7 years agoLinux
Greg Kroah-Hartman [Thu, 17 Feb 2011 23:14:38 +0000 (15:14 -0800)]

7 years agoxhci: Use GFP_NOIO during device reset.
Sarah Sharp [Tue, 28 Dec 2010 21:08:42 +0000 (13:08 -0800)]
xhci: Use GFP_NOIO during device reset.

commit a6d940dd759bf240d28624198660ed34582a327b upstream.

When xhci_discover_or_reset_device() is called after a host controller
power loss, the virtual device may need to be reallocated.  Make sure
xhci_alloc_dev() uses GFP_NOIO.  This avoid causing a deadlock by allowing
the kernel to flush pending I/O while reallocating memory for a virtual
device for a USB mass storage device that's holding the backing store for
dirty memory buffers.

This patch should be queued for the 2.6.37 stable tree.

Signed-off-by: Sarah Sharp <>
Signed-off-by: Greg Kroah-Hartman <>
7 years agousb: Realloc xHCI structures after a hub is verified.
Sarah Sharp [Thu, 23 Dec 2010 19:12:42 +0000 (11:12 -0800)]
usb: Realloc xHCI structures after a hub is verified.

commit 653a39d1f61bdc9f277766736d21d2e9be0391cb upstream.

When there's an xHCI host power loss after a suspend from memory, the USB
core attempts to reset and verify the USB devices that are attached to the
system.  The xHCI driver has to reallocate those devices, since the
hardware lost all knowledge of them during the power loss.

When a hub is plugged in, and the host loses power, the xHCI hardware
structures are not updated to say the device is a hub.  This is usually
done in hub_configure() when the USB hub is detected.  That function is
skipped during a reset and verify by the USB core, since the core restores
the old configuration and alternate settings, and the hub driver has no
idea this happened.  This bug makes the xHCI host controller reject the
enumeration of low speed devices under the resumed hub.

Therefore, make the USB core re-setup the internal xHCI hub device
information by calling update_hub_device() when hub_activate() is called
for a hub reset resume.  After a host power loss, all devices under the
roothub get a reset-resume or a disconnect.

This patch should be queued for the 2.6.37 stable tree.

Signed-off-by: Sarah Sharp <>
Signed-off-by: Greg Kroah-Hartman <>
7 years agoxhci: Do not run xhci_cleanup_msix with irq disabled
Zhang Rui [Fri, 17 Dec 2010 21:17:04 +0000 (13:17 -0800)]
xhci: Do not run xhci_cleanup_msix with irq disabled

commit 40a9fb17f32dbe54de3d636142a59288544deed7 upstream.

when unloading xhci_hcd, I got:
[  134.856813] xhci_hcd 0000:02:00.0: remove, state 4
[  134.858140] usb usb3: USB disconnect, address 1
[  134.874956] xhci_hcd 0000:02:00.0: Host controller not halted, aborting reset.
[  134.876351] BUG: sleeping function called from invalid context at kernel/mutex.c:85
[  134.877657] in_atomic(): 0, irqs_disabled(): 1, pid: 1451, name: modprobe
[  134.878975] Pid: 1451, comm: modprobe Not tainted 2.6.37-rc5+ #162
[  134.880298] Call Trace:
[  134.881602]  [<ffffffff8104156a>] __might_sleep+0xeb/0xf0
[  134.882921]  [<ffffffff814763dc>] mutex_lock+0x24/0x50
[  134.884229]  [<ffffffff810a745c>] free_desc+0x2e/0x5f
[  134.885538]  [<ffffffff810a74c8>] irq_free_descs+0x3b/0x71
[  134.886853]  [<ffffffff8102584d>] free_irq_at+0x31/0x36
[  134.888167]  [<ffffffff8102723f>] destroy_irq+0x69/0x71
[  134.889486]  [<ffffffff8102747a>] native_teardown_msi_irq+0xe/0x10
[  134.890820]  [<ffffffff8124c382>] default_teardown_msi_irqs+0x57/0x80
[  134.892158]  [<ffffffff8124be46>] free_msi_irqs+0x8b/0xe9
[  134.893504]  [<ffffffff8124cd46>] pci_disable_msix+0x35/0x39
[  134.894844]  [<ffffffffa01b444a>] xhci_cleanup_msix+0x31/0x51 [xhci_hcd]
[  134.896186]  [<ffffffffa01b4b3a>] xhci_stop+0x3a/0x80 [xhci_hcd]
[  134.897521]  [<ffffffff81341dd4>] usb_remove_hcd+0xfd/0x14a
[  134.898859]  [<ffffffff813500ae>] usb_hcd_pci_remove+0x5c/0xc6
[  134.900193]  [<ffffffff8123c606>] pci_device_remove+0x3f/0x91
[  134.901535]  [<ffffffff812e7ea4>] __device_release_driver+0x83/0xd9
[  134.902899]  [<ffffffff812e8571>] driver_detach+0x86/0xad
[  134.904222]  [<ffffffff812e7d56>] bus_remove_driver+0xb2/0xd8
[  134.905540]  [<ffffffff812e8633>] driver_unregister+0x6c/0x74
[  134.906839]  [<ffffffff8123c8e4>] pci_unregister_driver+0x44/0x89
[  134.908121]  [<ffffffffa01b940e>] xhci_unregister_pci+0x15/0x17 [xhci_hcd]
[  134.909396]  [<ffffffffa01bd7d2>] xhci_hcd_cleanup+0xe/0x10 [xhci_hcd]
[  134.910652]  [<ffffffff8107fcd1>] sys_delete_module+0x1ca/0x23b
[  134.911882]  [<ffffffff81123932>] ? path_put+0x22/0x26
[  134.913104]  [<ffffffff8109a800>] ? audit_syscall_entry+0x2c/0x148
[  134.914333]  [<ffffffff8100ac82>] system_call_fastpath+0x16/0x1b
[  134.915658] xhci_hcd 0000:02:00.0: USB bus 3 deregistered
[  134.916465] xhci_hcd 0000:02:00.0: PCI INT A disabled

and the same issue when xhci_suspend is invoked.  (Note from Sarah: That's
fixed by Andiry's patch before this, by synchronizing the irqs rather than
freeing them on suspend.)

Do not run xhci_cleanup_msix with irq disabled.

This patch should be queued for the 2.6.37 stable tree.

Signed-off-by: Zhang Rui <>
Signed-off-by: Sarah Sharp <>
Signed-off-by: Greg Kroah-Hartman <>
7 years agodrm/i915: Only bind to function 0 of the PCI device
Chris Wilson [Tue, 1 Feb 2011 19:43:02 +0000 (19:43 +0000)]
drm/i915: Only bind to function 0 of the PCI device

commit 5fe49d86f9d01044abf687a8cd21edef636d58aa upstream.

Early chipsets (gen2/3) used function 1 as a placeholder for multi-head.
We used to ignore these since they were not assigned to
PCI_CLASS_DISPLAY_VGA. However with 934f992c7 we attempt to bind to all
Intel PCI_CLASS_DISPLAY devices (and functions) to work in multi-gpu
systems. This fails hard on gen2/3.

Reported-by: Ferenc Wágner <>
Tested-by: Ferenc Wágner <>
Signed-off-by: Chris Wilson <>
Signed-off-by: Greg Kroah-Hartman <>
7 years agoagp: ensure GART has an address before enabling it
Stephen Kitt [Mon, 31 Jan 2011 22:25:43 +0000 (14:25 -0800)]
agp: ensure GART has an address before enabling it

commit a70b95c017e8b518e1e069853355e4e497453dbb upstream.

Some BIOSs (eg.  the AMI BIOS on the Asus P4P800 motherboard) don't
initialise the GART address, and pcibios_assign_resources() can ignore it
because it can be marked as a host bridge (see for details).  This
was handled correctly up to 2.6.35, but the pci_enable_device() cleanup in
2.6.36 96576a9e1a0cdb8 ("agp: intel-agp: do not use PCI resources before
pci_enable_device()") means that the kernel tries to enable the GART
before assigning it an address; in such cases the GART overlaps with other
device assignments and ends up being disabled.

This patch fixes

Note that I imagine efficeon-agp.c probably has the same problem, but
I can't test that and I'd like to make sure this patch is suitable for
-stable (since 2.6.36 and 2.6.37 are affected).

Signed-off-by: Stephen Kitt <>
Cc: Bjorn Helgaas <>
Cc: Maciej Rutecki <>
Cc: "Rafael J. Wysocki" <>
Cc: Kulikov Vasiliy <>
Cc: Florian Mickler <>
Signed-off-by: Andrew Morton <>
Signed-off-by: Dave Airlie <>
Signed-off-by: Greg Kroah-Hartman <>
7 years agox86, mm: avoid possible bogus tlb entries by clearing prev mm_cpumask after switching mm
Suresh Siddha [Thu, 3 Feb 2011 20:20:04 +0000 (12:20 -0800)]
x86, mm: avoid possible bogus tlb entries by clearing prev mm_cpumask after switching mm

commit 831d52bc153971b70e64eccfbed2b232394f22f8 upstream.

Clearing the cpu in prev's mm_cpumask early will avoid the flush tlb
IPI's while the cr3 is still pointing to the prev mm.  And this window
can lead to the possibility of bogus TLB fills resulting in strange
failures.  One such problematic scenario is mentioned below.

 T1. CPU-1 is context switching from mm1 to mm2 context and got a NMI
     etc between the point of clearing the cpu from the mm_cpumask(mm1)
     and before reloading the cr3 with the new mm2.

 T2. CPU-2 is tearing down a specific vma for mm1 and will proceed with
     flushing the TLB for mm1.  It doesn't send the flush TLB to CPU-1
     as it doesn't see that cpu listed in the mm_cpumask(mm1).

 T3. After the TLB flush is complete, CPU-2 goes ahead and frees the
     page-table pages associated with the removed vma mapping.

 T4. CPU-2 now allocates those freed page-table pages for something

 T5. As the CR3 and TLB caches for mm1 is still active on CPU-1, CPU-1
     can potentially speculate and walk through the page-table caches
     and can insert new TLB entries.  As the page-table pages are
     already freed and being used on CPU-2, this page walk can
     potentially insert a bogus global TLB entry depending on the
     (random) contents of the page that is being used on CPU-2.

 T6. This bogus TLB entry being global will be active across future CR3
     changes and can result in weird memory corruption etc.

To avoid this issue, for the prev mm that is handing over the cpu to
another mm, clear the cpu from the mm_cpumask(prev) after the cr3 is

Marking it for -stable, though we haven't seen any reported failure that
can be attributed to this.

Signed-off-by: Suresh Siddha <>
Acked-by: Ingo Molnar <>
Signed-off-by: Linus Torvalds <>
Signed-off-by: Greg Kroah-Hartman <>
7 years agodrm/i915: Recognise non-VGA display devices
Chris Wilson [Thu, 20 Jan 2011 13:09:12 +0000 (13:09 +0000)]
drm/i915: Recognise non-VGA display devices

commit 934f992c763ae1e5eefcce8af769c16444085df7 upstream.

Starting with SandyBridge (though possible with earlier hacked BIOSes),
the BIOS may initialise the IGFX as secondary to a discrete GPU. Prior,
it would simply disable the integrated GPU. So we adjust our PCI class
mask to match any DISPLAY_CLASS device.

In such a configuration, the IGFX is not a primary VGA controller and
so should not take part in VGA arbitration, and the error return from
vga_client_register() is expected.

Signed-off-by: Chris Wilson <>
Signed-off-by: Greg Kroah-Hartman <>
7 years agodrm/i915: Add dependency on CONFIG_TMPFS
Chris Wilson [Thu, 20 Jan 2011 10:03:24 +0000 (10:03 +0000)]
drm/i915: Add dependency on CONFIG_TMPFS

commit f7ab9b407b3bc83161c2aa74c992ba4782e87c9c upstream.

Without tmpfs, shmem_readpage() is not compiled in causing an OOPS as
soon as we try to allocate some swappable pages for GEM.

Jan 19 22:52:26 harlie kernel: Modules linked in: i915(+) drm_kms_helper cfbcopyarea video backlight cfbimgblt cfbfillrect
Jan 19 22:52:26 harlie kernel:
Jan 19 22:52:26 harlie kernel: Pid: 1125, comm: modprobe Not tainted 2.6.37Harlie #10 To be filled by O.E.M./To be filled by O.E.M.
Jan 19 22:52:26 harlie kernel: EIP: 0060:[<00000000>] EFLAGS: 00010246 CPU: 3
Jan 19 22:52:26 harlie kernel: EIP is at 0x0
Jan 19 22:52:26 harlie kernel: EAX: 00000000 EBX: f7b7d000 ECX: f3383100 EDX: f7b7d000
Jan 19 22:52:26 harlie kernel: ESI: f1456118 EDI: 00000000 EBP: f2303c98 ESP: f2303c7c
Jan 19 22:52:26 harlie kernel:  DS: 007b ES: 007b FS: 00d8 GS: 0033 SS: 0068
Jan 19 22:52:26 harlie kernel: Process modprobe (pid: 1125, ti=f2302000 task=f259cd80 task.ti=f2302000)
Jan 19 22:52:26 harlie kernel: Stack:
Jan 19 22:52:26 harlie udevd-work[1072]: '/sbin/modprobe -b pci:v00008086d00000046sv00000000sd00000000bc03sc00i00' unexpected exit with status 0x0009
Jan 19 22:52:26 harlie kernel:  c1074061 000000d0 f2f42b80 00000000 000a13d2 f2d5dcc0 00000001 f2303cac
Jan 19 22:52:26 harlie kernel:  c107416f 00000000 000a13d2 00000000 f2303cd4 f8d620ed f2cee620 00001000
Jan 19 22:52:26 harlie kernel:  00000000 000a13d2 f1456118 f2d5dcc0 f1a40000 00001000 f2303d04 f8d637ab
Jan 19 22:52:26 harlie kernel: Call Trace:
Jan 19 22:52:26 harlie kernel:  [<c1074061>] ? do_read_cache_page+0x71/0x160
Jan 19 22:52:26 harlie kernel:  [<c107416f>] ? read_cache_page_gfp+0x1f/0x30
Jan 19 22:52:26 harlie kernel:  [<f8d620ed>] ? i915_gem_object_get_pages+0xad/0x1d0 [i915]
Jan 19 22:52:26 harlie kernel:  [<f8d637ab>] ? i915_gem_object_bind_to_gtt+0xeb/0x2d0 [i915]
Jan 19 22:52:26 harlie kernel:  [<f8d65961>] ? i915_gem_object_pin+0x151/0x190 [i915]
Jan 19 22:52:26 harlie kernel:  [<c11e16ed>] ? drm_gem_object_init+0x3d/0x60
Jan 19 22:52:26 harlie kernel:  [<f8d65aa5>] ? i915_gem_init_ringbuffer+0x105/0x1e0 [i915]
Jan 19 22:52:26 harlie kernel:  [<f8d571b7>] ? i915_driver_load+0x667/0x1160 [i915]

Reported-by: John J. Stimson-III <>
Signed-off-by: Chris Wilson <>
Signed-off-by: Greg Kroah-Hartman <>
7 years agodrm/i915/lvds: Add AOpen i915GMm-HFS to the list of false-positive LVDS
Knut Petersen [Fri, 14 Jan 2011 15:38:10 +0000 (15:38 +0000)]
drm/i915/lvds: Add AOpen i915GMm-HFS to the list of false-positive LVDS

commit 22ab70d3262ddb6e69b3c246a34e2967ba5eb1e8 upstream.

