GDB broken on MIPS targets with unmarked binaries

GDB broken on MIPS targets with unmarked binaries

[Daniel Jacobowitz]

There's an interesting crash on startup in a mips-elf GDB nowadays... the
code reads:

    default:
      tdep->mips_abi_string = "default";
      tdep->mips_default_saved_regsize = MIPS_REGSIZE;
      tdep->mips_default_stack_argsize = MIPS_REGSIZE;
      tdep->mips_fp_register_double = (REGISTER_VIRTUAL_SIZE (FP0_REGNUM) == 8);
      tdep->mips_last_arg_regnum = A0_REGNUM + 8 - 1;
      tdep->mips_last_fp_arg_regnum = FPA0_REGNUM + 8 - 1;
      tdep->mips_regs_have_home_p = 1;
      tdep->gdb_target_is_mips64 = 0;
      tdep->default_mask_address_p = 0;
      set_gdbarch_long_bit (gdbarch, 32);
      set_gdbarch_ptr_bit (gdbarch, 32);
      set_gdbarch_long_long_bit (gdbarch, 64);
      break;

Something brought in recently, probably by the OS ABI changes if I had to
guess, is causing the check of REGISTER_VIRTUAL_SIZE to fail.  I've no idea
why.  Compiling a mips-elf (not mips-linux, which sets MIPS_DEFAULT_ABI) and
running gdb on an unmarked ELF binary reproduces this.

-- 
Daniel Jacobowitz                           Carnegie Mellon University
MontaVista Software                         Debian GNU/Linux Developer

Re: GDB broken on MIPS targets with unmarked binaries

[Andrew Cagney]

> There's an interesting crash on startup in a mips-elf GDB nowadays... the
> code reads:
> 
>     default:
>       tdep->mips_abi_string = "default";
>       tdep->mips_default_saved_regsize = MIPS_REGSIZE;
>       tdep->mips_default_stack_argsize = MIPS_REGSIZE;
>       tdep->mips_fp_register_double = (REGISTER_VIRTUAL_SIZE (FP0_REGNUM) == 8);
>       tdep->mips_last_arg_regnum = A0_REGNUM + 8 - 1;
>       tdep->mips_last_fp_arg_regnum = FPA0_REGNUM + 8 - 1;
>       tdep->mips_regs_have_home_p = 1;
>       tdep->gdb_target_is_mips64 = 0;
>       tdep->default_mask_address_p = 0;
>       set_gdbarch_long_bit (gdbarch, 32);
>       set_gdbarch_ptr_bit (gdbarch, 32);
>       set_gdbarch_long_long_bit (gdbarch, 64);
>       break;
> 
> Something brought in recently, probably by the OS ABI changes if I had to
> guess, is causing the check of REGISTER_VIRTUAL_SIZE to fail.  I've no idea
> why.  Compiling a mips-elf (not mips-linux, which sets MIPS_DEFAULT_ABI) and
> running gdb on an unmarked ELF binary reproduces this.

I'd suspect this:

2002-05-08  Andrew Cagney  <ac131313@redhat.com>

         * gdbarch.sh (init_gdbarch_swap): Do not clear the swap section.
         (clear_gdbarch_swap): New function.
         (initialize_non_multiarch): Call.
         (gdbarch_update_p): Before calling init(), swap out and clear the
         existing architecture.
         * gdbarch.c: Regenerate.

The comment from arm-tdep.c should help explain why the above code is 
broken:

   /* We can't use NUM_REGS nor NUM_PSEUDO_REGS here, since that still
      references the old architecture vector, not the one we are
      building here.  */
   prologue_cache.saved_regs = (CORE_ADDR *)
     xcalloc (1, (sizeof (CORE_ADDR)
                  * (gdbarch_num_regs (gdbarch)
                     + gdbarch_num_pseudo_regs (gdbarch))));

As far as I can tell though, REGISTER_VIRTUAL_SIZE expands to 
REGISTER_VIRTUAL_TYPE which is still a macro?

