* Re: build failure of latest ethtool
From: Ben Hutchings @ 2012-12-06 16:29 UTC (permalink / raw)
To: Or Gerlitz; +Cc: netdev
In-Reply-To: <50C0C276.5060809@mellanox.com>
On Thu, 2012-12-06 at 18:06 +0200, Or Gerlitz wrote:
> Hi Ben,
>
> FYI, building the latest ethtool git fails since sfpid.c is not present
> in the Makefile.
[...]
If you want to build from git you have to run:
./autogen.sh && ./configure && make
Ben.
--
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: build failure of latest ethtool
From: Or Gerlitz @ 2012-12-06 16:33 UTC (permalink / raw)
To: Ben Hutchings; +Cc: netdev
In-Reply-To: <1354811376.2828.1.camel@bwh-desktop.uk.solarflarecom.com>
On 06/12/2012 18:29, Ben Hutchings wrote:
> If you want to build from git you have to run: ./autogen.sh &&
> ./configure && make
OK, sorry for the spam, I pulled the tree but didn't re-run ./autogen.sh
&& ./configure
^ permalink raw reply
* Re: [RFC PATCH v2 3/3] tun: fix LSM/SELinux labeling of tun/tap devices
From: Paul Moore @ 2012-12-06 16:56 UTC (permalink / raw)
To: Michael S. Tsirkin; +Cc: netdev, linux-security-module, selinux, jasowang
In-Reply-To: <20121206161200.GA4340@redhat.com>
On Thursday, December 06, 2012 06:12:00 PM Michael S. Tsirkin wrote:
> On Thu, Dec 06, 2012 at 10:46:11AM -0500, Paul Moore wrote:
> > On Thursday, December 06, 2012 12:33:25 PM Michael S. Tsirkin wrote:
> > > OK so just to verify: this can be used to ensure that qemu
> > > process that has the queue fd can only attach it to
> > > a specific device, right?
> >
> > Whenever a new queue is created via TUNSETQUEUE/tun_set_queue() the
> > security_tun_dev_create_queue() LSM hook is called. When SELinux is
> > enabled this hook ends up calling selinux_tun_dev_create_queue() which
> > checks that the calling process (process_t) is allowed to create a new
> > queue on the specified device (tundev_t) . If you are familiar with
> > SELinux security policy, the allow rule would look like this:
> >
> > allow process_t tundev_t:tun_socket create_queue;
> >
> > In practice, if we assume libvirt is creating the TUN device and running
> > with a SELinux label of virtd_t and that QEMU instances are running with
> > a SELinux label of svirt_t then the allow rule would look like this:
> >
> > allow svirt_t virtd_t:tun_socket create_queue;
> >
> > There is also the matter of the MLS/MCS constraints providing additional
> > separation but that is another level of detail which I don't believe is
> > important for our discussion.
>
> Hmm. How do the rules for SETIFF look ATM?
> I am just checking default policy does not let qemu do with
> SETQUEUE something with a device which it can not
> attach to using SETIFF.
The SETQUEUE/tun_socket:create_queue permissions do not yet exist in any
released SELinux policy as we are just now adding them with this patchset.
With current policies loaded into a kernel with this patchset applied the
SETQUEUE/tun_socket:create_queue permission would be treated according to the
policy's unknown permission setting.
--
paul moore
security and virtualization @ redhat
^ permalink raw reply
* [PATCH] sctp: proc: protect bind_addr->address_list accesses with rcu_read_lock()
From: Thomas Graf @ 2012-12-06 17:34 UTC (permalink / raw)
To: linux-sctp; +Cc: netdev, Vlad Yasevich, Neil Horman
address_list is protected via the socket lock or RCU. Since we don't want
to take the socket lock for each assoc we dump in procfs a RCU read-side
critical section must be entered.
Cc: Vlad Yasevich <vyasevich@gmail.com>
Cc: Neil Horman <nhorman@tuxdriver.com>
Signed-off-by: Thomas Graf <tgraf@suug.ch>
---
net/sctp/proc.c | 4 +++-
1 file changed, 3 insertions(+), 1 deletion(-)
diff --git a/net/sctp/proc.c b/net/sctp/proc.c
index 9966e7b..ec9b0c8 100644
--- a/net/sctp/proc.c
+++ b/net/sctp/proc.c
@@ -139,7 +139,8 @@ static void sctp_seq_dump_local_addrs(struct seq_file *seq, struct sctp_ep_commo
primary = &peer->saddr;
}
- list_for_each_entry(laddr, &epb->bind_addr.address_list, list) {
+ rcu_read_lock();
+ list_for_each_entry_rcu(laddr, &epb->bind_addr.address_list, list) {
addr = &laddr->a;
af = sctp_get_af_specific(addr->sa.sa_family);
if (primary && af->cmp_addr(addr, primary)) {
@@ -147,6 +148,7 @@ static void sctp_seq_dump_local_addrs(struct seq_file *seq, struct sctp_ep_commo
}
af->seq_dump_addr(seq, addr);
}
+ rcu_read_unlock();
}
/* Dump remote addresses of an association. */
--
1.7.11.7
^ permalink raw reply related
* Hi!
From: Marketing Commucation @ 2012-12-06 17:18 UTC (permalink / raw)
I am Mr. Joseph. I have a lucrative business proposal of mutual interest to
share with you.
^ permalink raw reply
* Re: [PATCH net-next 0/7] Allow to monitor multicast cache event via rtnetlink
From: Thomas Graf @ 2012-12-06 17:49 UTC (permalink / raw)
To: Nicolas Dichtel; +Cc: David Miller, David.Laight, netdev
In-Reply-To: <50C05AC2.1050504@6wind.com>
On 12/06/12 at 09:43am, Nicolas Dichtel wrote:
> Le 05/12/2012 18:54, David Miller a écrit :
> >From: "David Laight" <David.Laight@ACULAB.COM>
> >Date: Wed, 5 Dec 2012 11:41:33 -0000
> >
> >>Probably worth commenting that the 64bit items might only be 32bit aligned.
> >>Just to stop anyone trying to read/write them with pointer casts.
> >
> >Rather, let's not create this situation at all.
> >
> >It's totally inappropriate to have special code to handle every single
> >time we want to put 64-bit values into netlink messages.
> >
> >We need a real solution to this issue.
> >
> The easiest way is to update *_ALIGNTO values (maybe we can keep
> NLMSG_ALIGNTO to 4). But I think that many userland apps have these
> values hardcoded and, the most important thing, this may increase
> size of many netlink messages. Hence we need probably to find
> something better.
We can't do this, as you say, ALIGNTO is compiled into all the
binaries.
A simple backwards compatible workaround would be to include an
unknown, empty padding attribute if needed. That would be 4 bytes
in size and could be used to include padding as needed.
We could use nla_type = 0 as it is a reserved value that should
be available in all protocols. All readers (kernel and user space)
must ignore such an attribute just like any other unknown
attribute they encounter.
We could easily extend nla_put_u64() and variants to automatically
include such a padding attribute as needed.
