[WebAssembly] Stackify function prologs and epilogs

[dschuff]

Summary

The instructions are the same, but fewer locals are used.

[sunfish]

:-)

llvm/trunk/lib/Target/WebAssembly/WebAssemblyFrameLowering.cpp

 #include "llvm/CodeGen/MachineRegisterInfo.h"
 #include "llvm/Support/Debug.h"
 using namespace llvm;
 
 #define DEBUG_TYPE "wasm-frame-info"
 
 // TODO: Implement a red zone?
 // TODO: wasm64
-// TODO: Prolog/epilog should be stackified too. This pass runs after register
-//       stackification, so we'll have to do it manually.
 // TODO: Emit TargetOpcode::CFI_INSTRUCTION instructions
 
 /// Return true if the specified function should have a dedicated frame pointer
 /// register.
 bool WebAssemblyFrameLowering::hasFP(const MachineFunction &MF) const {
   const MachineFrameInfo *MFI = MF.getFrameInfo();
   const auto *RegInfo =
       MF.getSubtarget<WebAssemblySubtarget>().getRegisterInfo();

 }
 
 void WebAssemblyFrameLowering::emitPrologue(MachineFunction &MF,
                                             MachineBasicBlock &MBB) const {
   // TODO: Do ".setMIFlag(MachineInstr::FrameSetup)" on emitted instructions
   auto *MFI = MF.getFrameInfo();
   assert(MFI->getCalleeSavedInfo().empty() &&
          "WebAssembly should not have callee-saved registers");
+  auto *WFI = MF.getInfo<WebAssemblyFunctionInfo>();
 
   uint64_t StackSize = MFI->getStackSize();
   if (!StackSize && !MFI->adjustsStack() && !hasFP(MF)) return;
 
   const auto *TII = MF.getSubtarget<WebAssemblySubtarget>().getInstrInfo();
   auto &MRI = MF.getRegInfo();
 
   auto InsertPt = MBB.begin();
   DebugLoc DL;
 
+  unsigned SPAddr = MRI.createVirtualRegister(&WebAssembly::I32RegClass);
   unsigned SPReg = MRI.createVirtualRegister(&WebAssembly::I32RegClass);
   auto *SPSymbol = MF.createExternalSymbolName("__stack_pointer");
-  BuildMI(MBB, InsertPt, DL, TII->get(WebAssembly::CONST_I32), SPReg)
+  BuildMI(MBB, InsertPt, DL, TII->get(WebAssembly::CONST_I32), SPAddr)
       .addExternalSymbol(SPSymbol);
   // This MachinePointerInfo should reference __stack_pointer as well but
   // doesn't because MachinePointerInfo() takes a GV which we don't have for
   // __stack_pointer. TODO: check if PseudoSourceValue::ExternalSymbolCallEntry
   // is appropriate instead. (likewise for EmitEpologue below)
   auto *LoadMMO = new MachineMemOperand(MachinePointerInfo(),
                                         MachineMemOperand::MOLoad, 4, 4);
   // Load the SP value.
   BuildMI(MBB, InsertPt, DL, TII->get(WebAssembly::LOAD_I32),
           StackSize ? SPReg : (unsigned)WebAssembly::SP32)
-      .addImm(0)      // offset
-      .addReg(SPReg)  // addr
-      .addImm(2)      // p2align
+      .addImm(0)       // offset
+      .addReg(SPAddr)  // addr
+      .addImm(2)       // p2align
       .addMemOperand(LoadMMO);
+  WFI->stackifyVReg(SPAddr);
 
