fs/cifs/cifs_dfs_ref.c: fix potential memory leakage

commit 10b8c7dff5d3633b69e77f57d404dab54ead3787 upstream.

When it goes to error through line 144, the memory allocated to *devname is
not freed, and the caller doesn't free it either in line 250. So we free the
memroy of *devname in function cifs_compose_mount_options() when it goes to
error.

Signed-off-by: Cong Ding <dinggnu@gmail.com>
Reviewed-by: Jeff Layton <jlayton@redhat.com>
Signed-off-by: Steve French <smfrench@gmail.com>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>

tmpfs: fix use-after-free of mempolicy object

commit 5f00110f7273f9ff04ac69a5f85bb535a4fd0987 upstream.

The tmpfs remount logic preserves filesystem mempolicy if the mpol=M
option is not specified in the remount request.  A new policy can be
specified if mpol=M is given.

Before this patch remounting an mpol bound tmpfs without specifying
mpol= mount option in the remount request would set the filesystem's
mempolicy object to a freed mempolicy object.

To reproduce the problem boot a DEBUG_PAGEALLOC kernel and run:
    # mkdir /tmp/x

    # mount -t tmpfs -o size=100M,mpol=interleave nodev /tmp/x

    # grep /tmp/x /proc/mounts
    nodev /tmp/x tmpfs rw,relatime,size=102400k,mpol=interleave:0-3 0 0

    # mount -o remount,size=200M nodev /tmp/x

    # grep /tmp/x /proc/mounts
    nodev /tmp/x tmpfs rw,relatime,size=204800k,mpol=??? 0 0
        # note ? garbage in mpol=... output above

    # dd if=/dev/zero of=/tmp/x/f count=1
        # panic here

Panic:
    BUG: unable to handle kernel NULL pointer dereference at           (null)
    IP: [<          (null)>]           (null)
    [...]
    Oops: 0010 [#1] SMP DEBUG_PAGEALLOC
    Call Trace:
      mpol_shared_policy_init+0xa5/0x160
      shmem_get_inode+0x209/0x270
      shmem_mknod+0x3e/0xf0
      shmem_create+0x18/0x20
      vfs_create+0xb5/0x130
      do_last+0x9a1/0xea0
      path_openat+0xb3/0x4d0
      do_filp_open+0x42/0xa0
      do_sys_open+0xfe/0x1e0
      compat_sys_open+0x1b/0x20
      cstar_dispatch+0x7/0x1f

Non-debug kernels will not crash immediately because referencing the
dangling mpol will not cause a fault.  Instead the filesystem will
reference a freed mempolicy object, which will cause unpredictable
behavior.

The problem boils down to a dropped mpol reference below if
shmem_parse_options() does not allocate a new mpol:

    config = *sbinfo
    shmem_parse_options(data, &config, true)
    mpol_put(sbinfo->mpol)
    sbinfo->mpol = config.mpol  /* BUG: saves unreferenced mpol */

This patch avoids the crash by not releasing the mempolicy if
shmem_parse_options() doesn't create a new mpol.

How far back does this issue go? I see it in both 2.6.36 and 3.3.  I did
not look back further.

Signed-off-by: Greg Thelen <gthelen@google.com>
Acked-by: Hugh Dickins <hughd@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>

sysfs: sysfs_pathname/sysfs_add_one: Use strlcat() instead of strcat()

commit 66081a72517a131430dcf986775f3268aafcb546 upstream.

The warning check for duplicate sysfs entries can cause a buffer overflow
when printing the warning, as strcat() doesn't check buffer sizes.
Use strlcat() instead.

Since strlcat() doesn't return a pointer to the passed buffer, unlike
strcat(), I had to convert the nested concatenation in sysfs_add_one() to
an admittedly more obscure comma operator construct, to avoid emitting code
for the concatenation if CONFIG_BUG is disabled.

Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>

fs/fscache/stats.c: fix memory leak

commit ec686c9239b4d472052a271c505d04dae84214cc upstream.

There is a kernel memory leak observed when the proc file
/proc/fs/fscache/stats is read.

The reason is that in fscache_stats_open, single_open is called and the
respective release function is not called during release.  Hence fix
with correct release function - single_release().

Addresses https://bugzilla.kernel.org/show_bug.cgi?id=57101

Signed-off-by: Anurup m <anurup.m@huawei.com>
Cc: shyju pv <shyju.pv@huawei.com>
Cc: Sanil kumar <sanil.kumar@huawei.com>
Cc: Nataraj m <nataraj.m@huawei.com>
Cc: Li Zefan <lizefan@huawei.com>
Cc: David Howells <dhowells@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>

fs/compat_ioctl.c: VIDEO_SET_SPU_PALETTE missing error check

commit 12176503366885edd542389eed3aaf94be163fdb upstream.

The compat ioctl for VIDEO_SET_SPU_PALETTE was missing an error check
while converting ioctl arguments.  This could lead to leaking kernel
stack contents into userspace.

Patch extracted from existing fix in grsecurity.

Signed-off-by: Kees Cook <keescook@chromium.org>
Cc: David Miller <davem@davemloft.net>
Cc: Brad Spengler <spender@grsecurity.net>
Cc: PaX Team <pageexec@freemail.hu>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>

epoll: prevent missed events on EPOLL_CTL_MOD

commit 128dd1759d96ad36c379240f8b9463e8acfd37a1 upstream.

EPOLL_CTL_MOD sets the interest mask before calling f_op->poll() to
ensure events are not missed.  Since the modifications to the interest
mask are not protected by the same lock as ep_poll_callback, we need to
ensure the change is visible to other CPUs calling ep_poll_callback.

We also need to ensure f_op->poll() has an up-to-date view of past
events which occured before we modified the interest mask.  So this
barrier also pairs with the barrier in wq_has_sleeper().

This should guarantee either ep_poll_callback or f_op->poll() (or both)
will notice the readiness of a recently-ready/modified item.

This issue was encountered by Andreas Voellmy and Junchang(Jason) Wang in:
http://thread.gmane.org/gmane.linux.kernel/1408782/

Signed-off-by: Eric Wong <normalperson@yhbt.net>
Cc: Hans Verkuil <hans.verkuil@cisco.com>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Jonathan Corbet <corbet@lwn.net>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: Davide Libenzi <davidel@xmailserver.org>
Cc: Hans de Goede <hdegoede@redhat.com>
Cc: Mauro Carvalho Chehab <mchehab@infradead.org>
Cc: David Miller <davem@davemloft.net>
Cc: Eric Dumazet <eric.dumazet@gmail.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Andreas Voellmy <andreas.voellmy@yale.edu>
Tested-by: "Junchang(Jason) Wang" <junchang.wang@yale.edu>
Cc: netdev@vger.kernel.org
Cc: linux-fsdevel@vger.kernel.org
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>

usbdevfs: Correct amount of data copied to user in processcompl_compat

commit 2102e06a5f2e414694921f23591f072a5ba7db9f upstream.

iso data buffers may have holes in them if some packets were short, so for
iso urbs we should always copy the entire buffer, just like the regular
processcompl does.

Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Acked-by: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>

USB: serial: fix race between probe and open

commit a65a6f14dc24a90bde3f5d0073ba2364476200bf upstream.

Fix race between probe and open by making sure that the disconnected
flag is not cleared until all ports have been registered.

A call to tty_open while probe is running may get a reference to the
serial structure in serial_install before its ports have been
registered. This may lead to usb_serial_core calling driver open before
port is fully initialised.

With ftdi_sio this result in the following NULL-pointer dereference as
the private data has not been initialised at open:

[  199.698286] IP: [<f811a089>] ftdi_open+0x59/0xe0 [ftdi_sio]
[  199.698297] *pde = 00000000
[  199.698303] Oops: 0000 [#1] PREEMPT SMP
[  199.698313] Modules linked in: ftdi_sio usbserial
[  199.698323]
[  199.698327] Pid: 1146, comm: ftdi_open Not tainted 3.2.11 #70 Dell Inc. Vostro 1520/0T816J
[  199.698339] EIP: 0060:[<f811a089>] EFLAGS: 00010286 CPU: 0
[  199.698344] EIP is at ftdi_open+0x59/0xe0 [ftdi_sio]
[  199.698348] EAX: 0000003e EBX: f5067000 ECX: 00000000 EDX: 80000600
[  199.698352] ESI: f48d8800 EDI: 00000001 EBP: f515dd54 ESP: f515dcfc
[  199.698356]  DS: 007b ES: 007b FS: 00d8 GS: 0033 SS: 0068
[  199.698361] Process ftdi_open (pid: 1146, ti=f515c000 task=f481e040 task.ti=f515c000)
[  199.698364] Stack:
[  199.698368]  f811a9fe f811a9e0 f811b3ef 00000000 00000000 00001388 00000000 f4a86800
[  199.698387]  00000002 00000000 f806e68e 00000000 f532765c f481e040 00000246 22222222
[  199.698479]  22222222 22222222 22222222 f5067004 f5327600 f5327638 f515dd74 f806e6ab
[  199.698496] Call Trace:
[  199.698504]  [<f806e68e>] ? serial_activate+0x2e/0x70 [usbserial]
[  199.698511]  [<f806e6ab>] serial_activate+0x4b/0x70 [usbserial]
[  199.698521]  [<c126380c>] tty_port_open+0x7c/0xd0
[  199.698527]  [<f806e660>] ? serial_set_termios+0xa0/0xa0 [usbserial]
[  199.698534]  [<f806e76f>] serial_open+0x2f/0x70 [usbserial]
[  199.698540]  [<c125d07c>] tty_open+0x20c/0x510
[  199.698546]  [<c10e9eb7>] chrdev_open+0xe7/0x230
[  199.698553]  [<c10e48f2>] __dentry_open+0x1f2/0x390
[  199.698559]  [<c144bfec>] ? _raw_spin_unlock+0x2c/0x50
[  199.698565]  [<c10e4b76>] nameidata_to_filp+0x66/0x80
[  199.698570]  [<c10e9dd0>] ? cdev_put+0x20/0x20
[  199.698576]  [<c10f3e08>] do_last+0x198/0x730
[  199.698581]  [<c10f4440>] path_openat+0xa0/0x350
[  199.698587]  [<c10f47d5>] do_filp_open+0x35/0x80
[  199.698593]  [<c144bfec>] ? _raw_spin_unlock+0x2c/0x50
[  199.698599]  [<c10ff110>] ? alloc_fd+0xc0/0x100
[  199.698605]  [<c10f0b72>] ? getname_flags+0x72/0x120
[  199.698611]  [<c10e4450>] do_sys_open+0xf0/0x1c0
[  199.698617]  [<c11fcc08>] ? trace_hardirqs_on_thunk+0xc/0x10
[  199.698623]  [<c10e458e>] sys_open+0x2e/0x40
[  199.698628]  [<c144c990>] sysenter_do_call+0x12/0x36
[  199.698632] Code: 85 89 00 00 00 8b 16 8b 4d c0 c1 e2 08 c7 44 24 14 88 13 00 00 81 ca 00 00 00 80 c7 44 24 10 00 00 00 00 c7 44 24 0c 00 00 00 00 <0f> b7 41 78 31 c9 89 44 24 08 c7 44 24 04 00 00 00 00 c7 04 24
[  199.698884] EIP: [<f811a089>] ftdi_open+0x59/0xe0 [ftdi_sio] SS:ESP 0068:f515dcfc
[  199.698893] CR2: 0000000000000078
[  199.698925] ---[ end trace 77c43ec023940cff ]---

Reported-and-tested-by: Ken Huang <csuhgw@gmail.com>
Signed-off-by: Johan Hovold <jhovold@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>

USB: cdc-wdm: fix lockup on error in wdm_read

commit b086b6b10d9f182cd8d2f0dcfd7fd11edba93fc9 upstream.

Clear the WDM_READ flag on empty reads to avoid running
forever in an infinite tight loop, causing lockups:

Jul  1 21:58:11 nemi kernel: [ 3658.898647] qmi_wwan 2-1:1.2: Unexpected error -71
Jul  1 21:58:36 nemi kernel: [ 3684.072021] BUG: soft lockup - CPU#0 stuck for 23s! [qmi.pl:12235]
Jul  1 21:58:36 nemi kernel: [ 3684.072212] CPU 0
Jul  1 21:58:36 nemi kernel: [ 3684.072355]
Jul  1 21:58:36 nemi kernel: [ 3684.072367] Pid: 12235, comm: qmi.pl Tainted: P           O 3.5.0-rc2+ #13 LENOVO 2776LEG/2776LEG
Jul  1 21:58:36 nemi kernel: [ 3684.072383] RIP: 0010:[<ffffffffa0635008>]  [<ffffffffa0635008>] spin_unlock_irq+0x8/0xc [cdc_wdm]
Jul  1 21:58:36 nemi kernel: [ 3684.072388] RSP: 0018:ffff88022dca1e70  EFLAGS: 00000282
Jul  1 21:58:36 nemi kernel: [ 3684.072393] RAX: ffff88022fc3f650 RBX: ffffffff811c56f7 RCX: 00000001000ce8c1
Jul  1 21:58:36 nemi kernel: [ 3684.072398] RDX: 0000000000000010 RSI: 000000000267d810 RDI: ffff88022fc3f650
Jul  1 21:58:36 nemi kernel: [ 3684.072403] RBP: ffff88022dca1eb0 R08: ffffffffa063578e R09: 0000000000000000
Jul  1 21:58:36 nemi kernel: [ 3684.072407] R10: 0000000000000008 R11: 0000000000000246 R12: 0000000000000002
Jul  1 21:58:36 nemi kernel: [ 3684.072412] R13: 0000000000000246 R14: ffffffff00000002 R15: ffff8802281d8c88
Jul  1 21:58:36 nemi kernel: [ 3684.072418] FS:  00007f666a260700(0000) GS:ffff88023bc00000(0000) knlGS:0000000000000000
Jul  1 21:58:36 nemi kernel: [ 3684.072423] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
Jul  1 21:58:36 nemi kernel: [ 3684.072428] CR2: 000000000270d9d8 CR3: 000000022e865000 CR4: 00000000000007f0
Jul  1 21:58:36 nemi kernel: [ 3684.072433] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
Jul  1 21:58:36 nemi kernel: [ 3684.072438] DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400
Jul  1 21:58:36 nemi kernel: [ 3684.072444] Process qmi.pl (pid: 12235, threadinfo ffff88022dca0000, task ffff88022ff76380)
Jul  1 21:58:36 nemi kernel: [ 3684.072448] Stack:
Jul  1 21:58:36 nemi kernel: [ 3684.072458]  ffffffffa063592e 0000000100020000 ffff88022fc3f650 ffff88022fc3f6a8
Jul  1 21:58:36 nemi kernel: [ 3684.072466]  0000000000000200 0000000100000000 000000000267d810 0000000000000000
Jul  1 21:58:36 nemi kernel: [ 3684.072475]  0000000000000000 ffff880212cfb6d0 0000000000000200 ffff880212cfb6c0
Jul  1 21:58:36 nemi kernel: [ 3684.072479] Call Trace:
Jul  1 21:58:36 nemi kernel: [ 3684.072489]  [<ffffffffa063592e>] ? wdm_read+0x1a0/0x263 [cdc_wdm]
Jul  1 21:58:36 nemi kernel: [ 3684.072500]  [<ffffffff8110adb7>] ? vfs_read+0xa1/0xfb
Jul  1 21:58:36 nemi kernel: [ 3684.072509]  [<ffffffff81040589>] ? alarm_setitimer+0x35/0x64
Jul  1 21:58:36 nemi kernel: [ 3684.072517]  [<ffffffff8110aec7>] ? sys_read+0x45/0x6e
Jul  1 21:58:36 nemi kernel: [ 3684.072525]  [<ffffffff813725f9>] ? system_call_fastpath+0x16/0x1b
Jul  1 21:58:36 nemi kernel: [ 3684.072557] Code: <66> 66 90 c3 83 ff ed 89 f8 74 16 7f 06 83 ff a1 75 0a c3 83 ff f4

The WDM_READ flag is normally cleared by wdm_int_callback
before resubmitting the read urb, and set by wdm_in_callback
when this urb returns with data or an error.  But a crashing
device may cause both a read error and cancelling all urbs.
Make sure that the flag is cleared by wdm_read if the buffer
is empty.

We don't clear the flag on errors, as there may be pending
data in the buffer which should be processed.  The flag will
instead be cleared on the next wdm_read call.

Signed-off-by: Bjørn Mork <bjorn@mork.no>
Acked-by: Oliver Neukum <oneukum@suse.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>

USB: kaweth.c: use GFP_ATOMIC under spin_lock

commit e4c7f259c5be99dcfc3d98f913590663b0305bf8 upstream.

The problem is that we call this with a spin lock held.  The call tree
is:
kaweth_start_xmit() holds kaweth->device_lock.
-> kaweth_async_set_rx_mode()
   -> kaweth_control()
      -> kaweth_internal_control_msg()

The kaweth_internal_control_msg() function is only called from
kaweth_control() which used GFP_ATOMIC for its allocations.

Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>

USB: echi-dbgp: increase the controller wait time to come out of halt.

commit f96a4216e85050c0a9d41a41ecb0ae9d8e39b509 upstream.

The default 10 microsecond delay for the controller to come out of
halt in dbgp_ehci_startup is too short, so increase it to 1 millisecond.

This is based on emperical testing on various USB debug ports on
modern machines such as a Lenovo X220i and an Ivybridge development
platform that needed to wait ~450-950 microseconds.

Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: Jason Wessel <jason.wessel@windriver.com>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>

usb: serial: mos7840: Fixup mos7840_chars_in_buffer()

commit 5c263b92f828af6a8cf54041db45ceae5af8f2ab upstream.

 * Use the buffer content length as opposed to the total buffer size.  This can
   be a real problem when using the mos7840 as a usb serial-console as all
   kernel output is truncated during boot.

Signed-off-by: Mark Ferrell <mferrell@uplogix.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>

USB: CDC ACM: Fix NULL pointer dereference

commit 99f347caa4568cb803862730b3b1f1942639523f upstream.

If a device specifies zero endpoints in its interface descriptor,
the kernel oopses in acm_probe(). Even though that's clearly an
invalid descriptor, we should test wether we have all endpoints.
This is especially bad as this oops can be triggered by just
plugging a USB device in.

Signed-off-by: Sven Schnelle <svens@stackframe.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>

USB: serial: ftdi_sio: Handle the old_termios == 0 case e.g. uart_resume_port()

commit c515598e0f5769916c31c00392cc2bfe6af74e55 upstream.

  Handle null old_termios in ftdi_set_termios() calls from uart_resume_port().

Signed-off-by: Andrew Worsley <amworsley@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>

USB: cdc-wdm: fix buffer overflow

commit c0f5ecee4e741667b2493c742b60b6218d40b3aa upstream.

The buffer for responses must not overflow.
If this would happen, set a flag, drop the data and return
an error after user space has read all remaining data.

Signed-off-by: Oliver Neukum <oliver@neukum.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
[PG: minor adjustment since RESET from 880442027569 isn't in .34]
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>

USB: io_ti: Fix NULL dereference in chase_port()

commit 1ee0a224bc9aad1de496c795f96bc6ba2c394811 upstream.

The tty is NULL when the port is hanging up.
chase_port() needs to check for this.

This patch is intended for stable series.
The behavior was observed and tested in Linux 3.2 and 3.7.1.

Johan Hovold submitted a more elaborate patch for the mainline kernel.

[   56.277883] usb 1-1: edge_bulk_in_callback - nonzero read bulk status received: -84
[   56.278811] usb 1-1: USB disconnect, device number 3
[   56.278856] usb 1-1: edge_bulk_in_callback - stopping read!
[   56.279562] BUG: unable to handle kernel NULL pointer dereference at 00000000000001c8
[   56.280536] IP: [<ffffffff8144e62a>] _raw_spin_lock_irqsave+0x19/0x35
[   56.281212] PGD 1dc1b067 PUD 1e0f7067 PMD 0
[   56.282085] Oops: 0002 [#1] SMP
[   56.282744] Modules linked in:
[   56.283512] CPU 1
[   56.283512] Pid: 25, comm: khubd Not tainted 3.7.1 #1 innotek GmbH VirtualBox/VirtualBox
[   56.283512] RIP: 0010:[<ffffffff8144e62a>]  [<ffffffff8144e62a>] _raw_spin_lock_irqsave+0x19/0x35
[   56.283512] RSP: 0018:ffff88001fa99ab0  EFLAGS: 00010046
[   56.283512] RAX: 0000000000000046 RBX: 00000000000001c8 RCX: 0000000000640064
[   56.283512] RDX: 0000000000010000 RSI: ffff88001fa99b20 RDI: 00000000000001c8
[   56.283512] RBP: ffff88001fa99b20 R08: 0000000000000000 R09: 0000000000000000
[   56.283512] R10: 0000000000000000 R11: ffffffff812fcb4c R12: ffff88001ddf53c0
[   56.283512] R13: 0000000000000000 R14: 00000000000001c8 R15: ffff88001e19b9f4
[   56.283512] FS:  0000000000000000(0000) GS:ffff88001fd00000(0000) knlGS:0000000000000000
[   56.283512] CS:  0010 DS: 0000 ES: 0000 CR0: 000000008005003b
[   56.283512] CR2: 00000000000001c8 CR3: 000000001dc51000 CR4: 00000000000006e0
[   56.283512] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
[   56.283512] DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400
[   56.283512] Process khubd (pid: 25, threadinfo ffff88001fa98000, task ffff88001fa94f80)
[   56.283512] Stack:
[   56.283512]  0000000000000046 00000000000001c8 ffffffff810578ec ffffffff812fcb4c
[   56.283512]  ffff88001e19b980 0000000000002710 ffffffff812ffe81 0000000000000001
[   56.283512]  ffff88001fa94f80 0000000000000202 ffffffff00000001 0000000000000296
[   56.283512] Call Trace:
[   56.283512]  [<ffffffff810578ec>] ? add_wait_queue+0x12/0x3c
[   56.283512]  [<ffffffff812fcb4c>] ? usb_serial_port_work+0x28/0x28
[   56.283512]  [<ffffffff812ffe81>] ? chase_port+0x84/0x2d6
[   56.283512]  [<ffffffff81063f27>] ? try_to_wake_up+0x199/0x199
[   56.283512]  [<ffffffff81263a5c>] ? tty_ldisc_hangup+0x222/0x298
[   56.283512]  [<ffffffff81300171>] ? edge_close+0x64/0x129
[   56.283512]  [<ffffffff810612f7>] ? __wake_up+0x35/0x46
[   56.283512]  [<ffffffff8106135b>] ? should_resched+0x5/0x23
[   56.283512]  [<ffffffff81264916>] ? tty_port_shutdown+0x39/0x44
[   56.283512]  [<ffffffff812fcb4c>] ? usb_serial_port_work+0x28/0x28
[   56.283512]  [<ffffffff8125d38c>] ? __tty_hangup+0x307/0x351
[   56.283512]  [<ffffffff812e6ddc>] ? usb_hcd_flush_endpoint+0xde/0xed
[   56.283512]  [<ffffffff8144e625>] ? _raw_spin_lock_irqsave+0x14/0x35
[   56.283512]  [<ffffffff812fd361>] ? usb_serial_disconnect+0x57/0xc2
[   56.283512]  [<ffffffff812ea99b>] ? usb_unbind_interface+0x5c/0x131
[   56.283512]  [<ffffffff8128d738>] ? __device_release_driver+0x7f/0xd5
[   56.283512]  [<ffffffff8128d9cd>] ? device_release_driver+0x1a/0x25
[   56.283512]  [<ffffffff8128d393>] ? bus_remove_device+0xd2/0xe7
[   56.283512]  [<ffffffff8128b7a3>] ? device_del+0x119/0x167
[   56.283512]  [<ffffffff812e8d9d>] ? usb_disable_device+0x6a/0x180
[   56.283512]  [<ffffffff812e2ae0>] ? usb_disconnect+0x81/0xe6
[   56.283512]  [<ffffffff812e4435>] ? hub_thread+0x577/0xe82
[   56.283512]  [<ffffffff8144daa7>] ? __schedule+0x490/0x4be
[   56.283512]  [<ffffffff8105798f>] ? abort_exclusive_wait+0x79/0x79
[   56.283512]  [<ffffffff812e3ebe>] ? usb_remote_wakeup+0x2f/0x2f
[   56.283512]  [<ffffffff812e3ebe>] ? usb_remote_wakeup+0x2f/0x2f
[   56.283512]  [<ffffffff810570b4>] ? kthread+0x81/0x89
[   56.283512]  [<ffffffff81057033>] ? __kthread_parkme+0x5c/0x5c
[   56.283512]  [<ffffffff8145387c>] ? ret_from_fork+0x7c/0xb0
[   56.283512]  [<ffffffff81057033>] ? __kthread_parkme+0x5c/0x5c
[   56.283512] Code: 8b 7c 24 08 e8 17 0b c3 ff 48 8b 04 24 48 83 c4 10 c3 53 48 89 fb 41 50 e8 e0 0a c3 ff 48 89 04 24 e8 e7 0a c3 ff ba 00 00 01 00
<f0> 0f c1 13 48 8b 04 24 89 d1 c1 ea 10 66 39 d1 74 07 f3 90 66
[   56.283512] RIP  [<ffffffff8144e62a>] _raw_spin_lock_irqsave+0x19/0x35
[   56.283512]  RSP <ffff88001fa99ab0>
[   56.283512] CR2: 00000000000001c8
[   56.283512] ---[ end trace 49714df27e1679ce ]---

Signed-off-by: Wolfgang Frisch <wfpub@roembden.net>
Cc: Johan Hovold <jhovold@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>

USB: garmin_gps: fix memory leak on disconnect

commit 618aa1068df29c37a58045fe940f9106664153fd upstream.

Remove bogus disconnect test introduced by 95bef012e ("USB: more serial
drivers writing after disconnect") which prevented queued data from
being freed on disconnect.

The possible IO it was supposed to prevent is long gone.

Signed-off-by: Johan Hovold <jhovold@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>

USB: mos7840: fix port-device leak in error path

commit 3eb55cc4ed88eee3b5230f66abcdbd2a91639eda upstream.

The driver set the usb-serial port pointers to NULL on errors in attach,
effectively preventing usb-serial core from decrementing the port ref
counters and releasing the port devices and associated data.

Signed-off-by: Johan Hovold <jhovold@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>

USB: mos7840: fix urb leak at release

commit 65a4cdbb170e4ec1a7fa0e94936d47e24a17b0e8 upstream.

Make sure control urb is freed at release.

Signed-off-by: Johan Hovold <jhovold@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>

USB: serial: Fix memory leak in sierra_release()

commit f7bc5051667b74c3861f79eed98c60d5c3b883f7 upstream.

I found a memory leak in sierra_release() (well sierra_probe() I guess)
that looses 8 bytes each time the driver releases a device.

Signed-off-by: Len Sorensen <lsorense@csclub.uwaterloo.ca>
Acked-by: Johan Hovold <jhovold@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>

USB: whiteheat: fix memory leak in error path

commit c129197c99550d356cf5f69b046994dd53cd1b9d upstream.

Make sure command buffer is deallocated in case of errors during attach.

Cc: <support@connecttech.com>
Signed-off-by: Johan Hovold <jhovold@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>

USB: EHCI: go back to using the system clock for QH unlinks

commit 004c19682884d4f40000ce1ded53f4a1d0b18206 upstream.

This patch (as1477) fixes a problem affecting a few types of EHCI
controller.  Contrary to what one might expect, these controllers
automatically stop their internal frame counter when no ports are
enabled.  Since ehci-hcd currently relies on the frame counter for
determining when it should unlink QHs from the async schedule, those
controllers run into trouble: The frame counter stops and the QHs
never get unlinked.

Some systems have also experienced other problems traced back to
commit b963801164618e25fbdc0cd452ce49c3628b46c8 (USB: ehci-hcd unlink
speedups), which made the original switch from using the system clock
to using the frame counter.  It never became clear what the reason was
for these problems, but evidently it is related to use of the frame
counter.

To fix all these problems, this patch more or less reverts that commit
and goes back to using the system clock.  But this can't be done
cleanly because other changes have since been made to the scan_async()
subroutine.  One of these changes involved the tricky logic that tries
to avoid rescanning QHs that have already been seen when the scanning
loop is restarted, which happens whenever an URB is given back.
Switching back to clock-based unlinks would make this logic even more
complicated.

Therefore the new code doesn't rescan the entire async list whenever a
giveback occurs.  Instead it rescans only the current QH and continues
on from there.  This requires the use of a separate pointer to keep
track of the next QH to scan, since the current QH may be unlinked
while the scanning is in progress.  That new pointer must be global,
so that it can be adjusted forward whenever the _next_ QH gets
unlinked.  (uhci-hcd uses this same trick.)

Simplification of the scanning loop removes a level of indentation,
which accounts for the size of the patch.  The amount of code changed
is relatively small, and it isn't exactly a reversion of the
b963801164 commit.

This fixes Bugzilla #32432.

Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
Tested-by: Matej Kenda <matejken@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>

xhci: Don't write zeroed pointers to xHC registers.

commit 159e1fcc9a60fc7daba23ee8fcdb99799de3fe84 upstream.

When xhci_mem_cleanup() is called, we can't be sure if the xHC is
actually halted.  We can ask the xHC to halt by writing to the RUN bit
in the command register, but that might timeout due to a HW hang.

If the host controller is still running, we should not write zeroed
values to the event ring dequeue pointers or base tables, the DCBAA
pointers, or the command ring pointers.  Eric Fu reports his VIA VL800
host accesses the event ring pointers after a failed register restore on
resume from suspend.  The hypothesis is that the host never actually
halted before the register write to change the event ring pointer to
zero.

Remove all writes of zeroed values to pointer registers in
xhci_mem_cleanup().  Instead, make all callers of the function reset the
host controller first, which will reset those registers to zero.
xhci_mem_init() is the only caller that doesn't first halt and reset the
host controller before calling xhci_mem_cleanup().

This should be backported to kernels as old as 2.6.32.

Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Tested-by: Elric Fu <elricfu1@gmail.com>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>

xHCI: Correct the #define XHCI_LEGACY_DISABLE_SMI

commit 95018a53f7653e791bba1f54c8d75d9cb700d1bd upstream.

Re-define XHCI_LEGACY_DISABLE_SMI and used it in right way. All SMI enable
bits will be cleared to zero and flag bits 29:31 are also cleared to zero.
Other bits should be presvered as Table 146.

This patch should be backported to kernels as old as 2.6.31.

Signed-off-by: Alex He <alex.he@amd.com>
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>

xhci: Reset reserved command ring TRBs on cleanup.

commit 33b2831ac870d50cc8e01c317b07fb1e69c13fe1 upstream.

When the xHCI driver needs to clean up memory (perhaps due to a failed
register restore on resume from S3 or resume from S4), it needs to reset
the number of reserved TRBs on the command ring to zero.  Otherwise,
several resume cycles (about 30) with a UAS device attached will
continually increment the number of reserved TRBs, until all command
submissions fail because there isn't enough room on the command ring.

This patch should be backported to kernels as old as 2.6.32,
that contain the commit 913a8a344ffcaf0b4a586d6662a2c66a7106557d
"USB: xhci: Change how xHCI commands are handled."

Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>

xhci: Increase reset timeout for Renesas 720201 host.

commit 22ceac191211cf6688b1bf6ecd93c8b6bf80ed9b upstream.

The NEC/Renesas 720201 xHCI host controller does not complete its reset
within 250 milliseconds.  In fact, it takes about 9 seconds to reset the
host controller, and 1 second for the host to be ready for doorbell
rings.  Extend the reset and CNR polling timeout to 10 seconds each.

This patch should be backported to kernels as old as 2.6.31, that
contain the commit 66d4eadd8d067269ea8fead1a50fe87c2979a80d "USB: xhci:
BIOS handoff and HW initialization."

Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Reported-by: Edwin Klein Mentink <e.kleinmentink@zonnet.nl>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>

xhci: Make handover code more robust

commit e955a1cd086de4d165ae0f4c7be7289d84b63bdc upstream.

My test platform (Intel DX79SI) boots reliably under BIOS, but frequently
crashes when booting via UEFI. I finally tracked this down to the xhci
handoff code. It seems that reads from the device occasionally just return
0xff, resulting in xhci_find_next_cap_offset generating a value that's
larger than the resource region. We then oops when attempting to read the
value. Sanity checking that value lets us avoid the crash.

I've no idea what's causing the underlying problem, and xhci still doesn't
actually *work* even with this, but the machine at least boots which will
probably make further debugging easier.

This should be backported to kernels as old as 2.6.31, that contain the
commit 66d4eadd8d067269ea8fead1a50fe87c2979a80d "USB: xhci: BIOS handoff
and HW initialization."

Signed-off-by: Matthew Garrett <mjg@redhat.com>
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>

Bluetooth: hci_ldisc: fix NULL-pointer dereference on tty_close

commit 33b69bf80a3704d45341928e4ff68b6ebd470686 upstream.

Do not close protocol driver until device has been unregistered.

This fixes a race between tty_close and hci_dev_open which can result in
a NULL-pointer dereference.

The line discipline closes the protocol driver while we may still have
hci_dev_open sleeping on the req_lock mutex resulting in a NULL-pointer
dereference when lock is acquired and hci_init_req called.