The only situation that I can think of where this would not work
is if we have code like this:
foo = nla_nest_start();
for ([..])
nla_put_u64([...])
nla_nest_end([...])
and a reader would stupidly do a nla_for_each_attr() in user space
and assume all attributes found must be NLA_U64 without even
checking the length of the attribute.
I would say we take that risk and let such code die horribly.
^ permalink raw reply
* [PATCH] sctp: Add RCU protection to assoc->transport_addr_list
From: Thomas Graf @ 2012-12-06 18:15 UTC (permalink / raw)
To: linux-sctp; +Cc: netdev, Vlad Yasevich, Neil Horman
peer.transport_addr_list is currently only protected by sk_sock
which is inpractical to acquire for procfs dumping purposes.
This patch adds RCU protection allowing for the procfs readers to
enter RCU read-side critical sections.
Modification of the list continues to be serialized via sk_lock.
Cc: Vlad Yasevich <vyasevich@gmail.com>
Cc: Neil Horman <nhorman@tuxdriver.com>
Signed-off-by: Thomas Graf <tgraf@suug.ch>
---
include/net/sctp/structs.h | 2 ++
net/sctp/associola.c | 4 ++--
net/sctp/proc.c | 8 ++++++--
net/sctp/transport.c | 18 +++++++++++++-----
4 files changed, 23 insertions(+), 9 deletions(-)
diff --git a/include/net/sctp/structs.h b/include/net/sctp/structs.h
index 64158aa..5d6987b 100644
--- a/include/net/sctp/structs.h
+++ b/include/net/sctp/structs.h
@@ -948,6 +948,8 @@ struct sctp_transport {
/* 64-bit random number sent with heartbeat. */
__u64 hb_nonce;
+
+ struct rcu_head rcu;
};
struct sctp_transport *sctp_transport_new(struct net *, const union sctp_addr *,
diff --git a/net/sctp/associola.c b/net/sctp/associola.c
index b1ef3bc..c826bb8 100644
--- a/net/sctp/associola.c
+++ b/net/sctp/associola.c
@@ -565,7 +565,7 @@ void sctp_assoc_rm_peer(struct sctp_association *asoc,
sctp_assoc_update_retran_path(asoc);
/* Remove this peer from the list. */
- list_del(&peer->transports);
+ list_del_rcu(&peer->transports);
/* Get the first transport of asoc. */
pos = asoc->peer.transport_addr_list.next;
@@ -765,7 +765,7 @@ struct sctp_transport *sctp_assoc_add_peer(struct sctp_association *asoc,
peer->state = peer_state;
/* Attach the remote transport to our asoc. */
- list_add_tail(&peer->transports, &asoc->peer.transport_addr_list);
+ list_add_tail_rcu(&peer->transports, &asoc->peer.transport_addr_list);
asoc->peer.transport_count++;
/* If we do not yet have a primary path, set one. */
diff --git a/net/sctp/proc.c b/net/sctp/proc.c
index ec9b0c8..36a3f9d 100644
--- a/net/sctp/proc.c
+++ b/net/sctp/proc.c
@@ -159,7 +159,8 @@ static void sctp_seq_dump_remote_addrs(struct seq_file *seq, struct sctp_associa
struct sctp_af *af;
primary = &assoc->peer.primary_addr;
- list_for_each_entry(transport, &assoc->peer.transport_addr_list,
+ rcu_read_lock();
+ list_for_each_entry_rcu(transport, &assoc->peer.transport_addr_list,
transports) {
addr = &transport->ipaddr;
af = sctp_get_af_specific(addr->sa.sa_family);
@@ -168,6 +169,7 @@ static void sctp_seq_dump_remote_addrs(struct seq_file *seq, struct sctp_associa
}
af->seq_dump_addr(seq, addr);
}
+ rcu_read_unlock();
}
static void * sctp_eps_seq_start(struct seq_file *seq, loff_t *pos)
@@ -438,11 +440,12 @@ static int sctp_remaddr_seq_show(struct seq_file *seq, void *v)
head = &sctp_assoc_hashtable[hash];
sctp_local_bh_disable();
read_lock(&head->lock);
+ rcu_read_lock();
sctp_for_each_hentry(epb, node, &head->chain) {
if (!net_eq(sock_net(epb->sk), seq_file_net(seq)))
continue;
assoc = sctp_assoc(epb);
- list_for_each_entry(tsp, &assoc->peer.transport_addr_list,
+ list_for_each_entry_rcu(tsp, &assoc->peer.transport_addr_list,
transports) {
/*
* The remote address (ADDR)
@@ -489,6 +492,7 @@ static int sctp_remaddr_seq_show(struct seq_file *seq, void *v)
}
}
+ rcu_read_unlock();
read_unlock(&head->lock);
sctp_local_bh_enable();
diff --git a/net/sctp/transport.c b/net/sctp/transport.c
index 206cf52..1295aec 100644
--- a/net/sctp/transport.c
+++ b/net/sctp/transport.c
@@ -163,13 +163,11 @@ void sctp_transport_free(struct sctp_transport *transport)
sctp_transport_put(transport);
}
-/* Destroy the transport data structure.
- * Assumes there are no more users of this structure.
- */
-static void sctp_transport_destroy(struct sctp_transport *transport)
+static void sctp_transport_destroy_rcu(struct rcu_head *head)
{
- SCTP_ASSERT(transport->dead, "Transport is not dead", return);
+ struct sctp_transport *transport;
+ transport = container_of(head, struct sctp_transport, rcu);
if (transport->asoc)
sctp_association_put(transport->asoc);
@@ -180,6 +178,16 @@ static void sctp_transport_destroy(struct sctp_transport *transport)
SCTP_DBG_OBJCNT_DEC(transport);
}
+/* Destroy the transport data structure.
+ * Assumes there are no more users of this structure.
+ */
+static void sctp_transport_destroy(struct sctp_transport *transport)
+{
+ SCTP_ASSERT(transport->dead, "Transport is not dead", return);
+
+ call_rcu(&transport->rcu, sctp_transport_destroy_rcu);
+}
+
/* Start T3_rtx timer if it is not already running and update the heartbeat
* timer. This routine is called every time a DATA chunk is sent.
*/
--
1.7.11.7
^ permalink raw reply related
* Re: [PATCH] sctp: proc: protect bind_addr->address_list accesses with rcu_read_lock()
From: Vlad Yasevich @ 2012-12-06 18:25 UTC (permalink / raw)
To: Thomas Graf; +Cc: linux-sctp, netdev, Neil Horman
In-Reply-To: <a803093c70e857e8e01e554288ee79f66f7acda4.1354814846.git.tgraf@suug.ch>
On 12/06/2012 12:34 PM, Thomas Graf wrote:
> address_list is protected via the socket lock or RCU. Since we don't want
> to take the socket lock for each assoc we dump in procfs a RCU read-side
> critical section must be entered.
