platform/kernel/linux-arm64.git
11 years agoMerge branch 'master' of git://git.kernel.org/pub/scm/linux/kernel/git/pablo/nf into...
David S. Miller [Fri, 12 Apr 2013 18:26:39 +0000 (14:26 -0400)]
Merge branch 'master' of git://git./linux/kernel/git/pablo/nf into netfilter

Pablo Neira Ayuso says:

====================
The following patchset contains late netfilter fixes for your net
tree, they are:

* Don't drop segmented TCP packets in the SIP helper, we've got reports
  from users that this was breaking communications when the SIP phone
  messages are larger than the MTU, from Patrick McHardy.

* Fix refcount leak in the ipset list set, from Jozsef Kadlecsik.

* On hash set resizing, the nomatch flag was lost, thus entirely inverting
  the logic of the set matching, from Jozsef Kadlecsik.

* Fix crash on NAT modules removal. Timer expiration may race with the
  module cleanup exit path while deleting conntracks, from Florian
  Westphal.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agonetfilter: nf_nat: fix race when unloading protocol modules
Florian Westphal [Thu, 11 Apr 2013 04:22:39 +0000 (04:22 +0000)]
netfilter: nf_nat: fix race when unloading protocol modules

following oops was reported:
RIP: 0010:[<ffffffffa03227f2>]  [<ffffffffa03227f2>] nf_nat_cleanup_conntrack+0x42/0x70 [nf_nat]
RSP: 0018:ffff880202c63d40  EFLAGS: 00010246
RAX: 0000000000000000 RBX: ffff8801ac7bec28 RCX: ffff8801d0eedbe0
RDX: dead000000200200 RSI: 0000000000000011 RDI: ffffffffa03265b8
[..]
Call Trace:
 [..]
 [<ffffffffa02febed>] destroy_conntrack+0xbd/0x110 [nf_conntrack]

Happens when a conntrack timeout expires right after first part
of the nat cleanup has completed (bysrc hash removal), but before
part 2 has completed (re-initialization of nat area).

[ destroy callback tries to delete bysrc again ]

Patrick suggested to just remove the affected conntracks -- the
connections won't work properly anyway without nat transformation.

So, lets do that.

Reported-by: CAI Qian <caiqian@redhat.com>
Cc: Patrick McHardy <kaber@trash.net>
Signed-off-by: Florian Westphal <fw@strlen.de>
Acked-by: Patrick McHardy <kaber@trash.net>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
11 years agotcp: Reallocate headroom if it would overflow csum_start
Thomas Graf [Thu, 11 Apr 2013 10:57:18 +0000 (10:57 +0000)]
tcp: Reallocate headroom if it would overflow csum_start

If a TCP retransmission gets partially ACKed and collapsed multiple
times it is possible for the headroom to grow beyond 64K which will
overflow the 16bit skb->csum_start which is based on the start of
the headroom. It has been observed rarely in the wild with IPoIB due
to the 64K MTU.

Verify if the acking and collapsing resulted in a headroom exceeding
what csum_start can cover and reallocate the headroom if so.

A big thank you to Jim Foraker <foraker1@llnl.gov> and the team at
LLNL for helping out with the investigation and testing.

Reported-by: Jim Foraker <foraker1@llnl.gov>
Signed-off-by: Thomas Graf <tgraf@suug.ch>
Acked-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agostmmac: prevent interrupt loop with MMC RX IPC Counter
Christian Ruppert [Tue, 9 Apr 2013 21:56:04 +0000 (21:56 +0000)]
stmmac: prevent interrupt loop with MMC RX IPC Counter

If the DesignWare MAC is synthesised with MMC RX IPC Counter, an unmanaged
and unacknowledged interrupt is generated after some time of operation.

This patch masks the undesired interrupts.

Signed-off-by: Christian Ruppert <christian.ruppert@abilis.com>
Acked-by: Giuseppe Cavallaro <peppe.cavallaro@st.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agobonding: IFF_BONDING is not stripped on enslave failure
nikolay@redhat.com [Thu, 11 Apr 2013 09:18:56 +0000 (09:18 +0000)]
bonding: IFF_BONDING is not stripped on enslave failure

While enslaving a new device and after IFF_BONDING flag is set, in case
of failure it is not stripped from the device's priv_flags while
cleaning up, which could lead to other problems.
Cleaning at err_close because the flag is set after dev_open().

v2: no change

Signed-off-by: Nikolay Aleksandrov <nikolay@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agobonding: fix netdev event NULL pointer dereference
nikolay@redhat.com [Thu, 11 Apr 2013 09:18:55 +0000 (09:18 +0000)]
bonding: fix netdev event NULL pointer dereference

In commit 471cb5a33dcbd7c529684a2ac7ba4451414ee4a7 ("bonding: remove
usage of dev->master") a bug was introduced which causes a NULL pointer
dereference. If a bond device is in mode 6 (ALB) and a slave is added
it will dereference a NULL pointer in bond_slave_netdev_event().
This is because in bond_enslave we have bond_alb_init_slave() which
changes the MAC address of the slave and causes a NETDEV_CHANGEADDR.
Then we have in bond_slave_netdev_event():
        struct slave *slave = bond_slave_get_rtnl(slave_dev);
        struct bonding *bond = slave->bond;
bond_slave_get_rtnl() dereferences slave_dev->rx_handler_data which at
that time is NULL since netdev_rx_handler_register() is called later.

This is fixed by checking if slave is NULL before dereferencing it.

v2: Comment style changed.

Signed-off-by: Nikolay Aleksandrov <nikolay@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agotcp: incoming connections might use wrong route under synflood
Dmitry Popov [Thu, 11 Apr 2013 08:55:07 +0000 (08:55 +0000)]
tcp: incoming connections might use wrong route under synflood

There is a bug in cookie_v4_check (net/ipv4/syncookies.c):
flowi4_init_output(&fl4, 0, sk->sk_mark, RT_CONN_FLAGS(sk),
   RT_SCOPE_UNIVERSE, IPPROTO_TCP,
   inet_sk_flowi_flags(sk),
   (opt && opt->srr) ? opt->faddr : ireq->rmt_addr,
   ireq->loc_addr, th->source, th->dest);

Here we do not respect sk->sk_bound_dev_if, therefore wrong dst_entry may be
taken. This dst_entry is used by new socket (get_cookie_sock ->
tcp_v4_syn_recv_sock), so its packets may take the wrong path.

Signed-off-by: Dmitry Popov <dp@highloadlab.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net
Linus Torvalds [Wed, 10 Apr 2013 21:15:27 +0000 (14:15 -0700)]
Merge git://git./linux/kernel/git/davem/net

Pull networking fixes from David Miller:

 1) cfg80211_conn_scan() must be called with the sched_scan_mutex, fix
    from Artem Savkov.

 2) Fix regression in TCP ICMPv6 processing, we do not want to treat
    redirects as socket errors, from Christoph Paasch.

 3) Fix several recvmsg() msg_name kernel memory leaks into userspace,
    in ATM, AX25, Bluetooth, CAIF, IRDA, s390 IUCV, L2TP, LLC, Netrom,
    NFC, Rose, TIPC, and VSOCK.  From Mathias Krause and Wei Yongjun.

 4) Fix AF_IUCV handling of segmented SKBs in recvmsg(), from Ursula
    Braun and Eric Dumazet.

 5) CAN gw.c code does kfree() on SLAB cache memory, use
    kmem_cache_free() instead.  Fix from Wei Yongjun.

 6) Fix LSM regression on TCP SYN/ACKs, some LSMs such as SELINUX want
    an skb->sk socket context available for these packets, but nothing
    else requires it.  From Eric Dumazet and Paul Moore.

 7) Fix ipv4 address lifetime processing so that we don't perform
    sleepable acts inside of rcu_read_lock() sections, do them in an
    rtnl_lock() section instead.  From Jiri Pirko.

 8) mvneta driver accidently sets HW features after device registry, it
    should do so beforehand.  Fix from Willy Tarreau.

 9) Fix bonding unload races more correctly, from Nikolay Aleksandrov
    and Veaceslav Falico.

10) rtnl_dump_ifinfo() and rtnl_calcit() invoke nlmsg_parse() with wrong
    header size argument.  Fix from Michael Riesch.

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net: (44 commits)
  lsm: add the missing documentation for the security_skb_owned_by() hook
  bnx2x: Prevent null pointer dereference in AFEX mode
  e100: Add dma mapping error check
  selinux: add a skb_owned_by() hook
  can: gw: use kmem_cache_free() instead of kfree()
  netrom: fix invalid use of sizeof in nr_recvmsg()
  qeth: fix qeth_wait_for_threads() deadlock for OSN devices
  af_iucv: fix recvmsg by replacing skb_pull() function
  rtnetlink: Call nlmsg_parse() with correct header length
  bonding: fix bonding_masters race condition in bond unloading
  Revert "bonding: remove sysfs before removing devices"
  net: mvneta: enable features before registering the driver
  hyperv: Fix RNDIS send_completion code path
  hyperv: Fix a kernel warning from netvsc_linkstatus_callback()
  net: ipv4: fix schedule while atomic bug in check_lifetime()
  net: ipv4: reset check_lifetime_work after changing lifetime
  bnx2x: Fix KR2 rapid link flap
  sctp: remove 'sridhar' from maintainers list
  VSOCK: Fix missing msg_namelen update in vsock_stream_recvmsg()
  VSOCK: vmci - fix possible info leak in vmci_transport_dgram_dequeue()
  ...

11 years agoMerge tag 'for-linus' of git://linux-c6x.org/git/projects/linux-c6x-upstreaming
Linus Torvalds [Wed, 10 Apr 2013 21:14:27 +0000 (14:14 -0700)]
Merge tag 'for-linus' of git://linux-c6x.org/git/projects/linux-c6x-upstreaming

Pull C6X fix from Mark Salter.

Final (?) fix from the barrier discussion.

* tag 'for-linus' of git://linux-c6x.org/git/projects/linux-c6x-upstreaming:
  add memory barrier to arch_local_irq_restore

11 years agolsm: add the missing documentation for the security_skb_owned_by() hook
Paul Moore [Wed, 10 Apr 2013 19:34:14 +0000 (15:34 -0400)]
lsm: add the missing documentation for the security_skb_owned_by() hook

Unfortunately we didn't catch the missing comments earlier when the
patch was merged.

Signed-off-by: Paul Moore <pmoore@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agobnx2x: Prevent null pointer dereference in AFEX mode
Yuval Mintz [Wed, 10 Apr 2013 10:34:39 +0000 (13:34 +0300)]
bnx2x: Prevent null pointer dereference in AFEX mode

The cnic module is responsible for initializing various bnx2x structs
via callbacks provided by the bnx2x module.
One such struct is the queue object for the FCoE queue.

If a device is working in AFEX mode and its configuration allows FCoE yet
the cnic module is not loaded, it's very likely a null pointer dereference
will occur, as the bnx2x will erroneously access the FCoE's queue object.

Prevent said access until cnic properly registers itself.

