user/sven/linux.git/net/caif/caif_dev.c, branch v3.2.32

caif: Fix access to freed pernet memory

2012-08-19T17:15:22Z

[ Upstream commit 96f80d123eff05c3cd4701463786b87952a6c3ac ] unregister_netdevice_notifier() must be called before unregister_pernet_subsys() to avoid accessing already freed pernet memory. This fixes the following oops when doing rmmod: Call Trace: [] caif_device_notify+0x4d/0x5a0 [caif] [] unregister_netdevice_notifier+0xb9/0x100 [] caif_device_exit+0x1c/0x250 [caif] [] sys_delete_module+0x1a4/0x300 [] ? trace_hardirqs_on_caller+0x15d/0x1e0 [] ? trace_hardirqs_on_thunk+0x3a/0x3 [] system_call_fastpath+0x1a/0x1f RIP [] caif_get+0x51/0xb0 [caif] Signed-off-by: Sjur Brændeland Acked-by: "Eric W. Biederman" Signed-off-by: David S. Miller Signed-off-by: Ben Hutchings

net caif: Register properly as a pernet subsystem.

2012-02-03T17:22:18Z

[ Upstream commit 8a8ee9aff6c3077dd9c2c7a77478e8ed362b96c6 ] caif is a subsystem and as such it needs to register with register_pernet_subsys instead of register_pernet_device. Among other problems using register_pernet_device was resulting in net_generic being called before the caif_net structure was allocated. Which has been causing net_generic to fail with either BUG_ON's or by return NULL pointers. A more ugly problem that could be caused is packets in flight why the subsystem is shutting down. To remove confusion also remove the cruft cause by inappropriately trying to fix this bug. With the aid of the previous patch I have tested this patch and confirmed that using register_pernet_subsys makes the failure go away as it should. Signed-off-by: Eric W. Biederman Acked-by: Sjur Brændeland Tested-by: Sasha Levin Signed-off-by: David S. Miller Signed-off-by: Greg Kroah-Hartman

net: Fix files explicitly needing to include module.h

2011-10-31T23:30:28Z

With calls to modular infrastructure, these files really needs the full module.h header. Call it out so some of the cleanups of implicit and unrequired includes elsewhere can be cleaned up. Signed-off-by: Paul Gortmaker

caif: Fix BUG() with network namespaces

2011-10-25T23:22:23Z

The caif code will register its own pernet_operations, and then register a netdevice_notifier. Each time the netdevice_notifier is triggered, it'll do some stuff... including a lookup of its own pernet stuff with net_generic(). If the net_generic() call ever returns NULL, the caif code will BUG(). That doesn't seem *so* unreasonable, I suppose — it does seem like it should never happen. However, it *does* happen. When we clone a network namespace, setup_net() runs through all the pernet_operations one at a time. It gets to loopback before it gets to caif. And loopback_net_init() registers a netdevice... while caif hasn't been initialised. So the caif netdevice notifier triggers, and immediately goes BUG(). We could imagine a complex and overengineered solution to this generic class of problems, but this patch takes the simple approach. It just makes caif_device_notify() *not* go looking for its pernet data structures if the device it's being notified about isn't a caif device in the first place. Cc: stable@kernel.org Signed-off-by: David Woodhouse Acked-by: Sjur Brændeland Signed-off-by: David S. Miller

caif: fix a potential NULL dereference

2011-09-16T21:40:34Z

Commit bd30ce4bc0b7 (caif: Use RCU instead of spin-lock in caif_dev.c) added a potential NULL dereference in case alloc_percpu() fails. caif_device_alloc() can also use GFP_KERNEL instead of GFP_ATOMIC. Signed-off-by: Eric Dumazet CC: Sjur Brændeland Acked-by: Sjur Brændeland Signed-off-by: David S. Miller

Remove redundant linux/version.h includes from net/

2011-06-21T23:03:17Z

It was suggested by "make versioncheck" that the follwing includes of linux/version.h are redundant: /home/jj/src/linux-2.6/net/caif/caif_dev.c: 14 linux/version.h not needed. /home/jj/src/linux-2.6/net/caif/chnl_net.c: 10 linux/version.h not needed. /home/jj/src/linux-2.6/net/ipv4/gre.c: 19 linux/version.h not needed. /home/jj/src/linux-2.6/net/netfilter/ipset/ip_set_core.c: 20 linux/version.h not needed. /home/jj/src/linux-2.6/net/netfilter/xt_set.c: 16 linux/version.h not needed. and it seems that it is right. Beyond manually inspecting the source files I also did a few build tests with various configs to confirm that including the header in those files is indeed not needed. Here's a patch to remove the pointless includes. Signed-off-by: Jesper Juhl Acked-by: Jozsef Kadlecsik Signed-off-by: David S. Miller

caif: Plug memory leak for checksum error

2011-05-23T00:11:49Z

In case of checksum error, the framing layer returns -EILSEQ, but does not free the packet. Plug this hole by freeing the packet if -EILSEQ is returned. Signed-off-by: Sjur Brændeland Signed-off-by: David S. Miller

caif: Handle dev_queue_xmit errors.

2011-05-15T21:45:56Z

Do proper handling of dev_queue_xmit errors in order to avoid double free of skb and leaks in error conditions. In cfctrl pending requests are removed when CAIF Link layer goes down. Signed-off-by: Sjur Brændeland Signed-off-by: David S. Miller

caif: prepare support for namespaces

2011-05-15T21:45:55Z

Use struct net to reference CAIF configuration object instead of static variables. Refactor functions caif_connect_client, caif_disconnect_client and squach files cfcnfg.c and caif_config_utils. Signed-off-by: Sjur Brændeland Signed-off-by: David S. Miller

caif: Use RCU instead of spin-lock in caif_dev.c

2011-05-15T21:45:54Z

RCU read_lock and refcount is used to protect in-flight packets. Use RCU and counters to manage freeing lower part of the CAIF stack if CAIF-link layer is removed. Old solution based on delaying removal of device is removed. When CAIF link layer goes down the use of CAIF link layer is disabled (by calling caif_set_phy_state()), but removal and freeing of the lower part of the CAIF stack is done when Link layer is unregistered. Signed-off-by: Sjur Brændeland Signed-off-by: David S. Miller