-  unsigned OffsetReg = 0;
   if (StackSize) {
     // Subtract the frame size
-    OffsetReg = MRI.createVirtualRegister(&WebAssembly::I32RegClass);
+    unsigned OffsetReg = MRI.createVirtualRegister(&WebAssembly::I32RegClass);
     BuildMI(MBB, InsertPt, DL, TII->get(WebAssembly::CONST_I32), OffsetReg)
         .addImm(StackSize);
     BuildMI(MBB, InsertPt, DL, TII->get(WebAssembly::SUB_I32),
             WebAssembly::SP32)
         .addReg(SPReg)
         .addReg(OffsetReg);
+    WFI->stackifyVReg(OffsetReg);
+    WFI->stackifyVReg(SPReg);
   }
   if (hasFP(MF)) {
     // Unlike most conventional targets (where FP points to the saved FP),
     // FP points to the bottom of the fixed-size locals, so we can use positive
     // offsets in load/store instructions.
     BuildMI(MBB, InsertPt, DL, TII->get(WebAssembly::COPY_LOCAL_I32),
             WebAssembly::FP32)
         .addReg(WebAssembly::SP32);
   }
   if (StackSize) {
-    assert(OffsetReg);
+    SPAddr = MRI.createVirtualRegister(&WebAssembly::I32RegClass);
     // The SP32 register now has the new stacktop. Also write it back to memory.
-    BuildMI(MBB, InsertPt, DL, TII->get(WebAssembly::CONST_I32), OffsetReg)
+    BuildMI(MBB, InsertPt, DL, TII->get(WebAssembly::CONST_I32), SPAddr)
         .addExternalSymbol(SPSymbol);
     auto *MMO = new MachineMemOperand(MachinePointerInfo(),
                                       MachineMemOperand::MOStore, 4, 4);
     BuildMI(MBB, InsertPt, DL, TII->get(WebAssembly::STORE_I32),
             WebAssembly::SP32)
         .addImm(0)
-        .addReg(OffsetReg)
+        .addReg(SPAddr)
         .addImm(2)  // p2align
         .addReg(WebAssembly::SP32)
         .addMemOperand(MMO);
+    WFI->stackifyVReg(SPAddr);
   }
 }
 
 void WebAssemblyFrameLowering::emitEpilogue(MachineFunction &MF,
                                             MachineBasicBlock &MBB) const {
   auto *MFI = MF.getFrameInfo();
   uint64_t StackSize = MFI->getStackSize();
   if (!StackSize && !MFI->adjustsStack() && !hasFP(MF)) return;
+  auto *WFI = MF.getInfo<WebAssemblyFunctionInfo>();
   const auto *TII = MF.getSubtarget<WebAssemblySubtarget>().getInstrInfo();
   auto &MRI = MF.getRegInfo();
-  unsigned OffsetReg = MRI.createVirtualRegister(&WebAssembly::I32RegClass);
   auto InsertPt = MBB.getFirstTerminator();
   DebugLoc DL;
 
   if (InsertPt != MBB.end()) {
     DL = InsertPt->getDebugLoc();
   }
 
   // Restore the stack pointer. If we had fixed-size locals, add the offset
   // subtracted in the prolog.
   if (StackSize) {
+    unsigned OffsetReg = MRI.createVirtualRegister(&WebAssembly::I32RegClass);
     BuildMI(MBB, InsertPt, DL, TII->get(WebAssembly::CONST_I32), OffsetReg)
         .addImm(StackSize);
     BuildMI(MBB, InsertPt, DL, TII->get(WebAssembly::ADD_I32),
             WebAssembly::SP32)
         .addReg(hasFP(MF) ? WebAssembly::FP32 : WebAssembly::SP32)
         .addReg(OffsetReg);
+    WFI->stackifyVReg(OffsetReg);
   }
 
   auto *SPSymbol = MF.createExternalSymbolName("__stack_pointer");
-  // Re-use OffsetReg to hold the address of the stacktop
-  BuildMI(MBB, InsertPt, DL, TII->get(WebAssembly::CONST_I32), OffsetReg)
+  unsigned SPAddr = MRI.createVirtualRegister(&WebAssembly::I32RegClass);
+  BuildMI(MBB, InsertPt, DL, TII->get(WebAssembly::CONST_I32), SPAddr)
       .addExternalSymbol(SPSymbol);
   auto *MMO = new MachineMemOperand(MachinePointerInfo(),
                                     MachineMemOperand::MOStore, 4, 4);
   BuildMI(MBB, InsertPt, DL, TII->get(WebAssembly::STORE_I32),
           WebAssembly::SP32)
       .addImm(0)
-      .addReg(OffsetReg)
+      .addReg(SPAddr)
       .addImm(2)  // p2align
       .addReg((!StackSize && hasFP(MF)) ? WebAssembly::FP32 : WebAssembly::SP32)
       .addMemOperand(MMO);
+  WFI->stackifyVReg(SPAddr);
 }

llvm/trunk/test/CodeGen/WebAssembly/byval.ll

 declare void @ext_byval_func_alignedstruct(%AlignedStruct* byval)
 declare void @ext_byval_func_bigarray(%BigArray* byval)
 declare void @ext_byval_func_empty(%EmptyStruct* byval)
 
