[llvm] 2442310 - [LV] Add additional trip count expansion tests for #92177.

Florian Hahn via llvm-commits llvm-commits at lists.llvm.org
Tue Oct 15 23:40:24 PDT 2024


Author: Florian Hahn
Date: 2024-10-16T07:40:04+01:00
New Revision: 24423107abc23a24d465189ba05e51d1bc31bbf2

URL: https://github.com/llvm/llvm-project/commit/24423107abc23a24d465189ba05e51d1bc31bbf2
DIFF: https://github.com/llvm/llvm-project/commit/24423107abc23a24d465189ba05e51d1bc31bbf2.diff

LOG: [LV] Add additional trip count expansion tests for #92177.

Extra tests for https://github.com/llvm/llvm-project/pull/92177, split
off the PR.

Added: 
    

Modified: 
    llvm/test/Transforms/LoopVectorize/trip-count-expansion-may-introduce-ub.ll

Removed: 
    


################################################################################
diff  --git a/llvm/test/Transforms/LoopVectorize/trip-count-expansion-may-introduce-ub.ll b/llvm/test/Transforms/LoopVectorize/trip-count-expansion-may-introduce-ub.ll
index c63dc9979bce3c..7dfd80a688f312 100644
--- a/llvm/test/Transforms/LoopVectorize/trip-count-expansion-may-introduce-ub.ll
+++ b/llvm/test/Transforms/LoopVectorize/trip-count-expansion-may-introduce-ub.ll
@@ -459,6 +459,7 @@ exit:
   ret i64 %p
 }
 
+; FIXME: currently the expansion of the loop bounds may introduce UB through the division.
 define i64 @multi_exit_4_exit_count_with_udiv_by_value_in_latch(ptr %dst, i64 %N) {
 ; CHECK-LABEL: define i64 @multi_exit_4_exit_count_with_udiv_by_value_in_latch(
 ; CHECK-SAME: ptr [[DST:%.*]], i64 [[N:%.*]]) {
@@ -526,6 +527,152 @@ exit:
   ret i64 %p
 }
 
