[AArch64, ARM] Enable GlobalMerge with -O3 rather than -O1.

The pass used to be enabled by default with CodeGenOpt::Less (-O1).
This is too aggressive, considering the pass indiscriminately merges
all globals together.

Currently, performance doesn't always improve, and, on code that uses
few globals (e.g., the odd file- or function- static), more often than
not is degraded by the optimization.  Lengthy discussion can be found
on llvmdev (AArch64-focused;  ARM has similar problems):
  http://lists.cs.uiuc.edu/pipermail/llvmdev/2015-February/082800.html
Also, it makes tooling and debuggers less useful when dealing with
globals and data sections.

GlobalMerge needs to better identify those cases that benefit, and this
will be done separately.  In the meantime, move the pass to run with
-O3 rather than -O1, on both ARM and AArch64.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@233024 91177308-0d34-0410-b5e6-96231b3b80d8

diff --git a/lib/Target/AArch64/AArch64TargetMachine.cpp b/lib/Target/AArch64/AArch64TargetMachine.cpp
index 187d2ce2088ee403a10d63a629e55682831cb9a9..f902f6486564064dc369de967b183711c61c2578 100644 (file)
--- a/lib/Target/AArch64/AArch64TargetMachine.cpp
+++ b/lib/Target/AArch64/AArch64TargetMachine.cpp
@@ -245,7 +245,7 @@ bool AArch64PassConfig::addPreISel() {
   // FIXME: On AArch64, this depends on the type.
   // Basically, the addressable offsets are up to 4095 * Ty.getSizeInBytes().
   // and the offset has to be a multiple of the related size in bytes.
-  if (TM->getOptLevel() != CodeGenOpt::None)
+  if (TM->getOptLevel() == CodeGenOpt::Aggressive)
     addPass(createGlobalMergePass(TM, 4095));
   if (TM->getOptLevel() != CodeGenOpt::None)
     addPass(createAArch64AddressTypePromotionPass());

diff --git a/lib/Target/ARM/ARMTargetMachine.cpp b/lib/Target/ARM/ARMTargetMachine.cpp
index 4586a9f45bf0296176310fd9fc2a145e99c1cbc8..fcb96e08039e703a8fba3d4de9f9360614552ebc 100644 (file)
--- a/lib/Target/ARM/ARMTargetMachine.cpp
+++ b/lib/Target/ARM/ARMTargetMachine.cpp
@@ -327,7 +327,7 @@ void ARMPassConfig::addIRPasses() {
 }
 
 bool ARMPassConfig::addPreISel() {
-  if (TM->getOptLevel() != CodeGenOpt::None)
+  if (TM->getOptLevel() == CodeGenOpt::Aggressive)
     // FIXME: This is using the thumb1 only constant value for
     // maximal global offset for merging globals. We may want
     // to look into using the old value for non-thumb1 code of

diff --git a/test/CodeGen/AArch64/global-merge-1.ll b/test/CodeGen/AArch64/global-merge-1.ll
index 7dc8da1c444b758923720bd93c4a539c11dae546..b40438900ab634dc56d030f6bd7aabc989d5abc8 100644 (file)
--- a/test/CodeGen/AArch64/global-merge-1.ll
+++ b/test/CodeGen/AArch64/global-merge-1.ll
@@ -1,11 +1,11 @@
-; RUN: llc %s -mtriple=aarch64-none-linux-gnu -enable-global-merge -o - | FileCheck %s
-; RUN: llc %s -mtriple=aarch64-none-linux-gnu -enable-global-merge -global-merge-on-external -o - | FileCheck %s
+; RUN: llc %s -mtriple=aarch64-none-linux-gnu -O3 -enable-global-merge -o - | FileCheck %s
+; RUN: llc %s -mtriple=aarch64-none-linux-gnu -O3 -enable-global-merge -global-merge-on-external -o - | FileCheck %s
 
-; RUN: llc %s -mtriple=aarch64-linux-gnuabi -enable-global-merge -o - | FileCheck %s
-; RUN: llc %s -mtriple=aarch64-linux-gnuabi -enable-global-merge -global-merge-on-external -o - | FileCheck %s
+; RUN: llc %s -mtriple=aarch64-linux-gnuabi -O3 -enable-global-merge -o - | FileCheck %s
+; RUN: llc %s -mtriple=aarch64-linux-gnuabi -O3 -enable-global-merge -global-merge-on-external -o - | FileCheck %s
 