 ; CHECK-LABEL: byval_arg
 define void @byval_arg(%SmallStruct* %ptr) {
  ; CHECK: .param i32
  ; Subtract 16 from SP (SP is 16-byte aligned)
- ; CHECK: i32.const [[L1:.+]]=, __stack_pointer
- ; CHECK-NEXT: i32.load [[L1]]=, 0([[L1]])
- ; CHECK-NEXT: i32.const [[L2:.+]]=, 16
- ; CHECK-NEXT: i32.sub [[SP:.+]]=, [[L1]], [[L2]]
+ ; CHECK: i32.const $push[[L1:.+]]=, __stack_pointer
+ ; CHECK-NEXT: i32.load $push[[L2:.+]]=, 0($pop[[L1]])
+ ; CHECK-NEXT: i32.const $push[[L3:.+]]=, 16
+ ; CHECK-NEXT: i32.sub [[SP:.+]]=, $pop[[L2]], $pop[[L3]]
  ; Ensure SP is stored back before the call
- ; CHECK-NEXT: i32.const [[L3:.+]]=, __stack_pointer
- ; CHECK-NEXT: i32.store {{.*}}=, 0([[L3]]), [[SP]]
+ ; CHECK-NEXT: i32.const $push[[L4:.+]]=, __stack_pointer
+ ; CHECK-NEXT: i32.store {{.*}}=, 0($pop[[L4]]), [[SP]]
  ; Copy the SmallStruct argument to the stack (SP+12, original SP-4)
  ; CHECK-NEXT: i32.load $push[[L4:.+]]=, 0($0)
  ; CHECK-NEXT: i32.store {{.*}}=, 12([[SP]]), $pop[[L4]]
  ; Pass a pointer to the stack slot to the function
  ; CHECK-NEXT: i32.const [[L5:.+]]=, 12
  ; CHECK-NEXT: i32.add [[ARG:.+]]=, [[SP]], [[L5]]
  ; CHECK-NEXT: call ext_byval_func@FUNCTION, [[L5]]
  call void @ext_byval_func(%SmallStruct* byval %ptr)
  ; Restore the stack
- ; CHECK-NEXT: i32.const [[L6:.+]]=, 16
- ; CHECK-NEXT: i32.add [[SP]]=, [[SP]], [[L6]]
- ; CHECK-NEXT: i32.const [[L7:.+]]=, __stack_pointer
- ; CHECK-NEXT: i32.store {{.*}}=, 0([[L7]]), [[SP]]
+ ; CHECK-NEXT: i32.const $push[[L6:.+]]=, 16
+ ; CHECK-NEXT: i32.add [[SP]]=, [[SP]], $pop[[L6]]
+ ; CHECK-NEXT: i32.const $push[[L7:.+]]=, __stack_pointer
+ ; CHECK-NEXT: i32.store {{.*}}=, 0($pop[[L7]]), [[SP]]
  ; CHECK-NEXT: return
  ret void
 }
 
 ; CHECK-LABEL: byval_arg_align8
 define void @byval_arg_align8(%SmallStruct* %ptr) {
  ; CHECK: .param i32
  ; Don't check the entire SP sequence, just enough to get the alignment.
- ; CHECK: i32.const [[L2:.+]]=, 16
- ; CHECK-NEXT: i32.sub [[SP:.+]]=, {{.+}}, [[L2]]
+ ; CHECK: i32.const $push[[L1:.+]]=, 16
+ ; CHECK-NEXT: i32.sub [[SP:.+]]=, {{.+}}, $pop[[L1]]
  ; Copy the SmallStruct argument to the stack (SP+8, original SP-8)
  ; CHECK: i32.load $push[[L4:.+]]=, 0($0):p2align=3
  ; CHECK-NEXT: i32.store {{.*}}=, 8([[SP]]):p2align=3, $pop[[L4]]
  ; Pass a pointer to the stack slot to the function
  ; CHECK-NEXT: i32.const [[L5:.+]]=, 8
  ; CHECK-NEXT: i32.add [[ARG:.+]]=, [[SP]], [[L5]]
  ; CHECK-NEXT: call ext_byval_func_align8@FUNCTION, [[L5]]
  call void @ext_byval_func_align8(%SmallStruct* byval align 8 %ptr)
  ret void
 }
 
