[LoongArch][NFC] Pre-commit tests for sign-extension removal with vectors

Author: heiher

llvm/test/CodeGen/LoongArch/sextw-removal.ll

@@ -1334,3 +1334,201 @@ entry:
   %a = call i32 @llvm.loongarch.lsx.vpickve2gr.w(<4 x i32> %v, i32 3)
   ret i32 %a
 }
+
+define fastcc ptr @test21(ptr %B, ptr %Op0, ptr %Op1, ptr %P, ptr %M, i1 zeroext %I, i64 %0, ptr %a.1, i64 %1, i1 %c, i32 %2) {
+; CHECK-LABEL: test21:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    addi.d $sp, $sp, -80
+; CHECK-NEXT:    .cfi_def_cfa_offset 80
+; CHECK-NEXT:    st.d $ra, $sp, 72 # 8-byte Folded Spill
+; CHECK-NEXT:    st.d $fp, $sp, 64 # 8-byte Folded Spill
+; CHECK-NEXT:    st.d $s0, $sp, 56 # 8-byte Folded Spill
+; CHECK-NEXT:    st.d $s1, $sp, 48 # 8-byte Folded Spill
+; CHECK-NEXT:    st.d $s2, $sp, 40 # 8-byte Folded Spill
+; CHECK-NEXT:    st.d $s3, $sp, 32 # 8-byte Folded Spill
+; CHECK-NEXT:    st.d $s4, $sp, 24 # 8-byte Folded Spill
+; CHECK-NEXT:    st.d $s5, $sp, 16 # 8-byte Folded Spill
+; CHECK-NEXT:    st.d $s6, $sp, 8 # 8-byte Folded Spill
+; CHECK-NEXT:    .cfi_offset 1, -8
+; CHECK-NEXT:    .cfi_offset 22, -16
+; CHECK-NEXT:    .cfi_offset 23, -24
+; CHECK-NEXT:    .cfi_offset 24, -32
+; CHECK-NEXT:    .cfi_offset 25, -40
+; CHECK-NEXT:    .cfi_offset 26, -48
+; CHECK-NEXT:    .cfi_offset 27, -56
+; CHECK-NEXT:    .cfi_offset 28, -64
+; CHECK-NEXT:    .cfi_offset 29, -72
+; CHECK-NEXT:    ld.d $s6, $sp, 80
+; CHECK-NEXT:    move $s2, $a7
+; CHECK-NEXT:    move $s4, $a5
+; CHECK-NEXT:    move $s0, $a4
+; CHECK-NEXT:    move $fp, $a3
+; CHECK-NEXT:    move $s5, $a2
+; CHECK-NEXT:    move $s3, $a1
+; CHECK-NEXT:    move $s1, $a0
+; CHECK-NEXT:    move $a0, $zero
+; CHECK-NEXT:    jirl $ra, $zero, 0
+; CHECK-NEXT:    beqz $s4, .LBB24_2
+; CHECK-NEXT:  # %bb.1: # %if.then26
+; CHECK-NEXT:    addi.w $s6, $s6, 1
+; CHECK-NEXT:    beqz $s4, .LBB24_3
+; CHECK-NEXT:    b .LBB24_6
+; CHECK-NEXT:  .LBB24_2:
+; CHECK-NEXT:    move $s3, $s5
+; CHECK-NEXT:    bnez $s4, .LBB24_6
+; CHECK-NEXT:  .LBB24_3: # %for.cond32.preheader.preheader
+; CHECK-NEXT:    ld.d $a0, $sp, 96
+; CHECK-NEXT:    ld.d $a1, $sp, 88
+; CHECK-NEXT:    addi.w $a2, $s6, 0
+; CHECK-NEXT:    sltui $a2, $a2, 1
+; CHECK-NEXT:    masknez $a0, $a0, $a2
+; CHECK-NEXT:    vreplgr2vr.w $vr0, $s6
+; CHECK-NEXT:    andi $a1, $a1, 1
+; CHECK-NEXT:    .p2align 4, , 16
+; CHECK-NEXT:  .LBB24_4: # %for.cond32.preheader
+; CHECK-NEXT:    # =>This Inner Loop Header: Depth=1
+; CHECK-NEXT:    st.w $a0, $zero, 0
+; CHECK-NEXT:    vst $vr0, $s2, 0
+; CHECK-NEXT:    bnez $a1, .LBB24_4
+; CHECK-NEXT:  # %bb.5: # %for.cond.cleanup
+; CHECK-NEXT:    move $a0, $zero
+; CHECK-NEXT:    move $a1, $s3
+; CHECK-NEXT:    move $a2, $zero
+; CHECK-NEXT:    move $a3, $zero
+; CHECK-NEXT:    move $a4, $zero
+; CHECK-NEXT:    move $a5, $zero
+; CHECK-NEXT:    jirl $ra, $zero, 0
+; CHECK-NEXT:    move $a0, $s1
