[llvm] 6f00170 - [LSR] Rework term-fold tests
Philip Reames via llvm-commits
llvm-commits at lists.llvm.org
Fri Mar 17 16:03:16 PDT 2023
Author: Philip Reames
Date: 2023-03-17T16:03:08-07:00
New Revision: 6f00170159f0427607f8a7847448ee9e36d3fcf7
URL: https://github.com/llvm/llvm-project/commit/6f00170159f0427607f8a7847448ee9e36d3fcf7
DIFF: https://github.com/llvm/llvm-project/commit/6f00170159f0427607f8a7847448ee9e36d3fcf7.diff
LOG: [LSR] Rework term-fold tests
There were two major problems with the tests.
First, with the pointer size being 32 bit and the original IVs also being 32 bit, almost all of the positive tests were actually unsound. An upcoming change will add the appropriate safety check, but the test diffs are really hard to understand without switching the tests to 64 bit pointers first.
Second, checking debug messages for failures is a major bad practice. This should not have been accepted in review at all. The reason is that it makes the *order* of legality checks visibile and modifying any of them becomes annoying and tedious.
Added:
Modified:
llvm/test/Transforms/LoopStrengthReduce/lsr-term-fold-negative-testcase.ll
llvm/test/Transforms/LoopStrengthReduce/lsr-term-fold.ll
Removed:
################################################################################
diff --git a/llvm/test/Transforms/LoopStrengthReduce/lsr-term-fold-negative-testcase.ll b/llvm/test/Transforms/LoopStrengthReduce/lsr-term-fold-negative-testcase.ll
index 2856da967ffa2..895ce1c6a6965 100644
--- a/llvm/test/Transforms/LoopStrengthReduce/lsr-term-fold-negative-testcase.ll
+++ b/llvm/test/Transforms/LoopStrengthReduce/lsr-term-fold-negative-testcase.ll
@@ -1,10 +1,23 @@
+; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --version 2
; REQUIRES: asserts
; RUN: opt < %s -passes="loop-reduce" -S -debug -lsr-term-fold 2>&1 | FileCheck %s
-target datalayout = "e-p:32:32:32-n32"
+target datalayout = "e-p:64:64:64-n64"
define i32 @loop_variant(ptr %ar, i32 %n, i32 %m) {
-; CHECK: Cannot fold on backedge that is loop variant
+; CHECK-LABEL: define i32 @loop_variant
+; CHECK-SAME: (ptr [[AR:%.*]], i32 [[N:%.*]], i32 [[M:%.*]]) {
+; CHECK-NEXT: entry:
+; CHECK-NEXT: br label [[FOR_COND:%.*]]
+; CHECK: for.cond:
+; CHECK-NEXT: [[N_ADDR_0:%.*]] = phi i32 [ [[N]], [[ENTRY:%.*]] ], [ [[MUL:%.*]], [[FOR_COND]] ]
+; CHECK-NEXT: [[CMP:%.*]] = icmp slt i32 [[N_ADDR_0]], [[M]]
+; CHECK-NEXT: [[MUL]] = shl nsw i32 [[N_ADDR_0]], 1
+; CHECK-NEXT: br i1 [[CMP]], label [[FOR_COND]], label [[FOR_END:%.*]]
+; CHECK: for.end:
+; CHECK-NEXT: [[N_ADDR_0_LCSSA:%.*]] = phi i32 [ [[N_ADDR_0]], [[FOR_COND]] ]
+; CHECK-NEXT: ret i32 [[N_ADDR_0_LCSSA]]
+;
entry:
br label %for.cond
@@ -19,8 +32,43 @@ for.end: ; preds = %for.cond
}
define i32 @nested_loop(ptr %ar, i32 %n, i32 %m, i32 %o) {