+declare void @foo()
+
+; FIXME: currently the expansion of the loop bounds may introduce UB through the division.
+define i64 @multi_exit_4_exit_count_with_udiv_by_value_in_latch_call_before_loop(ptr %dst, i64 %N) {
+; CHECK-LABEL: define i64 @multi_exit_4_exit_count_with_udiv_by_value_in_latch_call_before_loop(
+; CHECK-SAME: ptr [[DST:%.*]], i64 [[N:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    call void @foo()
+; CHECK-NEXT:    [[TMP2:%.*]] = udiv i64 42, [[N]]
+; CHECK-NEXT:    [[TMP3:%.*]] = freeze i64 [[TMP2]]
+; CHECK-NEXT:    [[SMAX:%.*]] = call i64 @llvm.smax.i64(i64 [[N]], i64 0)
+; CHECK-NEXT:    [[UMIN:%.*]] = call i64 @llvm.umin.i64(i64 [[TMP3]], i64 [[SMAX]])
+; CHECK-NEXT:    [[TMP4:%.*]] = add nuw nsw i64 [[UMIN]], 1
+; CHECK-NEXT:    [[MIN_ITERS_CHECK:%.*]] = icmp ule i64 [[TMP4]], 4
+; CHECK-NEXT:    br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
+; CHECK:       vector.ph:
+; CHECK-NEXT:    [[N_MOD_VF:%.*]] = urem i64 [[TMP4]], 4
+; CHECK-NEXT:    [[TMP5:%.*]] = icmp eq i64 [[N_MOD_VF]], 0
+; CHECK-NEXT:    [[TMP6:%.*]] = select i1 [[TMP5]], i64 4, i64 [[N_MOD_VF]]
+; CHECK-NEXT:    [[N_VEC:%.*]] = sub i64 [[TMP4]], [[TMP6]]
+; CHECK-NEXT:    br label [[VECTOR_BODY:%.*]]
+; CHECK:       vector.body:
+; CHECK-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
+; CHECK-NEXT:    [[TMP7:%.*]] = add i64 [[INDEX]], 0
+; CHECK-NEXT:    [[TMP8:%.*]] = getelementptr inbounds i32, ptr [[DST]], i64 [[TMP7]]
+; CHECK-NEXT:    [[TMP9:%.*]] = getelementptr inbounds i32, ptr [[TMP8]], i32 0
+; CHECK-NEXT:    store <4 x i32> <i32 1, i32 1, i32 1, i32 1>, ptr [[TMP9]], align 4
+; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
+; CHECK-NEXT:    [[TMP10:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
+; CHECK-NEXT:    br i1 [[TMP10]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP12:![0-9]+]]
+; CHECK:       middle.block:
+; CHECK-NEXT:    br label [[SCALAR_PH]]
+; CHECK:       scalar.ph:
+; CHECK-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
+; CHECK-NEXT:    br label [[LOOP_HEADER:%.*]]
+; CHECK:       loop.header:
+; CHECK-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP_LATCH:%.*]] ]
+; CHECK-NEXT:    [[GEP:%.*]] = getelementptr inbounds i32, ptr [[DST]], i64 [[IV]]
+; CHECK-NEXT:    store i32 1, ptr [[GEP]], align 4
+; CHECK-NEXT:    [[C_0:%.*]] = icmp slt i64 [[IV]], [[N]]
+; CHECK-NEXT:    br i1 [[C_0]], label [[LOOP_LATCH]], label [[EXIT:%.*]]
+; CHECK:       loop.latch:
+; CHECK-NEXT:    [[IV_NEXT]] = add i64 [[IV]], 1
+; CHECK-NEXT:    [[D:%.*]] = udiv i64 42, [[N]]
+; CHECK-NEXT:    [[C_1:%.*]] = icmp slt i64 [[IV]], [[D]]
+; CHECK-NEXT:    br i1 [[C_1]], label [[LOOP_HEADER]], label [[EXIT]], !llvm.loop [[LOOP13:![0-9]+]]
+; CHECK:       exit:
+; CHECK-NEXT:    [[P:%.*]] = phi i64 [ 1, [[LOOP_HEADER]] ], [ 0, [[LOOP_LATCH]] ]
+; CHECK-NEXT:    ret i64 [[P]]
+;
+entry:
+  call void @foo()
+  br label %loop.header
+
+loop.header:
+  %iv = phi i64 [ 0, %entry ], [ %iv.next, %loop.latch ]
+  %gep = getelementptr inbounds i32, ptr %dst, i64 %iv
+  store i32 1, ptr %gep
+  %c.0 = icmp slt i64 %iv, %N
+  br i1 %c.0, label %loop.latch, label %exit
+
+loop.latch:
+  %iv.next = add i64 %iv, 1
+  %d = udiv i64 42, %N
+  %c.1 = icmp slt i64 %iv, %d
+  br i1 %c.1, label %loop.header, label %exit
+
+exit:
+  %p = phi i64 [ 1, %loop.header ], [ 0, %loop.latch]
+  ret i64 %p
+}
+
+; FIXME: currently the expansion of the loop bounds may introduce UB through the division.
+define i64 @multi_exit_4_exit_count_with_udiv_by_value_in_latch_loop_may_not_execute(ptr %dst, i64 %N, i1 %c) {
+; CHECK-LABEL: define i64 @multi_exit_4_exit_count_with_udiv_by_value_in_latch_loop_may_not_execute(
+; CHECK-SAME: ptr [[DST:%.*]], i64 [[N:%.*]], i1 [[C:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    br i1 [[C]], label [[LOOP_HEADER_PREHEADER:%.*]], label [[EXIT:%.*]]
+; CHECK:       loop.header.preheader:
+; CHECK-NEXT:    [[TMP2:%.*]] = udiv i64 42, [[N]]
+; CHECK-NEXT:    [[TMP3:%.*]] = freeze i64 [[TMP2]]
+; CHECK-NEXT:    [[SMAX:%.*]] = call i64 @llvm.smax.i64(i64 [[N]], i64 0)
+; CHECK-NEXT:    [[UMIN:%.*]] = call i64 @llvm.umin.i64(i64 [[TMP3]], i64 [[SMAX]])
+; CHECK-NEXT:    [[TMP4:%.*]] = add nuw nsw i64 [[UMIN]], 1
+; CHECK-NEXT:    [[MIN_ITERS_CHECK:%.*]] = icmp ule i64 [[TMP4]], 4
+; CHECK-NEXT:    br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
