* Re: [PATCH 2.6.36] vlan: Avoid hwaccel vlan packets when vid not used
From: Jesse Gross @ 2011-01-02 0:27 UTC (permalink / raw)
To: Eric Dumazet
Cc: Matt Carlson, Michael Leun, Michael Chan, David Miller,
Ben Greear, linux-kernel@vger.kernel.org, netdev@vger.kernel.org
In-Reply-To: <1293901393.2535.49.camel@edumazet-laptop>
On Sat, Jan 1, 2011 at 12:03 PM, Eric Dumazet <eric.dumazet@gmail.com> wrote:
> Le mardi 14 décembre 2010 à 11:15 -0800, Matt Carlson a écrit :
>
>> Thanks for the comments Jesse. Below is an updated patch.
>>
>> Michael, I'm wondering if the difference in behavior can be explained by
>> the presence or absence of management firmware. Can you look at the
>> driver sign-on messages in your syslogs for ASF[]? I'm half expecting
>> the 5752 to show "ASF[0]" and the 5714 to show "ASF[1]". If you see
>> this, and the below patch doesn't fix the problem, let me know. I have
>> another test I'd like you to run.
>>
>> ----
>>
>> [PATCH] tg3: Use new VLAN code
>>
>> This patch pivots the tg3 driver to the new VLAN infrastructure.
>> All references to vlgrp have been removed and all VLAN code is
>> unconditionally active.
>>
>> Signed-off-by: Matt Carlson <mcarlson@broadcom.com>
[...]
> Hi Matt.
>
> Any news on this patch ?
>
> Without it, net-next-2.6 doesnt work for me on a vlan setup on top of
> bonding.
>
> (bond0 : eth1 & eth2, eth1 being bnx2, eth2 beging tg3)
>
> ip link add link bond0 vlan.103 type vlan id 103
> ip addr add 192.168.20.110/24 dev vlan.103
> ip link set vlan.103 up
>
>
> If active slave is eth1 (bnx2), everything works, but if active slave is
> eth2 (tg3), incoming tagged frames (on vlan 103) are lost.
This patch isn't quite right - it always disables vlan stripping
unless management firmware is in use, so it's not really a correct
fix.
You said that this used to work correctly on this NIC? Does it work
without a bond, just a vlan on the tg3 device? It sounds like Michael
has a problem with vlan stripping on one of his NICs but if it works
with just a vlan or on older kernels, it's probably not the same
thing.
If it works on bnx2, it would seem to be a driver problem but it would
be good to confirm that the tag in skb->vlan_tci is not being
delievered to the networking core in this case.
^ permalink raw reply
* [PATCH net-next] netdev: Update status of 8390 based drivers in MAINTAINERS
From: Paul Gortmaker @ 2011-01-01 23:28 UTC (permalink / raw)
To: davem; +Cc: netdev, Paul Gortmaker
With the original 8 bit ISA ne1000 card being over 20 years old, it
only makes sense to consider ne.c and all the other toplevel 8390
based driver files as legacy for obsolete hardware. The most
recent thing made in large quantities that was 8390 based were
those crazy PCI ne2k clones - and even they are now 10+ years old.
Also remove myself as maintainer, since the only changes to these
drivers going forward will be the generic API type changes that
touch all drivers.
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>
---
MAINTAINERS | 3 +--
1 files changed, 1 insertions(+), 2 deletions(-)
diff --git a/MAINTAINERS b/MAINTAINERS
index 4641289..aa835f7 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -166,9 +166,8 @@ F: drivers/serial/8250*
F: include/linux/serial_8250.h
8390 NETWORK DRIVERS [WD80x3/SMC-ELITE, SMC-ULTRA, NE2000, 3C503, etc.]
-M: Paul Gortmaker <p_gortmaker@yahoo.com>
L: netdev@vger.kernel.org
-S: Maintained
+S: Orphan / Obsolete
F: drivers/net/*8390*
F: drivers/net/ax88796.c
--
1.7.3.3
^ permalink raw reply related
* [PATCH net-next] net/Space: delete orphaned externs from deleted drivers
From: Paul Gortmaker @ 2011-01-01 23:15 UTC (permalink / raw)
To: davem; +Cc: netdev, Paul Gortmaker
The drivers associated with the prototypes in this commit have
been deleted some time ago, but the externs escaped detection.
Using a simple "git grep" shows that these references are
historical artefacts, only mentioned by the deleted lines.
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>
---
drivers/net/Space.c | 5 -----
1 files changed, 0 insertions(+), 5 deletions(-)
diff --git a/drivers/net/Space.c b/drivers/net/Space.c
index 9bb405b..068c356 100644
--- a/drivers/net/Space.c
+++ b/drivers/net/Space.c
@@ -55,8 +55,6 @@ extern struct net_device *eth16i_probe(int unit);
extern struct net_device *i82596_probe(int unit);
extern struct net_device *ewrk3_probe(int unit);
extern struct net_device *el1_probe(int unit);
-extern struct net_device *wavelan_probe(int unit);
-extern struct net_device *arlan_probe(int unit);
extern struct net_device *el16_probe(int unit);
extern struct net_device *elmc_probe(int unit);
extern struct net_device *elplus_probe(int unit);
@@ -68,7 +66,6 @@ extern struct net_device *ni5010_probe(int unit);
extern struct net_device *ni52_probe(int unit);
extern struct net_device *ni65_probe(int unit);
extern struct net_device *sonic_probe(int unit);
-extern struct net_device *SK_init(int unit);
extern struct net_device *seeq8005_probe(int unit);
extern struct net_device *smc_init(int unit);
extern struct net_device *atarilance_probe(int unit);
@@ -76,8 +73,6 @@ extern struct net_device *sun3lance_probe(int unit);
extern struct net_device *sun3_82586_probe(int unit);
extern struct net_device *apne_probe(int unit);
extern struct net_device *cs89x0_probe(int unit);
-extern struct net_device *hplance_probe(int unit);
-extern struct net_device *bagetlance_probe(int unit);
extern struct net_device *mvme147lance_probe(int unit);
extern struct net_device *tc515_probe(int unit);
extern struct net_device *lance_probe(int unit);
--
1.7.3.3
^ permalink raw reply related
* Re: [PATCH net-next 00/20] Delete more semi-useless stuff from TIPC.
From: David Miller @ 2011-01-01 22:56 UTC (permalink / raw)
To: paul.gortmaker; +Cc: netdev, allan.stephens
In-Reply-To: <20110101225330.GF30150@windriver.com>
From: Paul Gortmaker <paul.gortmaker@windriver.com>
Date: Sat, 1 Jan 2011 17:53:31 -0500
> Subject: [PATCH] tipc: update log.h re-include protection to reflect new name
>
> The tipc/dbg.h file was recently renamed to tipc/log.h,
> but the re-include define was not updated accordingly.
>
> Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>
Applied, thanks Paul.
^ permalink raw reply
* Re: [PATCH net-next 00/20] Delete more semi-useless stuff from TIPC.
From: Paul Gortmaker @ 2011-01-01 22:53 UTC (permalink / raw)
To: David Miller; +Cc: netdev, allan.stephens
In-Reply-To: <20110101.135952.226766850.davem@davemloft.net>
[Re: [PATCH net-next 00/20] Delete more semi-useless stuff from TIPC.] On 01/01/2011 (Sat 13:59) David Miller wrote:
> From: Paul Gortmaker <paul.gortmaker@windriver.com>
> Date: Fri, 31 Dec 2010 23:59:15 -0500
>
> > There really isn't a lot to see here technically. There is a bunch of
> > stuff in TIPC that I guess falls into the category of "seemed like a
> > good idea at the time", but is now just inactive or at best unstable
> > prototype code for things like slave nodes and routing features and
> > mulitcluster support. My favourite is this one:
> >
> > ----------
> > config TIPC_CLUSTERS
> > int "Maximum number of clusters in a zone"
> > depends on TIPC_ADVANCED
> > range 1 1
> > default "1"
> > help
> > Specifies how many clusters can be supported in a TIPC zone.
> >
> > *** Currently TIPC only supports a single cluster per zone. ***
> > ----------
> >
> > The remaining commits are just de-uglifying what is left, fixing whitespace
> > crimes and other common errors typically picked up in janitor patches.
>
> All applied, thanks Paul.
>
> net/tipc/log.h still uses "_TIPC_DBG_H" for it's multiple-include protection
> tests. Please send me a patch to rename that to _TIPC_LOG_H
Ah crap, I should have spotted that. Thanks for noticing it. Patch
follows.
Paul.
----
>From 443eb42b4e3e45b54698ee787ce7e5fb16acb57d Mon Sep 17 00:00:00 2001
From: Paul Gortmaker <paul.gortmaker@windriver.com>
Date: Sat, 1 Jan 2011 17:32:04 -0500
Subject: [PATCH] tipc: update log.h re-include protection to reflect new name
The tipc/dbg.h file was recently renamed to tipc/log.h,
but the re-include define was not updated accordingly.
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>
---
net/tipc/log.h | 4 ++--
1 files changed, 2 insertions(+), 2 deletions(-)
diff --git a/net/tipc/log.h b/net/tipc/log.h
index f4343bb..2248d96 100644
--- a/net/tipc/log.h
+++ b/net/tipc/log.h
@@ -34,8 +34,8 @@
* POSSIBILITY OF SUCH DAMAGE.
*/
-#ifndef _TIPC_DBG_H
-#define _TIPC_DBG_H
+#ifndef _TIPC_LOG_H
+#define _TIPC_LOG_H
/**
* struct print_buf - TIPC print buffer structure
--
1.7.3.3
^ permalink raw reply related
* Re: [PATCH] new UDPCP Communication Protocol
From: Eric Dumazet @ 2011-01-01 22:23 UTC (permalink / raw)
To: stefani; +Cc: linux-kernel, akpm, davem, netdev, shemminger
In-Reply-To: <1293918276-28773-1-git-send-email-stefani@seibold.net>
Le samedi 01 janvier 2011 à 22:44 +0100, stefani@seibold.net a écrit :
> From: Stefani Seibold <stefani@seibold.net>
>
> Changelog:
> 31.12.2010 first proposal
> 01.01.2011 code cleanup and fixes suggest by Eric Dumazet
>
> UDPCP is a communication protocol specified by the Open Base Station
> Architecture Initiative Special Interest Group (OBSAI SIG). The
> protocol is based on UDP and is designed to meet the needs of "Mobile
> Communcation Base Station" internal communications. It is widely used by
> the major networks infrastructure supplier.
>
> The UDPCP communication service supports the following features:
>
> -Connectionless communication for serial mode data transfer
> -Acknowledged and unacknowledged transfer modes
> -Retransmissions Algorithm
> -Checksum Algorithm using Adler32
> -Fragmentation of long messages (disassembly/reassembly) to match to the MTU
> during transport:
> -Broadcasting and multicasting messages to multiple peers in unacknowledged
> transfer mode
>
> UDPCP supports application level messages up to 64 KBytes (limited by 16-bit
> packet data length field). Messages that are longer than the MTU will be
> fragmented to the MTU.
>
> UDPCP provides a reliable transport service that will perform message
> retransmissions in case transport failures occur.
>
> The code is also a nice example how to implement a UDP based protocol as
> a kernel socket modules.
>
> Due the nature of UDPCP which has no sliding windows support, the latency has a
> huge impact. The perfomance increase by implementing as a kernel module is
> about the factor 10, because there are no context switches and data packets or
> ACKs will be handled in the interrupt service.
>
> There are no side effects to the network subsystems so i ask for merge it
> into linux-next. Hope you like it.
>
> The patch is against 2.6.37-rc8
>
Hi Stefani
1) Please base your next trys on net-next-2.6 : This is the reference
for stuff like that. It probably does not matter a lot, but still...
2) I still see some _irq() variants of spinlock(). Its not necessary in
network stack at the level you are working (process context, and
softirqs)
Please only use _bh() variants, it's enough.
3) I see UDPLITE references in your code. Are you sure UDPCP protocol
can really work on top of UDPLITE ? I think not, so please remove dead
code.
4) udpcp_release_sock() seems expensive to me. Why not testing
usk->timeout before releasing sock lock, and save a lock/unlock pair ?
static inline void udpcp_release_sock(struct sock *sk)
{
struct udpcp_sock *usk = udpcp_sk(sk);
if (usk->timeout)
udpcp_handle_timeout(sk);
release_sock(sk);
}
5) In udpcp_timeout(), if you find socket is locked by user, you set
timeout to one and rearm a timer to udpcp_timer(sk, jiffies + 1);
Why is it needed, since user process is going to handle the timeout
indication from udpcp_release_sock() ?
Thanks
^ permalink raw reply
* Re: [PATCH net-next] sky2: implement 64 bit stats
From: David Miller @ 2011-01-01 22:02 UTC (permalink / raw)
To: shemminger; +Cc: netdev
In-Reply-To: <20101231173427.17f07ad3@nehalam>
From: Stephen Hemminger <shemminger@vyatta.com>
Date: Fri, 31 Dec 2010 17:34:27 -0800
> This implements 64 bit statistics support and fixes races when reading
> counter values. The PHY counters can only be accessed 16 bits at a time,
> so they are subject to carry races.
>
> NB:
> * TX/RX counters are maintained in software because the the hardware packet count
> is only a 32 bit value.
>
> * Error counters are really only 32 bit.
>
> * Old 32 bit counter fields in dev->stats still used for some
> software counters
>
> Signed-off-by: Stephen Hemminger <shemminger@vyatta.com>
Applied, thanks Stephen.
^ permalink raw reply
* Re: [PATCH net-next 00/20] Delete more semi-useless stuff from TIPC.
From: David Miller @ 2011-01-01 21:59 UTC (permalink / raw)
To: paul.gortmaker; +Cc: netdev, allan.stephens
In-Reply-To: <1293857975-30267-1-git-send-email-paul.gortmaker@windriver.com>
From: Paul Gortmaker <paul.gortmaker@windriver.com>
Date: Fri, 31 Dec 2010 23:59:15 -0500
> There really isn't a lot to see here technically. There is a bunch of
> stuff in TIPC that I guess falls into the category of "seemed like a
> good idea at the time", but is now just inactive or at best unstable
> prototype code for things like slave nodes and routing features and
> mulitcluster support. My favourite is this one:
>
> ----------
> config TIPC_CLUSTERS
> int "Maximum number of clusters in a zone"
> depends on TIPC_ADVANCED
> range 1 1
> default "1"
> help
> Specifies how many clusters can be supported in a TIPC zone.
>
> *** Currently TIPC only supports a single cluster per zone. ***
> ----------
>
> The remaining commits are just de-uglifying what is left, fixing whitespace
> crimes and other common errors typically picked up in janitor patches.
All applied, thanks Paul.
net/tipc/log.h still uses "_TIPC_DBG_H" for it's multiple-include protection
tests. Please send me a patch to rename that to _TIPC_LOG_H
Thanks.
^ permalink raw reply
* Re: [PATCH net-next-2.6] tg3: fix warnings
From: David Miller @ 2011-01-01 21:51 UTC (permalink / raw)
To: eric.dumazet; +Cc: rjw, netdev, mcarlson, mchan, linux-pm, jesse
In-Reply-To: <1293895366.2535.43.camel@edumazet-laptop>
From: Eric Dumazet <eric.dumazet@gmail.com>
Date: Sat, 01 Jan 2011 16:22:46 +0100
> Happy new year everybody ;)
Same to you Eric :)
> [PATCH net-next-2.6] tg3: fix warnings
>
> In case CONFIG_PM_SLEEP is disabled, we dont need tg3_suspend() and
> tg3_resume().
>
> drivers/net/tg3.c:15056: warning: ‘tg3_suspend’ defined but not used
> drivers/net/tg3.c:15110: warning: ‘tg3_resume’ defined but not used
>
> Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Applied, thanks.
^ permalink raw reply
* [PATCH] new UDPCP Communication Protocol
From: stefani @ 2011-01-01 21:44 UTC (permalink / raw)
To: linux-kernel, akpm, davem, netdev, eric.dumazet, shemminger; +Cc: stefani
From: Stefani Seibold <stefani@seibold.net>
Changelog:
31.12.2010 first proposal
01.01.2011 code cleanup and fixes suggest by Eric Dumazet
UDPCP is a communication protocol specified by the Open Base Station
Architecture Initiative Special Interest Group (OBSAI SIG). The
protocol is based on UDP and is designed to meet the needs of "Mobile
Communcation Base Station" internal communications. It is widely used by
the major networks infrastructure supplier.
The UDPCP communication service supports the following features:
-Connectionless communication for serial mode data transfer
-Acknowledged and unacknowledged transfer modes
-Retransmissions Algorithm
-Checksum Algorithm using Adler32
-Fragmentation of long messages (disassembly/reassembly) to match to the MTU
during transport:
-Broadcasting and multicasting messages to multiple peers in unacknowledged
transfer mode
UDPCP supports application level messages up to 64 KBytes (limited by 16-bit
packet data length field). Messages that are longer than the MTU will be
fragmented to the MTU.
UDPCP provides a reliable transport service that will perform message
retransmissions in case transport failures occur.
The code is also a nice example how to implement a UDP based protocol as
a kernel socket modules.
Due the nature of UDPCP which has no sliding windows support, the latency has a
huge impact. The perfomance increase by implementing as a kernel module is
about the factor 10, because there are no context switches and data packets or
ACKs will be handled in the interrupt service.
There are no side effects to the network subsystems so i ask for merge it
into linux-next. Hope you like it.
The patch is against 2.6.37-rc8
- Stefani
Signed-off-by: Stefani Seibold <stefani@seibold.net>
---
include/linux/socket.h | 7 +-
include/net/udp.h | 1 +
include/net/udpcp.h | 47 +
net/Kconfig | 1 +
net/Makefile | 1 +
net/ipv4/ip_output.c | 2 +
net/ipv4/ip_sockglue.c | 2 +
net/udpcp/Kconfig | 34 +
net/udpcp/Makefile | 5 +
net/udpcp/udpcp.c | 2854 ++++++++++++++++++++++++++++++++++++++++++++++++
10 files changed, 2952 insertions(+), 2 deletions(-)
create mode 100644 include/net/udpcp.h
create mode 100644 net/udpcp/Kconfig
create mode 100644 net/udpcp/Makefile
create mode 100644 net/udpcp/udpcp.c
diff --git a/include/linux/socket.h b/include/linux/socket.h
index 86b652f..755eeb8 100644
--- a/include/linux/socket.h
+++ b/include/linux/socket.h
@@ -193,7 +193,8 @@ struct ucred {
#define AF_PHONET 35 /* Phonet sockets */
#define AF_IEEE802154 36 /* IEEE802154 sockets */
#define AF_CAIF 37 /* CAIF sockets */
-#define AF_MAX 38 /* For now.. */
+#define AF_UDPCP 38 /* UDPCP sockets */
+#define AF_MAX 39 /* For now.. */
/* Protocol families, same as address families. */
#define PF_UNSPEC AF_UNSPEC
@@ -203,7 +204,7 @@ struct ucred {
#define PF_AX25 AF_AX25
#define PF_IPX AF_IPX
#define PF_APPLETALK AF_APPLETALK
-#define PF_NETROM AF_NETROM
+#define PF_NETROM AF_NETROM
#define PF_BRIDGE AF_BRIDGE
#define PF_ATMPVC AF_ATMPVC
#define PF_X25 AF_X25
@@ -234,6 +235,7 @@ struct ucred {
#define PF_PHONET AF_PHONET
#define PF_IEEE802154 AF_IEEE802154
#define PF_CAIF AF_CAIF
+#define PF_UDPCP AF_UDPCP
#define PF_MAX AF_MAX
/* Maximum queue length specifiable by listen. */
@@ -307,6 +309,7 @@ struct ucred {
#define SOL_RDS 276
#define SOL_IUCV 277
#define SOL_CAIF 278
+#define SOL_UDPCP 279
/* IPX options */
#define IPX_TYPE 1
diff --git a/include/net/udp.h b/include/net/udp.h
index bb967dd..82c95a7 100644
--- a/include/net/udp.h
+++ b/include/net/udp.h
@@ -47,6 +47,7 @@ struct udp_skb_cb {
} header;
__u16 cscov;
__u8 partial_cov;
+ __u8 udpcp_flag;
};
#define UDP_SKB_CB(__skb) ((struct udp_skb_cb *)((__skb)->cb))
diff --git a/include/net/udpcp.h b/include/net/udpcp.h
new file mode 100644
index 0000000..45180a5
--- /dev/null
+++ b/include/net/udpcp.h
@@ -0,0 +1,47 @@
+/* Definitions for UDPCP sockets. */
+
+#ifndef __LINUX_IF_UDPCP
+#define __LINUX_IF_UDPCP
+
+#include "linux/ioctl.h"
+
+#define UDPCP_MAX_MSGSIZE 65487
+
+#define UDPCP_MAX_WAIT_SEC 60
+
+#define UDPCP_OPT_TRANSFER_MODE 0
+#define UDPCP_OPT_CHECKSUM_MODE 1
+#define UDPCP_OPT_TX_TIMEOUT 2
+#define UDPCP_OPT_RX_TIMEOUT 3
+#define UDPCP_OPT_MAXTRY 4
+#define UDPCP_OPT_OUTSTANDING_ACKS 5
+
+#define UDPCP_NOACK 0
+#define UDPCP_ACK 1
+#define UDPCP_SINGLE_ACK 2
+#define UDPCP_NOCHECKSUM 3
+#define UDPCP_CHECKSUM 4
+
+#define UDPCP_IOC_MAGIC 251
+
+#define UDPCP_IOCTL_GET_STATISTICS \
+ _IOR(UDPCP_IOC_MAGIC, 0x01, struct udpcp_statistics *)
+#define UDPCP_IOCTL_RESET_STATISTICS \
+ _IO(UDPCP_IOC_MAGIC, 0x02)
+#define UDPCP_IOCTL_SYNC \
+ _IOR(UDPCP_IOC_MAGIC, 0x03, unsigned long)
+
+struct udpcp_statistics {
+ unsigned int txMsgs; /* Num of transmitted messages */
+ unsigned int rxMsgs; /* Num of received messages */
+ unsigned int txNodes; /* Num of receiver nodes */
+ unsigned int rxNodes; /* Num of transmitter nodes */
+ unsigned int txTimeout; /* Num of unsuccessful transmissions */
+ unsigned int rxTimeout; /* Num of partial message receptions */
+ unsigned int txRetries; /* Num of resends */
+ unsigned int rxDiscardedFrags; /* Num of discarded fragments */
+ unsigned int crcErrors; /* Num of crc errors detected */
+};
+
+#endif
+
diff --git a/net/Kconfig b/net/Kconfig
index 55fd82e..4a206fc 100644
--- a/net/Kconfig
+++ b/net/Kconfig
@@ -294,6 +294,7 @@ source "net/rfkill/Kconfig"
source "net/9p/Kconfig"
source "net/caif/Kconfig"
source "net/ceph/Kconfig"
+source "net/udpcp/Kconfig"
endif # if NET
diff --git a/net/Makefile b/net/Makefile
index 6b7bfd7..a17ae27 100644
--- a/net/Makefile
+++ b/net/Makefile
@@ -69,3 +69,4 @@ endif
obj-$(CONFIG_WIMAX) += wimax/
obj-$(CONFIG_DNS_RESOLVER) += dns_resolver/
obj-$(CONFIG_CEPH_LIB) += ceph/
+obj-$(CONFIG_UDPCP) += udpcp/
diff --git a/net/ipv4/ip_output.c b/net/ipv4/ip_output.c
index 439d2a3..55b2d0c 100644
--- a/net/ipv4/ip_output.c
+++ b/net/ipv4/ip_output.c
@@ -1085,6 +1085,7 @@ error:
IP_INC_STATS(sock_net(sk), IPSTATS_MIB_OUTDISCARDS);
return err;
}
+EXPORT_SYMBOL(ip_append_data);
ssize_t ip_append_page(struct sock *sk, struct page *page,
int offset, size_t size, int flags)
@@ -1341,6 +1342,7 @@ error:
IP_INC_STATS(net, IPSTATS_MIB_OUTDISCARDS);
goto out;
}
+EXPORT_SYMBOL(ip_push_pending_frames);
/*
* Throw away all pending data on the socket.
diff --git a/net/ipv4/ip_sockglue.c b/net/ipv4/ip_sockglue.c
index 3948c86..310369c 100644
--- a/net/ipv4/ip_sockglue.c
+++ b/net/ipv4/ip_sockglue.c
@@ -226,6 +226,7 @@ int ip_cmsg_send(struct net *net, struct msghdr *msg, struct ipcm_cookie *ipc)
}
return 0;
}
+EXPORT_SYMBOL(ip_cmsg_send);
/* Special input handler for packets caught by router alert option.