Signed-off-by: Yuval Mintz <yuvalmin@broadcom.com>
Signed-off-by: Ariel Elior <ariele@broadcom.com>
Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoe100: Add dma mapping error check
Neil Horman [Tue, 9 Apr 2013 23:19:00 +0000 (23:19 +0000)]
e100: Add dma mapping error check

e100 uses pci_map_single, but fails to check for a dma mapping error after its
use, resulting in a stack trace:

[   46.656594] ------------[ cut here ]------------
[   46.657004] WARNING: at lib/dma-debug.c:933 check_unmap+0x47b/0x950()
[   46.657004] Hardware name: To Be Filled By O.E.M.
[   46.657004] e100 0000:00:0e.0: DMA-API: device driver failed to check map
error[device address=0x000000007a4540fa] [size=90 bytes] [mapped as single]
[   46.657004] Modules linked in:
[   46.657004]  w83627hf hwmon_vid snd_via82xx ppdev snd_ac97_codec ac97_bus
snd_seq snd_pcm snd_mpu401 snd_mpu401_uart ns558 snd_rawmidi gameport parport_pc
e100 snd_seq_device parport snd_page_alloc snd_timer snd soundcore skge shpchp
k8temp mii edac_core i2c_viapro edac_mce_amd nfsd auth_rpcgss nfs_acl lockd
sunrpc binfmt_misc uinput ata_generic pata_acpi radeon i2c_algo_bit
drm_kms_helper ttm firewire_ohci drm firewire_core pata_via sata_via i2c_core
sata_promise crc_itu_t
[   46.657004] Pid: 792, comm: ip Not tainted 3.8.0-0.rc6.git0.1.fc19.x86_64 #1
[   46.657004] Call Trace:
[   46.657004]  <IRQ>  [<ffffffff81065ed0>] warn_slowpath_common+0x70/0xa0
[   46.657004]  [<ffffffff81065f4c>] warn_slowpath_fmt+0x4c/0x50
[   46.657004]  [<ffffffff81364cfb>] check_unmap+0x47b/0x950
[   46.657004]  [<ffffffff8136522f>] debug_dma_unmap_page+0x5f/0x70
[   46.657004]  [<ffffffffa030f0f0>] ? e100_tx_clean+0x30/0x210 [e100]
[   46.657004]  [<ffffffffa030f1a8>] e100_tx_clean+0xe8/0x210 [e100]
[   46.657004]  [<ffffffffa030fc6f>] e100_poll+0x56f/0x6c0 [e100]
[   46.657004]  [<ffffffff8159dce1>] ? net_rx_action+0xa1/0x370
[   46.657004]  [<ffffffff8159ddb2>] net_rx_action+0x172/0x370
[   46.657004]  [<ffffffff810703bf>] __do_softirq+0xef/0x3d0
[   46.657004]  [<ffffffff816e4ebc>] call_softirq+0x1c/0x30
[   46.657004]  [<ffffffff8101c485>] do_softirq+0x85/0xc0
[   46.657004]  [<ffffffff81070885>] irq_exit+0xd5/0xe0
[   46.657004]  [<ffffffff816e5756>] do_IRQ+0x56/0xc0
[   46.657004]  [<ffffffff816dacb2>] common_interrupt+0x72/0x72
[   46.657004]  <EOI>  [<ffffffff816da1eb>] ?
_raw_spin_unlock_irqrestore+0x3b/0x70
[   46.657004]  [<ffffffff816d124d>] __slab_free+0x58/0x38b
[   46.657004]  [<ffffffff81214424>] ? fsnotify_clear_marks_by_inode+0x34/0x120
[   46.657004]  [<ffffffff811b0417>] ? kmem_cache_free+0x97/0x320
[   46.657004]  [<ffffffff8157fc14>] ? sock_destroy_inode+0x34/0x40
[   46.657004]  [<ffffffff8157fc14>] ? sock_destroy_inode+0x34/0x40
[   46.657004]  [<ffffffff811b0692>] kmem_cache_free+0x312/0x320
[   46.657004]  [<ffffffff8157fc14>] sock_destroy_inode+0x34/0x40
[   46.657004]  [<ffffffff811e8c28>] destroy_inode+0x38/0x60
[   46.657004]  [<ffffffff811e8d5e>] evict+0x10e/0x1a0
[   46.657004]  [<ffffffff811e9605>] iput+0xf5/0x180
[   46.657004]  [<ffffffff811e4338>] dput+0x248/0x310
[   46.657004]  [<ffffffff811ce0e1>] __fput+0x171/0x240
[   46.657004]  [<ffffffff811ce26e>] ____fput+0xe/0x10
[   46.657004]  [<ffffffff8108d54c>] task_work_run+0xac/0xe0
[   46.657004]  [<ffffffff8106c6ed>] do_exit+0x26d/0xc30
[   46.657004]  [<ffffffff8109eccc>] ? finish_task_switch+0x7c/0x120
[   46.657004]  [<ffffffff816dad58>] ? retint_swapgs+0x13/0x1b
[   46.657004]  [<ffffffff8106d139>] do_group_exit+0x49/0xc0
[   46.657004]  [<ffffffff8106d1c4>] sys_exit_group+0x14/0x20
[   46.657004]  [<ffffffff816e3b19>] system_call_fastpath+0x16/0x1b
[   46.657004] ---[ end trace 4468c44e2156e7d1 ]---
[   46.657004] Mapped at:
[   46.657004]  [<ffffffff813663d1>] debug_dma_map_page+0x91/0x140
[   46.657004]  [<ffffffffa030e8eb>] e100_xmit_prepare+0x12b/0x1c0 [e100]
[   46.657004]  [<ffffffffa030c924>] e100_exec_cb+0x84/0x140 [e100]
[   46.657004]  [<ffffffffa030e56a>] e100_xmit_frame+0x3a/0x190 [e100]
[   46.657004]  [<ffffffff8159ee89>] dev_hard_start_xmit+0x259/0x6c0

Easy fix, modify the cb paramter to e100_exec_cb to return an error, and do the
dma_mapping_error check in the obvious place

This was reported previously here:
http://article.gmane.org/gmane.linux.network/257893

But nobody stepped up and fixed it.

CC: Josh Boyer <jwboyer@redhat.com>
CC: e1000-devel@lists.sourceforge.net
Signed-off-by: Neil Horman <nhorman@tuxdriver.com>
Reported-by: Michal Jaegermann <michal@harddata.com>
Tested-by: Aaron Brown <aaron.f.brown@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoMerge tag 'nfs-for-3.9-5' of git://git.linux-nfs.org/projects/trondmy/linux-nfs
Linus Torvalds [Wed, 10 Apr 2013 17:26:49 +0000 (10:26 -0700)]
Merge tag 'nfs-for-3.9-5' of git://git.linux-nfs.org/projects/trondmy/linux-nfs

Pull another nfs fixlet from Trond Myklebust:
 "I suddenly noticed that a one-line issue that I _thought_ I had fixed
  with the nfs41_walk_client_list patch was apparently still there in
  the pull request I sent earlier today.  I'm very sorry for not
  catching that in time.

   - Fix a brain fart in nfs41_walk_client_list"

* tag 'nfs-for-3.9-5' of git://git.linux-nfs.org/projects/trondmy/linux-nfs:
  NFSv4: Doh! Typo in the fix to nfs41_walk_client_list

11 years agoNFSv4: Doh! Typo in the fix to nfs41_walk_client_list
Trond Myklebust [Wed, 10 Apr 2013 16:44:18 +0000 (12:44 -0400)]
NFSv4: Doh! Typo in the fix to nfs41_walk_client_list

Make sure that we set the status to 0 on success. Missed in testing
because it never appears when doing multiple mounts to _different_
servers.

Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
Cc: <stable@vger.kernel.org> # 3.7.x: 7b1f1fd: NFSv4/4.1: Fix bugs in nfs4[01]_walk_client_list
11 years agoMerge tag 'nfs-for-3.9-4' of git://git.linux-nfs.org/projects/trondmy/linux-nfs
Linus Torvalds [Wed, 10 Apr 2013 16:00:51 +0000 (09:00 -0700)]
Merge tag 'nfs-for-3.9-4' of git://git.linux-nfs.org/projects/trondmy/linux-nfs

Pull NFS client bugfixes from Trond Myklebust:
 - fix for memory corruption issues in nfs4[01]_walk_client_list (stable)
 - fix for an Oopsable bug in rpc_clone_client (stable)
 - another state manager deadlock in the NFSv4 open code
 - memory leaks in nfs4_discover_server_trunking and rpc_new_client

* tag 'nfs-for-3.9-4' of git://git.linux-nfs.org/projects/trondmy/linux-nfs:
  NFSv4: Fix another potential state manager deadlock
  SUNRPC: Fix a potential memory leak in rpc_new_client
  NFSv4/4.1: Fix bugs in nfs4[01]_walk_client_list
  NFSv4: Fix a memory leak in nfs4_discover_server_trunking
  SUNRPC: Remove extra xprt_put()

11 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/herbert/crypto-2.6
Linus Torvalds [Wed, 10 Apr 2013 03:19:50 +0000 (20:19 -0700)]
Merge git://git./linux/kernel/git/herbert/crypto-2.6

Pull crypto fixes from Herbert Xu:
 "This fixes a GCM bug that breaks IPsec and a compile problem in
  ux500."

* git://git.kernel.org/pub/scm/linux/kernel/git/herbert/crypto-2.6:
  crypto: ux500 - add missing comma
  crypto: gcm - fix assumption that assoc has one segment

11 years agoMerge branch 'drm-fixes' of git://people.freedesktop.org/~airlied/linux
Linus Torvalds [Wed, 10 Apr 2013 00:41:26 +0000 (17:41 -0700)]
Merge branch 'drm-fixes' of git://people.freedesktop.org/~airlied/linux

Pull drm fixes from Dave Airlie:
 "Just a spare semicolon in nouveau that caused some issues, and an
  mgag200 fix"

* 'drm-fixes' of git://people.freedesktop.org/~airlied/linux:
  drm/mgag200: Index 24 in extended CRTC registers is 24 in hex, not decimal.
  drm/nouveau: fix unconditional return waiting on memory

11 years agodrm/mgag200: Index 24 in extended CRTC registers is 24 in hex, not decimal.
Christopher Harvey [Fri, 5 Apr 2013 14:51:15 +0000 (10:51 -0400)]
drm/mgag200: Index 24 in extended CRTC registers is 24 in hex, not decimal.

This change properly enables the "requester" in G200ER cards that is
responsible for getting pixels out of memory and clocking them out to
the screen.

Signed-off-by: Christopher Harvey <charvey@matrox.com>
Cc: stable@vger.kernel.org
Signed-off-by: Dave Airlie <airlied@redhat.com>
11 years agoadd memory barrier to arch_local_irq_restore
Mark Salter [Tue, 9 Apr 2013 19:35:46 +0000 (15:35 -0400)]
add memory barrier to arch_local_irq_restore

arch_local_irq_save() and friends are required to act as compiler
memory barriers. This patch adds a "memory" clobber to the inline
asm code in arch_local_irq_restore() which is used as the building
block for other functions needing to set/clear the interrupt enable
in the CSR register.

Signed-off-by: Mark Salter <msalter@redhat.com>
11 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs
Linus Torvalds [Tue, 9 Apr 2013 19:22:49 +0000 (12:22 -0700)]
Merge branch 'for-linus' of git://git./linux/kernel/git/viro/vfs

Pull vfs fixes from Al Viro:
 "A nasty bug in fs/namespace.c caught by Andrey + a couple of less
  serious unpleasantness - ecryptfs misc device playing hopeless games
  with try_module_get() and palinfo procfs support being...  not quite
  correctly done, to be polite."