+; CHECK:       vector.ph:
+; CHECK-NEXT:    [[N_MOD_VF:%.*]] = urem i64 [[TMP4]], 4
+; CHECK-NEXT:    [[TMP5:%.*]] = icmp eq i64 [[N_MOD_VF]], 0
+; CHECK-NEXT:    [[TMP6:%.*]] = select i1 [[TMP5]], i64 4, i64 [[N_MOD_VF]]
+; CHECK-NEXT:    [[N_VEC:%.*]] = sub i64 [[TMP4]], [[TMP6]]
+; CHECK-NEXT:    br label [[VECTOR_BODY:%.*]]
+; CHECK:       vector.body:
+; CHECK-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
+; CHECK-NEXT:    [[TMP7:%.*]] = add i64 [[INDEX]], 0
+; CHECK-NEXT:    [[TMP8:%.*]] = getelementptr inbounds i32, ptr [[DST]], i64 [[TMP7]]
+; CHECK-NEXT:    [[TMP9:%.*]] = getelementptr inbounds i32, ptr [[TMP8]], i32 0
+; CHECK-NEXT:    store <4 x i32> <i32 1, i32 1, i32 1, i32 1>, ptr [[TMP9]], align 4
+; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
+; CHECK-NEXT:    [[TMP10:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
+; CHECK-NEXT:    br i1 [[TMP10]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP14:![0-9]+]]
+; CHECK:       middle.block:
+; CHECK-NEXT:    br label [[SCALAR_PH]]
+; CHECK:       scalar.ph:
+; CHECK-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[LOOP_HEADER_PREHEADER]] ]
+; CHECK-NEXT:    br label [[LOOP_HEADER:%.*]]
+; CHECK:       loop.header:
+; CHECK-NEXT:    [[IV:%.*]] = phi i64 [ [[IV_NEXT:%.*]], [[LOOP_LATCH:%.*]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
+; CHECK-NEXT:    [[GEP:%.*]] = getelementptr inbounds i32, ptr [[DST]], i64 [[IV]]
+; CHECK-NEXT:    store i32 1, ptr [[GEP]], align 4
+; CHECK-NEXT:    [[C_0:%.*]] = icmp slt i64 [[IV]], [[N]]
+; CHECK-NEXT:    br i1 [[C_0]], label [[LOOP_LATCH]], label [[EXIT_LOOPEXIT:%.*]]
+; CHECK:       loop.latch:
+; CHECK-NEXT:    [[IV_NEXT]] = add i64 [[IV]], 1
+; CHECK-NEXT:    [[D:%.*]] = udiv i64 42, [[N]]
+; CHECK-NEXT:    [[C_1:%.*]] = icmp slt i64 [[IV]], [[D]]
+; CHECK-NEXT:    br i1 [[C_1]], label [[LOOP_HEADER]], label [[EXIT_LOOPEXIT]], !llvm.loop [[LOOP15:![0-9]+]]
+; CHECK:       exit.loopexit:
+; CHECK-NEXT:    [[P_PH:%.*]] = phi i64 [ 0, [[LOOP_LATCH]] ], [ 1, [[LOOP_HEADER]] ]
+; CHECK-NEXT:    br label [[EXIT]]
+; CHECK:       exit:
+; CHECK-NEXT:    [[P:%.*]] = phi i64 [ 2, [[ENTRY:%.*]] ], [ [[P_PH]], [[EXIT_LOOPEXIT]] ]
+; CHECK-NEXT:    ret i64 [[P]]
+;
+entry:
+  br i1 %c, label %loop.header, label %exit
+
+loop.header:
+  %iv = phi i64 [ 0, %entry ], [ %iv.next, %loop.latch ]
+  %gep = getelementptr inbounds i32, ptr %dst, i64 %iv
+  store i32 1, ptr %gep
+  %c.0 = icmp slt i64 %iv, %N
+  br i1 %c.0, label %loop.latch, label %exit
+
+loop.latch:
+  %iv.next = add i64 %iv, 1
+  %d = udiv i64 42, %N
+  %c.1 = icmp slt i64 %iv, %d
+  br i1 %c.1, label %loop.header, label %exit
+
+exit:
+  %p = phi i64 [ 1, %loop.header ], [ 0, %loop.latch], [ 2, %entry ]
+  ret i64 %p
+}
+
+; FIXME: currently the expansion of the loop bounds may introduce UB through the division.
 define i64 @multi_exit_4_exit_count_with_udiv_by_value_in_latch_
diff erent_bounds(ptr %dst, i64 %N, i64 %M) {
 ; CHECK-LABEL: define i64 @multi_exit_4_exit_count_with_udiv_by_value_in_latch_
diff erent_bounds(
 ; CHECK-SAME: ptr [[DST:%.*]], i64 [[N:%.*]], i64 [[M:%.*]]) {
@@ -551,7 +698,7 @@ define i64 @multi_exit_4_exit_count_with_udiv_by_value_in_latch_
diff erent_bounds
 ; CHECK-NEXT:    store <4 x i32> <i32 1, i32 1, i32 1, i32 1>, ptr [[TMP9]], align 4
 ; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
 ; CHECK-NEXT:    [[TMP10:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
-; CHECK-NEXT:    br i1 [[TMP10]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP12:![0-9]+]]
+; CHECK-NEXT:    br i1 [[TMP10]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP16:![0-9]+]]
 ; CHECK:       middle.block:
 ; CHECK-NEXT:    br label [[SCALAR_PH]]
 ; CHECK:       scalar.ph:
@@ -567,7 +714,7 @@ define i64 @multi_exit_4_exit_count_with_udiv_by_value_in_latch_
diff erent_bounds
 ; CHECK-NEXT:    [[IV_NEXT]] = add i64 [[IV]], 1
 ; CHECK-NEXT:    [[D:%.*]] = udiv i64 42, [[M]]
 ; CHECK-NEXT:    [[C_1:%.*]] = icmp slt i64 [[IV]], [[D]]
-; CHECK-NEXT:    br i1 [[C_1]], label [[LOOP_HEADER]], label [[EXIT]], !llvm.loop [[LOOP13:![0-9]+]]
+; CHECK-NEXT:    br i1 [[C_1]], label [[LOOP_HEADER]], label [[EXIT]], !llvm.loop [[LOOP17:![0-9]+]]
 ; CHECK:       exit:
 ; CHECK-NEXT:    [[P:%.*]] = phi i64 [ 1, [[LOOP_HEADER]] ], [ 0, [[LOOP_LATCH]] ]
 ; CHECK-NEXT:    ret i64 [[P]]
@@ -593,16 +740,16 @@ exit:
   ret i64 %p
 }
 
