[LegalizeTypes][RISCV] Correctly sign-extend comparison for ATOMIC_CMP_XCHG

Currently, the comparison argument used for ATOMIC_CMP_XCHG is legalised
with GetPromotedInteger, which leaves the upper bits of the value
undefind. Since this is used for comparing in an LR/SC loop with a
full-width comparison, we must sign extend it on RISC-V.

This is related to https://reviews.llvm.org/D58829, which solved the
issue for ATOMIC_CMP_SWAP_WITH_SUCCESS, but not the simpler
ATOMIC_CMP_SWAP.

This patch is a modified form of
616289e by Jessica Clarke. It localises
the changes to LegalizeIntegerTypes and avoids adding a new virtual
method to TargetLowering to avoid changing the ABI of libLLVM.so.

Author: asb

llvm/lib/CodeGen/SelectionDAG/LegalizeIntegerTypes.cpp

@@ -271,8 +271,20 @@ SDValue DAGTypeLegalizer::PromoteIntRes_AtomicCmpSwap(AtomicSDNode *N,
     return Res.getValue(1);
   }
 
-  SDValue Op2 = GetPromotedInteger(N->getOperand(2));
+  // Op2 is used for the comparison and thus must be extended according to the
+  // target's atomic operations. Op3 is merely stored and so can be left alone.
+  SDValue Op2 = N->getOperand(2);
   SDValue Op3 = GetPromotedInteger(N->getOperand(3));
+  if (TLI.getTargetMachine().getTargetTriple().isRISCV()) {
+    // The comparison argument must be sign-extended for RISC-V. This is
+    // abstracted using a new TargetLowering hook in the main LLVM development
+    // branch, but handled here directly in order to fix the codegen bug for
+    // 10.x without breaking the libLLVM.so ABI.
+    Op2 = SExtPromotedInteger(Op2);
+  } else {
+    Op2 = GetPromotedInteger(Op2);
+  }
+
   SDVTList VTs =
       DAG.getVTList(Op2.getValueType(), N->getValueType(1), MVT::Other);
   SDValue Res = DAG.getAtomicCmpSwap(