Bug is 100% reproducible using hciattach and a disconnected serial port:

0. # hciattach -n ttyO1 any noflow

1. hci_dev_open called from hci_power_on grabs req lock
2. hci_init_req executes but device fails to initialise (times out
   eventually)
3. hci_dev_open is called from hci_sock_ioctl and sleeps on req lock
4. hci_uart_tty_close detaches protocol driver and cancels init req
5. hci_dev_open (1) releases req lock
6. hci_dev_open (3) grabs req lock, calls hci_init_req, which triggers oops
   when request is prepared in hci_uart_send_frame

[  137.201263] Unable to handle kernel NULL pointer dereference at virtual address 00000028
[  137.209838] pgd = c0004000
[  137.212677] [00000028] *pgd=00000000
[  137.216430] Internal error: Oops: 17 [#1]
[  137.220642] Modules linked in:
[  137.223846] CPU: 0    Tainted: G        W     (3.3.0-rc6-dirty #406)
[  137.230529] PC is at __lock_acquire+0x5c/0x1ab0
[  137.235290] LR is at lock_acquire+0x9c/0x128
[  137.239776] pc : [<c0071490>]    lr : [<c00733f8>]    psr: 20000093
[  137.239776] sp : cf869dd8  ip : c0529554  fp : c051c730
[  137.251800] r10: 00000000  r9 : cf8673c0  r8 : 00000080
[  137.257293] r7 : 00000028  r6 : 00000002  r5 : 00000000  r4 : c053fd70
[  137.264129] r3 : 00000000  r2 : 00000000  r1 : 00000000  r0 : 00000001
[  137.270965] Flags: nzCv  IRQs off  FIQs on  Mode SVC_32  ISA ARM  Segment kernel
[  137.278717] Control: 10c5387d  Table: 8f0f4019  DAC: 00000015
[  137.284729] Process kworker/u:1 (pid: 7, stack limit = 0xcf8682e8)
[  137.291229] Stack: (0xcf869dd8 to 0xcf86a000)
[  137.295776] 9dc0:                                                       c0529554 00000000
[  137.304351] 9de0: cf8673c0 cf868000 d03ea1ef cf868000 000001ef 00000470 00000000 00000002
[  137.312927] 9e00: cf8673c0 00000001 c051c730 c00716ec 0000000c 00000440 c0529554 00000001
[  137.321533] 9e20: c051c730 cf868000 d03ea1f3 00000000 c053b978 00000000 00000028 cf868000
[  137.330078] 9e40: 00000000 00000000 00000002 00000000 00000000 c00733f8 00000002 00000080
[  137.338684] 9e60: 00000000 c02a1d50 00000000 00000001 60000013 c0969a1c 60000093 c053b96c
[  137.347259] 9e80: 00000002 00000018 20000013 c02a1d50 cf0ac000 00000000 00000002 cf868000
[  137.355834] 9ea0: 00000089 c0374130 00000002 00000000 c02a1d50 cf0ac000 0000000c cf0fc540
[  137.364410] 9ec0: 00000018 c02a1d50 cf0fc540 00000000 cf0fc540 c0282238 c028220c cf178d80
[  137.372985] 9ee0: 127525d8 c02821cc 9a1fa451 c032727c 9a1fa451 127525d8 cf0fc540 cf0ac4ec
[  137.381561] 9f00: cf0ac000 cf0fc540 cf0ac584 c03285f4 c0328580 cf0ac4ec cf85c740 c05510cc
[  137.390136] 9f20: ce825400 c004c914 00000002 00000000 c004c884 ce8254f5 cf869f48 00000000
[  137.398712] 9f40: c0328580 ce825415 c0a7f914 c061af64 00000000 c048cf3c cf8673c0 cf85c740
[  137.407287] 9f60: c05510cc c051a66c c05510ec c05510c4 cf85c750 cf868000 00000089 c004d6ac
[  137.415863] 9f80: 00000000 c0073d14 00000001 cf853ed8 cf85c740 c004d558 00000013 00000000
[  137.424438] 9fa0: 00000000 00000000 00000000 c00516b0 00000000 00000000 cf85c740 00000000
[  137.433013] 9fc0: 00000001 dead4ead ffffffff ffffffff c0551674 00000000 00000000 c0450aa4
[  137.441589] 9fe0: cf869fe0 cf869fe0 cf853ed8 c005162c c0013b30 c0013b30 00ffff00 00ffff00
[  137.450164] [<c0071490>] (__lock_acquire+0x5c/0x1ab0) from [<c00733f8>] (lock_acquire+0x9c/0x128)
[  137.459503] [<c00733f8>] (lock_acquire+0x9c/0x128) from [<c0374130>] (_raw_spin_lock_irqsave+0x44/0x58)
[  137.469360] [<c0374130>] (_raw_spin_lock_irqsave+0x44/0x58) from [<c02a1d50>] (skb_queue_tail+0x18/0x48)
[  137.479339] [<c02a1d50>] (skb_queue_tail+0x18/0x48) from [<c0282238>] (h4_enqueue+0x2c/0x34)
[  137.488189] [<c0282238>] (h4_enqueue+0x2c/0x34) from [<c02821cc>] (hci_uart_send_frame+0x34/0x68)
[  137.497497] [<c02821cc>] (hci_uart_send_frame+0x34/0x68) from [<c032727c>] (hci_send_frame+0x50/0x88)
[  137.507171] [<c032727c>] (hci_send_frame+0x50/0x88) from [<c03285f4>] (hci_cmd_work+0x74/0xd4)
[  137.516204] [<c03285f4>] (hci_cmd_work+0x74/0xd4) from [<c004c914>] (process_one_work+0x1a0/0x4ec)
[  137.525604] [<c004c914>] (process_one_work+0x1a0/0x4ec) from [<c004d6ac>] (worker_thread+0x154/0x344)
[  137.535278] [<c004d6ac>] (worker_thread+0x154/0x344) from [<c00516b0>] (kthread+0x84/0x90)
[  137.543975] [<c00516b0>] (kthread+0x84/0x90) from [<c0013b30>] (kernel_thread_exit+0x0/0x8)
[  137.552734] Code: e59f4e5c e5941000 e3510000 0a000031 (e5971000)
[  137.559234] ---[ end trace 1b75b31a2719ed1e ]---

Signed-off-by: Johan Hovold <jhovold@gmail.com>
Acked-by: Marcel Holtmann <marcel@holtmann.org>
Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>

Bluetooth: add NULL pointer check in HCI

commit d9319560b86839506c2011346b1f2e61438a3c73 upstream.

If we fail to find a hci device pointer in hci_uart, don't try
to deref the NULL one we do have.

Signed-off-by: Jun Nie <njun@marvell.com>
Signed-off-by: Gustavo F. Padovan <padovan@profusion.mobi>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>

Bluetooth: fix possible info leak in bt_sock_recvmsg()

commit 4683f42fde3977bdb4e8a09622788cc8b5313778 upstream.

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>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>

Bluetooth: L2CAP - Fix info leak via getsockname()

commit 792039c73cf176c8e39a6e8beef2c94ff46522ed upstream.

The L2CAP code fails to initialize the l2_bdaddr_type member of struct
sockaddr_l2 and the padding byte added for alignment. It that for leaks
two bytes kernel stack via the getsockname() syscall. Add an explicit
memset(0) before filling the structure to avoid the info leak.

Signed-off-by: Mathias Krause <minipli@googlemail.com>
Cc: Marcel Holtmann <marcel@holtmann.org>
Cc: Gustavo Padovan <gustavo@padovan.org>
Cc: Johan Hedberg <johan.hedberg@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
[PG: net/bluetooth/l2cap_sock.c --> net/bluetooth/l2cap.c in .34]
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>

Bluetooth: RFCOMM - Fix missing msg_namelen update in rfcomm_sock_recvmsg()

commit e11e0455c0d7d3d62276a0c55d9dfbc16779d691 upstream.

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>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>

Bluetooth: RFCOMM - Fix info leak via getsockname()

commit 9344a972961d1a6d2c04d9008b13617bcb6ec2ef upstream.

The RFCOMM code fails to initialize the trailing padding byte of struct
sockaddr_rc added for alignment. It that for leaks one byte kernel stack
via the getsockname() syscall. Add an explicit memset(0) before filling
the structure to avoid the info leak.

Signed-off-by: Mathias Krause <minipli@googlemail.com>
Cc: Marcel Holtmann <marcel@holtmann.org>
Cc: Gustavo Padovan <gustavo@padovan.org>
Cc: Johan Hedberg <johan.hedberg@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>

Bluetooth: HCI - Fix info leak in getsockopt(HCI_FILTER)

commit e15ca9a0ef9a86f0477530b0f44a725d67f889ee upstream.

The HCI code fails to initialize the two padding bytes of struct
hci_ufilter before copying it to userland -- that for leaking two
bytes kernel stack. Add an explicit memset(0) before filling the
structure to avoid the info leak.

Signed-off-by: Mathias Krause <minipli@googlemail.com>
Cc: Marcel Holtmann <marcel@holtmann.org>
Cc: Gustavo Padovan <gustavo@padovan.org>
Cc: Johan Hedberg <johan.hedberg@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>

Bluetooth: Fix incorrect strncpy() in hidp_setup_hid()

commit 0a9ab9bdb3e891762553f667066190c1d22ad62b upstream.

The length parameter should be sizeof(req->name) - 1 because there is no
guarantee that string provided by userspace will contain the trailing
'\0'.

Can be easily reproduced by manually setting req->name to 128 non-zero
bytes prior to ioctl(HIDPCONNADD) and checking the device name setup on
input subsystem:

$ cat /sys/devices/pnp0/00\:04/tty/ttyS0/hci0/hci0\:1/input8/name
AAAAAA[...]AAAAAAAAf0:af:f0:af:f0:af

("f0:af:f0:af:f0:af" is the device bluetooth address, taken from "phys"
field in struct hid_device due to overflow.)

Signed-off-by: Anderson Lizardo <anderson.lizardo@openbossa.org>
Acked-by: Marcel Holtmann <marcel@holtmann.org>
Signed-off-by: Gustavo Padovan <gustavo.padovan@collabora.co.uk>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>

IPoIB: Fix use-after-free of multicast object

commit bea1e22df494a729978e7f2c54f7bda328f74bc3 upstream.

Fix a crash in ipoib_mcast_join_task().  (with help from Or Gerlitz)

Commit c8c2afe360b7 ("IPoIB: Use rtnl lock/unlock when changing device
flags") added a call to rtnl_lock() in ipoib_mcast_join_task(), which
is run from the ipoib_workqueue, and hence the workqueue can't be
flushed from the context of ipoib_stop().

In the current code, ipoib_stop() (which doesn't flush the workqueue)
calls ipoib_mcast_dev_flush(), which goes and deletes all the
multicast entries.  This takes place without any synchronization with
a possible running instance of ipoib_mcast_join_task() for the same
ipoib device, leading to a crash due to NULL pointer dereference.

Fix this by making sure that the workqueue is flushed before
ipoib_mcast_dev_flush() is called.  To make that possible, we move the
RTNL-lock wrapped code to ipoib_mcast_join_finish().

Signed-off-by: Patrick McHardy <kaber@trash.net>
Signed-off-by: Roland Dreier <roland@purestorage.com>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>

crypto: cryptd - disable softirqs in cryptd_queue_worker to prevent data corruption

commit 9efade1b3e981f5064f9db9ca971b4dc7557ae42 upstream.

cryptd_queue_worker attempts to prevent simultaneous accesses to crypto
workqueue by cryptd_enqueue_request using preempt_disable/preempt_enable.
However cryptd_enqueue_request might be called from softirq context,
so add local_bh_disable/local_bh_enable to prevent data corruption and
panics.

Bug report at http://marc.info/?l=linux-crypto-vger&m=134858649616319&w=2

v2:
 - Disable software interrupts instead of hardware interrupts

Reported-by: Gurucharan Shetty <gurucharan.shetty@gmail.com>
Signed-off-by: Jussi Kivilinna <jussi.kivilinna@mbnet.fi>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>

keys: fix race with concurrent install_user_keyrings()

commit 0da9dfdd2cd9889201bc6f6f43580c99165cd087 upstream.

This fixes CVE-2013-1792.

There is a race in install_user_keyrings() that can cause a NULL pointer
dereference when called concurrently for the same user if the uid and
uid-session keyrings are not yet created.  It might be possible for an
unprivileged user to trigger this by calling keyctl() from userspace in
parallel immediately after logging in.

Assume that we have two threads both executing lookup_user_key(), both
looking for KEY_SPEC_USER_SESSION_KEYRING.

THREAD A THREAD B
=============================== ===============================
==>call install_user_keyrings();
if (!cred->user->session_keyring)
==>call install_user_keyrings()
...
user->uid_keyring = uid_keyring;
if (user->uid_keyring)
return 0;
<==
key = cred->user->session_keyring [== NULL]
user->session_keyring = session_keyring;
atomic_inc(&key->usage); [oops]

At the point thread A dereferences cred->user->session_keyring, thread B
hasn't updated user->session_keyring yet, but thread A assumes it is
populated because install_user_keyrings() returned ok.

The race window is really small but can be exploited if, for example,
thread B is interrupted or preempted after initializing uid_keyring, but
before doing setting session_keyring.

This couldn't be reproduced on a stock kernel.  However, after placing
systemtap probe on 'user->session_keyring = session_keyring;' that
introduced some delay, the kernel could be crashed reliably.

Fix this by checking both pointers before deciding whether to return.
Alternatively, the test could be done away with entirely as it is checked
inside the mutex - but since the mutex is global, that may not be the best
way.

Signed-off-by: David Howells <dhowells@redhat.com>
Reported-by: Mateusz Guzik <mguzik@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: James Morris <james.l.morris@oracle.com>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>

bnx2i: Fixed NULL ptr deference for 1G bnx2 Linux iSCSI offload

commit d6532207116307eb7ecbfa7b9e02c53230096a50 upstream.

This patch fixes the following kernel panic invoked by uninitialized fields
in the chip initialization for the 1G bnx2 iSCSI offload.

One of the bits in the chip initialization is being used by the latest
firmware to control overflow packets.  When this control bit gets enabled
erroneously, it would ultimately result in a bad packet placement which would
cause the bnx2 driver to dereference a NULL ptr in the placement handler.

This can happen under certain stress I/O environment under the Linux
iSCSI offload operation.

This change only affects Broadcom's 5709 chipset.

Unable to handle kernel NULL pointer dereference at 0000000000000008 RIP:
 [<ffffffff881f0e7d>] :bnx2:bnx2_poll_work+0xd0d/0x13c5
Pid: 0, comm: swapper Tainted: G     ---- 2.6.18-333.el5debug #2
RIP: 0010:[<ffffffff881f0e7d>]  [<ffffffff881f0e7d>] :bnx2:bnx2_poll_work+0xd0d/0x13c5
RSP: 0018:ffff8101b575bd50  EFLAGS: 00010216
RAX: 0000000000000005 RBX: ffff81007c5fb180 RCX: 0000000000000000
RDX: 0000000000000ffc RSI: 00000000817e8000 RDI: 0000000000000220
RBP: ffff81015bbd7ec0 R08: ffff8100817e9000 R09: 0000000000000000
R10: ffff81007c5fb180 R11: 00000000000000c8 R12: 000000007a25a010
R13: 0000000000000000 R14: 0000000000000005 R15: ffff810159f80558
FS:  0000000000000000(0000) GS:ffff8101afebc240(0000) knlGS:0000000000000000
CS:  0010 DS: 0018 ES: 0018 CR0: 000000008005003b
CR2: 0000000000000008 CR3: 0000000000201000 CR4: 00000000000006a0
Process swapper (pid: 0, threadinfo ffff8101b5754000, task ffff8101afebd820)
Stack:  000000000000000b ffff810159f80000 0000000000000040 ffff810159f80520
 ffff810159f80500 00cf00cf8008e84b ffffc200100939e0 ffff810009035b20
 0000502900000000 000000be00000001 ffff8100817e7810 00d08101b575bea8
Call Trace:
 <IRQ>  [<ffffffff8008e0d0>] show_schedstat+0x1c2/0x25b
 [<ffffffff881f1886>] :bnx2:bnx2_poll+0xf6/0x231
 [<ffffffff8000c9b9>] net_rx_action+0xac/0x1b1
 [<ffffffff800125a0>] __do_softirq+0x89/0x133
 [<ffffffff8005e30c>] call_softirq+0x1c/0x28
 [<ffffffff8006d5de>] do_softirq+0x2c/0x7d
 [<ffffffff8006d46e>] do_IRQ+0xee/0xf7
 [<ffffffff8005d625>] ret_from_intr+0x0/0xa
 <EOI>  [<ffffffff801a5780>] acpi_processor_idle_simple+0x1c5/0x341
 [<ffffffff801a573d>] acpi_processor_idle_simple+0x182/0x341
 [<ffffffff801a55bb>] acpi_processor_idle_simple+0x0/0x341
 [<ffffffff80049560>] cpu_idle+0x95/0xb8
 [<ffffffff80078b1c>] start_secondary+0x479/0x488

Signed-off-by: Eddie Wai <eddie.wai@broadcom.com>
Reviewed-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>

fix crash in scsi_dispatch_cmd()

commit bfe159a51203c15d23cb3158fffdc25ec4b4dda1 upstream.

USB surprise removal of sr is triggering an oops in
scsi_dispatch_command().  What seems to be happening is that USB is
hanging on to a queue reference until the last close of the upper
device, so the crash is caused by surprise remove of a mounted CD
followed by attempted unmount.

