Merge branch 'main' into libcxx-overriden-function-detection

Author: ldionne

.ci/monolithic-linux.sh

@@ -34,8 +34,11 @@ function at-exit {
 
   # If building fails there will be no results files.
   shopt -s nullglob
-  python3 "${MONOREPO_ROOT}"/.ci/generate_test_report.py ":linux: Linux x64 Test Results" \
-    "linux-x64-test-results" "${BUILD_DIR}"/test-results.*.xml
+  if command -v buildkite-agent 2>&1 >/dev/null
+  then
+    python3 "${MONOREPO_ROOT}"/.ci/generate_test_report.py ":linux: Linux x64 Test Results" \
+      "linux-x64-test-results" "${BUILD_DIR}"/test-results.*.xml
+  fi
 }
 trap at-exit EXIT
 

.ci/monolithic-windows.sh

@@ -33,8 +33,11 @@ function at-exit {
 
   # If building fails there will be no results files.
   shopt -s nullglob
-  python "${MONOREPO_ROOT}"/.ci/generate_test_report.py ":windows: Windows x64 Test Results" \
-    "windows-x64-test-results" "${BUILD_DIR}"/test-results.*.xml
+  if command -v buildkite-agent 2>&1 >/dev/null
+  then
+    python "${MONOREPO_ROOT}"/.ci/generate_test_report.py ":windows: Windows x64 Test Results" \
+      "windows-x64-test-results" "${BUILD_DIR}"/test-results.*.xml
+  fi
 }
 trap at-exit EXIT
 

bolt/include/bolt/Passes/ADRRelaxationPass.h

@@ -25,7 +25,8 @@ namespace bolt {
 
 class ADRRelaxationPass : public BinaryFunctionPass {
 public:
-  explicit ADRRelaxationPass() : BinaryFunctionPass(false) {}
+  explicit ADRRelaxationPass(const cl::opt<bool> &PrintPass)
+      : BinaryFunctionPass(PrintPass) {}
 
   const char *getName() const override { return "adr-relaxation"; }
 

bolt/lib/Passes/LongJmp.cpp

@@ -138,9 +138,9 @@ BinaryBasicBlock *LongJmpPass::lookupStubFromGroup(
           const std::pair<uint64_t, BinaryBasicBlock *> &RHS) {
         return LHS.first < RHS.first;
       });
-  if (Cand == Candidates.end())
-    return nullptr;
-  if (Cand != Candidates.begin()) {
+  if (Cand == Candidates.end()) {
+    Cand = std::prev(Cand);
+  } else if (Cand != Candidates.begin()) {
     const StubTy *LeftCand = std::prev(Cand);
     if (Cand->first - DotAddress > DotAddress - LeftCand->first)
       Cand = LeftCand;

bolt/lib/Rewrite/BinaryPassManager.cpp

@@ -126,6 +126,11 @@ static cl::opt<bool> PrintJTFootprintReduction(
     cl::desc("print function after jt-footprint-reduction pass"), cl::Hidden,
     cl::cat(BoltOptCategory));
 
+static cl::opt<bool>
+    PrintAdrRelaxation("print-adr-relaxation",
+                       cl::desc("print functions after ADR Relaxation pass"),
+                       cl::Hidden, cl::cat(BoltOptCategory));
+
 static cl::opt<bool>
     PrintLongJmp("print-longjmp",
                  cl::desc("print functions after longjmp pass"), cl::Hidden,
@@ -493,7 +498,8 @@ Error BinaryFunctionPassManager::runAllPasses(BinaryContext &BC) {
   Manager.registerPass(std::make_unique<ReorderData>());
 
   if (BC.isAArch64()) {
-    Manager.registerPass(std::make_unique<ADRRelaxationPass>());
+    Manager.registerPass(
+        std::make_unique<ADRRelaxationPass>(PrintAdrRelaxation));
 
