Refine the change in r52258 for avoiding use-before-def conditions

when changing the stride of a comparison so that it's slightly
more precise, by having it scan the instruction list to determine
if there is a use of the condition after the point where the
condition will be inserted.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@52371 91177308-0d34-0410-b5e6-96231b3b80d8

diff --git a/lib/Transforms/Scalar/LoopStrengthReduce.cpp b/lib/Transforms/Scalar/LoopStrengthReduce.cpp
index e9976c59971861f4738b9c3b4c72eef91b48bf47..f44729204a8da3889157745e84f8fe3fcb41c969 100644 (file)
--- a/lib/Transforms/Scalar/LoopStrengthReduce.cpp
+++ b/lib/Transforms/Scalar/LoopStrengthReduce.cpp
@@ -1530,15 +1530,6 @@ namespace {
 ICmpInst *LoopStrengthReduce::ChangeCompareStride(Loop *L, ICmpInst *Cond,
                                                 IVStrideUse* &CondUse,
                                                 const SCEVHandle* &CondStride) {
-  // Forgo this transformation if the condition has multiple uses. This is
-  // over-conservative, but simpler than alternatives. It guards against
-  // comparisons with a use that occurs earlier than the add instruction for the
-  // new stride index.  See
-  // test/Transforms/LoopStrengthReduce/change-compare-stride-trickiness.ll
-  // for an example of this situation.
-  if (!Cond->hasOneUse())
-    return Cond;
-
   if (StrideOrder.size() < 2 ||
       IVUsesByStride[*CondStride].Users.size() != 1)
     return Cond;
@@ -1653,6 +1644,18 @@ ICmpInst *LoopStrengthReduce::ChangeCompareStride(Loop *L, ICmpInst *Cond,
     }
   }
 
+  // Forgo this transformation if it the increment happens to be
+  // unfortunately positioned after the condition, and the condition
+  // has multiple uses which prevent it from being moved immediately
+  // before the branch. See
+  // test/Transforms/LoopStrengthReduce/change-compare-stride-trickiness-*.ll
+  // for an example of this situation.
+  if (!Cond->hasOneUse())
+    for (BasicBlock::iterator I = Cond, E = Cond->getParent()->end();
+         I != E; ++I)
+      if (I == NewIncV)
+        return Cond;
+
   if (NewCmpVal != CmpVal) {
     // Create a new compare instruction using new stride / iv.
     ICmpInst *OldCond = Cond;

diff --git a/test/Transforms/LoopStrengthReduce/change-compare-stride-trickiness-0.ll b/test/Transforms/LoopStrengthReduce/change-compare-stride-trickiness-0.ll
new file mode 100644 (file)
index 0000000..33b0522
--- /dev/null
+++ b/test/Transforms/LoopStrengthReduce/change-compare-stride-trickiness-0.ll
@@ -0,0 +1,24 @@
+; RUN: llvm-as < %s | llc --x86-asm-syntax=att | grep {cmpl    \$4}
+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"
+target triple = "x86_64-apple-darwin9"
+
+; This is like change-compare-stride-trickiness-1.ll except the comparison
+; happens before the relevant use, so the comparison stride can't be
+; easily changed.
+
+define void @foo() {
+entry:
+       br label %loop
+
+loop:
+       %indvar = phi i32 [ 0, %entry ], [ %i.2.0.us1534, %loop ]               ; <i32> [#uses=1]
+       %i.2.0.us1534 = add i32 %indvar, 1              ; <i32> [#uses=3]
+       %tmp611.us1535 = icmp eq i32 %i.2.0.us1534, 4           ; <i1> [#uses=2]
+       %tmp623.us1538 = select i1 %tmp611.us1535, i32 6, i32 0         ; <i32> [#uses=0]
+       %tmp628.us1540 = shl i32 %i.2.0.us1534, 1               ; <i32> [#uses=1]
+       %tmp645646647.us1547 = sext i32 %tmp628.us1540 to i64           ; <i64> [#uses=0]
+       br i1 %tmp611.us1535, label %exit, label %loop
+
+exit:
+       ret void
+}

diff --git a/test/Transforms/LoopStrengthReduce/change-compare-stride-trickiness-1.ll b/test/Transforms/LoopStrengthReduce/change-compare-stride-trickiness-1.ll
new file mode 100644 (file)
index 0000000..06e2312
--- /dev/null
+++ b/test/Transforms/LoopStrengthReduce/change-compare-stride-trickiness-1.ll
@@ -0,0 +1,24 @@
+; RUN: llvm-as < %s | llc --x86-asm-syntax=att | grep {cmpl    \$8}
+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"
+target triple = "x86_64-apple-darwin9"
+
+; This is like change-compare-stride-trickiness-0.ll except the comparison
+; happens after the relevant use, so the comparison stride can be
+; easily changed.
+
+define void @foo() {
+entry:
+       br label %loop
+
+loop:
+       %indvar = phi i32 [ 0, %entry ], [ %i.2.0.us1534, %loop ]               ; <i32> [#uses=1]
+       %i.2.0.us1534 = add i32 %indvar, 1              ; <i32> [#uses=3]
+       %tmp628.us1540 = shl i32 %i.2.0.us1534, 1               ; <i32> [#uses=1]
+       %tmp645646647.us1547 = sext i32 %tmp628.us1540 to i64           ; <i64> [#uses=0]
+       %tmp611.us1535 = icmp eq i32 %i.2.0.us1534, 4           ; <i1> [#uses=2]
+       %tmp623.us1538 = select i1 %tmp611.us1535, i32 6, i32 0         ; <i32> [#uses=0]
+       br i1 %tmp611.us1535, label %exit, label %loop
+
+exit:
+       ret void
+}

diff --git a/test/Transforms/LoopStrengthReduce/change-compare-stride-trickiness.ll b/test/Transforms/LoopStrengthReduce/change-compare-stride-trickiness.ll
deleted file mode 100644 (file)
index 9a54b7b..0000000
--- a/test/Transforms/LoopStrengthReduce/change-compare-stride-trickiness.ll
+++ /dev/null
@@ -1,20 +0,0 @@
-; RUN: llvm-as < %s | llc
-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"
-target triple = "x86_64-apple-darwin9"
-
-define void @foo() {
-entry:
-       br label %loop
-
-loop:
-       %indvar = phi i32 [ 0, %entry ], [ %i.2.0.us1534, %loop ]               ; <i32> [#uses=1]
-       %i.2.0.us1534 = add i32 %indvar, 1              ; <i32> [#uses=3]
-       %tmp611.us1535 = icmp eq i32 %i.2.0.us1534, 4           ; <i1> [#uses=2]
-       %tmp623.us1538 = select i1 %tmp611.us1535, i32 6, i32 0         ; <i32> [#uses=0]
-       %tmp628.us1540 = shl i32 %i.2.0.us1534, 1               ; <i32> [#uses=1]
-       %tmp645646647.us1547 = sext i32 %tmp628.us1540 to i64           ; <i64> [#uses=0]
-       br i1 %tmp611.us1535, label %exit, label %loop
-
-exit:
-       ret void
-}