Andrew

Re: GDB broken on MIPS targets with unmarked binaries

[Daniel Jacobowitz]

On Wed, Jun 05, 2002 at 04:44:21PM -0400, Andrew Cagney wrote:
> >There's an interesting crash on startup in a mips-elf GDB nowadays... the
> >code reads:
> >
> >    default:
> >      tdep->mips_abi_string = "default";
> >      tdep->mips_default_saved_regsize = MIPS_REGSIZE;
> >      tdep->mips_default_stack_argsize = MIPS_REGSIZE;
> >      tdep->mips_fp_register_double = (REGISTER_VIRTUAL_SIZE (FP0_REGNUM) 
> >      == 8);
> >      tdep->mips_last_arg_regnum = A0_REGNUM + 8 - 1;
> >      tdep->mips_last_fp_arg_regnum = FPA0_REGNUM + 8 - 1;
> >      tdep->mips_regs_have_home_p = 1;
> >      tdep->gdb_target_is_mips64 = 0;
> >      tdep->default_mask_address_p = 0;
> >      set_gdbarch_long_bit (gdbarch, 32);
> >      set_gdbarch_ptr_bit (gdbarch, 32);
> >      set_gdbarch_long_long_bit (gdbarch, 64);
> >      break;
> >
> >Something brought in recently, probably by the OS ABI changes if I had to
> >guess, is causing the check of REGISTER_VIRTUAL_SIZE to fail.  I've no idea
> >why.  Compiling a mips-elf (not mips-linux, which sets MIPS_DEFAULT_ABI) 
> >and
> >running gdb on an unmarked ELF binary reproduces this.
> 
> I'd suspect this:
> 
> 2002-05-08  Andrew Cagney  <ac131313@redhat.com>
> 
>         * gdbarch.sh (init_gdbarch_swap): Do not clear the swap section.
>         (clear_gdbarch_swap): New function.
>         (initialize_non_multiarch): Call.
>         (gdbarch_update_p): Before calling init(), swap out and clear the
>         existing architecture.
>         * gdbarch.c: Regenerate.
> 
> The comment from arm-tdep.c should help explain why the above code is 
> broken:
> 
>   /* We can't use NUM_REGS nor NUM_PSEUDO_REGS here, since that still
>      references the old architecture vector, not the one we are
>      building here.  */
>   prologue_cache.saved_regs = (CORE_ADDR *)
>     xcalloc (1, (sizeof (CORE_ADDR)
>                  * (gdbarch_num_regs (gdbarch)
>                     + gdbarch_num_pseudo_regs (gdbarch))));
> 
> As far as I can tell though, REGISTER_VIRTUAL_SIZE expands to 
> REGISTER_VIRTUAL_TYPE which is still a macro?

The expansion appears to be:
tdep->mips_fp_register_double = (((((38) >= 38 && (38) < 38 +32) ?
	builtin_type_float : ((38) == 32 ) ?
        builtin_type_uint32 : ((38) >= 70 && (38) <= 89) ?
	builtin_type_uint32 :
        builtin_type_int))->main_type->length == 8);

I can't see any reason for it to crash; I haven't reproduced this
locally yet, although I'm working on it.  The above expansion came from
the second person to mention this to me.

-- 
Daniel Jacobowitz                           Carnegie Mellon University
MontaVista Software                         Debian GNU/Linux Developer

Re: GDB broken on MIPS targets with unmarked binaries

[Daniel Jacobowitz]

