Re: [PUSHED] vect: Add back testcases for trapping math with comparisons [PR36190]

On 2026-08-07 19:07, Andrea Pinski wrote:
> On Fri, Aug 7, 2026 at 10:01 AM Torbjorn SVENSSON
> <[email protected]> wrote:
>>
>> Hi,
>>
>> These test fail for arm-none-eabi with -mcpu=cortex-a7 -mfloat-abi=hard -mfpu=auto.
>>
>> Should the tests include the following line
>>
>> /* { dg-warning "'-fassociative-math' disabled" "" { target *-*-* } 0 } */
>>
>> or something similar to remove the warning (above line is from gcc.dg/tree-ssa/pr23109.c)?
> 
> See https://gcc.gnu.org/bugzilla/show_bug.cgi?id=126339 and
> https://gcc.gnu.org/pipermail/gcc-regression/2026-July/084147.html
> where I was figuring out the best way of solving this.
> I think s/-ffast-math/-funsafe-math-optimizations/ in
> testsuite/lib/target-supports.exp but I have not a chance to test it
> out yet.

I just tested the change locally here and it does not help.

Testing vect/vect-ifcvt-13.c
doing compile
Executing on host: /build/r17-2833-g32657f29f91871/bin/arm-none-eabi-gcc  /build/gcc_src/gcc/testsuite/gcc.dg/vect/vect-ifcvt-13.c  -mthumb -march=armv7ve+simd -mcpu=cortex-a7 -mfloat-abi=hard -mfpu=auto   -dumpbase "" -fdiagnostics-plain-output   -funsafe-math-optimizations -ftree-vectorize -fno-tree-loop-distribute-patterns -fno-vect-cost-model -fno-common -O2 -fdump-tree-vect-details -ftrapping-math      --specs=rdimon.specs -Wl,--start-group -lc -lm -Wl,--end-group --specs=nosys.specs -Wl,--allow-multiple-definition -Wl,-u,_isatty,-u,_fstat  -Wl,-wrap,exit -Wl,-wrap,_exit -Wl,-wrap,main -Wl,-wrap,abort -Wl,gcc_tg.o -lm -T qemu.ld -o ./vect-ifcvt-13.exe    (timeout = 800)
...
  status 0
FAIL: gcc.dg/vect/vect-ifcvt-15.c (test for excess errors)
Excess errors:
cc1: warning: '-fassociative-math' disabled; other options take precedence

Kind regards,
Torbjörn

> 
>>
>> This is what I see in my logs:
>>
>> Testing vect/vect-ifcvt-15.c
>> doing compile
>> Executing on host: /build/r17-2833-g32657f29f91871/bin/arm-none-eabi-gcc  /build/gcc_src/gcc/testsuite/gcc.dg/vect/vect-ifcvt-15.c  -mthumb -march=armv7ve+simd -mcpu=cortex-a7 -mfloat-abi=hard -mfpu=auto   -dumpbase "" -fdiagnostics-plain-output   -ffast-math -ftree-vectorize -fno-tree-loop-distribute-patterns -fno-vect-cost-model -fno-common -O2 -fdump-tree-vect-details -ftrapping-math      --specs=rdimon.specs -Wl,--start-group -lc -lm -Wl,--end-group --specs=nosys.specs -Wl,--allow-multiple-definition -Wl,-u,_isatty,-u,_fstat  -Wl,-wrap,exit -Wl,-wrap,_exit -Wl,-wrap,main -Wl,-wrap,abort -Wl,gcc_tg.o -lm -T qemu.ld -o ./vect-ifcvt-15.exe    (timeout = 800)
>> ...
>>    status 0
>> FAIL: gcc.dg/vect/vect-ifcvt-15.c (test for excess errors)
>> Excess errors:
>> cc1: warning: '-fassociative-math' disabled; other options take precedence
>>
>>
>> Kind regards,
>> Torbjörn
>>
>> On 2026-07-11 02:00, Andrew Pinski wrote:
>>> These were originally renamed to no-trapping-math-vect-*.c in r0-87196-g04b03edb56633f
>>> but were fixed in r0-107313-gf9613c9a919462 but never added back to the
>>> trapping versions.