>
> Cc: Vlad Yasevich <vyasevich@gmail.com>
> Cc: Neil Horman <nhorman@tuxdriver.com>
> Signed-off-by: Thomas Graf <tgraf@suug.ch>
> ---
> net/sctp/proc.c | 4 +++-
> 1 file changed, 3 insertions(+), 1 deletion(-)
>
> diff --git a/net/sctp/proc.c b/net/sctp/proc.c
> index 9966e7b..ec9b0c8 100644
> --- a/net/sctp/proc.c
> +++ b/net/sctp/proc.c
> @@ -139,7 +139,8 @@ static void sctp_seq_dump_local_addrs(struct seq_file *seq, struct sctp_ep_commo
> primary = &peer->saddr;
> }
>
> - list_for_each_entry(laddr, &epb->bind_addr.address_list, list) {
> + rcu_read_lock();
> + list_for_each_entry_rcu(laddr, &epb->bind_addr.address_list, list) {
> addr = &laddr->a;
> af = sctp_get_af_specific(addr->sa.sa_family);
> if (primary && af->cmp_addr(addr, primary)) {
> @@ -147,6 +148,7 @@ static void sctp_seq_dump_local_addrs(struct seq_file *seq, struct sctp_ep_commo
> }
> af->seq_dump_addr(seq, addr);
> }
> + rcu_read_unlock();
> }
>
> /* Dump remote addresses of an association. */
>
May want to avoid printing addresses that are !addr->valid.
Otherwise looks good.
-vlad
^ permalink raw reply
* Re: [PATCH] sctp: proc: protect bind_addr->address_list accesses with rcu_read_lock()
From: Thomas Graf @ 2012-12-06 18:28 UTC (permalink / raw)
To: Vlad Yasevich; +Cc: linux-sctp, netdev, Neil Horman
In-Reply-To: <50C0E300.3000703@gmail.com>
On 12/06/12 at 01:25pm, Vlad Yasevich wrote:
> May want to avoid printing addresses that are !addr->valid.
>
> Otherwise looks good.
Good point, I'll send a follow-up patch if that is OK. Same
applies to dead transports obviously.
^ permalink raw reply
* Re: [PATCH] sctp: Add RCU protection to assoc->transport_addr_list
From: Vlad Yasevich @ 2012-12-06 18:35 UTC (permalink / raw)
To: Thomas Graf; +Cc: linux-sctp, netdev, Neil Horman
In-Reply-To: <16453bea94a6fc43d657139dff2ce0b5924e2a1f.1354817574.git.tgraf@suug.ch>
On 12/06/2012 01:15 PM, Thomas Graf wrote:
> peer.transport_addr_list is currently only protected by sk_sock
> which is inpractical to acquire for procfs dumping purposes.
>
> This patch adds RCU protection allowing for the procfs readers to
> enter RCU read-side critical sections.
>
> Modification of the list continues to be serialized via sk_lock.
>
> Cc: Vlad Yasevich <vyasevich@gmail.com>
> Cc: Neil Horman <nhorman@tuxdriver.com>
> Signed-off-by: Thomas Graf <tgraf@suug.ch>
> ---
> include/net/sctp/structs.h | 2 ++
> net/sctp/associola.c | 4 ++--
> net/sctp/proc.c | 8 ++++++--
> net/sctp/transport.c | 18 +++++++++++++-----
> 4 files changed, 23 insertions(+), 9 deletions(-)
>
> diff --git a/include/net/sctp/structs.h b/include/net/sctp/structs.h
> index 64158aa..5d6987b 100644
> --- a/include/net/sctp/structs.h
> +++ b/include/net/sctp/structs.h
> @@ -948,6 +948,8 @@ struct sctp_transport {
>
> /* 64-bit random number sent with heartbeat. */
> __u64 hb_nonce;
> +
> + struct rcu_head rcu;
> };
>
> struct sctp_transport *sctp_transport_new(struct net *, const union sctp_addr *,
> diff --git a/net/sctp/associola.c b/net/sctp/associola.c
> index b1ef3bc..c826bb8 100644
> --- a/net/sctp/associola.c
> +++ b/net/sctp/associola.c
> @@ -565,7 +565,7 @@ void sctp_assoc_rm_peer(struct sctp_association *asoc,
> sctp_assoc_update_retran_path(asoc);
>
> /* Remove this peer from the list. */
> - list_del(&peer->transports);
> + list_del_rcu(&peer->transports);
>
> /* Get the first transport of asoc. */
> pos = asoc->peer.transport_addr_list.next;
> @@ -765,7 +765,7 @@ struct sctp_transport *sctp_assoc_add_peer(struct sctp_association *asoc,
> peer->state = peer_state;
>
> /* Attach the remote transport to our asoc. */
> - list_add_tail(&peer->transports, &asoc->peer.transport_addr_list);
> + list_add_tail_rcu(&peer->transports, &asoc->peer.transport_addr_list);
> asoc->peer.transport_count++;
>
> /* If we do not yet have a primary path, set one. */
> diff --git a/net/sctp/proc.c b/net/sctp/proc.c
> index ec9b0c8..36a3f9d 100644
> --- a/net/sctp/proc.c
> +++ b/net/sctp/proc.c
> @@ -159,7 +159,8 @@ static void sctp_seq_dump_remote_addrs(struct seq_file *seq, struct sctp_associa
> struct sctp_af *af;
>
> primary = &assoc->peer.primary_addr;
> - list_for_each_entry(transport, &assoc->peer.transport_addr_list,
> + rcu_read_lock();
> + list_for_each_entry_rcu(transport, &assoc->peer.transport_addr_list,
> transports) {
> addr = &transport->ipaddr;
> af = sctp_get_af_specific(addr->sa.sa_family);
> @@ -168,6 +169,7 @@ static void sctp_seq_dump_remote_addrs(struct seq_file *seq, struct sctp_associa
> }
> af->seq_dump_addr(seq, addr);
> }
> + rcu_read_unlock();
> }
>
> static void * sctp_eps_seq_start(struct seq_file *seq, loff_t *pos)
> @@ -438,11 +440,12 @@ static int sctp_remaddr_seq_show(struct seq_file *seq, void *v)
> head = &sctp_assoc_hashtable[hash];
> sctp_local_bh_disable();
> read_lock(&head->lock);
> + rcu_read_lock();
> sctp_for_each_hentry(epb, node, &head->chain) {
> if (!net_eq(sock_net(epb->sk), seq_file_net(seq)))
> continue;
> assoc = sctp_assoc(epb);
> - list_for_each_entry(tsp, &assoc->peer.transport_addr_list,
> + list_for_each_entry_rcu(tsp, &assoc->peer.transport_addr_list,
> transports) {
> /*
> * The remote address (ADDR)
> @@ -489,6 +492,7 @@ static int sctp_remaddr_seq_show(struct seq_file *seq, void *v)
> }
> }
>
> + rcu_read_unlock();
> read_unlock(&head->lock);
> sctp_local_bh_enable();
>
> diff --git a/net/sctp/transport.c b/net/sctp/transport.c
> index 206cf52..1295aec 100644
> --- a/net/sctp/transport.c
> +++ b/net/sctp/transport.c
> @@ -163,13 +163,11 @@ void sctp_transport_free(struct sctp_transport *transport)
> sctp_transport_put(transport);
> }
>
> -/* Destroy the transport data structure.
