The ability to use vector initializer lists is a GNU vector extension

and is unrelated to the NEON intrinsics in arm_neon.h. On little
endian machines it works fine, however on big endian machines it
exhibits surprising behaviour:

    uint32x2_t x = {42, 64};
    return vget_lane_u32(x, 0); // Will return 64.

Because of this, explicitly call out that it is unsupported on big
endian machines.

This patch will emit the following warning in big-endian mode:

    test.c:3:15: warning: vector initializers are a GNU extension and are not compatible with NEON intrinsics [-Wgnu]
    int32x4_t x = {0, 1, 2, 3};
                  ^
    test.c:3:15: note: consider using vld1q_s32() to initialize a vector from memory, or vcombine_s32(vcreate_s32(), vcreate_s32()) to initialize from integer constants
    1 warning generated.

llvm-svn: 211362

1 files changed, 13 insertions, 0 deletions

diff --git a/clang/test/Sema/big-endian-neon-initializers.c b/clang/test/Sema/big-endian-neon-initializers.c
new file mode 100644
index 00000000000..ffe310903bc
--- /dev/null
+++ b/clang/test/Sema/big-endian-neon-initializers.c
@@ -0,0 +1,13 @@
+// RUN: %clang_cc1 %s -triple arm64_be -target-feature +neon -verify -fsyntax-only -ffreestanding
+// RUN: %clang_cc1 %s -triple armebv7 -target-cpu cortex-a8 -verify -fsyntax-only -ffreestanding
+
+#include <arm_neon.h>
+
+int32x4_t x = {1, 2, 3, 4}; // expected-warning{{vector initializers are not compatible with NEON intrinsics}} expected-note{{consider using vld1q_s32() to initialize a vector from memory, or vcombine_s32(vcreate_s32(), vcreate_s32()) to initialize from integer constants}}
+int16x4_t y = {1, 2, 3, 4}; // expected-warning{{vector initializers are not compatible with NEON intrinsics}} expected-note{{consider using vld1_s16() to initialize a vector from memory, or vcreate_s16() to initialize from an integer constant}}
+int64x2_t z = {1, 2}; // expected-warning{{vector initializers are not compatible with NEON intrinsics}} expected-note{{consider using vld1q_s64() to initialize a vector from memory, or vcombine_s64(vcreate_s64(), vcreate_s64()) to initialize from integer constants}}
+float32x2_t b = {1, 2}; // expected-warning{{vector initializers are not compatible with NEON intrinsics}} expected-note{{consider using vld1_f32() to initialize a vector from memory, or vcreate_f32() to initialize from an integer constant}}
+
+// No warning expected here.
+typedef int v4si __attribute__ ((vector_size (16)));
+v4si c = {1, 2, 3, 4};