Fixes for PR68356, PR81210, and PR81693

The following patch fixes PR68356, PR81210, and PR81693 on darwin.

Is it OK?

TIA

Dominique

On Sun, Nov 12, 2017 at 6:22 AM, Dominique d'Humières
<dominiq@lps.ens.fr> wrote:
> The following patch fixes PR68356, PR81210, and PR81693 on darwin.
>
> --- ../_clean/gcc/testsuite/gcc.dg/torture/pr68264.c    2016-01-28 00:30:03.000000000 +0100
> +++ gcc/testsuite/gcc.dg/torture/pr68264.c      2017-11-11 17:16:58.000000000 +0100
> @@ -1,4 +1,5 @@
>  /* { dg-do run } */
> +/* { dg-xfail-run-if "PR68356 no math-errno on darwin" { "*-*-darwin*" } } */
>  /* { dg-add-options ieee } */
>  /* { dg-require-effective-target fenv_exceptions } */
>
> --- ../_clean/gcc/testsuite/gcc.dg/torture/pr68037-1.c  2016-06-10 15:22:50.000000000 +0200
> +++ gcc/testsuite/gcc.dg/torture/pr68037-1.c    2017-11-11 18:43:16.000000000 +0100
> @@ -1,4 +1,5 @@
>  /* { dg-do run { target i?86-*-* x86_64-*-* } } */
> +/* { dg-xfail-run-if "PR81210" { *-*-darwin* && lp64 } } */
>  /* { dg-options "-mgeneral-regs-only" } */
>
>  extern void exit (int);
> --- ../_clean/gcc/testsuite/gcc.dg/torture/pr68037-2.c  2016-06-10 15:22:50.000000000 +0200
> +++ gcc/testsuite/gcc.dg/torture/pr68037-2.c    2017-11-11 18:44:08.000000000 +0100
> @@ -1,4 +1,5 @@
>  /* { dg-do run { target i?86-*-* x86_64-*-* } } */
> +/* { dg-xfail-run-if "PR81210" { *-*-darwin* && lp64 } } */
>  /* { dg-options "-mgeneral-regs-only" } */
>
>  extern void exit (int);
> --- ../_clean/gcc/testsuite/gcc.dg/torture/pr68037-3.c  2016-06-10 15:22:50.000000000 +0200
> +++ gcc/testsuite/gcc.dg/torture/pr68037-3.c    2017-11-11 18:49:10.000000000 +0100
> @@ -1,4 +1,5 @@
>  /* { dg-do run { target i?86-*-* x86_64-*-* } } */
> +/* { dg-xfail-run-if "PR81210" { *-*-darwin* && lp64 } { "-O1" "-O2" "-O3" "-Os" } } */
>  /* { dg-options "-mgeneral-regs-only" } */
>
>  #include <stddef.h>
> --- ../_clean/gcc/testsuite/gcc.dg/torture/pr25967-1.c  2017-10-26 07:16:19.000000000 +0200
> +++ gcc/testsuite/gcc.dg/torture/pr25967-1.c    2017-11-11 19:36:30.000000000 +0100
> @@ -1,4 +1,5 @@
>  /* { dg-do run { target i?86-*-* x86_64-*-* } } */
> +/* { dg-xfail-run-if "PR81693" { "*-*-darwin*" } } */
>  /* { dg-options "-mgeneral-regs-only" } */
>
>  extern void exit (int);
> --- ../_clean/gcc/testsuite/gcc.dg/torture/pr25967-2.c  2017-10-26 07:16:19.000000000 +0200
> +++ gcc/testsuite/gcc.dg/torture/pr25967-2.c    2017-11-11 19:36:02.000000000 +0100
> @@ -1,4 +1,5 @@
>  /* { dg-do run { target i?86-*-* x86_64-*-* } } */
> +/* { dg-xfail-run-if "PR81693" { *-*-darwin* && ilp32 } } */
>  /* { dg-options "-mgeneral-regs-only" } */
>
>
> Is it OK?

I wrote these tests.  These tests don't align stack to 16 bytes and
should be skipped on Darwin.

