* [PATCH] testsuite: Sanitize fails for SP FPU on Arm
@ 2022-09-22 16:40 Torbjörn SVENSSON
2022-09-22 22:42 ` Joseph Myers
0 siblings, 1 reply; 3+ messages in thread
From: Torbjörn SVENSSON @ 2022-09-22 16:40 UTC (permalink / raw)
To: gcc-patches; +Cc: yvan.roux, ro, mikestump, Torbjörn SVENSSON
This patch stops reporting fails for Arm targets with single
precision floating point unit for types wider than 32 bits (the width
of float on arm-none-eabi).
As reported in PR102017, fenv is reported as supported in recent
versions of newlib. At the same time, for some Arm targets, the
implementation in libgcc does not support exceptions and thus, the
test fails with a call to abort().
gcc/testsuite/ChangeLog:
* gcc.dg/c2x-float-7.c: Invert the exception check for Arm
targets with SP FPU.
* gcc.dg/pr95115.c: Likewise.
* gcc.dg/torture/float32x-nan-floath.c: Likewise.
* gcc.dg/torture/float32x-nan.c: Likewise.
* gcc.dg/torture/float64-nan-floath.c: Likewise.
* gcc.dg/torture/float64-nan.c: Likewise.
* gcc.dg/torture/inf-compare-1.c: Likewise.
* gcc.dg/torture/inf-compare-2.c: Likewise.
* gcc.dg/torture/inf-compare-3.c: Likewise.
* gcc.dg/torture/inf-compare-4.c: Likewise.
Co-Authored-By: Yvan ROUX <yvan.roux@foss.st.com>
Signed-off-by: Torbjörn SVENSSON <torbjorn.svensson@foss.st.com>
---
gcc/testsuite/gcc.dg/c2x-float-7.c | 10 ++++++++++
gcc/testsuite/gcc.dg/pr95115.c | 5 +++++
gcc/testsuite/gcc.dg/torture/floatn-nan-floath.h | 5 +++++
gcc/testsuite/gcc.dg/torture/floatn-nan.h | 10 ++++++++++
gcc/testsuite/gcc.dg/torture/inf-compare-1.c | 5 +++++
gcc/testsuite/gcc.dg/torture/inf-compare-2.c | 5 +++++
gcc/testsuite/gcc.dg/torture/inf-compare-3.c | 5 +++++
gcc/testsuite/gcc.dg/torture/inf-compare-4.c | 5 +++++
8 files changed, 50 insertions(+)
diff --git a/gcc/testsuite/gcc.dg/c2x-float-7.c b/gcc/testsuite/gcc.dg/c2x-float-7.c
index 0c90ff24165..c699e94aff8 100644
--- a/gcc/testsuite/gcc.dg/c2x-float-7.c
+++ b/gcc/testsuite/gcc.dg/c2x-float-7.c
@@ -39,11 +39,21 @@ main (void)
abort ();
feclearexcept (FE_ALL_EXCEPT);
d += d;
+#if defined(__ARM_FP) && __ARM_FP == 4
+ /* Arm with SP FPU does not support exceptions (see pr102017). */
+ if (fetestexcept (FE_INVALID))
+#else
if (!fetestexcept (FE_INVALID))
+#endif
abort ();
feclearexcept (FE_ALL_EXCEPT);
ld += ld;
+#if defined(__ARM_FP) && __ARM_FP == 4
+ /* Arm with SP FPU does not support exceptions (see pr102017). */
+ if (fetestexcept (FE_INVALID))
+#else
if (!fetestexcept (FE_INVALID))
+#endif
abort ();
exit (0);
}
diff --git a/gcc/testsuite/gcc.dg/pr95115.c b/gcc/testsuite/gcc.dg/pr95115.c
index 46a95dfb698..15bc6854819 100644
--- a/gcc/testsuite/gcc.dg/pr95115.c
+++ b/gcc/testsuite/gcc.dg/pr95115.c
@@ -19,7 +19,12 @@ main (void)
double r = x ();
if (!__builtin_isnan (r))
abort ();
+#if defined(__ARM_FP) && __ARM_FP == 4
+ /* Arm with SP FPU does not support exceptions (see pr102017). */
+ if (fetestexcept (FE_INVALID))
+#else
if (!fetestexcept (FE_INVALID))
+#endif
abort ();
exit (0);
}
diff --git a/gcc/testsuite/gcc.dg/torture/floatn-nan-floath.h b/gcc/testsuite/gcc.dg/torture/floatn-nan-floath.h
index 9892fd0cf63..5c9f28d4fdc 100644
--- a/gcc/testsuite/gcc.dg/torture/floatn-nan-floath.h
+++ b/gcc/testsuite/gcc.dg/torture/floatn-nan-floath.h
@@ -30,7 +30,12 @@ main (void)
{
volatile TYPE r;
r = nans_cst + nans_cst;
+#if defined(__ARM_FP) && __ARM_FP == 4 && (EXT || WIDTH > 32)
+ /* Arm with SP FPU does not support exceptions (see pr102017). */
