[PATCH] [PATCH][ARM] Fix split-live-ranges-for-shrink-wrap.c testcase.

[PATCH] [PATCH][ARM] Fix split-live-ranges-for-shrink-wrap.c testcase.

[Alex Velenko]

Hi,

This patch limits testcase split-live-ranges-for-shrink-wrap.c runs to
supported achitecture versions.
Object size with -march=armv4t check fails because pop pc is not interworking
safe on armv4t.
This test is not supported for -march=armv7 as this test is for thumb1.

Is patch ok?

gcc/testsuite

2015-05-20  Alex Velenko  <Alex.Velenko@arm.com>

        * gcc.target/arm/split-live-ranges-for-shrink-wrap.c (dg-skip-if):
	Skip armv4t, armv7-a and later.
---
 gcc/testsuite/gcc.target/arm/split-live-ranges-for-shrink-wrap.c | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/gcc/testsuite/gcc.target/arm/split-live-ranges-for-shrink-wrap.c b/gcc/testsuite/gcc.target/arm/split-live-ranges-for-shrink-wrap.c
index e36000b..c649bc1 100644
--- a/gcc/testsuite/gcc.target/arm/split-live-ranges-for-shrink-wrap.c
+++ b/gcc/testsuite/gcc.target/arm/split-live-ranges-for-shrink-wrap.c
@@ -1,6 +1,8 @@
 /* { dg-do assemble } */
 /* { dg-options "-mthumb -Os -fdump-rtl-ira " }  */
 /* { dg-require-effective-target arm_thumb1_ok } */
+/* { dg-skip-if "" { arm_arch_v4t_ok } } */
+/* { dg-skip-if "" { arm_arch_v7a_ok } } */
 
 int foo (char *, char *, int);
 int test (int d, char * out, char *in, int len)
-- 
1.8.1.2

Re: [PATCH] [PATCH][ARM] Fix split-live-ranges-for-shrink-wrap.c testcase.

[Joseph Myers]

On Wed, 20 May 2015, Alex Velenko wrote:

> Hi,
> 
> This patch limits testcase split-live-ranges-for-shrink-wrap.c runs to
> supported achitecture versions.
> Object size with -march=armv4t check fails because pop pc is not interworking
> safe on armv4t.
> This test is not supported for -march=armv7 as this test is for thumb1.
> 
> Is patch ok?

Again, the condition you propose to add doesn't make sense.  arm_arch_X_ok 
is only appropriate for tests using an explicit -march=X.  Testing with 
-march=armv7* should automatically skip this test anyway because it would 
cause arm_thumb1_ok to fail.

-- 
Joseph S. Myers
joseph@codesourcery.com

Re: Re: [PATCH] [PATCH][ARM] Fix split-live-ranges-for-shrink-wrap.c testcase.

[Alex Velenko]

On 20/05/15 21:14, Joseph Myers wrote:
> Again, the condition you propose to add doesn't make sense.  arm_arch_X_ok
> is only appropriate for tests using an explicit -march=X.  Testing with
> -march=armv7* should automatically skip this test anyway because it would
> cause arm_thumb1_ok to fail.
>

Hi,

I adjusted the patch to skip execution split-live-ranges-for-shrink-wrap.c
with explicitly specified -march=armv4t and provide -march=armv5t flag =
for
arm_arch_v5t_ok targets.

Is patch ok?

Alex

gcc/testsuite

2015-06-22  Alex Velenko  <Alex.Velenko@arm.com>

          * gcc.target/arm/split-live-ranges-for-shrink-wrap.c (dg-skip-if):
	Skip -march=armv4t.
         (dg-additional-options): Set armv5t flag.

diff --git
a/gcc/testsuite/gcc.target/arm/split-live-ranges-for-shrink-wrap.c
b/gcc/testsuite/gcc.target/arm/split-live-ranges-for-shrink-wrap.c
index e36000b..3cb93dc 100644
--- a/gcc/testsuite/gcc.target/arm/split-live-ranges-for-shrink-wrap.c
+++ b/gcc/testsuite/gcc.target/arm/split-live-ranges-for-shrink-wrap.c
@@ -1,6 +1,8 @@
   /* { dg-do assemble } */
   /* { dg-options "-mthumb -Os -fdump-rtl-ira " }  */
   /* { dg-require-effective-target arm_thumb1_ok } */
+/* { dg-skip-if "do not test on armv4t" { *-*-* } { "-march=armv4t" } } =
*/
+/* { dg-additional-options "-march=armv5t" {target arm_arch_v5t_ok} } */

   int foo (char *, char *, int);
   int test (int d, char * out, char *in, int len)

--------------1.8.1.2--

Re: Re: [PATCH] [PATCH][ARM] Fix split-live-ranges-for-shrink-wrap.c testcase.

[Joseph Myers]

I have no more comments on this patch.

-- 
Joseph S. Myers
joseph@codesourcery.com

Re: Re: [PATCH] [PATCH][ARM] Fix split-live-ranges-for-shrink-wrap.c testcase.

