diff --git a/llvm/test/Transforms/LoopVectorize/runtime-checks-difference.ll b/llvm/test/Transforms/LoopVectorize/runtime-checks-difference.ll index ecdc4ed416d47..618f64315553d 100644 --- a/llvm/test/Transforms/LoopVectorize/runtime-checks-difference.ll +++ b/llvm/test/Transforms/LoopVectorize/runtime-checks-difference.ll @@ -1,18 +1,20 @@ +; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --check-globals none --filter-out-after "vector.ph" --version 5 ; RUN: opt %s -passes=loop-vectorize -hoist-runtime-checks=false -force-vector-width=4 -force-vector-interleave=1 -S | FileCheck %s target datalayout = "e-m:e-i64:64-i128:128-n32:64-S128" define void @same_step_and_size(ptr %a, ptr %b, i64 %n) { -; CHECK-LABEL: @same_step_and_size( -; CHECK-NEXT: entry: -; CHECK-NEXT: [[A2:%.*]] = ptrtoint ptr [[A:%.*]] to i64 -; CHECK-NEXT: [[B1:%.*]] = ptrtoint ptr [[B:%.*]] to i64 -; CHECK-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 [[N:%.*]], 4 -; CHECK-NEXT: br i1 [[MIN_ITERS_CHECK]], label %scalar.ph, label %vector.memcheck -; CHECK: vector.memcheck: +; CHECK-LABEL: define void @same_step_and_size( +; CHECK-SAME: ptr [[A:%.*]], ptr [[B:%.*]], i64 [[N:%.*]]) { +; CHECK-NEXT: [[ENTRY:.*:]] +; CHECK-NEXT: [[A2:%.*]] = ptrtoint ptr [[A]] to i64 +; CHECK-NEXT: [[B1:%.*]] = ptrtoint ptr [[B]] to i64 +; CHECK-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 [[N]], 4 +; CHECK-NEXT: br i1 [[MIN_ITERS_CHECK]], [[SCALAR_PH:label %.*]], label %[[VECTOR_MEMCHECK:.*]] +; CHECK: [[VECTOR_MEMCHECK]]: ; CHECK-NEXT: [[TMP0:%.*]] = sub i64 [[B1]], [[A2]] ; CHECK-NEXT: [[DIFF_CHECK:%.*]] = icmp ult i64 [[TMP0]], 16 -; CHECK-NEXT: br i1 [[DIFF_CHECK]], label %scalar.ph, label %vector.ph +; CHECK-NEXT: br i1 [[DIFF_CHECK]], [[SCALAR_PH]], [[VECTOR_PH:label %.*]] ; entry: br label %loop @@ -33,16 +35,17 @@ exit: } define void @same_step_and_size_no_dominance_between_accesses(ptr %a, ptr %b, i64 %n, i64 %x) { -; CHECK-LABEL: @same_step_and_size_no_dominance_between_accesses( -; CHECK-NEXT: entry: -; CHECK-NEXT: [[B2:%.*]] = ptrtoint ptr [[B:%.*]] to i64 -; CHECK-NEXT: [[A1:%.*]] = ptrtoint ptr [[A:%.*]] to i64 -; CHECK-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 [[N:%.*]], 4 -; CHECK-NEXT: br i1 [[MIN_ITERS_CHECK]], label %scalar.ph, label %vector.memcheck -; CHECK: vector.memcheck: +; CHECK-LABEL: define void @same_step_and_size_no_dominance_between_accesses( +; CHECK-SAME: ptr [[A:%.*]], ptr [[B:%.*]], i64 [[N:%.*]], i64 [[X:%.*]]) { +; CHECK-NEXT: [[ENTRY:.*:]] +; CHECK-NEXT: [[B2:%.*]] = ptrtoint ptr [[B]] to i64 +; CHECK-NEXT: [[A1:%.*]] = ptrtoint ptr [[A]] to i64 +; CHECK-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 [[N]], 4 +; CHECK-NEXT: br i1 [[MIN_ITERS_CHECK]], [[SCALAR_PH:label %.