Merged master:5c31b8b94fdd into amd-gfx:84156f83019c

Local branch amd-gfx 84156f8 Merged master:3eaec1fead4d into amd-gfx:d56063d98a2c Remote branch master 5c31b8b Revert "Use uint64_t for branch weights instead of uint32_t"
jaebaek · Oct 31, 2020 · 7e9078f · 7e9078f
2 parents 84156f8 + 5c31b8b
commit 7e9078f
Show file tree

Hide file tree

Showing 126 changed files with 581 additions and 1,189 deletions.
diff --git a/clang/lib/CodeGen/CodeGenPGO.cpp b/clang/lib/CodeGen/CodeGenPGO.cpp
@@ -1013,21 +1013,42 @@ void CodeGenPGO::loadRegionCounts(llvm::IndexedInstrProfReader *PGOReader,
   RegionCounts = ProfRecord->Counts;
 }
 
-/// Scale an individual branch weight (add 1).
+/// Calculate what to divide by to scale weights.
+///
+/// Given the maximum weight, calculate a divisor that will scale all the
+/// weights to strictly less than UINT32_MAX.
+static uint64_t calculateWeightScale(uint64_t MaxWeight) {
+  return MaxWeight < UINT32_MAX ? 1 : MaxWeight / UINT32_MAX + 1;
+}
+
+/// Scale an individual branch weight (and add 1).
+///
+/// Scale a 64-bit weight down to 32-bits using \c Scale.
 ///
 /// According to Laplace's Rule of Succession, it is better to compute the
 /// weight based on the count plus 1, so universally add 1 to the value.
-static uint64_t scaleBranchWeight(uint64_t Weight) { return Weight + 1; }
+///
+/// \pre \c Scale was calculated by \a calculateWeightScale() with a weight no
+/// greater than \c Weight.
+static uint32_t scaleBranchWeight(uint64_t Weight, uint64_t Scale) {
+  assert(Scale && "scale by 0?");
+  uint64_t Scaled = Weight / Scale + 1;
+  assert(Scaled <= UINT32_MAX && "overflow 32-bits");
+  return Scaled;
+}
 
 llvm::MDNode *CodeGenFunction::createProfileWeights(uint64_t TrueCount,
                                                     uint64_t FalseCount) const {
   // Check for empty weights.
   if (!TrueCount && !FalseCount)
     return nullptr;
 
+  // Calculate how to scale down to 32-bits.
+  uint64_t Scale = calculateWeightScale(std::max(TrueCount, FalseCount));
+
   llvm::MDBuilder MDHelper(CGM.getLLVMContext());
-  return MDHelper.createBranchWeights(scaleBranchWeight(TrueCount),
-                                      scaleBranchWeight(FalseCount));
+  return MDHelper.createBranchWeights(scaleBranchWeight(TrueCount, Scale),
+                                      scaleBranchWeight(FalseCount, Scale));
 }
 
 llvm::MDNode *
@@ -1041,10 +1062,13 @@ CodeGenFunction::createProfileWeights(ArrayRef<uint64_t> Weights) const {
   if (MaxWeight == 0)
     return nullptr;
 
-  SmallVector<uint64_t, 16> ScaledWeights;
+  // Calculate how to scale down to 32-bits.
+  uint64_t Scale = calculateWeightScale(MaxWeight);
+
+  SmallVector<uint32_t, 16> ScaledWeights;
   ScaledWeights.reserve(Weights.size());
   for (uint64_t W : Weights)
-    ScaledWeights.push_back(scaleBranchWeight(W));
+    ScaledWeights.push_back(scaleBranchWeight(W, Scale));
 
   llvm::MDBuilder MDHelper(CGM.getLLVMContext());
   return MDHelper.createBranchWeights(ScaledWeights);

diff --git a/clang/test/CodeGen/catch-undef-behavior.c b/clang/test/CodeGen/catch-undef-behavior.c
@@ -385,6 +385,6 @@ void call_nonnull_variadic(int a, int *b) {
   nonnull_variadic(a, b);
 }
 