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs:
  mnt: release locks on error path in do_loopback
  palinfo fixes
  procfs: add proc_remove_subtree()
  ecryptfs: close rmmod race

11 years agoMerge tag 'vfio-v3.9-rc7' of git://github.com/awilliam/linux-vfio
Linus Torvalds [Tue, 9 Apr 2013 19:07:01 +0000 (12:07 -0700)]
Merge tag 'vfio-v3.9-rc7' of git://github.com/awilliam/linux-vfio

Pull vfio overflow fix from Alex Williamson.

* tag 'vfio-v3.9-rc7' of git://github.com/awilliam/linux-vfio:
  vfio-pci: Fix possible integer overflow

11 years agoMerge tag 'for-linus-20130409' of git://git.kernel.dk/linux-block
Linus Torvalds [Tue, 9 Apr 2013 19:05:41 +0000 (12:05 -0700)]
Merge tag 'for-linus-20130409' of git://git.kernel.dk/linux-block

Pull block fixes from Jens Axboe:
 "I've got a few smaller fixes queued up for 3.9 that should go in.  The
  major one is the loop regression, the others are nice fixes on their
  own though.  It contains:

   - Fix for unitialized var in the block sysfs code, courtesy of Arnd
     and gcc-4.8.

   - Two fixes for mtip32xx, fixing probe and command timeout.  Also a
     debug measure that could have waited for 3.10, but it's driver
     only, so I let it slip in.

   - Revert the loop partition cleanup fix, it could cause a deadlock on
     auto-teardown as part of umount.  The fix is clear, but at this
     point we just want to revert it and get a real fix in for 3.10."

* tag 'for-linus-20130409' of git://git.kernel.dk/linux-block:
  Revert "loop: cleanup partitions when detaching loop device"
  mtip32xx: fix two smatch warnings
  mtip32xx: Add debugfs entry device_status
  mtip32xx: return 0 from pci probe in case of rebuild
  mtip32xx: recovery from command timeout
  block: avoid using uninitialized value in from queue_var_store

11 years agonetfilter: ipset: hash:*net*: nomatch flag not excluded on set resize
Jozsef Kadlecsik [Tue, 9 Apr 2013 08:57:20 +0000 (08:57 +0000)]
netfilter: ipset: hash:*net*: nomatch flag not excluded on set resize

If a resize is triggered the nomatch flag is not excluded at hashing,
which leads to the element missed at lookup in the resized set.

Signed-off-by: Jozsef Kadlecsik <kadlec@blackhole.kfki.hu>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
11 years agonetfilter: ipset: list:set: fix reference counter update
Jozsef Kadlecsik [Tue, 9 Apr 2013 08:57:19 +0000 (08:57 +0000)]
netfilter: ipset: list:set: fix reference counter update

The last element can be replaced or pushed off and in both
cases the reference counter must be updated.

Signed-off-by: Jozsef Kadlecsik <kadlec@blackhole.kfki.hu>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
11 years agomnt: release locks on error path in do_loopback
Andrey Vagin [Tue, 9 Apr 2013 13:33:29 +0000 (17:33 +0400)]
mnt: release locks on error path in do_loopback

do_loopback calls lock_mount(path) and forget to unlock_mount
if clone_mnt or copy_mnt fails.

[   77.661566] ================================================
[   77.662939] [ BUG: lock held when returning to user space! ]
[   77.664104] 3.9.0-rc5+ #17 Not tainted
[   77.664982] ------------------------------------------------
[   77.666488] mount/514 is leaving the kernel with locks still held!
[   77.668027] 2 locks held by mount/514:
[   77.668817]  #0:  (&sb->s_type->i_mutex_key#7){+.+.+.}, at: [<ffffffff811cca22>] lock_mount+0x32/0xe0
[   77.671755]  #1:  (&namespace_sem){+++++.}, at: [<ffffffff811cca3a>] lock_mount+0x4a/0xe0

Signed-off-by: Andrey Vagin <avagin@openvz.org>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
11 years agopalinfo fixes
Al Viro [Mon, 1 Apr 2013 02:34:37 +0000 (22:34 -0400)]
palinfo fixes

* check for proc_mkdir() failures
* fix buffer overrun - sizeof(format string) is *not* enough to
hold sprintf() result.
* use proc_remove_subtree(); life's much easier with it

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
11 years agoprocfs: add proc_remove_subtree()
Al Viro [Sun, 31 Mar 2013 00:13:46 +0000 (20:13 -0400)]
procfs: add proc_remove_subtree()

just what it sounds like; do that only to procfs subtrees you've
created - doing that to something shared with another driver is
not only antisocial, but might cause interesting races with
proc_create() and its ilk.

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
11 years agoecryptfs: close rmmod race
Al Viro [Thu, 28 Mar 2013 17:30:23 +0000 (13:30 -0400)]
ecryptfs: close rmmod race

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
11 years agospinlocks and preemption points need to be at least compiler barriers
Linus Torvalds [Tue, 9 Apr 2013 17:48:33 +0000 (10:48 -0700)]
spinlocks and preemption points need to be at least compiler barriers

In UP and non-preempt respectively, the spinlocks and preemption
disable/enable points are stubbed out entirely, because there is no
regular code that can ever hit the kind of concurrency they are meant to
protect against.

However, while there is no regular code that can cause scheduling, we
_do_ end up having some exceptional (literally!) code that can do so,
and that we need to make sure does not ever get moved into the critical
region by the compiler.

In particular, get_user() and put_user() is generally implemented as
inline asm statements (even if the inline asm may then make a call
instruction to call out-of-line), and can obviously cause a page fault
and IO as a result.  If that inline asm has been scheduled into the
middle of a preemption-safe (or spinlock-protected) code region, we
obviously lose.

Now, admittedly this is *very* unlikely to actually ever happen, and
we've not seen examples of actual bugs related to this.  But partly
exactly because it's so hard to trigger and the resulting bug is so
subtle, we should be extra careful to get this right.

So make sure that even when preemption is disabled, and we don't have to
generate any actual *code* to explicitly tell the system that we are in
a preemption-disabled region, we need to at least tell the compiler not
to move things around the critical region.

This patch grew out of the same discussion that caused commits
79e5f05edcbf ("ARC: Add implicit compiler barrier to raw_local_irq*
functions") and 3e2e0d2c222b ("tile: comment assumption about
__insn_mtspr for <asm/irqflags.h>") to come about.

Note for stable: use discretion when/if applying this.  As mentioned,
this bug may never have actually bitten anybody, and gcc may never have
done the required code motion for it to possibly ever trigger in
practice.

Cc: stable@vger.kernel.org
Cc: Steven Rostedt <srostedt@redhat.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
11 years agoMerge branch 'fixes-for-3.9' of git://gitorious.org/linux-can/linux-can
David S. Miller [Tue, 9 Apr 2013 17:24:35 +0000 (13:24 -0400)]
Merge branch 'fixes-for-3.9' of git://gitorious.org/linux-can/linux-can

Marc Kleine-Budde says:

====================
here's a fix for the v3.9 release cycle, if not too late:

Wei Yongjun contributes a patch for the can-gw protocoll. The patch fixes the
memory allocated with kmem_cache_alloc(), is now freed using kmem_cache_free(),
not kfree().
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoselinux: add a skb_owned_by() hook
Eric Dumazet [Mon, 8 Apr 2013 17:58:11 +0000 (17:58 +0000)]
selinux: add a skb_owned_by() hook

Commit 90ba9b1986b5ac (tcp: tcp_make_synack() can use alloc_skb())
broke certain SELinux/NetLabel configurations by no longer correctly
assigning the sock to the outgoing SYNACK packet.

Cost of atomic operations on the LISTEN socket is quite big,
and we would like it to happen only if really needed.

This patch introduces a new security_ops->skb_owned_by() method,
that is a void operation unless selinux is active.

Reported-by: Miroslav Vadkerti <mvadkert@redhat.com>
Diagnosed-by: Paul Moore <pmoore@redhat.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: "David S. Miller" <davem@davemloft.net>
Cc: linux-security-module@vger.kernel.org
Acked-by: James Morris <james.l.morris@oracle.com>
Tested-by: Paul Moore <pmoore@redhat.com>
Acked-by: Paul Moore <pmoore@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoNFSv4: Fix another potential state manager deadlock
Trond Myklebust [Tue, 9 Apr 2013 16:56:52 +0000 (12:56 -0400)]
NFSv4: Fix another potential state manager deadlock

Don't hold the NFSv4 sequence id while we check for open permission.
The call to ACCESS may block due to reboot recovery.

Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
11 years agotile: comment assumption about __insn_mtspr for <asm/irqflags.h>
Chris Metcalf [Tue, 9 Apr 2013 16:33:07 +0000 (12:33 -0400)]
tile: comment assumption about __insn_mtspr for <asm/irqflags.h>

The arch_local_irq_save(), etc., routines are required to function
as compiler barriers.  They do, but it's subtle and requires knowing
that the gcc builtin __insn_mtspr() is marked as a memory clobber.
Provide a comment explaining the assumption.

Signed-off-by: Chris Metcalf <cmetcalf@tilera.com>
[ This came about from me wondering about the synchronization rules of
  __insn_mtspr()   - Linus ]
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
11 years agocan: gw: use kmem_cache_free() instead of kfree()
Wei Yongjun [Tue, 9 Apr 2013 06:16:04 +0000 (14:16 +0800)]
can: gw: use kmem_cache_free() instead of kfree()

Memory allocated by kmem_cache_alloc() should be freed using
kmem_cache_free(), not kfree().

Cc: linux-stable <stable@vger.kernel.org> # >= v3.2
Signed-off-by: Wei Yongjun <yongjun_wei@trendmicro.com.cn>
Acked-by: Oliver Hartkopp <socketcan@hartkopp.net>
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
11 years agonetrom: fix invalid use of sizeof in nr_recvmsg()
Wei Yongjun [Tue, 9 Apr 2013 02:07:19 +0000 (10:07 +0800)]
netrom: fix invalid use of sizeof in nr_recvmsg()

sizeof() when applied to a pointer typed expression gives the size of the
pointer, not that of the pointed data.
Introduced by commit 3ce5ef(netrom: fix info leak via msg_name in nr_recvmsg)

Signed-off-by: Wei Yongjun <yongjun_wei@trendmicro.com.cn>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoMerge tag 'hwspinlock-3.9-fix' of git://git.kernel.org/pub/scm/linux/kernel/git/ohad...
Linus Torvalds [Tue, 9 Apr 2013 01:53:57 +0000 (18:53 -0700)]
Merge tag 'hwspinlock-3.9-fix' of git://git./linux/kernel/git/ohad/hwspinlock

Pull a hwspinlock fix from Ohad Ben-Cohen:
 "An hwspinlock fix from Li Fei, taking care of a faulty error path."

* tag 'hwspinlock-3.9-fix' of git://git.kernel.org/pub/scm/linux/kernel/git/ohad/hwspinlock:
  hwspinlock: fix __hwspin_lock_request error path

11 years agoMerge tag 'remoteproc-3.9-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Tue, 9 Apr 2013 01:53:13 +0000 (18:53 -0700)]
Merge tag 'remoteproc-3.9-fixes' of git://git./linux/kernel/git/ohad/remoteproc

Pull remoteproc fixes from Ohad Ben-Cohen:
 "Four small remoteproc fixes:
   - Suman fixed an issue that crawled in with the move to the new
     idr_alloc interface in 3.9
   - Dmitry fixed an STE specific memory leak
   - Sjur fixed an error path in the core
   - Rob fixed a Kconfig typo"

* tag 'remoteproc-3.9-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/ohad/remoteproc:
  remoteproc: fix FW_CONFIG typo
  remoteproc: fix error path of handle_vdev
  remoteproc/ste: fix memory leak on shutdown
  remoteproc: fix the error check for idr_alloc

11 years agoMerge branch 'drm-nouveau-fixes-3.9' of git://anongit.freedesktop.org/git/nouveau...
Dave Airlie [Mon, 8 Apr 2013 23:28:55 +0000 (09:28 +1000)]
Merge branch 'drm-nouveau-fixes-3.9' of git://anongit.freedesktop.org/git/nouveau/linux-2.6 into drm-fixes

too many semicolons.

