From 66f0cce05278a31fb53297bb248ab9cd1db33104 Mon Sep 17 00:00:00 2001 From: Philip Reames Date: Thu, 16 May 2019 20:33:41 +0000 Subject: [PATCH] [Tests] Start consolidating lftr tests into a single file git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@360934 91177308-0d34-0410-b5e6-96231b3b80d8 --- .../IndVarSimplify/lftr-udiv-tripcount.ll | 46 ------- test/Transforms/IndVarSimplify/lftr-zext.ll | 41 ------ test/Transforms/IndVarSimplify/lftr.ll | 117 ++++++++++++++++++ test/Transforms/IndVarSimplify/lftr_simple.ll | 39 ------ 4 files changed, 117 insertions(+), 126 deletions(-) delete mode 100644 test/Transforms/IndVarSimplify/lftr-udiv-tripcount.ll delete mode 100644 test/Transforms/IndVarSimplify/lftr-zext.ll create mode 100644 test/Transforms/IndVarSimplify/lftr.ll delete mode 100644 test/Transforms/IndVarSimplify/lftr_simple.ll diff --git a/test/Transforms/IndVarSimplify/lftr-udiv-tripcount.ll b/test/Transforms/IndVarSimplify/lftr-udiv-tripcount.ll deleted file mode 100644 index 371b8e58c23..00000000000 --- a/test/Transforms/IndVarSimplify/lftr-udiv-tripcount.ll +++ /dev/null @@ -1,46 +0,0 @@ -; NOTE: Assertions have been autogenerated by utils/update_test_checks.py -; RUN: opt < %s -indvars -S | FileCheck %s - -target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128-n8:16:32:64-S128" - -; It is okay to do LFTR on this loop even though the trip count is a -; division because in this case the division can be optimized to a -; shift. - -define void @foo(i8* %a, i8 %n) nounwind uwtable ssp { -; CHECK-LABEL: @foo( -; CHECK-NEXT: entry: -; CHECK-NEXT: [[E:%.*]] = icmp sgt i8 [[N:%.*]], 3 -; CHECK-NEXT: br i1 [[E]], label [[LOOP_PREHEADER:%.*]], label [[EXIT:%.*]] -; CHECK: loop.preheader: -; CHECK-NEXT: [[TMP0:%.*]] = add i8 [[N]], 3 -; CHECK-NEXT: [[TMP1:%.*]] = lshr i8 [[TMP0]], 2 -; CHECK-NEXT: [[TMP2:%.*]] = add i8 [[TMP1]], 1 -; CHECK-NEXT: br label [[LOOP:%.*]] -; CHECK: loop: -; CHECK-NEXT: [[I1:%.*]] = phi i8 [ [[I1_INC:%.*]], [[LOOP]] ], [ 0, [[LOOP_PREHEADER]] ] -; CHECK-NEXT: [[I1_INC]] = add nuw nsw i8 [[I1]], 1 -; CHECK-NEXT: store volatile i8 0, i8* [[A:%.*]] -; CHECK-NEXT: [[EXITCOND:%.*]] = icmp ne i8 [[I1_INC]], [[TMP2]] -; CHECK-NEXT: br i1 [[EXITCOND]], label [[LOOP]], label [[EXIT_LOOPEXIT:%.