> - * Assumes there are no more users of this structure.
> - */
> -static void sctp_transport_destroy(struct sctp_transport *transport)
> +static void sctp_transport_destroy_rcu(struct rcu_head *head)
> {
> - SCTP_ASSERT(transport->dead, "Transport is not dead", return);
> + struct sctp_transport *transport;
>
> + transport = container_of(head, struct sctp_transport, rcu);
> if (transport->asoc)
> sctp_association_put(transport->asoc);
>
> @@ -180,6 +178,16 @@ static void sctp_transport_destroy(struct sctp_transport *transport)
> SCTP_DBG_OBJCNT_DEC(transport);
> }
>
> +/* Destroy the transport data structure.
> + * Assumes there are no more users of this structure.
> + */
> +static void sctp_transport_destroy(struct sctp_transport *transport)
> +{
> + SCTP_ASSERT(transport->dead, "Transport is not dead", return);
> +
> + call_rcu(&transport->rcu, sctp_transport_destroy_rcu);
> +}
> +
> /* Start T3_rtx timer if it is not already running and update the heartbeat
> * timer. This routine is called every time a DATA chunk is sent.
> */
>
We may want to mark transports as dead sooner. Probably right about the
time we pull them off the list. When displaying, we may want to
look at transport->dead, and skip them. It will reduce the probability
that we would be looking at a transport that's about to go away.
Otherwise, looks very nice.
-vlad
^ permalink raw reply
* Re: [PATCH] sctp: Add RCU protection to assoc->transport_addr_list
From: Thomas Graf @ 2012-12-06 18:44 UTC (permalink / raw)
To: Vlad Yasevich; +Cc: linux-sctp, netdev, Neil Horman
In-Reply-To: <50C0E585.1080701@gmail.com>
On 12/06/12 at 01:35pm, Vlad Yasevich wrote:
> We may want to mark transports as dead sooner. Probably right about
> the time we pull them off the list.
We mark it dead in sctp_transport_free() which is called at the
end of sctp_assoc_rm_peer(). Do you want to mark it dead at the
beginning of sctp_assoc_rm_peer() as well? (We still need to
mark in sctp_transport_free() anyway).
> When displaying, we may want to
> look at transport->dead, and skip them. It will reduce the probability
> that we would be looking at a transport that's about to go away.
Agreed.
^ permalink raw reply
* [PATCH] tcp: bug fix Fast Open client retransmission
From: Yuchung Cheng @ 2012-12-06 18:45 UTC (permalink / raw)
To: davem, ncardwell, nanditad, edumazet; +Cc: netdev, Yuchung Cheng
If SYN-ACK partially acks SYN-data, the client retransmits the
remaining data by tcp_retransmit_skb(). This increments lost recovery
state variables like tp->retrans_out in Open state. If loss recovery
happens before the retransmission is acked, it triggers the WARN_ON
check in tcp_fastretrans_alert(). For example: the client sends
SYN-data, gets SYN-ACK acking only ISN, retransmits data, sends
another 4 data packets and get 3 dupacks.
Since the retransmission is not caused by network drop it should not
update the recovery state variables. Further the server may return a
smaller MSS than the cached MSS used for SYN-data, so the retranmission
needs a loop. Otherwise some data will not be retransmitted until timeout
or other loss recovery events.
Signed-off-by: Yuchung Cheng <ycheng@google.com>
---
include/net/tcp.h | 1 +
net/ipv4/tcp_input.c | 6 +++++-
net/ipv4/tcp_output.c | 15 ++++++++++-----
3 files changed, 16 insertions(+), 6 deletions(-)
diff --git a/include/net/tcp.h b/include/net/tcp.h
index 3202bde..aed42c7 100644
--- a/include/net/tcp.h
+++ b/include/net/tcp.h
@@ -524,6 +524,7 @@ static inline __u32 cookie_v6_init_sequence(struct sock *sk,
extern void __tcp_push_pending_frames(struct sock *sk, unsigned int cur_mss,
int nonagle);
extern bool tcp_may_send_now(struct sock *sk);
+extern int __tcp_retransmit_skb(struct sock *, struct sk_buff *);
extern int tcp_retransmit_skb(struct sock *, struct sk_buff *);
extern void tcp_retransmit_timer(struct sock *sk);
extern void tcp_xmit_retransmit_queue(struct sock *);
diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c
index fc67831..a136925 100644
--- a/net/ipv4/tcp_input.c
+++ b/net/ipv4/tcp_input.c
@@ -5652,7 +5652,11 @@ static bool tcp_rcv_fastopen_synack(struct sock *sk, struct sk_buff *synack,
tcp_fastopen_cache_set(sk, mss, cookie, syn_drop);
if (data) { /* Retransmit unacked data in SYN */
- tcp_retransmit_skb(sk, data);
+ tcp_for_write_queue_from(data, sk) {
+ if (data == tcp_send_head(sk) ||
+ __tcp_retransmit_skb(sk, data))
+ break;
+ }
tcp_rearm_rto(sk);
return true;
}
diff --git a/net/ipv4/tcp_output.c b/net/ipv4/tcp_output.c
index 8ac0855..5d45159 100644
--- a/net/ipv4/tcp_output.c
+++ b/net/ipv4/tcp_output.c
@@ -2309,12 +2309,11 @@ static void tcp_retrans_try_collapse(struct sock *sk, struct sk_buff *to,
* state updates are done by the caller. Returns non-zero if an
* error occurred which prevented the send.
*/
-int tcp_retransmit_skb(struct sock *sk, struct sk_buff *skb)
+int __tcp_retransmit_skb(struct sock *sk, struct sk_buff *skb)
{
struct tcp_sock *tp = tcp_sk(sk);
struct inet_connection_sock *icsk = inet_csk(sk);
unsigned int cur_mss;
- int err;
/* Inconslusive MTU probe */
if (icsk->icsk_mtup.probe_size) {
@@ -2387,11 +2386,17 @@ int tcp_retransmit_skb(struct sock *sk, struct sk_buff *skb)
if (unlikely(NET_IP_ALIGN && ((unsigned long)skb->data & 3))) {
struct sk_buff *nskb = __pskb_copy(skb, MAX_TCP_HEADER,
GFP_ATOMIC);
- err = nskb ? tcp_transmit_skb(sk, nskb, 0, GFP_ATOMIC) :
- -ENOBUFS;
+ return nskb ? tcp_transmit_skb(sk, nskb, 0, GFP_ATOMIC) :
+ -ENOBUFS;
} else {
- err = tcp_transmit_skb(sk, skb, 1, GFP_ATOMIC);
+ return tcp_transmit_skb(sk, skb, 1, GFP_ATOMIC);
}
+}
+
+int tcp_retransmit_skb(struct sock *sk, struct sk_buff *skb)
+{
+ struct tcp_sock *tp = tcp_sk(sk);
+ int err = __tcp_retransmit_skb(sk, skb);
if (err == 0) {
/* Update global TCP statistics. */
--
1.7.7.3
^ permalink raw reply related
* Re: [PATCH net next] team: remove team parameter from port_enter callback.