@@ -369,6 +370,7 @@ void ip_local_error(struct sock *sk, int err, __be32 daddr, __be16 port, u32 inf
if (sock_queue_err_skb(sk, skb))
kfree_skb(skb);
}
+EXPORT_SYMBOL(ip_local_error);
/*
* Handle MSG_ERRQUEUE
diff --git a/net/udpcp/Kconfig b/net/udpcp/Kconfig
new file mode 100644
index 0000000..a58c1b0
--- /dev/null
+++ b/net/udpcp/Kconfig
@@ -0,0 +1,34 @@
+#
+# UDPCP protocol
+#
+
+config UDPCP
+ tristate "UDPCP Communication Protocol"
+ depends on INET
+ ---help---
+ UDPCP is a communication protocol specified by the Open Base Station
+ Architecture Initiative Special Interest Group (OBSAI SIG). The
+ protocol is based on UDP and is designed to meet the needs of "Mobile
+ Communcation Base Station" internal communications.
+
+ The UDPCP communication service supports the following features:
+
+ -Connectionless communication for serial mode data transfer
+ -Acknowledged and unacknowledged transfer modes
+ -Retransmissions Algorithm
+ -Checksum Algorithm using Adler32
+ -Fragmentation of long messages (disassembly/reassembly) to
+ match to the MTU during transport:
+ -Broadcasting and multicasting messages to multiple peers in
+ unacknowledged transfer mode
+
+ UDPCP supports application level messages up to 64 KBytes (limited
+ by 16-bit packet data length field). Messages that are longer than the
+ MTU will be fragmented to the MTU.
+
+ UDPCP provides a reliable transport service that will perform message
+ retransmissions in case transport failures occur.
+
+ To compile this driver as a module, choose M here: the module
+ will be called udpcp.
+
diff --git a/net/udpcp/Makefile b/net/udpcp/Makefile
new file mode 100644
index 0000000..37f87c5
--- /dev/null
+++ b/net/udpcp/Makefile
@@ -0,0 +1,5 @@
+#
+# Makefile for UDPCP support code.
+#
+
+obj-$(CONFIG_UDPCP) += udpcp.o
diff --git a/net/udpcp/udpcp.c b/net/udpcp/udpcp.c
new file mode 100644
index 0000000..c7990c2
--- /dev/null
+++ b/net/udpcp/udpcp.c
@@ -0,0 +1,2854 @@
+/*
+ * UDPCP communication protocol
+ *
+ * Copyright (C) 2010 Stefani Seibold <stefani@seibold.net>
+ * in order of NSN Ulm/Germany
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
+ *
+ */
+
+#include <net/xfrm.h>
+#include <net/protocol.h>
+#include <net/ip.h>
+#include <net/udp.h>
+#include <net/udplite.h>
+#include <net/inet_common.h>
+#include <linux/zutil.h>
+#include <linux/module.h>
+#include <linux/proc_fs.h>
+#include <linux/spinlock.h>
+#include <linux/errqueue.h>
+#include <linux/atomic.h>
+
+#include <net/udpcp.h>
+
+#define VERSION "0.70"
+
+/*
+ * UDPCP Protocol default parameters
+ */
+#define UDPCP_TX_TIMEOUT 100 /* milliseconds */
+#define UDPCP_RX_TIMEOUT 1000 /* milliseconds */
+#define UDPCP_TX_MAXTRY 5
+#define UDPCP_OUTSTANDING_ACKS 1
+
+/*
+ * UDPCP Protocol definitions
+ */
+#define UDPCP_MSG_TYPE_BIT 14
+#define UDPCP_PROTOCOL_VERSION_BIT 11
+#define UDPCP_NO_ACK_BIT 10
+#define UDPCP_CHECKSUM_BIT 9
+#define UDPCP_SINGLE_ACK_BIT 8
+#define UDPCP_DUPLICATE_BIT 7
+
+#define UDPCP_MSG_TYPE_MASK (3 << UDPCP_MSG_TYPE_BIT)
+#define UDPCP_PROTOCOL_MASK (7 << UDPCP_PROTOCOL_VERSION_BIT)
+
+#define UDPCP_MSG_TYPE_DATA (1 << UDPCP_MSG_TYPE_BIT)
+#define UDPCP_MSG_TYPE_ACK (2 << UDPCP_MSG_TYPE_BIT)
+#define UDPCP_PROTOCOL_VERSION_2 (2 << UDPCP_PROTOCOL_VERSION_BIT)
+
+#define UDPCP_NO_ACK_FLAG (1 << UDPCP_NO_ACK_BIT)
+#define UDPCP_CHECKSUM_FLAG (1 << UDPCP_CHECKSUM_BIT)
+#define UDPCP_SINGLE_ACK_FLAG (1 << UDPCP_SINGLE_ACK_BIT)
+#define UDPCP_DUPLICATE_FLAG (1 << UDPCP_DUPLICATE_BIT)
+
+/*
+ * helper macros
+ */
+#define list_to_udpcpdest(d) container_of(d, struct udpcp_dest, list)
+#define list_to_udpcpsock(d) container_of(d, struct udpcp_sock, udpcplist)
+
+#define UDPCP_HDRSIZE (sizeof(struct udpcphdr)-sizeof(struct udphdr))
+
+#define RX_NODE 1
+#define TX_NODE 2
+
+/*
+ * name of the /proc entry
+ */
+#define UDPCP_PROC "driver/udpcp"
+
+/*
+ * UDPCP message header
+ */
+struct udpcphdr {
+ struct udphdr udphdr;
+ __be32 chksum;
+ __be16 msginfo;
+ u8 fragamount;
+ u8 fragnum;
+ __be16 msgid;
+ __be16 length;
+};
+
+/*
+ * UDPCP destination descriptor
+ *
+ * For each communication address an individual destination descriptor will
+ * be create.
+ *
+ * The fields has the following meanings:
+ *
+ * list: link list: part of udpcp_sock.destlist
+ * xmit: messages fragments to be transmit
+ * tx_time: timestamp of the last transmitted message fragment
+ * rx_time: timestamp ot the last received message fragment
+ * txTimeout: statistic use only: number of transmit timeout
+ * rxTimeout: statistic use only: number of receive timeout
+ * txRetries: statistic use only: number of transmit retries
+ * rxDiscardedFrags: statistic use only: number of discarded messages
+ * xmit_wait: message fragment which is waiting for an ACK
+ * xmit_last: last fragment transmitted
+ * recv_msg: first fragment of the received message
+ * recv_last: last fragment of the received message
+ * lastmsg: last messages fragment header received
+ * ipc: linux internal ipc cookie
+ * fl: flow/routing information
+ * rt: routing entry currently used for this destination
+ * addr: ipv4 destination address
+ * port: destination port number
+ * msgid: current message id for outgoing data messages
+ * use_flag: statistic use only: flag for dest using TX and/or RX
+ * insync: flag for protocol synchronization
+ * ackmode; ack mode for the current assembled message
+ * chkmode; checksum mode for the current assembled message
+ * try: current number of retries xmit_wait message
+ * acks: number of outstandig ack's
+ */
+struct udpcp_dest {
+ struct list_head list;
+ struct sk_buff_head xmit;
+ unsigned long tx_time;
+ unsigned long rx_time;
+ u32 txTimeout;
+ u32 rxTimeout;
+ u32 txRetries;
+ u32 rxDiscardedFrags;
+ struct sk_buff *xmit_wait;
+ struct sk_buff *xmit_last;
+ struct sk_buff *recv_msg;
+ struct sk_buff *recv_last;
+ struct udpcphdr lastmsg;
+ struct ipcm_cookie ipc;
+ struct flowi fl;
+ struct rtable *rt;
+ __be32 addr;
+ __be16 port;
+ u16 msgid;
+ u8 use_flag;
+ u8 insync;
+ u8 ackmode;
+ u8 chkmode;
+ u8 try;
+ u8 acks;
+};
+
+/*
+ * UDPCP socket descriptor
+ *
+ * For each opened socket individual socket descriptor will
+ * be created
+ *
+ * The fields has the following meanings:
+ *
+ * udpsock: UDP socket has to be the first member of udpcp_sock
+ * assembly: messages fragments currently assembled
+ * assembly_len: current length of the assembled message
+ * assembly_dest: current destination assembled
+ * wq: wait queue for UDPCP_IOCTL_SYNC
+ * destlist: head of destination descriptors link list
+ * udpcplist: link list: part of udpcp_list
+ * timer: timeout handler
+ * stat: statistics for this socket
+ * pending: number of pending messages fragment in the queues
+ * tx_timeout: transmit timeout in jiffies
+ * rx_timeout: receive timeout in jiffies
+ * udp_data_ready: original data_ready handler for this socket
+ * ackmode: default ack mode
+ * chkmode: default checksum mode
+ * maxtry: max. number of resends
+ * acks: max. number of outstandig ack's
+ * timeout: flag for unhandled timeout
+ */
+struct udpcp_sock {
+ struct udp_sock udpsock;
+ struct sk_buff_head assembly;
+ u32 assembly_len;
+ struct udpcp_dest *assembly_dest;
+ wait_queue_head_t wq;
+ struct list_head destlist;
+ struct list_head udpcplist;
+ struct timer_list timer;
+ struct udpcp_statistics stat;
+ u32 pending;
+ unsigned long tx_timeout;
+ unsigned long rx_timeout;
+ void (*udp_data_ready) (struct sock *sk, int bytes);
+ u8 ackmode;
+ u8 chkmode;
+ u8 maxtry;
+ u8 acks;
+ u8 timeout;
+};
+
+/* head of struct udpcp_sock.udpcplist link list */
+static struct list_head udpcp_list;
+
+/* spinlock for race free access to the static variables */
+static spinlock_t udpcp_lock;
+
+/* debug flag, set != 0 to enable debug */
+static int debug;
+
+/* overall UDPCP statistics */
+static atomic_t udpcp_txMsgs;
+static atomic_t udpcp_rxMsgs;
+static atomic_t udpcp_txNodes;
+static atomic_t udpcp_rxNodes;
+static atomic_t udpcp_txTimeout;
+static atomic_t udpcp_rxTimeout;
+static atomic_t udpcp_txRetries;
+static atomic_t udpcp_rxDiscardedFrags;
+static atomic_t udpcp_crcErrors;
+
+module_param(debug, int, 0);
+MODULE_PARM_DESC(debug, "Debug enabled or not");
+
+#ifdef CONFIG_PROC_FS
+/*
+ * Handle /proc/driver/udpcp
+ *
+ * Show the statistics information
+ */
+static int udpcp_proc(char *page, char **start, off_t off, int count, int *eof,
+ void *data)
+{
+ int len;
+
+ len = snprintf(page, count,
+ "txMsgs: %u\n"
+ "rxMsgs: %u\n"
+ "txNodes: %u\n"
+ "rxNodes: %u\n"
+ "txTimeout: %u\n"
+ "rxTimeout: %u\n"
+ "txRetries: %u\n"
+ "rxDiscaredFrags: %u\n"
+ "crcErrors: %u\n",
+ atomic_read(&udpcp_txMsgs),
+ atomic_read(&udpcp_rxMsgs),
+ atomic_read(&udpcp_txNodes),
+ atomic_read(&udpcp_rxNodes),
+ atomic_read(&udpcp_txTimeout),
+ atomic_read(&udpcp_rxTimeout),
+ atomic_read(&udpcp_txRetries),
+ atomic_read(&udpcp_rxDiscardedFrags),
+ atomic_read(&udpcp_crcErrors)
+ );
+
+ if (len <= off)
+ return 0;
+
+ len -= off;
+
+ if (len > count)
+ return count;
+
+ return len;
+}
+#endif
+
+/*
+ * Helper for the UDPCP header from a socket buffer
+ */
+static inline struct udpcphdr *udpcp_hdr(const struct sk_buff *skb)
+{
+ return (struct udpcphdr *)skb_transport_header(skb);
+}
+
+/*
+ * Helper for conversion a basic socket into a UDPCP socket
+ */
+static inline struct udpcp_sock *udpcp_sk(const struct sock *sk)
+{
+ return (struct udpcp_sock *)sk;
+}
+
+/*
+ * Dump the transport data of a socket buffer
+ */
+static inline void dump_data(struct sk_buff *skb, unsigned int max)
+{
+ unsigned int i;
+ unsigned char *data;
+ int data_len;
+
+ data = skb_transport_header(skb) + sizeof(struct udpcphdr);
+ data_len = skb_tail_pointer(skb) - data;
+
+ pr_debug(" data: ");
+
+ if (!data_len) {
+ pr_cont("<none>\n");
+ return;
+ }
+
+ if (max > data_len)
+ max = data_len;
+
+ for (i = 0; i < max; i++)
+ pr_cont("%02x ", data[i]);
+
+ if (data_len > max)
+ pr_cont("...");
+ pr_cont("\n");
+}
+
+/*
+ * Dump and decode a msginfo value
+ */
+static inline void dump_msginfo(u16 msginfo)
+{
+ pr_debug(" msginfo:0x%04x (", msginfo);
+
+ pr_cont("PCKT:");
+ switch (msginfo & UDPCP_MSG_TYPE_MASK) {
+ case UDPCP_MSG_TYPE_DATA:
+ pr_cont("DATA");
+ break;
+ case UDPCP_MSG_TYPE_ACK:
+ pr_cont("ACK");
+ break;
+ default:
+ pr_cont("UNKNOWN");
+ break;
+ }
+ pr_cont(" VER:%d",
+ (msginfo & UDPCP_PROTOCOL_MASK) >> UDPCP_PROTOCOL_VERSION_BIT);
+
+ if (msginfo & UDPCP_NO_ACK_FLAG)
+ pr_cont(" NO_ACK");
+ if (msginfo & UDPCP_CHECKSUM_FLAG)
+ pr_cont(" CHECKSUM");
+ if (msginfo & UDPCP_SINGLE_ACK_FLAG)
+ pr_cont(" SINGLE_ACK");
+ if (msginfo & UDPCP_DUPLICATE_FLAG)
+ pr_cont(" DUPLICATE");
+ pr_cont(")\n");
+}
+
+/*
+ * Dump and decode a UDPCP message fragment
+ */
+static void dump_msg(const char *action, struct sk_buff *skb, __be32 saddr,
+ __be32 daddr)
+{
+ struct udpcphdr *uh = udpcp_hdr(skb);
+
+ pr_debug("udpcp: %s (%lu)\n", action, jiffies);
+
+ pr_debug(" src:0x%08x:%d dst:0x%08x:%d fraglen:%d\n",
+ saddr, uh->udphdr.source, daddr, uh->udphdr.dest, skb->len);
+
+ pr_debug(" fragamount:%u fragnum:%u msgid:%u%s"
+ " length:%u checksum:0x%08x\n",
+ uh->fragamount, uh->fragnum, ntohs(uh->msgid),
+ (!uh->msgid) ? "(Sync)" : "", ntohs(uh->length),
+ ntohl(uh->chksum)
+ );
+
+ dump_msginfo(ntohs(uh->msginfo));
+ dump_data(skb, 16);
+}
+
+/*
+ * Create a new destination descriptor for the given IPV4 address and port
+ */
+static struct udpcp_dest *new_dest(struct sock *sk, __be32 addr, __be16 port)
+{
+ struct udpcp_dest *dest;
+ struct udpcp_sock *usk = udpcp_sk(sk);
+
+ dest = kzalloc(sizeof(*dest), sk->sk_allocation);
+
+ if (dest) {
+ skb_queue_head_init(&dest->xmit);
+ dest->addr = addr;
+ dest->port = port;
+ dest->ackmode = UDPCP_ACK;
+ list_add_tail(&dest->list, &usk->destlist);
+ }
+
+ return dest;
+}
+
+/*
+ * Lookup for a destination descriptor for the given IPV4 address and port
+ */
+static struct udpcp_dest *__find_dest(struct sock *sk, __be32 addr, __be16 port)
+{
+ struct udpcp_dest *dest;
+ struct list_head *p;
+ struct udpcp_sock *usk = udpcp_sk(sk);
+
+ list_for_each(p, &usk->destlist) {
+ dest = list_to_udpcpdest(p);
+
+ if ((dest->addr == addr) && (dest->port == port))
+ return dest;
+ }
+ return NULL;
+}
+
+/*
+ * Lookup for a destination descriptor and create a new one if no
+ * descriptor was found.
+ */
+static struct udpcp_dest *find_dest(struct sock *sk, __be32 addr, __be16 port)
+{
+ struct udpcp_dest *dest;
+
+ dest = __find_dest(sk, addr, port);
+
+ if (!dest)
+ dest = new_dest(sk, addr, port);
+
+ return dest;
+}
+
+/*
+ * Calculate udp checksum, mostly stolen from udp stack
+ */
+static void udpcp_do_csum(struct sock *sk, struct sk_buff *skb,
+ struct udpcp_dest *dest)
+{
+ struct flowi *fl = &dest->fl;
+ struct udphdr *uh = udp_hdr(skb);
+ __wsum csum = 0;
+ unsigned short len = ntohs(uh->len);
+
+ if (IS_UDPLITE(sk)) {
+ int cscov = udplite_sender_cscov(udp_sk(sk), uh);
+ int off = skb_transport_offset(skb);
+ int n = skb->len - off;
+
+ skb->ip_summed = CHECKSUM_NONE;
+ csum = skb_checksum(skb, off, (cscov > n) ? n : cscov, csum);
+ } else {
+ if (sk->sk_no_check == UDP_CSUM_NOXMIT) {
+ skb->ip_summed = CHECKSUM_NONE;
+ return;
+ }
+ if (skb->ip_summed == CHECKSUM_PARTIAL) {
+ /* UDP hardware csum */
+ skb->csum_start = skb_transport_header(skb) - skb->head;
+ skb->csum_offset = offsetof(struct udphdr, check);
+ uh->check =
+ ~csum_tcpudp_magic(fl->fl4_src, fl->fl4_dst, len,
+ sk->sk_protocol, 0);
+ return;
+ }
+ csum = csum_partial(uh, sizeof(struct udpcphdr), 0);
+ csum = csum_add(csum, skb->csum);
+ }
+
+ /* add protocol-dependent pseudo-header */
+ uh->check =
+ csum_tcpudp_magic(fl->fl4_src, fl->fl4_dst, len, sk->sk_protocol,
+ csum);
+ if (uh->check == 0)
+ uh->check = CSUM_MANGLED_0;
+}
+
+/*
+ * Fetch data from kernel space and fill in checksum if needed.
