[Karmic] SRU: Update to stable kernel

Stefan Bader stefan.bader at canonical.com
Thu Nov 12 09:29:08 UTC 2009

Leann Ogasawara wrote:
> Hi All,
> The stable kernel has recently been released:
> https://bugs.launchpad.net/ubuntu/karmic/+source/linux/+bug/480144
> The stable release contains 101 patches.  I've applied and
> pushed these patches to a stable branch in my tree for review.  I'll
> reply to this email and inline some additional comments I have for some
> patches.
> git://kernel.ubuntu.com/ogasawara/ubuntu-karmic stable-
> SRU Justification/Impact: The upstream process for stable tree updates
> is quite similar in scope to the Ubuntu SRU process, e.g., each patch
> has to demonstrably fix a bug, and each patch is vetted by upstream by
> originating either directly from Linus' tree or in a minimally
> backported form of that patch. The upstream stable patch set is
> now available. It should be included in the Ubuntu karmic kernel as
> well.
> http://kernel.org/pub/linux/kernel/v2.6/ChangeLog-
> http://kernel.org/pub/linux/kernel/v2.6/patch-
> The 101 stable patches are as follows:
ACK with the limitation to remove the two patches that cause modules/drivers
to go away (as Leann also mentioned). Below the writeup during my review.


commit 120f68c426e746771e8c09736c0f753822ff3f52
Author: Greg Kroah-Hartman <gregkh at suse.de>


commit ee2c54041db5404353aa768ba1ec6e4fbda7ef5f
Author: Rafael J. Wysocki <rjw at sisk.pl>

    PM / yenta: Split resume into early and late parts (rev. 4)

SABIB: A somewhat larger change to split resume for PCMCIA
into two parts. Fixes an upstream regression and should only
affect systems with yenta cardbus sockets.

commit 54fb902e50fee7b3c39d31853805f849edfdd62e
Author: JosephChan at via.com.tw <JosephChan at via.com.tw>

    pata_via: extend the rev_max for VT6330

Just a change to a probe table to include more revisions of the
same ID.

commit 8def1ff84d14a68872dd998909a0398c7cfecdc4
Author: JosephChan at via.com.tw <JosephChan at via.com.tw>

    sata_via: Remove redundant device ID for VIA VT8261

Removing one ID which apparently was decided to be used for some-
thing else.

commit 20cde25b833745a42ae7591592ec527653b569b4
Author: Benjamin Herrenschmidt <benh at kernel.crashing.org>

    powerpc: Fix some late PowerMac G5 with PCIe ATI graphics

This changes only powerpc code and ands a specific fixup handler for
a certain device ID. As the changes are quite hw specific, hard to
understand but also won't affect the main architectures.

commit 01ab8be189e37f37ede8bd9532bd22198062d631
Author: Andre Detsch <adetsch at br.ibm.com>

    powerpc/pci: Fix regression in powerpc MSI-X

Adresses msi-x init for the powerpc platform specifically.

commit 6b05a888adf307827a8cc960a6f660555bf9a683
Author: Benjamin Herrenschmidt <benh at kernel.crashing.org>

    powerpc/pmac: Fix issues with sleep on some powerbooks

Again powerpc, the change mostly heads for using IRQF_TIMER
instead of any interrupt. There is some change to pmu_adb_reset_bus()
request data I cannot really comment on. And the useage of the arg
to __pmu_adb_autopoll() seems a bit cheesy but looks ok as it is

commit e434e9fc1661989db1a39b6f5be29c39e13b419a
Author: Benjamin Herrenschmidt <benh at kernel.crashing.org>

    powerpc/pmac: Fix PowerSurge SMP IPI allocation

One line fix (and powerpc again) to use a wrapper function to
construct an IRQ number instead of some immediate value.

commit 6fa5a4d48a5cb7aa452cc34120f9122301ec8c93
Author: Darrick J. Wong <djwong at us.ibm.com>