Signed-off-by: Knut Petersen <>
Signed-off-by: Chris Wilson <>
Signed-off-by: Greg Kroah-Hartman <>
7 years agodrm/i915: fix calculation of eDP signal levels on Sandybridge
Yuanhan Liu [Thu, 6 Jan 2011 10:26:08 +0000 (18:26 +0800)]
drm/i915: fix calculation of eDP signal levels on Sandybridge

commit 3c5a62b5226ca5db993660281e9c2a7275d9fb02 upstream.

Some voltage swing/pre-emphasis level use the same value on eDP
Sandybridge, like 400mv_0db and 600mv_0db are with the same value
of (0x0 << 22). So, fix them, and point out the value if it isn't
a supported voltage swing/pre-emphasis level.

Signed-off-by: Yuanhan Liu <>
Signed-off-by: Chris Wilson <>
Signed-off-by: Greg Kroah-Hartman <>
7 years agodrm: Restore the old_fb upon modeset failure
Chris Wilson [Sat, 8 Jan 2011 15:10:41 +0000 (15:10 +0000)]
drm: Restore the old_fb upon modeset failure

commit 0ba41e449fd0f45f5b29c1009020ab1b298bedda upstream.

... or else we may end up disabling the wrong framebuffer, leading to an
OOPS, e.g:

[ 6033.229012] kernel BUG at drivers/gpu/drm/i915/i915_gem.c:3271!
[ 6033.229012] invalid opcode: 0000 [#1] SMP
[ 6033.229012] last sysfs file:
[ 6033.229012] Modules linked in: sunrpc cpufreq_ondemand acpi_cpufreq
mperf snd_hda_codec_analog snd_hda_intel snd_hda_codec snd_hwdep snd_seq
snd_seq_device snd_pcm snd_timer thinkpad_acpi ppdev snd r852 sm_common
iTCO_wdt uvcvideo i2c_i801 iTCO_vendor_support microcode wmi nand
videodev nand_ids nand_ecc snd_page_alloc parport_pc parport mtd
soundcore joydev v4l1_compat pcspkr uinput ipv6 sdhci_pci sdhci mmc_core
yenta_socket i915 drm_kms_helper drm i2c_algo_bit i2c_core video output
[last unloaded: scsi_wait_scan]
[ 6033.229012]
[ 6033.229012] Pid: 4834, comm: Xorg Not tainted 2.6.37-rc8+ #25 7661BL5/7661BL5
[ 6033.229012] EIP: 0060:[<f86fda5e>] EFLAGS: 00013246 CPU: 0
[ 6033.229012] EIP is at i915_gem_object_unpin+0x23/0x76 [i915]
[ 6033.229012] EAX: f68a4000 EBX: f6831f00 ECX: 000600fa EDX: f68a8000
[ 6033.229012] ESI: f68a4014 EDI: f68a42b8 EBP: f2169c44 ESP: f2169c3c
[ 6033.229012]  DS: 007b ES: 007b FS: 00d8 GS: 00e0 SS: 0068
[ 6033.229012] Process Xorg (pid: 4834, ti=f2168000 task=f21c8000 task.ti=f2168000)
[ 6033.229012] Stack:
[ 6033.229012]  f3a84800 f68a4014 f2169c54 f87045d8 f3a84800 f872d9a8 f2169c68 f7fd8091
[ 6033.229012]  f3b952a4 00000000 f68a414c f2169cf0 f7fd9377 00000000 00000000 f7fd98b0
[ 6033.229012]  f7fd9f4e 0000000f f7f328a0 00000000 00000000 00000000 f2169ca4 f68a414c
[ 6033.229012] Call Trace:
[ 6033.229012]  [<f87045d8>] ? intel_crtc_disable+0x36/0x41 [i915]
[ 6033.229012]  [<f7fd8091>] ?  drm_helper_disable_unused_functions+0xcd/0xf9 [drm_kms_helper]
[ 6033.229012]  [<f7fd9377>] ? drm_crtc_helper_set_config+0x62a/0x7f7 [drm_kms_helper]
[ 6033.229012]  [<c04daa10>] ? __slab_free+0x1b/0xa4
[ 6033.229012]  [<f7fd7e62>] ? drm_fb_helper_initial_config+0x466/0x497 [drm_kms_helper]
[ 6033.229012]  [<f7fd7ea3>] ? drm_fb_helper_restore+0x10/0x2a [drm_kms_helper]
[ 6033.229012]  [<f86f2577>] ? i915_driver_lastclose+0x2a/0x57 [i915]
[ 6033.229012]  [<f7f1989f>] ? drm_lastclose+0x45/0x23e [drm]
[ 6033.229012]  [<f7f1a0b4>] ? drm_release+0x462/0x4d7 [drm]

Signed-off-by: Chris Wilson <>
Signed-off-by: Dave Airlie <>
Signed-off-by: Greg Kroah-Hartman <>
7 years agodrm/radeon/kms: fix s/r issues with bios scratch regs
Alex Deucher [Thu, 3 Feb 2011 00:46:06 +0000 (19:46 -0500)]
drm/radeon/kms: fix s/r issues with bios scratch regs

commit 87364760de5d631390c478fcbac8db1b926e0adf upstream.

The accelerate mode bit gets checked by certain atom
command tables to set up some register state.  It needs
to be clear when setting modes and set when not.


Signed-off-by: Alex Deucher <>
Signed-off-by: Dave Airlie <>
Signed-off-by: Greg Kroah-Hartman <>
7 years agodrm/radeon: remove 0x4243 pci id
Alex Deucher [Wed, 2 Feb 2011 00:06:46 +0000 (19:06 -0500)]
drm/radeon: remove 0x4243 pci id

commit 63a507800c8aca5a1891d598ae13f829346e8e39 upstream.

0x4243 is a PCI bridge, not a GPU.


Signed-off-by: Alex Deucher <>
Signed-off-by: Dave Airlie <>
Signed-off-by: Greg Kroah-Hartman <>
7 years agodrm/radeon/kms: add pll debugging output
Alex Deucher [Mon, 31 Jan 2011 21:48:51 +0000 (16:48 -0500)]
drm/radeon/kms: add pll debugging output

commit 51d4bf840a27fe02c883ddc6d9708af056773769 upstream.

Signed-off-by: Alex Deucher <>
Signed-off-by: Dave Airlie <>
Signed-off-by: Greg Kroah-Hartman <>
7 years agoradeon/kms: fix dp displayport mode validation
Jerome Glisse [Wed, 26 Jan 2011 22:51:03 +0000 (17:51 -0500)]
radeon/kms: fix dp displayport mode validation

commit 6bba2e116808ca12e30c8d88dfedabf8b8d67390 upstream.

Check if there is a big enough dp clock & enough dp lane to
drive the video mode provided.

Signed-off-by: Jerome Glisse <>
Reviewed-By: Alex Deucher <>
Signed-off-by: Dave Airlie <>
Signed-off-by: Greg Kroah-Hartman <>
7 years agodrm/radeon/kms: make the mac rv630 quirk generic
Alex Deucher [Tue, 18 Jan 2011 18:26:11 +0000 (18:26 +0000)]
drm/radeon/kms: make the mac rv630 quirk generic

commit be23da8ad219650517cbbb7acbeaeb235667113a upstream.

Seems some other boards do this as well.

Reported-by: Andrea Merello <>
Signed-off-by: Alex Deucher <>
Signed-off-by: Dave Airlie <>
Signed-off-by: Greg Kroah-Hartman <>
7 years agodrm/radeon/kms: adjust quirk for acer laptop
Alex Deucher [Tue, 4 Jan 2011 22:42:20 +0000 (17:42 -0500)]
drm/radeon/kms: adjust quirk for acer laptop

commit 2f299d5de02da3ffb1f9e1a05c91dcd1173ebd3c upstream.

Acer laptop (TravelMate 5730G) has an HDMI connector
on the laptop and a DVI connector on the docking station
and both share the same encoder, hpd pin, and ddc line.
The bios connector table reflects this and is technically
correct, however, we drop the DVI connector here since
xrandr has no concept of encoders (only crtcs and connectors)
and will try and drive both connectors with different crtcs
which isn't possible on the hardware side and leaves no crtcs
for LVDS or VGA.


Signed-off-by: Alex Deucher <>
Signed-off-by: Dave Airlie <>
Signed-off-by: Greg Kroah-Hartman <>
7 years agodrm/radeon/kms: add quirk for Mac Radeon HD 2600 card
Alex Deucher [Tue, 4 Jan 2011 05:43:39 +0000 (00:43 -0500)]
drm/radeon/kms: add quirk for Mac Radeon HD 2600 card

commit f598aa7593427ffe3a61e7767c34bd695a5e7ed0 upstream.

Reported-by: 屋国遥 <>
Signed-off-by: Alex Deucher <>
Signed-off-by: Dave Airlie <>
Signed-off-by: Greg Kroah-Hartman <>
7 years agonet/fec: fix MMFR_OP type in fec_enet_mdio_write
Shawn Guo [Wed, 5 Jan 2011 21:13:09 +0000 (21:13 +0000)]
net/fec: fix MMFR_OP type in fec_enet_mdio_write

commit 862f0982eadcea0e114576c57ea426d3d51a69a6 upstream.

FEC_MMFR_OP_WRITE should be used than FEC_MMFR_OP_READ in
a mdio write operation.

It's probably a typo introduced by commit:

netdev/fec.c: add phylib supporting to enable carrier detection (v2)

Signed-off-by: Shawn Guo <>
Signed-off-by: David S. Miller <>
Signed-off-by: Greg Kroah-Hartman <>
7 years agodm mpath: disable blk_abort_queue
Mike Snitzer [Thu, 13 Jan 2011 19:59:46 +0000 (19:59 +0000)]
dm mpath: disable blk_abort_queue

commit 09c9d4c9b6a2b5909ae3c6265e4cd3820b636863 upstream.

Revert commit 224cb3e981f1b2f9f93dbd49eaef505d17d894c2
  dm: Call blk_abort_queue on failed paths

Multipath began to use blk_abort_queue() to allow for
lower latency path deactivation.  This was found to
cause list corruption:

   the cmd gets blk_abort_queued/timedout run on it and the scsi eh
   somehow is able to complete and run scsi_queue_insert while
   scsi_request_fn is still trying to process the request.

Signed-off-by: Mike Snitzer <>
Signed-off-by: Alasdair G Kergon <>
Cc: Mike Anderson <>
Cc: Mike Christie <>
Signed-off-by: Greg Kroah-Hartman <>
7 years agodm: dont take i_mutex to change device size
Mike Snitzer [Thu, 13 Jan 2011 19:53:46 +0000 (19:53 +0000)]
dm: dont take i_mutex to change device size

commit c217649bf2d60ac119afd71d938278cffd55962b upstream.

No longer needlessly hold md->bdev->bd_inode->i_mutex when changing the
size of a DM device.  This additional locking is unnecessary because
i_size_write() is already protected by the existing critical section in
dm_swap_table().  DM already has a reference on md->bdev so the
associated bd_inode may be changed without lifetime concerns.

A negative side-effect of having held md->bdev->bd_inode->i_mutex was
that a concurrent DM device resize and flush (via fsync) would deadlock.
Dropping md->bdev->bd_inode->i_mutex eliminates this potential for
deadlock.  The following reproducer no longer deadlocks:

Signed-off-by: Mike Snitzer <>
Signed-off-by: Mikulas Patocka <>
Signed-off-by: Alasdair G Kergon <>
Signed-off-by: Greg Kroah-Hartman <>
7 years agoieee80211: correct IEEE80211_ADDBA_PARAM_BUF_SIZE_MASK macro
Amitkumar Karwar [Wed, 12 Jan 2011 00:14:24 +0000 (16:14 -0800)]
ieee80211: correct IEEE80211_ADDBA_PARAM_BUF_SIZE_MASK macro

commit 8d661f1e462d50bd83de87ee628aaf820ce3c66c upstream.

It is defined in include/linux/ieee80211.h. As per IEEE spec.
bit6 to bit15 in block ack parameter represents buffer size.
So the bitmask should be 0xFFC0.

