Implemented stack symbol table ordering/packing optimization to improve data locality and code size from SP/FP offset encoding.

Differential Revision: http://reviews.llvm.org/D15393

llvm-svn: 260917

20 files changed, 213 insertions, 29 deletions

diff --git a/llvm/test/CodeGen/X86/2009-02-12-DebugInfoVLA.ll b/llvm/test/CodeGen/X86/2009-02-12-DebugInfoVLA.ll
index 2abb5ba7cd5..5a5a6eed741 100644
--- a/llvm/test/CodeGen/X86/2009-02-12-DebugInfoVLA.ll
+++ b/llvm/test/CodeGen/X86/2009-02-12-DebugInfoVLA.ll
@@ -1,5 +1,5 @@
 ; RUN: llc < %s
-; RUN: llc < %s -march=x86-64 -verify-machineinstrs | FileCheck %s
+; RUN: llc < %s -stack-symbol-ordering=0 -march=x86-64 -verify-machineinstrs | FileCheck %s
 ; PR3538
 target datalayout = "e-p:32:32:32-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:32:64-f32:32:32-f64:32:64-v64:64:64-v128:128:128-a0:0:64-f80:128:128"
 target triple = "i386-apple-darwin9"
diff --git a/llvm/test/CodeGen/X86/aligned-variadic.ll b/llvm/test/CodeGen/X86/aligned-variadic.ll
index 29415922062..1ea57296a70 100644
--- a/llvm/test/CodeGen/X86/aligned-variadic.ll
+++ b/llvm/test/CodeGen/X86/aligned-variadic.ll
@@ -1,5 +1,5 @@
-; RUN: llc < %s -mtriple=x86_64-apple-darwin | FileCheck %s -check-prefix=X64
-; RUN: llc < %s -mtriple=i686-apple-darwin   | FileCheck %s -check-prefix=X32
+; RUN: llc < %s -mtriple=x86_64-apple-darwin -stack-symbol-ordering=0 | FileCheck %s -check-prefix=X64
+; RUN: llc < %s -mtriple=i686-apple-darwin -stack-symbol-ordering=0 | FileCheck %s -check-prefix=X32
 
 %struct.Baz = type { [17 x i8] }
 %struct.__va_list_tag = type { i32, i32, i8*, i8* }
diff --git a/llvm/test/CodeGen/X86/avx512-mask-op.ll b/llvm/test/CodeGen/X86/avx512-mask-op.ll
index 7ae6c8a0efa..66307cd9f2c 100644
--- a/llvm/test/CodeGen/X86/avx512-mask-op.ll
+++ b/llvm/test/CodeGen/X86/avx512-mask-op.ll
@@ -1,6 +1,6 @@
 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
-; RUN: llc < %s -march=x86-64 -mtriple=x86_64-apple-darwin -mcpu=knl | FileCheck %s --check-prefix=CHECK --check-prefix=KNL
-; RUN: llc < %s -march=x86-64 -mtriple=x86_64-apple-darwin -mcpu=skx | FileCheck %s --check-prefix=CHECK --check-prefix=SKX
+; RUN: llc < %s -stack-symbol-ordering=0 -march=x86-64 -mtriple=x86_64-apple-darwin -mcpu=knl | FileCheck %s --check-prefix=CHECK --check-prefix=KNL
+; RUN: llc < %s -stack-symbol-ordering=0 -march=x86-64 -mtriple=x86_64-apple-darwin -mcpu=skx | FileCheck %s --check-prefix=CHECK --check-prefix=SKX
 