-
+; FIXME: currently the expansion of the loop bounds may introduce UB through the division.
 define i64 @multi_exit_4_exit_count_with_udiv_by_frozen_value_in_latch(ptr %dst, i64 %N) {
 ; CHECK-LABEL: define i64 @multi_exit_4_exit_count_with_udiv_by_frozen_value_in_latch(
 ; CHECK-SAME: ptr [[DST:%.*]], i64 [[N:%.*]]) {
 ; CHECK-NEXT:  entry:
 ; CHECK-NEXT:    [[FR_N:%.*]] = freeze i64 [[N]]
-; CHECK-NEXT:    [[TMP0:%.*]] = udiv i64 42, [[FR_N]]
-; CHECK-NEXT:    [[TMP1:%.*]] = freeze i64 [[TMP0]]
+; CHECK-NEXT:    [[TMP2:%.*]] = udiv i64 42, [[FR_N]]
+; CHECK-NEXT:    [[TMP10:%.*]] = freeze i64 [[TMP2]]
 ; CHECK-NEXT:    [[SMAX:%.*]] = call i64 @llvm.smax.i64(i64 [[N]], i64 0)
-; CHECK-NEXT:    [[UMIN:%.*]] = call i64 @llvm.umin.i64(i64 [[TMP1]], i64 [[SMAX]])
+; CHECK-NEXT:    [[UMIN:%.*]] = call i64 @llvm.umin.i64(i64 [[TMP10]], i64 [[SMAX]])
 ; CHECK-NEXT:    [[TMP3:%.*]] = add nuw nsw i64 [[UMIN]], 1
 ; CHECK-NEXT:    [[MIN_ITERS_CHECK:%.*]] = icmp ule i64 [[TMP3]], 4
 ; CHECK-NEXT:    br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
@@ -620,7 +767,7 @@ define i64 @multi_exit_4_exit_count_with_udiv_by_frozen_value_in_latch(ptr %dst,
 ; CHECK-NEXT:    store <4 x i32> <i32 1, i32 1, i32 1, i32 1>, ptr [[TMP8]], align 4
 ; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
 ; CHECK-NEXT:    [[TMP9:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
-; CHECK-NEXT:    br i1 [[TMP9]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP14:![0-9]+]]
+; CHECK-NEXT:    br i1 [[TMP9]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP18:![0-9]+]]
 ; CHECK:       middle.block:
 ; CHECK-NEXT:    br label [[SCALAR_PH]]
 ; CHECK:       scalar.ph:
@@ -636,7 +783,7 @@ define i64 @multi_exit_4_exit_count_with_udiv_by_frozen_value_in_latch(ptr %dst,
 ; CHECK-NEXT:    [[IV_NEXT]] = add i64 [[IV]], 1
 ; CHECK-NEXT:    [[D:%.*]] = udiv i64 42, [[FR_N]]
 ; CHECK-NEXT:    [[C_1:%.*]] = icmp slt i64 [[IV]], [[D]]
-; CHECK-NEXT:    br i1 [[C_1]], label [[LOOP_HEADER]], label [[EXIT]], !llvm.loop [[LOOP15:![0-9]+]]
+; CHECK-NEXT:    br i1 [[C_1]], label [[LOOP_HEADER]], label [[EXIT]], !llvm.loop [[LOOP19:![0-9]+]]
 ; CHECK:       exit:
 ; CHECK-NEXT:    [[P:%.*]] = phi i64 [ 1, [[LOOP_HEADER]] ], [ 0, [[LOOP_LATCH]] ]
 ; CHECK-NEXT:    ret i64 [[P]]
@@ -688,7 +835,7 @@ define i64 @multi_exit_4_exit_count_with_udiv_by_constant_in_latch(ptr %dst, i64
 ; CHECK-NEXT:    store <4 x i32> <i32 1, i32 1, i32 1, i32 1>, ptr [[TMP6]], align 4
 ; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
 ; CHECK-NEXT:    [[TMP7:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
-; CHECK-NEXT:    br i1 [[TMP7]], label [[MIDDLE_BLOCK:%.*]], label [[LOOP_HEADER]], !llvm.loop [[LOOP16:![0-9]+]]
+; CHECK-NEXT:    br i1 [[TMP7]], label [[MIDDLE_BLOCK:%.*]], label [[LOOP_HEADER]], !llvm.loop [[LOOP20:![0-9]+]]
 ; CHECK:       middle.block:
 ; CHECK-NEXT:    br label [[SCALAR_PH]]
 ; CHECK:       scalar.ph:
@@ -704,7 +851,7 @@ define i64 @multi_exit_4_exit_count_with_udiv_by_constant_in_latch(ptr %dst, i64
 ; CHECK-NEXT:    [[IV_NEXT]] = add i64 [[IV1]], 1
 ; CHECK-NEXT:    [[D:%.*]] = udiv i64 [[N]], 42
 ; CHECK-NEXT:    [[C_1:%.*]] = icmp slt i64 [[IV1]], [[D]]
-; CHECK-NEXT:    br i1 [[C_1]], label [[LOOP_HEADER1]], label [[EXIT]], !llvm.loop [[LOOP17:![0-9]+]]
+; CHECK-NEXT:    br i1 [[C_1]], label [[LOOP_HEADER1]], label [[EXIT]], !llvm.loop [[LOOP21:![0-9]+]]
 ; CHECK:       exit:
 ; CHECK-NEXT:    [[P:%.*]] = phi i64 [ 1, [[LOOP_HEADER1]] ], [ 0, [[LOOP_LATCH]] ]
 ; CHECK-NEXT:    ret i64 [[P]]
@@ -750,7 +897,7 @@ define void @single_exit_tc_with_udiv(ptr %dst, i64 %N) {
 ; CHECK-NEXT:    store <4 x i32> <i32 1, i32 1, i32 1, i32 1>, ptr [[TMP4]], align 4
 ; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
 ; CHECK-NEXT:    [[TMP5:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
-; CHECK-NEXT:    br i1 [[TMP5]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP18:![0-9]+]]
+; CHECK-NEXT:    br i1 [[TMP5]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP22:![0-9]+]]
 ; CHECK:       middle.block:
 ; CHECK-NEXT:    [[CMP_N:%.*]] = icmp eq i64 [[TMP1]], [[N_VEC]]
 ; CHECK-NEXT:    br i1 [[CMP_N]], label [[EXIT:%.*]], label [[SCALAR_PH]]
@@ -764,7 +911,7 @@ define void @single_exit_tc_with_udiv(ptr %dst, i64 %N) {
 ; CHECK-NEXT:    [[IV_NEXT]] = add i64 [[IV]], 1
 ; CHECK-NEXT:    [[D:%.*]] = udiv i64 42, [[N]]
 ; CHECK-NEXT:    [[C_1:%.*]] = icmp slt i64 [[IV]], [[D]]
-; CHECK-NEXT:    br i1 [[C_1]], label [[LOOP]], label [[EXIT]], !llvm.loop [[LOOP19:![0-9]+]]
+; CHECK-NEXT:    br i1 [[C_1]], label [[LOOP]], label [[EXIT]], !llvm.loop [[LOOP23:![0-9]+]]
 ; CHECK:       exit:
 ; CHECK-NEXT:    ret void
 ;
@@ -784,6 +931,7 @@ exit:
   ret void
 }
 