Signed-off-by: Amitkumar Karwar <>
Signed-off-by: Bing Zhao <>
Reviewed-by: Johannes Berg <>
Signed-off-by: John W. Linville <>
Signed-off-by: Greg Kroah-Hartman <>
7 years agotpm: Autodetect itpm devices
Matthew Garrett [Thu, 21 Oct 2010 21:42:40 +0000 (17:42 -0400)]
tpm: Autodetect itpm devices

commit 3f0d3d016d89a5efb8b926d4707eb21fa13f3d27 upstream.

Some Lenovos have TPMs that require a quirk to function correctly. This can
be autodetected by checking whether the device has a _HID of INTC0102. This
is an invalid PNPid, and as such is discarded by the pnp layer - however
it's still present in the ACPI code, so we can pull it out that way. This
means that the quirk won't be automatically applied on non-ACPI systems,
but without ACPI we don't have any way to identify the chip anyway so I
don't think that's a great concern.

Signed-off-by: Matthew Garrett <>
Acked-by: Rajiv Andrade <>
Tested-by: Jiri Kosina <>
Tested-by: Andy Isaacson <>
Signed-off-by: James Morris <>
Signed-off-by: Greg Kroah-Hartman <>
7 years agoSELinux: do not compute transition labels on mountpoint labeled filesystems
Eric Paris [Thu, 2 Dec 2010 21:13:40 +0000 (16:13 -0500)]
SELinux: do not compute transition labels on mountpoint labeled filesystems

commit 415103f9932d45f7927f4b17e3a9a13834cdb9a1 upstream.

selinux_inode_init_security computes transitions sids even for filesystems
that use mount point labeling.  It shouldn't do that.  It should just use
the mount point label always and no matter what.

This causes 2 problems.  1) it makes file creation slower than it needs to be
since we calculate the transition sid and 2) it allows files to be created
with a different label than the mount point!

# id -Z
# sesearch --type --class file --source sysadm_t --target tmp_t
Found 1 semantic te rules:
   type_transition sysadm_t tmp_t : file user_tmp_t;

# mount -o loop,context="system_u:object_r:tmp_t:s0"  /tmp/fs /mnt/tmp

# ls -lZ /mnt/tmp
drwx------. root root system_u:object_r:tmp_t:s0       lost+found
# touch /mnt/tmp/file1
# ls -lZ /mnt/tmp
-rw-r--r--. root root staff_u:object_r:user_tmp_t:s0   file1
drwx------. root root system_u:object_r:tmp_t:s0       lost+found

Whoops, we have a mount point labeled filesystem tmp_t with a user_tmp_t
labeled file!

Signed-off-by: Eric Paris <>
Reviewed-by: Reviewed-by: James Morris <>
Signed-off-by: Greg Kroah-Hartman <>
7 years agoSELinux: define permissions for DCB netlink messages
Eric Paris [Thu, 16 Dec 2010 16:46:51 +0000 (11:46 -0500)]
SELinux: define permissions for DCB netlink messages

commit 350e4f31e0eaf56dfc3b328d24a11bdf42a41fb8 upstream.

Commit 2f90b865 added two new netlink message types to the netlink route
socket.  SELinux has hooks to define if netlink messages are allowed to
be sent or received, but it did not know about these two new message
types.  By default we allow such actions so noone likely noticed.  This
patch adds the proper definitions and thus proper permissions

Signed-off-by: Eric Paris <>
Cc: James Morris <>
Signed-off-by: Greg Kroah-Hartman <>
7 years agoKEYS: Don't call up_write() if __key_link_begin() returns an error
David Howells [Wed, 22 Dec 2010 16:24:13 +0000 (16:24 +0000)]
KEYS: Don't call up_write() if __key_link_begin() returns an error

commit 3fc5e98d8cf85e0d77fc597b49e9268dff67400e upstream.

In construct_alloc_key(), up_write() is called in the error path if
__key_link_begin() fails, but this is incorrect as __key_link_begin() only
returns with the nominated keyring locked if it returns successfully.

Without this patch, you might see the following in dmesg:

[ BUG: bad unlock balance detected! ]
mount.cifs/5769 is trying to release lock (&key->sem) at:
[<ffffffff81201159>] request_key_and_link+0x263/0x3fc
but there are no more locks to release!

other info that might help us debug this:
3 locks held by mount.cifs/5769:
 #0:  (&type->s_umount_key#41/1){+.+.+.}, at: [<ffffffff81131321>] sget+0x278/0x3e7
 #1:  (&ret_buf->session_mutex){+.+.+.}, at: [<ffffffffa0258e59>] cifs_get_smb_ses+0x35a/0x443 [cifs]
 #2:  (root_key_user.cons_lock){+.+.+.}, at: [<ffffffff81201000>] request_key_and_link+0x10a/0x3fc

stack backtrace:
Pid: 5769, comm: mount.cifs Not tainted 2.6.37-rc6+ #1
Call Trace:
 [<ffffffff81201159>] ? request_key_and_link+0x263/0x3fc
 [<ffffffff81081601>] print_unlock_inbalance_bug+0xca/0xd5
 [<ffffffff81083248>] lock_release_non_nested+0xc1/0x263
 [<ffffffff81201159>] ? request_key_and_link+0x263/0x3fc
 [<ffffffff81201159>] ? request_key_and_link+0x263/0x3fc
 [<ffffffff81083567>] lock_release+0x17d/0x1a4
 [<ffffffff81073f45>] up_write+0x23/0x3b
 [<ffffffff81201159>] request_key_and_link+0x263/0x3fc
 [<ffffffffa026fe9e>] ? cifs_get_spnego_key+0x61/0x21f [cifs]
 [<ffffffff812013c5>] request_key+0x41/0x74
 [<ffffffffa027003d>] cifs_get_spnego_key+0x200/0x21f [cifs]
 [<ffffffffa026e296>] CIFS_SessSetup+0x55d/0x1273 [cifs]
 [<ffffffffa02589e1>] cifs_setup_session+0x90/0x1ae [cifs]
 [<ffffffffa0258e7e>] cifs_get_smb_ses+0x37f/0x443 [cifs]
 [<ffffffffa025a9e3>] cifs_mount+0x1aa1/0x23f3 [cifs]
 [<ffffffff8111fd94>] ? alloc_debug_processing+0xdb/0x120
 [<ffffffffa027002c>] ? cifs_get_spnego_key+0x1ef/0x21f [cifs]
 [<ffffffffa024cc71>] cifs_do_mount+0x165/0x2b3 [cifs]
 [<ffffffff81130e72>] vfs_kern_mount+0xaf/0x1dc
 [<ffffffff81131007>] do_kern_mount+0x4d/0xef
 [<ffffffff811483b9>] do_mount+0x6f4/0x733
 [<ffffffff8114861f>] sys_mount+0x88/0xc2
 [<ffffffff8100ac42>] system_call_fastpath+0x16/0x1b

Reported-by: Jeff Layton <>
Signed-off-by: David Howells <>
Reviewed-and-Tested-by: Jeff Layton <>
Signed-off-by: Linus Torvalds <>
Cc: James Morris <>
Signed-off-by: Greg Kroah-Hartman <>
7 years agotpm_tis: Use timeouts returned from TPM
Stefan Berger [Tue, 11 Jan 2011 19:37:29 +0000 (14:37 -0500)]
tpm_tis: Use timeouts returned from TPM

commit 9b29050f8f75916f974a2d231ae5d3cd59792296 upstream.

The current TPM TIS driver in git discards the timeout values returned
from the TPM. The check of the response packet needs to consider that
the return_code field is 0 on success and the size of the expected
packet is equivalent to the header size + u32 length indicator for the
TPM_GetCapability() result + 3 timeout indicators of type u32.

I am also adding a sysfs entry 'timeouts' showing the timeouts that are
being used.

Signed-off-by: Stefan Berger <>
Tested-by: Guillaume Chazarain <>
Signed-off-by: Rajiv Andrade <>
Signed-off-by: Greg Kroah-Hartman <>
7 years agoTPM: Long default timeout fix
Rajiv Andrade [Fri, 12 Nov 2010 21:30:02 +0000 (22:30 +0100)]
TPM: Long default timeout fix

commit c4ff4b829ef9e6353c0b133b7adb564a68054979 upstream.

If duration variable value is 0 at this point, it's because
chip->vendor.duration wasn't filled by tpm_get_timeouts() yet.
This patch sets then the lowest timeout just to give enough
time for tpm_get_timeouts() to further succeed.

This fix avoids long boot times in case another entity attempts
to send commands to the TPM when the TPM isn't accessible.

Signed-off-by: Rajiv Andrade <>
Signed-off-by: James Morris <>
Signed-off-by: Greg Kroah-Hartman <>
7 years agonilfs2: fix crash after one superblock became unavailable
Ryusuke Konishi [Fri, 21 Jan 2011 07:40:31 +0000 (16:40 +0900)]
nilfs2: fix crash after one superblock became unavailable

commit 0ca7a5b9ac5d301845dd6382ff25a699b6263a81 upstream.

Fixes the following kernel oops in nilfs_setup_super() which could
arise if one of two super-blocks is unavailable.

> BUG: unable to handle kernel NULL pointer dereference at   (null)
> Pid: 3529, comm: mount.nilfs2 Not tainted 2.6.37 #1 /
> EIP: 0060:[<c03196bc>] EFLAGS: 00010202 CPU: 3
> EIP is at memcpy+0xc/0x1b
> Call Trace:
>  [<f953720e>] ? nilfs_setup_super+0x6c/0xa5 [nilfs2]
>  [<f95369e9>] ? nilfs_get_root_dentry+0x81/0xcb [nilfs2]
>  [<f9537a08>] ? nilfs_mount+0x4f9/0x62c [nilfs2]
>  [<c02745cf>] ? kstrdup+0x36/0x3f
>  [<f953750f>] ? nilfs_mount+0x0/0x62c [nilfs2]
>  [<c0293940>] ? vfs_kern_mount+0x4d/0x12c
>  [<c02a5100>] ? get_fs_type+0x76/0x8f
>  [<c0293a68>] ? do_kern_mount+0x33/0xbf
>  [<c02a784a>] ? do_mount+0x2ed/0x714
>  [<c02a6171>] ? copy_mount_options+0x28/0xfc
>  [<c02a7ce3>] ? sys_mount+0x72/0xaf
>  [<c0473085>] ? syscall_call+0x7/0xb

Reported-by: Wakko Warner <>
Signed-off-by: Ryusuke Konishi <>
Tested-by: Wakko Warner <>
LKML-Reference: <>
Signed-off-by: Greg Kroah-Hartman <>
7 years agoKVM: MMU: Fix 32 bit legacy paging with NPT
Joerg Roedel [Fri, 14 Jan 2011 12:10:19 +0000 (10:10 -0200)]
KVM: MMU: Fix 32 bit legacy paging with NPT

commit f87f928882d080eaec8b0d76aecff003d664697d upstream.

This patch fixes 32 bit legacy paging with NPT enabled. The
mmu_check_root call on the top-level of the loop causes
root_gfn to take values (in the tdp_enabled path) which are
outside of guest memory. So the mmu_check_root call fails at
some point in the loop interation causing the guest to
This patch changes the mmu_check_root calls to the places
where they are really necessary. As a side-effect it
introduces a check for the root of a pae page table too.

Signed-off-by: Joerg Roedel <>
Signed-off-by: Marcelo Tosatti <>
Signed-off-by: Greg Kroah-Hartman <>
7 years agoKVM: MMU: Fix incorrect direct gfn for unpaged mode shadow
Avi Kivity [Fri, 14 Jan 2011 12:10:18 +0000 (10:10 -0200)]
KVM: MMU: Fix incorrect direct gfn for unpaged mode shadow

commit c093b8b46c5f0dd12d799f0d6a3b579863df72f6 upstream.

We use the physical address instead of the base gfn for the four
PAE page directories we use in unpaged mode.  When the guest accesses
an address above 1GB that is backed by a large host page, a BUG_ON()
in kvm_mmu_set_gfn() triggers.

Reported-and-tested-by: Nicolas Prochazka <>
Signed-off-by: Avi Kivity <>
Cc: Marcelo Tosatti <>
Signed-off-by: Greg Kroah-Hartman <>
7 years agoKVM: i8259: initialize isr_ack
Avi Kivity [Fri, 14 Jan 2011 12:10:17 +0000 (10:10 -0200)]
KVM: i8259: initialize isr_ack

commit a0272630bb594b4eac03a79e77957df7dad8eade upstream.

isr_ack is never initialized.  So, until the first PIC reset, interrupts
may fail to be injected.  This can cause Windows XP to fail to boot, as
reported in the fallout from the fix to

Reported-and-tested-by: Nicolas Prochazka <>
Signed-off-by: Avi Kivity <>
Signed-off-by: Greg Kroah-Hartman <>
7 years agomd_make_request: don't touch the bio after calling make_request
Chris Mason [Tue, 8 Feb 2011 00:21:48 +0000 (19:21 -0500)]
md_make_request: don't touch the bio after calling make_request

commit e91ece5590b3c728624ab57043fc7a05069c604a upstream.

md_make_request was calling bio_sectors() for part_stat_add
after it was calling the make_request function.  This is
bad because the make_request function can free the bio and
because the bi_size field can change around.

The fix here was suggested by Jens Axboe.  It saves the
sector count before the make_request call.  I hit this
with CONFIG_DEBUG_PAGEALLOC turned on while trying to break
his pretty fusionio card.

Signed-off-by: Chris Mason <>
Signed-off-by: NeilBrown <>
Signed-off-by: Greg Kroah-Hartman <>
7 years agopata_mpc52xx: inherit from ata_bmdma_port_ops
Tejun Heo [Sun, 9 Jan 2011 22:48:20 +0000 (17:48 -0500)]
pata_mpc52xx: inherit from ata_bmdma_port_ops

commit 77c5fd19075d299fe820bb59bb21b0b113676e20 upstream.

pata_mpc52xx supports BMDMA but inherits ata_sff_port_ops which
triggers BUG_ON() when a DMA command is issued.  Fix it.