+ */
+static int ip_reply_glue_bits(void *dptr, char *to, int offset,
+ int len, int odd, struct sk_buff *skb)
+{
+ __wsum csum;
+
+ csum = csum_partial_copy_nocheck(dptr+offset, to, len, 0);
+ skb->csum = csum_block_add(skb->csum, csum, odd);
+ return 0;
+}
+
+/*
+ * Send an ack for a received data message fragment
+ *
+ * If the argument duplicate is true a ACK with UDPCP_DUPLICATE_FLAG set will
+ * be send
+ */
+static void udpcp_send_ack(struct sock *sk, struct sk_buff *skb,
+ struct udpcp_dest *dest, int duplicate)
+{
+ struct inet_sock *inet = inet_sk(sk);
+ struct udpcphdr *uh = udpcp_hdr(skb);
+ struct rtable *rt = NULL;
+ __wsum csum;
+ struct ipcm_cookie ipc;
+ struct udpcphdr rep;
+
+ memset(&rep, 0, sizeof(rep));
+
+ /* Swap the send and the receive ports. */
+ rep.udphdr.source = uh->udphdr.dest;
+ rep.udphdr.dest = uh->udphdr.source;
+ rep.udphdr.len = htons(sizeof(struct udpcphdr));
+
+ rep.msginfo = htons(UDPCP_MSG_TYPE_ACK |
+ UDPCP_NO_ACK_FLAG |
+ UDPCP_SINGLE_ACK_FLAG | UDPCP_PROTOCOL_VERSION_2);
+ if (duplicate)
+ rep.msginfo |= htons(UDPCP_DUPLICATE_FLAG);
+ else
+ memcpy(&dest->lastmsg, uh, sizeof(dest->lastmsg));
+ rep.msgid = uh->msgid;
+ rep.fragamount = uh->fragamount;
+ rep.fragnum = uh->fragnum;
+ rep.length = 0;
+ rep.chksum = 0;
+ if (ntohs(uh->msginfo) & UDPCP_CHECKSUM_FLAG) {
+ u8 *data;
+ u32 data_len;
+
+ data = (u8 *) &rep + sizeof(struct udphdr);
+ data_len = sizeof(struct udpcphdr)-sizeof(struct udphdr);
+
+ rep.msginfo |= htons(UDPCP_CHECKSUM_FLAG);
+ rep.chksum = htonl(zlib_adler32(1, data, data_len));
+ }
+
+ if (unlikely(debug)) {
+ struct sk_buff tmp;
+
+ tmp.len = ntohs(rep.udphdr.len);
+ tmp.head = tmp.transport_header = tmp.data = (void *)&rep;
+ tmp.tail = tmp.head + tmp.len;
+
+ dump_msg("ack msg", &tmp, ip_hdr(skb)->daddr,
+ ip_hdr(skb)->saddr);
+ }
+
+ csum = csum_tcpudp_nofold(ip_hdr(skb)->daddr,
+ ip_hdr(skb)->saddr,
+ sizeof(rep), sk->sk_protocol, 0);
+
+ ipc.addr = dest->addr;
+ ipc.opt = NULL;
+ ipc.tx_flags = 0;
+
+ {
+ struct flowi fl = {
+ .nl_u = { .ip4_u = {
+ .daddr = ipc.addr,
+ .saddr = ip_hdr(skb)->daddr,
+ .tos = RT_TOS(ip_hdr(skb)->tos)
+ }
+ },
+ .uli_u = { .ports = {
+ .sport = udp_hdr(skb)->dest,
+ .dport = udp_hdr(skb)->source
+ }
+ },
+ .proto = sk->sk_protocol,
+ };
+ security_skb_classify_flow(skb, &fl);
+ if (ip_route_output_key(sock_net(sk), &rt, &fl))
+ return;
+ }
+
+ inet->tos = ip_hdr(skb)->tos;
+ sk->sk_priority = skb->priority;
+ sk->sk_protocol = ip_hdr(skb)->protocol;
+ sk->sk_bound_dev_if = 0;
+ ip_append_data(sk, ip_reply_glue_bits, &rep, sizeof(rep),
+ 0, &ipc, &rt, MSG_DONTWAIT);
+ skb = skb_peek(&sk->sk_write_queue);
+ if (skb) {
+ *((__sum16 *)skb_transport_header(skb) +
+ offsetof(struct udphdr, check) / 2) =
+ csum_fold(csum_add(skb->csum, csum));
+ skb->ip_summed = CHECKSUM_NONE;
+ ip_push_pending_frames(sk);
+ }
+
+ ip_rt_put(rt);
+
+ UDP_INC_STATS_BH(sock_net(sk), UDP_MIB_OUTDATAGRAMS, IS_UDPLITE(sk));
+}
+
+/*
+ * Pass a UDPCP skb buffer to the ip stack and send it
+ */
+static int udpcp_send_skb(struct sock *sk, struct sk_buff *skb,
+ struct udpcp_dest *dest, struct ip_options *opt)
+{
+ int err;
+
+ skb_dst_set(skb, dst_clone(&dest->rt->dst));
+
+ err = ip_build_and_send_pkt(skb, sk, dest->fl.fl4_src,
+ dest->fl.fl4_dst, opt);
+
+ if (!err)
+ UDP_INC_STATS_USER(sock_net(sk), UDP_MIB_OUTDATAGRAMS,
+ IS_UDPLITE(sk));
+ return err;
+}
+
+/*
+ * Release a routing table entry if no packed will be assembled
+ */
+static void udpcp_dst_release(struct udpcp_sock *usk, struct udpcp_dest *dest)
+{
+ if (usk->assembly_dest != dest) {
+ dst_release(&dest->rt->dst);
+ dest->rt = NULL;
+ }
+}
+
+/*
+ * Return true it the passed skb socket buffer is the last in the list
+ */
+static inline bool skb_is_eoq(const struct sk_buff_head *list,
+ const struct sk_buff *skb)
+{
+ return (skb->next == (struct sk_buff *)list);
+}
+
+/*
+ * Arm the timeout handler for the socket
+ */
+static void udpcp_timer(struct sock *sk, unsigned long timeout)
+{
+ struct udpcp_sock *usk = udpcp_sk(sk);
+
+ mod_timer(&usk->timer, timeout);
+}
+
+/*
+ * Decrement the socket pending counter and wakeup a waiting UDPCP_IOCTL_SYNC
+ */
+static inline void udpcp_dec_pending(struct sock *sk)
+{
+ struct udpcp_sock *usk = udpcp_sk(sk);
+
+ if (!--usk->pending) {
+ if (waitqueue_active(&usk->wq))
+ wake_up_interruptible(&usk->wq);
+ }
+}
+
+/*
+ * Returns true is the passed message fragment is the last fragment
+ */
+static inline int udpcp_is_last_frag(struct udpcphdr *uh)
+{
+ return uh->fragamount == uh->fragnum + 1;
+}
+
+/*
+ * Transmit data message fragments
+ */
+static int _udpcp_xmit(struct sock *sk, struct udpcp_dest *dest)
+{
+ struct udpcp_sock *usk = udpcp_sk(sk);
+ struct sk_buff *skb = NULL;
+ struct sk_buff *skbc;
+ struct udpcphdr *uh;
+ int err = 0;
+
+ if (dest->acks >= usk->acks)
+ goto out;
+
+ if (!dest->xmit_last) {
+ /*
+ * handle data message fragments without an ack
+ */
+ while ((skb = skb_peek(&dest->xmit))) {
+ uh = udpcp_hdr(skb);
+
+ if (!(ntohs(uh->msginfo) & UDPCP_NO_ACK_FLAG))
+ break;
+ if (udpcp_is_last_frag(uh)) {
+ usk->stat.txMsgs++;
+ atomic_inc(&udpcp_txMsgs);
+ }
+ skb_unlink(skb, &dest->xmit);
+ udpcp_dec_pending(sk);
+ if (unlikely(debug))
+ dump_msg("send msg", skb, dest->fl.fl4_src,
+ dest->fl.fl4_dst);
+ err = udpcp_send_skb(sk, skb, dest,
+ (struct ip_options *)skb->cb);
+ if (err) {
+ kfree_skb(skb);
+ skb = NULL;
+ break;
+ }
+ }
+ dest->xmit_wait = skb;
+ } else {
+ /*
+ * handle next data message fragment waiting for an ack
+ */
+ uh = udpcp_hdr(dest->xmit_last);
+
+ if (udpcp_is_last_frag(uh))
+ goto out;
+
+ /*
+ * get next data message fragment
+ */
+ skb = dest->xmit_last->next;
+ }
+
+ /*
+ * send all data message fragment till the first which must be acked
+ */
+ while (skb) {
+ skbc = skb_clone(skb, sk->sk_allocation);
+
+ if (!skbc)
+ break;
+
+ if (unlikely(debug))
+ dump_msg("send msg", skbc, dest->fl.fl4_src,
+ dest->fl.fl4_dst);
+ err = udpcp_send_skb(sk, skbc, dest,
+ (struct ip_options *)skb->cb);
+ if (err) {
+ kfree_skb(skbc);
+ break;
+ }
+
+ uh = udpcp_hdr(skb);
+
+ if (!(ntohs(uh->msginfo) & UDPCP_SINGLE_ACK_FLAG)
+ || udpcp_is_last_frag(uh)) {
+ dest->xmit_last = skb;
+
+ if (++dest->acks >= usk->acks || udpcp_is_last_frag(uh))
+ break;
+ }
+
+ skb = skb_is_eoq(&dest->xmit, skb) ? NULL : skb->next;
+ }
+
+out:
+ if (skb_queue_empty(&dest->xmit))
+ udpcp_dst_release(usk, dest);
+
+ return err;
+}
+
+/*
+ * Transmit data message fragments and rearm the timeout handler if necessary
+ */
+static int udpcp_xmit(struct sock *sk, struct udpcp_dest *dest)
+{
+ struct udpcp_sock *usk = udpcp_sk(sk);
+ int ret;
+
+ ret = _udpcp_xmit(sk, dest);
+
+ if (dest->xmit_wait) {
+ dest->tx_time = jiffies;
+
+ if (!timer_pending(&usk->timer))
+ udpcp_timer(sk, dest->tx_time + usk->tx_timeout);
+ }
+ return ret;
+}
+
+/*
+ * Queue the assembled message fragment into the transmit queue
+ */
+static void udpcp_queue_xmit(struct sock *sk, struct udpcp_dest *dest,
+ u8 ackmode, u8 chkmode)
+{
+ struct udpcp_sock *usk = udpcp_sk(sk);
+ struct udpcphdr *uh;
+ struct sk_buff *skb;
+ u8 fragamount;
+ u8 fragnum;
+ unsigned short msginfo;
+ struct flowi *fl = &dest->fl;
+
+ msginfo = UDPCP_MSG_TYPE_DATA | UDPCP_PROTOCOL_VERSION_2;
+ switch (ackmode) {
+ case UDPCP_NOACK:
+ msginfo |= UDPCP_NO_ACK_FLAG;
+ break;
+ case UDPCP_SINGLE_ACK:
+ msginfo |= UDPCP_SINGLE_ACK_FLAG;
+ break;
+ case UDPCP_ACK:
+ default:
+ break;
+ }
+ switch (chkmode) {
+ case UDPCP_NOCHECKSUM:
+ break;
+ case UDPCP_CHECKSUM:
+ default:
+ msginfo |= UDPCP_CHECKSUM_FLAG;
+ break;
+ }
+
+ fragamount = skb_queue_len(&usk->assembly);
+
+ udpcp_sk(sk)->pending += fragamount;
+
+ for (fragnum = 0; fragnum != fragamount; fragnum++) {
+ unsigned char *data;
+ int data_len;
+
+ skb = skb_dequeue(&usk->assembly);
+ uh = udpcp_hdr(skb);
+
+ /*
+ * setup a UDPCP header
+ */
+ uh->chksum = 0;
+ uh->msginfo = htons(msginfo);
+ uh->fragnum = fragnum;
+ uh->fragamount = fragamount;
+ uh->msgid = htons(dest->msgid);
+ uh->length = htons(usk->assembly_len);
+
+ data = skb_transport_header(skb) + sizeof(struct udphdr);
+ data_len = skb_tail_pointer(skb) - data;
+
+ if (chkmode == UDPCP_CHECKSUM)
+ uh->chksum = htonl(zlib_adler32(1, data, data_len));
+ /*
+ * create a UDP header
+ */
+ uh->udphdr.source = fl->fl_ip_sport;
+ uh->udphdr.dest = fl->fl_ip_dport;
+ uh->udphdr.len = htons(sizeof(struct udphdr) + data_len);
+ uh->udphdr.check = 0;
+
+ /*
+ * create UDP checksum
+ */
+ udpcp_do_csum(sk, skb, dest);
+
+ /*
+ * add to xmit queue
+ */
+ skb_queue_tail(&dest->xmit, skb);
+ }
+
+ dest->msgid++;
+ usk->assembly_len = 0;
+ usk->assembly_dest = NULL;
+}
+
+/*
+ * Remove all data message fragments of the first message from the transmit
+ * queue all fragments will be merged together
+ */
+static struct sk_buff *udpcp_dequeue_msg(struct sock *sk,
+ struct udpcp_dest *dest)
+{
+ struct sk_buff *msg;
+ struct sk_buff *skb;
+ struct sk_buff **next;
+ struct udpcphdr *uh;
+
+ msg = skb_dequeue(&dest->xmit);
+ if (!msg)
+ return NULL;
+ skb_orphan(msg);
+
+ uh = udpcp_hdr(msg);
+ if (!uh->msgid) {
+ /*
+ * sync message
+ */
+ kfree_skb(msg);
+ return NULL;
+ }
+
+ skb_pull(msg, sizeof(struct udpcphdr));
+ if (udpcp_is_last_frag(uh))
+ return msg;
+
+ next = &skb_shinfo(msg)->frag_list;
+ for (;;) {
+ skb = skb_dequeue(&dest->xmit);
+ if (!skb)
+ break;
+ skb_orphan(skb);
+ uh = udpcp_hdr(skb);
+ skb_pull(msg, sizeof(struct udpcphdr));
+ msg->len += skb->len;
+ msg->data_len += skb->len;
+ *next = skb;
+ if (udpcp_is_last_frag(uh))
+ break;
+ next = &skb->next;
+ }
+ return msg;
+}
+
+static void udpcp_flush_err(struct sock *sk, struct udpcp_dest *dest)
+{
+ struct inet_sock *inet = inet_sk(sk);
+ struct udpcp_sock *usk = udpcp_sk(sk);
+
+ if (!inet->recverr)
+ skb_queue_purge(&dest->xmit);
+ else {
+ struct sock_exterr_skb *serr;
+ struct iphdr *iph;
+ struct sk_buff *skb;
+
+ while (!skb_queue_empty(&dest->xmit)) {
+ skb = udpcp_dequeue_msg(sk, dest);
+ if (!skb)
+ continue;
+
+ if (unlikely(debug))
+ dump_msg("flush outgoing message", skb,
+ dest->fl.fl4_src, dest->fl.fl4_dst);
+
+ skb_push(skb, sizeof(struct iphdr));
+ skb_reset_network_header(skb);
+ iph = ip_hdr(skb);
+ iph->daddr = dest->rt->rt_dst;
+
+ serr = SKB_EXT_ERR(skb);
+ serr->ee.ee_errno = EPROTO;
+ serr->ee.ee_origin = SO_EE_ORIGIN_LOCAL;
+ serr->ee.ee_type = 0;
+ serr->ee.ee_code = 0;
+ serr->ee.ee_pad = 0;
+ serr->ee.ee_info = 0;
+ serr->ee.ee_data = 0;
+ serr->addr_offset = (u8 *) &iph->daddr -
+ skb_network_header(skb);
+ serr->port = dest->fl.fl_ip_dport;
+
+ skb_reset_transport_header(skb);
+ skb_pull(skb, sizeof(struct iphdr));
+
+ /*
+ * set a flag for UDPCP message
+ */
+ UDP_SKB_CB(skb)->udpcp_flag = 1;
+
+ /*
+ * pass the dequeued message to the error queue of the
+ * socket
+ */
+ skb_set_owner_r(skb, sk);
+ skb_queue_tail(&sk->sk_error_queue, skb);
+ if (!sock_flag(sk, SOCK_DEAD)) {
+ if (usk->udp_data_ready)
+ usk->udp_data_ready(sk, skb->len);
+ }
+ }
+ }
+
+ dest->xmit_wait = 0;
+ dest->xmit_last = 0;
+ dest->try = 0;
+ dest->acks = 0;
+
+ usk->pending = 0;
+ if (waitqueue_active(&usk->wq))
+ wake_up_interruptible(&usk->wq);
+}
+
+/*
+ * Purge the current incoming data message
+ */
+static void udpcp_purge_incoming(struct sock *sk, struct udpcp_dest *dest)
+{
+ struct udpcp_sock *usk = udpcp_sk(sk);
+
+ if (dest->recv_last) {
+ u32 fragnum = udpcp_hdr(dest->recv_last)->fragnum + 1;
+
+ dest->rxDiscardedFrags += fragnum;
+ usk->stat.rxDiscardedFrags += fragnum;
+ atomic_add(fragnum, &udpcp_rxDiscardedFrags);
+
+ dest->lastmsg.msgid = 0;
+
+ if (unlikely(debug))
+ dump_msg("purge incoming message", dest->recv_msg,
+ dest->fl.fl4_src, dest->fl.fl4_dst);
+ }
+
+ kfree_skb(dest->recv_msg);
+ dest->recv_msg = 0;
+ dest->recv_last = 0;
+}
+
+/*
+ * Resend all data message fragments to the one which is currently waiting for
+ * an ack
+ */
+static int udpcp_resend(struct sock *sk, struct udpcp_dest *dest)
+{
+ struct sk_buff *skb;
+ struct sk_buff *skbc;
+ struct udpcp_sock *usk = udpcp_sk(sk);
+ int err;
+
+ if (++dest->try >= usk->maxtry) {
+ dest->insync = 0;
+ udpcp_flush_err(sk, dest);
+ udpcp_purge_incoming(sk, dest);
+ udpcp_dst_release(usk, dest);
+ return 0;
+ }
+
+ dest->txRetries++;
+ usk->stat.txRetries++;
+ atomic_inc(&udpcp_txRetries);
+
+ if (!dest->xmit_last)
+ _udpcp_xmit(sk, dest);
+ else {
+ skb = dest->xmit_wait;
+
+ for (;;) {
+ skbc = skb_clone(skb, sk->sk_allocation);
+
+ if (skbc == NULL)
+ break;
+
+ if (unlikely(debug))
+ dump_msg("resend msg", skbc, dest->fl.fl4_src,
+ dest->fl.fl4_dst);
+ err = udpcp_send_skb(sk, skbc, dest,
+ (struct ip_options *)skb->cb);
+ if (err) {
+ kfree_skb(skbc);
+ break;
+ }
+
+ if (skb == dest->xmit_last) {
+ _udpcp_xmit(sk, dest);
+ break;
+ }
+
+ skb = skb->next;
+ }
+ }
+ dest->tx_time = jiffies;
+
+ return 1;
+}
+
+/*
+ * Handle udpcp timeout
+ */
+static void udpcp_handle_timeout(struct sock *sk)
+{
+ struct udpcp_dest *dest;
+ struct list_head *p;
+ struct udpcp_sock *usk = udpcp_sk(sk);
+ int wflag = 0;
+ unsigned long t = jiffies + UDPCP_MAX_WAIT_SEC * HZ + 1;
+
+ usk->timeout = 0;
+
+ /*
+ * walk through all destinations
+ */
+ list_for_each(p, &usk->destlist) {
+ dest = list_to_udpcpdest(p);
+
+ if (dest->xmit_wait) {
+ if (time_is_before_eq_jiffies
+ (dest->tx_time + usk->tx_timeout)) {
+ /*
+ * transmit timeout expired
+ */
+ if (unlikely(debug))
+ dump_msg("send timeout",
+ dest->xmit_wait,
+ dest->fl.fl4_src,
+ dest->fl.fl4_dst);
+ if (udpcp_resend(sk, dest) == 0) {
+ dest->txTimeout++;
+ usk->stat.txTimeout++;
+ atomic_inc(&udpcp_txTimeout);
+ goto check_incoming;
+ }
+ wflag = 1;
+ }
+ if (time_before(dest->tx_time + usk->tx_timeout, t)) {
+ /*
+ * calculate new timeout timer value
+ */
+ t = dest->tx_time + usk->tx_timeout;
+ wflag = 1;
+ }
+ }
+check_incoming:
+ if (dest->recv_msg) {
+ if (time_is_before_eq_jiffies
+ (dest->rx_time + usk->rx_timeout)) {
+ /*
+ * receive timeout occurred
+ */
+ if (unlikely(debug))
+ dump_msg("receive timeout",
+ dest->recv_last,
+ dest->fl.fl4_src,
+ dest->fl.fl4_dst);
+ udpcp_purge_incoming(sk, dest);
+ dest->rxTimeout++;
+ usk->stat.rxTimeout++;
+ atomic_inc(&udpcp_rxTimeout);
+ } else
+ if (time_before(dest->rx_time + usk->rx_timeout, t)) {
+ /*
+ * calculate new timeout timer value
+ */
+ t = dest->rx_time + usk->rx_timeout;
+ wflag = 1;
+ }
+ }
+ }
+ /*
+ * restart timer if necessary
+ */
+ if (wflag)
+ udpcp_timer(sk, t);
+}
+
+/*
+ * Timeout function
+ */
+static void udpcp_timeout(unsigned long data)
+{
+ struct sock *sk = (struct sock *)data;
+ struct udpcp_sock *usk = udpcp_sk(sk);
+
+ bh_lock_sock(sk);
+ if (!sock_owned_by_user(sk))
+ udpcp_handle_timeout(sk);
+ else {
+ /*
+ * bad, cannot handle the timeout because the socket is in use
+ * set flag for unhandled timeout and rearm the timer
+ */
+ usk->timeout = 1;
+ udpcp_timer(sk, jiffies + 1);
+ }
+ bh_unlock_sock(sk);
+}
+
+/*
+ * Handle timeout if an the unhandled timeout flag is set
+ */
+static inline void check_timeout(struct sock *sk)
+{
+ struct udpcp_sock *usk = udpcp_sk(sk);
+
+ while (usk->timeout) {
+ lock_sock(sk);
+ if (usk->timeout)
+ udpcp_handle_timeout(sk);
+ release_sock(sk);
+ }
+}
+
+/*
+ * Release the socket lock and test for unhandled timeouts
+ */
+static inline void udpcp_release_sock(struct sock *sk)
+{
+ release_sock(sk);
+ check_timeout(sk);
+}
+
+/*
+ * Parse sendmsg() control message
+ */
+static int udpcp_cmsg_send(struct msghdr *msg, u8 * ackmode, u8 * chkmode)
+{
+ struct cmsghdr *cmsg;
+
+ for (cmsg = CMSG_FIRSTHDR(msg); cmsg; cmsg = CMSG_NXTHDR(msg, cmsg)) {
+ if (!CMSG_OK(msg, cmsg))
+ return -EINVAL;
+ if (cmsg->cmsg_level != SOL_UDPCP)
+ continue;
+ switch (cmsg->cmsg_type) {
+ case UDPCP_NOACK:
+ case UDPCP_ACK:
+ case UDPCP_SINGLE_ACK:
+ *ackmode = cmsg->cmsg_type;
+ break;
+ case UDPCP_CHECKSUM:
+ case UDPCP_NOCHECKSUM:
+ *chkmode = cmsg->cmsg_type;
+ break;
+ default:
+ return -EINVAL;
+ }
+ }
+ return 0;
+}
+
+/*
+ * Validate a skb buffer
+ */
+static int udpcp_validate_skb(struct sk_buff *skb)
+{
+ if (skb->next) {
+ pr_err("udpcp: unexpected skb_buff->next != NULL\n");
+ BUG();
+ return 1;
+ }
+ if (skb_shinfo(skb)->frag_list) {
+ pr_err("udpcp: unexpected skb_shinfo(skb)->frag_list != NULL\n");
+ BUG();
+ return 1;
+ }
+ return 0;
+}
+
+/*
+ * Split a message into fragments and store it into the assemble queue
+ * mostly stolen from UDP stack
+ */
+static int udpcp_data(struct sock *sk, struct udpcp_dest *dest,
+ int getfrag(void *from, char *to, int offset, int len,
+ int odd, struct sk_buff *skb),
+ struct iovec *from, int length, unsigned int flags)
+{
+ struct udpcp_sock *usk = udpcp_sk(sk);
+ struct inet_sock *inet = inet_sk(sk);
+ struct sk_buff *skb;
+ struct ipcm_cookie *ipc = &dest->ipc;
+ struct ip_options *opt = ipc->opt;
+ int hh_len;
+ int exthdrlen;
+ int mtu;
+ int copy;
+ int err;
+ int offset = 0;
+ unsigned int maxfraglen, fragheaderlen;
+ int csummode = CHECKSUM_NONE;
+ int transhdrlen = sizeof(struct udpcphdr);
+ struct rtable *rt = dest->rt;
+
+ if (opt && sizeof(skb->cb) < optlength(opt)) {
+ err = -EFAULT;
+ goto error;
+ }
+
+ usk->assembly_len += length;
+ usk->assembly_dest = dest;
+
+ if (usk->assembly_len > UDPCP_MAX_MSGSIZE) {
+ ip_local_error(sk, EMSGSIZE, rt->rt_dst, dest->fl.fl_ip_dport,
+ usk->assembly_len);
+ err = -EMSGSIZE;
+ goto error;
+ }
+
+ mtu = (inet->pmtudisc == IP_PMTUDISC_PROBE) ?