+; FIXME: currently the expansion of the loop bounds may introduce UB through the division.
 define i64 @multi_exit_4_exit_count_with_urem_by_value_in_latch(ptr %dst, i64 %N) {
 ; CHECK-LABEL: define i64 @multi_exit_4_exit_count_with_urem_by_value_in_latch(
 ; CHECK-SAME: ptr [[DST:%.*]], i64 [[N:%.*]]) {
@@ -812,7 +960,7 @@ define i64 @multi_exit_4_exit_count_with_urem_by_value_in_latch(ptr %dst, i64 %N
 ; CHECK-NEXT:    store <4 x i32> <i32 1, i32 1, i32 1, i32 1>, ptr [[TMP8]], align 4
 ; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
 ; CHECK-NEXT:    [[TMP9:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
-; CHECK-NEXT:    br i1 [[TMP9]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP20:![0-9]+]]
+; CHECK-NEXT:    br i1 [[TMP9]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP24:![0-9]+]]
 ; CHECK:       middle.block:
 ; CHECK-NEXT:    br label [[SCALAR_PH]]
 ; CHECK:       scalar.ph:
@@ -828,7 +976,7 @@ define i64 @multi_exit_4_exit_count_with_urem_by_value_in_latch(ptr %dst, i64 %N
 ; CHECK-NEXT:    [[IV_NEXT]] = add i64 [[IV]], 1
 ; CHECK-NEXT:    [[D:%.*]] = urem i64 42, [[N]]
 ; CHECK-NEXT:    [[C_1:%.*]] = icmp slt i64 [[IV]], [[D]]
-; CHECK-NEXT:    br i1 [[C_1]], label [[LOOP_HEADER]], label [[EXIT]], !llvm.loop [[LOOP21:![0-9]+]]
+; CHECK-NEXT:    br i1 [[C_1]], label [[LOOP_HEADER]], label [[EXIT]], !llvm.loop [[LOOP25:![0-9]+]]
 ; CHECK:       exit:
 ; CHECK-NEXT:    [[P:%.*]] = phi i64 [ 1, [[LOOP_HEADER]] ], [ 0, [[LOOP_LATCH]] ]
 ; CHECK-NEXT:    ret i64 [[P]]
@@ -854,6 +1002,7 @@ exit:
   ret i64 %p
 }
 