Signed-off-by: Tejun Heo <>
Reported-by: Roman Fietze <>
Cc: Sergei Shtylyov <>
Signed-off-by: Jeff Garzik <>
Signed-off-by: Greg Kroah-Hartman <>
7 years agomd: Fix removal of extra drives when converting RAID6 to RAID5
NeilBrown [Thu, 13 Jan 2011 22:14:34 +0000 (09:14 +1100)]
md: Fix removal of extra drives when converting RAID6 to RAID5

commit bf2cb0dab8c97f00a71875d9b13dbac17a2f47ca upstream.

When a RAID6 is converted to a RAID5, the extra drive should
be discarded.  However it isn't due to a typo in a comparison.

This bug was introduced in commit e93f68a1fc6 in 2.6.35-rc4
and is suitable for any -stable since than.

As the extra drive is not removed, the 'degraded' counter is wrong and
so the RAID5 will not respond correctly to a subsequent failure.

Signed-off-by: NeilBrown <>
Signed-off-by: Greg Kroah-Hartman <>
7 years agomd: Ensure no IO request to get md device before it is properly initialised.
NeilBrown [Thu, 13 Jan 2011 22:14:33 +0000 (09:14 +1100)]
md: Ensure no IO request to get md device before it is properly initialised.

commit 0ca69886a8273ac1350143d562280bfcbe4760dc upstream.

When an md device is in the process of coming on line it is possible
for an IO request (typically a partition table probe) to get through
before the array is fully initialised, which can cause unexpected
behaviour (e.g. a crash).

So explicitly record when the array is ready for IO and don't allow IO
through until then.

There is no possibility for a similar problem when the array is going
off-line as there must only be one 'open' at that time, and it is busy
off-lining the array and so cannot send IO requests.  So no memory
barrier is needed in md_stop()

This has been a bug since commit 409c57f3801 in 2.6.30 which
introduced md_make_request.  Before then, each personality would
register its own make_request_fn when it was ready.
This is suitable for any stable kernel from 2.6.30.y onwards.

Signed-off-by: NeilBrown <>
Reported-by: "Hawrylewicz Czarnowski, Przemyslaw" <>
Signed-off-by: Greg Kroah-Hartman <>
7 years agomd: fix regression resulting in delays in clearing bits in a bitmap
NeilBrown [Thu, 13 Jan 2011 22:13:53 +0000 (09:13 +1100)]
md: fix regression resulting in delays in clearing bits in a bitmap

commit 6c9879101442b08581e8a0e3ae6b7f643a78fd63 upstream.

commit 589a594be1fb (2.6.37-rc4) fixed a problem were md_thread would
sometimes call the ->run function at a bad time.

If an error is detected during array start up after the md_thread has
been started, the md_thread is killed.  This resulted in the ->run
function being called once.  However the array may not be in a state
that it is safe to call ->run.

However the fix imposed meant that  ->run was not called on a timeout.
This means that when an array goes idle, bitmap bits do not get
cleared promptly.  While the array is busy the bits will still be
cleared when appropriate so this is not very serious.  There is no
risk to data.

Change the test so that we only avoid calling ->run when the thread
is being stopped.  This more explicitly addresses the problem situation.

This is suitable for 2.6.37-stable and any -stable kernel to which
589a594be1fb was applied.

Signed-off-by: NeilBrown <>
Signed-off-by: Greg Kroah-Hartman <>
7 years agomd: fix regression with re-adding devices to arrays with no metadata
NeilBrown [Tue, 11 Jan 2011 22:03:35 +0000 (09:03 +1100)]
md: fix regression with re-adding devices to arrays with no metadata

commit bf572541ab44240163eaa2d486b06f306a31d45a upstream.

Commit 1a855a0606 (2.6.37-rc4) fixed a problem where devices were
re-added when they shouldn't be but caused a regression in a less
common case that means sometimes devices cannot be re-added when they
should be.

In particular, when re-adding a device to an array without metadata
we should always access the device, but after the above commit we

This patch sets the In_sync flag in that case so that the re-add

This patch is suitable for any -stable kernel to which 1a855a0606 was

Signed-off-by: NeilBrown <>
Signed-off-by: Greg Kroah-Hartman <>
7 years agokernel/smp.c: fix smp_call_function_many() SMP race
Anton Blanchard [Thu, 20 Jan 2011 22:44:33 +0000 (14:44 -0800)]
kernel/smp.c: fix smp_call_function_many() SMP race

commit 6dc19899958e420a931274b94019e267e2396d3e upstream.

I noticed a failure where we hit the following WARN_ON in

                if (!cpumask_test_and_clear_cpu(cpu, data->cpumask))


                refs = atomic_dec_return(&data->refs);
                WARN_ON(refs < 0);      <-------------------------

We atomically tested and cleared our bit in the cpumask, and yet the
number of cpus left (ie refs) was 0.  How can this be?

It turns out commit 54fdade1c3332391948ec43530c02c4794a38172
("generic-ipi: make struct call_function_data lockless") is at fault.  It
removes locking from smp_call_function_many and in doing so creates a
rather complicated race.

The problem comes about because:

 - The smp_call_function_many interrupt handler walks call_function.queue
   without any locking.
 - We reuse a percpu data structure in smp_call_function_many.
 - We do not wait for any RCU grace period before starting the next

Imagine a scenario where CPU A does two smp_call_functions back to back,
and CPU B does an smp_call_function in between.  We concentrate on how CPU
C handles the calls:

CPU A            CPU B                  CPU C              CPU D

call_function.queue sees
data from CPU A on list



                                        call_function.queue sees
                                          (stale) CPU A on list
   smp_call_function int
   clears last ref on A
   list_del_rcu, unlock
smp_call_function reuses
percpu *data A
                                         data->cpumask sees and
                                         clears bit in cpumask
                                         might be using old or new fn!
                                         decrements refs below 0

set data->refs (too late!)

The important thing to note is since the interrupt handler walks a
potentially stale call_function.queue without any locking, then another
cpu can view the percpu *data structure at any time, even when the owner
is in the process of initialising it.

The following test case hits the WARN_ON 100% of the time on my PowerPC
box (having 128 threads does help :)

#include <linux/module.h>
#include <linux/init.h>

#define ITERATIONS 100

static void do_nothing_ipi(void *dummy)

static void do_ipis(struct work_struct *dummy)
int i;

for (i = 0; i < ITERATIONS; i++)
smp_call_function(do_nothing_ipi, NULL, 1);

printk(KERN_DEBUG "cpu %d finished\n", smp_processor_id());

static struct work_struct work[NR_CPUS];

static int __init testcase_init(void)
int cpu;

for_each_online_cpu(cpu) {
INIT_WORK(&work[cpu], do_ipis);
schedule_work_on(cpu, &work[cpu]);

return 0;

static void __exit testcase_exit(void)

MODULE_AUTHOR("Anton Blanchard");

I tried to fix it by ordering the read and the write of ->cpumask and
->refs.  In doing so I missed a critical case but Paul McKenney was able
to spot my bug thankfully :) To ensure we arent viewing previous
iterations the interrupt handler needs to read ->refs then ->cpumask then
->refs _again_.

Thanks to Milton Miller and Paul McKenney for helping to debug this issue.

[ add WARN_ON and BUG_ON, remove extra read of refs before initial read of mask that doesn't help (also noted by Peter Zijlstra), adjust comments, hopefully clarify scenario ]
[ remove excess tests]
Signed-off-by: Anton Blanchard <>
Signed-off-by: Milton Miller <>
Cc: Ingo Molnar <>
Cc: "Paul E. McKenney" <>
Signed-off-by: Andrew Morton <>
Signed-off-by: Linus Torvalds <>
Signed-off-by: Greg Kroah-Hartman <>
7 years agofs/direct-io.c: don't try to allocate more than BIO_MAX_PAGES in a bio
David Dillow [Thu, 20 Jan 2011 22:44:22 +0000 (14:44 -0800)]
fs/direct-io.c: don't try to allocate more than BIO_MAX_PAGES in a bio

commit 20d9600cb407b0b55fef6ee814b60345c6f58264 upstream.

When using devices that support max_segments > BIO_MAX_PAGES (256), direct
IO tries to allocate a bio with more pages than allowed, which leads to an
oops in dio_bio_alloc().  Clamp the request to the supported maximum, and
change dio_bio_alloc() to reflect that bio_alloc() will always return a
bio when called with __GFP_WAIT and a valid number of vectors.

[ remove redundant BUG_ON()]
Signed-off-by: David Dillow <>
Reviewed-by: Jeff Moyer <>
Signed-off-by: Andrew Morton <>
Signed-off-by: Linus Torvalds <>
Signed-off-by: Greg Kroah-Hartman <>
7 years agobacklight: fix 88pm860x_bl macro collision
Randy Dunlap [Thu, 20 Jan 2011 22:44:31 +0000 (14:44 -0800)]
backlight: fix 88pm860x_bl macro collision

commit 2550326ac7a062fdfc204f9a3b98bdb9179638fc upstream.

Fix collision with kernel-supplied #define:

  drivers/video/backlight/88pm860x_bl.c:24:1: warning: "CURRENT_MASK" redefined
  arch/x86/include/asm/page_64_types.h:6:1: warning: this is the location of the previous definition

Signed-off-by: Randy Dunlap <>
Cc: Haojian Zhuang <>
Cc: Richard Purdie <>
Signed-off-by: Andrew Morton <>
Signed-off-by: Linus Torvalds <>
Signed-off-by: Greg Kroah-Hartman <>
7 years agossb-pcmcia: Fix parsing of invariants tuples
Michael Büsch [Fri, 4 Feb 2011 22:34:45 +0000 (23:34 +0100)]
ssb-pcmcia: Fix parsing of invariants tuples

commit dd3cb633078fb12e06ce6cebbdfbf55a7562e929 upstream.

This fixes parsing of the device invariants (MAC address)
for PCMCIA SSB devices.

ssb_pcmcia_do_get_invariants expects an iv pointer as data

Tested-by: dylan cristiani <>
Signed-off-by: Michael Buesch <>
Signed-off-by: John W. Linville <>
Signed-off-by: Greg Kroah-Hartman <>
7 years agochar/ipmi: fix OOPS caused by pnp_unregister_driver on unregistered driver
Corey Minyard [Thu, 10 Feb 2011 22:08:38 +0000 (16:08 -0600)]
char/ipmi: fix OOPS caused by pnp_unregister_driver on unregistered driver

commit d2478521afc20227658a10a8c5c2bf1a2aa615b3 upstream.

This patch fixes an OOPS triggered when calling modprobe ipmi_si a
second time after the first modprobe returned without finding any ipmi
devices.  This can happen if you reload the module after having the
first module load fail.  The driver was not deregistering from PNP in
that case.

Peter Huewe originally reported this patch and supplied a fix, I have a
different patch based on Linus' suggestion that cleans things up a bit

Reviewed-by: Peter Huewe <>
Cc: Randy Dunlap <>
Signed-off-by: Corey Minyard <>
Signed-off-by: Linus Torvalds <>
Signed-off-by: Greg Kroah-Hartman <>
7 years agowatchdog: Don't change watchdog state on read of sysctl
Marcin Slusarz [Fri, 28 Jan 2011 16:00:33 +0000 (11:00 -0500)]
watchdog: Don't change watchdog state on read of sysctl

commit 9ffdc6c37df131f89d52001e0ef03091b158826f upstream.

Signed-off-by: Marcin Slusarz <>
[ add {}'s to fix a warning ]
Signed-off-by: Don Zickus <>
Cc: Peter Zijlstra <>
Cc: Frederic Weisbecker <>
LKML-Reference: <>
Signed-off-by: Ingo Molnar <>
Signed-off-by: Greg Kroah-Hartman <>
7 years agowatchdog: Fix sysctl consistency
Marcin Slusarz [Fri, 28 Jan 2011 16:00:32 +0000 (11:00 -0500)]
watchdog: Fix sysctl consistency

commit 397357666de6b5b6adb5fa99f9758ec8cf30ac34 upstream.

If it was not possible to enable watchdog for any cpu, switch
watchdog_enabled back to 0, because it's visible via
kernel.watchdog sysctl.

Signed-off-by: Marcin Slusarz <>
Signed-off-by: Don Zickus <>
Cc: Peter Zijlstra <>
Cc: Frederic Weisbecker <>
LKML-Reference: <>
Signed-off-by: Ingo Molnar <>
Signed-off-by: Greg Kroah-Hartman <>
7 years agoparisc : Remove broken line wrapping handling pdc_iodc_print()
Guy Martin [Mon, 6 Dec 2010 15:48:04 +0000 (16:48 +0100)]
parisc : Remove broken line wrapping handling pdc_iodc_print()

commit fbea668498e93bb38ac9226c7af9120a25957375 upstream.

Remove the broken line wrapping handling in pdc_iodc_print().
It is broken in 3 ways :
  - It doesn't keep track of the current screen position, it just
    assumes that the new buffer will be printed at the begining of the
  - It doesn't take in account that non printable characters won't
    increase the current position on the screen.
  - And last but not least, it triggers a kernel panic if a backspace
    is the first char in the provided buffer :

  [<0000000040128ec4>] pdc_console_write+0x44/0x78
  [<0000000040128f18>] pdc_console_tty_write+0x20/0x38
  [<000000004032f1ac>] n_tty_write+0x2a4/0x550
  [<000000004032b158>] tty_write+0x1e0/0x2d8
  [<00000000401bb420>] vfs_write+0xb8/0x188
  [<00000000401bb630>] sys_write+0x68/0xb8
  [<0000000040104eb8>] syscall_exit+0x0/0x14

Most terminals handle the line wrapping just fine. I've confirmed that
it works correctly on a C8000 with both vga and serial output.

Signed-off-by: Guy Martin <>
Signed-off-by: James Bottomley <>
Signed-off-by: Greg Kroah-Hartman <>
7 years agofirewire: core: fix unstable I/O with Canon camcorder
Stefan Richter [Sat, 15 Jan 2011 17:19:48 +0000 (18:19 +0100)]
firewire: core: fix unstable I/O with Canon camcorder

commit 6044565af458e7fa6e748bff437ecc49dea88d79 upstream.

