[PATCH] 64-bit support for Irix 6

[PATCH] 64-bit support for Irix 6

[Kevin Buettner]

I've just committed the change below.

This change (along with some other changes coming soon) allow us to
debug 64-bit applications on Irix 6.X.  I've verified that this change
does not affect our ability to debug o32 and n32 applications.

	* config/mips/tm-irix6.h (MIPS_REGSIZE): Define to be 8.
	(REGISTER_VIRTUAL_TYPE): Some registers are now 64 bits wide.

Index: config/mips/tm-irix6.h
===================================================================
RCS file: /cvs/src/src/gdb/config/mips/tm-irix6.h,v
retrieving revision 1.2
diff -u -p -r1.2 tm-irix6.h
--- config/mips/tm-irix6.h	11 Jun 2002 17:51:56 -0000	1.2
+++ config/mips/tm-irix6.h	31 Jul 2002 17:04:03 -0000
@@ -21,6 +21,9 @@
 
 #include "mips/tm-bigmips.h"
 
+#undef MIPS_REGSIZE
+#define MIPS_REGSIZE 8
+
 /* SGI's assembler doesn't grok dollar signs in identifiers.
    So we use dots instead.  This item must be coordinated with G++. */
 #undef CPLUS_MARKER
@@ -89,11 +92,12 @@
       32 * sizeof(double) + ((N) - 32) * MIPS_REGSIZE)
 
 #undef  REGISTER_VIRTUAL_TYPE
+/* define 8 byte register type */
 #define REGISTER_VIRTUAL_TYPE(N) \
 	(((N) >= FP0_REGNUM && (N) < FP0_REGNUM+32) ? builtin_type_double \
 	 : ((N) == 32 /*SR*/) ? builtin_type_uint32 \
 	 : ((N) >= 70 && (N) <= 89) ? builtin_type_uint32 \
-	 : builtin_type_int)
+	 : builtin_type_long_long)
 
 /* Force N32 ABI as the default. */
 #define MIPS_DEFAULT_ABI MIPS_ABI_N32

Re: [PATCH] 64-bit support for Irix 6

[Andrew Cagney]

>  
>  #include "mips/tm-bigmips.h"
>  
> +#undef MIPS_REGSIZE
> +#define MIPS_REGSIZE 8
> +
>  /* SGI's assembler doesn't grok dollar signs in identifiers.
>     So we use dots instead.  This item must be coordinated with G++. */
>  #undef CPLUS_MARKER
> @@ -89,11 +92,12 @@
>        32 * sizeof(double) + ((N) - 32) * MIPS_REGSIZE)
>  
>  #undef  REGISTER_VIRTUAL_TYPE
> +/* define 8 byte register type */
>  #define REGISTER_VIRTUAL_TYPE(N) \
>  	(((N) >= FP0_REGNUM && (N) < FP0_REGNUM+32) ? builtin_type_double \
>  	 : ((N) == 32 /*SR*/) ? builtin_type_uint32 \
>  	 : ((N) >= 70 && (N) <= 89) ? builtin_type_uint32 \
> -	 : builtin_type_int)
> +	 : builtin_type_long_long)
>  
>  /* Force N32 ABI as the default. */
>  #define MIPS_DEFAULT_ABI MIPS_ABI_N32
> 

Kevin, this is wrong.

It is no longer acceptable to add support for an additional sub-target 
using non-multi-arch mechanisms.

Andrew

Re: [PATCH] 64-bit support for Irix 6

[Kevin Buettner]

On Aug 1, 11:11am, Andrew Cagney wrote:

> Kevin, this is wrong.
> 
> It is no longer acceptable to add support for an additional sub-target 
> using non-multi-arch mechanisms.

How about this then?

	* config/mips/tm-irix6.h: Include mips/tm-bigmips64.h instead of
	mips/tm-bigmips.h.
	(MIPS_REGSIZE, REGISTER_VIRTUAL_SIZE): Delete.

