[Mips] Fix missing masking in fast-isel of br (PR40325)

Fixes https://bugs.llvm.org/show_bug.cgi?id=40325 by zero extending
(and x, 1) the condition before branching on it.

To avoid regressing trivial cases, I'm combining emission of cmp+br
sequences for the single-use + same block case (similar to what we
do in x86). icmpbr1.ll still regresses due to the cross-bb usage
of the condition.

Differential Revision: https://reviews.llvm.org/D58576

llvm-svn: 354808

llvm/lib/Target/Mips/MipsFastISel.cpp

@@ -953,21 +953,34 @@ bool MipsFastISel::selectBranch(const Instruction *I) {
   //
   MachineBasicBlock *TBB = FuncInfo.MBBMap[BI->getSuccessor(0)];
   MachineBasicBlock *FBB = FuncInfo.MBBMap[BI->getSuccessor(1)];
-  // For now, just try the simplest case where it's fed by a compare.
+
+  // Fold the common case of a conditional branch with a comparison
+  // in the same block.
+  unsigned ZExtCondReg = 0;
   if (const CmpInst *CI = dyn_cast<CmpInst>(BI->getCondition())) {
-    MVT CIMVT =
-        TLI.getValueType(DL, CI->getOperand(0)->getType(), true).getSimpleVT();
-    if (CIMVT == MVT::i1)
+    if (CI->hasOneUse() && CI->getParent() == I->getParent()) {
+      ZExtCondReg = createResultReg(&Mips::GPR32RegClass);
+      if (!emitCmp(ZExtCondReg, CI))
+        return false;
+    }
+  }
+
+  // For the general case, we need to mask with 1.
+  if (ZExtCondReg == 0) {
+    unsigned CondReg = getRegForValue(BI->getCondition());
+    if (CondReg == 0)
       return false;
 
-    unsigned CondReg = getRegForValue(CI);
-    BuildMI(*BrBB, FuncInfo.InsertPt, DbgLoc, TII.get(Mips::BGTZ))
-        .addReg(CondReg)
-        .addMBB(TBB);
-    finishCondBranch(BI->getParent(), TBB, FBB);
-    return true;
+    ZExtCondReg = emitIntExt(MVT::i1, CondReg, MVT::i32, true);
+    if (ZExtCondReg == 0)
+      return false;
   }
-  return false;
+
+  BuildMI(*BrBB, FuncInfo.InsertPt, DbgLoc, TII.get(Mips::BGTZ))
+      .addReg(ZExtCondReg)
+      .addMBB(TBB);
+  finishCondBranch(BI->getParent(), TBB, FBB);
+  return true;
 }
 
 bool MipsFastISel::selectCmp(const Instruction *I) {

llvm/test/CodeGen/Mips/Fast-ISel/icmpbr1.ll

@@ -17,7 +17,8 @@ bb0:
 bb1:
 ; CHECK:       # %bb.1: # %bb1
 ; CHECK-NEXT:    lw $[[REG2:[0-9]+]], [[SPILL]]($sp) # 4-byte Folded Reload
-; CHECK-NEXT:    bgtz $[[REG2]], $BB0_3
+; CHECK-NEXT:    andi $[[REG3:[0-9]+]], $[[REG2]], 1
+; CHECK-NEXT:    bgtz $[[REG3]], $BB0_3
   br i1 %2, label %bb2, label %bb3
 bb2:
 ; CHECK:         $BB0_3: # %bb2

llvm/test/CodeGen/Mips/Fast-ISel/pr40325.ll

@@ -0,0 +1,23 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc -mtriple=mipsel -relocation-model=pic -O0 -mcpu=mips32 < %s | FileCheck %s
+
+define void @test(i32 %x, i1* %p) nounwind {
+; CHECK-LABEL: test:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    move $1, $4
+; CHECK-NEXT:    andi $4, $4, 1
+; CHECK-NEXT:    sb $4, 0($5)
+; CHECK-NEXT:    andi $1, $1, 1
+; CHECK-NEXT:    bgtz $1, $BB0_1
+; CHECK-NEXT:    nop
+; CHECK-NEXT:  # %bb.1: # %foo
+; CHECK-NEXT:    jr $ra
+; CHECK-NEXT:    nop
+  %y = and i32 %x, 1
+  %c = icmp eq i32 %y, 1
+  store i1 %c, i1* %p
+  br i1 %c, label %foo, label %foo
+
+foo:
+  ret void
+}