+; FIXME: currently the expansion of the loop bounds may introduce UB through the division.
 define i64 @multi_exit_4_exit_count_with_urem_by_constant_in_latch(ptr %dst, i64 %N) {
 ; CHECK-LABEL: define i64 @multi_exit_4_exit_count_with_urem_by_constant_in_latch(
 ; CHECK-SAME: ptr [[DST:%.*]], i64 [[N:%.*]]) {
@@ -879,7 +1028,7 @@ define i64 @multi_exit_4_exit_count_with_urem_by_constant_in_latch(ptr %dst, i64
 ; CHECK-NEXT:    store <4 x i32> <i32 1, i32 1, i32 1, i32 1>, ptr [[TMP6]], align 4
 ; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
 ; CHECK-NEXT:    [[TMP7:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
-; CHECK-NEXT:    br i1 [[TMP7]], label [[MIDDLE_BLOCK:%.*]], label [[LOOP_HEADER]], !llvm.loop [[LOOP22:![0-9]+]]
+; CHECK-NEXT:    br i1 [[TMP7]], label [[MIDDLE_BLOCK:%.*]], label [[LOOP_HEADER]], !llvm.loop [[LOOP26:![0-9]+]]
 ; CHECK:       middle.block:
 ; CHECK-NEXT:    br label [[SCALAR_PH]]
 ; CHECK:       scalar.ph:
@@ -895,7 +1044,7 @@ define i64 @multi_exit_4_exit_count_with_urem_by_constant_in_latch(ptr %dst, i64
 ; CHECK-NEXT:    [[IV_NEXT]] = add i64 [[IV1]], 1
 ; CHECK-NEXT:    [[D:%.*]] = urem i64 [[N]], 42
 ; CHECK-NEXT:    [[C_1:%.*]] = icmp slt i64 [[IV1]], [[D]]
-; CHECK-NEXT:    br i1 [[C_1]], label [[LOOP_HEADER1]], label [[EXIT]], !llvm.loop [[LOOP23:![0-9]+]]
+; CHECK-NEXT:    br i1 [[C_1]], label [[LOOP_HEADER1]], label [[EXIT]], !llvm.loop [[LOOP27:![0-9]+]]
 ; CHECK:       exit:
 ; CHECK-NEXT:    [[P:%.*]] = phi i64 [ 1, [[LOOP_HEADER1]] ], [ 0, [[LOOP_LATCH]] ]
 ; CHECK-NEXT:    ret i64 [[P]]
@@ -1007,10 +1156,10 @@ define i64 @multi_exit_4_exit_count_with_udiv_by_value_in_latch1(ptr %dst, i64 %
 ; CHECK-LABEL: define i64 @multi_exit_4_exit_count_with_udiv_by_value_in_latch1(
 ; CHECK-SAME: ptr [[DST:%.*]], i64 [[N:%.*]]) {
 ; CHECK-NEXT:  entry:
-; CHECK-NEXT:    [[TMP0:%.*]] = udiv i64 42, [[N]]
-; CHECK-NEXT:    [[TMP8:%.*]] = freeze i64 [[TMP0]]
+; CHECK-NEXT:    [[TMP9:%.*]] = udiv i64 42, [[N]]
+; CHECK-NEXT:    [[TMP10:%.*]] = freeze i64 [[TMP9]]
 ; CHECK-NEXT:    [[SMAX:%.*]] = call i64 @llvm.smax.i64(i64 [[N]], i64 0)
-; CHECK-NEXT:    [[UMIN:%.*]] = call i64 @llvm.umin.i64(i64 [[TMP8]], i64 [[SMAX]])
+; CHECK-NEXT:    [[UMIN:%.*]] = call i64 @llvm.umin.i64(i64 [[TMP10]], i64 [[SMAX]])
 ; CHECK-NEXT:    [[TMP1:%.*]] = add nuw nsw i64 [[UMIN]], 1
 ; CHECK-NEXT:    [[MIN_ITERS_CHECK:%.*]] = icmp ule i64 [[TMP1]], 4
 ; CHECK-NEXT:    br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
@@ -1028,7 +1177,7 @@ define i64 @multi_exit_4_exit_count_with_udiv_by_value_in_latch1(ptr %dst, i64 %
 ; CHECK-NEXT:    store <4 x i32> <i32 1, i32 1, i32 1, i32 1>, ptr [[TMP6]], align 4
 ; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
 ; CHECK-NEXT:    [[TMP7:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
-; CHECK-NEXT:    br i1 [[TMP7]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP24:![0-9]+]]
+; CHECK-NEXT:    br i1 [[TMP7]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP28:![0-9]+]]
 ; CHECK:       middle.block:
 ; CHECK-NEXT:    br label [[SCALAR_PH]]
 ; CHECK:       scalar.ph:
@@ -1045,7 +1194,7 @@ define i64 @multi_exit_4_exit_count_with_udiv_by_value_in_latch1(ptr %dst, i64 %
 ; CHECK-NEXT:    [[D:%.*]] = udiv i64 42, [[N]]
 ; CHECK-NEXT:    [[X:%.*]] = sub i64 100, [[D]]
 ; CHECK-NEXT:    [[C_1:%.*]] = icmp slt i64 [[IV]], [[D]]
-; CHECK-NEXT:    br i1 [[C_1]], label [[LOOP_HEADER]], label [[EXIT]], !llvm.loop [[LOOP25:![0-9]+]]
+; CHECK-NEXT:    br i1 [[C_1]], label [[LOOP_HEADER]], label [[EXIT]], !llvm.loop [[LOOP29:![0-9]+]]
 ; CHECK:       exit:
 ; CHECK-NEXT:    [[P:%.*]] = phi i64 [ 1, [[LOOP_HEADER]] ], [ 0, [[LOOP_LATCH]] ]
 ; CHECK-NEXT:    ret i64 [[P]]
@@ -1072,6 +1221,119 @@ exit:
   ret i64 %p
 }
 
