[RISCV] When matching SROIW, check all 64 bits of the OR mask

[craig.topper]

Summary

We need to make sure the upper 32 bits are all ones to ensure the result is properly sign extended. Previously we only checked the lower 32 bits of the mask. I've also added a check that the shift amount is less than 32. Without that the original code asserts if the SROI check is removed or the SROIW pattern is checked first. I've refactored the code to use early outs to reduce nesting.

I've also updated SLOIW matching with the same changes, but I couldn't find a broken test case with the existing code.

[craig.topper]

Update SelectSLOIW in the same way. Check all bits of the mask and check the shift amount is less than 32. I haven't been able to produce any incorrect tests for sloiw though.

[craig.topper]

Ping

[frasercrmck]

Apart from my nits, this makes sense to me. The only thing I don't fully get it where/why the original code asserted if the shift amount was >= 32.

Regarding SLOIW I'm not sure you'll find a test case where the mask isn't wholly contained in the 32 lower bits since the or and the shl are i32 operations. But I don't think it hurts to check all 64 bits.

llvm/lib/Target/RISCV/RISCVISelDAGToDAG.cpp

317

Your changes to SROIW also mention the constraint VC2 < 32 so I reckon it should be stated here too

341

Comment here should say "SLOIW"

[craig.topper]

In D90961#2397314, @frasercrmck wrote:

Apart from my nits, this makes sense to me. The only thing I don't fully get it where/why the original code asserted if the shift amount was >= 32.

The assert is inside maskLeadingOnes<uint32_t>. There's a check that the bit count passed in is <= that size of the type passed as the template parameter.

Regarding SLOIW I'm not sure you'll find a test case where the mask isn't wholly contained in the 32 lower bits since the or and the shl are i32 operations. But I don't think it hurts to check all 64 bits.

llvm/lib/Target/RISCV/RISCVISelDAGToDAG.cpp

 // Check that it is a SLOIW (Shift Left Ones Immediate i32 on RV64).
 // We first check that it is the right node tree:
 //
 //  (SIGN_EXTEND_INREG (OR (SHL RS1, VC2), VC1))
 //
 // and then we check that VC1, the mask used to fill with ones, is compatible
 // with VC2, the shamt:
 //
 //  VC1 == maskTrailingOnes<uint32_t>(VC2)
 
 bool RISCVDAGToDAGISel::SelectSLOIW(SDValue N, SDValue &RS1, SDValue &Shamt) {
   if (Subtarget->getXLenVT() == MVT::i64 &&
       N.getOpcode() == ISD::SIGN_EXTEND_INREG &&
       cast<VTSDNode>(N.getOperand(1))->getVT() == MVT::i32) {
     if (N.getOperand(0).getOpcode() == ISD::OR) {
       SDValue Or = N.getOperand(0);
       if (Or.getOperand(0).getOpcode() == ISD::SHL) {
         SDValue Shl = Or.getOperand(0);
         if (isa<ConstantSDNode>(Shl.getOperand(1)) &&
             isa<ConstantSDNode>(Or.getOperand(1))) {
           uint32_t VC1 = Or.getConstantOperandVal(1);
           uint32_t VC2 = Shl.getConstantOperandVal(1);
           if (VC1 == maskTrailingOnes<uint32_t>(VC2)) {
             RS1 = Shl.getOperand(0);
             Shamt = CurDAG->getTargetConstant(VC2, SDLoc(N),
                                               Shl.getOperand(1).getValueType());
             return true;
           }
         }
       }
     }
   }
   return false;
 }
 