* 'drm-nouveau-fixes-3.9' of git://anongit.freedesktop.org/git/nouveau/linux-2.6:
  drm/nouveau: fix unconditional return waiting on memory

11 years agodrm/nouveau: fix unconditional return waiting on memory
Calvin Owens [Mon, 8 Apr 2013 02:01:19 +0000 (21:01 -0500)]
drm/nouveau: fix unconditional return waiting on memory

Typo in nv50_display_flip_wait allows page flipping to run ahead before
memory has time to settle.

Signed-off-by: Calvin Owens <jcalvinowens@gmail.com>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
11 years agoMerge tag 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/sfr/next-fixes
Linus Torvalds [Mon, 8 Apr 2013 23:10:43 +0000 (16:10 -0700)]
Merge tag 'for-linus' of git://git./linux/kernel/git/sfr/next-fixes

Pull powerpc bugfix from Stephen Rothwell:
 "A single BUG_ON fix for a condition that could happen for machines
  with certain hardware installed."

* tag 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/sfr/next-fixes:
  powerpc: pSeries_lpar_hpte_remove fails from Adjunct partition being performed before the ANDCOND test

11 years agoARC: Add implicit compiler barrier to raw_local_irq* functions
Christian Ruppert [Mon, 8 Apr 2013 07:35:30 +0000 (13:05 +0530)]
ARC: Add implicit compiler barrier to raw_local_irq* functions

ARC irqsave/restore macros were missing the compiler barrier, causing a
stale load in irq-enabled region be used in irq-safe region, despite
being changed, because the register holding the value was still live.

The problem manifested as random crashes in timer code when stress
testing ARCLinux (3.9-rc3) on a !SMP && !PREEMPT_COUNT

Here's the exact sequence which caused this:
 (0). tv1[x] <----> t1 <---> t2
 (1). mod_timer(t1) interrupted after it calls timer_pending()
 (2). mod_timer(t2) completes
 (3). mod_timer(t1) resumes but messes up the list
 (4). __runt_timers( ) uses bogus timer_list entry / crashes in
      timer->function

Essentially mod_timer() was racing against itself and while the spinlock
serialized the tv1[] timer link list, timer_pending() called outside the
spinlock, cached timer link list element in a register.
With low register pressure (and a deep register file), lack of barrier
in raw_local_irqsave() as well as preempt_disable (!PREEMPT_COUNT
version), there was nothing to force gcc to reload across the spinlock,
causing a stale value in reg be used for link list manipulation - ensuing
a corruption.

ARcompact disassembly which shows the culprit generated code:

mod_timer:
    push_s blink
    mov_s r13,r0 # timer, timer
..
    ###### timer_pending( )
    ld_s r3,[r13]       # <------ <variable>.entry.next LOADED
    brne r3, 0, @.L163

.L163:
..
    ###### spin_lock_irq( )
    lr  r5, [status32]  # flags
    bic r4, r5, 6       # temp, flags,
    and.f 0, r5, 6      # flags,
    flag.nz r4

    ###### detach_if_pending( ) begins

    tst_s r3,r3  <--------------
# timer_pending( ) checks timer->entry.next
                        # r3 is NOT reloaded by gcc, using stale value
    beq.d @.L169
    mov.eq r0,0

    #####  detach_timer( ): __list_del( )

    ld r4,[r13,4]     # <variable>.entry.prev, D.31439
    st r4,[r3,4]      # <variable>.prev, D.31439
    st r3,[r4]        # <variable>.next, D.30246

We initially tried to fix this by adding barrier() to preempt_* macros
for !PREEMPT_COUNT but Linus clarified that it was anything but wrong.
http://www.spinics.net/lists/kernel/msg1512709.html

[vgupta: updated commitlog]

Reported-by/Signed-off-by: Christian Ruppert <christian.ruppert@abilis.com>
Cc: Christian Ruppert <christian.ruppert@abilis.com>
Cc: Pierrick Hascoet <pierrick.hascoet@abilis.com>
Debugged-by/Signed-off-by: Vineet Gupta <vgupta@synopsys.com>

Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
11 years agoMerge tag 'upstream-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jgarzik...
Linus Torvalds [Mon, 8 Apr 2013 22:15:22 +0000 (15:15 -0700)]
Merge tag 'upstream-linus' of git://git./linux/kernel/git/jgarzik/libata-dev

Pull libata fixes from Jeff Garzik:
 "The HDIO_DRIVE_* fix is really the biggie.

  1) Fix ATAPI regression, noticed mainly on tape drives, due to a
     commit which mistakenly changed an 'int' return type to a 'bool'.
     Broken by commit 4dce8ba94c75 ("libata: Use 'bool' return value for
     ata_id_XXX")

  2) Add Slimtype DVD A DS8A8SH ATAPI quirk

  3) ata_piix: Intel Haswell platform quirk

  4) Avoid DMA'ing to stack buffer, when obtaining DEVSLP timings.  IMO
     a mild regression, given that libata previously did not DMA to a
     stack buffer.  Broken by commit commit 803739d25c23 ("[libata]
     replace sata_settings with devslp_timing")

  5) Fix regression impacting SMART and smartd, broken by commit
     84a9a8cd9d0a ("[libata] Set proper SK when CK_COND is set")"

* tag 'upstream-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jgarzik/libata-dev:
  [libata] Fix HDIO_DRIVE_* ioctl() Linux 3.9 regression
  libata: fix DMA to stack in reading devslp_timing parameters
  ata_piix: Fix DVD not dectected at some Haswell platforms
  libata: Set max sector to 65535 for Slimtype DVD A DS8A8SH drive
  libata: Use integer return value for atapi_command_packet_set

11 years agoMerge tag 'trace-fixes-3.9-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Mon, 8 Apr 2013 22:14:11 +0000 (15:14 -0700)]
Merge tag 'trace-fixes-3.9-rc6' of git://git./linux/kernel/git/rostedt/linux-trace

Pull tracing fixes from Steven Rostedt:
 "This includes three fixes.  Two fix features added in 3.9 and one
  fixes a long time minor bug.

  The first patch fixes a race that can happen if the user switches from
  the irqsoff tracer to another tracer.  If a irqs off latency is
  detected, it will try to use the snapshot buffer, but the new tracer
  wont have it allocated.  There's a nasty warning that gets printed and
  the trace is ignored.  Nothing crashes, just a nasty WARN_ON is shown.

  The second patch fixes an issue where if the sysctl is used to disable
  and enable function tracing, it can put the function tracing into an
  unstable state.

  The third patch fixes an issue with perf using the function tracer.
  An update was done, where the stub function could be called during the
  perf function tracing, and that stub function wont have the "control"
  flag set and cause a nasty warning when running perf."

* tag 'trace-fixes-3.9-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/rostedt/linux-trace:
  ftrace: Do not call stub functions in control loop
  ftrace: Consistently restore trace function on sysctl enabling
  tracing: Fix race with update_max_tr_single and changing tracers

11 years agoqeth: fix qeth_wait_for_threads() deadlock for OSN devices
Stefan Raspl [Sun, 7 Apr 2013 22:19:27 +0000 (22:19 +0000)]
qeth: fix qeth_wait_for_threads() deadlock for OSN devices

Any recovery thread will deadlock when calling qeth_wait_for_threads(), most
notably when triggering a recovery on an OSN device.
This patch will store the recovery thread's task pointer on recovery
invocation and check in qeth_wait_for_threads() respectively to avoid
deadlocks.

Signed-off-by: Stefan Raspl <raspl@linux.vnet.ibm.com>
Signed-off-by: Frank Blaschka <blaschka@linux.vnet.ibm.com>
Reviewed-by: Ursula Braun <ursula.braun@de.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoaf_iucv: fix recvmsg by replacing skb_pull() function
Ursula Braun [Sun, 7 Apr 2013 22:19:26 +0000 (22:19 +0000)]
af_iucv: fix recvmsg by replacing skb_pull() function

When receiving data messages, the "BUG_ON(skb->len < skb->data_len)" in
the skb_pull() function triggers a kernel panic.

Replace the skb_pull logic by a per skb offset as advised by
Eric Dumazet.

Signed-off-by: Ursula Braun <ursula.braun@de.ibm.com>
Signed-off-by: Frank Blaschka <blaschka@linux.vnet.ibm.com>
Reviewed-by: Hendrik Brueckner <brueckner@linux.vnet.ibm.com>
Acked-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agortnetlink: Call nlmsg_parse() with correct header length
Michael Riesch [Mon, 8 Apr 2013 05:45:26 +0000 (05:45 +0000)]
rtnetlink: Call nlmsg_parse() with correct header length

Signed-off-by: Michael Riesch <michael.riesch@omicron.at>
Cc: "David S. Miller" <davem@davemloft.net>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: Jiri Benc <jbenc@redhat.com>
Cc: "Theodore Ts'o" <tytso@mit.edu>
Cc: linux-kernel@vger.kernel.org
Acked-by: Mark Rustad <mark.d.rustad@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agobonding: fix bonding_masters race condition in bond unloading
nikolay@redhat.com [Sat, 6 Apr 2013 00:54:38 +0000 (00:54 +0000)]
bonding: fix bonding_masters race condition in bond unloading

While the bonding module is unloading, it is considered that after
rtnl_link_unregister all bond devices are destroyed but since no
synchronization mechanism exists, a new bond device can be created
via bonding_masters before unregister_pernet_subsys which would
lead to multiple problems (e.g. NULL pointer dereference, wrong RIP,
list corruption).

This patch fixes the issue by removing any bond devices left in the
netns after bonding_masters is removed from sysfs.

Signed-off-by: Nikolay Aleksandrov <nikolay@redhat.com>
Acked-by: Veaceslav Falico <vfalico@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoRevert "bonding: remove sysfs before removing devices"
nikolay@redhat.com [Sat, 6 Apr 2013 00:54:37 +0000 (00:54 +0000)]
Revert "bonding: remove sysfs before removing devices"

This reverts commit 4de79c737b200492195ebc54a887075327e1ec1d.

This patch introduces a new bug which causes access to freed memory.
In bond_uninit: list_del(&bond->bond_list);
bond_list is linked in bond_net's dev_list which is freed by
unregister_pernet_subsys.

Signed-off-by: Nikolay Aleksandrov <nikolay@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoMerge branch 'wireless'
David S. Miller [Mon, 8 Apr 2013 20:39:55 +0000 (16:39 -0400)]
Merge branch 'wireless'

John W. Linville says:

====================
For the cfg80211 fix, Johannes says:

"I have another straggler for 3.9, adding locking forgotten in a previous
fix."

On top of that:

Bing Zhao provides an mwifiex fix to properly order a scan completion.

Franky Lin gives us a brcmfmac fix to fail at the firmware loading
stage if the nvram cannot be downloaded.

Gabor Juhos brings what at first looks like a rather big rt2x00 patch.
I think it is OK because it is really just reorganizing some code
within the rt2x00 driver in order to fix a build failure.

