Index: llvm/test/Transforms/IRCE/pre_post_loops.ll =================================================================== --- llvm/test/Transforms/IRCE/pre_post_loops.ll +++ llvm/test/Transforms/IRCE/pre_post_loops.ll @@ -1,3 +1,4 @@ +; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --version 3 ; RUN: opt -verify-loop-info -irce-print-changed-loops -passes=irce -S < %s 2>&1 | FileCheck %s ; RUN: opt -verify-loop-info -irce-print-changed-loops -passes='require,irce' -S < %s 2>&1 | FileCheck %s @@ -6,37 +7,56 @@ ; Iterate from 0 to SINT_MAX, check that the post-loop is generated. define void @test_01(ptr %arr, ptr %a_len_ptr) { - -; CHECK: test_01( -; CHECK: entry: -; CHECK-NEXT: %exit.mainloop.at = load i32, ptr %a_len_ptr +; CHECK-LABEL: define void @test_01 +; CHECK-SAME: (ptr [[ARR:%.*]], ptr [[A_LEN_PTR:%.*]]) { +; CHECK-NEXT: entry: +; CHECK-NEXT: [[EXIT_MAINLOOP_AT:%.*]] = load i32, ptr [[A_LEN_PTR]], align 4, !range [[RNG0:![0-9]+]] +; CHECK-NEXT: [[TMP0:%.*]] = icmp slt i32 0, [[EXIT_MAINLOOP_AT]] +; CHECK-NEXT: br i1 [[TMP0]], label [[LOOP_PREHEADER:%.*]], label [[MAIN_PSEUDO_EXIT:%.*]] +; CHECK: loop.preheader: +; CHECK-NEXT: br label [[LOOP:%.*]] ; CHECK: loop: -; CHECK-NEXT: %idx = phi i32 [ %idx.next, %in.bounds ], [ 0, %loop.preheader ] -; CHECK-NEXT: %idx.next = add i32 %idx, 1 -; CHECK-NEXT: %abc = icmp slt i32 %idx, %exit.mainloop.at -; CHECK-NEXT: br i1 true, label %in.bounds, +; CHECK-NEXT: [[IDX:%.*]] = phi i32 [ [[IDX_NEXT:%.*]], [[IN_BOUNDS:%.*]] ], [ 0, [[LOOP_PREHEADER]] ] +; CHECK-NEXT: [[IDX_NEXT]] = add i32 [[IDX]], 1 +; CHECK-NEXT: [[ABC:%.*]] = icmp slt i32 [[IDX]], [[EXIT_MAINLOOP_AT]] +; CHECK-NEXT: br i1 true, label [[IN_BOUNDS]], label [[OUT_OF_BOUNDS_LOOPEXIT1:%.*]] ; CHECK: in.bounds: -; CHECK-NEXT: %addr = getelementptr i32, ptr %arr, i32 %idx -; CHECK-NEXT: store i32 0, ptr %addr -; CHECK-NEXT: %next = icmp slt i32 %idx.next, 2147483647 -; CHECK-NEXT: [[COND:%[^ ]+]] = icmp slt i32 %idx.next, %exit.mainloop.at -; CHECK-NEXT: br i1 [[COND]], label %loop, label %main.exit.selector +; CHECK-NEXT: [[ADDR:%.*]] = getelementptr i32, ptr [[ARR]], i32 [[IDX]] +; CHECK-NEXT: store i32 0, ptr [[ADDR]], align 4 +; CHECK-NEXT: [[NEXT:%.*]] = icmp slt i32 [[IDX_NEXT]], 2147483647 +; CHECK-NEXT: [[TMP1:%.*]] = icmp slt i32 [[IDX_NEXT]], [[EXIT_MAINLOOP_AT]] +; CHECK-NEXT: br i1 [[TMP1]], label [[LOOP]], label [[MAIN_EXIT_SELECTOR:%.*]] +; CHECK: main.exit.selector: +; CHECK-NEXT: [[IDX_NEXT_LCSSA:%.*]] = phi i32 [ [[IDX_NEXT]], [[IN_BOUNDS]] ] +; CHECK-NEXT: [[TMP2:%.*]] = icmp slt i32 [[IDX_NEXT_LCSSA]], 2147483647 +; CHECK-NEXT: br i1 [[TMP2]], label [[MAIN_PSEUDO_EXIT]], label [[EXIT:%.*]] ; CHECK: main.pseudo.exit: -; CHECK-NEXT: %idx.copy = phi i32 [ 0, %entry ], [ %idx.next.lcssa, %main.exit.selector ] -; CHECK-NEXT: %indvar.end = phi i32 [ 0, %entry ], [ %idx.next.lcssa, %main.exit.selector ] -; CHECK-NEXT: br label %postloop +; CHECK-NEXT: [[IDX_COPY:%.*]] = phi i32 [ 0, [[ENTRY:%.*]] ], [ [[IDX_NEXT_LCSSA]], [[MAIN_EXIT_SELECTOR]] ] +; CHECK-NEXT: [[INDVAR_END:%.*]] = phi i32 [ 0, [[ENTRY]] ], [ [[IDX_NEXT_LCSSA]], [[MAIN_EXIT_SELECTOR]] ] +; CHECK-NEXT: br label [[POSTLOOP:%.*]] +; CHECK: out.of.bounds.loopexit: +; CHECK-NEXT: br label [[OUT_OF_BOUNDS:%.*]] +; CHECK: out.of.bounds.loopexit1: +; CHECK-NEXT: br label [[OUT_OF_BOUNDS]] +; CHECK: out.of.bounds: +; CHECK-NEXT: ret void +; CHECK: exit.loopexit: +; CHECK-NEXT: br label [[EXIT]] +; CHECK: exit: +; CHECK-NEXT: ret void ; CHECK: postloop: -; CHECK-NEXT: br label %loop.postloop +; CHECK-NEXT: br label [[LOOP_POSTLOOP:%.*]] ; CHECK: loop.postloop: -; CHECK-NEXT: %idx.postloop = phi i32 [ %idx.copy, %postloop ], [ %idx.next.postloop, %in.bounds.postloop ] -; CHECK-NEXT: %idx.next.postloop = add i32 %idx.postloop, 1 -; CHECK-NEXT: %abc.postloop = icmp slt i32 %idx.postloop, %exit.mainloop.at -; CHECK-NEXT: br i1 %abc.postloop, label %in.bounds.postloop, label %out.of.bounds.loopexit +; CHECK-NEXT: [[IDX_POSTLOOP:%.*]] = phi i32 [ [[IDX_COPY]], [[POSTLOOP]] ], [ [[IDX_NEXT_POSTLOOP:%.*]], [[IN_BOUNDS_POSTLOOP:%.*]] ] +; CHECK-NEXT: [[IDX_NEXT_POSTLOOP]] = add i32 [[IDX_POSTLOOP]], 1 +; CHECK-NEXT: [[ABC_POSTLOOP:%.*]] = icmp slt i32 [[IDX_POSTLOOP]], [[EXIT_MAINLOOP_AT]] +; CHECK-NEXT: br i1 [[ABC_POSTLOOP]], label [[IN_BOUNDS_POSTLOOP]], label [[OUT_OF_BOUNDS_LOOPEXIT:%.*]] ; CHECK: in.bounds.postloop: -; CHECK-NEXT: %addr.postloop = getelementptr i32, ptr %arr, i32 %idx.postloop -; CHECK-NEXT: store i32 0, ptr %addr.postloop -; CHECK-NEXT: %next.postloop = icmp slt i32 %idx.next.postloop, 2147483647 -; CHECK-NEXT: br i1 %next.postloop, label %loop.postloop, label %exit.loopexit +; CHECK-NEXT: [[ADDR_POSTLOOP:%.*]] = getelementptr i32, ptr [[ARR]], i32 [[IDX_POSTLOOP]] +; CHECK-NEXT: store i32 0, ptr [[ADDR_POSTLOOP]], align 4 +; CHECK-NEXT: [[NEXT_POSTLOOP:%.*]] = icmp slt i32 [[IDX_NEXT_POSTLOOP]], 2147483647 +; CHECK-NEXT: br i1 [[NEXT_POSTLOOP]], label [[LOOP_POSTLOOP]], label [[EXIT_LOOPEXIT:%.*]], !llvm.loop [[LOOP1:![0-9]+]], !irce.loop.clone [[META6:![0-9]+]] +; entry: %len = load i32, ptr %a_len_ptr, !range !0 @@ -63,34 +83,55 @@ ; Iterate from SINT_MAX to 0, check that the pre-loop is generated. define void @test_02(ptr %arr, ptr %a_len_ptr) { - -; CHECK: test_02( -; CHECK: entry: -; CHECK-NEXT: %len = load i32, ptr %a_len_ptr, align 4, !range !0 -; CHECK-NEXT: br i1 true, label %loop.preloop.preheader -; CHECK: mainloop: -; CHECK-NEXT: br label %loop -; CHECK: loop: -; CHECK-NEXT: %idx = phi i32 [ %idx.preloop.copy, %mainloop ], [ %idx.next, %in.bounds ] -; CHECK-NEXT: %idx.next = add i32 %idx, -1 -; CHECK-NEXT: %abc = icmp slt i32 %idx, %len -; CHECK-NEXT: br i1 true, label %in.bounds -; CHECK: in.bounds: -; CHECK-NEXT: %addr = getelementptr i32, ptr %arr, i32 %idx -; CHECK-NEXT: store i32 0, ptr %addr -; CHECK-NEXT: %next = icmp sgt i32 %idx.next, -1 -; CHECK-NEXT: br i1 %next, label %loop, label %exit.loopexit -; CHECK: loop.preloop: -; CHECK-NEXT: %idx.preloop = phi i32 [ %idx.next.preloop, %in.bounds.preloop ], [ 2147483647, %loop.preloop.preheader ] -; CHECK-NEXT: %idx.next.preloop = add i32 %idx.preloop, -1 -; CHECK-NEXT: %abc.preloop = icmp slt i32 %idx.preloop, %len -; CHECK-NEXT: br i1 %abc.preloop, label %in.bounds.preloop, label %out.of.bounds.loopexit -; CHECK: in.bounds.preloop: -; CHECK-NEXT: %addr.preloop = getelementptr i32, ptr %arr, i32 %idx.preloop -; CHECK-NEXT: store i32 0, ptr %addr.preloop -; CHECK-NEXT: %next.preloop = icmp sgt i32 %idx.next.preloop, -1 -; CHECK-NEXT: [[COND:%[^ ]+]] = icmp sgt i32 %idx.next.preloop, -1 -; CHECK-NEXT: br i1 [[COND]], label %loop.preloop, label %preloop.exit.selector +; CHECK-LABEL: define void @test_02 +; CHECK-SAME: (ptr [[ARR:%.*]], ptr [[A_LEN_PTR:%.*]]) { +; CHECK-NEXT: entry: +; CHECK-NEXT: [[LEN:%.*]] = load i32, ptr [[A_LEN_PTR]], align 4, !range [[RNG0]] +; CHECK-NEXT: br i1 true, label [[LOOP_PRELOOP_PREHEADER:%.*]], label [[PRELOOP_PSEUDO_EXIT:%.*]] +; CHECK: loop.preloop.preheader: +; CHECK-NEXT: br label [[LOOP_PRELOOP:%.*]] +; CHECK: mainloop: +; CHECK-NEXT: br label [[LOOP:%.*]] +; CHECK: loop: +; CHECK-NEXT: [[IDX:%.*]] = phi i32 [ [[IDX_PRELOOP_COPY:%.*]], [[MAINLOOP:%.*]] ], [ [[IDX_NEXT:%.*]], [[IN_BOUNDS:%.*]] ] +; CHECK-NEXT: [[IDX_NEXT]] = add i32 [[IDX]], -1 +; CHECK-NEXT: [[ABC:%.*]] = icmp slt i32 [[IDX]], [[LEN]] +; CHECK-NEXT: br i1 true, label [[IN_BOUNDS]], label [[OUT_OF_BOUNDS_LOOPEXIT1:%.*]] +; CHECK: in.bounds: +; CHECK-NEXT: [[ADDR:%.*]] = getelementptr i32, ptr [[ARR]], i32 [[IDX]] +; CHECK-NEXT: store i32 0, ptr [[ADDR]], align 4 +; CHECK-NEXT: [[NEXT:%.*]] = icmp sgt i32 [[IDX_NEXT]], -1 +; CHECK-NEXT: br i1 [[NEXT]], label [[LOOP]], label [[EXIT_LOOPEXIT:%.*]] +; CHECK: out.of.bounds.loopexit: +; CHECK-NEXT: br label [[OUT_OF_BOUNDS:%.*]] +; CHECK: out.of.bounds.loopexit1: +; CHECK-NEXT: br label [[OUT_OF_BOUNDS]] +; CHECK: out.of.bounds: +; CHECK-NEXT: ret void +; CHECK: exit.loopexit: +; CHECK-NEXT: br label [[EXIT:%.*]] +; CHECK: exit: +; CHECK-NEXT: ret void +; CHECK: loop.preloop: +; CHECK-NEXT: [[IDX_PRELOOP:%.*]] = phi i32 [ [[IDX_NEXT_PRELOOP:%.*]], [[IN_BOUNDS_PRELOOP:%.*]] ], [ 2147483647, [[LOOP_PRELOOP_PREHEADER]] ] +; CHECK-NEXT: [[IDX_NEXT_PRELOOP]] = add i32 [[IDX_PRELOOP]], -1 +; CHECK-NEXT: [[ABC_PRELOOP:%.*]] = icmp slt i32 [[IDX_PRELOOP]], [[LEN]] +; CHECK-NEXT: br i1 [[ABC_PRELOOP]], label [[IN_BOUNDS_PRELOOP]], label [[OUT_OF_BOUNDS_LOOPEXIT:%.*]] +; CHECK: in.bounds.preloop: +; CHECK-NEXT: [[ADDR_PRELOOP:%.*]] = getelementptr i32, ptr [[ARR]], i32 [[IDX_PRELOOP]] +; CHECK-NEXT: store i32 0, ptr [[ADDR_PRELOOP]], align 4 +; CHECK-NEXT: [[NEXT_PRELOOP:%.*]] = icmp sgt i32 [[IDX_NEXT_PRELOOP]], -1 +; CHECK-NEXT: [[TMP0:%.*]] = icmp sgt i32 [[IDX_NEXT_PRELOOP]], -1 +; CHECK-NEXT: br i1 [[TMP0]], label [[LOOP_PRELOOP]], label [[PRELOOP_EXIT_SELECTOR:%.*]], !llvm.loop [[LOOP7:![0-9]+]], !irce.loop.clone [[META6]] +; CHECK: preloop.exit.selector: +; CHECK-NEXT: [[IDX_NEXT_PRELOOP_LCSSA:%.*]] = phi i32 [ [[IDX_NEXT_PRELOOP]], [[IN_BOUNDS_PRELOOP]] ] +; CHECK-NEXT: [[TMP1:%.*]] = icmp sgt i32 [[IDX_NEXT_PRELOOP_LCSSA]], -1 +; CHECK-NEXT: br i1 [[TMP1]], label [[PRELOOP_PSEUDO_EXIT]], label [[EXIT]] +; CHECK: preloop.pseudo.exit: +; CHECK-NEXT: [[IDX_PRELOOP_COPY]] = phi i32 [ 2147483647, [[ENTRY:%.*]] ], [ [[IDX_NEXT_PRELOOP_LCSSA]], [[PRELOOP_EXIT_SELECTOR]] ] +; CHECK-NEXT: [[INDVAR_END:%.*]] = phi i32 [ 2147483647, [[ENTRY]] ], [ [[IDX_NEXT_PRELOOP_LCSSA]], [[PRELOOP_EXIT_SELECTOR]] ] +; CHECK-NEXT: br label [[MAINLOOP]] +; entry: %len = load i32, ptr %a_len_ptr, !range !0 @@ -116,3 +157,13 @@ } !0 = !{i32 0, i32 50} +;. +; CHECK: [[RNG0]] = !{i32 0, i32 50} +; CHECK: [[LOOP1]] = distinct !{[[LOOP1]], [[META2:![0-9]+]], [[META3:![0-9]+]], [[META4:![0-9]+]], [[META5:![0-9]+]]} +; CHECK: [[META2]] = !{!"llvm.loop.unroll.disable"} +; CHECK: [[META3]] = !{!"llvm.loop.vectorize.enable", i1 false} +; CHECK: [[META4]] = !{!"llvm.loop.licm_versioning.disable"} +; CHECK: [[META5]] = !{!"llvm.loop.distribute.enable", i1 false} +; CHECK: [[META6]] = !{} +; CHECK: [[LOOP7]] = distinct !