On 12/09/14 11:13, Richard Earnshaw wrote:
On 09/09/14 16:14, Kyrill Tkachov wrote:
Hi all,
As Christophe mentioned at
https://gcc.gnu.org/ml/gcc-patches/2014-09/msg00202.html
These tests fail on big-endian. The reason is that the input is not
aligned to 128 bit forcing the use of a movmisalign which we don't
support on big-endian.
A solution is to force the alignment of the arrays, allowing for the use
of normal loads and stores.
We can look into enabling misaligned loads on big-endian with the
appropriate reversal logic as a separate
piece of work...
So surely these tests *ought* to pass on big-endian without this change
and the fact that they don't is a bug.
The backend disables misaligned loads in neon for big endian. You can't
do a vld1.8 on a movmisalignv2si and expect it to work correctly, can you ?
So in that case, surely we should be (at most XFAILing them), rather
than papering over a real problem.
I disagree - The tests are to detect vectorization of lceilf, lfloor and
all the other functions.
So until the backend is fixed, I asked for the tests to be changed to
test what they are supposed to - vectorizing calls to these builtin
functions which is more useful than XFAILing these.
regards
Ramana
R.
Ok for trunk?
2014-09-09 Kyrylo Tkachov <kyrylo.tkac...@arm.com>
* gcc.target/arm/vect-lceilf_1.c: Make input and output arrays global
and 16-byte aligned.
* gcc.target/arm/vect-lfloorf_1.c: Likewise.
* gcc.target/arm/vect-lroundf_1.c: Likewise.
* gcc.target/arm/vect-rounding-btruncf.c: Likewise.
* gcc.target/arm/vect-rounding-ceilf.c: Likewise.
* gcc.target/arm/vect-rounding-floorf.c: Likewise.
* gcc.target/arm/vect-rounding-roundf.c: Likewise.
arm-vect-tests-align.patch
diff --git a/gcc/testsuite/gcc.target/arm/vect-lceilf_1.c
b/gcc/testsuite/gcc.target/arm/vect-lceilf_1.c
index 75705ae..5e98b74 100644
--- a/gcc/testsuite/gcc.target/arm/vect-lceilf_1.c
+++ b/gcc/testsuite/gcc.target/arm/vect-lceilf_1.c
@@ -5,8 +5,11 @@
#define N 32
+float __attribute__((aligned(16))) input[N];
+int __attribute__((aligned(16))) output[N];
+
void
-foo (int *output, float *input)
+foo ()
{
int i = 0;
/* Vectorizable. */
diff --git a/gcc/testsuite/gcc.target/arm/vect-lfloorf_1.c
b/gcc/testsuite/gcc.target/arm/vect-lfloorf_1.c
index 298d54e..655f437 100644
--- a/gcc/testsuite/gcc.target/arm/vect-lfloorf_1.c
+++ b/gcc/testsuite/gcc.target/arm/vect-lfloorf_1.c
@@ -5,8 +5,11 @@
#define N 32
+float __attribute__((aligned(16))) input[N];
+int __attribute__((aligned(16))) output[N];
+
void
-foo (int *output, float *input)
+foo ()
{
int i = 0;
/* Vectorizable. */
diff --git a/gcc/testsuite/gcc.target/arm/vect-lroundf_1.c
b/gcc/testsuite/gcc.target/arm/vect-lroundf_1.c
index 6443821..92a722e 100644
--- a/gcc/testsuite/gcc.target/arm/vect-lroundf_1.c
+++ b/gcc/testsuite/gcc.target/arm/vect-lroundf_1.c
@@ -5,8 +5,11 @@
#define N 32
+float __attribute__((aligned(16))) input[N];
+int __attribute__((aligned(16))) output[N];
+
void
-foo (int *output, float *input)
+foo ()
{
int i = 0;
/* Vectorizable. */
diff --git a/gcc/testsuite/gcc.target/arm/vect-rounding-btruncf.c
b/gcc/testsuite/gcc.target/arm/vect-rounding-btruncf.c
index 5616837..372ddc5 100644
--- a/gcc/testsuite/gcc.target/arm/vect-rounding-btruncf.c
+++ b/gcc/testsuite/gcc.target/arm/vect-rounding-btruncf.c
@@ -5,8 +5,11 @@
#define N 32
+float __attribute__((aligned(16))) input[N];
+float __attribute__((aligned(16))) output[N];
+
void
-foo (float *output, float *input)
+foo ()
{
int i = 0;
/* Vectorizable. */
diff --git a/gcc/testsuite/gcc.target/arm/vect-rounding-ceilf.c
b/gcc/testsuite/gcc.target/arm/vect-rounding-ceilf.c
index cb8f1d5..3c786d4 100644
--- a/gcc/testsuite/gcc.target/arm/vect-rounding-ceilf.c
+++ b/gcc/testsuite/gcc.target/arm/vect-rounding-ceilf.c
@@ -5,8 +5,11 @@
#define N 32
+float __attribute__((aligned(16))) input[N];
+float __attribute__((aligned(16))) output[N];
+
void
-foo (float *output, float *input)
+foo ()
{
int i = 0;
/* Vectorizable. */
diff --git a/gcc/testsuite/gcc.target/arm/vect-rounding-floorf.c
b/gcc/testsuite/gcc.target/arm/vect-rounding-floorf.c
index bf68af7..eedb295 100644
--- a/gcc/testsuite/gcc.target/arm/vect-rounding-floorf.c
+++ b/gcc/testsuite/gcc.target/arm/vect-rounding-floorf.c
@@ -5,8 +5,11 @@
#define N 32
+float __attribute__((aligned(16))) input[N];
+float __attribute__((aligned(16))) output[N];
+
void
-foo (float *output, float *input)
+foo ()
{
int i = 0;
/* Vectorizable. */
diff --git a/gcc/testsuite/gcc.target/arm/vect-rounding-roundf.c
b/gcc/testsuite/gcc.target/arm/vect-rounding-roundf.c
index 7c0a1b4..360b2b9 100644
--- a/gcc/testsuite/gcc.target/arm/vect-rounding-roundf.c
+++ b/gcc/testsuite/gcc.target/arm/vect-rounding-roundf.c
@@ -5,8 +5,11 @@
#define N 32
+float __attribute__((aligned(16))) input[N];
+float __attribute__((aligned(16))) output[N];
+
void
-foo (float *output, float *input)
+foo ()
{
int i = 0;
/* Vectorizable. */
