linux/net, branch v3.0.13

cfg80211: amend regulatory NULL dereference fix

2011-12-09T16:52:45Z

commit 0bac71af6e66dc798bf07d0c0dd14ee5503362f9 upstream. Johannes' patch for "cfg80211: fix regulatory NULL dereference" broke user regulaotry hints and it did not address the fact that last_request was left populated even if the previous regulatory hint was stale due to the wiphy disappearing. Fix user reguluatory hints by only bailing out if for those regulatory hints where a request_wiphy is expected. The stale last_request considerations are addressed through the previous fixes on last_request where we reset the last_request to a static world regdom request upon reset_regdomains(). In this case though we further enhance the effect by simply restoring reguluatory settings completely. Cc: Johannes Berg Signed-off-by: Luis R. Rodriguez Reviewed-by: Johannes Berg Signed-off-by: John W. Linville Signed-off-by: Greg Kroah-Hartman

cfg80211: fix race on init and driver registration

2011-12-09T16:52:45Z

commit a042994dd377d86bff9446ee76151ceb6267c9ba upstream. There is a theoretical race that if hit will trigger a crash. The race is between when we issue the first regulatory hint, regulatory_hint_core(), gets processed by the workqueue and between when the first device gets registered to the wireless core. This is not easy to reproduce but it was easy to do so through the regulatory simulator I have been working on. This is a port of the fix I implemented there [1]. [1] https://github.com/mcgrof/regsim/commit/a246ccf81f059cb662eee288aa13100f631e4cc8 Cc: Johannes Berg Signed-off-by: Luis R. Rodriguez Signed-off-by: John W. Linville Signed-off-by: Greg Kroah-Hartman

mac80211: fix race between the AGG SM and the Tx data path

2011-12-09T16:52:33Z

commit 2a1e0fd175dcfd72096ba9291d31e3b1b5342e60 upstream. When a packet is supposed to sent be as an a-MPDU, mac80211 sets IEEE80211_TX_CTL_AMPDU to let the driver know. On the other hand, mac80211 configures the driver for aggregration with the ampdu_action callback. There is race between these two mechanisms since the following scenario can occur when the BA agreement is torn down: Tx softIRQ drv configuration ========== ================= check OPERATIONAL bit Set the TX_CTL_AMPDU bit in the packet clear OPERATIONAL bit stop Tx AGG Pass Tx packet to the driver. In that case the driver would get a packet with TX_CTL_AMPDU set although it has already been notified that the BA session has been torn down. To fix this, we need to synchronize all the Qdisc activity after we cleared the OPERATIONAL bit. After that step, all the following packets will be buffered until the driver reports it is ready to get new packets for this RA / TID. This buffering allows not to run into another race that would send packets with TX_CTL_AMPDU unset while the driver hasn't been requested to tear down the BA session yet. This race occurs in practice and iwlwifi complains with a WARN_ON when it happens. Signed-off-by: Emmanuel Grumbach Reviewed-by: Johannes Berg Signed-off-by: John W. Linville Signed-off-by: Greg Kroah-Hartman

mac80211: don't stop a single aggregation session twice

2011-12-09T16:52:33Z

commit 24f50a9d165745fd0701c6e089d35f58a229ea69 upstream. Nikolay noticed (by code review) that mac80211 can attempt to stop an aggregation session while it is already being stopped. So to fix it, check whether stop is already being done and bail out if so. Also move setting the STOPPING state into the lock so things are properly atomic. Reported-by: Nikolay Martynov Signed-off-by: Johannes Berg Signed-off-by: John W. Linville Signed-off-by: Greg Kroah-Hartman

cfg80211: fix regulatory NULL dereference

2011-12-09T16:52:32Z

commit de3584bd62d87b4c250129fbc46ca52c80330add upstream. By the time userspace returns with a response to the regulatory domain request, the wiphy causing the request might have gone away. If this is so, reject the update but mark the request as having been processed anyway. Cc: Luis R. Rodriguez Signed-off-by: Johannes Berg Signed-off-by: John W. Linville Signed-off-by: Greg Kroah-Hartman

nl80211: fix MAC address validation

2011-12-09T16:52:31Z

commit e007b857e88097c96c45620bf3b04a4e309053d1 upstream. MAC addresses have a fixed length. The current policy allows passing < ETH_ALEN bytes, which might result in reading beyond the buffer. Signed-off-by: Eliad Peller Signed-off-by: John W. Linville Signed-off-by: Greg Kroah-Hartman

SUNRPC: Ensure we return EAGAIN in xs_nospace if congestion is cleared

2011-12-09T16:52:27Z

commit 24ca9a847791fd53d9b217330b15f3c285827a18 upstream. By returning '0' instead of 'EAGAIN' when the tests in xs_nospace() fail to find evidence of socket congestion, we are making the RPC engine believe that the message was incorrectly sent and so it disconnects the socket instead of just retrying. The bug appears to have been introduced by commit 5e3771ce2d6a69e10fcc870cdf226d121d868491 (SUNRPC: Ensure that xs_nospace return values are propagated). Reported-by: Andrew Cooper Signed-off-by: Trond Myklebust Tested-by: Andrew Cooper Signed-off-by: Greg Kroah-Hartman