*]], label %[[VECTOR_MEMCHECK:.*]] +; CHECK: [[VECTOR_MEMCHECK]]: ; CHECK-NEXT: [[TMP0:%.*]] = sub i64 [[A1]], [[B2]] ; CHECK-NEXT: [[DIFF_CHECK:%.*]] = icmp ult i64 [[TMP0]], 16 -; CHECK-NEXT: br i1 [[DIFF_CHECK]], label %scalar.ph, label %vector.ph +; CHECK-NEXT: br i1 [[DIFF_CHECK]], [[SCALAR_PH]], [[VECTOR_PH:label %.*]] ; entry: br label %loop @@ -72,19 +75,20 @@ exit: } define void @different_steps_and_different_access_sizes(ptr %a, ptr %b, i64 %n) { -; CHECK-LABEL: @different_steps_and_different_access_sizes( -; CHECK-NEXT: entry: -; CHECK-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 [[N:%.*]], 4 -; CHECK-NEXT: br i1 [[MIN_ITERS_CHECK]], label %scalar.ph, label %vector.memcheck -; CHECK: vector.memcheck: -; CHECK-NEXT: [[N_SHL_2:%.]] = shl i64 %n, 2 -; CHECK-NEXT: [[SCEVGEP:%.*]] = getelementptr i8, ptr %b, i64 [[N_SHL_2]] -; CHECK-NEXT: [[N_SHL_1:%.]] = shl i64 %n, 1 -; CHECK-NEXT: [[SCEVGEP4:%.*]] = getelementptr i8, ptr %a, i64 [[N_SHL_1]] -; CHECK-NEXT: [[BOUND0:%.*]] = icmp ult ptr %b, [[SCEVGEP4]] -; CHECK-NEXT: [[BOUND1:%.*]] = icmp ult ptr %a, [[SCEVGEP]] +; CHECK-LABEL: define void @different_steps_and_different_access_sizes( +; CHECK-SAME: ptr [[A:%.*]], ptr [[B:%.*]], i64 [[N:%.*]]) { +; CHECK-NEXT: [[ENTRY:.*:]] +; CHECK-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 [[N]], 4 +; CHECK-NEXT: br i1 [[MIN_ITERS_CHECK]], [[SCALAR_PH:label %.*]], label %[[VECTOR_MEMCHECK:.*]] +; CHECK: [[VECTOR_MEMCHECK]]: +; CHECK-NEXT: [[TMP0:%.*]] = shl i64 [[N]], 2 +; CHECK-NEXT: [[SCEVGEP:%.*]] = getelementptr i8, ptr [[B]], i64 [[TMP0]] +; CHECK-NEXT: [[TMP1:%.*]] = shl i64 [[N]], 1 +; CHECK-NEXT: [[SCEVGEP1:%.*]] = getelementptr i8, ptr [[A]], i64 [[TMP1]] +; CHECK-NEXT: [[BOUND0:%.*]] = icmp ult ptr [[B]], [[SCEVGEP1]] +; CHECK-NEXT: [[BOUND1:%.*]] = icmp ult ptr [[A]], [[SCEVGEP]] ; CHECK-NEXT: [[FOUND_CONFLICT:%.*]] = and i1 [[BOUND0]], [[BOUND1]] -; CHECK-NEXT: br i1 [[FOUND_CONFLICT]], label %scalar.ph, label %vector.ph +; CHECK-NEXT: br i1 [[FOUND_CONFLICT]], [[SCALAR_PH]], [[VECTOR_PH:label %.*]] ; entry: br label %loop @@ -106,17 +110,18 @@ exit: } define void @steps_match_but_different_access_sizes_1(ptr %a, ptr %b, i64 %n) { -; CHECK-LABEL: @steps_match_but_different_access_sizes_1( -; CHECK-NEXT: entry: -; CHECK-NEXT: [[A2:%.*]] = ptrtoint ptr [[A:%.*]] to i64 -; CHECK-NEXT: [[B1:%.*]] = ptrtoint ptr [[B:%.*]] to i64 -; CHECK-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 [[N:%.*]], 4 -; CHECK-NEXT: br i1 [[MIN_ITERS_CHECK]], label %scalar.ph, label %vector.memcheck -; CHECK: vector.memcheck: +; CHECK-LABEL: define void @steps_match_but_different_access_sizes_1( +; CHECK-SAME: ptr [[A:%.*]], ptr [[B:%.