 define i16 @mask16(i16 %x) {
 ; CHECK-LABEL: mask16:
diff --git a/llvm/test/CodeGen/X86/cleanuppad-realign.ll b/llvm/test/CodeGen/X86/cleanuppad-realign.ll
index 5a565cc1570..314d5da07d7 100644
--- a/llvm/test/CodeGen/X86/cleanuppad-realign.ll
+++ b/llvm/test/CodeGen/X86/cleanuppad-realign.ll
@@ -1,5 +1,5 @@
-; RUN: llc -mtriple=i686-pc-windows-msvc < %s | FileCheck --check-prefix=X86 %s
-; RUN: llc -mtriple=x86_64-pc-windows-msvc < %s | FileCheck --check-prefix=X64 %s
+; RUN: llc -mtriple=i686-pc-windows-msvc -stack-symbol-ordering=0 < %s | FileCheck --check-prefix=X86 %s
+; RUN: llc -mtriple=x86_64-pc-windows-msvc -stack-symbol-ordering=0 < %s | FileCheck --check-prefix=X64 %s
 
 declare i32 @__CxxFrameHandler3(...)
 declare void @Dtor(i64* %o)
diff --git a/llvm/test/CodeGen/X86/dynamic-allocas-VLAs.ll b/llvm/test/CodeGen/X86/dynamic-allocas-VLAs.ll
index b0334d6a63e..7e30192c65b 100644
--- a/llvm/test/CodeGen/X86/dynamic-allocas-VLAs.ll
+++ b/llvm/test/CodeGen/X86/dynamic-allocas-VLAs.ll
@@ -1,5 +1,5 @@
-; RUN: llc < %s -mcpu=generic -march=x86-64 -mattr=+avx -mtriple=i686-apple-darwin10 | FileCheck %s
-; RUN: llc < %s -mcpu=generic -stackrealign -stack-alignment=32 -march=x86-64 -mattr=+avx -mtriple=i686-apple-darwin10 | FileCheck %s -check-prefix=FORCE-ALIGN
+; RUN: llc < %s -stack-symbol-ordering=0 -mcpu=generic -march=x86-64 -mattr=+avx -mtriple=i686-apple-darwin10 | FileCheck %s
+; RUN: llc < %s -stack-symbol-ordering=0 -mcpu=generic -stackrealign -stack-alignment=32 -march=x86-64 -mattr=+avx -mtriple=i686-apple-darwin10 | FileCheck %s -check-prefix=FORCE-ALIGN
 ; rdar://11496434
 
 ; no VLAs or dynamic alignment
diff --git a/llvm/test/CodeGen/X86/hipe-cc.ll b/llvm/test/CodeGen/X86/hipe-cc.ll
index e3808e75422..efea8967b75 100644
--- a/llvm/test/CodeGen/X86/hipe-cc.ll
+++ b/llvm/test/CodeGen/X86/hipe-cc.ll
@@ -1,4 +1,4 @@
-; RUN: llc < %s -tailcallopt -code-model=medium -stack-alignment=4 -mtriple=i686-linux-gnu -mcpu=pentium | FileCheck %s
+; RUN: llc < %s -stack-symbol-ordering=0 -tailcallopt -code-model=medium -stack-alignment=4 -mtriple=i686-linux-gnu -mcpu=pentium | FileCheck %s
 
 ; Check the HiPE calling convention works (x86-32)
 
diff --git a/llvm/test/CodeGen/X86/hipe-cc64.ll b/llvm/test/CodeGen/X86/hipe-cc64.ll
index 28d90399d85..0df4b3daad1 100644
--- a/llvm/test/CodeGen/X86/hipe-cc64.ll
+++ b/llvm/test/CodeGen/X86/hipe-cc64.ll
@@ -1,4 +1,4 @@
-; RUN: llc < %s -tailcallopt -code-model=medium -stack-alignment=8 -mtriple=x86_64-linux-gnu -mcpu=opteron | FileCheck %s
+; RUN: llc < %s -stack-symbol-ordering=0 -tailcallopt -code-model=medium -stack-alignment=8 -mtriple=x86_64-linux-gnu -mcpu=opteron | FileCheck %s
 
