Re: [RFA] Fix mips32_relative_offset()

[Andrew Cagney <cagney at gnu dot org>]

Mark Kettenis wrote:
> The current code assumes that sizeof(long) == 32, which obviously
> isn't true on LP64 platforms.  The attached patch fixes that by using
> the sign-extension idiom that's also used for SPARC.  It still assumes
> that a long can hold a 18-bit signed integer, but ISO C guarantees
> that it can.
>
> OK?

Yes.

> P.S. Incidentally I think the function should return a LONGEST instead
>      of CORE_ADDR.

and take a ULONGEST as input.  If you want to do that, ok.

Andrew

> Index: ChangeLog
> from  Mark Kettenis  <kettenis@gnu.org>
>
> 	* mips-tdep.c (mips32_relative_offset): Avoid ussumption that
> 	sizeof(long) == 32.
>
> Index: mips-tdep.c
> ===================================================================
> RCS file: /cvs/src/src/gdb/mips-tdep.c,v
> retrieving revision 1.362
> diff -u -p -r1.362 mips-tdep.c
> --- mips-tdep.c 31 Oct 2004 21:51:58 -0000 1.362
> +++ mips-tdep.c 9 Nov 2004 12:51:40 -0000
> @@ -881,14 +881,8 @@ mips_fetch_instruction (CORE_ADDR addr)
>  static CORE_ADDR
>  mips32_relative_offset (unsigned long inst)
>  {
> -  long x;
> -  x = itype_immediate (inst);
> -  if (x & 0x8000)		/* sign bit set */
> -    {
> -      x |= 0xffff0000;		/* sign extension */
> -    }
> -  x = x << 2;
> -  return x;
> +  long x = ((itype_immediate (inst) ^ 0x8000) - 0x8000);
> +  return x << 2;
>  }
>  
>  /* Determine whate to set a single step breakpoint while considering