-// CHECK-UBSAN: ![[WEIGHT_MD]] = !{!"branch_weights", i64 1048575, i64 1}
+// CHECK-UBSAN: ![[WEIGHT_MD]] = !{!"branch_weights", i32 1048575, i32 1}
 
 // CHECK-TRAP: attributes [[NR_NUW]] = { noreturn nounwind }
diff --git a/clang/test/CodeGenCXX/attr-likelihood-if-branch-weights.cpp b/clang/test/CodeGenCXX/attr-likelihood-if-branch-weights.cpp
@@ -142,5 +142,5 @@ void SwitchStmt() {
   }
 }
 
-// CHECK: !7 = !{!"branch_weights", i64 [[UNLIKELY]], i64 [[LIKELY]]}
-// CHECK: !8 = !{!"branch_weights", i64 [[LIKELY]], i64 [[UNLIKELY]]}
+// CHECK: !7 = !{!"branch_weights", i32 [[UNLIKELY]], i32 [[LIKELY]]}
+// CHECK: !8 = !{!"branch_weights", i32 [[LIKELY]], i32 [[UNLIKELY]]}
diff --git a/clang/test/CodeGenCXX/attr-likelihood-if-vs-builtin-expect.cpp b/clang/test/CodeGenCXX/attr-likelihood-if-vs-builtin-expect.cpp
@@ -221,5 +221,5 @@ void tu2(int &i) {
   }
 }
 
-// CHECK: !6 = !{!"branch_weights", i64 2000, i64 1}
-// CHECK: !10 = !{!"branch_weights", i64 1, i64 2000}
+// CHECK: !6 = !{!"branch_weights", i32 2000, i32 1}
+// CHECK: !10 = !{!"branch_weights", i32 1, i32 2000}
diff --git a/clang/test/CodeGenCXX/attr-likelihood-switch-branch-weights.cpp b/clang/test/CodeGenCXX/attr-likelihood-switch-branch-weights.cpp
@@ -177,18 +177,18 @@ void TwoCasesDefaultLNU() {
   }
 }
 
