1de3dc7d23
Currently we fail to detect the case where BTC + 1 wraps, i.e. the vector trip count is 0, In those cases, the minimum iteration count check will fail, and the vector code will never be executed. Explicitly check for this condition in computeMaxVF and avoid trying to vectorize alltogether. Note that a number of tests needed to be updated, because the vector loop would never be executed given the input IR. Fixes https://github.com/llvm/llvm-project/issues/122558.
36 lines
1.3 KiB
LLVM
36 lines
1.3 KiB
LLVM
; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --version 5
|
|
; RUN: opt -p loop-vectorize -force-vector-width=2 -epilogue-vectorization-force-VF=2 -S %s | FileCheck %s
|
|
|
|
target datalayout = "E-m:e-i1:8:16-i8:8:16-i64:64-f128:64-v128:64-a:8:16-n32:64"
|
|
|
|
; Test case for https://github.com/llvm/llvm-project/issues/122558.
|
|
define void @vector_trip_count_0_as_btc_is_all_1(ptr %dst) #0 {
|
|
; CHECK-LABEL: define void @vector_trip_count_0_as_btc_is_all_1(
|
|
; CHECK-SAME: ptr [[DST:%.*]]) {
|
|
; CHECK-NEXT: [[ENTRY:.*]]:
|
|
; CHECK-NEXT: br label %[[LOOP:.*]]
|
|
; CHECK: [[LOOP]]:
|
|
; CHECK-NEXT: [[IV:%.*]] = phi i32 [ 0, %[[ENTRY]] ], [ [[IV_NEXT:%.*]], %[[LOOP]] ]
|
|
; CHECK-NEXT: [[IV_NEXT]] = add i32 [[IV]], -1
|
|
; CHECK-NEXT: [[GEP:%.*]] = getelementptr inbounds i32, ptr [[DST]], i32 [[IV]]
|
|
; CHECK-NEXT: store i32 [[IV_NEXT]], ptr [[GEP]], align 4
|
|
; CHECK-NEXT: [[EC:%.*]] = icmp eq i32 [[IV_NEXT]], 0
|
|
; CHECK-NEXT: br i1 [[EC]], label %[[EXIT:.*]], label %[[LOOP]]
|
|
; CHECK: [[EXIT]]:
|
|
; CHECK-NEXT: ret void
|
|
;
|
|
entry:
|
|
br label %loop
|
|
|
|
loop:
|
|
%iv = phi i32 [ 0, %entry ], [ %iv.next, %loop ]
|
|
%iv.next = add i32 %iv, -1
|
|
%gep = getelementptr inbounds i32, ptr %dst, i32 %iv
|
|
store i32 %iv.next, ptr %gep, align 4
|
|
%ec = icmp eq i32 %iv.next, 0
|
|
br i1 %ec, label %exit, label %loop
|
|
|
|
exit:
|
|
ret void
|
|
}
|