hanchenye-llvm-project/llvm/test/Transforms/GVN
Taewook Oh 75acec8a14 Do not propagate DebugLoc across basic blocks
Summary:
DebugLoc shouldn't be propagated across basic blocks to prevent incorrect stepping and imprecise sample profile result. rL288903 addressed the wrong DebugLoc propagation issue by limiting the copy of DebugLoc when GVN removes a fully redundant load that is dominated by some other load. However, DebugLoc is still incorrectly propagated in the following example:


```
1:  extern int g;
2: 
3:  void foo(int x, int y, int z) {
4:    if (x)
5:      g = 0;
6:    else
7:      g = 1;
8:
9:    int i = 0;
10:   for ( ; i < y ; i++)
11:     if (i > z)
12:       g++;
13: }

```
Below is LLVM IR representation of the program before GVN:


```
@g = external local_unnamed_addr global i32, align 4

; Function Attrs: nounwind uwtable
define void @foo(i32 %x, i32 %y, i32 %z) local_unnamed_addr #0 !dbg !4 {
entry:
  %not.tobool = icmp eq i32 %x, 0, !dbg !8
  %.sink = zext i1 %not.tobool to i32, !dbg !8
  store i32 %.sink, i32* @g, align 4, !tbaa !9
  %cmp8 = icmp sgt i32 %y, 0, !dbg !13
  br i1 %cmp8, label %for.body.preheader, label %for.end, !dbg !17

for.body.preheader:                               ; preds = %entry
  br label %for.body, !dbg !19

for.body:                                         ; preds = %for.body.preheader, %for.inc
  %i.09 = phi i32 [ %inc4, %for.inc ], [ 0, %for.body.preheader ]
  %cmp1 = icmp sgt i32 %i.09, %z, !dbg !19
  br i1 %cmp1, label %if.then2, label %for.inc, !dbg !21

if.then2:                                         ; preds = %for.body
  %0 = load i32, i32* @g, align 4, !dbg !22, !tbaa !9
  %inc = add nsw i32 %0, 1, !dbg !22
  store i32 %inc, i32* @g, align 4, !dbg !22, !tbaa !9
  br label %for.inc, !dbg !23

for.inc:                                          ; preds = %for.body, %if.then2
  %inc4 = add nuw nsw i32 %i.09, 1, !dbg !24
  %exitcond = icmp ne i32 %inc4, %y, !dbg !13
  br i1 %exitcond, label %for.body, label %for.end.loopexit, !dbg !17

for.end.loopexit:                                 ; preds = %for.inc
  br label %for.end, !dbg !26

for.end:                                          ; preds = %for.end.loopexit, %entry
  ret void, !dbg !26
}

```
where 


```
!21 = !DILocation(line: 11, column: 9, scope: !15)
!22 = !DILocation(line: 12, column: 8, scope: !20)
!23 = !DILocation(line: 12, column: 7, scope: !20)
!24 = !DILocation(line: 10, column: 20, scope: !25)
```