-; RUN: llc %s -mtriple=aarch64-apple-ios -enable-global-merge -o - | FileCheck %s --check-prefix=CHECK-APPLE-IOS
-; RUN: llc %s -mtriple=aarch64-apple-ios -enable-global-merge -global-merge-on-external -o - | FileCheck %s --check-prefix=CHECK-APPLE-IOS
+; RUN: llc %s -mtriple=aarch64-apple-ios -O3 -enable-global-merge -o - | FileCheck %s --check-prefix=CHECK-APPLE-IOS
+; RUN: llc %s -mtriple=aarch64-apple-ios -O3 -enable-global-merge -global-merge-on-external -o - | FileCheck %s --check-prefix=CHECK-APPLE-IOS
 
 @m = internal global i32 0, align 4
 @n = internal global i32 0, align 4

diff --git a/test/CodeGen/AArch64/global-merge-2.ll b/test/CodeGen/AArch64/global-merge-2.ll
index 70b700c7e57a0b4952d412e273180458a53282a3..d5967b9383159727176441bf7ef910f935eec97f 100644 (file)
--- a/test/CodeGen/AArch64/global-merge-2.ll
+++ b/test/CodeGen/AArch64/global-merge-2.ll
@@ -1,6 +1,6 @@
-; RUN: llc %s -mtriple=aarch64-none-linux-gnu -enable-global-merge -global-merge-on-external -o - | FileCheck %s
-; RUN: llc %s -mtriple=aarch64-linux-gnuabi -enable-global-merge -global-merge-on-external -o - | FileCheck %s
-; RUN: llc %s -mtriple=aarch64-apple-ios -enable-global-merge -global-merge-on-external -o - | FileCheck %s --check-prefix=CHECK-APPLE-IOS
+; RUN: llc %s -mtriple=aarch64-none-linux-gnu -O3 -enable-global-merge -global-merge-on-external -o - | FileCheck %s
+; RUN: llc %s -mtriple=aarch64-linux-gnuabi -O3 -enable-global-merge -global-merge-on-external -o - | FileCheck %s
+; RUN: llc %s -mtriple=aarch64-apple-ios -O3 -enable-global-merge -global-merge-on-external -o - | FileCheck %s --check-prefix=CHECK-APPLE-IOS
 
 @x = global i32 0, align 4
 @y = global i32 0, align 4

diff --git a/test/CodeGen/AArch64/global-merge-3.ll b/test/CodeGen/AArch64/global-merge-3.ll
index 2364d8de5880ac6f6d0064c3ec71d712b598828f..15035c0032effaac27501d5125413323d69a536a 100644 (file)
--- a/test/CodeGen/AArch64/global-merge-3.ll
+++ b/test/CodeGen/AArch64/global-merge-3.ll
@@ -1,6 +1,6 @@
-; RUN: llc %s -mtriple=aarch64-none-linux-gnu -enable-global-merge -global-merge-on-external -o - | FileCheck %s
-; RUN: llc %s -mtriple=aarch64-linux-gnuabi -enable-global-merge -global-merge-on-external -o - | FileCheck %s
-; RUN: llc %s -mtriple=aarch64-apple-ios -enable-global-merge -global-merge-on-external -o - | FileCheck %s --check-prefix=CHECK-APPLE-IOS
+; RUN: llc %s -mtriple=aarch64-none-linux-gnu -O3 -enable-global-merge -global-merge-on-external -o - | FileCheck %s
+; RUN: llc %s -mtriple=aarch64-linux-gnuabi -O3 -enable-global-merge -global-merge-on-external -o - | FileCheck %s
+; RUN: llc %s -mtriple=aarch64-apple-ios -O3 -enable-global-merge -global-merge-on-external -o - | FileCheck %s --check-prefix=CHECK-APPLE-IOS
 