+define i64 @multi_exit_exit_count_with_udiv_by_0_in_latch(ptr %dst, i64 %N) {
+; CHECK-LABEL: define i64 @multi_exit_exit_count_with_udiv_by_0_in_latch(
+; CHECK-SAME: ptr [[DST:%.*]], i64 [[N:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    br label [[LOOP_HEADER:%.*]]
+; CHECK:       loop.header:
+; CHECK-NEXT:    [[IV:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[IV_NEXT:%.*]], [[LOOP_LATCH:%.*]] ]
+; CHECK-NEXT:    [[GEP:%.*]] = getelementptr inbounds i32, ptr [[DST]], i64 [[IV]]
+; CHECK-NEXT:    store i32 1, ptr [[GEP]], align 4
+; CHECK-NEXT:    [[C_0:%.*]] = icmp slt i64 [[IV]], [[N]]
+; CHECK-NEXT:    br i1 [[C_0]], label [[LOOP_LATCH]], label [[EXIT:%.*]]
+; CHECK:       loop.latch:
+; CHECK-NEXT:    [[IV_NEXT]] = add i64 [[IV]], 1
+; CHECK-NEXT:    [[D:%.*]] = udiv i64 42, 0
+; CHECK-NEXT:    [[C_1:%.*]] = icmp slt i64 [[IV]], [[D]]
+; CHECK-NEXT:    br i1 [[C_1]], label [[LOOP_HEADER]], label [[EXIT]]
+; CHECK:       exit:
+; CHECK-NEXT:    [[P:%.*]] = phi i64 [ 1, [[LOOP_HEADER]] ], [ 0, [[LOOP_LATCH]] ]
+; CHECK-NEXT:    ret i64 [[P]]
+;
+entry:
+  br label %loop.header
+
+loop.header:
+  %iv = phi i64 [ 0, %entry ], [ %iv.next, %loop.latch ]
+  %gep = getelementptr inbounds i32, ptr %dst, i64 %iv
+  store i32 1, ptr %gep
+  %c.0 = icmp slt i64 %iv, %N
+  br i1 %c.0, label %loop.latch, label %exit
+
+loop.latch:
+  %iv.next = add i64 %iv, 1
+  %d = udiv i64 42, 0
+  %c.1 = icmp slt i64 %iv, %d
+  br i1 %c.1, label %loop.header, label %exit
+
+exit:
+  %p = phi i64 [ 1, %loop.header ], [ 0, %loop.latch]
+  ret i64 %p
+}
+
+; FIXME: currently the expansion of the loop bounds may introduce UB through the division.
+define i64 @multi_exit_count_with_udiv_by_value_in_latch_
diff erent_bounds_divisor_non_zero_may_be_poison(ptr %dst, i64 %N, i64 %M) {
+; CHECK-LABEL: define i64 @multi_exit_count_with_udiv_by_value_in_latch_
diff erent_bounds_divisor_non_zero_may_be_poison(
+; CHECK-SAME: ptr [[DST:%.*]], i64 [[N:%.*]], i64 [[M:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[M_1:%.*]] = call i64 @llvm.umax.i64(i64 [[M]], i64 1)
+; CHECK-NEXT:    [[TMP0:%.*]] = udiv i64 42, [[M_1]]
+; CHECK-NEXT:    [[TMP1:%.*]] = freeze i64 [[TMP0]]
+; CHECK-NEXT:    [[SMAX:%.*]] = call i64 @llvm.smax.i64(i64 [[N]], i64 0)
+; CHECK-NEXT:    [[UMIN:%.*]] = call i64 @llvm.umin.i64(i64 [[TMP1]], i64 [[SMAX]])
+; CHECK-NEXT:    [[TMP2:%.*]] = add nuw nsw i64 [[UMIN]], 1
+; CHECK-NEXT:    [[MIN_ITERS_CHECK:%.*]] = icmp ule i64 [[TMP2]], 4
+; CHECK-NEXT:    br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
+; CHECK:       vector.ph:
+; CHECK-NEXT:    [[N_MOD_VF:%.*]] = urem i64 [[TMP2]], 4
+; CHECK-NEXT:    [[TMP3:%.*]] = icmp eq i64 [[N_MOD_VF]], 0
+; CHECK-NEXT:    [[TMP4:%.*]] = select i1 [[TMP3]], i64 4, i64 [[N_MOD_VF]]
+; CHECK-NEXT:    [[N_VEC:%.*]] = sub i64 [[TMP2]], [[TMP4]]
+; CHECK-NEXT:    br label [[VECTOR_BODY:%.*]]
+; CHECK:       vector.body:
+; CHECK-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
+; CHECK-NEXT:    [[TMP5:%.*]] = add i64 [[INDEX]], 0
+; CHECK-NEXT:    [[TMP6:%.*]] = getelementptr inbounds i32, ptr [[DST]], i64 [[TMP5]]
+; CHECK-NEXT:    [[TMP7:%.*]] = getelementptr inbounds i32, ptr [[TMP6]], i32 0
+; CHECK-NEXT:    store <4 x i32> <i32 1, i32 1, i32 1, i32 1>, ptr [[TMP7]], align 4
+; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
+; CHECK-NEXT:    [[TMP8:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
+; CHECK-NEXT:    br i1 [[TMP8]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP30:![0-9]+]]
+; CHECK:       middle.block:
+; CHECK-NEXT:    br label [[SCALAR_PH]]
+; CHECK:       scalar.ph:
+; CHECK-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
+; CHECK-NEXT:    br label [[LOOP_HEADER:%.*]]
+; CHECK:       loop.header:
+; CHECK-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP_LATCH:%.*]] ]
+; CHECK-NEXT:    [[GEP:%.*]] = getelementptr inbounds i32, ptr [[DST]], i64 [[IV]]
+; CHECK-NEXT:    store i32 1, ptr [[GEP]], align 4
+; CHECK-NEXT:    [[C_0:%.*]] = icmp slt i64 [[IV]], [[N]]
+; CHECK-NEXT:    br i1 [[C_0]], label [[LOOP_LATCH]], label [[EXIT:%.*]]
+; CHECK:       loop.latch:
+; CHECK-NEXT:    [[IV_NEXT]] = add i64 [[IV]], 1
+; CHECK-NEXT:    [[D:%.*]] = udiv i64 42, [[M_1]]
+; CHECK-NEXT:    [[C_1:%.*]] = icmp slt i64 [[IV]], [[D]]
+; CHECK-NEXT:    br i1 [[C_1]], label [[LOOP_HEADER]], label [[EXIT]], !llvm.loop [[LOOP31:![0-9]+]]
+; CHECK:       exit:
+; CHECK-NEXT:    [[P:%.*]] = phi i64 [ 1, [[LOOP_HEADER]] ], [ 0, [[LOOP_LATCH]] ]
+; CHECK-NEXT:    ret i64 [[P]]
+;
+entry:
+  %M.1 = call i64 @llvm.umax.i64(i64 %M, i64 1)
+  br label %loop.header
+
+loop.header:
+  %iv = phi i64 [ 0, %entry ], [ %iv.next, %loop.latch ]
+  %gep = getelementptr inbounds i32, ptr %dst, i64 %iv
+  store i32 1, ptr %gep
+  %c.0 = icmp slt i64 %iv, %N
+  br i1 %c.0, label %loop.latch, label %exit
+
+loop.latch:
+  %iv.next = add i64 %iv, 1
+  %d = udiv i64 42, %M.1
+  %c.1 = icmp slt i64 %iv, %d
+  br i1 %c.1, label %loop.header, label %exit
+
+exit:
+  %p = phi i64 [ 1, %loop.header ], [ 0, %loop.latch]
+  ret i64 %p
+}
+
+declare i64 @llvm.umax.i64(i64, i64)
+
 