{[[LOOP7]], [[META2]], [[META3]], [[META4]], [[META5]]} +;. Index: llvm/test/Transforms/IRCE/range_intersect_miscompile.ll =================================================================== --- llvm/test/Transforms/IRCE/range_intersect_miscompile.ll +++ llvm/test/Transforms/IRCE/range_intersect_miscompile.ll @@ -1,3 +1,4 @@ +; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --version 3 ; RUN: opt -verify-loop-info -irce-print-changed-loops -passes=irce -S < %s 2>&1 | FileCheck %s ; RUN: opt -verify-loop-info -irce-print-changed-loops -passes='require,irce' -S < %s 2>&1 | FileCheck %s @@ -15,26 +16,75 @@ ; postloop. define void @test_01() { - -; CHECK-LABEL: test_01 -; CHECK-NOT: preloop -; CHECK: range_check_block: ; preds = %inner_loop -; CHECK-NEXT: %range_check = icmp slt i32 %iv, 331 -; CHECK-NEXT: br i1 true, label %loop_latch -; CHECK: loop_latch: -; CHECK-NEXT: %iv_next = add i32 %iv, 1 -; CHECK-NEXT: %loop_cond = icmp ult i32 %iv_next, 400 -; CHECK-NEXT: [[COND:%[^ ]+]] = icmp ult i32 %iv_next, 331 -; CHECK-NEXT: br i1 [[COND]], label %loop_header, label %main.exit.selector -; CHECK: main.exit.selector: ; preds = %loop_latch -; CHECK-NEXT: %iv_next.lcssa = phi i32 [ %iv_next, %loop_latch ] -; CHECK-NEXT: %iv.lcssa = phi i32 [ %iv, %loop_latch ] -; CHECK-NEXT: [[MES_COND:%[^ ]+]] = icmp ult i32 %iv_next.lcssa, 400 -; CHECK-NEXT: br i1 [[MES_COND]], label %main.pseudo.exit, label %exit -; CHECK: loop_latch.postloop: ; preds = %range_check_block.postloop -; CHECK-NEXT: %iv_next.postloop = add i32 %iv.postloop, 1 -; CHECK-NEXT: %loop_cond.postloop = icmp ult i32 %iv_next.postloop, 400 -; CHECK-NEXT: br i1 %loop_cond.postloop, label %loop_header.postloop, label %exit.loopexit +; CHECK-LABEL: define void @test_01() { +; CHECK-NEXT: entry: +; CHECK-NEXT: br i1 true, label [[LOOP_HEADER_PREHEADER:%.*]], label [[MAIN_PSEUDO_EXIT:%.*]] +; CHECK: loop_header.preheader: +; CHECK-NEXT: br label [[LOOP_HEADER:%.*]] +; CHECK: loop_header: +; CHECK-NEXT: [[IV:%.*]] = phi i32 [ [[IV_NEXT:%.*]], [[LOOP_LATCH:%.*]] ], [ 2, [[LOOP_HEADER_PREHEADER]] ] +; CHECK-NEXT: [[IV_PREV:%.*]] = phi i32 [ [[IV]], [[LOOP_LATCH]] ], [ 1, [[LOOP_HEADER_PREHEADER]] ] +; CHECK-NEXT: [[TMP2:%.*]] = icmp sgt i32 [[IV_PREV]], -1 +; CHECK-NEXT: br i1 true, label [[LOOP_HEADER_SPLIT_US:%.*]], label [[EXIT_LOOPEXIT1:%.*]] +; CHECK: loop_header.split.us: +; CHECK-NEXT: br label [[INNER_LOOP:%.*]] +; CHECK: inner_loop: +; CHECK-NEXT: [[INNER_IV:%.*]] = phi i32 [ 1, [[LOOP_HEADER_SPLIT_US]] ], [ [[INNER_IV_NEXT:%.*]], [[INNER_LOOP]] ] +; CHECK-NEXT: [[INNER_IV_NEXT]] = add nuw nsw i32 [[INNER_IV]], 1 +; CHECK-NEXT: [[INNER_COND:%.*]] = icmp ult i32 [[INNER_IV_NEXT]], 31 +; CHECK-NEXT: br i1 [[INNER_COND]], label [[INNER_LOOP]], label [[RANGE_CHECK_BLOCK:%.*]] +; CHECK: exit.loopexit: +; CHECK-NEXT: br label [[EXIT:%.*]] +; CHECK: exit.loopexit1: +; CHECK-NEXT: br label [[EXIT]] +; CHECK: exit: +; CHECK-NEXT: ret void +; CHECK: range_check_block: +; CHECK-NEXT: [[RANGE_CHECK:%.*]] = icmp slt i32 [[IV]], 331 +; CHECK-NEXT: br i1 true, label [[LOOP_LATCH]], label [[DEOPT_LOOPEXIT2:%.*]] +; CHECK: loop_latch: +; CHECK-NEXT: [[IV_NEXT]] = add i32 [[IV]], 1 +; CHECK-NEXT: [[LOOP_COND:%.*]] = icmp ult i32 [[IV_NEXT]], 400 +; CHECK-NEXT: [[TMP0:%.*]] = icmp ult i32 [[IV_NEXT]], 331 +; CHECK-NEXT: br i1 [[TMP0]], label [[LOOP_HEADER]], label [[MAIN_EXIT_SELECTOR:%.*]] +; CHECK: main.exit.selector: +; CHECK-NEXT: [[IV_NEXT_LCSSA:%.*]] = phi i32 [ [[IV_NEXT]], [[LOOP_LATCH]] ] +; CHECK-NEXT: [[IV_LCSSA:%.*]] = phi i32 [ [[IV]], [[LOOP_LATCH]] ] +; CHECK-NEXT: [[TMP1:%.*]] = icmp ult i32 [[IV_NEXT_LCSSA]], 400 +; CHECK-NEXT: br i1 [[TMP1]], label [[MAIN_PSEUDO_EXIT]], label [[EXIT]] +; CHECK: main.pseudo.exit: +; CHECK-NEXT: [[IV_COPY:%.*]] = phi i32 [ 2, [[ENTRY:%.*]] ], [ [[IV_NEXT_LCSSA]], [[MAIN_EXIT_SELECTOR]] ] +; CHECK-NEXT: [[IV_PREV_COPY:%.*]] = phi i32 [ 1, [[ENTRY]] ], [ [[IV_LCSSA]], [[MAIN_EXIT_SELECTOR]] ] +; CHECK-NEXT: [[INDVAR_END:%.*]] = phi i32 [ 2, [[ENTRY]] ], [ [[IV_NEXT_LCSSA]], [[MAIN_EXIT_SELECTOR]] ] +; CHECK-NEXT: br label [[POSTLOOP:%.*]] +; CHECK: deopt.loopexit: +; CHECK-NEXT: br label [[DEOPT:%.*]] +; CHECK: deopt.loopexit2: +; CHECK-NEXT: br label [[DEOPT]] +; CHECK: deopt: +; CHECK-NEXT: ret void +; CHECK: postloop: +; CHECK-NEXT: br label [[LOOP_HEADER_POSTLOOP:%.*]] +; CHECK: loop_header.postloop: +; CHECK-NEXT: [[IV_POSTLOOP:%.*]] = phi i32 [ [[IV_COPY]], [[POSTLOOP]] ], [ [[IV_NEXT_POSTLOOP:%.*]], [[LOOP_LATCH_POSTLOOP:%.*]] ] +; CHECK-NEXT: [[IV_PREV_POSTLOOP:%.*]] = phi i32 [ [[IV_PREV_COPY]], [[POSTLOOP]] ], [ [[IV_POSTLOOP]], [[LOOP_LATCH_POSTLOOP]] ] +; CHECK-NEXT: [[TMP2_POSTLOOP:%.*]] = icmp sgt i32 [[IV_PREV_POSTLOOP]], -1 +; CHECK-NEXT: br i1 [[TMP2_POSTLOOP]], label [[LOOP_HEADER_SPLIT_US_POSTLOOP:%.*]], label [[EXIT_LOOPEXIT:%.*]] +; CHECK: loop_header.split.us.postloop: +; CHECK-NEXT: br label [[INNER_LOOP_POSTLOOP:%.*]] +; CHECK: inner_loop.postloop: +; CHECK-NEXT: [[INNER_IV_POSTLOOP:%.*]] = phi i32 [ 1, [[LOOP_HEADER_SPLIT_US_POSTLOOP]] ], [ [[INNER_IV_NEXT_POSTLOOP:%.*]], [[INNER_LOOP_POSTLOOP]] ] +; CHECK-NEXT: [[INNER_IV_NEXT_POSTLOOP]] = add nuw nsw i32 [[INNER_IV_POSTLOOP]], 1 +; CHECK-NEXT: [[INNER_COND_POSTLOOP:%.*]] = icmp ult i32 [[INNER_IV_NEXT_POSTLOOP]], 31 +; CHECK-NEXT: br i1 [[INNER_COND_POSTLOOP]], label [[INNER_LOOP_POSTLOOP]], label [[RANGE_CHECK_BLOCK_POSTLOOP:%.*]] +; CHECK: range_check_block.postloop: +; CHECK-NEXT: [[RANGE_CHECK_POSTLOOP:%.*]] = icmp slt i32 [[IV_POSTLOOP]], 331 +; CHECK-NEXT: br i1 [[RANGE_CHECK_POSTLOOP]], label [[LOOP_LATCH_POSTLOOP]], label [[DEOPT_LOOPEXIT:%.*]] +; CHECK: loop_latch.postloop: +; CHECK-NEXT: [[IV_NEXT_POSTLOOP]] = add i32 [[IV_POSTLOOP]], 1 +; CHECK-NEXT: [[LOOP_COND_POSTLOOP:%.*]] = icmp ult i32 [[IV_NEXT_POSTLOOP]], 400 +; CHECK-NEXT: br i1 [[LOOP_COND_POSTLOOP]], label [[LOOP_HEADER_POSTLOOP]], label [[EXIT_LOOPEXIT]], !llvm.loop [[LOOP0:![0-9]+]], !irce.loop.clone [[META5:![0-9]+]] +; entry: br label %loop_header @@ -74,17 +124,33 @@ ; is required. define void @test_02() { - -; CHECK-LABEL: test_02 -; CHECK-NOT: preloop -; CHECK-NOT: postloop -; CHECK: range_check_block: ; preds = %inner_loop -; CHECK-NEXT: %range_check = icmp slt i32 %iv, 450 -; CHECK-NEXT: br i1 true, label %loop_latch -; CHECK: loop_latch: ; preds = %range_check_block -; CHECK-NEXT: %iv_next = add i32 %iv, 1 -; CHECK-NEXT: %loop_cond = icmp ult i32 %iv_next, 400 -; CHECK-NEXT: br i1 %loop_cond, label %loop_header, label %exit +; CHECK-LABEL: define void @test_02() { +; CHECK-NEXT: entry: +; CHECK-NEXT: br label [[LOOP_HEADER:%.*]] +; CHECK: loop_header: +; CHECK-NEXT: [[IV:%.*]] = phi i32 [ 2, [[ENTRY:%.*]] ], [ [[IV_NEXT:%.*]], [[LOOP_LATCH:%.*]] ] +; CHECK-NEXT: [[IV_PREV:%.*]] = phi i32 [ 1, [[ENTRY]] ], [ [[IV]], [[LOOP_LATCH]] ] +; CHECK-NEXT: [[TMP2:%.*]] = icmp sgt i32 [[IV_PREV]], -1 +; CHECK-NEXT: br i1 true, label [[LOOP_HEADER_SPLIT_US:%.*]], label [[EXIT:%.*]] +; CHECK: loop_header.split.us: +; CHECK-NEXT: br label [[INNER_LOOP:%.*]] +; CHECK: inner_loop: +; CHECK-NEXT: [[INNER_IV:%.*]] = phi i32 [ 1, [[LOOP_HEADER_SPLIT_US]] ], [ [[INNER_IV_NEXT:%.*]], [[INNER_LOOP]] ] +; CHECK-NEXT: [[INNER_IV_NEXT]] = add nuw nsw i32 [[INNER_IV]], 1 +; CHECK-NEXT: [[INNER_COND:%.*]] = icmp ult i32 [[INNER_IV_NEXT]], 31 +; CHECK-NEXT: br i1 [[INNER_COND]], label [[INNER_LOOP]], label [[RANGE_CHECK_BLOCK:%.*]] +; CHECK: exit: +; CHECK-NEXT: ret void +; CHECK: range_check_block: +; CHECK-NEXT: [[RANGE_CHECK:%.*]] = icmp slt i32 [[IV]], 450 +; CHECK-NEXT: br i1 true, label [[LOOP_LATCH]], label [[DEOPT:%.*]] +; CHECK: loop_latch: +; CHECK-NEXT: [[IV_NEXT]] = add i32 [[IV]], 1 +; CHECK-NEXT: [[LOOP_COND:%.*]] = icmp ult i32 [[IV_NEXT]], 400 +; CHECK-NEXT: br i1 [[LOOP_COND]], label [[LOOP_HEADER]], label [[EXIT]] +; CHECK: deopt: +; CHECK-NEXT: ret void +; entry: br label %loop_header @@ -124,15 +190,33 @@ ; should not apply to the inner loop. The condition %tmp2 can be eliminated. define void @test_03() { - -; CHECK-LABEL: test_03 -; CHECK-NOT: preloop -; CHECK-NOT: postloop -; CHECK: %tmp2 = icmp sgt i32 %iv.prev, -1 -; CHECK-NEXT: br i1 true, label %loop_header.split.us, label %exit +; CHECK-LABEL: define void @test_03() { +; CHECK-NEXT: entry: +; CHECK-NEXT: br label [[LOOP_HEADER:%.*]] +; CHECK: loop_header: +; CHECK-NEXT: [[IV:%.*]] = phi i32 [ 2, [[ENTRY:%.*]] ], [ [[IV_NEXT:%.*]], [[LOOP_LATCH:%.*]] ] +; CHECK-NEXT: [[IV_PREV:%.*]] = phi i32 [ 1, [[ENTRY]] ], [ [[IV]], [[LOOP_LATCH]] ] +; CHECK-NEXT: [[TMP2:%.*]] = icmp sgt i32 [[IV_PREV]], -1 +; CHECK-NEXT: br i1 true, label [[LOOP_HEADER_SPLIT_US:%.*]], label [[EXIT:%.*]] +; CHECK: loop_header.split.us: +; CHECK-NEXT: br label [[INNER_LOOP:%.*]] +; CHECK: inner_loop: +; CHECK-NEXT: [[INNER_IV:%.*]] = phi i32 [ 1, [[LOOP_HEADER_SPLIT_US]] ], [ [[INNER_IV_NEXT:%.*]], [[INNER_LOOP]] ] +; CHECK-NEXT: [[INNER_IV_NEXT]] = add nuw nsw i32 [[INNER_IV]], 1 +; CHECK-NEXT: [[INNER_COND:%.*]] = icmp ult i32 [[INNER_IV_NEXT]], 31 +; CHECK-NEXT: br i1 [[INNER_COND]], label [[INNER_LOOP]], label [[RANGE_CHECK_BLOCK:%.*]] +; CHECK: exit: +; CHECK-NEXT: ret void ; CHECK: range_check_block: -; CHECK-NEXT: %range_check = icmp slt i32 %iv, 0 -; CHECK-NEXT: br i1 %range_check, label %loop_latch, label %deopt +; CHECK-NEXT: [[RANGE_CHECK:%.*]] = icmp slt i32 [[IV]], 0 +; CHECK-NEXT: br i1 [[RANGE_CHECK]], label [[LOOP_LATCH]], label [[DEOPT:%.*]] +; CHECK: loop_latch: +; CHECK-NEXT: [[IV_NEXT]] = add i32 [[IV]], 1 +; CHECK-NEXT: [[LOOP_COND:%.*]] = icmp ult i32 [[IV_NEXT]], 400 +; CHECK-NEXT: br i1 [[LOOP_COND]], label [[LOOP_HEADER]], label [[EXIT]] +; CHECK: deopt: +; CHECK-NEXT: ret void +; entry: br label %loop_header @@ -172,16 +256,85 @@ ; known positive. define void @test_04(ptr %p) { - -; CHECK-LABEL: test_04 -; CHECK: entry -; CHECK-NOT: preloop -; CHECK: %tmp2 = icmp sgt i32 %iv.prev, -1 -; CHECK-NEXT: br i1 true, label %loop_header.split.us, label %exit +; CHECK-LABEL: define void @test_04 +; CHECK-SAME: (ptr [[P:%.*]]) { +; CHECK-NEXT: entry: +; CHECK-NEXT: [[N:%.*]] = load i32, ptr [[P]], align 4 +; CHECK-NEXT: [[SMIN:%.*]] = call i32 @llvm.smin.i32(i32 [[N]], i32 0) +; CHECK-NEXT: [[TMP0:%.*]] = sub i32 [[N]], [[SMIN]] +; CHECK-NEXT: [[SMAX:%.*]] = call i32 @llvm.smax.i32(i32 [[SMIN]], i32 -1) +; CHECK-NEXT: [[TMP1:%.*]] = add nsw i32 [[SMAX]], 1 +; CHECK-NEXT: [[TMP2:%.*]] = mul i32 [[TMP0]], [[TMP1]] +; CHECK-NEXT: [[UMIN:%.*]] = call i32 @llvm.umin.i32(i32 [[TMP2]], i32 400) +; CHECK-NEXT: [[EXIT_MAINLOOP_AT:%.*]] = call i32 @llvm.umax.i32(i32 [[UMIN]], i32 2) +; CHECK-NEXT: [[TMP3:%.*]] = icmp ult i32 2, [[EXIT_MAINLOOP_AT]] +; CHECK-NEXT: br i1 [[TMP3]], label [[LOOP_HEADER_PREHEADER:%.*]], label [[MAIN_PSEUDO_EXIT:%.*]] +; CHECK: loop_header.preheader: +; CHECK-NEXT: br label [[LOOP_HEADER:%.*]] +; CHECK: loop_header: +; CHECK-NEXT: [[IV:%.*]] = phi i32 [ [[IV_NEXT:%.*]], [[LOOP_LATCH:%.*]] ], [ 2, [[LOOP_HEADER_PREHEADER]] ] +; CHECK-NEXT: [[IV_PREV:%.*]] = phi i32 [ [[IV]], [[LOOP_LATCH]] ], [ 1, [[LOOP_HEADER_PREHEADER]] ] +; CHECK-NEXT: [[TMP2:%.*]] = icmp sgt i32 [[IV_PREV]], -1 +; CHECK-NEXT: br i1 true, label [[LOOP_HEADER_SPLIT_US:%.*]], label [[EXIT_LOOPEXIT1:%.*]] +; CHECK: loop_header.split.us: +; CHECK-NEXT: br label [[INNER_LOOP:%.*]] +; CHECK: inner_loop: +; CHECK-NEXT: [[INNER_IV:%.*]] = phi i32 [ 1, [[LOOP_HEADER_SPLIT_US]] ], [ [[INNER_IV_NEXT:%.*]], [[INNER_LOOP]] ] +; CHECK-NEXT: [[INNER_IV_NEXT]] = add nuw nsw i32 [[INNER_IV]], 1 +; CHECK-NEXT: [[INNER_COND:%.*]] = icmp ult i32 [[INNER_IV_NEXT]], 31 +; CHECK-NEXT: br i1 [[INNER_COND]], label [[INNER_LOOP]], label [[RANGE_CHECK_BLOCK:%.*]] +; CHECK: exit.loopexit: +; CHECK-NEXT: br label [[EXIT:%.*]] +; CHECK: exit.loopexit1: +; CHECK-NEXT: br label [[EXIT]] +; CHECK: exit: +; CHECK-NEXT: ret void ; CHECK: range_check_block: -; CHECK-NEXT: %range_check = icmp slt i32 %iv, %n -; CHECK-NEXT: br i1 true, label %loop_latch, label %deopt +; CHECK-NEXT: [[RANGE_CHECK:%.