Hante Meuleman offers a trio of brcmfmac fixes related to running in
AP mode.

Robert Shade sends an ath9k fix to reenable interrupts even if a
channel change fails.

Tim Gardner gives us an rt2x00 fix to cut-down on some log SPAM.

Please let me know if there are problems!
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoMerge branch 'master' of git://git.kernel.org/pub/scm/linux/kernel/git/linville/wirel...
John W. Linville [Mon, 8 Apr 2013 18:26:57 +0000 (14:26 -0400)]
Merge branch 'master' of git://git./linux/kernel/git/linville/wireless into for-davem

11 years agoftrace: Do not call stub functions in control loop
Steven Rostedt (Red Hat) [Wed, 27 Mar 2013 13:31:28 +0000 (09:31 -0400)]
ftrace: Do not call stub functions in control loop

The function tracing control loop used by perf spits out a warning
if the called function is not a control function. This is because
the control function references a per cpu allocated data structure
on struct ftrace_ops that is not allocated for other types of
functions.

commit 0a016409e42 "ftrace: Optimize the function tracer list loop"

Had an optimization done to all function tracing loops to optimize
for a single registered ops. Unfortunately, this allows for a slight
race when tracing starts or ends, where the stub function might be
called after the current registered ops is removed. In this case we
get the following dump:

root# perf stat -e ftrace:function sleep 1
[   74.339105] WARNING: at include/linux/ftrace.h:209 ftrace_ops_control_func+0xde/0xf0()
[   74.349522] Hardware name: PRIMERGY RX200 S6
[   74.357149] Modules linked in: sg igb iTCO_wdt ptp pps_core iTCO_vendor_support i7core_edac dca lpc_ich i2c_i801 coretemp edac_core crc32c_intel mfd_core ghash_clmulni_intel dm_multipath acpi_power_meter pcspk
r microcode vhost_net tun macvtap macvlan nfsd kvm_intel kvm auth_rpcgss nfs_acl lockd sunrpc uinput xfs libcrc32c sd_mod crc_t10dif sr_mod cdrom mgag200 i2c_algo_bit drm_kms_helper ttm qla2xxx mptsas ahci drm li
bahci scsi_transport_sas mptscsih libata scsi_transport_fc i2c_core mptbase scsi_tgt dm_mirror dm_region_hash dm_log dm_mod
[   74.446233] Pid: 1377, comm: perf Tainted: G        W    3.9.0-rc1 #1
[   74.453458] Call Trace:
[   74.456233]  [<ffffffff81062e3f>] warn_slowpath_common+0x7f/0xc0
[   74.462997]  [<ffffffff810fbc60>] ? rcu_note_context_switch+0xa0/0xa0
[   74.470272]  [<ffffffff811041a2>] ? __unregister_ftrace_function+0xa2/0x1a0
[   74.478117]  [<ffffffff81062e9a>] warn_slowpath_null+0x1a/0x20
[   74.484681]  [<ffffffff81102ede>] ftrace_ops_control_func+0xde/0xf0
[   74.491760]  [<ffffffff8162f400>] ftrace_call+0x5/0x2f
[   74.497511]  [<ffffffff8162f400>] ? ftrace_call+0x5/0x2f
[   74.503486]  [<ffffffff8162f400>] ? ftrace_call+0x5/0x2f
[   74.509500]  [<ffffffff810fbc65>] ? synchronize_sched+0x5/0x50
[   74.516088]  [<ffffffff816254d5>] ? _cond_resched+0x5/0x40
[   74.522268]  [<ffffffff810fbc65>] ? synchronize_sched+0x5/0x50
[   74.528837]  [<ffffffff811041a2>] ? __unregister_ftrace_function+0xa2/0x1a0
[   74.536696]  [<ffffffff816254d5>] ? _cond_resched+0x5/0x40
[   74.542878]  [<ffffffff8162402d>] ? mutex_lock+0x1d/0x50
[   74.548869]  [<ffffffff81105c67>] unregister_ftrace_function+0x27/0x50
[   74.556243]  [<ffffffff8111eadf>] perf_ftrace_event_register+0x9f/0x140
[   74.563709]  [<ffffffff816254d5>] ? _cond_resched+0x5/0x40
[   74.569887]  [<ffffffff8162402d>] ? mutex_lock+0x1d/0x50
[   74.575898]  [<ffffffff8111e94e>] perf_trace_destroy+0x2e/0x50
[   74.582505]  [<ffffffff81127ba9>] tp_perf_event_destroy+0x9/0x10
[   74.589298]  [<ffffffff811295d0>] free_event+0x70/0x1a0
[   74.595208]  [<ffffffff8112a579>] perf_event_release_kernel+0x69/0xa0
[   74.602460]  [<ffffffff816254d5>] ? _cond_resched+0x5/0x40
[   74.608667]  [<ffffffff8112a640>] put_event+0x90/0xc0
[   74.614373]  [<ffffffff8112a740>] perf_release+0x10/0x20
[   74.620367]  [<ffffffff811a3044>] __fput+0xf4/0x280
[   74.625894]  [<ffffffff811a31de>] ____fput+0xe/0x10
[   74.631387]  [<ffffffff81083697>] task_work_run+0xa7/0xe0
[   74.637452]  [<ffffffff81014981>] do_notify_resume+0x71/0xb0
[   74.643843]  [<ffffffff8162fa92>] int_signal+0x12/0x17

To fix this a new ftrace_ops flag is added that denotes the ftrace_list_end
ftrace_ops stub as just that, a stub. This flag is now checked in the
control loop and the function is not called if the flag is set.

Thanks to Jovi for not just reporting the bug, but also pointing out
where the bug was in the code.

Link: http://lkml.kernel.org/r/514A8855.7090402@redhat.com
Link: http://lkml.kernel.org/r/1364377499-1900-15-git-send-email-jovi.zhangwei@huawei.com
Tested-by: WANG Chao <chaowang@redhat.com>
Reported-by: WANG Chao <chaowang@redhat.com>
Reported-by: zhangwei(Jovi) <jovi.zhangwei@huawei.com>
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
11 years agoftrace: Consistently restore trace function on sysctl enabling
Jan Kiszka [Tue, 26 Mar 2013 16:53:03 +0000 (17:53 +0100)]
ftrace: Consistently restore trace function on sysctl enabling

If we reenable ftrace via syctl, we currently set ftrace_trace_function
based on the previous simplistic algorithm. This is inconsistent with
what update_ftrace_function does. So better call that helper instead.

Link: http://lkml.kernel.org/r/5151D26F.1070702@siemens.com
Cc: stable@vger.kernel.org
Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
11 years agotracing: Fix race with update_max_tr_single and changing tracers
Steven Rostedt (Red Hat) [Tue, 26 Mar 2013 21:33:00 +0000 (17:33 -0400)]
tracing: Fix race with update_max_tr_single and changing tracers

The commit 34600f0e9 "tracing: Fix race with max_tr and changing tracers"
fixed the updating of the main buffers with the race of changing
tracers, but left out the fix to the updating of just a per cpu buffer.

Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
11 years agonet: mvneta: enable features before registering the driver
willy tarreau [Sat, 6 Apr 2013 08:47:01 +0000 (08:47 +0000)]
net: mvneta: enable features before registering the driver

It seems that the reason why the dev features were ignored was because
they were enabled after registeration.

Signed-off-by: Willy Tarreau <w@1wt.eu>
Acked-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agohyperv: Fix RNDIS send_completion code path
Haiyang Zhang [Fri, 5 Apr 2013 11:44:40 +0000 (11:44 +0000)]
hyperv: Fix RNDIS send_completion code path

In some cases, the VM_PKT_COMP message can arrive later than RNDIS completion
message, which will free the packet memory. This may cause panic due to access
to freed memory in netvsc_send_completion().

This patch fixes this problem by removing rndis_filter_send_request_completion()
from the code path. The function was a no-op.

Reported-by: Long Li <longli@microsoft.com>
Tested-by: Long Li <longli@microsoft.com>
Signed-off-by: Haiyang Zhang <haiyangz@microsoft.com>
Reviewed-by: K. Y. Srinivasan <kys@microsoft.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agohyperv: Fix a kernel warning from netvsc_linkstatus_callback()
Haiyang Zhang [Fri, 5 Apr 2013 11:44:39 +0000 (11:44 +0000)]
hyperv: Fix a kernel warning from netvsc_linkstatus_callback()

The warning about local_bh_enable inside IRQ happens when disconnecting a
virtual NIC.

The reason for the warning is -- netif_tx_disable() is called when the NIC
is disconnected. And it's called within irq context. netif_tx_disable() calls
local_bh_enable() which displays warning if in irq.

The fix is to remove the unnecessary netif_tx_disable & wake_queue() in the
netvsc_linkstatus_callback().

Reported-by: Richard Genoud <richard.genoud@gmail.com>
Tested-by: Long Li <longli@microsoft.com>
Tested-by: Richard Genoud <richard.genoud@gmail.com>
Signed-off-by: Haiyang Zhang <haiyangz@microsoft.com>
Reviewed-by: K. Y. Srinivasan <kys@microsoft.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agonet: ipv4: fix schedule while atomic bug in check_lifetime()
Jiri Pirko [Thu, 4 Apr 2013 23:39:39 +0000 (23:39 +0000)]
net: ipv4: fix schedule while atomic bug in check_lifetime()

move might_sleep operations out of the rcu_read_lock() section.
Also fix iterating over ifa_dev->ifa_list

Introduced by: commit 5c766d642bcaf "ipv4: introduce address lifetime"

Signed-off-by: Jiri Pirko <jiri@resnulli.us>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agonet: ipv4: reset check_lifetime_work after changing lifetime
Jiri Pirko [Thu, 4 Apr 2013 23:39:38 +0000 (23:39 +0000)]
net: ipv4: reset check_lifetime_work after changing lifetime

This will result in calling check_lifetime in nearest opportunity and
that function will adjust next time to call check_lifetime correctly.
Without this, check_lifetime is called in time computed by previous run,
not affecting modified lifetime.

Introduced by: commit 5c766d642bcaf "ipv4: introduce address lifetime"

Signed-off-by: Jiri Pirko <jiri@resnulli.us>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoRevert "loop: cleanup partitions when detaching loop device"
Jens Axboe [Mon, 8 Apr 2013 08:12:11 +0000 (10:12 +0200)]
Revert "loop: cleanup partitions when detaching loop device"

This reverts commit 8761a3dc1f07b163414e2215a2cadbb4cfe2a107.

There are situations where the destruction path is called
with the bdev->bd_mutex already held, which then deadlocks in
loop_clr_fd(). The normal partition cleanup does a trylock()
on the mutex, but it'd be nice to have a more bullet proof
method in loop. So punt this more involved fix to the next
merge window, and just back out this buggy fix for now.

Signed-off-by: Jens Axboe <axboe@kernel.dk>
11 years agopowerpc: pSeries_lpar_hpte_remove fails from Adjunct partition being performed before...
Michael Wolf [Fri, 5 Apr 2013 10:41:40 +0000 (10:41 +0000)]
powerpc: pSeries_lpar_hpte_remove fails from Adjunct partition being performed before the ANDCOND test

Some versions of pHyp will perform the adjunct partition test before the
ANDCOND test.  The result of this is that H_RESOURCE can be returned and
cause the BUG_ON condition to occur. The HPTE is not removed.  So add a
check for H_RESOURCE, it is ok if this HPTE is not removed as
pSeries_lpar_hpte_remove is looking for an HPTE to remove and not a
specific HPTE to remove.  So it is ok to just move on to the next slot
and try again.

