[llvm] a2e9157 - [LV] Make create-induction-resume.ll more robust by adding store.
Florian Hahn via llvm-commits
llvm-commits at lists.llvm.org
Tue Jun 25 03:14:39 PDT 2024
Author: Florian Hahn
Date: 2024-06-25T11:14:13+01:00
New Revision: a2e915704ffb040e544b13bf565ecad3aa661d98
URL: https://github.com/llvm/llvm-project/commit/a2e915704ffb040e544b13bf565ecad3aa661d98
DIFF: https://github.com/llvm/llvm-project/commit/a2e915704ffb040e544b13bf565ecad3aa661d98.diff
LOG: [LV] Make create-induction-resume.ll more robust by adding store.
Without the store, the vector loop body is empty. Add a store to avoid
that, while not impacting the induction resume values that are created.
Added:
Modified:
llvm/test/Transforms/LoopVectorize/create-induction-resume.ll
Removed:
################################################################################
diff --git a/llvm/test/Transforms/LoopVectorize/create-induction-resume.ll b/llvm/test/Transforms/LoopVectorize/create-induction-resume.ll
index d92fbe3a77d6d..aea383226c71e 100644
--- a/llvm/test/Transforms/LoopVectorize/create-induction-resume.ll
+++ b/llvm/test/Transforms/LoopVectorize/create-induction-resume.ll
@@ -1,12 +1,12 @@
; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
-; RUN: opt -S -passes=loop-vectorize < %s | FileCheck %s
+; RUN: opt -S -passes=loop-vectorize -force-vector-width=4 < %s | FileCheck %s
; This is a regression test. Without the fix it crashes on SSAUpdater due to
; LoopVectroizer created a phi node placeholder without incoming values but
; SSAUpdater expects that phi node is completely filled.
target datalayout = "e-m:e-p270:32:32-p271:32:32-p272:64:64-i64:64-f80:128-n8:16:32:64-S128-ni:1-p2:32:8:8:32-ni:2"
-define void @test(i32 %arg, i32 %L1.limit, i32 %L2.switch, i1 %c) {
+define void @test(i32 %arg, i32 %L1.limit, i32 %L2.switch, i1 %c, ptr %dst) {
; CHECK-LABEL: @test(
; CHECK-NEXT: L1.preheader:
; CHECK-NEXT: [[TMP0:%.*]] = sub i32 -1, [[ARG:%.*]]
@@ -36,8 +36,8 @@ define void @test(i32 %arg, i32 %L1.limit, i32 %L2.switch, i1 %c) {
; CHECK-NEXT: br label [[L2_HEADER_BACKEDGE:%.*]]
; CHECK: L2.header:
; CHECK-NEXT: switch i32 [[L2_SWITCH:%.*]], label [[L2_HEADER_BACKEDGE]] [
-; CHECK-NEXT: i32 8, label [[L2_EXIT:%.*]]
-; CHECK-NEXT: i32 20, label [[L2_INNER_HEADER_PREHEADER:%.*]]
+; CHECK-NEXT: i32 8, label [[L2_EXIT:%.*]]
+; CHECK-NEXT: i32 20, label [[L2_INNER_HEADER_PREHEADER:%.*]]
; CHECK-NEXT: ]
; CHECK: L2.header.backedge:
; CHECK-NEXT: br label [[L2_HEADER]]
@@ -46,12 +46,30 @@ define void @test(i32 %arg, i32 %L1.limit, i32 %L2.switch, i1 %c) {
; CHECK: vector.ph:
; CHECK-NEXT: [[TMP3:%.*]] = mul i32 12, [[INDUCTION_IV_LCSSA1]]
; CHECK-NEXT: [[IND_END:%.*]] = add i32 1, [[TMP3]]
+; CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <4 x i32> poison, i32 [[INDUCTION_IV_LCSSA1]], i64 0
+; CHECK-NEXT: [[DOTSPLAT:%.*]] = shufflevector <4 x i32> [[DOTSPLATINSERT]], <4 x i32> poison, <4 x i32> zeroinitializer
+; CHECK-NEXT: [[TMP4:%.*]] = mul <4 x i32> <i32 0, i32 1, i32 2, i32 3>, [[DOTSPLAT]]
+; CHECK-NEXT: [[INDUCTION:%.*]] = add <4 x i32> <i32 1, i32 1, i32 1, i32 1>, [[TMP4]]