 ; Check the HiPE calling convention works (x86-64)
 
diff --git a/llvm/test/CodeGen/X86/local_stack_symbol_ordering.ll b/llvm/test/CodeGen/X86/local_stack_symbol_ordering.ll
new file mode 100644
index 00000000000..998c14565ce
--- /dev/null
+++ b/llvm/test/CodeGen/X86/local_stack_symbol_ordering.ll
@@ -0,0 +1,184 @@
+; RUN: llc < %s -mtriple=x86_64-unknown-linux-gnu | FileCheck %s -check-prefix=X64
+; RUN: llc < %s -mtriple=i686-unknown-linux-gnu | FileCheck %s -check-prefix=X32
+
+; CHECK-LABEL: foo
+
+; Check the functionality of the local stack symbol table ordering
+; heuristics.
+; The test has a bunch of locals of various sizes that are referenced a
+; different number of times.
+;
+; a   : 120B, 9 uses,   density = 0.075
+; aa  : 4000B, 1 use,   density = 0.00025
+; b   : 4B, 1 use,      density = 0.25
+; cc  : 4000B, 2 uses   density = 0.0005
+; d   : 4B, 2 uses      density = 0.5
+; e   : 4B, 3 uses      density = 0.75
+; f   : 4B, 4 uses      density = 1
+;
+; Given the size, number of uses and calculated density (uses / size), we're
+; going to hope that f gets allocated closest to the stack pointer,
+; followed by e, d, b, then a (to check for just a few).
+; We use gnu-inline asm between calls to prevent registerization of addresses
+; so that we get exact counts.
+;
+; The test is taken from something like this:
+; void foo()
+; {
+;   int f; // 4 uses.          4 / 4 = 1
+;   int a[30]; // 9 uses.      8 / 120 = 0.06
+;   int aa[1000]; // 1 use.    1 / 4000 =
+;   int e; // 3 uses.          3 / 4 = 0.75
+;   int cc[1000]; // 2 uses.   2 / 4000 = 
+;   int b; // 1 use.           1 / 4 = 0.25
+;   int d; // 2 uses.          2 / 4 = 0.5
+;   int aaa[1000]; // 2 uses.  2 / 4000
+;
+; 
+;   check_a(&a);
+;   bar1(&aaa);
+;   asm ("":::"esi","edi","ebp","ebx","rbx","r12","r13","r14","r15","rbp");
+;   bar1(&a);
+;   check_f(&f);
+;   asm ("":::"esi","edi","ebp","ebx","rbx","r12","r13","r14","r15","rbp");
+;   bar1(&a);
+;   bar3(&aa, &aaa, &cc);
+;   asm ("":::"esi","edi","ebp","ebx","rbx","r12","r13","r14","r15","rbp");
+;   bar2(&a,&cc);
+;   check_b(&b);
+;   asm ("":::"esi","edi","ebp","ebx","rbx","r12","r13","r14","r15","rbp");
+;   bar1(&a);
+;   asm ("":::"esi","edi","ebp","ebx","rbx","r12","r13","r14","r15","rbp");
+;   bar2(&a, &f);
+;   check_e(&e);
+;   asm ("":::"esi","edi","ebp","ebx","rbx","r12","r13","r14","r15","rbp");
+;   bar1(&a);
+;   asm ("":::"esi","edi","ebp","ebx","rbx","r12","r13","r14","r15","rbp");
+;   bar2(&e, &f);
+;   check_d(&d);
+;   bar1(&a);
+;   asm ("":::"esi","edi","ebp","ebx","rbx","r12","r13","r14","r15","rbp");
+;   bar3(&d, &e, &f);
+;   asm ("":::"esi","edi","ebp","ebx","rbx","r12","r13","r14","r15","rbp");
+;   bar1(&a);
+; }
+;
+; X64: leaq 16(%rsp), %rdi
+; X64: callq check_a
+; X64: callq bar1
+; X64: callq bar1
+; X64: leaq (%rsp), %rdi
+; X64: callq check_f
+; X64: callq bar1
+; X64: callq bar3
+; X64: callq bar2
+; X64: leaq 12(%rsp), %rdi
+; X64: callq check_b
+; X64: callq bar1
+; X64: callq bar2
+; X64: leaq 4(%rsp), %rdi
+; X64: callq check_e
+; X64: callq bar1
+; X64: callq bar2
+; X64: leaq 8(%rsp), %rdi
+; X64: callq check_d
+
+; X32: leal 32(%esp)
+; X32: calll check_a
+; X32: calll bar1
+; X32: calll bar1
+; X32: leal 16(%esp)
+; X32: calll check_f