Regression since commit 10389536742c, "firewire: core: check for 1394a
compliant IRM, fix inaccessibility of Sony camcorder":

The camcorder Canon MV5i generates lots of bus resets when asynchronous
requests are sent to it (e.g. Config ROM read requests or FCP Command
write requests) if the camcorder is not root node.  This causes drop-
outs in videos or makes the camcorder entirely inaccessible.

Fix this by allowing any Canon device, even if it is a pre-1394a IRM
like MV5i are, to remain root node (if it is at least Cycle Master
capable).  With the FireWire controller cards that I tested, MV5i always
becomes root node when plugged in and left to its own devices.

Reported-by: Ralf Lange
Signed-off-by: Stefan Richter <>
Signed-off-by: Greg Kroah-Hartman <>
7 years agopowerpc: Fix some 6xx/7xxx CPU setup functions
Benjamin Herrenschmidt [Thu, 20 Jan 2011 20:35:23 +0000 (20:35 +0000)]
powerpc: Fix some 6xx/7xxx CPU setup functions

commit 1f1936ff3febf38d582177ea319eaa278f32c91f upstream.

Some of those functions try to adjust the CPU features, for example
to remove NAP support on some revisions. However, they seem to use
r5 as an index into the CPU table entry, which might have been right
a long time ago but no longer is. r4 is the right register to use.

This probably caused some off behaviours on some PowerMac variants
using 750cx or 7455 processor revisions.

Signed-off-by: Benjamin Herrenschmidt <>
Signed-off-by: Greg Kroah-Hartman <>
7 years agopowerpc/numa: Fix bug in unmap_cpu_from_node
Anton Blanchard [Sat, 29 Jan 2011 12:37:16 +0000 (12:37 +0000)]
powerpc/numa: Fix bug in unmap_cpu_from_node

commit 429f4d8d20b91e4a6c239f951c06a56a6ac22957 upstream.

When converting to the new cpumask code I screwed up:

-       if (cpu_isset(cpu, numa_cpumask_lookup_table[node])) {
-               cpu_clear(cpu, numa_cpumask_lookup_table[node]);
+       if (cpumask_test_cpu(cpu, node_to_cpumask_map[node])) {
+               cpumask_set_cpu(cpu, node_to_cpumask_map[node]);

This was introduced in commit 25863de07af9 (powerpc/cpumask: Convert NUMA code
to new cpumask API)

Fix it.

Signed-off-by: Anton Blanchard <>
Signed-off-by: Benjamin Herrenschmidt <>
Signed-off-by: Greg Kroah-Hartman <>
7 years agopowerpc: Fix hcall tracepoint recursion
Anton Blanchard [Thu, 21 Oct 2010 00:52:12 +0000 (00:52 +0000)]
powerpc: Fix hcall tracepoint recursion

commit 57cdfdf829a850a317425ed93c6a576c9ee6329c upstream.

Spinlocks on shared processor partitions use H_YIELD to notify the
hypervisor we are waiting on another virtual CPU. Unfortunately this means
the hcall tracepoints can recurse.

The patch below adds a percpu depth and checks it on both the entry and
exit hcall tracepoints.

Signed-off-by: Anton Blanchard <>
Acked-by: Steven Rostedt <>
Signed-off-by: Benjamin Herrenschmidt <>
Signed-off-by: Greg Kroah-Hartman <>
7 years agopowerpc/85xx: fix compatible properties of the P1022DS DMA nodes used for audio
Timur Tabi [Fri, 3 Dec 2010 16:52:14 +0000 (10:52 -0600)]
powerpc/85xx: fix compatible properties of the P1022DS DMA nodes used for audio

commit b2e0861e51f2961954330dcafe6d148ee3ab5cff upstream.

In order to prevent the fsl_dma driver from claiming the DMA channels that the
P1022DS audio driver needs, the compatible properties for those nodes must say
"fsl,ssi-dma-channel" instead of "fsl,eloplus-dma-channel".

Signed-off-by: Timur Tabi <>
Signed-off-by: Kumar Gala <>
Signed-off-by: Greg Kroah-Hartman <>
7 years agocfq-iosched: Don't wait if queue already has requests.
Justin TerAvest [Wed, 9 Feb 2011 13:20:03 +0000 (14:20 +0100)]
cfq-iosched: Don't wait if queue already has requests.

commit 02a8f01b5a9f396d0327977af4c232d0f94c45fd upstream.

Commit 7667aa0630407bc07dc38dcc79d29cc0a65553c1 added logic to wait for
the last queue of the group to become busy (have at least one request),
so that the group does not lose out for not being continuously
backlogged. The commit did not check for the condition that the last
queue already has some requests. As a result, if the queue already has
requests, wait_busy is set. Later on, cfq_select_queue() checks the
flag, and decides that since the queue has a request now and wait_busy
is set, the queue is expired.  This results in early expiration of the

This patch fixes the problem by adding a check to see if queue already
has requests. If it does, wait_busy is not set. As a result, time slices
do not expire early.

The queues with more than one request are usually buffered writers.
Testing shows improvement in isolation between buffered writers.

Signed-off-by: Justin TerAvest <>
Reviewed-by: Gui Jianfeng <>
Acked-by: Vivek Goyal <>
Signed-off-by: Jens Axboe <>
Signed-off-by: Greg Kroah-Hartman <>
7 years agoklist: Fix object alignment on 64-bit.
David Miller [Mon, 14 Feb 2011 00:37:07 +0000 (16:37 -0800)]
klist: Fix object alignment on 64-bit.

commit 795abaf1e4e188c4171e3cd3dbb11a9fcacaf505 upstream.

Commit c0e69a5bbc6f ("klist.c: bit 0 in pointer can't be used as flag")
intended to make sure that all klist objects were at least pointer size
aligned, but used the constant "4" which only works on 32-bit.

Use "sizeof(void *)" which is correct in all cases.

Signed-off-by: David S. Miller <>
Acked-by: Jesper Nilsson <>
Signed-off-by: Linus Torvalds <>
Signed-off-by: Greg Kroah-Hartman <>
7 years agomm: page allocator: adjust the per-cpu counter threshold when memory is low
Mel Gorman [Thu, 13 Jan 2011 23:45:41 +0000 (15:45 -0800)]
mm: page allocator: adjust the per-cpu counter threshold when memory is low

commit 88f5acf88ae6a9778f6d25d0d5d7ec2d57764a97 upstream.

Commit aa45484 ("calculate a better estimate of NR_FREE_PAGES when memory
is low") noted that watermarks were based on the vmstat NR_FREE_PAGES.  To
avoid synchronization overhead, these counters are maintained on a per-cpu
basis and drained both periodically and when a threshold is above a
threshold.  On large CPU systems, the difference between the estimate and
real value of NR_FREE_PAGES can be very high.  The system can get into a
case where pages are allocated far below the min watermark potentially
causing livelock issues.  The commit solved the problem by taking a better
reading of NR_FREE_PAGES when memory was low.

Unfortately, as reported by Shaohua Li this accurate reading can consume a
large amount of CPU time on systems with many sockets due to cache line
bouncing.  This patch takes a different approach.  For large machines
where counter drift might be unsafe and while kswapd is awake, the per-cpu
thresholds for the target pgdat are reduced to limit the level of drift to
what should be a safe level.  This incurs a performance penalty in heavy
memory pressure by a factor that depends on the workload and the machine
but the machine should function correctly without accidentally exhausting
all memory on a node.  There is an additional cost when kswapd wakes and
sleeps but the event is not expected to be frequent - in Shaohua's test
case, there was one recorded sleep and wake event at least.

To ensure that kswapd wakes up, a safe version of zone_watermark_ok() is
introduced that takes a more accurate reading of NR_FREE_PAGES when called
from wakeup_kswapd, when deciding whether it is really safe to go back to
sleep in sleeping_prematurely() and when deciding if a zone is really
balanced or not in balance_pgdat().  We are still using an expensive
function but limiting how often it is called.

When the test case is reproduced, the time spent in the watermark
functions is reduced.  The following report is on the percentage of time
spent cumulatively spent in the functions zone_nr_free_pages(),
zone_watermark_ok(), __zone_watermark_ok(), zone_watermark_ok_safe(),
zone_page_state_snapshot(), zone_page_state().

vanilla                      11.6615%
disable-threshold            0.2584%

David said:

: We had to pull aa454840 "mm: page allocator: calculate a better estimate
: of NR_FREE_PAGES when memory is low and kswapd is awake" from 2.6.36
: internally because tests showed that it would cause the machine to stall
: as the result of heavy kswapd activity.  I merged it back with this fix as
: it is pending in the -mm tree and it solves the issue we were seeing, so I
: definitely think this should be pushed to -stable (and I would seriously
: consider it for 2.6.37 inclusion even at this late date).

Signed-off-by: Mel Gorman <>
Reported-by: Shaohua Li <>
Reviewed-by: Christoph Lameter <>
Tested-by: Nicolas Bareil <>
Cc: David Rientjes <>
Cc: Kyle McMartin <>
Signed-off-by: Andrew Morton <>
Signed-off-by: Linus Torvalds <>
Signed-off-by: Greg Kroah-Hartman <>
7 years agommc: bfin_sdh: fix alloc size for private data
Sonic Zhang [Wed, 12 Jan 2011 03:39:35 +0000 (22:39 -0500)]
mmc: bfin_sdh: fix alloc size for private data

commit a34650f0f1ca589cda09c48cb62baf15e680a247 upstream.

The bfin_sdh driver allocates the wrong size for the private data
in the mmc_host.  The first parameter of mmc_alloc_host should be
the size of the local driver struct rather than the common mmc_host.

Signed-off-by: Sonic Zhang <>
Signed-off-by: Mike Frysinger <>
Signed-off-by: Chris Ball <>
Signed-off-by: Greg Kroah-Hartman <>
7 years agomemcg: fix account leak at failure of memsw acconting
KAMEZAWA Hiroyuki [Tue, 25 Jan 2011 23:07:27 +0000 (15:07 -0800)]
memcg: fix account leak at failure of memsw acconting

commit 01c88e2d6b7330c0cc5867fe2297e7d826e1337d upstream.

Commit 4b53433468 ("memcg: clean up try_charge main loop") removes a
cancel of charge at case: memory charge-> success.  mem+swap charge->

This leaks usage of memory.  Fix it.

Signed-off-by: KAMEZAWA Hiroyuki <>
Reviewed-by: Johannes Weiner <>
Acked-by: Daisuke Nishimura <>
Cc: Balbir Singh <>
Signed-off-by: Andrew Morton <>
Signed-off-by: Linus Torvalds <>
Signed-off-by: Greg Kroah-Hartman <>
7 years agoARM: initrd: disable initrd if passed address overlaps reserved region
Russell King [Sun, 30 Jan 2011 11:21:05 +0000 (11:21 +0000)]
ARM: initrd: disable initrd if passed address overlaps reserved region

commit b0a2679d27408d97ce31e5f800b44227d3388b84 upstream.

Disable the initrd if the passed address already overlaps the reserved
region.  This avoids oopses on Netwinders when NeTTrom tells the kernel
that an initrd is located at mem+4MB, but this overlaps the BSS,
resulting in the kernels in-use BSS being freed.

This should be applied to v2.6.37-stable.

Signed-off-by: Russell King <>
Signed-off-by: Greg Kroah-Hartman <>
7 years agoFix prlimit64 for suid/sgid processes
Kacper Kornet [Fri, 28 Jan 2011 23:21:04 +0000 (00:21 +0100)]
Fix prlimit64 for suid/sgid processes

commit aa5bd67dcfdf9af34c7fa36ebc87d4e1f7e91873 upstream.

Since check_prlimit_permission always fails in the case of SUID/GUID
processes, such processes are not able to read or set their own limits.
This commit changes this by assuming that process can always read/change
its own limits.

Signed-off-by: Kacper Kornet <>
Acked-by: Jiri Slaby <>
Signed-off-by: Linus Torvalds <>
Signed-off-by: Greg Kroah-Hartman <>
7 years agoixgbe: fix for 82599 erratum on Header Splitting
Don Skidmore [Tue, 18 Jan 2011 22:53:47 +0000 (22:53 +0000)]
ixgbe: fix for 82599 erratum on Header Splitting

commit a124339ad28389093ed15eca990d39c51c5736cc upstream.

We have found a hardware erratum on 82599 hardware that can lead to
unpredictable behavior when Header Splitting mode is enabled.  So
we are no longer enabling this feature on affected hardware.

Please see the 82599 Specification Update for more information.

Signed-off-by: Don Skidmore <>
Tested-by: Stephen Ko <>
Signed-off-by: Jeff Kirsher <>
Signed-off-by: Greg Kroah-Hartman <>
7 years agofix jiffy calculations in calibrate_delay_direct to handle overflow
Tim Deegan [Thu, 10 Feb 2011 08:50:41 +0000 (08:50 +0000)]
fix jiffy calculations in calibrate_delay_direct to handle overflow

commit 70a062286b9dfcbd24d2e11601aecfead5cf709a upstream.

Fixes a hang when booting as dom0 under Xen, when jiffies can be
quite large by the time the kernel init gets this far.

Signed-off-by: Tim Deegan <>
[ !time_after() -> time_before_eq() as suggested by Jiri Slaby]
Signed-off-by: Jan Beulich <>
Cc: Jiri Slaby <>
Cc: Jeremy Fitzhardinge <>
Signed-off-by: Linus Torvalds <>
Signed-off-by: Greg Kroah-Hartman <>
7 years agox86, mtrr: Avoid MTRR reprogramming on BP during boot on UP platforms
Suresh Siddha [Thu, 3 Feb 2011 01:02:55 +0000 (17:02 -0800)]
x86, mtrr: Avoid MTRR reprogramming on BP during boot on UP platforms

commit f7448548a9f32db38f243ccd4271617758ddfe2c upstream.