*]] -; CHECK: exit.loopexit: -; CHECK-NEXT: br label [[EXIT]] -; CHECK: exit: -; CHECK-NEXT: ret void -; -entry: - %e = icmp sgt i8 %n, 3 - br i1 %e, label %loop, label %exit - -loop: - %i = phi i8 [ 0, %entry ], [ %i.inc, %loop ] - %i1 = phi i8 [ 0, %entry ], [ %i1.inc, %loop ] - %i.inc = add nsw i8 %i, 4 - %i1.inc = add i8 %i1, 1 - store volatile i8 0, i8* %a - %c = icmp slt i8 %i, %n - br i1 %c, label %loop, label %exit - -exit: - ret void -} diff --git a/test/Transforms/IndVarSimplify/lftr-zext.ll b/test/Transforms/IndVarSimplify/lftr-zext.ll deleted file mode 100644 index 0e49bac550b..00000000000 --- a/test/Transforms/IndVarSimplify/lftr-zext.ll +++ /dev/null @@ -1,41 +0,0 @@ -; NOTE: Assertions have been autogenerated by utils/update_test_checks.py -; RUN: opt < %s -indvars -S | FileCheck %s -target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128-n8:16:32:64-S128" -target triple = "x86_64-apple-macosx10.9.0" - -@data = common global [240 x i8] zeroinitializer, align 16 - -define void @foo(i8* %a) #0 { -; CHECK-LABEL: @foo( -; CHECK-NEXT: entry: -; CHECK-NEXT: br label [[LOOP:%.*]] -; CHECK: loop: -; CHECK-NEXT: [[P_0:%.*]] = phi i8* [ getelementptr inbounds ([240 x i8], [240 x i8]* @data, i64 0, i64 0), [[ENTRY:%.*]] ], [ [[TMP3:%.*]], [[LOOP]] ] -; CHECK-NEXT: [[DOT0:%.*]] = phi i8* [ [[A:%.*]], [[ENTRY]] ], [ [[TMP:%.*]], [[LOOP]] ] -; CHECK-NEXT: [[TMP]] = getelementptr inbounds i8, i8* [[DOT0]], i64 1 -; CHECK-NEXT: [[TMP2:%.*]] = load i8, i8* [[DOT0]], align 1 -; CHECK-NEXT: [[TMP3]] = getelementptr inbounds i8, i8* [[P_0]], i64 1 -; CHECK-NEXT: store i8 [[TMP2]], i8* [[P_0]], align 1 -; CHECK-NEXT: [[EXITCOND:%.*]] = icmp ne i8* [[TMP3]], getelementptr (i8, i8* getelementptr inbounds ([240 x i8], [240 x i8]* @data, i64 0, i64 0), i64 240) -; CHECK-NEXT: br i1 [[EXITCOND]], label [[LOOP]], label [[EXIT:%.*]] -; CHECK: exit: -; CHECK-NEXT: ret void -; -entry: - br label %loop - -loop: - %i.0 = phi i8 [ 0, %entry ], [ %tmp4, %loop ] - %p.0 = phi i8* [ getelementptr inbounds ([240 x i8], [240 x i8]* @data, i64 0, i64 0), %entry ], [ %tmp3, %loop ] - %.0 = phi i8* [ %a, %entry ], [ %tmp, %loop ] - %tmp = getelementptr inbounds i8, i8* %.0, i64 1 - %tmp2 = load i8, i8* %.0, align 1 - %tmp3 = getelementptr inbounds i8, i8* %p.0, i64 1 - store i8 %tmp2, i8* %p.0, align 1 - %tmp4 = add i8 %i.0, 1 - %tmp5 = icmp ult i8 %tmp4, -16 - br i1 %tmp5, label %loop, label %exit - -exit: - ret void -} diff --git a/test/Transforms/IndVarSimplify/lftr.ll b/test/Transforms/IndVarSimplify/lftr.ll new file mode 100644 index 00000000000..57560e2931b --- /dev/null +++ b/test/Transforms/IndVarSimplify/lftr.ll @@ -0,0 +1,117 @@ +; NOTE: Assertions have been autogenerated by utils/update_test_checks.py +; RUN: opt < %s -indvars -dce -S | FileCheck %s + +; LFTR should eliminate the need for the computation of i*i completely. It +; is only used to compute the exit value. + +; Provide legal integer types. +target datalayout = "n8:16:32:64" + +@A = external global i32 + +define i32 @quadratic_setlt() { +; CHECK-LABEL: @quadratic_setlt( +; CHECK-NEXT: entry: +; CHECK-NEXT: br label [[LOOP:%.