+; X32: calll bar1
+; X32: calll bar3
+; X32: calll bar2
+; X32: leal 28(%esp)
+; X32: calll check_b
+; X32: calll bar1
+; X32: calll bar2
+; X32: leal 20(%esp)
+; X32: calll check_e
+; X32: calll bar1
+; X32: calll bar2
+; X32: leal 24(%esp)
+; X32: calll check_d
+
+
+define void @foo() nounwind uwtable {
+entry:
+  %f = alloca i32, align 4
+  %a = alloca [30 x i32], align 16
+  %aa = alloca [1000 x i32], align 16
+  %e = alloca i32, align 4
+  %cc = alloca [1000 x i32], align 16
+  %b = alloca i32, align 4
+  %d = alloca i32, align 4
+  %aaa = alloca [1000 x i32], align 16
+  %0 = bitcast i32* %f to i8*
+  call void @llvm.lifetime.start(i64 4, i8* %0) #1
+  %1 = bitcast [30 x i32]* %a to i8*
+  call void @llvm.lifetime.start(i64 120, i8* %1) #1
+  %2 = bitcast [1000 x i32]* %aa to i8*
+  call void @llvm.lifetime.start(i64 4000, i8* %2) #1
+  %3 = bitcast i32* %e to i8*
+  call void @llvm.lifetime.start(i64 4, i8* %3) #1
+  %4 = bitcast [1000 x i32]* %cc to i8*
+  call void @llvm.lifetime.start(i64 4000, i8* %4) #1
+  %5 = bitcast i32* %b to i8*
+  call void @llvm.lifetime.start(i64 4, i8* %5) #1
+  %6 = bitcast i32* %d to i8*
+  call void @llvm.lifetime.start(i64 4, i8* %6) #1
+  %7 = bitcast [1000 x i32]* %aaa to i8*
+  call void @llvm.lifetime.start(i64 4000, i8* %7) #1
+  %call = call i32 ([30 x i32]*, ...) bitcast (i32 (...)* @check_a to i32 ([30 x i32]*, ...)*)([30 x i32]* %a)
+  %call1 = call i32 ([1000 x i32]*, ...) bitcast (i32 (...)* @bar1 to i32 ([1000 x i32]*, ...)*)([1000 x i32]* %aaa)
+  call void asm sideeffect "", "~{esi},~{edi},~{ebp},~{ebx},~{rbx},~{r12},~{r13},~{r14},~{r15},~{rbp},~{dirflag},~{fpsr},~{flags}"() #1
+  %call2 = call i32 ([30 x i32]*, ...) bitcast (i32 (...)* @bar1 to i32 ([30 x i32]*, ...)*)([30 x i32]* %a)
+  %call3 = call i32 (i32*, ...) bitcast (i32 (...)* @check_f to i32 (i32*, ...)*)(i32* %f)
+  call void asm sideeffect "", "~{esi},~{edi},~{ebp},~{ebx},~{rbx},~{r12},~{r13},~{r14},~{r15},~{rbp},~{dirflag},~{fpsr},~{flags}"() #1
+  %call4 = call i32 ([30 x i32]*, ...) bitcast (i32 (...)* @bar1 to i32 ([30 x i32]*, ...)*)([30 x i32]* %a)
+  %call5 = call i32 ([1000 x i32]*, [1000 x i32]*, [1000 x i32]*, ...) bitcast (i32 (...)* @bar3 to i32 ([1000 x i32]*, [1000 x i32]*, [1000 x i32]*, ...)*)([1000 x i32]* %aa, [1000 x i32]* %aaa, [1000 x i32]* %cc)
+  call void asm sideeffect "", "~{esi},~{edi},~{ebp},~{ebx},~{rbx},~{r12},~{r13},~{r14},~{r15},~{rbp},~{dirflag},~{fpsr},~{flags}"() #1
+  %call6 = call i32 ([30 x i32]*, [1000 x i32]*, ...) bitcast (i32 (...)* @bar2 to i32 ([30 x i32]*, [1000 x i32]*, ...)*)([30 x i32]* %a, [1000 x i32]* %cc)
+  %call7 = call i32 (i32*, ...) bitcast (i32 (...)* @check_b to i32 (i32*, ...)*)(i32* %b)
+  call void asm sideeffect "", "~{esi},~{edi},~{ebp},~{ebx},~{rbx},~{r12},~{r13},~{r14},~{r15},~{rbp},~{dirflag},~{fpsr},~{flags}"() #1
+  %call8 = call i32 ([30 x i32]*, ...) bitcast (i32 (...)* @bar1 to i32 ([30 x i32]*, ...)*)([30 x i32]* %a)
+  call void asm sideeffect "", "~{esi},~{edi},~{ebp},~{ebx},~{rbx},~{r12},~{r13},~{r14},~{r15},~{rbp},~{dirflag},~{fpsr},~{flags}"() #1
+  %call9 = call i32 ([30 x i32]*, i32*, ...) bitcast (i32 (...)* @bar2 to i32 ([30 x i32]*, i32*, ...)*)([30 x i32]* %a, i32* %f)
+  %call10 = call i32 (i32*, ...) bitcast (i32 (...)* @check_e to i32 (i32*, ...)*)(i32* %e)
