Fix the nested PR lock calling issue in ACL

commit 845b6cf34150100deb5f58c8a37a372b111f2918 upstream.

Hi,

Thanks a lot for all the review and comments so far;) I'd like to send
the improved (V4) version of this patch.

This patch fixes a deadlock in OCFS2 ACL. We found this bug in OCFS2
and Samba integration using scenario, the symptom is several smbd
processes will be hung under heavy workload. Finally we found out it
is the nested PR lock calling that leads to this deadlock:

 node1        node2
              gr PR
                |
                V
 PR(EX)---> BAST:OCFS2_LOCK_BLOCKED
                |
                V
              rq PR
                |
                V
              wait=1

After requesting the 2nd PR lock, the process "smbd" went into D
state. It can only be woken up when the 1st PR lock's RO holder equals
zero. There should be an ocfs2_inode_unlock in the calling path later
on, which can decrement the RO holder. But since it has been in
uninterruptible sleep, the unlock function has no chance to be called.

The related stack trace is:
smbd          D ffff8800013d0600     0  9522   5608 0x00000000
 ffff88002ca7fb18 0000000000000282 ffff88002f964500 ffff88002ca7fa98
 ffff8800013d0600 ffff88002ca7fae0 ffff88002f964340 ffff88002f964340
 ffff88002ca7ffd8 ffff88002ca7ffd8 ffff88002f964340 ffff88002f964340
Call Trace:
[<ffffffff80350425>] schedule_timeout+0x175/0x210
[<ffffffff8034f580>] wait_for_common+0xf0/0x210
[<ffffffffa03e12b9>] __ocfs2_cluster_lock+0x3b9/0xa90 [ocfs2]
[<ffffffffa03e7665>] ocfs2_inode_lock_full_nested+0x255/0xdb0 [ocfs2]
[<ffffffffa0446019>] ocfs2_get_acl+0x69/0x120 [ocfs2]
[<ffffffffa0446368>] ocfs2_check_acl+0x28/0x80 [ocfs2]
[<ffffffff800e3507>] acl_permission_check+0x57/0xb0
[<ffffffff800e357d>] generic_permission+0x1d/0xc0
[<ffffffffa03eecea>] ocfs2_permission+0x10a/0x1d0 [ocfs2]
[<ffffffff800e3f65>] inode_permission+0x45/0x100
[<ffffffff800d86b3>] sys_chdir+0x53/0x90
[<ffffffff80007458>] system_call_fastpath+0x16/0x1b
[<00007f34a4ef6927>] 0x7f34a4ef6927

For details, please see:
https://bugzilla.novell.com/show_bug.cgi?id=614332 and
http://oss.oracle.com/bugzilla/show_bug.cgi?id=1278

Signed-off-by: Jiaju Zhang <jjzhang@suse.de>
Acked-by: Mark Fasheh <mfasheh@suse.com>
Signed-off-by: Joel Becker <joel.becker@oracle.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

x86, apic: Fix apic=debug boot crash

commit 05e407603e527f9d808dd3866d3a17c2ce4dfcc5 upstream.

Fix a boot crash when apic=debug is used and the APIC is
not properly initialized.

This issue appears during Xen Dom0 kernel boot but the
fix is generic and the crash could occur on real hardware
as well.

Signed-off-by: Daniel Kiper <dkiper@net-space.pl>
Cc: xen-devel@lists.xensource.com
Cc: konrad.wilk@oracle.com
Cc: jeremy@goop.org
LKML-Reference: <20100819224616.GB9967@router-fw-old.local.net-space.pl>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

perf, x86: Fix Intel-nhm PMU programming errata workaround

commit 351af0725e5222e35741011d1ea62215c1ed06db upstream.

Fix the Errata AAK100/AAP53/BD53 workaround, the officialy documented
workaround we implemented in:

 11164cd: perf, x86: Add Nehelem PMU programming errata workaround

doesn't actually work fully and causes a stuck PMU state
under load and non-functioning perf profiling.

A functional workaround was found by trial & error.

Affects all Nehalem-class Intel PMUs.

Signed-off-by: Zhang Yanmin <yanmin_zhang@linux.intel.com>
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
LKML-Reference: <1281073148.2125.63.camel@ymzhang.sh.intel.com>
Cc: Arjan van de Ven <arjan@linux.intel.com>
Cc: "H. Peter Anvin" <hpa@zytor.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

x86, hotplug: Serialize CPU hotplug to avoid bringup concurrency issues

commit d7c53c9e822a4fefa13a0cae76f3190bfd0d5c11 upstream.

When testing cpu hotplug code on 32-bit we kept hitting the "CPU%d:
Stuck ??" message due to multiple cores concurrently accessing the
cpu_callin_mask, among others.

Since these codepaths are not protected from concurrent access due to
the fact that there's no sane reason for making an already complex
code unnecessarily more complex - we hit the issue only when insanely
switching cores off- and online - serialize hotplugging cores on the
sysfs level and be done with it.

[ v2.1: fix !HOTPLUG_CPU build ]

Signed-off-by: Borislav Petkov <borislav.petkov@amd.com>
LKML-Reference: <20100819181029.GC17171@aftab>
Signed-off-by: H. Peter Anvin <hpa@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

platform/x86: move rfkill for Dell Mini 1012 to compal-laptop

commit c3f755e3842108c1cffe570fe9802239810352b6 upstream.

Like others in the Mini series, the Dell Mini 1012 does not support
the smbios hook required by dell-laptop.

Signed-off-by: Victor van den Elzen <victor.vde@gmail.com>
Signed-off-by: Matthew Garrett <mjg@redhat.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

x86, apic: Map the local apic when parsing the MP table.

commit 5989cd6a1cbf86587edcc856791f960978087311 upstream.

This fixes a regression in 2.6.35 from 2.6.34, that is
present for select models of Intel cpus when people are
using an MP table.

The commit cf7500c0ea133d66f8449d86392d83f840102632
"x86, ioapic: In mpparse use mp_register_ioapic" started
calling mp_register_ioapic from MP_ioapic_info.  An extremely
simple change that was obviously correct.  Unfortunately
mp_register_ioapic did just a little more than the previous
hand crafted code and so we gained this call path.

The problem call path is:
MP_ioapic_info()
  mp_register_ioapic()
   io_apic_unique_id()
     io_apic_get_unique_id()
       get_physical_broadcast()
         modern_apic()
           lapic_get_version()
             apic_read(APIC_LVR)

Which turned out to be a problem because the local apic
was not mapped, at that point, unlike the similar point
in the ACPI parsing code.

This problem is fixed by mapping the local apic when
parsing the mptable as soon as we reasonably can.

Looking at the number of places we setup the fixmap for
the local apic, I see some serious simplification opportunities.
For the moment except for not duplicating the setting up of the
fixmap in init_apic_mappings, I have not acted on them.

The regression from 2.6.34 is tracked in bug
https://bugzilla.kernel.org/show_bug.cgi?id=16173

Reported-by: David Hill <hilld@binarystorm.net>
Reported-by: Tvrtko Ursulin <tvrtko.ursulin@sophos.com>
Tested-by: Tvrtko Ursulin <tvrtko.ursulin@sophos.com>
Signed-off-by: Eric W. Biederman <ebiederm@xmission.com>
LKML-Reference: <m1eiee86jg.fsf_-_@fess.ebiederm.org>
Signed-off-by: H. Peter Anvin <hpa@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

perf, x86: P4 PMU -- update nmi irq statistics and unmask lvt entry properly

commit 1c250d709fdc8aa5bf42d90be99428a01a256a55 upstream.

In case if last active performance counter is not overflowed at
moment of NMI being triggered by another counter, the irq
statistics may miss an update stage. As a more serious
consequence -- apic quirk may not be triggered so apic lvt entry
stay masked.

Tested-by: Lin Ming <ming.m.lin@intel.com>
Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
Cc: Stephane Eranian <eranian@google.com>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
LKML-Reference: <20100805150917.GA6311@lenovo>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

x86, asm: Refactor atomic64_386_32.S to support old binutils and be cleaner

commit 30246557a06bb20618bed906a06d1e1e0faa8bb4 upstream.

The old code didn't work on binutils 2.12 because setting a symbol to
a register apparently requires a fairly recent version.

This commit refactors the code to use the C preprocessor instead, and
in the process makes the whole code a bit easier to understand.

The object code produced is unchanged as expected.

This fixes kernel bugzilla 16506.

Reported-by: Dieter Stussy <kd6lvw+software@kd6lvw.ampr.org>
Signed-off-by: Luca Barbieri <luca@luca-barbieri.com>
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
LKML-Reference: <tip-*@git.kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

cfg80211: fix locking in action frame TX

commit fe100acddf438591ecf3582cb57241e560da70b7 upstream.

Accesses to "wdev->current_bss" must be
locked with the wdev lock, which action
frame transmission is missing.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

tracing: Fix ring_buffer_read_page reading out of page boundary

commit 18fab912d4fa70133df164d2dcf3310be0c38c34 upstream.

With the configuration: CONFIG_DEBUG_PAGEALLOC=y and Shaohua's patch:

[PATCH]x86: make spurious_fault check correct pte bit

Function call graph trace with the following will trigger a page fault.

# cd /sys/kernel/debug/tracing/
# echo function_graph > current_tracer
# cat per_cpu/cpu1/trace_pipe_raw > /dev/null

BUG: unable to handle kernel paging request at ffff880006e99000
IP: [<ffffffff81085572>] rb_event_length+0x1/0x3f
PGD 1b19063 PUD 1b1d063 PMD 3f067 PTE 6e99160
Oops: 0000 [#1] SMP DEBUG_PAGEALLOC
last sysfs file: /sys/devices/virtual/net/lo/operstate
CPU 1
Modules linked in:

Pid: 1982, comm: cat Not tainted 2.6.35-rc6-aes+ #300 /Bochs
RIP: 0010:[<ffffffff81085572>]  [<ffffffff81085572>] rb_event_length+0x1/0x3f
RSP: 0018:ffff880006475e38  EFLAGS: 00010006
RAX: 0000000000000ff0 RBX: ffff88000786c630 RCX: 000000000000001d
RDX: ffff880006e98000 RSI: 0000000000000ff0 RDI: ffff880006e99000
RBP: ffff880006475eb8 R08: 000000145d7008bd R09: 0000000000000000
R10: 0000000000008000 R11: ffffffff815d9336 R12: ffff880006d08000
R13: ffff880006e605d8 R14: 0000000000000000 R15: 0000000000000018
FS:  00007f2b83e456f0(0000) GS:ffff880002100000(0000) knlGS:0000000000000000
CS:  0010 DS: 0000 ES: 0000 CR0: 000000008005003b
CR2: ffff880006e99000 CR3: 00000000064a8000 CR4: 00000000000006e0
DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400
Process cat (pid: 1982, threadinfo ffff880006474000, task ffff880006e40770)
Stack:
 ffff880006475eb8 ffffffff8108730f 0000000000000ff0 000000145d7008bd
<0> ffff880006e98010 ffff880006d08010 0000000000000296 ffff88000786c640
<0> ffffffff81002956 0000000000000000 ffff8800071f4680 ffff8800071f4680
Call Trace:
 [<ffffffff8108730f>] ? ring_buffer_read_page+0x15a/0x24a
 [<ffffffff81002956>] ? return_to_handler+0x15/0x2f
 [<ffffffff8108a575>] tracing_buffers_read+0xb9/0x164
 [<ffffffff810debfe>] vfs_read+0xaf/0x150
 [<ffffffff81002941>] return_to_handler+0x0/0x2f
 [<ffffffff810248b0>] __bad_area_nosemaphore+0x17e/0x1a1
 [<ffffffff81002941>] return_to_handler+0x0/0x2f
 [<ffffffff810248e6>] bad_area_nosemaphore+0x13/0x15
Code: 80 25 b2 16 b3 00 fe c9 c3 55 48 89 e5 f0 80 0d a4 16 b3 00 02 c9 c3 55 31 c0 48 89 e5 48 83 3d 94 16 b3 00 01 c9 0f 94 c0 c3 55 <8a> 0f 48 89 e5 83 e1 1f b8 08 00 00 00 0f b6 d1 83 fa 1e 74 27
RIP  [<ffffffff81085572>] rb_event_length+0x1/0x3f
 RSP <ffff880006475e38>
CR2: ffff880006e99000
---[ end trace a6877bb92ccb36bb ]---

The root cause is that ring_buffer_read_page() may read out of page
boundary, because the boundary checking is done after reading. This is
fixed via doing boundary checking before reading.

