[PATCH][AArch64] Two intrinsics are expected to return float64 not return float32.

Hao Liu Hao.Liu at arm.com
Thu Nov 28 18:30:29 PST 2013



This patch fixes a minor problem about arm_neon.h. 2 ALCE intrinsics have
incorrect return type.

For example in arm_neon.td, we define vcvt_f64:

def VCVT_F64 : SInst<"vcvt_f64", "fd",  "QlQUl">;

But in arm_neon.h, we get following incorrect intrinsics:

__ai float32x4_t vcvtq_f64_s64(int64x2_t __a) {

(float32x4_t)__builtin_neon_vcvtq_f64_v((int8x16_t)__a, 35); }

__ai float32x4_t vcvtq_f64_u64(uint64x2_t __a) {

(float32x4_t)__builtin_neon_vcvtq_f64_v((int8x16_t)__a, 51); }

The return type should be flaot64x2_t.


There are already many tests about such intrinsics. I don't add any tests.


As this is only a minor bug and I think it needs no code review, I'll commit
the patch firstly. 

If someone has comment about this, please tell me.





-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/cfe-commits/attachments/20131129/92ab4fd9/attachment.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: ret_f64.patch
Type: application/octet-stream
Size: 3303 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/cfe-commits/attachments/20131129/92ab4fd9/attachment.obj>

More information about the cfe-commits mailing list