+  call void asm sideeffect "", "~{esi},~{edi},~{ebp},~{ebx},~{rbx},~{r12},~{r13},~{r14},~{r15},~{rbp},~{dirflag},~{fpsr},~{flags}"() #1
+  %call11 = call i32 ([30 x i32]*, ...) bitcast (i32 (...)* @bar1 to i32 ([30 x i32]*, ...)*)([30 x i32]* %a)
+  call void asm sideeffect "", "~{esi},~{edi},~{ebp},~{ebx},~{rbx},~{r12},~{r13},~{r14},~{r15},~{rbp},~{dirflag},~{fpsr},~{flags}"() #1
+  %call12 = call i32 (i32*, i32*, ...) bitcast (i32 (...)* @bar2 to i32 (i32*, i32*, ...)*)(i32* %e, i32* %f)
+  %call13 = call i32 (i32*, ...) bitcast (i32 (...)* @check_d to i32 (i32*, ...)*)(i32* %d)
+  %call14 = call i32 ([30 x i32]*, ...) bitcast (i32 (...)* @bar1 to i32 ([30 x i32]*, ...)*)([30 x i32]* %a)
+  call void asm sideeffect "", "~{esi},~{edi},~{ebp},~{ebx},~{rbx},~{r12},~{r13},~{r14},~{r15},~{rbp},~{dirflag},~{fpsr},~{flags}"() #1
+  %call15 = call i32 (i32*, i32*, i32*, ...) bitcast (i32 (...)* @bar3 to i32 (i32*, i32*, i32*, ...)*)(i32* %d, i32* %e, i32* %f)
+  call void asm sideeffect "", "~{esi},~{edi},~{ebp},~{ebx},~{rbx},~{r12},~{r13},~{r14},~{r15},~{rbp},~{dirflag},~{fpsr},~{flags}"() #1
+  %call16 = call i32 ([30 x i32]*, ...) bitcast (i32 (...)* @bar1 to i32 ([30 x i32]*, ...)*)([30 x i32]* %a)
+  call void @llvm.lifetime.end(i64 4000, i8* %7) #1
+  call void @llvm.lifetime.end(i64 4, i8* %6) #1
+  call void @llvm.lifetime.end(i64 4, i8* %5) #1
+  call void @llvm.lifetime.end(i64 4000, i8* %4) #1
+  call void @llvm.lifetime.end(i64 4, i8* %3) #1
+  call void @llvm.lifetime.end(i64 4000, i8* %2) #1
+  call void @llvm.lifetime.end(i64 120, i8* %1) #1
+  call void @llvm.lifetime.end(i64 4, i8* %0) #1
+  ret void
+}
+
+; Function Attrs: nounwind
+declare void @llvm.lifetime.start(i64, i8* nocapture) #1
+
+declare i32 @check_a(...) #2
+declare i32 @bar1(...) #2
+declare i32 @check_f(...) #2
+declare i32 @bar3(...) #2
+declare i32 @bar2(...) #2
+declare i32 @check_b(...) #2
+declare i32 @check_e(...) #2
+declare i32 @check_d(...) #2
+
+; Function Attrs: nounwind
+declare void @llvm.lifetime.end(i64, i8* nocapture) #1
+
diff --git a/llvm/test/CodeGen/X86/phys-reg-local-regalloc.ll b/llvm/test/CodeGen/X86/phys-reg-local-regalloc.ll
index a0adba0f833..b1fc7ac5459 100644
--- a/llvm/test/CodeGen/X86/phys-reg-local-regalloc.ll
+++ b/llvm/test/CodeGen/X86/phys-reg-local-regalloc.ll
@@ -1,6 +1,6 @@
-; RUN: llc < %s -march=x86 -mtriple=i386-apple-darwin9 -mcpu=generic -regalloc=fast -optimize-regalloc=0 | FileCheck %s
-; RUN: llc -O0 < %s -march=x86 -mtriple=i386-apple-darwin9 -mcpu=generic -regalloc=fast | FileCheck %s
-; RUN: llc < %s -march=x86 -mtriple=i386-apple-darwin9 -mcpu=atom -regalloc=fast -optimize-regalloc=0 | FileCheck -check-prefix=ATOM %s
+; RUN: llc < %s -stack-symbol-ordering=0 -march=x86 -mtriple=i386-apple-darwin9 -mcpu=generic -regalloc=fast -optimize-regalloc=0 | FileCheck %s
+; RUN: llc -O0 < %s -stack-symbol-ordering=0 -march=x86 -mtriple=i386-apple-darwin9 -mcpu=generic -regalloc=fast | FileCheck %s
+; RUN: llc < %s -stack-symbol-ordering=0 -march=x86 -mtriple=i386-apple-darwin9 -mcpu=atom -regalloc=fast -optimize-regalloc=0 | FileCheck -check-prefix=ATOM %s
 ; CHECKed instructions should be the same with or without -O0 except on Intel Atom due to instruction scheduling.
 