The problem is that USB doesn't issue its final commands as part of
the SCSI teardown path, but on last close when the block queue is long
gone.  The long term fix is probably to make sr do the teardown in the
same way as sd (so remove all the lower bits on ejection, but keep the
upper disk alive until last close of user space).  However, the
current oops can be simply fixed by not allowing any commands to be
sent to a dead queue.

Signed-off-by: James Bottomley <JBottomley@Parallels.com>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>

fix Null pointer dereference on disk error

commit 26cd4d65deba587f3cf2329b6869ce02bcbe68ec upstream.

Following oops were observed when disk error happened:

[ 4272.896937] sd 0:0:0:0: [sda] Unhandled error code
[ 4272.896939] sd 0:0:0:0: [sda] Result: hostbyte=DID_BAD_TARGET driverbyte=DRIVER_OK
[ 4272.896942] sd 0:0:0:0: [sda] CDB: Read(10): 28 00 00 5a de a7 00 00 08 00
[ 4272.896951] end_request: I/O error, dev sda, sector 5955239
[ 4291.574947] BUG: unable to handle kernel NULL pointer dereference at (null)
[ 4291.658305] IP: [] ahci_activity_show+0x1/0x40
[ 4291.730090] PGD 76dbbc067 PUD 6c4fba067 PMD 0
[ 4291.783408] Oops: 0000 [#1] SMP
[ 4291.822100] last sysfs file: /sys/devices/pci0000:00/0000:00:1f.2/host0/target0:0:0/0:0:0:0/sw_activity
[ 4291.934235] CPU 9
[ 4291.958301] Pid: 27942, comm: hwinfo ......

ata_scsi_find_dev could return NULL, so ata_scsi_activity_{show,store} should check if atadev is NULL.

Signed-off-by: Xiaotian Feng <dannyfeng@tencent.com>
Cc: James Bottomley <JBottomley@Parallels.com>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>

w1: fix oops when w1_search is called from netlink connector

commit 9d1817cab2f030f6af360e961cc69bb1da8ad765 upstream.

On Sat, Mar 02, 2013 at 10:45:10AM +0100, Sven Geggus wrote:
> This is the bad commit I found doing git bisect:
> 04f482faf50535229a5a5c8d629cf963899f857c is the first bad commit
> commit 04f482faf50535229a5a5c8d629cf963899f857c
> Author: Patrick McHardy <kaber@trash.net>
> Date:   Mon Mar 28 08:39:36 2011 +0000

Good job. I was too lazy to bisect for bad commit;)

Reading the code I found problematic kthread_should_stop call from netlink
connector which causes the oops. After applying a patch, I've been testing
owfs+w1 setup for nearly two days and it seems to work very reliable (no
hangs, no memleaks etc).
More detailed description and possible fix is given below:

Function w1_search can be called from either kthread or netlink callback.
While the former works fine, the latter causes oops due to kthread_should_stop
invocation.

This patch adds a check if w1_search is serving netlink command, skipping
kthread_should_stop invocation if so.

Signed-off-by: Marcin Jurkowski <marcin1j@gmail.com>
Acked-by: Evgeniy Polyakov <zbr@ioremap.net>
Cc: Josh Boyer <jwboyer@gmail.com>
Tested-by: Sven Geggus <lists@fuchsschwanzdomain.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>

drivers/char/ipmi: memcpy, need additional 2 bytes to avoid memory overflow

commit a5f2b3d6a738e7d4180012fe7b541172f8c8dcea upstream.

When calling memcpy, read_data and write_data need additional 2 bytes.

  write_data:
    for checking:  "if (size > IPMI_MAX_MSG_LENGTH)"
    for operating: "memcpy(bt->write_data + 3, data + 1, size - 1)"

  read_data:
    for checking:  "if (msg_len < 3 || msg_len > IPMI_MAX_MSG_LENGTH)"
    for operating: "memcpy(data + 2, bt->read_data + 4, msg_len - 2)"

Signed-off-by: Chen Gang <gang.chen@asianux.com>
Signed-off-by: Corey Minyard <cminyard@mvista.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>

serial: 8250, increase PASS_LIMIT

commit e7328ae1848966181a7ac47e8ae6cddbd2cf55f3 upstream.

With virtual machines like qemu, it's pretty common to see "too much
work for irq4" messages nowadays. This happens when a bunch of output
is printed on the emulated serial console. This is caused by too low
PASS_LIMIT. When ISR loops more than the limit, it spits the message.

I've been using a kernel with doubled the limit and I couldn't see no
problems. Maybe it's time to get rid of the message now?

Signed-off-by: Jiri Slaby <jirislaby@gmail.com>
Cc: Alan Cox <alan@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
[PG: drivers/tty/serial/8250.c ---> drivers/serial/8250.c in 2.6.34]
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>

ACPI / cpuidle: Fix NULL pointer issues when cpuidle is disabled

commit b88a634a903d9670aa5f2f785aa890628ce0dece upstream.

If cpuidle is disabled, that means that:

per_cpu(acpi_cpuidle_device, pr->id)

is set to NULL as the acpi_processor_power_init ends up failing at

 retval = cpuidle_register_driver(&acpi_idle_driver)

(in acpi_processor_power_init) and never sets the per_cpu idle
device.  So when acpi_processor_hotplug on CPU online notification
tries to reference said device it crashes:

cpu 3 spinlock event irq 62
BUG: unable to handle kernel NULL pointer dereference at 0000000000000004
IP: [<ffffffff81381013>] acpi_processor_setup_cpuidle_cx+0x3f/0x105
PGD a259b067 PUD ab38b067 PMD 0
Oops: 0002 [#1] SMP
odules linked in: dm_multipath dm_mod xen_evtchn iscsi_boot_sysfs iscsi_tcp libiscsi_tcp libiscsi scsi_transport_iscsi libcrc32c crc32c nouveau mxm_wmi wmi radeon ttm sg sr_mod sd_mod cdrom ata_generic ata_piix libata crc32c_intel scsi_mod atl1c i915 fbcon tileblit font bitblit softcursor drm_kms_helper video xen_blkfront xen_netfront fb_sys_fops sysimgblt sysfillrect syscopyarea xenfs xen_privcmd mperf
CPU 1
Pid: 3047, comm: bash Not tainted 3.8.0-rc3upstream-00250-g165c029 #1 MSI MS-7680/H61M-P23 (MS-7680)
RIP: e030:[<ffffffff81381013>]  [<ffffffff81381013>] acpi_processor_setup_cpuidle_cx+0x3f/0x105
RSP: e02b:ffff88001742dca8  EFLAGS: 00010202
RAX: 0000000000010be9 RBX: ffff8800a0a61800 RCX: ffff880105380000
RDX: 0000000000000003 RSI: 0000000000000200 RDI: ffff8800a0a61800
RBP: ffff88001742dce8 R08: ffffffff81812360 R09: 0000000000000200
R10: aaaaaaaaaaaaaaaa R11: 0000000000000001 R12: ffff8800a0a61800
R13: 00000000ffffff01 R14: 0000000000000000 R15: ffffffff81a907a0
FS:  00007fd6942f7700(0000) GS:ffff880105280000(0000) knlGS:0000000000000000
CS:  e033 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: 0000000000000004 CR3: 00000000a6773000 CR4: 0000000000042660
DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400
Process bash (pid: 3047, threadinfo ffff88001742c000, task ffff880017944000)
Stack:
 0000000000000150 ffff880100f59e00 ffff88001742dcd8 ffff8800a0a61800
 0000000000000000 00000000ffffff01 0000000000000000 ffffffff81a907a0
 ffff88001742dd18 ffffffff813815b1 ffff88001742dd08 ffffffff810ae336
Call Trace:
 [<ffffffff813815b1>] acpi_processor_hotplug+0x7c/0x9f
 [<ffffffff810ae336>] ? schedule_delayed_work_on+0x16/0x20
 [<ffffffff8137ee8f>] acpi_cpu_soft_notify+0x90/0xca
 [<ffffffff8166023d>] notifier_call_chain+0x4d/0x70
 [<ffffffff810bc369>] __raw_notifier_call_chain+0x9/0x10
 [<ffffffff81094a4b>] __cpu_notify+0x1b/0x30
 [<ffffffff81652cf7>] _cpu_up+0x103/0x14b
 [<ffffffff81652e18>] cpu_up+0xd9/0xec
 [<ffffffff8164a254>] store_online+0x94/0xd0
 [<ffffffff814122fb>] dev_attr_store+0x1b/0x20
 [<ffffffff81216404>] sysfs_write_file+0xf4/0x170

This patch fixes it.

Signed-off-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>

MCE: Fix vm86 handling for 32bit mce handler

commit a129a7c84582629741e5fa6f40026efcd7a65bd4 upstream.

When running on 32bit the mce handler could misinterpret
vm86 mode as ring 0. This can affect whether it does recovery
or not; it was possible to panic when recovery was actually
possible.

Fix this by always forcing vm86 to look like ring 3.

[ Backport to 3.0 notes:
Things changed there slightly:
   - move mce_get_rip() up. It fills up m->cs and m->ip values which
     are evaluated in mce_severity(). Therefore move it up right before
     the mce_severity call. This seem to be another bug in 3.0?
   - Place the backport (fix m->cs in V86 case) to where m->cs gets
     filled which is mce_get_rip() in 3.0
]

Signed-off-by: Andi Kleen <ak@linux.intel.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
Signed-off-by: Thomas Renninger <trenn@suse.de>
Reviewed-by: Tony Luck <tony.luck@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
[PG: commit 8ef8fa7479fff9313387b873413f5ae233a2bd04 in v3.0.44]
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>

KVM: Fix bounds checking in ioapic indirect register reads (CVE-2013-1798)

commit a2c118bfab8bc6b8bb213abfc35201e441693d55 upstream.

If the guest specifies a IOAPIC_REG_SELECT with an invalid value and follows
that with a read of the IOAPIC_REG_WINDOW KVM does not properly validate
that request.  ioapic_read_indirect contains an
ASSERT(redir_index < IOAPIC_NUM_PINS), but the ASSERT has no effect in
non-debug builds.  In recent kernels this allows a guest to cause a kernel
oops by reading invalid memory.  In older kernels (pre-3.3) this allows a
guest to read from large ranges of host memory.

Tested: tested against apic unit tests.

Signed-off-by: Andrew Honig <ahonig@google.com>
Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>

KVM: x86: fix for buffer overflow in handling of MSR_KVM_SYSTEM_TIME (CVE-2013-1796)

commit c300aa64ddf57d9c5d9c898a64b36877345dd4a9 upstream.

If the guest sets the GPA of the time_page so that the request to update the
time straddles a page then KVM will write onto an incorrect page.  The
write is done byusing kmap atomic to get a pointer to the page for the time
structure and then performing a memcpy to that page starting at an offset
that the guest controls.  Well behaved guests always provide a 32-byte aligned
address, however a malicious guest could use this to corrupt host kernel
memory.

Tested: Tested against kvmclock unit test.

Signed-off-by: Andrew Honig <ahonig@google.com>
Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>

xen/bootup: allow {read|write}_cr8 pvops call.

commit 1a7bbda5b1ab0e02622761305a32dc38735b90b2 upstream.

We actually do not do anything about it. Just return a default
value of zero and if the kernel tries to write anything but 0
we BUG_ON.

This fixes the case when an user tries to suspend the machine
and it blows up in save_processor_state b/c 'read_cr8' is set
to NULL and we get:

kernel BUG at /home/konrad/ssd/linux/arch/x86/include/asm/paravirt.h:100!
invalid opcode: 0000 [#1] SMP
Pid: 2687, comm: init.late Tainted: G           O 3.6.0upstream-00002-gac264ac-dirty #4 Bochs Bochs
RIP: e030:[<ffffffff814d5f42>]  [<ffffffff814d5f42>] save_processor_state+0x212/0x270

.. snip..
Call Trace:
 [<ffffffff810733bf>] do_suspend_lowlevel+0xf/0xac
 [<ffffffff8107330c>] ? x86_acpi_suspend_lowlevel+0x10c/0x150
 [<ffffffff81342ee2>] acpi_suspend_enter+0x57/0xd5

Signed-off-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>

xen/bootup: allow read_tscp call for Xen PV guests.

commit cd0608e71e9757f4dae35bcfb4e88f4d1a03a8ab upstream.

The hypervisor will trap it. However without this patch,
we would crash as the .read_tscp is set to NULL. This patch
fixes it and sets it to the native_read_tscp call.

Signed-off-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>

x86, mm, paravirt: Fix vmalloc_fault oops during lazy MMU updates

commit 1160c2779b826c6f5c08e5cc542de58fd1f667d5 upstream.

In paravirtualized x86_64 kernels, vmalloc_fault may cause an oops
when lazy MMU updates are enabled, because set_pgd effects are being
deferred.

One instance of this problem is during process mm cleanup with memory
cgroups enabled. The chain of events is as follows:

- zap_pte_range enables lazy MMU updates
- zap_pte_range eventually calls mem_cgroup_charge_statistics,
  which accesses the vmalloc'd mem_cgroup per-cpu stat area
- vmalloc_fault is triggered which tries to sync the corresponding
  PGD entry with set_pgd, but the update is deferred
- vmalloc_fault oopses due to a mismatch in the PUD entries

The OOPs usually looks as so:

------------[ cut here ]------------
kernel BUG at arch/x86/mm/fault.c:396!
invalid opcode: 0000 [#1] SMP
.. snip ..
CPU 1
Pid: 10866, comm: httpd Not tainted 3.6.10-4.fc18.x86_64 #1
RIP: e030:[<ffffffff816271bf>]  [<ffffffff816271bf>] vmalloc_fault+0x11f/0x208
.. snip ..
Call Trace:
 [<ffffffff81627759>] do_page_fault+0x399/0x4b0
 [<ffffffff81004f4c>] ? xen_mc_extend_args+0xec/0x110
 [<ffffffff81624065>] page_fault+0x25/0x30
 [<ffffffff81184d03>] ? mem_cgroup_charge_statistics.isra.13+0x13/0x50
 [<ffffffff81186f78>] __mem_cgroup_uncharge_common+0xd8/0x350
 [<ffffffff8118aac7>] mem_cgroup_uncharge_page+0x57/0x60
 [<ffffffff8115fbc0>] page_remove_rmap+0xe0/0x150
 [<ffffffff8115311a>] ? vm_normal_page+0x1a/0x80
 [<ffffffff81153e61>] unmap_single_vma+0x531/0x870
 [<ffffffff81154962>] unmap_vmas+0x52/0xa0
 [<ffffffff81007442>] ? pte_mfn_to_pfn+0x72/0x100
 [<ffffffff8115c8f8>] exit_mmap+0x98/0x170
 [<ffffffff810050d9>] ? __raw_callee_save_xen_pmd_val+0x11/0x1e
 [<ffffffff81059ce3>] mmput+0x83/0xf0
 [<ffffffff810624c4>] exit_mm+0x104/0x130
 [<ffffffff8106264a>] do_exit+0x15a/0x8c0
 [<ffffffff810630ff>] do_group_exit+0x3f/0xa0
 [<ffffffff81063177>] sys_exit_group+0x17/0x20
 [<ffffffff8162bae9>] system_call_fastpath+0x16/0x1b

Calling arch_flush_lazy_mmu_mode immediately after set_pgd makes the
changes visible to the consistency checks.

RedHat-Bugzilla: https://bugzilla.redhat.com/show_bug.cgi?id=914737
Tested-by: Josh Boyer <jwboyer@redhat.com>
Reported-and-Tested-by: Krishna Raman <kraman@redhat.com>
Signed-off-by: Samu Kallio <samu.kallio@aberdeencloud.com>
Link: http://lkml.kernel.org/r/1364045796-10720-1-git-send-email-konrad.wilk@oracle.com
Tested-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Signed-off-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Signed-off-by: H. Peter Anvin <hpa@linux.intel.com>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>

x86/mm: Check if PUD is large when validating a kernel address

commit 0ee364eb316348ddf3e0dfcd986f5f13f528f821 upstream.

A user reported the following oops when a backup process reads
/proc/kcore:

 BUG: unable to handle kernel paging request at ffffbb00ff33b000
 IP: [<ffffffff8103157e>] kern_addr_valid+0xbe/0x110
 [...]

 Call Trace:
  [<ffffffff811b8aaa>] read_kcore+0x17a/0x370
  [<ffffffff811ad847>] proc_reg_read+0x77/0xc0
  [<ffffffff81151687>] vfs_read+0xc7/0x130
  [<ffffffff811517f3>] sys_read+0x53/0xa0
  [<ffffffff81449692>] system_call_fastpath+0x16/0x1b

Investigation determined that the bug triggered when reading
system RAM at the 4G mark. On this system, that was the first
address using 1G pages for the virt->phys direct mapping so the
PUD is pointing to a physical address, not a PMD page.

The problem is that the page table walker in kern_addr_valid() is
not checking pud_large() and treats the physical address as if
it was a PMD.  If it happens to look like pmd_none then it'll
silently fail, probably returning zeros instead of real data. If
the data happens to look like a present PMD though, it will be
walked resulting in the oops above.