 @x = global [1000 x i32] zeroinitializer, align 1
 @y = global [1000 x i32] zeroinitializer, align 1

diff --git a/test/CodeGen/AArch64/global-merge-4.ll b/test/CodeGen/AArch64/global-merge-4.ll
index d3ae32d3b5c5282a7a327c9125486603a3e7eec3..8fb7747977cd4669dffc5676f61525b47b3bb2c8 100644 (file)
--- a/test/CodeGen/AArch64/global-merge-4.ll
+++ b/test/CodeGen/AArch64/global-merge-4.ll
@@ -1,4 +1,4 @@
-; RUN: llc %s -mtriple=aarch64-linux-gnuabi -enable-global-merge -o - | FileCheck %s
+; RUN: llc %s -mtriple=aarch64-linux-gnuabi -O3 -enable-global-merge -o - | FileCheck %s
 
 target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-n32:64-S128"
 target triple = "arm64-apple-ios7.0.0"

diff --git a/test/CodeGen/ARM/2010-12-15-elf-lcomm.ll b/test/CodeGen/ARM/2010-12-15-elf-lcomm.ll
index 8db166afec6ef72ced208fdcd14fd365ba729aed..d3c0feeab45a975c1fab5e840fa5679f0557a4cc 100644 (file)
--- a/test/CodeGen/ARM/2010-12-15-elf-lcomm.ll
+++ b/test/CodeGen/ARM/2010-12-15-elf-lcomm.ll
@@ -1,6 +1,6 @@
-; RUN: llc  %s -mtriple=armv7-linux-gnueabi -filetype=obj -o - | \
+; RUN: llc  %s -mtriple=armv7-linux-gnueabi -O3 -filetype=obj -o - | \
 ; RUN:    llvm-readobj -s -t | FileCheck  -check-prefix=OBJ %s
-; RUN: llc  %s -mtriple=armv7-linux-gnueabi -o - | \
+; RUN: llc  %s -mtriple=armv7-linux-gnueabi -O3 -o - | \
 ; RUN:    FileCheck  -check-prefix=ASM %s
 
 

diff --git a/test/CodeGen/ARM/2011-01-19-MergedGlobalDbg.ll b/test/CodeGen/ARM/2011-01-19-MergedGlobalDbg.ll
index b361c13941c19875b561d61ae935df560fa0ce69..7196d2d37bd00dfec3eeac2d329048cd70f2c710 100644 (file)
--- a/test/CodeGen/ARM/2011-01-19-MergedGlobalDbg.ll
+++ b/test/CodeGen/ARM/2011-01-19-MergedGlobalDbg.ll
@@ -1,4 +1,4 @@
-; RUN: llc -filetype=obj < %s | llvm-dwarfdump -debug-dump=info - | FileCheck %s
+; RUN: llc -O3 -filetype=obj < %s | llvm-dwarfdump -debug-dump=info - | FileCheck %s
 
 target datalayout = "e-p:32:32:32-i1:8:32-i8:8:32-i16:16:32-i32:32:32-i64:32:64-f32:32:32-f64:32:64-v64:32:64-v128:32:128-a0:0:32-n32"
 target triple = "thumbv7-apple-darwin10"

diff --git a/test/CodeGen/ARM/2011-06-29-MergeGlobalsAlign.ll b/test/CodeGen/ARM/2011-06-29-MergeGlobalsAlign.ll
index 091d037356dea213e04c886047f441deab035fe3..236dc1c0dbcb560cc953d0b881251a114af94e25 100644 (file)
--- a/test/CodeGen/ARM/2011-06-29-MergeGlobalsAlign.ll
+++ b/test/CodeGen/ARM/2011-06-29-MergeGlobalsAlign.ll
@@ -1,4 +1,4 @@
-; RUN: llc < %s -mtriple=thumbv7-apple-darwin10 | FileCheck %s
+; RUN: llc < %s -mtriple=thumbv7-apple-darwin10 -O3 | FileCheck %s
 ; CHECK: .zerofill __DATA,__bss,__MergedGlobals,16,2
 