*]], i64 [[N:%.*]]) { +; CHECK-NEXT: [[ENTRY:.*:]] +; CHECK-NEXT: [[A2:%.*]] = ptrtoint ptr [[A]] to i64 +; CHECK-NEXT: [[B1:%.*]] = ptrtoint ptr [[B]] to i64 +; CHECK-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 [[N]], 4 +; CHECK-NEXT: br i1 [[MIN_ITERS_CHECK]], [[SCALAR_PH:label %.*]], label %[[VECTOR_MEMCHECK:.*]] +; CHECK: [[VECTOR_MEMCHECK]]: ; CHECK-NEXT: [[TMP0:%.*]] = add i64 [[B1]], -2 ; CHECK-NEXT: [[TMP1:%.*]] = sub i64 [[TMP0]], [[A2]] ; CHECK-NEXT: [[DIFF_CHECK:%.*]] = icmp ult i64 [[TMP1]], 16 -; CHECK-NEXT: br i1 [[DIFF_CHECK]], label %scalar.ph, label %vector.ph +; CHECK-NEXT: br i1 [[DIFF_CHECK]], [[SCALAR_PH]], [[VECTOR_PH:label %.*]] ; entry: br label %loop @@ -140,17 +145,18 @@ exit: ; Same as @steps_match_but_different_access_sizes_1, but with source and sink ; accesses flipped. define void @steps_match_but_different_access_sizes_2(ptr %a, ptr %b, i64 %n) { -; CHECK-LABEL: @steps_match_but_different_access_sizes_2( -; CHECK-NEXT: entry: -; CHECK-NEXT: [[B2:%.*]] = ptrtoint ptr [[B:%.*]] to i64 -; CHECK-NEXT: [[A1:%.*]] = ptrtoint ptr [[A:%.*]] to i64 -; CHECK-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 [[N:%.*]], 4 -; CHECK-NEXT: br i1 [[MIN_ITERS_CHECK]], label %scalar.ph, label %vector.memcheck -; CHECK: vector.memcheck: +; CHECK-LABEL: define void @steps_match_but_different_access_sizes_2( +; CHECK-SAME: ptr [[A:%.*]], ptr [[B:%.*]], i64 [[N:%.*]]) { +; CHECK-NEXT: [[ENTRY:.*:]] +; CHECK-NEXT: [[B2:%.*]] = ptrtoint ptr [[B]] to i64 +; CHECK-NEXT: [[A1:%.*]] = ptrtoint ptr [[A]] to i64 +; CHECK-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 [[N]], 4 +; CHECK-NEXT: br i1 [[MIN_ITERS_CHECK]], [[SCALAR_PH:label %.*]], label %[[VECTOR_MEMCHECK:.*]] +; CHECK: [[VECTOR_MEMCHECK]]: ; CHECK-NEXT: [[TMP0:%.*]] = add i64 [[A1]], 2 ; CHECK-NEXT: [[TMP1:%.*]] = sub i64 [[TMP0]], [[B2]] ; CHECK-NEXT: [[DIFF_CHECK:%.*]] = icmp ult i64 [[TMP1]], 16 -; CHECK-NEXT: br i1 [[DIFF_CHECK]], label %scalar.ph, label %vector.ph +; CHECK-NEXT: br i1 [[DIFF_CHECK]], [[SCALAR_PH]], [[VECTOR_PH:label %.*]] ; entry: br label %loop @@ -175,26 +181,29 @@ exit: ; one of the add-recs used is invariant in the inner loop. ; Test case for PR57315. define void @nested_loop_outer_iv_addrec_invariant_in_inner1(ptr %a, ptr %b, i64 %n) { -; CHECK-LABEL: @nested_loop_outer_iv_addrec_invariant_in_inner1( -; CHECK: entry: -; CHECK-NEXT: [[N_SHL_2:%.]] = shl i64 %n, 2 -; CHECK-NEXT: [[B_GEP_UPPER:%.*]] = getelementptr i8, ptr %b, i64 [[N_SHL_2]] -; CHECK-NEXT: br label %outer - -; CHECK: outer.header: -; CHECK: [[OUTER_IV_SHL_2:%.]] = shl i64 %outer.iv, 2 -; CHECK-NEXT: [[A_GEP_UPPER:%.*]] = getelementptr nuw i8, ptr %a, i64 [[OUTER_IV_SHL_2]] -; CHECK-NEXT: [[OUTER_IV_4:%.]] = add i64 [[OUTER_IV_SHL_2]], 4 -; CHECK-NEXT: [[A_GEP_UPPER_4:%.