On Nov 12, 2017, at 6:58 AM, H.J. Lu <hjl.tools@gmail.com> wrote:
> 
> On Sun, Nov 12, 2017 at 6:22 AM, Dominique d'Humières
> <dominiq@lps.ens.fr> wrote:
>> The following patch fixes PR68356, PR81210, and PR81693 on darwin.
>> 
>> --- ../_clean/gcc/testsuite/gcc.dg/torture/pr68264.c    2016-01-28 00:30:03.000000000 +0100
>> +++ gcc/testsuite/gcc.dg/torture/pr68264.c      2017-11-11 17:16:58.000000000 +0100
>> @@ -1,4 +1,5 @@
>> /* { dg-do run } */
>> +/* { dg-xfail-run-if "PR68356 no math-errno on darwin" { "*-*-darwin*" } } */
>> /* { dg-add-options ieee } */
>> /* { dg-require-effective-target fenv_exceptions } */
>> 
>> --- ../_clean/gcc/testsuite/gcc.dg/torture/pr68037-1.c  2016-06-10 15:22:50.000000000 +0200
>> +++ gcc/testsuite/gcc.dg/torture/pr68037-1.c    2017-11-11 18:43:16.000000000 +0100
>> @@ -1,4 +1,5 @@
>> /* { dg-do run { target i?86-*-* x86_64-*-* } } */
>> +/* { dg-xfail-run-if "PR81210" { *-*-darwin* && lp64 } } */
>> /* { dg-options "-mgeneral-regs-only" } */
>> 
>> extern void exit (int);
>> --- ../_clean/gcc/testsuite/gcc.dg/torture/pr68037-2.c  2016-06-10 15:22:50.000000000 +0200
>> +++ gcc/testsuite/gcc.dg/torture/pr68037-2.c    2017-11-11 18:44:08.000000000 +0100
>> @@ -1,4 +1,5 @@
>> /* { dg-do run { target i?86-*-* x86_64-*-* } } */
>> +/* { dg-xfail-run-if "PR81210" { *-*-darwin* && lp64 } } */
>> /* { dg-options "-mgeneral-regs-only" } */
>> 
>> extern void exit (int);
>> --- ../_clean/gcc/testsuite/gcc.dg/torture/pr68037-3.c  2016-06-10 15:22:50.000000000 +0200
>> +++ gcc/testsuite/gcc.dg/torture/pr68037-3.c    2017-11-11 18:49:10.000000000 +0100
>> @@ -1,4 +1,5 @@
>> /* { dg-do run { target i?86-*-* x86_64-*-* } } */
>> +/* { dg-xfail-run-if "PR81210" { *-*-darwin* && lp64 } { "-O1" "-O2" "-O3" "-Os" } } */
>> /* { dg-options "-mgeneral-regs-only" } */
>> 
>> #include <stddef.h>
>> --- ../_clean/gcc/testsuite/gcc.dg/torture/pr25967-1.c  2017-10-26 07:16:19.000000000 +0200
>> +++ gcc/testsuite/gcc.dg/torture/pr25967-1.c    2017-11-11 19:36:30.000000000 +0100
>> @@ -1,4 +1,5 @@
>> /* { dg-do run { target i?86-*-* x86_64-*-* } } */
>> +/* { dg-xfail-run-if "PR81693" { "*-*-darwin*" } } */
>> /* { dg-options "-mgeneral-regs-only" } */
>> 
>> extern void exit (int);
>> --- ../_clean/gcc/testsuite/gcc.dg/torture/pr25967-2.c  2017-10-26 07:16:19.000000000 +0200
>> +++ gcc/testsuite/gcc.dg/torture/pr25967-2.c    2017-11-11 19:36:02.000000000 +0100
>> @@ -1,4 +1,5 @@
>> /* { dg-do run { target i?86-*-* x86_64-*-* } } */
>> +/* { dg-xfail-run-if "PR81693" { *-*-darwin* && ilp32 } } */
>> /* { dg-options "-mgeneral-regs-only" } */
>> 
>> 
>> Is it OK?
> 
> I wrote these tests.  These tests don't align stack to 16 bytes and
> should be skipped on Darwin.

So, sounds like something based on:

  /* { dg-skip-if "sp not aligned to 16 bytes" { *-*-darwin } } */

then.  Ok with that change.

> Le 13 nov. 2017 à 18:40, Mike Stump <mikestump@comcast.net> a écrit :
> 
> On Nov 12, 2017, at 6:58 AM, H.J. Lu <hjl.tools@gmail.com> wrote:
>> 
>> On Sun, Nov 12, 2017 at 6:22 AM, Dominique d'Humières
>> <dominiq@lps.ens.fr> wrote:
>>> The following patch fixes PR68356, PR81210, and PR81693 on darwin.
>>> ...
>> 
>> I wrote these tests.  These tests don't align stack to 16 bytes and
>> should be skipped on Darwin.
> 
> So, sounds like something based on:
> 
>  /* { dg-skip-if "sp not aligned to 16 bytes" { *-*-darwin } } */
> 
> then.  Ok with that change.

Thanks for the review.

Am I correct to understand that this apply to pr25967-*.c only?

I’ld like to keep the PR numbers. Is it OK?

TIA

Dominique

On Nov 14, 2017, at 3:33 AM, Dominique d'Humières <dominiq@lps.ens.fr> wrote:
> 
>> Le 13 nov. 2017 à 18:40, Mike Stump <mikestump@comcast.net> a écrit :
>> On Nov 12, 2017, at 6:58 AM, H.J. Lu <hjl.tools@gmail.com> wrote:
>>> 
>>> On Sun, Nov 12, 2017 at 6:22 AM, Dominique d'Humières
>>> <dominiq@lps.ens.fr> wrote:
>>>> The following patch fixes PR68356, PR81210, and PR81693 on darwin.
>>>> ...
>>> 
>>> I wrote these tests.  These tests don't align stack to 16 bytes and
>>> should be skipped on Darwin.
>> 
>> So, sounds like something based on:
>> 
>> /* { dg-skip-if "sp not aligned to 16 bytes" { *-*-darwin } } */
>> 
>> then.  Ok with that change.
> 
> Thanks for the review.
> 
> Am I correct to understand that this apply to pr25967-*.c only?
> 
> I’ld like to keep the PR numbers. Is it OK?

Feel free to keep the PR, but, I'd close the other two as dups of the first, and then just list the first.  The issue is that all of these test cases are 1 problem, so there should only be 1 PR.  HJ's comment applies to all test cases in your patch, and the problem appears to be the single non-16 byte stack on all of them.  Because of that alone, the test cases should be skipped for that one reason, no matter the existence of other reasons why that test case should be skipped.