Merge tag 'upstream-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jgarzik/libata-dev

Pull libata fixes from Jeff Garzik:
 "If you were going to shoot me for not sending these earlier, you would
  be right.  -rc6 beat me by ~2 hours it seems, and they really should
  have gone out long before that.

  These have been in libata-dev.git for a day or so (unfortunately
  linux-next is on vacation).  The main one is #1, with the others being
  minor bits.  #1 has multiple tested-by, and can be considered a
  regression fix IMO.

   1) Fix ACPI oops:

        https://bugzilla.kernel.org/show_bug.cgi?id=48211

   2) Temporary WARN_ONCE() debugging patch for further ACPI debugging.

      The code already oopses here, and so this merely gives slightly
      better info.  Related to

        https://bugzilla.kernel.org/show_bug.cgi?id=49151

      which has been bisected down to a patch that _exposes_ a latest
      bug, but said bisection target does not actually appear to be the
      root cause itself.

   3) sata_svw: fix longstanding error recovery bug, which was
      preventing kdump, by adding missing DMA-start bit check.  Core
      code was already checking DMA-start, but ancillary, less-used
      routines were not.  Fixed.

   4) sata_highbank: fix minor __init/__devinit warning

   5) Fix minor warning, if CONFIG_PM is set, but CONFIG_PM_SLEEP is not
      set

   6) pata_arasan: proper functioning requires clock setting"

* tag 'upstream-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jgarzik/libata-dev:
  [libata] PM callbacks should be conditionally compiled on CONFIG_PM_SLEEP
  sata_svw: check DMA start bit before reset
  libata debugging: Warn when unable to find timing descriptor based on xfer_mode
  sata_highbank: mark ahci_highbank_probe as __devinit
  pata_arasan: Initialize cf clock to 166MHz
  libata-acpi: Fix NULL ptr derference in ata_acpi_dev_handle