Index: config/mips/tm-irix6.h
===================================================================
RCS file: /cvs/src/src/gdb/config/mips/tm-irix6.h,v
retrieving revision 1.3
diff -u -p -r1.3 tm-irix6.h
--- config/mips/tm-irix6.h	31 Jul 2002 17:22:06 -0000	1.3
+++ config/mips/tm-irix6.h	1 Aug 2002 16:33:24 -0000
@@ -19,10 +19,7 @@
    Foundation, Inc., 59 Temple Place - Suite 330,
    Boston, MA 02111-1307, USA.  */
 
-#include "mips/tm-bigmips.h"
-
-#undef MIPS_REGSIZE
-#define MIPS_REGSIZE 8
+#include "mips/tm-bigmips64.h"
 
 /* SGI's assembler doesn't grok dollar signs in identifiers.
    So we use dots instead.  This item must be coordinated with G++. */
@@ -90,14 +87,6 @@
       FP0_REGNUM * MIPS_REGSIZE + \
       ((N) - FP0_REGNUM) * sizeof(double) : \
       32 * sizeof(double) + ((N) - 32) * MIPS_REGSIZE)
-
-#undef  REGISTER_VIRTUAL_TYPE
-/* define 8 byte register type */
-#define REGISTER_VIRTUAL_TYPE(N) \
-	(((N) >= FP0_REGNUM && (N) < FP0_REGNUM+32) ? builtin_type_double \
-	 : ((N) == 32 /*SR*/) ? builtin_type_uint32 \
-	 : ((N) >= 70 && (N) <= 89) ? builtin_type_uint32 \
-	 : builtin_type_long_long)
 
 /* Force N32 ABI as the default. */
 #define MIPS_DEFAULT_ABI MIPS_ABI_N32

Re: [PATCH] 64-bit support for Irix 6

[Andrew Cagney]

> On Aug 1, 11:11am, Andrew Cagney wrote:
> 
> 
>> Kevin, this is wrong.
>> 
>> It is no longer acceptable to add support for an additional sub-target 
>> using non-multi-arch mechanisms.
> 
> 
> How about this then?
> 
> 	* config/mips/tm-irix6.h: Include mips/tm-bigmips64.h instead of
> 	mips/tm-bigmips.h.
> 	(MIPS_REGSIZE, REGISTER_VIRTUAL_SIZE): Delete.

Even though it removes a macro, I think it is still fudging things.  The 
patch should be converting REGISTER_VIRTUAL_TYPE() to a function and 
then adding it to the multi-arch vector.

Add an #undef to tm-irix6.h.

Andrew


> Index: config/mips/tm-irix6.h
> ===================================================================
> RCS file: /cvs/src/src/gdb/config/mips/tm-irix6.h,v
> retrieving revision 1.3
> diff -u -p -r1.3 tm-irix6.h
> --- config/mips/tm-irix6.h	31 Jul 2002 17:22:06 -0000	1.3
> +++ config/mips/tm-irix6.h	1 Aug 2002 16:33:24 -0000
> @@ -19,10 +19,7 @@
>     Foundation, Inc., 59 Temple Place - Suite 330,
>     Boston, MA 02111-1307, USA.  */
>  
> -#include "mips/tm-bigmips.h"
> -
> -#undef MIPS_REGSIZE
> -#define MIPS_REGSIZE 8
> +#include "mips/tm-bigmips64.h"
>  
>  /* SGI's assembler doesn't grok dollar signs in identifiers.
>     So we use dots instead.  This item must be coordinated with G++. */
> @@ -90,14 +87,6 @@
>        FP0_REGNUM * MIPS_REGSIZE + \
>        ((N) - FP0_REGNUM) * sizeof(double) : \
>        32 * sizeof(double) + ((N) - 32) * MIPS_REGSIZE)
> -
> -#undef  REGISTER_VIRTUAL_TYPE
> -/* define 8 byte register type */
> -#define REGISTER_VIRTUAL_TYPE(N) \
> -	(((N) >= FP0_REGNUM && (N) < FP0_REGNUM+32) ? builtin_type_double \
> -	 : ((N) == 32 /*SR*/) ? builtin_type_uint32 \
> -	 : ((N) >= 70 && (N) <= 89) ? builtin_type_uint32 \
> -	 : builtin_type_long_long)
>  
>  /* Force N32 ABI as the default. */
>  #define MIPS_DEFAULT_ABI MIPS_ABI_N32
> 
> 

Re: [PATCH] 64-bit support for Irix 6

[Kevin Buettner]

On Aug 1,  1:42pm, Andrew Cagney wrote:

> Even though it removes a macro, I think it is still fudging things.

Okay.

> The patch should be converting REGISTER_VIRTUAL_TYPE() to a function and 
> then adding it to the multi-arch vector.
> 
> Add an #undef to tm-irix6.h.

