[RFC] gdb.base/frame-args.exp

[RFC] gdb.base/frame-args.exp

[Luis Machado]

Hi,

It looks like this test is a bit focused on results from only a few 
targets. On x86 it works correctly and displays the arguments just fine. 
When set to display only scalars, then GDB proceeds to display '...' on 
the fields that are not scalars.

On PowerPC 32-bit though, this is what i get:

(gdb) PASS: gdb.base/frame-args.exp: set print frame-arguments all
frame 1^M
#1  0x100006dc in call_me (i=3, f=5, s=<error reading variable: value 
has been optimized out>, ss=0xbffff788, u=<error reading variable: value 
has been optimized out>, e=green) at gdb.base/frame-args.c:40^M
40        break_me ();^M
(gdb) FAIL: gdb.base/frame-args.exp: frame 1 with print frame-arguments 
set to all
set print frame-arguments scalars^M
(gdb) PASS: gdb.base/frame-args.exp: set print frame-arguments scalars
frame 1^M
#1  0x100006dc in call_me (i=3, f=5, s=<error reading variable: value 
has been optimized out>, ss=0xbffff788, u=<error reading variable: value 
has been optimized out>, e=green) at gdb.base/frame-args.c:40^M
40        break_me ();^M
(gdb) FAIL: gdb.base/frame-args.exp: frame 1 with print frame-arguments 
set to scalars
set print frame-arguments none^M

So it looks like we are taking a different route and GDB ends up 
catching an error before displaying the optimized out information, thus 
not matching what the testcase expects.

I could add a different pattern to the testcase and that would solve the 
couple failures i see, but i wonder if we should tweak the testcase to 
dodge optimizations?

Thoughts?

Luis

Re: [RFC] gdb.base/frame-args.exp

[Yao Qi]

On 07/01/2014 03:56 PM, Luis Machado wrote:
> So it looks like we are taking a different route and GDB ends up 
> catching an error before displaying the optimized out information, thus 
> not matching what the testcase expects.
> 
> I could add a different pattern to the testcase and that would solve the 
> couple failures i see, but i wonder if we should tweak the testcase to 
> dodge optimizations?

Hi Luis,
Andrew Burgess had a patch changing "<error reading variable: value has
been optimized out>" to "<optimized out>" in this thread,

  PATCH: error reading variable: value has been optimized out
  https://sourceware.org/ml/gdb-patches/2012-08/msg00715.html

so I think "<optimized out>" is the right direction to go.  We should
fix gdb rather than update test case to match the current output, IMO.

-- 
Yao (齐尧)

Re: [RFC] gdb.base/frame-args.exp

[Luis Machado]

Hi Yao,

On 07/03/2014 07:44 AM, Yao Qi wrote:
> On 07/01/2014 03:56 PM, Luis Machado wrote:
>> So it looks like we are taking a different route and GDB ends up
>> catching an error before displaying the optimized out information, thus
>> not matching what the testcase expects.
>>
>> I could add a different pattern to the testcase and that would solve the
>> couple failures i see, but i wonder if we should tweak the testcase to
>> dodge optimizations?
>
> Hi Luis,
> Andrew Burgess had a patch changing "<error reading variable: value has
> been optimized out>" to "<optimized out>" in this thread,
>
>    PATCH: error reading variable: value has been optimized out
>    https://sourceware.org/ml/gdb-patches/2012-08/msg00715.html
>
> so I think "<optimized out>" is the right direction to go.  We should
> fix gdb rather than update test case to match the current output, IMO.
>

Excellent! Thanks for pointing that out. Seeing the date of the patch, i 
suppose it has not gone in.

Regards,
Luis

Re: [RFC] gdb.base/frame-args.exp

[Yao Qi]

On 07/03/2014 03:03 PM, Luis Machado wrote:
> Excellent! Thanks for pointing that out. Seeing the date of the patch, i 
> suppose it has not gone in.

The patch was committed in September.

2012-09-21  Andrew Burgess  <aburgess@broadcom.com>

        * findvar.c (read_frame_register_value): Mark the result value as
        optimized out if any of the input registers have been optimized out.

https://sourceware.org/ml/gdb-patches/2012-09/msg00393.html

-- 
Yao (齐尧)