+; CHECK-NEXT:    move $a1, $s0
+; CHECK-NEXT:    move $a2, $zero
+; CHECK-NEXT:    move $a3, $fp
+; CHECK-NEXT:    jirl $ra, $zero, 0
+; CHECK-NEXT:  .LBB24_6: # %for.cond32.preheader.us.preheader
+; CHECK-NEXT:    move $a0, $zero
+; CHECK-NEXT:    ld.d $s6, $sp, 8 # 8-byte Folded Reload
+; CHECK-NEXT:    ld.d $s5, $sp, 16 # 8-byte Folded Reload
+; CHECK-NEXT:    ld.d $s4, $sp, 24 # 8-byte Folded Reload
+; CHECK-NEXT:    ld.d $s3, $sp, 32 # 8-byte Folded Reload
+; CHECK-NEXT:    ld.d $s2, $sp, 40 # 8-byte Folded Reload
+; CHECK-NEXT:    ld.d $s1, $sp, 48 # 8-byte Folded Reload
+; CHECK-NEXT:    ld.d $s0, $sp, 56 # 8-byte Folded Reload
+; CHECK-NEXT:    ld.d $fp, $sp, 64 # 8-byte Folded Reload
+; CHECK-NEXT:    ld.d $ra, $sp, 72 # 8-byte Folded Reload
+; CHECK-NEXT:    addi.d $sp, $sp, 80
+; CHECK-NEXT:    ret
+;
+; NORMV-LABEL: test21:
+; NORMV:       # %bb.0: # %entry
+; NORMV-NEXT:    addi.d $sp, $sp, -80
+; NORMV-NEXT:    .cfi_def_cfa_offset 80
+; NORMV-NEXT:    st.d $ra, $sp, 72 # 8-byte Folded Spill
+; NORMV-NEXT:    st.d $fp, $sp, 64 # 8-byte Folded Spill
+; NORMV-NEXT:    st.d $s0, $sp, 56 # 8-byte Folded Spill
+; NORMV-NEXT:    st.d $s1, $sp, 48 # 8-byte Folded Spill
+; NORMV-NEXT:    st.d $s2, $sp, 40 # 8-byte Folded Spill
+; NORMV-NEXT:    st.d $s3, $sp, 32 # 8-byte Folded Spill
+; NORMV-NEXT:    st.d $s4, $sp, 24 # 8-byte Folded Spill
+; NORMV-NEXT:    st.d $s5, $sp, 16 # 8-byte Folded Spill
+; NORMV-NEXT:    st.d $s6, $sp, 8 # 8-byte Folded Spill
+; NORMV-NEXT:    .cfi_offset 1, -8
+; NORMV-NEXT:    .cfi_offset 22, -16
+; NORMV-NEXT:    .cfi_offset 23, -24
+; NORMV-NEXT:    .cfi_offset 24, -32
+; NORMV-NEXT:    .cfi_offset 25, -40
+; NORMV-NEXT:    .cfi_offset 26, -48
+; NORMV-NEXT:    .cfi_offset 27, -56
+; NORMV-NEXT:    .cfi_offset 28, -64
+; NORMV-NEXT:    .cfi_offset 29, -72
+; NORMV-NEXT:    ld.d $s6, $sp, 80
+; NORMV-NEXT:    move $s2, $a7
+; NORMV-NEXT:    move $s4, $a5
+; NORMV-NEXT:    move $s0, $a4
+; NORMV-NEXT:    move $fp, $a3
+; NORMV-NEXT:    move $s5, $a2
+; NORMV-NEXT:    move $s3, $a1
+; NORMV-NEXT:    move $s1, $a0
+; NORMV-NEXT:    move $a0, $zero
+; NORMV-NEXT:    jirl $ra, $zero, 0
+; NORMV-NEXT:    beqz $s4, .LBB24_2
+; NORMV-NEXT:  # %bb.1: # %if.then26
+; NORMV-NEXT:    addi.d $a0, $s6, 1
+; NORMV-NEXT:    addi.w $s6, $a0, 0
+; NORMV-NEXT:    beqz $s4, .LBB24_3
+; NORMV-NEXT:    b .LBB24_6
+; NORMV-NEXT:  .LBB24_2:
+; NORMV-NEXT:    move $s3, $s5
+; NORMV-NEXT:    bnez $s4, .LBB24_6
+; NORMV-NEXT:  .LBB24_3: # %for.cond32.preheader.preheader
+; NORMV-NEXT:    ld.d $a0, $sp, 96
+; NORMV-NEXT:    ld.d $a1, $sp, 88
+; NORMV-NEXT:    addi.w $a2, $s6, 0
+; NORMV-NEXT:    sltui $a2, $a2, 1
+; NORMV-NEXT:    masknez $a0, $a0, $a2
+; NORMV-NEXT:    vreplgr2vr.w $vr0, $s6
+; NORMV-NEXT:    andi $a1, $a1, 1
+; NORMV-NEXT:    .p2align 4, , 16
+; NORMV-NEXT:  .LBB24_4: # %for.cond32.preheader
