RE: [PATCH] [PR51534] [arm] split out pr51534 test for softfp

2023-03-03 Thread Kyrylo Tkachov via Gcc-patches 



> -Original Message-
> From: Alexandre Oliva 
> Sent: Friday, February 17, 2023 7:14 AM
> To: gcc-patches@gcc.gnu.org
> Cc: ni...@redhat.com; Richard Earnshaw 
> Subject: [PATCH] [PR51534] [arm] split out pr51534 test for softfp
> 
> 
> The test uses arm_hard_ok and arm_softfp_ok as if they were mutually
> exclusive, but they test whether the corresponding -mfloat-abi= flag
> is usable, not whether it is in effect, so it is possible for both to
> pass, and then the test comes out with incorrect expectations
> whichever the default float-abi is.
> 
> Separate the test into hard and softfp variants, and extend the softfp
> variant to accept both ARM and Thumb opcodes; it unwarrantedly assumed
> the latter.
> 
> Regstrapped on x86_64-linux-gnu.
> Tested on arm-vxworks7 (gcc-12) and arm-eabi (trunk).  Ok to install?

Ok.
Thanks,
Kyrill

> 
> for  gcc/testsuite/ChangeLog
> 
>   PR target/51534
>   * gcc.target/arm/pr51534.c: Split softfp variant into...
>   * gcc.target/arm/pr51534s.c: ... this, and support ARM too.
> ---
>  gcc/testsuite/gcc.target/arm/pr51534.c  |9 ++--
>  gcc/testsuite/gcc.target/arm/pr51534s.c |   72
> +++
>  2 files changed, 76 insertions(+), 5 deletions(-)
>  create mode 100644 gcc/testsuite/gcc.target/arm/pr51534s.c
> 
> diff --git a/gcc/testsuite/gcc.target/arm/pr51534.c
> b/gcc/testsuite/gcc.target/arm/pr51534.c
> index 5e121f5fb9946..ba21259bee554 100644
> --- a/gcc/testsuite/gcc.target/arm/pr51534.c
> +++ b/gcc/testsuite/gcc.target/arm/pr51534.c
> @@ -1,9 +1,9 @@
>  /* Test the vector comparison intrinsics when comparing to immediate zero.
> */
> 
> -/* { dg-do assemble } */
> +/* { dg-do assemble { target { arm_hard_ok } } } */
>  /* { dg-require-effective-target arm_neon_ok } */
> -/* { dg-options "-save-temps -O3" } */
> +/* { dg-options "-save-temps -mfloat-abi=hard -O3" } */
>  /* { dg-add-options arm_neon } */
> 
>  #include 
> @@ -64,9 +64,8 @@ GEN_COND_TESTS(vceq)
>  /* { dg-final { scan-assembler-times "vceq\.i8\[ \]+\[qQ\]\[0-9\]+,
> \[qQ\]\[0-9\]+, #0" 4 } } */
>  /* { dg-final { scan-assembler-times "vceq\.i16\[\]+\[qQ\]\[0-9\]+,
> \[qQ\]\[0-9\]+, #0" 4 } } */
>  /* { dg-final { scan-assembler-times "vceq\.i32\[\]+\[qQ\]\[0-9\]+,
> \[qQ\]\[0-9\]+, #0" 4 } } */
> -/* { dg-final { scan-assembler-times "vmov\.i32\[\]+\[dD\]\[0-9\]+,
> #0x" 3 { target { arm_hard_ok } } } } */
> -/* { dg-final { scan-assembler-times "vmov\.i32\[\]+\[qQ\]\[0-9\]+,
> #4294967295" 3 { target { arm_hard_ok } } } } */
> -/* { dg-final { scan-assembler-times "mov\[  \]+r\[0-9\]+, #-1" 6 { target {
> arm_softfp_ok } } } } */
> +/* { dg-final { scan-assembler-times "vmov\.i32\[\]+\[dD\]\[0-9\]+,
> #0x" 3 } } */
> +/* { dg-final { scan-assembler-times "vmov\.i32\[\]+\[qQ\]\[0-9\]+,
> #4294967295" 3 } } */
> 
>  /* And ensure we don't have unexpected output too.  */
>  /* { dg-final { scan-assembler-not "vc\[gl\]\[te\]\.u\[0-9\]+\[
>   \]+\[qQdD\]\[0-9\]+, \[qQdD\]\[0-9\]+, #0" } } */
> diff --git a/gcc/testsuite/gcc.target/arm/pr51534s.c
> b/gcc/testsuite/gcc.target/arm/pr51534s.c
> new file mode 100644
> index 0..b1638919c2f75
> --- /dev/null
> +++ b/gcc/testsuite/gcc.target/arm/pr51534s.c
> @@ -0,0 +1,72 @@
> +/* Test the vector comparison intrinsics when comparing to immediate
> zero.
> +   */
> +
> +/* { dg-do assemble { target { arm_softfp_ok } } } */
> +/* { dg-require-effective-target arm_neon_ok } */
> +/* { dg-options "-save-temps -mfloat-abi=softfp -O3" } */
> +/* { dg-add-options arm_neon } */
> +
> +#include 
> +
> +#define GEN_TEST(T, D, C, R) \
> +  R test_##C##_##T (T a) { return C (a, D (0)); }
> +
> +#define GEN_DOUBLE_TESTS(S, T, C) \
> +  GEN_TEST (T, vdup_n_s##S, C##_s##S, u##T) \
> +  GEN_TEST (u##T, vdup_n_u##S, C##_u##S, u##T)
> +
> +#define GEN_QUAD_TESTS(S, T, C) \
> +  GEN_TEST (T, vdupq_n_s##S, C##q_s##S, u##T) \
> +  GEN_TEST (u##T, vdupq_n_u##S, C##q_u##S, u##T)
> +
> +#define GEN_COND_TESTS(C) \
> +  GEN_DOUBLE_TESTS (8, int8x8_t, C) \
> +  GEN_DOUBLE_TESTS (16, int16x4_t, C) \
> +  GEN_DOUBLE_TESTS (32, int32x2_t, C) \
> +  GEN_QUAD_TESTS (8, int8x16_t, C) \
> +  GEN_QUAD_TESTS (16, int16x8_t, C) \
> +  GEN_QUAD_TESTS (32, int32x4_t, C)
> +
> +GEN_COND_TESTS(vcgt)
> +GEN_COND_TESTS(vcge)
> +GEN_COND_TESTS(vclt)
> +GEN_COND_TESTS(vcle)
> +GEN_COND_TESTS(vceq)
> +
> +/* Scan for expected outputs.  */
> +/* { dg-final { scan-assembler "vcgt\.s8\[   \]+\[dD\]\[0-9\]+, \[dD\]\[0-
> 9\]+, #0&q