    KVM: Prevent kvm_init from corrupting debugfs structures

Fixes an oops caused by trying to free debugfs structure twice when
both arch modules where probed and removed. Only one of them gets to
finish the init anyways, so the debugfs init is moved to a later
place which is only reached when all goes well.

commit 77608cc64e305111725bbefe7d238d9f6da38c5d
Author: Joerg Roedel <joerg.roedel at amd.com>

    x86/amd-iommu: Un__init function required on shutdown

Simple change to declare a shutdown function _not_ being __init
code. So it will not cause a panic on shutdown.

commit de59170f391ea08e7ebb1e85e687a5df96985fcb
Author: Reinette Chatre <reinette.chatre at intel.com>

    iwlwifi: reduce noise when skb allocation fails

Rather cosmetical change to prevent an order-2 allocation failure
together with an error message should it fail. Unless the driver
really is low on buffers.

commit a8d37f2ebd5ef08a169ecf7c7fb52c2ef4037cec
Author: Reinette Chatre <reinette.chatre at intel.com>

    iwlwifi: fix potential rx buffer loss

Fix to prevent potential loss of rx buffers. This would cause a
non-functional device at some point.

commit 49fae454bb8ca92b5ff999cc823ac60caf2ccb61
Author: Ingo Molnar <mingo at elte.hu>

    x86: mce: Fix thermal throttling message storm

Limit the number of messages when thermal throtteling is active.
Not directly fixing a serious problem but might cause big/quickly
filling log files.

commit fe2414ca987a3a49007bec6fd12254c7defae2a9
Author: Ingo Molnar <mingo at elte.hu>

    x86: mce: Clean up thermal throttling state tracking code

A bit questionable from the viewpoint of SRU criteria. Though it
simplifies code readability and size. There is no functional

commit 94f48f549e455e4abbb3f0d6e65185830cbec8f1
Author: Wu Fengguang <fengguang.wu at intel.com>

    vmscan: limit VM_EXEC protection to file pages

Prevent OOM by only protecting mapped executable files from
being ignored on the shrink_active_list() run.

commit 781e844b1db70e0fe4911fa049e675a010d79cf9
Author: Bastian Blank <bastian at waldi.eu.org>

    powerpc: Remove SMP warning from PowerMac cpufreq

Powerpc, removes a compile warning only, which apparently is not
applicable anyways.

commit 73d1f971858ae19b7b755f244b025dd4ddbfe860
Author: Trond Myklebust <Trond.Myklebust at netapp.com>

    NFSv4: The link() operation should return any delegation on the file

Sounds rather like a performance improvement that a bug fix, though
things might be hidden.

commit 463cbfdf160340cbc5e91db88546471f697f98d9
Author: Trond Myklebust <Trond.Myklebust at netapp.com>

    NFSv4: Kill nfs4_renewd_prepare_shutdown()

Again, could be only peformance or worse. As the workqueue seems to be
clared too often it could also cause hangs with nfsv4.

commit 85c60e5dfa4c00bb4deacc36cd719db557c9d2c6
Author: Trond Myklebust <Trond.Myklebust at netapp.com>

    NFSv4: Fix two unbalanced put_rpccred() issues.

Fixes kernel OOPS caused by dropping referenced that were
never claimed.

commit 4886f073309eeb9d99d9f05d944f8b0874c60bbd
Author: Trond Myklebust <Trond.Myklebust at netapp.com>

    NFSv4: Fix a bug when the server returns NFS4ERR_RESOURCE

Limited to nfsv4, the actual consequences unclear. If the
sequence numbers where incorrect due to the bug, this might
cause IO errors/failures. The fix looks reasonable in the

commit 2169536c156b6bc876c0a80e5cf98319cc3b9aab
Author: Terry Loftin <terry.loftin at hp.com>

    nfs: Panic when commit fails

Fixes a panic by adding an informational structure to a data

commit 55a6efb39afcfe70754a733cdca91b6e1599fb6b
Author: Ben Hutchings <ben at decadent.org.uk>

    nfs: Avoid overrun when copying client IP address string

Replaces memcpy with strlcpy to avoid data corruption by string

commit cd45ad45a38aa500f96254ce21890ae7611cef46
Author: Jiri Pirko <jpirko at redhat.com>

    netlink: fix typo in initialization (CVE-2009-3612)