*]] +; CHECK: loop: +; CHECK-NEXT: [[I:%.*]] = phi i32 [ 7, [[ENTRY:%.*]] ], [ [[I_NEXT:%.*]], [[LOOP]] ] +; CHECK-NEXT: [[I_NEXT]] = add nuw nsw i32 [[I]], 1 +; CHECK-NEXT: store i32 [[I]], i32* @A +; CHECK-NEXT: [[EXITCOND:%.*]] = icmp ne i32 [[I_NEXT]], 33 +; CHECK-NEXT: br i1 [[EXITCOND]], label [[LOOP]], label [[LOOPEXIT:%.*]] +; CHECK: loopexit: +; CHECK-NEXT: ret i32 32 +; +entry: + br label %loop + +loop: + %i = phi i32 [ 7, %entry ], [ %i.next, %loop ] + %i.next = add i32 %i, 1 + store i32 %i, i32* @A + %i2 = mul i32 %i, %i + %c = icmp slt i32 %i2, 1000 + br i1 %c, label %loop, label %loopexit + +loopexit: + ret i32 %i +} + + +@data = common global [240 x i8] zeroinitializer, align 16 + +define void @test_zext(i8* %a) #0 { +; CHECK-LABEL: @test_zext( +; CHECK-NEXT: entry: +; CHECK-NEXT: br label [[LOOP:%.*]] +; CHECK: loop: +; CHECK-NEXT: [[P_0:%.*]] = phi i8* [ getelementptr inbounds ([240 x i8], [240 x i8]* @data, i64 0, i64 0), [[ENTRY:%.*]] ], [ [[TMP3:%.*]], [[LOOP]] ] +; CHECK-NEXT: [[DOT0:%.*]] = phi i8* [ [[A:%.*]], [[ENTRY]] ], [ [[TMP:%.*]], [[LOOP]] ] +; CHECK-NEXT: [[TMP]] = getelementptr inbounds i8, i8* [[DOT0]], i64 1 +; CHECK-NEXT: [[TMP2:%.*]] = load i8, i8* [[DOT0]], align 1 +; CHECK-NEXT: [[TMP3]] = getelementptr inbounds i8, i8* [[P_0]], i64 1 +; CHECK-NEXT: store i8 [[TMP2]], i8* [[P_0]], align 1 +; CHECK-NEXT: [[EXITCOND:%.*]] = icmp ne i8* [[TMP3]], getelementptr (i8, i8* getelementptr inbounds ([240 x i8], [240 x i8]* @data, i64 0, i64 0), i64 240) +; CHECK-NEXT: br i1 [[EXITCOND]], label [[LOOP]], label [[EXIT:%.*]] +; CHECK: exit: +; CHECK-NEXT: ret void +; +entry: + br label %loop + +loop: + %i.0 = phi i8 [ 0, %entry ], [ %tmp4, %loop ] + %p.0 = phi i8* [ getelementptr inbounds ([240 x i8], [240 x i8]* @data, i64 0, i64 0), %entry ], [ %tmp3, %loop ] + %.0 = phi i8* [ %a, %entry ], [ %tmp, %loop ] + %tmp = getelementptr inbounds i8, i8* %.0, i64 1 + %tmp2 = load i8, i8* %.0, align 1 + %tmp3 = getelementptr inbounds i8, i8* %p.0, i64 1 + store i8 %tmp2, i8* %p.0, align 1 + %tmp4 = add i8 %i.0, 1 + %tmp5 = icmp ult i8 %tmp4, -16 + br i1 %tmp5, label %loop, label %exit + +exit: + ret void +} + +; It is okay to do LFTR on this loop even though the trip count is a +; division because in this case the division can be optimized to a +; shift. +define void @test_udiv_as_shift(i8* %a, i8 %n) nounwind uwtable ssp { +; CHECK-LABEL: @test_udiv_as_shift( +; CHECK-NEXT: entry: +; CHECK-NEXT: [[E:%.