+ rt->dst.dev->mtu : dst_mtu(rt->dst.path);
+ sk->sk_sndmsg_page = NULL;
+ sk->sk_sndmsg_off = 0;
+ exthdrlen = rt->dst.header_len;
+ length += exthdrlen;
+ transhdrlen += exthdrlen;
+
+ hh_len = LL_RESERVED_SPACE(rt->dst.dev);
+
+ fragheaderlen = sizeof(struct iphdr) + (opt ? opt->optlen : 0);
+ maxfraglen = ((mtu - fragheaderlen) & ~7) + fragheaderlen;
+
+ if (rt->dst.dev->features & NETIF_F_V4_CSUM && !exthdrlen)
+ csummode = CHECKSUM_PARTIAL;
+
+ skb = skb_peek_tail(&usk->assembly);
+ if (skb) {
+ unsigned int off;
+
+ off = skb->len;
+
+ copy = mtu - skb->len;
+ if (copy > length)
+ copy = length;
+
+ if (copy > 0 &&
+ getfrag(from, skb_put(skb, copy), 0, copy, off, skb) < 0) {
+ __skb_trim(skb, off);
+ err = -EFAULT;
+ goto error;
+ }
+ length -= copy;
+ offset += copy;
+
+ if (!length)
+ return 0;
+ }
+
+ do {
+ char *data;
+ unsigned int datalen;
+ unsigned int fraglen;
+ unsigned int alloclen;
+
+ length += transhdrlen;
+ /*
+ * If remaining data exceeds the mtu,
+ * we know we need more fragment(s).
+ */
+ datalen = length;
+ if (datalen > mtu - fragheaderlen)
+ datalen = maxfraglen - fragheaderlen;
+ fraglen = datalen + fragheaderlen;
+
+ if ((flags & MSG_MORE)
+ && !(rt->dst.dev->features & NETIF_F_SG))
+ alloclen = mtu;
+ else
+ alloclen = fraglen;
+
+ alloclen += rt->dst.trailer_len + hh_len + 15;
+
+ udpcp_release_sock(sk);
+ skb = sock_alloc_send_skb(sk, alloclen,
+ (flags & MSG_DONTWAIT), &err);
+ lock_sock(sk);
+ if (skb == NULL)
+ goto error;
+
+ if (udpcp_validate_skb(skb)) {
+ kfree_skb(skb);
+
+ goto error;
+ }
+
+ /*
+ * Fill in the control structures
+ */
+ skb->ip_summed = csummode;
+ skb->csum = 0;
+ skb_reserve(skb, hh_len);
+
+ /*
+ * Find where to start putting bytes.
+ */
+ data = skb_put(skb, fraglen);
+ skb_set_network_header(skb, exthdrlen);
+ skb->transport_header = (skb->network_header + fragheaderlen);
+ data += fragheaderlen;
+
+ copy = datalen - transhdrlen;
+
+ if (copy > 0 &&
+ getfrag(from, data + transhdrlen, offset, copy, 0, skb) < 0) {
+ err = -EFAULT;
+ kfree_skb(skb);
+ goto error;
+ }
+
+ offset += copy;
+ length -= datalen;
+
+ if (ipc->opt)
+ memcpy(skb->cb, &ipc->opt, optlength(opt));
+
+ skb_pull(skb, fragheaderlen);
+ skb_queue_tail(&usk->assembly, skb);
+ } while (length > 0);
+
+ return 0;
+error:
+ skb_queue_purge(&usk->assembly);
+ usk->assembly_len = 0;
+
+ IP_INC_STATS(sock_net(sk), IPSTATS_MIB_OUTDISCARDS);
+ return err;
+}
+
+/*
+ * This function will be called by send(), sento() and sendmsg()
+ */
+static int udpcp_sendmsg(struct kiocb *iocb, struct sock *sk,
+ struct msghdr *msg, size_t len)
+{
+ struct inet_sock *inet = inet_sk(sk);
+ struct udpcp_sock *usk = udpcp_sk(sk);
+ struct ipcm_cookie *ipc;
+ struct rtable *rt = NULL;
+ int free = 0;
+ int connected = 0;
+ __be32 daddr, faddr, saddr;
+ __be16 dport;
+ u8 tos;
+ int err = 0;
+ int corkreq = usk->udpsock.corkflag || msg->msg_flags & MSG_MORE;
+ int (*getfrag) (void *, char *, int, int, int, struct sk_buff *);
+ struct udpcp_dest *dest;
+
+ if (len > UDPCP_MAX_MSGSIZE)
+ return -EMSGSIZE;
+
+ /*
+ * Check the flags.
+ */
+ if (msg->msg_flags & MSG_OOB)
+ return -EOPNOTSUPP;
+
+ /*
+ * check if socket is binded to a port
+ */
+ if (!(sk->sk_userlocks & SOCK_BINDPORT_LOCK) || !inet->inet_num)
+ return -ENOTCONN;
+
+ /*
+ * Get and verify the address.
+ */
+ if (msg->msg_name) {
+ struct sockaddr_in *usin = (struct sockaddr_in *)msg->msg_name;
+ if (msg->msg_namelen < sizeof(*usin))
+ return -EINVAL;
+ if (usin->sin_family != AF_INET) {
+ if (usin->sin_family != AF_UNSPEC)
+ return -EAFNOSUPPORT;
+ }
+
+ daddr = usin->sin_addr.s_addr;
+ dport = usin->sin_port;
+ } else {
+ if (sk->sk_state != TCP_ESTABLISHED)
+ return -EDESTADDRREQ;
+ daddr = inet->inet_daddr;
+ dport = inet->inet_dport;
+ /* Open fast path for connected socket.
+ Route will not be used, if at least one option is set.
+ */
+ connected = 1;
+ }
+
+ if (dport == 0)
+ return -EINVAL;
+
+ dest = find_dest(sk, daddr, dport);
+
+ if (!(dest->use_flag & TX_NODE)) {
+ dest->use_flag |= TX_NODE;
+ usk->stat.txNodes++;
+ atomic_inc(&udpcp_txNodes);
+ }
+
+ ipc = &dest->ipc;
+
+ getfrag = IS_UDPLITE(sk) ? udplite_getfrag : ip_generic_getfrag;
+
+ if (!skb_queue_empty(&usk->assembly)) {
+ /*
+ * assembly is ongoing
+ */
+ lock_sock(sk);
+ if (likely(!skb_queue_empty(&usk->assembly))) {
+ if (usk->assembly_dest != dest) {
+ udpcp_release_sock(sk);
+ return -EUSERS;
+ }
+ ipc->opt =
+ (struct ip_options *)skb_peek(&usk->assembly)->cb;
+ goto queue_data;
+ }
+ udpcp_release_sock(sk);
+ }
+
+ ipc->addr = inet->inet_saddr;
+ ipc->oif = sk->sk_bound_dev_if;
+
+ dest->ackmode = usk->ackmode;
+ dest->chkmode = usk->chkmode;
+
+ if (msg->msg_controllen) {
+ /*
+ * handle control message
+ */
+ err = udpcp_cmsg_send(msg, &dest->ackmode, &dest->chkmode);
+ if (err)
+ return err;
+ err = ip_cmsg_send(sock_net(sk), msg, ipc);
+ if (err)
+ return err;
+ if (ipc->opt)
+ free = 1;
+ connected = 0;
+ }
+
+ if (!ipc->opt)
+ ipc->opt = inet->opt;
+
+ saddr = ipc->addr;
+ ipc->addr = faddr = daddr;
+
+ if (ipc->opt && ipc->opt->srr) {
+ if (!daddr)
+ return -EINVAL;
+ faddr = ipc->opt->faddr;
+ connected = 0;
+ }
+ tos = RT_TOS(inet->tos);
+ if (sock_flag(sk, SOCK_LOCALROUTE) ||
+ (msg->msg_flags & MSG_DONTROUTE) ||
+ (ipc->opt && ipc->opt->is_strictroute)) {
+ tos |= RTO_ONLINK;
+ connected = 0;
+ }
+
+ if (ipv4_is_multicast(daddr)) {
+ if (dest->ackmode != UDPCP_NOACK) {
+ err = EOPNOTSUPP;
+ goto out;
+ }
+ if (!ipc->oif)
+ ipc->oif = inet->mc_index;
+ if (!saddr)
+ saddr = inet->mc_addr;
+ connected = 0;
+ }
+
+ lock_sock(sk);
+ rt = dest->rt;
+ if (rt)
+ goto queue_data;
+ udpcp_release_sock(sk);
+
+ /*
+ * calculate routing
+ */
+ if (connected)
+ rt = (struct rtable *)sk_dst_check(sk, 0);
+
+ if (rt == NULL) {
+ struct flowi fl = {.oif = ipc->oif,
+ .nl_u = {.ip4_u = {.daddr = faddr,
+ .saddr = saddr,
+ .tos = tos} },
+ .proto = sk->sk_protocol,
+ .uli_u = {.ports = {.sport = inet->inet_sport,
+ .dport = dport} }
+ };
+ struct net *net = sock_net(sk);
+
+ security_sk_classify_flow(sk, &fl);
+ err = ip_route_output_flow(net, &rt, &fl, sk, 1);
+ if (err) {
+ if (err == -ENETUNREACH)
+ IP_INC_STATS_BH(net, IPSTATS_MIB_OUTNOROUTES);
+ goto out;
+ }
+
+ err = -EACCES;
+ if ((rt->rt_flags & RTCF_BROADCAST) &&
+ !sock_flag(sk, SOCK_BROADCAST))
+ goto out;
+ if (connected)
+ sk_dst_set(sk, dst_clone(&rt->dst));
+ }
+
+ if (msg->msg_flags & MSG_CONFIRM)
+ goto do_confirm;
+back_from_confirm:
+
+ saddr = rt->rt_src;
+ if (!ipc->addr)
+ daddr = ipc->addr = rt->rt_dst;
+
+ lock_sock(sk);
+
+ dest->fl.fl4_dst = daddr;
+ dest->fl.fl_ip_dport = dport;
+ dest->fl.fl4_src = saddr;
+ dest->fl.fl_ip_sport = inet->inet_sport;
+ dest->rt = rt;
+
+queue_data:
+ if (msg->msg_flags & MSG_PROBE)
+ goto release;
+
+ if (!dest->insync && skb_queue_empty(&dest->xmit)) {
+ /*
+ * if not synced, queue a SYNC message
+ */
+ err = udpcp_data(sk, dest, getfrag, NULL, 0, 0);
+ if (err)
+ goto release;
+ dest->msgid = 0;
+ udpcp_queue_xmit(sk, dest, UDPCP_ACK, UDPCP_CHECKSUM);
+ }
+
+ /*
+ * split message and store it to the assembly queue
+ */
+ err = udpcp_data(sk, dest, getfrag, msg->msg_iov, len,
+ corkreq ? msg->msg_flags | MSG_MORE : msg->msg_flags);
+ if (err)
+ goto release;
+
+ if (!dest->msgid)
+ dest->msgid = 1;
+
+ if (!corkreq) {
+ /*
+ * message is complete, transfer it from the assembly queue
+ * into the transmit queue
+ */
+ udpcp_queue_xmit(sk, dest, dest->ackmode, dest->chkmode);
+ /*
+ * start transmit if possible
+ */
+ err = udpcp_xmit(sk, dest);
+ }
+release:
+ udpcp_release_sock(sk);
+out:
+ if (free)
+ kfree(ipc->opt);
+
+ if (!err)
+ return len;
+ /*
+ * ENOBUFS = no kernel mem, SOCK_NOSPACE = no sndbuf space. Reporting
+ * ENOBUFS might not be good (it's not tunable per se), but otherwise
+ * we don't have a good statistic (IpOutDiscards but it can be too many
+ * things). We could add another new stat but at least for now that
+ * seems like overkill.
+ */
+ if (err == -ENOBUFS || test_bit(SOCK_NOSPACE, &sk->sk_socket->flags)) {
+ UDP_INC_STATS_USER(sock_net(sk),
+ UDP_MIB_SNDBUFERRORS, IS_UDPLITE(sk));
+ }
+ return err;
+
+do_confirm:
+ dst_confirm(&rt->dst);
+ if (!(msg->msg_flags & MSG_PROBE) || len)
+ goto back_from_confirm;
+
+ err = 0;
+ goto out;
+}
+
+/*
+ * Sendpage() is not really implemented
+ */
+static int udpcp_sendpage(struct sock *sk, struct page *page, int offset,
+ size_t size, int flags)
+{
+ return sock_no_sendpage(sk->sk_socket, page, offset, size, flags);
+}
+
+/*
+ * Release all message fragments of the first in the transmit queue
+ */
+static void udpcp_release_xmit(struct sock *sk, struct udpcp_dest *dest)
+{
+ struct udpcp_sock *usk = udpcp_sk(sk);
+ struct sk_buff *skb;
+ struct udpcphdr *uh;
+
+ for (;;) {
+ skb = skb_dequeue(&dest->xmit);
+
+ uh = udpcp_hdr(skb);
+
+ if (udpcp_is_last_frag(uh) && uh->msgid) {
+ usk->stat.txMsgs++;
+ atomic_inc(&udpcp_txMsgs);
+ }
+
+ udpcp_dec_pending(sk);
+
+ kfree_skb(skb);
+ if (skb == dest->xmit_last)
+ break;
+ }
+
+ dest->xmit_wait = 0;
+ dest->xmit_last = 0;
+ dest->try = 0;
+}
+
+/*
+ * Set the sync state
+ */
+static void udpcp_sync(struct sock *sk, struct udpcp_dest *dest)
+{
+ dest->xmit_wait = 0;
+ dest->xmit_last = 0;
+ dest->try = 0;
+ dest->acks = 0;
+ dest->insync = 1;
+}
+
+/*
+ * Returns true if the first message in the transmit queue is a sync message
+ */
+static inline int udpcp_xmit_is_sync(struct udpcp_dest *dest)
+{
+ struct sk_buff *skb = skb_peek(&dest->xmit);
+
+ return skb && !udpcp_hdr(skb)->msgid;
+}
+
+static inline struct udpcphdr *udpcp_ack_scan(struct sk_buff *skb)
+{
+ struct udpcphdr *uh;
+
+ for (;;) {
+ uh = udpcp_hdr(skb);
+
+ if (!(ntohs(uh->msginfo) & UDPCP_SINGLE_ACK_FLAG)
+ || udpcp_is_last_frag(uh))
+ return uh;
+
+ skb = skb->next;
+ }
+}
+
+/*
+ * Handle an incoming ack
+ */
+static void udpcp_handle_ack(struct sock *sk, struct sk_buff *skb,
+ struct udpcp_dest *dest)
+{
+ struct udpcphdr *r_uh;
+ struct udpcphdr *q_uh;
+
+ if (!dest->acks)
+ return;
+
+ r_uh = udpcp_hdr(skb);
+
+ /*
+ * acks doesn't have a payload
+ */
+ if (r_uh->length)
+ return;
+
+ q_uh = udpcp_ack_scan(dest->xmit_wait);
+
+ /*
+ * message id, fragnum and fragamount must match the awaited message
+ * fragment
+ */
+ if (r_uh->msgid != q_uh->msgid)
+ return;
+
+ if (r_uh->fragnum != q_uh->fragnum)
+ return;
+
+ if (r_uh->fragamount != q_uh->fragamount)
+ return;
+
+ dest->acks--;
+
+ /*
+ * if last fragment release message
+ */
+ if (udpcp_is_last_frag(q_uh)) {
+ udpcp_release_xmit(sk, dest);
+
+ /*
+ * special handling for sync messages
+ */
+ if (r_uh->msgid == 0)
+ udpcp_sync(sk, dest);
+ } else
+ dest->xmit_wait = dest->xmit_wait->next;
+
+ /*
+ * try to transmit next message/fragment
+ */
+ udpcp_xmit(sk, dest);
+}
+
+/*
+ * Queue incoming message as owned by udpcp socket
+ */
+static void udpcp_set_owner_r(struct sock *sk, struct udpcp_dest *dest)
+{
+ struct sk_buff *skb;
+
+ skb = dest->recv_msg;
+ skb_set_owner_r(skb, sk);
+
+ skb = skb_shinfo(skb)->frag_list;
+ if (!skb)
+ return;
+
+ for (;;) {
+ skb_set_owner_r(skb, sk);
+ if (udpcp_is_last_frag(udpcp_hdr(skb)))
+ break;
+ skb = skb->next;
+ }
+}
+
+/*
+ * Handle an incoming data message fragment
+ */
+static int udpcp_handle_data(struct sock *sk, struct sk_buff *skb,
+ struct udpcp_dest *dest)
+{
+ struct udpcp_sock *usk = udpcp_sk(sk);
+ struct udpcphdr *uh = udpcp_hdr(skb);
+ unsigned short msginfo = ntohs(uh->msginfo);
+ unsigned short length = ntohs(uh->length);
+
+ /*
+ * special handling for sync messages
+ */
+ if (uh->msgid == 0) {
+ /*
+ * sync messages doesn't have a payload
+ */
+ if (length)
+ return 1;
+
+ /*
+ * sync messages doesn't have a ack rules
+ */
+ if (msginfo & (UDPCP_NO_ACK_FLAG | UDPCP_SINGLE_ACK_FLAG))
+ return 1;
+
+ udpcp_send_ack(sk, skb, dest,
+ memcmp(uh, &dest->lastmsg,
+ sizeof(dest->lastmsg)) ? 0 : 1);
+
+ udpcp_purge_incoming(sk, dest);
+
+ /*
+ * skip the first message in the queue if it is a sync messages
+ */
+ if (udpcp_xmit_is_sync(dest)) {
+ dest->acks--;
+ udpcp_dec_pending(sk);
+ kfree_skb(skb_dequeue(&dest->xmit));
+ }
+
+ if (!dest->insync)
+ udpcp_sync(sk, dest);
+
+ udpcp_xmit(sk, dest);
+
+ return -1;
+ }
+
+ if (!dest->insync)
+ return 1;
+
+ if (length > UDPCP_MAX_MSGSIZE)
+ return 1;
+
+ length += sizeof(struct udpcphdr);
+
+ /*
+ * if the message was still handled, send a duplicate ack
+ */
+ if (!memcmp(uh, &dest->lastmsg, sizeof(dest->lastmsg))) {
+ udpcp_send_ack(sk, skb, dest, 1);
+ return 1;
+ }
+
+ if (dest->recv_msg) {
+ /*
+ * if a fragment is already received validate the fragment
+ */
+ if ((uh->msgid != udpcp_hdr(dest->recv_msg)->msgid) ||
+ (uh->msginfo != udpcp_hdr(dest->recv_msg)->msginfo) ||
+ (uh->length != udpcp_hdr(dest->recv_msg)->length) ||
+ (uh->fragamount != udpcp_hdr(dest->recv_msg)->fragamount)
+ ) {
+ udpcp_purge_incoming(sk, dest);
+ goto newmsg;
+ }
+
+ if (uh->fragnum != udpcp_hdr(dest->recv_last)->fragnum + 1)
+ return 1;
+
+ if (dest->recv_msg->len + skb->len - sizeof(struct udpcphdr) >
+ length)
+ return 1;
+ } else {
+newmsg:
+ /*
+ * first fragment must have the number 0
+ */
+ if (uh->fragnum != 0)
+ return 1;
+
+ /*
+ * UDPCP data length cannot be smaller then the UDP data length
+ */
+ if (skb->len > length)
+ return 1;
+
+ /*
+ * id of the last received is not valid
+ */
+ if (dest->lastmsg.msgid == uh->msgid)
+ return 1;
+
+ /*
+ * check against receive buffer limit
+ */
+ if (atomic_read(&sk->sk_rmem_alloc) + length > sk->sk_rcvbuf)
+ return 1;
+ }
+
+ memset(&dest->lastmsg, 0, sizeof(dest->lastmsg));
+
+ if (!dest->recv_msg) {
+ /*
+ * store the first message fragment
+ */
+ if (skb->cloned) {
+ struct sk_buff *skbc;
+
+ skbc = skb_copy(skb, sk->sk_allocation);
+ if (skbc == NULL)
+ return 1;
+ kfree_skb(skb);
+ skb = skbc;
+ }
+ dest->recv_msg = skb;
+ } else {
+ /*
+ * store the consecutively message fragment
+ */
+ struct skb_shared_info *shinfo;
+
+ shinfo = skb_shinfo(dest->recv_msg);
+
+ if (!shinfo->frag_list)
+ shinfo->frag_list = skb;
+ else
+ dest->recv_last->next = skb;
+
+ skb_pull(skb, sizeof(struct udpcphdr));
+ dest->recv_msg->len += skb->len;
+ dest->recv_msg->data_len += skb->len;
+ }
+ dest->recv_last = skb;
+
+ msginfo = ntohs(uh->msginfo);
+
+ if (udpcp_is_last_frag(uh) || uh->fragamount == 0) {
+ /*
+ * last fragment: queue it to the socket sk_receive_queue
+ * and ack it
+ */
+
+ if (dest->recv_msg->len != length) {
+ udpcp_purge_incoming(sk, dest);
+ return 0;
+ }
+
+ if (!(msginfo & UDPCP_NO_ACK_FLAG))
+ udpcp_send_ack(sk, skb, dest, 0);
+
+ memcpy(dest->recv_msg->data + UDPCP_HDRSIZE,
+ dest->recv_msg->data, sizeof(struct udphdr));
+ skb_pull(dest->recv_msg, UDPCP_HDRSIZE);
+
+ usk->stat.rxMsgs++;
+ atomic_inc(&udpcp_rxMsgs);
+
+ /*
+ * set a flag for UDPCP message
+ */
+ UDP_SKB_CB(skb)->udpcp_flag = 1;
+
+ udpcp_set_owner_r(sk, dest);
+ skb_queue_tail(&sk->sk_receive_queue, dest->recv_msg);
+
+ /*
+ * call the original data available handler
+ */
+ if (usk->udp_data_ready)
+ usk->udp_data_ready(sk, dest->recv_msg->len);
+
+ dest->recv_msg = 0;
+ dest->recv_last = 0;
+ } else {
+ /*
+ * ack fragment if requiered
+ */
+ if (!(msginfo & UDPCP_NO_ACK_FLAG)
+ && !(msginfo & UDPCP_SINGLE_ACK_FLAG))
+ udpcp_send_ack(sk, skb, dest, 0);
+
+ /*
+ * setup timeout handler
+ */
+ dest->rx_time = jiffies;
+
+ if (!timer_pending(&usk->timer))
+ udpcp_timer(sk, dest->rx_time + usk->rx_timeout);
+ }
+
+ return 0;
+}
+
+/*
+ * Deal with received UDPCP frames - sort out what type source it is
+ * and hand of it to the udpcp_handle_packet function.