-// CHECK: !6 = !{!"branch_weights", i64 357913942, i64 715827883}
-// CHECK: !7 = !{!"branch_weights", i64 536870912, i64 1}
-// CHECK: !8 = !{!"branch_weights", i64 238609295, i64 715827883, i64 238609295}
-// CHECK: !9 = !{!"branch_weights", i64 357913942, i64 1, i64 357913942}
-// CHECK: !10 = !{!"branch_weights", i64 357913942, i64 715827883, i64 1}
-// CHECK: !11 = !{!"branch_weights", i64 143165577, i64 143165577, i64 143165577, i64 715827883, i64 143165577}
-// CHECK: !12 = !{!"branch_weights", i64 214748365, i64 214748365, i64 214748365, i64 1, i64 214748365}
-// CHECK: !13 = !{!"branch_weights", i64 79536432, i64 79536432, i64 79536432, i64 79536432, i64 79536432, i64 79536432, i64 79536432, i64 715827883, i64 79536432}
-// CHECK: !14 = !{!"branch_weights", i64 119304648, i64 119304648, i64 119304648, i64 119304648, i64 119304648, i64 119304648, i64 119304648, i64 1, i64 119304648}
-// CHECK: !15 = !{!"branch_weights", i64 2000, i64 1}
-// CHECK: !16 = !{!"branch_weights", i64 1, i64 2000}
-// CHECK: !17 = !{!"branch_weights", i64 715827883, i64 357913942}
-// CHECK: !18 = !{!"branch_weights", i64 1, i64 536870912}
-// CHECK: !19 = !{!"branch_weights", i64 536870912, i64 536870912, i64 268435456}
-// CHECK: !20 = !{!"branch_weights", i64 1, i64 715827883, i64 357913942}
+// CHECK: !6 = !{!"branch_weights", i32 357913942, i32 715827883}
+// CHECK: !7 = !{!"branch_weights", i32 536870912, i32 1}
+// CHECK: !8 = !{!"branch_weights", i32 238609295, i32 715827883, i32 238609295}
+// CHECK: !9 = !{!"branch_weights", i32 357913942, i32 1, i32 357913942}
+// CHECK: !10 = !{!"branch_weights", i32 357913942, i32 715827883, i32 1}
+// CHECK: !11 = !{!"branch_weights", i32 143165577, i32 143165577, i32 143165577, i32 715827883, i32 143165577}
+// CHECK: !12 = !{!"branch_weights", i32 214748365, i32 214748365, i32 214748365, i32 1, i32 214748365}
+// CHECK: !13 = !{!"branch_weights", i32 79536432, i32 79536432, i32 79536432, i32 79536432, i32 79536432, i32 79536432, i32 79536432, i32 715827883, i32 79536432}
+// CHECK: !14 = !{!"branch_weights", i32 119304648, i32 119304648, i32 119304648, i32 119304648, i32 119304648, i32 119304648, i32 119304648, i32 1, i32 119304648}
+// CHECK: !15 = !{!"branch_weights", i32 2000, i32 1}
+// CHECK: !16 = !{!"branch_weights", i32 1, i32 2000}
+// CHECK: !17 = !{!"branch_weights", i32 715827883, i32 357913942}
+// CHECK: !18 = !{!"branch_weights", i32 1, i32 536870912}
+// CHECK: !19 = !{!"branch_weights", i32 536870912, i32 536870912, i32 268435456}
+// CHECK: !20 = !{!"branch_weights", i32 1, i32 715827883, i32 357913942}
diff --git a/clang/test/CodeGenCXX/microsoft-abi-thread-safe-statics.cpp b/clang/test/CodeGenCXX/microsoft-abi-thread-safe-statics.cpp
@@ -96,5 +96,5 @@ int g1() {
   return i;
 }
 
-// CHECK-DAG: ![[unlikely_threadlocal]] = !{!"branch_weights", i64 1, i64 1023}
-// CHECK-DAG: ![[unlikely_staticlocal]] = !{!"branch_weights", i64 1, i64 1048575}
+// CHECK-DAG: ![[unlikely_threadlocal]] = !{!"branch_weights", i32 1, i32 1023}
+// CHECK-DAG: ![[unlikely_staticlocal]] = !{!"branch_weights", i32 1, i32 1048575}
diff --git a/clang/test/CodeGenCXX/profile-remap.cpp b/clang/test/CodeGenCXX/profile-remap.cpp
@@ -24,8 +24,8 @@ namespace Foo {
 //
 // FIXME: Laplace's rule of succession is applied to sample profiles...
 // CHECK-SAMPLES-DAG: [[FUNC_ENTRY]] = !{!"function_entry_count", i64 1}
-// CHECK-SAMPLES-DAG: [[BR_WEIGHTS]] = !{!"branch_weights", i64 11, i64 91}
+// CHECK-SAMPLES-DAG: [[BR_WEIGHTS]] = !{!"branch_weights", i32 11, i32 91}
 //
 // ... but not to instruction profiles.
 // CHECK-INSTR-DAG: [[FUNC_ENTRY]] = !{!"function_entry_count", i64 100}
-// CHECK-INSTR-DAG: [[BR_WEIGHTS]] = !{!"branch_weights", i64 10, i64 90}
+// CHECK-INSTR-DAG: [[BR_WEIGHTS]] = !{!"branch_weights", i32 10, i32 90}
diff --git a/clang/test/CodeGenCXX/static-initializer-branch-weights.cpp b/clang/test/CodeGenCXX/static-initializer-branch-weights.cpp
@@ -122,5 +122,5 @@ void use_b() {
 // CHECK: icmp eq i8 {{.*}}, 0
 // CHECK: br i1 {{.*}}, !prof ![[WEIGHTS_THREAD_LOCAL]]
 