*]] = icmp slt i32 [[IV]], [[N]] +; CHECK-NEXT: br i1 true, label [[LOOP_LATCH]], label [[DEOPT_LOOPEXIT2:%.*]] +; CHECK: loop_latch: +; CHECK-NEXT: [[IV_NEXT]] = add i32 [[IV]], 1 +; CHECK-NEXT: [[LOOP_COND:%.*]] = icmp ult i32 [[IV_NEXT]], 400 +; CHECK-NEXT: [[TMP4:%.*]] = icmp ult i32 [[IV_NEXT]], [[EXIT_MAINLOOP_AT]] +; CHECK-NEXT: br i1 [[TMP4]], label [[LOOP_HEADER]], label [[MAIN_EXIT_SELECTOR:%.*]] +; CHECK: main.exit.selector: +; CHECK-NEXT: [[IV_NEXT_LCSSA:%.*]] = phi i32 [ [[IV_NEXT]], [[LOOP_LATCH]] ] +; CHECK-NEXT: [[IV_LCSSA:%.*]] = phi i32 [ [[IV]], [[LOOP_LATCH]] ] +; CHECK-NEXT: [[TMP5:%.*]] = icmp ult i32 [[IV_NEXT_LCSSA]], 400 +; CHECK-NEXT: br i1 [[TMP5]], label [[MAIN_PSEUDO_EXIT]], label [[EXIT]] +; CHECK: main.pseudo.exit: +; CHECK-NEXT: [[IV_COPY:%.*]] = phi i32 [ 2, [[ENTRY:%.*]] ], [ [[IV_NEXT_LCSSA]], [[MAIN_EXIT_SELECTOR]] ] +; CHECK-NEXT: [[IV_PREV_COPY:%.*]] = phi i32 [ 1, [[ENTRY]] ], [ [[IV_LCSSA]], [[MAIN_EXIT_SELECTOR]] ] +; CHECK-NEXT: [[INDVAR_END:%.*]] = phi i32 [ 2, [[ENTRY]] ], [ [[IV_NEXT_LCSSA]], [[MAIN_EXIT_SELECTOR]] ] +; CHECK-NEXT: br label [[POSTLOOP:%.*]] +; CHECK: deopt.loopexit: +; CHECK-NEXT: br label [[DEOPT:%.*]] +; CHECK: deopt.loopexit2: +; CHECK-NEXT: br label [[DEOPT]] +; CHECK: deopt: +; CHECK-NEXT: ret void ; CHECK: postloop: +; CHECK-NEXT: br label [[LOOP_HEADER_POSTLOOP:%.*]] +; CHECK: loop_header.postloop: +; CHECK-NEXT: [[IV_POSTLOOP:%.*]] = phi i32 [ [[IV_COPY]], [[POSTLOOP]] ], [ [[IV_NEXT_POSTLOOP:%.*]], [[LOOP_LATCH_POSTLOOP:%.*]] ] +; CHECK-NEXT: [[IV_PREV_POSTLOOP:%.*]] = phi i32 [ [[IV_PREV_COPY]], [[POSTLOOP]] ], [ [[IV_POSTLOOP]], [[LOOP_LATCH_POSTLOOP]] ] +; CHECK-NEXT: [[TMP2_POSTLOOP:%.*]] = icmp sgt i32 [[IV_PREV_POSTLOOP]], -1 +; CHECK-NEXT: br i1 [[TMP2_POSTLOOP]], label [[LOOP_HEADER_SPLIT_US_POSTLOOP:%.*]], label [[EXIT_LOOPEXIT:%.*]] +; CHECK: loop_header.split.us.postloop: +; CHECK-NEXT: br label [[INNER_LOOP_POSTLOOP:%.*]] +; CHECK: inner_loop.postloop: +; CHECK-NEXT: [[INNER_IV_POSTLOOP:%.*]] = phi i32 [ 1, [[LOOP_HEADER_SPLIT_US_POSTLOOP]] ], [ [[INNER_IV_NEXT_POSTLOOP:%.*]], [[INNER_LOOP_POSTLOOP]] ] +; CHECK-NEXT: [[INNER_IV_NEXT_POSTLOOP]] = add nuw nsw i32 [[INNER_IV_POSTLOOP]], 1 +; CHECK-NEXT: [[INNER_COND_POSTLOOP:%.*]] = icmp ult i32 [[INNER_IV_NEXT_POSTLOOP]], 31 +; CHECK-NEXT: br i1 [[INNER_COND_POSTLOOP]], label [[INNER_LOOP_POSTLOOP]], label [[RANGE_CHECK_BLOCK_POSTLOOP:%.*]] +; CHECK: range_check_block.postloop: +; CHECK-NEXT: [[RANGE_CHECK_POSTLOOP:%.*]] = icmp slt i32 [[IV_POSTLOOP]], [[N]] +; CHECK-NEXT: br i1 [[RANGE_CHECK_POSTLOOP]], label [[LOOP_LATCH_POSTLOOP]], label [[DEOPT_LOOPEXIT:%.*]] +; CHECK: loop_latch.postloop: +; CHECK-NEXT: [[IV_NEXT_POSTLOOP]] = add i32 [[IV_POSTLOOP]], 1 +; CHECK-NEXT: [[LOOP_COND_POSTLOOP:%.*]] = icmp ult i32 [[IV_NEXT_POSTLOOP]], 400 +; CHECK-NEXT: br i1 [[LOOP_COND_POSTLOOP]], label [[LOOP_HEADER_POSTLOOP]], label [[EXIT_LOOPEXIT]], !llvm.loop [[LOOP6:![0-9]+]], !irce.loop.clone [[META5]] +; entry: %n = load i32, ptr %p @@ -222,31 +375,79 @@ ; intersect ranges (with insertion of postloop). define void @test_05(ptr %p) { - -; CHECK-LABEL: test_05 -; CHECK-NOT: preloop -; CHECK: entry: -; CHECK-NEXT: %n = load i32, ptr %p, align 4, !range ! -; CHECK-NEXT: %exit.mainloop.at = call i32 @llvm.umax.i32(i32 %n, i32 2) -; CHECK-NEXT: [[CMP_2:%[^ ]+]] = icmp ult i32 2, %exit.mainloop.at -; CHECK-NEXT: br i1 [[CMP_2]], label %loop_header.preheader, label %main.pseudo.exit -; CHECK: range_check_block: ; preds = %inner_loop -; CHECK-NEXT: %range_check = icmp slt i32 %iv, %n -; CHECK-NEXT: br i1 true, label %loop_latch, label %deopt.loopexit2 -; CHECK: loop_latch: ; preds = %range_check_block -; CHECK-NEXT: %iv_next = add i32 %iv, 1 -; CHECK-NEXT: %loop_cond = icmp ult i32 %iv_next, 400 -; CHECK-NEXT: [[COND:%[^ ]+]] = icmp ult i32 %iv_next, %exit.mainloop.at -; CHECK-NEXT: br i1 [[COND]], label %loop_header, label %main.exit.selector -; CHECK: main.exit.selector: ; preds = %loop_latch -; CHECK-NEXT: %iv_next.lcssa = phi i32 [ %iv_next, %loop_latch ] -; CHECK-NEXT: %iv.lcssa = phi i32 [ %iv, %loop_latch ] -; CHECK-NEXT: [[MES_COND:%[^ ]+]] = icmp ult i32 %iv_next.lcssa, 400 -; CHECK-NEXT: br i1 [[MES_COND]], label %main.pseudo.exit, label %exit -; CHECK: loop_latch.postloop: ; preds = %range_check_block.postloop -; CHECK-NEXT: %iv_next.postloop = add i32 %iv.postloop, 1 -; CHECK-NEXT: %loop_cond.postloop = icmp ult i32 %iv_next.postloop, 400 -; CHECK-NEXT: br i1 %loop_cond.postloop, label %loop_header.postloop, label %exit.loopexit +; CHECK-LABEL: define void @test_05 +; CHECK-SAME: (ptr [[P:%.*]]) { +; CHECK-NEXT: entry: +; CHECK-NEXT: [[N:%.*]] = load i32, ptr [[P]], align 4, !range [[RNG7:![0-9]+]] +; CHECK-NEXT: [[EXIT_MAINLOOP_AT:%.*]] = call i32 @llvm.umax.i32(i32 [[N]], i32 2) +; CHECK-NEXT: [[TMP0:%.*]] = icmp ult i32 2, [[EXIT_MAINLOOP_AT]] +; CHECK-NEXT: br i1 [[TMP0]], label [[LOOP_HEADER_PREHEADER:%.*]], label [[MAIN_PSEUDO_EXIT:%.*]] +; CHECK: loop_header.preheader: +; CHECK-NEXT: br label [[LOOP_HEADER:%.*]] +; CHECK: loop_header: +; CHECK-NEXT: [[IV:%.*]] = phi i32 [ [[IV_NEXT:%.*]], [[LOOP_LATCH:%.*]] ], [ 2, [[LOOP_HEADER_PREHEADER]] ] +; CHECK-NEXT: [[IV_PREV:%.*]] = phi i32 [ [[IV]], [[LOOP_LATCH]] ], [ 1, [[LOOP_HEADER_PREHEADER]] ] +; CHECK-NEXT: [[TMP2:%.*]] = icmp sgt i32 [[IV_PREV]], -1 +; CHECK-NEXT: br i1 true, label [[LOOP_HEADER_SPLIT_US:%.*]], label [[EXIT_LOOPEXIT1:%.*]] +; CHECK: loop_header.split.us: +; CHECK-NEXT: br label [[INNER_LOOP:%.*]] +; CHECK: inner_loop: +; CHECK-NEXT: [[INNER_IV:%.*]] = phi i32 [ 1, [[LOOP_HEADER_SPLIT_US]] ], [ [[INNER_IV_NEXT:%.*]], [[INNER_LOOP]] ] +; CHECK-NEXT: [[INNER_IV_NEXT]] = add nuw nsw i32 [[INNER_IV]], 1 +; CHECK-NEXT: [[INNER_COND:%.*]] = icmp ult i32 [[INNER_IV_NEXT]], 31 +; CHECK-NEXT: br i1 [[INNER_COND]], label [[INNER_LOOP]], label [[RANGE_CHECK_BLOCK:%.*]] +; CHECK: exit.loopexit: +; CHECK-NEXT: br label [[EXIT:%.*]] +; CHECK: exit.loopexit1: +; CHECK-NEXT: br label [[EXIT]] +; CHECK: exit: +; CHECK-NEXT: ret void +; CHECK: range_check_block: +; CHECK-NEXT: [[RANGE_CHECK:%.*]] = icmp slt i32 [[IV]], [[N]] +; CHECK-NEXT: br i1 true, label [[LOOP_LATCH]], label [[DEOPT_LOOPEXIT2:%.*]] +; CHECK: loop_latch: +; CHECK-NEXT: [[IV_NEXT]] = add i32 [[IV]], 1 +; CHECK-NEXT: [[LOOP_COND:%.*]] = icmp ult i32 [[IV_NEXT]], 400 +; CHECK-NEXT: [[TMP1:%.*]] = icmp ult i32 [[IV_NEXT]], [[EXIT_MAINLOOP_AT]] +; CHECK-NEXT: br i1 [[TMP1]], label [[LOOP_HEADER]], label [[MAIN_EXIT_SELECTOR:%.*]] +; CHECK: main.exit.selector: +; CHECK-NEXT: [[IV_NEXT_LCSSA:%.*]] = phi i32 [ [[IV_NEXT]], [[LOOP_LATCH]] ] +; CHECK-NEXT: [[IV_LCSSA:%.*]] = phi i32 [ [[IV]], [[LOOP_LATCH]] ] +; CHECK-NEXT: [[TMP2:%.*]] = icmp ult i32 [[IV_NEXT_LCSSA]], 400 +; CHECK-NEXT: br i1 [[TMP2]], label [[MAIN_PSEUDO_EXIT]], label [[EXIT]] +; CHECK: main.pseudo.exit: +; CHECK-NEXT: [[IV_COPY:%.*]] = phi i32 [ 2, [[ENTRY:%.*]] ], [ [[IV_NEXT_LCSSA]], [[MAIN_EXIT_SELECTOR]] ] +; CHECK-NEXT: [[IV_PREV_COPY:%.*]] = phi i32 [ 1, [[ENTRY]] ], [ [[IV_LCSSA]], [[MAIN_EXIT_SELECTOR]] ] +; CHECK-NEXT: [[INDVAR_END:%.*]] = phi i32 [ 2, [[ENTRY]] ], [ [[IV_NEXT_LCSSA]], [[MAIN_EXIT_SELECTOR]] ] +; CHECK-NEXT: br label [[POSTLOOP:%.*]] +; CHECK: deopt.loopexit: +; CHECK-NEXT: br label [[DEOPT:%.*]] +; CHECK: deopt.loopexit2: +; CHECK-NEXT: br label [[DEOPT]] +; CHECK: deopt: +; CHECK-NEXT: ret void +; CHECK: postloop: +; CHECK-NEXT: br label [[LOOP_HEADER_POSTLOOP:%.*]] +; CHECK: loop_header.postloop: +; CHECK-NEXT: [[IV_POSTLOOP:%.*]] = phi i32 [ [[IV_COPY]], [[POSTLOOP]] ], [ [[IV_NEXT_POSTLOOP:%.*]], [[LOOP_LATCH_POSTLOOP:%.*]] ] +; CHECK-NEXT: [[IV_PREV_POSTLOOP:%.*]] = phi i32 [ [[IV_PREV_COPY]], [[POSTLOOP]] ], [ [[IV_POSTLOOP]], [[LOOP_LATCH_POSTLOOP]] ] +; CHECK-NEXT: [[TMP2_POSTLOOP:%.*]] = icmp sgt i32 [[IV_PREV_POSTLOOP]], -1 +; CHECK-NEXT: br i1 [[TMP2_POSTLOOP]], label [[LOOP_HEADER_SPLIT_US_POSTLOOP:%.*]], label [[EXIT_LOOPEXIT:%.*]] +; CHECK: loop_header.split.us.postloop: +; CHECK-NEXT: br label [[INNER_LOOP_POSTLOOP:%.*]] +; CHECK: inner_loop.postloop: +; CHECK-NEXT: [[INNER_IV_POSTLOOP:%.*]] = phi i32 [ 1, [[LOOP_HEADER_SPLIT_US_POSTLOOP]] ], [ [[INNER_IV_NEXT_POSTLOOP:%.*]], [[INNER_LOOP_POSTLOOP]] ] +; CHECK-NEXT: [[INNER_IV_NEXT_POSTLOOP]] = add nuw nsw i32 [[INNER_IV_POSTLOOP]], 1 +; CHECK-NEXT: [[INNER_COND_POSTLOOP:%.*]] = icmp ult i32 [[INNER_IV_NEXT_POSTLOOP]], 31 +; CHECK-NEXT: br i1 [[INNER_COND_POSTLOOP]], label [[INNER_LOOP_POSTLOOP]], label [[RANGE_CHECK_BLOCK_POSTLOOP:%.*]] +; CHECK: range_check_block.postloop: +; CHECK-NEXT: [[RANGE_CHECK_POSTLOOP:%.*]] = icmp slt i32 [[IV_POSTLOOP]], [[N]] +; CHECK-NEXT: br i1 [[RANGE_CHECK_POSTLOOP]], label [[LOOP_LATCH_POSTLOOP]], label [[DEOPT_LOOPEXIT:%.*]] +; CHECK: loop_latch.postloop: +; CHECK-NEXT: [[IV_NEXT_POSTLOOP]] = add i32 [[IV_POSTLOOP]], 1 +; CHECK-NEXT: [[LOOP_COND_POSTLOOP:%.*]] = icmp ult i32 [[IV_NEXT_POSTLOOP]], 400 +; CHECK-NEXT: br i1 [[LOOP_COND_POSTLOOP]], label [[LOOP_HEADER_POSTLOOP]], label [[EXIT_LOOPEXIT]], !llvm.loop [[LOOP8:![0-9]+]], !irce.loop.clone [[META5]] +; entry: %n = load i32, ptr %p, !range !0 @@ -284,3 +485,14 @@ } !0 = !{i32 0, i32 50} +;. +; CHECK: [[LOOP0]] = distinct !{[[LOOP0]], [[META1:![0-9]+]], [[META2:![0-9]+]], [[META3:![0-9]+]], [[META4:![0-9]+]]} +; CHECK: [[META1]] = !{!"llvm.loop.unroll.disable"} +; CHECK: [[META2]] = !{!"llvm.loop.vectorize.enable", i1 false} +; CHECK: [[META3]] = !{!"llvm.loop.licm_versioning.disable"} +; CHECK: [[META4]] = !{!"llvm.loop.distribute.enable", i1 false} +; CHECK: [[META5]] = !{} +; CHECK: [[LOOP6]] = distinct !{[[LOOP6]], [[META1]], [[META2]], [[META3]], [[META4]]} +; CHECK: [[RNG7]] = !{i32 0, i32 50} +; CHECK: [[LOOP8]] = distinct !{[[LOOP8]], [[META1]], [[META2]], [[META3]], [[META4]]} +;. Index: llvm/test/Transforms/IRCE/unsigned_comparisons_ugt.ll =================================================================== --- llvm/test/Transforms/IRCE/unsigned_comparisons_ugt.ll +++ llvm/test/Transforms/IRCE/unsigned_comparisons_ugt.ll @@ -1,3 +1,4 @@ +; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --version 3 ; RUN: opt -verify-loop-info -irce-print-changed-loops -passes=irce -S < %s 2>&1 | FileCheck %s ; RUN: opt -verify-loop-info -irce-print-changed-loops -passes='require,irce' -S < %s 2>&1 | FileCheck %s @@ -10,23 +11,57 @@ ; UGT condition for increasing loop. define void @test_01(ptr %arr, ptr %a_len_ptr) #0 { - -; CHECK: test_01( -; CHECK: entry: -; CHECK-NEXT: %exit.mainloop.at = load i32, ptr %a_len_ptr, align 4, !range !0 -; CHECK-NEXT: [[COND:%[^ ]+]] = icmp ult i32 0, %exit.mainloop.at -; CHECK-NEXT: br i1 [[COND]], label %loop.preheader, label %main.pseudo.exit -; CHECK: loop: -; CHECK-NEXT: %idx = phi i32 [ %idx.next, %in.bounds ], [ 0, %loop.preheader ] -; CHECK-NEXT: %idx.next = add nuw nsw i32 %idx, 1 -; CHECK-NEXT: %abc = icmp ult i32 %idx, %exit.mainloop.at -; CHECK-NEXT: br i1 true, label %in.bounds, label %out.of.bounds.loopexit1 -; CHECK-NOT: loop.preloop: -; CHECK: loop.postloop: -; CHECK-NEXT: %idx.postloop = phi i32 [ %idx.copy, %postloop ], [ %idx.next.postloop, %in.bounds.postloop ] -; CHECK-NEXT: %idx.next.postloop = add nuw nsw i32 %idx.postloop, 1 -; CHECK-NEXT: %abc.postloop = icmp ult i32 %idx.postloop, %exit.mainloop.at -; CHECK-NEXT: br i1 %abc.postloop, label %in.bounds.postloop, label %out.of.bounds.loopexit +; CHECK-LABEL: define void @test_01 +; CHECK-SAME: (ptr [[ARR:%.