+ */
+static void udpcp_data_ready(struct sock *sk, int slen)
+{
+ struct udpcp_sock *usk = udpcp_sk(sk);
+ struct sk_buff *skb;
+ struct udpcp_dest *dest;
+ struct udpcphdr *uh;
+ unsigned short msginfo;
+ int ret;
+
+ skb = skb_peek_tail(&sk->sk_receive_queue);
+
+ /*
+ * don't handle NULL pointer buffer and UDPCP messages
+ */
+ if (skb == NULL || UDP_SKB_CB(skb)->udpcp_flag) {
+ if (usk->udp_data_ready)
+ usk->udp_data_ready(sk, slen);
+ return;
+ }
+
+ __skb_unlink(skb, &sk->sk_receive_queue);
+ if (udpcp_validate_skb(skb)) {
+ kfree_skb(skb);
+
+ return;
+ }
+
+ skb_orphan(skb);
+
+ /*
+ * do UDP checksum
+ */
+ if (udp_lib_checksum_complete(skb)) {
+ UDP_INC_STATS_BH(sock_net(sk), UDP_MIB_INERRORS,
+ IS_UDPLITE(sk));
+ return;
+ }
+
+ if (unlikely(debug))
+ dump_msg("receive", skb, ip_hdr(skb)->saddr,
+ ip_hdr(skb)->daddr);
+
+ uh = udpcp_hdr(skb);
+ msginfo = ntohs(uh->msginfo);
+
+ /*
+ * handle only UDPCP protocol version 2
+ */
+ if ((msginfo & UDPCP_PROTOCOL_MASK) != UDPCP_PROTOCOL_VERSION_2) {
+ kfree_skb(skb);
+ return;
+ }
+
+ /*
+ * handle UDPCP checksum
+ */
+ if (msginfo & UDPCP_CHECKSUM_FLAG) {
+ u8 *data;
+ u32 data_len;
+ u32 chksum;
+
+ chksum = ntohl(uh->chksum);
+ data = (u8 *) skb->data + sizeof(struct udphdr);
+ data_len = skb->len - sizeof(struct udphdr);
+
+ uh->chksum = 0;
+
+ if (chksum != zlib_adler32(1, data, data_len)) {
+ kfree_skb(skb);
+ usk->stat.crcErrors++;
+ atomic_inc(&udpcp_crcErrors);
+ return;
+ }
+ }
+
+ dest = __find_dest(sk, ip_hdr(skb)->saddr, udp_hdr(skb)->source);
+
+ if (!dest) {
+ /*
+ * new communication destination must start with an sync message
+ */
+ if (((msginfo & UDPCP_MSG_TYPE_MASK) != UDPCP_MSG_TYPE_DATA) ||
+ (uh->msgid != 0)) {
+ kfree_skb(skb);
+ return;
+ }
+
+ dest = new_dest(sk, ip_hdr(skb)->saddr, udp_hdr(skb)->source);
+
+ if (!dest) {
+ kfree_skb(skb);
+ return;
+ }
+ }
+
+ /*
+ * handle message type
+ */
+ switch (msginfo & UDPCP_MSG_TYPE_MASK) {
+ case UDPCP_MSG_TYPE_DATA:
+ if (!(dest->use_flag & RX_NODE)) {
+ dest->use_flag |= RX_NODE;
+ usk->stat.rxNodes++;
+ atomic_inc(&udpcp_rxNodes);
+ }
+
+ ret = udpcp_handle_data(sk, skb, dest);
+
+ if (ret > 0) {
+ dest->rxDiscardedFrags++;
+ usk->stat.rxDiscardedFrags++;
+ atomic_inc(&udpcp_rxDiscardedFrags);
+ }
+ break;
+ case UDPCP_MSG_TYPE_ACK:
+ udpcp_handle_ack(sk, skb, dest);
+ default:
+ ret = 1;
+ break;
+ }
+ if (ret)
+ kfree_skb(skb);
+}
+
+/*
+ * Set socket options
+ */
+static int udpcp_setsockopt(struct sock *sk, int level, int optname,
+ char __user *optval, unsigned int optlen)
+{
+ struct udpcp_sock *usk = udpcp_sk(sk);
+ int val, ret;
+
+ if (level != SOL_UDPCP) {
+ if (udp_prot.setsockopt) {
+ ret = udp_prot.setsockopt(sk, level, optname, optval,
+ optlen);
+ check_timeout(sk);
+ return ret;
+ }
+ return -ENOPROTOOPT;
+ }
+
+ if (optlen < sizeof(int))
+ return -EINVAL;
+
+ if (get_user(val, (int __user *)optval))
+ return -EFAULT;
+
+ switch (optname) {
+ case UDPCP_OPT_TRANSFER_MODE:
+ switch (val) {
+ case UDPCP_NOACK:
+ case UDPCP_ACK:
+ case UDPCP_SINGLE_ACK:
+ usk->ackmode = val;
+ break;
+ default:
+ return -EINVAL;
+ }
+ break;
+ case UDPCP_OPT_CHECKSUM_MODE:
+ switch (val) {
+ case UDPCP_NOCHECKSUM:
+ case UDPCP_CHECKSUM:
+ usk->chkmode = val;
+ break;
+ default:
+ return -EINVAL;
+ }
+ break;
+
+ case UDPCP_OPT_TX_TIMEOUT:
+ if ((val < 1) || (val > UDPCP_MAX_WAIT_SEC * 1000))
+ return -EINVAL;
+ usk->tx_timeout = msecs_to_jiffies(val);
+ break;
+
+ case UDPCP_OPT_RX_TIMEOUT:
+ if ((val < 1) || (val > UDPCP_MAX_WAIT_SEC * 1000))
+ return -EINVAL;
+ usk->rx_timeout = msecs_to_jiffies(val);
+ break;
+
+ case UDPCP_OPT_MAXTRY:
+ if ((val < 1) || (val > 10))
+ return -EINVAL;
+ usk->maxtry = val;
+ break;
+
+ case UDPCP_OPT_OUTSTANDING_ACKS:
+ if ((val < 1) || (val > 255))
+ return -EINVAL;
+ usk->acks = val;
+ break;
+
+ default:
+ return -ENOPROTOOPT;
+ }
+ return 0;
+}
+
+/*
+ * Get socket options
+ */
+static int udpcp_getsockopt(struct sock *sk, int level, int optname,
+ char __user *optval, int __user *optlen)
+{
+ struct udpcp_sock *usk = udpcp_sk(sk);
+ int val, len, ret;
+
+ if (level != SOL_UDPCP) {
+ if (udp_prot.getsockopt) {
+ ret = udp_prot.getsockopt(sk, level, optname, optval,
+ optlen);
+ check_timeout(sk);
+ return ret;
+ }
+ return -ENOPROTOOPT;
+ }
+
+ if (get_user(len, optlen))
+ return -EFAULT;
+
+ len = min_t(unsigned int, len, sizeof(int));
+
+ if (len < 0)
+ return -EINVAL;
+
+ switch (optname) {
+ case UDPCP_OPT_TRANSFER_MODE:
+ val = usk->ackmode;
+ break;
+
+ case UDPCP_OPT_CHECKSUM_MODE:
+ val = usk->chkmode;
+ break;
+
+ case UDPCP_OPT_TX_TIMEOUT:
+ val = jiffies_to_msecs(usk->tx_timeout);
+ break;
+
+ case UDPCP_OPT_MAXTRY:
+ val = usk->maxtry;
+ break;
+
+ case UDPCP_OPT_OUTSTANDING_ACKS:
+ val = usk->acks;
+ break;
+
+ default:
+ return -ENOPROTOOPT;
+ }
+
+ if (put_user(len, optlen))
+ return -EFAULT;
+ if (copy_to_user(optval, &val, len))
+ return -EFAULT;
+ return 0;
+}
+
+/*
+ * ioctl() requests applicable to the UDPCP protocol
+ */
+int udpcp_ioctl(struct sock *sk, int cmd, unsigned long arg)
+{
+ struct udpcp_sock *usk = udpcp_sk(sk);
+ int ret = 0;
+
+ switch (cmd) {
+ case UDPCP_IOCTL_GET_STATISTICS:
+ lock_sock(sk);
+ if (copy_to_user((void *)arg, &usk->stat, sizeof(usk->stat)))
+ ret = -EFAULT;
+ udpcp_release_sock(sk);
+ break;
+
+ case UDPCP_IOCTL_RESET_STATISTICS:
+ lock_sock(sk);
+ usk->stat.txMsgs = 0;
+ usk->stat.rxMsgs = 0;
+ usk->stat.txTimeout = 0;
+ usk->stat.rxTimeout = 0;
+ usk->stat.txRetries = 0;
+ usk->stat.rxDiscardedFrags = 0;
+ usk->stat.crcErrors = 0;
+ udpcp_release_sock(sk);
+ break;
+
+ case UDPCP_IOCTL_SYNC:
+ if (arg)
+ ret = wait_event_interruptible_timeout(usk->wq,
+ !usk->pending, msecs_to_jiffies(arg));
+ else
+ ret = wait_event_interruptible(usk->wq, !usk->pending);
+
+ break;
+
+ default:
+ if (udp_prot.ioctl) {
+ ret = udp_prot.ioctl(sk, cmd, arg);
+ check_timeout(sk);
+ } else
+ ret = -ENOIOCTLCMD;
+ break;
+ }
+ return ret;
+}
+
+/*
+ * This function will be called by recv(), recvfrom() and revmsg()
+ */
+int udpcp_recvmsg(struct kiocb *iocb, struct sock *sk, struct msghdr *msg,
+ size_t len, int noblock, int flags, int *addr_len)
+{
+ int ret;
+
+ ret = udp_prot.recvmsg(iocb, sk, msg, len, noblock, flags, addr_len);
+ check_timeout(sk);
+ return ret;
+}
+
+/*
+ * This function will be called by socket() and initialized the socket
+ */
+static int udpcp_sockinit(struct sock *sk)
+{
+ int ret;
+ struct udpcp_sock *usk;
+
+ sk->sk_protocol = SOL_UDP;
+ sk->sk_allocation = GFP_ATOMIC;
+ if (udp_prot.init) {
+ ret = udp_prot.init(sk);
+
+ if (ret)
+ return ret;
+ }
+
+ usk = udpcp_sk(sk);
+ usk->timer.expires = 0;
+ usk->timer.function = udpcp_timeout;
+ usk->timer.data = (long)sk;
+ init_timer(&usk->timer);
+ INIT_LIST_HEAD(&usk->destlist);
+ init_waitqueue_head(&usk->wq);
+ usk->pending = 0;
+ usk->ackmode = UDPCP_ACK;
+ usk->chkmode = UDPCP_CHECKSUM;
+ usk->maxtry = UDPCP_TX_MAXTRY;
+ usk->acks = UDPCP_OUTSTANDING_ACKS;
+ usk->tx_timeout = msecs_to_jiffies(UDPCP_TX_TIMEOUT);
+ usk->rx_timeout = msecs_to_jiffies(UDPCP_RX_TIMEOUT);
+ usk->udp_data_ready = sk->sk_data_ready;
+ sk->sk_data_ready = udpcp_data_ready;
+ usk->udpsock.pending = 0;
+ skb_queue_head_init(&usk->assembly);
+ usk->assembly_len = 0;
+ usk->assembly_dest = NULL;
+
+ spin_lock_irq(&udpcp_lock);
+ list_add_tail(&usk->udpcplist, &udpcp_list);
+ spin_unlock_irq(&udpcp_lock);
+
+#ifdef MODULE
+ try_module_get(THIS_MODULE);
+#endif
+ return 0;
+}
+
+/*
+ * This function will be called by close()
+ */
+static void udpcp_destroy(struct sock *sk)
+{
+ struct list_head *p;
+ struct list_head *n;
+ struct udpcp_sock *usk = udpcp_sk(sk);
+
+ spin_lock_irq(&udpcp_lock);
+ list_del(&usk->udpcplist);
+ spin_unlock_irq(&udpcp_lock);
+
+ if (udp_prot.destroy)
+ udp_prot.destroy(sk);
+
+ lock_sock(sk);
+
+ del_timer_sync(&usk->timer);
+ sk->sk_data_ready = usk->udp_data_ready;
+
+ skb_queue_purge(&usk->assembly);
+
+ list_for_each_safe(p, n, &usk->destlist) {
+ struct udpcp_dest *dest;
+
+ dest = list_to_udpcpdest(p);
+
+ skb_queue_purge(&dest->xmit);
+
+ kfree_skb(dest->recv_msg);
+
+ if (dest->rt)
+ dst_release(&dest->rt->dst);
+
+ kfree(dest);
+ }
+
+ atomic_sub(usk->stat.txNodes, &udpcp_txNodes);
+ atomic_sub(usk->stat.rxNodes, &udpcp_rxNodes);
+
+ usk->pending = 0;
+
+ if (waitqueue_active(&usk->wq))
+ wake_up_interruptible(&usk->wq);
+
+ release_sock(sk);
+
+#ifdef MODULE
+ module_put(THIS_MODULE);
+#endif
+}
+
+static struct proto udpcp_prot;
+
+/*
+ * inet protocol stack descriptor
+ */
+static struct inet_protosw udpcp_protosw = {
+ .type = SOCK_DGRAM,
+ .protocol = PF_UDPCP,
+ .prot = &udpcp_prot,
+ .ops = &inet_dgram_ops,
+ .no_check = UDP_CSUM_DEFAULT,
+ .flags = 0,
+};
+
+#ifdef CONFIG_PROC_FS
+/*
+ * The following functions handles the /proc/net/udpcp entry
+ */
+struct udpcp_seq_afinfo {
+ char *name;
+ const struct file_operations seq_fops;
+ const struct seq_operations seq_ops;
+};
+
+struct udpcp_iter_state {
+ struct seq_net_private p;
+ struct sock *sk;
+ struct list_head *list;
+ int bucket;
+};
+
+static int udpcp_get_destlist(struct udpcp_sock *usk,
+ struct udpcp_iter_state *state)
+{
+ struct sock *sk = (struct sock *)usk;
+
+ if (sock_flag(sk, SOCK_DEAD))
+ return 0;
+
+ sock_hold(sk);
+ if (!list_empty(&usk->destlist)) {
+ state->sk = sk;
+ state->list = &usk->destlist;
+ return 1;
+ }
+ sock_put(sk);
+
+ return 0;
+}
+
+static inline int udpcp_next_dest(struct udpcp_iter_state *state)
+{
+ struct sock *sk = state->sk;
+ struct udpcp_sock *usk = udpcp_sk(sk);
+ int found = 0;
+
+ if (sock_flag(sk, SOCK_DEAD))
+ return 0;
+
+ lock_sock(sk);
+ if (!list_is_last(state->list, &usk->destlist)) {
+ state->list = state->list->next;
+ state->bucket++;
+ found = 1;
+ }
+ udpcp_release_sock(sk);
+ return found;
+}
+
+static void *udpcp_get_next(struct seq_file *seq)
+{
+ struct udpcp_iter_state *state = seq->private;
+ struct udpcp_sock *usk;
+ struct sock *sk;
+
+ while (state) {
+ if (udpcp_next_dest(state))
+ return state;
+
+ sk = state->sk;
+ usk = udpcp_sk(sk);
+
+ spin_lock_irq(&udpcp_lock);
+ while (!list_is_last(&usk->udpcplist, &udpcp_list)) {
+ usk = list_entry(usk->udpcplist.next, struct udpcp_sock,
+ udpcplist);
+
+ if (udpcp_get_destlist(usk, state))
+ goto found;
+ }
+ state->sk = NULL;
+ state = NULL;
+found:
+ spin_unlock_irq(&udpcp_lock);
+ sock_put(sk);
+ }
+ return state;
+}
+
+static void *udpcp_get_first(struct seq_file *seq)
+{
+ struct list_head *p;
+ struct udpcp_iter_state *state = seq->private;
+ int found = 0;
+
+ if (!state)
+ return NULL;
+
+ spin_lock_irq(&udpcp_lock);
+ list_for_each(p, &udpcp_list) {
+ found = udpcp_get_destlist(list_to_udpcpsock(p), state);
+ if (found)
+ goto found;
+ }
+found:
+ spin_unlock_irq(&udpcp_lock);
+
+ if (!found)
+ return NULL;
+ return udpcp_get_next(seq);
+}
+
+static void *udpcp_get_idx(struct seq_file *seq, loff_t pos)
+{
+ if (!udpcp_get_first(seq))
+ return NULL;
+
+ while (pos--) {
+ if (!udpcp_get_next(seq))
+ return NULL;
+ }
+ return seq->private;
+}
+
+static void *udpcp_seq_start(struct seq_file *seq, loff_t * pos)
+{
+ return *pos ? udpcp_get_idx(seq, *pos - 1) : SEQ_START_TOKEN;
+}
+
+static void *udpcp_seq_next(struct seq_file *seq, void *v, loff_t * pos)
+{
+ void *private;
+
+ if (v == SEQ_START_TOKEN)
+ private = udpcp_get_idx(seq, 0);
+ else
+ private = udpcp_get_next(seq);
+
+ ++*pos;
+ return private;
+}
+
+static void udpcp_seq_stop(struct seq_file *seq, void *v)
+{
+ struct udpcp_iter_state *state = seq->private;
+
+ if (state->sk)
+ sock_put(state->sk);
+}
+
+static int udpcp_seq_open(struct inode *inode, struct file *file)
+{
+ struct udpcp_seq_afinfo *afinfo = PDE(inode)->data;
+ int err;
+
+ err = seq_open_net(inode, file, &afinfo->seq_ops,
+ sizeof(struct udpcp_iter_state));
+ if (err < 0)
+ return err;
+
+ return err;
+}
+
+int udpcp_proc_register(struct net *net, struct udpcp_seq_afinfo *afinfo)
+{
+ struct proc_dir_entry *p;
+ int rc = 0;
+
+ p = proc_create_data(afinfo->name, S_IRUGO, net->proc_net,
+ &afinfo->seq_fops, afinfo);
+ if (!p)
+ rc = -ENOMEM;
+ return rc;
+}
+
+void udpcp_proc_unregister(struct net *net, struct udpcp_seq_afinfo *afinfo)
+{
+ proc_net_remove(net, afinfo->name);
+}
+
+static unsigned int udpcp_tx_queue_len(struct sock *sk, struct udpcp_dest *dest)
+{
+ struct sk_buff *skb;
+ unsigned int n;
+
+ n = 0;
+ skb_queue_walk(&dest->xmit, skb)
+ n += skb->len;
+ return n;
+}
+
+static unsigned int udpcp_rx_queue_len(struct sock *sk, struct udpcp_dest *dest)
+{
+ struct sk_buff *skb;
+ unsigned int n;
+
+ n = 0;
+ skb_queue_walk(&sk->sk_receive_queue, skb) {
+ if (udp_hdr(skb)->source == dest->port
+ && ip_hdr(skb)->saddr == dest->addr)
+ n += skb->len;
+ }
+ return n;
+}
+
+static void udpcp_format_sock(struct seq_file *seq, int *len)
+{
+ struct udpcp_iter_state *state = seq->private;
+ struct sock *sk = state->sk;
+ struct inet_sock *inet = inet_sk(sk);
+ struct udpcp_dest *p = list_to_udpcpdest(state->list);
+ __be32 src = inet->inet_rcv_saddr;
+ __u16 srcp = ntohs(inet->inet_sport);
+ __be32 dest = p->addr;
+ __u16 destp = ntohs(p->port);
+
+ lock_sock(sk);
+ seq_printf(seq, "%4d: %08X:%04X %08X:%04X"
+ " %02X %08X:%08X %02X:%08lX %08X %5d %8d %lu %d %p %u%n",
+ state->bucket, src, srcp, dest, destp, sk->sk_state,
+ udpcp_tx_queue_len(sk, p),
+ udpcp_rx_queue_len(sk, p),
+ 0, 0L, p->txRetries, sock_i_uid(sk),
+ p->txTimeout, sock_i_ino(sk),
+ atomic_read(&sk->sk_refcnt), sk, p->rxTimeout,
+ len);
+ udpcp_release_sock(sk);
+}
+
+int udpcp_seq_show(struct seq_file *seq, void *v)
+{
+ if (v == SEQ_START_TOKEN)
+ seq_printf(seq, "%-127s\n",
+ " sl local_address rem_address st tx_queue "
+ "rx_queue tr tm->when retrnsmt uid timeout "
+ "inode ref pointer drops");
+ else {
+ int len;
+
+ udpcp_format_sock(seq, &len);
+ seq_printf(seq, "%*s\n", 127 - len, "");
+ }
+ return 0;
+}
+
+static struct udpcp_seq_afinfo udpcp_seq_afinfo = {
+ .name = "udpcp",
+ .seq_fops = {
+ .owner = THIS_MODULE,
+ .open = udpcp_seq_open,
+ .read = seq_read,
+ .llseek = seq_lseek,
+ .release = seq_release_net,
+ },
+ .seq_ops = {
+ .show = udpcp_seq_show,
+ .start = udpcp_seq_start,
+ .next = udpcp_seq_next,
+ .stop = udpcp_seq_stop,
+ },
+};
+
+static int udpcp_proc_init_net(struct net *net)
+{
+ return udpcp_proc_register(net, &udpcp_seq_afinfo);
+}
+
+static void udpcp_proc_exit_net(struct net *net)
+{
+ udpcp_proc_unregister(net, &udpcp_seq_afinfo);
+}
+
+static struct pernet_operations udpcp_net_ops = {
+ .init = udpcp_proc_init_net,
+ .exit = udpcp_proc_exit_net,
+};
+
+int __init udpcp_proc_init(void)
+{
+ return register_pernet_subsys(&udpcp_net_ops);
+}
+
+void udpcp_proc_exit(void)
+{
+ unregister_pernet_subsys(&udpcp_net_ops);
+}
+#endif /* CONFIG_PROC_FS */
+
+/*
+ * Install and init module
+ */
+static int __init udpcp_init(void)
+{
+ int ret;
+ struct proc_dir_entry *proc_entry = NULL;
+
+ spin_lock_init(&udpcp_lock);
+
+ INIT_LIST_HEAD(&udpcp_list);
+
+ /*
+ * to prevent to rewrite the whole UDP protocol,
+ * assign struct proto udp to the struct proto udpcp
+ */
+ udpcp_prot = udp_prot;
+
+ /*
+ * change the protocol name
+ */
+ strcpy(udpcp_prot.name, "UDPCP");
+
+ /*
+ * overload the following function, all other
+ * functions will use the UDP protocol functions
+ */
+ udpcp_prot.sendmsg = udpcp_sendmsg;
+ udpcp_prot.sendpage = udpcp_sendpage;
+ udpcp_prot.init = udpcp_sockinit;
+ udpcp_prot.destroy = udpcp_destroy;
+ udpcp_prot.setsockopt = udpcp_setsockopt;
+ udpcp_prot.getsockopt = udpcp_getsockopt;
+ udpcp_prot.ioctl = udpcp_ioctl;
+ udpcp_prot.recvmsg = udpcp_recvmsg;
+
+ /*
+ * fix the object size for the embedded udpcp_sock structure
+ */
+ udpcp_prot.obj_size = sizeof(struct udpcp_sock);
+
+ /*
+ * register the UDPCP protocol
+ */
+ ret = proto_register(&udpcp_prot, 1);
+ if (ret)
+ return ret;
+
+ /*
+ * register the inet socket for UDPCP
+ */
+ inet_register_protosw(&udpcp_protosw);
+
+#ifdef CONFIG_PROC_FS
+ /*
+ * register /proc/driver/udpcp entry
+ */
+ proc_entry =
+ create_proc_read_entry(UDPCP_PROC, S_IRUSR | S_IRGRP | S_IROTH,
+ NULL, udpcp_proc, NULL);
+
+ if (!proc_entry) {
+ ret = -ENOMEM;
+ goto err;
+ }
+
+ /*
+ * register /proc/net/udpcp entry
+ */
+ ret = udpcp_proc_init();
+
+ if (ret)
+ goto err;
+#endif
+ pr_info("UDPCP protocol stack version " VERSION "\n");
+ return 0;
+#ifdef CONFIG_PROC_FS
+err:
+ if (proc_entry)
+ remove_proc_entry(UDPCP_PROC, NULL);
+ proto_unregister(&udpcp_prot);
+ return ret;
+#endif
+}
+
+/*
+ * Cleanup and exit module
+ */
+static void __exit udpcp_exit(void)
+{
+#ifdef CONFIG_PROC_FS
+ udpcp_proc_exit();
+ remove_proc_entry(UDPCP_PROC, NULL);
+#endif
+ inet_unregister_protosw(&udpcp_protosw);
+ proto_unregister(&udpcp_prot);
+}
+
+module_init(udpcp_init);
+module_exit(udpcp_exit);
+
+MODULE_AUTHOR("Stefani Seibold <stefani@seibold.net>");
+MODULE_DESCRIPTION("UDPCP protocol stack v" VERSION);
+MODULE_LICENSE("GPL");
+
--
1.7.3.4
^ permalink raw reply related
* Re: [PATCH] UDPCP Communication Protocol
From: Stefani Seibold @ 2011-01-01 21:40 UTC (permalink / raw)
To: Eric Dumazet; +Cc: linux-kernel, akpm, davem, netdev
In-Reply-To: <1293796458.2973.59.camel@edumazet-laptop>
Am Freitag, den 31.12.2010, 12:54 +0100 schrieb Eric Dumazet:
> Le vendredi 31 décembre 2010 à 12:23 +0100, Stefani Seibold a écrit :
> > Am Freitag, den 31.12.2010, 11:41 +0100 schrieb Eric Dumazet:
> > > Le vendredi 31 décembre 2010 à 11:22 +0100, Stefani Seibold a écrit :
> > > > Am Freitag, den 31.12.2010, 11:00 +0100 schrieb Eric Dumazet:
> > > > > Le vendredi 31 décembre 2010 à 10:29 +0100, stefani@seibold.net a
> > > > > écrit :
> > > > > > From: Stefani Seibold <stefani@seibold.net>
> > > > > >
> > > This is not how things are dealed in linux, really.