Markus Kohn ran into a hard hang regression on an acer aspire
1310, when acpi is enabled. git bisect showed the following
commit as the bad one that introduced the boot regression.

commit d0af9eed5aa91b6b7b5049cae69e5ea956fd85c3
Author: Suresh Siddha <>
Date:   Wed Aug 19 18:05:36 2009 -0700

    x86, pat/mtrr: Rendezvous all the cpus for MTRR/PAT init

Because of the UP configuration of that platform,
native_smp_prepare_cpus() bailed out (in smp_sanity_check())
before doing the set_mtrr_aps_delayed_init()

Further down the boot path, native_smp_cpus_done() will call the
delayed MTRR initialization for the AP's (mtrr_aps_init()) with
mtrr_aps_delayed_init not set. This resulted in the boot
processor reprogramming its MTRR's to the values seen during the
start of the OS boot. While this is not needed ideally, this
shouldn't have caused any side-effects. This is because the
reprogramming of MTRR's (set_mtrr_state() that gets called via
set_mtrr()) will check if the live register contents are
different from what is being asked to write and will do the actual
write only if they are different.

BP's mtrr state is read during the start of the OS boot and
typically nothing would have changed when we ask to reprogram it
on BP again because of the above scenario on an UP platform. So
on a normal UP platform no reprogramming of BP MTRR MSR's
happens and all is well.

However, on this platform, bios seems to be modifying the fixed
mtrr range registers between the start of OS boot and when we
double check the live registers for reprogramming BP MTRR
registers. And as the live registers are modified, we end up
reprogramming the MTRR's to the state seen during the start of
the OS boot.

During ACPI initialization, something in the bios (probably smi
handler?) don't like this fact and results in a hard lockup.

We didn't see this boot hang issue on this platform before the
commit d0af9eed5aa91b6b7b5049cae69e5ea956fd85c3, because only
the AP's (if any) will program its MTRR's to the value that BP
had at the start of the OS boot.

Fix this issue by checking mtrr_aps_delayed_init before
continuing further in the mtrr_aps_init(). Now, only AP's (if
any) will program its MTRR's to the BP values during boot.


  [ By the way, this behavior of the bios modifying MTRR's after the start
    of the OS boot is not common and the kernel is not prepared to
    handle this situation well. Irrespective of this issue, during
    suspend/resume, linux kernel will try to reprogram the BP's MTRR values
    to the values seen during the start of the OS boot. So suspend/resume might
    be already broken on this platform for all linux kernel versions. ]

Reported-and-bisected-by: Markus Kohn <>
Tested-by: Markus Kohn <>
Signed-off-by: Suresh Siddha <>
Cc: Thomas Renninger <>
Cc: Rafael Wysocki <>
Cc: Venkatesh Pallipadi <>
LKML-Reference: <>
Signed-off-by: Ingo Molnar <>
Signed-off-by: Greg Kroah-Hartman <>
7 years agonet: Fix ip link add netns oops
Eric W. Biederman [Sat, 29 Jan 2011 14:57:22 +0000 (14:57 +0000)]
net: Fix ip link add netns oops

commit 13ad17745c2cbd437d9e24b2d97393e0be11c439 upstream.

Ed Swierk <> writes:
> On
>  ip link add link eth0 netns 9999 type macvlan
> where 9999 is a nonexistent PID triggers an oops and causes all network functions to hang:
> [10663.821898] BUG: unable to handle kernel NULL pointer dereference at 000000000000006d
>  [10663.821917] IP: [<ffffffff8149c2fa>] __dev_alloc_name+0x9a/0x170
>  [10663.821933] PGD 1d3927067 PUD 22f5c5067 PMD 0
>  [10663.821944] Oops: 0000 [#1] SMP
>  [10663.821953] last sysfs file: /sys/devices/system/cpu/cpu0/cpufreq/scaling_cur_freq
>  [10663.821959] CPU 3
>  [10663.821963] Modules linked in: macvlan ip6table_filter ip6_tables rfcomm ipt_MASQUERADE binfmt_misc iptable_nat nf_nat nf_conntrack_ipv4 nf_defrag_ipv4 xt_state nf_conntrack sco ipt_REJECT bnep l2cap xt_tcpudp iptable_filter ip_tables x_tables bridge stp vboxnetadp vboxnetflt vboxdrv kvm_intel kvm parport_pc ppdev snd_hda_codec_intelhdmi snd_hda_codec_conexant arc4 iwlagn iwlcore mac80211 snd_hda_intel snd_hda_codec snd_hwdep snd_pcm snd_seq_midi snd_rawmidi i915 snd_seq_midi_event snd_seq thinkpad_acpi drm_kms_helper btusb tpm_tis nvram uvcvideo snd_timer snd_seq_device bluetooth videodev v4l1_compat v4l2_compat_ioctl32 tpm drm tpm_bios snd cfg80211 psmouse serio_raw intel_ips soundcore snd_page_alloc intel_agp i2c_algo_bit video output netconsole configfs lp parport usbhid hid e1000e sdhci_pci ahci libahci sdhci led_class
>  [10663.822155]
>  [10663.822161] Pid: 6000, comm: ip Not tainted 2.6.35-23-generic #41-Ubuntu 2901CTO/2901CTO
>  [10663.822167] RIP: 0010:[<ffffffff8149c2fa>] [<ffffffff8149c2fa>] __dev_alloc_name+0x9a/0x170
>  [10663.822177] RSP: 0018:ffff88014aebf7b8 EFLAGS: 00010286
>  [10663.822182] RAX: 00000000fffffff4 RBX: ffff8801ad900800 RCX: 0000000000000000
>  [10663.822187] RDX: ffff880000000000 RSI: 0000000000000000 RDI: ffff88014ad63000
>  [10663.822191] RBP: ffff88014aebf808 R08: 0000000000000041 R09: 0000000000000041
>  [10663.822196] R10: 0000000000000000 R11: dead000000200200 R12: ffff88014aebf818
>  [10663.822201] R13: fffffffffffffffd R14: ffff88014aebf918 R15: ffff88014ad62000
>  [10663.822207] FS: 00007f00c487f700(0000) GS:ffff880001f80000(0000) knlGS:0000000000000000
>  [10663.822212] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
>  [10663.822216] CR2: 000000000000006d CR3: 0000000231f19000 CR4: 00000000000026e0
>  [10663.822221] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
>  [10663.822226] DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400
>  [10663.822231] Process ip (pid: 6000, threadinfo ffff88014aebe000, task ffff88014afb16e0)
>  [10663.822236] Stack:
>  [10663.822240] ffff88014aebf808 ffffffff814a2bb5 ffff88014aebf7e8 00000000a00ee8d6
>  [10663.822251] <0> 0000000000000000 ffffffffa00ef940 ffff8801ad900800 ffff88014aebf818
>  [10663.822265] <0> ffff88014aebf918 ffff8801ad900800 ffff88014aebf858 ffffffff8149c413
>  [10663.822281] Call Trace:
>  [10663.822290] [<ffffffff814a2bb5>] ? dev_addr_init+0x75/0xb0
>  [10663.822298] [<ffffffff8149c413>] dev_alloc_name+0x43/0x90
>  [10663.822307] [<ffffffff814a85ee>] rtnl_create_link+0xbe/0x1b0
>  [10663.822314] [<ffffffff814ab2aa>] rtnl_newlink+0x48a/0x570
>  [10663.822321] [<ffffffff814aafcc>] ? rtnl_newlink+0x1ac/0x570
>  [10663.822332] [<ffffffff81030064>] ? native_x2apic_icr_read+0x4/0x20
>  [10663.822339] [<ffffffff814a8c17>] rtnetlink_rcv_msg+0x177/0x290
>  [10663.822346] [<ffffffff814a8aa0>] ? rtnetlink_rcv_msg+0x0/0x290
>  [10663.822354] [<ffffffff814c25d9>] netlink_rcv_skb+0xa9/0xd0
>  [10663.822360] [<ffffffff814a8a85>] rtnetlink_rcv+0x25/0x40
>  [10663.822367] [<ffffffff814c223e>] netlink_unicast+0x2de/0x2f0
>  [10663.822374] [<ffffffff814c303e>] netlink_sendmsg+0x1fe/0x2e0
>  [10663.822383] [<ffffffff81488533>] sock_sendmsg+0xf3/0x120
>  [10663.822391] [<ffffffff815899fe>] ? _raw_spin_lock+0xe/0x20
>  [10663.822400] [<ffffffff81168656>] ? __d_lookup+0x136/0x150
>  [10663.822406] [<ffffffff815899fe>] ? _raw_spin_lock+0xe/0x20
>  [10663.822414] [<ffffffff812b7a0d>] ? _atomic_dec_and_lock+0x4d/0x80
>  [10663.822422] [<ffffffff8116ea90>] ? mntput_no_expire+0x30/0x110
>  [10663.822429] [<ffffffff81486ff5>] ? move_addr_to_kernel+0x65/0x70
>  [10663.822435] [<ffffffff81493308>] ? verify_iovec+0x88/0xe0
>  [10663.822442] [<ffffffff81489020>] sys_sendmsg+0x240/0x3a0
> [10663.822450] [<ffffffff8111e2a9>] ? __do_fault+0x479/0x560
>  [10663.822457] [<ffffffff815899fe>] ? _raw_spin_lock+0xe/0x20
>  [10663.822465] [<ffffffff8116cf4a>] ? alloc_fd+0x10a/0x150
>  [10663.822473] [<ffffffff8158d76e>] ? do_page_fault+0x15e/0x350
>  [10663.822482] [<ffffffff8100a0f2>] system_call_fastpath+0x16/0x1b
>  [10663.822487] Code: 90 48 8d 78 02 be 25 00 00 00 e8 92 1d e2 ff 48 85 c0 75 cf bf 20 00 00 00 e8 c3 b1 c6 ff 49 89 c7 b8 f4 ff ff ff 4d 85 ff 74 bd <4d> 8b 75 70 49 8d 45 70 48 89 45 b8 49 83 ee 58 eb 28 48 8d 55
>  [10663.822618] RIP [<ffffffff8149c2fa>] __dev_alloc_name+0x9a/0x170
>  [10663.822627] RSP <ffff88014aebf7b8>
>  [10663.822631] CR2: 000000000000006d
>  [10663.822636] ---[ end trace 3dfd6c3ad5327ca7 ]---

This bug was introduced in:
commit 81adee47dfb608df3ad0b91d230fb3cef75f0060
Author: Eric W. Biederman <>
Date:   Sun Nov 8 00:53:51 2009 -0800

    net: Support specifying the network namespace upon device creation.

    There is no good reason to not support userspace specifying the
    network namespace during device creation, and it makes it easier
    to create a network device and pass it to a child network namespace
    with a well known name.

    We have to be careful to ensure that the target network namespace
    for the new device exists through the life of the call.  To keep
    that logic clear I have factored out the network namespace grabbing
    logic into rtnl_link_get_net.

    In addtion we need to continue to pass the source network namespace
    to the rtnl_link_ops.newlink method so that we can find the base
    device source network namespace.

Signed-off-by: Eric W. Biederman <>
Acked-by: Eric Dumazet <>
Where apparently I forgot to add error handling to the path where we create
a new network device in a new network namespace, and pass in an invalid pid.

Reported-by: Ed Swierk <>
Signed-off-by: "Eric W. Biederman" <>
Signed-off-by: David S. Miller <>
Signed-off-by: Greg Kroah-Hartman <>
7 years agoptrace: use safer wake up on ptrace_detach()
Tejun Heo [Thu, 10 Feb 2011 23:01:22 +0000 (15:01 -0800)]
ptrace: use safer wake up on ptrace_detach()

commit 01e05e9a90b8f4c3997ae0537e87720eb475e532 upstream.

The wake_up_process() call in ptrace_detach() is spurious and not
interlocked with the tracee state.  IOW, the tracee could be running or
sleeping in any place in the kernel by the time wake_up_process() is
called.  This can lead to the tracee waking up unexpectedly which can be

The wake_up is spurious and should be removed but for now reduce its
toxicity by only waking up if the tracee is in TRACED or STOPPED state.

This bug can possibly be used as an attack vector.  I don't think it
will take too much effort to come up with an attack which triggers oops
somewhere.  Most sleeps are wrapped in condition test loops and should
be safe but we have quite a number of places where sleep and wakeup
conditions are expected to be interlocked.  Although the window of
opportunity is tiny, ptrace can be used by non-privileged users and with
some loading the window can definitely be extended and exploited.

Signed-off-by: Tejun Heo <>
Acked-by: Roland McGrath <>
Acked-by: Oleg Nesterov <>
Signed-off-by: Andrew Morton <>
Signed-off-by: Linus Torvalds <>
Signed-off-by: Greg Kroah-Hartman <>
7 years agoserial: unbreak billionton CF card
Pavel Machek [Sun, 9 Jan 2011 07:38:48 +0000 (08:38 +0100)]
serial: unbreak billionton CF card

commit d0694e2aeb815042aa0f3e5036728b3db4446f1d upstream.

Unbreak Billionton CF bluetooth card. This actually fixes a regression
on zaurus.

Signed-off-by: Pavel Machek <>
Signed-off-by: Greg Kroah-Hartman <>
7 years agodrivers: update to pl2303 usb-serial to support Motorola cables
Dario Lombardo [Fri, 21 Jan 2011 14:35:19 +0000 (15:35 +0100)]
drivers: update to pl2303 usb-serial to support Motorola cables

commit 96a3e79edff6f41b0f115a82f1a39d66218077a7 upstream.

Added 0x0307 device id to support Motorola cables to the pl2303 usb
serial driver. This cable has a modified chip that is a pl2303, but
declares itself as 0307. Fixed by adding the right device id to the
supported devices list, assigning it the code labeled

Signed-off-by: Dario Lombardo <>
Signed-off-by: Greg Kroah-Hartman <>
7 years agovirtio: console: Wake up outvq on host notifications
Amit Shah [Mon, 31 Jan 2011 07:36:36 +0000 (13:06 +0530)]
virtio: console: Wake up outvq on host notifications

commit 2770c5ea501be69989a7acf54ec4cb3554c47191 upstream.