Classified as being related to CVE-2009-3612. Apparantly the init
code had a copy and paste error, leaving one byte uninitialized.

commit 532d2b259d9cd9450eb7b500405cfe3e6c2f7aa7
Author: Jean Delvare <khali at linux-fr.org>

    hwmon: (it87) Fix VID reading on IT8718F/IT8720F

Fixes use of wrong data type for comparision. The result might range
from just getting the wrong value to poking completly wrong hw registers.

commit 44791e49b282eb1901b28d27d9756ace8f873dca
Author: Jiri Bohac <jbohac at suse.cz>

    bonding: fix a race condition in calls to slave MII ioctls

Preventing a seriously sounding race condition in bonding by
replacing a static variable by a dynamically allocated.

commit 474ac4951f31a678cda82833c6ec6dbb4b58bacb
Author: David Henningsson <launchpad.web at epost.diwic.se>

    ALSA: ice1724 - Make call to set hw params succeed on ESI Juli@

Fixes a non-critically sounding issue with a specific sound card,
allowing concurrnet playback and recording.

commit 027590f053888a282d09f420a39ad08a17dda76f
Author: Tomoki Sekiyama <tomoki.sekiyama.qu at hitachi.com>

    AF_UNIX: Fix deadlock on connecting to shutdown socket (CVE-2009-3621)

This fixes a potential DoS loophole simply by adding propper
exit condition.

commit 1917888ae8076efca1ae51c7517806572ad8db02
Author: Alan Cox <alan at linux.intel.com>

    pata_sc1200: Fix crash on boot

Adds NULL termination to prevent crash on boot.

commit 01a6a52a3cadb6a6db03199f3ec7ccb023aac45f
Author: Larry Finger <Larry.Finger at lwfinger.net>

    b43: Fix Bugzilla #14181 and the bug from the previous 'fix'

We got a bug report specifically for that: LP#476154 and some positive
feedback on a test kernel. Fixes a required unload reload cycle in case
the radio switch was turned off.

commit 7a99333e851ef087c7cd836950900602f0843c24
Author: David Howells <dhowells at redhat.com>

    KEYS: get_instantiation_keyring() should inc the keyring refcount in all cases

Adds aquisition of a reference to prevent a panic when the object gets
released too early.

commit 3c802f78271bd558d152a5f8632ea8395ed27a9b
Author: Len Brown <len.brown at intel.com>

    Revert "ACPI: Attach the ACPI device to the ACPI handle as early as possible"

This fixes a non-bootable Inspiron 2650 (lp#327499) which would or should
be a regression-release.

commit 5621e89d8bdf6934b3c0a4893e1a568f4a01061d
Author: Rafael J. Wysocki <rjw at sisk.pl>

    ACPI / PCI: Fix NULL pointer dereference in acpi_get_pci_dev() (rev. 2)

Fixes a NULL pointer violation (upstream regression between 2.6.30 and 31)
by adding a pointer check.

commit e259007f0090e5c2c4c0df1e316431a25287633e
Author: Sam Ravnborg <sam at ravnborg.org>

    alpha: fix build after vmlinux.lds.S cleanup

Build fix for alpha. Has no impact.

commit 039b7b2ecddd52d3e1d5cc4337d2b1f7053916eb
Author: Manuel Lauss <manuel.lauss at gmail.com>

    mips: fix build of vmlinux.lds

Build fix for mips. Has no impact.

commit 55af1d631af064ed53eea5e4805ffac0deb66e2d
Author: Eric Paris <eparis at redhat.com>

    fsnotify: do not set group for a mark before it is on the i_list

Fix a race condition. The effect is unknown but the fix moves assignment
of some elements into the case that handles things when no other entry
exists which looks right.

commit 6a3f97dfa2464e8186bd8bdadd29daca411d07c1
Author: Joerg Roedel <joerg.roedel at amd.com>

    x86/amd-iommu: Workaround for erratum 63

Seems to prevent IO_PAGE_FAULTS on amd64 hw. Not quite sure why the change
needs to be applied on attaching and detaching but tend to believe the
manufacturer (patch comes from amd).

commit c385d8d439907b6cd7b47da86b1837b1006c7228
Author: Jens Axboe <jens.axboe at oracle.com>

    fuse: fix kunmap in fuse_ioctl_copy_user

Changes the argument to a function call to the right pointer. Looks like
that might cause data corruption.

commit 82d2d504aeaad7ada524e9e3bfd8eede3ca2bf15
Author: Anand V. Avati <avati at gluster.com>

    fuse: prevent fuse_put_request on invalid pointer

Looks like an OOPS cause. Changes a put from unconditionally to only
happen on success.

commit 15076d40b0ce2e04cf383425a5a26e01f53b2aee
Author: Gleb Natapov <gleb at redhat.com>

    KVM: get_tss_base_addr() should return a gpa_t

PABIB: Change the return type of a function which would otherwise
return truncated addresses with more than 4G of memory. Patch is
simple and only affects kvm.

commit 42d7bdfc3320039bb9310703d6475a62f5c74772
Author: Linus Torvalds <torvalds at linux-foundation.org>

    sgi-gru: decrapfiy options_write() function

A near complete rewrite of a procfs write function. Luckily only affecting
one driver. It looks correct, too.

commit 9416d12bcaa77126bd55f7df2cb4e78b711f97bd
Author: Jeremy Fitzhardinge <jeremy.fitzhardinge at citrix.com>

    xen: mask extended topology info in cpuid

Fixes a crash in Xen. Looks safe and also affects only xen which is
not officially supported.

commit b318606e41620d2fee78444ae54533a323108e2b
Author: Jeremy Fitzhardinge <jeremy.fitzhardinge at citrix.com>

    xen/hvc: make sure console output is always emitted, with explicit polling

Replacing the write_console() implementation in a xen driver to improve
debugging. Not necessarily a SRU worthy change but does not affect the
normal kernel.

commit e05947b4e09d7cc3e4811a3fe0e6a1d83a9805cd
Author: Fabian Henze <hoacha at quantentunnel.de>

    drm/i915: add B43 chipset support

Adding a new chipset to the driver. Should not regress existing
chipsets. And as Leann found, we are already carrying that anyways.

commit 685ca021e44e2f6bc7e54050305ce9a4685804c6
Author: Fabian Henze <hoacha at quantentunnel.de>

    agp/intel: Add B43 chipset support

Again adding new device IDs to the driver.

commit f19f531a3257ea2e709de9e423c0a4a72358a4be
Author: Zhenyu Wang <zhenyuw at linux.intel.com>

    drm/i915: fix panel fitting filter coefficient select for Ironlake

Seems to fix HP video corruption issues. The change is limited to the
case when panel fitting is enabled and then sets an additional bit.
Somehow I would expect us to have a bug for that. Steve?

commit 82a1c0522cf747742461ac7874f463af46e3ba18
Author: Zhenyu Wang <zhenyuw at linux.intel.com>

    drm/i915: fix to setup display reference clock control on Ironlake

This affects only Intel chips with device id 0x0042 and 0x0046 (Ironlake).
We should have the ability to test this, shouldn't we?

commit a9d48b3eae33aeb6bc1c366913778d77267e17b7
Author: Zhenyu Wang <zhenyuw at linux.intel.com>

    drm/i915: Fix FDI M/N setting according with correct color depth

This affects only Intel chips with device id 0x0042 and 0x0046 (Ironlake).
As Leann pointed out we already carry this in Karmic.

commit 61ac9e400eb6c0c3d3e9da24c78176fbe272fc1b
Author: Ryusuke Konishi <konishi.ryusuke at lab.ntt.co.jp>

    nilfs2: fix dirty page accounting leak causing hang at write

Fixes a hang on arm (it claims), but generically changes the code in nilfs2
to use the same function in both cases of the if construct. So it sounds

commit 092afb745d0dfdd6cf5ed86ed301023285700fd2
Author: Jan Beulich <JBeulich at novell.com>

    x86-64: Fix register leak in 32-bit syscall audting

This sounds like protentially coming in as a CVE in the future.
Though I must admit that I don't really understand the assembly
well enough.

commit 82a90819b2992515f794b032ee7d75fba4771e33
Author: Bo Liu <bo-liu at hotmail.com>

    mm: remove incorrect swap_count() from try_to_unuse()

Simple enough looking but not really critically. But being faster on
swapof sounds ok.

commit cd4ad4b9dda686bce57f939671cc066a554b7f92
Author: David Howells <dhowells at redhat.com>

    NOMMU: Don't pass NULL pointers to fput() in do_mmap_pgoff()

Prevent OOPS by not passing NULL pointers. Looks sane.

commit 7af6e31fc7ab7235ae38be7847f346d8d776389d
Author: Robin Holt <holt at sgi.com>

    x86, UV: Set DELIVERY_MODE=4 for vector=NMI_VECTOR in uv_hub_send_ipi()

SGI Ultraviolet: corrects the sort of IRQ used. Implementation looks ok.

commit 39652527bce2d4e12ac4008ec93e41a8b28238a3
Author: Robin Holt <holt at sgi.com>

    x86, UV: Fix information in __uv_hub_info structure

Sounds a bit gibberish to me but as its only affecting Ultraviolet
I see not big regression potential.

commit 958da8f522180518a1f04b65d4eef61eb4ecbe72
Author: Johannes Berg <johannes at sipsolutions.net>

    mac80211: check interface is down before type change


Might fix a lot of bugs we have by moving a test whether the interface is
busy to the place it should be (before changing).

commit db6ad3e1187477afb4a1fc79bc48f0a8cf916ec3
Author: Björn Smedman <bjorn.smedman at venatech.se>

    mac80211: fix for incorrect sequence number on hostapd injected frames

A small change that could fix a lot of association problems. Hard to relate
to specific bugs though.

commit 49178370a647948ca8b529cc65f0c214bec33d42
Author: Steve French <sfrench at us.ibm.com>

    CIFS: Fixing to avoid invalid kfree() in cifs_get_tcp_session()

Fix a kernel BUG by avoiding double kfree. Sane

commit 48fe10afb355d88be5a9c0b7d5a198c5e5b9b4aa
Author: Michael S. Tsirkin <mst at redhat.com>

    virtio: order used ring after used index read

Prevent guest crashes by adding a memory barrier. Safe.

commit bed202356361d3620a51974eadca78261e049ef0
Author: Kevin Hilman <khilman at deeprootsystems.com>

    cpuidle: always return with interrupts enabled

Re-enable interrupts before leaving. Might not cause a real bug
but caused WARN_ONs on arm. Looks ok.

commit 1133a6e3632b5897f5a25084e38b221cb92f3eb6
Author: David Woodhouse <dwmw2 at infradead.org>

    libertas if_usb: Fix crash on 64-bit machines

Crash and potential data corruption. Fix looks very simple.

commit fb20e3f85ece49f23b72a7ad6aa441e004f4f3e7
Author: Bryan Wu <bryan.wu at canonical.com>

    USB: option: TLAYTECH TUE800 support

-> LP#456264

Additional ID added.

commit ad1d49cce87e86038f7756c68626a2760d7bfdad
Author: Huzaifa Sidhpurwala <sidhpurwala.huzaifa at gmail.com>

    USB: option: Support for AIRPLUS MCD650 Datacard

Another enablement patch only adding an ID.

commit e4613c5f3a957475b1f5b011e7833cc2b57b6ed4
Author: Ronnie Furuskog <rofu7 at hotmail.com>

    USB: option: Patch for Huawei Mobile Broadband E270+ Modem

Another one.

commit def3834da355e930d694b7373fe651feb0162cf4
Author: Elina Pasheva <epasheva at sierrawireless.com>

    USB: serial: sierra driver send_setup() autopm fix

Not critical but sort of straight forward. Fixes autosuspend in
the sierra driver.

commit 2367aa9bfd515bd76728b28d62a0fd00ef70832e
Author: Rusty Russell <rusty at rustcorp.com.au>

    param: fix setting arrays of bool

Should fix passing bool arguments on the kernel command line. As the
function changed is static it should not affect the ABI.

commit a8381266f673b5597f11fb9bad9ae196e9915f87
Author: Rusty Russell <rusty at rustcorp.com.au>

    param: fix NULL comparison on oom

Simple fix for crash on OOM. Ok.

commit 9f00eee2ffab59cb85ecf1de54282c7fb4669565
Author: Rusty Russell <rusty at rustcorp.com.au>

    param: fix lots of bugs with writing charp params from sysfs, by leaking mem.

This seems to trade OOPSes for leaking a bit of memory on charp

commit 05fe6c842ab23f6e823ad87a84adab73ab60beaa
Author: Alan Cox <alan at linux.intel.com>

    tty: Mark generic_serial users as BROKEN

As this removes several modules, I agree with Leann, that we should not take
this one. Even as the modules are broken in usage (or assumed to be so). We
do not have a way to say how broken. They might work under some circumstances
and removing them would be a regression we do not want to do on stable.

commit b4e01e264884a0acbb11c85e3e6d22d23b3f8194
Author: Stas Sergeev <stsp at aknet.ru>

    Driver core: fix driver_register() return value

As said in subject fix of return value which sounds reasonable.

commit 6acffd270b393f4f81e2eebe9aaf5d6837afb07e
Author: Ben Hutchings <ben at decadent.org.uk>

    hfsplus: refuse to mount volumes larger than 2TB

Avoid data loss which is caused by using too small sector number
variables by refusing to mount devices that are too big. Not a
final thing to do but at least safe.

commit 623fefd5b367d3d2e614157a3eaba03e2accbd85
Author: OGAWA Hirofumi <hirofumi at mail.parknet.co.jp>

    dpt_i2o: Fix typo of EINVAL

Fix return code to be negative. Made in the patch below

commit 01019ab600200ce6f43a52d0e8f7ab5d24a82da3
Author: Alan Cox <alan at linux.intel.com>

    dpt_i2o: Fix up copy*user

Limit the size passed to the kernel. Looks like it might prevent
data corruption and/or security issues.

commit 4d4c684cd460ca0e1f3daacde10e201776616c3f
Author: Benjamin Herrenschmidt <benh at kernel.crashing.org>

    8250_pci: add IBM Saturn serial card

Adding configuration data for a card that would otherwise be
incorrectly handled. Neither strict enabled nor kernel bug fix,
but should not cause regressions either.

commit 7b6f062f8c8252063ab1bf8c2c6efc9bbcaeedcc
Author: KOSAKI Motohiro <kosaki.motohiro at jp.fujitsu.com>

    do_mbind(): fix memory leak

Straight forward fix for a memory leak.

commit 71ad01d06d0803e034dae2fcd33aeb5e9662c046
Author: KOSAKI Motohiro <kosaki.motohiro at jp.fujitsu.com>

    mbind(): fix leak of never putback pages

Fixes a page leak on error. Ok.

commit bf501c3e8bb75e9f22f79eca39dc6c8ca5a32432
Author: Alan Cox <alan at linux.intel.com>

    ray_cs: Fix copy_from_user handling

Looks like a possible security hole. If count is large enough
to be treated as negative int memory corruption can be caused.

commit e676b2bd4553802a0b6211d5cca7fb3358ce66cf
Author: Mikulas Patocka <mpatocka at redhat.com>

    dm snapshot: use unsigned integer chunk size

PABIB: While claiming to fix a compile failure on 32bit systems with
large block devices, this likely causes an ABI bump. The change itself
should be safe if all places have been hit. On the other hand dm-snapshots
are claimed to be broken anyways.

commit 537c93c1c116409f1418c99286aa81c8b9ccbdee
Author: Mikulas Patocka <mpatocka at redhat.com>

    dm snapshot: require non zero chunk size by end of ctr

Prevent chunk_size from being zero which is claimed to possible crash
in many places. Looks sane.

commit 651385e4bbbff823f111e4e8a68b2e1613c34c79
Author: Mikulas Patocka <mpatocka at redhat.com>

    dm snapshot: lock snapshot while supplying status

Add locking. No return within the locked area. Ok.

commit e61f48eb0ac00c6d73dd53b3d79abe2c578328dc
Author: Mikulas Patocka <mpatocka at redhat.com>

    dm exception store: fix failed set_chunk_size error path

Would not be critical but would block devices from getting released

commit c5d0c81248f5b0d14f464f4170f76fb756df5ac0
Author: Kiyoshi Ueda <k-ueda at ct.jp.nec.com>

    dm: dec_pending needs locking to save error value

Preventive fix but looks reasonable.

commit 17f34f975f03e1cdad57fc04c590a750e2558d42
Author: Zdenek Kabelac <zkabelac at redhat.com>

    dm: add missing del_gendisk to alloc_dev error path

Pobably rare case but IMO could cause inability to create device-mapper
targets until reboot.

commit 67060bc39921c6fea6c8ce2eceaf348038dbba89
Author: Andrew Morton <akpm at linux-foundation.org>

    dm log: userspace fix incorrect luid cast in userspace_ctr

Fixes a compile warning. As likely both values are pointers/unsigned
longs it would be purely cosmetical.

commit 028e4a12f691559d367c12cb0612e1635b5c7a32
Author: Mikulas Patocka <mpatocka at redhat.com>

    dm snapshot: sort by chunk size to fix race

Data corruption fix on snapshots. Looks simple enough.

commit 1b0fd1e25a9393003f7034a83f5932ce26242577
Author: Jonathan Brassow <jbrassow at redhat.com>

    dm snapshot: free exception store on init failure

Fix for failure path. Not really sure about the implication.
But fix looks reasonable.

commit af57e833c96884bf0aa231039a25c853f7e83864
Author: Dmitry Torokhov <dmitry.torokhov at gmail.com>

    Input: synaptics - add another Protege M300 to rate blacklist

BugLink: http://bugs.launchpad.net/bugs/433801
Already applied in post relase update. Just replaced by this one. In order
to avoid double "closed by" mails of the janitor, we would not add the
buglink here as it should be in updates with -15.50 (hopefully).

commit bfdf85fe73f1b697430bbeb763f8456e2ce52e9d
Author: Thomas Gleixner <tglx at linutronix.de>

    watchdog: Fix rio watchdog probe function

Fixes a seemingly long standing bug by not running into the error case
on success.

commit e1f7f4182a52b850379c0d4f35a8085d9552786a
Author: David S. Miller <davem at davemloft.net>

    sparc: Kill PROM console driver.

Removes the procon driver which is only used on sparc. We have enabled that
though recommendation is no. But it has been in the release. It automatically
will go away in Lucid, but we should keep the current state.

commit b03a3764e5605f63f4a8eabe416c0d8156b93108
Author: David S. Miller <davem at davemloft.net>

    sparc64: Set IRQF_DISABLED on LDC channel IRQs.