 ; CHECK-LABEL: byval_arg_double
 define void @byval_arg_double(%AlignedStruct* %ptr) {
  ; CHECK: .param i32
  ; Subtract 16 from SP (SP is 16-byte aligned)
- ; CHECK: i32.const [[L2:.+]]=, 16
- ; CHECK-NEXT: i32.sub [[SP:.+]]=, {{.+}}, [[L2]]
+ ; CHECK: i32.const $push[[L1:.+]]=, 16
+ ; CHECK-NEXT: i32.sub [[SP:.+]]=, {{.+}}, $pop[[L1]]
  ; Copy the AlignedStruct argument to the stack (SP+0, original SP-16)
  ; Just check the last load/store pair of the memcpy
  ; CHECK: i64.load $push[[L4:.+]]=, 0($0)
  ; CHECK-NEXT: i64.store {{.*}}=, 0([[SP]]), $pop[[L4]]
  ; Pass a pointer to the stack slot to the function
  ; CHECK-NEXT: call ext_byval_func_alignedstruct@FUNCTION, [[SP]]
  tail call void @ext_byval_func_alignedstruct(%AlignedStruct* byval %ptr)
  ret void

 define void @byval_empty_callee(%EmptyStruct* byval %ptr) {
  ; CHECK: .param i32
  ; CHECK: call ext_func_empty@FUNCTION, $0
  call void @ext_func_empty(%EmptyStruct* %ptr)
  ret void
 }
 
 ; Call memcpy for "big" byvals.
-; TODO: When the prolog/epilog sequences are optimized, refine these checks to
-; be more specific.
-
 ; CHECK-LABEL: big_byval:
+; CHECK: i32.const $push[[L1:.+]]=, __stack_pointer
+; CHECK-NEXT: i32.load $push[[L2:.+]]=, 0($pop[[L1]])
+; CHECK-NEXT: i32.const $push[[L3:.+]]=, 131072
+; CHECK-NEXT: i32.sub [[SP:.+]]=, $pop[[L2]], $pop[[L3]]
 ; CHECK:      i32.call       ${{[^,]+}}=, memcpy@FUNCTION,
 ; CHECK-NEXT: call           big_byval_callee@FUNCTION,
 %big = type [131072 x i8]
 declare void @big_byval_callee(%big* byval align 1)
 define void @big_byval(%big* byval align 1 %x) {
   call void @big_byval_callee(%big* byval align 1 %x)
   ret void
 }

llvm/trunk/test/CodeGen/WebAssembly/mem-intrinsics.ll

 ; CHECK-NEXT: return{{$}}
 define void @set_no(i8* %dst, i8 %src, i32 %len) {
   call void @llvm.memset.p0i8.i32(i8* %dst, i8 %src, i32 %len, i32 1, i1 false)
   ret void
 }
 
 
 ; CHECK-LABEL: frame_index:
-; CHECK: i32.call $discard=, memset@FUNCTION, $3, $pop1, $pop0{{$}}
-; CHECK: i32.call $discard=, memset@FUNCTION, $4, $pop3, $pop2{{$}}
+; CHECK: i32.call $discard=, memset@FUNCTION, $0, $pop1, $pop0{{$}}
+; CHECK: i32.call $discard=, memset@FUNCTION, $1, $pop3, $pop2{{$}}
 ; CHECK: return{{$}}
 define void @frame_index() {
 entry:
   %a = alloca [2048 x i8], align 16
   %b = alloca [2048 x i8], align 16
   %0 = getelementptr inbounds [2048 x i8], [2048 x i8]* %a, i32 0, i32 0
   %1 = getelementptr inbounds [2048 x i8], [2048 x i8]* %b, i32 0, i32 0
   call void @llvm.memset.p0i8.i32(i8* %0, i8 256, i32 1024, i32 16, i1 false)

llvm/trunk/test/CodeGen/WebAssembly/userstack.ll

 ; RUN: llc < %s -asm-verbose=false | FileCheck %s
 ; RUN: llc < %s -asm-verbose=false -fast-isel | FileCheck %s
 
 
 target datalayout = "e-m:e-p:32:32-i64:64-n32:64-S128"
 target triple = "wasm32-unknown-unknown"
 
