Revert "[LSR][TTI][RISCV] Disable terminator folding for RISC-V."

This reverts commit fdb8764, and thus
re-enables terminator folding for RISCV.  The reported miscompile has
been fixed in f5dd70c.

Author: preames

llvm/lib/Target/RISCV/RISCVTargetTransformInfo.h

@@ -362,8 +362,7 @@ class RISCVTTIImpl : public BasicTTIImplBase<RISCVTTIImpl> {
                      const TargetTransformInfo::LSRCost &C2);
 
   bool shouldFoldTerminatingConditionAfterLSR() const {
-    // FIXME: Enabling this causes miscompiles.
-    return false;
+    return true;
   }
 };
 

llvm/test/CodeGen/RISCV/branch-on-zero.ll

@@ -120,36 +120,45 @@ define i32 @test_lshr2(ptr nocapture %x, ptr nocapture readonly %y, i32 %n) {
 ; RV32-LABEL: test_lshr2:
 ; RV32:       # %bb.0: # %entry
 ; RV32-NEXT:    srli a2, a2, 2
-; RV32-NEXT:    beqz a2, .LBB3_2
-; RV32-NEXT:  .LBB3_1: # %while.body
+; RV32-NEXT:    beqz a2, .LBB3_3
+; RV32-NEXT:  # %bb.1: # %while.body.preheader
+; RV32-NEXT:    slli a2, a2, 2
+; RV32-NEXT:    add a2, a1, a2
+; RV32-NEXT:  .LBB3_2: # %while.body
 ; RV32-NEXT:    # =>This Inner Loop Header: Depth=1
 ; RV32-NEXT:    lw a3, 0(a1)
-; RV32-NEXT:    addi a1, a1, 4
+; RV32-NEXT:    addi a4, a1, 4
 ; RV32-NEXT:    slli a3, a3, 1
-; RV32-NEXT:    addi a4, a0, 4
-; RV32-NEXT:    addi a2, a2, -1
+; RV32-NEXT:    addi a1, a0, 4
 ; RV32-NEXT:    sw a3, 0(a0)
-; RV32-NEXT:    mv a0, a4
-; RV32-NEXT:    bnez a2, .LBB3_1
-; RV32-NEXT:  .LBB3_2: # %while.end
+; RV32-NEXT:    mv a0, a1
+; RV32-NEXT:    mv a1, a4
+; RV32-NEXT:    bne a4, a2, .LBB3_2
+; RV32-NEXT:  .LBB3_3: # %while.end
 ; RV32-NEXT:    li a0, 0
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: test_lshr2:
 ; RV64:       # %bb.0: # %entry
 ; RV64-NEXT:    srliw a2, a2, 2
-; RV64-NEXT:    beqz a2, .LBB3_2
-; RV64-NEXT:  .LBB3_1: # %while.body
+; RV64-NEXT:    beqz a2, .LBB3_3
+; RV64-NEXT:  # %bb.1: # %while.body.preheader
+; RV64-NEXT:    addi a2, a2, -1
+; RV64-NEXT:    slli a2, a2, 32
+; RV64-NEXT:    srli a2, a2, 30
+; RV64-NEXT:    add a2, a2, a1
+; RV64-NEXT:    addi a2, a2, 4
+; RV64-NEXT:  .LBB3_2: # %while.body
 ; RV64-NEXT:    # =>This Inner Loop Header: Depth=1
 ; RV64-NEXT:    lw a3, 0(a1)
-; RV64-NEXT:    addi a1, a1, 4
+; RV64-NEXT:    addi a4, a1, 4
 ; RV64-NEXT:    slli a3, a3, 1
-; RV64-NEXT:    addi a4, a0, 4
-; RV64-NEXT:    addiw a2, a2, -1
+; RV64-NEXT:    addi a1, a0, 4
 ; RV64-NEXT:    sw a3, 0(a0)
-; RV64-NEXT:    mv a0, a4
-; RV64-NEXT:    bnez a2, .LBB3_1
-; RV64-NEXT:  .LBB3_2: # %while.end
+; RV64-NEXT:    mv a0, a1
+; RV64-NEXT:    mv a1, a4
+; RV64-NEXT:    bne a4, a2, .LBB3_2
+; RV64-NEXT:  .LBB3_3: # %while.end
 ; RV64-NEXT:    li a0, 0
 ; RV64-NEXT:    ret
 entry:

llvm/test/CodeGen/RISCV/riscv-codegenprepare-asm.ll

@@ -8,16 +8,18 @@
 define void @test1(ptr nocapture noundef %a, i32 noundef signext %n) {
 ; CHECK-LABEL: test1:
 ; CHECK:       # %bb.0: # %entry
-; CHECK-NEXT:    blez a1, .LBB0_2
-; CHECK-NEXT:  .LBB0_1: # %for.body
+; CHECK-NEXT:    blez a1, .LBB0_3
+; CHECK-NEXT:  # %bb.1: # %for.body.preheader
+; CHECK-NEXT:    slli a1, a1, 2
+; CHECK-NEXT:    add a1, a0, a1
+; CHECK-NEXT:  .LBB0_2: # %for.body
 ; CHECK-NEXT:    # =>This Inner Loop Header: Depth=1
 ; CHECK-NEXT:    lw a2, 0(a0)
 ; CHECK-NEXT:    addi a2, a2, 4
 ; CHECK-NEXT:    sw a2, 0(a0)
-; CHECK-NEXT:    addi a1, a1, -1
 ; CHECK-NEXT:    addi a0, a0, 4
-; CHECK-NEXT:    bnez a1, .LBB0_1
-; CHECK-NEXT:  .LBB0_2: # %for.cond.cleanup
+; CHECK-NEXT:    bne a0, a1, .LBB0_2
+; CHECK-NEXT:  .LBB0_3: # %for.cond.cleanup
 ; CHECK-NEXT:    ret
 entry:
   %cmp3 = icmp sgt i32 %n, 0

llvm/test/CodeGen/RISCV/rvv/fixed-vectors-fp-buildvec.ll

@@ -206,33 +206,19 @@ define <8 x float> @splat_idx_v8f32(<8 x float> %v, i64 %idx) {
 
 ; Test that we pull the vlse of the constant pool out of the loop.
 define dso_local void @splat_load_licm(float* %0) {
-; RV32-LABEL: splat_load_licm:
-; RV32:       # %bb.0:
-; RV32-NEXT:    li a1, 1024
-; RV32-NEXT:    lui a2, 263168
-; RV32-NEXT:    vsetivli zero, 4, e32, m1, ta, ma
-; RV32-NEXT:    vmv.v.x v8, a2
-; RV32-NEXT:  .LBB12_1: # =>This Inner Loop Header: Depth=1
-; RV32-NEXT:    vse32.v v8, (a0)
-; RV32-NEXT:    addi a1, a1, -4
-; RV32-NEXT:    addi a0, a0, 16
-; RV32-NEXT:    bnez a1, .LBB12_1
-; RV32-NEXT:  # %bb.2:
-; RV32-NEXT:    ret
-;
-; RV64-LABEL: splat_load_licm:
-; RV64:       # %bb.0:
-; RV64-NEXT:    li a1, 1024
-; RV64-NEXT:    lui a2, 263168
-; RV64-NEXT:    vsetivli zero, 4, e32, m1, ta, ma
-; RV64-NEXT:    vmv.v.x v8, a2
-; RV64-NEXT:  .LBB12_1: # =>This Inner Loop Header: Depth=1
-; RV64-NEXT:    vse32.v v8, (a0)
-; RV64-NEXT:    addiw a1, a1, -4
-; RV64-NEXT:    addi a0, a0, 16
-; RV64-NEXT:    bnez a1, .LBB12_1
-; RV64-NEXT:  # %bb.2:
-; RV64-NEXT:    ret
+; CHECK-LABEL: splat_load_licm:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    lui a1, 1
+; CHECK-NEXT:    add a1, a0, a1
+; CHECK-NEXT:    lui a2, 263168
+; CHECK-NEXT:    vsetivli zero, 4, e32, m1, ta, ma
+; CHECK-NEXT:    vmv.v.x v8, a2
+; CHECK-NEXT:  .LBB12_1: # =>This Inner Loop Header: Depth=1
+; CHECK-NEXT:    vse32.v v8, (a0)
+; CHECK-NEXT:    addi a0, a0, 16
+; CHECK-NEXT:    bne a0, a1, .LBB12_1
+; CHECK-NEXT:  # %bb.2:
+; CHECK-NEXT:    ret
   br label %2
 
 2:                                                ; preds = %2, %1
@@ -1408,3 +1394,6 @@ define <2 x double> @vid_step2_v2f64() {
 ; CHECK-NEXT:    ret
   ret <2 x double> <double 0.0, double 2.0>
 }
+;; NOTE: These prefixes are unused and the list is autogenerated. Do not add tests below this line:
+; RV32: {{.*}}
+; RV64: {{.*}}