-// CHECK-DAG: ![[WEIGHTS_THREAD_LOCAL]] = !{!"branch_weights", i64 1, i64 1023}
-// CHECK-DAG: ![[WEIGHTS_LOCAL]] = !{!"branch_weights", i64 1, i64 1048575}
+// CHECK-DAG: ![[WEIGHTS_THREAD_LOCAL]] = !{!"branch_weights", i32 1, i32 1023}
+// CHECK-DAG: ![[WEIGHTS_LOCAL]] = !{!"branch_weights", i32 1, i32 1048575}
diff --git a/clang/test/Driver/aarch64-security-options.c b/clang/test/Driver/aarch64-security-options.c
@@ -9,15 +9,6 @@
 // RUN: %clang -target aarch64--none-eabi -c %s -### -msign-return-address=all                              2>&1 | \
 // RUN: FileCheck %s --check-prefix=RA-ALL      --check-prefix=KEY-A --check-prefix=BTE-OFF
 
-// Check that the -msign-return-address= option can also accept the signing key
-// to use.
-
-// RUN: %clang -target aarch64--none-eabi -c %s -### -msign-return-address=non-leaf                   2>&1 | \
-// RUN: FileCheck %s --check-prefix=RA-NON-LEAF --check-prefix=KEY-B --check-prefix=BTE-OFF
-
-// RUN: %clang -target aarch64--none-eabi -c %s -### -msign-return-address=all                        2>&1 | \
-// RUN: FileCheck %s --check-prefix=RA-ALL      --check-prefix=KEY-B --check-prefix=BTE-OFF
-
 // -mbranch-protection with standard
 // RUN: %clang -target aarch64--none-eabi -c %s -### -mbranch-protection=standard                                2>&1 | \
 // RUN: FileCheck %s --check-prefix=RA-NON-LEAF --check-prefix=KEY-A --check-prefix=BTE-ON

diff --git a/clang/test/Driver/arch-specific-libdir-rpath.c b/clang/test/Driver/arch-specific-libdir-rpath.c
@@ -84,7 +84,7 @@
 // NO-RPATH-X86_64-NOT:   "-rpath" "[[RESDIR]]{{(/|\\\\)lib(/|\\\\)linux(/|\\\\)x86_64}}"
 //
 // LIBPATH-AARCH64: -L[[RESDIR]]{{(/|\\\\)lib(/|\\\\)linux(/|\\\\)aarch64}}
-// RPATH-AAARCH4:   "-rpath" "[[RESDIR]]{{(/|\\\\)lib(/|\\\\)linux(/|\\\\)aarch64}}"
+// RPATH-AARCH64:   "-rpath" "[[RESDIR]]{{(/|\\\\)lib(/|\\\\)linux(/|\\\\)aarch64}}"
 //
 // NO-LIBPATH-NOT: "-L{{[^"]*Inputs(/|\\\\)resource_dir}}"
 // NO-RPATH-NOT:   "-rpath" {{.*(/|\\\\)Inputs(/|\\\\)resource_dir}}
diff --git a/clang/test/Driver/arm-execute-only.c b/clang/test/Driver/arm-execute-only.c
@@ -11,7 +11,7 @@
 // CHECK-EXECUTE-ONLY-LONG-CALLS: error: option '-mexecute-only' cannot be specified with '-mlong-calls'
 
 // RUN: %clang -target armv7m-eabi -x assembler -mexecute-only %s -c -### 2>&1 \
-// RUN:    | FileCheck %s -check-prefix CHECK-NO-EXECUTE-ONLY -check-prefix CHECK-NO-EXECUTE-ONLY-ASM
+// RUN:    | FileCheck %s --check-prefix=CHECK-NO-EXECUTE-ONLY-ASM
 // CHECK-NO-EXECUTE-ONLY-ASM: warning: argument unused during compilation: '-mexecute-only'
 