 ; CHECK-LABEL: alloca32:
 ; Check that there is an extra local for the stack pointer.
-; CHECK: .local i32, i32, i32, i32{{$}}
+; CHECK: .local i32{{$}}
 define void @alloca32() {
- ; CHECK: i32.const [[L1:.+]]=, __stack_pointer
- ; CHECK-NEXT: i32.load [[L1]]=, 0([[L1]])
- ; CHECK-NEXT: i32.const [[L2:.+]]=, 16
- ; CHECK-NEXT: i32.sub [[SP:.+]]=, [[L1]], [[L2]]
+ ; CHECK: i32.const $push[[L1:.+]]=, __stack_pointer{{$}}
+ ; CHECK-NEXT: i32.load $push[[L2:.+]]=, 0($pop[[L1]])
+ ; CHECK-NEXT: i32.const $push[[L3:.+]]=, 16
+ ; CHECK-NEXT: i32.sub [[SP:.+]]=, $pop[[L2]], $pop[[L3]]
+ ; CHECK-NEXT: i32.const $push[[L4:.+]]=, __stack_pointer{{$}}
+ ; CHECK-NEXT: i32.store [[SP]]=, 0($pop[[L4]]), [[SP]]
  %retval = alloca i32
- ; CHECK: i32.const $push[[L3:.+]]=, 0
- ; CHECK: i32.store {{.*}}=, 12([[SP]]), $pop[[L3]]
+ ; CHECK: i32.const $push[[L0:.+]]=, 0
+ ; CHECK: i32.store {{.*}}=, 12([[SP]]), $pop[[L0]]
  store i32 0, i32* %retval
- ; CHECK: i32.const [[L4:.+]]=, 16
- ; CHECK-NEXT: i32.add [[SP]]=, [[SP]], [[L4]]
- ; CHECK-NEXT: i32.const [[L5:.+]]=, __stack_pointer
- ; CHECK-NEXT: i32.store [[SP]]=, 0([[L5]]), [[SP]]
+ ; CHECK: i32.const $push[[L5:.+]]=, 16
+ ; CHECK-NEXT: i32.add [[SP]]=, [[SP]], $pop[[L5]]
+ ; CHECK-NEXT: i32.const $push[[L6:.+]]=, __stack_pointer
+ ; CHECK-NEXT: i32.store [[SP]]=, 0($pop[[L6]]), [[SP]]
  ret void
 }
 
 ; CHECK-LABEL: alloca3264:
-; CHECK: .local i32, i32, i32, i32{{$}}
+; CHECK: .local i32{{$}}
 define void @alloca3264() {
- ; CHECK: i32.const [[L1:.+]]=, __stack_pointer
- ; CHECK-NEXT: i32.load [[L1]]=, 0([[L1]])
- ; CHECK-NEXT: i32.const [[L2:.+]]=, 16
- ; CHECK-NEXT: i32.sub [[SP:.+]]=, [[L1]], [[L2]]
+ ; CHECK: i32.const $push[[L1:.+]]=, __stack_pointer
+ ; CHECK-NEXT: i32.load $push[[L2:.+]]=, 0($pop[[L1]])
+ ; CHECK-NEXT: i32.const $push[[L3:.+]]=, 16
+ ; CHECK-NEXT: i32.sub [[SP:.+]]=, $pop[[L2]], $pop[[L3]]
+ ; CHECK-NEXT: i32.const $push[[L4:.+]]=, __stack_pointer{{$}}
+ ; CHECK-NEXT: i32.store [[SP]]=, 0($pop[[L4]]), [[SP]]
  %r1 = alloca i32
  %r2 = alloca double
  ; CHECK: i32.const $push[[L3:.+]]=, 0
  ; CHECK: i32.store {{.*}}=, 12([[SP]]), $pop[[L3]]
  store i32 0, i32* %r1
- ; CHECK: i64.const $push[[L4:.+]]=, 0
- ; CHECK: i64.store {{.*}}=, 0([[SP]]), $pop[[L4]]
+ ; CHECK: i64.const $push[[L0:.+]]=, 0
+ ; CHECK: i64.store {{.*}}=, 0([[SP]]), $pop[[L0]]
  store double 0.0, double* %r2
- ; CHECK: i32.const [[L4:.+]]=, 16
- ; CHECK-NEXT: i32.add [[SP]]=, [[SP]], [[L4]]
- ; CHECK-NEXT: i32.const [[L5:.+]]=, __stack_pointer
- ; CHECK-NEXT: i32.store [[SP]]=, 0([[L5]]), [[SP]]
+ ; CHECK: i32.const $push[[L5:.+]]=, 16
+ ; CHECK-NEXT: i32.add [[SP]]=, [[SP]], $pop[[L5]]
+ ; CHECK-NEXT: i32.const $push[[L6:.+]]=, __stack_pointer
+ ; CHECK-NEXT: i32.store [[SP]]=, 0($pop[[L6]]), [[SP]]
  ret void
 }
 
