update checkpatch to version 0.20

Following this email is the next batch of updates to checkpatch.
This marks a change in how I intend to send updates.  The previous
"single delta" update caused us to lose a patch from Greg KH (which is
re-included in this update), bad.  By sending the actual patches we avoid
regressing anything I have missed, it also has the advantage of maintaining
individual changes to carry signoff more accurately and generally being
more transparent.

This version brings a few new checks and the usual slew of fixes for
false positives mostly in type detection.  Of note:

  - allow printk text strings to bust the width restrictions for
    grepability,
  - major fixes for possible type/modifier handling,
  - trailing statement checks for case and default,
  - trailing statement checks for for and while,
  - spacing checks for square brackets '[' ']',
  - conditions/loops indent checks, and
  - checks for new use of __initcall.

Complete changelog below.

-apw

Andy Whitcroft (20):
      checkpatch: Version: 0.20
      checkpatch: return is not a function -- parentheses for casts are ok too
      checkpatch: types: some types may also be identifiers
      checkpatch: possible types: __asm__ is never a type
      checkpatch: comment detection: ignore macro continuation when
	      detecting associated comments
      checkpatch: types: unary -- goto introduces unary context
      checkpatch: macros: fix statement counting block end detection
      checkpatch: trailing statement indent: fix end of statement location
      checkpatch: allow printk strings to exceed 80 characters to
	      maintain their searchability
      checkpatch: switch -- report trailing statements on case and default
      checkpatch: check spacing for square brackets
      checkpatch: toughen trailing if statement checks and extend them
	      to while and for
      checkpatch: condition/loop indent checks
      checkpatch: allow for type modifiers on multiple declarations
      checkpatch: improve type matcher debug
      checkpatch: possible modifiers are not being correctly matched
      checkpatch: macro complexity checks are meaningless in linker scripts
      checkpatch: handle return types of pointers to functions
      checkpatch: possible types -- known modifiers cannot be types
      checkpatch: possible modifiers -- handle multiple modifiers and trailing

Greg Kroah-Hartman (1):
      checkpatch: usb_free_urb() can take NULL

Michael Ellerman (1):
      checkpatch: add a checkpatch warning for new uses of __initcall().

Wolfram Sang (1):
      checkpatch: correct spelling in kfree checks