I've committed the change below.  As the MIPS maintainer, you may want
to ponder my note in the mips_register_virtual_type() comment.  (It
doesn't affect Irix because (for good or ill) Irix redefines
FP_REGNUM.)

	* mips-tdep.c (mips_register_virtual_type): New function.
	(mips_gdbarch_init): Register mips_register_virtual_type()
	with gdbarch machinery.
	* config/mips/tm-irix6.h (mips/tm-bigmips64.h): Include
	this file instead of tm-bigmips.h.
	(MIPS_REGSIZE): Delete this macro.
	(REGISTER_VIRTUAL_TYPE): Delete macro.  Undef macro so that
	multiarch version in mips-tdep.c will be found.

Index: mips-tdep.c
===================================================================
RCS file: /cvs/src/src/gdb/mips-tdep.c,v
retrieving revision 1.85
diff -u -p -r1.85 mips-tdep.c
--- mips-tdep.c	31 Jul 2002 20:26:49 -0000	1.85
+++ mips-tdep.c	1 Aug 2002 19:49:59 -0000
@@ -483,6 +483,36 @@ mips_register_convert_to_raw (struct typ
 	    TYPE_LENGTH (virtual_type));
 }
 
+/* Return the GDB type object for the "standard" data type
+   of data in register REG.  
+   
+   Note: kevinb/2002-08-01: The definition below should faithfully
+   reproduce the behavior of each of the REGISTER_VIRTUAL_TYPE
+   definitions found in config/mips/tm-*.h.  I'm concerned about
+   the ``FCRCS_REGNUM <= reg && reg <= LAST_EMBED_REGNUM'' clause
+   though.  In some cases FP_REGNUM is in this range, and I doubt
+   that this code is correct for the 64-bit case.  */
+
+static struct type *
+mips_register_virtual_type (int reg)
+{
+  if (FP0_REGNUM <= reg && reg < FP0_REGNUM + 32)
+    return builtin_type_double;
+  else if (reg == PS_REGNUM /* CR */)
+    return builtin_type_uint32;
+  else if (FCRCS_REGNUM <= reg && reg <= LAST_EMBED_REGNUM)
+    return builtin_type_uint32;
+  else
+    {
+      /* Everything else... return ``long long'' when registers
+         are 64-bits wide, ``int'' otherwise.  */
+      if (MIPS_REGSIZE == TYPE_LENGTH (builtin_type_long_long))
+	return builtin_type_long_long;
+      else
+	return builtin_type_int;
+    }
+}
+
 /* Should the upper word of 64-bit addresses be zeroed? */
 enum auto_boolean mask_address_var = AUTO_BOOLEAN_AUTO;
 
@@ -4634,6 +4664,8 @@ mips_gdbarch_init (struct gdbarch_info i
   set_gdbarch_pointer_to_address (gdbarch, signed_pointer_to_address);
   set_gdbarch_address_to_pointer (gdbarch, address_to_signed_pointer);
   set_gdbarch_integer_to_address (gdbarch, mips_integer_to_address);
+
+  set_gdbarch_register_virtual_type (gdbarch, mips_register_virtual_type);
 
   /* Hook in OS ABI-specific overrides, if they have been registered.  */
   gdbarch_init_osabi (info, gdbarch, osabi);
Index: config/mips/tm-irix6.h
===================================================================
RCS file: /cvs/src/src/gdb/config/mips/tm-irix6.h,v
retrieving revision 1.3
diff -u -p -r1.3 tm-irix6.h
--- config/mips/tm-irix6.h	31 Jul 2002 17:22:06 -0000	1.3
+++ config/mips/tm-irix6.h	1 Aug 2002 19:49:59 -0000
@@ -19,10 +19,7 @@
    Foundation, Inc., 59 Temple Place - Suite 330,
    Boston, MA 02111-1307, USA.  */
 
-#include "mips/tm-bigmips.h"
-
-#undef MIPS_REGSIZE
-#define MIPS_REGSIZE 8
+#include "mips/tm-bigmips64.h"
 
 /* SGI's assembler doesn't grok dollar signs in identifiers.
    So we use dots instead.  This item must be coordinated with G++. */
@@ -91,14 +88,6 @@
       ((N) - FP0_REGNUM) * sizeof(double) : \
       32 * sizeof(double) + ((N) - 32) * MIPS_REGSIZE)
 