Sounds ok, for sparc only.

commit 99bea2784294c3bbec4d1d69d7c6e3b01e59edbb
Author: Chuck Ebbert <cebbert at redhat.com>

    ahci: revert "Restore SB600 sata controller 64 bit DMA"

Possibly quick fixing IO problems with that chipset. As that has
been working before and was only changed to 64bit as problems were
thought to be fixed, this will be safe. Only affects a certain sata

commit fc7df048bf93e52c04f4c39104a5a388b73725b8
Author: Thomas Gleixner <tglx at linutronix.de>

    futex: Fix spurious wakeup for requeue_pi really

Fixes an endless loop caused by a patch below (before)
    futex: Handle spurious wake up

commit 81e6fd571d574da95cf977ff3266d95a802b1820
Author: Darren Hart <dvhltc at us.ibm.com>

    futex: Move drop_futex_key_refs out of spinlock'ed region

Ok, after looking with apw, ther two places where the the reference
counter is marked to be dropped later. And in both cases the function
which got the drop removed is called and in case of a 1 as return it
needs the reference drop. So looks ok but hard to grasp.

commit 6d57fbdd82f6012a4d451bacd75451928cbc29be
Author: Darren Hart <dvhltc at us.ibm.com>

    futex: Check for NULL keys in match_futex

Fix OOPS in case of using a NULL pointer. A bit odd by ok.

commit e68e25e608880c149c2080930295227a79ceb5ac
Author: Thomas Gleixner <tglx at linutronix.de>

    futex: Handle spurious wake up

Not really understanding. RETRY later

commit 866b5a4dc9682101678d42071378a168744b15b0
Author: Andre Przywara <andre.przywara at amd.com>

    KVM: ignore reads from AMDs C1E enabled MSR

Its unclear whether the guest would crash or cause some implications
on the host. Preventing the access likely is ok as the guest won't
really cause a real cpu to go into c1e.

commit c66415b2216cebf6625a2ab01c347a2fc329671a
Author: Marcelo Tosatti <mtosatti at redhat.com>

    KVM: use proper hrtimer function to retrieve expiration time

Read as causing an OOPS in kvm. Use of a different function which
is claimed more robust.

commit 9c367e53d2780b5131baecd1874835ce8ddc4ed0
Author: Yinghai Lu <yinghai at kernel.org>

    nfs: Fix nfs_parse_mount_options() kfree() leak

Straight forward memleak fix.

commit 7613217197c8927891e554c1734c1c417b22672b
Author: Tejun Heo <tj at kernel.org>

    sata_nv: make sure link is brough up online when skipping hardreset

The explanation and the code look reasonable.

commit 32f4683db485b809264115327a2a34fe51d4e7ba
Author: Tejun Heo <tj at kernel.org>

    libata: fix PMP initialization

Probably not critical but fix looks straight forward.

commit 9982364654c186acd48c3070dcf6a76c69e540cc
Author: Tejun Heo <tj at kernel.org>

    libata: fix internal command failure handling

We might have WARNING reports on those (would need to see which)
Looks ok but is not sounding critical.

commit 2f37b165767202d7438aac98710f20038d197542
Author: Yinghai Lu <yinghai at kernel.org>

    pci: increase alignment to make more space for hidden code

We have some reports and waiting for some more validations to come in
for that harmlessy sounding fix. But this seems to fix regressions
on some devices not being detected/operational anymore.

commit c40ca2aab8a3d3ba213b7b174cb2c90bd3b51235
Author: Earl Chew <earl_chew at agilent.com>

    fs: pipe.c null pointer dereference

In essence adds check for the pipe being valid to open calls.
Look ok.

> == Fixes ==
> Bug 327499: "Revert "ACPI: Attach the ACPI device to the ACPI handle as early as possible"
> Bug 456264: "USB: option: TLAYTECH TUE800 support"
> Bug 476154: "b43: Fix Bugzilla #14181 and the bug from the previous 'fix'"
> Bug 407824: "pci: increase alignment to make more space for hidden code"
> Bug 407040: "pci: increase alignment to make more space for hidden code"
> Bug 453444: "x86: mce: Fix thermal throttling message storm"
> Bug 459265: "NFSv4: Fix two unbalanced put_rpccred() issues."
> == Possibly Fixes ==
> Bug 442966: "fsnotify: do not set group for a mark before it is on the i_list"
> Bug 457523: "mac80211: check interface is down before type change"

More information about the kernel-team mailing list