 // -mpure-code flag for GCC compatibility

diff --git a/clang/test/Driver/cl-denorms-are-zero.cl b/clang/test/Driver/cl-denorms-are-zero.cl
@@ -18,6 +18,7 @@
 // RUN: %clang -### -target amdgcn--amdhsa -nogpulib -c %s 2>&1 | FileCheck -check-prefixes=AMDGCN,AMDGCN-DENORM %s
 // RUN: %clang -### -cl-denorms-are-zero -o - -target amdgcn--amdhsa -nogpulib -c %s 2>&1 | FileCheck -check-prefixes=AMDGCN,AMDGCN-FLUSH %s
 
+// AMDGCN: "-triple" "amdgcn-unknown-amdhsa"
 // AMDGCN-FLUSH: "-fdenormal-fp-math-f32=preserve-sign,preserve-sign"
 
 // This should be omitted and default to ieee

diff --git a/clang/test/Driver/fopenmp.c b/clang/test/Driver/fopenmp.c
@@ -114,7 +114,7 @@
 // CHECK-LD-STATIC-GOMP: "{{.*}}ld{{(.exe)?}}"
 // CHECK-LD-STATIC-GOMP: "-Bstatic" "-lgomp" "-Bdynamic"
 // CHECK-LD-STATIC-GOMP-RT: "-lrt"
-// CHECK-LD-STATIC-NO-GOMP-RT-NOT: "-lrt"
+// CHECK-LD-STATIC-GOMP-NO-RT-NOT: "-lrt"
 //
 // CHECK-LD-STATIC-IOMP5: "{{.*}}ld{{(.exe)?}}"
 // CHECK-LD-STATIC-IOMP5: "-Bstatic" "-liomp5" "-Bdynamic"

diff --git a/clang/test/Driver/fsanitize-blacklist.c b/clang/test/Driver/fsanitize-blacklist.c
@@ -30,7 +30,7 @@
 // CHECK-DEFAULT-UBSAN-BLACKLIST: -fsanitize-system-blacklist={{.*}}ubsan_blacklist.txt
 
 // Check that combining ubsan and another sanitizer results in both blacklists being used.
-// RUN: %clang -target x86_64-linux-gnu -fsanitize=undefined,address -resource-dir=%S/Inputs/resource_dir %s -### 2>&1 | FileCheck %s --check-prefix=CHECK-DEFAULT-UBSAN-BLACKLIST --check-prefix=CHECK-DEFAULT-ASAN-BLACKLIST --implicit-check-not=fdepfile-entry --implicit-check-not=-fsanitize-blacklist=
+// RUN: %clang -target x86_64-linux-gnu -fsanitize=undefined,address -resource-dir=%S/Inputs/resource_dir %s -### 2>&1 | FileCheck %s --check-prefix=CHECK-DEFAULT-UBSAN-BLACKLIST --check-prefix=CHECK-DEFAULT-BLACKLIST-ASAN --implicit-check-not=fdepfile-entry --implicit-check-not=-fsanitize-blacklist=
 
 // Ignore -fsanitize-blacklist flag if there is no -fsanitize flag.
 // RUN: %clang -target x86_64-linux-gnu -fsanitize-blacklist=%t.good %s -### 2>&1 | FileCheck %s --check-prefix=CHECK-NO-SANITIZE --check-prefix=DELIMITERS

diff --git a/clang/test/Driver/fuzzer.c b/clang/test/Driver/fuzzer.c
@@ -1,12 +1,12 @@
 // Test flags inserted by -fsanitize=fuzzer.
 
-// RUN: %clang -fsanitize=fuzzer %s -target x86_64-apple-darwin14 -### 2>&1 | FileCheck --check-prefixes=CHECK-FUZZER-LIB,CHECK-COVERAGE-FLAGS %s
+// RUN: %clang -fsanitize=fuzzer %s -target x86_64-apple-darwin14 -### 2>&1 | FileCheck --check-prefixes=CHECK-FUZZER-LIB,CHECK-COVERAGE %s
 //
-// CHECK-FUZZER-LIB: libclang_rt.fuzzer
-// CHECK-COVERAGE: -fsanitize-coverage-inline-8bit-counters
-// CHECK-COVERAGE-SAME: -fsanitize-coverage-indirect-calls
+// CHECK-COVERAGE:      -fsanitize-coverage-indirect-calls
 // CHECK-COVERAGE-SAME: -fsanitize-coverage-trace-cmp
+// CHECK-COVERAGE-SAME: -fsanitize-coverage-inline-8bit-counters
 // CHECK-COVERAGE-SAME: -fsanitize-coverage-pc-table
+// CHECK-FUZZER-LIB: libclang_rt.fuzzer
 
 // RUN: %clang -fsanitize=fuzzer -target i386-unknown-linux -stdlib=platform %s -### 2>&1 | FileCheck --check-prefixes=CHECK-LIBCXX-LINUX %s
 //

diff --git a/clang/test/Driver/hip-phases.hip b/clang/test/Driver/hip-phases.hip
@@ -98,7 +98,7 @@
 //
 // RUN: %clang -x hip -target x86_64-unknown-linux-gnu -ccc-print-phases \
 // RUN: --cuda-gpu-arch=gfx803 --cuda-gpu-arch=gfx900 %s -fgpu-rdc 2>&1 \
-// RUN: | FileCheck -check-prefixes=RDC2,CL2,RCL2 %s
+// RUN: | FileCheck -check-prefixes=RDC2,RCL2 %s
 
 // RUN: %clang -x hip -target x86_64-unknown-linux-gnu -ccc-print-phases \
 // RUN: --cuda-gpu-arch=gfx803 --cuda-gpu-arch=gfx900 %s -fgpu-rdc -c 2>&1 \

diff --git a/clang/test/Driver/rocm-detect.cl b/clang/test/Driver/rocm-detect.cl
@@ -13,9 +13,10 @@
 
 // RUN: %clang -### -v -target amdgcn-amd-amdhsa -mcpu=gfx902 -nogpulib \
 // RUN:   --rocm-path=%S/Inputs/rocm %s 2>&1 \
-// RUN:   | FileCheck -check-prefixes=COMMON,GFX902,NODEFAULTLIBS %s
-
+// RUN:   | FileCheck -check-prefixes=COMMON,NODEFAULTLIBS %s
 
 // GFX902-DEFAULTLIBS: error: cannot find ROCm device library for gfx902. Provide its path via --rocm-path or --rocm-device-lib-path, or pass -nogpulib to build without ROCm device library
 
 // NODEFAULTLIBS-NOT: error: cannot find
+
+// COMMON: "-triple" "amdgcn-amd-amdhsa"
diff --git a/clang/test/Driver/rocm-detect.hip b/clang/test/Driver/rocm-detect.hip
@@ -19,9 +19,11 @@
 
 // RUN: %clang -### -v -target x86_64-linux-gnu --cuda-gpu-arch=gfx902 -nogpulib \
 // RUN:   --rocm-path=%S/Inputs/rocm %s 2>&1 \
-// RUN:   | FileCheck -check-prefixes=COMMON,GFX902,NODEFAULTLIBS %s
+// RUN:   | FileCheck -check-prefixes=COMMON,NODEFAULTLIBS %s
 
 
 // GFX902-DEFAULTLIBS: error: cannot find ROCm device library for gfx902. Provide its path via --rocm-path or --rocm-device-lib-path, or pass -nogpulib to build without ROCm device library
 
 // NODEFAULTLIBS-NOT: error: cannot find
+
+// COMMON: "-triple" "amdgcn-amd-amdhsa"
diff --git a/clang/test/Profile/c-captured.c b/clang/test/Profile/c-captured.c
@@ -47,11 +47,11 @@ void debug_captured() {
   if (x) {} // This is DC2. Checked above.
 }
 
