release/19.x: [RISCV] Don't create BuildPairF64 or SplitF64 nodes without D or Zdinx. (#116159)

[topperc]

The fix in ReplaceNodeResults is the only one really required for the known crash.

I couldn't hit the case in LowerOperation because that requires (f64 (bitcast i64)), but the result type is softened before the input so we don't get a chance to legalize the input.

The change to the setOperationAction call was an observation that a i64<->vector cast should not be custom legalized on RV32. The custom code already calls isTypeLegal on the scalar type.

[llvmbot]

@llvm/pr-subscribers-backend-risc-v

Author: Craig Topper (topperc)

Changes

The fix in ReplaceNodeResults is the only one really required for the known crash.

I couldn't hit the case in LowerOperation because that requires (f64 (bitcast i64)), but the result type is softened before the input so we don't get a chance to legalize the input.

The change to the setOperationAction call was an observation that a i64<->vector cast should not be custom legalized on RV32. The custom code already calls isTypeLegal on the scalar type.

---

Full diff: https://github.com/llvm/llvm-project/pull/121501.diff

2 Files Affected:

• (modified) llvm/lib/Target/RISCV/RISCVISelLowering.cpp (+7-4)
• (added) llvm/test/CodeGen/RISCV/rvv/rv32-zve-bitcast-crash.ll (+22)

diff --git a/llvm/lib/Target/RISCV/RISCVISelLowering.cpp b/llvm/lib/Target/RISCV/RISCVISelLowering.cpp
index 823fb428472ef3..badbb425997447 100644
--- a/llvm/lib/Target/RISCV/RISCVISelLowering.cpp
+++ b/llvm/lib/Target/RISCV/RISCVISelLowering.cpp
@@ -1396,8 +1396,9 @@ RISCVTargetLowering::RISCVTargetLowering(const TargetMachine &TM,
       }
 
       // Custom-legalize bitcasts from fixed-length vectors to scalar types.
-      setOperationAction(ISD::BITCAST, {MVT::i8, MVT::i16, MVT::i32, MVT::i64},
-                         Custom);
+      setOperationAction(ISD::BITCAST, {MVT::i8, MVT::i16, MVT::i32}, Custom);
+      if (Subtarget.is64Bit())
+        setOperationAction(ISD::BITCAST, MVT::i64, Custom);
       if (Subtarget.hasStdExtZfhminOrZhinxmin())
         setOperationAction(ISD::BITCAST, MVT::f16, Custom);
       if (Subtarget.hasStdExtFOrZfinx())
@@ -6317,7 +6318,8 @@ SDValue RISCVTargetLowering::LowerOperation(SDValue Op,
           DAG.getNode(RISCVISD::FMV_W_X_RV64, DL, MVT::f32, NewOp0);
       return FPConv;
     }
-    if (VT == MVT::f64 && Op0VT == MVT::i64 && XLenVT == MVT::i32) {
+    if (VT == MVT::f64 && Op0VT == MVT::i64 && !Subtarget.is64Bit() &&
+        Subtarget.hasStdExtDOrZdinx()) {
       SDValue Lo, Hi;
       std::tie(Lo, Hi) = DAG.SplitScalar(Op0, DL, MVT::i32, MVT::i32);
       SDValue RetReg =
@@ -12616,7 +12618,8 @@ void RISCVTargetLowering::ReplaceNodeResults(SDNode *N,
       SDValue FPConv =
           DAG.getNode(RISCVISD::FMV_X_ANYEXTW_RV64, DL, MVT::i64, Op0);
       Results.push_back(DAG.getNode(ISD::TRUNCATE, DL, MVT::i32, FPConv));
-    } else if (VT == MVT::i64 && Op0VT == MVT::f64 && XLenVT == MVT::i32) {
+    } else if (VT == MVT::i64 && Op0VT == MVT::f64 && !Subtarget.is64Bit() &&
+               Subtarget.hasStdExtDOrZdinx()) {
       SDValue NewReg = DAG.getNode(RISCVISD::SplitF64, DL,
                                    DAG.getVTList(MVT::i32, MVT::i32), Op0);
       SDValue RetReg = DAG.getNode(ISD::BUILD_PAIR, DL, MVT::i64,
diff --git a/llvm/test/CodeGen/RISCV/rvv/rv32-zve-bitcast-crash.ll b/llvm/test/CodeGen/RISCV/rvv/rv32-zve-bitcast-crash.ll
new file mode 100644
index 00000000000000..d6612c9d025afa
--- /dev/null
+++ b/llvm/test/CodeGen/RISCV/rvv/rv32-zve-bitcast-crash.ll
@@ -0,0 +1,22 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 4
+; RUN: llc < %s -mtriple=riscv32 -mattr=+zve32f,+zvl128b | FileCheck %s
+
+; This bitcast previously incorrectly produce a SplitF64 node.
+
+define i64 @foo(double %x) {
+; CHECK-LABEL: foo:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    addi sp, sp, -16
+; CHECK-NEXT:    .cfi_def_cfa_offset 16
+; CHECK-NEXT:    sw ra, 12(sp) # 4-byte Folded Spill
+; CHECK-NEXT:    .cfi_offset ra, -4
+; CHECK-NEXT:    lui a3, 261888
+; CHECK-NEXT:    li a2, 0
+; CHECK-NEXT:    call __adddf3
+; CHECK-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
+; CHECK-NEXT:    addi sp, sp, 16
+; CHECK-NEXT:    ret
+  %a = fadd double %x, 1.0
+  %b = bitcast double %a to i64
+  ret i64 %b
+}

[lenary]

LGTM

I guess this would have been easier had the original test had nounwind, but that's life. I don't think we should change it on the branch.

[github-actions]

@topperc (or anyone else). If you would like to add a note about this fix in the release notes (completely optional). Please reply to this comment with a one or two sentence description of the fix. When you are done, please add the release:note label to this PR.