 ; CHECK-LABEL: allocarray:
-; CHECK: .local i32, i32, i32, i32, i32{{$}}
+; CHECK: .local i32, i32{{$}}
 define void @allocarray() {
- ; CHECK-NEXT: i32.const [[L1:.+]]=, __stack_pointer
- ; CHECK-NEXT: i32.load [[L1]]=, 0([[L1]])
- ; CHECK-NEXT: i32.const [[L2:.+]]=, 32{{$}}
- ; CHECK-NEXT: i32.sub [[SP:.+]]=, [[L1]], [[L2]]
- ; CHECK-NEXT: i32.const [[L2]]=, __stack_pointer{{$}}
- ; CHECK-NEXT: i32.store [[SP]]=, 0([[L2]]), [[SP]]
+ ; CHECK: i32.const $push[[L1:.+]]=, __stack_pointer
+ ; CHECK-NEXT: i32.load $push[[L2:.+]]=, 0($pop[[L1]])
+ ; CHECK-NEXT: i32.const $push[[L3:.+]]=, 32{{$}}
+ ; CHECK-NEXT: i32.sub [[SP:.+]]=, $pop[[L2]], $pop[[L3]]
+ ; CHECK-NEXT: i32.const $push[[L4:.+]]=, __stack_pointer{{$}}
+ ; CHECK-NEXT: i32.store [[SP]]=, 0($pop[[L4]]), [[SP]]
  %r = alloca [5 x i32]
 
  ; CHECK-NEXT: i32.const $push[[L4:.+]]=, 12
  ; CHECK-NEXT: i32.const [[L5:.+]]=, 12
  ; CHECK-NEXT: i32.add [[L5]]=, [[SP]], [[L5]]
  ; CHECK-NEXT: i32.add $push[[L6:.+]]=, [[L5]], $pop[[L4]]
  ; CHECK-NEXT: i32.const $push[[L9:.+]]=, 1{{$}}
  ; CHECK-NEXT: i32.store $push[[L10:.+]]=, 12([[SP]]), $pop[[L9]]{{$}}
  ; CHECK-NEXT: i32.store $discard=, 0($pop3), $pop[[L10]]{{$}}
  %p = getelementptr [5 x i32], [5 x i32]* %r, i32 0, i32 0
  store i32 1, i32* %p
  %p2 = getelementptr [5 x i32], [5 x i32]* %r, i32 0, i32 3
  store i32 1, i32* %p2
 
- ; CHECK-NEXT: i32.const [[L7:.+]]=, 32
- ; CHECK-NEXT: i32.add [[SP]]=, [[SP]], [[L7]]
- ; CHECK-NEXT: i32.const [[L8:.+]]=, __stack_pointer
- ; CHECK-NEXT: i32.store [[SP]]=, 0([[L8]]), [[SP]]
+ ; CHECK: i32.const $push[[L11:.+]]=, 32
+ ; CHECK-NEXT: i32.add [[SP]]=, [[SP]], $pop[[L11]]
+ ; CHECK-NEXT: i32.const $push[[L12:.+]]=, __stack_pointer
+ ; CHECK-NEXT: i32.store [[SP]]=, 0($pop[[L12]]), [[SP]]
  ret void
 }
 
 declare void @ext_func(i64* %ptr)
 ; CHECK-LABEL: non_mem_use
 define void @non_mem_use(i8** %addr) {
- ; CHECK: i32.const [[L2:.+]]=, 48
- ; CHECK-NEXT: i32.sub [[SP:.+]]=, {{.+}}, [[L2]]
+ ; CHECK: i32.const $push[[L1:.+]]=, 48
+ ; CHECK-NEXT: i32.sub [[SP:.+]]=, {{.+}}, $pop[[L1]]
  %buf = alloca [27 x i8], align 16
  %r = alloca i64
  %r2 = alloca i64
  ; %r is at SP+8
  ; CHECK: i32.const [[OFF:.+]]=, 8
  ; CHECK-NEXT: i32.add [[ARG1:.+]]=, [[SP]], [[OFF]]
  ; CHECK-NEXT: call ext_func@FUNCTION, [[ARG1]]
  call void @ext_func(i64* %r)
  ; %r2 is at SP+0, no add needed
  ; CHECK-NEXT: call ext_func@FUNCTION, [[SP]]
  call void @ext_func(i64* %r2)
  ; Use as a value, but in a store
  ; %buf is at SP+16
  ; CHECK: i32.const [[OFF:.+]]=, 16
  ; CHECK-NEXT: i32.add [[VAL:.+]]=, [[SP]], [[OFF]]
  ; CHECK-NEXT: i32.store {{.*}}=, 0($0), [[VAL]]
  %gep = getelementptr inbounds [27 x i8], [27 x i8]* %buf, i32 0, i32 0
  store i8* %gep, i8** %addr
  ret void
 }
 