-; CHECK: Cannot fold on backedge that is loop variant
-; CHECK: Cannot fold on non-innermost loop
+; CHECK-LABEL: define i32 @nested_loop
+; CHECK-SAME: (ptr [[AR:%.*]], i32 [[N:%.*]], i32 [[M:%.*]], i32 [[O:%.*]]) {
+; CHECK-NEXT: entry:
+; CHECK-NEXT: [[CMP15:%.*]] = icmp sgt i32 [[O]], 0
+; CHECK-NEXT: br i1 [[CMP15]], label [[FOR_BODY_PREHEADER:%.*]], label [[FOR_COND_CLEANUP:%.*]]
+; CHECK: for.body.preheader:
+; CHECK-NEXT: br label [[FOR_BODY:%.*]]
+; CHECK: for.cond.cleanup.loopexit:
+; CHECK-NEXT: [[CNT_1_LCSSA_LCSSA:%.*]] = phi i32 [ [[CNT_1_LCSSA:%.*]], [[FOR_COND_CLEANUP3:%.*]] ]
+; CHECK-NEXT: br label [[FOR_COND_CLEANUP]]
+; CHECK: for.cond.cleanup:
+; CHECK-NEXT: [[CNT_0_LCSSA:%.*]] = phi i32 [ 0, [[ENTRY:%.*]] ], [ [[CNT_1_LCSSA_LCSSA]], [[FOR_COND_CLEANUP_LOOPEXIT:%.*]] ]
+; CHECK-NEXT: ret i32 [[CNT_0_LCSSA]]
+; CHECK: for.body:
+; CHECK-NEXT: [[I_017:%.*]] = phi i32 [ [[INC6:%.*]], [[FOR_COND_CLEANUP3]] ], [ 0, [[FOR_BODY_PREHEADER]] ]
+; CHECK-NEXT: [[CNT_016:%.*]] = phi i32 [ [[CNT_1_LCSSA]], [[FOR_COND_CLEANUP3]] ], [ 0, [[FOR_BODY_PREHEADER]] ]
+; CHECK-NEXT: [[SUB:%.*]] = sub nsw i32 [[N]], [[I_017]]
+; CHECK-NEXT: [[CMP212:%.*]] = icmp slt i32 [[SUB]], [[M]]
+; CHECK-NEXT: br i1 [[CMP212]], label [[FOR_BODY4_PREHEADER:%.*]], label [[FOR_COND_CLEANUP3]]
+; CHECK: for.body4.preheader:
+; CHECK-NEXT: br label [[FOR_BODY4:%.*]]
+; CHECK: for.cond.cleanup3.loopexit:
+; CHECK-NEXT: [[INC_LCSSA:%.*]] = phi i32 [ [[INC:%.*]], [[FOR_BODY4]] ]
+; CHECK-NEXT: br label [[FOR_COND_CLEANUP3]]
+; CHECK: for.cond.cleanup3:
+; CHECK-NEXT: [[CNT_1_LCSSA]] = phi i32 [ [[CNT_016]], [[FOR_BODY]] ], [ [[INC_LCSSA]], [[FOR_COND_CLEANUP3_LOOPEXIT:%.*]] ]
+; CHECK-NEXT: [[INC6]] = add nuw nsw i32 [[I_017]], 1
+; CHECK-NEXT: [[CMP:%.*]] = icmp slt i32 [[INC6]], [[O]]
+; CHECK-NEXT: br i1 [[CMP]], label [[FOR_BODY]], label [[FOR_COND_CLEANUP_LOOPEXIT]]
+; CHECK: for.body4:
+; CHECK-NEXT: [[J_014:%.*]] = phi i32 [ [[MUL:%.*]], [[FOR_BODY4]] ], [ [[SUB]], [[FOR_BODY4_PREHEADER]] ]
+; CHECK-NEXT: [[CNT_113:%.*]] = phi i32 [ [[INC]], [[FOR_BODY4]] ], [ [[CNT_016]], [[FOR_BODY4_PREHEADER]] ]
+; CHECK-NEXT: [[INC]] = add nsw i32 [[CNT_113]], 1
+; CHECK-NEXT: [[MUL]] = shl nsw i32 [[J_014]], 1
+; CHECK-NEXT: [[CMP2:%.*]] = icmp slt i32 [[MUL]], [[M]]
+; CHECK-NEXT: br i1 [[CMP2]], label [[FOR_BODY4]], label [[FOR_COND_CLEANUP3_LOOPEXIT]]
+;
entry:
%cmp15 = icmp sgt i32 %o, 0
br i1 %cmp15, label %for.body, label %for.cond.cleanup
@@ -54,8 +102,22 @@ for.body4: ; preds = %for.body, %for.body
declare void @foo(ptr)
define void @NonAddRecIV(ptr %a) {
-; CHECK: SCEV of phi ' %lsr.iv = phi i32 [ %lsr.iv.next, %for.body ], [ 1, %entry ]'
-; CHECK: is not an affine add recursion, not qualified for the terminating condition folding.