On Wed, Jun 05, 2002 at 04:58:34PM -0400, Daniel Jacobowitz wrote:
> On Wed, Jun 05, 2002 at 04:44:21PM -0400, Andrew Cagney wrote:
> > >There's an interesting crash on startup in a mips-elf GDB nowadays... the
> > >code reads:
> > >
> > >    default:
> > >      tdep->mips_abi_string = "default";
> > >      tdep->mips_default_saved_regsize = MIPS_REGSIZE;
> > >      tdep->mips_default_stack_argsize = MIPS_REGSIZE;
> > >      tdep->mips_fp_register_double = (REGISTER_VIRTUAL_SIZE (FP0_REGNUM) 
> > >      == 8);
> > >      tdep->mips_last_arg_regnum = A0_REGNUM + 8 - 1;
> > >      tdep->mips_last_fp_arg_regnum = FPA0_REGNUM + 8 - 1;
> > >      tdep->mips_regs_have_home_p = 1;
> > >      tdep->gdb_target_is_mips64 = 0;
> > >      tdep->default_mask_address_p = 0;
> > >      set_gdbarch_long_bit (gdbarch, 32);
> > >      set_gdbarch_ptr_bit (gdbarch, 32);
> > >      set_gdbarch_long_long_bit (gdbarch, 64);
> > >      break;
> > >
> > >Something brought in recently, probably by the OS ABI changes if I had to
> > >guess, is causing the check of REGISTER_VIRTUAL_SIZE to fail.  I've no idea
> > >why.  Compiling a mips-elf (not mips-linux, which sets MIPS_DEFAULT_ABI) 
> > >and
> > >running gdb on an unmarked ELF binary reproduces this.
> > 
> > I'd suspect this:
> > 
> > 2002-05-08  Andrew Cagney  <ac131313@redhat.com>
> > 
> >         * gdbarch.sh (init_gdbarch_swap): Do not clear the swap section.
> >         (clear_gdbarch_swap): New function.
> >         (initialize_non_multiarch): Call.
> >         (gdbarch_update_p): Before calling init(), swap out and clear the
> >         existing architecture.
> >         * gdbarch.c: Regenerate.
> > 
> > The comment from arm-tdep.c should help explain why the above code is 
> > broken:
> > 
> >   /* We can't use NUM_REGS nor NUM_PSEUDO_REGS here, since that still
> >      references the old architecture vector, not the one we are
> >      building here.  */
> >   prologue_cache.saved_regs = (CORE_ADDR *)
> >     xcalloc (1, (sizeof (CORE_ADDR)
> >                  * (gdbarch_num_regs (gdbarch)
> >                     + gdbarch_num_pseudo_regs (gdbarch))));
> > 
> > As far as I can tell though, REGISTER_VIRTUAL_SIZE expands to 
> > REGISTER_VIRTUAL_TYPE which is still a macro?
> 
> The expansion appears to be:
> tdep->mips_fp_register_double = (((((38) >= 38 && (38) < 38 +32) ?
> 	builtin_type_float : ((38) == 32 ) ?
>         builtin_type_uint32 : ((38) >= 70 && (38) <= 89) ?
> 	builtin_type_uint32 :
>         builtin_type_int))->main_type->length == 8);
> 
> I can't see any reason for it to crash; I haven't reproduced this
> locally yet, although I'm working on it.  The above expansion came from
> the second person to mention this to me.

I believe you're right about the patch at fault, though I don't know
what to do about it.  This makes builtin_type_int == NULL during
evaluation of that expression, along with all the others.

-- 
Daniel Jacobowitz                           Carnegie Mellon University
MontaVista Software                         Debian GNU/Linux Developer

Re: GDB broken on MIPS targets with unmarked binaries

[Andrew Cagney]

> The expansion appears to be:
>> tdep->mips_fp_register_double = (((((38) >= 38 && (38) < 38 +32) ?
>> 	builtin_type_float : ((38) == 32 ) ?
>>         builtin_type_uint32 : ((38) >= 70 && (38) <= 89) ?
>> 	builtin_type_uint32 :
>>         builtin_type_int))->main_type->length == 8);
>> 
>> I can't see any reason for it to crash; I haven't reproduced this
>> locally yet, although I'm working on it.  The above expansion came from
>> the second person to mention this to me.
> 
> 
> I believe you're right about the patch at fault, though I don't know
> what to do about it.  This makes builtin_type_int == NULL during
> evaluation of that expression, along with all the others.

The patch isn't `at fault' - it is detecting a fault in mips_tdep.c :-) 
  builtin_type_float would have contained a value from the previously 
selected ISA.

That code is trying to reverse engineer the header files to figure out 
what the default should be for the given target.

I think better would be to is set mips_fp_register_double=-1 and then 
change FP_REGISTER_DOUBLE to a real function (gdb multi-arch can now be 
assumed) that, when -1, apply the above logic (and even update the result).

enjoy,
Andrew

Re: GDB broken on MIPS targets with unmarked binaries

[Daniel Jacobowitz]

