last_syn_loss);
extern void tcp_fastopen_cache_set(struct sock *sk, u16 mss,
- struct tcp_fastopen_cookie *cookie);
+ struct tcp_fastopen_cookie *cookie,
+ bool syn_lost);
extern void tcp_fetch_timewait_stamp(struct sock *sk, struct dst_entry *dst);
extern void tcp_disable_fack(struct tcp_sock *tp);
extern void tcp_close(struct sock *sk, long timeout);
diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c
index 38b6a81..c49a4fc 100644
--- a/net/ipv4/tcp_input.c
+++ b/net/ipv4/tcp_input.c
@@ -5652,6 +5652,7 @@ static bool tcp_rcv_fastopen_synack(struct sock *sk, struct sk_buff *synack,
struct tcp_sock *tp = tcp_sk(sk);
struct sk_buff *data = tcp_write_queue_head(sk);
u16 mss = tp->rx_opt.mss_clamp;
+ bool syn_drop;
if (mss == tp->rx_opt.user_mss) {
struct tcp_options_received opt;
@@ -5664,7 +5665,14 @@ static bool tcp_rcv_fastopen_synack(struct sock *sk, struct sk_buff *synack,
mss = opt.mss_clamp;
}
- tcp_fastopen_cache_set(sk, mss, cookie);
+ /* The SYN-ACK neither has cookie nor acknowledges the data. Presumably
+ * the remote receives only the retransmitted (regular) SYNs: either
+ * the original SYN-data or the corresponding SYN-ACK is lost.
+ */
+ syn_drop = (cookie->len <= 0 && data &&
+ inet_csk(sk)->icsk_retransmits);
+
+ tcp_fastopen_cache_set(sk, mss, cookie, syn_drop);
if (data) { /* Retransmit unacked data in SYN */
tcp_retransmit_skb(sk, data);
diff --git a/net/ipv4/tcp_metrics.c b/net/ipv4/tcp_metrics.c
index d02ff37..99779ae 100644
--- a/net/ipv4/tcp_metrics.c
+++ b/net/ipv4/tcp_metrics.c
@@ -32,6 +32,8 @@ enum tcp_metric_index {
struct tcp_fastopen_metrics {
u16 mss;
+ u16 syn_loss:10; /* Recurring Fast Open SYN losses */
+ unsigned long last_syn_loss; /* Last Fast Open SYN loss */
struct tcp_fastopen_cookie cookie;
};
@@ -125,6 +127,7 @@ static void tcpm_suck_dst(struct tcp_metrics_block *tm, struct dst_entry *dst)
tm->tcpm_ts = 0;
tm->tcpm_ts_stamp = 0;
tm->tcpm_fastopen.mss = 0;
+ tm->tcpm_fastopen.syn_loss = 0;
tm->tcpm_fastopen.cookie.len = 0;
}
@@ -644,7 +647,8 @@ bool tcp_tw_remember_stamp(struct inet_timewait_sock *tw)
static DEFINE_SEQLOCK(fastopen_seqlock);
void tcp_fastopen_cache_get(struct sock *sk, u16 *mss,
- struct tcp_fastopen_cookie *cookie)
+ struct tcp_fastopen_cookie *cookie,
+ int *syn_loss, unsigned long *last_syn_loss)
{
struct tcp_metrics_block *tm;
@@ -659,14 +663,15 @@ void tcp_fastopen_cache_get(struct sock *sk, u16 *mss,
if (tfom->mss)
*mss = tfom->mss;
*cookie = tfom->cookie;
+ *syn_loss = tfom->syn_loss;
+ *last_syn_loss = *syn_loss ? tfom->last_syn_loss : 0;
} while (read_seqretry(&fastopen_seqlock, seq));
}
rcu_read_unlock();
}
-
void tcp_fastopen_cache_set(struct sock *sk, u16 mss,
- struct tcp_fastopen_cookie *cookie)
+ struct tcp_fastopen_cookie *cookie, bool syn_lost)
{
struct tcp_metrics_block *tm;
@@ -679,6 +684,11 @@ void tcp_fastopen_cache_set(struct sock *sk, u16 mss,
tfom->mss = mss;
if (cookie->len > 0)
tfom->cookie = *cookie;
+ if (syn_lost) {
+ ++tfom->syn_loss;
+ tfom->last_syn_loss = jiffies;
+ } else
+ tfom->syn_loss = 0;
write_sequnlock_bh(&fastopen_seqlock);
}
rcu_read_unlock();
diff --git a/net/ipv4/tcp_output.c b/net/ipv4/tcp_output.c
index 8869328..c5cfd5e 100644
--- a/net/ipv4/tcp_output.c
+++ b/net/ipv4/tcp_output.c
@@ -2860,10 +2860,19 @@ static int tcp_send_syn_data(struct sock *sk, struct sk_buff *syn)
{
struct tcp_sock *tp = tcp_sk(sk);
struct tcp_fastopen_request *fo = tp->fastopen_req;
- int space, i, err = 0, iovlen = fo->data->msg_iovlen;
+ int syn_loss = 0, space, i, err = 0, iovlen = fo->data->msg_iovlen;
struct sk_buff *syn_data = NULL, *data;
+ unsigned long last_syn_loss = 0;
+
+ tcp_fastopen_cache_get(sk, &tp->rx_opt.mss_clamp, &fo->cookie,
+ &syn_loss, &last_syn_loss);
+ /* Recurring FO SYN losses: revert to regular handshake temporarily */
+ if (syn_loss > 1 &&
+ time_before(jiffies, last_syn_loss + (60*HZ << syn_loss))) {
+ fo->cookie.len = -1;
+ goto fallback;
+ }
- tcp_fastopen_cache_get(sk, &tp->rx_opt.mss_clamp, &fo->cookie);
if (fo->cookie.len <= 0)
goto fallback;
--
1.7.7.3
^ permalink raw reply related
page: next (older) | prev (newer) | latest
- recent:[subjects (threaded)|topics (new)|topics (active)]
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox