Stable kernel 2.6.25.3 released

We (the -stable team) are announcing the release of the 2.6.25.3 kernel.

Yes, its a few hours earlier than originally expected, but there are two
security bugs fixed in here that just recently were made public, so we
figured we would err on the safe side.  All users of the 2.6.25 series
are strongly encouraged to upgrade, and anyone running 2.6.24 should
also move up to 2.6.25 at this time if possible.

I'll also be replying to this message with a copy of the patch between
2.6.25.2 and 2.6.25.3

The updated 2.6.25.y git tree can be found at:
        git://git.kernel.org/pub/scm/linux/kernel/git/stable/linux-2.6.25.y.git
and can be browsed at the normal kernel.org git web browser:
        http://git.kernel.org/?p=linux/kernel/git/stable/linux-2....

thanks,

greg k-h

--------

 Makefile                         |    2 -
 arch/arm/kernel/kprobes-decode.c |    2 -
 arch/arm/kernel/kprobes.c        |    2 -
 arch/sparc/kernel/sys_sparc.c    |    3 -
 arch/sparc64/kernel/sys_sparc.c  |    4 +-
 arch/x86/pci/common.c            |    7 ++--
 arch/x86/pci/init.c              |    2 +
 arch/x86/pci/pci.h               |    2 +
 crypto/authenc.c                 |    5 +-
 crypto/cryptd.c                  |    4 +-
 crypto/eseqiv.c                  |    3 +
 drivers/acpi/processor_idle.c    |    4 ++
 drivers/md/md.c                  |    2 -
 drivers/net/wireless/b43/b43.h   |    4 ++
 drivers/net/wireless/b43/main.c  |   40 ++++++++++++++++++-----
 drivers/serial/mpc52xx_uart.c    |    2 -
 fs/reiserfs/ioctl.c              |    4 --
 fs/reiserfs/super.c              |   17 +++++++---
 fs/utimes.c                      |   17 ++++++++--
 include/crypto/scatterwalk.h     |    4 ++
 include/linux/reiserfs_fs.h      |    1 
 kernel/sched.c                   |   66 ++++++++++++++++++++++++++++++++++++++-
 mm/page_alloc.c                  |   14 +++++++-
 net/ipv6/sit.c                   |    2 -
 24 files changed, 175 insertions(+), 38 deletions(-)


Dan Williams (1):
      md: fix use after free when removing rdev via sysfs

David S. Miller (2):
      sparc: Fix mmap VA span checking.
      sit: Add missing kfree_skb() on pskb_may_pull() failure.

Grant Likely (1):
      POWERPC: mpc5200: Fix unterminated of_device_id table

Greg Kroah-Hartman (1):
      Linux 2.6.25.3

Herbert Xu (2):
      CRYPTO: api: Fix scatterwalk_sg_chain
      CRYPTO: eseqiv: Fix off-by-one encryption

Jan Kara (1):
      reiserfs: Unpack tails on quota files

Julia Lawall (1):
      CRYPTO: cryptd: Correct kzalloc error test

KAMEZAWA Hiroyuki (1):
      mm: fix usemap initialization

Lennert Buytenhek (1):
      kprobes/arm: fix decoding of arithmetic immediate instructions

Michael Buesch (2):
      b43: Fix dual-PHY devices
      b43: Fix some TX/RX locking issues

Miklos Szeredi (1):
      vfs: fix permission checking in sys_utimensat

Nicolas Pitre (1):
      kprobes/arm: fix cache flush address for instruction stub

Patrick McHardy (1):
      CRYPTO: authenc: Fix async crypto crash in crypto_authenc_genicv()

Peter Zijlstra (1):
      sched: fix hrtick_start_fair and CPU-Hotplug

Venkatesh Pallipadi (1):
      2.6.25 regression: powertop says 120K wakeups/sec

Yinghai Lu (1):
      x86 PCI: call dmi_check_pciprobe()

why upgrade from 2.6.24.x to 2.6.25.3?  same reason you would upgrade from 2.6.25.2 to
2.6.25.3.  same reason enterprises pay for Linux.  continued support.

unless someone is backporting security fixes to your current kernel revision, you should
consider doing it yourself or upgrading to 2.6.25.3.  or maybe you've done a risk analysis and
determined that an upgrade to a more recent kernel (2.6.24.x -> 2.6.25.3) is "riskier" (higher
probability of disruption/failure) than a security breach with your current version.

