maxjhandsome
/
llvm-project
forked from OSchip/llvm-project
1
0
Fork 0
Code Issues Pull Requests Packages Releases Wiki Activity

[test][LoopUnroll] Cleanup FullUnroll.ll 

This is in preparation for enabling proper handling of optnone under the
NPM. Most optimizations won't run on an optnone function.

Previously the test would rely on lots of optimizations to optimize the
IR into a simple infinite loop. This is an optnone function, so clearly
that shouldn't be the case.

This IR was found by printing the module before the LoopFullUnrollerPass ran.

Reviewed By: asbirlea

Differential Revision: https://reviews.llvm.org/D85578
This commit is contained in:
Arthur Eubanks 2020-08-07 17:56:31 -07:00
parent e6ea8779c2
commit 72effd8d5b
1 changed files with 24 additions and 53 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

77
llvm/test/Transforms/LoopUnroll/FullUnroll.ll
View File

@ -1,70 +1,41 @@
; RUN: opt -passes='default<O1>' -disable-verify --mtriple x86_64-pc-linux-gnu -disable-loop-unrolling=true \
; RUN: -S -o - %s | FileCheck %s
; RUN: opt -passes='loop-unroll-full' -disable-verify --mtriple x86_64-pc-linux-gnu -S -o - %s | FileCheck %s
; This checks that the loop full unroller will fire in the new pass manager
; when forced via #pragma in the source (or annotation in the code).
target datalayout = "e-m:e-p270:32:32-p271:32:32-p272:64:64-i64:64-f80:128-n8:16:32:64-S128"
target triple = "x86_64-unknown-linux-gnu"
; We don't end up deleting the loop, merely turning it infinite, but we remove
; everything inside of it so check for the loop structure and absence of
; conditional branches.
; CHECK-LABEL: bb
; CHECK: br label
; CHECK-NOT: br i1
; CHECK: br label
; Completely unroll the inner loop
; CHECK-LABEL: @foo
; CHECK: br i1
; CHECK-NOT: br i1
; Function Attrs: noinline nounwind optnone uwtable
define void @foo() #0 {
define void @foo() local_unnamed_addr #0 {
bb:
%tmp = alloca [5 x i32*], align 16
%tmp1 = alloca i32, align 4
%tmp2 = alloca i32, align 4
store i32 5, i32* %tmp1, align 4
br label %bb3
br label %bb7.preheader
bb3: ; preds = %bb23, %bb
%tmp4 = load i32, i32* %tmp1, align 4
%tmp5 = icmp ne i32 %tmp4, 0
br i1 %tmp5, label %bb6, label %bb24
bb3.loopexit: ; preds = %bb10
%spec.select.lcssa = phi i32 [ %spec.select, %bb10 ]
%tmp5.not = icmp eq i32 %spec.select.lcssa, 0
br i1 %tmp5.not, label %bb24, label %bb7.preheader
bb6: ; preds = %bb3
store i32 0, i32* %tmp2, align 4
br label %bb7
bb7.preheader: ; preds = %bb3.loopexit, %bb
%tmp1.06 = phi i32 [ 5, %bb ], [ %spec.select.lcssa, %bb3.loopexit ]
br label %bb10
bb7: ; preds = %bb20, %bb6
%tmp8 = load i32, i32* %tmp2, align 4
%tmp9 = icmp slt i32 %tmp8, 5
br i1 %tmp9, label %bb10, label %bb23
bb10: ; preds = %bb7
%tmp11 = load i32, i32* %tmp2, align 4
%tmp12 = sext i32 %tmp11 to i64
%tmp13 = getelementptr inbounds [5 x i32*], [5 x i32*]* %tmp, i64 0, i64 %tmp12
bb10: ; preds = %bb10, %bb7.preheader
%indvars.iv = phi i64 [ 0, %bb7.preheader ], [ %indvars.iv.next, %bb10 ]
%tmp1.14 = phi i32 [ %tmp1.06, %bb7.preheader ], [ %spec.select, %bb10 ]
%tmp13 = getelementptr inbounds [5 x i32*], [5 x i32*]* %tmp, i64 0, i64 %indvars.iv
%tmp14 = load i32*, i32** %tmp13, align 8
%tmp15 = icmp ne i32* %tmp14, null
br i1 %tmp15, label %bb16, label %bb19
%tmp15.not = icmp ne i32* %tmp14, null
%tmp18 = sext i1 %tmp15.not to i32
%spec.select = add nsw i32 %tmp1.14, %tmp18
%indvars.iv.next = add nuw nsw i64 %indvars.iv, 1
%exitcond.not = icmp eq i64 %indvars.iv.next, 5
br i1 %exitcond.not, label %bb3.loopexit, label %bb10, !llvm.loop !1
bb16: ; preds = %bb10
%tmp17 = load i32, i32* %tmp1, align 4
%tmp18 = add nsw i32 %tmp17, -1
store i32 %tmp18, i32* %tmp1, align 4
br label %bb19
bb19: ; preds = %bb16, %bb10
br label %bb20
bb20: ; preds = %bb19
%tmp21 = load i32, i32* %tmp2, align 4
%tmp22 = add nsw i32 %tmp21, 1
store i32 %tmp22, i32* %tmp2, align 4
br label %bb7, !llvm.loop !1
bb23: ; preds = %bb7
br label %bb3
bb24: ; preds = %bb3
bb24: ; preds = %bb3.loopexit
ret void
}