[Hardy] Update kernel to Linux 2.6.24.4

This is a tracking report for the merge with upstream stable kernel 2.6.24.4. The following updates are included:

  [Upstream Kernel Changes]

  * NET: Fix race in dev_close(). (Bug 9750)
  * IPV6: Fix IPsec datagram fragmentation
  * IPV6: dst_entry leak in ip4ip6_err.
  * IPV4: Remove IP_TOS setting privilege checks.
  * IPCONFIG: The kernel gets no IP from some DHCP servers
  * IPCOMP: Disable BH on output when using shared tfm
  * IRQ_NOPROBE helper functions
  * MIPS: Mark all but i8259 interrupts as no-probe.
  * ub: fix up the conversion to sg_init_table()
  * x86: adjust enable_NMI_through_LVT0()
  * SCSI ips: handle scsi_add_host() failure, and other err cleanups
  * CRYPTO xcbc: Fix crash with IPsec
  * CRYPTO xts: Use proper alignment
  * SCSI ips: fix data buffer accessors conversion bug
  * SCSI aic94xx: fix REQ_TASK_ABORT and REQ_DEVICE_RESET
  * x86: replace LOCK_PREFIX in futex.h
  * ARM pxa: fix clock lookup to find specific device clocks
  * futex: fix init order
  * futex: runtime enable pi and robust functionality
  * file capabilities: simplify signal check
  * hugetlb: ensure we do not reference a surplus page after handing it to
    buddy
  * ufs: fix parenthesisation in ufs_set_fs_state()
  * spi: pxa2xx_spi clock polarity fix
  * NETFILTER: Fix incorrect use of skb_make_writable
  * NETFILTER: fix ebtable targets return
  * SCSI advansys: fix overrun_buf aligned bug
  * pata_hpt*, pata_serverworks: fix UDMA masking
  * moduleparam: fix alpha, ia64 and ppc64 compile failures
  * PCI x86: always use conf1 to access config space below 256 bytes
  * e1000e: Fix CRC stripping in hardware context bug
  * atmel_spi: fix clock polarity
  * x86: move out tick_nohz_stop_sched_tick() call from the loop
  * macb: Fix speed setting
  * ioat: fix 'ack' handling, driver must ensure that 'ack' is zero
  * VT notifier fix for VT switch
  * USB: ftdi_sio: Workaround for broken Matrix Orbital serial port
  * USB: ftdi_sio - really enable EM1010PC
  * SCSI: fix BUG when sum(scatterlist) > bufflen
  * x86: don't use P6_NOPs if compiling with CONFIG_X86_GENERIC
  * Fix default compose table initialization
  * SCSI: gdth: bugfix for the at-exit problems
  * sched: fix race in schedule()
  * nfsd: fix oops on access from high-numbered ports
  * sched_nr_migrate wrong mode bits
  * NETFILTER: xt_time: fix failure to match on Sundays
  * NETFILTER: nfnetlink_queue: fix computation of allocated size for
    netlink skb
  * NETFILTER: nfnetlink_log: fix computation of netlink skb size
  * zisofs: fix readpage() outside i_size
  * jbd2: correctly unescape journal data blocks
  * jbd: correctly unescape journal data blocks
  * aio: bad AIO race in aio_complete() leads to process hang
  * async_tx: avoid the async xor_zero_sum path when src_cnt >
    device->max_xor
  * SCSI advansys: Fix bug in AdvLoadMicrocode
  * BLUETOOTH: Fix bugs in previous conn add/del workqueue changes.
  * relay: fix subbuf_splice_actor() adding too many pages
  * slab: NUMA slab allocator migration bugfix
  * S390 futex: let futex_atomic_cmpxchg_pt survive early functional tests.