The outvq needs to be woken up on host notifications so that buffers
consumed by the host can be reclaimed, outvq freed, and application
writes may proceed again.

The need for this is now finally noticed when I have qemu patches ready
to use nonblocking IO and flow control.

CC: Hans de Goede <>
Signed-off-by: Amit Shah <>
Signed-off-by: Rusty Russell <>
Acked-by: Hans de Goede <>
Signed-off-by: Greg Kroah-Hartman <>
7 years agovirtio_net: Add schedule check to napi_enable call
Bruce Rogers [Thu, 10 Feb 2011 19:03:31 +0000 (11:03 -0800)]
virtio_net: Add schedule check to napi_enable call

commit 3e9d08ec0a68f6faf718d5a7e050fe5ca0ba004f upstream.

Under harsh testing conditions, including low memory, the guest would
stop receiving packets. With this patch applied we no longer see any
problems in the driver while performing these tests for extended periods
of time.

Make sure napi is scheduled subsequent to each napi_enable.

Signed-off-by: Bruce Rogers <>
Signed-off-by: Olaf Kirch <>
Signed-off-by: Rusty Russell <>
Signed-off-by: David S. Miller <>
Signed-off-by: Greg Kroah-Hartman <>
7 years agoASoC: Create an AIF1ADCDAT signal widget to match AIF2
Mark Brown [Thu, 3 Feb 2011 16:27:34 +0000 (16:27 +0000)]
ASoC: Create an AIF1ADCDAT signal widget to match AIF2

commit 7f94de483f4e37e14d646ad6e85a3c82f66fb487 upstream.

Due to the different routing for AIF1 and AIF2 we weren't using a
single widget to represent the ADCDAT signal. For consistency add

Signed-off-by: Mark Brown <>
Acked-by: Liam Girdwood <>
Signed-off-by: Greg Kroah-Hartman <>
7 years agoASoC: WM8994: fix wrong value in tristate function
Qiao Zhou [Wed, 19 Jan 2011 11:10:47 +0000 (19:10 +0800)]
ASoC: WM8994: fix wrong value in tristate function

commit 78b3fb46753872fc79bffecc8d50355a8622b39b upstream.

fix wrong value in wm8994_set_tristate func. when updating reg bits,
it should use "value", not "reg".

Signed-off-by: Qiao Zhou <>
Acked-by: Liam Girdwood <>
Signed-off-by: Mark Brown <>
Signed-off-by: Greg Kroah-Hartman <>
7 years agoASoC: Handle low measured DC offsets for wm_hubs devices
Mark Brown [Fri, 21 Jan 2011 12:47:33 +0000 (12:47 +0000)]
ASoC: Handle low measured DC offsets for wm_hubs devices

commit 20a4e7fc7e213365ea3771d7bf1e10a6bab853be upstream.

The DC servo codes are actually signed numbers so need to be treated as

Signed-off-by: Mark Brown <>
Acked-by: Liam Girdwood <>
Signed-off-by: Greg Kroah-Hartman <>
7 years agoi2c: Unregister dummy devices last on adapter removal
Jean Delvare [Fri, 14 Jan 2011 21:03:49 +0000 (22:03 +0100)]
i2c: Unregister dummy devices last on adapter removal

commit 5219bf884b6e2b54e734ca1799b6f0014bb2b4b7 upstream.

Remove real devices first and dummy devices last. This gives device
driver which instantiated dummy devices themselves a chance to clean
them up before we do.

Signed-off-by: Jean Delvare <>
Tested-by: Hans Verkuil <>
Signed-off-by: Greg Kroah-Hartman <>
7 years agop54: fix sequence no. accounting off-by-one error
Christian Lamparter [Thu, 6 Jan 2011 22:47:52 +0000 (23:47 +0100)]
p54: fix sequence no. accounting off-by-one error

commit 3b5c5827d1f80ad8ae844a8b1183f59ddb90fe25 upstream.

P54_HDR_FLAG_DATA_OUT_SEQNR is meant to tell the
firmware that "the frame's sequence number has
already been set by the application."

Whereas IEEE80211_TX_CTL_ASSIGN_SEQ is set for
frames which lack a valid sequence number and
either the driver or firmware has to assign one.

Yup, it's the exact opposite!

Signed-off-by: Christian Lamparter <>
Signed-off-by: John W. Linville <>
Signed-off-by: Greg Kroah-Hartman <>
7 years agocpuidle: Make cpuidle_enable_device() call poll_idle_init()
Rafael J. Wysocki [Fri, 7 Jan 2011 23:29:20 +0000 (00:29 +0100)]
cpuidle: Make cpuidle_enable_device() call poll_idle_init()

commit d8c216cfa57e8a579f41729cbb88c97835d9ac8d upstream.

The following scenario is possible with the current cpuidle code and
the ACPI cpuidle driver:
(1) acpi_processor_cst_has_changed() is called,
(2) cpuidle_disable_device() is called,
(3) cpuidle_remove_state_sysfs() is called to remove the (presumably
    outdated) states info from sysfs,
(3) acpi_processor_get_power_info() is called, the first entry in the
    pr->power.states[] table is filled with zeros,
(4) acpi_processor_setup_cpuidle() is called and it doesn't fill the
    first entry in pr->power.states[],
(5) cpuidle_enable_device() is called,
(6) __cpuidle_register_device() is _not_ called, since the device has
    already been registered,
(7) Consequently, poll_idle_init() is _not_ called either,
(8) cpuidle_add_state_sysfs() is called to create the sysfs attributes
    for the new states and it uses the bogus first table entry from
    acpi_processor_get_power_info() for creating state0.

This problem is avoided if cpuidle_enable_device()
unconditionally calls poll_idle_init().

Reported-by: Len Brown <>
Signed-off-by: Rafael J. Wysocki <>
Signed-off-by: Len Brown <>
Signed-off-by: Greg Kroah-Hartman <>
7 years agoavr32: use syscall prototypes from asm-generic instead of arch
Hans-Christian Egtvedt [Wed, 8 Dec 2010 23:19:33 +0000 (00:19 +0100)]
avr32: use syscall prototypes from asm-generic instead of arch

commit 664cb7142ced8b827e92e1851d1ed2cae922f225 upstream.

This patch removes the redundant syscalls prototypes in the architecture
specific syscalls.h header file. These were identical with the ones in

Signed-off-by: Hans-Christian Egtvedt <>
Reported-by: Peter Huewe <>
Reported-by: Sven Schnelle <>
Signed-off-by: Greg Kroah-Hartman <>
7 years agods2760_battery: Fix calculation of time_to_empty_now
Sven Neumann [Fri, 12 Nov 2010 10:36:22 +0000 (11:36 +0100)]
ds2760_battery: Fix calculation of time_to_empty_now

commit 86af95039b69a90db15294eb1f9c147f1df0a8ea upstream.

A check against division by zero was modified in commit b0525b48.
Since this change time_to_empty_now is always reported as zero
while the battery is discharging and as a negative value while
the battery is charging. This is because current is negative while
the battery is discharging.

Fix the check introduced by commit b0525b48 so that time_to_empty_now
is reported correctly during discharge and as zero while charging.

Signed-off-by: Sven Neumann <>
Acked-by: Daniel Mack <>
Signed-off-by: Anton Vorontsov <>
Signed-off-by: Greg Kroah-Hartman <>
7 years agojz4740-battery: Protect against concurrent battery readings
Lars-Peter Clausen [Thu, 11 Nov 2010 18:00:52 +0000 (19:00 +0100)]
jz4740-battery: Protect against concurrent battery readings

commit 8ec98fe0b4ffdedce4c1caa9fb3d550f52ad1c6b upstream.

We can not handle more then one ADC request at a time to the battery.
The patch adds a mutex around the ADC read code to ensure this.

Signed-off-by: Lars-Peter Clausen <>
Signed-off-by: Anton Vorontsov <>
Signed-off-by: Greg Kroah-Hartman <>
7 years agovirtio: remove virtio-pci root device
Milton Miller [Fri, 7 Jan 2011 08:55:06 +0000 (02:55 -0600)]
virtio: remove virtio-pci root device

commit 8b3bb3ecf1934ac4a7005ad9017de1127e2fbd2f upstream.

We sometimes need to map between the virtio device and
the given pci device. One such use is OS installer that
gets the boot pci device from BIOS and needs to
find the relevant block device. Since it can't,
installation fails.

Instead of creating a top-level devices/virtio-pci
directory, create each device under the corresponding
pci device node.  Symlinks to all virtio-pci
devices can be found under the pci driver link in
bus/pci/drivers/virtio-pci/devices, and all virtio
devices under drivers/bus/virtio/devices.

Signed-off-by: Milton Miller <>
Signed-off-by: Rusty Russell <>
Acked-by: Michael S. Tsirkin <>
Tested-by: Michael S. Tsirkin <>
Acked-by: Gleb Natapov <>
Tested-by: "Daniel P. Berrange" <>
Signed-off-by: Greg Kroah-Hartman <>
7 years agoPCI: pci-stub: ignore zero-length id parameters
Tejun Heo [Wed, 22 Dec 2010 09:06:36 +0000 (10:06 +0100)]
PCI: pci-stub: ignore zero-length id parameters

commit 99a0fadf561e1f553c08f0a29f8b2578f55dd5f0 upstream.

pci-stub uses strsep() to separate list of ids and generates a warning
message when it fails to parse an id.  However, not specifying the
parameter results in ids set to an empty string.  strsep() happily
returns the empty string as the first token and thus triggers the
warning message spuriously.

Make the tokner ignore zero length ids.

Reported-by: Chris Wright <>
Reported-by: Prasad Joshi <>
Signed-off-by: Jesse Barnes <>
Signed-off-by: Greg Kroah-Hartman <>
7 years agosh: Fix up legacy PTEA space attribute mapping.
Paul Mundt [Tue, 11 Jan 2011 06:02:59 +0000 (15:02 +0900)]
sh: Fix up legacy PTEA space attribute mapping.

commit efb3e34b6176d30c4fe8635fa8e1beb6280cc2cd upstream.

When p3_ioremap() was converted to ioremap_prot() there was some breakage
introduced where the 29-bit segmentation logic would trap the area range
and return an identity mapping without having allowed the area
specification to force mapping through page tables. This wires up a PCC
mask for pgprot verification to work out whether to short-circuit the
identity mapping on legacy parts, restoring the previous behaviour.

Reported-by: Nobuhiro Iwamatsu <>
Signed-off-by: Paul Mundt <>
Signed-off-by: Greg Kroah-Hartman <>
7 years agorapidio: fix hang on RapidIO doorbell queue full condition
Thomas Taranowski [Thu, 13 Jan 2011 01:00:44 +0000 (17:00 -0800)]
rapidio: fix hang on RapidIO doorbell queue full condition

commit 12a4dc43911785f51a596f771ae0701b18d436f1 upstream.

In fsl_rio_dbell_handler() the code currently simply acknowledges the QFI
queue full interrupt, but does nothing to resolve the queue full
condition.  Instead, it jumps to the end of the isr.  When a queue full
condition occurs, the isr is then re-entered immediately and continually,

The fix is to just fall through and read out current doorbell entries.

Signed-off-by: Thomas Taranowski <>
Cc: Alexandre Bounine <>
Cc: Kumar Gala <>
Cc: Matt Porter <>
Cc: Li Yang <>
Cc: Thomas Moll <>
Cc: Micha Nelissen <>
Cc: Benjamin Herrenschmidt <>
Cc: Grant Likely <>
Signed-off-by: Andrew Morton <>
Signed-off-by: Linus Torvalds <>
Signed-off-by: Greg Kroah-Hartman <>
7 years agoext4: fix memory leak in ext4_free_branches
Theodore Ts'o [Mon, 10 Jan 2011 17:51:28 +0000 (12:51 -0500)]
ext4: fix memory leak in ext4_free_branches

commit 1c5b9e9065567876c2d4a7a16d78f0fed154a5bf upstream.

Commit 40389687 moved a call to ext4_forget() out of
ext4_free_branches and let ext4_free_blocks() handle calling
bforget().  But that change unfortunately did not replace the call to
ext4_forget() with brelse(), which was needed to drop the in-use count
of the indirect block's buffer head, which lead to a memory leak when
deleting files that used indirect blocks.  Fix this.

Thanks to Hugh Dickins for pointing this out.

Signed-off-by: "Theodore Ts'o" <>
Signed-off-by: Greg Kroah-Hartman <>
7 years agomm: fix migration hangs on anon_vma lock
Hugh Dickins [Thu, 13 Jan 2011 23:47:30 +0000 (15:47 -0800)]
mm: fix migration hangs on anon_vma lock

commit 1ce82b69e96c838d007f316b8347b911fdfa9842 upstream.

Increased usage of page migration in mmotm reveals that the anon_vma
locking in unmap_and_move() has been deficient since 2.6.36 (or even
earlier).  Review at the time of f18194275c39835cb84563500995e0d503a32d9a
("mm: fix hang on anon_vma->root->lock") missed the issue here: the
anon_vma to which we get a reference may already have been freed back to
its slab (it is in use when we check page_mapped, but that can change),
and so its anon_vma->root may be switched at any moment by reuse in

Perhaps we could fix that with a get_anon_vma_unless_zero(), but let's
not: just rely on page_lock_anon_vma() to do all the hard thinking for us,
then we don't need any rcu read locking over here.

In removing the rcu_unlock label: since PageAnon is a bit in
page->mapping, it's impossible for a !page->mapping page to be anon; but
insert VM_BUG_ON in case the implementation ever changes.