> > >
> > > You'll have to find a way so that things work well for everybody, not
> > > only for you.
> > >
> > > I guess you must fix UDPCP protocol stack, not 'fix linux'
> > >
> >
> > I cannot fix it, because the information is still lost, and i need it.
> >
>
> You can fix it. Really. If not, you can pay me and I'll fix it for you.
>
I am a poor girl, so i did it by me self. Hope you like the new
approach.
^ permalink raw reply
* Re: [PATCH] UDPCP Communication Protocol
From: Stefani Seibold @ 2011-01-01 21:28 UTC (permalink / raw)
To: Eric Dumazet; +Cc: linux-kernel, akpm, davem, netdev
In-Reply-To: <1293796805.2973.62.camel@edumazet-laptop>
Am Freitag, den 31.12.2010, 13:00 +0100 schrieb Eric Dumazet:
> Le vendredi 31 décembre 2010 à 12:25 +0100, Eric Dumazet a écrit :
> > Le vendredi 31 décembre 2010 à 10:29 +0100, stefani@seibold.net a
> > écrit :
> > > + if (!list_empty(&usk->destlist)) {
> > > + state->sk = (struct sock *)usk;
> > > + state->dest = list_first_entry(&usk->destlist,
> > > + struct udpcp_dest, list);
> > > + sock_hold(state->sk);
> > > +
> > > + if (atomic_read(&state->sk->sk_refcnt) != 1) {
> > > + spin_unlock_irqrestore(&spinlock, flags);
> > > + return state;
> > > + }
> > > + atomic_dec(&state->sk->sk_refcnt);
> > > + }
> > > +
> >
> > I am trying to understand what you are doing here.
> >
> > It seems racy to me.
> >
> > Apparently, what you want is to take a reference only if actual
> > sk_refcnt is not zero.
> >
> > I suggest using atomic_inc_notzero(&state->sk->sk_refcnt) to avoid the
> > race in atomic_dec().
> >
> >
>
> Before you ask why its racy, this is because UDP sockets are RCU
> protected, and RCU lookups depend on sk_refcnt being zero or not.
>
> Doing an sk_refcnt increment/decrement opens a race window for the
> concurrent lookups.
>
I still revamped the whole /proc/net/udpcp thing and hope it is now race
free.
^ permalink raw reply
* Re: [PATCH V8 02/13] ntp: add ADJ_SETOFFSET mode bit
From: Kuwahara,T. @ 2011-01-01 20:38 UTC (permalink / raw)
To: Richard Cochran
Cc: linux-kernel-u79uwXL29TY76Z2rM5mHXA,
linux-api-u79uwXL29TY76Z2rM5mHXA, netdev-u79uwXL29TY76Z2rM5mHXA,
Alan Cox, Arnd Bergmann, Christoph Lameter, David Miller,
John Stultz, Krzysztof Halasa, Peter Zijlstra, Rodolfo Giometti,
Thomas Gleixner
In-Reply-To: <fe2bec763bec155381d62721aa9da84aeca2785c.1293820862.git.richard.cochran-3mrvs1K0uXizZXS1Dc/lvw@public.gmane.org>
On Sat, Jan 1, 2011 at 4:12 AM, Richard Cochran
<richardcochran-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org> wrote:
> +#define ADJ_SETOFFSET 0x0040 /* add 'time' to current time */
As you know, it conflicts with MOD_PPSMAX. And also, it is logically
the same as ADJ_OFFSET, unless the kernel PLL is enabled explicitly.
So here's my simple solution:
---
diff --git a/kernel/time/ntp.c b/kernel/time/ntp.c
index c631168..d492887 100644
--- a/kernel/time/ntp.c
+++ b/kernel/time/ntp.c
@@ -119,14 +119,21 @@
return div_s64(offset64 << (NTP_SCALE_SHIFT - SHIFT_FLL), secs);
}
-static void ntp_update_offset(long offset)
+static void ntp_update_offset(long offset, struct timespec *ts)
{
s64 freq_adj;
s64 offset64;
long secs;
- if (!(time_status & STA_PLL))
+ if (!(time_status & STA_PLL)) {
+ offset64 = offset;
+ if (!(time_status & STA_NANO))
+ offset64 *= NSEC_PER_USEC;
+
+ set_normalized_timespec(ts, ts->tv_sec, offset64 + ts->tv_nsec);
+
return;
+ }
if (!(time_status & STA_NANO))
offset *= NSEC_PER_USEC;
@@ -430,7 +437,7 @@
time_tai = txc->constant;
if (txc->modes & ADJ_OFFSET)
- ntp_update_offset(txc->offset);
+ ntp_update_offset(txc->offset, ts);
if (txc->modes & ADJ_TICK)
tick_usec = txc->tick;
@@ -526,6 +533,9 @@
write_sequnlock_irq(&xtime_lock);
+ if ((txc->modes & ADJ_OFFSET) && !(time_status & STA_PLL))
+ do_settimeofday(&ts);
+
txc->time.tv_sec = ts.tv_sec;
txc->time.tv_usec = ts.tv_nsec;
if (!(time_status & STA_NANO))
--
^ permalink raw reply related
* Re: [PATCH 2.6.36] vlan: Avoid hwaccel vlan packets when vid not used
From: Eric Dumazet @ 2011-01-01 17:03 UTC (permalink / raw)
To: Matt Carlson
Cc: Jesse Gross, Michael Leun, Michael Chan, David Miller, Ben Greear,
linux-kernel@vger.kernel.org, netdev@vger.kernel.org
In-Reply-To: <20101214191500.GD19951@mcarlson.broadcom.com>
Le mardi 14 décembre 2010 à 11:15 -0800, Matt Carlson a écrit :
> Thanks for the comments Jesse. Below is an updated patch.
>
> Michael, I'm wondering if the difference in behavior can be explained by
> the presence or absence of management firmware. Can you look at the
> driver sign-on messages in your syslogs for ASF[]? I'm half expecting
> the 5752 to show "ASF[0]" and the 5714 to show "ASF[1]". If you see
> this, and the below patch doesn't fix the problem, let me know. I have
> another test I'd like you to run.
>
> ----
>
> [PATCH] tg3: Use new VLAN code
>
> This patch pivots the tg3 driver to the new VLAN infrastructure.
> All references to vlgrp have been removed and all VLAN code is
> unconditionally active.
>
> Signed-off-by: Matt Carlson <mcarlson@broadcom.com>
> ---
> drivers/net/tg3.c | 95 +++++------------------------------------------------
> drivers/net/tg3.h | 3 --
> 2 files changed, 9 insertions(+), 89 deletions(-)
>
> diff --git a/drivers/net/tg3.c b/drivers/net/tg3.c
> index 5faa87d..3682205 100644
> --- a/drivers/net/tg3.c
> +++ b/drivers/net/tg3.c
> @@ -60,12 +60,6 @@
> #define BAR_0 0
> #define BAR_2 2
>
> -#if defined(CONFIG_VLAN_8021Q) || defined(CONFIG_VLAN_8021Q_MODULE)
> -#define TG3_VLAN_TAG_USED 1
> -#else
> -#define TG3_VLAN_TAG_USED 0
> -#endif
> -
> #include "tg3.h"
>
> #define DRV_MODULE_NAME "tg3"
> @@ -134,9 +128,6 @@
> TG3_TX_RING_SIZE)
> #define NEXT_TX(N) (((N) + 1) & (TG3_TX_RING_SIZE - 1))
>
> -#define TG3_RX_DMA_ALIGN 16
> -#define TG3_RX_HEADROOM ALIGN(VLAN_HLEN, TG3_RX_DMA_ALIGN)
> -
> #define TG3_DMA_BYTE_ENAB 64
>
> #define TG3_RX_STD_DMA_SZ 1536
> @@ -4725,8 +4716,6 @@ static int tg3_rx(struct tg3_napi *tnapi, int budget)
> struct sk_buff *skb;
> dma_addr_t dma_addr;
> u32 opaque_key, desc_idx, *post_ptr;
> - bool hw_vlan __maybe_unused = false;
> - u16 vtag __maybe_unused = 0;
>
> desc_idx = desc->opaque & RXD_OPAQUE_INDEX_MASK;
> opaque_key = desc->opaque & RXD_OPAQUE_RING_MASK;
> @@ -4785,12 +4774,12 @@ static int tg3_rx(struct tg3_napi *tnapi, int budget)
> tg3_recycle_rx(tnapi, tpr, opaque_key,
> desc_idx, *post_ptr);
>
> - copy_skb = netdev_alloc_skb(tp->dev, len + VLAN_HLEN +
> + copy_skb = netdev_alloc_skb(tp->dev, len +
> TG3_RAW_IP_ALIGN);
> if (copy_skb == NULL)
> goto drop_it_no_recycle;
>
> - skb_reserve(copy_skb, TG3_RAW_IP_ALIGN + VLAN_HLEN);
> + skb_reserve(copy_skb, TG3_RAW_IP_ALIGN);
> skb_put(copy_skb, len);
> pci_dma_sync_single_for_cpu(tp->pdev, dma_addr, len, PCI_DMA_FROMDEVICE);
> skb_copy_from_linear_data(skb, copy_skb->data, len);
> @@ -4817,30 +4806,11 @@ static int tg3_rx(struct tg3_napi *tnapi, int budget)
> }
>
> if (desc->type_flags & RXD_FLAG_VLAN &&
> - !(tp->rx_mode & RX_MODE_KEEP_VLAN_TAG)) {
> - vtag = desc->err_vlan & RXD_VLAN_MASK;
> -#if TG3_VLAN_TAG_USED
> - if (tp->vlgrp)
> - hw_vlan = true;
> - else
> -#endif
> - {
> - struct vlan_ethhdr *ve = (struct vlan_ethhdr *)
> - __skb_push(skb, VLAN_HLEN);
> -
> - memmove(ve, skb->data + VLAN_HLEN,
> - ETH_ALEN * 2);
> - ve->h_vlan_proto = htons(ETH_P_8021Q);
> - ve->h_vlan_TCI = htons(vtag);
> - }
> - }
> + !(tp->rx_mode & RX_MODE_KEEP_VLAN_TAG))
> + __vlan_hwaccel_put_tag(skb,
> + desc->err_vlan & RXD_VLAN_MASK);
>
> -#if TG3_VLAN_TAG_USED
> - if (hw_vlan)
> - vlan_gro_receive(&tnapi->napi, tp->vlgrp, vtag, skb);
> - else
> -#endif
> - napi_gro_receive(&tnapi->napi, skb);
> + napi_gro_receive(&tnapi->napi, skb);
>
> received++;
> budget--;
> @@ -5743,11 +5713,9 @@ static netdev_tx_t tg3_start_xmit(struct sk_buff *skb,
> base_flags |= TXD_FLAG_TCPUDP_CSUM;
> }
>
> -#if TG3_VLAN_TAG_USED
> if (vlan_tx_tag_present(skb))
> base_flags |= (TXD_FLAG_VLAN |
> (vlan_tx_tag_get(skb) << 16));
> -#endif
>
> len = skb_headlen(skb);
>
> @@ -5989,11 +5957,10 @@ static netdev_tx_t tg3_start_xmit_dma_bug(struct sk_buff *skb,
> }
> }
> }
> -#if TG3_VLAN_TAG_USED
> +
> if (vlan_tx_tag_present(skb))
> base_flags |= (TXD_FLAG_VLAN |
> (vlan_tx_tag_get(skb) << 16));
> -#endif
>
> if ((tp->tg3_flags3 & TG3_FLG3_USE_JUMBO_BDFLAG) &&
> !mss && skb->len > VLAN_ETH_FRAME_LEN)
> @@ -9538,17 +9505,8 @@ static void __tg3_set_rx_mode(struct net_device *dev)
> /* When ASF is in use, we always keep the RX_MODE_KEEP_VLAN_TAG
> * flag clear.
> */
> -#if TG3_VLAN_TAG_USED
> - if (!tp->vlgrp &&
> - !(tp->tg3_flags & TG3_FLAG_ENABLE_ASF))
> - rx_mode |= RX_MODE_KEEP_VLAN_TAG;
> -#else
> - /* By definition, VLAN is disabled always in this
> - * case.
> - */
> if (!(tp->tg3_flags & TG3_FLAG_ENABLE_ASF))
> rx_mode |= RX_MODE_KEEP_VLAN_TAG;
> -#endif
>
> if (dev->flags & IFF_PROMISC) {
> /* Promiscuous mode. */
> @@ -11233,31 +11191,6 @@ static int tg3_ioctl(struct net_device *dev, struct ifreq *ifr, int cmd)
> return -EOPNOTSUPP;
> }
>
> -#if TG3_VLAN_TAG_USED
> -static void tg3_vlan_rx_register(struct net_device *dev, struct vlan_group *grp)
> -{
> - struct tg3 *tp = netdev_priv(dev);
> -
> - if (!netif_running(dev)) {
> - tp->vlgrp = grp;
> - return;
> - }
> -
> - tg3_netif_stop(tp);
> -
> - tg3_full_lock(tp, 0);
> -
> - tp->vlgrp = grp;
> -
> - /* Update RX_MODE_KEEP_VLAN_TAG bit in RX_MODE register. */
> - __tg3_set_rx_mode(dev);
> -
> - tg3_netif_start(tp);
> -
> - tg3_full_unlock(tp);
> -}
> -#endif
> -
> static int tg3_get_coalesce(struct net_device *dev, struct ethtool_coalesce *ec)
> {
> struct tg3 *tp = netdev_priv(dev);
> @@ -13069,9 +13002,7 @@ static struct pci_dev * __devinit tg3_find_peer(struct tg3 *);
>
> static void inline vlan_features_add(struct net_device *dev, unsigned long flags)
> {
> -#if TG3_VLAN_TAG_USED
> dev->vlan_features |= flags;
> -#endif
> }
>
> static inline u32 tg3_rx_ret_ring_size(struct tg3 *tp)
> @@ -13866,11 +13797,11 @@ static int __devinit tg3_get_invariants(struct tg3 *tp)
> else
> tp->tg3_flags &= ~TG3_FLAG_POLL_SERDES;
>
> - tp->rx_offset = NET_IP_ALIGN + TG3_RX_HEADROOM;
> + tp->rx_offset = NET_IP_ALIGN;
> tp->rx_copy_thresh = TG3_RX_COPY_THRESHOLD;
> if (GET_ASIC_REV(tp->pci_chip_rev_id) == ASIC_REV_5701 &&
> (tp->tg3_flags & TG3_FLAG_PCIX_MODE) != 0) {
> - tp->rx_offset -= NET_IP_ALIGN;
> + tp->rx_offset = 0;
> #ifndef CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS
> tp->rx_copy_thresh = ~(u16)0;
> #endif
> @@ -14634,9 +14565,6 @@ static const struct net_device_ops tg3_netdev_ops = {
> .ndo_do_ioctl = tg3_ioctl,
> .ndo_tx_timeout = tg3_tx_timeout,
> .ndo_change_mtu = tg3_change_mtu,
> -#if TG3_VLAN_TAG_USED
> - .ndo_vlan_rx_register = tg3_vlan_rx_register,
> -#endif
> #ifdef CONFIG_NET_POLL_CONTROLLER
> .ndo_poll_controller = tg3_poll_controller,
> #endif
> @@ -14653,9 +14581,6 @@ static const struct net_device_ops tg3_netdev_ops_dma_bug = {
> .ndo_do_ioctl = tg3_ioctl,
> .ndo_tx_timeout = tg3_tx_timeout,
> .ndo_change_mtu = tg3_change_mtu,
> -#if TG3_VLAN_TAG_USED
> - .ndo_vlan_rx_register = tg3_vlan_rx_register,
> -#endif
> #ifdef CONFIG_NET_POLL_CONTROLLER
> .ndo_poll_controller = tg3_poll_controller,
> #endif
> @@ -14705,9 +14630,7 @@ static int __devinit tg3_init_one(struct pci_dev *pdev,
>
> SET_NETDEV_DEV(dev, &pdev->dev);
>
> -#if TG3_VLAN_TAG_USED
> dev->features |= NETIF_F_HW_VLAN_TX | NETIF_F_HW_VLAN_RX;
> -#endif
>
> tp = netdev_priv(dev);
> tp->pdev = pdev;
> diff --git a/drivers/net/tg3.h b/drivers/net/tg3.h
> index d62c8d9..f528243 100644
> --- a/drivers/net/tg3.h
> +++ b/drivers/net/tg3.h
> @@ -2808,9 +2808,6 @@ struct tg3 {
> u32 rx_std_max_post;
> u32 rx_offset;
> u32 rx_pkt_map_sz;
> -#if TG3_VLAN_TAG_USED
> - struct vlan_group *vlgrp;
> -#endif
>
>
> /* begin "everything else" cacheline(s) section */
Hi Matt.
Any news on this patch ?
Without it, net-next-2.6 doesnt work for me on a vlan setup on top of
bonding.
(bond0 : eth1 & eth2, eth1 being bnx2, eth2 beging tg3)
ip link add link bond0 vlan.103 type vlan id 103
ip addr add 192.168.20.110/24 dev vlan.103
ip link set vlan.103 up
If active slave is eth1 (bnx2), everything works, but if active slave is
eth2 (tg3), incoming tagged frames (on vlan 103) are lost.
^ permalink raw reply
* [PATCH net-next-2.6] tg3: fix warnings
From: Eric Dumazet @ 2011-01-01 15:22 UTC (permalink / raw)
To: David Miller; +Cc: rjw, netdev, mcarlson, mchan, linux-pm, Jesse Gross
In-Reply-To: <20101231.111440.28810446.davem@davemloft.net>
Le vendredi 31 décembre 2010 à 11:14 -0800, David Miller a écrit :
> From: "Rafael J. Wysocki" <rjw@sisk.pl>
> Date: Sat, 25 Dec 2010 23:56:23 +0100
>
> > From: Rafael J. Wysocki <rjw@sisk.pl>
> >
> > The tg3 driver uses the legacy PCI power management, so it has to do
> > some PCI-specific things in its ->suspend() and ->resume() callbacks,
> > which isn't necessary and should better be done by the PCI
> > sybsystem-level power management code.
> >
> > Convert tg3 to the new PCI power management framework and make it
> > let the PCI subsystem take care of all the PCI-specific aspects of
> > device handling during system power transitions.
> >
> > Tested on HP nx6325 with a NetXtreme BCM5788 adapter.
> >
> > Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
>
> Applied.
Happy new year everybody ;)
Here is a followup to avoid some warnings if CONFIG_PM_SLEEP=n
Now I have to understand why vlan/bonding doesnt work anymore with tg3
on current net-next-2.6 :(
Is anybody already on this problem ?
(Cc Jesse Gross)
Thanks
[PATCH net-next-2.6] tg3: fix warnings
In case CONFIG_PM_SLEEP is disabled, we dont need tg3_suspend() and
tg3_resume().
drivers/net/tg3.c:15056: warning: ‘tg3_suspend’ defined but not used
drivers/net/tg3.c:15110: warning: ‘tg3_resume’ defined but not used
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Cc: Rafael J. Wysocki <rjw@sisk.pl>
Cc: Michael Chan <mchan@broadcom.com>
Cc: Matt Carlson <mcarlson@broadcom.com>
---
drivers/net/tg3.c | 10 +++++++++-
1 files changed, 9 insertions(+), 1 deletion(-)
diff --git a/drivers/net/tg3.c b/drivers/net/tg3.c
index 6137869..e3d80c9 100644
--- a/drivers/net/tg3.c
+++ b/drivers/net/tg3.c
@@ -15052,6 +15052,7 @@ static void __devexit tg3_remove_one(struct pci_dev *pdev)
}
}
+#ifdef CONFIG_PM_SLEEP
static int tg3_suspend(struct device *device)
{
struct pci_dev *pdev = to_pci_dev(device);
@@ -15140,13 +15141,20 @@ out:
}
static SIMPLE_DEV_PM_OPS(tg3_pm_ops, tg3_suspend, tg3_resume);
+#define TG3_PM_OPS (&tg3_pm_ops)
+
+#else
+
+#define TG3_PM_OPS NULL
+
+#endif /* CONFIG_PM_SLEEP */
static struct pci_driver tg3_driver = {
.name = DRV_MODULE_NAME,
.id_table = tg3_pci_tbl,
.probe = tg3_init_one,
.remove = __devexit_p(tg3_remove_one),
- .driver.pm = &tg3_pm_ops,
+ .driver.pm = TG3_PM_OPS,
};
static int __init tg3_init(void)
^ permalink raw reply related
* [PATCH] atl1: fix oops when changing tx/rx ring params
From: J. K. Cliburn @ 2011-01-01 15:02 UTC (permalink / raw)
To: David Miller; +Cc: netdev, stable, jussuf, chris.snook, kronos.it, Xiong.Huang
Commit 3f5a2a713aad28480d86b0add00c68484b54febc zeroes out the statistics
message block (SMB) and coalescing message block (CMB) when adapter ring
resources are freed. This is desirable behavior, but, as a side effect,
the commit leads to an oops when atl1_set_ringparam() attempts to alter
the number of rx or tx elements in the ring buffer (by using ethtool
-G, for example). We don't want SMB or CMB to change during this
operation.