From: Jiri Pirko @ 2012-12-06 18:50 UTC (permalink / raw)
To: Rami Rosen; +Cc: davem, netdev
In-Reply-To: <1354799382-23808-1-git-send-email-ramirose@gmail.com>
Thu, Dec 06, 2012 at 02:09:42PM CET, ramirose@gmail.com wrote:
>This patch removes an unused parameter (team) from port_enter callback in
>team_mode_ops and fixes accordingly its invocations in 3 modes and in team.c.
Please do not do this. This parameter is ment to be used in future.
Thanks.
Jiri
>
>Signed-off-by: Rami Rosen <ramirose@gmail.com>
>---
> drivers/net/team/team.c | 2 +-
> drivers/net/team/team_mode_broadcast.c | 2 +-
> drivers/net/team/team_mode_loadbalance.c | 2 +-
> drivers/net/team/team_mode_roundrobin.c | 2 +-
> include/linux/if_team.h | 2 +-
> 5 files changed, 5 insertions(+), 5 deletions(-)
>
>diff --git a/drivers/net/team/team.c b/drivers/net/team/team.c
>index ad86660..4203808 100644
>--- a/drivers/net/team/team.c
>+++ b/drivers/net/team/team.c
>@@ -887,7 +887,7 @@ static int team_port_enter(struct team *team, struct team_port *port)
> dev_hold(team->dev);
> port->dev->priv_flags |= IFF_TEAM_PORT;
> if (team->ops.port_enter) {
>- err = team->ops.port_enter(team, port);
>+ err = team->ops.port_enter(port);
> if (err) {
> netdev_err(team->dev, "Device %s failed to enter team mode\n",
> port->dev->name);
>diff --git a/drivers/net/team/team_mode_broadcast.c b/drivers/net/team/team_mode_broadcast.c
>index c5db428..c3840bc 100644
>--- a/drivers/net/team/team_mode_broadcast.c
>+++ b/drivers/net/team/team_mode_broadcast.c
>@@ -46,7 +46,7 @@ static bool bc_transmit(struct team *team, struct sk_buff *skb)
> return sum_ret;
> }
>
>-static int bc_port_enter(struct team *team, struct team_port *port)
>+static int bc_port_enter(struct team_port *port)
> {
> return team_port_set_team_dev_addr(port);
> }
>diff --git a/drivers/net/team/team_mode_loadbalance.c b/drivers/net/team/team_mode_loadbalance.c
>index cdc31b5..1e84c77 100644
>--- a/drivers/net/team/team_mode_loadbalance.c
>+++ b/drivers/net/team/team_mode_loadbalance.c
>@@ -614,7 +614,7 @@ static void lb_exit(struct team *team)
> kfree(lb_priv->ex);
> }
>
>-static int lb_port_enter(struct team *team, struct team_port *port)
>+static int lb_port_enter(struct team_port *port)
> {
> struct lb_port_priv *lb_port_priv = get_lb_port_priv(port);
>
>diff --git a/drivers/net/team/team_mode_roundrobin.c b/drivers/net/team/team_mode_roundrobin.c
>index 105135a..abe889e 100644
>--- a/drivers/net/team/team_mode_roundrobin.c
>+++ b/drivers/net/team/team_mode_roundrobin.c
>@@ -64,7 +64,7 @@ drop:
> return false;
> }
>
>-static int rr_port_enter(struct team *team, struct team_port *port)
>+static int rr_port_enter(struct team_port *port)
> {
> return team_port_set_team_dev_addr(port);
> }
>diff --git a/include/linux/if_team.h b/include/linux/if_team.h
>index 0245def..3366453 100644
>--- a/include/linux/if_team.h
>+++ b/include/linux/if_team.h
>@@ -105,7 +105,7 @@ struct team_mode_ops {
> struct team_port *port,
> struct sk_buff *skb);
> bool (*transmit)(struct team *team, struct sk_buff *skb);
>- int (*port_enter)(struct team *team, struct team_port *port);
>+ int (*port_enter)(struct team_port *port);
> void (*port_leave)(struct team *team, struct team_port *port);
> void (*port_change_dev_addr)(struct team *team, struct team_port *port);
> void (*port_enabled)(struct team *team, struct team_port *port);
>--
>1.7.11.7
>
^ permalink raw reply
* Re: [PATCH] net : enable tx time stamping in the vde driver.
From: Richard Cochran @ 2012-12-06 18:57 UTC (permalink / raw)
To: Paul Chavent; +Cc: jdike, richard, user-mode-linux-devel, netdev
In-Reply-To: <1354807505-21222-1-git-send-email-paul.chavent@onera.fr>
On Thu, Dec 06, 2012 at 04:25:05PM +0100, Paul Chavent wrote:
> This new version moves the skb_tx_timestamp in the main uml
> driver. This should avoid the need to call this function in each
> transport (vde, slirp, tuntap, ...). It also add support for ethtool
> get_ts_info.
>
> Signed-off-by: Paul Chavent <paul.chavent@onera.fr>
Acked-by: Richard Cochran <richardcochran@gmail.com>
^ permalink raw reply
* Re: [PATCH] sctp: Add RCU protection to assoc->transport_addr_list
From: Vlad Yasevich @ 2012-12-06 18:57 UTC (permalink / raw)
To: Thomas Graf; +Cc: linux-sctp, netdev, Neil Horman
In-Reply-To: <20121206184433.GE16122@casper.infradead.org>
On 12/06/2012 01:44 PM, Thomas Graf wrote:
> On 12/06/12 at 01:35pm, Vlad Yasevich wrote:
>> We may want to mark transports as dead sooner. Probably right about
>> the time we pull them off the list.
>
> We mark it dead in sctp_transport_free() which is called at the
> end of sctp_assoc_rm_peer(). Do you want to mark it dead at the
> beginning of sctp_assoc_rm_peer() as well? (We still need to
> mark in sctp_transport_free() anyway).
Crud.. sctp_transport_free() is called directly in places... Hmm...
the one in sctp_association_free() may need to be list_del_rcu()...
Ok, we can leave the dead handling the way it is..
-vlad
>
>> When displaying, we may want to
>> look at transport->dead, and skip them. It will reduce the probability
>> that we would be looking at a transport that's about to go away.
>
> Agreed.
>
^ permalink raw reply
* Re: [PATCH] sctp: Add RCU protection to assoc->transport_addr_list
From: Thomas Graf @ 2012-12-06 19:08 UTC (permalink / raw)
To: Vlad Yasevich; +Cc: linux-sctp, netdev, Neil Horman
In-Reply-To: <50C0EAB5.3050303@gmail.com>
On 12/06/12 at 01:57pm, Vlad Yasevich wrote:
> On 12/06/2012 01:44 PM, Thomas Graf wrote:
> >On 12/06/12 at 01:35pm, Vlad Yasevich wrote:
> >>We may want to mark transports as dead sooner. Probably right about
> >>the time we pull them off the list.