+; CHECK-LABEL: define void @NonAddRecIV
+; CHECK-SAME: (ptr [[A:%.*]]) {
+; CHECK-NEXT: entry:
+; CHECK-NEXT: [[UGLYGEP:%.*]] = getelementptr i8, ptr [[A]], i32 84
+; CHECK-NEXT: br label [[FOR_BODY:%.*]]
+; CHECK: for.body:
+; CHECK-NEXT: [[LSR_IV1:%.*]] = phi ptr [ [[UGLYGEP2:%.*]], [[FOR_BODY]] ], [ [[UGLYGEP]], [[ENTRY:%.*]] ]
+; CHECK-NEXT: [[LSR_IV:%.*]] = phi i32 [ [[LSR_IV_NEXT:%.*]], [[FOR_BODY]] ], [ 1, [[ENTRY]] ]
+; CHECK-NEXT: store i32 1, ptr [[LSR_IV1]], align 4
+; CHECK-NEXT: [[LSR_IV_NEXT]] = mul nsw i32 [[LSR_IV]], 2
+; CHECK-NEXT: [[UGLYGEP2]] = getelementptr i8, ptr [[LSR_IV1]], i64 4
+; CHECK-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i32 [[LSR_IV_NEXT]], 65536
+; CHECK-NEXT: br i1 [[EXITCOND_NOT]], label [[FOR_END:%.*]], label [[FOR_BODY]]
+; CHECK: for.end:
+; CHECK-NEXT: ret void
+;
entry:
%uglygep = getelementptr i8, ptr %a, i32 84
br label %for.body
@@ -76,8 +138,25 @@ for.end: ; preds = %for.body
@fp_inc = common global float 0.000000e+00, align 4
define void @NonSCEVableIV(float %init, float* %A, i32 %N) {
-; CHECK: IV of phi ' %x.05 = phi float [ %init, %entry ], [ %add, %for.body ]'
-; CHECK: is not SCEV-able, not qualified for the terminating condition folding.
+; CHECK-LABEL: define void @NonSCEVableIV
+; CHECK-SAME: (float [[INIT:%.*]], ptr [[A:%.*]], i32 [[N:%.*]]) {
+; CHECK-NEXT: entry:
+; CHECK-NEXT: [[TMP0:%.*]] = load float, ptr @fp_inc, align 4
+; CHECK-NEXT: br label [[FOR_BODY:%.*]]
+; CHECK: for.body:
+; CHECK-NEXT: [[LSR_IV1:%.*]] = phi ptr [ [[SCEVGEP:%.*]], [[FOR_BODY]] ], [ [[A]], [[ENTRY:%.*]] ]
+; CHECK-NEXT: [[LSR_IV:%.*]] = phi i64 [ [[LSR_IV_NEXT:%.*]], [[FOR_BODY]] ], [ 1, [[ENTRY]] ]
+; CHECK-NEXT: [[X_05:%.*]] = phi float [ [[INIT]], [[ENTRY]] ], [ [[ADD:%.*]], [[FOR_BODY]] ]
+; CHECK-NEXT: store float [[X_05]], ptr [[LSR_IV1]], align 4
+; CHECK-NEXT: [[ADD]] = fsub float [[X_05]], [[TMP0]]
+; CHECK-NEXT: [[LFTR_WIDEIV:%.*]] = trunc i64 [[LSR_IV]] to i32
+; CHECK-NEXT: [[EXITCOND:%.*]] = icmp eq i32 [[LFTR_WIDEIV]], [[N]]
+; CHECK-NEXT: [[LSR_IV_NEXT]] = add nuw nsw i64 [[LSR_IV]], 1
+; CHECK-NEXT: [[SCEVGEP]] = getelementptr i8, ptr [[LSR_IV1]], i64 4
+; CHECK-NEXT: br i1 [[EXITCOND]], label [[FOR_END:%.*]], label [[FOR_BODY]]
+; CHECK: for.end:
+; CHECK-NEXT: ret void
+;
entry:
%0 = load float, float* @fp_inc, align 4
br label %for.body
@@ -98,7 +177,22 @@ for.end: ; preds = %for.end
}
define void @NonIcmpEqNe(ptr %a) {
-; CHECK: Cannot fold on branching condition that is not an ICmpInst::eq / ICmpInst::ne
+; CHECK-LABEL: define void @NonIcmpEqNe
+; CHECK-SAME: (ptr [[A:%.*]]) {
+; CHECK-NEXT: entry:
+; CHECK-NEXT: [[UGLYGEP:%.*]] = getelementptr i8, ptr [[A]], i64 84
+; CHECK-NEXT: br label [[FOR_BODY:%.*]]
+; CHECK: for.body:
+; CHECK-NEXT: [[LSR_IV1:%.*]] = phi ptr [ [[UGLYGEP2:%.*]], [[FOR_BODY]] ], [ [[UGLYGEP]], [[ENTRY:%.*]] ]
+; CHECK-NEXT: [[LSR_IV:%.*]] = phi i64 [ [[LSR_IV_NEXT:%.*]], [[FOR_BODY]] ], [ 379, [[ENTRY]] ]
+; CHECK-NEXT: store i32 1, ptr [[LSR_IV1]], align 4
+; CHECK-NEXT: [[LSR_IV_NEXT]] = add nsw i64 [[LSR_IV]], -1
+; CHECK-NEXT: [[UGLYGEP2]] = getelementptr i8, ptr [[LSR_IV1]], i64 4
+; CHECK-NEXT: [[EXITCOND_NOT:%.*]] = icmp sle i64 [[LSR_IV_NEXT]], 0
+; CHECK-NEXT: br i1 [[EXITCOND_NOT]], label [[FOR_END:%.*]], label [[FOR_BODY]]
+; CHECK: for.end:
+; CHECK-NEXT: ret void
+;
entry:
%uglygep = getelementptr i8, ptr %a, i64 84
br label %for.body
@@ -117,7 +211,25 @@ for.end: ; preds = %for.body
}
define void @TermCondMoreThanOneUse(ptr %a) {
-;CHECK: Cannot replace terminating condition with more than one use
+; CHECK-LABEL: define void @TermCondMoreThanOneUse
+; CHECK-SAME: (ptr [[A:%.*]]) {
+; CHECK-NEXT: entry:
+; CHECK-NEXT: [[UGLYGEP:%.*]] = getelementptr i8, ptr [[A]], i64 84
+; CHECK-NEXT: [[SCEVGEP:%.*]] = getelementptr i8, ptr [[A]], i64 1600
+; CHECK-NEXT: br label [[FOR_BODY:%.*]]
+; CHECK: for.body:
+; CHECK-NEXT: [[LSR_IV2:%.*]] = phi i64 [ [[LSR_IV_NEXT3:%.*]], [[FOR_BODY]] ], [ -378, [[ENTRY:%.*]] ]
+; CHECK-NEXT: [[LSR_IV1:%.*]] = phi ptr [ [[UGLYGEP2:%.*]], [[FOR_BODY]] ], [ [[UGLYGEP]], [[ENTRY]] ]
+; CHECK-NEXT: store i32 1, ptr [[LSR_IV1]], align 4
+; CHECK-NEXT: [[UGLYGEP2]] = getelementptr i8, ptr [[LSR_IV1]], i64 4
+; CHECK-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i64 [[LSR_IV2]], 0
+; CHECK-NEXT: [[DUMMY:%.*]] = select i1 [[EXITCOND_NOT]], i8 0, i8 1
+; CHECK-NEXT: [[LSR_IV_NEXT3]] = add nsw i64 [[LSR_IV2]], 1
+; CHECK-NEXT: [[LSR_FOLD_TERM_COND_REPLACED_TERM_COND:%.*]] = icmp eq ptr [[UGLYGEP2]], [[SCEVGEP]]
+; CHECK-NEXT: br i1 [[LSR_FOLD_TERM_COND_REPLACED_TERM_COND]], label [[FOR_END:%.*]], label [[FOR_BODY]]
+; CHECK: for.end:
+; CHECK-NEXT: ret void
+;
entry:
%uglygep = getelementptr i8, ptr %a, i64 84
br label %for.body
@@ -143,7 +255,59 @@ for.end: ; preds = %for.body
@histogram_energy_boundaries = global [1001 x double] zeroinitializer, align 8
define void @ebur128_calc_gating_block(ptr %st, ptr %optional_output) {
-; CHECK: Is not safe to expand terminating value for phi node %i.026 = phi i64 [ 0, %for.body7.lr.ph ], [ %inc, %for.body7 ]
+; CHECK-LABEL: define void @ebur128_calc_gating_block
+; CHECK-SAME: (ptr [[ST:%.*]], ptr [[OPTIONAL_OUTPUT:%.*]]) {
+; CHECK-NEXT: entry:
+; CHECK-NEXT: [[TMP0:%.*]] = load i32, ptr [[ST]], align 8
+; CHECK-NEXT: [[CONV:%.*]] = zext i32 [[TMP0]] to i64
+; CHECK-NEXT: [[CMP28_NOT:%.*]] = icmp eq i32 [[TMP0]], 0
+; CHECK-NEXT: br i1 [[CMP28_NOT]], label [[FOR_END13:%.*]], label [[FOR_COND2_PREHEADER_LR_PH:%.*]]