-#undef  REGISTER_VIRTUAL_TYPE
-/* define 8 byte register type */
-#define REGISTER_VIRTUAL_TYPE(N) \
-	(((N) >= FP0_REGNUM && (N) < FP0_REGNUM+32) ? builtin_type_double \
-	 : ((N) == 32 /*SR*/) ? builtin_type_uint32 \
-	 : ((N) >= 70 && (N) <= 89) ? builtin_type_uint32 \
-	 : builtin_type_long_long)
-
 /* Force N32 ABI as the default. */
 #define MIPS_DEFAULT_ABI MIPS_ABI_N32
 
@@ -129,3 +118,6 @@
 /* Select the disassembler */
 #undef TM_PRINT_INSN_MACH
 #define TM_PRINT_INSN_MACH bfd_mach_mips8000
+
+/* Undefine those methods which have been multiarched.  */
+#undef REGISTER_VIRTUAL_TYPE

Re: [PATCH] 64-bit support for Irix 6

[Andrew Cagney]

> 
> I've committed the change below.

Where is the fire?

>    set_gdbarch_integer_to_address (gdbarch, mips_integer_to_address);
> +
> +  set_gdbarch_register_virtual_type (gdbarch, mips_register_virtual_type);
>  

Can you please add a comment here explaining that many MIPS targets do 
not yet use this method because they are still defining the macro.

As a separate patch, could you please update things so that:

> +static struct type *
> +mips_register_virtual_type (int reg)
> +{
> +  if (FP0_REGNUM <= reg && reg < FP0_REGNUM + 32)
> +    return builtin_type_double;

This returns ieee_double big/little.

> +  else if (reg == PS_REGNUM /* CR */)
> +    return builtin_type_uint32;
> +  else if (FCRCS_REGNUM <= reg && reg <= LAST_EMBED_REGNUM)
> +    return builtin_type_uint32;
> +  else
> +    {
> +      /* Everything else... return ``long long'' when registers
> +         are 64-bits wide, ``int'' otherwise.  */
> +      if (MIPS_REGSIZE == TYPE_LENGTH (builtin_type_long_long))
> +	return builtin_type_long_long;

This returns builtin_type_uint64.

> +      else

> +	return builtin_type_int;

This returns builtin_type_uint32.

Andrew

Re: [PATCH] 64-bit support for Irix 6

[Michael Snyder]

Andrew Cagney wrote:
> 
> >
> >  #include "mips/tm-bigmips.h"
> >
> > +#undef MIPS_REGSIZE
> > +#define MIPS_REGSIZE 8
> > +
> >  /* SGI's assembler doesn't grok dollar signs in identifiers.
> >     So we use dots instead.  This item must be coordinated with G++. */
> >  #undef CPLUS_MARKER
> > @@ -89,11 +92,12 @@
> >        32 * sizeof(double) + ((N) - 32) * MIPS_REGSIZE)
> >
> >  #undef  REGISTER_VIRTUAL_TYPE
> > +/* define 8 byte register type */
> >  #define REGISTER_VIRTUAL_TYPE(N) \
> >       (((N) >= FP0_REGNUM && (N) < FP0_REGNUM+32) ? builtin_type_double \
> >        : ((N) == 32 /*SR*/) ? builtin_type_uint32 \
> >        : ((N) >= 70 && (N) <= 89) ? builtin_type_uint32 \
> > -      : builtin_type_int)
> > +      : builtin_type_long_long)
> >
> >  /* Force N32 ABI as the default. */
> >  #define MIPS_DEFAULT_ABI MIPS_ABI_N32
> >
> 
> Kevin, this is wrong.
> 
> It is no longer acceptable to add support for an additional sub-target
> using non-multi-arch mechanisms.

Does that mean that if we touch a macro, we have to multi-arch it?
For instance, if we touch the PUSH_ARGUMENTS function, and
PUSH_ARGUMENTS 
hasn't been multi-arched, do we have to multi-arch it?