This patch adds the necessary pud_large() check.

Unfortunately the problem was not readily reproducible and now
they are running the backup program without accessing
/proc/kcore so the patch has not been validated but I think it
makes sense.

Signed-off-by: Mel Gorman <mgorman@suse.de>
Reviewed-by: Rik van Riel <riel@redhat.coM>
Reviewed-by: Michal Hocko <mhocko@suse.cz>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Cc: linux-mm@kvack.org
Link: http://lkml.kernel.org/r/20130211145236.GX21389@suse.de
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>

x86, tls: Off by one limit check

commit 8f0750f19789cf352d7e24a6cc50f2ab1b4f1372 upstream.

These are used as offsets into an array of GDT_ENTRY_TLS_ENTRIES members
so GDT_ENTRY_TLS_ENTRIES is one past the end of the array.

Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Link: http://lkml.kernel.org/r/20120324075250.GA28258@elgon.mountain
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>

x86/msr: Add capabilities check

commit c903f0456bc69176912dee6dd25c6a66ee1aed00 upstream.

At the moment the MSR driver only relies upon file system
checks. This means that anything as root with any capability set
can write to MSRs. Historically that wasn't very interesting but
on modern processors the MSRs are such that writing to them
provides several ways to execute arbitary code in kernel space.
Sample code and documentation on doing this is circulating and
MSR attacks are used on Windows 64bit rootkits already.

In the Linux case you still need to be able to open the device
file so the impact is fairly limited and reduces the security of
some capability and security model based systems down towards
that of a generic "root owns the box" setup.

Therefore they should require CAP_SYS_RAWIO to prevent an
elevation of capabilities. The impact of this is fairly minimal
on most setups because they don't have heavy use of
capabilities. Those using SELinux, SMACK or AppArmor rules might
want to consider if their rulesets on the MSR driver could be
tighter.

Signed-off-by: Alan Cox <alan@linux.intel.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>

x86/xen: don't assume %ds is usable in xen_iret for 32-bit PVOPS.

commit 13d2b4d11d69a92574a55bfd985cfb0ca77aebdc upstream.

This fixes CVE-2013-0228 / XSA-42

Drew Jones while working on CVE-2013-0190 found that that unprivileged guest user
in 32bit PV guest can use to crash the > guest with the panic like this:

-------------
general protection fault: 0000 [#1] SMP
last sysfs file: /sys/devices/vbd-51712/block/xvda/dev
Modules linked in: sunrpc ipt_REJECT nf_conntrack_ipv4 nf_defrag_ipv4
iptable_filter ip_tables ip6t_REJECT nf_conntrack_ipv6 nf_defrag_ipv6
xt_state nf_conntrack ip6table_filter ip6_tables ipv6 xen_netfront ext4
mbcache jbd2 xen_blkfront dm_mirror dm_region_hash dm_log dm_mod [last
unloaded: scsi_wait_scan]

Pid: 1250, comm: r Not tainted 2.6.32-356.el6.i686 #1
EIP: 0061:[<c0407462>] EFLAGS: 00010086 CPU: 0
EIP is at xen_iret+0x12/0x2b
EAX: eb8d0000 EBX: 00000001 ECX: 08049860 EDX: 00000010
ESI: 00000000 EDI: 003d0f00 EBP: b77f8388 ESP: eb8d1fe0
 DS: 0000 ES: 007b FS: 0000 GS: 00e0 SS: 0069
Process r (pid: 1250, ti=eb8d0000 task=c2953550 task.ti=eb8d0000)
Stack:
 00000000 0027f416 00000073 00000206 b77f8364 0000007b 00000000 00000000
Call Trace:
Code: c3 8b 44 24 18 81 4c 24 38 00 02 00 00 8d 64 24 30 e9 03 00 00 00
8d 76 00 f7 44 24 08 00 00 02 80 75 33 50 b8 00 e0 ff ff 21 e0 <8b> 40
10 8b 04 85 a0 f6 ab c0 8b 80 0c b0 b3 c0 f6 44 24 0d 02
EIP: [<c0407462>] xen_iret+0x12/0x2b SS:ESP 0069:eb8d1fe0
general protection fault: 0000 [#2]
---[ end trace ab0d29a492dcd330 ]---
Kernel panic - not syncing: Fatal exception
Pid: 1250, comm: r Tainted: G      D    ---------------
2.6.32-356.el6.i686 #1
Call Trace:
 [<c08476df>] ? panic+0x6e/0x122
 [<c084b63c>] ? oops_end+0xbc/0xd0
 [<c084b260>] ? do_general_protection+0x0/0x210
 [<c084a9b7>] ? error_code+0x73/
-------------

Petr says: "
 I've analysed the bug and I think that xen_iret() cannot cope with
 mangled DS, in this case zeroed out (null selector/descriptor) by either
 xen_failsafe_callback() or RESTORE_REGS because the corresponding LDT
 entry was invalidated by the reproducer. "

Jan took a look at the preliminary patch and came up a fix that solves
this problem:

"This code gets called after all registers other than those handled by
IRET got already restored, hence a null selector in %ds or a non-null
one that got loaded from a code or read-only data descriptor would
cause a kernel mode fault (with the potential of crashing the kernel
as a whole, if panic_on_oops is set)."

The way to fix this is to realize that the we can only relay on the
registers that IRET restores. The two that are guaranteed are the
%cs and %ss as they are always fixed GDT selectors. Also they are
inaccessible from user mode - so they cannot be altered. This is
the approach taken in this patch.

Another alternative option suggested by Jan would be to relay on
the subtle realization that using the %ebp or %esp relative references uses
the %ss segment.  In which case we could switch from using %eax to %ebp and
would not need the %ss over-rides. That would also require one extra
instruction to compensate for the one place where the register is used
as scaled index. However Andrew pointed out that is too subtle and if
further work was to be done in this code-path it could escape folks attention
and lead to accidents.

Reviewed-by: Petr Matousek <pmatouse@redhat.com>
Reported-by: Petr Matousek <pmatouse@redhat.com>
Reviewed-by: Andrew Cooper <andrew.cooper3@citrix.com>
Signed-off-by: Jan Beulich <jbeulich@suse.com>
Signed-off-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>

ALSA: seq: Fix missing error handling in snd_seq_timer_open()

commit 66efdc71d95887b652a742a5dae51fa834d71465 upstream.

snd_seq_timer_open() didn't catch the whole error path but let through
if the timer id is a slave.  This may lead to Oops by accessing the
uninitialized pointer.

 BUG: unable to handle kernel NULL pointer dereference at 00000000000002ae
 IP: [<ffffffff819b3477>] snd_seq_timer_open+0xe7/0x130
 PGD 785cd067 PUD 76964067 PMD 0
 Oops: 0002 [#4] SMP
 CPU 0
 Pid: 4288, comm: trinity-child7 Tainted: G      D W 3.9.0-rc1+ #100 Bochs Bochs
 RIP: 0010:[<ffffffff819b3477>]  [<ffffffff819b3477>] snd_seq_timer_open+0xe7/0x130
 RSP: 0018:ffff88006ece7d38  EFLAGS: 00010246
 RAX: 0000000000000286 RBX: ffff88007851b400 RCX: 0000000000000000
 RDX: 000000000000ffff RSI: ffff88006ece7d58 RDI: ffff88006ece7d38
 RBP: ffff88006ece7d98 R08: 000000000000000a R09: 000000000000fffe
 R10: 0000000000000000 R11: 0000000000000000 R12: 0000000000000000
 R13: ffff8800792c5400 R14: 0000000000e8f000 R15: 0000000000000007
 FS:  00007f7aaa650700(0000) GS:ffff88007f800000(0000) GS:0000000000000000
 CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
 CR2: 00000000000002ae CR3: 000000006efec000 CR4: 00000000000006f0
 DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
 DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400
 Process trinity-child7 (pid: 4288, threadinfo ffff88006ece6000, task ffff880076a8a290)
 Stack:
  0000000000000286 ffffffff828f2be0 ffff88006ece7d58 ffffffff810f354d
  65636e6575716573 2065756575712072 ffff8800792c0030 0000000000000000
  ffff88006ece7d98 ffff8800792c5400 ffff88007851b400 ffff8800792c5520
 Call Trace:
  [<ffffffff810f354d>] ? trace_hardirqs_on+0xd/0x10
  [<ffffffff819b17e9>] snd_seq_queue_timer_open+0x29/0x70
  [<ffffffff819ae01a>] snd_seq_ioctl_set_queue_timer+0xda/0x120
  [<ffffffff819acb9b>] snd_seq_do_ioctl+0x9b/0xd0
  [<ffffffff819acbe0>] snd_seq_ioctl+0x10/0x20
  [<ffffffff811b9542>] do_vfs_ioctl+0x522/0x570
  [<ffffffff8130a4b3>] ? file_has_perm+0x83/0xa0
  [<ffffffff810f354d>] ? trace_hardirqs_on+0xd/0x10
  [<ffffffff811b95ed>] sys_ioctl+0x5d/0xa0
  [<ffffffff813663fe>] ? trace_hardirqs_on_thunk+0x3a/0x3f
  [<ffffffff81faed69>] system_call_fastpath+0x16/0x1b

Reported-and-tested-by: Tommi Rantala <tt.rantala@gmail.com>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>

mempolicy: fix a race in shared_policy_replace()

commit b22d127a39ddd10d93deee3d96e643657ad53a49 upstream.

shared_policy_replace() use of sp_alloc() is unsafe.  1) sp_node cannot
be dereferenced if sp->lock is not held and 2) another thread can modify
sp_node between spin_unlock for allocating a new sp node and next
spin_lock.  The bug was introduced before 2.6.12-rc2.

Kosaki's original patch for this problem was to allocate an sp node and
policy within shared_policy_replace and initialise it when the lock is
reacquired.  I was not keen on this approach because it partially
duplicates sp_alloc().  As the paths were sp->lock is taken are not that
performance critical this patch converts sp->lock to sp->mutex so it can
sleep when calling sp_alloc().

[kosaki.motohiro@jp.fujitsu.com: Original patch]
Signed-off-by: Mel Gorman <mgorman@suse.de>
Acked-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Reviewed-by: Christoph Lameter <cl@linux.com>
Cc: Josh Boyer <jwboyer@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>

mm: Hold a file reference in madvise_remove

commit 9ab4233dd08036fe34a89c7dc6f47a8bf2eb29eb upstream.

Otherwise the code races with munmap (causing a use-after-free
of the vma) or with close (causing a use-after-free of the struct
file).

The bug was introduced by commit 90ed52ebe481 ("[PATCH] holepunch: fix
mmap_sem i_mutex deadlock")

[bwh: Backported to 3.2:
 - Adjust context
 - madvise_remove() calls vmtruncate_range(), not do_fallocate()]
[luto: Backported to 3.0: Adjust context]

Cc: Hugh Dickins <hugh@veritas.com>
Cc: Miklos Szeredi <mszeredi@suse.cz>
Cc: Badari Pulavarty <pbadari@us.ibm.com>
Cc: Nick Piggin <npiggin@suse.de>
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
Signed-off-by: Andy Lutomirski <luto@amacapital.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
[PG: commit e12fcd38abe8a869cbabd77724008f1cf812a3e7 in v3.0.37]
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>

mm: mmu_notifier: fix freed page still mapped in secondary MMU

commit 3ad3d901bbcfb15a5e4690e55350db0899095a68 upstream.

mmu_notifier_release() is called when the process is exiting.  It will
delete all the mmu notifiers.  But at this time the page belonging to the
process is still present in page tables and is present on the LRU list, so
this race will happen:

      CPU 0                 CPU 1
mmu_notifier_release:    try_to_unmap:
   hlist_del_init_rcu(&mn->hlist);
                            ptep_clear_flush_notify:
                                  mmu nofifler not found
                            free page  !!!!!!
                            /*
                             * At the point, the page has been
                             * freed, but it is still mapped in
                             * the secondary MMU.
                             */

  mn->ops->release(mn, mm);

Then the box is not stable and sometimes we can get this bug:

[  738.075923] BUG: Bad page state in process migrate-perf  pfn:03bec
[  738.075931] page:ffffea00000efb00 count:0 mapcount:0 mapping:          (null) index:0x8076
[  738.075936] page flags: 0x20000000000014(referenced|dirty)

The same issue is present in mmu_notifier_unregister().

We can call ->release before deleting the notifier to ensure the page has
been unmapped from the secondary MMU before it is freed.

Signed-off-by: Xiao Guangrong <xiaoguangrong@linux.vnet.ibm.com>
Cc: Avi Kivity <avi@redhat.com>
Cc: Marcelo Tosatti <mtosatti@redhat.com>
Cc: Paul Gortmaker <paul.gortmaker@windriver.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>

mm: fix invalidate_complete_page2() lock ordering

commit ec4d9f626d5908b6052c2973f37992f1db52e967 upstream.

In fuzzing with trinity, lockdep protested "possible irq lock inversion
dependency detected" when isolate_lru_page() reenabled interrupts while
still holding the supposedly irq-safe tree_lock:

invalidate_inode_pages2
  invalidate_complete_page2
    spin_lock_irq(&mapping->tree_lock)
    clear_page_mlock
      isolate_lru_page
        spin_unlock_irq(&zone->lru_lock)

isolate_lru_page() is correct to enable interrupts unconditionally:
invalidate_complete_page2() is incorrect to call clear_page_mlock() while
holding tree_lock, which is supposed to nest inside lru_lock.

Both truncate_complete_page() and invalidate_complete_page() call
clear_page_mlock() before taking tree_lock to remove page from radix_tree.
 I guess invalidate_complete_page2() preferred to test PageDirty (again)
under tree_lock before committing to the munlock; but since the page has
already been unmapped, its state is already somewhat inconsistent, and no
worse if clear_page_mlock() moved up.

Reported-by: Sasha Levin <levinsasha928@gmail.com>
Deciphered-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Hugh Dickins <hughd@google.com>
Acked-by: Mel Gorman <mel@csn.ul.ie>
Cc: Rik van Riel <riel@redhat.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Michel Lespinasse <walken@google.com>
Cc: Ying Han <yinghan@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>

mm: bugfix: set current->reclaim_state to NULL while returning from kswapd()

commit b0a8cc58e6b9aaae3045752059e5e6260c0b94bc upstream.

In kswapd(), set current->reclaim_state to NULL before returning, as
current->reclaim_state holds reference to variable on kswapd()'s stack.

In rare cases, while returning from kswapd() during memory offlining,
__free_slab() and freepages() can access the dangling pointer of
current->reclaim_state.

Signed-off-by: Takamori Yamaguchi <takamori.yamaguchi@jp.sony.com>
Signed-off-by: Aaditya Kumar <aaditya.kumar@ap.sony.com>
Acked-by: David Rientjes <rientjes@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>

mm: Fix PageHead when !CONFIG_PAGEFLAGS_EXTENDED

commit ad4b3fb7ff9940bcdb1e4cd62bd189d10fa636ba upstream.

Unfortunately with !CONFIG_PAGEFLAGS_EXTENDED, (!PageHead) is false, and
(PageHead) is true, for tail pages.  If this is indeed the intended
behavior, which I doubt because it breaks cache cleaning on some ARM
systems, then the nomenclature is highly problematic.

This patch makes sure PageHead is only true for head pages and PageTail
is only true for tail pages, and neither is true for non-compound pages.

[ This buglet seems ancient - seems to have been introduced back in Apr
  2008 in commit 6a1e7f777f61: "pageflags: convert to the use of new
  macros".  And the reason nobody noticed is because the PageHead()
  tests are almost all about just sanity-checking, and only used on
  pages that are actual page heads.  The fact that the old code returned
  true for tail pages too was thus not really noticeable.   - Linus ]

Signed-off-by: Christoffer Dall <cdall@cs.columbia.edu>
Acked-by: Andrea Arcangeli <aarcange@redhat.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Will Deacon <Will.Deacon@arm.com>
Cc: Steve Capper <Steve.Capper@arm.com>
Cc: Christoph Lameter <cl@linux.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>

mm: fix vma_resv_map() NULL pointer

commit 4523e1458566a0e8ecfaff90f380dd23acc44d27 upstream.

hugetlb_reserve_pages() can be used for either normal file-backed
hugetlbfs mappings, or MAP_HUGETLB.  In the MAP_HUGETLB, semi-anonymous
mode, there is not a VMA around.  The new call to resv_map_put() assumed
that there was, and resulted in a NULL pointer dereference:

  BUG: unable to handle kernel NULL pointer dereference at 0000000000000030
  IP: vma_resv_map+0x9/0x30
  PGD 141453067 PUD 1421e1067 PMD 0
  Oops: 0000 [#1] PREEMPT SMP
  ...
  Pid: 14006, comm: trinity-child6 Not tainted 3.4.0+ #36
  RIP: vma_resv_map+0x9/0x30
  ...
  Process trinity-child6 (pid: 14006, threadinfo ffff8801414e0000, task ffff8801414f26b0)
  Call Trace:
    resv_map_put+0xe/0x40
    hugetlb_reserve_pages+0xa6/0x1d0
    hugetlb_file_setup+0x102/0x2c0
    newseg+0x115/0x360
    ipcget+0x1ce/0x310
    sys_shmget+0x5a/0x60
    system_call_fastpath+0x16/0x1b

This was reported by Dave Jones, but was reproducible with the
libhugetlbfs test cases, so shame on me for not running them in the
first place.