+; CHECK: for.cond2.preheader.lr.ph:
+; CHECK-NEXT: [[AUDIO_DATA_INDEX:%.*]] = getelementptr inbounds [[STRUCT_FFEBUR128STATE:%.*]], ptr [[ST]], i64 0, i32 3
+; CHECK-NEXT: [[TMP1:%.*]] = load i64, ptr [[AUDIO_DATA_INDEX]], align 8
+; CHECK-NEXT: [[DIV:%.*]] = udiv i64 [[TMP1]], [[CONV]]
+; CHECK-NEXT: [[CMP525_NOT:%.*]] = icmp ult i64 [[TMP1]], [[CONV]]
+; CHECK-NEXT: [[AUDIO_DATA:%.*]] = getelementptr inbounds [[STRUCT_FFEBUR128STATE]], ptr [[ST]], i64 0, i32 1
+; CHECK-NEXT: [[UMAX:%.*]] = tail call i64 @llvm.umax.i64(i64 [[DIV]], i64 1)
+; CHECK-NEXT: [[TMP2:%.*]] = shl nuw nsw i64 [[CONV]], 3
+; CHECK-NEXT: br label [[FOR_COND2_PREHEADER:%.*]]
+; CHECK: for.cond2.preheader:
+; CHECK-NEXT: [[LSR_IV1:%.*]] = phi i64 [ [[LSR_IV_NEXT2:%.*]], [[FOR_INC11:%.*]] ], [ 0, [[FOR_COND2_PREHEADER_LR_PH]] ]
+; CHECK-NEXT: [[CHANNEL_SUM_030:%.*]] = phi double [ 0.000000e+00, [[FOR_COND2_PREHEADER_LR_PH]] ], [ [[CHANNEL_SUM_1_LCSSA:%.*]], [[FOR_INC11]] ]
+; CHECK-NEXT: [[C_029:%.*]] = phi i64 [ 0, [[FOR_COND2_PREHEADER_LR_PH]] ], [ [[INC12:%.*]], [[FOR_INC11]] ]
+; CHECK-NEXT: br i1 [[CMP525_NOT]], label [[FOR_INC11]], label [[FOR_BODY7_LR_PH:%.*]]
+; CHECK: for.body7.lr.ph:
+; CHECK-NEXT: [[TMP3:%.*]] = load ptr, ptr [[AUDIO_DATA]], align 8
+; CHECK-NEXT: [[SCEVGEP:%.*]] = getelementptr i8, ptr [[TMP3]], i64 [[LSR_IV1]]
+; CHECK-NEXT: br label [[FOR_BODY7:%.*]]
+; CHECK: for.body7:
+; CHECK-NEXT: [[LSR_IV3:%.*]] = phi ptr [ [[SCEVGEP4:%.*]], [[FOR_BODY7]] ], [ [[SCEVGEP]], [[FOR_BODY7_LR_PH]] ]
+; CHECK-NEXT: [[LSR_IV:%.*]] = phi i64 [ [[LSR_IV_NEXT:%.*]], [[FOR_BODY7]] ], [ [[UMAX]], [[FOR_BODY7_LR_PH]] ]
+; CHECK-NEXT: [[CHANNEL_SUM_127:%.*]] = phi double [ [[CHANNEL_SUM_030]], [[FOR_BODY7_LR_PH]] ], [ [[ADD10:%.*]], [[FOR_BODY7]] ]
+; CHECK-NEXT: [[TMP4:%.*]] = load double, ptr [[LSR_IV3]], align 8
+; CHECK-NEXT: [[ADD10]] = fadd double [[CHANNEL_SUM_127]], [[TMP4]]
+; CHECK-NEXT: [[LSR_IV_NEXT]] = add i64 [[LSR_IV]], -1
+; CHECK-NEXT: [[SCEVGEP4]] = getelementptr i8, ptr [[LSR_IV3]], i64 [[TMP2]]
+; CHECK-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i64 [[LSR_IV_NEXT]], 0
+; CHECK-NEXT: br i1 [[EXITCOND_NOT]], label [[FOR_INC11_LOOPEXIT:%.*]], label [[FOR_BODY7]]
+; CHECK: for.inc11.loopexit:
+; CHECK-NEXT: [[ADD10_LCSSA:%.*]] = phi double [ [[ADD10]], [[FOR_BODY7]] ]
+; CHECK-NEXT: br label [[FOR_INC11]]
+; CHECK: for.inc11:
+; CHECK-NEXT: [[CHANNEL_SUM_1_LCSSA]] = phi double [ [[CHANNEL_SUM_030]], [[FOR_COND2_PREHEADER]] ], [ [[ADD10_LCSSA]], [[FOR_INC11_LOOPEXIT]] ]
+; CHECK-NEXT: [[INC12]] = add nuw nsw i64 [[C_029]], 1
+; CHECK-NEXT: [[LSR_IV_NEXT2]] = add nuw nsw i64 [[LSR_IV1]], 8
+; CHECK-NEXT: [[EXITCOND32_NOT:%.*]] = icmp eq i64 [[INC12]], [[CONV]]