Modify atl1_set_ringparam() to preserve SMB and CMB when changing ring
parameters.
Cc: stable@kernel.org
Signed-off-by: Jay Cliburn <jcliburn@gmail.com>
Reported-by: Tõnu Raitviir <jussuf@linux.ee>
---
drivers/net/atlx/atl1.c | 10 ++++++++++
1 files changed, 10 insertions(+), 0 deletions(-)
diff --git a/drivers/net/atlx/atl1.c b/drivers/net/atlx/atl1.c
index 5336310..3acf512 100644
--- a/drivers/net/atlx/atl1.c
+++ b/drivers/net/atlx/atl1.c
@@ -3504,6 +3504,8 @@ static int atl1_set_ringparam(struct net_device *netdev,
struct atl1_rfd_ring rfd_old, rfd_new;
struct atl1_rrd_ring rrd_old, rrd_new;
struct atl1_ring_header rhdr_old, rhdr_new;
+ struct atl1_smb smb;
+ struct atl1_cmb cmb;
int err;
tpd_old = adapter->tpd_ring;
@@ -3544,11 +3546,19 @@ static int atl1_set_ringparam(struct net_device *netdev,
adapter->rrd_ring = rrd_old;
adapter->tpd_ring = tpd_old;
adapter->ring_header = rhdr_old;
+ /*
+ * Save SMB and CMB, since atl1_free_ring_resources
+ * will clear them.
+ */
+ smb = adapter->smb;
+ cmb = adapter->cmb;
atl1_free_ring_resources(adapter);
adapter->rfd_ring = rfd_new;
adapter->rrd_ring = rrd_new;
adapter->tpd_ring = tpd_new;
adapter->ring_header = rhdr_new;
+ adapter->smb = smb;
+ adapter->cmb = cmb;
err = atl1_up(adapter);
if (err)
--
1.7.2.3
^ permalink raw reply related
* Re: [*v3 PATCH 02/22] IPVS: netns to services part 1
From: Jan Engelhardt @ 2011-01-01 14:57 UTC (permalink / raw)
To: hans
Cc: horms, ja, daniel.lezcano, wensong, lvs-devel, netdev,
netfilter-devel, Hans Schillstrom
In-Reply-To: <1293706266-27152-3-git-send-email-hans@schillstrom.com>
On Thursday 2010-12-30 11:50, hans@schillstrom.com wrote:
>+/*
>+ * Get net ptr from skb in traffic cases
>+ * use skb_sknet when call is from userland (ioctl or netlink)
>+ */
>+static inline struct net *skb_net(struct sk_buff *skb) {
>+#ifdef CONFIG_NET_NS
>+#ifdef CONFIG_IP_VS_DEBUG
>+ /*
>+ * This is used for debug only.
>+ * Start with the most likely hit
>+ * End with BUG
>+ */
>+ if (likely(skb->dev && skb->dev->nd_net))
>+ return dev_net(skb->dev);
>+ if (skb_dst(skb)->dev)
>+ return dev_net(skb_dst(skb)->dev);
>+ WARN(skb->sk,"Maybe skb_sknet should be used instead in %s() line:%d\n",
>+ __func__, __LINE__);
>+ if (likely(skb->sk && skb->sk->sk_net))
>+ return sock_net(skb->sk);
>+ pr_err("There is no net ptr to find in the skb in %s() line:%d\n",
>+ __func__, __LINE__);
>+ BUG();
>+#else
>+ return dev_net(skb->dev ? : skb_dst(skb)->dev);
>+#endif
>+#else
>+ return &init_net;
>+#endif
>+}
Whether NETNS is disabled or not, dev_net(skb->dev) can be used in either case,
so the extra return &init_net case is not really required AFAICS.
>@@ -3446,43 +3471,48 @@ static struct pernet_operations ipvs_control_ops = {
>- smp_wmb();
>+
>+ smp_wmb(); /* Do wee really need it now ? */
not "whee" :)
^ permalink raw reply
* Re: [PATCH 03/15]drivers:staging:rtl8187se:r8180_hw.h Typo change diable to disable.
From: Justin P. Mattock @ 2011-01-01 14:53 UTC (permalink / raw)
To: Dan Carpenter, Finn Thain, devel-gWbeCf7V1WCQmaza687I9mD2FQJk+8+b,
trivial-DgEjT+Ai2ygdnm+yROfE0A, linux-scsi-u79uwXL29TY76Z2rM5mHXA,
netdev-u79uwXL29TY76Z2rM5mHXA, linux-
In-Reply-To: <20110101090931.GH1886@bicker>
On 01/01/2011 01:09 AM, Dan Carpenter wrote:
> On Fri, Dec 31, 2010 at 11:43:30PM -0800, Justin P. Mattock wrote:
>> On 12/31/2010 10:48 PM, Finn Thain wrote:
>>>> -/* BIT[8-9] is for SW Antenna Diversity. Only the value EEPROM_SW_AD_ENABLE means enable, other values are diable. */
>>>> +/* BIT[8-9] is for SW Antenna Diversity. Only the value EEPROM_SW_AD_ENABLE means enable, other values are disabled. */
>>>
>>> I think, "other values disable" was what you meant?
>>>
>>> Finn
>>>
>>>> #define EEPROM_SW_AD_MASK 0x0300
>>>> #define EEPROM_SW_AD_ENABLE 0x0100
>>>>
>>>>
>>>
>>
>> no! I changed it to disabled to make it proper..
>
> Finn is obviously right, but maybe a compromise would be:
>
> Only the value EEPROM_SW_AD_ENABLE means "enable", other values mean
> "disable".
>
> regards,
> dan carpenter
>
ahh.. I see what you your saying now.. alright let me send this out that
way..
Justin P. Mattock
--
To unsubscribe from this list: send the line "unsubscribe linux-wireless" 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: [*v3 PATCH 00/22] IPVS Network Name Space aware
From: Julian Anastasov @ 2011-01-01 12:27 UTC (permalink / raw)
To: hans
Cc: horms, daniel.lezcano, wensong, lvs-devel, netdev,
netfilter-devel, Hans Schillstrom
In-Reply-To: <1293706266-27152-1-git-send-email-hans@schillstrom.com>
Hello,
On Thu, 30 Dec 2010, hans@schillstrom.com wrote:
> From: Hans Schillstrom <hans.schillstrom@ericsson.com>
>
> This patch series adds network name space support to the LVS.
>
> REVISION
>
> This is version 3
>
> OVERVIEW
>
> The patch doesn't remove or add any functionality except for netns.
> For users that don't use network name space (netns) this patch is
> completely transparent.
>
> Now it's possible to run LVS in a Linux container (see lxc-tools)
> i.e. a light weight visualization. For example it's possible to run
> one or several lvs on a real server in their own network name spaces.
>> From the LVS point of view it looks like it runs on it's own machine.
Only some comments for two of the patches:
v3 PATCH 10/22 use ip_vs_proto_data as param
- Can ip_vs_protocol_timeout_change() walk
proto_data_table instead of ip_vs_proto_table to avoid the
__ipvs_proto_data_get call?
v3 PATCH 15/22 ip_vs_stats
- Is ustats_seq allocated with alloc_percpu?
Such reader sections should be changed to use tmp vars because
on retry we risk to add the values multiple times. For example:
do {
start = read_seqcount_begin(seq_count);
ipvs->ctl_stats->ustats.inbytes += u->inbytes;
ipvs->ctl_stats->ustats.outbytes += u->outbytes;
} while (read_seqcount_retry(seq_count, start));
should be changed as follows:
u64 inbytes, outbytes;
do {
start = read_seqcount_begin(seq_count);
inbytes = u->inbytes;
outbytes = u->outbytes;
} while (read_seqcount_retry(seq_count, start));
ipvs->ctl_stats->ustats.inbytes += inbytes;
ipvs->ctl_stats->ustats.outbytes += outbytes;
Or it is better to create new struct for percpu stats,
they will have their own syncp, because we can not
change struct ip_vs_stats_user. syncp should be percpu
because we remove locks.
For example:
struct ip_vs_cpu_stats {
struct ip_vs_stats_user ustats;
struct u64_stats_sync syncp;
};
Then we can add this in struct netns_ipvs:
struct ip_vs_cpu_stats __percpu *stats; /* Statistics */
without the seqcount_t * ustats_seq;
Then syncp does not need any initialization, it seems
alloc_percpu returns zeroed area.
When we use percpu stats for all places (dest and svc) we
can create new struct struct ip_vs_counters, so that we
can reduce the memory usage from percpu data. Now stats
include counters and estimated values. The estimated
values should not be percpu. Then ip_vs_cpu_stats
will be shorter (it is not visible to user space):
struct ip_vs_cpu_stats {
struct ip_vs_counters ustats;
struct u64_stats_sync syncp;
};
For writer side in softirq context we should protect the whole
section with u64 counters, for example this code:
spin_lock(&ip_vs_stats.lock);
ip_vs_stats.ustats.outpkts++;
ip_vs_stats.ustats.outbytes += skb->len;
spin_unlock(&ip_vs_stats.lock);
should be changed to:
struct ip_vs_cpu_stats *s = this_cpu_ptr(ipvs->stats);
u64_stats_update_begin(&s->syncp);
s->ustats.outpkts++;
s->ustats.outbytes += skb->len;
u64_stats_update_end(&s->syncp);
Readers should look in similar way, only that _bh fetching
should be used only for the u64 counters because writer is in
softirq context:
u64 inbytes, outbytes;
for_each_possible_cpu(i) {
const struct ip_vs_cpu_stats *s = per_cpu_ptr(ipvs->stats, i);
unsigned int start;
...
do {
start = u64_stats_fetch_begin_bh(&s->syncp);
inbytes = s->ustats.inbytes;
outbytes = s->ustats.outbytes;
} while (u64_stats_fetch_retry_bh(&s->syncp, start));
ipvs->ctl_stats->ustats.inbytes += inbytes;
ipvs->ctl_stats->ustats.outbytes += outbytes;
}
Then IPVS_STAT_ADD and IPVS_STAT_INC can not access
the syncp. They do not look generic enough, in case we
decide to use struct ip_vs_cpu_stats for dest and svc stats.
I think, these macros are not needed.
It is possible to have other macros for write side,
for percpu stats:
#define IP_VS_STATS_UPDATE_BEGIN(stats) \
{ struct ip_vs_cpu_stats *s = this_cpu_ptr(stats); \
u64_stats_update_begin(&s->syncp)
#define IP_VS_STATS_UPDATE_END() \
u64_stats_update_end(&s->syncp); \
}
Then usage can be:
IP_VS_STATS_UPDATE_BEGIN(ipvs->stats);
s->ustats.outpkts++;
s->ustats.outbytes += skb->len;
IP_VS_STATS_UPDATE_END();
Then we can rename ctl_stats to total_stats or full_stats.
get_stats() can be changed to ip_vs_read_cpu_stats(), it
will be used to read all percpu stats as sum in the
total_stats/full_stats structure or tmp space:
/* Get sum of percpu stats */
... ip_vs_read_cpu_stats(struct ip_vs_stats_user *sum,
struct ip_vs_cpu_stats *stats)
{
...
for_each_possible_cpu(i) {
const struct ip_vs_cpu_stats *s = per_cpu_ptr(stats, i);
unsigned int start;
if (i) {...
...
do {
start = u64_stats_fetch_begin_bh(&s->syncp);
inbytes = s->ustats.inbytes;
outbytes = s->ustats.outbytes;
} while (u64_stats_fetch_retry_bh(&s->syncp, start));
sum->inbytes += inbytes;
sum->outbytes += outbytes;
}
}
As result, ip_vs_read_cpu_stats() will be used in
estimation_timer() instead of get_stats():
ip_vs_read_cpu_stats(&ipvs->total_stats->ustats, ipvs->stats);
but also in ip_vs_stats_show()
{
// I hope struct ip_vs_stats_user can fit in stack:
struct ip_vs_stats_user sum;
here we do not need to use spin_lock_bh(&ctl_stats->lock);
because now the stats are lockless, estimation_timer
does not use ctl_stats->lock, so we have to call
ip_vs_read_cpu_stats to avoid problems with
reading incorrect u64 values directly from
ipvs->total_stats->ustats.
ip_vs_read_cpu_stats(&sum, ipvs->stats);
seq_printf for sum
}
ip_vs_read_cpu_stats can be used also in ip_vs_copy_stats()
which copies values to user space and needs proper u64 reading.
But it is used only for svc and dest stats which are not
percpu yet.
Now this code does not look ok:
ip_vs_zero_stats(net_ipvs(net)->ctl_stats);
May be we need new func ip_vs_zero_percpu_stats that will
reset stats for all CPUs, i.e. ipvs->stats in our case?
Even if this operation is not very safe if done from
user space while the softirq can modify u64 counters
on another CPU.
Happy New Year!
^ permalink raw reply
* Re: [PATCH 03/15]drivers:staging:rtl8187se:r8180_hw.h Typo change diable to disable.
From: Dan Carpenter @ 2011-01-01 9:09 UTC (permalink / raw)
To: Justin P. Mattock
Cc: devel, trivial, linux-scsi, netdev, linux-usb, linux-wireless,
linux-kernel, Finn Thain, linux-m68k, ivtv-devel,
spi-devel-general, linux-media
In-Reply-To: <4D1EDB22.2020308@gmail.com>
On Fri, Dec 31, 2010 at 11:43:30PM -0800, Justin P. Mattock wrote:
> On 12/31/2010 10:48 PM, Finn Thain wrote:
> >>-/* BIT[8-9] is for SW Antenna Diversity. Only the value EEPROM_SW_AD_ENABLE means enable, other values are diable. */
> >>+/* BIT[8-9] is for SW Antenna Diversity. Only the value EEPROM_SW_AD_ENABLE means enable, other values are disabled. */
> >
> >I think, "other values disable" was what you meant?
> >
> >Finn
> >
> >> #define EEPROM_SW_AD_MASK 0x0300
> >> #define EEPROM_SW_AD_ENABLE 0x0100
> >>
> >>
> >
>
> no! I changed it to disabled to make it proper..
Finn is obviously right, but maybe a compromise would be:
Only the value EEPROM_SW_AD_ENABLE means "enable", other values mean
"disable".
regards,
dan carpenter
^ permalink raw reply
* Re: [PATCH 03/15]drivers:staging:rtl8187se:r8180_hw.h Typo change diable to disable.
From: Justin P. Mattock @ 2011-01-01 7:43 UTC (permalink / raw)
To: Finn Thain
Cc: devel, trivial, linux-scsi, netdev, linux-usb, linux-wireless,
linux-kernel, ivtv-devel, linux-m68k, spi-devel-general,
linux-media
In-Reply-To: <alpine.LNX.2.00.1012311722580.24460@nippy.intranet>
On 12/31/2010 10:48 PM, Finn Thain wrote:
>
> On Thu, 30 Dec 2010, Justin P. Mattock wrote:
>
>> The below patch fixes a typo "diable" to "disable". Please let me know if this
>> is correct or not.
>>
>> Signed-off-by: Justin P. Mattock<justinmattock@gmail.com>
>>
>> ---
>> drivers/staging/rtl8187se/r8180_hw.h | 2 +-
>> 1 files changed, 1 insertions(+), 1 deletions(-)
>>
>> diff --git a/drivers/staging/rtl8187se/r8180_hw.h b/drivers/staging/rtl8187se/r8180_hw.h
>> index 3fca144..2911d40 100644
>> --- a/drivers/staging/rtl8187se/r8180_hw.h
>> +++ b/drivers/staging/rtl8187se/r8180_hw.h
>> @@ -554,7 +554,7 @@
>> /* by amy for power save */
>> /* by amy for antenna */
>> #define EEPROM_SW_REVD_OFFSET 0x3f
>> -/* BIT[8-9] is for SW Antenna Diversity. Only the value EEPROM_SW_AD_ENABLE means enable, other values are diable. */
>> +/* BIT[8-9] is for SW Antenna Diversity. Only the value EEPROM_SW_AD_ENABLE means enable, other values are disabled. */
>
> I think, "other values disable" was what you meant?
>
> Finn
>
>> #define EEPROM_SW_AD_MASK 0x0300
>> #define EEPROM_SW_AD_ENABLE 0x0100
>>
>>
>
no! I changed it to disabled to make it proper..
Justin P. Mattock
^ permalink raw reply
* Re: [PATCH 03/15]drivers:staging:rtl8187se:r8180_hw.h Typo change diable to disable.
From: Finn Thain @ 2011-01-01 6:48 UTC (permalink / raw)
To: Justin P. Mattock
Cc: devel, trivial, linux-scsi, netdev, linux-usb, linux-wireless,
linux-kernel, ivtv-devel, linux-m68k, spi-devel-general,
linux-media
In-Reply-To: <1293750484-1161-3-git-send-email-justinmattock@gmail.com>
On Thu, 30 Dec 2010, Justin P. Mattock wrote:
> The below patch fixes a typo "diable" to "disable". Please let me know if this
> is correct or not.
>
> Signed-off-by: Justin P. Mattock <justinmattock@gmail.com>
>
> ---
> drivers/staging/rtl8187se/r8180_hw.h | 2 +-
> 1 files changed, 1 insertions(+), 1 deletions(-)
>
> diff --git a/drivers/staging/rtl8187se/r8180_hw.h b/drivers/staging/rtl8187se/r8180_hw.h
> index 3fca144..2911d40 100644
> --- a/drivers/staging/rtl8187se/r8180_hw.h
> +++ b/drivers/staging/rtl8187se/r8180_hw.h
> @@ -554,7 +554,7 @@
> /* by amy for power save */
> /* by amy for antenna */
> #define EEPROM_SW_REVD_OFFSET 0x3f
> -/* BIT[8-9] is for SW Antenna Diversity. Only the value EEPROM_SW_AD_ENABLE means enable, other values are diable. */
> +/* BIT[8-9] is for SW Antenna Diversity. Only the value EEPROM_SW_AD_ENABLE means enable, other values are disabled. */
I think, "other values disable" was what you meant?
Finn
> #define EEPROM_SW_AD_MASK 0x0300
> #define EEPROM_SW_AD_ENABLE 0x0100
>
>
^ permalink raw reply
* [PATCH net-next 20/20] tipc: remove extraneous braces from single statements
From: Paul Gortmaker @ 2011-01-01 4:59 UTC (permalink / raw)
To: davem; +Cc: netdev, allan.stephens, Allan Stephens, Paul Gortmaker
In-Reply-To: <1293857975-30267-1-git-send-email-paul.gortmaker@windriver.com>
From: Allan Stephens <Allan.Stephens@windriver.com>
Cleans up TIPC's source code to eliminate the presence of unnecessary
use of {} around single statements.
These changes are purely cosmetic and do not alter the operation of TIPC
in any way.