Reported-by: Shaohua Li <shaohua.li@intel.com>
Signed-off-by: Huang Ying <ying.huang@intel.com>
LKML-Reference: <1280297641.2771.307.camel@yhuang-dev>
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

tracing: Fix an unallocated memory access in function_graph

commit 575570f02761bd680ba5731c1dfd4701062e7fb2 upstream.

With CONFIG_DEBUG_PAGEALLOC, I observed an unallocated memory access in
function_graph trace. It appears we find a small size entry in ring buffer,
but we access it as a big size entry. The access overflows the page size
and touches an unallocated page.

Signed-off-by: Shaohua Li <shaohua.li@intel.com>
LKML-Reference: <1280217994.32400.76.camel@sli10-desk.sh.intel.com>
[ Added a comment to explain the problem - SDR ]
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

nilfs2: fix list corruption after ifile creation failure

commit af4e36318edb848fcc0a8d5f75000ca00cdc7595 upstream.

If nilfs_attach_checkpoint() gets a memory allocation failure during
creation of ifile, it will return without removing nilfs_sb_info
struct from ns_supers list.  When a concurrently mounted snapshot is
unmounted or another new snapshot is mounted after that, this causes
kernel oops as below:

> BUG: unable to handle kernel NULL pointer dereference at (null)
> IP: [<f83662ff>] nilfs_find_sbinfo+0x74/0xa4 [nilfs2]
> *pde = 00000000
> Oops: 0000 [#1] SMP
<snip>
> Call Trace:
>  [<f835dc29>] ? nilfs_get_sb+0x165/0x532 [nilfs2]
>  [<c1173c87>] ? ida_get_new_above+0x16d/0x187
>  [<c109a7f8>] ? alloc_vfsmnt+0x7e/0x10a
>  [<c1070790>] ? kstrdup+0x2c/0x40
>  [<c1089041>] ? vfs_kern_mount+0x96/0x14e
>  [<c108913d>] ? do_kern_mount+0x32/0xbd
>  [<c109b331>] ? do_mount+0x642/0x6a1
>  [<c101a415>] ? do_page_fault+0x0/0x2d1
>  [<c1099c00>] ? copy_mount_options+0x80/0xe2
>  [<c10705d8>] ? strndup_user+0x48/0x67
>  [<c109b3f1>] ? sys_mount+0x61/0x90
>  [<c10027cc>] ? sysenter_do_call+0x12/0x22

This fixes the problem.

Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
Tested-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

wl1251: fix trigger scan timeout usage

commit fe0dbcc9d2e941328b3269dab102b94ad697ade5 upstream.

Use appropriate command (CMD_TRIGGER_SCAN_TO) instead of scan command
(CMD_SCAN) to configure trigger scan timeout.

This was broken in commit 3a98c30f3e8bb1f32b5bcb74a39647b3670de275.

This fix address the bug reported here:

https://bugzilla.kernel.org/show_bug.cgi?id=16554

Signed-off-by: Yuri Ershov <ext-yuri.ershov@nokia.com>
Signed-off-by: Yuri Kululin <ext-yuri.kululin@nokia.com>
Acked-by: Kalle Valo <kvalo@adurom.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

ocfs2/dlm: remove potential deadlock -V3

commit b11f1f1ab73fd358b1b734a9427744802202ba68 upstream.

When we need to take both dlm_domain_lock and dlm->spinlock, we should take
them in order of: dlm_domain_lock then dlm->spinlock.

There is pathes disobey this order. That is calling dlm_lockres_put() with
dlm->spinlock held in dlm_run_purge_list. dlm_lockres_put() calls dlm_put() at
the ref and dlm_put() locks on dlm_domain_lock.

Fix:
Don't grab/put the dlm when the initialising/releasing lockres.
That grab is not required because we don't call dlm_unregister_domain()
based on refcount.

Signed-off-by: Wengang Wang <wen.gang.wang@oracle.com>
Signed-off-by: Joel Becker <joel.becker@oracle.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

ocfs2/dlm: avoid incorrect bit set in refmap on recovery master

commit a524812b7eaa7783d7811198921100f079034e61 upstream.

In the following situation, there remains an incorrect bit in refmap on the
recovery master. Finally the recovery master will fail at purging the lockres
due to the incorrect bit in refmap.

1) node A has no interest on lockres A any longer, so it is purging it.
2) the owner of lockres A is node B, so node A is sending de-ref message
to node B.
3) at this time, node B crashed. node C becomes the recovery master. it recovers
lockres A(because the master is the dead node B).
4) node A migrated lockres A to node C with a refbit there.
5) node A failed to send de-ref message to node B because it crashed. The failure
is ignored. no other action is done for lockres A any more.

For mormal, re-send the deref message to it to recovery master can fix it. Well,
ignoring the failure of deref to the original master and not recovering the lockres
to recovery master has the same effect. And the later is simpler.

Signed-off-by: Wengang Wang <wen.gang.wang@oracle.com>
Acked-by: Srinivas Eeda <srinivas.eeda@oracle.com>
Signed-off-by: Joel Becker <joel.becker@oracle.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

ocfs2: Count more refcount records in file system fragmentation.

commit 8a2e70c40ff58f82dde67770e6623ca45f0cb0c8 upstream.

The refcount record calculation in ocfs2_calc_refcount_meta_credits
is too optimistic that we can always allocate contiguous clusters
and handle an already existed refcount rec as a whole. Actually
because of file system fragmentation, we may have the chance to split
a refcount record into 3 parts during the transaction. So consider
the worst case in record calculation.

Signed-off-by: Tao Ma <tao.ma@oracle.com>
Signed-off-by: Joel Becker <joel.becker@oracle.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

ocfs2 fix o2dlm dlm run purgelist (rev 3)

commit 7beaf243787f85a2ef9213ccf13ab4a243283fde upstream.

This patch fixes two problems in dlm_run_purgelist

1. If a lockres is found to be in use, dlm_run_purgelist keeps trying to purge
the same lockres instead of trying the next lockres.

2. When a lockres is found unused, dlm_run_purgelist releases lockres spinlock
before setting DLM_LOCK_RES_DROPPING_REF and calls dlm_purge_lockres.
spinlock is reacquired but in this window lockres can get reused. This leads
to BUG.

This patch modifies dlm_run_purgelist to skip lockres if it's in use and purge
 next lockres. It also sets DLM_LOCK_RES_DROPPING_REF before releasing the
lockres spinlock protecting it from getting reused.

Signed-off-by: Srinivas Eeda <srinivas.eeda@oracle.com>
Acked-by: Sunil Mushran <sunil.mushran@oracle.com>
Signed-off-by: Joel Becker <joel.becker@oracle.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

ocfs2/dlm: fix a dead lock

commit 6d98c3ccb52f692f1a60339dde7c700686a5568b upstream.

When we have to take both dlm->master_lock and lockres->spinlock,
take them in order

lockres->spinlock and then dlm->master_lock.

The patch fixes a violation of the rule.
We can simply move taking dlm->master_lock to where we have dropped res->spinlock
since when we access res->state and free mle memory we don't need master_lock's
protection.

Signed-off-by: Wengang Wang <wen.gang.wang@oracle.com>
Signed-off-by: Joel Becker <joel.becker@oracle.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

ocfs2: do not overwrite error codes in ocfs2_init_acl

commit 6eda3dd33f8a0ce58ee56a11351758643a698db4 upstream.

Setting the acl while creating a new inode depends on
the error codes of posix_acl_create_masq. This patch fix
a issue of overwriting the error codes of it.

Reported-by: Pawel Zawora <pzawora@gmail.com>
Signed-off-by: Tiger Yang <tiger.yang@oracle.com>
Signed-off-by: Joel Becker <joel.becker@oracle.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

ALSA: hda - Add quirk for Dell Vostro 1220

commit c3e68fad88143fd1fe8fe640207fb19c0f087dbc upstream.

model=dell-vostro is needed for Dell Vostro 1220 with Coexnat 5067.

Reference: Novell bnc#631066
https://bugzilla.novell.com/show_bug.cgi?id=631066

Signed-off-by: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

ALSA: hda - Fix missing stream for second ADC on Realtek ALC260 HDA codec

commit 53bacfbbb2ddd981287b58a511c8b8f5df179886 upstream.

I discovered tonight that ALSA no longer sets up a stream for the second ADC
provided by the Realtek ALC260 HDA codec.  At some point alc_build_pcms()
started using stream_analog_alt_capture when constructing the second ADC
stream, but patch_alc260() was never updated accordingly.  I have no idea
when this regression occurred.  The trivial patch to patch_alc260() given
below fixes the problem as far as I can tell.  The patch is against 2.6.35.

Signed-off-by: Jonathan Woithe <jwoithe@physics.adelaide.edu.au>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

ALSA: emu10k1 - delay the PCM interrupts (add pcm_irq_delay parameter)

commit 56385a12d9bb9e173751f74b6c430742018cafc0 upstream.

With some hardware combinations, the PCM interrupts are acknowledged
before the period boundary from the emu10k1 chip. The midlevel PCM code
gets confused and the playback stream is interrupted.

It seems that the interrupt processing shift by 2 samples is enough
to fix this issue. This default value does not harm other,
non-affected hardware.