On Wed, Jun 05, 2002 at 05:41:42PM -0400, Andrew Cagney wrote:
> >The expansion appears to be:
> >>tdep->mips_fp_register_double = (((((38) >= 38 && (38) < 38 +32) ?
> >>	builtin_type_float : ((38) == 32 ) ?
> >>        builtin_type_uint32 : ((38) >= 70 && (38) <= 89) ?
> >>	builtin_type_uint32 :
> >>        builtin_type_int))->main_type->length == 8);
> >>
> >>I can't see any reason for it to crash; I haven't reproduced this
> >>locally yet, although I'm working on it.  The above expansion came from
> >>the second person to mention this to me.
> >
> >
> >I believe you're right about the patch at fault, though I don't know
> >what to do about it.  This makes builtin_type_int == NULL during
> >evaluation of that expression, along with all the others.
> 
> The patch isn't `at fault' - it is detecting a fault in mips_tdep.c :-) 
>  builtin_type_float would have contained a value from the previously 
> selected ISA.
> 
> That code is trying to reverse engineer the header files to figure out 
> what the default should be for the given target.
> 
> I think better would be to is set mips_fp_register_double=-1 and then 
> change FP_REGISTER_DOUBLE to a real function (gdb multi-arch can now be 
> assumed) that, when -1, apply the above logic (and even update the result).

How about setting something in each header that redefines
REGISTER_VIRTUAL_TYPE, instead?  I think that's much clearer, and when
we multi-arch the targets which do this it will transition more
naturally to setting it at gdbarch_init time.  gdbarch_data() is a very
nice tool, but this is the wrong kind of nail, IMO.

I'm testing this patch.  Yes, it's a little bit of a step backwards for
multi-arching, but I still think it's correct; rather than "reverse
engineer the header files to figure out the default", let's have the
header files tell us what it is.

It works fine in my testing; OK to commit, or would you prefer a
different way?

-- 
Daniel Jacobowitz                           Carnegie Mellon University
MontaVista Software                         Debian GNU/Linux Developer

2002-06-06  Daniel Jacobowitz  <drow@mvista.com>

	* mips-tdep.c (FP_REGISTER_DOUBLE):  Delete obsolete definition.
	Define unconditionally.
	(mips_gdbarch_init): Use MIPS_FP_DOUBLE.
	* config/mips/tm-mips.h (MIPS_FP_DOUBLE): New macro.
	* config/mips/tm-irix5.h (MIPS_FP_DOUBLE): New macro.
	* config/mips/tm-irix6.h (MIPS_FP_DOUBLE): New macro.
	* config/mips/tm-mips64.h (MIPS_FP_DOUBLE): New macro.

Index: mips-tdep.c
===================================================================
RCS file: /cvs/src/src/gdb/mips-tdep.c,v
retrieving revision 1.74
diff -u -p -u -r1.74 mips-tdep.c
--- mips-tdep.c	21 May 2002 15:36:03 -0000	1.74
+++ mips-tdep.c	6 Jun 2002 17:34:17 -0000
@@ -108,11 +108,6 @@ static int mips_fpu_type_auto = 1;
 static enum mips_fpu_type mips_fpu_type = MIPS_DEFAULT_FPU_TYPE;
 #define MIPS_FPU_TYPE mips_fpu_type
 
-/* Do not use "TARGET_IS_MIPS64" to test the size of floating point registers */
-#ifndef FP_REGISTER_DOUBLE
-#define FP_REGISTER_DOUBLE (REGISTER_VIRTUAL_SIZE(FP0_REGNUM) == 8)
-#endif
-
 static int mips_debug = 0;
 
 /* MIPS specific per-architecture information */
@@ -212,10 +207,7 @@ mips2_fp_compat (void)
    form double-precision values).  Do not use "TARGET_IS_MIPS64" to
    determine if the ABI is using double-precision registers.  See also
    MIPS_FPU_TYPE. */
-#if GDB_MULTI_ARCH
-#undef FP_REGISTER_DOUBLE
 #define FP_REGISTER_DOUBLE (gdbarch_tdep (current_gdbarch)->mips_fp_register_double)
-#endif
 
 /* Does the caller allocate a ``home'' for each register used in the
    function call?  The N32 ABI and MIPS_EABI do not, the others do. */
@@ -4332,7 +4324,7 @@ mips_gdbarch_init (struct gdbarch_info i
       tdep->mips_abi_string = "default";
       tdep->mips_default_saved_regsize = MIPS_REGSIZE;
       tdep->mips_default_stack_argsize = MIPS_REGSIZE;
-      tdep->mips_fp_register_double = (REGISTER_VIRTUAL_SIZE (FP0_REGNUM) == 8);
+      tdep->mips_fp_register_double = MIPS_FP_DOUBLE;
       tdep->mips_last_arg_regnum = A0_REGNUM + 8 - 1;
       tdep->mips_last_fp_arg_regnum = FPA0_REGNUM + 8 - 1;
       tdep->mips_regs_have_home_p = 1;
Index: config/mips/tm-irix5.h
===================================================================
RCS file: /cvs/src/src/gdb/config/mips/tm-irix5.h,v
retrieving revision 1.6
diff -u -p -u -r1.6 tm-irix5.h
--- config/mips/tm-irix5.h	6 Mar 2001 08:21:32 -0000	1.6
+++ config/mips/tm-irix5.h	6 Jun 2002 17:34:17 -0000
@@ -41,6 +41,9 @@
       ((N) - FP0_REGNUM) * sizeof(double) : \
       32 * sizeof(double) + ((N) - 32) * MIPS_REGSIZE)
 
