| Justin Lebar | b080b63 | 2017-01-25 21:29:48 +0000 | [diff] [blame^] | 1 | // Check that when we link a bitcode module into a file using |
| 2 | // -mlink-cuda-bitcode, we apply the same attributes to the functions in that |
| 3 | // bitcode module as we apply to functions we generate. |
| 4 | // |
| 5 | // In particular, we check that ftz and unsafe-math are propagated into the |
| 6 | // bitcode library as appropriate. |
| 7 | // |
| 8 | // In addition, we set -ftrapping-math on the bitcode library, but then set |
| 9 | // -fno-trapping-math on the main compilations, and ensure that the latter flag |
| 10 | // overrides the flag on the bitcode library. |
| 11 | |
| 12 | // Build the bitcode library. This is not built in CUDA mode, otherwise it |
| 13 | // might have incompatible attributes. This mirrors how libdevice is built. |
| 14 | // RUN: %clang_cc1 -x c++ -emit-llvm-bc -ftrapping-math -DLIB \ |
| 15 | // RUN: %s -o %t.bc -triple nvptx-unknown-unknown |
| 16 | |
| 17 | // RUN: %clang_cc1 -x cuda %s -emit-llvm -mlink-cuda-bitcode %t.bc -o - \ |
| 18 | // RUN: -fno-trapping-math -fcuda-is-device -triple nvptx-unknown-unknown \ |
| 19 | // RUN: | FileCheck %s --check-prefix=CHECK --check-prefix=NOFTZ --check-prefix=NOFAST |
| 20 | |
| 21 | // RUN: %clang_cc1 -x cuda %s -emit-llvm -mlink-cuda-bitcode %t.bc \ |
| 22 | // RUN: -fno-trapping-math -fcuda-flush-denormals-to-zero -o - \ |
| 23 | // RUN: -fcuda-is-device -triple nvptx-unknown-unknown \ |
| 24 | // RUN: | FileCheck %s --check-prefix=CHECK --check-prefix=FTZ \ |
| 25 | // RUN: --check-prefix=NOFAST |
| 26 | |
| 27 | // RUN: %clang_cc1 -x cuda %s -emit-llvm -mlink-cuda-bitcode %t.bc \ |
| 28 | // RUN: -fno-trapping-math -fcuda-flush-denormals-to-zero -o - \ |
| 29 | // RUN: -fcuda-is-device -menable-unsafe-fp-math -triple nvptx-unknown-unknown \ |
| 30 | // RUN: | FileCheck %s --check-prefix=CHECK --check-prefix=FAST |
| 31 | |
| 32 | // Wrap everything in extern "C" so we don't ahve to worry about name mangling |
| 33 | // in the IR. |
| 34 | extern "C" { |
| 35 | #ifdef LIB |
| 36 | |
| 37 | // This function is defined in the library and only declared in the main |
| 38 | // compilation. |
| 39 | void lib_fn() {} |
| 40 | |
| 41 | #else |
| 42 | |
| 43 | #include "Inputs/cuda.h" |
| 44 | __device__ void lib_fn(); |
| 45 | __global__ void kernel() { lib_fn(); } |
| 46 | |
| 47 | #endif |
| 48 | } |
| 49 | |
| 50 | // The kernel and lib function should have the same attributes. |
| 51 | // CHECK: define void @kernel() [[attr:#[0-9]+]] |
| 52 | // CHECK: define internal void @lib_fn() [[attr]] |
| 53 | |
| 54 | // Check the attribute list. |
| 55 | // CHECK: attributes [[attr]] = { |
| 56 | // CHECK: "no-trapping-math"="true" |
| 57 | |
| 58 | // FTZ-SAME: "nvptx-f32ftz"="true" |
| 59 | // NOFTZ-NOT: "nvptx-f32ftz"="true" |
| 60 | |
| 61 | // FAST-SAME: "unsafe-fp-math"="true" |
| 62 | // NOFAST-NOT: "unsafe-fp-math"="true" |