More information: Kernel bugzilla bug#16300

[A copmile warning fixed by tiwai]

Signed-off-by: Jaroslav Kysela <perex@perex.cz>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

ALSA: riptide - Fix detection / load of firmware files

commit a5ba6beb839cfa288960c92cd2668a2601c24dda upstream.

The detection and loading of firmeware on riptide driver has been broken
due to rewrite of some codes, checking the presense wrongly.
This patch fixes the logic again.

Reference: kernel bug 16596
https://bugzilla.kernel.org/show_bug.cgi?id=16596

Signed-off-by: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

regulator: Default GPIO controlled WM8994 regulators to disabled

commit c4604e49c1a5832a58789a22eba7ca982933e1be upstream.

This ensures that if the GPIO was not enabled prior to the driver
starting the regulator API will insert the required powerup ramp
delay when it enables the regulator.  The gpiolib API does not
provide this information.

[Rewrote changelog to describe the actual change -- broonie.]

Signed-off-by: Joonyoung Shim <jy0922.shim@samsung.com>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
Signed-off-by: Liam Girdwood <lrg@slimlogic.co.uk>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

ASoC: register cache should be 1 byte aligned for 1 byte long register

commit ac770267a7cd85a747b6111db46f66d1515e7cd7 upstream.

Signed-off-by: Cliff Cai <cliff.cai@analog.com>
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
Acked-by: Liam Girdwood <lrg@slimlogic.co.uk>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

ASoC: Remove DSP mode support for WM8776

commit b2c1e07b81a126e5846dfc3d36f559d861df59f4 upstream.

This is not supported by current hardware revisions.

Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
Acked-by: Liam Girdwood <lrg@slimlogic.co.uk>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

ASoC: Fix inverted mute controls for WM8580

commit 4f0ed9a51bc8ef16c2589112fdb110479e4b0df1 upstream.

Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
Acked-by: Liam Girdwood <lrg@slimlogic.co.uk>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

acpi: fix bogus preemption logic

commit 0a7992c90828a65281c3c9cf180be3b432d277b2 upstream.

The ACPI_PREEMPTION_POINT() logic was introduced in commit 8bd108d
(ACPICA: add preemption point after each opcode parse).  The follow up
commits abe1dfab6, 138d15692, c084ca70 tried to fix the preemption logic
back and forth, but nobody noticed that the usage of
in_atomic_preempt_off() in that context is wrong.

The check which guards the call of cond_resched() is:

    if (!in_atomic_preempt_off() && !irqs_disabled())

in_atomic_preempt_off() is not intended for general use as the comment
above the macro definition clearly says:

 * Check whether we were atomic before we did preempt_disable():
 * (used by the scheduler, *after* releasing the kernel lock)

On a CONFIG_PREEMPT=n kernel the usage of in_atomic_preempt_off() works by
accident, but with CONFIG_PREEMPT=y it's just broken.

The whole purpose of the ACPI_PREEMPTION_POINT() is to reduce the latency
on a CONFIG_PREEMPT=n kernel, so make ACPI_PREEMPTION_POINT() depend on
CONFIG_PREEMPT=n and remove the in_atomic_preempt_off() check.

Addresses https://bugzilla.kernel.org/show_bug.cgi?id=16210

[akpm@linux-foundation.org: fix build]
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Cc: Len Brown <lenb@kernel.org>
Cc: Francois Valenduc <francois.valenduc@tvcablenet.be>
Cc: Lin Ming <ming.m.lin@intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

memstick: fix hangs on unexpected device removal in mspro_blk

commit d862b13bc8cbab9692fbe0ef44c40d0488b81af1 upstream.

mspro_block_remove() is called from detect thread that first calls the
mspro_block_stop(), which stops the request queue.  If we call
del_gendisk() with the queue stopped we get a deadlock.

Signed-off-by: Maxim Levitsky <maximlevitsky@gmail.com>
Cc: Alex Dubov <oakad@yahoo.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

memstick: init sysfs attributes

commit 21fd0495ea61d53e0ebe575330e343ce4e6d2a61 upstream.

Otherwise lockdep complains.

Signed-off-by: Maxim Levitsky <maximlevitsky@gmail.com>
Cc: Alex Dubov <oakad@yahoo.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

Linux 2.6.35.3

vmware: fix build error in vmware.c

This fixes a build error reported in vmware.c due to commit
9f242dc10e0c3c1eb32d8c83c18650a35fd7f80d

Reported-by: Sven Joachim <svenjoac@gmx.de>
Cc: Alok Kataria <akataria@vmware.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

mm: fix up some user-visible effects of the stack guard page

commit d7824370e26325c881b665350ce64fb0a4fde24a upstream.

This commit makes the stack guard page somewhat less visible to user
space. It does this by:

 - not showing the guard page in /proc/<pid>/maps

   It looks like lvm-tools will actually read /proc/self/maps to figure
   out where all its mappings are, and effectively do a specialized
   "mlockall()" in user space.  By not showing the guard page as part of
   the mapping (by just adding PAGE_SIZE to the start for grows-up
   pages), lvm-tools ends up not being aware of it.

 - by also teaching the _real_ mlock() functionality not to try to lock
   the guard page.

   That would just expand the mapping down to create a new guard page,
   so there really is no point in trying to lock it in place.

It would perhaps be nice to show the guard page specially in
/proc/<pid>/maps (or at least mark grow-down segments some way), but
let's not open ourselves up to more breakage by user space from programs
that depends on the exact deails of the 'maps' file.

Special thanks to Henrique de Moraes Holschuh for diving into lvm-tools
source code to see what was going on with the whole new warning.

Reported-and-tested-by: François Valenduc <francois.valenduc@tvcablenet.be
Reported-by: Henrique de Moraes Holschuh <hmh@hmh.eng.br>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

mm: fix page table unmap for stack guard page properly

commit 11ac552477e32835cb6970bf0a70c210807f5673 upstream.

We do in fact need to unmap the page table _before_ doing the whole
stack guard page logic, because if it is needed (mainly 32-bit x86 with
PAE and CONFIG_HIGHPTE, but other architectures may use it too) then it
will do a kmap_atomic/kunmap_atomic.

And those kmaps will create an atomic region that we cannot do
allocations in.  However, the whole stack expand code will need to do
anon_vma_prepare() and vma_lock_anon_vma() and they cannot do that in an
atomic region.

Now, a better model might actually be to do the anon_vma_prepare() when
_creating_ a VM_GROWSDOWN segment, and not have to worry about any of
this at page fault time.  But in the meantime, this is the
straightforward fix for the issue.

See https://bugzilla.kernel.org/show_bug.cgi?id=16588 for details.

Reported-by: Wylda <wylda@volny.cz>
Reported-by: Sedat Dilek <sedat.dilek@gmail.com>
Reported-by: Mike Pagano <mpagano@gentoo.org>
Reported-by: François Valenduc <francois.valenduc@tvcablenet.be>
Tested-by: Ed Tomlinson <edt@aei.ca>
Cc: Pekka Enberg <penberg@kernel.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

Linux 2.6.35.2

x86: don't send SIGBUS for kernel page faults

commit 96054569190bdec375fe824e48ca1f4e3b53dd36 upstream.

It's wrong for several reasons, but the most direct one is that the
fault may be for the stack accesses to set up a previous SIGBUS.  When
we have a kernel exception, the kernel exception handler does all the
fixups, not some user-level signal handler.

Even apart from the nested SIGBUS issue, it's also wrong to give out
kernel fault addresses in the signal handler info block, or to send a
SIGBUS when a system call already returns EFAULT.

Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

mm: fix missing page table unmap for stack guard page failure case

commit 5528f9132cf65d4d892bcbc5684c61e7822b21e9 upstream.

.. which didn't show up in my tests because it's a no-op on x86-64 and
most other architectures.  But we enter the function with the last-level
page table mapped, and should unmap it at exit.

Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

mm: keep a guard page below a grow-down stack segment

commit 320b2b8de12698082609ebbc1a17165727f4c893 upstream.

This is a rather minimally invasive patch to solve the problem of the
user stack growing into a memory mapped area below it.  Whenever we fill
the first page of the stack segment, expand the segment down by one
page.

Now, admittedly some odd application might _want_ the stack to grow down
into the preceding memory mapping, and so we may at some point need to
make this a process tunable (some people might also want to have more
than a single page of guarding), but let's try the minimal approach
first.

Tested with trivial application that maps a single page just below the
stack, and then starts recursing.  Without this, we will get a SIGSEGV
_after_ the stack has smashed the mapping.  With this patch, we'll get a
nice SIGBUS just as the stack touches the page just above the mapping.

Requested-by: Keith Packard <keithp@keithp.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

ext4: fix freeze deadlock under IO

commit 437f88cc031ffe7f37f3e705367f4fe1f4be8b0f upstream.

Commit 6b0310fbf087ad6 caused a regression resulting in deadlocks
when freezing a filesystem which had active IO; the vfs_check_frozen
level (SB_FREEZE_WRITE) did not let the freeze-related IO syncing
through.  Duh.

Changing the test to FREEZE_TRANS should let the normal freeze
syncing get through the fs, but still block any transactions from
starting once the fs is completely frozen.

I tested this by running fsstress in the background while periodically
snapshotting the fs and running fsck on the result.  I ran into
occasional deadlocks, but different ones.  I think this is a
fine fix for the problem at hand, and the other deadlocky things
will need more investigation.

Reported-by: Phillip Susi <psusi@cfl.rr.com>
Signed-off-by: Eric Sandeen <sandeen@redhat.com>
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

crypto: testmgr - add an option to disable cryptoalgos' self-tests

commit 0b767f96164b2b27488e3daa722ff16e89d49314 upstream.

By default, CONFIG_CRYPTO_MANAGER_TESTS will be enabled and thus
self-tests will still run, but it is now possible to disable them
to gain some time during bootup.

Signed-off-by: Alexander Shishkin <virtuoso@slind.org>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

xen: Do not suspend IPI IRQs.

commit 4877c737283813bdb4bebfa3168c1585f6e3a8ca upstream.

In general the semantics of IPIs are that they are are expected to
continue functioning after dpm_suspend_noirq().

Specifically I have seen a deadlock between the callfunc IPI and the
stop machine used by xen's do_suspend() routine. If one CPU has already
called dpm_suspend_noirq() then there is a window where it can be sent
a callfunc IPI before all the other CPUs have entered stop_cpu().

If this happens then the first CPU ends up spinning in stop_cpu()
waiting for the other to rendezvous in state STOPMACHINE_PREPARE while
the other is spinning in csd_lock_wait().

Signed-off-by: Ian Campbell <ian.campbell@citrix.com>
Cc: Jeremy Fitzhardinge <jeremy@goop.org>
Cc: xen-devel@lists.xensource.com
LKML-Reference: <1280398595-29708-4-git-send-email-ian.campbell@citrix.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

irq: Add new IRQ flag IRQF_NO_SUSPEND

commit 685fd0b4ea3f0f1d5385610b0d5b57775a8d5842 upstream.

A small number of users of IRQF_TIMER are using it for the implied no
suspend behaviour on interrupts which are not timer interrupts.