 @.str = private constant [12 x i8] c"x + y = %i\0A\00", align 1 ; <[12 x i8]*> [#uses=1]
diff --git a/llvm/test/CodeGen/X86/seh-catch-all-win32.ll b/llvm/test/CodeGen/X86/seh-catch-all-win32.ll
index b895e11e4b3..1195bbedd8f 100644
--- a/llvm/test/CodeGen/X86/seh-catch-all-win32.ll
+++ b/llvm/test/CodeGen/X86/seh-catch-all-win32.ll
@@ -1,4 +1,4 @@
-; RUN: llc -mtriple=i686-windows-msvc < %s | FileCheck %s
+; RUN: llc -stack-symbol-ordering=0 -mtriple=i686-windows-msvc < %s | FileCheck %s
 
 ; 32-bit catch-all has to use a filter function because that's how it saves the
 ; exception code.
diff --git a/llvm/test/CodeGen/X86/seh-stack-realign.ll b/llvm/test/CodeGen/X86/seh-stack-realign.ll
index 654cad347f6..a2eb3bc53b7 100644
--- a/llvm/test/CodeGen/X86/seh-stack-realign.ll
+++ b/llvm/test/CodeGen/X86/seh-stack-realign.ll
@@ -1,4 +1,4 @@
-; RUN: llc -mtriple=i686-windows-msvc < %s | FileCheck %s
+; RUN: llc -stack-symbol-ordering=0 -mtriple=i686-windows-msvc < %s | FileCheck %s
 
