Recommit r231324 with a fix to the ARM execution domain code

to disable lane switching if we don't actually have the instruction
set we want to switch to. Models the earlier check above the
conditional for the pass.

The testcase is one that triggered with the assert that's added
as part of the fix, use it to avoid adding a new testcase as it
highlights the same problem.

llvm-svn: 231539

diff --git a/llvm/lib/Target/ARM/ARMBaseInstrInfo.cpp b/llvm/lib/Target/ARM/ARMBaseInstrInfo.cpp
index 29ee22e..42e37fe 100644 (file)
--- a/llvm/lib/Target/ARM/ARMBaseInstrInfo.cpp
+++ b/llvm/lib/Target/ARM/ARMBaseInstrInfo.cpp
@@ -4115,19 +4115,21 @@ enum ARMExeDomain {
 //
 std::pair<uint16_t, uint16_t>
 ARMBaseInstrInfo::getExecutionDomain(const MachineInstr *MI) const {
-  // VMOVD, VMOVRS and VMOVSR are VFP instructions, but can be changed to NEON
-  // if they are not predicated.
-  if (MI->getOpcode() == ARM::VMOVD && !isPredicated(MI))
-    return std::make_pair(ExeVFP, (1<<ExeVFP) | (1<<ExeNEON));
-
-  // CortexA9 is particularly picky about mixing the two and wants these
-  // converted.
-  if (Subtarget.isCortexA9() && !isPredicated(MI) &&
-      (MI->getOpcode() == ARM::VMOVRS ||
-       MI->getOpcode() == ARM::VMOVSR ||
-       MI->getOpcode() == ARM::VMOVS))
-    return std::make_pair(ExeVFP, (1<<ExeVFP) | (1<<ExeNEON));
-
+  // If we don't have access to NEON instructions then we won't be able
+  // to swizzle anything to the NEON domain. Check to make sure.
+  if (Subtarget.hasNEON()) {
+    // VMOVD, VMOVRS and VMOVSR are VFP instructions, but can be changed to NEON
+    // if they are not predicated.
+    if (MI->getOpcode() == ARM::VMOVD && !isPredicated(MI))
+      return std::make_pair(ExeVFP, (1 << ExeVFP) | (1 << ExeNEON));
+
+    // CortexA9 is particularly picky about mixing the two and wants these
+    // converted.
+    if (Subtarget.isCortexA9() && !isPredicated(MI) &&
+        (MI->getOpcode() == ARM::VMOVRS || MI->getOpcode() == ARM::VMOVSR ||
+         MI->getOpcode() == ARM::VMOVS))
+      return std::make_pair(ExeVFP, (1 << ExeVFP) | (1 << ExeNEON));
+  }
   // No other instructions can be swizzled, so just determine their domain.
   unsigned Domain = MI->getDesc().TSFlags & ARMII::DomainMask;
 
@@ -4220,6 +4222,9 @@ ARMBaseInstrInfo::setExecutionDomain(MachineInstr *MI, unsigned Domain) const {
       // Zap the predicate operands.
       assert(!isPredicated(MI) && "Cannot predicate a VORRd");
 
+      // Make sure we've got NEON instructions.
+      assert(Subtarget.hasNEON() && "VORRd requires NEON");
+
       // Source instruction is %DDst = VMOVD %DSrc, 14, %noreg (; implicits)
       DstReg = MI->getOperand(0).getReg();
       SrcReg = MI->getOperand(1).getReg();

diff --git a/llvm/lib/Target/ARM/ARMTargetMachine.cpp b/llvm/lib/Target/ARM/ARMTargetMachine.cpp
index 34f7f78..e3e6228 100644 (file)
--- a/llvm/lib/Target/ARM/ARMTargetMachine.cpp
+++ b/llvm/lib/Target/ARM/ARMTargetMachine.cpp
@@ -358,9 +358,7 @@ void ARMPassConfig::addPreRegAlloc() {
 void ARMPassConfig::addPreSched2() {
   if (getOptLevel() != CodeGenOpt::None) {
     addPass(createARMLoadStoreOptimizationPass());
-
-    if (getARMSubtarget().hasNEON())
-      addPass(createExecutionDependencyFixPass(&ARM::DPRRegClass));
+    addPass(createExecutionDependencyFixPass(&ARM::DPRRegClass));
   }
 
   // Expand some pseudo instructions into multiple instructions to allow

diff --git a/llvm/test/CodeGen/ARM/2007-01-19-InfiniteLoop.ll b/llvm/test/CodeGen/ARM/2007-01-19-InfiniteLoop.ll
index 8b94b7b..4ba81e0 100644 (file)
--- a/llvm/test/CodeGen/ARM/2007-01-19-InfiniteLoop.ll
+++ b/llvm/test/CodeGen/ARM/2007-01-19-InfiniteLoop.ll
@@ -1,5 +1,9 @@
 ; RUN: llc < %s -mtriple=arm-apple-ios -mattr=+v6,+vfp2 | FileCheck %s
+; RUN: llc < %s -mtriple=arm-apple-ios -mattr=+v6,+vfp2 | FileCheck --check-prefix=DOMAIN %s
 
+; The execution domain checking code would translate vmovs to vorr whether or not
+; we had NEON instructions. Verify we don't if we're not compiled with NEON.
+; DOMAIN-NOT: vorr
 @quant_coef = external global [6 x [4 x [4 x i32]]]            ; <[6 x [4 x [4 x i32]]]*> [#uses=1]
 @dequant_coef = external global [6 x [4 x [4 x i32]]]          ; <[6 x [4 x [4 x i32]]]*> [#uses=1]
 @A = external global [4 x [4 x i32]]           ; <[4 x [4 x i32]]*> [#uses=1]