>>> This adds them back this time with an explict -ftrapping-math.  Note vect-ifcvt-11.c
>>> already exists now so need to name it vect-ifcvt-11a.c.
>>>
>>> Committed as obvious after testing these testcases work on x86_64-linux-gnu.
>>>
>>>        PR testsuite/36190
>>>
>>> gcc/testsuite/ChangeLog:
>>>
>>>        * gcc.dg/vect/vect-111.c: New test.
>>>        * gcc.dg/vect/vect-ifcvt-11a.c: New test.
>>>        * gcc.dg/vect/vect-ifcvt-12.c: New test.
>>>        * gcc.dg/vect/vect-ifcvt-13.c: New test.
>>>        * gcc.dg/vect/vect-ifcvt-14.c: New test.
>>>        * gcc.dg/vect/vect-ifcvt-15.c: New test.
>>>
>>> Signed-off-by: Andrew Pinski <[email protected]>
>>> ---
>>>    gcc/testsuite/gcc.dg/vect/vect-111.c       | 40 ++++++++++++++++++++++
>>>    gcc/testsuite/gcc.dg/vect/vect-ifcvt-11a.c | 34 ++++++++++++++++++
>>>    gcc/testsuite/gcc.dg/vect/vect-ifcvt-12.c  | 33 ++++++++++++++++++
>>>    gcc/testsuite/gcc.dg/vect/vect-ifcvt-13.c  | 34 ++++++++++++++++++
>>>    gcc/testsuite/gcc.dg/vect/vect-ifcvt-14.c  | 34 ++++++++++++++++++
>>>    gcc/testsuite/gcc.dg/vect/vect-ifcvt-15.c  | 34 ++++++++++++++++++
>>>    6 files changed, 209 insertions(+)
>>>    create mode 100644 gcc/testsuite/gcc.dg/vect/vect-111.c
>>>    create mode 100644 gcc/testsuite/gcc.dg/vect/vect-ifcvt-11a.c
>>>    create mode 100644 gcc/testsuite/gcc.dg/vect/vect-ifcvt-12.c
>>>    create mode 100644 gcc/testsuite/gcc.dg/vect/vect-ifcvt-13.c
>>>    create mode 100644 gcc/testsuite/gcc.dg/vect/vect-ifcvt-14.c
>>>    create mode 100644 gcc/testsuite/gcc.dg/vect/vect-ifcvt-15.c
>>>
>>> diff --git a/gcc/testsuite/gcc.dg/vect/vect-111.c b/gcc/testsuite/gcc.dg/vect/vect-111.c
>>> new file mode 100644
>>> index 00000000000..a8a6d15aa1d
>>> --- /dev/null
>>> +++ b/gcc/testsuite/gcc.dg/vect/vect-111.c
>>> @@ -0,0 +1,40 @@
>>> +/* { dg-require-effective-target vect_float } */
>>> +/* { dg-additional-options "-ftrapping-math" } */
>>> +
>>> +#include <stdarg.h>
>>> +#include "tree-vect.h"
>>> +
>>> +#define N 16
>>> +
>>> +float b[N] = {0,3,6,9,12,15,18,21,24,27,30,33,36,39,42,45};
>>> +
>>> +__attribute__ ((noinline)) int
>>> +main1 (void)
>>> +{
>>> +  int i;
>>> +  float a[N];
>>> +
>>> +  /* Condition in loop.  */
>>> +  /* This loop is vectorized on platforms that support vect_condition.  */
>>> +  for (i = 0; i < N; i++)
>>> +  {
>>> +    a[i] = (b[i] > 0 ? b[i] : 0);
>>> +  }
>>> +
>>> +#pragma GCC novector
>>> +  for (i = 0; i < N; i++)
>>> +  {
>>> +    if (a[i] != b[i])
>>> +     abort ();
>>> +  }
>>> +  return 0;
>>> +}
>>> +
>>> +int main (void)
>>> +{
>>> +  check_vect ();
>>> +  return main1 ();
>>> +}
>>> +
>>> +/* { dg-final { scan-tree-dump-times "vectorized 1 loops" 1 "vect" { target powerpc*-*-* } } } */
>>> +/* { dg-final { scan-tree-dump-times "vectorized 1 loops" 1 "vect" { target i?86-*-* x86_64-*-* ia64-*-* } } } */