Therefore add a new IRQF_NO_SUSPEND flag, rename IRQF_TIMER to
__IRQF_TIMER and redefine IRQF_TIMER in terms of these new flags.

Signed-off-by: Ian Campbell <ian.campbell@citrix.com>
Cc: Jeremy Fitzhardinge <jeremy@goop.org>
Cc: Dmitry Torokhov <dmitry.torokhov@gmail.com>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Grant Likely <grant.likely@secretlab.ca>
Cc: xen-devel@lists.xensource.com
Cc: linux-input@vger.kernel.org
Cc: linuxppc-dev@ozlabs.org
Cc: devicetree-discuss@lists.ozlabs.org
LKML-Reference: <1280398595-29708-1-git-send-email-ian.campbell@citrix.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

net: Fix NETDEV_NOTIFY_PEERS to not conflict with NETDEV_BONDING_DESLAVE.

commit 38117d1495e587fbb10d6e55733139a27893cef5 upstream.

Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

jfs: don't allow os2 xattr namespace overlap with others

commit aca0fa34bdaba39bfddddba8ca70dba4782e8fe6 upstream.

It's currently possible to bypass xattr namespace access rules by
prefixing valid xattr names with "os2.", since the os2 namespace stores
extended attributes in a legacy format with no prefix.

This patch adds checking to deny access to any valid namespace prefix
following "os2.".

Signed-off-by: Dave Kleikamp <shaggy@linux.vnet.ibm.com>
Reported-by: Sergey Vlasov <vsu@altlinux.ru>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

l2tp: fix export of header file for userspace

commit 4565956dc0847985c0403c9ebbf274b6a122e1e2 upstream.

The header file l2tp.h should be exported to the installed include/linux/
tree for userspace programs.

This patch fixes compilation errors in L2TP userspace apps which want to
use the new L2TP support introduced in 2.6.35.

Signed-off-by: James Chapman <jchapman@katalix.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

mtd: gen_nand: fix support for multiple chips

commit 81cbb0b17796d81cbd92defe113cf2a7c7a21fbb upstream.

This patch corrects a problem where gen_nand driver assumed there can be only
one chip and ignored the pdata->chip.nr_chips value.

Signed-off-by: Marek Vasut <marek.vasut@gmail.com>
Signed-off-by: Artem Bityutskiy <Artem.Bityutskiy@nokia.com>
Signed-off-by: David Woodhouse <David.Woodhouse@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

mtd: mxc_nand: fix unbalanced enable for IRQ

commit 6a8cfcfb0de881735df4031eb5cc99be3d0971e9 upstream.

* preset should be done before sending the reset command

* without this, I get the following error on an i.MX35 :
[    0.900000] ------------[ cut here ]------------
[    0.900000] WARNING: at kernel/irq/manage.c:290 __enable_irq+0x4c/0x88()
[    0.900000] Unbalanced enable for IRQ 33
[    0.900000] Modules linked in:
[    0.900000] [<c002ffb8>] (unwind_backtrace+0x0/0xf0) from [<c02f7d0c>] (dump_stack+0x18/0x1c)
[    0.900000] [<c02f7d0c>] (dump_stack+0x18/0x1c) from [<c0049a44>] (warn_slowpath_common+0x54/0x6c)
[    0.900000] [<c0049a44>] (warn_slowpath_common+0x54/0x6c) from [<c0049b00>] (warn_slowpath_fmt+0x38/0x40)
[    0.900000] [<c0049b00>] (warn_slowpath_fmt+0x38/0x40) from [<c008f65c>] (__enable_irq+0x4c/0x88)
[    0.900000] [<c008f65c>] (__enable_irq+0x4c/0x88) from [<c008fca0>] (enable_irq+0x54/0x98)
[    0.900000] [<c008fca0>] (enable_irq+0x54/0x98) from [<c021e618>] (wait_op_done+0x40/0x134)
[    0.900000] [<c021e618>] (wait_op_done+0x40/0x134) from [<c021e808>] (send_cmd+0x30/0x38)
[    0.900000] [<c021e808>] (send_cmd+0x30/0x38) from [<c021eb8c>] (mxc_nand_command+0x26c/0x328)
[    0.900000] [<c021eb8c>] (mxc_nand_command+0x26c/0x328) from [<c021aa60>] (nand_scan_ident+0x188/0x6c0)
[    0.900000] [<c021aa60>] (nand_scan_ident+0x188/0x6c0) from [<c001a9cc>] (mxcnd_probe+0x2b8/0x3d0)
[    0.900000] [<c001a9cc>] (mxcnd_probe+0x2b8/0x3d0) from [<c01f9e88>] (platform_drv_probe+0x20/0x24)
[    0.900000] [<c01f9e88>] (platform_drv_probe+0x20/0x24) from [<c01f8c38>] (driver_probe_device+0xb0/0x164)
[    0.900000] [<c01f8c38>] (driver_probe_device+0xb0/0x164) from [<c01f8d54>] (__driver_attach+0x68/0x8c)
[    0.900000] [<c01f8d54>] (__driver_attach+0x68/0x8c) from [<c01f8348>] (bus_for_each_dev+0x50/0x84)
[    0.900000] [<c01f8348>] (bus_for_each_dev+0x50/0x84) from [<c01f8a9c>] (driver_attach+0x20/0x28)
[    0.900000] [<c01f8a9c>] (driver_attach+0x20/0x28) from [<c01f7c00>] (bus_add_driver+0x144/0x2dc)
[    0.900000] [<c01f7c00>] (bus_add_driver+0x144/0x2dc) from [<c01f906c>] (driver_register+0xb0/0x13c)
[    0.900000] [<c01f906c>] (driver_register+0xb0/0x13c) from [<c01fa13c>] (platform_driver_register+0x4c/0x60)
[    0.900000] [<c01fa13c>] (platform_driver_register+0x4c/0x60) from [<c01fa170>] (platform_driver_probe+0x20/0xa0)
[    0.900000] [<c01fa170>] (platform_driver_probe+0x20/0xa0) from [<c001a708>] (mxc_nd_init+0x18/0x24)
[    0.900000] [<c001a708>] (mxc_nd_init+0x18/0x24) from [<c002938c>] (do_one_initcall+0x64/0x1bc)
[    0.900000] [<c002938c>] (do_one_initcall+0x64/0x1bc) from [<c00084c4>] (kernel_init+0xe8/0x1ac)
[    0.900000] [<c00084c4>] (kernel_init+0xe8/0x1ac) from [<c002aee8>] (kernel_thread_exit+0x0/0x8)
[    0.900000] ---[ end trace 8bf72ac6ba089a19 ]---
[    1.140000] NAND device: Manufacturer ID: 0x2c, Chip ID: 0xda (Micron NAND 256MiB 3,3V 8-bit)

Signed-off-by: Eric Bénard <eric@eukrea.com>
Signed-off-by: Artem Bityutskiy <Artem.Bityutskiy@nokia.com>
Signed-off-by: David Woodhouse <David.Woodhouse@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

sched: Revert nohz_ratelimit() for now

commit 396e894d289d69bacf5acd983c97cd6e21a14c08 upstream.

Norbert reported that nohz_ratelimit() causes his laptop to burn about
4W (40%) extra. For now back out the change and see if we can adjust
the power management code to make better decisions.

Reported-by: Norbert Preining <preining@logic.at>
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Acked-by: Mike Galbraith <efault@gmx.de>
Cc: Arjan van de Ven <arjan@infradead.org>
LKML-Reference: <new-submission>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

Bluetooth: Added support for controller shipped with iMac i5

commit 0a79f67445de50ca0a8dc1d34f3cc406d89c28b2 upstream.

Device class is ff(vend.) instead of e0(wlcon).

Output from command `usb-devices`:
T:  Bus=01 Lev=03 Prnt=03 Port=00 Cnt=01 Dev#=  6 Spd=12  MxCh= 0
D:  Ver= 2.00 Cls=ff(vend.) Sub=01 Prot=01 MxPS=64 #Cfgs=  1
P:  Vendor=05ac ProdID=8215 Rev=01.82
S:  Manufacturer=Apple Inc.
S:  Product=Bluetooth USB Host Controller
S:  SerialNumber=7C6D62936607
C:  #Ifs= 4 Cfg#= 1 Atr=e0 MxPwr=0mA
I:  If#= 0 Alt= 0 #EPs= 3 Cls=ff(vend.) Sub=01 Prot=01 Driver=btusb
I:  If#= 1 Alt= 0 #EPs= 2 Cls=e0(wlcon) Sub=01 Prot=01 Driver=btusb
I:  If#= 2 Alt= 0 #EPs= 2 Cls=ff(vend.) Sub=ff Prot=ff Driver=(none)
I:  If#= 3 Alt= 0 #EPs= 0 Cls=fe(app. ) Sub=01 Prot=00 Driver=(none)

Signed-off-by: Cyril Lacoux <clacoux@ifeelgood.org>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

USB: serial: fix stalled writes

commit b58af4066d240b18b43f202e07b9ec7461d90b17 upstream.

As David VomLehn points out, it was possible to receive an interrupt
before clearing the free-urb flag which could lead to the urb being
incorrectly marked as busy.

For the same reason, move tx_bytes accounting so that it will never be
negative.

Note that the free-flags set and clear operations do not need any
additional locking as they are manipulated while USB_SERIAL_WRITE_BUSY
is set.

Reported-by: David VomLehn <dvomlehn@cisco.com>
Tested-by: David VomLehn <dvomlehn@cisco.com>
Signed-off-by: Johan Hovold <jhovold@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

USB: fix thread-unsafe anchor utiliy routines

commit b3e670443b7fb8a2d29831b62b44a039c283e351 upstream.

This patch fixes a race condition in two utility routines
related to the removal/unlinking of urbs from an anchor.

If two threads are concurrently accessing the same anchor,
both could end up with the same urb - thinking they are
the exclusive owner.

Alan Stern pointed out a related issue in
usb_unlink_anchored_urbs:

"The URB isn't removed from the anchor until it completes
 (as a by-product of completion, in fact), which might not
 be for quite some time after the unlink call returns.
 In the meantime, the subroutine will keep trying to unlink
 it, over and over again."

Cc: Oliver Neukum <oneukum@suse.de>
Cc: Greg Kroah-Hartman <greg@kroah.com>
Acked-by: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Christian Lamparter <chunkeey@googlemail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

USB: usbtest: avoid to free coherent buffer in atomic context

commit e10e1bec8e6654de4591ef45ddd6a6d1e5b2591c upstream.