> >
> >We mark it dead in sctp_transport_free() which is called at the
> >end of sctp_assoc_rm_peer(). Do you want to mark it dead at the
> >beginning of sctp_assoc_rm_peer() as well? (We still need to
> >mark in sctp_transport_free() anyway).
>
> Crud.. sctp_transport_free() is called directly in places... Hmm...
> the one in sctp_association_free() may need to be list_del_rcu()...
It's not really needed but it wouldn't be wrong from a
documentation perspective. The assoc is always unhashed
while holding head->lock before sctp_association_free()
and all current RCU readers of transport_addr_list access
the the assoc while holding a read-lock on head->lock.
Let me respin this patch and do a list_del_rcu() there
to document the RCU'iness of it.
^ permalink raw reply
* Re: [PATCH] sctp: Add RCU protection to assoc->transport_addr_list
From: Vlad Yasevich @ 2012-12-06 19:14 UTC (permalink / raw)
To: Thomas Graf; +Cc: linux-sctp, netdev, Neil Horman
In-Reply-To: <20121206190835.GF16122@casper.infradead.org>
On 12/06/2012 02:08 PM, Thomas Graf wrote:
> On 12/06/12 at 01:57pm, Vlad Yasevich wrote:
>> On 12/06/2012 01:44 PM, Thomas Graf wrote:
>>> On 12/06/12 at 01:35pm, Vlad Yasevich wrote:
>>>> We may want to mark transports as dead sooner. Probably right about
>>>> the time we pull them off the list.
>>>
>>> We mark it dead in sctp_transport_free() which is called at the
>>> end of sctp_assoc_rm_peer(). Do you want to mark it dead at the
>>> beginning of sctp_assoc_rm_peer() as well? (We still need to
>>> mark in sctp_transport_free() anyway).
>>
>> Crud.. sctp_transport_free() is called directly in places... Hmm...
>> the one in sctp_association_free() may need to be list_del_rcu()...
>
> It's not really needed but it wouldn't be wrong from a
> documentation perspective. The assoc is always unhashed
> while holding head->lock before sctp_association_free()
> and all current RCU readers of transport_addr_list access
> the the assoc while holding a read-lock on head->lock.
>
> Let me respin this patch and do a list_del_rcu() there
> to document the RCU'iness of it.
>
Right, but there may be chunks that have cached association with a ref
before sctp_association_free() is called. Now, after free they may
be looking at the transport list for whatever reason... Most places
check assoc->dead, but I don't want to get caught. So, there is
a remote chance that someone may look at transports and would crash
without rcu.
-vlad
^ permalink raw reply
* Re: [PATCH] sctp: proc: protect bind_addr->address_list accesses with rcu_read_lock()
From: Neil Horman @ 2012-12-06 19:16 UTC (permalink / raw)
To: Thomas Graf; +Cc: Vlad Yasevich, linux-sctp, netdev
In-Reply-To: <20121206182840.GD16122@casper.infradead.org>
On Thu, Dec 06, 2012 at 06:28:40PM +0000, Thomas Graf wrote:
> On 12/06/12 at 01:25pm, Vlad Yasevich wrote:
> > May want to avoid printing addresses that are !addr->valid.
> >
> > Otherwise looks good.
>
> Good point, I'll send a follow-up patch if that is OK. Same
> applies to dead transports obviously.
Assuming a followup patch to avoid dead addresses..
Acked-by: Neil Horman <nhorman@tuxdriver.com>
> --
> To unsubscribe from this list: send the line "unsubscribe linux-sctp" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at http://vger.kernel.org/majordomo-info.html
>
^ permalink raw reply
* [PATCH 0/2 v2] sctp: RCU protection when accessing assoc members for procfs
From: Thomas Graf @ 2012-12-06 19:25 UTC (permalink / raw)
To: linux-sctp; +Cc: netdev, vyasevich, nhorman
@Dave: This is a respin of the following patches:
[PATCH] sctp: proc: protect bind_addr->address_list accesses with rcu_read_lock()
[PATCH] sctp: Add RCU protection to assoc->transport_addr_list
Thomas Graf (2):
sctp: proc: protect bind_addr->address_list accesses with
rcu_read_lock()
sctp: Add RCU protection to assoc->transport_addr_list
include/net/sctp/structs.h | 2 ++
net/sctp/associola.c | 6 +++---
net/sctp/proc.c | 21 ++++++++++++++++++---
net/sctp/transport.c | 18 +++++++++++++-----
4 files changed, 36 insertions(+), 11 deletions(-)
--
1.7.11.7
^ permalink raw reply
* Re: [PATCH] net : enable tx time stamping in the vde driver.
From: Richard Weinberger @ 2012-12-06 19:25 UTC (permalink / raw)
To: Richard Cochran, Paul Chavent; +Cc: jdike, user-mode-linux-devel, netdev
In-Reply-To: <20121206185742.GA2175@netboy.at.omicron.at>
Richard Cochran <richardcochran@gmail.com> schrieb:
>On Thu, Dec 06, 2012 at 04:25:05PM +0100, Paul Chavent wrote:
>> This new version moves the skb_tx_timestamp in the main uml
>> driver. This should avoid the need to call this function in each
>> transport (vde, slirp, tuntap, ...). It also add support for ethtool
>> get_ts_info.
>>
>> Signed-off-by: Paul Chavent <paul.chavent@onera.fr>
>
>Acked-by: Richard Cochran <richardcochran@gmail.com>
Queued for 3.8.
Thanks,
//richard
^ permalink raw reply
* [PATCH 1/2] sctp: proc: protect bind_addr->address_list accesses with rcu_read_lock()
From: Thomas Graf @ 2012-12-06 19:25 UTC (permalink / raw)
To: linux-sctp; +Cc: netdev, vyasevich, nhorman, Thomas Graf
In-Reply-To: <cover.1354821623.git.tgraf@suug.ch>
From: Thomas Graf <tgraf@redhat.com>
address_list is protected via the socket lock or RCU. Since we don't want
to take the socket lock for each assoc we dump in procfs a RCU read-side
critical section must be entered.
V2: Skip local addresses marked as dead
Cc: Vlad Yasevich <vyasevich@gmail.com>
Cc: Neil Horman <nhorman@tuxdriver.com>
Signed-off-by: Thomas Graf <tgraf@suug.ch>
---
net/sctp/proc.c | 7 ++++++-
1 file changed, 6 insertions(+), 1 deletion(-)
diff --git a/net/sctp/proc.c b/net/sctp/proc.c
index 9966e7b..06b05ee 100644
--- a/net/sctp/proc.c
+++ b/net/sctp/proc.c
@@ -139,7 +139,11 @@ static void sctp_seq_dump_local_addrs(struct seq_file *seq, struct sctp_ep_commo
primary = &peer->saddr;
}
- list_for_each_entry(laddr, &epb->bind_addr.address_list, list) {
+ rcu_read_lock();
+ list_for_each_entry_rcu(laddr, &epb->bind_addr.address_list, list) {
+ if (!laddr->valid)
+ continue;
+
addr = &laddr->a;
af = sctp_get_af_specific(addr->sa.sa_family);
if (primary && af->cmp_addr(addr, primary)) {
@@ -147,6 +151,7 @@ static void sctp_seq_dump_local_addrs(struct seq_file *seq, struct sctp_ep_commo
}
af->seq_dump_addr(seq, addr);
}
+ rcu_read_unlock();
}
/* Dump remote addresses of an association. */
--
1.7.11.7
^ permalink raw reply related
* [PATCH 2/2] sctp: Add RCU protection to assoc->transport_addr_list
From: Thomas Graf @ 2012-12-06 19:25 UTC (permalink / raw)
To: linux-sctp; +Cc: netdev, vyasevich, nhorman
In-Reply-To: <cover.1354821623.git.tgraf@suug.ch>
peer.transport_addr_list is currently only protected by sk_sock
which is inpractical to acquire for procfs dumping purposes.
