InstrProf: Emit runtime hook directly in IRGen

-u behaviour is apparently not portable between linkers (see cfe-commits
discussions for r204379 and r205012).  I've moved the logic to IRGen,
where it should have been in the first place.

I don't have a Linux system to test this on, so it's possible this logic
*still* doesn't pull in the instrumented profiling runtime on Linux.

I'm in the process of getting tests going on the compiler-rt side
(llvm-commits "[PATCH] InstrProf: Add initial compiler-rt test").  Once
we have tests for the full flow there, the runtime logic should get a
whole lot less brittle.

<rdar://problem/16458307>

llvm-svn: 205023

2 files changed, 14 insertions, 2 deletions

diff --git a/clang/test/Profile/c-linkage.c b/clang/test/Profile/c-linkage.c
index a688197d494..72eae1f0f32 100644
--- a/clang/test/Profile/c-linkage.c
+++ b/clang/test/Profile/c-linkage.c
@@ -1,6 +1,7 @@
 // Check the data structures emitted by instrumentation.
 // RUN: %clang_cc1 -triple x86_64-apple-macosx10.9 -main-file-name c-linkage.c %s -o - -emit-llvm -fprofile-instr-generate | FileCheck %s
 
+// CHECK: @__llvm_profile_runtime = external global i32
 // CHECK: @__llvm_profile_counters_foo = global [1 x i64] zeroinitializer, section "__DATA,__llvm_prf_cnts", align 8
 // CHECK: @__llvm_profile_name_foo = constant [3 x i8] c"foo", section "__DATA,__llvm_prf_names", align 1
 // CHECK: @__llvm_profile_data_foo = constant { i32, i32, i64, i8*, i64* } { i32 3, i32 1, i64 1, i8* getelementptr inbounds ([3 x i8]* @__llvm_profile_name_foo, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64]* @__llvm_profile_counters_foo, i32 0, i32 0) }, section "__DATA,__llvm_prf_data", align 8
@@ -28,4 +29,9 @@ int main(void) {
 // CHECK: @__llvm_profile_data_foo_internal = internal constant { i32, i32, i64, i8*, i64* } { i32 24, i32 3, i64 3, i8* getelementptr inbounds ([24 x i8]* @__llvm_profile_name_foo_internal, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64]* @__llvm_profile_counters_foo_internal, i32 0, i32 0) }, section "__DATA,__llvm_prf_data", align 8
 static void foo_internal(void) { if (0){} if (0){} }
 
-// CHECK: @llvm.used = appending global [4 x i8*] [i8* bitcast ({ i32, i32, i64, i8*, i64* }* @__llvm_profile_data_foo to i8*), i8* bitcast ({ i32, i32, i64, i8*, i64* }* @__llvm_profile_data_foo_weak to i8*), i8* bitcast ({ i32, i32, i64, i8*, i64* }* @__llvm_profile_data_main to i8*), i8* bitcast ({ i32, i32, i64, i8*, i64* }* @__llvm_profile_data_foo_internal to i8*)], section "llvm.metadata"
+// CHECK: @llvm.used = appending global [5 x i8*] [i8* bitcast (i32 ()* @__llvm_profile_runtime_user to i8*), i8* bitcast ({ i32, i32, i64, i8*, i64* }* @__llvm_profile_data_foo to i8*), i8* bitcast ({ i32, i32, i64, i8*, i64* }* @__llvm_profile_data_foo_weak to i8*), i8* bitcast ({ i32, i32, i64, i8*, i64* }* @__llvm_profile_data_main to i8*), i8* bitcast ({ i32, i32, i64, i8*, i64* }* @__llvm_profile_data_foo_internal to i8*)], section "llvm.metadata"
+
+// CHECK: define linkonce_odr i32 @__llvm_profile_runtime_user() {{.*}} {
+// CHECK:   %[[REG:.*]] = load i32* @__llvm_profile_runtime
+// CHECK:   ret i32 %[[REG]]
+// CHECK: }
diff --git a/clang/test/Profile/cxx-linkage.cpp b/clang/test/Profile/cxx-linkage.cpp
index 16f84a68b02..44c1983c849 100644
--- a/clang/test/Profile/cxx-linkage.cpp
+++ b/clang/test/Profile/cxx-linkage.cpp
@@ -1,5 +1,6 @@
 // RUN: %clang_cc1 -triple x86_64-apple-macosx10.9.0 -emit-llvm -main-file-name cxx-linkage.cpp %s -o - -fprofile-instr-generate | FileCheck %s
 
+// CHECK: @__llvm_profile_runtime = external global i32
 // CHECK: @__llvm_profile_counters__Z3foov = global [1 x i64] zeroinitializer, section "__DATA,__llvm_prf_cnts", align 8
 // CHECK: @__llvm_profile_name__Z3foov = constant [7 x i8] c"_Z3foov", section "__DATA,__llvm_prf_names", align 1
 // CHECK: @__llvm_profile_data__Z3foov = constant { i32, i32, i64, i8*, i64* } { i32 7, i32 1, i64 1, i8* getelementptr inbounds ([7 x i8]* @__llvm_profile_name__Z3foov, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64]* @__llvm_profile_counters__Z3foov, i32 0, i32 0) }, section "__DATA,__llvm_prf_data", align 8
@@ -27,4 +28,9 @@ int main(void) {
 // CHECK: @__llvm_profile_data__Z10foo_inlinev = linkonce_odr constant { i32, i32, i64, i8*, i64* } { i32 15, i32 7, i64 7, i8* getelementptr inbounds ([15 x i8]* @__llvm_profile_name__Z10foo_inlinev, i32 0, i32 0), i64* getelementptr inbounds ([7 x i64]* @__llvm_profile_counters__Z10foo_inlinev, i32 0, i32 0) }, section "__DATA,__llvm_prf_data", align 8
 inline void foo_inline(void) { if (0){} if (0){} if (0){} if (0){} if (0){} if (0){}}
 
-// CHECK: @llvm.used = appending global [4 x i8*] [i8* bitcast ({ i32, i32, i64, i8*, i64* }* @__llvm_profile_data__Z3foov to i8*), i8* bitcast ({ i32, i32, i64, i8*, i64* }* @__llvm_profile_data__Z8foo_weakv to i8*), i8* bitcast ({ i32, i32, i64, i8*, i64* }* @__llvm_profile_data_main to i8*), i8* bitcast ({ i32, i32, i64, i8*, i64* }* @__llvm_profile_data__Z10foo_inlinev to i8*)], section "llvm.metadata"
+// CHECK: @llvm.used = appending global [5 x i8*] [i8* bitcast (i32 ()* @__llvm_profile_runtime_user to i8*), i8* bitcast ({ i32, i32, i64, i8*, i64* }* @__llvm_profile_data__Z3foov to i8*), i8* bitcast ({ i32, i32, i64, i8*, i64* }* @__llvm_profile_data__Z8foo_weakv to i8*), i8* bitcast ({ i32, i32, i64, i8*, i64* }* @__llvm_profile_data_main to i8*), i8* bitcast ({ i32, i32, i64, i8*, i64* }* @__llvm_profile_data__Z10foo_inlinev to i8*)], section "llvm.metadata"
+
+// CHECK: define linkonce_odr i32 @__llvm_profile_runtime_user() {{.*}} {
+// CHECK:   %[[REG:.*]] = load i32* @__llvm_profile_runtime
+// CHECK:   ret i32 %[[REG]]
+// CHECK: }