+#undef MIPS_FP_DOUBLE
+#define MIPS_FP_DOUBLE 1
+
 #undef  REGISTER_VIRTUAL_TYPE
 #define REGISTER_VIRTUAL_TYPE(N) \
 	(((N) >= FP0_REGNUM && (N) < FP0_REGNUM+32) ? builtin_type_double \
Index: config/mips/tm-irix6.h
===================================================================
RCS file: /cvs/src/src/gdb/config/mips/tm-irix6.h,v
retrieving revision 1.1
diff -u -p -u -r1.1 tm-irix6.h
--- config/mips/tm-irix6.h	7 Jun 2001 15:57:57 -0000	1.1
+++ config/mips/tm-irix6.h	6 Jun 2002 17:34:17 -0000
@@ -88,6 +88,9 @@
       ((N) - FP0_REGNUM) * sizeof(double) : \
       32 * sizeof(double) + ((N) - 32) * MIPS_REGSIZE)
 
+#undef MIPS_FP_DOUBLE
+#define MIPS_FP_DOUBLE 1
+
 #undef  REGISTER_VIRTUAL_TYPE
 #define REGISTER_VIRTUAL_TYPE(N) \
 	(((N) >= FP0_REGNUM && (N) < FP0_REGNUM+32) ? builtin_type_double \
Index: config/mips/tm-mips.h
===================================================================
RCS file: /cvs/src/src/gdb/config/mips/tm-mips.h,v
retrieving revision 1.25
diff -u -p -u -r1.25 tm-mips.h
--- config/mips/tm-mips.h	10 Feb 2002 07:17:17 -0000	1.25
+++ config/mips/tm-mips.h	6 Jun 2002 17:34:19 -0000
@@ -89,6 +89,12 @@ extern int in_sigtramp (CORE_ADDR, char 
 #define MIPS_REGSIZE 4
 #endif
 
+/* Defined to 1 if this architecture uses double-precision floating point
+   registers.  */
+#ifndef MIPS_FP_DOUBLE
+#define MIPS_FP_DOUBLE 0
+#endif
+
 /* Number of machine registers */
 
 #ifndef NUM_REGS
Index: config/mips/tm-mips64.h
===================================================================
RCS file: /cvs/src/src/gdb/config/mips/tm-mips64.h,v
retrieving revision 1.5
diff -u -p -u -r1.5 tm-mips64.h
--- config/mips/tm-mips64.h	5 Jun 2002 19:18:25 -0000	1.5
+++ config/mips/tm-mips64.h	6 Jun 2002 17:34:19 -0000
@@ -22,6 +22,9 @@
 /* Use eight byte registers.  */
 #define MIPS_REGSIZE 8
 
+/* Use double-precision floating point registers.  */
+#define MIPS_FP_DOUBLE 1
+
 /* define 8 byte register type */
 #define REGISTER_VIRTUAL_TYPE(N) \
         (((N) >= FP0_REGNUM && (N) < FP0_REGNUM+32) ? builtin_type_double \

Re: GDB broken on MIPS targets with unmarked binaries

[Andrew Cagney]

> 
> How about setting something in each header that redefines
> REGISTER_VIRTUAL_TYPE, instead?  I think that's much clearer, and when
> we multi-arch the targets which do this it will transition more
> naturally to setting it at gdbarch_init time.  gdbarch_data() is a very
> nice tool, but this is the wrong kind of nail, IMO.
> 
> I'm testing this patch.  Yes, it's a little bit of a step backwards for
> multi-arching, but I still think it's correct; rather than "reverse
> engineer the header files to figure out the default", let's have the
> header files tell us what it is.
> 
> It works fine in my testing; OK to commit, or would you prefer a
> different way?

try something like:

-/* Do not use "TARGET_IS_MIPS64" to test the size of floating point 
registers */
-#ifndef FP_REGISTER_DOUBLE
-#define FP_REGISTER_DOUBLE (REGISTER_VIRTUAL_SIZE(FP0_REGNUM) == 8)
-#endif

#define FP_REGISTER_DOUBLE mip_fp_register-double ()

mips_fp_register_double ()
{
   if (current_gdbarch->tdep->fp_register_double < 0)
     ...->fp_register_double = REGISTER_VIRTUAL_SIZE(FP0_REGNUM) == 8;
   return gdbarch->tdep->fp_register_double;
}

And:

-      tdep->mips_fp_register_double = (REGISTER_VIRTUAL_SIZE 
(FP0_REGNUM) == 8);

   tdep->mips_fp_register_double = -1;

It delays the computation until it is needed.  It also happens to keep 
the old code working without having to add more macros :-)

enjoy,
Andrew

Re: GDB broken on MIPS targets with unmarked binaries

[Daniel Jacobowitz]

On Fri, Jun 07, 2002 at 03:55:11PM -0400, Andrew Cagney wrote:
> >
> >How about setting something in each header that redefines
> >REGISTER_VIRTUAL_TYPE, instead?  I think that's much clearer, and when
> >we multi-arch the targets which do this it will transition more
> >naturally to setting it at gdbarch_init time.  gdbarch_data() is a very
> >nice tool, but this is the wrong kind of nail, IMO.
> >
> >I'm testing this patch.  Yes, it's a little bit of a step backwards for
> >multi-arching, but I still think it's correct; rather than "reverse
> >engineer the header files to figure out the default", let's have the
> >header files tell us what it is.
> >
> >It works fine in my testing; OK to commit, or would you prefer a
> >different way?
> 
> try something like:
> 
> -/* Do not use "TARGET_IS_MIPS64" to test the size of floating point 
> registers */
> -#ifndef FP_REGISTER_DOUBLE
> -#define FP_REGISTER_DOUBLE (REGISTER_VIRTUAL_SIZE(FP0_REGNUM) == 8)
> -#endif
> 
> #define FP_REGISTER_DOUBLE mip_fp_register-double ()
> 
> mips_fp_register_double ()
> {
>   if (current_gdbarch->tdep->fp_register_double < 0)
>     ...->fp_register_double = REGISTER_VIRTUAL_SIZE(FP0_REGNUM) == 8;
>   return gdbarch->tdep->fp_register_double;
> }
> 
> And:
> 
> -      tdep->mips_fp_register_double = (REGISTER_VIRTUAL_SIZE 
> (FP0_REGNUM) == 8);
> 
>   tdep->mips_fp_register_double = -1;
> 
> It delays the computation until it is needed.  It also happens to keep 
> the old code working without having to add more macros :-)

Doesn't work.  It's referenced in creating the register cache, which is
still before we have an architecture.  We need to know this information
earlier.  Any other ideas, or is my original patch OK?

By the way, I think the existing default: case in that switch is pretty
pointless anyway.  Note that it selects an arbitrary number of argument
registers.  It doesn't work when all the tools are configured mips-elf
(default to O32), which is why I suggested some time ago that we just
default to O32 in that configuration at least... that would make the
test numbers suck a bit less, and reduce the black magic factor a bit.

-- 
Daniel Jacobowitz                           Carnegie Mellon University
MontaVista Software                         Debian GNU/Linux Developer

Re: GDB broken on MIPS targets with unmarked binaries

[Andrew Cagney]

> It delays the computation until it is needed.  It also happens to keep 
>> the old code working without having to add more macros :-)
> 
> 
> Doesn't work.  It's referenced in creating the register cache, which is
> still before we have an architecture.  We need to know this information
> earlier.  Any other ideas, or is my original patch OK?

I'll commit a work-around.

> By the way, I think the existing default: case in that switch is pretty
> pointless anyway.  Note that it selects an arbitrary number of argument
> registers.  It doesn't work when all the tools are configured mips-elf
> (default to O32), which is why I suggested some time ago that we just
> default to O32 in that configuration at least... that would make the
> test numbers suck a bit less, and reduce the black magic factor a bit.

Default for plain mips-elf?  Yes sounds like a good idea.  As the other 
variants get deleted they can fall back to that.

(I take it this is the default for mips-elf-{as,gcc}.

Andrew