With this, the oops is gone, and the output of libhugetlbfs's
run_tests.py is identical to plain 3.4 again.

[ Marked for stable, since this was introduced by commit c50ac050811d
  ("hugetlb: fix resv_map leak in error path") which was also marked for
  stable ]

Reported-by: Dave Jones <davej@redhat.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Christoph Lameter <cl@linux.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>

hugetlb: fix resv_map leak in error path

commit c50ac050811d6485616a193eb0f37bfbd191cc89 upstream.

When called for anonymous (non-shared) mappings, hugetlb_reserve_pages()
does a resv_map_alloc().  It depends on code in hugetlbfs's
vm_ops->close() to release that allocation.

However, in the mmap() failure path, we do a plain unmap_region() without
the remove_vma() which actually calls vm_ops->close().

This is a decent fix.  This leak could get reintroduced if new code (say,
after hugetlb_reserve_pages() in hugetlbfs_file_mmap()) decides to return
an error.  But, I think it would have to unroll the reservation anyway.

Christoph's test case:

http://marc.info/?l=linux-mm&m=133728900729735

This patch applies to 3.4 and later.  A version for earlier kernels is at
https://lkml.org/lkml/2012/5/22/418.

Signed-off-by: Dave Hansen <dave@linux.vnet.ibm.com>
Acked-by: Mel Gorman <mel@csn.ul.ie>
Acked-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Reported-by: Christoph Lameter <cl@linux.com>
Tested-by: Christoph Lameter <cl@linux.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>

tracing: Fix double free when function profile init failed

commit 83e03b3fe4daffdebbb42151d5410d730ae50bd1 upstream.

On the failure path, stat->start and stat->pages will refer same page.
So it'll attempt to free the same page again and get kernel panic.

Link: http://lkml.kernel.org/r/1364820385-32027-1-git-send-email-namhyung@kernel.org
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Namhyung Kim <namhyung.kim@lge.com>
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>

tracing: Don't call page_to_pfn() if page is NULL

commit 85f2a2ef1d0ab99523e0b947a2b723f5650ed6aa upstream.

When allocating memory fails, page is NULL. page_to_pfn() will
cause the kernel panicked if we don't use sparsemem vmemmap.

Link: http://lkml.kernel.org/r/505AB1FF.8020104@cn.fujitsu.com
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Acked-by: Mel Gorman <mel@csn.ul.ie>
Reviewed-by: Minchan Kim <minchan@kernel.org>
Signed-off-by: Wen Congyang <wency@cn.fujitsu.com>
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>

Prevent interface errors with Seagate FreeAgent GoFlex

commit c531077f40abc9f2129c4c83a30b3f8d6ce1c0e7 upstream.

When using my Seagate FreeAgent GoFlex eSATAp external disk enclosure,
interface errors are always seen until 1.5Gbps is negotiated [1]. This
occurs using any disk in the enclosure, and when the disk is connected
directly with a generic passive eSATAp cable, we see stable 3Gbps
operation as expected.

Blacklist 3Gbps mode to avoid dataloss and the ~30s delay bus reset
and renegotiation incurs.

Signed-off-by: Daniel J Blueman <daniel@quora.org>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>

Fix a dead loop in async_synchronize_full()

[commit 45516ddc16abc923104d78bb3eb772ac0a09e33e in v3.0.44 - paulg ]

[Fixed upstream by commits 2955b47d2c1983998a8c5915cb96884e67f7cb53 and
a4683487f90bfe3049686fc5c566bdc1ad03ace6 from Dan Williams, but they are much
more intrusive than this tiny fix, according to Andrew - gregkh]

This patch tries to fix a dead loop in  async_synchronize_full(), which
could be seen when preemption is disabled on a single cpu machine.

void async_synchronize_full(void)
{
        do {
                async_synchronize_cookie(next_cookie);
        } while (!list_empty(&async_running) || !
list_empty(&async_pending));
}

async_synchronize_cookie() calls async_synchronize_cookie_domain() with
&async_running as the default domain to synchronize.

However, there might be some works in the async_pending list from other
domains. On a single cpu system, without preemption, there is no chance
for the other works to finish, so async_synchronize_full() enters a dead
loop.

It seems async_synchronize_full() wants to synchronize all entries in
all running lists(domains), so maybe we could just check the entry_count
to know whether all works are finished.

Currently, async_synchronize_cookie_domain() expects a non-NULL running
list ( if NULL, there would be NULL pointer dereference ), so maybe a
NULL pointer could be used as an indication for the functions to
synchronize all works in all domains.

Reported-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Signed-off-by: Li Zhong <zhong@linux.vnet.ibm.com>
Tested-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Tested-by: Christian Kujau <lists@nerdbynature.de>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Dan Williams <dan.j.williams@gmail.com>
Cc: Christian Kujau <lists@nerdbynature.de>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Cong Wang <xiyou.wangcong@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>

cgroup: remove incorrect dget/dput() pair in cgroup_create_dir()

commit 175431635ec09b1d1bba04979b006b99e8305a83 upstream.

cgroup_create_dir() does weird dancing with dentry refcnt.  On
success, it gets and then puts it achieving nothing.  On failure, it
puts but there isn't no matching get anywhere leading to the following
oops if cgroup_create_file() fails for whatever reason.

  ------------[ cut here ]------------
  kernel BUG at /work/os/work/fs/dcache.c:552!
  invalid opcode: 0000 [#1] PREEMPT SMP DEBUG_PAGEALLOC
  Modules linked in:
  CPU 2
  Pid: 697, comm: mkdir Not tainted 3.7.0-rc4-work+ #3 Bochs Bochs
  RIP: 0010:[<ffffffff811d9c0c>]  [<ffffffff811d9c0c>] dput+0x1dc/0x1e0
  RSP: 0018:ffff88001a3ebef8  EFLAGS: 00010246
  RAX: 0000000000000000 RBX: ffff88000e5b1ef8 RCX: 0000000000000403
  RDX: 0000000000000303 RSI: 2000000000000000 RDI: ffff88000e5b1f58
  RBP: ffff88001a3ebf18 R08: ffffffff82c76960 R09: 0000000000000001
  R10: ffff880015022080 R11: ffd9bed70f48a041 R12: 00000000ffffffea
  R13: 0000000000000001 R14: ffff88000e5b1f58 R15: 00007fff57656d60
  FS:  00007ff05fcb3800(0000) GS:ffff88001fd00000(0000) knlGS:0000000000000000
  CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
  CR2: 00000000004046f0 CR3: 000000001315f000 CR4: 00000000000006e0
  DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
  DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400
  Process mkdir (pid: 697, threadinfo ffff88001a3ea000, task ffff880015022080)
  Stack:
   ffff88001a3ebf48 00000000ffffffea 0000000000000001 0000000000000000
   ffff88001a3ebf38 ffffffff811cc889 0000000000000001 ffff88000e5b1ef8
   ffff88001a3ebf68 ffffffff811d1fc9 ffff8800198d7f18 ffff880019106ef8
  Call Trace:
   [<ffffffff811cc889>] done_path_create+0x19/0x50
   [<ffffffff811d1fc9>] sys_mkdirat+0x59/0x80
   [<ffffffff811d2009>] sys_mkdir+0x19/0x20
   [<ffffffff81be1e02>] system_call_fastpath+0x16/0x1b
  Code: 00 48 8d 90 18 01 00 00 48 89 93 c0 00 00 00 4c 89 a0 18 01 00 00 48 8b 83 a0 00 00 00 83 80 28 01 00 00 01 e8 e6 6f a0 00 eb 92 <0f> 0b 66 90 0f 1f 44 00 00 55 48 89 e5 41 57 41 56 49 89 fe 41
  RIP  [<ffffffff811d9c0c>] dput+0x1dc/0x1e0
   RSP <ffff88001a3ebef8>
  ---[ end trace 1277bcfd9561ddb0 ]---

Fix it by dropping the unnecessary dget/dput() pair.

Signed-off-by: Tejun Heo <tj@kernel.org>
Acked-by: Li Zefan <lizefan@huawei.com>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>

Driver core: treat unregistered bus_types as having no devices

commit 4fa3e78be7e985ca814ce2aa0c09cbee404efcf7 upstream.

A bus_type has a list of devices (klist_devices), but the list and the
subsys_private structure that contains it are not initialized until the
bus_type is registered with bus_register().

The panic/reboot path has fixups that look up devices in pci_bus_type.  If
we panic before registering pci_bus_type, the bus_type exists but the list
does not, so mach_reboot_fixups() trips over a null pointer and panics
again:

    mach_reboot_fixups
      pci_get_device
        ..
          bus_find_device(&pci_bus_type, ...)
            bus->p is NULL

Joonsoo reported a problem when panicking before PCI was initialized.
I think this patch should be sufficient to replace the patch he posted
here: https://lkml.org/lkml/2012/12/28/75 ("[PATCH] x86, reboot: skip
reboot_fixups in early boot phase")

Reported-by: Joonsoo Kim <js1304@gmail.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>

kernel/resource.c: fix stack overflow in __reserve_region_with_split()

commit 4965f5667f36a95b41cda6638875bc992bd7d18b upstream.

Using a recursive call add a non-conflicting region in
__reserve_region_with_split() could result in a stack overflow in the case
that the recursive calls are too deep.  Convert the recursive calls to an
iterative loop to avoid the problem.

Tested on a machine containing 135 regions.  The kernel no longer panicked
with stack overflow.

Also tested with code arbitrarily adding regions with no conflict,
embedding two consecutive conflicts and embedding two non-consecutive
conflicts.

Signed-off-by: T Makphaibulchoke <tmac@hp.com>
Reviewed-by: Ram Pai <linuxram@us.ibm.com>
Cc: Paul Gortmaker <paul.gortmaker@gmail.com>
Cc: Wei Yang <weiyang@linux.vnet.ibm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>

kernel/sys.c: call disable_nonboot_cpus() in kernel_restart()

commit f96972f2dc6365421cf2366ebd61ee4cf060c8d5 upstream.

As kernel_power_off() calls disable_nonboot_cpus(), we may also want to
have kernel_restart() call disable_nonboot_cpus().  Doing so can help
machines that require boot cpu be the last alive cpu during reboot to
survive with kernel restart.

This fixes one reboot issue seen on imx6q (Cortex-A9 Quad).  The machine
requires that the restart routine be run on the primary cpu rather than
secondary ones.  Otherwise, the secondary core running the restart
routine will fail to come to online after reboot.

Signed-off-by: Shawn Guo <shawn.guo@linaro.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>

coredump: prevent double-free on an error path in core dumper

commit f34f9d186df35e5c39163444c43b4fc6255e39c5 upstream.

In !CORE_DUMP_USE_REGSET case, if elf_note_info_init fails to allocate
memory for info->fields, it frees already allocated stuff and returns
error to its caller, fill_note_info.  Which in turn returns error to its
caller, elf_core_dump.  Which jumps to cleanup label and calls
free_note_info, which will happily try to free all info->fields again.
BOOM.

This is the fix.

Signed-off-by: Oleg Nesterov <oleg@redhat.com>
Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
Cc: Venu Byravarasu <vbyravarasu@nvidia.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>

wake_up_process() should be never used to wakeup a TASK_STOPPED/TRACED task

commit 9067ac85d533651b98c2ff903182a20cbb361fcb upstream.

wake_up_process() should never wakeup a TASK_STOPPED/TRACED task.
Change it to use TASK_NORMAL and add the WARN_ON().

TASK_ALL has no other users, probably can be killed.

Signed-off-by: Oleg Nesterov <oleg@redhat.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
[PG: kernel/sched/core.c --> kernel/sched.c on 2.6.34]
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>

kernel/signal.c: stop info leak via the tkill and the tgkill syscalls

commit b9e146d8eb3b9ecae5086d373b50fa0c1f3e7f0f upstream.

This fixes a kernel memory contents leak via the tkill and tgkill syscalls
for compat processes.

This is visible in the siginfo_t->_sifields._rt.si_sigval.sival_ptr field
when handling signals delivered from tkill.

The place of the infoleak:

int copy_siginfo_to_user32(compat_siginfo_t __user *to, siginfo_t *from)
{
        ...
        put_user_ex(ptr_to_compat(from->si_ptr), &to->si_ptr);
        ...
}

Signed-off-by: Emese Revfy <re.emese@gmail.com>
Reviewed-by: PaX Team <pageexec@freemail.hu>
Signed-off-by: Kees Cook <keescook@chromium.org>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: Oleg Nesterov <oleg@redhat.com>
Cc: "Eric W. Biederman" <ebiederm@xmission.com>
Cc: Serge Hallyn <serge.hallyn@canonical.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>

ptrace: ptrace_resume() shouldn't wake up !TASK_TRACED thread

commit 0666fb51b1483f27506e212cc7f7b2645b5c7acc upstream.

It is not clear why ptrace_resume() does wake_up_process(). Unless the
caller is PTRACE_KILL the tracee should be TASK_TRACED so we can use
wake_up_state(__TASK_TRACED). If sys_ptrace() races with SIGKILL we do
not need the extra and potentionally spurious wakeup.

If the caller is PTRACE_KILL, wake_up_process() is even more wrong.
The tracee can sleep in any state in any place, and if we have a buggy
code which doesn't handle a spurious wakeup correctly PTRACE_KILL can
be used to exploit it. For example:

int main(void)
{
int child, status;

child = fork();
if (!child) {
int ret;

assert(ptrace(PTRACE_TRACEME, 0,0,0) == 0);

ret = pause();
printf("pause: %d %m\n", ret);

return 0x23;
}

sleep(1);
assert(ptrace(PTRACE_KILL, child, 0,0) == 0);

assert(child == wait(&status));
printf("wait: %x\n", status);

return 0;
}

prints "pause: -1 Unknown error 514", -ERESTARTNOHAND leaks to the
userland. In this case sys_pause() is buggy as well and should be
fixed.

I do not know what was the original rationality behind PTRACE_KILL.
The man page is simply wrong and afaics it was always wrong. Imho
it should be deprecated, or may be it should do send_sig(SIGKILL)
as Denys suggests, but in any case I do not think that the current
behaviour was intentional.

Note: there is another problem, ptrace_resume() changes ->exit_code
and this can race with SIGKILL too. Eventually we should change ptrace
to not use ->exit_code.

Signed-off-by: Oleg Nesterov <oleg@redhat.com>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>

exec: use -ELOOP for max recursion depth

commit d740269867021faf4ce38a449353d2b986c34a67 upstream.

To avoid an explosion of request_module calls on a chain of abusive
scripts, fail maximum recursion with -ELOOP instead of -ENOEXEC. As soon
as maximum recursion depth is hit, the error will fail all the way back
up the chain, aborting immediately.

This also has the side-effect of stopping the user's shell from attempting
to reexecute the top-level file as a shell script. As seen in the
dash source:

        if (cmd != path_bshell && errno == ENOEXEC) {
                *argv-- = cmd;
                *argv = cmd = path_bshell;
                goto repeat;
        }

The above logic was designed for running scripts automatically that lacked
the "#!" header, not to re-try failed recursion. On a legitimate -ENOEXEC,
things continue to behave as the shell expects.

Additionally, when tracking recursion, the binfmt handlers should not be
involved. The recursion being tracked is the depth of calls through
search_binary_handler(), so that function should be exclusively responsible
for tracking the depth.

Signed-off-by: Kees Cook <keescook@chromium.org>
Cc: halfdog <me@halfdog.net>
Cc: P J P <ppandit@redhat.com>
Cc: Alexander Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>

exec: do not leave bprm->interp on stack

commit b66c5984017533316fd1951770302649baf1aa33 upstream.

If a series of scripts are executed, each triggering module loading via
unprintable bytes in the script header, kernel stack contents can leak
into the command line.

Normally execution of binfmt_script and binfmt_misc happens recursively.
However, when modules are enabled, and unprintable bytes exist in the
bprm->buf, execution will restart after attempting to load matching
binfmt modules.  Unfortunately, the logic in binfmt_script and
binfmt_misc does not expect to get restarted.  They leave bprm->interp
pointing to their local stack.  This means on restart bprm->interp is
left pointing into unused stack memory which can then be copied into the
userspace argv areas.

After additional study, it seems that both recursion and restart remains
the desirable way to handle exec with scripts, misc, and modules.  As
such, we need to protect the changes to interp.

This changes the logic to require allocation for any changes to the
bprm->interp.  To avoid adding a new kmalloc to every exec, the default
value is left as-is.  Only when passing through binfmt_script or
binfmt_misc does an allocation take place.

For a proof of concept, see DoTest.sh from:

   http://www.halfdog.net/Security/2012/LinuxKernelBinfmtScriptStackDataDisclosure/

Signed-off-by: Kees Cook <keescook@chromium.org>
Cc: halfdog <me@halfdog.net>
Cc: P J P <ppandit@redhat.com>
Cc: Alexander Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>

gen_init_cpio: avoid stack overflow when expanding

commit 20f1de659b77364d55d4e7fad2ef657e7730323f upstream.

Fix possible overflow of the buffer used for expanding environment
variables when building file list.

In the extremely unlikely case of an attacker having control over the
environment variables visible to gen_init_cpio, control over the
contents of the file gen_init_cpio parses, and gen_init_cpio was built
without compiler hardening, the attacker can gain arbitrary execution
control via a stack buffer overflow.

  $ cat usr/crash.list
  file foo ${BIG}${BIG}${BIG}${BIG}${BIG}${BIG} 0755 0 0
  $ BIG=$(perl -e 'print "A" x 4096;') ./usr/gen_init_cpio usr/crash.list
  *** buffer overflow detected ***: ./usr/gen_init_cpio terminated

This also replaces the space-indenting with tabs.

Patch based on existing fix extracted from grsecurity.

Signed-off-by: Kees Cook <keescook@chromium.org>
Cc: Michal Marek <mmarek@suse.cz>
Cc: Brad Spengler <spender@grsecurity.net>
Cc: PaX Team <pageexec@freemail.hu>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>

tick: Cleanup NOHZ per cpu data on cpu down

commit 4b0c0f294f60abcdd20994a8341a95c8ac5eeb96 upstream.

Prarit reported a crash on CPU offline/online. The reason is that on
CPU down the NOHZ related per cpu data of the dead cpu is not cleaned
up. If at cpu online an interrupt happens before the per cpu tick
device is registered the irq_enter() check potentially sees stale data
and dereferences a NULL pointer.

Cleanup the data after the cpu is dead.

Reported-by: Prarit Bhargava <prarit@redhat.com>
Cc: Mike Galbraith <bitbucket@online.de>
Link: http://lkml.kernel.org/r/alpine.LFD.2.02.1305031451561.2886@ionos
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>

timer: Don't reinitialize the cpu base lock during CPU_UP_PREPARE

commit 42a5cf46cd56f46267d2a9fcf2655f4078cd3042 upstream.

An inactive timer's base can refer to a offline cpu's base.

In the current code, cpu_base's lock is blindly reinitialized each
time a CPU is brought up. If a CPU is brought online during the period
that another thread is trying to modify an inactive timer on that CPU
with holding its timer base lock, then the lock will be reinitialized
under its feet. This leads to following SPIN_BUG().

<0> BUG: spinlock already unlocked on CPU#3, kworker/u:3/1466
<0> lock: 0xe3ebe000, .magic: dead4ead, .owner: kworker/u:3/1466, .owner_cpu: 1
<4> [<c0013dc4>] (unwind_backtrace+0x0/0x11c) from [<c026e794>] (do_raw_spin_unlock+0x40/0xcc)
<4> [<c026e794>] (do_raw_spin_unlock+0x40/0xcc) from [<c076c160>] (_raw_spin_unlock+0x8/0x30)
<4> [<c076c160>] (_raw_spin_unlock+0x8/0x30) from [<c009b858>] (mod_timer+0x294/0x310)
<4> [<c009b858>] (mod_timer+0x294/0x310) from [<c00a5e04>] (queue_delayed_work_on+0x104/0x120)
<4> [<c00a5e04>] (queue_delayed_work_on+0x104/0x120) from [<c04eae00>] (sdhci_msm_bus_voting+0x88/0x9c)
<4> [<c04eae00>] (sdhci_msm_bus_voting+0x88/0x9c) from [<c04d8780>] (sdhci_disable+0x40/0x48)
<4> [<c04d8780>] (sdhci_disable+0x40/0x48) from [<c04bf300>] (mmc_release_host+0x4c/0xb0)
<4> [<c04bf300>] (mmc_release_host+0x4c/0xb0) from [<c04c7aac>] (mmc_sd_detect+0x90/0xfc)
<4> [<c04c7aac>] (mmc_sd_detect+0x90/0xfc) from [<c04c2504>] (mmc_rescan+0x7c/0x2c4)
<4> [<c04c2504>] (mmc_rescan+0x7c/0x2c4) from [<c00a6a7c>] (process_one_work+0x27c/0x484)
<4> [<c00a6a7c>] (process_one_work+0x27c/0x484) from [<c00a6e94>] (worker_thread+0x210/0x3b0)
<4> [<c00a6e94>] (worker_thread+0x210/0x3b0) from [<c00aad9c>] (kthread+0x80/0x8c)
<4> [<c00aad9c>] (kthread+0x80/0x8c) from [<c000ea80>] (kernel_thread_exit+0x0/0x8)

As an example, this particular crash occurred when CPU #3 is executing
mod_timer() on an inactive timer whose base is refered to offlined CPU
#2.  The code locked the timer_base corresponding to CPU #2. Before it
could proceed, CPU #2 came online and reinitialized the spinlock
corresponding to its base. Thus now CPU #3 held a lock which was
reinitialized. When CPU #3 finally ended up unlocking the old cpu_base
corresponding to CPU #2, we hit the above SPIN_BUG().

CPU #0 CPU #3        CPU #2
------ -------        -------
.....  ......       <Offline>
mod_timer()
 lock_timer_base
   spin_lock_irqsave(&base->lock)

cpu_up(2)  .....         ......
init_timers_cpu()
....  .....      spin_lock_init(&base->lock)
.....    spin_unlock_irqrestore(&base->lock)  ......
   <spin_bug>

Allocation of per_cpu timer vector bases is done only once under
"tvec_base_done[]" check. In the current code, spinlock_initialization
of base->lock isn't under this check. When a CPU is up each time the
base lock is reinitialized. Move base spinlock initialization under
the check.

Signed-off-by: Tirupathi Reddy <tirupath@codeaurora.org>
Link: http://lkml.kernel.org/r/1368520142-4136-1-git-send-email-tirupath@codeaurora.org
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>

posix-cpu-timers: Fix nanosleep task_struct leak

commit e6c42c295e071dd74a66b5a9fcf4f44049888ed8 upstream.

The trinity fuzzer triggered a task_struct reference leak via
clock_nanosleep with CPU_TIMERs. do_cpu_nanosleep() calls
posic_cpu_timer_create(), but misses a corresponding
posix_cpu_timer_del() which leads to the task_struct reference leak.

Reported-and-tested-by: Tommi Rantala <tt.rantala@gmail.com>
Signed-off-by: Stanislaw Gruszka <sgruszka@redhat.com>
Cc: Dave Jones <davej@redhat.com>
Cc: John Stultz <john.stultz@linaro.org>
Cc: Oleg Nesterov <oleg@redhat.com>
Link: http://lkml.kernel.org/r/20130215100810.GF4392@redhat.com
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>

clockevents: Don't allow dummy broadcast timers

commit a7dc19b8652c862d5b7c4d2339bd3c428bd29c4a upstream.

Currently tick_check_broadcast_device doesn't reject clock_event_devices
with CLOCK_EVT_FEAT_DUMMY, and may select them in preference to real
hardware if they have a higher rating value. In this situation, the
dummy timer is responsible for broadcasting to itself, and the core
clockevents code may attempt to call non-existent callbacks for
programming the dummy, eventually leading to a panic.

This patch makes tick_check_broadcast_device always reject dummy timers,
preventing this problem.

Signed-off-by: Mark Rutland <mark.rutland@arm.com>
Cc: linux-arm-kernel@lists.infradead.org
Cc: Jon Medhurst (Tixy) <tixy@linaro.org>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>

hfsplus: fix potential overflow in hfsplus_file_truncate()

commit 12f267a20aecf8b84a2a9069b9011f1661c779b4 upstream.

Change a u32 to loff_t hfsplus_file_truncate().

Signed-off-by: Vyacheslav Dubeyko <slava@dubeyko.com>
Cc: Christoph Hellwig <hch@infradead.org>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: Hin-Tak Leung <htl10@users.sourceforge.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>

kernel panic when mount NFSv4

commit beb0f0a9fba1fa98b378329a9a5b0a73f25097ae upstream.

On Tue, 2010-12-14 at 16:58 +0800, Mi Jinlong wrote:
> Hi,
>
> When testing NFSv4 at RHEL6 with kernel 2.6.32, I got a kernel panic
> at NFS client's __rpc_create_common function.
>
> The panic place is:
>   rpc_mkpipe
>     __rpc_lookup_create()          <=== find pipefile *idmap*
>     __rpc_mkpipe()                 <=== pipefile is *idmap*
>       __rpc_create_common()
>        ******  BUG_ON(!d_unhashed(dentry)); ******    *panic*
>
> It means that the dentry's d_flags have be set DCACHE_UNHASHED,
> but it should not be set here.
>
> Is someone known this bug? or give me some idea?
>
> A reproduce program is append, but it can't reproduce the bug every time.
> the export is: "/nfsroot       *(rw,no_root_squash,fsid=0,insecure)"
>
> And the panic message is append.
>
> ============================================================================
> #!/bin/sh
>
> LOOPTOTAL=768
> LOOPCOUNT=0
> ret=0
>
> while [ $LOOPCOUNT -ne $LOOPTOTAL ]
> do
>  ((LOOPCOUNT += 1))
>  service nfs restart
>  /usr/sbin/rpc.idmapd
>  mount -t nfs4 127.0.0.1:/ /mnt|| return 1;
>  ls -l /var/lib/nfs/rpc_pipefs/nfs/*/
>  umount /mnt
>  echo $LOOPCOUNT
> done
>
> ===============================================================================
> Code: af 60 01 00 00 89 fa 89 f0 e8 64 cf 89 f0 e8 5c 7c 64 cf 31 c0 8b 5c 24 10 8b
> 74 24 14 8b 7c 24 18 8b 6c 24 1c 83 c4 20 c3 <0f> 0b eb fc 8b 46 28 c7 44 24 08 20
> de ee f0 c7 44 24 04 56 ea
> EIP:[<f0ee92ea>] __rpc_create_common+0x8a/0xc0 [sunrpc] SS:ESP 0068:eccb5d28
> ---[ end trace 8f5606cd08928ed2]---
> Kernel panic - not syncing: Fatal exception
> Pid:7131, comm: mount.nfs4 Tainted: G     D   -------------------2.6.32 #1
> Call Trace:
>  [<c080ad18>] ? panic+0x42/0xed
>  [<c080e42c>] ? oops_end+0xbc/0xd0
>  [<c040b090>] ? do_invalid_op+0x0/0x90
>  [<c040b10f>] ? do_invalid_op+0x7f/0x90
>  [<f0ee92ea>] ? __rpc_create_common+0x8a/0xc0[sunrpc]
>  [<f0edc433>] ? rpc_free_task+0x33/0x70[sunrpc]
>  [<f0ed6508>] ? prc_call_sync+0x48/0x60[sunrpc]
>  [<f0ed656e>] ? rpc_ping+0x4e/0x60[sunrpc]
>  [<f0ed6eaf>] ? rpc_create+0x38f/0x4f0[sunrpc]
>  [<c080d80b>] ? error_code+0x73/0x78
>  [<f0ee92ea>] ? __rpc_create_common+0x8a/0xc0[sunrpc]
>  [<c0532bda>] ? d_lookup+0x2a/0x40
>  [<f0ee94b1>] ? rpc_mkpipe+0x111/0x1b0[sunrpc]
>  [<f10a59f4>] ? nfs_create_rpc_client+0xb4/0xf0[nfs]
>  [<f10d6c6d>] ? nfs_fscache_get_client_cookie+0x1d/0x50[nfs]
>  [<f10d3fcb>] ? nfs_idmap_new+0x7b/0x140[nfs]
>  [<c05e76aa>] ? strlcpy+0x3a/0x60
>  [<f10a60ca>] ? nfs4_set_client+0xea/0x2b0[nfs]
>  [<f10a6d0c>] ? nfs4_create_server+0xac/0x1b0[nfs]
>  [<c04f1400>] ? krealloc+0x40/0x50
>  [<f10b0e8b>] ? nfs4_remote_get_sb+0x6b/0x250[nfs]
>  [<c04f14ec>] ? kstrdup+0x3c/0x60
>  [<c0520739>] ? vfs_kern_mount+0x69/0x170
>  [<f10b1a3c>] ? nfs_do_root_mount+0x6c/0xa0[nfs]
>  [<f10b1b47>] ? nfs4_try_mount+0x37/0xa0[nfs]
>  [<f10afe6d>] ? nfs4_validate_text_mount_data+-x7d/0xf0[nfs]
>  [<f10b1c42>] ? nfs4_get_sb+0x92/0x2f0
>  [<c0520739>] ? vfs_kern_mount+0x69/0x170
>  [<c05366d2>] ? get_fs_type+0x32/0xb0
>  [<c052089f>] ? do_kern_mount+0x3f/0xe0
>  [<c053954f>] ? do_mount+0x2ef/0x740
>  [<c0537740>] ? copy_mount_options+0xb0/0x120
>  [<c0539a0e>] ? sys_mount+0x6e/0xa0

Hi,

Does the following patch fix the problem?

Cheers
  Trond

--------------------------
SUNRPC: Fix a BUG in __rpc_create_common

From: Trond Myklebust <Trond.Myklebust@netapp.com>

Mi Jinlong reports:

When testing NFSv4 at RHEL6 with kernel 2.6.32, I got a kernel panic
at NFS client's __rpc_create_common function.

The panic place is:
  rpc_mkpipe
      __rpc_lookup_create()          <=== find pipefile *idmap*
      __rpc_mkpipe()                 <=== pipefile is *idmap*
        __rpc_create_common()
         ******  BUG_ON(!d_unhashed(dentry)); ****** *panic*

The test is wrong: we can find ourselves with a hashed negative dentry here
if the idmapper tried to look up the file before we got round to creating
it.

Just replace the BUG_ON() with a d_drop(dentry).

Reported-by: Mi Jinlong <mijinlong@cn.fujitsu.com>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>

NFSv4: Revalidate uid/gid after open

This is a shorter (and more appropriate for stable kernels) analog to
the following upstream commit:

commit 6926afd1925a54a13684ebe05987868890665e2b
Author: Trond Myklebust <Trond.Myklebust@netapp.com>
Date:   Sat Jan 7 13:22:46 2012 -0500

    NFSv4: Save the owner/group name string when doing open

    ...so that we can do the uid/gid mapping outside the asynchronous RPC
    context.
    This fixes a bug in the current NFSv4 atomic open code where the client
    isn't able to determine what the true uid/gid fields of the file are,
    (because the asynchronous nature of the OPEN call denies it the ability
    to do an upcall) and so fills them with default values, marking the
    inode as needing revalidation.
    Unfortunately, in some cases, the VFS will do some additional sanity
    checks on the file, and may override the server's decision to allow
    the open because it sees the wrong owner/group fields.

Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
Without this patch, logging into two different machines with home
directories mounted over NFS4 and then running "vim" and typing ":q"
in each reliably produces the following error on the second machine:

E137: Viminfo file is not writable: /users/system/rtheys/.viminfo

This regression was introduced by 80e52aced138 ("NFSv4: Don't do
idmapper upcalls for asynchronous RPC calls", merged during the 2.6.32
cycle) --- after the OPEN call, .viminfo has the default values for
st_uid and st_gid (0xfffffffe) cached because we do not want to let
rpciod wait for an idmapper upcall to fill them in.

The fix used in mainline is to save the owner and group as strings and
perform the upcall in _nfs4_proc_open outside the rpciod context,
which takes about 600 lines.  For stable, we can do something similar
with a one-liner: make open check for the stale fields and make a
(synchronous) GETATTR call to fill them when needed.

Trond dictated the patch, I typed it in, and Rik tested it.

Addresses http://bugs.debian.org/659111 and
          https://bugzilla.redhat.com/789298

Reported-by: Rik Theys <Rik.Theys@esat.kuleuven.be>
Explained-by: David Flyn <davidf@rd.bbc.co.uk>
Signed-off-by: Jonathan Nieder <jrnieder@gmail.com>
Tested-by: Rik Theys <Rik.Theys@esat.kuleuven.be>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
[PG: commit 19165bdbb3622cfca0ff66e8b30248d469b849d6 in v3.0.32]
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>

NFSv3: Ensure that do_proc_get_root() reports errors correctly

commit 086600430493e04b802bee6e5b3ce0458e4eb77f upstream.

If the rpc call to NFS3PROC_FSINFO fails, then we need to report that
error so that the mount fails. Otherwise we can end up with a
superblock with completely unusable values for block sizes, maxfilesize,
etc.

Reported-by: Yuanming Chen <hikvision_linux@163.com>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>

nfsd4: fix oops on unusual readlike compound

commit d5f50b0c290431c65377c4afa1c764e2c3fe5305 upstream.

If the argument and reply together exceed the maximum payload size, then
a reply with a read-like operation can overlow the rq_pages array.