This patch fixes the warning below:
[30753.755998] ------------[ cut here ]------------
[30753.755998] WARNING: at /home/tom/git/linux-2.6/linux-2.6-next/arch/x86/include/asm/dma-mapping.h:155 hcd_buffer_free+0xb1/0xd4 [usbcore]()
[30753.755998] Hardware name: 6475EK2
[30753.755998] Modules linked in: uvcvideo ehci_hcd usbtest cdc_ether usbnet vfat fat usb_storage nfsd lockd nfs_acl auth_rpcgss exportfs mii tun videodev v4l1_compat v4l2_compat_ioctl32 fuse bridge stp llc sunrpc ipv6 cpufreq_ondemand acpi_cpufreq freq_table mperf kvm_intel kvm arc4 ecb ath5k usbhid mac80211 snd_hda_codec_conexant ch341 usbserial ath cfg80211 thinkpad_acpi snd_hda_intel pcspkr wmi hwmon yenta_socket iTCO_wdt iTCO_vendor_support i2c_i801 e1000e snd_hda_codec snd_hwdep snd_pcm snd_timer snd soundcore snd_page_alloc pata_acpi uhci_hcd ohci_hcd usbcore i915 drm_kms_helper drm i2c_algo_bit i2c_core video output [last unloaded: uvcvideo]
[30753.755998] Pid: 0, comm: swapper Tainted: G        W   2.6.35-rc6-gkh-wl+ #49
[30753.755998] Call Trace:
[30753.755998]  <IRQ>  [<ffffffff8104478a>] warn_slowpath_common+0x80/0x98
[30753.755998]  [<ffffffff810447b7>] warn_slowpath_null+0x15/0x17
[30753.755998]  [<ffffffffa00ce02d>] hcd_buffer_free+0xb1/0xd4 [usbcore]
[30753.755998]  [<ffffffffa00c1345>] usb_free_coherent+0x1c/0x1e [usbcore]
[30753.755998]  [<ffffffffa00b13e4>] simple_free_urb+0x23/0x2f [usbtest]
[30753.755998]  [<ffffffffa00b210b>] iso_callback+0xbb/0x10f [usbtest]
[30753.755998]  [<ffffffffa00c7390>] usb_hcd_giveback_urb+0x8c/0xc0 [usbcore]
[30753.755998]  [<ffffffffa0449b35>] ehci_urb_done+0x84/0x95 [ehci_hcd]
[30753.755998]  [<ffffffffa044b5a5>] ehci_work+0x41a/0x7dd [ehci_hcd]
[30753.755998]  [<ffffffffa044e298>] ehci_irq+0x33b/0x370 [ehci_hcd]
[30753.755998]  [<ffffffff8100fb05>] ? sched_clock+0x9/0xd
[30753.755998]  [<ffffffff8105e641>] ? sched_clock_local+0x1c/0x82
[30753.755998]  [<ffffffff8105e76a>] ? sched_clock_cpu+0xc3/0xce
[30753.755998]  [<ffffffff81067c7e>] ? trace_hardirqs_off+0xd/0xf
[30753.755998]  [<ffffffff8105e7b8>] ? cpu_clock+0x43/0x5e
[30753.755998]  [<ffffffffa00c6999>] usb_hcd_irq+0x45/0xa1 [usbcore]
[30753.755998]  [<ffffffff81092e02>] handle_IRQ_event+0x20/0xa5
[30753.755998]  [<ffffffff81094cea>] handle_fasteoi_irq+0x92/0xd2
[30753.755998]  [<ffffffff8100c0ed>] handle_irq+0x1f/0x2a
[30753.755998]  [<ffffffff8100b75d>] do_IRQ+0x57/0xbe
[30753.755998]  [<ffffffff8136a693>] ret_from_intr+0x0/0x16
[30753.755998]  <EOI>  [<ffffffff81223baa>] ? acpi_idle_enter_bm+0x231/0x269
[30753.755998]  [<ffffffff81223ba3>] ? acpi_idle_enter_bm+0x22a/0x269
[30753.755998]  [<ffffffff812c4b6b>] cpuidle_idle_call+0x99/0xce
[30753.755998]  [<ffffffff81008dd5>] cpu_idle+0x61/0xaa
[30753.755998]  [<ffffffff8136374b>] start_secondary+0x1c2/0x1c6
[30753.755998] ---[ end trace 904cfaf7ab4cb1a2 ]---

Signed-off-by: Ming Lei <tom.leiming@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

USB: cp210x: Add four new device IDs

commit 356c5a4834a74c621715f7a7f16ded914eecbd3c upstream.

Signed-off-by: Alessio Igor Bogani <abogani@texware.it>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

USB: ftdi_sio: device id for Navitator

commit b6180ef7c99574c3350bbffa2a3a9d675321543d upstream.

This patch is to add a US Interface, Inc. "Navigator" USB device.
Specifically, it's a HAM Radio USB sound modem that also
incorporates three pairs of unique FTDI serial ports.  The standard
Linux FTDI serial driver will only recognize the first two serial
ports of an unknown FDTI derived device and this patch adds in
recognition to these specific new IDs.

Signed-off-by: David A. Ranch <dranch@trinnet.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

USB: option: add huawei k3765 k4505 devices to work properly

commit 0372a754be9aa43e19fd86c9bc04796d43b55e38 upstream.

This patch adds the product IDs of Huawei's K3765 and K4505 mobile
broadband usb modems to option.c. It also adds a quirk to the option
probe function so that binding to the device's network interface(class
0xff) is avoided. This is necessary to allow another driver to bind to
that, and to avoid programs like wvdial opening a nonfunctioning tty
during modem discovery.

Signed-off-by: Andrew Bird <ajb@spheresystems.co.uk>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

USB: option: Huawei ETS 1220 support added

commit b972302b0a13aaddc9e90da2b4b52722e5d0e776 upstream.

The patch adds Huawei ETS 1220 product id into the list of supported
devices in 'option' usb serial driver.

Signed-off-by: Pavel Kazlou <p.i.kazlou@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

USB: serial: enabling support for Segway RMP in ftdi_sio

commit afad19648f70c6493193e0a774bd754b7790b4a0 upstream.

I have added the ProductID=0xe729 VendorID=FTDI_VID=0x0403 which will
enable support for the Segway Robotic Mobility Platform (RMP200) in the
ftdi_sio kernel module.  Currently, users of the Segway RMP200 must use
a RUN+="/sbin/modprobe -q ftdi-sio product=0xe729 vendor=0x0403 in a
udev rule to get the ftdi_sio module to handle the usb interface and
mount it on /dev/ttyXXX.  This is not a good solution because some users
will have multiple USB to Serial converters which will use the ftdi_sio
module.

Signed-off-by: John Rogers <jgrogers@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

USB delay init quirk for logitech Harmony 700-series devices

commit 93362a875fc69881ae69299efaf19a55a1f57db0 upstream.

The Logitech Harmony 700 series needs an extra delay during
initialization.  This patch adds a USB quirk which enables such a delay
and adds the device to the quirks list.

Signed-off-by: Phil Dibowitz <phil@ipom.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

USB: resizing usbmon binary interface buffer causes protection faults

commit 33d973ad88ceb83ed1449592b7574b5b5bb33ac6 upstream.

Enlarging the buffer size via the MON_IOCT_RING_SIZE ioctl causes
general protection faults. It appears the culprit is an incorrect
argument to mon_free_buff: instead of passing the size of the current
buffer being freed, the size of the new buffer is passed.

Use the correct size argument to mon_free_buff when changing the size of
the buffer.

Signed-off-by: Steven Robertson <steven@strobe.cc>
Acked-by: Pete Zaitcev <zaitcev@redhat.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

USB: EHCI: remove PCI assumption

commit ae68a83bdc1971cb02fefc7a686ba6d077065e71 upstream.

This patch (as1405) fixes a small bug in ehci-hcd's isochronous
scheduler.  Not all EHCI controllers are PCI, and the code shouldn't
assume that they are.  Instead, introduce a special flag for
controllers which need to delay iso scheduling for full-speed devices
beyond the scheduling threshold.

Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
CC: Sarah Sharp <sarah.a.sharp@linux.intel.com>
CC: David Brownell <david-b@pacbell.net>
Acked-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

USB: musb: use correct register widths in register dumps

commit 0936fb5e92a90476959447ad8ae5d780afbbd930 upstream.

DMA_ADDR and DMA_COUNT are 32-bit registers, not 16-bit.

Marking them as 16-bit in the table causes only the lower
16-bits to be dumped and this is misleading.

Signed-off-by: Anand Gadiyar <gadiyar@ti.com>
Acked-by: Felipe Balbi <felipe.balbi@nokia.com>
Signed-off-by: Ajay Kumar Gupta <ajay.gupta@ti.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

ARM: 6280/1: imx: Fix build failure when including <mach/gpio.h> without <linux/spinlock.h>

commit 868003ca7ad17ac6c1606dc36101f10a7825b399 upstream.

This is a follow up to

14cb0de (arm/imx/gpio: add spinlock protection)

and fixes the following build failure:

  CC      arch/arm/mach-imx/pcm970-baseboard.o
In file included from arch/arm/include/asm/gpio.h:6,
 from include/linux/gpio.h:8,
 from arch/arm/mach-imx/pcm970-baseboard.c:20:
arch/arm/plat-mxc/include/mach/gpio.h:40: error: expected specifier-qualifier-list before 'spinlock_t'

Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>

ARM: 6299/1: errata: TLBIASIDIS and TLBIMVAIS operations can broadcast a faulty ASID

commit cdf357f1e13a08a11261edacb3083746f65c1ed9 upstream.

On versions of the Cortex-A9 prior to r2p0, performing TLB invalidations by
ASID match can result in the incorrect ASID being broadcast to other CPUs.
As a consequence of this, the targetted TLB entries are not invalidated
across the system.

This workaround changes the TLB flushing routines to invalidate entries
regardless of the ASID.

Tested-by: Rob Clark <rob@ti.com>
Acked-by: Catalin Marinas <catalin.marinas@arm.com>
Signed-off-by: Will Deacon <will.deacon@arm.com>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

smsc911x: Add spinlocks around registers access

commit 492c5d943d6a04b124ba3a719dc746dc36b14cfb upstream.

On SMP systems, the SMSC911x registers may be accessed by multiple CPUs
and this seems to put the chip in an inconsistent state. The patch adds
spinlocks to the smsc911x_reg_read, smsc911x_reg_write,
smsc911x_rx_readfifo and smsc911x_tx_writefifo functions.

Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

pxa/cm-x300: fix ffuart registration

commit a6cd7eb374647b572ae9e7dbfe49871e6996e8e0 upstream.

ffuart is available on cm-x300 only with pxa300.

Signed-off-by: Igor Grinberg <grinberg@compulab.co.il>
Signed-off-by: Eric Miao <eric.y.miao@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

signalfd: fill in ssi_int for posix timers and message queues

commit a2a20c412c86e0bb46a9ab0dd31bcfe6d201b913 upstream.

If signalfd is used to consume a signal generated by a POSIX interval
timer or POSIX message queue, the ssi_int field does not reflect the data
(sigevent->sigev_value) supplied to timer_create(2) or mq_notify(3).  (The
ssi_ptr field, however, is filled in.)