*]] = icmp sgt i8 [[N:%.*]], 3 +; CHECK-NEXT: br i1 [[E]], label [[LOOP_PREHEADER:%.*]], label [[EXIT:%.*]] +; CHECK: loop.preheader: +; CHECK-NEXT: [[TMP0:%.*]] = add i8 [[N]], 3 +; CHECK-NEXT: [[TMP1:%.*]] = lshr i8 [[TMP0]], 2 +; CHECK-NEXT: [[TMP2:%.*]] = add i8 [[TMP1]], 1 +; CHECK-NEXT: br label [[LOOP:%.*]] +; CHECK: loop: +; CHECK-NEXT: [[I1:%.*]] = phi i8 [ [[I1_INC:%.*]], [[LOOP]] ], [ 0, [[LOOP_PREHEADER]] ] +; CHECK-NEXT: [[I1_INC]] = add nuw nsw i8 [[I1]], 1 +; CHECK-NEXT: store volatile i8 0, i8* [[A:%.*]] +; CHECK-NEXT: [[EXITCOND:%.*]] = icmp ne i8 [[I1_INC]], [[TMP2]] +; CHECK-NEXT: br i1 [[EXITCOND]], label [[LOOP]], label [[EXIT_LOOPEXIT:%.*]] +; CHECK: exit.loopexit: +; CHECK-NEXT: br label [[EXIT]] +; CHECK: exit: +; CHECK-NEXT: ret void +; +entry: + %e = icmp sgt i8 %n, 3 + br i1 %e, label %loop, label %exit + +loop: + %i = phi i8 [ 0, %entry ], [ %i.inc, %loop ] + %i1 = phi i8 [ 0, %entry ], [ %i1.inc, %loop ] + %i.inc = add nsw i8 %i, 4 + %i1.inc = add i8 %i1, 1 + store volatile i8 0, i8* %a + %c = icmp slt i8 %i, %n + br i1 %c, label %loop, label %exit + +exit: + ret void +} diff --git a/test/Transforms/IndVarSimplify/lftr_simple.ll b/test/Transforms/IndVarSimplify/lftr_simple.ll deleted file mode 100644 index fa59baddae4..00000000000 --- a/test/Transforms/IndVarSimplify/lftr_simple.ll +++ /dev/null @@ -1,39 +0,0 @@ -; NOTE: Assertions have been autogenerated by utils/update_test_checks.py -; RUN: opt < %s -indvars -dce -S | FileCheck %s - -; LFTR should eliminate the need for the computation of i*i completely. It -; is only used to compute the exit value. - -; Provide legal integer types. -target datalayout = "n8:16:32:64" - -@A = external global i32 - -define i32 @quadratic_setlt() { -; CHECK-LABEL: @quadratic_setlt( -; CHECK-NEXT: entry: -; CHECK-NEXT: br label [[LOOP:%.*]] -; CHECK: loop: -; CHECK-NEXT: [[I:%.*]] = phi i32 [ 7, [[ENTRY:%.*]] ], [ [[I_NEXT:%.*]], [[LOOP]] ] -; CHECK-NEXT: [[I_NEXT]] = add nuw nsw i32 [[I]], 1 -; CHECK-NEXT: store i32 [[I]], i32* @A -; CHECK-NEXT: [[EXITCOND:%.*]] = icmp ne i32 [[I_NEXT]], 33 -; CHECK-NEXT: br i1 [[EXITCOND]], label [[LOOP]], label [[LOOPEXIT:%.*]] -; CHECK: loopexit: -; CHECK-NEXT: ret i32 32 -; -entry: - br label %loop - -loop: - %i = phi i32 [ 7, %entry ], [ %i.next, %loop ] - %i.next = add i32 %i, 1 - store i32 %i, i32* @A - %i2 = mul i32 %i, %i - %c = icmp slt i32 %i2, 1000 - br i1 %c, label %loop, label %loopexit - -loopexit: - ret i32 %i -} - -- 2.40.0