>>> diff --git a/gcc/testsuite/gcc.dg/vect/vect-ifcvt-11a.c b/gcc/testsuite/gcc.dg/vect/vect-ifcvt-11a.c
>>> new file mode 100644
>>> index 00000000000..a7f2f93a614
>>> --- /dev/null
>>> +++ b/gcc/testsuite/gcc.dg/vect/vect-ifcvt-11a.c
>>> @@ -0,0 +1,34 @@
>>> +/* { dg-require-effective-target vect_condition } */
>>> +/* { dg-require-effective-target vect_float } */
>>> +/* { dg-additional-options "-fno-trapping-math" } */
>>> +
>>> +#include <stdarg.h>
>>> +#include "tree-vect.h"
>>> +
>>> +#define N 16
>>> +#define MAX 42
>>> +
>>> +float A[N] = {36,39,42,45,43,32,21,12,23,34,45,56,67,78,89,11};
>>> +float B[N] = {0,0,42,42,42,0,0,0,0,0,42,42,42,42,42,0};
>>> +
>>> +extern void abort(void);
>>> +
>>> +int main ()
>>> +{
>>> +  int i, j;
>>> +
>>> +  check_vect ();
>>> +
>>> +  for (i = 0; i < 16; i++)
>>> +    A[i] = ( A[i] >= MAX ? MAX : 0);
>>> +
>>> +  /* check results:  */
>>> +#pragma GCC novector
>>> +  for (i = 0; i < N; i++)
>>> +    if (A[i] != B[i])
>>> +      abort ();
>>> +
>>> +  return 0;
>>> +}
>>> +
>>> +/* { dg-final { scan-tree-dump-times "vectorized 1 loops" 1 "vect" } } */
>>> diff --git a/gcc/testsuite/gcc.dg/vect/vect-ifcvt-12.c b/gcc/testsuite/gcc.dg/vect/vect-ifcvt-12.c
>>> new file mode 100644
>>> index 00000000000..b8418dd5a4e
>>> --- /dev/null
>>> +++ b/gcc/testsuite/gcc.dg/vect/vect-ifcvt-12.c
>>> @@ -0,0 +1,33 @@
>>> +/* { dg-require-effective-target vect_condition } */
>>> +/* { dg-require-effective-target vect_float } */
>>> +/* { dg-additional-options "-ftrapping-math" } */
>>> +
>>> +#include <stdarg.h>
>>> +#include "tree-vect.h"
>>> +
>>> +#define N 16
>>> +#define MAX 42
>>> +
>>> +float A[N] = {36,39,42,45,43,32,21,12,23,34,45,56,67,78,89,11};
>>> +float B[N] = {0,0,0,42,42,0,0,0,0,0,42,42,42,42,42,0};
>>> +
>>> +extern void abort(void);
>>> +
>>> +int main ()
>>> +{
>>> +  int i, j;
>>> +
>>> +  check_vect ();
>>> +  for (i = 0; i < 16; i++)
>>> +    A[i] = ( A[i] > MAX ? MAX : 0);
>>> +
>>> +  /* check results:  */
>>> +#pragma GCC novector
>>> +  for (i = 0; i < N; i++)
>>> +    if (A[i] != B[i])
>>> +      abort ();
>>> +
>>> +  return 0;
>>> +}
>>> +
>>> +/* { dg-final { scan-tree-dump-times "vectorized 1 loops" 1 "vect" } } */
>>> diff --git a/gcc/testsuite/gcc.dg/vect/vect-ifcvt-13.c b/gcc/testsuite/gcc.dg/vect/vect-ifcvt-13.c
>>> new file mode 100644
>>> index 00000000000..3a62c8cadba
>>> --- /dev/null
>>> +++ b/gcc/testsuite/gcc.dg/vect/vect-ifcvt-13.c
>>> @@ -0,0 +1,34 @@
>>> +/* { dg-require-effective-target vect_condition } */
>>> +/* { dg-require-effective-target vect_float } */
>>> +/* { dg-additional-options "-ftrapping-math" } */
>>> +
>>> +#include <stdarg.h>
>>> +#include "tree-vect.h"
>>> +
>>> +#define N 16
>>> +#define MAX 42