Re: [PATCH] [PR51534] [arm] split out pr51534 test for softfp

2023-03-03 Thread Alexandre Oliva via Gcc-patches 
On Feb 17, 2023, Alexandre Oliva  wrote:

>   PR target/51534
>   * gcc.target/arm/pr51534.c: Split softfp variant into...
>   * gcc.target/arm/pr51534s.c: ... this, and support ARM too.

Ping?
https://gcc.gnu.org/pipermail/gcc-patches/2023-February/612189.html

-- 
Alexandre Oliva, happy hackerhttps://FSFLA.org/blogs/lxo/
   Free Software Activist   GNU Toolchain Engineer
Disinformation flourishes because many people care deeply about injustice
but very few check the facts.  Ask me about

[PATCH] [PR51534] [arm] split out pr51534 test for softfp

2023-02-16 Thread Alexandre Oliva via Gcc-patches 


The test uses arm_hard_ok and arm_softfp_ok as if they were mutually
exclusive, but they test whether the corresponding -mfloat-abi= flag
is usable, not whether it is in effect, so it is possible for both to
pass, and then the test comes out with incorrect expectations
whichever the default float-abi is.

Separate the test into hard and softfp variants, and extend the softfp
variant to accept both ARM and Thumb opcodes; it unwarrantedly assumed
the latter.

Regstrapped on x86_64-linux-gnu.
Tested on arm-vxworks7 (gcc-12) and arm-eabi (trunk).  Ok to install?

for  gcc/testsuite/ChangeLog

PR target/51534
* gcc.target/arm/pr51534.c: Split softfp variant into...
* gcc.target/arm/pr51534s.c: ... this, and support ARM too.
---
 gcc/testsuite/gcc.target/arm/pr51534.c  |9 ++--
 gcc/testsuite/gcc.target/arm/pr51534s.c |   72 +++
 2 files changed, 76 insertions(+), 5 deletions(-)
 create mode 100644 gcc/testsuite/gcc.target/arm/pr51534s.c

diff --git a/gcc/testsuite/gcc.target/arm/pr51534.c 
b/gcc/testsuite/gcc.target/arm/pr51534.c
index 5e121f5fb9946..ba21259bee554 100644
--- a/gcc/testsuite/gcc.target/arm/pr51534.c
+++ b/gcc/testsuite/gcc.target/arm/pr51534.c
@@ -1,9 +1,9 @@
 /* Test the vector comparison intrinsics when comparing to immediate zero.
*/
 
-/* { dg-do assemble } */
+/* { dg-do assemble { target { arm_hard_ok } } } */
 /* { dg-require-effective-target arm_neon_ok } */
-/* { dg-options "-save-temps -O3" } */
+/* { dg-options "-save-temps -mfloat-abi=hard -O3" } */
 /* { dg-add-options arm_neon } */
 