This patch adds RCU protection allowing for the procfs readers to
enter RCU read-side critical sections.
Modification of the list continues to be serialized via sk_lock.
V2: Use list_del_rcu() in sctp_association_free() to be safe
Skip transports marked dead when dumping for procfs
Cc: Vlad Yasevich <vyasevich@gmail.com>
Cc: Neil Horman <nhorman@tuxdriver.com>
Signed-off-by: Thomas Graf <tgraf@suug.ch>
---
include/net/sctp/structs.h | 2 ++
net/sctp/associola.c | 6 +++---
net/sctp/proc.c | 14 ++++++++++++--
net/sctp/transport.c | 18 +++++++++++++-----
4 files changed, 30 insertions(+), 10 deletions(-)
diff --git a/include/net/sctp/structs.h b/include/net/sctp/structs.h
index 64158aa..5d6987b 100644
--- a/include/net/sctp/structs.h
+++ b/include/net/sctp/structs.h
@@ -948,6 +948,8 @@ struct sctp_transport {
/* 64-bit random number sent with heartbeat. */
__u64 hb_nonce;
+
+ struct rcu_head rcu;
};
struct sctp_transport *sctp_transport_new(struct net *, const union sctp_addr *,
diff --git a/net/sctp/associola.c b/net/sctp/associola.c
index b1ef3bc..81b1236 100644
--- a/net/sctp/associola.c
+++ b/net/sctp/associola.c
@@ -445,7 +445,7 @@ void sctp_association_free(struct sctp_association *asoc)
/* Release the transport structures. */
list_for_each_safe(pos, temp, &asoc->peer.transport_addr_list) {
transport = list_entry(pos, struct sctp_transport, transports);
- list_del(pos);
+ list_del_rcu(pos);
sctp_transport_free(transport);
}
@@ -565,7 +565,7 @@ void sctp_assoc_rm_peer(struct sctp_association *asoc,
sctp_assoc_update_retran_path(asoc);
/* Remove this peer from the list. */
- list_del(&peer->transports);
+ list_del_rcu(&peer->transports);
/* Get the first transport of asoc. */
pos = asoc->peer.transport_addr_list.next;
@@ -765,7 +765,7 @@ struct sctp_transport *sctp_assoc_add_peer(struct sctp_association *asoc,
peer->state = peer_state;
/* Attach the remote transport to our asoc. */
- list_add_tail(&peer->transports, &asoc->peer.transport_addr_list);
+ list_add_tail_rcu(&peer->transports, &asoc->peer.transport_addr_list);
asoc->peer.transport_count++;
/* If we do not yet have a primary path, set one. */
diff --git a/net/sctp/proc.c b/net/sctp/proc.c
index 06b05ee..8c19e97 100644
--- a/net/sctp/proc.c
+++ b/net/sctp/proc.c
@@ -162,15 +162,20 @@ static void sctp_seq_dump_remote_addrs(struct seq_file *seq, struct sctp_associa
struct sctp_af *af;
primary = &assoc->peer.primary_addr;
- list_for_each_entry(transport, &assoc->peer.transport_addr_list,
+ rcu_read_lock();
+ list_for_each_entry_rcu(transport, &assoc->peer.transport_addr_list,
transports) {
addr = &transport->ipaddr;
+ if (transport->dead)
+ continue;
+
af = sctp_get_af_specific(addr->sa.sa_family);
if (af->cmp_addr(addr, primary)) {
seq_printf(seq, "*");
}
af->seq_dump_addr(seq, addr);
}
+ rcu_read_unlock();
}
static void * sctp_eps_seq_start(struct seq_file *seq, loff_t *pos)
@@ -441,12 +446,16 @@ static int sctp_remaddr_seq_show(struct seq_file *seq, void *v)
head = &sctp_assoc_hashtable[hash];
sctp_local_bh_disable();
read_lock(&head->lock);
+ rcu_read_lock();
sctp_for_each_hentry(epb, node, &head->chain) {
if (!net_eq(sock_net(epb->sk), seq_file_net(seq)))
continue;
assoc = sctp_assoc(epb);
- list_for_each_entry(tsp, &assoc->peer.transport_addr_list,
+ list_for_each_entry_rcu(tsp, &assoc->peer.transport_addr_list,
transports) {
+ if (tsp->dead)
+ continue;
+
/*
* The remote address (ADDR)
*/
@@ -492,6 +501,7 @@ static int sctp_remaddr_seq_show(struct seq_file *seq, void *v)
}
}
+ rcu_read_unlock();
read_unlock(&head->lock);
sctp_local_bh_enable();
diff --git a/net/sctp/transport.c b/net/sctp/transport.c
index 206cf52..1295aec 100644
--- a/net/sctp/transport.c
+++ b/net/sctp/transport.c
@@ -163,13 +163,11 @@ void sctp_transport_free(struct sctp_transport *transport)
sctp_transport_put(transport);
}
-/* Destroy the transport data structure.
- * Assumes there are no more users of this structure.
- */
-static void sctp_transport_destroy(struct sctp_transport *transport)
+static void sctp_transport_destroy_rcu(struct rcu_head *head)
{
- SCTP_ASSERT(transport->dead, "Transport is not dead", return);
+ struct sctp_transport *transport;
+ transport = container_of(head, struct sctp_transport, rcu);
if (transport->asoc)
sctp_association_put(transport->asoc);
@@ -180,6 +178,16 @@ static void sctp_transport_destroy(struct sctp_transport *transport)
SCTP_DBG_OBJCNT_DEC(transport);
}
+/* Destroy the transport data structure.
+ * Assumes there are no more users of this structure.
+ */
+static void sctp_transport_destroy(struct sctp_transport *transport)
+{
+ SCTP_ASSERT(transport->dead, "Transport is not dead", return);
+
+ call_rcu(&transport->rcu, sctp_transport_destroy_rcu);
+}
+
/* Start T3_rtx timer if it is not already running and update the heartbeat
* timer. This routine is called every time a DATA chunk is sent.