 // Check that it is a SROIW (Shift Right Ones Immediate i32 on RV64).
 // We first check that it is the right node tree:
 //
-//  (OR (SHL RS1, VC2), VC1)
+//  (OR (SRL RS1, VC2), VC1)
 //
 // and then we check that VC1, the mask used to fill with ones, is compatible
 // with VC2, the shamt:
 //
-//  VC1 == maskLeadingOnes<uint32_t>(VC2)
-
+//  VC2 < 32
+//  VC1 == maskTrailingZeros<uint64_t>(32 - VC2)
+//
 bool RISCVDAGToDAGISel::SelectSROIW(SDValue N, SDValue &RS1, SDValue &Shamt) {
-  if (N.getOpcode() == ISD::OR && Subtarget->getXLenVT() == MVT::i64) {
-    SDValue Or = N;
-    if (Or.getOperand(0).getOpcode() == ISD::SRL) {
-      SDValue Srl = Or.getOperand(0);
-      if (isa<ConstantSDNode>(Srl.getOperand(1)) &&
-          isa<ConstantSDNode>(Or.getOperand(1))) {
-        uint32_t VC1 = Or.getConstantOperandVal(1);
-        uint32_t VC2 = Srl.getConstantOperandVal(1);
-        if (VC1 == maskLeadingOnes<uint32_t>(VC2)) {
-          RS1 = Srl.getOperand(0);
-          Shamt = CurDAG->getTargetConstant(VC2, SDLoc(N),
-                                            Srl.getOperand(1).getValueType());
-          return true;
-        }
-      }
-    }
-  }
-  return false;
-}
+  assert(Subtarget->is64Bit() && "SROIW should only be matched on RV64");
+  if (N.getOpcode() != ISD::OR || !isa<ConstantSDNode>(N.getOperand(1)))
+    return false;
+
+  SDValue Srl = N.getOperand(0);
+  if (Srl.getOpcode() != ISD::SRL || !isa<ConstantSDNode>(Srl.getOperand(1)))
+    return false;
+
+  uint64_t VC1 = N.getConstantOperandVal(1);
+  uint64_t VC2 = Srl.getConstantOperandVal(1);
+
+  if (VC2 >= 32 || VC1 != maskTrailingZeros<uint64_t>(32 - VC2))
+    return false;
+
 
 // Check that it is a RORIW (i32 Right Rotate Immediate on RV64).
 // We first check that it is the right node tree:
 //
 //  (SIGN_EXTEND_INREG (OR (SHL RS1, VC2),
 //                         (SRL (AND RS1, VC3), VC1)))
 //
 // Then we check that the constant operands respect these constraints:

llvm/test/CodeGen/RISCV/rv64Zbb.ll

; RV64IBB-NEXT:    ret
   %neg = lshr i32 %a, 1
   %neg12 = or i32 %neg, -2147483648
   ret i32 %neg12
 }
 
 ; This is similar to the type legalized version of sroiw but the mask is 0 in
 ; the upper bits instead of 1 so the result is not sign extended. Make sure we
 ; don't match it to sroiw.
-; FIXME: We're matching it to sroiw.
 define i64 @sroiw_bug(i64 %a) nounwind {
 ; RV64I-LABEL: sroiw_bug:
 ; RV64I:       # %bb.0:
 ; RV64I-NEXT:    srli a0, a0, 1
 ; RV64I-NEXT:    addi a1, zero, 1
 ; RV64I-NEXT:    slli a1, a1, 31
 ; RV64I-NEXT:    or a0, a0, a1
 ; RV64I-NEXT:    ret
 ;
 ; RV64IB-LABEL: sroiw_bug:
 ; RV64IB:       # %bb.0:
-; RV64IB-NEXT:    sroiw a0, a0, 1
+; RV64IB-NEXT:    srli a0, a0, 1
+; RV64IB-NEXT:    addi a1, zero, 1
+; RV64IB-NEXT:    slli a1, a1, 31
+; RV64IB-NEXT:    or a0, a0, a1
 ; RV64IB-NEXT:    ret
 ;
 ; RV64IBB-LABEL: sroiw_bug:
 ; RV64IBB:       # %bb.0:
-; RV64IBB-NEXT:    sroiw a0, a0, 1
+; RV64IBB-NEXT:    srli a0, a0, 1
+; RV64IBB-NEXT:    addi a1, zero, 1
+; RV64IBB-NEXT:    slli a1, a1, 31
+; RV64IBB-NEXT:    or a0, a0, a1
 ; RV64IBB-NEXT:    ret
   %neg = lshr i64 %a, 1
   %neg12 = or i64 %neg, 2147483648
   ret i64 %neg12
 }
 
 define i64 @sroi_i64(i64 %a) nounwind {
 ; RV64I-LABEL: sroi_i64: