[llvm] r288200 - [StructurizeCFG] Fix infinite loop in rebuildSSA.

[Justin Lebar via llvm-commits]

Author: jlebar
Date: Tue Nov 29 15:49:02 2016
New Revision: 288200

URL: http://llvm.org/viewvc/llvm-project?rev=288200&view=rev
Log:
[StructurizeCFG] Fix infinite loop in rebuildSSA.

Michel Dänzer reported that r288051, "[StructurizeCFG] Use range-based
for loops", introduced a bug into rebuildSSA, wherein we were iterating
over an instruction's use list while modifying it, without taking care
to do this correctly.

Added:
    llvm/trunk/test/Transforms/StructurizeCFG/rebuild-ssa-infinite-loop.ll
Modified:
    llvm/trunk/lib/Transforms/Scalar/StructurizeCFG.cpp

Modified: llvm/trunk/lib/Transforms/Scalar/StructurizeCFG.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Transforms/Scalar/StructurizeCFG.cpp?rev=288200&r1=288199&r2=288200&view=diff
==============================================================================
--- llvm/trunk/lib/Transforms/Scalar/StructurizeCFG.cpp (original)
+++ llvm/trunk/lib/Transforms/Scalar/StructurizeCFG.cpp Tue Nov 29 15:49:02 2016
@@ -834,7 +834,10 @@ void StructurizeCFG::rebuildSSA() {
   for (BasicBlock *BB : ParentRegion->blocks())
     for (Instruction &I : *BB) {
       bool Initialized = false;
-      for (Use &U : I.uses()) {
+      // We may modify the use list as we iterate over it, so be careful to
+      // compute the next element in the use list at the top of the loop.
+      for (auto UI = I.use_begin(), E = I.use_end(); UI != E;) {
+        Use &U = *UI++;
         Instruction *User = cast<Instruction>(U.getUser());
         if (User->getParent() == BB) {
           continue;

Added: llvm/trunk/test/Transforms/StructurizeCFG/rebuild-ssa-infinite-loop.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/Transforms/StructurizeCFG/rebuild-ssa-infinite-loop.ll?rev=288200&view=auto
==============================================================================
--- llvm/trunk/test/Transforms/StructurizeCFG/rebuild-ssa-infinite-loop.ll (added)
+++ llvm/trunk/test/Transforms/StructurizeCFG/rebuild-ssa-infinite-loop.ll Tue Nov 29 15:49:02 2016
@@ -0,0 +1,51 @@
+; RUN: opt -o /dev/null -structurizecfg %s
+
+; The following function caused an infinite loop inside the structurizer's
+; rebuildSSA routine, where we were iterating over an instruction's uses while
+; modifying the use list, without taking care to do this safely.
+
+target triple = "amdgcn--"
+
+declare <4 x float> @llvm.SI.vs.load.input(<16 x i8>, i32, i32) #0
+declare <4 x float> @llvm.amdgcn.image.load.v4f32.v2i32.v8i32(<2 x i32>, <8 x i32>, i32, i1, i1, i1, i1) #1
+declare void @llvm.SI.export(i32, i32, i32, i32, i32, float, float, float, float) #2
+
+define amdgpu_vs void @wrapper(i32 inreg, i32) {
+main_body:
+  %2 = add i32 %1, %0
+  %3 = call <4 x float> @llvm.SI.vs.load.input(<16 x i8> undef, i32 0, i32 %2)
+  %4 = extractelement <4 x float> %3, i32 1
+  %5 = fptosi float %4 to i32
+  %6 = insertelement <2 x i32> undef, i32 %5, i32 1
+  br label %loop11.i
+
+loop11.i:                                         ; preds = %endif46.i, %main_body
+  %7 = phi i32 [ 0, %main_body ], [ %15, %endif46.i ]
+  %8 = icmp sgt i32 %7, 999
+  br i1 %8, label %main.exit, label %if16.i
+
+if16.i:                                           ; preds = %loop11.i
+  %9 = call <4 x float> @llvm.amdgcn.image.load.v4f32.v2i32.v8i32(<2 x i32> %6, <8 x i32> undef, i32 15, i1 true, i1 false, i1 false, i1 false)
+  %10 = extractelement <4 x float> %9, i32 0
+  %11 = fcmp ult float 0.000000e+00, %10
+  br i1 %11, label %if28.i, label %endif46.i
+
+if28.i:                                           ; preds = %if16.i
+  %12 = bitcast float %10 to i32
+  %13 = shl i32 %12, 16
+  %14 = bitcast i32 %13 to float
+  br label %main.exit
+
+endif46.i:                                        ; preds = %if16.i
+  %15 = add i32 %7, 1
+  br label %loop11.i
+
+main.exit:                                        ; preds = %if28.i, %loop11.i
+  %16 = phi float [ %14, %if28.i ], [ 0x36F0800000000000, %loop11.i ]
+  call void @llvm.SI.export(i32 15, i32 0, i32 0, i32 32, i32 0, float %16, float 0.000000e+00, float 0.000000e+00, float 0x36A0000000000000)
+  ret void
+}
+
+attributes #0 = { nounwind readnone }
+attributes #1 = { nounwind readonly }
+attributes #2 = { nounwind }