+ if (fetestexcept (FE_INVALID))
+#else
if (!fetestexcept (FE_INVALID))
+#endif
abort ();
exit (0);
}
diff --git a/gcc/testsuite/gcc.dg/torture/floatn-nan.h b/gcc/testsuite/gcc.dg/torture/floatn-nan.h
index 89d2e2eec34..0abb0668677 100644
--- a/gcc/testsuite/gcc.dg/torture/floatn-nan.h
+++ b/gcc/testsuite/gcc.dg/torture/floatn-nan.h
@@ -30,10 +30,20 @@ main (void)
{
volatile TYPE r;
r = nan_cst + nan_cst;
+#if defined(__ARM_FP) && __ARM_FP == 4 && (EXT || WIDTH > 32)
+ /* Arm with SP FPU does not support exceptions (see pr102017). */
+ if (!fetestexcept (FE_INVALID))
+#else
if (fetestexcept (FE_INVALID))
+#endif
abort ();
r = nans_cst + nans_cst;
+#if defined(__ARM_FP) && __ARM_FP == 4 && (EXT || WIDTH > 32)
+ /* Arm with SP FPU does not support exceptions (see pr102017). */
+ if (fetestexcept (FE_INVALID))
+#else
if (!fetestexcept (FE_INVALID))
+#endif
abort ();
exit (0);
}
diff --git a/gcc/testsuite/gcc.dg/torture/inf-compare-1.c b/gcc/testsuite/gcc.dg/torture/inf-compare-1.c
index 70f255e680a..df0e61d9f89 100644
--- a/gcc/testsuite/gcc.dg/torture/inf-compare-1.c
+++ b/gcc/testsuite/gcc.dg/torture/inf-compare-1.c
@@ -16,6 +16,11 @@ int
main (void)
{
i = x > __builtin_inf ();
+#if defined(__ARM_FP) && __ARM_FP == 4
+ /* Arm with SP FPU does not support exceptions (see pr102017). */
+ if (i != 0 || fetestexcept (FE_INVALID))
+#else
if (i != 0 || !fetestexcept (FE_INVALID))
+#endif
abort ();
}
diff --git a/gcc/testsuite/gcc.dg/torture/inf-compare-2.c b/gcc/testsuite/gcc.dg/torture/inf-compare-2.c
index 011f992d5a0..dcb43ccc444 100644
--- a/gcc/testsuite/gcc.dg/torture/inf-compare-2.c
+++ b/gcc/testsuite/gcc.dg/torture/inf-compare-2.c
@@ -16,6 +16,11 @@ int
main (void)
{
i = x < -__builtin_inf ();
+#if defined(__ARM_FP) && __ARM_FP == 4
+ /* Arm with SP FPU does not support exceptions (see pr102017). */
+ if (i != 0 || fetestexcept (FE_INVALID))
+#else
if (i != 0 || !fetestexcept (FE_INVALID))
+#endif
abort ();
}
diff --git a/gcc/testsuite/gcc.dg/torture/inf-compare-3.c b/gcc/testsuite/gcc.dg/torture/inf-compare-3.c
index de5c478a8d8..1cd4d3e8199 100644
--- a/gcc/testsuite/gcc.dg/torture/inf-compare-3.c
+++ b/gcc/testsuite/gcc.dg/torture/inf-compare-3.c
@@ -16,6 +16,11 @@ int
main (void)
{
i = x <= __builtin_inf ();
+#if defined(__ARM_FP) && __ARM_FP == 4
+ /* Arm with SP FPU does not support exceptions (see pr102017). */
+ if (i != 0 || fetestexcept (FE_INVALID))
+#else
if (i != 0 || !fetestexcept (FE_INVALID))
+#endif
abort ();
}
diff --git a/gcc/testsuite/gcc.dg/torture/inf-compare-4.c b/gcc/testsuite/gcc.dg/torture/inf-compare-4.c
index 685562d3a40..0cd27076079 100644
--- a/gcc/testsuite/gcc.dg/torture/inf-compare-4.c
+++ b/gcc/testsuite/gcc.dg/torture/inf-compare-4.c
@@ -16,6 +16,11 @@ int
main (void)
{
i = x >= -__builtin_inf ();
+#if defined(__ARM_FP) && __ARM_FP == 4
+ /* Arm with SP FPU does not support exceptions (see pr102017). */
+ if (i != 0 || fetestexcept (FE_INVALID))
+#else
if (i != 0 || !fetestexcept (FE_INVALID))
+#endif
abort ();
}
--
2.25.1
^ permalink raw reply [flat|nested] 3+ messages in thread
* Re: [PATCH] testsuite: Sanitize fails for SP FPU on Arm
2022-09-22 16:40 [PATCH] testsuite: Sanitize fails for SP FPU on Arm Torbjörn SVENSSON
@ 2022-09-22 22:42 ` Joseph Myers
2022-09-23 10:03 ` Torbjorn SVENSSON
0 siblings, 1 reply; 3+ messages in thread
From: Joseph Myers @ 2022-09-22 22:42 UTC (permalink / raw)
To: Torbjörn SVENSSON; +Cc: gcc-patches