 #include 
@@ -64,9 +64,8 @@ GEN_COND_TESTS(vceq)
 /* { dg-final { scan-assembler-times "vceq\.i8\[   \]+\[qQ\]\[0-9\]+, 
\[qQ\]\[0-9\]+, #0" 4 } } */
 /* { dg-final { scan-assembler-times "vceq\.i16\[  \]+\[qQ\]\[0-9\]+, 
\[qQ\]\[0-9\]+, #0" 4 } } */
 /* { dg-final { scan-assembler-times "vceq\.i32\[  \]+\[qQ\]\[0-9\]+, 
\[qQ\]\[0-9\]+, #0" 4 } } */
-/* { dg-final { scan-assembler-times "vmov\.i32\[  \]+\[dD\]\[0-9\]+, 
#0x" 3 { target { arm_hard_ok } } } } */
-/* { dg-final { scan-assembler-times "vmov\.i32\[  \]+\[qQ\]\[0-9\]+, 
#4294967295" 3 { target { arm_hard_ok } } } } */
-/* { dg-final { scan-assembler-times "mov\[\]+r\[0-9\]+, #-1" 6 { target { 
arm_softfp_ok } } } } */
+/* { dg-final { scan-assembler-times "vmov\.i32\[  \]+\[dD\]\[0-9\]+, 
#0x" 3 } } */
+/* { dg-final { scan-assembler-times "vmov\.i32\[  \]+\[qQ\]\[0-9\]+, 
#4294967295" 3 } } */
 
 /* And ensure we don't have unexpected output too.  */
 /* { dg-final { scan-assembler-not "vc\[gl\]\[te\]\.u\[0-9\]+\[
\]+\[qQdD\]\[0-9\]+, \[qQdD\]\[0-9\]+, #0" } } */
diff --git a/gcc/testsuite/gcc.target/arm/pr51534s.c 
b/gcc/testsuite/gcc.target/arm/pr51534s.c
new file mode 100644
index 0..b1638919c2f75
--- /dev/null
+++ b/gcc/testsuite/gcc.target/arm/pr51534s.c
@@ -0,0 +1,72 @@
+/* Test the vector comparison intrinsics when comparing to immediate zero.
+   */
+
+/* { dg-do assemble { target { arm_softfp_ok } } } */
+/* { dg-require-effective-target arm_neon_ok } */
+/* { dg-options "-save-temps -mfloat-abi=softfp -O3" } */
+/* { dg-add-options arm_neon } */
+
+#include 
+
+#define GEN_TEST(T, D, C, R) \
+  R test_##C##_##T (T a) { return C (a, D (0)); }
+
+#define GEN_DOUBLE_TESTS(S, T, C) \
+  GEN_TEST (T, vdup_n_s##S, C##_s##S, u##T) \
+  GEN_TEST (u##T, vdup_n_u##S, C##_u##S, u##T) 
+
+#define GEN_QUAD_TESTS(S, T, C) \
+  GEN_TEST (T, vdupq_n_s##S, C##q_s##S, u##T) \
+  GEN_TEST (u##T, vdupq_n_u##S, C##q_u##S, u##T) 
+
+#define GEN_COND_TESTS(C) \
+  GEN_DOUBLE_TESTS (8, int8x8_t, C) \
+  GEN_DOUBLE_TESTS (16, int16x4_t, C) \
+  GEN_DOUBLE_TESTS (32, int32x2_t, C) \
+  GEN_QUAD_TESTS (8, int8x16_t, C) \
+  GEN_QUAD_TESTS (16, int16x8_t, C) \
+  GEN_QUAD_TESTS (32, int32x4_t, C)
+
+GEN_COND_TESTS(vcgt)
+GEN_COND_TESTS(vcge)
+GEN_COND_TESTS(vclt)
+GEN_COND_TESTS(vcle)
+GEN_COND_TESTS(vceq)
+
+/* Scan for expected outputs.  */
+/* { dg-final { scan-assembler "vcgt\.s8\[ \]+\[dD\]\[0-9\]+, 
\[dD\]\[0-9\]+, #0" } } */
+/* { dg-final { scan-assembler "vcgt\.s16\[\]+\[dD\]\[0-9\]+, 
\[dD\]\[0-9\]+, #0" } } */
+/* { dg-final { scan-assembler "vcgt\.s32\[\]+\[dD\]\[0-9\]+, 
\[dD\]\[0-9\]+, #0" } } */
+/* { dg-final { scan-assembler "vcgt\.s8\[ \]+\[qQ\]\[0-9\]+, 
\[qQ\]\[0-9\]+, #0" } } */
+/* { dg-final { scan-assembler "vcgt\.s16\[\]+\[qQ\]\[0-9\]+, 
\[qQ\]\[0-9\]+, #0" } } */
+/* { dg-final { scan-assembler "vcgt\.s32\[\]+\[qQ\]\[0-9\]+, 
\[qQ\]\[0-9\]+, #0" } } */
+/* { dg-final { scan-assembler "vcge\.s8\[ \]+\[dD\]\[0-9\]+, 
\[dD\]\[0-9\]+, #0" } } */
+/* { dg-final { scan-assembler "vcge\.s16\[\]+\[dD\]\[0-9\]+, 
\[dD\]\[0-9\]+, #0" } } */
+/* { dg-final { scan-assembler "vcge\.s32\[\]+\[dD\]\[0-9\]+, 
\[dD\]\[0-9\]+, #0" } } */
+/* { dg-final { scan-assembler "vcge\.s8\[ \]+\[qQ\]\[0-9\]+, 
\[qQ\]\[0-9\]+, #0" } } */
+/* { dg-final { scan-assembler "vcge\.s16\[