*]], ptr [[A_LEN_PTR:%.*]]) { +; CHECK-NEXT: entry: +; CHECK-NEXT: [[EXIT_MAINLOOP_AT:%.*]] = load i32, ptr [[A_LEN_PTR]], align 4, !range [[RNG0:![0-9]+]] +; CHECK-NEXT: [[TMP0:%.*]] = icmp ult i32 0, [[EXIT_MAINLOOP_AT]] +; CHECK-NEXT: br i1 [[TMP0]], label [[LOOP_PREHEADER:%.*]], label [[MAIN_PSEUDO_EXIT:%.*]] +; CHECK: loop.preheader: +; CHECK-NEXT: br label [[LOOP:%.*]] +; CHECK: loop: +; CHECK-NEXT: [[IDX:%.*]] = phi i32 [ [[IDX_NEXT:%.*]], [[IN_BOUNDS:%.*]] ], [ 0, [[LOOP_PREHEADER]] ] +; CHECK-NEXT: [[IDX_NEXT]] = add nuw nsw i32 [[IDX]], 1 +; CHECK-NEXT: [[ABC:%.*]] = icmp ult i32 [[IDX]], [[EXIT_MAINLOOP_AT]] +; CHECK-NEXT: br i1 true, label [[IN_BOUNDS]], label [[OUT_OF_BOUNDS_LOOPEXIT1:%.*]] +; CHECK: in.bounds: +; CHECK-NEXT: [[ADDR:%.*]] = getelementptr i32, ptr [[ARR]], i32 [[IDX]] +; CHECK-NEXT: store i32 0, ptr [[ADDR]], align 4 +; CHECK-NEXT: [[NEXT:%.*]] = icmp ugt i32 [[IDX_NEXT]], 100 +; CHECK-NEXT: [[TMP1:%.*]] = icmp ult i32 [[IDX_NEXT]], [[EXIT_MAINLOOP_AT]] +; CHECK-NEXT: [[TMP2:%.*]] = xor i1 [[TMP1]], true +; CHECK-NEXT: br i1 [[TMP2]], label [[MAIN_EXIT_SELECTOR:%.*]], label [[LOOP]] +; CHECK: main.exit.selector: +; CHECK-NEXT: [[IDX_NEXT_LCSSA:%.*]] = phi i32 [ [[IDX_NEXT]], [[IN_BOUNDS]] ] +; CHECK-NEXT: [[TMP3:%.*]] = icmp ult i32 [[IDX_NEXT_LCSSA]], 101 +; CHECK-NEXT: br i1 [[TMP3]], label [[MAIN_PSEUDO_EXIT]], label [[EXIT:%.*]] +; CHECK: main.pseudo.exit: +; CHECK-NEXT: [[IDX_COPY:%.*]] = phi i32 [ 0, [[ENTRY:%.*]] ], [ [[IDX_NEXT_LCSSA]], [[MAIN_EXIT_SELECTOR]] ] +; CHECK-NEXT: [[INDVAR_END:%.*]] = phi i32 [ 0, [[ENTRY]] ], [ [[IDX_NEXT_LCSSA]], [[MAIN_EXIT_SELECTOR]] ] +; CHECK-NEXT: br label [[POSTLOOP:%.*]] +; CHECK: out.of.bounds.loopexit: +; CHECK-NEXT: br label [[OUT_OF_BOUNDS:%.*]] +; CHECK: out.of.bounds.loopexit1: +; CHECK-NEXT: br label [[OUT_OF_BOUNDS]] +; CHECK: out.of.bounds: +; CHECK-NEXT: ret void +; CHECK: exit.loopexit: +; CHECK-NEXT: br label [[EXIT]] +; CHECK: exit: +; CHECK-NEXT: ret void +; CHECK: postloop: +; CHECK-NEXT: br label [[LOOP_POSTLOOP:%.*]] +; CHECK: loop.postloop: +; CHECK-NEXT: [[IDX_POSTLOOP:%.*]] = phi i32 [ [[IDX_COPY]], [[POSTLOOP]] ], [ [[IDX_NEXT_POSTLOOP:%.*]], [[IN_BOUNDS_POSTLOOP:%.*]] ] +; CHECK-NEXT: [[IDX_NEXT_POSTLOOP]] = add nuw nsw i32 [[IDX_POSTLOOP]], 1 +; CHECK-NEXT: [[ABC_POSTLOOP:%.*]] = icmp ult i32 [[IDX_POSTLOOP]], [[EXIT_MAINLOOP_AT]] +; CHECK-NEXT: br i1 [[ABC_POSTLOOP]], label [[IN_BOUNDS_POSTLOOP]], label [[OUT_OF_BOUNDS_LOOPEXIT:%.*]] +; CHECK: in.bounds.postloop: +; CHECK-NEXT: [[ADDR_POSTLOOP:%.*]] = getelementptr i32, ptr [[ARR]], i32 [[IDX_POSTLOOP]] +; CHECK-NEXT: store i32 0, ptr [[ADDR_POSTLOOP]], align 4 +; CHECK-NEXT: [[NEXT_POSTLOOP:%.*]] = icmp ugt i32 [[IDX_NEXT_POSTLOOP]], 100 +; CHECK-NEXT: br i1 [[NEXT_POSTLOOP]], label [[EXIT_LOOPEXIT:%.*]], label [[LOOP_POSTLOOP]], !llvm.loop [[LOOP1:![0-9]+]], !irce.loop.clone [[META6:![0-9]+]] +; entry: %len = load i32, ptr %a_len_ptr, !range !0 @@ -53,27 +88,58 @@ ; UGT condition for decreasing loop. define void @test_02(ptr %arr, ptr %a_len_ptr) #0 { - -; CHECK: test_02( -; CHECK: entry: -; CHECK-NEXT: %len = load i32, ptr %a_len_ptr, align 4, !range !0 -; CHECK-NEXT: [[UMIN:%[^ ]+]] = call i32 @llvm.umax.i32(i32 %len, i32 1) -; CHECK-NEXT: %exit.preloop.at = add nsw i32 [[UMIN]], -1 -; CHECK-NEXT: [[COND2:%[^ ]+]] = icmp ugt i32 100, %exit.preloop.at -; CHECK-NEXT: br i1 [[COND2]], label %loop.preloop.preheader, label %preloop.pseudo.exit -; CHECK: mainloop: -; CHECK-NEXT: br label %loop -; CHECK: loop: -; CHECK-NEXT: %idx = phi i32 [ %idx.preloop.copy, %mainloop ], [ %idx.next, %in.bounds ] -; CHECK-NEXT: %idx.next = add i32 %idx, -1 -; CHECK-NEXT: %abc = icmp ult i32 %idx, %len -; CHECK-NEXT: br i1 true, label %in.bounds, label %out.of.bounds.loopexit1 -; CHECK-NOT: loop.postloop: -; CHECK: loop.preloop: -; CHECK-NEXT: %idx.preloop = phi i32 [ %idx.next.preloop, %in.bounds.preloop ], [ 100, %loop.preloop.preheader ] -; CHECK-NEXT: %idx.next.preloop = add i32 %idx.preloop, -1 -; CHECK-NEXT: %abc.preloop = icmp ult i32 %idx.preloop, %len -; CHECK-NEXT: br i1 %abc.preloop, label %in.bounds.preloop, label %out.of.bounds.loopexit +; CHECK-LABEL: define void @test_02 +; CHECK-SAME: (ptr [[ARR:%.*]], ptr [[A_LEN_PTR:%.*]]) { +; CHECK-NEXT: entry: +; CHECK-NEXT: [[LEN:%.*]] = load i32, ptr [[A_LEN_PTR]], align 4, !range [[RNG0]] +; CHECK-NEXT: [[UMAX:%.*]] = call i32 @llvm.umax.i32(i32 [[LEN]], i32 1) +; CHECK-NEXT: [[EXIT_PRELOOP_AT:%.*]] = add nsw i32 [[UMAX]], -1 +; CHECK-NEXT: [[TMP0:%.*]] = icmp ugt i32 100, [[EXIT_PRELOOP_AT]] +; CHECK-NEXT: br i1 [[TMP0]], label [[LOOP_PRELOOP_PREHEADER:%.*]], label [[PRELOOP_PSEUDO_EXIT:%.*]] +; CHECK: loop.preloop.preheader: +; CHECK-NEXT: br label [[LOOP_PRELOOP:%.*]] +; CHECK: mainloop: +; CHECK-NEXT: br label [[LOOP:%.*]] +; CHECK: loop: +; CHECK-NEXT: [[IDX:%.*]] = phi i32 [ [[IDX_PRELOOP_COPY:%.*]], [[MAINLOOP:%.*]] ], [ [[IDX_NEXT:%.*]], [[IN_BOUNDS:%.*]] ] +; CHECK-NEXT: [[IDX_NEXT]] = add i32 [[IDX]], -1 +; CHECK-NEXT: [[ABC:%.*]] = icmp ult i32 [[IDX]], [[LEN]] +; CHECK-NEXT: br i1 true, label [[IN_BOUNDS]], label [[OUT_OF_BOUNDS_LOOPEXIT1:%.*]] +; CHECK: in.bounds: +; CHECK-NEXT: [[ADDR:%.*]] = getelementptr i32, ptr [[ARR]], i32 [[IDX]] +; CHECK-NEXT: store i32 0, ptr [[ADDR]], align 4 +; CHECK-NEXT: [[NEXT:%.*]] = icmp ugt i32 [[IDX_NEXT]], 0 +; CHECK-NEXT: br i1 [[NEXT]], label [[LOOP]], label [[EXIT_LOOPEXIT:%.*]] +; CHECK: out.of.bounds.loopexit: +; CHECK-NEXT: br label [[OUT_OF_BOUNDS:%.*]] +; CHECK: out.of.bounds.loopexit1: +; CHECK-NEXT: br label [[OUT_OF_BOUNDS]] +; CHECK: out.of.bounds: +; CHECK-NEXT: ret void +; CHECK: exit.loopexit: +; CHECK-NEXT: br label [[EXIT:%.*]] +; CHECK: exit: +; CHECK-NEXT: ret void +; CHECK: loop.preloop: +; CHECK-NEXT: [[IDX_PRELOOP:%.*]] = phi i32 [ [[IDX_NEXT_PRELOOP:%.*]], [[IN_BOUNDS_PRELOOP:%.*]] ], [ 100, [[LOOP_PRELOOP_PREHEADER]] ] +; CHECK-NEXT: [[IDX_NEXT_PRELOOP]] = add i32 [[IDX_PRELOOP]], -1 +; CHECK-NEXT: [[ABC_PRELOOP:%.*]] = icmp ult i32 [[IDX_PRELOOP]], [[LEN]] +; CHECK-NEXT: br i1 [[ABC_PRELOOP]], label [[IN_BOUNDS_PRELOOP]], label [[OUT_OF_BOUNDS_LOOPEXIT:%.*]] +; CHECK: in.bounds.preloop: +; CHECK-NEXT: [[ADDR_PRELOOP:%.*]] = getelementptr i32, ptr [[ARR]], i32 [[IDX_PRELOOP]] +; CHECK-NEXT: store i32 0, ptr [[ADDR_PRELOOP]], align 4 +; CHECK-NEXT: [[NEXT_PRELOOP:%.*]] = icmp ugt i32 [[IDX_NEXT_PRELOOP]], 0 +; CHECK-NEXT: [[TMP1:%.*]] = icmp ugt i32 [[IDX_NEXT_PRELOOP]], [[EXIT_PRELOOP_AT]] +; CHECK-NEXT: br i1 [[TMP1]], label [[LOOP_PRELOOP]], label [[PRELOOP_EXIT_SELECTOR:%.*]], !llvm.loop [[LOOP7:![0-9]+]], !irce.loop.clone [[META6]] +; CHECK: preloop.exit.selector: +; CHECK-NEXT: [[IDX_NEXT_PRELOOP_LCSSA:%.*]] = phi i32 [ [[IDX_NEXT_PRELOOP]], [[IN_BOUNDS_PRELOOP]] ] +; CHECK-NEXT: [[TMP2:%.*]] = icmp ugt i32 [[IDX_NEXT_PRELOOP_LCSSA]], 0 +; CHECK-NEXT: br i1 [[TMP2]], label [[PRELOOP_PSEUDO_EXIT]], label [[EXIT]] +; CHECK: preloop.pseudo.exit: +; CHECK-NEXT: [[IDX_PRELOOP_COPY]] = phi i32 [ 100, [[ENTRY:%.*]] ], [ [[IDX_NEXT_PRELOOP_LCSSA]], [[PRELOOP_EXIT_SELECTOR]] ] +; CHECK-NEXT: [[INDVAR_END:%.*]] = phi i32 [ 100, [[ENTRY]] ], [ [[IDX_NEXT_PRELOOP_LCSSA]], [[PRELOOP_EXIT_SELECTOR]] ] +; CHECK-NEXT: br label [[MAINLOOP]] +; entry: %len = load i32, ptr %a_len_ptr, !range !0 @@ -100,23 +166,57 @@ ; Check SINT_MAX + 1, test is similar to test_01. define void @test_03(ptr %arr, ptr %a_len_ptr) #0 { - -; CHECK: test_03( -; CHECK: entry: -; CHECK-NEXT: %exit.mainloop.at = load i32, ptr %a_len_ptr, align 4, !range !0 -; CHECK-NEXT: [[COND:%[^ ]+]] = icmp ult i32 0, %exit.mainloop.at -; CHECK-NEXT: br i1 [[COND]], label %loop.preheader, label %main.pseudo.exit -; CHECK: loop: -; CHECK-NEXT: %idx = phi i32 [ %idx.next, %in.bounds ], [ 0, %loop.preheader ] -; CHECK-NEXT: %idx.next = add nuw nsw i32 %idx, 1 -; CHECK-NEXT: %abc = icmp ult i32 %idx, %exit.mainloop.at -; CHECK-NEXT: br i1 true, label %in.bounds, label %out.of.bounds.loopexit1 -; CHECK-NOT: loop.preloop: -; CHECK: loop.postloop: -; CHECK-NEXT: %idx.postloop = phi i32 [ %idx.copy, %postloop ], [ %idx.next.postloop, %in.bounds.postloop ] -; CHECK-NEXT: %idx.next.postloop = add nuw nsw i32 %idx.postloop, 1 -; CHECK-NEXT: %abc.postloop = icmp ult i32 %idx.postloop, %exit.mainloop.at -; CHECK-NEXT: br i1 %abc.postloop, label %in.bounds.postloop, label %out.of.bounds.loopexit +; CHECK-LABEL: define void @test_03 +; CHECK-SAME: (ptr [[ARR:%.*]], ptr [[A_LEN_PTR:%.*]]) { +; CHECK-NEXT: entry: +; CHECK-NEXT: [[EXIT_MAINLOOP_AT:%.*]] = load i32, ptr [[A_LEN_PTR]], align 4, !range [[RNG0]] +; CHECK-NEXT: [[TMP0:%.*]] = icmp ult i32 0, [[EXIT_MAINLOOP_AT]] +; CHECK-NEXT: br i1 [[TMP0]], label [[LOOP_PREHEADER:%.*]], label [[MAIN_PSEUDO_EXIT:%.*]] +; CHECK: loop.preheader: +; CHECK-NEXT: br label [[LOOP:%.*]] +; CHECK: loop: +; CHECK-NEXT: [[IDX:%.*]] = phi i32 [ [[IDX_NEXT:%.*]], [[IN_BOUNDS:%.*]] ], [ 0, [[LOOP_PREHEADER]] ] +; CHECK-NEXT: [[IDX_NEXT]] = add nuw nsw i32 [[IDX]], 1 +; CHECK-NEXT: [[ABC:%.*]] = icmp ult i32 [[IDX]], [[EXIT_MAINLOOP_AT]] +; CHECK-NEXT: br i1 true, label [[IN_BOUNDS]], label [[OUT_OF_BOUNDS_LOOPEXIT1:%.*]] +; CHECK: in.bounds: +; CHECK-NEXT: [[ADDR:%.*]] = getelementptr i32, ptr [[ARR]], i32 [[IDX]] +; CHECK-NEXT: store i32 0, ptr [[ADDR]], align 4 +; CHECK-NEXT: [[NEXT:%.*]] = icmp ugt i32 [[IDX_NEXT]], -2147483648 +; CHECK-NEXT: [[TMP1:%.*]] = icmp ult i32 [[IDX_NEXT]], [[EXIT_MAINLOOP_AT]] +; CHECK-NEXT: [[TMP2:%.*]] = xor i1 [[TMP1]], true +; CHECK-NEXT: br i1 [[TMP2]], label [[MAIN_EXIT_SELECTOR:%.*]], label [[LOOP]] +; CHECK: main.exit.selector: +; CHECK-NEXT: [[IDX_NEXT_LCSSA:%.*]] = phi i32 [ [[IDX_NEXT]], [[IN_BOUNDS]] ] +; CHECK-NEXT: [[TMP3:%.*]] = icmp ult i32 [[IDX_NEXT_LCSSA]], -2147483647 +; CHECK-NEXT: br i1 [[TMP3]], label [[MAIN_PSEUDO_EXIT]], label [[EXIT:%.*]] +; CHECK: main.pseudo.exit: +; CHECK-NEXT: [[IDX_COPY:%.*]] = phi i32 [ 0, [[ENTRY:%.*]] ], [ [[IDX_NEXT_LCSSA]], [[MAIN_EXIT_SELECTOR]] ] +; CHECK-NEXT: [[INDVAR_END:%.*]] = phi i32 [ 0, [[ENTRY]] ], [ [[IDX_NEXT_LCSSA]], [[MAIN_EXIT_SELECTOR]] ] +; CHECK-NEXT: br label [[POSTLOOP:%.*]] +; CHECK: out.of.bounds.loopexit: +; CHECK-NEXT: br label [[OUT_OF_BOUNDS:%.*]] +; CHECK: out.of.bounds.loopexit1: +; CHECK-NEXT: br label [[OUT_OF_BOUNDS]] +; CHECK: out.of.bounds: +; CHECK-NEXT: ret void +; CHECK: exit.loopexit: +; CHECK-NEXT: br label [[EXIT]] +; CHECK: exit: +; CHECK-NEXT: ret void +; CHECK: postloop: +; CHECK-NEXT: br label [[LOOP_POSTLOOP:%.