Cc: stable@vger.kernel.org
Signed-off-by: Michael Wolf <mjw@linux.vnet.ibm.com>
Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
11 years agoLinux 3.9-rc6
Linus Torvalds [Mon, 8 Apr 2013 03:49:54 +0000 (20:49 -0700)]
Linux 3.9-rc6

11 years agobnx2x: Fix KR2 rapid link flap
Yaniv Rosner [Sun, 7 Apr 2013 05:36:23 +0000 (05:36 +0000)]
bnx2x: Fix KR2 rapid link flap

Check KR2 recovery time at the beginning of the work-around function.

Signed-off-by: Yaniv Rosner <yanivr@broadcom.com>
Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agosctp: remove 'sridhar' from maintainers list
Sridhar Samudrala [Tue, 2 Apr 2013 12:35:31 +0000 (12:35 +0000)]
sctp: remove 'sridhar' from maintainers list

Update SCTP maintainers list.

Signed-off-by: Sridhar Samudrala <sri@us.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoMerge branch 'infoleaks'
David S. Miller [Sun, 7 Apr 2013 20:42:40 +0000 (16:42 -0400)]
Merge branch 'infoleaks'

Mathias Krause says:

====================
a few more info leak fixes in the recvmsg path. The error pattern here
is the protocol specific recvmsg function is missing the msg_namelen
assignment -- either completely or in early exit paths that do not
result in errors in __sys_recvmsg()/sys_recvfrom() and, in turn, make
them call move_addr_to_user(), leaking the then still uninitialized
sockaddr_storage stack variable to userland.

My audit was initiated by a rather coarse fix of the leak that can be
found in the grsecurity patch, putting a penalty on protocols complying
to the rules of recvmsg. So credits for finding the leak in the recvmsg
path in __sys_recvmsg() should go to Brad!

The buggy protocols/subsystems are rather obscure anyway. As a missing
assignment of msg_namelen coupled with a missing filling of msg_name
would only result in garbage -- the leak -- in case userland would care
about that information, i.e. would provide a msg_name pointer. But
obviously current userland does not.

While auditing the code for the above pattern I found a few more
'uninitialized members' kind of leaks related to the msg_name filling.
Those are fixed in this series, too.

I have to admit, I failed to test all of the patches due to missing
hardware, e.g. iucv depends on S390 -- hardware I've no access to :/
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoVSOCK: Fix missing msg_namelen update in vsock_stream_recvmsg()
Mathias Krause [Sun, 7 Apr 2013 01:52:02 +0000 (01:52 +0000)]
VSOCK: Fix missing msg_namelen update in vsock_stream_recvmsg()

The code misses to update the msg_namelen member to 0 and therefore
makes net/socket.c leak the local, uninitialized sockaddr_storage
variable to userland -- 128 bytes of kernel stack memory.

Cc: Andy King <acking@vmware.com>
Cc: Dmitry Torokhov <dtor@vmware.com>
Cc: George Zhang <georgezhang@vmware.com>
Signed-off-by: Mathias Krause <minipli@googlemail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoVSOCK: vmci - fix possible info leak in vmci_transport_dgram_dequeue()
Mathias Krause [Sun, 7 Apr 2013 01:52:01 +0000 (01:52 +0000)]
VSOCK: vmci - fix possible info leak in vmci_transport_dgram_dequeue()

In case we received no data on the call to skb_recv_datagram(), i.e.
skb->data is NULL, vmci_transport_dgram_dequeue() will return with 0
without updating msg_namelen leading to net/socket.c leaking the local,
uninitialized sockaddr_storage variable to userland -- 128 bytes of
kernel stack memory.

Fix this by moving the already existing msg_namelen assignment a few
lines above.

Cc: Andy King <acking@vmware.com>
Cc: Dmitry Torokhov <dtor@vmware.com>
Cc: George Zhang <georgezhang@vmware.com>
Signed-off-by: Mathias Krause <minipli@googlemail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agotipc: fix info leaks via msg_name in recv_msg/recv_stream
Mathias Krause [Sun, 7 Apr 2013 01:52:00 +0000 (01:52 +0000)]
tipc: fix info leaks via msg_name in recv_msg/recv_stream

The code in set_orig_addr() does not initialize all of the members of
struct sockaddr_tipc when filling the sockaddr info -- namely the union
is only partly filled. This will make recv_msg() and recv_stream() --
the only users of this function -- leak kernel stack memory as the
msg_name member is a local variable in net/socket.c.

Additionally to that both recv_msg() and recv_stream() fail to update
the msg_namelen member to 0 while otherwise returning with 0, i.e.
"success". This is the case for, e.g., non-blocking sockets. This will
lead to a 128 byte kernel stack leak in net/socket.c.

Fix the first issue by initializing the memory of the union with
memset(0). Fix the second one by setting msg_namelen to 0 early as it
will be updated later if we're going to fill the msg_name member.

Cc: Jon Maloy <jon.maloy@ericsson.com>
Cc: Allan Stephens <allan.stephens@windriver.com>
Signed-off-by: Mathias Krause <minipli@googlemail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agorose: fix info leak via msg_name in rose_recvmsg()
Mathias Krause [Sun, 7 Apr 2013 01:51:59 +0000 (01:51 +0000)]
rose: fix info leak via msg_name in rose_recvmsg()

The code in rose_recvmsg() does not initialize all of the members of
struct sockaddr_rose/full_sockaddr_rose when filling the sockaddr info.
Nor does it initialize the padding bytes of the structure inserted by
the compiler for alignment. This will lead to leaking uninitialized
kernel stack bytes in net/socket.c.

Fix the issue by initializing the memory used for sockaddr info with
memset(0).

Cc: Ralf Baechle <ralf@linux-mips.org>
Signed-off-by: Mathias Krause <minipli@googlemail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoNFC: llcp: fix info leaks via msg_name in llcp_sock_recvmsg()
Mathias Krause [Sun, 7 Apr 2013 01:51:58 +0000 (01:51 +0000)]
NFC: llcp: fix info leaks via msg_name in llcp_sock_recvmsg()

The code in llcp_sock_recvmsg() does not initialize all the members of
struct sockaddr_nfc_llcp when filling the sockaddr info. Nor does it
initialize the padding bytes of the structure inserted by the compiler
for alignment.

Also, if the socket is in state LLCP_CLOSED or is shutting down during
receive the msg_namelen member is not updated to 0 while otherwise
returning with 0, i.e. "success". The msg_namelen update is also
missing for stream and seqpacket sockets which don't fill the sockaddr
info.

Both issues lead to the fact that the code will leak uninitialized
kernel stack bytes in net/socket.c.

Fix the first issue by initializing the memory used for sockaddr info
with memset(0). Fix the second one by setting msg_namelen to 0 early.
It will be updated later if we're going to fill the msg_name member.

Cc: Lauro Ramos Venancio <lauro.venancio@openbossa.org>
Cc: Aloisio Almeida Jr <aloisio.almeida@openbossa.org>
Cc: Samuel Ortiz <sameo@linux.intel.com>
Signed-off-by: Mathias Krause <minipli@googlemail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agonetrom: fix info leak via msg_name in nr_recvmsg()
Mathias Krause [Sun, 7 Apr 2013 01:51:57 +0000 (01:51 +0000)]
netrom: fix info leak via msg_name in nr_recvmsg()

In case msg_name is set the sockaddr info gets filled out, as
requested, but the code fails to initialize the padding bytes of
struct sockaddr_ax25 inserted by the compiler for alignment. Also
the sax25_ndigis member does not get assigned, leaking four more
bytes.

Both issues lead to the fact that the code will leak uninitialized
kernel stack bytes in net/socket.c.

Fix both issues by initializing the memory with memset(0).

Cc: Ralf Baechle <ralf@linux-mips.org>
Signed-off-by: Mathias Krause <minipli@googlemail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agollc: Fix missing msg_namelen update in llc_ui_recvmsg()
Mathias Krause [Sun, 7 Apr 2013 01:51:56 +0000 (01:51 +0000)]
llc: Fix missing msg_namelen update in llc_ui_recvmsg()

For stream sockets the code misses to update the msg_namelen member
to 0 and therefore makes net/socket.c leak the local, uninitialized
sockaddr_storage variable to userland -- 128 bytes of kernel stack
memory. The msg_namelen update is also missing for datagram sockets
in case the socket is shutting down during receive.

Fix both issues by setting msg_namelen to 0 early. It will be
updated later if we're going to fill the msg_name member.

Cc: Arnaldo Carvalho de Melo <acme@ghostprotocols.net>
Signed-off-by: Mathias Krause <minipli@googlemail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agol2tp: fix info leak in l2tp_ip6_recvmsg()
Mathias Krause [Sun, 7 Apr 2013 01:51:55 +0000 (01:51 +0000)]
l2tp: fix info leak in l2tp_ip6_recvmsg()

The L2TP code for IPv6 fails to initialize the l2tp_conn_id member of
struct sockaddr_l2tpip6 and therefore leaks four bytes kernel stack
in l2tp_ip6_recvmsg() in case msg_name is set.

Initialize l2tp_conn_id with 0 to avoid the info leak.

Signed-off-by: Mathias Krause <minipli@googlemail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoiucv: Fix missing msg_namelen update in iucv_sock_recvmsg()
Mathias Krause [Sun, 7 Apr 2013 01:51:54 +0000 (01:51 +0000)]
iucv: Fix missing msg_namelen update in iucv_sock_recvmsg()

The current code does not fill the msg_name member in case it is set.
It also does not set the msg_namelen member to 0 and therefore makes
net/socket.c leak the local, uninitialized sockaddr_storage variable
to userland -- 128 bytes of kernel stack memory.

Fix that by simply setting msg_namelen to 0 as obviously nobody cared
about iucv_sock_recvmsg() not filling the msg_name in case it was set.

Cc: Ursula Braun <ursula.braun@de.ibm.com>
Signed-off-by: Mathias Krause <minipli@googlemail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoirda: Fix missing msg_namelen update in irda_recvmsg_dgram()
Mathias Krause [Sun, 7 Apr 2013 01:51:53 +0000 (01:51 +0000)]
irda: Fix missing msg_namelen update in irda_recvmsg_dgram()

The current code does not fill the msg_name member in case it is set.
It also does not set the msg_namelen member to 0 and therefore makes
net/socket.c leak the local, uninitialized sockaddr_storage variable
to userland -- 128 bytes of kernel stack memory.

Fix that by simply setting msg_namelen to 0 as obviously nobody cared
about irda_recvmsg_dgram() not filling the msg_name in case it was
set.

Cc: Samuel Ortiz <samuel@sortiz.org>
Signed-off-by: Mathias Krause <minipli@googlemail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agocaif: Fix missing msg_namelen update in caif_seqpkt_recvmsg()
Mathias Krause [Sun, 7 Apr 2013 01:51:52 +0000 (01:51 +0000)]
caif: Fix missing msg_namelen update in caif_seqpkt_recvmsg()

The current code does not fill the msg_name member in case it is set.
It also does not set the msg_namelen member to 0 and therefore makes
net/socket.c leak the local, uninitialized sockaddr_storage variable
to userland -- 128 bytes of kernel stack memory.

Fix that by simply setting msg_namelen to 0 as obviously nobody cared
about caif_seqpkt_recvmsg() not filling the msg_name in case it was
set.

