Recommit "Enable MachineOutliner by default under -Oz for AArch64"

Fixed the ASAN failure from before in r338148, so recommiting.

This patch enables the MachineOutliner by default in AArch64 under -Oz.

The MachineOutliner offers around a 4.5% improvement on the current -Oz code
size improvements.

We have done work into improving the debuggability of outlined code, so that
users of -Oz won't be surprised by the optimization. We have also been executing
the LLVM test suite and common external tests such as the SPEC suites
continuously with no issue. The outliner has a low compile-time overhead of
roughly 1%. At this point, the outliner would be a really good addition to the
-Oz pass pipeline!

llvm-svn: 338160

diff --git a/llvm/lib/Target/AArch64/AArch64InstrInfo.cpp b/llvm/lib/Target/AArch64/AArch64InstrInfo.cpp
index 230480c..b571c42 100644 (file)
--- a/llvm/lib/Target/AArch64/AArch64InstrInfo.cpp
+++ b/llvm/lib/Target/AArch64/AArch64InstrInfo.cpp
@@ -5482,3 +5482,8 @@ MachineBasicBlock::iterator AArch64InstrInfo::insertOutlinedCall(
 
   return CallPt;
 }
+
+bool AArch64InstrInfo::shouldOutlineFromFunctionByDefault(
+  MachineFunction &MF) const {
+  return MF.getFunction().optForMinSize();
+}

diff --git a/llvm/lib/Target/AArch64/AArch64InstrInfo.h b/llvm/lib/Target/AArch64/AArch64InstrInfo.h
index 0e5953f..585cbd4 100644 (file)
--- a/llvm/lib/Target/AArch64/AArch64InstrInfo.h
+++ b/llvm/lib/Target/AArch64/AArch64InstrInfo.h
@@ -249,6 +249,7 @@ public:
   insertOutlinedCall(Module &M, MachineBasicBlock &MBB,
                      MachineBasicBlock::iterator &It, MachineFunction &MF,
                      const outliner::Candidate &C) const override;
+  bool shouldOutlineFromFunctionByDefault(MachineFunction &MF) const override;
   /// Returns true if the instruction sets to an immediate value that can be
   /// executed more efficiently.
   bool isExynosResetFast(const MachineInstr &MI) const;

diff --git a/llvm/lib/Target/AArch64/AArch64TargetMachine.cpp b/llvm/lib/Target/AArch64/AArch64TargetMachine.cpp
index 01a997e..120d713 100644 (file)
--- a/llvm/lib/Target/AArch64/AArch64TargetMachine.cpp
+++ b/llvm/lib/Target/AArch64/AArch64TargetMachine.cpp
@@ -255,6 +255,9 @@ AArch64TargetMachine::AArch64TargetMachine(const Target &T, const Triple &TT,
 
   // AArch64 supports the MachineOutliner.
   setMachineOutliner(true);
+
+  // AArch64 supports default outlining behaviour.
+  setSupportsDefaultOutlining(true);
 }
 
 AArch64TargetMachine::~AArch64TargetMachine() = default;

diff --git a/llvm/test/CodeGen/AArch64/O3-pipeline.ll b/llvm/test/CodeGen/AArch64/O3-pipeline.ll
index e482682..f0c7e4e 100644 (file)
--- a/llvm/test/CodeGen/AArch64/O3-pipeline.ll
+++ b/llvm/test/CodeGen/AArch64/O3-pipeline.ll
@@ -154,6 +154,8 @@
 ; CHECK-NEXT:       Insert fentry calls
 ; CHECK-NEXT:       Insert XRay ops
 ; CHECK-NEXT:       Implement the 'patchable-function' attribute
+; CHECK-NEXT:     Machine Outliner
+; CHECK-NEXT:     FunctionPass Manager
 ; CHECK-NEXT:       Lazy Machine Block Frequency Analysis
 ; CHECK-NEXT:       Machine Optimization Remark Emitter
 ; CHECK-NEXT:       AArch64 Assembly Printer

diff --git a/llvm/test/CodeGen/AArch64/arm64-memset-to-bzero.ll b/llvm/test/CodeGen/AArch64/arm64-memset-to-bzero.ll
index 0afe1c0..ab819a4 100644 (file)
--- a/llvm/test/CodeGen/AArch64/arm64-memset-to-bzero.ll
+++ b/llvm/test/CodeGen/AArch64/arm64-memset-to-bzero.ll
@@ -1,6 +1,6 @@
-; RUN: llc %s -mtriple=arm64-apple-darwin -o - | \
-; RUN:   FileCheck --check-prefixes=CHECK,CHECK-DARWIN %s
-; RUN: llc %s -mtriple=arm64-linux-gnu -o - | \
+; RUN: llc %s -enable-machine-outliner=never -mtriple=arm64-apple-darwin -o - \
+; RUN: | FileCheck --check-prefixes=CHECK,CHECK-DARWIN %s
+; RUN: llc %s -enable-machine-outliner=never -mtriple=arm64-linux-gnu -o - | \
 ; RUN:   FileCheck --check-prefixes=CHECK,CHECK-LINUX %s
 ; <rdar://problem/14199482> ARM64: Calls to bzero() replaced with calls to memset()
 