[Ramana Radhakrishnan]

On Mon, Jun 22, 2015 at 5:56 PM, Alex Velenko <Alex.Velenko@arm.com> wrote:
> On 20/05/15 21:14, Joseph Myers wrote:
>>
>> Again, the condition you propose to add doesn't make sense.  arm_arch_X_ok
>> is only appropriate for tests using an explicit -march=X.  Testing with
>> -march=armv7* should automatically skip this test anyway because it would
>> cause arm_thumb1_ok to fail.
>>
>
> Hi,
>
> I adjusted the patch to skip execution split-live-ranges-for-shrink-wrap.c
> with explicitly specified -march=armv4t and provide -march=armv5t flag =
> for
> arm_arch_v5t_ok targets.
>
> Is patch ok?
>
> Alex
>
> gcc/testsuite
>
> 2015-06-22  Alex Velenko  <Alex.Velenko@arm.com>
>
>          * gcc.target/arm/split-live-ranges-for-shrink-wrap.c (dg-skip-if):
>         Skip -march=armv4t.
>         (dg-additional-options): Set armv5t flag.
>
> diff --git
> a/gcc/testsuite/gcc.target/arm/split-live-ranges-for-shrink-wrap.c
> b/gcc/testsuite/gcc.target/arm/split-live-ranges-for-shrink-wrap.c
> index e36000b..3cb93dc 100644
> --- a/gcc/testsuite/gcc.target/arm/split-live-ranges-for-shrink-wrap.c
> +++ b/gcc/testsuite/gcc.target/arm/split-live-ranges-for-shrink-wrap.c
> @@ -1,6 +1,8 @@
>   /* { dg-do assemble } */
>   /* { dg-options "-mthumb -Os -fdump-rtl-ira " }  */
>   /* { dg-require-effective-target arm_thumb1_ok } */
> +/* { dg-skip-if "do not test on armv4t" { *-*-* } { "-march=armv4t" } } =
> */
> +/* { dg-additional-options "-march=armv5t" {target arm_arch_v5t_ok} } */
>
>   int foo (char *, char *, int);
>   int test (int d, char * out, char *in, int len)


OK - please watch out for any multilibs fallout and apply this.

Ramana
>
> --------------1.8.1.2--
>

Re: Re: Re: [PATCH] [PATCH][ARM] Fix split-live-ranges-for-shrink-wrap.c testcase.

[Alex Velenko]

On 25/06/15 14:35, Ramana Radhakrishnan wrote:
> On Mon, Jun 22, 2015 at 5:56 PM, Alex Velenko <Alex.Velenko@arm.com> wrote:
>> On 20/05/15 21:14, Joseph Myers wrote:
>>>
>>> Again, the condition you propose to add doesn't make sense.  arm_arch_X_ok
>>> is only appropriate for tests using an explicit -march=X.  Testing with
>>> -march=armv7* should automatically skip this test anyway because it would
>>> cause arm_thumb1_ok to fail.
>>>
>>
>> Hi,
>>
>> I adjusted the patch to skip execution split-live-ranges-for-shrink-wrap.c
>> with explicitly specified -march=armv4t and provide -march=armv5t flag =
>> for
>> arm_arch_v5t_ok targets.
>>
>> Is patch ok?
>>
>> Alex
>>
>> gcc/testsuite
>>
>> 2015-06-22  Alex Velenko  <Alex.Velenko@arm.com>
>>
>>           * gcc.target/arm/split-live-ranges-for-shrink-wrap.c (dg-skip-if):
>>          Skip -march=armv4t.
>>          (dg-additional-options): Set armv5t flag.
>>
>> diff --git
>> a/gcc/testsuite/gcc.target/arm/split-live-ranges-for-shrink-wrap.c
>> b/gcc/testsuite/gcc.target/arm/split-live-ranges-for-shrink-wrap.c
>> index e36000b..3cb93dc 100644
>> --- a/gcc/testsuite/gcc.target/arm/split-live-ranges-for-shrink-wrap.c
>> +++ b/gcc/testsuite/gcc.target/arm/split-live-ranges-for-shrink-wrap.c
>> @@ -1,6 +1,8 @@
>>    /* { dg-do assemble } */
>>    /* { dg-options "-mthumb -Os -fdump-rtl-ira " }  */
>>    /* { dg-require-effective-target arm_thumb1_ok } */
>> +/* { dg-skip-if "do not test on armv4t" { *-*-* } { "-march=armv4t" } } =
>> */
>> +/* { dg-additional-options "-march=armv5t" {target arm_arch_v5t_ok} } */
>>
>>    int foo (char *, char *, int);
>>    int test (int d, char * out, char *in, int len)
>
>
> OK - please watch out for any multilibs fallout and apply this.
>
> Ramana
>>
>> --------------1.8.1.2--
>>
>
Committed to trunk r226036.
Is patch ok for fsf-5?
kind regards,
Alex

Re: [PATCH] [PATCH][ARM] Fix split-live-ranges-for-shrink-wrap.c testcase.

[Ramana Radhakrishnan]

>>
> Committed to trunk r226036.
> Is patch ok for fsf-5?

OK for all release branches where affected as this is a testism.


Ramana

> kind regards,
> Alex
>