SIM: Handle XScale COFF binaries

SIM: Handle XScale COFF binaries

[Nick Clifton]

Hi Guys,

  I have checked in the following patch.  It fixes the ARM simulator
  so that COFF based XScale binaries will be correctly simulated.  The
  problem was that the COFF header did not have enough bits to
  distinguish between the v5, v5T, v5TE and XScale architectures, and
  so it was defaulting to v5, disabling the simulation of the XScale
  instructions.

Cheers
        Nick

2002-02-04  Nick Clifton  <nickc@cambridge.redhat.com>

	* wrapper.c: If a v5 architecture is detected, assume it might be
	an XScale binary, since there is no way to distinguish between
	the two in the COFF file format.

Index: sim/arm/wrapper.c
===================================================================
RCS file: /cvs/src/src/sim/arm/wrapper.c,v
retrieving revision 1.15
retrieving revision 1.16
diff -c -3 -p -w -r1.15 -r1.16
*** wrapper.c	2002/01/09 15:08:21	1.15
--- wrapper.c	2002/02/04 16:27:22	1.16
*************** sim_create_inferior (sd, abfd, argv, env
*** 234,239 ****
--- 234,248 ----
        break;
  
      case bfd_mach_arm_5:
+       /* This is a special case in order to support COFF based ARM toolchains.
+ 	 The COFF header does not have enough room to store all the different
+ 	 kinds of ARM cpu, so the XScale, v5T and v5TE architectures all default
+ 	 to v5.  (See coff_set_flags() in bdf/coffcode.h).  So if we see a v5
+ 	 machine type here, we assume it could be any of the above architectures
+ 	 and so select the most feature-full.  */
+       ARMul_SelectProcessor (state, ARM_v5_Prop | ARM_v5e_Prop | ARM_XScale_Prop);
+       break;
+ 
      case bfd_mach_arm_5T:
        ARMul_SelectProcessor (state, ARM_v5_Prop);
        break;

Re: SIM: Handle XScale COFF binaries

[Richard Earnshaw]

> Hi Guys,
> 
>   I have checked in the following patch.  It fixes the ARM simulator
>   so that COFF based XScale binaries will be correctly simulated.  The
>   problem was that the COFF header did not have enough bits to
>   distinguish between the v5, v5T, v5TE and XScale architectures, and
>   so it was defaulting to v5, disabling the simulation of the XScale
>   instructions.
> 
> Cheers
>         Nick
> 
> 2002-02-04  Nick Clifton  <nickc@cambridge.redhat.com>
> 
> 	* wrapper.c: If a v5 architecture is detected, assume it might be
> 	an XScale binary, since there is no way to distinguish between
> 	the two in the COFF file format.

UG.

Can't we restrict this so that it only happens if we have COFF files (as 
it stands ELF files will do the same as well).

R.