 ; 32-bit catch-all has to use a filter function because that's how it saves the
 ; exception code.
diff --git a/llvm/test/CodeGen/X86/ssp-data-layout.ll b/llvm/test/CodeGen/X86/ssp-data-layout.ll
index 4a63aceb7cc..e954d9c1042 100644
--- a/llvm/test/CodeGen/X86/ssp-data-layout.ll
+++ b/llvm/test/CodeGen/X86/ssp-data-layout.ll
@@ -1,4 +1,4 @@
-; RUN: llc < %s -disable-fp-elim -mtriple=x86_64-pc-linux-gnu -mcpu=corei7 -o - | FileCheck %s
+; RUN: llc < %s -stack-symbol-ordering=0 -disable-fp-elim -mtriple=x86_64-pc-linux-gnu -mcpu=corei7 -o - | FileCheck %s
 ;  This test is fairly fragile.  The goal is to ensure that "large" stack
 ;  objects are allocated closest to the stack protector (i.e., farthest away 
 ;  from the Stack Pointer.)  In standard SSP mode this means that large (>=
diff --git a/llvm/test/CodeGen/X86/statepoint-stack-usage.ll b/llvm/test/CodeGen/X86/statepoint-stack-usage.ll
index d4784212810..d4bc7d47f66 100644
--- a/llvm/test/CodeGen/X86/statepoint-stack-usage.ll
+++ b/llvm/test/CodeGen/X86/statepoint-stack-usage.ll
@@ -1,4 +1,4 @@
-; RUN: llc < %s | FileCheck %s
+; RUN: llc -stack-symbol-ordering=0 < %s | FileCheck %s
 
 target datalayout = "e-i64:64-f80:128-n8:16:32:64-S128"
 target triple = "x86_64-pc-linux-gnu"
diff --git a/llvm/test/CodeGen/X86/statepoint-stackmap-format.ll b/llvm/test/CodeGen/X86/statepoint-stackmap-format.ll
index 9632360d723..2b1357a1179 100644
--- a/llvm/test/CodeGen/X86/statepoint-stackmap-format.ll
+++ b/llvm/test/CodeGen/X86/statepoint-stackmap-format.ll
@@ -1,5 +1,5 @@
-; RUN: llc < %s -mtriple="x86_64-pc-linux-gnu" | FileCheck %s
-; RUN: llc < %s -mtriple="x86_64-pc-unknown-elf" | FileCheck %s
+; RUN: llc < %s -stack-symbol-ordering=0 -mtriple="x86_64-pc-linux-gnu" | FileCheck %s
+; RUN: llc < %s -stack-symbol-ordering=0 -mtriple="x86_64-pc-unknown-elf" | FileCheck %s
 
 ; This test is a sanity check to ensure statepoints are generating StackMap
 ; sections correctly.  This is not intended to be a rigorous test of the 
diff --git a/llvm/test/CodeGen/X86/statepoint-vector.ll b/llvm/test/CodeGen/X86/statepoint-vector.ll
index 9d80e9217b4..6290c9e87d4 100644
--- a/llvm/test/CodeGen/X86/statepoint-vector.ll
+++ b/llvm/test/CodeGen/X86/statepoint-vector.ll
@@ -1,4 +1,4 @@
-; RUN: llc -mcpu=core-avx -debug-only=stackmaps < %s | FileCheck %s
+; RUN: llc -stack-symbol-ordering=0 -mcpu=core-avx -debug-only=stackmaps < %s | FileCheck %s
 ; REQUIRES: asserts
 
 target triple = "x86_64-pc-linux-gnu"
diff --git a/llvm/test/CodeGen/X86/stdarg.ll b/llvm/test/CodeGen/X86/stdarg.ll
index 42cbcb1008d..7b4f4e845fc 100644
--- a/llvm/test/CodeGen/X86/stdarg.ll
+++ b/llvm/test/CodeGen/X86/stdarg.ll
@@ -1,4 +1,4 @@
-; RUN: llc < %s -mtriple=x86_64-linux | FileCheck %s
+; RUN: llc -stack-symbol-ordering=0 < %s -mtriple=x86_64-linux | FileCheck %s
 