Cc: Sjur Braendeland <sjur.brandeland@stericsson.com>
Signed-off-by: Mathias Krause <minipli@googlemail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoBluetooth: SCO - Fix missing msg_namelen update in sco_sock_recvmsg()
Mathias Krause [Sun, 7 Apr 2013 01:51:51 +0000 (01:51 +0000)]
Bluetooth: SCO - Fix missing msg_namelen update in sco_sock_recvmsg()

If the socket is in state BT_CONNECT2 and BT_SK_DEFER_SETUP is set in
the flags, sco_sock_recvmsg() returns early with 0 without updating the
possibly set msg_namelen member. This, in turn, leads to a 128 byte
kernel stack leak in net/socket.c.

Fix this by updating msg_namelen in this case. For all other cases it
will be handled in bt_sock_recvmsg().

Cc: Marcel Holtmann <marcel@holtmann.org>
Cc: Gustavo Padovan <gustavo@padovan.org>
Cc: Johan Hedberg <johan.hedberg@gmail.com>
Signed-off-by: Mathias Krause <minipli@googlemail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoBluetooth: RFCOMM - Fix missing msg_namelen update in rfcomm_sock_recvmsg()
Mathias Krause [Sun, 7 Apr 2013 01:51:50 +0000 (01:51 +0000)]
Bluetooth: RFCOMM - Fix missing msg_namelen update in rfcomm_sock_recvmsg()

If RFCOMM_DEFER_SETUP is set in the flags, rfcomm_sock_recvmsg() returns
early with 0 without updating the possibly set msg_namelen member. This,
in turn, leads to a 128 byte kernel stack leak in net/socket.c.

Fix this by updating msg_namelen in this case. For all other cases it
will be handled in bt_sock_stream_recvmsg().

Cc: Marcel Holtmann <marcel@holtmann.org>
Cc: Gustavo Padovan <gustavo@padovan.org>
Cc: Johan Hedberg <johan.hedberg@gmail.com>
Signed-off-by: Mathias Krause <minipli@googlemail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoBluetooth: fix possible info leak in bt_sock_recvmsg()
Mathias Krause [Sun, 7 Apr 2013 01:51:49 +0000 (01:51 +0000)]
Bluetooth: fix possible info leak in bt_sock_recvmsg()

In case the socket is already shutting down, bt_sock_recvmsg() returns
with 0 without updating msg_namelen leading to net/socket.c leaking the
local, uninitialized sockaddr_storage variable to userland -- 128 bytes
of kernel stack memory.

Fix this by moving the msg_namelen assignment in front of the shutdown
test.

Cc: Marcel Holtmann <marcel@holtmann.org>
Cc: Gustavo Padovan <gustavo@padovan.org>
Cc: Johan Hedberg <johan.hedberg@gmail.com>
Signed-off-by: Mathias Krause <minipli@googlemail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoax25: fix info leak via msg_name in ax25_recvmsg()
Mathias Krause [Sun, 7 Apr 2013 01:51:48 +0000 (01:51 +0000)]
ax25: fix info leak via msg_name in ax25_recvmsg()

When msg_namelen is non-zero the sockaddr info gets filled out, as
requested, but the code fails to initialize the padding bytes of struct
sockaddr_ax25 inserted by the compiler for alignment. Additionally the
msg_namelen value is updated to sizeof(struct full_sockaddr_ax25) but is
not always filled up to this size.

Both issues lead to the fact that the code will leak uninitialized
kernel stack bytes in net/socket.c.

Fix both issues by initializing the memory with memset(0).

Cc: Ralf Baechle <ralf@linux-mips.org>
Signed-off-by: Mathias Krause <minipli@googlemail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoatm: update msg_namelen in vcc_recvmsg()
Mathias Krause [Sun, 7 Apr 2013 01:51:47 +0000 (01:51 +0000)]
atm: update msg_namelen in vcc_recvmsg()

The current code does not fill the msg_name member in case it is set.
It also does not set the msg_namelen member to 0 and therefore makes
net/socket.c leak the local, uninitialized sockaddr_storage variable
to userland -- 128 bytes of kernel stack memory.

Fix that by simply setting msg_namelen to 0 as obviously nobody cared
about vcc_recvmsg() not filling the msg_name in case it was set.

Signed-off-by: Mathias Krause <minipli@googlemail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoMerge git://git.kernel.org/pub/scm/virt/kvm/kvm
Linus Torvalds [Sun, 7 Apr 2013 20:01:25 +0000 (13:01 -0700)]
Merge git://git.kernel.org/pub/scm/virt/kvm/kvm

Pull KVM fix from Gleb Natapov:
 "Bugfix for the regression introduced by commit c300aa64ddf5"

* git://git.kernel.org/pub/scm/virt/kvm/kvm:
  KVM: Allow cross page reads and writes from cached translations.

11 years agoMerge branch 'x86-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Sun, 7 Apr 2013 19:59:55 +0000 (12:59 -0700)]
Merge branch 'x86-urgent-for-linus' of git://git./linux/kernel/git/tip/tip

Pull x86 fixes from Peter Anvin:
 "Two quite small fixes: one a build problem, and the other fixes
  seccomp filters on x32."

* 'x86-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  x86: Fix rebuild with EFI_STUB enabled
  x86: remove the x32 syscall bitmask from syscall_get_nr()

11 years agoalpha: irq: remove deprecated use of IRQF_DISABLED
Will Deacon [Sun, 7 Apr 2013 09:36:12 +0000 (21:36 +1200)]
alpha: irq: remove deprecated use of IRQF_DISABLED

Interrupt handlers are always invoked with interrupts disabled, so
remove all uses of the deprecated IRQF_DISABLED flag.

Cc: Richard Henderson <rth@twiddle.net>
Cc: Ivan Kokshaysky <ink@jurassic.park.msu.ru>
Reviewed-by: Matt Turner <mattst88@gmail.com>
Signed-off-by: Will Deacon <will.deacon@arm.com>
Signed-off-by: Michael Cree <mcree@orcon.net.nz>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
11 years agoalpha: irq: run all handlers with interrupts disabled
Will Deacon [Sun, 7 Apr 2013 09:36:11 +0000 (21:36 +1200)]
alpha: irq: run all handlers with interrupts disabled

Linux has expected that interrupt handlers are executed with local
interrupts disabled for a while now, so ensure that this is the case on
Alpha even for non-device interrupts such as IPIs.

Without this patch, secondary boot results in the following backtrace:

  warning: at kernel/softirq.c:139 __local_bh_enable+0xb8/0xd0()
  trace:
    __local_bh_enable+0xb8/0xd0
    irq_enter+0x74/0xa0
    scheduler_ipi+0x50/0x100
    handle_ipi+0x84/0x260
    do_entint+0x1ac/0x2e0
    irq_exit+0x60/0xa0
    handle_irq+0x98/0x100
    do_entint+0x2c8/0x2e0
    ret_from_sys_call+0x0/0x10
    load_balance+0x3e4/0x870
    cpu_idle+0x24/0x80
    rcu_eqs_enter_common.isra.38+0x0/0x120
    cpu_idle+0x40/0x80
    rest_init+0xc0/0xe0
    _stext+0x1c/0x20

A similar dump occurs if you try to reboot using magic-sysrq.

Cc: Richard Henderson <rth@twiddle.net>
Cc: Ivan Kokshaysky <ink@jurassic.park.msu.ru>
Cc: Matt Turner <mattst88@gmail.com>
Reviewed-by: Matt Turner <mattst88@gmail.com>
Signed-off-by: Will Deacon <will.deacon@arm.com>
Signed-off-by: Michael Cree <mcree@orcon.net.nz>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
11 years agoalpha: makefile: don't enforce small data model for kernel builds
Will Deacon [Sun, 7 Apr 2013 09:36:10 +0000 (21:36 +1200)]
alpha: makefile: don't enforce small data model for kernel builds

Due to all of the goodness being packed into today's kernels, the
resulting image isn't as slim as it once was.

In light of this, don't pass -msmall-data to gcc, which otherwise results
in link failures due to impossible relocations when compiling anything but
the most trivial configurations.

Cc: Richard Henderson <rth@twiddle.net>
Cc: Ivan Kokshaysky <ink@jurassic.park.msu.ru>
Reviewed-by: Matt Turner <mattst88@gmail.com>
Tested-by: Thorsten Kranzkowski <dl8bcu@dl8bcu.de>
Signed-off-by: Will Deacon <will.deacon@arm.com>
Signed-off-by: Michael Cree <mcree@orcon.net.nz>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
11 years agoalpha: Add irongate_io to PCI bus resources
Jay Estabrook [Sun, 7 Apr 2013 09:36:09 +0000 (21:36 +1200)]
alpha: Add irongate_io to PCI bus resources

Fixes a NULL pointer dereference at boot on UP1500.

Cc: stable@vger.kernel.org
Reviewed-and-Tested-by: Matt Turner <mattst88@gmail.com>
Signed-off-by: Jay Estabrook <jay.estabrook@gmail.com>
Signed-off-by: Matt Turner <mattst88@gmail.com>
Signed-off-by: Michael Cree <mcree@orcon.net.nz>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
11 years agoipv6/tcp: Stop processing ICMPv6 redirect messages
Christoph Paasch [Sun, 7 Apr 2013 04:53:15 +0000 (04:53 +0000)]
ipv6/tcp: Stop processing ICMPv6 redirect messages

Tetja Rediske found that if the host receives an ICMPv6 redirect message
after sending a SYN+ACK, the connection will be reset.

He bisected it down to 093d04d (ipv6: Change skb->data before using
icmpv6_notify() to propagate redirect), but the origin of the bug comes
from ec18d9a26 (ipv6: Add redirect support to all protocol icmp error
handlers.). The bug simply did not trigger prior to 093d04d, because
skb->data did not point to the inner IP header and thus icmpv6_notify
did not call the correct err_handler.

This patch adds the missing "goto out;" in tcp_v6_err. After receiving
an ICMPv6 Redirect, we should not continue processing the ICMP in
tcp_v6_err, as this may trigger the removal of request-socks or setting
sk_err(_soft).

Reported-by: Tetja Rediske <tetja@tetja.de>
Signed-off-by: Christoph Paasch <christoph.paasch@uclouvain.be>
Acked-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoremoteproc: fix FW_CONFIG typo
Robert Tivy [Fri, 29 Mar 2013 01:41:43 +0000 (18:41 -0700)]
remoteproc: fix FW_CONFIG typo

Fix obvious typo introduced in commit e121aefa7d9f10eee5cf26ed47129237a05d940b
("remoteproc: fix missing CONFIG_FW_LOADER configurations").

Cc: stable@vger.kernel.org
Signed-off-by: Robert Tivy <rtivy@ti.com>
[cc stable, slight subject change]
Signed-off-by: Ohad Ben-Cohen <ohad@wizery.com>
11 years agoKVM: Allow cross page reads and writes from cached translations.
Andrew Honig [Fri, 29 Mar 2013 16:35:21 +0000 (09:35 -0700)]
KVM: Allow cross page reads and writes from cached translations.

This patch adds support for kvm_gfn_to_hva_cache_init functions for
reads and writes that will cross a page.  If the range falls within
the same memslot, then this will be a fast operation.  If the range
is split between two memslots, then the slower kvm_read_guest and
kvm_write_guest are used.

Tested: Test against kvm_clock unit tests.