Re: [PATCH] 64-bit support for Irix 6

[Andrew Cagney]

> Andrew Cagney wrote:
> 
>> 
> 
>> >
>> >  #include "mips/tm-bigmips.h"
>> >
>> > +#undef MIPS_REGSIZE
>> > +#define MIPS_REGSIZE 8
>> > +
>> >  /* SGI's assembler doesn't grok dollar signs in identifiers.
>> >     So we use dots instead.  This item must be coordinated with G++. */
>> >  #undef CPLUS_MARKER
>> > @@ -89,11 +92,12 @@
>> >        32 * sizeof(double) + ((N) - 32) * MIPS_REGSIZE)
>> >
>> >  #undef  REGISTER_VIRTUAL_TYPE
>> > +/* define 8 byte register type */
>> >  #define REGISTER_VIRTUAL_TYPE(N) \
>> >       (((N) >= FP0_REGNUM && (N) < FP0_REGNUM+32) ? builtin_type_double \
>> >        : ((N) == 32 /*SR*/) ? builtin_type_uint32 \
>> >        : ((N) >= 70 && (N) <= 89) ? builtin_type_uint32 \
>> > -      : builtin_type_int)
>> > +      : builtin_type_long_long)
>> >
>> >  /* Force N32 ABI as the default. */
>> >  #define MIPS_DEFAULT_ABI MIPS_ABI_N32
>> >
> 
>> 
>> Kevin, this is wrong.
>> 
>> It is no longer acceptable to add support for an additional sub-target
>> using non-multi-arch mechanisms.
> 
> 
> Does that mean that if we touch a macro, we have to multi-arch it?

If you want to fix a bug in a macro then you should first (separatly) 
convert it to a macro/function pair.  I believe this is accepted as 
``current best pratice''.

If you want to change a bunch of macros to support an additional 
architecture variant then they need to be multi-arch.  GDB no longer 
accepts an architecture variants that isn't multi-arch.

> For instance, if we touch the PUSH_ARGUMENTS function, and
> PUSH_ARGUMENTS 
> hasn't been multi-arched, do we have to multi-arch it?

If you mean mips_push_arguments, then that is already part of the 
multi-arch vector.  Just don't forget that any changes to that function 
should be multi-arch friendly.

enjoy,
Andrew

Re: [PATCH] 64-bit support for Irix 6

[Michael Snyder]

Andrew Cagney wrote:
> 
> > Andrew Cagney wrote:
> >
> >>
> >
> >> >
> >> >  #include "mips/tm-bigmips.h"
> >> >
> >> > +#undef MIPS_REGSIZE
> >> > +#define MIPS_REGSIZE 8
> >> > +
> >> >  /* SGI's assembler doesn't grok dollar signs in identifiers.
> >> >     So we use dots instead.  This item must be coordinated with G++. */
> >> >  #undef CPLUS_MARKER
> >> > @@ -89,11 +92,12 @@
> >> >        32 * sizeof(double) + ((N) - 32) * MIPS_REGSIZE)
> >> >
> >> >  #undef  REGISTER_VIRTUAL_TYPE
> >> > +/* define 8 byte register type */
> >> >  #define REGISTER_VIRTUAL_TYPE(N) \
> >> >       (((N) >= FP0_REGNUM && (N) < FP0_REGNUM+32) ? builtin_type_double \
> >> >        : ((N) == 32 /*SR*/) ? builtin_type_uint32 \
> >> >        : ((N) >= 70 && (N) <= 89) ? builtin_type_uint32 \
> >> > -      : builtin_type_int)
> >> > +      : builtin_type_long_long)
> >> >
> >> >  /* Force N32 ABI as the default. */
> >> >  #define MIPS_DEFAULT_ABI MIPS_ABI_N32
> >> >
> >
> >>
> >> Kevin, this is wrong.
> >>
> >> It is no longer acceptable to add support for an additional sub-target
> >> using non-multi-arch mechanisms.
> >
> >
> > Does that mean that if we touch a macro, we have to multi-arch it?
> 
> If you want to fix a bug in a macro then you should first (separatly)
> convert it to a macro/function pair.  I believe this is accepted as
> ``current best pratice''.
> 
> If you want to change a bunch of macros to support an additional
> architecture variant then they need to be multi-arch.  GDB no longer
> accepts an architecture variants that isn't multi-arch.
> 
> > For instance, if we touch the PUSH_ARGUMENTS function, and
> > PUSH_ARGUMENTS
> > hasn't been multi-arched, do we have to multi-arch it?
> 
> If you mean mips_push_arguments, then that is already part of the
> multi-arch vector. 

Umm... no it isn't.  I can see why you think it is
(set_gdbarch_push_arguments is called), but the call to
PUSH_ARGUMENTS is still going thru the macro in tm-mips.h.

> Just don't forget that any changes to that function
> should be multi-arch friendly.

So I can use code similar to what is already there
(eg.  "if (tdep->mips_abi == MIPS_ABI_N32)")
rather than splitting the function into variously
mips_n32_push_argument etc.?

