Re: [patch, arm] Consistent display of registers in corefile

[Yao Qi <yao at codesourcery dot com>]

On 12/20/2010 02:23 AM, Daniel Jacobowitz wrote:
> On Mon, Dec 13, 2010 at 06:00:32PM +0800, Yao Qi wrote:
>> On 12/13/2010 10:57 AM, Daniel Jacobowitz wrote:
>>> On Fri, Dec 10, 2010 at 03:43:15PM +0100, Mark Kettenis wrote:
>>>> I would suspect that the proper thing to do would be to align the
>>>> tdesc with the code instead of the other way around.  The arm-core.xml
>>>> file seems to underspecify things by omitting the type=xxx clause on
>>>> many registers.  Whoever wrote arm_register_type() at least had to
>>>> make a conscious decision about the signedness of the type used for
>>>> the general purpose registers.
>>>
>>> Yeah, I agree.  It was probably my mistake.
>>>
>>
>> In this new patch, 'type="uint32"' is added for registers from r0 to r12
>> except r11.  r11 is 'type="data_ptr"'.  features/arm*.c files are
>> regenerated by Makefile.  Regression tested along with the other patch
>> arm_fps_group.patch on armv7l-unknown-linux-gnueabi, "corefile restored
>> general registers" failure in gdb.base/gcore.exp goes away.  Is it OK
>> for GDB mainline?
> 
> Please use uint32 for r11 also.  It's sometimes the frame pointer, but
> that's not required and it may have any arbitrary data in it.
> 
> Otherwise OK.
> 

Thanks, Dan.  I'll use uint32 for r11 in my commit.

Do you have comments to arm_fps_group.patch, which fixes the 2nd problem
I pointed out in [1]?  In short, this fix will make GDB treat fps
register as registers in float group, fps will disappear in output of
"info register".

gdb/
	* arm-tdep.c (arm_register_reggroup_p): New.
	(arm_gdbarch_init): Set arm_register_reggroup_p for hook
	register_reggroup_p.

arm_fps_group.patch is the 2nd patch I attached in [1].

[1] http://sourceware.org/ml/gdb-patches/2010-12/msg00134.html

-- 
Yao (éå)