+; CHECK-NEXT: br i1 [[EXITCOND32_NOT]], label [[FOR_END13_LOOPEXIT:%.*]], label [[FOR_COND2_PREHEADER]]
+; CHECK: for.end13.loopexit:
+; CHECK-NEXT: [[CHANNEL_SUM_1_LCSSA_LCSSA:%.*]] = phi double [ [[CHANNEL_SUM_1_LCSSA]], [[FOR_INC11]] ]
+; CHECK-NEXT: br label [[FOR_END13]]
+; CHECK: for.end13:
+; CHECK-NEXT: [[CHANNEL_SUM_0_LCSSA:%.*]] = phi double [ 0.000000e+00, [[ENTRY:%.*]] ], [ [[CHANNEL_SUM_1_LCSSA_LCSSA]], [[FOR_END13_LOOPEXIT]] ]
+; CHECK-NEXT: [[ADD14:%.*]] = fadd double [[CHANNEL_SUM_0_LCSSA]], 0.000000e+00
+; CHECK-NEXT: store double [[ADD14]], ptr [[OPTIONAL_OUTPUT]], align 8
+; CHECK-NEXT: ret void
+;
entry:
%0 = load i32, ptr %st, align 8
%conv = zext i32 %0 to i64
@@ -198,7 +362,24 @@ declare i64 @llvm.umax.i64(i64, i64)
%struct.PAKT_INFO = type { i32, i32, i32, [0 x i32] }
define i64 @alac_seek(ptr %0) {
-; CHECK: Is not safe to expand terminating value for phi node %indvars.iv.i = phi i64 [ 0, %entry ], [ %indvars.iv.next.i, %for.body.i ]
+; CHECK-LABEL: define i64 @alac_seek
+; CHECK-SAME: (ptr [[TMP0:%.*]]) {
+; CHECK-NEXT: entry:
+; CHECK-NEXT: [[DIV:%.*]] = udiv i64 1, 0
+; CHECK-NEXT: [[TMP1:%.*]] = add nuw nsw i64 [[DIV]], 1
+; CHECK-NEXT: [[SCEVGEP:%.*]] = getelementptr i8, ptr [[TMP0]], i64 12
+; CHECK-NEXT: br label [[FOR_BODY_I:%.*]]
+; CHECK: for.body.i:
+; CHECK-NEXT: [[LSR_IV1:%.*]] = phi ptr [ [[SCEVGEP2:%.*]], [[FOR_BODY_I]] ], [ [[SCEVGEP]], [[ENTRY:%.*]] ]
+; CHECK-NEXT: [[LSR_IV:%.*]] = phi i64 [ [[LSR_IV_NEXT:%.*]], [[FOR_BODY_I]] ], [ [[TMP1]], [[ENTRY]] ]
+; CHECK-NEXT: [[TMP2:%.*]] = load i32, ptr [[LSR_IV1]], align 4
+; CHECK-NEXT: [[LSR_IV_NEXT]] = add i64 [[LSR_IV]], -1
+; CHECK-NEXT: [[SCEVGEP2]] = getelementptr i8, ptr [[LSR_IV1]], i64 4
+; CHECK-NEXT: [[EXITCOND_NOT_I:%.*]] = icmp eq i64 [[LSR_IV_NEXT]], 0
+; CHECK-NEXT: br i1 [[EXITCOND_NOT_I]], label [[ALAC_PAKT_BLOCK_OFFSET_EXIT:%.*]], label [[FOR_BODY_I]]
+; CHECK: alac_pakt_block_offset.exit:
+; CHECK-NEXT: ret i64 0
+;
entry:
%div = udiv i64 1, 0
br label %for.body.i
diff --git a/llvm/test/Transforms/LoopStrengthReduce/lsr-term-fold.ll b/llvm/test/Transforms/LoopStrengthReduce/lsr-term-fold.ll
index 56b55bec54310..0755b5ee33589 100644
--- a/llvm/test/Transforms/LoopStrengthReduce/lsr-term-fold.ll
+++ b/llvm/test/Transforms/LoopStrengthReduce/lsr-term-fold.ll
@@ -1,21 +1,13 @@
; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
; RUN: opt < %s -passes="loop-reduce" -S -lsr-term-fold | FileCheck %s
-; There are 3 test cases here regarding the replacing the primary iv with
-; other affine AddRec IV.
-; 1. Loop with constant tripcount
-; 2. Loop with runtime tripcount
-; 3. The replacing AddRec IV is a complicated AddRec. This tests whether
-; the fold terminating condition transformation is writing new terminating
-; condition in the correct type.