This behavior differs from signalfd's treatment of sigqueue-generated
signals -- see the default case in signalfd_copyinfo.  It also gives
results that differ from the case when a signal is handled conventionally
via a sigaction-registered handler.

So, set signalfd_siginfo->ssi_int in the remaining cases (__SI_TIMER,
__SI_MESGQ) where ssi_ptr is set.

akpm: a non-back-compatible change.  Merge into -stable to minimise the
number of kernels which are in the field and which miss this feature.

Signed-off-by: Nathan Lynch <ntl@pobox.com>
Acked-by: Davide Libenzi <davidel@xmailserver.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

bio, fs: update RWA_MASK, READA and SWRITE to match the corresponding BIO_RW_* bits

commit aca27ba9618276dd2f777bcd5a1419589ccf1ca8 upstream.

Commit a82afdf (block: use the same failfast bits for bio and request)
moved BIO_RW_* bits around such that they match up with REQ_* bits.
Unfortunately, fs.h hard coded RW_MASK, RWA_MASK, READ, WRITE, READA
and SWRITE as 0, 1, 2 and 3, and expected them to match with BIO_RW_*
bits.  READ/WRITE didn't change but BIO_RW_AHEAD was moved to bit 4
instead of bit 1, breaking RWA_MASK, READA and SWRITE.

This patch updates RWA_MASK, READA and SWRITE such that they match the
BIO_RW_* bits again.  A follow up patch will update the definitions to
directly use BIO_RW_* bits so that this kind of breakage won't happen
again.

Neil also spotted missing RWA_MASK conversion.

Stable: The offending commit a82afdf was released with v2.6.32, so
this patch should be applied to all kernels since then but it must
_NOT_ be applied to kernels earlier than that.

Signed-off-by: Tejun Heo <tj@kernel.org>
Reported-and-bisected-by: Vladislav Bolkhovitin <vst@vlnb.net>
Root-caused-by: Neil Brown <neilb@suse.de>
Signed-off-by: Jens Axboe <jaxboe@fusionio.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

drbd: Disable delay probes for the upcomming release

commit 6710a5760355be8f2e51682f41b0d3fc76550309 upstream.

Signed-off-by: Philipp Reisner <philipp.reisner@linbit.com>
Signed-off-by: Lars Ellenberg <lars.ellenberg@linbit.com>
Signed-off-by: Jens Axboe <jaxboe@fusionio.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

drbd: Initialize all members of sync_conf to their defaults [Bugz 315]

commit 85f4cc17a62c3ac9edeaf120cdae7261df458053 upstream.

Signed-off-by: Philipp Reisner <philipp.reisner@linbit.com>
Signed-off-by: Lars Ellenberg <lars.ellenberg@linbit.com>
Signed-off-by: Jens Axboe <jaxboe@fusionio.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

fs/ecryptfs/file.c: introduce missing free

commit ceeab92971e8af05c1e81a4ff2c271124b55bb9b upstream.

The comments in the code indicate that file_info should be released if the
function fails.  This releasing is done at the label out_free, not out.

The semantic match that finds this problem is as follows:
(http://www.emn.fr/x-info/coccinelle/)

// <smpl>
@r exists@
local idexpression x;
statement S;
expression E;
identifier f,f1,l;
position p1,p2;
expression *ptr != NULL;
@@

x@p1 = kmem_cache_zalloc(...);
...
if (x == NULL) S
<... when != x
     when != if (...) { <+...x...+> }
(
x->f1 = E
|
 (x->f1 == NULL || ...)
|
 f(...,x->f1,...)
)
...>
(
 return <+...x...+>;
|
 return@p2 ...;
)

@script:python@
p1 << r.p1;
p2 << r.p2;
@@

print "* file: %s kmem_cache_zalloc %s" % (p1[0].file,p1[0].line)
// </smpl>

Signed-off-by: Julia Lawall <julia@diku.dk>
Signed-off-by: Tyler Hicks <tyhicks@linux.vnet.ibm.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

ecryptfs: release reference to lower mount if interpose fails

commit 31f73bee3e170b7cabb35db9e2f4bf7919b9d036 upstream.

In ecryptfs_lookup_and_interpose_lower() the lower mount is not decremented
if allocation of a dentry info struct failed. As a result the lower filesystem
cant be unmounted any more (since it is considered busy). This patch corrects
the reference counting.

Signed-off-by: Lino Sanfilippo <LinoSanfilippo@gmx.de>
Signed-off-by: Tyler Hicks <tyhicks@linux.vnet.ibm.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

eCryptfs: Handle ioctl calls with unlocked and compat functions

commit c43f7b8fb03be8bcc579bfc4e6ab70eac887ab55 upstream.

Lower filesystems that only implemented unlocked_ioctl weren't being
passed ioctl calls because eCryptfs only checked for
lower_file->f_op->ioctl and returned -ENOTTY if it was NULL.

eCryptfs shouldn't implement ioctl(), since it doesn't require the BKL.
This patch introduces ecryptfs_unlocked_ioctl() and
ecryptfs_compat_ioctl(), which passes the calls on to the lower file
system.

https://bugs.launchpad.net/ecryptfs/+bug/469664

Reported-by: James Dupin <james.dupin@gmail.com>
Signed-off-by: Tyler Hicks <tyhicks@linux.vnet.ibm.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

blkdev: cgroup whitelist permission fix

commit b7300b78d1a87625975a799a109a2f98d77757c8 upstream.

The cgroup device whitelist code gets confused when trying to grant
permission to a disk partition that is not currently open.  Part of
blkdev_open() includes __blkdev_get() on the whole disk.

Basically, the only ways to reliably allow a cgroup access to a partition
on a block device when using the whitelist are to 1) also give it access
to the whole block device or 2) make sure the partition is already open in
a different context.

The patch avoids the cgroup check for the whole disk case when opening a
partition.

Addresses https://bugzilla.redhat.com/show_bug.cgi?id=589662

Signed-off-by: Chris Wright <chrisw@sous-sol.org>
Acked-by: Serge E. Hallyn <serue@us.ibm.com>
Tested-by: Serge E. Hallyn <serue@us.ibm.com>
Reported-by: Vivek Goyal <vgoyal@redhat.com>
Cc: Al Viro <viro@ZenIV.linux.org.uk>
Cc: Christoph Hellwig <hch@infradead.org>
Cc: "Daniel P. Berrange" <berrange@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

Fix sget() race with failing mount

commit 7a4dec53897ecd3367efb1e12fe8a4edc47dc0e9 upstream.

If sget() finds a matching superblock being set up, it'll
grab an active reference to it and grab s_umount.  That's
fine - we'll wait for completion of foofs_get_sb() that way.
However, if said foofs_get_sb() fails we'll end up holding
the halfway-created superblock.  deactivate_locked_super()
called by foofs_get_sb() will just unlock the sucker since
we are holding another active reference to it.

What we need is a way to tell if superblock has been successfully
set up.  Unfortunately, neither ->s_root nor the check for
MS_ACTIVE quite fit.  Cheap and easy way, suitable for backport:
new flag set by the (only) caller of ->get_sb().  If that flag
isn't present by the time sget() grabbed s_umount on preexisting
superblock it has found, it's seeing a stillborn and should
just bury it with deactivate_locked_super() (and repeat the search).

Longer term we want to set that flag in ->get_sb() instances (and
check for it to distinguish between "sget() found us a live sb"
and "sget() has allocated an sb, we need to set it up" in there,
instead of checking ->s_root as we do now).

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

ioat2: catch and recover from broken vtd configurations v6

commit 556ab45f9a775bfa4762bacc0a4afb5b44b067bc upstream.

On some platforms (MacPro3,1) the BIOS assigns the ioatdma device to the
incorrect iommu causing faults when the driver initializes.  Add a quirk
to catch this misconfiguration and try falling back to untranslated
operation (which works in the MacPro3,1 case).

Assuming there are other platforms with misconfigured iommus teach the
ioatdma driver to treat initialization failures as non-fatal (just fail
the driver load and emit a warning instead of triggering a BUG_ON).

This can be classified as a boot regression since 2.6.32 on affected
platforms since the ioatdma module did not autoload prior to that
kernel.

Acked-by: David Woodhouse <David.Woodhouse@intel.com>
Reported-by: Chris Li <lkml@chrisli.org>
Tested-by: Chris Li <lkml@chrisli.org>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

e100/e1000*/igb*/ixgb*: Add missing read memory barrier

commit 2d0bb1c1f4524befe9f0fcf0d0cd3081a451223f upstream.

Based on patches from Sonny Rao and Milton Miller...

Combined the patches to fix up clean_tx_irq and clean_rx_irq.

The PowerPC architecture does not require loads to independent bytes
to be ordered without adding an explicit barrier.

In ixgbe_clean_rx_irq we load the status bit then load the packet data.
With packet split disabled if these loads go out of order we get a
stale packet, but we will notice the bad sequence numbers and drop it.

The problem occurs with packet split enabled where the TCP/IP header
and data are in different descriptors. If the reads go out of order
we may have data that doesn't match the TCP/IP header. Since we use
hardware checksumming this bad data is never verified and it makes it
all the way to the application.

This bug was found during stress testing and adding this barrier has
been shown to fix it.  The bug can manifest as a data integrity issue
(bad payload data) or as a BUG in skb_pull().

This was a nasty bug to hunt down, if people agree with the fix I think
it's a candidate for stable.

Previously Submitted to e1000-devel only for ixgbe

http://marc.info/?l=e1000-devel&m=126593062701537&w=3

We've now seen this problem hit with other device drivers (e1000e mostly)
So I'm resubmitting with fixes for other Intel Device Drivers with
similar issues.

CC: Milton Miller <miltonm@bga.com>
CC: Anton Blanchard <anton@samba.org>
CC: Sonny Rao <sonnyrao@us.ibm.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

md/raid10: fix deadlock with unaligned read during resync

commit 51e9ac77035a3dfcb6fc0a88a0d80b6f99b5edb1 upstream.

If the 'bio_split' path in raid10-read is used while
resync/recovery is happening it is possible to deadlock.
Fix this be elevating ->nr_waiting for the duration of both
parts of the split request.

This fixes a bug that has been present since 2.6.22
but has only started manifesting recently for unknown reasons.
It is suitable for and -stable since then.

Reported-by: Justin Bronder <jsbronder@gentoo.org>
Tested-by: Justin Bronder <jsbronder@gentoo.org>
Signed-off-by: NeilBrown <neilb@suse.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

md: fix another deadlock with removing sysfs attributes.

commit bb4f1e9d0e2ef93de8e36ca0f5f26625fcd70b7d upstream.

Move the deletion of sysfs attributes from reconfig_mutex to
open_mutex didn't really help as a process can try to take
open_mutex while holding reconfig_mutex, so the same deadlock can
happen, just requiring one more process to be involved in the chain.

I looks like I cannot easily use locking to wait for the sysfs
deletion to complete, so don't.