diff --git a/llvm/test/CodeGen/AArch64/arm64-opt-remarks-lazy-bfi.ll b/llvm/test/CodeGen/AArch64/arm64-opt-remarks-lazy-bfi.ll
index 7efb4bf..452c822 100644 (file)
--- a/llvm/test/CodeGen/AArch64/arm64-opt-remarks-lazy-bfi.ll
+++ b/llvm/test/CodeGen/AArch64/arm64-opt-remarks-lazy-bfi.ll
@@ -26,8 +26,8 @@
 ; requested.  (This hard-codes the previous pass to the Assembly Printer,
 ; please adjust accordingly.)
 
-; HOTNESS:      Executing Pass 'Implement the 'patchable-function' attribute'
-; HOTNESS-NEXT:  Freeing Pass 'Implement the 'patchable-function' attribute'
+; HOTNESS:      Freeing Pass 'Machine Outliner'
+; HOTNESS-NEXT:  Executing Pass 'Function Pass Manager'
 ; HOTNESS-NEXT: Executing Pass 'Lazy Machine Block Frequency Analysis'
 ; HOTNESS-NEXT: Executing Pass 'Machine Optimization Remark Emitter'
 ; HOTNESS-NEXT: Building MachineBlockFrequencyInfo on the fly
@@ -41,8 +41,8 @@
 ; HOTNESS: arm64-summary-remarks.ll:5:0: 1 instructions in function (hotness: 33)
 
 
-; NO_HOTNESS:      Executing Pass 'Implement the 'patchable-function' attribute'
-; NO_HOTNESS-NEXT:  Freeing Pass 'Implement the 'patchable-function' attribute'
+; NO_HOTNESS:      Freeing Pass 'Machine Outliner'
+; NO_HOTNESS-NEXT:  Executing Pass 'Function Pass Manager'
 ; NO_HOTNESS-NEXT: Executing Pass 'Lazy Machine Block Frequency Analysis'
 ; NO_HOTNESS-NEXT: Executing Pass 'Machine Optimization Remark Emitter'
 ; NO_HOTNESS-NEXT: Executing Pass 'AArch64 Assembly Printer'

diff --git a/llvm/test/CodeGen/AArch64/cond-sel.ll b/llvm/test/CodeGen/AArch64/cond-sel.ll
index b39cea1..691cbcf 100644 (file)
--- a/llvm/test/CodeGen/AArch64/cond-sel.ll
+++ b/llvm/test/CodeGen/AArch64/cond-sel.ll
@@ -1,5 +1,5 @@
-; RUN: llc -verify-machineinstrs < %s -mtriple=aarch64-none-linux-gnu -mcpu=cyclone | FileCheck %s
-; RUN: llc -verify-machineinstrs < %s -mtriple=aarch64-none-linux-gnu -mattr=-fp-armv8 | FileCheck --check-prefix=CHECK-NOFP %s
+; RUN: llc -enable-machine-outliner=never -verify-machineinstrs < %s -mtriple=aarch64-none-linux-gnu -mcpu=cyclone | FileCheck %s
+; RUN: llc -enable-machine-outliner=never -verify-machineinstrs < %s -mtriple=aarch64-none-linux-gnu -mattr=-fp-armv8 | FileCheck --check-prefix=CHECK-NOFP %s
 
