Re: [ping 2] [RFA][PATCH v4 0/5] Add TDB regset support

[Andreas Arnez <arnez at linux dot vnet dot ibm dot com>]

Luis Machado <lgustavo@codesourcery.com> writes:

> On 07/15/2013 12:34 PM, Andreas Arnez wrote:
>> Luis Machado <lgustavo@codesourcery.com> writes:
>>
>>> I didn't go through your last update of the patch, but FTR i still
>>> think we should make the core file sections static and store them in
>>> some form of array instead of hardcoding their contents in numerous
>>> function calls.
>>
>> In the PowerPC case the patch includes four call-back invocations, all
>> contained in a 20-line iterator function.  I'd hardly call that
>> "numerous function calls".  And I consider it an improvement over the
>> original code, which had six hard-coded static array initializers with
>> various copy-/pasted lines, plus the logic for selecting the correct
>> array.  The improvement is even more drastic for S/390.  Don't you
>> agree?  Or do you see even more potential for improvement?
>
> What i don't see now is an obvious way of telling which register sets
> are available for core files in PowerPC. You'd have to infer that
> based on dynamic data.
>
> It is my personal view on the change, really. I don't claim it is
> right or wrong.
>
> Also, why is the PowerPC backend being modified together with S390? Is
> this a change to account for POWER8? The introductory mail does not
> mention anything PowerPC-specific.

Because I saw the potential for code simplification here.  Also, in an
informal conversation with Ulrich Weigand he indicated that he was
interested in such a change.

In the scope of this patch set the PowerPC change is purely optional.
I'll just remove it from the next version of the patch set.