[-- Attachment #1: Type: text/plain, Size: 1128 bytes --]
On Thu, 22 Sep 2022, Torbjörn SVENSSON via Gcc-patches wrote:
> This patch stops reporting fails for Arm targets with single
> precision floating point unit for types wider than 32 bits (the width
> of float on arm-none-eabi).
>
> As reported in PR102017, fenv is reported as supported in recent
> versions of newlib. At the same time, for some Arm targets, the
> implementation in libgcc does not support exceptions and thus, the
> test fails with a call to abort().
It's definitely wrong to have this sort of Arm-specific conditional in
architecture-independent tests. Tests requiring floating-point exceptions
support should have an appropriate dg-require-effective-target; if that
dg-require-effective-target wrongly passes in certain configurations, fix
it (or e.g. add a new check_effective_target_fenv_exceptions_double to
verify that exceptions work for double, as opposed to the present
check_effective_target_fenv_exceptions which checks whether exceptions
work for float, and then adjust tests requiring exceptions for double to
use the new effective-target).
--
Joseph S. Myers
joseph@codesourcery.com
^ permalink raw reply [flat|nested] 3+ messages in thread
* Re: [PATCH] testsuite: Sanitize fails for SP FPU on Arm
2022-09-22 22:42 ` Joseph Myers
@ 2022-09-23 10:03 ` Torbjorn SVENSSON
0 siblings, 0 replies; 3+ messages in thread
From: Torbjorn SVENSSON @ 2022-09-23 10:03 UTC (permalink / raw)
To: Joseph Myers; +Cc: gcc-patches
Hi Joseph,
On 2022-09-23 00:42, Joseph Myers wrote:
> On Thu, 22 Sep 2022, Torbjörn SVENSSON via Gcc-patches wrote:
>
>> This patch stops reporting fails for Arm targets with single
>> precision floating point unit for types wider than 32 bits (the width
>> of float on arm-none-eabi).
>>
>> As reported in PR102017, fenv is reported as supported in recent
>> versions of newlib. At the same time, for some Arm targets, the
>> implementation in libgcc does not support exceptions and thus, the
>> test fails with a call to abort().
>
> It's definitely wrong to have this sort of Arm-specific conditional in
> architecture-independent tests. Tests requiring floating-point exceptions
> support should have an appropriate dg-require-effective-target; if that
> dg-require-effective-target wrongly passes in certain configurations, fix
> it (or e.g. add a new check_effective_target_fenv_exceptions_double to
> verify that exceptions work for double, as opposed to the present
> check_effective_target_fenv_exceptions which checks whether exceptions
> work for float, and then adjust tests requiring exceptions for double to
> use the new effective-target).
>
Okay, thanks for your review.
I will split this test case into 3 files as for SP FPU, we would like to
verify that the exception handling for the float part work, but at the
same time we know that exception handling for double and long double
should not work.
Do you think it's preferable to have the double and long double should
be reported as UNSUPPORRTED, or as XFAIL for the SP FPU case?
^ permalink raw reply [flat|nested] 3+ messages in thread
end of thread, other threads:[~2022-09-23 10:03 UTC | newest]
Thread overview: 3+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2022-09-22 16:40 [PATCH] testsuite: Sanitize fails for SP FPU on Arm Torbjörn SVENSSON
2022-09-22 22:42 ` Joseph Myers
2022-09-23 10:03 ` Torbjorn SVENSSON
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).