*]] = getelementptr i8, ptr %a, i64 [[OUTER_IV_4]] -; CHECK: [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 [[N:%.*]], 4 -; CHECK-NEXT: br i1 [[MIN_ITERS_CHECK]], label %scalar.ph, label %vector.memcheck - -; CHECK: vector.memcheck: -; CHECK-NEXT: [[BOUND0:%.*]] = icmp ult ptr [[A_GEP_UPPER]], [[B_GEP_UPPER]] -; CHECK-NEXT: [[BOUND1:%.*]] = icmp ult ptr %b, [[A_GEP_UPPER_4]] +; CHECK-LABEL: define void @nested_loop_outer_iv_addrec_invariant_in_inner1( +; CHECK-SAME: ptr [[A:%.*]], ptr [[B:%.*]], i64 [[N:%.*]]) { +; CHECK-NEXT: [[ENTRY:.*]]: +; CHECK-NEXT: [[TMP0:%.*]] = shl i64 [[N]], 2 +; CHECK-NEXT: [[SCEVGEP2:%.*]] = getelementptr i8, ptr [[B]], i64 [[TMP0]] +; CHECK-NEXT: br label %[[OUTER_HEADER:.*]] +; CHECK: [[OUTER_HEADER]]: +; CHECK-NEXT: [[OUTER_IV:%.*]] = phi i64 [ [[OUTER_IV_NEXT:%.*]], [[OUTER_LATCH:%.*]] ], [ 0, %[[ENTRY]] ] +; CHECK-NEXT: [[TMP1:%.*]] = shl i64 [[OUTER_IV]], 2 +; CHECK-NEXT: [[SCEVGEP:%.*]] = getelementptr nuw i8, ptr [[A]], i64 [[TMP1]] +; CHECK-NEXT: [[TMP2:%.*]] = add i64 [[TMP1]], 4 +; CHECK-NEXT: [[SCEVGEP1:%.*]] = getelementptr i8, ptr [[A]], i64 [[TMP2]] +; CHECK-NEXT: [[GEP_A:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[OUTER_IV]] +; CHECK-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 [[N]], 4 +; CHECK-NEXT: br i1 [[MIN_ITERS_CHECK]], [[SCALAR_PH:label %.*]], label %[[VECTOR_MEMCHECK:.*]] +; CHECK: [[VECTOR_MEMCHECK]]: +; CHECK-NEXT: [[BOUND0:%.*]] = icmp ult ptr [[SCEVGEP]], [[SCEVGEP2]] +; CHECK-NEXT: [[BOUND1:%.*]] = icmp ult ptr [[B]], [[SCEVGEP1]] ; CHECK-NEXT: [[FOUND_CONFLICT:%.*]] = and i1 [[BOUND0]], [[BOUND1]] -; CHECK-NEXT: br i1 [[FOUND_CONFLICT]], label %scalar.ph, label %vector.ph +; CHECK-NEXT: br i1 [[FOUND_CONFLICT]], [[SCALAR_PH]], [[VECTOR_PH:label %.*]] ; + + entry: br label %outer.header @@ -225,26 +234,29 @@ exit: ; Same as @nested_loop_outer_iv_addrec_invariant_in_inner1 but with dependence ; sink and source swapped. define void @nested_loop_outer_iv_addrec_invariant_in_inner2(ptr %a, ptr %b, i64 %n) { -; CHECK-LABEL: @nested_loop_outer_iv_addrec_invariant_in_inner2( -; CHECK: entry: -; CHECK-NEXT: [[N_SHL_2:%.]] = shl i64 %n, 2 -; CHECK-NEXT: [[B_GEP_UPPER:%.*]] = getelementptr i8, ptr %b, i64 [[N_SHL_2]] -; CHECK-NEXT: br label %outer - -; CHECK: outer.header: -; CHECK: [[OUTER_IV_SHL_2:%.]] = shl i64 %outer.iv, 2 -; CHECK-NEXT: [[A_GEP_UPPER:%.*]] = getelementptr nuw i8, ptr %a, i64 [[OUTER_IV_SHL_2]] -; CHECK-NEXT: [[OUTER_IV_4:%.]] = add i64 [[OUTER_IV_SHL_2]], 4 -; CHECK-NEXT: [[A_GEP_UPPER_4:%.*]] = getelementptr i8, ptr %a, i64 [[OUTER_IV_4]] -; CHECK: [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 [[N:%.