*]] +; CHECK: loop.postloop: +; CHECK-NEXT: [[IDX_POSTLOOP:%.*]] = phi i32 [ [[IDX_COPY]], [[POSTLOOP]] ], [ [[IDX_NEXT_POSTLOOP:%.*]], [[IN_BOUNDS_POSTLOOP:%.*]] ] +; CHECK-NEXT: [[IDX_NEXT_POSTLOOP]] = add nuw nsw i32 [[IDX_POSTLOOP]], 1 +; CHECK-NEXT: [[ABC_POSTLOOP:%.*]] = icmp ult i32 [[IDX_POSTLOOP]], [[EXIT_MAINLOOP_AT]] +; CHECK-NEXT: br i1 [[ABC_POSTLOOP]], label [[IN_BOUNDS_POSTLOOP]], label [[OUT_OF_BOUNDS_LOOPEXIT:%.*]] +; CHECK: in.bounds.postloop: +; CHECK-NEXT: [[ADDR_POSTLOOP:%.*]] = getelementptr i32, ptr [[ARR]], i32 [[IDX_POSTLOOP]] +; CHECK-NEXT: store i32 0, ptr [[ADDR_POSTLOOP]], align 4 +; CHECK-NEXT: [[NEXT_POSTLOOP:%.*]] = icmp ugt i32 [[IDX_NEXT_POSTLOOP]], -2147483648 +; CHECK-NEXT: br i1 [[NEXT_POSTLOOP]], label [[EXIT_LOOPEXIT:%.*]], label [[LOOP_POSTLOOP]], !llvm.loop [[LOOP8:![0-9]+]], !irce.loop.clone [[META6]] +; entry: %len = load i32, ptr %a_len_ptr, !range !0 @@ -143,27 +243,58 @@ ; Check SINT_MAX + 1, test is similar to test_02. define void @test_04(ptr %arr, ptr %a_len_ptr) #0 { - -; CHECK: test_04( -; CHECK: entry: -; CHECK-NEXT: %len = load i32, ptr %a_len_ptr, align 4, !range !0 -; CHECK-NEXT: [[UMIN:%[^ ]+]] = call i32 @llvm.umax.i32(i32 %len, i32 1) -; CHECK-NEXT: %exit.preloop.at = add nsw i32 [[UMIN]], -1 -; CHECK-NEXT: [[COND2:%[^ ]+]] = icmp ugt i32 -2147483648, %exit.preloop.at -; CHECK-NEXT: br i1 [[COND2]], label %loop.preloop.preheader, label %preloop.pseudo.exit -; CHECK: mainloop: -; CHECK-NEXT: br label %loop -; CHECK: loop: -; CHECK-NEXT: %idx = phi i32 [ %idx.preloop.copy, %mainloop ], [ %idx.next, %in.bounds ] -; CHECK-NEXT: %idx.next = add i32 %idx, -1 -; CHECK-NEXT: %abc = icmp ult i32 %idx, %len -; CHECK-NEXT: br i1 true, label %in.bounds, label %out.of.bounds.loopexit1 -; CHECK-NOT: loop.postloop: -; CHECK: loop.preloop: -; CHECK-NEXT: %idx.preloop = phi i32 [ %idx.next.preloop, %in.bounds.preloop ], [ -2147483648, %loop.preloop.preheader ] -; CHECK-NEXT: %idx.next.preloop = add i32 %idx.preloop, -1 -; CHECK-NEXT: %abc.preloop = icmp ult i32 %idx.preloop, %len -; CHECK-NEXT: br i1 %abc.preloop, label %in.bounds.preloop, label %out.of.bounds.loopexit +; CHECK-LABEL: define void @test_04 +; CHECK-SAME: (ptr [[ARR:%.*]], ptr [[A_LEN_PTR:%.*]]) { +; CHECK-NEXT: entry: +; CHECK-NEXT: [[LEN:%.*]] = load i32, ptr [[A_LEN_PTR]], align 4, !range [[RNG0]] +; CHECK-NEXT: [[UMAX:%.*]] = call i32 @llvm.umax.i32(i32 [[LEN]], i32 1) +; CHECK-NEXT: [[EXIT_PRELOOP_AT:%.*]] = add nsw i32 [[UMAX]], -1 +; CHECK-NEXT: [[TMP0:%.*]] = icmp ugt i32 -2147483648, [[EXIT_PRELOOP_AT]] +; CHECK-NEXT: br i1 [[TMP0]], label [[LOOP_PRELOOP_PREHEADER:%.*]], label [[PRELOOP_PSEUDO_EXIT:%.*]] +; CHECK: loop.preloop.preheader: +; CHECK-NEXT: br label [[LOOP_PRELOOP:%.*]] +; CHECK: mainloop: +; CHECK-NEXT: br label [[LOOP:%.*]] +; CHECK: loop: +; CHECK-NEXT: [[IDX:%.*]] = phi i32 [ [[IDX_PRELOOP_COPY:%.*]], [[MAINLOOP:%.*]] ], [ [[IDX_NEXT:%.*]], [[IN_BOUNDS:%.*]] ] +; CHECK-NEXT: [[IDX_NEXT]] = add i32 [[IDX]], -1 +; CHECK-NEXT: [[ABC:%.*]] = icmp ult i32 [[IDX]], [[LEN]] +; CHECK-NEXT: br i1 true, label [[IN_BOUNDS]], label [[OUT_OF_BOUNDS_LOOPEXIT1:%.*]] +; CHECK: in.bounds: +; CHECK-NEXT: [[ADDR:%.*]] = getelementptr i32, ptr [[ARR]], i32 [[IDX]] +; CHECK-NEXT: store i32 0, ptr [[ADDR]], align 4 +; CHECK-NEXT: [[NEXT:%.*]] = icmp ugt i32 [[IDX_NEXT]], 0 +; CHECK-NEXT: br i1 [[NEXT]], label [[LOOP]], label [[EXIT_LOOPEXIT:%.*]] +; CHECK: out.of.bounds.loopexit: +; CHECK-NEXT: br label [[OUT_OF_BOUNDS:%.*]] +; CHECK: out.of.bounds.loopexit1: +; CHECK-NEXT: br label [[OUT_OF_BOUNDS]] +; CHECK: out.of.bounds: +; CHECK-NEXT: ret void +; CHECK: exit.loopexit: +; CHECK-NEXT: br label [[EXIT:%.*]] +; CHECK: exit: +; CHECK-NEXT: ret void +; CHECK: loop.preloop: +; CHECK-NEXT: [[IDX_PRELOOP:%.*]] = phi i32 [ [[IDX_NEXT_PRELOOP:%.*]], [[IN_BOUNDS_PRELOOP:%.*]] ], [ -2147483648, [[LOOP_PRELOOP_PREHEADER]] ] +; CHECK-NEXT: [[IDX_NEXT_PRELOOP]] = add i32 [[IDX_PRELOOP]], -1 +; CHECK-NEXT: [[ABC_PRELOOP:%.*]] = icmp ult i32 [[IDX_PRELOOP]], [[LEN]] +; CHECK-NEXT: br i1 [[ABC_PRELOOP]], label [[IN_BOUNDS_PRELOOP]], label [[OUT_OF_BOUNDS_LOOPEXIT:%.*]] +; CHECK: in.bounds.preloop: +; CHECK-NEXT: [[ADDR_PRELOOP:%.*]] = getelementptr i32, ptr [[ARR]], i32 [[IDX_PRELOOP]] +; CHECK-NEXT: store i32 0, ptr [[ADDR_PRELOOP]], align 4 +; CHECK-NEXT: [[NEXT_PRELOOP:%.*]] = icmp ugt i32 [[IDX_NEXT_PRELOOP]], 0 +; CHECK-NEXT: [[TMP1:%.*]] = icmp ugt i32 [[IDX_NEXT_PRELOOP]], [[EXIT_PRELOOP_AT]] +; CHECK-NEXT: br i1 [[TMP1]], label [[LOOP_PRELOOP]], label [[PRELOOP_EXIT_SELECTOR:%.*]], !llvm.loop [[LOOP9:![0-9]+]], !irce.loop.clone [[META6]] +; CHECK: preloop.exit.selector: +; CHECK-NEXT: [[IDX_NEXT_PRELOOP_LCSSA:%.*]] = phi i32 [ [[IDX_NEXT_PRELOOP]], [[IN_BOUNDS_PRELOOP]] ] +; CHECK-NEXT: [[TMP2:%.*]] = icmp ugt i32 [[IDX_NEXT_PRELOOP_LCSSA]], 0 +; CHECK-NEXT: br i1 [[TMP2]], label [[PRELOOP_PSEUDO_EXIT]], label [[EXIT]] +; CHECK: preloop.pseudo.exit: +; CHECK-NEXT: [[IDX_PRELOOP_COPY]] = phi i32 [ -2147483648, [[ENTRY:%.*]] ], [ [[IDX_NEXT_PRELOOP_LCSSA]], [[PRELOOP_EXIT_SELECTOR]] ] +; CHECK-NEXT: [[INDVAR_END:%.*]] = phi i32 [ -2147483648, [[ENTRY]] ], [ [[IDX_NEXT_PRELOOP_LCSSA]], [[PRELOOP_EXIT_SELECTOR]] ] +; CHECK-NEXT: br label [[MAINLOOP]] +; entry: %len = load i32, ptr %a_len_ptr, !range !0 @@ -190,10 +321,26 @@ ; Increasing loop, UINT_MAX. Negative test: we cannot add 1 to UINT_MAX. define void @test_05(ptr %arr, ptr %a_len_ptr) #0 { - -; CHECK: test_05( -; CHECK-NOT: loop.preloop: -; CHECK-NOT: loop.postloop: +; CHECK-LABEL: define void @test_05 +; CHECK-SAME: (ptr [[ARR:%.*]], ptr [[A_LEN_PTR:%.*]]) { +; CHECK-NEXT: entry: +; CHECK-NEXT: [[LEN:%.*]] = load i32, ptr [[A_LEN_PTR]], align 4, !range [[RNG0]] +; CHECK-NEXT: br label [[LOOP:%.*]] +; CHECK: loop: +; CHECK-NEXT: [[IDX:%.*]] = phi i32 [ 0, [[ENTRY:%.*]] ], [ [[IDX_NEXT:%.*]], [[IN_BOUNDS:%.*]] ] +; CHECK-NEXT: [[IDX_NEXT]] = add nuw nsw i32 [[IDX]], 1 +; CHECK-NEXT: [[ABC:%.*]] = icmp ult i32 [[IDX]], [[LEN]] +; CHECK-NEXT: br i1 [[ABC]], label [[IN_BOUNDS]], label [[OUT_OF_BOUNDS:%.*]] +; CHECK: in.bounds: +; CHECK-NEXT: [[ADDR:%.*]] = getelementptr i32, ptr [[ARR]], i32 [[IDX]] +; CHECK-NEXT: store i32 0, ptr [[ADDR]], align 4 +; CHECK-NEXT: [[NEXT:%.*]] = icmp ugt i32 [[IDX_NEXT]], -1 +; CHECK-NEXT: br i1 [[NEXT]], label [[EXIT:%.*]], label [[LOOP]] +; CHECK: out.of.bounds: +; CHECK-NEXT: ret void +; CHECK: exit: +; CHECK-NEXT: ret void +; entry: %len = load i32, ptr %a_len_ptr, !range !0 @@ -220,21 +367,55 @@ ; Decreasing loop, UINT_MAX. Positive test. define void @test_06(ptr %arr, ptr %a_len_ptr) #0 { - -; CHECK: test_06( -; CHECK: mainloop: -; CHECK-NEXT: br label %loop -; CHECK: loop: -; CHECK-NEXT: %idx = phi i32 [ %idx.preloop.copy, %mainloop ], [ %idx.next, %in.bounds ] -; CHECK-NEXT: %idx.next = add nuw i32 %idx, -1 -; CHECK-NEXT: %abc = icmp ult i32 %idx, %len -; CHECK-NEXT: br i1 true, label %in.bounds, label %out.of.bounds.loopexit1 -; CHECK-NOT: loop.postloop: -; CHECK: loop.preloop: -; CHECK-NEXT: %idx.preloop = phi i32 [ %idx.next.preloop, %in.bounds.preloop ], [ -1, %loop.preloop.preheader ] -; CHECK-NEXT: %idx.next.preloop = add nuw i32 %idx.preloop, -1 -; CHECK-NEXT: %abc.preloop = icmp ult i32 %idx.preloop, %len -; CHECK-NEXT: br i1 %abc.preloop, label %in.bounds.preloop, label %out.of.bounds.loopexit +; CHECK-LABEL: define void @test_06 +; CHECK-SAME: (ptr [[ARR:%.*]], ptr [[A_LEN_PTR:%.*]]) { +; CHECK-NEXT: entry: +; CHECK-NEXT: [[LEN:%.*]] = load i32, ptr [[A_LEN_PTR]], align 4, !range [[RNG0]] +; CHECK-NEXT: br i1 true, label [[LOOP_PRELOOP_PREHEADER:%.*]], label [[PRELOOP_PSEUDO_EXIT:%.*]] +; CHECK: loop.preloop.preheader: +; CHECK-NEXT: br label [[LOOP_PRELOOP:%.*]] +; CHECK: mainloop: +; CHECK-NEXT: br label [[LOOP:%.*]] +; CHECK: loop: +; CHECK-NEXT: [[IDX:%.*]] = phi i32 [ [[IDX_PRELOOP_COPY:%.*]], [[MAINLOOP:%.*]] ], [ [[IDX_NEXT:%.*]], [[IN_BOUNDS:%.*]] ] +; CHECK-NEXT: [[IDX_NEXT]] = add nuw i32 [[IDX]], -1 +; CHECK-NEXT: [[ABC:%.*]] = icmp ult i32 [[IDX]], [[LEN]] +; CHECK-NEXT: br i1 true, label [[IN_BOUNDS]], label [[OUT_OF_BOUNDS_LOOPEXIT1:%.*]] +; CHECK: in.bounds: +; CHECK-NEXT: [[ADDR:%.*]] = getelementptr i32, ptr [[ARR]], i32 [[IDX]] +; CHECK-NEXT: store i32 0, ptr [[ADDR]], align 4 +; CHECK-NEXT: [[NEXT:%.*]] = icmp ugt i32 [[IDX_NEXT]], 0 +; CHECK-NEXT: br i1 [[NEXT]], label [[LOOP]], label [[EXIT_LOOPEXIT:%.*]] +; CHECK: out.of.bounds.loopexit: +; CHECK-NEXT: br label [[OUT_OF_BOUNDS:%.*]] +; CHECK: out.of.bounds.loopexit1: +; CHECK-NEXT: br label [[OUT_OF_BOUNDS]] +; CHECK: out.of.bounds: +; CHECK-NEXT: ret void +; CHECK: exit.loopexit: +; CHECK-NEXT: br label [[EXIT:%.*]] +; CHECK: exit: +; CHECK-NEXT: ret void +; CHECK: loop.preloop: +; CHECK-NEXT: [[IDX_PRELOOP:%.*]] = phi i32 [ [[IDX_NEXT_PRELOOP:%.*]], [[IN_BOUNDS_PRELOOP:%.*]] ], [ -1, [[LOOP_PRELOOP_PREHEADER]] ] +; CHECK-NEXT: [[IDX_NEXT_PRELOOP]] = add nuw i32 [[IDX_PRELOOP]], -1 +; CHECK-NEXT: [[ABC_PRELOOP:%.*]] = icmp ult i32 [[IDX_PRELOOP]], [[LEN]] +; CHECK-NEXT: br i1 [[ABC_PRELOOP]], label [[IN_BOUNDS_PRELOOP]], label [[OUT_OF_BOUNDS_LOOPEXIT:%.*]] +; CHECK: in.bounds.preloop: +; CHECK-NEXT: [[ADDR_PRELOOP:%.*]] = getelementptr i32, ptr [[ARR]], i32 [[IDX_PRELOOP]] +; CHECK-NEXT: store i32 0, ptr [[ADDR_PRELOOP]], align 4 +; CHECK-NEXT: [[NEXT_PRELOOP:%.*]] = icmp ugt i32 [[IDX_NEXT_PRELOOP]], 0 +; CHECK-NEXT: [[TMP0:%.*]] = icmp ugt i32 [[IDX_NEXT_PRELOOP]], 0 +; CHECK-NEXT: br i1 [[TMP0]], label [[LOOP_PRELOOP]], label [[PRELOOP_EXIT_SELECTOR:%.*]], !llvm.loop [[LOOP10:![0-9]+]], !irce.loop.clone [[META6]] +; CHECK: preloop.exit.selector: +; CHECK-NEXT: [[IDX_NEXT_PRELOOP_LCSSA:%.*]] = phi i32 [ [[IDX_NEXT_PRELOOP]], [[IN_BOUNDS_PRELOOP]] ] +; CHECK-NEXT: [[TMP1:%.*]] = icmp ugt i32 [[IDX_NEXT_PRELOOP_LCSSA]], 0 +; CHECK-NEXT: br i1 [[TMP1]], label [[PRELOOP_PSEUDO_EXIT]], label [[EXIT]] +; CHECK: preloop.pseudo.exit: +; CHECK-NEXT: [[IDX_PRELOOP_COPY]] = phi i32 [ -1, [[ENTRY:%.*]] ], [ [[IDX_NEXT_PRELOOP_LCSSA]], [[PRELOOP_EXIT_SELECTOR]] ] +; CHECK-NEXT: [[INDVAR_END:%.*]] = phi i32 [ -1, [[ENTRY]] ], [ [[IDX_NEXT_PRELOOP_LCSSA]], [[PRELOOP_EXIT_SELECTOR]] ] +; CHECK-NEXT: br label [[MAINLOOP]] +; entry: %len = load i32, ptr %a_len_ptr, !range !0 @@ -260,3 +441,16 @@ } !0 = !{i32 0, i32 50} +;. +; CHECK: [[RNG0]] = !{i32 0, i32 50} +; CHECK: [[LOOP1]] = distinct !{[[LOOP1]], [[META2:![0-9]+]], [[META3:![0-9]+]], [[META4:![0-9]+]], [[META5:![0-9]+]]} +; CHECK: [[META2]] = !{!"llvm.loop.unroll.disable"} +; CHECK: [[META3]] = !{!"llvm.loop.vectorize.enable", i1 false} +; CHECK: [[META4]] = !{!"llvm.loop.licm_versioning.disable"} +; CHECK: [[META5]] = !{!"llvm.loop.distribute.enable", i1 false} +; CHECK: [[META6]] = !{} +; CHECK: [[LOOP7]] = distinct !{[[LOOP7]], [[META2]], [[META3]], [[META4]], [[META5]]} +; CHECK: [[LOOP8]] = distinct !{[[LOOP8]], [[META2]], [[META3]], [[META4]], [[META5]]} +; CHECK: [[LOOP9]] = distinct !