patch: init exponent for sim_fpu_class_zero

patch: init exponent for sim_fpu_class_zero

[Aldy Hernandez]

howdy.

in sim-fpu.c, the exponent is left undefined for zeros (sim_fpu_class_zero).

i am working on a port whose multiply and divide instructions check the 
value of the operands' exponent and sign to determine overflow bits to set.  
this is obviously not working when the exponent is undefined ;-).

i have verified that i can build the sim and gdb after my change (for
powerpc-eabi).

ok to install?

2002-06-09  Aldy Hernandez  <aldyh@redhat.com>

        * sim-fpu.c (unpack_fpu): Initialize exponent for
        sim_fpu_class_zero.
        (i2fpu): Same.
        (sim_fpu_sqrt): Same.

Index: sim-fpu.c
===================================================================
RCS file: /cvs/src/src/sim/common/sim-fpu.c,v
retrieving revision 1.3
diff -c -r1.3 sim-fpu.c
*** sim-fpu.c	23 Dec 2000 11:51:04 -0000	1.3
--- sim-fpu.c	11 Jun 2002 05:14:16 -0000
***************
*** 330,335 ****
--- 330,336 ----
  	  /* tastes like zero */
  	  dst->class = sim_fpu_class_zero;
  	  dst->sign = sign;
+ 	  dst->normal_exp = 0;
  	}
        else
  	{
***************
*** 520,525 ****
--- 521,527 ----
      {
        f->class = sim_fpu_class_zero;
        f->sign = 0;
+       f->normal_exp = 0;
      }
    else
      {
***************
*** 648,653 ****
--- 650,656 ----
      {
        f->class = sim_fpu_class_zero;
        f->sign = 0;
+       f->normal_exp = 0;
      }
    else
      {
***************
*** 1799,1804 ****
--- 1802,1808 ----
      {
        f->class = sim_fpu_class_zero;
        f->sign = r->sign;
+       f->normal_exp = 0;
        return 0;
      }
    if (sim_fpu_is_infinity (r))

Re: patch: init exponent for sim_fpu_class_zero

[Andrew Cagney]

> howdy.
> 
> in sim-fpu.c, the exponent is left undefined for zeros (sim_fpu_class_zero).
> 
> i am working on a port whose multiply and divide instructions check the 
> value of the operands' exponent and sign to determine overflow bits to set.  
> this is obviously not working when the exponent is undefined ;-).
> 
> i have verified that i can build the sim and gdb after my change (for
> powerpc-eabi).
> 
> ok to install?

Hmm, wierd,

> + 	  dst->normal_exp = 0;

of the possible values that is the most intuitive so yes.

Can you, separatly, commit/post a patch adding your self to the 
gdb/MAINTAINERS file under write after approva.

Andrew

Re: patch: init exponent for sim_fpu_class_zero

[Aldy Hernandez]

> Hmm, wierd,
> 
> >+ 	  dst->normal_exp = 0;
> 
> of the possible values that is the most intuitive so yes.

committed.

> Can you, separatly, commit/post a patch adding your self to the 
> gdb/MAINTAINERS file under write after approva.

committed the following as well.

thanks.

2002-06-09  Aldy Hernandez  <aldyh@redhat.com>

        * MAINTAINERS: Add self.

Index: MAINTAINERS
===================================================================
RCS file: /cvs/src/src/gdb/MAINTAINERS,v
retrieving revision 1.180
diff -c -p -r1.180 MAINTAINERS
*** MAINTAINERS	2 Jun 2002 03:09:31 -0000	1.180
--- MAINTAINERS	12 Jun 2002 00:54:15 -0000
*************** Fred Fish					fnf@ninemoons.com
*** 391,396 ****
--- 391,397 ----
  Orjan Friberg					orjanf@axis.com
  Ben Harris					bjh21@netbsd.org
  Richard Henderson				rth@redhat.com
+ Aldy Hernandez					aldyh@redhat.com
  Paul Hilfinger					hilfinger@gnat.com
  Matt Hiller					hiller@redhat.com
  Kazu Hirata					kazu@hxi.com