The only things that we cannot do while the deletions are still
pending is other things which can change the sysfs namespace: run,
takeover, stop.  Each of these can fail with -EBUSY.
So set a flag while doing a sysfs deletion, and fail run, takeover,
stop if that flag is set.

This is suitable for 2.6.35.x

Signed-off-by: NeilBrown <neilb@suse.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

md: move revalidate_disk() back outside open_mutex

commit 147e0b6a639ac581ca3bf627bedc3f4a6d3eca66 upstream.

Commit b821eaa5 "md: remove ->changed and related code" moved
revalidate_disk() under open_mutex, and lockdep noticed.

[ INFO: possible circular locking dependency detected ]
2.6.32-mdadm-locking #1
-------------------------------------------------------
mdadm/3640 is trying to acquire lock:
 (&bdev->bd_mutex){+.+.+.}, at: [<ffffffff811acecb>] revalidate_disk+0x5b/0x90

but task is already holding lock:
 (&mddev->open_mutex){+.+...}, at: [<ffffffffa055e07a>] do_md_stop+0x4a/0x4d0 [md_mod]

which lock already depends on the new lock.

It is suitable for 2.6.35.x

Reported-by: Przemyslaw Czarnowski <przemyslaw.hawrylewicz.czarnowski@intel.com>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Signed-off-by: NeilBrown <neilb@suse.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

ide-cd: Do not access completed requests in the irq handler

commit 110712828365ccafcc61a7f4db44c31ed4cf8793 upstream.

ide_cd_error_cmd() can complete an erroneous request with leftover
buffers. Signal this with its return value so that the request is not
accessed after its completion in the irq handler and we oops.

Signed-off-by: Borislav Petkov <bp@alien8.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

ipmi: fix ACPI detection with regspacing

commit d9e1b6c45059ccfff647a120769ae86da78dfdc4 upstream.

After the commit that changed ipmi_si detecting sequence from SMBIOS/ACPI
to ACPI/SMBIOS,

| commit 754d453185275951d39792865927ec494fa1ebd8
| Author: Matthew Garrett <mjg@redhat.com>
| Date:   Wed May 26 14:43:47 2010 -0700
|
|    ipmi: change device discovery order
|
|    The ipmi spec provides an ordering for si discovery.  Change the driver to
|    match, with the exception of preferring smbios to SPMI as HPs (at least)
|    contain accurate information in the former but not the latter.

ipmi_si can not be initialized.

[  138.799739] calling  init_ipmi_devintf+0x0/0x109 @ 1
[  138.805050] ipmi device interface
[  138.818131] initcall init_ipmi_devintf+0x0/0x109 returned 0 after 12797 usecs
[  138.822998] calling  init_ipmi_si+0x0/0xa90 @ 1
[  138.840276] IPMI System Interface driver.
[  138.846137] ipmi_si: probing via ACPI
[  138.849225] ipmi_si 00:09: [io  0x0ca2] regsize 1 spacing 1 irq 0
[  138.864438] ipmi_si: Adding ACPI-specified kcs state machine
[  138.870893] ipmi_si: probing via SMBIOS
[  138.880945] ipmi_si: Adding SMBIOS-specified kcs state machineipmi_si: duplicate interface
[  138.896511] ipmi_si: probing via SPMI
[  138.899861] ipmi_si: Adding SPMI-specified kcs state machineipmi_si: duplicate interface
[  138.917095] ipmi_si: Trying ACPI-specified kcs state machine at i/o address 0xca2, slave address 0x0, irq 0
[  138.928658] ipmi_si: Interface detection failed
[  138.953411] initcall init_ipmi_si+0x0/0xa90 returned 0 after 110847 usecs

in smbios has
DMI/SMBIOS
Handle 0x00C5, DMI type 38, 18 bytes
IPMI Device Information
        Interface Type: KCS (Keyboard Control Style)
        Specification Version: 2.0
        I2C Slave Address: 0x00
        NV Storage Device: Not Present
        Base Address: 0x0000000000000CA2 (I/O)
        Register Spacing: 32-bit Boundaries
in DSDT has
                    Device (BMC)
                    {

                        Name (_HID, EisaId ("IPI0001"))
                        Method (_STA, 0, NotSerialized)
                        {
                            If (LEqual (OSN, Zero))
                            {
                                Return (Zero)
                            }

                            Return (0x0F)
                        }

                        Name (_STR, Unicode ("IPMI_KCS"))
                        Name (_UID, Zero)
                        Name (_CRS, ResourceTemplate ()
                        {
                            IO (Decode16,
                                0x0CA2,             // Range Minimum
                                0x0CA2,             // Range Maximum
                                0x00,               // Alignment
                                0x01,               // Length
                                )
                            IO (Decode16,
                                0x0CA6,             // Range Minimum
                                0x0CA6,             // Range Maximum
                                0x00,               // Alignment
                                0x01,               // Length
                                )
                        })
                        Method (_IFT, 0, NotSerialized)
                        {
                            Return (One)
                        }

                        Method (_SRV, 0, NotSerialized)
                        {
                            Return (0x0200)
                        }
                    }

so the reg spacing should be 4 instead of 1.

Try to calculate regspacing for this kind of system.

Observed on a Sun Fire X4800.  Other OSes work and pass certification.

Signed-off-by: Yinghai Lu <yinghai@kernel.org>
Cc: Bjorn Helgaas <bjorn.helgaas@hp.com>
Acked-by: Matthew Garrett <mjg@redhat.com>
Cc: Len Brown <len.brown@intel.com>
Cc: Myron Stowe <myron.stowe@hp.com>
Cc: Corey Minyard <minyard@acm.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

rtl8180: avoid potential NULL deref in rtl8180_beacon_work

commit 8f1d2d2be73a98c21e68fe2a26f633892d4abdd1 upstream.

ieee80211_beacon_get can return NULL...

Signed-off-by: John W. Linville <linville@tuxdriver.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

iwlwifi: fix TX tracer

commit e95b743536937a72e1560c85696b425c5d1a1c18 upstream.

The TX tracing code copies with the wrong length,
which will typically copy too little data. Fix
this by using the correct length variable.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

drivers/video/w100fb.c: ignore void return value / fix build failure

commit fa260c00c1aa5c657793a7221e40d2400df5afd8 upstream.

Fix a build failure "error: void value not ignored as it ought to be"
by removing an assignment of a void return value.  The functionality of
the code is not changed.

Signed-off-by: Peter Huewe <peterhuewe@gmx.de>
Acked-by: Henrik Kretzschmar <henne@nachtwindheim.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

mm: fix corruption of hibernation caused by reusing swap during image saving

commit 966cca029f739716fbcc8068b8c6dfe381f86fc3 upstream.

Since 2.6.31, swap_map[]'s refcounting was changed to show that a used
swap entry is just for swap-cache, can be reused.  Then, while scanning
free entry in swap_map[], a swap entry may be able to be reclaimed and
reused.  It was caused by commit c9e444103b5e7a5 ("mm: reuse unused swap
entry if necessary").

But this caused deta corruption at resume. The scenario is

- Assume a clean-swap cache, but mapped.

- at hibernation_snapshot[], clean-swap-cache is saved as
  clean-swap-cache and swap_map[] is marked as SWAP_HAS_CACHE.

- then, save_image() is called.  And reuse SWAP_HAS_CACHE entry to save
  image, and break the contents.

After resume:

- the memory reclaim runs and finds clean-not-referenced-swap-cache and
  discards it because it's marked as clean.  But here, the contents on
  disk and swap-cache is inconsistent.

Hance memory is corrupted.

This patch avoids the bug by not reclaiming swap-entry during hibernation.
This is a quick fix for backporting.

Signed-off-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Rafael J. Wysocki <rjw@sisk.pl>
Reported-by: Ondreg Zary <linux@rainbow-software.org>
Tested-by: Ondreg Zary <linux@rainbow-software.org>
Tested-by: Andrea Gelmini <andrea.gelmini@gmail.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: Greg Kroah-Hartman <gregkh@suse.de>

Char: nozomi, set tty->driver_data appropriately

commit bf9c1fca9ae9a79ed209e7ab2c10b3862f3f6f72 upstream.

Sorry, one more fix, this one depends on the other, so this is rather 2/2.
--

tty->driver_data is used all over the code, but never set. This
results in oopses like:
BUG: unable to handle kernel NULL pointer dereference at 0000000000000130
IP: [<ffffffff814a0040>] mutex_lock+0x10/0x40
...
Pid: 2157, comm: modem-manager Not tainted 2.6.34.1-0.1-desktop #1 2768DR7/2768DR7
RIP: 0010:[<ffffffff814a0040>]  [<ffffffff814a0040>] mutex_lock+0x10/0x40
RSP: 0018:ffff88007b16fa50  EFLAGS: 00010286
RAX: 0000000000000000 RBX: 0000000000000130 RCX: 0000000000000003
RDX: 0000000000000003 RSI: 0000000000000286 RDI: 0000000000000130
RBP: 0000000000001000 R08: 0000000000000000 R09: 0000000000000000
R10: 0000000000000001 R11: 0000000000000000 R12: 0000000000000130
R13: 0000000000000001 R14: 0000000000000000 R15: ffff88007b16feb4
...
Call Trace:
 [<ffffffffa077690d>] ntty_write_room+0x4d/0x90 [nozomi]
...

Set tty->driver_data to the computed port in .install to not recompute it in
every place where needed. Switch .open to use driver_data too.

Signed-off-by: Jiri Slaby <jslaby@suse.cz>
Cc: Alan Cox <alan@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

Char: nozomi, fix tty->count counting

commit ee78bb95b7bea08b7774a02073ea2bb45611a9e1 upstream.

Currently ntty_install omits to increment tty count and we get the
following warnings:
Warning: dev (noz2) tty->count(0) != #fd's(1) in tty_open

So to fix that, add one tty->count++ there.

Signed-off-by: Jiri Slaby <jslaby@suse.cz>
Cc: Alan Cox <alan@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

ALSA: hda - Add PC-beep whitelist for an Intel board

commit e096c8e6d5ed965f346d94befbbec2275dde3621 upstream.

An Intel board needs a white-list entry to enable PC-beep.
Otherwise the driver misdetects (due to bogus BIOS info) and ignores
the PC-beep on 2.6.35.

Reported-and-tested-by: Leandro Lucarella <luca@llucax.com.ar>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

x86, mtrr: Use stop machine context to rendezvous all the cpu's

commit 68f202e4e87cfab4439568bf397fcc5c7cf8d729 upstream.

Use the stop machine context rather than IPI's to rendezvous all the cpus for
MTRR initialization that happens during cpu bringup or for MTRR modifications
during runtime.

This avoids deadlock scenario (reported by Prarit) like:

cpu A holds a read_lock (tasklist_lock for example) with irqs enabled
cpu B waits for the same lock with irqs disabled using write_lock_irq
cpu C doing set_mtrr() (during AP bringup for example), which will try to
rendezvous all the cpus using IPI's

This will result in C and A come to the rendezvous point and waiting
for B. B is stuck forever waiting for the lock and thus not
reaching the rendezvous point.