 ; CHECK-LABEL: allocarray_inbounds:
-; CHECK: .local i32, i32, i32, i32{{$}}
+; CHECK: .local i32{{$}}
 define void @allocarray_inbounds() {
- ; CHECK: i32.const [[L1:.+]]=, __stack_pointer
- ; CHECK-NEXT: i32.load [[L1]]=, 0([[L1]])
- ; CHECK-NEXT: i32.const [[L2:.+]]=, 32
- ; CHECK-NEXT: i32.sub [[SP:.+]]=, [[L1]], [[L2]]
+ ; CHECK: i32.const $push[[L1:.+]]=, __stack_pointer
+ ; CHECK-NEXT: i32.load $push[[L2:.+]]=, 0($pop[[L1]])
+ ; CHECK-NEXT: i32.const $push[[L3:.+]]=, 32{{$}}
+ ; CHECK-NEXT: i32.sub [[SP:.+]]=, $pop[[L2]], $pop[[L3]]
  %r = alloca [5 x i32]
  ; CHECK: i32.const $push[[L3:.+]]=, 1
  ; CHECK: i32.store {{.*}}=, 12([[SP]]), $pop[[L3]]
  %p = getelementptr inbounds [5 x i32], [5 x i32]* %r, i32 0, i32 0
  store i32 1, i32* %p
  ; This store should have both the GEP and the FI folded into it.
  ; CHECK-NEXT: i32.store {{.*}}=, 24([[SP]]), $pop
  %p2 = getelementptr inbounds [5 x i32], [5 x i32]* %r, i32 0, i32 3
  store i32 1, i32* %p2
- ; CHECK: i32.const [[L7:.+]]=, 32
- ; CHECK-NEXT: i32.add [[SP]]=, [[SP]], [[L7]]
- ; CHECK-NEXT: i32.const [[L8:.+]]=, __stack_pointer
- ; CHECK-NEXT: i32.store [[SP]]=, 0([[L7]]), [[SP]]
+ ; CHECK: i32.const $push[[L5:.+]]=, 32
+ ; CHECK-NEXT: i32.add [[SP]]=, [[SP]], $pop[[L5]]
+ ; CHECK-NEXT: i32.const $push[[L6:.+]]=, __stack_pointer
+ ; CHECK-NEXT: i32.store [[SP]]=, 0($pop[[L6]]), [[SP]]
  ret void
 }
 
 ; CHECK-LABEL: dynamic_alloca:
 define void @dynamic_alloca(i32 %alloc) {
- ; CHECK: i32.const [[L0:.+]]=, __stack_pointer
- ; CHECK-NEXT: i32.load [[SP:.+]]=, 0([[L0]])
+ ; CHECK: i32.const $push[[L1:.+]]=, __stack_pointer
+ ; CHECK-NEXT: i32.load [[SP:.+]]=, 0($pop[[L1]])
  ; CHECK-NEXT: copy_local [[FP:.+]]=, [[SP]]
  ; Target independent codegen bumps the stack pointer
  ; FIXME: we need to write the value back to memory
  %r = alloca i32, i32 %alloc
  ; Target-independent codegen also calculates the store addr
  store i32 0, i32* %r
- ; CHECK: i32.const [[L3:.+]]=, __stack_pointer
- ; CHECK-NEXT: i32.store [[SP]]=, 0([[L3]]), [[FP]]
+ ; CHECK: i32.const $push[[L3:.+]]=, __stack_pointer
+ ; CHECK-NEXT: i32.store [[SP]]=, 0($pop[[L3]]), [[FP]]
  ret void
 }
 