*/
--
1.7.11.7
^ permalink raw reply related
* Re: [PATCH] sctp: Add RCU protection to assoc->transport_addr_list
From: Thomas Graf @ 2012-12-06 19:28 UTC (permalink / raw)
To: Vlad Yasevich; +Cc: linux-sctp, netdev, Neil Horman
In-Reply-To: <50C0EE8E.3080505@gmail.com>
On 12/06/12 at 02:14pm, Vlad Yasevich wrote:
> Right, but there may be chunks that have cached association with a ref
> before sctp_association_free() is called. Now, after free they may
> be looking at the transport list for whatever reason... Most places
> check assoc->dead, but I don't want to get caught. So, there is
> a remote chance that someone may look at transports and would crash
> without rcu.
You are right, sctp_associate_free() can be called even though there
may still be multiple refs on that assoc around. We are currently
fine as I believe sk_lock serializes all the accesses but better
be safe than sorry.
I've respun the patches in a v2 patch series.
^ permalink raw reply
* Re: Random MAC address from smsc75xx: How to permanently set?
From: Bjørn Mork @ 2012-12-06 19:44 UTC (permalink / raw)
To: Steve Glendinning
Cc: Dan Williams, Cunningham, Robert, netdev-u79uwXL29TY76Z2rM5mHXA,
linux-usb-u79uwXL29TY76Z2rM5mHXA
In-Reply-To: <1354819495.8021.32.camel-wKZy7rqYPVb5EHUCmHmTqw@public.gmane.org>
[-- Attachment #1: Type: text/plain, Size: 3017 bytes --]
Dan Williams <dcbw-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org> writes:
> On Thu, 2012-12-06 at 18:35 +0000, Cunningham, Robert wrote:
>
>> I'm trying to bring up an OMAP4 system based on Variscite's OM44
>> module running Linaro's Ubuntu Precise in a headless configuration.
>> The system contains two SMSC LAN7500 USB-GigE chips (not dongles),
>> both of which are fully functional.
>>
>> The GigE chips don't have EEPROMS, so no permanent MAC addresses can
>> be assigned in hardware. As expected, the smsc75xx driver assigns a
>> random MAC address when the interface is discovered and initialized.
>> However, we need to provide consistent MAC addresses on these
>> interfaces. (Yes, we could respin the board to add EEPROMS, but
>> that's a last, and expensive, resort.)
>>
>> After the system boots, I'd like to change the MAC addresses to
>> specific values. While there are multiple ways to do this (using
>> commands such as ifconfig, ip, macchanger, and others), it seems the
>> updated MAC address is always overridden when I do "ifconfig ethX
>> up", which causes yet another different random MAC address to be
>> created and assigned. Simply repeating ifconfig up/down causes an
>> endless list of random MAC addresses to be generated.
>>
>> I created a udev rule that I hoped would handle the situation, but it is also overridden:
>> /etc/udev/rules.d/99-mac-address.rules:
>> SUBSYSTEM=="net", KERNEL=="eth0", RUN+="/sbin/ip link set dev %k address XX:XX:XX:XX:XX:00"
>> SUBSYSTEM=="net", KERNEL=="eth1", RUN+="/sbin/ip link set dev %k address XX:XX:XX:XX:XX:01"
>>
>> For a single interface, I can use u-boot variables or kernel boot
>> arguments, but they seem to work only for the first interface, and I
>> have two.
>
> Just matching on interface name won't guarantee that you get the same
> MAC assigned to the same physical interface each time you boot. You
> can't rely on bus probing order. What you really want to do is enhance
> the udev rule to match the network interfaces based on stuff like PCI
> bus address, SDIO details, or whatever bus type the interfaces are
> hanging off.
Creating a rule matching on the USB device should do, provided the
driver is fixed to prevent changing the address on every open.
>> I've searched extensively for other solutions, but have found none that will survive "ifconfig ethX up".
>>
>> Is there a way to "permanently" tell the smsc75xx device driver to
>> use my desired MAC addresses instead of generating random ones every
>> time? And to have the MAC addresses remain unchanged when I bring
>> the interfaces up and down?
Well, that is clearly a driver bug. It should reuse the same address.
What do you think about something like the completely untested attached
patch, Steve?
You may want to go over the other parts of .reset as well. It does a
lot, and I suspect most of it should be a onetime initialization thing
and not done on every device open.
Bjørn
[-- Attachment #2: 0001-smsc75xx-dont-reset-macaddress-in-open.patch --]
[-- Type: text/x-diff, Size: 1301 bytes --]
>From a51089b8ad13957ba72be2f411c06684225ab3e5 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Bj=C3=B8rn=20Mork?= <bjorn-yOkvZcmFvRU@public.gmane.org>
Date: Thu, 6 Dec 2012 20:33:25 +0100
Subject: [PATCH] smsc75xx: dont reset macaddress in open
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit
usbnet_open calls the minidriver .reset function. Moving the
macadress initialization to .bind to prevent the macaddress
from changing every time the device is brought up in case a
persistent macaddress is missing.
Signed-off-by: Bjørn Mork <bjorn-yOkvZcmFvRU@public.gmane.org>
---
drivers/net/usb/smsc75xx.c | 19 ++++++++-----------
1 file changed, 8 insertions(+), 11 deletions(-)
diff --git a/drivers/net/usb/smsc75xx.c b/drivers/net/usb/smsc75xx.c
index 1cbd936..971c679 100644
--- a/drivers/net/usb/smsc75xx.c
+++ b/drivers/net/usb/smsc75xx.c
@@ -1054,17 +1054,6 @@ static int smsc75xx_reset(struct usbnet *dev)
netif_dbg(dev, ifup, dev->net, "PHY reset complete\n");
- smsc75xx_init_mac_address(dev);
-
- ret = smsc75xx_set_mac_address(dev);
- if (ret < 0) {
- netdev_warn(dev->net, "Failed to set mac address\n");
- return ret;
- }
-
- netif_dbg(dev, ifup, dev->net, "MAC Address: %pM\n",
- dev->net->dev_addr);
^ permalink raw reply related
* Re: [PATCH] tcp: bug fix Fast Open client retransmission
From: Neal Cardwell @ 2012-12-06 19:50 UTC (permalink / raw)
To: Yuchung Cheng; +Cc: David Miller, Nandita Dukkipati, Eric Dumazet, Netdev
In-Reply-To: <1354819532-15676-1-git-send-email-ycheng@google.com>
On Thu, Dec 6, 2012 at 1:45 PM, Yuchung Cheng <ycheng@google.com> wrote:
> If SYN-ACK partially acks SYN-data, the client retransmits the
> remaining data by tcp_retransmit_skb(). This increments lost recovery
> state variables like tp->retrans_out in Open state. If loss recovery
> happens before the retransmission is acked, it triggers the WARN_ON
> check in tcp_fastretrans_alert(). For example: the client sends
> SYN-data, gets SYN-ACK acking only ISN, retransmits data, sends
> another 4 data packets and get 3 dupacks.
>
> Since the retransmission is not caused by network drop it should not
> update the recovery state variables. Further the server may return a
> smaller MSS than the cached MSS used for SYN-data, so the retranmission
> needs a loop. Otherwise some data will not be retransmitted until timeout
> or other loss recovery events.
>
> Signed-off-by: Yuchung Cheng <ycheng@google.com>
> ---
Acked-by: Neal Cardwell <ncardwell@google.com>
^ permalink raw reply
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