Re: arm long_call support patch.

[Nick Clifton <nickc at cygnus dot com>]

Hi Dimitri,

  Thanks once again for taking the time to rework and resubmit this
  patch.  I am sure that you will be pleased to hear that I have
  decided to accept your patch this time, although I have had to make
  a few small modifications on the way:

  * The #pragma [no_]long_calls support had no way to turn off the
    pragma's behaviour.  (#pragma no_long_calls forced all calls to be
    considered as short calls, rather than just stopping them from
    being considered as long calls).  So I added a third pragma
    definition, "#pragma long_calls_off", to restore the default
    behaviour.

  * The use of saveable_obstack is obsolete I believe.  Instead the
    function ggc_alloc_string() should be used (in this case), so I
    switched the code over to use that.

  * Your choice of '@' as the prefix character for short call
    function's clashes with the use of this character for DLL imported
    and exported functions in the arm-pe toolchain.  I changed the
    short call prefix character to '^' and also fixed
    STRIP_NAME_ENCODING so that it would strip both kinds of prefix.

Cheers
	Nick