Signed-off-by: Andrew Honig <ahonig@google.com>
Signed-off-by: Gleb Natapov <gleb@redhat.com>
11 years agonetfilter: nf_ct_sip: don't drop packets with offsets pointing outside the packet
Patrick McHardy [Fri, 5 Apr 2013 08:13:30 +0000 (08:13 +0000)]
netfilter: nf_ct_sip: don't drop packets with offsets pointing outside the packet

Some Cisco phones create huge messages that are spread over multiple packets.
After calculating the offset of the SIP body, it is validated to be within
the packet and the packet is dropped otherwise. This breaks operation of
these phones. Since connection tracking is supposed to be passive, just let
those packets pass unmodified and untracked.

Signed-off-by: Patrick McHardy <kaber@trash.net>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
11 years agoMerge tag 'dm-3.9-fixes-2' of git://git.kernel.org/pub/scm/linux/kernel/git/agk/linux-dm
Linus Torvalds [Sat, 6 Apr 2013 02:30:14 +0000 (19:30 -0700)]
Merge tag 'dm-3.9-fixes-2' of git://git./linux/kernel/git/agk/linux-dm

Pull device-mapper fixes from Alasdair Kergon:
 "A pair of patches to fix the writethrough mode of the device-mapper
  cache target when the device being cached is not itself wrapped with
  device-mapper."

* tag 'dm-3.9-fixes-2' of git://git.kernel.org/pub/scm/linux/kernel/git/agk/linux-dm:
  dm cache: reduce bio front_pad size in writeback mode
  dm cache: fix writes to cache device in writethrough mode

11 years agoMerge tag 'pci-v3.9-fixes-1' of git://git.kernel.org/pub/scm/linux/kernel/git/helgaas/pci
Linus Torvalds [Sat, 6 Apr 2013 02:29:36 +0000 (19:29 -0700)]
Merge tag 'pci-v3.9-fixes-1' of git://git./linux/kernel/git/helgaas/pci

Pull PCI fixes from Bjorn Helgaas:
 "PCI updates for v3.9:

  ASPM
      Revert "PCI/ACPI: Request _OSC control before scanning PCI root bus"
  kexec
      PCI: Don't try to disable Bus Master on disconnected PCI devices
  Platform ROM images
      PCI: Add PCI ROM helper for platform-provided ROM images
      nouveau: Attempt to use platform-provided ROM image
      radeon: Attempt to use platform-provided ROM image
  Hotplug
      PCI/ACPI: Always resume devices on ACPI wakeup notifications
      PCI/PM: Disable runtime PM of PCIe ports
  EISA
      EISA/PCI: Fix bus res reference
      EISA/PCI: Init EISA early, before PNP"

* tag 'pci-v3.9-fixes-1' of git://git.kernel.org/pub/scm/linux/kernel/git/helgaas/pci:
  PCI/PM: Disable runtime PM of PCIe ports
  PCI/ACPI: Always resume devices on ACPI wakeup notifications
  PCI: Don't try to disable Bus Master on disconnected PCI devices
  Revert "PCI/ACPI: Request _OSC control before scanning PCI root bus"
  radeon: Attempt to use platform-provided ROM image
  nouveau: Attempt to use platform-provided ROM image
  EISA/PCI: Init EISA early, before PNP
  EISA/PCI: Fix bus res reference
  PCI: Add PCI ROM helper for platform-provided ROM images

11 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net
Linus Torvalds [Fri, 5 Apr 2013 21:04:10 +0000 (14:04 -0700)]
Merge git://git./linux/kernel/git/davem/net

Pull networking fixes from David Miller:

 1) Fix erroneous sock_orphan() leading to crashes and double
    kfree_skb() in NFC protocol.  From Thierry Escande and Samuel Ortiz.

 2) Fix use after free in remain-on-channel mac80211 code, from Johannes
    Berg.

 3) nf_reset() needs to reset the NF tracing cookie, otherwise we can
    leak it from one namespace into another.  Fix from Gao Feng and
    Patrick McHardy.

 4) Fix overflow in channel scanning array of mwifiex driver, from Stone
    Piao.

 5) Fix loss of link after suspend/shutdown in r8169, from Hayes Wang.

 6) Synchronization of unicast address lists to the undelying device
    doesn't work because whether to sync is maintained as a boolean
    rather than a true count.  Fix from Vlad Yasevich.

 7) Fix corruption of TSO packets in atl1e by limiting the segmented
    packet length.  From Hannes Frederic Sowa.

 8) Revert bogus AF_UNIX credential passing change and fix the
    coalescing issue properly, from Eric W Biederman.

 9) Changes of ipv4 address lifetime settings needs to generate a
    notification, from Jiri Pirko.

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net: (22 commits)
  netfilter: don't reset nf_trace in nf_reset()
  net: ipv4: notify when address lifetime changes
  ixgbe: fix registration order of driver and DCA nofitication
  af_unix: If we don't care about credentials coallesce all messages
  Revert "af_unix: dont send SCM_CREDENTIAL when dest socket is NULL"
  bonding: remove sysfs before removing devices
  atl1e: limit gso segment size to prevent generation of wrong ip length fields
  net: count hw_addr syncs so that unsync works properly.
  r8169: fix auto speed down issue
  netfilter: ip6t_NPT: Fix translation for non-multiple of 32 prefix lengths
  mwifiex: limit channel number not to overflow memory
  NFC: microread: Fix build failure due to a new MEI bus API
  iwlwifi: dvm: fix the passive-no-RX workaround
  netfilter: nf_conntrack: fix error return code
  NFC: llcp: Keep the connected socket parent pointer alive
  mac80211: fix idle handling sequence
  netfilter: nfnetlink_acct: return -EINVAL if object name is empty
  netfilter: nfnetlink_queue: fix error return code in nfnetlink_queue_init()
  netfilter: reset nf_trace in nf_reset
  mac80211: fix remain-on-channel cancel crash
  ...

11 years agoSUNRPC: Fix a potential memory leak in rpc_new_client
Trond Myklebust [Fri, 5 Apr 2013 18:13:21 +0000 (14:13 -0400)]
SUNRPC: Fix a potential memory leak in rpc_new_client

If the call to rpciod_up() fails, we currently leak a reference to the
struct rpc_xprt.
As part of the fix, we also remove the redundant check for xprt!=NULL.
This is already taken care of by the callers.

Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
11 years agox86: Fix rebuild with EFI_STUB enabled
Jan Beulich [Wed, 3 Apr 2013 14:47:33 +0000 (15:47 +0100)]
x86: Fix rebuild with EFI_STUB enabled

eboot.o and efi_stub_$(BITS).o didn't get added to "targets", and hence
their .cmd files don't get included by the build machinery, leading to
the files always getting rebuilt.

Rather than adding the two files individually, take the opportunity and
add $(VMLINUX_OBJS) to "targets" instead, thus allowing the assignment
at the top of the file to be shrunk quite a bit.

At the same time, remove a pointless flags override line - the variable
assigned to was misspelled anyway, and the options added are
meaningless for assembly sources.

[ hpa: the patch is not minimal, but I am taking it for -urgent anyway
  since the excess impact of the patch seems to be small enough. ]

Signed-off-by: Jan Beulich <jbeulich@suse.com>
Link: http://lkml.kernel.org/r/515C5D2502000078000CA6AD@nat28.tlf.novell.com
Cc: Matthew Garrett <mjg@redhat.com>
Cc: Matt Fleming <matt.fleming@intel.com>
Signed-off-by: H. Peter Anvin <hpa@linux.intel.com>
11 years agoNFSv4/4.1: Fix bugs in nfs4[01]_walk_client_list
Trond Myklebust [Fri, 5 Apr 2013 20:11:11 +0000 (16:11 -0400)]
NFSv4/4.1: Fix bugs in nfs4[01]_walk_client_list

It is unsafe to use list_for_each_entry_safe() here, because
when we drop the nn->nfs_client_lock, we pin the _current_ list
entry and ensure that it stays in the list, but we don't do the
same for the _next_ list entry. Use of list_for_each_entry() is
therefore the correct thing to do.

Also fix the refcounting in nfs41_walk_client_list().

Finally, ensure that the nfs_client has finished being initialised
and, in the case of NFSv4.1, that the session is set up.

Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
Cc: Chuck Lever <chuck.lever@oracle.com>
Cc: Bryan Schumaker <bjschuma@netapp.com>
Cc: stable@vger.kernel.org [>= 3.7]
11 years agoNFSv4: Fix a memory leak in nfs4_discover_server_trunking
Trond Myklebust [Thu, 4 Apr 2013 19:55:00 +0000 (15:55 -0400)]
NFSv4: Fix a memory leak in nfs4_discover_server_trunking

When we assign a new rpc_client to clp->cl_rpcclient, we need to destroy
the old one.

Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
Cc: Chuck Lever <chuck.lever@oracle.com>
Cc: stable@vger.kernel.org [>=3.7]
11 years agoSUNRPC: Remove extra xprt_put()
Chuck Lever [Fri, 22 Mar 2013 16:52:59 +0000 (12:52 -0400)]
SUNRPC: Remove extra xprt_put()

While testing error cases where rpc_new_client() fails, I saw
some oopses.

If rpc_new_client() fails, it already invokes xprt_put().  Thus
__rpc_clone_client() does not need to invoke it again.

Introduced by commit 1b63a751 "SUNRPC: Refactor rpc_clone_client()"
Fri Sep 14, 2012.

Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Cc: stable@vger.kernel.org [>=3.7]
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
11 years agonetfilter: don't reset nf_trace in nf_reset()
Patrick McHardy [Fri, 5 Apr 2013 18:42:05 +0000 (20:42 +0200)]
netfilter: don't reset nf_trace in nf_reset()

Commit 130549fe ("netfilter: reset nf_trace in nf_reset") added code
to reset nf_trace in nf_reset(). This is wrong and unnecessary.

nf_reset() is used in the following cases:

- when passing packets up the the socket layer, at which point we want to
  release all netfilter references that might keep modules pinned while
  the packet is queued. nf_trace doesn't matter anymore at this point.

- when encapsulating or decapsulating IPsec packets. We want to continue
  tracing these packets after IPsec processing.

- when passing packets through virtual network devices. Only devices on
  that encapsulate in IPv4/v6 matter since otherwise nf_trace is not
  used anymore. Its not entirely clear whether those packets should
  be traced after that, however we've always done that.

- when passing packets through virtual network devices that make the
  packet cross network namespace boundaries. This is the only cases
  where we clearly want to reset nf_trace and is also what the
  original patch intended to fix.

Add a new function nf_reset_trace() and use it in dev_forward_skb() to
fix this properly.

Signed-off-by: Patrick McHardy <kaber@trash.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoMerge branch 'upstream' of git://git.linux-mips.org/pub/scm/ralf/upstream-linus
Linus Torvalds [Fri, 5 Apr 2013 19:23:12 +0000 (12:23 -0700)]
Merge branch 'upstream' of git://git.linux-mips.org/ralf/upstream-linus

Pull MIPS fixes from Ralf Baechle:
 "Fixes for a number of small glitches in various corners of the MIPS
  tree.  No particular areas is standing out.

  With this applied all MIPS defconfigs are building fine.  No merge
  conflicts are expected."

* 'upstream' of git://git.linux-mips.org/pub/scm/ralf/upstream-linus:
  MIPS: Delete definition of SA_RESTORER.
  MIPS: Fix ISA level which causes secondary cache init bypassing and more
  MIPS: Fix build error cavium-octeon without CONFIG_SMP
  MIPS: Kconfig: Rename SNIPROM too
  MIPS: Alchemy: Fix typo "CONFIG_DEBUG_PCI"
  MIPS: Unbreak function tracer for 64-bit kernel.