*]], 4 -; CHECK-NEXT: br i1 [[MIN_ITERS_CHECK]], label %scalar.ph, label %vector.memcheck - -; CHECK: vector.memcheck: -; CHECK-NEXT: [[BOUND0:%.*]] = icmp ult ptr %b, [[A_GEP_UPPER_4]] -; CHECK-NEXT: [[BOUND1:%.*]] = icmp ult ptr [[A_GEP_UPPER]], [[B_GEP_UPPER]] +; CHECK-LABEL: define void @nested_loop_outer_iv_addrec_invariant_in_inner2( +; CHECK-SAME: ptr [[A:%.*]], ptr [[B:%.*]], i64 [[N:%.*]]) { +; CHECK-NEXT: [[ENTRY:.*]]: +; CHECK-NEXT: [[TMP0:%.*]] = shl i64 [[N]], 2 +; CHECK-NEXT: [[SCEVGEP:%.*]] = getelementptr i8, ptr [[B]], i64 [[TMP0]] +; CHECK-NEXT: br label %[[OUTER_HEADER:.*]] +; CHECK: [[OUTER_HEADER]]: +; CHECK-NEXT: [[OUTER_IV:%.*]] = phi i64 [ [[OUTER_IV_NEXT:%.*]], [[OUTER_LATCH:%.*]] ], [ 0, %[[ENTRY]] ] +; CHECK-NEXT: [[TMP1:%.*]] = shl i64 [[OUTER_IV]], 2 +; CHECK-NEXT: [[SCEVGEP1:%.*]] = getelementptr nuw i8, ptr [[A]], i64 [[TMP1]] +; CHECK-NEXT: [[TMP2:%.*]] = add i64 [[TMP1]], 4 +; CHECK-NEXT: [[SCEVGEP2:%.*]] = getelementptr i8, ptr [[A]], i64 [[TMP2]] +; CHECK-NEXT: [[GEP_A:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[OUTER_IV]] +; CHECK-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 [[N]], 4 +; CHECK-NEXT: br i1 [[MIN_ITERS_CHECK]], [[SCALAR_PH:label %.*]], label %[[VECTOR_MEMCHECK:.*]] +; CHECK: [[VECTOR_MEMCHECK]]: +; CHECK-NEXT: [[BOUND0:%.*]] = icmp ult ptr [[B]], [[SCEVGEP2]] +; CHECK-NEXT: [[BOUND1:%.*]] = icmp ult ptr [[SCEVGEP1]], [[SCEVGEP]] ; CHECK-NEXT: [[FOUND_CONFLICT:%.*]] = and i1 [[BOUND0]], [[BOUND1]] -; CHECK-NEXT: br i1 [[FOUND_CONFLICT]], label %scalar.ph, label %vector.ph +; CHECK-NEXT: br i1 [[FOUND_CONFLICT]], [[SCALAR_PH]], [[VECTOR_PH:label %.*]] ; + + entry: br label %outer.header @@ -276,20 +288,21 @@ exit: ; of the outer loop as start value. It is sufficient to subtract the start ; values (%dst, %src) of the outer AddRecs. define void @nested_loop_start_of_inner_ptr_addrec_is_same_outer_addrec(ptr nocapture noundef %dst, ptr nocapture noundef readonly %src, i64 noundef %m, i64 noundef %n) { -; CHECK-LABEL: @nested_loop_start_of_inner_ptr_addrec_is_same_outer_addrec( -; CHECK-NEXT: entry: -; CHECK-NEXT: [[SRC2:%.*]] = ptrtoint ptr [[SRC:%.*]] to i64 -; CHECK-NEXT: [[DST1:%.*]] = ptrtoint ptr [[DST:%.*]] to i64 +; CHECK-LABEL: define void @nested_loop_start_of_inner_ptr_addrec_is_same_outer_addrec( +; CHECK-SAME: ptr noundef captures(none) [[DST:%.*]], ptr noundef readonly captures(none) [[SRC:%.*]], i64 noundef [[M:%.*]], i64 noundef [[N:%.*]]) { +; CHECK-NEXT: [[ENTRY:.*]]: +; CHECK-NEXT: [[SRC2:%.*]] = ptrtoint ptr [[SRC]] to i64 +; CHECK-NEXT: [[DST1:%.*]] = ptrtoint ptr [[DST]] to i64 ; CHECK-NEXT: [[SUB:%.*]] = sub i64 [[DST1]], [[SRC2]] -; CHECK-NEXT: br label [[OUTER_LOOP:%.