 @prev = external global [0 x i16]

diff --git a/test/CodeGen/ARM/2011-08-02-MergedGlobalDbg.ll b/test/CodeGen/ARM/2011-08-02-MergedGlobalDbg.ll
index dee4d67ec54fff3267050775967ce9d948a8b86a..1614e8d0c67fdbb4237943a56ac33ae3df648080 100644 (file)
--- a/test/CodeGen/ARM/2011-08-02-MergedGlobalDbg.ll
+++ b/test/CodeGen/ARM/2011-08-02-MergedGlobalDbg.ll
@@ -1,4 +1,4 @@
-; RUN: llc -filetype=obj < %s | llvm-dwarfdump -debug-dump=info - | FileCheck %s
+; RUN: llc -O3 -filetype=obj < %s | llvm-dwarfdump -debug-dump=info - | FileCheck %s
 
 ; Check debug info output for merged global.
 ; DW_AT_location

diff --git a/test/CodeGen/ARM/global-merge-1.ll b/test/CodeGen/ARM/global-merge-1.ll
index 5db4b0086ec4c59054a6439dd51019efa8ea1316..03a9d33c8879749e01da7910ff59bef2bb66fc27 100644 (file)
--- a/test/CodeGen/ARM/global-merge-1.ll
+++ b/test/CodeGen/ARM/global-merge-1.ll
@@ -1,9 +1,10 @@
 ; RUN: llc %s -O0 -o - | FileCheck -check-prefix=NO-MERGE %s
-; RUN: llc %s -O0 -o - -enable-global-merge=false | FileCheck -check-prefix=NO-MERGE %s
-; RUN: llc %s -O0 -o - -enable-global-merge=true | FileCheck -check-prefix=NO-MERGE %s
-; RUN: llc %s -O1 -o - | FileCheck -check-prefix=MERGE %s
+; RUN: llc %s -O1 -o - | FileCheck -check-prefix=NO-MERGE %s
 ; RUN: llc %s -O1 -o - -enable-global-merge=false | FileCheck -check-prefix=NO-MERGE %s
-; RUN: llc %s -O1 -o - -enable-global-merge=true | FileCheck -check-prefix=MERGE %s
+; RUN: llc %s -O1 -o - -enable-global-merge=true | FileCheck -check-prefix=NO-MERGE %s
+; RUN: llc %s -O3 -o - | FileCheck -check-prefix=MERGE %s
+; RUN: llc %s -O3 -o - -enable-global-merge=false | FileCheck -check-prefix=NO-MERGE %s
+; RUN: llc %s -O3 -o - -enable-global-merge=true | FileCheck -check-prefix=MERGE %s
 
 ; MERGE-NOT: .zerofill __DATA,__bss,_bar,20,2
 ; MERGE-NOT: .zerofill __DATA,__bss,_baz,20,2

diff --git a/test/CodeGen/ARM/global-merge-addrspace.ll b/test/CodeGen/ARM/global-merge-addrspace.ll
index 0efa690bde2858a29f6ef5b78927b4af63271ebe..7b71e89bbe7dde8e6a4b7bfec0e341a6616b4251 100644 (file)
--- a/test/CodeGen/ARM/global-merge-addrspace.ll
+++ b/test/CodeGen/ARM/global-merge-addrspace.ll
@@ -1,4 +1,4 @@
-; RUN: llc < %s -mtriple=thumb-apple-darwin | FileCheck %s
+; RUN: llc < %s -mtriple=thumb-apple-darwin -O3 | FileCheck %s
 ; Test the GlobalMerge pass. Check that the pass does not crash when using
 ; multiple address spaces.
 

diff --git a/test/CodeGen/ARM/global-merge.ll b/test/CodeGen/ARM/global-merge.ll
index f88e927961967fd999f4a0a5319c9d483f84ba09..d535ea7c0cebfe6bb869900eb20a6bd6d19009d6 100644 (file)
--- a/test/CodeGen/ARM/global-merge.ll
+++ b/test/CodeGen/ARM/global-merge.ll
@@ -1,4 +1,4 @@
-; RUN: llc < %s -mtriple=thumb-apple-darwin -global-merge-on-const=true | FileCheck %s
+; RUN: llc < %s -mtriple=thumb-apple-darwin -O3 -global-merge-on-const=true | FileCheck %s
 ; Test the ARMGlobalMerge pass.  Use -march=thumb because it has a small
 ; value for the maximum offset (127).