+; CHECK-NEXT: [[TMP5:%.*]] = mul i32 [[INDUCTION_IV_LCSSA1]], 4
+; CHECK-NEXT: [[DOTSPLATINSERT3:%.*]] = insertelement <4 x i32> poison, i32 [[TMP5]], i64 0
+; CHECK-NEXT: [[DOTSPLAT4:%.*]] = shufflevector <4 x i32> [[DOTSPLATINSERT3]], <4 x i32> poison, <4 x i32> zeroinitializer
+; CHECK-NEXT: [[BROADCAST_SPLATINSERT:%.*]] = insertelement <4 x i32> poison, i32 [[L1_EXIT_VAL]], i64 0
+; CHECK-NEXT: [[BROADCAST_SPLAT:%.*]] = shufflevector <4 x i32> [[BROADCAST_SPLATINSERT]], <4 x i32> poison, <4 x i32> zeroinitializer
; CHECK-NEXT: br label [[VECTOR_BODY:%.*]]
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
+; CHECK-NEXT: [[VEC_IND:%.*]] = phi <4 x i32> [ [[INDUCTION]], [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
+; CHECK-NEXT: [[OFFSET_IDX:%.*]] = add i64 1, [[INDEX]]
+; CHECK-NEXT: [[TMP6:%.*]] = add i64 [[OFFSET_IDX]], 0
+; CHECK-NEXT: [[TMP7:%.*]] = sub <4 x i32> [[VEC_IND]], [[BROADCAST_SPLAT]]
+; CHECK-NEXT: [[TMP8:%.*]] = sext <4 x i32> [[TMP7]] to <4 x i64>
+; CHECK-NEXT: [[TMP9:%.*]] = getelementptr inbounds i64, ptr [[DST:%.*]], i64 [[TMP6]]
+; CHECK-NEXT: [[TMP10:%.*]] = getelementptr inbounds i64, ptr [[TMP9]], i32 0
+; CHECK-NEXT: store <4 x i64> [[TMP8]], ptr [[TMP10]], align 8
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
-; CHECK-NEXT: [[TMP4:%.*]] = icmp eq i64 [[INDEX_NEXT]], 12
-; CHECK-NEXT: br i1 [[TMP4]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
+; CHECK-NEXT: [[VEC_IND_NEXT]] = add <4 x i32> [[VEC_IND]], [[DOTSPLAT4]]
+; CHECK-NEXT: [[TMP11:%.*]] = icmp eq i64 [[INDEX_NEXT]], 12
+; CHECK-NEXT: br i1 [[TMP11]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
; CHECK: middle.block:
; CHECK-NEXT: br i1 true, label [[L2_HEADER_LOOPEXIT:%.*]], label [[SCALAR_PH]]
; CHECK: scalar.ph:
@@ -63,6 +81,8 @@ define void @test(i32 %arg, i32 %L1.limit, i32 %L2.switch, i1 %c) {
; CHECK-NEXT: [[L2_IV:%.*]] = phi i64 [ [[L2_IV_NEXT:%.*]], [[L2_INNER_HEADER]] ], [ [[BC_RESUME_VAL2]], [[SCALAR_PH]] ]
; CHECK-NEXT: [[L2_ACCUM_NEXT]] = sub i32 [[L2_ACCUM]], [[L1_EXIT_VAL]]
; CHECK-NEXT: [[L2_DUMMY_BUT_NEED_IT:%.*]] = sext i32 [[L2_ACCUM_NEXT]] to i64
+; CHECK-NEXT: [[GEP:%.*]] = getelementptr inbounds i64, ptr [[DST]], i64 [[L2_IV]]
+; CHECK-NEXT: store i64 [[L2_DUMMY_BUT_NEED_IT]], ptr [[GEP]], align 8
; CHECK-NEXT: [[L2_IV_NEXT]] = add nuw nsw i64 [[L2_IV]], 1
; CHECK-NEXT: [[L2_EXIT_COND:%.*]] = icmp ugt i64 [[L2_IV]], 11
; CHECK-NEXT: br i1 [[L2_EXIT_COND]], label [[L2_HEADER_LOOPEXIT]], label [[L2_INNER_HEADER]], !llvm.loop [[LOOP3:![0-9]+]]
@@ -101,6 +121,8 @@ L2.Inner.header: ; preds = %L2.Inner.header, %L
%L2.iv = phi i64 [ %L2.iv.next, %L2.Inner.header ], [ 1, %L2.header ]
%L2.accum.next = sub i32 %L2.accum, %L1.exit.val
%L2.dummy.but.need.it = sext i32 %L2.accum.next to i64
+ %gep = getelementptr inbounds i64 , ptr %dst, i64 %L2.iv
+ store i64 %L2.dummy.but.need.it, ptr %gep
%L2.iv.next = add nuw nsw i64 %L2.iv, 1
%L2.exit_cond = icmp ugt i64 %L2.iv, 11
br i1 %L2.exit_cond, label %L2.header, label %L2.Inner.header
More information about the llvm-commits
mailing list