+; NORMV-NEXT:    # =>This Inner Loop Header: Depth=1
+; NORMV-NEXT:    st.w $a0, $zero, 0
+; NORMV-NEXT:    vst $vr0, $s2, 0
+; NORMV-NEXT:    bnez $a1, .LBB24_4
+; NORMV-NEXT:  # %bb.5: # %for.cond.cleanup
+; NORMV-NEXT:    move $a0, $zero
+; NORMV-NEXT:    move $a1, $s3
+; NORMV-NEXT:    move $a2, $zero
+; NORMV-NEXT:    move $a3, $zero
+; NORMV-NEXT:    move $a4, $zero
+; NORMV-NEXT:    move $a5, $zero
+; NORMV-NEXT:    jirl $ra, $zero, 0
+; NORMV-NEXT:    move $a0, $s1
+; NORMV-NEXT:    move $a1, $s0
+; NORMV-NEXT:    move $a2, $zero
+; NORMV-NEXT:    move $a3, $fp
+; NORMV-NEXT:    jirl $ra, $zero, 0
+; NORMV-NEXT:  .LBB24_6: # %for.cond32.preheader.us.preheader
+; NORMV-NEXT:    move $a0, $zero
+; NORMV-NEXT:    ld.d $s6, $sp, 8 # 8-byte Folded Reload
+; NORMV-NEXT:    ld.d $s5, $sp, 16 # 8-byte Folded Reload
+; NORMV-NEXT:    ld.d $s4, $sp, 24 # 8-byte Folded Reload
+; NORMV-NEXT:    ld.d $s3, $sp, 32 # 8-byte Folded Reload
+; NORMV-NEXT:    ld.d $s2, $sp, 40 # 8-byte Folded Reload
+; NORMV-NEXT:    ld.d $s1, $sp, 48 # 8-byte Folded Reload
+; NORMV-NEXT:    ld.d $s0, $sp, 56 # 8-byte Folded Reload
+; NORMV-NEXT:    ld.d $fp, $sp, 64 # 8-byte Folded Reload
+; NORMV-NEXT:    ld.d $ra, $sp, 72 # 8-byte Folded Reload
+; NORMV-NEXT:    addi.d $sp, $sp, 80
+; NORMV-NEXT:    ret
+entry:
+  %call4.i.i.i.i = tail call i32 null(ptr null)
+  %conv = trunc i64 %1 to i32
+  br i1 %I, label %if.then26, label %if.end30
+
+if.then26:                                        ; preds = %entry
+  %sub27 = add i32 %conv, 1
+  br label %if.end30
+
+if.end30:                                         ; preds = %if.then26, %entry
+  %Op1.addr.0 = phi ptr [ %Op0, %if.then26 ], [ %Op1, %entry ]
+  %ShiftVal.1 = phi i32 [ %sub27, %if.then26 ], [ %conv, %entry ]
+  br i1 %I, label %for.cond32.preheader.us.preheader, label %for.cond32.preheader.preheader
+
+for.cond32.preheader.preheader:                   ; preds = %if.end30
+  %cmp37 = icmp ult i32 %ShiftVal.1, 1
+  %spec.select = select i1 %cmp37, i32 0, i32 %2
+  br label %for.cond32.preheader
+
+for.cond32.preheader.us.preheader:                ; preds = %if.end30
+  ret ptr null
+
+for.cond32.preheader:                             ; preds = %for.cond32.preheader, %for.cond32.preheader.preheader
+  store i32 %spec.select, ptr null, align 4
+  %3 = insertelement <4 x i32> zeroinitializer, i32 %ShiftVal.1, i64 0
+  %4 = shufflevector <4 x i32> %3, <4 x i32> zeroinitializer, <4 x i32> zeroinitializer
+  store <4 x i32> %4, ptr %a.1, align 4
+  br i1 %c, label %for.cond32.preheader, label %for.cond.cleanup
+
+for.cond.cleanup:                                 ; preds = %for.cond32.preheader
+  %call48 = call ptr null(ptr null, ptr %Op1.addr.0, ptr null, [2 x i64] zeroinitializer, ptr null)
+  %call49 = call fastcc ptr null(ptr %B, ptr %M, ptr null, ptr %P)
+  ret ptr null
+}