     // Tighten branches according to offset differences between branch and
     // targets. No extra instructions after this pass, otherwise we may have

bolt/test/AArch64/long-jmp-one-stub.s

@@ -0,0 +1,32 @@
+## This test verifies that no unnecessary stubs are inserted when each DotAddress increases during a lookup.
+
+# REQUIRES: system-linux, asserts
+
+# RUN: llvm-mc -filetype=obj -triple aarch64-unknown-unknown %s -o %t.o
+# RUN: %clang %cflags -O0 %t.o -o %t.exe -Wl,-q
+# RUN: link_fdata %s %t.o %t.fdata
+# RUN: llvm-bolt %t.exe -o %t.bolt  \
+# RUN:   --data %t.fdata  | FileCheck %s
+
+# CHECK: BOLT-INFO: Inserted 1 stubs in the hot area and 0 stubs in the cold area. 
+
+  .section .text
+  .global _start
+  .global far_away_func
+
+  .align 4
+  .global _start
+  .type _start, %function
+_start:
+# FDATA: 0 [unknown] 0 1 _start 0 0 100
+    bl far_away_func
+    bl far_away_func
+    ret  
+  .space 0x8000000
+  .global far_away_func
+  .type far_away_func, %function
+far_away_func:
+    add x0, x0, #1
+    ret
+
+.reloc 0, R_AARCH64_NONE

bolt/test/X86/linux-static-keys.s

@@ -35,13 +35,13 @@ _start:
 .L0:
   jmp L1
 # CHECK:      jit
-# CHECK-SAME: # ID: 1 {{.*}} # Likely: 0 # InitValue: 1
+# CHECK-SAME: # ID: 1 {{.*}} # Likely: 1 # InitValue: 0
   nop
 L1:
   .nops 5
-  jmp .L0
 # CHECK:      jit
-# CHECK-SAME: # ID: 2 {{.*}} # Likely: 1 # InitValue: 1
+# CHECK-SAME: # ID: 2 {{.*}} # Likely: 0 # InitValue: 0
+  jmp .L0
 
 ## Check that a branch profile associated with a NOP is handled properly when
 ## dynamic branch is created.
@@ -67,18 +67,24 @@ foo:
   .type __start___jump_table, %object
 __start___jump_table:
 
-  .long .L0 - . # Jump address
-  .long L1 - . # Target address
-  .quad 1       # Key address
+  .long .L0 - .                   # Jump address
+  .long L1 - .                    # Target address
+  .quad fake_static_key + 1 - .   # Key address; LSB = 1 : likely
 
-  .long L1 - . # Jump address
-  .long L2 - . # Target address
-  .quad 0       # Key address
+  .long L1 - .                    # Jump address
+  .long L2 - .                    # Target address
+  .quad fake_static_key -.        # Key address; LSB = 0 : unlikely
 
   .globl __stop___jump_table
   .type __stop___jump_table, %object
 __stop___jump_table:
 
+## Staic keys (we just use the label ignoring the format of the keys).
+  .data
+  .align 8
+fake_static_key:
+  .quad 0
+
 ## Fake Linux Kernel sections.
   .section __ksymtab,"a",@progbits
   .section __ksymtab_gpl,"a",@progbits

bolt/test/merge-fdata-lbr-mode.test

@@ -7,9 +7,9 @@
 # RUN: FileCheck %s --input-file %t/merged.fdata
 
 # CHECK-NOT: no_lbr
-# CHECK: main 2
+# CHECK: 1 main 0 1 main 2 1 3
 
 #--- a.fdata
-main 1
+1 main 0 1 main 2 0 1
 #--- b.fdata
-main 1
+1 main 0 1 main 2 1 2

bolt/test/merge-fdata-mixed-bat-no-lbr.test

@@ -5,7 +5,7 @@
 # RUN: split-file %s %t
 # RUN: not merge-fdata %t/a.fdata %t/b.fdata 2>&1 | FileCheck %s
 
-# CHECK: cannot mix profile collected in BOLT and non-BOLT deployments
+# CHECK: cannot mix profile with and without boltedcollection
 
 #--- a.fdata
 boltedcollection

bolt/test/merge-fdata-mixed-mode.test

@@ -6,7 +6,7 @@
 # RUN: split-file %s %t
 # RUN: not merge-fdata %t/a.fdata %t/b.fdata 2>&1 | FileCheck %s
 
-# CHECK: cannot mix 'no_lbr' and 'lbr' profiles.
+# CHECK: cannot mix profile with and without no_lbr
 
 #--- a.fdata
 no_lbr