-
-target datalayout = "e-p:32:32:32-n32"
+target datalayout = "e-p:64:64:64-n64"
define void @const_tripcount(ptr %a) {
; CHECK-LABEL: @const_tripcount(
; CHECK-NEXT: entry:
; CHECK-NEXT: [[UGLYGEP:%.*]] = getelementptr i8, ptr [[A:%.*]], i64 84
-; CHECK-NEXT: [[SCEVGEP:%.*]] = getelementptr i8, ptr [[A]], i32 1600
+; CHECK-NEXT: [[SCEVGEP:%.*]] = getelementptr i8, ptr [[A]], i64 1600
; CHECK-NEXT: br label [[FOR_BODY:%.*]]
; CHECK: for.body:
; CHECK-NEXT: [[LSR_IV1:%.*]] = phi ptr [ [[UGLYGEP2:%.*]], [[FOR_BODY]] ], [ [[UGLYGEP]], [[ENTRY:%.*]] ]
@@ -47,9 +39,10 @@ define void @runtime_tripcount(ptr %a, i32 %N) {
; CHECK-LABEL: @runtime_tripcount(
; CHECK-NEXT: entry:
; CHECK-NEXT: [[UGLYGEP:%.*]] = getelementptr i8, ptr [[A:%.*]], i32 84
-; CHECK-NEXT: [[TMP0:%.*]] = shl i32 [[N:%.*]], 2
-; CHECK-NEXT: [[TMP1:%.*]] = add i32 [[TMP0]], 84
-; CHECK-NEXT: [[SCEVGEP:%.*]] = getelementptr i8, ptr [[A]], i32 [[TMP1]]
+; CHECK-NEXT: [[TMP0:%.*]] = zext i32 [[N:%.*]] to i64
+; CHECK-NEXT: [[TMP1:%.*]] = shl nuw nsw i64 [[TMP0]], 2
+; CHECK-NEXT: [[TMP2:%.*]] = add nuw nsw i64 [[TMP1]], 84
+; CHECK-NEXT: [[SCEVGEP:%.*]] = getelementptr i8, ptr [[A]], i64 [[TMP2]]
; CHECK-NEXT: br label [[FOR_BODY:%.*]]
; CHECK: for.body:
; CHECK-NEXT: [[LSR_IV1:%.*]] = phi ptr [ [[UGLYGEP2:%.*]], [[FOR_BODY]] ], [ [[UGLYGEP]], [[ENTRY:%.*]] ]
@@ -77,17 +70,21 @@ for.end: ; preds = %for.body
ret void
}
+; The replacing AddRec IV is a complicated AddRec. This tests whether
+; the fold terminating condition transformation is writing new terminating
+; condition in the correct type.
define void @ptr_of_ptr_addrec(ptr %ptrptr, i32 %length) {
; CHECK-LABEL: @ptr_of_ptr_addrec(
; CHECK-NEXT: entry:
; CHECK-NEXT: [[START_PTRPTR:%.*]] = getelementptr ptr, ptr [[PTRPTR:%.*]]
-; CHECK-NEXT: [[TMP0:%.*]] = shl i32 [[LENGTH:%.*]], 2
-; CHECK-NEXT: [[SCEVGEP:%.*]] = getelementptr i8, ptr [[START_PTRPTR]], i32 [[TMP0]]
+; CHECK-NEXT: [[TMP0:%.*]] = zext i32 [[LENGTH:%.*]] to i64
+; CHECK-NEXT: [[TMP1:%.*]] = shl nuw nsw i64 [[TMP0]], 3
+; CHECK-NEXT: [[SCEVGEP:%.*]] = getelementptr i8, ptr [[START_PTRPTR]], i64 [[TMP1]]
; CHECK-NEXT: br label [[FOR_BODY:%.*]]
; CHECK: for.body:
; CHECK-NEXT: [[IT_04:%.*]] = phi ptr [ [[INCDEC_PTR:%.*]], [[FOR_BODY]] ], [ [[START_PTRPTR]], [[ENTRY:%.*]] ]
-; CHECK-NEXT: [[TMP1:%.*]] = load ptr, ptr [[IT_04]], align 8
-; CHECK-NEXT: tail call void @foo(ptr [[TMP1]])
+; CHECK-NEXT: [[TMP2:%.*]] = load ptr, ptr [[IT_04]], align 8
+; CHECK-NEXT: tail call void @foo(ptr [[TMP2]])
; CHECK-NEXT: [[INCDEC_PTR]] = getelementptr inbounds ptr, ptr [[IT_04]], i64 1
; CHECK-NEXT: [[LSR_FOLD_TERM_COND_REPLACED_TERM_COND:%.*]] = icmp eq ptr [[INCDEC_PTR]], [[SCEVGEP]]
; CHECK-NEXT: br i1 [[LSR_FOLD_TERM_COND_REPLACED_TERM_COND]], label [[FOR_END:%.*]], label [[FOR_BODY]]