i want to personally thank Oliver Pinter for his continued maintenance of 2.6.22 [1].  the
stable version of linux-vserver is currently stuck at 2.6.22 (due to accommodating the new
container code in later kernels), so i'm forced to stick with 2.6.22 for right now, but
Oliver's "op" patchset has made that a supported possibility.

i only stumbled on his patchset because after the stable team dropped support for 2.6.22, i
searched for Willy Tarreau's patchset because he said he was going to rebase against it [2],
and in desperation i Googled for "2.6.22.20" [3], though i had heard nothing about it while
regularly reading lwn and kerneltrap (but it now appears that i haven't been paying enough
attention to the "Kernel Trees" section of the weekly edition).  i've been compiling his
releases since 2.6.22.21-op1, but honestly i haven't ran one yet because there hasn't been a
pressing enough need, and now that it includes security fixes, i'm waiting for a good time for
"planned downtime" to reboot into the latest.

so consider upgrading to 2.6.25.3 unless a great individual like Oliver Pinter is supporting
2.6.24 (or whatever your version) with backported fixes (security or otherwise).

[1] http://repo.or.cz/w/linux-2.6.22.y-op-patches.git
[2] http://kerneltrap.org/mailarchive/linux-kernel/2007/11/6/...
[3] http://www.google.com/search?q=2.6.22.20

> I've heard mixed feelings about the reliability of 2.6.25[.x].

And what are those feelings you have heard?  Do you have any bugs you have reported for this
kernel series?

Personally, I think it is one of the most used kernel versions in a very long time as both
Fedora and OpenSuSE have based their latest releases on it.  Don't you think that the hundreds
of thousands of users using those releases already would be a good enough reason to feel good?

And becides 2.6.24 isn't supported anymore by anyone that I know of, so if you feel that the
security problems posted are not relevant for you (which might be totally the case), feel free
to stay, but note that no developers are there to help you out if you have problems :(

I think 2.6.24 is the kernel in current Debian testing, so that version willll be in the next
release ("lenny") unless they move to 2.6.25 soon.

2.6.25 is in Debian unstable already.

>2.6.24 is the kernel of the last Ubuntu version 8.04 (Long Term Support)
> and it will be supported three years (desktop version) or five years
>(server version).

That's great if you are a Ubunutu customer, but that is not going to help you out if you are a
kernel.org 2.6.24 user as the Ubuntu developers do not contribute their fixes/changes
upstream, and their 2.6.24 kernel is very heavily modified from what the kernel.org version
looks like.

It's pretty easy to keep up to date with 2.6.x.y on Debian.

sudo apt-get install ketchup make-kpkg

Keep an unpacked copy of a recent version in your home directory, then:

cd linux-*
ketchup -r
make oldconfig
make-kpkg --rootcmd fakeroot clean
make-kpkg --rootcmd fakeroot kernel_image 
cd ..

and sudo dpkg -i to install the new kernel package.

Haven't tried this on Ubuntu.

It's less than a month old, which makes me less confidant in it than I'd like. Gentoo, at
least, likes to leave a kernel marked as "testing" for 3 weeks to collect bug reports before
deciding what needs to be followed up on before marking it "stable". In general, this has more
to do with getting other packages updated for changes (e.g., /sys/block/* being symlinks) than
kernel bugs. Of course, these aren't something to complain to kernel developers about, but
neither can they be ignored when choosing a kernel version. For example, old versions of
x11-drm use flush_agp_mappings, which has been removed in 2.6.25, while newer versions of
x11-drm don't work for some people. So, if you happen to have particular hardware and
particular external module versions for it, you have to wait for stuff to happen that's
outside of the kernel process before you can switch to 2.6.25.

Regardless of the internal quality of a kernel release, there's the question of whether it
will trigger bugs in other packages. FWIW, Gentoo just released a 2.6.24-based kernel last
night, backporting at least one of the security fixes from 2.6.25.3.

2.6.24 just felt uncomfortable with all those changes that gone into libata—and as such, the
reports that come trickling in on lkml.
There were some 100 more patches (287) between 2.6.23–2.6.24 than 2.6.22–2.6.23 (138)
2.6.24–2.6.25 (189).