>>> +
>>> +float A[N] = {36,39,42,45,43,32,21,12,23,34,45,56,67,78,89,11};
>>> +float B[N] = {42,42,42,0,0,42,42,42,42,42,0,0,0,0,0,42};
>>> +
>>> +extern void abort(void);
>>> +
>>> +int main ()
>>> +{
>>> +  int i, j;
>>> +
>>> +  check_vect ();
>>> +
>>> +  for (i = 0; i < 16; i++)
>>> +    A[i] = ( A[i] <= MAX ? MAX : 0);
>>> +
>>> +  /* check results:  */
>>> +#pragma GCC novector
>>> +  for (i = 0; i < N; i++)
>>> +    if (A[i] != B[i])
>>> +      abort ();
>>> +
>>> +  return 0;
>>> +}
>>> +
>>> +/* { dg-final { scan-tree-dump-times "vectorized 1 loops" 1 "vect" } } */
>>> diff --git a/gcc/testsuite/gcc.dg/vect/vect-ifcvt-14.c b/gcc/testsuite/gcc.dg/vect/vect-ifcvt-14.c
>>> new file mode 100644
>>> index 00000000000..3a62c8cadba
>>> --- /dev/null
>>> +++ b/gcc/testsuite/gcc.dg/vect/vect-ifcvt-14.c
>>> @@ -0,0 +1,34 @@
>>> +/* { dg-require-effective-target vect_condition } */
>>> +/* { dg-require-effective-target vect_float } */
>>> +/* { dg-additional-options "-ftrapping-math" } */
>>> +
>>> +#include <stdarg.h>
>>> +#include "tree-vect.h"
>>> +
>>> +#define N 16
>>> +#define MAX 42
>>> +
>>> +float A[N] = {36,39,42,45,43,32,21,12,23,34,45,56,67,78,89,11};
>>> +float B[N] = {42,42,42,0,0,42,42,42,42,42,0,0,0,0,0,42};
>>> +
>>> +extern void abort(void);
>>> +
>>> +int main ()
>>> +{
>>> +  int i, j;
>>> +
>>> +  check_vect ();
>>> +
>>> +  for (i = 0; i < 16; i++)
>>> +    A[i] = ( A[i] <= MAX ? MAX : 0);
>>> +
>>> +  /* check results:  */
>>> +#pragma GCC novector
>>> +  for (i = 0; i < N; i++)
>>> +    if (A[i] != B[i])
>>> +      abort ();
>>> +
>>> +  return 0;
>>> +}
>>> +
>>> +/* { dg-final { scan-tree-dump-times "vectorized 1 loops" 1 "vect" } } */
>>> diff --git a/gcc/testsuite/gcc.dg/vect/vect-ifcvt-15.c b/gcc/testsuite/gcc.dg/vect/vect-ifcvt-15.c
>>> new file mode 100644
>>> index 00000000000..3180b36eb78
>>> --- /dev/null
>>> +++ b/gcc/testsuite/gcc.dg/vect/vect-ifcvt-15.c
>>> @@ -0,0 +1,34 @@
>>> +/* { dg-require-effective-target vect_condition } */
>>> +/* { dg-require-effective-target vect_float } */
>>> +/* { dg-additional-options "-ftrapping-math" } */
>>> +
>>> +#include <stdarg.h>
>>> +#include "tree-vect.h"
>>> +
>>> +#define N 16
>>> +#define MAX 42
>>> +
>>> +float A[N] = {36,39,42,45,43,32,21,12,23,34,45,56,67,78,89,11};
>>> +float B[N] = {42,42,0,0,0,42,42,42,42,42,0,0,0,0,0,42};
>>> +
>>> +extern void abort(void);
>>> +
>>> +int main ()
>>> +{
>>> +  int i, j;
>>> +
>>> +  check_vect ();
>>> +
>>> +  for (i = 0; i < 16; i++)
>>> +    A[i] = ( A[i] < MAX ? MAX : 0);
>>> +
>>> +  /* check results:  */
>>> +#pragma GCC novector
>>> +  for (i = 0; i < N; i++)
>>> +    if (A[i] != B[i])
>>> +      abort ();
>>> +
>>> +  return 0;
>>> +}
>>> +
>>> +/* { dg-final { scan-tree-dump-times "vectorized 1 loops" 1 "vect" } } */
>>