{[[LOOP9]], [[META2]], [[META3]], [[META4]], [[META5]]} +; CHECK: [[LOOP10]] = distinct !{[[LOOP10]], [[META2]], [[META3]], [[META4]], [[META5]]} +;. Index: llvm/test/Transforms/IRCE/unsigned_comparisons_ult.ll =================================================================== --- llvm/test/Transforms/IRCE/unsigned_comparisons_ult.ll +++ llvm/test/Transforms/IRCE/unsigned_comparisons_ult.ll @@ -1,3 +1,4 @@ +; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --version 3 ; RUN: opt -verify-loop-info -irce-print-changed-loops -passes=irce -S < %s 2>&1 | FileCheck %s ; RUN: opt -verify-loop-info -irce-print-changed-loops -passes='require,irce' -S < %s 2>&1 | FileCheck %s @@ -13,23 +14,56 @@ ; ULT condition for increasing loop. define void @test_01(ptr %arr, ptr %a_len_ptr) #0 { - -; CHECK: test_01 -; CHECK: entry: -; CHECK-NEXT: %exit.mainloop.at = load i32, ptr %a_len_ptr, align 4, !range !0 -; CHECK-NEXT: [[COND:%[^ ]+]] = icmp ult i32 0, %exit.mainloop.at -; CHECK-NEXT: br i1 [[COND]], label %loop.preheader, label %main.pseudo.exit -; CHECK: loop: -; CHECK-NEXT: %idx = phi i32 [ %idx.next, %in.bounds ], [ 0, %loop.preheader ] -; CHECK-NEXT: %idx.next = add nuw nsw i32 %idx, 1 -; CHECK-NEXT: %abc = icmp ult i32 %idx, %exit.mainloop.at -; CHECK-NEXT: br i1 true, label %in.bounds, label %out.of.bounds.loopexit1 -; CHECK-NOT: loop.preloop: -; CHECK: loop.postloop: -; CHECK-NEXT: %idx.postloop = phi i32 [ %idx.copy, %postloop ], [ %idx.next.postloop, %in.bounds.postloop ] -; CHECK-NEXT: %idx.next.postloop = add nuw nsw i32 %idx.postloop, 1 -; CHECK-NEXT: %abc.postloop = icmp ult i32 %idx.postloop, %exit.mainloop.at -; CHECK-NEXT: br i1 %abc.postloop, label %in.bounds.postloop, label %out.of.bounds.loopexit +; CHECK-LABEL: define void @test_01 +; CHECK-SAME: (ptr [[ARR:%.*]], ptr [[A_LEN_PTR:%.*]]) { +; CHECK-NEXT: entry: +; CHECK-NEXT: [[EXIT_MAINLOOP_AT:%.*]] = load i32, ptr [[A_LEN_PTR]], align 4, !range [[RNG0:![0-9]+]] +; CHECK-NEXT: [[TMP0:%.*]] = icmp ult i32 0, [[EXIT_MAINLOOP_AT]] +; CHECK-NEXT: br i1 [[TMP0]], label [[LOOP_PREHEADER:%.*]], label [[MAIN_PSEUDO_EXIT:%.*]] +; CHECK: loop.preheader: +; CHECK-NEXT: br label [[LOOP:%.*]] +; CHECK: loop: +; CHECK-NEXT: [[IDX:%.*]] = phi i32 [ [[IDX_NEXT:%.*]], [[IN_BOUNDS:%.*]] ], [ 0, [[LOOP_PREHEADER]] ] +; CHECK-NEXT: [[IDX_NEXT]] = add nuw nsw i32 [[IDX]], 1 +; CHECK-NEXT: [[ABC:%.*]] = icmp ult i32 [[IDX]], [[EXIT_MAINLOOP_AT]] +; CHECK-NEXT: br i1 true, label [[IN_BOUNDS]], label [[OUT_OF_BOUNDS_LOOPEXIT1:%.*]] +; CHECK: in.bounds: +; CHECK-NEXT: [[ADDR:%.*]] = getelementptr i32, ptr [[ARR]], i32 [[IDX]] +; CHECK-NEXT: store i32 0, ptr [[ADDR]], align 4 +; CHECK-NEXT: [[NEXT:%.*]] = icmp ult i32 [[IDX_NEXT]], 100 +; CHECK-NEXT: [[TMP1:%.*]] = icmp ult i32 [[IDX_NEXT]], [[EXIT_MAINLOOP_AT]] +; CHECK-NEXT: br i1 [[TMP1]], label [[LOOP]], label [[MAIN_EXIT_SELECTOR:%.*]] +; CHECK: main.exit.selector: +; CHECK-NEXT: [[IDX_NEXT_LCSSA:%.*]] = phi i32 [ [[IDX_NEXT]], [[IN_BOUNDS]] ] +; CHECK-NEXT: [[TMP2:%.*]] = icmp ult i32 [[IDX_NEXT_LCSSA]], 100 +; CHECK-NEXT: br i1 [[TMP2]], label [[MAIN_PSEUDO_EXIT]], label [[EXIT:%.*]] +; CHECK: main.pseudo.exit: +; CHECK-NEXT: [[IDX_COPY:%.*]] = phi i32 [ 0, [[ENTRY:%.*]] ], [ [[IDX_NEXT_LCSSA]], [[MAIN_EXIT_SELECTOR]] ] +; CHECK-NEXT: [[INDVAR_END:%.*]] = phi i32 [ 0, [[ENTRY]] ], [ [[IDX_NEXT_LCSSA]], [[MAIN_EXIT_SELECTOR]] ] +; CHECK-NEXT: br label [[POSTLOOP:%.*]] +; CHECK: out.of.bounds.loopexit: +; CHECK-NEXT: br label [[OUT_OF_BOUNDS:%.*]] +; CHECK: out.of.bounds.loopexit1: +; CHECK-NEXT: br label [[OUT_OF_BOUNDS]] +; CHECK: out.of.bounds: +; CHECK-NEXT: ret void +; CHECK: exit.loopexit: +; CHECK-NEXT: br label [[EXIT]] +; CHECK: exit: +; CHECK-NEXT: ret void +; CHECK: postloop: +; CHECK-NEXT: br label [[LOOP_POSTLOOP:%.*]] +; CHECK: loop.postloop: +; CHECK-NEXT: [[IDX_POSTLOOP:%.*]] = phi i32 [ [[IDX_COPY]], [[POSTLOOP]] ], [ [[IDX_NEXT_POSTLOOP:%.*]], [[IN_BOUNDS_POSTLOOP:%.*]] ] +; CHECK-NEXT: [[IDX_NEXT_POSTLOOP]] = add nuw nsw i32 [[IDX_POSTLOOP]], 1 +; CHECK-NEXT: [[ABC_POSTLOOP:%.*]] = icmp ult i32 [[IDX_POSTLOOP]], [[EXIT_MAINLOOP_AT]] +; CHECK-NEXT: br i1 [[ABC_POSTLOOP]], label [[IN_BOUNDS_POSTLOOP]], label [[OUT_OF_BOUNDS_LOOPEXIT:%.*]] +; CHECK: in.bounds.postloop: +; CHECK-NEXT: [[ADDR_POSTLOOP:%.*]] = getelementptr i32, ptr [[ARR]], i32 [[IDX_POSTLOOP]] +; CHECK-NEXT: store i32 0, ptr [[ADDR_POSTLOOP]], align 4 +; CHECK-NEXT: [[NEXT_POSTLOOP:%.*]] = icmp ult i32 [[IDX_NEXT_POSTLOOP]], 100 +; CHECK-NEXT: br i1 [[NEXT_POSTLOOP]], label [[LOOP_POSTLOOP]], label [[EXIT_LOOPEXIT:%.*]], !llvm.loop [[LOOP1:![0-9]+]], !irce.loop.clone [[META6:![0-9]+]] +; entry: %len = load i32, ptr %a_len_ptr, !range !0 @@ -56,27 +90,59 @@ ; ULT condition for decreasing loops. define void @test_02(ptr %arr, ptr %a_len_ptr) #0 { - -; CHECK: test_02( -; CHECK: entry: -; CHECK-NEXT: %len = load i32, ptr %a_len_ptr, align 4, !range !0 -; CHECK-NEXT: [[UMIN:%[^ ]+]] = call i32 @llvm.umax.i32(i32 %len, i32 1) -; CHECK-NEXT: %exit.preloop.at = add nsw i32 [[UMIN]], -1 -; CHECK-NEXT: [[COND2:%[^ ]+]] = icmp ugt i32 100, %exit.preloop.at -; CHECK-NEXT: br i1 [[COND2]], label %loop.preloop.preheader, label %preloop.pseudo.exit -; CHECK: mainloop: -; CHECK-NEXT: br label %loop -; CHECK: loop: -; CHECK-NEXT: %idx = phi i32 [ %idx.preloop.copy, %mainloop ], [ %idx.next, %in.bounds ] -; CHECK-NEXT: %idx.next = add i32 %idx, -1 -; CHECK-NEXT: %abc = icmp ult i32 %idx, %len -; CHECK-NEXT: br i1 true, label %in.bounds, label %out.of.bounds.loopexit1 -; CHECK-NOT: loop.postloop: -; CHECK: loop.preloop: -; CHECK-NEXT: %idx.preloop = phi i32 [ %idx.next.preloop, %in.bounds.preloop ], [ 100, %loop.preloop.preheader ] -; CHECK-NEXT: %idx.next.preloop = add i32 %idx.preloop, -1 -; CHECK-NEXT: %abc.preloop = icmp ult i32 %idx.preloop, %len -; CHECK-NEXT: br i1 %abc.preloop, label %in.bounds.preloop, label %out.of.bounds.loopexit +; CHECK-LABEL: define void @test_02 +; CHECK-SAME: (ptr [[ARR:%.*]], ptr [[A_LEN_PTR:%.*]]) { +; CHECK-NEXT: entry: +; CHECK-NEXT: [[LEN:%.*]] = load i32, ptr [[A_LEN_PTR]], align 4, !range [[RNG0]] +; CHECK-NEXT: [[UMAX:%.*]] = call i32 @llvm.umax.i32(i32 [[LEN]], i32 1) +; CHECK-NEXT: [[EXIT_PRELOOP_AT:%.*]] = add nsw i32 [[UMAX]], -1 +; CHECK-NEXT: [[TMP0:%.*]] = icmp ugt i32 100, [[EXIT_PRELOOP_AT]] +; CHECK-NEXT: br i1 [[TMP0]], label [[LOOP_PRELOOP_PREHEADER:%.*]], label [[PRELOOP_PSEUDO_EXIT:%.*]] +; CHECK: loop.preloop.preheader: +; CHECK-NEXT: br label [[LOOP_PRELOOP:%.*]] +; CHECK: mainloop: +; CHECK-NEXT: br label [[LOOP:%.*]] +; CHECK: loop: +; CHECK-NEXT: [[IDX:%.*]] = phi i32 [ [[IDX_PRELOOP_COPY:%.*]], [[MAINLOOP:%.*]] ], [ [[IDX_NEXT:%.*]], [[IN_BOUNDS:%.*]] ] +; CHECK-NEXT: [[IDX_NEXT]] = add i32 [[IDX]], -1 +; CHECK-NEXT: [[ABC:%.*]] = icmp ult i32 [[IDX]], [[LEN]] +; CHECK-NEXT: br i1 true, label [[IN_BOUNDS]], label [[OUT_OF_BOUNDS_LOOPEXIT1:%.*]] +; CHECK: in.bounds: +; CHECK-NEXT: [[ADDR:%.*]] = getelementptr i32, ptr [[ARR]], i32 [[IDX]] +; CHECK-NEXT: store i32 0, ptr [[ADDR]], align 4 +; CHECK-NEXT: [[NEXT:%.*]] = icmp ult i32 [[IDX_NEXT]], 1 +; CHECK-NEXT: br i1 [[NEXT]], label [[EXIT_LOOPEXIT:%.*]], label [[LOOP]] +; CHECK: out.of.bounds.loopexit: +; CHECK-NEXT: br label [[OUT_OF_BOUNDS:%.*]] +; CHECK: out.of.bounds.loopexit1: +; CHECK-NEXT: br label [[OUT_OF_BOUNDS]] +; CHECK: out.of.bounds: +; CHECK-NEXT: ret void +; CHECK: exit.loopexit: +; CHECK-NEXT: br label [[EXIT:%.*]] +; CHECK: exit: +; CHECK-NEXT: ret void +; CHECK: loop.preloop: +; CHECK-NEXT: [[IDX_PRELOOP:%.*]] = phi i32 [ [[IDX_NEXT_PRELOOP:%.*]], [[IN_BOUNDS_PRELOOP:%.*]] ], [ 100, [[LOOP_PRELOOP_PREHEADER]] ] +; CHECK-NEXT: [[IDX_NEXT_PRELOOP]] = add i32 [[IDX_PRELOOP]], -1 +; CHECK-NEXT: [[ABC_PRELOOP:%.*]] = icmp ult i32 [[IDX_PRELOOP]], [[LEN]] +; CHECK-NEXT: br i1 [[ABC_PRELOOP]], label [[IN_BOUNDS_PRELOOP]], label [[OUT_OF_BOUNDS_LOOPEXIT:%.*]] +; CHECK: in.bounds.preloop: +; CHECK-NEXT: [[ADDR_PRELOOP:%.*]] = getelementptr i32, ptr [[ARR]], i32 [[IDX_PRELOOP]] +; CHECK-NEXT: store i32 0, ptr [[ADDR_PRELOOP]], align 4 +; CHECK-NEXT: [[NEXT_PRELOOP:%.*]] = icmp ult i32 [[IDX_NEXT_PRELOOP]], 1 +; CHECK-NEXT: [[TMP1:%.*]] = icmp ugt i32 [[IDX_NEXT_PRELOOP]], [[EXIT_PRELOOP_AT]] +; CHECK-NEXT: [[TMP2:%.*]] = xor i1 [[TMP1]], true +; CHECK-NEXT: br i1 [[TMP2]], label [[PRELOOP_EXIT_SELECTOR:%.*]], label [[LOOP_PRELOOP]], !llvm.loop [[LOOP7:![0-9]+]], !irce.loop.clone [[META6]] +; CHECK: preloop.exit.selector: +; CHECK-NEXT: [[IDX_NEXT_PRELOOP_LCSSA:%.*]] = phi i32 [ [[IDX_NEXT_PRELOOP]], [[IN_BOUNDS_PRELOOP]] ] +; CHECK-NEXT: [[TMP3:%.*]] = icmp ugt i32 [[IDX_NEXT_PRELOOP_LCSSA]], 0 +; CHECK-NEXT: br i1 [[TMP3]], label [[PRELOOP_PSEUDO_EXIT]], label [[EXIT]] +; CHECK: preloop.pseudo.exit: +; CHECK-NEXT: [[IDX_PRELOOP_COPY]] = phi i32 [ 100, [[ENTRY:%.*]] ], [ [[IDX_NEXT_PRELOOP_LCSSA]], [[PRELOOP_EXIT_SELECTOR]] ] +; CHECK-NEXT: [[INDVAR_END:%.*]] = phi i32 [ 100, [[ENTRY]] ], [ [[IDX_NEXT_PRELOOP_LCSSA]], [[PRELOOP_EXIT_SELECTOR]] ] +; CHECK-NEXT: br label [[MAINLOOP]] +; entry: %len = load i32, ptr %a_len_ptr, !range !0 @@ -103,23 +169,56 @@ ; Check SINT_MAX. define void @test_03(ptr %arr, ptr %a_len_ptr) #0 { - -; CHECK: test_03 -; CHECK: entry: -; CHECK-NEXT: %exit.mainloop.at = load i32, ptr %a_len_ptr, align 4, !range !0 -; CHECK-NEXT: [[COND:%[^ ]+]] = icmp ult i32 0, %exit.mainloop.at -; CHECK-NEXT: br i1 [[COND]], label %loop.preheader, label %main.pseudo.exit -; CHECK: loop: -; CHECK-NEXT: %idx = phi i32 [ %idx.next, %in.bounds ], [ 0, %loop.preheader ] -; CHECK-NEXT: %idx.next = add nuw nsw i32 %idx, 1 -; CHECK-NEXT: %abc = icmp ult i32 %idx, %exit.mainloop.at -; CHECK-NEXT: br i1 true, label %in.bounds, label %out.of.bounds.loopexit1 -; CHECK-NOT: loop.preloop: -; CHECK: loop.postloop: -; CHECK-NEXT: %idx.postloop = phi i32 [ %idx.copy, %postloop ], [ %idx.next.postloop, %in.bounds.postloop ] -; CHECK-NEXT: %idx.next.postloop = add nuw nsw i32 %idx.postloop, 1 -; CHECK-NEXT: %abc.postloop = icmp ult i32 %idx.postloop, %exit.mainloop.at -; CHECK-NEXT: br i1 %abc.postloop, label %in.bounds.postloop, label %out.of.bounds.loopexit +; CHECK-LABEL: define void @test_03 +; CHECK-SAME: (ptr [[ARR:%.*]], ptr [[A_LEN_PTR:%.*]]) { +; CHECK-NEXT: entry: +; CHECK-NEXT: [[EXIT_MAINLOOP_AT:%.*]] = load i32, ptr [[A_LEN_PTR]], align 4, !range [[RNG0]] +; CHECK-NEXT: [[TMP0:%.*]] = icmp ult i32 0, [[EXIT_MAINLOOP_AT]] +; CHECK-NEXT: br i1 [[TMP0]], label [[LOOP_PREHEADER:%.*]], label [[MAIN_PSEUDO_EXIT:%.*]] +; CHECK: loop.preheader: +; CHECK-NEXT: br label [[LOOP:%.*]] +; CHECK: loop: +; CHECK-NEXT: [[IDX:%.*]] = phi i32 [ [[IDX_NEXT:%.*]], [[IN_BOUNDS:%.*]] ], [ 0, [[LOOP_PREHEADER]] ] +; CHECK-NEXT: [[IDX_NEXT]] = add nuw nsw i32 [[IDX]], 1 +; CHECK-NEXT: [[ABC:%.*]] = icmp ult i32 [[IDX]], [[EXIT_MAINLOOP_AT]] +; CHECK-NEXT: br i1 true, label [[IN_BOUNDS]], label [[OUT_OF_BOUNDS_LOOPEXIT1:%.*]] +; CHECK: in.bounds: +; CHECK-NEXT: [[ADDR:%.*]] = getelementptr i32, ptr [[ARR]], i32 [[IDX]] +; CHECK-NEXT: store i32 0, ptr [[ADDR]], align 4 +; CHECK-NEXT: [[NEXT:%.