Re: a patch for PR68695

[Jeff Law <law@redhat.com>]

On 04/15/2016 05:06 AM, Kyrill Tkachov wrote:
>
> On 05/04/16 23:35, Segher Boessenkool wrote:
>> On Tue, Apr 05, 2016 at 10:48:58AM +0100, Kyrill Tkachov wrote:
>>> So for the test gcc.dg/pr10474.c on arm with -marm -O3 before this
>>> patch we
>>> perform shrink-wrapping:
>>>      cmp    r0, #0
>>>      bxeq    lr
>>>      push    {r4, lr}
>>>      mov    r4, r0
>>>      ...
>>>
>>> And after the patch we don't:
>>>      push    {r4, lr}
>>>      subs    r4, r0, #0
>>>      popeq    {r4, pc}
>>>      ...
>>>
>>> The assembly after the "..." is identical.
>>>
>>> So the resulting code is indeed shorter, though there is an
>>> extra stack push and pop on the early return path.
>>> A similar effect appears on gcc.dg/ira-shrinkwrap-prep-2.c.
>> The "new" code is better if there is no shrink-wrapping.  We can probably
>> teach prepare_shrink_wrap to do the extra register move if that will
>> allow
>> us to wrap more.
>>
>>> Though if so, it looks like a shrink-wrapping deficiency exposed by
>>> this patch, rather than caused by it.
>> Yes, and mostly a testcase problem even.
>>
>>> Jakub, do you happen to have the before and after codegen for these
>>> tests
>>> on ppc64? I wonder if the effect is more clearcut there.
>> RTL before shrink-wrapping would be useful, too.
>
> So what shall we do for these tests for GCC 6?
> Add an XFAIL for arm and powerpc?
We could just punt gcc-6 and focus on what we want for gcc-7 as this 
isn't a release critical issue.

jeff