llvm/test/CodeGen/RISCV/atomic-cmpxchg.ll

@@ -1628,6 +1628,7 @@ define void @cmpxchg_i32_monotonic_monotonic(i32* %ptr, i32 %cmp, i32 %val) noun
 ;
 ; RV64IA-LABEL: cmpxchg_i32_monotonic_monotonic:
 ; RV64IA:       # %bb.0:
+; RV64IA-NEXT:    sext.w a1, a1
 ; RV64IA-NEXT:  .LBB20_1: # =>This Inner Loop Header: Depth=1
 ; RV64IA-NEXT:    lr.w a3, (a0)
 ; RV64IA-NEXT:    bne a3, a1, .LBB20_3
@@ -1680,6 +1681,7 @@ define void @cmpxchg_i32_acquire_monotonic(i32* %ptr, i32 %cmp, i32 %val) nounwi
 ;
 ; RV64IA-LABEL: cmpxchg_i32_acquire_monotonic:
 ; RV64IA:       # %bb.0:
+; RV64IA-NEXT:    sext.w a1, a1
 ; RV64IA-NEXT:  .LBB21_1: # =>This Inner Loop Header: Depth=1
 ; RV64IA-NEXT:    lr.w.aq a3, (a0)
 ; RV64IA-NEXT:    bne a3, a1, .LBB21_3
@@ -1732,6 +1734,7 @@ define void @cmpxchg_i32_acquire_acquire(i32* %ptr, i32 %cmp, i32 %val) nounwind
 ;
 ; RV64IA-LABEL: cmpxchg_i32_acquire_acquire:
 ; RV64IA:       # %bb.0:
+; RV64IA-NEXT:    sext.w a1, a1
 ; RV64IA-NEXT:  .LBB22_1: # =>This Inner Loop Header: Depth=1
 ; RV64IA-NEXT:    lr.w.aq a3, (a0)
 ; RV64IA-NEXT:    bne a3, a1, .LBB22_3
@@ -1784,6 +1787,7 @@ define void @cmpxchg_i32_release_monotonic(i32* %ptr, i32 %cmp, i32 %val) nounwi
 ;
 ; RV64IA-LABEL: cmpxchg_i32_release_monotonic:
 ; RV64IA:       # %bb.0:
+; RV64IA-NEXT:    sext.w a1, a1
 ; RV64IA-NEXT:  .LBB23_1: # =>This Inner Loop Header: Depth=1
 ; RV64IA-NEXT:    lr.w a3, (a0)
 ; RV64IA-NEXT:    bne a3, a1, .LBB23_3
@@ -1836,6 +1840,7 @@ define void @cmpxchg_i32_release_acquire(i32* %ptr, i32 %cmp, i32 %val) nounwind
 ;
 ; RV64IA-LABEL: cmpxchg_i32_release_acquire:
 ; RV64IA:       # %bb.0:
+; RV64IA-NEXT:    sext.w a1, a1
 ; RV64IA-NEXT:  .LBB24_1: # =>This Inner Loop Header: Depth=1
 ; RV64IA-NEXT:    lr.w a3, (a0)
 ; RV64IA-NEXT:    bne a3, a1, .LBB24_3
@@ -1888,6 +1893,7 @@ define void @cmpxchg_i32_acq_rel_monotonic(i32* %ptr, i32 %cmp, i32 %val) nounwi
 ;
 ; RV64IA-LABEL: cmpxchg_i32_acq_rel_monotonic:
 ; RV64IA:       # %bb.0:
+; RV64IA-NEXT:    sext.w a1, a1
 ; RV64IA-NEXT:  .LBB25_1: # =>This Inner Loop Header: Depth=1
 ; RV64IA-NEXT:    lr.w.aq a3, (a0)
 ; RV64IA-NEXT:    bne a3, a1, .LBB25_3
@@ -1940,6 +1946,7 @@ define void @cmpxchg_i32_acq_rel_acquire(i32* %ptr, i32 %cmp, i32 %val) nounwind
 ;
 ; RV64IA-LABEL: cmpxchg_i32_acq_rel_acquire:
 ; RV64IA:       # %bb.0:
+; RV64IA-NEXT:    sext.w a1, a1
 ; RV64IA-NEXT:  .LBB26_1: # =>This Inner Loop Header: Depth=1
 ; RV64IA-NEXT:    lr.w.aq a3, (a0)
 ; RV64IA-NEXT:    bne a3, a1, .LBB26_3
@@ -1992,6 +1999,7 @@ define void @cmpxchg_i32_seq_cst_monotonic(i32* %ptr, i32 %cmp, i32 %val) nounwi
 ;
 ; RV64IA-LABEL: cmpxchg_i32_seq_cst_monotonic:
 ; RV64IA:       # %bb.0:
+; RV64IA-NEXT:    sext.w a1, a1
 ; RV64IA-NEXT:  .LBB27_1: # =>This Inner Loop Header: Depth=1
 ; RV64IA-NEXT:    lr.w.aqrl a3, (a0)
 ; RV64IA-NEXT:    bne a3, a1, .LBB27_3
@@ -2044,6 +2052,7 @@ define void @cmpxchg_i32_seq_cst_acquire(i32* %ptr, i32 %cmp, i32 %val) nounwind
 ;
 ; RV64IA-LABEL: cmpxchg_i32_seq_cst_acquire:
 ; RV64IA:       # %bb.0:
+; RV64IA-NEXT:    sext.w a1, a1
 ; RV64IA-NEXT:  .LBB28_1: # =>This Inner Loop Header: Depth=1
 ; RV64IA-NEXT:    lr.w.aqrl a3, (a0)
 ; RV64IA-NEXT:    bne a3, a1, .LBB28_3
@@ -2096,6 +2105,7 @@ define void @cmpxchg_i32_seq_cst_seq_cst(i32* %ptr, i32 %cmp, i32 %val) nounwind
 ;
 ; RV64IA-LABEL: cmpxchg_i32_seq_cst_seq_cst:
 ; RV64IA:       # %bb.0:
+; RV64IA-NEXT:    sext.w a1, a1
 ; RV64IA-NEXT:  .LBB29_1: # =>This Inner Loop Header: Depth=1
 ; RV64IA-NEXT:    lr.w.aqrl a3, (a0)
 ; RV64IA-NEXT:    bne a3, a1, .LBB29_3