Using stop cpu (run in the process context of per cpu based keventd) to do
this rendezvous, avoids this deadlock scenario.

Also make sure all the cpu's are in the rendezvous handler before we proceed
with the local_irq_save() on each cpu. This lock step disabling irqs on all
the cpus will avoid other deadlock scenarios (for example involving
with the blocking smp_call_function's etc).

   [ This problem is very old. Marking -stable only for 2.6.35 as the
     stop_one_cpu_nowait() API is present only in 2.6.35. Any older
     kernel interested in this fix need to do some more work in backporting
     this patch. ]

Reported-by: Prarit Bhargava <prarit@redhat.com>
Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com>
LKML-Reference: <1280515602.2682.10.camel@sbsiddha-MOBL3.sc.intel.com>
Acked-by: Prarit Bhargava <prarit@redhat.com>
Signed-off-by: H. Peter Anvin <hpa@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

solos-pci: Fix race condition in tasklet RX handling

commit 1f6ea6e511e5ec730d8e88651da1b7b6e8fd1333 upstream.

We were seeing faults in the solos-pci receive tasklet when packets
arrived for a VCC which was currently being closed:

[18842.727906] EIP: [<e082f490>] br2684_push+0x19/0x234 [br2684] SS:ESP 0068:dfb89d14

[18845.090712] [<c13ecff3>] ? do_page_fault+0x0/0x2e1
[18845.120042] [<e082f490>] ? br2684_push+0x19/0x234 [br2684]
[18845.153530] [<e084fa13>] solos_bh+0x28b/0x7c8 [solos_pci]
[18845.186488] [<e084f711>] ? solos_irq+0x2d/0x51 [solos_pci]
[18845.219960] [<c100387b>] ? handle_irq+0x3b/0x48
[18845.247732] [<c10265cb>] ? irq_exit+0x34/0x57
[18845.274437] [<c1025720>] tasklet_action+0x42/0x69
[18845.303247] [<c102643f>] __do_softirq+0x8e/0x129
[18845.331540] [<c10264ff>] do_softirq+0x25/0x2a
[18845.358274] [<c102664c>] _local_bh_enable_ip+0x5e/0x6a
[18845.389677] [<c102666d>] local_bh_enable+0xb/0xe
[18845.417944] [<e08490a8>] ppp_unregister_channel+0x32/0xbb [ppp_generic]
[18845.458193] [<e08731ad>] pppox_unbind_sock+0x18/0x1f [pppox]

This patch uses an RCU-inspired approach to fix it. In the RX tasklet's
find_vcc() function we first refuse to use a VCC which already has the
ATM_VF_READY bit cleared. And in the VCC close function, we synchronise
with the tasklet to ensure that it can't still be using the VCC before
we continue and allow the VCC to be destroyed.

Signed-off-by: David Woodhouse <David.Woodhouse@intel.com>
Tested-by: Nathan Williams <nathan@traverse.com.au>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

PCI: disable MSI on VIA K8M800

commit 549e15611b4ac1de51ef0e0a79c2704f50a638a2 upstream.

MSI delivery from on-board ahci controller doesn't work on K8M800.  At
this point, it's unclear whether the culprit is with the ahci
controller or the host bridge.  Given the track record and considering
the rather minimal impact of MSI, disabling it seems reasonable.

Signed-off-by: Tejun Heo <tj@kernel.org>
Reported-by: Rainer Hurtado Navarro <publio.escipion.el.africano@gmail.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

PCI: Do not run NVidia quirks related to MSI with MSI disabled

commit 3d2a531804d16cd8df6dbbb0429c6f143e756049 upstream.

There is no reason to run NVidia-specific quirks related to HT MSI
mappings with MSI disabled via pci=nomsi, so make
__nv_msi_ht_cap_quirk() return immediately in that case.

This allows at least one machine to boot 100% of the time with
pci=nomsi (it still doesn't boot reliably without that).

Addresses https://bugzilla.kernel.org/show_bug.cgi?id=16443 .

Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

serial: add support for OX16PCI958 card

commit e847003f00d5eca3e3b3a6a1199f82b51293faf6 upstream.

Signed-off-by: Lytochkin Boris <lytboris@gmail.com>
Tested-by: Lytochkin Boris <lytboris@gmail.com>
Signed-off-by: Alexander Beregalov <a.beregalov@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

staging: hv: Fix Kconfig dependency of hv_blkvsc

commit 54d2379c20d814ced657cbc2c4ead8d1f8389fa2 upstream.

LBDAF is not available nor necessary on 64BIT kernel. This patch
fixed the dependency for hv_blkvsc module on 64BIT kernel.
Thanks vrataj2 [vrataj2@comcast.net] for reporting this problem.

Reported-by: vrataj2 <vrataj2@comcast.net>
Signed-off-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Hank Janssen <hjanssen@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

Staging: panel: Prevent double-calling of parport_release - fix oops.

commit 060132ae42cce3f9d2fd34d9a17b98362b44b9f9 upstream.

This patch prevents the code from calling parport_release and
parport_unregister_device twice with the same arguments - and thus fixes an oops.

Rationale:
After the first call the parport is already released and the
handle isn't valid anymore and calling parport_release and
parport_unregister_device twice isn't a good idea.

Signed-off-by: Peter Huewe <peterhuewe@gmx.de>
Acked-by: Willy Tarreau <w@1wt.eu>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

Staging: line6: needs to select SND_PCM

commit e928c077e5cdcd72ee762125b37232aec1ff49f3 upstream.

line6 uses snd_pcm*() functions, so it should select SND_PCM.

ERROR: "snd_pcm_period_elapsed" [drivers/staging/line6/line6usb.ko] undefined!
ERROR: "snd_pcm_set_ops" [drivers/staging/line6/line6usb.ko] undefined!
ERROR: "snd_pcm_lib_free_pages" [drivers/staging/line6/line6usb.ko] undefined!
ERROR: "snd_pcm_lib_ioctl" [drivers/staging/line6/line6usb.ko] undefined!
ERROR: "snd_pcm_lib_malloc_pages" [drivers/staging/line6/line6usb.ko] undefined!
ERROR: "snd_pcm_hw_constraint_ratdens" [drivers/staging/line6/line6usb.ko] undefined!
ERROR: "snd_pcm_format_physical_width" [drivers/staging/line6/line6usb.ko] undefined!
ERROR: "snd_pcm_lib_preallocate_pages_for_all" [drivers/staging/line6/line6usb.ko] undefined!
ERROR: "snd_pcm_new" [drivers/staging/line6/line6usb.ko] undefined!

Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Cc: Markus Grabner <grabner@icg.tugraz.at>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

staging: rt2870: Add USB ID for Belkin F6D4050 v2

commit 5d92fe3387d086fc2f10426fbdb6b86d6cce5a47 upstream.

Device missing from current tables.

Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>
Tested-by: Rod Huffaker <rod.huffaker@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

x86: Add memory modify constraints to xchg() and cmpxchg()

commit 113fc5a6e8c2288619ff7e8187a6f556b7e0d372 upstream.

xchg() and cmpxchg() modify their memory operands, not merely read
them.  For some versions of gcc the "memory" clobber has apparently
dealt with the situation, but not for all.

Originally-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
Cc: Glauber Costa <glommer@redhat.com>
Cc: Avi Kivity <avi@redhat.com>
Cc: Peter Palfrader <peter@palfrader.org>
Cc: Greg KH <gregkh@suse.de>
Cc: Alan Cox <alan@lxorguk.ukuu.org.uk>
Cc: Zachary Amsden <zamsden@redhat.com>
Cc: Marcelo Tosatti <mtosatti@redhat.com>
LKML-Reference: <4C4F7277.8050306@zytor.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

pcmcia: avoid buffer overflow in pcmcia_setup_isa_irq

commit 127c03cdbad9bd5af5d7f33bd31a1015a90cb77f upstream.

NR_IRQS may be as low as 16, causing a (harmless?) buffer overflow in
pcmcia_setup_isa_irq():

static u8 pcmcia_used_irq[NR_IRQS];

...

if ((try < 32) && pcmcia_used_irq[irq])
continue;

This is read-only, so if this address would be non-zero, it would just
mean we would not attempt an IRQ >= NR_IRQS -- which would fail anyway!
And as request_irq() fails for an irq >= NR_IRQS, the setting code path:

pcmcia_used_irq[irq]++;

is never reached as well.

Reported-by: Christoph Fritz <chf.fritz@googlemail.com>
Signed-off-by: Dominik Brodowski <linux@dominikbrodowski.net>
Signed-off-by: Christoph Fritz <chf.fritz@googlemail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

x86/PCI: use host bridge _CRS info on ASRock ALiveSATA2-GLAN

commit 2491762cfb475dbdfa3db11ebea6de49f58b7fac upstream.

This DMI quirk turns on "pci=use_crs" for the ALiveSATA2-GLAN because
amd_bus.c doesn't handle this system correctly.

The system has a single HyperTransport I/O chain, but has two PCI host
bridges to buses 00 and 80.  amd_bus.c learns the MMIO range associated
with buses 00-ff and that this range is routed to the HT chain hosted at
node 0, link 0:

    bus: [00, ff] on node 0 link 0
    bus: 00 index 1 [mem 0x80000000-0xfcffffffff]

This includes the address space for both bus 00 and bus 80, and amd_bus.c
assumes it's all routed to bus 00.

We find device 80:01.0, which BIOS left in the middle of that space, but
we don't find a bridge from bus 00 to bus 80, so we conclude that 80:01.0
is unreachable from bus 00, and we move it from the original, working,
address to something outside the bus 00 aperture, which does not work:

    pci 0000:80:01.0: reg 10: [mem 0xfebfc000-0xfebfffff 64bit]
    pci 0000:80:01.0: BAR 0: assigned [mem 0xfd00000000-0xfd00003fff 64bit]

The BIOS told us everything we need to know to handle this correctly,
so we're better off if we just pay attention, which lets us leave the
80:01.0 device at the original, working, address:

    ACPI: PCI Root Bridge [PCI0] (domain 0000 [bus 00-7f])
    pci_root PNP0A03:00: host bridge window [mem 0x80000000-0xff37ffff]
    ACPI: PCI Root Bridge [PCI1] (domain 0000 [bus 80-ff])
    pci_root PNP0A08:00: host bridge window [mem 0xfebfc000-0xfebfffff]

This was a regression between 2.6.33 and 2.6.34.  In 2.6.33, amd_bus.c
was used only when we found multiple HT chains.  3e3da00c01d050, which
enabled amd_bus.c even on systems with a single HT chain, caused this
failure.

This quirk was written by Graham.  If we ever enable "pci=use_crs" for
machines from 2006 or earlir, this quirk should be removed.

Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=16007

Reported-by: Graham Ramsey <ramsey.graham@ntlworld.com>
Signed-off-by: Bjorn Helgaas <bjorn.helgaas@hp.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>