 
 ; CHECK-LABEL: dynamic_static_alloca:
 define void @dynamic_static_alloca(i32 %alloc) {
- ; CHECK: i32.const [[L0:.+]]=, __stack_pointer
- ; CHECK-NEXT: i32.load [[L0]]=, 0([[L0]])
- ; CHECK-NEXT: i32.const [[L2:.+]]=, 16
- ; CHECK-NEXT: i32.sub [[SP:.+]]=, [[L0]], [[L2]]
+ ; CHECK: i32.const $push[[L1:.+]]=, __stack_pointer
+ ; CHECK-NEXT: i32.load $push[[L2:.+]]=, 0($pop[[L1]])
+ ; CHECK-NEXT: i32.const $push[[L3:.+]]=, 16
+ ; CHECK-NEXT: i32.sub [[SP:.+]]=, $pop[[L2]], $pop[[L3]]
  ; CHECK-NEXT: copy_local [[FP:.+]]=, [[SP]]
- ; CHECK-NEXT: i32.const [[L3:.+]]=, __stack_pointer
- ; CHECK-NEXT: i32.store {{.*}}=, 0([[L3]]), [[SP]]
+ ; CHECK-NEXT: i32.const $push[[L4:.+]]=, __stack_pointer
+ ; CHECK-NEXT: i32.store {{.*}}=, 0($pop[[L4]]), [[SP]]
  %r1 = alloca i32
  %r = alloca i32, i32 %alloc
  store i32 0, i32* %r
- ; CHECK: i32.const [[L3:.+]]=, 16
- ; CHECK: i32.add [[SP]]=, [[FP]], [[L3]]
- ; CHECK: i32.const [[L4:.+]]=, __stack_pointer
- ; CHECK-NEXT: i32.store [[SP]]=, 0([[L4]]), [[SP]]
+ ; CHECK: i32.const $push[[L5:.+]]=, 16
+ ; CHECK-NEXT: i32.add [[SP]]=, [[FP]], $pop[[L5]]
+ ; CHECK-NEXT: i32.const $push[[L6:.+]]=, __stack_pointer
+ ; CHECK-NEXT: i32.store [[SP]]=, 0($pop[[L6]]), [[SP]]
  ret void
 }
 
 ; The use of the alloca in a phi causes a CopyToReg DAG node to be generated,
 ; which has to have special handling because CopyToReg can't have a FI operand
 ; CHECK-LABEL: copytoreg_fi:
 define void @copytoreg_fi(i1 %cond, i32* %b) {
 entry:
- ; CHECK: i32.const [[L2:.+]]=, 16
- ; CHECK-NEXT: i32.sub [[SP:.+]]=, {{.+}}, [[L2]]
+ ; CHECK: i32.const $push[[L1:.+]]=, 16
+ ; CHECK-NEXT: i32.sub [[SP:.+]]=, {{.+}}, $pop[[L1]]
  %addr = alloca i32
  ; CHECK: i32.const [[OFF:.+]]=, 12
  ; CHECK-NEXT: i32.add [[ADDR:.+]]=, [[SP]], [[OFF]]
  ; CHECK-NEXT: copy_local [[COPY:.+]]=, [[ADDR]]
  br label %body
 body:
  %a = phi i32* [%addr, %entry], [%b, %body]
  store i32 1, i32* %a
  ; CHECK: i32.store {{.*}}, 0([[COPY]]),
  br i1 %cond, label %body, label %exit
 exit:
  ret void
 }
 
 declare void @use_i8_star(i8*)
 declare i8* @llvm.frameaddress(i32)
 
 ; Test __builtin_frame_address(0).
-; TODO: When the prolog/epilog sequences are optimized, refine these checks to
-; be more specific.
-
 ; CHECK-LABEL: frameaddress_0:
-; CHECK: __stack_pointer
-; CHECK: load
-; CHECK: call use_i8_star
-; CHECK: __stack_pointer
-; CHECK: store
+; CHECK: i32.const $push[[L1:.+]]=, __stack_pointer
+; CHECK-NEXT: i32.load [[SP:.+]]=, 0($pop[[L1]])
+; CHECK-NEXT: copy_local [[FP:.+]]=, [[SP]]
+; CHECK-NEXT: call use_i8_star@FUNCTION, [[FP]]
+; CHECK-NEXT: i32.const $push[[L6:.+]]=, __stack_pointer
+; CHECK-NEXT: i32.store [[SP]]=, 0($pop[[L6]]), [[FP]]
 define void @frameaddress_0() {
   %t = call i8* @llvm.frameaddress(i32 0)
   call void @use_i8_star(i8* %t)
   ret void
 }
 
 ; Test __builtin_frame_address(1).