And below is after GVN:


```
@g = external local_unnamed_addr global i32, align 4

define void @foo(i32 %x, i32 %y, i32 %z) local_unnamed_addr !dbg !4 {
entry:
  %not.tobool = icmp eq i32 %x, 0, !dbg !8
  %.sink = zext i1 %not.tobool to i32, !dbg !8
  store i32 %.sink, i32* @g, align 4, !tbaa !9
  %cmp8 = icmp sgt i32 %y, 0, !dbg !13
  br i1 %cmp8, label %for.body.preheader, label %for.end, !dbg !17

for.body.preheader:                               ; preds = %entry
  br label %for.body, !dbg !19

for.body:                                         ; preds = %for.inc, %for.body.preheader
  %0 = phi i32 [ %1, %for.inc ], [ %.sink, %for.body.preheader ], !dbg !21
  %i.09 = phi i32 [ %inc4, %for.inc ], [ 0, %for.body.preheader ]
  %cmp1 = icmp sgt i32 %i.09, %z, !dbg !19
  br i1 %cmp1, label %if.then2, label %for.inc, !dbg !22

if.then2:                                         ; preds = %for.body
  %inc = add nsw i32 %0, 1, !dbg !21
  store i32 %inc, i32* @g, align 4, !dbg !21, !tbaa !9
  br label %for.inc, !dbg !23

for.inc:                                          ; preds = %if.then2, %for.body
  %1 = phi i32 [ %inc, %if.then2 ], [ %0, %for.body ]
  %inc4 = add nuw nsw i32 %i.09, 1, !dbg !24
  %exitcond = icmp ne i32 %inc4, %y, !dbg !13
  br i1 %exitcond, label %for.body, label %for.end.loopexit, !dbg !17

for.end.loopexit:                                 ; preds = %for.inc
  br label %for.end, !dbg !26

for.end:                                          ; preds = %for.end.loopexit, %entry
  ret void, !dbg !26
}

```
As you see, GVN removes the load in if.then2 block and creates a phi instruction in for.body for it. The problem is that DebugLoc of remove load instruction is propagated to the newly created phi instruction, which is wrong. rL288903 cannot handle this case because ValuesPerBlock.size() is not 1 in this example when the load is removed.