Re: [PATCH] 64-bit support for Irix 6

[Kevin Buettner]

On Aug 1,  4:28pm, Andrew Cagney wrote:

> > +  set_gdbarch_register_virtual_type (gdbarch, mips_register_virtual_type);
> 
> Can you please add a comment here explaining that many MIPS targets do 
> not yet use this method because they are still defining the macro.

Done.  (Separate patch committed a little while ago.)

> As a separate patch, could you please update things so that:
> 
> > +static struct type *
> > +mips_register_virtual_type (int reg)
> > +{
> > +  if (FP0_REGNUM <= reg && reg < FP0_REGNUM + 32)
> > +    return builtin_type_double;
> 
> This returns ieee_double big/little.
> 
> > +  else if (reg == PS_REGNUM /* CR */)
> > +    return builtin_type_uint32;
> > +  else if (FCRCS_REGNUM <= reg && reg <= LAST_EMBED_REGNUM)
> > +    return builtin_type_uint32;
> > +  else
> > +    {
> > +      /* Everything else... return ``long long'' when registers
> > +         are 64-bits wide, ``int'' otherwise.  */
> > +      if (MIPS_REGSIZE == TYPE_LENGTH (builtin_type_long_long))
> > +	return builtin_type_long_long;
> 
> This returns builtin_type_uint64.
> 
> > +      else
> 
> > +	return builtin_type_int;
> 
> This returns builtin_type_uint32.

Done.  See below:

	* mips-tdep.c (mips_register_virtual_type): Use architecture
	invariant return values.

Index: mips-tdep.c
===================================================================
RCS file: /cvs/src/src/gdb/mips-tdep.c,v
retrieving revision 1.87
diff -u -p -r1.87 mips-tdep.c
--- mips-tdep.c	1 Aug 2002 21:36:27 -0000	1.87
+++ mips-tdep.c	1 Aug 2002 23:02:18 -0000
@@ -497,19 +497,25 @@ static struct type *
 mips_register_virtual_type (int reg)
 {
   if (FP0_REGNUM <= reg && reg < FP0_REGNUM + 32)
-    return builtin_type_double;
+    {
+      /* Floating point registers...  */
+      if (TARGET_BYTE_ORDER == BFD_ENDIAN_BIG)
+	return builtin_type_ieee_double_big;
+      else
+	return builtin_type_ieee_double_little;
+    }
   else if (reg == PS_REGNUM /* CR */)
     return builtin_type_uint32;
   else if (FCRCS_REGNUM <= reg && reg <= LAST_EMBED_REGNUM)
     return builtin_type_uint32;
   else
     {
-      /* Everything else... return ``long long'' when registers
-         are 64-bits wide, ``int'' otherwise.  */
-      if (MIPS_REGSIZE == TYPE_LENGTH (builtin_type_long_long))
-	return builtin_type_long_long;
+      /* Everything else...
+         Return type appropriate for width of register.  */
+      if (MIPS_REGSIZE == TYPE_LENGTH (builtin_type_uint64))
+	return builtin_type_uint64;
       else
-	return builtin_type_int;
+	return builtin_type_uint32;
     }
 }
 

Re: [PATCH] 64-bit support for Irix 6

[Andrew Cagney]

> If you mean mips_push_arguments, then that is already part of the
>> multi-arch vector. 
> 
> 
> Umm... no it isn't.  I can see why you think it is
> (set_gdbarch_push_arguments is called), but the call to
> PUSH_ARGUMENTS is still going thru the macro in tm-mips.h.

I think the macro can be yanked.  Not sure why it has remained.

>> Just don't forget that any changes to that function
>> should be multi-arch friendly.
> 
> 
> So I can use code similar to what is already there
> (eg.  "if (tdep->mips_abi == MIPS_ABI_N32)")
> rather than splitting the function into variously
> mips_n32_push_argument etc.?

Adding ``if (.. == MIPS_ABI_N32)'' wouldn't be multi-arch unfriendly.

However, it would definitly be GDB developer/maintainer unfriendly.  The 
MIPS is held up as an example of how to not implement the push arguments 
function.

If the MIPS doesn't support your new ABI then I think you are way better 
off creating a new mips_ABI_push_arguments() function and model its 
implementation on the Arm.  Even cloning mips_push_arguments() and then 
stripping out all the irrelenvant guff would be better.

That way you would be sure it didn't break any of the other ABIs.

enjoy,
Andrew