Signed-off-by: J. Bruce Fields <bfields@redhat.com>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>

tg3: Avoid null pointer dereference in tg3_interrupt in netconsole mode

commit 9c13cb8bb477a83b9a3c9e5a5478a4e21294a760 upstream.

When netconsole is enabled, logging messages generated during tg3_open
can result in a null pointer dereference for the uninitialized tg3
status block. Use the irq_sync flag to disable polling in the early
stages. irq_sync is cleared when the driver is enabling interrupts after
all initialization is completed.

Signed-off-by: Nithin Nayak Sujir <nsujir@broadcom.com>
Signed-off-by: Michael Chan <mchan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
[PG: drivers/net/ethernet/broadcom/tg3.c --> drivers/net/tg3.c in .34]
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>

b43legacy: Fix crash on unload when firmware not available

commit 2d838bb608e2d1f6cb4280e76748cb812dc822e7 upstream.

When b43legacy is loaded without the firmware being available, a following
unload generates a kernel NULL pointer dereference BUG as follows:

[  214.330789] BUG: unable to handle kernel NULL pointer dereference at 0000004c
[  214.330997] IP: [<c104c395>] drain_workqueue+0x15/0x170
[  214.331179] *pde = 00000000
[  214.331311] Oops: 0000 [#1] SMP
[  214.331471] Modules linked in: b43legacy(-) ssb pcmcia mac80211 cfg80211 af_packet mperf arc4 ppdev sr_mod cdrom sg shpchp yenta_socket pcmcia_rsrc pci_hotplug pcmcia_core battery parport_pc parport floppy container ac button edd autofs4 ohci_hcd ehci_hcd usbcore usb_common thermal processor scsi_dh_rdac scsi_dh_hp_sw scsi_dh_emc scsi_dh_alua scsi_dh fan thermal_sys hwmon ata_generic pata_ali libata [last unloaded: cfg80211]
[  214.333421] Pid: 3639, comm: modprobe Not tainted 3.6.0-rc6-wl+ #163 Source Technology VIC 9921/ALI Based Notebook
[  214.333580] EIP: 0060:[<c104c395>] EFLAGS: 00010246 CPU: 0
[  214.333687] EIP is at drain_workqueue+0x15/0x170
[  214.333788] EAX: c162ac40 EBX: cdfb8360 ECX: 0000002a EDX: 00002a2a
[  214.333890] ESI: 00000000 EDI: 00000000 EBP: cd767e7c ESP: cd767e5c
[  214.333957]  DS: 007b ES: 007b FS: 00d8 GS: 0033 SS: 0068
[  214.333957] CR0: 8005003b CR2: 0000004c CR3: 0c96a000 CR4: 00000090
[  214.333957] DR0: 00000000 DR1: 00000000 DR2: 00000000 DR3: 00000000
[  214.333957] DR6: ffff0ff0 DR7: 00000400
[  214.333957] Process modprobe (pid: 3639, ti=cd766000 task=cf802e90 task.ti=cd766000)
[  214.333957] Stack:
[  214.333957]  00000292 cd767e74 c12c5e09 00000296 00000296 cdfb8360 cdfb9220 00000000
[  214.333957]  cd767e90 c104c4fd cdfb8360 cdfb9220 cd682800 cd767ea4 d0c10184 cd682800
[  214.333957]  cd767ea4 cba31064 cd767eb8 d0867908 cba31064 d087e09c cd96f034 cd767ec4
[  214.333957] Call Trace:
[  214.333957]  [<c12c5e09>] ? skb_dequeue+0x49/0x60
[  214.333957]  [<c104c4fd>] destroy_workqueue+0xd/0x150
[  214.333957]  [<d0c10184>] ieee80211_unregister_hw+0xc4/0x100 [mac80211]
[  214.333957]  [<d0867908>] b43legacy_remove+0x78/0x80 [b43legacy]
[  214.333957]  [<d083654d>] ssb_device_remove+0x1d/0x30 [ssb]
[  214.333957]  [<c126f15a>] __device_release_driver+0x5a/0xb0
[  214.333957]  [<c126fb07>] driver_detach+0x87/0x90
[  214.333957]  [<c126ef4c>] bus_remove_driver+0x6c/0xe0
[  214.333957]  [<c1270120>] driver_unregister+0x40/0x70
[  214.333957]  [<d083686b>] ssb_driver_unregister+0xb/0x10 [ssb]
[  214.333957]  [<d087c488>] b43legacy_exit+0xd/0xf [b43legacy]
[  214.333957]  [<c1089dde>] sys_delete_module+0x14e/0x2b0
[  214.333957]  [<c110a4a7>] ? vfs_write+0xf7/0x150
[  214.333957]  [<c1240050>] ? tty_write_lock+0x50/0x50
[  214.333957]  [<c110a6f8>] ? sys_write+0x38/0x70
[  214.333957]  [<c1397c55>] syscall_call+0x7/0xb
[  214.333957] Code: bc 27 00 00 00 00 a1 74 61 56 c1 55 89 e5 e8 a3 fc ff ff 5d c3 90 55 89 e5 57 56 89 c6 53 b8 40 ac 62 c1 83 ec 14 e8 bb b7 34 00 <8b> 46 4c 8d 50 01 85 c0 89 56 4c 75 03 83 0e 40 80 05 40 ac 62
[  214.333957] EIP: [<c104c395>] drain_workqueue+0x15/0x170 SS:ESP 0068:cd767e5c
[  214.333957] CR2: 000000000000004c
[  214.341110] ---[ end trace c7e90ec026d875a6 ]---Index: wireless-testing/drivers/net/wireless/b43legacy/main.c

The problem is fixed by making certain that the ucode pointer is not NULL
before deregistering the driver in mac80211.

Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>

xfrm_user: return error pointer instead of NULL #2

commit c25463722509fef0ed630b271576a8c9a70236f3 upstream.

When dump_one_policy() returns an error, e.g. because of a too small
buffer to dump the whole xfrm policy, xfrm_policy_netlink() returns
NULL instead of an error pointer. But its caller expects an error
pointer and therefore continues to operate on a NULL skbuff.

Signed-off-by: Mathias Krause <minipli@googlemail.com>
Acked-by: Steffen Klassert <steffen.klassert@secunet.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>

xfrm_user: return error pointer instead of NULL

commit 864745d291b5ba80ea0bd0edcbe67273de368836 upstream.

When dump_one_state() returns an error, e.g. because of a too small
buffer to dump the whole xfrm state, xfrm_state_netlink() returns NULL
instead of an error pointer. But its callers expect an error pointer
and therefore continue to operate on a NULL skbuff.

This could lead to a privilege escalation (execution of user code in
kernel context) if the attacker has CAP_NET_ADMIN and is able to map
address 0.

Signed-off-by: Mathias Krause <minipli@googlemail.com>
Acked-by: Steffen Klassert <steffen.klassert@secunet.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>

xfrm_user: fix info leak in copy_to_user_tmpl()

commit 1f86840f897717f86d523a13e99a447e6a5d2fa5 upstream.

The memory used for the template copy is a local stack variable. As
struct xfrm_user_tmpl contains multiple holes added by the compiler for
alignment, not initializing the memory will lead to leaking stack bytes
to userland. Add an explicit memset(0) to avoid the info leak.

Initial version of the patch by Brad Spengler.

Cc: Brad Spengler <spender@grsecurity.net>
Signed-off-by: Mathias Krause <minipli@googlemail.com>
Acked-by: Steffen Klassert <steffen.klassert@secunet.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>

xfrm_user: fix info leak in copy_to_user_policy()

commit 7b789836f434c87168eab067cfbed1ec4783dffd upstream.

The memory reserved to dump the xfrm policy includes multiple padding
bytes added by the compiler for alignment (padding bytes in struct
xfrm_selector and struct xfrm_userpolicy_info). Add an explicit
memset(0) before filling the buffer to avoid the heap info leak.

Signed-off-by: Mathias Krause <minipli@googlemail.com>
Acked-by: Steffen Klassert <steffen.klassert@secunet.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>

xfrm_user: fix info leak in copy_to_user_state()

commit f778a636713a435d3a922c60b1622a91136560c1 upstream.

The memory reserved to dump the xfrm state includes the padding bytes of
struct xfrm_usersa_info added by the compiler for alignment (7 for
amd64, 3 for i386). Add an explicit memset(0) before filling the buffer
to avoid the info leak.

Signed-off-by: Mathias Krause <minipli@googlemail.com>
Acked-by: Steffen Klassert <steffen.klassert@secunet.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>

drop_monitor: dont sleep in atomic context

commit bec4596b4e6770c7037f21f6bd27567b152dc0d6 upstream.

drop_monitor calls several sleeping functions while in atomic context.

 BUG: sleeping function called from invalid context at mm/slub.c:943
 in_atomic(): 1, irqs_disabled(): 0, pid: 2103, name: kworker/0:2
 Pid: 2103, comm: kworker/0:2 Not tainted 3.5.0-rc1+ #55
 Call Trace:
  [<ffffffff810697ca>] __might_sleep+0xca/0xf0
  [<ffffffff811345a3>] kmem_cache_alloc_node+0x1b3/0x1c0
  [<ffffffff8105578c>] ? queue_delayed_work_on+0x11c/0x130
  [<ffffffff815343fb>] __alloc_skb+0x4b/0x230
  [<ffffffffa00b0360>] ? reset_per_cpu_data+0x160/0x160 [drop_monitor]
  [<ffffffffa00b022f>] reset_per_cpu_data+0x2f/0x160 [drop_monitor]
  [<ffffffffa00b03ab>] send_dm_alert+0x4b/0xb0 [drop_monitor]
  [<ffffffff810568e0>] process_one_work+0x130/0x4c0
  [<ffffffff81058249>] worker_thread+0x159/0x360
  [<ffffffff810580f0>] ? manage_workers.isra.27+0x240/0x240
  [<ffffffff8105d403>] kthread+0x93/0xa0
  [<ffffffff816be6d4>] kernel_thread_helper+0x4/0x10
  [<ffffffff8105d370>] ? kthread_freezable_should_stop+0x80/0x80
  [<ffffffff816be6d0>] ? gs_change+0xb/0xb

Rework the logic to call the sleeping functions in right context.

Use standard timer/workqueue api to let system chose any cpu to perform
the allocation and netlink send.

Also avoid a loop if reset_per_cpu_data() cannot allocate memory :
use mod_timer() to wait 1/10 second before next try.

Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Neil Horman <nhorman@tuxdriver.com>
Reviewed-by: Neil Horman <nhorman@tuxdriver.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
[PG: diffstat here is less by one line due to blank line removal]
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>

drop_monitor: prevent init path from scheduling on the wrong cpu

commit 4fdcfa12843bca38d0c9deff70c8720e4e8f515f upstream.

I just noticed after some recent updates, that the init path for the drop
monitor protocol has a minor error.  drop monitor maintains a per cpu structure,
that gets initalized from a single cpu.  Normally this is fine, as the protocol
isn't in use yet, but I recently made a change that causes a failed skb
allocation to reschedule itself .  Given the current code, the implication is
that this workqueue reschedule will take place on the wrong cpu.  If drop
monitor is used early during the boot process, its possible that two cpus will
access a single per-cpu structure in parallel, possibly leading to data
corruption.

This patch fixes the situation, by storing the cpu number that a given instance
of this per-cpu data should be accessed from.  In the case of a need for a
reschedule, the cpu stored in the struct is assigned the rescheule, rather than
the currently executing cpu

Tested successfully by myself.

Signed-off-by: Neil Horman <nhorman@tuxdriver.com>
CC: David Miller <davem@davemloft.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>

drop_monitor: Make updating data->skb smp safe

commit 3885ca785a3618593226687ced84f3f336dc3860 upstream.

Eric Dumazet pointed out to me that the drop_monitor protocol has some holes in
its smp protections.  Specifically, its possible to replace data->skb while its
being written.  This patch corrects that by making data->skb an rcu protected
variable.  That will prevent it from being overwritten while a tracepoint is
modifying it.

Signed-off-by: Neil Horman <nhorman@tuxdriver.com>
Reported-by: Eric Dumazet <eric.dumazet@gmail.com>
CC: David Miller <davem@davemloft.net>
Acked-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
[PG: The "__rcu" sparse addition is only present in 2.6.37+ kernels.]
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>

drop_monitor: fix sleeping in invalid context warning

commit cde2e9a651b76d8db36ae94cd0febc82b637e5dd upstream.

Eric Dumazet pointed out this warning in the drop_monitor protocol to me:

[   38.352571] BUG: sleeping function called from invalid context at kernel/mutex.c:85
[   38.352576] in_atomic(): 1, irqs_disabled(): 0, pid: 4415, name: dropwatch
[   38.352580] Pid: 4415, comm: dropwatch Not tainted 3.4.0-rc2+ #71
[   38.352582] Call Trace:
[   38.352592]  [<ffffffff8153aaf0>] ? trace_napi_poll_hit+0xd0/0xd0
[   38.352599]  [<ffffffff81063f2a>] __might_sleep+0xca/0xf0
[   38.352606]  [<ffffffff81655b16>] mutex_lock+0x26/0x50
[   38.352610]  [<ffffffff8153aaf0>] ? trace_napi_poll_hit+0xd0/0xd0
[   38.352616]  [<ffffffff810b72d9>] tracepoint_probe_register+0x29/0x90
[   38.352621]  [<ffffffff8153a585>] set_all_monitor_traces+0x105/0x170
[   38.352625]  [<ffffffff8153a8ca>] net_dm_cmd_trace+0x2a/0x40
[   38.352630]  [<ffffffff8154a81a>] genl_rcv_msg+0x21a/0x2b0
[   38.352636]  [<ffffffff810f8029>] ? zone_statistics+0x99/0xc0
[   38.352640]  [<ffffffff8154a600>] ? genl_rcv+0x30/0x30
[   38.352645]  [<ffffffff8154a059>] netlink_rcv_skb+0xa9/0xd0
[   38.352649]  [<ffffffff8154a5f0>] genl_rcv+0x20/0x30
[   38.352653]  [<ffffffff81549a7e>] netlink_unicast+0x1ae/0x1f0
[   38.352658]  [<ffffffff81549d76>] netlink_sendmsg+0x2b6/0x310
[   38.352663]  [<ffffffff8150824f>] sock_sendmsg+0x10f/0x130
[   38.352668]  [<ffffffff8150abe0>] ? move_addr_to_kernel+0x60/0xb0
[   38.352673]  [<ffffffff81515f04>] ? verify_iovec+0x64/0xe0
[   38.352677]  [<ffffffff81509c46>] __sys_sendmsg+0x386/0x390
[   38.352682]  [<ffffffff810ffaf9>] ? handle_mm_fault+0x139/0x210
[   38.352687]  [<ffffffff8165b5bc>] ? do_page_fault+0x1ec/0x4f0
[   38.352693]  [<ffffffff8106ba4d>] ? set_next_entity+0x9d/0xb0
[   38.352699]  [<ffffffff81310b49>] ? tty_ldisc_deref+0x9/0x10
[   38.352703]  [<ffffffff8106d363>] ? pick_next_task_fair+0x63/0x140
[   38.352708]  [<ffffffff8150b8d4>] sys_sendmsg+0x44/0x80
[   38.352713]  [<ffffffff8165f8e2>] system_call_fastpath+0x16/0x1b

It stems from holding a spinlock (trace_state_lock) while attempting to register
or unregister tracepoint hooks, making in_atomic() true in this context, leading
to the warning when the tracepoint calls might_sleep() while its taking a mutex.
Since we only use the trace_state_lock to prevent trace protocol state races, as
well as hardware stat list updates on an rcu write side, we can just convert the
spinlock to a mutex to avoid this problem.

Signed-off-by: Neil Horman <nhorman@tuxdriver.com>
Reported-by: Eric Dumazet <eric.dumazet@gmail.com>
CC: David Miller <davem@davemloft.net>
Acked-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>

tcp: do_tcp_sendpages() must try to push data out on oom conditions

commit bad115cfe5b509043b684d3a007ab54b80090aa1 upstream.

Since recent changes on TCP splicing (starting with commits 2f533844
"tcp: allow splice() to build full TSO packets" and 35f9c09f "tcp:
tcp_sendpages() should call tcp_push() once"), I started seeing
massive stalls when forwarding traffic between two sockets using
splice() when pipe buffers were larger than socket buffers.

Latest changes (net: netdev_alloc_skb() use build_skb()) made the
problem even more apparent.

The reason seems to be that if do_tcp_sendpages() fails on out of memory
condition without being able to send at least one byte, tcp_push() is not
called and the buffers cannot be flushed.

After applying the attached patch, I cannot reproduce the stalls at all
and the data rate it perfectly stable and steady under any condition
which previously caused the problem to be permanent.

The issue seems to have been there since before the kernel migrated to
git, which makes me think that the stalls I occasionally experienced
with tux during stress-tests years ago were probably related to the
same issue.

This issue was first encountered on 3.0.31 and 3.2.17, so please backport
to -stable.

Signed-off-by: Willy Tarreau <w@1wt.eu>
Acked-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>