[regcache] raw vs. cooked...

[regcache] raw vs. cooked...

[Michael Snyder]

[-- Attachment #1: Type: text/plain, Size: 180 bytes --]

Andrew, 

I suspect the test below is not valid for pseudo-registers, since they
may not have a location in the reg cache at all.  Anyway, the change
makes sh4 work again.

Michael

[-- Attachment #2: regcache.diff --]
[-- Type: text/plain, Size: 1027 bytes --]

Index: regcache.c
===================================================================
RCS file: /cvs/src/src/gdb/regcache.c,v
retrieving revision 1.82
diff -p -r1.82 regcache.c
*** regcache.c	17 May 2003 05:59:58 -0000	1.82
--- regcache.c	7 Jun 2003 00:55:07 -0000
*************** init_regcache_descr (struct gdbarch *gdb
*** 235,241 ****
    /* Sanity check.  Confirm that there is agreement between the
       regcache and the target's redundant REGISTER_BYTE (new targets
       should not even be defining it).  */
!   for (i = 0; i < descr->nr_cooked_registers; i++)
      {
        if (REGISTER_BYTE_P ())
  	gdb_assert (descr->register_offset[i] == REGISTER_BYTE (i));
--- 235,241 ----
    /* Sanity check.  Confirm that there is agreement between the
       regcache and the target's redundant REGISTER_BYTE (new targets
       should not even be defining it).  */
!   for (i = 0; i < descr->nr_raw_registers; i++)
      {
        if (REGISTER_BYTE_P ())
  	gdb_assert (descr->register_offset[i] == REGISTER_BYTE (i));

Re: [regcache] raw vs. cooked...

[Andrew Cagney]

> Andrew, 
> 
> I suspect the test below is not valid for pseudo-registers, since they
> may not have a location in the reg cache at all.  Anyway, the change
> makes sh4 work again.

The check's valid.  See the thread: assertion failure in regcache.c
http://sources.redhat.com/ml/gdb/2003-05/msg00289.html

Kris, Elena, did a fix get committed?

Andrew



> Index: regcache.c
> ===================================================================
> RCS file: /cvs/src/src/gdb/regcache.c,v
> retrieving revision 1.82
> diff -p -r1.82 regcache.c
> *** regcache.c	17 May 2003 05:59:58 -0000	1.82
> --- regcache.c	7 Jun 2003 00:55:07 -0000
> *************** init_regcache_descr (struct gdbarch *gdb
> *** 235,241 ****
>     /* Sanity check.  Confirm that there is agreement between the
>        regcache and the target's redundant REGISTER_BYTE (new targets
>        should not even be defining it).  */
> !   for (i = 0; i < descr->nr_cooked_registers; i++)
>       {
>         if (REGISTER_BYTE_P ())
>   	gdb_assert (descr->register_offset[i] == REGISTER_BYTE (i));
> --- 235,241 ----
>     /* Sanity check.  Confirm that there is agreement between the
>        regcache and the target's redundant REGISTER_BYTE (new targets
>        should not even be defining it).  */
> !   for (i = 0; i < descr->nr_raw_registers; i++)
>       {
>         if (REGISTER_BYTE_P ())
>   	gdb_assert (descr->register_offset[i] == REGISTER_BYTE (i));

Re: [regcache] raw vs. cooked...

[Kris Warkentin]

> > I suspect the test below is not valid for pseudo-registers, since they
> > may not have a location in the reg cache at all.  Anyway, the change
> > makes sh4 work again.
>
> The check's valid.  See the thread: assertion failure in regcache.c
> http://sources.redhat.com/ml/gdb/2003-05/msg00289.html
>
> Kris, Elena, did a fix get committed?

I assume you're talking about the fix I described in this message:
http://sources.redhat.com/ml/gdb/2003-05/msg00322.html

It seems like this wouldn't work for SH64 though so I was waiting until I
had some more time to chase it. (or some SH experts piped up).  It certainly
seems to fix the problem for 32 bit SH though.  Perhaps we could do the fix
for all but the 64 bit version and then let someone using SH64 trip on the
assertion and have to fix it?

;-)

Kris

Re: [regcache] raw vs. cooked...

[Andrew Cagney]

>> > I suspect the test below is not valid for pseudo-registers, since they
>> > may not have a location in the reg cache at all.  Anyway, the change
>> > makes sh4 work again.
> 
>>
>> The check's valid.  See the thread: assertion failure in regcache.c
>> http://sources.redhat.com/ml/gdb/2003-05/msg00289.html
>>
>> Kris, Elena, did a fix get committed?
> 
> 
> I assume you're talking about the fix I described in this message:
> http://sources.redhat.com/ml/gdb/2003-05/msg00322.html
> 
> It seems like this wouldn't work for SH64 though so I was waiting until I
> had some more time to chase it. (or some SH experts piped up).  It certainly
> seems to fix the problem for 32 bit SH though.  Perhaps we could do the fix
> for all but the 64 bit version and then let someone using SH64 trip on the
> assertion and have to fix it?

To be honest, I suspect splitting sh-tdep along the lines of i386-tdep 
and x86-64-tdep might be the way to go.  Anway, yes, good suggestion,

Elena?

Andrew