ip6_tunnel: copy parms.name after register_netdevice

2011-11-26T17:09:55Z

commit 731abb9cb27aef6013ce60808a04e04a545f3f4e upstream. Commit 1c5cae815d removed an explicit call to dev_alloc_name in ip6_tnl_create because register_netdevice will now create a valid name. This works for the net_device itself. However the tunnel keeps a copy of the name in the parms structure for the ip6_tnl associated with the tunnel. parms.name is set by copying the net_device name in ip6_tnl_dev_init_gen. That function is called from ip6_tnl_dev_init in ip6_tnl_create, but it is done before register_netdevice is called so the name is set to a bogus value in the parms.name structure. This shows up if you do a simple tunnel add, followed by a tunnel show: [root@localhost ~]# ip -6 tunnel add remote fec0::100 local fec0::200 [root@localhost ~]# ip -6 tunnel show ip6tnl0: ipv6/ipv6 remote :: local :: encaplimit 0 hoplimit 0 tclass 0x00 flowlabel 0x00000 (flowinfo 0x00000000) ip6tnl%d: ipv6/ipv6 remote fec0::100 local fec0::200 encaplimit 4 hoplimit 64 tclass 0x00 flowlabel 0x00000 (flowinfo 0x00000000) [root@localhost ~]# Fix this by moving the strcpy out of ip6_tnl_dev_init_gen, and calling it after register_netdevice has successfully returned. Signed-off-by: Josh Boyer Signed-off-by: David S. Miller Signed-off-by: Greg Kroah-Hartman

cfg80211: fix bug on regulatory core exit on access to last_request

2011-11-26T17:09:55Z

commit 58ebacc66bd11be2327edcefc79de94bd6f5bb4a upstream. Commit 4d9d88d1 by Scott James Remnant added the .uevent() callback for the regulatory device used during the platform device registration. The change was done to account for queuing up udev change requests through udevadm triggers. The change also meant that upon regulatory core exit we will now send a uevent() but the uevent() callback, reg_device_uevent(), also accessed last_request. Right before commiting device suicide we free'd last_request but never set it to NULL so platform_device_unregister() would lead to bogus kernel paging request. Fix this and also simply supress uevents right before we commit suicide as they are pointless. This fix is required for kernels >= v2.6.39 $ git describe --contains 4d9d88d1 v2.6.39-rc1~468^2~25^2^2~21 The impact of not having this present is that a bogus paging access may occur (only read) upon cfg80211 unload time. You may also get this BUG complaint below. Although Johannes could not reproduce the issue this fix is theoretically correct. mac80211_hwsim: unregister radios mac80211_hwsim: closing netlink BUG: unable to handle kernel paging request at ffff88001a06b5ab IP: [] reg_device_uevent+0x1a/0x50 [cfg80211] PGD 1836063 PUD 183a063 PMD 1ffcb067 PTE 1a06b160 Oops: 0000 [#1] PREEMPT SMP DEBUG_PAGEALLOC CPU 0 Modules linked in: cfg80211(-) [last unloaded: mac80211] Pid: 2279, comm: rmmod Tainted: G W 3.1.0-wl+ #663 Bochs Bochs RIP: 0010:[] [] reg_device_uevent+0x1a/0x50 [cfg80211] RSP: 0000:ffff88001c5f9d58 EFLAGS: 00010286 RAX: 0000000000000000 RBX: ffff88001d2eda88 RCX: ffff88001c7468fc RDX: ffff88001a06b5a0 RSI: ffff88001c7467b0 RDI: ffff88001c7467b0 RBP: ffff88001c5f9d58 R08: 000000000000ffff R09: 000000000000ffff R10: 0000000000000000 R11: 0000000000000001 R12: ffff88001c7467b0 R13: ffff88001d2eda78 R14: ffffffff8164a840 R15: 0000000000000001 FS: 00007f8a91d8a6e0(0000) GS:ffff88001fc00000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 000000008005003b CR2: ffff88001a06b5ab CR3: 000000001c62e000 CR4: 00000000000006f0 DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000 DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400 Process rmmod (pid: 2279, threadinfo ffff88001c5f8000, task ffff88000023c780) Stack: ffff88001c5f9d98 ffffffff812ff7e5 ffffffff8176ab3d ffff88001c7468c2 000000000000ffff ffff88001d2eda88 ffff88001c7467b0 ffff880000114820 ffff88001c5f9e38 ffffffff81241dc7 ffff88001c5f9db8 ffffffff81040189 Call Trace: [] dev_uevent+0xc5/0x170 [] kobject_uevent_env+0x1f7/0x490 [] ? sub_preempt_count+0x29/0x60 [] ? _raw_spin_unlock_irqrestore+0x4a/0x90 [] ? devres_release_all+0x27/0x60 [] kobject_uevent+0xb/0x10 [] device_del+0x157/0x1b0 [] platform_device_del+0x1d/0x90 [] platform_device_unregister+0x16/0x30 [] regulatory_exit+0x5d/0x180 [cfg80211] [] cfg80211_exit+0x2b/0x45 [cfg80211] [] sys_delete_module+0x16c/0x220 [] ? trace_hardirqs_on_caller+0x7e/0x120 [] system_call_fastpath+0x16/0x1b Code: RIP [] reg_device_uevent+0x1a/0x50 [cfg80211] RSP CR2: ffff88001a06b5ab ---[ end trace 147c5099a411e8c0 ]--- Reported-by: Johannes Berg Cc: Scott James Remnant Signed-off-by: Luis R. Rodriguez Signed-off-by: John W. Linville Signed-off-by: Greg Kroah-Hartman

nl80211: fix HT capability attribute validation

2011-11-26T17:09:55Z

commit 6c7394197af90f6a332180e33f5d025d3037d883 upstream. Since the NL80211_ATTR_HT_CAPABILITY attribute is used as a struct, it needs a minimum, not maximum length. Enforce that properly. Not doing so could potentially lead to reading after the buffer. Signed-off-by: Johannes Berg Signed-off-by: John W. Linville Signed-off-by: Greg Kroah-Hartman