 @var32 = global i32 0
 @var64 = global i64 0

diff --git a/llvm/test/CodeGen/AArch64/machine-outliner-default.mir b/llvm/test/CodeGen/AArch64/machine-outliner-default.mir
new file mode 100644 (file)
index 0000000..698a2fc
--- /dev/null
+++ b/llvm/test/CodeGen/AArch64/machine-outliner-default.mir
@@ -0,0 +1,71 @@
+# RUN: llc -mtriple=aarch64--- -run-pass=machine-outliner \
+# RUN: -verify-machineinstrs %s -o - | FileCheck %s
+
+--- |
+  define void @outline_1() #0 { ret void }
+  define void @outline_2() #0 { ret void }
+  define void @outline_3() #0 { ret void }
+  define void @dont_outline() #1 { ret void }
+
+  attributes #0 = { noredzone minsize optsize }
+  attributes #1 = { noredzone }
+...
+---
+
+name:           outline_1
+tracksRegLiveness: true
+body:             |
+  bb.0:
+    ; CHECK-LABEL: bb.0:
+    ; CHECK: OUTLINED
+    liveins: $w8, $wzr
+    $w8 = ORRWri $wzr, 1
+    $w8 = ORRWri $wzr, 2
+    $w8 = ORRWri $wzr, 3
+    $w8 = ORRWri $wzr, 4
+    RET undef $lr
+...
+---
+
+name:           outline_2
+tracksRegLiveness: true
+body:             |
+  bb.0:
+    ; CHECK-LABEL: bb.0:
+    ; CHECK: OUTLINED
+    liveins: $w8, $wzr
+    $w8 = ORRWri $wzr, 1
+    $w8 = ORRWri $wzr, 2
+    $w8 = ORRWri $wzr, 3
+    $w8 = ORRWri $wzr, 4
+    RET undef $lr
+...
+---
+
+name:           outline_3
+tracksRegLiveness: true
+body:             |
+  bb.0:
+    ; CHECK-LABEL: bb.0:
+    ; CHECK: OUTLINED
+    liveins: $w8, $wzr
+    $w8 = ORRWri $wzr, 1
+    $w8 = ORRWri $wzr, 2
+    $w8 = ORRWri $wzr, 3
+    $w8 = ORRWri $wzr, 4
+    RET undef $lr
+...
+---
+
+name:           dont_outline
+tracksRegLiveness: true
+body:             |
+  bb.0:
+    ; CHECK-LABEL: bb.0:
+    ; CHECK-NOT: BL
+    liveins: $w8, $wzr
+    $w8 = ORRWri $wzr, 1
+    $w8 = ORRWri $wzr, 2
+    $w8 = ORRWri $wzr, 3
+    $w8 = ORRWri $wzr, 4
+    RET undef $lr

diff --git a/llvm/test/CodeGen/AArch64/machine-outliner-flags.ll b/llvm/test/CodeGen/AArch64/machine-outliner-flags.ll
index e00a190..c435093 100644 (file)
--- a/llvm/test/CodeGen/AArch64/machine-outliner-flags.ll
+++ b/llvm/test/CodeGen/AArch64/machine-outliner-flags.ll
@@ -14,7 +14,7 @@
 ; RUN: | FileCheck %s -check-prefix=NEVER
 
 ; RUN: llc %s -debug-pass=Structure -verify-machineinstrs \
-; RUN: -mtriple arm64---- -o /dev/null 2>&1 \
+; RUN: --debug-only=machine-outliner -mtriple arm64---- -o /dev/null 2>&1 \
 ; RUN: | FileCheck %s -check-prefix=NOT-ADDED
 
 ; RUN: llc %s -O=0 -debug-pass=Structure -verify-machineinstrs \
@@ -27,10 +27,11 @@
 ; Cases where it should be added:
 ;  * -enable-machine-outliner
 ;  * -enable-machine-outliner=always
+;  * -enable-machine-outliner is not passed (AArch64 supports
+;     target-default outlining)
 ;
 ; Cases where it should not be added:
 ;  * -O0 or equivalent
-;  * -enable-machine-outliner is not passed
 ;  * -enable-machine-outliner=never is passed
 
 ; ALWAYS: Machine Outliner
@@ -38,7 +39,8 @@
 ; ENABLE: Machine Outliner
 ; ENABLE: Machine Outliner: Running on all functions
 ; NEVER-NOT: Machine Outliner
-; NOT-ADDED-NOT: Machine Outliner
+; NOT-ADDED: Machine Outliner
+; NOT-ADDED: Machine Outliner: Running on target-default functions
 ; OPTNONE-NOT: Machine Outliner
 
 define void @foo() {

diff --git a/llvm/test/CodeGen/AArch64/max-jump-table.ll b/llvm/test/CodeGen/AArch64/max-jump-table.ll
index 612eba8..44dde7b 100644 (file)
--- a/llvm/test/CodeGen/AArch64/max-jump-table.ll
+++ b/llvm/test/CodeGen/AArch64/max-jump-table.ll
@@ -89,6 +89,7 @@ entry:
 ; CHECKM1-NOT: %jump-table.1
 ; CHECKM3-NEXT: %jump-table.0:  %bb.1 %bb.2 %bb.3 %bb.4 %bb.7 %bb.7 %bb.7 %bb.7 %bb.7 %bb.7 %bb.7 %bb.7 %bb.7 %bb.5 %bb.6{{$}}
 ; CHECKM3-NOT: %jump-table.1
+; CHECK-DAG: End machine code for function jt2.
 
 bb1: tail call void @ext(i32 1) br label %return
 bb2: tail call void @ext(i32 2) br label %return