 ;.
 ; CHECK: [[LOOP0]] = distinct !{[[LOOP0]], [[META1:![0-9]+]], [[META2:![0-9]+]]}
@@ -1100,4 +1362,10 @@ exit:
 ; CHECK: [[LOOP23]] = distinct !{[[LOOP23]], [[META2]], [[META1]]}
 ; CHECK: [[LOOP24]] = distinct !{[[LOOP24]], [[META1]], [[META2]]}
 ; CHECK: [[LOOP25]] = distinct !{[[LOOP25]], [[META2]], [[META1]]}
+; CHECK: [[LOOP26]] = distinct !{[[LOOP26]], [[META1]], [[META2]]}
+; CHECK: [[LOOP27]] = distinct !{[[LOOP27]], [[META2]], [[META1]]}
+; CHECK: [[LOOP28]] = distinct !{[[LOOP28]], [[META1]], [[META2]]}
+; CHECK: [[LOOP29]] = distinct !{[[LOOP29]], [[META2]], [[META1]]}
+; CHECK: [[LOOP30]] = distinct !{[[LOOP30]], [[META1]], [[META2]]}
+; CHECK: [[LOOP31]] = distinct !{[[LOOP31]], [[META2]], [[META1]]}
 ;.


        


More information about the llvm-commits mailing list