[ coding-style fixes]
Signed-off-by: Hugh Dickins <>
Reviewed-by: Mel Gorman <>
Reviewed-by: Rik van Riel <>
Cc: Naoya Horiguchi <>
Cc: "Jun'ichi Nomura" <>
Cc: Andi Kleen <>
Signed-off-by: Andrew Morton <>
Signed-off-by: Linus Torvalds <>
Signed-off-by: Greg Kroah-Hartman <>
7 years agortc-cmos: fix suspend/resume
Paul Fox [Thu, 13 Jan 2011 01:00:07 +0000 (17:00 -0800)]
rtc-cmos: fix suspend/resume

commit 2fb08e6ca9f00d1aedb3964983e9c8f84b36b807 upstream.

rtc-cmos was setting suspend/resume hooks at the device_driver level.
However, the platform bus code (drivers/base/platform.c) only looks for
resume hooks at the dev_pm_ops level, or within the platform_driver.

Switch rtc_cmos to use dev_pm_ops so that suspend/resume code is executed

Paul said:

: The user visible symptom in our (XO laptop) case was that rtcwake would
: fail to wake the laptop.  The RTC alarm would expire, but the wakeup
: wasn't unmasked.
: As for severity, the impact may have been reduced because if I recall
: correctly, the bug only affected platforms with CONFIG_PNP disabled.

Signed-off-by: Paul Fox <>
Signed-off-by: Daniel Drake <>
Acked-by: Rafael J. Wysocki <>
Signed-off-by: Andrew Morton <>
Signed-off-by: Linus Torvalds <>
Signed-off-by: Greg Kroah-Hartman <>
7 years ago/proc/kcore: fix seeking
Dave Anderson [Thu, 13 Jan 2011 01:00:36 +0000 (17:00 -0800)]
/proc/kcore: fix seeking

commit ceff1a770933e2ca2bf995b453dade4ec47a9878 upstream.

Commit 34aacb2920 ("procfs: Use generic_file_llseek in /proc/kcore") broke
seeking on /proc/kcore.  This changes it back to use default_llseek in
order to restore the original behavior.

The problem with generic_file_llseek is that it only allows seeks up to
inode->i_sb->s_maxbytes, which is 2GB-1 on procfs, where the memory file
offset values in the /proc/kcore PT_LOAD segments may exceed or start
beyond that offset value.

A similar revert was made for /proc/vmcore.

Signed-off-by: Dave Anderson <>
Acked-by: Frederic Weisbecker <>
Signed-off-by: Andrew Morton <>
Signed-off-by: Linus Torvalds <>
Signed-off-by: Greg Kroah-Hartman <>
7 years agoRDMA/cxgb4: Set the correct device physical function for iWARP connections
Steve Wise [Fri, 21 Jan 2011 17:00:34 +0000 (17:00 +0000)]
RDMA/cxgb4: Set the correct device physical function for iWARP connections

commit 94788657c94169171971968c9d4b6222c5e704aa upstream.

The PF passed to FW was 0, causing PCI failures in an SR-IOV environment.

Signed-off-by: Steve Wise <>
Signed-off-by: Roland Dreier <>
Signed-off-by: Greg Kroah-Hartman <>
7 years agoNFS: Fix "kernel BUG at fs/aio.c:554!"
Chuck Lever [Fri, 21 Jan 2011 15:54:57 +0000 (15:54 +0000)]
NFS: Fix "kernel BUG at fs/aio.c:554!"

commit 839f7ad6932d95f4d5ae7267b95c574714ff3d5b upstream.

Nick Piggin reports:

> I'm getting use after frees in aio code in NFS
> [ 2703.396766] Call Trace:
> [ 2703.396858]  [<ffffffff8100b057>] ? native_sched_clock+0x27/0x80
> [ 2703.396959]  [<ffffffff8108509e>] ? put_lock_stats+0xe/0x40
> [ 2703.397058]  [<ffffffff81088348>] ? lock_release_holdtime+0xa8/0x140
> [ 2703.397159]  [<ffffffff8108a2a5>] lock_acquire+0x95/0x1b0
> [ 2703.397260]  [<ffffffff811627db>] ? aio_put_req+0x2b/0x60
> [ 2703.397361]  [<ffffffff81039701>] ? get_parent_ip+0x11/0x50
> [ 2703.397464]  [<ffffffff81612a31>] _raw_spin_lock_irq+0x41/0x80
> [ 2703.397564]  [<ffffffff811627db>] ? aio_put_req+0x2b/0x60
> [ 2703.397662]  [<ffffffff811627db>] aio_put_req+0x2b/0x60
> [ 2703.397761]  [<ffffffff811647fe>] do_io_submit+0x2be/0x7c0
> [ 2703.397895]  [<ffffffff81164d0b>] sys_io_submit+0xb/0x10
> [ 2703.397995]  [<ffffffff8100307b>] system_call_fastpath+0x16/0x1b
> Adding some tracing, it is due to nfs completing the request then
> returning something other than -EIOCBQUEUED, so aio.c
> also completes the request.

To address this, prevent the NFS direct I/O engine from completing
async iocbs when the forward path returns an error without starting
any I/O.

This fix appears to survive ^C during both "xfstest no. 208" and "fsx

It's likely this bug has existed for a very long while, as we are seeing
very similar symptoms in OEL 5.  Copying stable.

Signed-off-by: Chuck Lever <>
Signed-off-by: Trond Myklebust <>
Signed-off-by: Greg Kroah-Hartman <>
7 years agoNFS: Fix an NFS client lockdep issue
Trond Myklebust [Thu, 27 Jan 2011 19:55:39 +0000 (14:55 -0500)]
NFS: Fix an NFS client lockdep issue

commit e00b8a24041f37e56b4b8415ce4eba1cbc238065 upstream.

There is no reason to be freeing the delegation cred in the rcu callback,
and doing so is resulting in a lockdep complaint that rpc_credcache_lock
is being called from both softirq and non-softirq contexts.

Reported-by: Chuck Lever <>
Signed-off-by: Trond Myklebust <>
Signed-off-by: Greg Kroah-Hartman <>
7 years agoASoC: Blackfin TDM: fix missed snd_soc_dai_get_drvdata update
Mike Frysinger [Wed, 12 Jan 2011 04:08:19 +0000 (23:08 -0500)]
ASoC: Blackfin TDM: fix missed snd_soc_dai_get_drvdata update

commit 15d2e22b820bad62854d6ad99d8af8320adf4a91 upstream.

One spot was missed in this driver when converting from
snd_soc_dai.private_data to snd_soc_dai_get_drvdata.

Signed-off-by: Mike Frysinger <>
Acked-by: Liam Girdwood <>
Signed-off-by: Mark Brown <>
Signed-off-by: Greg Kroah-Hartman <>
7 years agoASoC: Blackfin AC97: fix build error after multi-component update
Mike Frysinger [Wed, 12 Jan 2011 00:57:33 +0000 (19:57 -0500)]
ASoC: Blackfin AC97: fix build error after multi-component update

commit e9c2048915048d605fd76539ddd96f00d593e1eb upstream.

We need to tweak how we query the active capture/playback state after
the recent overhauls of common code.

Signed-off-by: Mike Frysinger <>
Acked-by: Liam Girdwood <>
Signed-off-by: Mark Brown <>
Signed-off-by: Greg Kroah-Hartman <>
7 years agoASoC: WM8990: msleep() takes milliseconds not jiffies
Dimitris Papastamos [Fri, 14 Jan 2011 15:59:13 +0000 (15:59 +0000)]
ASoC: WM8990: msleep() takes milliseconds not jiffies

commit 7ebcf5d6021a696680ee77d9162a2edec2d671dd upstream.

Signed-off-by: Dimitris Papastamos <>
Acked-by: Liam Girdwood <>
Signed-off-by: Mark Brown <>
Signed-off-by: Greg Kroah-Hartman <>
7 years agoASoC: When disabling WM8994 FLL force a source selection
Mark Brown [Fri, 3 Dec 2010 16:02:10 +0000 (16:02 +0000)]
ASoC: When disabling WM8994 FLL force a source selection

commit 4514e8997fbefd5befd6176ac9785e287b4daed4 upstream.

When we disable the WM8994 FLL code path sharing means that we end up
writing out a configuration. Currently this is the currently active
input and output frequency (which causes snd_soc_update_bits() to
suppress actual writes both immediately and in the common case where
we reenable the same configuration later) but we allow machine drivers
to pass through a source of zero. Since the register values written
are one less than the source constants this causes corruption of other
bitfields in the register.

Fix this by using the most recently configured FLL source when none is

Signed-off-by: Mark Brown <>
Acked-by: Liam Girdwood <>
Signed-off-by: Greg Kroah-Hartman <>
7 years agoALSA: HDA: Add subwoofer quirk for Acer Aspire 8942G
David Henningsson [Thu, 10 Feb 2011 14:39:19 +0000 (15:39 +0100)]
ALSA: HDA: Add subwoofer quirk for Acer Aspire 8942G

commit a6c47a85b8e7e4a8c47394607c5e5c43224b0892 upstream.

According to the reporter, node 0x15 needs to be muted for subwoofer
to stop sounding. This pin is marked as unused by BIOS, so fix that.

Reported-by: Hans Peter
Signed-off-by: David Henningsson <>
Signed-off-by: Takashi Iwai <>
Signed-off-by: Greg Kroah-Hartman <>
7 years agoALSA: hrtimer: handle delayed timer interrupts
Clemens Ladisch [Thu, 10 Feb 2011 15:15:44 +0000 (16:15 +0100)]
ALSA: hrtimer: handle delayed timer interrupts

commit b1d4f7f4bdcf9915c41ff8cfc4425c84dabb1fde upstream.

If a timer interrupt was delayed too much, hrtimer_forward_now() will
forward the timer expiry more than once.  When this happens, the
additional number of elapsed ALSA timer ticks must be passed to
snd_timer_interrupt() to prevent the ALSA timer from falling behind.

This mostly fixes MIDI slowdown problems on highly-loaded systems with
badly behaved interrupt handlers.

Signed-off-by: Clemens Ladisch <>
Reported-and-tested-by: Arthur Marsh <>
Signed-off-by: Takashi Iwai <>
Signed-off-by: Greg Kroah-Hartman <>
7 years agohwmon: (lm63) Consider LM64 temperature offset
Dirk Eibach [Wed, 9 Feb 2011 09:51:34 +0000 (04:51 -0500)]
hwmon: (lm63) Consider LM64 temperature offset

commit 2778fb13ba0fed1b3e4a040e71f7881d399610a3 upstream.

LM64 has 16 degrees Celsius temperature offset on all
remote sensor registers.
This was not considered When LM64 support was added to lm63.c.

Signed-off-by: Dirk Eibach <>
Signed-off-by: Guenter Roeck <>
Signed-off-by: Greg Kroah-Hartman <>
7 years agoinput: bcm5974: Add support for MacBookAir3
Edgar (gimli) Hucek [Tue, 9 Nov 2010 16:38:42 +0000 (17:38 +0100)]
input: bcm5974: Add support for MacBookAir3

commit 6021afcf19d8c6f5db6d11cadcfb6a22d0c28a48 upstream.

This patch adds support for the MacBookAir3,1 and MacBookAir3,2

[ touchpad range calibration]
Signed-off-by: Edgar (gimli) Hucek <>
Signed-off-by: Henrik Rydberg <>
Signed-off-by: Jiri Kosina <>
Signed-off-by: Greg Kroah-Hartman <>
7 years agoInput: i8042 - introduce 'notimeout' blacklist for Dell Vostro V13
Jiri Kosina [Sat, 8 Jan 2011 09:37:26 +0000 (01:37 -0800)]
Input: i8042 - introduce 'notimeout' blacklist for Dell Vostro V13

commit f8313ef1f448006207f12c107123522c8bc00f15 upstream.

i8042 controller present in Dell Vostro V13 errorneously signals spurious

Introduce i8042.notimeout parameter for ignoring i8042-signalled timeouts
and apply this quirk automatically for Dell Vostro V13, based on DMI match.

In addition to that, this machine also needs to be added to nomux blacklist.

Signed-off-by: Jiri Kosina <>
Signed-off-by: Dmitry Torokhov <>
Cc: Tim Gardner <>
Signed-off-by: Greg Kroah-Hartman <>
7 years agoALSA: hda - Fix memory leaks in conexant jack arrays
Takashi Iwai [Wed, 2 Feb 2011 16:16:38 +0000 (17:16 +0100)]
ALSA: hda - Fix memory leaks in conexant jack arrays

commit 70f7db11c45a313b23922cacf248c613c3b2144c upstream.

The Conexant codec driver adds the jack arrays in init callback which
may be called also in each PM resume.  This results in the addition of
new jack element at each time.

The fix is to check whether the requested jack is already present in
the array.

Reference: Novell bug 668929

Signed-off-by: Takashi Iwai <>
Signed-off-by: Greg Kroah-Hartman <>
7 years agoALSA: HDA: Fix dmesg output of HDMI supported bits
David Henningsson [Tue, 25 Jan 2011 18:44:26 +0000 (19:44 +0100)]
ALSA: HDA: Fix dmesg output of HDMI supported bits

commit d757534ed15387202e322854cd72dc58bbb975de upstream.

This typo caused the dmesg output of the supported bits of HDMI
to be cut off early.

Signed-off-by: David Henningsson <>
Signed-off-by: Takashi Iwai <>
Signed-off-by: Greg Kroah-Hartman <>
7 years agoALSA: fix invalid hardware.h include in ac97c for AVR32 architecture
Hans-Christian Egtvedt [Mon, 24 Jan 2011 15:09:56 +0000 (16:09 +0100)]
ALSA: fix invalid hardware.h include in ac97c for AVR32 architecture

commit fd76804f3f5484b35e6a51214c91e916ebba05aa upstream.

This patch fixes the non-compiling AC97C driver for AVR32 architecture by
include mach/hardware.h only for AT91 architecture. The AVR32 architecture does
not supply the hardware.h include file.

Signed-off-by: Hans-Christian Egtvedt <>
Signed-off-by: Takashi Iwai <>
Signed-off-by: Greg Kroah-Hartman <>