Reviewers: aprantl, andreadb, wolfgangp

Reviewed By: andreadb

Subscribers: davide, llvm-commits

Differential Revision: https://reviews.llvm.org/D29254

llvm-svn: 293688
2017-01-31 20:57:13 +00:00
..
PRE [DWARF] Null out the debug locs of load instructions that have been moved by GVN 2017-01-04 23:58:26 +00:00
2007-07-25-DominatedLoop.ll
2007-07-25-InfiniteLoop.ll [GVN] Use FileCheck instead of grep for tests. 2016-07-17 23:21:26 +00:00
2007-07-25-Loop.ll
2007-07-25-NestedLoop.ll
2007-07-25-SinglePredecessor.ll
2007-07-26-InterlockingLoops.ll
2007-07-26-NonRedundant.ll
2007-07-26-PhiErasure.ll [GVN] Use FileCheck instead of grep for tests. 2016-07-17 23:21:26 +00:00
2007-07-30-PredIDom.ll
2007-07-31-NoDomInherit.ll [GVN] Use FileCheck instead of grep for tests. 2016-07-17 23:21:26 +00:00
2007-07-31-RedundantPhi.ll [GVN] Use FileCheck instead of grep for tests. 2016-07-17 23:21:26 +00:00
2008-02-12-UndefLoad.ll [GVN] Use FileCheck instead of grep for tests. 2016-07-17 23:21:26 +00:00
2008-02-13-NewPHI.ll
2008-07-02-Unreachable.ll [GVN] Use FileCheck instead of grep for tests. 2016-07-17 23:21:26 +00:00
2008-12-09-SelfRemove.ll [GVN] Use FileCheck instead of grep for tests. 2016-07-17 23:21:26 +00:00
2008-12-12-RLE-Crash.ll
2008-12-14-rle-reanalyze.ll
2008-12-15-CacheVisited.ll
2009-01-21-SortInvalidation.ll
2009-01-22-SortInvalidation.ll
2009-03-10-PREOnVoid.ll
2009-07-13-MemDepSortFail.ll
2009-11-12-MemDepMallocBitCast.ll
2010-03-31-RedundantPHIs.ll
2010-05-08-OneBit.ll
2010-11-13-Simplify.ll
2011-04-27-phioperands.ll
2011-07-07-MatchIntrinsicExtract.ll
2011-09-07-TypeIdFor.ll
2012-05-22-PreCrash.ll
2016-08-30-MaskedScatterGather.ll IntrArgMemOnly is only defined (and current AA machinery only sanely supports) pointer arguments, and these intrinsics have vector of pointer arguments. Remove ArgMemOnly until we either have the machinery, define a new attribute, or something similar 2016-08-30 19:58:48 +00:00
MemdepMiscompile.ll
assume-equal.ll [Devirtualization] MemDep returns non-local !invariant.group dependencies 2017-01-12 11:33:58 +00:00
basic-undef-test.ll
basic.ll [GVN] Use FileCheck instead of grep for tests. 2016-07-17 23:21:26 +00:00
big-endian.ll Do not widen load for different variable in GVN. 2016-09-09 18:42:35 +00:00
bitcast-of-call.ll [GVN] Use FileCheck instead of grep for tests. 2016-07-17 23:21:26 +00:00
br-identical.ll
calloc-load-removal.ll
calls-nonlocal.ll [GVN] Use FileCheck instead of grep for tests. 2016-07-17 23:21:26 +00:00
calls-readonly.ll [GVN] Use FileCheck instead of grep for tests. 2016-07-17 23:21:26 +00:00
commute.ll
cond_br.ll
cond_br2.ll
condprop.ll
crash-no-aa.ll
crash.ll
dbg-redundant-load.ll When GVN removes a redundant load, it should not modify the debug location of the dominating load. 2016-12-07 12:31:36 +00:00
debugloc.ll Do not propagate DebugLoc across basic blocks 2017-01-31 20:57:13 +00:00
edge.ll
fence.ll [AliasAnalysis] Fences do not modify constant memory location 2017-01-20 00:21:33 +00:00
flags.ll
fold-const-expr.ll Fix some DOS-style line endings that I suspect snuck in from one of the 2016-12-23 02:02:26 +00:00
fpmath.ll
funclet.ll
invariant.group.ll [Devirtualization] MemDep returns non-local !invariant.group dependencies 2017-01-12 11:33:58 +00:00
invariant.start.ll [AliasAnalysis] Treat invariant.start as read-memory 2016-08-09 17:18:05 +00:00
lifetime-simple.ll
load-constant-mem.ll [GVN] Use FileCheck instead of grep for tests. 2016-07-17 23:21:26 +00:00
load-from-unreachable-predecessor.ll
malloc-load-removal.ll
no_speculative_loads_with_asan.ll Do not widen load for different variable in GVN. 2016-09-09 18:42:35 +00:00
noalias.ll
non-local-offset.ll
nonescaping-malloc.ll [GVN] Use FileCheck instead of grep for tests. 2016-07-17 23:21:26 +00:00
null-aliases-nothing.ll
opt-remarks.ll [GVN, OptDiag] Print the interesting instructions involved in missed load-elimination 2016-12-01 17:34:50 +00:00
phi-translate-partial-alias.ll
pr10820.ll
pr12979.ll
pr14166.ll [InstSimplify] Cast folding can be made more generic 2016-07-26 17:58:05 +00:00
pr17732.ll
pr17852.ll
pr24397.ll
pr24426.ll
pr25440.ll Do not widen load for different variable in GVN. 2016-09-09 18:42:35 +00:00
pr28562.ll [IR] andIRFlags and copyIRFlags needs to handle GEP 2016-07-15 05:02:31 +00:00
pre-compare.ll
pre-new-inst.ll
propagate-ir-flags.ll Fixed the lost FastMathFlags in GVN(Global Value Numbering). 2016-11-22 20:52:53 +00:00
range.ll
readattrs.ll
rle-must-alias.ll [GVN] Use FileCheck instead of grep for tests. 2016-07-17 23:21:26 +00:00
rle-no-phi-translate.ll
rle-nonlocal.ll
stale-loop-info.ll [GVN] When merging blocks update LoopInfo if it's available 2016-12-01 03:56:43 +00:00
tbaa.ll [Verifier] Add verification for TBAA metadata 2016-12-11 20:07:15 +00:00
unreachable_block_infinite_loop.ll
volatile-nonvolatile.ll