Merging r196046:

------------------------------------------------------------------------
r196046 | tnorthover | 2013-12-01 06:16:24 -0800 (Sun, 01 Dec 2013) | 8 lines

ARM: fix bug in -Oz stack adjustment folding

Previously, we clobbered callee-saved registers when folding an "add
sp, #N" into a "pop {rD, ...}" instruction. This change checks whether
a register we're going to add to the "pop" could actually be live
outside the function before doing so and should fix the issue.

This should fix PR18081.
------------------------------------------------------------------------

git-svn-id: https://llvm.org/svn/llvm-project/llvm/branches/release_34@196074 91177308-0d34-0410-b5e6-96231b3b80d8

diff --git a/lib/Target/ARM/ARMBaseInstrInfo.cpp b/lib/Target/ARM/ARMBaseInstrInfo.cpp
index 67534c4a44570ef37e41e9ed981136a2c8566655..df8c017515d977c141ec10202299aeeebeb5feca 100644 (file)
--- a/lib/Target/ARM/ARMBaseInstrInfo.cpp
+++ b/lib/Target/ARM/ARMBaseInstrInfo.cpp
@@ -1913,29 +1913,40 @@ bool llvm::tryFoldSPUpdateIntoPushPop(MachineFunction &MF,
 
   MachineBasicBlock *MBB = MI->getParent();
   const TargetRegisterInfo *TRI = MF.getRegInfo().getTargetRegisterInfo();
+  const MCPhysReg *CSRegs = TRI->getCalleeSavedRegs(&MF);
 
   // Now try to find enough space in the reglist to allocate NumBytes.
   for (unsigned CurReg = FirstReg - 1; CurReg >= RD0Reg && RegsNeeded;
-       --CurReg, --RegsNeeded) {
+       --CurReg) {
     if (!IsPop) {
       // Pushing any register is completely harmless, mark the
       // register involved as undef since we don't care about it in
       // the slightest.
       RegList.push_back(MachineOperand::CreateReg(CurReg, false, false,
                                                   false, false, true));
+      --RegsNeeded;
       continue;
     }
 
-    // However, we can only pop an extra register if it's not live. Otherwise we
-    // might clobber a return value register. We assume that once we find a live
-    // return register all lower ones will be too so there's no use proceeding.
-    if (MBB->computeRegisterLiveness(TRI, CurReg, MI) !=
-        MachineBasicBlock::LQR_Dead)
-      return false;
+    // However, we can only pop an extra register if it's not live. For
+    // registers live within the function we might clobber a return value
+    // register; the other way a register can be live here is if it's
+    // callee-saved.
+    if (isCalleeSavedRegister(CurReg, CSRegs) ||
+        MBB->computeRegisterLiveness(TRI, CurReg, MI) !=
+            MachineBasicBlock::LQR_Dead) {
+      // VFP pops don't allow holes in the register list, so any skip is fatal
+      // for our transformation. GPR pops do, so we should just keep looking.
+      if (IsVFPPushPop)
+        return false;
+      else
+        continue;
+    }
 
     // Mark the unimportant registers as <def,dead> in the POP.
     RegList.push_back(MachineOperand::CreateReg(CurReg, true, false, false,
                                                 true));
+    --RegsNeeded;
   }
 
   if (RegsNeeded > 0)

diff --git a/lib/Target/ARM/ARMBaseRegisterInfo.h b/lib/Target/ARM/ARMBaseRegisterInfo.h
index 0d4f54f89179f5391b3b4025b7e0029458377f4e..e28fff68f4e2cdf49a7643cc64f8154cb5db85fc 100644 (file)
--- a/lib/Target/ARM/ARMBaseRegisterInfo.h
+++ b/lib/Target/ARM/ARMBaseRegisterInfo.h
@@ -72,6 +72,14 @@ static inline bool isARMArea3Register(unsigned Reg, bool isIOS) {
   }
 }
 
