[PATCH][ARM/AArch64] PR 68088: Fix RTL checking ICE due to subregs inside accumulator forwarding check

[Ramana Radhakrishnan]

On 29/10/15 14:14, Kyrill Tkachov wrote:
> 
> On 29/10/15 14:00, Marcus Shawcroft wrote:
>> On 29 October 2015 at 13:50, Kyrill Tkachov <kyrylo.tkachov@arm.com> wrote:
>>
>>>>> Ok for trunk?
>>>> rtl.h exposes reg_or_subregno() already doesn't that do what we need here?
>>>
>>> reg_or_subregno assumes that what it's passed is REG or a SUBREG.
>>> It will ICE on any other rtx. Here I want to strip the subreg if it is
>>> a subreg, but leave it as it is otherwise.
>> OK, I follow.
>>
>>>> The test case is not aarch64 specific therefore I think convention is
>>>> that it should go into a generic directory.
>>>
>>> Ok, I'll put it in gcc.dg/
>>
>> OK with the test case moved. Thanks /Marcus
>>
> Thanks, but I'd like to do a slight respin.
> The testcase is moved to gcc.dg but I also avoid creating the new
> helper function and just do the SUBREG extraction once at the very end.
> This makes the patch smaller.
> 
> Since you're ok with the approach and this revision is logically equivalent,
> I just need an ok from an arm perspective.

OK, looks good to me and assuming you've tested this with a regression test run targeting cortex-a53 to stress the function ;)


regards
Ramana
> 
> Thanks,
> Kyrill
> 
> 2015-10-29  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
> 
>     PR target/68088
>     * config/arm/aarch-common.c (aarch_accumulator_forwarding): Strip
>     subregs from accumulator and make sure it's a register.
> 
> 2015-10-29  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
> 
>     PR target/68088
>     * gcc.dg/pr68088_1.c: New test.
>