*]] = icmp ult i32 [[IDX_NEXT]], 2147483647 +; CHECK-NEXT: [[TMP1:%.*]] = icmp ult i32 [[IDX_NEXT]], [[EXIT_MAINLOOP_AT]] +; CHECK-NEXT: br i1 [[TMP1]], label [[LOOP]], label [[MAIN_EXIT_SELECTOR:%.*]] +; CHECK: main.exit.selector: +; CHECK-NEXT: [[IDX_NEXT_LCSSA:%.*]] = phi i32 [ [[IDX_NEXT]], [[IN_BOUNDS]] ] +; CHECK-NEXT: [[TMP2:%.*]] = icmp ult i32 [[IDX_NEXT_LCSSA]], 2147483647 +; CHECK-NEXT: br i1 [[TMP2]], label [[MAIN_PSEUDO_EXIT]], label [[EXIT:%.*]] +; CHECK: main.pseudo.exit: +; CHECK-NEXT: [[IDX_COPY:%.*]] = phi i32 [ 0, [[ENTRY:%.*]] ], [ [[IDX_NEXT_LCSSA]], [[MAIN_EXIT_SELECTOR]] ] +; CHECK-NEXT: [[INDVAR_END:%.*]] = phi i32 [ 0, [[ENTRY]] ], [ [[IDX_NEXT_LCSSA]], [[MAIN_EXIT_SELECTOR]] ] +; CHECK-NEXT: br label [[POSTLOOP:%.*]] +; CHECK: out.of.bounds.loopexit: +; CHECK-NEXT: br label [[OUT_OF_BOUNDS:%.*]] +; CHECK: out.of.bounds.loopexit1: +; CHECK-NEXT: br label [[OUT_OF_BOUNDS]] +; CHECK: out.of.bounds: +; CHECK-NEXT: ret void +; CHECK: exit.loopexit: +; CHECK-NEXT: br label [[EXIT]] +; CHECK: exit: +; CHECK-NEXT: ret void +; CHECK: postloop: +; CHECK-NEXT: br label [[LOOP_POSTLOOP:%.*]] +; CHECK: loop.postloop: +; CHECK-NEXT: [[IDX_POSTLOOP:%.*]] = phi i32 [ [[IDX_COPY]], [[POSTLOOP]] ], [ [[IDX_NEXT_POSTLOOP:%.*]], [[IN_BOUNDS_POSTLOOP:%.*]] ] +; CHECK-NEXT: [[IDX_NEXT_POSTLOOP]] = add nuw nsw i32 [[IDX_POSTLOOP]], 1 +; CHECK-NEXT: [[ABC_POSTLOOP:%.*]] = icmp ult i32 [[IDX_POSTLOOP]], [[EXIT_MAINLOOP_AT]] +; CHECK-NEXT: br i1 [[ABC_POSTLOOP]], label [[IN_BOUNDS_POSTLOOP]], label [[OUT_OF_BOUNDS_LOOPEXIT:%.*]] +; CHECK: in.bounds.postloop: +; CHECK-NEXT: [[ADDR_POSTLOOP:%.*]] = getelementptr i32, ptr [[ARR]], i32 [[IDX_POSTLOOP]] +; CHECK-NEXT: store i32 0, ptr [[ADDR_POSTLOOP]], align 4 +; CHECK-NEXT: [[NEXT_POSTLOOP:%.*]] = icmp ult i32 [[IDX_NEXT_POSTLOOP]], 2147483647 +; CHECK-NEXT: br i1 [[NEXT_POSTLOOP]], label [[LOOP_POSTLOOP]], label [[EXIT_LOOPEXIT:%.*]], !llvm.loop [[LOOP8:![0-9]+]], !irce.loop.clone [[META6]] +; entry: %len = load i32, ptr %a_len_ptr, !range !0 @@ -146,23 +245,56 @@ ; Check SINT_MAX + 1, test is similar to test_01. define void @test_04(ptr %arr, ptr %a_len_ptr) #0 { - -; CHECK: test_04 -; CHECK: entry: -; CHECK-NEXT: %exit.mainloop.at = load i32, ptr %a_len_ptr, align 4, !range !0 -; CHECK-NEXT: [[COND:%[^ ]+]] = icmp ult i32 0, %exit.mainloop.at -; CHECK-NEXT: br i1 [[COND]], label %loop.preheader, label %main.pseudo.exit -; CHECK: loop: -; CHECK-NEXT: %idx = phi i32 [ %idx.next, %in.bounds ], [ 0, %loop.preheader ] -; CHECK-NEXT: %idx.next = add nuw nsw i32 %idx, 1 -; CHECK-NEXT: %abc = icmp ult i32 %idx, %exit.mainloop.at -; CHECK-NEXT: br i1 true, label %in.bounds, label %out.of.bounds.loopexit1 -; CHECK-NOT: loop.preloop: -; CHECK: loop.postloop: -; CHECK-NEXT: %idx.postloop = phi i32 [ %idx.copy, %postloop ], [ %idx.next.postloop, %in.bounds.postloop ] -; CHECK-NEXT: %idx.next.postloop = add nuw nsw i32 %idx.postloop, 1 -; CHECK-NEXT: %abc.postloop = icmp ult i32 %idx.postloop, %exit.mainloop.at -; CHECK-NEXT: br i1 %abc.postloop, label %in.bounds.postloop, label %out.of.bounds.loopexit +; CHECK-LABEL: define void @test_04 +; CHECK-SAME: (ptr [[ARR:%.*]], ptr [[A_LEN_PTR:%.*]]) { +; CHECK-NEXT: entry: +; CHECK-NEXT: [[EXIT_MAINLOOP_AT:%.*]] = load i32, ptr [[A_LEN_PTR]], align 4, !range [[RNG0]] +; CHECK-NEXT: [[TMP0:%.*]] = icmp ult i32 0, [[EXIT_MAINLOOP_AT]] +; CHECK-NEXT: br i1 [[TMP0]], label [[LOOP_PREHEADER:%.*]], label [[MAIN_PSEUDO_EXIT:%.*]] +; CHECK: loop.preheader: +; CHECK-NEXT: br label [[LOOP:%.*]] +; CHECK: loop: +; CHECK-NEXT: [[IDX:%.*]] = phi i32 [ [[IDX_NEXT:%.*]], [[IN_BOUNDS:%.*]] ], [ 0, [[LOOP_PREHEADER]] ] +; CHECK-NEXT: [[IDX_NEXT]] = add nuw nsw i32 [[IDX]], 1 +; CHECK-NEXT: [[ABC:%.*]] = icmp ult i32 [[IDX]], [[EXIT_MAINLOOP_AT]] +; CHECK-NEXT: br i1 true, label [[IN_BOUNDS]], label [[OUT_OF_BOUNDS_LOOPEXIT1:%.*]] +; CHECK: in.bounds: +; CHECK-NEXT: [[ADDR:%.*]] = getelementptr i32, ptr [[ARR]], i32 [[IDX]] +; CHECK-NEXT: store i32 0, ptr [[ADDR]], align 4 +; CHECK-NEXT: [[NEXT:%.*]] = icmp ult i32 [[IDX_NEXT]], -2147483648 +; CHECK-NEXT: [[TMP1:%.*]] = icmp ult i32 [[IDX_NEXT]], [[EXIT_MAINLOOP_AT]] +; CHECK-NEXT: br i1 [[TMP1]], label [[LOOP]], label [[MAIN_EXIT_SELECTOR:%.*]] +; CHECK: main.exit.selector: +; CHECK-NEXT: [[IDX_NEXT_LCSSA:%.*]] = phi i32 [ [[IDX_NEXT]], [[IN_BOUNDS]] ] +; CHECK-NEXT: [[TMP2:%.*]] = icmp ult i32 [[IDX_NEXT_LCSSA]], -2147483648 +; CHECK-NEXT: br i1 [[TMP2]], label [[MAIN_PSEUDO_EXIT]], label [[EXIT:%.*]] +; CHECK: main.pseudo.exit: +; CHECK-NEXT: [[IDX_COPY:%.*]] = phi i32 [ 0, [[ENTRY:%.*]] ], [ [[IDX_NEXT_LCSSA]], [[MAIN_EXIT_SELECTOR]] ] +; CHECK-NEXT: [[INDVAR_END:%.*]] = phi i32 [ 0, [[ENTRY]] ], [ [[IDX_NEXT_LCSSA]], [[MAIN_EXIT_SELECTOR]] ] +; CHECK-NEXT: br label [[POSTLOOP:%.*]] +; CHECK: out.of.bounds.loopexit: +; CHECK-NEXT: br label [[OUT_OF_BOUNDS:%.*]] +; CHECK: out.of.bounds.loopexit1: +; CHECK-NEXT: br label [[OUT_OF_BOUNDS]] +; CHECK: out.of.bounds: +; CHECK-NEXT: ret void +; CHECK: exit.loopexit: +; CHECK-NEXT: br label [[EXIT]] +; CHECK: exit: +; CHECK-NEXT: ret void +; CHECK: postloop: +; CHECK-NEXT: br label [[LOOP_POSTLOOP:%.*]] +; CHECK: loop.postloop: +; CHECK-NEXT: [[IDX_POSTLOOP:%.*]] = phi i32 [ [[IDX_COPY]], [[POSTLOOP]] ], [ [[IDX_NEXT_POSTLOOP:%.*]], [[IN_BOUNDS_POSTLOOP:%.*]] ] +; CHECK-NEXT: [[IDX_NEXT_POSTLOOP]] = add nuw nsw i32 [[IDX_POSTLOOP]], 1 +; CHECK-NEXT: [[ABC_POSTLOOP:%.*]] = icmp ult i32 [[IDX_POSTLOOP]], [[EXIT_MAINLOOP_AT]] +; CHECK-NEXT: br i1 [[ABC_POSTLOOP]], label [[IN_BOUNDS_POSTLOOP]], label [[OUT_OF_BOUNDS_LOOPEXIT:%.*]] +; CHECK: in.bounds.postloop: +; CHECK-NEXT: [[ADDR_POSTLOOP:%.*]] = getelementptr i32, ptr [[ARR]], i32 [[IDX_POSTLOOP]] +; CHECK-NEXT: store i32 0, ptr [[ADDR_POSTLOOP]], align 4 +; CHECK-NEXT: [[NEXT_POSTLOOP:%.*]] = icmp ult i32 [[IDX_NEXT_POSTLOOP]], -2147483648 +; CHECK-NEXT: br i1 [[NEXT_POSTLOOP]], label [[LOOP_POSTLOOP]], label [[EXIT_LOOPEXIT:%.*]], !llvm.loop [[LOOP9:![0-9]+]], !irce.loop.clone [[META6]] +; entry: %len = load i32, ptr %a_len_ptr, !range !0 @@ -189,26 +321,59 @@ ; Check SINT_MAX + 1, test is similar to test_02. define void @test_05(ptr %arr, ptr %a_len_ptr) #0 { -; CHECK: test_05( -; CHECK: entry: -; CHECK-NEXT: %len = load i32, ptr %a_len_ptr, align 4, !range !0 -; CHECK-NEXT: [[UMIN:%[^ ]+]] = call i32 @llvm.umax.i32(i32 %len, i32 1) -; CHECK-NEXT: %exit.preloop.at = add nsw i32 [[UMIN]], -1 -; CHECK-NEXT: [[COND2:%[^ ]+]] = icmp ugt i32 -2147483648, %exit.preloop.at -; CHECK-NEXT: br i1 [[COND2]], label %loop.preloop.preheader, label %preloop.pseudo.exit -; CHECK: mainloop: -; CHECK-NEXT: br label %loop -; CHECK: loop: -; CHECK-NEXT: %idx = phi i32 [ %idx.preloop.copy, %mainloop ], [ %idx.next, %in.bounds ] -; CHECK-NEXT: %idx.next = add i32 %idx, -1 -; CHECK-NEXT: %abc = icmp ult i32 %idx, %len -; CHECK-NEXT: br i1 true, label %in.bounds, label %out.of.bounds.loopexit1 -; CHECK-NOT: loop.postloop: -; CHECK: loop.preloop: -; CHECK-NEXT: %idx.preloop = phi i32 [ %idx.next.preloop, %in.bounds.preloop ], [ -2147483648, %loop.preloop.preheader ] -; CHECK-NEXT: %idx.next.preloop = add i32 %idx.preloop, -1 -; CHECK-NEXT: %abc.preloop = icmp ult i32 %idx.preloop, %len -; CHECK-NEXT: br i1 %abc.preloop, label %in.bounds.preloop, label %out.of.bounds.loopexit +; CHECK-LABEL: define void @test_05 +; CHECK-SAME: (ptr [[ARR:%.*]], ptr [[A_LEN_PTR:%.*]]) { +; CHECK-NEXT: entry: +; CHECK-NEXT: [[LEN:%.*]] = load i32, ptr [[A_LEN_PTR]], align 4, !range [[RNG0]] +; CHECK-NEXT: [[UMAX:%.*]] = call i32 @llvm.umax.i32(i32 [[LEN]], i32 1) +; CHECK-NEXT: [[EXIT_PRELOOP_AT:%.*]] = add nsw i32 [[UMAX]], -1 +; CHECK-NEXT: [[TMP0:%.*]] = icmp ugt i32 -2147483648, [[EXIT_PRELOOP_AT]] +; CHECK-NEXT: br i1 [[TMP0]], label [[LOOP_PRELOOP_PREHEADER:%.*]], label [[PRELOOP_PSEUDO_EXIT:%.*]] +; CHECK: loop.preloop.preheader: +; CHECK-NEXT: br label [[LOOP_PRELOOP:%.*]] +; CHECK: mainloop: +; CHECK-NEXT: br label [[LOOP:%.*]] +; CHECK: loop: +; CHECK-NEXT: [[IDX:%.*]] = phi i32 [ [[IDX_PRELOOP_COPY:%.*]], [[MAINLOOP:%.*]] ], [ [[IDX_NEXT:%.*]], [[IN_BOUNDS:%.*]] ] +; CHECK-NEXT: [[IDX_NEXT]] = add i32 [[IDX]], -1 +; CHECK-NEXT: [[ABC:%.*]] = icmp ult i32 [[IDX]], [[LEN]] +; CHECK-NEXT: br i1 true, label [[IN_BOUNDS]], label [[OUT_OF_BOUNDS_LOOPEXIT1:%.*]] +; CHECK: in.bounds: +; CHECK-NEXT: [[ADDR:%.*]] = getelementptr i32, ptr [[ARR]], i32 [[IDX]] +; CHECK-NEXT: store i32 0, ptr [[ADDR]], align 4 +; CHECK-NEXT: [[NEXT:%.*]] = icmp ult i32 [[IDX_NEXT]], 1 +; CHECK-NEXT: br i1 [[NEXT]], label [[EXIT_LOOPEXIT:%.*]], label [[LOOP]] +; CHECK: out.of.bounds.loopexit: +; CHECK-NEXT: br label [[OUT_OF_BOUNDS:%.*]] +; CHECK: out.of.bounds.loopexit1: +; CHECK-NEXT: br label [[OUT_OF_BOUNDS]] +; CHECK: out.of.bounds: +; CHECK-NEXT: ret void +; CHECK: exit.loopexit: +; CHECK-NEXT: br label [[EXIT:%.*]] +; CHECK: exit: +; CHECK-NEXT: ret void +; CHECK: loop.preloop: +; CHECK-NEXT: [[IDX_PRELOOP:%.*]] = phi i32 [ [[IDX_NEXT_PRELOOP:%.*]], [[IN_BOUNDS_PRELOOP:%.*]] ], [ -2147483648, [[LOOP_PRELOOP_PREHEADER]] ] +; CHECK-NEXT: [[IDX_NEXT_PRELOOP]] = add i32 [[IDX_PRELOOP]], -1 +; CHECK-NEXT: [[ABC_PRELOOP:%.*]] = icmp ult i32 [[IDX_PRELOOP]], [[LEN]] +; CHECK-NEXT: br i1 [[ABC_PRELOOP]], label [[IN_BOUNDS_PRELOOP]], label [[OUT_OF_BOUNDS_LOOPEXIT:%.*]] +; CHECK: in.bounds.preloop: +; CHECK-NEXT: [[ADDR_PRELOOP:%.*]] = getelementptr i32, ptr [[ARR]], i32 [[IDX_PRELOOP]] +; CHECK-NEXT: store i32 0, ptr [[ADDR_PRELOOP]], align 4 +; CHECK-NEXT: [[NEXT_PRELOOP:%.*]] = icmp ult i32 [[IDX_NEXT_PRELOOP]], 1 +; CHECK-NEXT: [[TMP1:%.*]] = icmp ugt i32 [[IDX_NEXT_PRELOOP]], [[EXIT_PRELOOP_AT]] +; CHECK-NEXT: [[TMP2:%.*]] = xor i1 [[TMP1]], true +; CHECK-NEXT: br i1 [[TMP2]], label [[PRELOOP_EXIT_SELECTOR:%.*]], label [[LOOP_PRELOOP]], !llvm.loop [[LOOP10:![0-9]+]], !irce.loop.clone [[META6]] +; CHECK: preloop.exit.selector: +; CHECK-NEXT: [[IDX_NEXT_PRELOOP_LCSSA:%.*]] = phi i32 [ [[IDX_NEXT_PRELOOP]], [[IN_BOUNDS_PRELOOP]] ] +; CHECK-NEXT: [[TMP3:%.*]] = icmp ugt i32 [[IDX_NEXT_PRELOOP_LCSSA]], 0 +; CHECK-NEXT: br i1 [[TMP3]], label [[PRELOOP_PSEUDO_EXIT]], label [[EXIT]] +; CHECK: preloop.pseudo.exit: +; CHECK-NEXT: [[IDX_PRELOOP_COPY]] = phi i32 [ -2147483648, [[ENTRY:%.*]] ], [ [[IDX_NEXT_PRELOOP_LCSSA]], [[PRELOOP_EXIT_SELECTOR]] ] +; CHECK-NEXT: [[INDVAR_END:%.