-// PGOUSE-DAG: ![[DC1]] = !{!"branch_weights", i64 2, i64 1}
-// PGOUSE-DAG: ![[DC2]] = !{!"branch_weights", i64 2, i64 1}
-// PGOUSE-DAG: ![[CS1]] = !{!"branch_weights", i64 2, i64 1}
-// PGOUSE-DAG: ![[C11]] = !{!"branch_weights", i64 11, i64 2}
-// PGOUSE-DAG: ![[C12]] = !{!"branch_weights", i64 2, i64 1}
+// PGOUSE-DAG: ![[DC1]] = !{!"branch_weights", i32 2, i32 1}
+// PGOUSE-DAG: ![[DC2]] = !{!"branch_weights", i32 2, i32 1}
+// PGOUSE-DAG: ![[CS1]] = !{!"branch_weights", i32 2, i32 1}
+// PGOUSE-DAG: ![[C11]] = !{!"branch_weights", i32 11, i32 2}
+// PGOUSE-DAG: ![[C12]] = !{!"branch_weights", i32 2, i32 1}
 
 int main(int argc, const char *argv[]) {
   debug_captured();

diff --git a/clang/test/Profile/c-counter-overflows.c b/clang/test/Profile/c-counter-overflows.c
@@ -0,0 +1,49 @@
+// Test that big branch weights get scaled down to 32-bits, rather than just
+// truncated.
+
+// RUN: llvm-profdata merge %S/Inputs/c-counter-overflows.proftext -o %t.profdata
+// RUN: %clang_cc1 -triple x86_64-apple-macosx10.9 -main-file-name c-counter-overflows.c %s -o - -emit-llvm -fprofile-instrument-use-path=%t.profdata | FileCheck %s
+
+typedef unsigned long long uint64_t;
+
+int main(int argc, const char *argv[]) {
+  // Need counts higher than 32-bits.
+  // CHECK: br {{.*}} !prof ![[FOR:[0-9]+]]
+  // max   = 0xffffffff0
+  // scale = 0xffffffff0 / 0xffffffff + 1 = 17
+  // loop-body: 0xffffffff0 / 17 + 1 = 0xf0f0f0f0 + 1 = 4042322161 => -252645135
+  // loop-exit: 0x000000001 / 17 + 1 = 0x00000000 + 1 =          1 =>          1
+  for (uint64_t I = 0; I < 0xffffffff0; ++I) {
+    // max   = 0xffffffff * 15 = 0xefffffff1
+    // scale = 0xefffffff1 / 0xffffffff + 1 = 16
+    // CHECK: br {{.*}} !prof ![[IF:[0-9]+]]
+    if (I & 0xf) {
+      // 0xefffffff1 / 16 + 1 = 0xefffffff + 1 = 4026531840 => -268435456
+    } else {
+      // 0x0ffffffff / 16 + 1 = 0x0fffffff + 1 =  268435456 =>  268435456
+    }
+
+    // max   = 0xffffffff * 5 = 0x4fffffffb
+    // scale = 0x4fffffffb / 0xffffffff + 1 = 6
+    // CHECK: ], !prof ![[SWITCH:[0-9]+]]
+    switch ((I & 0xf) / 5) {
+    case 0:
+      // 0x4fffffffb / 6 = 0xd5555554 + 1 = 3579139413 => -715827883
+      break;
+    case 1:
+      // 0x4fffffffb / 6 = 0xd5555554 + 1 = 3579139413 => -715827883
+      break;
+    case 2:
+      // 0x4fffffffb / 6 = 0xd5555554 + 1 = 3579139413 => -715827883
+      break;
+    default:
+      // 0x0ffffffff / 6 = 0x2aaaaaaa + 1 =  715827883 =>  715827883
+      break;
+    }
+  }
+  return 0;
+}
+
+// CHECK-DAG: ![[FOR]] = !{!"branch_weights", i32 -252645135, i32 1}
+// CHECK-DAG: ![[IF]]  = !{!"branch_weights", i32 -268435456, i32 268435456}
+// CHECK-DAG: ![[SWITCH]] = !{!"branch_weights", i32 715827883, i32 -715827883, i32 -715827883, i32 -715827883}