@@ -120,8 +117,9 @@ define void @iv_start_non_preheader(ptr %mark, i32 signext %length) {
; CHECK-NEXT: [[TOBOOL_NOT3:%.*]] = icmp eq i32 [[LENGTH:%.*]], 0
; CHECK-NEXT: br i1 [[TOBOOL_NOT3]], label [[FOR_COND_CLEANUP:%.*]], label [[FOR_BODY_PREHEADER:%.*]]
; CHECK: for.body.preheader:
-; CHECK-NEXT: [[TMP0:%.*]] = shl i32 [[LENGTH]], 2
-; CHECK-NEXT: [[SCEVGEP:%.*]] = getelementptr i8, ptr [[MARK:%.*]], i32 [[TMP0]]
+; CHECK-NEXT: [[TMP0:%.*]] = zext i32 [[LENGTH]] to i64
+; CHECK-NEXT: [[TMP1:%.*]] = shl nuw nsw i64 [[TMP0]], 3
+; CHECK-NEXT: [[SCEVGEP:%.*]] = getelementptr i8, ptr [[MARK:%.*]], i64 [[TMP1]]
; CHECK-NEXT: br label [[FOR_BODY:%.*]]
; CHECK: for.cond.cleanup.loopexit:
; CHECK-NEXT: br label [[FOR_COND_CLEANUP]]
@@ -129,8 +127,8 @@ define void @iv_start_non_preheader(ptr %mark, i32 signext %length) {
; CHECK-NEXT: ret void
; CHECK: for.body:
; CHECK-NEXT: [[DST_04:%.*]] = phi ptr [ [[INCDEC_PTR:%.*]], [[FOR_BODY]] ], [ [[MARK]], [[FOR_BODY_PREHEADER]] ]
-; CHECK-NEXT: [[TMP1:%.*]] = load ptr, ptr [[DST_04]], align 8
-; CHECK-NEXT: [[TMP2:%.*]] = call ptr @foo(ptr [[TMP1]])
+; CHECK-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DST_04]], align 8
+; CHECK-NEXT: [[TMP3:%.*]] = call ptr @foo(ptr [[TMP2]])
; CHECK-NEXT: [[INCDEC_PTR]] = getelementptr inbounds ptr, ptr [[DST_04]], i64 1
; CHECK-NEXT: [[LSR_FOLD_TERM_COND_REPLACED_TERM_COND:%.*]] = icmp eq ptr [[INCDEC_PTR]], [[SCEVGEP]]
; CHECK-NEXT: br i1 [[LSR_FOLD_TERM_COND_REPLACED_TERM_COND]], label [[FOR_COND_CLEANUP_LOOPEXIT:%.*]], label [[FOR_BODY]]
@@ -164,7 +162,7 @@ for.body: ; preds = %entry, %for.body
define void @inbounds_poison_use(ptr %a) {
; CHECK-LABEL: @inbounds_poison_use(
; CHECK-NEXT: entry:
-; CHECK-NEXT: [[SCEVGEP:%.*]] = getelementptr i8, ptr [[A:%.*]], i32 16
+; CHECK-NEXT: [[SCEVGEP:%.*]] = getelementptr i8, ptr [[A:%.*]], i64 16
; CHECK-NEXT: br label [[FOR_BODY:%.*]]
; CHECK: for.body:
; CHECK-NEXT: [[LSR_IV1:%.*]] = phi ptr [ [[UGLYGEP2:%.*]], [[FOR_BODY]] ], [ [[A]], [[ENTRY:%.*]] ]
@@ -198,9 +196,9 @@ for.end: ; preds = %for.body
define void @iv_size(ptr %a, i128 %N) {
; CHECK-LABEL: @iv_size(
; CHECK-NEXT: entry:
-; CHECK-NEXT: [[TMP0:%.*]] = trunc i128 [[N:%.*]] to i32
-; CHECK-NEXT: [[TMP1:%.*]] = shl i32 [[TMP0]], 2
-; CHECK-NEXT: [[SCEVGEP:%.*]] = getelementptr i8, ptr [[A:%.*]], i32 [[TMP1]]
+; CHECK-NEXT: [[TMP0:%.*]] = trunc i128 [[N:%.*]] to i64
+; CHECK-NEXT: [[TMP1:%.*]] = shl i64 [[TMP0]], 2
+; CHECK-NEXT: [[SCEVGEP:%.*]] = getelementptr i8, ptr [[A:%.*]], i64 [[TMP1]]
; CHECK-NEXT: br label [[FOR_BODY:%.*]]
; CHECK: for.body:
; CHECK-NEXT: [[LSR_IV1:%.*]] = phi ptr [ [[UGLYGEP2:%.*]], [[FOR_BODY]] ], [ [[A]], [[ENTRY:%.*]] ]
More information about the llvm-commits
mailing list