+static inline bool isCalleeSavedRegister(unsigned Reg,
+                                         const MCPhysReg *CSRegs) {
+  for (unsigned i = 0; CSRegs[i]; ++i)
+    if (Reg == CSRegs[i])
+      return true;
+  return false;
+}
+
 class ARMBaseRegisterInfo : public ARMGenRegisterInfo {
 protected:
   const ARMSubtarget &STI;

diff --git a/lib/Target/ARM/ARMFrameLowering.cpp b/lib/Target/ARM/ARMFrameLowering.cpp
index 7b02803c51fc3ae9823643e8ed1818330d24d7fd..3e72d3690aa59f4d8ba9cfac17adbb80ee203ee5 100644 (file)
--- a/lib/Target/ARM/ARMFrameLowering.cpp
+++ b/lib/Target/ARM/ARMFrameLowering.cpp
@@ -82,13 +82,6 @@ ARMFrameLowering::canSimplifyCallFramePseudos(const MachineFunction &MF) const {
   return hasReservedCallFrame(MF) || MF.getFrameInfo()->hasVarSizedObjects();
 }
 
-static bool isCalleeSavedRegister(unsigned Reg, const uint16_t *CSRegs) {
-  for (unsigned i = 0; CSRegs[i]; ++i)
-    if (Reg == CSRegs[i])
-      return true;
-  return false;
-}
-
 static bool isCSRestore(MachineInstr *MI,
                         const ARMBaseInstrInfo &TII,
                         const uint16_t *CSRegs) {

diff --git a/lib/Target/ARM/Thumb1FrameLowering.cpp b/lib/Target/ARM/Thumb1FrameLowering.cpp
index d921c82cfb091f0c48046db79275f8f65e766770..cfb33f5b8212ee5bb985015ef8869df7639b8876 100644 (file)
--- a/lib/Target/ARM/Thumb1FrameLowering.cpp
+++ b/lib/Target/ARM/Thumb1FrameLowering.cpp
@@ -215,13 +215,6 @@ void Thumb1FrameLowering::emitPrologue(MachineFunction &MF) const {
     AFI->setShouldRestoreSPFromFP(true);
 }
 
-static bool isCalleeSavedRegister(unsigned Reg, const uint16_t *CSRegs) {
-  for (unsigned i = 0; CSRegs[i]; ++i)
-    if (Reg == CSRegs[i])
-      return true;
-  return false;
-}
-
 static bool isCSRestore(MachineInstr *MI, const uint16_t *CSRegs) {
   if (MI->getOpcode() == ARM::tLDRspi &&
       MI->getOperand(1).isFI() &&

diff --git a/test/CodeGen/ARM/fold-stack-adjust.ll b/test/CodeGen/ARM/fold-stack-adjust.ll
index c8c48faffb2d5b20eb2b69257831af8b6445217d..8bda7683f902f1be1f7498ada2bb0c810a2a98b6 100644 (file)
--- a/test/CodeGen/ARM/fold-stack-adjust.ll
+++ b/test/CodeGen/ARM/fold-stack-adjust.ll
@@ -15,7 +15,7 @@ define void @check_simple() minsize {
 ; CHECK-NOT: sub sp, sp,
 ; ...
 ; CHECK-NOT: add sp, sp,
-; CHECK: pop.w {r7, r8, r9, r10, r11, pc}
+; CHECK: pop.w {r0, r1, r2, r3, r11, pc}
 
 ; CHECK-T1-LABEL: check_simple:
 ; CHECK-T1: push {r3, r4, r5, r6, r7, lr}
@@ -23,7 +23,7 @@ define void @check_simple() minsize {
 ; CHECK-T1-NOT: sub sp, sp,
 ; ...
 ; CHECK-T1-NOT: add sp, sp,
-; CHECK-T1: pop {r3, r4, r5, r6, r7, pc}
+; CHECK-T1: pop {r0, r1, r2, r3, r7, pc}
 
   ; iOS always has a frame pointer and messing with the push affects
   ; how it's set in the prologue. Make sure we get that right.