[i386] Fix PR 57756

Sriraman,

The tests gcc.target/i386/funcspec-5.c and gcc.target/i386/pr57756.c fail 
on targets for which -msse is the default (see
http://gcc.gnu.org/ml/gcc-testresults/2013-10/msg01365.html or
http://gcc.gnu.org/ml/gcc-testresults/2013-10/msg01345.html ).

This is fixed with the following patch:


My apologies if this has already reported in the thread (I may have missed some posts).

Dominique

On Fri, Oct 18, 2013 at 3:03 AM, Dominique Dhumieres <dominiq@lps.ens.fr> wrote:
> Sriraman,
>
> The tests gcc.target/i386/funcspec-5.c and gcc.target/i386/pr57756.c fail
> on targets for which -msse is the default (see
> http://gcc.gnu.org/ml/gcc-testresults/2013-10/msg01365.html or
> http://gcc.gnu.org/ml/gcc-testresults/2013-10/msg01345.html ).

pr57756.c is reported failing on x86_64-unknown-linux-gnu and I am
using the same config. I have both tests passing. What am I missing? I
do not see why pr57756 should be marked with -mno-sse.

Sri


>
> This is fixed with the following patch:
>
> diff -up ../_clean/gcc/testsuite/gcc.target/i386/funcspec-5.c gcc/testsuite/gcc.target/i386/funcspec-5.c
> --- ../_clean/gcc/testsuite/gcc.target/i386/funcspec-5.c        2011-08-23 21:54:27.000000000 +0200
> +++ gcc/testsuite/gcc.target/i386/funcspec-5.c  2013-10-17 09:45:20.000000000 +0200
> @@ -2,6 +2,7 @@
>     without error.  */
>  /* { dg-do compile } */
>  /* { dg-require-effective-target ia32 } */
> +/* { dg-options "-mno-sse" } */
>
>  extern void test_abm (void)                    __attribute__((__target__("abm")));
>  extern void test_aes (void)                    __attribute__((__target__("aes")));
> diff -up ../_clean/gcc/testsuite/gcc.target/i386/pr57756.c gcc/testsuite/gcc.target/i386/pr57756.c
> --- ../_clean/gcc/testsuite/gcc.target/i386/pr57756.c   2013-10-15 23:53:31.000000000 +0200
> +++ gcc/testsuite/gcc.target/i386/pr57756.c     2013-10-17 09:46:53.000000000 +0200
> @@ -1,6 +1,7 @@
>  /* callee cannot be inlined into caller because it has a higher
>     target ISA.  */
>  /* { dg-do compile } */
> +/* { dg-options "-mno-sse" } */
>
>  __attribute__((always_inline,target("sse4.2")))
>  __inline int callee () /* { dg-error "inlining failed in call to always_inline" }  */
>
> My apologies if this has already reported in the thread (I may have missed some posts).
>
> Dominique

On Fri, Oct 18, 2013 at 10:27 AM, Sriraman Tallam <tmsriram@google.com> wrote:
> On Fri, Oct 18, 2013 at 3:03 AM, Dominique Dhumieres <dominiq@lps.ens.fr> wrote:
>> Sriraman,
>>
>> The tests gcc.target/i386/funcspec-5.c and gcc.target/i386/pr57756.c fail
>> on targets for which -msse is the default (see
>> http://gcc.gnu.org/ml/gcc-testresults/2013-10/msg01365.html or
>> http://gcc.gnu.org/ml/gcc-testresults/2013-10/msg01345.html ).
>
> pr57756.c is reported failing on x86_64-unknown-linux-gnu and I am
> using the same config. I have both tests passing. What am I missing? I
> do not see why pr57756 should be marked with -mno-sse.

I see why pr57756.c could fail, if -msse4.2 is turned on by default. I
think this test needs {dg-options "-mno-sse4.2"}. With SSE4.2 by
default, the target attributes are no-ops and the inlining will
succeed causing the test to fail.

Thanks
Sri

>
> Sri
>
>
>>
>> This is fixed with the following patch:
>>
>> diff -up ../_clean/gcc/testsuite/gcc.target/i386/funcspec-5.c gcc/testsuite/gcc.target/i386/funcspec-5.c
>> --- ../_clean/gcc/testsuite/gcc.target/i386/funcspec-5.c        2011-08-23 21:54:27.000000000 +0200
>> +++ gcc/testsuite/gcc.target/i386/funcspec-5.c  2013-10-17 09:45:20.000000000 +0200
>> @@ -2,6 +2,7 @@
>>     without error.  */
>>  /* { dg-do compile } */
>>  /* { dg-require-effective-target ia32 } */
>> +/* { dg-options "-mno-sse" } */
>>
>>  extern void test_abm (void)                    __attribute__((__target__("abm")));
>>  extern void test_aes (void)                    __attribute__((__target__("aes")));
>> diff -up ../_clean/gcc/testsuite/gcc.target/i386/pr57756.c gcc/testsuite/gcc.target/i386/pr57756.c
>> --- ../_clean/gcc/testsuite/gcc.target/i386/pr57756.c   2013-10-15 23:53:31.000000000 +0200
>> +++ gcc/testsuite/gcc.target/i386/pr57756.c     2013-10-17 09:46:53.000000000 +0200
>> @@ -1,6 +1,7 @@
>>  /* callee cannot be inlined into caller because it has a higher
>>     target ISA.  */
>>  /* { dg-do compile } */
>> +/* { dg-options "-mno-sse" } */
>>
>>  __attribute__((always_inline,target("sse4.2")))
>>  __inline int callee () /* { dg-error "inlining failed in call to always_inline" }  */
>>
>> My apologies if this has already reported in the thread (I may have missed some posts).
>>
>> Dominique

> I see why pr57756.c could fail, if -msse4.2 is turned on by default. I
> think this test needs {dg-options "-mno-sse4.2"}.

This change allows the test to pass. The failure of gcc.target/i386/funcspec-5.c is

/opt/gcc/work/gcc/testsuite/gcc.target/i386/funcspec-5.c:34:1: warning: SSE instruction set disabled, using 387 arithmetics [enabled by default]
 extern void test_no_sse (void)   __attribute__((__target__("no-sse")));

This extra warning is silenced by dg-options "-mno-sse", but not by "-mno-sse4.2".

Dominique