*]] = phi i32 [ -2147483648, [[ENTRY]] ], [ [[IDX_NEXT_PRELOOP_LCSSA]], [[PRELOOP_EXIT_SELECTOR]] ] +; CHECK-NEXT: br label [[MAINLOOP]] +; entry: %len = load i32, ptr %a_len_ptr, !range !0 @@ -235,23 +400,56 @@ ; Increasing loop, UINT_MAX. Positive test. define void @test_06(ptr %arr, ptr %a_len_ptr) #0 { - -; CHECK: test_06 -; CHECK: entry: -; CHECK-NEXT: %exit.mainloop.at = load i32, ptr %a_len_ptr, align 4, !range !0 -; CHECK-NEXT: [[COND:%[^ ]+]] = icmp ult i32 0, %exit.mainloop.at -; CHECK-NEXT: br i1 [[COND]], label %loop.preheader, label %main.pseudo.exit -; CHECK: loop: -; CHECK-NEXT: %idx = phi i32 [ %idx.next, %in.bounds ], [ 0, %loop.preheader ] -; CHECK-NEXT: %idx.next = add nuw nsw i32 %idx, 1 -; CHECK-NEXT: %abc = icmp ult i32 %idx, %exit.mainloop.at -; CHECK-NEXT: br i1 true, label %in.bounds, label %out.of.bounds.loopexit1 -; CHECK-NOT: loop.preloop: -; CHECK: loop.postloop: -; CHECK-NEXT: %idx.postloop = phi i32 [ %idx.copy, %postloop ], [ %idx.next.postloop, %in.bounds.postloop ] -; CHECK-NEXT: %idx.next.postloop = add nuw nsw i32 %idx.postloop, 1 -; CHECK-NEXT: %abc.postloop = icmp ult i32 %idx.postloop, %exit.mainloop.at -; CHECK-NEXT: br i1 %abc.postloop, label %in.bounds.postloop, label %out.of.bounds.loopexit +; CHECK-LABEL: define void @test_06 +; CHECK-SAME: (ptr [[ARR:%.*]], ptr [[A_LEN_PTR:%.*]]) { +; CHECK-NEXT: entry: +; CHECK-NEXT: [[EXIT_MAINLOOP_AT:%.*]] = load i32, ptr [[A_LEN_PTR]], align 4, !range [[RNG0]] +; CHECK-NEXT: [[TMP0:%.*]] = icmp ult i32 0, [[EXIT_MAINLOOP_AT]] +; CHECK-NEXT: br i1 [[TMP0]], label [[LOOP_PREHEADER:%.*]], label [[MAIN_PSEUDO_EXIT:%.*]] +; CHECK: loop.preheader: +; CHECK-NEXT: br label [[LOOP:%.*]] +; CHECK: loop: +; CHECK-NEXT: [[IDX:%.*]] = phi i32 [ [[IDX_NEXT:%.*]], [[IN_BOUNDS:%.*]] ], [ 0, [[LOOP_PREHEADER]] ] +; CHECK-NEXT: [[IDX_NEXT]] = add nuw nsw i32 [[IDX]], 1 +; CHECK-NEXT: [[ABC:%.*]] = icmp ult i32 [[IDX]], [[EXIT_MAINLOOP_AT]] +; CHECK-NEXT: br i1 true, label [[IN_BOUNDS]], label [[OUT_OF_BOUNDS_LOOPEXIT1:%.*]] +; CHECK: in.bounds: +; CHECK-NEXT: [[ADDR:%.*]] = getelementptr i32, ptr [[ARR]], i32 [[IDX]] +; CHECK-NEXT: store i32 0, ptr [[ADDR]], align 4 +; CHECK-NEXT: [[NEXT:%.*]] = icmp ult i32 [[IDX_NEXT]], -1 +; CHECK-NEXT: [[TMP1:%.*]] = icmp ult i32 [[IDX_NEXT]], [[EXIT_MAINLOOP_AT]] +; CHECK-NEXT: br i1 [[TMP1]], label [[LOOP]], label [[MAIN_EXIT_SELECTOR:%.*]] +; CHECK: main.exit.selector: +; CHECK-NEXT: [[IDX_NEXT_LCSSA:%.*]] = phi i32 [ [[IDX_NEXT]], [[IN_BOUNDS]] ] +; CHECK-NEXT: [[TMP2:%.*]] = icmp ult i32 [[IDX_NEXT_LCSSA]], -1 +; CHECK-NEXT: br i1 [[TMP2]], label [[MAIN_PSEUDO_EXIT]], label [[EXIT:%.*]] +; CHECK: main.pseudo.exit: +; CHECK-NEXT: [[IDX_COPY:%.*]] = phi i32 [ 0, [[ENTRY:%.*]] ], [ [[IDX_NEXT_LCSSA]], [[MAIN_EXIT_SELECTOR]] ] +; CHECK-NEXT: [[INDVAR_END:%.*]] = phi i32 [ 0, [[ENTRY]] ], [ [[IDX_NEXT_LCSSA]], [[MAIN_EXIT_SELECTOR]] ] +; CHECK-NEXT: br label [[POSTLOOP:%.*]] +; CHECK: out.of.bounds.loopexit: +; CHECK-NEXT: br label [[OUT_OF_BOUNDS:%.*]] +; CHECK: out.of.bounds.loopexit1: +; CHECK-NEXT: br label [[OUT_OF_BOUNDS]] +; CHECK: out.of.bounds: +; CHECK-NEXT: ret void +; CHECK: exit.loopexit: +; CHECK-NEXT: br label [[EXIT]] +; CHECK: exit: +; CHECK-NEXT: ret void +; CHECK: postloop: +; CHECK-NEXT: br label [[LOOP_POSTLOOP:%.*]] +; CHECK: loop.postloop: +; CHECK-NEXT: [[IDX_POSTLOOP:%.*]] = phi i32 [ [[IDX_COPY]], [[POSTLOOP]] ], [ [[IDX_NEXT_POSTLOOP:%.*]], [[IN_BOUNDS_POSTLOOP:%.*]] ] +; CHECK-NEXT: [[IDX_NEXT_POSTLOOP]] = add nuw nsw i32 [[IDX_POSTLOOP]], 1 +; CHECK-NEXT: [[ABC_POSTLOOP:%.*]] = icmp ult i32 [[IDX_POSTLOOP]], [[EXIT_MAINLOOP_AT]] +; CHECK-NEXT: br i1 [[ABC_POSTLOOP]], label [[IN_BOUNDS_POSTLOOP]], label [[OUT_OF_BOUNDS_LOOPEXIT:%.*]] +; CHECK: in.bounds.postloop: +; CHECK-NEXT: [[ADDR_POSTLOOP:%.*]] = getelementptr i32, ptr [[ARR]], i32 [[IDX_POSTLOOP]] +; CHECK-NEXT: store i32 0, ptr [[ADDR_POSTLOOP]], align 4 +; CHECK-NEXT: [[NEXT_POSTLOOP:%.*]] = icmp ult i32 [[IDX_NEXT_POSTLOOP]], -1 +; CHECK-NEXT: br i1 [[NEXT_POSTLOOP]], label [[LOOP_POSTLOOP]], label [[EXIT_LOOPEXIT:%.*]], !llvm.loop [[LOOP11:![0-9]+]], !irce.loop.clone [[META6]] +; entry: %len = load i32, ptr %a_len_ptr, !range !0 @@ -278,10 +476,26 @@ ; Decreasing loop, UINT_MAX. Negative test: we cannot substract -1 from 0. define void @test_07(ptr %arr, ptr %a_len_ptr) #0 { - -; CHECK: test_07( -; CHECK-NOT: loop.preloop: -; CHECK-NOT: loop.postloop: +; CHECK-LABEL: define void @test_07 +; CHECK-SAME: (ptr [[ARR:%.*]], ptr [[A_LEN_PTR:%.*]]) { +; CHECK-NEXT: entry: +; CHECK-NEXT: [[LEN:%.*]] = load i32, ptr [[A_LEN_PTR]], align 4, !range [[RNG0]] +; CHECK-NEXT: br label [[LOOP:%.*]] +; CHECK: loop: +; CHECK-NEXT: [[IDX:%.*]] = phi i32 [ -1, [[ENTRY:%.*]] ], [ [[IDX_NEXT:%.*]], [[IN_BOUNDS:%.*]] ] +; CHECK-NEXT: [[IDX_NEXT]] = add nuw i32 [[IDX]], -1 +; CHECK-NEXT: [[ABC:%.*]] = icmp ult i32 [[IDX]], [[LEN]] +; CHECK-NEXT: br i1 [[ABC]], label [[IN_BOUNDS]], label [[OUT_OF_BOUNDS:%.*]] +; CHECK: in.bounds: +; CHECK-NEXT: [[ADDR:%.*]] = getelementptr i32, ptr [[ARR]], i32 [[IDX]] +; CHECK-NEXT: store i32 0, ptr [[ADDR]], align 4 +; CHECK-NEXT: [[NEXT:%.*]] = icmp ult i32 [[IDX_NEXT]], 0 +; CHECK-NEXT: br i1 [[NEXT]], label [[EXIT:%.*]], label [[LOOP]] +; CHECK: out.of.bounds: +; CHECK-NEXT: ret void +; CHECK: exit: +; CHECK-NEXT: ret void +; entry: %len = load i32, ptr %a_len_ptr, !range !0 @@ -311,23 +525,56 @@ ; range does not prevent us from performing IRCE. define void @test_08(ptr %arr, ptr %a_len_ptr) #0 { - -; CHECK: test_08 -; CHECK: entry: -; CHECK-NEXT: %exit.mainloop.at = load i32, ptr %a_len_ptr, align 4, !range !0 -; CHECK-NEXT: [[COND:%[^ ]+]] = icmp ult i32 0, %exit.mainloop.at -; CHECK-NEXT: br i1 [[COND]], label %loop.preheader, label %main.pseudo.exit -; CHECK: loop: -; CHECK-NEXT: %idx = phi i32 [ %idx.next, %in.bounds ], [ 0, %loop.preheader ] -; CHECK-NEXT: %idx.next = add i32 %idx, 1 -; CHECK-NEXT: %abc = icmp ult i32 %idx, %exit.mainloop.at -; CHECK-NEXT: br i1 true, label %in.bounds, label %out.of.bounds.loopexit1 -; CHECK-NOT: loop.preloop: -; CHECK: loop.postloop: -; CHECK-NEXT: %idx.postloop = phi i32 [ %idx.copy, %postloop ], [ %idx.next.postloop, %in.bounds.postloop ] -; CHECK-NEXT: %idx.next.postloop = add i32 %idx.postloop, 1 -; CHECK-NEXT: %abc.postloop = icmp ult i32 %idx.postloop, %exit.mainloop.at -; CHECK-NEXT: br i1 %abc.postloop, label %in.bounds.postloop, label %out.of.bounds.loopexit +; CHECK-LABEL: define void @test_08 +; CHECK-SAME: (ptr [[ARR:%.*]], ptr [[A_LEN_PTR:%.*]]) { +; CHECK-NEXT: entry: +; CHECK-NEXT: [[EXIT_MAINLOOP_AT:%.*]] = load i32, ptr [[A_LEN_PTR]], align 4, !range [[RNG0]] +; CHECK-NEXT: [[TMP0:%.*]] = icmp ult i32 0, [[EXIT_MAINLOOP_AT]] +; CHECK-NEXT: br i1 [[TMP0]], label [[LOOP_PREHEADER:%.*]], label [[MAIN_PSEUDO_EXIT:%.*]] +; CHECK: loop.preheader: +; CHECK-NEXT: br label [[LOOP:%.*]] +; CHECK: loop: +; CHECK-NEXT: [[IDX:%.*]] = phi i32 [ [[IDX_NEXT:%.*]], [[IN_BOUNDS:%.*]] ], [ 0, [[LOOP_PREHEADER]] ] +; CHECK-NEXT: [[IDX_NEXT]] = add i32 [[IDX]], 1 +; CHECK-NEXT: [[ABC:%.*]] = icmp ult i32 [[IDX]], [[EXIT_MAINLOOP_AT]] +; CHECK-NEXT: br i1 true, label [[IN_BOUNDS]], label [[OUT_OF_BOUNDS_LOOPEXIT1:%.*]] +; CHECK: in.bounds: +; CHECK-NEXT: [[ADDR:%.*]] = getelementptr i32, ptr [[ARR]], i32 [[IDX]] +; CHECK-NEXT: store i32 0, ptr [[ADDR]], align 4 +; CHECK-NEXT: [[NEXT:%.*]] = icmp ult i32 [[IDX_NEXT]], -100 +; CHECK-NEXT: [[TMP1:%.*]] = icmp ult i32 [[IDX_NEXT]], [[EXIT_MAINLOOP_AT]] +; CHECK-NEXT: br i1 [[TMP1]], label [[LOOP]], label [[MAIN_EXIT_SELECTOR:%.*]] +; CHECK: main.exit.selector: +; CHECK-NEXT: [[IDX_NEXT_LCSSA:%.*]] = phi i32 [ [[IDX_NEXT]], [[IN_BOUNDS]] ] +; CHECK-NEXT: [[TMP2:%.*]] = icmp ult i32 [[IDX_NEXT_LCSSA]], -100 +; CHECK-NEXT: br i1 [[TMP2]], label [[MAIN_PSEUDO_EXIT]], label [[EXIT:%.*]] +; CHECK: main.pseudo.exit: +; CHECK-NEXT: [[IDX_COPY:%.*]] = phi i32 [ 0, [[ENTRY:%.*]] ], [ [[IDX_NEXT_LCSSA]], [[MAIN_EXIT_SELECTOR]] ] +; CHECK-NEXT: [[INDVAR_END:%.*]] = phi i32 [ 0, [[ENTRY]] ], [ [[IDX_NEXT_LCSSA]], [[MAIN_EXIT_SELECTOR]] ] +; CHECK-NEXT: br label [[POSTLOOP:%.*]] +; CHECK: out.of.bounds.loopexit: +; CHECK-NEXT: br label [[OUT_OF_BOUNDS:%.*]] +; CHECK: out.of.bounds.loopexit1: +; CHECK-NEXT: br label [[OUT_OF_BOUNDS]] +; CHECK: out.of.bounds: +; CHECK-NEXT: ret void +; CHECK: exit.loopexit: +; CHECK-NEXT: br label [[EXIT]] +; CHECK: exit: +; CHECK-NEXT: ret void +; CHECK: postloop: +; CHECK-NEXT: br label [[LOOP_POSTLOOP:%.*]] +; CHECK: loop.postloop: +; CHECK-NEXT: [[IDX_POSTLOOP:%.*]] = phi i32 [ [[IDX_COPY]], [[POSTLOOP]] ], [ [[IDX_NEXT_POSTLOOP:%.*]], [[IN_BOUNDS_POSTLOOP:%.*]] ] +; CHECK-NEXT: [[IDX_NEXT_POSTLOOP]] = add i32 [[IDX_POSTLOOP]], 1 +; CHECK-NEXT: [[ABC_POSTLOOP:%.*]] = icmp ult i32 [[IDX_POSTLOOP]], [[EXIT_MAINLOOP_AT]] +; CHECK-NEXT: br i1 [[ABC_POSTLOOP]], label [[IN_BOUNDS_POSTLOOP]], label [[OUT_OF_BOUNDS_LOOPEXIT:%.*]] +; CHECK: in.bounds.postloop: +; CHECK-NEXT: [[ADDR_POSTLOOP:%.*]] = getelementptr i32, ptr [[ARR]], i32 [[IDX_POSTLOOP]] +; CHECK-NEXT: store i32 0, ptr [[ADDR_POSTLOOP]], align 4 +; CHECK-NEXT: [[NEXT_POSTLOOP:%.*]] = icmp ult i32 [[IDX_NEXT_POSTLOOP]], -100 +; CHECK-NEXT: br i1 [[NEXT_POSTLOOP]], label [[LOOP_POSTLOOP]], label [[EXIT_LOOPEXIT:%.*]], !llvm.loop [[LOOP12:![0-9]+]], !irce.loop.clone [[META6]] +; entry: %len = load i32, ptr %a_len_ptr, !range !0 @@ -356,12 +603,26 @@ ; (iteration space [-100; 100)). Negative test. define void @test_09(ptr %arr, ptr %a_len_ptr) #0 { - -; CHECK: test_09 -; CHECK-NOT: preloop -; CHECK-NOT: postloop -; CHECK-NOT: br i1 false -; CHECK-NOT: br i1 true +; CHECK-LABEL: define void @test_09 +; CHECK-SAME: (ptr [[ARR:%.*]], ptr [[A_LEN_PTR:%.*]]) { +; CHECK-NEXT: entry: +; CHECK-NEXT: [[LEN:%.*]] = load i32, ptr [[A_LEN_PTR]], align 4, !range [[RNG0]] +; CHECK-NEXT: br label [[LOOP:%.*]] +; CHECK: loop: +; CHECK-NEXT: [[IDX:%.*]] = phi i32 [ -100, [[ENTRY:%.*]] ], [ [[IDX_NEXT:%.*]], [[IN_BOUNDS:%.*]] ] +; CHECK-NEXT: [[IDX_NEXT]] = add i32 [[IDX]], 1 +; CHECK-NEXT: [[ABC:%.*]] = icmp ult i32 [[IDX]], [[LEN]] +; CHECK-NEXT: br i1 [[ABC]], label [[IN_BOUNDS]], label [[OUT_OF_BOUNDS:%.*]] +; CHECK: in.bounds: +; CHECK-NEXT: [[ADDR:%.*]] = getelementptr i32, ptr [[ARR]], i32 [[IDX]] +; CHECK-NEXT: store i32 0, ptr [[ADDR]], align 4 +; CHECK-NEXT: [[NEXT:%.*]] = icmp ult i32 [[IDX_NEXT]], 100 +; CHECK-NEXT: br i1 [[NEXT]], label [[LOOP]], label [[EXIT:%.*]] +; CHECK: out.of.bounds: +; CHECK-NEXT: ret void +; CHECK: exit: +; CHECK-NEXT: ret void +; entry: %len = load i32, ptr %a_len_ptr, !range !0 @@ -387,3 +648,18 @@ } !0 = !{i32 0, i32 50} +;. +; CHECK: [[RNG0]] = !{i32 0, i32 50} +; CHECK: [[LOOP1]] = distinct !{[[LOOP1]], [[META2:![0-9]+]], [[META3:![0-9]+]], [[META4:![0-9]+]], [[META5:![0-9]+]]} +; CHECK: [[META2]] = !{!"llvm.loop.unroll.disable"} +; CHECK: [[META3]] = !{!"llvm.loop.vectorize.enable", i1 false} +; CHECK: [[META4]] = !{!"llvm.loop.licm_versioning.disable"} +; CHECK: [[META5]] = !{!"llvm.loop.distribute.enable", i1 false} +; CHECK: [[META6]] = !{} +; CHECK: [[LOOP7]] = distinct !{[[LOOP7]], [[META2]], [[META3]], [[META4]], [[META5]]} +; CHECK: [[LOOP8]] = distinct !{[[LOOP8]], [[META2]], [[META3]], [[META4]], [[META5]]} +; CHECK: [[LOOP9]] = distinct !{[[LOOP9]], [[META2]], [[META3]], [[META4]], [[META5]]} +; CHECK: [[LOOP10]] = distinct !{[[LOOP10]], [[META2]], [[META3]], [[META4]], [[META5]]} +; CHECK: [[LOOP11]] = distinct !{[[LOOP11]], [[META2]], [[META3]], [[META4]], [[META5]]} +; CHECK: [[LOOP12]] = distinct !{[[LOOP12]], [[META2]], [[META3]], [[META4]], [[META5]]} +;.