 %struct.__va_list_tag = type { i32, i32, i8*, i8* }
 
diff --git a/llvm/test/CodeGen/X86/widen_load-1.ll b/llvm/test/CodeGen/X86/widen_load-1.ll
index c670b45df74..810e409c175 100644
--- a/llvm/test/CodeGen/X86/widen_load-1.ll
+++ b/llvm/test/CodeGen/X86/widen_load-1.ll
@@ -1,5 +1,5 @@
-; RUN: llc %s -o - -march=x86-64 -mattr=-avx -mtriple=x86_64-unknown-linux-gnu | FileCheck %s --check-prefix=SSE
-; RUN: llc %s -o - -march=x86-64 -mattr=+avx -mtriple=x86_64-unknown-linux-gnu | FileCheck %s --check-prefix=AVX
+; RUN: llc -stack-symbol-ordering=0 %s -o - -march=x86-64 -mattr=-avx -mtriple=x86_64-unknown-linux-gnu | FileCheck %s --check-prefix=SSE
+; RUN: llc -stack-symbol-ordering=0 %s -o - -march=x86-64 -mattr=+avx -mtriple=x86_64-unknown-linux-gnu | FileCheck %s --check-prefix=AVX
 ; PR4891
 ; PR5626
 
diff --git a/llvm/test/CodeGen/X86/win-catchpad-varargs.ll b/llvm/test/CodeGen/X86/win-catchpad-varargs.ll
index 6508f3bd7d6..a31b3d72c56 100644
--- a/llvm/test/CodeGen/X86/win-catchpad-varargs.ll
+++ b/llvm/test/CodeGen/X86/win-catchpad-varargs.ll
@@ -1,5 +1,5 @@
-; RUN: llc -mtriple=x86_64-windows-msvc < %s | FileCheck %s --check-prefix=X64
-; RUN: llc -mtriple=i686-windows-msvc < %s | FileCheck %s --check-prefix=X86
+; RUN: llc -stack-symbol-ordering=0 -mtriple=x86_64-windows-msvc < %s | FileCheck %s --check-prefix=X64
+; RUN: llc -stack-symbol-ordering=0 -mtriple=i686-windows-msvc < %s | FileCheck %s --check-prefix=X86
 
 declare void @llvm.va_start(i8*)
 declare void @llvm.va_end(i8*)
diff --git a/llvm/test/CodeGen/X86/win-catchpad.ll b/llvm/test/CodeGen/X86/win-catchpad.ll
index 836c53bda8e..dc1490fadda 100644
--- a/llvm/test/CodeGen/X86/win-catchpad.ll
+++ b/llvm/test/CodeGen/X86/win-catchpad.ll
@@ -1,5 +1,5 @@
-; RUN: llc -verify-machineinstrs -mtriple=i686-pc-windows-msvc < %s | FileCheck --check-prefix=X86 %s
-; RUN: llc -verify-machineinstrs -mtriple=x86_64-pc-windows-msvc < %s | FileCheck --check-prefix=X64 %s
+; RUN: llc -stack-symbol-ordering=0 -verify-machineinstrs -mtriple=i686-pc-windows-msvc < %s | FileCheck --check-prefix=X86 %s
+; RUN: llc -stack-symbol-ordering=0 -verify-machineinstrs -mtriple=x86_64-pc-windows-msvc < %s | FileCheck --check-prefix=X64 %s
 
 ; Loosely based on IR for this C++ source code:
 ;   void f(int p);
diff --git a/llvm/test/CodeGen/X86/win32-seh-catchpad-realign.ll b/llvm/test/CodeGen/X86/win32-seh-catchpad-realign.ll
index 23aeea37c11..1ba0c1a0efe 100644
--- a/llvm/test/CodeGen/X86/win32-seh-catchpad-realign.ll
+++ b/llvm/test/CodeGen/X86/win32-seh-catchpad-realign.ll
@@ -1,4 +1,4 @@
-; RUN: llc < %s | FileCheck %s
+; RUN: llc -stack-symbol-ordering=0 < %s | FileCheck %s
 
 ; The aligned alloca means that we have to realign the stack, which forces the
 ; use of ESI to address local variables.