Signed-off-by: Allan Stephens <Allan.Stephens@windriver.com>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>
---
net/tipc/bcast.c | 6 ++----
net/tipc/bearer.c | 3 +--
net/tipc/link.c | 25 +++++++++----------------
net/tipc/msg.c | 15 ++++++---------
net/tipc/name_table.c | 6 ++----
net/tipc/port.c | 6 ++----
net/tipc/socket.c | 14 +++++---------
7 files changed, 27 insertions(+), 48 deletions(-)
diff --git a/net/tipc/bcast.c b/net/tipc/bcast.c
index 99a1469..70ab5ef 100644
--- a/net/tipc/bcast.c
+++ b/net/tipc/bcast.c
@@ -196,9 +196,8 @@ static void bclink_retransmit_pkt(u32 after, u32 to)
struct sk_buff *buf;
buf = bcl->first_out;
- while (buf && less_eq(buf_seqno(buf), after)) {
+ while (buf && less_eq(buf_seqno(buf), after))
buf = buf->next;
- }
tipc_link_retransmit(bcl, buf, mod(to - after));
}
@@ -224,9 +223,8 @@ void tipc_bclink_acknowledge(struct tipc_node *n_ptr, u32 acked)
/* Skip over packets that node has previously acknowledged */
crs = bcl->first_out;
- while (crs && less_eq(buf_seqno(crs), n_ptr->bclink.acked)) {
+ while (crs && less_eq(buf_seqno(crs), n_ptr->bclink.acked))
crs = crs->next;
- }
/* Update packets that node is now acknowledging */
diff --git a/net/tipc/bearer.c b/net/tipc/bearer.c
index 68e7290..837b7a4 100644
--- a/net/tipc/bearer.c
+++ b/net/tipc/bearer.c
@@ -194,9 +194,8 @@ void tipc_media_addr_printf(struct print_buf *pb, struct tipc_media_addr *a)
unchar *addr = (unchar *)&a->dev_addr;
tipc_printf(pb, "UNKNOWN(%u)", media_type);
- for (i = 0; i < (sizeof(*a) - sizeof(a->type)); i++) {
+ for (i = 0; i < (sizeof(*a) - sizeof(a->type)); i++)
tipc_printf(pb, "-%02x", addr[i]);
- }
}
}
diff --git a/net/tipc/link.c b/net/tipc/link.c
index de9d491..18702f5 100644
--- a/net/tipc/link.c
+++ b/net/tipc/link.c
@@ -615,9 +615,8 @@ static void link_state_event(struct link *l_ptr, unsigned event)
return; /* Not yet. */
if (link_blocked(l_ptr)) {
- if (event == TIMEOUT_EVT) {
+ if (event == TIMEOUT_EVT)
link_set_timer(l_ptr, cont_intv);
- }
return; /* Changeover going on */
}
@@ -940,11 +939,10 @@ int tipc_link_send(struct sk_buff *buf, u32 dest, u32 selector)
if (n_ptr) {
tipc_node_lock(n_ptr);
l_ptr = n_ptr->active_links[selector & 1];
- if (l_ptr) {
+ if (l_ptr)
res = tipc_link_send_buf(l_ptr, buf);
- } else {
+ else
buf_discard(buf);
- }
tipc_node_unlock(n_ptr);
} else {
buf_discard(buf);
@@ -1626,9 +1624,8 @@ void tipc_recv_msg(struct sk_buff *head, struct tipc_bearer *tb_ptr)
/* Ensure message data is a single contiguous unit */
- if (unlikely(buf_linearize(buf))) {
+ if (unlikely(buf_linearize(buf)))
goto cont;
- }
/* Handle arrival of a non-unicast link message */
@@ -1843,9 +1840,8 @@ u32 tipc_link_defer_pkt(struct sk_buff **head,
*head = buf;
return 1;
}
- if (seq_no == msg_seqno(msg)) {
+ if (seq_no == msg_seqno(msg))
break;
- }
prev = crs;
crs = crs->next;
} while (crs);
@@ -1959,11 +1955,10 @@ void tipc_link_send_proto_msg(struct link *l_ptr, u32 msg_typ, int probe_msg,
msg_set_max_pkt(msg, l_ptr->max_pkt_target);
}
- if (tipc_node_has_redundant_links(l_ptr->owner)) {
+ if (tipc_node_has_redundant_links(l_ptr->owner))
msg_set_redundant_link(msg);
- } else {
+ else
msg_clear_redundant_link(msg);
- }
msg_set_linkprio(msg, l_ptr->priority);
/* Ensure sequence number will not fit : */
@@ -2071,9 +2066,8 @@ static void link_recv_proto_msg(struct link *l_ptr, struct sk_buff *buf)
l_ptr->peer_bearer_id = msg_bearer_id(msg);
/* Synchronize broadcast sequence numbers */
- if (!tipc_node_has_redundant_links(l_ptr->owner)) {
+ if (!tipc_node_has_redundant_links(l_ptr->owner))
l_ptr->owner->bclink.last_in = mod(msg_last_bcast(msg));
- }
break;
case STATE_MSG:
@@ -2108,9 +2102,8 @@ static void link_recv_proto_msg(struct link *l_ptr, struct sk_buff *buf)
max_pkt_ack = 0;
if (msg_probe(msg)) {
l_ptr->stats.recv_probes++;
- if (msg_size(msg) > sizeof(l_ptr->proto_msg)) {
+ if (msg_size(msg) > sizeof(l_ptr->proto_msg))
max_pkt_ack = msg_size(msg);
- }
}
/* Protocol message before retransmits, reduce loss risk */
diff --git a/net/tipc/msg.c b/net/tipc/msg.c
index e81d43a..bb6180c 100644
--- a/net/tipc/msg.c
+++ b/net/tipc/msg.c
@@ -348,7 +348,8 @@ void tipc_msg_dbg(struct print_buf *buf, struct tipc_msg *msg, const char *str)
tipc_printf(buf, "UNKNOWN ERROR(%x):",
msg_errcode(msg));
}
- default:{}
+ default:
+ break;
}
tipc_printf(buf, "HZ(%u):", msg_hdr_sz(msg));
@@ -357,9 +358,8 @@ void tipc_msg_dbg(struct print_buf *buf, struct tipc_msg *msg, const char *str)
if (msg_non_seq(msg))
tipc_printf(buf, "NOSEQ:");
- else {
+ else
tipc_printf(buf, "ACK(%u):", msg_ack(msg));
- }
tipc_printf(buf, "BACK(%u):", msg_bcast_ack(msg));
tipc_printf(buf, "PRND(%x)", msg_prevnode(msg));
@@ -387,9 +387,8 @@ void tipc_msg_dbg(struct print_buf *buf, struct tipc_msg *msg, const char *str)
if (msg_user(msg) == NAME_DISTRIBUTOR) {
tipc_printf(buf, ":ONOD(%x):", msg_orignode(msg));
tipc_printf(buf, ":DNOD(%x):", msg_destnode(msg));
- if (msg_routed(msg)) {
+ if (msg_routed(msg))
tipc_printf(buf, ":CSEQN(%u)", msg_transp_seqno(msg));
- }
}
if (msg_user(msg) == LINK_CONFIG) {
@@ -405,12 +404,10 @@ void tipc_msg_dbg(struct print_buf *buf, struct tipc_msg *msg, const char *str)
tipc_printf(buf, "TO(%u):", msg_bcgap_to(msg));
}
tipc_printf(buf, "\n");
- if ((usr == CHANGEOVER_PROTOCOL) && (msg_msgcnt(msg))) {
+ if ((usr == CHANGEOVER_PROTOCOL) && (msg_msgcnt(msg)))
tipc_msg_dbg(buf, msg_get_wrapped(msg), " /");
- }
- if ((usr == MSG_FRAGMENTER) && (msg_type(msg) == FIRST_FRAGMENT)) {
+ if ((usr == MSG_FRAGMENTER) && (msg_type(msg) == FIRST_FRAGMENT))
tipc_msg_dbg(buf, msg_get_wrapped(msg), " /");
- }
}
#endif
diff --git a/net/tipc/name_table.c b/net/tipc/name_table.c
index ea3e94e..205ed4a 100644
--- a/net/tipc/name_table.c
+++ b/net/tipc/name_table.c
@@ -751,9 +751,8 @@ struct publication *tipc_nametbl_publish(u32 type, u32 lower, u32 upper,
table.local_publ_count++;
publ = tipc_nametbl_insert_publ(type, lower, upper, scope,
tipc_own_addr, port_ref, key);
- if (publ && (scope != TIPC_NODE_SCOPE)) {
+ if (publ && (scope != TIPC_NODE_SCOPE))
tipc_named_publish(publ);
- }
write_unlock_bh(&tipc_nametbl_lock);
return publ;
}
@@ -795,9 +794,8 @@ void tipc_nametbl_subscribe(struct subscription *s)
write_lock_bh(&tipc_nametbl_lock);
seq = nametbl_find_seq(type);
- if (!seq) {
+ if (!seq)
seq = tipc_nameseq_create(type, &table.types[hash(type)]);
- }
if (seq) {
spin_lock_bh(&seq->lock);
tipc_nameseq_subscribe(seq, s);
diff --git a/net/tipc/port.c b/net/tipc/port.c
index c23f338..067bab2 100644
--- a/net/tipc/port.c
+++ b/net/tipc/port.c
@@ -131,9 +131,8 @@ int tipc_multicast(u32 ref, struct tipc_name_seq const *seq,
}
}
res = tipc_bclink_send_msg(buf);
- if ((res < 0) && (dports.count != 0)) {
+ if ((res < 0) && (dports.count != 0))
buf_discard(ibuf);
- }
} else {
ibuf = buf;
}
@@ -190,9 +189,8 @@ void tipc_port_recv_mcast(struct sk_buff *buf, struct port_list *dp)
warn("Unable to deliver multicast message(s)\n");
goto exit;
}
- if ((index == 0) && (cnt != 0)) {
+ if ((index == 0) && (cnt != 0))
item = item->next;
- }
msg_set_destport(buf_msg(b), item->ports[index]);
tipc_port_recv_msg(b);
}
diff --git a/net/tipc/socket.c b/net/tipc/socket.c
index 18aad57..2b02a3a 100644
--- a/net/tipc/socket.c
+++ b/net/tipc/socket.c
@@ -590,9 +590,8 @@ static int send_msg(struct kiocb *iocb, struct socket *sock,
m->msg_iov);
}
if (likely(res != -ELINKCONG)) {
- if (needs_conn && (res >= 0)) {
+ if (needs_conn && (res >= 0))
sock->state = SS_CONNECTING;
- }
break;
}
if (m->msg_flags & MSG_DONTWAIT) {
@@ -651,9 +650,8 @@ static int send_packet(struct kiocb *iocb, struct socket *sock,
}
res = tipc_send(tport->ref, m->msg_iovlen, m->msg_iov);
- if (likely(res != -ELINKCONG)) {
+ if (likely(res != -ELINKCONG))
break;
- }
if (m->msg_flags & MSG_DONTWAIT) {
res = -EWOULDBLOCK;
break;
@@ -1418,9 +1416,8 @@ static int connect(struct socket *sock, struct sockaddr *dest, int destlen,
m.msg_name = dest;
m.msg_namelen = destlen;
res = send_msg(NULL, sock, &m, 0);
- if (res < 0) {
+ if (res < 0)
goto exit;
- }
/* Wait until an 'ACK' or 'RST' arrives, or a timeout occurs */
@@ -1442,11 +1439,10 @@ static int connect(struct socket *sock, struct sockaddr *dest, int destlen,
advance_rx_queue(sk);
}
} else {
- if (sock->state == SS_CONNECTED) {
+ if (sock->state == SS_CONNECTED)
res = -EISCONN;
- } else {
+ else
res = -ECONNREFUSED;
- }
}
} else {
if (res == 0)
--
1.7.3.3
^ permalink raw reply related
* [PATCH net-next 19/20] tipc: remove zeroing assignments to static global variables
From: Paul Gortmaker @ 2011-01-01 4:59 UTC (permalink / raw)
To: davem; +Cc: netdev, allan.stephens, Allan Stephens, Paul Gortmaker
In-Reply-To: <1293857975-30267-1-git-send-email-paul.gortmaker@windriver.com>
From: Allan Stephens <Allan.Stephens@windriver.com>
Cleans up TIPC's source code to eliminate the needless initialization
of static variables to zero.
These changes are purely cosmetic and do not alter the operation of TIPC
in any way.
Signed-off-by: Allan Stephens <Allan.Stephens@windriver.com>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>
---
net/tipc/bcast.c | 8 ++++----
net/tipc/bearer.c | 2 +-
net/tipc/eth_media.c | 2 +-
net/tipc/handler.c | 2 +-
net/tipc/name_distr.c | 2 +-
net/tipc/name_table.c | 2 +-
net/tipc/node.c | 2 +-
net/tipc/port.c | 4 ++--
net/tipc/ref.c | 2 +-
net/tipc/socket.c | 2 +-
net/tipc/subscr.c | 2 +-
11 files changed, 15 insertions(+), 15 deletions(-)
diff --git a/net/tipc/bcast.c b/net/tipc/bcast.c
index 9de3256..99a1469 100644
--- a/net/tipc/bcast.c
+++ b/net/tipc/bcast.c
@@ -103,9 +103,9 @@ struct bclink {
};
-static struct bcbearer *bcbearer = NULL;
-static struct bclink *bclink = NULL;
-static struct link *bcl = NULL;
+static struct bcbearer *bcbearer;
+static struct bclink *bclink;
+static struct link *bcl;
static DEFINE_SPINLOCK(bc_lock);
/* broadcast-capable node map */
@@ -425,7 +425,7 @@ int tipc_bclink_send_msg(struct sk_buff *buf)
void tipc_bclink_recv_pkt(struct sk_buff *buf)
{
#if (TIPC_BCAST_LOSS_RATE)
- static int rx_count = 0;
+ static int rx_count;
#endif
struct tipc_msg *msg = buf_msg(buf);
struct tipc_node *node = tipc_node_find(msg_prevnode(msg));
diff --git a/net/tipc/bearer.c b/net/tipc/bearer.c
index e9136f0..68e7290 100644
--- a/net/tipc/bearer.c
+++ b/net/tipc/bearer.c
@@ -42,7 +42,7 @@
#define MAX_ADDR_STR 32
static struct media media_list[MAX_MEDIA];
-static u32 media_count = 0;
+static u32 media_count;
struct bearer tipc_bearers[MAX_BEARERS];
diff --git a/net/tipc/eth_media.c b/net/tipc/eth_media.c
index 5dfe663..b69092e 100644
--- a/net/tipc/eth_media.c
+++ b/net/tipc/eth_media.c
@@ -56,7 +56,7 @@ struct eth_bearer {
};
static struct eth_bearer eth_bearers[MAX_ETH_BEARERS];
-static int eth_started = 0;
+static int eth_started;
static struct notifier_block notifier;
/**
diff --git a/net/tipc/handler.c b/net/tipc/handler.c
index 0c70010..274c98e 100644
--- a/net/tipc/handler.c
+++ b/net/tipc/handler.c
@@ -45,7 +45,7 @@ struct queue_item {
static struct kmem_cache *tipc_queue_item_cache;
static struct list_head signal_queue_head;
static DEFINE_SPINLOCK(qitem_lock);
-static int handler_enabled = 0;
+static int handler_enabled;
static void process_signal_queue(unsigned long dummy);
diff --git a/net/tipc/name_distr.c b/net/tipc/name_distr.c
index a6c989f..483c226 100644
--- a/net/tipc/name_distr.c
+++ b/net/tipc/name_distr.c
@@ -73,7 +73,7 @@ struct distr_item {
*/
static LIST_HEAD(publ_root);
-static u32 publ_cnt = 0;
+static u32 publ_cnt;
/**
* publ_to_item - add publication info to a publication message
diff --git a/net/tipc/name_table.c b/net/tipc/name_table.c
index 46b2f31..ea3e94e 100644
--- a/net/tipc/name_table.c
+++ b/net/tipc/name_table.c
@@ -104,7 +104,7 @@ struct name_table {
u32 local_publ_count;
};
-static struct name_table table = { NULL } ;
+static struct name_table table;
static atomic_t rsv_publ_ok = ATOMIC_INIT(0);
DEFINE_RWLOCK(tipc_nametbl_lock);
diff --git a/net/tipc/node.c b/net/tipc/node.c
index 2ed162f..3af53e3 100644
--- a/net/tipc/node.c
+++ b/net/tipc/node.c
@@ -44,7 +44,7 @@ static void node_established_contact(struct tipc_node *n_ptr);
static DEFINE_SPINLOCK(node_create_lock);
-u32 tipc_own_tag = 0;
+u32 tipc_own_tag;
/**
* tipc_node_create - create neighboring node
diff --git a/net/tipc/port.c b/net/tipc/port.c
index 78fcb75..c23f338 100644
--- a/net/tipc/port.c
+++ b/net/tipc/port.c
@@ -46,8 +46,8 @@
#define MAX_REJECT_SIZE 1024
-static struct sk_buff *msg_queue_head = NULL;
-static struct sk_buff *msg_queue_tail = NULL;
+static struct sk_buff *msg_queue_head;
+static struct sk_buff *msg_queue_tail;
DEFINE_SPINLOCK(tipc_port_list_lock);
static DEFINE_SPINLOCK(queue_lock);
diff --git a/net/tipc/ref.c b/net/tipc/ref.c
index 3974529..8311689 100644
--- a/net/tipc/ref.c
+++ b/net/tipc/ref.c
@@ -89,7 +89,7 @@ struct ref_table {
* have a reference value of 0 (although this is unlikely).
*/
-static struct ref_table tipc_ref_table = { NULL };
+static struct ref_table tipc_ref_table;
static DEFINE_RWLOCK(ref_table_lock);
diff --git a/net/tipc/socket.c b/net/tipc/socket.c
index 0895dec..18aad57 100644
--- a/net/tipc/socket.c
+++ b/net/tipc/socket.c
@@ -68,7 +68,7 @@ static const struct proto_ops msg_ops;
static struct proto tipc_proto;
-static int sockets_enabled = 0;
+static int sockets_enabled;
static atomic_t tipc_queue_size = ATOMIC_INIT(0);
diff --git a/net/tipc/subscr.c b/net/tipc/subscr.c
index acc30e1..ca04479 100644
--- a/net/tipc/subscr.c
+++ b/net/tipc/subscr.c
@@ -70,7 +70,7 @@ struct top_srv {
spinlock_t lock;
};
-static struct top_srv topsrv = { 0 };
+static struct top_srv topsrv;
/**
* htohl - convert value to endianness used by destination
--
1.7.3.3
^ permalink raw reply related
* [PATCH net-next 18/20] tipc: split variable assignments out of conditional expressions
From: Paul Gortmaker @ 2011-01-01 4:59 UTC (permalink / raw)
To: davem; +Cc: netdev, allan.stephens, Allan Stephens, Paul Gortmaker
In-Reply-To: <1293857975-30267-1-git-send-email-paul.gortmaker@windriver.com>
From: Allan Stephens <Allan.Stephens@windriver.com>
Cleans up TIPC's source code to eliminate assigning values to variables
within conditional expressions, improving code readability and reducing
warnings from various code checker tools.
These changes are purely cosmetic and do not alter the operation of TIPC
in any way.
Signed-off-by: Allan Stephens <Allan.Stephens@windriver.com>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>
---
net/tipc/bearer.c | 3 ++-
net/tipc/core.c | 33 +++++++++++++++++++++------------
net/tipc/link.c | 16 ++++++++++------
net/tipc/socket.c | 36 +++++++++++++++++++++++-------------
4 files changed, 56 insertions(+), 32 deletions(-)
diff --git a/net/tipc/bearer.c b/net/tipc/bearer.c
index 040f3ed..e9136f0 100644
--- a/net/tipc/bearer.c
+++ b/net/tipc/bearer.c
@@ -251,7 +251,8 @@ static int bearer_name_validate(const char *name,
/* ensure all component parts of bearer name are present */
media_name = name_copy;
- if ((if_name = strchr(media_name, ':')) == NULL)
+ if_name = strchr(media_name, ':');
+ if (if_name == NULL)
return 0;
*(if_name++) = 0;
media_len = if_name - media_name;
diff --git a/net/tipc/core.c b/net/tipc/core.c
index 60b85ec..e071579 100644
--- a/net/tipc/core.c
+++ b/net/tipc/core.c
@@ -115,10 +115,11 @@ int tipc_core_start_net(unsigned long addr)
{
int res;
- if ((res = tipc_net_start(addr)) ||
- (res = tipc_eth_media_start())) {
+ res = tipc_net_start(addr);
+ if (!res)
+ res = tipc_eth_media_start();
+ if (res)
tipc_core_stop_net();
- }
return res;
}
@@ -157,15 +158,22 @@ static int tipc_core_start(void)
get_random_bytes(&tipc_random, sizeof(tipc_random));
tipc_mode = TIPC_NODE_MODE;
- if ((res = tipc_handler_start()) ||
- (res = tipc_ref_table_init(tipc_max_ports, tipc_random)) ||
- (res = tipc_nametbl_init()) ||
- (res = tipc_k_signal((Handler)tipc_subscr_start, 0)) ||
- (res = tipc_k_signal((Handler)tipc_cfg_init, 0)) ||
- (res = tipc_netlink_start()) ||
- (res = tipc_socket_init())) {
+ res = tipc_handler_start();
+ if (!res)
+ res = tipc_ref_table_init(tipc_max_ports, tipc_random);
+ if (!res)
+ res = tipc_nametbl_init();
+ if (!res)
+ res = tipc_k_signal((Handler)tipc_subscr_start, 0);
+ if (!res)
+ res = tipc_k_signal((Handler)tipc_cfg_init, 0);
+ if (!res)
+ res = tipc_netlink_start();
+ if (!res)
+ res = tipc_socket_init();
+ if (res)
tipc_core_stop();
- }
+
return res;
}
@@ -188,7 +196,8 @@ static int __init tipc_init(void)
tipc_max_nodes = CONFIG_TIPC_NODES;
tipc_net_id = 4711;
- if ((res = tipc_core_start()))
+ res = tipc_core_start();
+ if (res)
err("Unable to start in single node mode\n");
else
info("Started in single node mode\n");
diff --git a/net/tipc/link.c b/net/tipc/link.c
index ef203a1..de9d491 100644
--- a/net/tipc/link.c
+++ b/net/tipc/link.c
@@ -187,14 +187,17 @@ static int link_name_validate(const char *name, struct link_name *name_parts)
/* ensure all component parts of link name are present */
addr_local = name_copy;
- if ((if_local = strchr(addr_local, ':')) == NULL)
+ if_local = strchr(addr_local, ':');
+ if (if_local == NULL)
return 0;
*(if_local++) = 0;
- if ((addr_peer = strchr(if_local, '-')) == NULL)
+ addr_peer = strchr(if_local, '-');
+ if (addr_peer == NULL)
return 0;
*(addr_peer++) = 0;
if_local_len = addr_peer - if_local;
- if ((if_peer = strchr(addr_peer, ':')) == NULL)
+ if_peer = strchr(addr_peer, ':');
+ if (if_peer == NULL)
return 0;
*(if_peer++) = 0;
if_peer_len = strlen(if_peer) + 1;
@@ -2044,8 +2047,8 @@ static void link_recv_proto_msg(struct link *l_ptr, struct sk_buff *buf)
strcpy((strrchr(l_ptr->name, ':') + 1), (char *)msg_data(msg));
- if ((msg_tol = msg_link_tolerance(msg)) &&
- (msg_tol > l_ptr->tolerance))
+ msg_tol = msg_link_tolerance(msg);
+ if (msg_tol > l_ptr->tolerance)
link_set_supervision_props(l_ptr, msg_tol);
if (msg_linkprio(msg) > l_ptr->priority)
@@ -2074,7 +2077,8 @@ static void link_recv_proto_msg(struct link *l_ptr, struct sk_buff *buf)
break;
case STATE_MSG:
- if ((msg_tol = msg_link_tolerance(msg)))
+ msg_tol = msg_link_tolerance(msg);
+ if (msg_tol)
link_set_supervision_props(l_ptr, msg_tol);
if (msg_linkprio(msg) &&
diff --git a/net/tipc/socket.c b/net/tipc/socket.c
index e9fc5df..0895dec 100644
--- a/net/tipc/socket.c
+++ b/net/tipc/socket.c
@@ -563,7 +563,8 @@ static int send_msg(struct kiocb *iocb, struct socket *sock,
do {
if (dest->addrtype == TIPC_ADDR_NAME) {
- if ((res = dest_name_check(dest, m)))
+ res = dest_name_check(dest, m);
+ if (res)
break;
res = tipc_send2name(tport->ref,
&dest->addr.name.name,
@@ -580,7 +581,8 @@ static int send_msg(struct kiocb *iocb, struct socket *sock,
res = -EOPNOTSUPP;
break;
}
- if ((res = dest_name_check(dest, m)))
+ res = dest_name_check(dest, m);
+ if (res)
break;
res = tipc_multicast(tport->ref,
&dest->addr.nameseq,
@@ -750,7 +752,8 @@ static int send_stream(struct kiocb *iocb, struct socket *sock,
bytes_to_send = curr_left;
my_iov.iov_base = curr_start;
my_iov.iov_len = bytes_to_send;
- if ((res = send_packet(NULL, sock, &my_msg, 0)) < 0) {
+ res = send_packet(NULL, sock, &my_msg, 0);
+ if (res < 0) {
if (bytes_sent)
res = bytes_sent;
goto exit;
@@ -845,12 +848,15 @@ static int anc_data_recv(struct msghdr *m, struct tipc_msg *msg,
if (unlikely(err)) {
anc_data[0] = err;
anc_data[1] = msg_data_sz(msg);
- if ((res = put_cmsg(m, SOL_TIPC, TIPC_ERRINFO, 8, anc_data)))
- return res;
- if (anc_data[1] &&
- (res = put_cmsg(m, SOL_TIPC, TIPC_RETDATA, anc_data[1],
- msg_data(msg))))
+ res = put_cmsg(m, SOL_TIPC, TIPC_ERRINFO, 8, anc_data);
+ if (res)
return res;
+ if (anc_data[1]) {
+ res = put_cmsg(m, SOL_TIPC, TIPC_RETDATA, anc_data[1],
+ msg_data(msg));
+ if (res)
+ return res;
+ }
}
/* Optionally capture message destination object */
@@ -878,9 +884,11 @@ static int anc_data_recv(struct msghdr *m, struct tipc_msg *msg,
default:
has_name = 0;
}
- if (has_name &&
- (res = put_cmsg(m, SOL_TIPC, TIPC_DESTNAME, 12, anc_data)))
- return res;
+ if (has_name) {
+ res = put_cmsg(m, SOL_TIPC, TIPC_DESTNAME, 12, anc_data);
+ if (res)
+ return res;
+ }
return 0;
}
@@ -1664,7 +1672,8 @@ static int setsockopt(struct socket *sock,
return -ENOPROTOOPT;
if (ol < sizeof(value))
return -EINVAL;
- if ((res = get_user(value, (u32 __user *)ov)))
+ res = get_user(value, (u32 __user *)ov);
+ if (res)
return res;
lock_sock(sk);
@@ -1722,7 +1731,8 @@ static int getsockopt(struct socket *sock,
return put_user(0, ol);
if (lvl != SOL_TIPC)
return -ENOPROTOOPT;
- if ((res = get_user(len, ol)))
+ res = get_user(len, ol);
+ if (res)
return res;
lock_sock(sk);
--
1.7.3.3
^ 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