1 // Test for linking with CUDA's libdevice as outlined in
2 // http://llvm.org/docs/NVPTXUsage.html#linking-with-libdevice
3 //
4 // REQUIRES: nvptx-registered-target
5 //
6 // Prepare bitcode file to link with
7 // RUN: %clang_cc1 -triple nvptx-unknown-cuda -emit-llvm-bc \
8 // RUN: -disable-llvm-passes -o %t.bc %S/Inputs/device-code.ll
9 // RUN: %clang_cc1 -triple nvptx-unknown-cuda -emit-llvm-bc \
10 // RUN: -disable-llvm-passes -o %t-2.bc %S/Inputs/device-code-2.ll
11 //
12 // Make sure function in device-code gets linked in and internalized.
13 // RUN: %clang_cc1 -triple nvptx-unknown-cuda -fcuda-is-device \
14 // RUN: -mlink-cuda-bitcode %t.bc -emit-llvm \
15 // RUN: -disable-llvm-passes -o - %s \
16 // RUN: | FileCheck %s -check-prefix CHECK-IR
17 //
18 // Make sure we can link two bitcode files.
19 // RUN: %clang_cc1 -triple nvptx-unknown-cuda -fcuda-is-device \
20 // RUN: -mlink-cuda-bitcode %t.bc -mlink-cuda-bitcode %t-2.bc \
21 // RUN: -emit-llvm -disable-llvm-passes -o - %s \
22 // RUN: | FileCheck %s -check-prefix CHECK-IR -check-prefix CHECK-IR-2
23 //
24 // Make sure function in device-code gets linked but is not internalized
25 // without -fcuda-uses-libdevice
26 // RUN: %clang_cc1 -triple nvptx-unknown-cuda -fcuda-is-device \
27 // RUN: -mlink-bitcode-file %t.bc -emit-llvm \
28 // RUN: -disable-llvm-passes -o - %s \
29 // RUN: | FileCheck %s -check-prefix CHECK-IR-NLD
30 //
31 // Make sure NVVMReflect pass is enabled in NVPTX back-end.
32 // RUN: %clang_cc1 -triple nvptx-unknown-cuda -fcuda-is-device \
33 // RUN: -mlink-cuda-bitcode %t.bc -S -o /dev/null %s \
34 // RUN: -backend-option -debug-pass=Structure 2>&1 \
35 // RUN: | FileCheck %s -check-prefix CHECK-REFLECT
36
37 #include "Inputs/cuda.h"
38
39 __device__ float device_mul_or_add(float a, float b);
40 extern "C" __device__ double __nv_sin(double x);
41 extern "C" __device__ double __nv_exp(double x);
42
43 // CHECK-IR-LABEL: define void @_Z26should_not_be_internalizedPf(
44 // CHECK-PTX-LABEL: .visible .func _Z26should_not_be_internalizedPf(
should_not_be_internalized(float * data)45 __device__ void should_not_be_internalized(float *data) {}
46
47 // Make sure kernel call has not been internalized.
48 // CHECK-IR-LABEL: define void @_Z6kernelPfS_
49 // CHECK-PTX-LABEL: .visible .entry _Z6kernelPfS_(
kernel(float * out,float * in)50 __global__ __attribute__((used)) void kernel(float *out, float *in) {
51 *out = device_mul_or_add(in[0], in[1]);
52 *out += __nv_exp(__nv_sin(*out));
53 should_not_be_internalized(out);
54 }
55
56 // Make sure device_mul_or_add() is present in IR, is internal and
57 // calls __nvvm_reflect().
58 // CHECK-IR-LABEL: define internal float @_Z17device_mul_or_addff(
59 // CHECK-IR-NLD-LABEL: define float @_Z17device_mul_or_addff(
60 // CHECK-IR: call i32 @__nvvm_reflect
61 // CHECK-IR: ret float
62
63 // Make sure we've linked in and internalized only needed functions
64 // from the second bitcode file.
65 // CHECK-IR-2-LABEL: define internal double @__nv_sin
66 // CHECK-IR-2-LABEL: define internal double @__nv_exp
67 // CHECK-IR-2-NOT: double @__unused
68
69 // Verify that NVVMReflect pass is among the passes run by NVPTX back-end.
70 // CHECK-REFLECT: Replace occurrences of __nvvm_reflect() calls with 0/1
71