*]] -; CHECK: outer.loop: -; CHECK-NEXT: [[OUTER_IV:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[OUTER_IV_NEXT:%.*]], [[INNER_EXIT:%.*]] ] +; CHECK-NEXT: br label %[[OUTER_LOOP:.*]] +; CHECK: [[OUTER_LOOP]]: +; CHECK-NEXT: [[OUTER_IV:%.*]] = phi i64 [ 0, %[[ENTRY]] ], [ [[OUTER_IV_NEXT:%.*]], [[INNER_EXIT:%.*]] ] ; CHECK-NEXT: [[MUL:%.*]] = mul nsw i64 [[OUTER_IV]], [[N]] ; CHECK-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 [[N]], 4 -; CHECK-NEXT: br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_MEMCHECK:%.*]] -; CHECK: vector.memcheck: +; CHECK-NEXT: br i1 [[MIN_ITERS_CHECK]], [[SCALAR_PH:label %.*]], label %[[VECTOR_MEMCHECK:.*]] +; CHECK: [[VECTOR_MEMCHECK]]: ; CHECK-NEXT: [[DIFF_CHECK:%.*]] = icmp ult i64 [[SUB]], 16 -; CHECK-NEXT: br i1 [[DIFF_CHECK]], label [[SCALAR_PH]], label [[VECTOR_PH:%.*]] +; CHECK-NEXT: br i1 [[DIFF_CHECK]], [[SCALAR_PH]], [[VECTOR_PH:label %.*]] ; entry: br label %outer.loop @@ -321,15 +334,16 @@ outer.exit: } define void @use_diff_checks_when_retrying_with_rt_checks(i64 %off, ptr %dst, ptr %src) { -; CHECK-LABEL: @use_diff_checks_when_retrying_with_rt_checks( -; CHECK-NEXT: entry: -; CHECK-NEXT: [[SRC2:%.*]] = ptrtoint ptr %src to i64 -; CHECK-NEXT: [[DST1:%.*]] = ptrtoint ptr %dst to i64 -; CHECK-NEXT: br i1 false, label %scalar.ph, label %vector.memcheck -; CHECK: vector.memcheck: -; CHECK-NEXT: [[TMP0:%.*]] = mul i64 %off, -8 +; CHECK-LABEL: define void @use_diff_checks_when_retrying_with_rt_checks( +; CHECK-SAME: i64 [[OFF:%.*]], ptr [[DST:%.*]], ptr [[SRC:%.*]]) { +; CHECK-NEXT: [[ENTRY:.*:]] +; CHECK-NEXT: [[SRC2:%.*]] = ptrtoint ptr [[SRC]] to i64 +; CHECK-NEXT: [[DST1:%.*]] = ptrtoint ptr [[DST]] to i64 +; CHECK-NEXT: br i1 false, [[SCALAR_PH:label %.*]], label %[[VECTOR_MEMCHECK:.*]] +; CHECK: [[VECTOR_MEMCHECK]]: +; CHECK-NEXT: [[TMP0:%.*]] = mul i64 [[OFF]], -8 ; CHECK-NEXT: [[DIFF_CHECK:%.*]] = icmp ult i64 [[TMP0]], 32 -; CHECK-NEXT: [[TMP1:%.*]] = shl i64 %off, 3 +; CHECK-NEXT: [[TMP1:%.*]] = shl i64 [[OFF]], 3 ; CHECK-NEXT: [[TMP2:%.*]] = add i64 [[DST1]], [[TMP1]] ; CHECK-NEXT: [[TMP3:%.*]] = sub i64 [[TMP2]], [[SRC2]] ; CHECK-NEXT: [[DIFF_CHECK3:%.*]] = icmp ult i64 [[TMP3]], 32 @@ -346,9 +360,7 @@ define void @use_diff_checks_when_retrying_with_rt_checks(i64 %off, ptr %dst, pt ; CHECK-NEXT: [[TMP9:%.*]] = sub i64 [[TMP8]], [[SRC2]] ; CHECK-NEXT: [[DIFF_CHECK8:%.*]] = icmp ult i64 [[TMP9]], 32 ; CHECK-NEXT: [[CONFLICT_RDX9:%.*]] = or i1 [[CONFLICT_RDX7]], [[DIFF_CHECK8]] -; CHECK-NEXT: br i1 [[CONFLICT_RDX9]], label %scalar.ph, label %vector.ph -; CHECK: vector.ph: -; CHECK-NEXT: br label %vector.body +; CHECK-NEXT: br i1 [[CONFLICT_RDX9]], [[SCALAR_PH]], [[VECTOR_PH:label %.*]] ; entry: br label %loop